TFs (transcription factors) are modular proteins minimally containing a DBD (DNA-binding domain) and a TRD (transcription regulatory domain). NAC [for NAM (no apical meristem), ATAF, CUC (cup-shaped cotyledon)] proteins comprise one of the largest plant TF families. They are key regulators of stress perception and developmental programmes, and most share an N-terminal NAC domain. On the basis of analyses of gene expression data and the phylogeny of Arabidopsis thaliana NAC TFs we systematically decipher structural and functional specificities of the conserved NAC domains and the divergent C-termini. Nine of the ten NAC domains analysed bind a previously identified conserved DNA target sequence with a CGT[GA] core, although with different affinities. Likewise, all but one of the NAC proteins analysed is dependent on the C-terminal region for transactivational activity. In silico analyses show that the NAC TRDs contain group-specific sequence motifs and are characterized by a high degree of intrinsic disorder. Furthermore, ANAC019 was identified as a new positive regulator of ABA (abscisic acid) signalling, conferring ABA hypersensitivity when ectopically expressed in plants. Interestingly, ectopic expression of the ANAC019 DBD or TRD alone also resulted in ABA hypersensitivity. Expression of stress-responsive marker genes [COR47 (cold-responsive 47), RD29b (responsive-to-desiccation 29b) and ERD11 (early-responsive-to-dehydration 11)] were also induced by full-length and truncated ANAC019. Domain-swapping experiments were used to analyse the specificity of this function. Chimaeric proteins, where the NAC domain of ANAC019 was replaced with the analogous regions from other NAC TFs, also have the ability to positively regulate ABA signalling. In contrast, replacing the ANAC019 TRD with other TRDs abolished ANAC019-mediated ABA hypersensitivity. In conclusion, our results demonstrate that the biochemical and functional specificity of NAC TFs is associated with both the DBDs and the TRDs.

INTRODUCTION

Gene-specific TFs (transcription factors) are DNA-binding regulatory proteins which activate or repress the basal transcription apparatus at target gene promoters. TFs are grouped into families on the basis of sequence similarities, most often in the DBD (DNA-binding domain) [1]. TFs from the same family often bind DNA in a sequence-specific manner, thereby only targeting promoters with a given consensus sequence [1]. Apart from the DBD, TFs are characterized by having a TRD (transcription regulatory domain). In contrast with the sequence similarities and well-defined structures of DBDs in members of a specific TF family, TRDs have traditionally been classified according to their amino acid profile, i.e. as acidic, glutamine-, proline- or serine/threonine-rich. Moreover, TRDs often have a high degree of low-complexity sequences and a propensity for flexible protein segments that fail to self-fold into an ordered three-dimensional structure, commonly referred to as ID (intrinsic disorder) [2].

Ongoing efforts to understand the structural and functional modularity of TFs have traditionally been based on the examination of the activity of truncated versions of the proteins; an early discovery was that the DBD could be separated from the TRD with no loss of function of either module which is illustrated by the yeast GAL4 TF, a paradigm for eukaryotic transcriptional regulation studies [3]. Additionally, Johnston et al. [3] showed that expression of a cytoplasmically located version of GAL4, without the N-terminal DBD, in a wild-type background enabled induction of the expression of reporter genes to the same extent as expression of full-length GAL4. This implied that GAL4 consists of separate domains conferring nuclear localization, specific protein–protein interactions, DNA-binding and transcriptional activation [3]. This seemingly independent modularity has proven to be a general model for many TFs and has led to the question as to how functional specificity is substantiated [4,5].

Although a wealth of studies have aimed to characterize TF functionality, a few of them can be highlighted as important attempts to shed light on functional specificity from analysis of independent TF modules. In yeast, this question has been addressed by deciphering the functions of regions within the transcriptional activator Rap1p (repressor activator protein 1), showing that the C-terminal TRD has only a minor effect on the cell growth effect modulated by Rap1p [6]. Furthermore, domain-swapping experiments with combinations of elements from DBDs of Rap1p homologues from different yeasts revealed that major changes can be made to the amino acid sequence of this region without affecting the telomere-binding function of Rap1p [6]. In another study, using a series of domain-swap chimaeras in which different parts of Drosophila melanogastor T-domain TFs, encoded by the genes omb (optomotor-blind) and org-1 (optomotor-blind related-1), were mutually exchanged, Porsch et al. [7] investigated the relevance of individual domains for proper eye development. Their findings suggested that both transcriptional activation and repression properties, as well as the DNA-binding specificity, can contribute to the functional characteristics of T-domain factors [7], highlighting the complexity of parsing the functional specificity of seemingly independent TF domains.

In plants, the NAC [for NAM (no apical meristem), ATAF, CUC (cup-shaped cotyledon)] family constitutes a prominent group of TFs [8]. The genomes of Arabidopsis thaliana (thale cress), tobacco and rice all contain more than 100 genes encoding NAC TFs, making it one of the largest gene families in plants [9,10]. Genes encoding NAC TFs were originally identified from forward genetic screens as key regulators of developmental processes [8]. More recently, NAC TFs have also been shown to be involved in the regulation of stress responses in both model plants and agronomically important crops [11,12].

We have shown previously the NAC protein ANAC019 interacts with the RING-finger H2-type E3 ubiquitin-protein ligase RHA2a, and we have solved the crystal structure of the ANAC019 NAC domain [13,14]. The structure revealed a novel protein fold consisting of a twisted β-sheet packing against an α-helix on both sides. Moreover, using ANAC019 as a reference we elucidated residues important for both NAC homo- and hetero-dimerization and for binding to a core consensus DNA element of sequence CGT[GA] [15]. Studies of ANAC019 gene expression have implicated a role in different types of stress responses [14,16] and overexpression of ANAC019 caused increased drought tolerance [16]. Moreover, it has been shown that ANAC019 regulates defence responses upon fungal attack through a signal pathway related to the plant hormone jasmonic acid [17]. Recently, Bu et al. [18] verified the ANAC019–RHA2a interaction and defined RHA2a as a new positive regulator of ABA (abscisic acid) phytohormone signalling [18].

With regard to parsing functional determinants of NAC DBDs and TRDs, Taoka et al. [19] have shown that for members of the CUC subfamily of NAC TFs the DBD confers functional specificity in terms of inducing adventitious shoot formation on calli [19,20]. TRDs from three different NAC TF origins could be fused to the NAC domain of CUC2 and still induce shoot formations. In contrast, replacing the CUC NAC domains with a NAC domain from a member of the ATAF subfamily of NAC TFs did not induce shoot formation. However, the NAC domain of the CUC subfamily member CUC1 did not retain shoot-formation-ability when fused to the potent TRD from the herpes simplex virus VP16 protein. This suggested that certain structural features of the TRD, which are difficult to predict from the amino acid sequence, were necessary for the function of CUC [19].

In the present study, we have used a genome-wide approach, taking advantage of the functional diversity of the NAC proteins, to dissect structure–function aspects of NAC TF modularity. Our analyses showed that phylogenetic relationships based on NAC domain sequences are generally in accordance with the C-terminal sequence motifs of the intrinsically disordered TRDs and global NAC gene expression clusters. Additionally, our study includes examination of truncated versions of ten phylogenetically distinct NAC proteins to characterize the structural specificity of NAC DBDs and TRDs. We also use a chimaeric strategy, based on our identification of ANAC019 as a novel ABA regulator, to investigate the structure–function relationship of the ANAC019 DBD and TRD. The findings suggested multi-specificity of NAC TFs, in terms of the DNA-binding sites in target genes, and provided evidence that positive ABA-regulatory functionality is associated with both the ANAC019 TRD and DBD.

EXPERIMENTAL

Oligonucleotides and restriction enzymes

Oligonucleotides and restriction enzymes used in the present study are listed in Supplementary Table S1 (available at http://www.BiochemJ.org/bj/426/bj4260183add.htm). The DNA sequence integrity of all expression constructs was confirmed by sequencing.

Accession numbers

Sequence data from this article can be found in TAIR (The Arabidopsis Information Resource) and EMBL (European Molecular Biology Laboratory) data libraries under the following NAC nomenclature names and accession numbers: ANAC019, At1g52890; ANAC003, At1g02220; ANAC002 (ATAF1), At1g01720; VOZ2 (vascular plant one-zinc finger protein 2); At2g42400; ANAC008 [(SOG1 (suppressor of gamma response 1)], At1g25580; ANAC105 [VND3 (vascular-related NAC domain 3)], At5g66300; ANAC030 (VND7), At1g71930; ANAC092 (NAC2/ORE1), At5g39610; ANAC062 (NTL6), At3g49530; and ANAC040 (NTL8), At2g27300.

Bioinformatic analysis of the NAC family

BLASTP and PSI (position-specific iterated)-BLAST searches were performed using the NCBI (National Center for Biotechnology Information) BLAST services. Searches were performed against a non-redundant database, but were limited to include only hits within the Arabidopsis genome. All sequences were annotated according to the latest Arabidopsis genome annotation. Multiple alignments were generated using ClustalW or ClustalX [21] and BoxShade (http://www.ch.embnet.org/software/BOX_form.html) was then used for producing graphical presentations of the alignments and for manual adjustments. Phylogenetic analysis of the NAC domain sequences was performed using MEGA4 software [22]. The input was a multiple alignment of NAC domain sequences generated in ClustalX. The reliability of the different groupings of the tree was measured by bootstrapping. MEME [multiple EM (expectation maximization) for motif elicitation] was used to search for statistically significant motifs in the NAC proteins using either all NAC sequences or group- and subgroup-specific sequences as a data set. MAST (motif alignment and search tool) was used for further analysis of the NAC proteins. P values were used to determine the overall match of the sequence to the input motif. In addition, the relevance of the motifs identified was evaluated by analysis of its occurrence in control groups, e.g. the NAC domain served as a control group for motifs identified in the C-terminal regions and vice versa. Amino acid profiling was performed using Composition Profiler [23] taking advantage of the standard amino acid data set, SwissProt51, provided by the program. PONDR VL3 [24] was used for prediction of structural ID.

For global expression analysis, normalized At-TAX (tiling array express) expression data for 107 NAC genes during selected developmental stages and in response to different environmental stresses was obtained [25]. Expression values were imported into the R-software for processing and heatmap output using the Bioconductor microarray analysis packages affy, heatplus and gplots [26].

Plant materials, growth conditions and Arabidopsis transformation

Arabidopsis Col-0 (Columbia ecotype 0) wild-type, the mutant homozygotes anac019 (SALK_096303) and transgenic seeds were grown in growth chambers at 22 °C and with light intensity of 150 μmol/m2 per s from a combination of fluorescent and incandescent lamps. Transformation of Arabidopsis Col-0 plants was performed by a modified floral dip infiltration method [27] using the Agrobacterium tumefaciens strain GV3101 (pMP90) and the binary vector pCAMBIA3300 (Cambia) containing the CaMV (cauliflower mosaic virus) 35S promoter for constituent expression of transgenes and the BAR selectable marker conferring tolerance to the herbicide Basta (50 mg/l glufosinate ammonium). Transgenic plants were selected by Basta spraying and T3 seeds from transformants expressing transgenes were used for subsequent analyses. An empty pCAMBIA3300 vector was transformed into Col-0 wild-type plants as a control.

ABA sensitivity assay

To determine the effect of ABA on germination and post-germination growth efficiency, surface-sterilized and stratified (3 days) seeds of Arabidopsis Col-0 wild-type, anac019 mutant alleles (SALK_096303) and transgenic lines expressing full-length, truncated or chimaeric constructs of ANAC019, were grown on half-strength Murashige and Skoog medium supplemented with 0–5 μM ABA (mixed isomers; Sigma–Aldrich). A seed was considered as germinated when the radicle penetrated the seed coat. The percentage of germinated seeds was scored at 3 days after stratification in three independent experiments (40 seeds/experiment per treatment). Post-germination growth efficiency was scored as primary root length at 7 days after stratification. Seeds from Col-0 wild-type, anac019 mutants and transgenic lines used for phenotypic analysis were harvested and collected at the same time. Root elongation assays were performed as described in [28] and significance levels were assessed by a two-sided t-test.

Production of recombinant proteins and in planta expression

For production of recombinant NAC proteins, cDNA clones obtained from the ABRC (Arabidopsis Biological Resource Center) and were amplified by PCR to obtain the complete coding sequence or the region encoding the NAC domain of ANAC019, ATAF1, NAC2/ORE1, VOZ2, VND3, VND7, NTL8, SOG1, NTL6 and ANAC003 using the primer sets listed in Supplementary Table S1. The PCR products were inserted into pENTR/D-TOPO (Invitrogen) followed by recombination into pDEST-15 (Invitrogen) to obtain GST (glutathione transferease)-tagged NAC recombinant proteins. The recombinant proteins were named to show the borders of the NAC domain fragment, e.g. GST–ANAC019-(1–168), which consists of GST fused at the N-terminus of the 168 N-terminal residues of ANAC019. Escherichia coli strain BL21(DE3) was transformed with the expression constructs and used for expression of GST–NAC recombinant proteins. Transformed E. coli cultures were grown to a cell density (D600) of 0.6 in Luria–Bertani medium before induction with 0.5 mM IPTG (isopropyl β-D-thiogalactoside). After 3 h of incubation at 37 °C the cells were harvested by centrifugation at 10000 g for 20 min. The cell pellet was resuspended in PBS, pH 7.5, containing 1 mM DTT (dithiothreitol), and was incubated on ice for 30 min before sonication using a Soniprep 150 instrument (MSE) at amplitude 7 μ for 30 s, pausing for 1 min, and then repeating the process three times. Following sonication, the lysate was cleared by centrifugation at 10000 g for 20 min. The GST–NAC fusion proteins were purified by affinity chromatography using a 2 ml glutathione–Sepharose 4B column (GE Healthcare) with 250 ml of E. coli culture (at the time of harvest D600 was approx. 3.0–4.0). The column was washed with ten column volumes of PBS, pH 7.5, containing 1 mM DTT. The protein was eluted in 1-ml fractions in a buffer of 50 mM Tris/HCl, pH 8.0, containing 10 mM GSH. If required, the proteins were purified further by anion-exchange chromatography. The protein solutions were dialysed overnight against a buffer of 20 mM Tris/HCl, pH 8.0, containing 1 mM DTT and loaded on to a 1-ml RESOURCE Q column (GE Healthcare) and eluted using a gradient of NaCl.

The vector pCAMBIA3300 and the primers and restriction enzymes listed in Supplementary Table S1 were used for generation of chimaeric constructs for in planta expression. Full-length ANAC019 (residues 1–317) was amplified using a cDNA clone obtained from the ABRC. Chimaeric constructs for domain-swapping experiments that contained the ANAC019 NAC domain and divergent TRDs were constructed. First, the DNA fragment corresponding to the ANAC019 NAC domain (residues 1–168 of ANAC019) was PCR amplified using a full-length cDNA clone from ABRC; for overlap extension PCR with fragments encoding the ANAC019 NAC domain, the reverse primers included a 12-bp extension at their 5′-termini. Then DNA fragments encoding the TRDs [NAC2/ORE1-(175–285), NTL8-(158–335) and SOG1-(216–449)] were PCR amplified using cDNA clones from ABRC; for overlap extension PCR, forward primers for DNA amplification of sequences encoding TRDs, included a 12-bp extension in their 5′-termini. Upon successful generation of the individual fragments, the two amplification products were mixed and amplified using the following PCR program: four cycles of 94 °C for 3 min, 53 °C for 30 s and 72 °C for 3 min. The PCR was performed with 2.5 units of Pfu DNA polymerase (Fermentas), 0.2 mM dNTPs and approx. 150 ng of individual DNA fragments in a 100 μl reaction volume. Immediately afterwards, 0.3 μM external primers were added and the constructed insert was amplified using the following PCR program: 95 °C for 2 min; 30 cycles of 94 °C for 1 min, 52 °C for 1 min, 72 °C for 2 min; 72 °C of 15 min; and holding at 4 °C. Chimaeric constructs that contained the ANAC019 TRD and divergent NAC domains were constructed as described above, amplifying the NAC domains instead of the TRDs [NAC2/ORE1-(1–174), NTL8-(1–157) and SOG1-(1–215)], and the TRD of ANAC019 (residues 169–317) instead of the ANAC019 NAC domain.

EMSA (electrophoretic mobility-shift assay)

A double-stranded oligonucleotide containing an optimized palindromic version of the NAC-BS (binding site) selected by CASTing (cyclic amplification and selection of targets) [15] was prepared by annealing complementary single-stranded oligonucleotides of the sequence 5′-CAGTCTTGCGTGTTGGAACACGCAACAGTC-3′. The double-stranded oligonucleotide was labelled at the 3′ end with DIG (digoxigenin) using a DIG gel-shift kit (second generation; Roche). DNA binding reactions contained 4 μl of 5× binding buffer {100 mM Hepes, pH 7.6, containing 5 mM EDTA, 50 mM (NH4)2SO4, 5 mM DTT, 1% (w/v) Tween-20, 150 mM KCl, 1 μg of poly[d(I-C)]}, 31 fmol of DIG-labelled DNA and the indicated amount of GST–NAC domain protein in a final volume of 20 μl. The binding reactions were incubated for 15 min at room temperature (20 °C) and separated by PAGE (10% PAGE gold precast gels; Cambrex). Electroblotting was performed using a Hybond-N nylon membrane (Amersham Biosciences) and chemiluminescent detection was carried out according to the manufacturer's instructions.

Transactivation analysis in yeast

NAC proteins were examined for the presence of an activation domain using a yeast-based reporter assay. Full-length or NAC-domain-truncated coding sequences of the ten NAC candidates were cloned into the DBD vector pGBKT7 (Clontech). Primers and restriction enzymes used are listed in Supplementary Table S1. Subsequently, constructs encoding the GAL4 DBD fused to full-length or NAC-domain-truncated versions of NAC protein coding sequences were transformed into the yeast strain pJ694A, which harbours the HIS3 and ADE2 reporter genes. The transformed yeast cultures were plated on to SD (synthetic defined) plates without tryptophan (−Trp) and SD plates without tryptophan, histidine and adenosine (−Trp/−His/−Ade) for 7 days at 30 °C before inspection of transactivation properties of the reporter constructs. For easy assessment of transactivational activity, overnight cultures of transformants were diluted to a D600 of 0.6, and then 10-fold serial dilutions were prepared. Finally, 5 μl of each dilution was spotted on to SD −Trp plates and SD −Trp/−His/−Ade plates, which were then incubated at 30 °C for another 7 days. Constructs pVA3–1 (containing the p53 DBD) and pTD1–1 (containing the T-antigen activation domain) were used as positive controls. Yeast cells transformed with pGBKT7 vector alone cannot activate the GAL1 promoter and thus served as a negative control.

RNA extraction, cDNA synthesis and quantitative real-time RT (reverse transcription)–PCR analysis

For Arabidopsis transcript analysis, total RNA was isolated from three complete rosettes for each genotype, using the RNeasy mini kit (Qiagen). Purified DNase-treated RNA (1 μg; Promega) was used for cDNA synthesis using the Superscript III cDNA synthesis kit (Invitrogen) according to the manufacturer's instructions. Quantitative real-time PCRs were performed in triplicate for each individual line, and quantification of CT (cycle threshold) values was achieved by calculating means of normalized expression using Q-gene software [29]. Actin2 was used as a reference to determine the relative expression of the ABA-responsive genes COR47 (cold-responsive 47; At1g20440) and RD29b (responsive-to-desiccation 29b; At5g52300), and the stress-responsive genes FER1 (ferritin precursor 1; At5g01600) and ERD11 (early-responsive-to-dehydration 11; At1g02930). Experiments were repeated three times with similar results. Primers for quantitative real-time PCR are listed in Supplementary Table S1.

RESULTS

The Arabidopsis NAC family

To uncover the expansion of the Arabidopsis NAC TF family we searched for NAC sequences using BLASTP in latest genome annotation, the translated Arabidopsis genome, using the NAC domain of ANAC019 [14] as a query. This approach identified 92 non-redundant Arabidopsis proteins, and 18 additional NAC proteins were uncovered in an iterative PSI-BLAST search. A multiple ClustalX alignment of the 110 sequences revealed one sequence with only partial similarity to the NAC domain consensus, leaving 109 NAC sequences (see Supplementary Table S2 available at http://www.BiochemJ.org/bj/426/bj4260183add.htm). To study the Arabidopsis NAC TF family in more detail, a phylogenetic tree was constructed with the conserved NAC domains in MEGA4 [22], using a neighbour-joining method and bootstrap analysis to assess branch reliability (Figure 1). As MEGA4 was unable to construct a tree based on alignments including At1g60240, this sequence was left out. The phylogeny of the 108 Arabidopsis NAC is presented in Figure 1 and an alignment of selected NAC domains is presented in Figure 2(A). On the basis of the phylogenetic analysis, the Arabidopsis NAC family was divided into ten major polytomies by collapsing branches with bootstraps <40%. Aside from the ten major groupings, several minor groupings were observed. To keep the Arabidopsis NAC nomenclature consistent, names from previous nomenclature systems [9,30] are maintained when possible in this updated nomenclature (see also Supplementary Table S2).

Phylogenetic and expression analyses of the Arabidopsis NAC family

Figure 1
Phylogenetic and expression analyses of the Arabidopsis NAC family

Left-hand panel: phylogenetic presentation of the NAC TF family. The analysis is based on 108 NAC domain protein sequences and the tree was constructed in MEGA4 [22]. Individual NAC proteins are listed according to their TAIR9 annotation, ANAC numbering is as suggested by Ooka et al. [9] and, when possible, their trivial name (for references see Supplementary Table S2). Middle and right-hand panels: heatmap presentation of normalized At-TAX expression data for 107 NAC genes during six different developmental stages and in response to six different stress treatments. Averages of three replicate samples were used for all developmental stages and treatments analysed. For the stress samples, expression profiles are presented relative to the appropriate mock-treated samples. The colour key at the bottom displays correlation between colour and scaled log2 fold changes. Values below mean (developmental stages) or mock sample values (stress data set) are coloured red and those above mean or mock sample values are coloured white. Mean values and genes unchanged in their expression relative to the mock control are coloured orange. NAC TFs examined in more detail in the present study are highlighted (*).

Figure 1
Phylogenetic and expression analyses of the Arabidopsis NAC family

Left-hand panel: phylogenetic presentation of the NAC TF family. The analysis is based on 108 NAC domain protein sequences and the tree was constructed in MEGA4 [22]. Individual NAC proteins are listed according to their TAIR9 annotation, ANAC numbering is as suggested by Ooka et al. [9] and, when possible, their trivial name (for references see Supplementary Table S2). Middle and right-hand panels: heatmap presentation of normalized At-TAX expression data for 107 NAC genes during six different developmental stages and in response to six different stress treatments. Averages of three replicate samples were used for all developmental stages and treatments analysed. For the stress samples, expression profiles are presented relative to the appropriate mock-treated samples. The colour key at the bottom displays correlation between colour and scaled log2 fold changes. Values below mean (developmental stages) or mock sample values (stress data set) are coloured red and those above mean or mock sample values are coloured white. Mean values and genes unchanged in their expression relative to the mock control are coloured orange. NAC TFs examined in more detail in the present study are highlighted (*).

Structural features of NAC proteins

Figure 2
Structural features of NAC proteins

(A) Alignment of the NAC domain sequence of the ten Arabidopsis NAC proteins analysed in the present study. Residues surrounded by black are common to at least half of the sequences, whereas residues surrounded by grey are chemically similar in more than half of the sequences or similar to the dominating residue. (*) denotes residues forming a salt bridge stabilising the dimerization interface (underlined) [13]. (#) denotes residues in a region containing several highly conserved residues of importance to DNA binding [15]. The secondary structure elements, α-helix (α) and β-strands (β), of the ANAC019 NAC domain are indicated above the alignment. The element enclosed in parentheses was only present in one of the monomers of the NAC domain dimer [13]. The dominating sequence motifs identified by MEME/MAST analysis [49,50] are shown by the letters A–E. The consensus sequences of the motifs are shown in Supplementary Table S2. Proteins are labelled by their trivial name or ANAC numbering (for references see Supplementary Table S2). (B) WebLogos of MEME/MAST sequence motifs for motif I, found in the N-terminal extension of group IX-1 proteins, and representative sequence motifs from NAC TRDs (for motif assignments to specific groups, subgroups and proteins, see Supplementary Table S2). The height of a letter in the WebLogo indicates its relative frequency at the given position (x-axis) in the motif. (C) Schematic domain structures of typical NAC proteins, N-terminally extended subgroup IX-1 NAC proteins, VOZ proteins with the zinc finger indicated (Zinc F), and tandem NAC domain NACs. Black bars indicate NAC domain boundaries and black arrowheads indicate intron positions. The figure is not drawn to scale. (D) The structure of the ANAC019 NAC domain homodimer (PDB accession code 1UT7) is shown in ribbon format with different colours (green and magenta) for each monomer. The residues (Arg-19 and Glu-26) of the dimer interface (encircled) and Arg-88 of the probable DNA-binding motif (boxed) are shown as sticks with nitrogen and oxygen atoms shown in blue and red respectively. Residues Lys-79 to Arg-85 were not traced in the NAC019 crystal structure [13].

Figure 2
Structural features of NAC proteins

(A) Alignment of the NAC domain sequence of the ten Arabidopsis NAC proteins analysed in the present study. Residues surrounded by black are common to at least half of the sequences, whereas residues surrounded by grey are chemically similar in more than half of the sequences or similar to the dominating residue. (*) denotes residues forming a salt bridge stabilising the dimerization interface (underlined) [13]. (#) denotes residues in a region containing several highly conserved residues of importance to DNA binding [15]. The secondary structure elements, α-helix (α) and β-strands (β), of the ANAC019 NAC domain are indicated above the alignment. The element enclosed in parentheses was only present in one of the monomers of the NAC domain dimer [13]. The dominating sequence motifs identified by MEME/MAST analysis [49,50] are shown by the letters A–E. The consensus sequences of the motifs are shown in Supplementary Table S2. Proteins are labelled by their trivial name or ANAC numbering (for references see Supplementary Table S2). (B) WebLogos of MEME/MAST sequence motifs for motif I, found in the N-terminal extension of group IX-1 proteins, and representative sequence motifs from NAC TRDs (for motif assignments to specific groups, subgroups and proteins, see Supplementary Table S2). The height of a letter in the WebLogo indicates its relative frequency at the given position (x-axis) in the motif. (C) Schematic domain structures of typical NAC proteins, N-terminally extended subgroup IX-1 NAC proteins, VOZ proteins with the zinc finger indicated (Zinc F), and tandem NAC domain NACs. Black bars indicate NAC domain boundaries and black arrowheads indicate intron positions. The figure is not drawn to scale. (D) The structure of the ANAC019 NAC domain homodimer (PDB accession code 1UT7) is shown in ribbon format with different colours (green and magenta) for each monomer. The residues (Arg-19 and Glu-26) of the dimer interface (encircled) and Arg-88 of the probable DNA-binding motif (boxed) are shown as sticks with nitrogen and oxygen atoms shown in blue and red respectively. Residues Lys-79 to Arg-85 were not traced in the NAC019 crystal structure [13].

Global NAC gene expression analysis

Knowledge of NAC gene expression patterns has contributed significantly to our understanding of NAC gene involvement in plant stress tolerance and development [25]. To provide a global overview of the NAC transcriptome in relation to different developmental stages and environmental stresses we analysed the Arabidopsis gene expression data accessible online at the At-TAX website (http://www.weigelworld.org/resources/microarray/at-tax) [25]. In contrast with the limited number of NAC gene probe sets available from the widely used ATH1 GeneChip, At-TAX expression data is available for almost all of the 108 NAC genes listed in Figure 1 (locus At1g64105 overlaps At1g64100 and has no unique probes), and therefore provided the basis for the expression analysis.

The At-TAX expression profiles showed that, overall, NAC gene expression is detected in all selected tissues and developmental stages (Figure 1). Likewise, NAC transcript levels are widely affected by a range of environmental stresses. Additionally, without posing any clustering of the expression data, minor groupings with similar expression patterns correlate with the phylogenetic analysis based upon the NAC domain sequences. This is particularly evident for the tissue-specific expression patterns, e.g. a fraction of the VND subgroup (II-1) of NAC genes are predominantly expressed in both stems and roots, whereas others are expressed only in roots or stems. These data support previous studies of VND members which identified them as important regulators in transdifferentiation of various cells into vessel elements [31,32]. Another prominent subgroup is the ATAF (III-3) group, which shows the most abundant co-expression levels in old leaves and roots. This is also the case for the more distantly related subfamily VII-2 and for members of the transmembrane-anchored subgroups I-1 and I-2. Moreover, gene members of subgroup III-3 are highly salt-inducible, whereas those of subfamily I-1 and VII-2 mainly respond to elevated temperatures. Finally, the co-expression of the five NAC genes encoding two NAC domains (VIII-1) in whole inflorescence and fruits deserves to be highlighted as a yet-to-be characterized subgroup with a distinct co-expression pattern correlating with phylogenetic distribution. Overall, among the displayed stress data sets, factors affecting the expression of most NAC genes are salt and extreme temperatures. With respect to the latter, most NAC genes shown to be induced by cold are repressed by heat and vice versa. In conclusion, several phylogenetically related NAC genes are co-expressed in a developmental/tissue-specific, and to a lesser extent stress-specific, manner.

Structural characterization and classification of the NAC proteins

To further analyse the structural diversity of Arabidopsis NAC proteins, sequence motifs were predicted using the programs MEME and MAST. This identified both shared and group-specific sequence motifs in the NAC domains (Figures 2A and 2B; Supplementary Table S2). The major motifs mapped to the α-helices and β-strands of the NAC domain [13]. Whereas most of the NAC domains contained motifs we designated as A–F, domains from groups VI–X shared fewer motifs with the rest of the NAC domains, and some had subgroup specific motifs, such as motif R of subgroup IX-1 (Figure 2B).

The divergent C-terminal regions of NAC proteins, for which no tertiary structure information is available, were also examined for sequence motifs (Figure 2B and Supplementary Table S2). Subgroup II-1 of NST/VND proteins shared motifs WQ and LP, of which WQ has been shown to be important for transactivational activity [33]. Three motifs, LP (also named L; [19]), V and W, dominate the C-termini of the CUC-like proteins, which includes NAC2/ORE1 [34], of subgroup II-3. Motif W was reported to be of importance for transcriptional activity [19,33]. Thus most of the C-termini of group II proteins share motif LP and probably depend on a motif with a prominent tryptophan for activity (Figure 2B). Additional motifs, such as low-complexity histidine- and glutamine-rich Q, U and T motifs were identified in some group II proteins. Group I includes most of the fourteen NAC proteins with a C-terminal transmembrane region [8,35]. For these proteins, the majority of the C-terminal motifs (e.g. l, Y, b, K and k) are dominated by polar and negatively charged residues with conserved hydrophobic residues embedded in the polar matrix (e.g. isoleucine and leucine residues in motif l and tryptophan and phenylalanine residues in motif K). Subgroup III-3 contains the stress-related proteins, ANAC019, ANAC055, ANAC072, ATAF1 and ATAF2, which also have motifs found only in their subgroup (o, g, h, n and i). These motifs, and motif O of subgroup V-II, are also dominated by a negatively charged matrix with a few conserved bulky and hydrophobic amino acid residues (Figure 2B; Supplementary Table S2).

The groups discussed above represent the most highly studied NAC proteins, and comply with the classical modular outline presented in Figure 2(C). In contrast, the proteins in subgroup VIII-1 consist of two tandemly repeated NAC domains (Figure 2C) and have genes without annotated introns. Subgroup IX-1 forms a distinct clade, as supported by high bootstrap values, and contains the recently characterized SOG1 gene [12]. The seven members of this subgroup deviate from the characteristic NAC structure by having N-terminal extensions of approx. 40 amino acid residues with a pair of conserved cysteine residues (motif R; Figures 2B and 2C). Some of these proteins also have typical, but specific C-terminal motifs (motifs d and e). In the phylogenetic tree, subgroup IX-1 is closely related to the VOZ proteins (subgroup VIII-2 [36]), which were identified as NAC proteins in PSI-BLAST searches [15]. The VOZ proteins have an N-terminal region of approx. 240 residues followed by a conserved DBD, comprising of a zinc co-ordinating motif and a C-terminal basic region with structural similarity to the NAC domain (Figures 2A and 2C). In the N-terminal domain, the VOZ proteins contain motifs a and d which are chemically similar to characteristic C-terminal NAC motifs. In conclusion, some subgroups have specific motifs in the NAC domain, reflecting their divergence from the core NAC domain. Moreover, C-termini of NAC proteins contain several group-specific motifs, which are enriched in polar and acidic residues.

Selection of representative NAC proteins

Most studies of NAC family members have focused on members from groups I–III, which have typical NAC domains that bind DNA and a C-terminal TRD (Figure 2C) [15,16,19]. To understand the modular structures of NAC TFs with respect to functionality, ten NAC proteins, representing functionally important clades and spanning the phylogenetic diversity of the NAC family (Figure 1), were selected for characterization. On the basis of the extensive characterization of ANAC019 [13,1517], we decided to use this protein as a reference in the comparative studies. The remaining nine NAC proteins were ATAF1, VND3, VND7, ANAC003, NAC2/ORE1, SOG1, VOZ2, NTL6 and NTL8 (Figures 1, 2A and 2B).

In vitro DNA-binding of structurally variant NAC domains

In our previous study we determined a consensus NAC-BS with the core sequence CGT[GA] using a reiterative selection procedure on random oligonucleotides and ANAC019 and NAC2/ORE1 as baits [15]. A CGT[GA] or CGT core consensus sequence has also been identified as a DNA-binding sites for other NAC TFs from both Arabidopsis and wheat [15,16,37]. To examine the in vitro DNA-binding properties of selected NAC proteins, N-terminally GST-tagged NAC domains were produced heterologously. All proteins were purified by affinity chromatography and some were purified further by ion-exchange chromatography. The quality of the recombinant proteins was examined by SDS/PAGE. The GST–NAC domain fusion proteins had the expected molecular mass values (of between 47 and 54.1 kDa) and only minor contaminants (Figure 3A).

Analysis of NAC domain DNA binding using EMSAs

Figure 3
Analysis of NAC domain DNA binding using EMSAs

(A) SDS/PAGE analysis and Coomassie Blue staining of: gel molecular-mass-markers (lane 1; molecular masses in kDa are shown on the left); approx. 5 μg of affinity-purified recombinant GST–ANAC019-(1–168) (lane 2); GST–ATAF1-(1–165) (lane 3); GST–NAC2/ORE1-(1–176) (lane 4); GST–VOZ2-(205–420) (lane 5); GST–SOG1-(1–220) (lane 6); GST–ANAC003-(1–152) (lane 7); GST–NTL8-(1–157) (lane 8); GST–VND7-(1–163) (lane 9); GST–VND3-(1–165) (lane 10); and GST–NTL6-(1–168) (lane 11). (B) MEME WebLogos for the NAC-binding sequence selected for ANAC019 (top panel) and NAC2/ORE1 (middle panel). The selected sequences were used to design the palindromic NAC-BS, pal-NAC-BS (bottom panel), which was used for EMSA-based analysis of NAC DNA-binding [15]. (C) Representative EMSA using oligonucleotide palNAC-BS and 0 (lane 1), 5 ng (lane 2), 50 ng (lane 3) or 500 ng (lane 4) of the recombinant proteins indicated. All experiments were repeated at least three times with new preparations of the recombinant proteins, showing similar results with the major high-molecular-mass band corresponding in size to the palNAC-BS–GST–ANAC019-(1–168) complex.

Figure 3
Analysis of NAC domain DNA binding using EMSAs

(A) SDS/PAGE analysis and Coomassie Blue staining of: gel molecular-mass-markers (lane 1; molecular masses in kDa are shown on the left); approx. 5 μg of affinity-purified recombinant GST–ANAC019-(1–168) (lane 2); GST–ATAF1-(1–165) (lane 3); GST–NAC2/ORE1-(1–176) (lane 4); GST–VOZ2-(205–420) (lane 5); GST–SOG1-(1–220) (lane 6); GST–ANAC003-(1–152) (lane 7); GST–NTL8-(1–157) (lane 8); GST–VND7-(1–163) (lane 9); GST–VND3-(1–165) (lane 10); and GST–NTL6-(1–168) (lane 11). (B) MEME WebLogos for the NAC-binding sequence selected for ANAC019 (top panel) and NAC2/ORE1 (middle panel). The selected sequences were used to design the palindromic NAC-BS, pal-NAC-BS (bottom panel), which was used for EMSA-based analysis of NAC DNA-binding [15]. (C) Representative EMSA using oligonucleotide palNAC-BS and 0 (lane 1), 5 ng (lane 2), 50 ng (lane 3) or 500 ng (lane 4) of the recombinant proteins indicated. All experiments were repeated at least three times with new preparations of the recombinant proteins, showing similar results with the major high-molecular-mass band corresponding in size to the palNAC-BS–GST–ANAC019-(1–168) complex.

Using the consensus NAC-BS, a palindromic NAC-BS, palNAC-BS, was designed for EMSA-based analysis of the DNA-binding of the NAC domain (Figure 3B) [15]. The palNAC-BS was used to examine the ability of GST–NAC proteins to bind DNA in a titration series using increasing amounts (0–500 ng) of GST–NAC with GST alone as control (Figure 3C). Preliminary experiments showed that GST–NAC domain proteins bound palNAC-BS with the same affinity as the corresponding NAC domain without a GST tag and that full-length NAC2/ORE1 bound palNAC-BS with the same affinity as the NAC2/ORE1 NAC domain (results not shown). For each protein a minimum of three EMSAs were performed, each time with an independent protein batch, and these all showed the same binding pattern. As expected, GST–ANAC019-(1–168) and GST–NAC2/ORE1-(1–165) bound palNAC-BS, and binding could be detected using 5 ng of the recombinant proteins. GST–ATAF1-(1–165) also bound palNAC-BS, with an affinity similar to that of ANAC019 (measured as the amount of protein needed to produced a shifted band). This is in accordance with the close relationship between ANAC019 and ATAF1 (Figure 1). In addition to the major band formed in the presence of GST–NAC and palNAC-BS, we sometimes observed a minor band, most often of slower migration (Figure 3C). This may represent non-native protein–protein interactions and has been observed in studies of other DNA-binding proteins [38]. A similar binding pattern was also seen for GST–VND7-(1–163), whereas the closely related GST–VND3-(1–165) bound with lower affinity. To our knowledge, the in vitro DNA-binding properties of proteins from the well-studied subgroup II-1 of VND proteins have not been analysed previously. GST–VOZ2-(205–420) also bound palNAC-BS, although with lower affinity than GST–ANAC019-(1–168), which was also the case for both GST–NTL8-(1–157) and GST–NTL6-(1–168).

So far neither ANAC003 nor SOG1, or members of their clades, have been characterized with respect to their biochemical properties and their sequences show a relatively low degree of similarity to the typical NAC domains. They both have substitutions in some of the positions corresponding to the residues Lys-79, Arg-85 and Arg-88 in ANAC019 (Figure 2A), which are of known importance to DNA-binding [15]. GST–SOG1-(1–220) showed a weak affinity for palNAC-BS. In contrast, attempts to show binding of GST–ANAC003-(1–152) to palNAC-BS in EMSAs were unsuccessful. In addition, the recombinant protein has no detectable binding to a 30-bp fragment of the CaMV 35S promoter (positions −84 to −55), which is bound by several NAC proteins (results not shown) [39,40], and attempts to select binding sequences for GST–ANAC003-(1–152) using CASTing were also unsuccessful. In conclusion, nine of the ten NAC domains examined bound palNAC-BS, although with differing affinities which generally reflected their structural relations. No DNA-binding ability was detected for ANAC003.

C-terminal regions of NAC proteins are intrinsically disordered

The C-terminal regions of NAC TFs were analysed further using the Composition Profiler software [23], which analyses the amino acid composition of a protein. This showed an under-representation of hydrophobic and structure-stabilizing amino acid residues and an over-representation of polar and negatively charged residues in the C-termini compared with the frequency in the reference dataset, SwissProt51 (Figure 4A). The only frequent basic residue in the C-termini is histidine as reflected by the histidine-rich motifs Q and U (Figure 2B). The N-terminal regions of the VOZ proteins have similar profiles (see Figure 2C for a schematic comparison of the canonical NAC modularity against the VOZ modularity). In contrast, the NAC domains have a higher frequency of basic and aromatic residues (Figure 4A).

Structural characteristics and predictions of NAC domains and adjacent regions

Figure 4
Structural characteristics and predictions of NAC domains and adjacent regions

(A) Amino acid composition profiles of NAC C-terminal domains (not including C-termini of subgroups VIII-1 and VIII-2 proteins; left-hand panel), VOZ N-terminal domains (middle panel) and NAC domains (right-hand panel) compared with the reference data set, SwissProt51. (B) From left to right a PONDR VL3 [24] analysis of NAC2/ORE1, representing group II NAC proteins, NTL8, representing subgroup I proteins, ANAC019 representing group III proteins and SOG1, representing group IX proteins in shown. These proteins were used for construction of the chimaeric proteins (see Figure 7). A threshold is applied with disorder assigned to values greater than or equal to 0.5 as indicated by the black bar. The position of the NAC domain is shown by a broken bar. Positions of various motifs are indicated.

Figure 4
Structural characteristics and predictions of NAC domains and adjacent regions

(A) Amino acid composition profiles of NAC C-terminal domains (not including C-termini of subgroups VIII-1 and VIII-2 proteins; left-hand panel), VOZ N-terminal domains (middle panel) and NAC domains (right-hand panel) compared with the reference data set, SwissProt51. (B) From left to right a PONDR VL3 [24] analysis of NAC2/ORE1, representing group II NAC proteins, NTL8, representing subgroup I proteins, ANAC019 representing group III proteins and SOG1, representing group IX proteins in shown. These proteins were used for construction of the chimaeric proteins (see Figure 7). A threshold is applied with disorder assigned to values greater than or equal to 0.5 as indicated by the black bar. The position of the NAC domain is shown by a broken bar. Positions of various motifs are indicated.

The compositional profile of the NAC TF C-termini is suggestive of ID [2]. For this reason, all NAC proteins were examined for disorder using the predictor PONDR VL3 [24]. This analysis confirmed our assumptions and showed that the C-termini have a large degree of ID, and that they were more highly disordered than the NAC domains (Figure 4B). The analysis was supported by a circular dichroism analysis of the C-terminal region of a barley NAC protein indicating that the region is largely unfolded (T. Kjaersgaard and K. Skriver, unpublished work). For some proteins, a subgroup-specific conserved motif is located in a dip in the disorder prediction, as exemplified by motif l of the subgroup I-4 proteins, motif i of some subgroup III-3 proteins and motif WQ of the VND subgroup II-1 proteins. In other cases, the motif is located in similar positions in the border between ordered and disordered regions (e.g. motif LP of subgroup II-3, motif K of subgroup I-2 and motif N of subgroup VII-2 proteins) (Figure 4B and results not shown). Such motifs may represent MoRFs (molecular recognition features) [41] with a propensity for target-induced folding. In conclusion, in silico analysis of the NAC protein C-termini showed that they are enriched in polar and acidic residues and have a large degree of ID.

Divergent NAC transcription regulatory domains

In several cases NAC proteins have been reported to have a C-terminal TRD [5,42]. This correlates with the high degree of ID in the NAC C-termini (Figure 4B), which is characteristic of TRDs [2]. To provide a systematic characterization of NAC TRDs we examined the transactivational activity of full-length and N-terminally truncated versions of the candidate NAC proteins. For this purpose fragments encoding full-length and truncated versions of the ten NAC proteins were fused in-frame to the GAL4 DBD in the yeast expression vector pGBKT7. The GAL4-(DBD)–NAC-(TRD) fusion proteins were subsequently expressed in a yeast strain containing the HIS3 and ADE2 reporter genes.

The transactivation analysis of full-length and truncated versions of ANAC019 confirmed our previous findings, identifying the C-terminal of ANAC019 as harbouring a transactivation domain (Figure 5A) [14]. However, and in accordance with Bu et al. [17], full-length ANAC019 did not show any transactivation potential in this assay, though Tran et al. [16] have shown that ANAC019 transactivates expression in a plant-based reporter assay. Interestingly, removing the extreme N-terminal region, ANAC019-(1–40), improved the transactivational potential (Figure 5A), a result reminiscent of NAC1 transactivational activity [5]. Taken together, these results showed that the ANAC019 C-terminal region has transactivational activity in yeast.

NAC transcription factors are transcriptional activators

Figure 5
NAC transcription factors are transcriptional activators

(A) Fusion proteins of Gal4-(DBD)–ANAC019-(1–317), Gal4-(DBD)–ANAC019-(1–168), Gal4-(DBD)–ANAC019-(41–317) and Gal4-(DBD)–ANAC019-(169–317) were expressed in yeast and screened after seven days for their transactivation activity of the HIS3 and ADE2 reporter genes. (B) Full-length and truncated versions of ten NAC proteins were fused to the GAL4-(DBD) and analysed for their transactivation activity for HIS3 and ADE2 7 days after yeast transformation. pVA3–1, pTD1–1 is a positive control; empty pGBKT7 and Gal4-(DBD)-ANAC019-(1–168) are negative controls.

Figure 5
NAC transcription factors are transcriptional activators

(A) Fusion proteins of Gal4-(DBD)–ANAC019-(1–317), Gal4-(DBD)–ANAC019-(1–168), Gal4-(DBD)–ANAC019-(41–317) and Gal4-(DBD)–ANAC019-(169–317) were expressed in yeast and screened after seven days for their transactivation activity of the HIS3 and ADE2 reporter genes. (B) Full-length and truncated versions of ten NAC proteins were fused to the GAL4-(DBD) and analysed for their transactivation activity for HIS3 and ADE2 7 days after yeast transformation. pVA3–1, pTD1–1 is a positive control; empty pGBKT7 and Gal4-(DBD)-ANAC019-(1–168) are negative controls.

Secondly, full-length VND7, NTL8, ATAF1 and VOZ2 have all been shown to activate transcription in yeast or plant-based reporter assays [32,36,43,44]. Furthermore, it has been shown that VND3 activates transcription through heterodimerization with VND7 [32], and that the TRDs of NTL8 and ATAF1 are located in the C-terminal part of the protein [43,44]. The reporter gene assay in the present study corroborates the findings on the full-length and C-terminal constructs analysed previously, with the exception of full-length NTL8 where the construct did not show transactivation, in contrast with observations by Kim et al. [43] (Figure 5B). However, apart from ANAC019 and NTL8, all full-length constructs were able to transactivate expression of the HIS3 and ADE2 reporter genes. Additionally, all the N-terminal truncation constructs analysed showed transactivational activity, apart from the non-canonical NAC transcription factor VOZ2, which was dependent on its N-terminal to confer transactivation (Figure 5B). This protein has its transactivating activity in the N-terminus (see Figure 2C). In conclusion, NAC domains are dispensable for conferring a transactivational potential, and the TRD usually resides in the C-terminal part of the protein.

ANAC019 regulates ABA signalling during seedling development

Expression of the ANAC019 gene is induced by drought, salt and ABA (Figure 1; [14,16]). ABA plays a key role in seed germination and early seedling development, as well as during plant drought and salt-stress perception [45]. To investigate the functional relationship between ABA and ANAC019, the ABA-dose response of Col-0 wild-type plants was compared with that of transgenic plants overexpressing full-length ANAC019-(1–317), ANAC019-(169–317) or ANAC019-(1–168) and to anac019 mutant plants using quantitative bioassays of germination (radicle emergence) and seedling development (Figures 6A and 6B). Ectopic expression of transgenes was verified using quantitative real-time RT–PCR (see Supplementary Figure S1 available at http://www.BiochemJ.org/bj/426/bj4260183add.htm). In the absence of exogenous ABA, all genotypes showed similar germination by day 3, except anac019 mutant plants which showed a significant decrease in the ability to germinate (P<0.05) (Figure 6A). In the presence of low ABA concentrations ANAC019-(1–317) and ANAC019-(1–168) showed lower germination efficiency compared with wild-type and anac019 mutant plants. At 2 μM ABA plants expressing either full-length or truncated versions of ANAC019 displayed ABA hypersensitivity compared with wild-type and especially anac019 mutant plants, and at 5 μM ABA only anac019 mutant seeds germinated (Figure 6A). During early seedling development, primary root length was scored at 7 days after stratification in the absence or presence of different concentrations of ABA (Figure 6B). At low ABA concentrations, plants expressing either ANAC019-(1–317) or ANAC019-(1–168) were more sensitive than the wild-type and anac019 mutant plants, as shown by shortened primary root elongation (0.05<P<0.001). At 2 μM ABA, plants expressing ANAC019-(1–317) and ANAC019-(169–317), but not ANAC019-(1–168), were significantly hypersensitive to ABA compared with wild-type and anac019 mutant plants (P<0.01). In summary, ANAC019-(1–317) and truncated versions of ANAC019 displayed ABA hypersensitivity during germination and early seedling development, suggesting that ANAC019 is a positive regulator of ABA signalling.

ANAC019 is a positive regulator of ABA signalling

Figure 6
ANAC019 is a positive regulator of ABA signalling

Transgenic plants overexpressing full-length ANAC019-(1–317), ANAC019-(1–168) and ANAC019-(169–317) are more sensitive to ABA than the Col-0 wild-type and anac019 mutant. (A) Germination efficiency of Col-0, anac019 mutant, 35S:ANAC019-(1–317), 35S:ANAC019-(1–168) or 35S:ANAC019-(169–317) seeds grown on medium supplemented with different concentrations of ABA (0, 0.1, 0.5, 1, 2 or 5 μM) for 3 days after stratification. (B) Post-germination growth efficiency of Col-0, anac019 mutant, 35S:ANAC019(1–317), 35S:ANAC019(1–168) or 35S:ANAC019(169–317) seedlings grown on medium supplemented with different concentrations of ABA for 7 days after stratification. (C) Expression profiles (means±S.E.M.) of ABA-signalling and stress-responsive genes in Col-0, empty vector control, anac019 mutant, 35S:ANAC019-(1–317), 35S:ANAC019(1–168) or 35S:ANAC019-(169–317) plants. The expressions of ABA-responsive (COR47 and RD29B) and stress-responsive (FER1 and ERD11) genes were analysed by quantitative real-time RT–PCR using actin2 as a reference.

Figure 6
ANAC019 is a positive regulator of ABA signalling

Transgenic plants overexpressing full-length ANAC019-(1–317), ANAC019-(1–168) and ANAC019-(169–317) are more sensitive to ABA than the Col-0 wild-type and anac019 mutant. (A) Germination efficiency of Col-0, anac019 mutant, 35S:ANAC019-(1–317), 35S:ANAC019-(1–168) or 35S:ANAC019-(169–317) seeds grown on medium supplemented with different concentrations of ABA (0, 0.1, 0.5, 1, 2 or 5 μM) for 3 days after stratification. (B) Post-germination growth efficiency of Col-0, anac019 mutant, 35S:ANAC019(1–317), 35S:ANAC019(1–168) or 35S:ANAC019(169–317) seedlings grown on medium supplemented with different concentrations of ABA for 7 days after stratification. (C) Expression profiles (means±S.E.M.) of ABA-signalling and stress-responsive genes in Col-0, empty vector control, anac019 mutant, 35S:ANAC019-(1–317), 35S:ANAC019(1–168) or 35S:ANAC019-(169–317) plants. The expressions of ABA-responsive (COR47 and RD29B) and stress-responsive (FER1 and ERD11) genes were analysed by quantitative real-time RT–PCR using actin2 as a reference.

To further understand the molecular basis of ANAC019-mediated ABA hypersensitivity, the expression profile of several ABA- and stress-responsive genes (RD29b, COR47, FER1 and ERD11 [16]) were investigated by real-time PCR (Figure 6C). Three out of the four genes analysed displayed increased transcript levels in plants expressing ANAC019-(1–317), ANAC019-(1–168) and ANAC019-(169–317) compared with control and anac019 mutant plants. The induction of FER1 expression was dependent on the C-terminal TRD of ANAC019 (Figures 6C and 7). Taken together, the up-regulation of the ABA-responsive marker genes in plants expressing full-length or truncated versions of ANAC019 largely correlated with the increased ABA sensitivity of ectopically expressing ANAC019 plants. The presence of a NAC-BS in the 1-kb promoter of all four genes (results not shown) highlights that ANAC019 is a possible direct positive upstream regulator of ABA-signalling genes.

Analysis of ANAC019-mediated ABA sensitivity by domain-swap analysis

Figure 7
Analysis of ANAC019-mediated ABA sensitivity by domain-swap analysis

(A) Schematic representation of ANAC019-derived chimaeric constructs expressed from the 35S promoter of pCAMBIA3300. For analysis of ANAC019 functional specificity, TRDs (grey) of NAC2/ORE1, SOG1 and NTL8 were fused to the ANAC019 NAC domain (black) NAC2/ORE1, NTL8 and SOG1 NAC domains (grey) were fused to ANAC019 TRD (black). (B) ABA sensitivity of plants ectopically expressing ANAC019 and domain-swapped NAC gene products relative to empty-vector-transformed plants was scored 5 days after the end of stratification. The assay was performed under the same condition as shown in Figures 6(A) and 6(B). *P<0.05 significance from control plants compared with test plants as analysed by a two-sided t-test. Scale bar =1 cm.

Figure 7
Analysis of ANAC019-mediated ABA sensitivity by domain-swap analysis

(A) Schematic representation of ANAC019-derived chimaeric constructs expressed from the 35S promoter of pCAMBIA3300. For analysis of ANAC019 functional specificity, TRDs (grey) of NAC2/ORE1, SOG1 and NTL8 were fused to the ANAC019 NAC domain (black) NAC2/ORE1, NTL8 and SOG1 NAC domains (grey) were fused to ANAC019 TRD (black). (B) ABA sensitivity of plants ectopically expressing ANAC019 and domain-swapped NAC gene products relative to empty-vector-transformed plants was scored 5 days after the end of stratification. The assay was performed under the same condition as shown in Figures 6(A) and 6(B). *P<0.05 significance from control plants compared with test plants as analysed by a two-sided t-test. Scale bar =1 cm.

Modular interdependency of ANAC019-mediated ABA signalling

Taoka et al. [19] showed that among close homologues of the CUC proteins of the NAC family, functional specificity resides in the N-terminal NAC domain. To examine if ANAC019-mediated perturbation of ABA-sensitivity can be mapped to specific domains, six chimaeric constructs were generated in which NAC domains of NAC2/ORE1, NTL8 and SOG1 were individually fused to the ANAC019 TRD and the TRDs of NAC2/ORE1, NTL8 and SOG1 were fused to the ANAC019 NAC domain (Figure 7A). All constructs were transformed into Col-0 wild-type plants. Selection of these candidate fusion partners enabled analysis of both distant and close homologues of ANAC019, which have both overlapping and non-overlapping expression profiles (see Figure 1). The effect of ectopic expression of these constructs was examined by scoring root length in the absence or presence of 2 μM ABA (Figure 7). In addition to conferring reduced root length in plants expressing ANAC019, this ABA concentration has also been shown to compromise root length when other NAC members are ectopically expressed [28].

Interestingly, when replacing the ANAC019 NAC domain with the NAC domains of NAC2/ORE1 and NTL8 plants became ABA hypersensitive, whereas introducing the SOG1 NAC domain had no effect on root elongation compared with control plants (Figure 7B). Substituting the ANAC019 C-terminal TRD with NAC2/ORE1 and NTL8 TRDs had no effect on root elongation compared with control plants. Replacing the ANAC019 TRD with the SOG1 TRD could have a lethal effect on plants or hinder T-DNA (transfer DNA) insertion events, as no transformants were recovered from several attempts to ectopically express the ANAC019–SOG1 chimaera (Figure 7B). Real-time RT–PCR analysis was used to analyse whether the observed phenotypes correlated with the chimaeric transcript dosage. All transcripts were expressed in the same range, although transcripts of ANAC019-(1–168)–NTL8-(158–335) and NAC2/ORE1-(1–174)–ANAC019-(169–317) showed the lowest abundances (Supplementary Figure S1). In conclusion, these results imply that the C-terminal region of ANAC019 encodes a major level of functional specificity alone and when fused to both distantly and closely related NAC domains of ABA-inducible origin.

DISCUSSION

In this manuscript the modular structure of plant-specific NAC transcription factors was parsed to improve the understanding of their structure–function relationship. Most NAC proteins have an N-terminal NAC DBD and a C-terminal TRD. Such modularity is a paradigm of TF architecture, and members of individual TF families often target similar core cis-elements [1]. In accordance with this, DNA-binding studies of NAC proteins from distant clades have shown that NAC TFs bind a palindromic consensus sequence containing a core CGT[GA] sequence [15,16,36,37], supporting the existence of conserved TF family DNA-target sites. However, CASTing using the barley NAC protein IDEF2 and the Arabidopsis calmodulin-binding NAC CBNAC (a group I protein) selected unrelated NAC-BSs with a GCTT core sequence [46,47]. In the present study the structural divergence of the NAC proteins was used to examine their DNA-binding properties using a consensus sequence selected by two NAC proteins, ANAC019 and NAC2/ORE1, which belong to different NAC groups (Figure 1). Apart from the ANAC019 and NAC2/ORE1 NAC domains, the NAC domain derived from ATAF1 and VND7 bound palNAC-BS with comparable affinity, whereas the rest of the NAC domains bound this oligonucleotide probe with lower affinity or, in the case of ANAC003, with no detectable affinity. This difference in affinity cannot be explained by the quality of the recombinant proteins which were all soluble, homogenous and of high purity. Therefore the most probable explanation is that NAC TFs from different groups recognize different DNA-target sites.

The findings in the present paper raise other intriguing themes associated with transcriptional regulators. The phylogenetic analysis of the NAC domains shows the major subgroups correlate with conserved motifs in the C-terminal TRDs of NAC proteins (Figure 2B and Supplementary Table 2). Although divergent with respect to sequences, these motifs have common features including a dominance of polar residues and a few highly conserved hydrophobic residues, characteristic of TRDs [2]. The analyses also suggested that C-terminal NAC regions have a high degree of ID. On the basis of these properties, the NAC proteins are well-suited as model proteins for systematic analysis of structural ID, a research topic of great current interest. Most NAC TRDs contain sequence motifs that may be recognized both by specific and general proteins of the transcriptional apparatus, and, in cases such as motif l (Figure 4B), where they interrupt the large disordered TRD, they could serve as MoRFs for partner-induced protein folding [41]. A few of these motifs, e.g. motif W and motif WQ, have already been shown to be essential for TRD activity [19]. Further characterization of these motifs may depend on the identification of interaction partners; screening using motif-based consensus sequences could represent a useful strategy for identifying such interactors.

The phylogenetic analysis also revealed tissue-specific NAC gene expression clusters correlating with phylogenetic clades (Figure 1). Such observations have been reported for members of other large gene families [48] and suggest that the NAC gene family has expanded throughout evolution by gene duplication creating paralogous genes with a high degree of sequence similarity and functional redundancy. However, the less prominent correlation between phylogenetic clusters and stress expression patterns could indicate that functional redundancy of regulators from the same tissue is diluted during the fine-tuned environmental stress perceptions needed for an adequate response of the given plant tissue. Future systems-orientated temporal studies of selected tissues in response to environmental stresses are needed for an improved understanding of overlapping NAC gene expression patterns and potential functional redundancy.

Another interesting observation from the characterization of NAC TF modularity was the ability of ANAC019 DBD and TRD to mimic ABA hypersensitivity associated with ectopic expression of full-length ANAC019 in a wild-type background (Figures 6A and 6B). This suggested that powerful effects may result from engineering of NAC TFs. The ability of the ANAC019 DBD to positively regulate ABA signalling may be explained by its ability to dimerize with endogenous NAC domains and thereby enhance promoter binding [15] and subsequent activation of ABA-signalling genes. In contrast, the ability of a NAC TRD to function without its natural molecular context has not been reported previously. The functional specificity may be explained by the ability of isolated regions to functionally interact with other proteins of the transcriptional complex. That specific interactions are important for ANAC019 TRD function is supported by the inability of the TRD from NAC2/ORE1, NTL8 and SOG1 to functionally replace ANAC019 TRD in the chimaeric proteins (Figure 7B). Although knowledge of how the isolated ANAC019 TRD exerts its function in vivo remains elusive, it is intriguing to acknowledge research performed more than two decades ago on the GAL4 transcription factor [3]. Herein, an N-terminally truncated version of GAL4, disabled in DNA-binding, was able to induce the expression of a reporter gene and even localize to the nucleus when overexpressed in a GAL4 wild-type background [3]. All together, Johnston et al. [3] speculated that the truncated GAL4 protein was transported into the nucleus in association with the wild-type GAL4. This scenario is somewhat analogous to our findings on the expression perturbations of ABA- and stress-signalling reporter genes induced when full-length and truncated versions of ANAC019 are ectopically expressed (Figure 6C). Additionally, the hypothesis on co-transportation should be addressed by analysing the subcellular localization of ANAC019-(169–317) and reporter gene expression in a 35S:ANAC019-(169–317)-expressing anac019 mutant background.

Although truncated versions of ANAC019 excluding either the DBD or the TRD of ANAC019 retain ABA hypersensitivity when ectopically expressed in a wild-type background, the potential of the ANAC019 TRD is of particular interest. In addition to its autonomous effect when ectopically expressed, the ANAC019 TRD encodes a major level of functionality during ABA signalling even when fused to NAC domains from NAC2/ORE1 and NTL8 which are members of distantly related groups. The chimaera, NAC2/ORE1-(1–174)–ANAC019-(169–317) and NTL8-(1–157)–ANAC019-(169–317), also mimic the ANAC019-mediated phenotype of ABA hypersensitivity. In contrast, replacing ANAC019-(1–168) with SOG1-(1–215) did not confer ABA hypersensitivity when fused to the ANAC019 TRD. Although seemingly contrasting with the results showing that ability of CUC proteins to enhance adventitious shoot formation is dependent on the NAC domain [19] we propose that the functionality of the NAC domain of ANAC019 during ABA signalling is exchangeable with NAC2/ORE1 and NTL8 NAC domains due to the conservation of key residues. Hence, several studies have demonstrated that NAC proteins form both homo- and hetero-dimers [5,13,39], and the structure of the NAC domain fold contains both dimerization and DNA-binding interfaces [13,15]. In the NAC domain salt-bridge, Arg-19, Glu-26 and conserved hydrophobic residues are important for dimerization [15]. Arg-19 of ANAC019 is conserved in 81% of the NAC proteins, including NAC2/ORE1 and NTL8. In SOG1, and the additional subgroup IX-1 members, Arg-19 is replaced by a lysine residue and the sequence of the dimerization region shows a relatively low degree of conservation in these proteins (Figures 2A and 2D). This supports the hypothesis that residues in this region play an important role in the ‘exchangeability’ of the ANAC019 NAC domain with the NAC domain of NTL8 and NAC2/ORE1 during ABA signalling. This is further substantiated by our previous demonstration of in vitro ANAC019–NAC2/ORE1 heterodimerization [15]. Future studies using site-directed mutagenesis of SOG1, NTL8 and NAC2/ORE1 are needed to fully elucidate the roles of these residues.

The DNA-binding motif of the NAC domain has not been finally identified and no atomic-level structure information is available for a NAC–DNA complex. However, studies of mutant ANAC019 NAC domains pinpoint the loop/β-strand region containing Lys-79, Arg-85 and Arg-88 of ANAC019 as the DNA-binding, or part of the DNA-binding, motif (Figures 2A and 2D; [15]). Whereas Arg-88 is conserved in all NAC proteins, Arg-85 is replaced by a glutamine residue in NTL8 and ANAC003, and Lys-79 is replaced by an alanine residue in SOG1 and VOZ2. These changes may explain the decreased palNAC-BS affinity of these proteins. As Lys-79, Arg-85 and Arg-88 are conserved in NTL6 and VND3, which also bound palNAC-BS with lower affinity, additional regions or residues, such as Arg-78 or Arg-76, must be of importance to DNA binding. It is possible that the ability to dimerize with endogenous ANAC019 could overrule suboptimal DNA binding by, for example, the NTL8 NAC domain in enabling ANAC019-mediated ABA sensing. However, in the case of the SOG1 NAC domain, which binds palNAC-BS with relatively low affinity, dimerization may not be sufficient to compensate for weak target-DNA binding. Although ANAC019 mutants deficient in dimerization cannot bind cognate DNA-binding sites in vitro, it is not known whether two NAC-BSs are needed for in vivo DNA-binding by NAC TFs [15]. Hence, although the C-terminal TRD of ANAC019 is needed for conferring ABA hypersensitivity in NAC proteins with both a DBD and a TRD, residues in the NAC domain could modulate this activity through inter-domain communication for identification of interaction partners and target promoters from which the ANAC019 TRD exerts its transactivation activity. All together, our results have laid out future directions for the engineering of NAC TFs for an improved understanding and use of NAC networks and for investigating their role in plant stress perception.

Abbreviations

     
  • ABA

    abscisic acid

  •  
  • ABRC

    Arabidopsis Biological Resource Center

  •  
  • CaMV

    cauliflower mosaic virus

  •  
  • CASTing

    cyclic amplification and selection of targets

  •  
  • Col-0

    Columbia ecotype 0

  •  
  • COR47

    cold-responsive 47

  •  
  • CUC

    cup-shaped cotyledon

  •  
  • DBD

    DNA-binding domain

  •  
  • DIG

    digoxigenin

  •  
  • DTT

    dithiothreitol

  •  
  • EMSA

    electrophoretic mobility-shift assay

  •  
  • ERD11

    early-responsive-to-dehydration 11

  •  
  • FER1

    ferritin precursor 1

  •  
  • GST

    glutathione transferase

  •  
  • ID

    intrinsic disorder

  •  
  • MAST

    motif alignment and search tool

  •  
  • MEME

    multiple expectation maximization for motif elicitation

  •  
  • MoRF

    molecular recognition feature

  •  
  • NAC

    NAM (no apical meristem), ATAF, CUC

  •  
  • NAC-BS

    NAC-binding site

  •  
  • palNAC-BS

    palindromic NAC-BS

  •  
  • PSI

    position-specific iterated

  •  
  • Rap1p

    repressor activator protein 1

  •  
  • RD29b

    responsive-to-desiccation 29b

  •  
  • RT

    reverse transcription

  •  
  • SOG1

    suppressor of gamma response 1

  •  
  • SD

    synthetic defined

  •  
  • TAIR

    The Arabidopsis Information Resource

  •  
  • TAX

    tiling array express

  •  
  • TF

    transcription factor

  •  
  • TRD

    transcription regulatory domain

  •  
  • VND

    vascular-related NAC domain

  •  
  • VOZ

    vascular plant one-zinc finger protein

AUTHOR CONTRIBUTION

The planning of experiments and writing of the manuscript were performed by Michael Jensen, Trine Kjaersgaard and Karen Skriver. Michael Jensen, Trine Kjaersgaard, Pernille Galberg, Charlotte O'Shea and Michael Nielsen contributed to experimental characterization of NAC structure–function. Klaus Petersen and Michael Jensen performed work relating to gene expression analysis.

We thank Dr Lars Ellgaard for critical revision of manuscript prior to submission.

FUNDING

This work was supported by the Danish Agency for Science Technology and Innovation (DFF, Technology and Production) [grant numbers 274-07-0173 (to K.S.) and 274-07-0385 (to M.K.J.)].

References

References
1
Luscombe
N. M.
Thornton
J. M.
Protein–DNA interactions: amino acid conservation and the effects of mutations on binding specificity
J. Mol. Biol.
2002
, vol. 
320
 (pg. 
991
-
1009
)
2
Dyson
H. J.
Wright
P. E.
Intrinsically unstructured proteins and their functions
Nat. Rev. Mol. Cell Biol.
2005
, vol. 
6
 (pg. 
197
-
208
)
3
Johnston
S. A.
Zavortink
M. J.
Debouck
C.
Hopper
J. E.
Functional domains of the yeast regulatory protein GAL4
Proc. Natl. Acad. Sci. U.S.A.
1986
, vol. 
83
 (pg. 
6553
-
6557
)
4
Huang
J.
Li
X.
Yi
P.
Hilf
R.
Bambara
R. A.
Muyan
M.
Targeting estrogen responsive elements (EREs): design of potent transactivators for ERE-containing genes
Mol. Cell. Endocrinol.
2004
, vol. 
218
 (pg. 
65
-
78
)
5
Xie
Q.
Frugis
G.
Colgan
D.
Chua
N. H.
Arabidopsis NAC1 transduces auxin signal downstream of TIR1 to promote lateral root development
Genes Dev.
2000
, vol. 
14
 (pg. 
3024
-
3036
)
6
Graham
I. R.
Haw
R. A.
Spink
K. G.
Halden
K. A.
Chambers
A.
In vivo analysis of functional regions within yeast Rap1p
Mol. Cell. Biol.
1999
, vol. 
19
 (pg. 
7481
-
7490
)
7
Porsch
M.
Sauer
M.
Schulze
S.
Bahlo
A.
Roth
M.
Pflugfelder
G. O.
The relative role of the T-domain and flanking sequences for developmental control and transcriptional regulation in protein chimeras of Drosophila OMB and ORG-1
Mech. Dev.
2005
, vol. 
122
 (pg. 
81
-
96
)
8
Olsen
A. N.
Ernst
H. A.
Leggio
L. L.
Skriver
K.
NAC transcription factors: structurally distinct, functionally diverse
Trends Plant Sci.
2005
, vol. 
10
 (pg. 
79
-
87
)
9
Ooka
H.
Satoh
K.
Doi
K.
Nagata
T.
Otomo
Y.
Murakami
K.
Matsubara
K.
Osato
N.
Kawai
J.
Carninci
P.
, et al. 
Comprehensive analysis of NAC family genes in Oryza sativa and Arabidopsis thaliana
DNA Res.
2003
, vol. 
10
 (pg. 
239
-
247
)
10
Rushton
P. J.
Bokowiec
M. T.
Han
S.
Zhang
H.
Brannock
J. F.
Chen
X.
Laudeman
T. W.
Timko
M. P.
Tobacco transcription factors: novel insights into transcriptional regulation in the Solanaceae
Plant Physiol.
2008
, vol. 
147
 (pg. 
280
-
295
)
11
Jensen
M. K.
Rung
J. H.
Gregersen
P. L.
Gjetting
T.
Fuglsang
A. T.
Hansen
M.
Joehnk
N.
Lyngkjaer
M. F.
Collinge
D. B.
The HvNAC6 transcription factor: a positive regulator of penetration resistance in barley and Arabidopsis
Plant Mol. Biol.
2007
, vol. 
65
 (pg. 
137
-
150
)
12
Yoshiyama
K.
Conklin
P. A.
Huefner
N. D.
Britt
A. B.
Suppressor of gamma response 1 (SOG1) encodes a putative transcription factor governing multiple responses to DNA damage
Proc. Natl. Acad. Sci. U.S.A.
2009
, vol. 
106
 (pg. 
12843
-
12848
)
13
Ernst
H. A.
Olsen
A. N.
Larsen
S.
Lo
L. L.
Structure of the conserved domain of ANAC, a member of the NAC family of transcription factors
EMBO Rep.
2004
, vol. 
5
 (pg. 
297
-
303
)
14
Greve
K.
La Cour
T.
Jensen
M. K.
Poulsen
F. M.
Skriver
K.
Interactions between plant RING-H2 and plant-specific NAC (NAM/ATAF1/2/CUC2) proteins: RING-H2 molecular specificity and cellular localization
Biochem. J.
2003
, vol. 
371
 (pg. 
97
-
108
)
15
Olsen
A. N.
Ernst
H. A.
Leggio
L. L.
Skriver
K.
DNA-binding specificity and molecular functions of NAC transcription factors
Plant Science
2005
, vol. 
169
 (pg. 
785
-
797
)
16
Tran
L. S.
Nakashima
K.
Sakuma
Y.
Simpson
S. D.
Fujita
Y.
Maruyama
K.
Fujita
M.
Seki
M.
Shinozaki
K.
Yamaguchi-Shinozaki
K.
Isolation and functional analysis of Arabidopsis stress-inducible NAC transcription factors that bind to a drought-responsive cis-element in the early responsive to dehydration stress 1 promoter
Plant Cell
2004
, vol. 
16
 (pg. 
2481
-
2498
)
17
Bu
Q.
Jiang
H.
Li
C. B.
Zhai
Q.
Zhang
J.
Wu
X.
Sun
J.
Xie
Q.
Li
C.
Role of the Arabidopsis thaliana NAC transcription factors ANAC019 and ANAC055 in regulating jasmonic acid-signaled defense responses
Cell Res.
2008
, vol. 
18
 (pg. 
756
-
767
)
18
Bu
Q.
Li
H.
Zhao
Q.
Jiang
H.
Zhai
Q.
Zhang
J.
Wu
X.
Sun
J.
Xie
Q.
Wang
D.
Li
C.
The Arabidopsis RING finger E3 ligase RHA2a is a novel positive regulator of abscisic acid signaling during seed germination and early seedling development
Plant Physiol.
2009
, vol. 
150
 (pg. 
463
-
481
)
19
Taoka
K.
Yanagimoto
Y.
Daimon
Y.
Hibara
K.
Aida
M.
Tasaka
M.
The NAC domain mediates functional specificity of CUP-SHAPED COTYLEDON proteins
Plant J.
2004
, vol. 
40
 (pg. 
462
-
473
)
20
Daimon
Y.
Takabe
K.
Tasaka
M.
The CUP-SHAPED COTYLEDON genes promote adventitious shoot formation on calli
Plant Cell Physiol
2003
, vol. 
44
 (pg. 
113
-
121
)
21
Thompson
J. D.
Gibson
T. J.
Plewniak
F.
Jeanmougin
F.
Higgins
D. G.
The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools
Nucleic Acids Res.
1997
, vol. 
25
 (pg. 
4876
-
4882
)
22
Tamura
K.
Dudley
J.
Nei
M.
Kumar
S.
MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0
Mol. Biol. Evol.
2007
, vol. 
24
 (pg. 
1596
-
1599
)
23
Vacic
V.
Uversky
V. N.
Dunker
A. K.
Lonardi
S.
Composition Profiler: a tool for discovery and visualization of amino acid composition differences
BMC Bioinformatics
2007
, vol. 
8
 pg. 
211
 
24
Peng
K.
Vucetic
S.
Radivojac
P.
Brown
C. J.
Dunker
A. K.
Obradovic
Z.
Optimizing long intrinsic disorder predictors with protein evolutionary information
J. Bioinform. Comput. Biol.
2005
, vol. 
3
 (pg. 
35
-
60
)
25
Zeller
G.
Henz
S. R.
Widmer
C. K.
Sachsenberg
T.
Ratsch
G.
Weigel
D.
Laubinger
S.
Stress-induced changes in the Arabidopsis thaliana transcriptome analyzed using whole-genome tiling arrays
Plant J.
2009
, vol. 
58
 (pg. 
1068
-
1082
)
26
Gentleman
R. C.
Carey
V. J.
Bates
D. M.
Bolstad
B.
Dettling
M.
Dudoit
S.
Ellis
B.
Gautier
L.
Ge
Y.
Gentry
J.
, et al. 
Bioconductor: open software development for computational biology and bioinformatics
Genome Biol.
2004
, vol. 
5
 pg. 
R80
 
27
Clough
S. J.
Bent
A. F.
Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana
Plant J.
1998
, vol. 
16
 (pg. 
735
-
743
)
28
Jensen
M. K.
Hagedorn
P. H.
de Torres-Zabala
M.
Grant
M. R.
Rung
J. H.
Collinge
D. B.
Lyngkjaer
M. F.
Transcriptional regulation by an NAC (NAM–ATAF1,2–CUC2) transcription factor attenuates ABA signalling for efficient basal defence towards Blumeria graminis f. sp. hordei in Arabidopsis
Plant J.
2008
, vol. 
56
 (pg. 
867
-
880
)
29
Muller
P. Y.
Janovjak
H.
Miserez
A. R.
Dobbie
Z.
Processing of gene expression data generated by quantitative real-time RT-PCR
Biotechniques
2002
, vol. 
32
 (pg. 
1372
-
1379
)
30
Mitsuda
N.
Seki
M.
Shinozaki
K.
Ohme-Takagi
M.
The NAC transcription factors NST1 and NST2 of Arabidopsis regulate secondary wall thickenings and are required for anther dehiscence
Plant Cell
2005
, vol. 
17
 (pg. 
2993
-
3006
)
31
Kubo
M.
Udagawa
M.
Nishikubo
N.
Horiguchi
G.
Yamaguchi
M.
Ito
J.
Mimura
T.
Fukuda
H.
Demura
T.
Transcription switches for protoxylem and metaxylem vessel formation
Genes Dev.
2005
, vol. 
19
 (pg. 
1855
-
1860
)
32
Yamaguchi
M.
Kubo
M.
Fukuda
H.
Demura
T.
Vascular-related NAC-DOMAIN7 is involved in the differentiation of all types of xylem vessels in Arabidopsis roots and shoots
Plant J.
2008
, vol. 
55
 (pg. 
652
-
664
)
33
Ko
J. H.
Yang
S. H.
Park
A. H.
Lerouxel
O.
Han
K. H.
ANAC012, a member of the plant-specific NAC transcription factor family, negatively regulates xylary fiber development in Arabidopsis thaliana
Plant J.
2007
, vol. 
50
 (pg. 
1035
-
1048
)
34
Kim
J. H.
Woo
H. R.
Kim
J.
Lim
P. O.
Lee
I. C.
Choi
S. H.
Hwang
D.
Nam
H. G.
Trifurcate feed-forward regulation of age-dependent cell death involving miR164 in Arabidopsis
Science
2009
, vol. 
323
 (pg. 
1053
-
1057
)
35
Kim
Y. S.
Kim
S. G.
Park
J. E.
Park
H. Y.
Lim
M. H.
Chua
N. H.
Park
C. M.
A membrane-bound NAC transcription factor regulates cell division in Arabidopsis
Plant Cell
2006
, vol. 
18
 (pg. 
3132
-
3144
)
36
Mitsuda
N.
Hisabori
T.
Takeyasu
K.
Sato
M. H.
VOZ: isolation and characterization of novel vascular plant transcription factors with a one-zinc finger from Arabidopsis thaliana
Plant Cell Physiol.
2004
, vol. 
45
 (pg. 
845
-
854
)
37
Xue
G.-P.
Bower
N. I.
McIntyre
M. C.
Riding
G. A.
Kazan
K.
Shorter
R.
TaNAC69 from the NAC superfamily of transcription factors is up-egulated by abiotic stresses in wheat and recognises two consensus DNA-binding sequences
Funct. Plant Biol.
2006
, vol. 
33
 (pg. 
43
-
57
)
38
Bianchi
A.
Stansel
R. M.
Fairall
L.
Griffith
J. D.
Rhodes
D.
de Lange
T.
TRF1 binds a bipartite telomeric site with extreme spatial flexibility
EMBO J.
1999
, vol. 
18
 (pg. 
5735
-
5744
)
39
Hegedus
D.
Yu
M.
Baldwin
D.
Gruber
M.
Sharpe
A.
Parkin
I.
Whitwill
S.
Lydiate
D.
Molecular characterization of Brassica napus NAC domain transcriptional activators induced in response to biotic and abiotic stress
Plant Mol. Biol.
2003
, vol. 
53
 (pg. 
383
-
397
)
40
Aida
M.
Ishida
T.
Tasaka
M.
Shoot apical meristem and cotyledon formation during Arabidopsis embryogenesis: interaction among the CUP-SHAPED COTYLEDON and SHOOT MERISTEMLESS genes
Development
1999
, vol. 
126
 (pg. 
1563
-
1570
)
41
Vacic
V.
Oldfield
C. J.
Mohan
A.
Radivojac
P.
Cortese
M. S.
Uversky
V. N.
Dunker
A. K.
Characterization of molecular recognition features, MoRFs, and their binding partners
J. Proteome Res.
2007
, vol. 
6
 (pg. 
2351
-
2366
)
42
Zheng
X.
Chen
B.
Lu
G.
Han
B.
Overexpression of a NAC transcription factor enhances rice drought and salt tolerance
Biochem. Biophys. Res. Commun.
2009
, vol. 
379
 (pg. 
985
-
989
)
43
Kim
S. Y.
Kim
S. G.
Kim
Y. S.
Seo
P. J.
Bae
M.
Yoon
H. K.
Park
C. M.
Exploring membrane-associated NAC transcription factors in Arabidopsis: implications for membrane biology in genome regulation
Nucleic Acids Res.
2007
, vol. 
35
 (pg. 
203
-
213
)
44
Lu
P. L.
Chen
N. Z.
An
R.
Su
Z.
Qi
B. S.
Ren
F.
Chen
J.
Wang
X. C.
A novel drought-inducible gene, ATAF1, encodes a NAC family protein that negatively regulates the expression of stress-responsive genes in Arabidopsis
Plant Mol. Biol.
2007
, vol. 
63
 (pg. 
289
-
305
)
45
Finkelstein
R. R.
Gibson
S. I.
ABA and sugar interactions regulating development: cross-talk or voices in a crowd?
Curr. Opin. Plant Biol.
2002
, vol. 
5
 (pg. 
26
-
32
)
46
Kim
H. S.
Park
B. O.
Yoo
J. H.
Jung
M. S.
Lee
S. M.
Han
H. J.
Kim
K. E.
Kim
S. H.
Lim
C. O.
Yun
D. J.
, et al. 
Identification of a calmodulin-binding NAC protein as a transcriptional repressor in Arabidopsis
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
36292
-
36302
)
47
Ogo
Y.
Kobayashi
T.
Nakanishi
I. R.
Nakanishi
H.
Kakei
Y.
Takahashi
M.
Toki
S.
Mori
S.
Nishizawa
N. K.
A novel NAC transcription factor, IDEF2, that recognizes the iron deficiency-responsive element 2 regulates the genes involved in iron homeostasis in plants
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
13407
-
13417
)
48
Parenicova
L.
de Folter
S.
Kieffer
M.
Horner
D. S.
Favalli
C.
Busscher
J.
Cook
H. E.
Ingram
R. M.
Kater
M. M.
Davies
B.
, et al. 
Molecular and phylogenetic analyses of the complete MADS-box transcription factor family in Arabidopsis: new openings to the MADS world
Plant Cell
2003
, vol. 
15
 (pg. 
1538
-
1551
)
49
Bailey
T. L.
Elkan
C.
Fitting a mixture model by expectation maximization to discover motifs in biopolymers
Proc. Int. Conf. Intell. Syst. Mol. Biol.
1994
, vol. 
2
 (pg. 
28
-
36
)
50
Bailey
T. L.
Gribskov
M.
Combining evidence using p-values: application to sequence homology searches
Bioinformatics
1998
, vol. 
14
 (pg. 
48
-
54
)

Supplementary data