DELAYED HEADING DATE3 , Encoding a Heat Shock Transcription Factor, Delays Flowering Time and Improves Yield in Rice ( Oryza sativa L.)

: Heading date is an essential agronomic trait that affects adaptability and yield in rice ( Oryza sativa ). HSFs (heat shock transcription factors) are a type of transcription factor that responds to environmental stress in organisms. The relationship between the heading date and HSFs has been seldom reported so far. Here, we identiﬁed a new heat shock transcription factor, named DELAYED HEADING DATE3 (DHD3), which can signiﬁcantly delay the heading date by about 14 days and provide improvements of about 77% potential yield in rice. DHD3 protein is localized in the nucleus and has weak transactivation activity. DHD3 delays the heading date by signiﬁcantly suppressing Hd3a and RFT1 expression under long-day (LD) and short-day (SD) conditions. Furthermore, the low-temperature condition greatly enhances the delay effect of DHD3 on the heading date (from 16.1% to more than 89.3%). We propose that DHD3 may involve the temperature-regulated signaling pathway of ﬂowering time in rice and has the potential to improve crop yield.


Introduction
Rice is an important food crop in the world. Heading date is an important agronomic trait that affects rice yield [1]. There are two florigen genes in rice, Hd3a and RFT1, which play important roles in regulating the flowering time of rice under short-day (SD) and longday (LD) conditions, respectively [2]. Rice is a typical short-day plant. Under SD conditions, the initiation of rice flowering is mainly controlled by the GI-Hd1-Hd3a (GIGANTEA-Heading date 1-Heading date 3a) pathway [3]. This pathway is similar to the flowering signaling pathway in Arabidopsis [4]. OsGI can activate the expression of Hd1 and promote the early heading date of rice under SD conditions [3]. However, under LD conditions, the expression of Hd1 inhibited the expression of Hd3a and delayed the flowering time of rice [5]. The opposite effects of Hd1 under SD and LD conditions may be regulated by multiple genes such as PHYB (Phytochrome B), Hd6 (Heading date-6), DTH8 (Days To Heading on Chromosome 8), and Ghd7 (Grain number, plant height, and heading date 7) [6][7][8][9][10]. In addition, the initiation of rice flowering can also be regulated by another flowering pathway, Ehd1-Hd3a/RFT1 (Early heading date 1-hd3a/RICE FLOWERING LOCUS T 1), which is unique to rice [11]. Ehd1, a B-type response factor, is a unique heading date regulation protein in rice and promotes flowering [12]. Early heading date 2 (Ehd2) [13], Early heading date 3 (Ehd3) [14], Early heading date 4 (Ehd4) [15], SDG724 (SET domain group protein 724) [16], Ghd7 [17], CONSTANS-Like 4 (OsCOL4) [18], OsLFL1 (O. sativa LEC2 and FUSCA3 Like 1) [19], OsRE1 [20], and other proteins regulate the heading date of rice by controlling the expression of Ehd1. Environmental factors, such as low temperature, drought, and gibberellic acid (GA), can also affect the heading date of rice through Ehd1 [21][22][23].
Heat shock proteins (HSPs) are molecular chaperones that express ubiquitously in all organisms that maintain or restore protein homeostasis under stressful conditions [24,25]. Heat shock transcription factors (HSFs) are a type of transcription factor that controls the gene expression of HSPs by binding to the heat shock elements (HSEs) of the gene promotor sites of HSPs [26]. Unlike animals that generally encode only a few HSFs, most plants have more than 20 HSFs, which are essential for plant adaptation to various stressful environments [26]. Based on the sequence homology and conserved domains of these HSFs, HSFs in plants can be divided into three classes: class A, B, and C. The conserved domains of HSF proteins include DNA binding domain (DBD), oligomerization domain (OD), nuclear localization signal (NLS), nuclear export signal (NES), activator motifs (aromatic and large hydrophobic amino acid residues embedded in an acidic surrounding, AHA motifs) [27], and repressor domain (RD). Among them, AHA and RD are the unique structure domains of class A and class B, respectively [28]. Many previous studies reported the functions of HSFs in plants, most of which are related to abiotic stresses such as heat, drought, salt, and cold [29]. The first reported HSF genes in plants were three HSF genes in tomatoes, which were induced by heat stress [30]. Liu et al. found that the Arabidopsis HSFA1 quadruple knockout mutant (hsfa1a, 1b, 1c, 1d mutant) had impaired thermotolerance and reduced the expression of most heat-induced genes [31]. HSFA4A, HSFA6B, HSFA8, and HSFC1 genes are induced by low temperature [32]. Under low temperatures, HSFA1 cooperates with NPR1 to promote the expression of HSFA1-related genes and cold acclimation [33]. In addition to responding to extreme temperatures, HSFs are also critical in responding to drought, salinity, osmotic, oxidative stress, and pathogen defense [29].
In rice, spl7 mutants, allelic mutants of OsHSFA4D, show disease mimic spotted leaf phenotype under high temperature [34]. The overexpression of OsHSFA2E increases the high-temperature and high-salinity stress tolerance in transgenic Arabidopsis [35]. Decreased OsHSFA4A expression impairs tolerance to Cadmium in rice [36]. Recently, Zhu et al. reported that the overexpression of OsHSFA3 in Arabidopsis can improve drought tolerance by reducing water loss and reactive oxygen species (ROS) levels [37]. Overexpressing OsHSFB4D in rice exhibited enhanced resistance to Xanthomonas Oryzae by increasing the expression of OsHSP18.0-CI [38]. So far, no HSF has been reported to be related to the heading date in rice.
So far, most reports related to the heading date in rice are focused on the photoperiod pathway. The effect of environmental factors, especially temperature, on rice flowering has been less reported. The mechanism of temperature affecting rice flowering needs more exploration. Here, we identified an HSF associated with rice heading date and named it DELAYED HEADING DATE 3 (DHD3). When DHD3 was overexpressed in rice, the transgenic plants delayed heading and increased yield under both LD and SD conditions. DHD3 protein is located in the nucleus and has weak transcription activation activity. DHD3 inhibits rice flowering by inhibiting the gene expression of Hd3a and RFT1 under both LD and SD conditions. The inhibitory effect of DHD3 was significantly enhanced when the temperature of the growth environment was lower, suggesting that DHD3 may be involved in temperature regulation of flowering. In brief, DHD3 is a novel transcription factor controlling rice heading and has the potential to increase crop yield.

Plant Material and Growth Conditions
Oryza sativa L. ssp. japonica cv. Nipponbare and Kitaake were used for the transformation of rice in this study. All plants were grown in paddy fields in Beijing (116 • 13 E, 40 • 13 N) during summer as the natural long-day (NLD) condition. Plants were also planted in artificial light incubators under a relative humidity of 70%. The controlled longday (LD) condition is 14 h light, 30 • C/10 h dark, and 25 • C, and the controlled short day (SD) condition is 10 h light, 30 • C/14 h dark, and 25 • C. To investigate the flowering time in low temperature treatment, plants were grown at 20 • C in SD conditions. In 2017, the ehd1, ehd2, ehd4, hd1, ghd7, dth8, osprr1, osprr37, osprr59, osprr73, and osprr95 [10,12,13,15,17,[39][40][41] mutants in the Nipponbare background used in the study were isolated in our laboratory using CRISPR/Cas9 technology.

Vectors' Construction and Transgenic Plants' Generation
To generate the overexpression construct of DHD3, the full-length coding sequence (CDS) of DHD3 gene was amplified by specific primers, and the PCR (polymerase chain reaction) product was subcloned into binary vector pCAMBIA1390 using an In-Fusion Advantage PCR Cloning Kit (Clontech, Beijing, China). DHD3 was driven by the cauliflower mosaic virus 35S promoter. To knockout the DHD3 gene using CRISPR/Cas9 technology, a 20 bp gene-specific sgRNA sequence of the target gene was cloned into the entry vector pOs-sgRNA and then subcloned into the destination vector containing the Cas9 expression cassette using the Gateway LR Clonase II Enzyme mix (Invitrogen, Shanghai, China) [42]. Primer sequences used for vectors' construction are listed in Supplementary Table S1.
The constructed vectors were introduced into the callus of rice variety Kitaake or Nipponbare by Agrobacterium tumefaciens (strain EHA105)-mediated transformation [43]. To detect genotyping of transgenic plants, overexpressed plants are detected by amplifying the vector fragment inserted into the genome and agarose gel running. The mutant sites of genes from knockout plants were detected by amplifying flanking fragments containing the target site and then sending them to the company (Invitrogen, Shanghai, China) for Sanger sequencing. Primer pairs used for detection are listed in Supplementary Table S1.

RNA Extraction and RT-qPCR Assay
Total RNA was extracted from the leaves of plants using a ZR plant RNA MiniPrep Kit (ZYMO Research, Beijing, China) and reverse transcribed using a QuantiTect reverse transcription kit (Qiagen, Shanghai, China). RT-qPCR assay was performed using an SYBR premix Ex Taq Kit (Takara, Dalian, China) according to the user manual on ABI 7500 equipment. For normalization, the rice Ubiquitin (UBQ) gene was used as an internal control. All data were collected in biological triplicate and analyzed following the relative quantification method [44]. Primer sequences for RT-qPCR are listed in Supplementary Table S1.

Measurement of Agronomic Traits
The WT (wild-type) Kitaake and OE (overexpressed) lines were planted on 19 May 2016 and harvested on 3 May 2016. The WT Nipponbare and KO (knockout) lines were planted on 10 May 2017 and harvested on 9 May 2017. Multiple agronomic traits, including plant height, tiller number, panicle length, grains per panicle, primary branches number per panicle, secondary branches number per panicle, thousand-grain weight, and days to heading, were manually measured. The date of heading was recorded at the first panicle heading of about 1-2 cm of each plant and heading days were calculated by subtracting the seed sowing date [45]. The harvested seeds were air-dried in a glasshouse and ovendried at 50 • C until~14.0% moisture content remained when the weight was recorded as thousand-grain weight. Panicle length, number of primary branches, number of secondary branches, and number of grains per panicle were measured using the main panicles from WT and transgenic plants, respectively [46].

Subcellular Localization
The full-length CDS of DHD3 was amplified and cloned onto pAN580 vector while fusing the GFP (green fluorescent protein) tag at the C-terminal to generate DHD3-GFP. The DHD3-GFP fusion construct and marker vector (D53-mCherry) [47] were transiently co-transformed into rice, as previously reported [48]. All images were generated by a ZESS LSM880 confocal microscope system.

Transactivation Activity Assay
Transactivation activity assay was performed by the Matchmaker GAL4 Two-Hybrid System (Clontech). The CDS of DHD3 was cloned onto a vector (pGBKT7) while fusing the GAL4 DNA binding domain at the N-terminal. Then, the construction was transformed into the yeast strain AH109. The BD-Ehd4 was also transformed as a positive control [15], and the empty vector pGBKT7 (BD) was used as a negative control, respectively. The βgalactosidase activity was measured according to the Yeast Protocols Handbook (Clontech).

DHD3 Is a Negative Regulator of Flowering Time and Has the Potential to Increase Yield in Rice
To explore new regulators of rice heading date, a transcriptional factor library was constructed using pUbi of maize as a promotor and VP64 (tetrameric repeats of VP16) fused by the rice transcription factors (TFs) as an activation domain. Hereafter, the library was introduced into the japonica rice cultivar Kitaake by an agrobacterium tumefaciensmediated method and overexpressed transgenic plant lines were obtained [49]. So far, several genes regulating the rice heading date from the library have been identified and reported [20,[50][51][52][53][54][55]. In this work, we identified six independent lines that contained the VP64-LOC_Os03g12370 sequence, and these transgenic lines delayed the heading date under both NLD and NSD conditions. Thus, we named LOC_Os03g12370 Delayed Heading Date 3 (DHD3).
To verify the function of DHD3, we constructed a vector that overexpressed DHD3 under the control of pUbi promoter, and then transformed the vector into Kitaake. After two generations of plants, positive transgenic lines were obtained and confirmed by genomic PCR analysis. Three independent homozygous plant lines (DHD3-OE) showed high expression levels detected by RT-qPCR analysis ( Figure 1A,B). Under the NLD condition, all three DHD3-OE lines delayed the heading date by about two weeks ( Figure 1C). We selected one of the lines, DHD3-OE3, for further study. The plants of DHD3-OE3 grown in artificial light incubators showed a delayed flowering time under both LD and SD conditions ( Figure 1D), whereas no significant difference in the leaf emergence rate of DHD3-OE3 and WT was observed before heading ( Figure 1E,F). This suggested that the delayed flowering time of DHD3-OE3 plants was not due to retarded vegetative growth.
The overexpression of DHD3 has the potential to increase grain yield. The grains of the main panicle in DHD3-OE plants increased by about 77% (74.1%, 76.5%, and 81.7% for each line) compared with WT (Figure 2A,B), which suggests a corresponding increase in the potential yield. Besides, other agronomic traits such as plant height, panicle length, number of primary branches, and number of secondary branches were increased significantly ( Figure 2C-F), though tiller number and thousand-grain weight did not change significantly ( Figure 2G,H). Therefore, the increase in grain number per panicle was caused by the increase in the number of primary and secondary branches.

Gene Expression Analysis of DHD3
In order to investigate whether DHD3 has a rhythmical expression pattern, the diurnal expression level of DHD3 in rice leaves was examined by RT-qPCR every 4 h for 48 h under LD and SD conditions. The result showed that DHD3 exhibited a similar rhythmical expression pattern of DHD3 under both SD and LD conditions ( Figure 3A,B). The transcription level of DHD3 peaked at dawn and was relatively lowly expressed at dusk (Figure 3A,B). Besides, RT-qPCR analysis showed that DHD3 was constitutively expressed in various rice tissues including the root, stem, sheath, leaf, and panicle, with the relative highest expression in sheath ( Figure 3C).

DHD3 Protein Is Localized in the Nucleus and Has Weak Transactivation Activity
To investigate the subcellar localization of DHD3, the fusion protein DHD3-GFP and the nuclear marker (D53-mCherry) were transiently co-expressed in rice protoplast. The result showed that the green fluorescence signal exactly overlapped with the nuclear marker, indicating that DHD3 was localized in the nucleus ( Figure 4A). DHD3 belongs to the group of heat shock transcription factors (HSFs), which can regulate the expression of genes encoding HSPs by binding HSEs in their promoters. Thus, we performed the transcriptional activation assays of DHD3 in the yeast GAL4 system. The result showed that the yeast contains BD-DHD3 survive arduously on SD/-Trp-His-Ade medium ( Figure 4B). Furthermore, β-galactosidase activity assay suggested a relatively weak transactivation activity of DHD3, as compared with transcriptional activator Ehd4 ( Figure 4C) [15].

DHD3 Delays the Heading Date by Down-Regulating Hd3a and RFT1
To explore the downstream photoperiodic flowering genes regulated by DHD3, we examined the expression levels of rice flowering-related genes in DHD3-OE3 and WT plants by RT-qPCR. We found that the expression levels of Hd3a, RFT1, and OsMADS14 are significant decreased in DHD3-OE3 plants under both LD and SD conditions ( Figure 5A,B). Under the SD condition, Ehd1 is expressed to a much lower extent in DHD3-OE3 plants than in WT plants at night, while OsCOL4 increases compared with WT at dawn ( Figure 5B). Under the LD condition, these two genes do not exhibit much change ( Figure 5A). On the other side, there is not much change in the expression of Hd1 under both LD and SD conditions ( Figure 5A,B). This suggested that DHD3 suppresses the expression of Hd3a and RFT1 to delay the heading date in rice, but may regulate flowering time by different pathways under LD and SD conditions.

The Expression of DHD3 Is Independent of Some Other Heading Date-Related Genes, but Regulated by Circadian Rhythm-Related Genes
To study the role of DHD3 in the regulatory network of rice heading date, we examined the expression level of DHD3 in partial rice heading date mutants. The results show that the DHD3 expression level does not much change in ehd1, ehd2, ehd4, hd1, ghd7, and dth8 mutants ( Figure 6A,B). On the other side, we found that DHD3 expression was significantly impacted in some circadian rhythm-related gene mutants ( Figure 6C,D). Under the LD condition, the expression of DHD3 is more elevated in osprr1, osprr37, osprr59, osprr73, and osprr95 ( Figure 6C). In contrast, the DHD3 expression is more decreased in these mutants under the SD condition ( Figure 6D). These inconsistent results may be due to the different regulatory mechanisms under LD and SD conditions in rice. The exact role of DHD3 in the rice heading regulatory network remains obscure and needs further in-depth exploration.

Suppression Effect of DHD3 on Heading Date Is Enhanced by Low Temperature
In order to validate whether DHD3 deficiency affects the heading date in rice, we generated DHD3 knockout (KO) plant lines in the japonica rice Nipponbare using CRISPR/Cas9 technology. There is no difference in the heading date in two homozygous frameshift mutated DHD3 KO (dhd3-cr1, dhd3-cr4) plant lines compared with WT ( Figure 7A,B and Supplementary Figure S1). Considering that there are 26 HSF genes encoded by rice and 13 of them belong to Class A as DHD3 [56], it is likely that these genes share function redundancy and there are redundant genes that substitute the function of DHD3. It is no surprise that just knockout DHD3 has no expected earlier heading date phenotype. Previous studies showed that DHD3 expression is elevated in response to cold conditions [56]. To explore whether DHD3 controls the heading date in rice under low temperature, we tested the cold response of the WT and DHD3-OE3 plants under cold conditions (constant 20 • C, SD). We found that, compared with normal temperature (SD, 30 • C, 10 h day/25 • C, 14 h night), WT Kitaake delayed flowering time for about 30 days, but DHD3-OE3 plants did not flower during the entire experimental duration (>163 days, the plants were grown in the incubator for too long to grow) ( Figure 7C). This suggested that a low temperature can greatly enhance the flowering delay effect (delay 16.1% in 30 • C versus delay more than 89.3% in 20 • C) of DHD3 in rice. Meanwhile, both the WT Nipponbare and dhd3-cr1 plants delayed flowering time by about 45 days under low temperature conditions, and there is no difference between these plants ( Figure 7D).

DHD3 Has the Potential to Increase Yield in Rice
HSFs are usually reported as stress-responding proteins. Here, we describe an HSF family gene DHD3 that increases grain yield significantly by extending the vegetative period in rice for the first time. A delayed heading date in rice usually increases yield because it prolongs the vegetative stage of rice and allows the plants to accumulate higher biomass before flowering [57,58]. However, too late of a heading date can cause the rice to experience low temperatures in reproductive growth, resulting in reduced yield. Therefore, it has been the goal of breeders to finely regulate the heading date of rice to adapt to different areas and cropping seasons and to achieve the maximum yield [59]. In our study, overexpression of DHD3 in the japonica variety Kitaake delays the heading date and can increase the grains per panicle by 74%~81%, while the number of tillers and 1000-grains weight shows no significant changes (Figures 1 and 2). Hence, DHD3 has the potential to increase yield in the rice breeding process.

DHD3 Is Involved in Temperature-Regulated Heading Date Pathway in Rice
HSFs have been widely studied in various species as a class of ubiquitous important environmental responsive transcription factors. HSFs have mainly been studied as a type of transcription factor that responds to various biotic and abiotic stresses in plant-related studies [29]. There are few reports that described the relationship between HSFs and plant development, such as flowering time. HSFB2B can bind to the promoter region and suppress the expression of PRR7. When HSFB2B is overexpressed, the circadian rhythm period of plants is changed under different temperatures and delays the flowering time in Arabidopsis [60]. The overexpression of VsHSFA9 regulates AtLED expression, which is a GA signal pathway gene, and delays bloom in Arabidopsis [61]. Recently, HSFA2 was reported to regulate flowering time in Arabidopsis under heat stress. When plants are under heat stress, HSFA2 is activated by H3K27me3 demethylation and upregulates the downstream gene HTT5, which induces earlier bloom in Arabidopsis [62]. So far, there is no HSF reported to be related to the heading date in rice. The heading date is delayed by low temperature in rice under both LD and SD conditions, and the expression levels of Ehd1, Hd3a, and RFT1 are decreased under both day lengths [21,63]. However, DHD3 (LOC_Os03g12370) expression is induced by cold treatment [56]. The promoter binding site assay shows that the binding site motifs of AP2s, DREBs transcriptional factors, which had been reported to be associated with cold response, were found on the promoter of DHD3 [64]. Furthermore, the delayed heading date phenotype becomes much more dramatic in DHD3-OE3 plants under low temperatures ( Figure 7C). These observations suggested that DHD3 may be a flowering time regulator that responds to a low-temperature environment in rice.

DHD3 May Be Involved in the Circadian Clock Signaling Pathway
The circadian clock is regulated by temperature rhythm through the HSF1 oscillation activity in mammals [65,66]. HSFB2B can bind to the promoter of circadian gene PRR7 to regulate its expression [60]. Meanwhile, DHD3 expression was significantly impacted in osprr1, osprr37, osprr59, osprr73, and osprr95 mutants, which are circadian rhythm-related gene mutants. OsPRR37, OsPRR59, and OsPRR73 have been reported to negatively regulate the heading date [67][68][69]. OsPRR1 is a rice ortholog of the Arabidopsis TOC1/PRR1 gene, which is a central element in one of the feedback loops of the circadian clock; it has been reported that it regulates panicle and grain size in rice [70]. In our work, the expression of DHD3 is more elevated in these mutants under LD conditions ( Figure 6C). In contrast, DHD3 expression is more decreased in these mutants under SD conditions ( Figure 6D), which implies that DHD3 expression is regulated by these OsPRR genes under different light conditions. Hence, these results suggested that DHD3 may also be involved in signaling pathways that cross-regulate circadian rhythms between photoperiod and temperature. The functions of DHD3 in rice may be multiple and complex, and further research is needed.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support this study are available in the article and accompanying online Supplementary material.

Conflicts of Interest:
The authors have no conflicts of interest to declare.