An F₂ population comprising 849 individuals derived from a cross between “V2709” (purple hypocotyl) and “Sulv1” (green hypocotyl) was used for gene mapping of the VrP gene. The F₂ population together with their parents was planted in growth chambers with a 14/10 h photoperiod, 250 μE m^–2 s^–1 light intensity, 28°C daytime temperature, and 25°C nighttime temperature. A set of 124 mungbean accessions (Supplementary Table 1) was used to investigate the association between VrP sequence variations and hypocotyl colors.

An amount of 50 mg of different mungbean or Arabidopsis tissue samples were ground into powder in liquid nitrogen, followed by soaking in 1 ml of 1% (v/v) HCl/methanol in the dark for 12 h. The mixture solution was centrifuged at 12,000 × g for 10 min. The supernatant was measured for light absorption at 530 and 657 nm by using an ultraviolet spectrophotometer UV-7500 (Shimadzu, Kyoto, Japan). The total anthocyanin content was calculated as described previously (Weiss and Halevy, 1989). The quantification was replicated five times. In the mungbean, the quantification was done for only Sulv1, V2709, and Guilv1.

Sections of 1 cm of the hypocotyl of a 7-day-old seedling of V2709 and Sulv1 were obtained and embedded in 3% agarose. The sections were made using a double-edge razor blade by hand. Ten sections were placed on a slide with dH₂O. Images were observed and taken under Olympus BX51 light microscope (Olympus, Tokyo, Japan).

The seedlings of the F₂ mapping population were recorded for hypocotyl color (purple or green) 2 weeks after sowing, and genomic DNA was extracted from each F₂ progeny. DNA extraction, primer design, and DNA marker genotyping were the same as described in our previous study (Lin et al., 2020). Briefly, the DNA was extracted using a modified CTAB method. Primers for simple sequence repeats (SSRs) and Insertions/Deletions (InDels; Supplementary Table 2) were designed from all the chromosomes of the mungbean reference sequence “VC1973A” (Kang et al., 2014). In the case of the InDel markers, primers were developed by comparing sequences of Sulv1 (Yan et al., 2020) with the reference sequence. The SSR and InDel markers were screened for polymorphism between the parents. The polymerase chain reaction (PCR) contained 20 ng of DNA template, 10 mM Tris–HCl, 50 mM KCl, 1.5 mM MgCl₂, 50 mM dNTP, 0.2 mM mixed primers, and 0.5 U Taq DNA polymerase. The PCR program consisted of a denaturation step (94°C/5 min), followed by 33 cycles of 94°C/30 s, 55°C/30 s, and 72°C/60 s, and a final extension step of 72°C/5 min. The PCR products were electrophoresed using 8% polyacrylamide gels and DNA bands were visualized by silver staining.

In this study, the green hypocotyl was a recessive trait, thus we considered it the mutant type, while the purple hypocotyl was considered the wild type, thus the hypocotyl color was used as a phenotypic marker. In the F₂ population, the phenotypic marker, together with 17 molecular markers, including 8 SSR markers and 9 InDel markers (Supplementary Table 2), was used to construct linkage map. The linkage map was constructed using the QTL IciMapping 4.2 software (Meng et al., 2015). A logarithm of odds (LOD) value of 3.0 was used for grouping the markers. Markers were ordered using a recombination counting and ordering algorithm (RECORD) function (van Os et al., 2005). The genetic map distance was calculated using Kosambi’s mapping function (Kosambi, 1944).

To finely map the VrP locus, we initially identified markers associated with this locus by analyzing 10 F₂ mutant-type (green hypocotyl) individuals using the markers showing polymorphism between their parents (Supplementary Table 2). Markers revealing the same homozygous genotype between all the 10 individuals and their green-hypocotyl parents indicated linkage between the markers and the VrP locus. All the markers identifying linkage with the VrP locus were used to genotype the whole F₂ population. Then, the markers were mapped to the mungbean reference genome sequence (Kang et al., 2014), and the location of VrP was identified by examining marker recombination events in the F₂ population.

Genes located between markers flanking the VrP locus were identified as candidate genes using the mungbean reference genome sequence annotated by the National Center for Biotechnology Information (NCBI Vigna radiata Annotation Release 101)¹.

We determined the phylogeny of VrMYB90. Protein sequences from azuki bean [Vigna angularis (Ohwi) Ohwi and Ohashi], cowpea [Vigna unguiculata (L.) Walp.], soybean (Glycine max L. Merr.), petunia (Petunia bajeensis T. Ando and Hashim), barrel medic (Medicago truncatula Gaertn.), lupin (Lupinus angustifolius L.), kiwifruit (Actinidia chinensis A. Chev), apple (Malus × domestica Borkh.), and Arabidopsis, matching that of VrMYB90, were identified using the BLASTP algorithm and were downloaded from GenBank. A rooted phylogenetic tree was constructed from the sequences using the neighbor-joining method using the MEGA 10.0 software (Kumar et al., 2018). The robustness of the phylogeny was tested by bootstrapping (1,000 replicates) with the DNAMAN5.2 software (Lynnon Biosoft Corp., Quebec City, QC, Canada).

Eight open-reading frames (ORFs)/genes were identified in a genomic region covering the VrP locus (Supplementary Table 3). Coding sequences (CDS) of the eight ORFs in V2709 and Sulv1 were sequenced. In addition, the full ORF of the VrMYB90 and its 2.56-kb promoter region in V2709, Sulv1, Guilv1 (purple hypocotyl), and KPS2 (green hypocotyl) were sequenced. PCR amplification and Sanger sequencing were performed as per the methods described by Yun et al. (2002). In brief, PCR was conducted using KOD-plus DNA polymerase (Toyobo, Shanghai, China). DNA bands with the expected size were purified using (E.Z.N.A.)^® Gel Extraction Kit (Omega Bio-tek, Norcross, GA, United States) and then sequenced using an ABI 3730xl DNA Analyzer (Applied Biosystems, CA, United States) by TSINGKE Biotechnology Co., Ltd. (Nanjing, China).

The coding region without the termination codon of VrMYB90 from Guilv1 was amplified by polymerase chain reaction (PCR) and recombined into the vector pCAMBIA1305.1-GFP. The constructed 35S::VrMYB90-GFP was used in the following experiments.

Agrobacterium tumefaciens strain EHA105 containing the construct 35S::VrMYB90 was infiltrated into the 6-week-old tobacco (Nicotiana benthamiana L.) leaf for transient transformation. The transient expression assay was performed as previously described (Yoo et al., 2007). To observe the color change, the infiltrated plants were placed in an artificial illumination incubator for 4–5 days and photographed using a camera.

Vector 35S::VrMYB90 was introduced into the Agrobacterium rhizogenes K599 strain. Notably, 1-week-old seedlings of Guilv1 were used as receptors. The hairy root transformation was performed as previously described (Kereszt et al., 2007).

The plasmid was transformed into the Agrobacterium tumefaciens EHA105 strain, and an A. thaliana transformation was performed using the floral dipping method as described previously (Clough and Bent, 1998). The seeds of transgenic plants were selected on 1/2 Murashige and Skoog media containing 30 μg/ml hygromycin.

Total RNA was extracted using RNAprep Pure Plant Kit (Tiangen, Beijing, China) following the manufacturer’s instructions. The RNAs from hypocotyls of various mungbean varieties were used for expression analysis of VrMYB90. The RNAs from different tissues of Guilv1 were used for the expression pattern investigation of VrMYB90. The RNAs from hypocotyls of Guilv1 and transgenic Arabidopsis lines were used for expression analysis of anthocyanin-related genes by quantitative real-time PCR (qRT-PCR). First-strand cDNA was synthesized using the Goldenstar^® RT6 cDNA Synthesis Kit (Tsingke, Beijing, China). VrActin3 (LOC106757568) gene was used as an endogenous control. Primers used for the qRT-PCR are listed in Supplementary Table 2. The qRT-PCR was carried out according to Lin et al. (2020). The 2^–ΔΔCT method was used to calculate relative levels of gene expression.

All values and data points presented were means ± standard deviation (SD). All statistically significant difference tests in this study were conducted using an unpaired two-tailed Student’s t-test.

Microscopic observation showed that anthocyanins were accumulated in the second epidermal cell layer that gave rise to a purplish hypocotyl in V2709, and no anthocyanin accumulation was observed in the hypocotyl of Sulv1 (Figures 1A–D). Quantification of anthocyanins revealed that the total anthocyanin contents in the hypocotyl of V2709 were approximately eight times higher than that of Sulv1 (Figure 1E). A color investigation of the hypocotyl in the F₂ population showed that the segregation ratio of purple individuals to green individuals was consistently 3:1 (χ² = 0.55 < χ2 0.05 = 3.84, P = 0.46). This result was in line with previous studies (Swindell and Poehlman, 1978; Pandey et al., 1989) and indicated that the purple hypocotyl color is a dominant trait controlled by a single locus, designated as VrP.

Linkage analysis using 132 F2 plants mapped the VrP locus to the region between markers PS36 and LG4-15 at a genetic distance of 3.0 and 0.76 cM, respectively (Figure 2A). With all the plants in the population (849 plants) and more markers, the location of the VrP locus was precisely mapped between markers WS-28 and WS-43 on chromosome 4 of the mungbean reference genome (Figure 2B).

With more polymorphemic markers explored in the population, the VrP locus was narrowed down to a 78.9-kb region, delimited by SSR markers WS-2 and WS-16 (Figure 2B). Based on gene annotation of the mungbean reference sequence by NCBI, there were eight open reading frames (ORFs) predicted in this 78.9-kb region (Figure 2B and Supplementary Table 3). Sequencing and alignment of the CDS of all the eight ORFs between Sulv1 and V2709 revealed polymorphism only in the ORF3.

A qRT-PCR experiment showed that ORF1 and ORF2 were almost undetectable in both accessions; ORF5, 6, and 8 have a slight expression difference in the hypocotyls of Sulv1 and V2709; ORF7 was expressed significantly higher in Sulv1 than in V2709, while ORF3 was highly expressed in the hypocotyl of V2709 and undetectable in that of Sulv1 (Figure 2C). According to NCBI annotation, ORF3 was composed of three exons with a full CDS of 702 bp in length and encoded an R2R3-MYB transcription factor, VrMYB90. The ORF3 was named “VrMYB90.” The VrMYB90 is highly homologous to AtMYB90 (AtPAP2) in Arabidopsis (Zimmermann et al., 2004) and AcMYB1 in Actinidia chinensis Planch (Wang et al., 2019; Supplementary Figure 1B), which forms the MBW complex and promotes the biosynthesis of anthocyanins. Thus, VrMYB90 was selected as the candidate gene for the VrP locus.

Since the transcript of VrMYB90 was almost undetectable in the hypocotyl of Sulv1 (Figure 2C), we then investigated the expression level of VrMYB90 in four green- and four purple-hypocotyl mungbean cultivars. The expression of VrMYB90 was nearly nil in all the green-hypocotyl mungbean accessions (KPS1, CN60, CN84, and V2817), but very high in all the accessions with purple hypocotyls (Jilv7, Guilv1, NM10-12, and KUML8) (Figure 3A). These results indicate that the altered expression of VrMYB90 may be responsible for the variation in the anthocyanin content of hypocotyl in mungbeans.

To explain the difference in the expression of VrMYB90, amplified products of the genomic sequence (ATG to TAA) and a 2.56-kb promoter region of VrMYB90 in four mungbean accessions were sequenced. Sequence alignment revealed numerous SNPs and InDels in the promoter region of the four mungbeans (Supplementary Figure 1A); however, none of those polymorphisms was associated with the phenotype. VrMYB90 from Sulv1 was the same as the reference sequence (VC1973A; green hypocotyl). VrMYB90 from KPS2 had the same exonic sequence as Sulv1 with the only difference between them being TA repeats in intron 2 (Figure 3B). Compared with Sulv1, 21 SNPs and a 70-bp deletion existed in VrMYB90 coding region of V2709, while 13 SNPs and a 42-bp insertion were found in that of Guilv1 (Figure 3B). Among these nucleotide differences, the 13 SNPs in exon 3 were unique in both V2709 and Guilv1. These results suggest that the 13 nucleotide differences in exon 3 of VrMYB90 are associated with hypocotyl colors in mungbean.

To confirm that VrMYB90 has a similar function as AtMYB90 in the regulation of anthocyanin biosynthesis, we overexpressed the CDS of VrMYB90 from V2709 in different systems. In tobacco leaf, the transient expression of VrMYB90 led to anthocyanin accumulation in the infection position, while the part injected with an empty 35S::GFP vector showed no color change (Figure 4A). In the mungbean hairy-root transgenic system, ectopic expression of VrMYB90 induced anthocyanin accumulation, resulting in a dark purple root, while the mungbean injected by K599 carrying the 35S::GFP empty vector grew a white root (Figures 4B,C).

After functional verification in transient expression systems, the heterologous expression of VrMYB90 in Arabidopsis was carried out to determine its function in a stable genetic transformation system. The results show that the T₂-positive transgenic lines exhibited purple leaves, purple stems, and dark brown seeds in the T₂ generation (Figures 4D–G). These results confirm that VrMYB90 promotes anthocyanin accumulation in different transgenic plants.

Our sequencing results of VrMYB90 from four parental mungbeans suggested that 13 SNPs in exon 3 were associated with the expression levels (Figures 3A,B). To corroborate the correlation between hypocotyl color and VrMYB90, we analyzed CDS variations of VrMYB90 (Figure 5A) in 124 mungbean accessions by Sanger sequencing. Phenotypic analysis of these 124 varieties showed that 40 accessions (32.23%) possessed green hypocotyl and 84 accessions (63.07%) possessed purple hypocotyl (Supplementary Table 1).

The 124 accessions were genotypically classified into five groups of haplotypes based on natural variations, including 21 SNPs and one 3-bp InDel (Figure 5B). All the accessions showing green hypocotyls shared the same haplotype, Hap1, while the accessions displaying purple hypocotyls belonged to either haplotypes Hap2, Hap3, Hap4, or Hap5 (Figure 5C). Among the four haplotypes of purple hypocotyl accessions, Hap3 was the most prevalent haplotype at a frequency of 70.16%, while the haplotypes Hap2, Hap3, and Hap4 were rare, with each haplotype having only one accession. Comparing the Hap1 with other Haps revealed 10 SNPs in exon 3 (positions 456, 564, 573, 574, 603, 649, 657, 661, 666, and 683) associated with the phenotype (Figure 5B). Based on these results, we deduced that these ten SNPs cause the repression of the VrMYB90 gene.

Based on color appearance, we noticed differences in anthocyanin accumulation in different mungbean organs. Anthocyanin accumulation was observed in the hypocotyl, stem, petiole, and mature flower, while a few anthocyanins were accumulated in non-purple organs such as the root, leaf, pod, and seed (Figures 6A–G). qRT-PCR analysis revealed that VrMYB90 showed a great expression level in the hypocotyl and petiole, a low expression level in the stem and flower, and no expression in the leaf and root (Figure 6I). In addition, the expression level appeared to be positively correlated with the anthocyanin content; both the expression level and anthocyanin content were highest in the hypocotyl, followed by the stem and petiole, respectively (Figures 6H,I). Taken together, these results show that VrMYB90 controls anthocyanin biosynthesis in different organs/tissues, except for the seeds and roots, and the expression level is tightly correlated with a purple color appearance in mungbean.

Pigment accumulation is sensitive to environmental factors, such as light, temperature, and drought, and the anthocyanin content varies greatly depending on the climate and external stresses (Yan et al., 2021). In this study, we observed that the anthocyanin completely disappeared when mungbeans were grown in the dark (Supplementary Figure 2A), and the difference in expression of VrMYB90 was greatly reduced in the dark (Supplementary Figure 2B). However, when the mungbeans were transferred and exposed to light, VrMYB90 started to express after 2 h of light exposure, and subsequently, anthocyanin accumulation appeared on the surface of the hypocotyl. The expression level was increased constantly for hours and maintained at a high level (Supplementary Figure 2C). These results show that VrMYB90 was expressed in several organs of mungbean and was sensitive to light treatment, suggesting that the expression of VrMYB90 is regulated spatiotemporally.

Expression analysis demonstrated that VrMYB90 was expressed at a substantially higher level in the hypocotyl than in other organs (Figure 6I). Thus, the RNA samples from the hypocotyl of Guilv1 were prepared for expression analysis of the structural genes. BLAST against the mungbean reference genome (Kang et al., 2014) showed that there are multiple genes encoding different structural genes involved in anthocyanin biosynthesis, including VrCHI (LOC106761096 and LOC106762868), VrF3H (LOC106769061 and LOC106763868), VrF5′H (LOC106775152 and LOC106767100), VrDFR (LOC106763833 and LOC106773575), VrANS (LOC106757264, LOC106773578, LOC106771626, and LOC106779041), VrLDOX (LOC106777796 and LOC106776917), and VrUFGT (LOC106778549, LOC106778555, LOC106778578, LOC106779200, and LOC106780203). The expression analysis of these genes demonstrated that most of the genes had no significant difference in expression, while an F3′H gene (LOC106769061), a DFR gene (LOC106763833), two LDOX genes (LOC106777796 and LOC106776917), and a UFGT gene (LOC106779200) showed significantly higher expression in Guilv1 than in Sulv1 (Figures 7A,B). F3′H belongs to the EBG group, while DFR, LDOX, and UFGT belong to the LBG group. Among these genes, VrDFR-1 (LOC106763833) and VrLDOX-1 (LOC106777796) had a notable expression difference. DFR catalyzes the conversion of dihydroflavonols to leucoanthocyanidins (Meng et al., 2021), and LDOX catalyzes the conversion of leucoanthocyanidins into anthocyanidins (Abrahams et al., 2003), both of which are crucial steps in the biosynthesis of anthocyanins. Additionally, the expression of structural genes in transgenic Arabidopsis leaves was measured. The results show that all the structural genes were upregulated in T₃-positive VrMYB90 overexpression lines, compared with wild-type Arabidopsis, and AtDFR exhibited the largest expression difference (Figure 7C). These results indicate that VrMYB90 prefers to promote LBG genes in anthocyanin biosynthesis and that VrDFR-1 is the most probable target gene of VrMYB90.

Pigmentation in mungbean caused by anthocyanin accumulation was observed on the hypocotyls, stem, leaf rachis, petiole, and peduncle. Although the genetics of the coloration of the mungbean hypocotyls is simple, governed by a single dominant locus, VrP, and was reported more than six decades ago (Sen and Ghosh, 1959), the molecular basis of the gene underlying the VrP locus has never been identified and investigated. The main reason for this is the lack of reference genome information for mungbean. In this study, we exploited the draft genome sequence of mungbean and successfully identified the gene underlying the VrP locus.

In this study, we used an F₂ population to map VrP to a 78.9-kb region on chromosome 4 (Figures 2A,B), and the gene VrMYB90 was isolated and determined to be the factor for anthocyanin accumulation in mungbean hypocotyl. VrMYB90 has high homology with AtMYB90 (PAP2) and AtMYB113 in amino acid sequences (Supplementary Figure 1B) that influenced the color of Arabidopsis (Borevitz et al., 2000; Zimmermann et al., 2004). In Sulv1 and other mungbean accessions with green hypocotyls, the transcript of VrMYB90 was almost abolished, which blocked the biosynthesis of anthocyanins (Figures 2C, 3A). The introduction of VrMYB90 into different plant systems, e.g., tobacco leaf, mungbean hairy root, and Arabidopsis, triggered the expression of structural genes in the anthocyanin synthesis pathway and promoted purple pigment accumulation (Figure 4). qPCR analysis indicated that VrMYB90 upregulates the expression of anthocyanin biosynthesis genes in Arabidopsis-positive lines (Figure 7C), which implies it has a similar role to AtMYB90. Moreover, our results indicated that VrMYB90 predominated in the anthocyanin metabolism pathway of mungbean, and the content of anthocyanin positively correlates with the expression level of VrMYB90 (Figure 6). Together, this study reveals that VrMYB90 on chromosome 4 is the determiner of anthocyanin synthesis in mungbean.

Various nucleotide differences in MYB genes lead to function loss or expression silence and cause a lessened or complete absence of anthocyanin accumulation. A gypsy-like retrotransposon insertion in the promoter of Vvmyba1 inhibits the expression and turns off anthocyanin synthesis in fruit peels (Kobayashi et al., 2004). A 4-bp insertion in the first exon of RsMYB1 resulting in a frameshift mutation and premature termination of translation produced white radish cultivars (Kim et al., 2021). A 3.6–3.7 kb fragment insertion in the first intron of BrMYB2 caused the white-head trait of Chinese cabbage (Brassica rapa L.) (He et al., 2020). In this study, sequencing of a 2.6-kb promoter of eight mungbean lines showed several differences; however, none of these uniquely belonged to purple-hypocotyl lines, and thus we believe that the variations in the promoter were not correlated with the expression change of VrMYB90. Meanwhile, sequencing the CDS of VrMYB90 revealed that ten SNPs in exon 3 were tightly related to phenotype (Figure 5). The haplotype analysis revealed that the ten SNPs existed in all green-hypocotyl mungbean accessions (Supplementary Table 1) and further supported that the nucleotide variations in exon 3 caused the disappearance of the VrMYB90 transcript. Mutations of exons may happen at the target of microRNA or methylation sites, which can lead to an expression change in genes: a non-synonymous SNP mutation in the third exon of CsCEN caused a significantly lower expression in shoot apexes and axillary buds in D226, which resulted in determinate growth (Njogu et al., 2020), and an SNP in the third exon of OsSPL14 perturbed the repression of OsmiR156, resulting in a strong expression of OsSPL14 in the shoot apex, generated an ideal plant architecture and enhanced grain yield (Jiao et al., 2010). However, the specific transcriptional regulation mechanism of the ten SNPs in VrMYB90 remains unknown.

The tissue expression specificity of MYB genes is common in different species. For example, AcMYB123 expressed in the inner pericarp of kiwifruit produced the red-fleshed variety “Hongyang” (Wang et al., 2019), while an R2R3 MYB gene DEEP PURPLE was uniquely expressed in the petal of Petunia exserta J.R. Stehm. and caused diminished red color (Berardi et al., 2021). In our study, measurement of the VrMYB90 transcript in different organs showed that this gene has different expression levels in different tissues, leading to different anthocyanin contents. Notably, VrMYB90 was undetectable in mungbean seeds. Noble et al. (2018) reported that anthocyanins induce the diversity of seed coat colors in mungbean and VrMYB113, which is highly homologous to VrMYB90 and may be a determinant gene controlling seed color. In contrast to this report, the results from our study suggest that the seed coat color of some black seed mungbean accessions may be governed by another MBW complex. This is supported by the limited expression of VrMYB90 in the seeds and the imperfect correlation between the hypocotyl color and the seed coat color in the 124 mungbean accessions (Supplementary Table 1); for example, Anhei2, a mungbean accession, expressed a green hypocotyl but a black seed coat. Altogether, we believe that a complex regulatory mechanism of anthocyanin production exists in mungbean.