Rice cultivar Zhonghua 11 is a heat sensitive japonica variety and was used as WT throughout the study. The hst1 mutant was isolated from the T₇ generation of ZH 11 T-DNA insertion mutant lines. Rice seeds were sterilized with 10% (v/v) sodium hypochlorite for 15 min before rinsing five times with sterilized water, then soaked in the water at 37°C for 2 d in the dark. After germination in Petri dishes with wet filter paper at 37°C, the most uniformly germinated seeds were transferred to a 96-well plate, from which the bottom was removed. The plate was floated on water for 5 days in a growth chamber with a 13-h light (28°C)/11-h dark (23°C) photoperiod and 65% relative humidity. After 4 days, the seedlings were cultured with Yoshida’s culture solution (Yoshida et al., 1976). For heat-tolerant screening, rice seedlings were subjected to treatment of heat stress at 42°C with a 13h light/11h dark photoperiod.

For heat treatment at the vegetative stage, WT and hst1 mutant seedlings that were grown hydroponically in 4.5 liters of Yoshida’s culture solution were transferred to a chamber and treated for 3 days at 42°C with a 13h light/11h dark photoperiod. The plants were then transferred to the normal temperature for recovery. The daily loss of the volume of Yoshida’s culture solution was very small (<5%) and was replenished with water daily during the heat treatment and recovery period. The heat tolerance phenotypes of the plants were examined, and plants were photographed after different days of recovery at normal temperature. For analysis of the effect of heat stress on reproductive performance, rice plants were moved to a growth chamber one day prior to heading and subjected to heat treatment for 7 days with a 13h light/11h dark photoperiod. The heat treatment (40°C) lasted for 6 hours daily during the light period from 9:00AM to 15:00PM. The plants were grown at 30°C for the remaining 7 hours (3 and 4 hours before and after heat treatment, respectively). Control plants were grown at 28°C during the 13-hour light period. All plants were grown at 23°C during the 11-hour dark period. Plants were grown under normal growth conditions after treatment and their reproductive traits were evaluated after they reached full maturity.

For salt treatment, WT and hst1 mutant seedlings grown in pots under normal conditions were transferred to plastic containers containing Yoshida nutrient solution with 0.6% NaCl for 12 days. After the treatment, the plants were transferred to the normal nutrient solution for recovery.

The hst1 mutant was crossed with IR29, a heat-sensitive indica rice variety. All F1 progeny showed a heat-sensitive phenotype. In the F2 population, heat-sensitive plants and heat tolerant plants segregated in a ratio close to 3 to one. Preliminary mapping with about 300 F2 plants located the hst1 gene on chromosome 3 using PCR-based molecular markers ( Supplemental Table 1 ). Further mapping placed the mutant gene in a region on chromosome 3. The hst1 gene was putatively identified by sequencing the genes in the region and confirmed by complementation with the WT HST1 gene.

For complementation of the hst1 mutant, a 3970-bp DNA fragment of the full-length HST1 genomic sequence was amplified from genomic DNA and cloned into pCAMBIA1301 with Kpn I and Sal I enzymes. Transgenic lines were achieved by co-cultivation of hst1 rice calli with Agrobacterium tumefaciens strain EHA105 containing the HST1 construct (Hiei et al., 1994). Positive T0 transgenic plants were screened using PCR analysis of hpt with genomic DNA from their leaves.

Full-length coding sequences of WT DST and mutant DST^hst1 proteins were amplified by PCR using gene-specific primers and cloned into the expression vector pET-32a (Novagen) and transformed into E. coli strain BL21(DE3). Expression and purification of recombinant proteins were performed as previously described (Lai et al., 2011). EMSA was performed with purified recombinant proteins and ³²P-labeled double-stranded synthetic oligonucleotides as described previously (Lai et al., 2011).

Water loss from the detached leaves of WT plants and hst1 mutants was measured according to the method described by (Tian et al., 2004) with minor modifications. Detached leaves were placed at 24°C with 40% relative humidity and their fresh weights were determined at various time points. Water loss was expressed as the percentage of initial fresh weight at each time point. The measurement of relative water content was performed as described (Ascenzi and Gantt, 1999) with slight modifications. Briefly, eight fully expanded leaves were detached from plants after various periods of heat treatment and immediately determined for the fresh weight. The leaves were then submerged in ddH₂O overnight. Water on the surface of the leaves was removed by quick blotting using filter papers and the turgid weight was determined immediately. The leaves were then dried at 105°C for 4 hours before their dry weight was determined.

For assays of stomatal conductance, 10-day-old plants were grown in a growth chamber. The stomatal conductance was measured using a portable photosynthesis system (LI-6400 LI-COR, Lincoln, USA). For stomatal density measurement, middle sections of fully expanded leaves were sampled from the same position of rice plants of WT and hst1 mutant. A leaf surface imprint method was used. Nail polish was painted on the sample. After the nail polish was dry, the smear layer was obtained and spread onto the glycerin and observed under a light microscope as previously described (Berger and Altmann, 2000).

The peroxidase activity was measured using a peroxidase assay kit from Nanjing Jiancheng Bioengineering Institute (Nanjing, China) as described previously (Li et al., 2013a). H₂O₂ content was measured using an Amplex^® Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen) following the vendor’s instructions. For sample preparation, 0.5g leaf segments of 10-day-old seedlings were ground in liquid nitrogen and thoroughly mixed with 10 volumes of 50 mM Na₃PO₄ (pH 7.4). After centrifugation at 12000 rpm for 20 minutes at 4 °C, the supernatants were used for the H₂O₂ assay. The levels of MDA were determined based on the spectrophotometer measurement of the product from the reaction of MDA and thiobarbituric acid as previously described (Li et al., 2013a).

Total RNA was isolated using Trizol reagent (Invitrogen) from WT and hst1 mutant seedlings and flowers. Total RNA was treated with 5 U of RNase-free DNase I (TaKaRa) to remove any DNA contamination. DNase I-treated RNA was reverse transcribed using an oligo(dT) primer and a PrimeScript RT reagent kit (TaKara, Japan) to generate cDNA. Real-time PCR was carried out using SYBR Premix Ex Taq (TaKara, Japan) for detection of PCR products (parameters: 95°C for 1 min, followed by 40 cycles of 95°C for 15 s, 60°C for 15 s, and 72°C for 15 s). All reactions were performed in triplicate. Quantification of gene expression was done using the comparative Ct method. The gene encoding ACTIN (LOC_Os03g50885) was chosen as a reference gene. The primers used for qRT-PCR were shown in Supplemental Table 2 .

Total proteins from rice leaf tissues were extracted with P-PER^® Plant Protein Extraction Kit (Pierce, Cat # 89803). Proteins were fractionated by SDS-PAGE and the HSP70 protein levels were detected by western blotting for detection using an anti-HSP70 antibody (Abcam, Cat # ab5439). The Rubisco large subunit, which was detected with a Rubisco-specific antibody (Agrisera, Cat # AS07 259), was used an loading control. The antigen-antibody complexes were detected by enhanced chemiluminescence using luminal as substrate. The western blot films were scanned, and the intensities of protein bands were quantified using the Image J software. The quantification reflected the relative amounts as a ratio of each HSP protein band to the lane’s Rubisco loading control.

Pollen viability was tested with the peroxidase reaction method as described previously (Wang et al., 2012). Briefly, pollens were collected onto a glass slide and stained with two reagents (reagent I: 0.5% benzidine, 0.5% α-naphthol and 0.25% sodium carbonate at 1:1:1 ratio and reagent II: 0.3% hydrogen peroxide). After incubation at 30°C for 10 minutes, the pollen viability was examined under a microscope with viable pollens stained blue and unviable pollens stained colorless or yellowish.

To gain important insights into the genes and their network in controlling heat stress responses in rice, we performed a large-scale screen of the rice ZH11 T-DNA insertion lines for mutants with altered levels of wilting immediately after heat stress, or survival after recovery at normal temperature. We isolated several mutants that showed increased heat tolerance and retesting of these putative heat-stress-tolerant mutants confirmed one of them to be highly heat-tolerant in the subsequent generations and was named heat-stress tolerant 1 (hst1). After heat treatment at 42°C for 3 days (13 h light/11h dark photoperiod), the hst1 mutant seedlings showed less severe wilting than wild-type (WT) plants (data not shown). After 12 days of recovery at 25°C following the heat treatment, approximately 44% of the hst1 mutant seedlings survived but almost all of WT seedlings died ( Figure 1 ). Based on the wilting symptoms and survival rates after heat treatment, we concluded that the hst1 mutant had substantially improved heat tolerance.

Since the hst1 mutant was isolated from a rice T-DNA insertion population, we first determined whether the mutation was caused by a T-DNA insertion. The T-DNA insertion in the binary vector contains a hygromycin phosphotransferase (hpt) gene as a selection marker that confers resistance to the antibiotic hygromycin. Therefore, we first determined resistance of hst1 to hygromycin and, surprisingly, discovered that the mutant was as sensitive to the antibiotic as WT plants. To ensure that the hygromycin sensitivity was not due to silencing of the selection marker, we also performed PCR using hpt-specific primers but failed to amplify the antibiotic resistance gene from the mutant. Therefore, the hst1 mutant plant is unlikely to be a T-DNA insertion line. To further determine the genetic nature of the hst1 mutant, we crossed the mutant with WT plants and found hst1 to be a recessive mutation based on phenotyping of their F1 and F2 progeny. These results collectively suggested that the hst1 mutant resulted from a loss-of-function mutation rather than by a T-DNA insertion.

We then took a map-based cloning approach to isolate the HST1 gene. We first crossed the hst1 mutant (as female) with a heat sensitive Indica rice variety IR29 (as male parent). All F1 progeny showed a WT heat-sensitive phenotype based on the survival and recovered growth after heat treatment. In the F2 population, the ratio of the heat-sensitive plants to the heat-tolerant plants was close to 3 (220 heat-sensitive vs. 80 heat-tolerant; x^{2 =} 0.444<x² _0.5 ⁼ 0.455; P>0.5), indicating that the hst1 mutant was caused by a single recessive nuclear gene mutation. The HST1 locus was fine-mapped to a region of approximately 80 kb on chromosome 3 ( Supplemental Figure 1 ). The genes within the region were sequenced, and it was discovered that one of them (LOC_Os03g57240) contains a four-nucleotide insertion (TGGG) between the 194^th and 195^th nucleotides of its coding sequence in the hst1 mutant ( Figure 2A ). The insertion would lead to a frameshift in translation after the first 64 amino acid residues ( Figure 2 ). To confirm that LOC_Os03g57240 is HST1, we performed a complementation test. A 3966-nucleotide WT LOC_Os03g57240 genomic fragment containing both its promoter and coding region was amplified and cloned into the binary vector pCAMBIA1300. The recombinant vector pCAMBIA1300-HST1 was transformed into the callus of the hst1 mutant by an Agrobacterium-mediated method. Five positive transgenic hst1/Hst1 lines were obtained and were as heat-sensitive as WT based on increased wilting and reduced survival rate when compared to the hst1 mutant after heat treatment ( Figure 1 ). Thus, the TGGG insertion in LOC_Os03g57240 is responsible for the increased heat-tolerant phenotype of the hst1 mutant. These results indicate that HST1 has a negative role in heat tolerance in rice.

Interestingly, LOC_Os03g57240 was previously named Drought and Salt Tolerance (DST), which negatively regulates drought and salt tolerance in rice plants based on the analysis of a dst mutant (Huang et al., 2009; Cui et al., 2015). The DST gene encodes a 301-residue transcription factor containing a nuclear localization signal (NLS) and a C2H2-type zinc finger motif at the N-terminus. The C2H2-type zinc finger of DST is responsible for its sequence-specific DNA binding activity (Huang et al., 2009; Li et al., 2013b; Cui et al., 2015). The dst mutant is recessive with increased drought and salt tolerance due to a missense mutation that converted the codon for asparagine at residue 69 to a codon for threonine in the conserved C2H2 zinc finger motif ( Figure 2B ) (Huang et al., 2009). A second mutant, reg1, is semidominant with increased panicle branches and consequently improved grain number (Li et al., 2013b). The reg1 mutant contains an A insertion between the 214^th and 215^th nucleotides of the DST cDNA, which leads to a frameshift after the first 72 amino acids ( Figure 2B ) (Li et al., 2013b). As a result, the NLS and C2H2 zinc finger motif at the N-terminus of DST^reg1 are not altered but the remaining protein sequence is completely changed as a result of the frameshift ( Figure 2B ) (Li et al., 2013b). Both DST^dst and DST^reg1 mutant proteins still contain the conserved C2H2 residues in the zinc finger motif and, as a result, still bind to the TGNTANN(A/T)T sequence, a cis-acting element called DST-binding sequence present in DST target genes (Huang et al., 2009; Li et al., 2013b; Cui et al., 2015). By contrast, the two conserved histidine residues in the DNA-binding C2H2 zinc finger motif in our DST^hst1 mutant protein are eliminated by the frameshift mutation in the hst1 mutant ( Figure 2 ), which would abolish its sequence-specific DNA-binding activity. To test this, we produced recombinant DST and DST^hst1 proteins and analyzed them for sequence-specific DNA-binding activity using electrophoresis mobility shifting assays (EMSA). As shown in Figure 3 , only WT DST but not the DST^hst1 protein was able to bind DNA molecules containing the DST-binding sequence.

To further analyze the role of the rice DST gene, we analyzed its tissue-specific expression using qRT-PCR. As shown in Figure 4A , although DST transcripts were detected in all tissues, their levels were relatively low in callus, root, stem, node and spike. On the other hand, high levels of DST transcripts were detected in leaves ( Figure 4A ). As a negative regulator of plant responses to a range of abiotic stresses, high levels of DST transcripts in rice leaves may be necessary to suppress induction of stress responses under normal conditions and promote other processes such as photosynthesis in the green tissues that are important for plant growth.

We also analyzed the expression pattern of DST in rice leaves in response to heat stress. qRT-PCR analysis showed that the levels of DST transcripts were rapidly down-regulated after 1 hour of heat treatment, followed by gradual recovery over the subsequent 6 hours ( Figure 4B ). During the first 6 hours of heat treatment, the transcript levels of DST in heat-treated leaves were generally lower than those in the leaves with heat treatment ( Figure 4B ). Thus, heat stress reduced expression of DST during the first 6 hours of heat treatment. After 6 hours, the DST transcript levels decreased not only at 42°C but also at 25°C ( Figure 4B ). This reduction in DST transcripts was likely in response to the light-to-dark transition of the rice plants.

To understand the physiological basis of the reduced wilting phenotype of the hst1 mutant during heat treatment, we analyzed the relative water content in the WT and hst1 mutant plants. Detached WT and hst1 mutant leaves were placed at 24°C under 40% relative humidity and water loss was measured after various hours. As shown in Figure 5A , WT rice leaves lost 20-38% more water during the first 4 hours. After 4 hours at the high temperature, water loss in the detached WT leaves was substantially reduced likely due to limited amounts of residual water in the leaves. In the hst1 mutant, water loss also decreased after 4 hours under the heat stress but to a lesser extent than in WT probably due to higher levels of residual water in the mutant leaves ( Figure 5A ). We also subjected whole WT and hst1 mutant plants to heat stress (42°C) and monitored the relative water content for 36 hours. As shown in Figure 5B , significant difference in relative water content between WT and hst1 plants was first apparent after 24 hours under heat stress and became more pronounced in the next 12 hours. As a result, during the last 8-10 hours of the experiments, the hst1 mutant plants steadily maintained about 10% more water than the WT plants ( Figure 5B ). Thus, the hst1 mutant plants were more resistant to water loss than WT plants at the high temperature, which would allow for better survive under heat stress.

Stomata are vital for sensing and adapting to environmental changes including mitigating transpirational water loss under water or other stresses. Therefore, we investigated the stomatal conductance of the hst1 mutant and WT plants. Even under normal temperature, the stomatal conductance of the hst1 mutant was only about 60% of that of WT plants ( Figure 5C ). Under heat stress, the stomatal conductance of WT and the hst1 mutant plants was reduced by about 30% and 20%, respectively, when compared to those under normal temperature ( Figure 5C ). Even with the relatively small reduction for the hst1 mutant with elevated temperature, the stomatal conductance of the hst1 mutant was still about 30% lower than that of WT plants under heat stress ( Figure 5C ). Furthermore, we found that the stomatal density in the hst1 mutant plants was 15% lower than that in WT plants ( Figure 5D ). Thus, the enhanced heat tolerance of hst mutants was associated with decreased stomatal density and low stomatal conductance, which would make the mutant retain more water and survive better than WT plants under heat stress.

To determine whether altered water status of the hst1 mutant was caused by the DST gene mutation, we analyzed transgenic hst1 lines containing the WT DST^HST1 genomic fragment (hst1/HST1) for relative water content and water loss. Indeed, all transgenic hst1/HST1 plants were similar to WT with reduced water content and increased water loss after heat treatment when compared to those of the hst1 mutant ( Figures 5A, B ). Furthermore, the stomatal conductance and stomata number of the hst1 mutant were restored to those of WT in the complemented lines ( Figures 5C, D ). These results further indicate that the TGGG insertion in the DST^HST1 gene is responsible for the increased heat tolerance of the hst1 mutant, demonstrating that the DST transcription factor is a negative regulator of heat tolerance in rice.

To determine whether the enhanced heat tolerance of the hst1 mutant was entirely or partially due to its reduced stomatal conductance, which reduces water loss under heat stress, we compared survival rates of WT and hst1 mutant plants after heat treatment under 65% and 100% of relatively air humidity. As shown in Figure 6 , increased air humidity improved survival rates of both WT and hst1 mutant plants after both 3 and 4 days of heat treatment. Furthermore, the differences in the survival rates after heat stress between WT and hst1 mutant plants decreased when the air humidity was increased from 65% to 100% ( Figure 6 ). It should be noted that at both 65% and 100% humidity, the rice plants were grown hydroponically in a large volume of growth medium (4.5 liters) that was replenished daily during the heat treatment and recovery period. Therefore, the differences in the survival rates after heat treatment between the two levels of air humidity and between WT and the hst1 mutant were not caused by reduced water supply or drought. Instead, the substantial effect of relative air humidity on the heat tolerance was mostly through their effect on the water status of shoots above the ground, which is controlled not only by the water supply from the roots but also by water loss from transpiration. Furthermore, even with 100% air humidity, there were still significant differences in the survival rates between WT and hst1 mutant after 3 or 4 days of heat stress ( Figure 6 ). These results indicate that stomatal conductance plays an important role in increased heat tolerance of the hst1 mutant but other factors also contribute to improved survival of hst1 under high temperatures.

It has been previously shown that DST binds to the TGCTANNATTG elements in its target gene promoters by the C2H2 zinc finger motif and activates their transcription (Huang et al., 2009; Li et al., 2013b; Cui et al., 2015). Expression profile analysis using the Affymetrix Rice Genome Genechip has identified DST-dependent genes, some of which are related to reactive oxygen species (ROS) homeostasis (Huang et al., 2009). Among the genes directly targeted by DST is Peroxidase 24 Precursor (Prx24), which encodes a peroxidase that directly scavenges H₂O₂ (Huang et al., 2009). To determine whether the expression of Prx24 was altered in the hst1 mutant, we compared WT and the hst1 mutant for the transcript levels of Prx24 in response to heat stress using qRT-PCR. As shown in Figure 7A , the transcript levels of Prx24 in WT were substantially higher than those in hst1 even prior to heat treatment. Thus, the loss-of-function mutation in DST gene in the hst1 mutant caused reduction in basal expression levels of Prx24. Upon heat treatment at 42°C, we observed rapid reduction in Prx24 transcripts in both WT and hst1 mutant during the first 0.5 and 1 hours at the high temperature ( Figure 7A ). Following the rapid reduction during the first hour, the transcript levels of Prx24 became relatively stable or even increased slightly during the remaining hours of the experiments ( Figure 7A ). Throughout the 12 hours at the high temperature, however, the Prx24 transcript levels in the hst1 mutant were generally lower than those in WT, particularly during the first 0.5 to 1 hour under heat stress ( Figure 7A ). We also found that the levels of peroxidase activity was substantially lower in the hst1 mutant than in WT during the first 24 hours under heat stress ( Figure 7B ).

To determine whether reduced expression of ROS-scavenging Prx24 was associated with altered ROS levels in the hst1 mutant, we compared WT and hst1 for the H₂O₂ levels. As shown in Figure 7C , the hst1 mutant had a significantly higher level of H₂O₂ than WT even prior to heat treatment. During the first two hours of treat stress, the H₂O₂ levels in the hst1 mutant remained higher than those in WT ( Figure 7C ). By the 12^th hour at the high temperature, the H₂O₂ levels became similar in WT and hst1 mutants ( Figure 7C ). Thus, the loss-of-function hst1 mutation elevated H₂O₂ levels prior to and during the early hours of heat treatment. We also compared WT and the hst1 mutant for the levels of malondialdehyde (MDA), which results from lipid peroxidation of polyunsaturated fatty acids at elevated levels of ROS. Again, we observed that the basal levels of MDA in the hst1 mutant prior to heat treatment were elevated when compared to those in WT ( Figure 7D ). During the first 6 hours at the high temperature, the MDA level remained significantly different between WT and hst1 mutant plants ( Figure 7C ). The difference in the MDA levels between WT and the hst1 mutant during the remaining hours of the experiments was also observed, albeit to a lesser extent ( Figure 7D ).

Heat shock responses are found universally in cells that are characterized by rapid transcriptional activation of genes encoding HSPs, many of which act as molecular chaperones in protein quality control (Craig et al., 1993; Jakob et al., 1993). To determine whether enhanced heat tolerance of the hst1 mutant was also associated with altered expression of genes associated with the heat shock response, we compared 45-day-old WT and the hst1 mutant for the transcript levels of six HSP genes, which have been previously shown to be strongly responsive to heat shock treatment (Zou et al., 2009). In WT, all of these six HSP genes displayed rapid induction by heat treatment. Transcript levels for four of the six tested HSP genes (Hsp17.0, Hsp24.1, Hsp26.7 and Hsp74.8) peaked after 1 or 2 hours at the high temperature ( Figure 8 ). Rice Hsp58.7 displayed relatively slow induction and continued to increase in its transcript level after 2 hours of heat treatment ( Figure 8 ). Rice Hsp80.2, on the other hand, had relatively low induction in its transcript level, although its peak levels occurred quite rapidly at the first 0.5 hour after heat treatment ( Figure 8 ). Importantly, the hst1 mutant was more rapid and robust in induction of these HSP genes under heat treatment ( Figure 8 ). As a result, the differences in the transcript levels for the six HSP genes were most pronounced during the first half hour of heat treatment ( Figure 8 ). The transcript levels of five of these six HSP genes peaked at 0.5 hour after heat treatment in the hst1 mutant, which were 0.5 to 2 hour earlier than those in WT ( Figure 8 ). Even though they peaked at 0.5 hour after heat treatment in both WT and the hst1 mutant, the transcript levels of Hsp80.2 were substantially higher in the hst1 mutant than in WT at the initial hours of heat stress ( Figure 8 ). We also analyzed heat-induced expression of the HSP genes in ten-day-old seedlings and found that the hst1 mutant again displayed significantly faster and stronger induction of the HSP genes than WT ( Supplemental Figure 2A ). Furthermore, we performed western blotting using an anti-HSP70 antibody and found that the hst1 mutant seedlings accumulated HSP70 proteins at rates significantly faster and stronger than those of WT plants during response to heat treatment ( Figure 9 ). Thus, enhanced heat tolerance of the hst1 mutant was also associated with rapid and increased induction of HSP gene expression and HSP protein accumulation, which would lead to enhanced heat shock responses.

Several studies have shown that H₂O₂ is involved in heat shock responses (Miller and Mittler, 2006). Since the hst1 mutant contained increased levels of H₂O₂, we investigated whether the elevated ROS is required for the more rapid and robust induction of rice HSP genes. For this purpose, we treated both 45-day-old WT and hst1 mutant seedlings with diphenyleneiodonium chloride (DPI), an inhibitor of ROS production, and compared WT and hst1 for the induction of the HSP genes during the first half hour of heat treatment, in which the differences in the transcript levels for the six HSP genes were most pronounced ( Figure 8 ). Indeed, treatment of DPI suppressed the induction of the HSP gene transcript levels in the hst1 mutant to the levels similar to those in WT ( Figure 10 ). Similar inhibitory effects of DPI on the stronger induction of HSP genes in the hst1 mutant were also observed at the seedling stage ( Supplemental Figure 2B ). These results suggest that increased ROS levels in the hst1 mutant are involved in the increased transcription of HSP genes in the hst1 mutant.

Heat stress at the flowering and grain-filling stages seriously affects rice grain quality and yield (Zafar et al., 2020a; Yan et al., 2021; Zafar et al., 2022). To determine the role of DST transcription factor in the rice heat tolerance at reproductive stages, we first analyzed the expression of DST in reproductive tissues. Rice plants at the heading and flowering stages were growth at grown at 28°C and 40°C and flowers were collected at various time points for determination of DST transcript levels in the reproductive tissues. As shown in Figure 11 , there was a reduction of DST transcripts during the first hour of treatment at both at 28°C and 40°C. At 28°C, the reduced DST transcripts were largely restored over the next few hours and were largely maintained at the levels similar to those at the beginning of the experiments ( Figure 11 ). At 40°C, the DST transcripts also recovered from the early reduction and displayed a 3- to 4-fold induction between the 3rd and 6th hours after treatment ( Figure 11 ). As a result, the DST transcript levels were 2-4 times higher at 40°C than at 28°C throughout the 12 hours of the experiments ( Figure 11 ). Thus, expression of DST appears to be more responsive to high temperature in the reproductive tissues than in the seedling leaves ( Figure 4B ).

We also compared grain production between WT and hst1 mutant grown under the normal and heat stress conditions. Under normal growth conditions, grain production has been previously studied in another rice mutant for DST transcription factor, reg1, which contains a single nucleotide insertion that leads to a frameshift after the first 72 amino acids ( Figure 2 ). As a result, the NLS and C2H2 zinc finger motif at the N-terminus of DST^reg1 are not altered but the remaining protein sequence is completely changed ( Figure 2B ) (Li et al., 2013b). The reg1 mutant is semi-dominant with increased panicle branches and consequently improved grain number (Li et al., 2013b). The reg1 mutant contained about 14 primary and secondary branches per panicle, compared to about 10 branches in WT plants (Li et al., 2013b). As a result, grain number per main panicle in the reg1 mutant was 63.8% higher than that of WT plants (Li et al., 2013b). The hst1 mutant contains a four-base insertion (tggg) between its 194^th and 195^th nucleotides in DST, which causes a frameshift after the first 64 amino acid residues ( Figure 2 ). Therefore, the mutation in the hst1 mutant not only disrupts the conserved C2H2 zinc finger motif but also alters the remaining protein sequence ( Figure 2 ). Intriguingly, the hst1 mutant had about 10 primary and secondary branches per panicle, similar to those of WT plants ( Figure 12A ). Furthermore, both the seed yield per plant and the grain number per main panicle in the hst1 mutant were about 20% higher than that of WT, primarily due to significantly higher seed-setting rates in the hst1 mutant ( Figures 12B, C ). Thus, the beneficial phenotypes of the hst1 mutant in grain production were not as strong as the reg1 mutant. Furthermore, unlike the semidominant nature of the reg1 mutant, the slightly improved grain production in the hst1 mutant was observed only in the hst1/hst1 homozygous mutant, but not in the Hst1/hst1 heterozygous plants ( Figures 2A–C ).

To determine the role of DST in rice heat tolerance at reproductive stages, we analyzed the effect of heat treatment on the seed yield of WT and hst1 mutant. As described in Materials and Methods, for heat treatment, rice plants at the heading and flowering stages were subjected to 6-hour daily heat treatment (40°C) during the daytime for 7 days and their seed yields were compared to those of rice plants without the daily heat treatment. As shown in Figure 12 , under normal conditions, the seed yield per plant of hst1 was about 20% higher than those of WT and hst1 complemented with HST1 (hst1/HST1). This increase in seed yield in hst1 was primarily resulted from the increase in the grain number per panicle ( Figure 12 ). On the other hand, WT and hst1 had similar numbers of branches per panicles. Daily heat treatment of 6 hours at 40°C for 7 days reduced seed yield and grain numbers by approximately 80-85% in WT and hst1/HST1, but only about 50% in hst1 ( Figure 12 ). Thus, hst1 mutant was also more tolerant to heat stress at the reproductive stages than WT.

The negative effects on heat stress on the grain numbers and seed yield in WT and hst1 mutants were strongly correlated with the seed setting rates ( Figure 12D ). Under normal growth conditions, WT, hst1 and hst1/HST1 had very similar seed setting rates of approximately 70% ( Figure 12D ). When grown under daily heat stress, the seed setting rates were reduced by 80-90% in WT and hst1/HST1, but only about 45% in hst1 ( Figure 12D ). We also observed that heat stress has a stronger effect on the pollen viability of WT and hst1/HST1 than on that of hst1 ( Figure 13 ). Heat treatment for 1 day resulted in reduction of pollen viability by 12% in WT and hst1/HST1 but only by 7% in hst1 ( Figure 13 ). Interestingly, increasing daily heat treatment from 1 day to 5 days, led to a significant recovery in pollen viability and as a result, the pollen viability of WT and hst1 was reduced only by 10 and 5%, respectively, when compared to those under normal growth conditions ( Figure 13 ). These results collectively indicate that heat stress negatively impact reproductive processes, ultimately leading to reduction in grain formation and seed yield.

As described earlier, two rice mutants, dst and reg1, for the rice DST gene have been previously reported with altered phenotypes in drought and salt tolerance and in grain production, respectively (Huang et al., 2009; Li et al., 2013b; Cui et al., 2015). The dst mutant is recessive with increased drought and salt tolerance due to a missense mutation that causes the substitution of an asparagine (N) at residue 69 for a threonine (T) in the conserved C2H2 zinc finger motif ( Figure 2B ). Despite the substitution of a conserved amino acid in the C2H2 zinc finger motif, the sequence-specific DNA-binding activity of the DST^dst mutant protein was not altered and the enhanced drought and salt tolerance of the dst has been attributed to abolished or reduced transcription-acting activity of the mutant protein (Huang et al., 2009). Therefore, we also examined whether the hst1 mutant was altered in tolerance to salt. When treated with 0.6% NaCl for 12 days, more than 50% of hst1 mutant plants but only about 20% WT plants survived ( Supplemental Figure 3 ). The enhanced salt tolerance was observed only in the hst1/hst1 homozygous lines, but not in the hst1 mutant complemented with the WT DST genomic fragment (hst1/HST1). Thus, the dst and hst1 mutants share the same phenotype of enhanced salt tolerance and as with the dst mutant, this salt-tolerant phenotype of the hst1 mutant is recessive. Like the dst mutant, the leaf width of the hst1 mutant was also substantially increased ( Supplemental Figure 4 ).

The identifiers for the rice genes described in this article are as follows: HST1 (LOC_Os03g57240), OsPrx24 (LOC_Os02g06630) and OsCKX2 (LOC_Os01g10110).