Natural Variation in OsLG3 Increases Drought Tolerance in Rice by Inducing ROS Scavenging^,

Natural variation in the promoter of the transcription factor gene OsLG3 is associated with drought tolerance, and OsLG3 positively regulates drought stress tolerance by inducing reactive oxygen species scavenging.

Abstract

Improving the performance of rice (Oryza sativa) under drought stress has the potential to significantly affect rice productivity. Here, we report that the ERF family transcription factor OsLG3 positively regulates drought tolerance in rice. In our previous work, we found that OsLG3 has a positive effect on rice grain length without affecting grain quality. In this study, we found that OsLG3 was more strongly expressed in upland rice than in lowland rice under drought stress conditions. By performing candidate gene association analysis, we found that natural variation in the promoter of OsLG3 is associated with tolerance to osmotic stress in germinating rice seeds. Overexpression of OsLG3 significantly improved the tolerance of rice plants to simulated drought, whereas suppression of OsLG3 resulted in greater susceptibility. Phylogenetic analysis indicated that the tolerant allele of OsLG3 may improve drought tolerance in cultivated japonica rice. Introgression lines and complementation transgenic lines containing the elite allele of OsLG3^IRAT109 showed increased drought tolerance, demonstrating that natural variation in OsLG3 contributes to drought tolerance in rice. Further investigation suggested that OsLG3 plays a positive role in drought stress tolerance in rice by inducing reactive oxygen species scavenging. Collectively, our findings reveal that natural variation in OsLG3 contributes to rice drought tolerance and that the elite allele of OsLG3 is a promising genetic resource for the development of drought-tolerant rice varieties.

More than 50% of average agricultural yield losses worldwide occur due to abiotic stress, especially drought (Qin et al., 2011; You et al., 2014). Rice (Oryza sativa) requires high quantities of water during growth, which results in a number of production challenges in case of water shortage and insufficient rainfall during the rice-growing season (Mohanty et al., 2013). In the face of these challenges, enhanced performance of rice under drought stress has the potential to significantly improve rice productivity.

Plants have a set of mechanisms to minimize the harmful effects of drought stress (Hirayama and Shinozaki, 2010; Hu and Xiong, 2014). The generation of reactive oxygen species (ROS) is a key process in plant responsiveness to drought stress (Miller et al., 2010). ROS such as singlet oxygen, hydrogen peroxide (H₂O₂), and superoxide anion act as important signal transduction molecules and also toxic by-products of stress metabolism, depending on their overall cellular amount (Miller et al., 2010). At low to moderate concentrations, ROS may act as secondary messengers in stress signaling pathways, triggering stress-defensive/adaptive responses. However, when ROS levels reach a certain threshold, they can trigger progressive oxidative damage, leading to retarded growth and eventual cell death (Hou et al., 2009). Some studies have indicated that increased expression of ROS scavenging-related genes could increase tolerance to drought stress. For example, overexpression of SNAC3 increases rice drought and heat tolerance by modulating ROS homeostasis through the regulation of the expression of genes involved in ROS scavenging (Fang et al., 2015). Moreover, overexpression of the MEK kinase gene DSM1 in rice increases drought stress tolerance by regulating ROS scavenging. Conversely, suppression of DSM1 results in decreased ROS scavenging and increased drought hypersensitivity (Ning et al., 2010).

ERF proteins have been suggested to play diverse roles in cellular processes involving flower development, floral meristem formation, plant growth, pathogen resistance, and abiotic stress tolerance (Boutilier et al., 2002; Komatsu et al., 2003; Nakano et al., 2006; Yaish et al., 2010; Iwase et al., 2011; Hofmann, 2012; Wang et al., 2012; Zhao et al., 2012; Jin et al., 2013; Jung et al., 2017). There is growing evidence that ERF proteins are involved in plant responses and adapt to drought stress. For instance, transgenic rice lines overexpressing ERF transcription factors (TFs), including SUB1A, OsEREBP1, AP37, AP59, HYR, OsERF71, and OsERF48, all show strong resistance to drought stress (Oh et al., 2009; Fukao et al., 2012; Ambavaram et al., 2014; Jisha et al., 2015; Lee et al., 2016, 2017; Jung et al., 2017; Li et al., 2018). Other ERF genes, including HARDY, TRANSLUCENT GREEN (Zhu et al., 2014), and DREB genes (Liu et al., 1998) from Arabidopsis (Arabidopsis thaliana); TSRF1 (Quan et al., 2010) from tomato (Solanum lycopersicum); TaERF3 (Rong et al., 2014) from wheat (Triticum aestivum); GmERF3 (Zhang et al., 2009) from soybean (Glycine max); and JERF3 (Wu et al., 2008) from tobacco (Nicotiana tabacum), also have been found to be involved in responses to water-deficit stress conditions. Overall, these findings suggest that ERF TFs provide the potential for engineering crops to be more efficient under drought stress conditions.

Linkage disequilibrium (LD)-based association mapping has been proven to be a powerful tool for dissecting complex agronomic traits and identifying alleles that can contribute to crop improvement (Huang et al., 2010, 2011; Setter et al., 2011). Candidate gene association analysis, an effective method to validate targets, has become easier and cheaper with the advances in next-generation sequencing technology, facilitating the discovery and detection of single-nucleotide polymorphisms (SNPs) and identifying alleles that can contribute to crop improvement (Setter et al., 2011; Yang et al., 2014; Mao et al., 2015; Wang et al., 2016; Yu et al., 2017). Some association studies on crop drought tolerance have been performed. For instance, Lu et al. (2010) and Xue et al. (2013) identified some quantitative trait loci underlying drought tolerance in maize (Zea mays) by genome-wide association analysis. Liu et al. (2013) found that DNA polymorphisms in the promoter region of ZmDREB2.7 were associated with maize drought tolerance. Analysis of the association found that an 82-bp insertion in ZmNAC111 and a 366-bp insertion in ZmVPP1 affected drought tolerance in maize (Mao et al., 2015; Wang et al., 2016). Recently, an association study of 136 wild and four cultivated rice accessions identified three coding SNPs and one haplotype in a DREB TF, OsDREB1F, which are potentially associated with drought tolerance (Singh et al., 2015), and nine candidate SNPs were identified by association mapping of the ratio of deep rooting in rice (Lou et al., 2015). However, the role of these candidate genes and their causative variants in improving drought tolerance remains to be confirmed experimentally.

Here, we characterize the role of an ERF family TF, OsLG3 (LOC_Os03g08470), in rice drought tolerance. In our previous work, we demonstrated that OsLG3 has a positive effect on rice grain length without affecting grain quality (Yu et al., 2017). In this study, we identified the nucleotide polymorphisms of OsLG3 that are associated with tolerance to drought stress among different rice accessions. Transgenic plants with OsLG3 overexpression and underexpression demonstrated that an increased expression level of OsLG3 can enhance rice drought tolerance. Introgression lines and complementation transgenic plants containing the elite allele of OsLG3^IRAT109 showed improved drought tolerance, providing evidence that natural variation in OsLG3 contributes to drought tolerance in rice. OsLG3 functions as a pleiotropic gene that contributes to rice grain length and drought stress tolerance together. These data provided insights that the elite allele of OsLG3 is a promising genetic resource for the genetic improvement of rice drought tolerance and yield.

RESULTS

OsLG3 Is Associated with Drought Stress Tolerance in Rice

In our previous cDNA microarray experiment, osmotic stress caused by 15% (w/v) polyethylene glycol (PEG) induced the expression of OsLG3 more strongly in upland rice (UR, IRAT109, and Haogelao) than in lowland rice (LR, Nipponbare, and Yuefu; Wang et al., 2007). In this study, we confirmed that OsLG3 was more highly expressed in IRAT109 than in Nipponbare under well-watered conditions and that the expression of OsLG3 was induced strongly by increasingly severe soil drought stress in IRAT109 but not in Nipponbare (Fig. 1A). Further reverse transcription quantitative PCR (RT-qPCR) analysis of well-watered rice seedlings (4 weeks old) indicated that 10 typical UR accessions (Taitung_upland328, Hongmisandanbai, Cunsanli, Zimangfeienuo, Shanjiugu, Lengshuinuo, Funingzipijingzi, IRAT109, Han502, and Yunlu103) showed significantly higher OsLG3 expression levels than 10 LR accessions (Sansuijin, Baxiang, Wuziluosi215, Gaoliqiu, Nabated A Smar, Qiutianxiaoting, Zhenfu8, Liaojing287, Yuefu, and Nipponbare; Fig. 1B). These results suggested that the change in OsLG3 expression may be related to the drought stress response of upland rice.

Figure 1.

OsLG3 is associated with drought tolerance. A, RT-qPCR analysis of OsLG3 in Nipponbare and IRAT109 under different soil drought stress levels. NS, No stress; SLD, slight drought; MOD, moderate drought; SED, severe drought. Values are means ± se (n = 3). Statistical significance was determined by Student’s t test. The letter a above the bar indicates a significant difference at P < 0.01. B, Relative expression level of OsLG3 in 10 UR and 10 LR accessions under well-watered conditions. Values are means ± se (n = 10). C, Analysis of the association between pairwise LD of DNA polymorphisms in the OsLG3 gene and water-deficit tolerance. A schematic of OsLG3 is shown on the x axis, and the significance of each variation associated with seedling RGR (the ratio of germination rates under the 15% PEG condition to germination rates under well-watered conditions) is shown on the y axis. The SNPs with significant variation (P < 1 × 10⁻²) between genotypes are connected to the pairwise LD diagram with a solid line. Black dots in the pairwise LD diagram highlight the strong LD of SNP_4352886 (closed red circle) and two significant variations: SNP_4352414 and SNP_4352960 (open red circles). SNP_4348903, SNP_4352166, SNP_4352793, SNP_4353076, and SNP_4353119, which are marginally significant (P < 1 × 10⁻²), are denoted by open red triangles.

We conducted a candidate gene association analysis to investigate whether the natural variation in OsLG3 is associated with rice drought tolerance. For this, we used a mini-core collection (MCC) panel (Zhang et al., 2011) of 173 varieties (Supplemental Table S1) that have undergone deep sequencing (http://www.rmbreeding.cn/Index/). A total of 97 SNPs within the OsLG3 locus from these accessions were identified. Seeds from MCC lines were germinated in water or 15% (w/v) PEG. By calculating the relative germination rate (RGR) of each line after 5 d, we found significant variations in tolerance to osmotic stress among different varieties (Supplemental Table S1). Candidate gene association analysis detected three significant SNPs (P < 1.0 × 10⁻³; SNP_4352414, SNP_4352886, and SNP_4352960) located within the promoter region of OsLG3 (Fig. 1C). SNP_4352886, located 2,449 bp upstream from the start codon of OsLG3, showed the greatest significant association with RGR (P = 2.66 × 10⁻⁶; Fig. 1C) and contributed to 13.9% of the phenotypic variation in the MCC population. SNP_4352886 was in strong LD with two other variations (SNP_4352414 and SNP_4352960) in the promoter (r² ≥ 0.8) but not with SNP_4348903, SNP_4352166, SNP_4352793, SNP_4353076, and SNP_4353119, which were identified as marginally significant (P < 1.0 × 10⁻²; Fig. 1C). SNPs identified within the coding region of OsLG3 were not significantly associated with the RGR trait. Based on the above results, we conclude that the nucleotide polymorphisms in the OsLG3 promoter are associated with differential germination rates under water-deficit conditions.

OsLG3 Is an ERF Family Transcription Activator

OsLG3 encodes a putative protein with 334 amino acids. Amino acids 110 to 159 contain a typical AP2 domain, including 11 putative DNA-binding sites, implying strong DNA-binding capacity, and one putative nuclear localization signal from amino acids 95 to 121 (Supplemental Fig. S1, A and B). OsLG3 is located at the same locus as OsERF62 (Nakano et al., 2006) and OsRAF (a Root Abundant Factor gene in rice; Hu et al., 2008). Phylogenetic analysis comparing OsLG3 with known ERF TFs (Nakano et al., 2006) indicated that OsLG3 belongs to group VII of the ERF subfamily (Fig. 2) and is closely related to OsERF71 (Lee et al., 2016, 2017; Li et al., 2018), OsEREBP1 (Jisha et al., 2015), and OsBIERF1 (Cao et al., 2006), which have been reported to be involved in the stress response. Transactivation activity assays with the full-length coding sequence (CDS) or a series of shortened CDS of OsLG3 indicated that the C-terminal region (amino acids 213–334) is required for the transcriptional activation of OsLG3 (Supplemental Fig. S2A). A dimerization test of OsLG3 protein in vivo using a yeast two-hybrid system indicated that OsLG3 potentially functions as a homodimer in rice (Supplemental Fig. S2B). Nicotiana benthamiana leaves infiltrated with Agrobacterium tumefaciens (strain EH105) containing 35S:OsLG3-GFP confirmed that OsLG3 is a nucleus-localized protein (Supplemental Fig. S2C).

Figure 2.

Phylogenetic tree of OsLG3 homologs. Neighbor-joining phylogenetic analysis is shown for the OsLG3 protein sequence in the context of other characterized AP2/ERF proteins from rice. The phylogenetic tree was constructed using the ClustalW and MEGA programs. Tree topology with bootstrap support is based on a percentage of 1,000 replicates. Those numbers on the nodes are bootstrap percentages, indicating the reliability of the cluster descending from that node. The accession numbers are as follows: ARAG1, LOC_Os02g43970; OsAP21, LOC_Os01g10370; OsDREB4-1, LOC_Os02g43940; OsDREB4-2, LOC_Os04g46400; OsDREB2A, LOC_Os01g07120; OsDREB2B, LOC_Os05g27930; OsBIERF3, LOC_Os02g43790; OsERF1, LOC_Os04g46220; OsERF922, LOC_Os01g54890; OsWR1, LOC_Os02g10760; OsWR4, LOC_Os06g08340; SNORKEL1, AB510478; SNORKEL2, AB510479; OsBIERF1, LOC_Os09g26420; OsEBP-89, LOC_Os03g08460; OsEREBP1, LOC_Os02g54160; OsLG3, LOC_Os03g08470; OsERF71, LOC_Os06g09390; Sub1A, DQ011598; Sub1C, LOC_Os09g11480; Sub1B, LOC_Os09g11460; MFS1, LOC_Os05g41760; OsAP2-39, LOC_Os04g52090; OsERF3/OsBIERF4/AP37, LOC_Os01g58420; FZP, LOC_Os07g47330; OsRap2.6, LOC_Os04g32620; OsAP2-1, LOC_Os11g03540; and OsDREB1D, LOC_Os06g06970.

Expression Profile of OsLG3 under Different Stress Treatments and in Different Plant Tissues

The expression profile of OsLG3 in response to abiotic stresses and hormones in IRAT109 was investigated using RT-qPCR. The results indicated that the transcript level of OsLG3 was increased significantly under dehydration, PEG, H₂O₂, NaCl, abscisic acid (ABA), ethylene (ETH), and gibberellic acid (GA) treatments but remained unchanged under cold treatment (Fig. 3A). To determine the spatiotemporal expression pattern of OsLG3 under normal growth conditions, we isolated total RNA in eight representative tissues (root, stem, and sheath at the seedling stage and root, stem, sheath, leaf, and panicle at the heading stage) from IRAT109 and performed RT-qPCR analysis (Fig. 3B). The data indicated that OsLG3 was expressed in all of the tissues tested and showed higher levels in roots compared with other tissues.

Figure 3.

Expression analysis of OsLG3. A, Expression levels of OsLG3 under various abiotic stresses and hormone treatments in IRAT109. Three-week-old seedlings were subjected to dehydration, NaCl (200 mm), PEG6000 (20%, w/v), cold (4°C), H₂O₂ (1 mm), ABA (100 μm), ETH (100 μm), and GA (100 μm) treatments. The relative expression level of OsLG3 was detected by RT-qPCR at the indicated times. Error bars indicate the se based on three replicates. B, Detection of OsLG3 expression in various tissues and organs of IRAT109 using RT-qPCR. Three-week-old seedlings were used to harvest the samples of the root, sheath, and leaf at the seedling stage. Plants in stages before the heading stage were used to harvest the samples of root, stem, sheath, leaf, and panicle at the reproductive growth stage. Error bars indicate the se based on three technical replicates.

Overexpression of OsLG3 Enhances Drought Stress Tolerance in Rice

To elucidate the biological function of OsLG3, transgenic rice plants with OsLG3 overexpressed under the control of the 35S promoter (OsLG3-OE) and suppressed by RNA interference (OsLG3-RNAi) were generated. Two independent OsLG3-OE transgenic lines (OE4 and OE7; Supplemental Fig. S3) and two OsLG3-RNAi lines (RI6 and RI10; Supplemental Fig. S4) of OsLG3 were selected for further analysis. Under dehydration treatment using 20% (w/v) PEG for about 3 d, OsLG3-OE lines showed greater resistance than wild-type plants (Fig. 4A). Almost 44% to 81% of the OsLG3-OE plants survived, while only 8% to 15% of the wild-type plants survived under this treatment (Fig. 4B). In contrast, when 4-week-old wild-type and OsLG3-RNAi plants were subjected to a slightly less severe dehydration stress (20% [w/v] PEG for about 2.5 d), the relative survival of OsLG3-RNAi lines (3%–11%) was lower than that of the wild-type (24%–42%; Fig. 4, C and D). Under severe soil drought stress conditions (no watering for about 7 d), the OsLG3-OE lines showed a better survival rate compared with wild-type plants. Almost 48% to 64% of OsLG3-OE plants survived, whereas only 17% to 28% of wild-type plants survived this treatment (Fig. 4, E and F). In contrast, when wild-type and OsLG3-RNAi plants were subjected to moderate drought conditions (no watering for about 6 d), 36% to 47% of the wild-type plants recovered 10 d after watering was restored, but only 8% to 28% of the RNAi plants recovered (Fig. 4, G and H). Under NaCl treatment (Fig. 5A), OE lines showed significantly lower inhibition of relative shoot growth (Fig. 5B) and less suppression of relative fresh weight (Fig. 5C) than the wild-type and RNAi lines (Supplemental Table S2). Under mannitol treatment (Fig. 5D), the relative shoot growth (Fig. 5E) and relative fresh weight (Fig. 5F) of OE plants were significantly higher than those of the wild-type and RNAi lines (Supplemental Table S3). OsLG3-OE lines showed improved phenotypes under PEG, soil drought, NaCl, and mannitol stress conditions. These findings indicate that OsLG3 may not be involved only in drought tolerance but also widely involved in tolerance of osmotic stress.

Figure 4.

OsLG3 increases rice survival under severe drought stress. A, Physiological dehydration stress tolerance assay with OsLG3-OE plants subjected to 20% PEG for 3 d before being allowed to recover for 7 d. B, Survival rates of transgenic and wild-type (WT) plants tested in A. C, Physiological dehydration stress tolerance assay with OsLG3-RNAi plants subjected to 20% (w/v) PEG for 2.5 d before being allowed to recover for 7 d. D, Survival rates of transgenic and wild-type plants tested in C. E, OsLG3-OE and wild-type plants were subjected to severe drought stress without water for 7 d and then recovered for 10 d. F, Survival rates of transgenic and wild-type plants tested in E. G, OsLG3-RNAi and wild-type plants were subjected to moderate drought stress without wat er for 6 d and then recovered for 10 d. H, Survival rates of transgenic and wild-type plants tested in G. Values are means ± se (n = 3). Statistical significance was determined by Student’s t test. The letter a, b, or c above the bar indicates a significant difference at P < 0.01, P < 0.05, or P ≥ 0.05, respectively. Bars = 50 mm.

Figure 5.

Growth of OsLG3 transgenic plants under high-salinity and osmotic stress conditions. A, Ten-day-old seedlings of wild-type (WT), OsLG3-OE, and RNAi plants grown in one-half-strength Murashige and Skoog (MS) medium containing 150 mmol L⁻¹ NaCl. B and C, The relative shoot length (B) and relative fresh weight (C) of the transgenic and wild-type plants from A were compared. D, Ten -day-old seedlings of wild-type, OsLG3-OE, and RNAi plants grown in one-half-strength MS medium containing 200 mmol L⁻¹ mannitol. E and F, The relative shoot length (E) and relative fresh weight (F) of the transgenic and wild-type plants from D were compared. Values are means ± se (n = 10). Statistical significance was determined by Student’s t test. The letter a, b, or c above the bar indicates a significant difference at P < 0.01, P < 0.05, or P ≥ 0.05, respectively. Bars = 50 mm.

We found that the leaves of OsLG3-OE plants had a slower rate of water loss than those of the wild-type and OsLG3-RNAi lines under dehydration conditions (Supplemental Fig. S5A), indicating a role of OsLG3 in reducing water loss, especially under water-deficit conditions. The expression levels of some characterized stress-response genes, like OsLEA3, OsAP37, SNAC1, RAB16C, RAB21, and OsbZIP73, were monitored in the wild-type, OsLG3-OE, and RNAi lines under fully irrigated and soil drought conditions. As shown in Supplemental Figure S5B, these genes showed significantly higher levels of expression under soil drought stress in OsLG3-OE plants when compared with the wild type and RNAi lines. These results indicated that changes in OsLG3 expression have a significant effect on drought stress tolerance in rice.

Natural Variation in OsLG3 Contributes to japonica Rice Drought Tolerance

To analyze the relationship between OsLG3 haplotypes and drought tolerance, we investigated the phylogenetic relationship of 1,058 deep-sequenced (depth ∼ 14.9×) rice accessions, including 251 UR, 415 LR, and 392 wild rice accessions originating from a wide geographic range (Supplemental Data Set S1). Forty-five SNPs were identified in a search for those with a minor allele frequency > 0.05 and missing rates ≤ 50%. Phylogenetic analysis based on these variations showed that there is a clear differentiation between japonica-UR and japonica-LR rice (Supplemental Fig. S6). To get further insight into the phylogenetic relationship, 10 elite japonica-UR varieties (IRAT109, IAC150/76, IRAT266, Guangkexiangnuo, Shanjiugu, Taitung_upland328, Jaeraeryukdo, Riku aikoku, Padi darawal, and Malandi 2) were chosen as a japonica-UR pool based on their strong drought resistance, and 10 typical japonica-LR varieties (Nipponbare, Yuefu, Guichao2, Koshihikari, Zhonghua11, Early_chongjin, Xiushui115, IR24, Ningjing3, and Xiushui114) were chosen as a japonica-LR pool. Seven SNPs (SNP_4352414, SNP_4352886, SNP_4352960, SNP_4352792, SNP_4352797, SNP_4353103, and SNP_4353347) that differed between the upland rice pool and the lowland rice pool were identified. These SNPs showed a clear phylogenetic distinction between japonica-UR and japonica-LR (Fig. 6A). On the basis of these seven SNPs, we could divide the sequences of the 1,058 cultivated varieties into nine haplotypes, of which there were four main variants: Hap 1, Hap 2, Hap 3, and Hap 4 (Fig. 6B). OsLG3^Nipponbare is representative of Hap 1, which is composed mainly of japonica-LR rice, whereas OsLG3^IRAT109 belongs to Hap 2, which is composed mainly of japonica-UR rice and is the second largest group. Hap 3 is composed mainly of indica rice, and Hap 4 is composed mainly of wild rice. These results indicated that OsLG3 has clear differentiation in japonica rice, which can be divided into japonica-UR and japonica-LR rice, while there is no clear division in indica rice.

Figure 6.

The elite allele of OsLG3 improves drought tolerance in japonica rice. A, Phylogenetic tree of OsLG3 in 1,058 varieties constructed based on seven SNPs (S1–S7). Different colors reflect the different subgroups. The pink and green stripes represent indica and japonica, respectively. Both the purple stripes and red lines represent wild rice accessions. The blue and gold lines indicate upland rice and lowland rice, respectively. B, Haplotype analysis of OsLG3 among 1,058 rice accessions of a worldwide rice collection. S1 to S7 denote SNP_4352414, SNP_4352792, SNP_4352797, SNP_4352886, SNP_4352960, SNP_4353103, and SNP_4353347, respectively. Hap, Haplotype; Japonica_UR, upland rice in japonica; Japonica_LR, lowland rice in japonica; Indica_UR, upland rice in indica; Indica_LR, lowland rice in indica. C and D, Dehydration stress treatment of OsLG3 in Yuefu (receptor parent), IRAT109 (donor parent), and introgression lines (IL342 and IL381). C, Haplotypes of OsLG3 in Yuefu, IL342, IL381, and IRAT109. D, RGR of Yuefu, IL342, IL381, and IRAT109 after germination in 15% (w/v) PEG and water. Values are means ± se (n = 3). Statistical significance was determined by Student’s t test. The letter a or b above the bar indicates a significant difference at P < 0.01 or P < 0.05, respectively. E and F, Complementation test of OsLG3. E, OsLG3-CL and wild-type (WT) plants were subjected to severe drought stress without water for 7 d and then recovered for 10 d. Bars = 50 mm. F, Survival rates of transgenic and wild-type plants tested in E. Values are means ± se (n = 3). G and H, Pedigree of selected rice varieties Zhenghan 9 (G) and Huhan 3 (H). The red star indicates the beneficial allele of OsLG3.

To confirm the genetic effect of different alleles of OsLG3 on rice drought tolerance, two introgression lines (IL342 and IL381) were selected from a cross between IRAT109 (donor parent) and Yuefu (receptor parent). Yuefu carries the same OsLG3 allele as Nipponbare, whereas L342 and IL381 carry the same OsLG3 allele as IRAT109 (Fig. 6C). When germinated on 15% (w/v) PEG, the RGR of Yuefu was 27%, while those of IL342, IL381, and IRAT109 were 35%, 36%, and 46%, respectively (Fig. 6D). In order to verify whether Hap 2 of OsLG3 improves drought tolerance in rice, we performed a genetic complementation test that transferred the 4.739-kb genome fragment of OsLG3 (Supplemental Fig. S7A) from IRAT109 to Nipponbare. We selected three independent complementation lines (CL) with significantly higher expression levels of OsLG3 (Supplemental Fig. S7B) for further drought stress testing. The results suggested that OsLG3-CL plants showed a higher survival rate than wild-type plants under drought stress conditions (no watering for 7 d; Fig. 6E). Almost 75% to 91% of OsLG3-CL plants survived and only 22% to 39% of wild-type plants survived (Fig. 6F). These results supported the hypothesis that Hap 2 of OsLG3 contributes to enhanced rice drought tolerance.

Huhan3 is an elite upland rice because of its high yield and strong water-retention capacity, and it is widely planted in Hubei, China. According to pedigree records, one of the parents of Huhan3 is IRAT109. A resequencing study showed that Huhan3 carries the allele of OsLG3 derived from IRAT109 (Fig. 6G). In addition, another elite upland rice, Zhenghan 9, derived from a hybrid of IRAT109 and Yuefu, also retained the allele of OsLG3^IRAT109 (Fig. 6H). These observations illustrated the success of the technique adopted by breeders of combining the OsLG3^IRAT109 allele with other unidentified drought tolerance-related genes to achieve good drought tolerance.

Analysis of Global Gene Expression Reveals Changes in the Expression of Stress-Related Genes and ROS Scavenging-Related Genes

Digital gene expression (DGE) analysis was performed to analyze global gene expression changes in the OsLG3-OE and RNAi lines. Two hundred twenty-three transcripts in transgenic plants were found to have abundance levels greater than 2-fold those in the wild type (P < 0.05, false discovery rate [FDR] < 0.05; Fig. 7A). In addition, 159 genes were up-regulated in OE plants and down-regulated in RNAi plants, while 64 genes were down-regulated in OE plants and up-regulated in RNAi plants (Fig. 7B). The expression of several of these genes was tested independently by RT-qPCR to validate the DGE results. Six out of eight genes in OE plants (OE versus wild-type > 2) and five out of eight genes in RNAi plants (RNAi versus wild-type < 0.5) showed expression patterns consistent with the DGE results (Supplemental Fig. S8). Gene Ontology (GO) analysis showed that the 223 differentially expressed genes affected by OsLG3 overexpression and suppression were enriched significantly for three GO terms (hypergeometric test, P < 0.01, FDR < 0.05): response to stress (GO: 0006950), response to stimulus (GO: 0050896), and response to abiotic stimulus (GO: 0009628; Fig. 7C; Supplemental Table S3). These results are consistent with our proposed role for OsLG3 in the regulation of drought stress tolerance.

Figure 7.

DGE analysis shows that altering OsLG3 expression in transgenic plants affects the transcription of stress-response genes. A, Scatterplots comparing the transcriptome of OsLG3-OE and RNAi with the wild type (WT). The red and green dots indicate transcripts from OsLG3-OE or RNAi that have signal ratios compared with the wild type of greater than 2 and less than 0.5, respectively. B, Venn diagram showing the number of up-regulated and down-regulated genes affected by the overexpression and suppression of OsLG3. C, Significantly enriched GO terms show representative biological processes of genes differentially expressed in OsLG3-OE and RNAi plants. The color saturation of each box is positively correlated to the enrichment level of the term. Solid and dotted lines represent two and zero enriched terms at both ends connected by the line, respectively. D, Transcript levels of genes related to ROS scavenging in wild-type, OsLG3-OE, and RNAi plants under normal or drought stress conditions (no water for 5 d). Error bars indicate the se based on three technical replicates.

Interestingly, DGE analysis showed that 10 ROS scavenging-related genes (APX1, APX2, APX4, APX6, APX8, CATB, POD1, POD2, SODcc1, and FeSOD) were up-regulated in the OE lines and down-regulated in the RNAi lines. This result indicates a role of OsLG3 in the control of ROS homeostasis. To confirm this possibility, we analyzed the expression levels of 15 genes by RT-qPCR (nine genes identified in the DGE analysis and six other ROS-related genes) in wild-type, OsLG3-OE, and RNAi plants under well-watered and drought conditions. Of the 15 tested genes, 13 genes (except APX3 and POD2) were significantly higher in OE plants than in wild-type and RNAi plants under drought stress conditions (Fig. 7D). Conversely, the expression of APX1, APX2, POX8, POX22.3, and POD1 was significantly lower in the OsLG3-RNAi plants than in wild-type plants under drought stress conditions, while the expression of the remaining nine genes was not significantly different between the wild type and RNAi lines (Fig. 7D).

OsLG3 Participates in H₂O₂ Homeostasis

The potential role of OsLG3 in oxidative stress tolerance was examined further by using two oxidative stress inducers, H₂O₂ and methyl viologen (MV; Suntres, 2002). Germinated wild-type, OsLG3-OE, and RNAi plants were sown on one-half-strength MS medium and one-half-strength MS medium containing 2 µm MV. The application of MV dramatically repressed seedling growth in all plants. OsLG3-OE lines showed less growth inhibition compared with the wild type, whereas RNAi plants showed more severe growth inhibition than the wild type (Fig. 8, A–C; Supplemental Table S4). Two-week-old seedlings were treated with 1 mm H₂O₂ or 3 µm MV for 24 h, followed by 3,3′-diaminobenzidine (DAB) staining to determine the presence of H₂O₂ and nitro blue tetrazolium (NBT) staining to show the presence of superoxide anion. Under control conditions, wild-type and transgenic plants showed similar basal levels of H₂O₂ and superoxide, but DAB and NBT staining were much stronger in wild-type plants than in OsLG3-OE plants under H₂O₂ and MV stress treatments (Fig. 8D). These results indicated that the overexpression of OsLG3 in rice can enhance tolerance to oxidative stress.

Figure 8.

OsLG3 is involved in the oxidative stress response. A, Enhanced tolerance of OsLG3-OE plants and enhanced sensitivity of OsLG3-RNAi plants to oxidative stress caused by MV. B and C, Relative shoot length (B) and relative fresh weight (C) measurements of wild-type (WT), OsLG3-OE, and RNAi seedlings under oxidative stress treatments. Data are means ± se (n = 10). D, DAB and NBT staining of leaves for H₂O₂ in wild-type, OsLG3-OE, and RNAi seedlings under oxidative stress treatments caused by H₂O₂ (100 mm) and MV (30 μm) stress treatments. E, DAB staining of leaves for H₂O₂ from wild-type, OsLG3-OE, and RNAi seedlings under normal conditions and stress treatment (3-week-old seedlings were subjected to dehydration for 6 h and 20% PEG6000 for 24 h). F, H₂O₂ content in leaves from wild-type, OsLG3-OE, and RNAi seedlings under normal conditions and slight drought stress treatment (withholding water for 5 d). FW, Fresh weight. G to I, Relative MDA content (G), SOD activity (H), and POD activity (I) in leaves from wild-type, OsLG3-OE, and RNAi seedlings under normal and drought conditions (no water for 5 d). Data are means ± se (n = 3). Statistical significance was determined by Student’s t test. The letter a, b, or c above the bar indicates a significant difference at P < 0.01, P < 0.05, or P ≥ 0.05, respectively. Bars = 50 mm (A) and 5 mm (D and E).

To investigate whether OsLG3 contributes to rice drought tolerance by regulating ROS homeostasis, we treated wild-type, OsLG3-OE, and RNAi plants with dehydration and 20% (w/v) PEG followed by DAB staining to qualitatively detect H₂O₂ accumulation. As shown in Figure 8E, OsLG3-OE leaves showed less H₂O₂ accumulation than wild-type leaves, and RNAi lines showed more H₂O₂ accumulation than wild-type lines. We also quantitatively detected H₂O₂ accumulation under normal and drought stress conditions (no watering for 5 d). As shown in Figure 8F, the production of H₂O₂ in OsLG3-OE leaves was significantly less than that in wild-type leaves, while the production of H₂O₂ in OsLG3-RNAi plants was significantly greater than that in wild-type plants under drought stress conditions. Malondialdehyde (MDA) production under normal growth conditions was similar in wild-type and all transgenic plants, whereas it was significantly lower in OsLG3-OE compared with wild-type and OsLG3-RNAi plants under drought stress (Fig. 8G). These results demonstrate that overexpression of OsLG3 can reduce the overaccumulation of ROS caused by drought stress.

Four-week-old seedlings were treated with drought stress for 5 d, and the activity of superoxide dismutase (SOD) and peroxidase (POD) was determined. Under normal growth conditions, OsLG3-OE lines have significantly higher SOD activity than wild-type and RNAi plants (Fig. 8H), while POD activity did not appear to be affected significantly in OsLG3-OE or RNAi plants (Fig. 8I). Under drought stress, the activities of POD and SOD were both significantly higher in OE plants and significantly lower in RNAi plants than in wild-type plants (Fig. 8, H and I). These results implied that the function of OsLG3 in drought tolerance may be associated with the enhanced antioxidant response to counteract oxidative stress under drought.

DISCUSSION

Natural Variation in the Promoter of OsLG3 Affects Drought Tolerance in Rice

Although many studies have investigated the transcriptional response to drought stress (Lyu et al., 2014; Cheah et al., 2015; Zhang et al., 2016), how natural sequence variation is associated with phenotypic variations in drought tolerance remains largely unknown. Because of polygenic inheritance, low heritability, and strong genotype-by-environment interactions of drought resistance-related traits, there are few reports on the cloning and identification of drought-resistant genes by association analysis and positive mutant screening methods (Kumar et al., 2014; Lou et al., 2015; Singh et al., 2015). In this study, candidate gene association analyses helped identify nucleotide polymorphisms in the promoter of OsLG3 as significantly associated with the RGR trait (Fig. 1B). Transgenic experiments with OsLG3 overexpression and down-expression further supported the hypothesis that increased expression of OsLG3 enhances rice drought stress tolerance (Fig. 4). As in OsLG3, natural variation in ZmDREB2 (Liu et al., 2013), ZmNAC111 (Mao et al., 2015), and ZmVPP1 (Wang et al., 2016) enhanced maize drought tolerance in a dose-dependent manner. To analyze the relationship between OsLG3 haplotypes and drought tolerance, we investigated the phylogenetic relationship of 1,058 deep-sequenced rice accessions. The variations in OsLG3 in these accessions showed that there is a clear differentiation between japonica-UR and japonica-LR rice but no clear division between indica and wild rice (Fig. 6A). Based on the seven major SNPs between the japonica-LR and japonica-UR pools, we divided the sequences of 1,058 cultivated varieties into nine haplotypes. Among them, there are four major variants: Hap 1 (OsLG3^Nipponbare), composed mainly of japonica-LR rice; Hap 2 (OsLG3^IRAT109), composed mainly of japonica-UR rice; Hap 3, consisting mainly of indica rice, without clear division between UR and LR; and Hap 4, consisting mainly of wild rice. Drought treatment of introgression lines (IL342 and IL381; Fig. 6, C and D) and the complement lines (OsLG3-CL; Fig. 6, E and F) containing the genomic fragment of OsLG3^IRAT109 (Hap 2) showed higher drought tolerance than the wild type, indicating that the Hap 2 of OsLG3 effectively increases drought tolerance in cultivated japonica rice (Fig. 6E). These results demonstrated that Hap 2 is an elite allele and contributes to drought tolerance in rice. Interestingly, our previous work indicated that OsLG3 acts as an important positive regulator of grain length and could improve rice yield (Yu et al., 2017). In fact, the pedigree records in upland rice breeding showed that the tolerant allele of OsLG3 had been incorporated into elite varieties via breeding (Fig. 6, G and H). We suggest that polymerizing the beneficial allele (Hap 2) of OsLG3 with other genes related to high yield and quality may promote elite rice breeding because of its pleiotropic effect on traits.

OsLG3 Was Cloned from Upland Rice and Plays a Positive Role in Rice Drought Tolerance

UR has evolved more enhanced drought resistance than LR, being derived from natural and artificial selection over time under drought conditions. It performs better under drought conditions, with greater water retention ability, larger root volumes, and higher biomass production (Wang et al., 2007; Lenka et al., 2011; Ding et al., 2013; Li et al., 2017). Therefore, UR is highly suitable as research material for studying the mechanisms underlying drought resistance. Our previous work on expression profiles from typical LR (Nipponbare and Yuefu, drought-sensitive japonica rice) and UR (IRAT109 and Haogelao, drought-resistant japonica rice) varieties under osmotic stress conditions using cDNA microarray (Wang et al., 2007) showed that the transcription of OsLG3 can be induced to a greater extent in UR varieties than in LR varieties during drought stress. The expression of OsLG3 in Nipponbare was not induced under soil drought stress, whereas it was induced significantly in IRAT109 under both slight drought and moderate drought conditions (Fig. 1A). By performing candidate gene association analysis, we found that nucleotide polymorphisms in the promoter region of OsLG3 are associated with different levels of water-deficit tolerance among rice varieties at the germination stage (Fig. 1B). All these results indicated that OsLG3 might play a role in the observed drought stress response of upland rice. To assess the effect of OsLG3 on water-deficit stress responses, we tested the growth response of transgenic plants with OsLG3 overexpression and down-expression under simulated drought stresses. OsLG3-OE lines showed higher survival rates under dehydration caused by 20% (w/v) PEG6000 and soil drought stress caused by lack of watering, while plants with reduced OsLG3 expression showed reduced survival rates (Fig. 4). Collectively, these results demonstrated that OsLG3 is a positive regulator of the drought stress response in rice. OsLG3-OE plants also showed enhanced growth compared with the wild-type and RNAi lines under mannitol and NaCl treatments, which suggested that OsLG3 may be involved in a cross-talk pathway between water-deficit, osmotic, and high-salinity stress response pathways.

OsLG3 Enhances Drought Stress Tolerance by Inducing ROS Scavenging

Under drought stress, plants rapidly perceive changes in the environment and initiate a series of intercellular and intracellular signal transduction pathways to deal with the stress (Hirayama and Shinozaki, 2010; Hu and Xiong, 2014). The transcriptional aspect of drought stress responses is the signaling mechanisms, which direct changes in gene expression to coordinate stress tolerance and acclimation responses (Selvaraj et al., 2017). We found that the expression levels of a set of stress-related genes like OsLEA3 (Duan and Cai, 2012), OsAP37 (Oh et al., 2009), and SNAC1 (You et al., 2013) were higher in OsLG3-OE lines than in wild-type plants and decreased in OsLG3-RNAi lines before and after drought treatment (Supplemental Fig. S5B). This was confirmed by DGE analysis (Supplemental Table S3). We investigated if the altered response to drought was associated with global changes in the expression of stress-related genes before stress application. Analyses indicated that many stress-related genes were up-regulated in OE plants and down-regulated in RNAi plants (Fig. 7C), which was consistent with the observed phenotypes of OsLG3-OE and RNAi lines. Interestingly, the expression levels of some ROS scavenging-related genes showed increased abundance in OE and decreased abundance in RNAi lines as well (Fig. 7D). For instance, the ascorbate peroxidase gene OsAPX1 plays a positive role in chilling tolerance by enhancing H₂O₂ scavenging (Sato et al., 2011). DSM1 mediates drought resistance through ROS scavenging in rice (Ning et al., 2010). OsCATB prevents excessive accumulation of H₂O₂ under water stress (Ye et al., 2011). Thus, these DGE analyses are consistent with our hypothesis that the function of OsLG3 in abiotic stress tolerance occurs through the transcriptional regulation of stress-related and ROS scavenging-related genes.

Overaccumulation of ROS is a frequent event in plants subjected to diverse abiotic stresses, including drought, high salinity, and extreme temperatures, and can cause damage to plants (Miller et al., 2010). Numerous studies have demonstrated that the plant response to abiotic stresses occurs through the regulation of ROS metabolism (Ning et al., 2010; Lee et al., 2012; Wu et al., 2012; Schmidt et al., 2013; Fang et al., 2015). For example, overexpression of a NAC protein, SNAC3, increases drought and heat tolerance by modulating ROS homeostasis through regulation of the expression of genes encoding ROS-scavenging and ROS production enzymes (Fang et al., 2015). Overexpression of an ERF protein, SERF1, improves salinity tolerance mainly through the regulation of ROS-dependent signaling during the initial phase of salt stress in rice (Schmidt et al., 2013). In this study, OsLG3-OE seedlings exhibited better growth under oxidative stress caused by MV. In contrast, OsLG3-RNAi plants showed enhanced sensitivity to oxidative stress (Fig. 8A). The H₂O₂ and MDA contents that accumulated in the leaves of OsLG3-OE plants were significantly lower than those in the wild-type and OsLG3-RNAi plants (Fig. 8, F and D), suggesting that the improved drought tolerance of OsLG3-OE plants may be due to efficient ROS scavenging and lower levels of MDA, thereby reducing membrane lipid peroxidation. Therefore, the function of OsLG3 in drought tolerance may be associated with the regulation of antioxidation ability.

To scavenge or detoxify excess stress-induced ROS, plants have developed a complex antioxidant system comprising nonenzymatic as well as enzymatic antioxidants (Noctor and Foyer, 1998; Miller et al., 2010). The maintenance of high activity of various antioxidant enzymes, such as POD, SOD, catalase (CAT), peroxidase (POX), and ascorbate peroxidase (APX), to scavenge the toxic ROS has been linked to the increased tolerance of plants to abiotic stresses (Noctor and Foyer, 1998; Mittler, 2002; Miller et al., 2010). Under drought stress, POD and SOD activities were found to be higher in the OsLG3-OE lines than in wild-type plants, and OsLG3-RNAi lines showed the reverse results (Fig. 8, H and I). These data suggested that the activity of ROS-scavenging enzymes is enhanced in OsLG3-OE plants, which contributes significantly to the reduction of ROS accumulation and, thereby, improved drought stress tolerance.

In conclusion, here we present evidence that OsLG3 is induced by water-deficit stresses and that its induction is greater in UR IRAT109 than in LR Nipponbare under drought. Nucleotide polymorphisms in the promoter region of OsLG3 are associated with water-deficit tolerance in germinating rice. Transgenic plants overexpressing OsLG3 showed improved growth under drought stress, probably via the induction of ROS scavenging by controlling downstream ROS-related genes. Phylogenetic analysis and a complementation test indicated that the tolerant allele of OsLG3, identified in drought-tolerant japonica rice varieties, could be introduced in rice to improve drought tolerance. The pedigree records in upland rice breeding showed that the tolerant allele of OsLG3 had been incorporated into elite varieties via breeding. Importantly, the tolerant allele of OsLG3 is a promising genetic resource for the development of drought-tolerant and high-yield rice varieties by using traditional breeding approaches or genetic engineering.

MATERIALS AND METHODS

Plant Materials and Stress Treatments

Rice (Oryza sativa) variety IRAT109 was used for RT-qPCR analysis of OsLG3 transcript levels under various stresses and hormone treatments, and Nipponbare was used for all transgenic experiments. For RT-qPCR analysis of the expression level of OsLG3 under soil drought conditions, the seeds of Nipponbare and IRAT109 were sown in flower pots (140 mm diameter × 160 mm deep) with well-mixed soil (forest soil:vermiculite in a ratio of 1:1) and grown in the greenhouse under well-watered conditions at 28°C/26°C and a 12-h-light/12-h-dark photoperiod. Three-week-old plants were subjected to drought treatment with no watering for 0, 5, 6, and 7 d. These were subjected to four levels of stress treatment categorized as no stress, slight drought, moderate drought, and severe drought. To analyze the expression level of OsLG3 under various abiotic stresses and phytohormone treatments, 3-week-old IRAT109 seedlings grown in Hoagland solution (PPFD of 400 µmol m⁻² s⁻¹ and 12 h of light [28°C]/12 h of dark [26°C]) were subjected to different treatments with 20% (w/v) PEG6000, NaCl (200 mm), cold (4°C), H₂O₂ (1 mm), ABA (100 μm), ETH (100 μm), GA (100 μm), and dehydration by exposure to air. Leaf tissue was harvested at 0, 1, 2, 4, 6, 9, 12, and 36 h after PEG, NaCl, cold, H₂O₂, ABA, ETH, and GA treatments and at 0, 1, 2, 3, 4, 5, 6, 7, and 8 h after dehydration treatment.

For dehydration treatment, uniformly germinated seeds of the wild-type, OsLG3-OE, and RNAi lines were transplanted onto 96-well PCR plates with the bottoms removed and hydroponically grown using Hoagland solution at 28°C/26°C (day/night) with a 12-h photoperiod. Three-week-old plants were treated with 20% (w/v) PEG6000 solution for about 3 d and recovered with water for 10 d. Each stress test was repeated three times. For soil drought stress treatment, uniformly germinated seeds of the wild-type and transgenic lines were transplanted to well-mixed soil (forest soil:vermiculite in a ratio of 1:1) and grown for 4 weeks under normal watering conditions. Drought stress treatment was then applied by stopping irrigation for about 7 d. When all leaves had completely rolled, watering was resumed for 10 d. The survival ratio (the number of surviving plants over the total number of treated plants in the pot) of each line was calculated. To evaluate the tolerance of rice seedlings to osmotic, high-salinity, and oxidative stress treatments, 3-d-old wild-type, OsLG3-OE, and RNAi seedlings (10 plants per replicate, three replicates) were transplanted to one-half-strength MS medium or one-half-strength MS medium containing 200 mm mannitol, 150 mm NaCl, or 2 μm MV, respectively. The plants were grown for 7 d under a photoperiod of 12 h of light (28°C)/12 h of dark (26°C), and then the shoot height and fresh weight of all plants were measured.

OsLG3 Gene Association Analysis of Rice Drought Tolerance among 173 Rice Genotypes

One hundred seventy-three cultivated varieties from the mini-core collection of Chinese cultivated rice, including 130 indica and 43 japonica rice varieties (Supplemental Table S1), were selected for the candidate gene association mapping. To perform the analysis, we obtained phenotypic data of RGRs (the ratio of germination rates under stress conditions to germination rates under water conditions) from growth in water and 15% (w/v) PEG6000 treatment. Briefly, 50 seeds of each line were placed in petri dishes (90 mm diameter) lined with filter paper. To each petri dish, 10 mL of water or 15% (w/v) PEG6000 was added as mock treatment or to induce osmotic stress, respectively. All petri dishes were placed in a 28°C greenhouse, and the germination rates were assessed after 5 d. Genotype data for each line were acquired from the 3,000 Rice Genomes Project (Li et al., 2014). The SNP data were filtered out. After filtering, a total of 97 SNPs remained in a 5-kb region surrounding the OsLG3 gene. Association analysis using a general linear model with the population structure (Q matrix) method was conducted using the TASSEL 5.2.28 software. The Q matrix was estimated from the genomic data to control for population structure.

RT-qPCR and DGE Analysis

Total RNA was extracted using RNAiso Plus (Takara) and reverse transcribed using M-MLV reverse transcriptase (TaKaRa) according to the manufacturer’s instructions. RT-qPCR was performed as described previously (Duan et al., 2012; Xiong et al., 2014). DGE profiling analysis was performed using the wild type and transgenic lines OE7 and RI10. Ten-day-old seedlings grown on one-half-strength MS medium were harvested for total RNA extraction as described above. DGE was performed at the Beijing Genomics Institute (http://www.genomics.cn) using Illumina HiSeq2000 sequencing technology. Transcripts with significant differential expression between OsLG3-OE and the wild type or between OsLG3-RNAi and the wild type were identified as those with P < 0.05 using FDR < 0.05 and a fold change cutoff > 2. GO analysis was performed using agriGO (Du et al., 2010; http://bioinfo.cau.edu.cn/agriGO/). Representative differentially expressed genes were confirmed by RT-qPCR. The primers are listed in Supplemental Table S5.

Subcellular Localization and Biochemical Assays in Yeast

The construction process for subcellular localization analysis was as described previously (Yu et al., 2017). The fusion constructs were transformed into Agrobacterium tumefaciens strain EH105 and then infiltrated to 5-week-old Nicotiana benthamiana leaves (Clough and Bent, 1998). After 2 to 3 d of transformation, the fluorescence signal was observed with a confocal microscope (Olympus FV1000).

For the transactivation assay, the full-length and truncated CDSs of OsLG3 were fused to the GAL4-binding domain in pGBKT7 (Invitrogen). The plasmids BD-FL (the full-length coding region of OsLG3, amino acids 1–334), BD-dC1 (amino acids 1–218), BD-dC2 (amino acids 1–167), BD-dC3 (amino acids 1–106), BD-dC4 (amino acids 107–334), BD-dC5 (amino acids 168–334), and BD-dC6 (amino acids 213–334) were constructed. The transactivation assay was performed as described previously (Yu et al., 2017). For the dimerization assay, AD-OsLG3 with the full-length CDS of OsLG3 fused to the GAL4-activating domain in pGADT7 (Invitrogen) was constructed. The constructs AD-OsLG3 and BD-dC2 were cotransformed into the yeast strain AH109. AD and BD empty, AD-OsLG3, and BD empty were cotransformed as negative controls. All transformed cells were screened on selective medium (SD) plates without Trp and Leu (SD/-Trp-Leu). Then, the PCR-verified transformants were transferred to SD medium with 5-bromo-4-chloro-3-indolyl-α-d-galactopyranoside acid (X-α-gal) and without Trp/His/adenine/Leu (SD/-Trp-Leu-Ade-His/X-α-gal) for 3 d.

Plasmid Construction and Rice Transformation

The construction process for overexpression was as described previously (Yu et al., 2017). To construct the RNAi plasmid, a 415-bp cDNA fragment of OsLG3 was amplified from IRAT109 and inserted into the vector pTCK303 as described previously (Wang et al., 2004). For the complementation test, the coding region of OsLG3 as well as 3,051 bp upstream of the transcription start site and 574 bp downstream of the termination site was amplified from IRAT109 to obtain a 4.739-kb fragment, which was inserted into the pMDC83 vector to generate a pOsLG3^IRAT109::OsLG3^IRAT109 expression cassette. The primers are listed in Supplemental Table S5. The cloned constructs were transformed into A. tumefaciens strain EH105 cells and transferred into Nipponbare as described previously (Yu et al., 2017). Transgenic homozygous lines in the T₂ generation were used for further analysis.

Physiological Measurements

The rate of water loss under dehydration conditions was measured as described previously (Xiong et al., 2014). Ten plants of each line were used in each replicate with three replicates for each line. Histochemical assays for ROS accumulation were determined according to a previously described method (Li et al., 2010; Wu et al., 2012). Briefly, the qualitative detection of H₂O₂ accumulation was detected by DAB staining. Excised leaves were treated with DAB staining solution (1 mg mL⁻¹ DAB, pH 3.8) at 28°C for 12 h in the dark. After staining, the leaves were decolorized with acetic acid:ethanol (1:3) for 60 min and rehydrated in 70% (v/v) alcohol for 24 h at 28°C. Each experiment was repeated on at least 10 different plants, and representative images are shown. Superoxide anion radical accumulation was detected by NBT staining as described previously. The leaf samples were excised and immediately placed in 50 mm sodium phosphate buffer (pH 7.5) containing 6 mm NBT at 28°C for 8 h in the dark. The quantitative measurement of H₂O₂ concentrations was performed with an Amplex Red Hydrogen/Peroxidase Assay Kit (Molecular Probes, Invitrogen) according to the manufacturer’s instructions. Briefly, leaf samples from both the well-watered and drought-stressed (without water for 5 d) plants were ground in liquid nitrogen, and 100 mg of ground frozen tissue from each sample was placed in a 2 mL Eppendorf tube and kept frozen. One milliliter of precooled sodium phosphate buffer (20 mm, pH 6.5) was immediately added into the tube and mixed. After centrifugation (10,000g, 4°C, 10 min), the supernatant was used for the assay. Measurements were performed using a 96-well microplate reader (PowerWave XS2; BioTek) at an absorbance of 560 nm. The MDA content was measured as described previously (Xiong et al., 2014).

The activity of antioxidant enzymes, including SOD and POD, was measured following the protocols described previously (Duan et al., 2012). The units of the antioxidant enzyme activities were defined as follows: 1 unit of SOD activity was defined as the quantity of enzyme required to cause 50% inhibition of the photochemical reduction of NBT per minute at 560 nm; 1 unit of POD activity was defined as the amount of enzyme required to cause a 0.01 absorbance increase per minute at 470 nm.

Phylogenetic Analysis

The phylogenetic tree was analyzed by MEGA6 software based on the neighbor-joining method and bootstrap analysis (1,000 replicates). The EvolView online tool (Zhang et al., 2012) was used for visualizing the phylogenetic tree. Multiple sequence alignment was performed with ClustalW. All SNP data were obtained from the rice functional genomics and breeding database (http://www.rmbreeding.cn/Index/).

Statistical Analysis

Significant differences between wild-type and transgenic plants were analyzed by Student’s t test in R. P < 0.05 was considered to indicate statistical significance.

Accession Numbers

The sequence data of this article can be found in the Rice Genome Annotation Project Database and Resource (http://rice.plantbiology.msu.edu) under the following accession numbers: OsLG3 (LOC_Os03g08470), Actin1 (LOC_Os10g36650), OsLEA3 (LOC_Os05g46480), AP37 (LOC_Os01g58420), SNAC1 (LOC_Os03g60080), RAB21 (LOC_Os11g26790), RAB16C (LOC_Os11g26760), OsbZIP23 (LOC_Os02g52780), Apx1 (LOC_Os03g17690), Apx2 (LOC_Os07g49400), Apx3 (LOC_Os04g14680), Apx5 (LOC_Os12g07830), Apx6 (LOC_Os12g07820), Apx8 (LOC_Os02g34810), OsPox8.1 (LOC_Os07g48010), Pox22.3 (LOC_Os07g48020), FeSOD (LOC_Os06g05110), SODcc1 (LOC_Os03g22810), SODcc2 (LOC_Os07g46990), POD1 (LOC_Os01g22370), POD2 (LOC_Os03g22010), CATB (LOC_Os06g51150), DSM1 (LOC_Os02g50970), RAB16D (LOC_Os11g26750), OsNCED4 (LOC_Os07g05940), OsNCED3 (LOC_Os03g44380), OsDhn1 (LOC_Os02g44870), OsSRO1C (LOC_Os03g12820), OsMYB48 (LOC_Os01g74410), OsITPK4 (LOC_Os02g26720), OsLEA3-2 (LOC_Os03g20680), and OsERF48 (LOC_Os08g31580).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Sequence analysis of OsLG3.

• Supplemental Figure S2. Transactivation assay and subcellular localization of OsLG3.

• Supplemental Figure S3. Overexpression of OsLG3.

• Supplemental Figure S4. Suppression of OsLG3 by RNAi.

• Supplemental Figure S5. Water loss from detached leaves and expression of drought stress-related genes in wild-type, OsLG3-OE, and RNAi plants under normal and drought stress conditions.

• Supplemental Figure S6. Phylogenetic analysis of the OsLG3 gene based on 45 SNPs in 1,058 accessions.

• Supplemental Figure S7. Complementation testing of OsLG3.

• Supplemental Figure S8. RT-qPCR validation of genes expressed differentially between wild-type, OsLG3-OE, and RNAi plants.

• Supplemental Table S1. Information on 173 MCC varieties used for association analysis.

• Supplemental Table S2. Growth values of wild-type, OsLG3-OE, and RNAi plants under NaCl and mannitol stress conditions.

• Supplemental Table S3. GO biological process analysis of differentially expressed genes.

• Supplemental Table S4. Growth values of wild-type, OsLG3-OE, and RNAi plants under MV stress condition.

• Supplemental Table S5. Primer sequences used in this study.

• Supplemental Data Set S1. Information of 1,058 rice varieties and the wild rice used for haplotype analysis.

• Supplemental Data Set S2. Genes expressed differentially between OsLG3-OE plants and wild-type plants.

• Supplemental Data Set S3. Genes expressed differentially between OsLG3-RNAi plants and wild-type plants.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• AP37 Gramene: LOC_Os01g58420

• AP37 Araport: LOC_Os01g58420

• APX1 Gramene: LOC_Os03g17690

• APX1 Araport: LOC_Os03g17690

• APX2 Gramene: LOC_Os07g49400

• APX2 Araport: LOC_Os07g49400

• APX3 Gramene: LOC_Os04g14680

• APX3 Araport: LOC_Os04g14680

• APX6 Gramene: LOC_Os12g07820

• APX6 Araport: LOC_Os12g07820

• APX8 Gramene: LOC_Os02g34810

• APX8 Araport: LOC_Os02g34810

• Actin1 Gramene: LOC_Os10g36650

• Actin1 Araport: LOC_Os10g36650

• Apx5 Gramene: LOC_Os12g07830

• Apx5 Araport: LOC_Os12g07830

• FeSOD Gramene: LOC_Os06g05110

• FeSOD Araport: LOC_Os06g05110

• HYR Gramene: LOC_Os03g02650

• HYR Araport: LOC_Os03g02650

• OsAPX1 Gramene: LOC_Os03g17690

• OsAPX1 Araport: LOC_Os03g17690

• OsCATB Gramene: LOC_Os06g51150

• OsCATB Araport: LOC_Os06g51150

• OsDREB1F Gramene: LOC_Os01g73770

• OsDREB1F Araport: LOC_Os01g73770

• OsDhn1 Gramene: LOC_Os02g44870

• OsDhn1 Araport: LOC_Os02g44870

• OsEREBP1 Gramene: LOC_Os02g54160

• OsEREBP1 Araport: LOC_Os02g54160

• OsERF48 Gramene: LOC_Os08g31580

• OsERF48 Araport: LOC_Os08g31580

• OsERF62 Gramene: LOC_Os03g08470

• OsERF62 Araport: LOC_Os03g08470

• OsERF71 Gramene: LOC_Os06g09390

• OsERF71 Araport: LOC_Os06g09390

• OsITPK4 Gramene: LOC_Os02g26720

• OsITPK4 Araport: LOC_Os02g26720

• OsLEA3 Gramene: LOC_Os05g46480

• OsLEA3 Araport: LOC_Os05g46480

• OsLEA3-2 Gramene: LOC_Os03g20680

• OsLEA3-2 Araport: LOC_Os03g20680

• OsLG3 Gramene: LOC_Os03g08470

• OsLG3 Araport: LOC_Os03g08470

• OsMYB48 Gramene: LOC_Os01g74410

• OsMYB48 Araport: LOC_Os01g74410

• OsNCED3 Gramene: LOC_Os03g44380

• OsNCED3 Araport: LOC_Os03g44380

• OsNCED4 Gramene: LOC_Os07g05940

• OsNCED4 Araport: LOC_Os07g05940

• OsPox81 Gramene: LOC_Os07g48010

• OsPox81 Araport: LOC_Os07g48010

• OsRAF Gramene: LOC_Os03g08470

• OsRAF Araport: LOC_Os03g08470

• OsSRO1c Gramene: LOC_Os03g12820

• OsSRO1c Araport: LOC_Os03g12820

• OsbZIP23 Gramene: LOC_Os02g52780

• OsbZIP23 Araport: LOC_Os02g52780

• POX223 Gramene: LOC_Os07g48020

• POX223 Araport: LOC_Os07g48020

• POX8 Gramene: LOC_Os07g48010

• POX8 Araport: LOC_Os07g48010

• RAB16C Gramene: LOC_Os11g26760

• RAB16C Araport: LOC_Os11g26760

• RAB16D Gramene: LOC_Os11g26750

• RAB16D Araport: LOC_Os11g26750

• SNAC3 Gramene: LOC_Os01g09550

• SNAC3 Araport: LOC_Os01g09550

• SODcc1 Gramene: LOC_Os03g22810

• SODcc1 Araport: LOC_Os03g22810

• SODcc2 Gramene: LOC_Os07g46990

• SODcc2 Araport: LOC_Os07g46990

• SUB1A Gramene: DQ011598

• SUB1A Araport: DQ011598