Single-molecule G-quadruplex
visualisation in live cells and
3D chromatin
Antanas Radzevicˇius
St Catharine’s College
This thesis is submitted in partial fulfilment
of the requirements for the award of the degree of
Doctor of Philosophy of University of Cambridge
Supervised by
Prof. Sir Shankar Balasubramanian
March 2022
Declaration
This thesis is submitted in partial fulfilment of the requirements for the degree of Doctor of
Philosophy of University of Cambridge.
This thesis is the result of my own work and includes nothing which is the outcome of work
done in collaboration except as declared in the preface and specified in the text. It describes
work carried out in the Department of Chemistry of University of Cambridge from October
2017 to March 2022.
It is not substantially the same as any work that has already been submitted before for any
degree or other qualification except as declared in the preface and specified in the text.
It does not exceed the prescribed word limit for the Department of Chemistry Degree Commit-
tee.
Antanas Radzevicˇius
i
Abstract
Single-molecule G-quadruplex visualisation in live cells and 3D chromatin
Antanas Radzevicˇius
The G-quadruplex (G4) is a secondary DNA structure formed by the self-assembly of guanine
nucleotides. Mapping G4 formation in the human genome shows that they are enriched in
functionally important regions such as promoters, enhancers, transcription factor and architec-
tural protein binding sites. However, the roles of G4s in biology remains elusive. Whilst G4s
can exist in a folded or unfolded state, methods are required to ensure that endogenous G4
detection does not perturb G4 folding dynamics. In this work, G4-specific fluorescent probes
are reported which enable single-molecule real-time visualisation of G4 structures in living cells.
Single-molecule imaging was performed at nanomolar probe concentrations providing measure-
ments of relative populations of G4s in living cells without global perturbation of G4 dynamics.
Chemical trapping of the unfolded G4 state reveals that G4s alternate between their folded
and unfolded state over a time course of 20 minutes. G4 formation in live cells was also shown
to be cell-cycle dependent and disrupted by inhibition of transcription and replication. The
observations presented here provide evidence to support dynamic formation of G4s in living
cells.
G4 visualisation methods were then deployed with the aim of investigating the potential in-
volvement of G4s in 3D chromatin organisation. For this, G4 imaging was extended into 3D by
super-resolution microscopy in fixed cells, and combined with single-cell Hi-C method. Hi-C
is a fixed nucleus chromatin conformation capture technique that can simulate 3D chromatin
structure via high-throughput sequencing of DNA strands in close proximity. 3D imaging of
G4s has revealed their localisation throughout the nucleus volume with several regions showing
clusters. When G4 mapping data were overlapped with single-cell Hi-C structures and G4s
were visualised in 3D chromatin, this showed that G4s co-localised with the A compartment
of the nucleus. By overlapping Hi-C structures with 3D super-resolved G4 images of the same
single cells and population derived G4 sequencing data, the aim of this work is to establish a
foundation to understand G4 relationships with respect to higher order 3D chromatin structure.
Part of this work has been published: M. Di Antonio, A. Ponjavic, A. Radzevicˇius, R. T.
Ranasinghe, M. Catalano, X. Zhang, J. Shen, L.-M. Needham, S. F. Lee, D. Klenerman, and
S. Balasubramanian, ”Single- molecule visualization of DNA G-quadruplex formation in live
cells,” Nat. Chem., vol. 12, pp. 832-837, 2020.
ii
Acknowledgements
First of all, I would like to thank Prof. Sir Shankar Balasubramanian for an outstanding
opportunity to work in his group and being such a great mentor. He gave me a great amount of
freedom to pursue my own research questions and ideas which was a wonderful way to creatively
grow as a scientist and as a person.
I would also like to thank Dr Chris Lowe, Dr Mark Farrow and Jo Lockhart for their everyday
support. Thank you to Dr David Tannahill for the numerous discussions that we had and all
his help throughout the PhD. Special thanks to Dr Marco Di Antonio and Dr Aleks Ponjavic
who taught me many things at the start of this scientific journey. It was a pleasure and an
amazing experience to work with them leading to my first journal publication.
A lot of work throughout my PhD has been part of collaborations and I cannot thank enough
my colleagues from Prof. Sir David Klenerman, Prof. Steven Lee and Prof. Ernest Laue groups.
Without them this work would not be possible and I would like to thank everyone for their
massive support. Thank you Ziwei Zhang, Dr Aleks Ponjavic, Edward Sanders, Dr Lisa-Maria
Needham, Dr David Lando and Aleksandra Jartseva.
I would like to thank all members of the Balasubramanian group for all the interesting discus-
sions throughout the years and good humour. This group has been a great place to work and
flourish. This PhD would really be meaningless if not for the people and friends I met along
the way.
My thanks goes to Dr Sam Roberts, Silvia Galli, Dr Jiazhen Shen, Dr Angela Simeone, Isabel
Esain-Garcia, Dr Angie Kirchner, Dr Xiaoyun Zhang, Dr Max Lee and Dr Joa˜o Nogueira who
helped me with their expertise at many points of my PhD.
Thank you to everyone who proofread this thesis and for any contributions however small that
helped to improve this work. Thank you Dr David Tannahill, Ben Mortishire-Smith, Dr Sean
Flynn, Dr Jochen Spiegel, Raminta Tomkute˙, Dr Mark Farrow, Matt Simpson, Dr Kim Liu,
Dr Jack Hardwick and Francesca Ridgeway Bishop.
My thanks goes to Areeb, Kim, Matt, Ben, Bel, Sam, Max and Jane for all the wonderful time
we spent in and away from the lab.
Finally, I would like to thank my family for their love and care. They make my world a brighter
place and help to cope with harder moments in life. I am grateful to my parents who made
it all possible by providing best opportunities in life through love and support, by teaching
strong morals and values. Thank you to my beloved Raminta for her endless support and
encouragement, for shining bright caring light towards me and the world. I am forever grateful.
iii
Abbreviations
 Extinction coefficient
λ Wavelength
τ Residency time
∆Tm Change in melting temperatures
δ Chemical shift
2i Two inhibitors CHIR99021 and PD0325901
3C Chromatin conformation capture
BCL-2 B-cell lymphoma 2
C. elegans Caenorhabditis elegans
E. coli Escherichia coli
HRAS Harvey rat sarcoma viral oncogene homolog
KRAS Kirsten rat sarcoma viral oncogene homolog
m/z Mass and charge ratio
NA Numerical aperture
NRAS Neuroblastoma RAS viral oncogene homolog
A.U. Arbitrary units
AF647 Alexa Fluor 647
AFM Atomic force microscopy
AFR Antibody-fluorophore ratio
Aph Aphidicolin
iv
ATAC-seq Assay for transposase-accessible chromatin with sequencing
A Adenine
BLM Bloom syndrome RecQ like helicase
BPO Benzoyl peroxide
BRCA1 Breast cancer type 1 susceptibility protein
BRCA2 Breast cancer type 2 susceptibility protein
BSA Bovine serum albumin
Cas9 CRISPR associated protein 9
CDR Complementarity-determining region
CD Circular dichroism
CENP-A Centromere protein A
ChIA Chromatin interaction analysis
ChIP Chromatin immunoprecipitation
Chr Chromosome
CNBP Cellular nucleic acid binding protein
comp Complementary
COSY Correlation spectroscopy
CpG C-phosphate-G
CPM Counts per million
CRISPR Clustered Regularly Interspaced Short Palindromic Repeats
CTCF CCCTC-binding factor
CuAAC Copper-catalysed azide-alkyne cycloaddition
CUT&Tag Cleavage under targets and tagmentation
CV Column volume
Cy3/Cy5 Cyanine dyes 3 and 5
v
C Cytosine or Celsius
DamC DNA adenine methyltransferase identification
DAPI 2-(4-Amidinophenyl)-1H -indole-6-carboxamidine
DB Differential binding
dCas9 Dead CRISPR associated protein 9
DCM Dichloromethane
DDX11 DEAD/H-box helicase 11
DEPT Distortionless enhancement by polarization transfer
DHPSF Double helix point spread function
DHX36 DEAH-box helicase 36
DMEM Dulbecco’s modified eagle medium
DMF N,N -Dimethyl formamide
DMSO Dimethyl sulfoxide
DMS Dimethyl sulphate
DNA Deoxyribonucleic acid
DRB 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole
dsDNA Double-stranded deoxyribonucleic acid
ds Double-stranded
D Dark state
EDCI N -(3-Dimethylaminopropyl)-N -ethylcarbodiimide hydrochloride
EDTA Ethylenediaminetetraacetic acid
EGTA Ethylene glycol-bis(β-aminoethyl ether)-N,N,N,N -tetraacetic acid
ELISA Enzyme-linked immunosorbent assay
EMCCD Electron multiplying charge-coupled device
epi Epifluorescence
vi
eq Equivalent
ESI Electrospray ionization
Et Ethyl
FACS Fluorescence-activated cell sorting
FAIRE-seq Formaldehyde-assisted isolation of regulatory elements with sequencing
FAM 6-Carboxyfluorescein
FANCJ Fanconi anemia group J protein
FBS Fetal bovine serum
FISH Fluorescence in situ hybridisation
FLIM Fluorescnece lifetime microscopy
FPLC Fast protein liquid chromatography
FRET Fo¨rster resonance energy transfer
F Fluorescence signal
G4 G-quadruplex
GAM Genome architecture mapping
GFP Green fluorescent protein
GMEM Glasgow’s modified eagle medium
G Guanine
H-telo Human telomere
HaCaT Human epidermak keratinocyte cells
HEG Hexaethylene glycol
HEK293T Human embryonic kidney cells with large T antigen expression
hESC Human embryonic stem cells
Hi-C High resolution chromosome conformation capture
HILO Highly inclined laminated optical sheet
vii
HIV-1 Human immunodeficiency virus 1
HMBC Heteronuclear multiple-bond correlation spectroscopy
hnRNP A1 Heterogeneous nuclear ribonucleoprotein A1
hnRNP K Heterogeneous nuclear ribonucleoprotein K
HOAt 1-Hydroxy-7-azabenzotriazole
HPLC High-performance liquid chromatography
HRMS High-resolution mass spectra
HRP Horseradish peroxidase
HSQC Heteronuclear single-quantum correlation spectroscopy
H Ramdomised A or T base, in oligo context
IF Immunofluorescence
IGS In situ genome sequencing
IPTG Isopropyl β-D-1-thiogalactopyranoside
iPyPDS Isomeric pyrrolidine pyridostatin
ISC Intersystem crossing
ITC Isothermal titration calorimetry
Kd Dissociation constant
k Rate constant
LAD Lamina-associated domain
LCMS Liquid chromatography mass spectrometry
LHS Left hand side
LIF Mouse cytokine leukemia factor
MAZ MYC associated zinc finger protein
MEA Mercaptoehylamine
mESC Mouse embryonic stem cells
viii
Me Methyl
mRNA Messenger ribonucleic acid
MS Mass spectrometry
mut Mutant
MW Microwave
NCS N -Chlorosuccinimide
NHEK Normal human epidermal keratinocytes
NHS N -Hydrosuccinimide
NM23-H2 Non-metastatic cells 2, helicase expressed in humans
NMR Nuclear magnetic resonance
NS Not significant or non-specific
NTHL1 Nth like DNA glycosylase 1
N A, C, G or T (randomised nucleobase)
PBS Phosphate-buffered saline
PCR Polymerase chain reaction
PDS Pyridostatin
PD Dibromopyridazinedione reagent
PEG Polyethylene glycol
Pen/strep Penicillin and streptomycin mixture
Ph Phenyl
PIC Protease inhibitor cocktail
PIPES 2,2-(Piperazine-1,4-diyl)di(ethane-1-sulfonic acid)
PML Promyelocytic leukemia
PM Phase mask
ppm Parts per million
ix
PriPur Primary purification
PyPDS Pyrrolidine pyridostatin
P Passage
QY Quantum yield
Rf Retention factor
rG4 RNA G-quadruplex
RHS Right hand side
RIC RNA in situ conformation
RIF1 Replication timing regulatory factor 1
RNA Ribonucleic acid
rpm Rounds per minute
RT Room temperature
S0 Singlet state
SBR Signal-to-background ratio
scFV Single-chain variable fragment (antibody)
sDIBO Dibenzocyclooctyne
SDS Sodium dodecyl sulphate
sd Standard deviation
SecAB Secondary antibody
SecPur Secondary purification
seq Sequencing
SiR Silicon rhodamine
sm Single-molecule
SNP Single-nucleotide polymorphism
SOC Super optimal broth with catabolite repression
x
SPRITE Split-tool recognition of interactions by tag extension
ss Single-stranded
STORM Stochastic optical reconstruction microscopy
t1/2 Half-life
T1 Triplet state
tr Retention time
TAD Topologically associated domain
TAMRA 6-Carboxytetramethylrhodamine
tBu Tert butyl
TB Terrific broth
TCEP Tris(2-carboxyethyl)phosphine
TERT Telomerase reverse transcriptase
TES 2-[1,3-Dihydroxy-2-(hydroxymethyl)propan-2-yl]aminoethane-1-sulfonic acid
TET Ten-eleven translocation methylcytosine dioxygenase
TFA Trifluoroacetic acid
TIRFM Total internal reflection fluorescence microscopy
TIR Total internal reflection
TLC Thin layer chromatography
TRF2 Telomeric repeat-binding factor 2
Tris Tris(hydroxymethyl)aminomethane
TSS Transcription start site
T Temperature or thymine
U2OS Human bone osteosarcoma epithelial cells
UTR Untranslated region
UV Ultraviolet
xi
U Uracil
v/v Volume ratio
VEGF Vascular Endothelial Growth Factor
vis visible
w/v Weight by volume
WRN Werner syndrome RecQ like helicase
XPB Xeroderma pigmentosum type B helicase
XPD Xeroderma pigmentosum type D helicase
YT Yeast extract tryptone
YY1 Yin Yang 1 protein
xii
Contents
Declaration i
Abstract ii
Acknowledgements iii
Abbreviations xii
Contents xiii
1 Introduction 1
1.1 G-quadruplexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 History and structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.2 G4s in the human genome . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 G4s can be stabilised by small molecules . . . . . . . . . . . . . . . . . . 5
1.1.4 G4s biological implications . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.5 G4s link to cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.1.6 G4s in chromatin context . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.1.7 Evolution of G4 visualisation . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2 Single-molecule fluorescence imaging . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.1 TIRFM, light-sheet and HILO imaging techniques . . . . . . . . . . . . . 18
1.2.2 STORM super-resolution microscopy in 3D . . . . . . . . . . . . . . . . . 18
1.3 3D Chromatin structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.3.1 3C, TADs and compartments . . . . . . . . . . . . . . . . . . . . . . . . 22
1.3.2 Phase separation, nuclear substructures and new methods . . . . . . . . 23
xiii
1.3.3 3D genome structure simulation by single-cell Hi-C . . . . . . . . . . . . 24
2 Single-molecule G4 visualisation in live cells 27
2.1 Research aims and rationale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2 Fluorophore probe design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.1 G4 ligands: PDS and PhenDC3 . . . . . . . . . . . . . . . . . . . . . . . 29
2.2.2 Silicon Rhodamine Dye - the reporter moiety . . . . . . . . . . . . . . . . 32
2.3 Probe preparation and biophysics . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3.1 PhenDC3-SiR synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.3.2 SiR-PyPDS and SiR-iPyPDS preparation . . . . . . . . . . . . . . . . . . 36
2.3.3 FRET melting experiments . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.3.4 Fluorophore properties characterisation . . . . . . . . . . . . . . . . . . . 40
2.3.5 Fluorescence light-up experiments . . . . . . . . . . . . . . . . . . . . . . 44
2.3.6 FRET cascade indicates direct G4 ligand binding to a G4 oligo . . . . . 45
2.3.7 G4 induction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.8 G4 unfolding kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.4 TIRFM in vitro single-molecule imaging . . . . . . . . . . . . . . . . . . . . . . 55
2.4.1 In vitro G4 binding validation . . . . . . . . . . . . . . . . . . . . . . . . 56
2.4.2 A case for single-molecule, single-G4 binding . . . . . . . . . . . . . . . . 60
2.4.3 Measurement of G4 abundance . . . . . . . . . . . . . . . . . . . . . . . 61
2.5 Single-molecule imaging in cells . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.5.1 SiR-PyPDS and SiR-iPyPDS in live cells . . . . . . . . . . . . . . . . . . 63
2.5.2 PhenDC3-SiR in live cells . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5.3 Single-molecule observation and binding event residency times . . . . . . 69
2.5.4 Pre-blocking of G4 binding sites control experiments . . . . . . . . . . . 72
2.5.5 G4 folding is dynamic in live cells . . . . . . . . . . . . . . . . . . . . . . 73
2.5.6 G4 cell cycle dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.5.7 Single-molecule imaging in fixed cells . . . . . . . . . . . . . . . . . . . . 75
2.5.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
xiv
3 3D STORM super-resolution imaging of G4s 81
3.1 Project rationale and research aims . . . . . . . . . . . . . . . . . . . . . . . . . 81
3.1.1 Project inspiration and idea . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.2 G4 probe development for 3D STORM imaging . . . . . . . . . . . . . . . . . . 86
3.2.1 Probe choice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.2.2 Direct BG4 labelling with AF647 . . . . . . . . . . . . . . . . . . . . . . 87
3.2.3 IgG-BG4 labelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.2.4 IgG-BG4 confocal imaging . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.2.5 BG4-HaloTag and E12-HaloTag bacterial expressions . . . . . . . . . . . 92
3.2.6 Development of E12-AF647 G4-selective probe . . . . . . . . . . . . . . . 92
3.3 Single-molecule imaging with E12-AF647 labelled mixture - controls and opti-
misation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.3.1 STORM imaging conditions . . . . . . . . . . . . . . . . . . . . . . . . . 100
3.3.2 Fixed mESC nuclei imaging with E12-AF647 . . . . . . . . . . . . . . . . 103
3.3.3 G4 labelling protocol compatibility with Hi-C . . . . . . . . . . . . . . . 108
3.4 Obtaining singly labelled E12-AF647 . . . . . . . . . . . . . . . . . . . . . . . . 109
3.4.1 FPLC purification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.4.2 ELISA to validate G4 binding of E12-AF647 . . . . . . . . . . . . . . . . 114
3.5 3D STORM imaging of G4s with E12-AF647 . . . . . . . . . . . . . . . . . . . . 115
3.5.1 G4 binding controls in mESC nuclei with purified E12-AF647 . . . . . . 116
3.5.2 3D image acquisition with drift correction . . . . . . . . . . . . . . . . . 117
3.5.3 Whole nucleus 3D imaging . . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.6 Issues with E12-AF647 imaging and acquisition . . . . . . . . . . . . . . . . . . 120
3.6.1 Autofluorescence issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.6.2 Probe degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
3.6.3 FPLC purification optimisation . . . . . . . . . . . . . . . . . . . . . . . 123
3.6.4 Probe inactivation after FPLC purification . . . . . . . . . . . . . . . . . 126
3.6.5 Thermal reactivation of E12-AF647 . . . . . . . . . . . . . . . . . . . . . 129
3.7 Two antibody layer G4 imaging with BG4 . . . . . . . . . . . . . . . . . . . . . 132
xv
3.7.1 Controls and optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . 133
3.7.2 3D STORM imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
3.8 G4 genomic CUT&Tag map overlap with single-cell Hi-C structures . . . . . . . 136
3.8.1 mESC differentiation states . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.8.2 G4 map changes during differentiation . . . . . . . . . . . . . . . . . . . 138
3.8.3 Overlap and 3D analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
3.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4 Discussion and outlook 145
4.1 G4 visualisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.2 G4s during differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.3 G4 dynamics and antibody labelling . . . . . . . . . . . . . . . . . . . . . . . . . 147
4.4 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5 Materials and methods 149
5.1 Biophysical experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.1.1 Oligonucleotides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.1.2 FRET melting experiments . . . . . . . . . . . . . . . . . . . . . . . . . 149
5.1.3 Fluorophore properties characterisation . . . . . . . . . . . . . . . . . . . 151
5.1.4 pH titration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
5.1.5 Fluorescence light-up titrations . . . . . . . . . . . . . . . . . . . . . . . 152
5.1.6 FRET cascade experiments . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.1.7 G4 induction measurements . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.1.8 G4 unfolding kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
5.1.9 Fluorescence quench controls . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.1.10 Heat refolding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.1.11 ELISA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
5.2 Cell tissue culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.2.1 U2OS cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
5.2.2 mES cells 2i/LIF conditions . . . . . . . . . . . . . . . . . . . . . . . . . 156
xvi
5.3 Cell labelling protocols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5.3.1 Live U2OS cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
5.3.2 Fixation of U2OS cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
5.3.3 Fixation of mES cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
5.4 Single-molecule imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
5.4.1 smFRET imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
5.4.2 In vitro TIRFM single-molecule imaging . . . . . . . . . . . . . . . . . . 162
5.4.3 Live cell single-molecule imaging . . . . . . . . . . . . . . . . . . . . . . 163
5.4.4 Extracted mESC nuclei single-molecule imaging . . . . . . . . . . . . . . 164
5.4.5 3D STORM imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
5.5 Image data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.5.1 Quantification of binding events . . . . . . . . . . . . . . . . . . . . . . . 165
5.5.2 Residency time determination . . . . . . . . . . . . . . . . . . . . . . . . 166
5.6 E. coli antibody expression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
5.6.1 BG4 scFv antibody expression by autoinduction and purification . . . . . 166
5.6.2 Production and purification of E12 nanobody in BL21(DE3) from the
pHEN2 plasmid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
5.7 General synthetic experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.7.1 Mass spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
5.7.2 A¨KTA FPLC chromatography . . . . . . . . . . . . . . . . . . . . . . . . 172
5.7.3 Spin column purifications . . . . . . . . . . . . . . . . . . . . . . . . . . 173
5.8 Experimental synthetic details . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
5.8.1 Antibody labelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
5.8.2 PhenDC3-SiR synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Bibliography 188
Appendix 223
xvii
Chapter 1
Introduction
1.1 G-quadruplexes
1.1.1 History and structure
The widely recognised DNA double helix structure [1–3], solved in 1953, has been crucial for
studying and explaining DNA function. Composed of two complementary strands joined by
Watson and Crick base pairing, this structure was immediately suggestive of a genetic material
copying mechanism known as replication, what was later on elegantly confirmed to be semi-
conservative in Meselson and Stahl experiment [4]. Even though the DNA double helix structure
is often depicted as highly stable, its 3D structure is dynamic and can fold into non-canonical
states, shaped by proteins such as histones and helicases [5, 6]. The structural dynamics of
DNA are imperative for biological processes in cells such as gene regulation, transcription,
recombination and replication [7]. Such processes have been linked to and could potentially
be regulated by an alternative DNA secondary structure called G-quadruplex (G4) which has
actively been studied in the last few decades [8, 9].
The history of G4s began with an observation of self-association of guanylic acid forming
polycrystalline gels, first reported in 1910 by I. Bang [10]. Only half a century later, with
the help of X-ray crystallography, the phenomenon was explained to be caused by a guanosine
tetrad assembly [11] joined by Hoogsteen base pairing [12] (Figure: 1.1, a). In 1988, in vitro
studies suggested that these tetrads in DNA can stack on top of each other, forming a G4, by
stabilisation of pi-pi interactions. It what was the first indirect observation of a G4 [13], while
the first G4 X-ray crystal structure was produced in 1992 [14]. This sparked the search for new
discoveries and especially for the biological relevance of these structures.
1
Chapter 1: Introduction
G4 self-assembly is templated by monovalent cations which are centrally located between two
tetrads while occupying the central cavity between eight O6 of the guanines as observed in
X-ray crystal structures (Figure: 1.1, b) [14–18]. Cations were found to provide G4 stability
in the order of preference: K+ > Rb+ > Na+ > Cs+, NH+4 > Li
+ [19, 20] and G4s have
been demonstrated to be stable under physiological K+ concentrations [21–25]. G4s can form
a variety of topologies such as unimolecular, bimolecular and tetramolecular complexes with
strands in parallel or anti-parallel orientation (Figure: 1.1, c) [26, 27]. The topology also
influences the thermodynamic stability of G4s [28] and may implicate unique recognition of
individual or groups of G4s by proteins or potentially even small molecules [29].
a
c
b
Figure 1.1: G4 structures. (a) Guanine tetrad joined by Hoogsteen hydrogen bonds and stabilised
by coordinating to a mono-cation in the centre (M+). (b) Top and side view of a X-ray crystal struc-
ture of a parallel G4 from human telomere sequence comprising of three tetrads (PDB: 1KF1)[17].
(c) Schematic representations of selected examples of unimolecular and tetramolecular G4 topologies.
Adapted from D. Varshney, J. Spiegel, K. Zyner, D. Tannahill, and S. Balasubramanian, “The reg-
ulation and functions of DNA and RNA G-quadruplexes,” Nat. Rev. Mol. Cell Biol., pp. 1–16, Apr.
2020 [9].
2
Chapter 1: Introduction
G4s have been found to form in vitro from DNA, RNA [30] and DNA/RNA hybrids [31, 32].
However, a major research interest is in their implications in biology, therefore approaches to
map and observe G4s in cells were sought.
Bioinformatic methods have been used for predicting specific positions in genomes which have
potential to form G4s [33, 34]. The studies have been based on biophysical data on G4 formation
from techniques such as circular dichroism spectroscopy [35, 36], NMR spectroscopy [37, 38],
X-ray crystallography [14–18] and ultraviolet (UV) thermal melting analysis [39]. An initial
folding rule was developed in 2005 stating that ’a sequence of the form:
G3+N1−7G3+N1−7G3+N1−7G3+
will fold into a G4 under near-physiological conditions’ where N denotes any nucleotide base
in loops which are joining tetrad stacks [33]. Even though the initial rule accounts only for
loop lengths of up to 7 nucleotides, loops have been observed in vitro to be up to 30 bases in
length [40]. Moreover, G4s containing bulges, which connect two guanines of the same column
of the G-tetrad core [41], are missed by the predictive algorithm. Further developments to the
computational tools have improved the inclusion of G4s with a wider range of features [42–45]
and also use machine learning methods for G4 prediction in the genome [46].
Bioinformatic approaches have indicated that G4s are not randomly distributed across the
human genome, but enriched in functionally important regions such as gene promoters [44, 47],
transcription start sites [48], recombination sites [49, 50], replication origins [51] and telomeres
[33], suggesting potential G4 roles in biology.
1.1.2 G4s in the human genome
350 000 potential G4 forming sequences have been predicted in the human genome on the basis
of the folding rule [33, 34]. In contrast, over 700 000 structures have been detected in vitro
by a sequencing method to map G4s named G4-seq [52]. This method is based on polymerase
stalling at G4s during Illumina next-generation sequencing [53] because these structures can
act as a steric block for polymerase progression in DNA chain synthesis [54]. As a result,
sequencing quality at G4 positions drops significantly. Carrying out two separate sequencing
runs, under G4 promoting and disfavouring conditions, and comparing the sequencing quality
of the two runs, G4 positions in the genome could be mapped. The sequencing approach high-
lights where G4 structures have the capability of forming to act as a reference map. Disparity
between computational and G4-seq approaches arose because the prediction model does not
3
Chapter 1: Introduction
systematically include structures containing long loops [55] or bulges [41]. Whilst these struc-
tures can be detected by G4-seq, because they can also stall polymerase progression by steric
blocking. Moreover, a similar method called rG4-seq was developed, which exploits reverse
transcriptase stalling for mapping G4s in the human transcriptome in vitro [56]. Both G4- and
rG4-seq approaches account for G4 formation under stabilising conditions on isolated nucleic
acids, taken away from their biological context. Therefore, these sequencing methods provide
a reference map of all possible sequences capable of potentially forming G4s in the genome.
G4-seq has been used to map G4s in 12 different species, finding that G4s are evolutionary con-
served and enriched at promoters and 5’ untranslated regions (UTRs) in three of the species
mapped: human, mouse and Trypanosoma, but were distributed randomly or depleted in oth-
ers, including yeast, bacteria, C. elegans, Zebrafish and Drosophila [57]. This may suggest the
evolutionary importance of G4s emerged primarily in some of the more complex organisms,
including humans.
The first G4 observation in cells was in Stylonychia lemnae ciliates within telomeric repeats
using a single-chain antibody fragment (scFv) probe specific to G4s [58]. Another structure-
specific antibody - BG4, also scFv, has been developed and applied for visualisation of DNA
G4s in fixed human U2OS cells by immunofluorescence (IF) imaging (Figure: 1.2) [59, 60].
BG4 was instrumental in establishing chromatin immunoprecipitation with sequencing (ChIP-
seq) for mapping of G4s in the human genome within chromatin context [61, 62]. The method
is carried out as follows: cell are fixed with formaldehyde, then their chromatin is extracted,
fragmented and BG4 is used to enrich and purify for G4 containing chromatin fragments which
are then sequenced by Illumina sequencing to determine G4 genomic locations enriched by BG4
[61].
The method indicated 10 000 G4 structures in immortalised human epidermal keratinocyte
HaCaT cells, predominantly (98 %) in open chromatin regions as shown by complementary
sequencing methods ATAC-seq (assay for transposase-accessible chromatin with sequencing)
and FAIRE-seq (formaldehyde-assisted isolation of regulatory elements with sequencing) [63,
64]. Furthermore, the study also demonstrated that chromatin accessibility by itself was not
sufficient for BG4 binding and consequently G4 site calling. 87 % of the peaks overlapped
with sequences detected by G4-seq [52], however representing only 1 % of all sequences. It
signified the importance of chromatin context when considering G4 formation in cells [65].
Other reported G4-ChIP-seq maps reported peak numbers roughly in a range of 1 000 - 30
000 [61, 62, 66–71] suggesting different G4 landscapes in different cell lines. G4-ChIP-seq maps
showed G4s to be enriched in promoters, 5’UTRs, enhancers and architectural protein binding
sites [61, 71, 72].
4
Chapter 1: Introduction
Figure 1.2: LHS - Immunofluorescence showing BG4 foci (red) in U2OS osteosarcoma cell nuclei.
RHS - Increase in BG4 foci number in U2OS cells after treatment with the G4 binding ligand pyri-
dostatin (PDS, 1). Nuclei were counter stained with DAPI (blue). Scale bar, 20 µm. Adapted from
G. Biffi, D. Tannahill, J. McCafferty, and S. Balasubramanian, “Quantitative visualization of DNA
G-quadruplex structures in human cells,” Nat. Chem., vol. 5, pp. 182–186, Mar. 2013 [59].
More recent developments in sequencing technologies has led to an emergence of cleavage under
targets and tagmentation (CUT&Tag) sequencing technique, where specific antibodies by teth-
ering protein A-Tn5 transposase fusion enzyme to loci in the genome, can fragment DNA and
integrate sequencing adapters in the regions of antibody binding [73, 74]. CUT&Tag had also
been integrated for sequencing G4s with BG4, providing better signal-to-noise ratio and having
much lower input requirements to G4-ChIP-seq [75–77]. G4-CUT&Tag has allowed sequencing
of G4s at a single-cell level, highlighting differences in G4 landscapes in distinct cell popula-
tions [77]. Sequencing of 5 different human cancer cell lines, demonstrated large heterogeneity
between them having 3 000 common G4 peaks, but 1 000 - 16 000 unique peaks [76]. G4s
were also found to be associated with DNA-RNA hybrids (R-loops) formed on the opposing
C-strand to the G4. 71-81 % of R-loop peaks, also mapped by CUT&Tag, overlapped with
G4 peaks, what could suggest a synergistic influence of these two structures with respect to
genome function [75, 76]. Furthermore, overlap with active enhancers was found to be at 46 %
in HEK293T cells and at 13 % in mouse embryonic stem cells (mESC), further associating G4s
to transcription [75, 76].
1.1.3 G4s can be stabilised by small molecules
One of the principal tools in the study of G4s are G4-selective small molecules. These G4 ligands
can stabilise G4 structure upon binding and they are a means of perturbing G4 environment
for experiments in vitro and in cells [78]. Selected examples of these small molecules are shown
5
Chapter 1: Introduction
in scheme 1.1: pyridostatin (PDS, 1) [79], PhenDC3 (2) [80], Quarfloxin (3) [81], telomestatin
(4) [82], CX-5461 (5) [83], TMPyP4 (6) [84] and Braco-19 (7) [85].
N
H
N
H
HN
N
O
N
O
N
H
N
N
N
H
N
N
NN
N
TMPyP4
N
O
O
N
H
O
NF
N
N
N
Quarfloxin
N SN
O
N
H
O
N
N
N
N
CX-5461
N
O
H2N
O
HN
O
NH
N N
O
NH2
O
H2N
Braco-19
Pyridostatin
N
O
N
O
O
N
O N
N O
O
N
N
S
H
Telomestatin
N N
NH
O
HN
O
NN
I I
PhenDC3
1
2
3
4
5
6
7
Scheme 1.1: G4 ligand examples.
6
Chapter 1: Introduction
Generally, G4 ligands are planar aromatic molecules, capable of pi− pi stacking on the guanine
tetrads as observed by multidimensional NMR and X-ray crystal structures (Figure: 1.3).
Notably ligands can occupy both G-tetrad faces and can demonstrate preference for 3’ or
5’ tetrad face of a G4 [86]. Simultaneous dual binding of G4s is possible too, as has been
demonstrated by BG4 antibody and PDS co-binding (1) [87], while mass spectrometry (MS)
studies showed 1 or 2 ligands binding per G4 with possible conformational changes, depending
on G4 topology and ligand used [88]. Dual binding has been utilised in profiling G4 binding
proteins by PDS (1) derivative. Upon co-binding the same G4 target as the protein, the G4
ligand could chemically tag the protein for pull-down [89]. Furthermore, the ligands tend to be
positively charged for having a strong favourable electrostatic interaction with the negatively
charged phosphate backbone of DNA. However, such binding modes and relatively common
structural features of G4s, compared, for example, to proteins, makes it seem unlikely that
small molecules could achieve unique recognition of a single G4 [78, 90].
Figure 1.3: LHS - NMR structure of TMPyP4 (6) bound to G4, RHS - crystal structure of Braco-
19 (7) co-crystallised with G4. Adapted from S. M. Haider, S. Neidle, and G. N. Parkinson, “A
structural analysis of G-quadruplex/ligand interactions,” Biochimie, vol. 93, pp. 1239–1251, Aug.
2011 [78], original data from: [17, 91].
It is vital for G4 ligands to have selectivity for G4s over double-stranded (ds) and single-stranded
(ss) DNA, the primary competitors inside a nucleus. This feature is crucial for being able to
recognise specifically folded G4 structure rather than the G-richness of the DNA sequence where
G4s are capable of forming. For small molecules, a range of 10 - 500 fold selectivity of G4s
against other forms of DNA has been reported [67, 86, 92–95].
7
Chapter 1: Introduction
1.1.4 G4s biological implications
Helicases
G4s have been associated with the three main processes that make up the biological central
dogma, namely - replication, transcription and translation (Figure: 1.4). All of which include
single-stranded nucleic acid formation in their mechanisms, a general condition that favours for-
mation of G4s due to a higher flexibility of the strand when compared to interaction constraints
in the double helix [96]. Helicases are proteins that are able to unwind the DNA double-helix
[97], therefore, in principle, they might be considered as facilitators of G4 formation. On the
other hand, a variety of helicases have been found with potent G4 unwinding ability and are
important in maintaining genome stability [98, 99]. For example, XPB and XPD helicases have
important roles in nucleotide excision repair and mutations in their genes lead to rare autosomal
recessive diseases such as xeroderma pigmentosum, Cockayne syndrome and trichothiodystro-
phy [100, 101]. XPB and XPD helicases targets in the genome have been mapped by ChIP
approach and it was found that 40 % of the sequences bound have potential to form a G4
[102]. Moreover, WRN and BLM helicases, associated with Werner and Bloom syndromes,
unfold G4s and regulate gene expression levels in the cell [103–105]. Pif1 helicases were shown
to be involved in suppression of genome instability at G4 sites [99]. DHX36 helicase, a strong,
nanomolar affinity G4 binder has been co-crystallised with G4 with MYC promoter region G4
structure (Figure: 1.5) [106]. A small artificial protein G4P was designed according to DHX36
- G4 interaction and was used to develop a ChIP-seq technique for G4 sequencing, alternative
to ones using BG4 [107]. Lastly, it has been reported that G4 stabilisation with PDS G4 ligand
causes transcription- and replication-dependent DNA damage [54]. Overall, these observations
indicate that G4s require tight regulation by helicases, suggesting a biological consequence of
these structures. In depth analysis of G4 function in biology can be found in comprehensive
reviews [8, 9, 108–112], while a brief overview is presented in this work.
Transcription
G4 formation in the promoter region of genes has been found to be associated with gene ex-
pression in cells [29, 108, 110]. The majority of early studies reported decreased transcriptional
activity upon treatment by small molecule G4 ligands, with notable examples of oncogenes
such as MYC [29, 113], KRAS [114, 115], BRCA1 [116], HRAS [117] and KIT [118]. A more
recent genome wide study reported similar observations for many genes [119]. This had left
a view that G4s are roadblocks to transcription by inhibiting transcriptional machinery [108].
However, it is important to distinguish that a G4 and a G4 bound with a ligand do not nec-
essarily have the same effect on transcription or any other biological process for that matter.
Contrastingly to the studies using small molecule G4 ligands, developments of genome wide G4
8
Chapter 1: Introduction
Figure 1.4: Possible G4 locations in cells and their biological implications on transcription (A),
telomere maintenance (B), replication (C) and translation (D). Modified and adapted from D. Rhodes
and H. J. Lipps, “G-quadruplexes and their regulatory roles in biology,” Nucleic Acids Res., vol. 43,
pp. 8627–8637, Oct. 2015 [108].
sequencing approaches by G4-ChIP-seq and CUT&Tag, have provided a different view, where
G4s have been associated with higher transcriptional levels. A first study to report this was
by Ha¨nsel-Hertsch and co-workers [61], where mapping of G4s by G4-ChIP-seq demonstrated
that G4s residing in promoter regions of genes, are associated with elevated transcriptional
activity i.e. genes with promoters containing a BG4 binding site had on average higher tran-
scriptional output than genes that did not. A potential way G4s could facilitate transcription is
by acting as transcription factor binding hubs as was demonstrated by Spiegel and co-workers,
consequently G4 ligands can block these sites to hinder transcription [69, 76]. Other studies
also report higher transcriptional levels in genes containing G4s in their promoters as well as
overlap with transcription factor binding sites [70–72]. In greater detail, a link of G4s to active
and bivalent promoters has been established, with bivalent promoters more likely to become
transcriptionally active upon cellular differentiation, they also demonstrated less transcriptional
noise than G4-free genes [71]. Mechanistic investigation has demonstrated that G4 formation at
promoters precedes transcription initiation [68]. Moreover, G4s have been associated to influ-
ence RNA polymerase II residency in genes [68, 76]. These findings together with observations
that majority of G4s resided in hypomethylated CpG islands [66] (which also strongly overlap
with promoters), generally representative of active chromatin [120], suggest a complex array of
G4-related processes in transcriptional control.
G4 stabilisation by ligands has been linked to elevated relaxin expression [121]. Moreover,
studies in Escherichia coli (E. coli) by G4 motif insertions in different locations around tran-
scription start site have shown that the effect on gene expression was dependent on the G4
position with both increase and decrease of transcription being observed upon G4 formation
[122]. Another study, in E. coli showed that inserted G4 formation on the antisense strand
reduced transcription while G4 formation on the sense strand generally showed no effect [123].
9
Chapter 1: Introduction
Figure 1.5: Co-crystal structure of DHX36 helicase bound to G4 within the MYC promoter region.
Adapted from M. C. Chen, R. Tippana, N. A. Demeshkina, P. Murat, S. Balasubramanian, S. Myong,
and A. R. Ferre´-D’Amare´, “Structural basis of G-quadruplex unfolding by the DEAH/RHA helicase
DHX36,” Nature, p. 1, June 2018 [106].
The oxidation of guanine to 8-oxo-guanine in a G4 forming region has been linked to increased
transcriptional levels of VEGF and NTHL1 genes [124, 125]. A later investigation on the
mechanism suggested that oxidative damage recruits base excision repair machinery to remove
8-oxo-G and the abasic site formed promotes G4 folding which in turn facilitates transcription
factor binding [126]. Another study has shown that oxidation of guanines on long loops of G4
in KRAS promoter region led to enhanced MAZ and A1 protein ligation to G4 which then
recruit RNA polymerase II [127]. These examples suggest that 8-oxo-G, when occurring in
sequences capable of G4 formation, may constitute an epigenetic marker for sensing oxidative
damage and induce changes in transcription [125].
Replication
Another vital process linked to formation of G4s is replication [128–130]. G4s have been found
to promote replication fork stalling [131–133] and this is supported by the presence of helicases,
e.g. FANCJ, WRN, BLM and DDX11, which actively unwind G4s to avoid DNA polymerase
stalling [103, 130, 134, 135]. Another study investigated how local epigenetic changes induced
by a G4 ligand during replication of DT40 chicken cells have led to heritable transcriptional
reprogramming of the BU-1 locus [136, 137].
Moreover, bioinformatic approaches have determined that about 80 % of replication origins
overlap with predicted G4 sequences [51, 138], these studies were supported by an observation
that the origin recognition complex preferentially binds to G4s rather than single-stranded
DNA [139]. Indeed, it was found that cell treatment with PhenDC3 (2) creates new replication
10
Chapter 1: Introduction
origins, and therefore G4s could be important in replication initiation [140].
Rif1 protein which is responsible for regulating replication timing, has been found to bind to
G4s and exert long-range suppression of replication origin firing [141], further suggesting that
these structures may act as important regulators of genetic information copying process. It was
suggested that the long-range effect of Rif1 could be imposed by DNA looping through Rif1
binding to multiple G4 structures [142].
Telomeres
Telomeres are functional chromosome ending regions with a repeating TTAGGG motif [143],
conserved among vertebrates. Due to their G-rich pattern, they have high potential for forming
G4 structures, as predicted bioinformatically by the folding rule [33] or indicated by obtained G4
crystal structures [14, 17]. Telomeric DNA can adopt inter- and intra-molecular G4 structures
with either parallel or anti-parallel topology [26]. Hybrid and anti-parallel G4 structures in
transfected oligos were observed in live cells by 19F NMR [144]. G4s at telomeres also have
been imaged in cells by immunofluorescence methods using G4 specific antibodies [58, 59]. To
further support the telomere and G4 connection, studies have shown that they can be folded
and unfolded by telomeric proteins [145–147]. Moreover, in vitro experiments have displayed
the G4 inhibition capacity of telomerase extension [145, 148, 149] which could be indicative of
their role as chromosome end capping structures [150, 151]. However, more recent studies have
reported telomerase ability to resolve and extend parallel G4 structures at the telomeres [152].
Recombination, translation and disease
Studies of human pathogen Neisseria gonorrhoeae have found that G4s were necessary for
recombination of pilin proteins [50], the mechanism required for antigenic variation of the
pathogen to avoid hosts immune system. Moreover, BLM helicase has been found to suppress
recombination at G4 motifs by their unwinding [153], while RNA G4s in HIV-1 virus were
suggested to promote recombination [154].
RNA can also fold into G4 structures and their formation has been shown to inhibit translation
[155, 156]. This was shown for RNA G4s in the 5’ untranslated region in the gene transcript
of the human NRAS and BCL-2 proto-oncogenes [157, 158]. Moreover, RNA G4s have been
found to regulate translation in ribosomal protein mRNA [159]. CNBP (Cellular nucleic acid
binding protein) has been reported to bind G-rich regions to prevent G4 formation and there-
fore facilitating translation [160] while DHX36 helicase was found to unwound RNA G4s in
translationally inactive mRNAs reducing their accumulation [161]. Other studies reported that
DHX36, DHX9 and eIF4A helicases regulate translation by unwinding G4 structures in mRNA
[162].
11
Chapter 1: Introduction
Another example is a study on Epstein-Barr virus where G4s were found to regulate antigen
1 mRNA translation which may act as a potential mechanism to coordinate an attack on the
host [163]. G4s have also been found to regulate promoter activity in HIV-1 virus [164, 165],
while G4 ligands showed anti-viral activity [166]. Consequently, G4 stabilising ligands have
been considered as an antiviral therapy method [167, 168].
G4s have been associated with a number of diseases, including neurodegenerative ones and
cancer [169], as an extension that G4s were found to bind human disease-related factors. An
array of neurodegenerative disorders were linked to G4s through ability to regulate transcription
or translation levels of genes involved in disease development mechanism [170, 171]. However,
the most notable G4 link to disease and potential for therapeutics is the one to cancer.
1.1.5 G4s link to cancer
G4s link to cancer has been investigated by multiple studies, which suggest general elevation of
G4 levels in cancerous cells. Whether this is a consequence of altered cancer biology or specific
role of G4s in the malignancy remains to be explored, however this link has drawn interest to
investigate G4s as potential targets for anti-cancer therapeutics.
A immunohistochemistry approach using the BG4 antibody1 has been used to visualise elevated
levels of G4s in human stomach and liver cancer tissues in comparison with their normal tissue
counterparts [60]. Moreover, immortalized human HaCaT cells [172] showed ∼4-fold more
G4 foci in G4-ChIP-seq analysis than corresponding normal human epidermal keratinocytes
(NHEK)2. Also, HaCaT cells were ∼7-fold more sensitive than NHEK to growth inhibition by
G4-selective ligand PDS (1) (Scheme: 1.1 compound 1) [61]. Further work in breast cancer
models have demonstrated that G4 DNA levels predict tumour sensitivity response to G4-
ligands and G4 formation was linked to cancer genome instability [67]. The general observation
of higher G4 abundance in cancer cells than in respective normal cells may be used to achieve
selectivity for tumours via G4 targetting therapeutics [29, 110, 111, 173, 174]. Furthermore,
G4s mark promoters of highly expressed cancer genes [61, 62].
As discussed in section 1.1.3, unique G4 recognition is unlikely with small molecule G4 ligands.
Nevertheless, G4 targeting as a group can still prove to be a therapeutically viable strategy.
1Thin fixed tissue samples were treated with BG4 which was then visualised by secondary-antibody and
peroxidase staining.
2HaCaT cell line has been derived from NHEK by spontaneous immortalisation. HaCaT cells are nontu-
mourigenic, but they are used as cancer mimics as they are more cancer-like than NHEK.
12
Chapter 1: Introduction
For example, synthetic lethality3 has been observed with G4 ligands: telomestatin (4) in com-
bination with WRN helicase inhibitor [175], PDS (1) [173, 176] and CX-5461 (5) [177] on cells
deficient in BRCA1/2 oncogenes. Moreover, a genome wide screen revealed synthetic lethality
for PDS (1) and PhenDC3 (2) against disease-related genetic vulnerabilities [178]. Moreover,
PDS (1) can cause DNA damage and demonstrated inhibition of cancer cell growth [54, 179].
Quarfloxin (3) and CX-5461 (5), G4-selective drugs against cancer have progressed to phase
II and I clinical trials respectively, however Quarfloxin (3) eventually failed as it did not reach
therapeutically relevant bioavailability [81, 83, 177]. Another anti-cancer approach of using G4
oligonucleotide AS1411 has reached phase II. The oligonucleotide with its G4 structure targets
nucleolin, a protein overexpressed on the surface of cancer cells, which sequentially can cause
cancer cell death [180, 181].
1.1.6 G4s in chromatin context
G4s have been found to have intricate relationships with the chromatin environment around
them. A question remains whether G4s are merely passengers influenced by this environment
or whether G4 can induce an influence themselves and possibly shape chromatin epigenetics.
The emergence of G4-ChIP-seq and then CUT&Tag techniques has revealed that G4s are pre-
dominantly detected by BG4 in open chromatin regions [61, 75, 76]. This link was established
further by chromatin compaction under hypoxia conditions which led to a decrease of observed
G4s in promoter regions [68]. Meanwhile chromatin relaxation with histone deacetylase in-
hibitor entinostat resulted in more G4s being observed [61, 70]. These findings suggest that G4
landscape is largely linked on chromatin context.
G4 formation at CpG islands has been found to influence DNA methylation at cytosines by
binding and inhibiting DNA methyltransferase 1 activity around the G4 in human K562 cells
[66]. G4 positive correlation with CpG islands and anti-correlation with their cytosine methy-
lation was also reported in mESC [75]. While DNA methylation has been reported to reduce
chromatin accessibility and consequently the transcriptional activity of those regions [182],
these studies suggest that G4s can influence CpG islands to retain their hypomethylated status
at the G4 site, keeping chromatin accessible and active.
G4s have also been found to be associated with histone epigenetics. Initial G4-ChIP-seq maps
showed G4 colocalisation with H3K4me34 (euchromatin and active promoters histone mark) and
no correlation to H3K9me3 (heterochromatin histone mark) and H3K27me3 (repressive histone
3A concept in which cell death is induced by a synergy of multiple gene deficiencies or drug treatments, while
their individual effects would not cause cell death.
4H3K4me3 name indicates trimethylation of lysine 4 on histone H3.
13
Chapter 1: Introduction
mark) [61]. G4s were predominantly associated with active promoters by multiple studies
[75, 76]. Also, a study in human embryonic stem cells showed that upon cell differentiation,
loss of G4s at promoters containing H3K4me3 led to increased levels of H3K27me3 repressive
histone mark while a gain of a G4 led to a decrease [71]. G4 loss also led to loss of chromatin
accessibility. These differentiation experiments established a firm link between G4s and histone
modifications. An example of G4 induced epigenetic reprogramming has been reported where
G4 stabilisation by G4 ligands during replication have led to a loss of H3K4me3 histone mark
and subsequent heritable loss of DNA cytosine methylation [136]. Furthermore, G4s were
found to mark both active and poised enhancers with high H3K4me1/H3K27ac ratios [75, 76]
and enhancer-promoter contacts containing G4s were found to be more likely which suggests
potential role of G4s in shaping higher order 3D chromatin structure [71, 72].
Long-range chromatin interactions could also be shaped by G4s through interaction with ar-
chitectural proteins. One such interaction has been reported for YY1, a zinc-finger protein
capable of binding to G4s and formation of DNA loops through dimerisation [183, 184]. The
study showed that the DNA looping is G4 dependent by removal of the G4 motif with CRISPR-
Cas9 gene editing. Another example indicated that DNA G4s contribute to architectural pro-
tein CTCF recruitment at CpG islands [185]. Bioinformatics analysis of overlap of BG4 and
architectural protein binding sites has also been reported [71, 72].
G4s detected by G4P coincided with negative supercoiling of DNA [107]. This result was in
agreement with earlier reports that negative superhelicity can facilitate G4 formation, however
is not sufficient to drive it [186–188].
G4 coincidence with alternative nucleic acid structures can also have its implications to the
chromatin environment. A G-rich DNA strand capable of forming a G4 will result in a C-rich
region on the complementary strand of the double-helix. When a G4 forms the C-rich strand
remains single-stranded and it has been suggested that it can either fold into a secondary DNA
structure called an i-motif [189] or can hybridise with RNA complementary sequence forming
an R-loop [190]. Multiple studies have reported overlap of G4 and R-loop ChIP-seq maps where
71-81 % of R-loop sites also contained a G4 [66, 75, 76, 191] and were shown to facilitate each
others formation [192]. Moreover, deficiencies in TET enzymes, which oxidise methylcytosine,
were associated with an increase of both G4 and R-loop levels [193]. Though TET might
not be directly involved with G4s or R-loops, the observation that the two were increased
together suggested that the formation of both structures might be linked. Contrastingly, in vitro
biophysical optical tweezers experiment showed that G4s and i-motif are mutually exclusive
[194].
14
Chapter 1: Introduction
1.1.7 Evolution of G4 visualisation
Direct G4 observation is important in establishing the presence of these structures in cells
and assessing their biological relevance. The first G4 observation in cells was in Stylonychia
lemnae ciliates within telomeric repeats using a scFv antibody specific to G4s [58]. The probe
uniformly stained the nucleus without distinct visualisation of foci. 11 years later, the first
reported example of G4 visualisation with a fluorescently labelled small molecule G4 ligand
PDS (1) in fixed cells [54]. However, micromolar concentrations of probe used to stain the
cells, primarily labelled nuclear bodies, with any foci in the nucleus being hard to visualise.
Moreover, the study lacked convincing controls to indicate that the probe was indeed labelling
G4s.
Quantitative visualisation of G4s as distinct foci was reported by using BG4 in a three antibody
layer IF in fixed cells (Figure: 1.2) [59]. This study provided more robust evidence of DNA
G4s existing in fixed human cells. Another G4 selective scFv antibody 1H6 has been used
for G4 imaging in fixed cells and demonstrated that knockout of FANCJ helicase led to an
increase of fluorescence signal coming from 1H6 [195]. Also, 1H6 demonstrated localisation
of G4s in herpes virus infected cells with most of the signal coming from virus replication
compartments at the internal layer of the cell membrane [196]. Interestingly, in contrast to
BG4, 1H6 primarily localised in heterochromatin, as indicated by immuno-electron microscopy
[197], however showed a correlation with RNA polymerase II localisation and an increase of
binding upon chromatin relaxation [198]. Moreover, a study found 1H6 to have cross-reactivity
to AT-rich genome regions which could outcompete G4 binding [199]. The limitation of using
antibodies for G4 visualisation is the requirement of cell fixation, which may introduce artefacts
and disrupt G4 folding dynamics. Antibodies having high binding affinities to G4s might also
induce G4 formation leading to a non-endogenous G4 visualisation.
Fluorescence lifetime microscopy (FLIM) has also been utilised for observation of G4 struc-
tures. This method relies on a property that G4 ligands have longer fluorescence lifetime when
bound to G4s rather than to other forms of DNA or RNA. First example of this was reported
with a DAOTA-M2 probe, which was also used in live cells [200]. G4 fluorescence lifetimes
measured with this probe in vitro were longer than any lifetimes detected in cells. Therefore,
the assumption about this method was attributing longest measured lifetimes to G4s, assum-
ing that observations made in vitro hold in the cell too. With the probe having good binding
affinity to both G4s and dsDNA, this can be an issue in true G4 detection. In a follow up
study, the amplitude of fluorescence lifetimes of probe binding to ss- and dsDNA was similar
to the amplitude of lifetimes detected in cells, again with the longest lifetimes being attributed
to G4s as an assumption, though other forms of DNA seemed to be able to explain lifetime
amplitudes observed in cells [201]. Another example of FLIM was with o-BMVC light-up G4
15
Chapter 1: Introduction
ligand, which was used to detect G4s to be 13 times more prevalent in head and neck cancer
cells than in corresponding normal cells taken from patient biopsies [202, 203]. The probe in-
terestingly strongly stains nuclear bodies, however longer fluorescence lifetimes, characteristic
of G4s were detected outside of them exclusively.
Other examples of G4 imaging have been reported with different G4 ligands: c-exNDI and
IMT in live and fixed cells detected foci in nuclear bodies and nucleus periphery [204, 205],
N-TASQ primarily labelled nuclear bodies and RNA G4s outside the nucleus [206, 207], while
QUMA-1, TPA-3 and thioflavin exclusively label RNA G4s outside the nucleus [208–210]. An
alternative method to fluorescence imaging for direct observation of G4s was live-cell 19F NMR
spectroscopy with transfected RNA G4 oligos enriched with 19F [211, 212].
However, all the mentioned small molecule imaging methods required relatively high micro-
molar concentrations of probe for nucleus staining with up to 24 h long treatments. Such
conditions are likely to influence the dynamic G4 landscape by stabilising these structures,
inducing their formation and thus making G4 detection non-endogenous. This issue is inves-
tigated and addressed in discussion of the work carried out for this thesis, a single-molecule
fluorescence imaging approach of G4s in live cells to ensure endogenous G4 detection without
perturbing global G4 dynamics [213].
1.2 Single-molecule fluorescence imaging
Single-molecule experiments have been powerful in providing mechanistic insights into impor-
tant processes in molecular biology [214, 215]. Force spectroscopy has used magnetic and optical
tweezers to study RNA polymerase [216, 217] and translating ribosomes [218] while observing
amino acid or nucleotide elongation one unit at a time. Single-molecule fluorescence spec-
troscopy methods have revealed bursting behaviour in transcription and translation i.e. chain
elongation occurs in short pulses proposed to be caused by the fluctuating stochastic nature of
the biochemical reactions. These observations were made by following the synthesis of single
molecules of yellow fluorescent protein [219] and by tracking individual mRNAs through the
use of selective M2S-GFP probes5 which bind to mRNA as it is being produced [220, 221].
Single-molecule Fo¨rster resonance energy transfer (smFRET) has also been instrumental in
studying protein and nucleic acid dynamics [222, 223]. The method adds two fluorescent tags
to a molecule of interest and measures FRET signal as a readout of the distance between the
two points of reference [224]. Moreover, this research has allowed the development of advanced
biotechnologies such as real-time single-molecule DNA sequencing [225] and the development
5The tagging approach is based on a M2S bacteriophage coat protein which forms a stable complex with
RNA. Green fluorescent protein (GFP) is attached to MS2 for sensitive fluorescence spectroscopy visualisation
of RNAs.
16
Chapter 1: Introduction
of nanopore sequencing which deciphers nucleotide bases in DNA by measuring the current as
the strand is being extruded through a pore [226].
G4s have been observed by atomic force microscopy in vitro at a single-molecule level [147, 227].
Furthermore, mechanisms of helicase unfolding of G4s have been studied in vitro by smFRET
[106, 228, 229]. However single-G4 imaging methods in cells, which would provide a valuable
new way of studying G4 dynamics in a cellular context and gain mechanistic insight of G4
function, have not yet been developed.
Single-molecule experiments are useful methods because they can detect rare events while
observing single event variability and the dynamics of the studied process. Moreover, such
experiments are capable of distinguishing multiple populations within a system, a feature that
can be missed by averaging based methods. Studies of the repetitive cycles of chain synthesis
in transcription, replication and translation have particularly benefited by such absence of
averaging [215]. In addition, single-molecule imaging can provide unique biological insights by
monitoring the distribution of a probe in individual cells and assess localisation variability in
organelles. All in all, single-molecule studies can provide greater depth of understanding of
biological processes by ability to recognise subsets of individual molecule behaviour rather than
population averaged measurement methods.
The sensitivity of fluorescence spectroscopy and the use of high-speed cameras has allowed
recording of videos of single-molecules in real time [230, 231]. Multiple wavelength detector
channels can be used for observing different colour probes simultaneously in order to track their
co-localisation [232, 233]. An elegant example of this has been used in a study [234] where three
proteins: telomerase protein TERT, telomere binding protein TRF2 and Cajal body element
Coilin were co-imaged in real-time to observe their co-localisation (Figure: 1.6). The results
led to the authors proposing a model in which telomerase uses a search in 3D of telomeres by
forming many frequent reversible short interactions before having a long stable association.
Figure 1.6: Frames of simultaneous visualisation of fluorescently labelled telomeres (TRF2 in red),
telomerase (TERT in green), after imaging Cajal bodies (Coilin in blue, white arrows indicate co-
localisation). Adapted from J. C. Schmidt, A. J. Zaug, and T. R. Cech, “Live Cell Imaging Reveals
the Dynamics of Telomerase Recruitment to Telomeres,” Cell, vol. 166, pp. 1188–1197.e9, Aug. 2016
[234].
17
Chapter 1: Introduction
1.2.1 TIRFM, light-sheet and HILO imaging techniques
The principal challenge of single-molecule imaging is to reduce background noise to such a
degree that even weak signals from individual molecules can be observed. One way of achieving
this is to irradiate a minimal volume of the sample to minimise background arising from out-
of-focus imaging planes. An example of this is highly inclined laminated optical (HILO) sheet
microscopy technique [235], which illuminates a coverslip at a high angle resulting in a large
refraction of light at the surface. This produces a thin optical layer through the specimen
(Figure: 1.7, a). The thickness of the layer denoted dz can generally be 6 µm. The result is that
only a narrow region around the object plane is irradiated, thus fluorophores outside the plane
are not excited and do not contribute to the background noise, which also significantly reduces
phototoxicity induced to a cell and allows longer imaging experiments. Another well established
technique is light sheet microscopy [236], where a narrow light sheet is generated by advanced
optics and then is directed through the specimen (Figure: 1.7, b). Though conceptually similar
to HILO, light sheet microscopy can generate sheets as thin as 1 µm [237]. Total internal
reflection fluorescence microscopy (TIRFM) is another commonly used method for limiting
irradiation volume of the sample. It utilises the evanescent waves formed from the light reflected
at the coverslip, reaching about 200 nm into the sample (Figure: 1.7, c). This generates good
signal-to-background ratio for single-molecule imaging, however TIRFM is limited in that it
can only image the surface of the specimen.
1.2.2 STORM super-resolution microscopy in 3D
Super-resolution microscopy is a type of optical microscopy which surpasses the light diffraction
limit, which as defined by Ernst Abbe is d = λ
2NA
, where d is the resolvable feature size, λ is
the wavelength of light and NA is the numerical aperture of the imaging system. For visible
light, the limit is in the order of ∼200 nm. One of the methods capable of surpassing the light
diffraction limit is stochastic optical reconstruction microscopy (STORM), a type of super-
resolution localisation microscopy where an image is reconstructed by repeat observation of a
photoswitching probe at a single-molecule level [240]. STORM imaging constitutes a series of
localisation cycles, during which only a fraction of the fluorophores are switched on, allowing
the resolution of single molecules. Repeat of localisation cycles constructs the overall image
with improved accuracy with lateral resolution in the order of ∼20 nm.
For constructing a 3D image, foci axial positioning determination is required. Most commonly
used techniques for this are astigmatism [241] and biplane [242], however they are limited to a
depth of field of ∼500 nm, which makes imaging of large structures like a cell nucleus (∼10 µm
diameter) very time consuming [243]. The double helix point spread function (DHPSF) method
18
Chapter 1: Introduction
a
c
b
Figure 1.7: (a) Schematic illustration of HILO microscopy. The HILO laser beam path indicated
by solid blue arrows, Epi - indicates epifluorescence mode laser beam path, TIR - indicates total
internal reflection laser beam path, adapted from M. Tokunaga, N. Imamoto, and K. Sakata-Sogawa,
“Highly inclined thin illumination enables clear single-molecule imaging in cells,” Nat. Methods, vol. 5,
pp. 159–161, Feb. 2008 [235]. (b) Schematic illustration of light sheet microscopy technique, adapted
from A. Ponjavic, Y. Ye, E. Laue, S. F. Lee, and D. Klenerman, “Sensitive light-sheet microscopy
in multiwell plates using an AFM cantilever,” Biomed. Opt. Express, BOE, vol. 9, pp. 5863–5880,
Dec. 2018 [238]. (c) Schematic illustration of TIRFM technique, adapted from A.-Y. Guo, Y.-M.
Zhang, L. Wang, D. Bai, Y.-P. Xu, and W.-Q. Wu, “Single-Molecule Imaging in Living Plant Cells:
A Methodological Review,” Int. J. Mol. Sci., vol. 22, p. 5071, Jan. 2021 [239].
is a useful alternative [244] which can provide a wider depth of field (∼4 µm). In DHPSF a
single focus is split into two lobes by a double helix phase mask (PM) and axial positioning can
be determined by the angle between the two lobes (Figure: 1.8, a). The localisation precision
is dependent on the number of detected photons and can reach up to 10-20 nm resolution in
the lateral and axial positioning (Figure: 1.8, b). Whole cell 3D images can be constructed by
imaging with DHPSF at different plane slices (in each imaging the depth of field limit) and
then combining them together (Figure: 1.8, c).
19
Chapter 1: Introduction
a b
c
Figure 1.8: (a) Schematic representation of a DHPSF method. PM splits the imaged foci into two
lobes which are observed by an EMCCD detector (Electron Multiplying Charge-Coupled Device).
(b) Localisation precision improves up to a limit with increasing number of detected photons. (c)
Schematic representation of multi-slice 3D imaging of a cell. Z-axis field of view range of a cell is
imaged per slice at different heights, then the slices are combined to give a coherent whole cell 3D
image. Adapted from A. R. Carr, A. Ponjavic, S. Basu, J. McColl, A. M. Santos, S. Davis, E. D.
Laue, D. Klenerman, and S. F. Lee, “Three-Dimensional Super-Resolution in Eukaryotic Cells Using
the Double-Helix Point Spread Function,” Biophysical Journal, vol. 112, pp. 1444–1454, Apr. 2017
[243].
1.3 3D Chromatin structure
3D nuclear architecture is shaped by a hierarchy of chromatin folding elements varying in scale.
DNA double helix wraps around histones to form nucleosomes, as a first order of compaction.
Nucleosomes can then join into chromatin fibres and fold into chromosome domains. These
domains are then segregated at a larger scale into chromatin compartments and chromosome
territories (Figure: 1.9). At every level, the nucleus is also shaped by its substructures such
as nucleoli, nuclear lamina, nuclear speckles, Cajal and promyelocytic leukemia (PML) bodies.
Nucleus structuring is non-random, as shown by early imaging experiments that revealed that
20
Chapter 1: Introduction
chromosomes have preferred positions with respect to the nucleus centre or periphery as well
as one another [245–247]. Centromere clustering could also contribute to the overall shaping of
the 3D chromatin structure [248].
Figure 1.9: Schematic representation of 3D genome organisation at different scales. Adapted from
E. S. Dog˘an and C. Liu, “Three-dimensional chromatin packing and positioning of plant genomes,”
Nat. Plants, vol. 4, pp. 521–529, Aug. 2018 [249].
From a different perspective, we have known that the genome is segregated into densely packed
heterochromatin and more lightly packed euchromatin since the pioneering work of Emil Heitz
21
Chapter 1: Introduction
and the emergence of nucleus staining methods in the first half of the 20th century [250]. Distin-
guishing transcriptionally active and inactive chromatin by their density remains an important
concept in the field to this day [251, 252]. Active and inactive chromatin have specific posi-
tioning with respect to the 3D folding of chromosomes [253, 254]. Epigenetic histone marks are
associated with the segregation of activity, for example H3K4me generally marks transcription-
ally competent euchromatin while H3K9me3 and H3K27me3 mark silenced heterochromatin
[247, 255].
1.3.1 3C, TADs and compartments
Development of chromatin conformation capture (3C) technologies have allowed the study of
chromatin folding in greater detail. The general principle underlying these technologies is that
following cross-linking and digestion of the genome, the frequency of ligation of two regions
provides a measure of their proximity in 3D space. Sequencing the DNA allows the contacts
made in 3D space to be traced [256]. One of the most powerful of these technologies is Hi-C
(high resolution chromosome conformation capture), where pair-wise chromatin contacts are
detected genome-wide [257]. Hi-C studies demonstrated that chromatin is organised into A
and B compartments which are defined as regions of the genome that have higher likelihood of
interaction within themselves across a cell population. The A compartment primarily occupies
the nuclear interior and closely represents euchromatin, while B compartment resides near
nucleoli and nuclear lamina, consequently closely representing heterochromatin [258].
At finer detail Hi-C studies demonstrated genome organisation into topologically associated
domains (TADs), smaller regions of chromatin within which the probability of frequent inter-
actions is high across a cell population [259–262]. TADs are of an average size of ∼900 kb in
mammalian cells, but also a smaller subset of sub-TADs (∼200 kb) have been distinguished
by higher resolution Hi-C maps [263]. The best maps can reach nucleosome resolution and
were found to support 30 nm zig-zag chromatin fibre modelling, previously suggested by elec-
tron microscopy [264–268]. TAD structure organisation has been also supported by imaging
approaches [269–271]. TADs formation is considered to be largely dependent on their bound-
aries which are enriched for CTCF, a zinc finger architectural protein [259, 272]. Disruption of
CTCF binding sites leads to abnormal TADs and consequently chromosome folding, followed
by long-range transcriptional dysregulation, emphasising their important role in cell biology
[261, 273, 274]. Many TADs are segregated at the boundary region by pinning of DNA loops,
which are proposed to form by interplay of CTCF and a circular cohesin protein complex in a
loop extrusion model [275, 276]. A single-cell analysis approach showed that cohesin was not
required for TAD formation, but places them in preferential boundary positions, potentially
determined by CTCF binding sites [269]. Single-cell Hi-C studies revealed TADs to be dynamic
22
Chapter 1: Introduction
with respect to their size, positioning and insulation ability across different cell states [277–
279]. Furthermore, TADs and A/B compartments are abolished in methaphase meaning that
complete TAD restructuring is required every cell cycle [280, 281].
1.3.2 Phase separation, nuclear substructures and new methods
Increasing amounts of evidence suggest that phase separation is a major mechanism of nucleus
structuring [282, 283]. It has been suggested that phase separation governs formation of hete-
rochromatin and nuclear substructures such as nucleoli or nuclear speckles [284, 285]. Studies
report that transcription factors can directly regulate phase separation [286, 287]. Coactivator
and super-enhancer concentration by phase separation to control transcription has also been re-
ported [288]. While the majority of studies discuss liquid-liquid phase separation in the nucleus,
one recent study suggested that chromatin behaves like a liquid at small scales, but displays
solid-like polymer properties at 100-1000 nm scale in live cells, acting as a solid scaffold and
providing mechanical stability to the nucleus and the cell [289].
Nuclear substructures can act as anchor points for shaping the overall 3D genome structure.
Nuclear lamina interacts with specific chromatin regions, called lamina-associated domains
(LADs), which generally are heterochromatic [290–292]. Novel sequencing techniques revealed
that genome positioning around nuclear bodies influences chromatin folding and gene activity
[293–295].
Various new methods have been developed for studying 3D genome structure. Ligation free
(in contrast to 3C) techniques can provide orthogonal validation - for example GAM (genome
architecture mapping) uses cryosectioning of the cell followed by sequencing [296] or ChIA-drop
(chromatin interaction analysis) uses microfluidics to separate and recognise chromatin contact
clusters [297]. Additional information can be obtained too, like proximity to nuclear bodies and
RNA-DNA interactions that can be obtained by SPRITE (split-tool recognition of interactions
by tag extension) [294, 295]. 3D RNA-RNA interactions have also been mapped by RIC-seq
(RNA in situ conformation sequencing) [298], while DamC (DNA adenine methyltransferase
identification) can study DNA contacts in live cells [299]. Multiplexed imaging approaches
have also been developed which can range from visualising a whole nucleus with low overall
resolution to visualising a single locus at nm resolution [270, 271, 300].
23
Chapter 1: Introduction
1.3.3 3D genome structure simulation by single-cell Hi-C
Novel ways of investigating 3D genome have stemmed from the development of single-cell Hi-C
methodologies. A first study of this kind revealed the levels of heterogeneity among single-
cells in their genome structures. While organisation into domains remained a largely conserved
feature in different cells of the same type, stochastic behaviour was observed for larger inter-
chromosomal contacts [301]. Furthermore, the Hi-C contacts and restriction cut site positions6
were used as constraints to simulate the 3D structure of a single-copy X chromosome, which
demonstrated that 3D chromatin folding can be visualised from sequencing data.
3D chromatin structure research was taken to another level when the whole genome structure of
haploid mESC was visualised [278, 302] (Figure: 1.10, a). The haploid genome was instrumental
for structure simulation as unique contacts could be assigned to one set of chromosomes present.
Discrete chromosome territories can be recognised in this data, however 5-10 % chromosome
intermingling was present. The method visualised the positioning of A and B compartments and
their relationship to LADs and regions of high transcriptional activity (Figure: 1.10, b). Regions
of highest transcriptional activity were found to cluster in 3D space with close association to
active enhancers. A/B compartment segregation and positioning was largely conserved between
different individual cells, while TAD structures and DNA loop positioning was found to be
varied. Large-scale cell structure heterogeneity was illustrated by overlap of interactions at
the Nanog gene locus across a cell population from a 4C7 dataset with a single nucleus Hi-C
structure (Figure: 1.10, c). It shows that the Nanog gene can interact with different parts of
the genome in different cells with considerable variation in nucleus structure. As there was
conservation of A/B compartment structures, this suggests that chromosomes can reshuffle
within the nucleus, but they maintain their A and B compartment positioning. Furthermore,
this study combined 3D fluorescence microscopy with Hi-C of the same single-cell, to overlap
centromere images on top of the simulated Hi-C structures. The centromere positions obtained
through imaging corresponded to their supposed sequence position in the Hi-C structures, which
validated the accuracy of the overlap and the simulation. Extension of this method to overlap
G4 imaging and Hi-C structures in 3D is one core of this thesis, discussed in greater detail in
chapter 3.
In situ genome sequencing (IGS) also achieved combination of simultaneous imaging and se-
quencing of the same single-cell, however at much lower sub-micrometer spatial resolution [303].
This approach utilises introduction of spatial barcodes for in situ sequencing and imaging of
6Restriction enzymes are used in Hi-C protocols to cut different DNA strands, then if they are in close
proximity they can be ligated to one another. Knowledge on the possible cut sites based on the restriction
enzymes used is helpful for introducing additional constraints in the simulations.
74C is an extension of 3C methods where whole genome interactions are mapped for a single locus.
24
Chapter 1: Introduction
a
cb
Figure 1.10: (a) Single-cell HiC simulated structure of a mESC nucleus. Different colours denote
chromosome territories and A/B compartments. 100 kb structure resolution. (b) High RNA expression
regions co-localise with the A compartment, while LADs associate with the B compartment. (c)
Interaction points involving the Nanog gene from population-derived 4C dataset overlapped on a
single-nucleus structure. Adapted from T. J. Stevens, D. Lando, S. Basu, L. P. Atkinson, Y. Cao,
S. F. Lee, M. Leeb, K. J. Wohlfahrt, W. Boucher, A. O’Shaughnessy-Kirwan, J. Cramard, A. J.
Faure, M. Ralser, E. Blanco, L. Morey, M. Sanso´, M. G. S. Palayret, B. Lehner, L. Di Croce, A. Wutz,
B. Hendrich, D. Klenerman, and E. D. Laue, “3D structures of individual mammalian genomes studied
by single-cell Hi-C,” Nature, vol. 544, pp. 59–64, Mar. 2017 [278].
their locations, followed by amplification and high-throughput sequencing. This novel method
was used to report considerable differences in genome structure between embryonic development
stages. 3D chromatin structure reconstruction has also been achieved with purely imaging ap-
proaches with multiplex fluorescence in situ hybridisation (FISH) methodology. Although high
(2 kb) resolution was achieved, it was limited to reconstruction of a small 100 kb region [271].
25
Chapter 1: Introduction
Another study used lower resolution to reconstruct the whole genome structure, simultaneously
imaging nuclear bodies, but was limited to imaging 1 000 genomic loci [304].
An updated version of chromatin capture technique has allowed 3D simulation of diploid hu-
man cell genome structures. This novel method, termed Dip-C, used single-nucleotide poly-
morphisms (SNPs) from known maternal and paternal allelles to distinguish between the two
haplotypes of both chromosomes [279]. It achieved up to 20 kb chromatin resolution which
corresponded to ∼200 nm resolution in 3D space and determined that different cell types have
consistent 3D structure signatures sufficient for cell type recognition. Follow up studies using
Dip-C investigated inverted genome structures in mouse rod photoreceptor cells where hete-
rochromatin primarily localises to the nucleus interior. This work also suggests that regions
with high CpG frequency should be used as proxy for euchromatin in single-cell studies rather
than the A compartment, whose definition requires analysis of a population of cells [305]. Inves-
tigation of olfactory sensory neurons structures demonstrated clustering of olfactory receptor
genes, which could serve as means of control and play a role in following the ’one-neuron,
one-receptor’ rule of olfaction. Another study combined Dip-C with single cell transcriptomics
and reported major genome structure and transcriptome transformation during development
[306]. Also, 3D genome restructuring was linked to changes in DNA methylation in non-CpG
sites. Furthermore, sensory deprivation had little influence on the structure transition during
development suggesting that the genome restructuring program may be pre-determined.
Taken together, developments in 3D genome capture techniques opens a new exciting realm of
possibilities for studying 3D chromatin structure and its function relationships in the genome.
26
Chapter 2
Single-molecule G4 visualisation in live
cells
2.1 Research aims and rationale
One unanswered question in the G4 field has been a lack of robust endogenous G4 detection
methods in living cells, an area which this thesis explores. Endogenous G4 detection has its
major challenges, especially since G4s are secondary structures, capable of intrinsic dynamics
with respect to their folded and unfolded state. Importantly, the detection method must
distinguish between a folded G4 structure rather than G-richness of the primary nucleic acid
sequence. This has led to the development of a variety of probe molecules which rely on their
selective binding to folded G4 structures for achieving general G4 recognition. Most fluorescent
imaging and chemical pull-down sequencing methods of G4 detection rely on selective probe
binding as primary means of G4 recognition, however such means have their issues.
To ensure endogenous G4 detection, a method should not cause any significant perturbations
which may shift endogenous G4 landscape in cells. This is difficult to achieve with certainty
as G4s are hypothesised to exist in a dynamic equilibrium between their folded and unfolded
states. Therefore, any methods using G4 stabilising probe molecules may shift the equilibria if
an excess of probe molecules is used. Further considerations include potential secondary effects
like probe toxicity and cell stress in live cells and fixation or cell processing artefacts in fixed
cells. The exploration of potential G4 folding/unfolding dynamics is an important challenge to
address for which a real-time G4 detection method would open up such studies.
To directly address such questions and improve previous G4 visualisation methods, a single-
molecule fluorescence imaging detection platform was envisaged. By conjugating a G4-selective
27
Chapter 2: Single-molecule G4 visualisation in live cells
probe molecule to a fluorescent dye molecule one could indirectly visualise G4s by following
probe binding using single-molecule fluorescent microscopy. One advantage of single-molecule
microscopy is that it requires very low probe concentrations for detection which can be in the
sub-nanomolar range. Generally, this is much lower than measured in vitro Kd values for small
molecule binding to G4s. Therefore, for most probes it is reasonable to hypothesise that at
very low concentrations used for single-molecule imaging, they would be unlikely to lead to
significant perturbations to the global G4 landscape of the cell by induction of G4 formation.
Furthermore, low concentrations used for imaging would minimise cell toxicity of the probe and
any potential to induce cell stress.
An endogenous G4 detection method applicable to live cells would enable further questions
related to the function and possible mechanisms of action of G4s to be addressed. Are G4s
dynamic in living cells? Can the dynamic lifetime of G4s be measured? What is the cell cycle
dependence of G4s in a living cell? Is G4 formation associated with DNA processing activities
such as transcription and/or replication? The latter question could be tackled with the use
of co-localisation microscopy against transcription factors or RNA polymerases. Moreover, the
method could measure relative G4 abundance in different types of cells, like cancer cell lines
versus normal cell lines or follow changes in G4 levels over the course of stem cell differentiation.
Chapter outline
In this chapter, first I will describe the synthesis of a novel fluorescent G4 probe. Then another
probe and its isomeric control molecule will be introduced, prepared by collaborators of this
study. All probes will be characterised biophysically in terms of their fluorescent properties
and G4 binding selectivity. G4 ligand effects on global induction of G4 formation will be
investigated as well as their effect on G4 unfolding dynamics. I will discuss the value of using
smallest possible probe concentrations for G4 labelling, in order to minimise perturbations
caused by the probe and thereby to ensure endogenous G4 detection. In the following section I
will describe in vitro single-molecule imaging experiments on a G4-oligo covered surfaces using
TIRFM. Single-G4 visualisation in vitro will be demonstrated using G4 ligand probes. Then
will be followed by a demonstration of single-molecule observation of G4s in live U2OS cells
using HILO microscopy. Mirror comparisons of single-molecule imaging control experiments in
vitro on a surface and in cells, will be discussed as means to prove G4 visualisation as observed
binding events in imaging. Finally, the single-molecule imaging platform will be utilized to
study G4 unfolding dynamics, cell cycle dependence and relation to DNA processing. Key aims
and objectives in this study are summarised below:
28
Chapter 2: Single-molecule G4 visualisation in live cells
Key aims and objectives
• Synthesise a novel fluorescent G4 ligand probe.
• Biophysical characterisation of G4 probes used in this study in terms of their fluorescent
properties and G4 binding selectivity.
• Demonstrate that conditions used for single-molecule imaging do not globally induce G4
formation or globally perturb their dynamics.
• Demonstrate in vitro single-molecule visualisation of G4s on a G4-covered surface.
• Visualise G4s in live U2OS cells with single-molecule microscopy, ensuring minimal global
G4 perturbations for endogenous detection.
• Use the novel imaging platform to study G4 unfolding dynamics, cell cycle dependence
and relation to DNA processing.
In the next section, I will discuss key details of fluorophore probe design, an essential component
to enable imaging of endogenous G4s in live cells.
2.2 Fluorophore probe design
The most important feature of a fluorophore-conjugated G4 probe is its selectivity towards
folded G4s against other nucleic acid structures, especially double-stranded DNA which would
be a major competitor in a nuclear environment. The probe also requires a sufficiently bright
fluorescent output to be detected as single molecules. Generally, dedicated synthetic small
molecule fluorescent dyes perform best in this regard. Therefore, the strategy of the design of
a probe was to conjugate a well-established G4 ligand to a bright fluorescent dye via a suitable
linker. It is important that the conjugated dye and the linker do not disrupt G4 affinity of the
probe and its selectivity for G4s over dsDNA.
2.2.1 G4 ligands: PDS and PhenDC3
Two G4 ligands, PDS (1) and PhenDC3 (2), were chosen to be tested out over the course of the
project (First introduced in section: 1.1.3). Initially PDS (1) (Scheme: 1.1) was the primary
ligand of choice for conjugation to a fluorescent dye to obtain a probe molecule, as PDS (1)
has high G4 affinity and selectivity against double-stranded DNA as well as good cellular
29
Chapter 2: Single-molecule G4 visualisation in live cells
permeability [79]. PDS (1) has been also relatively well studied for its effects on mammalian
cells [54].
A potential benefit to having a second probe was recognised. A second probe could replace the
primary probe if it were to provide cleaner and more reliable results. Importantly, it would
provide cross-validation of the single-molecule imaging technique, as if two distinct G4 ligand
molecules gave similar results, it would raise confidence that any observed phenomena are G4-
related. Co-localisation of both probes when using different colour dyes could help distinguish
real G4 binding events from any non-specific ones, since it is unlikely that two different molecules
would have identical specific targets that are non-G4s. It was also envisaged that measurements
of time taken for an individual G4 to be rebound by a different probe could potentially reveal
G4 folding and unfolding dynamics. The dynamics of the residency time of the probe may be
influenced by cellular machinery acting on G4s, therefore comparison of residency times could
potentially inform on G4-protein interactions.
PhenDC3 (2) is a broadly used and relatively well established G4 ligand in the field [80]. This
molecule was chosen to as a second probe, since it has a high G4 binding affinity and selectivity
against double-stranded DNA as well as good cellular permeability. A clickable (containing an
alkyne moiety) PhenDC3 (2) derivative synthesis has been reported in the literature [307, 308]
allowing quicker preparation and access to the probe.
Structure and mode of binding
Although both PhenDC3 (2) and PDS (1) contain aminoquinoline moieties in their structures,
these are differently substituted and overall, both molecules have sufficient structural differences
to make them chemically distinct from each other. Moreover, it is expected that both molecules
would be unlikely to have common specific binding targets, aside from G4s.
PhenDC3 (2) binds DNA G4 structure by interacting via its flattened frame on top of the
G4 tetrad surface as shown by structural characterisation via NMR in solution (Figure: 2.1).
Charge interactions of positively-charged PhenDC3 (2) molecule and negatively charged DNA
phosphate backbone as well as pi-pi interactions are important in PhenDC3 (2) binding.
No experimental structures of G4-bound PDS (1) have been reported, though computational
models do exist [310]. These suggest that PDS (1) has a slightly bent binding framework onto
a G4 tetrad, with charge interactions between the protonated amines1 and the DNA phosphate
backbone (Figure: 2.2).
1Under physiological conditions.
30
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.1: NMR solution structure of PhenDC3 (2) stacked on top of Pu24T G4. (A) Ten su-
perimposed refined structures. Guanine bases are colored in cyan; adenine bases, green; thymine
bases, orange; backbone and sugar moieties, gray; O4 atoms, yellow; P atoms, red; PhenDC3 (2),
magenta. (B) Top view showing the stacking of PhenDC3 (2) on the top G-tetrad of Pu24T. Adapted
from W. J. Chung, B. Heddi, F. Hamon, M.-P. Teulade-Fichou, and A. T. Phan, “Solution Structure
of a G-quadruplex Bound to the Bisquinolinium Compound Phen-DC3,” Angew. Chem., vol. 126,
pp. 1017–1020, Jan. 2014 [309].
PDS (1) and PhenDC3 (2) have been reported to bind both 3’ and 5’ tetrad faces of a G4 by a
fluorescence quench assay [86]. For Kit1 G4 oligo PhenDC3 (2) showcased a 6-fold selectivity
towards 5’ tetrad (Kappd (3’) = 300 nM, K
app
d (5’) = 50 nM) while PDS (1) showed only a small
difference (Kappd (3’) = 440 nM, K
app
d (5’) = 640 nM). Although it seems plausible for two G4
ligands to bind to both G4 tetrads simultaneously, the fluorescence quench assay data did not
provide evidence for this. While reports exist of PDS derivatives co-binding a G4 structure
together with proteins [87, 89], it is unclear whether proteins utilise the second G4 tetrad in a
co-binding scenario.
Biological implications
The effects of both PhenDC3 (2) and PDS (1) on DNA G4s have been extensively studied
in vitro and in cells. Mapping of G4 sites by G4-seq showed a strong overlap of 85 % of
observed G4 peaks resulting from PhenDC3- or PDS induced DNA polymerase stalling [52],
suggesting that both molecules have largely similar G4 targets. While PhenDC3 (2) shows high
affinity and selectivity for G4s over dsDNA [80], it discriminates poorly between different G4
conformations [311]. PDS (1) is also considered a general G4 structure binder [54]. PhenDC3
31
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.2: Computational docking and 2D interaction representation of PDS (1) with H-telo G4.
Adapted from R. Rocca, C. Talarico, F. Moraca, G. Costa, I. Romeo, F. Ortuso, S. Alcaro, and
A. Artese, “Molecular recognition of a carboxy pyridostatin toward G-quadruplex structures: Why
does it prefer RNA?,” Chem. Biol. Drug Des., vol. 90, no. 5, pp. 919–925, 2017 [310].
(2) can cause G4 dependent genome instability [312, 313] while PDS (1) has also been shown
to cause replication- and transcription-dependent DNA damage as well as triggering a DNA-
damage response at telomeres [54, 79]. Both PDS (1) and PhenDC3 (2) have been reported
to affect transcription [116, 314, 315] and translation [163, 316, 317] upon cell treatment and
in vitro experiments. Moreover, PhenDC3 (2) has been subjected to generate G4-associated
replication origins in cells [140]. PDS (1) and PhenDC3 (2) have been used in synthetic
lethality studies [54, 178], to identify genes in which down-regulation resulted in sensitivity to
G4 ligand treatment. These genes were found to relate to DNA damage repair, helicase activity,
chromatin remodellers, ubiquitin and splicing, as G4 ligand treatment sensitisers. Treatment
with PhenDC3 (2) has also been found to delay human cell differentiation [71].
2.2.2 Silicon Rhodamine Dye - the reporter moiety
A suitable fluorescent dye is essential for successful imaging. Silicon rhodamine (SiR) is ideally
suited for live-cell single-molecule imaging due its high brightness, photostability and biocom-
patibility. It has a sensitive fluorogenic character as it exists in an equilibrium between its
nonfluorescent spirolactone (OFF form) and a fluorescent zwitterion (ON form) (Scheme: 2.1).
This equilibrium is dependent on the dielectric constant of the solvent environment around
32
Chapter 2: Single-molecule G4 visualisation in live cells
the dye: for example, unspecific binding to hydrophobic surfaces favours the spirolactone OFF
form while highly polar environments like the solvent shell around nucleic acids greatly shift the
equilibrium towards the fluorescent zwitterion. A dielectric constant of 30 gives approximately
a 1:1 ratio of the two forms [318]. SiR blinks inside a cell by interchanging between the ON and
OFF states, and thus is present in the non-fluorescent state for considerable periods of time
resulting in reduced photobleaching and prolonged lifetime. Moreover, SiR has good cellular
permeability and low toxicity which is valuable in cellular studies. Being a far-red dye, it also
can be excited at wavelengths that avoid the majority of cell auto-fluorescence [319]. Silicon
rhodamine therefore has very suitable properties for real-time single-molecule imaging, and it
was thus chosen for making a fluorescent G4 probe.
O
H
NR
O
Si
O
N N
O
H
NR
O
Si
O
N N
ON formOFF form
Scheme 2.1: Silicon rhodamine dye equilibrium between nonfluorescent spirolactone (OFF form)
and a fluorescent zwitterion (ON form).
The second fluorophore probe was designed to include a PhenDC3 (2) scaffold (an active region
for G4 binding) covalently linked to SiR dye. The linker had to be sufficiently long to minimise
the effect of the attached SiR moiety on the binding strength of the original PhenDC3. Both
the dye and 2 have good bioavailability, therefore it was hypothesised that the linked molecule
would probably have a lower, but still sufficient cell membrane permeability to be used for live
cell imaging.
2.3 Probe preparation and biophysics
2.3.1 PhenDC3-SiR synthesis
PhenDC3 derivative synthesis (Scheme: 2.2) with a clickable alkyne handle (16) was adopted
from a method reported in the literature [307, 308]. The original synthesis procedure required
optimisation to obtain better results and higher yields. Final compound PhenDC3-SiR (17)
was prepared by reacting the alkyne of (16) with an azide linked to Silicon Rhodamine dye via
a copper-catalysed azide-alkyne cycloaddition (CuAAC) click reaction [320]. Detailed experi-
mental conditions are outlined in Materials and methods chapter (Section: 5.8)
33
Chapter 2: Single-molecule G4 visualisation in live cells
N HN
O
N NH2 N HN
O
O
O
O
110 oC
Ph2O
210 oC
64%
POCl3
95 oC
OO
OO
MeO OMe
OMe
N N
Cl
N N
Cl
OH
O
HO
O
NCS, BPO H2SO4, 85 oC
3-Aminoquinoline
EDCl, HOAt
DMFN
Cl3C
N
CCl3
Cl
CHCl3, 61 oC
95% 96% 98%
N N
Cl
NH
O
HN
O
NN
N N
HN
NH
O
HN
O
NN
NH2
H2N
NH2
MW, 120 oC
O
HO
EDCl, HOAt, DMF
1.
2. MeI, DMF, 40 oC
78% 97%
N N
HN
NH
O
HN
O
NN
NH
O
I I
O
NH
O
Si
O
N
N
SiR-azide N N
HN
NH
O
HN
O
NN
NH
O
•2 CF3COO
N N
N
CuSO4•5 H2O
Sodium ascorbate
H2O / tBuOH, 12 h
58%32%
8 9 10
11 12 13
14 15
16 17
Scheme 2.2: PhenDC3-SiR (17) fluorophore derivative synthesis with 8 % overall yield.
34
Chapter 2: Single-molecule G4 visualisation in live cells
The synthesis started by condensation of 2-methylquinolin-8-amine (8) with trimethylorthoac-
etate and Meldrum’s acid to give intermediate 9 which was then converted to compound 10 by
thermal decomposition and [3,3] pericyclic cyclisation. I propose that the reaction proceeded by
initially fragmenting the 1,3-dioxane ring moiety, evolving carbon dioxide and acetone, forming
a ketene which then participated in [3,3] electrocyclic ring closing reaction (Scheme: 2.3).
N HN
O
O
O
O
N HN
C
N HN
C
N HN
O
O
CO2
N HN
OO H
C
O
[3,3]
+H+
-H+
9
10
Scheme 2.3: Mechanism of compound 9 conversion to 10 via thermal fragmentation followed by
[3,3] electrocyclic ring closing reaction.
4-chloro-2,9-dimethyl-1,10-phenanthroline (11) intermediate synthesis followed by treatment
of 10 with phosphoryl chloride. Radical chlorination of the methyl groups on 11 with N -
chlorosuccinimide (NCS) and catalytic benzoyl peroxide (BPO) led to compound 12 which was
then hydrolysed in conc. sulphuric acid to 13 (Scheme: 2.2).
However, the literature described [307] procedure for hydrolysis of 12 with conc. sulphuric acid
(Scheme: 2.4, A) did not provide the desired result, therefore optimisation was required. The
reaction was run at different temperatures for both steps (conversion of 11 to 13) with and
without inclusion of extra water. Optimised conditions were obtained using conc. sulphuric
acid at 85 ◦C for 2.5 hours as a single step (Scheme: 2.4, B), affording 13 with an excellent
98 % yield. For characterisation purposes compound 18 was also synthesised (Scheme: 2.4, C)
by running the reaction at a higher temperature.
Amide coupling to 3-aminoquinoline followed next using EDCI and HOAt as coupling reagents
to afford 14 which was then reacted with putrescine using microwave (MW) irradiation to
give 15 (Scheme: 2.2). Note that this molecule has proven to be useful for synthesising other
PhenDC3 derivatives with different functional handles utilised in other projects. Another amide
coupling reaction was then carried out to couple compound 15 to 4-pentynoic acid, followed by
methylation of quinoline moieties by methyl iodide to afford clickable PhenDC3 derivative 16.
This was reacted with an azide-bearing SiR dye as previously described [320] to give 3.1 mg
35
Chapter 2: Single-molecule G4 visualisation in live cells
N N
OH
OH
O
HO
O
N
Cl3C
N
CCl3
Cl
(i) H2SO4, 100 oC, 3h
(ii) H2O, 120 oC, 1h
N N
Cl
OH
O
HO
O
H2SO4, 85 oC, 2.5h
N N
OH
OH
O
HO
O
(ii) H2O, 120 oC, 1h
(i) H2SO4, 150 oC, 2h
N N
Cl
OH
O
HO
O
+
2 1:A
B
C
98 %
88 %
12
13
13
18
18
Scheme 2.4: (A) Literature [307] hydrolysis method provided a 2:1 mixture of products inseparable
by prep. HPLC. (B) Optimised reaction conditions to give desired compound 13. (C) Reaction to
give 18 for characterisation purposes.
of the final compound PhenDC3-SiR (17) which could be used for single-molecule fluorescence
imaging. The overall yield of the synthesis over 10 steps was 8 %.
2.3.2 SiR-PyPDS and SiR-iPyPDS preparation
The probe molecule SiR-PyPDS (19) and a control probe molecule SiR-iPyPDS (20) were
prepared by Dr Marco Di Antonio and Marco Catalano [213]. SiR-PyPDS (19) has a linker
length optimised for highest binding affinity to G4s and was designed to have pyrolidine rather
than amine moieties for improved lipophilicity and simplified synthesis2 (Scheme: 2.5). An
isomer SiR-iPyPDS (20) was designed to be structurally similar but possess weaker G4-binding
properties for use in control experiments. By switching the 2-(pyrrolidin-1-yl)ethoxy- group
from the 4- position on quinoline ring to the 8- position, the PyPDS moiety would have a
different dominant conformation thus reducing G4 binding ability.
2Hence the name pyrolidine PDS (PyPDS) for distinction from PDS.
36
Chapter 2: Single-molecule G4 visualisation in live cells
N O
HN
N
O
O
NH
N
O O
NN
HN
O
HN
O
O
Si
O N
N
N O
HN
N
O
O
NH
N
HN
O
HN
O
O
Si
O N
N
OO
N N
SiR-PyPDS SiR-iPyPDS
19 20
Scheme 2.5: Structures of the probe molecule SiR-PyPDS (19) and an isomeric weaker G4 binder
control molecule SiR-iPyPDS (20).
2.3.3 FRET melting experiments
Having obtained PhenDC3-SiR (17), it was neccessary to validate that the compound retained
specificity for G4 structures as opposed to double-stranded DNA when compared to unmodified
PhenDC3 (2). To do this, a series of biophysical Fo¨rster resonance energy transfer (FRET)
melting experiments were set up with oligonucleotides which could fold into G4 structures,
together with G4 ligands [321]. For such experiments, the oligonucleotides were made with two
fluorophores attached to one of the strands in such a way that when a G4 folded correctly these
fluorophores become close in proximity leading to more efficient FRET, which shifts the emitted
light wavelength and alters its intensity. The melting temperature can be then determined, by
recording emitted light spectra at various temperatures. G4 ligands stabilise G4s and increase
their melting temperature, hence calculated ∆Tm values are an indirect measure of binding
strength.
37
Chapter 2: Single-molecule G4 visualisation in live cells
For FRET melting experiments PhenDC3-SiR (17) and unmodified PhenDC33 (2) were com-
pared. The results were used to interpret how the linker and silicon rhodamine dye affect
PhenDC3 derivative affinity for G4s.
Oligonucleotides synthesised for FRET were from genomic regions known to form G4s in vitro
[14, 18, 26] KIT1 and MYC genes as well as from human telomeres (H-telo). The oligonu-
cleotides were doubly labelled with fluorescent tags FAM and TAMRA which can partake
in efficient FRET when the G4 is folded but not when it is unfolded. FAM is the donor
fluorophore and TAMRA the acceptor fluorophore (Figure: 2.3). Moreover, control oligonu-
cleotides which are unable to form G4 structure were tested to assess ligand binding specificity
with double-stranded DNA. These oligonucleotides were dual-labelled 20-mers comprising a
self-complementary sequence with a central polyethylene glycol linker able to fold into a hair-
pin of double-stranded DNA4.
Figure 2.3: Schematic representation of FRET melting assay. When the G4 is folded, efficient
FRET can occur between FAM and TAMRA fluorophores because of the short distance between
them, therefore fluorescence emission of FAM is quenched. Unfolded G4s have FAM and TAMRA
further apart thus FRET is minimal and a characteristic FAM emission at 533 nm can be observed.
For FRET melting experiments 200 nM of each oligonucleotide was tested against a range of
ligand concentrations (0 to 10 µM), and curves of ∆Tm values with respect to ligand concen-
tration were plotted. The data (Figure: 2.4) displayed almost identical FRET melting curves
for PhenDC3 (2) and PhenDC3-SiR (17) indicating that the linker and dye moiety inclusion to
PhenDC3 (2) did not disrupt its binding ability to G4 structures. Importantly, hairpin binding
did not increase indicating that selectivity was retained. Note that SiR absorption maximum
is at 645 nm [318], which is sufficiently far from FAM emission at 533 nm not to interfere at
observation wavelengths.
3Prepared previously and kindly provided by Dr M. Di Antonio.
4Oligo sequences can be found in Materials and methods chapter (Section: 5.1.1).
38
Chapter 2: Single-molecule G4 visualisation in live cells
0 2 4 6 8 10
0
10
20
30
Concentration / µM
Δ
T 1
/2
 / 
K
FRET Melting with PhenDC3
H-telo
Myc
Kit1
hairpin dsDNA
0 2 4 6 8 10
0
10
20
30
Concentration / µM
Δ
T 1
/2
 / 
K
FRET Melting with PhenDC3-SiR
H-telo
Myc
Kit1
hairpin dsDNA
Figure 2.4: ∆Tm against ligand concentration curves displaying denaturation stabilisation of G4
structures in oligonucleotides by PhenDC3 (2) and a fluorescently labelled PhenDC3-SiR (17). Lig-
ands dissolved in ITC buffer (10 mM potassium phosphate, 70 mM KCl, 0.1 mM EDTA, pH 7.0) were
added in a 1:1 volume ratio to FRET oligo solutions at 200 nM in potassium cacodylate buffer (60
mM, pH 7.4).
Competition experiments
Competition experiments were then carried out to further test PhenDC3-SiR (17) selectivity for
G4s against dsDNA binding. When competing with a short hairpin dsDNA oligo (Figure: 2.5,
LHS), for H-telo and Kit1 G4s the melting temperature stabilisation by G4 ligand decreased
upon addition of the competitor. At 500 equivalents of dsDNA w.r.t. FRET G4 oligo, H-telo
was stabilised by 8.5 K and Kit1 by 15.2 K, suggesting that a much larger excess of dsDNA
competitor would be required to completely abolish probe binding to these G4 structures.
Figure 2.5: Competition FRET melting experiments with 1 µM PhenDC3-SiR (17) stabilising DNA
structures. LHS - competition with hairpin dsDNA oligo. RHS - competition with extracted genomic
U2OS DNA. Equivalents calculated by mass excess. FRET oligo concentration - 200 nM, pH 7.4.
In contrast, Myc G4 melting stabilisation remained the same on addition of dsDNA suggesting
very high PhenDC3-SiR (17) binding selectivity to this particular G4. Same trend was observed
when competing with extracted genomic U2OS cells DNA (Figure: 2.5, RHS), Myc showed
no change in FRET stabilisation and H-telo was the most sensitive to competition, being
39
Chapter 2: Single-molecule G4 visualisation in live cells
completely out-competed by dsDNA at 50 mass equivalents. This suggests that PhenDC3-SiR
(17) binds to Myc oligonucleotide more strongly and with better selectivity against other forms
of DNA. Note that extracted genomic U2OS DNA should contain G4 structures, but the small
fraction of G4s can be neglected when compared to amount of dsDNA5.
2.3.4 Fluorophore properties characterisation
The majority of results in this chapter have been published [213] and were part of multiple
collaborations. I explicitly state which results were not from my own contribution.
Absorption and emission
The fluorogenic properties of the compounds synthesised and presented above were charac-
terised. In particular, attention was paid to investigate SiR fluorogenic character upon G4
binding.
Absorption and emission spectra of PhenDC3-SiR (17) and SiR-PyPDS (19) were determined
(Figure: 2.6). By introducing 0.1 % SDS, a detergent used to stabilise the zwitterionic ON
fluorescent form (Scheme: 2.1), it was evident that both probe molecules were primarily present
in OFF fluorescent state in PBS pH 7.4 buffer6.
Figure 2.6: Absorption and emission spectra of PhenDC3-SiR (17) and SiR-PyPDS (19) at 1 µM.
Absorption spectra denoted by dashed lines and emission spectra denoted by solid lines. Experiment
was carried out at pH 7.4, 20 ◦C.
Next, the effect of G4 binding on fluorgenic character of PhenDC3-SiR (17) and SiR-PyPDS
5Counting 700 000 G4 structures in human genome by G4-seq [52] and 20 bases length for a single average
G4, then G4s comprise of about 14 Mb of genome sequence length. Human genome being 3.2 billion bases, then
G4s account for just 0.4 % of the genome.
6Which was main buffer used in our imaging experiments.
40
Chapter 2: Single-molecule G4 visualisation in live cells
(19) was measured. Emission of both probes increased significantly when bound to Kit1 and
Myc G4s, the latter showing a more pronounced response (Figure: 2.7). The effect on ab-
sorption was smaller, for SiR-PyPDS (19), whereby binding to Myc G4 increased the observed
absorption, but not for Kit1 G4. For PhenDC3-SiR (17), Myc G4 barely increased measured
absorption, but Kit1 increased it substantially. Changes to absorption and emission wavelength
maxima were minimal in all tested cases (Table: 2.1).
Figure 2.7: Absorption and emission spectra of PhenDC3-SiR (17) and SiR-PyPDS (19) at 5 µM
upon binding to G4 oligos at 10 µM. Absorption spectra denoted by dashed lines and emission spectra
denoted by solid lines. Experiment was carried out at pH 7.4, 20 ◦C.
PhenDC3-SiR (17) SiR-PyPDS (19)
Absorption max Emission max Absorption max Emission max
/ nm / nm / nm / nm
PBS 655 671 654 672
PBS + 0.1 % SDS 650 670 649 669
Myc G4 658 671 655 677
Kit1 G4 659 674 657 677
K+ buffer 656 671 656 674
Table 2.1: Absorption and emission maxima of PhenDC3-SiR (17) and SiR-PyPDS (19).
Quantum yield
Extinction coefficients and quantum yields of SiR-PyPDS (19) were determined with and with-
out being bound to Myc G4 (Figure: 2.8). Strikingly, the probe showed a six-fold increase in
extinction coefficient when bound to Myc G4 suggesting a large SiR equilibrium shift to the
zwitterionic ON fluorescence state. Moreover, the quantum yield increased almost two-fold.
When comparing SiR-PyPDS (19) bound to Myc G4 with unconjugated SiR dye fluorescence
41
Chapter 2: Single-molecule G4 visualisation in live cells
properties [318], there is a slight shift of absorption and emission maxima (Table: 2.2), sug-
gesting an interaction between SiR dye and PyPDS moieties in the probe molecule. Moreover,
the quantum yield in both instances was measured to be identical at 0.39, while the extinc-
tion coefficient decreased four-fold in comparison to the unconjugated dye. Assuming that
when SiR-PyPDS (19) binds to Myc, it shifts completely to the ON fluorescent equilibrium
state, the decrease in extinction coefficient but not in quantum yield, suggests that PyPDS
can quench SiR fluorescence brightness after conjugation via cis-interactions. This has been
observed previously as PDS and other G4 ligands were reported to quench the fluorescence of
different dyes [86, 322, 323].
Figure 2.8: Quantum yield of SiR-PyPDS (19) with and without being bound to Myc G4. Ex-
periment was carried out in K+ buffer at pH 7.4, 20
◦C. Experiments and analysis carried out in
collaboration with Dr Lisa-Maria Needham.
λabs / nm λem / nm  / M·cm QY
SiR-PyPDS 656 674 4 000 0.21
SiR-PyPDS + Myc G4 654 678 26 000 0.39
SiR-carboxyl [318] 645 661 100 000 0.39
Table 2.2: Quantum yield of SiR-PyPDS (19) with and without being bound to Myc G4 and its
comparison to the unconjugated dye. Experiments and analysis carried out in collaboration with Dr
Lisa-Maria Needham.
42
Chapter 2: Single-molecule G4 visualisation in live cells
Fluorescence dependence on pH
It is important to understand if probe fluorescence can be influenced by environmental fac-
tors, therefore absorption and fluorescence relationships against pH were investigated. The
PhenDC3-SiR (17) pH titration curve for absorption was of sigmoidal shape, having higher ab-
sorption at low pH (Figure: 2.9). Meanwhile, the SiR-PyPDS (19) absorption titration curve
curiously showed a dip at pH 5, but still overall followed the same trend of higher absorbance
at low pH. The dip was likely due to unexpected buffer effects upon changing the buffer from
PBS at pH 6 to citric acid buffer at pH 5.
Figure 2.9: Absorbance pH titration curves of PhenDC3-SiR and SiR-PyPDS at 1 µM. Experiment
was carried out at 20 ◦C. Error bars indicate mean ± sd.
Fluorescence relationship vs pH was also determined for SiR-PyPDS (19). The pH titration
curve was of sigmoidal shape with highest fluorescence at low pH (Figure: 2.10). It suggests
that SiR is primarily in ON state at low pH and in OFF state at high pH. This may arise
due to acidic conditions facilitating lactone opening and being better able to stabilise the
formed carboxylate, possibly by protonation, than in basic conditions. Moreover, highly basic
conditions may lead to hydroxide attacking the lactone and trapping it in closed form as a
lactone bis-olate. The inflection point of the pH titration curve was at about pH 7.5, suggesting
that the probe at physiological pH can have significant fluorescence fluctuations even with small
pH changes. However, this can turn out as an advantage for live cell imaging, as at physiological
pH, significant portion of the probe would be blinking while in solution. Blinking is desirable as
most of unbound probe would be in the OFF state and contribute less background from out of
focus planes. Overall, it helps with observing single molecules. Moreover, the probe would be
sensitive for fluorescence light-up, therefore it could switch to bright ON form for when bound
to DNA G4s.
43
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.10: Fluorescence pH titration curve of SiR-PyPDS at 1 µM concentration. Experiment
was carried out at 20 ◦C. Error bars indicate mean ± sd.
2.3.5 Fluorescence light-up experiments
The fluorogenic character of SiR moiety when connected to G4 ligands was further exploited to
determine the binding affinities of PhenDC3-SiR (17) and SiR-PyPDS (19) to different types
of G4s via fluorescence light-up experiments. These showed that PhenDC3-SiR (17) had sub-
micromolar binding affinity to DNA G4s, Kit1, H-telo, Kras, strongest binding being to Myc G4
with Kd of 0.034 ± 0.008 µM, with good selectivity versus ds- and ssDNA (Figure: 2.12, Table:
2.3). Myc mut (short for mutant) and Kit1 mut were altered G4 oligo mutant sequences where
GGG tracts were switched to GTG as to avoid G4 formation. Both of these G-rich sequences
showed a linear fluorescence dependence on their concentration suggesting weak non-specific
binding.
a
5 10
0
50,000
100,000
150,000
[DNA] / µM
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Fluorescence light-up titrations
with PhenDC3-SiR 100 nM
MYC
MYC mut
H-TELO
dsDNA
ssDNA
ssDNA comp
b
5 10
0
50,000
100,000
150,000
[DNA] / µM
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Fluorescence light-up titrations
with PhenDC3-SiR 100 nM
Kit1
Kit1 mut
KRAS
hairpin dsDNA
Figure 2.11: PhenDC3-SiR (100 nM) fluorescence light-up binding curves titrated with: (a) Myc,
H-telo, Myc mutant, dsDNA, ssDNA, ssDNA comp; (b) Kit1, Kras, Kit1 mutant, hairpin dsDNA.
Experiment was carried out in K+ buffer (pH 7.4) at 20 ◦C. Error bars indicate mean ± sd.
44
Chapter 2: Single-molecule G4 visualisation in live cells
PhenDC3-SiR (17) Kd / µM SiR-PyPDS (19) Kd / µM
Myc 0.034 ± 0.008 0.63 ± 0.08
Kit1 0.14 ± 0.09 1.0 ± 0.1
H-telo 0.089 ± 0.04 2.0 ± 0.8
KRAS 0.11 ± 0.06
Table 2.3: Kd parameters determined from fitting the binding curves of PhenDC3-SiR (17) and
SiR-PyPDS (19) fluorescent light-up upon G4 binding. Errors indicate sd.
SiR-PyPDS (19) fluorescence light-up experiments showed a slightly weaker probe binding to
G4s than PhenDC3-SiR, but with excellent selectivity against ss- and dsDNA (Figure: 2.12,
Table: 2.3). Moreover, SiR-iPyPDS isomeric control molecule showed poor binding to Myc G4,
validating its use for control experiments.
a b
Figure 2.12: (a) SiR-PyPDS (19) fluorescence light-up binding curves. Data are plotted as the ratio
of the SiR fluorescence emission at 633 nm for every titration point over the emission measured in
buffer alone and normalised to the highest fluorescence emission measured. (b) Differential G4-binding
of SiR-PyPDS (19) and SiR-iPyPDS (20). Experiments were carried out in K+ buffer (pH 7.4) at 20
◦C. Error bars indicate mean ± sd. Carried out by Dr Marco Di Antonio and taken from [213].
2.3.6 FRET cascade indicates direct G4 ligand binding to a G4 oligo
Next experiments investigated PhenDC3-SiR (17) and SiR-PyPDS (19) binding to FRET
oligos. It was envisaged that a double FRET cascade is possible from the FRET pair on a
G4 oligo to SiR dye on the G4 ligand. In greater detail, the FAM dye, excited at 492 nm can
undergo FRET to TAMRA dye when a G4 is folded, which then can emit at 580 nm, and if a
G4 ligand is bound, perform a second energy transfer to SiR dye, which then emits at 672 nm
(Figure: 2.13). This system was good for testing ligand binding as the double FRET cascade
would only be possible when both G4 was folded and bound by a G4 ligand.
45
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.13: Double FRET system with FAM (green) to TAMRA (yellow) to SiR (red).
FRET was observed between Kit1 oligo and both fluorescent probes evident by an increase of
fluorescence signal at 670 nm and a decrease at 580 nm (Figure: 2.14, a, b). However, the
signal at 670 nm did not increase steadily with increasing concentrations of probe molecules.
It increased up to 2.5 µM for both PhenDC3-SiR (17) and SiR-PyPDS (19) and then started
to decrease at higher concentrations (Figures: 2.15). The decrease is most likely due to self-
quenching of the probe molecules, which could happen by collisions of excited fluorophores, or
due to formation and energy transfer to non-fluorescent dimers, both mechanisms are favoured
at higher concentrations [324–326].
A similar trend was observed for H-telo oligo, but this time the maximum FRET signal at 670
nm was at 1.25 µM concentration of the probe (Figure: 2.14, c, d, 2.15). However, no FRET
cascade was observed for Myc oligo, only quenching of TAMRA signal at 580 nm (Figures:
2.14, e, f; 2.15). Possibly this is due to an unfavourable dye arrangement in space for double
FRET cascade upon probe binding to G4 and/or due to particularly strong quenching effects
between the dyes. Finally, dsDNA did not show a cascade FRET response and acted as a
control (Figure: 2.14, g, h). A small but noticeable increase of emission signal at 670 nm was
due to direct SiR excitation by 492 nm laser, demonstrating the extent of direct excitation.
Also, it suggests that the large increase of 670 nm signal for Kit1 and H-telo oligos must be
a result of FRET. Taken together, these results further demonstrate PhenDC3-SiR (17) and
SiR-PyPDS (19) ability to directly bind to G4s, also with selectivity over dsDNA.
46
Chapter 2: Single-molecule G4 visualisation in live cells
a b
c d
e f
g h
Figure 2.14: Double FRET cascade of SiR with different G4 FRET oligos. Experiments were carried
out with 1 µ of oligonucleotide in K+ buffer (pH 7.4) at 20 ◦C.
47
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.15: 670 nm FRET signal emission by SiR dye moiety in PhenDC3-SiR and SiR-PyPDS.
Experiments were carried out with 1 µ of oligonucleotide in K+ buffer (pH 7.4) at 20 ◦C.
2.3.7 G4 induction
It was important to determine whether the nanomolar probe concentrations used for single-
molecule imaging were capable of G4 folding induction at a global scale7 and how that relates
to higher probe concentrations. Here, I define G4 folding induction as creating and increasing
G4 formation in a system. For this study, an inducible in vitro G4 oligo system was sought
which had a majority of G4s in an unfolded state at equilibrium. It has been previously reported
that DNA G4 oligos annealed in 10 mM Tris pH 7 buffer were primarily in unfolded state which
can then be induced for folding upon K+ addition [96]. These conditions were tested on the
G4 oligos by CD spectroscopy. Kit1 showed characteristic peaks far a parallel G4 in K+ buffer
with a positive signal at 263 nm and a corresponding negative signal at 240 nm with absence
of these signals in Tris buffer (Figure: 2.16, a) [36, 327, 328]. These results suggested that
the Kit1 G4 oligo is a suitable system to study G4 folding induction. Myc on the other hand,
showed the characteristic parallel G4 peaks in both buffer conditions meaning that a Myc G4
can form in Tris buffer too (Figure: 2.16, b). Only the mutant Myc sequence in Tris buffer did
not indicate the presence of a G4. Consequently, it meant that Myc was not an ideal system
for studying G4 folding induction. Since Myc was a common model sequence used in other
experiments, it was decided to test Myc propensity for induction by G4 ligands nevertheless.
A Cy3/Cy5 dye FRET system on G4 oligos was utilised to study G4 induction (Figure: 2.17).
The FRET system was adopted from the previous G4 studies [329, 330]. The construct con-
sisted of a G4 forming DNA sequence labelled with Cy5 dye on 5’ end, while containing an
overhang for complementary sequence binding on the 3’ end. The Cy3 dye is positioned on the
complementary strand distant from the G4 to prevent interactions, which could have caused
quenching. It was also reasoned that Cy3 placement within the oligo sequence rather than at
7Global scale here meaning that a noticeable change in a property can be measured for the whole G4
population in a system, after averaging.
48
Chapter 2: Single-molecule G4 visualisation in live cells
a b
Figure 2.16: (a) CD spectra of Kit1 in K+ buffer (G4 stabilising condition, pH 7.4) and in Tris
buffer (G4 destabilising condition, pH 7.4). (b) CD spectra of Myc in different buffers and Myc
mutant sequence. Experiments were carried out with 10 µ of oligonucleotide at 20 ◦C.
the ends helps to protect the dye from further quenching by introduced G4 ligands, as this
was an encountered problem while using FAM/TAMRA FRET oligos used for FRET melting
experiments containing the dyes at the oligo ends (Section: 2.3.3).
Figure 2.17: Cy3/Cy5 dye oligo FRET system to study G4 induction by ligands. More efficient
FRET is observed when G4 is folded.
The degree of G4 induction upon addition of G4 ligand PDS was monitored by following the
relative intensities of Cy3 and Cy5 dye signals. When a G4 was unfolded, FRET would be
inefficient due to larger distance between the dyes, and led to stronger Cy3 emission signal at
567 nm (Figures: 2.17, 2.18). Meanwhile, when a G4 was folded it would reduce the average
distance between the dyes giving a more efficient FRET and a stronger Cy5 signal at 670 nm.
To indicate the extent of the effect of ligand induction on G4 formation, a formula for fold
induction has been derived:
Fold inductioni =
F (Cy5)i
F (Cy3)i
−F (Cy5)0F (Cy3)0
F (Cy5)0
F (Cy3)0
Where F(Cy5)i and F(Cy3)i denote the fluorescence intensity of Cy5 and Cy3 dyes respectively
at a particular ligand concentration and F(Cy5)0 and F(Cy3)0 denote the fluorescence intensity
of the respective dyes with no ligand present. Fold induction describes the relative increase
49
Chapter 2: Single-molecule G4 visualisation in live cells
of G4 formation compared to the G4 oligo state without any ligand binding in Tris buffer, in
which G4s were found to be destabilised.
a
550 600 650 700 750
0
200
400
600
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Myc
By itself
20 nM PDS
3 µM PDS
7 µM PDS
20 µM PDS
b
550 600 650 700 750
0
200
400
600
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Kit1
By itself
20 nM PDS
3 µM PDS
7 µM PDS
20 µM PDS
c
550 600 650 700 750
0
200
400
600
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
H-telo
By itself
20 nM PDS
3 µM PDS
7 µM PDS
20 µM PDS
d
Figure 2.18: Induction of G4-folding by increasing concentrations of SiR-PyPDS (19) measured
with dually labelled FRET oligos at 1 µM. (a-c) Fluorescence emission spectra under Cy3 excitation
for each G4 sequence. (d) Changes in the FRET ratio can be observed at µM concentrations of PDS
(1) for Kit1 and H-telo and at larger concentrations for Myc, indicative of G4 induction. Experiments
were carried out in 10 mM Tris buffer (pH 7.4) at 20 ◦C.
G4 induction experiments indicated that G4 formation can be induced for Myc, Kit1 and H-
telo G4 oligos (1 µM), at high µM concentrations of PDS (Figure: 2.18)8. Kit1 and H-telo
exhibited about 2-fold increase of G4 formation at 5 µM of PDS while 20 µM were required
to induce Myc by 1.5-fold. Importantly, 20 nM PDS, which would be used for single-molecule
imaging labelling experiments, showed no observable global G4 induction for all oligos tested.
Taken together, these results demonstrate the importance of using small concentrations of probe
molecules for endogenous G4 detection, to avoid perturbing the global G4 levels by induction of
G4 formation, which could happen at higher concentrations when G4 ligands can substantially
stabilise folded G4 structures.
8The G4 ligand effect on G4 induction was found to happen in seconds (and stay at equilibrium for hours),
therefore time-dependent G4 induction was not investigated.
50
Chapter 2: Single-molecule G4 visualisation in live cells
2.3.8 G4 unfolding kinetics
G4 ligand effects on the global G4 dynamics were studied next. In particular, it was of interest
to determine the ligand effect on the G4 unfolding kinetics. For this, kinetic trap experiments
were designed according to previous kinetic G4 oligos unfolding studies [329–331], in which a
G4 oligo with FAM/TAMRA dye FRET system was used for observing G4 fold state. The
unfolded state could be trapped by introducing an excess of complementary DNA sequence
(Figure: 2.19), allowing the measurement of the unfolding rate. The study also depends on the
complementary DNA hybridisation rate (khybridisation) which was found to be much faster than
G4 oligo refolding rate (kfold). The G4 unfolding therefore was the rate limiting step of the G4
oligo trapping process. Consequently, fluorescence signal increase of the FAM dye at 518 nm
is a surrogate for the rate of G4 unfolding due to the decrease of the FRET efficiency as G4
unfolds.
Figure 2.19: FAM/TAMRA dye oligo FRET system for studying G4 unfolding kinetics. A comple-
mentary oligo sequence can trap the G4 in an unfolded state by hybridisation and observe a FRET
efficiency change.
A two-phase exponential model fit the kinetic unfolding data substantially better than an one-
phase model. The model was fitted according to the following formula:
y = y0 + (A− y0)B(1− ekfastt)+
(A− y0)(1−B)(1− ekslowt)
Where y0 is the y value at time zero, A is the plateau y value at infinite times, kfast and kslow
are the two rate constants, B is the fraction of the span of the fast process component (from
y0 to plateau A).
The rate limiting step of the kinetic trapping process was the G4 unfolding, which was previously
reported [329]. This was confirmed by showing that the rate of G4 unfolding was independent
of the complementary strand concentration for multiple G4 structures. These results were
51
Chapter 2: Single-molecule G4 visualisation in live cells
independently validated for experiments presented here. For this, the unfolding experiments
were done with 1 µM of FRET G4 oligo by adding 10 µM or 100 µM concentration of the
complementary strand. No significant difference in unfolding rate was observed (Table: 2.4),
concluding that DNA strand hybridisation happens after the rate determining step in the
proposed kinetic model.
Kit1 10 µM complementary oligo 100 µM complementary oligo
Condition Slow t1/2
/ min
Fast t1/2
/ min
Percent
fast / %
Slow t1/2
/ min
Fast t1/2
/ min
Percent
fast / %
No ligand 238 ± 8 17 ± 3 34 ± 2 237 ± 7 16 ± 2 33 ± 1
SiR-PyPDS (19) 1 µM 296 ± 9 18 ± 2 31 ± 1 300 ± 7 18 ± 2 29 ± 1
Table 2.4: G4 unfolding rate was independent from complementary DNA concentration when in
excess. G4 two-phase unfolding kinetics were measured by introducing 10 µM or 100 µM of respective
complementary DNA oligonucleotide at t = 0 to trap the unfolded G4 oligonucleotide state. Data
presented are best fit of a two-phase association model. Plus minus values indicate the 95 % confidence
interval of the fit.
In preparation for experiments to determine G4 ligand effects on G4 unfolding rate, Kit1 and
H-telo G4 oligos were folded by heat annealing at 95 ◦C for 10 minutes, followed by slow cooling
down to 4 ◦C, and their fold state confirmed by following a characteristic FAM/TAMRA dye
FRET emission spectrum (Figure: 2.20, a, b). Then by introducing 10 µM final concentration
of the complementary oligo to folded G4s, the G4 unfolding kinetics were followed by measuring
the fluorescence signal from the FAM/TAMRA FRET system every 5 minutes from the 2nd
minute9. Comparing the change of the fluorescence spectra over time with and without the
presence of 1 µM SiR-PyPDS (19), it was evident that the ligand substantially slows down the
unfolding (Figure: 2.20, c, d). Following the FAM signal increase at 518 nm, the curve was
fitted to a double exponential kinetic model with good accuracy, R2 values 0.9993 and 0.9986
for no ligand and 1 µM SiR-PyPDS (19) conditions respectively (Figure: 2.20, e). Curiously,
the TAMRA signal at 580 nm decreased and then barely changed from the point when the
complementary oligo was introduced, possibly caused by the quenching of the dye. Finally,
a double FRET cascade from FAM to TAMRA to SiR was assessed as discussed previously
(Section: 2.3.6). The decrease of the SiR cascade FRET signal could be followed over the
course of the G4 unfolding (Figure: 2.20, f), suggesting that SiR-PyPDS (19) unbound from
the Kit1 G4 as it unfolded and hybridised to be become double stranded.
9The delay was required to allow for mixing, cuvette sealing and time required for the instrument to start
the run.
52
Chapter 2: Single-molecule G4 visualisation in live cells
a
500 520 540 560 580 600
0
50
100
150
200
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Kit1 FRET oligo emission spectrum b
520 540 560 580 600
0
50
100
150
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
H-telo FRET oligo emission spectrum
c
500 520 540 560 580 600
0
200
400
600
800
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Kit1 FRET oligo unfolding
1002 min
182 min
62 min
502 min
2 min
22 min
d
500 550 600
0
200
400
600
800
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Kit1 FRET oligo unfolding with 1 µM SiR-PyPDS
1002 min
182 min
62 min
502 min
2 min
22 min
e
0 500 1000
0
500
1000
Time / min
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Kit1 FRET oligo unfolding at 518 nm
SiR-PyPDS 1 µM
Two phase fit SiR-PyPDS
No ligand
Two phase fit no ligand
f
640 660 680 700
0
5
10
15
20
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
Decrease of SiR FRET upon Kit1 unfolding
1002 min
182 min
62 min
502 min
2 min
22 min
Figure 2.20: (a) Pre-kinetics-run emission spectra of Kit1 FRET oligo. (b) Pre-kinetics-run emission
spectra of H-telo FRET oligo. (c-d) Kit1 FRET oligo emission spectra by itself (c) and with SiR-
PyPDS (19) (d) after initiating unfolded state trapping with 10 µ complementary oligos. Coloured
lines indicate a spectra at particular time points after unfolding trapping initiation. (e) Fluorescence
intensity profile at 518 nm FAM signal over the time course of Kit1 G4 unfolding. (f) Doube FRET
cascade SiR fluorescence signal decrease as Kit1 G4 unfolded and SiR-PyPDS (19) became unbound.
Experiments were carried out with 1 µ of FRET oligonucleotide in K+ buffer (pH 7.4) at 20 ◦C.
Kinetic trapping experiments were done on Kit1, H-telo and Myc oligos with and without
the presence of SiR-PyPDS(19) or SiR-iPyPDS (20) (Table: 2.5). The biphasic nature of
fluorescence curve reveals a fast and a slow process in the kinetic trapping. The exact nature of
the biphasic behaviour can not be identified from these data, however, it may be hypothesised
that it arises from multiple populations of G4 fold states or is related to complementary oligo
interactions with the FAM/TAMRA FRET system, since a rapid TAMRA signal decrease was
observed. The fast process had its half-life reduced to a similar extent at greater SiR-PyPDS
53
Chapter 2: Single-molecule G4 visualisation in live cells
(19) and SiR-iPyPDS (20) concentrations for Kit1 and H-telo oligos, while the slow process
became slower for SiR-PyPDS (19) and had lesser changes for the SiR-iPyPDS (20) control
molecule. For Myc G4, both processes slowed down upon G4 ligand addition. Taken together,
I interpret the slow process as the G4 unfolding rate, and have avoided making interpretations
on the fast process due to its unclear origins and mechanism.
Kit1 (1 µM) unfolding
Condition Slow t1/2 / min Fast t1/2 / min Percent fast / % N
No ligand 234 ± 58 23 ± 12 46 ± 18 4
SiR-PyPDS (19) 20 nM 268 ± 47 20 ± 12 32.3 ± 4.7 4
SiR-PyPDS (19) 100 nM 283 15.2 28 1
SiR-PyPDS (19) 1 µM 331 ± 69 17.3 ± 3.0 30.0 ± 1.5 5
SiR-iPyPDS (20) 1 µM 300 ± 15 16.4 ± 3.8 30.0 ± 1.7 4
H-telo (1 µM) unfolding
No ligand 102 ± 16 9.6 ± 2.2 53.6 ± 1.5 2
SiR-PyPDS (19) 20 nM 106 ± 27 8.9 ± 1.5 54.5 ± 1.7 2
SiR-PyPDS (19) 100 nM 109 ± 29 9.1 ± 2.2 55.7 ± 1.7 2
SiR-PyPDS (19) 1 µM 119 ± 13 7.66 ± 0.61 53.89 ± 0.32 2
SiR-iPyPDS (20) 1 µM 101 ± 13 6.80 ± 0.11 57.4 ± 1.7 2
Myc (1 µM) unfolding
No ligand 1410 ± 190 17.2 ± 5.0 10.29 ± 0.21 2
SiR-PyPDS (19) 20 nM 1240 ± 140 21.0 ± 1.6 8.34 ± 0.50 2
SiR-PyPDS (19) 1 µM 2040 ± 280 48.5 ± 3.0 5.82 ± 0.22 2
Table 2.5: The effect on SiR-PyPDS (19) binding on unfolding kinetics of G4 DNA sequences in vitro.
G4 two-phase unfolding kinetics were measured by introducing 10 µM of respective complementary
DNA oligonucleotide at t=0 to trap the unfolded G4 oligonucleotide state. Data presented here are
of best fit of a two-phase association model. Errors indicate sd. N denotes the number of replicates.
G4 ligand binding slows down the G4 unfolding rate as reported previously [331, 332]. Im-
portantly, larger micromolar concentrations altered the global G4 unfolding dynamics while
smaller nanomolar concentrations gave a much less pronounced effect. Combined with these
findings with the G4 induction experiments (Section: 2.3.7), they demonstrate the requirement
of low probe concentrations to avoid significant global G4 landscape perturbations. Moreover, it
demonstrates the value of single-molecule imaging methods which enable the use of nanomolar
concentrations as a minimally perturbative tool for G4 studies.
54
Chapter 2: Single-molecule G4 visualisation in live cells
Section summary
I first described the synthesis of a novel fluorescent G4 probe PhenDC3-SiR (17). The ability
of this probe to selectively bind G4 structures versus other forms of DNA was tested by FRET
melting experiments. Further experiments then also included the G4 probe SiR-PyPDS (19).
The fluorescence properties of both molecules were characterised in terms of their absorption,
emission and quantum yield under different conditions. The probes G4 binding selectivity was
tested by fluorescence light-up titrations, with dissociation constants obtained for different G4
structures. A two-step FRET cascade was observed between FAM/TAMRA labelled G4 oligos
and SiR moiety on G4 ligands, and further validated their ability to bind to G4 structures. It
was next demonstrated that G4 ligands such as PDS, if used in excess, can cause perturbations
to G4 folding by inducing their formation, while lower concentrations avoid such population
scale effect. Finally, an excess of SiR-PyPDS (19) was shown to cause significant perturbation
of G4 dynamics by slowing down the unfolding rate of a G4 in kinetic trap experiments, while
low nanomolar concentrations showed a much less pronounced effect. Together, the induction
and G4 unfolding experiments demonstrate the value of using lowest possible concentrations
for detection of G4s as to avoid significant global G4 landscape perturbations.
2.4 TIRFM in vitro single-molecule imaging
In this section, I discuss experiments aimed to demonstrate that single G4s can be visualised
with confidence on an oligo-covered surface. Particular focus is drawn to control experiments
demonstrating probes binding selectivity to G4s. These controls, when transferred to a live cell
environment, would provide supporting evidence for probe cellular G4 selectivity versus other
nuclear features. Experiments mirroring in cells and in vitro are discussed in the following
sections.
The binding of G4 probes to oligos immobilised on a surface were imaged at single-molecule
resolution by total internal reflection fluorescence microscopy (TIRFM) as previously described
[333]. Biotinylated oligos were immobilised on glass coverslips, coated with PEG and NeutrA-
vidin. The oligos were also labelled with AlexaFluor 488 dye, to have a measure for control of
surface coverage10.
In this section, oligo coated surfaces for in vitro experiments were provided by Dr Rohan T.
Ranasinghe and TIRFM was carried out in collaboration with Dr Aleks Ponjavic.
10For oligo sequences see Materials and methods chapter (Section: 5.1.1).
55
Chapter 2: Single-molecule G4 visualisation in live cells
2.4.1 In vitro G4 binding validation
The oligos used in the study were validated for G4 folding by circular dichroism. A G4 structure
was confirmed by characteristic spectrum features for a parallel G4 fold (Figure: 2.21) - a
positive peak maxima at 265 nm and a negative peak minima at 245 nm [35, 36, 334–336]. A
Myc oligo mutant sequence, where four GGG tracts were replaced with GTG base sequences to
remove its ability to form a G4, showed less pronounced peak intensities at these wavelengths.
Figure 2.21: Circular Dichroism analysis of Myc oligonucleotides used for in vitro experiments: CD
spectra trace confirming folded state of the G4 Myc oligo used for surface binding experiments and
unfolded state of the mutant version Myc mut both used at 10 µM concentration and annealed in 100
mM K+ buffer pH 7.4, 20 ◦C. Carried out by Dr Marco Di Antonio and taken from [213].
SiR-PyPDS (19) could be visualised on a surface covered with Myc oligos via TIRFM (Figure:
2.22, a, d) whereas negligible binding seen on Myc mutant covered surfaces (Figure: 2.22, b,
e). The control molecule SiR-iPyPDS (20), a significantly weaker G4 binder, showed about an
order of magnitude less foci when binding to Myc oligo at an identical concentration (Figure:
2.22, c, f). Furthermore, the SiR-PyPDS (19) probe imaged after 250 pM treatment could
be displaced with 10 µM of another G4 ligand PhenDC3 (2) leading to a 98 % reduction in
foci number (Figure: 2.23, a, b). A similar result was obtained with PhenDC3-SiR (17) in an
excess of PDS (1) where a 95 % reduction in foci was observed. The displacement experiments
indicated that two different G4 ligands, PDS and PhenDC3, together with their SiR fluorophore
probe counterparts can both target and compete for the same G4 oligo binding sites.
Since, G4 ligands have been reported to be able to quench fluorophore dyes [86], it was important
to check whether PDS (1) or PhenDC3 (2) used for displacement could quench SiR dye at large
excess. For this, a series of titrations were carried out, by adding increasing amounts of G4
ligands to SiR-PyPDS (19) and PhenDC3-SiR (17) solutions. A decrease of fluorescence of
SiR-PyPDS (19) was initially observed upon titration with PhenDC3 (2) in PBS buffer (Figure:
2.24, a). On the other hand, when the titration was done with 0.1 % SDS and using a lower
56
Chapter 2: Single-molecule G4 visualisation in live cells
a b c
Figure 2.22: (a) Schematic of the methodology used for visualizing individual G4s. Pre-folded
G4 Myc was attached to a coverslip via a biotin-neutravidin linker. The fluorescent G4 probe SiR-
PyPDS (19) binds to G4 Myc, and is visualised using single-molecule fluorescence imaging. (b)
SiR-PyPDS (19) does not bind single-stranded mutated-Myc that cannot form a G4. (c) The inactive
isomer SiR-iPyPDS (20), with its 10-fold weaker binding affinity, is less likely to bind G4 Myc. (d)
Representative images (500-ms exposure) of individual SiR-PyPDS (19) molecules (250 pM) binding
to a surface coated with pre-folded Myc G4 oligonucleotide. Individual fluorescent puncta indicate
binding of single SiR-PyPDS (19) molecules. (e) SiR-PyPDS (19) (250 pM) binding to mutated-Myc.
(f) 250 pM SiR-iPyPDS (20) binding to pre-folded Myc. Experiments in d-f were repeated three times
independently with similar results.
57
Chapter 2: Single-molecule G4 visualisation in live cells
10 nM concentration of SiR-PyPDS (19)11, the opposite trend was observed with an increase
of fluorescence at larger PhenDC3 (2) concentrations (Figure: 2.24, b). These conflicting
observations were suggestive that a complex array of interactions and competing effects could
be in play here. However, since SiR-PyPDS (19) is mostly in its ON fluorescence state when
bound to G4s it was regarded that 0.1 % SDS buffer condition more closely resembles probe
displacement experimental conditions when the dye is highly fluorescent. Finally, titration of
both SiR-PyPDS (19) and PhenDC3 (2) with PDS (1) in PBS buffer leads to a significant
increase in fluorescence signal (Figure: 2.24, c, d), therefore dye quenching was regarded as
negligible in these displacement experiments.
c
Figure 2.23: (a) Representative image of SiR-PyPDS (19) single-molecules visualised on a Myc
G4 oligo covered surface prior displacement with PhenDC3. (b) After displacement with 10 µM
PhenDC3. (c) Quantification of SiR-PyPDS (19) binding to the G4 Myc, SiR-PyPDS (19) binding
to the mutated-Myc (Mut), SiR-iPyPDS (20) binding to the G4 Myc and SiR-PyPDS (19) binding
to the G4 Myc in the presence of 10 µM unlabelled PhenDC3 competitor. Error bars indicate mean ±
sd. P < 1 · 10−5, unpaired two-sided t-test, n = 12 measurements from three independent replicates.
The quantitative in vitro data is summarised (Figure: 2.23, c). Taken together, these results
were indicative of SiR-PyPDS (19) specific binding to a folded Myc G4 oligo structure and
further validated SiR-iPyPDS (20) as a useful negative control molecule. In vitro TIRFM
experiments were also carried out with other G4 oligos - Kit1 and H-telo. A similar trend was
seen as with Myc with about an order of magnitude more binding events for SiR-PyPDS (19)
than for SiR-iPyPDS (20). This gives further evidence for SiR-PyPDS (19) binding ability to
different G4 structures (Figure: 2.25).
11As shown earlier (Figure: 2.6), 0.1 % SDS addition switches the fluorogenic SiR dye to its fluorescent ON
state, allowing (19) fluorescence to be detected at much smaller concentrations and therefore a larger excess of
PhenDC3 (2) concentration could have been tested.
58
Chapter 2: Single-molecule G4 visualisation in live cells
a
640 660 680 700 720 740
0
200
400
600
800
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
SiR-PyPDS titrated with PhenDC3 in PBS
SiR-PyPDS 1 µM
+ PhenDC3 1 µM
+ PhenDC3 10 µM
+ PhenDC3 100 µM
+ PhenDC3 1000 µM
b
640 660 680 700 720 740
0
500
1000
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
SiR-PyPDS titration with PhenDC3 in PBS 0.1% SDS
SiR-PyPDS 10 nM
+ PhenDC3 20 nM
+ PhenDC3 100 nM
+ PhenDC3 1 µM
+ PhenDC3 10 µM
+ PhenDC3 100 µM
c
640 660 680 700 720 740
0
500
1000
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
SiR-PyPDS titrated with PDS in PBS buffer
SiR-PyPDS 1 µM
+ PDS 1 µM
+ PDS 10 µM
+ PDS 100 µM
+ PDS 1000 µM
d
640 660 680 700 720 740
0
500
1000
Wavelength / nm
Fl
uo
re
sc
en
ce
 in
te
ns
ity
PhenDC3-SiR titration with PDS in PBS
PhenDC3-SiR 1 µM
+ PDS 1 µM
+ PDS 10 µM
+ PDS 100 µM
+ PDS 1000 µM
Figure 2.24: Fluorescent G4 probe titrations with PhenDC3 (2) and PDS (1). The presence of other
small molecules does influence the fluorescence properties of SiR-PyPDS (19) and PhenDC3-SiR (17).
In b-d, at highest small molecule concentrations, the fluorescence intensity saturated the capabilities
of the detector. Experiments were carried out in pH 7.4 buffer at 20 ◦C.
Figure 2.25: Binding of SiR-PyPDS to different G4 structures in vitro. The number of binding events
observed for SiR-PyPDS and SiR-iPyPDS binding to G4s in vitro varies for different G4 sequences
(250 pM concentration of both probes in all cases except for Myc where it was 25 pM). Error bars
indicate mean ± sd. n = 6 measurements for each condition.
59
Chapter 2: Single-molecule G4 visualisation in live cells
2.4.2 A case for single-molecule, single-G4 binding
A single-step photobleaching experiment was used to confirm the single-molecule nature of G4
visualisation in vitro using SiR-PyPDS (19) (Figure: 2.26). Here, the fluorescence intensity
coming from the SiR dye could be seen to go up and down in single-steps as foci visualised on
the surface appeared/disappeared, therefore validating single-molecule binding.
Figure 2.26: Single-step photobleaching confirms detection of individual probes. (a) 25 pM SiR-
PyPDS (19) binding to Myc in vitro. The blue, red and green squares indicate a single binding
event. (b) Intensity traces from the three binding events in (a), showing probes undergoing single-
step photobleaching. Similar single-step photobleaching could be consistently observed in all single-
molecule video acquisitions.
A case for single-G4 visualisation can also be made by observation of smFRET between SiR dye
and Alexa Fluor 488 dye on the Myc oligo. Initially, FRET between the dyes was observed in
bulk experiment where after irradiation at 488 nm the Alexa Fluor 488 dye signal decreased with
increasing SiR-PyPDS (19) stoichiometric ratio while SiR dye signal was increasing (Figure:
2.27, a). An identical FRET donor/acceptor relationship was seen at a single-molecule level,
where Alexa Fluor 488 donor signal had increased in single steps as the SiR acceptor was
bleached by 635 nm excitation (Figure: 2.27, b - d). Such single-step FRET behaviour supports
an interaction between two single-molecules.
60
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.27: FRET between SiR-PyPDS (19) and Alexa Fluor 488-labelled Myc confirms direct
binding to G4s. (a) Emission spectrum of 488-Myc-G4 at 1 µM and SiR-PyPDS (19) at various stoi-
chiometric ratios. As the probe concentration increases, donor emission drops and acceptor emission
increases, indicating FRET. (b) In vitro G4 FRET experiment. 250 pM of SiR-PyPDS (19) (shown
in red with acceptor excitation) interacting with Alexa Fluor 488-labelled Myc-G4 (with 1 % surface
coverage). The green channel shows acceptor emission under donor excitation. FRET between Myc
and SiR-PyPDS (19) is highlighted with white arrows. (c) 10 nM SiR-PyPDS (19) interacting with
488-Myc-G4 (0.001 % surface coverage). Temporal intensity traces of donor (green) and acceptor (red)
emission under donor excitation. Anti-correlated intensity fluctuation upon acceptor photobleaching
indicates single-molecule FRET between SiR-PyPDS (19) and Myc. (d) Example time lapse of ac-
ceptor (top, red) and donor (bottom, green) emission from (c). Experiments a-d were performed as 3
independent replicates all providing similar results.
2.4.3 Measurement of G4 abundance
It was of interest to investigate whether single-molecule binding events could be used as a proxy
for G4 abundance12 on a surface, which would later be useful for comparing G4 levels in differ-
ently treated cells. It was found that the count of binding events was linearly dependent with
SiR-PyPDS (19) concentration in the concentration range tested (Figure: 2.28, a), suggesting
sub-labelling conditions, i.e. G4 labelling with SiR-PyPDS (19) has not reached saturation.
Then Myc G4 oligo abundance on the surface was systematically reduced by mixing Myc-biotin
oligo with a corresponding fraction of biotin with an ssDNA oligo attached which cannot form
a G4 structure, this way introducing a competition between biotinylated species for neutravidin
binding sites13.
12I.e. total amount of G4s in a system.
13To achieve a differently covered surface by dilution of Myc-biotin oligo, the method would have required
finding a concentration range where neutravidin binding sites on the surface would not be occupied to saturation.
61
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.28: Single-molecule imaging with SiR-PyPDS can be used to quantify Myc-G4 abundance
in vitro. (a) Number of detected binding events increases with probe concentrations. (b) Myc flu-
orescence showing that the concentration of Myc on the surface can be controlled by mixing with
a competing biotinylated oligomer. (c) Number of detected events increases with G4 concentration,
tested at 25 pM of SiR-PyPDS (19). Sample images for each condition is shown beneath each plot.
Error bars indicate mean ± sd. n = 12 measurements taken from 2 independent replicates.
Myc coverage levels on the surface were measured by AF488 fluorescence signal intensity, which
was also found to have a linear relationship (Figure: 2.28, b). Finally, SiR-PyPDS (19) binding
events were measured on these surfaces with different levels of G4 coverage, which showed that
observed events depend linearly on Myc G4 surface coverage (Figure: 2.28, c). Together, these
results validate that single-molecule binding event count can give a measure of relative G4
abundance levels even at sub-labelling conditions.
Section summary
In vitro single-molecule TIRFM experiments showed that G4 probes can be used to specifically
visualise different G4 structures on a surface. The number of binding events was also shown
to be a proxy for G4 density measurement. These experiments also validated G4 binding
specificity by testing the imaging platform on G4 forming and non-G4 forming oligo sequences,
by employing a negative control molecule SiR-iPyPDS (20) and by displacement experiments
using a non-fluorescent G4 ligand competitor. Finally, single-probe-molecule and single-G4
binding was validated by single-step photobleaching experiments and smFRET.
62
Chapter 2: Single-molecule G4 visualisation in live cells
2.5 Single-molecule imaging in cells
In this section, live cell visualisation of G4s will be discussed. A series of control experiments
will be introduced which were also mirrored in vitro to support the validity of single-molecule
G4 visualisation in terms of its selectivity towards G4 structures. In vitro controls were helpful
for building an in depth understanding of the G4-related processes we were studying in a simpler
system than a live human cell. Furthermore, the first evidence on G4s lifetime and dynamics
in live cells will be presented. Finally, G4 cell-cycle dependence is explored together with G4
relation to transcription and replication.
2.5.1 SiR-PyPDS and SiR-iPyPDS in live cells
SiR-PyPDS and SiR-iPyPDS comparison
The fluorescent G4 probes were utilised for visualising G4s in live U2OS cells, which were picked
because of their robustness and characterisation of G4 structures [59, 62] (Figure: 2.29, a). Both
SiR-PyPDS (19) and SiR-iPyPDS (20) were found to be internalised into the cells over a 30
minute treatment at 20 nM concentration to be later observed as single-molecules in the nucleus
by HILO imaging technique. With SiR-PyPDS (19), an order of magnitude more binding events
were observed than for the control molecule SiR-iPyPDS (20) (Figure: 2.29, b - d), as observed
in vitro (Figure: 2.25). Combined, these results served as an initial suggestion for probes G4
selectivity in cells. To ensure that this observation was not a result of different cellular uptake of
the two probes, their fluorescence per area was compared under a confocal microscope after 10
µM treatments of both probes. It was found that SiR-PyPDS (19) had a slightly lower nuclear
accumulation than SiR-iPyPDS (20), therefore the control molecule SiR-iPyPDS (20 showed
fewer binding events because of its reduced binding ability to G4s and not due to differences
in nuclear accumulation14 (Figure: 2.30).
The probes accumulate in the lysosomes
Bright spots of SiR-PyPDS (19) and SiR-iPyPDS (20) accumulation were visible just outside
the nucleus where fluorescence signal was too strong for detection of single-molecule binding
events. Their spherical shape and slow movement in the imaging videos suggested that these
might be cellular vesicles. Indeed, co-localisation imaging of lysosomes with LysoTracker in the
14Since SiR-iPyPDS (20) accumulated 35 % more, but showed 97 % less binding events than SiR-PyPDS
(19).
63
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.29: Single-molecule fluorescence imaging of G-quadruplexes in living cells using the fluo-
rescent probe SiR-PyPDS (19). (a) Schematic of G4s in the cell nucleus with a zoom-in showing G4s
stained by SiR-PyPDS (19). (b) Representative background-subtracted image (maximum projection
of 100 frames with 200 ms exposure) of SiR-PyPDS (19) binding events in a living U2OS cell treated
with 20 nM SiR-PyPDS (19) for at least 30 min before imaging; fluorescent puncta indicate binding of
single SiR-PyPDS (19) molecules. Blue color corresponds to nuclear staining with Hoechst. (c) Repre-
sentative image of SiR-iPyPDS (20) staining in living U2OS cell treated with 20 nM SiR-PyPDS (19)
for at least 30 min before imaging. (d) Quantification of the binding events within the nucleus lasting
more than one frame (100 ms per frame) per cell for SiR-PyPDS (19) and SiR-iPyPDS (20). Center
lines indicate the median; boxes show interquartile range; whiskers denote 5th and 95th percentiles.
*** P < 10−5, Mann-Whitney U-test.
green channel and of SiR-PyPDS (19) in the red channel confirmed that the majority of the
probe accumulates in lysosomes (Figure: 2.31). High lysosome brightness is not a surprise, since
they contain a low pH environment and SiR dye switches to the ON fluorogenic state in high
acid, as was discussed previously (Figure: 2.10). Therefore, it was the expectation that SiR
dye would be pre-dominantly in the ON state in the lysosomes, while elsewhere in the cell the
probe would be switching between ON and OFF states and allowing single-molecule detection.
Moreover, switching to the zwitterionic ON fluorescence state of SiR as well as protonation of
SiR-PyPDS (19) could make the probe less permeable through the membrane of the lysosome,
trapping the probe there and this way potentially favouring accumulation.
64
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.30: Total nuclear accumulation of SiR-PyPDS (19) and SiR-iPyPDS (20) in U2OS cells.
Total fluorescence intensity measured inside the nuclei of >300 U2OS cells after incubation with 10
µM SiR-PyPDS (19) or SiR-iPyPDS (20) by standard confocal microscopy at 633 nm. each point on
the graph represents the total fluorescence of SiR measured at 633 nm per nuclei, data are plotted
as the mean of >300 nuclei measured in 3 independent replicates. Total fluorescence measurement
revealed comparable ability of the two molecules to accumulate in the nuclei. Error bars indicate
mean ± sd. Carried out by Dr Marco di Antonio and taken from [213].
SiR-PyPDS does not cause lysosome activation
SiR-PyPDS (19) accumulation in the lysosomes raises the question of whether the probe might
be inducing formation of lysosomes and their activation. Lysosomal activation is a cellular
response linked to both autophagy and apoptosis [337, 338], which may become triggered by
compound administration. To test this, live U2OS cells had their lysosomes stained by 50 nM
of LysoTracker Green with and without the addition of G4 ligands. Signal-to-background ratio
(SBR) of the fluorescence intensity of the lysosomes was used to measure for the relative levels
of lysosome staining abundance, what can act as a measure of lysosome activity [337]. Using
SBR ensured that effects of sample thickness, photobleaching, small concentration differences
or any other variations were minimised in the data analysis. No lysosomal activation was
observed under our standard single-molecule imaging conditions (Figure: 2.32), and, neither
longer treatments nor higher concentration of a G4 ligand lead to lysosome activation. These
results showed that although the SiR-PyPDS (19) probe did accumulate in the lysosomes, it
did not trigger a lysosomal activation cellular response and supported our aim to not introduce
significant cellular perturbation when carrying out single-molecule G4 imaging.
65
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.31: SiR-PyPDS mainly accumulates in lysosomes. Representative confocal and HILO
microscopy images obtained in the presence of SiR-PyPDS (1 µM in confocal and 40 nM in HILO)
and LysoTracker Green (50 nM), confirming co-localisation of extra-nuclear staining with lysosomes.
Experiments have been repeated 3 times providing similar results. HILO imaging carried out in
collaboration with Dr Lisa-Maria Needham. Confocal images provided by Dr Marco di Antonio.
Taken from [213].
Figure 2.32: Lysosomes are unaffected by short duration treatment with SiR-PyPDS. LHS - Rep-
resentative HILO microscopy images of U2OS cells stained with 50 nM Lysotracker Green. White
arrows point to representative lysosomes, while orange ellipses indicate the nucleus. RHS - Signal-to-
background-ratio (SBR) of the fluorescence intensity in lysosomes does not change significantly upon
treatment with 40 nM SiR-PyPDS. Each data point represents the average SBR in a field of view with
at least one cell. Error bars indicate mean ± sd. HILO imaging carried out in collaboration with Dr
Lisa-Maria Needham.
66
Chapter 2: Single-molecule G4 visualisation in live cells
Cell viability study
The G4 probes effects on cellular viability were investigated. It was found that nanomolar
probe concentrations used for single-molecule imaging did not inhibit cell growth of U2OS
cells (Figure: 2.33)15. An effect on cell growth was only observed at probe concentrations
above 10 µM. It was noticeable that SiR-PyPDS (19) may have a slightly greater effect on cell
growth inhibition than SiR-iPyPDS (20), which may be reflective of the weaker G4 binding
ability of the latter.
These results confirmed that no gross cell stress was induced by the single-molecule imaging
conditions over 24 hours, suggesting that G4 visualisation can be performed with minimal
cellular perturbation.
Figure 2.33: Growth inhibition curves obtained with SiR-PyPDS (left) and SiRiPyPDS (right)
in U2OS cells. Growth inhibition studies indicate that no significant cellular toxicity is elicited by
treatment of either SiR-PyPDS or SiR-iPyPDS over a 24 h treatment for doses up to 10 µM. Error
bars indicate mean ± sd. n = 4 measurements taken from 4 independent replicates of each treatment
condition. Carried out by Jiazhen Shen. Taken from [213].
2.5.2 PhenDC3-SiR in live cells
PhenDC3-SiR (17) was imaged in live U2OS cells and was observed in the nucleus (Fig-
ure: 2.34, a) using HILO microscopy16. Just like with SiR-PyPDS (19) live cell imaging,
the bright white-yellow regions around the nucleus were mostly stained lysosomes, where SiR
15Cell viability experiments were carried out by Jiazhen Shen.
16Imaging experimental procedures were used as reported in the literature [333].
67
Chapter 2: Single-molecule G4 visualisation in live cells
dye switches to the ON fluorogenic state in a low pH environment. Preferential accumula-
tion in the lysosomes could also be a feature of both probes. Unfortunately, bright lysosomes
were problematic for single-molecule imaging of PhenDC3-SiR (17), much more so than for
SiR-PyPDS (19), as it appeared that the ratio between lysosome brightness and signal from
the nucleus was much higher, i.e. the bright regions were essentially blinding the fluorescence
detector, making it difficult to obtain good image contrast in the nucleus for single-molecule
detection. Testing 1 nM - 5 µM concentration range, varying treatment times, different washing
sequences did not help to achieve good single-molecule detection in the nucleus using HILO
technique. Consequently, it was decided to try using light-sheet microscopy, which can achieve
better signal-to-noise ratio.
Figure 2.34: Representative background subtracted images of live U2OS cells. Blue dashed lines
indicate the location of the nucleus. (a) HILO fluorescence microscopy image (maximum projection
of 1000 frames with 33 ms exposure) after treating with 2.5 µM PhenDC3-SiR (17) for 1 h followed
by 2 washes with growth media over 20 min and 3 washes with PBS. (b) Light-sheet fluorescence
microscopy image (maximum projection of 200 frames with 200 ms exposure) after 1 µM PhenDC3-
SiR (17) treatment for 30 min followed by 3 washes with PBS.
Although light-sheet presents a more complicated setup, the technique allows irradiation on a
smaller volume within the cell when compared to HILO, enabling reduction of background noise
stemming from irrelevant parts of the cell as well as out-of-focus planes in the nucleus. Indeed,
light-sheet imaging produced significantly better results and clear single-molecule localisations
were observed in live U2OS nuclei (Figure: 2.34, b). However, due to technical difficulties
related to the complex nature of light-sheet microscopy setup and due to time restraints of the
project, it was decided not to pursue further leads using PhenDC3-SiR (17) probe. Instead
focus was on using SiR-PyPDS (19) in HILO mode for further experiments. Moreover, the
68
Chapter 2: Single-molecule G4 visualisation in live cells
nanomolar concentrations of SiR-PyPDS (19) used for labelling were more suitable than the
micromolar concentrations used for PhenDC3-SiR (17) as to enable minimal effect of the probe
on cells.
2.5.3 Single-molecule observation and binding event residency times
For SiR-PyPDS (19), imaging of single-molecule binding events in live cells was confirmed
by a single-step photobleaching experiment (Figure: 2.35). Since fluorescence signal intensity
increased and decreased in a single-step binding event, it showed that the signal was emitted
by a single fluorophore. However, this does not necessarily correspond to the observation of a
single G4. Consider this example - the criteria chosen to classify G4 binding events were the
observation of a bright focus for at least two frames at 100 ms exposure i.e. a focus point must
be observed for a minimum duration to qualify as a binding event. In principle, binding to a
single G4 might be too short in order to be counted under such criteria. However, if the probe
resided in a cluster of G4s, then it could migrate back and forth among multiple G4s, this
way compounding the residency lifetimes and making them be long enough to be counted as
observed binding events. Such movement of the probe from one G4 to the next within a cluster,
could not be registered as separate events due to limits of resolution in a diffraction-limited
microscopy method. Then, in this example, only G4 clusters would be imaged and not single
G4s.
Figure 2.35: Single-step photobleaching confirms detection of individual probes in cells. (a) 20 nM
SiR-PyPDS binding to targets in a living cell. The red square indicates a single binding event. (b)
Intensity traces from three binding events in a, showing probes undergoing single-step photobleaching.
Similar single-step photobleaching could be observed in all single-molecule video acquisitions.
69
Chapter 2: Single-molecule G4 visualisation in live cells
A potential way to investigate whether single G4s were observed would be to measure binding
event residency times in live cells and in vitro for comparison. Moreover, probe residency times
could also give a better understanding of binding behaviour especially if long dwell events
could be more confidently identified as G4s. Time-lapse imaging was used to observe long-lived
events in vitro and in cells while using cycle times of 2 s and 3 s respectively to minimise
photobleaching effects (Figure: 2.36). A slightly longer cycle time and a longer exposure was
used for measurement in cells to reduce the contributions from the unbound ligands [339] and
photobleaching correction was implemented as reported previously [340].
Figure 2.36: (a) Single-molecule time-lapse imaging of SiR-PyPDS (19) in vitro (top) and in cells
(bottom). Individual images from the time-lapse stack are shown on the left, while kymographs on the
right show the dynamic binding kinetics of SiR-PyPDS (19) to G4s. The experiments were repeated
three times independently, with similar results. (b) The histograms of dwell times for each experiment
(three positions on a coverslip for in vitro experiments and six cells for the cell experiment) were fitted
with a single exponential fit to determine the binding lifetime in each condition.
A stark difference in residency times in cells (τ = 6.6 ± 0.5 s) versus Myc G4 in vitro (τ = 15.4 ±
0.6) was observed. Although, the disparity could arise from the different physical environments
in cells and in vitro, it was further investigated whether increased probe binding time to Myc
G4 could be an outlier. Indeed, additional experiments on Kit1 and H-telo G4 oligo covered
surfaces showed 2.5 times shorter residency times of SiR-PyPDS (19) than to Myc G4 (Table:
2.6), when imaged with shorter cycle times to explore faster binding dynamics. However, SiR-
PyPDS (19) binding residency time to Myc mutant oligo sequence, although shorter than for
Myc G4 binding, was similar to Kit1 and H-telo residency times, suggesting that residency
70
Chapter 2: Single-molecule G4 visualisation in live cells
time alone is not able to distinguish G4 structures. On the other hand, control molecule SiR-
iPyPDS (20) showed two and four-fold shorter residency times than SiR-PyPDS (19) when
measured for both long and short binding events respectively (with cycle times of 2 s and 0.1 s),
suggesting that longer lasting binding events are more likely to be attributed to G4s. Therefore,
filtering for longer-lasting binding events may provide confidence for selectively observing G4s.
Finally, residency time of PhenDC3-SiR (17) was two-fold shorter than for SiR-PyPDS (19)
on Myc G4 in vitro, suggesting that binding behaviour is primarily probe-dependent and does
not necessarily reflect single-G4 dynamics.
Ligand Condition τ / s Exposure / ms Cycle time / s
SiR-PyPDS (19) In cells 6.6 500 3
Myc G4 15.4 100 2
Myc G4 1.0 100 0.1
Myc Mut 0.40 100 0.1
H-telo G4 0.42 100 0.1
Kit1 G4 0.49 100 0.1
SiR-iPyPDS (20) Myc G4 8.1 100 2
Myc G4 0.23 100 0.1
PhenDC3-SiR (17) Myc G4 8.1 100 2
Table 2.6: Summary of photobleaching corrected residency times of SiR-PyPDS (19), SiR-iPyPDS
(20) and PhenDC3-SiR (17) binding in cells and to oligo sequences in vitro. Different cycle times
were used to explore fast and slow binding dynamics.
For a question whether single-G4s were imaged in cells, the residency times observed can be
used to give a clearer picture. Single-G4 observation in vitro was previously demonstrated
by smFRET (Figure: 2.27), therefore residency times observed to different G4s in vitro do
reflect single probe binding dynamics to a single G4. In cells single-G4 observation could be
implied from residency times comparison, under the assumption that SiR-PyPDS (19) binding
dynamics in cells and in vitro do not differ by orders of magnitude due to different physical
environments of the systems in question. Consequently, if multiple G4 binding events within
a cluster were to be visualised as a single focus, one would expect longer residency times for
a cluster than for a single-G4. The fact that the residency time measured in cells was not
significantly longer than in vitro (actually being shorter when compared to Myc), suggest that
the residency was not compounded by binding multiple G4s sequentially within a cluster and
likely arises from single G4 binding.
The above reasoning is speculative and is only suggestive of the true nature of G4s imaged by
this method. If the majority of G4s do actually form and reside in clusters, then probabilistically
71
Chapter 2: Single-molecule G4 visualisation in live cells
majority of single-G4 visualisations would be of those within clusters. The question of whether
G4s cluster in space could become important in search of their mechanism of action and function
in cell biology. However, different techniques need to be sought in order to answer it.
2.5.4 Pre-blocking of G4 binding sites control experiments
To further validate that SiR-PyPDS (19) single-molecule binding events recognise G4 struc-
tures, pre-blocking of G4 sites experiments were performed followed by imaging. Live U2OS
cells were pre-treated with 10 µM of unlabelled PDS (19) or PhenDC3 (2) to pre-block G4
binding sites before standard SiR-PyPDS (19) treatment for single-molecule imaging. Pre-
blocking led to an order of magnitude reduction of observed binding events for both G4 ligands
(Figure: 2.37), signifying that SiR-PyPDS (19) probe primarily binds to the same targets in
the nucleus as non-fluorescent G4 ligands. This control experiment mirrored the G4 ligand
competition in vitro (Figure: 2.23) leading to a similar conclusion that SiR-PyPDS (19) was
G4-selective.
Figure 2.37: Cellular pre-blocking experiments of SiR-PyPDS (19) with the established G4-ligands
PDS (1) and PhenDC3 (2). SiR-PyPDS (19) competed in cells with 10 µM of unlabelled G4-ligands
PDS (1) and PhenDC3 (2). Cells were pre-incubated 30 minutes with PDS (1) or PhenDC3 (2)
at 10 µM prior standard single-molecule imaging with SiR-PyPDS (19). Each point on the graph
depicts the number of long-lived SiR-PyPDS (19) events measured in independent replicates. Data
are plotted as the mean of 3 or more independent replicates. Error bars indicate mean ± sd. * P
< 0.05, two-sided Mann-Whitney U-test. n = 5, 3 and 3 measurements taken from 3 independent
replicates for no displacement, PDS (1) displacement and PhenDC3 (2) displacement respectively.
Pre-blocking experiment and analysis was carried out by Dr Marco di Antonio and Dr Aleks Ponjavic.
Taken from [213].
72
Chapter 2: Single-molecule G4 visualisation in live cells
2.5.5 G4 folding is dynamic in live cells
The formation of G-tetrads hinges on the ability of the N7 of guanine to act as an acceptor
in Hoogsteen base-pairing/hydrogen bonding (Figure: 1.1, a). By disrupting the N7 ability
to act as a hydrogen acceptor, guanines no longer possess an ability to form a G-tetrad and
consequently a G4. Such alteration is achievable by use of a DNA-methylating agent dimethyl
sulphate (DMS), which can preferentially methylate the nucleophillic N7 on guanines [341].
The N7 that participates in H-bonding is essentially protected from methylation when a G4 is
folded. The N7 of guanine is exposed when the G4 unfolds into an ssDNA or dsDNA state.
Subsequently alkylation by DMS results in the chemical trapping of the unfolded G4 state.
(Figure: 2.38, a). Myc G4 oligo treatment with 600 mM DMS for 5 minutes followed by 10 %
β-mercapto-ethanol quenching led to 87 % reduction of observed binding events on Myc G4
covered surface in vitro (Figure: 2.38, b), demonstrating the ability of DMS to trap the unfolded
G4 state. The experiment was next performed in cells where 20 mM DMS concentration was
used to prevent cell death during the experiment. A time-dependent SiR-PyPDS (19) foci
decrease was observed with 95 % reduction after 20 min of DMS treatment (Figure: 2.38, c).
This result suggests that G4s are dynamic in live cells and can undergo unfolding over a time
course of 20 minutes.
Figure 2.38: (a) Schematic of DMS-meditated chemical trapping of unfolded G4s. (b) Quantification
of G4-binding events for untreated and 600 mM DMS-treated G4 Myc for 20 min. Error bars indicate
mean ± sd. *P = 0.05, two-sided Mann-Whitney U-test. n = 3 (Myc) and n = 4 (DMS) measurements
taken from three independent replicates. (c) Quantification of G4-binding events detected in living cells
upon increased exposure to DMS (20 mM), showing a clear time-dependent depletion of G4s. Centre
lines indicate the median; boxes show interquartile range; whiskers denote 5th and 95th percentiles.
**P < 0.01, two-sided Mann-Whitney U-test. n = 5, 6, 7 and 8 cells for untreated, 5 min, 10 min and
20 min, respectively, taken from three independent replicates. Experiment and analysis were carried
out by Dr Marco di Antonio and Dr Aleks Ponjavic. Taken and adapted from [213].
73
Chapter 2: Single-molecule G4 visualisation in live cells
2.5.6 G4 cell cycle dependence
Single-molecule imaging was next used to investigate relative levels of G4 abundance over the
course of the cell cycle. There was an open question whether G4 levels change during active
DNA processing states in live cells and whether single-molecule imaging could confirm previous
observations made by fixed cell G4 imaging using antibodies [59]. Highest G4 prevalence was
detected in S cell cycle synchronised cells (208 events over 40 s of imaging), with a two-fold
decrease at G1/S phase (103 events) and negligible number of G4 localisations were observed in
G0/G1 phase (3 events) (Figure: 2.39). Taken together, the results indicated that G4 prevalence
was higher in cell cycles where DNA processing such as replication and transcription were more
active, namely in S and G1/S cell cycle phases, while G4 prevalence was severely reduced
during G0/G1 phase when DNA processing is primarily deactivated. It was reasoned that
opening of dsDNA to ssDNA during transcription and replication could enable G4 formation,
which requires ssDNA.
Figure 2.39: The observation of G4s in live cells is altered by cell-cycle phase and transcription. (a -
d) Representative single-molecule images of G4-binding events are shown for synchronized U2OS cells
in the S phase (a), G1/S phase (b) and G0/G1 phase (c) and for unsynchronized cells treated with
both the transcriptional inhibitor DRB and the replication inhibitor aphidicolin (d). Experiments
a - d were repeated three times independently, with similar results. (e) Quantification of binding
events lasting more than two frames (100 ms per frame) per cell in living U2OS cells at different
cell-cycle phases and after transcription/replication arrest. Centre lines indicate the median; boxes
show interquartile range; whiskers denote 5th and 95th percentiles. ***P < 1 · 106; *P = 0.01; NS,
P = 0.99; two-sided Mann-Whitney U test. n = 18, 19, 19 and 15 cells for S, G0, G1 and arrest,
respectively, taken from three independent replicates. Experiment and analysis were carried out by
Dr Marco di Antonio and Dr Aleks Ponjavic. Taken from [213].
74
Chapter 2: Single-molecule G4 visualisation in live cells
To further test G4 prevalence relationship with transcription and replication, unsynchronised
U2OS cells were treated with replication inhibitor amphidicolin and transcription inhibitor
5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB), as previously described [54], conditions
mimicking DNA processing state in the G0 phase of the cell cycle. Only a few binding events of
SiR-PyPDS (19) were observed (3 events) under transcription and replication arrest, showing
G4 link to these processes (Figure: 2.39, d, e).
To check whether different probe uptake at cell cycle phases could influence the measured G4
prevalence changes, total nuclear accumulation of SiR-PyPDS (19) was measured by confocal
microscopy (Figure: 2.30). The differences in uptake were negligible for unsynchronised and
G0 cells, as well as for transcription and replication arrest conditions.
Cell-cycle dependent changes in levels of G4 abundance were measured in live cells via single
molecule imaging. The observations agreed with previously reported results observed in fixed
cells, where G4 selective antibody BG4 or small molecule IMT was used in fluorescence imaging
[59, 205]. Since three independent probes and orthogonal approaches led to similar conclusions,
this gives confidence in the reliability of the single-molecule imaging method.
2.5.7 Single-molecule imaging in fixed cells
The visualisation of G4s in fixed U2OS cells was also attempted. The aim was to measure G4
dynamics after live cell processes were halted and to compare results with observations made
previously in live cells. SiR-PyPDS (19) and its isomeric control molecule SiR-iPyPDS (20)
were both observed to label the nucleus (Figure: 2.40). As expected, SiR-iPyPDS (20) showed
less foci, however, the difference to SiR-PyPDS (19) was less compared to live cells. Lysosomes
were no longer strongly labelled with the probes in fixed cells, suggesting that fewer lysosomes
were present, or that their pH was increased and no longer highly acidic.
Other control experiments in fixed cells were less successful. After formaldehyde fixation,
although SiR-iPyPDS (20) labelling led to 80 % drop in binding events when compared to SiR-
PyPDS (19), pre-blocking with 10 µM of PDS or PhenDC3 led to only 54 % and 13 % drop
respectively (Figure: 2.41, a). Treating fixed cells with 600 mM DMS, prior to (19) labelling,
led to an unexpected increase in tracked events17. It was hypothesised that an increase of SiR-
PyPDS (19) non-specific binding could overwhelm any signal coming from specific binding to
G4s, therefore the majority of tracked binding events would arise from unwanted noise. Conse-
quently, a range of labelling, washing protocols and blocking conditions were tested in attempts
to minimise any background contributions. Upon methanol fixation, which also permeabilises
17Acidification of the nucleus could give rise to an increased signal.
75
Chapter 2: Single-molecule G4 visualisation in live cells
Figure 2.40: Background subtracted maximum intensity projections of videos taken over 70 frames
500 ms exposure, cycle time 3 s. 4 % Formaldehyde fixed U2OS cells labelled with (a) 1 nM SiR-PyPDS
(19), (b) control molecule 1 nM SiR-iPyPDS (20).
the cells, DNA degradase treatment (which degrades DNA into individual nucleotides) led to
72 % reduction of observed foci, but PDS pre-blocking made negligible difference (Figure: 2.41,
b). It was concluded that though SiR-PyPDS (19) primarily labelled DNA, the pre-blocking
experiments made it unclear whether the targets were G4s.
a b
Figure 2.41: (a) Quantification of the binding events within formaldehyde fixed nucleus lasting more
than one frame (100 ms per frame, over 400 frames) per cell for SiR-PyPDS (19) and SiR-iPyPDS
(20) after 1 nM labelling. n = 21, 5, 4, 4 and 8 for respective conditions taken from 1 - 3 independent
replicates. (b) Quantification of observed foci within methanol fixed U2OS cells labelled with 1 nM
SiR-PyPDS (19). n = 6, 4 and 6 for respective conditions taken from 1 independent replicate. Error
bars indicate mean ± sd.
76
Chapter 2: Single-molecule G4 visualisation in live cells
The effect of pH was investigated in formaldehyde fixed cells. It was observed that while small
pH changes from PBS buffer pH of 7.4 only induced small changes in observed foci count,
acidification of the solution to pH 3, led to a large scale light-up of the nucleus (Figure: 2.42).
This result reflected previous in vitro investigation on SiR-PyPDS (19) fluorescence dependence
on pH (Figure: 2.10), where high sensitivity was seen in 6-8 pH range. Low pH sensitivity in
fixed cells was beneficial as it meant that the probe was not measuring small environmental pH
differences in the nucleus.
Figure 2.42: SiR-PyPDS (19) observed foci quantification, after 1 nM treatment, was pH dependent
in formaldehyde fixed U2OS cells. n = 5, 7, 9, 6 for respective conditions taken from 1 independent
replicate. Error bars indicate mean ± sd.
Optimising labelling conditions in fixed cells had proven to be difficult for minimising non-
specific and off-target binding, therefore experiments in fixed cells were discontinued due to
time constraints.
Section summary
G4 structures were visualised in the nucleus of live U2OS cells via single-molecule imaging of
SiR-PyPDS (19). Specific G4 binding was investigated by the following control experiments:
• Isomeric, weaker G4 binder control molecule SiR-iPyPDS (20) showed 40-fold fewer bind-
ing events than SiR-PyPDS (19) at the same concentration.
• SiR-PyPDS (19) can be outcompeted by pre-blocking G4 sites with an excess of estab-
lished G4 ligands PDS (1) and PhenDC3 (2).
• Chemical trapping of the unfolded G4 state by DMS showed a time-dependent reduction
of observed probe binding events. This experiment also revealed that G4 structures are
dynamic in live cells and undergo unfolding over a time course of 20 min.
77
Chapter 2: Single-molecule G4 visualisation in live cells
Live cell control experiments were mirrored in vitro as to help obtain better understanding
of the induced perturbations in a simpler system. Perturbations on G4s in vitro could be
confidently attributed to be G4-related, not to be an influence of secondary effects. In contrast,
in live cells any perturbations, though aimed at G4s, could potentially influence other cellular
processes. However, if an identical perturbation performed in live cells and in vitro resulted
in similar conclusions, then it provides confidence that the perturbation worked on G4s in live
cells too.
SiR-PyPDS (19) probe was found to co-localise with lysosomes but no influence on their for-
mation was observed at the conditions used for single-molecule imaging. Moreover, the probe
did not cause a decrease in cell viability at low concentrations.
Single-molecule, single-G4 observation was investigated and discussed by single-step photo-
bleaching and residency time determination experiments. While individual-probe observation
could be proved, residency time comparison in cells and in vitro could only establish that it is
most likely that a single-G4 was also being observed.
G4 abundance was also found to be cell-cycle-dependent and G4 formation was linked to DNA
processing processes of transcription and replication.
2.5.8 Conclusions
The project described in this chapter was a demonstration of endogenous visualisation of G4s
in live U2OS cells. My own work in this project started with a 10-step synthesis of a novel
G4 ligand probe - PhenDC3-SiR (17), and biophysical characterisation of its ability to bind
G4 structures which was carried out both in bulk- and single-molecule in vitro experiments.
However, live cell single-molecule observation of PhenDC3-SiR (17) required a complicated
light-sheet imaging setup which limited its utility in live cell experiments.
Another set of probes, SiR-PyPDS (19) and its isomeric weak G4 binding control molecule
SiR-iPyPDS (20) proved to be more convenient to use in HILO imaging mode in live cells.
Biophysical in vitro assays were used to demonstrate SiR-PyPDS (19) binding affinity to G4
structures with selectivity against other forms of DNA.
20 nM labelling concentrations of SiR-PyPDS (19) were used for observing G4s in live U2OS
cells. Low concentrations of the probe, below Kds of G4 binding, had allowed endogenous
visualisation of G4s at a single-molecule level with minimal perturbations to G4s globally, as
was shown by G4 induction and kinetic unfolding experiments (while higher µM concentrations
can substantially influence a system containing G4s). However, minimal effect on G4s at a global
78
Chapter 2: Single-molecule G4 visualisation in live cells
level should not be mistaken with probe effect on individual G4s, indeed, all the G4s that were
observed with SiR-PyPDS (19) as single-molecules were perturbed while bound to the probe
and being imaged. This is an observers effect - one cannot observe a system without perturbing
it; thus, as it was demonstrated in this case, it is important to ensure that perturbations are
minimal. The aim of endogenous observation of G4s was important in the field due to G4
ability to be dynamic between its folded and unfolded state, therefore it is desirable to observe
endogenously folded G4s rather than visualisation of a perturbed system by G4 induction. This
criterion is tricky to prove, however in vitro experiments demonstrated a linear dependence
of observed foci against G4 surface coverage. This suggested that the imaging method was
sensitive to account for changes in the number of G4s and thus likely reflect the already folded
structures on the surface. In principle, the G4s could still be folding and unfolding every time
the probe binds/unbinds, but such a view seems unlikely given the slow G4 unfolding kinetics
and high G4 thermodynamic stability (in the presence of K+ cations).
Single G4 observation with SiR-PyPDS (19) was demonstrated by direct smFRET and single-
step photobleaching experiments in vitro on a surface, but in cells, single G4 observation could
not be claimed with such certainty. While single-step photobleaching in cells showed single
probe molecule visualisation, shorter residency times in cells than in vitro can be suggestive of
observation of individual G4s.
The G4 imaging platform presented in this chapter was used to investigate G4 dynamics in
live cells. When DMS treatment was used to trap the unfolded G4 state, a severe drop in
SiR-PyPDS (19) foci indicated that the majority of G4 structures unfolded within 20 minutes.
To the best of my knowledge, this is the first report of the assessment of potential G4 lifetimes
within living cells. And results are in line with the timescales of transcription initiation and
bursting [342]. Though DMS and β-mercaptoethanol used for quenching of DMS are both toxic
chemicals which could have other influences in a live cell, G4 unfolding was considered to be
the simplest and most plausible explanation. The G4 imaging platform was also used to show
G4 cell cycle dependence in a live system, confirming the observations made in fixed cells. G4s
were found to be most prevalent in S and G1/S cell cycle phases when DNA processing is most
active and G4s were at much lower levels in G0/G1 phase when transcription and replication are
primarily quiescent. This link between G4 formation and DNA processing steps of transcription
and replication was further confirmed by their inhibition with DRB and amphidicolin, what
also resulted in a large decrease of G4 localisations. Additional experiments with the two
inhibitors treated separately would be beneficial for establishing a more robust link to either
transcription or replication. Moreover, a systematic approach to inhibit different steps in the
mechanism could also provide interesting findings.
Finally, the new single-molecule imaging platform provides a novel means to account for relative
79
Chapter 2: Single-molecule G4 visualisation in live cells
G4 population levels in live cells in real-time. The method could be used to study endogenous
G4 levels in different cell types and states to assess G4 relation to cancer, cellular differentiation
or DNA processing mechanisms. I anticipate that further work with this or similar imaging
platform will help unravel specific biological functions regulated by G4s within the human
genome.
80
Chapter 3
3D STORM super-resolution imaging
of G4s
3.1 Project rationale and research aims
G4s and their relation to higher order chromatin structure is an under-explored area in the
field. G4s, being a secondary DNA structure, have an inherent requirement to transition
through a single-stranded state shifting from the double helix. From a molecular point of view,
it is reasonable to hypothesise that G4 formation would lead to a significant influence on the
local chromatin structure in the vicinity. A question which follows is whether G4s could then
impose their effect only in the locus of their formation or whether it could also drive (or follow)
changes in long range chromatin interactions. A glimpse of such a question has been tackled by
recent G4-ChIP-seq studies, which have shown that G4s were primarily detected in nucleosome
depleted, open chromatin regions and were coupled to chromatin compaction/decompaction
[61, 68, 198].
A slightly different view is given by G4s’ association with active transcription, and a potential
G4 link to 3D chromatin structure. G4s have been found to be highly enriched in gene promot-
ers, supported by bioinformatic studies based on predicted G4 forming sequences [44, 47] as
well as G4-ChIP-seq [61, 71, 72]. Moreover, genes containing an observed G4 were, on average,
more highly transcribed [61, 71], which could be a result of G4s acting as transcription factor
binding hubs [69, 72]. In greater detail, G4s in hESC were found to be enriched in active and
bivalent promoters (the latter being defined as such by having both H3K4me3 and H3K27me3
histone marks), which when a G4 was present demonstrated less pronounced transcriptional
variability upon differentiation [71]. Moreover, mechanistic investigation has found that G4
formation preceded active transcription and was associated with RNA polymerase II residency
81
Chapter 3: 3D STORM super-resolution imaging of G4s
[68], implying that G4s are not just passengers of DNA processing (i. e. when ssDNA bubbles
form during transcription/replication, condition necessary for G4 formation), but rather could
be important in transcription machinery recruitment and positioning. It has been suggested
that G4 links to transcription maybe manifest through G4 influence on 3D chromatin struc-
ture or vice versa, as gene expression is dependent on 3D chromatin arrangement of genome
regions such as enhancers and silencers associated with the transcription site [343]. Spatial
clustering of groups of expressing genes to transcription factories is another means of achieving
higher gene expression efficiency [344, 345]. G4s could potentially act as an additional layer of
transcriptional control by influencing 3D chromatin positioning.
G4s in transcriptional control in 3D space has been explored bioinformatically by investigating
promoter/enhancer interactions containing a G4 using Hi-C and G4-ChIP-seq datasets in K562
cells [72]. While 3 % of enhancers and 29 % of promoters contained a G4, 39 % of promoter-
enhancer interaction pairs contained a G4, having a higher probability of interaction than
non-G4-containing pairs. Promoter-promoter contacts containing G4s were also found to have
increased probability of pairing, suggesting that G4-associated gene clustering could be a means
of 3D positional control. Moreover, 34 % of TAD boundaries were found to contain a G4, regions
enriched in architectural proteins involved in DNA looping and insulating different TADs from
one another [72]. Another study has also made similar observations in hESC that promoters
containing a G4 were more likely to contact other promoter-interacting regions with a G4 [71].
These studies only provide links of association, to investigate whether G4s could cause changes
to 3D chromatin structure, perturbation experiments are required and 3D interactions need to
be proven.
A recent study investigated YY1, a zinc-finger-containing transcription factor, that binds G4s
[183] and is a structural regulator of enhancer-promoter loops [184]. Li and co-workers provided
evidence supporting that DNA looping exerted by YY1 was G4 structure dependent. This was
suggested by observing a drop in the number of DNA loops mediated by YY1 sites from 36 %
down to 9 % upon PDS (1) treatment [183]. Moreover, mutation of a G4-forming sequence by
CRISPR-Cas9 genome editing appeared to disrupt its ability to form a G4, led to a drop of YY1
binding to the site and consequently a drop of YY1-mediated DNA looping. Changes in DNA
looping were also found to affect associated genes expression levels. This study focused on one
protein to investigate a potential G4 role in 3D chromatin structure, but a more general genome-
wide approach to study G4s is desired to firmly establish G4s as 3D structure regulators. The
development of new techniques could potentially provide such means, one of which could be an
extension of currently established chromatin capture methods.
82
Chapter 3: 3D STORM super-resolution imaging of G4s
3.1.1 Project inspiration and idea
Single-cell Hi-C is a fixed nucleus chromatin capture technique which, by carrying out high-
throughput sequencing of DNA strands in close physical proximity to each other, provides
a means to simulate a 3D chromatin structure [278]. Recent developments [278, 302], have
allowed introduction of 3D super-resolution imaging of the same single-nucleus prior to stan-
dard Hi-C protocol. Single-nucleus images can then be overlaid with simulated 3D chromatin
structure of the nucleus from Hi-C. It has been successfully performed with single-molecule 3D
imaging for an endogenous fluorescently labelled centromeric CENP-A protein. This provides
a proof-of-principle for cross-validation of the genome structure simulation result (Figure: 3.1).
Centromere positioning obtained through sequencing and Hi-C simulation results corresponded
to the overlapped positioning obtained through fluorescence imaging, this demonstrates that
the two techniques are co-related. The key breakthrough in this new method was that it al-
lowed a fluorescence image and its foci to be related to their 3D chromatin context and genome
sequence.
a b
Figure 3.1: Single cell Hi-C whole genome structure overlapped with 3D fluorescence centromere
CENP-A protein imaging. Red spheres denote the location of centromeres, one chromosome and its
corresponding centromere is denoted in green, others in blue. (a) Whole nucleus visualisation. (b)
Zoomed in visualisation of centromere and simulated chromatin strand overlap. Data taken from [278],
figure was adapted for the purposes of this thesis.
In an analogous manner, this method has the potential to be expanded by combining it with
single G4 imaging technology, which uses G4 selective fluorescent probes. Hi-C simulated
83
Chapter 3: 3D STORM super-resolution imaging of G4s
chromatin structure superimposed with 3D fluorescence imaging of G4s forms the cornerstone
idea of this chapter (Figure: 3.2). To achieve this, some important milestones are required to be
met. First, G4 visualisation should be expanded to 3D and preferably done by super-resolution
imaging methods like STORM for more precise G4 foci localisation. Secondly, G4 labelling
should be compatible with the single-cell Hi-C protocol, in both achieving clear and reliable G4
3D images and obtaining sufficient number of Hi-C contacts from sequencing to simulate a 3D
genome structure. Lastly, Hi-C structure and 3D G4 images need to overlap with precision.
Overlay
3D super-resolution fluorescence image of G4s Hi-C simulated chromatin structure
Chromosome 
territory
Nucleolus
N
N N
N
O
NH
H
H
N
N
N N
O
N
H
H H
N
NN
N
O
N H
H
H
N
N
NN
O
N
H
HH
G G
G G=M+
G4 foci detected by 
fluorescent probe (   )
DNA looping
Architectural 
proteins
N
N N
N
O
NH
H
H
N
N
N N
O
N
H
H H
N
NN
N
O
N H
H
H
N
N
NN
O
N
H
HH
G G
G G=M+
Transcription
factors
Enhancer
Promoter
N
N N
N
O
NH
H
H
N
N
N N
O
N
H
H H
N
NN
N
O
N H
H
H
N
N
NN
O
N
H
HH
G G
G G=M+
N
N N
N
O
NH
H
H
N
N
N N
O
N
H
H H
N
NN
N
O
N H
H
H
N
N
NN
O
N
H
HH
G G
G G=M+
Topologically 
associated domain
Open 
chromatin 
region
Figure 3.2: Project overview - 3D super-resolution fluorescence G4 images are overlaid with Hi-C
simulated chromatin structure maps. This enables study into relationships between G4s and the 3D
structural features of chromatin.
Once established this study could be extended. For example, by obtaining RNA-seq data in
parallel to determine actively transcribed regions in the nucleus. RNA-seq in combination with
structural Hi-C data and G4 imaging data, would provide a detailed picture to investigate G4
influences on 3D nucleus structure and its functional relationships. Moreover, G4-related per-
turbations, such as G4 ligand treatments, unfolded G4 state trapping by dimethyl sulphate or
transfection of G4 oligos into the nucleus, would help to establish G4 function links to chromatin
structure. G4 ligands could block G4 sites from architectural proteins, chemical trapping could
reduce the G4 levels in cells and G4 oligo transfection would introduce additional G4s which
could restructure genome 3D interactions by acting as competitors. Whether chromatin struc-
ture is consistently influenced by the perturbations induced on or by G4s could be addressed
by these approaches.
84
Chapter 3: 3D STORM super-resolution imaging of G4s
Hi-C methodology together with G4 imaging would be a powerful new tool for investigating
G4 effects on 3D chromatin structure and G4 spatial localisation patterns within the nucleus.
With super-resolution visualisation of G4s in 3D, the question of whether G4s are clustering
in space could be tackled and provide a platform to investigate what mechanisms G4s may
employ to influence such processes in the nucleus. Moreover, this novel imaging-based method
of G4 genomic mapping would be orthogonal to other existing techniques like G4-ChIP-seq
antibody pulldown [61], G4 CUT&Tag [77] or G4-seq utilising polymerase stalling [52]. Such
cross-validation is important for higher confidence of true G4 structure formation mapping
and determining off-target effects. This general approach is not limited to G4s, as it could
be extended to investigate histone marks, DNA modifications, phase separation or any other
nuclear structural features which can be fluorescently labelled.
Key aims and objectives
For this work, a collaboration with the groups of Prof. Ernest Laue and Prof. Sir David Klen-
erman was established. They had previously laid the foundation for single-cell Hi-C structure
simulation and 3D fluorescence image overlap technique development [278, 302]. The introduc-
tion of G4 imaging in 3D into this technique was the main practical goal of this project.
The major objectives were:
• Obtaining a G4 antibody probe.
• Validating probes binding ability by biophysical experiments and by imaging in nuclei.
• Demonstrating the ability to image G4s in 3D by a super-resolution technique.
• Overlaying 3D G4 images with same single-nucleus Hi-C simulated genome structures.
• Overlay population derived G4 sequencing data with single-nucleus Hi-C data for cross-
validation.
• Analysing the overlaid 3D data.
Chapter outline
In this chapter, directly labelled G4 antibody probe development is described. The probe
was then validated for G4 binding biophysically and by imaging mESC nuclei, followed by a
demonstration of 3D STORM imaging of G4s. Further progress was halted due to emergence
of issues with the probe and imaging quality, for which details are discussed as well as alternate
85
Chapter 3: 3D STORM super-resolution imaging of G4s
solutions. A different probe was then used to visualise G4s in 3D, finding staining of majority
of nucleus volume with some regions of clustering. G4-CUT&Tag data was obtained for mESC
differentiation states and was overlaid with single-cell Hi-C data. Visualisation of G4 sequencing
data in 3D chromatin revealed strong G4s association with the A compartment of the nucleus
organisation.
3.2 G4 probe development for 3D STORM imaging
A DHPSF setup STORM [244] was being used for 3D fluorescence visualisation by the Laue
and Klenerman groups, therefore the same approach was selected for imaging G4s. As Hi-C
simulated structure maps were obtained using mESC nuclei, the same cells were also used.
3.2.1 Probe choice
Antibodies vs small molecules
To achieve 3D visualisation of G4s, a suitable choice of probe is required. First, the double-
helix STORM technique, Alexa Fluor 647 (AF647) dye was chosen due to its high brightness,
photostability and good blinking properties for STORM [346]. Moreover, the probe should
be compatible with a formaldehyde fixation and nuclear extraction protocol steps required
for Hi-C. Thus, G4 antibody probes were preferable to small molecule G4 probes, like PDS
and PhenDC3, used in chapter 2, as difficulties were encountered using SiR-PyPDS (19) and
PhenDC3-SiR (17) in fixed cells (Subsection: 2.5.7). Moreover, intact fixed cell labelling was
preferable to labelling extracted nuclei, since the latter led to unusually high non-specific probe
binding, being difficult to wash out. Nuclear extraction using 0.1 % IGEPAL CA-630 detergent
may also remove the probe from the nucleus, therefore a secondary fixation of the probe to the
chromatin, post-labelling was used. Since it was unclear whether small molecule G4 ligands
can be reliably fixed in place to chromatin, due to a potential lack of nucleophilic groups in
physiological conditions, antibody G4 probes were chosen as protein fixation by formaldehyde
was a more common and established methodology.
Finally, G4 antibodies generally have tighter binding constants and better selectivity against
other forms of DNA [95] than small molecule G4 ligands (for example SiR-PyPDS Kd(Myc) =
630 nM, BG4 Kd(Myc) = 3 nM), enabling the use of lower probe concentrations for labelling
and consequently reduced non-specific background. All in all, antibodies were the preferred
probe of choice.
86
Chapter 3: 3D STORM super-resolution imaging of G4s
Homogeneous vs heterogeneous labelling
For Hi-C structure and 3D G4 image overlap, an important parameter is the spatial density
of G4s in different regions of the nucleus. Therefore, it is important to consider what kind of
labelling is most appropriate to acquire G4 density measurements with precision. For this, ho-
mogeneous labelling (i.e. a precise known number of fluorophores attached per probe molecule)
is highly desirable, since provided that a single antibody is bound to a single G4, then the fluo-
rescence signal coming from a single point can be traced back to the number of G4s residing in
that point. When STORM is used for super-resolution localisation, dye molecules are purpose-
fully made to switch on and off, therefore a probabilistic approach is required to account for
the mean number of switching cycles per dye molecule. Thus, a single fluorophore per antibody
would be ideal and allow more precise observation of G4 clustering in space (Poisson binomial
distribution of a number of switch cycles coming from one focus point from one dye has less
uncertainty than compounded distribution from multiple dyes).
By comparison, heterogeneous labelling would lead to unknown number of fluorophores asso-
ciated per G4. While probabilities on fluorophore numbers could be calculated based on the
labelling mixture properties, combined with Poisson distributions from number of dye switching
cycles, such an approach would be less accurate.
Consequently, homogeneous antibody labelling methods were prioritised to obtain a novel probe
for fixed cell G4 imaging in 3D.
3.2.2 Direct BG4 labelling with AF647
The BG4 scFv phage display antibody was the first choice for labelling as it has been most
widely established as a G4 antibody probe (See chapter 1 for description). It has been used for
G4 imaging in fixed cells [59, 60], G4-ChIP-seq [61] and more recently CUT&Tag [77].
The scFv BG4 antibody protein was expressed in BL21(DE3) E. coli with kanamycin selection
and then purified by His-select nickel affinity beads1.
BG4 antibody lysine labelling with NHS ester
Initially, BG4 was labelled on lysine amino acid moieties with AF647 N -hydroxy-succinimide
ester (AF647-NHS). BG4 has 16 lysines in its sequence, 4 of which are in the FLAG-tag region
and 2 in complementarity-determining region (CDR). Reactions were carried out at pH 8.3 in
1See Materials and methods chapter for details of the protocol (Section: 5.6.1).
87
Chapter 3: 3D STORM super-resolution imaging of G4s
100 mM NaHCO3 supplemented PBS buffer. AF647-NHS dissolved in anhydrous DMSO at 10
mM was added to the reaction mixture and then shaken 500 rpm at 25 ◦C for 1 hour. After
the reaction, excess dye was removed by spin column purification (Scheme: 3.1).
Centrifuge
Step 1: Labelling reaction Step 2: Removal of excess dye
Use spin column Recover labelled
antibody
Mix antibody with dye
Scheme 3.1: General antibody labelling scheme.
4 equivalents of AF647-NHS were used to obtain low-labelled BG4-AF647 conjugate and 40
equivalents for high-labelled conjugate. Since lysine labelling by NHS-esters is non-selective,
a Poisson binomial distribution of labelled products was expected. To check whether BG4
conjugates retained their G4 binding activity, enzyme-linked immunosorbent assay (ELISA)
with Myc oligo was carried out (Figure: 3.3, a). High-labelled conjugate with an average of 4.2
dyes conjugated per BG4 antibody, was completely inactive. Meanwhile, low-labelled sample,
with an average of 0.8 antibody-fluorophore ratio (AFR), showed some G4 binding activity,
however the activity may be due to unlabelled BG4 in the mixture.
a
0.1 1 10 100
0.0
0.5
1.0
1.5
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc using anti-FLAG HRP
Low labelled 
0.8 AFR
Unlabelled BG4
High labelled 
4.2 AFR
b
1 10 100
0
1
2
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc using anti-His HRP
Unlabelled BG4
BG4-Lys-AF647 1 AFR
BG4-Lys-AF647 0.5 AFR
Figure 3.3: ELISAs of lysine labelled BG4 by using: (a) anti-FLAG-tag HRP, Kd(BG4) = 5.2 ±
0.6 nM (b) anti-His-tag HRP, Kd(BG4) = 17 ± 2 nM (standard error). Error bars in figure indicate
mean ± sd.
In an ELISA experiment, BG4 antibody is detected by binding a secondary antibody attached
to horseradish peroxidase (HRP) to its FLAG-tag binding region, which could also be blocked
by dye conjugation (which might mean that BG4 was still G4 active), since FLAG-tag contains
88
Chapter 3: 3D STORM super-resolution imaging of G4s
lysines in its sequence. To test this, ELISA using an anti-HisTag HRP secondary antibody
was carried out, however it also showed that BG4 was inactive in both high and low labelled
cases (Figure: 3.3, b). From these results it was concluded that even single lysine modification
with AF647 significantly abrogates G4 binding activity. This may arise either from having the
most reactive amine on BG4 in its G4 binding site at CDR or dye addition disturbing protein
folding. Due to lack of success with lysine labelling of BG4, alternative methods were sought.
BG4 antibody cysteine labelling with AF647 maleimide
BG4 contains 4 cysteine amino acids, of which none are in CDR, thus cysteine-selective labelling
chemistry by Michael reaction could provide means of conjugation (Scheme: 3.2). However,
with 10-100 equivalent excess of AF647-C2-maleimide no labelling of BG4 was observed (data
not shown).
N+
SO3NaO
SO3
-
N
SO3Na
NaO3S
7.4 pH, PBS buffer
10 eq TCEP
25 oC, 2h
Cys-SH
BG4
100 eq AF647-C2-meleimide
Cys
BG4
N
O
O
N+
SO3NaO
SO3
-
N
SO3Na
NaO3S
N
O
O
S
Scheme 3.2: BG4 cysteine amino acid attempted labelling with AF647-C2-maleimide.
3.2.3 IgG-BG4 labelling
Due to a lack of labelling success on scFv BG4, a commercial goat IgG variant of BG4 (from
Absolute Antibody, mammalian expression) was the next target for conjugation. Initially, lysine
labelling with AF647-NHS was attempted, rationalising that the larger IgG-BG4 could retain
G4 binding activity upon conjugation due to more available lysine labelling sites.
Goat IgG-BG4 lysine modification was carried out successfully (same reaction conditions as
for BG4, Section: 3.2.2) as characterised by UV-vis. Multiple reactions conjugating with 4,
10 and 40 equivalents of NHS ester produced AFRs of 2.7 - Low, 5.6 - Medium and 14.6 -
89
Chapter 3: 3D STORM super-resolution imaging of G4s
High respectively. Anti-goat HRP ELISA showed that unlabelled BG4 and goat IgG-BG4 have
similar Kd values against Myc G4 binding: 4.9 ± 1.9 nM for IgG-BG4 and 2.5 ± 0.7 nM for
BG4 (Figure: 3.4, a). However, ELISA results indicated that 5.6 and 14.6 AFR antibodies
had completely abolished Myc binding ability or/and anti-goat HRP binding specificity. Low
2.7 AFR mixture showed diminished binding with Kd = 90 ± 28 nM, an 18-fold reduction in
comparison with unlabelled antibody. However, such lower apparent binding affinity could have
arisen from trace quantities (estimated 6 % would be sufficient) of unlabelled and still active
antibody in the low labelled mixture. Taken together that medium 5.6 AFR and high 14.6
AFR labelling densities showed no corresponding antibody activities, it was also considered
that IgG-BG4 binding affinity was too sensitive to lysine modification and is not viable to
obtain a reliable G4 probe.
a
0.1 1 10 100
0.00
0.05
0.10
Concentration / nM
A
bs
or
ba
nc
ee
ELISA on Myc using anti-Goat IgG-HRP
High 14.6 AFR
Medium 5.6 AFR
Low 2.7 AFR
IgG-BG4 unlabelled
BG4 unlabelled
b
0.1 1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc using anti-goat HRP
IgG-BG4
SiteClick
Figure 3.4: (a) ELISA binding graphs of BG4 and goat IgG-BG4 comparison for lysine labelled
species. (b) ELISA binding graphs of IgG-BG4 binding comparison for SiteClick labelling chemistry.
Error bars indicate mean ± sd.
Site-click labelling
A SiteClick methodology [347, 348] was tested on goat IgG-BG4 which relies on enzymatic
truncating of sugar side chains on the Fc portion of IgG antibody and then utilizing the exposed
reactivity to enzymatically introduce a monosaccharide with azide moiety. The introduced azide
on the antibody then can be used in Click Chemistry to add a label of choice. After carrying
out the reaction, UV-vis analysis indicated successful conjugation of IgG-BG4 with AF647 dye,
however ELISA on Myc G4 showed greatly diminished G4 activity for the SiteClick product
(Figure: 3.4). As the reaction also produced low yields of the product, it was concluded that
Site-Click labelling was not providing the desired means to obtain a reliable G4 probe.
90
Chapter 3: 3D STORM super-resolution imaging of G4s
Disulphide rebridging labelling
A heavy and light-chain antibody rebridging strategy was attempted next. This is a two-
step process for IgG antibodies involving disulphide bridge reduction followed by rebridging
using dibromopyridazinedione reagent (PD)2 [349] which can then be tagged with a fluorophore
AF647 via strain-promoted copper-free azide-alkyne cycloaddition click reaction (Figure: 3.3).
Scheme 3.3: PD rebridging of IgG-BG4 to label with AF647.
This antibody labelling approach should provide homogenously labelled antibody based on
the reliability of the reactions previously reported [349]. Several attempts of the reaction
were made, but with no success. High-resolution mass spec experiments indicated that PD
moiety was not being incorporated into the antibody by the rebridging reaction. There was a
possibility, however, that PD moiety gets fragmented from the antibody in mass spectrometer,
2The reagent PD was synthesised by Dr Maximilian Lee.
91
Chapter 3: 3D STORM super-resolution imaging of G4s
and therefore would not be observed. The click reaction was also attempted on the IgG-BG4-PD
mixture, but no conjugation to the dye was observed by UV-vis spectroscopy. It was concluded
that PD rebridging was not a viable strategy for G4 probe acquisition.
3.2.4 IgG-BG4 confocal imaging
Due to difficulties encountered with direct antibody labelling, other options for imaging G4s in
fixed cells were explored. Therefore, two-layer IF with IgG-BG4 was performed on U2OS cells
fixed in 2 % formaldehyde for 10 min. A secondary rat IgG anti-FLAG-antibody labelled with
AF647 (MA1-142-A647, ThermoFisher) was tested first, but showed excess background noise
when staining cells without primary antibody.
A different commercial anti-goat donkey IgG (A-21447, ThermoFisher) secondary antibody la-
belled with AF647 was tested next. Two-antibody layer IF showed foci in fixed U2OS cells
nucleus, and upon DNAse treatment the majority of the signal coming from the nucleus dis-
appeared while secondary antibody alone showed negligible staining (Figure: 3.5, a, b, d).
Optimal imaging conditions were found with less dense labelling for better single-foci recogni-
tion (Figure: 3.5, c).
Although this work provided promising preliminary results, the staining was heterogeneous,
therefore other methods of obtaining a homogeneous probe were explored.
3.2.5 BG4-HaloTag and E12-HaloTag bacterial expressions
Another labelling strategy was HaloTag fusion with a G4 selective antibody that was expressed
and isolated from bacteria. HaloTag is modified haloalkane dehalogenase designed to covalently
bind to synthetic ligands [350]. Since there is only one active site on the protein, only one ligand
can react, therefore homogeneous labelling can be achieved.
BG4 and G4 selective E12 nanobody were chosen. The recombinant cloning to obtain BG4-
HaloTag and E12-HaloTag plasmids was carried out by Dr Sam Roberts. Unfortunately neither
fusion led to isolation of the desired products over multiple attempts.
3.2.6 Development of E12-AF647 G4-selective probe
An independent available G4 antibody for labelling was E12 nanobody. It was selected against
a Myc G4 structure by phage display by commercial partner (Hybrigenics Service, France).
92
Chapter 3: 3D STORM super-resolution imaging of G4s
a b
c d
Figure 3.5: Confocal imaging of fixed U2OS cells. All conditions used 0.63 nM of anti-goat donkey
IgG secondary antibody labelled with AF647. (a) 6.9 nM goat IgG-BG4. (b) Cells pre-treated with
DNAse before labelling with 6.9 nM goat IgG-BG4. (c) 0.69 nM goat IgG-BG4. (d) Secondary
antibody alone.
The design was carried out by Dr David Tannahill and Dr Jochen Spiegel. Initial isolation,
purification and binding studies were carried out by Silvia Galli which demonstrated E12 ability
to selectively bind G4 structures (Kd(Myc) = 4.6 ± 0.7 nM, Kd(Kit1) = 7.4 ± 2 nM, Kd(H-
telo) = 27 ± 8 nM). The smaller size of the nanobody vs BG4 (18 vs 29 kDa) may allow for
improved access to G4s in chromatin3. Moreover, direct labelling of a small nanobody places
the fluorophore ∼2-4 nm away from the G4 target, a significant improvement over IgG antibody
(∼12 nm uncertainty) or indirect labelling with two-antibody layers (∼20 nm uncertainty) [351].
Compared to ∼20 nm lateral resolution in STORM, nanobodies can provide higher precision
imaging.
3G4 sequencing techniques using BG4 for detection, ChIP-seq and CUT&Tag [61, 77], find most of BG4
binding sites to be in open chromatin, but there is a possibility that BG4 is unable to access G4s where
chromatin is closed.
93
Chapter 3: 3D STORM super-resolution imaging of G4s
Cysteine modification labelling reactions of E12 were performed under the same conditions used
for BG4 (Figure: 3.2). However, no conjugation was achieved, possibly due to cysteine being
unavailable and buried in the nanobody structure. It was also found that P6 gel spin columns
worked best4 to separate unconjugated AF647 dye from the nanobody.
Lysine labelling with NHS esters was attempted in parallel. It was performed according to the
following reaction scheme with varying number of AF647-NHS ester equivalents (Scheme: 3.4).
For an initial attempt 4 and 40 equivalents were used and generated a low labelled mixture
with 1.4 AFR and a high labelled mixture with 2.7 AFR respectively. It was surprising that 4
and 40 equivalents of dye used in the reaction led to a small difference in AFR, possibly due to
only a few surface exposed lysines in the nanobody structure, which can be easily saturated.
Sequence analysis showed that there are 8 lysines in total, with none in CDR.
N+
SO3Na
O
O
N
O
O
SO3
-
N
SO3Na
NaO3S
8.3 pH, NaHCO3/PBS buffer
25 oC, 2h
Lys-NH2
E12
AF647-NHS ester
H
N
N+
SO3Na
O
SO3
-N
SO3Na
NaO3S
E12-AF647
E12
Scheme 3.4: E12 lysine labelling with AF647-NHS ester.
The reaction products were analysed by mass spectrometry. Low lys labelled E12 sample
showed the incorporation of 1 - 4 dye molecules on each nanobody consistent with average
1.4 AFR determined by UV-vis spectroscopy analysis. No unlabelled E12 was observed in the
mixture even with low AFR suggesting that there is one lysine in E12 which is much more
reactive. This could potentially be exploited by trying to obtain singly labelled E12. Native
and labelled E12 mass results are in the table 3.1.
ELISA analysis of E12-AF647 was then performed. E12 contains FLAG-Tag and His-Tag
epitopes either of which could potentially become blocked by the dye modification, therefore
ELISA was performed against both of these tags. ELISA was carried out on Myc G4 labelled
4ZebaSpin, Vivaspin and His affinity gel columns all led to substantial loss of nanobody.
94
Chapter 3: 3D STORM super-resolution imaging of G4s
Species Mass Difference Modification
Native 18031 0
1x modified 18872 841 1x841
2x modified 19713 1682 2x841
3x modified 20554 2523 3x841
4x modified 21396 3365 4x841
Table 3.1: Masses of E12 modification series with AF647-NHS.
surface with anti-FLAG-Tag and anti-His-Tag HRPs. Results showed that high Lys labelled
E12 with an average of 2.7 AFR had completely lost its binding affinity to Myc G4 (Figure:
3.6). Low Lys labelled E12, however, retained some of E12 original binding affinity (Table:
3.2). As mass spectrometry showed no unlabelled E12, this activity must have arisen from
labelled E12. Binding curves with FLAG and His-tags are roughly the same, and suggest that
modification must come from reduced G4 binding and not conjugation to FLAG-Tag or His-Tag
regions of the nanobody.
a
1 10 100
0.00
0.05
0.10
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc G4 using Anti-FLAG HRP b
1 10 100
0.0
0.1
0.2
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc G4 using Anti-His HRP
E12 native
High Lys labelled 2.68 ratio
Low Lys labelled 1.42 ratio
Figure 3.6: ELISA on Myc G4 with E12-AF647, low and high labelled mixtures. (a) Using anti-
FLAG HRP, (b) Using anti-His-Tag HRP. Error bars indicate mean ± sd.
Sample FLAG-tag Kd / nM His-tag Kd / nM
Native E12 4.2 ± 0.5 5.5 ± 1.4
Low Lys labelled 1.42 ratio 390 ± 200 180 ± 100
Table 3.2: Dissociation constants of native E12 and lysine labelled mixture. Errors indicate standard
error of the fit from at least two replicates.
Initial E12 modification results suggested that there is potential to obtain singly labelled
nanobody with retained G4 binding activity, therefore further work continued.
95
Chapter 3: 3D STORM super-resolution imaging of G4s
Reactions with 1-2 equivalents of NHS ester and allowing the reaction to go to completion5
over 3 hours gave a Poisson binomial distribution mixture of 1x - 3x labelled E12 as shown by
MS (Figure: 3.7). Note that 1.5 and 2 eq reactions resulted in negligible levels of native E12,
therefore any G4 binding affinity measured by ELISA would primarily come from modified E12
(Table: 3.3). Peak intensities from MS were used to determine the composition of the samples
since the species compared are essentially the same nanobody, only differing with number of
AF647 attached. It is therefore reasonable to assume that all species have similar ionisation
and flight patterns in MS.
Native E12
1X labelled
2X labelled
3X labelled
1 eq
1.5 eq
2 eq
Figure 3.7: MS of E12 AF647 labelling reactions indicating that 1.5 and 2 eq labelling had little
native E12 in the product mixture. x-axis - mass, y-axis - relative intensity.
5With respect to NHS-ester which also gets hydrolysed in aqueous buffer.
96
Chapter 3: 3D STORM super-resolution imaging of G4s
Sample E12 / % 1X / % 2X / % 3X / % 4X / %
1 eq labelled 4.4 58.7 30.2 6.8 0
1.5 eq labelled 0 31.3 40.1 22.1 6.4
2 eq labelled 1.5 40.1 38.9 18.2 1.4
Table 3.3: Composition of E12 labelled species according to MS intensities.
Indeed, 1.5 eq and 2 eq labelled mixtures gave Kds in a similar range of magnitude compared
with native E12 binding (Figure: 3.8, Table: 3.4), therefore demonstrating that E12 labelled
with AF647 is G4 active, albeit with reduced affinity. As the labelled mixtures had higher
Kds, this suggests that the 1x modification has weaker binding to G4s and that higher modified
species may have no activity. The latter would not be a problem if inactive probe can be washed
out from a cell after staining. But if the probe were to stain other targets than G4s, the higher
labelled species would be visualised more frequently by STORM since they are brighter (have
more dye attached per nanobody). It was desirable to have minimal amount of higher labelled
species in the product mixture. A solution was to purify the single labelled E12-AF647, however
due to similarities between products it proved to be difficult (probe purification discussed in
later sections).
a
0.1 1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc G4 using anti-FLAG HRP
E12 native
1 eq
1.5 eq
2 eq
b
0.1 1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc G4 using anti-His HRP
E12 native
1 eq
1.5 eq
2 eq
Figure 3.8: ELISA on Myc G4 with E12-AF647, reactions with 1-2 equivalents of AF647-NHS ester.
(a) Using anti-FLAG-Tag HRP. (b) Using anti-His-Tag HRP. See table 3.4 for dissociation constant
comparison.
Sample FLAG-tag Kd / nM His-tag Kd / nM
Native E12 2.3 ± 0.4 7.7 ± 0.8
1 eq labelled 14 ± 2 37 ± 5
1.5 eq labelled 20 ± 8 37 ± 13
2 eq labelled 16 ± 4 7.2± 2
Table 3.4: Dissociation constants of labelled E12 mixtures. Errors indicate standard error of the fit
from at least two replicates.
97
Chapter 3: 3D STORM super-resolution imaging of G4s
As an interim solution, a possible workaround with lysine labelling was envisaged in obtaining
a pseudo-homogenous labelled mixture. Since only E12 with a fluorophore attached would
be observed under the microscope, in principle, unlabelled E12 would not contribute to the
mixture of the observed molecules. Therefore, unlabelled E12 could be present during imaging
only having an effect of blocking some G4 binding sites, but at low concentrations used for single-
molecule imaging, that should not pose a problem. So labelling E12 with sub-stoichiometric
amounts of AF647 could provide a mixture primarily consisting of unlabelled E12 and singly
modified E12-AF647, and since only the latter could be observed by fluorescence, the mixture
would appear homogeneous under a microscope.
When E12 was labelled with 0.8 equivalents of AF647 NHS ester, a Poisson binomial distribution
mixture of 0x, 1x and 2x modified species were obtained as indicated by MS (Figure: 3.9). Small
levels of 2x and label were present in the mixtures of different reaction batches (Table: 3.5),
however, this was considered an acceptable compromise for pseudo-homogeneous labelling to
perform preliminary imaging experiments.
Figure 3.9: MS of psiaudo-homogeneously labelled E12-AF647 (AR072 batch). x-axis - relative
intensity, y-axis mass.
AR071 AR072 AR076
Native E12 0.28 0.26 0.32
1x labelled 0.50 0.52 0.55
2x labelled 0.18 0.18 0.11
3x labelled 0.04 0.03 0.02
AFR 0.98 0.97 0.83
Table 3.5: Labelled E12-AF647 species composition of different reaction batches.
98
Chapter 3: 3D STORM super-resolution imaging of G4s
ELISA of E12-AF647 reaction mixtures showed comparable G4 binding activity to native E12
and no binding to hairpin dsDNA in concentration range tested (Figure: 3.10). Unlabelled E12
did contribute to the binding affinity of the mixture, however similarity of Kds suggested that
labelled E12-AF647 should also be G4 active.
0.1 1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on Myc using FLAG-tag
E12 native Kd = 6.4 ± 0.8 nM
E12-AF647 Kd = 4.4 ± 0.3 nM
a
c
e f
b
d
0.1 1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on MYC using His-tag
E12 native Kd = 8.9 ± 0.9 nM
E12-AF647 Kd = 3.8 ± 0.3 nM
0.1 1 10 100
0.00
0.05
0.10
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on H-telo using FLAG-tag
E12 native Kd = 27 ± 8 nM
E12-AF647 Kd = 28 ± 3 nM
0.1 1 10 100
0.00
0.05
0.10
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on H-telo using His-tag
E12 native Kd = 75 ± 25 nM
E12-AF647 Kd = 49± 8 nM
0.1 1 10 100
0.11
0.12
0.13
0.14
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on Hairpin dsDNA using FLAG-tag
E12 native
E12-AF647
0.1 1 10 100
0.11
0.12
0.13
0.14
Concentration / nM
Fl
uo
re
sc
en
ce
(a
.u
.)
ELISA on Hairpin dsDNA using His-tag
E12 native
E12-AF647
Figure 3.10: ELISAs on various DNA oligos with E12-AF647 labelled mixture. (a) Baseline-corrected
ELISA on Myc using FLAG-tag. (b) Baseline-corrected ELISA on Myc using His-tag. (c) Baseline-
corrected ELISA on H-telo using FLAG-tag. (d) Baseline-corrected ELISA on H-telo using His-tag.
(e) ELISA on hairpin dsDNA using FLAG-tag. (f) ELISA on hairpin dsDNA using His-tag. Error bars
indicate standard error of the mean of 2 individual replicates. Binding constant errors are standard
errors of the fit.
99
Chapter 3: 3D STORM super-resolution imaging of G4s
Section summary
In this section, different approaches were presented on attempts to obtain a directly labelled
G4 selective antibody. The majority of the approaches were not successful, but provided in-
formation about applicability of modification methods available for G4 antibodies. Cysteine
labelling of BG4, IgG-BG4 and E12 with AF647-maleimide did not provide any observable
reaction products, suggesting that cysteines could be buried in protein structures. Additional
modification techniques were attempted on IgG-BG4, where PD rebridging and SiteClick la-
belling strategies were applicable, however neither gave positive results. Lysine modifications
with AF647-NHS ester, though successfully reacted, led to inactivation of BG4 and IgG-BG4
binding to a G4 structure even at low AFR. This suggests high reactivity of lysines in G4
binding site of the antibodies. While lysine labelling on E12, led to nanobody inactivation at
high AFR, controlled labelling demonstrated that singly labelled E12-AF647 had G4 binding
activity. This labelled E12-AF647 was used in the next section for single-molecule imaging
experiments in extracted mESC nuclei.
3.3 Single-molecule imaging with E12-AF647 labelled mix-
ture - controls and optimisation
The E12-AF647 labelled product mixture was imaged at single-molecule level in extracted
mESC nuclei. This section describes the development of nuclei preparation protocol for G4
imaging that is compatible with single-cell Hi-C. A discussion then follows on E12-AF647
imaging in mESC nuclei with emphasis on controls performed to provide evidence for G4
visualisation. Key aims of this section are:
• Development of G4 staining protocol in fixed mESC nuclei
• Optimisation of G4 imaging conditions
• Performing G4 imaging controls
• Demonstrating compatibility with Hi-C
3.3.1 STORM imaging conditions
STORM imaging utilises a specialised buffer to ensure stable photoswitching of fluorophores for
super-resolved localisation [346]. The buffer includes 50 mM Tris at pH 8, 10 mM NaCl, 10 %
100
Chapter 3: 3D STORM super-resolution imaging of G4s
glucose and 0.5 mg/mL glucose oxidase with 40 µg/mL catalase oxygen scavenging system.
Also, 10 mM mercaptoethylamine (MEA), a reactive thiol, which with red light irradiation
can perform nucleophilic attack on AF647 dye conjugated system, breaking conjugation and
switching it to the dark state (Figure: 3.11). The dark state can reverse back to the fluorescent
state by thermal activation or by UV irradiation at 405 nm [352]. The enzymatic oxygen
scavenging system leads to a production of an unfavourable byproduct - gluconic acid [353].
Over time its accumulation acidifies the imaging solution, resulting in a reduced photoswitching
rate of the fluorophores since MEA becomes less nucleophilic at lower pH. This essentially gives
a time limit of about 2 hours for a sample to be imaged under these conditions. The buffer
described, termed STORM buffer was used for all imaging experiments of this chapter.
HN
N+ SO3Na
O
SO3
-
N
SO3Na
NaO3S
E12
hν (red)S-
H2N
HN
N SO3Na
O
SO3
-
N
SO3Na
NaO3S
E12
SR
On-state Dark state
hν (UV) or ∆
Figure 3.11: Equilibrium between fluorescent on-state and dark state of AF647 dye in STORM
buffer conditions. The MEA thiolate anion undergo reversible nucleophilic addition to the AF647
conjugated system.
A Jablonski diagram can help in understanding the processes involved in fluorescence STORM
imaging (Figure: 3.12). The fluorophore initially resides in a ground electronic singlet state
(S0), but upon red laser irradiation a photon can transfer it to an excited S1 state, on one
of the higher vibrational energy levels. After relaxing to the lowest vibration level of S1, the
fluorophore can revert back to the ground S0 state by emitting a photon, to drop to a lower
energy than initially excited, causing a red-shifted emission also referred as Stokes-shift. The
emitted light is the fluorescence signal that is observed for single-molecule visualisation.
The excited S1 can also perform an intersystem crossing (ISC) to triplet T1 state, which is
relatively long-lived lasting µs to seconds, compared to singlet excited states lasting ns [354].
Both S1 and T1 excited states can react with MEA thiolate anion to form an adduct with
disrupted conjugation between the two aromatic systems of the AF647 (Figure: 3.11). The
adduct is long-lived, lasting more than ms and generally is referred as the dark state (D) [355].
T1 and D states are non-fluorescent and collectively are referred as the off-state, but they can
101
Chapter 3: 3D STORM super-resolution imaging of G4s
revert to ground state by emitting a photon or losing MEA. S1 and T1 excited states can react
with molecular oxygen in solution leading to photobleaching products which deplete the amount
of active fluorescent molecules. Since such photobleacing is essentially irreversible over the time
course of imaging experiments, it is important to reduce molecular oxygen concentration in
solution by oxygen scavenging systems.
On-state Off-state
S1
S0
D
T1hν hν
UV
ns
ISC
MEA
Photobleached products
µs
ms−min
E
O2
Figure 3.12: Jablonski diagram of AF647 with MEA present. Fluorescence-capable S0 and S1 states
are collectively referred as the On-state, while non-fluorescent states are collectively referred as the
Off-state. Solid arrows represent transitions between states, dashed arrow represent possible reaction
with MEA, dotted arrows represent possible reaction with oxygen leading to photobleached products.
The interplay of the processes involved in fluorescence imaging can be visualised experimentally
by following fluorescence intensity time profiles. During the initial 100 frames, fluorescence
of the E12-AF647 labelled sample was high as most of the fluorophore was in the On-state
(excitation and emission cycles between S0 and S1 states), but the fluorescence then sharply
decreased as more of AF647 entered T1 and D states (Figure: 3.13, a). Since these states can
revert back to S0 state, an equilibrium was reached after about a 1000 frames. A slow decrease
of fluorescence then follows as irreversible photobleaching continues. By following a fluorescence
intensity time profile of a blank sample, bleaching of the background autofluorescence of the
nucleus can be demonstrated (Figure: 3.13, b). After 500-1000 frames autofluorescence is
quenched to approximate steady state levels in the short term.
102
Chapter 3: 3D STORM super-resolution imaging of G4s
a
b
E12-AF647
Blank
Figure 3.13: Mean fluorescence intensity time profiles of: (a) 6 nM E12-AF647 labelled mESC
nucleus. (b) Blank nucleus. 30 ms exposure, 9.7 kW/cm2 laser power at 650 nm. y-axis - line profile
mean relative intensity.
Whenever a mESC nucleus was to be imaged, it was therefore bleached for 1000 frames first to
reach equilibrium between the On- and Off-states of AF647 and quench the autofluorescence.
3.3.2 Fixed mESC nuclei imaging with E12-AF647
Pluripotent mouse ES cells were cultured in a state of self-renewal, termed the ground state
[356]. This is achieved by growing the cells in a serum-free media with the addition of two
103
Chapter 3: 3D STORM super-resolution imaging of G4s
chemical inhibitors (collectively termed 2i), CHIR99021 and PD0325901 together with mouse
cytokine leukemia inhibitory factor (LIF) [357]. 2i/LIF conditions keep mESC pluripotent state
when cells grow in tight circular colonies6 (Figure: 3.14).
Figure 3.14: mESC colonies grown in serum-free NDiff 227 media supplemented with 2i/LIF under
a light microscope.
As adherent cell populations would experience different conditions with respect to labelling
depending on their position in the colony, it was decided to perform all IF treatments in a
cell suspension to ensure uniform treatment of floating cells. Consequently, separate steps
are required to pellet the cells/nuclei by centrifugation. Initially, this led to substantial loss
of samples over the course of the protocol, and very few nuclei (sometimes none) would be
available for microscopy at the end of the protocol. It was found that the losses were caused
by the cells/nuclei sticking to the surface of centrifugation tube this way avoiding formation of
a tight pellet. Switching from polypropylene falcon centrifugation tubes to eppendorf coated
protein LoBind R© centrifugation tubes, which have reduced surface binding of proteins, greatly
increased retention yielding thousands of nuclei by the end of IF labelling protocol.
Extracted nuclei were empirically observed to be have very high background on antibody treat-
ments. To avoid this issue, E12-AF647 antibody treatment was first done on formaldehyde
fixed, permeabilised mESCs as in a regular IF staining protocol7, followed by a secondary 2 %
formaldehyde fixation to fix the antibody to the chromatin, and with nuclear extraction using
IGEPAL CA-630 detergent. A secondary fixation is necessary to avoid the detergent of washing
the antibody out during the nuclei extraction step.
6Details of precise culturing procedures can be found in Materials and methods chapter (Section: 5.2.2).
7Details of the protocol can be found in Materials and methods chapter (Section: 5.3.3).
104
Chapter 3: 3D STORM super-resolution imaging of G4s
After nanobody treatment and imaging optimisation it was found that approximately 5 nM of
E12-AF647 provided a suitable labelling density for 3D STORM. This was by a rule of thumb
criteria which was to have about 1-2 localisations per frame, such density generally does not
lead to much overlap of the foci in a single frame when a double-helix setup is used. Initial
controls for G4 binding were carried out with benzonase (digests DNA and RNA) and RNAse
A (degrades single-stranded RNA at C and U residues). These control experiments indicated
that E12-AF647 does not target RNA in fixed nuclei and that it is primarily selective towards
DNA structures (Figure: 3.15).
5 n
M
5 n
M 
RN
As
e A
20
 nM
20
 nM
 R
NA
se
 A
20
 nM
 B
en
zo
na
se
0
1000
2000
3000
4000
O
bs
er
ve
d 
fo
ci
G4 binding controls mixture of E12-AF647 products
Figure 3.15: E12-AF647 primarily binds to DNA structures. The nuclei were labelled with 5 nM or
20 nM of E12-AF647 product mixture. The foci were observed over 1000 frames, 200 ms exposure.
Observed foci are counted per nucleus, error bars indicate mean ± sd. 3-5 nuclei imaged per condition,
taken from 1 independent replicate. Imaging experiments were performed in collaboration with Edward
Sanders.
Additional controls were carried out using excess of unlabelled G4 ligands to pre-block G4
binding sites (Figure: 3.16). BG4 and PDS pre-block reduced E12-AF647 foci number by 66 %
and 57 % respectively, while E12 pre-block led to a 68 % reduction (Figure: 3.17). The 98.8 %
reduction of observed foci upon benzonase digestion also suggests probe binding to nucleic
acids. These results strongly suggest that the majority of E12-AF647 binding targets in mESC
nuclei were common with other G4 ligands, suggesting G4-selectivity. It was interesting to
find that 3 different ligands all reduced the observed foci number approximately 3-fold, which
may indicate that the remainder third fraction of the foci could be attributed to non-G4 target
noise.
105
Chapter 3: 3D STORM super-resolution imaging of G4s
5 µM
a b
c d
e f
g
Figure 3.16: 5 nM E12-AF647 product mixture labelled mESC nuclei. Background subtracted
images of maximum projection of 2000 frames taken at 200 ms exposure. (a) Reference nucleus. (b)
Benzonase disgestion control. (c) 400 nM BG4 pre-block. (d) Unlabelled nucleus. (e) 10 µM PDS
pre-block. (f) 400 nM E12 pre-block. (g) 600 mM DMS pre-treatment of fixed cells at 37 ◦C. Imaging
experiments were performed in collaboration with Edward Sanders.
106
Chapter 3: 3D STORM super-resolution imaging of G4s
Note that all data collected up to this point has been with using a E12-AF647 labelled product
mixture. For example, the AR076 batch used for imaging contained 32 % unlabelled E12, 55 %
1X AF647 labelled, 11 % 2X labelled and 2 % 3X labelled with an overall AFR of 0.83 (as
determined by mass spectrometry). While majority of the probe is still with the desired 1X
label, the higher labelled impurities are unlikely to contribute to the overall picture, since they
most likely are G4 inactive (see section: 3.2.6). On the other hand, higher labelled species
might be notable contributors to the residual observed foci in pre-blocked G4 controls.
5 n
M
Be
nz
on
as
e
80
0 n
M 
BG
4 p
re-
blo
ck
10
 µM
 PD
S p
re-
blo
ck
80
0 n
M 
E1
2 p
re-
blo
ck
0
2000
4000
6000
8000
10000
12000
O
bs
er
ve
d 
fo
ci
G4 binding controls mixture of E12-AF647 products
Figure 3.17: G4 binding controls with observed foci counts being made per nucleus over 1000 frames
(200 ms). All samples labelled with 5 nM E12-AF647 product mixture. Error bars indicate mean
± sd. n = 12, 4, 8, 10 and 8 nuclei for 5 nM, benzonase, 800 nM BG4 pre-block, 10 µM PDS pre-block
and 800 nM E12 pre-block conditions respectively, taken from two to three independent replicates.
Imaging experiments were performed in collaboration with Edward Sanders.
An orthogonal control for providing evidence of G4 imaging, was to use DMS to trap the
unfolded G4 state in a fixed cell, this way reducing the number of foci observed by G4 binding.
However, pre-treating the fixed mES cells with 600 mM DMS, led to strong overall light-up of
the nucleus after E12-AF647 labelling (Figure: 3.16, g). So single-molecule foci were no longer
visible in the background noise. The increase of brightness was similar to the effect DMS had
on fixed U2OS cells when SiR-PyPDS (19) was used for imaging (Section: 2.41). These suggest
that DMS leads to a major increase in background fluorescence that swamps the single-molecule
signal.
107
Chapter 3: 3D STORM super-resolution imaging of G4s
3.3.3 G4 labelling protocol compatibility with Hi-C
Next, it was important to confirm that the optimised imaging protocol is compatible with Hi-C
and does not disturb regular cell chromatin contact densities. To check this, the G4 labelling
procedure of mESC nuclei was performed. The nuclei were then FACS (fluorescence-activated
cell sorting) sorted as single-nuclei, subjected to restriction enzyme digestion and tested via
PCR (polymerase chain reaction) by Dr David Lando. The PCR test determines the contact
densities of the chromatin of the nuclei in preparation of Hi-C libraries. Contact densities were
sufficient for further Hi-C processing for 8/10 nuclei (Figure: 3.18). This suggested that G4
labelling protocol did not degrade chromatin contacts and is compatible with the single-cell
Hi-C protocol.
Figure 3.18: PCR library test to see library contact densities for further steps in Hi-C protocol.
Nuclei producing library products which are short on average, are not suitable for further Hi-C pro-
cessing. ”-” denotes line of negative control without a nucleus. Performed in collaboration with Dr
David Lando.
Section summary
In this section, an E12-AF647 labelled product mixture was used to obtain single-molecule
imaging data in mESC nuclei under STORM conditions. First, a G4 staining protocol was
developed and optimised to show that 5 nM of probe mixture was suitable for robust STORM.
Then G4 binding controls were performed supporting that E12-AF647 targets DNA and that
two-thirds of its targets are G4s. Finally, G4 IF staining protocol was demonstrated to be
compatible with Hi-C protocol. These results together showed potential for using the E12-
AF647 nanobody for 3D G4 imaging in STORM.
108
Chapter 3: 3D STORM super-resolution imaging of G4s
3.4 Obtaining singly labelled E12-AF647
Given the promising results with the E12-AF647 labelled products mixture, it became desirable
and worthwhile to invest time in obtaining a purified version of singly labelled E12-AF647.
3.4.1 FPLC purification
The aim was to further purify E12-AF647 probe mixture by A¨KTA FPLC (Fast liquid protein
chromatography). What is more, the BioSpin6 column purifications for unreacted dye removal
from lysine labelling generally yielded 10-20 % of nanobody, therefore an improved method of
probe purification was desirable.
After lysine modification, the AF647 moiety carries net 3 negative charge, and since it also
changes a positive lysine group into an amide group, one modification changes the nanobody
overall by 4 negative charges. These charge differences between labelled species would change
the properties of the nanobody, giving grounds for separation by ion-exchange chromatography
[358]. This method can separate ions based on their affinity to an ion exchanger attached to the
column matrix, which can be positively and negatively charged groups for anion- and cation-
exchange respectively.
A¨KTA FPLC purifications were performed immediately after AF647-NHS ester labelling re-
actions of E12, to both remove the unreacted dye and attempt to separate nanobodies with
different degree of labelling. Being limited by E12 stocks, initial purifications were carried out
at a small scale of about 15 µg of E12 per run.
Cation exchange was attempted first with 1 mL HiTrap SP HP columns by Cytiva, with
sepharose matrix and sulfopropyl functional groups. As higher labelled species have more
negative AF647 attached the labelled nanobody would have weaker affinity to the negatively
charged resin and be eluted before 1X labelled product. Then unlabelled E12 would be eluted
last, but its presence can be reduced to low levels by reacting with AF647-NHS ester, as seen
previously (Figure: 3.7). However, cation exchange chromatography run at pH 5.5 in acetate
buffer did not provide any chromatography peak separation. It is possible that at pH 5.5 the
nanobody was not sufficiently positively charged to separate the species (data not shown).
Lower pH was not to tested to avoid damaging the nanobody.
Anion exchange was attempted next with 1 mL HiTrap Q HP columns by Cytiva, with sepharose
matrix and quaternary ammonium functional groups. In this case, the more negative highly
labelled E12 species would have higher affinity to the column and would be eluted last, while
unmodified E12 and singly labelled E12-AF647 should theoretically be the first to elute. The
109
Chapter 3: 3D STORM super-resolution imaging of G4s
purification run at pH 8 (in 50 mM Tris buffer), provided separation of 3 product peaks, eluted
by increasing NaCl concentration. However, the eluted fractions had too low concentrations of
nanobodies to be detected and confirmed by MS, even after concentrating. Observation of 3
separated peaks by UV-vis spectroscopy showed the principle that differently labelled products
could be separated.
The low yields suggested that the column had a high dead retention, which at small scale
purifications was too large to obtain good yields after elution. Because of this, another attempt
was made with a hydrophobic C18 column, which although had better elution yields, but was
completely unable to separate reaction products (data not shown).
Consequently, optimisation of anion exchange chromatography followed, with increased scale
(up to 75 µg of E12). The pH was also increased to 8.5. The chromatography run was successful
in separating the reaction products (Figure: 3.19). The first peak eluted at 29 mL and was
identified by mass spectrometry to be pure E12. The second peak at 30 mL consisted of 96 %
E12-AF647 with 4 % of unlabelled E12. The peak at 32 mL was unreacted dye and 2x and above
labelled E12. Overall, the reaction and purification provided 8 µg yield of E12-AF647 with 96 %
purity. Although such yield is sufficient for many single-molecule imaging experiments, higher
yields are required for performing biophysical characterisation of this novel probe.
Figure 3.19: Purification UV chromatogram of AR080 labelling reaction with 1 equivalent of AF647-
NHS to 75 µg of E12. Anion exchange chromatography, 50 mM tris pH 8.5 buffer. Blue and purple
lines denote chromatogram UV traces at 280 and 650 nm respectively. Green line denotes change in
NaCl concentration.
110
Chapter 3: 3D STORM super-resolution imaging of G4s
A high yielding E12 bacterial expression has provided the means to scale up the labelling
reaction and its purification even more. To test the limits of the reaction, it was scaled to 1.5
mg of E12. Unfortunately, at a larger scale, the column was overloaded during the purification
and complete separation of the reaction products was not achieved (Figure: 3.20). The first E12
peak and second E12-AF647 1X peak significantly overlapped, and depending on the fraction
E12-AF647 1X contribution varied from 69 to 84 % of total detected nanobody mass. The
third largest peak surprisingly contained 65 % of single label and 32 % of double label of
antibody indicating a strange E12-AF647 1X split into two peaks. E12 was still detected in all
the fractions suggesting significant streaking. All the peaks eluted at approximately 440 mM
NaCl concentration, meaning that the differently labelled species might have similar affinities
to the ion exchange column and the separation is mainly achieved by their different rates of
movement according to their charge. Because of this, possibly a longer column could achieve
better separation. Halting the gradient below 440 mM NaCl might also improve separation as
E12 might elute from the column slightly earlier. The 280 nm UV trace indicated that the
majority of E12 was unreacted, suggesting that greater equivalents of AF647-NHS should be
used in the labelling reaction. This would reduce the relative amounts of unlabelled E12 and
streaking. Finally, lowering the reaction scale could help achieve better separation.
E12
E12 (16-31%)
E12-AF647 1X (69-84%)
E12-AF647 3X
other protein impurities
E12 (3%)
E12-AF647 1X (65%)
E12-AF647 2X (32%)
Figure 3.20: AR081 reaction at 1.5 mg scale purification UV chromatogram. Blue and purple lines
denote chromatogram UV traces at 280 and 650 nm respectively. Green line denotes change in NaCl
concentration, orange - solution conductivity.
The guidelines outlined above were used for the next reaction and purification. The separation
of peaks was better (Figure: 3.21), however most of E12-AF647 (1X) was still mixed with
111
Chapter 3: 3D STORM super-resolution imaging of G4s
unlabelled E12, with fractions sequentially having 53-97 % of E12-AF647. The final fraction
of 28 mL peak had 99 % purity of E12-AF647 (1X) with respect to other nanobody species
(Figure: 3.22). The product fraction was desalted by dialysis into PBS overnight at 4 ◦C
through 3.5 kDa cut-off membrane, and to any residual unreacted dye in the fraction. Indeed,
no free AF647 was observed by MS. However, even though the reaction was done with 1 mg
scale of E12, pure product yield was only 10 µg, 1 %. This batch demonstrated that pure
E12-AF647 can be obtained to use in single-molecule experiments.
E12
E12 + E12-AF647
E12-AF647 1X, 2X + dye
One fraction was 99%
E12-AF647 (1X)
Figure 3.21: AR082 reaction batch purification. 1.5 equivalents of AF647-NHS dye was reacted with
1 mg of E12 (28FTB batch). Gradient was held at 350 mM NaCl, 20 mM Tris pH 8.5 for major peaks.
Two 1 mL HiTrap Q HP columns were used in parallel. Blue and purple lines denote chromatogram
UV traces at 280 and 650 nm respectively. Green line denotes change in NaCl concentration.
After AR081 and AR082 reaction batches purifications (Figure: 3.21), the majority of the
resulting products were only partially separated, approximately containing 20 % unlabelled E12
with 80 % of E12-AF647. Those fractions were then used in a second round of purification trying
to separate the species completely. A lower elution gradient was held at 300 mM NaCl and
two HiTrap columns were connected in sequence to improve separation. Indeed, the first E12
peak and E12-AF647 second peak became better separated, yielding 20 µg of pure E12-AF647
(1X) from all of the fractions (Figure: 3.23). Even though only fractions from a single peak
were used, these separated into multiple peaks after a secondary purification. The appearance
112
Chapter 3: 3D STORM super-resolution imaging of G4s
a
b
Figure 3.22: (a) Deconvoluted MS spectrum of ARO82 batch E12-AF647 (1X). (b) UV chro-
matogram at 280 nm. Single peak mass corresponds to mass of E12 and AF647.
of the third peak at 48 mL with strong absorbance at 650 nm but not 280 nm was unexpected,
since it suggested that not all AF647 dye was removed by primary purification. It also revealed
that the product fractions require additional purification, either by a secondary ion-exchange
purification as in the example above, or by dialysis through a suitable cut-off membrane.
113
Chapter 3: 3D STORM super-resolution imaging of G4s
E12
E12-AF647 (1X)
Unreacted dye
Figure 3.23: Secondary anion-exchange purifications of 2nd peak fractions from AR081 and AR082
reactions. Gradient held at 300 mM NaCl, 50 mM Tris pH 8.5 for major peaks. 0.25 mg of E12 scale.
Blue and purple lines denote chromatogram UV traces at 280 and 650 nm respectively. Green line
denotes change in NaCl concentration.
3.4.2 ELISA to validate G4 binding of E12-AF647
ELISA experiments were carried out to confirm purified E12-AF647 binding affinity to G4s
and whether purification diminishes activity. It was found that native E12, labelled E12-
AF647 (1X) and native E12 recovered from the purification had similar affinity to Myc G4 in
ELISA (Figure: 3.24, a; Table: 3.6). These results demonstrated that neither modification,
nor FPLC purification diminished E12 binding to G4s. E12-AF647 selectivity for G4s was
then tested and no binding to single-stranded8 and double stranded DNA was observed in the
concentration range tested (Figure: 3.24, b). Different batches of E12-AF647 were compared,
after primary FPLC purification, E12-AF647 (AR082) and E12-XP9 (AR082), and after sec-
ondary purification (batches II-42 and II-53) against native E12 (Table: 3.6). Overall, these
results demonstrated that labelled E12-AF647 is G4 active, but secondary FPLC purification
can reduce its G4 binding ability.
8Random ssDNA oligo had a randomised nucleotide sequence NHN NHN NHN NHN NHN NHN NHN N
with every 3rd base (H) being random between A and T, so the oligo could not form G-tetrads.
9E12-XP stands for E12 extra purified.
114
Chapter 3: 3D STORM super-resolution imaging of G4s
a
1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
ELISA on Myc G4 using FLAG HRP
E12
E12-AF647-AR082
E12-XP
A
bs
or
ba
nc
e
b
1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
E12-AF647 ELISA controls
A
bs
or
ba
nc
e
dsDNA
random ssDNA
hairpin dsDNA
Myc
Figure 3.24: ELISA against Myc G4 using FLAG-tag HRP for detection of E12 binding. (a) ELISA
comparison between unlabelled E12, E12-AF647 from batch AR082 and unlabelled E12-XP recovered
after FPLC purification. (b) E12-AF647 (batch II-42) ELISAs showing selectivity to Myc G4s. Error
bars indicate standard error of the mean of 2 individual replicates.
Batch Kd(FLAG-tag HRP) /nM K(His-tag HRP)d /nM
E12 native 5.0 ± 0.4 7.5 ± 0.6
E12-AF647(AR082) 3.7 ± 0.3 16 ± 2
E12-XP (AR082) 3.9 ± 0.3 16 ± 2
E12-AF647(II-42) 17 ± 1 28 ± 1
E12-XP (II-42) 2.0 ± 0.2 4.0 ± 0.3
E12-AF647(II-53) 42 ± 5
Table 3.6: Dissociation constants of different batches of E12 and E12-AF647. Errors indicate stan-
dard error of the fit from at least two replicates.
Section summary
Singly labelled E12-AF647 probe was successfully purified from other E12 products. Purified
E12-AF647 showed undiminished binding affinity to Myc G4 with selectivity against ssDNA
and dsDNA.
3.5 3D STORM imaging of G4s with E12-AF647
In this section, 3D STORM super-resolution imaging results with purified E12-AF647 probe
are presented. First, G4 imaging binding controls were repeated with the purified probe. Then
a slice of the mESC nucleus 3D image with drift-correction is presented. Finally, whole nucleus
imaging results are presented.
115
Chapter 3: 3D STORM super-resolution imaging of G4s
The key aim of this section is demonstrating the ability of imaging G4s in 3D with super-
resolution and drift-correction.
3.5.1 G4 binding controls in mESC nuclei with purified E12-AF647
Having a purified version of E12-AF647, it was important to repeat the G4 imaging binding
controls on the new version of the probe. The AR082 batch of E12-AF647 showed the same
general trend as previous probe mixture, with a decrease of observed foci after benzonase and
E12 pre-block treatments (Figure: 3.25).
5 n
M
Be
nz
on
as
e
80
0 n
M 
E1
2 p
re-
blo
ck
6 n
M
5 µ
M 
E1
2 p
re-
blo
ck
Bl
an
k
Liv
e c
ell
 D
MS
 20
 m
M
0
1000
2000
3000
4000
5000
6000
O
bs
er
ve
d 
fo
ci
G4 binding controls purified E12-AF647 multiple batches
AR082 at 5 nM II-42 at 6 nM
Figure 3.25: Analysis of observed foci count of E12-AF647 under different control conditions. Two
batches of E12-AF647 were used - AR082 which was A¨KTA FPLC purified once, followed by dialysis,
and II-42 batch which was A¨KTA FPLC purified twice, followed by dialysis. Error bars indicate mean
± sd. n = 3, 1, 3, 36, 29, 10 and 15 nuclei for 5 nM, benzonase, 800 nM E12 pre-block, 6 nM, 5 µM E12
pre-block, blank and live cell DMS 20 mM conditions respectively, taken from 1 independent replicate.
Imaging experiments were performed in collaboration with Edward Sanders and Ziwei Zhang.
However, it was expected that removal of higher labelled species (as compared to imaging with
labelled products probe mixture) would make the observed foci decrease of 59 % to be more
stark upon E12 pre-block treatment. The control imaging experiments with another batch
of E12-AF647 (II-42) were therefore done with a higher E12 pre-block concentration (5 µM
as compared to 800 nM previously). Unexpectedly, the difference observed was smaller than
116
Chapter 3: 3D STORM super-resolution imaging of G4s
expected - a 38 % decrease10. It is unclear whether the difference could arise from experimental
variability, or possibly a higher concentration of E12 could induce G4 formation in fixed cells,
this way creating an opposing influence of blocking G4 sites.
The inclusion of 20 mM DMS treatment to trap unfolded G4 state in live cells (30 min treat-
ment) did not decrease the number of observed localisations. A key experimental difference
compared to when the study was done on U2OS cells with SiR-PyPDS (19) was avoiding the use
of β-mercaptoethanol for quenching of DMS in case of killing the cells with β-mercaptoethanol
before fixation. Instead, DMS treated cells were fixed straight away with 2 % formaldehyde
and quenched only after further 10 min with glycine (which quenched both formaldehyde and
DMS simultaneously). Indeed, straight after DMS treatment 95 % of cells were alive according
to trypan blue staining. But overall, this control did not work as intended.
3.5.2 3D image acquisition with drift correction
The next major objective was to collect the first 3D G4 images with drift-correction in STORM.
Moving from using labelled product mixture to a purified probe required labelling density re-
optimisation and it was found that 6 nM of purified E12-AF647 provided the best results in
comparison to the probe mixture used previously. A first drift-corrected 3D image has was
obtained (Figure: 3.26, a, b). 3D localisations were obtained by using DHPSF method at
full depth of field of a single slice of mESC nucleus. The image contains 28 000 localisations
taken over 40 000 frames, 30 ms exposure. In the image, the white yellow colour represents
localisations which had >20 neighbouring localisations within 100 nm radius. This was a means
to visualise clustering in the image, which demonstrated that the E12-AF647 probe clusters in
3D space. However, more data are required to confirm that these clusters actually represent
G4s or are consistent features in different nuclei.
Drift-correction was carried out in x,y plane by cross-correlating red channel images of AF647
dye with green channel images of the whole cell. It was performed by taking 5 frames of
green channel images every 200 frames on a red channel (corresponds to about 6 s at 30 ms
exposure, therefore any nucleus movement was recorded every 6 seconds). Any drift observed
then is corrected to the original 3D image. Over ∼half an hour, the nucleus was seen to have
drift amplitudes of about 200 nm in x-axis and 40 nm in y-axis (Figure: 3.26, c). This is
rather small in comparison with the scale of the whole nucleus with a radius of 6 µm, but it
is significant with regard to STORM resolution which generally is about 20 nm. Pre-corrected
and post-corrected images look almost identical as a whole, but some changes can be better
10Note a vastly greater number of cells imaged with II-42 batch, this was due to switching to coated protein
LoBind R© centrifugation tubes from non-coated polypropylene tubes, as discussed previously (Section: 3.3.2).
117
Chapter 3: 3D STORM super-resolution imaging of G4s
a b
c
d
Figure 3.26: Drift-corrected 3D image of a mESC nucleus slice. 28 000 localisations. (a) Top view.
(b) Side view. White and yellow colour represent localisations which have >20 neighbours within 100
nm radius. Axes denote distance in µm. (c) Plot displaying the overall drift of the whole nucleus
over the course of imaging. (d) Pre- and post-corrected images of a mESC nucleus. Correction
differences are best visualised at the edge of the image highlighted in squares. Imaging was performed
in collaboration with Edward Sanders and Ziwei Zhang, drift-correction processing was carried out by
Aleksandra Jartseva. 118
Chapter 3: 3D STORM super-resolution imaging of G4s
visualised at the side of the image (Figure: 3.26, d). Z-axis drift is technically not corrected,
however it is minimised by using a perfect focus stage control on the microscope. It measures
the z position of the imaging plate with great precision using an infrared LED and adjusts the
focus of the microscope accordingly, so any drift in z would be accounted by that. However, it
relies on having the nucleus being perfectly attached to the plate and having the two moving
synchronously, which is assumed to generally be the case.
3.5.3 Whole nucleus 3D imaging
Initial images taken in 3D for optimisation purposes were of single slices of mESC nucleus. A
slice is an observable z-axis depth of field (∼4 µm) at a given focus plane observed by DHPSF
method. Whole nucleus in 3D was obtained by summing the imaged slices taken at different z
planes (Figure: 1.8).
Whole nucleus 3D images were obtained with using 6 nM E12-AF647 (II-42 batch). The
image was combined from 12 imaged slices, slightly overlapping with one another to exclude
localisations at extreme double-helix angles (Figure: 3.27). 5 000 frames at 30 ms exposure
were taken per slice and 6 images of whole nuclei were taken in total11. Clustering can be seen
at all 6 nuclei imaged, with the number of localisations in the images being between 10 000 and
18 000. However, more nuclei and preferably with more localisations should be imaged for a
robust clustering analysis, but issues with E12-AF647 acquisition emerged, and are discussed
next.
Section summary
This section presented 3D super-resolved STORM images obtained with E12-AF647 probe.
Drift-correction was utilised for more precise imaging. Preliminary E12-AF647 3D data in-
dicates that the probe could cluster in 3D space, but additional data is required for robust
analysis.
11For additional images see the appendix.
119
Chapter 3: 3D STORM super-resolution imaging of G4s
bba
Figure 3.27: Whole nucleus 3D super-resolved images of E12-AF647. 18 000 localisations. (a) Top
view. (b) Side view. Axes denote distance in µm. White and yellow colour represent localisations
which have >20 neighbours within 100 nm radius. Experiments were carried out in collaboration with
Ziwei Zhang.
3.6 Issues with E12-AF647 imaging and acquisition
In this section, problems that emerged with E12-AF647 imaging and acquisition are discussed.
Key challenges were:
• High nucleus autofluorescence
• E12-AF647 degradation
• Synthesis of G4 active E12-AF647
• Reactivation of E12-AF647 by heat induced refolding
3.6.1 Autofluorescence issues
Attempts to obtain more reliable 3D data were unsuccessful due to an emerged problem with
nuclei autofluorescence. Blank nuclei showed a comparable fluorescence output to labelled
nuclei and E12 pre-block control, also with substantially higher brightness than blank nuclei in
earlier imaging experiments (Figure: 3.28, a-d). Consequently, DNAse I and benzonase control
120
Chapter 3: 3D STORM super-resolution imaging of G4s
E12 pre-block
DNAse I
Autofluorescent blank
Benzonase
6 nM E12-AF647
Blank
a b
c d
e f
Figure 3.28: Maximum intensity z-projections of videos of E12-AF647 (II-42) labelled sample. Im-
ages taken after bleaching for 500 frames (15 s) at maximum laser power. (a) 6 nM E12-AF647 labelled
reference sample. (b) Unlabelled blank nucleus with high autofluorescence. (c) E12 pre-block control.
(d) Unlabelled blank nucleus showing the usually low levels of autofluorescence in other experiments.
(e) DNAse I digestion control. (f) Benzonase digestion control. Experiments were carried out in
collaboration with Ziwei Zhang.
121
Chapter 3: 3D STORM super-resolution imaging of G4s
treatments showed no reduction of observed foci (Figure: 3.28, e, f). Counting localisations
showed that the blank had half the localisations of a labelled sample (Figure: 3.29).
6 n
M
Be
nz
on
as
e
Bl
an
k
0
500
1000
1500
O
bs
er
ve
d 
fo
ci
G4 imaging E12-AF647 (II-42)
Figure 3.29: E12-AF647 (II-42) localisations on highly autofluorescent cells which make G4 imaging
unreliable. Error bars indicate mean ± sd. n = 15, 10 and 7 nuclei for 6 nM, benzonase and
blank conditions respectively, taken from 1 independent replicate. Experiments were carried out in
collaboration with Ziwei Zhang.
It is unknown why the cell and nucleus became autofluorescent. Passaging the same mESC
cultures for longer at later passages still demonstrated the autofluorescent behaviour. How-
ever, restarting the culture from a new frozen stock eliminated the issue and blank cells then
showed minimal autofluorescence. This suggested that the issue may be due to cell stress
and/or contamination in the culture, but no unusual changes were seen by light microscopy
and contamination tests for mycoplasma, bacteria and fungi turned out to be negative. Never-
theless, while these issues compromised much of 3D data obtained, the causes were not further
investigated since restarting of cell culture was a working solution.
3.6.2 Probe degradation
Another imaging session on G4 binding controls in mESC nuclei revealed a different problem.
This time nucleus labelling with the probe seemed to have worked as the blank sample was not
autofluorescent anymore and reference 6 nM labelled sample showed a much higher number
of localisations than the blank (Figure: 3.30). However, none of the G4 binding controls
showed the expected decrease of E12-AF647 localisations. In particular, benzonase treatment
control which showed 99 % reduction of observed foci previously (Figure: 3.17), now showed a
comparable number of observed foci as 6 nM reference treatment.
122
Chapter 3: 3D STORM super-resolution imaging of G4s
6 n
M
Bl
an
k
E1
2 p
re-
blo
ck
 5 
µM
E1
2 p
re-
blo
ck
 80
0 n
M
BG
4 p
re-
blo
ck
 80
0 n
M
Be
nz
on
as
e
DN
As
e I
RN
As
e A
Liv
e 2
0 m
M 
DM
S b
lan
k
Liv
e 2
0 m
M 
DM
S
Fix
ed
 D
MS
 60
0 m
M
0
1000
2000
3000
4000
5000
O
bs
er
ve
d 
fo
ci
G4 binding controls SecPur II-42 E12-AF647
Figure 3.30: G4 binding controls with a degraded batch of E12-AF647 (II-42). Error bars indicate
mean ± sd. n = 29, 10, 29, 30, 20, 30, 25, 30, 9, 19 and 30 nuclei for presented conditions respectively,
taken from 1 independent replicate.
Later it was found by MS that the probe had degraded (Figure: 3.31). This still seemed
strange that a degraded probe moiety containing the dye finds a tight binding partner that is
not washed out during washing steps while also not being specific to DNA as shown by DNAse
I and benzonase treatments. This was of concern as a partially degraded probe could skew the
G4 imaging results.
3.6.3 FPLC purification optimisation
After probe degradation, additional probe synthesis and purification was required. Unfortu-
nately, initial purifications achieving 99 % purity of E12-AF647 could not be repeated as other
purifications yielded∼90 % purity, with 10 % consisting of unlabelled E12. Therefore, to achieve
the previous standard of purity, optimisation of A¨KTA FPLC purifications was required.
123
Chapter 3: 3D STORM super-resolution imaging of G4s
a
b
Figure 3.31: Comparison of MS of E12 nanobody: (a) before degradation, (b) after degradation.
The following general changes were used to improve E12-AF647 purification:
1. Reduce the relative amount of unreacted E12 by increasing the equivalents of reactive
dye, more higher labelled species would be obtained but they are easier to separate from
1X labelled probe.
2. Lengthen the column by joining 3 HiTrap columns in sequence to give broader but also
better separated peaks.
3. Holding the gradient at a lower salt concentration of 250-300 mM NaCl (as compared to
350-440 mM used previously).
4. A second sequential purification on fractions containing majority of E12-AF647.
5. Use amicon concentration spin columns (3.5 kDa cut-off) to concentrate and desalt the
final purified fractions. This step should remove any residual unreacted AF647 dye.
6. Reaction and purifications were optimised for ∼1 mg of E12, to balance recovery yields
with separation of the peaks.
124
Chapter 3: 3D STORM super-resolution imaging of G4s
These changes greatly improved general peak separation, however, peaks also became much
broader and the product eluted at much lower concentrations (about 1 µM), which became too
low to detect by MS and required concentration with spin columns. Primary purification UV
chromatograms generally showed a purified E12 peak eluting first, followed by a double peak of
E12 and E12-AF647 (1X) mixture, and last - large peak consisting of unreacted AF647 dye and
higher labelled E12 species (Figure: 3.32, a). The E12 (∼25 %) and E12-AF647 (1X, ∼75 %)
mixture was then purified a second time, which also showed a similar peak order, E12 and
E12-AF647 (1X) still co-eluting with 10-90 ratio, but some fractions containing >95 % purity
E12-AF647 (Figure: 3.32, b). A smaller 3rd dye peak, about a third of total AF647 signal,
suggested that not all of AF647 dye was removed by primary purification. The eluted fractions
from secondary purification were also desalted and concentrated by amicon spin-columns (3.5
kDa cut-off membrane) to remove any remainder of AF647.
a b
Take E12-AF647 fractions
for a secondary purification
Final fractions have >95% E12-AF647
Figure 3.32: UV-vis chromatograms at 280 nm and 650 nm of optimised anion exchange purifica-
tions. (a) Primary purification chromatogram. (b) Secondary purification chromatogram. Blue and
purple traces denote 280 nm and 650 nm chromatograms respectively, green line denotes relative NaCl
concentration, orange line - conductivity of the solution.
Moreover, after increasing the labelling reaction to 2.5 equivalents of the dye, a significant
amount of unlabelled E12 remained in the mixture, giving compositions of 12 % E12, 70 %
E12-AF647 (1X), 12 % E12-AF647 (2X), 6 % E12-AF647 (3X). Since co-elution of E12 and
E12-AF647 (1X) was the primary problem of purification, the reaction was then attempted
with 5 equivalents of AF647-NHS, in attempt to react unlabelled E12 to completion. Though
the unlabelled E12 part in the mixture was successfully decreased, the E12-AF647 (1X) started
to co-elute with E12-AF647 (2X). This result suggested that E12 and its labelled species might
associate with each other and that their complete separation is difficult.
125
Chapter 3: 3D STORM super-resolution imaging of G4s
3.6.4 Probe inactivation after FPLC purification
Optimisation of A¨KTA FPLC purification has led to obtaining >95 % purity E12-AF647 mixed
with unlabelled E12. The yields also were improved with up to 200 µg of purified probe
being isolated from 2 mg of E12 reaction. However, testing the newly purified batches of
E12-AF647 by ELISA, revealed that E12-AF647 had lost G4 binding activity when testing
against both FLAG- and His-tag (Figure: 3.33). A much lowered binding curve of E12-AF647
PriPur and SecPur (PriPur and SecPur indicate products obtained after primary and secondary
purifications respectively) might be explained by residual 3-6 % of unlabelled E12 in the mixture
(Table: 3.7). Furthermore, only the labelled antibody showed a decrease in activity as the
unreacted E12 fractions from primary and secondary purifications (therefore having undergone
the same treatment as labelled E12-AF647) demonstrated similar Kds to native E12.
a
1 10 100
0.00
0.05
0.10
Concentration / nM
ELISA on Myc using FLAG-tag HRP
Silvia E12 63.6 µM
E12-AF647 SecPur A3-A4
E12 SecPur
E12-AF647 PriPur
E12-AF647 SecPur A5-A6
E12-AF647 SecPur A7-A8
A
bs
or
ba
nc
e
b
1 10 100
0.0
0.1
0.2
Concentration / nM
A
bs
or
ba
nc
e
ELISA on Myc using FLAG-tag HRP
E12-AF647 PriPur
E12-AF647 SecPur A7-A8
E12-AF647 SecPur A5-A6
E12 SecPur
Figure 3.33: ELISA on Myc G4 of AR092 batch of E12-AF647 purification using: (a) FLAG-tag
HRP for detection (b) His-tag HRP for detection. See table 3.7 for dissociation constants. Error bars
indicate standard error of the fit obtained from at least two replicates.
Batch Species Kd / nM E12 / % E12-AF647 / %
II-5 E12 native 5.1 ± 0.4 100 0
AR090 E12-XP SecPur 8.4 ± 3 100 0
AR090 E12-AF647 SecPur 2A2 40 ± 20 5 95
AR090 E12-AF647 SecPur 310 ± 200 2 98
II-5 E12 native 3.1 ± 0.6 100 0
AR091 E12-XP SecPur 14 ± 5 100 0
AR091 E12-AF647 SecPur B7-B8 250 ± 100 24 76
Continued on next page.
126
Chapter 3: 3D STORM super-resolution imaging of G4s
Batch Species Kd / nM E12 / % E12-AF647 / %
AR092 (FLAG) E12-XP PriPur 6.9 ± 0.4 100 0
AR092 (FLAG) E12-AF647 PriPur 77 ± 20 5 95
AR092 (FLAG) E12-XP SecPur 11 ± 1 100 0
AR092 (FLAG) E12-AF647 A3-A4 78 ± 10 6 94
AR092 (FLAG) E12-AF647 A5-A6 190 ± 50 5 95
AR092 (FLAG) E12-AF647 A7-A8 >1200 3 97
AR092 (His) E12-AF647 PriPur 100 ± 20 5 95
AR092 (His) E12-XP SecPur 16 ± 0.8 100 0
AR092 (His) E12-AF647 A5-A6 210 ± 40 5 95
AR092 (His) E12-AF647 A7-A8 >1300 3 97
II-108 E12 native 0.59 ± 0.06 100 0
AR093 E12-XP PriPur 1.00 ± 0.09 100 0
AR093 E12-AF647 PriPur 14 ± 2 4 96
AR093 E12-AF647 SecPur 38 ± 4 4 96
AR094 E12-XP PriPur 0.98 ± 0.08 100 0
AR094 E12-AF647 PriPur 8.2 ± 0.5 5 95
AR094 E12-AF647 SecPur 86 ± 20
II-128 E12 native 2.4 ± 0.2 100 0
AR095 E12-XP PriPur 4.0 ± 0.2 100 0
AR095 E12-AF647 PriPur 63 ± 10 3 97
AR095 E12-AF647 SecPur D1-D3 42 ± 4
AR095 E12-AF647 SecPur D4-D6 92 ± 20 0.4 99.6
AR095 E12-AF647 SecPur D7-E1 120 ± 30
AR096 E12-XP PriPur 1.7 ± 0.1 100 0
AR096 E12-AF647 PriPur 14 ± 2 12 88
AR099 E12-XP PriPur 0.91 ± 0.07 100 0
AR099 E12-AF647 PriPur A6 14 ± 2 10 90
AR099 E12-AF647 PriPur A7 31 ± 4 10 90
AR099 E12-AF647 PriPur A8 35 ± 6 7 93
AR099 E12-AF647 PriPur B1 14 ± 1 9 91
AR100 E12-XP PriPur 0.31 ± 0.03 100 0
AR100 E12-AF647 PriPur1 35 ± 5 4 96
AR100 E12-AF647 PriPur2 21 ± 3 12 88
Table 3.7: ELISA results of different purification batches and the corresponding E12 nanobody
species composition. Errors indicate standard error of the fit obtained from at least two replicates.
127
Chapter 3: 3D STORM super-resolution imaging of G4s
These data could be interpreted as inactivation of the probe by labelling with AF647, however,
early data with several batches of E12-AF647 showed activity to G4s by both ELISA and
imaging in nuclei, therefore it was likely that the E12-AF647 purification process was causing
deactivation for the newest batches. Consequently, different modifications to the purification
protocol were attempted to troubleshoot potential problems:
• Quicker purifications to shorten the time nanobody spends on the column.
• pH 8.5 and pH 8 purifications tested.
• Starting anion exchange gradient at 100 mM NaCl rather than at 0 mM to avoid nanobody
experiencing low salt environment.
• Testing larger 5 mL HiTrap columns to avoid concentrating the nanobody in small volume.
• For fraction desalting - switching between dialysis and concentrating amicon spin columns.
• Checking activity before and after flash freezing (carrying out 1 freeze-thaw cycle).
However, none of the modifications regained active E12-AF647 probe. This raises concerns
about the validity of 3D data taken with earlier batches (AR082 and II-42) of the probe.
Because of this, it was decided to test G4 binding controls with a G4 inactive E12-AF647
probe batch AR090. As expected, G4 controls did not suggest G4 binding in mESC nuclei
since E12-pre-block treatment showed no difference in observed foci (Figure: 3.34). While BG4
pre-block showed 39 % decrease, the BG4 control in isolation is not convincing to suggest G4
binding. DMS treatments on fixed cells did not show a difference either, both when performed
at 37 ◦C and 85 ◦C, higher temperature was tested to attempt to unfold G4 structures by heat,
hoping to trap the unfolded structure. DNAse I and RNAse A both led to ∼70 % reduction
which suggested DNA and RNA binding for the G4 inactive E12-AF647, while active probe
previously showed DNA binding exclusively (Figure: 3.15), but benzonase treatment (which
digests both DNA and RNA) showed little difference. This suggests that one of these treatments
on DNA and RNA were not effective.
Overall, the G4 controls did not suggest G4 binding with an inactive E12-AF647 probe, and
that the controls do work as intended to show G4 binding ability. Therefore there appears to
be an issue with the probe and not the controls.
128
Chapter 3: 3D STORM super-resolution imaging of G4s
Figure 3.34: G4 binding controls with a G4 inactive AR092 batch of E12-AF647 probe. Error bars
indicate mean ± sd. n = 20, 10, 9, 29, 10, 10, 9, 8 and 10 nuclei for presented conditions respectively,
taken from 1 independent replicate.
3.6.5 Thermal reactivation of E12-AF647
Based on a hypothesis that E12-AF647 inactivation could be caused by unfolding of the
nanobody and reports that camelid nanobodies can be thermally refolded [359], refolding inac-
tive E12-AF647 batches might make the nanobodies regain their G4 binding ability. Thermal
refolding experiments were performed by keeping the E12-AF647 sample at elevated temper-
ature for a specific period of time, then followed by slow cooling by 1 ◦C / min to room
temperature (RT). Initial experiments on AR093 batch of E12-AF647 showed a decrease of
Kd to 7.7 nM down from 31 nM after treating the sample for 30 minutes at 80
◦C (Figure:
3.35, a, Table: 3.8, a). The 7.7 nM Kd cannot be accounted by 4 % of unlabelled E12 in the
batch, suggesting E12-AF647 reactivation. The encouraging result led to further optimisation
of the thermal refolding process. Longer heat treatments up to 3 h were attempted, but those
showed only a small difference than when performed at 80 ◦C, whereas an increase of Kd was
seen at 90 ◦C (Table: 3.8, b). The latter suggested that nanobody can aggregate at higher
concentrations and that a balance between refolding and aggregation needs to be found. This
was also supported by higher concentration batches of E12-AF647 having a lesser reduction
129
Chapter 3: 3D STORM super-resolution imaging of G4s
of Kd upon thermal treatment. It was considered whether unlabelled E12 could also have its
binding ability altered. While none of the thermal treatments on E12 led to tighter binding to
G4s, at higher E12 concentrations than 1 µM, a decrease of binding was observed, most likely
due to aggregation (Table: 3.8, c). A 30 minute treatment at 80 ◦C to 90 µM E12, completely
abolished its G4 binding activity. Further experimets found that optimal thermal refolding
conditions were 30 min treatment at 80 ◦C at 1 µM concentration, giving up to 30-fold stronger
binding affinity to Myc G4 (Table: 3.8, d, e, Figure: 3.35).
a
1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
ELISA on Myc using FLAG-Tag HRP
AR093 E12-AF647 SecPur
80 oC 10 min
80 oC 30 min
A
bs
or
ba
nc
e
b
1 10 100
0.00
0.05
0.10
0.15
Concentration / nM
ELISA on Myc using FLAG-Tag HRP
Native E12
A
bs
or
ba
nc
e
AR092 SecPur-A5-A6 80 oC 30 min
AR092 SecPur-A5-A6 thawed
AR092 SecPur-A5-A6 1 µM 80 oC 30 min
Figure 3.35: (a) ELISA binding curves showing an increased binding ability of E12-AF647 to Myc
G4 upon thermal treatment. (b) ELISA binding curves showing E12-AF647 increased binding ability
to Myc G4 after optimised thermal treatment. See table 3.8 for dissociation constants. Error bars
indicate mean ± sd.
The reactivated E12-AF647 was then used for imaging in mESC nuclei. Although the nuclei
could be stained with the probe, neither DNAse I digestion or E12 pre-block controls showed
a clear decrease of foci, suggesting other binding targets than DNA G4s (Figure: 3.36). Nev-
ertheless, it was puzzling why native E12 does not block labelled E12-AF647 access to the
nucleus. Also, why initial active E12-AF647 batches demonstrated G4 binding by G4 imaging
controls while the new thermally reactivated E12-AF647 did not.
130
Chapter 3: 3D STORM super-resolution imaging of G4s
Part Batch Treatment Kd / nM
a AR093 3.9 µM None 31 ± 3
AR093 3.9 µM 70 ◦C 10 min 23 ± 3
AR093 3.9 µM 80 ◦C 10 min 13 ± 2
AR093 3.9 µM 90 ◦C 10 min 16 ± 2
AR093 3.9 µM 80 ◦C 30 min 7.7 ± 0.9
AR093 3.9 µM 90 ◦C 30 min 12 ± 2
b AR092 A7-A8 7.4 µM None >1200
AR092 A7-A8 7.4 µM None 44 ± 6
AR092 A7-A8 7.4 µM 80 ◦C 15 min 22 ± 3
AR092 A7-A8 7.4 µM 80 ◦C 3 h 28 ± 5
AR092 A7-A8 7.4 µM 90 ◦C 3 h 310 ± 70
AR092 A7-A8 1.0 µM 80 ◦C 30 min 20 ± 1
AR092 A7-A8 1.0 µM 80 ◦C 3 h 15 ± 2
c E12 90 µM None 4.9 ± 0.3
E12 90 µM 80 ◦C 30 min No binding
E12 6.6 µM None 0.83 ± 0.5
E12 6.6 µM 80 ◦C 30 min 7.2 ± 3
E12 1.0 µM None 1.8 ± 0.6
E12 1.0 µM 80 ◦C 30 min 1.8 ± 0.8
d AR092 A5-A6 13 µM None 67 ± 13
AR092 A5-A6 13 µM 80 ◦C 30 min 67 ± 13
AR092 A5-A6 1.0 µM 80 ◦C 30 min 7.9 ± 1
AR092 A5-A6 1.0 µM 80 ◦C 3 h 8.6 ± 2
e AR095 D4-D6 2.6 µM None 85 ± 25
AR095 D4-D6 1 µM 80 ◦C 30 min 2.8 ± 2
Table 3.8: ELISA results of different batches of E12-AF647 subjected to thermal treatment. ELISAs
were performed on Myc G4. Errors indicate standard error of the fit from at least two replicates.
In conclusion, the active version of E12-AF647 as used for initial experiments, could not be
re-produced. Even though thermal treatment of the probe could reactivate its binding ability
to Myc G4 in ELISAs, the probe did not label DNA G4 structures in mESC nuclei. Moreover,
the latest thermal reactivation results reduced the overall confidence in E12-AF647 as a G4
probe as consistent data was difficult to obtain. These issues with E12-AF647 acquisition have
led to a change in the G4 labelling approach and pivot to a different probe.
131
Chapter 3: 3D STORM super-resolution imaging of G4s
a b c
Figure 3.36: Imaging G4 controls do not suggest DNA G4 binding. Images of mESC nuclei using a
heat refolded E12-AF647 (30 min at 80 ◦C). Background-subtracted maximum intensity projections
from 1000 frames at 30 ms exposure. (a) Reference treatment with 6 nM of E12-AF647 AR095 D4-D6
batch. (b) DNAse I pre-treatment. (c) 2 µM E12 pre-block.
Section summary
Problems with E12-AF647 imaging and acquisition were discussed. mESC nuclei autofluo-
rescence diminished the quality of single-molecule imaging, which was eventually resolved by
restarting cell cultures. Then the remainder of E12-AF647 was lost due to degradation. A¨KTA
FPLC purification optimisation was described in efforts to re-obtain a purified E12-AF647
probe. Two-sequential purifications showed best results, however the probe was found to be
inactive to G4s by ELISA and by G4 imaging controls in mESC nuclei. Heat refolding led to
reactivation of E12-AF647 in ELISAs, however for imaging this did not demonstrate DNA G4
binding. Overall, these issues led to a need for an alternative and more reliable probe.
3.7 Two antibody layer G4 imaging with BG4
In this section, an alternative method for G4 staining in mESC nuclei with two layers of
antibodies in an indirect IF are described. Imaging controls are presented to demonstrate G4
binding, followed by presentation of 3D STORM imaging results.
BG4 was chosen as the principal probe for G4 recognition in fixed cells. BG4 is a well-established
G4 probe for IF imaging [59, 60, 360] and sequencing techniques - ChIP-seq, CUT&Tag [61,
77]. As earlier direct labelling approaches did not provide results with BG4 (Section: 3.2.2),
two antibody layer IF imaging was planned to be used. A secondary monoclonal rabbit-IgG
antibody labelled with AF647 was to target FLAG-tag on BG4.
132
Chapter 3: 3D STORM super-resolution imaging of G4s
3.7.1 Controls and optimisation
With a new staining method, re-optimisation of imaging is required, as well as demonstration of
G4 binding in cells with control experiments. Imaging mESC nuclei with 10 nM of BG4 followed
by 1 µg/mL of SecAB-AF647 gave 828 observed foci on average over 1000 frames (taken after
1000 frames of pre-bleaching at 30 ms exposure), which was close to the desired number of
localisations per frame for 3D STORM (which is 1-2). The average number of observed foci
decreased by 87 % and 82 % with DNAse I and benzonase digestions respectively, suggesting
BG4 binding to DNA (Figure: 3.37).
BG
4 1
0 n
M
DN
As
e B
G4
 10
 nM
Be
nz
on
as
e B
G4
 10
 nM
Se
cA
B 
on
ly
BG
4 1
 nM
BG
4 1
0 n
M
BG
4 1
 nM
Se
cA
B 
on
ly
0
500
1000
1500
O
bs
er
ve
d 
fo
ci
BG4 SecAB-AF647 
SecAB-AF647 1 µg/mL SecAB-AF647 0.1 µg/mL
Figure 3.37: Optimisation of BG4 and SecAB-AF647 imaging of G4s in mESC nuclei. DNAse I and
benzonase pre-treatment controls demonstrated DNA binding. Error bars indicate mean ± sd. n = 11,
15, 13, 11, 14, 25, 15 and 12 nuclei for presented conditions respectively, taken from 1-2 independent
replicates.
SecAB-AF647 treatment at 1 µg/mL without BG4 produced 86 % fewer foci localisations,
demonstrating the level of background noise stemming from the secondary antibody. Impor-
tantly, DNAse I and benzonase digestions led to signal to drop to this background level, suggest-
ing that the BG4 signal primarily arises from binding to DNA. Imaging with a 10 times lower
concentration of SecAB-AF647 at 0.1 µg/mL with 10 nM of BG4 gave 310 foci localisations on
average and secondary antibody alone had 94 % less localisations, therefore these conditions
had a better signal-to-noise ratio. However, the number of observed foci per frame was too low
for effective 3D STORM imaging, thus larger concentrations of both BG4 and SecAB-AF647
are required despite a poorer signal-to-noise ratio.
133
Chapter 3: 3D STORM super-resolution imaging of G4s
3.7.2 3D STORM imaging
BG412 and anti-FLAG-tag IgG-AF647 antibody labelling of mESC nuclei was visualised by
3D STORM using DHPSF method (Figure: 3.38). Initial imaging attempts provided too few
localisations per slice to see clear clustering of the probe on a single slice, nevertheless some
small regions of increased density could be distinguished. A higher nucleus background signal
has emerged when compared to imaging with E12-AF647. This reduces the imaging quality
and consequently the peak calling, hence the reduced number of localisations.
a b
Figure 3.38: 3D super-resolution image of a slice of mESC nucleus labelled with 10 nM BG4 and
1 µg/mL of anti-FLAG-AF647 secondary antibody (rabbit monoclonal). 8 000 localisations. Yellow-
white regions indicate localisations which have >20 neighbours within 100 nm radius. (a) Front view.
(b) Side view. Axes denote distance in µm.
Due to a persistent high background, re-optimisation of labelling protocol was carried out and
best blocking conditions were found (1 % BSA (bovine serum albumin), 0.1 % Tween20 in PBS
pH 7.4)13. Moreover, it was found that cell labelling at 37 ◦C introduced autofluorescence in
the nucleus, labelling at lower temperatures (initially RT, then 0 ◦C) substantially improved
signal-to-noise ratio.
Further work continued with imaging whole mESC nuclei while summing multiple slices ob-
tained at different heights. The best images had up to 20 000 localisations and showed mild
12BG4 used in this section has been kindly provided by Isabel Esain Garcia.
13Previously 5 % goat serum, 0.1 % Tween20 in PBS pH 7.4 was being used.
134
Chapter 3: 3D STORM super-resolution imaging of G4s
clustering of the probe in some parts of the nucleus, while generally staining most of its vol-
ume (Figure: 3.39). 8 nuclei have been imaged with good quality, however a larger number is
required for performing more robust 3D analysis, especially since some of the nuclei had too
few localisations for clear observation of probes clustering14.
a b
Figure 3.39: 3D super-resolution image of whole mESC nucleus labelled with 20 nM BG4 and 1
µg/mL of anti-FLAG-AF647 secondary antibody (rabbit monoclonal). 20 000 localisations. Yellow-
white regions indicate localisations which have >10 neighbours within 100 nm radius. (a) Front view.
(b) Side view. Axes denote distance in µm.
Section summary
G4 structures were visualised in extracted mESC nuclei by indirect IF using two antibody
layers - BG4 and a secondary anti-FLAG rabbit IgG antibody labelled with AF647. Control
experiments demonstrated visualisation specificity to DNA and that the secondary antibody
has minimal contribution to the background noise. G4 visualisation by 3D STORM revealed
that the probe stained majority of the nucleus volume with mild clustering at specific locations.
14Additional 3D STORM images can be found in the appendix.
135
Chapter 3: 3D STORM super-resolution imaging of G4s
3.8 G4 genomic CUT&Tag map overlap with single-cell
Hi-C structures
The major aim of this project - is to overlap of 3D G4 fluorescence image and simulated Hi-C
chromatin structure in the same single cell to provide approximate G4 density measures at
specific genome sequence regions. As sequencing data arises from Hi-C, the precision of G4
sequence position would depend on the resolution of single-cell Hi-C, which currently is down
to ∼20 kb [279]. This also depends on the 3D imaging and image overlap quality, but it is
difficult to quantify this limitation in terms of sequencing resolution. These limitations and
the novel nature of the technique being developed demands a means of cross-validation for the
accuracy of the approach.
Cross-validation can be provided by an alternative G4 sequencing technique such as G4-ChIP-
seq [61] or the newer G4-CUT&Tag [77]. Though these methods are generally applied to an
ensemble of cells, they can provide a reference map to compare to ones determined by G4
image and Hi-C overlap in a single-cell. Furthermore, both G4-ChIP-seq and G4-CUT&Tag
have better 150-500 base resolution when compared to Hi-C with up to 20 kb resolution. This
might indicate which G4s could be forming within the 20 kb region (or larger, depending on the
resolution achieved). Consequently, sequencing of G4s in mESC was desired and G4-CUT&Tag
was picked due to its lower cell input requirements than G4-ChIP-seq.
Overlap of population G4 sequencing data on single-cell Hi-C data may provide new insights
in mESCs. G4 associations to 3D chromatin structure could be investigated genome wide,
but it could also provide a picture of G4s in single cells. Individual G4 contacts and struc-
ture associations could be followed in single-cells, notifying any significant differences between
cells. Coupling these differences to G4 peak intensity as proxy for G4 abundance across a cell
population, could signify G4 formation variability with respect to chromatin structure.
G4-CUT&Tag sequencing data discussed in this section has been obtained by Isabel Esain
Garcia and Dr Angie Kirchner. Bioinformatics analysis was performed by Dr Angela Simeone.
Single-cell Hi-C data was provided by Ernest Laue’s group working in collaboration with Dr
David Lando. My own contribution was leading and planning the overall study which was part
of my plan in the overall investigation of G4s in 3D chromatin. I established the collaboration
with the Laue group and carried out 3D structural visualisation analysis.
136
Chapter 3: 3D STORM super-resolution imaging of G4s
3.8.1 mESC differentiation states
Our collaborator Ernest Laue’s group investigate mESC differentiation by implementing single-
cell Hi-C to follow chromatin structure transformation during the differentiation transition (un-
published work). By following G4 landscape variation by CUT&Tag, 3D chromatin structure
association with G4s can be established, through analysing experimentally induced perturba-
tions to differentiation. These experiments could be also related to a study in hESC where
large G4 landscape changes were reported during differentiation [71].
Differentiation was induced by removing 2i/LIF conditions from the mESC culture media ac-
cording to previous reports [357]. Throughout the transition na¨ıve, formative and primed cell
states were investigated (Figure: 3.40).
Figure 3.40: A diagram of mESC differentiation states. The na¨ıve state closely represents E4.5 stage
of mouse embryonic development and formative state closely represent E5.5 stage. Figure shared by
Dr David Lando.
The na¨ıve state closely represents epiblast cells of E4.5 stage of mouse embryonic development,
while the formative state represents post-implantation epiblast cells of the E5.5 stage. The for-
mative state, which forms after 24 hours after loss of differentiation inhibitors, can be regarded
as a tipping point of differentiation. The pluripotency marker Rex-1 expression levels, correlate
with cells committed to a differentiation pathway (Rex-1 at low levels) or are still capable of
reversing back to the na¨ıve state (Rex-1 at high levels). Consequently, the formative state is of
a mixed-population, a loosely defined state according to the time spent without 2i inhibitors.
The primed state forms after another 24 hours and is regarded as committed towards neuronal
differentiation pathway.
137
Chapter 3: 3D STORM super-resolution imaging of G4s
3.8.2 G4 map changes during differentiation
The different mESC states had their G4s mapped by CUT&Tag. Peaks observed in two out
of three biological replicates, termed consensus peaks, showed an increase of G4 structures
upon differentiation induction from the na¨ıve state (Figure: 3.41, a). The peaks of the three
biological replicates clustered together for the corresponding mESC states, indicating experi-
mental reproducibility and G4 landscape differences between the states. A subset of 10 000
G4s was present in all of the states, but each state also had unique subsets (Figure: 3.41, b).
Different groups of G4s can now be analysed to study how chromatin structure changes during
differentiation transitions coupled to G4 formation or loss at a genomic site. The 9 000 G4
peak subset unique to formative state is of particular interest, as these peaks were gained and
then lost throughout the differentiation. Investigating changes of G4 peak intensity changes by
differential binding analysis revealed that 30 000 G4 peaks were differentially binding (DB) up
after 24 h (against 2i/LIF) , followed by a decrease at 10 000 sites going to 48 h time point
primed state (48 h vs 24 h) out of 46 000 total sites analysed (Figure: 3.41, c, d).
3.8.3 Overlap and 3D analysis
BG4 CUT&Tag data was then overlapped onto a single-cell Hi-C structures of na¨ıve, formative
and primed mESC states. It was found that the CUT&Tag sites primarily overlapped with the
A compartment which closely represents euchromatin, more open and transcriptionally active
regions (Figure: 3.42). G4 sites were largely excluded from the B compartment and the nucleoli
that it surrounds. This was most clearly visualised in the na¨ıve nucleus structure where the
nucleolus forms a large cavity in the structure of the nucleus. Formative state structure did
not contain a cavity, had higher chromosome intermingling and mixing of A/B compartments.
It was visualised that G4 sites remained overlapped primarily with the A compartment after
the structural transformation of the nucleus to the formative and primed states15.
In a zoomed view of the structure of chromosome 10 in na¨ıve mESC state, a clear partitioning
into A and B compartments can be recognised (Figure: 3.43, a). Upon overlap with G4-
CUT&Tag, genome regions that overlap with A and B compartments can be visualised as
indicated by denoted region colour mixing (Figure: 3.43, b - d).
15Additional single-cell structures with G4-CUT&Tag overlap can be found in the appendix.
138
Chapter 3: 3D STORM super-resolution imaging of G4s
637
5406
82
9195
315
20415
10175
BC8.2i24hCT.bio2BC8.2i48hCT.bio2BC8.2iCT.bio2
24 h vs 2i/LIF
48 h vs 24 h
48 h vs 2i/LIF
b
d
24 h 48 h
2i/LIF
a
c
Figure 3.41: Summary of G4-CUT&Tag sequencing results. (a) Graph showing consensus peak
numbers at corresponding mESC state in the differentiation pathway. (b) Venn diagram demonstrating
the overlap between the three mESC states. (c, d) Differential binding analysis between the three
mESC states. CPM - counts per million. 3 biological replicates with each having 2 technical replicates.
Bioinformatics analysis performed by Dr Angela Simeone, CUT&Tag data obtained by Isabel Esain
Garcia and Dr Angie Kirchner.
139
Chapter 3: 3D STORM super-resolution imaging of G4s
Naı¨ve
Formative
Primed
Figure 3.42: G4s primarily reside in the A compartment in na¨ıve, formative and primed mESC
states. Cross-sections of single-cell Hi-C simulated structures overlaid with A/B compartment data and
with (LHS) or without (RHS) population derived G4-CUT&Tag sequencing data. 100 kb resolution
structures. Blue - A compartment, orange - B compartment, green - G4 sites.
140
Chapter 3: 3D STORM super-resolution imaging of G4s
a b
c d
Figure 3.43: G4s primarily reside in the A compartment. Na¨ıve state single-cell Hi-C simulated struc-
ture of chromosome 10 overlaid with A/B compartment data and population derived G4-CUT&Tag
sequencing data. (a) A/B compartments. (b) Regions containing CUT&Tag G4 sites. (c) Overlap of
G4 regions with the A compartment. (d) Overlap of G4 regions with B compartment. Colour scheme:
blue - A compartment, orange - B compartment, green - G4 sites, cyan - G4/A compartment overlap,
yellow, G4/B compartment overlap.
Visualising individual G4 peaks from CUT&Tag on chromosome 10 seems to provisionally
indicate some spatial regions which appear to have high spatial density of G4s (Figure: 3.44,
a). These regions could correspond to the spatial clustering observed by 3D STORM of G4s,
however more robust analysis is required (Figure: 3.38). In particular, this analysis needs to
141
Chapter 3: 3D STORM super-resolution imaging of G4s
take into account any clustering on 1D genome sequence for confirming any 3D clustering. BG4
CUT&Tag data was also visualised together with population RNA-seq data on a Hi-C structure
(Figure: 3.44). Some regions richer in G4s appeared to have higher densities of transcribed
genes too, which potentially represents transcriptional factories. More detailed analysis could
reveal how well G4 spatial density can predict genomic locus transcriptional output during
differentiation, when combined with the structural transformations. This way the association
could be established more firmly16.
a b
Figure 3.44: BG4 CUT&Tag overlap visualised with RNA-seq data on chromosome 10 Hi-C structure
of the na¨ıve mESC state. (a) Visualisation of G4 sites on A compartment. White arrows indicate
spatial regions dense with observed G4 sites by CUT&Tag (b) Overlap of G4 sites and population
RNA-seq data. Size of red circles represent relative transcriptional output as measured by RNA-seq.
The centre of the circle represents transcribed gene locus. Colour scheme: blue - A compartment,
green - G4 sites, red - RNA-seq.
Section summary
In this section, G4-CUT&Tag data was presented for mESC differentiation states. It was found
that G4 numbers increase about 4-fold upon transition from na¨ıve to formative state, followed
by a smaller decrease to the primed state. Subgroups of G4s were identified which remain
stable throughout the transition, which are unique to each state or are shared by two of three
states. G4-CUT&Tag overlap with single-cell Hi-C structures visualised G4 sequencing data in
3D chromatin and demonstrated strong overlap with the A compartment.
16This is a still ongoing project and due to time restrictions at the time of writing this thesis additional
analysis of Hi-C data and G4-CUT&Tag overlap could not be performed.
142
Chapter 3: 3D STORM super-resolution imaging of G4s
3.9 Conclusions
In this chapter the goal was to visualise G4s in 3D chromatin. The cornerstone idea of the
project was to overlap 3D super-resolved G4 images on top of single-cell Hi-C simulated struc-
tures, this way co-relating G4 spatial positioning from fluorescence images to sequence frag-
ments in Hi-C structures.
First, a suitable probe for imaging G4s in 3D was investigated. G4 binding antibodies were
chosen as most suitable for use in fixed cells17. Antibodies have higher binding affinities to G4s
in comparison to small molecules. Direct homogeneous labelling of the antibody was preferable
for a G4 imaging probe, hence a number of antibodies and labelling methods were attempted.
Unfortunately, majority of the approaches did not work, due to antibody deactivation upon
labelling, lack of reactivity or inability to express fusions to HaloTag protein. The E12 nanobody
was unique in that regard as its lysine modification with single AF647 dye appeared to have
G4 activity in ELISAs and when imaged in mESC nuclei.
3D STORM imaging with E12-AF647 followed next and good quality images were initially
obtained for a cross-section and the whole of nucleus. Spatial density calculations of probe
localisations allowed visualisation of some regions with E12-AF647 clustering, however good
quality 3D images were obtained of only 7 nuclei, insufficient for making strong conclusions. In
general, E12-AF647 appeared to stain majority of mESC nucleus volume.
Attempts to obtain additional 3D data with E12-AF647 were unsuccessful due to problems with
nucleus autofluorescence, probe stability and probe deactivation in newly synthesised batches.
Later experiments with BG4 revealed that antibody treatment at 37 ◦C, could raise nucleus
autofluorescence, so temperature could be an issue for E12-AF647 labelling too. Note that for
indirect IF with BG4 a secondary antibody labelling was used, raising the total treatment time
at 37 ◦C to 2 h as compared to 1 h for E12-AF647 treatment. Background increase in BG4
experiments therefore became more noticeable.
New attempts to remake E12-AF647 yielded a purified singly labelled probe, however it became
inactive in preparation as indicated by ELISA experiments. Moreover, the inactivated probe
did not demonstrate G4 binding in cells, suggesting the reliability of G4 controls used for
distinguishing imaging to be G4 selective. Heating G4 inactive E12-AF647 regained G4 binding
activity as indicated by ELISA, however G4 controls did not indicate G4 binding in cells during
imaging experiments. The finding that E12-AF647 had its binding affinity increased in ELISAs
by heating was hypothesised to be a result of nanobody refolding at 80 ◦C into an active form.
More detailed analysis such as CD melting to confirm the refolding was not carried out as it
17Since Hi-C is a fixed cell technique.
143
Chapter 3: 3D STORM super-resolution imaging of G4s
was decided to move away from E12 probe for G4 imaging due to its emerged unreliability.
It was not uncovered what caused a change in E12-AF647 in terms of its activity to G4s.
Previous attempts to label BG4 and IgG-BG4 with NHS esters led to their inactivation too.
This signifies the difficulty of labelling nucleic acid binding antibodies as they likely use pos-
itively charged amine groups for tighter DNA binding (as phosphate backbone is negatively
charged). BG4 contains two lysine amino acids in its CDR, while E12 contains none. Change
of E12-AF647 activity could have been caused by labelling a different reactive amine site, a
different lysine or the N-terminus of the protein. Different labelling could block the G4 binding
site or induce allosteric changes to the overall nanobody structure. Small pH or ionic strength
changes in the reaction buffer could be a potential cause for reactivity change or batch-to-batch
differences from bacterial antibody expressions.
Indirect IF with BG4 and a secondary antibody labelled with AF647 was the next probe used
for visualising G4s in 3D. Indirect IF is not as precise in STORM as direct labelling, however,
due to time limitations it was considered a useful approach, as obtaining a G4 binding antibody
conjugated to a dye proved to be difficult. Moreover, BG4 is a well established antibody, already
utilised in IF studies and G4 sequencing [59–61, 77].
BG4 IF demonstrated specificity to DNA in mESC nuclei validating the imaging approach. 3D
STORM images were obtained visualising general staining of majority of nucleus volume with
a few regions of clustering per nucleus. However, too few 3D images of sufficient quality were
obtained for making strong conclusions.
G4-CUT&Tag sequencing data provided a means to combine G4 information with 3D chro-
matin structure. The approach studied mESC in their initial differentiation stages going from
na¨ıve to formative and primed states. It was found that G4 peak numbers increased 4-fold at
the formative state and 3-fold upon reaching the primed state, as compared to the onset of
differentiation. Hi-C structures showed a transformation in 3D chromatin too, however addi-
tional analysis is required for investigating any associations with detected G4s coupled to the
transition. The overlap of G4 sequencing data on Hi-C structures visualised G4 positioning
in 3D and demonstrated G4s to be primarily located in the A compartment. This result was
to be expected due to previous G4 associations to open chromatin, CpG sites and active gene
promoters [61, 66, 71, 72].
Overall, the primary goal of the project to overlap 3D G4 images to simulated Hi-C structures
was only partially reached due to ongoing optimisation of G4 imaging by STORM. However,
with additional future work the overlap of G4 imaging in 3D, G4 sequencing and single-cell
Hi-C will reveal any connections G4s could have with 3D chromatin structure.
144
Chapter 4
Discussion and outlook
4.1 G4 visualisation
The work described in this thesis has focused on the detection and visualisation of G4s in live
cells and 3D chromatin. Robust detection methods of G4s are important in establishing G4
presence and relevance in nuclear biology. Single-molecule imaging has allowed visualisation of
G4s in live cells without large perturbations to global G4 dynamics. Meanwhile, 3D STORM
has given a snap-shot of G4 distribution to be ubiquitous in the whole volume of a fixed
nucleus. Finally, overlap of G4 structure sequencing data with Hi-C chromatin structures
has demonstrated which parts of the nucleus correspond to structures mapped by sequencing
methods. Combined, these methods provide insights into where G4s are localised and their
density. This information can inform on possible functions of G4s and mechanisms by which
G4s could impact nuclear biology. For example, the observed G4 clustering may be involved in
drawing different parts of the genome into compact space to influence 3D chromatin organisation
and function. Clustering could help achieve phase separation in specific parts of the genome.
Moreover, both live-cell and 3D fixed cell imaging appeared to stain the entirety of the nucleus
with no clear sites of exclusion in both human and mice cells indicating G4s widespread presence.
Observation of general nuclear staining and clustering at specific sites simultaneously by higher
resolution methods can connect findings from previous reports where foci or widespread staining
were observed with fluorescent G4 ligands. It is a possibility that different probes can detect
different subsets of G4s. Contrastingly, G4-CUT&Tag sequencing maps overlapped on single-
cell Hi-C structures showed G4 localisations primarily in the A compartment, generally residing
in nuclear interior and excluded from nucleoli-associated regions. In the na¨ıve state, the A
compartment and G4s were organised as a ring shape with the B compartment in its centre in
all nuclei structures. This did not precisely correspond to general nucleus volume G4 staining
145
Chapter 4: Discussion and outlook
in 3D STORM, where no ring shape of localisations was seen, or any regions of exclusion at and
around nucleoli. This difference could be explained by multiple factors. First, probe accessibility
could be dependent on the method used. G4-CUT&Tag might not be able to cleave and tag
regions of chromatin in the B compartment, steps requiring access of secondary and tertiary
antibodies to bring Tn5 to the locus. These steps are needed for successful sequencing but
are independent of G4 presence. Second, Hi-C structures were of nuclei in G1 cell-cycle phase,
ensured by FACS, while 3D STORM images were acquired without cell-cycle information. The
images obtained might not be in G1 due to a low number of nuclei imaged and consequently do
not exactly correspond to Hi-C structures. Both 3D chromatin structure and G4s are known to
be cell-cycle dependent, thus co-relating nuclei at the same phase is important [59, 281]. All in
all, different G4 visualisation methods can help paint a clearer picture of the presence of G4s
in different cells.
4.2 G4s during differentiation
G4-CUT&Tag has revealed differences in levels of G4s during the onset of pluripotent mESC
differentiation. Formative and primed states had a 3.5 and 2.9-fold increase in consensus G4
peaks than in the na¨ıve state. In another study, the trend was also seen during mESC differ-
entiation to neural progenitor cells where a 1.6-fold increase was observed [75]. However, these
cells are further along the differentiation pathway than the primed state, suggesting that the
G4 levels could continue to decrease after the primed state. This suggests that the increase in
G4s during formative and primed state could be only temporary during the onset of differenti-
ation, related to processes involved in the transition. In contrast, in hESCs differentiation led
to a 1.9-fold decrease of number of G4s for differentiation to cranial neural crest cells and a
4-fold decrease during the transition to neural stem cells [71]. This study in hESCs suggested
that G4s help maintain the pluripotent state. However, it is unclear if this role could be as-
cribed in mESC, given the opposite trend. Though studies investigated different timelines of
differentiation, with hESC being most similar to primed state in mESC. The hESC study also
reported a delay in differentiation upon treatment of cells with G4 ligands, further suggesting
G4 structures can influence the transition [71].
Hi-C structures of formative and primed states have higher A/B compartment mixing than in
na¨ıve nuclei, indicating large scale chromatin rearrangement during differentiation. G4 levels
undergo large changes in their levels too, suggesting 3D structure could be associated to G4s,
however additional analysis is required.
146
Chapter 4: Discussion and outlook
4.3 G4 dynamics and antibody labelling
Single-molecule imaging in live cells and chemical trapping with DMS demonstrated that G4s
can dynamically unfold over a time course of 20 min. The appreciation that G4 structures are
dynamic in live cells is important for understanding G4 related processes. Use of G4 ligands
is a common method in studying G4s, however ligand effects can be interpreted in different
ways. G4 ligands can stabilise G4 structures, inducing G4s in treated cells by shifting the
equilibrium towards their formation. An increase of G4 levels should consequently facilitate
G4 related processes (i.e. increase G4 activity). On the other hand, other G4 ligands can
block G4 binding sites and reduce G4-related influences on the cell. The opposing effect of
both mechanisms can make it unclear what effect a G4 ligand could have on G4 activity. A
balance between G4 induction and blocking could also provide an overall neutral influence on
activity. Knowledge of G4 dynamics can aid understanding of G4-related processes and also
the importance of endogenous detection.
A major challenge in this work was direct antibody labelling. This has been difficult for G4
binding proteins which would either be unreactive or lose their G4 binding activity with the
use of chemical methods. An alternative approach to achieve direct labelling could be to
utilise protein engineering strategies such as installing a cysteine site on the antibody’s surface.
SiteClick labelling of IgG-BG4 was successful with low yields, therefore scaling up of that
method could be another approach.
4.4 Outlook
A single-molecule G4 imaging platform with SiR-PyPDS (19) has allowed detection of rela-
tive G4 levels between different live cells in real-time. This approach could be extended to
investigate mechanisms of G4 formation and associated processes. For example, it was found
that simultaneous inhibition of transcription and replication, leads to a large decrease of G4
levels. Decoupling the inhibition would be the next step to investigate transcription and repli-
cation individually. Moreover, different steps of the transcription process could be inhibited
to investigate G4 association to DNA bubble formation, transcription initiation or elongation.
This has been investigated by G4-ChIP-seq, reporting that G4 formation preceded transcrip-
tion initiation by inhibition with triptolide [68]. This approach however, requires cell fixation
and pull-down with BG4 which may itself induce G4 formation through G4 stabilisation. G4
dynamics in fixed chromatin are unknown. Confirming the finding in live cells with endoge-
nous detection methods would provide useful insights. Real time changes in G4 levels could be
observed by inhibiting different steps of replication. A single-molecule G4 detection platform
could be used to determine changes in G4 levels during differentiation in greater detail and to
compare G4 levels in different cancer cell lines versus their normal cell counterparts.
147
Chapter 4: Discussion and outlook
G4-CUT&Tag data has been overlaid with single-cell Hi-C structures for 3 different mESC
differentiation states, however robust analysis has not been performed. Future work will inves-
tigate G4 associations with chromatin structural features, such as TAD boundaries, DNA loops
and promoter-enhancer contacts. Changes to the chromatin structure and G4 levels during
differentiation should help to reveal what features change simultaneously at sites of G4 forma-
tion or loss. Additional population derived data from our collaborators will allow comparisons
with wider range of datasets. RNA-seq and histone sequencing data for H3K27ac, H3K27me3,
H3K4me3 will help to investigate subsets of G4s at active, bivalent and repressed promoters
for detailed analysis of potential G4 influences on transcription with respect to 3D positioning
in the nucleus. G4 spatial density analysis will help reveal if G4s cluster in space and whether
it can predict transcriptional output of the region and its 3D structural arrangement. These
are planned investigations based on the data currently acquired.
Further work is required for overlapping 3D STORM images of G4s with single-cell Hi-C.
After some additional optimisation of the imaging to obtain higher number of localisations,
the imaging will be ready for implementing into single-cell Hi-C protocol established by the
Laue group. The next major challenge will be the precise overlap of 3D images and Hi-C
structures. When this was initially done for centromeres, a coherent point drift method was
used to maximise point spread function overlap between 3D image and centromere locations in
Hi-C structures. A similar approach could be applied for G4s, however they are of much greater
numbers than centromeres. Trying to maximise overlap with G4-CUT&Tag data could be a
way to do this. Moreover, overlap according to nuclear shape may be implemented, as this can
be implied from 3D imaging and Hi-C structures, but this could be imprecise for spherically
shaped nuclei. Finally, specific genomic locations could be marked by FISH or dCas9 methods
to act as points of reference for relation of 3D images to Hi-C structures [278, 302]. If the overlap
was successful, 3D analysis of G4s relationships to chromatin structure could be conducted in
single-nuclei.
Lastly, G4-related perturbations are required for investigating whether chromatin structure
could be manipulated through G4s. G4 ligands could be used in an attempt to perturb G4
activity in cells, however as discussed previously, the dynamic nature of G4 folding and unfolding
leaves it unclear whether G4 ligands should increase activity though G4 stabilisation or decrease
it by blocking G4 binding sites. Another means of perturbation is DMS treatment for chemically
trapping the G4 unfolded state. However, DMS is a toxic alkylating agent which does also
influence other processes in the cell by methylating everything nucleophilic, including RNA,
DNA and proteins. Transfection of G4 oligos into the cell could also be attempted to introduce
a competitor for G4 binding, which could disrupt genomic G4 function. Obtaining G4 images
after any of these perturbations, combined with single-cell Hi-C structures could help reveal
any G4 related influences on 3D chromatin.
148
Chapter 5
Materials and methods
This chapter contains the descriptions of methods and materials used in experiments performed
for this thesis.
5.1 Biophysical experiments
5.1.1 Oligonucleotides
All oligonucleotides were HPLC purified and used as supplied by manufacturer (Table 5.1).
Stock solutions of 100 µM were prepared in MilliQ purified water. Self-annealing of oligonu-
cleotides was performed as follows: heating at 95 ◦C for 10 min, followed by slowly cooling to
RT in a heat block and then held at 4 ◦C overnight to be used for experiments the next day.
5.1.2 FRET melting experiments
Potassium cacodylate buffer (60 mM, pH 7.4) was used to dilute oligonucleotides used in FRET
melting from 100 µM stock solution in nuclease free water.
Oligonucleotides at 200 nM final concentration were titrated with PhenDC3 derivatives: 2, 21,
16 and 17 from 0.2 to 10 µM concentration over 12 data points together with no ligand control
group in 96-well PCR plates. Every titration was carried out in duplicate. Ligands dissolved
in ITC buffer (10 mM potassium phosphate, 70 mM KCl, 0.1 mM EDTA, pH 7.0) were added
in a 1:1 volume ratio with potassium cacodylate buffer with 50 µL volume per sample.
149
Chapter 5: Materials and methods
Name Supplier Sequence (5’ to 3’)
FRET H-telo Biomers FAM-GGG TTA GGG TTA GGG TTA GGG-TAMRA
FRET Kit1 Biomers FAM-GGG AGG GCG CTG GGA GGA GGG-TAMRA
FRET Myc Biomers FAM-TGA GGG TGG GTA GGG TGG GTA A-TAMRA
FRET hairpin Biomers FAM-TAT AGC TAT A-HEG-T ATA
dsDNA GCT ATA-TAMRA
H-telo comp trap Sigma CCC TAA CCC TAA CCC TAA CCC
Kit1 comp trap Sigma CCC TCC TCC CAG CGC CCT CCC
FRET2 H-telo IDT Cy5-AGG GTT AGG GTT AGG GTT AGG GAG
AGG TAA AAG GAT AAT GGC CAC GGT GCG
GAC GGC-Biotin
FRET2 Myc IDT Cy5-TGG GTG GGT AGG GTG GGA GAG GTA AAA
GGA TAA TGG CCA CGG TGC GGA CGG C-Biotin
FRET2 Kit1 IDT Cy5-AGG GAG GGC GCT GGG AGG AGG GAG
AGG TAA AAG GAT AAT GGC CAC GGT GCG
GAC GGC-Biotin
FRET2 comp IDT GCC GTC CGC ACC GTG GCC ATT ATC
overhang Cy3 CTT T-Cy3-TACCTCT
Kit1 Sigma AGG GAG GGC GCT GGG AGG AGG G
Kit1 mut Sigma AGG GAG TGC GCT GTG AGG AGG G
Myc Sigma TGG GTG GGT AGG GTG GGT AA
Myc mut Sigma TGT GTG TGT AGT GTG TGT AA
H-telo Sigma AGG GTT AGG GTT AGG GTT AGG GT
KRAS Sigma AGG GCG GTG TGG GAA GAG
GGA AGA GGG GGA GG
ssDNA Sigma ACG TCA TGC TAT AGA TCG CT
ssDNA comp Sigma AGC GAT CTA TAG CAT GAC GT
Myc for sm binding Invitrogen Biotin-TGA GGG TGG GTA GGG TGG GTA A-AF488
Myc mut for Invitrogen Biotin-TGA GTG TGT GTA GTG
sm binding TGT GTA A-AF488
Random ssDNA1 Sigma Biotin-NHN NHN NHN NHN NHN NHN NHN N
Table 5.1: Oligo sequences used for experiments in this thesis. By approximate order of appearance.
Measurements were made in duplicate with an excitation wavelength of 483 nm and detection
wavelength of 533 nm. The oligos were heated up to 95 ◦C from 25 ◦C by ramping up the
temperature by 0.5 ◦C / min. The instrument used was BioRad CFX96TM Real-Time System
with a C1000 TouchTM Thermal Cycler. Final analysis of the data was carried out using Prism
150
Chapter 5: Materials and methods
5 data analysis and graphing software (Prism R©).
Data analysis of the obtained melting curves was as follows. The first temperature derivative of
the melting curves was taken and used to construct plots of dF/dT vs T. Melting temperature
was taken as a linear average of the maxima observed in dF/dT vs T plot, scaled by population
percentage of the estimated relative areas of G4-bound and unbound states. The ∆Tm value at
an individual ligand concentration was taken by subtracting the G4 melting temperature with
no ligand present. ∆Tm vs ligand concentration plots were used to visualise the data.
5.1.3 Fluorophore properties characterisation
Unless otherwise stated, all the fluorescence and absorbance measurement experiments were
performed at 20 ◦C with a standard quartz cuvette of 10 mm path length and 100 µL or 1 mL
chamber volume.
Spectroscopic properties of SiR-PyPDS
A 20 µM SiR-PyPDS (19) solution was prepared in K+ buffer (50 mM KH2PO4, 100 mM KCl,
58 mM LiOH, pH 7.4) and absorbance and fluorescence spectra were measured. Serial dilution
and measurement cycles were repeated down to 0.63 µM concentration of SiR-PyPDS (19).
Spectroscopic properties of SiR-PyPDS bound to Myc G4
10 µM Myc oligo was annealed into a folded G4 as above in K+ buffer. 10 µM SiR-PyPDS was
prepared in K+ buffer with 10 µM Myc oligo and absorbance and fluorescence of the solution
was measured. Serial dilution of SiR-PyPDS (19) and measurement cycles were repeated down
to 1.3 µM SiR-PyPDS (19), keeping the Myc oligo concentration constant at 10 µM.
Measurement parameters
Fluorescence excitation and emission spectra were recorded on Duetta Fluorescence and Ab-
sorbance Spectrometer (Horiba scientific) at RT. Absorption was measured in 300-800 nm
interval with 1 nm step size and 1 nm bandwidth. Fluorescence spectra were recorded by ex-
citing at 625 nm and measuring 630-800 nm range with 5 nm excitation slit, 5 nm emission
slit.
151
Chapter 5: Materials and methods
Quantum yield determination analysis
The fluorescence quantum yield (Φ) of SiR-PyPDS (17) and bound to Myc G4 were determined
via the following model:
Φ = ΦR(
∫
I∫
IR
1−10AR
1−10A
n2
n2R
)
Where ΦR is the fluorescence quantum yield of Cresyl Violet, used as a standard, I and IR are
the fluorescence intensities of SiR and Cresyl Violet respectively. A and AR are absorbances
of SiR and Cresyl Violet respectively. n is the refractive index of the K+ buffer used for SiR
(1.33) and nR is the refractive index of ethanol (1.36).
5.1.4 pH titration
For absorption pH titration, a range of 2.0-5.0 pH solutions of citric acid (100 mM) and a range
of 6.0-8.0 pH solutions in PBS were prepared by addition of either aqueous hydrochloric acid
or lithium hydroxide. SiR-PyPDS (19) or PhenDC3-SiR (17) were diluted to 1 µM in the
following buffers and measured by Cary 100 UV/Vis spectrophotometer. Measurements were
performed at 20 ◦C with a standard quartz cuvette of 10 mm path length and 100 µL chamber
volume.
For fluorescence pH titration, a range of 2.0-10.9 pH solutions were prepared from PBS (pH 7.4)
by addition of either aqueous hydrochloric acid or lithium hydroxide. SiR-PyPDS was diluted
to 1 µM in the following buffers and the samples were excited at 625 nm wavelength via 2.5
nm excitation slit and emission was measured at 630-800 nm via 5 nm emission slit. Detector
voltage was set to 920 V.
5.1.5 Fluorescence light-up titrations
25 µM solutions of DNA oligos were prepared in assay buffer (100 mM KCl, 50 mM KH2PO4,
pH 7.4, adjusted by LiOH). The oligo solutions were annealed as in section 5.1.1. A two-fold
serial dilution of oligos were prepared between 25 µM and 24.4 nM across 11 wells on 96 well
plate (last well without oligo was a negative control), 50 µL per well. SiR-PyPDS (19) or
PhenDC3-SiR (17) solutions at 200 nM were prepared and 50 µL were added to each well
(accounting for 2 fold dilution). The 96-well plates were agitated on a shaker at 450 rpm
152
Chapter 5: Materials and methods
(rounds per minute) for 2 hours. Fluorescence readings were read on BMG Pherastar Plus
plate reader.
Fluorescence intensity (FI) titration curves were fitted according to a single-site saturation
binding curve equation, taking into account non-specific (NS) binding using Prism software:
y = Bmax·xKd+x + NS · x + Background
5.1.6 FRET cascade experiments
Stock solutions at 1 µM of unannealed FRET oligos (Kit1, H-telo, Myc, hairpin dsDNA) were
prepared in MilliQ water. Then to 100 µL volume of oligo solution sequential additions of
SiR-PyPDS (19) or PhenDC3-SiR (17) were added in 0.039 µM - 10 µM range (from 100 µM
and 1 mM stock solutions in MilliQ water)2. Measurements of fluorescence intensity in 510 -
700 nm range were made after every addition of a G4 ligand. Measurements were performed
on Cary Eclipse fluorimeter with excitation at 493 nm with 10 nm excitation slit and 2.5 nm
emission slit. Detector voltages used depended on the FRET oligo used: Kit1 - 1000 V, H-telo
- 900 V, Myc - 850 V, hairpin dsDNA - 750 V.
5.1.7 G4 induction measurements
Stock solutions of FRET2 H-telo, FRET2 Myc and FRET2 Kit1 at 1 µM were mixed with
FRET2 comp overhang Cy3 oligo in 10 mM Tris buffer (pH 7.4) and annealed at 95 ◦C for
10 min, then allowed to cool to RT and stored at 4 ◦C overnight. Samples were excited at
a wavelength of 540 nm via a 5-nm excitation slit and emission was measured at 550-750 nm
via a 5-nm emission slit. The detector voltage was set to 600 V. Measurements were made on
Cary Eclipse fluorimeter. H-telo, Myc and Kit1 FRET systems at 1 µM concentration were
studied by titrating in PDS (10 mM in DMSO) followed by fluorescence measurements. Data
were analysed using Prism software.
5.1.8 G4 unfolding kinetics
Stock solutions of 2 µM FRET H-telo and Kit1 oligonucleotides and 20 µM H-telo and Kit1 com-
plementary oligonucleotides were prepared separately in K+ buffer (50 mM KH2PO4, 100 mM
2The additions contributed up to 12.8 % total dilution. Oligo concentrations were largely unchanged.
153
Chapter 5: Materials and methods
KCl, 58 mM LiOH, pH 7.4) and annealed as in section 5.1.1. The samples were then mixed
and excited at a wavelength of 493 nm via a 5-nm excitation slit and emission was measured at
518 nm via a 5-nm emission slit. The detector voltage was 570 V for the H-telo FRET oligo and
650 V for the Kit1 oligo. Data points were taken every 2 min for 20 h with 5 s of read averaging.
Kinetic runs were initiated by mixing 50 µL of 2 µM FRET oligo with 50 µL of 20 µM comple-
mentary trap oligo at t = 0 min and oligo unfolding progression was followed by the increase of
FAM fluorescence signal at 518 nm. Kinetics data were recorded using Cary Kinetics software
and analysed using Prism. Kinetics curves were fitted with a two-phase association model.
5.1.9 Fluorescence quench controls
1 µM solutions of SiR-PyPDS (19) or PhenDC3-SiR (17) in PBS pH 7.4 (with or without
supplementation of 0.1 % SDS (v/v)) were titrated with sequential additions of PDS (1) or
PhenDC3 (2) in 1 - 1000 µM range3. Measurements were performed on Cary Eclipse fluorimeter
with excitation at 620 nm with 5 nm excitation slit and 5 nm emission slit, detector voltage -
1000 V, fluorescence detection range - 635 - 750 nm.
5.1.10 Heat refolding
Samples of E12 or E12-AF647 were heated at 70, 80 or 90 ◦C temperature for 10 - 180 min, then
cooled 1 ◦C / min to RT in PeqStar thermal cycler. 10 - 200 µL sample volumes at 1 - 90 µM
concentration were being tested. Optimised conditions were found to be heating at 80 ◦C for
30 min, at 108 µL sample volume and 1 µM concentration.
5.1.11 ELISA
G4s were annealed in ELISA buffer at 1 µM (1.2 mL / plate) as in section 5.1.1. Outline of
protocol is as follows:
1. Wash Streptavidin-coated 96 well plates (PierceTM , ThermoFisher, cat. no. 15500) 4 ×
with PBS pH 7.4 (150 µL) each well.
2. Hydrate plate for 30 min with PBS (200 µL each well).
(a) Meanwhile prepare 25 mL 50 nM bio-oligo in ELISA buffer.
3. Incubate with 200 µL of 50 nM bio-oligo for 1 h, RT (for each well).
3The additions contributed up to 11.8 % total dilution with up to 9 % increase in DMSO fraction (v/v).
154
Chapter 5: Materials and methods
(a) Meanwhile prepare 50 mL block buffer (1.5 g BSA in 50 mL ELISA buffer (3 %));
vortex.
4. Wash 3 × with 150 µL ELISA buffer 1 min shaking at 450 rpm.
5. Block for 1 h in blocking buffer: 100 µL / well.
6. Discard the solutions and add fresh 50 µL block buffer to every well apart from the first.
7. Add to the first column 75 µL of protein (400 nM diluted in block buffer) and serial dilute
to 10 wells (mix at each step) 2-fold dilution series. Leave 25 µL in first column (to save
material).
8. Take out 50 µL from 11th column and discard so that same total volume in all columns
remains 50 µL.
9. Plates incubated for 1 h with serial dilutions of protein (from 400 to 0 nM) in blocking
buffer - shaking at 450 rpm. Ensure there are no bubbles.
(a) Meanwhile prepare ELISA buffer + 0.1 % TWEEN-20 (200 µL in 200 mL ELISA
buffer).
10. Wash with 3 × ELISA buffer / 0.1 % Tween 20, 150 µL each well and shake 450 rpm
1 min each.
11. Incubated 1 h HRP-conjugated antibody (anti-FLAG, ab1238, Abcam) dilute 1:15000 in
block buffer (100 µL per well). Same with anti-goat IgG 1 mg/mL (ab97110, abcam)
against IgG-BG4. Anti-His HRP use 0.5 µg / µL (1:1000 dilution in block buffer).
12. Plates washed 4 × with 150 µL ELISA wash buffer; 1 min shake.
13. Dry wells to remove bubbles - turn plate upside down on tissue.
14. Add 100 µL TMB (HRP substrate) RT add slowly and leave for approx. 2 min. Should
turn shades of blue.
15. Reaction stopped with 50 µL 2 M HCl (or 1 M HCl, or 2 M H2S04: any acid) - should
turn yellow with positive wells.
16. Signal intensity was measured at 450 nm on a BMG Pherastar Plus plate reader.
17. Dissociation constants were calculated from binding curves using Graphpad Prism with
background subtraction.
Solutions and buffers
ELISA Buffer: 100 mM KCl, 50 mM H2KPO4, (1 L, 7.455 g KCl, 6.80g KH2PO4) (adjust
pH to 7.4 using LiOH).
Block buffer: ELISA buffer + 3 % BSA (MP biomedicals, cat.no. 810033) 1.5 g/50 mL.
Wash buffer: ELISA buffer + 0.1 % Tween20.
155
Chapter 5: Materials and methods
5.2 Cell tissue culture
5.2.1 U2OS cells
U2OS cells were grown in standard media DMEM (Dulbeccos Modified Eagle Medium, Gibco)
supplemented with 1 % L-glutamine, 10 % FBS (fetal bovine serum, Gibco, cat. no. 26140079),
at 37 ◦C with 5 % CO2. Cells were usually split in 1:5 ratio every 2-4 days using PBS (pH 7.4)
for washing and trypsin (0.25 % with EDTA) for cell detachment from the surface.
5.2.2 mES cells 2i/LIF conditions
Growth and passage of ES cells were cultured on 0.1 % gelatin (Embryomax, Sigma-Aldrich,
cat. no. ES-006-B). The amounts of reagents provided are for T75 flasks which were used for
regular cell passaging. 1-2 T225 flasks were used for collecting cells for IF protocol, growth and
passaging amounts were scaled 3 times to account for a larger flask and cell number.
1. Gelatin coat a new flask by adding 10 mL of a 0.1 % gelatin/PBS mix. Incubate for at
least 10 minutes in 37 ◦C/5 % CO2 incubator.
2. Wash cells to be passaged with 10 mL of pre-warmed PBS.
3. Add 1 mL of accutase (StemPro, Gibco, cat. no. A1110501) and incubate for 2-3 min
until cells detach. Check that cells are detaching- if needed incubate longer.
4. Add 10 mL of pre-warmed ES serum medium to flask. Pipette the cells up and down 10
times to break ES colony up. Exposing ES cells briefly to serum helps them to cope with
2i conditions.
5. In a 15 mL centrifuge tube spin cells down 1000 rpm for 3 min. Remove supernatant and
re-suspend cell pellet in 1 mL of pre-warmed 2i/LIF media.
6. Remove gelatin/PBS solution from the new flask.
7. Then add cells to a density of ∼10 000 cells / cm2 with 12 mL of pre-warmed 2i/LIF
medium.
8. Grow cells at 37 ◦C / 5 % CO2 and passage every 2 days.
Defrosting cells
1. Gelatin coat a new flask by adding 10 mL of a 0.1 % gelatin/PBS mix. Incubate for at
least 10 minutes in incubator.
2. Rapidly defrost cells from liquid N2 quickly in a bead bath at 37
◦C.
3. Dilute the cells in warm ES serum media (15 mL).
156
Chapter 5: Materials and methods
4. Spin down 1000 rpm for 3 min.
5. Resuspend in 25 mL of 2i/LIF media and grow in T75 flask.
6. Change 2i/LIF media next day (12 mL).
7. Passage in 2 days after defrosting.
Media
CHIR99021 (10 mM DMSO stock). Aliquot 70 µL / tube store -20 ◦C (stable for 1 year).
PD0325901 (10 mM DMSO stock). Aliquot 25 µL / tube store -20 ◦C (stable for 1 year).
Defrosted aliquots kept in fridge and used up in 1-2 weeks.
N2B27 media (stable in fridge for 1 month). NDiff 227 media from Takara Biotech (cat. no.
Y40002), stored as 500 mL bottles at 20 ◦C. When needed thawed one bottle, supplemented
with 1x of Pen/Strep (Gibco, cat. no. 15070063).
2i/LIF media (stable for 1 week in fridge). N2B27 medium (50 mL) supplemented with 15 µL
CHIR99021 (10 mM stock), 5 µL PD0325901 (10 mM stock) and 5 µL Lif (100 µg/mL).
ES serum media (stable for 1 month in fridge). 500 mL GMEM (Gibco, cat. no. 11710035), 50
mL FBS (Embryomax ES cell qualified, Sigma, cat. no. ES-009-B), 550 µL β-mercaptoethanol
(100 mM), 5 mL glutamine 200 mM, 5 mL pyruvate 100 mM, 5.5 mL non-essential amino acid
solution 100x stock (Gibco, cat. no. 11140050), 5.5 mL Pen/Strep 100x stock.
Freezing media. 90 % FBS, 10 % DMSO.
5.3 Cell labelling protocols
5.3.1 Live U2OS cells
Cell plating
In preparation for imaging 100 000 - 200 000 U2OS cells diluted in 2 mL of DMEM supplemented
with 10 % FBS were plated in a 35 mm dish with a 14 mm glass coverslip at the bottom
(MatTek), then allowed to adhere and grow overnight. After ∼18 h, the medium was replaced
with 1 mL of fresh DMEM supplemented medium containing labelling solutions.
157
Chapter 5: Materials and methods
SiR-PyPDS and SiR-iPyPDS
The cells were treated with 1 mL of fresh DMEM medium containing SiR-PyPDS (19) or SiR-
iPyPDS (20) at a final concentration of 20 nM, then cells where further incubated for 30 min.
The DMEM medium containing probe molecules was then discarded and cells were washed
twice with PBS pre-warmed at 37 ◦C. Finally, the medium was replaced with PBS containing 2
µM Hoechst 33342 for nuclear staining, pre-warmed at 37 ◦C, which was immediately followed
by imaging at RT.
PhenDC3-SIR
The cells were treated with 1 mL of fresh DMEM medium containing PhenDC3-SiR in 1 nM -
5 µM range and incubated for 0.5 - 2 h. Then the cells were washed with growth media (1-2
× 2 mL) and incubated for a further 15 minutes. The cells had their growth medium removed
and they were washed with PBS pre-warmed at 37 ◦C (3 × 1 mL) until no pink colour from
the media was visible. The cells in PBS were then imaged immediately at RT.
Lysosome tracking
For lysosome tracking in reference samples without PDS treatment, DMEM media was removed
from the glass coverslips plated with U2OS cells, washed twice with PBS pre-warmed at 37 ◦C
and PBS solution was then added with 50 nM of Lysotracker green in PBS and cell imaging
followed.
Cells with additional G4 ligand treatments, were incubated with 40 nM of SiR-PyPDS (19) or
5 µM of PDS (1) in DMEM for 30 min or 4 h, then washed twice with PBS pre-warmed at 37
◦C and PBS solution was then added with 50 nM of Lysotracker green in PBS and cell imaging
followed.
5.3.2 Fixation of U2OS cells
1. Remove DMEM media from the coverslip.
2. Wash media off with 1 mL of pre-warmed at 37 ◦C PBS.
3. Fixation with formaldehyde (Pierce, Thermo Fisher, cat. no. 28908) or methanol:
(a) Fix the cells with 1 mL of 4 % formaldehyde in PBS at RT for 15 min.
(b) Fix and permeabilise the cells with MeOH 100 % at -20 ◦C for 10 min.
158
Chapter 5: Materials and methods
4. Perform 1 quick 1 mL PBS wash. Remove PBS.
5. Add 1 mL of fresh PBS and incubate at RT for 15 min. Remove PBS.
6. An additional 1 mL PBS wash for 5 min.
(a) (Optional for MeOH fixed and permeabilised cells) DNA degradase plus treatment
(0.5 mL, 60U / mL) at 37 ◦C for 3 h to remove genomic DNA.
7. Cells are kept in 2 mL PBS at 4 ◦C if imaged another day.
(a) (Optional) Cells pre-blocked with 10 µM of PDS (1) or PhenDC3 (2) at 37 ◦C for
1 h.
(b) (Optional) Pre-treat the cells with DMS 600 mM in 1 mL PBS for 30 min at RT.
i. Quench by adding 1 mL 20 % β-mercaptoethanol.
ii. Discard the liquid to bleach solution (neutralise thiol smell).
iii. Then wash with 1 mL PBS 4 times.
8. Remove PBS, add fresh 1 mL PBS with 1 nM SiR-PyPDS (19) or SiR-iPyPDS (20).
9. Cells taken to imaging.
Imaging at different pH
Labelled fixed cells had PBS (pH 7.4) with 1 nM SiR-PyPDS removed and replaced with PBS
at pH 3, pH5 or pH10 together with 1 nM of SiR-PyPDS. Cells then were imaged immediately.
5.3.3 Fixation of mES cells
1. Prepare a fresh single-cell suspension of 10-20 million mES cells in 12.7 mL of 2i/LIF
medium at RT in 15 mL centrifuge tube. Most experiments were performed on P13-P17
(passage) cells.
(a) (Optional) For DMS treatment, passage cells as per normal then treat with DMS 20
mM (0.25 %) for 20 min in the incubator.
(b) Fix immediately adding 1.43 mL 16 % formaldehyde (10 min), then quench with 777
µL 2 M glycine. Then go to step 5.
2. Add 1.8 mL of 16 % (w/v) formaldehyde (Pierce, Thermo Fisher, cat. no. 28908) to
obtain a final concentration of 2 % (w/v) in 14.5 mL.
(a) Start cooling PBS to ice cold.
3. Fix for exactly 10 min at RT. Invert tube 2-3 times during incubation.
4. Quench fixation by adding 1 mL of 2 M glycine (final concentration 0.133 M) and mix
gently by inverting the tube two-three times. The red media solution should turn yellow.
159
Chapter 5: Materials and methods
5. Centrifuge the tube in swing out rotor at 500 g for 5 min at RT.
6. Remove most of supernatant (leaving ∼0.25 mL) and resuspend the pellet (by liquid
stream from the pipette) in 15 mL of PBS.
7. Centrifuge the tube at 500 g for 5 min at RT.
8. Remove most of the supernatant. Gently tap the tube to loosen the pellet.
9. Resuspend the pellet by gently pipetting up and down in 12 mL 0.1 % (v/v) Triton-X 100
in PBS for 10 min at RT to permeabilise the cells.
10. Centrifuge the tube at 500 g for 5 min at 4 ◦C.
11. Wash with 2 × 13 mL ice-cold PBS. Centrifuge the tube at 500 g for 5 min at 4 ◦C.
12. Resuspend the pellet in ice-cold PBS (volume depending on number of samples) and split
the cells into fractions for how many differently pre-treated samples are required.
(a) For pre-treatments: Centrifuge the tube at 500 g for 5 min at 4 ◦C. Remove super-
natant.
(b) Benzonase treatment 1500 units in 3 mL (Buffer: 20 mM Tris·HCl pH 8, 1 mM
MgCl2 in MilliQ water.). Treat for 3 h at 37
◦C.
(c) DNAse I treatment (330 U in 3 mL) with 1x of the 10x reaction buffer. Treat 3 h
at 37 ◦C.
(d) RNAse A (0.25 mg / mL) in 3 mL of PBS. Treat for 3 h at 37 ◦C.
(e) DMS treatment on fixed cells. Pre-treat with 600 mM DMS (add 1 mL of 1.2 M
DMS in MilliQ water to 1 mL cell suspension). 30 min at 37 ◦C or 85 ◦C.
i. Quench by adding 2 mL 20 % β-mercaptoethanol.
ii. Centrifuge 500 g 5 min.
iii. Discard the liquid to bleach solution (to neutralise thiol smell).
iv. Then wash with block buffer (15 mL).
13. Centrifuge the tube at 500 g for 5 min at 4 ◦C. Remove supernatant.
14. Resuspend the pellet in blocking buffer (5 mL) and leave at 4 ◦C overnight. Blocking
buffer: 1 % BSA (w/v, MP biomedicals, cat.no. 810033), 2.25 % Gly (w/v), 0.1 %
Tween 20 (v/v) in PBS for E12-AF647 labelling; 1 % BSA (w/v, Fisher, fraction V, cat.
no. BP1605-100), 0.1 % Tween 20 (v/v) in PBS for BG4 labelling.
15. Centrifuge the tube at 500 g for 5 min at 4 ◦C. Remove supernatant leaving 500 µL
per future sample. Resuspend the cells and split into samples in 15 mL centrifuge tubes
500 µL / sample and keep on ice.
16. (Optional) For pre-blocking control experiments. Every sample should be in 500 µL
volume.
(a) Add 95 µL of BG4 antibody (5 µM) to give 800 nM solution (in 500 µL). Treat at
37 ◦C for 1 h.
(b) Add 0.5 µL of PDS (10 mM in DMSO) to give 10 µM solution (in 500 µL). Treat
160
Chapter 5: Materials and methods
at 37 ◦C for 1 h.
(c) Add 0.5 µL of PhenDC3 (10 mM in DMSO) to give 10 µM solution (in 500 µL).
Treat at 37 ◦C for 1 h.
(d) Add 68.1 µL of E12 (41.7 µM) to give 5000 nM solution (in 500 µL). Treat at 37 ◦C
for 1 h.
17. Add antibody labelling solutions taking into account 2x factor dilution. Add 500 µL
to 500 µL cell solution in block buffer (for samples that had increased volume by pre-
treatment additions, add appropriate volume to ensure correct concentration labelling).
(a) For E12-AF647 6 nM labelling, add 500 µL at 12 nM of E12-AF647. Incubate at 37
◦C for 1 h.
(b) For BG4 20 nM labelling, 500 µL at 40 nM. Incubate at 0 ◦C for 1 h.
i. Wash 3x. Dilute the samples up to 15 mL PBS, centrifuge the tube at 500 g for
5 min at 4 ◦C. Remove supernatant (leaving 0.5 mL).
ii. For secondary antibody labelling add 500 µL at 1 µg/mL of anti-FLAG-AF647
secondary antibody (rabbit monoclonal, Abcam, ab245893). Incubate at 0 ◦C
for 1 h.
18. Prepare 160 mL of ice-cold nuclei extraction buffer. 10 mM Tris-HCl (1.6 mL at 1 M),
10 mM NaCl (0.32 mL at 5 M), 0.2 % (v/v) NP-40 (IGEPAL-630, 0.32 mL), 157.8 mL
MilliQ water and cOmplete protease inhibitor cocktail tablets. Filter through 0.22 µm
filter.
19. Wash 4×. Dilute the samples up to 15 mL PBS, centrifuge the tube at 500 g for 5 min
at 4 ◦C. Remove supernatant (leaving 0.5 mL).
20. Fix the cells and the antibody in place with 10 mL 2 % formaldehyde solution in PBS
for 5 min at RT.
21. Quench by adding 686 µL 2 M glycine. Mix gently by inverting the tube.
22. Centrifuge the tube at 500 g for 5 min at 4 ◦C. Remove supernatant.
23. Resuspend the pellet in 10-15 mL cold PBS.
24. Centrifuge the tube at 500 g for 5 min at 4 ◦C. Remove supernatant. Resuspend with a
pipette.
CRITICAL STEP. Gentle mixing is important to prevent excessive bubbles from forming
in solution which can lead to poor nuclei extraction and cell clumping.
25. Resuspend the pellet in 15 mL of ice-cold nuclei extraction buffer (10 mM Tris·HCl,
pH 8.0, 10 mM NaCl, 0.2 % (v/v) NP-40 (IGEPAL-630) and cOmplete protease inhibitor
cocktail.
26. Incubate the tube on ice for 30 min (gently mix by inversion every 10 min).
27. Centrifuge the tube 600 g for 5 min at 4 ◦C.
161
Chapter 5: Materials and methods
28. Remove supernatant. Resuspend in 10-15 mL cold 50 mM Tris·HCl, 10 mM NaCl, 0.22 µm
filtered.
29. Centrifuge the tube 600 g in swing out rotor for 5 min at 4 ◦C.
30. Remove supernatant. Resuspend in 10 mL cold 50 mM Tris·HCl buffer, 10 mM NaCl,
0.22 µm filtered.
31. Use cell scrapper to remove the cells from the walls.
32. Centrifuge the tube 600 g in swing out rotor for 5 min at 4 ◦C. Remove supernatant
leaving 1 mL.
33. Leave overnight at 4 ◦C as a pellet if imaged next day.
34. Centrifuge the tube 600 g in swing out rotor for 5 min at 4 ◦C.
35. Remove supernatant leaving 1 mL. Add 9 mL 50 mM Tris·HCl, pH 8.0, 11.1 % glucose,
10 mM NaCl, 0.22 µm filtered.
36. Centrifuge the tube 600 g in swing out rotor for 3 min at 4 ◦C.
37. Remove supernatant leaving 1 mL.
38. Nuclei are now ready for imaging.
5.4 Single-molecule imaging
5.4.1 smFRET imaging
To directly visualize smFRET between the Alexa Fluor 488 tag on Myc and SiR-PyPDS (19)
upon binding to Myc, we captured two images of SiR-PyPDS (19) emission: (1) under donor
(488 nm) excitation (200 W·cm2) to observe FRET and (2) under acceptor (647 nm) excitation
(150 W·cm2) to observe all PyPDS-SiR molecules bound to the surface. A separate set of
smFRET experiments were also carried out at a 1,000-fold lower concentration of Myc (0.001 %
surface coverage) and higher concentrations of SiR-PyPDS (19) (10 nM) to monitor anti-
correlated donor and acceptor emission.
5.4.2 In vitro TIRFM single-molecule imaging
Biotinylated oligonucleotides (Myc or Myc-mutant, annealed as in section 5.1.1 at 100 nM
concentration in 100 mM KCl and 50 mM KH2PO4, pH 7.4) and diluted to 10 nM in PBS
containing 0.05 % Tween 20 and 10 µL was added to each well for 5 min. The wells were then
washed twice with 10 µL of 1× PBS containing 0.05 % Tween 20, then treated with 10 µL of
1× PBS containing 1 % Tween 20 for 10 min. The wells were then washed once with 250 pM
of G4 ligand solutions (SiR-PyPDS (19) or SiR-iPyPDS (20)) in PBS and the solution was
162
Chapter 5: Materials and methods
finally replaced with 9 µL of G4 ligand at 250 pM in PBS. For in vitro ligand displacement
experiments, 1 µL of 1 mM PhenDC3 (2) was added to the well. For DMS trapping, a pre-
annealed Myc oligonucleotide (100 nM) was treated with DMS 8 % (v/v) for 20 min, quenched
by adding 10 % (v/v) β-mercapto-ethanol and used for surface coating.
The general set-up used for TIRFM has been described previously [333]. For the in vitro
experiments, TIRFM was implemented on a Nikon Eclipse Ti2 inverted microscope with a
Perfect Focus System for maintaining focus during acquisition. We used 488 nm (MLD 488-200,
Cobolt) and 640 nm (LBX-638- 180-CSB-PP, Oxxius) lasers for excitation with clean-up filters.
The emission collected by the 1.49-NA oil immersion ×60 (×90 with internal magnification)
objective lens (Nikon) was filtered with long-pass and band-pass filters (520/36- 67030 and
692/40 - 67038, Edmund Optics) and imaged on an Evolve 512 Delta electron-magnifying
charge-coupled device (Photometrics) with a pixel size of 178 nm, confirmed using a Ronchi
ruling. The excitation power density was measured by determining the excitation power after
the objective and the beam size in the imaging plane, taking approximately fourfold near-
field enhancement into account. For binding event measurements, a field of view was acquired
for each condition with 500 ms exposure time at a power density of 1.4 kW·cm2. For longer
residency time measurements, time lapses of 300 frames were acquired every 2 s with an exposure
time of 100 ms and a power density of 0.4 kW·cm2. For shorter residency time measurements,
time lapses of 300 frames were acquired every 100 ms with an exposure time of 100 ms and a
power density of 0.4 kW·cm2.
5.4.3 Live cell single-molecule imaging
Binding of SiR-PyPDS (19) to nuclear G4s was visualized using HILO microscopy [235]. The
microscope set-up has been described previously [333]. The central plane of the nucleus in U2OS
cells was found with either bright-field microscopy or using Hoechst staining. For binding event
measurements, 400 frames were acquired for each cell with 100 ms exposure time at a power
density of 180 W·cm2. For residency time measurements, time lapses of 70 frames were acquired
every 3 s with an exposure time of 500 ms and a power density of 180 W·cm2.
Light-sheet microscopy for imaging PhenDC3-SiR (17) in live U2OS cells was set up as de-
scribed previously [238].
163
Chapter 5: Materials and methods
5.4.4 Extracted mESC nuclei single-molecule imaging
Extracted mESC nuclei (prepared as in section 5.3.3) were imaged using HILO with 60× 1.27
NA water immersion objective. 2000 frames were acquired for each nucleus with 30 ms exposure
time at a 640 nm laser power density of 180 W·cm2. The nuclei were deposited on 35 mm dish
with a 14 mm glass coverslip at the bottom (MatTek) and imaged at RT.
Nuclei were imaged STORM buffer which was prepared in situ by having nuclei deposited on
an imaging plate in 50 mM Tris·HCl, 10 mM NaCl, 10 % glucose, pH 8.0 buffer supplementing
with final concentrations of 50 mM MEA, 40 µg/mL catalase and 0.5 mg/mL glucose oxidase
just before imaging.
Stock buffers prepared for STORM imaging:
• 1 M of MEA (Merck, Cat. No. 30070) dissolved in 0.36 M HCl, stored frozen at -20 ◦C
(20x imaging solution of STORM).
• 40 mg/mL glucose oxidase from Aspergilus niger (Merck, Cat. No. G2133) dissolved in
buffer (24 mM PIPES, 4 mM MgCl2, 2 mM EGTA at pH 6.8) 0.22 µm filtered, flash
frozen in liquid nitrogen and stored at -20 ◦C (80x imaging solution of STORM)
• 5 mg/mL catalase (Merck, Cat. No. C40) in buffer (24 mM PIPES, 4 mM MgCl2, 2 mM
EGTA at pH 6.8) 0.22 µm filtered, flash frozen in liquid nitrogen and stored at -20 ◦C
(125x imaging solution of STORM)
5.4.5 3D STORM imaging
Double helix point spread function method of 3D STORM imaging was set up and used as de-
scribed previously [243]. Before start of imaging the microscope was calibrated with Tetraspek
beads to adjust phase mask position in the emission path and to adjust the objective correction
collar. This was to ensure symmetrical rotation of the two lobes from double helix imaging
around their centre position.
12 overlapping z -axis slices every 0.8 µm were taken and combined for obtaining whole nucleus
3D images. Every slice would be imaged for 200 frames at 20 ms exposure every iteration to a
total of 5 000 total frames per slice. Before and after every iteration of a slice, 20 frames of the
nucleus would be taken with a second camera in a collimated 535nm LED light for x,y drift-
correction post-processing. z drift was accounted for by perfect focus stage control on Nikon
microscope. The scanning order of slices was 0, 6, 1, 7, 2, 8, 3, 9, 4, 10, 5, 11 with numbers
164
Chapter 5: Materials and methods
allocated in sequence according to their height position in the nucleus. This is to minimise the
bleaching effect in adjacent slices and give time for the dye to recover for next iteration.
3D STORM data processing was carried out as follows. One cropped bead from the calibration
file was used to adjust a peak fit function on FIJI software [361]. A code is then used4 on the .tiff
data files of each imaged slice to find single-molecule localisations with x,y position determined
by centroid of DHPSF lobes and z position by angle between the lobes. Localisations at extreme
double helix angles were removed during post-processing to improve overall accuracy. A .3d
file is obtained for each slice which are then combined to get a 3D STORM image of the whole
nucleus. 3D files were visualised by ViSP software [362].
5.5 Image data analysis
5.5.1 Quantification of binding events
For in vitro binding event measurements, the number of events was determined by counting the
number of peaks in an image using the Find Maxima function in ImageJ with a noise threshold
of 5 500. For binding event measurements in cells, a single image typically only yielded a few
points, so it was necessary to acquire a time-lapse video to obtain a suitable number of binding
events. Before analysis, a rolling ball background subtraction of 5 pixel and a 1-pixel Gaussian
blur were applied to all images. In U2OS cells single-molecule tracking was then performed
using TrackMate in FIJI software [361], with the particle diameter set to 5 pixels, intensity
threshold set to 200, linking distance set to 3 pixels, gap closing distance set to 3 pixels and
gap frames set to 3. The number of binding events was then quantified as the number of tracks
with a track length of three or longer.
Before analysis of mESC nuclei imaging with E12-AF647, a rolling ball background subtraction
of 5 pixel and a 1-pixel Gaussian blur were applied to all images. The first 1000 of 2000
acquired imaging frames were removed to account for pre-bleaching in STORM buffer. Single-
molecule tracking was performed with particle diameter set to 4 pixels, linking distance set to
2.5 pixels, gap closing distance set to 2.5 pixels and gap frames set to 2. The setting of threshold
was variable from condition to condition due to differences in background levels, it would be
adjusted such that confident single-molecule events confirmed by eye would correspond to ones
detected automatically by the set parameters.
4Written by Dr Aleks Ponjavic and modified by Aleksandra Jartseva.
165
Chapter 5: Materials and methods
5.5.2 Residency time determination
Time lapses for both in vitro and cell measurements were analysed using image processing
and TrackMate [361]. The distribution of residence times was fitted to a single-component
exponential decay, to determine the characteristic residence time. This value was corrected for
photobleaching as previously described [340]. The photobleaching rate for in vitro experiments
was determined to be 0.001 s−1, compared to 0.066 s−1 for the off rate [213]. The photobleaching
rate in cells was determined to be 0.01 s−1, compared to 0.16 s−1 for the off rate [213].
5.6 E. coli antibody expression
5.6.1 BG4 scFv antibody expression by autoinduction and purifica-
tion
Transformation
1. Thaw a tube of BL21(DE3) E. coli (NEB, Cat. No. C2527H) on ice.
2. Add 0.5 µL of BG4 pSANG10-3F-scFV plasmid prep5 (350 ng/µL) to BL21 and mix by
gently flicking tube (5x); do not vortex.
3. Place mixture on ice (2 min).
4. Heat shock at 42 ◦C for 10 seconds.
5. Place mixture on ice (2 min).
6. Add 950 µL of SOC (super optimal broth with catabolite repression, NEB, Cat. No.
B9020) at RT to tube and incubate at 37 ◦C, 250 rpm for 1 h.
7. Warm selection plates, one with kanamycin and one with ampicillin (negative control).
8. Spread bacteria on selection plates and incubate inverted overnight at 37 ◦C. (Kanamycin
plate with 100 µL, ampicillin with 250 µL)
9. Count and pick 2 colonies (small, clearly defined), use 2 separate colonies to ensure at
least one grows overnight.
BG4 expression
10. Inoculate each colony into 2x 5 mL 2x YT medium with kanamycin (50 µg/mL) and 2 %
glucose and grow overnight at 30 ◦C, 200 rpm, in loosely closed 50 mL tube.
11. Inoculate 2 mL of the overnight cell culture in 1 L auto-induction medium and grow at
37 ◦C for 6 h shaking at 250 rpm. Transfer 4x 250 mL into 4 2 L flasks to allow proper
5Provided by Jiazhen Shen.
166
Chapter 5: Materials and methods
mixing.
12. Incubate this starter culture overnight at 25 ◦C, shaking at 280 rpm.
BG4 purification
13. Spin down cell culture for 30 minutes at 4 ◦C, at 4 000 g.
14. Equilibrate the His-select Nickel affinity beads (Sigma, Cat. No. P6611) 3 times 1 mL
using filtered PBS (spin 2 min at 200 g).
15. Gently resuspend pellet in 80 mL ice-cold TES, leave for 10 minutes on ice.
16. Add 120 ml ice-cold TES diluted 1:5, leave 15 minutes on ice.
17. Spin down for 20 minutes at 16 000 g at 4 ◦C.
18. Collect supernatant and rotate for 1 h at RT with PBS-washed Nickel affinity beads.
19. Load samples onto Proteus 1-step batch Midi Spin columns (Generon) and wash beads
twice with wash buffer. Use 2 washes to transfer beads and then 2 additional washes.
20. Elute antibody with ∼ 2 mL elute buffer.
21. Dialyse overnight against 4 L PBS (0.22 µm filtered) at 4 ◦C using GeBaflex tubes (Gen-
eron).
22. Dialyse again 4 h against fresh 4 L.
23. Store at 4 ◦C for up to a month or freeze aliquots (10-100 µL size) at -20C (avoid freeze-
thaw cycles).
Solutions and buffers
Auto-induction Medium (ZYM5052). Supplement basic ZY medium with: 2 mM MgSO4,
0.2x Metals Mix (a small precipitate forms on addition), 1x 5052 buffer, 1x M buffer, 50 µg/mL
kanamycin.
1000x metals mix (Stock Solution). Autoclave the stock solutions of the individual metals
(except the FeCl3 in HCl); Mix aseptically and store at RT. Add in the order written (per 10
mL, Table: 5.2):
167
Chapter 5: Materials and methods
Final concentration
/ mM
Chemical Volume added
/ µL
Stock
concentration / M
MilliQ water 3600
50 FeCl3·6H2O (dissolved in
∼ 0.1 M HCl)
5000 0.1
20 CaCl2 200 1
10 MnCl2·4H2O 100 1
10 ZnSO4·7H2O 100 1
2 CoCl2·6H2O 100 0.2
2 CuCl2·2H2O 200 0.1
2 NiCl2·6H2O 100 0.2
2 Na2MoO4·2H2O 200 0.1
2 Na2SeO3·5H2O 200 0.1
2 H3BO3 200 0.1
Table 5.2: Metals mix stock solution composition.
ZY medium (base medium). 1 L MilliQ water, N-Z amine AS (Sigma Aldrich, Cat. No.
N4517) 10 g, Yeast Extract (Sigma, Cat. No. Y1625) 5 g. Dissolve in H2O, autoclave and store
at RT.
50x 5052 (Stock Solution). 25 % (w/v) glycerol, 2.5 % (w/v) glucose, 10 % (w/v) α-lactose.
Dissolve in H2O, autoclave and store at RT.
50x M (Stock Solution). 1.25 M KH2PO4, 2.5 M NH4Cl, 0.25 M Na2SO4. Dissolve in H2O,
autoclave and store at RT.
TES buffer. 50 mM Tris·HCl pH 8.0, 1 mM EDTA pH 8.0, 20 % sucrose. Dissolve in H2O,
filter and store at 4 ◦C.
TES. 1:5 TES diluted with MilliQ water (24 mL TES + 96 mL water), 12 µL benzonase
(Sigma, Cat. No. N5661), 2 mM MgSO4. Dissolve in H2O, filter and store at 4
◦C. Add PIC
(protease inhibitor cocktail, EDTA free, cOmpleteTM Mini tablets, Roche) immediately before
use.
Wash Buffer. PBS pH 8, 100 mM NaCl, 10 mM imidazole. Dissolve in H2O, filter and store
at 4 ◦C.
Elution Buffer. PBS pH8, 250 mM imidazole. Dissolve in H2O, filter and store at 4
◦C.
168
Chapter 5: Materials and methods
5.6.2 Production and purification of E12 nanobody in BL21(DE3)
from the pHEN2 plasmid
Day 1: Transformation BL21(DE3) bacteria with pHEN2 plasmid
1. Start heating a water bath to 42 ◦C. Warm SOC to RT.
2. Thaw a tube of BL21 (50 µL) competent E. coli on ice for 10 min.
3. Add 1-5 µL containing 100 ng of plasmid DNA6 to the cell mixture. Carefully flick the
tube 4-5 times to mix cells and DNA. Do not vortex.
4. Place the mixture on ice for 30 min (high efficiency) or at least 2 min (low efficiency).
Do not mix.
5. Heat shock at 42 ◦C for 10 seconds. Do not mix.
6. Place on ice for 2-5 minutes. Do not mix.
7. Warm ampicillin selection plates to 37 ◦C in a bacterial culture incubator.
8. Add 950 µL of RT SOC into the mixture.
9. Place at 37 ◦C for 60 minutes. Shake vigorously (300 rpm).
10. Mix the cells thoroughly by flicking the tube and inverting, then add 100 µL of the culture
to a selection plate. Spread with a cell spreader and incubate overnight at 37 ◦C.
Day 2
11. Count and pick 2 small, clearly defined colonies, use 2 separate colonies to ensure at least
one grows overnight.
12. Grow overnight in 5 mL (for 100 mL culture) or 12 mL (for 250 mL culture) TB + 1 %
glucose + Ampicillin 100 µg/mL + 1 clone of bacteria at 37 ◦C shaking at 180 rpm.
Day 3
13. Culture of 100 mL or 250 mL TB + 0.1 % glucose + Ampicillin 100 µg/mL, with 2 mL
or 5 mL of the overnight culture. Incubate at 37 ◦C, shaking at 250 rpm, until OD at 600
nm is between 0.6 and 0.8. Generally takes 2-5 h.
14. Induce protein expression by IPTG (Sigma, Cat. No. I6758) at 0.5 mM final, and incubate
overnight at 28 ◦C and 280 rpm. Prepare IPTG fresh as it hydrolyses in water.
Day 4: Purification
Always work on ice.
15. Spin the culture for 15 min at 4 ◦C, at 4000 rpm (about 2300 g).
(a) Prepare sonication buffer.
6Kindly provided by Silvia Galli.
169
Chapter 5: Materials and methods
(b) Bring 1.8 L of MiliQ water to 4 ◦C for dialysis.
16. Discard the supernatant and put the pellet on ice.
17. Quick freeze and thaw in liquid nitrogen (if continuing on another day, store at -80 ◦C).
18. Resuspend the pellet in 15 mL Sonication buffer.
19. Incubate for 30 min in ice (time can vary).
(a) Prepare 130 mL (per 100 mL culture) of wash buffer and 2 mL elution buffer.
20. Sonicate 20 times for 30” sonication + 60” in ice. 40 % amplitude power with microprobe
on Qsonica (Q700) sonicator.
(a) For 100 ml of culture, count 500 µL of HIS-select Nickel Affinity Gel resin (Sigma).
(b) 2 washes: with PBS 1x (spin 2500 rpm for 2 min per wash).
(c) Equilibration: add 20 mL of Wash buffer. Turn the tube several times and rotate at
4 ◦C for 30 minutes minimum.
21. Spin at 4 ◦C for 30 min at 14000 rpm.
22. Take the supernatant. Keep on ice for same day purification, otherwise freeze it in liquid
nitrogen and store at -80 ◦C.
23. Filter the supernatant through stericup (Millipore) 0.22 µm filter.
24. Binding. Remove the Wash buffer from the equilibrated beads. Spin 3000 rpm, 5 min.
25. Mix the supernatant with His-select Nickel Affinity gel resin (Sigma). Rotate or shake at
4 ◦C for 1 hour.
26. Remove the liquid completely by column filtration (3000 rpm, 5-15 min)
27. Wash the resin ∼200 times the bead volume. For 100 mL culture, 500 µL of beads,
requires 100 mL of Wash buffer. Allow all the liquid run through the column before
adding more for next wash.
(a) While waiting between centrifugation steps, hydrate dialysis tube membrane with
MiliQ water for at least 30 min.
28. Elute with 3 × 500 µL of Elution buffer. On second elution allow the liquid to sit with
resin for 5 min.
29. Dialyse the elute overnight against 2 L PBS at 4 ◦C.
Day 5: Dialysis
30. Do second dialysis for 4 h against fresh 2 L PBS at 4 ◦C.
Solutions and buffers
Storage buffer: PBS pH 7.4 + 160 mM NaCl.
Sonication buffer: Prepare fresh 15 mL per 100 mL of culture. Storage buffer + PIC tablet
170
Chapter 5: Materials and methods
+ PMSF 1 mM final (stock at 50 mM in DMSO, 300 µL) + lysozyme (Sigma, Cat. No. L6876)
1 mg/mL final.
Wash buffer: Prepare fresh storage buffer + 15 mM imidazole.
Elution buffer: Prepare fresh storage buffer + 250 mM imidazole.
Terrific Broth (TB): tryptone 12 g/L, Yeast extract 24 g/L, 17 mM KH2PO4, 72 mM
K2HPO4, 0.4 % glycerol.
5.7 General synthetic experimental
All solvents and reagents were purified by standard procedures [363] or used as supplied from
commercial sources, unless otherwise stated. All reactions were monitored by thin layer chro-
matography (TLC), LCMS and 1H NMR spectra taken from the reaction. Microwave irradiation
was performed in a CEM Discover R© SP microwave reactor. All yields presented, unless noted
otherwise, are of dried pure compounds measured by weight.
Proton magnetic resonance spectra were recorded using an internal deuterium lock (at 298 K
unless stated otherwise) on Bruker DPX (400 MHz; 1H-13C DUL probe), Bruker Avance III
HD (400 MHz; Smart probe), Bruker Avance III HD (500 MHz; Smart probe) and Bruker
Avance III HD (500 MHz; DCH Cryoprobe) spectrometers. Proton assignments are supported
by 1H-1H COSY, 1H-13C HSQC or 1H-13C HMBC spectra, or by analogy. Chemical shifts (δH)
are quoted in ppm to the nearest 0.01 ppm and are referenced to the residual non-deuterated
solvent peak. Discernable coupling constants for mutually coupled protons are reported as
measured values in Hertz (Hz), rounded to the nearest 0.1 Hz. Data are reported as: chemical
shift, number of nuclei, multiplicity (br, broad; s, singlet; d, doublet; t, triplet; q, quartet;
m, multiplet; or a combination thereof), coupling constants and assignment. Diastereotopic
protons are assigned as X and X’ , where X’ designates the lower-field proton.
Carbon magnetic resonance spectra were recorded using an internal deuterium lock (at 298 K
unless stated otherwise) on Bruker DPX (101 MHz), Bruker Avance III HD (101 MHz) and
Bruker Avance III HD (126 MHz) spectrometers with broadband proton decoupling. Carbon
spectra assignments are supported by DEPT editing, 1H-13C HSQC or 1H-13C HMBC spectra,
or by analogy. Chemical shifts (δC) are quoted in ppm to the nearest 0.1 ppm and are referenced
to the deuterated solvent peak. Data are reported as: chemical shift, number of nuclei (if not
one), multiplicity (if not a singlet), coupling constants and assignment.
High performance liquid chromatography (HPLC) was carried out on an interchim Puriflash
prep. HPLC-MS, using a Varian Pursuit C18, 5 µm column (250 × 21.2 mm) and a gradient
171
Chapter 5: Materials and methods
elution with water/MeCN containing 0.1 vol.% TFA at a flow rate of 25 mL min−1 and UV
detection (λmax = 254 and 365 nm).
Analytical TLC was performed using pre-coated Merck glass backed silica gel plates (Silicagel
60, Cat. No. F254). Retention factors (Rf ) are quoted to 0.01. Flash column chromatography
was undertaken on Fluka or Material Harvest silica gel (230-400 mesh) under a positive pressure
of nitrogen. Visualization was achieved using UV light (254 nm) and chemical staining with
basic potassium permanganate solution or ninhydrin as appropriate.
Flash chromatographic purifications were carried out using CombiFlash Rf (Teledyne Isco)
with RediSepRf column.
5.7.1 Mass spectrometry
Waters Xevo G2-S or Waters Vion IMS Qtof MS were used for protein mass determination.
Mass spectra of protein chromatography peaks detected at 280 nm were deconvoluted with
Mass Lynx software using MaxEnt1 plugin with parameters set to 0.75 Da resolution, 33 %
minimum intensity ratios, uniform Gaussian width at half-height 1 Da, iteration to convergence.
High-resolution mass spectra (HRMS) of small molecules were measured on a Micromass LCT
Premier spectrometer using electron spray ionization (ESI) techniques. Masses are quoted
within the 5 ppm error limit.
Liquid chromatography - mass spectrometry (LC-MS) were measured on a Bruker amaZon X
Ion Trap MS, with Kinetex C18 column (Phenomenex, 50 × 2.1 mm, 2.6 µm).
5.7.2 A¨KTA FPLC chromatography
An optimised protocol used 3 HiTrap Q 1 mL columns connected in sequence. Buffers: A - 50
mM Tris·HCl pH 8.5, B - 50 mM Tris·HCl pH 8.5, 1.0 M NaCl. Buffers were 0.22 µm filtered
and degassed under 100 mbar pressure for 30 min.
1. The A¨KTA Pure system (+loop) was cleaned with H2O / NaOH / H2O sequence before
use (20 mL, 5 mL/min).
2. Flowrate of H2O was set to 1 mL/min and 20 mL was drawn each from P9A and P9B
valves to remove bubbles from the system.
3. Column alarms were set to 0.5 MPa and detection to 650 nm wavelength by UV-vis
detector.
172
Chapter 5: Materials and methods
4. With 0.5 mL/min H2O flowrate columns were connected drop to drop in down position.
5. Columns were washed with 3-5 column volumes (CV) of A / B / (90 % A, 10 % B) buffer
sequence, run the latter until conductivity reading stabilises.
6. Run was set up with the parameters:
• 0.7 mL/min flowrate.
• Equilibration: 3 mL (1 CV), 10 % B.
• Sample application: 3 mL or 7.5 mL (1.5x loop volume), 10 % B.
• Column wash: 3 mL (1 CV) 10 % B.
• Elution gradient:
– 10 - 30 % B over 15 mL.
– 30 % B hold gradient for 40 mL.
– 30 % - 50 % B over 10 mL.
– 50 % - 100 % B over 10 mL.
• 100 % B column wash - 15 mL.
• 0 % B column wash - 15 mL.
7. Apply E12-AF647 crude product mixture via syringe to the loop and start the run.
Secondary purification of E12-AF647 was carried out on a new set of HiTrap columns majority
containing fractions from the primary purification. As these fractions were in ∼300 mM NaCl
at that stage, before injecting them for a second round, they were diluted with buffer A 1:1
to give 150 mM NaCl concentration for nanobody injection. At this salt concentration, it was
observed that the E12-AF647 binds to the anion exchange column.
5.7.3 Spin column purifications
His-select affinity gel purification for E12-AF647.
1. Load 200 µL of His-select nickel affinity gel into 0.22 µm, 1 mL size spin column.
2. Spin the column, 1 min, 14 100 g.
3. Add the crude reaction mixture from antibody labelling reactions.
4. Shake the loaded beads for 30 min, 300 rpm.
5. Wash 5 times with 150 µL of wash buffer (PBS pH 8, 100 mM NaCl, 10 mM imidazole).
6. Elute the beads with 100 µL, followed by 50 µL of elution buffer (PBS pH 8, 250 mM
imidazole).
173
Chapter 5: Materials and methods
7. Dialyse against 2 L PBS pH 7.4 overnight.
8. Repeat dialysis again against fresh 2 L PBS pH 7.4 for 4 hours.
Pre-packed spin column purification.
1. Remove storage media from spin column. 1500 g, 1 min.
2. Exchange buffer by running 300 µL of PBS pH 7.4 three times through the column.
Centrifuge 1 500 g, 1 min.
3. Add crude reaction mixture to the centre of the resin.
4. Centrifuge 1 500 g, 2 min.
5. Collect purified and desalted products.
5.8 Experimental synthetic details
5.8.1 Antibody labelling
Lysine labelling general procedure
To antibody solutions in PBS, add NaHCO3 (1.0 M, to give 100 mM final concentration, pH
8.3). Then add AF647-NHS dye (10 mM, 0.5 - 40 equivalents w.r.t. antibody) dissolved in
anhydrous DMSO. Incubate at 25 ◦C for 90-180 min shaking at 300 rpm (Scheme: 3.4). The
unreacted dye from the antibodies was then purified by a number of different purification
methods.
Initial purifications were performed with spin columns exchanged into PBS. Later on E12-
AF647 purifications were performed with one or two runs on A¨KTA FPLC followed by desalting
overnight by dialysis at 4 ◦C or by amicon spin columns (3.5 kDa membrane cut-off).
Conditions used for different antibodies and their batches are summarised in a table 5.3:
174
Chapter 5: Materials and methods
Antibody
and batch
Concentration
/ µM
Reaction
volume / µL
AF647-NHS
equivalents
Purification
method
Yield /
µg
BG4 4.5 µM 30-75 µL 4 - 40 Zeba Spin columns
(7 kDa)
1 - 5
goat
IgG-BG4
6.0 µM 12 µL 4 - 40 Zeba Spin columns
(7 kDa)
4 - 5
E12
AR061
57 µM 9-11 µL 4 - 40 His-select affinity
gel spin columns
3 - 8
E12
AR064
57 µM 22-27 µL 4 - 40 Biorad P6 spin
columns
6 - 7
E12
AR068
57 µM 17-19 µL 1 - 2 Biorad P6 spin
columns
2 - 3
E12
AR071-076
57 µM 28 - 50 µL 0.8 Biorad P6 spin
columns
4
E12
AR079
8.6 µM 55-57 µL 0.5 - 40 Biorad P6 spin
columns
0.8 - 2
E12
AR082
57 µM 930 µL 1.5 Anion exchange
A¨KTA and dialysis
11
E12
AR080-100
14 - 57 µM 320 - 3100 µL 1 - 3.5 2x A¨KTA and
amicon filters (3.5
kDa)
0 - 180
Table 5.3: Lysine labelling conditions for different reactions.
Cysteine labelling
BG4 (100 µL in PBS pH 7.4, 5 µM) was sequentially combined with TCEP·HCl (5 µL in PBS,
1 mM, 10 eq.) and AF647-C2-maleimide (5 µL, 10 mM, 50 eq.). The reaction was incubated
at RT for 2 h (Scheme: 3.2). Then crude reaction product mixture was purified by His-select
affinity gel purification. No desired labelled product was observed by UV-vis.
E12 (20 µL in PBS pH 7.4, 63.6 µM) was sequentially combined with TCEP·HCl (5.1 µL in
PBS, 20 mM, 80 eq.) and AF647-C2-maleimide (56.4 µL, 10 mM, 50 eq.). The reaction was
incubated at RT for 2 hours. Then crude reaction product mixture was purified on Biorad P6
columns. No desired labelled product was observed by UV-vis.
175
Chapter 5: Materials and methods
PD rebridging
Goat IgG-BG4 (50 µL, 6.9 µM, Absolute Antibody, Cat. No. Ab00174-24.1) in BBS (borate
buffered saline) buffer (25 mM sodium borate, 25 mM NaCl, 0.5 mM EDTA, pH 8.0) was
combined with PD (4.0 µL, 20 mM in DMSO, 222 eq.) and kept at 4 ◦C for 1 hour. Then
TCEP·HCl (10 µL, 20 mM in H2O, 80 eq.) was added to the reaction mixture and the re-
action was stored at 4 ◦C overnight (Scheme: 3.3). Crude products were purified by Zeba
(ThermoFisher, Cat. No. 89890) pre-packed spin columns (7 kDa cut-off).
The second step of azide-alkyne Huisgen cycloaddition was performed on product solution
(50 µL) upon mixing with AF647-azide (1.8 µL, 10 mM in DMSO, 50 eq., ThermoFisher, Cat.
No. A10277). The solution was then purified by Zeba pre-packed spin columns (7 kDa cut-off).
No conjugation was observed by UV-vis and MS.
SiteClick labelling
Labelling was carried out using SiteClick sDIBO alkyne and azido modification labelling kit
from ThermoFisher (Cat. No. C20029, S20026).
1. Concentration and preparation antibody.
(a) Wash the antibody concentrator.
i. Add 500 µL of H2O into the concentrator.
ii. Centrifuge at 5000 g for 6 min.
iii. Discard flow through.
(b) Add 80 µL of IgG-BG4 (80 µg)
i. Add 420 µL of antibody preparation buffer.
ii. Centrifuge at 5000 g for 6 min.
iii. Discard flow through.
(c) Invert the concentrator into collection tube and collect concentrated, buffer ex-
changed antibody (about 50 µL). Centrifuge at 1000 g for 3 min.
2. Modify carbohydrate domain of IgG.
(a) Add 10 µL of β-galactosidase to the antibody solution.
(b) Seal the tube with parafilm and incubate the reaction overnight at 37 ◦C.
176
Chapter 5: Materials and methods
3. Azide attachment.
(a) Prepare azide modification solution. To the reaction mixture add UDP-GalNAz
previously combined with: 75 µL H2O, 12.5 µL 20x Tris buffer pH 7.0, 25 µL buffer
additive, 80 µL β-1,4-galactosyl transferase.
(b) Vortex.
(c) Briefly centrifuge, wrap with parafilm and incubate overnight at 30 ◦C.
4. Purify and concentrate the azide-modified antibody.
(a) Wash antibody concentrator with 1 mL of 1x Tris buffer pH 7.0. Centrifuge 1200 g
for 10 min.
(b) Add 1.6 mL of 1x Tris buffer and azide modified antibody (∼250 µL).
(c) Centrifuge at 1200 g for 6 min. Discard flow through.
(d) Add 1x Tris buffer up to 2 mL mark, centrifuge 1200 g for 6 min. Repeat 2 times.
5. Click reaction conjugation of AF647 to the azide modified antibody.
(a) Add 11 µL of Click-iT sDIBO-alkyne-AF647 to the azide modified antibody solution
(∼110 µL)
(b) Incubate at 25 ◦C for 24 hours.
6. Purification and concentration of the IgG-BG4-AF647 conjugate.
(a) Wash antibody concentrator with 2 mL of 1x Tris buffer. Centrifuge 1200 g for
10 min. Discard flow through.
(b) Add 1.6 mL of PBS pH 7.4 and the IgG-BG4-AF647 conjugation product mixture.
(c) Centrifuge at 1200 g for 10 min. Discard flow through.
(d) Add 1.8 mL PBS, centrifuge at 1200g for 15-20 min. Discard flow through. Repeat
twice.
(e) Invert concentrator and collect IgG-BG4-AF647 conjugate by centrifuging at 1000 g
for 3 min.
The overall reaction provided 1.7 µg of IgG-BG4-AF647 (30 µL at 0.40 µM).
177
Chapter 5: Materials and methods
5.8.2 PhenDC3-SiR synthesis
2,9-Dimethyl-1,10-phenanthrolin-4(1H)-one (10)
87
6
5 N4
3
2
11
109
HN114
13
12
O
1516
10
2,2-Dimethyl-1,3-dioxane-4,6-dione (4.11 g, 28.4 mmol) and trimethylorthoacetate (36.6 mL,
567 mmol) were mixed and brought to reflux at 110 ◦C for 15 min. The resulting yellow
solution was cooled to RT and then 2-methylquinolin-8amine (3.00 g, 19.0 mmol) was added
turning the solution dark green. The reaction mixture was refluxed at 110 ◦C or 1.5 h and
stirred at RT for 16 h.
The solvent was removed in vacuo to afford a red oil which was then suspended in diphenylether
(60 mL) and the mixture was heated at 230 ◦C for 1 h. After cooling to 50 ◦C petroleum ether
was added and a dark red brown powder was obtained after filtration. The product was purified
by flash column chromatography (silica gel, gradient elution: pure DCM to 20:80 EtOH:DCM)
to give the title compound7 as a dark red solid (2566 mg, 11.4 mmol, 60 %):
Rf = 0.19 [EtOAc : Hexane (1:1)];
1H NMR (400 MHz, CDCl3) δ 10.13 (1H, s, H1), 8.28
(1H, d, J = 8.8 Hz, H10), 8.11 (1H, d, J = 8.4 Hz, H7), 7.56 (1H, d, J = 8.8 Hz, H9), 7.45
(1H, d, J = 8.3 Hz, H6), 6.35 (1H, dd, J = 2.1, 0.8 Hz, H13), 2.79 (3H, s, H16), 2.55 (3H,
d, J = 0.7 Hz, H15); 13C NMR (101 MHz, CDCl3) δ 178.7 (C12), 158.3 (C5), 147.1 (C14),
138.1 (C3), 136.6 (C7), 136.5 (C2), 127.2 (C8), 124.4 (C6), 123.5 (C11), 122.3 (C10), 121.5
(C9), 112.6 (C13), 25.2 (C16), 20.5 (C15); HRMS (ESI+) calculated for [C14H12N2O + H]
+:
225.1023, m/z found: 225.1031 (∆ +3.6 ppm).
4-Chloro-2,9-dimethyl-1,10-phenanthroline (11)
87
6
5 N4
3
2
11
109
N1 14
13
12
Cl
1516
11
Phosphoryl chloride (30.0 mL, 321 mmol) was stirred under nitrogen atmosphere before addition
of 2,9-dimethyl-1,4-dihydro-1,10-phenantrolin-4-one (2.40 g, 10.7 mmol). The reaction mixture
7Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
178
Chapter 5: Materials and methods
was stirred at 90 ◦C for 3.5 h. While still hot, the reaction mixture was added to ice (100 g) and
water (50 mL) and stirred for 15 min. Then chloroform (120 mL) was added and the resulting
two-layer system was brought to pH 13-14 by NaOH addition. The aqueous layer was extracted
by DCM (3 × 150 mL), then washed by NaOH aqueous solution (1 M, 100 mL), dried over
anhydrous MgSO4 and concentrated in vacuo to afford the desired product 11 (2460 mg, 10.1
mmol, 95 %) as light tan crystals8:
1H NMR (400 MHz, CDCl3) δ 8.16 (1H, d, J = 9.0 Hz, H9), 8.16 (1H, d, J = 8.2 Hz, H7),
7.81 (1H, d, J = 9.0 Hz, H10), 7.60 (1H, s, H13), 7.53 (1H, d, J = 8.2 Hz, H6), 2.95 (3H, s,
H16), 2.93 (3H, s, H15); 13C NMR (101 MHz, CDCl3) δ 160.2 (C5), 159.5 (C14), 146.6 (C3),
145.1 (C2), 142.8 (C11), 136.5 (C7), 127.0 (C8), 126.6 (C10), 125.0 (C12), 124.2 (C6), 123.8
(C13), 121.3 (C9), 26.2 (C16), 25.9 (C15); HRMS (ESI+) calculated for [C14H11
35ClN2 + H]
+:
243.0684, m/z found: 243.0688 (∆ +1.8 ppm).
4-Chloro-2,9-bis(trichloromethyl)-1,10-phenanthroline (12)
87
6
5 N4
3
2
11
109
N1 14
13
12
Cl
CCl315Cl3C16
12
A stirred green solution of compound 11 (2.10 g, 8.65 mmol, 1 equiv.), N -chlorosuccinimide
(8.32 g, 62.3 mmol, 7.2 equiv.) and a catalytic amount of benzoyl peroxide (5.3 mg, 0.022 mmol,
0.0025 equiv.) was refluxed for 19 h in chloroform (160 mL). The obtained orange reaction
mixture was washed with saturated aqueous Na2CO3 (5 × 100 mL)9, dried over anhydrous
MgSO4 and concentrated in vacuo to afford the title compound 12 (3137 mg, 8.32 mmol,
96 %) as yellow crystals10:
LCMS tr = 5.5 min;
1H NMR (500 MHz, CDCl3) δ 8.48 (1H, d, J = 8.5 Hz, H7), 8.40 (1H,
s, H13), 8.37 (1H, d, J = 8.5 Hz, H10), 8.39 (1H, d, J = 9.0 Hz, H6), 8.07 (1H, d, J = 9.0
Hz, H9); 13C NMR (126 MHz, CDCl3) δ 158.6 (C14), 158.0 (C5), 144.5 (C2), 143.2 (C3),
138.4 (C7), 129.4 (C8), 128.7 (C9), 127.6 (C11), 123.8 (C12), 121.2 (C6), 120.9 (C13), 98.2
(C16), 97.4 (C15); HRMS (ESI+) calculated for C14H5
35Cl7N2: 445.8272, observed neutral
mass: 445.8291 (∆ +4.0 ppm).
8Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
9Extensive washing helped to completely remove unwanted reaction products and unreacted reagents, there-
fore chromatography purification was not required as opposed to the literature [307] method.
10Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
179
Chapter 5: Materials and methods
4-Chloro-1,10-phenanthroline-2,9-dicarboxylic acid (13)
87
6
5 N4
3
2
11
109
N1 14
13
12
Cl
1516
HO
17
O
OH
18
O
13
Sulphuric acid (conc., 4.7 mL) and compound 12 (1400 mg, 3.12 mmol) were stirred at 85 ◦C
for 2.5 h. The reaction mixture was then cooled to RT, before careful addition of water (20 mL).
The precipitate was filtered, washed by water (4 × 20 mL) and diethylether (3 × 20 mL) to
give the desired product 13 (927.6 mg, 3.07 mmol, 98 %) as a light tan solid11:
LCMS tr = 4.1 min;
1H NMR (400 MHz, DMSO-d6) δ 8.78 (1H, d, J = 8.3 Hz, H7), 8.48
(1H, s, H13), 8.43 (1H, d, J = 8.3 Hz, H6), 8.38 (1H, d, J = 9.1 Hz, H9), 8.34 (1H, d, J =
9.2 Hz, H10), 4.24 (2H, s, H17/H18); 13C NMR (101 MHz, DMSO-d6) δ 166.2 (C16), 165.4
(C15), 149.1 (C5), 148.5 (C14), 146.0 (C12), 144.5 (C3), 142.9 (C11), 138.5 (C7), 130.4 (C8),
130.1 (C9), 127.9 (C2), 124.0 (C6), 123.6 (C13), 123.3 (C10); HRMS (ESI+) calculated for
C14H7
35ClN2O4: 302.0094, observed neutral mass: 302.0106 (∆ +3.4 ppm).
4-Hydroxy-1,10-phenanthroline-2,9-dicarboxylic acid (18)
87
6
5 N4
3
2
11
109
N1 14
13
12
OH
19
1516
HO
17
O
OH
18
O
18
Sulphuric acid (conc., 0.6 mL) and compound 12 (150 mg, 0.33 mmol) were stirred at 150 ◦C
for 2 h. The reaction mixture was then cooled down to RT, before careful addition of water
(0.6 mL). The mixture was then stirred at 170 ◦C for 1 h and after cooling to RT water (70 mL)
was added. The precipitate formed was filtered, washed by water (2 × 10 mL) and diethylether
(3 × 25 mL) to give the desired product 18 (83.6 mg, 0.29 mmol, 88 %) as a light tan solid:
LCMS tr = 3.7 min;
1H NMR (500 MHz, DMSO-d6)
12 δ 13.93 (1H, s, H17/H18/H19), 12.56
(1H, s, H17/H18/H19), 11.50 (1H, s, H17/H18/H19), 8.73 (1H, d, J = 8.4 Hz, H7), 8.38 (1H,
11Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
12Assigned NMR spectra can be found in the appendix.
180
Chapter 5: Materials and methods
d, J = 8.4 Hz, H6), 8.23 (1H, d, J = 8.7 Hz, H10), 7.92 (1H, d, J = 8.9 Hz, H9) 6.90 (1H, s,
H13); 13C NMR (126 MHz, DMSO-d6) δ 177.1 (C12), 165.3 (C16), 163.4 (C15), 147.0 (C5),
138.6 (C7), 138.3 (C14), 137.9 (C3), 137.1 (C2), 130.7 (C8), 124.8 (C11), 124.2 (C10), 123.6
(C6), 122.8 (C9), 113.6 (C13); HRMS (ESI+) calculated for [C14H8N2O5 + H]
+: 285.0506,
m/z found: 285.0507 (∆ +0.4 ppm).
PhenDC3n-Cl (14)
87
6
5 N4
3
2
11
109
N1 14
13
12
Cl
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
14
3-Aminoiquinoline (870 mg, 6.0 mmol, 2.15 equiv.) and HOAt solution in DMF (0.5 M, 2.8 mL,
5.6 mmol, 2 equiv.) were successively added to a stirred solution of compound 13 (850 mg,
2.8 mmol, 1 equiv.) suspended in DMF (50 mL), under nitrogen atmosphere. The reaction
mixture was then cooled to 0 ◦C before addition of EDCI (1190 mg, 6.2 mmol, 2.2 equiv.).
The reaction was allowed to reach RT over 1 h and was stirred overnight. The pale yellow
precipitate obtained was filtered, washed with water (3 × 150 mL) and diethyl ether (3 ×
150 mL) to afford PhenDC3n-Cl (1221 mg, 2.20 mmol, 78 %) as a yellow solid13:
LCMS tr = 5.2 min;
1H NMR (500 MHz, DMSO-d6) δ 11.88 (1H, s, H20), 11.87 (1H, s,
H27), 9.674 (1H, s, H26), 9.669 (1H, s, H33), 9.15 (2H, t, J = 2.8 Hz, H22, H29), 8.97 (1H, d,
J = 8.4 Hz, H7), 8.74 (1H, s, H13), 8.73 (1H, d, J = 8.4 Hz, H6), 8.52 (1H, d, J = 9.2 Hz,
H10), 8.48 (1H, d, J = 9.2 Hz, H9), 8.14 - 8.07 (4H, m, H34, H37, H38, H41), 7.78 - 7.73 (2H,
m, H36, H40), 7.70 - 7.66 (2H, m, H35, H39); 13C NMR (126 MHz, DMSO-d6) δ 163.1 (C16),
162.4 (C15), 150.0 (C14), 149.6 (C5), 145.44 (C26), 145.4 (C33), 145.1 (C12), 144.7 (C24),
144.6 (C31), 143.8 (C11), 143.5 (C3), 139.2 (C7), 132.3 (C21), 132.2 (C28), 130.8 (C8), 130.8
(C9), 128.7 (C37), 128.5 (C41), 128.4 (C36), 128.3 (C2), 128.1 (C34), 128.0 (C38), 127.8 (C23),
127.8 (C30), 127.2 (C35) 127.2 (C39), 123.9 (C22), 123.7 (C29), 123.5 (C10), 122.3 (C6), 121.9
(C13); HRMS (ESI+) calculated for [C32H19
35ClN6O2 + H]
+: 555.1331, m/z found: 555.1331
(∆ 0 ppm).
13Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
181
Chapter 5: Materials and methods
PhenDC3-Cl (21)
87
6
5 N4
3
2
11
109
N1 14
13
12
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
4243 •2 CF3COO
Cl
21
A solution of PhenDC3n-Cl (14) (100 mg, 0.18 mmol, 1.0 equiv.) in DMF (5 mL) was heated
at 40 ◦C, under nitrogen atmosphere, before methyl iodide (0.67 mL, 10.8 mmol, 60 equiv.)
was added dropwise. The reaction mixture was stirred for 24 h. The solvent was evaporated
to dryness in vacuo and the resulting solid was washed with ethanol (2 × 10 mL). The crude
product was purified by flash column chromatography (C18 column, gradient elution: water
(0.1 % (v/v) TFA) to MeCN (0.1 % (v/v) TFA) over 30 min at a flow rate of 25 mL min−1) to
give the title compound 21 as a yellow powder (52.8 mg, 0.065 mmol, 36 %):
LCMS tr = 3.8 min;
1H NMR (500 MHz, DMSO-d6)
14 δ12.20 (1H, s, H20), 12.18 (1H, s,
H27), 10.31 (1H, s, H26), 10.30 (1H, s, H33), 9.91 (2H, s, H22, H29), 9.05 (1H, d, J = 8.4 Hz,
H7), 8.82 (1H, s, H13), 8.80 (1H, J = 7.6 Hz, H6), 8.60 - 8.49 (6H, m, H9, H10, H34, H37,
H38, H41), 8.23 - 8.29 (2H, m, H36, H40), 8.11 - 8.05 (2H, m, H35, H39), 4.74 (s, 6H, H42,
H43); 13C NMR (126 MHz, DMSO-d6) δ 163.4 (C15), 162.7 (C16), 149.0 (C14), 148.5 (C5),
145.7 (C26, C33), 145.1 (C2), 144.2 (C12), 143.6 (C3), 139.6 (C7), 135.6 (C24), 135.6 (C31),
134.9 (C22), 134.8 (C29), 134.0 (C36), 133.9 (C40), 132.7 (C21), 132.5 (C28), 131.3 (C8), 130.6
(C9), 130.3 (C35, C39), 130.0 (C34, C38), 129.1 (C23, C30), 128.8 (C11), 123.9 (C10), 122.7
(C6), 122.2 (C13), 119.2 (C37, C41), 46.1 (C42), 46.0 (C41); HRMS (ESI+) calculated for
[C36H25
35ClN6O4F3]
+: 697.1572, m/z found: 697.1563 (∆ -1.4 ppm).
14Assigned NMR spectra can be found in the appendix.
182
Chapter 5: Materials and methods
PhenDC3n-NH2 (15)
87
6
5 N4
3
2
11
109
N1 14
13
12
HN
42
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
43
44
45
46
NH2
47
15
A sealed vial containing PhenDC3n-Cl (14) (201 mg, 0.36 mmol) and 1,4-diaminobutane
(3.64 mL, 36 mmol) was heated at 120 ◦C under microwave irradiation in nitrogen atmosphere
for 1 h to give a homogeneous orange solution. The expected compound was precipitated upon
addition of MeCN (30 mL). The solid was filtered, washed with MeCN (3 × 30 mL) and Et2O
(3 × 40 mL) to afford title compound PhenDC3n-NH2 (15) as an orange powder15 (213 mg,
0.35 mmol, 97 %):
LCMS tr = 4.0 min;
1H NMR (400 MHz, DMSO-d6) δ 11.78 (1H, s), 9.67 (2H, d, J = 2.5
Hz), 9.13 (1H, d, J = 2.5 Hz), 9.08 (1H, d, J = 2.5 Hz), 8.78 (1H, d, J = 8.4 Hz), 8.59 (1H,
d, J = 8.3 Hz), 8.54 (1H, d, J = 9.2 Hz), 8.14 - 8.03 (2H, m), 7.87 (1H, s), 7.78 - 7.70 (2H,
m), 7.70 - 7.61 (3H, m), 3.57 - 3.44 (2H, m), 2.84 (2H, t, J = 7.3 Hz), 2.67 (2H, q, J = 4.8,
3.3 Hz), 1.83 (2H, q, J = 7.1 Hz), 1.70 (2H, p, J = 7.5 Hz); 13C NMR (126 MHz, DMSO-d6)
δ 165.8, 165.2, 164.2, 163.6, 152.2, 149.6, 149.0, 145.5, 145.4, 144.7, 144.6, 144.5, 144.2, 138.3,
132.6, 132.5, 130.4, 128.7, 128.3, 128.2, 128.0, 128.0, 127.9, 127.8, 127.1, 127.1, 127.1, 124.7,
123.6, 123.3, 122.6, 121.1, 119.1, 119.1, 50.5, 42.6, 28.2, 25.7; HRMS (ESI+) calculated for
[C36H30N8O2 + H]
+: 607.2565, m/z found: 607.2571 (∆ -1.0 ppm).
15Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
183
Chapter 5: Materials and methods
PhenDC3n-yne (22)
87
6
5 N4
3
2
11
109
N1 14
13
12
HN
42
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
43
44
45
46
NH
47
48
49
O
50 51
52
22
4-Pentyoic acid (24.9 mg, 0.25 mmol, 1.3 equiv.) was dissolved in DMF (7.3 mL) with
PhenDC3n-NH2 (15) (118.3 mg, 0.20 mmol, 1 equiv.), HOAt solution in DMF (0.5 M, 0.20 mL,
0.10 mmol, 0.5 equiv.) and EDCI (48.7 mg, 0.25 mmol, 1.3 equiv.). Then Et3N (35 µL, 0.25
mmol, 3 equiv.) was added and the reaction mixture was stirred overnight at RT with pro-
tection from light. DMF was removed in vacuo and the crude mixture was purified by flash
column chromatography (C18 column, gradient elution: water (0.1 % (v/v) TFA) to MeCN
(0.1 % (v/v) TFA) over 30 min at a flow rate of 25 mL min−1) to afford PhenDC3n-yne (22)
as a yellow powder16 (71.4 mg, 0.10 mmol, 53 %):
LCMS tr = 4.9 min;
1H NMR (400 MHz, DMSO-d6) δ 11.88 (1H, s), 11.81 (1H, s), 9.53 (1H,
s), 9.48 (1H, s), 9.07 (2H, d, J = 2.5 Hz), 8.89 (1H, d, J = 8.5 Hz), 8.69 - 8.59 (2H, m), 8.27
(1H, d, J = 9.1 Hz), 8.15 - 8.04 (4H, m), 7.98 (1H, t, J = 5.6 Hz), 7.81 - 7.62 (6H, m), 3.65
(2H, s), 3.19 (2H, q, J = 6.5 Hz), 2.74 (1H, t, J = 2.5 Hz), 2.35 (2H, dtt, J = 7.6, 6.4, 2.2 Hz),
2.28 (2H, dd, J = 7.9, 5.7 Hz), 1.82 (2H, p, J = 7.4 Hz), 1.63 (2H, p, J = 7.1 Hz); HRMS
(ESI+) calculated for [C41H34N8O3 + H]
+: 687.2827, m/z found: 687.2820 (∆ -1.0 ppm).
16Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
184
Chapter 5: Materials and methods
PhenDC3-yne (16)
87
6
5 N4
3
2
11
109
N1 14
13
12
HN
42
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
43
44
45
46
NH
47
48
49
O
50 51
52
•2 CF3COO
16
A solution of PhenDC3n-yne (22) in DMF (6 mL) was heated at 40 ◦C, under nitrogen atmo-
sphere, before methyl iodide (0.56 mL, 9.0 mmol) was added dropwise. The yellow reaction
mixture was stirred for 24 h. The solvent was evaporated to dryness in vacuo and the resulting
orange solid was washed with ethanol (2 × 10 mL). The crude product was purified by flash
column chromatography (C18 column, gradient elution: water (0.1 % (v/v) TFA) to MeCN
(0.1 % (v/v) TFA) over 30 min at a flow rate of 25 mL min−1) to give the title compound 16
as an orange powder17 (62.1 mg, 0.064 mmol):
LCMS tr = 3.7 min;
1H NMR (400 MHz, DMSO-d6) δ 12.15 (2H, s, H20, H27), 10.31 (1H,
d, J = 2.1 Hz, H33), 10.30 (1H, d, J = 1.9 Hz, H26), 9.89 (1H, s, H29), 9.88 (1H, s, H29), 8.87
(1H, d, J = 8.4 Hz, H7), 8.68 (1H, d, J = 8.3 Hz, H6), 8.62 (1H, d, J = 9.4 Hz, H10), 8.57
- 8.47 (4H, m, H34, H37, H38, H41), 8.28 - 8.22 (2H, m, H36, H40), 8.21 (1H, d, J = 9.5 Hz,
H9), 8.11 - 8.04 (3H, m, H35, H39, H42), 7.98 (1H, t, J = 5.6 Hz, H47), 7.70 (1H, s, H13), 4.73
(3H, s, H54), 4.72 (3H, s, H53), 3.52 (2H, q, J = 6.6 Hz, H43), 3.18 (2H, q, J = 6.8 Hz, H46),
2.75 (1H, t, J = 2.6 Hz, H52), 2.40 - 2.34 (2H, m, H50), 2.28 (2H, t, J = 7.0 Hz, H49), 1.80
(2H, p, J = 7.4 Hz, H44), 1.62 (2H, p, J = 7.1 Hz, H45); 13C NMR18 (101 MHz, DMSO-d6)
δ 170.2 (C48), 164.2 (C15), 163.7 (C16), 152.5 (C12), 148.0 (C5), 145.8 (C33), 145.7 (C26),
145.8 (C2), 145.7 (C3), 138.8 (C7), 135.5 (C31), 135.4 (C24), 134.7 (C29), 134.5 (C22), 133.9
(C40), 133.8 (C36), 132.8 (C21, C28), 131.0 (C8), 130.3 (C35, C39), 130.0 (C38), 129.9 (C34),
129.2 (C23, C30), 125.3 (C9), 122.9 (C10), 121.6 (C6), 119.4 (C11), 119.2 (C37, C41), 99.7
(C13), 83.8 (C51), 71.3 (C52), 46.0 (C53, C54), 42.6 (C43), 38.2 (C46), 34.3 (C49), 27.0 (C45),
25.1 (C44), 14.3 (C50); HRMS (ESI+) calculated for [C45H40N8O5F3]
+: 829.3068, m/z found:
829.3057 (∆ -1.4 ppm).
17Compound has been reported in the literature [307] and corresonding 1H NMR spectra were matched.
18C14 was not labelled due to a lack of HMBC coupling and it could not be determined with confidence.
185
Chapter 5: Materials and methods
PhenDC3-SiR (17)
87
6
5 N4
3
2
11
109
N1 14
13
12
HN
42
1516
HN20
O
NH27
O
21
28
22
23 24
N
2633
N
31 30
29
41
40 39
38 34
35 36
37
43
44
45
46
NH
47
7475
48
49
O
50
51
•2 CF3COO
62
61
60
59
58
63
57
O
NH
56
55
65 O67
64
Si 71
66
O
54
53
70
69
68
72
N
N
73
N N
N
52
17
Silicon Rhodamine dye azide (2.0 mg, 0.0036 mmol, 1.0 equiv.) and PhenDC3-yne (16) (14.0
mg, 0.014 mmol, 4.0 equiv.) were dissolved in water (200 µL) and tBuOH (300 µL). CuSO4
aqueous solution (1.8 M, 20 µL, 0.036 mmol, 10.0 equiv.) was added to the reaction mixture
followed by an aqueous solution of (+)-sodium L-ascorbate (2.2 M, 50 µL, 0.11 mmol, 30 equiv.).
The reaction was run at RT overnight while being monitored by LCMS. As no expected product
was yet to be observed, further CuSO4 aqueous solution (1.8 M, 50 µL, 0.090 mmol, 25.0 equiv.)
and aqueous solution of (+)-sodium L-ascorbate (2.2 M, 50 µL, 0.11 mmol, 30 equiv.) were
added to the reaction mixture followed by DMSO (50 µL) to improve the solubility. After
running the reaction for 2 h approximately 80 % conversion was observed and thus more CuSO4
aqueous solution (1.8 M, 50 µL, 0.090 mmol, 25.0 equiv.), aqueous solution of (+)-sodium L-
ascorbate (2.2 M, 50 µL, , 0.11 mmol, 30 equiv.) and DMSO (50 µL) were added. After another
2 h complete conversion to the product was observed by LCMS. The crude product was purified
by prep. HPLC (0-100 % MeCN (0.1 % (v/v) TFA)) to give PhenDC3-SiR (17) as a fine blue
powder (3.1 mg, 0.0021 mmol, 58 %)19:
LCMS tr = 3.1 min;
1H NMR (500 MHz, DMSO-d6)
20 δ 12.12 (1H, s, H27), 12.08 (1H, s,
H20), 10.31 (2H, s, H26, H33), 9.90 (1H, s, H29), 9.88 (1H, s, H22), 8.85 (1H, d, J = 8.3 Hz,
H7), 8.80 (1H, t, J = 5.3 Hz, H56), 8.68 (1H, dd, J = 8.3, 2.8 Hz, H6), 8.61 (1H, d, J = 9.4
Hz, H10), 8.57 - 8.45 (4H, m, H35, H36, H39, H40), 8.28 - 8.21 (2H, m, H34, H38), 8.18 (1H,
d, J = 9.2 Hz, H9), 8.11 - 8.04 (3H, m, H37, H41, H59), 8.01 (1H, d, J = 8.0 Hz, H60), 8.00
- 7.90 (2H, m, H42, H47), 7.82 (1H, s, H52), 7.72 - 7.64 (2H, m, H13, H63), 7.01 (2H, d, J
19After HPLC purification the product got contaminated by NH4CF3CO2 which was then sublimed off by
repeated cycles of freeze-drying.
20Full assigned NMR spectra can be found in the appendix.
186
Chapter 5: Materials and methods
= 2.2 Hz, H70), 6.66 - 6.57 (4H, m, H67, H68), 4.73 (3H, s, H75), 4.71 (3H, s, H74), 4.31
(2H, t, J = 7.0 Hz, H53), 3.44 (2H, s, H43)21, 3.21 (2H, q, J = 6.5 Hz, H55), 3.16 (2H, q, J
= 6.3 Hz, H46), 2.90 (12H, s, H73), 2.82 (2H, t, J = 7.8 Hz, H49), 2.45 - 2.38 (2H, m, H50),
2.00 (2H, p, J = 6.4 Hz, H54), 1.80 - 1.70 (2H, m, H44), 1.64 - 1.56 (2H, m, H45), 0.63 (3H,
s, H72/H79), 0.51 (3H, s, H72/H79); 13C NMR (126 MHz, DMSO-d6) δ 171.1 (C48), 169.3
(C64), 165.0 (C57), 164.4 (C15), 163.8 (C16), 154.8 (C58/C61/C62), 152.3 (C12), 149.2 (C69),
148.5 (C14), 147.9 (C5), 145.9 (C51), 145.8(C33), 145.7 (C26), 144.7 (C2), 144.2 (C3), 139.8
(C58/C61/C62), 138.8 (C7), 135.9 (C71), 135.5 (C31), 135.4 (C24), 134.7 (C29), 134.4(C22),
133.9 (C38), 133.7 (C34), 132.9 (C28), 132.8 (C21), 130.9 (C8), 130.4 (C66), 130.3 (C41), 130.2
(C37), 129.9 (C40), 129.9 (C36), 129.1 (C23, C30), 128.2 (C59), 127.6 (C67), 125.3 (C60), 125.2
(C9), 122.9 (C10), 122.8 (C63), 122.0 (C52), 121.5 (C6), 119.5 (C11), 119.2 (C35, C39), 116.4
(C70), 113.7 (C68), 99.5 (C13), 91.3 (C65), 47.1 (C53), 46.0 (C74, C75), 42.5 (C43), 38.1 (C46),
36.8 (C55), 35.0 (C50), 29.6 (C54), 27.0 (C45), 25.1 (C44), 21.4 (C49), 0.0 (C72/C79), -1.3
(C72/C79); HRMS (ESI+) calculated for [C73H74N14O6Si]
2+: 635.2837, m/z found: 635.2824
(∆ -2.0 ppm).
21The peak is overlapping with broad water signal, as determined by HSQC NMR.
187
Bibliography
[1] J. D. Watson and F. H. C. Crick, “Molecular Structure of Nucleic Acids: A Structure for
Deoxyribose Nucleic Acid,” Nature, vol. 171, pp. 737–738, Apr. 1953.
[2] M. H. F. Wilkins, A. R. Stokes, and H. R. Wilson, “Molecular Structure of Nucleic Acids:
Molecular Structure of Deoxypentose Nucleic Acids,” Nature, vol. 171, pp. 738–740, Apr.
1953.
[3] R. E. Franklin and R. G. Gosling, “Molecular Configuration in Sodium Thymonucleate,”
Nature, vol. 171, pp. 740–741, Apr. 1953.
[4] M. Meselson and F. W. Stahl, “The replication of DNA in Escherichia coli,” PNAS,
vol. 44, pp. 671–682, July 1958.
[5] J. Zhao, A. Bacolla, G. Wang, and K. M. Vasquez, “Non-B DNA structure-induced
genetic instability and evolution,” Cell. Mol. Life Sci., vol. 67, pp. 43–62, Jan. 2010.
[6] H. M. Wong and J. L. Huppert, “Stable G-quadruplexes are found outside nucleosome-
bound regions,” Mol. BioSyst., vol. 5, pp. 1713–1719, Nov. 2009.
[7] W. A. Bickmore and B. van Steensel, “Genome Architecture: Domain Organization of
Interphase Chromosomes,” Cell, vol. 152, pp. 1270–1284, Mar. 2013.
[8] J. Spiegel, S. Adhikari, and S. Balasubramanian, “The Structure and Function of DNA
G-Quadruplexes,” Trends in Chemistry, July 2019.
[9] D. Varshney, J. Spiegel, K. Zyner, D. Tannahill, and S. Balasubramanian, “The regulation
and functions of DNA and RNA G-quadruplexes,” Nat. Rev. Mol. Cell Biol., pp. 1–16,
Apr. 2020.
[10] I. Bang, “Untersuchungen u¨ber die Guanylsa¨ure,” Biochem. Z., vol. 26, pp. 293–311,
1910.
[11] M. Gellert, M. N. Lipsett, and D. R. Davies, “Helix Formation by Guanylic Acid,” PNAS,
vol. 48, pp. 2013–2018, Dec. 1962.
188
Bibliography
[12] K. Hoogsteen, “The crystal and molecular structure of a hydrogen-bonded complex be-
tween 1-methylthymine and 9-methyladenine,” Acta Crystallogr, vol. 16, pp. 907–916,
Sept. 1963.
[13] D. Sen and W. Gilbert, “Formation of parallel four-stranded complexes by guanine-rich
motifs in DNA and its implications for meiosis,” Nature, vol. 334, pp. 364–366, July 1988.
[14] C. Kang, X. Zhang, R. Ratliff, R. Moyzis, and A. Rich, “Crystal structure of four-stranded
Oxytricha telomeric DNA,” Nature, vol. 356, pp. 126–131, Mar. 1992.
[15] G. Laughlan, A. I. Murchie, D. G. Norman, M. H. Moore, P. C. Moody, D. M. Lilley, and
B. Luisi, “The high-resolution crystal structure of a parallel-stranded guanine tetraplex,”
Science, vol. 265, pp. 520–524, July 1994.
[16] K. Phillips, Z. Dauter, A. I. H. Murchie, D. M. J. Lilley, and B. Luisi, “The crystal
structure of a parallel-stranded guanine tetraplex at 0.95A˚ resolution,” J. Mol. Biol.,
vol. 273, pp. 171–182, Oct. 1997.
[17] G. N. Parkinson, M. P. H. Lee, and S. Neidle, “Crystal structure of parallel quadruplexes
from human telomeric DNA,” Nature, vol. 417, pp. 876–880, June 2002.
[18] A. T. Phan, V. Kuryavyi, S. Burge, S. Neidle, and D. J. Patel, “Structure of an Un-
precedented G-Quadruplex Scaffold in the Human c-kit Promoter,” J. Am. Chem. Soc.,
vol. 129, pp. 4386–4392, Apr. 2007.
[19] J. R. Williamson, M. K. Raghuraman, and T. R. Cech, “Monovalent cation-induced
structure of telomeric DNA: The G-quartet model,” Cell, vol. 59, pp. 871–880, Dec.
1989.
[20] D. Sen and W. Gilbert, “A sodium-potassium switch in the formation of four-stranded
G4-DNA,” Nature, vol. 344, pp. 410–414, Mar. 1990.
[21] P. Balagurumoorthy and S. K. Brahmachari, “Structure and stability of human telomeric
sequence.,” J. Biol. Chem., vol. 269, pp. 21858–21869, Aug. 1994.
[22] A. Risitano and K. R. Fox, “Stability of Intramolecular DNA Quadruplexes: Comparison
with DNA Duplexes,” Biochemistry, vol. 42, pp. 6507–6513, June 2003.
[23] W. Li, P. Wu, T. Ohmichi, and N. Sugimoto, “Characterization and thermodynamic
properties of quadruplex/duplex competition,” FEBS Lett., vol. 526, no. 1-3, pp. 77–81,
2002.
[24] A. N. Lane, J. B. Chaires, R. D. Gray, and J. O. Trent, “Stability and kinetics of G-
quadruplex structures,” Nucleic Acids Res., vol. 36, pp. 5482–5515, Oct. 2008.
189
Bibliography
[25] D. Miyoshi, H. Karimata, and N. Sugimoto, “Hydration Regulates Thermodynamics of
G-Quadruplex Formation under Molecular Crowding Conditions,” J. Am. Chem. Soc.,
vol. 128, pp. 7957–7963, June 2006.
[26] S. Burge, G. N. Parkinson, P. Hazel, A. K. Todd, and S. Neidle, “Quadruplex DNA:
Sequence, topology and structure,” Nucleic Acids Res., vol. 34, pp. 5402–5415, Nov.
2006.
[27] T. Simonsson, “G-Quadruplex DNA Structures Variations on a Theme,” Biol. Chem.,
vol. 382, no. 4, pp. 621–628, 2005.
[28] A. Bugaut and S. Balasubramanian, “A Sequence-Independent Study of the Influ-
ence of Short Loop Lengths on the Stability and Topology of Intramolecular DNA G-
Quadruplexes,” Biochemistry, vol. 47, pp. 689–697, Jan. 2008.
[29] S. Balasubramanian, L. H. Hurley, and S. Neidle, “Targeting G-quadruplexes in gene
promoters: A novel anticancer strategy?,” Nat. Rev. Drug Discov., vol. 10, pp. 261–275,
Apr. 2011.
[30] M. M. Fay, S. M. Lyons, and P. Ivanov, “RNA G-Quadruplexes in Biology: Principles
and Molecular Mechanisms,” J. Mol. Biol., vol. 429, pp. 2127–2147, July 2017.
[31] J.-y. Zhang, K.-w. Zheng, S. Xiao, Y.-h. Hao, and Z. Tan, “Mechanism and Manipulation
of DNA:RNA Hybrid G-Quadruplex Formation in Transcription of G-Rich DNA,” J. Am.
Chem. Soc., vol. 136, pp. 1381–1390, Jan. 2014.
[32] R.-y. Wu, K.-w. Zheng, J.-y. Zhang, Y.-h. Hao, and Z. Tan, “Formation of DNA:RNA
Hybrid G-Quadruplex in Bacterial Cells and Its Dominance over the Intramolecular DNA
G-Quadruplex in Mediating Transcription Termination,” Angew. Chem. Int. Ed., vol. 54,
no. 8, pp. 2447–2451, 2015.
[33] J. L. Huppert and S. Balasubramanian, “Prevalence of quadruplexes in the human
genome,” Nucleic Acids Res., vol. 33, no. 9, pp. 2908–2916, 2005.
[34] A. K. Todd, M. Johnston, and S. Neidle, “Highly prevalent putative quadruplex sequence
motifs in human DNA,” Nucleic Acids Res., vol. 33, no. 9, pp. 2901–2907, 2005.
[35] R. Giraldo, M. Suzuki, L. Chapman, and D. Rhodes, “Promotion of parallel DNA quadru-
plexes by a yeast telomere binding protein: A circular dichroism study,” PNAS, vol. 91,
pp. 7658–7662, Aug. 1994.
[36] P. Balagurumoorthy, S. K. Brahmachari, D. Mohanty, M. Bansal, and V. Sasisekharan,
“Hairpin and parallel quartet structures for telomeric sequences,” Nucleic Acids Res.,
vol. 20, pp. 4061–4067, Aug. 1992.
190
Bibliography
[37] Y. Wang and D. J. Patel, “Solution structure of the human telomeric repeat
d[AG3(T2AG3)3] G-tetraplex,” Structure, vol. 1, pp. 263–282, Dec. 1993.
[38] M. Cˇrnugelj, P. Sˇket, and J. Plavec, “Small Change in a G-Rich Sequence, a Dramatic
Change in Topology: New Dimeric G-Quadruplex Folding Motif with Unique Loop Ori-
entations,” J. Am. Chem. Soc., vol. 125, pp. 7866–7871, July 2003.
[39] J.-L. Mergny, A.-T. Phan, and L. Lacroix, “Following G-quartet formation by UV-
spectroscopy,” FEBS Lett., vol. 435, no. 1, pp. 74–78, 1998.
[40] A. Bourdoncle, A. Este´vez Torres, C. Gosse, L. Lacroix, P. Vekhoff, T. Le Saux, L. Jullien,
and J.-L. Mergny, “Quadruplex-Based Molecular Beacons as Tunable DNA Probes,” J.
Am. Chem. Soc., vol. 128, pp. 11094–11105, Aug. 2006.
[41] V. T. Mukundan and A. T. Phan, “Bulges in G-Quadruplexes: Broadening the Definition
of G-Quadruplex-Forming Sequences,” J. Am. Chem. Soc., vol. 135, pp. 5017–5028, Apr.
2013.
[42] A. Varizhuk, D. Ischenko, V. Tsvetkov, R. Novikov, N. Kulemin, D. Kaluzhny,
M. Vlasenok, V. Naumov, I. Smirnov, and G. Pozmogova, “The expanding repertoire
of G4 DNA structures,” Biochimie, vol. 135, pp. 54–62, Apr. 2017.
[43] O. Stegle, L. Payet, J.-L. Mergny, D. J. C. MacKay, and J. L. Huppert, “Predicting and
understanding the stability of G-quadruplexes,” Bioinformatics, vol. 25, pp. i374–i1382,
June 2009.
[44] A. Bedrat, L. Lacroix, and J.-L. Mergny, “Re-evaluation of G-quadruplex propensity with
G4Hunter,” Nucleic Acids Res., vol. 44, pp. 1746–1759, Feb. 2016.
[45] E. Belmonte-Reche and J. C. Morales, “G4-iM Grinder: When size and frequency mat-
ter. G-Quadruplex, i-Motif and higher order structure search and analysis tool,” NAR
Genomics and Bioinformatics, vol. 2, p. lqz005, Mar. 2020.
[46] A. B. Sahakyan, V. S. Chambers, G. Marsico, T. Santner, M. D. Antonio, and S. Balasub-
ramanian, “Machine learning model for sequence-driven DNA G-quadruplex formation,”
Sci. Rep., vol. 7, p. 14535, Nov. 2017.
[47] J. L. Huppert and S. Balasubramanian, “G-quadruplexes in promoters throughout the
human genome,” Nucleic Acids Res., vol. 35, pp. 406–413, Jan. 2007.
[48] J. A. Smestad and L. J. Maher, “Relationships between putative G-quadruplex-forming
sequences, RecQ helicases, and transcription,” BMC Med. Genet., vol. 16, p. 91, Oct.
2015. Pages 91 in PDF.
191
Bibliography
[49] A. Verma, V. K. Yadav, R. Basundra, A. Kumar, and S. Chowdhury, “Evidence of
genome-wide G4 DNA-mediated gene expression in human cancer cells,” Nucleic Acids
Res., vol. 37, pp. 4194–4204, July 2009.
[50] L. A. Cahoon and H. S. Seifert, “An Alternative DNA Structure Is Necessary for Pilin
Antigenic Variation in Neisseria gonorrhoeae,” Science, vol. 325, pp. 764–767, Aug. 2009.
[51] E. Besnard, A. Babled, L. Lapasset, O. Milhavet, H. Parrinello, C. Dantec, J.-M. Marin,
and J.-M. Lemaitre, “Unraveling cell type–specific and reprogrammable human replica-
tion origin signatures associated with G-quadruplex consensus motifs,” Nat. Struct. Mol.
Biol., vol. 19, pp. 837–844, Aug. 2012.
[52] V. S. Chambers, G. Marsico, J. M. Boutell, M. D. Antonio, G. P. Smith, and S. Balasub-
ramanian, “High-throughput sequencing of DNA G-quadruplex structures in the human
genome,” Nat. Biotechnol., vol. 33, pp. 877–881, Aug. 2015.
[53] D. R. Bentley, S. Balasubramanian, and H. P. Swerdlow, et al., “Accurate whole human
genome sequencing using reversible terminator chemistry,” Nature, vol. 456, pp. 53–59,
Nov. 2008.
[54] R. Rodriguez, K. M. Miller, J. V. Forment, C. R. Bradshaw, M. Nikan, S. Brit-
ton, T. Oelschlaegel, B. Xhemalce, S. Balasubramanian, and S. P. Jackson, “Small-
molecule–induced DNA damage identifies alternative DNA structures in human genes,”
Nat. Chem. Biol., vol. 8, pp. 301–310, Mar. 2012.
[55] A. Gue´din, J. Gros, P. Alberti, and J.-L. Mergny, “How long is too long? Effects of loop
size on G-quadruplex stability,” Nucleic Acids Res., vol. 38, pp. 7858–7868, Nov. 2010.
[56] C. K. Kwok, G. Marsico, A. B. Sahakyan, V. S. Chambers, and S. Balasubramanian,
“rG4-seq reveals widespread formation of G-quadruplex structures in the human tran-
scriptome,” Nat. Methods, vol. 13, pp. 841–844, Oct. 2016.
[57] G. Marsico, V. S. Chambers, A. B. Sahakyan, P. McCauley, J. M. Boutell, M. D. Antonio,
and S. Balasubramanian, “Whole genome experimental maps of DNA G-quadruplexes in
multiple species,” Nucleic Acids Res, vol. 47, pp. 3862–3874, May 2019.
[58] C. Schaffitzel, I. Berger, J. Postberg, J. Hanes, H. J. Lipps, and A. Plu¨ckthun, “In
vitro generated antibodies specific for telomeric guanine-quadruplex DNA react with
Stylonychia lemnae macronuclei,” Proc. Natl. Acad. Sci. U.S.A., vol. 98, pp. 8572–8577,
July 2001.
[59] G. Biffi, D. Tannahill, J. McCafferty, and S. Balasubramanian, “Quantitative visualiza-
tion of DNA G-quadruplex structures in human cells,” Nat. Chem., vol. 5, pp. 182–186,
Mar. 2013.
192
Bibliography
[60] G. Biffi, D. Tannahill, J. Miller, W. J. Howat, and S. Balasubramanian, “Elevated Levels
of G-Quadruplex Formation in Human Stomach and Liver Cancer Tissues,” PLOS ONE,
vol. 9, p. e102711, July 2014.
[61] R. Ha¨nsel-Hertsch, D. Beraldi, S. V. Lensing, G. Marsico, K. Zyner, A. Parry, M. D.
Antonio, J. Pike, H. Kimura, M. Narita, D. Tannahill, and S. Balasubramanian, “G-
quadruplex structures mark human regulatory chromatin,” Nat. Genet., vol. 48, pp. 1267–
1272, Oct. 2016.
[62] R. Ha¨nsel-Hertsch, J. Spiegel, G. Marsico, D. Tannahill, and S. Balasubramanian,
“Genome-wide mapping of endogenous G-quadruplex DNA structures by chromatin im-
munoprecipitation and high-throughput sequencing,” Nat. Protoc., vol. 13, pp. 551–564,
Mar. 2018.
[63] G. J. Hogan, C.-K. Lee, and J. D. Lieb, “Cell Cycle–Specified Fluctuation of Nucleosome
Occupancy at Gene Promoters,” PLOS Genet., vol. 2, p. e158, Sept. 2006.
[64] J. D. Buenrostro, P. G. Giresi, L. C. Zaba, H. Y. Chang, and W. J. Greenleaf, “Transpo-
sition of native chromatin for fast and sensitive epigenomic profiling of open chromatin,
DNA-binding proteins and nucleosome position,” Nat. Methods, vol. 10, pp. 1213–1218,
Dec. 2013.
[65] R. Rodriguez and K. M. Miller, “Unravelling the genomic targets of small molecules using
high-throughput sequencing,” Nat. Rev. Genet., vol. 15, pp. 783–796, Dec. 2014.
[66] S.-Q. Mao, A. T. Ghanbarian, J. Spiegel, S. M. Cuesta, D. Beraldi, M. D. Anto-
nio, G. Marsico, R. Ha¨nsel-Hertsch, D. Tannahill, and S. Balasubramanian, “DNA G-
quadruplex structures mold the DNA methylome,” Nat. Struct. Mol. Biol., p. 1, Oct.
2018.
[67] R. Ha¨nsel-Hertsch, A. Simeone, A. Shea, W. W. I. Hui, K. G. Zyner, G. Marsico, O. M.
Rueda, A. Bruna, A. Martin, X. Zhang, S. Adhikari, D. Tannahill, C. Caldas, and S. Bal-
asubramanian, “Landscape of G-quadruplex DNA structural regions in breast cancer,”
Nat. Genet., vol. 52, pp. 878–883, Sept. 2020.
[68] J. Shen, D. Varshney, A. Simeone, X. Zhang, S. Adhikari, D. Tannahill, and S. Balasub-
ramanian, “Promoter G-quadruplex folding precedes transcription and is controlled by
chromatin,” Genome Biology, vol. 22, p. 143, May 2021.
[69] J. Spiegel, S. M. Cuesta, S. Adhikari, R. Ha¨nsel-Hertsch, D. Tannahill, and S. Balasub-
ramanian, “G-quadruplexes are transcription factor binding hubs in human chromatin,”
Genome Biology, vol. 22, p. 117, Apr. 2021.
193
Bibliography
[70] S. Lago, M. Nadai, F. M. Cernilogar, M. Kazerani, H. Domı´niguez Moreno, G. Schotta,
and S. N. Richter, “Promoter G-quadruplexes and transcription factors cooperate to
shape the cell type-specific transcriptome,” Nat Commun, vol. 12, p. 3885, June 2021.
[71] K. G. Zyner, A. Simeone, S. M. Flynn, C. Doyle, G. Marsico, S. Adhikari, G. Portella,
D. Tannahill, and S. Balasubramanian, “G-quadruplex DNA structures in human stem
cells and differentiation,” Nat Commun, vol. 13, p. 142, Jan. 2022.
[72] Y. Hou, F. Li, R. Zhang, S. Li, H. Liu, Z. S. Qin, and X. Sun, “Integrative characterization
of G-Quadruplexes in the three-dimensional chromatin structure,” Epigenetics, vol. 14,
pp. 894–911, Sept. 2019.
[73] H. S. Kaya-Okur, S. J. Wu, C. A. Codomo, E. S. Pledger, T. D. Bryson, J. G. Henikoff,
K. Ahmad, and S. Henikoff, “CUT&Tag for efficient epigenomic profiling of small samples
and single cells,” Nat Commun, vol. 10, p. 1930, Apr. 2019.
[74] Q. Wang, H. Xiong, S. Ai, X. Yu, Y. Liu, J. Zhang, and A. He, “CoBATCH for High-
Throughput Single-Cell Epigenomic Profiling,” Molecular Cell, vol. 76, pp. 206–216.e7,
Oct. 2019.
[75] J. Lyu, R. Shao, P. Y. Kwong Yung, and S. J. Elsa¨sser, “Genome-wide mapping of G-
quadruplex structures with CUT&Tag,” Nucleic Acids Research, p. gkab1073, Nov. 2021.
[76] C. Li, H. Wang, Z. Yin, P. Fang, R. Xiao, Y. Xiang, W. Wang, Q. Li, B. Huang, J. Huang,
and K. Liang, “Ligand-induced native G-quadruplex stabilization impairs transcription
initiation,” Genome Res., vol. 31, pp. 1546–1560, Jan. 2021.
[77] W. W. I. Hui, A. Simeone, K. G. Zyner, D. Tannahill, and S. Balasubramanian, “Single-
cell mapping of DNA G-quadruplex structures in human cancer cells,” Sci Rep, vol. 11,
p. 23641, Dec. 2021.
[78] S. M. Haider, S. Neidle, and G. N. Parkinson, “A structural analysis of G-
quadruplex/ligand interactions,” Biochimie, vol. 93, pp. 1239–1251, Aug. 2011.
[79] R. Rodriguez, S. Mu¨ller, J. A. Yeoman, C. Trentesaux, J.-F. Riou, and S. Balasubra-
manian, “A Novel Small Molecule That Alters Shelterin Integrity and Triggers a DNA-
Damage Response at Telomeres,” J. Am. Chem. Soc., vol. 130, pp. 15758–15759, Nov.
2008.
[80] A. De Cian, E. DeLemos, J.-L. Mergny, M.-P. Teulade-Fichou, and D. Monchaud, “Highly
Efficient G-Quadruplex Recognition by Bisquinolinium Compounds,” J. Am. Chem. Soc.,
vol. 129, pp. 1856–1857, Feb. 2007.
194
Bibliography
[81] D. Drygin, A. Siddiqui-Jain, S. O’Brien, M. Schwaebe, A. Lin, J. Bliesath, C. B. Ho,
C. Proffitt, K. Trent, J. P. Whitten, J. K. C. Lim, D. V. Hoff, K. Anderes, and W. G.
Rice, “Anticancer Activity of CX-3543: A Direct Inhibitor of rRNA Biogenesis,” Cancer
Res., vol. 69, pp. 7653–7661, Oct. 2009.
[82] T. Tauchi, K. Shin-ya, G. Sashida, M. Sumi, A. Nakajima, T. Shimamoto, J. H. Ohyashiki,
and K. Ohyashiki, “Activity of a novel G-quadruplex-interactive telomerase inhibitor,
telomestatin (SOT-095), against human leukemia cells: Involvement of ATM-dependent
DNA damage response pathways,” Oncogene, vol. 22, pp. 5338–5347, Aug. 2003.
[83] D. Drygin, A. Lin, and J. Bliesath, et al, “Targeting RNA Polymerase I with an Oral
Small Molecule CX-5461 Inhibits Ribosomal RNA Synthesis and Solid Tumor Growth,”
Cancer Res., vol. 71, pp. 1418–1430, Feb. 2011.
[84] E. Izbicka, R. T. Wheelhouse, E. Raymond, K. K. Davidson, R. A. Lawrence, D. Sun,
B. E. Windle, L. H. Hurley, and D. D. V. Hoff, “Effects of Cationic Porphyrins as G-
Quadruplex Interactive Agents in Human Tumor Cells,” Cancer Res., vol. 59, pp. 639–
644, Feb. 1999.
[85] M. Read, R. J. Harrison, B. Romagnoli, F. A. Tanious, S. H. Gowan, A. P. Reszka, W. D.
Wilson, L. R. Kelland, and S. Neidle, “Structure-based design of selective and potent G
quadruplex-mediated telomerase inhibitors,” PNAS, vol. 98, pp. 4844–4849, Apr. 2001.
[86] D. D. Le, M. D. Antonio, L. K. M. Chan, and S. Balasubramanian, “G-quadruplex ligands
exhibit differential G-tetrad selectivity,” Chem. Commun., vol. 51, pp. 8048–8050, Apr.
2015.
[87] P. M. Yangyuoru, M. Di Antonio, C. Ghimire, G. Biffi, S. Balasubramanian, and H. Mao,
“Dual Binding of an Antibody and a Small Molecule Increases the Stability of TERRA
G-Quadruplex,” Angew. Chem. Int. Ed., vol. 54, no. 3, pp. 910–913, 2015.
[88] M. J. Lecours, A. Marchand, A. Anwar, C. Guetta, W. S. Hopkins, and V. Gabelica,
“What stoichiometries determined by mass spectrometry reveal about the ligand bind-
ing mode to G-quadruplex nucleic acids,” Biochim. Biophys. Acta BBA - Gen. Subj.,
vol. 1861, pp. 1353–1361, May 2017.
[89] X. Zhang, J. Spiegel, S. Mart´ınez Cuesta, S. Adhikari, and S. Balasubramanian, “Chemi-
cal profiling of DNA G-quadruplex-interacting proteins in live cells,” Nat. Chem., vol. 13,
pp. 626–633, July 2021.
[90] S. Neidle, “Quadruplex Nucleic Acids as Novel Therapeutic Targets,” J. Med. Chem.,
vol. 59, pp. 5987–6011, July 2016.
195
Bibliography
[91] A. T. Phan, V. Kuryavyi, H. Y. Gaw, and D. J. Patel, “Small-molecule interaction with a
five-guanine-tract G-quadruplex structure from the human MYC promoter,” Nat. Chem.
Biol., vol. 1, pp. 167–173, Aug. 2005.
[92] E. W. White, F. Tanious, M. A. Ismail, A. P. Reszka, S. Neidle, D. W. Boykin, and W. D.
Wilson, “Structure-specific recognition of quadruplex DNA by organic cations: Influence
of shape, substituents and charge,” Biophysical Chemistry, vol. 126, pp. 140–153, Mar.
2007.
[93] D. Monchaud and M.-P. Teulade-Fichou, “A hitchhiker’s guide to G-quadruplex ligands,”
Org. Biomol. Chem., vol. 6, no. 4, pp. 627–636, 2008.
[94] J. M. Nicoludis, S. P. Barrett, J.-L. Mergny, and L. A. Yatsunyk, “Interaction of human
telomeric DNA with N- methyl mesoporphyrin IX,” Nucleic Acids Research, vol. 40,
pp. 5432–5447, July 2012.
[95] S. Ray, D. Tillo, R. E. Boer, N. Assad, M. Barshai, G. Wu, Y. Orenstein, D. Yang,
J. Schneekloth, John S, and C. Vinson, “Custom DNA microarrays reveal diverse binding
preferences of proteins and small molecules to thousands of G-quadruplexes,” ACS Chem.
Biol., Mar. 2020.
[96] A. Y. Q. Zhang and S. Balasubramanian, “The Kinetics and Folding Pathways of In-
tramolecular G-Quadruplex Nucleic Acids,” J. Am. Chem. Soc., vol. 134, pp. 19297–
19308, Nov. 2012.
[97] S. S. Patel and I. Donmez, “Mechanisms of Helicases,” J. Biol. Chem., vol. 281, pp. 18265–
18268, July 2006.
[98] O. Mendoza, A. Bourdoncle, J.-B. Boule´, R. M. Brosh, and J.-L. Mergny, “G-
quadruplexes and helicases,” Nucleic Acids Res., vol. 44, pp. 1989–2006, Mar. 2016.
[99] K. Paeschke, M. L. Bochman, P. D. Garcia, P. Cejka, K. L. Friedman, S. C. Kowal-
czykowski, and V. A. Zakian, “Pif1 family helicases suppress genome instability at G-
quadruplex motifs,” Nature, vol. 497, pp. 458–462, May 2013.
[100] S. Hashimoto and J. M. Egly, “Trichothiodystrophy view from the molecular basis of
DNA repair/transcription factor TFIIH,” Hum. Mol. Genet., vol. 18, pp. R224–R230,
Oct. 2009.
[101] A. R. Lehmann, “DNA repair-deficient diseases, xeroderma pigmentosum, Cockayne syn-
drome and trichothiodystrophy,” Biochimie, vol. 85, pp. 1101–1111, Nov. 2003.
196
Bibliography
[102] L. T. Gray, A. C. Vallur, J. Eddy, and N. Maizels, “G quadruplexes are genomewide
targets of transcriptional helicases XPB and XPD,” Nat. Chem. Biol., vol. 10, pp. 313–
318, Apr. 2014.
[103] J. E. Johnson, K. Cao, P. Ryvkin, L.-S. Wang, and F. B. Johnson, “Altered gene ex-
pression in the Werner and Bloom syndromes is associated with sequences having G-
quadruplex forming potential,” Nucleic Acids Res., vol. 38, pp. 1114–1122, Mar. 2010.
[104] G. H. Nguyen, W. Tang, A. I. Robles, R. P. Beyer, L. T. Gray, J. A. Welsh, A. J.
Schetter, K. Kumamoto, X. W. Wang, I. D. Hickson, N. Maizels, R. J. Monnat, and
C. C. Harris, “Regulation of gene expression by the BLM helicase correlates with the
presence of G-quadruplex DNA motifs,” PNAS, vol. 111, pp. 9905–9910, July 2014.
[105] W. Tang, A. I. Robles, R. P. Beyer, L. T. Gray, G. H. Nguyen, J. Oshima, N. Maizels,
C. C. Harris, and R. J. Monnat, “The Werner syndrome RECQ helicase targets G4 DNA
in human cells to modulate transcription,” Hum. Mol. Genet., vol. 25, pp. 2060–2069,
May 2016.
[106] M. C. Chen, R. Tippana, N. A. Demeshkina, P. Murat, S. Balasubramanian, S. My-
ong, and A. R. Ferre´-D’Amare´, “Structural basis of G-quadruplex unfolding by the
DEAH/RHA helicase DHX36,” Nature, p. 1, June 2018.
[107] K.-w. Zheng, J.-y. Zhang, Y.-d. He, J.-y. Gong, C.-j. Wen, J.-n. Chen, Y.-h. Hao, Y. Zhao,
and Z. Tan, “Detection of genomic G-quadruplexes in living cells using a small artificial
protein,” Nucleic Acids Research, vol. 48, pp. 11706–11720, Nov. 2020.
[108] D. Rhodes and H. J. Lipps, “G-quadruplexes and their regulatory roles in biology,” Nu-
cleic Acids Res., vol. 43, pp. 8627–8637, Oct. 2015.
[109] M. L. Bochman, K. Paeschke, and V. A. Zakian, “DNA secondary structures: Stability
and function of G-quadruplex structures,” Nat. Rev. Genet., vol. 13, pp. 770–780, Nov.
2012.
[110] R. Rigo, M. Palumbo, and C. Sissi, “G-quadruplexes in human promoters: A challenge for
therapeutic applications,” Biochim. Biophys. Acta BBA - Gen. Subj., vol. 1861, pp. 1399–
1413, May 2017.
[111] R. Ha¨nsel-Hertsch, M. D. Antonio, and S. Balasubramanian, “DNA G-quadruplexes in
the human genome: Detection, functions and therapeutic potential,” Nat. Rev. Mol. Cell
Biol., vol. 18, pp. 279–284, May 2017.
[112] J. Robinson, F. Raguseo, S. P. Nuccio, D. Liano, and M. Di Antonio, “DNA G-quadruplex
structures: More than simple roadblocks to transcription?,” Nucleic Acids Research,
vol. 49, pp. 8419–8431, Sept. 2021.
197
Bibliography
[113] A. Siddiqui-Jain, C. L. Grand, D. J. Bearss, and L. H. Hurley, “Direct evidence for a
G-quadruplex in a promoter region and its targeting with a small molecule to repress
c-MYC transcription,” PNAS, vol. 99, pp. 11593–11598, Sept. 2002.
[114] S. Cogoi and L. E. Xodo, “G-quadruplex formation within the promoter of the KRAS
proto-oncogene and its effect on transcription,” Nucleic Acids Res., vol. 34, no. 9,
pp. 2536–2549, 2006.
[115] M. Paramasivam, A. Membrino, S. Cogoi, H. Fukuda, H. Nakagama, and L. E. Xodo,
“Protein hnRNP A1 and its derivative Up1 unfold quadruplex DNA in the human KRAS
promoter: Implications for transcription,” Nucleic Acids Res., vol. 37, pp. 2841–2853,
May 2009.
[116] J. F. Moruno-Manchon, E. C. Koellhoffer, J. Gopakumar, S. Hambarde, N. Kim, L. D.
McCullough, and A. S. Tsvetkov, “The G-quadruplex DNA stabilizing drug pyridostatin
promotes DNA damage and downregulates transcription of Brca1 in neurons,” Aging,
Sept. 2017.
[117] S. Cogoi, A. E. Shchekotikhin, and L. E. Xodo, “HRAS is silenced by two neighbor-
ing G-quadruplexes and activated by MAZ, a zinc-finger transcription factor with DNA
unfolding property,” Nucleic Acids Res., vol. 42, pp. 8379–8388, July 2014.
[118] M. Bejugam, S. Sewitz, P. S. Shirude, R. Rodriguez, R. Shahid, and S. Balasubrama-
nian, “Trisubstituted Isoalloxazines as a New Class of G-Quadruplex Binding Ligands:
Small Molecule Regulation of c-kit Oncogene Expression,” J. Am. Chem. Soc., vol. 129,
pp. 12926–12927, Oct. 2007.
[119] C. Marchetti, K. G. Zyner, S. A. Ohnmacht, M. Robson, S. M. Haider, J. P. Morton,
G. Marsico, T. Vo, S. Laughlin-Toth, A. A. Ahmed, G. Di Vita, I. Pazitna, M. Gunarat-
nam, R. J. Besser, A. C. G. Andrade, S. Diocou, J. A. Pike, D. Tannahill, R. B. Pedley,
T. R. J. Evans, W. D. Wilson, S. Balasubramanian, and S. Neidle, “Targeting Multiple
Effector Pathways in Pancreatic Ductal Adenocarcinoma with a G-Quadruplex-Binding
Small Molecule,” J. Med. Chem., vol. 61, pp. 2500–2517, Mar. 2018.
[120] Z. Siegfried, S. Eden, M. Mendelsohn, X. Feng, B.-Z. Tsuberi, and H. Cedar, “DNA
methylation represses transcription in vivo,” Nat Genet, vol. 22, pp. 203–206, June 1999.
[121] H.-P. Gu, S. Lin, M. Xu, H.-Y. Yu, X.-J. Du, Y.-Y. Zhang, G. Yuan, and W. Gao,
“Up-Regulating Relaxin Expression by G-Quadruplex Interactive Ligand to Achieve An-
tifibrotic Action,” Endocrinology, vol. 153, pp. 3692–3700, Aug. 2012.
198
Bibliography
[122] I. T. Holder and J. S. Hartig, “A Matter of Location: Influence of G-Quadruplexes on
Escherichia coli Gene Expression,” Chemistry & Biology, vol. 21, pp. 1511–1521, Nov.
2014.
[123] T. Agarwal, S. Roy, S. Kumar, T. K. Chakraborty, and S. Maiti, “In the Sense of Tran-
scription Regulation by G-Quadruplexes: Asymmetric Effects in Sense and Antisense
Strands,” Biochemistry, vol. 53, pp. 3711–3718, June 2014.
[124] A. M. Fleming, Y. Ding, and C. J. Burrows, “Oxidative DNA damage is epigenetic by
regulating gene transcription via base excision repair,” PNAS, vol. 114, pp. 2604–2609,
Mar. 2017.
[125] B. I. Fedeles, “G-quadruplex–forming promoter sequences enable transcriptional activa-
tion in response to oxidative stress,” PNAS, vol. 114, pp. 2788–2790, Mar. 2017.
[126] S. Roychoudhury, S. Pramanik, H. L. Harris, M. Tarpley, A. Sarkar, G. Spagnol, P. L.
Sorgen, D. Chowdhury, V. Band, D. Klinkebiel, and K. K. Bhakat, “Endogenous oxidized
DNA bases and APE1 regulate the formation of G-quadruplex structures in the genome,”
PNAS, vol. 117, pp. 11409–11420, May 2020.
[127] S. Cogoi, A. Ferino, G. Miglietta, E. B. Pedersen, and L. E. Xodo, “The regulatory G4
motif of the Kirsten ras (KRAS) gene is sensitive to guanine oxidation: Implications on
transcription,” Nucleic Acids Res., vol. 46, pp. 661–676, Jan. 2018.
[128] M.-N. Prioleau, “G-Quadruplexes and DNA Replication Origins,” in DNA Replication,
Advances in Experimental Medicine and Biology, pp. 273–286, Springer, Singapore, 2017.
[129] A.-L. Valton and M.-N. Prioleau, “G-Quadruplexes in DNA Replication: A Problem or
a Necessity?,” Trends Genet., vol. 32, pp. 697–706, Nov. 2016.
[130] L. K. Lerner and J. E. Sale, “Replication of G Quadruplex DNA,” Genes (Basel), vol. 10,
p. 95, Jan. 2019.
[131] J. Lopes, A. Piazza, R. Bermejo, B. Kriegsman, A. Colosio, M.-P. Teulade-Fichou,
M. Foiani, and A. Nicolas, “G-quadruplex-induced instability during leading-strand repli-
cation: G-quadruplex-induced instability,” EMBO J., vol. 30, pp. 4033–4046, Oct. 2011.
[132] P. Sarkies, C. Reams, L. J. Simpson, and J. E. Sale, “Epigenetic Instability due to
Defective Replication of Structured DNA,” Mol. Cell, vol. 40, pp. 703–713, Dec. 2010.
[133] P. Murat, G. Guilbaud, and J. E. Sale, “DNA polymerase stalling at structured DNA
constrains the expansion of short tandem repeats,” Genome Biology, vol. 21, p. 209, Aug.
2020.
199
Bibliography
[134] P. Sarkies, P. Murat, L. G. Phillips, K. Patel, S. Balasubramanian, and J. E. Sale,
“FANCJ coordinates two pathways that maintain epigenetic stability at G-quadruplex
DNA,” Nucleic Acids Res., vol. 40, pp. 1485–1498, Feb. 2012.
[135] T. B. C. London, L. J. Barber, G. Mosedale, G. P. Kelly, S. Balasubramanian, I. D.
Hickson, S. J. Boulton, and K. Hiom, “FANCJ Is a Structure-specific DNA Helicase
Associated with the Maintenance of Genomic G/C Tracts,” J. Biol. Chem., vol. 283,
pp. 36132–36139, Dec. 2008.
[136] G. Guilbaud, P. Murat, B. Recolin, B. C. Campbell, A. Maiter, J. E. Sale, and S. Bala-
subramanian, “Local epigenetic reprogramming induced by G-quadruplex ligands,” Nat.
Chem., vol. 9, pp. 1110–1117, Nov. 2017.
[137] C. Papadopoulou, G. Guilbaud, D. Schiavone, and J. E. Sale, “Nucleotide Pool Deple-
tion Induces G-Quadruplex-Dependent Perturbation of Gene Expression,” Cell Reports,
vol. 13, pp. 2491–2503, Dec. 2015.
[138] F. Picard, J.-C. Cadoret, B. Audit, A. Arneodo, A. Alberti, C. Battail, L. Duret, and
M.-N. Prioleau, “The Spatiotemporal Program of DNA Replication Is Associated with
Specific Combinations of Chromatin Marks in Human Cells,” PLOS Genet., vol. 10,
p. e1004282, May 2014.
[139] S. Hoshina, K. Yura, H. Teranishi, N. Kiyasu, A. Tominaga, H. Kadoma, A. Nakatsuka,
T. Kunichika, C. Obuse, and S. Waga, “Human origin recognition complex binds pref-
erentially to G-quadruplex-preferable RNA and single-stranded DNA,” J. Biol. Chem.,
vol. 288, pp. 30161–30171, Oct. 2013.
[140] P. Prorok, M. Artufel, A. Aze, P. Coulombe, I. Peiffer, L. Lacroix, A. Gue´din, J.-L.
Mergny, J. Damaschke, A. Schepers, B. Ballester, and M. Me´chali, “Involvement of G-
quadruplex regions in mammalian replication origin activity,” Nat Commun, vol. 10,
pp. 1–16, July 2019.
[141] Y. Kanoh, S. Matsumoto, R. Fukatsu, N. Kakusho, N. Kono, C. Renard-Guillet, K. Ma-
suda, K. Iida, K. Nagasawa, K. Shirahige, and H. Masai, “Rif1 binds to G quadruplexes
and suppresses replication over long distances,” Nat. Struct. Mol. Biol., vol. 22, pp. 889–
897, Nov. 2015.
[142] H. Masai, R. Fukatsu, N. Kakusho, Y. Kanoh, K. Moriyama, Y. Ma, K. Iida, and K. Na-
gasawa, “Rif1 promotes association of G-quadruplex (G4) by its specific G4 binding and
oligomerization activities,” Sci. Rep., vol. 9, p. 8618, June 2019.
[143] J. Meyne, R. L. Ratliff, and R. K. Moyzis, “Conservation of the human telomere sequence
(TTAGGG)n among vertebrates,” PNAS, vol. 86, pp. 7049–7053, Sept. 1989.
200
Bibliography
[144] H.-L. Bao, H.-s. Liu, and Y. Xu, “Hybrid-type and two-tetrad antiparallel telomere DNA
G-quadruplex structures in living human cells,” Nucleic Acids Research, vol. 47, pp. 4940–
4947, June 2019.
[145] A. J. Zaug, E. R. Podell, and T. R. Cech, “Human POT1 disrupts telomeric G-
quadruplexes allowing telomerase extension in vitro,” PNAS, vol. 102, pp. 10864–10869,
Aug. 2005.
[146] I. M. Pedroso, W. Hayward, and T. M. Fletcher, “The effect of the TRF2 N-terminal
and TRFH regions on telomeric G-quadruplex structures,” Nucleic Acids Res., vol. 37,
pp. 1541–1554, Apr. 2009.
[147] H. Wang, G. J. Nora, H. Ghodke, and P. L. Opresko, “Single Molecule Studies of Physio-
logically Relevant Telomeric Tails Reveal POT1 Mechanism for Promoting G-quadruplex
Unfolding,” J. Biol. Chem., vol. 286, pp. 7479–7489, Apr. 2011.
[148] A. M. Zahler, J. R. Williamson, T. R. Cech, and D. M. Prescott, “Inhibition of telomerase
by G-quartet DMA structures,” Nature, vol. 350, pp. 718–720, Apr. 1991.
[149] Q. Wang, J.-q. Liu, Z. Chen, K.-w. Zheng, C.-y. Chen, Y.-h. Hao, and Z. Tan, “G-
quadruplex formation at the 3′ end of telomere DNA inhibits its extension by telomerase,
polymerase and unwinding by helicase,” Nucleic Acids Res., vol. 39, pp. 6229–6237, Aug.
2011.
[150] J. S. Smith, Q. Chen, L. A. Yatsunyk, J. M. Nicoludis, M. S. Garcia, R. Kranaster, S. Bal-
asubramanian, D. Monchaud, M.-P. Teulade-Fichou, L. Abramowitz, D. C. Schultz, and
F. B. Johnson, “Rudimentary G-quadruplex–based telomere capping in Saccharomyces
cerevisiae,” Nat Struct Mol Biol, vol. 18, pp. 478–485, Apr. 2011.
[151] J. Nandakumar and T. R. Cech, “Finding the end: Recruitment of telomerase to telom-
eres,” Nat. Rev. Mol. Cell Biol., vol. 14, pp. 69–82, Feb. 2013.
[152] B. P. Paudel, A. L. Moye, H. Abou Assi, R. El-Khoury, S. B. Cohen, J. K. Holien,
M. L. Birrento, S. Samosorn, K. Intharapichai, C. G. Tomlinson, M.-P. Teulade-Fichou,
C. Gonza´lez, J. L. Beck, M. J. Damha, A. M. van Oijen, and T. M. Bryan, “A mechanism
for the extension and unfolding of parallel telomeric G-quadruplexes by human telomerase
at single-molecule resolution,” eLife, vol. 9, p. e56428, July 2020.
[153] N. van Wietmarschen, S. Merzouk, N. Halsema, D. C. J. Spierings, V. Guryev, and P. M.
Lansdorp, “BLM helicase suppresses recombination at G-quadruplex motifs in transcribed
genes,” Nat. Commun., vol. 9, p. 271, Jan. 2018.
201
Bibliography
[154] D. Piekna-Przybylska, M. A. Sullivan, G. Sharma, and R. A. Bambara, “U3 Region in
the HIV-1 Genome Adopts a G-Quadruplex Structure in Its RNA and DNA Sequence,”
Biochemistry, vol. 53, pp. 2581–2593, Apr. 2014.
[155] A. Bugaut and S. Balasubramanian, “5′-UTR RNA G-quadruplexes: Translation regula-
tion and targeting,” Nucleic Acids Res., vol. 40, pp. 4727–4741, June 2012.
[156] C. K. Kwok, G. Marsico, and S. Balasubramanian, “Detecting RNA G-Quadruplexes
(rG4s) in the Transcriptome,” Cold Spring Harb Perspect Biol, vol. 10, p. a032284, Jan.
2018.
[157] S. Kumari, A. Bugaut, J. L. Huppert, and S. Balasubramanian, “An RNA G-quadruplex
in the 5′ UTR of the NRAS proto-oncogene modulates translation,” Nat. Chem. Biol.,
vol. 3, pp. 218–221, Apr. 2007.
[158] R. Shahid, A. Bugaut, and S. Balasubramanian, “The BCL-2 5′ Untranslated Region
Contains an RNA G-Quadruplex-Forming Motif That Modulates Protein Expression,”
Biochemistry, vol. 49, pp. 8300–8306, Sept. 2010.
[159] D. Varshney, S. M. Cuesta, B. Herdy, U. B. Abdullah, D. Tannahill, and S. Balasubra-
manian, “RNA G-quadruplex structures control ribosomal protein production,” Sci Rep,
vol. 11, p. 22735, Nov. 2021.
[160] D. Benhalevy, S. K. Gupta, C. H. Danan, S. Ghosal, H.-W. Sun, H. G. Kazemier,
K. Paeschke, M. Hafner, and S. A. Juranek, “The Human CCHC-type Zinc Finger Nu-
cleic Acid-Binding Protein Binds G-Rich Elements in Target mRNA Coding Sequences
and Promotes Translation,” Cell Reports, vol. 18, pp. 2979–2990, Mar. 2017.
[161] M. Sauer, S. A. Juranek, J. Marks, A. D. Magis, H. G. Kazemier, D. Hilbig, D. Benhalevy,
X. Wang, M. Hafner, and K. Paeschke, “DHX36 prevents the accumulation of transla-
tionally inactive mRNAs with G4-structures in untranslated regions,” Nat. Commun.,
vol. 10, p. 2421, June 2019.
[162] P. Murat, G. Marsico, B. Herdy, A. Ghanbarian, G. Portella, and S. Balasubramanian,
“RNA G-quadruplexes at upstream open reading frames cause DHX36- and DHX9-
dependent translation of human mRNAs,” Genome Biol, vol. 19, p. 229, Dec. 2018.
[163] P. Murat, J. Zhong, L. Lekieffre, N. P. Cowieson, J. L. Clancy, T. Preiss, S. Balasubrama-
nian, R. Khanna, and J. Tellam, “G-quadruplexes regulate Epstein-Barr virus–encoded
nuclear antigen 1 mRNA translation,” Nat. Chem. Biol., vol. 10, pp. 358–364, May 2014.
[164] R. Perrone, M. Nadai, I. Frasson, J. A. Poe, E. Butovskaya, T. E. Smithgall, M. Palumbo,
G. Palu`, and S. N. Richter, “A Dynamic G-Quadruplex Region Regulates the HIV-1 Long
Terminal Repeat Promoter,” J. Med. Chem., vol. 56, pp. 6521–6530, Aug. 2013.
202
Bibliography
[165] M. Scalabrin, I. Frasson, E. Ruggiero, R. Perrone, E. Tosoni, S. Lago, M. Tassinari,
G. Palu`, and S. N. Richter, “The cellular protein hnRNP A2/B1 enhances HIV-1 tran-
scription by unfolding LTR promoter G-quadruplexes,” Sci Rep, vol. 7, p. 45244, Mar.
2017.
[166] R. Perrone, E. Butovskaya, D. Daelemans, G. Palu`, C. Pannecouque, and S. N. Richter,
“Anti-HIV-1 activity of the G-quadruplex ligand BRACO-19,” J Antimicrob Chemother,
vol. 69, pp. 3248–3258, Dec. 2014.
[167] E. Ruggiero and S. N. Richter, “G-quadruplexes and G-quadruplex ligands: Targets and
tools in antiviral therapy,” Nucleic Acids Res., vol. 46, pp. 3270–3283, Apr. 2018.
[168] M. Me´tifiot, S. Amrane, S. Litvak, and M.-L. Andreola, “G-quadruplexes in viruses:
Function and potential therapeutic applications,” Nucleic Acids Res, vol. 42, pp. 12352–
12366, Nov. 2014.
[169] N. Maizels, “G4-associated human diseases,” EMBO Rep, vol. 16, pp. 910–922, Aug.
2015.
[170] R. Simone, P. Fratta, S. Neidle, G. N. Parkinson, and A. M. Isaacs, “G-quadruplexes:
Emerging roles in neurodegenerative diseases and the non-coding transcriptome,” FEBS
Lett., vol. 589, no. 14, pp. 1653–1668, 2015.
[171] E. Wang, R. Thombre, Y. Shah, R. Latanich, and J. Wang, “G-Quadruplexes as
pathogenic drivers in neurodegenerative disorders,” Nucleic Acids Research, vol. 49,
pp. 4816–4830, May 2021.
[172] P. Boukamp, R. T. Petrussevska, D. Breitkreutz, J. Hornung, A. Markham, and N. E.
Fusenig, “Normal keratinization in a spontaneously immortalized aneuploid human ker-
atinocyte cell line.,” J. Cell Biol., vol. 106, pp. 761–771, Mar. 1988.
[173] K. I. E. McLuckie, M. Di Antonio, H. Zecchini, J. Xian, C. Caldas, B.-F. Krippendorff,
D. Tannahill, C. Lowe, and S. Balasubramanian, “G-Quadruplex DNA as a Molecular
Target for Induced Synthetic Lethality in Cancer Cells,” J. Am. Chem. Soc., vol. 135,
pp. 9640–9643, July 2013.
[174] J. Carvalho, J.-L. Mergny, G. F. Salgado, J. A. Queiroz, and C. Cruz, “G-quadruplex,
Friend or Foe: The Role of the G-quartet in Anticancer Strategies,” Trends in Molecular
Medicine, vol. 26, pp. 848–861, Sept. 2020.
[175] M. Aggarwal, J. A. Sommers, R. H. Shoemaker, and R. M. Brosh, “Inhibition of helicase
activity by a small molecule impairs Werner syndrome helicase (WRN) function in the
cellular response to DNA damage or replication stress,” PNAS, vol. 108, pp. 1525–1530,
Jan. 2011.
203
Bibliography
[176] J. Zimmer, E. M. C. Tacconi, C. Folio, S. Badie, M. Porru, K. Klare, M. Tumiati,
E. Markkanen, S. Halder, A. Ryan, S. P. Jackson, K. Ramadan, S. G. Kuznetsov, A. Biroc-
cio, J. E. Sale, and M. Tarsounas, “Targeting BRCA1 and BRCA2 Deficiencies with
G-Quadruplex-Interacting Compounds,” Mol. Cell, vol. 61, pp. 449–460, Feb. 2016.
[177] H. Xu, M. Di Antonio, and S. McKinney, et al, “CX-5461 is a DNA G-quadruplex stabi-
lizer with selective lethality in BRCA1/2 deficient tumours,” Nat. Commun., vol. 8, Feb.
2017.
[178] K. G. Zyner, D. S. Mulhearn, S. Adhikari, S. Mart´ınez Cuesta, M. Di Antonio, N. Er-
ard, G. J. Hannon, D. Tannahill, and S. Balasubramanian, “Genetic interactions of G-
quadruplexes in humans,” eLife, vol. 8, p. e46793, July 2019.
[179] S. Mu¨ller, D. A. Sanders, M. D. Antonio, S. Matsis, J.-F. Riou, R. Rodriguez, and
S. Balasubramanian, “Pyridostatin analogues promote telomere dysfunction and long-
term growth inhibition in human cancer cells,” Org. Biomol. Chem., vol. 10, pp. 6537–
6546, July 2012.
[180] J. E. Rosenberg, R. M. Bambury, E. M. Van Allen, H. A. Drabkin, P. N. Lara, A. L.
Harzstark, N. Wagle, R. A. Figlin, G. W. Smith, L. A. Garraway, T. Choueiri, F. Er-
landsson, and D. A. Laber, “A phase II trial of AS1411 (a novel nucleolin-targeted DNA
aptamer) in metastatic renal cell carcinoma,” Invest New Drugs, vol. 32, pp. 178–187,
Feb. 2014.
[181] P. J. Bates, E. M. Reyes-Reyes, M. T. Malik, E. M. Murphy, M. G. O’Toole, and
J. O. Trent, “G-quadruplex oligonucleotide AS1411 as a cancer-targeting agent: Uses
and mechanisms,” Biochim. Biophys. Acta BBA - Gen. Subj., vol. 1861, pp. 1414–1428,
May 2017.
[182] R. E. Thurman, E. Rynes, R. Humbert, J. Vierstra, M. T. Maurano, E. Haugen, N. C.
Sheffield, A. B. Stergachis, H. Wang, B. Vernot, K. Garg, S. John, R. Sandstrom,
D. Bates, L. Boatman, T. K. Canfield, M. Diegel, D. Dunn, A. K. Ebersol, T. Frum,
E. Giste, A. K. Johnson, E. M. Johnson, T. Kutyavin, B. Lajoie, B.-K. Lee, K. Lee,
D. London, D. Lotakis, S. Neph, F. Neri, E. D. Nguyen, H. Qu, A. P. Reynolds, V. Roach,
A. Safi, M. E. Sanchez, A. Sanyal, A. Shafer, J. M. Simon, L. Song, S. Vong, M. Weaver,
Y. Yan, Z. Zhang, Z. Zhang, B. Lenhard, M. Tewari, M. O. Dorschner, R. S. Hansen, P. A.
Navas, G. Stamatoyannopoulos, V. R. Iyer, J. D. Lieb, S. R. Sunyaev, J. M. Akey, P. J.
Sabo, R. Kaul, T. S. Furey, J. Dekker, G. E. Crawford, and J. A. Stamatoyannopoulos,
“The accessible chromatin landscape of the human genome,” Nature, vol. 489, pp. 75–82,
Sept. 2012.
204
Bibliography
[183] L. Li, P. Williams, W. Ren, M. Y. Wang, Z. Gao, W. Miao, M. Huang, J. Song, and
Y. Wang, “YY1 interacts with guanine quadruplexes to regulate DNA looping and gene
expression,” Nat. Chem. Biol., pp. 1–8, Nov. 2020.
[184] A. S. Weintraub, C. H. Li, A. V. Zamudio, A. A. Sigova, N. M. Hannett, D. S. Day, B. J.
Abraham, M. A. Cohen, B. Nabet, D. L. Buckley, Y. E. Guo, D. Hnisz, R. Jaenisch, J. E.
Bradner, N. S. Gray, and R. A. Young, “YY1 Is a Structural Regulator of Enhancer-
Promoter Loops,” Cell, vol. 171, pp. 1573–1588.e28, Dec. 2017.
[185] P. Tikhonova, I. Pavlova, E. Isaakova, V. Tsvetkov, A. Bogomazova, T. Vedekhina, A. V.
Luzhin, R. Sultanov, V. Severov, K. Klimina, O. L. Kantidze, G. Pozmogova, M. La-
garkova, and A. Varizhuk, “DNA G-Quadruplexes Contribute to CTCF Recruitment,”
Int. J. Mol. Sci., vol. 22, p. 7090, Jan. 2021.
[186] Y. Xia, K.-w. Zheng, Y.-d. He, H.-h. Liu, C.-j. Wen, Y.-h. Hao, and Z. Tan, “Transmis-
sion of dynamic supercoiling in linear and multi-way branched DNAs and its regulation
revealed by a fluorescent G-quadruplex torsion sensor,” Nucleic Acids Research, vol. 46,
pp. 7418–7424, Aug. 2018.
[187] D. Sun and L. H. Hurley, “The Importance of Negative Superhelicity in Inducing the
Formation of G-Quadruplex and i-Motif Structures in the c-Myc Promoter: Implications
for Drug Targeting and Control of Gene Expression,” J. Med. Chem., vol. 52, pp. 2863–
2874, May 2009.
[188] D. A. T. Sekibo and K. R. Fox, “The effects of DNA supercoiling on G-quadruplex
formation,” Nucleic Acids Res, vol. 45, pp. 12069–12079, Dec. 2017.
[189] M. Zeraati, D. B. Langley, P. Schofield, A. L. Moye, R. Rouet, W. E. Hughes, T. M.
Bryan, M. E. Dinger, and D. Christ, “I-motif DNA structures are formed in the nuclei of
human cells,” Nat. Chem., vol. 10, pp. 631–637, June 2018.
[190] P. A. Ginno, P. L. Lott, H. C. Christensen, I. Korf, and F. Che´din, “R-Loop Formation Is
a Distinctive Characteristic of Unmethylated Human CpG Island Promoters,” Molecular
Cell, vol. 45, pp. 814–825, Mar. 2012.
[191] L. Chen, J.-Y. Chen, X. Zhang, Y. Gu, R. Xiao, C. Shao, P. Tang, H. Qian, D. Luo, H. Li,
Y. Zhou, D.-E. Zhang, and X.-D. Fu, “R-ChIP Using Inactive RNase H Reveals Dynamic
Coupling of R-loops with Transcriptional Pausing at Gene Promoters,” Molecular Cell,
vol. 68, pp. 745–757.e5, Nov. 2017.
[192] C.-Y. Lee, C. McNerney, K. Ma, W. Zhao, A. Wang, and S. Myong, “R-loop induced
G-quadruplex in non-template promotes transcription by successive R-loop formation,”
Nat Commun, vol. 11, p. 3392, July 2020.
205
Bibliography
[193] V. Shukla, D. Samaniego-Castruita, Z. Dong, E. Gonza´lez-Avalos, Q. Yan, K. Sarma, and
A. Rao, “TET deficiency perturbs mature B cell homeostasis and promotes oncogenesis
associated with accumulation of G-quadruplex and R-loop structures,” Nat Immunol,
vol. 23, pp. 99–108, Jan. 2022.
[194] S. Dhakal, Z. Yu, R. Konik, Y. Cui, D. Koirala, and H. Mao, “G-Quadruplex and i-Motif
Are Mutually Exclusive in ILPR Double-Stranded DNA,” Biophys J, vol. 102, pp. 2575–
2584, June 2012.
[195] A. Henderson, Y. Wu, Y. C. Huang, E. A. Chavez, J. Platt, F. B. Johnson, R. M. Brosh,
D. Sen, and P. M. Lansdorp, “Detection of G-quadruplex DNA in mammalian cells,”
Nucleic Acids Res., vol. 42, pp. 860–869, Jan. 2014.
[196] S. Artusi, R. Perrone, S. Lago, P. Raffa, E. Di Iorio, G. Palu`, and S. N. Richter, “Visu-
alization of DNA G-quadruplexes in herpes simplex virus 1-infected cells,” Nucleic Acids
Res, vol. 44, pp. 10343–10353, Dec. 2016.
[197] R. F. Hoffmann, Y. M. Moshkin, S. Mouton, N. A. Grzeschik, R. D. Kalicharan,
J. Kuipers, A. H. Wolters, K. Nishida, A. V. Romashchenko, J. Postberg, H. Lipps,
E. Berezikov, O. C. Sibon, B. N. Giepmans, and P. M. Lansdorp, “Guanine quadruplex
structures localize to heterochromatin,” Nucleic Acids Research, vol. 44, pp. 152–163,
Jan. 2016.
[198] R. Hanna, A. Flamier, A. Barabino, and G. Bernier, “G-quadruplexes originating from
evolutionary conserved L1 elements interfere with neuronal gene expression in Alzheimer’s
disease,” Nat Commun, vol. 12, p. 1828, Mar. 2021.
[199] H. G. Kazemier, K. Paeschke, and P. M. Lansdorp, “Guanine quadruplex monoclonal
antibody 1H6 cross-reacts with restrained thymidine-rich single stranded DNA,” Nucleic
Acids Research, vol. 45, pp. 5913–5919, June 2017.
[200] A. Shivalingam, M. A. Izquierdo, A. L. Marois, A. Vysˇniauskas, K. Suhling, M. K.
Kuimova, and R. Vilar, “The interactions between a small molecule and G-quadruplexes
are visualized by fluorescence lifetime imaging microscopy,” Nat. Commun., vol. 6,
p. 8178, Sept. 2015.
[201] P. A. Summers, B. W. Lewis, J. Gonzalez-Garcia, R. M. Porreca, A. H. M. Lim, P. Cadinu,
N. Martin-Pintado, D. J. Mann, J. B. Edel, J. B. Vannier, M. K. Kuimova, and R. Vilar,
“Visualising G-quadruplex DNA dynamics in live cells by fluorescence lifetime imaging
microscopy,” Nat. Commun., vol. 12, p. 162, Jan. 2021.
[202] T.-Y. Tseng, W.-W. Chen, I.-T. Chu, C.-L. Wang, C.-C. Chang, M.-C. Lin, P.-J. Lou, and
T.-C. Chang, “The G-quadruplex fluorescent probe 3,6-bis(1-methyl-2-vinyl-pyridinium)
206
Bibliography
carbazole diiodide as a biosensor for human cancers,” Sci. Rep., vol. 8, p. 16082, Oct.
2018.
[203] T.-Y. Tseng, I.-T. Chu, S.-J. Lin, J. Li, and T.-C. Chang, “Binding of Small Molecules
to G-quadruplex DNA in Cells Revealed by Fluorescence Lifetime Imaging Microscopy
of o-BMVC Foci,” Molecules, vol. 24, p. 35, Jan. 2019.
[204] F. Doria, M. Nadai, M. Zuffo, R. Perrone, M. Freccero, and S. N. Richter, “A red-
NIR fluorescent dye detecting nuclear DNA G-quadruplexes: In vitro analysis and cell
imaging,” Chem. Commun., vol. 53, no. 14, pp. 2268–2271, 2017.
[205] S. Zhang, H. Sun, L. Wang, Y. Liu, H. Chen, Q. Li, A. Guan, M. Liu, and Y. Tang, “Real-
time monitoring of DNA G-quadruplexes in living cells with a small-molecule fluorescent
probe,” Nucleic Acids Res, vol. 46, pp. 7522–7532, Sept. 2018.
[206] A. Laguerre, J. M. Y. Wong, and D. Monchaud, “Direct visualization of both DNA and
RNA quadruplexes in human cells via an uncommon spectroscopic method,” Sci. Rep.,
vol. 6, Oct. 2016.
[207] S. Y. Yang, S. Amor, A. Laguerre, J. M. Wong, and D. Monchaud, “Real-time and
quantitative fluorescent live-cell imaging with quadruplex-specific red-edge probe (G4-
REP),” Biochim. Biophys. Acta BBA - Gen. Subj., vol. 1861, pp. 1312–1320, May 2017.
[208] X.-C. Chen, S.-B. Chen, J. Dai, J.-H. Yuan, T.-M. Ou, Z.-S. Huang, and J.-H. Tan,
“Tracking the Dynamic Folding and Unfolding of RNA G-Quadruplexes in Live Cells,”
Angew. Chem. Int. Ed., vol. 57, pp. 4702–4706, Apr. 2018.
[209] S. Zhang, H. Sun, H. Chen, Q. Li, A. Guan, L. Wang, Y. Shi, S. Xu, M. Liu, and Y. Tang,
“Direct visualization of nucleolar G-quadruplexes in live cells by using a fluorescent light-
up probe,” Biochimica et Biophysica Acta (BBA) - General Subjects, vol. 1862, pp. 1101–
1106, May 2018.
[210] M.-Q. Wang, L.-X. Gao, Y.-F. Yang, X.-N. Xiong, Z.-Y. Zheng, S. Li, Y. Wu, and L.-
L. Ma, “A triphenylamine derivative as a naked-eye and light-up fluorescent probe for
G-quadruplex DNA,” Tetrahedron Letters, vol. 57, pp. 5042–5046, Nov. 2016.
[211] H.-L. Bao, T. Ishizuka, T. Sakamoto, K. Fujimoto, T. Uechi, N. Kenmochi, and Y. Xu,
“Characterization of human telomere RNA G-quadruplex structures in vitro and in living
cells using 19F NMR spectroscopy,” Nucleic Acids Research, vol. 45, pp. 5501–5511, May
2017.
[212] H.-L. Bao and Y. Xu, “Investigation of higher-order RNA G-quadruplex structures in
vitro and in living cells by 19F NMR spectroscopy,” Nat. Protoc., vol. 13, pp. 652–665,
Apr. 2018.
207
Bibliography
[213] M. Di Antonio, A. Ponjavic, A. Radzevicˇius, R. T. Ranasinghe, M. Catalano, X. Zhang,
J. Shen, L.-M. Needham, S. F. Lee, D. Klenerman, and S. Balasubramanian, “Single-
molecule visualization of DNA G-quadruplex formation in live cells,” Nat. Chem., vol. 12,
pp. 832–837, Sept. 2020.
[214] G.-W. Li and X. S. Xie, “Central dogma at the single-molecule level in living cells,”
Nature, vol. 475, pp. 308–315, July 2011.
[215] D. Dulin, J. Lipfert, M. C. Moolman, and N. H. Dekker, “Studying genomic processes
at the single-molecule level: Introducing the tools and applications,” Nat. Rev. Genet.,
vol. 14, pp. 9–22, Jan. 2013.
[216] M. D. Wang, M. J. Schnitzer, H. Yin, R. Landick, J. Gelles, and S. M. Block, “Force and
Velocity Measured for Single Molecules of RNA Polymerase,” Science, vol. 282, pp. 902–
907, Oct. 1998.
[217] E. A. Abbondanzieri, W. J. Greenleaf, J. W. Shaevitz, R. Landick, and S. M. Block,
“Direct observation of base-pair stepping by RNA polymerase,” Nature, vol. 438, pp. 460–
465, Nov. 2005.
[218] J.-D. Wen, L. Lancaster, C. Hodges, A.-C. Zeri, S. H. Yoshimura, H. F. Noller, C. Busta-
mante, and I. Tinoco, “Following translation by single ribosomes one codon at a time,”
Nature, vol. 452, pp. 598–603, Apr. 2008.
[219] J. Yu, J. Xiao, X. Ren, K. Lao, and X. S. Xie, “Probing Gene Expression in Live Cells,
One Protein Molecule at a Time,” Science, vol. 311, pp. 1600–1603, Mar. 2006.
[220] E. Bertrand, P. Chartrand, M. Schaefer, S. M. Shenoy, R. H. Singer, and R. M. Long,
“Localization of ASH1 mRNA Particles in Living Yeast,” Mol. Cell, vol. 2, pp. 437–445,
Oct. 1998.
[221] I. Golding, J. Paulsson, S. M. Zawilski, and E. C. Cox, “Real-Time Kinetics of Gene
Activity in Individual Bacteria,” Cell, vol. 123, pp. 1025–1036, Dec. 2005.
[222] A. N. Kapanidis, E. Margeat, S. O. Ho, E. Kortkhonjia, S. Weiss, and R. H. Ebright,
“Initial Transcription by RNA Polymerase Proceeds Through a DNA-Scrunching Mech-
anism,” Science, vol. 314, pp. 1144–1147, Nov. 2006.
[223] T. Ha, “Single-molecule fluorescence methods for the study of nucleic acids,” Curr. Opin.
Struct. Biol., vol. 11, pp. 287–292, June 2001.
[224] S. Weiss, “Fluorescence Spectroscopy of Single Biomolecules,” Science, vol. 283, pp. 1676–
1683, Mar. 1999.
208
Bibliography
[225] J. Eid, A. Fehr, and J. Gray, et al, “Real-time DNA sequencing from single polymerase
molecules,” Science, vol. 323, pp. 133–138, Jan. 2009.
[226] M. Jain, H. E. Olsen, B. Paten, and M. Akeson, “The Oxford Nanopore MinION: Delivery
of nanopore sequencing to the genomics community,” Genome Biol., vol. 17, Nov. 2016.
[227] X.-Y. Zhang, E.-H. Cao, Y. Zhang, C. Zhou, and C. Bai, “K+ and Na+-Induced Self-
Assembly of Telomeric Oligonucleotide d(TTAGGG)n,” J. Biomol. Struct. Dyn., vol. 20,
pp. 693–701, Apr. 2003.
[228] R. Tippana, W. Xiao, and S. Myong, “G-quadruplex conformation and dynamics are
determined by loop length and sequence,” Nucleic Acids Res., vol. 42, pp. 8106–8114,
July 2014.
[229] R. Tippana, H. Hwang, P. L. Opresko, V. A. Bohr, and S. Myong, “Single-molecule imag-
ing reveals a common mechanism shared by G-quadruplex–resolving helicases,” PNAS,
vol. 113, pp. 8448–8453, July 2016.
[230] R. Tatavosian, H. N. Duc, T. N. Huynh, D. Fang, B. Schmitt, X. Shi, Y. Deng, C. Phiel,
T. Yao, Z. Zhang, H. Wang, and X. Ren, “Live-cell single-molecule dynamics of PcG
proteins imposed by the DIPG H3.3K27M mutation,” Nat. Commun., vol. 9, p. 2080,
May 2018.
[231] I. Izeddin, V. Re´camier, and L. Bosanac, et al, “Single-molecule tracking in live cells
reveals distinct target-search strategies of transcription factors in the nucleus,” eLife
Sciences, vol. 3, p. e02230, June 2014.
[232] M. Georgieva, D. I. Cattoni, J.-B. Fiche, T. Mutin, D. Chamousset, and M. Nollmann,
“Nanometer resolved single-molecule colocalization of nuclear factors by two-color super
resolution microscopy imaging,” Methods, vol. 105, pp. 44–55, Aug. 2016.
[233] J. D. Larson, M. L. Rodgers, and A. A. Hoskins, “Visualizing Cellular Machines with
Colocalization Single Molecule Microscopy,” Chem. Soc. Rev., vol. 43, pp. 1189–1200,
Feb. 2014.
[234] J. C. Schmidt, A. J. Zaug, and T. R. Cech, “Live Cell Imaging Reveals the Dynamics of
Telomerase Recruitment to Telomeres,” Cell, vol. 166, pp. 1188–1197.e9, Aug. 2016.
[235] M. Tokunaga, N. Imamoto, and K. Sakata-Sogawa, “Highly inclined thin illumination
enables clear single-molecule imaging in cells,” Nat. Methods, vol. 5, pp. 159–161, Feb.
2008.
209
Bibliography
[236] J. G. Ritter, R. Veith, A. Veenendaal, J. P. Siebrasse, and U. Kubitscheck, “Light Sheet
Microscopy for Single Molecule Tracking in Living Tissue,” PLOS ONE, vol. 5, p. e11639,
July 2010.
[237] K. M. Dean, P. Roudot, E. S. Welf, T. Pohlkamp, G. Garrelts, J. Herz, and R. Fiolka,
“Imaging subcellular dynamics with fast and light-efficient volumetrically parallelized
microscopy,” Optica, OPTICA, vol. 4, pp. 263–271, Feb. 2017.
[238] A. Ponjavic, Y. Ye, E. Laue, S. F. Lee, and D. Klenerman, “Sensitive light-sheet mi-
croscopy in multiwell plates using an AFM cantilever,” Biomed. Opt. Express, BOE,
vol. 9, pp. 5863–5880, Dec. 2018.
[239] A.-Y. Guo, Y.-M. Zhang, L. Wang, D. Bai, Y.-P. Xu, and W.-Q. Wu, “Single-Molecule
Imaging in Living Plant Cells: A Methodological Review,” Int. J. Mol. Sci., vol. 22,
p. 5071, Jan. 2021.
[240] M. J. Rust, M. Bates, and X. Zhuang, “Sub-diffraction-limit imaging by stochastic optical
reconstruction microscopy (STORM),” Nat Methods, vol. 3, pp. 793–796, Oct. 2006.
[241] B. Huang, W. Wang, M. Bates, and X. Zhuang, “Three-Dimensional Super-Resolution
Imaging by Stochastic Optical Reconstruction Microscopy,” Science, Feb. 2008.
[242] M. F. Juette, T. J. Gould, M. D. Lessard, M. J. Mlodzianoski, B. S. Nagpure, B. T.
Bennett, S. T. Hess, and J. Bewersdorf, “Three-dimensional sub–100 nm resolution fluo-
rescence microscopy of thick samples,” Nat Methods, vol. 5, pp. 527–529, June 2008.
[243] A. R. Carr, A. Ponjavic, S. Basu, J. McColl, A. M. Santos, S. Davis, E. D. Laue, D. Klen-
erman, and S. F. Lee, “Three-Dimensional Super-Resolution in Eukaryotic Cells Using
the Double-Helix Point Spread Function,” Biophysical Journal, vol. 112, pp. 1444–1454,
Apr. 2017.
[244] S. R. P. Pavani, M. A. Thompson, J. S. Biteen, S. J. Lord, N. Liu, R. J. Twieg, R. Piestun,
and W. E. Moerner, “Three-dimensional, single-molecule fluorescence imaging beyond the
diffraction limit by using a double-helix point spread function,” PNAS, vol. 106, pp. 2995–
2999, Mar. 2009.
[245] A. Bolzer, G. Kreth, I. Solovei, D. Koehler, K. Saracoglu, C. Fauth, S. Mu¨ller, R. Eils,
C. Cremer, M. R. Speicher, and T. Cremer, “Three-Dimensional Maps of All Chromo-
somes in Human Male Fibroblast Nuclei and Prometaphase Rosettes,” PLOS Biology,
vol. 3, p. e157, Apr. 2005.
[246] L. A. Parada and T. Misteli, “Chromosome positioning in the interphase nucleus,” Trends
in Cell Biology, vol. 12, pp. 425–432, Sept. 2002.
210
Bibliography
[247] I. Solovei, K. Thanisch, and Y. Feodorova, “How to rule the nucleus: Divide et impera,”
Current Opinion in Cell Biology, vol. 40, pp. 47–59, June 2016.
[248] H. Tjong, W. Li, R. Kalhor, C. Dai, S. Hao, K. Gong, Y. Zhou, H. Li, X. J. Zhou,
M. A. L. Gros, C. A. Larabell, L. Chen, and F. Alber, “Population-based 3D genome
structure analysis reveals driving forces in spatial genome organization,” PNAS, vol. 113,
pp. E1663–E1672, Mar. 2016.
[249] E. S. Dog˘an and C. Liu, “Three-dimensional chromatin packing and positioning of plant
genomes,” Nat. Plants, vol. 4, pp. 521–529, Aug. 2018.
[250] E. Heitz, “Das Heterochromatin der Moose,” Jahrb. Wiss. Bot., vol. 69, pp. 762–818,
1928.
[251] M. Simonis, P. Klous, E. Splinter, Y. Moshkin, R. Willemsen, E. de Wit, B. van Steensel,
and W. de Laat, “Nuclear organization of active and inactive chromatin domains uncov-
ered by chromosome conformation capture–on-chip (4C),” Nat Genet, vol. 38, pp. 1348–
1354, Nov. 2006.
[252] T. Cremer, M. Cremer, B. Hu¨bner, H. Strickfaden, D. Smeets, J. Popken, M. Sterr,
Y. Markaki, K. Rippe, and C. Cremer, “The 4D nucleome: Evidence for a dynamic
nuclear landscape based on co-aligned active and inactive nuclear compartments,” FEBS
Lett., vol. 589, no. 20PartA, pp. 2931–2943, 2015.
[253] A. Kurz, S. Lampel, R. M. Zirbel, T. Cremer, and P. Lichter, “Active and Inactive
Genes Locafize Preferentially in the Periphery of Chromosome Territories,” J. Cell Biol.,
vol. 135, p. 11, 1996.
[254] N. L. Mahy, P. E. Perry, S. Gilchrist, R. A. Baldock, and W. A. Bickmore, “Spatial orga-
nization of active and inactive genes and noncoding DNA within chromosome territories,”
Journal of Cell Biology, vol. 157, pp. 579–589, May 2002.
[255] A. J. Bannister and T. Kouzarides, “Regulation of chromatin by histone modifications,”
Cell Res, vol. 21, pp. 381–395, Mar. 2011.
[256] J. Dekker, K. Rippe, M. Dekker, and N. Kleckner, “Capturing Chromosome Conforma-
tion,” Science, vol. 295, pp. 1306–1311, Feb. 2002.
[257] E. Lieberman-Aiden, N. L. van Berkum, L. Williams, M. Imakaev, T. Ragoczy, A. Telling,
I. Amit, B. R. Lajoie, P. J. Sabo, M. O. Dorschner, R. Sandstrom, B. Bernstein, M. A.
Bender, M. Groudine, A. Gnirke, J. Stamatoyannopoulos, L. A. Mirny, E. S. Lander,
and J. Dekker, “Comprehensive Mapping of Long-Range Interactions Reveals Folding
Principles of the Human Genome,” Science, vol. 326, pp. 289–293, Oct. 2009.
211
Bibliography
[258] B. van Steensel and A. S. Belmont, “Lamina-Associated Domains: Links with Chromo-
some Architecture, Heterochromatin, and Gene Repression,” Cell, vol. 169, pp. 780–791,
May 2017.
[259] J. R. Dixon, S. Selvaraj, F. Yue, A. Kim, Y. Li, Y. Shen, M. Hu, J. S. Liu, and B. Ren,
“Topological domains in mammalian genomes identified by analysis of chromatin inter-
actions,” Nature, vol. 485, pp. 376–380, Apr. 2012.
[260] C. Hou, L. Li, Z. S. Qin, and V. G. Corces, “Gene Density, Transcription, and Insulators
Contribute to the Partition of the Drosophila Genome into Physical Domains,” Molecular
Cell, vol. 48, pp. 471–484, Nov. 2012.
[261] E. P. Nora, B. R. Lajoie, E. G. Schulz, L. Giorgetti, I. Okamoto, N. Servant, T. Piolot,
N. L. van Berkum, J. Meisig, J. Sedat, J. Gribnau, E. Barillot, N. Blu¨thgen, J. Dekker,
and E. Heard, “Spatial partitioning of the regulatory landscape of the X-inactivation
centre,” Nature, vol. 485, pp. 381–385, May 2012.
[262] T. Sexton, E. Yaffe, E. Kenigsberg, F. Bantignies, B. Leblanc, M. Hoichman, H. Par-
rinello, A. Tanay, and G. Cavalli, “Three-Dimensional Folding and Functional Organiza-
tion Principles of the Drosophila Genome,” Cell, vol. 148, pp. 458–472, Feb. 2012.
[263] S. S. P. Rao, M. H. Huntley, N. C. Durand, E. K. Stamenova, I. D. Bochkov, J. T.
Robinson, A. L. Sanborn, I. Machol, A. D. Omer, E. S. Lander, and E. L. Aiden, “A
3D Map of the Human Genome at Kilobase Resolution Reveals Principles of Chromatin
Looping,” Cell, vol. 159, pp. 1665–1680, Dec. 2014.
[264] N. Krietenstein, S. Abraham, S. V. Venev, N. Abdennur, J. Gibcus, T.-H. S. Hsieh, K. M.
Parsi, L. Yang, R. Maehr, L. A. Mirny, J. Dekker, and O. J. Rando, “Ultrastructural De-
tails of Mammalian Chromosome Architecture,” Molecular Cell, vol. 78, pp. 554–565.e7,
May 2020.
[265] T.-H. S. Hsieh, G. Fudenberg, A. Goloborodko, and O. J. Rando, “Micro-C XL: Assaying
chromosome conformation from the nucleosome to the entire genome,” Nat Methods,
vol. 13, pp. 1009–1011, Dec. 2016.
[266] S. A. Grigoryev, G. Arya, S. Correll, C. L. Woodcock, and T. Schlick, “Evidence for het-
eromorphic chromatin fibers from analysis of nucleosome interactions,” PNAS, vol. 106,
pp. 13317–13322, Aug. 2009.
[267] B. Dorigo, T. Schalch, A. Kulangara, S. Duda, R. R. Schroeder, and T. J. Richmond, “Nu-
cleosome Arrays Reveal the Two-Start Organization of the Chromatin Fiber,” Science,
vol. 306, pp. 1571–1573, Nov. 2004.
212
Bibliography
[268] H. D. Ou, S. Phan, T. J. Deerinck, A. Thor, M. H. Ellisman, and C. C. O’Shea,
“ChromEMT: Visualizing 3D chromatin structure and compaction in interphase and mi-
totic cells,” Science, vol. 357, July 2017.
[269] B. Bintu, L. J. Mateo, J.-H. Su, N. A. Sinnott-Armstrong, M. Parker, S. Kinrot, K. Ya-
maya, A. N. Boettiger, and X. Zhuang, “Super-resolution chromatin tracing reveals do-
mains and cooperative interactions in single cells,” Science, vol. 362, p. eaau1783, Oct.
2018.
[270] A. M. Cardozo Gizzi, D. I. Cattoni, J.-B. Fiche, S. M. Espinola, J. Gurgo, O. Messina,
C. Houbron, Y. Ogiyama, G. L. Papadopoulos, G. Cavalli, M. Lagha, and M. Nollmann,
“Microscopy-Based Chromosome Conformation Capture Enables Simultaneous Visual-
ization of Genome Organization and Transcription in Intact Organisms,” Molecular Cell,
vol. 74, pp. 212–222.e5, Apr. 2019.
[271] L. J. Mateo, S. E. Murphy, A. Hafner, I. S. Cinquini, C. A. Walker, and A. N. Boettiger,
“Visualizing DNA folding and RNA in embryos at single-cell resolution,” Nature, vol. 568,
p. 49, Apr. 2019.
[272] S. Kim, N.-K. Yu, and B.-K. Kaang, “CTCF as a multifunctional protein in genome
regulation and gene expression,” Exp Mol Med, vol. 47, pp. e166–e166, June 2015.
[273] V. Narendra, M. Bulajic´, J. Dekker, E. O. Mazzoni, and D. Reinberg, “CTCF-mediated
topological boundaries during development foster appropriate gene regulation,” Genes
Dev, vol. 30, pp. 2657–2662, Dec. 2016.
[274] D. G. Lupia´n˜ez, K. Kraft, V. Heinrich, P. Krawitz, F. Brancati, E. Klopocki, D. Horn,
H. Kayserili, J. M. Opitz, R. Laxova, F. Santos-Simarro, B. Gilbert-Dussardier, L. Wit-
tler, M. Borschiwer, S. A. Haas, M. Osterwalder, M. Franke, B. Timmermann, J. Hecht,
M. Spielmann, A. Visel, and S. Mundlos, “Disruptions of Topological Chromatin Domains
Cause Pathogenic Rewiring of Gene-Enhancer Interactions,” Cell, vol. 161, pp. 1012–1025,
May 2015.
[275] A. L. Sanborn, S. S. P. Rao, S.-C. Huang, N. C. Durand, M. H. Huntley, A. I. Jewett, I. D.
Bochkov, D. Chinnappan, A. Cutkosky, J. Li, K. P. Geeting, A. Gnirke, A. Melnikov,
D. McKenna, E. K. Stamenova, E. S. Lander, and E. L. Aiden, “Chromatin extrusion
explains key features of loop and domain formation in wild-type and engineered genomes,”
PNAS, vol. 112, pp. E6456–E6465, Nov. 2015.
[276] I. F. Davidson, B. Bauer, D. Goetz, W. Tang, G. Wutz, and J.-M. Peters, “DNA loop
extrusion by human cohesin,” Science, vol. 366, pp. 1338–1345, Dec. 2019.
213
Bibliography
[277] I. M. Flyamer, J. Gassler, M. Imakaev, H. B. Branda˜o, S. V. Ulianov, N. Abdennur, S. V.
Razin, L. A. Mirny, and K. Tachibana-Konwalski, “Single-nucleus Hi-C reveals unique
chromatin reorganization at oocyte-to-zygote transition,” Nature, vol. 544, pp. 110–114,
Apr. 2017.
[278] T. J. Stevens, D. Lando, S. Basu, L. P. Atkinson, Y. Cao, S. F. Lee, M. Leeb, K. J.
Wohlfahrt, W. Boucher, A. O’Shaughnessy-Kirwan, J. Cramard, A. J. Faure, M. Ralser,
E. Blanco, L. Morey, M. Sanso´, M. G. S. Palayret, B. Lehner, L. Di Croce, A. Wutz,
B. Hendrich, D. Klenerman, and E. D. Laue, “3D structures of individual mammalian
genomes studied by single-cell Hi-C,” Nature, vol. 544, pp. 59–64, Mar. 2017.
[279] L. Tan, D. Xing, C.-H. Chang, H. Li, and X. S. Xie, “Three-dimensional genome structures
of single diploid human cells,” Science, vol. 361, pp. 924–928, Aug. 2018.
[280] N. Naumova, M. Imakaev, G. Fudenberg, Y. Zhan, B. R. Lajoie, L. A. Mirny, and
J. Dekker, “Organization of the Mitotic Chromosome,” Science, vol. 342, pp. 948–953,
Nov. 2013.
[281] T. Nagano, Y. Lubling, C. Va´rnai, C. Dudley, W. Leung, Y. Baran, N. Mendelson Cohen,
S. Wingett, P. Fraser, and A. Tanay, “Cell-cycle dynamics of chromosomal organization
at single-cell resolution,” Nature, vol. 547, pp. 61–67, July 2017.
[282] B. A. Gibson, L. K. Doolittle, M. W. G. Schneider, L. E. Jensen, N. Gamarra, L. Henry,
D. W. Gerlich, S. Redding, and M. K. Rosen, “Organization of Chromatin by Intrinsic
and Regulated Phase Separation,” Cell, vol. 179, pp. 470–484.e21, Oct. 2019.
[283] S. F. Banani, H. O. Lee, A. A. Hyman, and M. K. Rosen, “Biomolecular condensates:
Organizers of cellular biochemistry,” Nat Rev Mol Cell Biol, vol. 18, pp. 285–298, May
2017.
[284] Y. Shin and C. P. Brangwynne, “Liquid phase condensation in cell physiology and dis-
ease,” Science, vol. 357, p. eaaf4382, Sept. 2017.
[285] A. R. Strom, A. V. Emelyanov, M. Mir, D. V. Fyodorov, X. Darzacq, and G. H. Karpen,
“Phase separation drives heterochromatin domain formation,” Nature, vol. 547, pp. 241–
245, July 2017.
[286] A. Boija, I. A. Klein, B. R. Sabari, A. Dall’Agnese, E. L. Coffey, A. V. Zamudio, C. H.
Li, K. Shrinivas, J. C. Manteiga, N. M. Hannett, B. J. Abraham, L. K. Afeyan, Y. E.
Guo, J. K. Rimel, C. B. Fant, J. Schuijers, T. I. Lee, D. J. Taatjes, and R. A. Young,
“Transcription Factors Activate Genes through the Phase-Separation Capacity of Their
Activation Domains,” Cell, vol. 175, pp. 1842–1855.e16, Dec. 2018.
214
Bibliography
[287] S. Chong, C. Dugast-Darzacq, Z. Liu, P. Dong, G. M. Dailey, C. Cattoglio, A. Heck-
ert, S. Banala, L. Lavis, X. Darzacq, and R. Tjian, “Imaging dynamic and selective
low-complexity domain interactions that control gene transcription,” Science, vol. 361,
p. eaar2555, July 2018.
[288] B. R. Sabari, A. Dall’Agnese, A. Boija, I. A. Klein, E. L. Coffey, K. Shrinivas, B. J.
Abraham, N. M. Hannett, A. V. Zamudio, J. C. Manteiga, C. H. Li, Y. E. Guo, D. S.
Day, J. Schuijers, E. Vasile, S. Malik, D. Hnisz, T. I. Lee, I. I. Cisse, R. G. Roeder,
P. A. Sharp, A. K. Chakraborty, and R. A. Young, “Coactivator condensation at super-
enhancers links phase separation and gene control,” Science, vol. 361, p. eaar3958, July
2018.
[289] H. Strickfaden, T. O. Tolsma, A. Sharma, D. A. Underhill, J. C. Hansen, and M. J.
Hendzel, “Condensed Chromatin Behaves like a Solid on the Mesoscale In Vitro and in
Living Cells,” Cell, vol. 183, pp. 1772–1784.e13, Dec. 2020.
[290] L. Guelen, L. Pagie, E. Brasset, W. Meuleman, M. B. Faza, W. Talhout, B. H. Eussen,
A. de Klein, L. Wessels, W. de Laat, and B. van Steensel, “Domain organization of hu-
man chromosomes revealed by mapping of nuclear lamina interactions,” Nature, vol. 453,
pp. 948–951, June 2008.
[291] D. Peric-Hupkes, W. Meuleman, L. Pagie, S. W. M. Bruggeman, I. Solovei, W. Brugman,
S. Gra¨f, P. Flicek, R. M. Kerkhoven, M. van Lohuizen, M. Reinders, L. Wessels, and B. van
Steensel, “Molecular Maps of the Reorganization of Genome-Nuclear Lamina Interactions
during Differentiation,” Molecular Cell, vol. 38, pp. 603–613, May 2010.
[292] A. S. Belmont, Y. Zhai, and A. Thilenius, “Lamin B distribution and association with pe-
ripheral chromatin revealed by optical sectioning and electron microscopy tomography.,”
Journal of Cell Biology, vol. 123, pp. 1671–1685, Dec. 1993.
[293] Y. Chen, Y. Zhang, Y. Wang, L. Zhang, E. K. Brinkman, S. A. Adam, R. Goldman,
B. van Steensel, J. Ma, and A. S. Belmont, “Mapping 3D genome organization relative
to nuclear compartments using TSA-Seq as a cytological ruler,” Journal of Cell Biology,
vol. 217, pp. 4025–4048, Aug. 2018.
[294] S. A. Quinodoz, N. Ollikainen, B. Tabak, A. Palla, J. M. Schmidt, E. Detmar, M. M.
Lai, A. A. Shishkin, P. Bhat, Y. Takei, V. Trinh, E. Aznauryan, P. Russell, C. Cheng,
M. Jovanovic, A. Chow, L. Cai, P. McDonel, M. Garber, and M. Guttman, “Higher-Order
Inter-chromosomal Hubs Shape 3D Genome Organization in the Nucleus,” Cell, vol. 174,
pp. 744–757.e24, July 2018.
[295] S. A. Quinodoz, J. W. Jachowicz, P. Bhat, N. Ollikainen, A. K. Banerjee, I. N. Goronzy,
M. R. Blanco, P. Chovanec, A. Chow, Y. Markaki, J. Thai, K. Plath, and M. Guttman,
215
Bibliography
“RNA promotes the formation of spatial compartments in the nucleus,” Cell, vol. 184,
pp. 5775–5790.e30, Nov. 2021.
[296] R. A. Beagrie, A. Scialdone, M. Schueler, D. C. A. Kraemer, M. Chotalia, S. Q. Xie,
M. Barbieri, I. de Santiago, L.-M. Lavitas, M. R. Branco, J. Fraser, J. Dostie, L. Game,
N. Dillon, P. A. W. Edwards, M. Nicodemi, and A. Pombo, “Complex multi-enhancer
contacts captured by genome architecture mapping,” Nature, vol. 543, pp. 519–524, Mar.
2017.
[297] M. Zheng, S. Z. Tian, D. Capurso, M. Kim, R. Maurya, B. Lee, E. Piecuch, L. Gong, J. J.
Zhu, Z. Li, C. H. Wong, C. Y. Ngan, P. Wang, X. Ruan, C.-L. Wei, and Y. Ruan, “Multi-
plex chromatin interactions with single-molecule precision,” Nature, vol. 566, pp. 558–562,
Feb. 2019.
[298] Z. Cai, C. Cao, L. Ji, R. Ye, D. Wang, C. Xia, S. Wang, Z. Du, N. Hu, X. Yu, J. Chen,
L. Wang, X. Yang, S. He, and Y. Xue, “RIC-seq for global in situ profiling of RNA–RNA
spatial interactions,” Nature, pp. 1–6, May 2020.
[299] J. Redolfi, Y. Zhan, C. Valdes-Quezada, M. Kryzhanovska, I. Guerreiro, V. Iesmantavi-
cius, T. Pollex, R. S. Grand, E. Mulugeta, J. Kind, G. Tiana, S. A. Smallwood, W. de
Laat, and L. Giorgetti, “DamC reveals principles of chromatin folding in vivo without
crosslinking and ligation,” Nat. Struct. Mol. Biol., vol. 26, pp. 471–480, June 2019.
[300] Y. Takei, J. Yun, S. Zheng, N. Ollikainen, N. Pierson, J. White, S. Shah, J. Thomassie,
S. Suo, C.-H. L. Eng, M. Guttman, G.-C. Yuan, and L. Cai, “Integrated spatial genomics
reveals global architecture of single nuclei,” Nature, pp. 1–7, Jan. 2021.
[301] T. Nagano, Y. Lubling, T. J. Stevens, S. Schoenfelder, E. Yaffe, W. Dean, E. D. Laue,
A. Tanay, and P. Fraser, “Single-cell Hi-C reveals cell-to-cell variability in chromosome
structure,” Nature, vol. 502, pp. 59–64, Oct. 2013.
[302] D. Lando, S. Basu, T. J. Stevens, A. Riddell, K. J. Wohlfahrt, Y. Cao, W. Boucher,
M. Leeb, L. P. Atkinson, S. F. Lee, B. Hendrich, D. Klenerman, and E. D. Laue, “Com-
bining fluorescence imaging with Hi-C to study 3D genome architecture of the same single
cell,” Nat. Protoc., vol. 13, pp. 1034–1061, May 2018.
[303] A. C. Payne, Z. D. Chiang, P. L. Reginato, S. M. Mangiameli, E. M. Murray, C.-C.
Yao, S. Markoulaki, A. S. Earl, A. S. Labade, R. Jaenisch, G. M. Church, E. S. Boyden,
J. D. Buenrostro, and F. Chen, “In situ genome sequencing resolves DNA sequence and
structure in intact biological samples,” Science, Dec. 2020.
216
Bibliography
[304] J.-H. Su, P. Zheng, S. S. Kinrot, B. Bintu, and X. Zhuang, “Genome-Scale Imaging of the
3D Organization and Transcriptional Activity of Chromatin,” Cell, vol. 182, pp. 1641–
1659.e26, Sept. 2020.
[305] L. Tan, D. Xing, N. Daley, and X. S. Xie, “Three-dimensional genome structures of single
sensory neurons in mouse visual and olfactory systems,” Nat. Struct. Mol. Biol., vol. 26,
pp. 297–307, Apr. 2019.
[306] L. Tan, W. Ma, H. Wu, Y. Zheng, D. Xing, R. Chen, X. Li, N. Daley, K. Deisseroth,
and X. S. Xie, “Changes in genome architecture and transcriptional dynamics progress
independently of sensory experience during post-natal brain development,” Cell, vol. 184,
pp. 741–758.e17, Feb. 2021.
[307] J. Lefebvre, C. Guetta, F. Poyer, F. Mahuteau-Betzer, and M.-P. Teulade-Fichou,
“Copper-Alkyne Complexation Responsible for the Nucleolar Localization of Quadru-
plex Nucleic Acid Drugs Labeled by Click Reactions,” Angew. Chem. Int. Ed., vol. 56,
pp. 11365–11369, Sept. 2017.
[308] A. F. Larsen and T. Ulven, “Efficient Synthesis of 4,7-Diamino Substituted 1,10-
Phenanthroline-2,9-dicarboxamides,” Org. Lett., vol. 13, pp. 3546–3548, July 2011.
[309] W. J. Chung, B. Heddi, F. Hamon, M.-P. Teulade-Fichou, and A. T. Phan, “Solu-
tion Structure of a G-quadruplex Bound to the Bisquinolinium Compound Phen-DC3,”
Angew. Chem., vol. 126, pp. 1017–1020, Jan. 2014.
[310] R. Rocca, C. Talarico, F. Moraca, G. Costa, I. Romeo, F. Ortuso, S. Alcaro, and A. Artese,
“Molecular recognition of a carboxy pyridostatin toward G-quadruplex structures: Why
does it prefer RNA?,” Chem. Biol. Drug Des., vol. 90, no. 5, pp. 919–925, 2017.
[311] E. Largy, F. Hamon, and M.-P. Teulade-Fichou, “Development of a high-throughput G4-
FID assay for screening and evaluation of small molecules binding quadruplex nucleic
acid structures,” Anal Bioanal Chem, vol. 400, pp. 3419–3427, July 2011.
[312] A. Piazza, J.-B. Boule´, J. Lopes, K. Mingo, E. Largy, M.-P. Teulade-Fichou, and A. Nico-
las, “Genetic instability triggered by G-quadruplex interacting Phen-DC compounds in
Saccharomyces cerevisiae,” Nucleic Acids Res., vol. 38, pp. 4337–4348, July 2010.
[313] A. Piazza, M. Adrian, F. Samazan, B. Heddi, F. Hamon, A. Serero, J. Lopes, M.-P.
Teulade-Fichou, A. T. Phan, and A. Nicolas, “Short loop length and high thermal stability
determine genomic instability induced by G-quadruplex-forming minisatellites,” EMBO
J., vol. 34, no. 12, pp. 1718–1734, 2015.
217
Bibliography
[314] B. Biswas, M. Kandpal, and P. Vivekanandan, “A G-quadruplex motif in an envelope
gene promoter regulates transcription and virion secretion in HBV genotype B,” Nucleic
Acids Research, vol. 45, pp. 11268–11280, Nov. 2017.
[315] R. Halder, J.-F. Riou, M.-P. Teulade-Fichou, T. Frickey, and J. S. Hartig, “Bisquinolinium
compounds induce quadruplex-specific transcriptome changes in HeLa S3 cell lines,” BMC
Research Notes, vol. 5, p. 138, Mar. 2012.
[316] D. Gomez, A. Gue´din, J.-L. Mergny, B. Salles, J.-F. Riou, M.-P. Teulade-Fichou, and
P. Calsou, “A G-quadruplex structure within the 5′-UTR of TRF2 mRNA represses
translation in human cells,” Nucleic Acids Res., vol. 38, pp. 7187–7198, Nov. 2010.
[317] M. J. Lista, R. P. Martins, and O. Billant, et al, “Nucleolin directly mediates Epstein-
Barr virus immune evasion through binding to G-quadruplexes of EBNA1 mRNA,” Nat.
Commun., vol. 8, p. 16043, July 2017.
[318] G. Lukinavicˇius, K. Umezawa, N. Olivier, A. Honigmann, G. Yang, T. Plass, V. Mueller,
L. Reymond, I. R. Correˆa Jr, Z.-G. Luo, C. Schultz, E. A. Lemke, P. Heppenstall,
C. Eggeling, S. Manley, and K. Johnsson, “A near-infrared fluorophore for live-cell super-
resolution microscopy of cellular proteins,” Nat. Chem., vol. 5, pp. 132–139, Feb. 2013.
[319] G. Lukinavicˇius, L. Reymond, E. D’Este, A. Masharina, F. Go¨ttfert, H. Ta, A. Gu¨ther,
M. Fournier, S. Rizzo, H. Waldmann, C. Blaukopf, C. Sommer, D. W. Gerlich, H.-D.
Arndt, S. W. Hell, and K. Johnsson, “Fluorogenic probes for live-cell imaging of the
cytoskeleton,” Nat. Methods, vol. 11, pp. 731–733, July 2014.
[320] F. Himo, T. Lovell, R. Hilgraf, V. V. Rostovtsev, L. Noodleman, K. B. Sharpless, and
V. V. Fokin, “Copper(I)-Catalyzed Synthesis of Azoles. DFT Study Predicts Unprece-
dented Reactivity and Intermediates,” J. Am. Chem. Soc., vol. 127, pp. 210–216, Jan.
2005.
[321] J.-L. Mergny and J.-C. Maurizot, “Fluorescence Resonance Energy Transfer as a Probe
for G-Quartet Formation by a Telomeric Repeat,” ChemBioChem, vol. 2, no. 2, pp. 124–
132, 2001.
[322] J. Alzeer, B. R. Vummidi, P. J. C. Roth, and N. W. Luedtke, “Guanidinium-Modified
Phthalocyanines as High-Affinity G-Quadruplex Fluorescent Probes and Transcriptional
Regulators,” Angew. Chem. Int. Ed., vol. 48, no. 49, pp. 9362–9365, 2009.
[323] J. Alzeer and N. W. Luedtke, “pH-Mediated Fluorescence and G-Quadruplex Binding of
Amido Phthalocyanines,” Biochemistry, vol. 49, pp. 4339–4348, May 2010.
218
Bibliography
[324] R. I. MacDonald, “Characteristics of self-quenching of the fluorescence of lipid-conjugated
rhodamine in membranes.,” Journal of Biological Chemistry, vol. 265, pp. 13533–13539,
Aug. 1990.
[325] T. R. Evans, “Singlet quenching mechanisms,” J. Am. Chem. Soc., vol. 93, pp. 2081–2082,
Apr. 1971.
[326] W. Bae, T.-Y. Yoon, and C. Jeong, “Direct evaluation of self-quenching behavior of
fluorophores at high concentrations using an evanescent field,” PLOS ONE, vol. 16,
p. e0247326, Feb. 2021.
[327] S. Rankin, A. P. Reszka, J. Huppert, M. Zloh, G. N. Parkinson, A. K. Todd, S. Ladame,
S. Balasubramanian, and S. Neidle, “Putative DNA Quadruplex Formation within the
Human c-kit Oncogene,” J. Am. Chem. Soc., vol. 127, pp. 10584–10589, Aug. 2005.
[328] H. Fernando, A. P. Reszka, J. Huppert, S. Ladame, S. Rankin, A. R. Venkitaraman,
S. Neidle, and S. Balasubramanian, “A Conserved Quadruplex Motif Located in a Tran-
scription Activation Site of the Human c-kit Oncogene,” Biochemistry, vol. 45, pp. 7854–
7860, June 2006.
[329] A. Fegan, P. S. Shirude, L. Ying, and S. Balasubramanian, “Ensemble and single molecule
FRET analysis of the structure and unfolding kinetics of the c-kit promoter quadru-
plexes,” Chem. Commun., vol. 46, pp. 946–948, Jan. 2010.
[330] L. Ying, J. J. Green, H. Li, D. Klenerman, and S. Balasubramanian, “Studies on the
structure and dynamics of the human telomeric G quadruplex by single-molecule fluores-
cence resonance energy transfer,” PNAS, vol. 100, pp. 14629–14634, Dec. 2003.
[331] J. J. Green, S. Ladame, L. Ying, D. Klenerman, and S. Balasubramanian, “Investigating
a Quadruplex-Ligand Interaction by Unfolding Kinetics,” J. Am. Chem. Soc., vol. 128,
pp. 9809–9812, Aug. 2006.
[332] B. Klejevskaja, A. L. B. Pyne, M. Reynolds, A. Shivalingam, R. Thorogate, B. W.
Hoogenboom, L. Ying, and R. Vilar, “Studies of G-quadruplexes formed within self-
assembled DNA mini-circles,” Chem. Commun., vol. 52, pp. 12454–12457, Oct. 2016.
[333] A. Ponjavic, J. McColl, A. R. Carr, A. M. Santos, K. Kulenkampff, A. Lippert, S. J.
Davis, D. Klenerman, and S. F. Lee, “Single-Molecule Light-Sheet Imaging of Suspended
T Cells,” Biophys. J., vol. 114, pp. 2200–2211, May 2018.
[334] M. Vorl´ıcˇkova´, I. Kejnovska´, J. Sagi, D. Rencˇiuk, K. Bedna´rˇova´, J. Motlova´, and J. Kypr,
“Circular dichroism and guanine quadruplexes,” Methods, vol. 57, pp. 64–75, May 2012.
219
Bibliography
[335] A. I. Karsisiotis, N. M. Hessari, E. Novellino, G. P. Spada, A. Randazzo, and M. Webba da
Silva, “Topological Characterization of Nucleic Acid G-Quadruplexes by UV Absorption
and Circular Dichroism,” Angew. Chem. Int. Ed., vol. 50, no. 45, pp. 10645–10648, 2011.
[336] S. Masiero, R. Trotta, S. Pieraccini, S. D. Tito, R. Perone, A. Randazzo, and G. P. Spada,
“A non-empirical chromophoric interpretation of CD spectra of DNA G-quadruplex struc-
tures,” Org. Biomol. Chem., vol. 8, pp. 2683–2692, June 2010.
[337] J. Zhou, S.-H. Tan, V. Nicolas, C. Bauvy, N.-D. Yang, J. Zhang, Y. Xue, P. Codogno, and
H.-M. Shen, “Activation of lysosomal function in the course of autophagy via mTORC1
suppression and autophagosome-lysosome fusion,” Cell Res, vol. 23, pp. 508–523, Apr.
2013.
[338] M. E. Guicciardi, M. Leist, and G. J. Gores, “Lysosomes in cell death,” Oncogene, vol. 23,
pp. 2881–2890, Apr. 2004.
[339] T. J. Etheridge, R. L. Boulineau, A. Herbert, A. T. Watson, Y. Daigaku, J. Tucker,
S. George, P. Jo¨nsson, M. Palayret, D. Lando, E. Laue, M. A. Osborne, D. Klenerman,
S. F. Lee, and A. M. Carr, “Quantification of DNA-associated proteins inside eukary-
otic cells using single-molecule localization microscopy,” Nucleic Acids Research, vol. 42,
p. e146, Oct. 2014.
[340] J. Chen, Z. Zhang, L. Li, B.-C. Chen, A. Revyakin, B. Hajj, W. Legant, M. Dahan,
T. Lionnet, E. Betzig, R. Tjian, and Z. Liu, “Single-Molecule Dynamics of Enhanceosome
Assembly in Embryonic Stem Cells,” Cell, vol. 156, pp. 1274–1285, Mar. 2014.
[341] P. D. Lawley and P. Brookes, “Further studies on the alkylation of nucleic acids and their
constituent nucleotides,” Biochem J, vol. 89, pp. 127–138, Oct. 1963.
[342] A. M. Corrigan, E. Tunnacliffe, D. Cannon, and J. R. Chubb, “A continuum model of
transcriptional bursting,” eLife, vol. 5, p. e13051, Feb. 2016.
[343] B. Mifsud, F. Tavares-Cadete, A. N. Young, R. Sugar, S. Schoenfelder, L. Ferreira, S. W.
Wingett, S. Andrews, W. Grey, P. A. Ewels, B. Herman, S. Happe, A. Higgs, E. LeProust,
G. A. Follows, P. Fraser, N. M. Luscombe, and C. S. Osborne, “Mapping long-range
promoter contacts in human cells with high-resolution capture Hi-C,” Nat Genet, vol. 47,
pp. 598–606, June 2015.
[344] F. Iborra, A. Pombo, D. Jackson, and P. Cook, “Active RNA polymerases are local-
ized within discrete transcription “factories’ in human nuclei,” Journal of Cell Science,
vol. 109, pp. 1427–1436, June 1996.
[345] A. Papantonis and P. R. Cook, “Transcription Factories: Genome Organization and Gene
Regulation,” Chem. Rev., vol. 113, pp. 8683–8705, Nov. 2013.
220
Bibliography
[346] G. T. Dempsey, J. C. Vaughan, K. H. Chen, M. Bates, and X. Zhuang, “Evaluation
of fluorophores for optimal performance in localization-based super-resolution imaging,”
Nat Methods, vol. 8, pp. 1027–1036, Dec. 2011.
[347] E. Boeggeman, B. Ramakrishnan, M. Pasek, M. Manzoni, A. Puri, K. H. Loomis, T. J.
Waybright, and P. K. Qasba, “Site Specific Conjugation of Fluoroprobes to the Remodeled
Fc N-Glycans of Monoclonal Antibodies Using Mutant Glycosyltransferases: Application
for Cell Surface Antigen Detection,” Bioconjugate Chem., vol. 20, pp. 1228–1236, June
2009.
[348] B. M. Zeglis, C. B. Davis, R. Aggeler, H. C. Kang, A. Chen, B. J. Agnew, and J. S. Lewis,
“An Enzyme-Mediated Methodology for the Site-Specific Radiolabeling of Antibodies
Based on Catalyst-Free Click Chemistry,” Bioconjug Chem, vol. 24, pp. 1057–1067, June
2013.
[349] M. T. W. Lee, A. Maruani, D. A. Richards, J. R. Baker, S. Caddick, and V. Chudasama,
“Enabling the controlled assembly of antibody conjugates with a loading of two modules
without antibody engineering,” Chem. Sci., vol. 8, pp. 2056–2060, Feb. 2017.
[350] G. V. Los, L. P. Encell, M. G. McDougall, D. D. Hartzell, N. Karassina, C. Zimprich,
M. G. Wood, R. Learish, R. F. Ohana, M. Urh, D. Simpson, J. Mendez, K. Zimmerman,
P. Otto, G. Vidugiris, J. Zhu, A. Darzins, D. H. Klaubert, R. F. Bulleit, and K. V. Wood,
“HaloTag: A Novel Protein Labeling Technology for Cell Imaging and Protein Analysis,”
ACS Chem. Biol., vol. 3, pp. 373–382, June 2008.
[351] G. Carrington, D. Tomlinson, and M. Peckham, “Exploiting nanobodies and Affimers for
superresolution imaging in light microscopy,” MBoC, vol. 30, pp. 2737–2740, Oct. 2019.
[352] G. T. Dempsey, M. Bates, W. E. Kowtoniuk, D. R. Liu, R. Y. Tsien, and X. Zhuang,
“Photoswitching Mechanism of Cyanine Dyes,” J. Am. Chem. Soc., vol. 131, pp. 18192–
18193, Dec. 2009.
[353] X. Shi, J. Lim, and T. Ha, “Acidification of the Oxygen Scavenging System in Single-
Molecule Fluorescence Studies: In Situ Sensing with a Ratiometric Dual-Emission Probe,”
Anal Chem, vol. 82, pp. 6132–6138, July 2010.
[354] A. K. Pati, O. E. Bakouri, S. Jockusch, Z. Zhou, R. B. Altman, G. A. Fitzgerald, W. B.
Asher, D. S. Terry, A. Borgia, M. D. Holsey, J. E. Batchelder, C. Abeywickrama, B. Hud-
dle, D. Rufa, J. A. Javitch, H. Ottosson, and S. C. Blanchard, “Tuning the Baird aromatic
triplet-state energy of cyclooctatetraene to maximize the self-healing mechanism in or-
ganic fluorophores,” PNAS, vol. 117, pp. 24305–24315, Sept. 2020.
221
Bibliography
[355] Y. Lin, J. J. Long, F. Huang, W. C. Duim, S. Kirschbaum, Y. Zhang, L. K. Schroeder,
A. A. Rebane, M. G. M. Velasco, A. Virrueta, D. W. Moonan, J. Jiao, S. Y. Hernandez,
Y. Zhang, and J. Bewersdorf, “Quantifying and Optimizing Single-Molecule Switching
Nanoscopy at High Speeds,” PLOS ONE, vol. 10, p. e0128135, May 2015.
[356] Q.-L. Ying, J. Wray, J. Nichols, L. Batlle-Morera, B. Doble, J. Woodgett, P. Cohen,
and A. Smith, “The ground state of embryonic stem cell self-renewal,” Nature, vol. 453,
pp. 519–523, May 2008.
[357] T. Kalkan, N. Olova, M. Roode, C. Mulas, H. J. Lee, I. Nett, H. Marks, R. Walker, H. G.
Stunnenberg, K. S. Lilley, J. Nichols, W. Reik, P. Bertone, and A. Smith, “Tracking the
embryonic stem cell transition from ground state pluripotency,” Development, vol. 144,
pp. 1221–1234, Apr. 2017.
[358] J. S. Fritz, “Ion Chromatography,” Anal. Chem., vol. 59, pp. 335A–344A, Feb. 1987.
[359] P. Kunz, K. Zinner, N. Mu¨cke, T. Bartoschik, S. Muyldermans, and J. D. Hoheisel, “The
structural basis of nanobody unfolding reversibility and thermoresistance,” Sci Rep, vol. 8,
p. 7934, May 2018.
[360] G. Biffi, M. Di Antonio, D. Tannahill, and S. Balasubramanian, “Visualization and se-
lective chemical targeting of RNA G-quadruplex structures in the cytoplasm of human
cells,” Nat. Chem., vol. 6, pp. 75–80, Jan. 2014.
[361] J.-Y. Tinevez, N. Perry, J. Schindelin, G. M. Hoopes, G. D. Reynolds, E. Laplantine,
S. Y. Bednarek, S. L. Shorte, and K. W. Eliceiri, “TrackMate: An open and extensible
platform for single-particle tracking,” Methods, vol. 115, pp. 80–90, Feb. 2017.
[362] E. Marchand, F. Spindler, and F. Chaumette, “ViSP for visual servoing: A generic
software platform with a wide class of robot control skills,” IEEE Robot. Automat. Mag.,
vol. 12, pp. 40–52, Dec. 2005.
[363] J. A. Riddick, W. B. Bunger, and T. K. Sakano, ”Organic Solvents : Physical Properties
and Methods of Purification”. New York: Wiley, fourth ed., 1986.
222
Appendix
Selected 1H and 13C NMR spectra of new compounds
PhenDC3-Cl (21)
223
Appendix
224
Appendix
4-Hydroxy-1,10-phenanthroline-2,9-dicarboxylic acid (18)
225
Appendix
PhenDC3-yne (16)
0.01.02.03.04.05.06.07.08.09.010.011.012.013.0
f1 (ppm)
2.
04
2.
06
2.
08
2.
33
1.
03
2.
16
2.
17
2.
90
2.
97
1.
03
1.
07
2.
83
1.
16
2.
00
4.
34
1.
00
1.
05
1.
00
2.
08
2.
09
1.
98
1.
16
1.
17
1.
19
1.
62
1.
80
2.
08
2.
26
2.
28
2.
30
2.
35
2.
36
2.
37
2.
49
2.
50
2.
50
2.
50
 D
M
SO
-d
6
2.
50
2.
51
2.
74
2.
75
2.
76
3.
17
3.
19
3.
51
4.
72
4.
73
7.
70
7.
98
8.
05
8.
07
8.
09
8.
20
8.
22
8.
25
8.
48
8.
51
8.
53
8.
55
8.
61
8.
63
8.
67
8.
69
8.
86
8.
89
9.
88
10
.3
0
10
.3
1
12
.1
5
54
53
13 5241
37
6
49720
27
9
39
42
35
33
26
50
29
22
34
38
46
10 36
40
47 43 45
44
7.607.707.807.908.008.108.208.308.408.508.608.708.808.90
f1 (ppm)
1.
03
1.
07
2.
83
1.
16
2.
00
4.
34
1.
00
1.
05
1.
00
7.
70
7.
98
8.
05
8.
07
8.
09
8.
20
8.
22
8.
25
8.
48
8.
51
8.
53
8.
55
8.
61
8.
63
8.
67
8.
69
8.
86
8.
89
13
37
41
6
7
9 35
42
39
34
38
10 40
36
47
226
Appendix
020406080100120140160180200220
f1 (ppm)
8.
63
14
.3
4
25
.1
4
26
.9
8
34
.2
8
38
.1
7
39
.5
2 
D
M
SO
-d
6
42
.6
0
45
.7
2
46
.0
0
71
.3
2
83
.8
1
99
.6
6
11
4.
58
11
7.
50
11
9.
22
11
9.
37
12
2.
90
12
9.
16
12
9.
94
12
9.
96
13
0.
27
13
0.
95
13
2.
83
13
3.
80
13
3.
87
13
4.
50
13
4.
75
13
5.
44
13
5.
49
13
8.
83
14
5.
69
14
5.
77
14
7.
97
15
7.
59
15
7.
94
15
8.
29
15
8.
63
16
3.
75
17
0.
22
50
52 4954
53
48
5123
30
37,41
11
28,21
4546
35
38
34
39
44
24
31
16
15
5 8 43
29
22
33
26
40
36
7 10
9 6
12 13
3
2
110112114116118120122124126128130132134136138140142144146148150
f1 (ppm)
11
1.
65
11
4.
58
11
7.
50
11
9.
22
11
9.
37
12
0.
42
12
1.
62
12
2.
90
12
5.
31
12
9.
16
12
9.
94
12
9.
96
13
0.
27
13
0.
95
13
2.
83
13
3.
80
13
3.
87
13
4.
50
13
4.
75
13
5.
44
13
5.
49
13
8.
83
14
5.
69
14
5.
77
14
7.
97
30
23
37,41
11
28,21
35
34
38
39
24
31
5 829
22
26
33
40
36
7
109 6
2
3
Note that a quartet at 116 ppm is NH4CF3CO2 trace impurity from HPLC.
227
Appendix
COSY
0.01.02.03.04.05.06.07.08.09.010.011.0
f2 (ppm)
0
1
2
3
4
5
6
7
8
9
10
11
f1
 (p
pm
)
33
26
29
22
7 6
10
41
37
38
34
40
36
9
42
39
35
47 13 54
53
43 46 52 50
49
44
45
33 26
29 22
7 6
104137 38 34
40 36 9 42 39 35
47
13
54 53
43
46
52
50 49
44 45
7.67.98.28.58.89.19.49.710.010.3
f2 (ppm)
7.5
8.0
8.5
9.0
9.5
10.0
10.5
f1
 (p
pm
)
33
26
29
22
7 6 10
41
37
38
34
40
36
9
42
39
35
47 13
3326
2922
7
610
41373834
40369
4239 35
47
13
228
Appendix
HSQC
1.02.03.04.05.06.07.08.09.010.0
f2 (ppm)
0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
f1
 (p
pm
)
33
26
29
22
7 6
10
41
37
38
34
40
36
9
42
39
35
47 13 54
53
43 46 52 50
49
44
45
33 26 2 3
7
3124 29 22 40 36
28,218
35 39 38 3430 23
9
10 6
1137,41
13
51
52
54 53
43
46
49
45 44
50
45
44
49
50
52
46
43
13
9
10
6
7
7.67.88.08.28.48.68.89.09.29.49.69.810.0
f2 (ppm)
100
105
110
115
120
125
130
135
140
f1
 (p
pm
)
29
22
7 6 10 41
37
38
34
40
36
9 42
39
35
47 13
3
7
3124 29
2240 36
28,21
8 35 39
38 34
30 23
9
10
6
1137,41
13
13
9
10
6
7
229
Appendix
HMBC
1.02.03.04.05.06.07.08.09.010.011.012.0
f2 (ppm)
10
30
50
70
90
110
130
150
170
190
f1
 (p
pm
)
20
27
33
26
29
22
7 6
10
41
37
38
34
40
36
9
42
39
35
47
13 54
53
43 46 52 50
49
44
45
48
15 16
12
5 33 26
2 3
7
31 24 29 22 40 36
28,218
35 3938 34 30 23
9 10
6 11
37,41
13
51
52
54 53
43
46
49
45 44
50
7.68.08.48.89.29.610.010.410.811.211.612.0
f2 (ppm)
120
125
130
135
140
145
150
155
160
165
170
f1
 (p
pm
)
20
27
33
26
29
22
7 6 10
41
37
38
34
40
36
9
42
39
35
47 13
48
15 16
12
5
33 26
2 3
7
31 24
29 22
40 36
28,21
8
35 39 38 34
30 23
9
10
6
1137,41
230
Appendix
PhenDC3-SiR (17)
231
Appendix
232
Appendix
COSY
0123456789101112
f2 (ppm)
0
1
2
3
4
5
6
7
8
9
10
11
12
f1
 (p
pm
)
27
20
26,33 29
22
7
56
6
10
40
36
35
39
34
38
9
41
37
59
60
47
42
52
13
63 70,0
67,0
68,0
75
74
53 43 55
46
0,73,0##
49 50 54
44
45 72,79
27 20
26,33
29 22
7 56 6
10 40 36 35 39
34 38 94137 59
60 47 4252
13 63
70,0
67,0 68,0
75 74
53
43
55 46
0,73,0##49
50
54
4445
72,79
6.67.07.47.88.28.69.09.49.810.2
f2 (ppm)
6.5
7.0
7.5
8.0
8.5
9.0
9.5
10.0
f1
 (p
pm
)
26,33 29
22
7 56 6 10 40
36
35
39
34
38
9 41
37
59
60
47
42
52 13
63
70,0 67,0
68,0
26,33
29 22
7
56
6
10
40 36 35 39
34 38
9
4137 59
6047 42
52
13 63
70,0
67,0 68,0
233
Appendix
HSQC
0123456789101112
f2 (ppm)
-10
10
30
50
70
90
110
130
150
170
f1
 (p
pm
)
27
20
26,33 29
22
7
56
6
10
40
36
35
39
34
38
9
41
37
59
60
47
42
52
13
63
70,0 67,0
68,0
75
74
53 43 55
46
0,73,0##
49 50 54 44
45
72,79
48 64
57 15 16
12
69,0 145
51 33 262 3
7
0#,7131 24 29 22 38 34
28 218
0,66 41 37 40 36 23,30 59
67,0
60 9
10 63 52 6
11 39 35
70,0
68,0
13
65
53 75 74
43 0,73,0##
46 55
50
54
4544
49
72,79
6.46.87.27.68.08.48.89.29.610.010.4
f2 (ppm)
100
105
110
115
120
125
130
135
140
145
f1
 (p
pm
)
26,33 29
22
7 56 6 10 40
36
35
39
34
38
9 41
37
59
60
47
42
52 13
63
70,0 67,0
68,0
5
51 33 26
2 3
7
0#,7131 24
2922
38 34
28 21
8 0,66
41 37 40 36
23,30
5967,0
60 9
10 63
526
11 39 35
70,0
68,0
13
234
Appendix
HMBC
0.01.02.03.04.05.06.07.08.09.010.0
f2 (ppm)
0
20
40
60
80
100
120
140
160
f1
 (p
pm
)
26,33 29
22
7
56
6
10
40
36
35
39
34
38
9
41
37
59
60
47
42
52
13
63
70,0 67,0
68,0
75
74
53 43 55
46
0,73,0##
49 50 54 44
45
72,79
48 64
57 15 16
12
69,0 145
51 33 262 3
7
0#,7131 24 29 22 38 34
28 218
0,66 41 37 40 36 23,30
59 67,0
60 9
10 63 52 6
11 39 35
70,0
68,0
13
65
53 75 74
43 0,73,0##
46 55
50
54
4544
49
72,79
6.46.87.27.68.08.48.89.29.610.010.410.8
f2 (ppm)
90
100
110
120
130
140
150
160
170
f1
 (p
pm
)
26,33 29
22
7 56 6 10 40
36
35
39
34
38
9 41
37
59
60
47
42
52 13
63
70,0 67,0
68,0
48
64
57 15
16
12
69,0 14
5
51 33 26
2 3
7
0#,7131 24
29 2238 34
28 21
8 0,66 41 37 40 36
23,3059
67,0
60 9
10 63 52
6
11 39 35
70,0
68,0
13
65
235
Appendix
LC-MS data of selected compounds
PhenDC3-yne (16)
UV chromatogram at 260 nm
Mass spectrum
236
Appendix
PhenDC3-SiR (17)
UV chromatogram at 260 nm
Mass spectrum
237
Appendix
3D STORM images of whole mESC nuclei imaged with
E12-AF647
14k localisations
9k localisations
14k localisations
Nucleus 1
Nucleus 2
Nucleus 3
238
Appendix
Nucleus 4 11k localisations
Nucleus 5 15k localisations
239
Appendix
3D STORM images of whole mESC nuclei imaged with
BG4 and SecAB-AF647
Nucleus 1
Nucleus 2
13k localisations
9k localisations
Nucleus 3 8k localisations
240
Appendix
Nucleus 4
Nucleus 5
Nucleus 6
6k localisations
5k localisations
3k localisations
241
Appendix
Cross-section of 3D Hi-C structure of na¨ıve mESC nuclei
overlapped with G4-CUT&Tag data
Nucleus 1
Nucleus 2
Nucleus 3
242
Appendix
Cross-section of 3D Hi-C structure of formative mESC
nuclei overlapped with G4-CUT&Tag data
Nucleus 1
Nucleus 2
243
Appendix
Cross-section of 3D Hi-C structure of primed mESC nu-
clei overlapped with G4-CUT&Tag data
Nucleus 1
Nucleus 2
Nucleus 3
244
Appendix
Structural differences in chromosome 5 and 8 in different
mESC states
Naı¨ve nucleus 1
Formative nucleus 1
Primed nucleus 1
245
Appendix
Naı¨ve nucleus 2
Formative nucleus 2
Primed nucleus 2
246