A SHORT INTERNODES (SHI) family transcription factor gene regulates awn elongation and pistil morphology in barley.

The awn, an apical extension from the lemma of the spikelet, plays important roles in seed dispersal, burial, and photosynthesis. Barley typically has long awns, but short-awn variants exist. The short awn 2 (lks2) gene, which produces awns about 50% shorter than normal, is a natural variant that is restricted to Eastern Asia. Positional cloning revealed that Lks2 encodes a SHI-family transcription factor. Allelism tests showed that lks2 is allelic to unbranched style 4 (ubs4) and breviaristatum-d (ari-d), for which the phenotypes are very short awn and sparse stigma hairs. The gene identity was validated by 25 mutant alleles with lesions in the Lks2 gene. Of these, 17 affected either or both conserved regions: the zinc-binding RING-finger motif and the IGGH domain. Lks2 is highly expressed in awns and pistils. Histological observations of longitudinal awn sections showed that the lks2 short-awn phenotype resulted from reduced cell number. Natural variants of lks2 were classified into three types, but all shared a single-nucleotide polymorphism (SNP) that causes a proline-to-leucine change at position 245 in the IGGH domain. All three lks2 natural variants were regarded as weak alleles because their awn and pistil phenotypes are mild compared with those of the 25 mutant alleles. Natural variants of lks2 found in the east of China and the Himalayas had considerably different sequences in the regions flanking the critical SNP, suggesting independent origins. The available results suggest that the lks2 allele might have a selective advantage in the adaptation of barley to high-precipitation areas of Eastern Asia.

Introduction

Barley (Hordeum vulgare L.) inflorescences are characterized by stiff bristles, called awns, one on the tip of each spikelet. Botanically, the awn is an apical extension of the distal end of the lemma; it is probably a modified leaf blade (Dahlgren ). Awns are also found in other grass species such as wheat, rye, oats, sorghum, and rice. The awn is considered to play various biological roles (Grundbacher, 1963). In the wild, awns with barbs aid in seed dispersal by attaching to animal fur (zoochory) (von Bothmer et al., 1995). The importance of the awn in seed burial has been well studied in wild tetraploid wheat, which has a pair of awns in each spikelet; the awn movement driven by the daily humidity cycle propels the seeds into the ground (Elbaum ). The awn also protects seeds from predation by birds and animals (Grundbacher, 1963). Under agricultural conditions, however, the long and stiff awn with barbs has persisted as a nuisance to farmers because they hinder manual harvesting. More importantly for modern agriculture, the injurious awn reduces the feed value of straw for livestock (Takahashi, 1955). Consequently, during domestication and subsequent varietal differentiation, much selection pressure has been applied to reduce awn length (Takahashi, 1955). Despite such selection pressure, most barleys have retained rather long awns. This is in contrast to other small grain crops, particularly rice (Takahashi ), where breeding has remarkably reduced awn length. This is probably because the awn in barley contributes significantly to photosynthesis (Kjack and Witters, 1974). In cross-sections, barley awns have a triangular shape with two zones of chlorenchema cells and three well-developed vascular bundles inside and stomata on the surface (Reid, 1985). In contrast, rice awns are round-shaped in cross-section with one vascular bundle in the centre (Toriba ) and probably do not contribute as much to photosynthesis. Grundbacher (1963) estimated that photosynthate derived from barley awns contributes more than 10% of the total kernel dry weight. In field tests, Qualset confirmed the significant yield advantage of barley accessions with full awns and half awns relative to those of quarter awns and awnless genotypes.

The molecular basis of awn elongation in grasses remains unclear because of a lack of cloned genes. In rice, many quantitative trait loci (QTLs) for awn length have been reported on all chromosomes except for chromosomes 1 and 5 (Cai and Morishima, 2002; Gu ). However, their causative genes have not yet been cloned. Barley provides excellent materials for such studies because many major genes subject to simple Mendelian inheritance are known to affect awn length in this species (Lundqvist and Franckowiack, 2003). In addition, genes pleiotropically affecting awn length and QTLs for awn length in barley have been reported (Sameri ; Gyenis ).

The short awn 2 (lks2, for length 2) gene on the long arm of chromosome 7H is a spontaneous recessive mutation that reduces awn length by about 50%, as well as awn thickness (Fig. 1A). This natural mutant allele shows a unique geographical distribution restricted to Eastern Asia, including India, Nepal, China, Korea, and Japan; the frequencies of short-awned varieties in Nepal, China, Korea, and Japan were reported to be 22, 3.2, 25.8 and 13.6%, respectively (Takahashi, 1987). This restricted geographical distribution raises the possibility of a selective advantage of lks2 in humid conditions. In light of the agricultural importance of lks2 in Eastern Asia, Taketa roughly mapped lks2. The present study reports the positional cloning and identification of the Lks2 gene. From lines of experimental evidence such as the gene structure, allelic mutations, expression pattern, and histology of awns, our results revealed that lks2 affects both awn elongation and pistil morphology. The origin of lks2 in barley is discussed based on natural allelic variation of the gene.

Fig. 1.

Morphology of long-awn and short-awn barley cultivars. (A) Spikes of long-awn Lks2 (Karafuto Zairai, left) and short-awn lks2 (Aizu Hadaka 3, right) cultivars used as mapping parents. Both cultivars are six-rowed. (B) Spikes of Bowman near-isogenic lines having the two-rowed genetic background. From left to right: wild-type cultivar Bowman, lks2.b-Bowman, ubs4.d-Bowman, and ari-d.15-Bowman. (C) Pistils of Bowman near-isogenic lines arranged in the same order as in (B). Bars, 1cm (A, B); 1mm (C).

Materials and methods

Plant materials and allelism tests

For mapping of lks2, 1512 F2 plants derived from a cross between Karafuto Zairai (Lks2, Okayama University accession no. OUJ301) and Aizu Hadaka 3 (lks2, OUJ323) were used (Fig. 1A). Aizu Hadaka 3 additionally carries the dense spike 1 (dsp1) gene (Takahashi et al., 1953a; Taketa ), but this gene did not affect phenotyping of lks2. For genetic analysis, each plant was scored qualitatively for awn phenotype as either long or short awn by visual inspection. F3 progeny tests were carried out for selected recombinant plants.

The following short-awn barley accessions were used for allelism tests with lks2. Short-awn mutant KM7 was induced by gamma-ray irradiation from Kanto Nijo 29 (a gift from Dr. N. Kawada, Tochigi Prefectural Agricultural Experiment Station, Japan). Bowman near-isogenic lines carrying lks2.b, unbranched style 4.d (ubs4.d), or breviaristatum-d.15 (ari-d.15) (Fig. 1B) were developed by back-crossing (a gift from Dr. H. Bockelman, National Small Grains Collection, USDA). Twenty-three lines of short-awn breviaristatum-d (ari-d) mutants and their wild types were used (a gift from Dr. F. Ottosson, NordGen, Alnarp, Sweden). The short-awn mutants used in this study are listed in Supplementary Table S1 at JXB online.

For allelism tests, short-awn mutant KM7 was crossed with Honen 6, a standard lks2 line used by Takahashi et al. (1953a). Allelism tests among the three short-awn loci of lks2.b, ubs4.d, and ari-d.15 were also conducted using Bowman near-isogenic lines; F1 hybrids and their parental lines were grown in a controlled environment room at a constant temperature of 15°C under natural light. About 3 weeks after flowering, awn phenotype was classified visually as long (wild-type Bowman), short (lks2.b type) or very short (ubs4.d or ari-d.15 type). Three plants per cross combination were tested. Quantitative data on awn length were also collected from measurements of at least five spikes per plant. Awn length was measured for samples on the second and third spikelets from the top of the spikes.

For studies of natural allelic variation, 39 lks2 accessions and eight long-awn accessions were used (Supplementary Table S2 at JXB online). Nine of the 39 short-awn accessions were confirmed to carry lks2 by crossing to Honen 6, a standard lks2 line. The other 30 lks2 carriers were selected based on phenotype. These 39 accessions were all six-rowed because spontaneous lks2 mutation is not present among two-rowed barley landrace (Takahashi, 1983). These accessions were examined for the Lks2 candidate gene by sequencing or marker genotyping of mutation points.

Genetic mapping

DNA was extracted according to methods described previously (Taketa ). PCRs were performed in 10 µl reaction mixtures containing 5 µl of Quick TaqTM HS DyeMix DNA polymerase (Toyobo Co.), 20ng genomic DNA, and 0.2 µM of each specific primer using the following conditions: initial denaturation for 2min at 94 °C, followed by 35 cycles of 0.5min at 94 °C, 0.5min at the annealing temperature of the respective primers, and 0.5min at 68 °C, with a final extension for 7min at 68 °C and a hold at 15 °C. For restriction assays, 10 µl of the PCR products was digested with 2U polymorphic enzyme and buffer in a final volume of 15 µl. These fragments were separated on 1.5–2% agarose gels or 3% Metaphor agarose gels.

PCR markers were developed as described previously (Taketa ). PCR primers were designed using Primer 3 (http://frodo.wi.mit.edu/primer3/) and synthesized commercially. Amplified PCR products from both parental accessions were sequenced according to the method of Taketa et al. (2004, 2006), and polymorphic markers were developed. For chromosome arm assignment of barley markers, wheat–barley chromosome addition lines (Islam ; Islam 1983) were used. Primer sequences are presented in Supplementary Table S3 at JXB online.

Recombination values were calculated using MAPMAKER/EXP Macintosh version 2.0 (Lander ), and a genetic linkage map was constructed based on LOD scores greater than 3.0.

Chromosome walking

A barley bacterial artificial chromosome (BAC) library of cultivar Haruna Nijo (Saisho ) was used. BAC clones near the lks2 locus were selected by PCR screening of DNA pools or by hybridizing the arrayed BAC colony filters with peroxidase-labelled probes. BAC DNA was extracted using a standard alkaline/SDS procedure. The size of the insert was analysed using pulsed-field gel electrophoresis after digestion with NotI or AscI. All BAC clones were end sequenced using M13 primers for designing PCR primers. Overlapping BACs were identified by Southern blot hybridization or by PCR marker analysis, and a BAC minimum tiling path was constructed.

BAC sequencing and annotation

Five BAC clones (HNB 389J01, 573P19, 164O16, 719A14, and 507D16) were shotgun sequenced (tenfold coverage) according to the standard dideoxy terminator chemistry protocol of the Rice Genome Research Program (International Rice Genome Sequencing Project, 2005). Annotation analyses were conducted as reported by Taketa et al. (2008). The 519kb BAC contig sequence reported in this paper has been deposited in EMBL/GenBank/DDBJ under accession number AB678347.

Expression analysis and rapid amplificationof cDNA ends (RACE)

Long-awn cultivar Haruna Nijo was grown in the greenhouse. From immature 5cm or 6cm long inflorescences, awns, pistils, anthers, glumes, lemmas, paleas, and flag leaves were collected separately and used for RNA extraction. Experiments were done with three biological replications, and each replication included one immature inflorescence. Total RNA was extracted from various tissues using Sepasol RNA I Super G (Nakalai Tesque). Real-time quantitative reverse transcription-PCR (qRT-PCR) was performed as described previously (Taketa ). Briefly, RNA samples were treated with DNase I (Promega). For qRT-PCR, first-strand cDNA was synthesized with ReverTra Ace (Toyobo Co.). Quantitative analyses was carried out on Thermal Cycler Dice TP800 (TaKaRa) using SYBR Premix DimerEraser (TaKaRa) according to the manufacture’s instructions. Normalization was carried out using the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene. For determination of 5′- and 3′-untranslated regions (UTRs), RACE was conducted as described by Taketa et al. (2008). The primer sequences are presented in Supplementary Table S3.

Histological observations

Awn samples for histological observation were collected from Bowman and an isogenic line of Bowman carrying lks2.b (lks2.b-Bowman) grown in a greenhouse. Awns were sampled from all positions of spikes at 3 weeks following flowering and were fixed with FAA solution (3.7% p-formaldehyde, 5% acetic acid) under low pressure, dehydrated, and embedded in Paraplast (McCormick Scientific). Using a microtome, 12 µm thick sections were prepared. Deparaffinization of the sections was conducted with xylene, with subsequent dehydration using a graded ethanol series. Staining was done with 1% safranin O (Waldeck GmbH and Co. KG). Awn dimensions (height: the distance from the mid-point of the longest side to the apex; and width: the length of the longest side) were measured in transverse sections prepared from the middle part of awns. In the longitudinal sections, the length of large parenchyma cells in the cell file adjacent to chlorenchyma tissues was measured. Measurements were made separately for the bottom, middle and top parts of awns. At least five awns were used for cell length measurement of each part of the awns. Cell numbers for entire length of an awn were estimated based on average lengths of awns measured at three positions within a spike: (i) awns on spikelets of the second and third spikelets from the top of spikes; (ii) those on two spikelets in the middle positions of the spikes; and (iii) those on the second and third spikelets from the bottom of spikes. This is because awns used for sectioning were the bulk of awns from all positions of the spikes. Student’s t-test was used for statistical comparisons.

Phylogenetic analysis

The deduced amino acid sequences of barley Lks2 homologues were compared with homologous proteins in Arabidopsis, wheat, and four grass species that possess known genome sequences (rice, maize, sorghum, and Brachypodium distachyon). Phylogenetic trees were generated by CLUSTALW sequence alignment of the complete proteins using the neighbour-joining method (Saitou and Nei, 1987) available in the MEGA4 software package (Tamura ). In tree construction, the gaps/missing data option was selected for the treatment of pairwise deletion.

Results

Positional cloning of Lks2

A total of 1512 F2 plants from a cross between Karafuto Zairai and Aizu Hadaka 3 were screened using flanking markers developed in a previous report (Taketa ), and lks2 was delimited to a 0.66 cM interval between expressed sequence tag markers k04151 and k06123. BLASTN analysis identified their respective rice homologues (Os06g0611200 and Os06g0710500) widely apart (5.6Mb) on rice chromosome 6, because of disrupted microsynteny in this chromosomal region (Taketa ). Because marker saturation based on barley/rice microsynteny was difficult, chromosome walking was initiated from the closest distal marker, k06123 (0.27 cM from the locus), using a barley BAC library (Saisho ). Thereby, a 613kb BAC contig was constructed (Fig. 2). Eight markers were developed on this BAC contig. Two recombination events between markers OIS56 and OIS55 correctly oriented the BAC contig relative to the barley genetic map. Database searches identified rice homologues for four other non-polymorphic markers. Six marker points on this 613kb barley BAC contig corresponded in reverse order to the syntenic rice chromosome 6 segment, identifying an inversion distal to lks2. The next rice gene (Os06g0711800) was converted into a marker (OIS62). It was mapped 0.03 cM distal to lks2. However, a marker (OIS63) from the second-next rice gene (Os06g0711900) was mapped 0.23 cM proximal of lks2. With one exception, the following four rice genes showed parallel correspondence between rice and barley. The barley homologue of the exceptional gene (Os06g0712500) was mapped on chromosome 4H. Two markers (OIS65 and OIS66) developed from rice genes Os06g0712300 and Os06g0712400 were mapped 0.2 cM proximal of lks2. Finally, the rice gene Os06g0712600 was converted into a marker (OIS71). It co-segregated perfectly with lks2 in 3024 gametes. This gene encoding a SHORT INTERNODES (SHI)-family transcription factor is considered a strong candidate for Lks2. The barley homologue of the next distal rice gene (Os06g0712700) was mapped 5.7 cM proximal of lks2, which suggests a recurrence of disrupted barley/rice microsynteny. Taken together, the observations described above indicated substantial reshuffling of the corresponding syntenic blocks between barley and rice in the Lks2 candidate region.

Fig. 2.

Positional cloning of the barley Lks2 gene. At the top, a high-resolution genetic map of barley based on 3024 gametes is shown. In the middle, a physical map of barley is presented. BAC clones representing the minimum tiling path are shown. Only the BAC contig shown in red was sequenced completely. At the bottom, a physical map of syntenic regions of rice chromosome 6 is shown. Identification numbers for each rice gene are those from the Rice Annotation Project Database, where the common prefix of Os06g has been omitted for brevity. Dotted lines connect common or homologous markers or genes. * denotes that the barley homologue was mapped to chromosome 4H; ** signifies that the barley homologue was mapped 5.7 cM proximal of lks2.

Starting from the Lks2 candidate gene, an additional BAC contig of 519kb was developed (Fig. 2). Five BAC clones representing the minimal tiling path of the contig were selected for shotgun sequencing. The SHI-family gene is the sole non-transposon-based gene on this BAC contig; approximately 74% of the contig sequence was classified as repetitive elements. Seven markers were developed on this BAC contig, but they showed no recombination with lks2. Consequently, the 519kb BAC contig was shorter than the lks2 candidate region (0.23 cM) that was delimited by markers OIS66 and OIS62 (Fig. 2). BAC contigs were also constructed starting from the flanking markers, but these BAC contigs did not overlap with others. The current physical map near the lks2 locus consists of five BAC contigs (about 1540kb in combined lengths) with four gaps.

Comparison between cDNA and genomic sequences in barley showed that the Lks2 candidate gene (SHI-family transcription factor) harboured two introns (one in the 5′-UTR) and that the open reading frame encoded a deduced protein of 344 aa with a calculated molecular mass of 34.3kDa (Fig. 3). This gene contained a zinc-binding RING-finger motif (Freemont, 1993) and an IGGH domain (Fridborg ).

Fig. 3.

Analyses of mutants allelic to lks2. Lesions detected in three natural variants and 25 mutants are summarized. The three thick downward arrows indicate amino acid substitutions in lks2.b1, lks2.b2, and lks2.b3 (natural allelic variants). Others are mutant alleles. Black arrows pointing down signify an amino acid substitution. Black arrows pointing up with a solid line signify a frame shift. The two black arrows pointing up with a dotted line represent a nucleotide substitution in the 5′-UTR region. The white arrow pointing up with a solid line represents a premature stop codon. The white arrow pointing up with a dotted line represents altered splicing as a result of a nucleotide substitution in the exon–intron boundary.

Screening and analysis of allelic mutants

No alleles at the lks2 locus were identified in previous studies of induced mutants (Kucera ; Franckowiack, 2007). In the present study, an induced short-awn mutant KM7 was crossed with Honen 6 (a standard lks2 line), and their F1 hybrids showed the short-awn phenotype, indicating that KM7 is allelic to lks2. KM7 also has reduced stigma hairs and partial female sterility. This phenotype is similar to that of ubs4 (Takahashi et al., 1953b), which was mapped only 1 cM distal to lks2 (Franckowiack, 1997). The ubs4 locus is known to be allelic to the ari-d short-awn mutant locus (Franckowiack and Lundqvist, 1997; Franckowiack, 2007). However, neither ubs4 nor ari-d has yet been cloned. To test allelism, intercrosses were made among near-isogenic lines of lks2.b, ubs4.d, and ari-d.15 that were bred in the genetic background of the cultivar Bowman (Fig. 1B, 1C). Regarding awn length (Fig. 1B), those in lks2.b, ubs4.d and ari-d.15 were 51, 28, and 29% of that in Bowman, respectively. In terms of pistil morphology (note, the term pistil will be used throughout the paper instead of gynoecium), lks2.b has slightly shorter styles, but has normal stigma hairs (Fig. 1C). However, ubs4.d and ari-d.15 have almost no stigma hairs on reduced styles (Fig. 1C). In allelism tests, the awn lengths in F1 hybrids between lks2.b and ubs4.d (62% of wild type) were almost comparable with that in the lks2.b parent (51% of wild type). The awn length in F1 hybrids between lks2.b and ari-d.15 (61% of wild type) was also comparable with that in the lks2.b parent. Therefore, it was concluded that lks2, ubs4, and ari-d occupy the same locus, and that lks2 is dominant over ubs4 and ari-d. The different locations of lks2 and ubs4.d in the genetic map by Franckowiack (1997) can be probably attributed to phenotyping errors in the segregating populations.

Sequences of the Lks2 candidate gene in 25 allelic mutants that had significantly shorter awns and reduced stigma hairs were compared with those of their respective wild type. All 25 mutants had lesions in the Lks2 candidate gene (Supplementary Table S1); 11 revealed a single amino acid substitution, four revealed truncation of the protein by a premature stop codon, four showed altered splicing as a result of a nucleotide substitution in the exon–intron boundary (altered splicing was confirmed by transcript sequencing), two were a nucleotide substitution in the 5′-non-coding region, and four were a complete deletion of the gene based on analysis of many flanking markers (Fig. 3). Most mutants have lesions in the functionally important RING motif or IGGH domain (Fig. 3). The mutant allele sequence data showed unequivocally that the SHI-family transcription factor gene comprised the lks2 locus. Therefore, the gene is referred to as Lks2 in the following text.

Natural variation of the Lks2 gene

To assess natural allelic variation (Supplementary Table S2), 39 lks2 accessions and eight long-awn (Lks2) accessions were examined. Of the eight long-awn accessions tested, six shared the same Lks2 gene sequence (the ‘standard’ haplotype). Relative to the standard haplotype, long-awn barley cultivars Bonus and Kristina had two nucleotide substitutions in the coding sequence, one of which is a non-synonymous substitution, plus three single-nucleotide polymorphisms (SNPs) and one indel in the 5′- and 3′-non-coding regions (Table 1). This type was named the ‘Bonus’ haplotype. The ‘standard’ and ‘Bonus’ haplotypes were thus regarded as wild-type Lks2 alleles (long awn). The sequences of lks2 accessions differed from the wild-type sequences. Three natural lks2 alleles were found: lks2.b1, lks2.b2, and lks2.b3 (Table 1 and Fig. 3). Allele symbols were assigned according to Franckowiack (2007). They all shared a C-to-T nucleotide substitution leading to a proline (P) to leucine (L) change at position 245 in the IGGH domain, indicating the significance of this domain in determination of awn length. Both lks2.b2 and lks2.b3 had a second amino acid substitution at different sites outside the conserved regions. The lks2.b2 allele had a C-to-T nucleotide substitution leading to an alanine (A) to valine (V) change at position 292. The lks2.b3 allele had a G-to-A nucleotide substitution leading to an alanine (A) to threonine (T) change at position 29, but this amino acid substitution was also present in the Bonus haplotype. All three natural lks2 alleles could be considered weak alleles because the plants had an awn length about half of the long-awn type and showed normal seed fertility. lks2.b1, lks2.b2, and lks2.b3, were found, respectively, at frequencies of 12.8, 56.4, and 30.8% among the 39 lks2-carrying accessions tested. lks2.b1 and lks2.b2 were distributed in eastern China, Korea and Japan, whereas lks2.b3 was localized in the Himalayas (regions including India, Nepal and Tibet) (Fig. 4 and Supplementary Table S2).

Table 1

Natural allelic variation in the Lks2 gene.

Bold letters show changes from the standard type.

Polymorphism	5’-UTR		2nd exon		3rd exon		3’-UTR
Nt position	–563	–48	85	195	829	970	1185	1245
Nt change	C→G	C→T	G→A	T→C	C→T	C→T	A→G	∆GA
Aa change	–	–	A→T	No	P→L	A→V	–	–
Aa position	–	–	29	65	245	292	–	-
Allele/haplotype	SNP-1	SNP-2	SNP-3	SNP-4	SNP-5	SNP-6	SNP-7	Indel-1
Standard type (Lks2)	C	C	G	T	C	C	A	No
lks2.b1	C	C	G	T	T	C	A	No
lks2.b2	C	C	G	T	T	T	A	No
Bonus type (Lks2)	G	T	A	C	C	C	G	∆GA
lks2.b3	G	T	A	C	T	C	G	∆GA

Fig. 4.

Geographical distribution of natural lks2 alleles in Eastern Asia. Each symbol represents one accession. Circles indicate covered (hulled) barley, while triangles represent naked (hull-less) barley.

Phylogenetic analysis of Lks2

The SHI-family transcription factor encodes proteins with a RING finger-like zinc-finger motif and an IGGH domain. Most likely, this family is plant specific (Kuusk ). The Arabidopsis SHI gene family comprises ten members: SHORT INTERNODES (SHI), STYLISH1 (STY1) and STYLISH2 (STY2), LATERAL ROOT PRIMORDIUM1 (LRP1), and SHI-RELATED SEQUENCE 3–8 (SRS3–SRS8). In the rice genome, five SHI family members are present (Kuusk ). B. distachyon, sorghum, and maize have four, five, and nine members of this family, respectively. In barley, database searching identified three additional LRP-like genes, while in wheat, a single orthologue (GenBank accession no. AK335496) of Lks2 was identified.

Phylogenetic trees of SHI-family proteins of Arabidopsis and six grass species were generated using the neighbour-joining method. The tree showed division into two major clades: one containing Arabidopsis and grass proteins, and the other containing only grass proteins (Supplementary Fig. S1 at JXB online). HvLks2 belonged to the grass-specific clade and was distinct from all ten members of the Arabidopsis SHI-family proteins. Within the grass-specific clade, HvLks2 formed a highly supported subclade with probable orthologues of other grass species. These results suggested that HvLks2 has no clear orthologue in Arabidopsis.

Expression of Lks2

Lks2 was highly expressed in pistils (including stigmas, styles, and ovaries), followed by awns; expression was weak in the anthers, glumes, lemmas, paleas, and flag leaves among the organs tested (Fig. 5). Gene expression data coincided with the organs of the types (pistils and awns) in which the phenotypes were clearly manifested in lks2.b/ubs4.d/ari-d mutants.

Fig. 5.

qRT-PCR analysis of Lks2 expression in a long-awn cultivar, Haruna Nijo. Transcript levels were normalized to the GAPDH gene as a reference. Error bars indicate standard deviation (n=3).

Histological analysis

To determine the basis of the short-awn phenotype in lks2, awn sections of Bowman and lks2.b-Bowman were compared (Fig. 6). In the transverse sections, the awns appeared triangular. Three vascular bundles (a large one in the centre and two smaller ones located in each peripheral corner) and two zones of chlorenchyma cells were observed in both genotypes, as reported for normal barley by Reid (1985). No apparent differences existed in the overall structures of the awns, but lks2.b-Bowman had significantly thinner awns both in height and width (Fig. 7C). No significant differences in longitudinal cell length existed between Bowman and lks2.b-Bowman at the three awn parts (bottom, middle, and top) measured (Fig. 7A). However, the longitudinal cell number constituting the awn in lks2.b-Bowman was reduced to 55.5% of that of Bowman (Fig. 7B). Therefore, it was concluded that the short-awn phenotype in lks2 resulted from a reduced cell number.

Fig. 6.

Histological observation of sections of long-awn (Lks2) Bowman and short-awn lks2.b-Bowman. (A, B) Transverse sections of awn in the middle part of Bowman (A) and lks2.b-Bowman (B) with the abaxial sides up. (C, D) Longitudinal sections of awn in the bottom part of Bowman (C) and lks2.b-Bowman (D). p, Parenchyma cells; vb, vascular bundle; ch, chlorenchyma tissue; W, awn width; H, awn height. Arrows indicate the cell files used for parenchyma cell length measurements in longitudinal sections. Bars, 200 µm.

Fig. 7.

Measurement of awn sections from Bowman and lks2.b-Bowman. (A) Parenchyma cell length in the cell file adjacent to chlorenchyma tissues measured in longitudinal awn sections. (B) Longitudinal parenchyma cell number for entire lengths of awns. (C) Heights and widths of awns measured in transverse sections from the middle part of awns. Bars represent standard deviation. ns, Not significant; **, significantly different at the 1% level.

Discussion

Positional cloning

Because the exploitation of barley/rice microsynteny in the search for an Lks2 candidate gene was hampered by highly disrupted collinearity, chromosome walking was initiated from the closest marker (k06123), 0.27 cM distal to lks2. A complete BAC contig spanning the lks2 locus could not be constructed, mainly because of low gene density, a high proportion of repetitive DNA sequences, and suppressed recombination near the target locus. However, these attempts identified local inversions of the corresponding syntenic blocks between barley and rice in the Lks2 candidate region. An SHI-family transcription factor gene was identified as a strong candidate for Lks2 because of perfect co-segregation of its marker with the phenotype in 3024 gametes (Fig. 2). Gene isolation was validated by: (i) genetic lesions found in all 25 mutant alleles and in three natural allelic variants at the lks2 locus; and (ii) strong gene expression in awns and pistils, the two tissues where strong visible phenotypes are observed. As pointed out previously by Pourkheirandish , the present study verifies the importance of implementing physical mapping strategies in the target species (barley) because of the possibility of a breakdown of microsynteny between barley and other related species.

Function of the barley Lks2 gene

The present study revealed that barley Lks2 is a SHI-family transcription factor that regulates awn elongation and pistil morphology. Four SHI-family genes have been found in barley, and mutation of Lks2 showed prominent phenotypes in the awn and pistil. The SHI-family transcription factors are already well studied in Arabidopsis. The ten members of the Arabidopsis SHI-family are highly divergent in sequence except for the two conserved regions (the RING motif and the IGGH domain), but show partial redundancy in function (Fridborg et al., 1999, 2001; Kuusk ). A loss-of-function sty1-1 mutant shows subtle morphological defects in the pistil, while mutations in seven other family members tested had no apparent effects on pistil development (Kuusk ). Therefore, among the Arabidopsis SHI-family, STY1 is most similar to barley Lks2 in terms of mutant phenotype affecting the pistil. Double or higher-order multiple mutants of sty1-1 with other members showed an enhanced mutant phenotype in pistil, stamen, and leaf development. STY1 was expressed not only in the apical parts of the developing pistil but also in many other organs such as hypocotyls, cotyledons, leaf and root primordia, and leaf margins (Kuusk ).

The present histological observations revealed that the short-awn phenotype in lks2.b resulted from a reduced longitudinal cell number, which was likely to be due to reduced cell divisions (Figs 6 and 7). Similarly, reduced stigma hairs and style length could be caused by reduced cell divisions during pistil development. It has been reported that Arabidopsis homologue STY1 protein regulates auxin homeostasis (Sohlberg ) by transcriptional activation (Eklund et al., 2010), and, by analogy, barley lks2 mutation may reduce cell divisions by affecting auxin levels in developing awns and pistils. This point needs to be examined in future studies.

The phylogenetic analyses of SHI-family proteins in Arabidopsis and six grass species showed that barley Lks2 belonged to the grass-specific clade (Supplementary Fig. S1). This may coincide with the fact that barley Lks2 is controlling elongation of the awn, which is an organ unique to grasses. Thus, Lks2 and its grass orthologues may have diversified functionally in the SHI family. However, barley Lks2 and the Arabidopsis SHI-family genes (especially STY1) share a common function in regulation of pistil morphology. These findings suggest the hypothesis that, to some extent, the highly conserved RING motif and IGGH domain may play common roles among different plant species in the regulation of plant development including pistil morphogenesis. In contrast, other parts of the SHI-family proteins with relatively low homology among different plant species could be contributing to diversity in functions of the SHI-family proteins and their regulatory mechanisms.

Although Lks2 orthologues were found in grass species whose genomes have been completely sequenced (Supplementary Fig S1), there is no report about their mutants and/or associated phenotypic changes. The barley Lks2 appears to be orthologous to rice gene Os06g0712600, but no QTL for awn elongation has been reported in the syntenic region of rice chromosome 6 (Cai and Morishima, 2002; Gu ). According to the rice gene expression database RiceXPro, Os06g0712600 is highly expressed in pistils, inflorescences, and embryos of the awnless cultivar Nipponbare (Sato ). Whether this gene is expressed in rice awns needs to be tested using long-awn varieties. Because barley and rice largely differ in their awn morphology (Reid, 1985; Toriba ), different genes may control awn elongation in these species. Similarly, a database search identified an orthologue in wheat (GenBank accession no. AK335496), but no awn gene is documented for wheat homologous group 7 chromosomes (McIntosh ). Thus, a grass-wide survey is required to test whether control of awn elongation is specific to barley Lks2. It also remains to be tested whether Lks2 orthologues in other grasses control pistil morphology.

In barley, the dominant Hooded mutation (Kap) transforms the awn into an extra flower of inverse polarity on the lemma; this homeotic Hooded mutation is known to be caused by a 305bp duplication in a homeobox gene intron (Müller ). Interestingly, recessive lks2 is epistatic over Kap and suppresses the Hooded phenotype (Takahashi et al., 1953a). This epistatic inhibition of Kap by lks2.b may be mediated by suppression of cell divisions in the awn, as was revealed by the present histological investigation (Figs 6 and 7). Suppressor mutant genes of Kap (suK), which show a short-awn phenotype, were induced, but they were non-allelic to lks2 (Roig ). A QTL for awn length was also detected near the row-type locus (vrs1) (Sameri ), suggesting its effects on awn length. The molecular bases of the interaction of lks2 with Kap, suK, and vrs1 need to be clarified by future investigations.

Origin of natural lks2 alleles

A total of 25 allelic mutants had severe reductive phenotypes in awns and stigmas accompanied by seed sterility. In contrast, barley accessions carrying one of the three natural lks2 variants, lks2.b1, lks2.b2, and lks2.b3, have awns about half the length of the long-awn type, normal stigma hairs, and normal seed fertility. These natural variants share a P-to-L change at position 245 in the IGGH domain, which is regarded as key to producing mild phenotypes. This critical mutation occurred within a short acidic cluster present in the IGGH domain (Fridborg ), where conservation is rather low among the SHI-family members of different species (Supplementary Fig. S2). Similar nucleotide sequences and geographical distribution suggest that the lks2.b1 and lks2.b2 alleles are closely related (Table 1, Fig. 4, and Supplementary Table S2). It is likely that lks2.b1 is the ancestral allele and that its descendant, lks2.b2, became predominant in the east of China. The allele lks2.b3, which is localized in the Himalayas, differs from lks2.b1 by having five SNPs and one indel in regions flanking the critical mutation. Creation of the lks2.b3 allele by intragenic recombination after hybridization between the Bonus haplotype and lks2.b1 is unlikely because it requires double crossovers within a small region. Gene conversion could be an alternative mechanism. However, no evidence of exchange of barley materials exists between the Himalayas and eastern China (Konishi, 1995; Taketa ). Therefore, it is proposed that critical mutations resulting in a P-to-L change at position 245 might have independent origins in the Himalayas and the east of China, although deeper sequence sampling is required in future to confirm this.

The short-awn trait conferred by the lks2.b allelic series is considered to be an example of varietal differences (Doebley ) that have occurred after domestication of the barley crop. This is because these mutant alleles are present among limited barley accessions from Eastern Asia (Takahashi, 1987). The frequency of varieties with lks2.b allele in Eastern Asia regions varies from 3.2 to 25.8% (Takahashi, 1987). The reasons why lks2.b established an ecological niche in Eastern Asia are not clear. It may be related to the Eastern Asian climate, which has high precipitation during the growing season of barley. Under such climatic conditions, the short-awn lks2.b allele may be advantageous in alleviating lodging, because short-awn spikes would hold less water during rain and would dry faster, as suggested by Grundbacher (1963). The effects of lks2.b on the degree of lodging need to be tested in future studies using isogenic lines.

This report is the first to describe the molecular isolation of a gene controlling awn elongation in grasses. The molecular cloning of Lks2 in barley revealed that an SHI transcription factor in barley plays important dual roles in controlling awn length and pistil morphology. The isolation of Lks2 and future analysis will give many insights into the control of floral morphology of both awns and pistils in grasses. In barley, smooth-awned mutant genes (raw1 and raw2) also reduce the number of stigma hairs (Lundqvist and Franckowiack, 2003), indicating that a morphological link appears to exist between the awn and stigma hair. Further work to isolate other awn-related genes of barley is underway for a better understanding of the genetic network controlling awn elongation. As awns appear to have an effect on grain yield, such studies may enable the design of barley awns for improving productivity in response to various climatic conditions.