Grass xylan structural variation suggests functional
specialization and distinctive interaction with cellulose
and lignin
Theodora Tryfona1 , Matthieu Bourdon2, Rita Delgado Marques1, Marta Busse-Wicher1, Francisco Vilaplana3 ,
Katherine Stott1 and Paul Dupree1,*
1Department of Biochemistry, School of Biological Sciences, University of Cambridge, Cambridge CB2 1QW, UK,
2Sainsbury Laboratory, University of Cambridge, Cambridge CB2 1LR, UK, and
3Department of Chemistry, School of Engineering Sciences in Chemistry, Biotechnology and Health, KTH Royal Institute of
Technology, Stockholm SE-10691, Sweden
Received 23 September 2022; revised 21 December 2022; accepted 23 December 2022.
*For correspondence (e-mail pd101@cam.ac.uk).
SUMMARY
Xylan is the most abundant non-cellulosic polysaccharide in grass cell walls, and it has important structural
roles. The name glucuronoarabinoxylan (GAX) is used to describe this variable hemicellulose. It has a linear
backbone of b-1,4-xylose (Xyl) residues that may be substituted with a-1,2-linked (4-O-methyl)-glucuronic
acid (GlcA), a-1,3-linked arabinofuranose (Araf), and sometimes acetylation at the O-2 and/or O-3 positions.
The role of these substitutions remains unclear, although there is increasing evidence that they affect the
way xylan interacts with other cell wall components, particularly cellulose and lignin. Here, we used
substitution-dependent endo-xylanase enzymes to investigate the variability of xylan substitution in grass
culm cell walls. We show that there are at least three different types of xylan: (i) an arabinoxylan with
evenly distributed Araf substitutions without GlcA (AXe); (ii) a glucuronoarabinoxylan with clustered GlcA
modifications (GAXc); and (iii) a highly substituted glucuronoarabinoxylan (hsGAX). Immunolocalization of
AXe and GAXc in Brachypodium distachyon culms revealed that these xylan types are not restricted to a
few cell types but are instead widely detected in Brachypodium cell walls. We hypothesize that there are
functionally specialized xylan types within the grass cell wall. The even substitutions of AXe may permit
folding and binding on the surface of cellulose fibrils, whereas the more complex substitutions of the other
xylans may support a role in the matrix and interaction with other cell wall components.
Keywords: Miscanthus sinensis, Brachypodium distachyon, grass xylan, polysaccharide substitution pat-
terns, grass cell wall, arabinosylation, glucuronidation, xylan–lignin interactions, xylan–cellulose interac-
tions, cell wall molecular architecture.
INTRODUCTION
Plant cell walls are intricate polysaccharide-based struc-
tures. There are two widespread and distinct developmen-
tal types of plant cell walls that differ in composition and
molecular cross-linking. Primary cell walls are highly
hydrated structures, having a plastic architecture enabling
the cells to expand and grow. On the other hand, sec-
ondary cell walls are less hydrated, form after the deposi-
tion of the primary cell wall, once cells cease to expand,
and are rigid structures that provide mechanical strength
(Cosgrove & Jarvis, 2012). It is therefore apparent that cell
wall components and their molecular interactions are criti-
cal in conferring the various properties that cell walls pos-
sess, but the detailed knowledge of the spatial
organization of each cell wall component ismissing. Gaining
such detailed knowledge would lead to a better understand-
ing of the underlying principles of cell wall architecture,
allowing the prediction of cell wall properties and function.
The primary and secondary cell walls of grasses are
distinct from eudicots in several ways. Most importantly,
xylan is the dominant hemicellulose in grass primary
walls, with mixed-linkage glucan being the second major
hemicellulose. Depending on the species, growth stage
and tissue, xylan may carry a number of different substitu-
tions. The molecule of xylan consists of a linear backbone
of b-1,4-xylose (Xyl) residues substituted with a-1,2-linked
(4-O-methyl)-glucuronic acid (GlcA), a-1,3-linked arabinofu-
ranose (Araf), and acetylation at the O-2 and/or O-3

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.
This is an open access article under the terms of the Creative Commons Attribution License,
which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
1
The Plant Journal (2023) doi: 10.1111/tpj.16096
positions (Ebringerov
a & Heinze, 2000). Grass xylan substi-
tution consists mainly of Araf, with comparatively fewer
GlcA substitutions (Ebringerov
a & Heinze, 2000; Tryfona
et al., 2019). However, grass xylans have some unique
structural features, such as Xylp–Araf substitutions
(Ebringerov
a & Heinze, 2000; Tryfona et al., 2019; Wende &
Fry, 1997; Zhong et al., 2022) or hydroxycinnamic acid
groups, for example, feruloyl (Fer) or coumaryl (Co)
groups, esterified to the O-5 position of Araf residues (Fei-
jao et al., 2021; Hatfield et al., 2017; Mueller-Harvey
et al., 1986). These hydroxycinnamic groups can be esteri-
fied or etherified to lignin (Hatfield et al., 2017). The ferula-
tion of grass xylan is known to facilitate covalent cross
links of xylan chains or xylan–lignin via diferulate bridge
formation (Feijao et al., 2021; Hatfield et al., 1999, 2017;
Terrett & Dupree, 2019).
The arrangement of substitutions on the backbone of
xylan influences how this molecule interacts with other cell
wall components. In eudicots, GlcA and acetyl group sub-
stitutions are positioned on alternate Xyl residues of the
xylan backbone at the O-2/O-3 positions (Bromley
et al., 2013; Busse-Wicher et al., 2014). It is this even pat-
tern of decoration that allows xylan to adopt a twofold heli-
cal screw conformation. Such conformation results in all
decorations being oriented on one side of the xylan chain,
which could then permit xylan molecules to hydrogen
bond with the hydrophilic surfaces of cellulose. Randomly
modified xylan, on the other hand, adopts a threefold con-
formation and therefore is more flexible and does not
interact with cellulose hydrophilic surfaces (Grantham
et al., 2017). The even spacing of xylan decoration is a
highly conserved feature and is also found in gymnosperm
xylans (Busse-Wicher, Li, et al., 2016; Mart
ınez-Abad
et al., 2017; Pereira et al., 2017; Terrett et al., 2019). There-
fore, it becomes apparent that in eudicot angiosperms and
gymnosperms the pattern of xylan substitution influences
the conformation of this polysaccharide and enables inter-
action with cellulose fibrils.
The Araf substitutions and hydroxycinnamic acid
modifications of grass xylan are likely to have an effect on
the conformation of xylan and its interaction with both cel-
lulose and lignin. Our recent solid-state nuclear magnetic
resonance (ssNMR) study on never-dried Brachypodium
distachyon tissues confirms that some xylan in grasses
binds to cellulose in the twofold conformation (Duan
et al., 2021). Furthermore, lignin–polysaccharide interac-
tions in secondary cell walls of Zea mays (maize) revealed
by ssNMR were also found to be xylan-conformation
dependent (Kang et al., 2019). It is clear, therefore, that dif-
ferent xylan conformations are present in grass cell walls.
However, the arrangement of substitutions on grass xylan
remains unknown, and so it is not yet possible to under-
stand fully how these arrangements might influence cell
wall assembly.
To investigate the arrangements of xylan decorations
in commelinid monocots, we analysed the pattern of xylan
substitutions on grass xylan. We demonstrate that grass
cell wall contains distinct types of xylan. These xylans have
different substitution patterns, can be partially separated
from each other, and are widely present on cell walls.
Based on these observations, we postulate that these
xylans may adapt different conformations and have differ-
ent interaction properties with other cell wall components.
RESULTS
Some Araf substitutions on grass xylan are evenly spaced
and GlcA decorations are clustered together
On Miscanthus culm xylan, the average frequency of Araf
substitutions is about 11%, whereas GlcA substitution is
significantly lower, at about 5% of Xyl residues (Tryfona
et al., 2019); however, it is not yet known whether Araf or
GlcA decoration is random or is regularly or irregularly
spaced. To investigate the spacing of these substitutions,
we hydrolysed xylan from Miscanthus culms with xyla-
nases from two different glycosyl hydrolase (GH) families,
GH5 (CtGH5_34) and GH30 (EcGH30) (Carbohydrate Active
EnZymes database, CAZy, http://www.cazy.org; Urb
anikov
a
et al., 2011). The GH5 enzyme is an arabinoxylan-specific
xylanase that displays an absolute requirement for xylans
that contain Araf side chains (Correia et al., 2011). This
enzyme requires the Araf appended to O-3 of the Xyl
bound in the 
1 subsite (Labourel et al., 2016). By sizing
the oligosaccharides released by the GH5 enzyme, the
spacing between Araf substitutions can be determined
(Figure 1a). On the other hand, the GH30 enzyme requires
GlcA substitutions to be attached at the 
2 Xyl residue
from the active site (Vrsansk
a et al., 2007), and the size of
the oligosaccharides released by this enzyme has been
previously used to determine the spacing between two
GlcA residues on the xylan backbone in Arabidopsis
and gymnosperms (Figure 1b; Bromley et al., 2013; Busse-
Wicher, Li, et al., 2016).
Alcohol-insoluble residue (AIR) from Miscanthus
culms was depectinated, saponified and subsequently
hydrolysed with the GH5 arabinoxylanase. The oligosac-
charide hydrolysis products were analysed using matrix-
assisted laser desorption/ionisation (MALDI) time-of-flight
(ToF) mass spectrometry (MS) and polysaccharide analysis
by carbohydrate gel electrophoresis (PACE). A range of
xylo-oligosaccharide products of a degree of polymerisa-
tion (DP) were released, as seen by PACE (Figure 1c,
lane 1). The migration differences suggested that the main
series of oligosaccharides differ in DP by two pentoses,
with the smallest oligosaccharide migrating slightly faster
than xylotriose (X3). Based on the specificity of the GH5
arabinoxylanase, all xylo-oligosaccharides should be sub-
stituted with an Araf at the reducing end. Consistent with

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
2 Theodora Tryfona et al.
this, all the oligosaccharides were sensitive to PaGH62 ara-
binofuranosidase (Figure 1c, lane 2). To investigate
whether further substitutions were present, we treated the
GH5 products with TrGH3 xylosidase, which removes ter-
minal xylosyl residues from the non-reducing end, but is
inhibited by substitutions at O-3 of Xyl on the +2 subsite
(Tenkanen et al., 1996). All the major GH5 oligosaccharide
products were sensitive to the xylosidase, yielding one
major oligosaccharide migrating slightly faster than X3
(Figure 1c, lane 3). The structure of this oligosaccharide
Figure 1. Analysis of the distribution of arabinose (Araf) and glucuronic acid (GlcA) on xylan from Miscanthus culms.
(a) Schematic representation of the GH5 arabinoxylanase action to determine Araf spacing on the xylan backbone. GH5 requires Araf substitutions to be a-1,3-
linked to the active site xylose (Xyl) residue and liberates Araf-modified oligosaccharides. A, Araf; NR, non-reducing end; Pent, pentose; RE, reducing end; U,
GlcA; X, Xyl.
(b) Schematic representation of the GH30 glucuronoxylanase to determine the spacing of GlcA. The GH30 enzyme requires GlcA substitutions to be attached at
the 
2 Xyl residue from the active site and liberates singly GlcA-substituted oligosaccharides.
(c) Saponified alcohol insoluble residue (AIR) from Miscanthus culms was hydrolysed with GH5 arabinoxylanase and analysed by polysaccharide analysis with
carbohydrate gel electrophoresis (PACE). GH5-hydrolysed samples were further hydrolysed either with PaGH62 arabinosidase or TrGH3 xylosidase. Xylan
oligosaccharides: X–X7.
(d) Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-ToF MS) spectra of 2-AA labelled GH5 hydrolysis products.
(e) Saponified AIR from Miscanthus culms was hydrolysed with GH30 glucuronoxylanase and analysed by PACE. GH30-hydrolysed samples were further hydrol-
ysed with BoGH115 glucuronidase, PaGH62 arabinosidase or TrGH3 xylosidase. Arrows with the same colour show the flow direction of bands after sequential
digestion.
(f) MALDI-ToF MS spectra of 2-AA labelled GH30 hydrolysis products.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 3
was determined by NMR spectroscopy and was identified
as b-Xylp-(1?4)-[a-Araf-(1?3)]-Xylp (XA3; Figure S1;
Table S1). Therefore, taking together the PACE and NMR
data, the majority of the GH5 hydrolysis products have just
a single Araf substitution at the reducing end. This is con-
sistent with the MALDI-ToF-MS spectra, which revealed
mainly neutral oligosaccharides (Figure 1d). Furthermore,
the dominance of oligosaccharides consisting of odd-
number pentose residues is consistent with the backbone
length of even numbers of xylosyl residues, plus a single
Araf on the reducing end xylose. Only minor acidic
oligosaccharide products were detected, modified by a sin-
gle methylated GlcA (MeGlcA) residue. The results indicate
that the xylan digested by the GH5 arabinoxylanase has
sparse and predominantly even-spaced Araf, and very few
MeGlcA residues.
Glucuronoxylanase GH30 hydrolysis of Miscanthus
culm AIR and analysis with PACE revealed several
oligosaccharides demonstrating migration between X4 and
X7, plus a few minor larger oligosaccharides; two major
(yellow and blue arrows; Figure 1e, lane 1) and two minor
oligosaccharides (grey and black arrows; Figure 1e, lane 1)
were studied further. All these oligosaccharides were sen-
sitive to BoGH115 a-glucuronidase, indicating the presence
of GlcA decoration, as expected (Figure 1e, lane 2). Three
of the GH115 a-glucuronidase products now co-migrated
with xylo-oligosaccharide standards: X7, X6 and X5. Given
the specificity of the GH30 glucuronoxylanase, this indi-
cates that the products were X5U
2X (yellow arrow,
XXXXXU2X), X4U
2X (grey arrow, XXXXU2X) and X3U
2X
(black arrow, XXXU2X), respectively. One oligosaccharide
did not migrate with any standard (Figure 1e, lane 2, blue
arrow), suggesting that it was further substituted. Indeed,
this GH30 product was sensitive to PaGH62 arabinofura-
nosidase (Figure 1e, lane 4). Hydrolysis of the GH30 prod-
ucts with GH3 b-xylosidase resulted in U2X from XnU
2X
oligosaccharides, as well as a resistant oligosaccharide
(migrating near X5), indicating a substitution (Figure 1e,
lane 3). The positions and linkages of the substitutions of
this oligosaccharide were determined by MALDI-ToF/ToF-
MS/MS as XA3XU2X (Figure S2). The GH30 products were
also investigated by MALDI-ToF-MS (Figure 1f), confirming
that MeGlcAPent5, MeGlcAPent6 and MeGlcAPent7 pre-
dominated. Together, the results indicate that GH30
cleaves the xylan into short oligosaccharides and suggests
most of the GlcA substitutions are clustered between five
and seven backbone residues apart.
These data revealed two quite different grass xylan
substitution patterns: (i) arabinoxylan with evenly dis-
tributed Araf substitutions (AXe); and (ii) glucuronoarabi-
noxylan with clustered GlcA modifications (GAXc). The
pattern of Miscanthus culm xylan substitution with GlcA
and Ara shown here is distinct from that found in gym-
nosperms and many angiosperms (Busse-Wicher, Li,
et al., 2016). To investigate whether grass substitution pat-
terns are conserved among grass species, we produced
AIR from culms of eight different grass species: Andro-
pogon gerardii, B. distachyon, Oryza sativa, Phragmites
australis, Phyllostachys viridiglaucescens, Saccharum spp.
(SP80-3280), Triticum aestivum and Z. mays. Digestion of
AIR with GH5 arabinoxylanase and analysis by PACE
showed that a number of oligosaccharides were released
from all grass species, which co-migrate with the typical
AXe oligosaccharides released from Miscanthus culms
(Figure 2a). Similarly, digestion of culm AIR with GH30 glu-
curonoxylanase and analysis by PACE showed that the
oligosaccharides released also co-migrate with the GAXc
oligosaccharides released from Miscanthus culms
(Figure 2b). The relative quantities of oligosaccharides
released by either of the two enzymes varied among the
different grass species analysed, which might reflect differ-
ences in development or genuine quantitative differences
in the substitution of xylan (Tryfona et al., 2019). Neverthe-
less, the even distribution of Araf substitutions and the
clustering of GlcA residues on xylan are common features
across all grasses analysed.
Extraction and separation of distinct domains of grass
xylan
We investigated whether AXe and GAXc are distinct
domains within a single type of xylan or, alternatively, rep-
resent distinct types of xylan chains. In addition, it
remained unclear what proportion of xylan was constituted
by these xylan types. We digested saponified Miscanthus
culm AIR with substitution-tolerant GH10 xylanase
(Figure S3), GH5 arabinoxylanase or GH30 glucuronoxy-
lanase. We then used 65% ethanol to separate, by precipita-
tion, any xylanase-resistant xylan from the released
oligosaccharides. The xylan in the pellet was then treated
further with each of the other xylanases (Figure 3a). This
revealed significant differences in the digestibility of xylan
by the three enzymes: GH10 endo-xylanase digested all the
AXe and GAXc, as the characteristic oligosaccharides were
not released by these enzymes (Figure 3b, lanes 1–3). Con-
sistent with the low GlcA substitution frequency of the AXe
xylan domain, the GH30 glucuronoxylanase did not digest
all the AXe (Figure 3b, lanes 7 and 9). On the other hand,
the GH5 arabinoxylanase digested almost all the GAXc (Fig-
ure 3b, lanes 4 and 6), consistent with the presence of Ara
substitutions in GAXc. Interestingly, a further fraction of
xylan was GH5 arabinoxylanase resistant but GH10 digesti-
ble (Figure 3b, lanes 4 and 5). Thus, most patterned Araf
and clustered GlcA modifications are indeed found on dis-
tinct domains of xylan, AXe and GAXc, respectively. The
proportions of xylan that were GH30 glucuronoxylanase
resistant but GH10 digestible (Figure 3b, lane 8) and GH5 ara-
binoxylanase resistant but GH10 digestible (Figure 3b, lane 5)
were significant, but did not represent all of the total xylan

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
4 Theodora Tryfona et al.
released by GH10 endoxylanase (Figure 3b, lane 1). Thus,
AXe and GAXc are both relatively substantial fractions of
wall xylan. Additionally, in culm cell walls there is also a
small population of xylan that is digestible only by GH10
xylanase, which we call highly substituted GAX (hsGAX).
Sequential cell wall extractions may provide informa-
tion on cell wall architecture by revealing how strongly gly-
cans are bound to cell walls. To investigate the interaction
of grass xylan with other cell wall components, polysac-
charides were sequentially chemically extracted from Mis-
canthus culm AIR. The solubilized polysaccharides were
digested with xylanases (GH10, GH5 and GH30) and anal-
ysed by PACE (Figure 4). As expected, only trace quantities
of xylan were released by 1,2-cyclohexanediamine tetra
acetic acid (CDTA) and Na2CO3 fractions, which solubilize
mainly pectic polysaccharides. The majority of all types of
xylan was solubilized by 1 M KOH, in accordance with pre-
vious studies (Kulkarni et al., 2012). A smaller but substan-
tial quantity of AXe (GH5-accessible) xylan was found in
the 4 M KOH fraction. Only trace quantities of GAXc (di-
gested by GH30 glucuronoxylanase) were released by 4 M
KOH. Treatment with 4 M KOH after sodium chlorite (PC
4 M KOH) is thought to solubilize polysaccharides that are
either directly or indirectly associated with lignin in the cell
wall matrix (da Costa et al., 2017). A small quantity of xylan
was released with 4 M KOH PC, as revealed by GH10
hydrolysis, but this was not AXe, and was probably mainly
hsGAX. We conclude that harsher conditions are required
for the extraction of all of the AXe compared with GAXc,
and therefore AXe may be more strongly bound in the cell
wall.
If GAXc and AXe are separate xylan molecule types,
then they would be differently charged and separable by
anion-exchange chromatography on diethylaminoethanol
(DEAE) cellulose. To investigate this question, xylan was
bound to DEAE cellulose and then eluted with increasing
concentrations of NaHCO3 (Figure 5). The presence of dif-
ferent types of xylan was determined by hydrolysis with
GH10 xylanase, GH5 arabinoxylanase or GH30 glu-
curonoxylanase and analysed by PACE. GH10 xylanase
hydrolysis showed xylan eluted in all fractions. Notably,
AXe and GAXc eluted with somewhat differing profiles:
the peak of the AXe elution was observed in the 10 mM
NaHCO3 fraction, with lower proportions eluted in the
Figure 2. Analysis of Araf and GlcA substitutions in grasses.
(a) Araf is regularly distributed on xylan in grasses. Saponified alcohol insoluble residue (AIR) of Andropogon gerardii, Brachypodium, Miscanthus, common
rice, Phragmites australis, Phyllostachys viridiglaucescens, sugar cane, common wheat and maize was hydrolysed with GH5 arabinoxylanase and analysed by
PACE. The major grass products correspond to substituted xylo-oligosaccharides with Araf side chains on evenly spaced Xyl residues, largely at 6-, 8-, 10-, 12-
or 14-residue intervals.
(b) GlcA substitutions are clustered together on xylan in grasses. Saponified AIR of Andropogon gerardii, Brachypodium, Miscanthus, common rice, Phragmites
australis, Phyllostachys viridiglaucescens, sugar cane, common wheat and maize was hydrolysed with GH30 glucuronoxylanase and analysed by PACE. The
major grass products correspond to substituted xylopentose XA3XU2X and xyloheptaose X5U
2X.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 5
Figure 3. Analysis of xylanase-accessible xylan.
(a) Overview of the technical procedure used to identify xylan accessible to different xylanases.
(b) Alcohol-insoluble residue (AIR) from Miscanthus culms was hydrolysed with GH10, GH5 or GH30 xylanases and the undigested material was precipitated
with 65% ethanol. Soluble oligosaccharides were reductively aminated with 8-aminonaphthalene-1,3,6-trisulfonic acid (ANTS) and analysed by PACE (S sam-
ples). The ethanol-precipitable pellet was incubated again with one of the remaining xylanases and soluble oligosaccharide products were analysed by PACE (P
samples).
Figure 4. Sequential cell wall extraction from Miscanthus sinensis culms.
Cell walls were sequentially extracted using CDTA, Na2CO3, 1 M KOH, 4 M KOH and 4 M KOH after sodium chlorite treatment (4 M KOH PC). Resulting fractions
were analysed by PACE using GH10, GH11, GH30 and GH5 xylanases. Fractions not treated with enzymes are also shown. Bands marked with asterisks (*) are
non-specific labelling products.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
6 Theodora Tryfona et al.
following fractions and with only traces of the AXe diag-
nostic oligosaccharides detected in the 70–100 mM NaHCO3
fractions (the smaller oligosaccharides from xylan eluting
later arise from other xylan types, perhaps including
GAXc). On the contrary, substantial proportions of GAXc
molecules eluted above 10 mM NaHCO3 fractions, suggest-
ing a more charged molecule compared with AXe. From
this we conclude that AXe and GAXc domains are found in
differing proportions or perhaps are found in distinct xylan
molecules.
Next, we investigated whether it was possible to
resolve the two xylan molecules based on size. Xylan was
subjected to size-exclusion chromatography (SEC) using
simultaneous detection with UV (280 nm) and refractive
index (RI), which detects dissolved substances non-
specifically. We performed SEC using a dimethyl sulfoxide
(DMSO)/LiBr elution system to avoid aggregation phenom-
ena during separation, which could bias the molecular
mass separations (Vilaplana & Gilbert, 2010). For these
experiments, Miscanthus culm AIR was depectinated with
ammonium oxalate and xylan was extracted with 4 M
NaOH. The extracted xylan was then desalted, dissolved in
DMSO and applied to an SEC column. The RI profile gave
a broad asymmetric peak (elution between 14 and 22 min),
whereas the UV signal indicated the presence of more than
one population, and hence we collected the xylan in two
fractions (Figure 6a): (i) fraction 1 (elution between 14 and
19 min), which contained molecules with high molecular
mass; and (ii) fraction 2 (elution between 19 and 22 min),
of lower abundance, which contained smaller molecular
mass polymers. The larger UV signal of fraction 2 com-
pared with fraction 1 might arise from the presence of phe-
nolic (lignin) components associated with this xylan
population that may be resistant to the alkaline treatment
with 4 M NaOH. To study the molecular mass of the SEC
fractions, fractions 1 (SEC_fr1) and 2 (SEC_fr2) were indi-
vidually subjected to SEC-MALS (Figure 6b). We found that
SEC_fr1 contained a single mass population centred
around 30 kDa, whereas SEC_fr2 contained two popula-
tions: a population with smaller molar mass (between 1
and 10 kDa) and a significant proportion of the same mate-
rial as in SEC_fr1. PACE analysis with GH10 hydrolysis con-
firmed that xylan was present in both SEC fractions, and
that it was more abundant in SEC_fr1 (Figure 6c). SEC_fr1
contained most of the GAXc, but a significant quantity also
eluted in SEC_fr2. AXe also eluted mainly in SEC_fr1, but
only trace quantities were observed in SEC_fr2 (Figure 6c).
Therefore, GAXc and AXe xylan molecules were not clearly
separated by size, and both elute mainly in the high molec-
ular mass fraction. GH30 (Figure 6d,e) and GH5 (Figure 6f,
g) digestion resulted in cleavage of all the high molecular
mass xylan in both fractions but did not reduce all the
xylan to small oligosaccharides, as a major peak between
1 and 10 kDa was observed after both enzymatic treat-
ments. The sensitivity of all the higher mass xylan to both
enzymes suggests these molecules contain both GAXc and
Figure 5. Ion-exchange chromatography of xylan.
Alcohol-insoluble residue (AIR) from Miscanthus culms was depectinated with (NH₄)₂C₂O₄, and the xylan extracted by 4 M NaOH was bound to diethy-
laminoethanol (DEAE) cellulose and subjected to stepwise elution with increasing concentrations of NaHCO3 buffer (10–100 mM). The xylan in each fraction was
digested with GH10, GH5 or GH30 enzymes to reveal xylan, Araf-substituted xylan or GlcA-substituted xylan, respectively. For each hydrolysis, three oligosac-
charides were selected and the band intensity was quantified by IMAGEJ to monitor the abundance of the corresponding oligosaccharides across the NaHCO3 buf-
fer concentration gradient. Values are means  standard deviations (SDs) of three independent DEAE experiments.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 7
Figure 6. Size-exclusion chromatography (SEC) analysis and fractionation of Miscanthus culm xylan.
(a) Xylan extracted from Miscanthus culms by 4 M NaOH was analysed and fractionated by SEC using a dimethyl sulfoxide (DMSO)/LiBr elution system. The
xylan eluted into two fractions: fraction 1 (collected after elution between 14 and 19 min) and fraction 2 (collected after elution between 19 and 22 min). The
dashed line indicates the switch from the collection of fraction 1 to the collection of fraction 2. The elugrams represent the signals from the refractive index (RI)
detector (in blue) and the UV detector (in red).
(b) Molar mass distributions (w log M) of the parent (intact) xylan sample and fractions 1 and 2. Fraction 1 shows a monomodal distribution with larger macro-
molecular populations between 10 and 100 kDa. Fraction 2 contains significantly less populations with molar masses between 1–10 kDa and larger macromolec-
ular populations (from fraction 1) that co-eluted during SEC fractionation due to band broadening.
(c) The xylan in each fraction was digested with GH10 xylanase, GH5 arabinoxylanase or GH30 glucuronoxylanase to show the xylan substitution pattern. The
GH10 hydrolysis of fraction 1 (SEC_fr1) and fraction 2 (SEC_fr2) indicates the presence of a significant quantity of xylan in both fractions. A large proportion of
GH30-digestible xylan is present in SEC_fr1, whereas a significant quantity is also present in SEC_fr2. Similarly, most of GH5-digestible xylan eluted in SEC_fr1,
with only minor quantities eluting in SEC_fr2.
(d) Xylan in fraction 1 purified by SEC was hydrolysed by GH30 glucuronoxylanase and subjected again to SEC analysis.
(e) Xylan in fraction 2 purified by SEC was hydrolysed by GH30 glucuronoxylanase and subjected again to SEC analysis.
(f) Xylan in fraction 1 purified by SEC was hydrolysed by GH5 xylanase and subjected again to SEC analysis.
(g) Xylan in fraction 2 purified by SEC was hydrolysed by GH5 xylanase and subjected again to SEC analysis.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
8 Theodora Tryfona et al.
AXe. The lower molecular mass xylan was partially resis-
tant to both enzymes. This is consistent with the presence
of hsGAX that eluted mainly in the small molecular mass
fraction. Monosaccharide analysis conducted with high-
performance anion-exchange chromatography and pulsed
amperometric detection (HPAEC-PAD) is consistent with
SEC_fr2 containing a proportion of more highly substituted
xylan (Figure S4).
Immunolocalization of AXe and GAXc in B. distachyon
culms
Monoclonal antibodies (MAbs) are versatile tools that in
conjunction with fluorescent imaging can detect and local-
ize cell wall glycans in cell walls with high sensitivity
through the specific recognition of oligosaccharide struc-
tures. To investigate whether AXe and GAXc are localized
in different cell types we employed three well-
characterized xylan-directed MAbs (Figure 7a): LM11,
which binds to the unsubstituted b-1,4-linked xylan back-
bone (Ruprecht et al., 2017); LM28 (Cornuault et al., 2015),
which binds to a glucuronosyl-containing epitope widely
present in heteroxylans, such as GAXc; and CCRC-M154,
which selectively binds to xylan oligosaccharides with a 3-
linked Araf substitution (Ruprecht et al., 2017) that should
recognize AXe. We performed antibody labelling of the
middle of the second internode from the base of 5-week-
old culms of B. distachyon (Brachypodium), which were
treated with 4 M NaOH and pectate lyase to improve the
binding of the antibodies (Verhertbruggen et al., 2017).
Transverse Brachypodium culm sections were either
directly labelled with LM11, LM28 or CCRC-M154 antibod-
ies or were treated with GH30 glucuronoxylanase or GH5
arabinoxylanase prior to MAb labelling. Sections were also
counterstained with Calcofluor white, which binds to cellu-
lose and other b-glycans and fluoresces under UV excita-
tion, allowing a very effective visualization of cell walls
(Wood, 1980). To identify lignified tissues, equivalent
transverse sections (not treated with 4 M NaOH) were
stained with Toluidine blue O, which labels poly-aromatic
substances such as lignin (Pradhan Mitra & Loqu
e, 2014).
First, we used Direct Red 23 (Anderson et al., 2009) to
label cellulose and assess the vascular tissue anatomy.
The labelling revealed vascular bundles of phloem cells,
flanked by two large metaxylem vessels and a central
protoxylem cell surrounded by sheaths of fibre cells
(Figure 7b). Secondary cell wall thickening was apparent in
epidermis, sclerenchyma, interfascicular parenchyma and
metaxylem cells. Toluidine blue O stained these walls
bright blue, indicating lignification. Phloem, sheath fibres
and pith parenchyma cell walls were stained dark blue/pur-
ple and therefore were not lignified (Figure 7c). The differ-
ent xylan epitopes recognized by LM11, LM28 and CCRC-
M154 were widely detected, indicating all cell walls con-
tained xylan molecules, and at least a proportion of these
have glucuronic acid and Araf substitutions. However, the
LM11 epitope (unsubstituted xylan) was absent from pith
parenchyma and phloem cell walls and an overall reduced
binding was notable for interfascicular parenchyma, epi-
dermis and sclerenchyma regions (Figures 7d and S4). This
suggests that these cell walls have relatively higher pro-
portions of frequently substituted xylan.
We were interested to understand to what extent the
LM28 anti-glucuronoxylan labelling reflects GAXc pres-
ence. We digested the GAXc by GH30 glucuronoxylanase
treatment of the sections prior to antibody labelling.
Remarkably, LM28 binding was completely abolished in all
cell walls, with only weak binding detected in the corner
cell regions (Figure 7d). The GH30 glucuronoxylanase
treatment exposed additional unsubstituted xylan epitopes
for LM11 binding in all cell walls, excepting primary cell
walls of pith parenchyma, unlignified cell walls of phloem,
sheath fibres and protoxylem. CCRC-M154 binding was
notably reduced in all cell walls but again some labelling
remained in the corner regions of the cell wall (Figures 7d
and S5). These results suggest that highly substituted
enzyme-resistant xylans are distributed in the corners of
cells.
The distribution of AXe was studied by GH5 arabi-
noxylanase treatment of sections prior to antibody label-
ling with CCRC-M154. This enzyme removed labelling from
all cell walls, indicating that this enzyme was very effective
in digesting Ara-substituted xylan in the sections. Similar
to GH30 glucuronoxylanase treatment, GH5 arabinoxy-
lanase treatment exposed unsubstituted xylan epitopes for
LM11 binding in the epidermis, sclerenchyma, vascular
bundles and interfascicular parenchyma. LM28 binding
was also affected by the arabinoxylanase GH5 treatment,
with complete loss of binding from epidermis, pith par-
enchyma, phloem, xylem and the surrounding sheath fibre
cells, whereas LM28 binding was significantly reduced
from sclerenchyma and interfascicular parenchyma cells
(Figures 7d and S4). These results are consistent with the
GAXc of epidermis, pith parenchyma, phloem and fibre
sheath cells possessing the XA3XU2X sequence that is sen-
sitive to both GH30 and GH5. The CCRC-M154 labelling
was more profoundly affected by the arabinoxylanase than
the glucuronoxylanase, consistent with it labelling AXe in
addition to the GAXc. Together, the demonstration of the
sensitivity of MAb labelling to enzyme digestion suggests
that neither AXe nor GAXc are restricted to certain cell
types and are found in cells with primary and with sec-
ondary cell walls.
DISCUSSION
There is increasing interest in the arrangement of substitu-
tions along the xylan backbone and the resulting implica-
tions for the interaction of this polysaccharide with
cellulose and lignin. It has been demonstrated that there is

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 9
a precise patterning of some xylan substitutions, not only
in eudicots (Bromley et al., 2013; Busse-Wicher et al., 2014;
Vrsansk
a et al., 2007) but also in gymnosperms (Busse-
Wicher, Li, et al., 2016; Mart
ınez-Abad et al., 2017). In cer-
tain plant groups, patterning has been observed for xylan
acetylation as well as for glucuronidation. This very
Figure 7. Indirect immunofluorescence detection of xylan epitopes in transverse sections of Brachypodium distachyon internodes before and after treatment of
sections with GH30 glucuronoxylanase or GH5 arabinoxylanase. Equivalent sections were also stained with Direct Red 23 and toluidine blue to show the tissue
anatomy and deposition of lignin in the walls of all cell types.
(a) Schematic representation of epitopes (highlighted in light blue) of cell wall-directed antibodies LM11, LM28 and CCRC-M154 according to Ruprecht
et al. (2017). Linkages that are marked with purple indicate positions that must not be substituted for antibody to bind. Light-grey sugar residues and linkages
indicate positions for substitutions that are allowed but not required for binding (Ruprecht et al., 2017). A, Araf; U, GlcA; X, Xyl.
(b) Direct Red 23 (red) showing tissue anatomy. Close-up panel shows an individual vascular bundle and surrounding cells, secondary cell wall thickening is vis-
ible for epidermis, sclerenchyma and metaxylem cell walls.
(c) Microscopic inspection of sections from culms from Brachypodium distachyon stained with toluidine blue to identify lignified cells.
(d) Xylan immunolabelling of LR-white serial sections with LM11 (relatively unsubstituted xylan), LM28 (GlcA-substituted xylan) and CCRC-M154 (Ara-
substituted xylan) MAbs, comparing the labelling pattern following no enzyme treatment, GH30 glucuronoxylanase hydrolysis or GH5 arabinoxylanase treat-
ment, respectively. All sections underwent Na2CO3 and alkali treatment (4 M NaOH), which removed ester-linked substituents from the cell walls, and pectate
lyase treatment to enhance mAb binding. Calcofluor staining of the GH5-treated sections are shown for all treatments as a comparison (right column). Key for
arrows: empty triangles, epidermis; white arrowheads, sclerenchyma; double arrowheads, phloem; white triangles, fibres; empty arrows, interfascicular par-
enchyma; white arrows, pith parenchyma. Ep, epidermis; F, fibres; iPa, interfascicular parenchyma; Mx, metaxylem; Ph, phloem; pPa, pith parenchyma; Px, pro-
toxylem; Sc, sclerenchyma. Scale bars: whole sections, 200 lm; close-ups, 50 lm.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
10 Theodora Tryfona et al.
specific spacing may facilitate the interaction of twofold
screw xylan with the hydrophilic surface of the cellulose
microfibrils of plant cell walls (Busse-Wicher et al., 2014;
Busse-Wicher, Grantham, et al., 2016; Busse-Wicher, Li,
et al., 2016; Grantham et al., 2017; Gupta et al., 2021; Kang
et al., 2019; Mart
ınez-Abad et al., 2017). Here, we show that
the precise patterning is even more widely spread in vas-
cular plants, by extending these observations to new pat-
terns and plant groups: the arabinosylation of xylan and
grasses. We further demonstrate that grass cell wall xylan
consists of distinct types with different substitution pat-
terns that are widely present in cell walls. We hypothesize
that these xylan populations in grasses reflect functional
specialization and interact differently with cellulose and
other cell wall components.
We studied the spacing of Araf and GlcA substitutions
along the xylan backbone. Interestingly, we found a xylan
fraction with an even number of Xylp residues between
each Araf, with the spacing ranging from two to at least 16
Xylp residues apart. As this arabinoxylan has evenly
spaced Araf substitutions, we named it AXe. In contrast, to
the even spacing of Araf, GlcA substitutions were detected
predominantly spaced five, six or seven Xyl residues apart.
One of the products of GH30 glucuronoxylanase was found
to carry an Araf substitution. We named this glucuronoara-
binoxylan with clustered GlcA, GAXc. The pattern of Araf
and GlcA substitutions was found to be remarkably similar
in all grass species analysed, supporting the view that
there is a relatively precise, non-random arrangement of
the substitutions. Our finding of the regular patterning of
Araf and GlcA substitutions in grass xylan is consistent
with earlier studies giving evidence of a repetitive structure
in grass xylan (Carpita et al., 2001; Faik, 2010; Gruppen
et al., 1993; Kozlova et al., 2012; Zeng et al., 2010).
AXe and GAXc are widely present in cell walls and are not
always found on separate molecules
Our finding of the structurally different domains of xylan
raised the question of whether the domains of xylan are
present on the same or distinct molecules and whether
they are present in specific types of cell walls, such as pri-
mary and secondary cell walls. We found that both AXe
and GAXc represent substantial proportions of cell wall
xylan. To separate any different populations of xylan from
Miscanthus culms, we performed sequential cell wall
extraction, ion exchange and SEC. The separation profiles
provided evidence for the somewhat differing extractability
of xylan domains from the cell wall and differently charged
xylan populations. SEC separation resulted into a low
molecular weight and a more abundant high molecular
weight xylan population containing the majority of both
AXe and GAXc. The low-molecular-weight fraction might
contain mostly hsGAX, as it was most easily digestible
with GH10 xylanase. On the other hand, the high-
molecular-weight fraction was sensitive to both GH5 arabi-
noxylanase and GH30 glucuronoxylanase, suggesting that
these high-molecular-weight molecules contain both AXe
and GAXc. Both SEC fractions absorbed UV. Therefore, we
speculate that AXe and GAXc are: (i) two distinct mole-
cules that are cross linked with alkali-resistant phenolic–
polysaccharide ether bonds (Burr & Fry, 2009); or (ii) con-
tiguous domains contained within the same xylan mole-
cule that is covalently linked to lignin through alkali-
resistant benzyl ether bonds (Iiyama et al., 1990; Lam
et al., 2003; Mart
ınez-Abad et al., 2018), forming lignin–car-
bohydrate complexes.
Our data on the sequential chemical extraction of
xylan from cell walls of Miscanthus culms are consistent
with previous studies on cell walls of various grass spe-
cies (Kulkarni et al., 2012). Xylans with varying degrees of
substitution have been reported for grasses: a GAX with a
high degree of Araf substitution (hsGAX) and a GAX with
a low degree of substitution (lsGAX) (Shrestha
et al., 2019; Tabuchi et al., 2011; Wang et al., 2014, 2016).
These xylans are reported to interact differently with cellu-
lose: hsGAX is thought to be more mobile than lsGAX
and is found in the interstitial matrix, playing the role of a
‘filler’, whereas the more rigid, relatively unsubstituted
lsGAX is tightly associated with cellulose microfibrils
(Wang et al., 2016). Consistent with these findings,
Kozlova et al. (2014) described three domains of GAX in
maize roots: one keeping cellulose microfibrils apart, one
interacting with them and a middle domain between the
two, which links them. It is therefore conceivable that
some grass GAX is tightly adherent to cellulose microfib-
rils and other cell wall components, providing mechanical
strength, whereas other GAX keeps cellulose microfibrils
separated, which would give cell walls plasticity and allow
them to expand.
Using fluorescence microscopy with MAb probes, and
arabinoxylan and glucuronoarabinoxylan-specific degrad-
ing enzymes, we found that AXe and GAXc molecules are
widespread in walls of different cell types of Brachy-
podium culm. This supports a role for these different
xylans in the same walls, across many cell types. We found
that pith parenchyma, phloem and sheath fibre cell walls
were distinctive, as they showed no LM11 epitopes (unsub-
stituted xylan) but had relatively abundant levels of LM28
(uronic acid substituted) and CCRC-M154 (3-Araf substi-
tuted). These walls do not have secondary thickening and
are not lignified, so these antibody labelling results sug-
gest that unlignified tissues contain more highly substi-
tuted xylan. This is consistent with earlier findings on
maize culms (Suzuki et al., 2000): low-substituted GAX was
found in lignified walls and highly substituted GAX was
found in unlignified and primary cell walls at an early
developmental stage. Although our probes are likely to
have revealed the most abundant xylan molecules, it is

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 11
also possible that there is masked xylan in the culm sec-
tions not initially bound by the antibodies or cleaved by
the hydrolase enzymes. Some LM11 epitopes, for example,
were revealed after GH5 and GH30 hydrolysis, whereas
both LM28 and CCRC-M154 epitopes localized strongly in
cell wall corner areas after GH30 enzymatic treatment.
Recently, transcriptome and immunocytochemistry
analysis during the elongation growth of maize roots
showed that homologous members of glycosyltransferase
families involved in GAX biosynthesis were expressed dif-
ferently during the early and late stages of development,
suggesting that two separate sets of genes may be respon-
sible for the biosynthesis of GAX for primary and sec-
ondary cell wall deposition. Glycosyltransferases (GTs)
involved in the synthesis of secondary cell wall GAX from
GT61 (XAX, MUCI) and GT8 (GUX1 and GUX2) families
were widely present at late stages of development
(Kozlova et al., 2020).
Structural variation may reflect functional specialisation
and distinct interactions with wall components
Here, we propose that the structurally different AXe, GAXc
and hsGAX molecules might have distinct interactions with
cellulose and lignin, reflecting functional specialization,
and speculative linkages and roles are shown in Figure 8.
We hypothesize that an even distribution of the hydrophilic
Araf residues on the AXe backbone would favour a twofold
conformation of the xylan molecule. Molecular dynamics
simulations predict evenly decorated xylan would adopt a
twofold helical screw conformation upon interaction with
the hydrophilic cellulose surface (Busse-Wicher et al., 2014;
Gupta et al., 2021). We do not yet know whether the AXe is
acetylated, as our samples were alkali-extracted, but an
even pattern of acetylation could further promote the xylan–
cellulose association (Gupta et al., 2021; Pereira et al., 2017).
Furthermore, our recent ssNMR study on never-dried B. dis-
tachyon tissues confirms that some xylan in grasses has
branches that permit binding to cellulose in the twofold
conformation (Duan et al., 2021). Indeed, this xylan is prefer-
entially acetylated, suggesting that AXe might be more
highly acetylated than other xylans in grasses.
Some Araf residues on grass xylan are modified with
ferulic acid (Feijao et al., 2021; Mueller-Harvey et al., 1986).
Strong evidence indicates that xylan in the secondary walls
of grasses is cross-linked to lignin by the extensive copoly-
merization of their ferulates (de Oliveira et al., 2015; Feijao
et al., 2021; Grabber et al., 1995; Grabber & Lu, 2007;
Ralph, 2010). Ferulic acid has also been described to cross-
link xylan chains through oxidative coupling with other fer-
ulic acid groups (Feijao et al., 2021; Ishii, 1997; Ralph, 2010).
Hence, we hypothesize that AXe could be cross-linked
either with other xylan molecules or with lignin via diferu-
late bridges (Figure 8, boxes A and B, respectively).
We postulate that clustered GlcA substitutions on
GAXc would obstruct the interaction of this molecule with
the hydrophilic cellulose surface through steric hindrance
(Busse-Wicher et al., 2014; Grantham et al., 2017). There-
fore, this type of xylan would adopt a threefold helical
screw conformation, would extend into the matrix and
would interact with other cell wall components, such as
lignin (Figure 8). Indeed, recent ssNMR data have indicated
that lignin could preferentially bind xylans with threefold
or distorted twofold helical screw conformations that are
not closely associated with cellulose (Kang et al., 2019). As
Figure 8. Molecular model indicating possible interactions of different grass secondary cell wall xylans with other cell wall components.
AXe (green chair conformations), with an even distribution of Araf residues, could adopt a twofold conformation and tightly associate with the hydrophilic sur-
face of cellulose (orange chair conformations). Some Araf residues could be feruloylated and therefore AXe could cross-link with other xylan chains (box A) or
lignin (box B). GAXc (brown chair conformations), with clustered distribution of GlcA substitutions, is hypothesized to adopt a threefold helical screw conforma-
tion, and if some of Araf residues are feruloylated may tether adjacent xylan chains (box A) or cross-link to lignin (box C). GAXc could also cross-link to lignin
via GlcA substitutions (box D). hsGAX (grey chair conformations) is most likely to be situated in the matrix and plays the role of “space filler”. For simplicity,
AXe and GAXc are shown as separate molecules. Additional phenolic–polysaccharide or phenolic–lignin ether linkages are possible.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
12 Theodora Tryfona et al.
GAXc is also modified with Araf residues, we hypothesize
that some of these residues could be feruloylated. There-
fore, GAXc could covalently cross-link to lignin or other
xylan chains through ferulic acid (Figure 8, box C). Hard-
wood glucuronoxylan can be esterified to lignin via its
GlcA substituents (Nishimura et al., 2022). It is therefore
plausible that some of the GlcA residues on GAXc are
esterified to lignin (Figure 8, box D).
hsGAX is probably in a threefold screw conformation
because of the dense substitutions and is likely to be found
in the matrix. This hypothesis is consistent with several
studies suggesting that a type of highly substituted mobile
GAX (hsGAX) is found in the interfibrillar space and plays
the role of a ‘filler’, whereas the relatively unsubstituted
GAX is more tightly associated with cellulose microfibrils
(Carpita et al., 2001; Kozlova et al., 2014; Wang et al., 2014,
2016) and contributes to cell wall strength and mechanics,
functioning as a cementing substance (Tabuchi et al.,
2011).
CONCLUSION
Over the last few years there has been an increasing num-
ber of studies into the patterning of decoration of xylan
and the implications this patterning has on interactions
with other cell wall components. Here, we demonstrated
that the even spacing of decorations is highly conserved in
plants and can be found not only in Arabidopsis (Bromley
et al., 2013; Busse-Wicher et al., 2014), hardwoods (Busse-
Wicher, Li, et al., 2016; Derba-Maceluch et al., 2015;
Vrsansk
a et al., 2007) and softwoods (Busse-Wicher, Li,
et al., 2016), but also in grasses. Furthermore, we demon-
strated that distinct xylan molecules are found in walls of
many cell types, suggesting the functional specialization of
xylan within individual walls. Our work highlights the need
of xylan biosynthesis-related genetic resources in plants
such as B. distachyon for the detailed study of grass
xylans, as it will be important to study the function of dif-
ferent xylans by creating mutants in the different xylan
types. The engineering of grass xylan could result in bio-
mass with distinctive and optimized properties for use in a
more sustainable bioeconomy.
EXPERIMENTAL PROCEDURES
Plant material
The plant materials used in this study were fresh stem material
from plants aged 2.5–3.0 months: Andropogon gerardii (collected
from the University of Cambridge Botanic garden), B. distachyon
(Brachypodium; grown in the University of Cambridge glass-
house), Miscanthus sinensis (Miscanthus; from Luisa Trindade
and Oene Dolstra, Wageningen University, the Netherlands),
O. sativa (rice; grown in the University of Cambridge glasshouse),
Phragmites australis (collected from the University of Cambridge
Botanic garden), Phyllostachys viridiglaucescens (bamboo; col-
lected from the University of Cambridge Botanic garden),
Saccharum spp. SP80-3280 (sugar cane; from Marcos Buckeridge,
University of Sao Paulo, Brazil,), T. aestivum (wheat; from Greg
Tucker, University of Nottingham, UK) and Z. mays (maize; grown
in the University of Cambridge glasshouse).
Extraction of AIR
Plant culms were harvested, submerged in 96% (v/v) ethanol and
boiled at 70°C for 30 min to inactivate the enzymes. Following
homogenization using a ball mixer mill (Glen Creston, now
Retsch, https://www.retsch.com), the pellet was collected by cen-
trifugation (4000 g for 15 min) and was washed with 100% (v/v)
ethanol, twice with chloroform:methanol (2:1), followed by suc-
cessive washes with 65% (v/v), 80% (v/v) and 100% (v/v) ethanol.
The remaining pellet of AIR was air dried.
Hemicellulose extraction
AIR preparations (100 mg) were depectinated in ammonium oxa-
late (0.5% w/v) at 85°C for 2 h, washed with H2O and subjected to
alkali extraction in 4 M NaOH for 1 h at room temperature (RT)
(21°C). NaOH was removed from the extracted xylan in the super-
natant by passing through a PD-10 column (GE Healthcare, https://
www.gehealthcare.com) and the recovered xylan was kept in ali-
quots for further analysis.
Enzymatic hydrolysis and enzymes
The enzymes used in this study were: GH10 endo-b-1,4-xylanase
CjGH10 from Cellvibrio japonicus (Charnock et al., 1998); GH115 a-
glucuronidase BoGH115 from Bacteroides ovatus (Rogowski
et al., 2014; Ryabova et al., 2009); GH62 a-arabinofuranosidase
PaGH62 from Penicillium aurantiogriseum (Peng et al., 2017) and
GH3 b-1,4 xylosidase TrGH3 from Trichoderma reesei (Margolles-
Clark et al., 1996), both gifts from Novozymes (https://www.
novozymes.com); endo-glucuronoxylanase EcGH30 from Dickeya
dadantii (formerly known as (Erwinia chrysanthemi) (Urb
anikov
a
et al., 2011); and endo arabinoxylanase CtGH5 from Clostridium
thermocellum (Correia et al., 2011), a gift from Professor Harry Gil-
bert (Newcastle University, UK). All enzymes were added at a final
concentration of 2 lM and incubated at 21°C (40°C for CtGH5)
under constant shaking for 24 h.
Five hundred micrograms of extracted xylan was used for
enzymatic hydrolysis. Xylan was hydrolysed in 50 mM ammonium
acetate buffer, pH 6.0 (pH 4.0 for CtGH5), overnight before boiling
for 30 min to heat-inactivate the enzymes. After digestion, sam-
ples were taken to dryness in vacuo before further processing,
depending upon the analytical technique used.
For the analysis of xylanase accessible xylan, enzymatic
hydrolyses were stopped by the addition of three reaction vol-
umes of 96% (v/v) ethanol before precipitation of the undigested
substrates and separation of the reaction products for further
enzymatic hydrolysis.
Polysaccharide analysis by carbohydrate gel
electrophoresis
The derivatization of carbohydrates was performed according to
previously developed protocols (Goubet et al., 2002). Carbohy-
drate electrophoresis and PACE gel scanning and quantification
was performed as described by Goubet et al. (2002, 2009). Control
experiments without substrates or enzymes were performed
under the same conditions to identify any non-specific com-
pounds in the enzymes, polysaccharides/cell walls or labelling
reagents.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 13
Preparation, reductive amination, purification of
xylooligosaccharides for MALDI-MS and MALDI-MS/MS
CID analysis
Xylooligosaccharides were desalted using HyperSep Hypercarb
cartridges (Thermo-Hypersil-Keystone, part of ThermoFisher Sci-
entific, https://www.thermofisher.com), as previously described
(Tryfona et al., 2019). Oligosaccharides were lyophilized and then
reductively aminated with 2-AA (Sigma-Aldrich, https://www.
sigmaaldrich.com) before being purified from the reductive ami-
nation reagents using a Glyko Clean S cartridge (Prozyme, now
Agilent, https://www.agilent.com), as previously described (Try-
fona & Stephens, 2010). One microliter of reaction supernatant
was mixed with an equal volume of 20 mg ml
1 2,5-
dihydroxybenzoic acid (Sigma-Aldrich) matrix in 50% methanol
containing 0.4 mg ml
1 ammonium sulphate ((NH4)2SO; Enebro &
Karlsson, 2006) and was spotted on an MTP 384 ground steel tar-
get plate (Bruker, https://www.bruker.com). The spotted samples
were then air dried before being analysed on an UltrafleXtreme
MALDI-TOF/TOF mass spectrometer (Bruker). MALDI-CID spectra
were acquired with an average 50 000 laser shots per spectrum.
The oligosaccharide ions were allowed to collide in the CID cell
with argon at a pressure of 7.2 9 10
6 T.
Solution NMR analysis of CtGH5 dependent trisaccharide
Depectinated, saponified Miscanthus culm AIR (250 mg) was
hydrolysed with CtGH5 arabinoxylanase, followed by TrGH3 b-1,4-
xylosidase, using the enzyme hydrolysis conditions described
above. The resulting oligosaccharide mixture was then fraction-
ated using HyperSep Hypercarb cartridges (Thermo-Hypersil-
Keystone) with increasing acetonitrile gradient. Consequently,
fractions containing the CtGH5-dependant trisaccharide were
pooled, solubilized in 0.6 ml of D2O and analysed by NMR. NMR
spectra were recorded at 298 K with a Bruker AVANCE III spec-
trometer operating at 600 MHz equipped with a TCI CryoProbe.
Two-dimensional 1H-1H TOCSY, ROESY, 13C HSQC, HSQC-TOCSY
and H2BC experiments were performed, using established meth-
ods (Cavanagh et al., 1996; Nyberg et al., 2005); the mixing times
were 70 and 200 msec for the TOCSY and ROESY experiments,
respectively. Chemical shifts were measured relative to internal
acetone (dH = 2.225, dC = 31.07 ppm). Data were processed using
AZARA 2.8 (copyright 1993–2022, Wayne Boucher and Department
of Biochemistry, University of Cambridge, unpublished) and
chemical-shift assignment was performed using ANALYSIS 2.4 (Vran-
ken et al., 2005).
Reducing-end reduction
The reducing end was reduced with 10 mg ml
1 NaBH4 in 0.5 M
NaOH for 2 h at RT. The reaction was terminated by adjustment to
pH 5.5 with glacial acetic acid on ice. The samples were desalted
by passing through a PD-10 column (GE Healthcare).
Cell-wall sequential fractionation
Cell-wall sequential fractionation was based upon the methods
described by Pattathil et al. (2012), with minor alterations. AIR
(200 mg) was suspended in 10 ml of 0.05 M CDTA (pH 6.5) for 24 h
at RT. The suspension was centrifuged (48 000 g) and the pellet
washed once with distilled H2O. The supernatants were combined
as the CDTA-soluble fraction. The AIR was subsequently extracted
using increasingly harsher chemicals: 0.05 M Na2CO3 containing
0.01 M NaBH4 for 24 h at 4°C (Na2CO3-soluble fraction), 1 M KOH
containing 0.01 M NaBH4 for 24 h at RT (1 M KOH-soluble fraction),
4 M KOH containing 0.01 M NaBH4 for 24 h at RT (4 M KOH-soluble
fraction) and then acidic sodium chlorite (3 mg ml
1 for 2 h at 80
°C) followed by 4 M KOH (4 M KOH post chlorite-soluble fraction)
containing 0.01 M NaBH4 for 24 h at RT. All fractions were filtered
through a GF/C glass fibre filter (Whatman, now Cytiva, https://
www.cytivalifesciences.com). The Na2CO3 and KOH fractions were
also chilled on ice and adjusted to pH 5.0 with glacial acetic acid.
All cell-wall fractions were then dialysed extensively against
deionized H2O for 5 days, and then lyophilized. Samples were fur-
ther desalted by passing through a PD-10 column (GE Healthcare)
before further analysis.
Ion-exchange chromatography
Ten milligrammes of depectinated, saponified and desalted AIR
(as described above) were mixed with activated DEAE DE 32 cellu-
lose (Whatman) in 10 mM NaHCO3, pH 6.0. The suspension was
loaded onto chromatography cartridges (Bio-Rad, https://www.
bio-rad.com), and the flow through collected. Bound xylan was
eluted buffer of increasing ionic strength of NaHCO3. Aliquots of
each sample were hydrolysed with CtGH5, EcGH30 or CjGH10 and
analysed by PACE. Three representative oligosaccharides from
each hydrolysis were selected, with signal intensity for three inde-
pendent PACE analyses calculated with IMAGEJ and the mean of
these three independent replicas was calculated.
Size-exclusion chromatography
The molar mass distributions of the xylan extracts from depecti-
nated, saponified Miscanthus culm AIR (100 mg) were analysed
by size-exclusion chromatography (SECcurity 1260; Polymer Stan-
dards Services, https://www.pss-polymer.com) coupled in series
to a multiple-angle laser light scattering detector (MALLS; BIC-
MwA7000; Brookhaven Instruments, https://www.
brookhaveninstruments.com), a refractive index (RI) detector
(SECcurity 1260; Polymer Standards Services) and a UV detector
(SECcurity 1260; Polymer Standards Services) at a wavelength of
280 nm, as previously described (Morais de Carvalho et al., 2017).
All SEC analyses were performed at a flow rate of 0.5 ml min
1
using DMSO (HPLC grade; Sigma-Aldrich) with 0.5% w/w LiBr
(ReagentPlus) as a mobile phase, using a column set consisting of
a GRAM PreColumn, 100 and 10 000 analytical columns (Polymer
Standards Services) held by thermostat at 60°C. Prior to the analy-
ses, the xylans were dissolved directly in the SEC eluent for 16 h
at 60°C. For the size fractionation of the xylan, a concentration of
10 g L
1 was injected repeatedly (five times), and the fractions
(SEC_fr1 and SEC_fr2) were separated manually after elution and
pooled together for further analysis. Each fraction was analysed
individually at a concentration of 1 g L
1 using the same set-up as
for the fractionation. Standard calibration was performed by the
injection of Pullulan Standards of known molar masses provided
by the Polymer Standards Services.
Trifluoroacetic acid hydrolysis and HPAEC-PAD monosac-
charide analysis
Samples (200 lg) were hydrolysed in 2 M trifluoroacetic acid for
1 h at 120°C. Trifluoroacetic acid was removed by evaporation
under vacuum. Following resuspension in 200 ll of water, the
monosaccharide sugars were separated as previously described
(Tryfona et al., 2012) on a Dionex ICS3000 system equipped with a
PA20 column, a PA20 guard column, and a borate trap (Dionex,
now ThermoFisher Scientific). In particular, for the separation of
neutral sugars, the initial isocratic wash with 12 mM KOH from 0
to 5 min followed a linear gradient from 5 to 20 min, down to

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
14 Theodora Tryfona et al.
1 mM KOH. Finally, an isocratic wash step in 100 mM KOH was
maintained for 5 min (from 20 to 25 min). The post-column addi-
tion of 100 mM KOH was performed to increase PAD sensitivity.
Histology staining and in situ immunolabelling
Approximately 4-mm-long culm sections of 5-week-old B. dis-
tachyon were dissected with razor blades and immediately fixed
in a fixative solution consisting of paraformaldehyde (4% v/v;
Sigma-Aldrich) with glutaraldehyde (0.5% v/v; Sigma-Aldrich) in
phosphate-buffered saline (PBS). Fixation, dehydration, resin infil-
tration and antibody washing steps were all preformed in a PELCO
BiowavePro with SteadyTempTM (Ted Pella, https://www.tedpella.
com). Fixation was microwave radiation assisted at 150 W, under
vacuum (20 Hg) for 1 min (five times) and samples were incubated
in fixative solution overnight at 4°C. Accordingly, samples were
washed three times with PBS buffer before culms were aligned
and embedded in 1% agarose in PBS buffer. Samples were dehy-
drated with a graded ethanol series and infiltrated (LR White med-
ium grade; Agar Scientific, https://www.agarscientific.com)
through increasing resin concentration assisted by microwave
radiation: 150 W; vacuum 20 Hg for 5 min. Resin was polymerized
at 60°C during 17 h. Transverse semi-thin sections (1 lm) were
then obtained with an EM UC7 ultra-microtome (Leica, https://
www.leicabiosystems.com) and laid on SuperFrost Ultra Plus
(ThermoFisher Scientific) microscopy slides.
Histology assessment through Direct Red 23 labelling was
performed by mounting the sections in a solution consisting of a
1:1 solution of AF1 antifading (Citifluor, https://www.citifluor.com)
with PBS containing 0.1 mg ml
1 of Direct Red 23 (Sigma-Aldrich).
Images were then acquired by confocal laser scanning microscopy
(LSM700; Zeiss, https://www.zeiss.com).
All toluidine blue stained sections were incubated with 0.02%
aqueous solution of toluidine blue O for 5 min at RT. Accordingly,
samples were washed five times with water, transferred onto a
microscope slide and observed with a Zeiss Axiomager.M2 under
bright-field lighting.
Prior to immunolocalization, all sections were treated with
fresh sodium carbonate (Na2CO3; 0.1 M) for 2 h at RT, according to
previously established protocols (Marcus et al., 2010; Xue
et al., 2013). Na2CO3 was removed by washing five times with
water, and sections were then incubated with 4 M NaOH for 1 h at
RT. Subsequently sections were washed five times with water.
The removal of pectic homogalacturonan to enhance antibody
binding was carried out using pectate lyase (Aspergillus sp.;
Megazyme) in 50 mM 3-(cylcohexylamino)-1-propanesulfonic acid,
2 mM CaCl2 buffer, pH 10, at 25 lg ml

1 for 2 h at RT. Subse-
quently, samples were washed three times with 50 mM ammo-
nium acetate, pH 6.0, buffer. In some cases, culm sections were
pre-treated, prior to immunolabelling, with enzymes to remove
specific cell wall polysaccharides. The removal of xylan was car-
ried out using CtGH5 in 50 mM ammonium acetate buffer, pH 7.0,
at 40°C for 24 h and EcGH30 in 50 mM ammonium acetate buffer,
pH 6.0, at 21°C for 24 h. Control sections not treated with enzymes
(CtGH5 or EcGH30) were incubated for an equivalent time with the
corresponding buffers alone. Following enzymatic treatment, sec-
tions were washed with PBS three times, then incubated for 1 h
with 2% bovine serum albumin protein/PBS (BSA/PBS) to prevent
non-specific binding, and then washed for 5 min with PBS. The
MAbs used in this study were the rat MAbs LM11 and LM28 and
the mouse MAb CCRC-M154. LM11 (Ruprecht et al., 2017) anti-
body binds to unsubstituted xylan, LM28 antibody (Cornuault
et al., 2015) is directed to a [Me]GlcA-containing epitope on
heteroxylan and CCRC-M154 (Ruprecht et al., 2017) antibody binds
strongly to 3-substituted arabinofuranoses on xylan. Primary
MAbs in 2% BSA/PBS buffer were incubated overnight at 4°C, with
15-fold dilutions for LM11 and LM28 and 1-fold dilution for CCRC-
M154. Sections were then washed five times with 2% BSA/PBS
buffer before secondary antibodies, anti-rat IgG-FITC (Sigma-
Aldrich) and anti-mouse IgG-Alexa 488 (ThermoFisher Scientific),
at a 100-fold dilution, were added in 2% BSA/PBS buffer and incu-
bated for 2 h at RT in the dark. Sections were washed with 2%
BSA/PBS buffer five times, then mounted in anti-fade reagent Citi-
fluor AF1 (Agar Scientific) containing 35 lg ml
1 of Calcofluor
White M2R (Sigma-Aldrich) for cell wall counter-staining. Sections
were observed with a confocal scanning microscope (Zeiss
LSM700) and images were processed with IMAGEJ.
AUTHOR CONTRIBUTIONS
TT designed and performed experiments and analysed the
data. MB and FV performed some of the experiments. KS
performed and analysed the NMR. RDM provided technical
assistance to TT. TT and PD conceived the original
research plans and wrote the article. PD agrees to serve as
the author responsible for correspondence.
ACKNOWLEDGEMENTS
We thank Paul Knox for the LM11 and LM28 antibodies, Dr Peter
Michna (Cambridge University Botanic Garden) for help collecting
grass samples and Ms Xiaolan Yu for technical support. This work
was supported as part of The Center for Lignocellulose Structure
and Formation, an Energy Frontier Research Center funded by the
US Department of Energy, Office of Science, Basic Energy
Sciences, under award number DE-SC0001090. The Cambridge
NMR facility infrastructure is funded by the Biotechnology and
Biological Sciences Research Council (BBSRC) and the Wellcome
Trust. FV acknowledges financial support from the Swedish
Research Council (project 2020-04720).
CONFLICT OF INTEREST
The authors declare that they have no conflicts of interest
associated with this work.
DATA AVAILABILITY STATEMENT
All relevant data can be found within the article and its
supporting materials. To obtain raw data or materials,
please contact the corresponding author.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online ver-
sion of this article.
Figure S1. Solution nuclear magnetic resonance (NMR) analysis.
Figure S2. MALDI-LIFT MS/MS of the XA3XU2X oligosaccharide
released by GH30 glucuronoxylanase from Miscanthus culms
labelled with 2-AA.
Figure S3. Schematic representation of the GH10 endoxylanase
action on the xylan backbone.
Figure S4. HPAEC-PAD analysis of SEC fraction 1 (SEC_fr1) and
fraction 2 (SEC_fr2).
Figure S5. Indirect immunofluorescence detection of xylan epi-
topes in whole transverse sections of Brachypodium distachyon
internodes before and after treatment of sections with GH30 glu-
curonidase or GH5 arabinoxylanase.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 15
Table S1. 1H and 13C NMR assignments of all oligosaccharide
structures in Figure S1, at 25°C in D2O.
REFERENCES
Anderson, C.T., Carroll, A., Akhmetova, L. & Somerville, C. (2009) Real-time
imaging of cellulose reorientation during cell wall expansion in Ara-
bidopsis roots. Plant Physiology, 152, 787–796.
Bromley, J.R., Busse-Wicher, M., Tryfona, T., Mortimer, J.C., Zhang, Z.,
Brown, D.M. et al. (2013) GUX1 and GUX2 glucuronyltransferases deco-
rate distinct domains of glucuronoxylan with different substitution pat-
terns. The Plant Journal, 74, 423–434.
Burr, S.J. & Fry, S.C. (2009) Extracellular cross-linking of maize arabinoxy-
lans by oxidation of feruloyl esters to form oligoferuloyl esters and
ether-like bonds. The Plant Journal, 58, 554–567.
Busse-Wicher, M., Gomes, T.C.F., Tryfona, T., Nikolovski, N., Stott, K., Gran-
tham, N.J. et al. (2014) The pattern of xylan acetylation suggests xylan
may interact with cellulose microfibrils as a twofold helical screw in the
secondary plant cell wall of Arabidopsis thaliana. The Plant Journal, 79,
492–506.
Busse-Wicher, M., Grantham, N.J., Lyczakowski, J.J., Nikolovski, N. &
Dupree, P. (2016) Xylan decoration patterns and the plant secondary cell
wall molecular architecture. Biochemical Society Transactions, 44, 74–78.
Busse-Wicher, M., Li, A., Silveira, R.L., Pereira, C.S., Tryfona, T., Gomes,
T.C.F. et al. (2016) Evolution of xylan substitution patterns in gym-
nosperms and angiosperms: implications for xylan interaction with cellu-
lose. Plant Physiology, 171, 2418–2431.
Carpita, N.C., Defernez, M., Findlay, K., Wells, B., Shoue, D.A., Catchpole, G.
et al. (2001) Cell Wall architecture of the elongating maize coleoptile.
Plant Physiology, 127, 551–565.
Cavanagh, J., Fairbrother, W.J., Palmer, A.G. & Skelton, N.J. (1996) Protein
NMR spectroscopy: principles and practice. San Diego, CA: Academic Press.
Charnock, S.J., Spurway, T.D., Xie, H., Beylot, M.-H., Virden, R., Warren,
R.A.J. et al. (1998) The topology of the substrate binding clefts of glyco-
syl hydrolase family 10 xylanases are not conserved. The Journal of Bio-
logical Chemistry, 273, 32187–32199.
Cornuault, V., Buffetto, F., Rydahl, M.G., Marcus, S.E., Torode, T.A., Xue, J.
et al. (2015) Monoclonal antibodies indicate low-abundance links
between heteroxylan and other glycans of plant cell walls. Planta, 242,
1321–1334.
Correia, M.A.S., Mazumder, K., Br
as, J.L.A., Firbank, S.J., Zhu, Y., Lewis,
R.J. et al. (2011) Structure and function of an arabinoxylan-specific xyla-
nase. The Journal of Biological Chemistry, 286, 22510–22520.
Cosgrove, D. & Jarvis, M. (2012) Comparative structure and biomechanics
of plant primary and secondary cell walls. Frontiers in Plant Science, 3,
204.
da Costa, R.M., Pattathil, S., Avci, U., Lee, S.J., Hazen, S.P., Winters, A.
et al. (2017) A cell wall reference profile for miscanthus bioenergy crops
highlights compositional and structural variations associated with devel-
opment and organ origin. The New Phytologist, 213, 1710–1725.
de Oliveira, D.M., Finger-Teixeira, A., Rodrigues Mota, T., Salvador, V.H.,
Moreira-Vilar, F.C., Correa Molinari, H.B. et al. (2015) Ferulic acid: a key
component in grass lignocellulose recalcitrance to hydrolysis. Plant
Biotechnology Journal, 13, 1224–1232.
Derba-Maceluch, M., Awano, T., Takahashi, J., Lucenius, J., Ratke, C., Kon-
tro, I. et al. (2015) Suppression of xylan endotransglycosylase
PtxtXyn10A affects cellulose microfibril angle in secondary wall in aspen
wood. New Phytologist, 205, 666–681.
Duan, P., Kaser, S., Lyczakowski, J.J., Phyo, P., Tryfona, T., Dupree, P. et al.
(2021) Xylan structure and dynamics in native Brachypodium grass cell
walls investigated by solid-state NMR spectroscopy. American Chemical
Society Omega, 6, 15460–15471.
Ebringerov
a, A. & Heinze, T. (2000) Xylan and xylan derivatives – biopoly-
mers with valuable properties. 1. Naturally occurring xylans structures,
isolation procedures and properties. Macromolecular Rapid Communica-
tions, 21, 542–556.
Enebro, J. & Karlsson, S. (2006) Improved matrix-assisted laser desorption/
ionisation time-of-flight mass spectrometry of carboxymethyl cellulose.
Rapid Communications in Mass Spectrometry, 20, 3693–3698.
Faik, A. (2010) Xylan biosynthesis: news from the grass. Plant Physiology,
153, 396–402.
Feijao, C., Morreel, K., Anders, N., Tryfona, T., Busse-Wicher, M., Kotake, T.
et al. (2021) Hydroxycinnamic acid modified xylan side chains and their
role in cell wall cross-linking in rice. The Plant Journal, 109, 1152–1167.
Goubet, F., Barton, C.J., Mortimer, J.C., Yu, X., Zhang, Z., Miles, G.P. et al.
(2009) Cell wall glucomannan in Arabidopsis is synthesised by CSLA gly-
cosyltransferases, and influences the progression of embryogenesis. The
Plant Journal, 60, 527–538.
Goubet, F., Jackson, P., Deery, M.J. & Dupree, P. (2002) Polysaccharide
analysis using carbohydrate gel electrophoresis: a method to study plant
cell wall polysaccharides and polysaccharide hydrolases. Analytical Bio-
chemistry, 300, 53–68.
Grabber, J.H., Hatfield, R.D., Ralph, J., Zo
n, J. & Amrhein, N. (1995) Ferulate
cross-linking in cell walls isolated from maize cell suspensions. Phyto-
chemistry, 40, 1077–1082.
Grabber, J.H. & Lu, F. (2007) Formation of syringyl-rich lignins in maize as
influenced by feruloylated xylans and p-coumaroylated monolignols.
Planta, 226, 741–751.
Grantham, N.J., Wurman-Rodrich, J., Terrett, O.M., Lyczakowski, J.J.,
Stott, K., Iuga, D. et al. (2017) An even pattern of xylan substitution is
critical for interaction with cellulose in plant cell walls. Nature Plants, 3,
859–865.
Gruppen, H., Kormelink, F.J.M. & Voragen, A.G.J. (1993) Water-
unextractable cell wall material from wheat flour. 3. A structural model
for arabinoxylans. Journal of Cereal Science, 18, 111–128.
Gupta, M., Rawal, T.B., Dupree, P., Smith, J.C. & Petridis, L. (2021) (2021)
spontaneous rearrangement of acetylated xylan on hydrophilic cellulose
surfaces. Cellulose, 28(6), 3327–3345.
Hatfield, R.D., Ralph, J. & Grabber, J.H. (1999) Cell wall cross-linking by fer-
ulates and diferulates in grasses. Journal of the Science of Food and
Agriculture, 79, 403–407.
Hatfield, R.D., Rancour, D.M. & Marita, J.M. (2017) Grass cell walls: a story
of cross-linking. Frontiers in Plant Science, 7, 2056.
Iiyama, K., Lam, T.B.T. & Stone, B.A. (1990) Phenolic acid bridges between
polysaccharides and lignin in wheat internodes. Phytochemistry, 29,
733–737.
Ishii, T. (1997) Structure and functions of feruloylated polysaccharides. Plant
Science, 127, 111–127.
Kang, X., Kirui, A., Dickwella Widanage, M.C., Mentink-Vigier, F., Cosgrove,
D.J. & Wang, T. (2019) Lignin-polysaccharide interactions in plant sec-
ondary cell walls revealed by solid-state NMR. Nature Communications,
10, 347.
Kozlova, L.V., Ageeva, M.V., Ibragimova, N.N. & Gorshkova, T.A. (2014)
Arrangement of mixed-linkage glucan and glucuronoarabinoxylan in
the cell walls of growing maize roots. Annals of Botany, 114, 1135–
1145.
Kozlova, L.V., Mikshina, P.V. & Gorshkova, T.A. (2012) Glucuronoarabinoxy-
lan extracted by treatment with endoxylanase from different zones of
growing maize root. Biochemistry (Moscow), 77, 395–403.
Kozlova, L.V., Nazipova, A.R., Gorshkov, O.V., Petrova, A.A. & Gorshkova,
T.A. (2020) Elongating maize root: zone-specific combinations of polysac-
charides from type I and type II primary cell walls. Scientific Reports, 10,
10956.
Kulkarni, A., Pattathil, S., Hahn, M.G., York, W.S. & O’Neill, M.A. (2012)
Comparison of arabinoxylan structure in bioenergy and model grasses.
Industrial Biotechnology, 8, 222–229.
Labourel, A., Crouch, L.I., Br
as, J.L.A., Jackson, A., Rogowski, A., Gray, J.
et al. (2016) The mechanism by which arabinoxylanases can recognize
highly decorated xylans. The Journal of Biological Chemistry, 291,
22149–22159.
Lam, T.B.-T., Iiyama, K. & Stone, B.A. (2003) Hot alkali-labile linkages in the
walls of the forage grass Phalaris aquatica and Lolium perenne and their
relation to in vitro wall digestibility. Phytochemistry, 64, 603–607.
Marcus, S.E., Blake, A.W., Benians, T.A.S., Lee, K.J.D., Poyser, C., Donald-
son, L. et al. (2010) Restricted access of proteins to mannan polysaccha-
rides in intact plant cell walls. The Plant Journal, 64, 191–203.
Margolles-Clark, E., Tenkanen, M., Nakari-Set€al€a, T. & Penttil€a, M. (1996)
Cloning of genes encoding alpha-L-arabinofuranosidase and beta-
xylosidase from Trichoderma reesei by expression in Saccharomyces
cerevisiae. Applied and Environmental Microbiology, 62, 3840–3846.
Mart
ınez-Abad, A., Berglund, J., Toriz, G., Gatenholm, P., Henriksson, G.,
Lindstr€om, M. et al. (2017) Regular motifs in xylan modulate molecular

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
16 Theodora Tryfona et al.
flexibility and interactions with cellulose surfaces. Plant Physiology, 175,
1579–1592.
Mart
ınez-Abad, A., Giummarella, N., Lawoko, M. & Vilaplana, F. (2018) Dif-
ferences in extractability under subcritical water reveal interconnected
hemicellulose and lignin recalcitrance in birch hardwoods. Green Chem-
istry, 20, 2534–2546.
Morais de Carvalho, D., Mart
ınez-Abad, A., Evtuguin, D.V., Colodette, J.L.,
Lindstr€om, M.E., Vilaplana, F. et al. (2017) Isolation and characterization
of acetylated glucuronoarabinoxylan from sugarcane bagasse and straw.
Carbohydrate Polymers, 156, 223–234.
Mueller-Harvey, I., Hartley, R.D., Harris, P.J. & Curzon, E.H. (1986) Linkage
of p-coumaroyl and feruloyl groups to cell-wall polysaccharides of barley
straw. Carbohydrate Research, 148, 71–85.
Nishimura, H., Nagata, K., Nagata, T., Katahira, M. & Watanabe, T. (2022)
Direct evidence of a ester linkages between lignin and glucuronoxylan
that reveal the robust heteropolymeric complex in plant cell walls. Nat-
ure Plants, 8, 6538.
Nyberg, N.T., Duus, J.Ø. & Sørensen, O.W. (2005) Heteronuclear two-bond
correlation: suppressing heteronuclear three-bond or higher NMR corre-
lations while enhancing two-bond correlations even for vanishing 2JCH.
Journal of the American Chemical Society, 127, 6154–6155.
Pattathil, S., Avci, U., Miller, J.S. & Hahn, M.G. (2012) Immunological
approaches to plant cell wall and biomass characterization: glycome pro-
filing. In: Himmel, M.E. (Ed.) Biomass conversion: methods and proto-
cols. Totowa, NJ: Humana Press, pp. 61–72.
Peng, W., Pedersen, N.R., Pettersson, D., Ekloff, J.M., Nymand-Grarup, S.,
Palmen, L.G. et al. (2017) Compositions comprising polypeptides having
xylanase activity and polypeptides having arabinofuranosidase activity.
US-2021332340-A1.
Pereira, C.S., Silveira, R.L., Dupree, P. & Skaf, M.S. (2017) Effects of xylan
side-chain substitutions on xylan–cellulose interactions and implications
for thermal pretreatment of cellulosic biomass. Biomacromolecules, 18,
1311–1321.
Pradhan Mitra, P. & Loqu
e, D. (2014) Histochemical staining of Arabidopsis
thaliana secondary cell wall elements. Journal of Visualized Experiments,
e51381.
Ralph, J. (2010) Hydroxycinnamates in lignification. Phytochemistry
Reviews, 9, 65–83.
Rogowski, A., Basle, A., Farinas, C.S., Solovyova, A., Mortimer, J.C.,
Dupree, P. et al. (2014) Evidence that GH115 a-glucuronidase activity,
which is required to degrade plant biomass, is dependent on conforma-
tional flexibility. The Journal of Biological Chemistry, 289, 53–64.
Ruprecht, C., Bartetzko, M.P., Senf, D., Dallabernadina, P., Boos, I., Ander-
sen, M.C.F. et al. (2017) A synthetic glycan microarray enables epitope
mapping of plant cell wall glycan-directed antibodies. Plant Physiology,
175, 1094–1104.
Ryabova, O., Vrsansk
a, M., Kaneko, S., van Zyl, W.H. & Biely, P. (2009) A
novel family of hemicellulolytic a-glucuronidase. FEBS Letters, 583,
1457–1462.
Shrestha, U.R., Smith, S., Pingali, S.V., Yang, H., Zahran, M., Breunig, L.
et al. (2019) Arabinose substitution effect on xylan rigidity and self-
aggregation. Cellulose, 26, 2267–2278.
Suzuki, K., Kitamura, S., Kato, Y. & Itoh, T. (2000) Highly substituted glu-
curonoarabinoxylans (hsGAXs) and low-branched xylans show a distinct
localization pattern in the tissues of Zea mays L. Plant and Cell Physiol-
ogy, 41, 948–959.
Tabuchi, A., Li, L.-C. & Cosgrove, D.J. (2011) Matrix solubilization and cell
wall weakening by b-expansin (group-1 allergen) from maize pollen. The
Plant Journal, 68, 546–559.
Tenkanen, M., Luonteri, E. & Teleman, A. (1996) Effect of side groups on
the action of b-xylosidase from Trichoderma reesei against substituted
xylo-oligosaccharides. FEBS Letters, 399, 303–306.
Terrett, O.M. & Dupree, P. (2019) Covalent interactions between lignin and
hemicelluloses in plant secondary cell walls. Current Opinion in Biotech-
nology, 56, 97–104.
Terrett, O.M., Lyczakowski, J.J., Yu, L., Iuga, D., Franks, W.T., Brown, S.P.
et al. (2019) Molecular architecture of softwood revealed by solid-state
NMR. Nature Communications, 10, 4978.
Tryfona, T., Liang, H.-C., Kotake, T., Tsumuraya, Y., Stephens, E. & Dupree,
P. (2012) Structural characterization of Arabidopsis leaf arabinogalactan
polysaccharides. Plant Physiology, 160, 653–666.
Tryfona, T., Sorieul, M., Feijao, C., Stott, K., Rubtsov, D.V., Anders, N. et al.
(2019) Development of an oligosaccharide library to characterise the
structural variation in glucuronoarabinoxylan in the cell walls of vegeta-
tive tissues in grasses. Biotechnology for Biofuels, 12, 109.
Tryfona, T. & Stephens, E. (2010) Analysis of carbohydrates on proteins by
offline normal-phase liquid chromatography MALDI-ToF/ToF-MS/MS.
Methods in Molecular Biology, 658, 137–151.
Urb
anikov
a, L’., Vrsansk
a, M., Mørkeberg Krogh, K.B.R., Hoff, T. & Biely, P.
(2011) Structural basis for substrate recognition by Erwinia chrysanthemi
GH30 glucuronoxylanase. The FEBS Journal, 278, 2105–2116.
Verhertbruggen, Y., Walker, J.L., Guillon, F. & Scheller, H.V. (2017) A com-
parative study of sample preparation for staining and immunodetection
of plant cell walls by light microscopy. Frontiers in Plant Science, 8,
1505.
Vilaplana, F. & Gilbert, R.G. (2010) Characterization of branched polysaccha-
rides using multiple-detection size separation techniques. Journal of
Separation Science, 33, 3537–3554.
Vranken, W.F., Boucher, W., Stevens, T.J., Fogh, R.H., Pajon, A., Llinas, M.
et al. (2005) The CCPN data model for NMR spectroscopy: development
of a software pipeline. Proteins: Structure, Function, and Bioinformatics,
59, 687–696.
Vrsansk
a, M., Kolenov
a, K., Puchart, V. & Biely, P. (2007) Mode of action of
glycoside hydrolase family5 glucuronoxylan xylanohydrolase from Erwi-
nia chrysanthemi. The FEBS Journal, 274, 1666–1677.
Wang, T., Chen, Y., Tabuchi, A., Cosgrove, D.J. & Hong, M. (2016) The tar-
get of b-expansin EXPB1 in maize cell walls from binding and solid-state
NMR studies. Plant Physiology, 172, 2107–2119.
Wang, T., Salazar, A., Zabotina, O.A. & Hong, M. (2014) Structure and
dynamics of Brachypodium primary cell wall polysaccharides from two-
dimensional 13C solid-state nuclear magnetic resonance spectroscopy.
Biochemistry, 53, 2840–2854.
Wende, G. & Fry, S.C. (1997) 2-O-b-D-xylopyranosyl-(5-O-feruloyl)-l-
arabinose, a widespread component of grass cell walls. Phytochemistry,
44, 1019–1030.
Wood, P.J. (1980) Specificity in the interaction of direct dyes with polysac-
charides. Carbohydrate Research, 85, 271–287.
Xue, J., Bosch, M. & Knox, J.P. (2013) Heterogeneity and glycan masking of
cell wall microstructures in the stems of Miscanthus 9 giganteus, and its
parents M. sinensis and M. sacchariflorus. PLoS One, 8, e82114.
Zeng, W., Jiang, N., Nadella, R., Killen, T.L., Nadella, V. & Faik, A. (2010) A
glucurono(arabino)xylan synthase complex from wheat contains mem-
bers of the GT43, GT47, and GT75 families and functions cooperatively.
Plant Physiology, 154, 78–97.
Zhong, R., Lee, C., Cui, D., Phillips, D.R., Adams, E.R., Jeong, H.Y. et al.
(2022) Identification of xylan arabinosyl 2-O-xylosyltransferases catalyz-
ing the addition of 2-O-xylosyl residue onto arabinosyl side chains of
xylan in grass species. The Plant Journal, 112, 193–206.

 2023 The Authors.
The Plant Journal published by Society for Experimental Biology and John Wiley & Sons Ltd.,
The Plant Journal, (2023), doi: 10.1111/tpj.16096
Grass xylan structural variation 17