Characterization of barley (Hordeum vulgare L.) NAC transcription factors suggests conserved functions compared to both monocots and dicots

Background The NAC transcription factor family is involved in the regulation of traits in both monocots and dicots of high agronomic importance. Understanding the precise functions of the NAC genes can be of utmost importance for the improvement of cereal crop plants through plant breeding. For the cereal crop plant barley (Hordeum vulgare L.) only a few NAC genes have so far been investigated. Results Through searches in publicly available barley sequence databases we have obtained a list of 48 barley NAC genes (HvNACs) with 43 of them representing full-length coding sequences. Phylogenetic comparisons to Brachypodium, rice, and Arabidopsis NAC proteins indicate that the barley NAC family includes members from all of the eight NAC subfamilies, although by comparison to these species a number of HvNACs still remains to be identified. Using qRT-PCR we investigated the expression profiles of 46 HvNACs across eight barley tissues (young flag leaf, senescing flag leaf, young ear, old ear, milk grain, late dough grain, roots, and developing stem) and two hormone treatments (abscisic acid and methyl jasmonate). Conclusions Comparisons of expression profiles of selected barley NAC genes with the published functions of closely related NAC genes from other plant species, including both monocots and dicots, suggest conserved functions in the areas of secondary cell wall biosynthesis, leaf senescence, root development, seed development, and hormone regulated stress responses.


Background
In recent years, research in the regulatory roles of members of the plant-specific NAC (NAM, ATAF-1,2, CUC) transcription factor family has increased considerably. It has become evident that these transcription factors are essential components in the regulation of a multitude of traits in plants, including traits of agronomic importance such as development, senescence, tolerance to both biotic and abiotic stresses, and hormone responses.
The first NAC transcription factor was described fifteen years ago by Souer et al. [1]. Since then the genomes of a number of plant species have been fully sequenced, revealing the NAC gene family to code for one of the largest families of transcription factors in plants [2]. Members of the family are identified by the presence of the NAC domain [3]. This domain consists of five subdomains: A-E [4,5] that make up motifs for both DNA-binding and protein-protein interactions [6].
A typical NAC transcription factor has the conserved NAC domain in the N-terminal [5] as well as a more variable, transcriptional activation or repression region in the C-terminal [7] (figure 1A). Thirteen Arabidopsis and six rice NAC transcription factors have been shown to contain an α-helical transmembrane motif in the far C-terminal region which anchors the NAC protein to intracellular membranes rendering them inert. Only through controlled proteolytic cleavage from this anchor are the proteins able to exert their function [8]. Examination of known NAC families reveals a few atypical NAC genes, such as genes encoding only the NAC domain or genes with the NAC domain in the C-terminal with the variable region preceding it (figure 1B). None of these atypical NAC genes have been among closely characterized NAC genes, and therefore their functions are still unknown.
There have been several attempts to classify the phylogenetic relationships among all members of the NAC family [2,5,9,10]. However, the most comprehensive classification to date is the one published by Shen et al. [2], who divided the NAC superfamily into eight distinct NAC-subfamilies, NAC-a to NAC-h, based on NAC domain alignments. Furthermore, each subfamily was divided into several subgroups based on C-terminal motifs.
Since the first NAC gene was described, a number of NAC transcription factors have been characterized, in particular from the Arabidopsis NAC family (ANACs, 115 members) and, to a lesser extent, from the rice NAC family (ONACs, 144 members). There have been only a few detailed studies on NAC genes from species with non-sequenced genomes, such as wheat and barley [11,12].
So far it is known that many aspects of secondary cell wall biosynthesis (SCWB) are regulated by NAC transcription factors. Arabidopsis NST1 (ANAC043) was found to promote secondary cell wall thickening in the stem fibers [13][14][15]. NST2 (ANAC066), a close homologue of NST1, seems to have some functional redundancy to NST1. However, it shows strong promoter activity in the anthers, suggesting this tissue to be the primary site of NST2 regulation [15]. A homologue of NST1 in Medicago truncatula (MtNST1) was recently found to regulate lignin biosynthesis in the interfascicular region of the cell wall [16]. SND1 (NST3/ANAC012) appears to specifically regulate xylary fiber thickness [13,14,17,18]. This regulation most likely involves SND2 and SND3 (ANAC073 and ANAC010, respectively), which are both downstream targets of SND1 [19]. In Populus an orthologue of SND2 and SND3, PopNAC154, was shown to affect stem elongation [20].
Several NACs have been shown to regulate leaf senescence. Guo and Gan [21] described how AtNAP (ANAC029) could induce or delay leaf senescence in overexpression or knockout plants, respectively. More recently, the two paralogues ORE1 (AtNAC2/ ANAC092) [22,23] and ORS1 (ANAC059) [24] were found to regulate salt-induced senescence. Positional cloning of a quantitative trait locus associated with increased grain protein content in wheat lead to the discovery of the wheat NAC transcription factor NAM-B1 [11]. As a possible orthologue of AtNAP NAM-B1 was also able to induce leaf senescence as well as increase grain protein content [25]. Besides its role in leaf senescence, ORE1 also plays a part in lateral root development as this formation is promoted in transgenic plants overexpressing ORE1 [26]. Another NAC transcription factor, NAC1 (ANAC021), has similarly been shown to induce the formation of lateral roots [27].
Untill now, only five barley NAC genes have been mentioned in the literature. The first study was done in 2003 by Scharrenberg et al. [42] who isolated cDNA clones from senescing barley flag leaves. One of these clones, named HvSF6, showed homology to wheat NAC transcription factor GRAB2 earlier reported to be involved in wheat Geminivirus defense [38]. The expression of this NAC gene was found to be induced by both age-dependent and dark-induced senescence, as well as a combination of the hormones ethylene and JA. Interestingly, neither hormone alone was enough to cause the induction of this gene. Robertson [43] identified, through yeast one-hybrid screens, a barley NAC transcription factor, HSINAC, interacting with HvSPY. HvSPY is a negative regulator of gibberellin signaling. This NAC gene was found to be highly expressed in shoot, mature blade, sheath, and mature aleurone, and lowly expressed in young blade and stem, and was barely detectable in coleoptile and root. Jensen et al. [44] published their results on the barley HvNAC6 transcription factor, a homologue of Arabidopsis ATAF-1. They presented evidence of a role for HvNAC6 in penetration resistance to powdery mildew fungus. Besides being up-regulated during powdery mildew infection, no other expression data was presented.
Ogo et al. [45] published a comprehensive study on a rice NAC transcription factor called IDEF2. IDEF2 was suggested to be involved in iron homeostasis and was identified by its specific binding to promoter DNA containing an iron deficiency response element (IDE). A barley IDE was used in the same way to identify a barley homologue of IDEF2 called HvIDEF2. It was isolated from a root cDNA library, but otherwise no expression data was presented.
Finally, there have been two publications on the barley HvNAM-1 [12,46] encoding a homologue of the wheat TtNAM-B1 NAC transcription factor shown to regulate both senescence and grain protein content (GPC) [11]. Gene polymorphisms of HvNAM-1 in three Hordeum species explained some of the variation in GPC [12] suggesting that the biological function of HvNAM-1 is similar to that of TtNAM-B1.
Barley (Hordeum vulgare L.) is an important food and feed crop worldwide. Besides its agronomic importance, barley has also been used as the model species of choice for many researchers in the field of small grain cereal crops. Although its genome has not yet been fully sequenced, an extensive EST (expressed sequence tags) collection is available containing samples from a wide variety of tissues, different developmental stages, and pathogen infections, as well as an increasing amount of microarray data from the Affymetrix Barley1 GeneChip (http://www.plexdb.org/plex.php?database=Barley). Considering that many of the NAC transcription factors in barley will most likely turn out to be key regulators of important agronomic traits, as they are in other species, studies of this gene family are pertinent.
This work presents the first phylogenetic classification of H. vulgare NAC genes (HvNACs). 48 individual HvNACs were identified with 43 of them representing full-length coding sequences. Specific, quantitative realtime PCR (qRT-PCR) primers could be designed for 46 HvNAC genes, and the expression profiles of these were investigated in eight tissues of the barley plant as well as after two different hormone treatments of leaves with ABA and methyl jasmonate (MeJA). In conclusion, the results provide good evidence that also in barley the NAC gene family is involved in regulatory pathways of importance for agronomic traits.

HvNAC genes
Searching all presently available barley nucleotide sequences for features of the characteristic NAC domains resulted in a list of 48 barley NAC genes either based on full-length cDNA, genomic sequences, or EST contigs. Nine of the genes were represented by only partial coding sequences based on homologous proteins in Brachypodium and rice. In order to obtain more fulllength coding sequences (CDS), partial length EST clones were resequenced or cloned and sequenced when possible. This procedure reduced the number of partial length genes from 9 to 5. The complete list of established barley NAC genes is shown in table 1. The barley NAC genes were named according to the recommendations by Gray et al. [47]. The numbering of the genes was arbitrary except for a few sequences, e.g. HvNAC001 and HvNAC006, where annotation had already been made for submitted sequences. Sequence alignments of the HvNACs and all NAC proteins known to contain a C-terminal transmembrane anchor revealed four barley NACs with this motif: HvNAC002, HvNAC007, HvNAC016 and HvNAC048.

Phylogenetics
The HvNACs were classified according to the system proposed by Shen et al. [2] based on positions in a phylogenetic tree made from the alignment of the conserved NAC domains of ONACs, ANACs, and HvNACs (additional file 1). The phylogenetic tree also includes Brachypodium distachyon NACs (BdNACs) as well a few selected NACs from other species discussed in this paper. Table 2 shows the distribution of HvNACs among the eight NAC subfamilies. In addition to the NAC subfamily classification, Shen et al. [2] further divided the NAC transcription factors into subgroups based on homology of C-terminal regions. By referring to the subgrouping of the ANACs and ONACs, each of the full-length HvNACs could in most cases be assigned to a NAC subgroup as well. These subgroups are listed in table 1. This table also lists putative orthologues of each HvNAC among BdNACs, ONACs, and ANACs. Through full-length protein alignments, the possible orthologues were identified as having a highly conserved NAC domain as well as one or more conserved C-terminal motifs. Figure 2 illustrates the identification of Cterminal motifs for putative orthologues exemplified by HvNAC033 and HvNAC011. In both cases at least three conserved, smaller motifs can be observed. After List of barley HvNACs with EMBL/GenBank accession numbers. The length of each established gene transcript is given as either full or partial coding sequence as compared to wheat or Brachypodium homologues. Phylogenetic grouping was performed according to Shen et al. [2] and was based only on the NAC domain of the encoded protein. For HvNACs with no close rice or Arabidopsis homologues, only the subfamily is noted. The putative orthologues are based on full-length protein alignments and the presence of conserved motifs in the C-terminal. No orthologues are listed for the partial length proteins. The percent identity, on fulllength amino acid level, between the barley NAC protein and the orthologue in question is given in parenthesis. Numbers in square brackets in the gene name column represent references in the bibliographic reference list.
identifying each of the possible orthologues through visual inspection of full-length protein alignments, their percent identity was calculated as noted in table 1.

HvNAC gene expression studies
In order to extend our characterization of the HvNACs, tissue-specific gene expression studies using qRT-PCR were performed. One of the partial length sequences HvNAC047 was excluded from this experiment, since the design of a specific set of qRT-PCR primers was not possible. HvNAC030 had aberrant expression, based on the dissociation curves, in most of the tissues investigated, and so it was excluded after analysis of the qRT-PCR data. The individual tissues were selected to cover a range of plant organs. As several NAC genes from rice and Arabidopsis have been shown to be hormone and stress responsive, treatments with the two stress hormones ABA and MeJA were included in the expression studies as well. The data collected from these experiments are presented in figure 3. The raw data from the qRT-PCR experiment, as well as calculated p-values, can be found in additional file 2. The relative expression levels in each tissue were determined using the Limma software with non-senescing flag leaf tissue as the reference. Prior to construction of the heatmap for the tissue experiment in figure 3, the relative expression values for each gene were re-scaled to have mean equal to zero in order to bring the different expression patterns into the same colour range centred on zero. We have included in the figure one non-NAC gene Rubisco, encoding the Rubisco small subunit, as a control that reflects the amount of green photosynthesizing tissue. From figure 3 Rubisco is, as expected, found to be up-regulated in young flag leaf, young ear and stem, and strongly downregulated in roots.
The eight tissues were selected in order to identify HvNAC genes associated with processes already known to be NAC regulated in other species. SCWB is represented here by the elongating stem tissue, and we see a small group of just three genes (HvNAC033, HvNAC034, and HvNAC039) clearly up-regulated. Leaf senescence is represented by the senescing flag leaf, where a number of genes are up-regulated. If we observe the old ear and late dough grain as tissues, where the senescence process is also taking place, as well as ABA induction, since ABA is a known senescence inducer [48], three genes stand out as putative senescence regulators: HvNAC005, HvNAC027, and HvNAC029. A few NAC transcription factors regulate root development, a process represented here by the root tissue. Only three genes are highly up-regulated in this tissue: HvNAC021, HvNAC022, and HvNAC032. The old ear, milk grain, and late dough grain all represent various stages of seed development and as such could help identify genes involved with regulation of this. Looking at these tissues together, four genes stand out quite clearly: HvNAC017, HvNAC018, HvNAC019, and HvNAC024.
In the hormone treatment experiment, seven genes were excluded from the heatmap due to expression data of poor quality (aberrant dissociation curves at low expression levels). The non-senescing flag leaf samples showed low expression for all of the excluded genes, and the poor quality might have been caused by an even lower expression in the used plant material, which was from young flag leaves harvested at an earlier developmental stage than the non-senescing flag leaves in order to avoid environmental stresses. With the two hormone treatments it is noticeable that only a few genes were up-or down-regulated after three hours of treatment, whereas more significant changes were observed after five hours. There appears to be some phylogeny correlations in the genes induced by ABA. All members of the NAC-a-6 subgroup, HvNAC005, HvNAC023, and HvNAC027, are significantly up-regulated by ABA, whereas both members of the closely related NAC-a-8 subgroup, HvNAC003 and HvNAC004, are down-regulated. All members of the NAC-d-7 subgroup, HvNAC025 and HvNAC032, are up-regulated as well. The remaining ABA-induced genes do not seem to be systematically induced, in terms of phylogeny, which seems also to be the case for all the MeJA-induced genes. Of all the genes HvNAC027 stands out as being highly responsive to both hormones.
Data from experiments using the Affymetrix 22 K Bar-ley1 GeneChip provide a wealth of information on gene expression in barley (http://www.plexdb.org/plex.php? database=Barley). However, comparisons of our data to Barley1 GeneChip data are complex, since, first, two important tissues in our study the naturally senescing flag leaf and the developing stem are not included in the Barley1 GeneChip experiment on development stages in barley [49], and, second, only 20 of the HvNAC genes  indicating strong conservation (non-identical aa, but with highly similar physico-chemical properties) and '.' indicating weak conservation (nonidentical aa, but with somewhat similar physico-chemical properties). B: Same as A, but for HvNAC011 and its homologues.   In brief, the presented expression data are based on three (tissues) or two (hormone treatments) biological replicates, each with three technical replicate PCR runs, followed by normalization to the expression of the 18 S ribosomal RNA gene. Prior to construction of the heatmap for the tissue experiment, the relative expression values for each gene were re-scaled to have mean equal to zero. Asterisks in the hormone experiments indicate significant differences at the 0.05 level between treatment and control (p-values adjusted for multiple comparisons according to default settings in the Limma software).
genes will be identified before the barley genome is fully sequenced and made available. Again, as the barley EST databases are extensive, it could be argued that the most essential HvNACs with regard to plant development, abiotic and biotic stress responses are probably among the 48 genes identified in this study. However, a gene with rare or very low expression does not equal a nonessential product, and thus important NAC genes could still be absent from EST collections.

Phylogenetics
A number of different suggestions for the classification of NAC transcription factors have been reported [2,5,9,10]. The most comprehensive one by Shen et al. [2] was adopted in this study on barley NAC genes. For identification of putative Brachypodium, rice, and Arabidopsis orthologues, full-length protein sequences with closely related NAC domain regions were aligned to reveal the C-terminal motifs. Generally, there is low conservation in the C-terminal region of the NAC proteins, however, when aligning orthologues across species, the presence of small C-terminal motifs becomes apparent. Table 1 shows putative orthologues, as well as their percent identity of each HvNAC, based on both a highly conserved N-terminal NAC domain and on conserved motifs in the C-terminal. Seven HvNAC genes encode both a NAC domain and C-terminal motifs conserved throughout Brachypodium, rice, and Arabidopsis. The orthologous NAC proteins among the grasses are highly similar, often sharing with barley more than 70% (for Brachypodium) or 60% (for rice) of their amino acid sequence. It was not possible to identify clear Arabidopsis orthologues for many of the HvNACs. However, there were a few which showed high conservation. In these cases the percent identity was above 50%. It could be speculated that NAC genes showing conservation throughout the monocots and dicots could be involved in important regulatory pathways. In support of this, five out of the seven Arabidopsis NAC proteins identified as HvNAC orthologues have been characterized in the literature and in all cases found to have profound effects on the plants if overexpressed or knocked out [13,41,[50][51][52].

HvNAC expression
To validate our expression data, we included Rubisco as an internal control, which showed an expression pattern as expected, however, further validation can be done by comparing our results with previous reports. HvSF6 (HvNAC008) was found to be up-regulated in senescing tissue [42], matching entirely with our data. HvSF6 was further seen to be induced by a combination of ethylene and MeJA, but not MeJA alone [42], which also agrees with the lack of induction by our MeJA treatment.
HSINAC (HvNAC031) was found to be up-regulated in mature blade and mature aleurone [43], fitting well with the up-regulation we observed in the senescing leaf and late dough grain, respectively. Furthermore, a low expression in young blade and stem was reported. We also observed low expression in the stem tissue, however, our young flag leaf tissue showed relatively high expression. It is possible that our young, but fully developed, flag leaf sample resembles their mature blade more than their young blade. Finally, Robertson [43] reported barely detectable expression levels in the roots, while we observed root levels similar to the low expression in the stem.
No expression patterns were reported for HvIDEF2 (HvNAC028), besides it being isolated from a root cDNA library [45]. Interestingly, this gene was one of the most stably expressed genes across all eight tissues. Likewise, no HvNAM-1 (HvNAC029) induction was reported [12,46]. However, we did observe a significant induction in the senescing flag leaf, which would be expected for an orthologue of TtNAM-B1.
The overall results presented in figure 3 emphasize the suggested involvement of NAC transcription factors as regulatory factors in a range of processes during plant development and stress responses. We will focus here on a few selected genes and discuss their expression patterns in comparison to the known and suggested functions of phylogenetically related NAC genes.
SCWB has been the focus of several NAC gene studies, and many NAC genes have been shown to regulate aspects of this process. HvNAC039 is clearly up-regulated in the developing stem tissue, where extensive SCWB is expected to take place. Considering that HvNAC039 belongs to subgroup NAC-g-9, as do both SND2, SND3 (ANAC073 and ANAC010, respectively), and PopNAC154, a role in SCWB is not unlikely for this NAC gene. HvNAC033 and HvNAC034 are also upregulated in the stem tissue and may thus regulate aspects of the SCWB as well. Protein alignments identified HvNAC033 as a possible orthologue of both NST1 (ANAC043) and MtNST1. Furthermore, considering the expression data, HvNAC033 is very likely an NST orthologue. HvNAC034 is very closely related to HvNAC033, both belonging to subgroup NAC-c-3, and therefore also related to NST1 and MtNST1. Both HvNAC033 and HvNAC034 are also up-regulated in the developing young ear, where a high degree of SCWB is also expected to take place, supporting the hypothesis that they could be regulators of this process.
Leaf senescence has also been the focus of several NAC gene studies. There are several HvNACs which are up-regulated in the senescing flag leaf, however, by including the relative expression levels across the old ear and late dough grain tissues, three genes stand out: HvNAC005, HvNAC027, and HvNAC029. HvNAC029 is an orthologue of NAM-B1 from wheat and has furthermore been mapped to a quantitative trait locus (QTL) for leaf senescence in barley [46], clearly supporting the hypothesis of a role for this transcription factor in leaf senescence regulation. HvNAC005 and HvNAC027 both belong to subgroup NAC-a-6, as does the Arabidopsis senescence regulator AtNAP (ANAC029), suggesting a possible link to senescence. Both are also found to be up-regulated following ABA treatment, adding further evidence, as ABA is known to be an inducer of senescence [48].
Three highly up-regulated HvNAC genes were found in the root tissue: HvNAC021, HvNAC022, and HvNAC032. This could indicate roles in root development, especially as they all belong to subfamily NAC-d, as do both NAC1 (ANAC021) and ORE1 (ANAC092). Although it is possible that all three are involved in regulating root development, HvNAC022 stands out as being phylogenetically very closely related to NAC1 as well as being down-regulated in the presence of ABA, which is known to have an inhibitory effect on lateral root formation [53]. Using the online GeneVestigator software (http://www.genevestigator.com/gv/), it was confirmed that Arabidopsis NAC1 is also down-regulated by ABA in an experiment very similar to the one presented here (Experiment ID: AT-00231). These results support the hypothesis that at least HvNAC022 is a putative regulator of root development.
HvNAC017, HvNAC018, HvNAC019 and HvNAC024 form, together with HvNAC009, a small barley-specific group of NAC transcription factors with no clear orthologues in other plant species investigated so far. The gene expression patterns in the data presented here indicate that these four genes could regulate aspects of seed development. This hypothesis might be supported by results of Guo et al. [54] and Verza et al. [55]. They reported on the expression of two maize NAC genes ZmNRP-1 and ZmAPN-1, respectively. Both of these genes are closely related to the barley specific group, although they do not appear as direct orthologues. It was found that ZmNRP-1 had endosperm specific expression [54] and that ZmAPN-1 expression was restricted to the aleurone cell layer [55].
Many NAC genes that have been found to be involved in various forms of stress responses and tolerances, and very often the same genes are also found to be induced by stress hormones. In order to get an initial overview of the involvement of the HvNAC family in stress regulation, we tested the effects of treatment with the two stress hormones ABA and MeJA on HvNAC expression. Notably, most of the significant changes happened after five hours of treatment. This could either be due to a secondary wave of gene induction or to generally slow inductions, perhaps related to our experimental setup. Another observation from our data is that some of the ABA-induced genes were phylogenetically related, i.e. belong to the same subgroup. HvNAC027 stood out as being highly responsive to both hormones, suggesting it could have a putative role in stress tolerance regulation. TaNAC69 from wheat, a direct orthologue of HvNAC027, displays a similar induction from ABA as well as induction by drought and cold [56].

Conclusions
In this work we have compiled available sequences from public databases that represent 48 members of the NAC gene family in barley. The list of barley NAC genes is, evidently, not exhaustive, since a number of additional members exist in NAC gene families of the closely related species Brachypodium and rice.
The results presented here support the idea that the functional roles of NAC transcription factors are conserved throughout both monocot and dicot species. Furthermore, they clearly demonstrate that the qRT-PCR approach for an initial characterization of the barley NAC family was a good choice. Most of the genes selected as putative regulators of particular processes, based on the qRT-PCR results, were either direct orthologues of, or at least very closely related to, NAC genes from other species, specifically known to regulate these processes.

Identification of HvNACs
All publicly available barley nucleotide sequences were collected from the Nucleotide, GSS, EST and UniGene databases of the National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/) and Plant Genomic Database (http://www.plantgdb.org/). These sequences were searched for the presence of barley NAC gene sequences using the tBLASTn algorithm. The NAC domain of all available Brachypodium distachyon and Oryza sativa NAC protein sequences were used as input for the BLAST function. Contig assembly was performed using the CAP3 algorithm [57]. As the vast majority of the collected sequences were in the form of expressed sequence tags (ESTs), each contig was manually inspected due to the error prone nature of ESTs. In the case that a contig was considered to be erroneous based on the translated protein sequence, all ESTs of the contig were carefully inspected and the contig manually corrected, if applicable. In order to obtain more full-length CDS sequences, partial length EST clones were re-sequenced or cloned and sequenced when possible. Primers used for cloning were based on homologous wheat or Brachypodium sequences.

Phylogenetic analysis
All phylogenetic analyses were performed on the NAC domain part of each gene. Each HvNAC protein sequence was manually truncated immediately following the E subdomain as defined in Ooka et al. [5]. Sequences lacking a defined E subdomain were truncated based on multiple alignments of all HvNACs proteins. All alignments and phylogenetic trees were made using ClustalX 2.1 [58]. The phylogenetic tree in additional file 1 was drawn using the ape add-on package [59] of the R software [60].

Plant material and treatments
Barley plants, cv. Golden Promise, were grown in greenhouse soil plots. Artificial illumination was used for supplementation and for ensuring a day/night cycle of 16/8 hr. All sampling were done between 12 noon and 2 pm, and samples were frozen in liquid nitrogen immediately after being harvested. For all tissue samples, three biological replicates were harvested and processed in parallel. The samples were stored at -80°C until RNA extraction was performed. Plants from which the roots were sampled were grown in individual pots containing perlite (Nordisk Perlite, Denmark), and sampling was performed when the plantlets were at the three leaf stage.
Young flag leaf material was harvested at the heading stage when the leaf was fully developed, and senescing flag leaves when green leaf area was reduced to approximately 50%. Young ear material was harvested immediately after heading, and the old ear material when the seeds were close to full maturity. The milk and late dough grain material comprised individual grains in the early milk stage and late dough stage, respectively. The stem samples comprised the lower approximately 5-6 cm of culms from the second internode from the top harvested just prior to heading. Young, light green leaves from small plantlets were sampled for genomic DNA (gDNA) isolation used for qRT-PCR primer testing.
For treatments with ABA and MeJA, young flag leaves were taken just prior to heading. In order to encounter developmental differences, the basal 10 cm of the leaves was divided into six pieces that were distributed across different time points of the time-course study. Hormone treatments were done in 10 ml water containing 50 μM hormone (dissolved in DMSO) and 0.005% Triton-X 100. Controls were treated with 0.005% Triton-X 100 and 0.1% DMSO. Two replicate samples comprising five leaf pieces were taken for RNA isolation at 0, 1, 3, and 5 hours after the unset of hormone treatment. At 0 hours only the control sample was taken.

RNA isolation and cDNA synthesis
RNA isolation was performed on approximately 100 mg of frozen homogenized material using the Spectrum™ Plant Total RNA kit (Sigma) according to the manufacturer's recommendations. First-strand cDNA synthesis was performed on 3-5 μg of RNA and a nonamer, random oligonucleotide primer (2.5 μM) by incubation at 65°C for 5 min followed by 10 min at RT in a volume of 18,4 μL. 200 U SuperscriptII (Invitrogen), 40 U RNAsin (Promega), 1 × FS Buffer (Invitrogen), 10 μM dichlorodiphenyltrichloroethane 1,4-dithiothreitol (DDT), and 2 mM dNTPs (GE Healthcare) were added to make a final volume of 30 μl, which was incubated for 1 hour at 42°C and 10 min at 70°C, followed by the addition of 70 μl of water.

Primer design and efficiency
To ensure specificity of each HvNAC primer pair, the online Primique software [61] was used to design primer sets for qRT-PCR. All other primers were designed with Oligo Explorer 1.2 from Gene Link, Inc. qRT-PCR primer sequences are listed in additional file 3. All primers were tested for their efficiency on gDNA. gDNA was used instead of cDNA, since many of the HvNAC genes were expected to have low expression levels. As such, measures were taken to avoid intron spanning amplicons during the design of the primers. Four dilutions of gDNA were used in the testing: 1, 1/8, 1/64, and 1/512 as well as a zero template control. Average values for three technical replicates were plotted against the concentration, and the efficiency was calculated from the slope of the best linear regression, using the formula: efficiency = 10( -1/slope) [62]. Efficiencies of 2 ± 0.2 were acceptable. Primers were also tested without any template to ensure that no primer dimer products would occur.

qRT-PCR
Quantitative real-time PCR was performed in 384 well plates using the ABI Prism7900HT Sequence Detection System with the Power SYBR ® Green PCR master mix (Applied Biosystems) according to the manufacturer's recommendations. 0.5 μl of cDNA was used for each qRT-PCR reaction and all tests were performed with three technical replicates. For each qRT-PCR run a dissociation stage was included, in order to be used in quality assessments of each sample during data analysis. The threshold cycle (Ct) for each run was determined using the automatic settings for baseline and threshold level of the ABI Prism 7900HT SDS software.

Analysis of qRT-PCR data
Ct values exported from the ABI Prism 7900HT SDS software were used as raw data for the analysis of qRT-PCR data. The R software [60] and the add-on packages HTqPCR [63] and Limma [64] were used for the manipulation and analysis of the Ct values, and for generation of heatmaps of gene expression data. qRT-PCR runs showing high variation among technical replicates were manually inspected, and clear outliers and runs with aberrant dissociation curves were excluded from the analysis. Several possible reference genes were tested for their stability across different tissues, using the tools in the R package SLqPCR [65]. Based on this analysis, the barley 18 S ribosomal RNA gene was selected as the most stable reference gene to be used in the normalization of gene expression of NAC genes in different tissues and treatments.

Additional material
Additional file 1: Phylogenetic tree. Suppl_file1_Phylogenetic_tree.pdf. Phylogenetic tree of all HvNACs, BdNACs, ONACs, ANACs, and a few selected NAC genes from other species discussed in the study.
Additional file 2: Data and statistics for qRT-PCR results. Suppl_file2_qRT-PCR data. Normalized qRT-PCR data used for heatmap construction, including statistics on significance.
Additional file 3: qRT-PCR primers. Suppl_file3_qRT-PCR primers. Primer sequences, presented in a 5' to 3' direction, for all qRT-PCR primers used in this study.