• Open Access

Emerging trends in the functional genomics of the abiotic stress response in crop plants

Authors


* Correspondence (fax 91-11-24115095; e-mail akhilesh@genomeindia.org)

Summary

Plants are exposed to different abiotic stresses, such as water deficit, high temperature, salinity, cold, heavy metals and mechanical wounding, under field conditions. It is estimated that such stress conditions can potentially reduce the yield of crop plants by more than 50%. Investigations of the physiological, biochemical and molecular aspects of stress tolerance have been conducted to unravel the intrinsic mechanisms developed during evolution to mitigate against stress by plants. Before the advent of the genomics era, researchers primarily used a gene-by-gene approach to decipher the function of the genes involved in the abiotic stress response. However, abiotic stress tolerance is a complex trait and, although large numbers of genes have been identified to be involved in the abiotic stress response, there remain large gaps in our understanding of the trait. The availability of the genome sequences of certain important plant species has enabled the use of strategies, such as genome-wide expression profiling, to identify the genes associated with the stress response, followed by the verification of gene function by the analysis of mutants and transgenics. Certain components of both abscisic acid-dependent and -independent cascades involved in the stress response have already been identified. Information originating from the genome-wide analysis of abiotic stress tolerance will help to provide an insight into the stress-responsive network(s), and may allow the modification of this network to reduce the loss caused by stress and to increase agricultural productivity.

Introduction

In order to understand the basis of stress tolerance, the diversity of the stress response and its utility for the survival of plants should be investigated. As environmental conditions begin to change, plants will sense this. After this initial perception, a signal is relayed via several signal transduction cascades that amplify the signal and notify parallel pathways. Various strategies have been employed to isolate the genes that are involved in the stress response. Several genes that are known to be responsive to stress or are involved in imparting tolerance can be related to cellular functions. Information about such stress-responsive genes has been obtained largely using conventional approaches. However, the challenge still remains to integrate the function of these genes logically to generate a global understanding of the stress response process (Bohnert et al., 2006; Valliyodan and Nguyen, 2006). The complete genome sequence of rice and Arabidopsis and emerging sequence information for several other plant genomes, such as Populus, Medicago, lotus, tomato and maize, have given rise to the use of tools which can aid in the determination of the function of many genes simultaneously (The Arabidopsis Genome Initiative, 2000; International Rice Genome Sequencing Project, 2005; Rensink and Buell, 2005; Vij et al., 2006). Functional genomics employs multiple parallel approaches, including global transcript profiling coupled with the use of mutants and transgenics, to study gene function in a high-throughput mode. The aim of these genome-wide efforts is to finally link the genome to the phenome (Figure 1). This change in the approach to the study of the abiotic stress response has undoubtedly reduced the time of completion of an otherwise arduous task in plants with well-established genomics platforms, such as Arabidopsis (Bray, 2002; Denby and Gehring, 2005; Yamaguchi-Shinozaki and Shinozaki, 2006). In this review, the use of such genome-wide strategies in understanding the basis of abiotic stress tolerance, with an emphasis on crop plants such as rice, is discussed.

Figure 1.

A typical functional genomics approach to improve crop performance under abiotic stress conditions. 2DGE, two-dimensional gel electrophoresis; EST, expression sequence tag; MALDI-TOF, matrix-assisted laser desorption/ionization-time of flight; MPSS, massively parallel signature sequencing; QTL, quantitative trait locus; SAGE, serial analysis of gene expression.

Global expression profiling during abiotic stress

The availability of a large volume of genomic data has provided information about the gene content of plants. Partial or complete sequences of cDNAs often provide a firm basis of the dimension of the transcriptome. There are three main databases [National Center for Biotechnology Information (NCBI) Unigenes, http://www.ncbi.nlm.nih.gov/; The Institute for Genomic Research (TIGR) Gene Indices, http://www.tigr.org; Sputnik, http://mips.gsf.de/proj/sputnik] which serve to organize the available plant expression sequence tags (ESTs), together with well-characterized genes, into non-redundant gene clusters. ESTs have formed the core of studies on the global gene expression of multiple stress tolerance traits in Arabidopsis and rice (Rensink and Buell, 2005). The analysis of ESTs generated from cDNA libraries of salt-stressed rice by Bohnert et al. (2001) showed that there was an increase in transcripts related to cell rescue, defence, transport, energy and metabolism, but the majority of stress-inducible genes could not be assigned a function. In a study performed to identify salt stress-inducible ESTs obtained from polymerase chain reaction (PCR) subtraction in salt-tolerant rice, 384 genes were identified as salt responsive, ∼5% of which were also confirmed by Northern analysis. Almost 50% of these genes were identified for involvement in detoxification, stress response, growth and development (Shiozaki et al., 2005). In addition, a large number of ESTs related to abiotic stress have been identified in rice (Babu et al., 2002; de los Reyes et al., 2003; Sahi et al., 2003, 2006). Recently, large-scale EST sequencing has been reported from cDNA libraries composed of wheat plants exposed to abiotic stress conditions. In total, 73 521 ESTs were generated from cDNA libraries composed of wheat plants subjected to various abiotic stress conditions and at different developmental stages. In addition, 196 041 ESTs available from two other sources [National Science Foundation (NSF) wheat EST sequencing program and DuPont] were also used for analysis. These two EST sources together generated 75 488 unique sequences enriched in stress-regulated genes, such as those coding for transport, cryoprotection, signalling cascades and transcription factors (Houde et al., 2006). This information is a valuable starting point for a researcher interested in working on abiotic stress-responsive wheat genes. Such large-scale EST projects are useful for other crop species, for which genome-level sequencing would probably never be performed, to aid in the identification of the genes involved in abiotic stress.

Serial analysis of gene expression (SAGE) is a powerful tool which can be used to quantify global gene expression. This technique depends on the generation of unique transcript-specific short sequences of 9–17 base pairs (Velculescu et al., 1995; Saha et al., 2002). The method can potentially identify > 49 (> 262 144) tags, which is much more than the estimated number of genes in Arabidopsis (25 498; The Arabidopsis Genome Initiative, 2000) and rice (37 544; International Rice Genome Sequencing Project, 2005). Quantification of a particular tag provides the expression level of the corresponding transcript. It also unravels novel expressed regions of the genome. SAGE was first used for the quantification of global gene expression in rice by Matsumura et al. (1999); 10 122 tags from 5921 expressed genes from rice seedlings were analysed; 18 genes were found to be anaerobically induced and six genes were repressed. The anaerobically induced genes were those coding for prolamin, expansin and glycine-rich cell wall protein (Matsumura et al., 1999). In Arabidopsis, SAGE has been used to analyse changes in gene expression in leaves on exposure to cold stress (Jung et al., 2003). Genes involved in cell rescue, defence, cell death, ageing, protein synthesis, metabolism, transport facilitation and protein destination were highly expressed and photosynthesis genes were down-regulated. Similarly, SAGE tags derived from unstressed pollen were compared with those derived from cold-stressed pollen by Lee and Lee (2003). Cold stress at 0 °C for 72 h did not affect the expression of the majority of transcripts in pollen. Further, many cold-responsive genes, such as those coding for cold-responsive (COR), lipid transfer protein and β-amylase, were expressed at normal levels or were induced to a lesser extent in cold-stressed pollen, suggesting that the cold sensitivity of Arabidopsis pollen may be attributed to a weak accumulation of proteins involved in the stress response (Lee and Lee, 2003).

Massively parallel signature sequencing (MPSS) is another powerful technique for transcription profiling on a genome-wide scale. This technique, like SAGE, can be used to obtain a representation of the mRNA population in the sample, which can be related to ESTs, mRNAs or the whole genome sequence, but the data generated are much larger in magnitude (Brenner et al., 2000; Pollock, 2002). An MPSS resource is available for three plant species (Arabidopsis, rice and grape) in a public database (http://mpss.udel.edu). The resource for rice includes 20 MPSS libraries constructed from different tissues, including three from abiotic stress conditions, namely cold, salt and dehydration (Nakano et al., 2006). However, there are very few reports of the use of MPSS in plant gene expression profiling in response to abiotic stress. In one such example, MPSS was used to compare abscisic acid (ABA)-responsive gene expression in ABA-insensitive mutants (abi1–1) and wild-type (WT) Arabidopsis plants. Only a small set of genes was regulated by ABA in the ABA-insensitive mutant (abi1-1), whereas the ABA-responsive gene expression of the majority of genes was impaired in the abi1-1 mutant, suggesting the presence of two ABA signalling pathways, only one of which is impaired in the abi1-1 mutant (Hoth et al., 2002). However, because of its cost, the full potential of MPSS in the global expression profiling of the abiotic stress response is yet to be realized.

Microarrays have revolutionized global gene expression profiling by allowing the entire gene complement of the genome to be studied in a single experiment (Duggan et al., 1999; Jiao et al., 2005; Li L. et al., 2005). cDNA and oligonucleotide microarrays have been widely used in plants, such as Arabidopsis, rice, maize, strawberry, petunia, ice plants and lima bean, to study and compare global gene expression patterns. Their use in the study of the genes induced in response to various abiotic stresses in crop plants is shown in Table 1. The use of a microarray to study global gene expression profiling in response to abiotic stress in rice was first reported by Kawasaki et al. (2001), who compared the gene expression profiles in salt-tolerant (var. Pokkali) and salt-sensitive (var. IR29) rice in response to salt stress. The analysis was performed using a cDNA microarray comprising 1728 cDNA clones prepared from unstressed or salt-stressed roots of Pokkali. The main difference between the expression patterns of the two varieties was the delayed timing of the IR29 response in terms of kinetics of gene expression, which could be responsible for its salt sensitivity (Kawasaki et al., 2001). A rice genome array containing 48 564 japonica and 1260 indica sequences has also been used to compare the transcriptome of salt-tolerant (FL478) and salt-sensitive (IR29) rice varieties. The response of the two varieties was strikingly different, with a much larger number of genes expressed in IR29 than in FL478 on exposure to salt stress. This difference in expression pattern is mainly attributed to the fact that FL478 is salt tolerant and maintains a low Na+ to K+ ratio (Walia et al., 2005). In another study, genes associated with quantitative trait loci (QTLs) for osmotic adjustment (OA) were identified by expression profiling of ∼21 000 rice genes in divergent accessions. Six hundred and sixty-two genes showed differential expression amongst the parental lines in response to drought stress. Interestingly, the high- and low-OA parents showed different responses to drought stress. Of the 69 genes up-regulated in all the high-OA lines, nine were induced only in high-OA lines. Several potential candidate genes were identified at each of the five QTL locations related to OA based on their expression patterns in the low- and high-OA lines, and these included genes coding for an small nuclear ribonucleoprotein partide (snRNP) auxiliary factor, a late embryogenesis abundant (LEA) protein, protein phosphatase 2C and a Sar1 homologue (Hazen et al., 2005). Oligonucleotide microarray analysis has also been used to study the expression profile of rice genes under various environmental, biological and chemical stress treatments. Interestingly, a correlation between the expression pattern under stress conditions and the developmental process was found (Cooper et al., 2003). Microarrays have also been used to compare the gene expression profiles during pollination/fertilization under drought and wounding stress. More than one-half of the pollination/fertilization-related genes were regulated by dehydration stress, indicating that water deficit may be a crucial factor during pollination and fertilization (Lan et al., 2005). In addition to cDNA and oligonucleotide microarrays, tiling-path arrays have been used to study gene expression in plants (Jiao et al., 2005; Li L. et al., 2005, 2006; Stolc et al., 2005). The advantage of tiling-path arrays over conventional microarrays is that they are not presumptuous about the gene structure, and hence provide unbiased and more accurate information about the transcriptome. In addition, they provide information about transcriptional control at the chromosomal level. The use of tiling-path arrays could help to provide novel information about the abiotic stress transcriptome at the genome-wide level.

Table 1.  Microarray analysis of gene expression profiles under abiotic stress conditions in crop plants (selected examples)
PlantStressReference
  1. ABA, abscisic acid.

Capsicum annumCold stressHwang et al. (2005)
Hordeum vulgareDrought and salinityOzturk et al. (2002)
Iron deficiencyNegishi et al. (2002)
Oryza sativaSalt stressKawasaki et al. (2001)
Multiple abiotic stressesCooper et al. (2003)
Cold, drought, salinity and ABA treatmentRabbani et al. (2003)
ABA treatmentYazaki et al. (2003)
Chilling stressYamaguchi et al. (2004)
Salt stressChao et al. (2005)
Drought stressHazen et al. (2005)
Drought stressLan et al. (2005)
Salt stressWalia et al. (2005)
Low nitrogen stressLian et al. (2006)
Populus euphraticaMultiple abiotic stressesBrosche et al. (2005)
Solanum tuberosumCold, heat and salt stressRensink et al. (2005)
Sorghum bicolorDehydration, salt and ABA treatmentBuchanan et al. (2005)
Triticum aestivumLow-temperature stressGulick et al. (2005)
Salt stressKawaura et al. (2006)
Zea maysWater stressYu and Setter (2003)

As an initial step, a reference library of gene expression through the life cycle of model plant species such as rice, together with global expression profiling in response to stress, would go a long way to decipher the plant stress-responsive network. To this end, several general databases have been made available which contain plant expression information obtained from microarray analysis, e.g. Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/projects/geo) and ArrayExpress (http://www.ebi.ac.uk/arrayexpress). Some of the databases containing rice microarray-related information are the Rice Expression Database (http://www.red.dna.affrc.go.jp/RED), NSF Rice Oligonucleotide Array Project and Database (http://www.ricearray.org) and Virtual Centre for Cellular Expression Profiling in Rice (http://130.132.8.83/rc) (Rensink and Buell, 2005). The integration of these resources would prove useful in establishing at least the preliminary response of rice genes under abiotic stress conditions.

Map-based cloning of abiotic stress-related gene loci and QTLs

A positional cloning strategy allows the use of a phenotype to determine the position of the mutated gene or natural allele by examining the linkage to markers whose physical location in the genome is already known. To date, mutant/variant analysis has been used to identify 1698 rice genes, for which the trait of stress tolerance and disease resistance accounted for more than 17% of the mutants analysed (Kurata et al., 2005). The availability of the Arabidopsis and rice genome sequences, together with improvements in the methods used to detect DNA polymorphisms, has made map-based cloning a viable option for global functional genomics (Jander et al., 2002). There are 37 344 single nucleotide polymorphisms (SNPs), 18 579 insertions/deletions (InDels) and 747 large InDels available at The Arabidopsis Information Resource (TAIR) website. A large number of markers are also available for rice, including 3267 markers released in 2000, 332 PCR-based genetic markers released in 2002 (rgp.dna.affrc.go.jp) and over 18 000 simple sequence repeats (SSRs) (International Rice Genome Sequencing Project, 2005).

A large number of genes in the abiotic stress response pathway have been identified using the map-based cloning approach, including SOS1, SOS2, SOS3, SOS4, SOS5, HOS1, SpI7, STT3, FRO1, LOS5/ABA3 and AtCesA8/IRX1 (Liu and Zhu, 1998; Liu et al., 2000; Shi et al., 2000, 2003; Lee H. et al., 2001; Xiong et al., 2001; Lee B.H. et al., 2002; Shi et al., 2002; Yamanouchi et al., 2002; Koiwa et al., 2003; Chen et al., 2005). In rice, about 60 000 mutants have been generated in the IR64 background using both chemical [ethyl methane sulphonate (EMS) and diepoxybutane] and irradiation (fast neutron and γ-ray) mutagenesis (Wu et al., 2005). More than 8% of the mutant population showed phenotypic changes with respect to 35 phenotypic traits. The mutant collection is also being screened for alterations in stress response (Wu et al., 2005). In addition, efforts are being made to use sequence-indexed knockout mutant resources developed by rice researchers world-wide to identify insertions in stress-associated genes (SAGs). The SAG mutant lines will be screened for response to drought stress, and this information, together with expression information (from a SAG microarray), will be made available in the near future (http://www.generationcp.org/research/2006_SP2/2005-09.pdf). Targeting induced local lesions in genomes (TILLING) is another high-throughput technology helpful in identifying mutations in a selected gene or a variant allele (Henikoff and Comai, 2003). TILLING projects are underway for plant species such as Arabidopsis, lotus, maize, wheat and Brassica (Gilchrist and Haughn, 2005). In rice, mutants generated in the IR64 background using EMS and diepoxybutane have been used to generate DNA pools for TILLING, and were screened for mutations in selected genes (Wu et al., 2005). With the advantage of amenability to high throughput, together with the capability of generating an allelic series at a given locus, TILLING will also find application in the identification of useful alleles for abiotic stress tolerance.

QTLs are specific genetic loci in the genome associated with a particular trait. Stress tolerance is a complex trait, and dissection of its QTLs would be of immense value in understanding the stress response and would be useful for plant breeders (Gorantla et al., 2005). Several QTLs involved in the stress response have been reported recently (Lin et al., 2004; Ren et al., 2005; Salvi and Tuberosa, 2005; Xu et al., 2005; Tuberosa and Salvi, 2006). Most of the plant QTLs cloned to date have been obtained using a map-based cloning strategy. Lin et al. (2004) mapped eight QTLs responsible for variation in K+ or Na+ content from an F2 population derived from a cross between a salt-tolerant indica variety (Nona Bokra) and a susceptible japonica variety (Koshihikari) Of these, SKC1, a major QTL for shoot K+ content, was mapped to chromosome 1. SKC1 was obtained by positional cloning and was found to code for an HKT-type transporter. Further physiological studies with SKC1 showed that it plays a vital role in maintaining the K+ to Na+ ratio. This role of SKC1 shows that it is an important determinant of salt stress tolerance, and hence would be a valuable QTL for engineering salt stress tolerance in rice (Ren et al., 2005). A map-based approach was also demonstrated to be useful in the identification of an important submergence tolerance QTL present on chromosome 9 of rice. Three genes belonging to the ethylene-response-factor (ERF) family were identified on this locus, designated as Sub1, the variation of one of which, Sub1A, results in tolerance/susceptibility to submergence. Sub1 from the submergence-tolerant variety was introgressed in flooding-susceptible local rice varieties. The new varieties showed submergence tolerance without compromising on yield or other agronomic traits, demonstrating the efficacy of this locus (Xu et al., 2006). The expression profile of genes in a QTL interval associated with the abiotic stress response is also being used to identify target genes. Gorantla et al. (2005) used information from ESTs sequenced from drought-stressed cDNA libraries to generate a transcript map of rice. Several known stress-responsive genes coding for mitogen-activated protein kinase, OSMYB1, EREBP-like protein, helicase-like transcription factor and 14-3-3 protein homologue could be identified at drought QTL locations. In another effort, over 20 000 introgression lines (ILs) have been developed in three elite rice genetic backgrounds for diverse traits, including physiological, morpho-agronomic and those related to resistance/tolerance to biotic and abiotic stresses. These ILs could provide a rich source for the identification of candidate genes and the cloning of QTLs involved in the stress response (Li Z.K. et al., 2005). Another possible resource for QTL analysis is the Oryza Map Alignment Project (OMAP). This project aims to develop bacterial artificial chromosome (BAC)/sequencing tag connector (STC) physical maps of 11 wild and one cultivated rice species, which will ultimately be aligned to the finished rice (Oryza sativa ssp. japonica cv. Nipponbare) genome sequence (Wing et al., 2005; Ammiraju et al., 2006), and will definitely prove to be useful for the identification of stress-related QTLs. The influence of natural alleles or mutated genes on the expression of downstream genes or functions will aid in the unravelling of the stress response mechanism in plants.

Gene tagging interrupts abiotic stress response genes to unravel their functions

Insertional mutagenesis in plants usually involves the use of T-DNA or transposable elements. It has the advantage of a genome-wide distribution with preferential insertion in gene-rich regions (An et al., 2005b). Insertional mutagenesis has been widely employed to characterize abiotic stress-responsive genes, including those coding for AtHKT1 (a high-affinity potassium transporter), CBL1 (calcineurin B-like protein), OsRLK1 (LRR-type receptor-like protein kinase), CIPK3 (calcium-associated protein kinase), OSM1/SYP61 (syntaxin) and HOS10 (R2R3-type MYB transcription factor) (Rus et al., 2001; Zhu et al., 2002, 2005; Cheong et al., 2003; Kim et al., 2003; Lee S. et al., 2004).

A large number of insertional mutants are available for both Arabidopsis (Krysan et al., 1999; Speulman et al., 1999; Tissier et al., 1999; Sessions et al., 2002; Szabados et al., 2002) and rice (An et al., 2005a,b). Jeon et al. (2000) have estimated that ∼200 000 tagged lines of Arabidopsis are required to obtain a 99% probability of finding a T-DNA insertion. In Arabidopsis, such a resource is available, as more than 225 000 T-DNA lines have been generated (Alonso et al., 2003). In rice also, lines with > 200 000 T-DNA tags are available, giving the probability of finding a knockout in a given rice gene at > 90% (An et al., 2005b). Thus, saturation mutagenesis has been achieved for both rice and Arabidopsis with tagged gene disruptions. Large-scale forward genetic screens have been used to identify abiotic stress response determinants in a T-DNA-mutagenized Arabidopsis population in the RD29a-LUC background. More than 200 mutants with altered stress/ABA response were identified from 250 000 independent insertion lines. These included mutations in genes coding for transcription factors, syntaxin, ABA biosynthetic enzyme, SUMO E3 ligase and the sodium transporter HKT1 (Koiwa et al., 2006). In a study carried out by Lee S. et al. (2004), T-DNA-tagged transgenic rice lines were evaluated for cold-responsive β-glucuronidase (GUS) expression. At least ∼0.5% of such lines were analysed, and detailed analysis led to the identification of two cold-responsive genes, namely OsDMKT1 (putative demethylmenaquinone methyl transferase) and OsRLK1 (putative LRR-type receptor-like protein kinase).

It is estimated that less than 10% of genes tagged in Arabidopsis and rice are expected to produce a visible phenotype (An et al., 2005b). This may be a result of redundant function, conditional mutation or loss-of-function causing a lethal phenotype. Thus, alternative means of gene tagging, such as the use of traps and activation tagging, are useful in studying gene function. Activation tagging makes use of enhancer elements in the construct which can activate the transcription of genes near the site of insertion. Activation tagging has been widely used for Arabidopsis (Kardailsky et al., 1999; Ito and Meyerowitz, 2000; Lee H. et al., 2000; Weigel et al., 2000; Kirik et al., 2004; Tani et al., 2004) and rice (Jeong et al., 2002). Enhancer (with reporter gene fused to a minimal promoter and expressing only if insertion occurs proximal to an enhancer element), promoter (promoter-less reporter expressing only if insertion occurs in an exon) and gene (reporter gene with one or more splice acceptor sites upstream and expressing only if insertion occurs in an intron) traps are used to tag a gene on the basis of its expression pattern (Springer, 2000). Such tagged populations have been generated in Arabidopsis and rice (Greco et al., 2003; Wu et al., 2003; Alvarado et al., 2004; Ito et al., 2004; Ryu et al., 2004; Sallaud et al., 2004; Zhang et al., 2006).

A retroelement, Tos17, has also been used for large-scale mutagenesis in rice; a total of 47 196 lines were generated through tissue culture. Tos17 insertion lines have been used to study the role of some stress-responsive rice genes, such as OsMT2B, a metallothionein gene involved in the scavenging of reactive oxygen species (ROS) (Wong et al., 2004), and OsTPC1, a putative voltage-gated Ca(2+)-permeable channel involved in the regulation of elicitor-induced hypersensitive cell death (Kurusu et al., 2005).

Tagged lines can be screened by either PCR-based screening or sequencing of the region flanking the insertion site. In the PCR-based approach, PCR is performed using gene- and insert-specific primers (Krysan et al., 1999). In Arabidopsis, DNA pools of 1000–5000 lines are usually used, whereas pools of about 1000 lines have been found to be appropriate for rice (Meissner et al., 1999; Lee S. et al., 2003). An alternative approach involves the sequencing of the region flanking the insertion site. Several databases provide useful information on such flanking sequence tags (FSTs) in Arabidopsis (http://signal.salk.edu/, http://www.tmri.org/, http://www.atidb.org/ and http://www.Arabidopsis.org/) and rice (http://www.tigr.org/, signal.salk.edu/cgi-bin/RiceGE, http://193.51.165.9/projects/FLAGdb++/HTML/data.shtml, http://www.gramene.org/ and http://orygenesdb.cirad.fr) (Droc et al., 2006), and it is possible to select a tagged gene of interest for functional analysis and association with a phenotype.

Transgenics for ectopic expression of abiotic stress response genes

In addition to over-expression and silencing of a target gene, targeted gene inactivation can be performed by homologous recombination using a transgenic approach (Tyagi and Mohanty, 2000; Terada et al., 2002; Zhu, 2002, 2003; An et al., 2005a; Iida and Terada, 2005). However, the application of homologous recombination in rice showed very low efficiency (Hanin and Paszkowski, 2003), and viral-induced gene silencing has not yet been shown to be very successful in rice (Robertson, 2004). Gene inactivation using anti-sense, co-suppression or RNAi strategies is largely based on a single gene approach and, at present, it is difficult to use these methods at the genome-wide level (Parinov and Sundaresan, 2000). A transgenic approach, however, has been widely employed to understand abiotic stress response in plants (Bajaj and Mohanty, 2005). Some examples in which the roles of genes in the stress response have been identified by over-expression or suppression of gene expression in crop plants, using a transgenic approach, are listed in Table 2. A major part of our present understanding of the plant stress-responsive network comes from the functional analysis of Arabidopsis genes in transgenic systems. However, in crop plants, the maximum number of transgenic studies have been performed in tobacco and, to a lesser extent, in rice. Interestingly, the majority of such studies have aimed to decipher the function of genes encoding downstream components (effectors), such as those coding for antiporters, heat-shock proteins, superoxide dismutases (SODs) and LEA proteins, rather than upstream components (regulators), such as those coding for transcription factors and kinases. Recently, the functions of genes representing QTLs for abiotic stress have also been confirmed by employing transgenics (Ren et al., 2005; Xu et al., 2006). Thus, the use of the transgenic system to study a range of genes involved in diverse aspects of the control of the plant adaptive response to abiotic stress would give a clearer picture. In addition, most of the stress-responsive genes have cis-acting conserved elements in their promoter region involved in regulating the stress response. A transgenic approach has also been used to dissect the role of stress-responsive promoters (Haralampidis et al., 2002; Trindade et al., 2003; Yamaguchi-Shinozaki and Shinozaki, 2005). The major cis-acting elements present in several abiotic stress-responsive promoters include the ABA-responsive element (ABRE) from the promoter of ABA-responsive genes (Ingram and Bartels, 1996; Grover et al., 2001), dehydration responsive element (DRE) from the promoter of cold- and drought-inducible genes (Yamaguchi-Shinozaki and Shinozaki, 1994; Thomashow, 1999), anaerobic response element (ARE) from the promoter of genes responsive to low oxygen conditions, such as maize Adh1, LDH1 and PDC1 (Dolferus et al., 1994; Dennis et al., 2000), and heat-shock element (HSE) from the promoter of heat stress-inducible genes (Schoffl et al., 1998).

Table 2.  Impact of expression of plant transgenes on response to abiotic stress in crop plants (selected examples)
Gene (plant/product)Phenotype (approach)Reference
Alfalfa
MnSOD (tobacco Mn superoxide dismutase)Tolerance to freezing stress (o)McKersie et al. (1993)
Alfin1 (alfalfa transcription factor)Salinity tolerance (o)Winicov and Bastola (1999)
FeSOD (tobacco Fe superoxide dismutase)Increased survival during cold stress (o)McKersie et al. (2000)
WXP1 (alfalfa transcription factor)Drought tolerance (o)Zhang et al. (2005)
Barley
ALMT1 (wheat malate transporter 1)Aluminium stress tolerance (o)Delhaize et al. (2004)
Carrot
HSP17.7 (carrot heat-shock protein 17.7)Thermotolerance (o)Malik et al. (1999)
Finger millet
PcSrp (Porteresia coarctata serine-rich protein)Salt tolerance (o)Mahalakshmi et al. (2006)
Maize
NPK1 (tobacco MAPKKK)Drought and freezing stress tolerance (o)Shou et al. (2004)
Rice
HVA1 (barley late embryogenesis abundant protein)Tolerance to water and salt stress (o)Xu et al. (1996)
GPAT (Arabidopsis glycerol 3-phosphate acyltransferase)Chilling tolerance (o)Yokoi et al. (1998)
GS2 (rice glutamine synthetase)Salt tolerance (o)Hoshida et al. (2000)
OsCDPK7 (rice calcium-dependent protein kinase)Cold, drought and salt tolerance (o)Saijo et al. (2000)
AgNHX1 (Atriplex gmelini Na+/H+ antiporter)Salt tolerance (o)Ohta et al. (2002)
Athsp101 (Arabidopsis heat-shock protein 101)Thermotolerance (o)Katiyar-Agarwal et al. (2003)
HvPIP2 (barley plasma membrane aquaporin)Salt sensitivity (o)Katsuhara et al. (2003)
OsMAPK5 (rice mitogen-activated protein kinase 5)Abiotic stress tolerance/sensitivity (o/r)Xiong and Yang (2003)
ABF3 (Arabidopsis ABRE-binding factor 3)Drought tolerance (o)Oh et al. (2005)
DREB1A (Arabidopsis DRE-binding protein 1)Drought and salt tolerance (o)Oh et al. (2005)
MnSOD (pea Mn superoxide dismutase)Drought tolerance (o)Wang et al. (2005)
SNAC1 (rice stress-responsive NAC1)Drought and salt tolerance (o)Hu et al. (2006)
OsDREB1 (rice DRE-binding protein 1)Drought, salt and cold stress tolerance (o)Ito et al. (2006)
SsNHX1 (Suaeda salsa Na+/H+ antiporter)Salt tolerance (o)Zhao et al. (2006)
Soybean
GmHsfA1 (soybean heat-shock transcription factor)Thermotolerance (o)Zhu et al. (2006)
Tobacco
MnSOD (tobacco Mn superoxide dismutase)Resistance to oxidative stressBowler et al. (1991)
GPAT (Cucurbita glycerol 3-phosphate acyltransferase)Less chilling damage (o)Murata et al. (1992)
Cu/ZnSOD (tobacco Cu/Zn superoxide dismutase)Protection from oxidative stress (o)Gupta et al. (1993)
SOD (Arabidopsis superoxide dismutase)Chilling stress tolerance (o)Sen Gupta et al. (1993)
TPS1 (Arabidopsis trehalose 6-phosphate synthase)Drought and salinity stress tolerance (o)Holmstrom et al. (1996)
FeSOD (Arabidopsis thaliana Fe superoxide dismutase)Oxidative stress tolerance (o)Van Camp et al. (1996)
BADH (spinach betaine aldehyde dehydrogenase)Tolerance to salt stress (o)Liang et al. (1997)
APX (tobacco ascorbate peroxidase)Susceptibility to ozone injury (a)Örvar and Ellis (1997)
IMT1 (tobacco myo-inositol O-methyl transferase)Tolerance to drought and salt stress (o)Sheveleva et al. (1997)
TPX2 (tobacco cell wall-associated peroxidase)Osmotic stress toleranceAmaya et al. (1999)
MsFer (alfalfa ferritin)Tolerance to oxidative stress (o)Deak et al. (1999)
GlyI (Brassica juncea glyoxalase I)Tolerance to salt stressVeena et al. (1999)
NPK1 (tobacco MAPKKK)Tolerance to environmental stresses (o)Ezaki et al. (2000)
RIC-Prx (rice peroxiredoxin)Tolerance to oxidative stress (o)Lee K.O. et al. (2000)
FAD7 (Arabidopsis omega-3 fatty acid desaturase)Tolerance to high temperature stressMurakami et al. (2000)
MsALR (tobacco NADPH-dependent aldose/aldehyde reductase)Tolerance to oxidative stress (o)Oberschall et al. (2000)
Bip (soybean molecular chaperone-binding protein)Tolerance/sensitivity to water deficit (o/a)Alvim et al. (2001)
Tsi (tobacco EREBP/AP2-type transcription factor)Tolerance to osmotic stress (o)Park et al. (2001)
Cat2 (maize catalase 2)Tolerance to oxidative stress (o)Polidoros et al. (2001)
fad7 (Arabidopsis omega-3 fatty acid desaturase)Reduced salt and drought tolerance (a)Im et al. (2002)
PvNCED1 (Phaseolis vulgaris 9-cis-epoxycarotenoid dioxygenase)Drought tolerance (o)Qin and Zeevaart (2002)
AhCMO (Atriplex hortensis choline mono-oxygenase)Salt tolerance (o)Shen et al. (2002)
GlyI/GlyII (Brassica juncea glyoxalase I/II)Salinity tolerance (o)Singla-Pareek et al. (2003)
NtC7 (tobacco membrane-located receptor-like protein)Osmotic stress tolerance (o)Tamura et al. (2003)
Gst-cr1 (cotton glutathione S-transferase)Oxidative stressYu et al. (2003)
OBP1 (tobacco osmotin-binding protein 1)Tolerance to salt stress (o)Guo et al. (2004)
TERF1 (tomato ethylene-response-factor 1)Osmotic stress tolerance (o)Huang et al. (2004)
CaERFLP1 (Capsicum ethylene-responsive factor-like)Tolerance to salt stress (o)Lee J.H. et al. (2004)
PINO1 (Porteresia l-myo-inositol-1-phosphate synthase)Salt tolerance (o)Majee et al. (2004)
OSISAP1 (rice A20-AN1-type zinc-finger protein)Drought, cold and salt tolerance (o)Mukhopadhyay et al. (2004)
MT-sHSP (tomato heat-shock protein)Thermotolerance/thermosensitivity (o/a)Sanmiya et al. (2004)
JERF3 (tomato jasmonate and ethylene-responsive factor 3)Salt tolerance (o)Wang et al. (2004)
GhNHX1 (cotton putative vacuolar Na+/H+ antiporter)Salt tolerance (o)Wu et al. (2004)
OsBIHD1 (rice benzothiadiazole activated transcription factor)Sensitivity to abiotic stress (o)Luo et al. (2005)
PDH45 (pea DNA helicase 45)Salt tolerance (o)Sanan-Mishra et al. (2005)
CABPR1 (Capsicum basic pathogenesis-related protein 1)Tolerance to heavy metals (o)Sarowar et al. (2005)
NtHSP70-1 (tobacco heat-shock protein)Drought stress tolerance (o)Cho and Hong (2006)
GmTP55 (soybean antiquitin-like protein)Salt, dehydration and oxidative stress tolerance (o)Rodrigues et al. (2006)
a, anti-sense expression; o, over-expression; r, RNA interference.

In an attempt to deploy RNA-mediated gene silencing for large-scale gene inactivation, a collection of gene-specific tags (GSTs) representing at least 21 500 genes has been created for Arabidopsis, and will be used to create RNAi vectors for functional genomics studies. GST hairpin RNA (hpRNA) expression clones have been confirmed for 8136 different GSTs to date and, side-by-side, a medium-scale Arabidopsis transformation project is underway wherein randomly selected hpRNA clones will be tested for efficacy. In one such preliminary analysis, effective silencing has been reported for three genes coding for vacuolar-type H+-ATPase subunit B3, a component of cellulose synthase and pentatricopeptide repeats (Hilson et al., 2004). In rice, RNAi vectors have been designed by Miki and Shimamoto (2004) for the functional analysis of rice genes. The efficacy of the vectors for silencing gene expression was tested using 11 rice genes and, in each case, silencing was observed in more than 90% of the transgenics analysed. One major drawback of the transgenic approach is that it is not useful for large-scale functional analysis. A possible way of employing transgenics for functional analysis would be the use of PLACs (plant artificial chromosomes) containing large-sized genomic fragments. This would significantly reduce the number of transgenic plants required for functional analysis (Somerville and Somerville, 1999). One such large-insert library has been prepared for Arabidopsis in BIBAC2, a plant transformation-competent binary vector, and represented 11.5 × coverage of the Arabidopsis genome (Chang et al., 2003).

Proteomics to identify targets beyond the gene

The systematic analysis of the entire protein complement or proteome is referred to as ‘proteomics’. Analysis of the proteome provides a direct link of genome sequence with biological activity (Pandey and Mann, 2000; Agrawal and Rakwal, 2006). Analysis of the proteome includes a knowledge of the entire protein repertoire as well as studies on other aspects, such as expression levels, post-translational modifications and interactions, to understand the cellular processes at the protein level (Peck, 2005). The combination of mass spectrometry (MS) with two-dimensional gel electrophoresis (2DGE) in the 1990s proved to be useful for proteome analysis. Matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) and electrospray ionization (ESI) are the two most commonly used MS techniques (Mann and Pandey, 2001; Mann et al., 2001).

Bae et al. (2003) used 2DGE and MALDI-TOF-MS to study the Arabidopsis nuclear proteome and changes in the nuclear proteome in response to cold stress. One hundred and eighty-four protein spots were identified, with the expression of almost 30% of these proteins altered in response to cold stress. These included several proteins previously reported to be involved in stress, including heat-shock proteins, transcription factors (AtMYB2 and OBF4), DNA-binding proteins (DRT102 and Dr1), catalytic enzymes (phosphoglycerate kinase, serine acetyltransferase and glyceraldehyde-3-phosphate dehydrogenase), syntaxin, calmodulin and germin-like proteins. Proteome analysis of rice was performed by Salekdeh et al. (2002) to study changes in response to drought stress. More than 1000 protein spots were identified on 2DGE, with the expression of 42 proteins altered by stress. The identities of 16 of these drought-responsive proteins were established by MS. The rice root proteome was also studied to identify salt stress-responsive proteins. 2DGE identified 54 proteins whose expression changed in response to salt stress. Twelve of the proteins were identified by MS, 50% of which were novel salt-responsive proteins, such as UGPase, Cox6b-1, GS root isozyme, α-NAC, putative splicing factor-like protein and putative ABP (Yan et al., 2005). The study of tobacco leaf apoplast proteome in response to salt stress identified 20 proteins whose expression changed in response to stress. These included several well-known stress-associated proteins, together with chitinases, germin-like protein and lipid transfer proteins (Dani et al., 2005). Proteome analysis was performed to study the effect of cold stress on rice anthers at the young microspore stage. More than 3000 proteins were resolved on 2DGE, 70 of which showed differential expression in response to cold stress. Seven of the 18 proteins identified by MALDI-TOF-MS were observed to be partially degraded, reflecting the effect of cold stress at the young microspore stage (Imin et al., 2004). Similarly, in an analysis of the rice cold stress proteome, proteins from unstressed seedlings were compared with those from seedlings exposed to temperatures of 15, 10 and 5 °C. Of a total of 1700 protein spots separated by 2DGE, 60 proteins were up-regulated with a decrease in temperature. The identities of 41 of these proteins were established by MALDI-TOF-MS or ESI/MS/MS, and these mainly included chaperones, proteases, detoxifying enzymes, and enzymes linked to cell wall biosynthesis, energy pathways and signal transduction. These results emphasize the importance of maintaining protein quality control via chaperones and proteases, together with an increase in cell wall components, during the cold stress response (Cui et al., 2005). In addition, a rice proteome database (http://gene64.dna.affrc.go.jp/RPD) is available which catalogues information from 23 reference maps of 2DGE analysis of proteins from diverse biological samples. The database contains, in total, 13 129 identified proteins and the amino acid sequences of 5092 proteins (Komatsu, 2005).

Protein–protein interactions are at the core of all cellular processes. The yeast two-hybrid method is the most commonly used technique to study protein–protein interactions. The yeast two-hybrid principle is based on the fact that most of the transcriptional activators have two functional domains: the DNA-binding domain, which binds to the promoter sequence, and the activation domain, which activates transcription. The yeast two-hybrid system exploits the fact that the DNA-binding domain of the transcriptional activator cannot activate transcription unless it physically interacts with the activation domain. The application of this technology to study protein–protein interactions was made by Fields and Song (1989).

A large-scale yeast two-hybrid analysis was performed by Cooper et al. (2003) to identify rice proteins involved in stress and development. The baits were made by PCR amplification of conserved domains from cDNA libraries and fused to the DNA-binding domain of the GAL4 transcription factor. Each of the bait constructs was screened for possible interactions with two prey cDNA libraries (these cDNAs were fused to the activation domain of GAL4). cDNAs for one set of prey libraries were prepared from cold-stressed, salt-stressed, drought-stressed, unstressed and ABA-treated rice seedlings, and the second was derived from different developmental stages of rice seed together with callus and panicles. In total, more than 5 million bait–prey interactions were studied, from which interaction domains could be established for 200 rice proteins. One of the proteins identified in the yeast two-hybrid analysis was protein phosphatase type 2A regulatory B subunit. This protein not only interacted with itself, but also with a carboxypeptidase, an orthologue of a wheat translation initiation factor, an inositol phosphatase-like protein (IPP) and a stress-regulated 14-3-3 protein. All of these proteins seem to be involved in the stress response. For instance, IPP is known to negatively regulate the response to both ABA and stress. Further, IPP also interacts with the dehydration-repressible zinc-finger protein. In addition, a large number of other interactions were established for proteins involved in different stress conditions and, on many occasions, this interaction network also overlapped with proteins involved in development, showing the complexity of the stress response (Cooper et al., 2003). A large number of bioinformatics tools are available for plant proteome analysis. These include the Proteins of Arabidopsis thaliana Database (PAT) (http://www.pat.sdsc.edu/), MIPS Arabidopsis thaliana Database (MAtDB) (http://mips.gsf.de/proj/thal/db) and Rice Proteome Database (RPD) (http://gene64.dna.affrc.go.jp/RPD/main_en.html).

Stress response mechanisms

The direct or indirect aims of all functional genomics programs focusing on the abiotic stress response are to define the function of the genes involved and to find candidate gene(s) to improve crop performance. Global expression profiling reveals the initial repertoire of genes involved in the abiotic stress response. Such information can be used to establish the roles of genes in abiotic stress and to identify their mechanisms of action. This step acts as a sieve to identify a smaller subset of genes which can be employed to draw up a final list of candidate genes using a multipronged strategy dependent on several check points, including the association with abiotic stress-related QTLs, mutant phenotypes and alleles from wild relatives (Valliyodan and Nguyen, 2006). For this purpose, databases such as the Gramene database (http://www.gramene.org/) are a valuable resource. This database presently hosts 59 types of trait in the category of abiotic stress, including QTLs for cold tolerance, drought sensitivity, drought tolerance, potassium concentration, root penetration index, salt sensitivity, sodium/potassium ratio, stomatal conductance, submergence tolerance and ultraviolet-B resistance. Ultimately, master regulators of the stress response may be used to impart stress tolerance.

Several components of regulatory systems are activated on the perception of stress. This leads to the production of effector molecules which are directly involved in mitigating stress. Regulatory systems are represented by transcription factors and signal transduction components such as kinases and phosphatases (Yamaguchi-Shinozaki and Shinozaki, 2006). The ABA-independent regulatory cascade is mainly represented by the DREB1 (cold) and DREB2 (salt, dehydration) families of proteins (Liu et al., 1998; Nakashima et al., 2000). Another major abiotic stress signal transduction pathway is the ABA-dependent pathway. This pathway is mainly associated with the bZIP class of transcription factors called ABRE-binding factors, ABFs (Choi et al., 2000; Uno et al., 2000). Several other transcription factors involved in the abiotic stress response have been identified and are being characterized (Yamaguchi-Shinozaki and Shinozaki, 2006).

Expression profiling of mutants and transgenics over-expressing transcription factors can aid in an understanding of their mechanism of action. The analysis of the CBF regulon in Arabidopsis revealed that, of 306 cold-responsive genes, only 70% were part of the CBF regulon which included 15 transcription factors (Fowler and Thomashow, 2002). ZAT12 has been identified as a new cold-responsive regulon which functions as a negative regulator of CBF2 (Vogel et al., 2005). In addition, studies with another cold-responsive regulon, i.e. ICE1, have shown that it acts as a master switch controlling many CBF-dependent and CBF-independent regulons (Lee B.H. et al., 2005). Microarray analysis has also been performed for transgenic Arabidopsis plants constitutively expressing DREB2A, a key transcription factor involved in salt and drought stress regulation. Twenty-one genes were up-regulated by DREB2A over-expression. The DRE core motif was found in 14 of these genes, indicating that they may be direct targets for regulation by DREB2A (Sakuma et al., 2006). Such microarray analyses to decipher abiotic stress regulation have also been performed for signalling components, such as kinases and phosphatases (Umezawa et al., 2004; Osakabe et al., 2005). J. Heard and coworkers from Monsanto have expressed an NF-YB class CCAAT-binding transcription factor from Arabidopsis in maize, and have shown improved drought tolerance of transgenic plants in the field (cited in Valliyodan and Nguyen, 2006). Such studies have shown that a finely tuned system of regulatory components seems to be in place, and further investigations may help in the identification of suitable targets for crop improvement.

A good parallel approach may be to search for candidates in naturally stress-tolerant plants (Xiong and Zhu, 2002). The significance of this approach is highlighted by the availability of several reports on contrasting stress-tolerant relatives of Arabidopsis, tomato, barley and maize (Bohnert et al., 2006). Comparison of Arabidopsis thaliana with Thellungiella halophila has shown that the latter does not seem to have any special set of stress-responsive genes and does not show a major change in its gene expression profile following stress; however, the major difference is in the expression level of stress-responsive genes prior to exposure to stress conditions (Taji et al., 2004; Kant et al., 2006; Wong et al., 2006). Similar studies performed to compare the gene expression patterns in desiccation-tolerant (ice plant) and non-tolerant plants have shown that the difference is in the expression pattern, rather than in the presence or absence of particular genes (Vinocur and Altman, 2005). The comparison of gene expression profiling between contrasting genotypes with respect to stress tolerance can be extended to transcription profiling at the QTL level, and the genes identified at such QTLs may potentially be better candidates for mediating stress tolerance (Salvi and Tuberosa, 2005).

Towards making plants abiotic stress tolerant

It must be kept in mind that the basic task of the identification of key gene(s), whose manipulation will ultimately affect crop performance in response to abiotic stress, is highly complex and difficult to decipher because of the polygenic nature of the abiotic response. In addition, the plant's response to each stress is unique, and thus the response to multiple stresses will also be different. Indeed, global expression profiling of a plant's response to abiotic stress conditions has shown that, although overlap may occur for different abiotic stresses, such as cold, salt, dehydration, heat, high light and mechanical stress, a set of genes unique to each stress response is also seen. Even in the case of ROS, which are known to play a central role in both biotic and abiotic stress conditions, it has been shown that different genes of the ROS network respond differently to different stress treatments (Mittler et al., 2004; Mittler, 2006; Noctor, 2006). However, most of the studies carried out to investigate the performance of plants under abiotic stress conditions have not focused on this aspect, making it an important area of concern, especially as it is known that plants are exposed to multiple environmental stresses in the field. Further, the response to abiotic stress is also developmentally regulated (Munns, 2002; Vinocur and Altman, 2005). For instance, in plant species such as rice, wheat, tomato, barley and corn, salt tolerance increases with an increase in plant age. Moreover, it has been shown that QTLs associated with salt tolerance in the germination stage in barley, tomato and Arabidopsis are different from QTLs associated with the early stage of growth. In transgenic studies on crop plants such as rice, the majority have not evaluated the effect of stress on grain yield. This aspect becomes especially important in plants such as rice, in which stresses such as salinity do not affect the vegetative growth as much as the grain yield (Yamaguchi and Blumwald, 2005).

It is apparent that an understanding of the abiotic stress-responsive network will require a considerable amount of time and resources, but a systematic and concerted effort will ensure that only the most suitable genes are identified for crop improvement. The task can be shortened by integrating the information already available and by avoiding the repetition of effort or branching away from the main focus. The final list of candidate genes and their alleles identified through this approach must be subjected ultimately to field trials to determine their efficacy (Figure 1). Some such field trials, e.g. the evaluation of transgenic wheat plants stably expressing the HVA1 gene for drought tolerance (Bahieldin et al., 2005), have already been performed with a handful of positive results, but there is still a long way to go.

Conclusions

The availability of the complete Arabidopsis and rice genome sequences, together with the collection of several plant ESTs, has understandably shifted the focus from determining the sequences to understanding their function (The Arabidopsis Genome Initiative, 2000; Hirochika et al., 2004; Rensink and Buell, 2004; International Rice Genome Sequencing Project, 2005; Wu et al., 2005; Vij et al., 2006). These sequences will greatly aid the process of determination of the function of the majority of plant genes, whose functions remain to be demonstrated experimentally (Sasaki and Antonio, 2005). Functional genomics employing genome-wide strategies, such as expression genomics, mutant analysis and proteomics, has been widely used to identify stress-responsive genes and to understand the mechanism of stress tolerance. Certain networks involved in the plant stress response have already been proposed, and their extension and refinement in relation to crop plants such as rice will become possible by functional genomics. In addition, small non-coding RNAs have recently been brought into focus as regulators of transcriptional and post-transcriptional gene silencing. A few reports are available in which their function has been correlated with abiotic stress, such as mechanical stress-responsive miRNAs in Populus, phosphate starvation-responsive miRNAs in Arabidopsis and dehydration-, cold-, salt- and ABA-responsive miRNAs in Arabidopsis (Sunkar and Zhu, 2004; Fujii et al., 2005; Lu et al., 2005). In the future, more detailed information will become available on expression analysis, together with a robust system to search for knockout mutants of crop plants and an integrated bioinformatics system to serve as the portal for the available information; this will aid the task of linking sequence to functional information (Tyagi et al., 2004; Sasaki et al., 2005). A large number of resources are now available in the form of well-catalogued and easily accessible databases. The outcome of such effort will not only reveal gene function, but will also identify the effective combination of genes (Denby and Gehring, 2005). Genome-wide strategies have accelerated the deciphering of complex stress-responsive networks, and will also help in the identification of key networks and their associated genes, which may be manipulated through either breeding strategies or genetic engineering. It should also be noted that certain bacterial genes have been found to improve the response of crop plants, such as rice, to abiotic stress (Garg et al., 2002; Mohanty et al., 2002; Su et al., 2006). Pyramiding of such genes will aid in the development of ‘the model stress-tolerant crop’, which can be considered as the ultimate aim of these efforts in terms of application and outreach.

Acknowledgements

The research work in our laboratory is supported by the Department of Biotechnology, Government of India.

Ancillary