Regiospecific Acetylation of Xylan is Mediated by a Group of DUF231-Containing O-Acetyltransferases

Abstract

Xylan is a major hemicellulose in the secondary walls of vessels and fibers, and its acetylation is essential for normal secondary wall assembly and properties. The acetylation of xylan can occur at multiple positions of its backbone xylosyl residues, including 2-O-monoacetylation, 3-O-monoacetylation, 2,3-di-O-acetylation and 3-O-acetylation of 2-O-glucuronic acid (GlcA)-substituted xylosyl residues, but the biochemical mechanism controlling the regiospecific acetylation of xylan is largely unknown. Here, we present biochemical characterization of a group of Arabidopsis thaliana DUF231-containing proteins, namely TBL28, ESK1/TBL29, TBL30, TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35, for their roles in catalyzing the regiospecific acetylation of xylan. Acetyltransferase activity assay of recombinant proteins demonstrated that all of these proteins possessed xylan acetyltransferase activities catalyzing the transfer of acetyl groups from acetyl-CoA onto xylooligomer acceptors albeit with differential specificities. Structural analysis of their reaction products revealed that TBL28, ESK1, TBL3, TBL31 and TBL34 catalyzed xylan 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation with differential positional preference, TBL30 carried out 2-O- and 3-O-monoacetylation, TBL35 catalyzed 2,3-di-O-acetylation, and TBL32 and TBL33 mediated 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues. Furthermore, mutations of the conserved GDS and DXXH motifs in ESK1 were found to result in a complete loss of its acetyltransferase activity. Together, these results establish that these nine DUF231-containing proteins are xylan acetyltransferases mediating the regiospecific acetylation of xylan and that the conserved GDS and DXXH motifs are critical for their acetyltransferase activity.

Introduction

Xylan is a major plant polysaccharide present in secondary walls of tracheary elements and fibers as well as in primary walls of parenchyma cells in grasses. It is composed of a linear chain of β-1,4-linked xylosyl residues that are often decorated with 2-O-linked glucuronic acid (GlcA)/methylated glucuronic acid (MeGlcA) (Teleman et al. 2000). In addition, xylans from bryophytes (Haghighat et al. 2016), seedless vascular plants (Haghighat et al. 2016) and angiosperms (Teleman et al. 2000, Teleman et al. 2002, Evtuguin et al. 2003, Goncalves et al. 2008, Naran et al. 2009, Marques et al. 2010, Prozil et al. 2012, Yuan et al. 2013) are acetylated at O-2 and/or O-3 with the degree of acetyl substitution ranging from 20% to 70%. The acetylation of xylan is critical for normal secondary wall deposition and assembly, as demonstrated in Arabidopsis mutants showing that an 85% reduction in xylan acetylation results in a drastic decrease in the amount of cellulose and xylan and a severe defect in secondary wall structure (Yuan et al. 2016a). On the other hand, the acetyl groups present in xylan, one of the major components in plant biomass targeted for biofuel production, hinder the efficient conversion of biomass into biofuels (Selig et al. 2009). Therefore, deciphering the biochemical mechanism controlling xylan acetylation is of importance not only to further our understanding of how secondary wall components are synthesized and assembled but also may lay the knowledge base for a rational design of biomass composition better suited for biofuel production.

Genetic studies in Arabidopsis have implicated the involvement of three groups of proteins, i.e. RWAs (reduced wall acetylation), AXY9 (altered xyloglucan 9) and several TBL (trichome birefringence-like) proteins belonging to the DUF231 family, in xylan acetylation. Four RWA genes have been shown to be expressed in secondary wall-forming cells and their simultaneous mutations cause an overall reduction in xylan acetylation (Lee et al. 2011). RWA proteins contain multiple transmembrane helices and their sequences show homology to the transmembrane region of fungal CAS1, a protein involved in capsular glucuronoxylomannan acetylation (Janbon et al. 2001), but their exact biochemical functions remain to be investigated. The AXY9 mutation results in a reduction in the acetylation of xyloglucan and xylan. AXY9 has no significant sequence similarity to other proteins and it is currently unknown how AXY9 exerts its effect on polysaccharide acetylation (Schultink et al. 2015). Mutations in several TBL genes belonging to the DUF231 family have also been shown to affect xylan acetylation. The esk1 mutation causes an apparent reduction in xylan 2-O- and 3-O-monoacetylation, a decrease in xylan acetyltransferase activity, a deformation of vessel morphology and a retardation in plant growth (Xiong et al. 2013, Yuan et al. 2013). The double mutants tbl3 tbl31, tbl32 tbl33 and tbl34 tbl35 showed a mild reduction in xylan 3-O-monoacetylation and 2,3-di-O-acetylation (Yuan et al. 2016a, Yuan et al. 2016b, Yuan et al. 2016c). Although these double mutants alone did not show any apparent defects in vessel morphology or plant growth, the triple mutants tbl34 tbl35 esk1 and tbl32 tbl33 esk1 exhibited highly exacerbated phenotypes with severely deformed vessels and strongly retarded plant growth (Yuan et al. 2016a, Yuan et al. 2016b). In addition, the tbl32 tbl33 double mutant displayed a loss of 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues (Yuan et al. 2016a). The potential involvement in xylan acetylation of two other TBL genes, TBL28 and TBL30, which are phylogenetically grouped together with the above-described TBLs, has not yet been studied. Together with a recent report showing that ESK1 (eskimo 1) exhibits an acetyltransferase activity for 2-O- and 3-O-monoacetylation of xylan (Urbanowicz et al. 2014), the findings from the comprehensive genetic analysis of these TBL genes indicate that different acetyltransferase activities might be involved in catalyzing the regiospecific acetylation of xylan.

In this study, we performed expression and functional analyses of the TBL28 and TBL30 genes and carried out biochemical studies of TBL28, ESK1/TBL29, TBL30, TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35 for their acetyltransferase activities for the regiospecific acetylation of xylan. We found that both TBL28 and TBL30 genes were expressed in secondary wall-forming cells and that TBL28 expression was able to rescue the defects conferred by the esk1 mutation. We demonstrated that both TBL28 and ESK1 possessed acetyltransferase activities for 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation of xylan, whereas TBL30 only displayed an activity for xylan 2-O- and 3-O-monoacetylation. We further revealed that TBL3, TBL31 and TBL34 exhibited acetyltransferase activities catalyzing 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation of xylan with positional preference for 3-O-monoacetylation, TBL35 carried out 2,3-di-O-acetylation, and TBL32 and TBL33 are acetyltransferases responsible for 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues. Mutational analysis demonstrated that the conserved GDS and DXXH motifs in ESK1 were essential for its xylan acetyltransferase activity. Our findings provide biochemical evidence establishing that these TBL proteins are xylan acetyltransferases with differential positional preferences, which function together to control the regiospecific acetylation of xylan.

Results

TBL28 and TBL30 are expressed in secondary wall-forming cells and their encoded proteins are targeted to the Golgi

TBL28 and TBL30 are Phylogenetically grouped together with ESK1 and six other TBL proteins (Fig. 1A) that have been genetically shown to play critical roles in xylan acetylation in Arabidopsis, indicating their possible involvement in xylan acetylation. We first investigated whether the expression of TBL28 and TBL30 was associated with cells undergoing xylan biosynthesis and whether their encoded proteins were targeted to the Golgi where xylan biosynthesis occurs. Real-time quantitative PCR analysis of different Arabidopsis organs showed that TBL28 was expressed preferentially in the stems, whereas TBL30 did not exhibit such a preferential expression (Fig. 1B). Further expression analysis at the cell type level using the β-glucuronidase (GUS) reporter gene revealed that TBL28 was expressed predominantly in both xylem cells and interfascicular fiber cells in the stems (Fig. 1C, D), whereas TBL30 expression was restricted to xylem cells (Fig. 1F, G). In root hypocotyls, the expression of TBL28 was seen in mature secondary xylem regions (Fig. 1E) and that of TBL30 was prevalent in developing secondary xylem regions (Fig. 1H). Their expression was also evident in the secondary phloem region where phloem fibers were developed.

Fig. 1

Expression analysis of TBL28 and TBL30. (A) Phylogenetic relationships of TBL28 and TBL30 with other previously studied TBL members in the Arabidopsis DUF231 family. PMR5 (TBL44), powdery mildew resistant 5; TBR, trichome birefringence; AXY4 (TBL27), altered xyloglucan 4; AXY4L (TBL22), altered xyloglucan 4-like. The Neighbor–Joining algorithm was used to construct the phylogenetic tree, and the 0.1 scale represents 10% change. The number at each node denotes the percentage of the bootstrap values from 1,000 replicates. (B) Quantitative PCR analysis of expression of TBL28 and TBL30 in different Arabidopsis organs. Error bars indicate the SD of three biological samples. (C–H) GUS reporter gene analysis of the expression of TBL28 (C–E) and TBL30 (F–H) in elongating internodes (C and F), non-elongating internodes (D and G) and root hypocotyls (E and H). The TBL28 or TBL30 gene fused with the GUS reporter gene was transformed into Arabidopsis and the transgenic plants were analyzed for GUS activity (stained blue). co, cortex; if, interfascicular fiber; sp, secondary phloem; sx, secondary xylem; xy, xylem. Scale bars = 122 µm.

To determine their subcellular localization, green fluorescence protein (GFP)-tagged TBL28 and TBL30 were co-transformed with mCherry-tagged fragile fiber8 (FRA8), which has been shown to be a Golgi-localized protein involved in xylan biosynthesis (Zhong et al. 2005), into leaf cells of Nicotiana benthamiana. Examination of fluorescent signals revealed a punctate-like pattern for both TBL28 and TBL30, which overlapped with the localization pattern of FRA8 (Supplementary Fig. S1), indicating their Golgi localization. The observed Golgi localization of TBL28 and TBL30 was consistent with computational predictions by the TMHMM2.0 program and the Golgi predictor designating them as Type II transmembrane proteins localized in the Golgi.

Expression of TBL28 rescues the esk1 mutant phenotypes

To study whether TBL28 and TBL30 played any roles in xylan acetylation, we first obtained two T-DNA lines for each of them, SALK_025118C (tbl28) and CS859671 (tbl28-2) for TBL28, and SALK_125287C (tbl30) and SALK_070679 (tbl30-2) for TBL30. Examination of the homozygous mutant lines did not reveal any apparent alterations in plant growth (Supplementary Fig. S2A), the morphology of xylem vessels and interfascicular fibers (Supplementary Fig. S2B–G) and the pattern of xylan acetylation (Supplementary Fig. S3). Since TBL28 and TBL30 are phylogentically closest to ESK1, we next examined whether they were functional paralogs of ESK1 by complementation analysis. TBL28 and TBL30, along with ESK1 as a positive control, were expressed under the promoter of the secondary wall-specific cellulose synthase A catalytic subunit 7 (CesA7) gene in the esk1 mutant. The esk1 mutation caused stunted plant growth with smaller rosette sizes and shorter inflorescence stems, a reduced stem strength and a deformation of xylem vessels (Yuan et al. 2013). Expression of TBL28 but not that of TBL30 in the esk1 mutant effectively rescued the mutant phenotypes to a level equivalent to that of the wild type (Supplementary Fig. S4). Matrix-assisted laser desorption ionization-time-of-fight mass spectrometry (MALDI-TOF-MS) analysis of xylooligomers generated from endoxylanase digestion of dimethylsulfoxide (DMSO)-extracted acetyl xylan from these transgenic plants revealed that expression of TBL28 but not that of TBL30 in esk1 restored the frequency of acetyl substitutions in xylan (Supplementary Fig. S5). Whereas the predominant ion peaks displayed by xylooligomers from esk1 and TBL30-expressing esk1 were at m/z 787 and 801 corresponding to Xyl₄ substituted with one GlcA/MeGlcA and one acetyl group, those from TBL28-expressing esk1 were shifted to m/z 1,149, 1,225 and 1,267, corresponding to Xyl₆(MeGlcA)(Ac)₃, Xyl₇(GlcA)(Ac)₂ and Xyl₆(GlcA)(Ac)₃, respectively, which is similar to those from the wild type (Supplementary Fig. S5).

We further analyzed the pattern of acetyl substitution in xylan using nuclear magnetic resonance (NMR) spectroscopy. The ¹H-NMR spectrum of the wild-type acetyl xylan displayed resonances for all the sugar residues between 3.0 and 5.5 p.p.m. and those for acetyl groups between 2.10 and 2.22 p.p.m. (Fig. 2B). Based on published data on Eucalyptus globulus xylan (Neumuller et al. 2015), the four separated resonance peaks for acetyl groups at 2.10, 2.15, 2.17 and 2.22 p.p.m. were attributed to 2,3-di-O-acetylated xylosyl residues (Xyl-2,3Ac), 3-O-monoacetylated xylosyl residues (Xyl-3Ac), 2-O-monoacetylated xylosyl residues (Xyl-2Ac) and 3-O-acetylated 2-O-GlcA-substituted xylosyl residues (Xyl-3Ac-2GlcA), respectively. Because the resonances between 2.10 and 2.22 p.p.m. are attributed to the methyl protons of acetyl groups attached to different positions of xylosyl residues and are free of any resonances for sugars, they were better suited for monitoring the quantitative changes of acetylated structural units in xylan than the resonances at the anomeric region that overlap with each other. Integration analysis of resonances between 2.10 and 2.22 p.p.m. (Fig. 2B, right panel) in wild-type xylan showed that the total degree of acetyl substitution was 49.7%, including 20.6% of the acetyl groups at O-2 and 16.1% at O-3 of monoacetylated xylosyl residues, 9.5% at O-2 and O-3 of 2,3-di-O-acetylated xylosyl residues, and 3.5% at O-3 of 2-O-GlcA-substituted xylosyl residues (Fig. 2C). The ¹H-NMR spectrum of esk1 xylan revealed a drastic reduction in the resonances at 2.10, 2.15 and 2.17 p.p.m. but not that at 2.22 p.p.m. (Fig. 2B), indicating that in addition to a reduction in 2-O- and 3-O-monoacetylation as reported previously (Yuan et al. 2013), the 2,3-di-O-acetylation of xylan in esk1 was also significantly reduced. Integration analysis demonstrated that the total degree of xylan acetylation in esk1 was reduced to 53.5% of that of the wild type (Fig. 2C), with 2-O-monoacetylation reduced to 57.3%, 3-O-monoacetylation to 32.3% and 2,3-di-O-acetylation to 17.9% of that of the wild type (Fig. 2C). It is interesting to note that the acetylation at O-3 of 2-O-GlcA-substituted xylosyl residues was doubled in the esk1 mutant compared with the wild type (Fig. 2C). These results indicate that ESK1 is involved in not only 2-O- and 3-O-monoacetylation but also 2,3-di-O-acetylation of xylan.

Fig. 2

¹H NMR spectroscopic analysis of the degree of xylan acetylation in the esk1 mutant and its complementation lines. (A) Diagram of the structural units of xylan with 2-O-monoacetylation (Xyl-2Ac), 3-O-monoacetylation (Xyl-3Ac), 2,3-di-O-acetylation (Xyl-2,3Ac) and 3-O-acetylation of 2-O-GlcA-substituted Xyl (Xyl-3Ac-2GlcA). (B) ¹H-NMR spectra of DMSO-extracted xylans from esk1, esk1 with expression of ESK1, TBL28 or TBL30 under the CesA7 promoter, and the wild type. The left panel shows the resonances corresponding to carbohydrate (3.0–5.5 p.p.m.) and acetyl groups (2.08–2.23 p.p.m.). The right panel highlights the resonances attributed to the acetyl groups in Xyl-2,3Ac (2.10 p.p.m.), Xyl-3Ac (2.15 p.p.m.), Xyl-2Ac (2.16 p.p.m.) and Xyl-3Ac-2GlcA (2.21 p.p.m.). A slight shift of the resonances for Xyl-3Ac-2GlcA occurred among different samples. (C) Determination of the degree of substitution of xylosyl residues by acetyl groups (DS_AC) based on integration of the signals for acetyl groups and carbohydrate as shown in (B). The relative amount of acetyl groups (% WT) is calculated from the ratio of acetyl groups in esk1 over that of the wild type (taken as 100). Note a significant increase in DS_AC for Xyl-2Ac, Xyl-3Ac and Xyl-2,3Ac in esk1 complemented with ESK1 or TBL28 compared with the esk1 mutant.

Consistent with the MALDI-TOF-MS data showing a restoration of the frequency of acetyl substitution in xylan (Supplementary Fig. S5), TBL28 expression in esk1 resulted in a significant elevation in the resonance signals for 2-O-monoacetylation, 3-O-monoacetylation and 2,3-di-O-acetylation of xylosyl residues, and a slight reduction in those for 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues compared with esk1 (Fig. 2B). The total degree of xylan acetylation was restored to 85.5% of that of the wild type (Fig. 2C). Although TBL30 expression in esk1 did not result in an apparent increase in the total degree of xylan acetylation, it led to an elevation in the signals for 3-O-monoacetylation of xylosyl residues compared with esk1 (Fig. 2B, C). These results demonstrate that TBL28 is a functional paralog of ESK1 involved in xylan 2-O-monoacetylation, 3-O-monoacetylation and 2,3-di-O-acetylation, whereas TBL30 might play a minor role in xylan 3-O-monoacetylation.

TBL28 and TBL30 are acetyltransferases catalyzing acetyl transfer onto xylan

The finding that expression of TBL28 and TBL30 in esk1 resulted in a differential elevation of acetylation at different positions of the xylosyl residues in xylan prompted us to investigate their enzymic activities and biochemical properties. To do so, we expressed His-tagged TBL28, TBL30 and ESK1 proteins as secreted forms in human embryonic kidney (HEK) 293F cells and subsequently purified these recombinant proteins for biochemical characterization (Fig. 3A). Acetyltransferase activity assays demonstrated that similar to ESK1, both TBL28 and TBL30 were capable of transferring acetyl groups from the ¹⁴C-labeled acetyl-CoA donor onto the Xyl₆ acceptor although their activities were 5–6 times lower than that of ESK1 (Fig. 3B). In contrast, no incorporation of ¹⁴C-labeled acetyl groups was detected for TBL28, TBL30 and ESK1 when mannohexaose or xyloglucan oligosaccharides were used as acceptors. Examination of the reaction products with MALDI-TOF-MS showed that the ESK1- and TBL28-catalyzed reaction products had two new ion species at m/z 875 and 917 with an increase in mass of 42 and 84 Da over the mass of Xyl₆ (m/z 833), respectively (Fig. 3C). Since the mass of one acetyl group (after loss of water) is 42 Da, these two new species at m/z 875 and 917 corresponded to Xyl₆ with one acetyl group and two acetyl groups, respectively. In the TBL30-catalyzed reaction products, one new ion species corresponding to Xyl₆ with one acetyl group (m/z 875) was detected (Fig. 3C). These results indicate that like ESK1, TBL28 and TBL30 are acetyltransferases catalyzing the transfer of acetyl groups onto xylan.

Fig. 3

Detection of xylan acetyltransferase activities of recombinant ESK1, TBL28 and TBL30 proteins. His-tagged ESK1, TBL28 and TBL30 proteins were expressed as secreted forms in human HEK293F cells. The proteins secreted into the culture medium were purified with nickel resin and then used in acetyltransferase activity assays. (A) SDS–PAGE detection of total proteins in the medium (medium) and nickel resin-purified ESK1, TBL28 and TBL30 (purified). A 15 µl aliquot of culture medium or 5 µg of purified protein was loaded in each lane. Proteins were detected by staining with Coomassie Blue. (B) Recombinant ESK1, TBL28 and TBL30 were able to transfer the radiolabeled acetyl group from acetyl-CoA onto the Xyl₆ acceptor. The recombinant proteins were incubated with Xyl₆ and [¹⁴C]acetyl-CoA for 16 h and the radiolabeled reaction products were counted for the amount of radioactivity. The data are the average of three independent assays. (C) MALDI-TOF-MS of reaction products catalyzed by recombinant ESK1, TBL28 and TBL30 after incubation with Xyl₆ and acetyl-CoA for 16 h. Note the appearance of two new ion peaks at m/z 875 and 917 (red arrows) with an increase in mass of 42 and 84 Da, respectively, in the enzyme-catalyzed reaction products compared with the m/z 833 of Xyl₆ alone in the control reaction without recombinant proteins. The peak at m/z 855 is attributed to the doubly sodiated ion of Xyl₆. The data shown are representative of three independent experiments.

The reaction products catalyzed by TBL28, TBL30 and ESK1 were further subjected to NMR spectroscopy to examine to which positions of the xylosyl residues the acetyl groups were added. It was revealed that the resonance signals attributed to 2-O-monoacetylated, 3-O-monoacetylated and 2,3-di-O-acetylated xylosyl residues were present in the reaction products catalyzed by TBL28 and ESK1, whereas those for 2-O-monoacetylated and 3-O-monoacetylated xylosyl residues were present in that of TBL30 (Fig. 4A; left panel). Integration analysis showed the prevalence of 2-O-monoacetylation of xylosyl residues in the reaction products catalyzed by TBL28 and ESK1 and the prevalence of 3-O-monoacetylation in that of TBL30 (Fig. 4B). We next tested whether TBL28, TBL30 and ESK1 were able to catalyze the 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues. When GlcA-substituted Xyl₄ was used as the acceptor in the reactions, no resonance signals for acetylated xylosyl residues were detected in the reaction products catalyzed by TBL28, TBL30 or ESK1 (Fig. 4A, right panel). These results demonstrate that TBL28 and ESK1 are acetyltransferases catalyzing xylan 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation with a preference of 2-O-monoacetylation and that TBL30 is an acetyltransferase catalyzing xylan 2-O- and 3-O-monoacetylation with a preference of 3-O-monoacetylation.

Fig. 4

Structural analysis of the pattern of acetyl substitutions in ESK1-, TBL28- and TBL30-catalyzed reaction products using ¹H-NMR spectroscopy. The data shown are representative of three independent experiments. (A) ¹H-NMR spectra of reaction products catalyzed by ESK1, TBL28 and TBL30 after incubation with acetyl-CoA and xylohexaose (Xyl₆; left panel) or GlcA-substituted xylotetraose [(GlcA)Xyl₄; right panel] for 16 h. The control reaction contained acetyl-CoA and the acceptor only without any proteins. The resonances corresponding to Xyl-2Ac, Xyl-3Ac and Xyl-2,3Ac are indicated. In the reactions using the GlcA-substituted Xyl₄ receptor (right panel), the resonance peak attributed to the acetyl group of acetyl-CoA was present (indicated by an arrowhead in the control) since it was not removed from the reaction products. (B) Relative amount of acetyl substitutions at different positions of xylosyl residues catalyzed by ESK1, TBL28 and TBL30 based on integration of the acetyl signals shown in the left panel of (A).

To gain further insight into the biochemical properties of TBL28, TBL30 and ESK1, we analyzed their acetyltransferase activities with different incubation times, various protein amounts and acceptor concentrations, and xylooligomer acceptors with various degrees of polymerization (DPs). The acetyltransferase activities of TBL28, TBL30 and ESK1 toward the Xyl₆ acceptor were both time and protein concentration dependent (Fig. 5A, B). Although they exhibited little acetyltransferase activities toward xylose and xylobiose, TBL28, TBL30 and ESK1 readily catalyzed the acetyl transfer onto xylooligomers with a DP of ≥3 and their activities increased with the increasing DP of xylooligomers (Fig. 5C). The acetyl transfer onto Xyl₆ by TBL28, TBL30 and ESK1 was acceptor concentration dependent, with apparent K_m values of 0.28, 1.55 and 0.39 mM, and V_max values of 7.0, 6.6 and 23.8 pmol min^–1 mg^–1 protein, respectively (Fig. 5D). These results indicate that TBL28 and ESK1 have a much higher affinity for Xyl₆ than TBL30 and that ESK1 exhibited a much higher rate of acetyltransferase activity than TBL28 and TBL30.

Fig. 5

Characterization of the biochemical properties of ESK1, TBL28 and TBL30 acetyltransferases. Recombinant ESK1, TBL28 and TBL30 were incubated with ¹⁴C-labeled acetyl-CoA and Xyl₆ (A, B and D) or xylooligomers with a different degree of polymerization (DP) (C) for 4 h unless otherwise indicated, and the reaction products were assayed for the amount of ¹⁴C-labeled acetyl transferred onto the xylooligomer acceptor. All data shown are the average of three independent assays. (A) Time course of the transfer of acetyl groups onto Xyl₆ by ESK1, TBL28 and TBL30. (B) The level of acetyltransferase activities of ESK1, TBL28 and TBL30 is correlated with their protein amount. (C) The acetyltransferase activities of ESK1, TBL28 and TBL30 toward xylooligomers with different DPs. (D) Characterization of the kinetic properties of ESK1, TBL28 and TBL30. The acetyltransferase activities of ESK1, TBL28 and TBL30 toward various concentrations of Xyl₆ were assayed and the data were displayed as Lineweaver–Burk plots and calculated for the K_m and V_max values.

Additional xylan acetyltransferases with differential regiospecificities

Previous genetic studies have implicated the involvement of six other TBL proteins, namely TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35, in xylan acetylation in Arabidopsis (Yuan et al. 2016a, Yuen et al. 2016b, Yuan et al. 2016c). Having demonstrated that TBL28, TBL30 and ESK1 are acetyltransferases catalyzing xylan 2-O- and 3-O-monoacetylation and/or 2,3-di-O-acetylation with differential specificities, we next investigated whether these six TBL proteins were acetyltransferases catalyzing the acetylation of xylan. His-tagged recombinant proteins of TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35 were expressed as secreted forms in HEK293F cells (Fig. 6A) and the purified recombinant proteins were analyzed for their acetyltransferase activities. When incubated with [¹⁴C]acetyl-CoA and the Xyl₆ acceptor, TBL3, TBL31, TBL34 and TBL35 were able to transfer the radiolabeled acetyl groups onto Xyl_6, albeit at a much lower efficiency than ESK1 (Fig. 6B), indicating that they possessed acetyltransferase activities catalyzing acetyl transfer onto xylan. In contrast, TBL32 and TBL33 did not exhibit any acetyltransferase activities toward Xyl₆ (Fig. 6B). None of these TBL proteins was able to incorporate ¹⁴C-labeled acetyl groups onto mannohexaose or xyloglucan oligosaccharides. Examination of the reaction products with NMR spectroscopy showed that the resonance signals attributed to 2-O-monoacetylated, 3-O-monoacetylated and 2,3-di-O-acetylated xylosyl residues were detected in the reactions catalyzed by TBL3, TBL31 and TBL34, and, among them, the signals for 3-O-monoacetylated xylosyl residues were dominant (Fig. 6C, left panel). Whereas only the resonance signals for 2,3-di-O-acetylated xylosyl residues were evident for TBL35, no signal peaks for acetylated xylosyl residues were observed in the reactions catalyzed by TBL32 and TBL33 (Fig. 6C, left panel). We next used GlcA-substituted Xyl₄ as the acceptor to test whether any of these proteins were able to catalyze the 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues. It was found that only the TBL32- and TBL33-catalyzed reaction products displayed resonance signals attributed to 3-O-acetylated 2-O-GlcA-substituted xylosyl residues (Fig. 6C, right panel). Similar to TBL28, TBL30 and ESK1 (Fig. 4A, right panel), TBL3, TBL31, TBL34 and TBL35 were unable to transfer acetyl groups onto GlcA-substituted Xyl₄ (Fig. 6C, right panel). These results demonstrate that TBL3, TBL31 and TBL34 are xylan acetyltransferases catalyzing 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation with preference for 3-O-monoacetylation, TBL35 is a xylan acetyltransferase catalyzing 2,3-di-O-acetylation, and TBL32 and TBL33 are xylan acetyltransferases responsible for 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues.

Fig. 6

Structural analysis of the pattern of acetyl substitutions in xylooligomers catalyzed by TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35. (A) Detection of recombinant TBL proteins by SDS–PAGE and Coomassie Blue staining. His-tagged recombinant proteins were expressed as secreted forms in human HEK293F cells and purified with nickel resin. A 5 µg aliquot of purified recombinant proteins was loaded in each lane. Note that TBL32 migrated more slowly than other proteins possibly due to its high level of glycosylation. (B) Detection of acetyltransferase activities of the recombinant TBL proteins by incubation with ¹⁴C-labeled acetyl-CoA and the Xyl₆ acceptor for 16 h. The data are the average of three independent assays. (C) ¹H-NMR spectra of the reaction products catalyzed by the recombinant TBL proteins incubated with acetyl-CoA and the Xyl₆ acceptor (left panel) or the GlcA-substituted Xyl₄ acceptor (right panel) for 16 h. The control reaction contained acetyl-CoA and the acceptor only without any proteins. The acetylated structural units in the reaction products, including Xyl-2Ac, Xyl-3Ac, Xyl-2,3Ac and Xyl-3Ac-2GlcA, were labeled. In the reactions using the GlcA-substituted Xyl₄ receptor (right panel), the resonance peak attributed to the acetyl group of acetyl-CoA is present (indicated by an arrowhead in the control) since it was not removed from the reaction products. The data shown are representative of three independent experiments.

Mutations of the conserved GDS and DXXH motifs in ESK1 abolish its acetyltransferase activity

All 46 members of the Arabidopsis DUF231 family contain a TBL domain at the N-terminus that has a conserved GDS (Gly–Asp–Ser) motif and a DUF231 domain at the C-terminus that harbors a conserved DXXH (Asp–X–X–His) motif (Bischoff et al. 2010; Supplementary Fig. S6). To ascertain the importance of these motifs, we performed site-specific amino acid substitutions with alanine (A) of each of the conserved residues in the GDS and DXXH motifs of ESK1 by site-directed mutagenesis. Five mutant proteins of ESK1 thus created, namely M1 (G214A), M2 (D215A), M3 (S216A), M4 (D462A) and M5 (H465A), were expressed in HEK293F cells as His-tagged secreted recombinant proteins (Fig. 7A, B). We also expressed the mutated version of the ESK1 protein, esk1-1 (G145R), which is encoded by the mutant allele in the esk1 mutant used for genetic analysis in this study and a previous report (Yuan et al. 2013). Acetyltransferase activity assay using [¹⁴C[acetyl-CoA and Xyl₆ revealed that none of these mutant proteins exhibited any detectable acetyltransferase activities (Fig. 7C), indicating that all of the conserved residues in the GDS and DXXH motifs in ESK1 are essential for its xylan acetyltransferase activity and that the missense mutation occurring in the esk1 mutant completely abolishes its acetyltransferase activity.

Fig. 7

Effects of mutations of the GDS and DXXH motifs on ESK1 acetyltransferase activity. (A) Diagram of the ESK1 protein. The transmembrane (TM) helix, the TBL domain and the DUF231 domain are indicated, with the positions of amino acid residues at the top. The positions of the GDS and DXXH motifs and their mutations are shown at the bottom. The glycine at position 145 was mutated to arginine (G145R) in the esk1-1 mutant protein and each of the conserved amino acids in the GDS motif and the DXXH motif was mutated to alanine as indicated (M1, M2, M3, M4 and M5). (B) Detection of mutated recombinant ESK1 proteins by SDS–PAGE and Coomassie Blue staining. A 5 µg aliquot of purified proteins was loaded in each lane. (C) Xylan acetyltransferase activity assay of the mutated ESK1 proteins by incubation with the Xyl₆ acceptor and ¹⁴C-labeled acetyl-CoA for 16 h. Note the loss of acetyltransferase activity in the mutated ESK1 proteins. The data are the average of three independent assays.

Discussion

Previous genetic studies have demsontrated that xylan acetylation is a complex process involving at least seven secondary wall-associated TBL genes, i.e. ESK1, TBL3, TBL31, TBL32, TBL33, TBL34 and TBL35, mutations of which cause differential effects on the degree of acetylation at different positions of xylosyl residues (Zhong and Ye 2015). In this report, we have demonstrated that two additional secondary wall-associated TBL genes, TBL28 and TBL30, are also involved in xylan acetylation. More importantly, we have established that these nine DUF231-containing proteins are xylan acetyltransferases with differential regiospecificities, which provides the first line of biochemical evidence that these acetyltransferases mediate the regiospecific acetylation of xylan. Our findings also shed light on the elucidation of the biochemical functions of other members of the DUF231 family. Among the 46 DUF231-containing proteins in Arabidopsis, only two other members, AXY4 and AXY4L, have been shown to be involved in acetylation of xyloglucan, another cell wall polysaccharide, but biochemical proof of their enzymatic activities is still lacking (Gille et al. 2011). Although the functions of the remaining 35 members are yet to be identified, it is possible that they are acetyltransferases catalyzing the regiospecific acetylation of other cell wall polysaccharides, including glucomannan (mono- or di-acetylated at O-2, O-3 and O-6 of mannose residues) (Teleman et al. 2003, Campestrini et al. 2013) and pectins (homogalacturonan and rhamnogalacturonan I; mono- or di-acetylated at O-2 and O-3 of galacturonic acid residues) (Ishii 1997, Ralet et al. 2005).

Our structural analysis of the acetyltransferase-catalyzed reaction products has led to the discovery that these nine acetyltransferases have differential positional specificities in catalyzing the regiospecific acetylation of xylan. Specifically, ESK1 and TBL28 catalyze 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation with positional preference for 2-O-monoacetylation, TBL3, TBL31 and TBL34 also mediate 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation but with positional preference for 3-O-monoacetylation, TBL30 carries out 2-O- and 3-O-monoacetylation with positional preference for 3-O-monoacetylation, TBL35 catalyzes 2,3-di-O-acetylation, and TBL32 and TBL33 are responsible for 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues (Table 1). Since these proteins are biochemically proven to be xylan acetyltransferases, they are accordingly named as XOAT1 (ESK1), XOAT2 (TBL28), XOAT3 (TBL30), XOAT4 (TBL3), XOAT5 (TBL31), XOAT6 (TBL32), XOAT7 (TBL33), XOAT8 (TBL34) and XOAT9 (TBL35) (Fig. 1A; Table 1). The finding that ESK1 catalyzes both mono- and di-acetylation at O-2 and O-3 is consistent with the xylan acetylation defect in the esk1 mutant showing a drastic reduction in xylan 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation (Fig. 2). A reduction in xylan 2,3-di-O-acetylation in the esk1 mutant was not observed in a previous report (Yuan et al. 2013) because the acetylated structural units examined were at the anomeric regions in the NMR spectra where the resonances for the reducing end xylosyl residues overlap with those for the 2,3-di-O-acetylated xylosyl residues, thus masking the differences (Zhong et al. 2014). Because the resonances between 2.10 and 2.22 p.p.m. attributed to the methyl protons of acetyl groups are free of any resonances for sugars, a drastic reduction in the acetylated structural units for 2,3-di-O-acetylation in addition to 2-O- and 3-O-monoacetylation was revealed in the esk1 mutant (Fig. 2). In addition, 2,3-di-O-acetylation was not observed in the ESK1-catalyzed reactions in another previous study (Urbanowicz et al. 2014), which could be due to the fact that the acetyl donor (acetyl-CoA) used in the reaction gives a resonance at 2.10 in the NMR spectrum, the same position for that of 2,3-di-O-acetylated xylosyl residues, thus masking it. Removal of acetyl-CoA from the reaction products enabled us to detect the 2,3-di-O-acetylated xylosyl residues in the ESK1-catalyzed reactions (Fig. 4A). TBL28 displays a similar regiospecificity for xylan acetylation as ESK1 (Fig. 4), which is in agreement with the complementation data showing that TBL28 expression rescues the esk1 mutant defects (Fig. 2; Supplementary Figs. S4, S5). In contrast, TBL30 possesses an acetyltransferase activity for xylan 2-O- and 3-O-monoacetylation only, and its expression is unable to rescue the esk1 defects, probably due to its lack of activity for xylan 2,3-di-O-acetylation and its much lower substrate binding affinity than ESK1 and TBL28.

Table 1

Summary of the differential positional specificities and the relative activities of xylan acetyltransferases

	Xyl-2Ac	Xyl-3Ac	Xyl-2,3Ac	Xyl-3Ac-2GlcA
ESK1/XOAT1	++++++++	+++++	+++++	–
TBL28/XOAT2	++	+	+	–
TBL30/XOAT3	+	++	–	–
TBL3/XOAT4	++	+++	+	–
TBL31/XOAT5	+	++	+	–
TBL32/XOAT6	–	–	–	+++++
TBL33/XOAT7	–	–	–	+++
TBL34/XOAT8	++	+++	+	–
TBL35/XOAT9	–	–	++	–

Xyl-2Ac, 2-O-monoacetylation of xylosyl residues; Xyl-3Ac, 3-O-monoacetylation of xylosyl residues; Xyl-2,3Ac, 2,3-di-O-acetylation of xylosyl residues; Xyl-3Ac-2GlcA, 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues.

‘+’ indicates the relative specific activity of acetyltransferases based on the activity assay data in Figs. 3–6. ‘–’ indicates the absence of detectable acetyltransferase activity.

The findings that TBL3, TBL31 and TBL34 are acetyltransferases catalyzing 2-O- and 3-O-monoacetylation and 2,3-di-O-acetylation with positional preference for 3-O-monoacetylation and that TBL35 is an acetyltransferase catalyzing 2,3-di-O-acetylation are also consistent with the xylan acetylation defects observed in the double mutants tbl3 tbl31 and tbl34 tbl35. Both of these double mutants displayed a reduction in the acetylated structural units for 3-O-acetylation and 2,3-di-O-acetylation (Yuan et al. 2006b, Yuan et al. 2016c). It is interesting to note that only 2,3-di-O-acetylated xylosyl residues are accumulated in the TBL35-catalyzed reactions; one possible explanation is that TBL35 adds an acetyl group to one position (e.g. O-2) immediately followed by adding another acetyl group to an adjacent position (e.g. O-3) of the same xylosyl residue, thus leading to the production of 2,3-di-O-acetylated xylosyl residues without detectable accumulation of 2- or 3-O-monoacetylated xylosyl residues. Among the nine xylan acetyltransferases, TBL32 and TBL33 are the only ones capable of catalyzing 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues (Fig. 6), which is corroborated by genetic analysis showing that their simultaneous mutation leads to a complete loss of 3-O-acetylated 2-O-GlcA-substituted xylosyl residues in xylan (Yuan et al. 2016a).

The acetyltransferase activity assay of recombinant proteins revealed that although all DUF231 members in the ESK1 group exhibit acetyltransferase activities albeit with different regiospecificities, ESK1 has a much higher activity than other members, indicating that ESK1 plays a dominant role in xylan acetylation. This finding is in agreement with previous mutational analysis showing that the esk1 mutant alone displays a significant reduction in xylan acetylation and a retardation in plant growth (Yuan et al., 2013), but single mutation of other members in the ESK1 group has no apparent effect on xylan acetylation and plant growth (Yuan et al. 2006a, Yuan et al. 2006b, Yuan et al. 2006c; Supplementary Figs. S2, S3). However, double mutations of TBL3 and TBL31, TBL32 and TBL33, and TBL34 and TBL35 cause minor defects in xylan acetylation and these double mutations in the esk1 mutant background further exacerbate the defect in xylan acetylation conferred by the esk1 mutant alone (Yuan et al. 2006a, Yuan et al. 2006b, Yuan et al. 2006c). In particular, the triple mutant tbl32 tbl33 esk1 has a xylan acetyl level reduced down to 15% of that of the wild type, and subsequently results in severely collapsed vessels, stunted plant growth and altered secondary wall structure (Yuan et al. 2006a). These findings indicate that although the ESK1 acetyltransferase activity is responsible for catalyzing addition of the bulk of acetyl groups onto xylan, other DUF231 acetyltransferase activities also contribute to this process albeit with a lower level.

It should be pointed out that although it is considered that the acetyl groups may migrate between different positions of the xylosyl residues under weak acidic conditions (Evtuguin et al. 2003), the acetyl xylan isolation procedure that was used in this study was shown not to cause migration of acetyl groups (Evtuguin et al. 2003). In addition, incubation of acetylated xylooligomers in the buffer that was used for the acetyltransferase assays did not result in any migration of acetyl groups (Supplementary Fig. S7). Therefore, the observed acetylated structural units of xylan reported in this study were unlikely to be due to spontaneous migration of acetyl groups between different positions of xylosyl residues, although we could not completely exclude the possibility of a low level of acetyl migration. Even if there were any spontaneous migration of acetyl groups, it should primarily occur between 2-O- and 3-O-monoacetylation and thereby have no effect on 2,3-di-O-acetylation and 3-O-acetylation of 2-O-GlcA-substituted xylosyl residues. Nevertheless, the acetyltransferase assay method used in this study represents the best possible approach to study xylan acetylation and it is currently not feasible to design an assay to block possible migration of acetyl groups. Furthermore, the regiospecific acetyltransferase activities exhibited by the recombinant TBL proteins are in agreement with their corresponding mutant phenotypes, further validating the observed acetyltransferase activity data. It was reported in a previous study that there was a slight lag (by 30 min) in the appearance of 3-O-monoacetylation compaed with 2-O-monoacetylation in the ESK1-catalyzed reaction and therefore it was concluded that the acetylation at O-3 was the result of spontaneous migration of the acetates between O-2 and O-3 (Urbanowicz et al. 2014). However, the reason for the lag could be due to a difference in the rate of transfer of acetyl groups to O-2 and O-3 as the acetyl transfer in vitro is a very slow process (Urbanowicz et al. 2014), and therefore it is uncertain whether the observed differential appearance of 2-O- and 3-O-acetyl groups is due to acetyl migration or differential ESK activity toward O-2 and O-3 of xylosyl residues.

All 46 members of Arabidopsis DUF231-containing proteins harbor two conserved motifs, the GDS motif located in the TBL domain and the DXXH motif in the DUF231 domain (Bischoff et al. 2010; Fig. 7a), but their functional signficance in these proteins is unclear. These two conserved motifs are also present in a family of GDSL-esterases/lipases, and the serine residue in the GDS motif together with histidine and aspartate in the DXXH motif have been shown to function as a catalytic triad for the mechanism of action of serine esterases and proteases (Pfeffer et al. 2013). The GDS and DXXH motifs are also found in a number of proteins involved in O-acetylation of carbohydrates in several other organisms, including fungal and human CAS1 proteins required for O-acetylation of glucuronoxylomannan and sialic acid, respectively (Janbon et al. 2001, Baumann et al. 2015), OatA and OatB responsible for O-acetylation of N-acetylmuramoyl and N-acetylglucosamyl residues, respectively, in Gram-positive bacteria (Bera et al. 2005, Crisostomo et al. 2006), PatB for O-acetylation of peptidoglycan in Gram-negative bacteria (Moynihan and Clarke, 2014), and AlgJ and AlgX for O-acetylation of alginate in Pseudomonas aeruginosa (Baker et al. 2014). The Ser–His–Asp catalytic triad has been shown to be required for the acetylesterase activity of AlgJ as well as for the acetyltransferase activities of AlgX and PatB (Baker et al. 2014, Moynihan and Clarke 2014). Our demonstration that the conserved residues in the GDS and DXXH motifs of ESK1 are required for its xylan acetyltransferase activities indicates that ESK1 may have evolved to utilize the Ser–His–Asp catalytic triad for the transfer of the acetyl group onto the hydroxyl of xylosyl residues. Since all of the conserved residues in the GDS and DXXH motifs are invariable in all 46 members of Arabidopsis DUF231 proteins (Bischoff et al. 2010; Supplementary Fig. S6), it is likely that all of them may have adopted a similar Ser–His–Asp catalytic triad for their mechanism of action. Mutation of the glycine residue at position 145 to arginine as in the esk1-1 mutant (Xin et al. 2007), which was used for genetic analysis in this study and in a previous report (Yuan et al. 2013), also completely abolishes the activity of ESK1. Although the exact role of Gly145 in the catalytic mechanism of ESK1 is unclear, this glycine residue is invariable in all 46 Arabidopsis DUF231 proteins and may also be critical for the functions of other DUF 231 proteins (Supplementary Fig. S6).

The available evidence indicates that xylan O-acetylation may have evolved to employ a two-component mechanism analogous to that for peptidoglycan O-acetylation in Gram-negative bacteria, in which a large transmembrane protein PatA transfers the acetyl donor to the periplasmic space where the peripheral membrane protein PatB functions as an O-acetyltransferase to transfer the acetyl group to peptidoglycan (Moynihan and Clarke 2014). By analogy, the four Arabidopsis RWA proteins, which are large transmembrane proteins with 12–13 transmembrane helices (Lee et al. 2011), may function as transporters translocating the acetyl donor to the Golgi where the xylan acetyltransferases catalyze the acetyl transfer onto xylan in a regiospecific manner. Our discovery that a group of xylan O-acetyltransferases mediate the regiospecific acetylation of xylan and that the conserved GDS and DXXH motifs in ESK1 are essential for its acetyltransferase activity not only greatly enriches our understanding of the complex biochemical mechanism that plants use for xylan O-acetylation but also sets the stage for further dissection of functional roles of other DUF231 proteins in controlling the O-acetylation of cell wall polysaccharides.

Materials and Methods

Gene expression analysis

For quantitative PCR analysis of gene expression, total RNA was isolated from Arabidopsis thaliana leaves, flowers, stems and roots of 6-week-old plants using the Qiagen RNA isolation kit (Qiagen). Stems used included the top (rapidly elongating internodes), middle (internodes near cessation of elongation) and bottom (non-elongating internodes) parts. After treatment with DNase, the RNA was used for reverse transcription–PCR and real-time quantitative PCR analysis. The primers used for PCR analysis were 5′-cattactgcattgtctgagtatcg-3′ and 5′-tcatcttcgagaaataatatgtgt-3′ for TBL28, and 5′-tacgaaatcgcactgaatgcgacc-3′ and 5′-attttgtagatgatatacagagag-3′ for TBL30. The copy number of transcripts of each gene was calculated based on the PCR threshold cycle numbers of the sample and the standard curve of the plasmid control. Three separate pools of samples for each tissue were used for RNA isolation and real-time quantitative PCR analysis.

For GUS reporter gene expression analysis, the gene sequences of TBL28 and TBL30, including the 3 kb 5′ upstream sequence, the entire coding region and the 2 kb 3′ downstream sequence were used. The GUS reporter gene was inserted in-frame before the stop codon of each gene and the TBL28 or TBL30 gene–GUS cassette thus created was cloned into a modified binary vector pBI101 (Clontech). Wild-type Arabidopsis plants were transformed with Agrobacterium containing these GUS reporter constructs and the transformed seeds were selected for transgenic plants by growing in a medium containing kanamycin. Stems and root hypocotyls from 7-week-old plants were examined for GUS gene expression as described previously (Zhong et al. 2005). At least 20 independent transgenic plants were analyzed for GUS staining, and representative data were presented.

Protein subcellular localization

The full-length cDNA sequence of TBL28 or TBL30 was cloned in-frame into the N-terminal end of the GFP gene driven by the Cauliflower mosaic virus (CaMV) 35S promoter in a modified pBI121 binary vector. The GFP-tagged constructs thus created had GFP located at the C-terminus of TBL28 (TBL28–GFP) or TBL30 (TBL30–GFP). These GFP-tagged constructs together with a mCherry-tagged FRA8 construct were transformed into leaves of N. benthamiana by Agrobacterium-mediated infiltration. After 2 d, the infiltrated leaf areas were imaged for GFP and mCherry fluorescence signals under a confocal microscope. At least five independently infiltrated leaves were used for examination of fluorescence signals, and representative images were shown.

Complementation analysis

The full-length cDNA of TBL28, TBL30 or ESK1 was ligated between the 2 kb CesA7 promoter and the nopaline synthase (NOS) terminator in a binary vector to create complementation constructs. These constructs were introduced into the esk1 mutant by Agrobacterium-mediated transformation. More than 100 transgenic plants were generated for examination of their phenotypes and subsequent cell wall analysis. For stem strength analysis, the bottom parts (2 cm) of mature stems of wild type (8 weeks old), esk1 (10 weeks old) and transgenic esk1 transformed with the complementation constructs (8 weeks old) were measured for breaking force using a digital force/length tester (Zhong et al. 1997). Stems from at least 20 plants for each construct were used for measurement. For examination of vessel morphology, thin sections (1 µm thick) from the bottom parts (0.5 cm) of mature stems were cut and stained with toluidine blue (Burk et al. 2006). Stems of at least 10 plants were examined, and representative data are shown.

Cell wall isolation

Inflorescence stems from at least 20 plants were pooled for cell wall isolation. Stems were first ground into powder in liquid N₂ and then extracted sequentially with 70% ethanol, 100% ethanol and 100% acetone to obtain alcohol-insoluble cell walls (Zhong et al. 2005). The cell walls were extracted with DMSO to release acetylated xylan (Evtuguin et al. 2003), which was further digested with β-endoxylanase M6 (Megazyme) to generate acetylated xylooligomers for structural analysis. For each experiment, cell walls isolated from three separate pools of stems were used for structural analysis of acetylated xylan by MALDI-TOF-MS and NMR.

Recombinant protein expression

The cDNA sequences of TBL28, TBL30, TBL3, TBL31, TBL32, TBL33, TBL34, TBL35 and ESK1 with deletion of the N-terminal transmembrane domain was cloned between the murine Igκ chain leader sequence (for protein secretion) and the c-myc epitope and six tandem histidine tags in the pSecTag2 mammalian expression vector (Invitrogen). The expression constructs were tranfected into HEK293F cells using the Invitrogen FreeStyle 293 Expression System according to the manufacturer’s protocol. The expressed recombinant proteins were purified from the culture medium by passing through the nickel resin column and stored at –80°C until used. The calculated molecular weights of recombinant proteins including the c-myc epitope and polyhistidine tags are 54.1 kDa (ESK1), 44.8 kDa (TBL28), 50.1 kDa (TBL30), 48.6 kDa (TBL3), 47.3 kDa (TBL31), 49.4 kDa (TBL32), 47.5 kDa (TBL33), 46.6 kDa (TBL34) and 52.7 kDa (TBL35). The purified recombinant proteins (5 µg) were examined by SDS–PAGE and staining with Coomassie Blue.

Acetyltransferase activity assay

The purified recombinant proteins were assayed for acetyltransferase activities using the ¹⁴C-labeled acetyl-CoA as an acetyl donor and xylooligomers with various DPs as acceptors. The recombinant proteins (20 µg for radiodetection and 100 µg for NMR analysis unless otherwise indicated) were incubated in a reaction mixture containing 50 mM HEPES buffer (pH 7.0), acetyl-1,2-[¹⁴C]CoA (0.1 µCi; American Radiolaled Chemicals) or non-radiolabeled acetyl CoA (1 mM) and various amounts of xylooligomers (Xyl₁–Xyl₆) or GlcA-substituted xylotetraose [(GlcA)Xyl₄] (30 µg for radiodetection and 200 µg for NMR analysis unless otherwise indicated) at 37°C. After incubation, xylooligomers were separated from acetyl-1,2-[¹⁴C]CoA by passing through Dowex 1X4 anion exchange resin and then counted for the amount of radioactivity that was incorporated onto xylooligomers with a Perkin Elmer scintillation counter. The recombinant proteins were also assayed for acetyltransferase activities using mannohexaose and xyloglucan oligosaccharides as acceptors as described above. Xylooligomers and mannohexaose were purchased from Megazyme, xyloglucan oligosaccharides were prepared from digestion with endo-1,4-β-glucanase of KOH-extracted cell walls of the wild-type Arabidopsis stems (Zhong et al. 2017), and (GlcA)Xyl₄ was generated from digestion with endo-1,4-β-xylanase of KOH-extracted cell walls of the gxm1/2/3 triple mutant that is devoid of MeGlcA substitution in xylan (Lee et al. 2012, Yuan et al. 2014). For MALDI-TOF-MS and NMR analyses, products in the reactions with non-GlcA-substituted xylooligomers as acceptors were passed through Dowex 1X4 resin for the removal of acetyl-CoA, whereas those in the reaction with GlcA-substituted acceptors were not due to the negative charge of GlcA residues. For each recombinant protein, the reaction products from three separate reactions were used for subsequent analyses.

MALDI-TOF-MS

The masses of acetylated xylooligomers were determined by MALDI-TOF-MS analysis as described (Zhong et al. 2005). The samples (1 µg) were mixed (1 : 1, v/v) with the MALDI matrix 2,5-dihydroxybenzoic acid and then subjected to MALDI-TOF-MS. The spectra were the averages of 250 laser shots. Xylooligomers from three separate pools of cell walls or reaction products were analyzed for each sample, and representative spectra are shown.

¹H-NMR spectroscopy

The acetylated structural units of xylooligomers (200 µg) were analyzed with a Bruker Avance III HD 400 MHz spectrometer. The NMR spectra were recorded with 512 transients. The proton positions and residue identities in the NMR spectra were assigned based on the NMR spectra data for acetylated xylan from E. globulus (Neumuller et al. 2015). Xylooligomers from three separate pools of cell walls or reaction products were examined for each sample, and representative NMR spectra were shown.

Accession numbers

The Arabidopsis genome initiative locus identifiers for the genes investigated in this study are At3g55990 (ESK1/XOAT1), At2g40150 (TBL28/XOAT2), At2g40160 (TBL30/XOAT3), At5g01360 (TBL3/XOAT4), At1g73140 (TBL31/XOAT5), At3g11030 (TBL32/XOAT6), At2g40320 (TBL33/XOAT7), At2g38320 (TBL34/XOAT8) and At5g01620 (TBL35/XOAT9).

Supplementary data

Supplementary data are available at PCP online.

Funding

This work was funded by the US Department of Energy, Office of Science, Office of Basic Energy Sciences [grant No. DE-FG02–03ER15415].

Abbreviations

Abbreviations
 
• Ac

  acetyl

 
• AXY

  altered xyloglucan

 
• DP

  degree of polymerization

 
• DMSO

  dimethylsulfoxide

 
• ESK

  eskimo

 
• FRA8

  fragile fiber8

 
• GFP

  green fluorescent protein

 
• GlcA

  glucuronic acid

 
• GUS

  β-glucuronidase

 
• HEK

  human embryonic kidney

 
• MALDI-TOF-MS

  matrix-assisted laser desorption ionization-time-of-flight mass spectrometry

 
• MeGlcA

  4-O-methyl-glucuronic acid

 
• NMR

  nuclear magnetic resonance

 
• RWA

  reduced wall acetylation

 
• TBL

  trichome birefringence-like

 
• XOAT

  xylan O-acetyltransferase

 
• Xyl

  xylose

Acknowledgments

We thank the Arabidopsis Biological Resource Center for the T-DNA mutant lines, Dr. D. Philips at the UGA PAMS Facility for technical help, and two anonymous reviewers for their constructive comments and suggestions.

Disclosures

The authors have no conflicts of interest to declare.