Identification and expression analysis of the glycosyltransferase GT43 family members in bamboo reveal their potential function in xylan biosynthesis during rapid growth

Xylan is one of the most abundant hemicelluloses and can crosslink cellulose and lignin to increase the stability of cell walls. A number of genes encoding glycosyltransferases play vital roles in xylan biosynthesis in plants, such as those of the GT43 family. However, little is known about glycosyltransferases in bamboo, especially woody bamboo which is a good substitute for timber. A total of 17 GT43 genes (PeGT43–1 ~ PeGT43–17) were identified in the genome of moso bamboo (Phyllostachys edulis), which belong to three subfamilies with specific motifs. The phylogenetic and collinearity analyses showed that PeGT43s may have undergone gene duplication, as a result of collinearity found in 12 pairs of PeGT43s, and between 17 PeGT43s and 10 OsGT43s. A set of cis-acting elements such as hormones, abiotic stress response and MYB binding elements were found in the promoter of PeGT43s. PeGT43s were expressed differently in 26 tissues, among which the highest expression level was found in the shoots, especially in the rapid elongation zone and nodes. The genes coexpressed with PeGT43s were annotated as associated with polysaccharide metabolism and cell wall biosynthesis. qRT–PCR results showed that the coexpressed genes had similar expression patterns with a significant increase in 4.0 m shoots and a peak in 6.0 m shoots during fast growth. In addition, the xylan content and structural polysaccharide staining intensity in bamboo shoots showed a strong positive correlation with the expression of PeGT43s. Yeast one-hybrid assays demonstrated that PeMYB35 could recognize the 5′ UTR/promoter of PeGT43–5 by binding to the SMRE cis-elements. PeGT43s were found to be adapted to the requirement of xylan biosynthesis during rapid cell elongation and cell wall accumulation, as evidenced by the expression profile of PeGT43s and the rate of xylan accumulation in bamboo shoots. Yeast one-hybrid analysis suggested that PeMYB35 might be involved in xylan biosynthesis by regulating the expression of PeGT43–5 by binding to its 5′ UTR/promoter. Our study provides a comprehensive understanding of PeGT43s in moso bamboo and lays a foundation for further functional analysis of PeGT43s for xylan biosynthesis during rapid growth.

accumulation and elaborate proportions of lignin, cellulose and hemicellulose. Cellulose forms the main structural framework of the cell wall. Hemicellulose covers cellulose microfibrils and crosslinks with them to form a stable network by hydrogen bonding. Lignin is a heterogeneous mixture of randomly polymerized phenolic monolignols that is interspersed and crosslinked with hemicellulose to make cell wall stronger. The biosynthesis of these major cell wall components is regulated by a set of transcription factors, some of which can regulate both cellulose and hemicellulose synthesis as well as lignin synthesis, such as MYB26, MYB46/83, SND1 and KNAT7 [1][2][3][4]. Genes associated with lignin, cellulose and hemicellulose biosynthesis in moso bamboo have been identified, such as the members of NAC, MYB, CESA, LAC and UGDH gene families [5][6][7][8].
It has been shown that the hemicelluloses of moso bamboo are dominated by xylan [9,10], which is composed of 1,4-β-D-xylose as the backbone, with glucuronic acid, arabinose and acetylation in the side chains [11]. Glycosyltransferases (GTs) play important roles in the biosynthesis of xylans, such as members of the GT8, GT43, GT47, GT61 and GT75 families. A set of IRREGULAR XYLEM (IRX) genes, IRX7, IRX8, IRX9, IRX9L, IRX10, IRX10L, IRX14 and IRX14L encoding GTs in plants, have been found to be for xylan biosynthesis [12]. The irx7, irx8, irx9, and irx14 mutants in Arabidopsis (Arabidopsis thaliana) exhibited a collapsed xylem phenotype and a significant decrease of xylose in the cell wall, whereas irx10 showed a moderate degree of xylose reduction. MALDI-TOF MS analysis indicated that the number of glucuronoxylans (GX) chains decreased and that the GX chain length increased in irx7 and irx8 plants, whereas the number of GX chains increased and the GX chain length decreased in irx9 and irx14 plants, suggesting that IRX7 and IRX8 may be involved in the synthesis of a xylan primer, whereas IRX9 and IRX14 may be required to synthesize the xylan backbone [13,14]. Coimmunoprecipitation and bimolecular fluorescence complementation analysis provided biochemical evidence for the xylan synthase complex (XSC) formed by AoIRX9, AoIRX10 and AoIRX14. AoIRX9 plays an important role in the structural properties of the complex, as the activity of AoIRX10 and AoIRX14 is significantly reduced in the absence of AoIRX9 [15]. Coexpression of TaGT43-4, TaGT47-13, TaGT75-3, and TaGT75-4 in Pichia pastoris confirmed that these proteins form a complex. Among them, TaGT43-4 acts as a scaffold protein that holds the other proteins [16]. The side chain α-D-glucuronic acid (GlcUA) of xylan was reduced in irx10 plants, suggesting that IRX10 was involved in side-chain substitution [17,18]. PARVUS is involved in both pectin and xylan synthesis, while the FRA8 and F8H families act to modulate xylan contents and polymerization [14,19].
The GT43 family has been reported in many model plants, because all members, such as IRX9, IRX9L, IRX14 and IRX14L, are involved in the biosynthesis of xylan. The biomass of the RNA interference lines showed higher digestibility, decreased xylan, arabinose and ferulic acid, and increased coumaric acid, supporting a causative role of BdGT43A in Brachypodium distachyon [20]. The BdGT43B2 knockout lines showed that the immunofluorescence labeling of xylan was greatly reduced, whereas cellulose labeling was unchanged or slightly increased [21]. Overexpression of OsIRX9, OsIRX9L and OsIRX14 in Arabidopsis mutants irx9 and irx14 resulted in xylosyltransferase (XylT) activity in the stems that was greater thandouble that in wild type plants, and the stem strength increased to 124% above that of wild-type plants, suggesting that OsIRX9/OsIRX9L, and OsIRX14, have similar functions to Arabidopsis IRX9 and IRX14, respectively [22]. In addition, other GT43 members in rice such as OsGT43A and OsGT43E can partially complement the defects of irx9 and irx14 mutants, indicating their similar functions in xylan biosynthesis [23]. GT43s in cotton and poplar also have similar functions as Arabidopsis IRX genes. Overexpression of GhGT43A1 and PoGT43B in irx9 rescued the defects in plant size and secondary wall thickness, and partially restored the xylose contents. Moreover, GhGT43C1 and PoGT43C were able to rescue the defect of irx14 [24,25].
Transcription factors (TFs) affect the biosynthesis of xylan by regulating GT43 genes. KNAT7 and OSH15 changed the transcript levels of AtIRX9 and OsIRX9 in Arabidopsis and rice by binding to their promoters respectively. The knat7 mutants had lower transcript levels of AtIRX9, reduced xylan contents, and showed a lack of xylose immunofluorescence labeling [26]. OSH15 regulates xylan biosynthesis via OsIRX9 and affects branching angle and plant shape [27]. MYB46 was found to activate the expression of KNAT7 and IRX14L using the estrogen-inducible direct activation system, and SND1 and VND7 also altered the expression levels of IRX14L and IRX10. Moreover, potential MYB-and NAC-binding elements were identified in the promoters of these genes [28]. The Miscanthus lutarioriparius GT43 gene family was also shown to be regulated by MYB and NAC, as MlSND1, MlMYB46 and MlVND7 bound to MlGT43A and MlGT43B upstream sequences and enhanced the fused GUS activity [29].
To investigate xylan accumulation in moso bamboo during rapid growth, the molecular characteristics and expression patterns of 17 GT43 genes were identified and analyzed in moso bamboo. Furthermore, a predicted regulatory network of PeGT43s was constructed. The expression correlation of genes in the network and the regulation between PeMYB35 and the hub gene of PeGT43-5 were verified. This study will serve as a useful reference for further functional analyses of PeGT43s involved in bamboo cell wall development in terms of xylan accumulation and for the future molecular mechanism of GT43 genes in related plant species.

Identification of GT43 genes in moso bamboo
The genomes of moso bamboo and other four bamboo species (Bonia amplexicaulis, Guadua angustifolia, Olyra latifolia, Raddia guianensis) have been sequenced with moso bamboo genome reaching the chromosome level [30,31], which enabled genome-wide identification of the GT43 gene for different bamboo species. A total of 24 potential GT43 genes were found in the moso bamboo genome. After screening the conserved domain by CD-Search program from the NCBI, only 17 genes encoding complete glycosyltransferase 43 (GT43) were identified and designated PeGT43-1 ~ PeGT43-17. For comparison with GT43 genes in other bamboos, the genomes of O. latifolia, R. guianensis, G. angustifolia and B. amplexicaulis were also used, in which 7, 6, 6 and 12 GT43 genes were identified, respectively. Detailed information regarding these GT43 genes such as gene number, protein length and molecular weight is listed in Table S1.
Comparative analysis revealed that woody bamboo had more GT43 gene family members than herbaceous bamboo. The most GT43 gene family members were found in moso bamboo, followed by B. amplexicaulis, while G. angustifolia had only six complete GT43 genes. In addition, there were 10, 10, 14, 4, 9 and 6 GT43 genes in the genomes of Brachypodium, rice, maize, Arabidopsis, cotton and poplar, respectively, in which the gene number of GT43 was greater in the monocots than in the dicots. The number difference of GT43 gene family members in different species may be related to their xylan biosynthesis requirements, considering the abundance of xylan in monocots [9,10].

Phylogenetic analysis of GT43s
To investigate the origin and evolution of GT43s, a maximum likelihood phylogenetic tree was constructed (Fig. 1). All 101 GT43 members from 11 plant species were classified into three subfamilies (A, B, and C), of which the largest one, C, was further divided into three groups (C1, C2 and C3). GT43s from monocots and dicots were clustered into different branches in each subfamily, but the C2 and C3 groups only contained GT43s from monocots. The highest protein homology of GT43s was found between moso bamboo and the other four bamboo species, followed by Brachypodium and rice. There was at least one GT43 homologous protein of moso bamboo compared with those in Arabidopsis and rice. IRX9 (AT2G37090), IRX9L (AT1G27600) and IRX14/ IRX14L (AT4G36890/AT5G67230) associated with xylan biosynthesis [22] were clustered in the A, B and C subfamilies, which were close to PeGT43-8, PeGT43-14 and PeGT43-6/PeGT43-9, respectively. PeGT43-15/ PeGT43-5 and PeGT43-4 were presumed to be involved in glucuronide xylan synthesis as they were homologous to OsGT43A (LOC Os05g03174) and OsGT43E (LOC Os05g48600), respectively [23]. PeGT43-17/PeGT43-2 were homologs of BdGT43B2 (Bradi5g24290), which is involved in arabinoxylan synthesis [21].
To examine the distribution of PeGT43s in moso bamboo scaffolds and their evolution, PeGT43s and OsGT43s were mapped onto moso bamboo and rice genomes for collinearity analysis (Fig. 2). Seventeen PeGT43s were located on 13 scaffolds of moso bamboo, of which 15 PeGT43s comprised 12 gene pairs. PeGT43-3, PeGT43-4, PeGT43-11 and PeGT43-14 from the B subfamily shared collinearity on different scaffolds, showing evidence of more than one duplication event in moso bamboo. PeGT43s were compared to the rice genome with 10 collinear OsGT43s. PeGT43s (PeGT43-1, PeGT43-3, PeGT43-4, PeGT43-5, PeGT43-6, PeGT43-8, PeGT43-9, PeGT43-11, PeGT43-13 and PeGT43- 14), which were homologs of OsIRX9/OsIRX9L (LOC Os07g49370/LOC Os01g48440) and OsIRX14 (LOC Os06g47340) were all found in pairs on different scaffolds and were twice the number of direct homologs in rice. However, PeGT43-7 and PeGT43-16 were collinear with LOC_Os01g06450 and LOC_Os10g13810, respectively, suggesting that no expansion occurred in them. Furthermore, the Ka/Ks ratio of duplication pairs and approximate divergence dates were estimated. The results showed that the duplication pairs underwent purifying selection as the Ka/Ks values of the duplication pairs were less than 1.0. Duplication events may have taken place from 62.14 million years ago (Mya) to 8.61 Mya (Table S2). The divergence of PeGT43-4 and PeGT43-14 in the B subfamily occurred at the earliest time, probably before the divergence of moso bamboo and rice. The majority of gene pairs with collinearity, however, diverged at a late time. According to the potential function of PeGT43s, it is believed that the expansion of these genes may be the result of partial gene duplication to adapt to biological function.

Conserved motifs of PeGT43s
To further clarify the specific motifs of PeGT43s, the distribution of 15 main conserved motifs was analyzed with the MEME online program (  [32].

Cis-acting elements in the promoter of PeGT43s
Cis-acting elements distributed in the promoters of target genes affect their expression. Screening for the cis-acting elements in upstream sequences of PeGT43s revealed a large number of TF binding elements, hormone response elements and abiotic stress response elements ( Fig. 4 and Table S3). The most abundant elements were MYB binding elements and MeJA-responsive elements, followed by MYC binding elements, light-responsive elements and stress-responsive elements. In addition, several promoters of PeGT43s contained wound-responsive elements, ABA-responsive elements, anaerobic-responsive elements and meristem expression elements. These results suggested that PeGT43s may be regulated by many internal factors and respond to environmental changes in moso bamboo.
TFs play an important role in plant growth and development by binding cis-acting elements of targeted genes. To investigate the potential transcriptional regulatory mechanisms, the potential regulatory relationships were analyzed by PlantRegMap ( Fig. 4 and Table S4). Three hundred and fifteen regulatory relationships between 60 TFs and 17 PeGT43s were identified under a cutoff p value ≤ 0.05, with MYBs and ERFs being the most enriched TFs. Fourteen PeGT43s promoters possessed ERF binding sites and 16 PeGT43s promoters possessed SMRE elements binding MYBs. GATA, LBD, C2H2 and NAC binding elements were also found in the promoter sequences of PeGT43s.

Expression profiles of PeGT43s in different tissues of moso bamboo
To reveal the potential function of PeGT43s, we analyzed the expression profiles of PeGT43s in various tissues with RNA-seq data. The results of 17 PeGT43s in 26 different tissues showed that most PeGT43s were expressed at high levels in roots and shoots, especially in the middle and base portions of 1.5 m to 6.0 m shoots, while PeGT43-12 was most highly expressed in 0.2 m shoots and sheath sheets ( Fig. 5 and Table S5). PeGT43-7 was nearly absent in all tissues. PeGT43s of the A subfamily were expressed mainly in mature roots and shoots but only to a small extentin leaves, while those of the C subfamily showed various levels of expression, not only in the roots and shoots but also in the buds. For example, PeGT43-5 of the A subfamily exhibited preferential transcriptional abundance in mature roots and the base of 1.5 m and 3.0 m shoots, but PeGT43-17 of the C subfamily was also exhibited a relatively high abundance in buds.
Due to the preferential abundance in shoots, the expression profiles of PeGT43s based on RNA-seq data from different growth stages and parts of the shoots were analyzed. The results showed that PeGT43s were expressed specifically in shoots of different heights, which was consistent across 26 different tissues (Fig. S1  Table S5). Furthermore, transcriptional analysis of different parts of shoots at the early stage of rapid growth (3.0 m shoots) was performed ( Fig. 6 and Table S5). The expression levels of most PeGT43s were significantly different (p < 0.05) among the start of division (SD), rapid division (RD) and rapid elongation (RE) sections ( Fig. 6 and Table S5). The genes belong to A and B subfamilies were highly expressed in RE section but nearly absent in the SD and RD sections. PeGT43-2, PeGT43-6, PeGT43-9 and PeGT43-17 belonging to the C subfamily were highly expressed in all portions of the 3.0 m shoot, with the highest expression level in the RE section. In addition, the transcripts of PeGT43-4 and PeGT43-5 accumulated substantially from the SD to the RE section.
Nodes are considered to be important tissues to provide mechanical support during the rapid growth of moso bamboo. Therefore, the PeGT43s expression patterns in the shoot apical meristem (SAM), young internode (YIN), young node (YNO), mature internode (MIN) and mature node (MNO) of shoots were analyzed ( Fig. 6 and Table S5). The expression levels of PeGT43s were lower in the SAM, and higher in nodes than in internodes, and those in mature tissues were higher than those in young tissues. Twelve PeGT43s were differentially expressed at significant levels (p < 0.05). PeGT43-2 and PeGT43-17 showed much higher expression levels in the nodes than other PeGT43s.

Co-expression network of PeGT43s in moso bamboo
To determine the genes coexpressed with PeGT43s, a coexpression network was constructed using the Bam-booNET online database. A total of 310 nodes and 1871 connections were found in the network coexpressed with PeGT43s. The GO terms showed that the majority of the coexpressed genes were enriched in biological pathways involved in polysaccharide metabolism, such as cellulose biosynthetic process (GO:0030244), cellular polysaccharide biosynthetic process (GO:0033692), beta-glucan biosynthetic process (GO:0051274) and cellular carbohydrate biosynthetic process (GO:0034637), which implied an important function of PeGT43s in cell wall reformulation (Fig. S2). The molecular function category showed that the coexpressed genes had UDP-glycosyltransferase activity (GO:0008194) and cellulose synthase activity (GO:0016759). A network containing 35 genes with coexpressed Pearson correlation coefficient (PCC) > 0.8 was created by identifying categories associated with cell wall biosynthesis (Fig. 7 and Table S6). Of 35 genes, four genes were selected as the most significantly coexpressed genes encoding TFs. In addition, multiple genes encoding cellulose synthases (CesAs), glycosyltransferases (GTs), glycosyl hydrolases (GHs), acetyltransferases (ATs) and glycosylphosphatidylinositol-anchored proteins (COBLs) were identified to be significantly coexpressed with PeGT43s in the network.

Expression abundance of PeGT43s and co-expressed genes consistent with xylan content
To confirm the correlation between PeGT43s and coexpressed genes in shoots of different heights, the expression of 9 representative PeGT43s and nine coexpressed genes (PCC > 0.9) within the network was explored in a qRT-PCR assay. The expression level of PeGT43s was low at the early growth stages (0.5 m ~ 2.0 m) of the shoots, and then showed a significant increase (p < 0.05) when the shoots grew to 4.0 m. The expression of most PeGT43s continued to rise significantly (p < 0.05) during growth from 4.0 m to 6.0 m and peaked in 6.0 m shoots, while a few PeGT43s showed similar expression levels in 4.0 m and 6.0 m shoots. Then, the expression of PeGT43s decreased and was maintained at a high level (p < 0.05) in 8.0 m shoots (Fig. 8).
The coexpressed genes showed similar expression profiles to those of PeGT43s, among which PeMYB35, PeMYB37, PeNAC42, PeCesA4, PeCesA7, PeCesA9, and PeGT47-12,063 showed great changes in transcription level. Most of them were expressed at low levels at the  (Fig. 9).
To determine the connection between PeGT43s expression and the abundance of xylan, the xylan contents and structural polysaccharide staining intensity in the same tissues were measured. Xylan contents increased significantly during the growth of shoots, especially in 8.0 m shoots (Fig. 10), which was reasonable as PeGT43s were highly expressed in the older shoots. The increase in xylan contents in 2.0 m shoots may be due to the upregulated expression accumulation of some PeGT43s. The intensity of structural polysaccharide staining also increased with the growth of shoots (Fig. 10), which was consistent with the expression of PeGT43s.

Subcellular localization analysis and validation of PeMYB35 binding 5′ UTR/promoter of PeGT43-5 in yeast
Correlation analysis based on the coexpression network and transcriptome data revealed that PeGT43-5 and PeGT43-6 had the strongest correlation with PeMYBs and PeNACs. In addition, PeGT43-5 was also highly positively correlated with PeGT47-07283, which may be relevant to the formation of glycosyltransferase polymers ( Fig. 7 and Fig. S3). Biosynthesis of xylan occurs in the Golgi apparatus. Previous study shows that coexpression of IRX9 and IRX14 contributes to better Golgi localization results [15]. Therefore, the sequences encoding PeGT43-5 and PeGT43-6 (Table S9), which were orthologs of IRX9 and IRX14, were fused in-frame to the sequence encoding green fluorescent protein (GFP) at the 3′ end respectively and then cotransfected tobacco leaves with a mRFP-tagged Golgi marker together. The resulting overlaid signals suggested that they are Golgi-localized glycosyltransferases (Fig. 11).
AtMYB103, the ortholog of PeMYB35, has been shown to act as the regulator of cell wall biosynthesis [3]. To better understand the underlying regulatory mechanism of PeGT43s, we isolated the untranslated region (5′ UTR, 1119 bp) and promoter (2539 bp) of PeGT43-5, as well as PeMYB35 (Table S9). Moreover, triple SMRE elements including SMRE2 (ACC AAC T), SMRE3 (ACC AAA C), SMRE4 (ACC AAC C), SMRE5 (ACC TAA T), SMRE6 (ACC AAC C) and SMRE7 (ACC TAA C) were synthesized. All fragments were inserted into yeast expression vectors. Potential binding sites of PeMYB35 in the 5′ UTR/promoter of PeGT43-5 were analyzed using a yeast one-hybrid (Y1H) assay. The positive control and the experimental group grew well on TDO medium containing 3-amino-1,2,4-triazole (3-AT). The transformants harboring triple elements showed the best growth, followed by the transformants harboring the 5′ UTR and those harboring the promoter (Fig. 12). In contrast, the negative control could not grow on the nutritional  Table S5 screening medium. The reason that transformants harboring a 5′ UTR grew better than those harboring a promoter might be the double tandem repeat of SMRE4 in the 5′ UTR. These experimental results indicated that PeMYB35 could bind the 5′ UTR/promoter of PeGT43-5, which suggests that PeMYB35 could regulate the expression of PeGT43-5.

Discussion
Xylan is the major hemicellulose in both primary and secondary cell walls. Since GT43 is one of the important glycosyltransferase families contributing to xylan biosynthesis, exploration of the GT43 genes in bamboo will potentially provide insight to modify the biomass composition of bamboo suitable for biofuel production. Genome-wide identification of GT43 genes from five bamboo species revealed that GT43s in moso bamboo were the most abundant, indicating that woody bamboo had more GT43s than herbaceous bamboo. Only 12 GT43 genes were identified in B. amplexicaulis. This lower-than-expected number is likely due to the lower quality assembly of the B. amplexicaulis genome (848 Mb of the scaffold with an N50 length of 1.76 Mb) compared to that of moso bamboo (1886 Mb of the scaffold with an N50 length of 79.90 Mb) [5,31]. There was a correlation between biological characteristics and the number of related genes. Monocots have more GT43 genes and a higher percentage of xylan in hemicellulose than dicots [9,10]. Therefore, GT43 may be one of the factors affecting xylan biosynthesis. Collinearity between moso bamboo and rice provided evidence for the gene duplication of PeGT43s. The divergence time of the four pairs of PeGT43s was approximately50 ~ 60 million years ago (Mya) and that of the two pairs of PeGT43s was approxi-mately10 Mya, which is similar to the time of the WGD events that occurred in moso bamboo [33].
The type and distribution of conserved motifs in GT43 proved the reliability of subfamily classification. GT43s from the same subfamily shared the same conserved motifs and those from different subfamilies possessed  Table S5 unique motifs, such as Motif 12 for the GT43-A subfamily and Motif 10 and Motif 14 for the GT43-B subfamily. There were differences between conserved motifs of GT43s from monocots and dicots. For example, Motif 14 and Motif 13 were specific to GT43-B and GT43-C subfamily members from monocots, respectively, whereas there were no specific motifs in dicots. In addition, different conserved motifs possessed various sugar donor and sugar acceptor sites, which presumably represents the functional differentiation of GT43s caused by conserved motif differentiation. Differentiated motifs of GT43s may function to recognize different glycosyl and xylan backbones [34,35].
The expression of GT43 showed significant spatial and temporal specificity. PeGT43s were highly expressed in shoots and showed an increasing expression trend with the growth of shoots, which was consistent with the expression trend of cell wall related genes [6]. The cellulose and lignin contents in the RE section of moso bamboo shoots were significantly higher than those in the SD and RD sections [36]. The variations in the expression of PeGT43s and the contents of these components suggested the changes related to cell wall biosynthesis during rapid cell elongation. Moso bamboo shoot internodes can elongate rapidly but not thicken in diameter during growth and development. Thus, the formation of nodes of moso bamboo shoots is of great biological importance in supporting rapid upright growth. Genes highly expressed in the MIN were mainly enriched in secondary cell wall biosynthesis pathways such as the phenylpropane biosynthesis pathway. Genes preferentially expressed in the MNO were enriched in cell wall organization and membrane metabolism pathways [37]. The expression of PeGT43s in the MIN and MNO was higher than that in the YIN and YNO. The preferential expression of PeGT43s in the RE sections and nodes reflected their role in xylan biosynthesis adapted to the rapid growth of moso bamboo.
There are a set of MeJA-responsive elements in the PeGT43s promoter. MeJA plays an important role in plant self-defense as a responsive signal. For example, the exogenous application of MeJA increased the total free phenolics, ferulic acid, caffeic acid, coumaric acid, and chlorogenic acid in the leaves of bread wheat. Upregulation of phenylpropanoid cascades in response to exogenous application of MeJA which reduces cell wall disruption and tissue disintegration and increases suberization and lignification of the plant cell may increase the resistance to spot blotch [38]. Genes for pectin methylesterase (PME), endo-polygalacturonase (PG), glycosyltransferase (GT) and cellulose synthase (CesA) are also Fig. 7 The co-expression network of PeGT43s significantly upregulated by MeJA, suggesting that MeJA treatment not only promotes the degradation of polysaccharides to destroy the cell wall but also promotes the biosynthesis of new polysaccharides [39]. In addition, cellulose, hemicellulose, and pectin polysaccharides induced by MeJA act as damage-associated molecular patterns (DAMPs) to activate plant innate immunity [40]. The MeJA-responsive and wound-responsive elements in the PeGT43s promoter suggest that PeGT43s may be involved in hormone signaling pathways of moso bamboo.
Xylan, as a linkage between cellulose and lignin, increases in content with the growth of bamboo shoots, which was similar to the content changes of other cell wall components [41]. This result implied that PeGT43s are inseparably related to other genes involved in cell wall biosynthesis [9]. In this study, the coexpression network of PeGT43s included genes encoding CESAs, GHs, COBLs and ATs. CESAs and COBLs are associated with cellulose accumulation, such as CESAs involved in the biosynthesis of cellulose and COBLs in the polar arrangement of cellulose to affect the toughness and stiffness [42,43]. Protein-protein interactions have been shown to constitute an important organizing principle for xylan biosynthetic enzymes [44]. Genes from the GT43 and GT47 families are assumed to form complexes in the Golgi apparatus involved in xylan backbone biosynthesis [45]. Wheat (Triticum aestivum) GTs were proposed to synthesize arabinoxylan via a glucurono(arabino)xylan synthase complex form. The complexes of TaGT43-4, TaGT47-13 and TaGT75-4 obtained by immunoprecipitation were characterized by xylosyltransferase, glucuronosyltransferase and arabinosyltransferase activities, respectively [46,47]. In this study, the significant correlation between PeGT43s and genes from the GT47 family may be related to complex formation in moso bamboo, which needs to be further validated.
GHs are involved in the degradation of cell wall polysaccharides [48] and induce lignin content changes and biosynthesis of cellulose during seed maturation [49], considering their coexpression with GTs and CESAs. OsGH9A3 and OsGH9B5 are coexpressed with OsCESA1, OsCESA3 and OsCESA8 and involved in primary cell wall cellulose synthesis [50]. ATs are involved in xylan biosynthesis by adding acetyl to xylan chains [11]. OsTBL1 and OsTBL2 are ATs that acetylate xylan in rice. They are responsible for acetylation of both xylan main and side chains and their mutants display a stunted growth phenotype with varying degrees of dwarfism [51]. Furthermore, the expression patterns of PeGHs and PeATs are similar to those of PeGT43s, suggesting their synergistic role in cell wall remodeling.
PeMYB35 and PeMYB37, coexpressed with PeGT43s, are homologs of AtMYB103, which play a critical role in secondary wall biosynthesis in A. thaliana [52]. PeMYB35 and PeMYB37 were expressed at low levels Furthermore, there are abundant MYB-binding elements on the promoter of PeGT43s, which belonged to the SMRE motif. SMRE elements are involved in the lignin, cellulose and hemicellulose biosynthetic pathways by binding to MYB, NAC and KNAT transcription factors [28,53]. In Arabidopsis, AtMYB83 increas the expression level of AtIRX14L significantly [28]. NAC and MYB in Mangifera have also been found to bind to the promoters of MlGT43A and MlGT43B [29]. In this study, the binding of PeMYB35 to the 5′ UTR/promoter of PeGT43-5 with different SMRE elements was verified in yeast, suggesting that PeMYB35 might play a regulatory role in the expression of PeGT43s for xylan biosynthesis. In addition, GT43 genes in other species like IRX9 and IRX14, as well as transcription factors that bind to the GT43 gene promoters, are related to xylan biosynthesis. Similarly involved in xylan biosynthesis, PeIRX10 has been identified in moso bamboo to have the ability to complement the deficiency of irx10 in Arabidopsis, which suggests a functional similarity between PeIRX10 and irx10 [54]. Therefore, it is assumed that PeGT43s are related to xylan biosynthesis as well. However, the content and structure of xylan differs in monocots and

Conclusions
GT43s are essential for xylan biosynthesis and function in cell wall modification. A total of 17 PeGT43s belonging to three subfamilies were identified in the moso bamboo genome. The cis-acting elements involved in MeJA, ABA and wound response, as well as MYB-binding sites were present in the promoters of PeGT43s. Transcriptome data revealed that PeGT43s were highly expressed in shoots, especially in rapid elongation potions and nodes. The expression of PeGT43s and their coexpressed genes associated with cell wall biosynthesis was consistent with the xylan content and structural polysaccharide staining intensity in bamboo shoots. In addition, Y1H assay confirmed that PeMYB35 could bind the 5′ UTR/ promoter of PeGT43-5, indicating the regulatory relationship between TFs and PeGT43s in moso bamboo. These results lay a foundation for further investigation of PeGT43s function in the biosynthesis of xylan during rapid growth of moso bamboo.

Plant materials
Representative sections (the fifteenth section from the base upwards) of moso bamboo shoots of different heights (0.5 m, 1.0 m, 2.0 m, 4.0 m, 6.0 m and 8.0 m) were collected with permission from the bamboo forest experimental site of Jiangxi Academy of Forestry located in Nanchang, Jiangxi Province, China. The bamboo samples were frozen in liquid nitrogen and stored at − 80 °C Fig. 12 Y1H assay for the interaction between PeMYB35 and the 5′ UTR/promoter of PeGT43-5. The 5′ UTR/promoter of PeGT43-5 as well as triple SMRE elements were inserted into pHIS2 vector. The reporter and effector constructs were co-transformed into yeast Y187 cells, and the transformed cells were identified by spotting serial dilutions of yeast onto selective medium SD/−Leu/−Trp (DDO) and SD/−Leu/− Trp/−His (TDO) containing 20 mmol·L − 1 3-amino-1,2,4-triazole (3-AT). P: positive control (p53::HIS2 + pGADT7::53); N: negative control (p53::HIS2 + pGADT7::PeMYB35); Pro: promoter before RNA extraction and determination of xylan contents. Three biological replicates were used for each experiment.

Phylogenetic and structure analysis
Phylogenetic analysis was conducted according to the maximum likelihood method (1000 bootstrap replicates) of the MEGA X software. The conserved motifs were identified using the Multiple EM for Motif Elicitation program v5.4.1 (https:// meme-suite. org/ meme/ tools/ meme) [56] and visualized with TBtools [57]. Collinear regions between PeGT43s and the Ka (nonsynonymous substitution) / Ks (synonymous substitution) ratio between the paralogous pairs were analyzed by MCScanX [57]. The divergence time of the duplication event was estimated with the rate λ = 6.5 × 10 − 9 in monocots [58].

qRT-PCR, xylan contents and structural polysaccharides staining analysis
Total RNA was isolated using TRIzol reagent (Invitrogen, USA). First-strand cDNA was synthesized with 1 μg RNA using a PrimeScript ™ RT Reagent Kit (TaKaRa, Japan) according to the manufacturer's instructions. qRT-PCR experiments were performed using SYBR Green chemistry (Roche, Mannheim, Germany) on a qTOWER 2.2 system (Analytik, Jena, Germany) according to the manufacturer's directions. Gene-specific primers were designed by Primer Premier 5.0 software (Table S7), and PeTIP41 was selected as an internal control [63]. Three biological replicates and three technical replicates were employed in each experiment. The final relative expression levels were calculated using the 2 -△△Ct method [64]. The xylan contents were measured based on the orcinol-HCl method [65] with a commercial kit (item number S0169O-1) from the Kete Biotechnology Company of Jiangsu, China. Structural polysaccharide staining was performed by the zinc iodide chloride method.

Subcellular localization and yeast one-hybrid assay
The coding sequences of PeGT43-5 and PeGT43-6 were cloned and fused in frame with that of GFP in pCAM-BIA1300 vector, respectively. The resulting constructs were cotransformed together into tobacco leaves with a construct for expression of the Golgi marker mRFP. The fluorescence signals were recorded with a confocal laser scanning microscope (LSM 980; Zeiss) after 3 days. Each inoculation was performed on at least triplicate leaves.