Oryza sativa ssp. japonica cv. Nipponbare and O. rufipogon (named CWR1) which were collected from Yuanjiang County, Yunnan Province, China, were studied in this study (Li et al., 2020a). They both display photoperiod sensitivity, in which flowering is delayed under LD conditions and induced under SD conditions. Seeds were germinated on ½ Murashige and Skoog medium for 10 days. Seedlings were transplanted to plastic pots and grown in the controlled growth room under either LD (14/10 light/dark cycle, 28/22°C) or SD (10/14 light/dark cycle, 28/22°C) conditions. Light intensity was approximately 1,000 μmol m^–2 s^–1 with humidity of approximately 50%.

The two genome assemblies of O. rufipogon (Li et al., 2020b) and Nipponbare (Ouyang et al., 2007) were retrieved to identify CCT TFs. The Nipponbare reference genome (RGAP_7) was downloaded from RGAP database¹. The longest isoforms were extracted using the Fast Get Representative program of TBtools². Unless otherwise stated, the longest isoform was used throughout the study. HMMER 3.0 was employed to screen the protein sets with the Hidden Markov Model (HMM)³ file of CCT (PF06203), BBOX (PF00643), and REC (PF00072) as queries (cutoff = 0.01, other parameters of default). The putative CCT proteins in which the length of the aligned domain is smaller than 50% of what HMM file annotated were filtered out. The redundant sequences were discarded after BLASTP searches (E-value < 10^––10). Proteins containing CCT domain and lacking BBOX and REC domains were classified as CMF genes. Proteins with CCT domain and additional BBOX or REC domain toward their amino-terminus were defined as COL or PRR genes, respectively. The deduced CCT TFs were further checked for the existence of the corresponding domain by using the Conserved Domain Database⁴. We named the CCT TFs with initials of genus and species and numerical symbols based on their chromosomal locations.

The molecular weight (D) and isoelectric point (Pi) of OrCCT TFs were calculated by ExPASy⁵. The web-server BUSCA was used to predict the subcellular localization of OrCCT proteins⁶. The information of position on chromosomes, exons, introns, and UTR regions of OrCCT genes was extracted from the gene finding format (GFF3) file. MEME software⁷ was used to identify the conserved motifs with the width of each motif = 10–100 amino acid residues, maximum number of motifs = 10, and other parameters of default values (Bailey et al., 2009). The visualization of gene structure and conserved domain (including classification) were conducted using the Gene Structure View tool of TBtools (Chen et al., 2020).

Multiple Collinearity Scan toolkit (MCScanX) is often used to scan multiple genomes to detect putative homologous chromosomal regions using genes as anchors (Wang et al., 2012). To identify the putative orthologous CCT genes between O. rufipogon and O. sativa, the inter-species collinear relationship was identified using MCScanX with the parameters recommended by MCScanX’s manual (Wang et al., 2012). The collinear and syntenic gene pairs of CCT genes were extracted from the MCScanX output files. In this step, the data sets include both paralogs and orthologs. To remove the possible paralogs, the genes that showed the same order on chromosomes were selected as orthologous CCT genes between O. rufipogon and O. sativa.

Gene duplication events within CCT TFs were detected by MCScanX (Wang et al., 2012), and then visualized by Advanced Circos software (see text footnote 2). Non-synonymous (ka) and synonymous (ks) substitution of the paired CCT genes were calculated using KaKs_Calculator 2.0 (Wang et al., 2010). Gene duplication events were approximately dated according to the eq. T = Ks/2λ (λ = 6.5 × 10^–9) (Yu et al., 2005). The comparative synteny relationships of CCT TFs between O. rufipogon and O. sativa were constructed by Multiple Synteny Plotter software (see text footnote 2).

All identified CCT TFs were divided into the three subfamilies according to their domains. The sequence of CCT TFs from Brachypodium distachyon and O. sativa ssp. indica was downloaded from the Phytozome database v13⁸. The sequence of OnCCT TFs was downloaded from the Gramene database⁹. Multiple sequence alignment of CCT full proteins from the four species was performed by using MAFFT 7.243 with E-INS-i algorithm (Katoh and Standley, 2013). The Neighbor-Joining (NJ) phylogenetic tree was inferred by MEGA6 (Kumar et al., 1994) with bootstraps = 1,000.

Total RNA was extracted from the leaves of 90-day-old plants using the QIAGEN plant RNA kit (Hilden, Germany). The concentration and quality of RNA were evaluated using NanoDrop 2000 UV-VIS spectrophotometer (NanoDrop Technologies, Wilmington, DE, United States). Paired-end reads were generated on a HiSeq 2000 platform following the manufacturer’s instructions (Illumina, United States). RNA-sequencing (RNA-seq) data were mapped on the reference genome with HISAT2 2.1.0 (Kim et al., 2019). FeatureCounts 1.6.2 was used to count the number of reads mapped on exons (Liao et al., 2014). Differentially expressed genes (DEGs) were evaluated by edgeR 3.32.0 (Robinson et al., 2010). Genes with p < 0.05 and log2 fold-changes >1 were considered as DEGs. Further screening among the initial DEGs was performed based on fragments per kilo-base per million fragments mapped (FPKM) values.

Total RNAs were extracted from the leaves using RNAiso Plus (TaKaRa, Shiga, Japan). The first cDNA strand was synthesized with 2 μg total RNA, using Moloney murine leukemia virus reverse transcriptase (Promega, Madison, WI, United States) with 10 ng of the oligo(dT) 18 primer and 2.5 mM deoxyribonucleotide triphosphate. Synthesized cDNAs were used as templates for quantitative real-time PCR (qRT-PCR) with SYBR Premix Ex Taq II (TaKaRa) and the Rotor-Gene 6000 instrument system (Corbett Research, Sydney, NSW, Australia). The primers used for qRT-PCR were designed according to O. rufipogon reference sequences. The specificity of primers in both O. rufipogon and Nipponbare was checked by melting curve. The relative expression levels were calculated with rice Ubi1 as an internal control. Each dataset was collected from five independent biological repeats. The primers used are listed in Supplementary Table 1.

The 2,427-bp full-length genome DNA sequence of OrCCT24 was amplified from CWR1 using PCR with specific primers (CATAAGCTTTATCCGTTCATGTCGATGGGA and CC GGTACCCTATCTGAACCATTGTCCAAGC, where underlined sequences indicate HindIII and KpnI enzyme sites, respectively). The PCR fragments were cloned into the pGEM-T Easy vector for blue-white screening. After checking the insert by DNA sequencing, the cloned fragment from the positive clone was moved into the overexpression binary vector pGA3426 under the control of the maize ubiquitin 1 promoter (Kim et al., 2009). After checking its quality by DNA-sequencing, the recombinant vector was transformed into Nipponbare via Agrobacterium-mediated co-cultivation (An et al., 1989). Transgenic rice plants were generated through the stable transformation method as previously reported (An et al., 1989). The putative positive calli were transferred to shoot induction medium that contains 40 mg L^–1 hygromycin.

We identified 41 candidate OrCCT TFs in O. rufipogon (PRJCA002637)¹⁰. The proteins were named as OrCCT01 to OrCCT41 according to their chromosomal locations (Figure 1 and Supplementary Table 2). In addition, 41 OsCCT TFs were identified in the Nipponbare reference genome, as previously reported (Zhang et al., 2017). The molecular weight of OrCCT proteins ranged from 9,689.86 D (OrCCT40) to 171,328.16 D (OrCCT13). Their isoelectric points varied from 4.09 (OrCCT20) to 11.44 (OrCCT40) (Supplementary Table 2). Our results suggest that OrCCT proteins varied greatly among molecular features. Our prediction using BUSCA (Savojardo et al., 2018) suggested that 32 OrCCTs were located in the nucleus, while others were in chloroplast (5), extracellular space (3), and mitochondrion (1) (Supplementary Table 2).

The phylogram of CCT genes in O. rufipogon showed that OrCCT TFs were grouped into the three clusters based on their conserved domains (Figure 1A). The first cluster was the CMF subfamily with 19 members, the second was the COL subfamily with 17 members, and the third was the PRR subfamily with five members. The number of the possessed exons ranged from 1 (OrCCT36) to 33 (OrCCT28) (Figure 1B). The motif number of OrCCT genes alternated from 1 to 6. All CCT members possessed motif 1. The CMF members, OrCCT38 and OrCCT39, had the most motif, which possessed additional motif 5, 6, 7, 8, and 9. COL members had additional motif 2, while motif 10 was specifically presented in OrCCT08 and OrCCT21. PRR members possessed motif 1 and motif 3 (Figure 1C). The results suggest that the classification of OrCCT genes is coincident with their conserved motifs. The sequence information for each motif was present in Supplementary Table 3.

Our results showed that the OrCCT genes were unevenly distributed on the 12 chromosomes of O. rufipogon. Chromosome 1 contained the largest number of OrCCT TFs (8), and chromosomes 1 and 4 had only one OrCCT TF (Figure 2A). The distribution of OsCCT genes on chromosomes is similar to that in O. rufipogon (Figure 2B). Our results showed that there were 11 duplicated OrCCT gene pairs in O. rufipogon (Figure 2A). OrCCT37, OrCCT38, OrCCT39, and OrCCT40 were present as tandem duplicated genes on Chromosome 12 (Figure 2A). Thirty-eight OrCCT genes had the orthologous genes in O. sativa (Figure 2 and Supplementary Table 4). However, the orthologs of OrCCT27, OrCCT36, and OrCCT40 were absent in O. sativa (Figure 2C). In addition, we failed to identify orthologs of OsCCT19, OsCCT25, and OsCCT37 in O. rufipogon, indicating that they are likely O. sativa-specific (Figure 2 and Supplementary Table 4).

Ten duplicated gene pairs were present in both O. rufipogon and O. sativa (Figure 2 and Supplementary Table 5). We used the formula T = Ks/2λ to evaluate approximate dates of duplicated genes (DEs). The dates of shared DEs of CCT genes varied from 23.05 to 89.31 million years ago (Mya) in O. rufipogon and O. sativa (Supplementary Table 5). The DE OsCCT37-OsCCT40, which was estimated to generate about 0.7 Mya, was absent in O. rufipogon, indicating that it probably occurred after the domestication of O. sativa. ω (dN/dS) is a good indicator of selective pressure at both nucleotide and protein levels. It is often expected that ω > 1, ω = 1, and ω < 1 imply positive selection, neutral selection, and purifying selection, respectively (Zhang et al., 2014). Our results suggest that nearly all duplicated CCT gene pairs underwent negative selection in both O. rufipogon and O. sativa (Supplementary Table 5).

We further investigated the diversification of CCT TFs in Brachypodium distachyon, Oryza nivara, and O. sativa ssp. indica (Supplementary Table 6). All investigated species possessed five PRR genes. Our results showed that O. rufipogon, O. nivara, and O. sativa ssp. japonica possessed the same composition of CCT subfamilies (19 CMF, 17 COL, and five PRR) (Figure 3A and Supplementary Table 6). A phylogenetic tree of CCT TFs was constructed using the complete protein sequences from the four species, including B. distachyon, O. rufipogon, O. nivara, O. sativa ssp. japonica, and O. sativa ssp. indica. As shown in Figure 3B, the CCT proteins could be divided into three clusters with nine clades (A to I). Clade A contained all PRR proteins. Clade B, C, F, and I consisted of CMF sub-family proteins. Clade D possessed COL proteins. The clade E, G, and H were composed of COL and few CMF proteins. Interestingly, the COL proteins were closely related to CMF proteins, suggesting that the COL proteins might originate from CMF proteins by gaining the BBOX domain. Alternatively, the CMF proteins were derived from COL proteins due to the loss of the BBOX domain.

Rice senses the day length by endogenous genetic factors to onset reproductive growth (Cho et al., 2017). Previous results suggested that 18 OsCCT genes are flowering regulators in O. sativa (Zhang et al., 2020). To investigate whether OrCCT can respond to photoperiod, the O. rufipogon plants were grown under LD and SD, and the second leaves from the top of main stems were collected at Zeitgeber time (ZT)-2 h, 8 h, and 15 h at 90 days after germination (DAG), respectively. RNA-seq experiments generated temporal expression profiles of all 41 OrCCT genes with three independent replicates.

Our results showed that thirty OrCCT genes showed significantly different expression levels among ZT-2 h, ZT-8 h, and ZT-15 h under LD (Figure 4A). Seven genes (OrCCT06, OrCCT14, OrCCT16, OrCCT22, OrCCT24, OrCCT30, and OrCCT34) were highly expressed at ZT-2 h and weakly expressed at ZT-15 h, suggesting that they are morning-peak genes. Ghd7, the rice ortholog of OrCCT24, was highly expressed in the morning (Xue et al., 2008). Twelve genes (OrCCT04, OrCCT08, OrCCT09, OrCCT12, OrCCT13, OrCCT19, OrCCT20, OrCCT21, OrCCT025, OrCCT027, OrCCT33, and OrCCT36) were highly expressed at ZT-15 h and weakly at ZT-2 h and ZT-8 h, suggesting that they are evening peak genes (Figure 4A). OrCCT20 is the ortholog of rice Hd1 that is highly expressed in the evening (Cho et al., 2018). Eleven genes (OrCCT01, OrCCT03, OrCCT05, OrCCT07, OrCCT11, OrCCT15, OrCCT17, OrCCT26, OrCCT31, OrCCT32, and OrCCT35) exhibited a high expression level at ZT-8 h compared to ZT-2 h and ZT-15 h (Figure 4A).

Under SD, daily expression patterns of OrCCT genes were similar to those observed from LD (Figure 4B). All seven genes that were expressed most highly at ZT-2 h also showed a similar morning-peak expression under SD. Among 12 evening-peak genes, nine exhibited similar daily expression patterns between LD and SD. However, three genes (OrCCT04, OrCCT33, and OrCCT36) were similarly expressed at ZT-8 h and ZT-15 h under SD conditions. Instead, five genes (OrCCT15, OrCCT18, OrCCT23, OrCCT38, and OrCCT39) that failed to show evening-peak under LD displayed a high expression at ZT-15 h under SD. It is well known that ZT-15 h is at the beginning of the dark period under LD whereas the time is at near midnight under SD. Thus, the difference in some CCT genes might be due to the day-length.

To validate the veracity of our RNA-seq results, we tested twenty OrCCT TFs that showed rhythmic expression by using qRT-PCR experiments. The relative expression patterns of the selected genes were almost consistent with those of RNA-seq analysis (Supplementary Figure 1).

The expression patterns of 16 OrCCT and 14 OsCCT genes were measured by qRT-PCR at different developmental stages under LD. Four flowering regulators, Ehd1, Hd3a, RFT1, and OsGI, were included to monitor the developmental stages of plants. The penultimate leaves of the main stems were sampled from O. rufipogon (CWR1) and Nipponbare plants at ZT-2 h, ZT-8 h, and ZT-15 h at 4 days intervals. The time points for qRT-PCR corresponded to the expression peak as shown in Figure 4. In Nipponbare, the transcript level of Ehd1 rapidly started to increase at 46 DAG, peaking at 75 DAG (Figure 5A). Hd3a and RFT1 also exhibited similar expression patterns with Ehd1 in Nipponbare plants (Figures 5B,C). However, all three genes did not express at a detectable level during the experimental period in CWR1 (Figures 5A–C). OsGI kept a high expression level until 54 DAG and then rapidly declined in Nipponbare, while it remained at a high level in CWR1 (Figure 5D). The phenotypic observation showed that Nipponbare flowered at 86–90 DAG, while CWR1 showed a non-flowering phenotype when grown for >213 DAG. Our results indicate that Nipponbare can complete the floral transition with the promotion of Ehd1, Hd3a, and RFT1 under LD. However, O. rufipogon remained at the vegetative growth phase during the investigated period.

The transcript level of Hd1 that is a major photoperiod-sensitive floral regulator stayed at a relatively constant level in Nipponbare (Figure 5L). A similar expression pattern was observed for OrCCT20 that is an ortholog of Hd1 in O. rufipogon. The expression level of Ghd7 decreased to a low level at 61 DAG after floral transition in Nipponbare, but the transcript amount of OrCCT24 remained at a much high level and did not decline during the experimental period in CWR1 (Figure 5N). Similarly, the transcript levels of Ghd2 and OsCCT22 decreased after floral transition in Nipponbare, while their orthologs in O. rufipogon, OrCCT09 and OrCCT21, respectively, remained at relatively high levels during the investigated stages (Figures 5H,M). The content of OsPRR59 was high before the floral transition and the level declined after floral transition in Nipponbare, but its ortholog, OrCCT35, was lowly expressed at all stages in CWR1 (Figure 5R).

Three CCT genes interestingly exhibited opposite expression patterns between CWR1 and Nipponbare plants. The transcript levels of OsCCT01, OsCCT04, and NRR decreased as the Nipponbare plants grew up, whereas gene expression levels of their orthologs, OrCCT01, OrCCT04, and OrCCT18, increased during the experimental period in CWR1 (Figures 5E,F,K). The expression level of OsPRR37 was relatively low and slightly increased after 75 DAG in Nipponbare, but the level of its ortholog OrCCT26 increased rapidly after 61 DAG in O. rufipogon (Figure 5O), suggesting that OrCCT26 is a strongly functional allele of OsPRR37. Transcript level of DTH2 that is a rice flowering activator gradually increased after floral transition in Nipponbare, while OrCCT08 remained at a relatively high level during the experiment in CWR1 (Figure 5G).

Several genes showed similar expression patterns between Nipponbare and O. rufipogon. The transcript levels of both OsCO3 and OrCCT30 were high at 42 DAG and declined to low levels at 68 DAG in Nipponbare and O. rufipogon (Figure 5Q). Expression levels of OsPRR73 (OrCCT11) and OsCOL15 (OrCCT29) did not vary significantly during the experimental period in both Nipponbare and CWR1 (Figures 5I,P). The developmental expression pattern of OsCOL10, a floral repressor downstream of Ghd7 (Tan et al., 2016), was similar to its ortholog OrCCT14 (Figure 5J), indicating that they may function similarly.

The transcript levels of O. rufipogon-specific CCT genes, OrCCT36 and OrCCT40, were at a relatively low level and did not change significantly during the experimental period, indicating that they may not involve in controlling flowering (Figures 5S,T). Sequence similarity and developmental expression patterns suggest that OrCCT08, OrCCT24, and OrCCT26 are the functional alleles of DTH2, Ghd7, and OsPRR37, respectively. Expression levels of OrCCT01, OrCCT04, OrCCT09, OrCCT18, OrCCT21, OrCCT24, and OrCCT26 were high during the vegetative phase, suggesting that they may function as the flowering suppressor in O. rufipogon under LD.

Flowering is induced by 1 week SD treatment in O. sativa (Doi et al., 2004). To examine whether SD treatment induces flowering in O. rufipogon, we applied SD treatment to LD-grown CWR1 with Nipponbare as a control. At 40 DAG, rice plants were transferred to the SD growth room. After 10 days treatment, these plants were transplanted back to the LD growth room until flowering (Figure 6A). All SD-treated Nipponbare plants flowered evenly 13.5 days earlier than the mock-control (continuously grown under LD) plants, suggesting that, as expected, 10 days SD treatment induced flowering in Nipponbare (Figures 6A,E,F). However, O. rufipogon plants treated in the same way did not induce flowering even at 180 days after the treatment (Figures 6A,E,G). For the Nipponbare plants, Ehd1 and Hd3a were induced after 3 days of SD treatment, and the transcript level of RFT1 increased after 7 days of treatment (Figures 6B–D). However, these three genes were expressed at low levels in the SD-treated O. rufipogon plants (Figures 6B–D).

Because SD treatment at 40 DAG did not induce the expression of flowering regulatory genes, we assumed that O. rufipogon requires a longer vegetative growth period than Nipponbare before the onset of phase transition. Therefore, SD treatment was imposed on the 80-DAG CWR1 plants that were first grown under LD (Figure 7A). The SD-treated O. rufipogon plants flowered at 132–140 DAG, while mock plants did not flower even after growing for >223 DAG (Figures 7A,F). Compared with the mock plants, the expression of OrEhd1 was induced after 5 days of SD treatment (Figure 7B). Similarly, transcript levels of OrRFT1 and OrHd3a were increased with the treatment (Figures 7C,D).

The expression levels of 41 OrCCT genes further showed that, compared with mock plant, the expression of OrCCT08, OrCCT11, OrCCT12, OrCCT22, OrCCT24, OrCCT26, OrCCT31, OrCCT32, and OrCCT35 were significantly downregulated in the SD-treated plants, indicating that these CCT genes function as flowering suppressors in O. rufipogon (Figure 7E). It was reported that DTH2, the orthologous gene of OrCCT08, may induce flowering under LD. The expression level of DTH2 peaked at the beginning of the dark period and gradually reduced after that time (Wu et al., 2013). ZT-15 h is at the beginning of dark in LD, whereas the time is 5 h after dark in SD treatment. Therefore, the significantly decreased expression of OrCCT08 in SD-treated plants might be due to the change of day-length.

OrCCT24, the ortholog of rice Ghd7, was highly expressed in O. rufipogon compared to Ghd7 in Nipponbare (Figure 5N). Sequence analysis showed that three single nucleotide polymorphisms (SNPs) were found in the coding region of OrCCT24 in CWR1 compared with Nipponbare. Among them, two SNPs caused amino acid substitutions, and one SNP was synonymous mutation (Figure 8A). To examine whether the high expression of OrCCT24 caused late flowering in O. rufipogon, we constructed the overexpressed OrCCT24 vector, and then transformed it into Nipponbare (Figure 8A). From 15 independently transformed plants, two lines with high levels of expression of OrCCT24 were selected (Figure 8B). The developmental expression patterns showed that the expression levels of Ehd1, Hd3a, and RFT1 rapidly increased from 50 DAG to 75 DAG in Nipponbare. Hence, the transcript levels of the three genes were measured at ZT-2 h from the transgenic plants at 60 DAG under LD. qRT-PCR experiments showed that the expression of Ehd1, Hd3a, and RFT1 was induced in the wild type (WT), whereas their expression was strongly suppressed in the overexpressed OrCCT24 plants (Figures 8C–E). The transgenic plants did not flower up to 220 DAG, while their WT controls flowered at 85–90 DAG (Figures 8F,G). Our results indicated that OrCCT24 is a strong inhibitor of flowering by suppressing the expression of Ehd1, Hd3a, and RFT1 in O. rufipogon.

Crop wild relatives play an extremely important role in crops’ adaptation to farming practices, market demands, and climatic conditions (Dempewolf et al., 2017). Over the past decade, the reference genomes of approximately 15 Oryza species have been deciphered, which have greatly facilitated comprehensive allele mining in these Oryza species (Zhang et al., 2014; Stein et al., 2018; Li et al., 2020b; Shi et al., 2020). There have been great successes in introducing desired traits from wild rice into cultivated rice, such as cytoplasmic male sterile source (Lin and Yuan, 1980). In this study, we obtained OrCCT genes with the strategy of the reference genome-based gene family identification, which is time-efficient compared with traditional methods of genetic mapping (Peng et al., 2019).

In rice and Arabidopsis, the PRR subfamily is a crucial component of feedback loops of the core oscillator for the circadian clock (Mizuno and Nakamichi, 2005). In the present study, the expression levels of PRR TFs (OrCCT07, OrCCT11, OrCCT26, OrCCT32, and OrCCT35) was significantly divergent among ZT-2 h, 8 h, and 15 h both under LD and SD in O. rufipogon plants, suggesting that OrPRR genes are relevant to the circadian clock. In rice, two florigens, Hd3a and RFT1, activate floral transition by inducing the expression of MADS14 and MADS15 (Shrestha et al., 2014). The OsGI-Ghd7-Ehd1-RFT1/Hd3a pathway regulates rice flowering under LD. In this pathway, Ghd7 represses the cereal-specific flowering inducer gene Ehd1, thereby delays flowering by decreasing expression of Hd3a and RFT1 (Doi et al., 2004; Xue et al., 2008; Cho et al., 2016). OsPRR37, OsCCT01, Ghd2 negatively regulate flowering by downregulating Ehd1 (Koo et al., 2013; Zhang L. et al., 2015; Liu et al., 2016). DTH2 activates flowering by directly upregulating Hd3a and RFT1 (Wu et al., 2013). Overexpressed NRR decreases the expression of Hd3a and RFT1, which consequently delays flowering (Zhang et al., 2013). In addition, Hd1 suppresses flowering under LD when functional Ghd7 is present (Fujino et al., 2019; Zhang et al., 2019).

In the present study, a typical O. rufipogon accession (CWR1) did not flower up to 213 DAG under LD condition. The developmental expression profiles revealed that the orthologs of Ehd1 and florigens were not expressed in CWR1 under LD condition. Several CCT genes, including OrCCT01, OrCCT04, OrCCT09, OrCCT18, OrCCT21, OrCCT24, and OrCCT26, were highly expressed in CWR1 compared to Nipponbare, suggesting that they are repressors of flowering in O. rufipogon. Among these genes, orthologs of six OrCCTs except for OrCCT04 are flowering suppressors in rice (Xue et al., 2008; Koo et al., 2013; Zhang et al., 2013; Zhang L. et al., 2015; Zhang et al., 2020; Liu et al., 2016). In addition, the expression of OrCCT08, OrCCT11, OrCCT12, OrCCT22, OrCCT24, OrCCT26, OrCCT31, OrCCT32, and OrCCT35 were significantly downregulated in the SD-treated plants. With the combination of previous findings and our obtained results in this study, we propose a model for flowering regulation of OrCCT TFs in O. rufipogon under LD (Figure 9). In the model, the florigen genes OrHd3a and OrRFT1 are induced by OrEhd1 that are repressed by OrCCT01, OrCCT09, OrCCT24, and OrCCT26. Among the repressors, OrCCT24 and OrCCT26 are the strongest suppressors. OrGI positively controls OrCCT20 and OrCCT24 expression. In addition, OrCCT11, OrCCT12, OrCCT22, OrCCT31, OrCCT32, and OrCCT35 may negatively regulate flowering (Figure 9). However, their up- and down-stream genes are still unknown. Further efforts are thus needed to elucidate the roles of OrCCT TFs under SD.

OrCCT24, the ortholog of Ghd7, was highly expressed during all examined developmental stages in CWR1 under LD compared to Nipponbare, suggesting that OrCCT24 is a strong repressor of flowering in O. rufipogon. We observed that OrCCT24 was highly expressed in the OrCCT24 over-expressed plants compared with WT. However, the transgenic plants did not flower up to 220 DAG. RFT1 is a major florigen that functions to induce reproductive development in the SAM. It is documented that overexpression of RFT1 resulted in the direct formation of spikelets from most of the transgenic calli (Pasriga et al., 2019). In all, we confirm that 13 OrCCT TFs have played important roles in controlling flowering time, but functional roles of other OrCCT genes remain largely unknown.

Crops are distinguished from their wild progenitors by some typical alterations, such as the loss of seed dormancy and shattering mechanisms, reduced branching, increased fruit or seed size, and changes in photoperiod sensitivity (Olsen and Wendel, 2013). The growth of O. rufipogon is limited to tropical regions (Gao, 2004; Zhao et al., 2013). In this study, O. rufipogon plants did not flower under LD conditions, which is indicative of its high photoperiod sensitivity. In most well-known examples, the members of CCT TFs are involved in the adaptation for photoperiod and flowering, including Tof11 and Tof12 in soybean (Lu et al., 2020), as well as Hd1, Ghd7, and OsPRR37 in rice (Koo et al., 2013; Zong et al., 2021). Our findings suggest that the daily expression patterns of 36 OrCCT genes (of a total of 41 members) changed with the circadian rhythm, indicating that they can respond to the light signal. We also found that 13 OrCCT genes are likely the flowering suppressors based on their expression patterns, and OrCCT08, OrCCT24, and OrCCT26 serve as the strong functional alleles of rice DTH2, Ghd7, and OsPRR37, respectively. Ghd7 and OsPRR37 are the pivotal determinants for strong photoperiod sensitivity in rice (Koo et al., 2013; Zhang J. et al., 2015; Zong et al., 2021). As discussed above, we conclude that the 13 OrCCT TFs have likely contributed to the strong photoperiod sensitivity in O. rufipogon, resulting in extremely delayed flowering under LD.

Rice cultivation has been expanded from its primitive domesticated regions to wide regions due to the long-term natural and artificial selection during rice domestication and subsequent modern improvement (Fujino et al., 2013). In this study, 10 days SD-treatment at the early developmental stage (40 DAG) did not promote flowering in O. rufipogon. Such a result indicates that the wild rice plants may require a long vegetative growth stage before responding to SD induction. When 80 days LD-grown plants were treated with 10 days of SD-induction, the treated O. rufipogon plants flowered at 52–60 days after the treatment while untreated control plants did not flower. During the SD treatment, nine OrCCT genes were significantly downregulated, indicating that they can respond to SD to regulate flowering in O. rufipogon. The photoperiod-responding OrCCT genes may be applied to breeding new rice varieties that possess higher biomass and increased grain yields.