Skip to main content
NCBI home page
Plant Physiology logoLink to Plant Physiology
. 2016 Aug 26;172(2):1334–1351. doi: 10.1104/pp.16.01100

A Transcriptional and Metabolic Framework for Secondary Wall Formation in Arabidopsis1

Zheng Li 1,2,3,4,2, Nooshin Omranian 1,2,3,4,2, Lutz Neumetzler 1,2,3,4,2, Ting Wang 1,2,3,4, Thomas Herter 1,2,3,4, Bjoern Usadel 1,2,3,4, Taku Demura 1,2,3,4, Patrick Giavalisco 1,2,3,4, Zoran Nikoloski 1,2,3,4, Staffan Persson 1,2,3,4,*
PMCID: PMC5047112  PMID: 27566165

Network analyses reveal metabolic pathways associated with secondary cell wall synthesis and provide a framework for a better understanding of plant biomass formation.

Abstract

Plant cell walls are essential for plant growth and development. The cell walls are traditionally divided into primary walls, which surround growing cells, and secondary walls, which provide structural support to certain cell types and promote their functions. While much information is available about the enzymes and components that contribute to the production of these two types of walls, much less is known about the transition from primary to secondary wall synthesis. To address this question, we made use of a transcription factor system in Arabidopsis (Arabidopsis thaliana) in which an overexpressed master secondary wall-inducing transcription factor, VASCULAR-RELATED NAC DOMAIN PROTEIN7, can be redirected into the nucleus by the addition of dexamethasone. We established the time frame during which primary wall synthesis changed into secondary wall production in dexamethasone-treated seedlings and measured transcript and metabolite abundance at eight time points after induction. Using cluster- and network-based analyses, we integrated the data sets to explore coordination between transcripts, metabolites, and the combination of the two across the time points. We provide the raw data as well as a range of network-based analyses. These data reveal links between hormone signaling and metabolic processes during the formation of secondary walls and provide a framework toward a deeper understanding of how primary walls transition into secondary walls.

Plant cells are encased by polysaccharide-rich cell walls. These walls provide a number of essential functions for plant growth and development (Carpita and McCann, 2000). For example, the cell walls define the shape of plant cells, thereby determining tissue and organ morphology; they provide protection against external stresses and aid in water transport throughout the plant (McFarlane et al., 2014). In addition, the components of the cell walls provide great raw material for a range of industrial applications, including textile, lumber, feed, food, and fuel.

Plants typically contain two different types of cell walls: a primary wall that is deposited around all growing cells, and a secondary wall that is produced in cells with specialized functions once they have ceased to grow. Prominent secondary wall-producing cell types are tracheary elements, fibers, and other sclerenchymatous cells that supply plants with water-transporting capacity and tissue strength (Zhong and Ye, 2015). Controlled deposition of secondary walls results in elaborate helical or reticulated wall patterns during tracheary element and fiber development (Zhong and Ye, 2015). These patterns provide a basis for the cells to withstand mechanical stress and support the xylem conduits.

While the primary walls typically consist of cellulose, hemicelluloses, and pectins, the secondary walls contain cellulose, hemicelluloses, and the polyphenolic polymer lignin (Turner and Somerville, 1997). Cellulose is the main contributor to cell wall mechanical strength and is synthesized by plasma membrane-based cellulose synthase (CesA) complexes fueled by a cytosolic UDP-Glc pool (Endler and Persson, 2011). In contrast, hemicelluloses and pectins are produced in the Golgi apparatus by a collection of glycosyltransferases and by the activity of nucleotide sugar-converting enzymes and transporters (Scheller and Ulvskov, 2010; Atmodjo et al., 2013). Precursors for the secondary wall-specific polymer lignin are made in the cytoplasm and then transported to the apoplast, presumably via monomeric lignin transporters (Boerjan et al., 2003; Bonawitz and Chapple, 2010).

Patterned secondary cell wall (SCW) deposition is the dominant feature of xylem vessel differentiation. The initial phase of xylem vessel development includes cell elongation that is underpinned by the action of multiple hormones (Li et al., 2016). Once cell elongation has ceased, SCW synthesis begins. Mature xylem vessels are referred to as functional corpses, as enzyme activity (e.g. lignin polymerization) is maintained after programmed cell death (PCD) has proceeded and SCW polysaccharides have been deposited. It is worthwhile noting that SCW formation and PCD are two processes that are difficult to untangle, in particular since SCW lignification occurs mainly after or during the death of the cell (Derbyshire et al., 2015).

Historically, zinnia (Zinnia elegans) culture cells have been used as an excellent experimental model system in which to study xylem vessel development (McCann, 1997). Transdifferentiation was induced in this model system using a hormone cocktail of cytokinins and auxins (Demura et al., 2002). Using such an inducible zinnia system, Demura et al. (2002) performed transcriptional profiling at seven time points during transdifferentiation. A similar approach was undertaken subsequently in Arabidopsis (Arabidopsis thaliana) suspension cultures using brassinolide and boric acid as transdifferentiating inducing agents (Kubo et al., 2005). That study revealed many genes whose expression was altered during secondary wall formation, including the CesA-interacting factors TED6 and TED7 and several NAC transcription factors (TFs; Demura et al., 2002; Kubo et al., 2005; Endo et al., 2009). A clade of these NAC TFs were referred to as VASCULAR-RELATED NAC DOMAIN PROTEIN1 (VND1) to VND7, and overexpression of two of the genes, VND6 and VND7, using 35S promoters, caused ectopic transdifferentiation (Kubo et al., 2005). Because VND6 and VND7 could independently induce metaxylem and protoxylem vessel formation, respectively, in cells that normally do not undergo transdifferentiation, the TFs were referred to as master regulators for vessel cell fate (Kubo et al., 2005).

The finding of the VNDs paved the road for the discovery of many other TFs, including MYB and NAC, that work in parallel or downstream of the VNDs, including SND2, SND3, MYB20, MYB42, MYB43, MYB46, MYB52, MYB54, MYB58, MYB63, MYB69, MYB83, MYB85, MYB103, and KNOTTED ARABIDOPSIS THALIANA7 (KNAT7; Zhong et al., 2008). These findings, together with a recent study that outlines a comprehensive transcriptional network for how secondary walls are regulated (Taylor-Teeples et al., 2015), have provided important information toward our understanding of how plants make xylem vessels. Interestingly, some of the TFs can specifically regulate certain biosynthetic pathways. For example, MYB58, MYB63, and MYB85 are specific activators for the lignin biosynthetic pathway (Zhou et al., 2009; Zhong and Ye, 2012; Cassan-Wang et al., 2013; Hirano et al., 2013).

In addition to the 35S-driven constructs, an inducible system of VND7 was established (Yamaguchi et al., 2010). In this system, VND7 is expressed as a fused protein with the transcriptional activation domain of the herpes virus VP16 protein and a glucocorticoid receptor domain, which promotes the protein’s redirection into the nucleus when induced by treatment with dexamethasone (DEX). Consequently, DEX induction led to the formation of protoxylem in all induced cells and an activation of the complete set of secondary wall genes (Yamaguchi et al., 2010). Therefore, this unique system is well suited to perform systems-related studies to dissect the sequence of events during xylem vessel differentiation in Arabidopsis.

Here, we took advantage of the DEX-inducible VND7 line and used Arabidopsis etiolated seedlings to induce SCW formation. We subsequently analyzed transcript and metabolic levels at eight time points after induction. We thus present a comprehensive map of transcriptional and metabolomic changes upon the induction of secondary wall formation. These data may form a foundation for future studies to understand the complex transcriptional and metabolic programs that underpin secondary wall synthesis.

RESULTS AND DISCUSSION

Experimental Design and Phenotypic Induction

To establish the time frame for the secondary wall induction, we grew seedlings of the inducible VND7 line for 7 d in liquid Murashige and Skoog (MS) medium and then supplemented the medium with DEX in the dark. The DEX treatment effectively induced secondary wall formation, as seen after 48 h of DEX treatment in light-grown seedlings (Supplemental Fig. S1). We also imaged Pontamine Fast Scarlet S4B (Anderson et al., 2012)-stained dark-grown seedlings using differential interference contrast microscopy to visualize the induction of helical bands surrounding epidermal cells that we used as a proxy for the production of secondary walls (Supplemental Fig. S2; Yamaguchi et al., 2011). From these experiments, we concluded that secondary walls had been established, and cell death was evident, 48 h after DEX induction (HAI). Hence, we chose nine time points within these 48 h (0, 1, 3, 6, 9, 12, 24, 30, and 48 h) to carry out transcript and metabolic profiling analyses (Fig. 1A).

Figure 1.

Figure 1.

Open in a new tab

Experimental design. A, Seeds of Arabidopsis (Columbia-0) plant lines harboring the DEX-inducible construct VND7-VP16-GR were added to liquid MS medium and grown for 6 d prior to DEX induction. After induction, cultures were harvested for microarray and metabolic profiling at the indicated time points (1, 3, 6, 9, 12, 24, 30, and 48 h) and at 0 h. Uninduced cultures (dimethyl sulfoxide [DMSO] instead of DEX) served as controls for each time point. B, Seeds of Arabidopsis plant lines (Columbia-0) harboring the empty vector control VP16-GR were used in a similar experimental setup (three time points for microarray and two time points for metabolic profiling) to ensure that DEX did not influence the measurements. *Only primary and lipid metabolites were measured for empty vector plants.

Total RNA was isolated from seedlings of either treated or control plants at each time point and used for transcriptomic analyses (see “Materials and Methods”). Three biological replicates were analyzed using the Affymetrix ATH1 Arabidopsis genome array, as described by Misson et al. (2005). We selected a threshold of 2 for log-fold expression differences and a P value of 0.05 or less after false discovery rate (FDR) correction. The full data set has been deposited in the Gene Expression Omnibus under accession number GSE77153 (for GC-robust multiarray-normalized data, see Supplemental Table S1).

Plant material at each time point was divided into aliquots for metabolic analyses using three to six biological replicates (see “Materials and Methods”). After applying the methyl-tert-butyl-ether method (Giavalisco et al., 2011) to extract the metabolites, two independent platforms were utilized to conduct the metabolic profile analysis: ultra-performance liquid chromatography-Fourier transform-mass spectrometry (UPLC-FT-MS) and tandem mass spectrometry (MS/MS) for organic phase (lipids) and the secondary metabolites from the polar phase, and gas chromatography-mass spectrometry for primary metabolites. This enabled us to detect, identify, and quantify 251 metabolites (Supplemental Table S2). Among them, 247 metabolites showed significant differences (P ≤ 0.05; FDR corrected) across the induction period and were included in the cluster and network analyses.

To ensure that the seedlings carrying the empty vector construct (VP16-GR) did not show differences associated with secondary wall induction, we analyzed material of VP16-GR plants (with and without DEX treatment) using both transcriptomic analyses and metabolic profiling (Fig. 1B). We first identified differentially expressed genes (DEGs) at 9, 12, and 24 HAI between conditions (i.e. with and without DEX treatment) using the R package limma (Smyth, 2005), with fold change = 1.5 and a significance level of 0.05 (FDR corrected). This analysis was performed separately on each expression data set from seedlings carrying the empty vector (VP16-GR) and seedlings of the inducible VND7 line. Then, for each time point, the ratio between the number of DEGs in both empty vector and VND7 data sets (ratio of shared difference [RSD]) and the number of DEGs under either the empty vector or VND7 data set was determined. These RSDs were 0.0047, 0.005, and 0.0045 for the time points 9, 12, and 24 HAI, respectively. This implies that about 0.5% of the DEGs are shared between the empty vector and VND7 data sets. Furthermore, profiling of primary and lipid metabolites at 9 and 24 HAI was performed to rule out the possibility that DEX might contribute to the metabolic changes we observed in the VND7-induced plants. Similarly, we first identified differentially behaving metabolites at each HAI between conditions (i.e. with and without DEX treatment) using the R package limma, with a significance level of 0.05. Then, the same analysis was applied to the VND7 metabolic data set, and the RSDs were determined. Only one metabolite was commonly altered in the two genotypes at each HAI (urea at 9 HAI and Gln at 24 HAI). The RSDs, therefore, were approximately 0 at both HAI for metabolite changes. Hence, the effect of DEX and/or the empty vector on transcript and metabolite changes was negligible during the VND7-induced secondary wall synthesis (Supplemental Tables S1 and S2).

Data Overview

The transcriptional analysis revealed that over 2,000 transcripts were significantly induced or repressed within the first HAI of VND7 compared with the control (Supplemental Table S3). In contrast, among the 253 metabolites we tested, only 13 were significantly changed at the same time point (Supplemental Table S4). However, during the progression of secondary wall synthesis (i.e. at later time points: 12–30 HAI), we observed an increasing number of metabolites whose levels changed. In contrast, most transcripts that changed during the early time points also maintained similar levels during subsequent time points. In addition, most metabolite and transcript levels declined during the last two time points (30 and 48 HAI; Fig. 2, A–D; Supplemental Tables S3 and S4). These data were anticipated, as the progression of secondary wall synthesis during vessel development involves PCD, which should lead to overall reductions in transcript and metabolic contents.

Figure 2.

Figure 2.

Open in a new tab

Overview of metabolite and transcript changes in VND7-induced seedlings. A and B, Venn diagrams (venneuler function in the R package venneuler; Wilkinson, 2012) of differentially behaving transcripts and metabolites between consecutive time points (compared with the corresponding control), respectively. Different colored circles correspond to the particular consecutive time points as indicated in the key. All analyses were based on absolute log-fold changes greater than 2 and FDR-corrected P < 0.05. C and D, Double-sided histograms of the number of differentially behaving transcripts and metabolites between consecutive time points, discerning up-/down-regulation with red/green color. E and F, Profiles of the selected transcripts and metabolites with altered levels during the VND7 induction period. CesA4, IRREGULAR XYLEM9 (IRX9), KNAT7, Phe (a precursor for lignin), and UDP-GlcA (a substrate for cell wall polymer synthesis) were selected due to their roles in secondary wall formation. The letters c and t represent control and treatment, respectively (E). Green and red lines correspond to control and treatment, respectively (F).

To ensure that transcripts and metabolites that typically are associated with secondary wall formation were induced, we monitored the levels of several genes (e.g. CesA4, IRX9, and KNAT7) and metabolites (lignin precursor and nucleotide sugars) for which such associations have been shown (Zhong and Ye, 2015). We observed that secondary wall-related genes were rapidly induced after DEX treatment (Fig. 2E; Supplemental Table S3; Supplemental Fig. S4). Metabolites expected to be associated with secondary wall synthesis, however, were generally increased at later time points (Fig. 2F; Supplemental Table S2), indicating that the proteins necessary for SCW formation need to be made before the metabolic scheme is changed. These data also confirmed that the DEX-treated seedlings, in terms of transcript changes, very rapidly changed their fate to produce SCWs.

Transcript and Metabolite Changes during Secondary Wall Formation

To get a more detailed overview of the processes that were affected in response to the VND7 induction, we analyzed Gene Ontology (GO; Biological Process [BP] and Molecular Function [MF]) terms associated with the transcripts and metabolites that were significantly changed during the induction. Not surprisingly, the ontology terms changed dramatically during the induction course, with the most notable changes occurring in the general ontology bins associated with phenylpropanoid and hormone metabolic processes as well as some transferase activities (Fig. 3). However, we also observed a clear change in the cell wall degradation-related term hydrolase activity, perhaps representing an active transition from primary cell wall to SCWs (Ohdaira et al., 2002). Full lists of significantly changed BP and MF terms can be found in Supplemental Tables S5 and S6.

Figure 3.

Figure 3.

Open in a new tab

GO terms associated with changing transcripts during VND7 induction. Ten highly enriched and SCW-related metabolic process terms (A) and molecular function terms (B) of DEGs (compared with time point 0 h) during the VND7 induction are shown. The y axis indicates the number of DEGs, and the x axis displays specific time points.

To further assess how transcripts and metabolites, which were associated with different onotology terms, changed over time, we used the tool PageMan using MapMan ontology (Usadel et al., 2009) to visualize the levels of gene expression and metabolites after the VND7 induction (Fig. 4; Supplemental Fig. S3). Photosynthesis was identified as one of the most prominent down-regulated processes in zinnia cell cultures after secondary wall induction (Demura et al., 2002). Photosynthesis does not occur in etiolated seedlings; however, seedlings grown in the dark are still primed to develop photosynthesis capacity once they are exposed to light (Thompson, 1988), and they do contain etioplast and/or etiochloroplast (Gruissem, 1989). Nevertheless, we found that photosynthesis-related genes were dramatically down-regulated after VND7 induction (Fig. 4; Supplemental Fig. S5). For example, the expression of genes encoding the PSII light reaction center subunit and the PSI reaction center subunit XI/PSI subunit V dropped very quickly at the first HAI (Supplemental Fig. S5). In terms of metabolites, most monogalactosyldiacylglycerols and digalactosyldiacylglycerols showed a substantial decrease during the mid to late induction stages. These lipids are major glycerolipid constituents of photosynthetic membranes in seed plant chloroplasts (Boudière et al., 2014). In addition, genes associated with starch synthesis were negatively regulated, supporting a coordinated transcriptional response of photosynthesis and starch metabolism. It is interesting that similar responses have been shown during pathogen-related responses (Berger et al., 2007). These similarities are most likely due to similar fates of pathogen-infected and secondary wall-induced cells (e.g. they are both undergoing wound responses and cell death; Cheong et al., 2002; Coll et al., 2011; Bollhöner et al., 2012).

Figure 4.

Figure 4.

Open in a new tab

Condensed PageMan display of the alteration in transcripts associated with selected pathways. Changes in transcript levels are presented as log2-fold changes in comparison with the control (DMSO-treated VND7-VP16-GR plants). The analyzed time points were 1, 3, 6, 9, 12, 24, 30, and 48 HAI. The data were subjected to a Wilcoxon test in PageMan, and the results are displayed in false-color code. Bins colored in red are significantly up-regulated, whereas bins colored in blue are significantly down-regulated. CHO, Carbohydrate; FA, fatty acid; PS, photosynthesis. Other pathways are shown in Supplemental Figure S3.

Primary metabolism is well studied in Arabidopsis, in particular via biochemical and genetic studies (Sweetlove et al., 2008). However, our understanding of how primary metabolism contributes to SCW formation is scant, although primary metabolism has been associated with biomass production in the past (Meyer et al., 2007; Sulpice et al., 2009). For example, constitutive down-regulation of genes associated with the tricarboxylic acid cycle in transgenic tomato (Solanum lycopersicum) lines had significant impact on the development of the vasculature, possibly involving the last steps of the tricarboxylic acid cycle (van der Merwe et al., 2010). We observed clear changes in several of the tricarboxylic acid cycle-related genes, and obvious reductions in ketoglutarate, succinate, fumarate, and malate happened after 12 HAI (Fig. 5). Since these tricarboxylic acid cycle metabolites also are associated with amino acid metabolism (Fig. 5), we further looked at changes in amino acids (Supplemental Fig. S6). Here, it is apparent that amino acids stemming from common precursors behave similarly. For example, Lys, Thr, Met, and Ile, which are made from oxaloacetate, showed very similar profiles with an accumulation at 24 HAI. By contrast, Glu, Gln, Pro, and Arg, which are synthesized from ketoglutarate, did not show any substantial changes during the later HAI. In summary, our data indicate that the Asp family pathway (Lys, Met, Ile, and Thr), shikimate pathway (Trp, Phe, Try, and His), and branched-chain amino acid pathway (Ile, Leu, and Val) are main contributors to SCW synthesis. Notably, Phe is the precursor of lignins, and the increase of this amino acid at 24 HAI might signify the onset of, or increase in, lignin biosynthesis. It might be instructive to assess the cross-regulation of amino acid pathways during SCW synthesis using inhibitors of key enzymes in these pathways (e.g. acetohydroxy acid synthase via sulfonylurea and imidazolinones; Manabe et al., 2007), or by the use of genetic means, in combination with the VND7 system. Such experiments may be helpful to delineate the importance of the tricarboxylic acid cycle and amino acid biosynthesis pathways in SCW synthesis.

Figure 5.

Figure 5.

Open in a new tab

Transcript and metabolite changes in the tricarboxylic acid cycle. The y axis indicates relative metabolite levels, and the x axis displays time points after treatment. Error bars represent sd. Changes in transcript levels of related enzyme genes are presented as log2-fold changes in comparison with the control (DMSO-treated VND7-VP16-GR plants) in colored boxes. Bins are colored in false-color code.

As SCW production would need a considerable increase in nucleotide sugars (Seifert, 2004), we anticipated that VND7 induction would cause changes in the pool sizes of these metabolites. As shown in Figure 6, most of the nucleotide sugars were increased already by the induction of VND7 around 3 HAI. To produce diverse nucleotide sugars, plants possess a large number of nucleotide sugar interconversion enzyme families, which can control the ratio of nucleotide sugar precursors for the different polysaccharides. Consistent with the nucleotide sugar data, most genes encoding the nucleotide sugar intermediate pool enzymes were up-regulated at early HAI (Fig. 6). Two mechanisms that can regulate nucleotide sugar fluxes have been proposed: feedback inhibition and redox sensing. The control via feedback inhibition relies on transcriptional control (Seifert, 2004), where different steps in the nucleotide sugar interconversion pathway can be transcriptionally repressed. Indeed, some of the genes encoding such enzymes were down-regulated over time (Fig. 6). Notably, some genes that encode different isozymes showed very different behavior. For example, PHOSPHOMANNOSE ISOMERASE1 (PMI1) was down-regulated already at the earliest time points, whereas PMI2 was highly induced. Maruta et al. (2008) reported that PMI1, but not PMI2, is essential for ascorbic acid biosynthesis in Arabidopsis. Therefore, the changes in transcripts might reflect a fine-tuning mechanism to relocate resources into different pathways. To support cell wall biosynthesis, nucleotide sugars need to be transported into the Golgi lumen, which is facilitated by nucleotide sugar transporters. We observed that genes encoding nucleotide sugar transporter protein also were generally up-regulated after VND7 induction (Fig. 6). However, the expression of the recently identified bifunctional UDP-Rha/UDP-Gal transporters (URGTs) deviated from this behavior and generally declined from 12 HAI onward (Fig. 6; Rautengarten et al., 2014). URGT1 transports UDP-Gal that is used mainly for pectic galactan (Rautengarten et al., 2014). It would be anticipated that the production of pectic polymers would decrease during SCW synthesis. While UDP-Gal still might be made and present in the cytosol, it would be expected that the transport of it into the Golgi lumen is decreased, as many of the URGTs are down-regulated. Taken together, we conclude that, during the development of vessels, plants boost the nucleotide sugar supply and Golgi content, presumably to provide enough building blocks for the cell wall synthesis that occurs during secondary wall formation.

Figure 6.

Figure 6.

Open in a new tab

Transcript and metabolite changes in the nucleotide sugar interconversion pathway. The y axis indicates the relative metabolite levels, and the x axis displays time points after treatment. Error bars represent sd. Changes in transcript levels encoding enzymes/nucleotide sugar transporters are presented as log2-fold changes in comparison with the control (DMSO-treated VND7-VP16-GR plants) as colored boxes. Bins are colored in false-color code.

We also anticipated that the lipid levels and contents would change during the VND7 induction, given the importance of lipids to membrane functionality and vesicle shuttling. From the microarray analysis, genes associated with fatty acid synthesis were largely down-regulated (Fig. 4; Supplemental Table S1). In addition, lipid degradation-related genes were largely up-regulated (Fig. 4; Supplemental Table S1). This suggests that the lipids are used to fuel SCW production, as it is an energy-consuming process. Furthermore, phosphatidylethanolamines and phosphatidycholines generally decreased, whereas diacylglycerol generally increased (Supplemental Table S2). As diacylglycerol is the substrate for the biosynthesis of phosphatidycholines and phosphatidylethanolamines via the CDP-choline and CDP-ethanolamine pathways, respectively, it is anticipated that these precursors build up if the downstream steps are blocked or reduced. In addition, many triacylglycerides (TAGs) increased at 24 to 48 HAI. These data are consistent with a link between TAG accumulation and cell death, as TAG accumulation was reported in senescing leaves (Kaup et al., 2002).

Tracheary elements need to be impregnated by the hydrophobic polymer lignin and to be connected transversely through pits by membrane degradation that is developmentally controlled by PCD. Several key genes and metabolites that are associated with PCD have been identified, such as reactive oxygen species (ROS), catalases, mitogen-activated protein kinase-related signal transduction pathways, and ethylene (de Jong et al., 2002; Ren et al., 2002; Hackenberg et al., 2013; Wang et al., 2013). In particular, spatiotemporal induction of ROS production is crucial to activate PCD, such as the hypersensitive response reaction trigged by several plant-pathogen interactions (Dickman and Fluhr, 2013). In addition, mitogen-activated protein kinase signal transduction networks and ethylene induce PCD, and blocking these pathways impairs the progression of PCD (de Jong et al., 2002; Ren et al., 2002). We found that transcripts in the peroxidase category, which are responsible for ROS generation, were at significantly higher levels in the induced plants compared with the control plants. Likewise, genes associated with ethylene generation also were increased in the VND7-induced seedlings (Fig. 4). Taken together, these analyses show that the VND7 system effectively induces secondary wall synthesis and that a combination between transcript and metabolite profiling gives useful insights into this process.

Clustering of Transcript and Metabolite Data

To look at the overall patterns of gene expression and metabolite levels during secondary wall formation, we first performed Short Time-series Expression Miner (STEM) clustering of transcripts and metabolites (Ernst et al., 2005). STEM clustering was applied to the log2-fold changes obtained from the differential behavior analysis at each time point, with significance level of 0.05, FDR corrected (for details, see “Materials and Methods”).

As a result of STEM clustering, we obtained four significant profiles that represent groups of genes that are coexpressed over the induction time course (Fig. 7; Supplemental Table S7). We argued that the clusters containing already known SCW-related genes are of particular interest, as these clusters might contain other genes potentially important or involved in the SCW transition. Many genes associated with secondary wall production were included in cluster 21. For example, genes encoding cellulose synthase catalytic subunits, CesA4 (At5g44030), CesA7 (At5g17420), and CesA8 (At4g18780; Taylor et al., 2003), were associated with this cluster and were rapidly and highly induced (Supplemental Fig. S4; Watanabe et al., 2015). In this cluster, we also found many other genes involved in cellulose and xylan biosynthesis and assembly. Another group of genes in this cluster are involved in lignin biosynthesis and polymerization. Many of the TF-related genes that are associated with SCW induction also appeared in cluster 21 (highlighted in Supplemental Table S7; Zhong et al., 2008).

Figure 7.

Figure 7.

Open in a new tab

Clustering of transcript and metabolite profiles. A and B illustrate the significant (P < 0.05) cluster profiles resulting from applying STEM clustering on transcript and metabolite profiles, respectively. STEM clustering has been applied to the log2-fold changes obtained from differential behavior analysis at each time point (considering DMSO-treated VND7-VP16-GR plants as controls). Red lines indicate average profiles for each cluster.

Apart from cluster 21, cluster 13 also contained many genes potentially associated with secondary wall synthesis, as their expression patterns increased substantially at early induction time points. Similar to cluster 21, several cellulose synthesis-related genes (COBL4, AtFLA12, IRX14, I14H/IRX14L, GXM1, GXM2, GXM3/GXMT1, RWA3, RWA4, and F8H) and TF-encoding genes (MYB83, MYB58, MYB7, MYB32, and BLH6) belong to this cluster. In tracheary elements and fibers of vascular plants, SCWs are deposited in elaborate patterns. Some genes encoding proteins that are necessary for patterned deposition of secondary walls also were included in this cluster (FRA1, MAP65-8, Kinesin-13A, ROP11, and ROPGAP3). In addition to clusters 21 and 13, cluster 5 also contains some previously reported secondary wall-related genes, such as CslA2 and CslA9, two genes encoding glucomannan synthases, MUCILAGE RELATED10, which is likely involved in the galactosylation of glucomannan (Voiniciuc et al., 2015), and AtPRX2, encoding a peroxidase that can catalyze the polymerization of monolignols. Hence, clusters 13 and 21, and to some extent cluster 5, contain genes associated with SCW production.

Six significant clusters were revealed from our STEM analysis of the metabolites (Fig. 7B; Supplemental Table S8). The largest cluster, cluster 10, contains 92 metabolites that showed a continual decrease during VND7 induction. Several fundamental compounds in the tricarboxylic acid cycle, such as succinate, fumarate, 2-oxoglutarate, and pyruvate, are associated with this cluster, corroborating a role of the tricarboxylic acid cycle during secondary wall synthesis (van der Merwe et al., 2010). Metabolites in clusters 21 and 11 are mostly lipid-related compounds. As mentioned above, the coordinated increase of these compounds at the end of the VND7 induction phase suggests that they are associated with PCD, similar to what has been shown for senescing leaves (Kaup et al., 2002).

Network Analyses of Transcript and Metabolite Data

To gain further insight into the relation between the transcript and metabolite changes, we generated network representations by dampening indirect effects based on two approaches: network deconvolution (Feizi et al., 2013) and global silencing (Barzel and Barabási, 2013; for details, see “Materials and Methods”).

Using the transcripts associated with the four major transcript-related clusters (Fig. 7A), we created two types of networks: one that included all the transcripts and four that included only the transcripts associated with each of the four clusters, respectively (Fig. 8A; Supplemental Fig. S7). From the cluster-based networks, it is clear that a wide variety of ontology terms are associated with each network, and it is difficult to directly associate a distinct process with a certain transcript profile (with the notable exception of photosynthesis; Supplemental Fig. S7). However, apart from the typical secondary wall-related genes described in the cluster analysis (Fig. 7 and discussions above), several interesting connections are evident when looking at individual transcripts. For example, several hormones, including auxin, cytokinin, gibberellin, and brassinosteroid, regulate vasculature patterning and vascular cell differentiation (Kubo et al., 2005). DIMINUTO1 (DIM1), encoding a protein that has a role in the conversion of 24-methylene-cholesterol to campesterol, affects SCW formation in Arabidopsis (Hossain et al., 2012). We found that DIM1 is associated with GIBBERELLIN INSENSITIVE, a gene encoding a DELLA protein that is the master negative regulator in gibberellin signaling, acting in the nucleus as a transcriptional regulator (Marín-de la Rosa et al., 2014), reinforcing the potential importance of gibberellin in SCW formation (Fig. 8). These transcript profiles are coordinated with the three TFs ASL19/LBD30, ASL20/LBD18, and KNAT7 (Soyano et al., 2008; Liu et al., 2014; Fig. 8). ASL19/LBD30 and ASL20/LBD18 mediate a feedback pathway downstream of VND7 (Soyano et al., 2008), and their expression is induced by auxin. Furthermore, in this network, these TFs are linked to several other hormone-related genes, such as SAR1 (Parry et al., 2006), PIN7 (Robert et al., 2013), IAA20 (Sato and Yamamoto, 2008), AtDAO2 (Voß et al., 2015), AFB5 (Walsh et al., 2006), ARR5 (Li et al., 2013), GASA6 (Zhong et al., 2015), and EXO (Müssig et al., 2006). Therefore, these data indicate a role for hormone cross talk during the downstream processes of VND7 (Fig. 8B).

Figure 8.

Figure 8.

Open in a new tab

Network analysis of transcript changes. A, Network of all transcripts associated with clusters 5, 6, 13, and 21 (see Fig. 7). Nodes represent genes, and edges represent significant similarities (FDR-corrected P < 0.05) between the differential behavior of transcripts during VND7 induction. B, Selected nodes associated with hormones are highlighted.

The metabolite-based network revealed that many amino acids, which are part of cluster 11, are closely associated in the network (Fig. 9A). Here, Trp is connected to Tyr, Met, Val, Leu, and Ser, further indicating a common behavior in amino acid levels during VND7 induction (Fig. 9). Given that the metabolite profiles in cluster 11 increase substantially during the final three time points, it appears likely that the increase in amino acids is due to protein degradation during PCD. In addition, Trp also is connected to indole-3-acetate, or auxin, suggesting a role both during the patterning of the vasculature (Ibañes et al., 2009) and during stages of secondary wall completion. These data are consistent with recent studies that link vascular auxin transport with secondary wall differentiation and maturation (Ranocha et al., 2010, 2013).

Figure 9.

Figure 9.

Open in a new tab

Network analysis of metabolite changes. A, Network of all metabolites associated with clusters 10, 11, 21, 29, 37, and 39 (see Fig. 7). Nodes represent metabolites, and edges represent significant similarities (FDR-corrected P < 0.05) between the differential behavior of metabolites during VND7 induction. B, Selected nodes associated with Trp.

To reveal potential relevant relationships between the transcripts and metabolites, we also constructed networks containing both types of profiling data. For this purpose, we used an elastic net regression model (Zou and Hastie, 2012) to predict the genes that best explain the metabolite profiles (for details, see “Materials and Methods”). We chose to approach the lignin-related phenylpropanoid pathway, as this pathway clearly is relevant to secondary wall formation and since very little is known about how the lignin pathway is coordinated with other metabolites and transcripts. We found that several metabolites associated with the phenylpropanoid pathway, such as sinapoyl malate, caffeate, coniferin, and syringin, are present in our network of metabolite clusters (node 10; Fig. 10A; Supplemental Data Set S1). Surprisingly, a variety of glucosinolate pathway-related compounds, including glucoberteroin, glucobrassicine, glucoerucin, 3-indolyl methyl desulfo glucosinolate, and others, were closely associated with the phenylpropanoid pathway-related compounds in the same cluster. Metabolic cross talk is an interesting topic, and a recent study by Kim et al. (2015) identified a Mediator-based cross talk between the glucosinolate and phenylpropanoid pathways, which probably occurs by genes associated in the first steps of the phenylpropanoid pathway.

Figure 10.

Figure 10.

Open in a new tab

Network analysis of transcript and metabolite clusters and of all metabolite and transcript changes. A, to C, Networks of metabolite, transcript, and metabolite-transcript clusters (Supplemental Figs. S7 and S8). Nodes represent clusters (the coloration of nodes indicates time points at which the number of differentially behaving transcripts or metabolites is enriched over a certain significance threshold; FDR-corrected P < 0.05); white nodes indicate that the differentially behaving metabolites or transcripts are not overrepresented in the corresponding cluster at any time point. The edges represent significantly similar cluster profiles (green/red edges represent positive/negative correlation between profile pairs). Time point coloration is indicated in each key. D, Regression-based network of all differentially behaving transcripts and metabolites. Nodes indicate genes/metabolites, and edges are drawn between two nodes if the regression coefficient in the model is not zero. Metabolites and transcripts are represented with red and blue nodes, respectively. Red/blue edges indicate positive/negative regression coefficients. All networks are available as Cytoscape-compatible files in Supplemental Data Set S1.

Moreover, we performed network analysis on the results obtained from the STEM clustering of transcripts and metabolites. For each cluster, we obtained the average of profiles across all time points. The average profiles were then used to infer the network between clusters (see “Materials and Methods”). This analysis was performed either separately on transcript and metabolite clusters or on the combination of transcript and metabolite clusters. In the resulting networks, each node thus represents a cluster, and edges are drawn if there is correlation (which is not due to indirect effects) between the average profiles of the corresponding nodes. We then sorted the clusters based on the time points at which the differentially behaving transcripts/metabolites are overrepresented by applying a hypergeometric test (Kachitvichyanukul and Schmeiser, 1985) using the phyper function in the R package stats (Johnson et al., 1992). The place and color of the nodes (clusters) in Figure 10, A to C, therefore correspond to the time points at which the differentially behaving transcripts/metabolites are overrepresented.

Cluster 10 is closely associated with transcript cluster 21 (Fig. 10A); therefore, we expected to see some genes associated with the metabolic changes observed in cluster 10. Indeed, in cluster 21, we find genes encoding a β-glucosidase gene, BGLU28, a plastid-localized arogenate dehydratase gene, ADT4, and the flavin-binding monooxygenase gene At5g07800, which are implicated in the glucosinolate or phenylpropanoid pathway (Cho et al., 2007; Malitsky et al., 2008). Notably, a large number of genes with functions in lignin biosynthesis are enriched in this cluster (e.g. 4CL5, CCR2, and CYP84A1), substantiating the coordination of glucosinolates and lignin-related pathways. In addition, several important TFs (e.g. KNAT7, LBD18, and LBD30) also are present in this cluster, supporting a connection between the pathways and secondary wall regulation.

CONCLUSION

The construction of a xylem vessel is an important developmental step in plant biology, and a better understanding of this process may impact fuel and biomaterials production. We analyzed transcript and metabolite changes during this process and paid particular attention to SCW-related changes. Our data provide insights into links between processes in central metabolism, including the tricarboxylic acid cycle and amino acid synthesis, and SCW synthesis. In addition, the combined data of nucleotide sugar levels and the expression of genes encoding enzymes and transporters of the nucleotide sugar interconversion pathway suggest that different nucleotide sugars might be transported into the Golgi apparatus differently during SCW formation. While we have mainly focused our efforts on cell wall-related aspects, the generated data sets also may be useful to better understand other processes of vessel development, such as PCD and the cessation of cell elongation. One possible way to achieve this would be to introgress mutants that affect cell wall synthesis or delay PCD into the VND7 system and then analyze what consequences these mutations have on the system. Likewise, it may be suitable to treat VND7 seedlings with inhibitors related to these processes during the induction. These efforts might improve our understanding of how these interlinked processes work during vessel formation. Furthermore, the comprehensive network analyses might provide a general framework for further investigation of multiple SCW-related biological processes in Arabidopsis and other plants.

MATERIALS AND METHODS

Plant Material and Culturing Conditions

An Arabidopsis (Arabidopsis thaliana) plant line (Columbia-0) that harbors a DEX-inducible pH35GVGR Gateway-based construct containing VND7-VP16-GR under the control of a 35S promoter was described previously (Yamaguchi et al., 2010). T3 seeds were surface sterilized (10 min, 3% sodium hypochlorite) and washed five times with 1 mL of sterilized water each time. About 15 to 20 mg of seeds (weight was carefully monitored) was added to each flask containing 100 mL of liquid MS medium supplemented with 1% Suc (3.9 g L−1 MS minerals including vitamins [Duchefa], adjusted with 0.45 g L−1 MES buffer and KOH to pH 5.7–5.8). Cultures were grown at 23°C for 6 d in the dark with 80 rpm on a GFL 3017 shaker (GFL). The cultures were induced by adding 10 µL of DEX stock solution (dissolved in DMSO) into liquid medium of each flask to a final concentration of 10 µm DEX and further cultivated for the time-course experiment. Flasks were harvested at the indicated time points after induction (1, 3, 6, 9, 12, 24, 30, and 48 h) and at 0 h. Six flasks were used at each time point and for both conditions (i.e. six flasks for 0 h and 12 flasks for each other time point; 102 flasks in total). Material in one flask harvested at each time point served as one biological replicate. Uninduced cultures (adding 10 µL of DMSO instead of DEX) served as controls for each time point; thus, there were 48 microarray analyses for eight time points (three biological replicates) for +/− induced VND7-VP16-GR cultures (three arrays for 0 h). Flasks were decanted over filter paper and washed three times with sterilized dest. water, dried on tissue paper, flash frozen, and ball milled (Retsch; 1–3 min, 30 Hz). Material from each flask was divided for different measurements, and the weight of each portion was carefully measured. To gain comparable data, all analyses, including transcriptomics, metabolomics, and nucleotide sugar determination, were undertaken using material from the same samples at each time point. For transcriptomics, three biological replicates were used, and for metabolomics and nucleotide sugar analyses at least three (three to six) biological replicates were taken. To ensure that the DEX did not influence the measurements, empty vector construct plants (VP16-GR; i.e. without the VND7 gene) were used at three time points (9, 12, and 24 h) for transcript analysis and two time points (9 and 24 h) for metabolic analysis.

Transcriptomics

Total RNA was isolated using RNAeasy Kits (Qiagen) following the manufacturer’s instructions and was sent for Affymetrix ATH1 expression profiling using the full microarray service at Atlas Biolabs. The obtained ATH1 chip and expression data (CEL files) were quality checked and evaluated using the Robin software package (Lohse et al., 2010). After chip quality control, data were analyzed as described in “Bioinformatics Analysis and Network Construction” below.

Metabolic Profiling

Primary, secondary, and lipid metabolite extraction and profiling were performed as described by Caldana et al. (2013) using 40 to 60 mg of frozen, fine-powdered plant material. In brief, the methyl-tert-butyl-ether method (Giavalisco et al., 2011) was used, and the lipid phase was separated from the polar phase (primary and secondary metabolites). The lipid phase was analyzed utilizing UPLC-FT-MS and MS/MS (Hummel et al., 2011), whereas the polar phase was split 1:3 (v/v) to analyze primary metabolites by derivatization and gas chromatography-mass spectrometry (Cuadros-Inostroza et al., 2009; Caldana et al., 2013) and secondary metabolites by UPLC-FT-MS and MS/MS (Giavalisco et al., 2011).

Nucleotide Sugar Extraction and Measurements

Frozen and ground plant materials from the VND7-VP16-GR induced and uninduced cultures were extracted from 10 to 25 mg fresh weight as described (Arrivault et al., 2009). The freeze-dried extract was dissolved in 1.2 mL of 10 mm ammonium bicarbonate before using the Envi-Carb (Sigma-Aldrich) column for solid-phase extraction as described (Räbinä et al., 2001). Purified extracts were lyophilized and dissolved in 100 µL of liquid chromatography-mass spectrometry-grade water. Nucleotide sugars were detected using a Thermo Vantage triple quadrupole mass spectrometer (Thermo Scientific) and an electrospray ionization source. The system was operated in negative ion mode using selected reaction monitoring mode. The voltage of the source was adjusted to −4,000 V. The capillary temperature was 320°C. Nitrogen was used as sheath and auxiliary gas (40 and 5 arbitrary units), and argon gas at a pressure of 1.1 mTorr was used for collision. The peak widths for quadrupoles Q1 and Q3 were set to 0.7 mass-to-charge ratio. For quantification, the Thermo-Finnigan software package LCQuan 2.5 was used. Nucleotide sugars were quantified by comparing the integrated peak areas with calibration curves obtained from the NDP sugar standards.

Liquid Chromatography and MS/MS Detection

Ion-pairing liquid chromatography was performed on a Thermo Accela system (Thermo Scientific) consisting of Accela pump 1250 and Accela autosampler with a column compartment. The injection volume was 25 µL, and each sample was additionally injected diluted 1:10. The nucleotide sugars were separated by reverse-phase chromatography on a Synergi 4u Hydro 80A 150- × 2-mm column with a security guard cartridge (C18, 4 × 3 mm; Phenomenex). The column temperature was adjusted to 28°C. The separation of nucleotide sugars was performed by a flow gradient of 20 mm buffered triethylamine/acetic acid (A), pH 6, at a flow rate of 0.2 to 0.3 mL min−1 in combination with a linear acetonitrile (B) gradient ranging from 0% to 90%. The initial flow rate and equilibration flow was set to 0.2 mL min−1. The parameters for the solvent gradient and flow were as follows: 0 to 10 min, flow from 0.2 to 0.25 mL min−1 and 0% solvent B; 10 to 19 min, flow from 0.25 to 0.3 mL min−1 and a 0% to 16% gradient of solvent B; 19 to 22 min, flow from 0.3 to 0.25 mL min−1 and a 16% to 90% gradient of solvent B; 22 to 26 min, flow of 0.25 mL min−1 and 90% solvent B; and 26 to 38 min, flow of 0.2 mL min−1 and 0% concentration of solvent B.

Bioinformatics Analysis and Network Construction

The analysis work flow is illustrated in Supplemental Figure S8 and described in detail below.

Data Preprocessing

RNA levels were measured by high-throughput technique using RNA microarray Affymetrix ATH1, and the probe sets were redefined according to Dai et al. (2005). The expression levels of probe sets were then estimated by GC-robust multiarray analysis using the R package gcrma (Wu et al., 2004). Furthermore, genes with nonreproducible replicates on the array were eliminated by applying ANOVA followed by Tukey’s posthoc test (significance level of 0.01). In addition, a gene was filtered out if the sd between its replicate pairs was above 0.2.

Finally, the primary, secondary, and lipid metabolite intensities were first corrected by the corresponding internal standard for each analysis ([13C]sorbitol for gas chromatography data, ampicillin for secondary metabolite data, and phosphatidylcholine 34:0 for lipid data). After this correction, the intensities were corrected for the variations in fresh weights and finally log (base 2) transformed to fulfill the normality assumption and exclude the dominant effect of extreme small/large values (Caldana et al., 2013).

Analysis of Differential Behavior

Analysis of differential behavior was conducted on the data after data preprocessing. Differential behavior was inspected between two conditions (i.e. control and treatment) for the following three scenarios: (1) at a specific time point; (2) between consecutive time points; and (3) over all time points. For the first and second scenarios, a linear model was employed to detect differential behavior. To this end, we applied the R package limma (Smyth, 2005). For the third scenario, differential behavior was detected by the statistical method capturing global differential behavior as implemented in the R packages timecourse (Tai, 2007), limma (Smyth, 2005), and samr (Tibshirani et al., 2011).

Clustering

We used the STEM approach to cluster both transcript and metabolite profiles (Ernst et al., 2005). STEM clustering has been applied to the log2-fold changes obtained from differential behavior analysis at each time point (i.e. scenario 1, considering DMSO-treated VND7-VP16-GR plants as a control). To select an appropriate number of model profiles (clusters), we set the maximum unit change between time points to two and repeated the STEM clustering with different values for the maximum number of model profiles (number of clusters): 15, 25, and 50. For each resulting clustering, we then calculated the average silhouette index from the R package cluster (Maechler et al., 2013) over all significant clusters (significance level of 0.05, FDR corrected). The maximum average silhouette indices for transcript and metabolite clustering have been captured with 25 and 50 as the maximum number of model profiles, respectively.

GO Enrichment Analysis

The R package GOstats (Falcon and Gentleman, 2007) was used to find the biological processes and molecular functions that are overenriched in the clustering results. The cutoff value for significance level was set at 0.05.

Network Analysis

Gene Regulatory Network

To infer a gene regulatory network, we applied two recent approaches: global silencing (Barzel and Barabási, 2013) and network deconvolution (Feizi et al., 2013), which detect and dampen indirect relationships based on (weighted) structural investigations of networks following eigenvalue decomposition and its derivatives, leading to higher inference power than the classically applied approaches.

First, for the given data set, we calculated the Pearson correlation coefficients matrix Sg×g. Given g1 regulators and g2 nonregulators, with g = g1 + g2, the correlation matrix can be modified as

Inline graphic

where O denotes the zero matrix, to include biological roles (TF and non-TF genes). We extracted the regulatory genes (TFs) from different databases, such as AGRIS (Palaniswamy et al., 2006), PlnTFDB (Pérez-Rodríguez et al., 2010), and DATF (Guo et al., 2005). We then applied the network deconvolution and global silencing methods to the modified correlation matrix S′. However, global silencing depends on finding the inverse of the correlation matrix that is rank deficient in the case p » n, where p is the number of genes and n is the number of features, as with the data analyzed here. Since finding an inverse for a rank-deficient matrix is an ill-posed problem, we resolved it by adding a noise term that renders the matrix positive definite. We then selected the best result, with respect to a match with experimentally verified regulatory interactions, from 10 runs of the procedure as a final outcome. The resulting distribution of weighted matrices for the regulatory interactions obtained by each method was decomposed into the mixture of two Gaussian distributions, and the value at which the two distributions intersect was taken as a cutoff for filtering the resulting interaction weight matrices. The latter was conducted to avoid arbitrary selection of a threshold value and prompted by the bimodality of the regulatory interaction weight matrices resulting from these methods. Finally, the gene regulatory network is attained by taking the shared regulatory interactions between the resulting filtered regulatory interactions obtained by the two approaches. The edges were rescored based on the geometric mean of the scores obtained by the two approaches.

Metabolite-Transcript Network

The relationships between all measured metabolites across different time points and the transcript profiles were obtained by employing elastic net regression (cvenet from the elasticnet package in R; Zou and Hastie, 2012). The profiles of the transcripts were considered as predictors (regressors). Regression models were then fitted for each metabolite profile separately. The regression coefficients were robustly estimated by 10-fold cross-validation based on the optimum value for the penalty parameter from the set {0.01, 0.05, 0.1, 0.5, 1, 1.5, 2, 10, 100}.

Supplemental Data

The following supplemental materials are available.

Acknowledgments

We thank Drs. Berit Ebert, Joshua Heazlewood, and Wei Zeng for useful comments on the article.

Glossary

SCW

secondary cell wall

PCD

programmed cell death

TF

transcription factor

DEX

dexamethasone

MS

Murashige and Skoog

HAI

hours after induction

FDR

false discovery rate

MS/MS

tandem mass spectrometry

UPLC-FT-MS

ultra-performance liquid chromatography-Fourier transform-mass spectrometry

DEGs

differentially expressed genes

RSD

ratio of shared difference

GO

Gene Ontology

BP

Biological Process

MF

Molecular Function

TAG

triacylglyceride

ROS

reactive oxygen species

STEM

Short Time-series Expression Miner

DMSO

dimethyl sulfoxide

Footnotes

1

This work was supported by the Max-Planck Gesellschaft, by MIFRS/MIRS scholarships from the University of Melbourne (to Z.L.), by the European Union and the European Seventh Framework Programme (grant no. 311804), by a R@MAP professorship at the University of Melbourne (to S.P.), and by the IRRTF-RNC (grant no. 501892 to S.P. and Z.N.).

2

These authors contributed equally to this article.

References

  1. Anderson CT, Wallace IS, Somerville CR (2012) Metabolic click-labeling with a fucose analog reveals pectin delivery, architecture, and dynamics in Arabidopsis cell walls. Proc Natl Acad Sci USA 109: 1329–1334 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Arrivault S, Guenther M, Ivakov A, Feil R, Vosloh D, van Dongen JT, Sulpice R, Stitt M (2009) Use of reverse-phase liquid chromatography, linked to tandem mass spectrometry, to profile the Calvin cycle and other metabolic intermediates in Arabidopsis rosettes at different carbon dioxide concentrations. Plant J 59: 826–839 [DOI] [PubMed] [Google Scholar]
  3. Atmodjo MA, Hao Z, Mohnen D (2013) Evolving views of pectin biosynthesis. Annu Rev Plant Biol 64: 747–779 [DOI] [PubMed] [Google Scholar]
  4. Barzel B, Barabási AL (2013) Network link prediction by global silencing of indirect correlations. Nat Biotechnol 31: 720–725 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Berger S, Sinha AK, Roitsch T (2007) Plant physiology meets phytopathology: plant primary metabolism and plant-pathogen interactions. J Exp Bot 58: 4019–4026 [DOI] [PubMed] [Google Scholar]
  6. Boerjan W, Ralph J, Baucher M (2003) Lignin biosynthesis. Annu Rev Plant Biol 54: 519–546 [DOI] [PubMed] [Google Scholar]
  7. Bollhöner B, Prestele J, Tuominen H (2012) Xylem cell death: emerging understanding of regulation and function. J Exp Bot 63: 1081–1094 [DOI] [PubMed] [Google Scholar]
  8. Bonawitz ND, Chapple C (2010) The genetics of lignin biosynthesis: connecting genotype to phenotype. Annu Rev Genet 44: 337–363 [DOI] [PubMed] [Google Scholar]
  9. Boudière L, Michaud M, Petroutsos D, Rébeillé F, Falconet D, Bastien O, Roy S, Finazzi G, Rolland N, Jouhet J, et al. (2014) Glycerolipids in photosynthesis: composition, synthesis and trafficking. Biochim Biophys Acta 1837: 470–480 [DOI] [PubMed] [Google Scholar]
  10. Caldana C, Li Y, Leisse A, Zhang Y, Bartholomaeus L, Fernie AR, Willmitzer L, Giavalisco P (2013) Systemic analysis of inducible target of rapamycin mutants reveal a general metabolic switch controlling growth in Arabidopsis thaliana. Plant J 73: 897–909 [DOI] [PubMed] [Google Scholar]
  11. Carpita N, McCann M (2000) The cell wall. In Buchanan BB, Wilhelm G, Jones RL, eds, Biochemistry and Molecular Biology of Plants. American Society of Plant Physiologists, Rockville, MD, pp 52–108 [Google Scholar]
  12. Cassan-Wang H, Goué N, Saidi MN, Legay S, Sivadon P, Goffner D, Grima-Pettenati J (2013) Identification of novel transcription factors regulating secondary cell wall formation in Arabidopsis. Front Plant Sci 4: 189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Cheong YH, Chang HS, Gupta R, Wang X, Zhu T, Luan S (2002) Transcriptional profiling reveals novel interactions between wounding, pathogen, abiotic stress, and hormonal responses in Arabidopsis. Plant Physiol 129: 661–677 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Cho MH, Corea ORA, Yang H, Bedgar DL, Laskar DD, Anterola AM, Moog-Anterola FA, Hood RL, Kohalmi SE, Bernards MA, et al. (2007) Phenylalanine biosynthesis in Arabidopsis thaliana: identification and characterization of arogenate dehydratases. J Biol Chem 282: 30827–30835 [DOI] [PubMed] [Google Scholar]
  15. Coll NS, Epple P, Dangl JL (2011) Programmed cell death in the plant immune system. Cell Death Differ 18: 1247–1256 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Cuadros-Inostroza A, Caldana C, Redestig H, Kusano M, Lisec J, Peña-Cortés H, Willmitzer L, Hannah MA (2009) TargetSearch: a Bioconductor package for the efficient preprocessing of GC-MS metabolite profiling data. BMC Bioinformatics 10: 428. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dai M, Wang P, Boyd AD, Kostov G, Athey B, Jones EG, Bunney WE, Myers RM, Speed TP, Akil H, et al. (2005) Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data. Nucleic Acids Res 33: e175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. de Jong AJ, Yakimova ET, Kapchina VM, Woltering EJ (2002) A critical role for ethylene in hydrogen peroxide release during programmed cell death in tomato suspension cells. Planta 214: 537–545 [DOI] [PubMed] [Google Scholar]
  19. Demura T, Tashiro G, Horiguchi G, Kishimoto N, Kubo M, Matsuoka N, Minami A, Nagata-Hiwatashi M, Nakamura K, Okamura Y, et al. (2002) Visualization by comprehensive microarray analysis of gene expression programs during transdifferentiation of mesophyll cells into xylem cells. Proc Natl Acad Sci USA 99: 15794–15799 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Derbyshire P, Ménard D, Green P, Saalbach G, Buschmann H, Lloyd CW, Pesquet E (2015) Proteomic Analysis of Microtubule Interacting Proteins over the Course of Xylem Tracheary Element Formation in Arabidopsis. Plant Cell 27: 2709–2726 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Dickman MB, Fluhr R (2013) Centrality of host cell death in plant-microbe interactions. Annu Rev Phytopathol 51: 543–570 [DOI] [PubMed] [Google Scholar]
  22. Endler A, Persson S (2011) Cellulose synthases and synthesis in Arabidopsis. Mol Plant 4: 199–211 [DOI] [PubMed] [Google Scholar]
  23. Endo S, Pesquet E, Yamaguchi M, Tashiro G, Sato M, Toyooka K, Nishikubo N, Udagawa-Motose M, Kubo M, Fukuda H, et al. (2009) Identifying new components participating in the secondary cell wall formation of vessel elements in Zinnia and Arabidopsis. Plant Cell 21: 1155–1165 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ernst J, Nau GJ, Bar-Joseph Z (2005) Clustering short time series gene expression data. Bioinformatics (Suppl 1) 21: i159–i168 [DOI] [PubMed] [Google Scholar]
  25. Falcon S, Gentleman R (2007) Using GOstats to test gene lists for GO term association. Bioinformatics 23: 257–258 [DOI] [PubMed] [Google Scholar]
  26. Feizi S, Marbach D, Médard M, Kellis M (2013) Network deconvolution as a general method to distinguish direct dependencies in networks. Nat Biotechnol 31: 726–733 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Giavalisco P, Li Y, Matthes A, Eckhardt A, Hubberten HM, Hesse H, Segu S, Hummel J, Köhl K, Willmitzer L (2011) Elemental formula annotation of polar and lipophilic metabolites using 13C, 15N and 34S isotope labelling, in combination with high-resolution mass spectrometry. Plant J 68: 364–376 [DOI] [PubMed] [Google Scholar]
  28. Guo A, He K, Liu D, Bai S, Gu X, Wei L, Luo J (2005) DATF: a database of Arabidopsis transcription factors. Bioinformatics 21: 2568–2569 [DOI] [PubMed] [Google Scholar]
  29. Gruissem W. (1989) Chloroplast gene expression: how plants turn their plastids on. Cell 56: 161–170 [DOI] [PubMed] [Google Scholar]
  30. Hackenberg T, Juul T, Auzina A, Gwiżdż S, Malolepszy A, Van Der Kelen K, Dam S, Bressendorff S, Lorentzen A, Roepstorff P, et al. (2013) Catalase and NO CATALASE ACTIVITY1 promote autophagy-dependent cell death in Arabidopsis. Plant Cell 25: 4616–4626 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Hirano K, Kondo M, Aya K, Miyao A, Sato Y, Antonio BA, Namiki N, Nagamura Y, Matsuoka M (2013) Identification of transcription factors involved in rice secondary cell wall formation. Plant Cell Physiol 54: 1791–1802 [DOI] [PubMed] [Google Scholar]
  32. Hossain Z, McGarvey B, Amyot L, Gruber M, Jung J, Hannoufa A (2012) DIMINUTO 1 affects the lignin profile and secondary cell wall formation in Arabidopsis. Planta 235: 485–498 [DOI] [PubMed] [Google Scholar]
  33. Hummel J, Segu S, Li Y, Irgang S, Jueppner J, Giavalisco P (2011) Ultra performance liquid chromatography and high resolution mass spectrometry for the analysis of plant lipids. Front Plant Sci 2: 54. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Ibañes M, Fàbregas N, Chory J, Caño-Delgado AI (2009) Brassinosteroid signaling and auxin transport are required to establish the periodic pattern of Arabidopsis shoot vascular bundles. Proc Natl Acad Sci USA 106: 13630–13635 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Johnson NL, Kotz S, Kemp AW (1992) Univariate Discrete Distributions, Ed 2 Wiley, New York [Google Scholar]
  36. Kachitvichyanukul V, Schmeiser B (1985) Computer generation of hypergeometric random variates. J Stat Comput Simul 22: 127–145 [Google Scholar]
  37. Kaup MT, Froese CD, Thompson JE (2002) A role for diacylglycerol acyltransferase during leaf senescence. Plant Physiol 129: 1616–1626 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Kim JI, Dolan WL, Anderson NA, Chapple C (2015) Indole glucosinolate biosynthesis limits phenylpropanoid accumulation in Arabidopsis thaliana. Plant Cell 27: 1529–1546 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Kubo M, Udagawa M, Nishikubo N, Horiguchi G, Yamaguchi M, Ito J, Mimura T, Fukuda H, Demura T (2005) Transcription switches for protoxylem and metaxylem vessel formation. Genes Dev 19: 1855–1860 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Li Y, Kurepa J, Smalle J (2013) AXR1 promotes the Arabidopsis cytokinin response by facilitating ARR5 proteolysis. Plant J 74: 13–24 [DOI] [PubMed] [Google Scholar]
  41. Li Z, Fernie AR, Persson S (2016) Transition of primary to secondary cell wall synthesis. Sci Bull 61: 838–846 [Google Scholar]
  42. Liu Y, You S, Taylor-Teeples M, Li WL, Schuetz M, Brady SM, Douglas CJ (2014) BEL1-LIKE HOMEODOMAIN6 and KNOTTED ARABIDOPSIS THALIANA7 interact and regulate secondary cell wall formation via repression of REVOLUTA. Plant Cell 26: 4843–4861 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Lohse M, Nunes-Nesi A, Krüger P, Nagel A, Hannemann J, Giorgi FM, Childs L, Osorio S, Walther D, Selbig J, et al. (2010) Robin: an intuitive wizard application for R-based expression microarray quality assessment and analysis. Plant Physiol 153: 642–651 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Maechler M, Rousseeuw P, Struyf A, Hubert M, Hornik K (2013) Cluster: cluster analysis basics and extensions. R package version 1(4), http://cran.r-project.org/package=cluster
  45. Malitsky S, Blum E, Less H, Venger I, Elbaz M, Morin S, Eshed Y, Aharoni A (2008) The transcript and metabolite networks affected by the two clades of Arabidopsis glucosinolate biosynthesis regulators. Plant Physiol 148: 2021–2049 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Manabe Y, Tinker N, Colville A, Miki B (2007) CSR1, the sole target of imidazolinone herbicide in Arabidopsis thaliana. Plant Cell Physiol 48: 1340–1358 [DOI] [PubMed] [Google Scholar]
  47. Marín-de la Rosa N, Sotillo B, Miskolczi P, Gibbs DJ, Vicente J, Carbonero P, Oñate-Sánchez L, Holdsworth MJ, Bhalerao R, Alabadí D, et al. (2014) Large-scale identification of gibberellin-related transcription factors defines group VII ETHYLENE RESPONSE FACTORS as functional DELLA partners. Plant Physiol 166: 1022–1032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Maruta T, Yonemitsu M, Yabuta Y, Tamoi M, Ishikawa T, Shigeoka S (2008) Arabidopsis phosphomannose isomerase 1, but not phosphomannose isomerase 2, is essential for ascorbic acid biosynthesis. J Biol Chem 283: 28842–28851 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. McCann M. (1997) Tracheary element formation: building up to a dead end. Trends Plant Sci 2: 333–338 [Google Scholar]
  50. McFarlane HE, Döring A, Persson S (2014) The cell biology of cellulose synthesis. Annu Rev Plant Biol 65: 69–94 [DOI] [PubMed] [Google Scholar]
  51. Meyer RC, Steinfath M, Lisec J, Becher M, Witucka-Wall H, Törjék O, Fiehn O, Eckardt A, Willmitzer L, Selbig J, et al. (2007) The metabolic signature related to high plant growth rate in Arabidopsis thaliana. Proc Natl Acad Sci USA 104: 4759–4764 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Misson J, Raghothama KG, Jain A, Jouhet J, Block MA, Bligny R, Ortet P, Creff A, Somerville S, Rolland N, et al. (2005) A genome-wide transcriptional analysis using Arabidopsis thaliana Affymetrix gene chips determined plant responses to phosphate deprivation. Proc Natl Acad Sci USA 102: 11934–11939 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Müssig C, Lisso J, Coll-Garcia D, Altmann T (2006) Molecular analysis of brassinosteroid action. Plant Biol (Stuttg) 8: 291–296 [DOI] [PubMed] [Google Scholar]
  54. Ohdaira Y, Kakegawa K, Amino S, Sugiyama M, Fukuda H (2002) Activity of cell-wall degradation associated with differentiation of isolated mesophyll cells of Zinnia elegans into tracheary elements. Planta 215: 177–184 [DOI] [PubMed] [Google Scholar]
  55. Palaniswamy SK, James S, Sun H, Lamb RS, Davuluri RV, Grotewold E (2006) AGRIS and AtRegNet: a platform to link cis-regulatory elements and transcription factors into regulatory networks. Plant Physiol 140: 818–829 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Parry G, Ward S, Cernac A, Dharmasiri S, Estelle M (2006) The Arabidopsis SUPPRESSOR OF AUXIN RESISTANCE proteins are nucleoporins with an important role in hormone signaling and development. Plant Cell 18: 1590–1603 [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Pérez-Rodríguez P, Riaño-Pachón DM, Corrêa LGG, Rensing SA, Kersten B, Mueller-Roeber B (2010) PlnTFDB: updated content and new features of the plant transcription factor database. Nucleic Acids Res 38: D822–D827 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Räbinä J, Mäki M, Savilahti EM, Järvinen N, Penttilä L, Renkonen R (2001) Analysis of nucleotide sugars from cell lysates by ion-pair solid-phase extraction and reversed-phase high-performance liquid chromatography. Glycoconj J 18: 799–805 [DOI] [PubMed] [Google Scholar]
  59. Ranocha P, Denancé N, Vanholme R, Freydier A, Martinez Y, Hoffmann L, Köhler L, Pouzet C, Renou JP, Sundberg B, et al. (2010) Walls are thin-1 (WAT1), an Arabidopsis homolog of Medicago truncatula NODULIN21, is a tonoplast-localized protein required for secondary wall formation in fibers. Plant J 63: 469–483 [DOI] [PubMed] [Google Scholar]
  60. Ranocha P, Dima O, Nagy R, Felten J, Corratgé-Faillie C, Novák O, Morreel K, Lacombe B, Martinez Y, Pfrunder S, et al. (2013) Arabidopsis WAT1 is a vacuolar auxin transport facilitator required for auxin homoeostasis. Nat Commun 4: 2625. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Rautengarten C, Ebert B, Moreno I, Temple H, Herter T, Link B, Doñas-Cofré D, Moreno A, Saéz-Aguayo S, Blanco F, et al. (2014) The Golgi localized bifunctional UDP-rhamnose/UDP-galactose transporter family of Arabidopsis. Proc Natl Acad Sci USA 111: 11563–11568 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Ren D, Yang H, Zhang S (2002) Cell death mediated by MAPK is associated with hydrogen peroxide production in Arabidopsis. J Biol Chem 277: 559–565 [DOI] [PubMed] [Google Scholar]
  63. Robert HS, Grones P, Stepanova AN, Robles LM, Lokerse AS, Alonso JM, Weijers D, Friml J (2013) Local auxin sources orient the apical-basal axis in Arabidopsis embryos. Curr Biol 23: 2506–2512 [DOI] [PubMed] [Google Scholar]
  64. Sato A, Yamamoto KT (2008) Overexpression of the non-canonical Aux/IAA genes causes auxin-related aberrant phenotypes in Arabidopsis. Physiol Plant 133: 397–405 [DOI] [PubMed] [Google Scholar]
  65. Scheller HV, Ulvskov P (2010) Hemicelluloses. Annu Rev Plant Biol 61: 263–289 [DOI] [PubMed] [Google Scholar]
  66. Seifert GJ. (2004) Nucleotide sugar interconversions and cell wall biosynthesis: how to bring the inside to the outside. Curr Opin Plant Biol 7: 277–284 [DOI] [PubMed] [Google Scholar]
  67. Smyth GK. (2005) Limma: linear models for microarray data. In Gentleman R, Carey V, Dudoit S, Irizarry R, Huber W, eds, Bioinformatics and Computational Biology Solutions using R and Bioconductor. Springer, New York, pp 397–420 [Google Scholar]
  68. Soyano T, Thitamadee S, Machida Y, Chua NH (2008) ASYMMETRIC LEAVES2-LIKE19/LATERAL ORGAN BOUNDARIES DOMAIN30 and ASL20/LBD18 regulate tracheary element differentiation in Arabidopsis. Plant Cell 20: 3359–3373 [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Sulpice R, Pyl ET, Ishihara H, Trenkamp S, Steinfath M, Witucka-Wall H, Gibon Y, Usadel B, Poree F, Piques MC, et al. (2009) Starch as a major integrator in the regulation of plant growth. Proc Natl Acad Sci USA 106: 10348–10353 [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Sweetlove LJ, Fell D, Fernie AR (2008) Getting to grips with the plant metabolic network. Biochem J 409: 27–41 [DOI] [PubMed] [Google Scholar]
  71. Tai YC (2007) Timecourse: statistical analysis for developmental microarray time course data. R package version 1(0), www.bioconductor.org
  72. Taylor NG, Howells RM, Huttly AK, Vickers K, Turner SR (2003) Interactions among three distinct CesA proteins essential for cellulose synthesis. Proc Natl Acad Sci USA 100: 1450–1455 [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Taylor-Teeples M, Lin L, de Lucas M, Turco G, Toal TW, Gaudinier A, Young NF, Trabucco GM, Veling MT, Lamothe R, et al. (2015) An Arabidopsis gene regulatory network for secondary cell wall synthesis. Nature 517: 571–575 [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Thompson WF. (1988) Photoregulation: diverse gene responses in greening seedlings. Plant Cell Environ 11: 319–328 [Google Scholar]
  75. Tibshirani R, Chu G, Hastie T, Narasimhan B (2011) samr: SAM: significance analysis of microarrays. R package version 2.0, http://cran.r-project.org/package=samr
  76. Turner SR, Somerville CR (1997) Collapsed xylem phenotype of Arabidopsis identifies mutants deficient in cellulose deposition in the secondary cell wall. Plant Cell 9: 689–701 [DOI] [PMC free article] [PubMed] [Google Scholar]
  77. Usadel B, Poree F, Nagel A, Lohse M, Czedik-Eysenberg A, Stitt M (2009) A guide to using MapMan to visualize and compare omics data in plants: a case study in the crop species, maize. Plant Cell Environ 32: 1211–1229 [DOI] [PubMed] [Google Scholar]
  78. van der Merwe MJ, Osorio S, Araújo WL, Balbo I, Nunes-Nesi A, Maximova E, Carrari F, Bunik VI, Persson S, Fernie AR (2010) Tricarboxylic acid cycle activity regulates tomato root growth via effects on secondary cell wall production. Plant Physiol 153: 611–621 [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Voiniciuc C, Schmidt MHW, Berger A, Yang B, Ebert B, Scheller HV, North HM, Usadel B, Günl M (2015) MUCILAGE-RELATED10 produces galactoglucomannan that maintains pectin and cellulose architecture in Arabidopsis seed mucilage. Plant Physiol 169: 403–420 [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Voß U, Wilson MH, Kenobi K, Gould PD, Robertson FC, Peer WA, Lucas M, Swarup K, Casimiro I, Holman TJ, et al. (2015) The circadian clock rephases during lateral root organ initiation in Arabidopsis thaliana. Nat Commun 6: 7641. [DOI] [PMC free article] [PubMed] [Google Scholar]
  81. Walsh TA, Neal R, Merlo AO, Honma M, Hicks GR, Wolff K, Matsumura W, Davies JP (2006) Mutations in an auxin receptor homolog AFB5 and in SGT1b confer resistance to synthetic picolinate auxins and not to 2,4-dichlorophenoxyacetic acid or indole-3-acetic acid in Arabidopsis. Plant Physiol 142: 542–552 [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Wang HH, Tang RJ, Liu H, Chen HY, Liu JY, Jiang XN, Zhang HX (2013) Chimeric repressor of PtSND2 severely affects wood formation in transgenic Populus. Tree Physiol 33: 878–886 [DOI] [PubMed] [Google Scholar]
  83. Watanabe Y, Meents MJ, McDonnell LM, Barkwill S, Sampathkumar A, Cartwright HN, Demura T, Ehrhardt DW, Samuels AL, Mansfield SD (2015) Visualization of cellulose synthases in Arabidopsis secondary cell walls. Science 350: 198–203 [DOI] [PubMed] [Google Scholar]
  84. Wilkinson L. (2012) Exact and approximate area-proportional circular Venn and Euler diagrams. IEEE Trans Vis Comput Graph 18: 321–331 [DOI] [PubMed] [Google Scholar]
  85. Wu Z, Irizarry RA, Gentleman R, Martinez-Murillo F, Spencer F (2004) A model-based background adjustment for oligonucleotide expression arrays. J Am Stat Assoc 99: 909–917 [Google Scholar]
  86. Yamaguchi M, Goué N, Igarashi H, Ohtani M, Nakano Y, Mortimer JC, Nishikubo N, Kubo M, Katayama Y, Kakegawa K, et al. (2010) VASCULAR-RELATED NAC-DOMAIN6 and VASCULAR-RELATED NAC-DOMAIN7 effectively induce transdifferentiation into xylem vessel elements under control of an induction system. Plant Physiol 153: 906–914 [DOI] [PMC free article] [PubMed] [Google Scholar]
  87. Yamaguchi M, Mitsuda N, Ohtani M, Ohme-Takagi M, Kato K, Demura T (2011) VASCULAR-RELATED NAC-DOMAIN7 directly regulates the expression of a broad range of genes for xylem vessel formation. Plant J 66: 579–590 [DOI] [PubMed] [Google Scholar]
  88. Zhong C, Xu H, Ye S, Wang S, Li L, Zhang S, Wang X (2015) Gibberellic Acid-Stimulated Arabidopsis6 serves as an integrator of gibberellin, abscisic acid, and glucose signaling during seed germination in Arabidopsis. Plant Physiol 169: 2288–2303 [DOI] [PMC free article] [PubMed] [Google Scholar]
  89. Zhong R, Lee C, Zhou J, McCarthy RL, Ye ZH (2008) A battery of transcription factors involved in the regulation of secondary cell wall biosynthesis in Arabidopsis. Plant Cell 20: 2763–2782 [DOI] [PMC free article] [PubMed] [Google Scholar]
  90. Zhong R, Ye ZH (2012) MYB46 and MYB83 bind to the SMRE sites and directly activate a suite of transcription factors and secondary wall biosynthetic genes. Plant Cell Physiol 53: 368–380 [DOI] [PubMed] [Google Scholar]
  91. Zhong R, Ye ZH (2015) Secondary cell walls: biosynthesis, patterned deposition and transcriptional regulation. Plant Cell Physiol 56: 195–214 [DOI] [PubMed] [Google Scholar]
  92. Zhou J, Lee C, Zhong R, Ye ZH (2009) MYB58 and MYB63 are transcriptional activators of the lignin biosynthetic pathway during secondary cell wall formation in Arabidopsis. Plant Cell 21: 248–266 [DOI] [PMC free article] [PubMed] [Google Scholar]
  93. Zou H, Hastie T (2012) Elastic-net for sparse estimation and sparse PCA. http://cran.r-project.org/package=elasticnet

Articles from Plant Physiology are provided here courtesy of Oxford University Press

RESOURCES