Transcriptional and developmental
consequences of aneuploidy during
male meiosis
Christina Ernst
Cancer Research UK Cambridge Institute
University of Cambridge
This dissertation is submitted for the degree of
Doctor of Philosophy
Darwin College September 2017

Declaration
I hereby declare that except where specific reference is made to
the work of others, the contents of this dissertation are original and
have not been submitted in whole or in part for consideration for any
other degree or qualification in this, or any other university.
This dissertation is my own work and contains nothing which is
the outcome of work done in collaboration with others, except as
specified in the text and Acknowledgements.
This dissertation does not exceed the specified word limit of 60,000
words as defined by the Degree Committee for the Faculty of Clinical
Medicine.
Christina Ernst
September 2017

Acknowledgements
I would like to thank and acknowledge all the people that contributed to this work
and supported me throughout my PhD:
First and foremost, I thank my supervisor Duncan Odom, for giving me the
opportunity to work in his lab and for his continuous mentorship and guidance
throughout my studies.
I would also like to thank my main collaborators Nils Eling, who has done the
majority of the single-cell analysis and provided me with scripts and tools to explore
the data, as well as John Marioni for his supervision on this joint project.
I would like to thank all current and previous members of the Odom lab who
have contributed to a fantastic work environment, with special thanks going to:
Claudia Kutter who has supervised me during my first two years in the lab and
introduced me to piRNAs; Sarah Aitken for her help with histological analysis and
the numerous scientific discussions throughout the years; Celia Martinez for her
support in generating single-cell RNA-Seq libraries; Frances Connor for managing
the Tc1 mouse colony for the first few months of my PhD and Tim Rayner for his
help with analysing smallRNA data sets.
I am very grateful to Cancer Research UK for providing me with the funding for my
studies and the excellent scientific support that is provided by the core facilities at CI.
I owe special thanks to Jeremy Pike from the Microscopy Core for his help with image
analysis, Richard Grenfell and Mateusz Stzelecki from the Flow Cytometry Core for
accommodating my sorting needs, Angela Mowbray from the Biological Resource
Unit for taking excellent care of my mice on a day-to-day basis, Julia Jones from the
Histopathology Core for optimising and performing sequential FISH stainings, as
well as the entire Genomics core for sequencing the hundreds of libraries that were
generated throughout my thesis.
I would also like to thank Hannah Long and Rob Klose for providing DNA methy-
lation data for the Tc1 mouse, Ramesh Pillai for providing Mili and Miwi antibodies,
Ilaria Falciatori for trouble-shooting the FACS sorting with me as well as James
Turner who has provided invaluable input on the work presented in this thesis.
Finally, a big thank you goes to my friends and family who have supported and
encouraged me throughout this entire time.

Transcriptional and developmental consequences of
aneuploidy during male meiosis
Christina Ernst
Summary
Eukaryotes have developed stringent regulatory mechanisms that control cell
division and ensure proper chromosome segregation. Maintaining genome integrity
is especially important during meiosis, the specialised cell division programme in the
germline that generates haploid gametes. As these cells transmit genetic information
to the next generation, the consequences of meiotic errors are not restricted to an
organismal level, but can directly impact the fitness of the offspring.
Mammals display a high degree of sexual dimorphism in meiosis with regard to
the stringency of regulatory mechanisms. This manifests in a relatively high degree
of maternally-derived aneuploidies due to weaker checkpoint control in females,
whereas more rigorous checkpoints in males frequently perturb fertility.
Mouse models of aneuploidy often exhibit complete male sterility and early germ
cell arrest, preventing the study of aneuploidy during late and post-meiotic stages
in males. In this thesis, we have used the trans-chromosomic mouse model, Tc1,
which carries a single copy of human chromosome 21 (HsChr21) and show that,
unlike other aneuploid mouse strains, the Tc1 mouse can successfully passage
the exogenous human chromosome through male meiosis and generate aneuploid
offspring.
Our investigations have shown that the presence of the aneuploid human chro-
mosome causes spermatogenic defects due to an arrest at the first meiotic division.
Despite this impairment, we found an unexpectedly high number of aneuploid ga-
metes in Tc1 males and the majority of males were able to produce aneuploid
offspring, albeit at a lower frequency. Transmission of HsChr21 through the male
germline was less efficient compared to female germline transmission, but allowed
us to study the impact of male germline-associated chromatin remodelling on the
transcriptional deployment of HsChr21 in the offspring. This revealed that, despite
fundamentally different developmental dynamics, male- versus female-germline
passage result in indistinguishable transcriptional and regulatory phenotypes.
viii
An important pathway in the male germline involves the expression of piRNAs, a
class of small non-coding RNAs that are commonly found in the germline of animals
where they defend cells against transposable elements. Profiling the expression
of small RNAs in the Tc1 mouse showed that conserved human piRNA clusters
can be successfully transcribed by the mouse piRNA machinery. In addition, we
detected Tc1-specific piRNA sequences that were neither present in human nor
mouse, mapping to a human-specific repeat element. In line with the previously
observed activation of human-specific repeat elements in the Tc1 mouse, this
suggests that novel transcripts arising from human repeats can trigger an adaptive
piRNA response, thereby demonstrating the plasticity of this pathway to newly
invading repeat elements.
Transcriptional profiling of spermatogenic cell populations on a single-cell level al-
lowed us to generate an atlas of gene expression over the course of spermatogenesis
and dissect meiotic silencing dynamics in the presence of aneuploidy. Transcriptional
silencing during meiosis occurs in response to unpaired chromosomes and, in male
germ cells, affects the sex chromosomes due to their largely unpaired nature. We
found that the presence of HsChr21 has no impact on the silencing of chromosome
X, however, the two chromosomes display drastically different silencing patterns with
HsChr21 showing a much weaker repression.
Taken together, this study revealed a higher than expected tolerance for ane-
uploidy in the mouse male germline thus allowing the characterisation of meiotic
checkpoint mechanisms, the meiotic silencing response to unpaired chromosomes
as well as piRNA expression in the presence of an exogenous human chromosome.
Table of contents
List of figures xiii
List of tables xvii
List of code xvii
List of abbreviations xix
1 Introduction 1
1.1 Mammalian gene regulation . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 Epigenetic regulation of gene expression . . . . . . . . . . . 1
1.2 Repeat elements in mammalian genomes . . . . . . . . . . . . . . . 6
1.2.1 Transposable elements . . . . . . . . . . . . . . . . . . . . . 6
1.2.2 Role of transposable elements in genome evolution . . . . . 7
1.2.3 Mechanisms to control transposable elements . . . . . . . . 8
1.3 Epigenetic reprogramming during mammalian development . . . . . 9
1.3.1 Early embryonic development . . . . . . . . . . . . . . . . . . 9
1.3.2 Female germline . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3.3 Male germline . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 Meiosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4.1 Programmed DNA double strand break formation . . . . . . . 16
1.4.2 Homologous chromosome synapsis and crossovers . . . . . 16
1.4.3 Chromosome segregation during meiosis . . . . . . . . . . . 18
1.4.4 Meiotic surveillance mechanisms . . . . . . . . . . . . . . . . 18
1.4.5 Meiotic silencing of unsynapsed chromosomes . . . . . . . . 19
1.5 Mammalian piRNA pathway . . . . . . . . . . . . . . . . . . . . . . . 21
1.5.1 Mouse PIWI proteins and piRNA biogenesis . . . . . . . . . 21
1.5.2 Pre-pachytene piRNAs and the regulation of de novo methylation 23
1.5.3 Pachytene piRNAs and their role in TE regulation . . . . . . . 25
1.5.4 Non-repeat related functions of pachytene piRNAs . . . . . . 27
1.6 The Tc1 mouse model . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.6.1 Previous findings in the Tc1 mouse . . . . . . . . . . . . . . . 29
x Table of contents
2 Materials and Methods 31
2.1 Mouse colony management . . . . . . . . . . . . . . . . . . . . . . . 31
2.2 Tissue preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.3 Histology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.4 Classification of seminiferous tubules . . . . . . . . . . . . . . . . . . 34
2.4.1 Immunofluorescence staining of seminiferous tubules . . . . 34
2.4.2 Interactive learning using ilastik . . . . . . . . . . . . . . . . . 35
2.5 Fluorescence-activated cell sorting of spermatogenic cell populations 36
2.6 Fluorescence in situ hybridisation (FISH) . . . . . . . . . . . . . . . 37
2.7 Chromatin immunoprecipitation followed by high-throughput
sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.8 Strand-specific total RNA-Seq . . . . . . . . . . . . . . . . . . . . . . 44
2.9 BioCAP-Sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.10 smallRNA-Seq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.11 RIP-Seq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.12 Single-cell RNA-Seq on spermatogenic cells . . . . . . . . . . . . . 48
2.13 ChIP-Seq analysis workflow . . . . . . . . . . . . . . . . . . . . . . . 51
2.14 RNA-Seq analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.15 Differential methylation analysis . . . . . . . . . . . . . . . . . . . . . 53
2.16 smallRNA-Seq analysis workflow . . . . . . . . . . . . . . . . . . . . 54
2.17 Single-cell RNA-Seq analysis . . . . . . . . . . . . . . . . . . . . . . 56
3 Transmission of a human chromosome through mouse male meiosis 59
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.2.1 Sub-fertility phenotype of Tc1 males . . . . . . . . . . . . . . 61
3.2.2 Metaphase arrest during the first meiotic division . . . . . . . 64
3.2.3 Male germline transmission of human chromosome 21 . . . . 72
3.2.4 Meiotic loss of human chromosome 21 . . . . . . . . . . . . . 77
3.2.5 Activation of repeat elements on human chromosome 21 . . 80
3.2.6 Transcriptional deployment of human chromosome 21 after
male germline transmission . . . . . . . . . . . . . . . . . . . 88
3.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Table of contents xi
4 Adaptive piRNA response against human-specific repeat elements 95
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.1.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.2.1 piRNA expression in the Tc1 mouse . . . . . . . . . . . . . . 97
4.2.2 Expression of exogenous piRNA sequences from human
chromosome 21 in the mouse . . . . . . . . . . . . . . . . . . 102
4.2.3 Tc1-specific piRNA sequences originating from human repeat
elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5 Single-cell profiling of meiotic silencing throughout spermatogenesis 117
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.1.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.2.1 Single-cell gene expression atlas of mouse spermatogenesis 119
5.2.2 Dynamic gene expression profile throughout spermatogenesis 128
5.2.3 Meiotic silencing dynamics of the X chromosome . . . . . . . 131
5.2.4 Meiotic silencing in response to autosomal aneuploidy . . . . 138
5.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
6 Discussion and Outlook 147
6.1 Repeat activation in the Tc1 mouse . . . . . . . . . . . . . . . . . . . 147
6.2 Low transmission rate of human chromosome 21 through the male
germline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
6.3 Tc1-specific piRNAs across human LTR array . . . . . . . . . . . . . 152
6.4 Meiotic silencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Publications 156
References 157
Appendix A Analysis scripts 191
Appendix B Mouse colony data 215

List of figures
1.1 Three-dimensional chromatin organisation . . . . . . . . . . . . . . . 4
1.2 Epigenetic reprogramming during mouse pre-implantation development 10
1.3 Dynamics of mouse post-implantation development . . . . . . . . . . 11
1.4 Mouse oocyte development . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Developmental dynamics of mouse spermatogenesis . . . . . . . . . 14
1.6 Meiotic chromosome synapsis . . . . . . . . . . . . . . . . . . . . . 17
1.7 Biogenesis and function of pre-pachytene piRNAs in the mouse . . . 24
1.8 Multiple roles of action for pachytene piRNAs during mouse spermato-
genesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.9 Species-specific transcriptional regulation in the Tc1 mouse . . . . . 30
2.1 Tc1 breeding colony . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1 Subfertility phenotype in Tc1 males . . . . . . . . . . . . . . . . . . . 62
3.2 Classification of hypo-spermatogenesis in Tc1 males . . . . . . . . . 63
3.3 Aberrant phospho-Histone H2A.X (Ser139) (γH2AFX) staining in Tc1
animals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.4 Binary decision code for epithelial staging of seminiferous tubules . 66
3.5 Staging of seminiferous tubules on PAS stained tissue sections . . . 67
3.6 Quantification of spermatogenic cell types . . . . . . . . . . . . . . . 69
3.7 Congression defects in Tc1 metaphase cells . . . . . . . . . . . . . 70
3.8 The majority of congression defects are caused by unaligned human
chromosome 21 (HsChr21) . . . . . . . . . . . . . . . . . . . . . . . 71
3.9 Increased apoptosis in stage XII tubules of Tc1 animals . . . . . . . 73
3.10 Transmission of HsChr21 through the male germline is less efficient
than through the female germline . . . . . . . . . . . . . . . . . . . . 74
3.11 Litter size and frequency for different mouse strains . . . . . . . . . . 75
3.12 Histological comparison across multiple somatic tissues . . . . . . . 77
3.13 Sorting of spermatogenic cell types . . . . . . . . . . . . . . . . . . . 78
3.14 Genotyping of spermatogenic cell types . . . . . . . . . . . . . . . . 79
3.15 H3K4me3 enrichment on HsChr21 in human and Tc1 mouse . . . . 81
3.16 Activation of repeat elements in Tc1 liver and testes . . . . . . . . . 82
xiv List of figures
3.17 Activation of distinct repeat elements on HsChr21 in liver and testes 83
3.18 Transcription initiation across HsChr21 . . . . . . . . . . . . . . . . 85
3.19 Transcription initiation across HsChr21 in multiple tissues . . . . . . 86
3.20 NF-Y enrichment at Tc1-specific repeat elements in liver . . . . . . . 87
3.21 Epigenetic profiling of HsChr21 after male germline transmission . . 89
3.22 Differential binding analysis for H3K4me3 and H3K27ac across the
mouse genome in Tc1 and BL6 mice . . . . . . . . . . . . . . . . . . 90
4.1 Size distribution of smallRNA libraries and piRNA cluster calling opti-
misation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.2 Mouse piRNA cluster expression . . . . . . . . . . . . . . . . . . . . 100
4.3 Genome-wide gene and repeat expression profile and repeat expres-
sion in Tc1 animals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.4 piRNA cluster on human chromosome 21 . . . . . . . . . . . . . . . 103
4.5 smallRNA coverage across chromosomes . . . . . . . . . . . . . . . 104
4.6 piRNA cluster detection in sorted cell populations . . . . . . . . . . . 106
4.7 Enrichment of Tc1 smallRNA reads across ERVL-MaLR elements . 107
4.8 Tc1-specific piRNA expression across human-specific LTR array . . 109
4.9 Tc1-specific piRNAs are MIWI-bound and only expressed after cells
enter the pachytene stage . . . . . . . . . . . . . . . . . . . . . . . . 110
4.10 Increased transcription of LTR array in testes from Tc1 mice compared
to human . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
5.1 FACS sorting strategy and single-cell capture on C1 Fluidigm platform 120
5.2 Expression profile of cells within spermatogonial stem cell population 121
5.3 Diffusion pseudotime analysis on spermatogonia . . . . . . . . . . . 123
5.4 Transcriptional profile of differentiating spermatocytes . . . . . . . . 125
5.5 Transcriptional profile of haploid spermatids . . . . . . . . . . . . . . 127
5.6 Marker gene expression throughout spermatogenesis . . . . . . . . 129
5.7 Differential gene expression throughout spermatogenesis . . . . . . 130
5.8 Chromosome X expression throughout spermatogenesis . . . . . . 132
5.9 Differential gene expression of chromosome X throughout spermato-
genesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.10 Expression of X-linked pseudogenes and functional paralogs . . . . 134
5.11 Expression of X-linked housekeeping genes and their retroposed
autosomal copies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
List of figures xv
5.12 Differential gene expression of X-linked genes between spermatogo-
nia and spermatids . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.13 X-linked gene expression in haploid spermatids . . . . . . . . . . . . 139
5.14 Human chromosome 21 expression throughout mouse spermatogen-
esis in Tc1 animals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5.15 Differential gene expression of human chromosome 21 throughout
mouse spermatogenesis . . . . . . . . . . . . . . . . . . . . . . . . . 142
5.16 Expression level of HsChr21-encoded genes in spermatogonia and
spermatocytes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143

List of tables
2.1 Primer sequences for genotyping . . . . . . . . . . . . . . . . . . . . 32
2.2 Antibodies used for immunofluorescence stainings . . . . . . . . . . 36
2.3 Antibodies used for chromatin immunoprecipitation . . . . . . . . . . 40
3.1 Cell type quantification across testis cross-sections using ilastik . . 68
3.2 Quantification of congression defects overlapping with HsChr21 . . . 73
3.3 Fur colour and gender ratios for female and male germline transmission 76
3.4 Quantification of FISH stainings . . . . . . . . . . . . . . . . . . . . . 80
4.1 Number of reads in smallRNA libraries before and after filtering . . . 99
4.2 Number of reads in human smallRNA libraries before and after filtering102
B.1 Transmission rates and breeding statistics for Tc1 females . . . . . . 215
B.2 Transmission rates and breeding statistics for Tc1 males . . . . . . . 216

List of code examples
A.1 Quantification of FISH stainings . . . . . . . . . . . . . . . . . . . . . 191
A.2 Filtering script for ChIP-Seq libraries . . . . . . . . . . . . . . . . . . 192
A.3 Peak calling using MACS2 with summits and pileup . . . . . . . . . . 193
A.4 DiffBind workflow - Human versus Tc1 . . . . . . . . . . . . . . . . . 193
A.5 Differential expression analysis of total RNA-seq using DeSeq2 . . . 194
A.6 Python script to remove cross-mapping smallRNA reads . . . . . . . 197
A.7 R function to call piRNA clusters and identify concordant cluster
across replicates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
A.8 Quality control and normalisation of single-cell RNA-Seq data . . . . 200
A.9 Example workflow for spermatid cell population . . . . . . . . . . . . 205
A.10 Differential gene expression between cell types and visualization
across pseudotime . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
A.11 Workflow to dissect chromosome silencing . . . . . . . . . . . . . . 210

List of abbreviations
γH2AFX Phospho-Histone H2A.X (Ser139)
2C 2-cell like cell
5hmC Hydroxy-methylcytosine
ABC ATP-binding cassette
AE Axial element
Ant4 Adenine nucleotide translocator 4
Anxa5 Annexin A5
APOBEC Apolipoprotein B mRNA-editing enzyme 3
ATM Ataxia-telangiectasia mutation
ATP Adenosine triphosphate
ATR Ataxia telangiectasia and Rad3 related
Aurkb Aurora kinase B
Bcrp1 Breakpoint cluster region pseudogene 1
BER Base excision repair
Bio-CAP Biotinylated CxxC affinity purification
BRCA1 Breast cancer 1
BRU Biological Resources Unit
BSA Bovine serum albumin
BWA Burrows-Wheeler Aligner
C2H2-ZF Cys2-His2 zinc finger
CBF CCAAT binding factor
CC3 Cleaved Caspase-3
Ccin Calicin
Cdca7l Cell division cycle associated 7 like
CDK2 Cyclin-dependent kinase 2
Cdk4 Cyclin dependent kinase 4
cDNA Complementary DNA
CE Central element
xxii List of abbreviations
ChIP-Seq Chromatin immunoprecipitation followed by
high-throughput sequencing
CO Crossover
CRISPR Clustered regularly interspersed short palin-
dromic repeat
CRUK CI Cancer Research UK - Cambridge Institute
CTCF CCCTC-binding factor
CTCFL CTCF-like
Dazl Deleted in azoospermia-like
DC Diffusion component
DMEM Dulbecco’s Modified Eagle Medium
DNMT DNA methyltransferase
DPT Diffusion pseudotime
DS Down syndrome
DSB Double strand break
E Embryonic day
EB Elution buffer
endo-siRNAs Endogenous siRNAs
ESC Embryonic stem cell
EtOH Ethanol
FACS Fluorescence-activated cell sorting
FCS Fetal calf serum
FFPE Formalin-fixed paraffin-embedded
FISH Fluorescence in situ hybridisation
Fkbp6 FK506 binding protein
FOA Fetal oocyte attrition
GLP G9a-like protein
GO Gene ontology
GSNAP Genomic Short-read Nucleotide Alignment
Program
H&E Hematoxylin and eosin
List of abbreviations xxiii
H1fnt H1 Histone Family Member N, Testis Specific
H3K27me3 Trimethylation of histone H3 on lysine 27
H3K4me3 Trimethylation of histone H3 on lysine 4
H3K9ac Acetylation of histone H3 on lysine 9
H3K9me3 Trimethylation of histone H3 on lysine 9
HAT Histone acetyltransferase
HDAC Histone deacetylase
HDM Histone demethylase
HFDs Histone-fold domains
HMR Hypomethylated region
HMT Histone methyltransferase
HNF4α Hepatocyte nuclear factor 4 α
HORMAD1 HORMA-domain protein 1
HR Homologous recombination
HS High sensitivity
HsChr21 Human chromosome 21
IAP Intracisternal particle A
ICM Inner cell mass
IF Immunofluorescence
IFC Integrated Fluidic Circuit
IGV Integrative Genomics Viewer
IHC Immunohistochemistry
IVF In vitro fertilisation
KAP1 KRAB-associated protein 1
KRAB Krüppel-associated box
KRAB-ZFP KRAB zinc finger protein
KZNF KRAB zinc-finger
LB Lysis buffer
LE Lateral element
LINE Long interspersed nuclear element
LTR Long terminal repeat
MACS Model-based Analysis of ChIP-Seq
xxiv List of abbreviations
Mael Maelstrom
Malat1 Metastasis associated lung adenocarcinoma
transcript 1
MaLR Mammalian apparent LTR-retrotransposon
Mb Megabases
Meioc Meiosis specific with coiled-coil domain
MI Metaphase of the first meiotic division
MII Metaphase II
miRNA MicroRNA
Mlh3 MutL homolog3
Mllt10 Histone lysine Methyltransferase DOT1L Co-
factor
mRNA Messenger RNA
MSCI Meiotic sex chromosome inactivation
MSUC Meiotic silencing of unpaired chromosomes
MYOD1 Myoblast determination protein 1
NBF Neutral buffered formalin
NF-Y Nuclear transcription factor
NGS Next-generation seqeuncing
NHEJ Non-homologous end joining
nt Nucleotides
Nxf2 Nuclear RNA export factor 2
P Postnatal day
PAR Pseudoautosomal region
PAS Periodic Acid Schiff
PBS Phosphate buffered saline
PCA Principal component analysis
PCR Polymerase chain reaction
PGC Primordial germ cell
pH3 Phospho-Histone H3 (Ser10)
piRNA Piwi-interacting RNA
PIWI P-element induced wimpy testis
PRDM9 PR domain-containing 9
Prm1 Sperm protamine 1
List of abbreviations xxv
RA Retinoic acid
RAG Recombinase-activating gene
RIP RNA immunoprecipitation
RNAP2 RNA polymerase II
RNF8 RING finger protein 8
RSB Resuspension buffer
RT Room temperature
S100a11 S100 calcium binding protein A11
SAC Spindle assembly checkpoint
Samd4 Sterile alpha motif domain containing 4
Sap30 Sin3A-associated protein, 30kDa
SC Synaptonemal complex
sc-RNA-Seq Single-cell RNA sequencing
SDS Sodium dodecyl sulfate
SETDB1 SET domain bifurcated 1
Sgo1 Shugosin
SINE Short interspersed nuclear element
siRNA Small-interfering RNA
Slc25a31 Solute carrier family 25, member 31
Slx1 Sycp3 like X-linked
Slxl1 Slx-like 1
smRNA Small RNA
SP Side population
Spaca1 Sperm acrosome associated 1
SRE Smaug recognition element
SSC Saline sodium citrate
STAR Spliced Transcripts Alignment to a Reference
Stra8 Stimulated by retinoic acid gene 8
SVA SINE-VNTR-Alu
SYCE Synaptonemal complex central element
SYCP Synaptonemal complex protein
TAD Topologically associating domain
TBS Tris-buffered saline
xxvi List of abbreviations
TE Transposable element
TET Ten eleven translocation
Tex11 Testis-expressed 11
Tex12 Testis expressed 12
TF Transcription factor
TMM Trimmed Mean of M-values
TPRT Target-primed reverse transcription
TRIM28 Tripartite motif containing 28
Tsga8 Testis-specific gene A8
TSP50 Testis specific protease 50
TSPY Testis-specific protein, Y-encoded
Tssk4 Testis-specific serine/threonine-protein ki-
nase 4
Tubb5 Tubulin, beta 5 class I
UNG Uracil-DNA glycosylase
UTR Untranslated region
XCI X chromosome inactivation
Xlr X-linked lymphocyte-regulated
XMMCT Irradiation microcell-mediated chromosome
transfer
Yipf5 Yip Domain Family member 5
Zfy Zinc finger protein Y-linked
ZGA Zygotic genome activation
1 | Introduction
Parts of the introduction and illustrations presented here have been published in
a similar form in Nature Communications (Ernst et al., 2017).
1.1 Mammalian gene regulation
Physiological diversity in mammals is not primarily caused by differences in gene
content, but rather by the fine-tuned mechanisms that control expression of the
genome. This is illustrated by the extensive phenotypic differences between human
and chimpanzees, yet their genomes only differ in ∼4%, comprising about 35 million
single nucleotide differences and ∼90 megabases (Mb) of insertions and deletions
acquired over the course of ∼5-7 million years (The Chimpanzee Sequencing and
Analysis Consortium, 2005; Varki and Altheide, 2005).
Phenotypic differences are not only observed between species, but also between
different cell types within one organism. To allow species- or tissue-specific regulation
of gene expression, regulatory elements embedded in the DNA are interpreted and
bound by sequence-specific transcription factors (TFs), which then orchestrate gene
expression in a combinatorial manner (Vaquerizas et al., 2009). Such TFs can be
expressed ubiquitously or in a tissue-specific manner and many tissues express
so-called master regulators of gene expression (Villar et al., 2014). Prime examples
are hepatocyte nuclear factor 4 α (HNF4α) in liver (Odom et al., 2004) and myoblast
determination protein 1 (MYOD1) in muscle (Lassar et al., 1989). Such tissue-
specific TFs will bind to DNA in a sequence-dependent manner and then recruit
general transcription factors and the RNA polymerase II (RNAP2) machinery to
initiate transcription (Hampsey, 1998).
1.1.1 Epigenetic regulation of gene expression
The existence of tissue-specific transcription factors alone does however not
explain how genetically identical cells within an organism can have different pheno-
types with unique expression patterns once they undergo differentiation (Jaenisch
and Bird, 2003). This requires another layer of regulation that controls the expression
and binding of TFs to the DNA sequence (Berger, 2007).
2 Introduction
This additional control layer was termed epigenetic regulation, describing the
mechanisms that result in heritable changes of gene expression without changes in
the underlying DNA sequence (Goldberg et al., 2007; Sandoval and Esteller, 2012).
The current definition of epigenetics was defined at the 2008 Cold Spring Harbor
epigenetics meeting describing ’an epigenetic trait’ as ’a stably heritable phenotype
resulting from changes in a chromosome without alterations in the DNA sequence’
(Berger, 2007). Epigenetic mechanisms employ different mechanistic layers, includ-
ing post-translational modifications of histone proteins, DNA methylation, chromatin
remodelling and non-coding RNAs, which together form an interactive network to
regulate gene expression (Allis and Jenuwein, 2016).
Histone modifications
Post-translational modifications of histone tails have a dramatic impact on chro-
matin packaging by influencing how densely the DNA is wrapped around the histone
octamer (Bannister and Kouzarides, 2011). Histone modifications are established
by a range of proteins such as histone methyltransferases (HMTs) and histone
demethylases (HDMs), histone acetyltransferases (HATs) and histone deacetylases
(HDACs), as well as other enzymes that establish or erase phosphorylation and
other modifications (Chi et al., 2010). Such histone modifiers often act together in
large protein complexes such as the repressing Polycomb or activating Trithorax
group complexes (Geisler and Paro, 2015).
The effect of histone modifications on gene expression is highly context-dependent,
on the residue that is modified and the type of modification (Jenuwein and Allis,
2001). For example, trimethylation of histone H3 on lysine 4 (H3K4me3) and acetyla-
tion of histone H3 on lysine 9 (H3K9ac) are common modifications of active promoter
regions, whereas trimethylation of histone H3 on lysine 27 (H3K27me3) or trimethy-
lation of histone H3 on lysine 9 (H3K9me3) is associated with inactive promoters
(The ENCODE Project Consortium, 2012) .
DNA methylation
DNA methylation is one of the most studied epigenetic modifications and de-
scribes the covalent modification of the cytosine ring at the 5’ position, typically
occurring at symmetrical CpG dinucleotides in mammals (Smith and Meissner,
2013). This modification is catalysed by enzymes belonging to the family of DNA
methyltransferases (DNMTs). Dnmt1 is known as the ’maintenance methyltrans-
ferase’ as it shows a 10-fold higher preference for hemimethylated DNA (Hermann
Introduction 3
et al., 2004; Song et al., 2012). In contrast to that, Dnmt3A and 3B are known as de
novo methyltransferases that, although capable of methylating hemimethylated DNA,
establish new methylation marks on CpG dinucleotides (Geiman and Muegge, 2010;
Okano et al., 1999). DNA methylation usually acts as a gene silencing mechanism
and the repressing effect is caused by the inhibition of transcription factor binding
and the recruitment of methyl-binding proteins that exert silencing effects (Tate and
Bird, 1993).
The removal of this chemical modification can occur via active or passive mecha-
nisms. Active DNA demethylation involves the activity of ten eleven translocation
(TET) enzymes that oxidise 5-methylcytosine (Ito et al., 2010; Tahiliani et al., 2009).
The resulting 5-hydroxymethylcytosine can then be further oxidised and eventually
be reverted to cytosine by the base excision repair (BER) pathway (Kohli and Zhang,
2013). In contrast, passive DNA demethylation is achieved via successive rounds of
DNA replication in the absence of the maintenance DNA methyltransferase (Chen
and Riggs, 2011).
Three-dimensional genome organisation
In addition to covalent modifications of the DNA itself or of associated proteins,
the three dimensional chromatin structure plays an important role in the regulation
of gene expression. Chromosomes occupy specific territories within the nucleus,
with more gene-rich chromosomes located towards the interior (Cremer et al., 2001).
Within chromosomes, two distinct compartments that correspond to open and closed
chromatin can be distinguished that are termed A and B compartments, respectively
(Figure 1.1) (Lieberman-Aiden et al., 2009). Within compartments, megabase-sized
domains that show a high degree of self-interaction can be identified, which show
distinct patterns of histone modifications, transcriptional activity and replication timing
(Dixon et al., 2012; Nora et al., 2012). Topologically associating domains (TADs)
are largely conserved across cell types as well as across species, which is to a
large extent attributed to the conservation of CTCF binding sites (Dixon et al., 2012;
Vietri Rudan et al., 2015). CCCTC-binding factor (CTCF) is a DNA-binding protein
with 11 zinc finger domains that recognises a large DNA motif (Kim et al., 2007) and,
in its insulator function, co-localises with cohesin to form chromatin loops (Parelho
et al., 2008; Wendt et al., 2008). Chromatin loops function in bringing spatially distal
regulatory elements such as promoters and enhancers into proximity to regulate
gene expression (Hadjur et al., 2009) and such long-range interactions are frequently
found to demarcate TADs resulting in loop domains.
4 Introduction
Figure 1.1 | Three-dimensional chromatin organisation
DNA within the nucleus has a highly ordered structure, with chromosomes occupying
distinct territories resulting in only low levels of inter-chromosomal contacts. Within
chromosomes, DNA can be subdivided into two compartments that resemble active
and inactive chromatin, termed A and B compartments, respectively. Within compart-
ments, domains that show a high degree of self interaction are termed topologically
associating domains (TADs) and are frequently defined by loop structures in which
the anchor points define the domain boundaries. Chromatin loops are enriched for
insulator proteins such as CTCF, Smc1/3, Rad21 and SA1/2. Chromatin loops func-
tion in bringing regulatory elements such as promoters and enhancer into proximity
and are often demarcated by CTCF binding sites that require convergent orientation
of the CTCF binding motif.
Introduction 5
Our current understanding of how these loops are formed is based on the loop
extrusion model, which postulates cohesin molecules to be loaded onto adjacent
DNA strands that slide along the DNA in opposite directions, thereby extruding DNA
to form a loop structure (Fudenberg et al., 2016). The model suggests that the
sliding continues until cohesin encounters CTCF anchor molecules that are oriented
in a convergent manner. This is in accord with the finding that CTCF, as well as the
components of the cohesin complex are enriched at loop boundaries, with the CTCF
binding motifs being mostly found in a convergent orientation and such convergent
CTCF sites show a higher degree of conservation throughout evolution (Rao et al.,
2014; Vietri Rudan et al., 2015).
RNA silencing
RNA silencing describes a gene silencing mechanism present in animals and
plants and is relying on the action of small non-coding RNAs (Ghildiyal and Zamore,
2009). The most important classes of small RNAs (smRNAs) are microRNAs
(miRNAs), small-interfering RNAs (siRNAs), and Piwi-interacting RNAs (piRNAs),
which all differ in their biogenesis pathway and mode of function (Siomi et al., 2011).
smallRNA-mediated gene silencing in mammals can act on both a transcriptional and
post-transcriptional level, depending on the different kinds of smallRNAs (Moazed,
2009).
microRNAs are short (20-23 nucleotides (nt)) endogenous single-stranded RNAs
that were the first small non-coding RNAs to be discovered in 1993 (Lee et al., 1993).
They provide an additional layer above transcriptional gene regulation and are
estimated to regulate approximately 30% of all protein-coding genes (Almeida et al.,
2011; Inui et al., 2010). MicroRNAs exert their function by directing the RNA-induced
silencing complex to complementary sequences in the 3’ untranslated region (UTR)
of messenger RNA (mRNA) (He and Hannon, 2004). For most metazoan microRNAs,
binding to a target mRNA results in translational repression or mRNA destabilisation
(Bartel, 2009), with mRNA degradation being the predominant mechanism of action
(Guo et al., 2010).
Small interfering RNAs can be of exogenous or endogenous origin and are pro-
cessed from double-stranded RNA. Genomic loci giving rise to double-stranded RNA
can either be inverted or direct repeat regions or loci with bi-directional transcription
(Chapman and Carrington, 2007). Endogenous siRNAs (endo-siRNAs) bind to
complementary target RNAs and result in their endonucleolytic cleavage but can
also be involved in transcriptional gene silencing (Malecová and Morris, 2010). As
6 Introduction
these molecules often arise from repeat sequences, they are frequently used by
the cell as a defence mechanism against the expansion of mobile repeat elements
(Girard and Hannon, 2008).
Piwi-interacting RNAs are specifically expressed in the germline and are slightly
longer than the two previously described classes of small RNAs, ranging from
24 - 32 nt in length (Ishizu et al., 2012). piRNAs function to protect the germline
genome from invasive mobile genetic elements, as mutagenic insertions would not
only decrease fertility, but could also impact the next generation. A more detailed
description of the developmental dynamics of piRNAs during spermatogenesis in
the mouse can be found in Chapter 1.5.
1.2 Repeat elements in mammalian genomes
Repetitive sequences make up a large proportion of mammalian genomes with
up to two thirds of the human genome being repeat-derived (de Koning et al., 2011;
The Initial Human Sequencing Consortium, 2001). Repetitive elements are classified
into various classes, depending on the origin, length and structure of the sequence
and whether they are found adjacent to each other (tandem repeats) or are spread
out through the genome (interspersed repeats) (Treangen and Salzberg, 2011).
A sub-fraction of repeats are mobile elements that have played a crucial role in
driving genome evolution by fuelling the development of new genes, introducing
novel immune strategies such as V(D)J recombination, and rewiring gene regulatory
circuitries (Kazazian, 2004; Slotkin and Martienssen, 2007).
1.2.1 Transposable elements
Transposable elements (TEs) are mobile repetitive elements that are dispersed
throughout the genome and are further classified into two broad classes: DNA
transposons and retrotransposons.
DNA transposons move as one piece in a ’cut-and-paste’ fashion in which the
entire element is excised by the element-encoded transposase that also mediates
the integration into a new genomic location (Muñoz-López and García-Pérez, 2010).
In contrast, retrotransposons move in a replicative ’copy-and-paste’ fashion that
involves an RNA intermediate (Smit, 1999). Retrotransposons are further classified
into long terminal repeat (LTR) and non-LTR retrotransposons.
Introduction 7
LTR retrotransposons are derived from retroviruses that lost their ability to move
between cells and are therefore referred to as endogenous retroviruses (ERVs).
Transcription of these elements is driven from the LTRs and the resulting mRNA
gets translated in the cytoplasm, encoding for Gag proteins, reverse transcriptase
and integrase proteins (Thompson et al., 2016). Reverse transcription of the RNA
transcript occurs in virus-like-particles in the cytoplasm and the complementary
DNA (cDNA) copy then gets transposed into the nucleus and integrates into a new
genomic location (Havecker et al., 2004).
Non-LTR retrotransposons also move via an RNA intermediate, but use an
entirely different mechanism for mobilisation. They employ target-primed reverse
transcription (TPRT), in which a DNA single-strand break is introduced by element-
encoded endonucleases. The freed DNA single-strand is then used to prime the
reverse transcription of an arising RNA transcript resulting in duplicated elements
that are often 5’ truncated and thereby rendered retrotransposition-defective (Levin
and Moran, 2011). Long interspersed nuclear elements (LINEs) are one of the main
representatives of non-LTR retrotransposons and are self-autonomous by encoding
all proteins required for retrotransposition themselves. In contrast to that, short
interspersed nuclear elements (SINEs) are non-autonomous and do rely on other
mobile elements for their retrotransposition (Slotkin and Martienssen, 2007).
1.2.2 Role of transposable elements in genome evolution
Many transposable elements, such as LTR retrotransposons, appear to be no
longer active in humans; however, they may have played an important role in shaping
the genome during hominid evolution (Slotkin and Martienssen, 2007). Transposition
of mobile elements can result in large DNA rearrangements and unequal crossing-
over between transposable elements from different genomic locations can lead to
deletions or duplications, thereby affecting genome structure (Kazazian, 2004).
Over time, mammalian genomes have co-opted different proteins from trans-
posable elements. An important example are recombinase-activating gene (RAG)
proteins that seem to originate from DNA transposons and are now important players
in the adaptive immune response of mammals by mediating V(D)J recombination
(Agrawal et al., 1998). In addition, telomerase is believed to originate from an ancient
retrotransposon due to its structural analogy with retroviral reverse transcriptase,
and is now a crucial player in maintaining telomere structure (Belfort et al., 2011).
Besides genes being co-opted by the host genomes, TEs are also a source of
regulatory elements such as binding sites for TFs or insulator proteins (Bourque
8 Introduction
et al., 2008). Species-specific expansion of certain repeat families carrying a specific
regulatory element can therefore dramatically alter the regulatory landscape of the
host genome (Schmidt et al., 2012).
1.2.3 Mechanisms to control transposable elements
Although host genomes can benefit from TEs, these elements pose a threat to
their host genomes due to the mutagenic nature of their expansion mechanism. In
humans, several genetic diseases have been linked to TE insertions and increased
activity of LINE-1 elements has been reported in various cancers (Hancks and
Kazazian, 2016; Kemp and Longworth, 2015). To minimise the negative impact of
TEs, host genomes evolved a number of repressive mechanisms that silence repeat
elements and prevent them from transposition (Prak and Kazazian, 2000).
One important repressive mechanism is DNA methylation at CpG dinucleotides,
which inhibits the transcription of TEs, preventing their expansion (Levin and Moran,
2011). Apart from DNA methylation, repressive histone marks, specifically H3K9me3,
play a crucial role in the transcriptional silencing of TEs.
Targeted silencing of TEs is frequently mediated by Cys2-His2 zinc finger (C2H2-
ZF) proteins that form one of the largest families of TFs in both the mouse and
human genome (Emerson and Thomas, 2009). DNA binding is mediated via the
use of several dozen C2H2-ZF domains in a combinatorial manner resulting in large
and diverse binding motifs. Silencing is achieved by coupling of these proteins
to effector domains, the most common being the Krüppel-associated box (KRAB).
Such combinations are termed KRAB zinc finger proteins (KRAB-ZFPs) in mice
or KRAB zinc-fingers (KZNFs) in humans (Urrutia, 2003). KRAB-ZFPs mediate
transcriptional silencing by interacting with the co-repressor protein KRAB-associated
protein 1 (KAP1) also known as tripartite motif containing 28 (TRIM28), which
serves as a scaffold domain to recruit chromatin modifiers like the histone lysine-
methyltransferase SET domain bifurcated 1 (SETDB1) (Schultz et al., 2002, 2001).
RNA silencing is mainly acting on a post-transcriptional level, defending the
genome against invading viruses and transposable elements (Girard and Hannon,
2008). Out of the small RNA repertoire, endo-siRNAs and piRNAs (discussed in
(Chapter 1.5) are the main players involved in TE defence.
Last but not least, also acting on a post-transcriptional level are cytosine deami-
nases such as apolipoprotein B mRNA-editing enzyme 3 (APOBEC) that deaminate
cytidines during the first strand cDNA synthesis, leading to either cDNA degradation,
or the integration of a mutated sequence (Chiu and Greene, 2008).
Introduction 9
1.3 Epigenetic reprogramming during mammalian
development
The fusion of gametes from the opposite sex generates a single-cell with the
potential to build a new organism - the zygote. For the zygote to acquire totipotency,
the ability to generate all cell types contributing to embryonic and extra-embryonic
tissues, extensive epigenetic reprogramming is necessary shortly after fertilisation.
A second wave of epigenetic reprogramming occurs after sex determination in
primordial germ cells (PGCs), after these cells committed to form the germline of
the developing embryo and is different between genders.
1.3.1 Early embryonic development
Epigenetic reprogramming in the developing embryo starts shortly after fertili-
sation and shows sexually dimorphic features for the maternal and paternal DNA
contribution. For the first 15 - 18 hours after fertilisation, the two parental genomes
co-exist in spatially separated compartments, the maternal and paternal pronuclei,
which gradually converge and fuse during syngamy just a few hours prior to the first
cleavage division (Wossidlo et al., 2010). Due to the different chromatin states in
which the parental genomes are transmitted to the zygote, different reprogramming
dynamics occur within pronuclei (Figure 1.2).
The maternal genome deposited in the oocyte is transcriptionally silent at fertili-
sation, but otherwise in a normal chromatin state with intermediate levels of DNA
methylation and bound to histone proteins (Fraser and Lin, 2016). In contrast to that,
the paternal genome is transmitted in a highly condensed state, bound to protamines,
and highly methylated. Hence, the first act of epigenetic remodelling of the paternal
genome is the exchange of protamines for maternally-derived histones which is com-
pleted within 3 hours post fertilisation (McLay and Clarke, 2003). Decondensation of
the paternal genome is followed by active DNA demethylation which is observed as
early as 3 hours after fertilisation and completed by 6 hours post fertilisation, well in
advance of DNA replication (Mayer et al., 2000; Messerschmidt et al., 2014; Oswald
et al., 2000). This process is mediated by TET3 which is specifically enriched in the
paternal pronucleus and leads to an enrichment of hydroxy-methylcytosine (5hmC)
(Gu et al., 2011), which is then removed by the BER (Hajkova et al., 2010). Among
the genomic regions escaping this global demethylation are some evolutionary young
repetitive elements and paternal imprints (Iqbal et al., 2011; Li et al., 2008).
10 Introduction
Figure 1.2 | Epigenetic reprogramming during mouse pre-implantation
development
Epigenetic reprogramming during mouse pre-implantation development displays
sexually dimorphic patterns for DNA demethylation of the paternal and maternal
DNA. Active DNA demethylation of the paternal genome occurs within a few hours
after fertilisation before DNA replication commences. The first cleavage division
usually occurs around 18 hours post-fertilisation generating the 2-cell embryo. This
stage lasts for almost 24 hours, during which zygotic genome activation occurs.
This is followed by a series of asynchronous cleavage divisions during which the
cells divide without cell growth until they form the morula which corresponds to
a 16-cell embryo. The morula then undergoes compaction to form the blastocyst
which consists of a layer of trophoectoderm cells that encloses a cavity called the
blastocoel and the inner cell mass.
In contrast, the maternal genome undergoes a steady replication-coupled DNA
demethylation, which is achieved by the exclusion of the maintenance DNA methyl-
transferase DNMT1 from the nucleus, resulting in DNA methylation levels matching
that of the paternal DNA around the morula stage with the exception of parental
imprints (Kurihara et al., 2008; Rougier et al., 1998).
Interestingly, despite certain repetitive elements escaping global DNA demethyla-
tion in the zygote, the two-cell embryo displays an unnaturally high level of repeat
activity. In mouse embryos, the activation of MERVL elements is necessary to
orchestrate gene expression by using these elements as alternative promoter se-
quences to drive the expression of nearby genes (Franke et al., 2017; Friedli et al.,
2014; Peaston et al., 2004). This upregulation appears to be highly specific to the
2-cell embryo and can also be observed in 2-cell like cells (2Cs), a transient stage of
embryonic stem cells (ESCs) that closely resemble the 2-cell embryo (Fadloun et al.,
2013; Ishiuchi et al., 2015; Macfarlan et al., 2012) .
Introduction 11
Figure 1.3 | Dynamics of mouse post-implantation development
Mouse post-implantation development begins with the implantation of the blastocyst
into the uterine wall, which is shortly followed by gastrulation. Primordial germ cells
(PGCs) are displayed as purple circles that first arise around E6.5 in the epiblast and
then migrate to the genital ridges, reaching the gonads between E10.5-E11.5. PGCs
continue to proliferate and undergo sex determination at E11.5 from which point
onwards the development differs between female and male PGCs and morphological
differences between the gonads can be observed as early as E12.5.
Interestingly, while such an activation of repeat elements appears to be conserved
across mammalian development, the class of repeat elements appear to differ
between species (Grow et al., 2015; Whiddon et al., 2017). The activation of
repetitive elements coincides with the zygotic genome activation (ZGA), the major
wave of transcriptional activation of the embryonic genome, which results in a
switch from maternal to zygotic transcripts (Lee et al., 2014). Shortly after genome
activation, these repeat elements are repressed in a H3K9me3-mediated manner,
most likely involving KRAB-ZFPs, which are also highly active in ESCs (Karimi et al.,
2011; Matsui et al., 2010; Walter et al., 2016).
After implantation, lineage-specific de novo methylation is observed in cells from
the inner cell mass (ICM) and cells commit to different lineages (Lees-Murdock
and Walsh, 2008). The cells committed to forming the germline of the developing
embryo first arise around embryonic day (E)6.5 from the epiblast and are termed
PGCs (Figure 1.3). By E7.5 a founding population of ∼40 PGCs have formed,
which then migrate to the genital ridges and colonise these around mid-gestation at
E10.5-E11.5 (Ginsburg et al., 1990). During their migration period, PGCs undergo
various epigenetic changes, such as global erasure of H3K9me1/2, which is linked
to decreased expression of the H3K9 methyltransferase G9a-like protein (GLP), as
well as an increase in H3K27me2 and various histone variants (Seki et al., 2007).
Reprogramming continues upon arrival at the genital ridge and includes a global
12 Introduction
loss of DNA methylation (Popp et al., 2010). Sex determination occurs around E11.5
and commits cells to the developmental programme of either the female or male
germline, which are strikingly different.
1.3.2 Female germline
Female PGCs undergo a phase of mitotic proliferation with incomplete cytokinesis
that leads to the formation of germ cell cysts around E13.5 (Figure 1.4) (Pepling,
2006). Oocytes then enter meiosis and progress through prophase, reaching and
completing the pachytene stage at E17 and E21 (1-2 days after birth), respectively
(Speed, 1982). Once the cells reach the diplotene stage, oocytes enter a special
arrest called dictyate, which can last several months in mice, or even decades
in humans (Pepling and Spradling, 2001). Around the time of birth, germ cell
cysts are breaking down to initiate follicle formation during which a substantial
number of oocytes (∼80%) are lost in a process termed fetal oocyte attrition (FOA)
(Burgoyne and Baker, 1984; Pepling and Spradling, 2001). The surviving oocytes
are maintained as primordial follicles which constitute the pool of germ cells for the
entire lifespan of the animal.
From puberty onwards, a set of primordial follicles will re-enter meiosis with each
estrus cycle (Morelli and Cohen, 2005) and enter a growth phase during which
the diameter of the oocyte increases from ∼20µm to ∼85µm while the surrounding
follicle cells are transformed into the cuboidal granulosa cells and the zona pellucida
is formed (Svoboda et al., 2015). Once oocytes reach a critical size during this
growth phase, de novo methylation of the maternal DNA is triggered just before
ovulation (Trasler, 2006). Ovulated oocytes then undergo the first meiotic division
which leads to the extrusion of the first polar body after which they arrest again
before the second metaphase which is only completed upon fertilisation.
1.3.3 Male germline
Male PGCs continue to proliferate in the gonads until they enter a mitotic G0/G1
arrest at E14.5, at which point ∼25.000 PGCs, called prospermatogonia, are found
in each gonad (Figure 1.5) (Tam and Snow, 1981). These cells remain mitotically
arrested until postnatal day (P)2, during which time de novo DNA methylation and
the establishment of paternal imprints takes place (Kato et al., 2007). This process
is mediated by the de novo methyltransferases DNMT3A and DNMT3B as well as
their catalytically inactive interaction partner DNMT3L (Bourc’his and Bestor, 2004).
Introduction 13
Figure 1.4 | Mouse oocyte development
Oogenesis in mice begins with female PGCs that divide mitotically to form germ
cell cysts before entering meiosis. Meiosis arrests at the diplotene stage of the first
meiotic prophase which is termed dictyate. Oocytes then remain quiescent until
they enter a growth phase during which the follicle is being built and the oocytes are
growing in size. Maturation involves gradual transition from secondary follicles to
antral follicles which grow further in size until ovulation is triggered that allows the
completion of the first meiotic division.
DNMT3L is essential for spermatogenesis by guiding de novo DNA methylation
and binding to unmethylated histone H3 lysine 4 tails (Ooi et al., 2007). De novo
methylation was shown to occur in two waves and many TEs manage to escape
the first ‘generic’ wave that occurs by E16.5. Interestingly, many of these evading
TEs (mainly LINE1 elements and intracisternal particle As (IAPs)) correspond to
evolutionary young elements (Molaro et al., 2014).
However, DNMT3L mutants only arrest at the pachytene stage, which occurs
several days and mitotic cell divisions later, and are characterised by a global
de-repression of LINE1 and IAP elements. This suggest that DNA methylation is
dispensable for TE silencing until the onset of meiosis due to additional transcriptional
silencing mediated by H3K9me2 (Di Giacomo et al., 2013).
14 Introduction
Figure 1.5 | Developmental dynamics of mouse spermatogenesis
Male PGCs in the gonads continue to proliferate until E14.5 when they enter a mitotic
arrest until P2 at which point they re-enter the mitotic cell cycle. Spermatogenesis
then initiates shortly after birth when type-A spermatogonia divide and give rise
to type-B spermatogonia that are committed to the meiotic programme. The first
primary spermatocytes appear around P10 and undergo meiosis with the first
haploid spermatids occurring around P20. These cells then undergo a differentiation
programme termed spermiogenesis during which these cells elongate and develop
sperm-specific structures such as the acrosome and the flagellum to eventually
form mature sperm cells. The process of gametogenesis and spermatogenesis
is associated with epigenetic reprogramming including drastic changes in DNA
methylation and histone modifications such as H3K9me2. During later stages of
spermatogenesis, global changes in histone composition and finally a histone-to-
protamine exchange results in chromatin compaction. Figure from Ernst et al. (2017).
This repressive histone mark however, displays a dynamic pattern during sper-
matogenesis and completely disappears at the pachytene stage (Figure 1.5) (Za-
mudio et al., 2015). Loss of H3K9me2 can be correlated with the expression of the
histone methyltransferase G9a which disappears at the leptotene stage preceding
the H3K9me2 loss (Tachibana et al., 2007). In DNA methylation mutants, the loss
of repressive histone marks is concomitant with the gain of active marks such as
H3K4me3.
Introduction 15
Spermatogenesis
Spermatogenesis occurs in continuous waves throughout the life span of an
animal and begins shortly after birth when pro-spermatogonia resume mitotic cell
divisions at P2. This gives rise to type-A spermatogonia, which are found on the
basement membrane of seminiferous tubules. These cells have self-renewing poten-
tial but also produce type-B spermatogonia which can enter meiosis and become
spermatocytes (Donovan, 1998). The first meiosis is initiated in a synchronous man-
ner around P10 after several rounds of mitotic divisions. As cells progress through
spermatogenesis, they move away from the basement membrane towards the lumen
of seminiferous tubules and enter preleptotene/leptotene, zygotene, pachytene and
diplotene stages on P10, 12, 14, and 17, respectively (Bellvé et al., 1977). The
two consecutive cell divisions occur within a short time frame and generate the first
set of round spermatids to be found at P20. Round spermatids then undergo a
complex differentiation programme called spermiogenesis, during which they form
sperm-specific structures such as the acrosome and flagellum and undergo nuclear
compaction (O’Donnell, 2015). Genome compaction is achieved via genome-wide
replacement of histones with transition proteins at first, followed by protamines which
results in an up to twenty-times more condensed structure due to the basic character
of these proteins (Balhorn, 2007; Dadoune, 1995). To allow the transition, a loose
chromatin structure is adopted and mediated in part by the incorporation of various
testis-specific histone variants which leads to a global transcriptional activation that
includes TEs (Rathke et al., 2014; Soumillon et al., 2013). During spermiogenesis,
spermatids undergo elongation and reach a stage of transcriptional shut-down after
which they are transcriptionally inactive. This is followed by cytoplasmic extrusion
generating mature spermatozoa which are then exported via the lumen of seminifer-
ous tubules into the epididymus where they undergo further maturation and quality
control (O’Donnell, 2015).
1.4 Meiosis
Meiosis is a specialised cell division programme that serves the generation of hap-
loid gametes and consists of two consecutive cell divisions without an intermediate S
phase. As such, meiosis has several unique features compared to mitosis, including
programmed DNA double strand break (DSB) formation, homologous recombination
and chromosome synapsis (Marston and Amon, 2004). These events take place
during prophase of the first meiotic division, which is extremely prolonged and, in
16 Introduction
mice, can last for several days or even months in males or females, respectively (Soh
et al., 2017). After replicating their DNA in S phase during which sister-chromatid
cohesion is established, cells enter prophase of meiosis I which resembles a mitotic
G2 phase and can be divided in four substages: leptonema, zygonema, pachynema
and diplonema.
1.4.1 Programmed DNA double strand break formation
One of the first events during leptonema is the programmed formation of DNA
DSBs across the entire genome by SPO11 (Keeney et al., 1997a). This occurs in a
non-random manner, as SPO11 is recruited via H3K4me3 to specific recombination
hotspots in the genome (Brick et al., 2012; Grey et al., 2011). These hotspots are
defined by PR domain-containing 9 (PRDM9), a histone methyltransferase, which
recognises a particular DNA binding motif via its zinc finger domain and methylates
histone H3 lysine 4 tails (Baudat et al., 2010; Myers et al., 2010; Parvanov et al.,
2010).
The induced DNA DSBs are then substrates for homologous recombination (HR)
which enables the generation of meiotic crossovers (COs). COs are essential for
chromosome segregation during the first meiotic division and thus, at least one CO
per chromosome is necessary (Dumont, 2017). Mice generally display between 200-
300 DSBs per nucleus which result in 20-25 crossovers (Cole et al., 2012; Moens
et al., 2002). During meiosis, HR is the favoured mechanism for DNA DSB repair
due to a downregulation of the components involved in non-homologous end joining
(NHEJ) (Goedecke et al., 1999). The SPO11-mediated induction of DNA DSBs
results in SPO11 covalently bound to the DNA (Keeney et al., 1997b), which upon
removal (Neale et al., 2005) exposes the DNA and result in a nucleolytic resection
of the 5’ strand (Bishop et al., 1992). This generates a single-stranded 3’ overhang
which can then invade neighbouring DNA to achieve HR (Sun et al., 1991).
1.4.2 Homologous chromosome synapsis and crossovers
Homologous recombination between non-sister chromatids requires physical
proximity between homologous chromosomes, which is achieved during the zygotene
stage in form of chromosome synapsis (Handel and Schimenti, 2010). Synapsis
involves the formation of a proteinaceous structure between homologous chromo-
somes which is termed the synaptonemal complex (SC) and is composed of several
meiosis-specific proteins (Zickler and Kleckner, 1999).
Introduction 17
Figure 1.6 | Meiotic chromosome synapsis
In leptonema of meiotic prophase I, axial elements associate with chromosomes
along their length which then facilitates the pairing between homologous chro-
mosomes during zygotene. Proximity between homologs allows the loading of
synaptonemal complex proteins such as transverse filaments and proteins of the
central element that act as a zipper-like structure leading to full synapsis of the
homologs in pachynema.
These include synaptonemal complex protein (SYCP) 1, 2 and 3, synaptonemal
complex central element (SYCE) 1, 2 and 3 as well as testis expressed 12 (Tex12).
To start with, chromosomes associate with axial elements (AEs) that are composed
of SYCP2 and SYCP3 along their length during leptonema, which mediates the
pairing between homologous chromosomes (Lammers et al., 1994; Offenberg et al.,
1998). This is followed by the assembly of central elements (CEs) between the two
homologs during zygonema, involving SYCE1/2/3 and TEX12 (Costa et al., 2005;
Hamer et al., 2006; Schramm et al., 2011). The SC is fully formed by the pachytene
stage, consisting of lateral elements (LEs) (former AEs) and the central region
containing CEs and transverse filaments that are comprised of SYCP1 (Meuwissen
et al., 1992). The SC acts in a zipper-like fashion and upon formation ensures the
close proximity of homologous chromosomes to allow meiotic recombination (Fraune
et al., 2012).
DNA DSB repair via strand invasion results in the formation of a D-loop in which
the homologous chromosome is used as a template to synthesise the resected DNA
fragment (Hunter and Kleckner, 2001). This leads to the formation of intermediate
structures called double Holliday junctions which can result in crossovers (Collins and
Newlon, 1994; Schwacha and Kleckner, 1995). At the end of the pachytene stage,
the SC is broken down and homologous chromosomes only remain attached to each
other via their crossovers, forming the characteristic structures called chiasmata.
18 Introduction
1.4.3 Chromosome segregation during meiosis
The events of meiosis and the segregation of homologous chromosomes during
the first division, followed by the segregation of sister chromatids during the second
division requires a fine-tuned regulation of cohesin molecules that establish cohesion
between sister chromatids. The cohesin complex functioning in meiosis has several
substitutions compared to mitotic cells, including the meiosis-specific kleisin subunit
Rec8 that replaces Rad21 (Bannister et al., 2004). After the breakdown of the SC
and the start of spindle attachment, homologous chromosomes are held together
via crossovers that exert tension onto the spindle (Hirose et al., 2011). Step-wise
removal of cohesin from the distal parts of the chromosomes, but sparing cohesin
molecules located near the centromere, ensures that sister chromatids remain
attached to each other resulting in the segregation of homologous chromosomes
(Clift and Marston, 2011). The protection of centromeric cohesin is mediated by
Shugosin (Sgo1) which recruits the protein phosphatase PP2A and thereby prevents
the phosphorylation of Rec8 which would lead to the breakdown of the cohesin
complex. Sgo1-PP2A is removed before anaphase of the second meiotic division
allowing the cleavage of Rec8 at centromeres and resulting in the segregation of
sister chromatids (Ishiguro et al., 2010).
1.4.4 Meiotic surveillance mechanisms
To ensure accurate chromosome segregation, meiosis is equipped with several
surveillance mechanisms that act either during prophase I or the first metaphase.
Most of these checkpoints are more stringent in males than in females and, when
triggered, often lead to male infertility.
One such mechanism that shows a higher stringency in males is the spindle
assembly checkpoint (SAC). The SAC acts during the first metaphase and surveils
the attachment to microtubules and proper tension on the spindle, and when triggered
prevents the progression to anaphase (Sun and Kim, 2012).
In contrast, surveillance mechanisms acting during prophase monitor chromo-
some synapsis and DNA DSB repair. Prolonged DNA damage persisting into
the pachytene stage due to a failure in meiotic recombination results in ataxia-
telangiectasia mutation (ATM) kinase-mediated pachytene arrest (Pacheco et al.,
2015). However, failures in meiotic recombination are not the only cause of persistent
DNA DSBs but they can also result from increased activity of repetitive elements
that cause DNA DSBs during retrotransposition (Soper et al., 2008).
Introduction 19
In addition to this DSB-dependent mechanism, studies in SPO11 and Mei1
mutant mice, which have no programmed meiotic DNA DSBs, have shown the
presence of a DSB-independent checkpoint (Baudat et al., 2000; Di Giacomo et al.,
2005; Libby et al., 2002; Romanienko and Camerini-Otero, 2000). Further proof
for such a DSB-independent checkpoint comes from XO female mice which show
increased oocyte loss during the pachytene stage. This oocyte loss is accompanied
by a shift from oocytes in which the X chromosomes is asynapsed to oocytes in
which the X chromosome is self-synapsed in a hairpin-like structure, suggesting that
cells with asynapsis are actively eliminated (Burgoyne and Baker, 1985; Hodges
et al., 2001). HORMA-domain protein 1 (HORMAD1) was identified to be a key
player in this pathway by recruiting the ataxia telangiectasia and Rad3 related (ATR)
kinase and breast cancer 1 (BRCA1) to unsynapsed chromosome axes which then
leads to the phosphorylation of H2AFX (γH2AFX) (Daniel et al., 2011; Kogo et al.,
2012). Extensive asynapsis leads to a sequestration of ATR, BRCA1 and γH2AFX
on autosomes, causing a failure in meiotic sex chromosome inactivation (MSCI) (see
below) (Shin et al., 2010).
1.4.5 Meiotic silencing of unsynapsed chromosomes
Meiotic silencing occurs in response to unpaired chromosomes during the
pachytene stage of prophase I and results in transcriptional silencing of the en-
tire asynapsed chromosome. While the exact purpose of meiotic silencing is still
under debate, it is believed to be part of the prophase surveillance mechanisms
that ensure chromosome synapsis (Turner, 2015). Meiotic silencing is induced by
so-called sensors at the AEs of chromosomes, which when unpaired results in a
signalling cascade and modification of so-called effectors that are located in distal
chromatin loops, therefore leading to the silencing of the entire chromosome. Sen-
sors for meiotic silencing include the AE elements SYCP3 and SmC1B (Biswas et al.,
2013; Kouznetsova et al., 2009), as well as the cell cycle kinase cyclin-dependent
kinase 2 (CDK2) (Viera et al., 2009) and other proteins that are recruited in re-
sponse to asynapsis such as HORMAD1/2 and BRCA1 (see above). This leads
to the recruitment of ATR, which then diffuses along the chromosome loops and
phosphorylates H2AFX leading to transcriptional gene silencing via the induction of
heterochromatin (Royo et al., 2013).
The current model by which meiotic silencing exerts a checkpoint function is
based on the idea that transcriptional silencing of an entire chromosome will starve
the cell of all the genes encoded on a given chromosome - the starvation model. This
20 Introduction
however means, that the precise mechanism of pachytene arrest differs, depending
on which chromosome is unsynapsed (Burgoyne et al., 2009). The starvation model
is supported by observations in XO females in which the unsynapsed X chromosome
is transcriptionally silenced which is lethal for oocytes due to the relative enrichment
of oocyte-specific genes on the X chromosome (Burgoyne et al., 2009).
However, it was observed that the meiotic silencing response appears to have a
dose-dependency, as it breaks down in cases with a high degree of asynapsis, due
to a limited pool of BRCA1 (Kouznetsova et al., 2009).
Meiotic sex chromosome inactivation
The best studied example of meiotic silencing occurs in male germ cells and
affects the sex chromosomes. The X and Y chromosomes are largely heteromorphic
except for a small region called the pseudoautosomal region (PAR) that allows for
the obligatory crossover between these chromosomes and spans about 1 megabase
in mice (Perry et al., 2001).
The phenomenon of meiotic silencing of the sex chromosomes has been long
recognised and was first observed as a more intense stain forming a nuclear com-
partment, which was termed the sex body (Monesi, 1965; Solari, 1964; Solari and
Tres, 1967). This structure is the result of heterochromatinisation of the X and Y
chromosomes and corresponds to the second wave of γH2AFX stainings that is
observed in pachytene spermatocytes (Hamer, 2002). MSCI starts at the zygotene-
to-pachytene transition, persists throughout meiosis and extends mostly throughout
post-meiotic spermiogenesis (Greaves et al., 2006; Turner et al., 2006). So far,
no genes have been identified that escape the silencing during meiosis, which is
in contrast to the number of genes escaping X chromosome inactivation (XCI) in
female somatic cells (3-7% in mouse, 15% in human) (Disteche and Berletch, 2015;
Greaves et al., 2006). However, haploid spermatids show partial re-activation of
X-linked genes (approximately 18% of X-linked genes), with a particular enrichment
for multicopy genes (Mueller et al., 2008).
Transcriptional silencing of the sex chromosomes at pachytene is essential for
spermatogenesis and upon failure results in mid-pachytene arrest (Mahadevaiah
et al., 2008). Spermatogenic arrest is due to the expression of spermatocyte-lethal
genes encoded on the X and Y chromosome and has been demonstrated in XYY
mice, in which the two Y chromosomes occasionally achieve synapsis resulting in
the expression of zinc finger protein Y-linked (Zfy) 1 and 2, which upon insertion onto
autosomes also cause pachytene arrest in spermatocytes (Royo et al., 2010).
Introduction 21
1.5 Mammalian piRNA pathway
The different waves of epigenetic reprogramming and global remodelling of the
genome allow repetitive elements to become transcriptionally active in the germline.
Since active cell divisions are a requirement for successful retrotransposition of TEs
(Shi et al., 2007), the burden is heavier in the male germline in mammals, which is
marked by continuous waves of spermatogenesis throughout the life span compared
to female germ cells of which a defined number arrests in meiosis I during embryonic
development (Handel and Schimenti, 2010). Mammalian male germlines therefore
have an active piRNA pathway to control this increased activity of TEs.
1.5.1 Mouse PIWI proteins and piRNA biogenesis
The mouse male germline expresses three proteins of the P-element induced
wimpy testis (PIWI) family: MILI, MIWI and MIWI2, all of which are essential for sper-
matogenesis and male fertility (Carmell et al., 2007; Deng and Lin, 2002; Kuramochi-
Miyagawa et al., 2001). Each PIWI protein exhibits a unique expression profile
throughout spermatogenesis and associates with a defined population of piRNAs
(Aravin et al., 2006; Girard et al., 2006; Grivna et al., 2006; Lau et al., 2006). PIWI
proteins are members of the highly conserved Argonaute family and are charac-
terised by their PIWI, MID and PAZ domains (Cerutti et al., 2000). The PIWI domain
is essential for the endonucleolytic slicer activity of PIWI proteins (Song et al., 2004),
while the MID and PAZ domains function in anchoring the 5’ and 3’ ends of piRNAs,
respectively (Ma et al., 2005; Parker et al., 2005; Vourekas et al., 2012). In addition
to their association with different PIWI proteins, piRNAs can be further distinguished
into primary and secondary piRNAs based on their biogenesis pathway.
Primary piRNAs
The biogenesis of primary piRNAs begins with the transcription of a single-
stranded piRNA precursor molecule by RNAP2 from defined genomic locations
termed piRNA clusters. piRNA precursor molecules are then exported into the
cytoplasm which might be facilitated by Maelstrom (MAEL) (Castañeda et al., 2014).
Once in the cytoplasm, piRNA precursors associate with the RNA helicase MOV10L1
that fuels them into the biogenesis pathway by facilitating the interaction with PLD6
(MITOPLD), the endonuclease that generates the 5’ ends of primary piRNAs (Ipsaro
et al., 2012; Nishimasu et al., 2012).
22 Introduction
Intermediate fragments with a 5’ uridine (1U) are preferentially bound by PIWI
proteins and thereby stabilised, leading to the strong 1U bias that is characteristic for
primary piRNAs (Cora et al., 2014). PIWI-bound piRNA intermediates then undergo
3’ end processing which most likely involves exonucleolytic shortening by a 3’-5’
trimmer that has not been identified in mammals yet. The final length of piRNAs
during 3’ end maturation is determined by the footprint of their associated PIWI
protein, which protects the 3’ end from further trimming (Kawaoka et al., 2011). This
generates size-specific piRNA populations for the different PIWI proteins with a mean
length of 26nt for MILI-, 28nt for MIWI2- and 30nt for MIWI-bound piRNAs (Aravin
et al., 2008, 2007; Girard et al., 2006). Mature 3’ ends are then 2’-O methylated by
the RNA methyltransferase HENMT1 resulting in increased stability and binding to
the PAZ domain of their PIWI protein (Lim et al., 2015).
Secondary piRNAs
Primary piRNAs can then guide the PIWI-mediated cleavage of antisense tran-
scripts between nucleotide 10 and 11 to produce secondary piRNAs, a process
known as the ping-pong cycle. Secondary piRNAs therefore have a defining 10nt
5’ overlap with the primary piRNA that mediated the cleavage and often have ade-
nine at their tenth position (10A) (Aravin et al., 2008). This 10A bias has long
been thought to originate from complementarity to the 5’ uridine of primary piRNAs.
However, base pairing at the first position is not necessary for efficient cleavage
by Argonaute proteins (Wang et al., 2009) and it appears that mammalian PIWI
proteins rather have a preference for adenine at the t1 position that is not determined
by base pairing (Wang et al., 2014). Secondary piRNAs can then in turn fuel the
production of the initiating primary piRNAs, thereby completing the ping-pong cycle.
In the mouse, secondary piRNA production is driven by the slicer activity of MILI
and most ping-pong cycles rely on homotypic MILI:MILI interactions with occasional
heterotypic MILI:MIWI2 associations (De Fazio et al., 2011). These heterotypic
ping-pong interactions are however not reciprocal and the loading of MIWI2 with
secondary piRNA is entirely dependent on MILI (Aravin et al., 2008).
Introduction 23
1.5.2 Pre-pachytene piRNAs and the regulation of de novo
methylation
MILI is the first PIWI protein to be expressed during development, starting at
E13.5 in PGCs with a dynamic expression profile up until the round spermatid
stage (Di Giacomo et al., 2013; Kuramochi-Miyagawa et al., 2004, 2001). In con-
trast, MIWI2 is only expressed during a short period of time, starting at E14.5 until
shortly after birth with protein levels being undetectable by P4 (Aravin et al., 2008;
Kuramochi-Miyagawa et al., 2004). This expression window largely coincides with
mitotic arrest of PGCs and de novo DNA methylation. Knockout of either MILI or
MIWI2 leads to upregulation of TEs resulting in a spermatogenic arrest; however,
not during the time of their joint expression but only several days and cell divisions
later when early pachytene spermatocytes are formed (Kuramochi-Miyagawa et al.,
2008). piRNAs expressed during this period are termed pre-pachytene piRNAs
and are characterised by a high occurrence of TEs in sense orientation, suggesting
that most primary pre-pachytene piRNAs derive directly from TE transcripts (Aravin
et al., 2008, 2007). These piRNAs associate with MILI and drive the production of
secondary piRNAs through recognition of antisense transcripts, which then associate
with either MILI or MIWI2. This amplification method results in fine-tuning of the
piRNA population and specific enrichment of those piRNAs that target the most
active TEs.
While MILI is found exclusively in the cytoplasm MIWI2 is predominantly found in
the nucleus (Aravin et al., 2009). Translocation of MIWI2 into the nucleus is however
dependent on its association with piRNAs and therefore on the slicer activity of MILI
that is necessary to load MIWI2 with secondary piRNAs (Aravin et al., 2008). In
contrast, the slicer activity of MIWI2 itself is dispensable for spermatogenesis, as cat-
alytically inactive MIWI2 mutants show normal fertility and only modest upregulation
of TEs (De Fazio et al., 2011). This strongly suggests that MIWI2’s essential function
during spermatogenesis defers from post-transcriptional gene silencing (PTGS) of
TE transcripts in the cytoplasm and is rather related to its nuclear function. Indeed,
MIWI2 is required for the establishment of de novo DNA methylation over TEs acting
upstream of DNMT3L in PGCs (Aravin et al., 2008).
De novo methylation was shown to occur in two waves and a defined set of TEs
manages to escape the first ‘generic’ wave that occurs by E16.5. Many of these
evading TEs (mainly long interspersed element 1 (LINE1s) and intracisternal particle
A (IAPs)) correspond to evolutionary young elements that become activated in Mili
24 Introduction
Figure 1.7 | Biogenesis and function of pre-pachytene piRNAs in the mouse
Pre-pachytene piRNAs are expressed during embryonic development from genomic
loci that often display bi-directional transcription. piRNA precursors are exported to
the cytoplasm and directed to the processing machinery. After 5’ ends processing
by PLD6, the piRNA intermediates associate with MILI followed by 3’ trimming and
methylation by Trimmer and HENMT1, respectively. This generates mature primary
piRNAs bound to MILI which can then engage in the ping-pong cycle leading to the
cleavage of complementary antisense transcripts and the production of secondary
piRNAs bound to either MILI or MIWI2. Secondary piRNAs are characterised
by a 10nt overlap in their 5’ ends with primary piRNAs as a result of MILI-directed
cleavage. Loading of MIWI2 with secondary piRNAs induces its translocation into the
nucleus where it functions in the transcriptional silencing of transposable elements
(TEs). This involves the recruitment of the de novo DNA methylation machinery
(DNMTs) as well as histone methyltransferases (HMTase) resulting in targeted DNA
methylation and H3K9me3-mediated heterochromatinisation of TE loci, such as
LINEs and IAPs. Figure from Ernst et al. (2017).
Introduction 25
knockout mice as a consequence of unloaded MIWI2 (Molaro et al., 2014). Inter-
estingly, a similar transcriptional activation as well as hypomethylation of evolutionary
young TEs was observed in mutants for DNMT3C, a recently identified fourth mem-
ber of the de novo DNA methyltransferase family in rodents, which is exclusively
expressed in the male germline (Barau et al., 2016). While the exact interplay be-
tween the piRNA pathway and DNMTs remains to be addressed, MIWI2 is believed
to use antisense piRNAs as guides to detect nascent TE transcripts in the nucleus
and recruit not only the de novo DNA methylation machinery (Itou et al., 2015),
but also histone methyltransferases to establish a repressive chromatin landscape
especially across LINE1 elements (Pezic et al., 2014).
Once MIWI2 expression ceases, MILI is the only PIWI protein expressed in
male germ cells with its own expression levels decreasing gradually to very low or
undetectable levels in preleptotene and leptotene spermatocytes until it increases
again during the zygotene to pachytene transition (Di Giacomo et al., 2013). Based
on the lack of detectable PIWI protein expression in preleptotene to late zygotene
spermatocytes, it is suggested that these cells do not have an active piRNA pathway.
This might be necessary to deplete the pool of pre-pachytene piRNAs from these
cells to free up MILI proteins to engage with the new class of piRNAs about to be
expressed (Di Giacomo et al., 2013).
1.5.3 Pachytene piRNAs and their role in TE regulation
As the first wave of spermatocytes reaches the pachytene stage of meiosis,
the transcription of the third PIWI family member MIWI starts together with a new
class of piRNAs termed pachytene piRNAs (Girard et al., 2006; Grivna et al., 2006).
Pachytene piRNAs are transcribed from large intergenic noncoding loci as well as 3’
untranslated regions of protein-coding gene loci (Robine et al., 2009; Vourekas et al.,
2012), which have no unifying genomic sequence or motif definition and are fast
evolving across mammals (Chirn et al., 2015). The majority of pachytene piRNAs
(>80%) map uniquely to the genome, thereby allowing unambiguous identification of
piRNA clusters (Girard et al., 2006). Pachytene piRNA clusters can be either uni-
or bidirectionally transcribed and although there is a fewer number of bidirectional
clusters annotated in the genome, they contribute the majority of pachytene piRNAs.
Transcription of these clusters is facilitated by the ancestral transcription factor A-
MYB and the resulting piRNA transcripts are 5’ capped and polyadenylated. However,
A-MYB is not specific to piRNA clusters as it also drives its own expression as well
as other members of the piRNA pathway in a positive feed-forward loop and thus
26 Introduction
Figure 1.8 | Multiple roles of action for pachytene piRNAs during mouse
spermatogenesis
Pachytene piRNA expression starts during prophase of the first meiotic division
when spermatocytes reach the pachytene stage. This expression is orchestrated by
the transcription factor A-MYB, which drives the expression of piRNA clusters and
other piRNA pathway related genes. Pachytene piRNAs are transcribed from distinct
clusters in the genome often carrying bidirectional promoters. piRNA precursors are
then processed in the same fashion as pre-pachytene piRNAs and mature pachytene
piRNAs associate with either MILI or MIWI.
Pachytene piRNAs engage in a myriad of functions throughout spermatogenesis,
including the post-transcriptional silencing of TE transcripts but also non-TE related
functions. After meiosis, round spermatids undergo extensive epigenetic remod-
elling which first involves a genome-wide derepression due to the incorporation of
histone variants. This is followed by association of transition proteins that lead to
a transcriptional shutdown and finally an almost complete replacement of histones
with protamines leading to chromatin condensation. During this differentiation pro-
cess, pachytene piRNAs regulate spermatogenic mRNAs and lncRNAs that become
transcribed due to the genome-wide derepression. Furthermore, MIWI is involved
in the piRNA-independent stabilisation of spermiogenic mRNAs to allow storage
and translation after the transcriptional shutdown. Towards later stages, pachytene
piRNAs direct global mRNA degradation in association with MIWI, which recruits
CAF1 to induce deadenylation resulting in mRNA decay. Figure from Ernst et al.
(2017).
Introduction 27
cannot be the sole determining factor that distinguishes piRNA cluster transcripts
from conventional mRNAs (Li et al., 2013).
In contrast to pre-pachytene piRNAs, pachytene piRNAs have a surprisingly low
repeat content (<20%) (Girard et al., 2006), that is in fact lower than the genome
average (The Mouse Genome Sequencing Consortium, 2002). However, it has been
demonstrated that MIWI’s slicer activity is necessary for the accurate silencing of
LINE1 elements in spermatocytes (Di Giacomo et al., 2013; Reuter et al., 2011).
Pachytene piRNA clusters have been shown to produce novel artificial piRNA
sequences upon introduction of transgenic sequences (Muerdter et al., 2012) sug-
gesting that active retrotransposition of TEs into already existing piRNA clusters
could be required for the piRNA pathway to adapt to new elements. However, in
contrast to observations in Drosophila that indeed show a higher TE insertion rate
into piRNA clusters (Ronsseray et al., 1991), no such bias has been described
in mammals so far and therefore appears rather ’time-ineffective’. It is therefore
still unclear how the pachytene piRNA system adapts to newly invading repetitive
elements and how quickly a new piRNA response can be installed.
1.5.4 Non-repeat related functions of pachytene piRNAs
MIWI remains actively transcribed until the round spermatid stage and is the only
PIWI protein detected in late round and elongating spermatids (Deng and Lin, 2002).
These late stages of spermatogenesis are associated with major epigenetic changes
and various functions have been implicated for non-repeat related pachytene piRNAs.
For example, the transitioning from histone- to protamine-bound DNA in round
spermatids requires open and accessible chromatin, which leads to a genome-
wide loosening of transcriptional control and the expression of spurious transcripts
including many long non-coding RNAs (lncRNAs) (Kimmins and Sassone-Corsi,
2005). Pachytene piRNAs have been shown to be essential in targeting many of
these mRNA and lncRNAs to regulate their expression via PTGS (Watanabe et al.,
2015) (Figure 1.8). Recent work cloning a human piRNA cluster into the mouse
and identifying mRNAs that are targeted by human ectopic piRNAs, shed light onto
the target specificity of this mechanism describing the requirement for a perfect
match between nucleotides 2-11 with a maximum of four mismatches between
nucleotides 12-21 (Goh et al., 2015). During later stages of spermatogenesis in
elongating spermatids, pachytene piRNAs were shown to direct the elimination of
mRNAs. This was independent of the slicer activity of MIWI but dependent on the
interaction with the deadenylase CAF1 that induces deadenylation and degradation
28 Introduction
of mRNAs (Gou et al., 2014). Further functional implications for MIWI were made
that are independent of associated piRNAs, such as binding to and stabilisation
of spermiogenic mRNAs that are stored for translation at later timepoints when
transcription has been switched off (Castañeda et al., 2014; Vourekas et al., 2012).
Studies on conditional knockouts of MILI after birth, therefore avoiding adverse
phenotypes from MILI’s prenatal role, have shown that MILI-bound piRNAs are not
necessary for the establishment of meiotic silencing across the X and Y chromosome
(Beyret and Lin, 2011). In accord with the transcriptional silencing of the X and
Y chromosome during the pachytene stage, no pachytene piRNA clusters are
annotated on these chromosomes in the mouse. However, profiling of the piRNA
repertoire in the marmoset identified several piRNA clusters on the X chromosome,
suggesting that these regions might escape meiotic silencing (Hirano et al., 2014)
1.6 The Tc1 mouse model
The Tc1 mouse model is an aneuploid transchromosomic mouse strain that
carries an almost complete copy of HsChr21 in addition to the mouse genome
(O’Doherty et al., 2005). The presence of HsChr21 alongside the orthologous
mouse genes that are spread mainly over mouse chromosome 10, 16 and 17 results
in an excess of these genes, creating a highly comparable situation to what is
observed in human Down syndrome (DS) caused by a trisomy of HsChr21 (Reeves
et al., 2001).
The Tc1 mouse model was generated by introducing HsChr21 into female mouse
ESCs using irradiation microcell-mediated chromosome transfer (XMMCT) followed
by injection of these ESC into host blastocysts. The resulting chimeras were bred
with C57BL/6J mice to achieve germline transmission of HsChr21. One female
chimera produced Tc1-positive offspring that were then bred to expand the line of
mice carrying the freely segregating copy of HsChr21. Detailed analysis of the human
DNA content in these mice showed that at least 42 Mb, including ∼92% of all known
genes encoded on HsChr21 were retained, which corresponds to about 90% of the
entire 46.9 Mb of HsChr21 (O’Doherty et al., 2005). Deep-sequencing of HsChr21
in the Tc1 mouse later defined deleted and duplicated regions and generated a
map for the various rearrangements that occurred in response to irradiation (Gribble
et al., 2013). While the Tc1 mouse serves as an excellent model to study DS, it
has also been a valuable tool for studying the mechanisms of species-specific gene
regulation as well as mechanisms of aneuploidy in a broader sense.
Introduction 29
Aneuploidy is characterised by an abnormal number of chromosomes within a
cell, but does not include overall changes in ploidy. In humans, aneuploidy affects
∼5% of all pregnancies and is the most prevalent cause of spontaneous abortions
(Hassold and Hunt, 2001). The most common aneuploidies in human affect the
sex chromosomes, such as Klinefelter syndrome (XXY, males with additional X)
and Turner syndrome (X, females with monosomy of X), which usually result from
non-disjunction during male meiosis, whereas autosomal aneuploidies, such as
trisomy 21, usually originate maternally (Heard and Turner, 2011).
In contrast to the majority of mouse models to study aneuploidy, the Tc1 mouse
contains an exogenous chromosome, whereas most commonly used trisomic mouse
models carry extra copies of segments of mouse chromosomes. These include the
Ts65DN and Ts1Cje mouse which both carry an extra copy of a segment of mouse
chromosome 16 and serve as models for human Down syndrome (Reeves et al.,
1995; Sago et al., 1998). However, a major drawback of these mouse models is
that they exhibit complete male sterility and therefore prevent a careful dissection of
meiotic events in the presence of aneuploidy in males (Sheppard et al., 2012).
1.6.1 Previous findings in the Tc1 mouse
The first inter-species studies, looking at the conservation of transcription factor
binding showed a strong and rapid divergence across species, indicating that DNA
sequence determines the binding of TFs in a cis-directed manner (Borneman et al.,
2007; Moses et al., 2006; Odom et al., 2007). However, the interpretation of these
results was slightly ambiguous as it could not be excluded that the observed differ-
ence in TF binding were not caused by environmental factors such as differential
interactions with cofactors that are influenced by a given nuclear environment.
Using the Tc1 mouse model to perform inter-species comparisons solved this
ambiguity, as the genetic material to be compared exists in the same nuclear
environment, thereby eliminating all technical and environmental variables. The
study by Wilson et al. (2008) compared the binding of three hepatocyte-specific
TFs - HNF1a, HNF4a and HNF6 - across HsChr21 in humans, Tc1 mice and the
orthologous regions on mouse chromosome 16. This comparison showed that
only between a third to half of all TF binding events were shared between human,
confirming the low level of conservation between human and mouse. The remainder
of binding sites were detected in a cis-directed manner, being only observed on
HsChr21 in human and Tc1 mouse, but not at orthologous regions on the mouse
chromosome.
30 Introduction
Figure 1.9 | Species-specific
transcriptional regulation in
the Tc1 mouse
Schematic representation of
shared, cis-directed and trans-
directed transcription factor
binding events detected in
species comparisons.
These results showed that the DNA sequence is the main determinant of gene
expression by regulating the binding of TFs in a cis-directed manner. In contrast to
that, environmental factors, such as the nuclear environment and the availability and
concentration of TFs and other cofactors, play only a minor role.
However, this study was performed using genome-tiling microarrays, which
excluded repetitive regions by design and only allowed the analysis of the non-
repetitive portion of the genome. Further studies, using high-throughput sequencing
instead of microarrays were able to include repetitive elements into the analysis
and found that many repeat regions on HsChr21 showed transcriptional activation
specific to the Tc1 mouse (Ward et al., 2013). Using H3K4me3 enrichment as a proxy
for transcriptional activation revealed 118 sites that were specifically enriched for this
mark in Tc1 mice compared to humans, whereas only 50 sites showed a human
bias for H3K4me3-enrichment. Detailed analysis of these regions showed that
Tc1-specific sites are enriched for repeat elements at their peak summit compared
to regions that share the H3K4me3 signal between human and Tc1 mouse. Most of
the repeat elements associated with Tc1-specific sites were transposable elements,
such as LTR retrotransposons, LINE and SINE elements. Over one third of all
Tc1-specific sites were associated with a repeat and detailed analysis of repeat
lineage revealed that most of these repeats were significantly younger and belonged
to the human- or primate-specific lineage. These results indicate that the mouse
regulatory machinery might not be able to accurately silence young repeat elements
of the primate- and human-specific lineage, as these elements emerged after the
divergence of human and mouse.
2 | Materials and Methods
2.1 Mouse colony management
Standard husbandry procedures including weaning and biopsies were performed
by the Biological Resources Unit staff at CRUK CI.
The Tc1 mouse strain was obtained from Dr. E. Fisher and Dr. V. Tybulewizc
(O’Doherty et al., 2005) and housed in the Biological Resources Unit (BRU) in
the Cancer Research UK - Cambridge Institute (CRUK CI) under the Home Office
Licence (PPL 70/7535). Colony maintenance is depicted in Figure 2.1. The genetic
background of the Tc1 mouse is based on the 129S8 mouse strain which was
maintained by sister-brother breedings. Female 129S8 mice were crossed with
male C57BL/6J mice to generate (129S8xC57BL/6J)F1 mice which were used to
cross to Tc1 animals. For maintenance of the line, Tc1 females were crossed with
(129S8xC57BL/6J)F1 males, transmitting HsChr21 through the female germline. For
male germline transmission, female (129S8xC57BL/6J)F1 mice were crossed with
Tc1-positive males. Tc1-negative littermates that did not inherit HsChr21 were used
as control animals and are referred to as Tc0. Genotyping was performed on ear
biopsies using primers designed against the SOD1 gene on HsChr21 (Table 2.1).
2.2 Tissue preparation
Tissues were harvested from male mice at an age between 8-14 weeks unless
specific timepoints were required. Mice were sacrificed by a schedule 1 method of
euthanasia and perfused with phosphate buffered saline (PBS) via the portal vein.
Unless processed directly for experiments, tissues were flash-frozen in liquid nitrogen
for RNA and DNA extractions, fixed with 4% formalin for histology or cross-linked for
chromatin immunoprecipitation followed by high-throughput sequencing (ChIP-Seq)
experiments. Cross-linking was performed for 20 minutes at room temperature (RT)
in crosslinking solution (1% formaldehyde, 50mM Hepes-KOH, 100mM NaCl, 1mM
EDTA and 0.5mM EGTA). The reaction was quenched with 1/20th volume of 2.5M
glycine for 10 minutes, washed with PBS and then flash-frozen.
32 Materials and Methods
Figure 2.1 | Tc1 breeding colony
Schematic representation of the breeding colony to maintain the Tc1 mouse strain.
Refer to text for details.
Primer Sequence
SOD1 forward CAGTAACTGAGAGTTTACCCTTTGGT
SOD1 reverse CACACTAATGCTCTGGGAAGAAAGA
SOD1 reporter CTGCACCTGATTTCAG
Table 2.1 | Primer sequences for genotyping
Materials and Methods 33
Mature sperm was obtained from male mice aged between 16-32 weeks accord-
ing to Hisano et al. (2013). In brief, caudal epididymides were dissected with several
small incisions and placed in 4 ml Donners medium containing 25mM NaHCO3,
20mg/ml BSA, 1mM sodium pyruvate and 0.53% (vol/vol) sodium DL-lactate. Epi-
didymi were incubated at 37◦C for 1 hour to allow sperm to swim-up, then the
upper 2.5ml of Donners medium were collected and sperm was counted using a
hemocytometer.
2.3 Histology
Standard histological stains and immunohistochemistry (IHC) were performed by
the Histopathology core at CRUK CI.
Tissues were fixed in neutral buffered formalin (NBF) for 24 hours, transferred
to 70% ethanol, machine processed and paraffin embedded. All formalin-fixed
paraffin-embedded (FFPE) sections used for histological stains and IHC were 3µm
in thickness. De-waxing and re-hydration prior to stainings and post-staining de-
hydration and clearing were performed on the automated Leica ST5020, mounting
was performed on the Leica CV5030.
Hematoxylin and eosin (H&E) stainings were entirely performed on the automated
Leica ST5020. In brief, for dewaxing and re-hydration, slides were placed in xylene
for 2 x 10 minutes followed by 100% ethanol for 2 x 5 minutes, 70% ethanol for 5 min
and rinsing in water. Slides were stained in Mayers Haemalum for 2 minutes, rinsed
in water for 5 minutes and then placed in 1% aqueous eosin for 30 seconds. After
washing, slides were dehydrated by dipping in 70% ethanol followed by incubation in
100% ethanol for 10 and 20 seconds and 2 x 2 minutes in xylene and then mounted
in DPX (Sigma, 06522).
For Periodic Acid Schiff (PAS) stainings slides were dewaxed, washed in water
and placed in 0.5% Periodic Acid (Sigma P0430) for 5 minutes. After three washes
in ultra-pure water, slides were placed in Schiff reagent (Thermo Fisher Scientific,
J/7300/PB08) for 15-30 minutes in a closed container and washed again three
times in ultra pure water. Counterstain was performed using Mayers Haematoxylin
(Thermo Fisher Scientific, LAMB/170-D) for 40 seconds. Slides were then rinsed in
tap water, dehydrated and mounted.
IHC was performed on FFPE sections using the BondTM Polymer Refine Kit
(DS9800, Leica Microsystems) on the automated Bond platform. Antibodies against
34 Materials and Methods
phospho-Histone H2A.X (Ser139) (Millipore, MABE205, 1:5000 dilution) and cleaved
Caspase-3 (CC3) (Cell Signaling Technology, 9664, 1:200 dilution) were used with
DAB Enhancer (Leica Microsystems, AR9432). Heat-induced epitope retrieval was
performed for 20 min at 100◦C on the Bond platform with sodium citrate (for H2AFX)
and Tris-EDTA (for CC3). Slides were scanned using Aperio XT (Leica Biosystems)
and CC3 quantification was performed using the Aperio eSlide Manager (Leica
Biosystems).
2.4 Classification of seminiferous tubules
Histological assessment was performed with support from Dr. Sarah Aitken.
H&E stained testis cross-sections were scored by two blinded individuals using
the Aperio eSlide Manager (Leica Biosystems) to assess the overall subfertility phe-
notype. Seminiferous tubules were scored according to the predominant histological
pattern of spermatogenesis in each individual tubule based on Creasy et al. (2012)
and Dohle et al. (2012). Grade I: normal spermatogenesis; Grade II: mild hypo-
spermatogenesis (all germ cell stages present but visible meiotic disruption and
suboptimal frequency of spermatozoa); Grade III: severe hypo-spermatogenesis (all
germ cell stages present including occasional spermatozoa); Grade IV: maturation
arrest (incomplete spermatogenesis, no cell types beyond the spermatocyte stage).
Classification of seminiferous tubules according to epithelial stage of spermatoge-
nesis was performed on PAS stained tissue sections. Different stages were identified
as described in the binary decision key by Meistrich and Hess (2013). Stage I-III:
Two generations of spermatids but no acrosome cap over the nucleus of round
spermatids; Stage IV-VI: Two generations of spermatids and acrosomic system
forming a cap over the nucleus of round spermatids; Stage VII-VIII: Two generations
of spermatids with elongated spermatids lining the lumen; Stage IX-XI: Only one
generation of spermatids but no visible meiotic figures or secondary spermatocytes;
Stage XII: Only one generation of spermatids and visible meiotic figures as well as
secondary spermatocytes.
2.4.1 Immunofluorescence staining of seminiferous tubules
FFPE sections (standard 3µm sections for widefield microscopy and thicker 10µm
sections for confocal microscopy) were de-waxed on the automated Leica ST5020
Materials and Methods 35
and antigen retrieval was performed by boiling the slides for 10 minutes in 10mM
sodium citrate containing 0.05% Tween-20. Slides were cooled down to RT for 30
minutes and then permeabilised in PBS containing 0.3% Triton-X100 for 30 minutes
followed by blocking with 5% BSA in PBS containing 0.3% Triton-X100 for 1 hour.
Primary antibodies were diluted in 5% BSA in PBS containing 0.1% Tween-20 (PBS-
T) and incubated with tissue for 2-3 hours at 37◦C in a humidifying chamber (see
Table 2.2 for antibodies and dilutions). Slides were washed in PBS-T followed by
incubation with secondary antibodies (Table 2.2) for 1 hour at RT.
Slides were washed in PBS, stained with Hoechst (1µg/ml) and mounted in
ProLong Diamond Antifade Mountant (ThermoFisher). Widefield microscopy of
tissue sections was performed using a Zeiss Axio Observer Z1 with a Pl APO 0.8
NA 20X dry objective (Carl Zeiss Microscopy) fitted with a CoolLED PE-4000 LED
light-source and Zeiss Axiocam 506 camera. A 2D tile-scan across the entire tissue
section was performed with 10% tile overlap. Voxel size was 0.23µm.
Confocal microscopy was performed using a Leica TCS SP8 STED 3X micro-
scope with an HC PL APO CS2 1.4NA 100x oil objective (Leica Microsystems). A
405nm diode laser was used to excite Hoechst at 405nm and a white light pulsed
laser (SuperK EXTREME, NKT Photonic) was used to excite the secondary antibody
fluorophores. Voxel size was 0.07µm and a Z-stack was acquired through the sample
with 0.3µm spacing. Each channel was acquired sequentially. Deconvolution was
performed post-acquisition using Huygens Professional (Scientific Volume Imaging,
Version 15.10.1p2).
2.4.2 Interactive learning using ilastik
Machine-learning based cell type classification was performed by Dr. Jeremy
Pike from the Light microscopy core at CRUK CI.
The open source interactive learning toolkit ilastik was used to segment and
classify cells in the tissue section stained for phospho-Histone H3 (Ser10) (pH3)
and α-tubulin (Sommer et al., 2011). Classification was done in a two-stage pro-
cess: pixel classification followed by object classification. Pixel classification was
performed to segment the nuclear regions, and object classification to split each
nucleus into one of five classes: germinal epithelium, primary spermatocytes, mei-
otic, round spermatids and elongating spermatids. The training data contained two
36 Materials and Methods
Antigen Species Dilution Catalog number Manufacturer
phospho-Histone
H3 (Ser10) Rabbit 1000 06-570 Millipore
phospho-Histone
H2A.X (Ser139) Mouse 500 MABE205 Millipore
α-tubulin Mouse 500 T9026 Sigma
MIWI Rabbit 1:500 G82 Cell Signaling
MILI Mouse 10µl of custom antibody from Ramesh Pillai
rabbit IgG
Alexa Fluor 488 Donkey 1:1000 A-21206 ThermoFisher
mouse IgG
Alexa Fluor 555 Donkey 1:1000 A-31570 ThermoFisher
Table 2.2 | Antibodies used for immunofluorescence stainings
cropped images from each tissue section and image annotation was done blindly
with respect to condition. Each tile was classified independently and Matlab (2015b,
MathWorks) was used to process the results and remove duplicated objects.
The subset of cells identified as meiotic by the ilastik toolbox were manually
classified as either pro-metaphase, metaphase, phenotypic metaphase, anaphase
or non-mitotic. A randomly selected subset of 100 meiotic cells were analysed from
each section in a randomised order to eliminate user bias and facilitate blind analysis.
This was performed using Matlab with a bridge to ImageJ for visualisation (Sage
et al., 2012).
2.5 Fluorescent-activated cell sorting (FACS) of
spermatogenic cell populations
Spermatogenic cells were isolated from adult mouse testes (16-36 weeks old)
as described in Goh et al. (2015) with minor modifications. One testis from wild-
type males or both testes from Tc1 males were used per experiment. In brief, the
albuginea was removed and tissue was digested in dissociation buffer (25 mg/ml
Collagenase A, 25mg/ml Dispase II and 2.5 mg/ml DNaseI) for 30 min at 37◦C.
Enzymatic digestion was quenched with Dulbecco’s Modified Eagle Medium (DMEM)
(GibcoTM) supplemented with 10% Fetal calf serum (FCS) (10270106, GibcoTM).
Materials and Methods 37
Cells were resuspended at a concentration of 1 million cells per ml and stained
with Hoechst 33342 (H3570, ThermoFisher) at a final concentration of 5µg/ml for
45 minutes at 37◦C. Cells were resuspended in PBS containing 1% FCS and 2mM
EDTA for sorting and propidium iodide was added to a final concentration of 1µg/ml
for dead cell exclusion.
Cells were sorted on an Aria IIu cell sorter (Becton Dickinson) using a 100µm
nozzle. Hoechst was excited with a UV laser at 355nm and fluorescence was
recorded with a 424/44 filter (Hoechst blue) and 675LP filter (Hoechst red). This
allows distinction of 5 different cell populations: spermatogonial stem cells (2N
with low Hoechst Red emission), cells in S phase (2N-4N DNA content), primary
spermatocytes (4N), secondary spermatocytes (2N with high Hoechst Red emission)
and spermatids (1N). Cells were collected into Trizol LS for bulk RNA extractions or
PBS containing 1% FCS and 2mM EDTA for single-cell RNA-Seq and stainings.
2.6 Fluorescence in situ hybridisation (FISH)
Fluorescence in situ hybridisation (FISH) stainings on tissue sections and sorted
spermatogenic cell populations were performed by Dr. Julia Jones from the Histo-
pathology core at CRUK CI.
Sequential FISH for mouse chromosome X and Y and HsChr21
on testis tissue sections
Tissue sections of 3µm thickness were collected on Superfrost slides (Ther-
moFisher Scientific) and dried at 60◦C for 1 hour. Dewaxing was performed in xylene
for 2 x 5 minutes and rehydration through graded alcohol series to MilliQ water was
performed for 5 minutes each in 2 x 100% EtOH, 1 x 70%, 1 x MilliQ water. Heat
pre-treatment was performed using the Cytocell pre-treatment kit at 98◦C for 10 min-
utes followed by two washes in MilliQ water for 3 minutes. Enzymatic pre-treatment
was performed at 37◦C for 5 minutes followed by two washes in MilliQ water for 3
minutes. Dehydration was performed through graded ethanol series for 2 minutes
each at 70%, 95% and 100% EtOH and then air dried. 20µl of ready-to-use probe
mix of whole chromosome paint for mouse X and Y (1:1 mixture, both FITC-labelled,
Cytocell, AMP0XG and AMP0YG) were applied, sealed with coverslip, denatured
at 60◦C for 2 minutes and hybridised overnight at 37◦C. Coverslip was removed
in 2X saline sodium citrate (SSC) + 0.05% Tween-20, then slides were washed
38 Materials and Methods
in 0.4X SSC at 72◦C for 2 minutes followed by 2X SSC + 0.05% Tween-20 for 30
seconds and mounted in ProLong Gold Antifade Mountant with DAPI (ThermoFisher
Scientific). Slides were imaged on Axio Scan.Z1 (Carl Zeiss Microscopy) with a
scanning resolution of 40X and 30ms exposure time for DAPI and 100ms exposure
time for FITC. Coverslip was removed, mountant was washed off in 2X SSC, slides
dehydrated through graded ethanol series as described above and 10µl of ready-to-
use HsChr21-specific probe XA 21q22 (Metasystems, D-5601-100-OR) was added.
Samples were sealed with coverslip, denatured at 80◦C for 2 minutes and hybridised
overnight at 37◦C. Post-hybridisation procedure was performed as described above
and slides were re-imaged on the Axio Scan.Z1 using exposure times of 13ms for
DAPI and 600ms for Cy3.
FISH for HsChr21 on sorted spermatogenic cell populations
Spermatogenic cell populations obtained from fluorescence-activated cell sorting
(FACS) were spun onto SuperfrostTM Plus slides (ThermoFisher Scientific) using a
Cytospin Cytocentrifuge (ThermoFisher Scientific) (∼ 50.000 cells per slide) at 1000
rpm for 3 minutes and fixed in Carnoy’s Fixative (3:1 ratio of methanol:glacial acetic
acid) for 30 minutes.
FISH for genotyping purposes was performed using the human chromosome 21
locus-specific probe XA 21q22 (Metasystems, D-5601-100-OR). In brief, the slides
were treated with 0.01M HCl + 0.5µg/ml pepsin at 37◦C for 10 minutes, washed in
water and dehydrated in 70% and 10% ethanol. 10µl of ready-to-use probe mix
were applied, sealed with coverslip and denatured at 80◦C for 2 minutes followed
by hybridisation for 16 hours at 37◦C. Coverslips were removed in 2X SSC buffer
+ 0.05% Tween-20 then slides were washed in 0.4X SSC for 2 minutes at 72◦C,
rinsed again in 2X SSC + 0.05% Tween-20 and mounted in ProLong Gold Antifade
Mountant with DAPI (ThermoFisher Scientific).
Widefield microscopy was performed using a Zeiss Axio Observer Z1 with a Pl
APO 0.8 NA 20X dry objective (Carl Zeiss Microscopy) fitted with a CoolLED PE-4000
LED light-source and Zeiss Axiocam 506 camera. A 2D tile-scan was performed with
10% tile overlap to obtain large enough cell numbers for quantification. Voxel size
was 0.23µm. The percentage of cells containing HsChr21 was quantified using FIJI
(Schindelin et al., 2012) (Code A.1). Representative high-resolution images were
captured using a Nikon TE-2000 inverted microscope with NIS-elements software
using a Plan Apochromat x100 objective and Andor Neo 5.5 sCMOS camera.
Materials and Methods 39
FISH for HsChr21 in epididymal sperm
FISH in sperm was performed as described in Hill et al. (2003) with minor modi-
fications. In brief, caudal epididymides were removed, nicked in several locations
while leaving the overall tissue structure intact and placed in 300µl 2.2% sodium
citrate at 32◦C for 5 minutes to allow sperm to swim into solution. 7µl of supernatant
were placed on SuperfrostTM Plus slides and air-dried overnight.
Sperm were then fixed with Carnoy’s fixative for 10 minutes followed by treatment
with 10mM DTT for 30 minutes on ice. Slides were rinsed in water at RT and air
dried before denaturation in 70% formamide in 2X SSC for 8 minutes at 78◦C. Slides
were then dehydrated in an ice-cold ethanol series of 70%, 85% and 100% for 2
minutes each. 10µl of ready-to-use HsChr21-specific probe XA 21q22 (Metasystems,
D-5601-100-OR) were applied, sealed with coverslip and denatured on hot plate at
80◦C for 2 minutes followed by hybridisation overnight at 37◦C. Post-hybridisation
washes were as follows: 3 washes in 50% formamide in 2X SSC, one wash in 2X
SSC and one wash in PN buffer (0.1M NaH2PO4/0.1M Na2HPO4 + 0.1% NP-40)
for 5 minutes each at 45◦C, followed by one wash in PN buffer at RT. Nuclei were
then stained with Hoechst 33342 (H3570, ThermoFisher) at a final concentration of
10µg/ml in PBS for 15 minutes at RT and mounted in ProLongTM Diamond Antifade
Mountant (ThermoFisher).
Imaging was performed on a Nikon TE-2000 inverted microscope with NIS-
elements software using a Plan Apochromat x63 objective and Andor Neo 5.5
sCMOS camera.
2.7 Chromatin immunoprecipitation followed by high-
throughput sequencing (ChIP-Seq)
Pre-block and antibody binding
Dynabeads® Protein G (ThermoFisher Scientific) were washed three times in
blocking solution containing 0.5% bovine serum albumin (BSA) in PBS and then
incubated with 5 - 10µg of antibody (see Table 2.3) in a volume of 250µl for at least
4 hours or overnight at 4◦C. Excess antibody was removed by washing three times
with blocking solution and antibody-bead mixture was resuspended in 100µl blocking
solution.
40 Materials and Methods
Cell lysis and sonication
Frozen cross-linked tissue was thawed on ice and homogenised in ice-cold PBS
using a dounce tissue grinder. Dounced tissue was then filtered through a 100µm
cell strainer and washed twice with ice-cold PBS, pelleting cells at 2500 g at 4◦C for
3 minutes inbetween. Cells were then lysed in three sequential lysis buffers (LBs)
containing 1X cOmpleteTM EDTA-free protease inhibitor cocktail (Roche). First lysis
was performed in 10mL LB1 (50mM Hepes-KOH, pH 7.5; 140mM NaCl; 1mM EDTA;
10% glycerol; 0.5% NP-40; 0.25% Triton-X100) for 10 minutes at 4◦C on a rotating
wheel and then pelleted by centrifugation.
Antigen Species Amount Catalog number Manufacturer
H3K4me3
mouse
monoclonal
5µg 05-1339 Millipore
H3K27ac
rabbit
polyclonal
10µg ab4729 Abcam
H3K36me3
rabbit
polyclonal
10µg 61101 Active Motif
RNA Pol II
mouse
monoclonal
10µg ab5408 Abcam
CEBPα
goat
polyclonal
10µg sc-9314
Santa Cruz
Biotechnology
HNF4α
rabbit
polyclonal
10µg ARP31946
Aviva Systems
Biology
CTCF
rabbit
polyclonal
10µg 07-729 Millipore
Rad21
rabbit
polyclonal
10µg ab992 Abcam
NF-YA
mouse
monoclonal
10µg sc-17753X
Santa Cruz
Biotechnology
NF-YB
rabbit
polyclonal
10µg C15410241 Diagenode
Table 2.3 | Antibodies used for chromatin immunoprecipitation
Materials and Methods 41
Supernatant was discarded and pellet was resuspended in 10mL LB2 (10mM Tris-
HCl pH 8.0; 200mM NaCl; 1mM EDTA; 0.5mM EGTA) and incubated for 5 minutes
at 4◦C on a rotating wheel followed by centrifugation. Pellet was then resuspended
in 3 mL LB3 (10mM Tris-HCl, pH 8.0; 100mM NaCl; 1mM EDTA; 0.5mM EGTA; 0.1%
Sodium-Deoxycholate; 0.5% N-lauroylsarcosine) and DNA was fragmented using a
Misonix Sonicator 3000 Homogeniser in an ice-water bath. 12 or 15 cycles of 30
seconds pulsing followed by 60 seconds of rest at a power output of 27-33 Watts
were used for liver or testis samples, respectively. 1/10th volume of 10% Triton X-100
was added to the lysate and debris was pelleted by centrifugation at 20,000 g for 10
minutes at 4◦C. An aliquot of 50µl of cell lysate was taken as input DNA. Cell lysate
obtained from one mouse liver was enough for 2 transcription factor or 3 histone
ChIPs. Cell lysate from two wildtype testes was used for one histone ChIP, whereas
testes from two Tc1 animals had to be pooled to obtain enough material.
Pre-clearing of lysates was performed for testis samples by incubation with 50µl
of Dynabeads® Protein G beads for 2 hours at 4◦C to remove unspecific binding.
Chromatin immunoprecipitation
Antibody-bead solution was added to cell lysate and incubated overnight at 4◦C
on a rotating wheel. Beads were washed five times with RIPA buffer (50mM KOH,
pH 7.6; 500mM LiCl; 1% NP-40; 0.7% Sodium-deoxycholate) and one time with tris-
buffered saline (TBS) (20mM Tris-HCl, pH 7.6; 150mM NaCl) at 4◦C, resuspending
beads inbetween washes on rotating wheel. Remaining wash buffer was removed by
centrifugation at 960 g for 3 minutes at 4◦C and beads were resuspended in 200µl
elution buffer (EB) (50mM Tris-HCl, pH 8.0; 10mM EDTA; 0.5% sodium dodecyl
sulfate (SDS)).
DNA elution and purification
Input DNA was diluted with 150µl EB and protein-DNA cross-link of input and
ChIP samples was reversed by incubation at 65◦C for 6 hours with vortexing every
5 minutes for the first 15 minutes. Beads were pelleted by centrifugation at 16,000
g for 1 minute and supernatant containing DNA was transferred and diluted with
200µl TE buffer. RNase treatment was performed with 0.02µg/µl RNaseA (Ambion)
for 30 minutes at 37◦C followed by treatment with 0.2µg/µl proteinase K (Invitrogen)
for 1-2 hours at 55◦C. DNA was then purified by phenol-chloroform extraction using
Phase Lock Gel (QuantaBio). An equal volume of phenol:chloroform:isoamyl alcohol
42 Materials and Methods
(P:C:IA) was added and phases were separated by centrifugation at 16,000 g for 5
minutes. Aqueous phase was transferred to a new tube and DNA was precipitated
by adding NaCl to a final concentration of 200mM, 0.5µg GlycoBlue (Ambion) and
800µl 100% ethanol (EtOH) followed by incubation at -80◦C for at least one hour.
DNA was pelleted by centrifugation at 20,000 g for 10 minutes at 4◦C, washed twice
with 500µl 80% EtOH, air-dried and resuspended in 10mM Tris-HCl, pH 8.0.
ChIP samples were quantified using the Qubit® double-stranded DNA high
sensitivity (HS) assay and input samples were quantified using a NanoDropTM
Spectrophotometer.
ChIP library preparation
Next-generation seqeuncing (NGS) libraries were prepared using either the
TruSeq ChIP Library Preparation Kit from Illumina® or the ThruPLEX® DNA-Seq Kit
from Rubicon Genomics.
TruSeq ChIP Library Preparation
A maximum of 50ng of ChIP DNA or input DNA were used in a total volume of
30µl and mixed with 40µl of End-repair mix and 30µl resuspension buffer (RSB),
followed by incubation at 30◦C for 30 minutes.
Agencourt® AMPure XP Bead clean-up was performed in a 1:1.6 ratio as follows:
1.6 volumes of AMPure beads were added to end-repair reaction and mixed by
pipetting. Beads were incubated with DNA for 15 minutes at RT and then placed
on magnetic stand for at least 5 minutes. Supernatant was removed and beads
were washed twice with 200µl fresh 80% EtOH for 30 seconds while remaining on
the magnet. Beads were air-dried for 10 minutes at RT to remove residual EtOH,
removed from magnet and resuspended in 20µl RSB for 2 minutes. Beads were
collected on magnet and 17.5µl of supernatant were transferred to a fresh tube.
12.5µl of A-tailing mix were added and incubated at 30◦C for 30 minutes. Adapter
Ligation was performed by adding 2.5µl ligation mix, 2.5µl RSB and 2.5µl of TruSeq
Adapters and incubation at 30◦C for 10 minutes. 5µl Stop Ligation Buffer were added
and excess adapters were removed using AMPure beads in a 1:1 ratio as described
above, eluting in 22.5µl RSB. 20µl of adapter-ligated DNA were then amplified via
polymerase chain reaction (PCR) by addition of 5µl of PCR Primer Cocktail and
25µl 2X KAPA HiFi Ready Mix (KAPA Biosystems) using the following programme:
Materials and Methods 43
30 seconds at 98◦C; 15 cycles of 10 seconds at 98◦C, 30 seconds at 60◦C and 30
seconds at 72◦C and final extension for 5 minutes at 72◦C.
PCR product was then double-size selected using AMPure beads to obtain
fragments between 200-400bp as follows: standard AMPure bead clean-up in
a 1:0.85 ratio, resuspending in 52.5µl and transferring 50µl to mix with AMPure
beads in a 1:0.75 ratio to bind fragments larger than 400bp. After binding to beads,
supernatant containing smaller fragments is transferred and mixed with AMPure
beads in a 1:1.2 ratio and cleaned up as previously described.
ThruPLEX DNA-Seq Library Preparation
ThruPLEX Library preparation was performed according to manufacturer’s in-
structions. In brief, a maximum of 50ng of ChIP DNA or input DNA were used in
a total volume of 10µl. 3µl of Template Preparation mix (2µl Template Preparation
Buffer and 1µl Template Preparation enzyme) were added followed by incubation
at 22◦C for 25 minutes and 55◦C for 20 minutes. 2µl of Library Synthesis mix (1µl
Library Synthesis buffer and 1µl Library Synthesis enzyme) were added followed by
incubation at 22◦C for 40 minutes. For adaptor ligation and library amplification, 5µl
of indexing reagent and 30µl of library amplification mix (25µl Library Amplification
buffer, 1µl Library Amplification enzyme, 4µl nuclease-free water) were added and
incubated as follows: 3 minutes at 72◦C, 2 minutes at 85◦C, 2 minutes at 98◦C
followed by 4 cycles of 20 seconds at 98◦C, 20 seconds at 67◦C and 40 seconds at
72◦C to add index primers . Library amplification was performed for 5 to 16 cycles of
20 seconds at 98◦C and 50 seconds at 72◦C depending on the amount of starting
material. PCR-amplified libraries were then quality assessed on a Bioanalyser DNA
High Sensitivity chip (Agilent) and purified using AMPure beads either in a 1:1 ra-
tio or performing double-size selection (as described above) if the average library
size was above 400bp. Quantification was then performed using the KAPA Library
Quantification Kit for Illumina sequencing platforms (Kapa Biosystems) according to
manufacturer’s instructions.
Next-generation sequencing
Sequencing was performed by the Genomics core facility at CRUK CI
ChIP-Seq libraries were sequenced on an Illumina HiSeq2500 according to
manufacturer’s instructions using a single-end 50bp run.
44 Materials and Methods
2.8 Strand-specific total RNA-Seq
RNA isolation
Total RNA was extracted from flash-frozen tissues using QIAzol® Lysis Reagent
(Qiagen) according to manufacturer’s instructions. 700µl of QIAzol Lysis Reagent
were added to 20-25mg of tissue and homogenised in Precellys Tissue Homogeniser
for 2 times 10 seconds with a 10 second break inbetween. Samples were incubated
at RT for 5 minutes then 140µl chloroform were added and mixed by shaking
vigorously. Phases were separated by centrifugation at 16,000 g for 15 minutes and
aqueous phase was transferred to a new tube. RNA was precipitated by adding 1
volume of isopropanol and incubation at RT for 10 minutes. RNA was pelleted at
16,000 g for 15 minutes at 4◦C, washed twice with 750µl 80% EtOH, air-dried and
resuspended in RNase-free water. RNA integrity was assessed on a Bioanalyser
Total RNA Nano ChIP.
DNase treatment
DNase treatment was performed on 5µg of total RNA using the TURBO DNA-free
Kit (Ambion) according to manufacturer’s instructions in a 15µl volume. Reaction
was incubated at 37◦C for 30 minutes, then 2µl of Stop Reaction was added. Sample
was spun down for 2 minutes at 10,000 g and 15 µl supernatant was transferred to a
new tube.
rRNA removal
rRNA removal was performed on 15µl DNase-treated RNA using the RiboZero
Gold Kit (formerly Epicentre, now Illumina) according to manufacturer’s instructions.
In brief, 225µl of magnetic beads were washed twice with RNase-free water and
resuspended in 65µl magnetic bead resuspension solution and 1µl of RiboGuard
RNase Inhibitor was added. 15µl of RNA were mixed with RiboZero rRNA removal
mix (4µl RiboZero buffer, 10µl RiboZero Removal Solution and 11µl RNase-free
water) and incubated at 68◦C for 10 minutes followed by 15 minutes at RT. Treated
RNA was then added to magnetic beads and mixed immediately by pipetting. After 5
minute incubation at RT, samples were vortexed and incubated for another 5 minutes
at 50◦C. Beads were collected on magnetic stand and supernatant was transferred
to a new tube. For purification of RNA, volume was topped up to 180µl and 18µl of
Materials and Methods 45
3M sodium acetate, 2µl glycogen and 600µl of ice-cold 100% EtOH were added and
incubated at -20◦C for at least one hour. RNA was spun down at 10,000 g for 30
minutes at 4◦C and washed twice with 600µl 70% EtOH. Pellet was air-dried and
resuspended in 19.5µl elute-prime frag solution for enzymatic fragmentation when
proceeding immediately to cDNA synthesis.
cDNA synthesis
RNA resuspended in elute-prime frag solution was incubated at 94◦C for 8
minutes for fragmentation. First strand cDNA synthesis was performed by adding
7µl First strand mix and 1µl SuperScriptTM II Reverse Transcriptase (ThermoFisher
Scientific) followed by incubation for 10 minutes at 25◦C, 50 minutes at 42◦C and 15
minutes at 70◦C. The reaction was cleaned up using the Qiagen PCR purification
kit according to manufacturer’s instructions by adding 5 volumes of PB buffer and
applying mixture to the spin column. DNA was washed twice with 750µl PE buffer
and eluted in 50µl preheated EB. Second strand synthesis was performed by adding
4µl 5X First strand buffer, 2µl 10mM DTT, 30µl 5X Second strand buffer, 3µl of 10mM
dUTP mix, 1µl E.coli DNA ligase (M0205, New England BioLabs), 3µl E.coli DNA
polymerase I (N0209, New England BioLabs), 5U RNaseH (M0297, New England
BioLabs) and topping up the volume to 150µl. The reaction was incubated for 1-2
hours at 16◦C and cleaned up again using the Qiagen PCR purification kit eluting in
50µl preheated EB.
Library preparation
Library preparation was performed as described for the TruSeq ChIP library
preparation procedure with one modification. After adapter ligation, before PCR
amplification, 1µl uracil-DNA glycosylase (UNG) was added and incubated at 35◦C
for 15 minutes.
Next-generation sequencing
Sequencing was performed by the Genomics core facility at CRUK CI
RNA-Seq libraries were sequenced on an Illumina HiSeq2500 according to
manufacturer’s instructions using a paired-end 125bp run.
46 Materials and Methods
2.9 BioCAP-Sequencing
BioCAP-Sequencing was performed by Dr. Hannah Long in the laboratory of Dr.
Rob Klose at the University of Oxford.
Biotinylated CxxC affinity purification (Bio-CAP)-Sequencing was performed as
described in Blackledge et al. (2012). In brief, flash-frozen tissue samples were
homogenised and DNA was purified using the Genomic-tip 100/G kit (Qiagen).
NeutrAvidin-coated magnetic beads (ThermoFisher Scientific, 7815-2104-011150)
were washed with BC100 buffer (10% glycerol, 0.2% Triton X-100, 20mM Tris pH
7.9 and 100mM NaCl), incubated with 25µg biotinylated hKDM2b-CxxC protein for 1
hour at 4◦C and the conjugate was then washed in CAP100 buffer (12.5% glycerol,
0.1% Triton X-100, 20 mM HEPES pH 7.9 and 100 mM NaCl). Genomic DNA was
sonicated to an average size of 150 - 250 bp using a Diagenode Bioruptor and
diluted in CAP100 buffer. 8µg of DNA were added to conjugated CxxC resin and
incubated for 1 hour at 4◦C. Resin was collected on magnetic stand and washed
twice with 1mL CAP100 buffer and then eluted twice in 50µl CAP300 buffer (12.5%
glycerol, 0.1% Triton X-100, 20mM HEPES pH 7.9 and 300mM NaCl) for 10 minutes
at RT. The elution process was repeated with buffers containing 500mM, 700mM
and 1M NaCl and 100µl fractions were purified using the QIAquick PCR Purification
kit (Qiagen) eluting in 50µl EB. The 700mM and 1M fractions were pooled for
sequencing and libraries were prepared using the NEBNextTM DNA Sample Prep
Kit (New England BioLabs). Sequencing was performed on an Illumina HiSeq2500
according to manufacturer’s instructions using a single-end 50bp run.
2.10 smallRNA-Seq
SmallRNA-Seq libraries were prepared using the Illumina TruSeq Small RNA
library Preparation kit, following manufacturer’s instructions with minor modifications.
For total testis RNA, 1µg of total RNA were mixed with 1µl RNA 3’ Adapter (RA3)
and incubated for 2 minutes at 70◦C and then placed on ice. 40µl of ligation
mix (2µl Ligation buffer (HML), 1µl RNaseOut Recombinant Ribonuclease Inhibitor
(ThermoFisher Scientific) and 1µl T4 RNA Ligase 2, truncated (M0242, New England
Biolabs)) were added, mixed by pipetting and incubated at 28◦C. After 1 hour
incubation, 1µl Stop Solution (STP) was added and incubated at 28◦C for another
15 minutes. 1µl of RNA 5’ Adapter (RA5) per sample were denatured at 70◦C for 2
Materials and Methods 47
minutes and added to 3’-ligated RNA together with 1µl 10mM adenosine triphosphate
(ATP) and 1µl T4 RNA ligase (M0204, New England Biolabs) followed by incubation
at 28◦C for 1 hour. 6µl of 5’ and 3’ adapter-ligated RNA (half of the reaction) were
then mixed with 1µl RNA RT Primer (RTP) and incubated at 70◦C for 2 minutes. 5.5µl
of reverse transcription mix (2µl 5X First Strand buffer (ThermoFisher Scientific),
0.5µl 12.5mM dNTP mix, 1µl 100mM DTT (ThermoFisher Scientific), 1µl RNaseOUT
(ThermoFisher Scientific), 1µl SuperScript II Reverse Transcriptase (ThermoFisher
Scientific)) were added and reaction was incubated at 50◦C for 1 hour. cDNA was
then amplified by adding 32.5µl PCR master mix (25µl 2x KAPA HiFi mix, 2µl RNA
PCR primer (RPI1), 2µl RNA PCR Primer Index and 8.5µl water) containing Illumina
index primers using the following programme: 30 seconds at 98◦C, 10 cycles of
10 seconds at 98◦C, 30 seconds at 60◦C and 15 seconds at 72◦C and a final
amplification at 72◦C for 10 minutes. Amplified PCR product was then purified from
2% agarose gel run at 80V for 2 hours. 150bp band was excised and purified using
the Qiagen GelExtraction Kit buffers with Zymo Clean-and-Concentrator columns to
allow elution in small volume of 10µl.
When starting with low input material from as little as 100,000 FACS-sorted cells
or after RNA immunoprecipitation (RIP), the protocol was slightly modified: 1:10
dilution of RNA 3’ adapter and 1:1 dilution of RNA 5’ adapter were used to reduce the
amount of adapter dimers. At reverse transcription step, the entire volume (13.5µl
of 5’- and 3’-adapter-ligated RNA was used, with only 0.5µl of RNA RT Primer and
slightly adjusted RT mix (4µl 5X First Strand buffer, 0.5µl 12.5mM dNTP mix, 2µl
100mM DTT , 1µl RNaseOUT and 1µl SuperScriptII RT. PCR amplification was then
performed by adding 25µl 2x KAPA HiFi Mix and 1µl RNA PCR Primer and 1µl RNA
PCR Primer Index using the same programme, but with 12 cycles.
smallRNA libraries were then quantified on Agilent High Sensitivity Bioanalyser
ChIP, multiplexed and sequenced on Illumina HiSeq2500 using a single-end 50bp
run.
2.11 RNA immunoprecipitation followed by high-
throughput sequencing (RIP-Seq)
Freshly dissected testes from adult animals were washed in ice-cold PBS three
times and then homogenised using a dounce tissue grinder. Dounced tissue was
then filtered through 100µm cell strainer and pelleted at 2500 g for 3 minutes at 4◦C.
Pellet was resuspended in equal volume of RIP Lysis buffer (50mM Tris-HCl, pH
48 Materials and Methods
8.0, 150mM NaCl, 5mM MgCl2, 1mM DTT, 0.5% Sodium deoxycholate, 1% Triton
X100, 1X Protease Inhibitor, 2nM Vanadyl RNase Inhibitor), mixed by pipetting and
incubated for 20 minutes on ice. Lysate was then centrifuged at 14,000 g for 10
minutes at 4◦C and 100µl of lysate were stored as input material. 100µl of lysate
were used per IP.
100µl Dynabeads® Protein G (ThermoFisher Scientific) were washed 3 times
in RIP Wash buffer (10mM Tris-HCl, pH8.0; 150mM NaCl; 0.05% NP-40) and
then resuspended in 250µl RIP Wash buffer. Beads were mixed with 5-10µg of
antibody and incubated at RT for 30 - 60 minutes. Antibody-bead mixture was
washed three times with RIP Wash buffer and then resuspended in 900µl RIP Wash
buffer containing 100U RNaseOUT and 1X Protease Inhibitor. 100µl of RIP lysate
were added to antibody-bead mix and incubated for at least 3 hours or overnight
at 4◦C on rotating wheel. IPs were washed three times in cold RIP Wash buffer
containing 100U/ml RNaseOUT and 1X Protease Inhibitor. RNA was purified by
adding 150µl Proteinase K buffer (117µl RIP Wash buffer, 15µl 10% SDS, 18µl
10mg/ml proteinase K) to beads and incubating at 55◦C for 30 minutes on shaker
(processing input sample alongside). Supernatant was transferred, topped up to
400µl with RIP Wash buffer and mixed with 400µl phenol:chloroform:isoamyl alcohol.
Phase separation was performed at 10,000 g for 10 minutes at RT and aqueous
phase was transferred to a new tube and mixed with 400µl chloroform and separated
again. Purified aqueous phase was then mixed with 166µl 3M sodium acetate, 6µl
linear acrylamide and 1200µl 100% ice-cold ethanol and precipitated at -80◦C for at
least one hour. RNA was precipitated at 14,000 g for 30 minutes at 4◦C and pellet
was washed once with 80% Ethanol. RNA pellet was then air dried and resuspended
in 6µl RNase-free water and used for smallRNA library preparation as described
above with modifications for low input RNA samples.
2.12 Single-cell RNA-Seq on spermatogenic cells
Testes were dissociated as described in Section 2.5 to yield single cell suspension
and FACS sorted to obtain different spermatogenic cell populations (spermatogonia,
primary spermatocytes and round spermatids) to be compared by single-cell RNA
sequencing (sc-RNA-Seq).
Materials and Methods 49
Fluidigm C1 technology
Capture on the Fluidigm C1 and subsequent Nextera library preparation was
performed by Dr. Celia Martinez.
Sorted cells were spun down in swing bucket centrifuge at 300 g for 3 minutes at
4◦C, resuspended at a concentration between 66,000-333,000 cells/mL and mixed
with C1 Suspension Reagent in a 3:2 ratio and kept on ice. This concentration
leads to a total of 200-1000 cells to be used for capture. Spermatogonia and round
spermatids were captured on 10-17µm C1TM Single-Cell mRNA Seq Integrated
Fluidic Circuit (IFC) and primary spermatocytes were captured on 17-25µm large
C1TM Single-Cell mRNA Seq IFC (Fluidigm).
Capture, lysis, reverse transcription and cDNA amplification were performed on
the C1 Fluidigm system following the instructions in the SMART-Seq® v4 Ultra® Low
Input RNA Kit for the Fluidigm® C1TM System (Clontech). In brief, the following
reagents were prepared in a pre-PCR and RNA-free workstation and kept on ice
until use:
• 10X Reaction buffer containing 19µl 10X Lysis buffer and 1µl RNase Inhibitor
• Lysis mix containing 1µl of ERCC spike-in RNA (Ambion) in a 1:20,000 dilution,
2.4µl 12µM 3’ SMART-Seq CDS Primer II A, 14µl nuclease-free water and
2.6µl 10X Reaction buffer.
• Reverse Transcription mix containing 1.2µl 20X C1 Loading Reagent, 11.2µl 5X
Ultra Low First-Strand buffer, 2.8µl 48µM SMART-Seq v4 oligonucleotide, 56U
RNase Inhibitor, 9.8µl nuclease-free water. 560U of SMARTScribe Reverse
Transcriptase were added just before use.
• PCR mix containing 4.5µl 20X C1 Loading Reagent, 4.4µl nuclease-free water,
3µl 12µM PCR Primer II A, 75.2µl 2X SeqAmp PCR buffer, 3.625U SeqAmp
DNA Polymerase
Priming of IFCs was performed by adding 200µl of C1 Harvest reagent into both
accumulators, 20µl of C1 Harvest reagent into 36 inlets around the accumulators,
20µl of C1 Harvest reagent into each pair of inlets on each side of the IFC and 15µl
C1 Blocking reagent into both cell inlets and running the SMART-Seq v4: Prime
script on the C1 system. After priming, blocking reagent was removed from both
inlets, 24µl of C1 Preloading Reagent were added to Inlet 2 on the left side and 7µl
50 Materials and Methods
of Cell Wash buffer were added to Inlet 6 on the right side of the chip. 20µl of Cell
suspension were added into one inlet previously filled with blocking solution, 5µl of
which will eventually enter the IFC. Capture was then performed on the C1 running
the SMART-Seq v4: Cell Load script. Capture sides were then visually inspected to
annotate wells that captured multiple cells or debris for filtering during data analysis.
After capture, remaining solution in flowthrough outlet was removed and 180µl of
C1 Harvest Reagent were added to all four reservoirs. 7µl of Lysis mix were added
into Inlet 3, 8µl of Reverse Transcription mix were added into Inlet 4 and 24µl of
PCR mix were added into Inlets 7 and 8. Lysis, reverse transcription and cDNA
amplification was then performed by running the SMART-Seq v4: Sample prep script.
Amplified cDNA was then collected the next day by adding 3.5µl of cDNA into a
96-well containing 10µl C1 DNA Dilution reagent. Concentration of diluted cDNA was
then checked on the Bioanalyser using the DNA High Sensitivity kit to ensure that
concentrations are within the optimal range for library preparation between 0.1-0.3
ng/µl.
Nextera XT library prep
Library preparation was performed using the Nextera XT Library Preparation Kit
(Illumina) according to manufacturer’s instructions. In brief, 1.25µl of diluted cDNA
sample were mixed with 2.5µl Tagment DNA buffer and 1.25µl Amplification Tagment
Mix and incubated at 55◦C for 10 minutes. After cooling down to 10◦C, 1.25µl of
NT buffer were added per sample to stop tagmentation reaction. 3.75µl of Nextera
PCR Master mix and 1.25µl of Index 1 and Index 2 primers were added and PCR
amplification was performed as follows: 72◦C for 3 minutes, 95◦C for 30 seconds,
12 cycles of 95◦C for 10 seconds, 55◦C for 40 seconds and 72◦C for 60 seconds
followed by a final incubation at 72◦C for 5 minutes.
1µl of each amplified library were then pooled together and sequenced on Illumina
HiSeq2500 using a paired-end 125bp run.
Materials and Methods 51
Computational data analysis
Initial data processing including de-multiplexing and alignment was performed
by the Genomics and Bioinformatics cores at CRUK CI and Drs. Tim Rayner and
Margus Lukk.
2.13 ChIP-Seq analysis workflow
Sequence alignments and filtering
ChIP-Seq libraries were aligned against the reference genome using the Burrows-
Wheeler Aligner (BWA) (Li and Durbin, 2009). Human and wild-type mouse libraries
were aligned against GRCh37/hg19 and GRCm38/mm10, respectively. Tc1 libraries
were aligned against a composite genome consisting of GRCm38/mm10 with the
addition of human chromosome 21 from GRCh37/hg19. Regions with a mapping
quality score of 0 were removed and only uniquely mapping reads were used for
downstream analysis. ChIP-Seq libraries were filtered against ENCODE blacklisted
regions (hg19/GRCh37 and mm9 liftover to mm10) (The ENCODE Project Consor-
tium, 2012). Known regions on HsChr21 that are deleted or duplicated in the Tc1
mouse based on Gribble et al. (2013) were removed from both Tc1 and human
libraries (Code A.2).
ChIP-Seq peak calling
Peak calling was performed using the Model-based Analysis of ChIP-Seq (MACS)
algorithm version 2.0 (MACS2) (Feng et al., 2012). Concatenated input samples of
higher complexity were used as controls per experimental condition. The callpeak
function was specified as well as SPMR to generate signal per million reads pileup
files for visualisation (Code A.3). For broad spanning factors such as H3K27ac
and RNA Polymerase II –broad was specified. macs2 bdgcmp -m FE was used to
generate signal tracks showing the fold enrichment of treatment over control. These
tracks were used for visualisation on the UCSC Genome Browser (Kent et al., 2002)
and Integrative Genomics Viewer (IGV) (Robinson et al., 2011).
52 Materials and Methods
Differential binding analysis
Differential binding analysis on enriched regions was performed as in Ross-Innes
et al. (2012), using the R/Bioconductor package DiffBind (version 1.14.5) (Stark and
Brown, 2011). Using dba.count, peaks were required to be present in at least half
(when comparing between Tc1 and human) or one fourth (when comparing between
different germline transmission) of all replicates and reads were normalised using
the Trimmed Mean of M-values (TMM) method using the effective library size after
subtracting control reads (Robinson and Oshlack, 2010). Differentially bound sites
were defined to have at least 3.5-fold (when comparing between Tc1 and human)
or 2.5-fold (when comparing between different germline transmission) difference
in binding intensity with an FDR of less than 0.1 between conditions (CodeA.4).
The log2 mean read concentration as defined by DiffBind was plotted using ggplot2
(Wickham, 2009) and correlation between samples was calculated using Pearson’s
correlation.
Functional annotation for genomic regions were obtained using the R/Bioconduc-
tor package compEpiTools function GRannotateSimple (Kishore et al., 2015).
Generation of ChIP-Seq intensity heatmaps
Pileup bedGraph files normalised to reads per million as generated by macs2
during peak calling were used to plot ChIP-Seq intensity heatmaps. BedGraph
files were converted to bigwig format using bedtools (Quinlan and Hall, 2010) and
uploaded to Galaxy (Afgan et al., 2016). Input reads were subtracted from ChIP
reads and the computeMatrix and plotHeatmap function from deepTools (Ramirez
et al., 2014) were used to generate intensity heatmaps centred on peak summits
and sorted by decreasing intensity in the Tc1 sample.
Repeat overlap
Peak summits for shared and Tc1-specific sites were either defined by macs2 or
redefined using DiffBind after differential binding analysis using dba.count (summits
= 25) and obtained using dba.peakset. 50bp windows centred on peak summits were
then overlapped with repetitive elements on HsChr21 or the mouse genome obtained
from RepeatMasker (Smit et al., 2015) with simple, telomeric and centromeric repeats
removed.
Materials and Methods 53
Sequence motif enrichment analysis
De novo sequence motif analysis was performed on repeat sequences on
HsChr21 that showed H3K4me3 enrichment in Tc1 mice using the MEME suite
(Bailey et al., 2015) with default settings and using dinucleotide scrambled versions
of the input sequences as control. The top 5 motifs were compared with known
motifs using Tomtom (Gupta et al., 2007) using the jolmaa2013.meme, JASPAR-
CORE-2014-vertebrates.meme and uniprobe-mouse.meme databases.
2.14 RNA-Seq analysis
RNA-Seq alignments were performed by Dr. Tim Rayner
RNA-Seq libraries from liver were aligned against the reference genome using
the Genomic Short-read Nucleotide Alignment Program (GSNAP) (Wu and Nacu,
2010). Tc1 libraries were aligned against GRCm38/mm10 with the addition of human
chromosome 21 from hg19. Differential expression analysis for liver was performed
using edgeR (Robinson et al., 2010) on at least 2 biological replicates per condition.
RNA-Seq libraries from testes were aligned using the Spliced Transcripts Align-
ment to a Reference (STAR) software (Dobin et al., 2013) and gene counts produced
by STAR were used for differential expression analysis using DESeq2 (CodeA.5)
(Love et al., 2014). Count data over repeat elements was calculated using the bamu-
tils count function from the NGSUtils suite (Breese and Liu, 2013) using -repeatfam
and -multiple partial or ignore parameters when using multi-mapping or only uniquely
mapping reads on .fa.out files obtained from RepeatMasker.
2.15 Differential methylation analysis
BioCAP-Seq libraries were aligned against a composite genome containing all
mouse chromosomes and human chromosome 21 (mm9 + hg19 Hschr21) using
bowtie (Langmead et al., 2009). Hypomethylated regions (HMRs) of DNA were
identified using MACS1.4 (Zhang et al., 2008) with settings -tsize = 50, -bw = 300
-mfold = 10,30 -pvalue = 1e-5 -verbose = 10 -g 4.8e+8 against an input control. Only
HMRs that were identified in both biological replicates were retained and HMRs
overlapping known breakpoints or deletions of HsChr21 in the Tc1 mouse were
removed.
54 Materials and Methods
For differential methylation analysis, HMRs obtained for female and male germline-
derived Tc1 mice were merged and read counts over genomic intervals were ob-
tained using bedtools genomecov (Quinlan and Hall, 2010).
2.16 smallRNA-Seq analysis workflow
smallRNA-Seq alignments and analysis scripts in python and R were provided
by Dr. Tim Rayner
Sequence alignments
Initial processing of smallRNA libraries was done using the Kraken software
(Davis et al., 2013). First, Reaper was used to demultiplex and remove adapter
sequences, followed by Tally to count identical sequences and remove redundancy.
Unique smallRNA reads with individual count information were then mapped to the
reference genome using BWA.
Filtering and size selection
Due to the short length of smallRNA reads, the amount of cross-mapping between
mouse and human smallRNA sequences is significantly higher than for ChIP-Seq
reads. We therefore determined the amount of cross-mapping by aligning Tc0
libraries to the Tc1 genome and compiled all sequences from wildtype mouse
smallRNA libraries that align to HsChr21 and removed these reads from all Tc1, Tc0
and human smallRNA libraries (Code A.6). Libraries were then size selected to only
include reads between 24 - 32 nucleotides in length using the bamutils filter function
with -minlen 24 and -maxlen 32 from the NGSUtils suite (Breese and Liu, 2013).
piRNA cluster definition
piRNA clusters were defined using a sliding window approach and counting
a minimum number of distinct piRNA sequences (peakDepth) across a defined
windowSize. Annotated genes were masked and adjacent windows that passed
the threshold were merged together in the end (CodeA.7). Clusters from biologi-
cal replicates were combined, calculating the average read count per cluster and
testing for convergent expression across replicates. Only convergent clusters that
Materials and Methods 55
showed consistent expression across biological replicates were used for downstream
analysis.
Differential expression analysis
Differential expression analysis across piRNA clusters was performed by counting
reads over defined clusters using the bedtools multicov function and using the
obtained count data in DESeq2 with default settings normalising against effective
library size (Love et al., 2014).
Coverage across chromosomes
Number of smallRNA reads mapping to chromosomes were calculated using
bedtools multicov function and count data was used in DESeq2 to calculate norm
factors and Mean of normalised counts per condition comparing Tc0 and Tc1 cov-
erage across chromosomes. For human samples, DESeq2 was performed without
any contrast to normalise reads for library size and extract the mean of normalised
counts for HsChr21 to plot against Tc1, therefore allowing normalisation against the
entire library size.
Coverage visualisation
Coverage tracks over genomic regions of interest were generated using the
R/Bioconductor package Gvis (Hahne and Ivanek, 2016).
Sequence motifs
Sequence logos were generated using all reads mapping to the LTR array from
Tc1 and human libraries using the R/Bioconductor package seqLogo (Bembom,
2017).
Testing for unique 5’ ends
The test whether Tc1-specific smallRNA reads were indeed unique, we pooled
reads from all biological replicates mapping to the megasatellite region, trimmed
these sequences to 20nt and extracted unique sequences. These sequences
were then queried against the entire human smallRNA libraries using only uniquely
mapping reads or including multi-mapping reads using grep.
56 Materials and Methods
2.17 Single-cell RNA-Seq analysis
single-cell RNA-Seq analysis was performed in collaboration with Nils Eling.
Genome reference and annotation
For alignment of Tc1 and wild-type libraries to a reference genome, fasta se-
quences as well as GTF annotation files from the mouse genome (GRCm38)
were merged with chromosome 21 from the human reference (GRCh38). Addi-
tionally, the fasta sequences of the ERCC spike-ins (https://assets.thermofisher.com/
TFS-Assets/LSG/manuals/ERCC92.zip) were added to the sequence and annotation
file. For the C1 Fluidigm data, the new sequence file was prepared for alignment
using the gmap_build function of the gsnap aligner.
Read alignment and counting
Read alignment to a reference genome was performed using gsnap with default
parameters while supplying a splice-site file. All samples were mapped against the
merged GRCm38 and chr21/GRCh38 genome. Transcript numbers for annotated
genes were counted using the HTSeq package with default options.
Quality control
Low quality cells were removed to avoid technical biases in data analysis based
on previously defined criteria (Brennecke et al., 2013; Martinez-Jimenez et al., 2017).
In the first step, we removed capture sites in which no cell, multiple cells or
cell debris were detected. We observed that the total number of ERCCs per cell
differ batch-specifically. Therefore, ERCCs were not used for quality control or
normalisation. Cells with less than 15% of reads mapping to the exonic region of
the genome were removed as well as cells with more than 3% of unmapped reads.
Additionally, cells with less than 3000 mouse genes detected (at least one count)
were excluded from downstream analysis.
In order to remove contaminations from different cell populations we pre-normalised
the data using the pooling approach implemented in the scran package in R where
cells were normalised within each batch based on mouse gene expression (Lun
et al., 2016). Scattered outliers on the first 2 principal components as well as sper-
Materials and Methods 57
matocytes and spermatogonia that showed expression of the spermatid marker
genes Prm1 and Prm2 (more than 10 counts) were removed.
Filtering on genes was performed in a two-step approach. First, mouse genes
that are expressed on average with less than one count were removed. Second,
human genes that were detected in all wild-type cells with a total count larger than
10 were excluded (CodeA.8).
Normalisation
We used the scran package to normalise the data. This method uses a cell
pooling approach to compute stable size factors and de-convolutes the averaged
cells to find cell-specific size factors (Lun et al., 2016). Size factors were computed
using the expressed mouse genes while only cells within the different cell-types were
pooled.
Diffusionmaps and diffusion pseudo-time
To computationally order cells along cell-to-cell transition paths, we calculated
diffusionmaps using the R package destiny (Angerer et al., 2016; Haghverdi et al.,
2016). Diffusion pseudo-time is estimated via diffusion-like random walks along the
transition path using the 20 nearest neighbours (CodeA.9).
Differential expression analysis
Differential gene expression between cell populations or genotypes was per-
formed using the find_markers function of the scran package. Genes with a log2-fold
change (log2FC) larger or smaller than zero and a false discovery rate smaller than
10% were selected as cell type specific genes (CodeA.10).
Chromosomal silencing and reactivation
Quantification of expression from human chromosome 21 and mouse chro-
mosome X was performed by calculating the ratio of normalised counts from the
chromosome of interest to all normalised counts from the mouse genome. Addition-
ally, the number of genes with a normalised count greater than zero located on each
chromosome was detected. Location of genes across human chromosome 21 were
visualised using the Gviz package (CodeA.11).
58 Materials and Methods
Gene enrichment analysis
The functional Annotation Tool from DAVID was used to perform gene ontology
analysis using the GOTERM-BP-DIRECT database (Huang et al., 2009).
3 | Transmission of a human
chromosome through mouse
male meiosis
Parts of the work presented in this chapter have been published in a similar form
in eLife (Ernst et al., 2016). Histological assessment was performed with help from
Dr. Sarah Aitken and machine-learning based image analysis was performed by Dr.
Jeremy Pike from the Light Microscopy core. The BioCAP-Sequencing data was
generated by Dr. Hannah Long and RNA-Seq analysis was assisted by Nils Eling.
3.1 Introduction
Mammalian gametogenesis is characterised by a high degree of sexual dimor-
phism which also affects the stringency of checkpoint control during meiosis. As a
result, female meiosis is more error-prone resulting in a relatively higher number
of oocytes carrying chromosomal abnormalities compared to sperm (Martin et al.,
1991). However, more rigorous checkpoint control in male meiosis comes at the
price of germ cell arrest leading to frequent perturbations of fertility (Hunt, 2002). As
such, the majority of mouse models used to study aneuploidy exhibit complete male
sterility due to a block in spermatogenesis during early stages of meiosis (Davisson
et al., 2006). The transmission of exogenous DNA through the male germline is
therefore extremely rare and, in the mouse, has only been reported for comparatively
small human artificial, or fragmented chromosomes (Tomizuka et al., 1997; Voet
et al., 2001; Weuts et al., 2012).
Transmission of genetic material through the male germline however, involves a
series of global remodelling of the chromatin landscape, This includes a genome-
wide derepression after meiosis followed by condensation of DNA during sperm
packaging and eventually de-packaging and reconstruction of the epigenetic land-
scape upon fertilisation (Dadoune, 1995; Soumillon et al., 2013). The effect of these
events on the transcriptional deployment of exogenous DNA passaged through the
male germline are entirely unknown.
60 Results
In particular, the transcriptional activation occurring post-meiosis includes numer-
ous repetitive elements that often require species-specific regulatory mechanisms
to ensure their proper silencing. It is therefore unclear whether exogenous repeat
sequences, for which the appropriate silencing mechanisms might be absent in the
mouse, could acquire a more permissive chromatin state in somatic tissues after
their transient activation in the male germline.
Here we make use of a trans-chromosomic mouse model, the Tc1 mouse, which
carries one copy of human chromosome 21 and therefore serves as a model for
human Down syndrome, by mimicking many phenotypes of the human disease
(O’Doherty et al., 2005). Unlike other aneuploid mouse models, we found that the
Tc1 mouse can passage the exogenous human chromosome through the male
germline allowing us to study male meiosis under aneuploid conditions and assess
the effects of male germline transmission on the epigenetic landscape of exogenous
DNA.
3.1.1 Aim
The aim of this project was to study the effect of pre-existing aneuploidy on
meiotic progression in males and test whether an exogenous chromosome can
be transmitted through the male germline, thereby allowing to study the effects of
male-specific reprogramming on the transcriptional deployment in the offspring.
Results 61
3.2 Results
3.2.1 Sub-fertility phenotype of Tc1 males
To assess the fertility of male mice carrying human chromosome 21 (Tc1), we
performed phenotypic and histological comparisons with wild-type littermates that
did not inherit human chromosome 21 (Tc0). Tc1 males showed signs of testicular
atrophy with a drastic reduction in testis size and weight as well as a lower number
of epididymal sperm (Figure 3.1A). These Tc1-associated phenotypes appeared
to be specific to testes, as total body weight and liver weight (as a representative
somatic tissue) were indistinguishable between Tc1 and Tc0 mice (Figure 3.1B).
Histologically, Tc1 males also displayed subfertility phenotypes when compared
to testes from Tc0 littermates. These observed histological abnormalities were
characterised by the absence (spermatogenic arrest) or reduced frequency (hypo-
spermatogenesis) of secondary spermatocytes, as well as the absence of any cell
types derived from these (Borg et al., 2010). The most pronounced change in
Tc1 males was the strong reduction or complete absence of mature spermatozoa
in a majority of the tubules (Figure 3.1C blue arrowheads). Neither Tc1 nor Tc0
mice demonstrated other subfertility phenotypes, such as Sertoli cell-only syndrome,
tubular sclerosis, or fibrosis (Dohle et al., 2012).
To further classify the sub-fertility phenotype in Tc1 males, we developed a sys-
tem based on Dohle et al. (2012) and Creasy et al. (2012) to grade spermatogenesis
according to the predominant histological pattern observed in individual seminiferous
tubules. Our system ranged from completely normal spermatogenesis (Grade I) to
maturation arrest (no production of mature sperm: Grade IV) with two intermediate
steps of mild and severe hypo-spermatogenesis (Grade II and Grade III, respectively)
and was mainly based on the presence and abundance of maturing spermatozoa, a
cell type present during all epithelial stages of spermatogenesis (Figure 3.2A). This
quantification showed that the vast majority of seminiferous tubules from wildtype
animals (>95%) showed normal spermatogenesis. In contrast to that, Tc1 males
showed a wide range of phenotypic variation both across tissue sections and be-
tween animals. While we found phenotypically normal tubules in all Tc1 males, their
frequency was strongly reduced and the majority of tubules displayed mild to severe
hypo-spermatogenesis with little or no sperm production (Figure 3.2B).
62 Results
Figure 3.1 | Subfertility phenotype in Tc1 males
A| Comparison of testis size from 12-week-old Tc0 and Tc1 littermates.
B| Testis weight, sperm count, body weight and liver weight from Tc0 and Tc1 mice.
Five mice of each genotype aged between 12-14 weeks were used for tissue and
body weight measurements. Sperm samples were from mice aged between 16-32
weeks, and were counted for 2 Tc0 and 5 Tc1 animals. Statistical analysis was a
student’s t-test (p<0.0001).
C and D| Photomicrographs of testis tissue sections from adult Tc0 and Tc1 mice
stained with H&E and IHC with anti-γH2AFX. Original magnification was 20x. Blue
arrowheads indicate mature sperm, red arrowheads indicate failure in chromosome
segregation. Infertile males did not produce any offspring over 6 month period kept
with the same female.
Interestingly, the few examples for Grade II-IV tubules found in wildtype animals
simply displayed an absence of mature spermatozoa, whereas similarly graded
tubules in Tc1 males often displayed failures in chromosome segregation corre-
sponding to metaphase of the first meiotic division (Figure 3.2C red arrowheads).
Results 63
Figure 3.2 | Classification of hypo-spermatogenesis in Tc1 males
A| Seminiferous tubules were classified from photomicrographs of Tc1 testis tissue
sections stained with H&E into Grade I - normal spermatogenesis; Grade II - mild
hypo-spermatogenesis (all germ cell stages present but visible meiotic disruption and
suboptimal frequency of spermatozoa); Grade III - severe hypo-spermatogenesis (all
germ cell stages present including occasional spermatozoa); Grade IV - maturation
arrest (incomplete spermatogenesis, no cell types beyond the spermatocyte stage).
Original magnification 20x.
B| Percentage of tubules per sub-fertility grade is shown for individual Tc0 and Tc1
males. Colour shading corresponds to individual animals.
C| Example images of Grade III tubules from wild-type and Tc1 mice, red arrowheads
highlighting defects in chromosome segregation.
DNA DSBs are a prominent feature during meiosis and are marked by an accu-
mulation of phosphorylated H2AFX. γH2AFX therefore shows a dynamic pattern
throughout male meiosis reflecting the genome-wide distribution of DNA DSBs during
early leptonema which become subsequently repaired via homologous recombina-
tion but persist into the pachytene stage on any chromosome that fails to undergo
synapsis. This leads to meiotic silencing of unpaired chromosomes (MSUC), which
in male germ cells affects the sex chromosomes that are largely unpaired, but also
any unsynapsed autosomes (Turner et al., 2005).
64 Results
To test whether the presence of HsChr21 has an effect on the localisation pattern of
DNA DSBs and γH2AFX during spermatogenesis we performed IHC for this chro-
matin mark in Tc1 and wild-type testes. As previously described in Hamer (2002)
and Mahadevaiah et al. (2001), wild-type mice display a homogeneous γH2AFX
staining across the nucleus of cells undergoing meiotic recombination at the lep-
totene stage (Figure 3.3A orange arrowheads), followed by its restriction to the XY
body in pachytene cells (Figure 3.3A purple arrowheads). In contrast, Tc1 animals
showed a more diffuse distribution of γH2AFX that does not appear to be restricted
to leptotene cells (Figure 3.3B). In addition, we detected strong γH2AFX signal in
cells undergoing meiotic cell divisions, at which stage DNA DSBs should be resolved
to allow meiotic progression (Figure 3.1D). Such increased staining intensity could
be an indicator of ongoing apoptosis, rather than persisting DSBs (Odorisio et al.,
1998), supporting an arrest at the meiotic cell division as previously suggested
(Cloutier et al., 2015).
Taken together, our phenotypic and histological profiling of Tc1 males revealed a
subfertility phenotype but no complete maturation arrest as frequently observed in
males of other aneuploid mouse models.
3.2.2 Metaphase arrest during the first meiotic division
Our previous observations suggest an arrest at metaphase of the first meiotic
division (MI) due to the increased number of cells with meiotic figures that also
displayed a higher staining intensity for γH2AFX.
Due to the synchronised nature of spermatogenesis and the developmental timings
required for different cell types to transition from one to the next, tubular cross-
sections can be classified into different stages based on the combination of cell
types that are visible (Figure 3.4).
We therefore stained testis cross-sections with the PAS technique that allows better
visualisation of the acrosome development and classified seminiferous tubules into
epithelial stages according to a binary decision code proposed by Meistrich and
Hess (2013) (Figure 3.5A). For a first approximation, tubules were assessed for the
presence of one or two generations of spermatids which allows distinction into late
or early/mid spermatogenesis, respectively. In tubules containing two generations of
spermatids we then assessed the structure of the acrosome in round spermatids
and the localisation of elongating spermatids within the tubules. Tubules with round
spermatids that had no or only small acrosomic granules were assigned as Stage
I-III, whereas tubules with acrosome caps forming over the nucleus
Results 65
Figure 3.3 | Aberrant γH2AFX staining in Tc1 animals
Photomicrographs of testis tissue sections with IHC against γH2AFX
A|Wildtype testes display homogenous nuclear distribution in leptotene cells (orange
arrowheads) and restriction of γH2AFX to a single foci in pachytene cells (purple
arrowheads).
B| γH2AFX in Tc1 testes shows similar distribution in leptotene cells but fails to be
restricted to a single foci in numerous pachytene spermatocytes (green arrowheads).
of round spermatids were staged as Stage IV-VI. Tubules in which elongating
spermatids were lining the lumen were assigned to epithelial Stage VII-VIII.
Late stages of spermatogenesis, in which tubules contain only one generation
of spermatids, can be further dissected based on chromatin structure in primary
spermatocytes and the presence of secondary spermatocytes. In the absence of
meiotic figures or secondary spermatocytes tubules were assigned to Stage IX-XI,
whereas the presence of these structures or cell type allowed the identification of
Stage XII tubules.
66 Results
Figure 3.4 | Binary decision code for epithelial staging of seminiferous tubules
A| Illustration of cell type composition at different epithelial stages during spermato-
genesis and annotation of binary decision key for staging (adapted from Meistrich
and Hess (2013)).
B| Example images for different cell types from PAS stained testis sections (original
magnification 60X).
According to this system, we staged a total of 6 wild-type and 9 Tc1 cross-
sections and calculated the percentage of tubules at a given stage per animals
(Figure 3.5).
Results 67
Figure 3.5 | Staging of seminiferous tubules on PAS stained tissue sections
A| Seminiferous tubules from wild-type and Tc1 testes were staged from photomi-
crographs of PAS stained tissue sections. Staging was performed according to the
binary decision key depicted in Figure 3.4 (Meistrich and Hess, 2013).
B| Percentage of tubules per spermatogenic stage. Males marked as infertile did
not produce offspring over a six month period kept with the same female. Statistical
analysis was a student’s t-test (*p<0.05; **p<0.005).
Even for wild-type animals we observed a high degree of variability between
animals with the exception of Stage XII tubules, of which only a small number was
consistently found in wild-type cross-sections (less than 5% of all tubules). Due to the
overall reduced number of spermatids in Tc1 animals, the distinction between tubules
in early and middle stages of spermatogenesis was not as clear as for wild-type
animals. However, Stage XII tubules could be reliably identified based on the meiotic
figures in cells undergoing meiotic cell divisions and showed a significant increase in
Tc1 compared to wild-type animals, consistent with an arrest at metaphase I.
68 Results
To accurately quantify the number of meiotic cells, we performed immunofluo-
rescence (IF) stainings for α-tubulin to visualise the meiotic spindle and phospho-
rylated histone H3 on Serine 10 (pH3), a histone modification commonly found
on metaphase chromosomes and therefore a good marker for meiotic and mitotic
cells (Wei et al., 1999) (Figure 3.6A). We then used these stainings to classify and
quantify different spermatogenic cell types across entire tissue cross-sections using
the interactive learning and segmentation toolkit ilastik (Sommer et al., 2011). Clas-
sification was performed in a two step process, first segmenting nuclear regions via
pixel classification followed by assignment of nuclear regions to different cell types
via object classification. This enabled the distinction between cells of the germinal
epithelium (purple), primary (4N) spermatocytes (green), meiotic cells (red) as well
as round and elongating spermatids (light and dark blue, respectively) (Figure 3.6B).
Separate identification of spermatogonia or Sertoli cells was not possible with
this approach, resulting in these cells being classified as either 4N spermatocytes or
round spermatids depending on cell size. However, we expect any mis-identification
to be consistent between Tc1 and wild-type mice, as neither cell type appeared
to be affected by the presence of HsChr21. The total number of cells quantified
per animal are shown in Table 3.1; however, due to the different size of the tissue
sections between Tc1 and wild-type animals, the data is presented in percentages
of the total cell number (Figure 3.6C). This revealed a significant increase in primary
spermatocytes in Tc1 animals compared to wild-types and a corresponding decrease
in both round and elongating spermatids. On average, the percentage of meiotic
cells was almost twice as high in Tc1 animals compared to wild-types corresponding
to the increased number of Stage XII tubules in Tc1 males. We then manually
classified 100 randomly selected pH3-positive cells per animal into different meiotic
Primary
spermatocytes Mitotic
Round
spermatids
Elongating
spermatids Total
Tc0
37621
17366
17757
737
493
1015
43564
28280
42176
30725
28575
38091
112647
74714
99039
Tc1
22186
16868
7620
12924
28941
881
570
239
488
748
11231
10141
4131
8403
22849
11966
10551
3972
7570
18798
46264
38130
15962
29385
71336
Table 3.1 | Cell type quantification across testis cross-sections using ilastik
Results 69
Figure 3.6 | Quantification of spermatogenic cell types
A| Representative immunofluorescent image of seminiferous tubules stained with
Hoechst (blue), anti-phospho histone H3 on serine 10 (green), and α-tubulin (red).
B| Illustration of the mark-up image created by ilastik for the tissue section presented
in A. Cells of the germinal epithelium (purple), 4N spermatocytes (green), meiotic
cells (red) as well as round spermatids (light blue) and elongating spermatids (dark
blue) were quantified across entire tissue cross-sections for 3 Tc0 and 5 Tc1 animals.
C| Quantification of different cell types for individual animals shown in % cell count.
Statistical analysis was a student’s t-test (*p<0.05; **p<0.005).
D| Manual classification of the percentage of cells into different meiotic stages
including pro-metaphase (P/M), metaphase (M), abnormal metaphase (M*), ana-
and telophase (A/T)) for individual animals. 100 cells were classified per animal,
however numbers are displayed as % of cells as some cells were classified as
non-mitotic.
stages based on DNA condensation and spindle structure to calculate the
metaphase to pro-metaphase ratio as described in Odorisio et al. (1998). De-
spite variability especially amongst Tc1 animals, we observed an almost 2-fold
increase in the ratio of metaphase to pro-metaphase cells (1.02 for Tc1 and 0.52
for wild-types), indicating an accumulation of metaphase cells in Tc1 animals and
therefore supporting an arrest at this stage (Figure 3.6D). Interestingly, Tc1 animals
frequently displayed abnormal metaphase cells with congression defects that could
be the underlying cause of the metaphase arrest as unaligned chromosomes have
been shown to activate the spindle checkpoint (Figure 3.7) (Eaker et al., 2001).
70 Results
Figure 3.7 | Congression defects in Tc1 metaphase cells
Representative high-resolution confocal images of tissue sections from wild-type (A)
and Tc1 (B) animals stained with Hoechst (blue), pH3 (green) and α-tubulin (red).
Tc1 snapshot shows a pro-metaphase cell (P/M), a normal metaphase cell (M) as
well as a metaphase with congression defect (M*).
In order to see whether the misaligned chromosome is the aneuploid HsChr21,
we performed FISH stainings on tissue sections from Tc1 animals with a locus-
specific probe for HsChr21 that does not cross-react with mouse sequences. DNA
counterstain was then used to identify stage XII tubules and mark up metaphase
cells as well as congression defects within these cells (Figure 3.8A, dotted white
and yellow circles, respectively). From the cross-sections of 3 Tc1 animals we
found a total of 19 Stage XII tubules containing 244 metaphase cells of which 110
(45%) displayed congression defects (Table 3.2). In 72 out of these 110 cells (65%)
the unaligned DNA identified in the DAPI stain overlapped with the FISH signal for
HsChr21. This shows that in the majority of cells that display congression defects
in the Tc1 mouse, HsChr21 is indeed the unaligned chromosome. This number is
however most likely an underestimation, since HsChr21 was not detected in all cells
with congression defects. We only found 15 cells (14%) with a congression defect in
which HsChr21 was properly aligned on the metaphase plate. The identity of the
misaligned chromosome in these 15 cells remains unclear.
Results 71
Figure 3.8 | The majority of congression defects are caused by unaligned
HsChr21
Example images of stage XII tubules from FISH stainings for HsChr21 (A) and mouse
chromosome X and Y (B) in Tc1 males. White dotted circles mark metaphase cells
and smaller yellow dotted circles mark DNA congression defects. Red triangles
highlight cells in which congression defect overlaps with signal for HsChr21; green
triangle highlights cell with congression defect but HsChr21 properly aligned on the
metaphase plate. Scale bars are 50 µm.
72 Results
Since it has been described that HsChr21 frequently colocalises with the X
and Y chromosomes in pachytene spermatocytes (Mahadevaiah et al., 2008), we
stained the same tissue sections with whole-chromosome paint for mouse X and Y
(Figure 3.8B) and scored for co-localisation of these chromosomes with HsChr21.
However, in metaphase spermatocytes we found HsChr21 in the proximity of mouse
chromosome X and Y in only 21 out of 72 cells (29%) that showed signal for all three
chromosomes.
The high number of cells with congression defects in Tc1 animals is expected to
activate the spindle checkpoint which prevents the metaphase-to-anaphase transition
and eventually leads to apoptosis (Eaker et al., 2001). We therefore performed IHC
for cleaved-caspase 3 and quantified the staining intensity in individual seminiferous
tubules. This revealed a drastic increase in CC3 positive cells specifically in stage
XII tubules of Tc1 animals, whereas tubules of other spermatogenic stages showed
no increased signal compared to wild-types (Figure 3.9).
Overall, these results demonstrate that Tc1 animals encounter an arrest at
metaphase of the first meiotic division due to a high number of congression defects
which are often caused by unaligned HsChr21. This metaphase arrest correlates
with increased apoptosis in stage XII tubules which is likely the cause of the re-
duced number of spermatids and thereby contributing to the sub-fertility phenotype
observed in Tc1 males.
3.2.3 Male germline transmission of human chromosome 21
We next wanted to test whether mouse sperm containing the transchromo-
somic HsChr21 can successfully fertilise wild-type eggs and produce aneuploid
offspring. Numerous trisomic mouse models and transchromosomic mouse strains
have reported that the transmission of extra-chromosomal material through the male
germline is difficult, if not impossible, depending on the size of the exogenous DNA
(Davisson et al., 2006; O’Doherty et al., 2005; Voet et al., 2001).
Indeed, the established protocol to maintain the Tc1 mouse line is by passaging
HsChr21 through the female germline by breeding Tc1 females with wild-type males
of an F1 cross between 129S8 and BL6 mice. Breeding with inbred males instead of
F1 hybrids leads to loss of HsChr21 after 2-3 generations (O’Doherty et al., 2005). In
our breeding colony, transmission of HsChr21 through the female germline occurred
in ∼35% of offspring born to aneuploid Tc1 mothers. Out of the 824 offspring
produced from this breeding set-up from 38 actively breeding Tc1 females, 290 mice
inherited HsChr21 from their mothers (Figure 3.10).
Results 73
Figure 3.9 | Increased apoptosis in stage XII tubules of Tc1 animals
Representative tissue sections from Tc0 and Tc1 testes stained with anti-CC3
antibody. Original magnification 20X, scale bar is 100 µm. Nuclei positive for CC3
were quantified in epithelial stage VII-VIII and stage XII tubules.
Stage XII
tubules
Metaphase
cells
Congression
defects
HsChr21
positive
Unaligned
HsChr21
BL134971 6
7
18
9
12
19
9
3
8
5
6
8
1
4
8
6
7
7
3
2
4
3
4
5
1
BL134884 8
15
16
10
13
14
18
15
25
7
11
7
4
6
9
8
9
6
10
5
4
6
11
11
15
3
7
3
2
2
6
8
8
BL135049 5
10
10
5
10
9
5
2
1
6
4
6
4
1
6
5
3
2
1
4
4
Total 19 244 110 206 72
Table 3.2 | Quantification of congression defects overlapping with HsChr21
74 Results
Figure 3.10 | Transmission of HsChr21 through the male germline is less effi-
cient than through the female germline
Oogenesis (left, top panel) can allow the transmission of epigenetic information de-
posited on retained maternal histones, whereas the majority of histones are replaced
by protamines during spermatogenesis (right, top panel). Germline transmission
of the full aneuploid chromosome HsChr21 was successful using male Tc1 mice
as transmitters, but at a substantially reduced frequency compared with female
transmission via eggs (13% versus 35%).
Parallel breeding experiments with aneuploid Tc1 males passaging HsChr21
through sperm resulted in successful transmission of HsChr21 in only ∼12.7% of
offspring (Figure 3.10). Out of a total of 1214 offspring produced from 38 Tc1 male
breeders, 135 inherited HsChr21 from their fathers.
Results 75
Figure 3.11 | Litter size and frequency for different mouse strains
Violin plots displaying the number of pups per litter (A) and days between litters (B)
for different mouse strains and breeding set-ups. Violin plots display kernel density
estimation to illustrate distribution of data.
Despite the subfertility phenotypes we observed for Tc1 males, the majority of
males used for breeding (32 out of 44) were fertile and 26 out of these 32 breeders
transmitted HsChr21 at least once to their offspring. However, in comparison to Tc1
females, individual males showed varying transmission rates (Table B.1 and B.2)
as well as inconsistent litter frequencies and numbers (Figure 3.11). Overall, litter
size was slightly increased as was the time inbetween litters when breeding with Tc1
males. Male germline-derived offspring were themselves fertile and were able to
transmit HsChr21 through the male germline for at least three generations, producing
aneuploid offspring at a similar rate (8 out of 65 animals inherited HsChr21).
Human Down syndrome, especially when paternally inherited, exhibits a strong
sex bias with a male-to-female ratio of up to 3.5. This has been attributed to a
preference of the extra copy of human chromosome 21 to segregate with the human
Y chromosome (Hassold et al., 1984; Nicolaidis and Petersen, 1998; Patersen et al.,
1993). We therefore tested whether the transmission of HsChr21 displays a similar
bias when passaged through the male germline in the Tc1 mouse. Unlike in human
Down syndrome, we observed no sex bias after male germline transmission in the
Tc1 mouse and obtained very similar male-to-female ratios for female and male
germline transmission of HsChr21 (1.04 and 1.08, respectively) (Table 3.3).
76 Results
Female germline transmission Male germline transmission
Agouti Black Agouti Black
524 (73%) 201 (27%) 937 (77%) 275 (23%)
Male Female Male Female Male Female Male Female
264 260 112 89 494 443 130 145
Sex ratio 1.03 1.07
Tc1-positive Tc1-positive
Agouti Black Agouti Black
214 (78%) 59 (22%) 100 (74%) 35 (26%)
Male Female Male Female Male Female Male Female
103 111 36 23 53 47 17 18
Sex ratio 1.04 1.08
Table 3.3 | Fur colour and gender ratios for female and male germline transmission
This is in line with the low level of co-localisation of HsChr21 and mouse chro-
mosome X and Y that we observed during metaphase (Figure 3.8B). Furthermore,
not only gender but also fur colour showed a nearly perfect pattern of mendelian
inheritance, suggesting that there is no obvious bias in the segregation pattern of
HsChr21 associated with either gender or strain background.
Male germline-derived offspring are macroscopically indistinguishable from fe-
male germline-derived offspring and histological assessment of various tissues
revealed no novel histological alterations (Figure 3.12). Neither hepatic nor renal
tissues displayed any histological differences between wild-type mice and female- or
male-germline-derived offspring (Figure 3.12A and B). Brain tissue sections from Tc1
animals demonstrated the previously described hippocampal phenotype of reduced
granule cells in the dentate gyrus that is associated with Down syndrome (Lorenzi
and Reeves, 2006) (Figure 3.12C). However, there was no difference in severity
between female and male germline-derived Tc1 animals regarding this phenotype.
Taken together, we show that HsChr21 can be transmitted through the male
germline of Tc1 mice for several generations, but at an appreciably lower frequency
than via female germline transmission.
Results 77
Figure 3.12 | Histological comparison across multiple somatic tissues
Representative photomicrographs of H&E stained tissue sections for liver, kidney
and brain from wild-type animals as well as female and male germline-derived Tc1
mice. Original magnification for hepatic and renal tissue sections was 20X; original
magnification for brain sections was 10X.
3.2.4 Meiotic loss of human chromosome 21
The lower transmission rate of HsChr21 through the male germline could suggest
a specific loss of aneuploid cells either during spermatogenesis or fertilisation and
early embryonic development. We therefore tested whether HsChr21-carrying cells
are eliminated at a particular stage during spermatogenesis by genotyping meiotic
and post-meiotic cells as well as mature sperm using FISH.
We isolated specific spermatogenic cell populations based on DNA content using
FACS as previously described in Bastos et al. (2005). In order to capture cells before
and after the meiotic division we sorted primary spermatocytes (4N) with the highest
DNA content as well as post-meiotic haploid spermatids (1N) with the lowest DNA
content (Figure 3.13A). The sorting profiles obtained for wild-type and Tc1 males
further confirmed our previous quantification of cell types based on IF and showed
an increase in 4N spermatocytes as well as a reduction in haploid spermatids in
78 Results
Figure 3.13 | Sorting of spermatogenic cell types
A| Cells were gated based on their Hoechst Red versus Hoechst Blue emission
profile to distinguish spermatogonia (unmarked cell population), cells undergoing
DNA replication (yellow), primary (4N) spermatocytes (orange), secondary (2N)
spermatocytes (purple), and haploid (1N) spermatids (green).
B| Bar charts representing the average percentages for different spermatogenic cell
types quantified for 3 individual Tc0 and 6 individual Tc1 animals.
C and D| Representative histograms for Tc0 and Tc1 animals displaying the distribu-
tion of gated cell populations from a total of 50.000 events.
Tc1 males (Figure 3.13B-D). Genotyping of 4N spermatocytes using the HsChr21-
specific probe revealed that ∼94% of these cells carried HsChr21, showing a sur-
prisingly low level of mosaicism compared to previously published rates in somatic
tissues (O’Doherty et al., 2005; Wilson et al., 2008). In contrast to that, only 34% of
haploid spermatids (1N) carried HsChr21, deviating by ∼16% from the maximum
number of HsChr21-positive spermatids as only 50% can inherit the aneuploid
human chromosome. This illustrates that there is only a modest loss of HsChr21-
carrying cells during male meiosis, resulting in the generation of a high number of
aneuploid spermatids.
Since our sorting strategy biased the haploid cell population towards earlier
stages with rounder morphology, we wanted to test whether HsChr21-positive cells
might be eliminated during later stages of spermatogenesis. We therefore isolated
mature sperm from caudal epididymides and quantified the number of HsChr21
carrying sperm for 4 Tc1 males (Figure 3.14A and B).
Results 79
Figure 3.14 | Genotyping of spermatogenic cell types
A and B| Representative images of FISH stainings of 4N spermatocytes, 1N sper-
matids and mature sperm from Tc0 (A) and Tc1 (B) animals using a HsChr21-specific
probe. White dotted circles highlight HsChr21 signal in mature sperm. Scale bars
are 20µm.
80 Results
Tc0 Tc1
Cells
counted
HsChr21-
positive
Cells
counted
HsChr21-
positive Percentage
4N
spermatocytes 236 n.d. 3848 3615 93.9%
1N spermatids 303 n.d. 3894 1333 34.2%
Mature sperm 131 n.d. 680 191 28%
Table 3.4 | Quantification of FISH stainings
On average we found 28% of sperm still carrying HsChr21, showing only a minor
drop from the number of HsChr21-positive spermatids (Table 3.4). This shows that
Tc1 males have an unexpectedly high number of aneuploid sperm and while there is
some specific loss of HsChr21-carrying cells during spermatogenesis, this cannot
fully account for the low transmission rate of HsChr21 through the male germline.
3.2.5 Activation of repeat elements on human chromosome 21
Previous studies in the Tc1 mouse have shown that primate-specific repeat
elements encoded on HsChr21 are activated in somatic tissues in comparison to
humans (Ward et al., 2013). Spermatogenesis involves major chromatin remodelling
which temporarily leads to a genome-wide transcriptional de-repression and therefore
holds the potential to activate additional repetitive elements on HsChr21.
To assess the epigenetic state of HsChr21 in the male germline, we performed
ChIP-Seq from whole testis lysates from Tc1 mice and humans using H3K4me3
as a proxy for transcription initiation. In line with previous studies reporting a highly
transcriptionally active chromatin landscape in testis (Soumillon et al., 2013), the
number of sites that show H3K4me3 occupancy is significantly larger in testis
compared to liver. Genome-wide, we found a total number of 59,420 and 40,690
peaks in testis, compared to 18,231 and 16,417 peaks in liver of Tc1 mouse and
humans, respectively (present in at least two replicates). For HsChr21 in particular
we detect a total of 517 peaks in testis compared to 176 peaks in liver (peaks present
in at least half of all replicates) (Figure 3.15).
Differential binding analysis between Tc1 and humans showed that the majority
of regions show a shared occupancy for H3K4me3. However, as previously reported
(Ward et al., 2013), a substantial number of regions on HsChr21 gain an active
chromatin state in the Tc1 mouse compared to humans, which affects ∼32% of all
Results 81
Figure 3.15 | H3K4me3 enrichment on HsChr21 in human and Tc1 mouse
Differential binding analysis of H3K4me3 occupancy between human and Tc1 mouse
in liver and testis. ChIP-Seq mean read concentration (log2) is plotted and signifi-
cantly differentially bound sites with a fold change greater than 3.5 and an FDR of
less than 0.1 are highlighted in blue.
regions within a tissue (55 out of 176 regions in liver compared to 170 out of
517 regions in testis). It is noteworthy that out of the 55 sites identified in liver,
the majority (31) are differentially bound with stronger binding in Tc1 compared to
human, whereas only 24 regions represent novel binding sites in the Tc1 mouse. In
contrast to that, the vast majority of sites (139 out of 170) in testis are novel binding
sites in Tc1 compared to humans, with only 31 sites showing differential binding.
We then overlapped H3K4me3-enriched sites on HsChr21 from liver and testis
with the repeat annotation for HsChr21 obtained from RepeatMasker (Smit et al.,
2015) and found that the number of regions that harbor a repeat element in a 50bp
window around the peak summit is drastically increased in testes (Figure 3.16).
Interestingly, repeat enrichment was not only observed for Tc1-specific sites as is the
case in liver, but also shared regions displayed a higher repeat-association in testes
(132 out of 343). This allowed us to define four different sets of repeat elements on
HsChr21: Set 1 that showed activation in Tc1 and human testis but silenced in liver
(108); Set 2 that showed Tc1-specific activation in liver but was active in both human
and Tc1 testis (17); Set 3 that showed Tc1-specific activation in both liver and testis
(8); Set 4 that showed Tc1-specific activation only in testis (73).
82 Results
Figure 3.16 | Activation of repeat elements in Tc1 mouse liver and testes
ChIP-Seq intensity heatmaps for liver and testis displaying regions with shared
enrichment between human and Tc1 mouse or Tc1-specific sites as identified via
differential binding analysis. Regions in heatmaps are sorted by descending signal
strength in the Tc1 mouse, each row representing a 5-kb window centred on the
H3K4me3 peak summit. 50bp-windows centred on the peak summit were then
overlapped with the RepeatMasker annotation revealing a high repeat-association of
H3K4me3-enriched sites in testis as well as at Tc1-specific sites in the liver. Repeat
families are depicted in the form of bar chart and the overlap between liver and testis
is illustrated by lines.
We then looked at the composition of repeat families within these subsets and
estimated their evolutionary age based on the number of substitutions from the
repeat consensus (Figure 3.17A). Set 1 contained repeat elements from all different
subfamilies, however the majority, with the exception of AluY and AluSx elements,
were evolutionary old elements with an average age of ∼100 million years. This is
representative of the genome-wide transcriptional activation during spermatogenesis.
Results 83
Figure 3.17 | Activation of distinct repeat elements on HsChr21 in liver and
testes
A| Different sets of repeat elements were defined based on their activity in liver and
testis compared between human and Tc1 mouse. Number of repeat elements with
active chromatin state per repeat set is displayed for individual subfamilies (circle
size) and repeat age was estimated by dividing the number of substitutions from the
repeat consensus by the estimated mutation rate for mammalian species (2.2x109
per bp per year) (colour gradient).
B| Top five de novo sequence motifs enriched in different repeat sets displayed with
enrichment value, number of matched sites and best transcription factor match.
84 Results
In contrast to that, the majority of repeat elements in Set 2 and 3 belonged to the
LTR family of a younger evolutionary age. Set 4, which contains repeat elements that
only appear active in Tc1 testis again contained elements across all repeat families,
however compared to the repeats from Set 1 that are also active in human testis,
the elements had an average age of ∼60 million years.
Interestingly, out of the different repeat subfamilies, the youngest human-specific
repeat elements (L1PAs and SINE-VNTR-Alus (SVAs)) were exclusively found in
an active chromatin state in the Tc1 mouse (Set 3 and 4). This suggests that these
young elements still pose a threat to the human genome and need to be silenced
even under the more permissive chromatin state during spermatogenesis.
To test whether the transmission of HsChr21 through the male germline and the
transient activation of human-specific repeats in the testes could cause persistent
activation throughout fertilisation and development, we profiled H3K4me3 in livers of
male germline-derived Tc1 mice. We then assessed the signal enrichment across
regions that have H3K4me3-enrichment in Tc1 testes, separating between regions of
shared enrichment in human and Tc1 mouse and Tc1-specific sites (top and bottom
panel, respectively) (Figure 3.18). This showed that male germline-derived offspring
display the same pattern of transcription initiation as female-germline derived mice,
demonstrating that the transient activation in testes does not alter the epigenetic
state of these repeats in the offspring. To ensure that this is not specific to liver,
we profiled H3K4me3 in a number of adult somatic tissues to cover all three germ
layers. However, as observed for liver (endoderm), male and female germline-
derived offspring displayed identical patterns for transcription initiation also in kidney
(mesoderm), muscle (mesoderm) and brain (ectoderm) (Figure 3.19). These results
argue that the epigenetic remodelling associated with male germline passage does
not alter the transcriptional deployment of HsChr21 in the offspring.
To gain further insight into what might distinguish repeat elements that are
activated only in Tc1 testes but not in liver, we tested whether any of these repeats
carry a particular DNA binding motif as it was observed for B2 SINEs in the mouse
(Schmidt et al., 2012). Motif analysis on the different repeat sets as well as repeat
subfamilies using MEME revealed an enrichment for the CCAAT binding motif in
repeat elements belonging to Set 2 (amongst others), which was caused by the high
number of LTR12 elements in this set (Figure 3.17B). CCAAT is the binding motif
for Nuclear transcription factor (NF-Y) (also known as CCAAT binding factor (CBF))
and is found at a large number of eukaryotic promoter sequences (Bi et al., 1997;
Romier et al., 2003).
Results 85
Figure 3.18 | Transcription initiation across HsChr21 after male germline trans-
mission
ChIP-Seq intensity heatmaps for H3K4me3 across regions identified via differential
binding analysis between human and Tc1 in testes. Shared regions are plotted in
the top panel, regions with Tc1-specific activation in testes are plotted in the bottom
panel. H3K4me3 signal in livers from female and male germline-derived offspring
were plotted onto to same regions to visualise shared transcription initiation events.
Regions in heatmaps were sorted by descending signal strength in Tc1 testes, each
row representing a 5-kb window centred on the H3K4me3 peak summit.
NF-Y is a conserved heterotrimeric TF consisting of NF-YA which confers DNA-
binding specificity and NF-YB and NF-YC which contain histone-fold domains (HFDs)
that allow the formation of a stable histone-like heterodimer that is bound by NF-YA
(Sinha et al., 1995). The NF-Y complex has been shown to be crucial for embryonic
development (Bhattacharya et al., 2003) and has been described as a pioneer factor,
opening chromatin to make it more accessible for other TFs (Oldfield et al., 2014).
86 Results
Figure 3.19 | Transcription initiation across HsChr21 in multiple tissues
H3K4me3 was profiled genome-wide in kidney (A), brain (B) and muscle (C). Active
regions on HsChr21 are shown as heatmaps of ChIP-Seq intensities, each row
representing a 5kb window around a H3K4me3 peak summit and regions being
sorted by descending signal strength in tissues from female germline-derived Tc1
mice.
However, NF-Y was also found to bind a large number of repetitive elements in
human cell lines, particularly LTR elements (Fleming et al., 2013). To test whether
NF-Y associates with LTR12 elements that show H3K4me3-enrichment in the Tc1
mouse, we performed ChIP-Seq for two of its subunits, NF-YA and NF-YB, in Tc1
livers. As previously described for human cell lines (Fleming et al., 2013), ChIP-Seq
for NF-YB yielded more peaks genome-wide (22,331) compared to NF-YA (5,534),
Results 87
Figure 3.20 | NF-Y enrichment at human and mouse repeat elements in liver
A| ChIP-Seq intensity heatmaps for H3K4me3, NF-YA and NF-YB in livers from Tc1
animals across LTR12 elements encoded on HsChr21. Heatmaps were positioned
based on start location of LTR12 elements, displaying signal 0.5kb up and 1.5kb
downstream.
B| Overlap between ChIP-Seq peaks obtained for NF-YA and NF-YB across the
entire Tc1 genome (mm10 + hg19 HsChr21).
C| NF-Y ChIP-Seq peaks were overlapped with genomic annotation for promoter,
inter- and intragenic regions for the mouse genome and HsChr21.
D| NF-YB peak summits were overlapped with the RepeatMasker annotation for
the mouse genome and number of repeats for different families and different LTR
subfamilies are displayed.
however the two factors showed a near perfect overlap with 99.8% of NF-YA peaks
being present in NF-YB ChIP-Seq libraries. We then plotted ChIP-Seq intensity
signal for NF-YA and NF-YB across all LTR12 elements that can be assessed on
HsChr21 (30) and found that all elements that showed Tc1-specific enrichment for
H3K4me3 in the liver were bound by NF-Y (Figure 3.20A and B).
88 Results
Interestingly, NF-Y showed a particular binding pattern located further away from
the repeat start site for LTR12C elements, whereas it was found more towards the
center region of LTR12D repeats. We then checked whether a similar enrichment of
NF-Y at repetitive elements can be observed across the mouse genome, or whether
NF-Y binding to these human-specific LTR elements could represent a species-
specific function of this factor. Although occurring at a lower frequency than reported
for K562 cells in human (Fleming et al., 2013), we find an enrichment of repetitive
elements at NF-Y binding sites also in the mouse. For the smaller dataset obtained
for NF-YA, a large proportion of peaks (83%) was found at promoter sequences
(Figure 3.20C). In contrast to that, a higher proportion of peaks (47%) identified
for NF-YB was found at intra- and intergenic regions. We then overlapped the
peak summits of NF-YB-bound sites with the RepeatMasker annotation for the
mouse genome and found that particularly intra- and intergenic regions overlap
with repetitive elements (1046 out of 5989 (17%) and 1239 out of 4892 (25%)
respectively). As described for humans, LTR elements were overrepresented at
NF-YB binding sites, with a particular enrichment of the ERVK subfamily (Figure
3.20D). This shows that the binding of NF-Y to repetitive elements is not a species-
specific feature, but occurs in both human and mouse. Given the described function
of NF-Y as a pioneer factor, binding of NF-Y to LTR12 elements on HsChr21 could
lead to a more open chromatin state and be the cause of the observed H3K4me3
enrichment. However, we do not have the corresponding binding profile for NF-Y in
human liver and can therefore not exclude the possibility that these elements are
equally enriched for NF-Y in humans.
3.2.6 Transcriptional deployment of human chromosome 21
after male germline transmission
We considered the possibility that the process of decondensation during sper-
matogenesis could adversely affect other layers of transcriptional control on the
human chromosome, including enhancers, promoters, DNA methylation, tissue-
specific transcription factor binding, and the core transcriptional machinery.
Using liver as a representative somatic tissue, in addition to H3K4me3 (marking
active transcription initiation at promoters) we profiled the genomic occupancy of
H3K27ac (active enhancer regions), two tissue specific TFs (CEBPA and HNF4A)
and RNA polymerase 2, as well as total RNA-Seq and non-methylated DNA across
human chromosome 21 in female and male germline-derived Tc1 mice (Figure 3.21).
Results 89
Figure 3.21 | Epigenetic profiling of HsChr21 after male germline transmission
Comparison of ChIP-Seq mean read concentration (log2) across human chromo-
some 21 in livers of female and male germline-derived offspring for (A) H3K4me3
(transcription initiation), (B) H3K27ac (enhancer activity), (C) CEBPA (tissue-specific
TF), (D) HNF4A (tissue-specific TF) and (E) RNA polymerase II. Differentially bound
sites are highlighted in pink representing a fold change greater than 2.5 and an FDR
of lower than 0.1. Pearson correlation was applied to obtain correlation coefficient.
(F) Comparison of DNA hypomethylation patterns between male and female germline
offspring profiled by BioCAP-Sequencing (log2 read count).
(G) Differential gene expression analysis of RNA-Seq in liver between male and
female germline-derived offspring (log10 mean expression).
Promoter and enhancer activity is highly correlated between male and female
germline-derived mice (r2 = 0.99 and 0.96, respectively) with no significantly differen-
tially bound sites on chromosome 21 (Figure 3.21A and B). We used the presence
of the entire mouse genome in both Tc1 samples as internal technical controls to
evaluate what genome-wide correlation in transcription initiation would be expected
between diploid individuals.
Differential binding analysis across the mouse genome revealed a total of 10 out
of 20,001 sites for H3K4me3 and 471 out of 47,254 sites for H3K27ac that showed
a fold change greater than 2.5 (FDR < 0.1) (Figure 3.22A and B). These modest
differences did not appear to be due to the presence of HsChr21, as these numbers
are comparable to technical noise levels we calculated using previously published
wild-type mouse replicates for H3K27ac (599 out of 41,327) (Figure 3.22C) (Villar
et al., 2015).
90 Results
Figure 3.22 | Differential binding analysis for H3K4me3 and H3K27ac across
the mouse genome in Tc1 and BL6 mice
Results 91
Figure 3.22 | Differential binding analysis for (A) H3K4me3 and (B) H3K27ac across
the entire mouse genome comparing female and male germline-derived Tc1 mice.
(C) To estimate technical noise levels for ChIP-Seq experiments, DBA was performed
on four biological replicates of H3K27ac ChIPs in livers of BL6 wild-type mice that
were randomly assigned to two conditions. Differentially bound sites are highlighted
in pink representing a fold change greater than 2.5 and an FDR of lower than 0.1.
Pearson correlation was applied to obtain correlation coefficient. Identified regions
were annotated as promoter, intragenic and intergenic.
To identify whether different germline passaging would lead to alterations in the
DNA methylation underlying the chromatin, we identified hypomethylated regions on
human chromosome 21 passaged via either egg or sperm using BioCAP-Sequencing
(Blackledge et al., 2012). As with the chromatin marks above, DNA methylation
in adult somatic tissues converges on the same molecular phenotype (r2 = 0.97)
(Figure 3.21F).
Unsurprisingly, the correlation in genomic occupancy between male and female
germline-derived mice for two liver-specific transcription factors (CEBPA and HNF4A)
and RNA Pol II were slightly noisier (r2 = 0.72-0.83) (Stefflova et al., 2013); neverthe-
less, almost no sites were identified reliably as differentially bound (Figure 3.21C - E).
In most cases, differences were due to modest changes in overall ChIP intensity, not
the occurrence of entirely novel occupancy in male- or female-derived Tc1 mice. The
noisier correlation of Pol II is likely due to the more distributed nature of polymerase
genome occupancy: Pol II typically binds across tens of kilobases at a comparatively
low intensity, as opposed to more sharply defined regions occupied by modified his-
tones in active regions of the genome. Notably, the modest differences observed in
polymerase occupancy do not impact the transcriptome, which shows exceptionally
high correlation between livers of Tc1 female- and male-derived offspring (r2 = 0.97),
with no genes identified as differentially expressed.
Overall, despite the extensive epigenetic remodelling and chromatin condensation
associated with male germline transmission, human chromosome 21 is accurately
deployed in multiple tissues during development by the mouse epigenetic machinery.
92 Results
3.3 Discussion
The creation of haploid gametes is one of the most tightly regulated processes in
cell biology, as failure to accurately evaluate DNA content can result in catastrophic
organismal aneuploidies. As such, embryonic aneuploidies are one of the leading
causes for spontaneous abortions during human pregnancy, with only a few, such as
trisomy 21, being viable but resulting in developmental defects (Hassold and Hunt,
2001).
Aneuploid female mice that carry large amounts of exogenous DNA are generally
fertile and can pass the aneuploid DNA on to their offspring. In contrast, aneuploid
male mice usually have strongly suppressed fertility or are entirely sterile (Borowski
et al., 2000). Male sterility is commonly observed amongst transchromosomic mouse
models and is attributed to the presence of an extra chromosome rather than the
trisomic gene content, causing spermatogenic arrest in meiosis (Hernandez and
Fisher, 1999). Previous studies in the Tc1 mouse, investigating the pachytene
stage of meiosis showed that the presence of HsChr21 does not lead to increased
pachytene apoptosis in epithelial stage IV tubules (Mahadevaiah et al., 2008). How-
ever, whether this would allow the production of viable sperm and the transmission
of chromosome 21 to the next generation was not tested.
We demonstrate that mouse male meiosis can produce sperm that both carry
and transmit the complete 42 Mb copy of human chromosome 21 to generate viable
aneuploid offspring. Consistent with the reduced fertility generally observed for ane-
uploid male mice, Tc1 testes showed macro- and microscopically visible disruptions
to their tissue architecture. Nevertheless, despite frequent spermatogenic arrest
at metaphase I due to congression defects resulting in increased apoptosis, the
majority of Tc1 males tested in this study were able to produce viable aneuploid
offspring, albeit at a lower frequency. The observed lower transmission rate of
HsChr21 through males, however, cannot be fully accounted for by the reduced
number of HsChr21 positive cells produced during male meiosis. This indicates that
failures are likely to occur further downstream, i.e. at fertilisation or during early em-
bryonic development, eventually resulting in only ∼13% aneuploid offspring. While
female meiosis is more error-prone than male meiosis due to weaker checkpoint
mechanisms (Morelli and Cohen, 2005), the percentage of aneuploid cells produced
via male meiosis (34%) is strikingly similar to the percentage of aneuploid offspring
generated via female germline transmission.
Results 93
Attempts to establish the Tc1 chromosome on three different inbred genetic back-
grounds led to a complete loss of HsChr21 over only a few generations (O’Doherty
et al., 2005). This loss was independent of the sex of the transmitting parent and
stable transmission was only observed when Tc1 mice were crossed with F1 hybrids
between BL6 and 129S8 mice. It is currently unclear why a heterozygous genetic
background is necessary to maintain stable transmission of HsChr21, or to what
extent strain-specific checkpoint mechanisms may be involved. Studies in XO fe-
males have suggested that C3H mice have a weaker SAC compared to C57/BL6
(LeMaire-Adkins and Hunt, 2000; Nagaoka et al., 2011).
Equivalent data for the 129S8 strain is not available, however genetic background
has been implicated in the length of the synaptonemal complex and the number
of meiotic crossovers (Vranis et al., 2010). Interestingly, a study investigating the
spermatogenic function of piRNA pathway protein Maelstrom resulted in increased
non-meiotic DNA damage and failure in chromosome synapsis on a BL6 background,
whereas back-crossing to the 129Sv/Jae strain allowed the completion of meiosis
and partial spermiogenesis despite carrying the same genetic defect (Castañeda
et al., 2014). While the underlying genetic or molecular causes for this phenomenon
are not clear, this suggests that there are striking differences between inbred mouse
strains that affect the meiotic process.
The first successful passage of human DNA via the mouse germline followed the
development of microcell-mediated chromosome transfer of human chromosome
fragments into mouse ES cells. Mice derived from these ES cells were able to
passage fragmented regions of human DNA via the female and occasionally male
germline (Tomizuka et al., 1997). To date, the largest successful and stable germline
transmission of human DNA via mouse sperm was of a circularised 5-10 Mb human
artificial chromosome (Voet et al., 2001); of which a linearised version containing
short telomeres can also be passaged via sperm (Weuts et al., 2012).
94 Results
The successful passage of the human chromosome through the male germline
held the potential to further unmask human-specific repetitive elements, because of
the genome-wide transcriptional activation occurring during spermatogenesis and
the assumed absence of species-specific mechanisms that co-evolved to repress
human repeats (Jacobs et al., 2014; Zamudio and Bourc’his, 2010). However, we did
not detect any additional repeat elements that gained an active chromatin state in
somatic tissues following male germline transmission of HsChr21. Instead, extensive
epigenetic profiling of the chromatin landscape in livers of male germline-derived
offspring revealed that, despite the fundamentally different developmental dynamics,
male and female germline transmission resulted in indistinguishable transcriptional
and regulatory phenotypes in this tissue.
4 | Adaptive piRNA response against
human-specific repeat elements
smallRNA-Seq analysis was assisted by Dr. Tim Rayner.
4.1 Introduction
PIWI-interacting RNAs (piRNAs) are a class of germline-specific non-coding
RNAs that are critical for germ cell development. One of the main functions of
piRNAs in the gonads is the silencing of repetitive elements that become activated
as a result of epigenetic reprogramming during germ cell development (Siomi et al.,
2011).
In evolutionary terms, the mammalian piRNA pathway is the most dynamic
genome defence mechanism against invading repetitive elements and mediates si-
lencing of the youngest retroelements. This is in contrast to other adaptive pathways,
such as KRAB zinc finger proteins, that require prolonged co-existence to establish
silencing of new retroelements as the mechanism relies on gene duplication and
mutation (Molaro and Malik, 2016). However, the actual timing of adaptation of the
mammalian piRNA machinery to newly invading repeat elements is largely unknown.
Meiotic piRNAs that start their expression at the pachytene stage of the first
meiotic division are depleted of repetitive sequences compared to the class of
piRNAs that are expressed during embryonic development, with only ∼20% of
murine meiotic piRNAs mapping to repetitive elements (Girard et al., 2006). Yet,
this relatively small fraction of meiotic piRNAs fulfils a crucial function in genome
defence during meiosis when genome-wide chromatin changes lead to a loosening
of repeat repression.
It has been shown that upon its re-expression at the zygotene-to-pachytene
transition, Mili is involved in post-transcriptional silencing of L1 transcripts together
with Miwi, which is expressed during pachytene and is the sole PIWI protein for
most of post-meiotic development in round spermatids (Di Giacomo et al., 2013;
Reuter et al., 2011). While meiotic piRNAs in mammals indisputably have functions
beyond transposon control (Gou et al., 2014; Zhang et al., 2015), their role in post-
transcriptional silencing of repetitive elements is crucial to ensure genome stability.
96 Results
This is demonstrated by the effect of slicer-inactive Mili mutants or Miwi knockout
mice, which both show defects in spermatogenesis that is accompanied with an
increase in L1 transcripts (Di Giacomo et al., 2013; Reuter et al., 2011).
To address the adaptability of this pathway, we exploit the trans-chromosomic
mouse model Tc1 that carries a copy of human chromosome 21. As previously
described, several human-specific repeat elements are activated in this mouse
model, both in somatic and germline tissues. Particularly in the male germline, we
observed a significant activation of human repeat elements that do not gain an active
chromatin state in humans. We can therefore utilise transcripts arising from these
species-specific repeats to test whether foreign repeat transcripts are recognised
by the mouse piRNA machinery to trigger an adaptive piRNA response and test
whether the piRNA system can function as an impromptu genome defender.
4.1.1 Aim
The aim of this project was to test whether the transcriptional activation of human
repeat elements in the male germline can trigger an adaptive piRNA response in the
mouse and lead to the generation of novel piRNA sequences.
Results 97
4.2 Results
4.2.1 piRNA expression in the Tc1 mouse
The previously described sub-fertility phenotype observed in Tc1 males is distinct
from the spermatogenic defects observed in PIWI protein mutants, which all result in
complete male sterility at different stages during spermatogenesis (Carmell et al.,
2007; Deng and Lin, 2002; Kuramochi-Miyagawa et al., 2004). In contrast to MILI
and MIWI2 mutants that arrest at the early pachytene stage, or MIWI mutants that
arrest at the round spermatid stage, spermatogenesis is impaired at metaphase of
the first meiotic division in Tc1 animals (Section 3.2.2).
To test whether the overall expression of piRNAs is impacted in Tc1 animals,
we profiled smallRNAs from adult males using whole testis RNA extractions and
compared wild-type with Tc1 animals. Due to the relatively short length of smallRNAs
that could cause extensive cross-mapping of mouse sequences to HsChr21, we
filtered out all smallRNAs reads from wild-type animals that aligned to HsChr21. This
removed on average 0.1% of total reads and included the majority of reads mapping
to the five microRNA genes encoded on HsChr21 (Table 4.1), as these sequences
are highly conserved between human and mouse and it is therefore not possible to
distinguish whether these reads originate from the mouse or human genome.
Comparison of the size distribution of smallRNA reads between Tc0 and Tc1
animals did not show any notable differences between the two genotypes, suggesting
that there is no large-scale shift in the expression of different smallRNA classes.
We observed a pronounced peak around 30nt, which corresponds to the size of
pachytene piRNAs (29-32nt), which make up the majority (90%) of the piRNA pool
in adult animals (Figure 4.1A) (Li et al., 2013). We then selected smallRNA reads
between 24 and 32 nucleotides in length, corresponding to the size range of MILI-
and MIWI-bound piRNAs for further analysis, which represented between 82-91% of
total reads (Table 4.1).
Pachytene piRNA clusters within the genome are not characterised by specific
genomic sequence features, but are defined by piRNA sequences mapping to
a genomic location at a given density. As such, sequence conservation within
piRNA clusters is extremely low across species, however, piRNA clusters tend to
be conserved in syntenic regions across mammalian genomes (Aravin et al., 2006;
Chirn et al., 2015; Girard et al., 2006; Hirano et al., 2014).
98 Results
Figure 4.1 | Size distribution of smallRNA libraries and piRNA cluster calling
optimisation
A| Size distribution of smallRNA libraries from whole testis lysates of wild-type and
Tc1 males.
B| peakDepth optimisation for piRNA cluster calling on all mouse smallRNA libraries.
To define piRNA clusters in our smallRNA libraries, we used a sliding window
approach, counting reads falling into 1kb windows. We determined the appropriate
peakDepth (threshold for minimum number of reads falling into 1kb window) for our
libraries by sampling through different numbers and smallRNAs and determining the
total number of reads captured in piRNA clusters that are called at a given peakDepth
Results 99
Before filtering After filtering After size selection
Tc0
do3316
do3731
do3733
do3734
do3737
do3739
do3741
do3742
39,085,132
25,481,676
14,296,350
18,037,645
16,808,944
33,110,031
22,042,573
20,348,837
39,054,104
25,460,847
14,275,458
18,021,922
16,794,776
33,076,402
22,022,095
20,335,360
34,027,682
23,187,064
12,278,896
16,188,434
15,144,289
28,772,330
19,711,087
18,438,335
Tc1
do3317
do3469
do3474
do3728
do3730
do3735
do3736
do3738
do3740
do3743
36,077,146
26,919,725
36,462,917
20,766,837
16,779,287
18,777,916
19,357,213
15,085,161
35,476,140
41,874,913
36,045,202
26,895,371
36,424,383
20,743,805
16,756,345
18,755,125
19,329,489
15,071,972
35,440,758
41,840,014
30,545,464
23,468,319
31,889,582
17,753,226
13,789,693
16,066,693
15,856,901
13,617,568
30,748,599
37,781,312
Table 4.1 | Number of reads in smallRNA libraries before and after filtering
(Figure 4.1A). This showed that for all of our mouse libraries, a peakDepth of 50,
requiring a minimum of 50 individual smallRNA sequences mapping per 1kb window
yielded good results, with the number of reads falling into piRNA clusters being
comparable when adjusted to library size.
We then performed piRNA cluster calling on Tc0 and Tc1 smallRNA libraries
using these parameters and identified 334 and 394 piRNA clusters that showed
concordant expression across Tc0 and Tc1 replicates, respectively. The majority of
Tc0 clusters were also called in Tc1 samples, with 324 clusters overlapping between
the two genotypes and most clusters that were unique to one condition showed
low expression values (Figure 4.2A). We then performed differential expression
analysis across all piRNA clusters which showed that only 42 out of the 403 clusters
were differentially expressed based on an FDR smaller than 0.001 (21 each up- or
down-regulated). However, none of these clusters had a log2 fold change greater
than 1 and therefore only display minor changes in gene expression (Figure 4.2B).
This suggests that the mouse piRNA machinery is fully functional in Tc1 animals.
100 Results
Figure 4.2 | Mouse piRNA cluster expression
A| Venn diagram displaying the overlap between piRNA clusters called in Tc0
and Tc1 libraries. Only concordant clusters showing consistent expression across
replicates were considered.
B| Differential expression analysis across the total of 403 concordant piRNA clusters
between Tc1 and Tc0 animals. Differentially expressed clusters with a FDR>0.001
are highlighted in green.
In accord with this, we performed differential expression analysis on total RNA
from whole testes for protein-coding genes and repeat families and observed no
major changes in gene expression between wild-type and Tc1 animals except for
human genes encoded on HsChr21 (Figure 4.3A and B). Indeed, all major piRNA
pathway components were expressed at similar levels between the two genotypes
and IF stainings for the two major PIWI proteins expressed in adult testes, MILI
and MIWI, confirmed the accurate localisation of these proteins within seminiferous
tubules. MILI was found in the cytoplasm of spermatocytes, whereas MIWI signal
was strongest in round spermatids where it localised to distinct foci corresponding
to RNA processing center in these cells termed chromatoid body (Figure 4.3C and
D). Repeat analysis revealed no major upregulation of individual repeat classes,
which is in contrast to MILI or MIWI knockout mouse models that show increased
L1 expression (Di Giacomo et al., 2013; Reuter et al., 2011), suggesting that these
repeats are accurately silenced by the mouse piRNA machinery also in Tc1 animals.
These findings together showed that the mouse piRNA pathway functions unper-
turbed throughout spermatogenesis in Tc1 males.
Results 101
Figure 4.3 | Genome-wide gene expression profile and repeat expression in
Tc1 animals
A| MA plot displaying differential gene expression between Tc1 and Tc0 whole testis
lysates. Genes differentially expressed with a log2 fold change greater than 1 and
an FDR lower than 0.001 are highlighted in green. Human genes are displayed as
diamonds and mouse genes are displayed as circles. Genes with a log2 fold change
greater than 1.5 are labeled.
B| MA plot displaying differential expression of repeat elements between Tc1 and
Tc0 whole testis lysates. Repeats with a log2 fold change greater than 0.25 and
an FDR lower than 0.001 are highlighted in green and different repeat classes are
labeled.
C| Heatmap displaying log2 normalised read counts for piRNA pathway genes in
Tc0 and Tc1 replicates.
D| Representative immunofluorescent image of seminiferous tubules stained with
Hoechst (blue), anti-MILI (red) and anti-MIWI (green) antibodies. Primary spermato-
cytes are marked with orange triangle and round spermatids are marked by purple
triangle; scale bar is 50µm .
102 Results
4.2.2 Expression of exogenous piRNA sequences from human
chromosome 21 in the mouse
We then wanted to examine the piRNA expression from HsChr21 in Tc1 ani-
mals, as previous studies suggest that human piRNA clusters, when inserted into
the mouse genome, can be successfully deployed (Goh et al., 2015). Published
annotations for human piRNA clusters include one bi-directional piRNA cluster on
HsChr21, which is conserved in syntenic regions in the mouse and rat (Girard et al.,
2006).
To confirm the genomic location of this cluster we generated smallRNA libraries
from human testis total RNA obtained from Ambion for different biological replicates.
Performing the same filtering as described for mouse libraries, we observed that
human smallRNA libraries contained a much higher percentage of reads in the length
between 19 and 22 nucleotides with a sharp peak at 21 nucleotides, which corre-
sponds to the length of microRNAs and endo-siRNAs. While this could represent a
difference in abundance of testis microRNAs between mouse and human, we cannot
exclude the possibility that this observation is due to differences in RNA isolation,
since we used commercially available human RNA. After size-selection for smallRNA
reads between 24-32nt, only 11-55% of the original smallRNA reads remained for
downstream analysis (Table 4.2). Performing the same piRNA cluster analysis on our
human smallRNA libraries also identified the conserved cluster on HsChr21, being
the tenth most highly expressed cluster out of 359 clusters called across the human
genome. However we did not detect this cluster in our Tc1 libraries using the same
threshold. Lowering the threshold parameter to 25 reads per 1kb window instead
of 50 yielded a cluster in Tc1 libraries that overlapped in part with the coordinates
of the cluster called in human samples. Calculating the RPKM value for individual
replicates showed that the cluster on HsChr21 has extremely low coverage in Tc1
animals (Figure 4.4B).
Before filtering After filtering After size selection
Hsa
do3814
do4040
do4114
do8249
do8250
14455591
14266290
16021153
6867257
10628270
14413554
14251711
15950305
6839427
10577750
6996363
9175028
4921789
2755769
1223000
Table 4.2 | Number of reads in human smallRNA libraries before and after filtering
Results 103
Figure 4.4 | piRNA cluster on human chromosome 21
A| Size distribution of human smallRNA libraries. Size-selected reads are highlighted
in orange.
B| RPKM value for piRNA cluster on HsChr21 in individual replicates of human and
Tc1 smallRNA libraries.
C and D| Coverage plot over piRNA cluster on HsChr21 for human and Tc1 mouse.
However, plotting coverage across the piRNA cluster showed that the overall
pattern of expression and bi-directionality is very similar between human and Tc1
libraries (Figure 4.4C and D). The low coverage in the Tc1 mouse is most likely
due to an overall lower expression of HsChr21, since Tc1 mice only have one copy
which, in addition, is affected by meiotic silencing, the transcriptional silencing of
unpaired chromosomes occurring at the zygotene-to-pachytene transition during
meiotic prophase I (Mahadevaiah et al., 2008).
104 Results
Figure 4.5 | smallRNA coverage across chromosomes
A| Scatter plot displaying the mean normalised counts per mouse chromosome for
Tc1 and Tc0 libraries. Mean normalised reads from Tc1 samples for HsChr21 were
plotted against mean normalised reads in human samples (Read counts are not
normalised by chromosome length).
B| Mean normalised read counts across chromosomes for total RNA (light green
circles) and smallRNA (dark green circles) libraries from FACS-sorted Tc1 sperma-
tocytes and spermatids. Zoom-in windows highlight chromosomes with low read
counts.
However, the two chromosomes that are naturally affected by meiotic silencing
in male germ cells, X and Y, (Turner, 2007) are depleted of piRNA clusters in the
mouse (Girard et al., 2006; Li et al., 2013), and thus, the effect of meiotic silencing
on piRNA clusters has not been explored previously. To assess the overall levels
of smallRNAs across the genome, we calculated how many reads mapped to each
chromosome and observed that indeed the three heteromorphic chromosomes in
Results 105
the Tc1 mouse - X, Y and HsChr21 - captured the lowest number of smallRNA
reads, whereas HsChr21 in human libraries had comparable read numbers with
other autosomes (Figure 4.5A).
Meiotic silencing, particularly affecting the X chromosome, was shown to be more
permissive post-meiosis in round spermatids compared to spermatocytes (Mueller
et al., 2008). Since primary spermatocytes represent the majority of cells within Tc1
testes (Figure 3.6), we wanted to test whether we would observe increased signal
for HsChr21 when specifically enriching for post-meiotic cells. For this purpose we
purified primary spermatocytes and round spermatids via FACS sorting as previously
described and prepared smallRNA libraries from 250,000-500,000 cells. We then
calculated the read distribution across chromosomes comparing between the two
cell types and indeed observed a slight increase in the number of reads mapping to
HsChr21 in spermatids compared to spermatocytes, which was more pronounced
for chromosome X (Figure 4.5B). Since the dynamics of post-meiotic silencing on
piRNA clusters are largely unknown, we used total RNA-seq libraries from the same
cell types to look at the level of de-repression in round spermatids. We observed a
much stronger increase in reads mapping to the X chromosome in spermatids over
spermatocytes as well as a moderate increase in reads for HsChr21 in total RNA-seq
data. This could indicate different dynamics in post-meiotic silencing for protein-
coding genes and piRNA-encoding genomic locations. However, the re-expression
of specific genes on chromosome X post-meiosis is essential for spermatogenesis
and might therefore represent a more targeted de-repression, whereas none of the
genes encoded on HsChr21 are essential for spermatogenesis in the mouse.
The isolation of specific cell types increased the number of smallRNA reads
mapping to HsChr21 only minimally; however, performing piRNA cluster analysis
in these libraries confirmed the reproducibility of our previously identified piRNA
producing loci including the cluster on HsChr21 which was also detected in the
spermatid population (using a peakDepth of 25 reads per 1kb window). Overall,
concordant clusters detected in Tc1 spermatocytes and spermatids overlapped to
∼90% (Figure 4.6A). Differentially called clusters had generally low read counts and
are therefore most likely due to thresholding issues. All but two of the piRNA clusters
identified in the sorted cell populations were also present in the clusters detected in
whole tissue lysates.
106 Results
Figure 4.6 | piRNA cluster detection in sorted cell populations
A| Venn diagram displaying the overlap between piRNA clusters identified in Tc1
spermatocytes (dark purple), Tc1 spermatids (light purple) and Tc1 whole testis
lysates (green).
B| Average read counts per piRNA cluster are displayed in violin plots for piRNA
clusters that are commonly identified across all samples and piRNA clusters that are
only detected in whole tissue lysates but not in sorted cell populations.
Comparing the average read counts between clusters that are commonly identi-
fied across sorted cell types and whole tissue lysates and clusters that are uniquely
identified in whole tissue lysates revealed a much stronger expression of the clus-
ters that are consistent across samples (Figure 4.6B) suggesting that these 206
commonly identified piRNA clusters are high confidence piRNA loci.
In sum, these findings suggest that meiotic silencing, which persists into post-
meiotic cells, drastically reduces the amount of piRNAs produced from HsChr21.
However, the pattern observed for the conserved human piRNA cluster in the Tc1
mouse, although at a considerably lower expression level, is highly similar compared
to humans.
4.2.3 Tc1-specific piRNA sequences originating from human
repeat elements
Despite the low read counts for HsChr21 we wanted to test whether we can
detect Tc1-specific piRNA sequences originating from human repetitive elements.
We therefore calculated the number of reads mapping to the different repeat families
encoded on HsChr21 in both human and Tc1 whole testis smallRNA libraries.
Results 107
Figure 4.7 | Enrichment of Tc1 smallRNA reads across ERVL-MaLR elements
A| Heatmap displaying the normalised smallRNA counts (log2) across repeat families
encoded on HsChr21 from Tc1 and human smallRNA libraries.
B| Heatmap displaying the normalised smallRNA counts (log2) across ERVL-MaLR
subfamilies on HsChr21.
While piRNAs originating from piRNA clusters are mostly uniquely mapping to
the genome, a smaller fraction of ∼20% pachytene piRNAs is repeat-derived (Girard
et al., 2006). To ensure that we can confidently identify the locus from where such
repeat-derived piRNAs were generated, we only used reads that mapped to the
reference genome with up to two mismatches but no alternative mapping locations.
Comparison between the different repeat classes using the effective library size to
normalise (reads mapping to repetitive elements on HsChr21) revealed an increase
in Tc1-derived piRNAs mapping to the LTR/ERVL-MaLR (mammalian apparent LTR-
retrotransposon (MaLR)) family (Figure 4.7A). When examining the distribution of
Tc1-derived piRNAs across the different ERVL-MaLR subfamilies, we observed that
the vast majority of reads mapped to MSTA and MSTA-int elements, with a few reads
mapping to THE1D and THE1D-int (Figure 4.7B). In contrast to that human piRNAs
mapping to the ERVL-MaLR family were spread out across different subclasses with
the only enrichment being observed in THE1D elements.
MSTA and THE1 elements are primate-specific LTR elements (Smit, 1993) and
are the main component of LTR megasatellite arrays of which there are ∼25 in the
human genome (Warburton et al., 2008). Such megasatellites consist of ∼3.5kb
tandem duplicated repeat units, containing MSTA, MSTA-ints (internal open reading
frames) and THE-1 LTR elements. One such LTR array of ∼54kb in size is found in
the pericentromeric region of HsChr21 and the terminal regions of this LTR array
were called as piRNA clusters in our Tc1 whole testis libraries at a peakDepth
108 Results
of 25. Plotting the coverage of smallRNA reads across the LTR array showed a
specific pattern in Tc1 libraries with distinct peaks at the terminal regions, whereas
human smallRNA libraries showed no enrichment over this region with only ∼0.4%
of all reads that map to HsChr21 falling into the megasatellite regions (Figure 4.8A).
To determine the identity of these Tc1-specific sequences, we checked the size
distribution of reads mapping to the LTR array and found that the majority of reads
from the Tc1 mouse were between 28-31 nucleotides in length, corresponding to the
characteristic size of MIWI-bound pachytene piRNAs (Figure 4.8B).
In contrast, smallRNAs originating from human libraries did not show a distinct
pattern, however the number of reads obtained for this region is extremely low.
Generating a position weight matrix for the first 20 nucleotides of all smallRNA reads
mapping to the LTR array revealed a strong bias for uridine at the first position in
Tc1 libraries, a characteristic feature of PIWI-bound RNAs. In contrast to that no
sequence bias for adenine was observed at the tenth base position, which is a
characteristic feature of secondary piRNAs, therefore suggesting that these are
primary piRNAs and not the products of piRNA-mediated cleavage (Figure 4.8C).
Human smallRNAs also showed a bias for uridine in the first position which was
however less pronounced compared to Tc1 libraries.
To test whether these smallRNA sequences are unique to the Tc1 mouse and
therefore potential novel piRNAs, we screened for identical 5’ ends in human small-
RNA libraries. For this we trimmed all Tc1 reads mapping to the LTR array down to
20nt in length and queried whether identical sequences exist in our human smallRNA
libraries. We found that out of a total of 437 unique 20nt sequences mapping to
the LTR array in the Tc1 mouse, 20 or 65 were also detected in human smallRNA
libraries when using uniquely mapping or multi-mapping human reads, respectively.
In contrast to that, out of a total of 603 unique 20nt sequences that map to the
conserved piRNA cluster on HsChr21 in the Tc1 mouse, 418 or 486 were also
detected in human uniquely mapping and multi-mapping reads, respectively.
To confirm that these Tc1-specific smallRNAs are indeed piRNAs, we wanted
to test their interaction with PIWI proteins. We therefore performed RNA-immuno-
precipitation for the two major PIWI proteins expressed in adult mouse testes, MILI
and MIWI and generated smallRNA libraries from the RNA that was enriched with
these proteins.
Results 109
Figure 4.8 | Tc1-specific piRNA expression across human-specific LTR array
A| Coverage plot for smallRNA reads spanning the LTR array on HsChr21 from
human (top) and Tc1 (bottom) smallRNA libraries. Repeat subunits are presented in
different colours and positioned based on strandness. (Coverage visualisation does
not represent strandness of smallRNA reads - the majority of reads for both human
and Tc1 map to the minus strand).
B| smallRNA length distribution of all reads mapping to the LTR array in human (top)
and Tc1 mouse (bottom).
C| Position weight matrix across the first 20 nucleotides of all smallRNA reads
mapping to LTR array.
We indeed observed an enrichment of the smallRNA sequences mapping to the
LTR array in MIWI immunoprecipitates but not in MILI-IPs, which is in accord with
the size distribution of smallRNAs that we are observing (Figure 4.9A). We further
confirmed these smallRNAs to be pachytene piRNAs by profiling smallRNAs from
testis of 7-day old pups, an age at which no spermatocytes will have reached the
pachytene stage. Consistent with our findings, we did not detect any smallRNAs
mapping to the LTR array in libraries from 7d-old pups (Figure 4.9B), however, we did
observe a higher number of reads mapping to HsChr21, which supports our previous
conclusion that the observed low read count for HsChr21 in adult Tc1 libraries is due
to meiotic silencing which first occurs at the zygotene-to-pachytene transition (Figure
4.9C). These data together suggest that the Tc1-specific smallRNAs mapping to the
LTR array on HsChr21 are novel pachytene piRNAs that are generated in adult Tc1
testes despite meiotic silencing of HsChr21.
We then looked at the chromatin landscape and genomic location of this repeat
element to see whether it displays Tc1-specific activation compared to human.
110 Results
Figure 4.9 | Tc1-specific piRNAs are MIWI-bound and only expressed after
cells enter the pachytene stage
A| Coverage plot for smallRNA reads spanning the LTR array on HsChr21 from MIWI-
and MILI-RIP smallRNA libraries. Repeat subunits are presented in different colours
and positioned based on strandness. (Coverage visualisation does not represent
strandness of smallRNA reads - the majority of reads map to the minus strand).
B| Percentage of reads on HsChr21 that map to the LTR array in different conditions.
C| Scatter plot displaying the mean normalised counts per chromosome for small-
RNAs from adult and 7-day old Tc1 mice. (Read counts are not normalised by
chromosome length).
Using our H3K4me3 ChIP-seq data we found that this repeat does not have an
active chromatin state in the liver, however it showed Tc1-specific activation in testes
with an H3K4me3 peak located at one end of the LTR array (Figure 4.10A).
Results 111
Figure 4.10 | Increased transcription of LTR array in testes from Tc1 mice com-
pared to human
A| Log2 fold enrichment of H3K4me3 ChIP-seq signal over input samples from Tc1
and human whole testis plotted across LTR array. Repeat subunits are presented in
different colours and positioned based on strandness.
B| MA plot displaying differential expression of retroelements encoded on HsChr21
between Tc1 mouse and human from total RNA-Seq libraries from whole testis,
using the effective library size (all reads mapping to HsChr21) for normalisation.
Repeats with log2 fold change greater than 2.5 and an FDR lower than 0.001 are
highlighted in green and labels mark repeat classes with a log2 fold change greater
than 3.5.
C| MA plot displaying differential gene expression, including the LTR array on
HsChr21 between Tc1 mouse and human from total RNA-Seq libraries from whole
testis, using the effective library size (all reads mapping to HsChr21) for normalisa-
tion. Genes with a log2 fold change greater than 1.5 and an FDR lower than 0.001
are highlighted in green and labelled.
112 Results
When looking at the genomic location of this repeat, taking the annotated rear-
rangements of the Tc1 genome into account (Gribble et al., 2013), we found that this
peri-centromeric array was moved to a more central position as a result of HsChr21
becoming meta-centric in the mouse. This rearrangement could potentially allow
a higher transcription of the LTR array in the Tc1 mouse, as it places it into a new
genomic environment, potentially becoming part of a more active chromatin domain.
Comparing the expression levels of repeat families encoded on HsChr21 between
Tc1 mouse and humans from total RNA-Seq showed that there was a higher expres-
sion of mainly LTR elements. However, the specific sub-classes of the LTR array
did not appear upregulated in this approach when using the average expression
of all elements across HsChr21 (Figure 4.10B). We then considered only reads
mapping to the LTR array and compared the expression of this repeat between Tc1
and human alongside protein-coding genes on HsChr21. While we observed an
upregulation of transcripts mapping to the LTR array in the Tc1 mouse (log2 fold
change 1.52), we found a surprisingly high expression of this repeat also in humans
(Figure 4.10C). The relative absence of smallRNAs mapping to this repeat in humans
despite highly abundant transcripts could suggest that there are species-specific
differences in the recognition of these repeat-derived RNAs between human and
mouse.
Taken together, our data suggest that the human-specific LTR array gains a more
active chromatin state and higher transcriptional activity in the Tc1 mouse, leading to
an adaptive piRNA response against this foreign RNA molecule and the generation
of novel pachytene piRNAs.
4.3 Discussion
Defending the germline genome against invading repetitive elements is the unify-
ing function of piRNAs across all animal germlines. In mammals, the piRNA pathway
diverged into two functionally distinct arms that act during different developmental
time windows, the first during embyronic development and the second during sper-
matogenesis. While pre- and perinatal piRNAs function exclusively in the silencing
of repeat elements on both a transcriptional and post-transcriptional level, pachytene
piRNAs, which are expressed during spermatogenesis, have acquired a myriad of
additional functions. Nevertheless, repeat-targeting pachytene piRNAs, which mainly
act via post-transcriptional silencing employing the slicer activity of their associated
PIWI protein, are essential for spermatogenesis.
Results 113
It is currently still unclear how repetitive elements first become targets of piRNA-
mediated silencing and what determines whether they are silenced during embyronic
development or escape transcriptional silencing and re-gain expression at the onset
of meiosis (Manakov et al., 2015; Nagamori et al., 2015; Pezic et al., 2014). Our
current understanding places repeat-associated pachytene piRNAs as a safeguard
mechanism that enforces silencing of repeat elements, mostly L1 elements, that
escape transcriptional silencing during embryonic development (Di Giacomo et al.,
2013; Reuter et al., 2011). However, whether this mechanism requires adaptation,
or whether it can immediately react to newly invading repeat elements is unknown.
This work provides evidence for an adaptive function of the pachytene piRNA
pathway that generates novel piRNA sequences from transcripts arising from a
human-specific repeat element. This suggests that piRNAs can be generated from
new transcripts without any prior adaptation or education of the system.
We observed low, but consistent expression of piRNA sequences originating
from the conserved piRNA cluster on HsChr21 as well as from a peri-centromeric
LTR array which showed similar expression levels. The low expression of HsChr21
was due to meiotic silencing which affected all three chromosomes that do not
synapse during meiotic prophase in the Tc1 mouse, HsChr21 as well as mouse X
and Y (Turner, 2015). While the mouse sex chromosomes are depleted of piRNA
clusters due to negative selection, investigating smallRNA expression in the Tc1
mouse showed that piRNA producing loci are also subjected to meiotic silencing as
demonstrated by the low expression of the piRNA cluster on HsChr21. Two recent
studies investigating the expression of microRNAs from the X and Y chromosome
during meiosis have reached conflicting conclusions regarding the effect of meiotic
silencing on the expression of smallRNAs (Royo et al., 2015; Song et al., 2009).
While we were not able to assess HsChr21-encoded microRNAs due to the high
sequence conservation between human and mouse, our data suggests that, while
not completely shut down, meiotic silencing results in a repression of piRNA clusters.
The low expression of HsChr21 poses a challenge for data analysis and inter-
pretation as the low read numbers are more sensitive to normalisation procedures
and therefore need to be handled with caution. However, despite the low expres-
sion, the piRNA expression from HsChr21 was too consistent and in accord with
biological processes to be due to technical artefacts. The consistent detection of
these sequences in adult testis libraries, sorted meiotic cell populations and MIWI
IPs, but absence in young animals prior to reaching the pachytene stage, strongly
supports that these smallRNA are true piRNA sequences functioning by association
114 Results
with MIWI. It remains however an open question whether the sequences that we
detect are the result of low level transcription across all cells within a tissue or due to
a high expression from a few cells in which HsChr21 self-synapses and does not
undergo meiotic silencing (Mahadevaiah et al., 2008).
Possible ways to examine this would be to perform single-cell smallRNA-Seq
as described in Faridani et al. (2016). However, a general problem with single-cell
sequencing is the capture efficiency, which for total RNA still lies between only 5 -
60% depending on the technology and cell type that are used (Wen and Tang, 2016).
The published protocol only investigates the expression of microRNAs and while
piRNAs are expressed in high numbers, piRNA sequences are much more diverse
than microRNAs resulting in individual piRNA sequences with lower count numbers.
It might therefore not be possible to detect piRNA sequences with lower expression
using this technique, especially originating from HsChr21.
An alternative would be to visualise HsChr21-derived piRNAs via in situ hybridis-
ation to see whether expression is observed in all or only a subset of cells. However,
depending on the nature of the expression, low levels across all cells might not be
detectable using ISH either, as signal amplification is limited due to the short probes.
Previous studies that generated exogenous piRNAs in the mouse either inserted
exogenous sequences into duplicated mouse clusters or transplanted an entire
human piRNA cluster into the mouse to examine the expression (Goh et al., 2015;
Muerdter et al., 2012). In both cases, a strong expression was observed for exoge-
nous piRNA sequences, indicating that the genomic context in which piRNA clusters
are found is less important than the sequence content. The stronger expression
of exogenous piRNAs in these studies compared to our observations suggest that
the insertion of up to 80kb does not result in enough asynapsis to triggers meiotic
silencing of these insertions.
Even though piRNA clusters show very little sequence conservation across
mammals, the fact that more than 80% of piRNAs produced from the human cluster
in Tc1 mice have identical 5’ ends in humans suggests that there is a distinct
processing pattern encoded in the piRNA precursor transcript, which is correctly
interpreted across species. What defines piRNA precursors from other transcripts
in mammals is currently still unknown, but is most likely an RNA-based signal. For
example, G-quadruplex structures were shown to mediate an interaction with the
RNA helicase MOV10L and might be involved in guiding piRNA precursors into the
piRNA biogenesis pathway (Vourekas et al., 2015).
Results 115
The high number of LTR elements that show an increased expression in the
Tc1 mouse compared to human is in accord with the increase in H3K4me3 that we
observed in testes which is mostly enriched at LTR elements. This could be due
to a lack of piRNA-mediated de novo methylation of these repeat elements during
embryonic development. However, this increased expression of repetitive elements
does not appear to cause spermatogenic defects in contrast to repeat upregulation
observed in various piRNA pathway mutants (Castañeda et al., 2014; Di Giacomo
et al., 2013; Manakov et al., 2015; Reuter et al., 2011; Zamudio et al., 2015). This is
most likely due to the relative small size and therefore low number of repeat elements
encoded on HsChr21 that become expressed.
The discrepancy between the level of transcription from the LTR array and the
output of smallRNAs in human and Tc1 mouse suggests that there is a mechanism of
self/non-self recognition that recognises these transcripts as foreign in the mouse and
fuels them into the piRNA biogenesis pathway. However, very little is known about
the function and expression of these LTR arrays in the human genome. Therefore
a detailed dissection of the transcription of these LTR arrays in human testes will
be necessary to determine their overall contribution to and regulation by pachytene
piRNAs.
Overall, we present evidence that meiotic silencing results in reduced expression
of piRNA-producing loci and detect novel piRNA sequences in the Tc1 mouse that
map to a human-specific LTR array, suggesting that transcripts arising from this
repeat element are readily processed into piRNAs in the Tc1 mouse.

5 | Single-cell profiling of meiotic
silencing throughout
spermatogenesis
Single-cell RNA-Seq libraries were generated with support from Dr. Celia Mar-
tinez and the analysis was performed in collaboration with Nils Eling.
5.1 Introduction
Spermatogenesis is a complex process that occurs in continuous waves through-
out the adult lifespan of an animal and extends over several weeks, resulting in the
generation of haploid gametes (Russell et al., 1993). This involves a number of highly
specialised processes, starting with meiosis, the reductive cell division programme
halving the number of chromosomes per cell, followed by spermiogenesis, a multi-
day differentiation programme that generates mature sperm cells (Oakberg, 1956,
1957). Due to this complex and extended nature, spermatogenesis includes several
drastic changes in gene expression to facilitate these cell transitions (Soumillon
et al., 2013).
Meiotic prophase employs a unique gene silencing mechanism, termed meiotic
silencing of unpaired chromosomes (MSUC), that acts in response to asynapsed
chromosomes and leads to the transcriptional silencing of chromosomes that fail
to undergo synapsis by the late zygotene stage (Turner, 2015). Meiotic silencing
is thought to be part of a prophase I surveillance mechanism that monitors proper
chromosome pairing and can lead to pachytene apoptosis in cells with extensive
asynapsis (Burgoyne et al., 2009; Mahadevaiah et al., 2008). A specialised man-
ifestation of meiotic silencing occurs in male germ cells in the form of meiotic sex
chromosome inactivation (MSCI) (Turner et al., 2005).
Meiotic silencing was shown to display sexually dimorphic features, with weaker
silencing in females compared to males. This was observed when comparing the
level of silencing of chromosome X that naturally occurs in spermatocytes with
silencing of the X chromosome in XO females (Cloutier et al., 2016).
118 Results
The mechanisms that underlie meiotic silencing targeted at unsynapsed auto-
somes and the largely unpaired sex chromosomes in male germ cells are believed
to be the same (Homolka et al., 2007; Mahadevaiah et al., 2008). However, so
far comparative studies relied on low-throughput technologies such as RNA FISH,
which allow the assessment of only a small number of genes and cannot infer
chromosome-wide expression levels.
Here we use single-cell RNA-sequencing of different spermatogenic cell pop-
ulations that allows us to capture the temporal trajectory of spermatogenesis by
profiling cells throughout their differentiation process. By making use of the trans-
chromosomic mouse model Tc1, that carries a single copy of HsChr21, we were
further able to dissect the dynamics of meiotic silencing allowing the comparison
between silencing of the sex chromosomes and an unpaired autosome within the
same cell.
The use of single-cell sequencing enabled a more extensive profiling of the overall
expression state of chromosomes affected by meiotic silencing, thereby avoiding
any bias in the selection of genes to be assayed. This approach revealed drastic
differences in the level of transcriptional silencing across sex chromosomes and
autosomes that result in mosaicism in gene expression between spermatocytes that
carry an unsynapsed autosome.
5.1.1 Aim
The aim of this project is the characterisation of gene expression profiles through-
out mouse spermatogenesis and dissect the dynamics of meiotic silencing
in response to unpaired chromosomes.
Results 119
5.2 Results
5.2.1 Single-cell gene expression atlas of mouse
spermatogenesis
In order to profile the gene expression changes throughout the complex process
of spermatogenesis in the mouse, we performed single-cell RNA-sequencing on
different spermatogenic cell populations. Spermatogonia, primary spermatocytes
and round spermatids were isolated via FACS based on their DNA content and
captured on a C1 Fluidigm platform, which allows the visual confirmation of single
cell capture prior to cDNA synthesis. After filtering and quality control, a total of 458
cells remained, allowing the detection of a total of 17,299 mouse genes as well as
354 genes encoded on HsChr21. We then performed principal component analysis
(PCA) on all cells that passed the quality control, which clearly distinguished the
different cell types (Figure 5.1).
Spermatogonial stem cells
Our sorting strategy using Hoechst staining allows the distinction between cell
types based on their different DNA content as well as the side population (SP)
phenomenon. Side populations, when using viable fluorescent dyes, are created
due to the expression of ATP-binding cassette (ABC) transporters in stem cells that
result in an active efflux of the dye and therefore a lower emission from these cells
(Bunting, 2002). In spermatogonia, this process is mediated by the ABC transporter
breakpoint cluster region pseudogene 1 (Bcrp1) (Lassalle et al., 2004). However,
since we did not use any additional surface marker alongside the Hoechst stain to
unambiguously identify these cells, we do not expect to obtain a pure population.
We therefore performed PCA on spermatogonia alone and indeed observed
two distinct groups within this cell population (Figure 5.2A). Differential expression
analysis between the two groups identified marker genes that included solute carrier
family 25, member 31 (Slc25a31) (also known as adenine nucleotide translocator
4 (Ant4)), nuclear RNA export factor 2 (Nxf2), Maelstrom (Mael), Testis-expressed
11 (Tex11) and FK506 binding protein (Fkbp6) as the top five genes for Group1.
The top five marker genes for Group2 included S100 calcium binding protein A11
(S100a11), the long-noncoding RNA metastasis associated lung adenocarcinoma
transcript 1 (Malat1), annexin A5 (Anxa5) as well as two pseudogenes GM12854
and GM37376 (Figure 5.2B).
120 Results
Figure 5.1 | FACS sorting strategy and single-cell capture on C1 Fluidigm
platform
A| FACS sorting strategy based on Lassalle et al. (2004) that allows the distinction
between different spermatogenic cell populations based on Hoechst staining.
B| Representative image of primary spermatocyte captured on a large IFC with a
capture size between 17-25µm.
C| Representative image of spermatogonium captured on a medium IFC with a
capture size between 10-17µm.
D| Principal component analysis (PCA) on all remaining 439 cells after initial quality
control and filtering.
E| Total cell number and number for individual cell types that were used for down-
stream analysis.
Comparison with published marker genes for spermatogonia showed significant
overlap with markers identified for Group1 and the genes reported in Wang and Pan
(2007). Indeed plotting the expression values for all genes reported in Wang and
Pan (2007) showed a clear distinction between the two groups identified by PCA
(Figure 5.2C), allowing us to confidently label Group1 as spermatogonia.
Results 121
Figure 5.2 | Expression profile of cells within spermatogonial stem cell
population
A| PCA on cell population sorted as spermatogonial stem cells showed two distinct
cell populations. Marker genes were identified for the two populations and the two
top hits per group are displayed as log10 of normalised expression values per cell.
B| Top 15 marker genes per group identified using R/Bioconductor package scran
(Lun et al., 2016).
C| Heatmap displaying the normalised expression values (log10) for spermatogonia
markers described in Wang and Pan (2007) based on the two cell populations
identified by PCA.
Marker genes identified for Group 2 did not allow a clear identification of a spe-
cific cell type, however, included markers for both Sertoli and Leydig cells, the two
main somatic cell types in testes. For example, while some of the genes that are
upregulated in Group 2, showed a homogenous expression across all cells within
this group, other genes, such as Malat1 were only upregulated in a subset of cells,
suggesting that there are several cell populations within Group 2. However, due to
the low cell number, a further dissection of cells within this group did not allow a
stable marker identification.
122 Results
We therefore restricted any further analysis to cells within Group 1 that showed a
clear expression profile as expected for spermatogonia.
Interestingly, when looking at the expression of Stimulated by retinoic acid gene
8 (Stra8) across spermatogonia, we found a heterogenous expression profile (Figure
5.2C), which is in accord with findings that Stra8 is periodically upregulated in
response to retinoic acid (RA) to initiate spermatogonial differentiation as well as
meiotic initiation (Endo et al., 2015). We therefore wanted to test whether Stra8
expression could be used as a marker to further distinguish between undifferentiated
and differentiating spermatogonia. For this purpose, we calculated the diffusion
pseudotime for spermatogonia that allows the temporal ordering of differentiating
cells based on their intrinsic expression profile (Haghverdi et al., 2016). We were
able to obtain a developmental trajectory for these cells, which when overlapped
with the expression of Stra8 showed a mild enrichment of this gene along the first
diffusion component (DC) (Figure 5.3A). To test whether this indeed reflects the
differentiation programme of spermatogonia leading up to meiotic initiation, we
tested the expression of meiosis specific with coiled-coil domain (Meioc), which
was shown to be amongst early meiotic genes that are induced in a partially Stra8-
independent manner (Soh et al., 2017). Indeed, we found that Meioc expression
also increased along the differentiation trajectory, suggesting that we indeed capture
the differentiation process of spermatogonia committing to meiosis.
To identify further genes involved in this transition, we visualised the Top 15
genes underlying the separation of this cell population and indeed found very het-
erogeneous expression patterns (Figure 5.3B). In accord with our interpretation that
the diffusion pseudotime (DPT) leads up to meiotic initiation, we found Prss50 (also
known as testis specific protease 50 (TSP50)), to be strongly induced with DPT. This
gene was previously shown to be specifically induced in type B spermatogonia as
well as leptotene spermatocytes by the transiently expressed CTCF-like (CTCFL)
(Sleutels et al., 2012). Among the genes that are enriched in earlier cells we found
cell division cycle associated 7 like (Cdca7l) and tubulin, beta 5 class I (Tubb5) which
were both previously described to be enriched in spermatogonia (Kent Hamra et al.,
2004; Wu et al., 2009).
While the role of most of these genes during spermatogonial differentiation is
elusive, this dataset presents a valuable resource to identify genes with potential
spermatogenesis-specific function.
Results 123
Figure 5.3 | Diffusion pseudotime analysis on spermatogonia
A| Diffusion pseudotime of spermatogonia overlaid with the log10 normalised ex-
pression value for Stra8 and Meioc across the differentiation trajectory.
B| Heatmap displaying the log10 normalised counts of the Top 15 genes negatively
or positively correlated with pseudotime.
Primary spermatocytes
Primary spermatocytes were isolated based on their 4N DNA content and can
therefore comprise all meiotic prophase stages, as well as cells in the first metaphase.
Given the timing of meiotic prophase and the percentages of cells found within
tubules, we expect the majority of cells to be pachytene spermatocytes (Mays-
Hoopes et al., 1995). However, the pachytene stage itself is a complex process
spanning several days and can in turn be sub-divided into 5 stages, warranting
heterogeneity within this cell population (Greenbaum et al., 1986). Potential cell
contaminants of this population can include mitotically active spermatogonia and
somatic cells, which have a transient DNA content of 4N after DNA replication prior
to mitotic cell division.
To dissect the spermatocyte population further, we performed PCA on these cells
to identify any potential subpopulations. This revealed that, while the separation was
124 Results
not as distinct as for spermatogonia, spermatocytes also displayed a heterogeneous
expression profile (Figure 5.4A).
As these differences most likely reflect a differentiation process, we calculated the
diffusion pseudotime for spermatocytes and compared the differentiation trajectory
with the groups identified via PCA. We found that for both PCA and DPT, the
expression of the germ cell marker deleted in azoospermia-like (Dazl) was the major
driver in separating the attributes. Dazl is highly expressed in spermatogonia and
shows a dynamic pattern in spermatocytes, steadily decreasing in expression as
spermatocytes differentiate (Kim et al., 2012; Ruggiu et al., 1997). This suggests that
both PCA and DPT capture the progression of spermatocytes throughout meiotic
prophase.
To test whether we can further sub-divide spermatocytes into different stages of
meiotic prophase, we explored the expression of marker genes that are known to be
induced at a specific stage. Some of the best studied genes during spermatogenesis
are genes involved in the piRNA pathway that can serve as markers for meiotic
substages. For example, the expression of the second wave of piRNAs in mammals
starts at the pachytene stage and is orchestrated by the transcription factor A-Myb,
which first appears in the earliest pachytene spermatocytes (Bolcun-Filas et al.,
2011). We therefore tested the expression of Piwil1 (also known as MIWI in mice),
which is a target of A-Myb and therefore only found in pachytene spermatocytes or in
later post-meiotic cell types. This showed that MIWI is expressed in all cells captured
in our spermatocyte population, suggesting that we only captured spermatocytes
that already entered the pachytene stage (Figure 5.4B). Based on this we would
expect the differentiation trajectory captured in DPT to reflect the progression of
spermatocytes throughout the 7-day long pachytene stage leading up to the meiotic
cell division. To confirm this, we tested the expression of two genes that were
previously reported to show differential expression throughout the pachytene stage.
The Cyclin dependent kinase 4 (Cdk4) is essential for spermatogenesis (Cohen et al.,
2006) and localises to the synaptonemal complex, disappearing in mid-pachytene
which is in accord with a decrease in expression along our trajectory (Ashley et al.,
2001; Zindy et al., 2001). In contrast to that, the expression of Sox17, which initiates
during pachytene (Kanai et al., 1996), showed a consistent increase along our
differentiation trajectory (Figure 5.4B).
Results 125
Figure 5.4 | Transcriptional profile of differentiating spermatocytes
A| PCA and DPT of spermatocyte cell population with log10 normalised counts for
germ-cell marker Dazl.
B| Gene expression across spermatocyte pseudotime for marker genes expressed
during pachytene (MIWI), early pachytene (Cdk4) and late pachytene (Sox17).
C| Heatmap displaying the log10 normalised counts of the Top 15 genes that underlie
the separation of the first principal component, sorted based on pseudotime.
Our dataset therefore captures gene expression changes during the pachytene
stage and could identify key genes involved in the prophase I to metaphase I
transition, such as Aurora kinase B (Aurkb) (Kimmins et al., 2007), which is among
the Top 15 genes that increase in expression in late spermatocytes (Figure 5.4C).
126 Results
Spermatids
The isolation of spermatids was solely based on their low DNA content as these
cells are the only haploid cells within testes and can be easily separated from other
cells using the Hoechst stain. We therefore expect no major cell contaminations in
this population, however, due to the sorting our population might be biased towards
earlier spermatids which have a rounder structure and are therefore less likely to
disturb the integrity of the droplet during the sort.
PCA on the spermatid population yielded a similar result as for spermatocytes,
with no separate groups identified but a spread of the cells most likely representing
a differentiation process. We therefore calculated the pseudotime for these cells
and overlaid both the PCA and pseudotime trajectory with the expression for sperm
protamine 1 (Prm1). This gene initiates expression in round spermatids and steadily
increases throughout spermiogenesis, which is captured in both the PCA as well as
DPT thereby confirming the directionality of the differentiation process (Figure 5.5A).
Amongst the genes that are enriched in early spermatids we found Yip Domain
Family member 5 (Yipf5) (Figure 5.5B), which is involved in transport between the
endoplasmic reticulum and the Golgi apparatus (Yoshida et al., 2008), which could
suggest an involvement in the formation of the acrosome that commences shortly
after the meiotic cell divisions (Huang and Ho, 2006). We also observed an increase
of several mitochondrial genes that function in the respiratory chain (Figure 5.5B),
which could reflect changes in mitochondria associated with the formation of the
axoneme (Borg et al., 2010).
Genes that were positively correlated with pseudotime showed a clear association
with spermiogenesis and were related to processes occurring in elongating sper-
matids (Figure 5.5B). These include H1 Histone Family Member N, Testis Specific
(H1fnt), which is an essential mediator of the histone-protamine transition (Martianov
et al., 2005) as well as calicin (Ccin), a sperm-specific basic protein that is part
of the cytoskeleton in sperm heads (von Bülow et al., 1995). A sperm-specific
head structure is only initiated during mid-spermiogenesis at step 8 of spermatid
development, when both nucleus and acrosome start to polarise (O’Donnell, 2015),
suggesting that we capture the differentiation process until at least this stage.
Compaction of chromatin eventually prevents any new transcription and therefore
requires the storage of mRNAs for translation at a later point in elongating spermatids
(Braun, 2000).
Results 127
Figure 5.5 | Transcriptional profile of haploid spermatids
A| PCA and DPT of spermatid cell population with log10 normalised counts for Prm1.
B| Heatmap displaying the log10 normalised counts of the Top 15 genes that underlie
the separation of the first principal component, sorted based on pseudotime.
Sterile alpha motif domain containing 4 (Samd4), also known as mammalian
Smaug, is an RNA-binding protein which in mammals has been shown to act as
a translational repressor of Smaug recognition element (SRE)-containing mRNAs
in neurons (Baez and Boccaccio, 2005). Thus far, translational repression during
spermiogenesis has mostly been linked to Y-box proteins (Paronetto and Sette,
2010), however, the increased expression of Samd4 in elongating spermatids could
point towards a role for this gene in translational repression during spermiogenesis.
Taken together, the differentiation trajectory obtained for the spermatid population
resembles at least early-to-mid-spermiogenesis, corresponding to the phase with
the most pronounced gene expression changes.
128 Results
5.2.2 Dynamic gene expression profile throughout
spermatogenesis
Dissecting the different cell types individually allowed us to generate an overall
differentiation trajectory throughout spermatogenesis by ordering all single cells
according to their developmental timing. Based on this we are able to visualise gene
expression changes throughout spermatogenesis at an unprecedented temporal
resolution.
Visualisation of well-described marker genes with known functions during meiosis
or spermiogenesis emphasises the tight transcriptional regulation between cell types
as well as during the differentiation process (Figure 5.6). This showed that many
genes in spermatogonia or spermatocytes had a clear ’on’ or ’off’ state, whereas
gene expression patterns in spermatids appeared more sporadic. While we detected
some genes, especially ones involved in chromatin compaction, that showed a clear
expression pattern along the differentiation trajectory in spermatids, many of the
genes also expressed in spermatogonia and spermatocytes showed a heteroge-
neous expression in spermatids. This could reflect the genome-wide upregulation
and spurious transcription that occurs in spermatids, preceding chromatin com-
paction and transcriptional shut down in these cells. In accord with this, we detected
the highest number of expressed genes in the spermatid cell population (13,314
genes compared to 11602 and 12046 in spermatocytes and spermatogonia, respec-
tively). This suggests that our data could be helpful in distinguishing functional genes
in spermatids based on their gene expression pattern throughout differentiation,
compared to sporadically expressed genes as a by-product of chromatin remod-
elling, which would be masked in bulk sequencing experiments even in purified cell
populations.
To identify further genes that show distinct expression changes throughout sper-
matogenesis, we performed differential expression analysis between the different cell
populations (Figure 5.7). This comparison yielded a number of genes, such as Taf7l
and Nxf2 specifically expressed in spermatogonia when compared to spermatocytes
that have been previously described (Wang and Pan, 2007). Similarly, the genes
upregulated in spermatids compared to spermatocytes included genes with a clear
function during spermiogenesis, such as sperm acrosome associated 1 (Spaca1)
and Testis-specific serine/threonine-protein kinase 4 (Tssk4) (Fujihara et al., 2012;
Wang et al., 2015).
Results 129
Figure 5.6 | Marker gene expression throughout spermatogenesis
Heatmap displaying log 10 normalised counts for known marker genes that display
differential expression throughout spermatogenesis. Cells in heatmap were ordered
based on diffusion pseudotime calculated within each individual cell population and
are coloured based on cell type.
Genes that were specifically upregulated in spermatocytes compared to sper-
matids included chromatin modifying enzymes such as the histone deacetylase
complex subunit Sin3A-associated protein, 30kDa (Sap30) as well as the histone
methyltransferase histone lysine Methyltransferase DOT1L Cofactor (Mllt10), which
were both shown to be expressed in metaphase II (MII) oocytes (Wang et al.,
2010). We also detected MutL homolog3 (Mlh3), a member of the meiotic DNA
mismatch repair system that is involved in resolving double Holliday junctions to
result in crossovers (Lipkin et al., 2002). Genes that were upregulated in sperma-
tocytes compared to spermatogonia mostly showed a high expression throughout
spermiogenesis as well and included members of the metalloproteinases family
ADAM (Adam2 and Adam32) which play a role in regulating apoptosis throughout
spermatogenesis (Moreno et al., 2011).
130 Results
Figure 5.7 | Differential gene expression throughout spermatogenesis
Heatmap displaying log 10 normalised counts for the Top 15 genes identified in pair-
wise comparisons for differentially expressed genes between spermatogonia and
spermatocytes as well as spermatocytes and spermatids. Cells are ordered based
on diffusion pseudotime calculated within each cell population and are coloured
based on cell type.
Taken together, our data present the most comprehensive dissection of gene
expression during mouse spermatogenesis to date, and allows the identification of
more subtle gene expression changes that are masked in bulk sequencing experi-
ments.
Results 131
5.2.3 Meiotic silencing dynamics of the X chromosome
We next wanted to examine the levels of meiotic silencing throughout spermato-
genesis. To profile the physiological dynamics of this process, we first looked at the
expression level of mouse chromosome X, which due to its largely unpaired nature
becomes transcriptionally silenced at the zygotene-to-pachytene transition together
with the Y chromosome.
Looking at the overall expression of the X chromosome across the three different
cell types we observed a steep drop both in read numbers as well as genes that are
detected between spermatogonia and spermatocytes (Figure 5.8A and B). This is in
accord with previous studies showing that gene expression from the X chromosome
is entirely shut-down at the transition between late zygotene to early pachytene
spermatocytes. Overall, we find that out of 768 X-linked genes that are detected
in our dataset, 427 are significantly downregulated between spermatogonia and
spermatocytes (Figure 5.8C). In contrast to that we only detect 19 genes that show a
relative increase in expression in spermatocytes compared to spermatogonia. These
genes are almost exclusively pseudogenes as well as one protein-coding gene testis-
specific gene A8 (Tsga8), which is involved in chromatin condensation in spermatids
(Figure 5.9B) (Good et al., 2011). To test whether the detection of these X-linked
pseudogenes might be caused by cross-mapping of reads from their autosomal
paralogs, we tested the expression of the functional paralogs that we were able
to identify. Indeed we found that almost all paralogs of the X-linked pseudogenes
detected in spermatocytes showed a high expression throughout spermatogenesis
(Figure 5.10). We can therefore not exclude that the signal detected for X-linked
pseudogenes in spermatocytes originates from their functional paralogs, which
will require a more thorough investigation of the sequence conservation between
the pseudogenes and their functional paralogs to exclude potential cross-mapping.
Furthermore, due to the absence of leptotene and zygotene spermatocytes in our
data, we cannot exclude that the X-linked pseudogenes are expressed during early
prophase and their transcripts remain throughout pachytene.
Due to the evolutionary forces that act on the X chromosome, it is relatively
enriched for male-specific genes that function in early spermatogenesis and are
expressed in spermatogonia, but depleted of genes acting during male meiosis as a
result of MSCI (Khil et al., 2004; Wang et al., 2001). However, while meiosis-specific
genes are not affected, MSCI presents a challenge for ubiquitously expressed
housekeeping genes that are encoded on the X chromosome.
132 Results
Figure 5.8 | Chromosome X expression throughout spermatogenesis
A| Relative expression level of chromosome X (all reads mapping to X divided by all
reads mapping to the mouse genome) across cells from different populations sorted
based on their X expression level.
B| Number of X-linked genes expressed (at least one read) in cells ordered based
on their X expression level.
C| Differential expression analysis for genes encoded on X chromosome comparing
between spermatocytes and spermatogonia. Genes downregulated in spermato-
cytes are highlighted in red and genes upregulated in spermatocytes are highlighted
in purple (with a log2 fold change greater than 0.5).
D| Differential expression analysis for genes encoded on X chromosome comparing
between spermatids and spermatocytes. Genes downregulated in spermatids are
highlighted in purple and genes upregulated in spermatids are highlighted in blue
(with a log2 fold change greater than 0.5).
As a result, many essential X-linked housekeeping genes have retroposed copies
on autosomes, which can rescue the expression of these genes in spermatocytes,
when the parental copies are inactivated by MSCI (Wang, 2004). We therefore tested
the expression of X-linked parental genes and their autosomal retrogene copies in
our data set and indeed observed that the majority of autosomal retrogenes were
specifically upregulated upon MSCI onset in spermatocytes (Figure 5.11).
Results 133
Figure 5.9 | Differential gene expression of chromosome X throughout
spermatogenesis
Heatmaps displaying log10 normalised counts for Top50 differentially expressed
genes between spermatogonia and spermatocytes (A) as well as spermatocytes and
spermatids (B). Cells are ordered based on X chromosome expression decreasing
in spermatogonia and spermatocytes and increasing in spermatids.
134 Results
Figure 5.10 | Expression of X-linked pseudogenes and functional paralogs
Heatmaps display the log10 normalised counts for pairs of X-linked pseudogenes
that appeared upregulated in spermatocytes compared to spermatogonia with their
functional orthologs encoded on autosomes. Cells are ordered based on develop-
mental pseudotime and coloured based on cell type.
The transcriptional silencing that is established across the sex chromosomes
during early pachytene was shown to persist throughout post-meiotic development
for the majority of genes (Greaves et al., 2006; Turner et al., 2006). However, a
number of X- and Y-encoded genes regain expression in spermatids and have
been termed ’escape genes’. This is in part mediated by RING finger protein 8
(RNF8), which is an E3 ubiquitin-protein ligase. RNF8 catalyses the ubiquitination of
histone H2A on the sex chromosomes during meiosis and subsequently mediates
the establishment of active histone marks, priming genes for re-expression (Lu et al.,
2010; Sin et al., 2012a).
In addition to this process, RNF8-independent activation of escape genes mostly
involves genes that exist in multiple copies on the sex chromosomes (Sin and
Namekawa, 2013). Throughout evolution, the mammalian sex chromosomes were
shaped by intrachromosomal rearrangements, generating large ampliconic structures
that cover up to 31% of the mouse X chromosome (Warburton et al., 2004).
Results 135
Figure 5.11 | Expression of X-linked housekeeping genes and their retroposed
autosomal copies
Heatmaps display the log10 normalised counts for pairs of X-linked housekeeping
genes and their retroposed copies on autosomes. Cells are ordered based on
developmental pseudotime and coloured based on cell type.
These ampliconic regions are enriched for multi-copy genes with up to 30 copies
per gene that are arranged in palindromic head-to-head or tail-to-tail structures
(Mueller et al., 2008, 2013). The current hypothesis is that the increase in copy
number enables a higher RNA output; however, how transcriptional activation is
achieved in the first place is currently unclear (Turner, 2015). Previous studies have
identified at least 33 multi-copy gene families that add up to ∼273 gene copies,
thereby accounting for ∼18% of genes encoded on the mouse X chromosome under
the previous genome build (NCBI 37.1) (Mueller et al., 2008).
136 Results
Multi-copy genes were originally found to have a higher expression level com-
pared to single-copy genes expressed in spermatids (Mueller et al., 2008), however
more recent studies identified a number of highly expressed single-copy genes post
meiosis both in mice and humans (Sin et al., 2012b). Thus, the functional divide
between X-linked multi-copy and single-copy genes remains mostly unclear.
Investigating the amount of re-expression from the X chromosome in our dataset,
revealed a moderate increase in the relative read counts mapping to the X chro-
mosome in the spermatid population (Figure 5.8A). However, when looking at the
number of genes expressed, we found a surprisingly large number of X-linked genes
reactivated in spermatids (Figure 5.8B). Differential gene expression for genes en-
coded on the X chromosome between spermatocytes and spermatids showed that a
total of 393 genes were upregulated post meiosis, with a total number of 645 genes
being detected across the spermatid population (Figure 5.8D).
Previous studies reported partially contradicting findings regarding the expres-
sion pattern of escape genes throughout spermatogenesis. While Mueller et al.
(2008) reported that the majority of multi-copy genes expressed in spermatids are
also expressed in spermatogonia, therefore proposing a silencing by MSCI and
post-meiotic derepression, Sin et al. (2012b) reported most escape genes to be
spermatid-specific. Comparing the gene expression between spermatids and sper-
matogonia showed that 284 genes were specifically upregulated in spermatids of
which 136 genes were exclusively detected in spermatids. In contrast to that, 340
genes showed a higher expression in spermatogonia compared to spermatids, with
120 genes being specific to spermatogonia (Figure 5.12A). Gene ontology analysis
on differentially expressed genes showed an enrichment of genes functioning in sper-
matid development in both spermatids and spermatogonia (Figure 5.12B). Among
the genes identified in spermatids were Sycp3 like X-linked (Slx1) and Slx-like 1
(Slxl1), which are both essential for spermiogenesis (Cocquet et al., 2010), whereas
the majority of genes identified for this GO term in spermatogonia belonged to the
X-linked lymphocyte-regulated (Xlr) multi-copy gene family.
Overall, we seem to detect a considerably larger number of genes that escape
post-meiotic silencing in spermatids compared to previous studies. However, a
thorough analysis of multi-copy versus single-copy genes and their expression
pattern throughout spermatogenesis will be necessary to better understand the
mechanistic basis of escape genes.
Results 137
Figure 5.12 | Differential gene expression of X-linked genes between sper-
matogonia and spermatids
A| Differential expression analysis for genes encoded on X chromosome comparing
between spermatogonia and spermatids. Genes that are upregulated in spermato-
gonia compared to spermatids are highlighted in red and genes that are upregulated
in spermatids compared to spermatogonia are highlighted in blue (with a log2 fold
change greater than 0.5). Venn diagram illustrates the number of genes detected
per cell population and the overlap between spermatids and spermatogonia.
B| Gene ontology analysis was performed on genes upregulated in spermatids or
spermatogonia. Enrichment scores (-log10 of p-value) for GO terms are depicted
and the number of genes identified per category is annotated inside the individual
bar charts.
The paradoxical enrichment of genes on the X chromosome that are essential
for spermiogenesis not only requires the escape from meiotic silencing of these
genes post meiosis, but also creates the need for these gene products to be shared
with Y-carrying spermatids. This is achieved via cytoplasmic bridges that connect
spermatids via a syncytium, thereby allowing the exchange of RNA and protein
between cells, making them phenotypically diploid (Braun et al., 1989). However,
little is known about the level of exchange between spermatids and whether the
exchange is driven by passive diffusion or active transport of selected mRNAs or
proteins.
138 Results
The profiling of spermatids on a single-cell level allows us to assess the amount
of transcript sharing by investigating the expression pattern of X-linked transcripts
in individual cells. When looking at the expression of X-linked genes across the
spermatid population, we find that out of the 475 genes that are detected in at least
10% of spermatids, 276 X-linked genes are detected in more than 50% of cells, with
70 genes being detected in at least 90% of all spermatids (Figure 5.13A).
We then tested whether there is a correlation between the expression level of X-
linked genes and the percentage of spermatids they are detected in. When averaging
gene expression for individual genes by the number of cells in which we detect the
transcript, we observed a mild trend of increased expression for transcripts that are
present in a higher number of cells (Figure 5.13A). Interestingly, gene ontology (GO)
analysis on X-linked genes present in more than 50% of spermatids revealed an
enrichment for spermatid development as well as protein ubiquitination and organisa-
tion of actin filaments (Figure 5.13B). In contrast, GO terms enriched among genes
that are detected in only 10-50% of spermatids showed no spermatogenesis-related
terms, but were related to transcriptional regulation and chromatin modifications.
The enrichment of spermiogenesis-related genes among the transcripts that
are found in the majority of cells could suggest a positive selection of these genes,
either via active transport or increased expression of these genes, thereby driving
passive diffusion. However, due to the relatively low number of spermatids in
our population spanning an extended differentiation process, any difference in
the number of cells expressing a certain gene in our current analysis could also
reflect a temporal expression pattern. We therefore require a higher number of
cells that more accurately recapitulate the complex differentiation process that
underlies spermiogenesis, to dissect the mechanisms of transcript sharing between
spermatids.
5.2.4 Meiotic silencing in response to autosomal aneuploidy
Meiotic silencing of asynapsed chromosomes is thought to be mediated by
the same machinery, irrespective of whether sex chromosomes or autosomes are
targeted. To test whether the transcriptional behaviour of an unsynapsed autosome
is similar to what we observed for chromosome X, we profiled the expression of
HsChr21 in the Tc1 mouse throughout spermatogenesis.
The meiotic silencing response was previously shown to be dose-dependent,
leading to MSCI failure in the presence of extensive asynapsis affecting 2-3 chromo-
some pairs (Kouznetsova et al., 2009; Mahadevaiah et al., 2008).
Results 139
Figure 5.13 | X-linked gene expression in haploid spermatids
A| Violin plots displaying the mean normalised counts (average expression across
cells expressing a given gene) of genes that are detected in a defined percentage of
spermatids. Total number of genes within each interval are displayed inside violin
plot.
B| Enrichment score (-log10 of p-value) of gene ontology terms for genes that are
found in 10-50% of spermatids (light blue, 199 genes) and genes that are found
in 50-100% (dark blue, 276 genes) of spermatids. Number of genes identified per
category is annotated inside individual bar charts.
140 Results
In accord with this, we did not observe any effect on X chromosome silencing
in Tc1 animals, suggesting that the amount of additional asynapsis caused by
HsChr21 does not perturb MSCI. Looking at the relative expression of HsChr21
across our cell populations we observed a steady decrease in reads mapping to
HsChr21 within the spermatocyte population in contrast to the steep drop that
we observed for chromosome X (Figure 5.14A). This is also recapitulated at a
gene level, with some spermatocytes even showing an increase in the number of
HsChr21-encoded genes (Figure 5.14B). Overall, we observe a consistent number
of 40-60 genes expressed in spermatogonia out of the 354 human genes that can
be identified without cross-mapping issues. Differential gene expression between
spermatogonia and spermatocytes identified 20 genes specifically upregulated
in spermatocytes compared to 63 genes that are downregulated (Figure 5.14C).
Looking at the expression of genes in individual cells we found that many genes with
a lower expression in spermatocytes nevertheless remained expressed at a higher
level than we had observed for X-linked genes. In contrast to that, the majority of
upregulated genes were specific to spermatocytes (Figure 5.15A and B).
These findings are surprising in context of previous studies in the Tc1 mouse
showing that HsChr21 is unsynapsed in ∼50% of cells resulting in transcriptional
silencing as was shown by the absence of human Cot1 RNA (Mahadevaiah et al.,
2008). Similarly, gene-specific RNA FISH for nascent transcripts of three HsChr21-
encoded genes, USP25, NRIP1 and TPTE, in spermatocytes confirmed robust
transcriptional silencing of HsChr21 in almost all spermatocytes that were analysed
(Cloutier et al., 2016). It is possible that this discrepancy stems from the detection of
nascent RNA transcripts in previously published studies compared to the detection
of mature mRNA in our approach. However, testing the RNA expression level of the
three genes analysed by nascent RNA FISH in Cloutier et al. (2016), we observed
very low or undetectable levels of mature mRNA for these genes in spermatogonia
and spermatocytes (Figure 5.16A).
As shown by Mahadevaiah et al. (2008), HsChr21 can self-synapse and thereby
escape transcriptional silencing. It is therefore possible that genes located in specific
areas might be involved in self-synapsis of the chromosome and show a higher
expression level. We therefore tested where the genes that are upregulated in
spermatocytes compared to spermatogonia are located on HsChr21. This revealed
an accumulation of 15 genes across a ∼9kb region at the end of the long arm of
HsChr21, which is however more centrally positioned on HsChr21 in the Tc1 mouse
due to rearrangements of the chromosome (Figure 5.14E) (Gribble et al., 2013).
Results 141
Figure 5.14 | Human chromosome 21 expression throughout mouse
spermatogenesis in Tc1 animals
A| Relative expression level of HsChr21 (all reads mapping to HsChr21 divided by all
reads mapping to the mouse genome) across cells from different populations sorted
based on HsChr21 expression level.
B| Number of HsChr21-encoded genes expressed (at least one read) in cells ordered
based on HsChr21 expression level.
C| Differential expression analysis for genes encoded on HsChr21 comparing be-
tween spermatocytes and spermatogonia. Genes downregulated in spermatocytes
are highlighted in red and genes upregulated in spermatocytes are highlighted in
purple (with a log2 fold change greater than 0.5).
D| Differential expression analysis for genes encoded on HsChr21 comparing be-
tween spermatids and spermatocytes. Genes downregulated in spermatids are
highlighted in purple and genes upregulated in spermatids are highlighted in blue
(with a log2 fold change greater than 0.5)
E-F| Location of genes differentially expressed between spermatocytes and sper-
matogonia (E) and between spermatids and spermatocytes (F) across HsChr21.
142 Results
Figure 5.15 | Differential gene expression of human chromosome 21
throughout mouse spermatogenesis
Heatmaps display the log10 normalised counts for differentially expressed genes
between spermatogonia and spermatocytes (A and B) or spermatocytes and sper-
matids (C). Only Top25 genes are displayed in A.
Results 143
Figure 5.16 | Expression level of HsChr21-encoded genes in spermatogonia
and spermatocytes
A| Boxplots display the log10 normalised expression of HsChr21-encoded genes
in spermatogonia and spermatocytes. Top3 genes upregulated in spermatogonia
and spermatocytes as well as genes tested via nascent RNA FISH in Cloutier et al.
(2016) are displayed.
B| Heatmap displaying the log10 normalised counts of HsChr21-encoded genes that
are expressed in spermatogonia, downregulated in spermatocytes and re-expressed
in spermatids.
In contrast to that, the genes expressed in spermatogonia were more evenly
distributed across the length of HsChr21. This could suggest that only parts of
HsChr21 remain expressed during meiosis depending on their position on the
chromosome, however active transcription from this area will need to be verified via
RNA FISH for nascent transcripts.
Focusing on the post-meiotic expression of HsChr21 we observed an almost
equal up- and down-regulation of genes, with 25 genes increasing in expression
whereas 23 genes are downregulated post-meiosis (Figure 5.14D and Figure 5.15C).
When looking at the location of genes that are differentially expressed between
spermatocytes and spermatids we did not observe any hotspots across HsChr21
144 Results
(Figure 5.14F). Among the genes upregulated in spermatids, we only found 8
genes that were also expressed in spermatogonia and subsequently downregulated
in spermatocytes, a pattern that is frequently observed for genes encoded on
the X chromosome (183 genes) (Figure 5.16B). This could reflect the enrichment
of spermatogenesis-essential genes on the X chromosome, whereas any gene
products from HsChr21 are dispensable, especially for mouse spermatogenesis.
Looking at the number of spermatids expressing HsChr21-encoded genes, we
found that the most highly expressed genes are found in the majority of cells even
though a maximum of 50% of spermatids carry HsChr21. This suggest that also
non-essential transcripts are shared between neighbouring spermatids. Our previous
genotyping of round spermatids from Tc1 animals showed that the number of haploid
cells carrying HsChr21 is actually closer to 35%, potentially due to meiotic loss of
HsChr21 (Figure 3.14) and (Table 3.4). In accord with this we observed a small
number of spermatids that have almost no detectable signal for HsChr21, which
could reflect cells in which the human chromosome was lost either in the progenitor
cell already or during meiosis (Figure 5.14B).
Taken together, we observed a surprisingly high expression of HsChr21 in sper-
matocytes which could reflect differences in the robustness of meiotic silencing of
autosomes compared to sex chromosomes within the same cell.
5.3 Discussion
Spermatogenesis lasts over several weeks upon meiotic entry and involves
drastic changes in both gene expression as well as cell morphology and function.
As a consequence, the cells obtained even from purified cell populations undergo
complex differentiation programmes of which the fine-tuning is concealed across the
population average.
Our dataset is, to the best of our knowledge, the first representation of the
transcriptional profile throughout mouse spermatogenesis on a single-cell level.
Enabling the temporal dissection of individual cell types along their differentiation
trajectory, we generated a valuable resource to identify novel regulators of meiosis
and spermiogenesis. The dissection of individual cell types identified numerous
genes that change in their expression profile along the differentiation trajectory
and for example in spermatocytes included genes like Aurkb, which is involved in
the prophase to metaphase transition and increased as spermatocytes progressed
through prophase I.
Results 145
However, in addition to genes with well described functions during spermatogenesis,
we also detect a large number of genes that have not been previously described in
the context of spermatogenesis or meiosis.
To ensure the accurate progression throughout meiosis, several surveillance
mechanisms are in place that monitor chromosome synapsis, DNA damage repair as
well as chromosome segregation. As part of the chromosome synapsis checkpoint,
meiotic silencing functions in both female and male germ cells to transcriptionally
silence unsynapsed chromosomes. In male germ cells, a specialised installation
of this response to unsynapsed chromosomes is termed meiotic sex chromosome
inactivation. MSCI is conserved throughout mammals and also occurs in marsupials,
therefore predating the separation of the eutherian and metatherian lineage (Franco
et al., 2007; Namekawa et al., 2007). MSCI results in complete silencing of the X and
Y chromosome at the onset of pachytene and persists throughout meiosis and partly
also throughout post-meiotic development. As a consequence of X silencing during
meiosis, X-encoded housekeeping genes have retroposed copies on autosomes
which can compensate for the loss of X expression and we observed a consistent
upregulation of these autosomal retrogenes in spermatocytes. In accord with the
notion that all X- and Y-encoded genes are transcriptionally silenced in pachytene
spermatocytes, we observed a sharp decrease in relative X expression between
spermatogonia and spermatocytes. The low level expression that was detected
for some X-linked genes mainly affected pseudogenes, which could be due to
cross-mapping of reads from their autosomal functional paralogs. However, we
also detected one protein-coding gene Tsga8, which was specifically upregulated in
spermatocytes compared to spermatogonia and further increased in expression in
spermatids. Tsga8 was described as the most rapidly evolving gene in the mouse
genome, diverging even between closely related mouse strains (Good et al., 2011).
However, while this would be an interesting gene to investigate, active transcription
of this gene needs to be confirmed by performing RNA FISH for nascent transcripts.
One very interesting aspect that can be addressed in single-cell data is the
dynamics of post-meiotic escape from silencing and the sharing of X- and Y-linked
transcripts between individual haploid spermatids. Compared to previous studies we
detected a larger number of genes escaping post-meiotic silencing which will require
a thorough dissection of the relation between multi-copy and single-copy genes and
their expression levels. Furthermore the correlation between expression and the
extent of ’transcript sharing’ between haploid spermatids will be interesting to assess.
However, one current limitation of our data are the comparably low cell numbers,
146 Results
especially for spermatids, which do not provide accurate coverage of the complex
differentiation process underlying this cell population. It would therefore be useful to
generate a similar data set using a more high-throughput approach such as the 10X
Genomics single-cell RNA-Seq methodology. Droplet-based single-cell sequencing
techniques allow the capture of much larger cell numbers and could therefore
be used to generate an unbiased expression profile of spermatogenesis. When
performed on whole tissue lysates without prior FACS sorting, such an approach
should capture a snapshot of the tissue composition and therefore contain a larger
number of spermatids as these cells make up the majority of the tissue. Using our
current dataset to annotate the captured cell populations, this should allow a more
temporal dissection of spermiogenesis and could shed light onto the pattern and
order of post-meiotic gene activation on the sex chromosomes.
Profiling gene expression throughout spermatogenesis in Tc1 animals revealed
that the presence of HsChr21 did not affect the silencing of the sex chromosome,
which is consistent with the absence of a mid-pachytene arrest in prophase I in these
mice (Mahadevaiah et al., 2008). The discrepancy between previously published
studies and our data regarding the expression level of HsChr21 in spermatocytes
is most likely due to the different detection methods used. One previous study was
based on RNA FISH for human Cot1 RNA, which is a commonly used marker for
nascent RNA transcripts in RNA FISH experiments (Namekawa and Lee, 2011).
However, Cot1 can only be used as a proxy for transcription as it corresponds to
a repetitive sequence that shows pervasive transcription. The more recent study
performing gene-specific RNA FISH for nascent transcripts in comparison with
our data showed that the genes used for visualisation showed either very low or
undetectable expression in our data set and might therefore not be representative
of overall HsChr21 expression. Further confirmation of our sequencing data by
performing RNA FISH for nascent transcripts of HsChr21-encoded genes that are
specifically upregulated in Tc1 spermatocytes will be necessary to confirm active
transcription in these cells.
Taken together, we have generated a comprehensive gene expression atlas for
mouse spermatogenesis that captures the dynamics of meiotic silencing of sex
chromosomes and autosomes.
6 | Discussion and Outlook
The work presented in this thesis provides valuable insight into mammalian
meiosis and the impact of aneuploidy during this developmental programme.
6.1 Repeat activation in the Tc1 mouse
In accord with previous studies in the Tc1 mouse, we find human-specific repet-
itive elements that acquire active chromatin marks in both liver and testes, and
observe persistent activation of these elements after male germline transmission.
This suggests that the observed activation of these elements is inherent to the DNA
sequence and is stable throughout fertilisation and development, irrespective of the
amount of epigenetic information transmitted from the parent.
However, studies in mouse and human have shown that DNA transmitted through
sperm retains a low percentage of histones bound to the DNA, which comprises
about 10-15% in humans and 5-10% in mouse (Oliva and Luís Ballescà, 2012). It
would therefore be interesting to profile regions on HsChr21 that retain histones in the
Tc1 mouse, however, due to the low sperm count in Tc1 males, these experiments
have proven difficult to obtain enough material. Publicly available data sets profiling
histone-bound DNA in human sperm show discrepancies between studies, which
are most likely due to technical differences, however these differences make it hard
to draw conclusions about which regions in the human genome stably retain histones
throughout sperm development (Erkek et al., 2013; Hammoud et al., 2009; Saitou
and Kurimoto, 2014).
The consistent activation of a specific set of evolutionary young, human-specific,
repetitive elements suggests that the mouse might lack a species-specific silencing
mechanism to accurately control these elements encoded on HsChr21. Prime candi-
dates for this are KRAB-zinc finger proteins that mediate transcriptional repression
via interaction with the repressor protein KAP1 that acts as a scaffold domain to
recruit chromatin modifiers (Urrutia, 2003). KZFPs represent the largest family of
transcription factors in both the human and mouse genome and are rapidly evolving
in correlation to the emergence of species-specific TEs (Emerson and Thomas,
2009; Thomas and Schneider, 2011).
148 Discussion and Outlook
Indeed, a recent study using transchromosomic mouse ES cells that carry one
copy of human chromosome 11 (HsChr11), therefore using a similar approach to
the Tc1 mouse, except that the larger size of human chromosomes 11 most likely
prevents the generation of a mouse from these cells, was able to link two human
KZNFs to the repression of recently evolved human-specific repeats (Jacobs et al.,
2014). Similar to our observations, the authors found primate-specific TEs encoded
on HsChr11 gained H3K4me3 in TC11-mESCs resulting in increased transcription
of the majority of SVA and human-specific L1 elements (61% and 93%, respectively).
Luciferase reporter-based assays, using these human-specific repeat elements as
promoter sequences, and co-transfection with human KZNFs into mouse ES cells
allowed the identification of two human-specific KZNFs, ZNF91 and ZNF93, that
mediate repression of SVA and L1PA3 elements, respectively (Jacobs et al., 2014).
In a similar fashion, the Tc1 mouse might lack a human KZNF that is necessary to
mediate the repression of LTR12 elements, which is the predominant repeat family
derepressed in Tc1 somatic tissues.
Further dissection of the repeat classes that show differential activation between
somatic and germline tissues in the Tc1 mouse could potentially shed light on
silencing dynamics. For example, screening of these different repeat classes against
the recently created collection of binding motifs for mouse KRAB-ZFPs and human
KZNFs could point towards potential interactions (Ecco et al., 2016; Imbeault et al.,
2017; Najafabadi et al., 2015). It would be interesting to see whether repetitive
elements that only show transient activation in Tc1 testis, but accurate repression in
adult somatic tissues, are targeted by mouse-specific KRAB-ZFPs or an evolutionary
older protein that might be shared between human and mouse.
Interestingly, ChIP-Seq experiments for NF-Y in the Tc1 mouse revealed that
this factor binds specifically to all LTR12 elements on HsChr21 that gain an active
chromatin state. NF-Y is a ubiquitously expressed conserved transcription factor
that binds to the CCAAT motif found at numerous eukaryotic promoters (Bi et al.,
1997; Sinha et al., 1995). However, NF-Y has also been proposed to function as a
pioneer factor, promoting chromatin accessibility and could therefore be involved
in the acquisition of an active chromatin state of these repeat elements (Oldfield
et al., 2014). Previous genome-wide binding studies in human cell lines revealed an
enrichment of NF-Y at intergenic regions, particularly at repetitive elements of the
LTR family (Fleming et al., 2013). Consistent with these observations, our ChIP-seq
data in the Tc1 mouse also demonstrates the binding of NF-Y to repetitive elements
across the mouse genome, with a particular preference for LTR elements of the
Discussion and Outlook 149
ERVK family. These findings are interesting in light of a recent report implicating NF-
Y in the formation of accessible chromatin regions during zygotic genome activation
at the 2-cell stage, particularly at intergenic regions comprising repetitive elements
(Lu et al., 2016). However, the exact function of NF-Y binding to repetitive elements
both in early embryos and somatic cells remains to be determined.
6.2 Low transmission rate of human chromosome 21
through the male germline
The low transmission rate of HsChr21 through the male germline cannot be fully
explained by loss of HsChr21-carrying cells during spermatogenesis. In fact, the
number of Tc1-positive cells after male meiosis and even in mature sperm cells was
strikingly similar to the transmission rate that is observed via females. This suggests
that further loss downstream must occur to result in the low transmission rate of
around 12% aneuploid offspring via the male germline.
Our experiments have excluded meiotic loss as well as elimination during spermio-
genesis, as we still observe higher numbers of Tc1-carrying sperm. We therefore
expect this discrepancy to be due to an impairment either at fertilisation or during
early embryonic development.
One possibility is that HsChr21-carrying sperm have aberrant chromatin conden-
sation due to the presence of the extra chromosome, resulting in reduced fitness
and/or mobility of the sperm (Acloque et al., 2013). It is rather unlikely that the
presence of HsChr21 by itself, in terms of the amount of DNA, is problematic since
it is smaller in size than the difference between the X and Y chromosome (Soh
et al., 2014). If the extra ’weight’ of the DNA would indeed be the limiting factor, we
would expect a bias towards males inheriting HsChr21 via sperm since the weight
of a sperm carrying a Y chromosome and HsChr21 would be roughly the same as
an X-carrying sperm. However, pure weight let aside, chromosomes within sperm
cells are organised in specific territories (Zalenskaya and Zalensky, 2004) and the
presence of an extra chromosome might interfere with the overall organisation and
result in structural abnormalities that might impact the sperms ability to fertilise an
egg. When genotyping Tc1 sperm for the presence of HsChr21 via FISH, we did not
notice any obvious changes in sperm anatomy; however, given the small structure of
these cells, high-resolution imaging and staining of sperm-specific structures would
be necessary to further investigate this possibility.
150 Discussion and Outlook
Another scenario that could lead to a lower transmission rate of HsChr21 via
the male germline involves defects during early embryonic development. Any loss
of HsChr21-carrying embryos before implantation would most likely go unnoticed
during conventional breeding experiments and therefore only manifest in a lower
transmission rate via the male germline.
Pre-implantation loss of embryos carrying HsChr21 could be due to increased
DNA damage that is either already inherited from the father, or by specific damage to
HsChr21 in the early embryo. When investigating γH2AFX, a marker for DNA DSBs,
in the Tc1 mouse we observed a higher level of DNA damage in spermatocytes
which appeared to be on a genome-wide level. However, it is hard to assess whether
the cells with increased γH2AFX would be eliminated during spermatogenesis.
Persisting DNA damage into the sperm cell that would be transmitted to the egg,
could then result in checkpoint activation in the zygote and lead to a loss of HsChr21-
carrying embryos (Gawecka et al., 2013).
Alternatively, differences in embryonic development depending on whether HsChr21
is transmitted via the egg or the sperm could be caused by sexual dimorphism in
the zygote with respect to epigenetic remodelling. During the first few hours after
fertilisation, the two parental genomes exist in separate compartments and are
subjected to different epigenetic reprogramming. As such, the paternal genome
undergoes protamine-to-histone exchange which is completed at ∼3 hours after
fertilisation and followed by active DNA demethylation, which is independent of cell
division and mediated by Tet3 (McLay and Clarke, 2003; Oswald et al., 2000). This
active DNA demethylation of the paternal genome was recently shown to result in
increased DNA damage, which in repair-deficient mutants triggered a checkpoint
preventing the progression to the 2-cell stage (Ladstätter and Tachibana-Konwalski,
2016). Both the BER and replication-coupled repair that requires Rad21 have been
implicated in repairing Tet3-induced lesions; however, the exact mechanism of the
repair process is still unknown. It is possible that the repair relies on homologous
recombination with non-sister chromatids since the DNA damage occurs before
DNA replication and would have been copied onto the new sister strand. In such a
scenario, extensive damage on HsChr21 when inherited via the sperm could not be
repaired since there is no maternally inherited homolog for this chromosome.
This would present an interesting control mechanism for the zygote, not neces-
sarily for aneuploidies as conventional models for trisomies would have a maternal
homolog available for repair, but for protecting species-boundaries and controlling
the amount of paternal DNA contributed to the embryo. Natural examples for this are
Discussion and Outlook 151
the few viable inter-species hybrids in which the parents have different numbers of
chromosomes. These include equine hybrids involving horses, donkeys and zebras
as well as hybrids between sheep and goats (McGovern, 1975). Inter-breeding
between horses and donkeys produces mules or hinnies, depending on whether
the female was a horse or a donkey, respectively. However, mules are much more
common than hinnies. While it cannot be excluded that this is due to environmental
factors, it is noteworthy that horses have 64 chromosomes, whereas donkeys have
62. Therefore, the breed in which the larger genome would be contributed by the
sperm, is less common than the reciprocal cross (Allen and Short, 1997). The same
observation can be made for hybrids between sheep (54 chromosomes) and goats
(60 chromosomes) (Fehilly et al., 6 22; Mine et al., 2000; Pauciullo et al., 2016;
Stewart-Scott et al., 1990). Thus, for viable inter-species hybrids with different num-
bers of chromosomes, the cross in which the larger chromosome set is contributed
via the egg is more common.
To test whether a similar bias underlies the low transmission rate of HsChr21
via the male germline in the Tc1 mouse, embryonic development of the reciprocal
crosses could be examined via live-cell microscopy to test for differences in the
developmental dynamics. As gametes cannot be easily selected to carry HsChr21
and at least 50% of embryos from conventional in vitro fertilisation (IVF) experi-
ments would be wild-type controls, recently developed live-cell DNA visualisation
techniques could be extremely useful to allow genotyping of living embryos (Ma
et al., 2016; Miyanari et al., 2013). For example, methods based on using clustered
regularly interspersed short palindromic repeat (CRISPR) guides to tile larger areas
of HsChr21 to recruit fluorescently-labelled dCas9 to the human chromosome would
allow live-cell visualisation (Qin et al., 2017; Zhou et al., 2017). This would not
only allow the immediate distinction between HsChr21-carrying embryos to avoid
laborious genotyping methods after imaging, but also generate imaging data of an
aneuploid chromosome during the first embryonic cleavage division. If the low trans-
mission rate through the male germline is indeed due to increased DNA damage on
HsChr21 induced by active DNA methylation which cannot be repaired due to the
lack of a homologous chromosome, one would expect a proportion of embryos that
inherited HsChr21 via the sperm to arrest at the 1-cell stage due to activation of the
Chk1-checkpoint as demonstrated in Ladstätter and Tachibana-Konwalski (2016).
152 Discussion and Outlook
6.3 Tc1-specific piRNAs across human LTR array
Profiling the expression of total and smallRNA in Tc1 testes revealed a consistent
downregulation of HsChr21 in meiotic and post-meiotic cells due to meiotic silencing,
with similar patterns being observed for the mouse sex chromosomes. Nevertheless,
we were able to detect HsChr21-derived piRNA sequences in the Tc1 mouse, that
mainly mapped to two locations on the human chromosomes: a conserved piRNA
cluster and a human-specific LTR array.
The LTR array that showed increased piRNA signal in Tc1 animals is one of ∼25
megasatellites that are human-specific expansions of the MaLR class and make up
a large proportion of this repeat class, comprising ∼460kb of DNA in the human
genome (Warburton et al., 2008). Repeat units within these arrays consist of MSTA,
MSTA-int and THE1 elements, and are organised in tandem repeats with varying
numbers of copies. Interestingly, this repeat is proximal to a breakpoint position
of a translocation event in the Tc1 mouse, during which the human centromere
acquired a metacentric position (Gribble et al., 2013). This resulted in the LTR
array being adjacent to a more gene-rich region in the Tc1 mouse, which could
potentially influence its expression. It would therefore be interesting to investigate
the chromatin conformation in the Tc1 mouse to see what effect these translocations
have on the 3D organisation of HsChr21 and whether new chromatin domains are
formed as observed in Franke et al. (2016). However, expression analysis in human
testes also showed high levels of transcription of this repeat in humans, which
is surprising given its pericentromeric position, which is usually associated with
constitutive heterochromatin and transcriptional silencing (Guenatri et al., 2004). It is
possible that the tandem duplicated organisation of this repeat is involved in enabling
a more open chromatin conformation as similar observations were made for protein-
coding tandem arrays. For example, the testis-specific protein, Y-encoded (TSPY)
cluster consists of up to 35 tandemly duplicated copies of the TSPY gene and is one
of the few genes on the Y chromosome that escapes meiotic silencing in humans
(Skaletsky et al., 2003). Similarly, many X-linked genes that show re-expression
post-meiosis are multi-copy genes that are found in palindromic or tandem duplicated
segments (Mueller et al., 2008). It would therefore be interesting to see whether the
specific organisation of these elements makes them more permissive to transcription,
possibly via the formation of secondary structures.
As pachytene piRNAs were shown to also initiate the cleavage of complemen-
tary spermatogenic mRNAs and lncRNAs (Goh et al., 2015; Zhang et al., 2015),
Discussion and Outlook 153
an integration between our piRNA data and our single-cell expression profiling of
spermatogenic cell populations could yield insight into whether piRNAs might be re-
sponsible for some of the observed gene expression changes. For example, mouse
spermatogenic mRNAs that are negatively correlated with HsChr21 expression lev-
els could be potential targets of HsChr21-derived piRNAs. Therefore, overlapping
these mRNA sequences with Tc1-specific piRNAs based on the target specificity
criteria defined by Goh et al. (2015) could show whether some of the gene expres-
sion changes could be explained by piRNA-mediated mRNA cleavage. Potential
candidate mRNAs that show complementarity to HsChr21-derived piRNAs and are
negatively correlated with HsChr21 expression could then be validated via 5’ RACE
(rapid amplification of cDNA ends) to detect novel 5’ ends that are generated through
piRNA-mediated cleavage.
6.4 Meiotic silencing
Meiotic sex chromosome inactivation (MSCI) is the best studied meiotic silencing
response to unpaired chromosomes due to its physiological occurrence in sperma-
tocytes. In contrast to that, the mechanistics underlying the silencing response to
unpaired autosomes are less well understood. Studies in females have shown that
several key components, including BRCA1 and γH2AFX, can be found on unsy-
napsed chromosomes also in oocytes (Garcia-Cruz et al., 2009; Kouznetsova et al.,
2009). However, the silencing response in females is weaker than in males (Cloutier
et al., 2016) and male-specific features, such as an accumulation of H3K9me3 in
the sex body have been identified (Taketo and Naumova, 2013).
Our analyses revealed differences in the level of transcriptional silencing between
the mouse X chromosome and human chromosome 21 within the same cell. As
previously demonstrated by Mahadevaiah et al. (2008), HsChr21 can self-synapse
during meiosis and escape meiotic silencing, which was shown to occur in ∼50% of
cells. However, our data show a gradual decrease in HsChr21 expression across
the spermatocyte population, which is in stark contrast to the pattern we observed
for chromosome X. This could reflect a specific mechanism that reinforces the
silencing of the sex chromosomes due to the otherwise lethal consequences for
spermatocytes upon expression of the Y-encoded Zfy1 and Zfy2 genes (Royo et al.,
2010). In contrast to that, the silencing of autosomes in response to asynapsis can
be more stochastic without having deleterious effects for the cell, but still fulfil its
checkpoint function. As such, no evolutionary pressure enforces a robust silencing
154 Discussion and Outlook
of autosomes in the MSUC response, whereas incomplete MSCI results in apoptosis
(Royo et al., 2010). It would therefore be interesting to test whether potential MSCI-
specific factors exist that enforce the silencing across the sex chromosomes, for
example by establishing a more stable heterochromatic landscape.
Besides differences in the silencing of sex chromosomes and autosomes, the
different behaviour of mouse chromosome X and human chromosome 21 could
also reflect species-specific aspects of the meiotic silencing response. Meiotic
silencing is conserved across the eutherian and metatherian lineage, but is not well
characterised in humans. While the human X and Y chromosomes also form a sex
body enriched for γH2AFX and BRCA1 (Sciurano et al., 2007), a higher degree
of variation in gene expression was observed in both human spermatocytes and
round spermatids (de Vries et al., 2012). This variability in the human silencing
response, if directed by DNA sequence rather than nuclear environment, could be
the underlying reason of the more permissive expression of HsChr21 observed in
Tc1 spermatocytes.
Another interesting aspect that can be addressed using single-cell RNA-Seq
throughout spermatogenesis is the genetic conflict between X- and Y-carrying sper-
matids. Murine sex chromosomes have acquired selfish genetic elements that have
antagonistic functions during spermiogenesis and influence the sex ratio of the
offspring when unbalanced (Good, 2012). The Y-encoded Sly gene is involved
in maintaining meiotic silencing post meiosis (Cocquet et al., 2009), whereas the
X-encoded Slx and Slxl1 genes activate expression from the X and Y chromosome in
spermatids (Cocquet et al., 2012). Interestingly, while depletion of either the Y-linked
or the X-linked genes results in spermatogenic defects and female- or male-biased
sex ratios, respectively (Cocquet et al., 2010; Ellis et al., 2005), depletion of both
genes rescues these phenotypes, demonstrating the interdependency between
these genes (Cocquet et al., 2012).
How the distortion in sex ratio is achieved when one gene outcompetes the other,
and which genes are regulated by Sly and Slx, is however unclear. To address
this questions, one could calculate the Sly to Slx/Slxl1 ratios in individual cells and
correlate the overall expression of the sex chromosomes, as well as the expression
of individual sex-linked genes to this ratio. This could allow the identification of
target genes that are regulated in a Sly/Slx-dependent way and shed light on the
mechanism that underlies the competition between these genes.
Publications
Journal articles published during PhD:
Dixon, J.R., Xu, J., Dileep, V., Zhan, Y., Song, F., Le, V.T., Yardimci, G.G.,
Chakraborty, A., Bann, D.V., Wang, Y., Clark, R., Zhang, L., Yang, H., Liu, T., Iyyanki,
S., An, L., Pool, C., Sasaki, T., Rivera-Mulia, J.C., Ozadam, H., Lajoie, B.R., Kaul,
R., Buckley, M., Lee, K., Diegel, M., Pezic, D., Ernst, C., Hadjur, S., Odom, D.T.,
Stamatoyannopoulos, J.A., Broach, J.R., Hardison, R., Ay, F., Noble, W.S., Dekker,
J., Gilbert, D.M., Yue, F. "Integrative Detection and Analysis of Structural Variation in
Cancer Genomes."
Nature Genetics (In Press)
Lowe, R., Barton, C., Jenkins, C.A., Ernst, C., Forman, O., Fernandez-Twinn,
D.S., Bock, C., Rossiter, S.J., Faulkes, C.G., Ozanne, S.E., Walter, L., Odom,
D.T., Mellersh, C., Rakyan, V. "Ageing-associated DNA methylation dynamics are a
molecular readout of lifespan variation among mammalian species"
Genome Biology (2018) 19, 22
Ernst, C., Odom, D.T., Kutter, C. "The emergence of piRNAs against transposon
invasion to preserve mammalian genome integrity"
Nature Communications (2017) 8, 1411
Ernst, C., Pike, J., Aitken, S.J., Long, H.K., Eling, N., Stojic, L., Ward, M.C.,
Connor, F., Rayner, T.F., Lukk, M., Klose, R.J., Kutter, C., Odom, D.T. "Successful
transmission and transcriptional deployment of a human chromosome via mouse
male meiosis."
eLife (2016) 5, e20235
Rudan, M.V., Barrington, C., Henderson, S., Ernst, C., Odom D.T., Tanay, A.,
Hadjur, S. "Comparative Hi-C reveals that CTCF underlies evolution of chromosomal
domain architecture."
Cell Reports (2015) 10(8):1297-309

References
Acloque, H., Bonnet-Garnier, A., Mompart, F., Pinton, A., and Yerle-Bouissou,
M. (2013). Sperm Nuclear Architecture Is Locally Modified in Presence of a
Robertsonian Translocation t(13;17). PLOS ONE, 8(10):e78005.
Afgan, E., Baker, D., van den Beek, M., Blankenberg, D., Bouvier, D., Cˇech, M.,
Chilton, J., Clements, D., Coraor, N., Eberhard, C., Grüning, B., Guerler, A.,
Hillman-Jackson, J., Von Kuster, G., Rasche, E., Soranzo, N., Turaga, N., Taylor,
J., Nekrutenko, A., and Goecks, J. (2016). The Galaxy platform for accessible,
reproducible and collaborative biomedical analyses: 2016 update. Nucleic Acids
Research, page gkw343.
Agrawal, A., Eastman, Q. M., and Schatz, D. G. (1998). Transposition mediated
by RAG1 and RAG2 and its implications for the evolution of the immune system.
Nature, 394(6695):744–751.
Allen, W. and Short, R. (1997). Interspecific and Extraspecific Pregancies in Equids:
Anything Goes. Journal of Heredity, 88:384–392.
Allis, C. D. and Jenuwein, T. (2016). The molecular hallmarks of epigenetic control.
Nature Reviews Genetics, 17(8):487–500.
Almeida, M. I., Reis, R. M., and Calin, G. A. (2011). MicroRNA history: Discovery,
recent applications, and next frontiers. Mutation Research, 717(1-2):1–8.
Angerer, P., Haghverdi, L., Büttner, M., Theis, F. J., Marr, C., and Buettner, F. (2016).
Destiny: Diffusion maps for large-scale single-cell data in R. Bioinformatics
(Oxford, England), 32(8):1241–1243.
Aravin, A., Gaidatzis, D., Pfeffer, S., Lagos-Quintana, M., Landgraf, P., Iovino,
N., Morris, P., Brownstein, M. J., Kuramochi-Miyagawa, S., Nakano, T., Chien,
M., Russo, J. J., Ju, J., Sheridan, R., Sander, C., Zavolan, M., and Tuschl, T.
(2006). A novel class of small RNAs bind to MILI protein in mouse testes. Nature,
442(7099):203–207.
Aravin, A. A., Sachidanandam, R., Bourc’his, D., Schaefer, C., Pezic, D., Toth,
K. F., Bestor, T., and Hannon, G. J. (2008). A piRNA pathway primed by individ-
ual transposons is linked to de novo DNA methylation in mice. Molecular Cell,
31(6):785–799.
Aravin, A. A., Sachidanandam, R., Girard, A., Fejes-Toth, K., and Hannon, G. J.
(2007). Developmentally regulated piRNA clusters implicate MILI in transposon
control. Science (New York, N.Y.), 316(5825):744–747.
Aravin, A. A., van der Heijden, G. W., Castañeda, J., Vagin, V. V., Hannon, G. J.,
and Bortvin, A. (2009). Cytoplasmic Compartmentalization of the Fetal piRNA
Pathway in Mice. PLOS Genetics, 5(12):e1000764.
158 References
Ashley, T., Walpita, D., and de Rooij, D. G. (2001). Localization of two mammalian
cyclin dependent kinases during mammalian meiosis. Journal of Cell Science,
114(Pt 4):685–693.
Baez, M. V. and Boccaccio, G. L. (2005). Mammalian Smaug is a translational
repressor that forms cytoplasmic foci similar to stress granules. The Journal of
Biological Chemistry, 280(52):43131–43140.
Bailey, T. L., Johnson, J., Grant, C. E., and Noble, W. S. (2015). The MEME Suite.
Nucleic Acids Research, 43(W1):W39–49.
Balhorn, R. (2007). The protamine family of sperm nuclear proteins. Genome
Biology, 8(9):227.
Bannister, A. J. and Kouzarides, T. (2011). Regulation of chromatin by histone
modifications. Cell Research, 21(3):381–395.
Bannister, L. A., Reinholdt, L. G., Munroe, R. J., and Schimenti, J. C. (2004).
Positional cloning and characterization of mouse mei8, a disrupted allelle of the
meiotic cohesin Rec8. Genesis (New York, N.Y.: 2000), 40(3):184–194.
Barau, J., Teissandier, A., Zamudio, N., Roy, S., Nalesso, V., Hérault, Y., Guillou, F.,
and Bourc’his, D. (2016). The DNA methyltransferase DNMT3C protects male
germ cells from transposon activity. Science, 354(6314):909–912.
Bartel, D. P. (2009). MicroRNAs: Target recognition and regulatory functions. Cell,
136(2):215–233.
Bastos, H., Lassalle, B., Chicheportiche, A., Riou, L., Testart, J., Allemand, I.,
and Fouchet, P. (2005). Flow cytometric characterization of viable meiotic and
postmeiotic cells by Hoechst 33342 in mouse spermatogenesis. Cytometry Part
A, 65A(1):40–49.
Baudat, F., Buard, J., Grey, C., Fledel-Alon, A., Ober, C., Przeworski, M., Coop, G.,
and de Massy, B. (2010). PRDM9 is a major determinant of meiotic recombination
hotspots in humans and mice. Science (New York, N.Y.), 327(5967):836–840.
Baudat, F., Manova, K., Yuen, J. P., Jasin, M., and Keeney, S. (2000). Chromosome
synapsis defects and sexually dimorphic meiotic progression in mice lacking
Spo11. Molecular Cell, 6(5):989–998.
Belfort, M., Curcio, M. J., and Lue, N. F. (2011). Telomerase and retrotransposons:
Reverse transcriptases that shaped genomes. Proceedings of the National
Academy of Sciences, 108(51):20304–20310.
Bellvé, A. R., Cavicchia, J. C., Millette, C. F., O’Brien, D. A., Bhatnagar, Y. M., and
Dym, M. (1977). Spermatogenic cells of the prepuberal mouse. Isolation and
morphological characterization. The Journal of Cell Biology, 74(1):68–85.
Bembom, O. (2017). seqLogo: Sequence logos for DNA sequence alignments. R
package version 1.42.0.
References 159
Berger, S. L. (2007). The complex language of chromatin regulation during transcrip-
tion. Nature, 447(7143):407–412.
Beyret, E. and Lin, H. (2011). Pinpointing the expression of piRNAs and function of
the PIWI protein subfamily during spermatogenesis in the mouse. Developmental
Biology, 355(2):215–226.
Bhattacharya, A., Deng, J. M., Zhang, Z., Behringer, R., de Crombrugghe, B.,
and Maity, S. N. (2003). The B subunit of the CCAAT box binding transcription
factor complex (CBF/NF-Y) is essential for early mouse development and cell
proliferation. Cancer Research, 63(23):8167–8172.
Bi, W., Wu, L., Coustry, F., de Crombrugghe, B., and Maity, S. N. (1997). DNA
Binding Specificity of the CCAAT-binding Factor CBF/NF-Y. Journal of Biological
Chemistry, 272(42):26562–26572.
Bishop, D. K., Park, D., Xu, L., and Kleckner, N. (1992). DMC1: A meiosis-specific
yeast homolog of E. coli recA required for recombination, synaptonemal complex
formation, and cell cycle progression. Cell, 69(3):439–456.
Biswas, U., Wetzker, C., Lange, J., Christodoulou, E. G., Seifert, M., Beyer, A., and
Jessberger, R. (2013). Meiotic cohesin SMC1β provides prophase I centromeric
cohesion and is required for multiple synapsis-associated functions. PLoS genetics,
9(12):e1003985.
Blackledge, N. P., Long, H. K., Zhou, J. C., Kriaucionis, S., Patient, R., and Klose,
R. J. (2012). Bio-CAP: A versatile and highly sensitive technique to purify and
characterise regions of non-methylated DNA. Nucleic Acids Research, 40(4):e32–
e32.
Bolcun-Filas, E., Bannister, L. A., Barash, A., Schimenti, K. J., Hartford, S. A.,
Eppig, J. J., Handel, M. A., Shen, L., and Schimenti, J. C. (2011). A-MYB
(MYBL1) transcription factor is a master regulator of male meiosis. Development
(Cambridge, England), 138(15):3319–3330.
Borg, C. L., Wolski, K. M., Gibbs, G. M., and O’Bryan, M. K. (2010). Phenotyping
male infertility in the mouse: How to get the most out of a ’non-performer’. Human
Reproduction Update, 16(2):205–224.
Borneman, A. R., Gianoulis, T. A., Zhang, Z. D., Yu, H., Rozowsky, J., Seringhaus,
M. R., Wang, L. Y., Gerstein, M., and Snyder, M. (2007). Divergence of transcription
factor binding sites across related yeast species. Science (New York, N.Y.),
317(5839):815–819.
Borowski, A. H., Leung, J. D., van der Kaa, J., Hengst, S., Platenburg, G. J., Pieper,
F. R., Perez, C. F., Jirik, F. R., Drayer, J. I., and others (2000). Generation of
transgenic mice and germline transmission of a mammalian artificial chromosome
introduced into embryos by pronuclear microinjection. Chromosome Research,
8(3):183–191.
Bourc’his, D. and Bestor, T. H. (2004). Meiotic catastrophe and retrotransposon
reactivation in male germ cells lacking Dnmt3L. Nature, 431(7004):96–99.
160 References
Bourque, G., Leong, B., Vega, V. B., Chen, X., Lee, Y. L., Srinivasan, K. G., Chew,
J.-L., Ruan, Y., Wei, C.-L., Ng, H. H., and Liu, E. T. (2008). Evolution of the
mammalian transcription factor binding repertoire via transposable elements.
Genome Research, 18(11):1752–1762.
Braun, R. E. (2000). Temporal control of protein synthesis during spermatogenesis.
International Journal of Andrology, 23 Suppl 2:92–94.
Braun, R. E., Behringer, R. R., Peschon, J. J., Brinster, R. L., and Palmiter,
R. D. (1989). Genetically haploid spermatids are phenotypically diploid. Na-
ture, 337(6205):373–376.
Breese, M. R. and Liu, Y. (2013). NGSUtils: A software suite for analyzing and
manipulating next-generation sequencing datasets. Bioinformatics, 29(4):494–
496.
Brennecke, P., Anders, S., Kim, J. K., Kołodziejczyk, A. A., Zhang, X., Proserpio,
V., Baying, B., Benes, V., Teichmann, S. A., Marioni, J. C., and Heisler, M. G.
(2013). Accounting for technical noise in single-cell RNA-seq experiments. Nature
Methods, 10(11):1093–1095.
Brick, K., Smagulova, F., Khil, P., Camerini-Otero, R. D., and Petukhova, G. V. (2012).
Genetic recombination is directed away from functional genomic elements in mice.
Nature, 485(7400):642–645.
Bunting, K. D. (2002). ABC transporters as phenotypic markers and functional
regulators of stem cells. Stem Cells (Dayton, Ohio), 20(1):11–20.
Burgoyne, P. S. and Baker, T. G. (1984). Meiotic pairing and gametogenic failure.
Symposia of the Society for Experimental Biology, 38:349–362.
Burgoyne, P. S. and Baker, T. G. (1985). Perinatal oocyte loss in XO mice and its
implications for the aetiology of gonadal dysgenesis in XO women. Journal of
Reproduction and Fertility, 75(2):633–645.
Burgoyne, P. S., Mahadevaiah, S. K., and Turner, J. M. (2009). The consequences
of asynapsis for mammalian meiosis. Nature Reviews Genetics, 10(3):207–216.
Carmell, M. A., Girard, A., van de Kant, H. J., Bourc’his, D., Bestor, T. H., de Rooij,
D. G., and Hannon, G. J. (2007). MIWI2 Is Essential for Spermatogenesis and
Repression of Transposons in the Mouse Male Germline. Developmental Cell,
12(4):503–514.
Castañeda, J., Genzor, P., van der Heijden, G. W., Sarkeshik, A., Yates, J. R.,
Ingolia, N. T., and Bortvin, A. (2014). Reduced pachytene piRNAs and translation
underlie spermiogenic arrest in Maelstrom mutant mice. The EMBO journal,
33(18):1999–2019.
Cerutti, L., Mian, N., and Bateman, A. (2000). Domains in gene silencing and cell
differentiation proteins: The novel PAZ domain and redefinition of the Piwi domain.
Trends in Biochemical Sciences, 25(10):481–482.
References 161
Chapman, E. J. and Carrington, J. C. (2007). Specialization and evolution of
endogenous small RNA pathways. Nature Reviews. Genetics, 8(11):884–896.
Chen, Z.-x. and Riggs, A. D. (2011). DNA Methylation and Demethylation in Mam-
mals. Journal of Biological Chemistry, 286(21):18347–18353.
Chi, P., Allis, C. D., and Wang, G. G. (2010). Covalent histone modifications:
Miswritten, misinterpreted, and miserased in human cancers. Nature Reviews.
Cancer, 10(7):457–469.
Chirn, G.-w., Rahman, R., Sytnikova, Y. A., Matts, J. A., Zeng, M., Gerlach, D., Yu,
M., Berger, B., Naramura, M., Kile, B. T., and Lau, N. C. (2015). Conserved piRNA
Expression from a Distinct Set of piRNA Cluster Loci in Eutherian Mammals.
PLOS Genetics, 11(11):e1005652.
Chiu, Y.-L. and Greene, W. C. (2008). The APOBEC3 cytidine deaminases: An
innate defensive network opposing exogenous retroviruses and endogenous
retroelements. Annual Review of Immunology, 26:317–353.
Clift, D. and Marston, A. (2011). The Role of Shugoshin in Meiotic Chromosome
Segregation. Cytogenetic and Genome Research, 133(2-4):234–242.
Cloutier, J. M., Mahadevaiah, S. K., ElInati, E., Nussenzweig, A., Tóth, A., and
Turner, J. M. A. (2015). Histone H2AFX Links Meiotic Chromosome Asynapsis to
Prophase I Oocyte Loss in Mammals. PLOS Genetics, 11(10):e1005462.
Cloutier, J. M., Mahadevaiah, S. K., ElInati, E., Tóth, A., and Turner, J. (2016).
Mammalian meiotic silencing exhibits sexually dimorphic features. Chromosoma,
125(2):215–226.
Cocquet, J., Ellis, P. J. I., Mahadevaiah, S. K., Affara, N. A., Vaiman, D., and
Burgoyne, P. S. (2012). A Genetic Basis for a Postmeiotic X Versus Y Chromosome
Intragenomic Conflict in the Mouse. PLOS Genetics, 8(9):e1002900.
Cocquet, J., Ellis, P. J. I., Yamauchi, Y., Mahadevaiah, S. K., Affara, N. A., Ward,
M. A., and Burgoyne, P. S. (2009). The Multicopy Gene Sly Represses the
Sex Chromosomes in the Male Mouse Germline after Meiosis. PLOS Biology,
7(11):e1000244.
Cocquet, J., Ellis, P. J. I., Yamauchi, Y., Riel, J. M., Karacs, T. P. S., Rattigan, A.,
Ojarikre, O. A., Affara, N. A., Ward, M. A., and Burgoyne, P. S. (2010). Deficiency
in the multicopy Sycp3-like X-linked genes Slx and Slxl1 causes major defects in
spermatid differentiation. Molecular Biology of the Cell, 21(20):3497–3505.
Cohen, P. E., Pollack, S. E., and Pollard, J. W. (2006). Genetic Analysis of Chro-
mosome Pairing, Recombination, and Cell Cycle Control during First Meiotic
Prophase in Mammals. Endocrine Reviews, 27(4):398–426.
Cole, F., Kauppi, L., Lange, J., Roig, I., Wang, R., Keeney, S., and Jasin, M. (2012).
Homeostatic control of recombination is implemented progressively in mouse
meiosis. Nature Cell Biology, 14(4):424–430.
162 References
Collins, I. and Newlon, C. S. (1994). Meiosis-specific formation of joint DNA
molecules containing sequences from homologous chromosomes. Cell, 76(1):65–
75.
Cora, E., Pandey, R. R., Xiol, J., Taylor, J., Sachidanandam, R., McCarthy, A. A., and
Pillai, R. S. (2014). The MID-PIWI module of Piwi proteins specifies nucleotide-
and strand-biases of piRNAs. RNA (New York, N.Y.), 20(6):773–781.
Costa, Y., Speed, R., Ollinger, R., Alsheimer, M., Semple, C. A., Gautier, P., Maratou,
K., Novak, I., Höög, C., Benavente, R., and Cooke, H. J. (2005). Two novel proteins
recruited by synaptonemal complex protein 1 (SYCP1) are at the centre of meiosis.
Journal of Cell Science, 118(Pt 12):2755–2762.
Creasy, D., Bube, A., de Rijk, E., Kandori, H., Kuwahara, M., Masson, R., Nolte,
T., Reams, R., Regan, K., Rehm, S., Rogerson, P., and Whitney, K. (2012).
Proliferative and Nonproliferative Lesions of the Rat and Mouse Male Reproductive
System. Toxicologic Pathology, 40(6 suppl):40S–121S.
Cremer, M., von Hase, J., Volm, T., Brero, A., Kreth, G., Walter, J., Fischer, C.,
Solovei, I., Cremer, C., and Cremer, T. (2001). Non-random radial higher-order
chromatin arrangements in nuclei of diploid human cells. Chromosome Research,
9(7):541–567.
Dadoune, J. P. (1995). The nuclear status of human sperm cells. Micron (Oxford,
England: 1993), 26(4):323–345.
Daniel, K., Lange, J., Hached, K., Fu, J., Anastassiadis, K., Roig, I., Cooke, H. J.,
Stewart, A. F., Wassmann, K., Jasin, M., Keeney, S., and Tóth, A. (2011). Mei-
otic homologue alignment and its quality surveillance are controlled by mouse
HORMAD1. Nature Cell Biology, 13(5):599–610.
Davis, M. P. A., van Dongen, S., Abreu-Goodger, C., Bartonicek, N., and Enright, A. J.
(2013). Kraken: A set of tools for quality control and analysis of high-throughput
sequence data. Methods (San Diego, Calif.), 63(1):41–49.
Davisson, M., Akeson, E., Schmidt, C., Harris, B., Farley, J., and Handel, M. (2006).
Impact of trisomy on fertility and meiosis in male mice. Human Reproduction,
22(2):468–476.
De Fazio, S., Bartonicek, N., Di Giacomo, M., Abreu-Goodger, C., Sankar, A.,
Funaya, C., Antony, C., Moreira, P. N., Enright, A. J., and O’Carroll, D. (2011).
The endonuclease activity of Mili fuels piRNA amplification that silences LINE1
elements. Nature, 480(7376):259–263.
de Koning, A. P. J., Gu, W., Castoe, T. A., Batzer, M. A., and Pollock, D. D. (2011).
Repetitive Elements May Comprise Over Two-Thirds of the Human Genome.
PLOS Genetics, 7(12):e1002384.
de Vries, M., Vosters, S., Merkx, G., D’Hauwers, K., Wansink, D. G., Ramos, L., and
de Boer, P. (2012). Human Male Meiotic Sex Chromosome Inactivation. PLOS
ONE, 7(2):e31485.
References 163
Deng, W. and Lin, H. (2002). Miwi, a murine homolog of piwi, encodes a cytoplasmic
protein essential for spermatogenesis. Developmental Cell, 2(6):819–830.
Di Giacomo, M., Barchi, M., Baudat, F., Edelmann, W., Keeney, S., and Jasin, M.
(2005). Distinct DNA-damage-dependent and -independent responses drive the
loss of oocytes in recombination-defective mouse mutants. Proceedings of the
National Academy of Sciences of the United States of America, 102(3):737–742.
Di Giacomo, M., Comazzetto, S., Saini, H., De Fazio, S., Carrieri, C., Morgan, M.,
Vasiliauskaite, L., Benes, V., Enright, A. J., and O’Carroll, D. (2013). Multiple
epigenetic mechanisms and the piRNA pathway enforce LINE1 silencing during
adult spermatogenesis. Molecular Cell, 50(4):601–608.
Disteche, C. M. and Berletch, J. B. (2015). X-chromosome inactivation and escape.
Journal of genetics, 94(4):591–599.
Dixon, J. R., Selvaraj, S., Yue, F., Kim, A., Li, Y., Shen, Y., Hu, M., Liu, J. S., and Ren,
B. (2012). Topological domains in mammalian genomes identified by analysis of
chromatin interactions. Nature, 485(7398):376–380.
Dobin, A., Davis, C. A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P.,
Chaisson, M., and Gingeras, T. R. (2013). STAR: Ultrafast universal RNA-seq
aligner. Bioinformatics, 29(1):15–21.
Dohle, G. R., Elzanaty, S., and van Casteren, N. J. (2012). Testicular biopsy: Clinical
practice and interpretation. Asian Journal of Andrology, 14(1):88–93.
Donovan, P. J. (1998). The germ cell–the mother of all stem cells. The International
Journal of Developmental Biology, 42(7):1043–1050.
Dumont, B. L. (2017). Variation and Evolution of the Meiotic Requirement for
Crossing Over in Mammals. Genetics, 205(1):155–168.
Eaker, S., Pyle, A., Cobb, J., and Handel, M. A. (2001). Evidence for meiotic spin-
dle checkpoint from analysis of spermatocytes from Robertsonian-chromosome
heterozygous mice. Journal of Cell Science, 114(16):2953–2965.
Ecco, G., Cassano, M., Kauzlaric, A., Duc, J., Coluccio, A., Offner, S., Imbeault, M.,
Rowe, H. M., Turelli, P., and Trono, D. (2016). Transposable Elements and Their
KRAB-ZFP Controllers Regulate Gene Expression in Adult Tissues. Developmen-
tal Cell, 36(6):611–623.
Ellis, P. J. I., Clemente, E. J., Ball, P., Touré, A., Ferguson, L., Turner, J. M. A.,
Loveland, K. L., Affara, N. A., and Burgoyne, P. S. (2005). Deletions on mouse Yq
lead to upregulation of multiple X- and Y-linked transcripts in spermatids. Human
Molecular Genetics, 14(18):2705–2715.
Emerson, R. O. and Thomas, J. H. (2009). Adaptive Evolution in Zinc Finger
Transcription Factors. PLOS Genetics, 5(1):e1000325.
164 References
Endo, T., Romer, K. A., Anderson, E. L., Baltus, A. E., de Rooij, D. G., and Page, D. C.
(2015). Periodic retinoic acid–STRA8 signaling intersects with periodic germ-cell
competencies to regulate spermatogenesis. Proceedings of the National Academy
of Sciences, 112(18):E2347–E2356.
Erkek, S., Hisano, M., Liang, C.-Y., Gill, M., Murr, R., Dieker, J., Schübeler, D., van der
Vlag, J., Stadler, M. B., and Peters, A. H. F. M. (2013). Molecular determinants
of nucleosome retention at CpG-rich sequences in mouse spermatozoa. Nature
Structural & Molecular Biology, 20(7):868–875.
Ernst, C., Odom, D. T., and Kutter, C. (2017). The emergence of piRNAs against
transposon invasion to preserve mammalian genome integrity. Nature Communi-
cations, 8(1):1411.
Ernst, C., Pike, J., Aitken, S. J., Long, H. K., Eling, N., Stojic, L., Ward, M. C.,
Connor, F., Rayner, T. F., Lukk, M., Klose, R. J., Kutter, C., and Odom, D. T. (2016).
Successful transmission and transcriptional deployment of a human chromosome
via mouse male meiosis. eLife, 5:e20235.
Fadloun, A., Le Gras, S., Jost, B., Ziegler-Birling, C., Takahashi, H., Gorab, E.,
Carninci, P., and Torres-Padilla, M.-E. (2013). Chromatin signatures and retro-
transposon profiling in mouse embryos reveal regulation of LINE-1 by RNA. Nature
Structural & Molecular Biology, 20(3):332–338.
Faridani, O. R., Abdullayev, I., Hagemann-Jensen, M., Schell, J. P., Lanner, F., and
Sandberg, R. (2016). Single-cell sequencing of the small-RNA transcriptome.
Nature Biotechnology, 34(12):1264–1266.
Fehilly, C. B., Willadsen, S. M., and Tucker, E. M. (1984 Feb 16-22). Interspecific
chimaerism between sheep and goat. Nature, 307(5952):634–636.
Feng, J., Liu, T., Qin, B., Zhang, Y., and Liu, X. S. (2012). Identifying ChIP-seq
enrichment using MACS. Nature Protocols, 7(9):1728–1740.
Fleming, J. D., Pavesi, G., Benatti, P., Imbriano, C., Mantovani, R., and Struhl, K.
(2013). NF-Y coassociates with FOS at promoters, enhancers, repetitive elements,
and inactive chromatin regions, and is stereo-positioned with growth-controlling
transcription factors. Genome Research, 23(8):1195–1209.
Franco, M. J., Sciurano, R. B., and Solari, A. J. (2007). Protein immunolocalization
supports the presence of identical mechanisms of XY body formation in eutherians
and marsupials. Chromosome Research, 15(6):815–824.
Franke, M., Ibrahim, D. M., Andrey, G., Schwarzer, W., Heinrich, V., Schöpflin, R.,
Kraft, K., Kempfer, R., Jerkovic´, I., Chan, W.-L., Spielmann, M., Timmermann, B.,
Wittler, L., Kurth, I., Cambiaso, P., Zuffardi, O., Houge, G., Lambie, L., Brancati,
F., Pombo, A., Vingron, M., Spitz, F., and Mundlos, S. (2016). Formation of new
chromatin domains determines pathogenicity of genomic duplications. Nature,
538(7624):265–269.
References 165
Franke, V., Ganesh, S., Karlic, R., Malik, R., Pasulka, J., Horvat, F., Kuzman, M.,
Fulka, H., Cernohorska, M., Urbanova, J., Svobodova, E., Ma, J., Suzuki, Y., Aoki,
F., Schultz, R. M., Vlahovicek, K., and Svoboda, P. (2017). Long terminal repeats
power evolution of genes and gene expression programs in mammalian oocytes
and zygotes. Genome Research, page gr.216150.116.
Fraser, R. and Lin, C.-J. (2016). Epigenetic reprogramming of the zygote in mice
and men: On your marks, get set, go! Reproduction (Cambridge, England),
152(6):R211.
Fraune, J., Schramm, S., Alsheimer, M., and Benavente, R. (2012). The mam-
malian synaptonemal complex: Protein components, assembly and role in meiotic
recombination. Experimental Cell Research, 318(12):1340–1346.
Friedli, M., Turelli, P., Kapopoulou, A., Rauwel, B., Castro-Díaz, N., Rowe, H. M.,
Ecco, G., Unzu, C., Planet, E., Lombardo, A., Mangeat, B., Wildhaber, B. E.,
Naldini, L., and Trono, D. (2014). Loss of transcriptional control over endogenous
retroelements during reprogramming to pluripotency. Genome Research, page
gr.172809.114.
Fudenberg, G., Imakaev, M., Lu, C., Goloborodko, A., Abdennur, N., and Mirny, L. A.
(2016). Formation of Chromosomal Domains by Loop Extrusion. Cell reports,
15(9):2038–2049.
Fujihara, Y., Satouh, Y., Inoue, N., Isotani, A., Ikawa, M., and Okabe, M.
(2012). SPACA1-deficient male mice are infertile with abnormally shaped sperm
heads reminiscent of globozoospermia. Development (Cambridge, England),
139(19):3583–3589.
Garcia-Cruz, R., Roig, I., Robles, P., Scherthan, H., and Caldés, M. G. (2009). ATR,
BRCA1 and γH2AX localize to unsynapsed chromosomes at the pachytene stage
in human oocytes. Reproductive BioMedicine Online, 18(1):37–44.
Gawecka, J. E., Marh, J., Ortega, M., Yamauchi, Y., Ward, M. A., and Ward, W. S.
(2013). Mouse Zygotes Respond to Severe Sperm DNA Damage by Delaying
Paternal DNA Replication and Embryonic Development. PLOS ONE, 8(2):e56385.
Geiman, T. M. and Muegge, K. (2010). DNA methylation in early development.
Molecular Reproduction and Development, 77(2):105–113.
Geisler, S. J. and Paro, R. (2015). Trithorax and Polycomb group-dependent regula-
tion: A tale of opposing activities. Development, 142(17):2876–2887.
Ghildiyal, M. and Zamore, P. D. (2009). Small silencing RNAs: An expanding
universe. Nature Reviews Genetics, 10(2):94–108.
Ginsburg, M., Snow, M. H., and McLaren, A. (1990). Primordial germ cells in
the mouse embryo during gastrulation. Development (Cambridge, England),
110(2):521–528.
Girard, A. and Hannon, G. J. (2008). Conserved themes in small-RNA-mediated
transposon control. Trends in Cell Biology, 18(3):136–148.
166 References
Girard, A., Sachidanandam, R., Hannon, G. J., and Carmell, M. A. (2006). A
germline-specific class of small RNAs binds mammalian Piwi proteins. Nature,
442(7099):199–202.
Goedecke, W., Eijpe, M., Offenberg, H. H., van Aalderen, M., and Heyting, C. (1999).
Mre11 and Ku70 interact in somatic cells, but are differentially expressed in early
meiosis. Nature Genetics, 23(2):194–198.
Goh, W. S. S., Falciatori, I., Tam, O. H., Burgess, R., Meikar, O., Kotaja, N., Hammell,
M., and Hannon, G. J. (2015). piRNA-directed cleavage of meiotic transcripts
regulates spermatogenesis. Genes & Development, 29(10):1032–1044.
Goldberg, A. D., Allis, C. D., and Bernstein, E. (2007). Epigenetics: A Landscape
Takes Shape. Cell, 128(4):635–638.
Good, J. M. (2012). The Conflict within and the Escalating War between the Sex
Chromosomes. PLOS Genetics, 8(9):e1002955.
Good, J. M., Vanderpool, D., Smith, K. L., and Nachman, M. W. (2011). Extraor-
dinary Sequence Divergence at Tsga8, an X-linked Gene Involved in Mouse
Spermiogenesis. Molecular Biology and Evolution, 28(5):1675–1686.
Gou, L.-T., Dai, P., Yang, J.-H., Xue, Y., Hu, Y.-P., Zhou, Y., Kang, J.-Y., Wang, X.,
Li, H., Hua, M.-M., Zhao, S., Hu, S.-D., Wu, L.-G., Shi, H.-J., Li, Y., Fu, X.-D., Qu,
L.-H., Wang, E.-D., and Liu, M.-F. (2014). Pachytene piRNAs instruct massive
mRNA elimination during late spermiogenesis. Cell Research, 24(6):680–700.
Greaves, I. K., Rangasamy, D., Devoy, M., Marshall Graves, J. A., and Tremethick,
D. J. (2006). The X and Y chromosomes assemble into H2A.Z-containing [cor-
rected] facultative heterochromatin [corrected] following meiosis. Molecular and
Cellular Biology, 26(14):5394–5405.
Greenbaum, I. F., Hale, D. W., and Fuxa, K. P. (1986). The mechanism of autosomal
synapsis and the substaging of zygonema and pachynema from deer mouse
spermatocytes. Chromosoma, 93(3):203–212.
Grey, C., Barthès, P., Chauveau-Le Friec, G., Langa, F., Baudat, F., and de Massy,
B. (2011). Mouse PRDM9 DNA-Binding Specificity Determines Sites of Histone
H3 Lysine 4 Trimethylation for Initiation of Meiotic Recombination. PLoS Biology,
9(10).
Gribble, S. M., Wiseman, F. K., Clayton, S., Prigmore, E., Langley, E., Yang, F.,
Maguire, S., Fu, B., Rajan, D., Sheppard, O., Scott, C., Hauser, H., Stephens, P. J.,
Stebbings, L. A., Ng, B. L., Fitzgerald, T., Quail, M. A., Banerjee, R., Rothkamm, K.,
Tybulewicz, V. L. J., Fisher, E. M. C., and Carter, N. P. (2013). Massively Parallel
Sequencing Reveals the Complex Structure of an Irradiated Human Chromosome
on a Mouse Background in the Tc1 Model of Down Syndrome. PLoS ONE,
8(4):e60482.
Grivna, S. T., Beyret, E., Wang, Z., and Lin, H. (2006). A novel class of small RNAs
in mouse spermatogenic cells. Genes & Development, 20(13):1709–1714.
References 167
Grow, E. J., Flynn, R. A., Chavez, S. L., Bayless, N. L., Wossidlo, M., Wesche, D. J.,
Martin, L., Ware, C. B., Blish, C. A., Chang, H. Y., Pera, R. A. R., and Wysocka,
J. (2015). Intrinsic retroviral reactivation in human preimplantation embryos and
pluripotent cells. Nature, 522(7555):221–225.
Gu, T.-P., Guo, F., Yang, H., Wu, H.-P., Xu, G.-F., Liu, W., Xie, Z.-G., Shi, L., He, X.,
Jin, S.-g., Iqbal, K., Shi, Y. G., Deng, Z., Szabó, P. E., Pfeifer, G. P., Li, J., and Xu,
G.-L. (2011). The role of Tet3 DNA dioxygenase in epigenetic reprogramming by
oocytes. Nature, 477(7366):606–610.
Guenatri, M., Bailly, D., Maison, C., and Almouzni, G. (2004). Mouse centric and
pericentric satellite repeats form distinct functional heterochromatin. The Journal
of Cell Biology, 166(4):493–505.
Guo, H., Ingolia, N. T., Weissman, J. S., and Bartel, D. P. (2010). Mammalian microR-
NAs predominantly act to decrease target mRNA levels. Nature, 466(7308):835–
840.
Gupta, S., Stamatoyannopoulos, J. A., Bailey, T. L., and Noble, W. S. (2007). Quanti-
fying similarity between motifs. Genome Biology, 8:R24.
Hadjur, S., Williams, L. M., Ryan, N. K., Cobb, B. S., Sexton, T., Fraser, P.,
Fisher, A. G., and Merkenschlager, M. (2009). Cohesins form chromosomal cis-
interactions at the developmentally regulated IFNG locus. Nature, 460(7253):410–
413.
Haghverdi, L., Büttner, M., Wolf, F. A., Buettner, F., and Theis, F. J. (2016). Diffusion
pseudotime robustly reconstructs lineage branching. Nature Methods, 13(10):845–
848.
Hahne, F. and Ivanek, R. (2016). Visualizing Genomic Data Using Gviz and Biocon-
ductor. In Statistical Genomics, Methods in Molecular Biology, pages 335–351.
Humana Press, New York, NY.
Hajkova, P., Jeffries, S. J., Lee, C., Miller, N., Jackson, S. P., and Surani, M. A. (2010).
Genome-wide reprogramming in the mouse germ line entails the base excision
repair pathway. Science (New York, N.Y.), 329(5987):78–82.
Hamer, G. (2002). DNA Double-Strand Breaks and -H2AX Signaling in the Testis.
Biology of Reproduction, 68(2):628–634.
Hamer, G., Gell, K., Kouznetsova, A., Novak, I., Benavente, R., and Höög, C. (2006).
Characterization of a novel meiosis-specific protein within the central element of
the synaptonemal complex. Journal of Cell Science, 119(Pt 19):4025–4032.
Hammoud, S. S., Nix, D. A., Zhang, H., Purwar, J., Carrell, D. T., and Cairns,
B. R. (2009). Distinctive chromatin in human sperm packages genes for embryo
development. Nature, 460(7254):473–478.
Hampsey, M. (1998). Molecular genetics of the RNA polymerase II general transcrip-
tional machinery. Microbiology and molecular biology reviews: MMBR, 62(2):465–
503.
168 References
Hancks, D. C. and Kazazian, H. H. (2016). Roles for retrotransposon insertions in
human disease. Mobile DNA, 7:9.
Handel, M. A. and Schimenti, J. C. (2010). Genetics of mammalian meiosis: Regula-
tion, dynamics and impact on fertility. Nature Reviews Genetics, 11(2):124–136.
Hassold, T., Chiu, D., and Yamane, J. A. (1984). Parental origin of autosomal
trisomies. Annals of Human Genetics, 48(2):129–144.
Hassold, T. and Hunt, P. (2001). To err (meiotically) is human: The genesis of human
aneuploidy. Nature Reviews Genetics, 2(4):280–291.
Havecker, E. R., Gao, X., and Voytas, D. F. (2004). The diversity of LTR retrotrans-
posons. Genome Biology, 5:225.
He, L. and Hannon, G. J. (2004). MicroRNAs: Small RNAs with a big role in gene
regulation. Nature Reviews. Genetics, 5(7):522–531.
Heard, E. and Turner, J. (2011). Function of the Sex Chromosomes in Mammalian
Fertility. Cold Spring Harbor Perspectives in Biology, page a002675.
Hermann, A., Goyal, R., and Jeltsch, A. (2004). The Dnmt1 DNA-(cytosine-
C5)-methyltransferase methylates DNA processively with high preference for
hemimethylated target sites. The Journal of Biological Chemistry, 279(46):48350–
48359.
Hernandez, D. and Fisher, E. M. (1999). Mouse autosomal trisomy: Two’s company,
three’sa crowd. Trends in Genetics, 15(6):241–247.
Hill, F. S., Marchetti, F., Liechty, M., Bishop, J., Hozier, J., and Wyrobek, A. J. (2003).
A new FISH assay to simultaneously detect structural and numerical chromoso-
mal abnormalities in mouse sperm. Molecular Reproduction and Development,
66(2):172–180.
Hirano, T., Iwasaki, Y. W., Lin, Z. Y.-C., Imamura, M., Seki, N. M., Sasaki, E., Saito,
K., Okano, H., Siomi, M. C., and Siomi, H. (2014). Small RNA profiling and
characterization of piRNA clusters in the adult testes of the common marmoset, a
model primate. RNA (New York, N.Y.), 20(8):1223–1237.
Hirose, Y., Suzuki, R., Ohba, T., Hinohara, Y., Matsuhara, H., Yoshida, M., Itabashi,
Y., Murakami, H., and Yamamoto, A. (2011). Chiasmata Promote Monopolar
Attachment of Sister Chromatids and Their Co-Segregation toward the Proper
Pole during Meiosis I. PLoS Genetics, 7(3).
Hisano, M., Erkek, S., Dessus-Babus, S., Ramos, L., Stadler, M. B., and Peters, A.
H. F. M. (2013). Genome-wide chromatin analysis in mature mouse and human
spermatozoa. Nature Protocols, 8(12):2449–2470.
Hodges, C. A., LeMaire-Adkins, R., and Hunt, P. A. (2001). Coordinating the
segregation of sister chromatids during the first meiotic division: Evidence for
sexual dimorphism. Journal of Cell Science, 114(Pt 13):2417–2426.
References 169
Homolka, D., Ivanek, R., Capkova, J., Jansa, P., and Forejt, J. (2007). Chromoso-
mal rearrangement interferes with meiotic X chromosome inactivation. Genome
Research, 17(10):1431–1437.
Huang, D. W., Sherman, B. T., and Lempicki, R. A. (2009). Systematic and inte-
grative analysis of large gene lists using DAVID bioinformatics resources. Nature
Protocols, 4(1):44–57.
Huang, W.-P. and Ho, H.-C. (2006). Role of microtubule-dependent membrane
trafficking in acrosomal biogenesis. Cell and Tissue Research, 323(3):495–503.
Hunt, P. A. (2002). Sex Matters in Meiosis. Science, 296(5576):2181–2183.
Hunter, N. and Kleckner, N. (2001). The single-end invasion: An asymmetric
intermediate at the double-strand break to double-holliday junction transition of
meiotic recombination. Cell, 106(1):59–70.
Imbeault, M., Helleboid, P.-Y., and Trono, D. (2017). KRAB zinc-finger proteins
contribute to the evolution of gene regulatory networks. Nature, 543(7646):550–
554.
Inui, M., Martello, G., and Piccolo, S. (2010). MicroRNA control of signal transduction.
Nature Reviews. Molecular Cell Biology, 11(4):252–263.
Ipsaro, J. J., Haase, A. D., Knott, S. R., Joshua-Tor, L., and Hannon, G. J. (2012). The
structural biochemistry of Zucchini implicates it as a nuclease in piRNA biogenesis.
Nature, 491(7423):279–283.
Iqbal, K., Jin, S.-G., Pfeifer, G. P., and Szabó, P. E. (2011). Reprogramming of
the paternal genome upon fertilization involves genome-wide oxidation of 5-
methylcytosine. Proceedings of the National Academy of Sciences of the United
States of America, 108(9):3642–3647.
Ishiguro, T., Tanaka, K., Sakuno, T., and Watanabe, Y. (2010). Shugoshin-PP2A
counteracts casein-kinase-1-dependent cleavage of Rec8 by separase. Nature
Cell Biology, 12(5):500–506.
Ishiuchi, T., Enriquez-Gasca, R., Mizutani, E., Boškovic´, A., Ziegler-Birling, C.,
Rodriguez-Terrones, D., Wakayama, T., Vaquerizas, J. M., and Torres-Padilla,
M.-E. (2015). Early embryonic-like cells are induced by downregulating replication-
dependent chromatin assembly. Nature Structural & Molecular Biology, 22(9):662–
671.
Ishizu, H., Siomi, H., and Siomi, M. C. (2012). Biology of PIWI-interacting RNAs:
New insights into biogenesis and function inside and outside of germlines. Genes
& Development, 26(21):2361–2373.
Ito, S., D’Alessio, A. C., Taranova, O. V., Hong, K., Sowers, L. C., and Zhang, Y.
(2010). Role of Tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal and
inner cell mass specification. Nature, 466(7310):1129–1133.
170 References
Itou, D., Shiromoto, Y., Yukiho, S.-y., Ishii, C., Nishimura, T., Ogonuki, N., Ogura,
A., Hasuwa, H., Fujihara, Y., Kuramochi-Miyagawa, S., and Nakano, T. (2015).
Induction of DNA methylation by artificial piRNA production in male germ cells.
Current biology: CB, 25(7):901–906.
Jacobs, F. M. J., Greenberg, D., Nguyen, N., Haeussler, M., Ewing, A. D., Katzman,
S., Paten, B., Salama, S. R., and Haussler, D. (2014). An evolutionary arms race
between KRAB zinc-finger genes ZNF91/93 and SVA/L1 retrotransposons. Nature,
516(7530):242–245.
Jaenisch, R. and Bird, A. (2003). Epigenetic regulation of gene expression: How
the genome integrates intrinsic and environmental signals. Nature Genetics,
33:245–254.
Jenuwein, T. and Allis, C. D. (2001). Translating the Histone Code. Science,
293(5532):1074–1080.
Kanai, Y., Kanai-Azuma, M., Noce, T., Saido, T. C., Shiroishi, T., Hayashi, Y., and
Yazaki, K. (1996). Identification of two Sox17 messenger RNA isoforms, with
and without the high mobility group box region, and their differential expression in
mouse spermatogenesis. The Journal of Cell Biology, 133(3):667–681.
Karimi, M. M., Goyal, P., Maksakova, I. A., Bilenky, M., Leung, D., Tang, J. X.,
Shinkai, Y., Mager, D. L., Jones, S., Hirst, M., and Lorincz, M. C. (2011). DNA
methylation and SETDB1/H3K9me3 regulate predominantly distinct sets of genes,
retroelements, and chimeric transcripts in mESCs. Cell Stem Cell, 8(6):676–687.
Kato, Y., Kaneda, M., Hata, K., Kumaki, K., Hisano, M., Kohara, Y., Okano, M., Li, E.,
Nozaki, M., and Sasaki, H. (2007). Role of the Dnmt3 family in de novo methylation
of imprinted and repetitive sequences during male germ cell development in the
mouse. Human Molecular Genetics, 16(19):2272–2280.
Kawaoka, S., Izumi, N., Katsuma, S., and Tomari, Y. (2011). 3’ end formation of
PIWI-interacting RNAs in vitro. Molecular Cell, 43(6):1015–1022.
Kazazian, H. H. (2004). Mobile Elements: Drivers of Genome Evolution. Science,
303(5664):1626–1632.
Keeney, S., Giroux, C. N., and Kleckner, N. (1997a). Meiosis-specific DNA double-
strand breaks are catalyzed by Spo11, a member of a widely conserved protein
family. Cell, 88(3):375–384.
Keeney, S., Giroux, C. N., and Kleckner, N. (1997b). Meiosis-specific DNA double-
strand breaks are catalyzed by Spo11, a member of a widely conserved protein
family. Cell, 88(3):375–384.
Kemp, J. R. and Longworth, M. S. (2015). Crossing the LINE Toward Genomic
Instability: LINE-1 Retrotransposition in Cancer. Frontiers in Chemistry, 3:68.
Kent, W. J., Sugnet, C. W., Furey, T. S., Roskin, K. M., Pringle, T. H., Zahler,
A. M., and Haussler, D. (2002). The human genome browser at UCSC. Genome
research, 12(6):996–1006.
References 171
Kent Hamra, F., Schultz, N., Chapman, K. M., Grellhesl, D. M., Cronkhite, J. T.,
Hammer, R. E., and Garbers, D. L. (2004). Defining the spermatogonial stem cell.
Developmental Biology, 269(2):393–410.
Khil, P. P., Smirnova, N. A., Romanienko, P. J., and Camerini-Otero, R. D. (2004). The
mouse X chromosome is enriched for sex-biased genes not subject to selection
by meiotic sex chromosome inactivation. Nature Genetics, 36(6):642–646.
Kim, B., Cooke, H. J., and Rhee, K. (2012). DAZL is essential for stress granule for-
mation implicated in germ cell survival upon heat stress. Development (Cambridge,
England), 139(3):568–578.
Kim, T. H., Abdullaev, Z. K., Smith, A. D., Ching, K. A., Loukinov, D. I., Green,
R. D., Zhang, M. Q., Lobanenkov, V. V., and Ren, B. (2007). Analysis of the
Vertebrate Insulator Protein CTCF-Binding Sites in the Human Genome. Cell,
128(6):1231–1245.
Kimmins, S., Crosio, C., Kotaja, N., Hirayama, J., Monaco, L., Höög, C., van Duin, M.,
Gossen, J. A., and Sassone-Corsi, P. (2007). Differential functions of the Aurora-B
and Aurora-C kinases in mammalian spermatogenesis. Molecular Endocrinology
(Baltimore, Md.), 21(3):726–739.
Kimmins, S. and Sassone-Corsi, P. (2005). Chromatin remodelling and epigenetic
features of germ cells. Nature, 434(7033):583–589.
Kishore, K., de Pretis, S., Lister, R., Morelli, M. J., Bianchi, V., Amati, B., Ecker, J. R.,
and Pelizzola, M. (2015). methylPipe and compEpiTools: A suite of R packages
for the integrative analysis of epigenomics data. BMC Bioinformatics, 16(1).
Kogo, H., Tsutsumi, M., Ohye, T., Inagaki, H., Abe, T., and Kurahashi, H. (2012).
HORMAD1-dependent checkpoint/surveillance mechanism eliminates asynaptic
oocytes. Genes to Cells, 17(6):439–454.
Kohli, R. M. and Zhang, Y. (2013). TET enzymes, TDG and the dynamics of DNA
demethylation. Nature, 502(7472):472–479.
Kouznetsova, A., Wang, H., Bellani, M., Camerini-Otero, R. D., Jessberger, R., and
Höög, C. (2009). BRCA1-mediated chromatin silencing is limited to oocytes with
a small number of asynapsed chromosomes. Journal of Cell Science, 122(Pt
14):2446–2452.
Kuramochi-Miyagawa, S., Kimura, T., Ijiri, T. W., Isobe, T., Asada, N., Fujita, Y., Ikawa,
M., Iwai, N., Okabe, M., Deng, W., Lin, H., Matsuda, Y., and Nakano, T. (2004).
Mili, a mammalian member of piwi family gene, is essential for spermatogenesis.
Development (Cambridge, England), 131(4):839–849.
Kuramochi-Miyagawa, S., Kimura, T., Yomogida, K., Kuroiwa, A., Tadokoro, Y., Fujita,
Y., Sato, M., Matsuda, Y., and Nakano, T. (2001). Two mouse piwi-related genes:
Miwi and mili. Mechanisms of Development, 108(1-2):121–133.
172 References
Kuramochi-Miyagawa, S., Watanabe, T., Gotoh, K., Totoki, Y., Toyoda, A., Ikawa,
M., Asada, N., Kojima, K., Yamaguchi, Y., Ijiri, T. W., Hata, K., Li, E., Matsuda,
Y., Kimura, T., Okabe, M., Sakaki, Y., Sasaki, H., and Nakano, T. (2008). DNA
methylation of retrotransposon genes is regulated by Piwi family members MILI
and MIWI2 in murine fetal testes. Genes & Development, 22(7):908–917.
Kurihara, Y., Kawamura, Y., Uchijima, Y., Amamo, T., Kobayashi, H., Asano, T.,
and Kurihara, H. (2008). Maintenance of genomic methylation patterns during
preimplantation development requires the somatic form of DNA methyltransferase
1. Developmental Biology, 313(1):335–346.
Ladstätter, S. and Tachibana-Konwalski, K. (2016). A Surveillance Mechanism
Ensures Repair of DNA Lesions during Zygotic Reprogramming. Cell, 167(7):1774–
1787.e13.
Lammers, J. H., Offenberg, H. H., van Aalderen, M., Vink, A. C., Dietrich, A. J., and
Heyting, C. (1994). The gene encoding a major component of the lateral elements
of synaptonemal complexes of the rat is related to X-linked lymphocyte-regulated
genes. Molecular and Cellular Biology, 14(2):1137–1146.
Langmead, B., Trapnell, C., Pop, M., and Salzberg, S. L. (2009). Ultrafast and
memory-efficient alignment of short DNA sequences to the human genome.
Genome Biology, 10(3):R25.
Lassalle, B., Bastos, H., Louis, J. P., Riou, L., Testart, J., Dutrillaux, B., Fouchet, P.,
and Allemand, I. (2004). ’Side Population’ cells in adult mouse testis express Bcrp1
gene and are enriched in spermatogonia and germinal stem cells. Development
(Cambridge, England), 131(2):479–487.
Lassar, A. B., Buskin, J. N., Lockshon, D., Davis, R. L., Apone, S., Hauschka,
S. D., and Weintraub, H. (1989). MyoD is a sequence-specific DNA binding protein
requiring a region of myc homology to bind to the muscle creatine kinase enhancer.
Cell, 58(5):823–831.
Lau, N. C., Seto, A. G., Kim, J., Kuramochi-Miyagawa, S., Nakano, T., Bartel, D. P.,
and Kingston, R. E. (2006). Characterization of the piRNA complex from rat testes.
Science (New York, N.Y.), 313(5785):363–367.
Lee, M. T., Bonneau, A. R., and Giraldez, A. J. (2014). Zygotic genome activation
during the maternal-to-zygotic transition. Annual review of cell and developmental
biology, 30:581–613.
Lee, R. C., Feinbaum, R. L., and Ambros, V. (1993). The C. elegans heterochronic
gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell,
75(5):843–854.
Lees-Murdock, D. J. and Walsh, C. P. (2008). DNA methylation reprogramming in
the germ line. Advances in Experimental Medicine and Biology, 626:1–15.
LeMaire-Adkins, R. and Hunt, P. A. (2000). Nonrandom segregation of the mouse
univalent X chromosome: Evidence of spindle-mediated meiotic drive. Genetics,
156(2):775–783.
References 173
Levin, H. L. and Moran, J. V. (2011). Dynamic interactions between transposable
elements and their hosts. Nature Reviews. Genetics, 12(9):615–627.
Li, H. and Durbin, R. (2009). Fast and accurate short read alignment with Burrows-
Wheeler transform. Bioinformatics, 25(14):1754–1760.
Li, X., Ito, M., Zhou, F., Youngson, N., Zuo, X., Leder, P., and Ferguson-Smith,
A. C. (2008). A maternal-zygotic effect gene, Zfp57, maintains both maternal and
paternal imprints. Developmental Cell, 15(4):547–557.
Li, X. Z., Roy, C. K., Dong, X., Bolcun-Filas, E., Wang, J., Han, B. W., Xu, J.,
Moore, M. J., Schimenti, J. C., Weng, Z., and Zamore, P. D. (2013). An Ancient
Transcription Factor Initiates the Burst of piRNA Production during Early Meiosis
in Mouse Testes. Molecular Cell, 50(1):67–81.
Libby, B. J., De La Fuente, R., O’Brien, M. J., Wigglesworth, K., Cobb, J., Inselman,
A., Eaker, S., Handel, M. A., Eppig, J. J., and Schimenti, J. C. (2002). The mouse
meiotic mutation mei1 disrupts chromosome synapsis with sexually dimorphic
consequences for meiotic progression. Developmental Biology, 242(2):174–187.
Lieberman-Aiden, E., van Berkum, N. L., Williams, L., Imakaev, M., Ragoczy, T.,
Telling, A., Amit, I., Lajoie, B. R., Sabo, P. J., Dorschner, M. O., Sandstrom, R.,
Bernstein, B., Bender, M. A., Groudine, M., Gnirke, A., Stamatoyannopoulos,
J., Mirny, L. A., Lander, E. S., and Dekker, J. (2009). Comprehensive Mapping
of Long-Range Interactions Reveals Folding Principles of the Human Genome.
Science, 326(5950):289–293.
Lim, S. L., Qu, Z. P., Kortschak, R. D., Lawrence, D. M., Geoghegan, J., Hempfling,
A.-L., Bergmann, M., Goodnow, C. C., Ormandy, C. J., Wong, L., Mann, J., Scott,
H. S., Jamsai, D., Adelson, D. L., and O’Bryan, M. K. (2015). HENMT1 and piRNA
Stability Are Required for Adult Male Germ Cell Transposon Repression and to De-
fine the Spermatogenic Program in the Mouse. PLOS Genetics, 11(10):e1005620.
Lipkin, S. M., Moens, P. B., Wang, V., Lenzi, M., Shanmugarajah, D., Gilgeous, A.,
Thomas, J., Cheng, J., Touchman, J. W., Green, E. D., Schwartzberg, P., Collins,
F. S., and Cohen, P. E. (2002). Meiotic arrest and aneuploidy in MLH3-deficient
mice. Nature Genetics, 31(4):385–390.
Lorenzi, H. A. and Reeves, R. H. (2006). Hippocampal hypocellularity in the Ts65Dn
mouse originates early in development. Brain Research, 1104(1):153–159.
Love, M. I., Huber, W., and Anders, S. (2014). Moderated estimation of fold change
and dispersion for RNA-seq data with DESeq2. Genome Biology, 15:550.
Lu, F., Liu, Y., Inoue, A., Suzuki, T., Zhao, K., and Zhang, Y. (2016). Establishing
Chromatin Regulatory Landscape during Mouse Preimplantation Development.
Cell, 165(6):1375–1388.
Lu, L.-Y., Wu, J., Ye, L., Gavrilina, G. B., Saunders, T. L., and Yu, X. (2010). RNF8-
dependent histone modifications regulate nucleosome removal during spermato-
genesis. Developmental cell, 18(3):371–384.
174 References
Lun, A. T., McCarthy, D. J., and Marioni, J. C. (2016). A step-by-step workflow for
low-level analysis of single-cell RNA-seq data with Bioconductor. F1000Research,
5:2122.
Ma, H., Tu, L.-C., Naseri, A., Huisman, M., Zhang, S., Grunwald, D., and Pederson, T.
(2016). Multiplexed labeling of genomic loci with dCas9 and engineered sgRNAs
using CRISPRainbow. Nature Biotechnology, 34(5):528–530.
Ma, J.-B., Yuan, Y.-R., Meister, G., Pei, Y., Tuschl, T., and Patel, D. J. (2005).
Structural basis for 5’-end-specific recognition of guide RNA by the A. fulgidus
Piwi protein. Nature, 434(7033):666–670.
Macfarlan, T. S., Gifford, W. D., Driscoll, S., Lettieri, K., Rowe, H. M., Bonanomi,
D., Firth, A., Singer, O., Trono, D., and Pfaff, S. L. (2012). Embryonic stem cell
potency fluctuates with endogenous retrovirus activity. Nature, 487(7405):57–63.
Mahadevaiah, S. K., Bourc’his, D., de Rooij, D. G., Bestor, T. H., Turner, J. M. A.,
and Burgoyne, P. S. (2008). Extensive meiotic asynapsis in mice antagonises
meiotic silencing of unsynapsed chromatin and consequently disrupts meiotic sex
chromosome inactivation. The Journal of Cell Biology, 182(2):263–276.
Mahadevaiah, S. K., Turner, J. M., Baudat, F., Rogakou, E. P., de Boer, P., Blanco-
Rodríguez, J., Jasin, M., Keeney, S., Bonner, W. M., and Burgoyne, P. S. (2001).
Recombinational DNA double-strand breaks in mice precede synapsis. Nature
Genetics, 27(3):271–276.
Malecová, B. and Morris, K. V. (2010). Transcriptional gene silencing through
epigenetic changes mediated by non-coding RNAs. Current Opinion in Molecular
Therapeutics, 12(2):214–222.
Manakov, S. A., Pezic, D., Marinov, G. K., Pastor, W. A., Sachidanandam, R.,
and Aravin, A. A. (2015). MIWI2 and MILI Have Differential Effects on piRNA
Biogenesis and DNA Methylation. Cell Reports, 12(8):1234–1243.
Marston, A. L. and Amon, A. (2004). Meiosis: Cell-cycle controls shuffle and deal.
Nature Reviews Molecular Cell Biology, 5(12):983–997.
Martianov, I., Brancorsini, S., Catena, R., Gansmuller, A., Kotaja, N., Parvinen, M.,
Sassone-Corsi, P., and Davidson, I. (2005). Polar nuclear localization of H1T2,
a histone H1 variant, required for spermatid elongation and DNA condensation
during spermiogenesis. Proceedings of the National Academy of Sciences of the
United States of America, 102(8):2808–2813.
Martin, R. H., Ko, E., and Rademaker, A. (1991). Distribution of aneuploidy in human
gametes: Comparison between human sperm and oocytes. American journal of
medical genetics, 39(3):321–331.
Martinez-Jimenez, C. P., Eling, N., Chen, H.-C., Vallejos, C. A., Kolodziejczyk, A. A.,
Connor, F., Stojic, L., Rayner, T. F., Stubbington, M. J. T., Teichmann, S. A., de la
Roche, M., Marioni, J. C., and Odom, D. T. (2017). Aging increases cell-to-cell
transcriptional variability upon immune stimulation. Science (New York, N.Y.),
355(6332):1433–1436.
References 175
Matsui, T., Leung, D., Miyashita, H., Maksakova, I. A., Miyachi, H., Kimura, H.,
Tachibana, M., Lorincz, M. C., and Shinkai, Y. (2010). Proviral silencing in
embryonic stem cells requires the histone methyltransferase ESET. Nature,
464(7290):927–931.
Mayer, W., Niveleau, A., Walter, J., Fundele, R., and Haaf, T. (2000). Embryogenesis:
Demethylation of the zygotic paternal genome. Nature, 403(6769):501–502.
Mays-Hoopes, L. L., Bolen, J., Riggs, A. D., and Singer-Sam, J. (1995). Preparation
of Spermatogonia, Spermatocytes, and Round Spermatids for Analysis of Gene
Expression Using Fluorescence-Activated Cell Sorting. Biology of Reproduction,
53(5):1003–1011.
McGovern, P. (1975). The Barriers to Interspecific Hybridization in Domestic and
Laboratory Mammals. British Veterinary Journal, 131(6).
McLay, D. W. and Clarke, H. J. (2003). Remodelling the paternal chromatin at
fertilization in mammals. Reproduction (Cambridge, England), 125(5):625–633.
Meistrich, M. L. and Hess, R. A. (2013). Assessment of Spermatogenesis Through
Staging of Seminiferous Tubules. In Carrell, D. T. and Aston, K. I., editors, Sper-
matogenesis, volume 927, pages 299–307. Humana Press, Totowa, NJ.
Messerschmidt, D. M., Knowles, B. B., and Solter, D. (2014). DNA methylation
dynamics during epigenetic reprogramming in the germline and preimplantation
embryos. Genes & Development, 28(8):812–828.
Meuwissen, R. L., Offenberg, H. H., Dietrich, A. J., Riesewijk, A., van Iersel, M., and
Heyting, C. (1992). A coiled-coil related protein specific for synapsed regions of
meiotic prophase chromosomes. The EMBO journal, 11(13):5091–5100.
Mine, n., Kedikilwe, n., Ndebele, n., and Nsoso, n. (2000). Sheep-goat hybrid
born under natural conditions. Small Ruminant Research: The Journal of the
International Goat Association, 37(1-2):141–145.
Miyanari, Y., Ziegler-Birling, C., and Torres-Padilla, M.-E. (2013). Live visualization
of chromatin dynamics with fluorescent TALEs. Nature Structural & Molecular
Biology, 20(11):1321–1324.
Moazed, D. (2009). Small RNAs in transcriptional gene silencing and genome
defence. Nature, 457(7228):413–420.
Moens, P. B., Kolas, N. K., Tarsounas, M., Marcon, E., Cohen, P. E., and Spyropoulos,
B. (2002). The time course and chromosomal localization of recombination-related
proteins at meiosis in the mouse are compatible with models that can resolve
the early DNA-DNA interactions without reciprocal recombination. Journal of Cell
Science, 115(Pt 8):1611–1622.
Molaro, A., Falciatori, I., Hodges, E., Aravin, A. A., Marran, K., Rafii, S., McCombie,
W. R., Smith, A. D., and Hannon, G. J. (2014). Two waves of de novo methylation
during mouse germ cell development. Genes & Development, 28(14):1544–1549.
176 References
Molaro, A. and Malik, H. S. (2016). Hide and seek: How chromatin-based pathways
silence retroelements in the mammalian germline. Current Opinion in Genetics &
Development, 37:51–58.
Monesi, V. (1965). Differential rate of ribonucleic acid synthesis in the autosomes and
sex chromosomes during male meiosis in the mouse. Chromosoma, 17(1):11–21.
Morelli, M. A. and Cohen, P. E. (2005). Not all germ cells are created equal: Aspects
of sexual dimorphism in mammalian meiosis. Reproduction, 130(6):761–781.
Moreno, R. D., Urriola-Muñoz, P., and Lagos-Cabré, R. (2011). The emerging
role of matrix metalloproteases of the ADAM family in male germ cell apoptosis.
Spermatogenesis, 1(3):195–208.
Moses, A. M., Pollard, D. A., Nix, D. A., Iyer, V. N., Li, X.-Y., Biggin, M. D., and Eisen,
M. B. (2006). Large-Scale Turnover of Functional Transcription Factor Binding
Sites in Drosophila. PLOS Computational Biology, 2(10):e130.
Mueller, J. L., Mahadevaiah, S. K., Park, P. J., Warburton, P. E., Page, D. C., and
Turner, J. M. A. (2008). The mouse X chromosome is enriched for multicopy testis
genes showing postmeiotic expression. Nature Genetics, 40(6):794–799.
Mueller, J. L., Skaletsky, H., Brown, L. G., Zaghlul, S., Rock, S., Graves, T., Auger, K.,
Warren, W. C., Wilson, R. K., and Page, D. C. (2013). Independent specialization
of the human and mouse X chromosomes for the male germ line. Nature Genetics,
45(9):1083–1087.
Muerdter, F., Olovnikov, I., Molaro, A., Rozhkov, N. V., Czech, B., Gordon, A., Hannon,
G. J., and Aravin, A. A. (2012). Production of artificial piRNAs in flies and mice.
RNA (New York, N.Y.), 18(1):42–52.
Muñoz-López, M. and García-Pérez, J. L. (2010). DNA transposons: Nature and
applications in genomics. Current Genomics, 11(2):115–128.
Myers, S., Bowden, R., Tumian, A., Bontrop, R. E., Freeman, C., MacFie, T. S.,
McVean, G., and Donnelly, P. (2010). Drive against hotspot motifs in primates
implicates the PRDM9 gene in meiotic recombination. Science (New York, N.Y.),
327(5967):876–879.
Nagamori, I., Kobayashi, H., Shiromoto, Y., Nishimura, T., Kuramochi-Miyagawa, S.,
Kono, T., and Nakano, T. (2015). Comprehensive DNA Methylation Analysis of
Retrotransposons in Male Germ Cells. Cell Reports, 12(10):1541–1547.
Nagaoka, S. I., Hodges, C. A., Albertini, D. F., and Hunt, P. A. (2011). Oocyte-specific
differences in cell-cycle control create an innate susceptibility to meiotic errors.
Current biology: CB, 21(8):651–657.
Najafabadi, H. S., Mnaimneh, S., Schmitges, F. W., Garton, M., Lam, K. N., Yang,
A., Albu, M., Weirauch, M. T., Radovani, E., Kim, P. M., Greenblatt, J., Frey, B. J.,
and Hughes, T. R. (2015). C2H2 zinc finger proteins greatly expand the human
regulatory lexicon. Nature Biotechnology, 33(5):555–562.
References 177
Namekawa, S. H. and Lee, J. T. (2011). Detection of nascent RNA, single-copy DNA
and protein localization by immunoFISH in mouse germ cells and preimplantation
embryos. Nature Protocols, 6(3):270–284.
Namekawa, S. H., VandeBerg, J. L., McCarrey, J. R., and Lee, J. T. (2007). Sex
chromosome silencing in the marsupial male germ line. Proceedings of the
National Academy of Sciences, 104(23):9730–9735.
Neale, M. J., Pan, J., and Keeney, S. (2005). Endonucleolytic processing of covalent
protein-linked DNA double-strand breaks. Nature, 436(7053):1053–1057.
Nicolaidis, P. and Petersen, M. B. (1998). Origin and mechanisms of non-disjunction
in human autosomal trisomies. Human Reproduction (Oxford, England), 13(2):313–
319.
Nishimasu, H., Ishizu, H., Saito, K., Fukuhara, S., Kamatani, M. K., Bonnefond, L.,
Matsumoto, N., Nishizawa, T., Nakanaga, K., Aoki, J., Ishitani, R., Siomi, H., Siomi,
M. C., and Nureki, O. (2012). Structure and function of Zucchini endoribonuclease
in piRNA biogenesis. Nature, 491(7423):284–287.
Nora, E. P., Lajoie, B. R., Schulz, E. G., Giorgetti, L., Okamoto, I., Servant, N., Piolot,
T., van Berkum, N. L., Meisig, J., Sedat, J., Gribnau, J., Barillot, E., Blüthgen, N.,
Dekker, J., and Heard, E. (2012). Spatial partitioning of the regulatory landscape
of the X-inactivation centre. Nature, 485(7398):381–385.
Oakberg, E. F. (1956). A description of spermiogenesis in the mouse and its use
in analysis of the cycle of the seminiferous epithelium and germ cell renewal.
American Journal of Anatomy, 99(3):391–413.
Oakberg, E. F. (1957). Duration of spermatogenesis in the mouse. Nature,
180(4595):1137–1138.
O’Doherty, A., Ruf, S., Mulligan, C., Hildreth, V., Errington, M. L., Cooke, S., Sesay,
A., Modino, S., Vanes, L., Hernandez, D., Linehan, J. M., Sharpe, P. T., Brandner,
S., Bliss, T. V. P., Henderson, D. J., Nizetic, D., Tybulewicz, V. L. J., and Fisher, E.
M. C. (2005). An Aneuploid Mouse Strain Carrying Human Chromosome 21 with
Down Syndrome Phenotypes. Science, 309(5743):2033–2037.
Odom, D. T., Dowell, R. D., Jacobsen, E. S., Gordon, W., Danford, T. W., MacIsaac,
K. D., Rolfe, P. A., Conboy, C. M., Gifford, D. K., and Fraenkel, E. (2007). Tissue-
specific transcriptional regulation has diverged significantly between human and
mouse. Nature Genetics, 39(6):730–732.
Odom, D. T., Zizlsperger, N., Gordon, D. B., Bell, G. W., Rinaldi, N. J., Murray, H. L.,
Volkert, T. L., Schreiber, J., Rolfe, P. A., Gifford, D. K., Fraenkel, E., Bell, G. I.,
and Young, R. A. (2004). Control of pancreas and liver gene expression by HNF
transcription factors. Science (New York, N.Y.), 303(5662):1378–1381.
O’Donnell, L. (2015). Mechanisms of spermiogenesis and spermiation and how they
are disturbed. Spermatogenesis, 4(2).
178 References
Odorisio, T., Rodriguez, T. A., Evans, E. P., Clarke, A. R., and Burgoyne, P. S.
(1998). The meiotic checkpoint monitoring sypapsis eliminates spermatocytes via
p53-independent apoptosis. Nature Genetics, 18(3):257–261.
Offenberg, H. H., Schalk, J. A., Meuwissen, R. L., van Aalderen, M., Kester, H. A.,
Dietrich, A. J., and Heyting, C. (1998). SCP2: A major protein component of the
axial elements of synaptonemal complexes of the rat. Nucleic Acids Research,
26(11):2572–2579.
Okano, M., Bell, D. W., Haber, D. A., and Li, E. (1999). DNA methyltransferases
Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian devel-
opment. Cell, 99(3):247–257.
Oldfield, A. J., Yang, P., Conway, A. E., Cinghu, S., Freudenberg, J. M., Yellaboina,
S., and Jothi, R. (2014). Histone-fold domain protein NF-Y promotes chromatin
accessibility for cell type-specific master transcription factors. Molecular cell,
55(5):708–722.
Oliva, R. and Luís Ballescà, J. (2012). Altered histone retention and epigenetic
modifications in the sperm of infertile men. Asian Journal of Andrology, 14(2):239–
240.
Ooi, S. K. T., Qiu, C., Bernstein, E., Li, K., Jia, D., Yang, Z., Erdjument-Bromage, H.,
Tempst, P., Lin, S.-P., Allis, C. D., Cheng, X., and Bestor, T. H. (2007). DNMT3L
connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA.
Nature, 448(7154):714–717.
Oswald, J., Engemann, S., Lane, N., Mayer, W., Olek, A., Fundele, R., Dean, W.,
Reik, W., and Walter, J. (2000). Active demethylation of the paternal genome in
the mouse zygote. Current Biology, 10(8):475–478.
Pacheco, S., Marcet-Ortega, M., Lange, J., Jasin, M., Keeney, S., and Roig, I. (2015).
The ATM Signaling Cascade Promotes Recombination-Dependent Pachytene
Arrest in Mouse Spermatocytes. PLOS Genetics, 11(3):e1005017.
Parelho, V., Hadjur, S., Spivakov, M., Leleu, M., Sauer, S., Gregson, H. C., Jarmuz,
A., Canzonetta, C., Webster, Z., Nesterova, T., Cobb, B. S., Yokomori, K., Dillon, N.,
Aragon, L., Fisher, A. G., and Merkenschlager, M. (2008). Cohesins functionally
associate with CTCF on mammalian chromosome arms. Cell, 132(3):422–433.
Parker, J. S., Roe, S. M., and Barford, D. (2005). Structural insights into mRNA
recognition from a PIWI domain-siRNA guide complex. Nature, 434(7033):663–
666.
Paronetto, M. P. and Sette, C. (2010). Role of RNA-binding proteins in mammalian
spermatogenesis. International Journal of Andrology, 33(1):2–12.
Parvanov, E. D., Petkov, P. M., and Paigen, K. (2010). Prdm9 controls activation of
mammalian recombination hotspots. Science (New York, N.Y.), 327(5967):835.
References 179
Patersen, M. B., Antonarakis, S. E., Hassold, T. J., Freeman, S. B., Sherman, S. L.,
Avramopoulos, D., and Mikkelsen, M. (1993). Paternal nondisjunction in trisomy
21: Excess of male patients. Human Molecular Genetics, 2(10):1691–1695.
Pauciullo, A., Knorr, C., Perucatti, A., Iannuzzi, A., Iannuzzi, L., and Erhardt, G.
(2016). Characterization of a very rare case of living ewe-buck hybrid using
classical and molecular cytogenetics. Scientific Reports, 6:srep34781.
Peaston, A. E., Evsikov, A. V., Graber, J. H., de Vries, W. N., Holbrook, A. E., Solter,
D., and Knowles, B. B. (2004). Retrotransposons regulate host genes in mouse
oocytes and preimplantation embryos. Developmental Cell, 7(4):597–606.
Pepling, M. E. (2006). From primordial germ cell to primordial follicle: Mammalian
female germ cell development. Genesis (New York, N.Y.: 2000), 44(12):622–632.
Pepling, M. E. and Spradling, A. C. (2001). Mouse Ovarian Germ Cell Cysts Undergo
Programmed Breakdown to Form Primordial Follicles. Developmental Biology,
234(2):339–351.
Perry, J., Palmer, S., Gabriel, A., and Ashworth, A. (2001). A short pseudoautosomal
region in laboratory mice. Genome Research, 11(11):1826–1832.
Pezic, D., Manakov, S. A., Sachidanandam, R., and Aravin, A. A. (2014). piRNA
pathway targets active LINE1 elements to establish the repressive H3K9me3 mark
in germ cells. Genes & Development.
Popp, C., Dean, W., Feng, S., Cokus, S. J., Andrews, S., Pellegrini, M., Jacobsen,
S. E., and Reik, W. (2010). Genome-wide erasure of DNA methylation in mouse
primordial germ cells is affected by AID deficiency. Nature, 463(7284):1101–1105.
Prak, E. T. and Kazazian, H. H. (2000). Mobile elements and the human genome.
Nature Reviews. Genetics, 1(2):134–144.
Qin, P., Parlak, M., Kuscu, C., Bandaria, J., Mir, M., Szlachta, K., Singh, R., Darzacq,
X., Yildiz, A., and Adli, M. (2017). Live cell imaging of low- and non-repetitive
chromosome loci using CRISPR-Cas9. Nature Communications, 8:14725.
Quinlan, A. R. and Hall, I. M. (2010). BEDTools: A flexible suite of utilities for
comparing genomic features. Bioinformatics, 26(6):841–842.
Ramirez, F., Dundar, F., Diehl, S., Gruning, B. A., and Manke, T. (2014). deepTools:
A flexible platform for exploring deep-sequencing data. Nucleic Acids Research,
42(W1):W187–W191.
Rao, S. S. P., Huntley, M. H., Durand, N. C., Stamenova, E. K., Bochkov, I. D.,
Robinson, J. T., Sanborn, A. L., Machol, I., Omer, A. D., Lander, E. S., and Aiden,
E. L. (2014). A 3D map of the human genome at kilobase resolution reveals
principles of chromatin looping. Cell, 159(7):1665–1680.
Rathke, C., Baarends, W. M., Awe, S., and Renkawitz-Pohl, R. (2014). Chromatin
dynamics during spermiogenesis. Biochimica et Biophysica Acta (BBA) - Gene
Regulatory Mechanisms, 1839(3):155–168.
180 References
Reeves, R. H., Baxter, L. L., and Richtsmeier, J. T. (2001). Too much of a good
thing: Mechanisms of gene action in Down syndrome. Trends in genetics: TIG,
17(2):83–88.
Reeves, R. H., Irving, N. G., Moran, T. H., Wohn, A., Kitt, C., Sisodia, S. S., Schmidt,
C., Bronson, R. T., and Davisson, M. T. (1995). A mouse model for Down syndrome
exhibits learning and behaviour deficits. Nature Genetics, 11(2):177–184.
Reuter, M., Berninger, P., Chuma, S., Shah, H., Hosokawa, M., Funaya, C.,
Antony, C., Sachidanandam, R., and Pillai, R. S. (2011). Miwi catalysis is re-
quired for piRNA amplification-independent LINE1 transposon silencing. Nature,
480(7376):264–267.
Robine, N., Lau, N., Balla, S., Jin, Z., Okamura, K., Kuramochi-Miyagawa, S.,
Blower, M. D., and Lai, E. C. (2009). A broadly conserved pathway generates 3’
UTR-directed primary piRNAs. Current biology : CB, 19(24):2066–2076.
Robinson, J. T., Thorvaldsdóttir, H., Winckler, W., Guttman, M., Lander, E. S., Getz,
G., and Mesirov, J. P. (2011). Integrative genomics viewer. Nature Biotechnology,
29(1):24–26.
Robinson, M. D., McCarthy, D. J., and Smyth, G. K. (2010). edgeR: A Bioconduc-
tor package for differential expression analysis of digital gene expression data.
Bioinformatics, 26(1):139–140.
Robinson, M. D. and Oshlack, A. (2010). A scaling normalization method for differ-
ential expression analysis of RNA-seq data. Genome biology, 11(3):1.
Romanienko, P. J. and Camerini-Otero, R. D. (2000). The Mouse Spo11 Gene Is
Required for Meiotic Chromosome Synapsis. Molecular Cell, 6(5):975–987.
Romier, C., Cocchiarella, F., Mantovani, R., and Moras, D. (2003). The NF-YB/NF-YC
structure gives insight into DNA binding and transcription regulation by CCAAT
factor NF-Y. The Journal of Biological Chemistry, 278(2):1336–1345.
Ronsseray, S., Lehmann, M., and Anxolabéhère, D. (1991). The maternally inherited
regulation of P elements in Drosophila melanogaster can be elicited by two P
copies at cytological site 1A on the X chromosome. Genetics, 129(2):501–512.
Ross-Innes, C. S., Stark, R., Teschendorff, A. E., Holmes, K. A., Ali, H. R., Dunning,
M. J., Brown, G. D., Gojis, O., Ellis, I. O., Green, A. R., Ali, S., Chin, S.-F., Palmieri,
C., Caldas, C., and Carroll, J. S. (2012). Differential oestrogen receptor binding is
associated with clinical outcome in breast cancer. Nature.
Rougier, N., Bourc’his, D., Gomes, D. M., Niveleau, A., Plachot, M., Pàldi, A.,
and Viegas-Péquignot, E. (1998). Chromosome methylation patterns during
mammalian preimplantation development. Genes & Development, 12(14):2108–
2113.
References 181
Royo, H., Polikiewicz, G., Mahadevaiah, S. K., Prosser, H., Mitchell, M., Bradley, A.,
de Rooij, D. G., Burgoyne, P. S., and Turner, J. M. A. (2010). Evidence that meiotic
sex chromosome inactivation is essential for male fertility. Current biology: CB,
20(23):2117–2123.
Royo, H., Prosser, H., Ruzankina, Y., Mahadevaiah, S. K., Cloutier, J. M., Baumann,
M., Fukuda, T., Höög, C., Tóth, A., de Rooij, D. G., Bradley, A., Brown, E. J., and
Turner, J. M. A. (2013). ATR acts stage specifically to regulate multiple aspects of
mammalian meiotic silencing. Genes & Development, 27(13):1484–1494.
Royo, H., Seitz, H., ElInati, E., Peters, A. H. F. M., Stadler, M. B., and Turner, J.
M. A. (2015). Silencing of X-Linked MicroRNAs by Meiotic Sex Chromosome
Inactivation. PLOS Genetics, 11(10):e1005461.
Ruggiu, M., Speed, R., Taggart, M., McKay, S. J., Kilanowski, F., Saunders, P., Dorin,
J., and Cooke, H. J. (1997). The mouse Dazla gene encodes a cytoplasmic protein
essential for gametogenesis. Nature, 389(6646):73–77.
Russell, L. D., Ettlin, R. A., Hikim, A. P. S., and Clegg, E. D. (1993). Histological
and Histopathological Evaluation of the Testis. International Journal of Andrology,
16(1):83–83.
Sage, D., Schindelin, J., l., J., P., and J.-Y., T. (2012). MIJ: Making interoperability
between ImageJ and Matlab possible. In ImageJ User & Developer Conference.
Sago, H., Carlson, E. J., Smith, D. J., Kilbridge, J., Rubin, E. M., Mobley, W. C.,
Epstein, C. J., and Huang, T. T. (1998). Ts1Cje, a partial trisomy 16 mouse model
for Down syndrome, exhibits learning and behavioral abnormalities. Proceedings
of the National Academy of Sciences of the United States of America, 95(11):6256–
6261.
Saitou, M. and Kurimoto, K. (2014). Paternal nucleosomes: Are they retained in
developmental promoters or gene deserts? Developmental Cell, 30(1):6–8.
Sandoval, J. and Esteller, M. (2012). Cancer epigenomics: Beyond genomics.
Current Opinion in Genetics & Development, 22(1):50–55.
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T.,
Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., Tinevez, J.-Y., White, D. J.,
Hartenstein, V., Eliceiri, K., Tomancak, P., and Cardona, A. (2012). Fiji: An open-
source platform for biological-image analysis. Nature Methods, 9(7):676–682.
Schmidt, D., Schwalie, P. C., Wilson, M. D., Ballester, B., Gonçalves, Â., Kutter,
C., Brown, G. D., Marshall, A., Flicek, P., and Odom, D. T. (2012). Waves of
Retrotransposon Expansion Remodel Genome Organization and CTCF Binding in
Multiple Mammalian Lineages. Cell, 148(4):832.
Schramm, S., Fraune, J., Naumann, R., Hernandez-Hernandez, A., Höög, C., Cooke,
H. J., Alsheimer, M., and Benavente, R. (2011). A novel mouse synaptonemal
complex protein is essential for loading of central element proteins, recombination,
and fertility. PLoS genetics, 7(5):e1002088.
182 References
Schultz, D. C., Ayyanathan, K., Negorev, D., Maul, G. G., and Rauscher, F. J. (2002).
SETDB1: A novel KAP-1-associated histone H3, lysine 9-specific methyltrans-
ferase that contributes to HP1-mediated silencing of euchromatic genes by KRAB
zinc-finger proteins. Genes & Development, 16(8):919–932.
Schultz, D. C., Friedman, J. R., and Rauscher, F. J. (2001). Targeting histone deacety-
lase complexes via KRAB-zinc finger proteins: The PHD and bromodomains of
KAP-1 form a cooperative unit that recruits a novel isoform of the Mi-2alpha subunit
of NuRD. Genes & Development, 15(4):428–443.
Schwacha, A. and Kleckner, N. (1995). Identification of double Holliday junctions as
intermediates in meiotic recombination. Cell, 83(5):783–791.
Sciurano, R., Rahn, M., Rey-Valzacchi, G., and Solari, A. J. (2007). The asynaptic
chromatin in spermatocytes of translocation carriers contains the histone variant
gamma-H2AX and associates with the XY body. Human Reproduction (Oxford,
England), 22(1):142–150.
Seki, Y., Yamaji, M., Yabuta, Y., Sano, M., Shigeta, M., Matsui, Y., Saga, Y., Tachibana,
M., Shinkai, Y., and Saitou, M. (2007). Cellular dynamics associated with the
genome-wide epigenetic reprogramming in migrating primordial germ cells in mice.
Development, 134(14):2627–2638.
Sheppard, O., Wiseman, F. K., Ruparelia, A., Tybulewicz, V. L. J., and Fisher, E. M. C.
(2012). Mouse Models of Aneuploidy. The Scientific World Journal, 2012:1–6.
Shi, X., Seluanov, A., and Gorbunova, V. (2007). Cell Divisions Are Required for L1
Retrotransposition. Molecular and Cellular Biology, 27(4):1264–1270.
Shin, Y.-H., Choi, Y., Erdin, S. U., Yatsenko, S. A., Kloc, M., Yang, F., Wang, P. J.,
Meistrich, M. L., and Rajkovic, A. (2010). Hormad1 mutation disrupts synap-
tonemal complex formation, recombination, and chromosome segregation in
mammalian meiosis. PLoS genetics, 6(11):e1001190.
Sin, H.-S., Barski, A., Zhang, F., Kartashov, A. V., Nussenzweig, A., Chen, J., An-
dreassen, P. R., and Namekawa, S. H. (2012a). RNF8 regulates active epigenetic
modifications and escape gene activation from inactive sex chromosomes in
post-meiotic spermatids. Genes & Development, 26(24):2737–2748.
Sin, H.-S., Ichijima, Y., Koh, E., Namiki, M., and Namekawa, S. H. (2012b). Human
postmeiotic sex chromatin and its impact on sex chromosome evolution. Genome
Research, 22(5):827–836.
Sin, H.-S. and Namekawa, S. H. (2013). The great escape: Active genes on inactive
sex chromosomes and their evolutionary implications. Epigenetics, 8(9):887–892.
Sinha, S., Maity, S. N., Lu, J., and de Crombrugghe, B. (1995). Recombinant rat
CBF-C, the third subunit of CBF/NFY, allows formation of a protein-DNA complex
with CBF-A and CBF-B and with yeast HAP2 and HAP3. Proceedings of the
National Academy of Sciences of the United States of America, 92(5):1624–1628.
References 183
Siomi, M. C., Sato, K., Pezic, D., and Aravin, A. A. (2011). PIWI-interacting small
RNAs: The vanguard of genome defence. Nature Reviews. Molecular Cell Biology,
12(4):246–258.
Skaletsky, H., Kuroda-Kawaguchi, T., Minx, P. J., Cordum, H. S., Hillier, L., Brown,
L. G., Repping, S., Pyntikova, T., Ali, J., Bieri, T., Chinwalla, A., Delehaunty, A.,
Delehaunty, K., Du, H., Fewell, G., Fulton, L., Fulton, R., Graves, T., Hou, S.-F.,
Latrielle, P., Leonard, S., Mardis, E., Maupin, R., McPherson, J., Miner, T., Nash,
W., Nguyen, C., Ozersky, P., Pepin, K., Rock, S., Rohlfing, T., Scott, K., Schultz, B.,
Strong, C., Tin-Wollam, A., Yang, S.-P., Waterston, R. H., Wilson, R. K., Rozen, S.,
and Page, D. C. (2003). The male-specific region of the human Y chromosome is
a mosaic of discrete sequence classes. Nature, 423(6942):825–837.
Sleutels, F., Soochit, W., Bartkuhn, M., Heath, H., Dienstbach, S., Bergmaier, P.,
Franke, V., Rosa-Garrido, M., van de Nobelen, S., Caesar, L., van der Reijden,
M., Bryne, J. C., van IJcken, W., Grootegoed, J. A., Delgado, M. D., Lenhard,
B., Renkawitz, R., Grosveld, F., and Galjart, N. (2012). The male germ cell
gene regulator CTCFL is functionally different from CTCF and binds CTCF-like
consensus sites in a nucleosome composition-dependent manner. Epigenetics &
Chromatin, 5:8.
Slotkin, R. K. and Martienssen, R. (2007). Transposable elements and the epigenetic
regulation of the genome. Nature Reviews Genetics, 8(4):272–285.
Smit, A. F. (1993). Identification of a new, abundant superfamily of mammalian
LTR-transposons. Nucleic Acids Research, 21(8):1863–1872.
Smit, A. F. (1999). Interspersed repeats and other mementos of transposable
elements in mammalian genomes. Current Opinion in Genetics & Development,
9(6):657–663.
Smit, A. F. A., Hubley, R., and Green, P. (2015). RepeatMasker Open-4.0. 2013-2015.
Smith, Z. D. and Meissner, A. (2013). DNA methylation: Roles in mammalian
development. Nature Reviews Genetics, 14(3):204–220.
Soh, Y. Q. S., Alföldi, J., Pyntikova, T., Brown, L. G., Graves, T., Minx, P. J., Ful-
ton, R. S., Kremitzki, C., Koutseva, N., Mueller, J. L., Rozen, S., Hughes, J. F.,
Owens, E., Womack, J. E., Murphy, W. J., Cao, Q., de Jong, P., Warren, W. C.,
Wilson, R. K., Skaletsky, H., and Page, D. C. (2014). Sequencing the Mouse Y
Chromosome Reveals Convergent Gene Acquisition and Amplification on Both
Sex Chromosomes. Cell, 159(4):800–813.
Soh, Y. Q. S., Mikedis, M. M., Kojima, M., Godfrey, A. K., de Rooij, D. G., and Page,
D. C. (2017). Meioc maintains an extended meiotic prophase I in mice. PLOS
Genetics, 13(4):e1006704.
Solari, A. J. (1964). THE MORPHOLOGY AND ULTRASTRUCTURE OF THE SEX
VESICLE IN THE MOUSE. Experimental Cell Research, 36:160–168.
Solari, A. J. and Tres, L. L. (1967). The ultrastructure of the human sex vesicle.
Chromosoma, 22(1):16–31.
184 References
Sommer, C., Straehle, C., Köthe, U., and Hamprecht, F. A. (2011). Ilastik: Interactive
learning and segmentation toolkit. In 2011 IEEE International Symposium on
Biomedical Imaging: From Nano to Macro, pages 230–233.
Song, J., Teplova, M., Ishibe-Murakami, S., and Patel, D. J. (2012). Structure-
based mechanistic insights into DNMT1-mediated maintenance DNA methylation.
Science (New York, N.Y.), 335(6069):709–712.
Song, J.-J., Smith, S. K., Hannon, G. J., and Joshua-Tor, L. (2004). Crystal structure
of Argonaute and its implications for RISC slicer activity. Science (New York, N.Y.),
305(5689):1434–1437.
Song, R., Ro, S., Michaels, J. D., Park, C., McCarrey, J. R., and Yan, W. (2009).
Many X-linked microRNAs escape meiotic sex chromosome inactivation. Nature
genetics, 41(4):488–493.
Soper, S. F., van der Heijden, G. W., Hardiman, T. C., Goodheart, M., Martin,
S. L., de Boer, P., and Bortvin, A. (2008). Mouse Maelstrom, a component of
nuage, is essential for spermatogenesis and transposon repression in meiosis.
Developmental cell, 15(2):285–297.
Soumillon, M., Necsulea, A., Weier, M., Brawand, D., Zhang, X., Gu, H., Barthès, P.,
Kokkinaki, M., Nef, S., Gnirke, A., Dym, M., de Massy, B., Mikkelsen, T. S., and
Kaessmann, H. (2013). Cellular Source and Mechanisms of High Transcriptome
Complexity in the Mammalian Testis. Cell Reports, 3(6):2179–2190.
Speed, R. M. (1982). Meiosis in the foetal mouse ovary. I. An analysis at the light
microscope level using surface-spreading. Chromosoma, 85(3):427–437.
Stark, R. and Brown, G. D. (2011). DiffBind: Differential binding analysis of ChIP-Seq
peak data. https://bioconductor.org/packages/release/bioc/html/DiffBind.html.
Stefflova, K., Thybert, D., Wilson, M. D., Streeter, I., Aleksic, J., Karagianni, P.,
Brazma, A., Adams, D. J., Talianidis, I., Marioni, J. C., Flicek, P., and Odom, D. T.
(2013). Cooperativity and Rapid Evolution of Cobound Transcription Factors in
Closely Related Mammals. Cell, 154(3):530–540.
Stewart-Scott, I. A., Pearce, P. D., Dewes, H. F., and Thompson, J. W. (1990). A
case of a sheep-goat hybrid in New Zealand. New Zealand Veterinary Journal,
38(1):7–9.
Sun, H., Treco, D., and Szostak, J. W. (1991). Extensive 3’-overhanging, single-
stranded DNA associated with the meiosis-specific double-strand breaks at the
ARG4 recombination initiation site. Cell, 64(6):1155–1161.
Sun, S.-C. and Kim, N.-H. (2012). Spindle assembly checkpoint and its regulators in
meiosis. Human Reproduction Update, 18(1):60–72.
Svoboda, P., Franke, V., and Schultz, R. M. (2015). Sculpting the Transcriptome Dur-
ing the Oocyte-to-Embryo Transition in Mouse. Current Topics in Developmental
Biology, 113:305–349.
References 185
Tachibana, M., Nozaki, M., Takeda, N., and Shinkai, Y. (2007). Functional dynamics
of H3K9 methylation during meiotic prophase progression. The EMBO journal,
26(14):3346–3359.
Tahiliani, M., Koh, K. P., Shen, Y., Pastor, W. A., Bandukwala, H., Brudno, Y.,
Agarwal, S., Iyer, L. M., Liu, D. R., Aravind, L., and Rao, A. (2009). Conversion of
5-Methylcytosine to 5-Hydroxymethylcytosine in Mammalian DNA by MLL Partner
TET1. Science, 324(5929):930–935.
Taketo, T. and Naumova, A. K. (2013). Oocyte heterogeneity with respect to the
meiotic silencing of unsynapsed X chromosomes in the XY female mouse. Chro-
mosoma, 122(5):337–349.
Tam, P. P. and Snow, M. H. (1981). Proliferation and migration of primordial germ
cells during compensatory growth in mouse embryos. Journal of Embryology and
Experimental Morphology, 64:133–147.
Tate, P. H. and Bird, A. P. (1993). Effects of DNA methylation on DNA-binding proteins
and gene expression. Current Opinion in Genetics & Development, 3(2):226–231.
The Chimpanzee Sequencing and Analysis Consortium (2005). Initial sequence
of the chimpanzee genome and comparison with the human genome. Nature,
437(7055):69–87.
The ENCODE Project Consortium (2012). An integrated encyclopedia of DNA
elements in the human genome. Nature, 489(7414):57–74.
The Initial Human Sequencing Consortium (2001). Initial sequencing and analysis
of the human genome. Nature, 409(6822):860–921.
The Mouse Genome Sequencing Consortium (2002). Initial sequencing and com-
parative analysis of the mouse genome. Nature, 420(6915):520–562.
Thomas, J. H. and Schneider, S. (2011). Coevolution of retroelements and tandem
zinc finger genes. Genome Research, 21(11):1800–1812.
Thompson, P. J., Macfarlan, T. S., and Lorincz, M. C. (2016). Long Terminal Repeats:
From Parasitic Elements to Building Blocks of the Transcriptional Regulatory
Repertoire. Molecular Cell, 62(5):766–776.
Tomizuka, K., Yoshida, H., Uejima, H., Kugoh, H., Sato, K., Ohguma, A., Hayasaka,
M., Hanaoka, K., Oshimura, M., and Ishida, I. (1997). Functional expression
and germline transmission of a human chromosome fragment in chimaeric mice.
Nature Genetics, 16(2):133–143.
Trasler, J. M. (2006). Gamete imprinting: Setting epigenetic patterns for the next
generation. Reproduction, Fertility, and Development, 18(1-2):63–69.
Treangen, T. J. and Salzberg, S. L. (2011). Repetitive DNA and next-generation
sequencing: Computational challenges and solutions. Nature Reviews. Genetics,
13(1):36–46.
186 References
Turner, J. M. A. (2007). Meiotic sex chromosome inactivation. Development,
134(10):1823–1831.
Turner, J. M. A. (2015). Meiotic Silencing in Mammals. Annual Review of Genetics,
49:395–412.
Turner, J. M. A., Mahadevaiah, S. K., Ellis, P. J. I., Mitchell, M. J., and Burgoyne, P. S.
(2006). Pachytene asynapsis drives meiotic sex chromosome inactivation and
leads to substantial postmeiotic repression in spermatids. Developmental Cell,
10(4):521–529.
Turner, J. M. A., Mahadevaiah, S. K., Fernandez-Capetillo, O., Nussenzweig, A., Xu,
X., Deng, C.-X., and Burgoyne, P. S. (2005). Silencing of unsynapsed meiotic
chromosomes in the mouse. Nature Genetics, 37(1):41–47.
Urrutia, R. (2003). KRAB-containing zinc-finger repressor proteins. Genome Biology,
4:231.
Vaquerizas, J. M., Kummerfeld, S. K., Teichmann, S. A., and Luscombe, N. M. (2009).
A census of human transcription factors: Function, expression and evolution.
Nature Reviews Genetics, 10(4):252–263.
Varki, A. and Altheide, T. K. (2005). Comparing the human and chimpanzee genomes:
Searching for needles in a haystack. Genome Research, 15(12):1746–1758.
Viera, A., Rufas, J. S., Martínez, I., Barbero, J. L., Ortega, S., and Suja, J. A.
(2009). CDK2 is required for proper homologous pairing, recombination and
sex-body formation during male mouse meiosis. Journal of Cell Science, 122(Pt
12):2149–2159.
Vietri Rudan, M., Barrington, C., Henderson, S., Ernst, C., Odom, D. T., Tanay, A.,
and Hadjur, S. (2015). Comparative Hi-C reveals that CTCF underlies evolution of
chromosomal domain architecture. Cell Reports, 10(8):1297–1309.
Villar, D., Berthelot, C., Aldridge, S., Rayner, T. F., Lukk, M., Pignatelli, M., Park,
T. J., Deaville, R., Erichsen, J. T., Jasinska, A. J., Turner, J. M. A., Bertelsen, M. F.,
Murchison, E. P., Flicek, P., and Odom, D. T. (2015). Enhancer Evolution across
20 Mammalian Species. Cell, 160(3):554–566.
Villar, D., Flicek, P., and Odom, D. T. (2014). Evolution of transcription factor binding in
metazoans — mechanisms and functional implications. Nature Reviews Genetics,
15(4):221–233.
Voet, T., Vermeesch, J., Carens, A., Dürr, J., Labaere, C., Duhamel, H., David, G.,
and Marynen, P. (2001). Efficient male and female germline transmission of a
human chromosomal vector in mice. Genome research, 11(1):124–136.
von Bülow, M., Heid, H., Hess, H., and Franke, W. W. (1995). Molecular Nature
of Calicin, a Major Basic Protein of the Mammalian Sperm Head Cytoskeleton.
Experimental Cell Research, 219(2):407–413.
References 187
Vourekas, A., Zheng, K., Fu, Q., Maragkakis, M., Alexiou, P., Ma, J., Pillai, R. S.,
Mourelatos, Z., and Wang, P. J. (2015). The RNA helicase MOV10L1 binds piRNA
precursors to initiate piRNA processing. Genes & Development, 29(6):617–629.
Vourekas, A., Zheng, Q., Alexiou, P., Maragkakis, M., Kirino, Y., Gregory, B. D.,
and Mourelatos, Z. (2012). Mili and Miwi target RNA repertoire reveals piRNA
biogenesis and function of Miwi in spermiogenesis. Nature Structural & Molecular
Biology, 19(8):773–781.
Vranis, N. M., van der Heijden, G. W., Malki, S., and Bortvin, A. (2010). Synaptone-
mal Complex Length Variation in Wild-Type Male Mice. Genes, 1(3):505–520.
Walter, M., Teissandier, A., Pérez-Palacios, R., and Bourc’his, D. (2016). An epige-
netic switch ensures transposon repression upon dynamic loss of DNA methylation
in embryonic stem cells. eLife, 5:e11418.
Wang, P. J. (2004). X chromosomes, retrogenes and their role in male reproduction.
Trends in endocrinology and metabolism: TEM, 15(2):79–83.
Wang, P. J., McCarrey, J. R., Yang, F., and Page, D. C. (2001). An abundance of
X-linked genes expressed in spermatogonia. Nature Genetics, 27(4):422–426.
Wang, P. J. and Pan, J. (2007). The role of spermatogonially expressed germ cell-
specific genes in mammalian meiosis. Chromosome Research, 15(5):623–632.
Wang, S., Kou, Z., Jing, Z., Zhang, Y., Guo, X., Dong, M., Wilmut, I., and Gao, S.
(2010). Proteome of mouse oocytes at different developmental stages. Proceed-
ings of the National Academy of Sciences, 107(41):17639–17644.
Wang, W., Yoshikawa, M., Han, B. W., Izumi, N., Tomari, Y., Weng, Z., and Zamore,
P. D. (2014). The Initial Uridine of Primary piRNAs Does Not Create the Tenth
Adenine that Is the Hallmark of Secondary piRNAs. Molecular Cell, 56(5):708–716.
Wang, X., Wei, Y., Fu, G., Li, H., Saiyin, H., Lin, G., Wang, Z., Chen, S., and Yu, L.
(2015). Tssk4 is essential for maintaining the structural integrity of sperm flagellum.
Molecular Human Reproduction, 21(2):136–145.
Wang, Y., Juranek, S., Li, H., Sheng, G., Wardle, G. S., Tuschl, T., and Patel, D. J.
(2009). Nucleation, propagation and cleavage of target RNAs in Ago silencing
complexes. Nature, 461(7265):754–761.
Warburton, P. E., Giordano, J., Cheung, F., Gelfand, Y., and Benson, G. (2004).
Inverted Repeat Structure of the Human Genome: The X-Chromosome Contains
a Preponderance of Large, Highly Homologous Inverted Repeats That Contain
Testes Genes. Genome Research, 14(10a):1861–1869.
Warburton, P. E., Hasson, D., Guillem, F., Lescale, C., Jin, X., and Abrusan, G.
(2008). Analysis of the largest tandemly repeated DNA families in the human
genome. BMC Genomics, 9:533.
188 References
Ward, M. C., Wilson, M. D., Barbosa-Morais, N. L., Schmidt, D., Stark, R., Pan, Q.,
Schwalie, P. C., Menon, S., Lukk, M., Watt, S., Thybert, D., Kutter, C., Kirschner,
K., Flicek, P., Blencowe, B. J., and Odom, D. T. (2013). Latent Regulatory Potential
of Human-Specific Repetitive Elements. Molecular Cell, 49(2):262–272.
Watanabe, T., Cheng, E.-c., Zhong, M., and Lin, H. (2015). Retrotransposons
and pseudogenes regulate mRNAs and lncRNAs via the piRNA pathway in the
germline. Genome Research, 25(3):368–380.
Wei, Y., Yu, L., Bowen, J., Gorovsky, M. A., and Allis, C. D. (1999). Phosphorylation
of histone H3 is required for proper chromosome condensation and segregation.
Cell, 97(1):99–109.
Wen, L. and Tang, F. (2016). Single-cell sequencing in stem cell biology. Genome
Biology, 17:71.
Wendt, K. S., Yoshida, K., Itoh, T., Bando, M., Koch, B., Schirghuber, E., Tsutsumi,
S., Nagae, G., Ishihara, K., Mishiro, T., Yahata, K., Imamoto, F., Aburatani, H.,
Nakao, M., Imamoto, N., Maeshima, K., Shirahige, K., and Peters, J.-M. (2008).
Cohesin mediates transcriptional insulation by CCCTC-binding factor. Nature,
451(7180):796–801.
Weuts, A., Voet, T., Verbeeck, J., Lambrechts, N., Wirix, E., Schoonjans, L., Danloy,
S., Marynen, P., and Froyen, G. (2012). Telomere length homeostasis and telomere
position effect on a linear human artificial chromosome are dictated by the genetic
background. Nucleic Acids Research, 40(22):11477–11489.
Whiddon, J. L., Langford, A. T., Wong, C.-J., Zhong, J. W., and Tapscott, S. J. (2017).
Conservation and innovation in the DUX4-family gene network. Nature Genetics,
49(6):935–940.
Wickham, H. (2009). Ggplot2 Elegant Graphics for Data Analysis. Springer, Dor-
drecht; New York. OCLC: 656394958.
Wilson, M. D., Barbosa-Morais, N. L., Schmidt, D., Conboy, C. M., Vanes, L., Ty-
bulewicz, V. L. J., Fisher, E. M. C., Tavaré, S., and Odom, D. T. (2008). Species-
specific transcription in mice carrying human chromosome 21. Science (New York,
N.Y.), 322(5900):434–438.
Wossidlo, M., Arand, J., Sebastiano, V., Lepikhov, K., Boiani, M., Reinhardt, R.,
Schöler, H., and Walter, J. (2010). Dynamic link of DNA demethylation, DNA
strand breaks and repair in mouse zygotes. The EMBO Journal, 29(11):1877–
1888.
Wu, T. D. and Nacu, S. (2010). Fast and SNP-tolerant detection of complex variants
and splicing in short reads. Bioinformatics, 26(7):873–881.
Wu, X., Schmidt, J. A., Avarbock, M. R., Tobias, J. W., Carlson, C. A., Kolon, T. F.,
Ginsberg, J. P., and Brinster, R. L. (2009). Prepubertal human spermatogo-
nia and mouse gonocytes share conserved gene expression of germline stem
cell regulatory molecules. Proceedings of the National Academy of Sciences,
106(51):21672–21677.
References 189
Yoshida, Y., Suzuki, K., Yamamoto, A., Sakai, N., Bando, M., Tanimoto, K., Yam-
aguchi, Y., Sakaguchi, T., Akhter, H., Fujii, G., Yoshimura, S.-i., Ogata, S., Sohda,
M., Misumi, Y., and Nakamura, N. (2008). YIPF5 and YIF1A recycle between
the ER and the Golgi apparatus and are involved in the maintenance of the Golgi
structure. Experimental Cell Research, 314(19):3427–3443.
Zalenskaya, I. A. and Zalensky, A. O. (2004). Non-random positioning of chromo-
somes in human sperm nuclei. Chromosome research : an international journal on
the molecular, supramolecular and evolutionary aspects of chromosome biology,
12(2):163–173.
Zamudio, N., Barau, J., Teissandier, A., Walter, M., Borsos, M., Servant, N., and
Bourc’his, D. (2015). DNA methylation restrains transposons from adopting a
chromatin signature permissive for meiotic recombination. Genes & Development,
29(12):1256–1270.
Zamudio, N. and Bourc’his, D. (2010). Transposable elements in the mammalian
germline: A comfortable niche or a deadly trap&quest. Heredity, 105(1):92–104.
Zhang, P., Kang, J.-Y., Gou, L.-T., Wang, J., Xue, Y., Skogerboe, G., Dai, P.,
Huang, D.-W., Chen, R., Fu, X.-D., Liu, M.-F., and He, S. (2015). MIWI and
piRNA-mediated cleavage of messenger RNAs in mouse testes. Cell Research,
25(2):193–207.
Zhang, Y., Liu, T., Meyer, C. A., Eeckhoute, J., Johnson, D. S., Bernstein, B. E.,
Nusbaum, C., Myers, R. M., Brown, M., Li, W., and Liu, X. S. (2008). Model-based
Analysis of ChIP-Seq (MACS). Genome Biology, 9(9):R137.
Zhou, Y., Wang, P., Tian, F., Gao, G., Huang, L., Wei, W., and Xie, X. S. (2017).
Painting a specific chromosome with CRISPR/Cas9 for live-cell imaging. Cell
Research, 27(2):298–301.
Zickler, D. and Kleckner, N. (1999). Meiotic Chromosomes: Integrating Structure
and Function. Annual Review of Genetics, 33(1):603–754.
Zindy, F., den Besten, W., Chen, B., Rehg, J. E., Latres, E., Barbacid, M., Pollard,
J. W., Sherr, C. J., Cohen, P. E., and Roussel, M. F. (2001). Control of Spermatoge-
nesis in Mice by the Cyclin D-Dependent Kinase Inhibitors p18Ink4c and p19Ink4d.
Molecular and Cellular Biology, 21(9):3244–3255.

A | Analysis scripts
Image analysis using FIJI
Code A.1 | Quantification of FISH stainings
# This script serves to quantify HsChr21−positive cells using spermatids as an example.
open("DAPI−Tc1_19736_spermatids_max_1−4.tif");
# Otsu threshold works best for distinguishing the fainter background cells that are likely dead cells from centrifugation.
setAutoThreshold("Otsu dark");
# setThreshold(47, 255);
setOption("BlackBackground", false);
run("Convert to Mask");
# Distinguish between connected cells
run("Watershed");
# Most spermatids are around 40um in area most likley due to spreading out from centrifugation, 20−60 excludes smaller
debris particles as well as the larger cells that are most likely wrongly sorted. For spermatocytes use 40−120.
run("Analyze Particles...", "size=20−60 show=[Count Masks] display exclude summarize add");
selectWindow("Count Masks of DAPI−Tc1_19736_spermatids_max_1−4.tif");
# Add Gaussian Blur to the count mask to increase the radius of cells slightly as the FISH signal is often located towards
the nuclear membrane since thresholding decreased cell size a bit.
run("Gaussian Blur...", "sigma=3.50");
open("HsChr21−Tc1_19736_spermatids_max_1−4.tif");
# Best ’dot detection’ technique for FISH signal is to generate two blurred images and subtract them from each other to get
sharp defined points.
run("Duplicate...", "title=HsChr21−Tc1_19736_spermatids_max_1_bigGaussian.tif");
selectWindow("HsChr21−Tc1_19736_spermatids_max_1−4.tif");
run("Duplicate...", "title=HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif");
run("Gaussian Blur...", "sigma=2.50");
selectWindow("HsChr21−Tc1_19736_spermatids_max_1_bigGaussian.tif");
run("Gaussian Blur...", "sigma=3.50");
imageCalculator("Subtract create", "HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif","HsChr21−
Tc1_19736_spermatids_max_1_bigGaussian.tif");
selectWindow("Result of HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif");
run("Enhance Contrast", "saturated=0.35");
# Apply the same threshold method to the created image
setAutoThreshold("Otsu dark");
# Count everything up to a diameter of 20um, to exclude the larger background signal
run("Analyze Particles...", "size=0−20 show=[Count Masks] display exclude summarize add");
# Multiplication of the count masks for both the nuclei and the FISH signal will ony result in signal when the two overlap (
otherwise multiplication by 0).
192 Analysis scripts
imageCalculator("Multiply create", "Count Masks of Result of HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif
","Count Masks of DAPI−Tc1_19736_spermatids_max_1−4.tif");
selectWindow("Result of Count Masks of Result of HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif");
setAutoThreshold("Otsu dark");
# Count regions that overlap DAPI and HsChr21 signal
run("Analyze Particles...", "size=0−60 show=[Count Masks] display exclude summarize add");
selectWindow("Result of Count Masks of Result of HsChr21−Tc1_19736_spermatids_max_1_smallGaussian.tif");
ChIP-seq analysis
Filtering
Code A.2 | Filtering script for ChIP-Seq libraries
−−−
#!/bin/bash
# $Id: filtering_chip_libraries.sh 2017−05−12 ernst01 $
−−−
name=’${1}’
# Filter out multi−mapping reads form BED file
date +"%d−%m−%Y %H:%M:%S Info: Filter out non−unique reads"
grep −v $’\t0\t’ ${1}.bed > ${1}_unique.bed
echo "Info: file ${1}_unique.bed generated"
# Filter against ENCODE blacklist
date +"%d−%m−%Y %H:%M:%S Info: remove ENCODE blacklisted regions for mm10 and chr21_hg19 (Use hg19
blacklist for human libraries)"
bedtools subtract −a ${1}_unique.bed −b /data03/users/ernst01/thesis/filtering_lists/blacklist_tc1_mm10_hg19.bed > ${1}
_unique_blackfilter.bed
echo "Info: ${1}_unique_blackfilter.bed generated"
# Filter out reads that are deleted or duplicated in the Tc1 mouse
date +"%d−%m−%Y %H:%M:%S Info: Filter out deleted or duplicated regions in the Tc1 mouse"
bedtools subtract −a ${1}_unique_blackfilter.bed −b /data03/users/ernst01/thesis/filtering_lists/deldup_hg19.bed > ${1}
_filtered.bed
echo "Info: ${1}_unique_blackgreydeldupfilter.bed generated"
# Count reads that are going to be fed into peak calling
wc −l ${1}_filtered.bed
Analysis scripts 193
Peak calling
Code A.3 | Peak calling using MACS2 with summits and pileup
−−−
#!/bin/bash
# $Id: peak_calling_tc1_testis.sh 2017−05−26 ernst01 $
−−−
# Call peaks with MACS2 using −B and −SPMR to retain summit read pileup information
date +"%d−%m−%Y %H:%M:%S Info: call peaks with Macs2 against Input_tc1_fg_testes.bed, output including sequence
tag pileup files"
macs2 callpeak −t ${1}_filtered.bed −c /data03/users/ernst01/thesis/testis/Tc1_FG/Input_tc1_fg_testes.bed −f BED −g
mm −B −−SPMR −n ${1}
echo "Info: macs peakcalling finished"
# Generate read pileup files using fold enrichment method
date +"%d−%m−%Y %H:%M:%S Info: Generate bedGraph file for fold enrichment"
macs2 bdgcmp −t ${1}_treat_pileup.bdg −c ${1}_control_lambda.bdg −o ${1}_FE.bdg −m FE
echo "Info: generated bedGraph for fold enrichment"
# Bedgraph file of FE for chr21 only
date +"%d−%m−%Y %H:%M:%S Info: Generate bedGraph file for chr21 only"
awk ’/chr21/{print}’ ${1}_FE.bdg > ${1}_FE_chr21.bdg
echo "Info: generated bedGraph for fold enrichment for chr21"
DiffBind
Code A.4 | DiffBind workflow - Human versus Tc1
−−−
title: "Differential binding analysis between Human and Tc1 mouse livers"
author: "Christina Ernst"
date: "8/19/2016"
output: pdf_document
−−−
setwd("~/Documents/PhD/Analysis_thesis/DiffBind/Liver")
library(DiffBind)
# Read in sample sheet
liver_chr21 = dba(sampleSheet = "H3K4me3_HsavsFG_chr21_liver.csv")
# Take read counts into consideration requiring peaks to be present in at least half of the replicates. TMM normalisation
from egeR is being using, using ChIP read counts minus Control read counts and full library size.
liver_chr21_counts = dba.count(liver_chr21, minOverlap = 0.5, score = DBA_SCORE_TMM_MINUS_EFFECTIVE,
bScaleControl = TRUE, summits = TRUE, bParallel = TRUE)
# This calculates the fraction of reads in peaks (FRiP) which can be used as a quality metric, libraries with a low FRiP will
be excluded from analysis
# Establish contrast between human and Tc1 mouse
194 Analysis scripts
liver_chr21_contrast = dba.contrast(liver_chr21_counts, categories = DBA_CONDITION)
# Differential binding analysis
liver_chr21_dba = dba.analyze(liver_chr21_contrast)
# Set threshold to 1 to obtain all regions
liver_chr21_db = dba.report(liver_chr21_dba, th = 1)
liver_chr21_DB = as.data.frame(liver_chr21_db)
write.csv(liver_chr21_DB, "H3K4me3_liver_Human_vs_Tc1_2.csv")
# Manipulate data obtained for differential binding analysis for plotting in ggplot2 using dplyr
library(ggplot2)
library(dplyr)
liver_chr21_ggplot = liver_chr21_DB
# Minimum fold change of 2.5 to be considered as differentially bound
liver_chr21_ggplot←mutate(liver_chr21_ggplot, Fold_change = Fold < −3.5 | Fold > 3.5)
# Significantly differentially bound with fold change of 2.5 and FDR <0.1
liver_chr21_ggplot←mutate(liver_chr21_ggplot, Significant = FDR < 0.1 & Fold_change == "TRUE")
# Set negative values to zero
liver_chr21_ggplot←liver_chr21_ggplot %>% mutate(Conc_FG = replace(Conc_FG, Conc_FG < 0, 0))
liver_chr21_ggplot←liver_chr21_ggplot %>% mutate(Conc_Has = replace(Conc_Has, Conc_Has < 0, 0))
ggplot(liver_chr21_ggplot, aes(x = Conc_Has, y = Conc_FG, colour = Significant)) + geom_point(size = 1.5, shape = 19) +
ggtitle("H3K4me3 liver Human vs Tc1 − chr21") + labs(x = "Human", y = "Tc1") + scale_colour_manual(values = c("
gray28", "deeppink3"))
# Defined peak summits − summit=25 recenters the peaks on the summit that is determined based on read counts and
extracts 25 bp up− and downstream of the summit
liver_summits = dba.count(liver_chr21, minOverlap = 0.5, score=DBA_SCORE_TMM_MINUS_EFFECTIVE, bScaleControl
= TRUE, summits = 25)
liver_summits_50 = dba.peakset(liver_summits, bRetrieve = T)
liver_summits_50 = as.data.frame(liver_summits_50)
write.csv(liver_summits_50, "liver_summits_50.csv")
write.table(liver_summits_50, file="liver_summits_50.bed", quote = F, sep = "\t", row.names = F, col.names = F)
RNA-seq
Differential expression
Code A.5 | Differential expression analysis of total RNA-seq using DeSeq2
−−−
title: "DESeq on adult total RNAseq libraries"
author: "Christina Ernst"
date: "1/19/2017"
output: pdf_document
−−−
library(ggplot2)
library(geneplotter)
library(plyr)
library(LSD)
library(DESeq2)
Analysis scripts 195
library(gplots)
library(RColorBrewer)
library(stringr)
library("topGO", lib.loc="/sw/gentoo/usr/lib/R/library")
library(genefilter)
library(biomaRt)
library(dplyr)
# Generate metadata file including sample name, condition and library name
metadata_adult←data.frame(sampleNo = c("do4029", "do4030", "do4031", "do4032", "do4033", "do4034", "do4035"),
condition = c("tc0", "tc0", "tc0", "tc1", "tc1", "tc1", "tc0"), libraryName = c("do4029_testis", "do4030_testis", "do4031_
testis", "do4032_testis", "do4033_testis", "do4034_testis", "do4035_testis"))
metadata_adult = mutate(metadata_adult, countFile = paste0(metadata_adult$libraryName, "_genecount.txt"))
metadata_adult←lapply(metadata_adult, as.character)
#Create DESeq2 summarized experiment object from the count files containing metadata on the genomic features
accesible via rowData
sampleTable←data.frame(sampleName = metadata_adult$libraryName, fileName = metadata_adult$countFile, condition =
metadata_adult$condition, sampleNo=metadata_adult$sampleNo)
dds_adult←DESeqDataSetFromHTSeqCount(sampleTable = sampleTable, design = ~ condition)
# Keep only genes with at least 5 counts
dds_adult←dds_adult[ rowSums(counts(dds_adult)) >4,]
# 33418 out of 44216 genes have at least two counts, 31892 genes have at least 5 counts.
# Add additional annotation information using biomaRt − in this case gene symbols and description of gene
mmu←useMart("ensembl", dataset="mmusculus_gene_ensembl")
hsa←useMart("ensembl", dataset="hsapiens_gene_ensembl")
bm_mmu←getBM(attributes=c("ensembl_gene_id", "external_gene_name", "description"), filter="ensembl_gene_id",
values=rownames(dds_total), mart=mmu)
bm_hsa←getBM(attributes=c("ensembl_gene_id", "external_gene_name", "description"), filter="ensembl_gene_id", values
=rownames(dds_total), mart=hsa)
# Combine all mouse and HsChr21 annottations ot have Tc1 annotation
bm_tc1←rbind(bm_hsa, bm_mmu)
save(bm_tc1, file= "bm_tc1.RData")
load(’bm_tc1.RData’)
# Add additional annotation data to the DESeqDataSet
DESeq2Features←data.frame(ensembl_gene_id = rownames(dds_adult))
DESeq2Features$ensembl_gene_id←as.character(DESeq2Features$ensembl_gene_id)
rowData←dplyr::left_join(DESeq2Features, bm_tc1, by = "ensembl_gene_id")
rowData←as(rowData, "DataFrame")
mcols(rowRanges(dds_adult))←c(mcols(rowRanges(dds_adult)),rowData)
# Normalization − estimating the size factor − in this case the raw read counts are not actually altered, DESeq2 defines a
virtual reference sample by taking the median of each gene’s values across samples and then computes size factors
as the median of ratios of each sample to the reference sample
GeneCounts←counts(dds_adult)
idx.nz←apply(GeneCounts, 1, function(x) { all(x >0)})
sum(idx.nz)
con←as.character(colData(dds_adult)$condition)
colData(dds_adult)$condition←factor(dds_adult$condition, levels = c("tc0", "tc1"))
dds_adult←estimateSizeFactors((dds_adult))
196 Analysis scripts
sizeFactors(dds_adult)
# Plot densities of counts for the different samples to assess their distributions
multiecdf(counts(dds_adult, normalized = T)[idx.nz ,], xlab = "mean counts", xlim=c(0,10000))
multidensity(counts(dds_adult, normalized=T)[idx.nz,], xlab="mean counts", xlim=c(0,1000))
#PCA and sample heatmaps
#Produce rlog−transformed data to stabilize the variance of the data and avoid that the distance measure is dominated by
a few highly variable genes.
rld_adult←rlogTransformation(dds_adult, blind=TRUE)
distsRL_adult←dist(t(assay(rld_adult)))
mat_adult←as.matrix(distsRL_adult)
hmcol←colorRampPalette(brewer.pal(9, "GnBu"))(100)
heatmap.2(mat_adult, trace="none", col=rev(hmcol), margin=c(13,13))
#PCA plot
DESeq2::plotPCA(rld, intgroup=c("condition"))
# Tc0 samples cluster nicely together and are separated from Tc1 samples which show a higher variance
# Statistical testing of differential expression
dds_adult←estimateDispersions((dds_adult))
plotDispEsts(dds_adult)
# Statistical testing of differential expression, use FDR of 0.001
dds_adult←DESeq(dds_adult, betaPrior = TRUE)
results_adult←results(dds_adult, alpha=.001)
summary(results_adult)
results_adult
# Export results data into data frame
results_adult_df←results(dds_adult, alpha=.01, tidy=TRUE)
#Remove NA values in padj column with 1
results_adult_df[is.na(results_adult_df)]←1
#Add column for significance based on FDR <0.001
mutate_adultddf←mutate(results_adult_df, sig=ifelse(results_adult_df$padj<0.001, "Sig", "Not Sig"))
#Add column for log2FoldChange threshold of 1
adult_fold←mutate(mutate_adultddf, fold=log2FoldChange < −1 | log2FoldChange >1 )
#Add column for genes to be colored in plot
adult_anno←mutate(adult_fold, anno=fold=="TRUE" & sig =="Sig")
#Annoate human genes
test_human←mutate(adult_anno, human = row %like% "^ENSG")
#Add column for labeling genes with a log2Fold change greater than 1.5
ub_fold1.5←mutate(test_human, fold15=log2FoldChange < −1.5 | log2FoldChange >1.5 )
sub←subset(sub_fold1.5, anno == "TRUE" & human == "FALSE" & fold15 == "TRUE")
colnames(sub)[1]←"ensembl_gene_id"
test_sub← merge(x = sub, y = bm_tc1, by = "ensembl_gene_id")
#Generate ggplo2 object for MA plot
adult_MA = ggplot(test_human, aes(baseMean, log2FoldChange)) + scale_x_log10() + geom_point(aes(col=anno, shape=
human, size=2.5)) + scale_color_manual(values=c("gray28", "darkgreen")) + scale_shape_manual(values=c(16, 5)) +
geom_text_repel(data = test_sub, aes(label = external_gene_name))
adult_MA
Analysis scripts 197
smallRNA-seq analysis
Filtering of cross-mapping reads
Code A.6 | Python script to remove cross-mapping smallRNA reads
−−−
#!/usr/bin/env python
# $Id: filter_bam_smrna.py 3344 2014−07−31 10:11:23Z tfrayner $
−−−
Script to remove all the small−RNA reads listed in a given fasta file from a bam file.
import os
import re
import pysam
import string
from tempfile import NamedTemporaryFile
REVCOMP = string.maketrans(’ACTGactg’, ’TGACtgac’)
def read_fasta_sequences(fasta):
seqre = re.compile(’^[^>]’)
# pysam.Fastafile seems underdeveloped at the moment.
print "Reading fasta file..."
with open(fasta, ’rb’) as infasta:
unwanted = [ x.strip() for x in infasta if seqre.match(x) ]
return unwanted
def filter_bam_reads(infile, outfile, unwanted):
tmpfiles = []
# Use sets for fast lookup.
print "Calculating reverse complement sequences..."
unwanted = set(unwanted + [x.translate(REVCOMP)[::−1] for x in unwanted])
bai = "%s.bai" % infile
if not os.path.exists(bai):
print "Indexing bam file %s..." % infile
pysam.index(infile)
if not os.path.exists(bai):
# Failure most likely due to unsorted bam
print "Indexing failed; retrying on sorted file..."
sortfile = NamedTemporaryFile(suffix=’.bam’)
pysam.sort(infile, os.path.splitext(sortfile.name)[0])
infile = sortfile.name
bai = "%s.bai" % infile
tmpfiles.extend([bai])
pysam.index(infile)
if not os.path.exists(bai):
raise StandardError("Retry failed. Unable to autogenerate bam index.")
print "Filtering bam file..."
with pysam.Samfile(infile, ’rb’) as inbam:
with pysam.Samfile(outfile, ’wb’, header=inbam.header) as outbam:
198 Analysis scripts
# until_eof=True returns unmapped as well as mapped reads.
for read in inbam.fetch(until_eof=True):
if not str(read.seq) in unwanted:
outbam.write(read)
outbam.reset()
for tmp in tmpfiles:
os.unlink(tmp)
if __name__ == ’__main__’:
from argparse import ArgumentParser
PARSER = ArgumentParser(
description=’Filter a bam file by removing reads as specified in a fasta file.’)
PARSER.add_argument(’−b’, ’−−bam’, dest=’bam’, type=str, required=True,
help=’The bam file to be filtered.’)
PARSER.add_argument(’−f’, ’−−fasta’, dest=’fasta’, type=str, required=True,
help=’The fasta file containing the sequences to remove.’)
PARSER.add_argument(’−o’, ’−−out’, dest=’output’, type=str, required=True,
help=’The name of the output bam file.’)
ARGS = PARSER.parse_args()
SEQS = read_fasta_sequences(ARGS.fasta)
filter_bam_reads(ARGS.bam, ARGS.output, SEQS)
print "Done."
piRNA cluster calling
Code A.7 | R function to call piRNA clusters and identify concordant cluster across
replicates
−−−
piRNAClusterAnalsys←function (bam, chrLenFile, mart, repeatGR, peakDepth = NULL,
annotGR = NULL, drawPlots = TRUE, bidirGapWidth = 20000,
transcriptGR = NULL, extendedStats = FALSE, windowSize = 1000,
ignoreReadStrand = FALSE)
{
chrLens←readChrLens(chrLenFile)
if (inherits(bam, "character")) {
if (is.null(annotGR))
annotGR←getEnsemblAnnot(mart, chrLens = chrLens)
gr←cleanReadGR(bam, annotGR, repeatGR)
}
else {
gr←bam
}
statmat←NA
Analysis scripts 199
if (is.null(peakDepth)) {
depths←seq(5, 300, by = 5)
statmat←.optimisePeakDepth(gr, depths, chrLens, bidirGapWidth,
transcriptGR, extendedStats = extendedStats, windowSize = windowSize)
if (drawPlots) {
message("Writing results to PDF...")
rc←try(.drawPeakDepthPlot(bam, depths, statmat))
if (inherits(rc, "try−error"))
warning("Unable to create PDF plot output.")
}
peakDepth←attr(statmat, "optimized")
message("Will use peakDepth of ", peakDepth, " for final analysis.")
}
clusterGR←.makeWindowedGR(gr, peakDepth = peakDepth, chrLens = chrLens,
bidirGapWidth = bidirGapWidth, windowSize = windowSize)
message("Trimming bam reads to the clusters...")
gr←subsetByOverlaps(gr, clusterGR, ignore.strand = ignoreReadStrand)
overlapMat←findOverlaps(clusterGR, gr, ignore.strand = ignoreReadStrand)
clusterGR←clusterGR[unique(queryHits(overlapMat))]
values(clusterGR)[, "count"]←sapply(split(as.numeric(values(gr)[,
"count"])[subjectHits(overlapMat)], queryHits(overlapMat)),
sum)
clusterGR←.annotateCoords(clusterGR)
clusterGR←clusterGR[order(values(clusterGR)$count, decreasing = TRUE)]
return(list(readsGR = gr, clusterGR = clusterGR, peakDepthOpt = statmat))
}
<bytecode: 0x52add7f8>
<environment: namespace:OdomLabPiRNA>
combineCluster←function (x, h3k4me3 = NULL, mybl1 = NULL, idtag = "", val.column = "count")
{
if (!all(sapply(x, function(y) "concordant" %in% colnames(values(y)))))
x←annotateConcordance(x)
allranges←Reduce(c, x)
gr.out←reduce(allranges)
gr.out←annotateBidirectionality(gr.out)
ol←findOverlaps(gr.out, allranges)
counts←aggregate(values(allranges[subjectHits(ol)])[,
val.column], list(queryHits(ol)), mean, na.rm = TRUE)
values(gr.out)[, val.column]←vector(length = length(gr.out),
mode = "numeric")
values(gr.out)[counts[, 1], val.column]←counts$x
gr.out←.annotateCoords(gr.out)
if (!is.null(h3k4me3))
gr.out$h3k4me3←overlapsAny(gr.out, h3k4me3)
if (!is.null(mybl1))
gr.out$mybl1←overlapsAny(gr.out, mybl1)
gr.out$ID←paste(idtag, 1:length(gr.out), sep = "_")
concordancy←aggregate(allranges[subjectHits(ol)]$concordant,
list(queryHits(ol)), any, na.rm = TRUE)
gr.out$concordant←NA
gr.out[counts[, 1]]$concordant←concordancy$x
return(gr.out)
}
<environment: namespace:OdomLabPiRNA>
200 Analysis scripts
Single-cell RNA-Seq analysis
Quality control
Code A.8 | Quality control and normalisation of single-cell RNA-Seq data
−−−
title: "Quality assessment of TC1/TC0 single spermatogonia, spermatocyes and spermatids"
author: "Nils Eling"
date: "29/05/2017"
output: html_document
−−−
edited: "Christina Ernst"
date: "11/09/2017"
## Load libraries
library(RColorBrewer)
suppressMessages(library(scran))
## Collecting meta information on single cells
input.SST3_7←read.table("data/raw_reads/SST4−7_all.txt", sep="\t", stringsAsFactors = FALSE)
input.SST7←read.table("data/raw_reads/SST7−2_all.txt", sep="\t", stringsAsFactors = FALSE)
input.SST8←read.table("data/raw_reads/SST8_all.txt", sep="\t", stringsAsFactors = FALSE)
input.SST5_3←read.table("data/raw_reads/SST5−3_all.txt", sep="\t", stringsAsFactors = FALSE)
input.SST3_7←cbind(input.SST3_7, input.SST7, input.SST8, input.SST5_3)
Decoding.SST3.1←read.table("data/decoding_files/SST3−1.txt", sep="\t")
Decoding.SST3.2←read.table("data/decoding_files/SST3−2.txt", sep="\t")
Decoding.SST4.1←read.table("data/decoding_files/SST4−1.txt", sep="\t")
Decoding.SST4.2←read.table("data/decoding_files/SST4−2.txt", sep="\t")
Decoding.SST5.1←read.table("data/decoding_files/SST5−1.txt", sep="\t")
Decoding.SST5.3←read.table("data/decoding_files/SST5−2.txt", sep="\t")
Decoding.SST6.1←read.table("data/decoding_files/SST6−1.txt", sep="\t")
Decoding.SST6.2←read.table("data/decoding_files/SST6−2.txt", sep="\t")
Decoding.SST7.1←read.table("data/decoding_files/SST7−1.txt", sep="\t")
Decoding.SST7.2←read.table("data/decoding_files/SST7−2.txt", sep="\t")
Decoding.SST8.1←read.table("data/decoding_files/SST8−1.txt", sep="\t")
Decoding.SST8.2←read.table("data/decoding_files/SST8−2.txt", sep="\t")
## The libraries are odered based on their position on the plate. Therefore, we order the decoding information based on the
second column (Position of plate).
ordering←order(Decoding.SST3.1[,2])
## Large dataframe keeps track of all features of each cell. Information is stored in rows, names of cells are composed of
the batch ID, cell type and position on the chip.
Master.QC←as.data.frame(matrix(data=NA,
nrow=ncol(input.SST3_7),
ncol=3))
colnames(Master.QC)←c("PositionOnChip", "PositionOnPlate", "Microscopy")
rownames(Master.QC)←c(paste("SST3_TC1_Spermatocyte", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST3_TC1_Stem_cell", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST4_TC1_Spermatocyte", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST4_TC1_Stem_cell", Decoding.SST3.1[ordering,1], sep="_"),
Analysis scripts 201
paste("SST5_TC1_Spermatocyte", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST6_TC0_Spermatocyte", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST6_TC0_Stem_cell", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST7_TC0_Spermatocyte", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST7_TC0_Stem_cell", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST8_TC1_Spermatids", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST8_TC0_Spermatids", Decoding.SST3.1[ordering,1], sep="_"),
paste("SST5_TC1_Spermatids", Decoding.SST3.1[ordering,1], sep="_"))
## First three columns contain info on: position on chip, position on plate and the visual observation.
Master.QC[,1:3]←rbind(Decoding.SST3.1[ordering,],
Decoding.SST3.2[ordering,],
Decoding.SST4.1[ordering,],
Decoding.SST4.2[ordering,],
Decoding.SST5.1[ordering,],
Decoding.SST6.1[ordering,],
Decoding.SST6.2[ordering,],
Decoding.SST7.1[ordering,],
Decoding.SST7.2[ordering,],
Decoding.SST8.1[ordering,],
Decoding.SST8.2[ordering,],
Decoding.SST5.3[ordering,])
## Add other features of the cells for meta−analysis.
Master.QC$Genomic←NA
Master.QC$ERCC←NA
Master.QC$NoFeature←NA
Master.QC$Ambigous←NA
Master.QC$LowQual←NA
Master.QC$NotAligned←NA
Master.QC$NotUnique←NA
Master.QC$TotalReads←NA
Master.QC$Mitochondrial←NA
Master.QC$MouseGenesDetected←NA
Master.QC$HumanGenesDetected←NA
Master.QC$Libraries←c(colnames(input.SST3_7))
## Read in mitochondrial genes
mit_genes←read.table("data/MT_genes.txt")
Master.QC[,4:14]←matrix(
data = as.numeric(c(colSums(input.SST3_7[grepl("ENS", rownames(input.SST3_7)),]),
colSums(input.SST3_7[grepl("ERCC", rownames(input.SST3_7)),]),
input.SST3_7[nrow(input.SST3_7)−4,],
input.SST3_7[nrow(input.SST3_7)−3,],
input.SST3_7[nrow(input.SST3_7)−2,],
input.SST3_7[nrow(input.SST3_7)−1,],
input.SST3_7[nrow(input.SST3_7),],
colSums(input.SST3_7),
colSums(input.SST3_7[as.character(mit_genes[,1]),]),
unlist(apply(input.SST3_7[grepl("ENSM", rownames(input.SST3_7)),], 2,
function(n){length(which(n > 0))})),
unlist(apply(input.SST3_7[grepl("ENSG", rownames(input.SST3_7)),], 2,
function(n){length(which(n > 0))})))),
ncol=11, nrow=ncol(input.SST3_7), byrow = FALSE
)
202 Analysis scripts
## Add batch info to df
chips←sapply(rownames(Master.QC),
function(n){paste(unlist(strsplit(n, "_"))[1:3], collapse = "_")})
Master.QC$chips←chips
Master.QC$col= brewer.pal(n = 12, name = "Set3")[match(Master.QC$chips, unique(Master.QC$chips))]
## Visualization of features and filtering
## Plot the percentage of reads mapped to the exonic part of the genome
# Colored by visual inspection
plot(Master.QC$Genomic/Master.QC$TotalReads,
pch=16, xlab="Cells", ylab="log10[Total mapped reads]",
col= ifelse(Master.QC$Microscopy == "−", "red",
ifelse(Master.QC$Microscopy == "2 cells", "yellow",
ifelse(Master.QC$Microscopy == "single", "black", "green"))), cex=0.7)
## Colored by batches
plot(Master.QC$Genomic/Master.QC$TotalReads,
pch=16, xlab="Cells", ylab="log10[Total mapped reads]",
col= Master.QC$col, cex=0.7)
## Remove duplets, empty wells and debris − select only single cells for further analysis
Singles←Master.QC[Master.QC$Microscopy == "single",]
## Spike−in reads − Plot percentage of reads mapped to ERCCs
plot(Singles$ERCC/Singles$TotalReads, col= Singles$col, ylab="%ERCC", pch=16)
plot(log10(Singles$ERCC + 1), col= Singles$col, ylab="Total number of ERCCs per well", pch=16)
## Two batches − SST4 Spermatogonia and SST6 Spermatogonia − have higher amounts of ERCCs. We will therefore not
use ERCCs for QC and normalisation.
## Exonic reads
Filter on the percentage of reads mapped to exonic regions.
plot(Singles$Genomic/Singles$TotalReads, col= Singles$col, ylab="%Genomic", pch=16)
Singles←Singles[Singles$Genomic/Singles$TotalReads > 0.15,]
## Mapping stats
## Removal of cells with unusual mapping and counting stats.
# Reads not mapped to exons
plot(Singles$NoFeature/Singles$TotalReads,
pch=16, xlab="Cells", ylab="log[No feature reads]",
col= Singles$col, cex=0.7)
# Multimapping reads
plot(Singles$Ambigous/Singles$TotalReads,
pch=16, xlab="Cells", ylab="log[Ambigous Reads]",
col= Singles$col, cex=0.7)
# Low quality reads
plot(Singles$LowQual/Singles$TotalReads,
pch=16, xlab="Cells", ylab="log[LowQual reads]",
col= Singles$col, cex=0.7)
plot(Singles$NotAligned/Singles$TotalReads,
pch=16, xlab="Cells", ylab="log[Not aligned reads]",
col= Singles$col, cex=0.7)
Analysis scripts 203
## No major outliers in these statistics, we therefore only remove cells in which more than 3\% of reads could not be aligned
.
Singles←Singles[Singles$NotAligned/Singles$TotalReads < 0.03,]
plot(Singles$MouseGenesDetected,
pch=16, col=Singles$col, xlab="Cells", ylab="# Mouse Genes Detected")
Singles←Singles[which(Singles$MouseGenesDetected > 3000),]
QC_pass←input.SST3_7[,Singles$Libraries]
QC_pass←QC_pass[grepl("E", rownames(input.SST3_7)),]
colnames(QC_pass)←rownames(Singles)
## SST3 batch from pilot experiment which was performed on smaller IFCs is a consistent outlier in all these plots and will
therefore be removed from downstream analysis
QC_pass←QC_pass[,!grepl("SST3", colnames(QC_pass))]
## Pre−normalize data and remove spermatocytes that cluster with spermatogonia − normalisation is done based on
mouse genes expressed.
Mouse←QC_pass[grepl("ENSM", rownames(QC_pass)),]
Mouse←Mouse[rowMeans(Mouse) > 1,]
# Generate SCE dataset
sce←newSCESet(countData=as.data.frame(Mouse))
# Keep cell types separate during normalization
clusters←factor(x = as.numeric(grepl("Stem", colnames(Mouse)))+1)
cell_types←sub(".*Stem.*", "2", colnames(Mouse), ignore.case=TRUE)
cell_types←sub(".*Spermatocyte.*", "1", cell_types, ignore.case=TRUE)
cell_types←sub(".*Spermatids.*", "3", cell_types, ignore.case=TRUE)
cluster←factor(cell_types)
# Calculate size factors
sce←computeSumFactors(sce, clusters = cluster)
sf←sizeFactors(sce)
# Normalize
norm.counts←t(t(Mouse)/sf)
## Calculate PCA of normalised data
pca←prcomp(log10(t(norm.counts) + 1))
plot(pca$x[,1], pca$x[,2],
col = alpha(ifelse(grepl("Stem", colnames(norm.counts)), "indianred1",
ifelse(grepl("Spermatids", colnames(norm.counts)), "navy", "darkorchid3")),0.75), pch=16)
## Remove two stem cell outliers
stem←colnames(norm.counts)[pca$x[,2] > 25]
stem←stem[grepl("Stem", stem)]
## Remove spermatocyte outliers
spermatocyte←colnames(norm.counts)[pca$x[,1] > 5 | pca$x[,2] > −10]
spermatocyte←spermatocyte[grepl("Spermatocyte", spermatocyte)]
## Remove spermatid outliers
spermatid←colnames(norm.counts)[pca$x[,1] > 0]
spermatid←spermatid[grepl("Spermatids", spermatid)]
QC_pass_new←QC_pass[,−which(colnames(QC_pass) %in% c(stem, spermatocyte, spermatid))]
204 Analysis scripts
## Repeat PCA after removing these cells
Mouse_new←QC_pass_new[grepl("ENSM", rownames(QC_pass_new)),]
Mouse_new←Mouse_new[rowMeans(Mouse_new) > 1,]
## Generate SCE dataset
sce_new←newSCESet(countData=as.data.frame(Mouse_new))
## Keep spermatocytes and spermatogonia seperately during normalization
cell_types←sub(".*Stem.*", "2", colnames(Mouse_new), ignore.case=TRUE)
cell_types←sub(".*Spermatocyte.*", "1", cell_types, ignore.case=TRUE)
cell_types←sub(".*Spermatids.*", "3", cell_types, ignore.case=TRUE)
cluster_new←factor(cell_types)
## Calculate size factors
sce_new←computeSumFactors(sce_new, clusters = cluster_new)
sf_new←sizeFactors(sce_new)
## Normalize
norm.counts_new←t(t(Mouse_new)/sf_new)
pca_new←prcomp(log10(t(norm.counts_new) + 1))
plot(pca_new$x[,1], pca_new$x[,2],
col = alpha(ifelse(grepl("Stem", colnames(norm.counts_new)), "indianred1",
ifelse(grepl("Spermatids", colnames(norm.counts_new)), "navy", "darkorchid3")), 0.75), pch=16)
## Check whether there is still protamine expression in cell types other than spermatids
plot(as.numeric(log10(QC_pass_new["ENSMUSG00000022501",] + 1)),
as.numeric(log10(QC_pass_new["ENSMUSG00000038015",] + 1)),
pch = 16, col = ifelse(grepl("Spermato", colnames(QC_pass_new)), "darkorchid3",
ifelse(grepl("Spermatid", colnames(QC_pass_new)), "navy", "indianred1")))
## Filtering on genes
Mouse←QC_pass_new[grepl("ENSM", rownames(QC_pass_new)),]
Human←QC_pass_new[grepl("ENSG", rownames(QC_pass_new)),]
## To filter mouse genes, we use an arbitrary threshold of an average count of 1
dim(Mouse)
Mouse←Mouse[rowMeans(Mouse) > 1,]
dim(Mouse)
## To filter human genes, we split cells into TC0 and TC1 partitions
TC1←Human[,grepl("TC1", colnames(Human))]
TC0←Human[,grepl("TC0", colnames(Human))]
## Remove genes that show cross mapping between human and mouse in TC0 mice
plot(log10(rowSums(TC1) + 1), log10(rowSums(TC0) + 1),
pch = 16, xlab = "Chr21 expr TC1", ylab = "Chr21 expr TC0")
Human←Human[−which(log10(rowSums(TC0) + 1) > 1 | rowSums(TC1) == 0),]
## Merge Datasets again
QC_pass_final←rbind(Human, Mouse)
## Write file
write.table(as.data.frame(QC_pass_final), "data/raw_reads/All_pass_ce.txt", quote = FALSE, sep = "\t")
norm.counts_new←t(t(QC_pass_final)/sf_new)
write.table(as.data.frame(norm.counts_new), "data/norm_counts/All_norm_counts_ce.txt", quote = FALSE, sep = "\t")
Analysis scripts 205
PCA and pseudotime for individual cell types
Code A.9 | Example workflow for spermatid cell population
−−−
title: "Spermatids"
author: "Nils Eling"
date: "14/07/2017"
output: html_document
−−−
edited: "Christina Ernst"
date: "14/09/2017"
## Load libraries and data
library(RColorBrewer)
suppressMessages(library(scran))
library(Rtsne)
library(plot3D)
library(destiny)
library(pheatmap)
library(ggplot2)
library(readxl)
## Load the normalized data and focus on spermatids
norm.data←read.table("data/norm_counts/All_norm_counts_ce.txt", sep = ’\t’)
sperm←norm.data[!grepl("ERCC", rownames(norm.data)),grepl("Spermatids", colnames(norm.data))]
Mouse←sperm[grepl("ENSM", rownames(sperm)),]
Mouse←Mouse[rowMeans(Mouse) > 0.1,]
sperm←rbind(Mouse, sperm[grepl("ENSG", rownames(sperm)),])
chips←sapply(colnames(sperm), function(n){unlist(strsplit(n, split = "_"))[1]})
# Genenames
chr21←read.table("data/chr21.txt", sep = ’,’, header = TRUE, stringsAsFactors = FALSE)
rownames(chr21)←chr21$Gene.stable.ID
mouse.genes←read.table("data/Genenames.txt", sep = ’\t’, header = TRUE, stringsAsFactors = FALSE)
rownames(mouse.genes)←mouse.genes[,1]
## Dimensionality reduction − calculate PCA and visualise
pca←prcomp(log10(t(sperm + 1)))
scatter2D(pca$x[,1], pca$x[,2],
pch = 16, col = "navy",
xlab = "PC1", ylab = "PC2")
# Color by library size
scatter2D(pca$x[,1], pca$x[,2],
pch = 16, colvar = log10(colSums(sperm)),
xlab = "PC1", ylab = "PC2")
# Color by chr21 content
scatter2D(pca$x[,1], pca$x[,2],
pch = 16, colvar = log10(colSums(sperm[grepl("ENSG", rownames(sperm)),] + 1)),
xlab = "PC1", ylab = "PC2")
206 Analysis scripts
genes←rownames(Mouse)
# Protamine 1
match("ENSMUSG00000022501", genes)
scatter2D(pca$x[,1], pca$x[,2], pch = 16, colvar = as.numeric(log10(sperm[genes[2967],] + 1)), clim=c(0,4), col =
colorRampPalette(c("white", "royalblue3", "navy"))(100))
# Protamine 2
match("ENSMUSG00000038015", genes)
scatter2D(pca$x[,1], pca$x[,2], pch = 16, colvar = as.numeric(log10(sperm[genes[8591],] + 1)), clim=c(0,4), col =
colorRampPalette(c("white", "royalblue3", "navy"))(100))
## Diffusionmap − calculate based on all mouse genes
genes←sperm[grepl("ENSM", rownames(sperm)),]
genes←genes[rowMeans(genes) > 1,]
dm←DiffusionMap(log10(t(genes) + 1), k = 20)
genes_dm←rownames(genes)
# Color based on chr21 expression
scatter2D(dm$DC1, dm$DC2, pch = 16, colvar = log10(colSums(sperm[grepl("ENSG", rownames(sperm)),]) + 1), clab = "
log(chr21/allGenes)")
# Color based on RNA content
scatter2D(dm$DC1, dm$DC2, pch = 16, colvar = log10(colSums(genes)), clab = "log10(RNA content)")
# Protamine 1
match("ENSMUSG00000022501", genes_dm)
scatter2D(dm$DC1, dm$DC2, pch = 16, colvar = as.numeric(log10(sperm[genes_dm[2281],] + 1)), clim=c(0,4), col =
colorRampPalette(c("white", "royalblue3", "navy"))(100))
## Find genes that correalte with diffusion pseudotime.
pt←DPT(dm, tips = which.min(dm$DC1), branching = FALSE)
dpt_flat←branch_divide(pt, integer(0L))
root←min(dpt_flat@branch[, 1], na.rm = TRUE)
dpt.pt←destiny:::dpt_for_branch(dpt_flat, root)
plot.DPT(pt, pch = 16)
## Find genes underlying the PCA separation
rotation_up←head(sort(pca$rotation[,1]), n=50)
rotation_down←tail(sort(pca$rotation[,1]), n=50)
## Define heatmap color gradient
breaks←seq(0,5, by=0.1)
color.palette←colorRampPalette(c("white", "royalblue3", "navy"))(length(breaks) − 1)
## Plot heatmap for genes correalting with early spermatids
pheatmap(log10(genes[names(rotation_down)[1:25],order(dpt.pt, decreasing = FALSE)] + 1), cluster_rows = FALSE, cluster
_cols = FALSE, color = color.palette, breaks = breaks, labels_row = mouse.genes[names(rotation_down)[1:25],2],
cellheight = 7, fontsize = 7, annotation_col = data.frame(row.names = colnames(genes)))
## Plot heatmap for genes correalting with late spermatids
pheatmap(log10(genes[names(rotation_up)[1:25],order(dpt.pt, decreasing = FALSE)] + 1), cluster_rows = FALSE, cluster_
cols = FALSE, color = color.palette, breaks = breaks, labels_row = mouse.genes[names(rotation_up)[1:25],2],
cellheight = 7, fontsize = 7, annotation_col = data.frame(row.names = colnames(genes)))
Analysis scripts 207
Differential gene expression and visualisation
Code A.10 | Differential gene expression between cell types and visualization across
pseudotime
−−−
title: "Marker_genes"
author: "Christina Ernst"
date: "18/09/2017"
output: html_document
−−−
## Load libraries and data
library(RColorBrewer)
suppressMessages(library(scran))
library(Rtsne)
library(plot3D)
library(destiny)
library(pheatmap)
library(ggplot2)
## Load the normalized data
norm.data←read.table("data/norm_counts/All_norm_counts_ce.txt", sep = ’\t’)
## Remove ERCCs for analyis
norm.data←norm.data[!grepl("ERCC−", rownames(norm.data)),]
chips←sapply(colnames(norm.data), function(n){unlist(strsplit(n, split = "_"))[1]})
# Genenames
chr21←read.table("data/chr21.txt", sep = ’\t’, header = TRUE, stringsAsFactors = FALSE)
rownames(chr21)←chr21$Gene.stable.ID
mouse.genes←read.table("data/Genenames.txt", sep = ’\t’, header = TRUE, stringsAsFactors = FALSE)
rownames(mouse.genes)←mouse.genes[,1]
## Establish ordering throuhgout spermatogenesis based on pseudotime
## Define groups based on PCA
pca←prcomp(log10(t(norm.data[,!grepl("Spermatids", colnames(norm.data))]) + 1))
plot(pca$x[,1], pca$x[,2], pch = 16,
col = ifelse(grepl("Sperm", colnames(norm.data[,!grepl("Spermatids", colnames(norm.data))])), "blue", "red"))
groups←ifelse(pca$x[,1] < 10, "Spermatocytes", ifelse(pca$x[,1] > 10 & pca$x[,2] > 0, "Spermatogonia 1", "Spermatogonia
2"))
groups←c(groups, rep("Spermatids", length(which(grepl("Spermati", colnames(norm.data))))))
names(groups)←colnames(norm.data)
## Calculate pseudotime for spermatocytes
genes←norm.data[,groups == "Spermatocytes"]
genes←genes[rowMeans(genes) > 1,]
dm←DiffusionMap(log10(t(genes) + 1), k = 20)
plot(dm$DC1, dm$DC2, pch=16)
pt←DPT(dm, tips = which.max(dm$DC2), branching = FALSE)
dpt_flat←branch_divide(pt, integer(0L))
root←min(dpt_flat@branch[, 1], na.rm = TRUE)
spermatocytes.pt←destiny:::dpt_for_branch(dpt_flat, root)
names(spermatocytes.pt)←colnames(genes)
208 Analysis scripts
## Calculate pseudotime for spermatogonia
genes←norm.data[,groups == "Spermatogonia 1"]
genes←genes[rowMeans(genes) > 1,]
dm←DiffusionMap(log10(t(genes) + 1), k = 20)
plot(dm$DC1, dm$DC2, pch=16)
pt←DPT(dm, tips = which.min(dm$DC1), branching = FALSE)
dpt_flat←branch_divide(pt, integer(0L))
root←min(dpt_flat@branch[, 1], na.rm = TRUE)
spermatogonia.pt←destiny:::dpt_for_branch(dpt_flat, root)
names(spermatogonia.pt)←colnames(genes)
## Calculate pseudotime for spermatids
genes←norm.data[,groups == "Spermatids"]
genes←genes[rowMeans(genes) > 1,]
dm←DiffusionMap(log10(t(genes) + 1), k = 20)
plot(dm$DC1, dm$DC2, pch=16)
pt←DPT(dm, tips = which.min(dm$DC1), branching = FALSE)
dpt_flat←branch_divide(pt, integer(0L))
root←min(dpt_flat@branch[, 1], na.rm = TRUE)
spermatids.pt←destiny:::dpt_for_branch(dpt_flat, root)
names(spermatids.pt)←colnames(genes)
## Differential gene expression between spermtogonia and spermatocytes
stem_spermatocyte←norm.data[,groups == "Spermatogonia 1" | groups == "Spermatocytes"]
stem_spermatocyte_markers←findMarkers(as.matrix(log2(stem_spermatocyte + 1)),
ifelse(grepl("Stem", colnames(stem_spermatocyte)),
"Spermatogonia 1","Spermatocytes"))
stem_spermatocyte_markers$‘Spermatocytes‘$HumanGenes←chr21[stem_spermatocyte_markers$‘Spermatocytes‘$
Gene,2]
stem_spermatocyte_markers$‘Spermatocytes‘$MouseGenes←mouse.genes[stem_spermatocyte_markers$‘
Spermatocytes‘$Gene,2]
head(stem_spermatocyte_markers$‘Spermatocytes‘[stem_spermatocyte_markers$‘Spermatocytes‘$‘Spermatogonia 1‘ >
0,], n = 50)
head(stem_spermatocyte_markers$‘Spermatocytes‘[stem_spermatocyte_markers$‘Spermatocytes‘$‘Spermatogonia 1‘ <
0,], n = 50)
## Differential gene expression between spermatocytes and spermatids
spermatocyte_spermatid←norm.data[,groups == "Spermatocytes" | groups == "Spermatids"]
spermatocyte_spermatid_markers←findMarkers(as.matrix(log2(spermatocyte_spermatid + 1)),
ifelse(grepl("Spermatocyte", colnames(spermatocyte_spermatid)),
"Spermatocytes","Spermatids"))
spermatocyte_spermatid_markers$‘Spermatocytes‘$HumanGenes←chr21[spermatocyte_spermatid_markers$‘
Spermatocytes‘$Gene,2]
spermatocyte_spermatid_markers$‘Spermatocytes‘$MouseGenes←mouse.genes[spermatocyte_spermatid_markers$‘
Spermatocytes‘$Gene,2]
head(spermatocyte_spermatid_markers$‘Spermatocytes‘[spermatocyte_spermatid_markers$‘Spermatocytes‘$‘Spermatids‘
> 0,], n = 50)
head(spermatocyte_spermatid_markers$‘Spermatocytes‘[spermatocyte_spermatid_markers$‘Spermatocytes‘$‘Spermatids‘
< 0,], n = 50)
Analysis scripts 209
## Define protein−coding genes to be plotted in heatmap
spermatocyte.vs.stem.up←c("Adam2", "Ldhc", "Fbp1", "Zmynd10", "Kctd19", "Pgam2", "Fabp9", "Ldhal6b", "Tmbim7", "
Morc2b", "Sept12", "Als2cr12", "Adam32", "Abcc12", "Ttc30a2")
stem.vs.spermatocyte.up←c("Taf7l", "Eif1ax", "Nxf2", "Mif", "Bcap31", "Tex11", "Hmgb3", "Hprt", "Ppp1ca", "Prdx4", "
Crabp1", "Taf9b", "Tsc22d3", "Rhox13", "Ftl1")
spermatocyte.vs.spermatid.up←c("Ube2t", "Lsm3", "Mlh3", "Sap30", "Fam221a", "Dut", "Dcaf13", "Pcbp1", "Ankrd10", "
Calm2", "Bcl2l12", "Eif2b1", "Tomm34", "Tdrd5", "Mllt10")
spermatid.vs.spermatocyte.up←c("Tex33", "Acsl1", "Tex36", "Slc22a14", "Nipsnap3a", "Ttll7", "Tmc7", "Ttc23l", "Spaca1", "
Erich3", "Spag6l", "Sqrdl", "Nt5c1b", "Catsper4", "Tssk4")
expr←norm.data[mouse.genes[match(c(stem.vs.spermatocyte.up, spermatocyte.vs.stem.up, spermatocyte.vs.spermatid.
up, spermatid.vs.spermatocyte.up), mouse.genes[,2]),1],]
expr[is.na(expr)]←0
rownames(expr)←c(stem.vs.spermatocyte.up, spermatocyte.vs.stem.up, spermatocyte.vs.spermatid.up, spermatid.vs.
spermatocyte.up)
rownames(expr)
## Define different color gradients
breaks←seq(0,5, by=0.05)
color.palette←colorRampPalette(c("gray100", "royalblue1", "navy"))(length(breaks) − 1)
color.palette←colorRampPalette(c("white", "darkorchid2", "darkorchid4"))(length(breaks) − 1)
color.palette←colorRampPalette(c("white", "indianred2", "indianred4"))(length(breaks) − 1)
pheatmap(log10(norm.data[mouse.genes[match(c(stem.vs.spermatocyte.up, spermatocyte.vs.stem.up, spermatocyte.vs.
spermatid.up, spermatid.vs.spermatocyte.up), mouse.genes[,2]),1],
c(names(spermatogonia.pt)[order(spermatogonia.pt, decreasing = TRUE)],
names(spermatocytes.pt)[order(spermatocytes.pt, decreasing = FALSE)],
names(spermatids.pt)[order(spermatids.pt, decreasing = FALSE)])] + 1),
cluster_rows = FALSE,
cluster_cols = FALSE, show_colnames = FALSE, color = color.palette, breaks = breaks,
labels_row = rownames(expr),
annotation_col = data.frame(
row.names = c(names(spermatogonia.pt)[order(spermatogonia.pt, decreasing = TRUE)],
names(spermatocytes.pt)[order(spermatocytes.pt, decreasing = FALSE)],
names(spermatids.pt)[order(spermatids.pt, decreasing = FALSE)]),
cell_type = c(rep("Spermatogonia", length(spermatogonia.pt)),
rep("Spermatocytes", length(spermatocytes.pt)),
rep("Spermatids", length(spermatids.pt))),
pseudotime = c(spermatogonia.pt[order(spermatogonia.pt, decreasing = TRUE)],
spermatocytes.pt[order(spermatocytes.pt, decreasing = FALSE)],
spermatids.pt[order(spermatids.pt, decreasing = FALSE)])))
210 Analysis scripts
Chromosome silencing
Code A.11 | Workflow to dissect chromosome silencing
−−−
title: "Chromosome silencing"
author: "Nils Eling"
date: "6 June 2017"
output: html_document
−−−
edited: "Christina Ernst"
date: "21/09/2017"
## Load libraries and data
library(RColorBrewer)
suppressMessages(library(scran))
library(Rtsne)
library(plot3D)
library(destiny)
library(pheatmap)
library(ggplot2)
library(GenomicRanges)
library(Gviz)
## Load normalised data
norm.data←read.table("data/norm_counts/All_norm_counts_ce.txt", sep = ’\t’)
# Remove ERCCs for analyis
norm.data←norm.data[!grepl("ERCC−", rownames(norm.data)),]
chips←sapply(colnames(norm.data), function(n){unlist(strsplit(n, split = "_"))[1]})
# Genenames
# Chr21
chr21←read.csv("data/chr21.txt", stringsAsFactors = FALSE)
rownames(chr21)←chr21$Gene.stable.ID
chr21←chr21[rownames(norm.data)[grepl("ENSG", rownames(norm.data))],]
# All genes
mouse.genes←read.table("data/Genenames.txt", sep = ’\t’, header = TRUE, stringsAsFactors = FALSE)
rownames(mouse.genes)←mouse.genes[,1]
# ChrX
chrX←read.csv("data/chrX.txt", stringsAsFactors = FALSE)
chrX←chrX[match(unique(chrX$Gene.stable.ID), chrX$Gene.stable.ID),]
rownames(chrX)←chrX$Gene.stable.ID
chrX←chrX[!is.na(match(chrX$Gene.stable.ID, rownames(norm.data))),]
## Selection of groups of interest
pca←prcomp(log10(t(norm.data[,!grepl("Spermatids", colnames(norm.data))]) + 1))
plot(pca$x[,1], pca$x[,2], pch = 16,
col = ifelse(grepl("Sperm", colnames(norm.data[,!grepl("Spermatids", colnames(norm.data))])), "blue", "red"))
groups←ifelse(pca$x[,1] < 10, "Spermatocytes", ifelse(pca$x[,1] > 10 & pca$x[,2] > 0, "Spermatogonia 1", "Spermatogonia
2"))
groups←c(groups, rep("Spermatids", length(which(grepl("Spermati", colnames(norm.data))))))
Analysis scripts 211
names(groups)←colnames(norm.data)
## Order based on X expression
Xchr←norm.data[as.character(chrX$Gene.stable.ID),groups != "Spermatogonia 2"]
cur_groups←groups[colnames(Xchr)]
glob.Xchr←colSums(Xchr)/colSums(norm.data[,groups != "Spermatogonia 2"])
spermatocytes.orderX←glob.Xchr[cur_groups == "Spermatocytes"]
spermatogonia1.orderX←glob.Xchr[cur_groups == "Spermatogonia 1"]
spermatids.orderX←glob.Xchr[cur_groups == "Spermatids"]
## Chromosome X inactivation
Xchr←norm.data[as.character(chrX$Gene.stable.ID),groups != "Spermatogonia 2"]
cur_groups←groups[colnames(Xchr)]
glob.Xchr←colSums(Xchr)/colSums(norm.data[,groups != "Spermatogonia 2"])
# Order cells in a way that 1st: Spermatogonia decrease, Spermatocytes increase, Spermatids increase
glob.Xchr.1←glob.Xchr[!grepl("Spermatids", names(glob.Xchr))]
glob.Xchr.2←glob.Xchr[grepl("Spermatids", names(glob.Xchr))]
glob.Xchr←c(glob.Xchr.1[order(glob.Xchr.1, decreasing = TRUE)],
glob.Xchr.2[order(glob.Xchr.2, decreasing = FALSE)])
plot(log10(glob.Xchr + 1), pch = 16, xlab = "Cells ordered by chrX expression",
ylab = "Relative reads on chrX",
col = c("navy", "darkorchid3", "indianred1")[as.factor(cur_groups[names(glob.Xchr)])])
## Number of X−linked genes expressed
plot(apply(Xchr[names(glob.Xchr)], 2, function(n){length(which(n > 0))}),
pch = 16, xlab = "Cells ordered by chrX expression",
ylab = "Genes expressed on chrX",
col = c("navy", "darkorchid3", "indianred1")[as.factor(cur_groups[names(glob.Xchr)])])
## Downregulation of X chromosome
Xchr←norm.data[chrX$Gene.stable.ID,groups == "Spermatogonia 1" | groups == "Spermatocytes"]
markers←findMarkers(as.matrix(log2(Xchr + 1)), ifelse(grepl("Stem", colnames(Xchr)), "Spermatogonia 1", "Spermatocytes
"))
markers_up←markers$‘Spermatogonia 1‘[markers$‘Spermatogonia 1‘$‘Spermatocytes‘ < −0.5,]
downregulated←markers$Spermatocytes[markers$Spermatocytes$‘Spermatogonia 1‘ < −0.5,]
# MA plot
plot(log2(rowMeans(Xchr[markers$‘Spermatocytes‘$Gene,])),
markers$‘Spermatocytes‘$‘Spermatogonia 1‘,
xlab = "Mean expression", ylab = "log2FC",
col = alpha(ifelse(markers$‘Spermatocytes‘$‘Spermatogonia 1‘ < −0.5, "indianred2",
ifelse(markers$‘Spermatocytes‘$‘Spermatogonia 1‘ > 0.5, "darkorchid3", "black")), 0.75), pch=16)
# Reactivation of X chromosome
Xchr←norm.data[chrX$Gene.stable.ID,groups == "Spermatocytes" | groups == "Spermatids"]
markers←findMarkers(as.matrix(log2(Xchr + 1)), ifelse(grepl("Spermato", colnames(Xchr)), "Spermatocytes", "Spermatids"
))
# MA plot
plot(log2(rowMeans(Xchr[markers$‘Spermatids‘$Gene,])),
markers$‘Spermatids‘$‘Spermatocytes‘,
xlab = "Mean expression", ylab = "log2FC",
212 Analysis scripts
col = alpha(ifelse(markers$‘Spermatids‘$‘Spermatocytes‘ < −0.5, "darkorchid3",
ifelse(markers$‘Spermatids‘$‘Spermatocytes‘ > 0.5, "navy", "black")), 0.75), pch=16)
## Focus on chromosome 21 silencing
norm.data.21←norm.data[,grepl("TC1", colnames(norm.data))]
groups.1←groups[colnames(norm.data.21)]
chr21.genes←norm.data.21[grepl("ENSG", rownames(norm.data.21)),groups.1 != "Spermatogonia 2"]
cur_groups←groups.1[colnames(chr21.genes)]
glob.21chr←colSums(chr21.genes)/colSums(norm.data.21[,groups.1 != "Spermatogonia 2"])
# Order cells in a way that 1st: Spermatogonia decrease, Spermatocytes increase, Spermatids increase
glob.21chr.1←glob.21chr[!grepl("Spermatids", names(glob.21chr))]
glob.21chr.2←glob.21chr[grepl("Spermatids", names(glob.21chr))]
glob.21chr←c(glob.21chr.1[order(glob.21chr.1, decreasing = TRUE)],
glob.21chr.2[order(glob.21chr.2, decreasing = FALSE)])
## Relative reads across chromosome 21
plot(log10(glob.21chr + 1), pch = 16, xlab = "Cells ordered by chr21 expression",
ylab = "Relative reads on chr21",
col = c("navy", "darkorchid3", "indianred1")[as.factor(cur_groups[names(glob.21chr)])])
## Gene expression across chromosome 21
plot(apply(chr21.genes[names(glob.21chr)], 2, function(n){length(which(n > 0))}),
pch = 16, xlab = "Cells ordered by chr21 expression",
ylab = "Genes expressed on chr21",
col = c("navy", "darkorchid3", "indianred1")[as.factor(cur_groups[names(glob.21chr)])])
## Downregulation of chromosome 21
chr21.genes←norm.data.21[grepl("ENSG", rownames(norm.data.21)),
groups.1 == "Spermatogonia 1" | groups.1 == "Spermatocytes" | groups.1 == "Spermatids"]
markers←findMarkers(as.matrix(log2(chr21.genes + 1)),
ifelse(grepl("Stem", colnames(chr21.genes)),
"Spermatogonia 1", "Spermatocytes"))
plot(log2(rowMeans(chr21.genes[markers$‘Spermatocytes‘$Gene,])),
markers$‘Spermatocytes‘$‘Spermatogonia 1‘,
xlab = "Mean expression", ylab = "log2FC",
col = alpha(ifelse(markers$‘Spermatocytes‘$‘Spermatogonia 1‘ < −0.5, "indianred2",
ifelse(markers$‘Spermatocytes‘$‘Spermatogonia 1‘ > 0.5, "darkorchid3", "black")), 0.75), pch=16)
## Localisation of genes along chromosome 21
genes←markers$‘Spermatocytes‘
rownames(genes)←genes$Gene
## Generate Genomic Range that has chromosome coordiates for gene downregulated in spermatocytes
downregulated←markers$Spermatocytes[markers$Spermatocytes$‘Spermatogonia 1‘ < −0.5,]
downregulated←downregulated[order(downregulated$‘Spermatogonia 1‘, decreasing = FALSE),]
downregulated_genes←downregulated$Gene
downregulated_genes←chr21[match(downregulated_genes, chr21$Gene.stable.ID),]
downregulated_genes[order(downregulated_genes$Gene.start..bp., decreasing = FALSE),]
chromosome←rep("chr21", 64)
downregulated_genes["chromosome"]←chromosome
Analysis scripts 213
downregulated_genes_coordinates←data.frame(downregulated_genes$chromosome, downregulated_genes$Gene.start..
bp., downregulated_genes$Gene.end..bp.)
cols←c("chr", "start", "end")
colnames(downregulated_genes_coordinates)←cols
downregulated_gr←GenomicRanges::makeGRangesFromDataFrame(downregulated_genes_coordinates)
## Generate Genomic Range that has chromosome coordiates for gene that are not differentially expressed between
spermatogonia and spermatocytes
no_change←markers$Spermatocytes[markers$Spermatocytes$‘Spermatogonia 1‘ > −0.5 & markers$Spermatocytes$‘
Spermatogonia 1‘ <0.5,]
no_change←no_change[order(no_change$‘Spermatogonia 1‘, decreasing = FALSE),]
no_change_genes←no_change$Gene
no_change_genes←chr21[match(no_change_genes, chr21$Gene.stable.ID),]
no_change_genes[order(no_change_genes$Gene.start..bp., decreasing = FALSE),]
chromosome←rep("chr21", 270)
no_change_genes["chromosome"]←chromosome
no_change_genes_coordinates←data.frame(no_change_genes$chromosome, no_change_genes$Gene.start..bp., no_
change_genes$Gene.end..bp.)
colnames(no_change_genes_coordinates)←cols
no_change_genes_coordinates←na.omit(no_change_genes_coordinates)
no_change_gr←GenomicRanges::makeGRangesFromDataFrame(no_change_genes_coordinates)
## Generate Genomic Range that has chromosome coordiates for gene that are upregulated in spermatocytes
escape←markers$Spermatocytes[markers$Spermatocytes$‘Spermatogonia 1‘ > 0.5,]
escape←escape[order(escape$‘Spermatogonia 1‘, decreasing = FALSE),]
escape_genes←escape$Gene
escape_genes←chr21[match(escape_genes, chr21$Gene.stable.ID),]
escape_genes[order(escape_genes$Gene.start..bp., decreasing = FALSE),]
chromosome←rep("chr21", 20)
escape_genes["chromosome"]←chromosome
escape_genes_coordinates←data.frame(escape_genes$chromosome, escape_genes$Gene.start..bp., escape_genes$
Gene.end..bp.)
colnames(escape_genes_coordinates)←cols
escape_gr←GenomicRanges::makeGRangesFromDataFrame(escape_genes_coordinates)
##Genearete visualisation tracks with Gviz package
downregulated_21←AnnotationTrack(downregulated_gr, name = "Downregulated on 21")
no_change_21←AnnotationTrack(no_change_gr, name = "No change on Chr21")
escape_21←AnnotationTrack(escape_gr, name = "Escape 21")
gtrack←GenomeAxisTrack()
itrack←IdeogramTrack(genome = "hg19", chromosome = "chr21")
plotTracks(list(itrack, gtrack, escape_21, no_change_21, downregulated_21))
## Reactivation of chromosome 21
chr21.genes←norm.data.21[grepl("ENSG", rownames(norm.data.21)),
groups.1 == "Spermatocytes" | groups.1 == "Spermatids"]
markers←findMarkers(as.matrix(log2(chr21.genes + 1)),
ifelse(grepl("Spermato", colnames(chr21.genes)),
"Spermatocytes", "Spermatids"))
214 Analysis scripts
plot(log2(rowMeans(chr21.genes[markers$‘Spermatids‘$Gene,])),
markers$‘Spermatids‘$‘Spermatocytes‘,
xlab = "Mean expression", ylab = "log2FC",
col = alpha(ifelse(markers$‘Spermatids‘$‘Spermatocytes‘ < −0.5, "darkorchid3",
ifelse(markers$‘Spermatids‘$‘Spermatocytes‘ > 0.5, "navy", "black")), 0.75), pch=16)
## Probe expression of some specific genes
# USP25 − ENSG00000155313; NRIP1 − ENSG00000180530l; TPTE − ENSG00000274391; SOD1 −
ENSG00000142168; CCT8 − ENSG00000156261; CXADR − ENSG00000154639; AL133499.1 −
ENSG00000267857; LCA5L − ENSG00000157578; RSPH1 − ENSG00000160188
boxplot(list(spermatogonia = as.numeric(log10(chr21.genes["ENSG00000142168", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000142168", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000156261", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000156261", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000154639", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000154639", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000155313", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000155313", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000180530", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000180530", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000274391", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000274391", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000267857", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000267857", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000157578", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000157578", groups.1 == "Spermatocytes"] + 1)),
spermatogonia = as.numeric(log10(chr21.genes["ENSG00000160188", groups.1 == "Spermatogonia 1"] + 1)),
spermatocytes = as.numeric(log10(chr21.genes["ENSG00000160188", groups.1 == "Spermatocytes"] + 1))),
col = c("indianred2", "darkorchid3", "indianred2", "darkorchid3", "indianred2", "darkorchid3", "indianred2", "
darkorchid3", "indianred2", "darkorchid3", "indianred2", "darkorchid3", "indianred2", "darkorchid3", "indianred2
", "darkorchid3", "indianred2", "darkorchid3"), outline = FALSE)
B | Mouse colony data
Mouse ID
Number
of litters
Number
of pups
Number of Tc1-
positive offspring
Transmission
rate
38929 3 12 9 75%
99697 2 15 10 67%
97139 2 3 2 67%
71676 5 30 19 63%
76494 2 17 10 59%
97886 3 9 5 56%
38928 5 26 14 54%
38930 4 31 16 52%
79649 3 25 13 52%
100048 3 10 5 50%
18093 3 6 3 50%
65223 1 4 2 50%
99696 1 4 2 50%
25419 1 2 1 50%
85098 5 37 17 46%
85099 6 38 17 45%
97887 7 38 17 45%
82041 3 11 5 45%
76495 4 18 8 44%
11676 2 7 3 43%
36838 5 19 8 42%
100050 5 15 6 40%
73885 1 3 1 33%
100047 8 62 19 31%
10732 7 45 18 31%
7088 9 51 16 31%
42257 4 16 5 31%
73887 4 10 3 30%
85831 3 7 2 29%
68166 2 4 1 25%
98981 4 19 4 21%
94095 5 23 13 17%
94099 6 41 7 17%
25422 8 50 8 16%
100049 4 18 1 6%
94094 1 2 0 0%
7087 9 66 0 0%
42196 3 30 0 0%
Table B.1 | Transmission rates and breeding statistics for Tc1 females
216 Mouse colony data
Mouse ID
Number
of litters
Number
of pups
Number of Tc1-
positive offspring
Transmission
rate
Comments
14/16872 2 10 6 60%
15/42199 4 12 5 42%
14/93006 9 39 16 41%
14/92043 3 9 3 33%
16/36047 1 6 2 33%
16/36047 1 7 2 29%
17/2830 5 43 11 26%
15/4695 8 33 8 24%
14/34436 3 17 4 24%
16/35150 4 20 4 20%
14/34496 10 77 12 16%
13/50197 6 27 4 15%
15/43687 4 27 4 15%
15/11532 6 30 4 13%
16/49057 3 17 2 12%
15/11372 8 54 6 11%
14/34497 3 27 3 11%
13/93659 12 102 9 9%
16/2647 10 83 7 8%
16/36046 6 49 4 8%
14/34437 12 86 5 6%
15/26233 9 86 5 6%
15/48772 3 21 1 5%
16/21529 4 21 1 4%
14/98984 17 117 2 2%
14/34495 12 84 1 1%
13/92051 5 49 0 0%
15/26218 1 1 0 0%
13/43686 1 2 0 0%
16/11470 4 30 0 0%
16/37479 1 2 0 0%
16/24392 4 40 n/a n/a Litters did not survive until weaning
13/92037 n/a n/a n/a n/a No offspring over 3 month period
13/93169 n/a n/a n/a n/a No offspring over 3 month period
16/9499 n/a n/a n/a n/a No offspring over 3 month period
16/9501 n/a n/a n/a n/a No offspring over 3 month period
16/34778 n/a n/a n/a n/a No offspring over 3 month period
16/51186 n/a n/a n/a n/a No offspring over 3 month period
14/36869 n/a n/a n/a n/a No offspring over 6 month period
14/36870 n/a n/a n/a n/a No offspring over 6 month period
15/15024 n/a n/a n/a n/a No offspring over 6 month period
15/26244 n/a n/a n/a n/a No offspring over 6 month period
16/14339 n/a n/a n/a n/a No offspring over 6 month period
16/15156 n/a n/a n/a n/a No offspring over 6 month period
Table B.2 | Transmission rates and breeding statistics for Tc1 males