VASCULAR-RELATED NAC-DOMAIN 7 directly regulates the expression of a broad range of genes for xylem vessel formation

Authors

  • Masatoshi Yamaguchi,

    1. Graduate School of Biological Sciences, Nara Institute of Science and Technology, Ikoma, Nara 630-0192, Japan
    2. RIKEN Plant Science Center, Yokohama, Kanagawa 230-0045, Japan
    Search for more papers by this author
  • Nobutaka Mitsuda,

    1. Bioproduction Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8562, Japan
    Search for more papers by this author
  • Misato Ohtani,

    1. RIKEN Plant Science Center, Yokohama, Kanagawa 230-0045, Japan
    2. RIKEN Biomass Engeneering Program, Yokohama, Kanagawa 230-0045, Japan
    Search for more papers by this author
  • Masaru Ohme-Takagi,

    1. Bioproduction Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Ibaraki 305-8562, Japan
    Search for more papers by this author
  • Ko Kato,

    1. Graduate School of Biological Sciences, Nara Institute of Science and Technology, Ikoma, Nara 630-0192, Japan
    Search for more papers by this author
  • Taku Demura

    Corresponding author
    1. Graduate School of Biological Sciences, Nara Institute of Science and Technology, Ikoma, Nara 630-0192, Japan
    2. RIKEN Plant Science Center, Yokohama, Kanagawa 230-0045, Japan
    3. RIKEN Biomass Engeneering Program, Yokohama, Kanagawa 230-0045, Japan
      (fax +81 743 72 5460; e-mail demura@bs.naist.jp).
    Search for more papers by this author

(fax +81 743 72 5460; e-mail demura@bs.naist.jp).

Summary

The Arabidopsis thaliana NAC domain transcription factor, VASCULAR-RELATED NAC-DOMAIN7 (VND7), acts as a key regulator of xylem vessel differentiation. In order to identify direct target genes of VND7, we performed global transcriptome analysis using Arabidopsis transgenic lines in which VND7 activity could be induced post-translationally. This analysis identified 63 putative direct target genes of VND7, which encode a broad range of proteins, such as transcription factors, IRREGULAR XYLEM proteins and proteolytic enzymes, known to be closely associated with xylem vessel formation. Recombinant VND7 protein binds to several promoter sequences present in candidate direct target genes: specifically, in the promoter of XYLEM CYSTEINE PEPTIDASE1, two distinct regions were demonstrated to be responsible for VND7 binding. We also found that expression of VND7 restores secondary cell wall formation in the fiber cells of inflorescence stems of nst1 nst3 double mutants, as well as expression of NAC SECONDARY WALL THICKENING PROMOTING FACTOR3 (NST3, however, the vessel-type secondary wall deposition was observed only as a result of VND7 expression. These findings indicated that VND7 upregulates, directly and/or indirectly, many genes involved in a wide range of processes in xylem vessel differentiation, and that its target genes are partially different from those of NSTs.

Introduction

Xylem vessels function in the transport of water and soluble minerals from the roots throughout the plant. Xylem vessels are composed of a special type of cells, tracheary elements (TEs). TEs undergo well-defined developmental processes of differentiation, including specification, enlargement, secondary cell wall deposition, programmed cell death and removal of the cell wall prior to cell-cell fusion (Fukuda, 2004). In order to build up continuous networks, TE differentiation must be coordinated between adjacent TEs. Thus, xylem vessel differentiation requires a strict regulating system.

Recent work in Arabidopsis has revealed a part of the regulatory system for xylem vessel formation. The NAC domain transcription factors, named VASCULAR-RELATED NAC-DOMAIN1–7 (VND1VND7), were originally isolated as genes for which expression levels are elevated during transdifferentiation into TEs, in an induction system using Arabidopsis suspension cultured cells (Kubo et al., 2005). VND7, in particular is expressed in all types of differentiating vessels (Kubo et al., 2005; Yamaguchi et al., 2008). Overexpression of VND7 induces the transdifferentiation of various types of cells into xylem vessel cells (Kubo et al., 2005), suggesting that VND7 plays a crucial role in xylem vessel differentiation. VND1VND6 are also preferentially expressed in developing vascular cells; the VND7 protein forms homodimers as well as heterodimers with any of the other VND proteins (Kubo et al., 2005; Yamaguchi et al., 2008). These data suggest that VND1–VND6 function redundantly and/or cooperatively with VND7 in the formation of xylem vessels. The genes NAC SECONDARY WALL THICKENING PROMOTING FACTOR1 (NST1) and NST3 (also called SECONDARY WALL-ASSOCIATED NAC DOMAIN PROTEIN1 or ARABIDOPSIS NAC DOMAIN CONTAINING PROTEIN012) belong to the gene family that is phylogenetically closest to the VND gene family, and are known to be the key regulators of differentiation of Arabidopsis fiber cells (Zhong et al., 2006, 2007b; Ko et al., 2007; Mitsuda et al., 2007).

We have demonstrated that continuous overexpression of VND7 induces the expression of a number of genes involved in various aspects of xylem vessel differentiation. These functions include biosynthesis of the secondary wall components, such as cellulose, hemicellulose and lignin, and programmed cell death (Yamaguchi et al., 2010b). We also showed that several transcription factors, such as MYB46 and LOB DOMAIN PROTEIN30 (LBD30)/ASYMMETRIC LEAVES2 LIKE19 (ASL19)/JAGGED LATERAL ORGANS (JLO), are upregulated by VND7. These genes induce ectopic secondary wall formation (Zhong et al., 2007a; Soyano et al., 2008; Ko et al., 2009), suggesting that VND7 is involved in a complex transcriptional network regulating xylem vessel differentiation (Caño-Delgado et al., 2010; Demura and Ye, 2010; Yamaguchi and Demura, 2010). Thus, we must distinguish direct targets of VND7 from indirect downstream genes by using carefully controlled experimental systems. Several studies reported that NSTs, and possibly VND6 and VND7, directly regulate several transcription factors (Zhong et al., 2007a, 2008; McCarthy et al., 2009). However, information about the direct targets of VND7 is still limited.

Recently, we have shown that upregulation of VND7 activity using a glucocorticoid receptor (GR)-mediated post-translational induction system effectively induces transdifferentiation of various cell types into xylem vessel cells (Yamaguchi et al., 2010a). This post-translational induction system is a useful system for the identification of direct target genes (Sablowski and Meyerowitz, 1998; Sakai et al., 2001; Ito et al., 2004; Okushima et al., 2007). In brief, the system functions as follows: in the absence of glucocorticoid, a GR fusion protein is localized in the cytosol, and is inactivated by forming protein complexes with heat shock proteins. When glucocorticoid binds to GR, GR dissociates with the heat shock proteins and translocates into the nucleus, where it influences gene expression (Aoyama and Chua, 1997). In this study, we used this system to isolate a number of candidates for direct target genes of VND7; several genes known to be associated with xylem vessel differentiation were included. We confirmed the direct binding of VND7 protein to the promoters of several genes. We also found that expression of VND7 restores secondary cell wall formation in fiber cells of inflorescence stems of the nst1 nst3 double mutant as well as expression of NST3; however, the vessel-type secondary wall deposition was observed only in response to VND7 expression.

Results

VND7 directly regulates a number of genes involved in xylem vessel differentiation

As shown previously, overexpression of VND7 fused to yellow fluorescence protein (VND7-YFP) induces the expression of many genes related to the differentiation of vascular vessels (Yamaguchi et al., 2010b). Microarray analysis revealed that 300 genes are upregulated by more than twofold in the transgenic VND7-YFP plant, with a false discovery rate (FDR) of <0.1 (< 0.028) (Table S1). To identify direct target genes of VND7 among the upregulated genes, we used a glucocorticoid-mediated post-translational induction system (Sablowski and Meyerowitz, 1998). We generated a transgenic Arabidopsis plant expressing VND7 fused C-terminally to the activation domain of VP16 and the hormone-binding domain of GR, under the control of the cauliflower mosaic virus (CaMV) 35S promoter (VND7-VP16-GR; Yamaguchi et al., 2010a). In this transgenic plant, transdifferentiation into xylem vessel cells was effectively and tightly dependent on treatment with dexamethasone (DEX), a glucocorticoid derivative (Yamaguchi et al., 2010a). We subjected 10-day-old seedlings of VND7-VP16-GR plants to pre-treatment with cycloheximide (CHX), a protein synthesis inhibitor to prevent the production of secondary transcription factors that could then influence gene transcription, followed by treatment with or without DEX. Microarray analysis using these samples revealed that, among the 300 genes upregulated in the VND7-YFP plants, 63 were also upregulated (fold change > 2; FDR < 0.1; P < 0.026) in the VND7-VP16-GR plants in response to DEX treatment (Figure 1; Table 1). No genes among these 63 were upregulated by DEX treatment in transgenic plants harboring vector control (fold change > 2; P < 0.05); thus, we concluded that these genes are the candidates for direct target genes of VND7. It should be noted that only three genes were selected as downregulated in the VND7-VP16-GR plants in response to DEX treatment (At2g14520, At5g02650 and At5g47340; fold change < 0.5; P < 0.05).

Figure 1.

 Expression profiles of downstream genes of VND7 during differentiation to tracheary elements (TEs).
Expression profiles of 300 downstream genes of VND7, identified in an analysis of VND7-YFP plants (Table S1; Yamaguchi et al., 2010b) during transdifferentiation to TEs in an Arabidopsis cell culture system (Kubo et al., 2005). Results are summarized as a heat map. Colors in the heat map represent the fold changes of gene expression during transdifferentiation to TEs, with reference to samples on day 0. Red marks on the left side represent 63 putative direct target genes of VND7 that are newly defined by this study (see Tables 1 and S2).

Table 1.   Sixty-three putative direct target genes of VND7
AGI no.DescriptionVND7-YFP/YFPVND7−VP16−GR + DEX ± CHX/CHXVP16−GR + DEX ± CHX/CHX
Fold changePFold changePFold changeP
  1. The fold changes (VND7-YFP/YFP), (VND7-VP16-GR + DEX + CHX/CHX) and (VP16-GR + DEX + CHX/CHX) calculated from normalized values; P values are given. *Genes containing TERE–like motifs in their promoter regions (the regions 1000-bp upstream from the transcription initiation sites).

Transcription factors
At4G00220*LBD30/ASL19/JLO6.113.94E–0520.222.15E–041.3720.570
At5G12870MYB465.978.17E–048.971.22E–071.6600.358
At2G40470*LBD15/ASL112.801.48E–035.714.19E–040.9370.924
At1G63910MYB1032.701.89E–023.381.70E–030.8830.748
Signal transduction
At3G52820Purple acid phosphatase PAP225.964.32E–066.532.05E–052.1950.398
At1G55180Phospholipase Dε7.271.04E–044.481.50E–020.7410.536
At1G01900*Subtilase family protein/AtSBT1.14.644.19E–043.424.97E–041.0580.812
At2G25440Receptor like protein 20/AtRLP203.465.52E–033.362.20E–041.6840.258
At1G54790*GDSL-motif lipase4.621.42E–032.562.60E–020.9690.970
At1G14240Nucleoside phosphatase family protein GDA12.543.02E–032.412.90E–031.3030.285
At4G18550Lipase class-3 family protein2.931.04E–032.111.30E–030.9570.928
At4G11470*Cys rich receptor like kinase 31/CRK313.073.02E–032.061.10E–031.2460.321
Cell wall
At1G70500*Polygalacturonase, putative12.362.33E–068.812.55E–061.6890.525
At1G23460Polygalacturonase4.506.00E–037.796.07E–051.4730.543
At3G47400Pectinesterase family protein4.234.09E–057.242.07E–052.0770.123
At5G38610Invertase/Pectin methylesterase inhibitor family protein6.672.11E–056.021.99E–061.3830.424
At5G01930(1-4)-beta-mannan endohydrolase, putative2.294.18E–054.308.49E–051.0430.876
At4G18780Cellulose synthase IRX1/CesA85.641.13E–063.531.34E–031.0720.743
At2G38080Laccase 4/LAC4/IRX124.111.02E–042.703.50E–030.9510.844
At5G44030*Cellulose synthase IRX5/CesA44.802.19E–052.697.60E–041.3010.142
At5G03170*IRX13/FLA112.021.96E–032.329.04E–041.1380.460
At5G07080Transferase family protein6.861.17E–042.311.79E–021.1060.701
At4G33330Plant glycogenin-like starch initiation protein 3/GSIP33.411.86E–032.211.18E–031.2730.471
At4G08160Glycosyl hydrolase family 10/AtXyn34.031.72E–052.051.22E–030.9620.890
Cell death
At2G31980Cys proteinase inhibitor-related7.341.16E–049.271.50E–051.4600.507
At4G35350*Xylem Cys protease1/XCP15.785.77E–074.457.61E–051.6030.494
At5G04200*Metacaspase 9/ATMC94.775.62E–054.112.01E–040.9430.894
At4G12910Ser carboxypeptidase-like 20/SCPL209.942.98E–064.011.39E–041.1640.556
At1G20850*Xylem Cys protease2/XCP23.854.89E–073.462.34E–051.0910.779
At4G04460Aspartyl protease family protein3.276.41E–053.231.32E–031.2830.297
At3G45010Ser carboxypeptidase-like 48/SCPL484.272.69E–072.889.02E–041.1650.473
Unknown function
At5G40020Thaumatin family protein4.476.22E–0510.076.99E–070.6240.603
At1G58300Heme oxygenase 4/HO47.826.21E–047.944.45E–051.1100.872
At1G23470Pseudogene4.506.00E–037.796.07E–051.4730.543
At1G24600Unknown protein2.431.94E–037.348.60E–040.8740.779
At2G46760FAD-binding domain-containing protein12.701.69E–057.324.31E–051.8950.127
At5G60720Unknown protein4.113.76E–054.831.91E–051.1910.623
At1G10800Unknown protein5.794.89E–064.724.48E–070.9420.934
At5G19870Unknown protein5.816.48E–054.497.19E–041.2840.596
At3G52900Unknown protein2.223.44E–044.451.23E–041.1000.762
At5G11540FAD-binding domain-containing protein3.784.90E–034.431.58E–051.3790.527
At1G29200Unknown protein3.622.44E–054.103.98E–061.7650.122
At3G27200Plastocyanin-like protein4.002.17E–053.753.39E–041.3680.414
At5G01360Unknown protein4.162.45E–053.611.03E–021.3720.382
At2G38320Unknown protein6.494.44E–053.591.82E–041.3980.385
At2G04850*Auxin-responsive protein-related3.994.11E–063.546.00E–041.3550.620
At3G53350ROP interacting partner 4/RIP44.111.04E–043.471.18E–041.5340.163
At4G23690Disease resistance-responsive family protein2.514.39E–053.394.88E–031.4330.210
At2G46710RAC GTPase activating protein, putative2.071.44E–033.392.32E–050.9770.924
At5G45320Unknown protein2.102.66E–032.879.01E–041.0060.983
At5G45970Arabidopsis RAC-like2/ARAC22.092.22E–022.758.44E–041.0440.864
At3G62020*Germin-like protein 10/GLP105.639.66E–072.722.10E–031.1160.791
At5G49900*Catalytic/glucosylceramidase3.451.37E–042.698.41E–041.2170.486
At5G49690UDP-glucoronosyl/UDP-glucosyl transferase family protein3.369.12E–032.607.55E–041.1570.613
At1G69680*Unknown protein2.031.77E–042.473.85E–041.1090.672
At1G05450Protease inhibitor prtein/lipid transfer protein2.801.64E–022.301.08E–021.0270.923
At5G53590Auxin-responsive family protein2.602.13E–052.249.64E–051.0300.896
At5G53588CPuORF502.602.13E–052.249.64E–051.0300.896
At1G63300Unknown protein4.012.10E–032.188.47E–051.1390.639
At5G15490UDP-glucose 6-dehydrogenase, putative2.344.35E–052.161.30E–031.2260.167
At1G75090Methyladenine glycosylase family protein3.831.32E–032.104.54E–031.4690.229
At1G27920Microtuble-associated protein 8/MAP65-83.222.39E–042.071.31E–020.9940.982
At5G07800Flavin-containing monooxygenase family protein2.683.02E–032.031.18E–031.2250.394

Most of the 300 downstream genes of VND7 are transiently induced during in vitro TE transdifferentiation of Arabidopsis suspension cells (Figure 1; Table S2; Kubo et al., 2005). We previously defined 1705 genes showing changes in expression over the time course, and classified them into 23 sets according to their expression patterns (Kubo et al., 2005). Downstream genes, including putative direct targets of VND7, are apparently enriched in sets 3 and 8, which showed upregulated expression just when the xylem vessel elements were actively forming, and included VND7 itself (Table S3). These genes encode well-known factors of xylem vessel differentiation, such as transcriptional factors (MYB46 and LBD30/ASL19/JLO; Zhong et al., 2007a; Soyano et al., 2008; Nakano et al., 2010), cellulose synthase subunits (CesA4/IRX5 and CesA8/IRX1; Taylor et al., 2003), a laccase (LAC4/IRX12; Brown et al., 2005), xylem cysteine proteases (XCP1 and XCP2; Zhao et al., 2000; Funk et al., 2002) and a serine protease AtSBT1.1 catalyzing the maturation of a phytosulfokine peptide hormone (Matsubayashi et al., 1999; Srivastava et al., 2008; Motose et al., 2009), were transiently induced during differentiation into vessel elements. These data suggest that our attempt to identify the direct target genes of VND7 was quite reasonable. Importantly, several transcription factors were found among the candidates for direct target genes of VND7 (Table 1). The expression levels of MYB46 and MYB103 were increased in the VND7-VP16-GR seedling after DEX treatment, consistent with the reports that MYB46 and MYB103 are directly regulated by NST3, and possibly by VND7 (Zhong et al., 2007a, 2008; McCarthy et al., 2009). The upregulation of two LBD/ASL transcription family genes, LBD15/ASL11 and LBD30/ASL19/JLO, was also detected in the VND7-VP16-GR plants after DEX treatment. As LBD30/ASL19/JLO appears to be involved in a positive feedback loop with VND7 expression (Soyano et al., 2008), there might be a rather complex transcriptional network in control of xylem vessel differentiation.

Several IRX genes are directly regulated by VND7

In order to validate our microarray screen for direct targets of VND7, we performed quantitative RT-PCR (qRT-PCR) for several candidate genes (Figure 2). The mRNA levels of MYB46, XCP1 and LBD30/ASL19/JLO genes significantly rose in the VND7-VP16-GR plants in response to DEX treatment in the presence of CHX (11.33-, 7.65- and 7.26-fold, respectively).

Figure 2.

 Expression analysis of candidate genes for VND7 direct target.
Quantitative RT-PCR analysis was performed for several putative direct target genes of VND7 in the VP16-GR and VND7-VP16-GR plants. Seedlings were treated with cycloheximide (CHX) solution, with (+) or without (−) dexamethasone (DEX) for 4 h. mRNA levels for each gene were normalized to UBQ10 mRNA. Error bars represent SDs (n = 3). *More than threefold changes with statistically significant differences (Student’s t-test, P < 0.01) between the conditions with (+) and without (−) DEX.

Previously, we showed that constitutive overexpression of VND7 increases the expression of IRREGULAR XYLEM genes (IRX1, -3, -5, -6, -8, -9, -10, -12 and -13; Yamaguchi et al., 2010b). In the VND7-VP16-GR plants, four IRX genes (IRX1, -5, -12 and -13) are significantly upregulated by DEX treatment (Table 1). qRT-PCR analysis showed that the levels of IRX1/CesA8 and IRX5/CesA4 mRNAs in VND7-VP16-GR plants rose more than threefold in response to DEX treatment, even in the presence of CHX. In contrast, in the presence of DEX, the expression levels of IRX3/CesA7, IRX8/GALACTURONOSYLTRANSFERASE12 (GAUT12) and IRX10 were unchanged in the VND7-VP16-GR plants (Figure 2). These observations were consistent with the results of the microarray analysis (Table 1). Taken together with the previously reported expression data, we conclude that the expression of several IRX genes is directly regulated by VND7.

XCP1 is specifically expressed in differentiating xylem vessels

XCP1 is a cysteine protease that seems to be involved in programmed cell death during xylem vessel differentiation (Avci et al., 2008). In contrast to VND7, overexpression of NST1 or NST3, key regulators of fiber cell differentiation, does not induce XCP1 expression (Mitsuda et al., 2005; Zhong et al., 2006). Thus, XCP1 is a downstream gene specific to VND7. To further investigate, we checked the expression pattern of XCP1 after the induction of VND7. qRT-PCR revealed that XCP1 expression was elevated, starting 1 hour after treatment with DEX, and continued to increase during culture (Figure 3a). Chimeric reporter genes encoding YFP fused to the SV40 nuclear localization signal (NLS), or β-glucuronidase (GUS), under the control of the XCP1 promoter (Figure 3b) showed strong expression in differentiating xylem vessel cells in both roots and leaves (Figure 3c–f). This expression pattern is similar to that of VND7 (Yamaguchi et al., 2008). Moreover, when a vector construct harboring the VND7 gene driven by the 35S promoter was introduced into the transgenic plants expressing the GUS gene driven by the XCP1 promoter by particle bombardment, ectopic GUS staining was observed in various types of cells, such as epidermal cells and trichome cells, in addition to the differentiating xylem vessel cells (Figure S1). These data suggest that the promoter region of XCP1 includes sequences that respond to VND7.

Figure 3.

 Expression patterns of the XCP1 gene.
(a) Quantitative RT-PCR analysis of XCP1 expression in VP16-GR and VND7-VP16-GR plants after treatment with dexamethasone (DEX). Error bars indicate SDs (n = 3). mRNA levels were normalized against UBQ10 mRNA.
(b) Schematic diagram of the XCP1 gene for the construction of reporter genes. Black, grey and white boxes represent exons, introns and untranslated regions (UTRs), respectively. A genomic fragment corresponding to the region from −705 to +9 bp (indicated by the arrow) was used as the XCP1 promoter region.
(c–f) XCP1 expression levels in differentiating xylem vessel cells. The expression of reporter genes, the YFP-NSL (b and d) or GUS (c and e) gene under the control of the XCP1 promoter, is specifically detected at differentiating xylem vessel cells in roots (b, d) and leaves (c, e). Scale bars: 500 μm in (b, c) and 100 μm in (d, e).

Determination of the XCP1 promoter region responsible for its expression regulated by VND7

To define the promoter region responsible for the upregulation of gene expression by VND7, we carried out transient reporter assays using the XCP1 promoter sequence. We constructed reporter plasmids by linking various lengths of XCP1 promoter sequences to a minimal CaMV 35S promoter driving the firefly luciferase gene (Figure 4a); the VND7 gene driven by the CaMV 35S promoter was used as an effector plasmid. When the region from −705 to −79 bp of the XCP1 promoter was used as the reporter, strong luciferase activity was detected, as reported previously (Figure 4b; Yamaguchi et al., 2010b). A 5′ deletion series of XCP1 promoter (proB-F; Figure 4a) indicated that the luciferase activity obviously dropped in the proF construct (from −173 to −79 bp), suggesting that 98 bp of the XCP1 promoter fragment, corresponding to the region from −271 to −174 bp, includes the crucial region for VND7-inducible XCP1 expression. Sequential 5′ deletion analysis revealed that a 38-bp fragment of the XCP1 promoter (from −211 to −174 bp) is necessary for XCP1 induction by VND7 (compare results of proI and proJ; Figure 4c).

Figure 4.

 Deletion analysis of the XCP1 promoter.
(a) Schematic diagrams of the reporter constructs used for the transient reporter assay. The reporter constructs contain the firefly luciferase gene under the control of various lengths of XCP1 promoter fused to a minimal CaMV 35S promoter (min pro). Shaded regions (from −211 to −173 bp and from −136 to −96 bp) on the XCP1 promoter represent the regions responsible for XCP1 expression, which were identified in this study. The position of the tracheary element-regulating cis-element (TERE)-like motif is indicated as a dark-gray box, in the region between −122 and −112 bp, on the XCP1 promoter (see the Discussion). XCP1 promoter fragments with strong luciferase activity were denoted by ‘O’.
(b–e) Results of transient reporter assays. Both of the effector plasmids containing VND7 driven by the CaMV 35S promoter and the reporter constructs were introduced into Arabidopsis leaves by particle bombardment, and the luciferase activity was measured. Error bars indicate SDs (n = 3).

To investigate the possibility that a promoter region other than the 38-bp region described above (from −211 to −174 bp) contributes to the induction by VND7, we performed 3′ deletion of proE (from −271 to −79 bp) and proI (from −211 to −79 bp), to make proK (from −271 to −173 bp) and proL (from −211 to −173 bp), respectively. Interestingly, the expression levels of luciferase driven by proK and proL were lower than when driven by proF (Figure 4d). Thus, the proL fragment is not sufficient for the induction of gene expression by VND7. To determine an additional responsible region between −173 and −79 bp, we prepared a 3′ deletion series of the XCP1 promoter (proN–proQ; Figure 4a). A decrease of luciferase activity was observed in the cases of proN, proP and proQ; in particular, proP and proQ lacked almost all luciferase activity, as shown for proL (Figure 4e). These results show the importance of the upstream region starting at −96 bp for VND7-inducible gene expression.

We conclude that the region of the XCP1 promoter between residues −211 and −96 is necessary and sufficient for gene expression induced by VND7.

VND7 binds to the XCP1 promoter at two distinct locations

In order to investigate the direct DNA–protein interaction between the XCP1 promoter sequence and the VND7 protein, we carried out electrophoretic mobility shift assay (EMSA). For EMSA, we prepared the poly-His-tagged N-terminal region of VND7 that includes the whole NAC domain (His-VND71−161) and the 138-bp XCP1 promoter fragment (from −233 to −96 bp), as we expected that it gives full XCP1 expression level in response to VND7, based on the transient reporter assay (Figure 4). When this fragment was labeled with biotin and incubated with the recombinant His-VND71−161 protein, we observed two gel-shift bands (Figure 5a). Because VND7 is expected to form homo- and/or heterodimers with VND family proteins, including VND7 (Yamaguchi et al., 2008), these two bands may correspond to the VND7 homodimer-DNA complex (upper band) and the VND7 monomer-DNA complex (lower band) (Figure 5a). Competitive binding experiments were performed with unlabeled XCP1 promoter fragments (from −233 to −96 bp), showing dose-dependent inhibition effects on the band shift. Notably, the upper band disappeared more rapidly than the lower band (Figure 5a). These results suggested that sensitivity to the competitive probes is a function of the nature of the protein-DNA complex, specifically the number of VND7 molecules present.

Figure 5.

 VND7 protein binds to the XCP1 promoter.
(a) Electrophoretic mobility shift assay (EMSA) using recombinant His-tagged VND7 protein and the 138-bp XCP1 promoter fragment. Biotin-labeled XCP1 promoter fragments (from −233 to −96 bp; Biotin-XCP1p) were incubated with (+) or without (−) His-VND71−161 protein. For competitive binding experiments, a 40-, 100- or 200-fold excess of unlabeled XCP1 promoter fragments was added. Arrowheads indicate the shifted bands.
(b) VND7 binds to two distinct regions within the XCP1 promoter. Biotin-labeled XCP1 promoter fragments X1E1 (from −148 to −96 bp) and X1E2 (from −233 to −148 bp) were incubated with (+) or without (−) His-VND71−161 protein. For competitive binding experiments, a 200-fold excess of unlabeled fragments was added. Arrowheads indicate the shifted bands.

When the 138-bp XCP1 fragment was divided into two fragments, one of 53 bp (from −148 to −96 bp; X1E1) and the other of 85 bp (from −233 to −149 bp; X1E2), gel-shifted bands were also detected for both fragments (Figure 5b). Thus, the XCP1 promoter has at least two distinct binding sites for VND7. For both X1E1 and X1E2, no band was observed that corresponded to the VND7 homodimer-DNA complex present in the case of the 138-bp XCP1 promoter fragment (Figure 5b). These results led us to the additional conclusion that although VND7 can bind to two distinct regions of the XCP1 promoter, both regions are required in order to form the strict VND7 dimer–DNA complex.

VND7 binds to promoter regions of several candidate genes for direct targeting of VND7

We subsequently performed EMSA on the promoters of the other candidates for VND7 direct target genes: XCP2 (from −191 to −68 bp), IRX5/CesA4 (from −367 to −71 bp) and MYB83 (from −1000 to −498 bp). These promoter fragments contain tracheary element-regulating cis-element (TERE)-like sequences (Pyo et al., 2007). As a result of transient expression analysis, MYB83 has been proposed to be the direct target of NST3 and its close homologs (including VND7) (McCarthy et al., 2009). As the basal expression level of MYB83 was quite low, our microarray analysis excluded MYB83 from the list of putative direct targets of VND7 (Table 1); however, to check the proposed hypothesis about MYB83, we also carried out EMSA on the promoter of MYB83. As shown in Figure 6, shifted bands were observed in the promoters of all three genes tested. These signals were significantly decreased by the application of competitive unlabeled fragments, as in the case of the XCP1 promoter (Figures 5 and 6). Therefore, XCP2, IRX5/CesA4 and MYB83 should be considered direct targets of the VND7 protein.

Figure 6.

 VASCULAR-RELATED NAC-DOMAIN7 (VND7) protein binds to promoters of candidate direct target genes.
Biotin-labeled fragments of the promoter regions of XCP2, IRX5/CesA4 and MYB83 genes were incubated with (+) or without (−) His-VND71−161 protein. For competitive binding experiments, a 200-fold excess of unlabeled fragments was added. Arrowheads indicate the shifted bands.

VND7 regulates patterns of secondary cell wall deposition in vascular vessels

Ectopic secondary walls are observed when some NSTs and VNDs are ectopically expressed. In epidermal cells, these secondary walls showed similar striated patterns like vascular vessels (Kubo et al., 2005; Mitsuda et al., 2005; Zhong et al., 2006). From these findings, a question arose regarding whether both VND7 and NSTs are similarly able to regulate secondary wall deposition. To examine this question, we designed an experiment in which we expressed VND7 and NST3 under the control of the NST3 promoter in the nst1-1 nst3-1 double mutant, which has a pendant stem because of a lack of secondary walls in fiber cells (Figure 7a,b; Mitsuda et al., 2007). Because VND7 or NST3 expression is induced in fiber cells by the NST3 promoter, we can exclude other factors, e.g. tissue-specific regulators, and focus entirely on differences in transcriptional activity between VND7 and NST proteins. All chimeric genes suppressed the pendant stem phenotype of nst1-1 nst3-1, and most of the resulting plants could stand erect like the wild type. Cross sections of stem showed that secondary walls in fiber cells seemed to be restored by VND7 expression as well as by NST3 expression (Figure 7c,e; Mitsuda et al., 2007). Vertical sections revealed that NST3 induced uniform non-striated deposition of secondary walls, which are found in the fiber cells of wild-type plants (Figure 7d). However, VND7 induced striated deposition of secondary walls like those that appear in vessels, even in the fiber cells (Figure 7f). According to these observations, the patterns of secondary wall depositions are preferentially regulated by the nature of transcription factors. Furthermore, the striated deposition of secondary walls is regulated by VND7, not by NST proteins, in intact Arabidopsis plants.

Figure 7.

 Patterns of secondary cell wall formed by VND7 or NST3 in fiber cells.
Sections of the stem of the nst1-1 nst3-1 mutant (a, b) and the double mutant expressing either NST3pro:NST3 (c, d), or NST3pro:VND7 (e, f). (a, c, e) Cross sections (100-μm thick) of inflorescence stem viewed under UV illumination. (b, d, f) Vertical sections (8-μm thick) of inflorescence stem stained with Toluidine Blue O. Scale bars: 50 μm.

Discussion

VND7 directly regulates the expression of a number of genes involved in a wide range of processes in xylem vessel differentiation

Overexpression of VND7 can induce transdifferentation of cells into TE, not only in Arabidopsis, but also in Nicotiana tabacum (tobacco) culture cells and Populus tremula × tremuloides (poplar) (Kubo et al., 2005; Yamaguchi et al., 2010a). Thus, VND7 is sufficient to function as a key regulator for xylem vessel differentiation. In the present study, we identified more than 60 genes as putative direct targets of VND7 through microarray analysis of VND7-VP16-GR plants. Recombinant VND7 protein bound the promoter sequences of several candidate genes, such as XCP1, XCP2, CesA4/IRX5 and MYB83; thus, we believe that our list of putative direct targets of VND7 is reliable.

Our list of the direct targets of VND7 includes genes of several functional classes. Several well-known factors involved in xylem vessel differentiation are on the list, including the XCP1 and XCP2 genes. These genes encode cysteine proteases; the double mutation of xcp1 and xcp2 delays autolysis during xylogenesis in Arabidopsis roots (Avci et al., 2008). Our results showed that the expression of XCP1 is regulated by the VND7 protein acting at two distinct XCP1 promoter regions, resulting in specific expression in the vascular tissues of plants (Figures 2–5). In addition to XCP1 and XCP2, several genes encoding peptidase-like proteins were found among the candidates, suggesting that one of the important downstream processes directly regulated by VND7 is proteolysis related to programmed cell death. We also identified twelve genes functioning in cell wall biosynthesis as the direct targets of VND7 (Table 1). These genes, including CesA4/IRX5, CesA8/IRX1 and LAC4/IRX12 (Taylor et al., 2003; Brown et al., 2005), would contribute the secondary cell wall deposition, which is one of the essential steps for TE development. These findings imply that VND7 promotes xylem vessel formation by the direct control of cell wall biosynthesis as well as programmed cell death.

VND7 also indirectly regulates many genes to control xylem vessel formation

We also identified several transcription factors, MYBs (MYB46, MYB83 and MYB103) and LBD/ASL genes (LBD30/ASL19/JLO and LBD15/ASL11), as direct targets of VND7. Analysis of MYB46 and MYB83 have shown that these two MYB genes function in secondary cell wall deposition by activating biosynthetic pathways for cellulose, hemicellulose and lignin (Zhong et al., 2007a; Ko et al., 2009; McCarthy et al., 2009). The VND7 protein directly targets the IRX1/CesA8 and IRX5/CesA4 genes, encoding subunits of the cellulose synthase complexes, which are also the downstream genes of MYB46, MYB83 and MYB103 (Zhong et al., 2007a, 2008; Ko et al., 2009; McCarthy et al., 2009). Thus, the expression of IRX1/CesA8 and IRX5/CesA4 is cooperatively regulated by VND7 and its downstream transcription factors, the MYBs (Figure 8). Our results indicated that IRX3/CesA7, IRX8/GAUT12 and IRX10 are upregulated by VND7 overexpression, but are not direct targets of VND7 (Figure 2; Yamaguchi et al., 2010a,b). IRX3/CesA7 and IRX8/GAUT12 genes are induced by the overexpression of MYB genes (Zhong et al., 2007a; McCarthy et al., 2009), meaning that VND7 can increase the expression of these IRX genes through the upregulation of the MYB genes. This also suggests that VND7 can indirectly control a wide range of gene functions during xylem vessel differentiation.

Figure 8.

 Schematic diagram of the gene network interacting with VND7 during xylem vessel differentiation.
VND7 directly regulates many genes involved in various events during xylem vessel differentiation. For several genes, VND7 regulates their expressions through the MYB transcription factors. Red arrows represent the direct transcriptional regulation by VND7 revealed in this paper. (VNI2; Yamaguchi et al., 2010b; LBD30/ASL19; Soyano et al., 2008; MYBs; Ko et al., 2009; McCarthy et al., 2009; Zhong et al., 2007a, 2008).

Soyano et al. (2008) reported that VND6 and VND7, together with LBD30/ASL19/JLO and LBD18/ASL20, form a positive feedback loop during xylem vessel formation. LBD30/ASL19/JLO was also isolated as a direct target of VND7, supporting previous reports. In addition to these transcription factors, which are positive regulators of xylem vessel differentiation, a negative regulator of VND7 function, VNI2, was recently discovered (Yamaguchi et al., 2010b). VNI2 was isolated as an interacting factor of VND7: it represses the transcriptional activation activity of VND7 by forming a protein complex. Taken together, these data indicate that there is a complex network of transcription factors acting during xylem vessel differentiation (Figure 8).

VND7 binding to the XCP1 promoter, probably through the TERE motif

We showed that a 116-bp region (from −211 to −96 bp) of the XCP1 promoter is important for its regulation by VND7 (Figure 4), and that VND7 binds to at least two distinct sites in the region (Figure 5). A previous study reported a cis-element related to the differentiating xylem vessel cell-specific promoter activity: the TERE (Pyo et al., 2007). TERE-like sequences are found in promoters of many Arabidopsis TE differentiation-related genes, and are apparently over-represented among direct target genes of VND7 (15 out of 63; Table 1). In the XCP1 promoter, there is one TERE-like sequence in the region between −122 and −112 bp (Figure 4a). Transient expression assays showed that the deletion of the TERE motif resulted in a decrease in luciferase activity, either partial (proO) or complete (proP and proQ) (Figure 4e), suggesting that the TERE motif is important for XCP1 expression regulated by VND7. On the other hand, the deletion analysis and EMSA analysis indicated that an additional sequence in the XCP1 promoter contributed to VND7 binding: this element is located at the region between −211 and −174 bp (Figure 4). We speculate that under the control of VND7, the TERE motif at the region between −122 and −112 bp is required for basal XCP1 promoter activity, whereas the region between −211 and −174 bp participates in boosting the expression level.

Diversity of the downstream events between VND7 and NST proteins

As described above, VND7 directly binds to the promoters of XCP1 and XCP2, which are involved in programmed cell death during xylem vessel differentiation. In addition to the genes related to cell death, we found a significant number of genes involved in the modification of pectin polysaccharide among the putative direct targets of VND7 (Table 1). In contrast to VND7, the NST proteins have not been reported to upregulate the expression of these genes when they are ectopically overexpressed (Mitsuda et al., 2005; Zhong et al., 2006). These facts suggest that there are differences between VND7 and NST proteins, with respect to both the identity of target genes and the diversity of the downstream events.

On this point, valuable information was obtained from the experiments in which VND7 or NST3 was expressed under the control of the NST3 promoter in the nst1-1 nst3-1 double mutant. Although both VND7 and NST3 restored stem strength through secondary cell wall formation in the fiber cells, the structures of the secondary cell walls were clearly different (Figure 7). Thus, downstream genes of VND7 could be partially, but not completely, overlapped with those of NST proteins. It is supposed that these can be attributed to the differences in the cis-elements and/or the binding activities of the target gene promoters.

Results of recent works have proposed transcriptional gene networks for xylem vessel differentiation (Caño-Delgado et al., 2010; Yamaguchi and Demura, 2010). Moreover, the presence of a negative regulator, VNI2, which functions at the post-translational level, provided us with insight into more complex regulation around VNDs, not only at the transcriptional level, but also at the post-translational level (Figure 8; Yamaguchi et al., 2010b). Further investigation of interactions among the VND/NST proteins, VNI2 and the promoter sequences of target genes will help elucidate the regulation of gene expression in xylem vessel formation. In addition, in our list of the direct targets of VND7, the molecular functions of more than half of the direct target genes are uncertain. To obtain a view of xylem vessel differentiation, we also need to reveal how these genes function during xylem vessel formation.

Experimental procedures

Plasmid constructions

The promoter fragment of XCP1 or truncated cDNA of VND7 were subcloned into pENTR/D-TOPO (Invitrogen, http://www.invitrogen.com) using the primers described in Table S4. For construction of reporter genes for XCP1 expression, the promoter fragments of XCP1 were integrated into pBGYN and pBGGUS (Kubo et al., 2005); for recombinant VND7 expression, the truncated cDNA fragment of VND7 was integrated into pDEST17 (Invitrogen) using LR clonase (Invitrogen). For the deletion series of the XCP1 promoter, the XCP1 promoter fragments were inserted into the HindIII-blunted/XbaI sites of the GAL4UAS:TATA:LUC reporter plasmid (Yamaguchi et al., 2010b). Production of NST3pro:VND7 constructs followed methods described previously for the NST3pro:NST3 construct (Mitsuda et al., 2007).

Plant materials and growth conditions

The VND7-VP16-GR and VP16-GR plants (Col-0 transgenic plants of Arabidopsis thaliana) were described in Yamaguchi et al. (2010a). The nst1 nst3 double mutant was described in Mitsuda et al. (2007, 2008). The plasmids constructed for the YFP-NSL or GUS genes under the control of the XCP1 promoter, or the NST3pro:VND7/NST3 genes, were transformed into wild-type A. thaliana Col-0 plants by a protocol described previously (Yamaguchi et al., 2010a,b). The transgenic seedlings were selected on germination medium (Kubo et al., 2005) supplemented with 10 μg ml−1 bialaphos, and then planted in Jiffy 7 pots (Sakata Seed, http://www.sakata.com) under a 16-h light/8-h dark cycle. For complementation tests, the chimeric NST3pro:NST3/VND7 genes were transformed into the nst1 nst3 double mutant.

Expression analyses using the VND7-VP16-GR plants

Ten-day-old seedlings of VND7-VP16-GR and VP16-GR grown on a germination medium (Kubo et al., 2005) were soaked with sterilized water containing 10 μm CHX, a protein synthesis inhibitor, for 2 h to prevent the production of secondary transcription factors that could then influence gene transcription. After removal of the solution, the seedlings were re-soaked with sterilized water containing 10 μm CHX, with or without 10 μm DEX, for 4 h. Total RNA was extracted using the Nucleospin RNA Plant System (Macherey-Nagel, http://www.mn-net.com). Microarray analysis was performed using GeneChip ATH1 Arabidopsis genome arrays (Affymetrix, http://www.affymetrix.com/index.affx), with five independent biological replicates. For qRT-PCR analysis, cDNA was synthesized from the total RNA samples by treatment with reverse transcriptase (Transcriptor First Strand cDNA Synthesis Kit; Roche, http://www.roche.com) and oligo(dT) primer (Invitrogen). qRT-PCR was performed in the LightCycler 480 system II (Roche) using the LightCycler 480 SYBR green I master reagent (Roche). Primer sets used for the expression analysis are described in Table S4.

Microarray data analysis

Signal values of the probe set were calculated using microarraysuite (mas) v5.0, supplied by Affymetrix. All signal values were divided by the median value among all non-control probe sets to enable comparisons with other microarray data. Probe set-to-gene conversion was accomplished based on a table provided by TAIR (ftp://ftp.arabidopsis.org/home/tair/Microarrays/Affymetrix/affy_ATH1_array_elements-2009-7-29.txt). The average values were used for genes corresponding to two or more probe sets. We defined a detection value for each probe set as 0, 1 or 2, when the detection call by mas v5.0 was A, M or P, respectively. Only genes with an average detection value ≥1 in the ‘test’ sample or ‘control’ sample were analyzed for the purpose of selecting upregulated or downregulated genes, respectively. P and Q values for estimating the FDR were calculated as described previously (Yamaguchi et al., 2010b). Microarray data presented in this study was submitted to NCBI GEO (http://www.ncbi.nlm.nih.gov/geo), and can be retrieved via accession number GSE24169.

Microscopy analysis

The reporter lines for XCP1 were observed according to the methods described in Kubo et al. (2005). Cross sections of stems of NST3pro:NST3/VND7 plants were produced and observed as described previously (Mitsuda et al., 2007). For vertical sections, stem segments were embedded in paraffin (Paraplast plus; TYCO healthcare, Mansfield, MA, USA) after fixation in FAA solution (45% ethanol, 2.5% acetic acid and 2.5% formalin). Sections (8-mm thick) were prepared using a microtome, and stained with Toluidine Blue solution (0.5% Toluidine Blue O and 0.1% Na2CO3) after the removal of paraffin with xylene.

Transient reporter assay

The effector, reporter and reference plasmids, referenced in Yamaguchi et al. (2010b), were delivered into the rosette leaves of 4-week-old Arabidopsis plants by particle bombardment (GE Healthcare, http://www.gehealthcare.com). After overnight incubation, the leaf samples were pulverized in liquid nitrogen, and luciferase activity was assayed using the Dual-Luciferase Reporter Assay System (Promega, http://www.promega.com) using a Mithras LB940 reader (Berthold Technologies, http://www.berthold.com).

Electrophoretic mobility shift assay

Promoter fragments were labeled with biotin using the Biotin 3′ End DNA Labeling Kit (Pierce, http://www.piercenet.com). Labeled probes were separated from unincorporated biotin-dUTP using Micro Bio-Spin6 chromatography columns (Bio-Rad, http://www.bio-rad.com). The poly-His-tagged N-terminal region of VND7 (His-VND71−161) was expressed in Escherichia coli using the plasmid described in ‘Plasmid construction’, above, and purified as previously described in Yamaguchi et al. (2010b). Approximately 20 fmol of biotin-labeled promoter fragments were incubated in reaction buffer (LightShift Chemiluminescent EMSA Kit; Pierce) for 30 min at 4°C with 2 pmol of the purified recombinant His-VND71−161 protein and/or an excess of unlabeled fragments as a competitor. The VND7-bound DNA fragments were separated by polyacrylamide gel electrophoresis. The DNA was electroblotted onto nitrocellulose membrane (Biodyne Plus; Pall, http://www.pall.com) and detected using the LightShift Chemiluminescent EMSA Kit according to the manufacturer’s instructions.

Acknowledgements

We thank Dr Nam-Hai Chua (The Rockefeller University, USA) and Dr Takashi Aoyama (Kyoto University, Japan) for providing us with the VP16-GR vector. We also thank Hiromi Ogawa, Ayumi Ihara, Mitsutaka Araki, Sachiko Oyama, Tomoko Matsumoto, Ryoko Hiroyama, Toshie Kita, Kayo Kitaura, Akiko Sato, Minami Shimizu (RIKEN, Japan) and Nobuko Shizawa (NAIST, Japan) for excellent technical assistance. This work was supported by a Grant-in-Aid for Scientific Research (grant no. 20770041, MY; no. 21770064 to NM; and nos 21027031 and 22370020 to TD) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Note added in proof

While this manuscript was under review, direct targets of NST3 (also called SECONDARY WALL-ASSOCIATED NAC DOMAIN PROTEIN1 or ARABIDOPSIS NAC DOMAIN CONTAINING PROTEIN012) and VND7 (Zhong et al., 2010), and those of VND6 and NST3 (Ohashi-Ito et al., 2010), were described independently.

Ohashi-Ito, K., Oda, Y. and Fukuda, H. (2010) Arabidopsis VASCULAR-RELATED NAC-DOMAIN6 directly regulates the genes that govern programmed cell death and secondary wall formation during xylem differentiation. Plant Cell, 22, 3461–3473.

Zhong, R., Lee, C. and Ye, Z.-H. (2010) Global analysis of direct targets of secondary wall NAC master switches in Arabidopsis. Mol. Plant, 3, 1087–1103.

Ancillary