Stress-inducible expression of AtDREB1A transcription factor greatly improves drought stress tolerance in transgenic indica rice

The cultivation of rice (Oryza sativa L.), a major food crop, requires ample water (30 % of the fresh water available worldwide), and its productivity is greatly affected by drought, the most significant environmental factor. Much research has focussed on identifying quantitative trait loci, stress-regulated genes and transcription factors that will contribute towards the development of climate-resilient/tolerant crop plants in general and rice in particular. The transcription factor DREB1A, identified from the model plant Arabidopsis thaliana, has been reported to enhance stress tolerance against drought stress. We developed transgenic rice plants with AtDREB1A in the background of indica rice cultivar Samba Mahsuri through Agrobacterium-mediated transformation. The AtDREB1A gene was stably inherited and expressed in T1 and T2 plants and in subsequent generations, as indicated by the results of PCR, Southern blot and RT-PCR analyses. Expression of AtDREB1A was induced by drought stress in transgenic rice lines, which were highly tolerant to severe water deficit stress in both the vegetative and reproductive stages without affecting their morphological or agronomic traits. The physiological studies revealed that the expression of AtDREB1A was associated with an increased accumulation of the osmotic substance proline, maintenance of chlorophyll, increased relative water content and decreased ion leakage under drought stress. Most of the homozygous lines were highly tolerant to drought stress and showed significantly a higher grain yield and spikelet fertility relative to the nontransgenic control plants under both stressed and unstressed conditions. The improvement in drought stress tolerance in combination with agronomic traits is very essential in high premium indica rice cultivars, such as Samba Mahsuri, so that farmers can benefit in times of seasonal droughts and water scarcity. Electronic supplementary material The online version of this article (doi:10.1007/s11248-013-9776-6) contains supplementary material, which is available to authorized users.


Introduction
Drought is the most significant environmental stress on agricultural production worldwide (Cattivelli et al. 2008), and a tremendous effort is being applied to improve crop yields in the face of increasing water scarcity. Rice (Oryza sativa L.) is the most widely Electronic supplementary material The online version of this article (doi:10.1007/s11248-013-9776-6) contains supplementary material, which is available to authorized users. consumed food crop and is grown on 160 million hectares worldwide (FAO 2007). Globally, more than 3 billion people from Asia and other countries depend on rice as their staple food, and by 2025 about 60 % more rice must be produced to meet the needs of the growing population. Drought affects plant growth, yield, membrane integrity, pigment content, osmotic adjustments, water relations and photosynthetic activity (Benjamin and Nielsen 2006). Drought-prone regions and potential agricultural land with no irrigation system in place have been less exploited than those with developed irrigation systems or more reliable rainfall due to difficulties and high costs of developing improved technologies. As a result, rice yields are showing a steady decrease worldwide in unirrigated and drought-prone areas. Therefore, developing drought-tolerant rice varieties and reducing water consumption during rice production is crucial to increased rice yield.
Due to the complex polygenic nature of drought tolerance, attempts to improve this trait through conventional breeding have met with little success. Alternatively, the identification and transfer of genes that confer resistance/tolerance to drought stress through transgenic technology is often projected as one solution for protecting crops against a water stress environment and increasing crop yields worldwide, particularly in less developed areas that are threatened by food scarcity and low crop productivity (Nelson et al. 2007). The transgenic approach involves structurally modifying traits by transferring desired genes from one species to other (Ashraf 2010) without any barrier, and it has been employed to overexpress genes from the model dicotyledonous plant Arabidopsis to many crop plants.
Transcription factors (TFs: activators and repressors) are key regulators of the changes in gene expression and environmental stress responses. They have been proved to be useful for improving plant stress tolerance through inducing the expression of a number of stress-related target genes (Thomashow 2001). Both transcription activators and repressors have been shown to confer drought stress tolerance Sakuma et al. 2006). Most of such TFs have been identified and analysed in Arabidopsis, in which genome-wide microarray analyses have helped to identify several potential target genes (Bray 2004;Denby and Gehring 2005;Shinozaki et al. 2003). The best characterized TF groups are ABA responsive element binding protein1 (AREB1), ABA responsive binding factor 2 (ABF2), dehydrationresponsive binding protein (DREB) genes, MYB genes, bZIP encoding genes and the protein kinases such as receptor like kinase 1, SNF1-related protein kinase 2C or guard cell expressing calcium dependent protein kinases (Choi et al. 2000;Osakabe et al. 2005;Umezawa et al. 2004;Uno et al. 2000). Many TF genes have been used to produce transgenic rice lines with either constitutive or inducible promoters, such as HvCBF4 (Oh et al. 2007), AP37 (Oh et al. 2009), TaSTRG (Zhou et al. 2009), ERF protein TSRF1 (Quan et al. 2010), ERF protein JERF3 , OsDREB2A with the 4ABRC promoter (Cui et al. 2011), OsDREB2A with the rd29A promoter (Mallikarjuna et al. 2011), SbDREB2 (Bihani et al. 2011), OsSD1R1 (Gao et al. 2011), OsDREB1A, OsDREB1B and AtDREB1A (Datta et al. 2012).
The Arabidopsis gene CBF3/DREB1A has been used to improve abiotic stress tolerance in japonica rice (Oryza sativa cv. Nakdong) by constitutive expression (Oh et al. 2005) and in indica rice (O. sativa cv. BR29) by inducible expression (Datta et al. 2012). In the study reported here, we improved the drought tolerance of highly recalcitrant indica rice cultivar Samba Mahsuri (BPT 5204), which has medium slender grains with excellent cooking quality, by introducing the Arabidopsis CBF3/ DREB1A gene driven by the stress-inducible rd29A promoter.

Vector and gene construct
Agrobacterium tumefaciens strain LBA4404 carrying the binary vector pC1200-rd29A-AtDREB1A with a hygromycin-resistant gene (hpt) as plant selection marker and chloramphenicol as bacterial selection marker was used for transformation. The target gene (TF) was derived from Arabidopsis thaliana driven by the stress-inducible rd29A promoter and a nos terminator (Fig. 1a). Initially, the Agrobacterium culture was streaked on petri plates containing solid YEB medium (yeast extract 0.1 %, beef extract 0.5 %, sucrose 0.5 %, peptone 0.5 %, magnesium sulphate heptahydrate 0.05 %, agar-agar 1.5 %) containing 10 mg/l rifampicin, 25 mg/l chloramophenicol and incubated at 28°C for 48 h. Starter cultures (5 ml) were obtained from the streaked plates and incubated in YEB broth supplemented with same antibiotics (5 ll) at 28°C on a shaker at 220 rpm for 24 h.
Plant material used for genetic transformation and selection of callus Matured seeds of the indica rice cultivar Samba Mahsuri (BPT-5204) were used for the induction of embryogenic callus. Sterilized seeds were inoculated on MS (Murashige and Skoog 1962) medium supplemented with 2,4-dichlorophenoxyacetic acid (2,4-D; 2 mg/l) and kinetin (0.5 mg/ml) and incubated at 25°C in the dark for callus induction. Embryogenic calli aged 15-21 days were selected under the microscope and subcultured on fresh MS medium for 5 days. Agrobacterium starter cultures incubated overnight were transferred to fresh YEB liquid medium and incubated at 28°C for 6 h. At 1.0 OD, bacterial cells were harvested by centrifugation at 3,200 rpm for 20 min and re-suspended in MS medium containing 100 lM acetosyringone (MS-AS medium). Pre-incubation of bacterial suspensions was carried out at room temperature for 45 min, and the infection of embryogenic calli was done by vacuum infiltration for 20 min, following which all of the bacterial solution was removed. The calli were then co-cultivated on a sterile filter paper mat in cocultivation medium supplemented with 200 lM AS for 3 days at 25°C, following which the calli were washed three times, 15 min each wash, in sterile distilled water and liquid MS medium containing 250 mg/l cefotaxim and 250 mg/l carbenicillin antibiotics. All washed calli were selected on 50 mg/l hygromycin selection medium for three cycles of 15 days each cycle. Resistant calli were transferred onto regeneration medium containing kinetin (2 mg/l) and 1-naphthaleneacetic acid (NAA; 1 mg/l). Rooting of shoots/plantlets was done in half-strength MS medium in culture tubes. All of the plantlets thus derived were hardened in Yoshida's culture solution and transplanted in earthen pots under biosafety glasshouse conditions. The protocols for callus induction, transformation, selection of transformed calli and regeneration of plantlets were as described in the DRR Laboratory Manual (Balachandran 2011).

PCR analysis of putative transgenic plants
The molecular characterization of putative transgenic plants involved PCR and Southern blot analyses. For PCR, genomic DNA was isolated from a small piece of the leaves collected from putative transgenic plants at the hardening stage in Yoshida's solution by using the CTAB method as described in the (DRR Laboratory Manual). A 10-ll PCR reaction volume was prepared with 50 ng genomic DNA isolated from leaves, 2.5 mM of dNTP, 1 U Taq DNA polymerase, 2 pM of primers and buffer with 1.5 mM MgCl 2 . The PCR reaction was set up with an initial denaturation at 94°C for 5 min, followed by 35 cycles with denaturation at 94°C for 30 s, annealing at 60°C for 30 s, elongation at 72°C for 1 min and a final extension step at 72°C for 7 min. The amplified product was loaded onto a 1.2 % agarose gel and checked for amplified fragments of 524 bp. The primer sequences used were 5 0 AAGAAGTTTCGTGAGACTCG3 0 for the forward primer and 5 0 CTCCATAACGATACGTCGTC3 0 for the reverse primer.

Southern blot analysis
Southern blot analysis was carried out by isolating genomic DNA from 2 g of leaf tissue from both transgenic and nontransgenic control plants. About 20 lg of genomic DNA was digested with BstX1 and HindIII for 16 h at 37°C. Restricted samples were loaded onto 0.8 % agarose gel and subjected to electrophoresis overnight at 20 V. Restricted fragments in the gel were depurinated (0.025 M HCl) for 15 min, followed by denaturation (0.4 N NaOH) for 30 min and then transferred to nylon membrane by vacuum blotting for 20 min. Membranes were prehybridized for 6 h and then hybridized with probe DNA labelled with 32 P in hybridization solution [0.5 M phosphate buffer pH 7.2, 14 % sodium dodecyl sulfate (SDS), 0.5 mM EDTA, 10 % bovine serum albumin (BSA) and 5 % SSS DNA] for 16 h. About 100 ng (45 ll) of eluted plasmid DNA was denatured and mixed with the beads (GE Healthcare, Little Chalfont, UK) containing dNTPs. A 5-ll aliquote of alpha 32 P was added to 45 ll of bead solution and incubated for 45 min at 37°C. Labelled DNA was denatured and added to the hybridization solution containing the membranes. The next day blots were washed with Solution I (29 SSC, 0.1 % SDS) for 15 min, with Solution II (19 SSC, 0.1 % SDS) for 15 min and with Solution III (0.19 SSC, 0.1 % SDS) for 10 min. Washed blots were wrapped with saran wrap and loaded inside an X-ray film cassette and exposed for 24 h to detect signals. Further, copy number detection was made by restriction digestion of genomic DNA of transgenic and control plants with enzyme BstX1 alone, which cuts at a unique site, followed by the same procedure for blotting and hybridization with probe DNA as described above.

Reverse transcriptase-PCR analysis
Southern-and PCR-positive transgenic plants were subjected to drought stress by withholding water for 7 days to study the expression of the AtDREB1A gene under the control of the stress-inducible promoter rd29A. Total RNA of the plants was extracted from the leaf tissue by using Trizol Reagent (Invitrogen, Carlsbad, CA). Reverse transcriptase (RT)-PCR was carried out using RNase-free DNase I-treated total RNA according to the RT PCR TM kit of Promega (Madison, WI). A 1-lg aliquot of total RNA from each sample was used in a 10-ll RT reaction volume using oligo dT as a primer; the cycling program consisted of 25°C for 5 min, 42°C for 45 min and 72°C for 10 min. A 2-ll sample of the RT reaction volume after the completion of cDNA synthesis was used for amplification of AtDREB1A in a 10-ll PCR reaction volume. The same primers and procedure described above were used for PCR amplification of AtDREB1A cDNA from the transgenic plants, with the exception that 28 cycles were used instead of 35. During drought stress, the RT-PCR analysis was done on T 3 homozygous lines with gene-specific AtDREB1A primers and OsActin primers as internal controls. The Actin primer sequences used were 5 0 -CCTTACCGACAACCT-TATGA-3 0 for the forward primer and 5 0 -AT-GGAGTTGTATGTGGCTTC-3 0 for the reverse primer.

Generation advancement and progeny analysis of transgenic plants
Fifty seeds collected from each of the three T 0 Southern-positive plants were used to raise T 1 plants for segregation studies in the T 1 generation. All T 1 plants were analysed by PCR to determine the inheritance pattern of the AtDREB1A gene. Southern analysis was also carried out for all 40 plants of each T 1 line. All PCR-and Southern-positive T 1 plants were harvested, and 25 seeds of each T 2 plant were sown from 12 lines of BD-33, eight lines of BD-38 and seven lines of BD-45. PCR analysis of T 2 plants of all three events was carried out, and the results confirmed integration of the target gene. The harvested T 3 seeds from T 2 homozygous lines (from all three events) were advanced further and confirmed the gene inheritance and homozygosity by PCR. These homozygous lines were selected for gene expression and evaluation studies against drought stress in the T 3 generation.

Evaluation of transgenic rice for drought tolerance
Evaluation of homozygous (T 3 ) transgenic plants for expression of the targeted trait (drought stress) was carried out in hydroponic, vegetative and reproductive stages. Ten 4-week-old plants from each of five homozygous T 3 lines and nontransformed control plants were subjected to an initial water stress treatment in Yoshida's medium containing 20 % polyethylene glycol (PEG; hydroponics).
Homozygous plants and nontransformed controls were planted at the fifth leaf stage in the same type of pots containing homogenous soil composed of loam with a low level of organic matter; the pots were kept in a biosafety glasshouse. At 20 days post-transplantation, when the plants had nine to ten leaves, the control wild type (WT) and transgenic plants were divided into two groups. One group was maintained under well-watered normal conditions (unstressed arm) and the other group was subjected to drought stress by withdrawing water initially for 14 days inside the biosafety glass house and then sufficiently re-watered for 7 days. The highly tolerant plants (from the vegetative stage) were checked for drought stress tolerance also in the reproductive stage. During the reproductive stage, both transgenic and control plants were subjected to drought stress by withholding water for 14 days followed by re-watering. During the drought stress period, the water content of the soil was kept at about 15 % by restricting the amount of water applied. Agronomic traits, such as spikelet fertility, grain yield and days for maturity, were observed.
For the severe drought stress studies, homozygous transgenic (T 5 ) lines, along with nontransgenic, drought-tolerant checks (Rasi, and WT BPT controls were subjected to water deficit stress for 28 days in the vegetative stage. Similarly, the same lines were also subjected to drought stress for 15 days in the booting and anthesis stages along with nontransgenic controls. Another set of plants were screened in the real-time condition by planting them in soil in the biosafety screen house, and drought stress response was recorded after 30 days.

Physiological studies
All physiological parameters were analysed at four stages, namely, before stress, 7 and 14 days of stress, and after re-watering. Relative water content (RWC) was measured by taking the fresh weight, turgid weight and dry weight of transgenic and control leaves. RWC was calculated from the equation of Schonfeld et al. (1988) Sullivan and Ross (1979) and Deshmukh et al. (1991) was followed for determining the percentage of ion leakage for both transgenic and control plants. Fully expanded leaves were collected from top second and third position from the main culm of both transgenic and control plants. Uniformly, 1-cm 2 leaf bits totaling 10 cm 2 in area were immersed in 25 ml distilled water, and the initial electrolyte conductivity (EC a ) was measured. The leaf sample was then kept in the water bath at 50°C for 1 h, and the conductivity (EC b ) was measured. The contents were then incubated at ambient temperature overnight to complete ion leakage from tissues, and the electrolyte conductivity (EC c ) was measured once again. The electrolyte leakage was calculated using the following formula: Chlorophyll content was measured by using a soil and plant analysis device (SPAD) meter (Minolta Model 502; Minolta Co., Osaka, Japan) on leaves of both the transgenic and control plants. Proline content of the plants was estimated by the 3 % sulphosalysilic acid method described by Bates et al. (1973). The amount of the proline content was expressed in micrograms per gram plant FW.
Spikelet fertility and grain yield per plant were used as the major criteria to evaluate the drought resistance performance of each T 3 homozygous transgenic line under biosafety glasshouse conditions. Spikelet fertility and grain yield were calculated by counting the total number of filled grains per panicle against total grains (filled and unfilled).

Transformation
A total of 12,000 embryogenic calli induced on callus induction medium were infected by Agrobacterium strain LBA4404 carrying the binary vector pC1200-rd29A-AtDREB1A (Fig. 1a) in 15 different batches. Transformed calli were selected on media containing 50 mg/l hygromycin for three cycles of 15 days each. As many as 750 putative transgenic plants were regenerated from about the 2,500 resistant calli selected. All of the regenerated plantlets were hardened in Yoshida's culture medium and maintained in transgenic biosafety glass house [Electronic Supplementary Material (ESM) Fig. S1].

Molecular characterization
All of the 750 putative transgenic rice plants were screened by PCR using gene-specific primers; the 524-bp fragment, specific to the AtDREB1A gene, was amplified in 75 of these ( Fig. 1b). These transgenic rice plants were further subjected to Southern hybridization analysis to confirm the integration of the AtDREB1A gene into the genome. Southern analysis clearly indicated the presence of hybridized bands of 2.4 kb from the AtDREB1A gene in three T 0 transgenic plants, namely, BD-33, BD-38 and BD-45, thereby confirming the stable integration of the target gene ( Fig. 1c). No hybridization signal was found in the nontransformed control plants. Determination of the copy number also revealed the presence of different hybridization signals, thereby confirming all three lines (BD-33, BD-38 and BD-45) as transgenic plants; these signals were confirmed to be indicative of singlecopy insertions and independent events (Fig. 1d). The RT-PCR analysis involved the total RNA isolated from the three putative transgenic plants (BD-33, BD-38 and BD-45) and the control plants on days 3, 5 and 7 of the water stress treatment during the dry-down experiment in pots. The results of this analysis showed the amplification of the 524-and 366-bp fragments from the cDNAs of these three putative transgenic plants with the DREB primers ( Fig. 1e) and actin primers, respectively (Fig. 1f), but not in the controls. All three Southern-positive plants (BD-33, BD-38 and BD-45) also showed gene expression as evidenced by RT-PCR (Fig. 1e).

Generation advancement and progeny analysis of transgenic plants
Fifty seeds collected from each of the three T 0 Southern-positive plants were used to raise T 1 plants for segregation studies in the T 1 generation. All of the T 1 plants were analysed by PCR to determine the inheritance pattern of the AtDREB1A gene; 32 of the  Table 1). Southern blot analysis was carried out for the T 1 lines of the event BD-33, revealing that 32 of the 42 plants were positive (Fig. 2b). T 1 transgenic lines of all three events that followed the Mendelian ratio of 3:1 were further advanced to the T 2 generation. All PCR-and Southern-positive T 1 plants were harvested, and 25 seeds each of the three T 2 were sown from 12 lines of BD-33, eight lines of BD-38 and seven lines of BD-45. PCR analysis of the T 2 plants of all three events confirmed the integration of the target gene, showing the presence of the 524-bp fragment among most of the transgenic lines but not in the controls (ESM Fig.  S2a). The Southern blot analysis of BD-33 T 2 plants further confirmed the stable integration of the transgene in T 2 generation (ESM Fig. S2b). In total, 13 c Time course analysis of gene expression of T 3 homozygous lines by RT-PCR with Actin (internal control) and gene-specific (DREB) primers during drought stress. Lanes: M 1-kb DNA ladder, B blank, C nontransformed control plant, 1-7 BD-33-24-4 lines (4-3, 4-4, 4-5, 4-6, 4-9, 4-10, 4-11), 8-14 BD-33-24-5 lines (5-3, 5-4, 5-7, 5-8, 5-9, 5-10, 5-12), US unstressed transgenic plant, P positive control (plasmid) Transgenic Res (2014) 23:421-439  427 homozygous lines [6 lines from BD-33 (ESM Table 2), four lines from BD-38 and three lines from BD-45] were identified in the T 2 generation based upon their PCR analysis and the 1:0 segregation pattern of the AtDREB1A gene. The harvested T 3 seeds from T 2 homozygous lines (from all three events) were advanced further and the gene inheritance and homozygosity confirmed by PCR (ESM Fig.S2c). The selected homozygous lines were analysed for gene expression and in evaluation studies against drought stress in the T 3 generation.
Initial screening for drought tolerance at hydroponic stage Ten 4-week-old plants from each of the five T 3 homozygous lines (BD-45-3-1, BD-45-10-1, BD-33-24-1, BD-33-24-2 and BD-38-33-1) along with the nontransformed controls were subjected to 20 % PEG treatment in Yoshida's culture medium to study the response to drought stress. Of these five lines, BD-33-24-1 and BD-33-24-2 showed good growth without wilting, whereas the control plants showed no growth and wilted badly (ESM Fig. S3). Based on these results, we selected lines from the BD-33-24 series for generation advancement. Thirty T 3 seeds were sown from each of the 15 lines of the BD-33-24 series (from BD-33-24-1 to BD-33-24-15) and transplanted into pots for the dry-down experiment in the biosafety glasshouse.

Evaluation of T 3 homozygous lines against drought stress
Drought tolerance at vegetative stage A set of 50 plants at the tenth leaf stage from five different lines (6, 6, 6, 6 and 6 plants from the BD-33-24-4, BD-33-24-5, BD-33-24-6, BD-33-24-7 and BD-33-24-9 lines, respectively; 20 plants from other than the BD-33-24 series) were subjected to the dry-down treatment consisting of the withholding of water for 14 days; another set of 50 plants were maintained as unstressed. During this experiment, physiological parameters, such as chlorophyll content, free proline content, RWC and ion leakage were analysed in transgenic rice lines and nontransgenic controls (both in stressed and unstressed conditions  Fig. S4).

Confirmation of gene expression under drought stress
Leaf samples from the transgenic rice T 3 lines (BD-33-24-4 and BD-33-24-5, each 7 plants) were collected on days 3, 6, 9, 12 and 14 of the water stress treatment at regular intervals and RT-PCR analysis was carried out.
AtDREB1A gene expression was confirmed by RT-PCR analysis; gene expression was very low on day 3 of the water stress treatment but showed continuous increases thereafter, ultimately reaching the highest level on day 15 of the water stress treatment, as evidenced by the amplicon size of 336 and 524 bp with reference to actin and DREB primers, respectively (Fig. 2c).

Analysis of physiological parameters
Based on the results of the PCR analysis and the hydroponic experiment described above, T 3 segregant lines from the BD-33-24 series were selected for the analysis of all physiological parameters under stressed and unstressed conditions. During the dry-down experiment, the lines for analysis in the physiological assays were further selected based on their morphological features, such as tillering mass, leaf rolling, drying, among others. Since the measurement of all these physiological experiments is a destructive process (more leaves should be used), it was carried out in both transgenic and control plants that had an abundance of tillers and healthy leaves. RWC was measured before and after subjecting the transgenic rice plants and controls to the water stress treatment. Before stress, there were no obvious differences in the leaf RWC between control and transgenic plants, and the RWC was within the range of 85-95 %. After the plants had been subjected to water stress for 7 days, the RWC of the nontransformed control leaves fell quickly with respect to their first reading (before stress) from 81 to 63 %, whereas the RWC of most of the transgenic plants declined very slowly (BD-33-24-4, 69 %; BD-33-24-5, 68 %; BD-33-24-6, 75 %; BD-33-24-7, 69 %; BD-33-24-9, 72 %. After 14 days of drought stress, the RWC of the transgenic plants had declined by just 15-30 % as compared to 42 % in the control plants. Lines BD-33-24-4, BD-33-24-5, BD-33-24-6, BD-33-24-7 and BD-33-24-9 showed very impressive RWC on day 14, with RWC percentages of 55.0, 75.5, 65.0, 53.0 and 63.0, respectively (Fig. 3a). The rapid decline of the RWC (minimum being 47.0 %) was observed in the control plants after 14 days. Five days after re-watering, the RWC of the most of the transgenic plants recovered up to 79-89 %, which was significantly higher than that observed in the control plants (48 %) whose leaves had almost dried out. However, there was no significant change in the RWCs of the transgenic plants and controls in the unstressed condition.
Chlorophyll content is usually measured using a SPAD meter. Transgenic and control plants were measured for their chlorophyll content before and after water stress. Before stress, the SPAD values of the transgenic and nontransgenic control plants were similar, ranging from 35 to 45. After commencement of the drought stress treatment, the chlorophyll values of the control plants showed a very drastic change (30-35) compared to the first reading; in contrast, most of the transgenic plants maintained the same chlorophyll content (40-45) even on day 14 of the stress (Fig. 3c). However, after re-watering for 2 days, transgenic plants showed increased levels of chlorophyll whereas the controls did not.
Proline, which is a common compatible osmoprotectant, accumulates in response to drought stress. During the drought stress treatment, the proline level in both the transgenic and control plants rose continuously. This phenomenon was especially significant after 7 days of stress. After 14 days of stress treatment, transgenic lines BD-33-24-4 (62 lg/g of FW), BD-33-24-5 (63 lg/g of FW) and BD-33-24-6 (60 lg/g of FW) had a twofold higher content of proline, followed by lines BD-33-24-7 (59 lg/g of FW) and BD-33-24-9 (58 lg/g of FW), whereas the control plants showed the lowest level of proline (25 lg/g of FW) (Fig. 3d). There was no difference in the proline contents of unstressed transgenic and control plants.

Discussion
Drought is a major constraint affecting rice production, especially in rainfed areas across Asia and sub-Saharan Africa. India accounts for the largest share (59 %) of the total drought-prone rice area in Asia, and most of these areas are rainfed. As drought tolerance is a very complex characteristic, breeding rice varieties with improved drought resistance has progressed slowly over the past few decades, mainly due to a lack of understanding of the traits and genetic mechanisms that confer adaptation to water deficit in the different environments and to the inability of breeders to select for these traits. A few major quantitative trait loci for drought stress have been identified recently and introgressed into the upland varieties. The development of rice varieties with increased tolerance to drought, by both conventional and molecular breeding methods and by genetic engineering, is an important strategy to meet global food demands in a world with less water. Considerable progress has been made in our understanding of gene expression, transcriptional regulation and signal transduction in plant responses related to drought (Zhu et al. 2010). On the other hand, molecular and genomic analyses have facilitated gene discovery Seki et al. 2001Seki et al. , 2007Tran et al. 2004) and enabled genetic engineering approaches to be used to activate or repress several specific functional or regulatory genes or broad pathways related to drought tolerance in plants (Trujillo et al. 2009). Over/ inducible expression of the genes that regulate the transcription of a number of downstream droughtresponsive genes seems to be a better approach to developing drought-resistant/tolerant transgenic plants when compared to engineering individual functional genes (Bartels and Hussain 2008). Since drought mainly depends on the ecosystem of the upland or lowland rice in terms of its severity and randomness, rice breeders facing major challenges in their attempts to improve drought tolerance in rice varieties due to the diversified rainfed and variable environment conditions across the world. In our study, we attempted to improve drought stress tolerance of the indica rice variety Samba Mahsuri (BPT 5204), which is a premium variety having excellent cooking quality and high marketability. We used the TF AtDREB1A from Arabidopsis to transform this variety, which binds to dehydration responsive element (DRE/CRT), a cis-acting promoter element which plays an important role in regulating its downstream target gene expression in response to stress in an abscisic acid (ABA)-independent pathway.
The successful transformation and subsequent regeneration of transgenic rice plants using Agrobacterium-mediated methods are heavily dependent on many factors, including the type of explants, nutritional supplements, culture conditions prior to and during inoculation, composition of the medium used, duration of co-cultivation, virulence of Agrobacterium strain, concentration and composition of the antibacterial agent used, washing of callus, duration of selection and concentration of plant selection agents, conditions of tissue culture, including a robust system of plant regeneration; all are of critical importance (Mohanty et al. 2002;Yookongkaew et al. 2007). In rice, a considerable difference exists between indica and japonica types in terms of response to callus induction, plant regeneration and transformation (Hiei et al. 1997). Tissue culture and genetic transformation in japonica rice has been well studied when compared to indica types (Lin and Qifa 2005;Yookongkaew et al. 2007;Yukoh and Toshihiko 2006). The genotypic influence can be overcome by modifying the nutrient medium or transformation conditions, since the same nutrient medium is not ideal for all varieties (Ge et al. 2006). Therefore, we established an efficient transformation and regeneration system for the highly recalcitrant indica rice cultivar Samba Mahsuri (BPT-5204) by modifying the 2,4-D concentration in the callus induction medium and NAA concentration in the regeneration medium. The addition of AS in both the pre-culture and co-cultivation medium has been reported to induce Vir genes and extend host range of some Agrobacterium strains and found to be essential for rice transformation (Saharan et al. 2004). Our observations also confirmed these findings, and the addition of AS in liquid (100 lM) and solid (200 lM) medium was found to be more reliable for transformation with the LBA4404 strain which is a low in virulent strain. Our observations are consistent with a number of previous reports of rice transformation (Kumar et al. 2005). Various researchers have used 50 mg/l hygromycin B for selecting the putative transformants of different indica-type rice genotypes (Kant et al. 2007;Sridevi et al. 2005). In our study, we used the same concentration of hygromycin for selecting the callus for three 15-day cycles. We regenerated many putative transgenic plants and obtained three independent events confirmed by PCR and Southern blot analyses. The confirmed transgenic plants did not show any phenotypic changes, such as growth retardation or stunted growth, and were similar to the WT control plants. Oh et al. (2005) suggested that rice is more tolerant in the evolutionary context to the expression of stress-regulated genes than dicots, including Arabidopsis. However, plants developed with the OsCDPK, Ubi1:CBF3 and Ubi1: ABF3 genes also did not exhibit either growth inhibition or visible phenotypic alterations in rice, despite constitutive expression of the transgenes (Oh et al. 2005;Saijo et al. 2000). Similar results were also observed using the stress-inducible rd29A promoter in transgenic Arabidopsis, tobacco (Kasuga et al. 2004) and rice (Datta et al. 2012). We selected independent singlecopy events (Fig. 1d) and advanced to T 3 , T 4 , and T 5 generations to obtain homozygous lines. The advancement of progenies was based on the results of PCR analysis-not by hygromycin selection-in order to confirm the presence or absence of the transgene by analysing the exact segregation ratio of the integrated transgene AtDREB1A (whether 3:1 or approx. 3:1). Although the expression pattern was similar in the three events  in the T 0 generation, based on the RT-PCR results, their real osmotic tolerance was different in the homozygous T 3 lines under the physiological drought induced by the PEG treatment in the hydroponic experiment (ESM Fig. S3). The BD-33 event T 3 lines performed showed better stress tolerance than the BD-38 and BD-45 lines, thereby confirming the differential expression pattern of the three independent insertional events. The selected BD-33-24 T 3 homozygous lines were subjected to drought stress in two important stages, namely, the vegetative and reproductive stages. To confirm gene expression under drought stress we performed RT-PCR analyses and observed increasing gene expression with increasing severity of the stress. Gene expression was highest on day 15 of the stress treatment, suggesting that the expression of At-DREB1A was under the strict control of the strong inducible rd29A promoter.
From the physiological perspective, leaf chlorophyll content has a relationship with photosynthetic rate. The decrease in chlorophyll SPAD values under drought stress could be considered as a typical symptom of oxidative stress and may be the result of pigment photo-oxidation and chlorophyll degradation (Farooq et al. 2009). We recorded very high chlorophyll SPAD values among all of the transgenic plants, and their leaves were more greenish than those of the control plants, thereby confirming normal photosynthesis in transgenic plants.
The process of osmolyte solute accumulation under drought stress is known as osmotic adjustment, and it strongly depends on the rate of plant-water stress. Among these solutes, proline is the most widely studied because of its considerable importance in stress tolerance. Proline accumulation is the first response of plants exposed to a water-deficit condition in order to reduce injury to cells. It affects protein salvation, maintenance of quaternary structure of complex proteins and membrane integrity under dehydration stress and it lowers the oxidation of lipid membranes or photoinhibition (Demiral and Turkan 2004). We observed that transgenic rice lines accumulated highest levels of proline on day 14 of the stress, whereas the WT control plants had very low levels of proline at this time point. The increased proline levels in the transgenic plants indicate that proline accumulation could be one of the factors responsible for tolerance to drought stress. Lines BD-33-24-4 and BD-33-24-5 had the longest healthy period among the transgenic plants, possibly due to their higher level of proline accumulation. However, although proline accumulation under drought stress could be one of the factors responsible for the tolerance, whether the transgene AtDREB1A positively regulates the native P5CS gene (responsible for proline accumulation) as its downstream gene has yet to be determined.
EC percentage values are the direct indices of ion leakage from stressed plant cells, and it has been reported that the maintenance of their stability and integrity of cell membranes under drought stress is the major component of drought tolerance (Sullivan and Ross 1979). Our study reveals that the cell membrane in most of the transgenic plants was highly intact, i.e., it was stable, compared to that of the controls. RWC is considered to be a measure of plant water status, reflecting the metabolic activity in tissues, and is used as the most meaningful index of water stress tolerance (Nayyar and Gupta 2006). RWC is related to water uptake by the roots as well as water loss by transpiration. During transpiration, stomata closure has been observed as a result of reduced water loss in transgenic plants and is considered to be a characteristic feature of enhanced water stress tolerance (Datta et al. 2012;Pardo 2010). A decrease in the RWC in response to drought stress has been noted in wide variety of plants, as reported by Nayyar and Gupta (2006): when leaves are subjected to drought, leaves exhibit large reductions in RWC and water potential. In our study, RWC percentage values confirmed that the transgenic drought-tolerant lines could retain more water than the controls during the peak stage of the water deficit condition. Moreover, our analysis of gene expression and RWC were done in same leaf; therefore, we can suggest that AtDREB1A strongly regulates the drought-responsive genes in our transgenic lines.
Apart from the physiological parameters, gene expression, as evidenced by RT-PCR analysis, and our phenotypic observations of transgenic plants clearly demonstrate that expression of the TF AtDREB1A under the control of the stress-inducible promoter rd29A substantially elevated drought stress tolerance in transgenic lines in the both vegetative and reproductive stages. A large number of studies have been performed on the characterization and expression of CBF genes from different plant species using Arabidopsis and other model plants, such as rice and tobacco. Many DREB type genes from different sources have been overexpressed in various crop plants to improve drought tolerance, including MbDREB1 in Arabidopsis (Yang et al. 2011), At-DREB1A in Lolium perenne , TsCBF1 in maize , OsDREB1F in Arabidopsis  and OsDREB1B in tobacco (Gutha and Reddy 2008). In rice, various genes of the DREB1 class (from both rice and Arabidopsis), such as OsDREB1A, OsDREB1B, OsDREB1Ci, At-DREB1A, AtDREBIB and AtDREB1C, have been overexpressed under the control of either droughtresponsive promoters, such as rd29A and OsHVA22p, or various constitutive promoters Hsieh et al. 2002;Ito et al. 2006;Oh et al. 2005;Xiao et al. 2009;Yang et al. 2010). The gene AtDREB1A has also been successfully introduced in various crops for improving drought tolerance, such as wheat (Pellegrineschi et al. 2004), tobacco (Kasuga et al. 2004), tall fescue (Zhao et al. 2007), bahiagrass (James et al. 2008), groundnut ), peanut ), maize (Al-Abed et al. 2007) and soya bean (Paiva Rolla et al. 2013). However, the development of DREB-transformed transgenic plants in different rice genotypes with strong tolerance to drought is still in the experimental phase. Drought phenotyping to identify high-performance lines under realistic greenhouse screening is essential due to the restrictions on open field trials involving transgenic material for biosafety issues. Several of the studies conducted on transgenic crops to date have achieved successful transgene expression, but may have given rise to misleading conclusions from an agronomic or physiological perspective (Bhatnagar-Mathur et al. 2008) mainly due to the fact that testing transgenic lines is generally conducted under artificial stress conditions, in small pots and at early seedling growth stages (Yang et al. 2010). In our study, the evaluation of transgenic lines was based on the high survival of transgenic plants after severe water deficit stress under greenhouse conditions. It is clear from the results shown in Fig. 4a that transgenic lines L1 (BD-33-24-4-3-1-2, 4) and L2 (BD-33-24-4-9-3-4, 5) were more healthy after 28 days of severe stress than after 2 days of re-watering, indicating that the drought-inducible expression of the AtDREB1A gene driven by the strong promoter rd29A functions in a drought-tolerant pathway. It has been suggested that this improved survival of the transgenic plants after severe drought conditions could be associated with either the activation of genes related to drought resistance or a more conservative growth pattern in the transgenics compared with controls (Bhatnagar-Mathur et al. 2008;Morran et al. 2011). The activation of genes in transgenic plants may involve the stressinducible genes, such as LEA/COR/DHN genes, and is yet to be determined.
The improvement of drought tolerance in rice genotypes should be achieved without a parallel limitation of plant growth and yield potential (Cattivelli et al. 2008). In our study, we demonstrated that stress-inducible expression of the AtDREB1A TF in rice improved the drought tolerance in association with grain yield under bipartite conditions, i.e. stressed and unstressed conditions (ESM Table 3). Based on our observations, a number of factors contributed to the improved grain yield of transgenic rice compared to the WT controls under drought conditions. Firstly, the transgenic lines were more tolerant to drought because they had elevated levels of proline compared to the WT controls which provided protection from osmotic damage (Fig. 3d). Secondly, transgenic lines had a higher number of tillers, lower degree of leaf rolling and drying, quicker recovery percentages (Table 1) and larger panicle sizes than controls, all factors which directly contribute to grain yield. Most importantly, the flag leaf is thought to make the greatest contribution to spikelet fertility, thereby increasing grain filling (ESM Table 3), compared with the other leaves of the same plant (Chen et al. 2007;Mahmood et al. 1991). It is reasonable to predict that grain yield can be substantially improved if the photosynthetic capacity of the flag leaves is raised. Transgenic rice had a larger flag leaf area (ESM Fig.  S4) than control plants which significantly increases the photosynthetic rate and net photosynthate level, thus increasing the grain yield. Furthermore, pollen viability of transgenic plants was high when compared to control plants (data not shown). Similarly, transgenic plants under unstressed conditions also showed a significant higher grain yield than control plants due to the contribution of better agronomical characters, such as an increased number of tillers and panicles, larger panicle sizes, increased spikelet fertility (ESM Table 3) and higher pollen viability (data not shown). Since transgenic plants have undergone tissue culture cycle, it is presumed that there could be a positive pleiotropic effect, resulting in the improved agronomic and morphological traits in transgenic plants under unstressed conditions. Datta et al. (2012) reported that a knowledge of the effects of drought treatment in the pre-flowering and post-flowering stages was essential to determine the dehydration effects on flowering and seed development. In our study, the T 5 individual sub-lines of BD-33-24-4-10 were used to separately check for drought tolerance (but not the same plants; ESM Fig. S7) at the booting (L1: BD-33-24-4-10-1-7,8, 9 and L2: BD-33-24-4-10-3-1,2, 3) and anthesis (L3: BD-33-24-4-10-4-1,2, 3 and L4: BD-33-24-4-10-5-2,3, 4) stages. Hence, all of these lines were tolerant at either the booting or anthesis stage up to 14 days, and no sterility was observed in any line, apart from their normal spikelet fertility and grain yields (similar to their parents in the earlier generations T 3 and T 4 ) which were higher in the transgenic plants than in the WT control plants. Spikelet fertility, grain filling and grain yield of the T 5 transgenic plants were also clearly significantly higher than those of the WT control plants in unstressed conditions (data not shown).
Although there has been considerable progress in developing transgenic rice, their effective phenotyping for improving drought tolerance under field conditions remains to be assessed. Most of the researchers evaluated transgenic plants by subjecting them to the stress at the seedling stage for periods ranging from 1 h to few days in small containers; this represents a 'shock' treatment and provides only unrealistic conditions (Babu et al. 2004;Rabbani et al. 2003). Hence, it is important to conduct experiments under real time conditions that will help researchers to closely estimate the drought tolerance in the field in terms of timing, duration and severity (Hu et al. 2006). However, from an agricultural point of view, drought occurs simultaneously with other abiotic stresses, and the evaluation of transgenic lines in the field or under field-like conditions is essential (Deikman et al. 2011;Peleg et al. 2011;Varshney et al. 2011). In another experiment, we imposed severe water stress alone (no other abiotic stresses) on transgenic plants continuously for 4 weeks (28 days) in pots and as well as in the screenhouse under biosafety real time conditions (Fig. 4b) which approximately resembled the stress condition in the field. Transgenic plants showed greater drought tolerance and a more rapid recovery upon re-watering when compared with nontransgenic and WT controls such as Rasi, Vandana, N-22 and BPT5204. Drought phenotyping of transgenic lines in pots and under real time conditions resulted in the selection of the best events in terms of tolerance among the same individual lines. However, as reported earlier by Datta et al. (2012), variation in level of tolerance among the same individual lines was observed, possibly due to the differential regulation of downstream genes by the AtDREB1A TF in each individual plant, which is to be confirmed by the microarray analysis.
The improvement of drought stress tolerance is a major goal of genetic engineering in rice, and our results have confirmed that stress-inducible expression of the AtDREB1A gene in transgenic plants greatly enhanced drought tolerance at both the vegetative and reproductive stages without jeopardizing the physiological responses of the plants. The Arabidopsis gene CBF3/DREB1A functioned in the stress tolerance pathway and substantially enhanced drought stress tolerance in highly susceptible indica rice cultivar Samba Mahsuri without affecting its agronomic traits. Since this variety is a premium variety with high marketability-but susceptible to drought stress-the selected transgenic lines would be a valuable resource and can be used as a pre-breeding and direct variety for the improvement of rice under water stress conditions.