Immunoprofiling reveals unique cell-specific patterns of wall epitopes in the expanding Arabidopsis stem


  • Hardy C. Hall,

    1. Michael Smith Laboratories, University of British Columbia, 2185 East Mall, Vancouver, BC, Canada
    Current affiliation:
    1. Umeå Plant Science Centre, Swedish University of Agricultural Sciences, Umeå, Sweden
    Search for more papers by this author
  • Jingling Cheung,

    1. Department of Botany and Michael Smith Laboratories, University of British Columbia, Vancouver, BC, Canada
    Search for more papers by this author
  • Brian E. Ellis

    Corresponding author
    1. Department of Botany and Michael Smith Laboratories, University of British Columbia, Vancouver, BC, Canada
    • Michael Smith Laboratories, University of British Columbia, 2185 East Mall, Vancouver, BC, Canada
    Search for more papers by this author

For correspondence (e-mail


The Arabidopsis inflorescence stem undergoes rapid directional growth, requiring massive axial cell-wall extension in all its tissues, but, at maturity, these tissues are composed of cell types that exhibit markedly different cell-wall structures. It is not clear whether the cell-wall compositions of these cell types diverge rapidly following axial growth cessation, or whether compositional divergence occurs at earlier stages in differentiation, despite the common requirement for cell-wall extensibility. To examine this question, seven cell types were assayed for the abundance and distribution of 18 major cell-wall glycan classes at three developmental stages along the developing inflorescence stem, using a high-throughput immunolabelling strategy. These stages represent a phase of juvenile growth, a phase displaying the maximum rate of stem extension, and a phase in which extension growth is ceasing. The immunolabelling patterns detected demonstrate that the cell-wall composition of most stem tissues undergoes pronounced changes both during and after rapid extension growth. Hierarchical clustering of the immunolabelling signals identified cell-specific binding patterns for some antibodies, including a sub-group of arabinogalactan side chain-directed antibodies whose epitope targets are specifically associated with the inter-fascicular fibre region during the rapid cell expansion phase. The data reveal dynamic, cell type-specific changes in cell-wall chemistry across diverse cell types during cell-wall expansion and maturation in the Arabidopsis inflorescence stem, and highlight the paradox between this structural diversity and the uniform anisotropic cell expansion taking place across all tissues during stem growth.


Anisotropic cell-wall expansion is a process of fundamental importance to plant growth and development, and is integral to establishing plant organ shape. The properties of cellulose microfibrils are central to the ability of the cell to generate the differences in tensile strength among its walls that are necessary for controlled directional cell expansion (Fujita et al., 2011). However, many other cell-wall carbohydrate constituents, such as cross-linking glycans and glycoproteins, interact with the cellulose microfibrils to condition the degree of cell-wall extensibility and to affect the direction of cell growth. Non-enzymatic and enzymatic modifications to these components dynamically alter the way in which they interact with cellulose (Cosgrove, 1997).

The developing Arabidopsis inflorescence stem provides an experimentally accessible model for studying the coordinated interaction of diverse tissue types during a period of dramatic, directional organ elongation. Although there are various functions and structural endpoints for cell walls across the various tissue types within this organ, the cells within these tissues must elongate in a unified fashion as the stem grows. One model for such coordinated growth is that the lateral (expanding) walls of the various cell types all share a relatively invariant composition until elongation ceases, at which point the differentiation program that specifies individual tissues triggers more specialized processes, such as secondary cell-wall formation. Alternatively, specialized cell types may be envisaged to diverge in wall structure/composition much earlier, during their elongation phase, but still display matched, cell-wall expansion behaviour despite these different wall properties.

To obtain insight into the apparent contradiction between uniform, anisotropic cell expansion characteristics and the diversity of structural endpoints, we have probed the composition of the cell walls of a range of inflorescence stem cell types at distinct stages in their elongative development, using a large panel of immunochemical reagents. Monoclonal antibodies (mAb) raised against cell-wall components are expected to be specific to a particular epitope structure and are predicted to bind their corresponding epitope targets in planta in a 1:1 fashion. Through indirect fluorescent labelling approaches, such mAbs provide quantifiable, cell type-specific and subcellular localization information for epitopes in plant cell walls. An extensive array of cell wall-directed mAbs is now publicly available (Pattathil et al., 2010), and these allow simultaneous assessment of many cell-wall constituents across multiple cell types in replicate tissue sections. This general strategy has been used in a limited number of recent studies, including use of 27 different antibodies to examine sub-regions of G–fibres in the tension wood of sweetgum (Liquidambar stryraciflua; Hamamelidaceae) (Bowling and Vaughn, 2008), and use of a panel of 24 antibodies to explore cell-wall composition within the phloem fibres of hemp plants (Cannabis sativa L.) at three regions along the primary stem (Blake et al., 2008).

Here, we present a more comprehensive view of epitope abundance for 55 carbohydrate-directed antibodies in the developing inflorescence stem of Arabidopsis thaliana using tissues from three experimentally defined growth stages along a continuum of elongative development identified by growth kinematic profiling (Hall and Ellis, 2012). The resulting growth stage-specific immunolabelling micrographs enabled us to (i) highlight major shifts in epitope abundance/accessibility across developmentally distinct tissues, (ii) identify similarly behaving antibodies that displayed parallel reporting patterns, and (iii) detect novel cell type-specific patterns of immunolabelling. The complete set of micrograph images is available as a public research resource (


High-throughput stage-specific epitope profiling

Regions representing distinct elongative developmental states in the flowering stem were identified by growth kinematic profiling of individual Arabidopsis plants, a procedure that establishes relative growth rates for segments of individual stems and thereby permits direct comparison of developmentally matched tissues (Hall and Ellis, 2012). This sampling approach enabled us to perform comparative immunohistochemical analysis of stem sections representing three stages of elongative development: (i) an apical region in which directional cell growth is initializing (termed ‘young’/'YNG'), (ii) a region in which this directional growth is most rapid (termed ‘maximum growth rate’/'MGR'), and (iii) a region in which such elongation is ceasing (termed ‘cessation’/'CSS') (Figure S1).

The use of an Alexa-Fluor™ pre-labelled secondary antibody whose chromophore emits in a spectral range distinct from that of the Congo Red xylan/cellulose-specific counter-stain (Wood, 1980; Anderson et al., 2009) allowed us to generate two distinct signal channels for each immunolabelled sample. The resulting confocal image stacks provided the morphological context that is necessary for accurate tissue identification (561 nm emission channel) while permitting observation of immunofluorescence intensity and distribution (488 nm channel). These images were assembled into antibody-specific views for browsing (interim access:, user:, password: 1mMuN01) and subsequent analysis of ‘regions of interest’. For each antibody, the immunofluorescence intensity was then quantified in seven cell types (Figure 1), yielding immunolabelling profiles for 55 antibodies across seven stem tissues in three biological replicate plants, each sampled at three developmental stages.

Figure 1.

Representative developmental series for fluorescence intensity scoring. A developmental series for the JIM11 antibody (antigen = extensin and/or potato lectin) in a single primary flowering stem at YNG (a, b), MGR (c, d) and CSS (e, f) stages. Tissues were identified from maximum-projection overlays of the Congo Red counter-stain (blue) channel with the Alexa Fluor™ antibody fluorescence (green) channel (a, c, e). Ep, epidermis; Co, cortex; En, endodermis; IFF, inter-fascicular fibre region; Pi, pith; Ph, phloem; Xy, xylem. The antibody-only channel (b, d, f) was examined separately to quantify fluorescence signals of seven cell types and background (Bk) for each 40 μm section.

Similarity of epitope patterning among cell wall-directed antibodies at three stages of stem development

Hierarchical clustering (HCL) may be used to reveal similarities in epitope abundance patterning produced by antibodies at each stage of elongative development. For a given antibody, we used the background-corrected, mean fluorescence ‘grey value’ intensities (Ībc) measured in the seven tissues of interest across biological replicate sections to establish a pattern, or signature, that reflects that antibody's performance (tissue response) at a given stage of development. This pattern may be described in terms of relative fluorescence intensity scores (Īrel) for each of the seven tissues, and thus represents the distribution of signal across the tissue. This quantitative measure of fluorescence distribution may be used as the basis for grouping antibodies that behave similarly at a particular developmental stage, which presumably reflects epitope co-occurrence. In addition, the 55 mean Īrel scores, one for each antibody, may be correlated with tissue type, thereby providing a measure of the similarity of epitope composition among tissue types. In this analytical approach, which addresses the three developmental stages (YNG, MGR and CSS) individually, we used HCL methods to assess the degree of similarity among (i) antibodies, based on their associated Īrel scores for the seven tissue types, and (ii) tissue types, based on their associated relative abundance scores for all antibodies, thus grouping ‘like antibodies’ and ‘like tissues’, respectively.

For this purpose, antibodies were grouped according to binding pattern similarity by use of HCL, and then organized in dendrogram form (vertical axis), while a heatmap displays their Īrel scores on a nine-level colorimetric scale for the seven cell types (Figures 2-4 for the YNG, MGR and CSS stages, respectively). Tissue types were also grouped according to their similarity in Īrel scores to yield a two-axis clustering view that provides the clearest presentation of biologically relevant patterns. Mean grey value scores (Ībc, see Experimental Procedures) indicate overall antibody signal abundance.

Figure 2.

Cluster analysis in YNG segments of relative fluorescence intensities associated with 55 antibodies. Hierarchical clustering of both antibodies (vertical axis) and tissue types (horizontal axis) based upon relative mean antibody fluorescence. Numerical values within heatmap cells indicate mean intensities used to derive relative intensity clustering values. Horizontal dotted lines indicate major clade divisions, which are numbered on the left. Cell colours representing relative intensities are restricted to nine levels. The histogram (inset, top left) indicates the frequency of relative intensity values (nine levels).

Figure 3.

Cluster analysis in MGR segments of relative fluorescence intensities associated with 55 antibodies. Details as in Figure 2.

Figure 4.

Cluster analysis in CSS segments of relative fluorescence intensities associated with 55 antibodies. Details as in Figure 2.

Many of the antibodies used in this study had been earlier assigned to specified groupings (clades) based on similarities in their ability to bind particular cell-wall glycans in vitro (Pattathil et al., 2010). To place our in planta analysis in the context of these higher-order epitope classes, clade assignments for the featured antibodies are provided (Figures 2-4, right margin). Additional published information on the target specificity of many of the antibodies used here is presented in Table S1, and is also provided at WallBioNet (glycomics.ccrc.uga.eduwall2index.html). We will first briefly discuss the clustering results for each developmental stage, followed by a discussion of the important overall co-abundance groupings.

‘Young’ (YNG) growth-stage clustering of epitope binding patterns

Two-axis clustering on the basis of similarity of relative abundance scores for the 55 antibodies over seven tissue types in ‘young’ (YNG) stem samples yielded six distinct clusters (Figure 2). With minor exceptions, these may be characterized as cluster #1, whose signal is dominant in the vascular tissue; cluster #2, whose signal is dominant in the peripheral tissues (cortex and epidermis); cluster #3, for which the endodermis signal is dominant to those of the phloem and the inter-fascicular fibre (IFF) region; cluster #4, for which the signal is xylem-dominant but also present in other tissues; cluster #5, antibodies with either widely distributed signal and/or low intensity signal scores across all tissues. In addition, two antibodies (JIM5 and CCRC–M34) appear as clustering outgroups whose signals are highly specific to particular tissues: JIM5 is specific to the protoxylem and CCRC–M34 is specific to the cortex (Figure S2). Both antibodies have been previously characterized as recognizing partially methyl-esterified homogalacturonan (HG) (Clausen et al., 2003; Verhertbruggen et al., 2009; (glycomics.ccrc.uga.eduwall2index.html), and were therefore placed in the ‘pectic backbone’ clade by Pattathil et al. (2010). The unique Īrel distribution patterns for the JIM5 and CCRC–M34 signals suggest that partially methyl-esterified forms of HG are deposited in a very spatially specific manner during the early stages of stem extension.

The tissue type-specific clustering analysis in YNG stem samples identified the phloem and xylem as being most similar in terms of overall antibody binding profile, while the epidermis formed an outgroup. The parenchymal tissues are less clearly distinguished from each other at this growth stage, and no strong patterning is evident in glycan-specific clade assignments (Figure 2, right column). The occasional occurrence of some homogeneous groups within this clustering may reflect epitope co-occurrence or cross-reactivity.

‘Maximum growth-rate’ (MGR) growth-stage clustering

In the MGR-stage stem samples, all cell types are expected to be expanding anisotropically in unison, a condition that requires that the axial cell walls are capable of rapid extension. Two-axis HCL of the MGR antibody binding patterns yielded seven well-resolved clusters of patterns (Figure 3): cluster #1 (an outgroup), antibodies whose signals are dominant in the IFF region; cluster #2, antibodies whose signals are dominant in the cortex; cluster #3, antibodies whose signals are largely absent from the IFF region; cluster #4, antibodies whose signals are most prominent in the endodermis but also present in adjacent interior tissues of phloem and the IFF region; cluster #5, antibodies whose signals are dominant in the pith but absent in the endodermis; cluster #6, antibodies whose signals are present in the epidermis and either of the cortex or pith; cluster #7, antibodies whose signals are widespread and/or low or absent in all tissues. The high signal abundance of JIM4 (‘AG–3' clade) in the pith may account for its occurrence as an outgroup for the common branch to clusters #3–7 (Figure 3).

In terms of binding pattern similarity across tissues, the IFF region in the MGR samples forms a prominent outgroup, but, in contrast to the YNG stem sections, phloem and xylem in the MGR samples no longer appear most similar to each other, and no clear parenchymal grouping is observed.

The glycan clade distribution for clusters at the MGR stage (Figure 3, right column) presents a very different picture to that observed in YNG tissue. Glycan clade-specific antibodies now form homogeneous groups of up to seven members (Figure 3, cluster #4), and groups of three or four members of the same super-clade are common (Figure 3, clusters #3, 4 and 6). Six of twelve ‘RG–I/AG’ clade members examined in this study are tightly sub-clustered (Figure 3, cluster #5) and exhibit their highest signal in the pith, while ‘AG–1’ and ‘AG–2’ cluster together but separately from ‘AG–3’ in most but not all cases.

‘Cessation’ (CSS) growth-stage clustering

At the point of cessation of extension growth in the stem, primary walls in all tissues are predicted to have discontinued anisotropic expansion, and secondary cell-wall formation (thickening and rigidification) is proceeding on the walls of xylary and IFF cells. The antibody binding patterns at this growth stage appear to group in six distinct clusters (Figure 4): cluster #1 (an outgroup), antibodies whose signals are dominant in the IFF region and xylem; cluster #2, antibodies whose signals are dominant in the peripheral tissue of the cortex and/or the epidermis; cluster #3, antibodies whose signals are present in the phloem but also in the adjacent IFF region and endodermis; cluster #4, antibodies whose signals are dominant in the pith and/or highly abundant in all tissues; cluster #5, antibodies whose signals are most prominent in the fibre-possessing tissue of the IFF region and the xylem; cluster #6, antibodies with ubiquitous and/or low abundance across tissues. Notable outgroups to the main clusters #2–6 are the sub-clustered antibodies JIM93 and JIM94, both of which have been previously assigned to the ‘AG–1’ clade (Pattathil et al., 2010). JIM93 and JIM94 signal patterns were also found to sub-cluster in the MGR-stage tissues, together with other ‘AG–1’ and ‘AG–2’ clade members (Figure 3, cluster #4), but, interestingly, only JIM94 signals showed appreciable abundance in YNG tissue (Figure 2, cluster #1). The structures of the epitope(s) recognized by JIM94 and JIM93 are unknown, but the differential behaviour of the two antibodies in YNG tissue suggests that their epitopes must occur at independent sites of common or different antigens in plant cell walls.

As observed for the MGR growth stage, the IFF region in CSS sections again forms a pronounced immunolabelling outgroup among the various tissue types, with the cortex being separated from other tissues to a lesser degree. While the MGR-stage binding patterns reveal the largest homogeneous groupings of antibody clade members, the distribution of ‘RG–I/AG’ clade member signal patterns remains concentrated within a few HCL clusters in the CSS tissues (Figure 4, clusters #4 and 6), as do certain core groupings of ‘AG–3/4’ clade members (Figure 4, cluster #1).

Immunodetection of early stages of secondary cell-wall deposition

One of the most striking patterns of antibody binding across the three inflorescence stem growth stages was that displayed by the ‘AG–3’ glycan clade members JIM8, JIM15 and PN16.4B4, together with an ‘AG–4’ clade member, JIM13 (Figures 2-4, cluster #1). In YNG tissues, the signals for all four antibodies are restricted to the vessel elements of the protoxylem and to a small group of cells in the protophloem. Their signals then appear within the developing inter-fascicular fibre region during the phase of most rapid stem elongation (MGR), and they are found in all tissues by the CSS stage. Interestingly, the signal associated with these four antibodies appears limited to the inner cell walls, as opposed to the outer cell walls, which stain exclusively with Congo Red (Figure 5). These inner layers also appear to be physically distinct, as they tend to separate from the outer layers during transverse sectioning.

Figure 5.

Immunofluorescent labelling of YNG (left), MGR (centre) and CSS (right) tissue with antibodies identified by hierarchical clustering as specific to inner walls of rapidly expanding fibres. Maximum-projection, two-channel views of arabinogalactan-specific JIM8 (a–c), JIM13 (d–f), JIM15 (g–i) and PN16.4B4 (j–l). Antibodies that bind later in development (CSS) were added for comparison, and include JIM84 (m–o), which is also arabinogalactan-specific, and xylan-specific LM10 (p–r). The Congo Red counter-stain channel is represented as blue, and antibody fluorescence is represented as green. All images are for biological replicate 1 (BR1). See supplemental image series of these antibodies to view all biological replicate images as well as view the antibody-specific channel independently (

As JIM8, 13 and 15 have been reported to display a high degree of cross-reactivity in vitro (Pattathil et al., 2010) (PN16.4B4 was not included in that study), they may recognize the same or similar structures, but the binding patterns of JIM13 place it in a different glycan-specific clade (‘AG–4’), implying that it may be recognizing a class of arabinogalactan side-chain distinct from that bound by the three members of the ‘AG–3’ clade. The association of epitopes recognized by these four antibodies with sites of secondary cell-wall deposition, even in younger, rapidly elongating tissues, implies that the presence of these epitopes may be diagnostic for the earliest stages of secondary cell-wall formation.

At the CSS stage, two additional antibodies appear to be tightly clustered with this same group (Figure 4, cluster #1): JIM84 (‘AG–3’ clade) and LM10 (‘xylan’ clade), which is known to target low methyl-substituted xylans in secondary walls (glycomics.ccrc.uga.eduwall2index.html). JIM84 was raised against a carrot (Daucus carotus) ‘coated vesicle’ preparation, and has been observed to bind Golgi bodies as well as the plasma membrane in many cell types; biochemical analysis suggests that the protein portion of a glycoprotein is the likely epitope for this antibody (Horsley et al., 1993). Thus, localization of both JIM84 and LM10 to IFF cell walls at the CSS growth stage may indicate co-association of one or more arabinogalactan glycoprotein(s) with unsubstituted xylan in cells whose extension growth has ceased.

Some epitopes are unique to the endodermis

In maturing stems (MGR and CSS growth stages), a discrete subset of antibodies show pronounced localization to the endodermis, a layer of cells that is morphologically distinct from both the neighbouring cortical cells and the inter-fascicular fibres (Figure 6). While the endodermis is found in aerial plant organs of many species (Lersten, 1997), our knowledge of its biological role in Arabidopsis is limited to its participation in regulation of the shoot gravitropic response (Fukaki et al., 1998). At the MGR growth stage, the ‘endodermis-specific’ antibody cluster includes JIM11 (‘AG–1’ clade), JIM19 (‘AG–2’ clade), JIM20 (‘AG–1’ clade), MAC204 (‘AG–1’ clade) and JIM12 (‘AG–2’ clade) (Figure 3, cluster #4). As observed with other AG clade members surveyed, labelling generated by members of this ‘endodermis cluster’ is restricted to the inner wall of particular cells, which may indicate that the antibodies only label endodermis cells cut open by the sectioning process.

Figure 6.

Immunofluorescent labelling of YNG, MGR and CSS tissue with antibodies identified by hierarchical clustering as abundant within the endodermis at MGR and CSS stages. Maximum-projection, two-channel views of arabinogalactan-specific JIM11 (a–c), JIM19 (d–f), JIM20 (g–i) and MAC204 (j–l) at three stages of development: YNG, MGR and CSS. The Congo Red counter-stain channel is represented as blue, and antibody fluorescence is represented as green. All images are for biological replicate 1 (BR1), chosen arbitrarily. Antibodies sub-clustered at both the MGR stage (cluster #4) and the CSS stage (cluster #3). JIM12 (m–o) is included as it sub-clusters with JIM11, JIM19, JIM20 and MAC204 at the MGR stage. See supplemental image series of these antibodies to view all biological replicate images as well as view the antibody-specific channel independently.

Pectin methyl-esterification trajectories

One widely accepted model of cell-wall extensibility proposes that this property is controlled, in part, by the degree of methyl-esterification of the acidic side chains on the pectic HG backbone (Kim and Carpita, 1992; Derbyshire et al., 2007; Siedlecka et al., 2008; Hongo et al., 2012), with a lower abundance of methyl ester groups being associated with more extensive calcium ion cross-polymer bridging and general rigidification of the wall. Thus, reduced levels of pectin methyl-esterification have been correlated with reduced cell-wall expansion in Arabidopsis hypocotyls, for which 60% pectin methyl-esterification has been shown to be sufficient to support normal elongative growth (Derbyshire et al., 2007). The degree of methyl-esterification of deposited, methylated homogalacturonans is thought to be controlled largely by the amount of active pectin methyl esterase secreted to the cell wall (Micheli, 2001), and enhancement of methyl esterase activity in transgenic aspen (Populus tremula) stems was shown to result in reduced cell expansion during secondary xylem formation (Siedlecka et al., 2008). In Arabidopsis, mutation of AtPME35 led to elevated binding of JIM7 (low degree of methyl-esterification, DM) in more mature regions of the inflorescence stems, but did not affect the extent of binding of JIM5 (intermediate DM) (Hongo et al., 2012).

Among the stage-specific hierarchical clusters detected in this study is a group of ‘pectic backbone’ clade members whose signals are distributed ubiquitously across various tissues at the MGR growth stage (Figure 3, cluster #3), but then become strongly localized to the IFF region and xylem at the CSS stage (Figure 4, cluster #5). This group includes LM18 and LM19, which are reported to bind de-esterified HG, and LM20, which exhibits bias for highly methyl-esterified HG (Verhertbruggen et al., 2009). In addition, we assayed four other mAbs (JIM5, JIM7, CCRC–M34 and CCRC–M38) that were previously assigned to the ‘pectic backbone’ HG clade described by Pattathil et al. (2010), and whose affinity for pectin varies with the degree of pectin methyl-esterification. JIM5 and JIM7 epitopes have been described as partially methyl-esterified HG (Knox et al., 1990; Verhertbruggen et al., 2009) CCRC–M34 and CCRC–M38 are reported in the WallBioNet database (glycomics.ccrc.uga.eduwall2index.html) to bind partially and fully de-esterified pectins, respectively (Pattathil et al., 2010).

To examine the relationship between anisotropic growth and pectin methyl-esterification, we compared normalized, absolute signal intensities (Īn.abs) for these seven antibodies in specific tissues at the three sampled growth stages (Figure S2). By calculating the sum of intensities across all seven tissues at each growth stage, we were also able to generate a summary view of the overall pattern of methyl-esterification within the entire stem segment(s). This immunolabelling-derived view is analogous to the DM data derived from whole segments using either chemical or biochemical assays of uronic acid methyl ester abundance (Kim and Carpita, 1992; Hongo et al., 2012).

Our immunoprofiling analysis revealed that there is indeed an overall decline in abundance of highly methyl-esterified HG epitopes as the stem tissues mature, accompanied by a linear increase in signals from epitopes with low degree of methyl-esterification. This pattern is clearest in the epidermis, which exhibits a steady decline in higher-DM epitopes (LM20 and JIM7) as the stem matures, accompanied by uniform increases in the abundance of lower-DM epitopes recognized by JIM5, CCRC–M32, LM18 and LM19). Overall, this pattern is consistent with the hypothesis that one factor leading to reduced cell-wall extensibility is lower levels of pectin methyl-esterification. However, this conformity is not consistent across all the tissues assayed, as the inter-fascicular fibres, for example, do not show a similar pattern of maturity-related de-methyl-esterification.

In addition to the tissue specificity displayed by these ‘methyl-esterified’ antibodies, we found evidence for even more localized variation in signal distribution (7). One striking pattern that is common to JIM5 as well as LM18 and 19 is a peripheral zone of absence of signal in the phloem, resulting in a pronounced ‘halo effect’ in the CSS-stage vascular tissue (7). However, within the pith, JIM5 and LM19 signals are localized specifically to the middle lamella in three-cell intersection points at both the YNG and MGR stages (7). Regions of appressed cell walls between these three-cell junction points (i.e. the two-cell boundaries) subsequently display strong JIM5/LM19 signals at the CSS stage.


The publicly available library of plant cell wall-directed monoclonal antibodies provides a powerful tool to assess the cellular location and relative abundance of many important classes of wall-associated glycans, including pectins, glycoproteins and cross-linking xylans (hemicellulose). Our analysis of the binding patterns of a large set of these antibodies in Arabidopsis inflorescence stem sections sampled at experimentally defined developmental stages provides a remarkably detailed view of the spatial and temporal changes in cell-wall chemistry that accompany the transition of stem tissue from early cell expansion, through a period of maximal extension growth, to full stem maturation. The changes in epitope presence and patterning observed along the stem development continuum appear to reflect the existence of a highly dynamic and tissue-specific wall structure whose physicochemical properties are being constantly modified as different cells pass through sequential phases of directional cell-wall expansion, while also executing points their cell-fate trajectories. These data indicate that, within the constraints of the set of chosen mAb probes, there is little evidence for large-scale commonality in cell-wall chemistry among cells that are following different developmental trajectories but concurrently undergoing rapid anisotropic expansion.

Probes for early stages of fibre cell-wall deposition

One of the interesting patterns detected in this study is the spatiotemporal co-occurrence of signals specific to discrete sets of cell-wall antibodies. A prominent example of this involves three ‘AG–3’ glycan clade members (JIM13, JIM8 and PN16.4B4), which appear to recognize a set of epitopes specific to the developing inter-fascicular fibres at the MGR and CSS growth stages (Figures 3 and 4). While the JIM8 and PN16.4B4 signals are strongest at the MGR and CSS stages, the onset of JIM13 signals at the earlier YNG stage suggests that JIM13 targets an arabinogalactan-bearing cell-wall component that is distinct from those targeted by JIM8 and PN16.4B4, and that may be associated with the early stages of fibre cell-wall development. The JIM13 antibody was raised against a preparation of ARABINOGALACTAN PROTEIN 2 (AGP2) and is known to bind its glycan component, (β)GlcA1→3(α)GalA1→Rha (Yates et al., 1996). Consistent with this, the distribution of the JIM13 signal correlates with the behaviour of the AGP2 gene, which is expressed early in the stem elongation process and whose expression is sustained until cessation of elongation (Hall and Ellis, 2013).

The epitope specificity of the JIM8 and PN16.4B4 antibodies is less clear, as they were raised against different constituents of suspension-cultured cells of two species: whole protoplasts of Beta vulgaris (sugar beet) (Pennell et al., 1991) and cell membranes of tobacco (Nicotiana glutinosa) (Norman et al., 1986), respectively. Antibodies assigned to the ‘AG–3’ clade have previously been described as ‘AGP-specific’ (Moller et al., 2008; Pattathil et al., 2010), and it is possible that JIM8 and PN16.4B4 target one or more of the 40 known Arabidopsis AGPs (Knox, 2005). Six genes encoding AGPs (AGP12, 13, 14, 21, 22 and 24) have been found to be significantly up-regulated in MGR-stage stem tissues compared with the YNG stage (Hall and Ellis, 2013), and expression of AGP12 and 13 peaked in MGR and CSS tissues, respectively. However, it should be noted that seven other ‘AG–3’ clade member antibodies examined in this study did not share this tissue specificity. Even in the absence of more detailed knowledge of JIM13 and PN16.4B4 targets, the present data suggest the existence of a unique class of epitopes, possibly associated with AGP12 and/or AGP13, whose occurrence is specific to early stages of inner cell-wall thickening, and that start to accumulate well before formal cessation of elongation when secondary cell-wall formation is generally thought to be active.

Probes for endodermis cell walls

Another striking antibody cluster includes those whose signals are concentrated to varying degrees around the endodermis: JIM11 (‘AG–1’ clade), JIM19 (‘AG–2’ clade), JIM20 (‘AG–1’ clade), MAC204 (‘AG–1’ clade) and JIM12 (‘AG–2’ clade) (Figure 6). JIM11, JIM12, JIM19 and JIM20 all recognize putative extensin antigens (Smallwood et al., 1994; Knox et al., 1995), and bind very strongly both to the endodermis and to IFF cell walls. However, JIM11 is unique within this group in that it does not have targets in the IFF region at the CSS growth stage; indeed, the specificity exhibited by JIM11, whose signal appears almost exclusively within the endodermis across all three developmental stages (Figure 6a–c), is remarkable. JIM11 was previously reported to label regions of the carrot (Daucus carotus) root pericycle (Smallwood et al., 1994), a tissue that is thought to be developmentally analogous to the endodermis of aerial organs (Lersten, 1997). The exact biochemical target of JIM11 has not been defined, but the available data suggest that JIM11 is specific to the oligosaccharide portion of either extensin or lectins (Smallwood et al., 1994). Extensins have been proposed to be involved in cell-wall assembly (Humphrey et al., 2007), possibly through their propensity to cross-link with pectins and thereby act as cell-wall stabilizers (Lamport et al., 2011), especially within cells that do not undergo appreciable fortification by lignification.

Probes for pectin across development

Hierarchical clustering of the signal patterns for the 55 antibodies revealed remarkable heterogeneity in the signal distribution associated with HG-directed epitopes, which vary both between tissues (spatially) and over the course of stem development (temporally). For example, while JIM5 and CCRC–M34 signals co-occur in YNG-stage tissues (Figure 2, cluster #1), this spatial relationship did not persist at later growth stages. In fact, no pectin antibodies sub-clustered together across all three growth stages (Figures 2-4).

The divergent labelling displayed by pectin-related antibodies may be related, in part, to their differing affinities for HG with various DM, but even mAbs that are described as binding preferentially to pectins with low DM, such as JIM5, LM18 and LM19, did not routinely label the same sites within stem tissue sections. Adding to this complexity, signal patterns in the MGR and CSS growth stages reveal juxtaposition of some antibodies that differ markedly in their reported DM affinities (Figures 3, 4 and 7).

Figure 7.

Immunofluorescent labelling of YNG, MGR and CSS tissue with antibodies identified by hierarchical clustering as specific to xylem and IFF region at the CSS stage. Maximum-projection, two-channel views of arabinogalactan-specific LM20 (a–c), JIM7 (d–f), JIM5 (g–i), CCRC–M34 (j–l), LM18 (m–o) and LM19 (p–r) at three stages of development: YNG, MGR, and CSS. The Congo Red counter-stain channel is represented as blue, and antibody fluorescence is represented as green.

Despite the complex antibody-specific patterns revealed by clustering analysis, the pattern of absolute signal intensities summed over all tissues (Figure S2) provides broad support to the ‘low DM promotes rigidification’ hypothesis (Kim and Carpita, 1992; Derbyshire et al., 2007; Siedlecka et al., 2008; Hongo et al., 2012). Nevertheless, it is clear that, at the level of individual tissues, the model is more relevant for some cell types (e.g. epidermis) than for others (e.g. inter-fascicular fibres).

The lack of noticeable correlation between DM and maturity among cell types other than the epidermis does not necessarily preclude a role for pectin-DM in regulating cell-wall development in those cell types. The observation that epitopes of varying DM are specifically localized at discrete subcellular regions of the pith and cortical cells (Figure 7) may be indicative of the need to maintain a balance between cell-wall flexibility (high DM) and rigidity (low DM) during turgor-driven cell separation. Similar punctate immunolabelling by JIM5 (lower DM) and JIM7 (higher DM) has been reported in cortical cells of pea (Pisum sativum L.) (VandenBosch et al., 1989; Willats et al., 2001) and tobacco (Nicotiana tabacum L.) (Verhertbruggen et al., 2009).

Although this study has generated a wealth of novel information about patterns of cell-wall composition during and after rapid cell elongation, knowledge of the precise structural binding determinants of many of the available antibodies is still limited (Clausen et al., 2003; Moller et al., 2008; Verhertbruggen et al., 2009; Pattathil et al., 2010) In addition, changes in the accessibility of particular antigenic site(s) to an antibody (i.e. ‘epitope masking’) may potentially confound interpretation of immunolabelling data (Marcus et al., 2008). It is therefore necessary to use caution when drawing conclusions about the absolute abundance of specific polysaccharides in a particular tissue sample. However, the observed immunolabelling trends between growth stages (YNG, MGR and CSS), and the remarkably diverse signal distribution patterns amongst tissue types, together provide a wide range of potentially important leads for more detailed investigation of how the chemistry of plant cell walls is modified between stages of cell-wall differentiation in a rapidly expanding plant organ.

Experimental Procedures

Plant material and growth conditions

Cold-treated Columbia (Col–0) seeds were sown in 32-plug tray inserts with soil-less potting mix (Sunshine Mix #5, Sun Gro Horticulture Canada Ltd, ( supplemented with liquid fertilizer Plant-Prod 20N/20P/20K (soluble fertilizer (Plant Products Co. Ltd, (, then grown under short-day conditions (8 h light, 21°C/16 h dark, 19°C) for 6 weeks. To induce bolting, plants were transferred to long-day conditions (16 h light, 21°C/8 h dark, 21°C) until the inflorescence reached a height of 10–15 cm. Plants were then removed from the growth chamber for image analysis and harvesting.

Growth kinematic profiling

Developmentally specific tissues for young (YNG), maximum growth rate (MGR) and cessation (CSS) tissues were isolated essentially as described by Hall and Ellis (2012), utilizing 1 mm diameter disks cut from 45 μm polyvinylidene fluoride membrane (IPVH 00010, Millipore, ( using a CTR 6500 laser microdissection system (Leica, ( as optical markers. These were applied to the stem at 5 mm intervals via electrostatic charge by holding in close proximity to the stem with forceps. Feature tracking (ImageJ, was performed as described by Hall and Ellis (2012). Specifically, an oval of fixed dimension (red) is best-fitted by eye for each disk through the time series. The pixelated nature of the shape in ImageJ allows tracking of left and right edges of the oval through the series as described by Hall and Ellis (2012). Growth kinematic profiles of relative growth rates for marker-defined segments were used to identify points of maximum growth rate and cessation for three biological replicate plants, and are shown in Figure S1.

Tissue harvesting for microscopic analysis

Stem segments between optical markers were harvested from individual plants in sequence from top to bottom. Upon excision, segments were immersed in 150 μl fixation buffer (stock 2× PME: 50 mm PIPES, 2 mm MgSO4, 2 mm EGTA), within 0.2 ml dome-cap thermal cycler tubes (Thermo Scientific, ( Segments were then subjected to three consecutive 21°C cycles of 5 min vacuum infiltration at 68 kPa, and washed three times in 1× PME (21°C, 68kPa) prior to long-term storage at 4°C in 1× PME. Segments were individually encased in 1 cm3 blocks of 5% agar at 65°C, and stored at 4°C to set. Transverse sections (40 μm thick) were cut from segments using a VT100S vibrating microtome (Leica), separated from agar encasement using a sable hair (‘00’) brush, then blocked for at least 1 h in 5% bovine serum albumin in 1× TBST (10 mm Tris, 0.25 m NaCl, 0.1% Tween).

Immunolabelling and staining

Sections were mixed to randomize developmental difference, and randomly allocated from each biological replicate pool, together with 100 μl fresh blocking solution, to wells of a 96-well plate (BD Falcon, (, grouping three biological replicates row-wise for each antibody or negative control. Antibodies included in this experiment are listed in Table S1, which indicates 96-well plate position, source animal, supplier, immunogen, epitope structure and antigen (information derived from the Complex Carbohydrates Research Center database WallBioNet, ( These 55 mAbs collectively provide broad coverage of the major cell-wall glycan classes relative to the coverage offered by the entire set of available cell-wall glycan-directed antibodies (Figure S3), most of which have been previously described in the literature. Blocking solutions were swapped with 15 μl 1:36 dilutions of supplied antibody solutions (see Table S1) using gel-loading tips, then sections were incubated at 4°C for 16 h. Sections were washed twice in 100 μl 1× TBST, then incubated for 1 h at 21°C in the dark in secondary antibody: either 15 μl of 2 μg μl−1 Alexa Fluor™ 488 donkey anti-rat IgG (H + L) (Invitrogen, (www. or 15 μl of 2 μg μl−1 Alexa Fluor™ 488 goat anti-mouse IgG (H + L) (www., based upon the requirements of the primary antibodies (see Table S1). Sections were again washed twice in 100 μl 1× TBST prior to counter-staining with 0.015% Congo Red (Sigma-Aldrich, ( Sections were again washed twice in 100 μl 1× TBST to remove excess counter-stain and unbound secondary antibody.

Confocal imaging

Sections were imaged row-wise on a confocal imaging dish (series GWSt-5030, WillCo Wells, ( using a spinning disk confocal microscope (Ultraview VoX, Perkin-Elmer, ( fitted with an electron multiplier CCD camera (9100–02, Hamamatsu Photonics, ( using an inverted microscope (DMI6000, Leica). Congo Red was excited with a 561 nm laser, and emitted light was then filtered with an emission wheel, transmitting 525 nm (width = 50nm) and 640 nm (width = 120nm) wavelengths. Alexa Fluor™ 488 secondary antibodies were excited with a 488 nm laser, and emitted light was filtered with an emission wheel allowing 527 nm (width = 55). Laser powers were adjusted section-wise to minimize background fluorescence and maximize contrast. Z–stacks were captured from the center 20 μm of each 40 μm section using Volocity software (Perkin-Elmer, (

Image processing and analysis

Region-of-interest (ROI) quantification was performed on 488 nm channels of two-channel maximum projections images using the ‘ROI’ tool within ImageJ, identifying seven tissue types (and background) discernable on the basis of morphology in the 561 nm channel (see Figure 1 for a list of tissue types). Background signals for each image (single well in 96-well layout) were subtracted from grey value measurements to yield background-corrected (Ībc) values (Table S2) for each of the seven cell types using the statistical programming environment, R ((

For assessment of relative intensity among tissue types, Ībc values for each tissue type were computed as the proportion of the sum of all Ībc values (seven tissue types) for each antibody/stage combination. Relative intensity scores (Īrel) were computed as the proportion of each Ībc value to the sum of all seven intensities for that antibody (∑Ībc). Īrel scores were adjusted by a correction factor using a custom logistic threshold function in order to reduce the contribution of background noise and residual fluorescence anomalies in cases where all seven values were below 20 units of the 256-level grey scale (Figure S4). Normalized absolute intensity values (Īn.abs) utilized for the pectin mAb-specific plotting (Figure S2) were computed using the correction factor parameters optimized for Irel computation as detailed in Figure S4. Īn.abs values are provided in Table S3.


We would like to acknowledge invaluable discussions with Lacey Samuels regarding immunolabelling procedures and interpretation. We are greatly indebted to Paul Knox (Centre for Plant Science, Faculty of Biological Sciences, University of Leeds) for additional JIM5, JIM7 and LM10 antibodies, obtained through Lacey Samuels (Department of Botany, University of British Columbia). This project would not have been possible without funding from the Natural Sciences and Engineering Council of Canada, and the resources and knowledge available through the Complex Carbohydrate Research Centre (University of Georgia). In particular, we thank Michael Hahn (Complex Carbohydrate Research Centre, University of Georgia) for useful insights on interpretation of immunolabelling data. Finally, we are grateful to Colin MacLeod for his technical assistance with immunolabelling and Rick White [The Statistical Consulting and Research Laboratory (SCARL), University of British Columbia) for statistical consultation].