Three mutants, namely, mib1-1 (A001), mib1-2 (A006), and mib1-3 (I007), were identified from M₂ generation of the gamma ray mutagenized cultivar Sulu (Li et al., 2022). For phenotype analysis of wild-type (WT) plants, mutants were grown in the greenhouse at 28 ± 2°C, with a 16-h/8-h day/night photoperiod. The allelic tests for three mutants were carried out by crossing the mib1-1 mutant with the mib1-2 and mib1-3 mutants, respectively. All plants of F₁ generation showed the mutated phenotype.

The terminal leaflets of the fifth compound leaves were fixed in FAA solution and then the samples were dehydrated in the ethanol/tert-butanol series. Field emission scanning electron microscopic (SU8010, Hitachi, Tokyo, Japan) analysis was conducted as previously described (Jiao et al., 2019).

The mib1-3 mutants were crossed with cultivar AL127 to generate a population for genetic mapping. A total of 150 plants with mutant phenotype isolated from 642 plants in the F₂ population were used to map the MIB1 gene. The primers of the molecular markers used in present study are listed in Supplementary Table 1 . The DNA were extracted via a plant Genomic DNA Kit DP305 (Tiangen, Beijing, China). The polymerase chain reaction (PCR) was carried out and the polymorphisms of the markers were analyzed as previously described (Jiao et al., 2016).

The PCR of the MIB1 genomic region was conducted by the primers in Supplementary Table 1 . The PCR products were cloned into the pMD18-T (TaKaRa, Dalian, China) and sequenced.

Shoot buds (2 weeks after germination) and the young pods (2 days after pollination) of WT and mib1-3 mutants were collected with three biological replicates. RNA was extracted by the RNA Kit R6827-01 (Omega, Shanghai, China). We performed RNA-seq using the Illumina HiSeq X Ten platform (Illumina, San Diego, California, USA). The raw sequences were submitted to the NCBI SRA database with accession numbers SRR16944233–SRR16944244. Number of reads per kilobase of exon region in a gene per million mapped reads (RPKM) was used to value expression levels (Mortazavi et al., 2008), and VC1973A version 1.0 was used as the reference genome (Kang et al., 2014). Based on the methods described by Audic and Claverie (1997), DEGs were identified. Heat maps were generated by the pheatmap package (https://cran.r-project.org).

For qRT-PCR, the first strand cDNA was synthesized via Takara PrimeScript™ RT reagent Kit RR047A (TaKaRa, Dalian, China). qRT-PCR analysis was conducted using TB Green™ Premix Ex™ RR420A (TaKaRa) and the ABI StepOnePlus machine (Applied Biosystems, Foster City, CA, USA). Three biological replicates with three technical repeats were conducted.

The WT (Col-0) and dtx54 mutant (WiscDsLoxHs046_04F) were used in the present study. The CDS of the MIB1 gene was cloned into pCAMBIA1304 using primers in Supplementary Table 1 . The construct was transformed into the dtx54 mutants through floral dip transformation as previously described (Clough and Bent, 1998). T₃ progeny lines of 35S::MIB1/dtx54 (L04 and L06) were used for phenotype analysis in this study.

WT strain K12 and ΔacrB mutant strain of E. coli were obtained from Professor Chuanzhen Jiang (South China Agricultural University). The CDS of the mungbean MIB1 gene was cloned into the pET32a vector using primers in Supplementary Table 1 , and the vectors were transformed into K12 and mutant strain. Transformants were selected on Luria-Bertani (LB) plate medium with 100 μg/ml ampicillin. The positive clones were then grown in liquid medium containing ampicillin and 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) to induce the expression of MIB1. The cells were diluted and spotted on medium plates with or without tetrabutyl ammonium (TBA) at 37°C for 24 h. Cell growth curves were determined by the absorbance at 600 nm of the cultures grown at 37°C for 24 h.

Plant hormone levels of indole-3-acetic acid (IAA) and abscisic acid (ABA) in young pods of the WT plant and mutants were determined by high-performance liquid chromatography–mass spectrum/mass spectrum (HPLC/MS/MS) by Agilent 1290 HPLC (Agilent, Santa Clara, CA, USA) and SCIEX-6500 Qtrap (AB Sciex, Foster, CA, USA), as described previously (Pan et al., 2010).

In this study, the MIB1 protein sequence was used to search against the mungbean database (Kang et al., 2014). The phylogenetic analysis was conducted by MEGA (version 7.0) using the neighbor-joining method with 1,000 replications (Kumar et al., 2016). The tree was displayed by the Interactive Tree of Life (iTOL; Letunic and Bork, 2016). Protein sequences from this study are listed in Supplementary Table 2 .

To investigate key components regulating plant architecture in mungbean, we screened mutants with altered branch number and shape and size of lateral organs from the mutagenesis population (Li et al., 2022). Three allelic mutants affecting plant architecture were isolated in mungbean ( Figure 1 ). We named these mutants mini body1-1 (mib1-1), mib1-2, and mib1-3, respectively.

The leaf production rate in the mib1 mutants was accelerated, compared with that of WT ( Figures 1A, B ). The juvenile leaves of the mutants were normal, but the adult leaves displayed pentafoliate form, compared to those of WT with trifoliate compound leaves ( Figure 1C ). In the mib1 mutants, the size of the leaflets was severely reduced by 43.61%–60.93% ( Figure 1D ). The outgrowth of axillary buds in the mib1 mutants was faster than those in WT ( Figure 1E ). The number of branches in the mib1 mutants increased significantly ( Figure 1F ). At 4 weeks after germination, there was only one branch in each WT plant, while each mib1 mutant had four branches ( Figure 1F ). At 8 weeks after germination, there was no difference in the number of primary branches between WT and mutants of mib1-2 and mib1-3 (the mib1-1 mutant has about two more primary branches than WT; Supplementary Figure 1 ). However, the secondary branches in the three mib1 alleles increased significantly ( Supplementary Figure 1 ). Thus, the increased branch number in the mutants was caused by accelerated bud outgrowth and sustained branching capacity among early developed primary branches. Additionally, the mib1 mutants had a compact plant architecture, compared with WT ( Figure 1E and Supplementary Figure 1A ).

The flowers and young pods of the mib1 mutants were smaller than those of WT ( Supplementary Figure 2 ). The matured pods of the mutants were shorter, with decreased seed number and size ( Figures 2A–E ). The pod length of the mib1 mutants (6.6± 0.03, 6.7± 0.05, and 7.3 ± 0.09 cm, respectively) was decreased, compared to that of WT (9.8 ± 0.11 cm). The seed number per pod of three mib1 mutants (8.5 ± 0.11, 8.7 ± 0.20, and 9.9 ± 0.09, respectively) was much lower than that of WT (11.3 ± 0.65). Compared with the WT, mature seeds of mib1 mutants were rounder and showed significantly decreased length, width, and thickness ( Figures 2D, F ). Therefore, the 100-seed weight was decreased by 26.82%, 18.63%, and 27.42% in mib1-1, mib1-2, and mib1-3 mutants, compared with that of WT, respectively ( Figure 2E ).

The plant organ size is regulated by the coordination of two connected processes, cell division and expansion (Gonzalez et al., 2012). Microscopic examination of leaflet epidermal cells showed that the cell size decreased significantly in the mib1-3 mutants in comparison with that of WT ( Supplementary Figure 3A ). The area of epidermal cells in the mib1-3 mutants was only about half that in the WT plants ( Supplementary Figure 3B ), suggesting that MIB1 augments organ size mainly by increasing the cell size.

Genetic analysis of the mib1 mutants was conducted by backcrossing mib1-3 mutants with the WT plants. All F₁ plants were similar to WT. In the F₂ population, the WT plants and mutant plants segregated with a 3:1 ratio (87 WT plants and 25 mutants, χ ² = 0.42 < χ ² _0.05 = 3.84), indicating that mib1 was a single recessive locus.

We conducted map-based cloning to identify the MIB1 gene (Jiao et al., 2016). The MIB1 gene was preliminarily mapped on chromosome 1 of the VC1973A genome (Kang et al., 2014), linked with the markers ID244 and ID171 ( Figure 3A ). By developing new markers, the mib1 mapping region was narrowed down to a 1.71-Mb region flanked by the markers ID218 and ID201 ( Figure 3A ). Based on the functional annotation (Kang et al., 2014) and the mutant phenotype, Vradi01g10280 (LOC106766026) in the mapping region was identified as the candidate ( Figure 3B ). Sequencing of the PCR products of Vradi01g10280 from WT and mib1 mutants displayed mutations ( Figure 3B ), showing that three alleles, mib1-1, mib1-2, and mib1-3, carried different deletions (1-bp deletion, 1-bp deletion, and 21-bp deletion, respectively). qRT-PCR analysis of shoot buds (2 weeks after germination) revealed that there were decreased expression of the Vradi01g10280 gene in the mib1 mutants ( Figure 3C ).

Segregation analysis showed that 150 mutated plants out of a total of 642 individuals from the F₂ mapping population were homozygous for the 21-bp deletion in Vradi02g10020, indicating that the deletion co-segregates with the mutant phenotype. Therefore, MIB1 (Vradi01g10280) encoded a member of MATE/DTX proteins ( Figure 4 ), which was closely related to DTX54/BIGE1A in Arabidopsis and BIGE in maize (Suzuki et al., 2015), affecting plant architecture in mungbean.

The relative expression of the MIB1 gene in different tissues of mungbean was analyzed by qRT-PCR. It was found that the MIB1 gene was expressed in all tissues, with the highest expression level in the young inflorescence ( Figure 3D ).

Multiple amino acid sequence alignments of the MIB1 protein (XP_014506278.1) with its homologs indicated that it shared a conserved MatE domain ( Figure 4A ). The MIB1 protein was predicted to have 12 transmembrane domains with N-termini towards the inside of the cell ( Supplementary Figure 4 ).

We conducted a BLASTP search for sequences with homology to MIB1 to characterize the MATE/DTX family in the mungbean database (Kang et al., 2014) and found 56 MATE/DTX proteins in the mungbean genome ( Supplementary Table 2 ). These mungbean MATE/DTX proteins were classified into four groups by phylogenetic analysis with Arabidopsis MATE/DTX proteins ( Figure 4B ; Wang et al., 2016). It was found that MIB1 had two other closely related homologs in mungbean, XP_022635913.1 (Vradi05g00900) and XP_014506743.1 (Vradi07g25110, Figure 4B ).

In order to investigate the origin of MIB1 in legume plants, we identified MIB1 closed homologs from a number of public databases ( Supplementary Table 2 ). The phylogenetic tree of aligned legume DTX54 and DTX55 orthologs was constructed ( Figure 4B ). It was found that one copy encoding the ortholog to DTX54 in legume formed the LegDTX54 clade, which was distinct from the LegDTX55 clade ( Figure 4C ). In contrast, within the LegDTX55 clade, there were different copies in legume, such as two copies in adzuki bean (V. angularis) and mungbean, and one copy in Medicago truncatula and Lotus japonicus ( Figure 4C ). The best phylogeny places the legume DTX55A (LegDTX55A) subclade and the legume DTX55B (LegDTX55B) subclade sister together, forming the LegDTX55 clade in legume ( Figure 4C ).

To investigate the functional character of the MIB1 protein, the expression vector containing the MIB1 gene or empty vector was transformed into the WT strain (K12) and mutant strain (ΔacrB) in E. coli. The ΔacrB mutant strain lacks the functional AcrAB complex, the multidrug efflux carrier (Seo et al., 2012), and cannot grow under toxic conditions. The transformed cells were grown on the medium with and without different concentrations of TBA. The ΔacrB mutant cells with empty expressing vector (pET32a) did not grow on an LB plate with 10 and 15 mg/ml TBA ( Figure 5A ), while the MIB1-expressing ΔacrB cells continued their growth on the LB medium with 10 and 15 mg/ml TBA ( Figure 5A ), suggesting that MIB1 in mungbean potentially acts as a MATE/DTX family transporter.

In order to further verify the results of the plate experiment, we determined the growth curve of the strains under 0, 10, and 15 mg/ml TBA treatment in liquid LB medium ( Figures 5B–D ). Compared to those of expressing MIB1 cells and the WT strain, TBA treatment significantly inhibited the growth of the mutant strain ( Figures 5B–D ). Under 10 and 15 mg/ml TBA treatments for 24 h, the growth curve of the mutant strain expressing MIB1 was similar to those of the WT strain with and without expressing MIB1 ( Figures 5C, D ). The above results showed that heterologous expression of mungbean MIB1 increased TBA tolerance of the ΔAcrB mutant strain.

It has been reported that loss of function of DTX54/BIGE1A, MIB1 ortholog in Arabidopsis, leads to the shortened pods (Suzuki et al., 2015). To test whether MIB1 performs a similar function to DTX54/BIGE1A in the control of pod development, the coding sequence of MIB1 driven by the cauliflower mosaic virus 35S (CaMV35S) promoter was transformed into the Arabidopsis dtx54 mutant (35S::MIB1/dtx54, two representative lines L04 and L06). It was found that the shortened pods were partially rescued in 35S::MIB1/dtx54 transgenic lines ( Figures 6A, B ). The expression of MIB1 was detected in 35S::MIB1/dtx54 transgenic lines ( Figure 6C ). The above results indicated that the mungbean MIB1 gene plays a conserved role in the pod development.

To investigate the potential genes whose expression was altered in the mutants, RNA-sequencing (RNA-seq) analysis was conducted to compare the gene expression levels in shoot buds and young pods between WT and mib1-3 mutants. A total of 3,173 and 875 differentially expressed genes (DEGs) were identified at the two developmental stages, respectively ( Figure 7 , Supplementary Tables 3 and 4 ). The qRT-PCR analysis confirmed the results of the RNA-Seq ( Figures 8A, B ).

Consistent with the mutant phenotype of plant architecture, the expression levels of key components of the plant hormone pathway and transcription factors related to plant development and growth were changed in the mib1 mutants ( Figures 8C, D ). Among DEGs of the shoot buds between WT and mib1 mutants, three auxin biosynthesis genes (LOC106761734, LOC106761746, and LOC106764471, YUCCA 4, YUC4) and a gene encoding auxin transporter (LOC106760087, PIN-FORMED 1C, PIN1C) were downregulated ( Figure 8C ). Several transcription factor encoding genes involved in plant development, such as LOC06756958 (Auxin Response Factor 2, ARF2), LOC106778614 (TCP family gene, TCP4), LOC106769314 and LOC106758337 (SPL family genes, SPL7 and SPL8), and LOC06765209 (AP2/ERF family gene AINTEGUMENTA, ANT), were downregulated ( Figure 8C ). Additionally, LOC106767323 (DWARF 14, D14), encoding a key component of the SL signaling pathway (Zhou et al., 2013), was downregulated in the shoot buds of the mib1 mutants ( Figure 8C ).

It has been shown that auxin and cytokinin pathways play a key role in the control of pod development and seed number per pod (Liu et al., 2021; Qadir et al., 2021; Yu et al., 2022). We found that the auxin biosynthesis gene (LOC106778822, YUC11) and the auxin transporter encoding gene (LOC106761294, PIN2) were downregulated in the young pods of the mib1 mutants ( Figure 8D ). Consistently, there was a significant reduction in IAA level in young pods of the mib1 mutants, compared to that of WT ( Supplementary Figure 5 ). In addition, the expression levels of LOC106759647 (Cytokinin dehydrogenase 3, CKX3) and LOC106764037 (LONELY GUY 3, LOG3), related to the cytokinin pathway, were also significantly changed in the young pods of the mib1 mutants ( Figure 8D ).

The plant architecture significantly affects the seed yield of grain legume. However, the underlying molecular mechanism is still poorly understood (Liu et al., 2020). In this study, The mutations of the MIB1 gene in mungbean resulted in bushy and compact plant architecture ( Figure 1 ) and shortened pods with smaller and rounder seeds ( Figure 2 ). Map-based cloning showed that the MIB1 gene encoded a MATE/DTX family protein in mungbean, which was an ortholog of DTX54/BIGE1A in Arabidopsis and BIGE in maize ( Figure 4 ). It has been reported that loss of function of the DTX54/BIGA1A gene, MIB1 ortholog in Arabidopsis, gives rise to increased branch numbers and shortened pods (Suzuki et al., 2015). We found that heterologous expression of the MIB1 gene partially rescued the phenotype of dtx54/bige1a mutant in Arabidopsis, suggesting that MIB1 plays a conserved role in the control of pod development.

MIB1 belonged to group IV of the MATE/DTX family ( Figure 4A ). Complementation assays in E. coli showed that MIB1 potentially acted as a MATE/DTX transporter in mungbean. Meanwhile, there was a significant reduction in IAA levels in young pods of the mib1 mutants ( Supplementary Figure 5 ). Consistently, transcriptome analysis revealed that expression levels of the genes related to auxin synthesis and transport were decreased ( Figure 8 ). Thus, our results suggested that auxin plays a key role in regulating plant architecture in mungbean. The alteration of plant architecture in the mib1 mutants was probably due to the modulated levels of auxin and other plant hormones and then the altered expression of the downstream genes related to plant growth and development ( Figure 8E ).

It has been reported that the group IV MATE/DTX transporters are able to modulate plant hormone levels such as auxin and ABA in Arabidopsis and rice (Li et al., 2014; Zhang et al., 2014; Qin et al., 2021). Thus, how plant hormone level is modulated by the MIB1 protein should be investigated in more detail in the future.

It has been shown that the DTX54/BIGE1A and DTX55/BIGE1B in Arabidopsis, two paralogs, have partial functional redundancy and diversity (Suzuki et al., 2015). The mutant of the Arabidopsis DTX54/BIGE1A gene shows increased number of leaves (Suzuki et al., 2015). By contrast, the dtx55 mutant exhibits a slight increase in leaf number, suggesting that Arabidopsis DTX54 has a greater role in the control of leaf initiation, while the leaf number of the double mutants of DTX54/BIGE1A and DTX55/BIGE1B is enhanced compared to the single mutants, indicating that there is an additive interaction between the two genes (Suzuki et al., 2015).

It was found that there were 56 MATE/DTX family proteins in mungbean genome ( Figure 4 ), among which two other MATE/DTX proteins are closely related to MIB1 and might redundantly affect plant development and growth in mungbean. Moreover, based on the public sequences, we identified DTX54 and DTX55 orthologs in legume. We found that there was a single copy encoding the DTX54 orthologs in legume ( Figure 4 ). In contrast, the legDTX55 clade in legume could be further divided into two subclades, LegDTX55A and LegDTX55B ( Figure 4B ).

At present, mutant libraries for several legume species such as M. truncatula, L. japonicus, and G. max are available (Tadege et al., 2008; Małolepszy et al., 2016; Li et al., 2017), and it would be worth identifying the mutant lines of the LegDTX54 and LegDTX55 clade genes in these species to dissect their function in plant architecture establishment. Moreover, it is also interesting to study the interactions between the LegDTX54 clade and LegDTX55 clade genes during plant development and growth in legume.