Characterization of Mungbean CONSTANS-LIKE Genes and Functional Analysis of CONSTANS-LIKE 2 in the Regulation of Flowering Time in Arabidopsis

CONSTANS-LIKE (COL) genes play important roles in the regulation of plant growth and development, and they have been analyzed in many plant species. However, few studies have examined COL genes in mungbean (Vigna radiata). In this study, we identified and characterized 31 mungbean genes whose proteins contained B-Box domains. Fourteen were designated as VrCOL genes and were distributed on 7 of the 11 mungbean chromosomes. Based on their phylogenetic relationships, VrCOLs were clustered into three groups (I, II, and III), which contained 4, 6, and 4 members, respectively. The gene structures and conserved motifs of the VrCOL genes were analyzed, and two duplicated gene pairs, VrCOL1/VrCOL2 and VrCOL8/VrCOL9, were identified. A total of 82 cis-acting elements were found in the VrCOL promoter regions, and the numbers and types of cis-acting elements in each VrCOL promoter region differed. As a result, the expression patterns of VrCOLs varied in different tissues and throughout the day under long-day and short-day conditions. Among these VrCOL genes, VrCOL2 showed a close phylogenetic relationship with Arabidopsis thaliana CO and displayed daily oscillations in expression under short-day conditions but not long-day conditions. In addition, overexpression of VrCOL2 accelerated flowering in Arabidopsis under short-day conditions by affecting the expression of the flowering time genes AtFT and AtTSF. Our study lays the foundation for further investigation of VrCOL gene functions.


INTRODUCTION
Flowering time is a key factor that influences crop growth and development, and crops achieve higher yields when they flower at the correct time. To regulate flowering time, crops sense the interactions between endogenous and environmental factors to determine the transition from vegetative to reproductive growth (Wickland and Hanzawa, 2015;Beinecke et al., 2018;Eom et al., 2018;Xu and Chong, 2018). Several functional pathways have been identified that regulate the switch from vegetative to reproductive development. These include the photoperiodic, vernalization, ambient temperature, plant hormone, and autonomous flowering pathways (Boss et al., 2004;Jack, 2004;Baurle and Dean, 2006;Wickland and Hanzawa, 2015;Xu and Chong, 2018;Ronald and Davis, 2019;Taylor et al., 2019;Zhang et al., 2019). A number of genes in these pathways are known to be involved in flowering time regulation, including CONSTANS-LIKE (COL) genes, phosphatidyl ethanolamine-binding protein (PEBP) genes, and several members of the MADS-box gene family (Gangappa and Botto, 2014;Wickland and Hanzawa, 2015;Beinecke et al., 2018;del-Olmo et al., 2019;Jin et al., 2019Jin et al., , 2020Jing et al., 2019;Lee et al., 2019;Nam et al., 2019;Ning et al., 2019;Parenicova et al., 2019).
COL genes belong to the zinc-finger transcription factor family and play central roles in plant growth and development (Khanna et al., 2009;Gangappa and Botto, 2014). COL proteins are identified based on their conserved structure, which includes one or two BBX (B-Box) domains and one CCT (CONSTANS, CO-like, and TIMING of CAB1) domain (Khanna et al., 2009;Gangappa and Botto, 2014). The BBX domain can be further divided into two types, B-Box1 and B-Box2, which are recognized by their consensus sequences and the distances between their zinc-binding residues, which are considered to be involved in protein-protein interactions (Khanna et al., 2009). The CCT domain has important functions in transcriptional regulation and nuclear protein transport Khanna et al., 2009;Yan et al., 2011;Gangappa and Botto, 2014). The COL proteins are grouped into three classes based on the number and type of their conserved domains. Classes I and II have two distinct BBX domains and one CCT domain, whereas class III has only one BBX and one CCT domain. Classes I, II, and III contain 6, 7, and 4 members in Arabidopsis, respectively. In addition, several COL proteins contain valine-proline (VP) motifs in their C termini (Khanna et al., 2009;Gangappa and Botto, 2014).
Among these COL members, AtCO (AtBBX1) and its homologs are well studied in many plant species (Khanna et al., 2009;Gangappa and Botto, 2014;Luo et al., 2018;Luccioni et al., 2019;Serrano-Bueno et al., 2020). AtCO is expressed in a rhythmic manner and coordinates light pathway and circadian clock signal inputs in Arabidopsis (Putterill et al., 1993(Putterill et al., , 1995Andres and Coupland, 2012;Song et al., 2013). Thus, AtCO plays an important role in the regulation of flowering time by the photoperiod-dependent pathway. Atco mutants exhibit delayed flowering time under long-day conditions (LD), but under shortday conditions (SD), their flowering times are similar to those of wild-type plants. By contrast, AtCO overexpression plants show early flowering time under both LD and SD conditions (Khanna et al., 2009;Gangappa and Botto, 2014). The AtCO protein binds to cis-acting elements in the promoter region of the flowering activator FLOWERING LOCUS T (AtFT) to active AtFT expression. Moreover, AtCO is regulated by many flowering factors, such as AtGI (GIGANTEA), AtCDF1 (CYCLING DOF FACTOR 1) and AtFKF1 (FLAVIN BINDING, KELCH REPEAT, F-BOX1) (Imaizumi et al., 2005;Sawa et al., 2007). OsHd1 (Heading date 1), the AtCO ortholog in rice, accelerates flowering under SD conditions but delays flowering under LD conditions through the regulation of the AtFT orthologs OsHd3a (Heading date 3a) and OsRFT1 (RICE FLOWERING LOCUS T1) (Yano et al., 2000;Komiya et al., 2008Komiya et al., , 2009). The soybean AtCO orthologs GmCOL1, GmCOL2, GmCOL3, and GmCOL4 can complement the late flowering phenotype of Atco mutants (Wu et al., 2014). In addition to their functions in flowering time and circadian clock regulation, some COL proteins are also involved in abiotic or biotic stress responses, root development and stomatal opening (Khanna et al., 2009;Gangappa and Botto, 2014).
Mungbean is a diploid legume crop, and its seeds contain proteins and nutrients that are essential for human nutrition (Keatinge et al., 2011). The cultivated mungbean is thought to have been domesticated in India, from which it then spread to other areas (Fuller, 2007). Mungbean is considered to be an SD crop, and flowering time is a critical factor influencing its production (Vas Aggarwal and Poehlman, 1977;Imrie, 1996;Kim et al., 2015). Mungbean plants produce a large number of flowers, but only a few set pods. Approximately 70-90% of the flowers are shed, mainly the later-formed flowers of the racemes (Kumari and Verma, 1983;Mondal et al., 2011). Thus, it has been suggested that the prevention of late flowering is an important way to increase mungbean yield (Isobe et al., 1995;Kuroda et al., 1998;Mondal et al., 2011). The sequencing of the mungbean genome provides genetic resources for the investigation of gene functions (Kang et al., 2014), and the study of mungbean flowering time genes can therefore provide essential information for further modification of mungbean cultivars to increase yield. Until now, there has been limited information on the functions of genes involved in mungbean flowering time regulation. In this study, we identified mungbean COL genes and investigated their characteristics, including chromosomal distributions, gene structures, cis-acting elements and gene expression patterns. We also analyzed the functions of VrCOL2 in the regulation of flowering time. Our findings will provide useful information for further characterization of mungbean COL gene functions.

Plant Materials and Growth Conditions
The mungbean reference genome variety VC1973A was provided by Suk-Ha Lee at Seoul National University, Seoul, South Korea (Kang et al., 2014) and used for all experiments in this study. Mungbean seeds were geminated in tap water for 1 day and then planted in soil-filled pots. Seedlings were grown in growth chambers with 16 h 25 • C light/8 h 25 • C dark cycles for LD conditions and 10 h 25 • C light/14 h 25 • C dark cycles for SD conditions. Leaves of 5-week-old mungbean plants were sampled every 4 h after lights-on and used to analyze the diurnal rhythm of gene expression. Multiple tissues were collected from field-grown mungbean plants sown at the end of May in Qingdao, China, including roots, nodule roots, shoot apices, stems, leaves, flowers, pods and seeds (Shi et al., 2021). Tissues were collected in the afternoon (ZT 10-12) in early July for gene expression analysis, and all samples were stored at −80 • C before RNA extraction. Arabidopsis plants were grown in growth chambers with 16 h 23 • C light/8 h 21 • C dark cycles for LD conditions and 10 h 23 • C light/14 h

Identification of Mungbean VrCOL Members
The amino acid sequences of Arabidopsis BBXs were used as blast queries against the National Center for Biotechnology Information (NCBI) and mungbean genome databases 1 to search for mungbean VrBBX proteins (Kang et al., 2014). The presence of conserved BBX and CCT domains in candidate genes was confirmed using the Pfam database and InterPro program with default parameters (Finn et al., 2017;El-Gebali et al., 2019).

Phylogenetic Analysis
The amino acid sequences of CO and COL proteins from Arabidopsis, soybean, Medicago, mungbean, rice, and maize were aligned using ClustalW2 (Oliver et al., 2005), and the resulting alignment was used to construct a phylogenetic tree in MEGA 1 http://plantgenomics.snu.ac.kr/mediawiki-1.21.3/index.php/Main_Page 7.0 using the neighbor-joining method with default parameters (Kumar et al., 2016). In addition, VrBBX proteins were aligned separately in ClustalW2 and used to construct a phylogenetic tree in MEGA 7.0 with the neighbor-joining method.

Chromosomal Distribution and Duplication Analyses
The physical positions of VrCOL genes were obtained from NCBI, and a chromosomal location map was constructed using MapInspect software (Mike Lischke, Berlin, Germany). Duplicated gene pairs were identified using OrthoMCL software as described by Fischer et al. (2011) andJin et al. (2020). The duplicated gene pairs were defined as having greater than 60% amino acid sequence similarity and were visualized using Circos software (Krzywinski et al., 2009).  Structure Display Server (GSDS) to analyze their gene structures (Hu et al., 2015). The full-length amino acid sequences of VrCOL proteins were used to analyze the positions of the conserved BBX and CCT domains using the InterPro program (Finn et al., 2017). The sequence logos of the conserved BBX1, BBX2, and CCT domains were analyzed using the WebLogo platform (Crooks et al., 2004). The conserved motifs present in the VrCOL proteins were identified using MEME tools, with an optimum motif width of 11-50 amino acid residues (Bailey et al., 2009). The cis-acting elements in each VrCOL promoter, 2 kb upstream of the initiation codon, were predicted by PlantCARE (Lescot et al., 2002).

Plasmid Construction and Plant Transformation
To investigate the functions of VrCOL2, a 35S: CDS-VrCOL2 plasmid was constructed. The VrCOL2 CDS was amplified from the cDNA of the sequenced mungbean variety VC1973A using primers with XhoI and XbaI digestion site sequences.
FIGURE 3 | Evolutionary relationships among VrCOL proteins and COL proteins from other species. The amino acid sequences of COL proteins from Arabidopsis, soybean, Medicago, mungbean, rice and maize were used to construct a phylogenetic tree in MEGA 7.0 with the neighbor-joining method. VrCOL proteins are grouped into three classes and indicated with different colors.
The resulting PCR fragment was digested by the restriction endonucleases XhoI and XbaI to generate sticky ends. The pPTN1171 vector was digested with XhoI and XbaI to generate a linearized plasmid (Ping et al., 2014). Then the VrCOL2 and pPTN1171 fragments were ligated using T4 DNA ligase (Promega). The constructed plasmid was verified by sequencing. It was then introduced into Arabidopsis using the floral dip method (Bent, 2006), and successful transformation was confirmed by PCR. All primers are listed in Supplementary Table 1.

RNA Extraction and Transcription Analysis
RNA isolation and quantitative real-time PCR (qRT-PCR) analysis were carried out as described in Li et al. (2019).
Gene expression levels were normalized to an Actin gene from mungbean (Vradi03g00210) . Each sample was analyzed using three biological replicates. All primers are listed in Supplementary Table 1.

Identification of VrCOL Genes in Mungbean
To search for mungbean VrCOL genes, we first identified mungbean proteins that contained BBX domains. The amino acid sequences of the conserved BBX domain (PF00643) and of Arabidopsis BBX proteins were used as blast queries against the mungbean genome database at NCBI. The presence of conserved BBX domains in each candidate mungbean gene was confirmed using Pfam and InterPro software, and a total of 31 VrBBX genes were identified in the mungbean genome (Figure 1). Among the VrBBX proteins, 17 contained only BBX domains, and 14 contained both BBX and CCT domains. The latter were designated VrCOL proteins (Figure 1 and Table 1). We then analyzed the numbers and types of BBX and CCT domains in the VrCOL proteins, and found two distinct BBX domains (BBX1 and BBX2) and one CCT domain (Supplementary Figure 1). Sequence logos of the BBX1 (CX 2 CX 8 CX 4 AXLCX 2 CDX 3 HX 8 HXR), BBX2 (CX 2 CX 4 AX 3 CX 7 CX 2 CDX 3 HX 8 H), and CCT (RYX 2 KX 3 RX 3 KX 2 RYX 2 RKX 2 AX 2 RXR) domains were produced using WebLogo (Figure 2 and Supplementary Figure 1). Nine VrCOL proteins contained one BBX1, one BBX2, and one CCT domain, and five VrCOL proteins contained one BBX1 and one CCT domain (Figure 1 and Table 1).
Multiple characteristics of the VrCOL members were analyzed based on their genomic and protein sequences ( Table 1). The genomic lengths of VrCOL genes ranged from 1,506 (XP_014502470) to 14,007 bp (XP_014523701), the CDS lengths ranged from 933 (XP_014502470) to 1,329 bp (XP_022637309), and the amino acid numbers ranged from 310 to 442. The isoelectric points of VrCOL proteins varied from 4.86 (XP_014523701) to 9.22 (XP_014523547), and their molecular weights ranged from 33,756.82 Da (XP_014502470) to 48,806.9 Da (XP_022637309). The GC content, which influences gene stability to some degree, ranged from 34.64 to 50.39%, and 12 of the 14 VrCOL genes had lower than 50% GC content ( Table 1).

Phylogenetic Analysis of the VrCOL Proteins
To analyze the evolutionary relationships among the VrCOL genes and obtain information from well-studied CO homologs in other species, a phylogenetic tree was constructed using 17 Arabidopsis, 26 soybean, 11 Medicago, 16 rice, 18 maize, and 14 mungbean CO and COL proteins (Gangappa and Botto, 2014;Wu et al., 2014Wu et al., , 2017Hu et al., 2018). The VrCOL genes were named VrCOL1 to VrCOL13 based on their phylogenetic relationships with their soybean orthologs (Figure 3 and Table 1). The COL proteins were grouped into three classes based on their phylogenetic relationships (Khanna et al., 2009;Gangappa and Botto, 2014; Figure 3). Classes I, II, and III contained 4, 6, and 4 VrCOL members, respectively (Figure 3). The BBX1 and BBX2 domains were located close to one another in the class I and II proteins, with the exception of VrCOL10 (Figure 1), whereas class III proteins contained only one BBX domain (Figures 1, 3). Among these VrCOL members, VrCOL1 and VrCOL2 showed close relationships to Arabidopsis AtCO, soybean GmCOL1a, GmCOL1b, GmCOL2a, and GmCOL2b and rice OsHd1 (OsCOL-A), all of which have documented roles in the regulation of flowering time (Khanna et al., 2009;Gangappa and Botto, 2014;Wu et al., 2014). This result suggests that VrCOL1 and VrCOL2 may play critical roles in the flowering time regulation of mungbean.

Gene Structures and Conserved Motifs of the VrCOL Genes
To investigate the gene structures of the VrCOL genes, we downloaded their genomic and CDS sequences from NCBI and analyzed them using the GSDS program (Hu et al., 2015). All the VrCOL members contained 5 UTR and 3 UTR regions. Their exon numbers ranged from two to six, and their intron numbers ranged from one to six. All the group I and III VrCOL members contained two exons and one intron (Figure 4). By contrast, group II members contained various numbers of exons (3-6) and introns (2-6), suggesting potential functional diversity among these genes (Figure 4). To further investigate the conservation and diversity of VrCOL protein structures, we analyzed putative protein motifs in the VrCOLs. A total of 17 distinct motifs were identified, and all VrCOL proteins contained motifs 1 and 2, which appeared to represent the conserved BBX1 and CCT domains, respectively (Figure 4 and  Supplementary Figure 2). Most members of the same class shared some conserved motifs. For example, class I proteins shared motifs 1, 2, 3, 9, and 16, class II members shared motifs 1, 2, 3, and 5, and class III members shared motifs 1, 2, 4, 8, 12, and 13 (Figure 4).

Chromosomal Distribution and Duplication Analysis of the VrCOL Genes
Some genes have evolved from common ancestors, and the chromosomal locations of COL genes may provide insight into changes in gene distribution during evolution. To visualize the chromosomal locations of the VrCOL genes, we mapped them to their physical positions in the mungbean genome. VrCOL7b was discarded due to a lack of related positional information. Seven of the 14 VrCOL genes were located on the positive strand. Seven of the 11 mungbean chromosomes contained VrCOL genes, with the exception of chromosomes 2, 9, 10, and 11 ( Figure 5 and Table 1). Chromosome 5 contained the greatest number of VrCOL genes (three), followed by chromosomes 1, 4, 7, and 8, with two genes on each. In addition, most of the VrCOL genes were located on the relatively long chromosomes (1, 5, 6, 7, and 8). Only three members (VrCOL1, VrCOL11, and VrCOL13) were located on the relatively short chromosomes 3 and 4 ( Figure 5).
Mungbean has experienced one round of whole-genome duplication that produced many duplicated gene pairs (Kang et al., 2014;Li et al., 2019). To investigate the evolutionary relationships among the VrCOLs, we searched for duplicated gene pairs among them. Two interchromosomal duplication events were identified in chromosomes 1, 4, 5, and 6, producing the duplicated gene pairs VrCOL1/VrCOL2 and VrCOL8/VrCOL9 (Figure 6). The duplicated genes were clustered together in the phylogenetic tree (Figure 1). All the duplicated genes contained one BBX1, one BBX2 and one CCT domain and belonged to groups I and II; no duplicated gene pairs were found in group III. The duplicated genes VrCOL1 and VrCOL2 showed similar exon-intron organization and similar motifs, as did VrCOL8 and VrCOL9 (Figure 4), indicating that the duplicates may share similar functions.

Cis-Acting Element Analysis of the VrCOL Promoter Regions
To predict the potential expression responses of VrCOL genes, we investigated the cis-acting elements in their promoters using PlantCARE (Lescot et al., 2002). A total of 82 cis-acting elements were found across the 14 VrCOL promoter regions (2 kb upstream of the initiation codon) (Supplementary Table 2). Forty-five of them had predicted functions, including six development-related elements, four environmental-stress-related elements, three site-bindingrelated elements, nine hormone-responsive elements, three promoter-related elements and twenty light-responsive elements ( Table 2 and Supplementary Table 2). The various VrCOL promoter regions had different numbers and types of cis-acting elements, highlighting the functional diversity of these genes. All VrCOL promoters contained hormone-responsive elements, light-responsive elements and promoter-related elements. Light-responsive elements were the most abundant element in each VrCOL promoter, with the exception of VrCOL8 ( Table 2), indicating that VrCOL genes may play critical roles in lightdependent signaling pathways. Environmental-stress-related elements were the most abundant element in the VrCOL8 promoter (nine elements), indicating that VrCOL8 may function in stress response ( Table 2). All the VrCOL genes contained the promoter-related elements CAAT-Box and TATA-Box, which are basic promoter components. Thirteen of the 14 VrCOLs contained the hormone-responsive elements CGTCA-motif and TGACG-motif and the light-responsive element Box 4 (Supplementary Table 2), suggesting potential functions of these genes in related signaling pathways.

Transcription Patterns of VrCOL Genes in Different Tissues
To shed light on the potential functions of VrCOL genes during plant development, we analyzed the expression of VrCOL genes in different tissues, including roots, nodule roots, shoot apices, stems, leaves, flowers, pods and seeds. VrCOL genes showed distinct expression patterns in different tissues (Figure 7). For example, VrCOL3 was highly expressed in all tissues examined, whereas VrCOL2 and VrCOL7a showed low expression in most tissues. Some genes were expressed at high levels in specific tissues, suggesting that they may have critical functions in those tissues. For example, VrCOL6 showed high expression in leaves but low expression in nodule roots and flowers.
Duplicated genes may retain some common functions and evolve some new functions (Kondrashov et al., 2002;Wang et al., 2015). To investigate the conservation and diversity of duplicated genes, we also analyzed their tissue-specific expression patterns. VrCOL1 and VrCOL2 differed in their expression levels across all tissues examined, indicating that they may have different responses to the environment in these tissues. VrCOL8 and VrCOL9 showed similar expression levels in roots and nodule roots but different expression levels in other tissues (Figure 7 and  Supplementary Figure 3). FIGURE 7 | Relative expression levels of VrCOL genes in different tissues. The expression levels of VrCOL genes were analyzed by qRT-PCR. The expression level of VrCOL1 in flowers was set to 1, and other values were adjusted accordingly. The gene expression results were visualized using a heatmap generated with Multiple Experiment Viewer 4.9.0 (Saeed et al., 2003). Different colors in the heatmap indicate different expression levels.

Diurnal Rhythm of VrCOL Gene Expression
In Arabidopsis, the expression levels of CO, COL1, and COL2 are regulated by the circadian clock and show diurnal oscillations (Suárez-López et al., 2001;Gangappa and Botto, 2014). We therefore investigated whether VrCOL genes exhibited diurnal expression rhythms in mungbean leaves under LD and SD conditions. Gene expression analysis revealed that VrCOL4, VrCOL6, VrCOL12, and VrCOL13 showed daily oscillations under both LD and SD conditions, whereas VrCOL1, VrCOL2, VrCOL5, VrCOL7a, VrCOL7b, VrCOL10, and VrCOL11 showed daily oscillations only under SD conditions (Figure 8). The duplicated genes VrCOL8 and VrCOL9 exhibited similar expression patterns under both LD and SD conditions, whereas VrCOL1 and VrCOL2 showed distinct expression patterns under both LD and SD throughout the day (Figure 8).

Overexpression of VrCOL2 Accelerates Flowering in Arabidopsis Under SD Conditions
VrCOL1 and VrCOL2 displayed close phylogenetic relationships with AtCO (Figure 1), and the amino acid sequences of VrCOL1 and VrCOL2 showed 49.35 and 50.93% similarities with AtCO, respectively. We speculated that VrCOL1 and VrCOL2 might influence flowering time in mungbean, and we therefore first analyzed the function of VrCOL2 in the regulation of flowering time in Arabidopsis in this study. To investigate the potential functions of VrCOL2 in flowering time regulation, VrCOL2 was transformed into Arabidopsis under the control of the 35S promoter. The empty vector was also transformed into Arabidopsis, and the transgenic plants showed no differences from wild type under both LD and SD conditions (Supplementary Figure 4). The VrCOL2 transgenic Arabidopsis lines showed high FIGURE 8 | Relative expressions of VrCOLs in mungbean leaves throughout the day under SD and LD conditions. The SD condition was set as 8:00 am-6:00 pm light, 6:00 pm-8:00 am dark; the LD condition was set as 8:00 am-0:00 am light, 0:00 am-8:00 am dark. ZT, Zeitgeber Time. Expression level of VrCOLs was normalized to an Actin gene from mungbean. The gray and red lines indicate VrCOL expression levels under LD and SD conditions, respectively. Each sample was analyzed using three biological replicates.
levels of VrCOL2 expression (Supplementary Figure 5). The wild-type Arabidopsis plants and three VrCOL2 overexpression lines exhibited approximately 14 rosette leaves after bolting under LD conditions, indicating that they had similar flowering times. By contrast, the wildtype Arabidopsis plants and three VrCOL2 overexpression lines showed approximately 47, 34, 34, and 31 rosette leaves after bolting under SD conditions, suggesting that VrCOL2 transgenic plants had earlier flowering times than wild-type plants (Figure 9). These results indicated that VrCOL2 regulated flowering time through a photoperiod-dependent pathway.
AtFT and AtTSF accelerate flowering and are regulated by AtCO in Arabidopsis (Khanna et al., 2009;Gangappa and Botto, 2014), and we therefore investigated the expression of AtFT and AtTSF in wild-type and VrCOL2 transgenic plants under LD and SD conditions throughout the day. AtFT and AtTSF showed similar expression levels in VrCOL2 transgenic and wild-type plants in both light and dark conditions under LD treatment. By contrast, AtFT and AtTSF showed higher expression levels in VrCOL2 transgenic plants at several time points than in wildtype plants under SD conditions (Figure 9). These results further support the conclusion that VrCOL2 is involved in flowering time regulation under SD conditions.

DISCUSSION
In recent decades, the investigation of CO and COL genes in many plant species has greatly increased our knowledge about the molecular mechanisms of flowering time regulation, stress response and root development (Khanna et al., 2009;Gangappa and Botto, 2014). Mungbean is a globally important legume crop, and the mechanisms of its flowering time regulation are still largely unknown. In this study, we identified and characterized 14 VrCOL genes from the mungbean genome and investigated the function of VrCOL2 in flowering time regulation.
Plant genome evolution produces many duplicated gene pairs and provides resources for new gene functions (Kondrashov et al., 2002). Two duplicated gene pairs, VrCOL1/VrCOL2 and VrCOL8/VrCOL9 (Figure 6), were found among the mungbean VrCOLs. The duplicated genes showed close relationships in the phylogenetic tree and contained similar motifs (Figures 1, 4), indicating that they evolved from the same origin and may share similar functions. However, the duplicated gene pairs contained different numbers and types of cis-acting elements in their promoter regions and exhibited different expression levels in some tissues (Figure 7 and Table 2), suggesting that they might have evolved novel functions compared with their original gene. For example, VrCOL8 and VrCOL9 shared similar numbers of several cis-acting elements in their promoter regions, including promoter-related elements and site-binding related elements, but differed in the numbers of development-related elements, environmental-stress-related elements, hormoneresponsive elements and light-responsive elements ( Table 2  and Supplementary Table 2). VrCOL8 and VrCOL9 showed similar expression levels in roots and nodule roots, but their expression differed in flowers, pods, leaves, seeds, stems, and shoot apices (Figure 7 and Supplementary Figure 3). This result suggests that they may have retained some common functions from the original gene in roots and nodule roots but evolved novel functions in other tissues.
The expression of VrCOL genes in different tissues provides clues to their potential functions, and many VrCOL genes (such as VrCOL6 and VrCOL12) showed tissue-specific expression patterns (Figure 7). However, several VrCOL genes (including VrCOL2, VrCOL7a, and VrCOL10) showed low expression levels in all tissues tested, despite the fact that their promoter regions contained many cis-acting elements (Figure 7, Table 2, and Supplementary Table 2). Gene expression is influenced by many factors. For example, many circadian clock and flowering time regulation genes are controlled by photoperiod. Their expression changes under different photoperiods and during the day and night (Suárez-López et al., 2001;Jack, 2004;Wickland and Hanzawa, 2015;Xu and Chong, 2018). For example, VrCOL2 appeared to be a daily oscillation gene whose expression changed during the day under SD conditions but was low throughout the day under LD conditions (Figure 8). The different fieldgrown mungbean tissues were collected in the afternoon under relatively LD conditions in July, and that may explain why VrCOL2 showed low expression levels in the tissue expression analysis (Figure 7).
CO and CO-homologous genes, such as OsHd1, play critical roles in flowering time regulation (Khanna et al., 2009;Gangappa and Botto, 2014). VrCOL2 showed close relationships with Arabidopsis CO, soybean GmCOL1a, GmCOL1b, GmCOL2a, and GmCOL2b and rice OsHd1 (OsCOL-A), and accelerated flowering under SD but not LD conditions in transgenic Arabidopsis lines (Figure 9). AtCO regulates AtFT and AtTSF to accelerate flowering (Putterill et al., 1993(Putterill et al., , 1995Andres and Coupland, 2012;Song et al., 2013), and the expression of AtFT and AtTSF increased in VrCOL2 transgenic Arabidopsis lines at several time points under SD but not LD conditions (Figure 9). Moreover, VrCOL2 showed daily oscillations only under SD conditions, but not LD conditions (Figure 8), indicating that VrCOL2 might only have functions under SD conditions. VrCOL2 therefore affects the expression of downstream AtFT and AtTSF genes via photoperiod-dependent pathways. Moreover, AtCO protein accumulation is also regulated by the circadian clock. AtCO mRNA is highly abundant from late afternoon to dawn, but AtCO protein accumulates only in the late afternoon under LD conditions (Putterill et al., 1995;Shim and Imaizumi, 2015;Song et al., 2015;Shim et al., 2017). Although VrCOL2 was controlled by the 35S promoter and expressed under both LD and SD conditions, the accumulation of VrCOL2 proteins was unknown. Whether the accumulation of VrCOL2 protein depends on day length, in turn affecting flowering time by influencing AtFT and AtTSF expression requires further investigation. In addition, AtCO promotes flowering under LD conditions and suppresses flowering time under SD conditions (Luccioni et al., 2019), but rice OsHd1 accelerates flowering under SD conditions and delays flowering under LD conditions (Yano et al., 2000;Komiya et al., 2008Komiya et al., , 2009. Mungbean (Imrie, 1996;Kim et al., 2015) and rice are SD plants, and Arabidopsis is an LD plant, and this may explain why CO homologs have different functions in different plant species. These results suggest that CO and its homologs are involved in flowering time regulation under photoperiod-dependent pathways and have distinct roles in different plant species. Thus, in summer LD conditions, the expression of VrCOL2 may be low and have little effect on the acceleration of flowering. In the autumn, as days become shorter, the expression of VrCOL2 may increase and accelerate mungbean flowering. In addition, VrCOL1 and VrCOL2 form a duplicated gene pair and show a close relationship with one another (Figures 1, 6), and VrCOL1 showed high expression levels in many tissues, indicating that VrCOL1 may share similar functions to VrCOL2 in flowering time regulation, a possibility that requires further investigation. Much more work is needed to fully elucidate the mechanisms by which VrCOL2 affects flowering time and circadian clock regulation in mungbean.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
SL conceived and designed the research. CL, QZ, HZ, and CC conducted the experiments and analyzed the data. SL and HZ wrote the manuscript. All authors read and approved the manuscript.

ACKNOWLEDGMENTS
We thank Suk-Ha Lee at Seoul National University, Seoul, South Korea, for supplying mungbean VC1973A seeds. This manuscript has been released as a pre-print at Researchsquare, (Liu et al., 2020).