Introduction

Water deficit is one of the prevalent causes of crop yield loss. Increased plant tolerance to water deficit is considered among the most important abiotic parameters that can contribute to increased grain production, in particular in Mediterranean regions (Araus et al. 2003). Plant genetic adaptation to environmental stress is displayed in physiological and biochemical responses, controlled by changes in gene expression. Several major classes of genes have been noted that are altered in response to water-deficit stress; genes involved in signaling and gene regulation and gene products that are proposed to support cellular adaptation to water-deficit stress are among the most frequently altered in gene expression (Hazen et al. 2003; Hazen et al. 2005; Ito et al. 2006; Yamaguchi-Shinozaki and Shinozaki 2005). Although a large number of genes and proteins responding to stresses have been studied, most molecular components of the signaling transduction pathway involved in gene regulation under stress are still unidentified, and their precise functions in either tolerance or sensitivity remain unclear. Comprehensive profiling of stress-associated metabolites, combined with stress metabolomics of major crop plants may be a key factor in molecular breeding for tolerance (Vinocur and Altman 2006).

Numerous studies have shown that wild progenitors of cultivars comprise one of the major genetic recourses of plant tolerance to stressful environments (Ellis et al. 2000; Nevo et al. 2002). Habitats of wild barley (Hordeum spontaneum), the progenitor of cultivated barley, differ in water availability, temperature, soil type, altitude, and vegetation, generating thereby a high potential of adaptive diversity in wild barley to abiotic stresses. Adaptive genetic diversity in natural populations was revealed by protein and DNA markers (Nevo et al. 1979; Owuor et al. 2003; Turpeinen et al. 2003; Suprunova et al. 2004; Ozkan et al. 2005), indicate the potential of wild barley as a source for drought-resistance alleles for breeding purposes.

Functional genomic tools can be applied for identifying and isolating the genes involved in plant abiotic stress tolerance (Langridge et al. 2006). These tools include variety of molecular techniques available to identify and clone differentially expressed genes. Among the techniques based on assaying nucleic acid hybridization, microarray technology plays an ever-increasing role in unraveling the molecular genetic basis of plant reaction to stress (Chao et al. 2005; Gulick et al. 2005; Kim and von Arnim 2006). However, microarray technology has relatively high start-up costs and its utility is limited by low sensitivity for detection of rarely expressed transcripts and difficulties in distinguishing transcripts from homologous genes. Among the genome-wide expression analysis techniques based on PCR and gel separation and visualization procedures, cDNA amplified fragment length polymorphism (cDNA-AFLP) (Bachem et al. 1996, 1998) has proved the most popular technique. cDNA-AFLP overcomes some of the limitations in hybridization-based techniques and is considered a valid alternative/complementation to microarrays (Volkmuth et al. 2003). This technique does not require prior sequence information and has good reproducibility and sensitivity compared to microarray technologies (Reijans et al. 2003). cDNA-AFLP results correlate well with Northern, quantitative expression analysis by real-time PCR (Q-PCR), and microarray analysis (Donson et al. 2002; Avrova et al. 2003; Breyne et al. 2003).

In the present study, we have used the cDNA-AFLP technique to screen for candidate transcripts, which are differentially expressed between sensitive and tolerant wild barley genotypes under dehydration stress. Several transcript-derived fragments (TDFs) that were differentially expressed in the tolerant genotype, as compared to the sensitive genotype, were isolated. TDF4 was analyzed as a novel candidate gene for drought tolerance in wild barley. The novel water stress inducible gene; designated Hsdr4 (Hordeum spontaneum dehydration-responsive 4) was sequenced and analyzed and its involvement in drought stress tolerance in barley is discussed.

Materials and methods

Plant material

Two water stress sensitive genotypes (JS1 and JS2) and two water stress tolerant genotypes (JR1 and JR2) of wild barley (Hordeum spontaneum) were used in this study. These genotypes were selected from a collection of 400 genotypes originating from diverse eco-geographic regions in Israel and Jordan and surroundings. Selection was done based on measurements of water loss rate (WLR) and relative water content (RWC) after severe dehydration stress, and displayed different patterns in the dynamics of drought-induced expression of dehydrin genes in response to dehydration stress (described in details by Suprunova et al. 2004).

Stress treatments

Seedlings were grown in a greenhouse at 22°C, with a photoperiod of 12 h light/12 h dark, in Murashige and Skoog basal salt mixture (MS) (Sigma) solution, circulated by air pumps. Ten-day-old seedlings were subjected to water stress by complete draining of the MS solution from the container. Leaf tissues were harvested from control plants (time 0), and after 3 h and 12 h of stress. Leaves were frozen in liquid nitrogen and stored at −80°C for RNA extraction. Two to three seedlings from each genotype were tested independently in each step of the gene expression analysis.

cDNA-AFLP analysis

Total RNA was extracted from barley leaf tissue using EZ-RNA Total RNA Isolation Kit (Biological Industries, Beit Haemek LTD, Israel). Poly(A)+ RNA was prepared from 30 μg of total RNA using oligo (dT) coupled to paramagnetic beads (PolyA Tract mRNA Isolation System IV, Promega). Double-stranded cDNA synthesis was carried out with the Universal RiboClone cDNA Synthesis System (Promega) according to the manufacturer’s instruction.

cDNA-AFLP analysis was conducted according to Bachem et al. (1998) with a slight modification. Double-stranded cDNA (100 ng) was digested by MseI/EcoRI enzyme combination and than ligated to appropriated adaptors. For pre-amplification reactions non-selective MseI + 0 and EcoRI + 0 primers were used. Following the pre-amplification step, the products were diluted (10×) with TE buffer and 5 μl were used for final selective amplifications using the following primer combinations: EcoRI-C/MseI-TC; EcoRI-C/MseI-CG; EcoRI-AC/MseI-GG; EcoRI-AC/MseI-CC; EcoRI-CA/MseI-TC; EcoRI-CA/MseI-CC; EcoRI-CA/MseI-CG; EcoRI-CA/MseI-CCA; EcoRI-CA/MseI-CCC; EcoRI-CA/MseI-CCG; EcoRI-CA/MseI-CGA; EcoRI-CA/MseI-CGC; EcoRI-CA/MseI-CGG; EcoRI-CC/MseI-GG; EcoRI-CC/MseI-TC. The EcoRI primers were radioactively labeled by [γ-33P]ATP. Selective amplification products were separated on 5% polyacrylamide sequencing gels using Sequi-Gen GT System (Bio-Rad). The polyacrylamide gels were dried onto 3 MM Whatmann paper and positionally marked before being exposed to Kodak Biomax MR Films for 3 days.

Isolation and sequencing of cDNA-AFLP fragments

X-ray films and gels were aligned according to the markers on the gels, and transcript-derived fragments (TDFs) that showed clear differential amplification in sensitive vs. tolerant barley genotypes under dehydration were excised from the gels and incubated in 50 μl H2O overnight at room temperature. The eluted TDFs were re-amplified using the method developed by Brugmans et al. (2003) for converting AFLP markers into single-locus markers. The method includes three steps: (1) Each of the TDFs was excised from the cDNA-AFLP gel and re-amplified with its corresponding selective cDNA-AFLP primers under the same PCR conditions that were used for the active PCR step. Recovered TDFs were separated on 2% agarose gel and purified (QIAquick Gel Extraction Kit, Qiagen); (2) The re-amplified fragments were used as templates for second PCR using a set of 16 degenerated MseI primers and 16 degenerated EcoRI primers. The amplification products were separated on 2% agarose gel. Based on the results of the quality and quantity of the different amplification products we were able to determine the extra selective nucleotides adjacent to the corresponding ends of MseI and EcoRI selective primers; (3) Based on this analysis, new specific primers were designed for each TDF, that were used for final PCR amplification using the first re-amplified TDF as a template. The final PCR products were separated on 2% agarose gel, purified, and directly sequenced. Each TDF was sequenced using the corresponding new selective primers as sequencing primers.

The homologues for all TDF sequences were determined using BLAST algorithm (Altshul et al. 1997) by comparison with database at National Center for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/BLAST/) and the Triticeae EST database (“HarvEST”) (http://harvest.ucr.edu/).

Reverse Transcription (RT-PCR) analysis

Equal amounts of total RNA from leaves were treated with RNase-free DNase I (DNA Free kit, Ambion, USA), following the manufacturer’s instruction. First-strand cDNA was synthesized from 3 μg of total RNA, using universal oligo(dT)15 primer and 200 units of SuperScript II reverse transcriptase (GIBCO-BRL), at 42°C for 1 h in a 20 μl reaction volume. The resulting single-strand cDNA was amplified using specific primers for each TDF that were designed, based on the obtained TDFs sequences. To preserve the highest RT-PCR stringency, only primers with annealing temperatures higher than 55°C were designed. PCR products were sampled after 20, 25, 30, and 35 cycles to determine the linearity of the PCRs. For each TDF specific PCR conditions were determined, including annealing temperature and cycle’s number that provided the sensitivity needed to detect the differences in expression (Table 1). Barley α-tubulin gene amplified with the specific primers 5′-AGTGTCCTGTCCACCCACTC-3′ and 5′-CCAAGGATCCACTTGATGCT-3′ (acc. no. U40042) was used as a constitutive control in all experiments. The RT-PCR products were visualized by electrophoresis on 2% agarose gel.

Table 1 Primer sequences and PCR conditions used in RT-PCR analysis of seven TDFs
Full size table

Generation of the 3′- and 5′-ends of cDNA

The FirstChoice RLM-RACE kit (Ambion, USA) was used for rapid amplification of cDNA ends according to the manufacturer’s instruction. Two gene-specific primers (5′-TCGGCTGGAATATCTATGGG-3′ and 5′- ACCATTTGTCTTCATACCTACC-3′) and two gene-specific inner primers (5′-TGAGATTCCAGATTTTGTTACT-3′ and 5′-TGGTCGGTGCTCAAGTATCT-3′) were designed based on the 223-bp TDF-4 sequence. Total RNA of wild barley genotype (JR1) dehydrated for 12 h was used as a template for the RACE procedure. RACE-PCR products were cloned in a pGEM-Teasy vector (Promega, Madison, USA), and six independent clones were sequenced. Based on the sequences of the 5′- and 3′-untranslated regions the specific primers (5′-CAGCAGGAGGGGCGGCCGGC-3′ and 5′-GCATCGCCTTTTGCTATGACAT-3′) were designed and used for PCR amplification of the whole Hsdr4 gene from genomic DNA of wild barley.

BAC library screening

The Bacterial Artificial Chromosome (BAC) library of H. vulgare L. cv. Morex (Yu et al. 2000) was used for isolation of the Hsdr4 promoter region. A set of 17 high-density filters printed with the BAC library clones in a 4 × 4 double-spotted array was screened with a 143 bp fragment from Hsdr4 cDNA as a 32P-labeled probe. Hybridizations of the BAC library high-density filters and genomic Southern blots were performed as described by Dubcovsky and Dvorak (1994). Five positive BAC clones (61L6, 76G22, 76G24, 275F6, and 422I22) carried the Hsdr4 gene were confirmed by PCR using Hsdr4 specific cDNA primers. The DNA from these BAC clones was isolated using PCIψClone BACDNA Kit (Princeton Separation, USA). HindIII digested fragments of BAC clone’s DNA were separated on 1% agarose gel using pulse-field electrophoresis system (Bio-Rad), blotted onto a Hybond N+ membrane and hybridized with [32P]-labeled 143 bp fragment from cDNA of Hsdr4 according to standard procedures (Sambrook et al. 1989). BAC clone 76G24 was chosen for isolation of the Hsdr4 promoter region.

Quantitative expression analysis of Hsdr4 by real-time PCR (Q-PCR)

Gene quantification of Hsdr4 gene was performed using ABI PRISM 7000 Sequence Detection System and SYBR Green PCR Master Mix (Applied Biosystems, USA). Total RNA was extracted from seedling shoots of two sensitive (JS1 and JS2) and two tolerant (JR1 and JR2) genotypes under control and after 3 h and 12 h of dehydration. A specific primer pair was designed, based on sequence Hsdr4 gene considering the exon–intron structure (forward 5′-CCGGGCTTTATTCCTGGCT-3′ and reverse 5′-TTTCCAGTACAACCCTCCGCT-3′). The standard curve was generated for the Hsdr4 gene using serial dilutions of an experimental cDNA sample that showed the maximal amount of target gene in preliminary RT-PCR analysis (cDNA sample of JR1 genotype after 12 h dehydration). In order to account for differences in target RNA presented in each sample, Hsdr4 gene quantities were normalized to the barley α-tubulin as an internal housekeeping gene, which was amplified with forward (5′-TCCATGATGGCCAAGTGTGA-3′) and reverse (5′-CTCATGTACCGTGGGGATGTC-3′) primers (acc. no. U40042). Five independent plant samples for each genotype were examined in triplicates. Student’s t-test and StatSoft package (Version 6.0) were used to evaluate the expression data.

Genetic mapping of the Hsdr4 gene

A wild barley mapping population derived from a cross of H. spontaneum accessions MA10-30 × WQ23-38 (Chen 2005) was used to determine the chromosomal location of the Hsdr4 gene. Using pair of primers (5′-CGAGGAGTGGCACGACTGCGT-3′ and 5′-CCTCTGGCAACTCGGTGCGGAG-3′), one of the introns of Hsdr4 gene was amplified and sequenced for both parental lines. The genomic DNA of 135 plants was used for PCR amplification followed by Taq1 restriction of PCR products. Using the presence/absence restriction site polymorphism of F2 plants and aligning our data with existing mapping data for the MA10-30 × WQ23-38 mapping population, the map location of the barley Hsdr4 gene was determined. The MultiPoint program for efficient multipoint mapping (http://www.multiqtl.com), based on Evolutionary Strategy algorithm (Mester et al. 2004), was used for mapping.

Analysis of the Hsdr4 promoter and gene sequence

The obtained nucleotide and deduced amino acid sequences of Hsdr4 were analyzed with DS Gene program (Accelrys, Software Inc., England). Analysis of the promoter sequence was performed using the Eukaryotic Promoter Database, EPD, (http://www.epd.isb-sib.ch). To search for known promoter motifs and transcription factor binding sites, the software Signal Scan of the Plant Cis-acting Regulatory DNA elements database was used (Higo et al. 1999), (http://www.dna.affrc.go.jp/htdocs/PLACE/signalscan.html).

Results

Differential expression of wild barley under water stress by cDNA-AFLP analysis

cDNA-AFLP analysis was performed using seedlings of two contrasting genotypes (JS1 and JR1) subjected to dehydration stress in order to identify genes that are involved in water-stress tolerance of wild barley. Differentially expressed fragments were detected by selective amplifications using fifteen primer combinations (PCs). An example of a typical cDNA-AFLP banding pattern with three primer combinations is presented in Fig. 1. Number and length of observed transcript-derived fragments (TDFs), varying from 50 bp to 500 bp, were dependent on PCs. In total, approximately 1100 TDFs were detected in the two barley genotypes under control and stress conditions. Of these, 70 fragments (6.3%) were differentially expressed in the control plants as compared to water stressed plants (3 h or 12 h dehydration). We selected 11 TDFs that displayed clear differences between sensitive and tolerant genotypes under 3 h or 12 h dehydration stress. Differential amplification was confirmed by repeating the cDNA-AFLP PCR reaction with primer combinations that produced bands of interest.

Fig. 1
figure 1

cDNA-AFLP fingerprints of a sensitive (JS1) and tolerant (JR1) wild barley genotypes under control and dehydration stress. Lanes: 1. control plants; 2. 3 h dehydration; 3. 12 h dehydration. The primer combinations used here: I—EcoRI-C/MseI-CG; II—EcoRI-CA/MseI-CCC; III—EcoRI-CA/MseI-CCG. Arrows indicate several differentially expressed bands, which were isolated and sequenced

Full size image

cDNA-AFLP employs a highly stringent PCR regime. However, there remains the problem of overlapping and co-migration of bands in the gels, since each AFLP band may be composed of a number of fragments with identical size that differ in sequence (Meksem et al. 2001). To overcome this problem, we used a few steps of the procedure developed by Brugmans et al. (2003) for converting AFLP markers into single-locus markers. By using a generalized set of 16 degenerated MseI primers and 16 degenerated EcoRI primers, we could determine the extra selective nucleotides adjacent to the MseI and EcoRI primers for the 11 TDFs. For example, TDF-11 that was originally amplified with MseI-CG/EcoRI-C primers was used as template for PCR amplification with MseI-CG and a set of 16 degenerated EcoRI primers: (1) EcoRI primer + N + A or C or G or T; (2) EcoRI primer + NN + A or C or G or T primers; (3) EcoRI primer + NNN + A or C or G or T; (4) EcoRI primer + NNNN + A or C or G or T. The results of PCR with first four primers (+N) allowed the determination of the second selective nucleotide since the quantity of only one of the amplification products was high. The same happened with the amplification products of the second set of four primers (+NN) that allowed the determination of the third selective nucleotide. The third set of four primers (+NNN) allowed the determination of the fourth selective nucleotide, while the amplification with the fourth sets of primers (+NNNN) allowed the determination of the fifth selective nucleotide. According to these PCR results, we could determine four extra selective nucleotides adjacent to the EcoRI-C primers for the specific TDF. The same was done with the set of MseI degenerated primers. Based on this analysis, new specific primers were designed for final PCR amplification of each TDF. The final PCR products were used for direct sequencing with no need in sub-cloning and sequencing of a few bands from each TDF. The TDFs gave good-quality sequences and the presence of the selective primers with additional extra four selective nucleotides in the 5′ or 3′ ends of all eleven TDFs confirmed the validity of our technique that was developed to overcome the problem of overlapping and co-migration of multiple bands in the gels.

The eleven TDFs were used to search the GenBank database and the Triticeae EST database (“HarvEST”). Hits were considered significant if the expected value (E-value) was less than 1.0E−5. The results of sequence comparisons are summarized in Table 2. From the eleven TDFs, six fragments showed significant homology with known-function genes. Five TDFs were homologous to EST sequences from H. vulgare, among them three corresponded to clones of EST libraries from disease and drought-stressed plants. TDF-2 and TDF-10 displayed very weak similarities to EST sequences. Of the six known-function genes, four are possibly involved in general cellular metabolism and organization, including pectin glucuronyltransferase (TDF-5), cystathionine β-lyase (TDF-11), 3-β-glucuronosyltransferase (TDF-13), and cullin3 (TDF-14). Another two drought responsive genes encoding for S-adenosylmethionine decarboxylase (TDF-12) and Rho-GTPase-activating protein (TDF-4), belong to the defense/stress and signal transduction category.

Table 2 Homologies of TDF sequences isolated by cDNA-AFLP analysis
Full size table

Expression analysis of selected TDFs by RT-PCR

To verify the cDNA-AFLP results, an independent expression study was performed for seven TDFs by RT-PCR. New cDNA samples were prepared from a new set of JS1 and JR1 barley plants grown under the same conditions as used for the initial cDNA-AFLP analysis. Internal specific primer pairs were designed for fragments TDF-2, TDF-3, TDF-4, TDF-5, TDF-11, TDF-13, and TDF-14 based on their sequences and used for RT-PCR. The suitable primer pairs were designed for seven (relatively long) out of the sequenced 11 TDFs. Individual PCR conditions and primer pair sequences for each TDF are shown in Table 1.

RT-PCR analysis proved that the seven isolated cDNA-AFLP fragments are differentially expressed in sensitive and tolerant wild barley genotypes under water-stress condition (Fig. 2). All of the TDFs were up regulated in the tolerant genotype as compared to the sensitive genotype after 12 h of dehydration, and very good correspondence was found between cDNA-AFLP and RT-PCR expression patterns in JS1 and JR1 genotypes under control and dehydration stress (Fig. 2).

Fig. 2
figure 2

Comparison of TDF expression patterns obtained by (A) RT-PCR and (B) cDNA-AFLP techniques. cDNA was obtained from seedlings of drought-sensitive (JS1) and drought tolerant (JR1) genotypes after dehydration treatments of 3 h and 12 h and control conditions (time 0). The expression of TDF-4 was compared by using a second pair of tolerant and sensitive genotypes JS2 and JR2. Barley α-tubulin gene was used as control for relative amount of RNA

Full size image

TDF-4 was selected as a promising candidate gene for drought tolerance due to high correspondence of its cDNA-AFLP and RT-PCR expression patterns and high homology to the gene encoding for GTPases known to be involved in many signal transduction pathways. Furthermore, RT-PCR using primers of TDF-4 confirmed the differential expression in another pair of sensitive (JS2) and tolerant (JR2) barley genotypes that was included in this expression assay.

Expression analysis of the Hsdr4 gene tested by quantitative RT-PCR (Q-PCR)

To quantitatively determine the expression pattern of the Hsdr4 gene in tolerant and sensitive wild barley genotypes under dehydration stress, Q-PCR (quantitative expression analysis by real-time PCR) was used. Expression analysis of the Hsdr4 gene under dehydration stress tested by Q-PCR analysis revealed no significant difference between sensitive and tolerant genotypes under control and after 3 h drought stress (Fig. 3). However, after 12 h dehydration, the tolerant genotypes (JR1 and JR2) showed 61% (P = 0.011) higher expression of the Hsdr4 gene as compared to the sensitive ones (JS1 and JS2). These Q-PCR analysis of the Hsdr4 gene validated the results obtained by both cDNA-AFLP and RT-PCR analysis and revealed higher expression level of the Hsdr4 gene after 12 h dehydration stress in both tolerant genotypes (JR1 and JR2) as compared to the sensitive ones (JS1 and JS2).

Fig. 3
figure 3

Hsdr4 gene expression upon dehydration stress detected by quantitative real time PCR (Q-PCR). Q-PCR was carried out with cDNA obtained from two sensitive (JS1 and JS2) and two tolerant (JR1 and JR2) wild barley genotypes under control (c) and after 3 h and 12 h of dehydration. Quantification is based on Ct values that were normalized using the Ct value corresponding to a barley α-tubulin gene. Each value is the mean ± SE (n = 5)

Full size image

Cloning and structure analysis of the Hsdr4 gene

To obtain the complete nucleotide sequence of the gene corresponding to TDF-4, both 5′ - and 3′ - amplification of cDNA ends (RACE) was performed using TDF-4-specific primers and cDNA synthesized from dehydrated (12 h) wild barley genotype (JR1). The cloned 980 bp full-length cDNA, designated Hsdr4 (Hordeum spontaneum drought-responsive), contains one open reading frame of 753 bp, encoding a protein of 250 amino acid residues. The 5′-UTR and 3′-UTR comprised of 34 bp and 193 bp, respectively. The database BLASTP search showed that the protein encoded by Hsdr4 cDNA is 82% identical (E-value 3E−122, score 440 bits, 91% positivities) to the Rho-GTPase-activating protein-like of Oryza sativa (acc. no. BAD87212) and is 61% identical (E-value 1E−81, score 305 bits, 77% positivities) to an unknown protein of Arabidopsis thaliana (acc. no. NP_566369). Figure 4 illustrates the alignment of the deduced amino acid sequence of barley Hsdr4 with the two most similar deduced proteins. Alignment revealed the SEC14 domain (Sec14p-like lipid-binding domain) (E-value 2E−11, score 63.5 bits, 100% aligned), that was first described in Saccharomyces cerevisiae phosphatidylinositol transfer protein (Sec14p), which plays an essential role in the formation of secretory vesicles and in protein transport from the Golgi complex (Bankaitis et al. 1990).

Fig. 4
figure 4

Alignment of the full-length amino acid sequence of the Hsdr4 with the two most similar proteins found in the databases, the rice Rho-GTPase-activating protein-like (acc. no. BAD87212) and unknown protein of Arabidopsis (acc. no. NP_566369). Enclosed boxes indicate identical amino acids. The SEC14 domain is indicated as a solid line. The alignment was generated by ClustalW algorithm using the MegAlign tool of the DS Gene program

Full size image

The whole Hsdr4 gene was amplified from genomic DNA of wild barley using specific primers for 5′- and 3′-UTR. Comparison of the genomic DNA sequence with Hsdr4 cDNA revealed that the gene consists of two introns and three exons. The splice junctions of both introns follow the GT/AG rule. The genomic sequence of the Hsdr4 gene is available in the NCBI databases (acc. no. DQ464370). Southern blot of wild barley genomic DNA revealed that there is only one copy of this gene present in the wild barley genome (data not shown).

Isolation of Hsdr4 promoter

The promoter region of the Hsdr4 gene was isolated by screening the BAC library of cultivated barley (H. vulgare L. cv. Morex) (Yu et al. 2000). A positive BAC clone, 76G24, was used to subclone a 3.1-kb HindIII fragment carrying the Hsdr4 gene into pBlueScript KS+. This fragment was sequenced using Hsdr4 and vector specific primers. Then, a new forward primer specific for 5′-end of Hsdr4 of cultivated barley and reverse wild-specific primer of Hsdr4 were used to PCR amplify the promoter region of the Hsdr4 gene from wild barley, yielding a 1.7 kb fragment upstream of the translation start codon. Sequence analysis of the 5′-region of Hsdr4 gene by comparison with the Eukaryotic Promoter Database (EPD) from NCBI showed no overall sequence homology to any other promoter. The putative transcription initiation site (TIS) was determined experimentally with FirstChoice RLM-RACE Kit (Ambion, USA) at 34 bp upstream of the first in-frame ATG codon. Using SignalScan software of PLACE database (Higo et al. 1999), the putative CCAAT motif was recognized at 52 bp upstream of the TIS. However, a putative TATA box was not revealed, suggesting that Hsdr4 promoter belongs to the TATA-less class of promoters (Fig. 5).

Fig. 5
figure 5

Nucleotide sequence of the promoter region of the Hsdr4 gene sequenced from tolerant barley genotype. Sequence of the 5′-flanking region and partial amino acid sequence of the first exon of the Hsdr4 gene are shown together. The numbering of nucleotides relative to the translation start site (ATG) is shown on the right. The transcription initiation site is labeled with an arrow, and the putative CCAAT box is in bold letters. Putative promoter cis-elements are labeled with boxes (boxes A–O). The stress-related cis-elements are shown in shaded boxes. The miniature inverted-repeat transposable element (MITE), which potentially can form the hairpin structure, is underlined. The sequence differences in Hsdr4 promoter of sensitive genotypes and cultivated barley are represented in bold and italic bold letters, respectively, under corresponding nucleotides in the Hsdr4 promoter of tolerant genotypes

Full size image

Analysis of Hsdr4 promoter

Putative cis-elements

For the identification of putative cis-acting elements in Hsdr4 promoter the plant-oriented collection of transcription regulatory elements PLACE database was used. Fifteen putative regulatory cis-elements were found in the Hsdr4 promoter (boxes A–N, Fig. 5). Their functions, sequences, and positions in Hsdr4 promoter are summarized in Table 3. These elements from Hsdr4 promoter could be separated into two groups according to their known functions. More than half of the identified putative signals belong to the group of typical stress-related cis-acting elements (boxes A–H, Fig. 5 and Table 3). Another group of putative binding sites for transcription factors, found in Hsdr4 promoter, is composed of elements involved in hormone responses and developmental processes (boxes I-O, Fig. 5 and Table 3).

Table 3 Homology between Hsdr4 promoter and several transcription factors (TF) binding sites
Full size table

Putative MITE structure

Structural analysis of 5′ region of Hsdr4 gene revealed some additional interesting features. The portion (299 bp in length) of the proximal part of this promoter, located at −222 bp to −521 bp upstream of the start codon (Fig. 5), can potentially form a secondary structure because of several inverted repeats. A more detailed analysis revealed high resemblance of this sequence to the miniature inverted-repeat transposable elements (MITEs). MITEs are non-autonomous DNA (class II) transposable elements characterized by short length (∼60–700 bp), no coding capacity, usually short (10–30 bp) terminal inverted repeats (TIR), and target site (2–3 bp, rich in A and T) duplicated (TSD) at insertion, so that the MITE is flanked by a direct repeat (Fig. 6). MITEs have a potential to form a hairpin-like secondary structure and show preference to insert in introns or near the 5′ or 3′ ends of genes, but not in coding regions (Wessler et al. 1995; Feschotte et al. 2002).

Fig. 6
figure 6

Miniature inverted repeat transposable element (MITE) in the promoter region of the barley Hsdr4 gene. MITE (299 bp in length) was identified −222 bp to −521 bp upstream of the start codon. The empty arrows refer to the target site duplication (TSD) sequences and filled arrows denote terminal inverted repeats (TIRs)

Full size image

The sequence of this 299 bp portion of Hsdr4 promoter exhibited no homology to known repetitive elements in the Plant Repeat Database (http://www.tigr.org/tdb/e2 k1/plant.repeats/index.shtml). Nevertheless, its structure resembled the MITEs features. In our putative MITE, the insertion is flanked by 3 bp direct repeats (TTA), and the ends of the element are represented by 11 bp inverted terminal repeats (TIRs) (GGCCTCGTTTG) (Fig. 6). Its internal sequence displayed high similarity (E-value 5E−92, score 345 bits, 91% identities) to a part of the first intron of cyclic nucleotide-gated ion channel 4 (nec1) gene of barley (acc. no. AY972619). The detected MITE is especially capable of forming hairpin-like secondary structure (Fig. 7).

Fig. 7
figure 7

Potential secondary structures formed by 299-bp putative MITE of promoter region Hsdr4 gene from cultivated barley (A), and wild barley - sensitive (B) and tolerant (C) genotypes. The minimum-energy folding of elements and free energies (ΔG) are predicted by DNA MFOLD program (version 3.1) (Zuker 2003; http://www.bioinfo.rpi.edu/applications/mfold/old/dna/)

Full size image

Chromosomal assignment of the Hsdr4 gene

To map the Hsdr4 gene on wild barley chromosomes one of the introns of this gene was amplified and sequenced from the two parental lines of the MA10-30 × WQ23-38 wild barley mapping population, which was evaluated for drought tolerance (Chen 2005). Sequence alignment revealed a few SNPs (single nucleotide polymorphism) between the parental lines. One of the SNPs corresponded to the Taq1 restriction site that enabled us to map the cloned gene. Using the presence versus absence of restriction site and aligning our data with existing mapping data for this mapping population, the Hsdr4 gene was mapped to the long arm of chromosome 3H between markers EBmac541 and EBmag705 (Fig. 8).

Fig. 8
figure 8

Chromosome assignment of Hsdr4 gene on barley chromosome 3H. The map location of Hsdr4 gene is boxed. Units are in cM. Genetic mapping was conducted using MultiPoint program (http://www.multiqtl.com)

Full size image

Sequence comparison of Hsdr4 between tolerant and sensitive barley genotypes

Since the expression pattern of Hsdr4 gene under dehydration stress was different between tolerant and sensitive genotypes, we wondered whether it may derive from sequence differences in the promoter region. Moreover, keeping in mind our hypothesized importance of this gene in drought resistance, it was also interesting to look for possible differences between tolerant and sensitive genotypes in its transcribed part. Hsdr4 gene and its promoter region of two tolerant and two sensitive genotypes were sequenced. Sequence alignment showed a high degree of sequence similarity between the four genotypes. The only differences found between tolerant and sensitive genotypes were in some SNPs in intron (data not shown) and in the promoter region (Fig. 5). Several SNPs were found in the putative MITE in comparisons between tolerant and sensitive wild barley genotypes and cultivated barley that may cause different folding patterns (Fig. 7).

Discussion

Differential expression of wild barley under dehydration by cDNA-AFLP analysis

Plant adaptation to environmental stress is regulated through multiple physiological mechanisms at the cellular, tissue, and whole-plant levels (Hazen et al. 2003; Hazen et al. 2005; Ito et al. 2006; Yamaguchi-Shinozaki and Shinozaki 2005). Identification and detailed analysis of a large number of candidate genes involved in drought tolerance may enable to elucidate the molecular basis of drought resistance complexity.

The aim of our study was to investigate expression pattern of genes that are altered in response to dehydration stress and are potentially involved in the tolerance to dehydration in wild barley, by using the cDNA-AFLP fingerprinting method. cDNA-AFLP is one of the technologies that is successfully employed for characterization of gene expression under abiotic and biotic stresses in plants including water deficit (Dubos and Plomion 2003; Yang et al. 2003; Knight et al. 2006; Rodriguez et al. 2006). The cDNA-AFLP technique has the advantage of technical simplicity as compared with microarray, substracctive hybridization and less labor intensive as compared with other mRNA fingerprinting methods such as differential display (DDRT-PCR) (Liang and Pardee, 1992).

We used the cDNA-AFLP method to compare expression patterns of genotypes that under dehydration stress showed higher relative water content (RWC) and lower water loss rate (WLR), as compared to genotypes with lower RWC and lower WLR (Suprunova et al. 2004). The expression patterns of these genotypes that were regarded as tolerant and susceptible to water stress, were compared under control and dehydration stress. This kind of comparison enabled us to relate the function of the identified gene not only to those that are induced by dehydration stress but to those that are involved in mechanisms of tolerance or resistance to water stress. We have used this approach previously by comparing patterns of dehydrin gene expression in the same tolerant and sensitive wild barley genotypes (Suprunova et al. 2004). Recently, the same approach was described by Rampino et al. (2006) with resistant and tolerant wheat genotypes. This aspect makes the main difference between the present study and most of others that use expression analysis by cDNA-AFLP to provide a genome-wise description of gene expression profiles. Moreover, we used for our study wild relative of cultivated barley, Hordeum spontaneum, which is a unique genetic resource harboring adaptive mechanisms of stress resistance from a unique (desert) ecogeographic region.

For the efficiency of sequencing efforts, we carefully selected the candidate TDFs that were sequenced and further analyzed. For that, the method developed by Brugmans et al. (2003) was employed. By using this method we were able to isolate one amplicon per each TDF for direct sequencing, omitting other co-migrating fragments usually presented within the PCR products of the TDF excised from the gel. This was achieved by three steps of PCR amplifications: (a) re-amplification of each TDF with its corresponding selective primers, (b) second amplification by a set of degenerated primers in order to define additional four bases adjacent to the first one or two selective nucleotides; (c) third amplification with a pair of EcoRI primer + four additional nucleotides and MseI primer + four additional nucleotides. Usually, DNA-AFLP fragments are cloned, confirmed by PCR or restriction analysis and then sequenced. Thus, Umezawa et al. (2002) used 64 PCs for selective amplification. They describe that from 130,000 fragments 140 bands were differentially expressed; their putative gene function according to BLAST comparison was described, but the expression of only four fragments was confirmed by RNA dot blot. Other studies describe the isolation of TDFs and re-amplification with non-selective primers (Bruggmann et al. 2005). These authors indeed used the maximal number of primer combinations (256); and out of the 363 TDFs isolated from the cDNA-AFLP gel, the induced expression of 92 TDFs was confirmed by gel blot; 1–8 clones per each TDF were sequenced because of mixed PCR products. Eventually, the expression of seven fragments was confirmed by quantitative RT-PCR. In the extensive work by Blanco et al. (2005) 128 PCs were used for amplification; however, these authors focused on analysis of 59 fragments out of 5680 TDFs. Eventually, the expression of 12 candidates was extensively described and analyzed. The resulting patterns of cDNA-AFLP are highly reproducible and sensitive and correlate well with Northern analysis, Q-PCR and microarrays (Donson et al. 2002; Avrova et al. 2003; Bruggmann et al. 2005). In our study, out of seven TDFs chosen for RT-PCR analysis, four displayed high coincidence of cDNA-AFLP and RT-PCR expression patterns. Q-PCR expression analysis of the Hsdr4 gene validated the results obtained by cDNA-AFLP and RT- PCR analysis and revealed a higher expression level of the Hsdr4 gene after 12 h drought stress in both tolerant genotypes (JR1 and JR2) as compared to the sensitive ones (JS1 and JS2).

Although cDNA-AFLP is widely used for expression analysis, there is still a disadvantage of low genuine “yield”, when the objective is to isolate real relevant genes. By using the approach of (a) selecting only those TDFs that were up-regulated in tolerant plants under stress and (b) adopting the procedure (with some modifications) developed by Brugmans et al. (2003), we managed to achieve the goal with relatively low number of primer combinations.

Sequence analysis of the differentially expressed TDFs

Sequence analysis of the eleven TDFs revealed that they represent genes involved in basic metabolic activity as well as genes encoding for stress related proteins (Table 3). Pectin glucuronyltransferase (TDF-5) and 3-β-glucuronosyltransferase (TDF-13) belong to a large group of glycosyltransferases (GTs). One of the biological functions of GTs is to catalyze the biosynthesis of polysaccharides including cellulose, hemicelluloses, and pectins the main components of the cell wall (Zhong and Ye 2003). Cystathionine β-lyase (TDF-11) is an enzyme that plays a central role in the methionine (Met) biosynthesis pathway and, therefore, in plant growth and development (Maimann et al. 2000). Cullin3 (TDF-14) is a member of the family, which comprises the cullin-dependent ubiquitin ligases that control the rapid and selective degradation of important regulatory proteins involved in cell cycle progression and development (Thomann et al. 2005).

TDF-12 induced in the tolerant genotype under 12 h dehydration stress exhibited homology with T. aestivum S-adenosylmethionine decarboxylase protein (SAMDC). SAMDC is one of the rate-limiting enzymes in the biosynthesis of polyamines, which play a crucial role in morphogenesis, embryogenesis, floral and fruit development, and root formation (Walden et al. 1997). SAMDC was previously described as a stress-related protein involved in salt, drought, and cold tolerances of a variety of plants (Li and Chen 2000; Roy and Wu 2002; Hao et al. 2005).

The signal transduction and stress responding category also includes a gene encoding for Rho-GTPase-activating protein, homologous to TDF-4. Rho-GTPase-activating proteins (RhoGAPs) are among the components regulating the small GTPases of the Rho family that play an important role in molecular switches in many signal transduction pathways and functions. The Rho-GTPase family has emerged as a key regulator of actin cytoskeleton in yeast and animal cells (Hall 1998). In plants, Rho-GTPases have been shown to control different cellular processes such as pollen and root hair tip growth and H2O2 production as a second messenger in plant signaling (Yang 2002). One of the most interesting findings about Rho-GTPases is the demonstration of its involvement in the control of stomatal closure and therefore plant water homeostasis. The Arabidopsis Rho-related GTPase protein, AtRac1, was identified as a central component in ABA-mediated stomatal closure process. It was shown that ABA treatment induced inactivation of AtRac GTPases, leading to stomatal closure through the disruption of the guard cell actin cytoskeleton (Lemichez et al. 2001).

Novel drought stress responsive gene, Hsdr4 from wild barley

The isolated Hsdr4 gene is composed of 3 exons and encodes 250 amino acid putative Rho-GTPase-activating protein that contains a conserved domain very similar to the SEC14 domain (Sec14p-like lipid-binding domain). Sec14p-related proteins are widespread and found in mammals, fungi, and plants. For example, Arabidopsis patellin1 (PATL1), characterized also by Golgi dynamics domains (GOLD), is involved in plant cell cytokinesis (Peterman et al. 2004). Another Arabidopsis phosphatidylinositol transfer protein, AtSfh1p, possesses the Sec14p-nodulin domain and acts as regulator of polarized membrane growth of root hairs (Vincent et al. 2005). Sec14p-nodulin domain is also shared by four members of PITP-like protein family (LjPLPs) from Lotus japonicus. These proteins may function in lipid-signaling pathways that regulate membrane biogenesis during nodulation (Kapranov et al. 2001). One of the members of the Sec14p family from soybean, Ssh1p, was characterized as a component of a stress response pathway that serves to protect the adult plant under osmotic stress (Monks et al. 2001).

All hitherto characterized plant proteins with the Sec-14-like domain belong to the phosphatidylinositol transfer protein (PITP) family. To the best of our knowledge, the Sec-14-like domain was not previously reported in any Rho GTPase-activating proteins (RhoGAPs) in plants. The GAP catalytic domain and the Cdc42/Rac-interactive binding (CRIB) domain are characteristic of this family (Borg et al. 1999; Wu et al. 2000). Nevertheless, several mammalian RhoGAP-like proteins, including neurofibromin NF1 and members of the multifunctional Dbl family were reported recently to possess Sec14p lipid binding domain (Aravind et al. 1999). It was proposed that these proteins with Sec14p-like domain play a role in lipid regulation of the Rho-mediated signaling pathway. Here, we report for the first time on a new putative Rho GTPase-activating protein from wild barley that possesses the Sec14p-lipid binding domain. Studying of the function of the protein encoded by this transcript will be the next major step towards determining its role in stress response, including water stress.

Identification and mapping of relevant QTLs and/or co-localization of QTLs with candidate genes are useful approaches for dissecting the genetic basis of complex traits such as drought tolerance. Hsdr4 gene was mapped on the long arm of the barley chromosome 3H between markers EBmac541 and EBmag705. None of the QTLs associated with drought tolerance were revealed in this chromosome region in our mapping population of wild barley (Chen 2005). However, according to QTL mapping of another barley population, Tadmor × Er/Apm, this region contains a QTL for osmotic potential (OP) and a QTL that affects the relative water content (RWC) (Diab et al. 2004). Moreover, the Hsdr4 rice orthologue, encoding for Rho-GTPase-activating protein-like (BAC clone AP003259), is located in rice chromosome 1 at position 167.2–169.5 cM. In this region two QTLs, associated with total root dry weight (TRDW) and penetrated root thickness (PRT) related to drought resistance, were identified in rice (Nguyen et al. 2004). Our finding on the higher expression level of Hsdr4 gene under dehydration stress in drought tolerant as compared with drought sensitive genotypes and its co-localization with QTLs associated with drought tolerance allow us to suggest that Hsdr4 could be a viable candidate gene for the determinant of water-stress tolerance. To investigate the gene further to validate its possible role in tolerance to dehydration stress, transgenic work will be conducted.

Sequence analysis of the Hsdr4 promoter region

Analysis of the 1.7-kb sequence corresponding to the Hsdr4 promoter resulted in the identification of several potential binding sites for transcription factors. More than half of the identified motifs are typical stress-related cis-acting elements that have been identified in the promoter regions of a number of other drought- and ABA-induced genes. For example, MYC and MYB recognition sites (box A and C, Fig. 5 and Table 3) in Arabidopsis rd22 promoter function as cis-acting elements in drought-inducible expression of rd22 gene (Abe et al. 1997, 2003). Another MYC-like motif (box B) was found in drought-, salt-, and dark-induced Arabidopsis erd1 gene. GCC-motif (box D) seems to be associated with jasmonate-responsive gene expression (Brown et al. 2003; Liu et al. 2006). Box E (Fig. 4) is a light-responsive element that is sufficient to mediate circadian cycling of the barley chloroplast psbD gene promoter activated by blue, white, or UV-light (Thum et al. 2001). Low temperature responsive element, LTRE (box F), seems to be involved in the expression of such cold-regulated genes as cor15a from A. thaliana (Baker et al. 1994), bn115 from winter Brassica napus (Jiang et al. 1996), and wcs120 from wheat (Ouellet et al. 1998). GT-1 cis-element (box G) interacts with a GT-1-like transcription factor that plays a role in pathogen- and salt-induced SCaM-4 gene expression in both soybean and Arabidopsis (Park et al. 2004). Finally, box H (Fig. 5) showed similarity with TAAAG motif found in the promoter of the potato KST1 gene encoding for a K+ influx channel of guard cells. TAAAG elements are target sites for trans-acting Dof (zinc finger) transcription factors controlling guard cell-specific gene expression (Plesch et al. 2001). Our results of Hsdr4 promoter analysis are consistent with the expression data of this gene. Strong indications of Hsdr4 induction by dehydration stress and preliminary results on salt induction (not shown) allow us to speculate that the revealed MYC, MYB, LTRE, and GT-1 elements may be related to stress responsiveness. Whether different responsive elements of the Hsdr4 promoter are in vitro targets for corresponding transcription factors remains to be elucidated in future studies.

The MITE insertion, which was identified in the promoter region of the Hsdr4 gene, seems to be a new transposable element in barley. MITEs are high-copied DNA transposons associated with the non-coding part of genes and, therefore, are useful in plant genome analysis as molecular markers and potential tools in plant systematic and phylogenetic studies (Feng 2003). The close proximity of MITEs to the coding regions suggests their role in transcription, splicing, and translational regulation of the genes. For example, MITE insertion in intron 2 of one of the nec1 gene alleles in barley caused alternative splicing, frame shift, and production of a defective protein (Rostoks et al. 2006). MITEs have been suggested to supply cis-acting elements affecting, presumably, gene expression (Bureau and Wessler 1994; Pozueta-Romero et al. 1996; Yang et al. 2001). Several features known to be critical for the characteristic of MITEs, are also characteristic of the MITE-like insertion detected in our study: (i) the proximity to the start codon of Hsdr4; (ii) presence of a few putative cis-element in the internal sequence; and (iii) the predicted ability to form sequence-depending alternative hairpin structures with different free energy (Fig. 7). These features may provide the different regulatory mechanism since such structures are known to be critical for transcriptional regulation (Wadkins et al. 2000).

In summary, we demonstrated that the study of wild relatives of cultivated plants from diverse ecogeographic regions combined with functional genomic tools such as the cDNA-AFLP technique has great potential for identification of novel candidate genes related to stress tolerance. We described here the sequence, structure and expression, and genetic mapping of a new candidate gene (Hsdr4) for water-stress tolerance in wild barley. Hsdr4, induced by dehydration stress in tolerant young plants of wild barley, encodes a putative Rho-GTPase-activating protein with a Sec14p-like lipid-binding domain, and has several potential stress-related cis-elements in the promoter region. Due to its expression patterns and structure, Hsdr4 is a promising candidate gene for further studies on the molecular genetic basis of water-stress tolerance in plants.