Introduction

The grasses (Poaceae) form a monophyletic family of monocotyledonous plants which includes all the cereal crops, notably rice (Oryza sativa L.), maize (Zea mays L.), wheat (Triticum aestivum L.), barley (Hordeum vulgare L.), and sorghum (Sorghum bicolor L.). These cereals share a common ancestor from which they have diverged over a period of some 60 million years ago (Devos 2005); nevertheless, some synteny has been retained between them (Devos 2005; Gale and Devos 1998; Lu and Faris 2006). For example, rice chromosome 4 and 7 align well with chromosome 2 of barley and wheat (Chen et al. 2009; Devos 2005; Moore et al. 1995). With the complete rice genomic sequence to hand (International Rice Genome Sequencing Project 2005), it has become possible to demonstrate both where collinearity has been retained at the fine-scale level (Bennetzen and Ma 2003; Bossolini et al. 2007; Faris et al. 2008; Srinivasachary et al. 2007; Yan et al. 2003), and where it has collapsed as a result of inversions, deletions, duplications, and other intrachromosomal rearrangements (Ilic et al. 2003; La Rota and Sorrells 2004; Li and Gill 2002; Liu et al. 2006; Tarchini et al. 2000). Other full grass species genome sequencing project either completed or underway include those for sorghum (Paterson et al. 2009; Sasaki and Antonio 2009) and Brachypodium distachyon, a small genome, short growth cycle, self-fertile, model temperate grass(Ozdemir et al. 2008).

Inflorescence structure is one of the main determinants of grain yield in the cereals. The inflorescence can take the form of a panicle (rice, sorghum, and maize) or a spike (wheat, barley, and Brachypodium). Some evidence supports the notion that the spike has evolved from the panicle (Vegetti and Anton 1995). The barley spike carries a set of three spikelets at each rachis node. In “two-rowed” barley, the two lateral spikelets are reduced in size and sterile, but in the “six-rowed” type, all three spikelets are fertile. The six-rowed phenotype is genetically determined by homozygosity for the recessive allele (previously referred to as vrs1) at the vrs1 locus, which has been identified as a homeobox gene (HvHox1) encoding a transcription factor containing a homeodomain (HD) with a leucine zipper motif (Zip; Komatsuda et al. 2007). HD-Zip proteins have been grouped into four families (Ariel et al. 2007), with the Vrs1 gene product (VRS1) belonging to the family I. Although HD can be found in all eukaryotic genomes, the HD-Zip family is restricted to the plant kingdom. The HD-Zip protein is dimerized by the Zip domain, and uses the HD to bind specifically to dyad-symmetrical DNA recognition sequences, based on the strict spatial relationship between HD and Zip (Sessa et al. 1993). VRS1 is thought to suppress the development of the lateral spikelets, since its expression was restricted to the lateral-spikelet primordia in the immature spikes (Komatsuda et al. 2007). The loss of Vrs1 function resulted in the complete conversion of the rudimentary lateral spikelets of a two-rowed barley into fully developed fertile spikelets, just as in the six-rowed type.

Phylogenetic analysis demonstrated that the origin of the six-rowed phenotype was probably polyphyletic, both temporally and spatially, and occurred via a series of independent mutations at the Vrs1 (Komatsuda et al. 2007). The higher seed set of the six-rowed type would have been readily selected during the domestication process (Harlan et al. 1973). Micro-collinearity between rice and barley is disrupted in the Vrs1 region, but a Vrs1 ortholog has been identified on rice chromosome 7 (Pourkheirandish et al. 2007). The barley EST (scsnp06322), mapping to the centromere region of chromosome 2H, is homologous to rice Os07g0581000 (LOC_Os07g39280), which co-locates with the rice Vrs1 ortholog Os07g0581700 (LOC_Os07g39320), (Pourkheirandish et al. 2007; Rostoks et al. 2005). This genomic location suggests the original site of Vrs1 to be the centromere region of chromosome 2H prior to the chromosomal rearrangement, which has been responsible for the local loss of synteny between rice and barley, but it is plausible that Vrs1 evolved as a ‘copy’ of an indispensable ‘master’ gene, which is still present in its ancestral location on chromosome 2H (Pourkheirandish et al. 2007). Neither the structure nor the function of Vrs1 orthologs in any of the other Poaceae members has been elucidated. The objective of this study was to compare the genomic organization of the regions containing a Vrs1 ortholog in a set of Poaceae species, as a means of inferring the refinement of the function of Vrs1 by gene duplication in the speciation of barley.

Materials and methods

Plant materials

The two-rowed barley cv. Kanto Nakate Gold (KNG, NIAS accession number JP 15436) and the six-rowed barley cv. Azumamugi (AZ, JP 17209; maintained in the Gene Bank, NIAS, Tsukuba, Japan) were intercrossed to allow the development of a population of 99 F12 recombinant inbred lines (RILs).The wild barley (H. vulgare ssp. spontaneum) strain OUH602 was obtained from Research Institute for Bioresources, Okayama University, Kurashiki, and used to generate a population of 186 F7 RILs from the cross OUH602 × KNG. A set of chromosome addition lines (CALs), in which six of barley cv. Betzes chromosomes (2H–7H) are present, in turn, in a background of bread wheat cv. Chinese Spring (Shepherd and Islam 1981) were kindly provided by Dr. A. K. M. R. Islam, Department of Plant Science, Waite Institute, University of Adelaide, Australia.

Barley full-length cDNA library

Seedling shoots and roots of cv. Haruna Nijo were used as a source of mRNAs to construct full-length cDNA libraries, following the methods described by Carninci et al. (1996). A sample of clones was end-sequenced (both 5’ and 3’). A detailed description of this library and its construction will appear elsewhere.

Searching for Vrs1 orthologs in Poaceae

Nucleotide-BLAST (BLASTN), protein–protein BLAST (BLASTP), and translated nucleotide-protein BLAST (TBLASTN) searches were made against the following sequence databases: barley, Barley Full-Length cDNA End Sequence Database of NIAS (unpublished); rice, Rice Annotation Project Database (http://rapdb.dna.affrc.go.jp/) and The Institute for Genomic Research (TIGR) Rice genome annotation (http://rice.plantbiology.msu.edu/); maize, MaizeSequence.org (http://www.maizesequence.org/index.html); sorghum, Department of Energy Joint Genome Institute (JGI) S. bicolor (http://genome.jgi-psf.org/Sorbi1/Sorbi1.download.html); B. distachyon, BRACHYPODIUM.ORG (http://www.brachypodium.org/); wheat, TIGR Wheat Genome database (http://www.tigr.org/tdb/e2k1/tae1/), and NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Phylogenetic and peptide motif analysis

Sequence data were aligned using ClustalW2 software (http://www.ebi.ac.uk/Tools/clustalw2/). A phylogenetic tree was constructed by the neighbor-joining method, using PAUP 4.0b10 software (Sinauer, Sunderland, Massachusetts) employing 100 bootstrap replicates. Insertion/deletion characters were not included. Peptide motifs were analyzed using the Surveyed conserved motif ALignment diagram and the Associating Dendrogram (SALAD) database (http://salad.dna.affrc.go.jp/salad/en/; Mihara et al. 2008). A graphical display of motif composition was constructed using Interactive SALAD analysis.

Barley ESTs data

Barley ESTs giving the best match to rice genes on chromosome 7 were selected from the Gramene database (http://www.gramene.org/Oryza_sativa_japonica/index.html). The copy number of each EST was investigated using Plant Repeat Database at Michigan State University (http://plantrepeats.plantbiology.msu.edu/index.html). Exon–intron junctions were assumed to be conserved between rice and barley, and were extracted from NCBI BLAST 2 SEQUENCES (http://blast.ncbi.nlm.nih.gov/bl2seq/wblast2.cgi).

PCR amplification

Plant genomic DNA was extracted as described by Komatsuda et al. (1998). PCR primers were designed from the predicted exon regions with Oligo5 software (W. Rychlick, National Bioscience, Plymouth, MN, USA) and synthesized commercially (BEX, Tokyo, Japan) (Supplementary Table 2). PCR amplification was carried out in 10μl reactions containing 0.25 U ExTaq polymerase (Takara, Tokyo, Japan), 1× ExTaq polymerase buffer, 0.3 μM of each primer, 200 μM dNTP, 2 mM MgCl2, 0–5% v/v dimethyl sulphoxide (DMSO), and 20 ng genomic DNA. Each PCR was cycled through a denaturation step (94°C/5 min), followed by 30 cycles of 94°C/30 s, 55–65°C (primer-dependent)/30 s, 72°C/30–90 s with a final incubation of 72°C/7 min. Amplicons were electrophoresed through either agarose (Agarose ME, Iwai Kagaku, Tokyo, Japan) or a MetaPhor agarose (Cambrex Bio Science Rockland Inc., Rockland, MA, USA) gels, depending on their size, and were visualized by ethidium bromide staining.

Development of CAPS and dCAPS markers

PCR products were purified using the QIAquick PCR purification Kit (Qiagen, Germantown, MD, USA) and subjected to cycle sequencing using a Big Dye Terminator Kit (Applied Biosystem, Foster, CA, USA). Sequencing reactions were purified by Agencourt CleanSEQ (Beckman, Beverly, MA, USA), and analyzed with an ABI prism 3130 genetic analyzer (Applied Biosystem). Sequence data were aligned by Sequencher DNA Sequencing Software (HitachiSoft, Yokohama, Japan). Polymorphic restriction sites were identified via the Restriction Maps option of Molecular Toolkit (http://arbl.cvmbs.colostate.edu/molkit/mapper/) or with dCAPS Finder 2.0 (http://helix.wustl.edu/dcaps/dcaps.html). PCR products were digested at the recommended temperature for 2–3 h in reactions of 15 μl of reaction mixture containing 10 μl PCR products, 1× reaction buffer, and 1 U restriction enzyme.

Genetic mapping

The AZ × KNG and OUH602 x KNG RILs population were genotyped using polymorphic CAPS and dCAPS markers, and MAPMAKER/EXP ver.3.0 (Lander et al. 1987) was used to integrate the resulting loci into the linkage map constructed by Komatsuda and Tanno (2004).

RNA extraction and RT-PCR assay

Total RNA was extracted from various tissues by using TRIzol (Invitrogen, Carlsbad, CA). A first-strand cDNA was synthesized with SuperScript II (Invitrogen). The first-strand cDNA was used as a template, and amplification was performed for 30 or 40 PCR cycles (1 min at 94°C, 30 s at 60/65°C, 30 s at 72°C) followed by 7 min at 72°C. The primers for HvHox2 were 5′-GCGTGGTCGAGTGGTTTAGCCTGT-3′ (sense) and 5′-GAGAGCTACCGGTACTACACTTGC-3′ (antisense). Vrs1 primers were 5′-GGTTTTTAGCATGAATTAGAGTTTA-3′ (sense) and 5′-TATACAGGCTAAAAACCAAAGATTA-3′ (antisense). Actin primers were 5′-GTCCTTTTCCAGCCATCTTTC-3′ (sense) and 5′-CAAGAATCGACCCTCCAATCC-3′ (antisense). RT-PCR assay was performed at least twice for each sample.

Results

Identification of a Vrs1 paralog in barley

The best hit (E-value, e-110) from a BLASTN search based on the Vrs1 cDNA sequence (accession no. AB259783) as a query was the barley full-length cDNA clone Hv2074A10 (accession no. AB490233) extracted from the seedling shoots and root of cv. Haruna Nijo (NIAS barley database). Four lesser hits (E-values < e-10) were also identified from the same library. The search using Vrs1 HD-Zip coding region as the query sequence hit three entries (E-values < e-10) in the same database. Vrs1 and Hv2074A10 shared an identical exon/intron structure (Supplementary Fig. 1). The second and third exons of both genes contained homeodomain-leucine zipper motif. As a result, the Hv2074A10 was named “HvHox2”. Alignment of the Vrs1 and HvHox2 cDNA sequences identified a 300-bp insertion in Vrs1 and a 44 bp insertion in the third exon of HvHox2 (Supplementary Fig. 2). Neither the 300 nor the 44 bp sequences were homologous to any accession in the public domain. The Vrs1 insertion resulted in the generation of a stop codon, producing a polypeptide 14 residues shorter than the HvHox2 product (Supplementary Fig. 3). The deduced polypeptide sequences of the HvHox2 (HvHOX2, 236 aa) and Vrs1 (VRS1, 222 aa) shared 88% identity in the HD-Zip domain and 84% identity in the whole protein.

Identification of Vrs1 homologues in the Poaceae

The Vrs1 cDNA sequence was used to query each species database in turn. Vrs1 homologues were present in rice (Os07g0581700, Oshox14), B. distachyon (Bradi1g23460.1), sorghum (Sb02g03750), and maize (AC216056.2_FG006 and AC187394.3_FG016). The presence of two (rather than just one) Vrs1 homologues in maize probably reflects the cryptic tetraploidy of the maize genome (Paterson et al. 2004; Swigonova et al. 2004; Wei et al. 2007). BLASTN search using HvHox2 as the query sequence produced the same hits. All the Vrs1 homologues contained an HD-Zip motif (Supplementary Fig. 3).

Phylogenetic analysis of VRS1 homologues

Polypeptide sequences matching that of VRS1 were selected by BLASTP or TBLASTN search against the wheat, B. distachyon, rice, sorghum, and maize databases (Supplementary Table 1). The top five sequences with significant similarity to VRS1 were selected from each taxon to form a phylogenetic tree. Arabidopsis thaliana HD-Zip family (classes I–IV) proteins were included to infer the class to which the Poaceae proteins belonged. The resulting neighbor-joining tree showed a clear separation of the four families of HD-Zip families, with each clad being strongly supported by bootstrap analysis (Fig. 1). All the VRS1 homologues clustered within the HD-Zip I family (along with A. thaliana ATHB40, ATHB21, and ATHB53), suggesting that these sequences are orthologous. The high degree of similarity between VRS1 and HvHOX2 (bootstrap support level 88%) suggests that they became separated from one another after the separation of barley (Fig. 1). The sister clad, containing the rice, B. distachyon, wheat, sorghum and maize sequences, was also highly supported (100%). The peptide motifs 1 and 7 within VRS1 form the homeodomain and motif 2 forms the leucine zipper (Fig. 2). Motif 3, 4, and 6 were highly conserved across all the orthologs (Supplementary Fig. 4)—their function is unknown. Motif 8, in the C-terminal region of the gene, was highly conserved across all the non-barley orthologs and HvHOX2, but not in VRS1 (Fig. 2). Thus, we concluded that HvHox2, rather than Vrs1, is probably the original representative of the gene in barley.

Fig. 1
figure 1

Phylogenetic analysis of the predicted peptide sequences of the Vrs1 homologues among the Poaceae and HD-Zip family genes in A. thaliana. The phylogenetic tree was constructed by the neighbor-joining method using PAUP 4.0b10. Local bootstrap values after 100 replicates are indicated near the branching points

Full size image
Fig. 2
figure 2

Comparison between the deduced peptide sequences of VRS1, HvHOX2 and its Poaceae orthologs. The three homeodomain helical sequences are shown boxed, and a leucine zipper is indicated by staring and underlining. Hyphens indicate a gap introduced to facilitate alignment. Colored boxes show the motifs. Asterisks show the position of Ala 84 (A) and Tyr 88 (Y) in VRS1

Full size image

Genetic mapping of HvHox2 in barley

HvHox2 mapped 27.1 cM from vrs1 on chromosome 2H (OUH602 × KNG; Fig. 3). A more saturated map of the HvHox2 region was obtained by including the loci corresponding to 24 barley ESTs with homology to rice genes in the chromosome 7 containing the rice HvHox2 ortholog Oshox14 (syn. Os07g0581700). PCR primers targeting the predicted exon regions generated a single-PCR product for 20 of these 24 ESTs (Supplementary Table 2). CALs analysis assigned eight of 20 to chromosome 2H, with the remaining 12 non-assigned, because the wheat and barley amplicons co-migrated on the gels. The OUH602 and KNG amplicon sequence differed for ten of the EST sequences, but in the contrast between AZ and KNG, only one amplicon sequence was polymorphic. Based on the OUH602 × KNG RILs, five of the loci co-segregated with HvHox2, mapping to a location 0.6 cM proximal to the AV924741 and 3.7 cM proximal to ABG602 on the short arm of chromosome 2H (Fig. 3).

Fig. 3
figure 3

Micro-collinearity among the Poaceae in the Vrs1 and HvHox2 regions. The maps of Brachypodium, rice, sorghum, and maize are physical, and barley genetical. Dotted lines connect barley markers with their Poaceae orthologs. Broken lines connect HvHox2 with their Poaceae orthologs. In the barley genetic maps, OUH602 × KNG (left), AZ × KNG (right), markers shown in italics are PCR marked loci derived from RFLPs and AFLPs, while other markers are EST-derived. Markers with ‘BC’ denote barley EST clusters: BC15586 (CB872532 and CA025026), BC11013 (BF064588 and BJ453556), BC12348 (AJ468022 and CB881790), BC12063 (BU973565 and BU987455), BC14030 (CA005338 and CA019164), BC17525 (AJ486478 and BQ664265), BC09675 (AV920420and BM100120) and BC02453 (BM816591; after Pourkheirandish et al. 2007). Scale bar shows 1 Mb

Full size image

Micro-collinearity in the Poaceae

The outcome of a study of fine-scale micro-collinearity in the HvHox2 homologue region between rice chromosome 7 and B. distachyon Bd. 1 is shown in Supplementary Table 3. A series of BLASTN searches based on rice cDNA queries showed that both gene contents and orthologs order was well conserved between these two species (Fig. 3). Similarly, micro-collinearity in the HvHox2 region was conserved between rice chromosome 7, sorghum chromosome 6, and maize chromosomes 2 S and 10 (Fig. 3). Micro-collinearity was also conserved in the Vrs1 region (rice chromosome 4, B. distachyon Bd. 5, sorghum chromosome 2 and maize chromosome 2 L and 7—see Fig. 3). However, there were no Vrs1 orthologs present in this region in any of these species.

Expression analysis of HvHox2 and Vrs1

RT-PCR analysis showed that HvHox2 was expressed in all four organs tested (leaves, shoots and roots of the seedlings, and immature inflorescences in 5 mm long; Fig. 4). Vrs1 was strongly expressed in the immature inflorescences but slightly or not expressed in the other organs. There was no difference of gene expression pattern between AZ and KNG.

Fig. 4
figure 4

Expression analysis of HvHox2 and Vrs1 by RT-PCR. SP, immature inflorescence (5 mm in length); L leaves of seedling, M coleoptile node segment of seedling, R roots of seedling. A six-rowed barley cv. Azumamugi, K two-rowed barley cv. Kanto Nakate Gold

Full size image

Discussion

A duplication of HvHox2 gave rise to Vrs1

Micro-collinearity can be disrupted by inversion, tandem duplication, indel formation, or transposition (Devos 2005). Two alternative hypotheses have been proposed to account for the loss of micro-collinearity between barley and rice in the Vrs1 region: (1) the chromosomal segment containing Vrs1 was transposed from the short arm to the long arm within chromosome 2H; or (2) Vrs1 evolved from a copy of an indispensable master gene, which is still present in its ancestral location on chromosome 2H (Pourkheirandish et al. 2007). The results of the present study are supportive of the latter hypothesis, since the phylogenetic analysis suggested that HvHox2 and Vrs1 are paralog (Fig. 1), while the comparative genetic mapping showed that the ancestral copy is HvHox2, not Vrs1 (Fig. 3). The duplication of HvHox2 must have occurred after the separation of B. distachyon from the Pooideae (Fig. 1), an event which has been timed at ~35–40 million years ago (Catalan and Olmstead 2000). It seems probable that HvHox2 and Vrs1 diverged after the split between Triticum and Hordeum, because no Vrs1 homologue is represented in the wheat EST database, whereas a wheat HvHox2 ortholog has been identified. Vrs1 is present in H. vulgare ssp. spontaneum (Komatsuda et al. 2007; this study) and H. bulbosum (M. Pourkheirandish et al. unpublished) but until its distribution among Hordeum spp. more generally has been established, it is not possible to estimate how early during the evolution of this tribe the duplication event occurred. Although it is not possible to exclude the possibility of the ‘duplicated’ Vrs1 pseudogene formation in process of the speciation in the other taxa of Poaceae, a BLAST search did not detect any pseudogenes located in the Vrs1 region in other species (Fig. 3, Supplementary Table 4).

The HvHox2 sequence is highly conserved among the Poaceae

The HD-Zip I genes share intron and exon distribution (Ariel et al. 2007), and HvHox2 and its Poaceae orthologs have a similar genomic structure, including HD-Zip motif and peptide motifs (Fig. 2). The expression of HD-Zip I genes is regulated by drought, cold, and osmotic stresses, and by various hormones (Dezar et al. 2005; Himmelbach et al. 2002; Olsson et al. 2004; Rueda et al. 2005). Treatment of A. thaliana plants with either abscisic acid or salt increased the transcription of ATHB21, -40, -53 genes which are phylogenetically closest to the HvHox2 (Fig. 1). Most HD-Zip genes (including the three above) are expressed (in seedlings, roots, leaves, stems, and flowers), but a few show organ-specific patterns of expression (Agalou et al. 2008; Henriksson et al. 2005). HvHox2 was expressed in the leaves, shoots, roots, and immature inflorescences, while Vrs1 was expressed only in the immature inflorescences (Fig. 4). In situ hybridization showed that Vrs1 expression was very much localized to the lateral-spikelet primordia in spikes at the triple-mound stage and glum-primordium stages (Komatsuda et al. 2007), with zero detectable expression in young leaves (Fig. 4). While HvHox2 and Vrs1 may regulate a wholly or partially overlapping set of downstream genes, their biological functions may have been differentiated by their expression profile. On the basis of polypeptide sequences, it seems likely that the molecular role of HvHox2 and its Poaceae orthologs is identical. Within barley itself, the genome sequence of HvHox2 (accession no. AB490234) in the two-rowed type (KNG and Haruna Nijo) is the same as that in the six-rowed type (AZ and Morex) (Supplementary Fig. 1), whereas the Vrs1 sequence is remarkably variable among these same cultivars (Komatsuda et al. 2007). The lack of a Oshox14 mutant in the Tos17 insertion mutant collection (Hirochika 2001) does not contradict with the notion that HvHox2 is essential for plant growth and development in the cereals. In contrast, the loss of Vrs1 function both in natural variants, and in many induced mutants (including the total deletion of the gene) indicates that Vrs1 is non-essential for plant growth and development (Komatsuda et al. 2007).

Neo-functionalization of Vrs1

Gene duplication is closely associated with creation of novel gene functions (Force et al. 1999). Most paralogs are lost after few million years, but those which either gain a new function (“neo-functionalization”) or cover old function (“sub-functionalization”) survive. In rice, the two C-class MADS box genes OSMADS3 and OSMADS58, which were generated by a gene duplication, have been partially sub-functionalized (Yamaguchi et al. 2006). While HvHox2 probably has retained its ancestral functions, Vrs1 may have acquired the role of suppressing the lateral-spikelet development (Fig. 4). Vrs1 expression was localized to the lateral-spikelet primordia (Komatsuda et al. 2007). Presumably, this phenotype is associated with a selective advantage in nature (Pourkheirandish and Komatsuda 2007). It is unclear, however, whether the recruitment of Vrs1 to its present function occurred before or after the gain of the triplet in Hordeum.

HD-Zip proteins form dimers that recognize the pseudo-palindromic DNA sequence (Chan et al. 1998; Palena et al. 1999, 2001). Two residues in helix II, and one in the loop between helix I and helix II make contact with the target DNA (Tron et al. 2004), and these contacts are critical for aligning the recognition helix correctly. In the VRS1 protein, the glycine 84 and histidine 88 residues present in the helix II region of HvHOX2 and its Poaceae orthologs have been mutated to alanine 84 and tyrosine 88, respectively (Fig. 2). One, or both of these alterations may change either binding affinity or the target DNA sequence. However, if HvHOX2 and VRS1 still share the same target DNA sequence and retain the same level of affinity, it is possible that VRS1 competes with HvHOX2 to bind to cis-elements of downstream genes. Since Vrs1 is expressed only in the lateral-spikelet primordia, VRS1 would tend to out-compete HvHOX2 for binding in the lateral-spikelet primordia. The 300-bp insertion into the Vrs1 sequence both introduced a new stop codon (Supplementary Fig. 3) and removed a conserved peptide in motif 8 in the C-terminal region (Fig. 2). The role of the motif 8 is not annotated in the protein family (Pfam) database (http://pfam.sanger.ac.uk/), but its loss may be sufficient to differentiate the functions of the Vrs1 and HvHox2 gene products. Transcriptional co-activators enhance transcription by interacting with both general and gene-specific transcription factors (Zanetti et al. 2004). Thus, motif 8 in HvHOX2 could interact with certain classes of transcriptional co-activators, and then act as an activator for the transcription of downstream genes, while VRS1 would not be able to replicate this interaction as it lacks motif 8. The formation of HvHOX2/VRS1 heterodimers may reduce the concentration of HvHOX2 homodimer in the cell. None of the above three hypotheses are mutually exclusive, so may be combined to explain the function of Vrs1.