There are 10 genes in the Arabidopsis genome that contain a domain described in the Pfam database as domain of unknown function 579 (DUF579). Although DUF579 is widely distributed in eukaryotic species, there is no direct experimental evidence to assign a function to it. Five of the 10 Arabidopsis DUF579 family members are co-expressed with marker genes for secondary cell wall formation. Plants in which two closely related members of the DUF579 family have been disrupted by T-DNA insertions contain less xylose in the secondary cell wall as a result of decreased xylan content, and exhibit mildly distorted xylem vessels. Consequently we have named these genes IRREGULAR XYLEM 15 (IRX15) and IRX15L. These mutant plants exhibit many features of previously described xylan synthesis mutants, such as the replacement of glucuronic acid side chains with methylglucuronic acid side chains. By contrast, immunostaining of xylan and transmission electron microscopy (TEM) reveals that the walls of these irx15irx15l double mutants are disorganized, compared with the wild type or other previously described xylan mutants, and exhibit dramatic increases in the quantity of sugar released in cell wall digestibility assays. Furthermore, localization studies using fluorescent fusion proteins label both the Golgi and also an unknown intracellular compartment. These data are consistent with irx15 and irx15l defining a new class of genes involved in xylan biosynthesis. How these genes function during xylan biosynthesis and deposition is discussed.
Xylan is an abundant non-cellulose polysaccharide that may constitute up to 30% of plant biomass. It is composed of a linear backbone of β1,4-linked xylose with a variety of short side branches that vary considerably between different species (Ebringerova and Heinze, 2000). Arabidopsis xylan has a structure typical of many dicotyledonous plants, including birch wood, in which the linear xylan backbone is substituted with both glucuronic and methyl glucuronic acid side branches (Brown et al., 2007; Peña et al., 2007).
Whereas xylan potentially represents an abundant source of carbohydrate that may be used for the synthesis of ethanol or other biofuels, it also poses considerable problems for the efficient use of the plant biomass. Xylan is reported to limit the ability of cell wall degrading enzymes to access other cell wall polymers such as cellulose (Yang and Wyman, 2004). Furthermore, the modification of sugars by acetylation and the potential cross-linking of xylan to lignin and other phenolic compounds is believed to have a major role in limiting cell wall breakdown and use (Grabber, 2005).
Expression analysis has made a large contribution to identifying the genes required for xylan biosynthesis (Brown et al., 2005; Persson et al., 2005). However, how these genes contribute to different aspects of xylan biosynthesis remains unclear. At least three proteins, IRX7/FRA8, IRX8 and PARVUS, appear to be required to synthesize a short oligonucleotide sequence found at the reducing end of the xylose chain (Brown et al., 2007; Peña et al., 2007). Additionally, three separate putative glycosyltransferases, IRX9, IRX14 and IRX10/IRX10L, are required for the proper elongation of the xylose backbone (Brown et al., 2007, 2009; Peña et al., 2007; Wu et al., 2009). The situation is further complicated by the presence of functional homologues of several of these genes (Lee et al., 2009, 2010; Wu et al., 2010). Although progress has been made in identifying the genes required for xylan biosynthesis, the exact mechanism and the exact reaction catalysed by each of these individual proteins is yet to be unambiguously established (York and O’Neill, 2008). Furthermore, in common with all plant cell wall carbohydrates, how xylan is transported within the cell and how it is targeted into particular areas of the cell wall remains poorly understood.
Despite being more than a decade since the first publication of the Arabidopsis Genome Sequence (AGI, 2000), the functional annotation of the majority of the open reading frames are not supported by any experimental data. In particular, attributing a function to novel gene families with no homology to any genes of known function remains particularly challenging. A large number of protein domains have been identified and classified in the Pfam database, and this includes many that have been classified as domains of unknown function (DUFs) (Finn et al., 2010). Many DUFs are represented in the Arabidopsis genome. DUF579 is widely distributed among eukaryotic species, including plants, yeast and mammals; however, only a single example has been identified in prokaryotes (Finn et al., 2010). Despite the fact that NMR has been used to solve the structure of a DUF579-containing protein from the yeast Saccharomyces cerevisiae, little or no information on the actual function of this domain is available. Ten genes in the Arabidopsis genome encode proteins containing a DUF579 (Finn et al., 2010), but there is no experimental data to support the function of any of these genes.
Functional analysis of secondary cell wall deposition has proved particularly amenable to expression analysis using microarrays. At certain stages of development, genes required for secondary cell wall formation are dramatically upregulated. This has been used to identify genes that are co-expressed with the cellulose biosynthesis genes CESA4, CESA7 and CESA8 that function as specific marker genes for secondary cell wall deposition (Brown et al., 2005; Persson et al., 2005). In this study, we investigate the function of five members of the DUF579 gene family that are all expressed during secondary cell wall deposition. Our results suggest that at least two members of the DUF579 gene family are essential for normal xylan deposition during secondary cell wall deposition, but may not be directly involved in catalysing the addition of sugars to the growing xylan polymer.
Identification of DUF579 family members expressed during secondary cell wall formation
In Arabidopsis, 10 genes encode predicted proteins that contain a DUF579. All 10 predicted proteins have a conserved structure and range in size from 282 to 329 amino acids. The Aramemnom database (Schwacke et al., 2003) suggests that in all 10 proteins the DUF579 domain is preceded by a single predicted transmembrane anchor close to the amino terminus, and predicts that the protein is localized in the secretory system (Figure S1). Endomembrane localization is supported by proteomic analysis that suggests at least two members of the Arabidopsis DUF579 family reside within the Golgi (Dunkley et al., 2006; Sadowski et al., 2008).
Previously published analysis of genes co-expressed with CESA4, CESA7 and CESA8 using expression data from developing stems have identified At1g09610 and At1g33800 within the 15 most closely co-expressed genes (Brown et al., 2005). Further analysis of this data suggests that in total five members of the Arabidopsis DUF579 family are among the 150 genes most closely co-expressed with CESA4, CESA7 and CESA8 (Table S1). At least some of these genes have also been identified in separate studies using different algorithms to identify genes co-expressed with CESA4, CESA7 and CESA8 (Persson et al., 2005; Srinivasasainagendra et al., 2008). Many of the genes identified in these co-expression studies have been subjected to functional analysis that has confirmed their functions in some aspect of secondary cell wall biosynthesis (e.g. Brown et al., 2007, 2009; Persson et al., 2007; Mortimer et al., 2010). Consequently, it appeared likely that at least some members of the DUF579 gene family are involved in secondary cell wall formation. To further confirm the expression of these five members of the DUF579 gene family during secondary cell wall formation we analysed the cell-specific expression patterns of these genes in the Arabidopsis root using publicly available expression data (Birnbaum et al., 2003; Brady et al., 2007). We have previously used this approach to visualize tissue-specific expression patterns (Brown et al., 2005), but can now exploit the expanded and improved data sets that have recently become available (Brady et al., 2007). Using this analysis CESA7 expression is very specifically limited to the xylem, and is highest on the oldest parts of the root (Figure S2). The expression of those five members of the DUF579 gene family identified from co-expression show expression patterns in the root very similar to that of CESA7, whereas none of the other DUF579 family members exhibit a similar expression pattern (Figure S2).
In order to identify the function of DUF579 family members in secondary cell wall formation, we obtained homozygous T-DNA insertion lines in all five family members expressed during secondary cell wall deposition (Figure S3a; Table S1). We examined hand-cut sections of the stem of all these lines in order to determine whether any of these plants exhibited a phenotype characteristic of a secondary cell wall defect. None of the single mutant plants exhibited an obvious secondary cell wall defect, and the plants grew in a very similar manner to the wild type (Figures 1 and S4). Examination of the phylogenetic relationship of DUF579 family members reveals that the members expressed during secondary cell wall formation are located within two distinct subclades (Figure S2). At3g50220 and At5g67210 are closely related, and both fall within the same subclade (Figure S2). As neither of the lines that were homozygous for insertions in these two genes exhibited a clear phenotype, real-time PCR was used to confirm that the mRNA levels were almost undetectable in these lines (Figure S3b). When we analysed double mutants in which both of these genes were disrupted, stem cross sections showed that some of the xylem vessels exhibited a mild collapsed xylem phenotype that is indicative of a secondary cell wall defect (Figure 1; Brown et al., 2005). In addition, the inflorescence stem of these plants were slightly shorter and brittle (Figure S4 and data not shown). As these genes are closely related and appeared to exhibit some degree of functional redundancy, we focused on these genes for the subsequent analysis aimed at determining the nature of the defect. We named At3g50220 IRREGULAR XYLEM 15 (IRX15) and named At5g67210 IRX15L.
We also carried out preliminary analysis of the remaining three DUF579 genes that are upregulated during secondary cell wall deposition that fall into a separate subclade to that of IRX15 and IRX15L. Double mutants between insertion mutants in At1g09610 and At4g09990 exhibit no obvious phenotype. We have been unable to recover a double mutant between At1g09610 and At1g33800. Although these genes are on the same chromosome, it should be possible to recover double mutants at a reasonable rate. Segregating crosses exhibit no obvious signs of embryo seedling lethality (data not shown). The reason that we are unable to recover the double mutant in this case is likely to stem from a chromosomal alteration, which are common in T-DNA insertion lines (Clark and Krysan, 2010). We also examined whether mutations in irx15 or irx15l were enhanced by mutants in At1g09610; however, preliminary analysis of double mutant lines with insertion irx15/At1g09610 or irx15l/At1g09610 gave no obvious phenotype.
DUF579 mutants exhibit large alterations in cell wall composition
In order to identify the nature of the secondary wall defect in the irx15irx15l double mutant line, the sugar composition of the non-cellulosic fraction of the cell wall was analysed by gas chromatography (GC). The most obvious difference was a large decrease in the level of xylose in the irx15 irx15l double mutant, which was reduced to 65% of that of the wild type (Figure 2a). Insertions in irx15 also appeared to give a significant but smaller decrease in xylose, but levels of xylose appear to be unaffected in irx15l. In order to further characterize the nature of the xylose decrease in the double mutant line, we subjected the cell walls of irx15 irx15l to chemical fractionation and analysed the sugar composition of each fraction. Consistent with previous analysis, most of the xylose was extracted in either the 1 m or 4 m KOH fractions that contain the majority of the cell wall xylans. In irx15 irx15l the xylose content of the 1 m KOH fractions were similar to those of the wild-type lines (Figure 3). In contrast, in both the 4 m KOH fractions and the residue, the proportion of the xylose was significantly reduced in irx15 irx15l compared with the wild type. The quantity of 4 m KOH extractable polysaccharides was also less in the irx15 irx15l mutant. These changes are consistent with the mutant having less xylan in the secondary cell wall.
In order to measure xylan levels directly, cell wall material from both wild-type and mutant lines were subjected to polysaccharide analysis by carbohydrate gel electrophoresis (PACE). The results show directly that irx15 irx15l exhibits a large reduction in the level of xylan, which is reduced to half that of the wild-type level (Figures 2c and S5). Also consistent with the sugar composition analysis, there is a suggestion that irx15 exhibits a decrease in xylan, whereas irx15l appears similar to the wild type. We also used PACE analysis to determine the proportion of glucuronic acid or methyl glucuronic acid side branches attached to the xylan chain (Figures 2d and S5). None of the mutants exhibited any significant difference from the wild type, indicating that the frequency of side branch addition is unaffected by the mutations.
To determine whether irx15 irx15l had more widespread affects on the secondary cell wall deposition, we compared the cellulose content to that of the parental wild types. The double mutant did exhibit a significant decrease in the cellulose content compared with the wild types; however, this decrease was not as large as that observed for a severe cellulose-deficient mutant such as irx3 (Figure 2b), and is comparable with that observed in other previously described xylan mutants (Brown et al., 2007, 2009).
Xylan structure is altered in DUF579 mutant lines
In order to investigate the structure of xylan in the mutant lines, the xylanase digests were further analysed using matrix-assisted laser desorption/ionisation-time of flight mass spectrometry (MALDI-ToF). Xylan from both of the single mutants (irx15 and irx15l) and the double mutant (irx15 irx15l) still contain the short oligosaccharide sequence (Figure 4) previously identified at the reducing end of xylan chains (Brown et al., 2007, 2009; Peña et al., 2007). In common with a number of the other xylan-deficient mutants, however, the ratio of methyl glucuronic acid to glucuronic acid side branches was dramatically altered. This is particularly apparent in the double mutant, in which glucuronic acid side chains are almost undetectable, whereas methyl glucuronic acid side branches were still apparent (Figure 4). Given that the frequency of side chain addition does not change, this would suggest that the glucuronic acid side branches are replaced with methyl glucuronic side branches.
The cell wall fractionation data shows that a higher proportion of the xylan in the double mutants is extracted in the 1 m KOH fraction. This suggests that there may be some alteration in the length and/or structure of the xylan. To investigate this further, purified xylan was subjected to size exclusion chromatography (Figure 5a). None of the major peaks were apparent if the samples were first subjected to digestion with xylanases, confirming that the material being examined was indeed xylan. Compared with the wild type, the majority of the xylan from irx15 irx15l eluted later from the column, suggesting that the xylan in this line was of a lower molecular weight than that of the wild type. Three mutants have previously been reported with decreased xylan chain length, and in all three cases these mutants exhibit decreased xylosyltransferase activity (Brown et al., 2007, 2009; Peña et al., 2007). Consequently, we examined the xylosyltransferease activity of irx15/irx15L; however, in this case the xylosyltransferase activity of microsomal extracts exhibited only a small decrease compared with the wild type, suggesting that the apparent chain length decrease in these mutants did not result from a reduction in the rate of xylan backbone elongation (Figure 5b).
DUF579 alters xylan distribution and cell wall morphology
In order to examine whether mutations in IRX15 IRX15L altered the distribution of xylan in the wall, we used the well-characterized LM10 antibody (McCartney et al., 2005) to examine xylan distribution in cross sections. The double mutant line appeared to have xylan in both the interfascicular region and the xylem, but compared with the wild type the distribution of xylan appeared much more uneven. This uneven distribution is most apparent in the interfascicular region, where some regions of the secondary cell wall appeared to stain very intensely, whereas adjacent areas of the same cell wall appeared to exhibit almost no fluorescence (Figure 6).
To examine the implications of this uneven distribution of xylan in more detail, we examined the ultrastructure of the secondary cell walls using TEM. As has been shown in previous studies (Turner and Somerville, 1997), the secondary cell wall of the interfascicular region appears quite smooth and, with the exception of the interfascicular regions, stains relatively evenly (Figure 7). Consistent with previous reports of xylan-deficient mutants (Zhong et al., 2005; Peña et al., 2007), irx7 and irx9 appeared similar to the wild type, except the walls were much thinner. In contrast, walls of the irx15 irx15l double mutant exhibited a much more uneven surface, and stained more unevenly. Cell walls from the wild type exhibit very pronounced dark staining mid-lamellae, as do walls from irx7 and irx9. In contrast, in irx15 irx15l the mid-lamellae were also much less well pronounced, and were hard to distinguish from the secondary cell wall (Figure 7).
DUF579 proteins localize to peripheral Golgi and an additional unidentified intracellular compartment
In order to identify the subcellular localization of the DUF579-containing proteins, we constructed a series of fusion proteins in which either YFP was fused at the N terminus of both IRX15 and IRX15L, or mCherry was fused to the C terminus of IRX15L and transformed into the irx15 irx15l double mutant. IRX15 and IRX15L were identified on the basis of their co-expression with IRX3. Abundant tissue- and cell-specific data indicates similar tissue-specific expression patterns and expression levels for these three genes. Consequently, we used the well-characterized IRX3 promoter to drive the expression of the transgene, as it will give an expression pattern identical, or very similar, to that of the endogenous gene. All three fusions showed a similar intracellular distribution that labelled several compartments (Figure 8; Video Clip S1). All constructs clearly labelled compartments that co-localize with a Golgi marker (Figure 8); however, only the C-terminal IRX15L:mCherry fusion complemented the irx phenotype of irx15 irx15l (data not shown). It was also notable that using this construct the fusion protein labelled a ring around the periphery of the Golgi, as has been described previously for CESA proteins (Paredez et al., 2006; Desprez et al., 2007; Wightman and Turner, 2008). In addition to the Golgi, there is also more widespread labelling within the cell (Figure 8). This distribution resembles that of ER staining in xylem vessels (data not shown); however, prominent small, but intensely stained intracellular compartments were also clearly visible that did not co-localize with Golgi markers (Figure 8). Four of the compartments were tracked in time-lapse movies of developing xylem vessels (Video Clip S1). Golgi movement in xylem vessels has previously been shown to proceed in a single direction around the cell in an actin-dependent manner (Wightman and Turner, 2008). The smaller compartments, however, did not always proceed with the bulk flow of faint YFP material as expected. In Video Clip S1, the compartments labelled blue and green exhibit erratic movements within a small area of the cell. The compartment labelled red initially exhibits these erratic movements, but then follows a unidirectional high-velocity path consistent with movement along the thick actin cables (Wightman and Turner, 2008). Another path (labelled cyan) also exhibits high-velocity movement in the same direction. This, together with the co-localization data of Figure 8, shows the bright compartments to be distinct and to have different properties to the Golgi apparatus.
DUF579 genes exhibit a dramatic affect on cell wall digestibility
The altered cell wall morphology exhibited by DUF579 mutants led us to examine what affect they might have on cell wall digestibility. Consequently, we compared the sugars released from the irx15 and irx15l mutants using a standard mixture of enzyme designed to release cell wall sugars. The irx15 mutant alone is sufficient to give a dramatic increase in cell wall sugar release that is not significantly enhanced in the irx15 irx15l double mutant (Figure 9). Sugar release from irx15 is significantly higher throughout the course of an 18-h incubation, and after 18 h, when sugar release levels off, the sugar release is still 46% higher than in the corresponding wild type (Figure 9).
Proteins that contain DUF579 are predicted to be widespread among eukaryotic organisms; however, there is little, or no, experimental data that can be used to ascribe a specific function to it. In Arabidopsis there are 10 genes that possess a DUF579 motif. These genes all have a conserved structure and are predicted to be type-II membrane proteins with a single transmembrane spanning domain at the N terminus, with the bulk of the protein located within the endomembrane system. An analysis of co-expression patterns (Table S1) suggests that at least five members of this gene family are expressed during secondary cell wall formation, and as such are likely to play some role in this process. The fact that some of these five genes are closely related, and that lines in which only one of these five genes is disrupted do not exhibit a severe phenotype, suggests that a degree of functional redundancy exists within this family. This is confirmed by the analysis of a double mutant derived from cDNA insertion mutants in two closely related genes. The irx15 irx15l double mutant plants are slightly shorter and bushier than either of the wild-type parents, and exhibit a weak irregular xylem phenotype, characterized by slightly misshapen xylem vessels in the stem, indicative of secondary cell wall defects (Figure 1). The function of the other three DUF579-containing genes expressed during secondary cell wall formation is unclear at present; however, the irregular wall phenotype of irx15 1rx15l is most pronounced in the interfascicular region, and so it is conceivable that some DUF579 are more important in the interfascicular region, whereas others are more important within the xylem. This is supported by recent analysis of gene expression in xylem vessels, as opposed to xylem fibres (Ohashi-Ito et al., 2010), that show both IRX15 and IRX15L are expressed much more highly in xylem fibres than in vessels.
A number of lines of evidence indicate that the secondary cell wall defects in the irx15 irx15l mutant are at least in part caused by a decrease in xylan deposition. Analysis of the non-cellulose sugar composition of total cell wall from stems shows a large decrease in xylose in the double mutants (Figure 2a). Consistent with this observation, fractionation studies of the cell wall suggest that the xylose content is decreased in the 4 m KOH-extractable fractions and the insoluble residue (Figure 3). This analysis is similar to the decrease in xylose content reported for other xylan biosynthesis mutants (Zhong et al., 2005; Brown et al., 2007, 2009; Peña et al., 2007). Direct measurement of xylan using PACE analysis demonstrates a 50% decrease in xylan that is likely to account for the decrease in cell wall xylose (Figure 2c).
Whereas all of the data described above support a role for the Arabidopsis DUF579 family members in xylan biosynthesis, the irx15 irx15l double mutant also exhibits a significant decrease in cellulose content (Figure 2b). This decrease is, however, not as large as that observed in severe cellulose mutants, such as irx3. A decrease in cellulose content has been observed in several other xylan biosynthesis mutants; however, no reason for this link has been established (Persson et al., 2007; Brown et al., 2009). Known cellulose-deficient mutants such as irx3 do not exhibit any change in either xylose content of the cell wall or the replacement of glucuronic acid side branches of xylan with methylglucuronic acid (Brown et al., 2007). Consequently, it is unlikely that the xylan phenotype described for irx15 irx15l is a consequence of a primary defect in cellulose biosynthesis. It is more likely that a decrease in xylan deficiency in some way affects the ability of cells to properly deposit cellulose into the wall. We cannot entirely rule out, however, that irx15 irx15l is defective in some process that affects both cellulose and xylan deposition.
The PACE analysis indicates that the frequency of side branch addition to the xylan backbone is unchanged in the mutant (Figure 2c). Further analysis of the xylan structure using MS indicates that the xylan from the double mutants contains no glucuronic acid side branches, and that these have been replaced with methyl glucuronic acid (Figure 4). Both phenotypes are characteristic of previously described mutants that are putative glycosyltransferases believed to be directly involved in xylan biosynthesis (Brown et al., 2007, 2009; Peña et al., 2007). MS analysis also indicates that the xylan retains a characteristic reducing end oligosaccharide that is absent in some previously described xylan-deficient mutants that are believed to be directly involved in the biosynthesis of this reducing end sequence (e.g. Brown et al., 2007; Peña et al., 2007). The replacement of glucuronic acid by methylglucuronic acid branches of xylan is characteristic of xylan synthesis mutants (Brown et al., 2007), and further supports a direct role for IRX15 and IRX15L in xylan synthesis. There are, however, unique characteristics of the xylan-defective phenotype of irx15 irx15l. The chain length of the extractable xylan appears to be shorter in the double mutants compared with the wild type, but the double mutant does not exhibit the large decreases in xylosyltransferase activity observed in other mutants that exhibit decreases in xylan backbone chain length (Brown et al., 2007; Lee et al., 2007a). Consequently, although irx15 irx15l clearly exhibits a xylan-deficient phenotype, it affects xylan biosynthesis in a distinctive way and appears to specify a new class of genes required for xylan biosynthesis. This conclusion is supported by both the uneven distribution of xylan in the cell wall (Figure 6) and the uneven morphology of the secondary cell wall when viewed by TEM (Figure 7). Neither of these characteristics have been previously reported for xylan-deficient mutants (Lee et al., 2007b; Peña et al., 2007).
The irx phenotype of irx15 irx15l is complemented using the IRX15L gene with a C-terminal mCherry fusion. Several xylan biosynthesis genes localize to the Golgi, where they form punctate structures that appear identical to the Golgi markers used (Zhong et al., 2005; Peña et al., 2007). In contrast, the IRX15L:mCherry fusion appears to form a distinctive ring structure in the Golgi, with the mannosidase Golgi marker localizing in the centre of this ring. This is reminiscent of the distribution of the cellulose synthase complex in the Golgi, and may be a common feature of some genes involved in cell wall biosynthesis (Paredez et al., 2006; Desprez et al., 2007; Wightman and Turner, 2010). In addition, the fusion proteins also appeared to identify a brightly stained smaller intracellular compartment that is distinctive from the Golgi (Figure 8). Although some of these bodies appear to undergo rapid cytoplasmic streaming, reminiscent of the Golgi, a proportion of these DUF579-labelled compartments are confined within a small area, where they exhibit only short erratic movements (Video Clip S1). Two other small intracellular compartments have been identified with CESA fusion proteins; however, neither MASC/SmaCCs nor the VHAa1-associated compartments (Crowell et al., 2009; Gutierrez et al., 2009) exhibit the characteristics of the IRX15L-labelled compartment described above. The exact function of this compartment is unclear, but it is possible it may have some role in the transport of cell wall polysaccharides into the cell wall. A role for DUF579 proteins in the delivery of cell wall assembly is consistent with both immunolabelling of xylan distribution and direct observation of cell wall morphology in the mutants using TEM. Both studies appear to indicate a disruption in normal cell wall organization that is not observed in previously studied xylan biosynthesis mutants. It should be noted, however, that the disrupted wall visualized by TEM is somewhat reminiscent of the uneven wall observed in severe cellulose-deficient mutants (Turner and Somerville, 1997). Consequently, the cell wall phenotype observed in the double mutants may be a consequence of both reduced xylan and alterations in cellulose biosynthesis. Several altered cell wall properties are also indicated by the digestibility assay that clearly shows a dramatic increase in sugar release. The increase in digestibility is comparable with lignin biosynthesis mutants, known to dramatically improve digestibility (Chen and Dixon, 2007), analysed under the same conditions (L. Gomez, unpublished data).
All the data described above suggest that the irx15 irx15l mutant represents a new class of xylan-deficient mutant, but it is not clear exactly where it acts in the pathway of xylan biosynthesis and deposition. The replacement of glucuronic acid side branches with methylglucuronic acid is a feature of all known xylan-deficient mutants, and different hypotheses have been proposed that involve the availability of nucleotide sugars (Brown et al., 2007; Peña et al., 2007), and consequently this phenotype provides little insight into the specific roles of IRX15 or IRX15L. The fact that xylan from irx15 irx15l has normal levels of the reducing end oligosaccharide suggests that IRX15 and IRX15L are unlikely to be directly involved in the synthesis of this structure. Paradoxically, although the xylosyltransferase activity of irx15 irx15l microsomes is only slightly less than that of the wild type, the xylan elutes more slowly from size exclusion columns (SECs). Although this is generally accepted to be an indication of reduced size, it is unclear what affect altering the xylan side chains will have on xylan mobility. The replacement of glucuronic acid with methylglucuronic acid could affect the xylan hydrodynamic shape or alter the way in which xylan chains interact with one another, both of which would have the potential to alter the way in which the xylan behaves during SEC.
The sequence of the DUF579 domain offers no support for a direct role as a glycosyltransferase: for example, there are no conserved DxD motifs, frequently associated with activated sugar binding. Although DUF579 could indirectly be required for glycosyltransferase activity, for example, by targeting proteins to the Golgi, this is not supported by the morphological studies of the wall. The uneven nature of cell wall deposition in irx15 irx15l is not seen in either irx7 or irx9 mutants (Figure 7), which are believed to be glycosyltransferases required for the synthesis of the reducing end oligosaccharide or the xylose backbone, respectively (Brown et al., 2007; Peña et al., 2007). Other possibilities for how IRX15 and IRX15L function could be in the modification of xylan, for example by the addition of acetate groups, or that they are somehow involved in the transport or targeting of xylan to the cell wall. This latter possibility would be consistent with their location in the endomembrane system, which appears to be predominantly within the Golgi, but the fluorescent protein fusions also label the endoplasmic reticulum (ER) and a small compartment.
A role for DUF579 in transport through the endomembrane system is consistent with information gained from the study of DUF579-containing proteins from other species. The structure DUF579 from the yeast S. cerevisiae has been solved using NMR, and contains a coiled-coil domain, suggestive of protein–protein interaction. Furthermore, high-throughput screens of the yeast YPL225W gene indicate genetic interactions with various components of the cytoskeleton and the endomembrane systems, including the ARP2/3 complex and the GET1 complex, suggesting that YPL225W may be part of the system involved in protein trafficking/polysaccharide trafficking, and/or interactions with the cytoskeleton (Costanzo et al., 2010).
During the course of this study it came to our attention that Curtis Wilkerson and colleagues have independently been studying the role of IRX15 and IRX15L in cell wall deposition, and have come to broadly similar conclusions to those described here. In summary, we have characterized a family of Arabidopsis proteins containing a DUF579, and have established a role for some of these family members in secondary cell wall formation. Specifically, two of these genes are essential for normal xylan deposition in the secondary cell wall. The mutants exhibit unique characteristics that suggest that they define a new and distinct class of proteins required for xylan biosynthesis. Our analysis suggests that IRX15 or IRX15L are unlikely to be glycosyltransferases directly involved in the synthesis of xylan, and offer an opportunity to study a new process in xylan biosynthesis.
Members of the DUF579 family were identified in Pfam and TAIR. Identification of the Pfam domain was carried out by comparison with the Pfam consensus sequence. Putative transmembrane domains were identified using the Aramemnon database.
Plants were germinated and grown as described by Brown et al. (2009). Briefly, seedlings were germinated on MS and B5 vitamins (Murashige and Skoog, 1962) in 0.8% (w/v) agar. Seedlings were grown for approximately 14 days before transfer to compost containing vermiculite and perlite (10:1:1). Plants were grown under a regime of constant temperature (22°C) and light (150–180 μmol m−2 sec−1).
Transverse stem sections (5-μm thick) were cut from paraffin wax-embedded stem segments and stained with toluidine blue O (Sigma-Aldrich, http://www.sigmaaldrich.com). Sections were examined using microscopy, as described in the microscopy section below.
Cell wall analysis
Non-cellulosic cell wall sugars were analysed using trimethylsyl (TMS) ethers of methyl glycosides, and the cellulose content of stems from 6-week-old plants was measured as previously described (Brown et al., 2005).
Immunolocalization of cell wall xylan
Stems from 6-week-old plants were fixed and embedded in paraffin wax, as previously described (Brown et al., 2009). Transverse sections (5-μm thick) were cut with a microtome and incubated with LM10 antibody (1:10 dilution) (McCartney et al., 2005). They were then washed and incubated with rabbit anti-rat fluorescein isothiocyanate (FITC)-conjugated secondary antibodies (Sigma-Aldrich), diluted 1:100. Sections were viewed under fluorescence microscopy, as described in the microscopy section below.
Xylan analysis by size-exclusion chromatography (SEC)
Xylan analysis by SEC was performed as previously described (Brown et al., 2009). Briefly, alcohol-insoluble cell wall material was sequentially fractionated using endopolygalacturonase, CDTA and Na2CO3 to remove pectin. Xylan was then extracted using 4 m KOH under oxygen-free conditions. The extracted xylan was dialysed extensively against de-ionized water, filtered through a GF/C glass fibre filter (Whatman) and then lyophilized. Treatment with Trichoderma viride endo-1,4-β-xylanase (Sigma-Aldrich) was carried out for 16 h at pH 4.5 and 30°C. Both untreated and xylanase-treated fractions were then analysed by SEC using a Superdex 200 10/300 column (Amersham, now part of GE Healthcare, http://www.gehealthcare.com) at a flow rate of 0.5 ml min−1 in 50 mm sodium acetate. Data were normalized to the largest peak in the chromatograph.
Microsomal membrane isolation and xylosyltransferase assay
Microsomal membrane isolation and xylosyltransferase assays were performed as described previously (Brown et al., 2009). Microsomal membranes were isolated by grinding the stems of 5-week-old plants in a buffer containing 50 mm HEPES-KOH, pH 7.2, 1 mm DTT, 1 mm MgCl2, 5 mm EDTA, 100 mm sodium ascorbate, 0.4 m sucrose, proteinase inhibitor cocktail (Roche Applied Science, http://roche.com) and 1% w/v polyvinylpolypyrrolidone (PVPP). The ground suspension was then filtered through nylon (50-μm mesh), and large particulates were removed by centrifugation (2500 g for 15 min). Microsomal membranes were isolated by centrifuging the resulting supernatant at 100 000 g for 1 h and re-suspending in extraction buffer. Xylosyltransferase assays containing 300 μg of microsomal membranes in 50 mm HEPES-KOH, pH 7.2, 5 mm MgCl2, 10 mm NaF, 1 mm DTT and 3.7 μm UDP-[14C]-d-Xyl (1.5 kBq) were performed with or without 1 mmβ(1,4)Xyl6 (Megazyme) at room temperature (25°C) for 30 min. Assays were terminated by adding 50% w/v Dowex-1X8 (Sigma-Aldrich), and xylosyltransferase activity was determined by measuring the total radioactivity of the eluate by liquid scintillation counting.
Construction of fluorescent protein fusions
The vectors for expression of N-terminal YFP-tagged IRX15 and YFP-tagged IRX15L constructs were assembled using Gateway technology (Invitrogen, http://www.invitrogen.com), as described by the manufacturer. The p3HSYC Gateway destination vector was constructed by inserting an EYFP sequence (Clontech, http://www.clontech.com) into the previously described p3HSC destination vector (Atanassov et al., 2009), downstream of the attR1/CmR/ccdB/attR2 cassette. The entry clones were obtained by the PCR amplification of cDNA fragments containing the full-length IRX15 and IRX15L genes and BP clonase II cloning into pDONR/Zeo (Invitrogen). The entry clones were subcloned into the p3HSYC destination vector using LR clonase II cloning. The resulting YFP:IRX15 and YFP:IRX15L expression vectors contained the YFP:IRX15 and YFP:IRX15L gene fusions under the 1.7-kb CESA7 promoter sequence.
The IRX15L:mCherry fusion was made by amplifying the coding sequence of IRX15L cDNA using a forward primer that incorporated a NheI site to the beginning of IRX15L and a reverse primer that incorporated a ClaI site. ClaI/NheI-digested IRX15L DNA was then ligated into ClaI/NheI-cut pG58 consisting of ApaI-CESA7 promoter-NheI-ClaI-mCHERRY-XbaI-NosT-NotI in pBluescript II KS+. Annealed oligonucleotides containing SalI flanked with either ApaI or NotI were cloned into the ApaI sites and NotI sites, respectively, of pG58. The entire construct was subcloned as a SalI fragment into SalI-cut pCB1300.
All constructs were transformed into irx15 irx15l plants using the floral-dip method. The xylem-specific CFP-mannosidase I Golgi reporter has been described previously (Wightman and Turner, 2008).
Live imaging was carried out on a Leica DM5500 microscope equipped with a SPOT RT3 camera (Imsol, http://www.imsol.co.uk). The objectives and filter sets used to view YFP, CFP and mCherry were as described in Wightman and Turner, 2008. For the high-detail image of IRX15L:mCherry and CFP-mannosidase I (Figure 8c), seedlings were treated with 1 μm lantrunculin B for 15 min prior to imaging using the Leica DM5500. A profile plot taken through the centre of the Golgi was carried out in ImageJ (W. Rasband, National Institutes of Health). Concurrent images of YFP:IRX15L and CFP-mannosidase I (Golgi marker) were taken using a Leica TCS SP5 confocal microscope. Light microscopy of toluidine blue O-stained stem sections and immunofluorescence microscopy of samples treated with the anti-xylan antibody were carried out using a Leica DMR fitted with a SPOT Xplorer 4MP camera and 40× and 100× objectives.
Transverse sections of mature stems were vacuum infiltrated in fixative (1% glutaraldehyde and 4% formaldehyde) and embedded in TAAB low-viscosity resin and polymerized at 60°C for 24 h. Ultrathin sections were cut on a Reichert–Jung Ultracut microtome, stained with 1% uranyl acetate and 0.3% lead citrate, and then viewed on a FEI Tecnai 12 electron microscope (FEI, http://www.fei.com/).
Xylan in alcohol insoluble residue (AIR) was solubilized with 4 m NaOH and quantitated by PACE according to the method described by Brown et al. (2007), using xylanases from Cellvibrio japonicus Xyl10B (GH10) or Neocallimastix patriciarum Xyl11A (GH11) (gifts from Harry Gilbert, CCRC, The University of Georgia, Athens, GA, USA). For the calculation of branching frequency, data from NpXyl11A was used to estimate the ratio of [Me]GlcA to Xyl in the xylanase-released oligosaccharides. To determine xylan structure, xylan was solubilized by 4 m NaOH for 1 h at room temperature, and the reducing end was reduced with 10 mg ml−1 NaBH4 in 0.5 m NaOH for 2 h at room temperature. The reaction was terminated by adjusting to pH 5.5 with glacial acetic acid on ice. The xylan was extensively dialysed against water (molecular weight cut-off 500 Da) and dried. The xylan was then digested by CjXyl10B, deuteropermethylated and analysed by MALDI-TOF MS according to the protocol described by Brown et al. (2009).
Cell wall digestibility assay
Assays were performed as recently described (Gomez et al., 2010). Briefly, samples were prepared using the Grinding and Loading robot (Labman Automation Ltd, http://www.labman.co.uk), and formatted in 96-well plates to contain 4 mg of each sample either as four or eight replicates. The pre-treatment, hydrolysis and sugar determination was performed using the automated liquid handling station (Tecan, http://www.tecan.com). The powdered samples were pre-treated with 1% sulfuric acid at 90°C for 30 min, after which, the biomass was rinsed six times with 500 μl sodium acetate buffer. The samples were incubated while shaking at 50°C for 8 h in the presence of enzyme cocktail (4:1 ratio of Celluclast and Novozyme 188). Enzyme loading was 6.3 filter paper unit g−1 of material, and automated sugar determination was carried out as previously described (Gomez et al., 2010).
The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007–2013) under the grant agreement no. 211982. We are grateful to the BBSRC (grant reference BB/C505632/1) for supporting DB, IA and J-PB.