Suppressor of rid1 (SID1) shares common targets with RID1 on florigen genes to initiate floral transition in rice

The transition from vegetative to reproductive growth is a critical process in the life cycle of higher plants. Previously, we cloned Rice Indeterminate 1 (RID1), which acts as the master switch for the transition from the vegetative to reproductive phase in rice. Although the photoperiod pathway of RID1 inducing expression of the florigen genes Hd3a and RFT1 via Ehd1 has been established, the alternative pathways for the essential flowering transition need to be further examined. Here, we identified a Suppressor of rid1 (SID1), which rescues the never-flowering phenotype of rid1. SID1 encodes an INDETERMINATE DOMAIN (IDD) transcription factor. Mutation in SID1 showed the delayed flowering phenotype. Gain-of-function of SID1, OsIDD1, or OsIDD6 could restore the rid1 to flowering. Further analyses showed SID1 and RID1 directly target the promoter regions of Hd3a and RFT1, two florigen genes in rice. Taken together, our results reveal an autonomous flowering pathway might be mediated by RID1, thereby controlling the phase transition from vegetative to reproductive development in rice.


Author summary
Transition from vegetative to reproductive phase is a critical developmental switch in the life cycle of higher plants. In rice, our previous work suggested Rice Indeterminate 1 (RID1) acts as the master switch for the transition to flowering. Mutation in RID1 results in a never-flowering phenotype. In order to uncover the molecular network regulated by RID1, a Suppressor of rid1 (SID1) was identified in this study. Both SID1 and RID1 encode a plant-specific INDETERMINATE DOMAIN (IDD) transcription factor. Overexpression of SID1, OsIDD1, or OsIDD6 could rescue the never-flowering phenotype of rid1. Molecular data indicate both SID1 and RID1 physically bind the promoters of the florigen genes Hd3a and RFT1 in rice. Thus, we propose that the transition to flowering could be regulated by RID1 through the autonomous pathway, in addition to the photoperiod pathway. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

Introduction
The post-embryonic development of flowering plants can be divided into two major phases: the vegetative and reproductive growth stages. During vegetative development, shoot apical meristems continue to produce leaves for the generation of organic materials through photosynthesis. After a given number of leaves are generated, endogenous genetic factors and environmental signals control the time of flowering [1]. Molecular regulatory networks that monitor the changes in the environment and complex endogenous signals determine the timing of the developmental transition [2][3][4]. Great progress has been made in elucidating the molecular basis for the flowering transition in Arabidopsis, which represents a long-day (LD) plant [5][6][7][8][9][10]. Numerous genes were identified and integrated into six major pathways: the photoperiod, vernalization, age, autonomous, gibberellin, and ambient temperature pathways [11]. Rice is not only a leading cereal crop in the world, but also a representative short-day (SD) plant for flowering time (heading date) studies. As an important agronomic trait, heading date is crucial for determining the regional adaptability and grain yields [12][13][14].
The molecular mechanisms for flowering time control have been well studied in Arabidopsis. However, studies of heading date control in rice have almost exclusively focused on the photoperiodic pathway [15]. Although rice is regarded as a SD plant, it also has evolved its flowering pathway to induce flowering under LD conditions during artificial domestication at high latitudes [16][17][18][19][20]. Thus, photoperiodic flowering in rice can be artificially considered as two distinct pathways: the evolutionarily conserved OsGI-Hd1-Hd3a pathway for adaption under SD conditions, which is parallel to the GI-CO-FT module in Arabidopsis [21,22], and the uniquely evolved Ghd7-Ehd1-Hd3a/RFT1 pathway for adaptation under LD conditions [12,13,15,23]. To understand the photoperiodic control of flowering in rice more comprehensively, recent investigations have identified some flowering mutants that are insensitive to photoperiod variations. Mutants with RID1/OsID1/Ehd2, a rice ortholog of the maize INDE-TERMINATE1 (ID1) gene, showed a late-or never-flowering phenotype under SD or LD conditions [24][25][26], indicating that RID1 might function as an autonomous factor to induce the floral transition in rice [26]. Ehd3 encodes a plant homeodomain finger-containing protein [27]. Mutation in Ehd3 results in no flowering under LD conditions, suggesting that Ehd3 acts as a flowering inducer in the unique genetic pathway Ehd3-Ghd7-Ehd1 in rice [27]. In addition, Ehd4, encoding a novel CCCH-type zinc finger protein, was identified as a critical regulator promoting flowering under both SD and LD conditions [16]. ehd4 also showed a neverflowering phenotype under LD conditions [16]. All these flowering switches (RID1, Ehd3, Ehd4) thus far identified in rice and have no direct homologs in Arabidopsis [16,26,27]. Thus, it appears that RID1, Ehd3, and Ehd4 may participate in a rice-specific flowering transition pathway, the underlying molecular mechanisms of which are still not well understood.
RID1/OsID1/Ehd2 encodes a highly conserved zinc finger protein in plants [24][25][26]. The zinc fingers and its surrounding sequence compose a so-called INDETERMINATE DOMAIN (IDD), which was identified in all higher plant genomes [28]. Maize ID1 is the founding member of the IDD family and controls the transition to flowering in maize [29]. In vitro DNA binding experiments showed that ID1 binds selectively to an 11-bp DNA sequence with the consensus motif TTTGTCG/CT/CT/aT/aT via the IDD [30]. Sixteen and fifteen IDD members were identified in the genomes of Arabidopsis and rice, respectively [28]. Previous studies of IDD members in Arabidopsis revealed that IDD genes participate in multiple developmental processes. AtIDD8 is involved in photoperiodic flowering by modulating sugar transport and metabolism [31]. AtIDD8, AtIDD3, and AtIDD10, either physically or genetically interact with the GRAS domain proteins SHR and SCR to regulate root development or patterning [32][33][34]. AtIDD1 is required for seed maturation and germination [35]. AtIDD14, AtIDD15, and AtIDD16 play a critical role in lateral organ morphogenesis and gravitropism by regulating spatial auxin accumulation [36]. Recent investigations showed that some IDD members (AtIDD2, AtIDD3, AtIDD4, AtIDD5, AtIDD9, and AtIDD10) interact with DELLAs to control gibberellin homeostasis and signaling and modulate flowering time in Arabidopsis [37,38]. RID1 is the only IDD member being functionally analyzed in rice. RID1 and its putative orthologs, ID1 in maize and SbID in Sorghum, are preferentially expressed in immature leaves and may exhibit conserved function for flowering transition [26,28,29].
In maize, ID1 is a key regulator of the transition from vegetative to reproductive growth [29,39]. The id1 mutant has prolonged vegetative growth and retains vegetative features in the inflorescence [29]. The ID1 gene was proposed to regulate the production or transmission of a mobile florigenic signal [29,40]. Transcript and metabolite profiles indicated that expression levels of major sucrose and starch metabolism genes were altered in the id1 mutant, suggesting that ID1 might be involved in the starch to sucrose transition and sucrose utilization within the leaf [39]. However, similar changes in transitory starch and sucrose are not observed in the photoperiodic flowering plants [39]. Thus, it appears that ID1 is likely engaged in a novel autonomous flowering pathway that is distinct from the photoperiod induction pathway [39,41]. Our previous study showed that RID1 acts as a master switch of the flowering transition in rice [26]. Loss of function of RID1 seriously suppressed the expression of Ehd1 and florigen genes Hd3a and RFT1, suggesting that RID1 plays important roles in photoperiodic flowering promotion in rice [26]. At present, the direct target of RID1 and whether RID1 controls an autonomous flowering pathway in rice remain unclear.
In this study, a gain-of-function mutant suppressor of rid1-D (sid1-D) was identified. sid1-D restored the rid1 mutant to flowering successfully. SID1 belongs to the IDD family in rice. Loss-of-function mutants of SID1 exhibit late flowering under LD conditions. Moreover, our results show that RID1 and SID1 directly regulate the expression of Hd3a and RFT1, two florigen genes in rice. Our results indicate that the original function of RID1 might trigger the expression of florigen genes, thus controlling the flowering transition in rice.

Identification of a suppressor of rid1
Our previous study identified a RID1 knockout mutant (rid1), which shows a never-flowering phenotype under LD or SD conditions [26]. Further examination showed that T-DNA insertion at the second intron of the RID1 gene caused the never-flowering phenotype, and a 5.7-kb genomic fragment containing the entire RID1 coding region and its promoter could successfully rescue the mutant phenotype [26]. To examine whether the full-length cDNA of RID1 could rescue the mutant phenotype, we generated genetic complementary plants transformed with construct (pUBQ::RID1) harboring the RID1 cDNA fragment driven by the Ubiquitin promoter. Among 80 independent transgenic plants, we analyzed 7 positive transgenic plants with the restored normal flowering phenotype. Surprisingly, we discovered one line (#4) in which the transcript of RID1 was undetectable but exhibited a restored flowering (Fig 1A). Its progeny of 200 plants exhibited a phenotypic segregation of flowering to never-flowering of 3:1 (143:57, χ 2 = 1.13, P < 0.05), in which all flowering plants contained the selection marker Kanamycin gene. This observation suggests that the restored flowering plant results from a dominant mutation of a single gene that is likely to co-segregate with a T-DNA insertion event. Thus, we designated this mutant as suppressor of rid1-D (sid1-D) ( Fig 1B).
Next, we investigated the heading date of sid1-D compared to wild-type plants. Under natural-long-day (NLD) conditions during the growing season at Wuhan, China, the heading date of sid1-D was delayed 10 days compared with the wild type. In the growth control room, the heading date of sid1-D (72.4 ± 1.1 days for SD; 101.1 ± 1.4 days for LD) was delayed about 2 weeks compared with the wild type under SD or LD conditions ( Fig 1C). Furthermore, sid1-D exhibited a similar leaf emergence rate as that of rid1 under both SD and LD conditions (S1A Fig). The heterozygotes and homozygous sid1-D exhibited an indistinguishable heading date under distinct day length conditions (S1B Fig). These results indicate that sid1-D is a dominant mutant that partially rescued the never-flowering phenotype of rid1.

SID1 is the rice INDETERMINATE DOMAIN 4 (OsIDD4) transcription factor
Because sid1-D is generated by a single gene mutation and co-segregates with a T-DNA insertion, the genomic sequence flanking the left border of the T-DNA insertion site was isolated by thermal asymmetric interlaced PCR [42]. BLAST analysis of the flanking sequence indicated that a T-DNA was inserted into the intergenic region between the annotated genes LOC_Os02g45054 (OsIDD4) and LOC_Os02g45040 (Fig 2A). PCR analysis using the primers P1, P2, and P3 [26] indicated that the genomic background is homozygous for rid1 (S2A Fig). We determined the genotypes of the re-introduced T-DNA insertion site by PCR amplification using the primers P4, P5, and P6 (Fig 2A and S2A Fig). All the plants homozygous or heterozygous for the T-DNA insertion showed the restored heading phenotype, whereas the plants without T-DNA insertion showed the never-flowering phenotype, like that of rid1. Our further analysis of the T-DNA sequence integrated into the genome indicated that a truncated T-DNA insertion event occurred. The truncated T-DNA only remained in the left region containing the selection marker Kanamycin driven by the CaMV 35S promoter and a sequence of the 3' untranslated region of RID1 (162 bp) (Fig 2A).
Because sid1-D is a dominant mutant with re-introduced T-DNA inserted in the intergenic region, we examined the expression levels of genes flanking the T-DNA insertion. Quantitative reverse transcription PCR (QRT-PCR) analysis indicated that the transcript of LOC_Os02g45054 (OsIDD4) was significantly increased, while the other genes showed identical expression patterns in sid1-D and rid1 (Fig 2B). To determine whether the elevated transcript level of OsIDD4 is responsible for the rescued flowering in sid1-D, we introduced a In rice, SID1 encodes a typical Cys-2/His-2 (C2H2) zinc finger protein belonging to the plant-specific IDD protein family, comprising 15 members in rice [28]. Phylogenetic analysis showed that SID1 belongs to a different clade than that of RID1 (

Expression patterns of SID1 and its transcriptional activity
To determine the spatial expression profile of SID1, we examined the expression level of SID1 in various tissues by qRT-PCR at seedling stage (S5A Fig). The analysis showed that SID1 was preferentially expressed in vegetative tissues ( Fig 4A). We also made a construct pSID1::GUS and generated transgenic plants to precisely examine SID1 expression patterns. GUS staining was detected in mature leaves, young leaves, sheath, and root tips and was most abundant in mature leaves and young leaves (Fig 4B to 4H). To examine SID1 expression during the vegetative stage, we harvested young and expanding leaves from wild-type plants every 5 days from day 15 until floral transition. SID1 showed an expression pattern similar to that of RID1, with continual expression in all examined points of young leaves (S5B Fig). In expanding leaves, the transcription level of SID1 was higher than that of RID1 in less than 30-day-old seedling, and Considering that SID1 encodes a C2H2-type zinc finger transcription factor, we also assayed the subcellular localization of SID1. The construct 35S::SID1::GFP was transiently transformed into rice protoplasts. The SID1-GFP exclusively co-localized with the Ghd7-CFP fusion protein (an established nuclear marker; [43]) (Fig 4I), indicating that SID1 is localized in the nucleus. We further examined the transcriptional activity of SID1 in rice protoplasts using a dual luciferase reporter (DLR) assay system. All fragments of SID1, especially its N terminus (amino acids 1-251aa) enhanced the relative luciferase activity compared with the GAL4 binding domain negative control (Fig 4J). These results suggest that SID1 is a nuclear protein showing transcription activation activity.

SID1 is required for normal heading date in rice
To examine the function of SID1 in rice, we generated sid1 mutants using the CRISPR-Cas9 system [44]. The construct containing the Cas9 and sgRNA targeting the first exon of SID1 was designed, and 97 transgenic plants were generated (Fig 5A). PCR amplification products containing the target region were digested by CELI enzyme to detect potential mutations ( Fig  5B). Confirmation of the mutations by sequencing showed that the target region had small deletions of 1-7 bp and three mutant lines were used for further analysis (Fig 5C). T1 family of 40 homozygotes for each sid1 plants (D1, D5, D7) presented a small but statistically significant (P < 0.05) delayed flowering in NLD (82.1±1.4 days for D1; 81.6±1.4 days for D5; 81.5±1.2 days for D7) compared to the wild type (79.2±1.3 days) (Fig 5D and 5E). These results suggest that mutation of SID1 results in late heading.
A previous study reported that the expression of Ehd1, Hd3a, and RFT1 were suppressed in rid1 under both SD and LD conditions [26]. Therefore, we performed qRT-PCR analysis to detect the expression levels of Ehd1, Hd3a, RFT1, and Hd1 in sid1 under SD and LD conditions ( Fig 5F). Hd1 showed an almost identical expression level in sid1 and the wild type under both conditions, suggesting that SID1 had no effect on the expression of Hd1. In sid1, the transcript levels of Ehd1 were partially reduced under LD conditions. Transcript levels of Hd3a and RFT1 were largely reduced in the sid1 mutants under both conditions. These results suggest that SID1 might be involved in flowering regulation through modulation of the expression of Ehd1, Hd3a, and RFT1.
To investigate the possible regulation of Hd3a and RFT1 by SID1 and RID1, we generated transgenic plants overexpressing SID1 or RID1, respectively. The independent transgenic plants overexpressing SID1 showed a similar heading date as that in wild type (S6A to S6C Gain-of-function SID1 performs the role of flowering transition in rid1 Because overexpression of SID1 rescued the failed flowering transition in rid1, we wondered whether overexpression of SID1 would recover flowering pathways activated by RID1. As shown in Fig 6, the transcript levels of Hd3a, RFT1, and Ehd1 were completely suppressed in rid1 under either LD or SD conditions, which were the same results as in our previous investigation [26]. However, in sid1-D background, with the overexpression of SID1, the expression of Hd3a, RFT1, and Ehd1 were elevated and diurnal under both SDs and LDs at all-time points examined during the 24 h period (Fig 6). Because RID1 had only a slight effect on the expression of Hd1 [26], Hd1 showed identical expression patterns in rid1 and sid1-D under both conditions (Fig 6). Thus, overexpression of SID1 might take over the role of initiating the flowering transition in RID1-dependent photoperiodic flowering pathways in sid1-D.
Because SID1 has a highly conserved IDD belonging to members of the plant-specific zinc finger protein family, we speculated that they might exhibit the same DNA binding characteristic. Previous experiments demonstrated maize ID1 selectively binds to the consensus motifs TTTGTCG/CT/CT/aT/aT and TTTTGTCG/C by IDD in vitro, but the consensus motifs without T in the 5' position did not affect the binding affinity of ID1 [30]. To identify possible targets of the SID1, we surveyed the consensus motifs in the promoter regions of genes controlling flowering time in rice. As shown in Fig 7A, a core sequence containing TTTGTC was found at -2191 (I region) and -1923 (II region) in the promoter regions of Hd3a and RFT1, respectively. To examine whether SID1 could directly bind to these fragments, we performed electrophoresis mobility shift assays (EMSA) to assess the potential binding ability in vitro. The recombinant SID1 protein was able to bind to the fragments containing the consensus motif TTTGTC in the promoter regions of Hd3a or RFT1, respectively (Fig 7B). However, the negative controls with mutation in TTTGTC (bio-Hd3a-M or bio-RFT1-M) abolished these binding evens (Fig 7B), indicating that SID1 could specially bind to the TTTGTC motif. The results suggest that SID1 might have the ability to drive the expression of Hd3a and RFT1.
We further examined the transcriptional activity of SID1 using a DLR assay system in rid1 protoplasts. Hd3a or RFT1 promoter driving the firefly luciferase gene was used as reporter and transfected into protoplasts of rid1, respectively. The construct harboring the SID1 gene driven by CaMV 35S promoter was used as the effector. With the increasing effector construct containing the SID1 gene, the luciferase activity was gradually enhanced (Fig 7C and 7D). This result further confirms that a considerable amount of SID1 has the transcriptional activation ability to drive the expression of Hd3a and RFT1 when RID1 was abolished.

RID1 directly activates the expression of Hd3a and RFT1
RID1 is also a member of the IDD family in rice, and QRT-PCR results demonstrated that transcription of Hd3a and RFT1 were seriously reduced in the rid1 mutant [26]. Likewise, we performed EMSA to test the potential interactions between RID1 and the promoters of Hd3a and RFT1. EMSA competition experiments demonstrated that the recombinant RID1 protein could bind to the fragments containing the consensus motif TTTGTC (Fig 8A). The binding activities of RID1 were abolished when the consensus motif was mutated to TTAATC (bio-Hd3a-M or bio-RFT1-M), indicating RID1 could specially bind to the fragments containing the consensus motif TTTGTC in the promoter regions of Hd3a or RFT1, respectively ( Fig 8A). Next, we generated the construct ProRID1::RID1:FLAG:HA and introduced it into the rid1 mutant background by Agrobacterium-mediated transformation. The transgenic plants successfully rescued the flowering of rid1 (S7 Fig). Using the ProRID1::RID1:FLAG:HA transgenic plants, a chromatin immunoprecipitation (ChIP)-QPCR assays was carried out in the young leaves using HA antibody. As expected, the selected regions I and II of the Hd3a and RFT1 promoters were significantly enriched in young leaves (Fig 8B). These results indicate that RID1 may initiate the flowering transition through its direct targets Hd3a and RFT1.
Overexpressing Hd3a restored the flowering of rid1 Because the two florigen genes Hd3a and RFT1 were shown to be the direct targets of RID1, we generated transgenic plants overexpressing Hd3a to investigate whether they can rescue the flowering transition in rid1. We introduced the p35S::Hd3a construct into rid1 and obtained more than 40 transgenic plants overexpressing Hd3a (Fig 9A). Interestingly, all the positive transgenic plants reached flowering at the seedling stage ( Fig 9B). Thus, overexpression of Hd3a caused early flowering in rid1.
Our previous investigation demonstrated that the expression of Ehd1 and Hd3a were completely repressed in the rid1 mutants [26]. Subsequently, we generated transgenic plants with overexpression of Ehd1 in rid1 (Fig 9C). All of the transgenic plants exhibited the neverflowering phenotype, similar to rid1 (Fig 9D). This observation shows that overexpression of Ehd1 is not sufficient to restore flowering transition in rid1. Our results further confirm that RID1-Ehd1-Hd3a/RFT1 is not the sole pathway for floral induction mediated by RID1 in rice [26].

Discussion
Our previous investigation established the rough photoperiodic flowering pathway mediated by RID1 in rice [26], but the detailed molecular mechanism of RID1 initiating the flowering transition remained unclear. In Arabidopsis and yeast, dosage suppression genetic interaction has been known to apply extensively to map functional connections among genes [45][46][47]. To our knowledge, this is the first report to identify a suppressor functionally related to a certain gene in rice. SID1, a suppressor of rid1, was identified as a rice flowering promoter in this study. Gain of function of SID1 led to rescue of the never-flowering phenotype in rid1. SID1 and RID1 showed the binding ability with the promoter regions of Hd3a and RFT1 to drive expression of Hd3a and RFT1. Our new findings indicate that RID1 and SID1 might be involved in the autonomous flowering pathway regulating the transition to flowering in rice.

SID1 is a flowering promoter
Although the identification of SID1 could be due to mere chance in the transgenic events, our genetic and molecular analyses clearly suggested that SID1 is required for promoting flowering in rice. SID1 encodes an IDD-type zinc finger transcription factor, preferentially expressed in mature leaves (Fig 4A), where floral inductive cues are perceived or initiated [22,29]. Mutation in SID1 caused delayed flowering time compared to that of the wild type (Fig 5D and 5E). Overexpression of SID1 could successfully restore the flowering transition in rid1 (Fig 2C and  2D). In addition, the expression levels of Hd3a and RFT1 were greatly suppressed in sid1 ( Fig   Fig 9. Overexpressing of Hd3a and Ehd1 in rid1 plants. 5F). However, the expression of Ehd1 was slightly reduced in sid1 plants (Fig 5F), but was almost completely repressed in the rid1 mutants [26], suggesting that the Ehd1-mediated flowering pathways may differ between rid1 and sid1 mutants. This observation coincides with evidence that rid1 shows the strongest phenotype, never flowering, whereas sid1 shows only slightly delayed flowering (Fig 5D and 5E).
RID1 acts as a master switch for floral transition. SID1 and RID1 might exert their function in the flowering transition with RID1 having priority for driving the expression of Hd3a and RFT1. When the function of RID1 is abolished, only the normal expressing level of SID1 may not enough to trigger the expression of Hd3a and RFT1, or due to non-overlapping expression patterns between SID1 and RID1. Thus, rid1 plants remain in the vegetative growth stage. However, increasing or ectopic expressing SID1 transcripts is responsible for reverting rid1 to the phase of flowering. Subsequently, we demonstrated that florigen genes, Hd3a and RFT1, are up-regulated in sid1-D (Fig 6). Furthermore, SID1 binds Hd3a and RFT1 promoter region in vitro (Fig 7B), and the LUC activity enhanced with increasing levels of SID1 in rid1 protoplasts (Fig 7C and 7D). These evidences support the function of SID1 recovering, at least in part, the SID1-Hd3a/RFT1 pathway to elicit flowering when RID1 is abolished. Expression of Ehd1 was also elevated in sid1-D plants (Fig 6), suggests that other pathways regulated by RID1 may also be activated by overexpression of SID1.

Roles of IDD protein in promoting flowering transition
Both SID1 and RID1 are IDD zinc finger proteins. Proteins containing an IDD comprise a family of zinc finger transcription factors that are unique to plants [28]. The recognition of the DNA consensus sequence is likely to be mediated by the zinc finger modules located in the IDD [30]. The highly conserved IDD is composed of four putative zinc finger domains with spacer sequences between them [28,30]. In vitro DNA binding experiments showed that the second and third zinc fingers in the IDD are required for interaction with the DNA consensus motif [30]. Moreover, a different spacer between these zinc fingers modules in the IDD does not alter DNA binding specificity [30]. In this study, genetic evidences suggest that overexpression of SID1, OsIDD1 or OsIDD6 could restore the rid1 mutant to flowering successfully. We propose the function of SID1, OsIDD1, and OsIDD6 are redundant and that overexpression any of them could take the place of RID1 to initiate the flowering transition when RID1 is absent. However, because mutation of any of zinc fingers of SID1 abolished rescuing the never-flowering phenotype of rid1 (S4A and S4B Fig), suggesting that the first and fourth zinc fingers in the IDD also have critical roles in mediating DNA-protein interactions. Given that SID1, OsIDD1, and OsIDD6 are some co-expression (S8A and S8B Fig) and sid1 null mutants displayed moderate late flowering phenotypes (Fig 5D and 5E), future research is needed to develop the double and triple mutants to understand whether these OsIDDs coordinately modulate flowering time or not.

RID1 may participate in flowering transition through an autonomous pathway in rice
Our previous investigation indicated that RID1 activates the expression of florigen genes (Hd3a, RFT1) mainly by regulating the expression of Ehd1 and Hd1 [26]. These findings suggest that RID1 is involved in two independent photoperiod pathways, mediated by Ehd1 and Hd1, respectively. However, plants harboring nonfunctional alleles of Hd1 and Ehd1 still flower under either SD or LD conditions [48], demonstrating that RID1 may play a role in alternative flowering pathway(s) for the flowering transition. In this study, overexpression of Ehd1 could not reverse the never-flowering phenotype of rid1 (Fig 9C and 9D), further verifying this speculation.
A striking finding of our study is that Hd3a and RFT1 are the direct targets of RID1 (Fig 8). Transcription of Hd3a and RFT1 was completely repressed in rid1 [26], and genetic analysis showed that ectopic expression of Hd3a could reverse the never-flowering phenotype of rid1 (Fig 9A and 9B), indicating that florigen genes indeed act downstream of RID1. Furthermore, ChIP investigations indicated that RID1 could bind the promoter region of Hd3a and RFT1 in young leaves (Fig 8B), suggested that RID1 initiates the expression of Hd3a and RFT1 most likely occurred in the early vegetative stages. Hd3a and RFT1, two florigens in rice, be synthesized in leaves and transported to the shoot apex, and induced flowering [49][50][51]. As previously reported, photoperiod variation, temperature, gibberellin, age, and nutrition have been implicated in floral induction [11,52]; thus, the transition to flowering is complex and involves the convergence of multiple signals onto the florigen genes. RID1 acts as the master switch of phase transition and may function in initiating the expression of florigen genes (Hd3a and RFT1) at the early stage, and then several floral induction cues converge to accumulate florigens to promote floral transition.
Indeed, RID1 expression was detected most abundantly in young leaves at early seedling stages and is unaffected by photoperiod, indicating that RID1 may regulate an autonomous signal for flowering transition. Recent transcription and metabolism analyses showed that maize ID1 actually affects primary carbohydrate metabolism-related genes' function and is closely associated with florigen production in maize mature leaves [39]. Although no clear ortholog of RID1 exists in Arabidopsis, the AtIDD8 was found to regulate sugar transport and metabolism contributing to photoperiodic flowering time [31]. Therefore, mediation of autonomous floral induction by RID1 may involve coordinating the state of carbohydrate metabolism in rice. Characterizing the transcript and metabolite signature changes in rid1 would further provide clues to help us further understand the mechanism(s) underlying the vegetative-reproductive phase transition in rice.

Plant materials and growth conditions
The rice variety used in this study was Oryza sativa subsp. japonica 'Zhonghua11' (ZH11). Plants were grown under NLD conditions in the experimental field during the rice growing season of Huazhong Agriculture University in Wuhan, China, and in a greenhouse during the winter. All transgenic plants were grown under similar growth conditions. Plants were grown in controlled-growth chambers (Conviron) under SD (10 h light at 26˚C/14 h dark at 24˚C) or LD (14 h light at 26˚C/10 h dark at 24˚C) conditions with a relative humidity of 70%. The light intensity was 800 μmol m -2 s -1 .

Plasmid construction and rice transformation
To generate pUBQ::OsIDDs transgenic plants, the OsIDD genomic DNA sequence was amplified and then cloned into pU2301, which was modified from pC2301 vector with the maize Ubiquitin promoter, and then verified by sequencing. An empty pU2301 vector was used as a negative control. For overexpression of Ehd1, the Ehd1 genomic DNA sequence was amplified with primer pair Ehd1-OX-F/Ehd1-OX-R and then cloned into pU2301 by KpnI-BamHI digestion. For overexpression of Hd3a, the Hd3a genomic DNA sequence was amplified with primer pair Hd3a-OX-F/Hd3a-OX-R and then cloned into pS2300 by XbaI-KpnI digestion. To obtain ProRID1::RID1:FLAG:HA transgenic plants, genomic fragments containing the RID1 promoter and coding region lacking a stop codon were amplified with primer pair PFA2300-RID1-F/PFA2300-RID1-R and then cloned in frame into pFA2300 (kindly provided by Saifeng Cheng, Huazhong Agricultural University) by KpnI digestion. The constructs were introduced into Agrobacterium tumefaciens EHA105 and homozygous callus from rid1 was used as the transformation recipient.
To generate pUBQ::SID1(cDNA) and pUBQ::RID1(cDNA) transgenic plants, the coding sequences were amplified by RT-PCR and ligated into the pEASY-T3 vector (TransGen Biotech), and then verified by sequencing. The resulting plasmids were used as templates. Fulllength cDNA of SID1 were amplified with primer pair SID1(CDs)-OX-F/SID1(CDs)-OX-R and then cloned into pU2301 by KpnI-BamHI digestion; full-length cDNA of RID1 was amplified with primer pair RID1(cDNA)-OX-F/RID1(cDNA)-OX-R and then cloned into pU2301 by KpnI-BamHI digestion. To investigate the tissue-specific expression of SID1, approximately 3-kb promoter fragments of SID1 were amplified from genomic DNA and then cloned into pC2300-EX-GUS [53] to create pSID1::GUS. To introduce targeted mutations in SID1 protein, sgRNA:Cas9 expression vector of the SID1 gene was constructed as described previously [44]. The constructs were introduced into A. tumefaciens EHA105 and transformed into the callus derived from ZH11. All primers used for genotyping and vector construction are listed in S1 Table. Site-directed mutagenesis Site-directed mutagenesis was introduced by three-step PCR. Full-length SID1 CDs in pEA-SY-T3 vector were used as templates in the first and second PCR amplifications. In the first PCR, the forward primer SID1(CDs)-OX-F and reverse primers containing the desired mutation were used. In the second PCR, the forward primers containing the desired mutation, which was the complement sequence of the first PCR reverse primer, and reverse primer SID1 (CDs)-OX-R were used. The first and second PCR products were purified, and this mixture was used as a template for the final PCR amplification with primers SID1(CDs)-OX-F/SID1 (CDs)-OX-R. The final products were inserted into the pU2301 vector and confirmed by sequence analyses. The resulting plasmids were introduced into A. tumefaciens EHA105 and homozygous callus from rid1 was used as the transformation recipient. All primers for sitedirected mutagenesis are listed in S1 Table. Identification mutants by CRISPR-Cas9 system Genomic DNA from individual transgenic plants was extracted for PCR analysis. The CELI assay was used to identify the potential mutations. The PCR products amplified with SID1-specific primers SID1-CE-F/SID1-CE-R from individual mutant plants were cloned into pEA-SY-T3 vector (TransGen Biotech) for sequencing.

GUS assay
GUS staining and imaging were carried out as described previously [26].

RT-PCR and qRT-PCR analyses
Total RNA was extracted using TRIzol reagent (Invitrogen). RNA (2 μg) was treated with RNase-free DNaseI (Invitrogen), and first-strand cDNA was synthesized by M-MLV reverse transcriptase (Invitrogen) in a volume of 150 μl. For RT-PCR analysis, 3 μl of the first-strand cDNA described above was used as a template for PCR in a reaction volume of 20 μl. GAPDH served as a control for mRNA levels. qRT-PCR was run in a total volume of 10 μl containing 4.4 μl of the reverse-transcribed product described above, 0.3 μM gene-specific primers, and 5 μl FastStart Universal SYBR Green Master (Rox) superMIX (Roche) on an Applied Biosystems ViiA 7 Real-Time PCR system or the ABI PRISM 7500 sequence detection system according to the manufacturer's instructions. Rice Ubiquitin was set as an internal control. The measurements were obtained using the relative quantification method. All primers are listed in S1 Table.

Subcellular localization of SID1
To construct the subcellular localization plasmids, the full-length CDs of SID1 were amplified with primers SID1-pM999-F and SID1-pM999-R with EcoRI-KpnI digestion sites and inserted into pM999-GFP for fusion with the reporter gene. Rice protoplasts were isolated from 13day-old etiolated seedlings and transformed with the tested pairs of constructs. Fluorescence in the transformed protoplasts was imaged using a confocal laser scanning microscope (TCS SP2; Leica) after incubation at 23˚C for 12-16 h.

Transcriptional activity analysis
The transcriptional activity of SID1 was analyzed using the DLR assay system in rice protoplasts prepared from etiolated seedlings [54]. The firefly luciferase gene driven by the minimal TATA box of the CaMV 35S promoter following five copies of the GAL4 binding element was used as a reporter. The Renilla luciferase gene driven by CaMV 35S was used as an internal control. The different deletion fragments of SID1 were amplified and then fused with the yeast GAL4 DNA-binding domain as effectors, driven by CaMV 35S followed by the translational enhancer O from tobacco mosaic virus. For each assay, 2.5 μg reporter plasmid DNA, 2.5 μg effector plasmid DNA, and 0.5 μg internal control plasmid DNA were co-transfected. After incubating for 12-16 h at 23˚C, the relative luciferase activity was measured using the DLR assay system and the TECAN Infinite M200 microplate reader.
To assess the specific binding and activity of Hd3a and RFT1 promoters, protoplasts were prepared from 2-week-old fully green tissue of rid1 [55]. The coding sequence of SID1 was amplified and then fused into the NONE vector, an effector plasmid driven by the CaMV 35S promoter followed with the translational enhancer O sequence. To generate the Hd3a::LUC and RFT1::LUC reporter genes, the Hd3a and RFT1 promoters were amplified (specific primers are listed in S1 Table) and inserted into 190-LUC vector, respectively. The Renilla luciferase gene driven by CaMV 35S was used as a normalization control. The recovery assays were performed with Hd3a::LUC or RFT1::LUC plus 35S::SID1 at various dosages, respectively.

Electrophoretic mobility shift assays
To express the RID1 protein in Escherichia coli, the CDs of RID1 were amplified with primers 32a-RID1-F and 32a-RID1-R cloned into the BamHI-EcoRI sites of the pET-32a expression vector (Novagen) and then introduced into Transetta (DE3) cells (TransGen Biotech). The target protein was purified with Ni-NTA agarose (Qiagen). To express the SID1 protein in E. coli, the CDs of SID1 were amplified with primers pGEX-4T-SID1-F and pGEX-4T-SID1-R cloned into the BamHI-EcoRI sites of the pGEX-4T-1 expression vector (GE Healthcare) and then introduced into Transetta (DE3) cells (TransGen Biotech). The target protein was purified with GST Fast Flow (GE Healthcare).

Chromatin co-immunoprecipitation assay
Chromatin co-immunoprecipitations (ChIPs) were performed as described previously [56]. In brief, the young leaves of ProRID1::RID1:FLAG:HA transgenic plants were fixed in formaldehyde in a vacuum. The chromatin solution was sonicated and the soluble chromatin fragments were obtained from isolated nuclei. Pre-adsorption with Dynabeads protein G (Invitrogen) was performed to remove nonspecific binding DNA. Immunoprecipitation with anti-HA specific antibody (Pierce HA Tag IP/Co-IP; #26180) and IgG were performed as described previously [56]. Immunoprecipitated DNA was analyzed by qRT-PCR, and the primers are listed in S1 Table online.