Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences, Qingdao , China

Transcription

1 The Plant Journal (216) 88, doi: /tpj UDP-glycosyltransferase 72B1 catalyzes the glucose conjugation of monolignols and is essential for the normal cell wall lignification in Arabidopsis thaliana Ji-Shan Lin 1, Xu-Xu Huang 1, Qin Li 1, Yingping Cao 2, Yan Bao 2, Xia-Fei Meng 1, Yan-Jie Li 1, Chunxiang Fu 2 and Bing-Kai Hou 1, * 1 Key Laboratory of Plant Cell Engineering and Germplasm Innovation, Ministry of Education of China, School of Life Sciences, Shandong University, Jinan 251, China, and 2 Key Laboratory of Biofuels, Qingdao Engineering Research Center of Biomass Resources and Environment, Qingdao Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences, Qingdao 26611, China Received 3 December 215; revised 11 May 216; accepted 2 June 216; published online 13 August 216. *For correspondence ( bkhou@sdu.edu.cn). SUMMARY Glycosylation of monolignols has been found to be widespread in land plants since the 197s. However, whether monolignol glycosylation is crucial for cell wall lignification and how it exerts effects are still unknown. Here, we report the identification of a mutant ugt72b1 showing aggravated and ectopic lignification in floral stems along with arrested growth and anthocyanin accumulation. Histochemical assays and thioacidolysis analysis confirmed the enhanced lignification and increased lignin biosynthesis in the ugt72b1 mutant. The loss of UDP-glycosyltransferase UGT72B1 function was responsible for the lignification phenotype, as demonstrated by complementation experiments. Enzyme activity analysis indicated that UGT72B1 could catalyze the glucose conjugation of monolignols, especially coniferyl alcohol and coniferyl aldehyde, which was confirmed by analyzing monolignol glucosides of UGT72B1 transgenic plants. Furthermore, the UGT72B1 gene was strongly expressed in young stem tissues, especially xylem tissues. However, UGT72B1 paralogs, such as UGT72B2 and UGT72B3, had weak enzyme activity toward monolignols and weak expression in stem tissues. Transcriptomic profiling showed that UGT72B1 knockout resulted in extensively increased transcript levels of genes involved in monolignol biosynthesis, lignin polymerization and cell wall-related transcription factors, which was confirmed by quantitative real-time PCR assays. These results provided evidence that monolignol glucosylation catalyzed by UGT72B1 was essential for normal cell wall lignification, thus offering insight into the molecular mechanism of cell wall development and cell wall lignification. Keywords: Arabidopsis thaliana, monolignols, glycosylation, glycosyltransferase, cell wall, lignification, transcriptome. INTRODUCTION Lignin, produced by the oxidative polymerization of p- hydroxycinnamyl alcohol monomers (termed monolignols) on plant cell walls, is the second most abundant terrestrial biopolymer after cellulose (Boerjan et al., 23). By providing mechanical support to plant stems, enabling water conduction from roots to leaves and protecting plants against bacterial and fungal pathogens, lignin is essential to plant growth and development (Bhuiyan et al., 29; Zhao and Dixon, 214; Barros et al., 215). It is clearly recognized that monolignols are derived from phenylalanine via a series of enzymatic reactions catalyzed by phenylalanine ammonia lyase (PAL), cinnamic 26 acid 4-hydroxylase (C4H), 4-hydroxycinnamoyl CoA ligase (4CL), ferulic acid 5-hydroxylase (F5H), p-coumaroylshikimate 3 -hydroxylase (C3H), hydroxycinnamoyl CoA:shikimate hydroxycinnamoyl transferase (HCT), caffeoyl CoA Omethyltransferase (CCoAOMT), hydroxycinnamoyl CoA reductase (CCR), caffeic acid/5-hydroxyferulic acid O- methyltransferase (COMT) and (hydroxy)cinnamyl alcohol dehydrogenase (CAD) (Vanholme et al., 28). Recently, the mechanism of monolignol transport and lignin polymerization/deposition was revealed step by step. First, monolignols are actively transported to the cell wall through the plasma membrane possibly by ABC The Plant Journal 216 John Wiley & Sons Ltd

2 UGT72B1 modulates cell wall lignification 27 transporters, although only p-coumaryl alcohol s transporter AtABCG29 has been identified (Alejandro et al., 212). Then, a plasma membrane platform, probably CASP-like protein, is needed to bring together the NADPH oxidase and peroxidase (and possibly other cell wall-modifying enzymes) to mediate localized lignin polymerization (Lee et al., 213). After that, NADPH oxidase produces hydrogen peroxide, which is used by peroxidases to oxidize and polymerize the monolignols (Lee et al., 213). In addition, at least three laccases (LAC4, LAC11 and LAC17) were also genetically demonstrated to be important regulators of lignification in Arabidopsis thaliana fiber cell walls and the lac4 lac11 lac17 triple mutant is severely dwarf (Berthet et al., 211; Zhao et al., 213). In dicot plants, lignin is mainly composed of the monolignols coniferyl and sinapyl alcohol, with typically minor amounts of p-coumaryl alcohol, that give rise to the guaiacyl (G), syringyl (S) and hydroxyl-cinnamyl (H) units of the lignin polymer, respectively. These polymers are deposited predominantly in the walls of secondarily thickened cells, making them rigid and impervious (Vanholme et al., 21). Recently, MYB and NAC family transcription factors (TFs) were shown to play key roles in regulating lignin deposition in secondary cell walls of Arabidopsis (Mitsuda et al., 27; Zhong et al., 27, 28; Mitsuda and Ohme-Takagi, 28; Zhou et al., 29). MYB TFs have key functions in the regulation of plant development and metabolism, especially the synthesis of lignin-related and phenylpropanoidderived compounds in plants. For instance, AtMYB58, AtMYB61, AtMYB63 and AtMYB85 have been identified as true lignin-specific TFs, and these MYBs could directly regulate most monolignol pathway genes (Newman et al., 24; Zhong et al., 28; Zhou et al., 29). The NAC TFs, NST1, NST2, NST3 and AtVND6/7, are master transcriptional switches in regulating the entire secondary cell wall formation (including cellulose, xylan and lignin synthesis), and overexpression of NSTs causes ectopic lignification (Mitsuda et al., 27; Mitsuda and Ohme-Takagi, 28). These NACs have been proposed to act through a cascade of downstream TFs, such as MYB46, MYB58, MYB63, MYB83 and MYB85, which could directly regulate most monolignol pathway genes (Zhong et al., 26). Glycosylation of monolignols is widespread in land plants including gymnosperms and angiosperms. Since the 197s, the glucosides of monolignols, mainly including coniferyl alcohol glucoside (coniferin), synapyl alcohol glucoside (syringin) and p-coumaryl alcohol glucoside, have been found in various plant tissues such as vascular bundles, cambia, stems, roots and leaves (Marcinowski and Grisebach, 1977; Terazawa et al., 1984; Terazawa and Miyake, 1984; Whetten and Sederoff, 1995; Ito et al., 2; Hemm et al., 24; Tsuji and Fukushima, 24; Huis et al., 212). However, the physiological significance of monolignol glycosylation is still largely unknown. Several researchers assessed the monolignol glucoside content and distribution in several dicot trees and gymnosperm species, and found that the monolignol glucosides were present primarily in protoplasts of developing xylem of gymnosperm species and some angiosperm species (Terazawa and Miyake, 1984; Savidge, 1989; Dharmawardhana et al., 1995; Whetten and Sederoff, 1995). This leads to the assumption that the high level of accumulated monolignol glucosides are most likely stored in vacuoles and used as the storage or transport forms of the monolignols, suggesting that lignin biosynthesis and mobilization of the synthetic precursors may be affected by monolignol glucosides. Monolignol glycosylation is catalyzed by soluble UDP-sugar glycosyltransferases (UGTs). The b-glycosidases are proposed to release monolignols from the glycoconjugates. In Arabidopsis, a cluster of glucosyltransferases, including UGT72E1, UGT72E2 and UGT72E3, was identified to be responsible for the glucose conjugation of monolignols (Lim et al., 21, 25). Two b- glucosidases, BGLU45 and BGLU46, were also isolated from Arabidopsis and demonstrated to hydrolyze monolignol glucosides (Escamilla-Trevino et al., 26). To explore the possible roles of monolignol glucosides in lignin synthesis, several studies have been performed to disturb the expression of monolignol glycosyltransferases or glycosidases. For instance, downregulation or upregulation of glucosyltransferases UGT72E1, UGT72E2 and UGT72E3 resulted in the corresponding reduction or accumulation of the monolignol glucosides in transgenic Arabidopsis; however, no significant change in lignin deposition was observed (Lanot et al., 26; Vanholme et al., 28). The mutants of monolignol glycosidases BGLU45, BGLU46 and BGLU47 were also investigated in Arabidopsis, and results confirmed the substrates of glycosidases but did not support the notion of monolignol glucosides working as the direct precursors of lignin (Chapelle et al., 212). A recent study was performed for monolignol transport experiments using poplar xylem tissues. Its results indicated that membrane vesicles prepared from differentiating xylem tissues had clear ATP-dependent transport activity toward coniferin, suggesting likely involvement of monolignol glucosides (especially coniferin) in active lignification (Tsuyama et al., 213). However, these data mentioned above do not provide a clear conclusion about whether monolignol glycosylation has a crucial role and how it affects lignin deposition and cell wall lignification. This has remained unresolved since the discovery of monolignol glucosides in planta. In this study, we identified an Arabidopsis glycosyltransferase mutant ugt72b1. Its inflorescence stem displayed aggravated and ectopic lignification with an arrested growth phenotype. Lignin content was also substantially increased in mutant cell walls. However, rescued lines restored the wild-type phenotypes in regard to both

3 28 Ji-Shan Lin et al. growth and lignin deposition. We analyzed the expression pattern of the UGT72B1 gene and found that it was expressed specifically in differentiating xylem tissues. Enzyme activity analysis indicated that glycosyltransferase UGT72B1 could catalyze the glucose conjugation of monolignols, especially coniferyl alcohol and coniferyl aldehyde. Furthermore, transcriptomic profiling showed that knockout of UGT72B1 resulted in extensive upregulation of genes related to monolignol biosynthesis and lignin polymerization. These results provide evidence that monolignol glucosylation catalyzed by UGT72B1 is essential for normal cell wall lignification. RESULTS Growth phenotypes of ugt72b1 mutants and rescued lines In our attempt to screen lignin-related glycosyltransferase mutants, two T-DNA insertion lines of the UGT72B1 gene (At4g17) were identified from the European Arabidopsis Stock Centre (NASC). Both the mutant ugt72b1ko-1 (SALK_49597) and ugt72b1ko-2 (SAIL_611_E4) had a T- DNA insertion located in the exon of the target gene (Figure 1a). Two T-DNA lines were null mutants for UGT72B1 expression demonstrated by reverse transcription PCR (RT- PCR) (Figure 1b). Two ugt72b1 mutants exhibited the same phenotypes their shoot tips exhibited the phenomena of anthocyanin accumulation and growth repression after bolting (Figure 1c e; Figure S1). Due to repression of shoot growth, ugt72b1 mutants were still very short and had only several young branches occurring from very close sites after about 6 7 weeks of growth (Figure 1e). At full maturity, mutant plants developed a similar stature but shorter and fewer siliques compared with the wild-type (Figure S2). The development of floral and reproductive tissues in the mutant was monitored to determine why it bore smaller and fewer siliques. The mutant stamen filaments were shorter than that of wild-type, resulting in fewer pollen grains released on the pistil (Figure S3). In addition, scanning electronic microscopy and in vitro germination assay on mature pollen grains at stage 13 were performed. Abnormal pollen morphology and a low percentage of germination were observed in mutant pollen grains (Figure S3). These alterations in reproductive tissues may be responsible for the low fertility observed in mutants. To confirm that the phenotype observed in ugt72b1 mutants was due to the loss of function of UGT72B1, we introduced the cdna of UGT72B1 under control of native UGT72B1 promoter into the ugt72b1 mutant (SALK_49597) to generate rescued lines. RT-PCR analysis showed that the UGT72B1 transcription in rescued lines was restored to wild-type level (Figure 1b). Correspondingly, the mutant phenotype was also restored to wild-type in rescued lines (Figure 1c e; Figures S2 and S3). These data demonstrated that growth phenotype of ugt72b1 mutants was a result of loss of function of UGT72B1. Knockout of UGT72B1 caused an aggravated and ectopic lignification To screen lignin-related mutants, the cross-sections of mutant floral stems were prepared at two developmental stages corresponding to the wild-type: the shoot tip area of stage 6.; and the basal area between the first and second lateral branch of stage 6.5 (Boyes et al., 21). (a) ugt72b1 ko-2 (SAIL_611_E4) UGT72B1 ugt72b1 ko-1 (SALK_49597) Exon UTR T-DNA insert (c) (d) (e) Figure 1. Identification and phenotype of ugt72b1 mutants. (a) Schematic diagram showing the T-DNA insertion locations of ugt72b1 mutant ko-1 and ko-2. (b)transcript levels of UGT72B1 gene in wild-type (WT), mutants (ko-1, ko-2) and rescued lines (RE-1, RE-2) determined by RT-PCR. TUBULIN2 (TUB) was used as internal control. (c and d) Phenotypes of wild-type, mutants and rescued lines at growth stage 6.. (e) Phenotypes of wild-type, mutants and rescued lines at growth stage 6.5. Scale bars: 1 cm. (b) WT ugt72b1 ko-1 RE-1 ugt72b1 ko-2 RE-2 UGT72B1 TUB WT KO-1 KO-2 RE-1 RE-2

4 UGT72B1 modulates cell wall lignification 29 Wiesner (phloroglucinol HCl) staining was firstly performed on stem cross-sections. This staining is often considered as diagnostic with red color for lignin content (Nakano and Meshitsuka, 1992). Wiesner staining showed that the ugt72b1 mutant had an aggravated and ectopic lignification phenotype compared with wild-type and rescued lines (Figure 2). In the stem tissue of growth stage 6., red staining was seen only in xylem vessels in wildtype and rescued lines (Figure 2a and c); however, red coloration was also observed in the pith and interfascicular fibers of the inflorescence stems of the ugt72b1 mutant (Figure 2b). Sections of ugt72b1 mutant stem from growth stage 6.5 showed a more aggravated and intense lignification not only in xylem but also in areas around pith tissues and interfascicular fibers (Figure 2e). (a) (b) (c) (d) (e) (f) (g) (h) (i) Figure 2. Knockout of UGT72B1 caused an aggravated and ectopic lignification. (a c) Wiesner staining of floral stems of (a) WT, (b) mutants and (c) rescued lines at growth stage 6., showing the aggravated and ectopic lignification in pith cells and interfasicular cells in ugt72b1 mutant. (d f) Wiesner staining of floral stems of (d) WT, (e) mutants and (f) rescued lines at growth stage 6.5, showing the enhanced lignification in pith and interfasicular cells in ugt72b1 mutant. (g i) M aule staining of floral stems of (g) WT, (h) mutants and (i) rescued lines at growth stage 6.5, showing the substantially increased S lignin and G lignin in ugt72b1 mutant. Scale bars: 2 lm.

5 3 Ji-Shan Lin et al. We then performed M aule staining, which provides qualitative indications of lignin monomer composition by staining G lignin brown and S lignin red. Compared with the wild-type, which exhibited a typical pattern with brown staining in the vascular bundles and red staining in the fibers (Chapple et al., 1992), ugt72b1 mutant plants showed a widespread and stronger red and brown staining in xylem, interfascicular fiber and pith (Figure 2h), suggesting increased S and G lignins in the mutant cell walls. The enhanced and ectopic lignification areas in mutants indicated by the two different methods of Wiesner and Ma ule reagents were basically in accordance, indicating that lignin deposition has been increased and misplaced. Lignin composition and total yield were substantially altered in ugt72b1 mutants To further investigate the impact of UGT72B1 functional deficiency on the lignification in mutants, lignin composition of the ugt72b1 mutants and the rescued lines was determined by thioacidolysis using inflorescence stems of 5-week-old plants at growth stage 6. (Table 1). The thioacidolysis yield of the rescued lines was similar to that of wild-type plants. However, the ugt72b1 mutants released significantly increased amounts of H, G and S monomers (Table 1). A much more significant increase in S units than in G units was observed in mutants, resulting in a higher S/G ratio. The total thioacidolysis yield (H + G + S) of the entire inflorescence stems of 5-week-old plants was also significantly increased in mutants compared with wild-type, further confirming the impact of UGT72B1 deficiency on lignin deposition. Here, thioacidolysis chemical analysis of lignin composition showed results consistent with those of histochemical staining. Table 1 Lignin composition, total yield and cell wall thickness Lines WT KO RE H(lmol g 1 ) ** G(lmol g 1 ) * S(lmol g 1 ) ** S/G **.73.4 Total yield ** (lmol g 1 ) Cell wall thickness (lm) ** WT, wild-type plants; KO, ugt72b1 mutants; RE, rescued lines; H, G, S, yields of the thioethylated products of p-hydroxyphenyl (H), guaiacyl (G) and syringyl (S) lignin units expressed as lmoles per gram of CWR. S/G = ratio of G to S lignin units. Total yield = H + S + G lignin units expressed as lmoles per gram of CWR. All data are means SE (n = 3). Asterisks indicate values that were determined by the Student s t- test to be significantly different from their equivalent control (*P <.5; **P <.1). Knockout of UGT72B1 resulted in secondary cell wall thickening To investigate the potential effects of UGT72B1 knockout on cell wall formation, the inflorescence stems of plants were sectioned and stained by toluidine blue-o, which is often used to differentially stain polysaccharides and lignin and show the secondary cell wall. The ugt72b1 mutant displayed a thicker cell wall of pith cells at both the shoot tip area and basal area (Figure 3). Ultrathin sectioning and transmission electron microscopy were used to further investigate the cell wall of mutants. The mutant cell walls of pith cells were much thicker than that of wild-type (Figure 3). The measurement of pith cell walls of inflorescence stems indicated that the cell wall thickness in ugt72b1 mutants was increased by approximately three times that of wild-type (Table 1). In addition, thickness of cells that develop secondary cell walls such as tracheary elements or interfascicular fiber cells was also observed. We found that knockout of UGT72B1 resulted in thicker secondary cell walls or many more cells with secondary walls in interfascicular fibers and xylems. Together, these results demonstrated that UGT72B1 knockout had a crucial impact on wall thickening in floral stems. UGT72B1 catalyzed the glucose conjugation of monolignols UGT72B1 is a member of Arabidopsis family 1 glycosyltransferases (Li et al., 21). Previously, it was reported that UGT72B1 had activity toward xenobiotic pollutants 3,4-dichloroaniline and 2,4,5-trichlorophenol (Loutre et al., 23; Brazier-Hicks and Edwards, 25). However, the naturally occurring substrates of this enzyme were not known until now. To understand how UGT72B1 exerts its effects on cell wall lignification, we tested its possible glucosylating activity toward dozens of natural compounds including monolignols, phenolic acids, flavonoids, anthocyanins and plant hormones (Table S1). Recombinant UGT72B1 had glucosylating activity only for monolignols such as coniferyl alcohol, coniferyl aldehyde, dihydroconiferyl alcohol, p-coumaryl alcohol and p-coumaryl aldehyde. No activity was found toward other tested substrates (Table S1). Recombinant UGT72B1 catalyzed the formation of reaction products in reaction mixes with coniferyl alcohol, coniferyl aldehyde or dihydroconiferyl alcohol as the acceptors and UDP-glucose as sugar donor (Figure 4a c). We predicted that these reaction products were monolignol glucosides. In contrast, the negative control reactions with glutathione S-transferase (GST) peptide did not give the product peaks. Liquid chromatography mass spectrometry (LC-MS) analysis was used to further verify the identity of reaction products. It was known that molecular weight (M) of

6 UGT72B1 modulates cell wall lignification 31 (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) Figure 3. Knockout of UGT72B1 resulted in thicker secondary cell wall. (a c) Toluidine blue-o staining of floral stems of (a) WT, (b) mutants and (c) rescued lines at growth stage 6., showing the thicker pith cell wall in ugt72b1 mutant. (d f) Ultrathin sectioning and electron microscope pictures showed the thicker pith cell wall in (e) mutant than in (d) WT and (f) rescued line at growth stage 6.. (g i) Toluidine blue-o staining of floral stems of (g) WT, (h) mutants and (i) rescued lines at growth stage 6.5, showing the thicker pith cell wall in ugt72b1 mutant. (j l) Ultrathin sectioning and electron microscope pictures showed the thicker pith cell wall in (k) mutant than in (j) WT and (l) rescued line at growth stage 6.5. Scale bars: 2 lm and 5 lm, respectively, for microscopic pictures and electron microscopic pictures.

7 32 Ji-Shan Lin et al. (a) A285 nm µv (d) Intensity M+H + -Glc-OMe M+H + -Glc M+H M+NH Retention time (min) m/z (b) A3 nm µv (e) Intensity M+H M+ NH 4 + M+Na Retention time (min) m/z (c) µv (f) M+ NH 4 + -Glc M+ NH A285 nm Retention time (min) Intensity M+ NH 4 + -Glc-OH M+H m/z Figure 4. Enzyme activity of UGT72B1 toward monolignols. (a c) HPLC analysis of reaction products of UGT72B1 with (a) coniferyl alcohol, (b) coniferyl aldehyde and (c) dihydroconiferyl alcohol as the substrates, respectively. 1: authentic substrate standards; 2: negative control reactions of GST protein and different substrates; 3: reactions of UGT72B1 fusion protein and different substrates; 4: authentic coniferin standard. (d f) Mass spectrum identification of glucosylation products: (d) coniferin, (e) coniferyl aldehyde glucoside and (f) dihydroconiferyl alcohol glucoside. coniferyl alcohol glucoside (coniferin) is 342. In the positive ionization mode of MS analysis, the reaction product of coniferyl alcohol showed ion peaks at m/z (M+H + ), m/z (M+NH 4 + ), m/z (M+H + -Glc), m/z (M+H + -Glc-OMe), which were consistent with the expected protonated molecular ions of coniferin (Figure 4d). In addition, LC-MS analysis also confirmed the identity of putative reaction products resulting from coniferyl aldehyde and dihydroconiferyl alcohol (Figure 4e and f). All these m/z data were in accordance with their monolignol glucosides. Thus, the results showed that UGT72B1 had activity catalyzing the glucose conjugation of monolignols. In previous research, members of the UGT72E cluster were also demonstrated to have activity of glucosyl-conjugating

8 UGT72B1 modulates cell wall lignification 33 monolignols, with UGT72E2 having the highest activity toward monolignols (Lim et al., 21, 25). Thus, we compared the catalytic activity of UGT72B1 and UGT72E2 toward two important monolignols, coniferyl alcohol and coniferyl aldehyde. UGT72E2 showed much higher catalytic activities than UGT72B1 (Table 2). UGT72B1 was mainly expressed in young stem tissues To determine the tissue specificity of UGT72B1 function, we used b-glucuronidase (GUS) staining to analyze the expression pattern of UGT72B1pro:GUS for various tissues during all of plant development. A total of three independent transgenic lines were examined, and showed consistent GUS staining patterns. GUS staining mainly appeared in the upper and young part of floral stems, but slightly expressed in the main vein of young cauline leaf (Figure 5). Other tissues did not display clear GUS activity. The floral stem was further sectioned and stained to detect GUS activity. Interestingly, sections displayed extensive GUS staining in younger stem tissues including cortex, xylem and pith, but mainly in xylem (Figure 5g). However, sections on the older stems with differentiated interfascicular fibers gave GUS reaction only in the xylem (Figure 5h). In addition to GUS staining, we used quantitative real-time PCR (qrt-pcr) to analyze relative UGT72B1 expression in different organs and different stem developmental stages. The experimental results again showed the strongest UGT72B1 expression in young stem (Figure S4). These data indicated that UGT72B1 was predominantly expressed in young stem tissues, suggesting its important role in xylem formation or cell wall lignification, especially in early stem development. Chemical analysis of the monolignol glucosides in young stems Because UGT72B1 had in vitro activity toward monolignols and main expression in young stems, it was necessary to know whether or not the loss or gain of UGT72B1 function affected the levels of monolignol glucosides in young stems. Because UGT72B1 had the highest activity toward coniferyl alcohol and standard coniferin is available commercially, we examined the coniferin contents of plants with different genotypes. Our data indicated that UGT72B1 Table 2 Enzyme kinetics of UGT72B1 and UGT72E2 Substrates UGTs K m (mm) K cat (sec 1 ) K cat /K m (mm 1 s 1 ) Coniferyl alcohol UGT72B UGT72E Coniferyl aldehyde UGT72B UGT72E overexpression lines could synthesize much more coniferin in young stems than wild-type with or without the application of coniferyl alcohol (Figures S5 and S6; Table 3). The in vivo synthesized coniferin was also verified by LC-MS analysis (Figure S7). These results suggested that UGT72B1 has in vivo glucose-conjugating activity to monolignols in stem tissues. The coniferin extracted from ugt72b1 mutant young stems was also analyzed. Beyond our expectations, mutants did not possess less coniferin, but had more coniferin in stems than wild-type and the UGT72B1 overexpressors (Figure S5; Table 3). For the leaf sample, however, we did not find an obvious difference in monolignol glucoside level in different genotypes. To determine the reason for the enhanced biosynthesis of coniferin in mutant stems, we examined the expressions of three paralogs of UGT72B1: UGT72B2, UGT72B3 and UGT72E2. There was higher expression of UGT72B3 and UGT72E2 in young stems of mutants than in wild-type, but no difference in seedlings or leaves of both genotypes (Figure S8). Possibly, mutant plants overcompensated for UGT72B1 loss-offunction in stems by increased synthesis of monolignols, triggering the increased conjugation rate of UGT72B3 and UGT72E2. Expression patterns and in vitro enzymatic activity of UGT72B1 paralogs UGT72B1, UGT72B2 and UGT72B3 belong to group E of the UGT family 1 in Arabidopsis (Li et al., 21), which also includes UGT72E1, UGT72E2 and UGT72E3 (Figure S9). UGT72E members have been demonstrated to possess activity of glucosyl-conjugating monolignols, and UGT72E2 has the highest activity (Lim et al., 21, 25). Thus, it appears that UGT72B1 and UGT72E2 are redundant in function. In addition, UGT72B2 and UGT72B3 are most similar to UGT72B1 in sequences, and they share 82.7% amino acid identity with UGT72B1. Because so many paralogs exist in the same genome, why is that they do not complement the single mutant of UGT72B1 gene in the lignification phenotype mentioned above? To answer this question, we investigated the expression patterns and enzymatic activities for UGT72B1 paralogs. Analysis of the expression pattern revealed that UGT72B2 was mainly expressed in veins of cotyledons, rosette leaves, cauline leaves and sepals, but almost not in the floral stem (Figure S1). However, UGT72B3 was mainly expressed in basal petiole, young leaf margins and slightly in the floral stem (Figure S11). We also examined the expression of UGT72E2 in stem tissues of 72E2pro:: GUS transgenic plants, but found no obvious GUS staining (Figure S11f and g). To provide further evidence to evaluate gene expression, qrt-pcr was used to analyze the UGT72B1 gene and its paralogs. Experimental results showed that UGT72B3 and UGT72E2 had only a low level

9 34 Ji-Shan Lin et al. (a) (c) (b) (d) Figure 5. Expression pattern of UGT72B1 gene in different tissues. UGT72B1 expression was observed mainly at the young floral stem (c,e), but slightly in the main vein of young cauline leaf (f). Other tissues did not display clear expression activity (a,b,d). For stem section, UGT72B1 expression was observed mainly in differentiating xylems, cortex and pith in younger stem (g), while only observed in xylems in older stem with interfascicular fiber cells (h). Scale bars: 1mm(a f); 2 lm (g, h). (e) (f) (g) (h) of expression in stems, while their highest expression was in 14-day-old seedlings (Figure S4). These results indicated that UGT72B1, which was mainly expressed in stem tissues, had a different expression pattern from its paralogs. To examine the possible enzymatic activity of UGT72B2 and UGT72B3 toward monolignols, we tested seven monolignols as substrates: coniferyl alcohol, dihydroconiferyl alcohol, coniferyl aldehyde, sinapyl alcohol, sinapyl aldehyde, p-coumaryl alcohol and p-coumaryl aldehyde (Table S1). Enzymatic assays indicated that UGT72B3 could catalyze, to some extent, the glucosylation of only sinapyl aldehyde and coniferyl aldehyde (Figure S12). However, no enzymatic activity was detected for UGT72B2 toward the monolignols tested. Constitutively expressed UGT72B3 but not UGT72B2 rescued ugt72b1 mutant phenotypes To further investigate the potential in vivo enzyme activity of UGT72B2 and UGT72B3 toward monolignols and their effects on lignin synthesis, UGT72B2 and UGT72B3 were constitutively expressed under control of the CaMV 35S promoter in ugt72b1 mutant background. We found that UGT72B3 could rescue the ugt72b1 growth phenotype, but UGT72B2 could not (Figure S13). This experiment

10 UGT72B1 modulates cell wall lignification 35 Table 3 Chemical analysis of the monolignol glucosides from plants Plant materials Genotypes demonstrated that only UGT72B3 had a similar enzymatic activity to UGT72B1. Transcriptomic profiling of ugt72b1 mutants Coniferin (lmol/1 g) Young stem WT RE OE OE KO Young stem applied with.5 mm coniferyl alcohol WT OE OE WT, wild-type; RE, rescued lines; OE, overexpression lines; KO, mutants. Coniferin was used as the representative glucoside in this experiment. All data are means SE for three biological replicates (n = 3). As mentioned above, the level of monolignol glucosides was increased in ugt72b1 mutants, while an enhanced cell wall lignification was also observed in mutants. To understand the possible mechanism, we investigated the consequence of ugt72b1 mutation on the transcriptome by a genome-wide RNA sequencing. Transcript profiling was performed on two biologically independent sets of the wild-type and mutants of 4 5-week-old inflorescence stems at growth stage 6.. Each single sample was a pool of at least 3 individual plants to normalize for the expected variability in gene expression of individual plants. In the expression profiles, differentially expressed genes were identified by their increase/decrease in the mutant compared with the wild-type. Among the genes examined, 75 genes were remarkably upregulated and 226 were repressed in ugt72b1 mutant (Figure 6; Data S1). A functional classification of these significantly changed genes using Arabidopsis gene ontology annotations revealed that a large number of genes were related to phenylpropanoid biosynthesis, plant pathogen interaction, starch and sucrose metabolism, and pentose and glucoronate interconversion (Figure S14; Table S2). Perturbations in the UGT72B1 function seem to lead to widespread adjustments in carbohydrate metabolism, stress adaptation and development. It is noteworthy that the genes in almost every step of lignin synthesis pathway were widely upregulated. These genes included not only those involved in monolignol biosynthesis (PAL, CHS, C4H, CCR, 4CL, CAD, COMT, HCT and CCoAOMT), monolignol transporter (ABCG29) and putative transporter (ABCG4), active oxygen supporter (RBOHD) and lignin polymerization (PER64, PRX34 and PRX37), but also those involved in the regulation of secondary wall formation (NST1, MYB7 and MYB43; Figure 6c; Table S2). qrt-pcr analysis confirmed the enhanced transcription of lignification-related genes in ugt72b1 mutants Because transcriptomic profiling revealed a substantial and extensive alteration in expression level for many lignification pathway genes in ugt72b1 mutants, we further conducted qrt-pcr analysis to investigate the consequences of UGT72B1 disruption on the transcription of these genes related to lignin biosynthesis. Many genes associated with monolignol biosynthesis, monolignol transport, lignin polymerization and lignification-related TFs were selected for this investigation. We found that 3 genes including CCR2, COMT1, COMT2, HCT2, CAD1, CAD5, CAD8, 4CL, C4H, CCoAOMT1, PAL1, PRX34, PRX37, PRX71, PER64, LAC5, LAC12, LAC15, PME, RBOHA, RBOHD, RBOHF, ABCG29, ABCG4, MYB7, MYB43, MYB58, MYB63, NST1 and NST3 were all upregulated in ugt72b1 mutants, but restored to wild-type level in rescued lines (Figure 7). In addition, flavonoid and anthocyanin synthesis genes such as CHS and DFR were also upregulated in mutants in our analysis. The qrt-pcr results validated the transcriptomic profiling data. Thus, our research suggested that the enhanced expression of lignin pathway-related genes was possibly responsible for the aggravated and ectopic lignification resulting from the loss of function of UGT72B1, a glycosyltransferase catalyzing the glucose conjugation of monolignols. DISCUSSION The important role of UGT72B1 in lignin synthesis and cell wall lignification In previous research, several glycosyltransferases were demonstrated to possess activity toward monolignol glycosylation. For example, in vitro activity of Arabidopsis UGT72E1, UGT72E2 and UGT72E3 was identified for glucosyl-conjugating monolignols (Lim et al., 21, 25). When the UGT72E cluster was knocked down or overexpressed in Arabidopsis, the levels of coniferyl and sinapyl alcohol glucosides were significantly changed in roots or leaves, demonstrating the important role that glucosylation of secondary metabolites can play in cellular homeostasis (Lanot et al., 26). However, no clear alteration in growth phenotype or cell wall lignification was reported. UGT72B1 is another different monolignol glycosyltransferase identified in this study. Differing from the UGT72E cluster, the loss of UGT72B1 function resulted in obvious growth arrest and ectopic lignification in floral stems. Our analysis of expression pattern indicated that UGT72B1 and UGT72E2 (a member with strongest activity in its cluster) had different tissue specificity, in which UGT72B1 was mainly expressed in young stems, whereas UGT72E2 was mainly expressed in seedlings. Thus, specific tissue expression may be why

11 36 Ji-Shan Lin et al. (a) 4 (b) log1 (gene expression level of ko) Down-regulated Up-regulated log1 (gene expression level of WT) 4 (c) F5H LAC PER RBOH LAC PER RBOH LAC PER RBOH Figure 6. Transcriptomic profiling of ugt72b1 mutant. (a) Scatter plots of all expressed genes in pair WT-vs-KO. The x- and y-axes present log2 value of gene expression. Blue means downregulation gene, orange means upregulation gene, brown means non-regulation gene. (b) Number of up/downregulated genes in ugt72b1 mutant. (c) Upregulated genes involved in lignin synthesis pathway.

12 UGT72B1 modulates cell wall lignification (a) WT KO RE 12 1 (b) CCR2 COMT2 HCT2 CAD1 CAD5 CAD8 4CL C4H CCoAMOT1 PAL1 COMT (c) 3 25 (d) ABCG4 ABCG29 DFR CHS PRX34 PRX37 PER64 PRX71 LAC5 LAC12 LAC15 PME (e) (f) RBOH A RBOH D RBOH F MYB7 MYB43 MYB58 MYB63 NST1 NST3 Figure 7. Quantitative RT-PCR confirmed the activation of genes responsible for secondary wall biosynthesis in ugt72b1 mutant. (a,b) The activation of monolignol biosynthetic genes in mutant. (c) The activation of ABC transporter genes (ABCG19, ABCG4) and anthocyanin biosynthetic genes (CHS, DFR). (d,e) The activation of genes involved in lignin polymerization, such as PRXs, LACs, RBOHs. (f) The expression of transcription factor genes involved in the regulation of secondary wall formation was also upregulated in the ugt72b1 mutant. Values are means SD of three biological replicates. UGT72B1 and UGT72E exerted different phenotypic influences after loss of their functions. Chapelle et al. (212) investigated the impact of the absence of stem-specific beta-glucosidases (BGLU45, BGLU46 and BGLU47) on lignin and monolignols. BGLU45 and BGLU46 proteins are mainly located in the interfascicular fibers and in the protoxylem. Although BGLU45 and BGLU46 mutant lines displayed a significant increase in coniferin content, knockout mutants for BGLU45 or BGLU46 do not have a lignin-deficient phenotype, suggesting that monolignol glucosides are the storage form of monolignols in Arabidopsis, but not the direct precursors of lignin (Chapelle et al., 212). Recently, a monolignol transport experiment was performed using poplar xylem tissues. Its results indicated that membrane vesicles prepared from differentiating xylem tissues had clear ATP-dependent transport activity toward coniferin, suggesting involvement of monolignol glucosides (especially coniferin) in active lignification (Tsuyama et al., 213). In contrast, the present study highlights an important role of monolignol glycosylation in maintaining the homeostasis of the monolignol biosynthesis pathway and lignin deposition by investigation of ugt72b1 mutants. Our results suggested that monolignol glycosylation catalyzed by UGT72B1 played an essential role in maintaining the balance of monolignol metabolism and normal cell wall lignification. Thus, UGT72B1 working as a glycosyltransferase may be a key component in modulating cell wall lignification of plants.

13 38 Ji-Shan Lin et al. Phenotype of ugt72b1 mutant reflects the complicated feedback regulation mechanism of lignin metabolism Lignification is a very complex metabolism network and includes monolignol biosynthesis, glycosylation, transport, lignin polymerization and other related processes. The transcriptomic profiling and qrt-pcr analysis of the ugt72b1 mutant revealed that monolignol glycosylation catalyzed by UGT72B1 could give rise to extensive effects on cellular metabolic pathways. Global changes in gene expressions involved in the whole lignification pathway were observed, including upstream TFs, monolignol biosynthetic enzymes, monolignol transporters and various enzymes of lignin polymerization. In addition, there were multiple aspects of the mutant phenotype, such as abnormal lignification, growth arrest, fertility reduction and pigment deposition. These observations implied that the disturbance of monolignol glycosylation had an intricate and global feedback regulation on whole lignin synthesis and even on other pathways. Several previous studies proposed that monolignols would be conserved as monolignol glucosides in vacuolar vesicles, which could maintain the balance of monolignol metabolism and maintain normal cell growth (Whetten and Sederoff, 1995; Boerjan et al., 23; Miao and Liu, 21; Liu et al., 211). In this study, the 72B1 mutation resulted in the enhanced lignification. Thus, it is possible that competition between monolignol glucosylation and lignin formation exists. However, the mechanism involved may be more intricate. For example, our analysis indicated that ugt72b1 mutants contained more monolignol glucosides and showed multiple growth defects, suggesting that a more complex regulation may be involved in the relationship of monolignol glucosylation and lignin formation. Feedback regulation of metabolic pathways is a common phenomenon; for example, a mutation of REF8 gene of lignin biosynthesis pathway led to lignin deficiency and also stunted growth due to the loss of C3 H enzyme activity. Bonawitz et al. (214) found that a mutation of transcriptional co-regulator mediator could rescue the lignin and stunted growth, suggesting that an unknown signaling pathway or sensor is responsible for the feedback of lignin deficiency information to the nucleus and the subsequent massive changes in gene expression. This research would be valuable for understanding the ugt72b1 mutant. Considered together with the established UGT72B1 enzyme activity, metabolite analysis and lignin-related transcriptomic data presented in this study led us to propose a possible working model for UGT72B1 (Figure 8). When UGT72B1 works normally, monolignol glucosylation catalyzed by UGT72B1 maintains a balance between monolignols and their glucosides. Here, the level of monolignol glucosides could be used as an indicator to mark the normal level of monolignols inventory. This situation may be sensed by an as-yet-undiscovered sensor, fed back to the upstream TFs and inhibiting the TF expression, thus Phe Sensor? NST1/3 MYB58 MYB61 MYB63 PAL C4H 4CL HCT CCoAOMT CCR COMT CAD p-coumaroyl-coa CHS DFR Anthocyanins Monolignol glucosides 72B1 Monolignols ABCG RBOH LAC PER PME Lignin Figure 8. UGT72B1 working model proposed by this study. When UGT72B1 works normally, monolignol glucosylation catalyzed by UGT72B1 maintains a balance between monolignols and their glucosides. Here, the level of monolignol glucosides could be used as an indicator to mark the normal level of monolignols inventory. This situation may be sensed by an as-yetundiscovered sensor, fed back to the upstream TFs and inhibiting the TF expression, thus keeping the lignin-related genes at relatively low expression levels and normal cell wall lignification. However, when UGT72B1 does not work, the balance between monolignols and glucosides may be broken. The changed level of monolignol glucosides could indicate the inventory is not enough for the monolignol supply. This abnormal situation may be sensed by the same sensor, fed back to the upstream TFs and thus releasing the expression inhibition. Upstream TFs activate their own expressions, upregulate various kinds of ligninrelated genes and increase the phenylpropanoid flux, leading to the aggravated and ectopic lignification. In addition, the increased phenylpropanoid flux might strengthen branch pathways (e.g. flavonoid and anthocyanin biosynthesis), leading to the observed anthocyanin accumulation.

14 UGT72B1 modulates cell wall lignification 39 keeping the lignin-related genes at relatively low expression levels and normal cell wall lignification. However, when UGT72B1 does not work, the balance between monolignols and glucosides may be broken. The changed monolignol glucosides could indicate that the inventory is not enough for the monolignol supply. This abnormal situation may be sensed by the same sensor, fed back to the upstream TFs and thus releasing the expression inhibition. Upstream TFs activate their own expressions, upregulate various kinds of lignin-related genes and increase the phenylpropanoid flux, leading to the aggravated and ectopic lignification. Here, we proposed that mutant plants possibly overcompensate for UGT72B1 loss-of-function by the increased synthesis (metabolic flux) of monolignols, thus the increased monolignol glucosides in mutants may be regarded a secondary effect of UGT72B1 loss-of-function. In addition, the increased phenylpropanoid flux might strengthen branch pathways (e.g. flavonoid and anthocyanin biosynthesis), leading to the observed anthocyanin accumulation (Figure 8). Expression specificity and functional differentiation of UGT72B1 and its paralogs Why did the developmental deficiency exist only in the floral stem in ugt72b1 mutants? We propose that this was due to the tissue specificity of UGT72B1 expression. Expression pattern analysis of GUS staining showed that UGT72B1 was mainly expressed in the upper and young part of floral stems, which should be the area of fastest stem elongation. Furthermore, GUS staining of stem sections indicated that UGT72B1 was predominantly expressed in differentiating xylem as well as the cortex and pith, suggesting its important function in xylem formation or cell wall lignification. This is consistent with the deficient phenotypes of ectopic lignification and inhibited growth occurring mainly in floral stems of ugt72b1 mutants. In addition, although UGT72B1 was mainly expressed in xylem, its expression was actually distributed in whole stem tissues (e.g. pith, cortex and xylem) in the early development stage. Thus, it is understandable that functional disruption of UGT72B1 resulted in ectopic lignification in pith cells and other cells of the stem. The phenylpropanoid pathway is used in biosynthesis of a wide range of soluble secondary metabolites, including hydroxycinnamic acid esters, flavonoids and the precursors of lignin and lignans. For example, flavonoids and anthocyanins are direct metabolites of the phenylpropanoid pathway. In the ugt72b1 mutant, the metabolic flux of the phenylpropanoid pathway to monolignols and lignin was enhanced, and possibly the flux to anthocyanin synthesis was also increased. Thus, anthocyanin accumulation was observed in shoot tips of mutants. Many other studies have shown that cell wall lignification is important for plant growth and development (Zhou et al., 29; Zhao et al., 21, 213). In the present study, enhanced and ectopic lignification also caused growth arrest and fertility problems. Our results are consistent with previous findings. In Arabidopsis, three other glycosyltransferases (UGT72E1, UGT72E2 and UGT72E3) have demonstrated activity in catalyzing monolignol glycosylation, and UGT72E2 has the highest activity toward monolignols (Lim et al., 21, 25). We compared UGT72E2 and UGT72B1 activities for their shared substrates. The Kcat/Km of UGT72E2 and UGT72B1 for coniferyl alcohol was approximately 2.78 and.74, respectively; and correspondingly for coniferyl aldehyde, approximately and.3. UGT72E2 showed much higher activity than UGT72B1. However, no significant change in lignin deposition was previously observed in knockdown or knockout lines of UGT72E2 (Lanot et al., 26). We investigated UGT72E2 expression in different organs or tissues and found that it was mainly expressed in roots and leaves, only at a very low level in stems, which contrasted with the UGT72B1 expression pattern and may explain why the UGT72E2 mutant did not show obvious lignin phenotype. UGT72B2 and UGT72B3 are the closest phylogenetic members to UGT72B1, and we found that UGT72B2 was not expressed but UGT72B3 was slightly expressed in floral stems. Moreover, our biochemical analysis indicated that UGT72B3 could catalyze the glucosylation of sinapyl aldehyde and coniferyl aldehyde, whereas UGT72B2 had no activity toward monolignols. Thus, when constitutively expressing UGT72B3 in the ugt72b1 mutant, phenotypes were restored to wild-type. How the levels of lignin are controlled within particular bounds is a particularly interesting question, considering that lignin is an extracellular insoluble polymer. In this research, the substrate preference and special expression pattern of UGT72B1 and its paralogs might suggest their functional differentiation in maintaining lignin levels within particular tissues and organs. CONCLUSIONS This study demonstrated the essential role of monolignol glycosylation in lignin biosynthesis. Acting as a monolignol glycosyltransferase and functioning specifically in stems, UGT72B1 may be a key component in promising normal cell wall lignification of plants. However, the mechanism by which monolignol glycosylation catalyzed by UGT72B1 is fed back to the whole lignin pathway remains to be identified. EXPERIMENTAL PROCEDURES Plant materials and growth conditions All wild-type and mutant A. thaliana were in the Columbia- background. Two T-DNA insertion mutants, ugt72b1ko-1 (SALK_49597) and ugt72b1 ko-2 (SAIL_611_E4), were used for the functional analysis of glycosyltransferase gene UGT72B1 (At4 g17). These

15 4 Ji-Shan Lin et al. knockout mutants were ordered from NASC ( and confirmed by PCR and RT-PCR (the primers are listed in Table S3). Arabidopsis plants were grown in a greenhouse at 22 1 C under 16-h light (approximately 1 lmol m 2 sec 1 ) and 8-h darkness. When plants reached growth stage ~6. after growing for 4 5 weeks and growth stage ~6.5 after growing for 6 7 weeks, the floral stems were used for this research. Complementation analysis of the Arabidopsis ugt72b1 mutant Firstly, a HindIII EcoRI restriction fragment containing CaMV35S promoter GUS gene Nos terminator cassette was cut from plasmid pbi121 and inserted between HidIII and EcoRI sites of binary plasmid pcambia132 to create a modified binary vector. Then, the 1.5-kb promoter sequence upstream of the start site (ATG) of UGT72B1 gene (TAIR, was amplified from Arabidopsis genomic DNA using PCR and cloned into the modified pcambia132 to replace the CaMV35S promoter. The 1443-bp cdna of UGT72B1 gene (TAIR) was amplified by RT-PCR and cloned downstream of the 1.5-kb UGT72B1 promoter to replace the GUS gene. The resulting plasmid construct was transformed into ugt72b1 mutant by the floral dip method (Clough and Bent, 1998). The generated transformants were selected on solid Murashige and Skoog (MS) medium supplemented with 4 mg L 1 of hygromycin B. Independent T 1 kanamycin-resistant lines were selected, and homozygous T 3 progeny of rescued lines were isolated. The expression level of UGT72B1 in rescued lines was determined by RT-PCR. TUBULIN2 (TUB) gene was used as the internal control for RT-PCR analysis. Histochemical lignin assay Plants were collected at development stages 6. and 6.5 (Boyes et al., 21), and the cross-sections of floral stems were made at the regions between the first and second branch (from base to apex). For visualization of lignified tissues, paraffin sectioning was performed, and Wiesner, Ma ule and toluidine blue staining were carried out. Paraffin sections (thickness 7 lm) were placed onto glass slides and stained with appropriate reagents as previously described (Pomar et al., 22; Shigeto et al., 215). Transmission and scanning electron microscopy For observation of subcellular structures of floral stems, samples were fixed in 2.5% (v/v) glutaraldehyde and 4% paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA, USA) in phosphatebuffered saline (PBS, ph 7.2) at 4 C for 12 h, then washed with PBS, dehydrated in a graded ethanol series, and embedded in LR White resin (London Resin Berkshire, Reading, UK). The resin was polymerized at 55 C for 3 days. Cross-sections (.5 lm) were cut with a diamond knife on a Leica EM UC7 ultramicrotome (Leica Mikrosysteme, Wetzlar, Germany). Sections of 85 nm thickness were post-stained with uranyl acetate and lead citrate, and observed using a Zeiss EM 92A transmission electron microscope (Carl Zeiss, Oberkochen, Germany). Anthers from newly opened flowers were used for scanning electron microscopy directly, and images were captured using a FEI Quanta25 FEG environmental scanning electron microscope with a 5-kV beam, and the signal collected using the GSED detector. Pollen germination assay In vitro pollen germination was performed according to Wu et al. (21). Briefly, pollen was isolated from newly opened flowers and placed on pollen germination medium [1 mm CaCl 2, 1 mm MgSO 4, 1 mm Ca(NO 3 ) 2,.1% (w/v) H 3 BO 3, 18% (w/v) sucrose and.5% (w/v) agar, ph 7.]. Pollen were cultured in slide glass with pollen germination medium at 25 C under moist conditions and observed using an Olympus BX61 fluorescence microscope. Analysis of lignin composition The Arabidopsis floral stem tissues (2 3 g fresh weight) were harvested at the growth stage ~6. after growing for 4 5 weeks. The collected samples were ground in liquid nitrogen and lyophilized. The extractive-free cell wall residue (CWR) samples were prepared as described by Chen and Dixon (27). Lyophilized extractive-free material was used for lignin analysis. The thioacidolysis method was used to determine lignin composition (Lapierre et al., 1995). Tissue-specific expression assay The 1.5-kb promoter sequences upstream of the start site (ATG) of UGT72B1 (At4g17), UGT72B2 (At1g139), UGT72B3 (At1g142) and UGT72E2 (At5 g6669) (TAIR) were amplified from Arabidopsis genomic DNA using PCR and cloned into plasmid pbluescript SK(+). Then the promoters were separately subcloned into the pbi121 binary vector to replace the CaMV35S promoter and create the GUS reporter constructs. The resulting constructs were transformed into wild-type Arabidopsis, and transgenic plants were selected on plates containing 5 mg L 1 kanamycin. Homozygous T 3 transgenic plants were visualized by staining for GUS activity for 4 6 h as described by Wang et al. (212). Purification and enzyme activity assays of UGT72B1, UGT72B2, UGT72B3 and UGT72E2 Full-length cdna of UGT72B1, UGT72B2, UGT72B3 and UGT72E2 were amplified from Arabidopsis by RT-PCR with primers (listed in Table S3), and cloned into prokaryote expression plasmid pgex-2t. Soluble recombinant proteins corresponding to UGT72B1, UGT72B2, UGT72B3 and UGT72E2 were respectively induced to express in Escherichia coli, and were purified according to Hou et al. (24). Natural compounds used for enzyme activity assay were purchased from several companies. Monolignols, phenolic acids, coniferin, syringin and UDP-glucose were from Meryer (Shanghai) Chemical Technology. Flavonoids and anthocyanidins were from Carbosynth China. Plant hormones were from OlChemIm (Czech Republic) or Sigma-Aldrich. The glycosyltransferase activity assay was carried out as described by Lim et al. (21) with modification. The assay mix (2 ll) contained.2 lg of recombinant protein, 14 mm 2-mercaptoethanol, 5 mm UDP glucose and 1 mm substrate (Table S1). Initial screening for activity of the enzymes against different substrates was carried out in the buffer of 1 mm Tris HCl (ph 7.) at 3 C for 1 h. For kinetic analysis of the enzymes, reactions were carried out at ph 7. and 3 C for 3 min due to linearity of the reactions. The reaction was stopped by addition of 2 ll trichloroacetic acid (24 mg ml 1 ), quick-frozen and stored at 2 C prior to the reverse-phase high-performance liquid chromatography (HPLC) analysis. Extraction and analysis of soluble monolignol glucosides from plants The one tube method was used to extract soluble monolignol glucosides from young stems or seedlings. In detail,.2 g of frozen tissue was added into a 2-ml centrifuge tube and ground into