Germplasm screening and genome-wide PEBP profiling identify key regulators of photoperiod-insensitive flowering in winged bean (Psophocarpus tetragonolobus L)

Abstract

Winged bean (Psophocarpus tetragonolobus L.) is a nutritionally rich underutilized tropical legume. It requires short-day (SD) conditions for flowering, which limits its cultivation across seasons and regions. In this study, 81 winged bean accessions were evaluated over three years under contrasting natural photoperiods, identifying five genotypes capable of flowering under both short- and long-day conditions. However, only one accession (IIAB-PIS1) achieved full reproductive development under long-day (LD) conditions. To understand the molecular basis of this trait, we conducted a genome-wide identification and analysis of the phosphatidylethanolamine-binding protein (PEBP) genes, which are key regulators of flowering-time pathways. Ten PEBP genes were identified and classified into FT-like, TFL1/CEN-like, and MFT-like groups, each of which displayed conserved domain structures and features. Divergence time analysis indicated recent diversification within subfamilies. Expression profiling under LD conditions showed that FT2 may be the potential candidate gene for florigen production in both photoperiod-sensitive and -insensitive genotypes. In contrast, differential expression of FT1, FT4, TFL1, and CEN2 suggested genotype-specific regulation of floral transition pathways. The discovery of a truly photo-insensitive genotype, along with the identification of key regulatory PEBP genes, offers valuable resources for developing high-yielding, photoperiod-insensitive winged bean cultivars.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-026-12857-x.

Introduction

Flowering is a critical developmental transition in the plant life cycle. It is regulated by a coordinated interplay of endogenous factors, such as developmental age, hormonal status, and nutrition, as well as external cues, such as temperature and photoperiod [1, 2]. Based on their photoperiodic responses, plants are broadly classified as long-day (LD), short-day (SD), or day-neutral species. LD plants flower when exposed to daylengths longer than a critical threshold, SD plants require extended periods of darkness (short days) to induce flowering, whereas day-neutral plants flower independently of daylength [3, 4].

The phosphatidylethanolamine-binding protein (PEBP) genes form a conserved regulatory module that integrates endogenous and environmental signals to modulate the floral transition in both LD and SD species. In angiosperms, PEBP genes fall into three major clades: FLOWERING LOCUS T (FT), TERMINAL FLOWER 1 (TFL1), and MOTHER OF FT AND TFL1 (MFT) [5, 6]. MFT represents the ancestral lineage from which FT-like and TFL1-like genes evolved [6, 7]. In Arabidopsis thaliana, FT protein functions as a systemic florigen produced in leaves and is translocated to the shoot apical meristem to promote flowering, whereas TFL1 antagonizes FT activity and maintains inflorescence meristem identity [8–10]. The quantitative balance between FT and TFL1 proteins acts as a developmental rheostat controlling the timing of floral induction and reproductive determinacy [11, 12]. MFT and additional PEBP gene family members, including TWIN SISTER OF FT (TSF), BROTHER OF FT AND TFL1 (BFT), and CENTRORADIALIS (CEN), contribute to diverse developmental functions. MFT protein integrates abscisic acid and gibberellin signaling to repress seed germination in a PHYOCHROME-INTERACTING-FACTOR1 (PIF1)-dependent manner [13, 14]. TSF protein shows functional redundancy with FT and contributes to florigenic activity under SD conditions [15]. BFT and CEN proteins function as floral repressors, and their overexpression phenotypes resemble those of TFL1 [6, 16, 17].

Photoperiodic control of FT-like gene expression forms the core molecular link between daylength perception and flowering initiation. In LD species such as A. thaliana, circadian- and photoreceptor-mediated stabilization of CONSTANS (CO) protein under extended light periods promotes FT transcription in phloem companion cells, and the resulting FT protein moves to the shoot apical meristem, where it interacts with the bZIP transcription factor FLOWERING LOCUS D (FD) to activate floral meristem identity genes and initiate reproductive development [18–20]. Conversely, SD plants, such as rice (Oryza sativa), initiate flowering via a distinct yet evolutionarily conserved photoperiodic pathway. Under inductive SD conditions, the SD-specific transcription factor EARLY HEADING DATE 1 (Ehd1) is activated and strongly induces the FT-like florigen genes Hd3a and RFT1. These proteins function analogously to FT, moving from leaves to the shoot apical meristem to trigger the vegetative-to-reproductive transition. The CO ortholog Hd1 protein acts as a dual-function regulator, activating Hd3a/RFT1 under SD while repressing their transcription under LD, thereby preventing premature floral induction. Integration of SD-specific activation of Ehd1 protein with the photoperiod-dependent switch in Hd1 activity ensures that flowering occurs strictly under SD conditions [21–23].

Winged bean (Psophocarpus tetragonolobus L.) is an underutilized tropical legume with high nutritional value, and nearly all plant parts are edible, making it a highly versatile crop. Its seeds contain ~ 34% crude protein, 23–40% carbohydrate, and 15–20% lipid, with a favorable saturated–unsaturated fatty acid ratio [24]. The seeds can also be processed into flour or brewed into a coffee-like beverage, while the tubers provide an additional nutrient-rich component containing 12–19% protein [25]. Moreover, it shows effective symbiosis with a broad range of rhizobial strains, making it a good nitrogen fixer for low-input, self-resilient agricultural systems [26, 27]. Despite these advantages, its cultivation remains geographically limited because the crop is highly SD-sensitive, restricting flowering and seed set to specific latitudes and sowing periods. Under LD or non-inductive conditions, plants remain vegetative and often fail to flower, reducing their suitability for subtropical and temperate regions. Developing photoperiod-insensitive or day-neutral genotypes would greatly expand the crop's adaptation range, support more reliable seed production, and enable off-season nurseries and multi-location testing to accelerate breeding progress.

Although PEBP genes are well established as key regulators of flowering in model species, their roles in winged bean remain largely unexplored. In this study, we screened 81 winged bean accessions under both SD and LD conditions and identified IIAB-PIS1 as a photoperiod-insensitive genotype. Genome-wide analysis using the winged bean reference genome and Cajanus cajan PEBP gene protein sequences as queries identified 10 PEBP genes with conserved domains and motifs characteristic of the PEBP family. Multiple protein sequence alignment and phylogenetic analysis of A. thaliana and C. cajan grouped the winged bean PEBP genes into FT-like, TFL1/CEN-like, and MFT-like clades [5]. Conserved exon–intron architecture and divergence-time estimates indicated both structural conservation and recent functional specialization. Expression profiling under LD conditions identified FT2 as the predominant candidate gene for florigen production in both genotypes, consistent with its conserved role in florigenesis [28, 29]. In contrast, FT1 and FT4 were preferentially upregulated in IIAB-PIS1, suggesting potential involvement in photoperiod insensitivity. TFL1 expression was stable across genotypes, whereas CEN2 expression was restricted to AKWB1 (photoperiod-sensitive), consistent with the indeterminate growth habit of winged bean and the known roles of CEN-like genes in meristem maintenance [9, 30]. These findings provide a foundational understanding of the genetic control of photoperiodic flowering in winged bean and may be highly useful for future breeding strategies to improve adaptability.

Materials and methods

Screening of winged bean accessions for photoperiod sensitivity

Eighty-one genetically diverse winged bean accessions (Supplementary Table S1) were evaluated for photoperiodic flowering response under both SD and LD conditions. SD conditions corresponded to natural daylengths from July to October (11.5–12.5 h), and sowing was carried out during this period to ensure that plants experienced short days from germination. LD conditions corresponded to natural daylengths from March to June (13.0–14.2 h), and sowing dates were selected within this window so that plants were exposed to long days from the beginning. This evaluation was conducted over three consecutive years (2022–2024) at the experimental farm of the ICAR-Indian Institute of Agricultural Biotechnology, Ranchi, Jharkhand, India (23.34°N, 85.31°E). The experiment followed a randomized block design. Each accession was represented by three plants (P₁, P₂, P₃) per photoperiod and year, yielding nine biological replicates per accession per season. Standard agronomic practices, including soil preparation, irrigation, nutrient management, and pest and disease control, were uniformly applied across all experimental units and temporal replicates to maintain environmental consistency. For each plant, days to first flower (DTF) was defined as the duration from sowing to anthesis of the first floral bud. Flowering incidence was recorded as a binary trait (presence or absence of flowering), and seed yield per plant was measured at physiological maturity.

Estimation of days to flowering (DTF) across replications and years

DTF were recorded for all winged bean accessions under both LD and SD photoperiods across replications and years. For each accession, DTF data from all replications and years were pooled separately by photoperiod, and the mean and standard deviation (S_d) were calculated. All subsequent analyses were performed using the mean DTF values derived from this pooled dataset for each photoperiod treatment.

Assessment of photoperiod-induced variation in flowering time

To evaluate the effect of photoperiod on flowering time, a paired t-test was used to determine whether DTF under LD differed significantly from DTF under SD among accessions of winged bean capable of flowering in both environments. The test statistic was computed as:

where

is the mean of paired differences,

is the standard deviation of the differences, and is the number of paired observations.

Effect size was quantified using Cohen’s d, calculated as:

Photoperiod sensitivity indices

Multiple photoperiod sensitivity indices were calculated to quantify the magnitude and direction of flowering-time responses under LD and SD conditions.

Absolute difference in flowering time

The absolute difference () in flowering time between photoperiods was computed as:

Photoperiod sensitivity index (PSI)

The PSI was estimated according to Wang et al. (2019) [31]:

PSI values closer to zero indicated weak or negligible photoperiod sensitivity.

Relative difference (RD)

Relative difference in flowering time (RD) was computed following Fei et al. (2022) [32]:

This metric represents the proportional change in DTF under LD relative to SD.

Clustering analysis

Clustering of LD-flowering accessions of winged bean was performed using days to flowering under LD (DTF_LD) and the PSI as input variables. Before analysis, both traits were standardized using z-scores. k-means clustering (k = 2), implemented in scikit-learn v1.4 [33], was used to delineate phenology-based groups. Cluster robustness and optimal partitioning were evaluated using silhouette coefficients, and final cluster assignments were determined based on the fitted model.

Principal component analysis (PCA)

PCA was conducted on the complete morpho-physiological trait dataset using the prcomp function in R v4.3 [34]. Trait values were mean-centred and scaled to unit variance prior to analysis. Scores for individual winged bean genotypes along the first two principal components (PC1 and PC2) were extracted to visualize multivariate trait structure. Photoperiod-insensitive and photoperiod-sensitive genotypes were subsequently mapped onto the PCA space to evaluate class separation.

Genome-wide identification of PEBP genes

Genome-wide identification of PEBP genes was performed using the in-house, high-quality telomere-to-telomere (T2T) genome assembly of winged bean. The complete genome dataset is accessible at NCBI Accession No. GCA_053477545.1. The statistics of the winged bean genome assembly and annotation are provided in Table 1. First, protein sequences of C. cajan PEBP genes were retrieved from the Legume Information System (LIS) and used as queries in BLASTP against the winged bean genome to identify homologous sequences. Candidate loci were subsequently subjected to gene prediction using FGENESH, with Glycine max as the species-specific model. Finally, predicted genes were examined using the Conserved Domain Database (CDD) to confirm the presence of the conserved PEBP domain.

Table 1.

Summary statistics of winged bean genome assembly

Parameter	Value
Assembled genome size (Mb)	697
Total number of scaffolds	15
Largest scaffold (Mb)	111
Scaffold N50 (Mb)	85
GC content (%)	31.08
BUSCO completeness	C: 97.61% [S:94.5%, D:3.1%], F: 0.4%, M: 1.9%, n: 5366
Total length of repetitive elements (Mb)	409
Total number of annotated genes	53,745

Conserved domain and motif analysis

Functional domain architecture of PEBP proteins was determined using the batch CD-search utility on the NCBI platform, followed by visualization in TBtools [35]. Conserved motif composition was predicted using the MEME Suite [36], with a maximum of 10 motifs and optimal width parameters adjusted to capture conserved functional regions.

Phylogenetic analysis

To classify winged bean PEBP genes into FT-like, TFL1-like, and MFT-like subclades, protein sequences from A. thaliana and C. cajan were included along with those from winged bean in MUSCLE-generated multiple sequence alignments. Phylogenetic trees were subsequently constructed in MEGA11 using the Maximum Likelihood method with the Jones–Taylor–Thornton (JTT) model, complete deletion of gaps, and 1,000 bootstrap replications [37]. To further infer evolutionary relationships, an extended comparative phylogenetic analysis was performed by aligning winged bean PEBP genes with those from other legumes, including Vigna unguiculata, Medicago truncatula, G. max, and Vigna radiata, as well as model species such as A. thaliana and O. sativa.

Gene structure analysis

Exon–intron organization was analyzed by aligning coding sequences (CDS) with genomic sequences using the Gene Structure Display Server (GSDS) (http://gsds.gao-lab.org/), allowing comparative visualization of structural variation among PEBP genes.

Divergence time estimation

Divergence times for PEBP genes in winged bean were estimated using the RelTime-ML method in MEGA v10.0.5 with the Tamura–Nei substitution model [37]. Analyses used AtMFT as the outgroup and default parameters; since external calibration was not applied, the results provided relative ordering and spacing of speciation events.

Real-time quantitative PCR (RT-qPCR) analysis of the PEBP genes

Total RNA was extracted from leaf samples of IIAB-PIS1 (photo-insensitive) and AKWB1 (photosensitive) genotypes of winged bean. Plants were grown under LD conditions at the experimental farm of the ICAR–Indian Institute of Agricultural Biotechnology, Ranchi, Jharkhand, India. RNA extraction was performed using the RNeasy Plant Mini Kit (Qiagen) according to the manufacturer’s protocol. Leaf tissues were collected at two developmental stages: the vegetative phase (30 days after sowing, DAS) and the floral induction stage, defined by the appearance of the first visible bud. For each genotype, three biological replicates were obtained. Sampling was conducted in the late afternoon (3:00–3:30 PM) when FT-like genes are known to peak in other LD species, and all samples were collected at the same time of day across genotypes to minimize circadian variation. First-strand cDNA synthesis from the extracted RNA was performed using the SuperScript IV CellsDirect cDNA Synthesis Kit (Invitrogen). Gene-specific primers were designed using the PrimerQuest tool (Integrated DNA Technologies; Table 2). Real-time quantitative PCR (RT-qPCR) was carried out on a CFX Opus 96 Real-Time PCR System (Bio-Rad). Each 10 µL reaction comprised 100 ng cDNA, 0.5 µL each of forward and reverse primers, and 5 µL PowerUp SYBR Green Master Mix (Applied Biosystems). Cycling conditions were as follows: 94 °C for 3 min; 40 cycles of 94 °C for 30 s, 60 °C for 15 s, and 72 °C for 20 s. The Actin gene served as an endogenous reference to normalize transcript abundance. Relative expression levels were calculated using the 2^ − ΔΔCt method across biological and technical replicates. Non-template controls were included to confirm amplification specificity.

Table 2.

List of primers used in the study

S. No	Genes Primer ID	Primer sequence (5′−3′)
1	PT03_G14584_MFT_F	CTGGTGGGACAAACCCATT
	PT03_G14584_MFT_R	TGGCTGCTCCACAATTCC
2	PT03_G18217_CEN1_F	GCACAACAGACGCCACAT
	PT03_G18217_CEN1_R	TTCATCACTCCCTGCCTCT
3	PT04_G20188_FT1_F	CCTAGCCCTGGAAATCCTAATC
	PT04_G20188_FT1_R	GGTCGTGGACATTCATACCC
4	PT04_G20189_FT2_F	CCCTTCACGAGTTCTCTTTCTC
	PT04_G20189_FT2_R	CAATTCTTGGTTGGTTGACG
5	PT04_G23167_FT3_F	TTCCTTTCAGGGTTACCTATGG
	PT04_G23167_FT3_R	GGAGCATCAGGATCCACTAAAG
6	PT04_G23169_FT4_F	GCAATGGCTTGGAACTTAGG
	PT04_G23169_FT4_R	GCAGGTATTCCCTCAAGATAGG
7	PT04_G24383_TFL1_F	CCAGGCACAACAGATGCTA
	PT04_G24383_TFL1_R	CCCTTGAACTAGGTGGAGTAAC
8	PT05_G27155_CEN2_F	TGAAGTGCCAAAGCCTAACA
	PT05_G27155_CEN2_R	CCTTGAAGATGGTGGAGTAACA
9	PT09_G43917_TFL2_F	GTTGCCAATGGTCATGAGTTAAT
	PT09_G43917_TFL2_R	CCCACAATCTCTTTCCCTACAA
10	PT09_G44365_FT5_F	CCAGCAACAACTGGTCCTAAT
	PT09_G44365_FT5_R	AGCATACACTGTCTCTCTACCC
11	WB_Actin_F	CCGTTCTATCCCTGTATG
	WB_Actin_R	GATCACGTCCTGCTAAG

Statistical software

All statistical analyses, including descriptive statistics, paired t-tests, and computation of photoperiod sensitivity indices, were performed in R v4.3 [34] using the stats package for hypothesis testing and custom scripts for index computation.

Results

Flowering response of winged bean accessions across photoperiods

Among the 81 winged bean accessions of winged bean evaluated under contrasting photoperiods, only five genotypes (6.17%)—IIAB-PIS1, IIABWB1, IC0041980, EC0142652, and EC0142654—flowered under both SD and LD conditions. These accessions had a PSI of ≤ 0.30 and were classified as photoperiod-insensitive. In contrast, the remaining 76 accessions (93.83%) did not flower under LD and were classified as photoperiod-sensitive. Among the five photoperiod-insensitive accessions, time to flowering was comparable under SD and LD, with only a slight delay under LD. A paired t-test comparing DTF between photoperiods revealed a non-significant mean difference of 2.15 days (S_d = 2.11; t = 2.27; p = 0.085; Cohen’s d = 1.02), indicating that daylength exerted minimal influence on these genotypes. Notably, only IIAB-PIS1 completed reproductive development under LD, with a mean seed yield of 0.43 kg plant⁻¹, whereas the other accessions flowered but failed to set viable seed. Thus, IIAB-PIS1 was the only photo-thermo-insensitive genotype identified in the panel. Table 3 summarizes mean DTF, PSI, and accession classifications.

Table 3.

Mean days to first flower (DTF), photoperiod sensitivity indices, and classification of accessions flowering under both short-day (SD)and long-day (LD) conditions

S. No	Accession	Mean DTF (SD)	Mean DTF (LD)	Δ (days)	PSI	RD (%)	LD seed yield (kg plant⁻1)	Classification
1	IIAB-PIS1	61.33 ± 1.87	62.67 ± 2.31	1.33	0.021	+ 2.17	0.43	PIS
2	IIABWB1	65.22 ± 3.14	70.44 ± 2.86	5.22	0.074	+ 8.00	0.00	PIS
3	IC0041980	61.00 ± 3.27	60.00 ± 2.00	1.00	0.017	− 1.64	0.00	PIS
4	EC0142652	91.33 ± 2.24	89.33 ± 1.63	2.00	0.022	− 2.19	0.00	PIS
5	EC0142654	90.56 ± 3.02	88.33 ± 2.06	2.22	0.025	− 2.46	0.00	PIS

Values represent means of nine plants (3 plants × 3 years) per photoperiod per accession. Δ, PSI, and RD computed from mean DTF values

Cluster analysis of flowering behaviour under contrasting photoperiods

K-means clustering based on DTF under LD (DTF_LD) and the PSI separated LD-flowering genotypes of winged bean into two groups (Fig. 1). IIABWB1 was distinct, exhibiting a higher PSI and a persistent delay in flowering under LD conditions. The remaining four accessions clustered closely together despite differences in DTF_LD, reflecting their consistently low PSI values and indicating that photoperiod had minimal influence on their flowering behaviour. This pattern suggests that flowering in these genotypes is regulated predominantly by autonomous or developmental pathways rather than by daylength-dependent cues.

Fig. 1.

Clustering of long-day (LD)–flowering winged bean accessions based on phenology and photoperiod sensitivity. Accessions are plotted by days to flowering under LD (DTF_LD; x-axis) and photoperiod sensitivity index (PSI; y-axis), with individual genotypes labelled. Points are colored by cluster assignment (Cluster 0, green; Cluster 1, orange), delineating two distinct groups that differ in flowering time and the magnitude of the photoperiod response

Multivariate differentiation of photoperiod-insensitive and photo-thermo-insensitive accessions

PCA of flowering and yield traits in winged bean provided finer resolution of photoperiod and temperature responses (Fig. 2). The fully insensitive genotype IIAB-PIS1 was clearly separated along the first principal component, driven by its minimal differences in DTF between short- and long-day conditions, low PSI, and its ability to set seed under LD. In contrast, the other LD-flowering accessions clustered together; although they initiated flowering under LD, they failed to sustain reproductive growth or set viable seed. This pattern indicates that while these genotypes are photoperiod-independent for floral initiation, their subsequent reproductive development remains sensitive to LD or associated temperature stress. Accordingly, two categories of LD-flowering accessions were distinguished: one fully insensitive throughout the reproductive phase and another insensitive only at the stage of flowering initiation.

Fig. 2.

Principal component analysis using flowering and yield traits in photoperiod-insensitive (PIS) and photoperiod-sensitive (PTIS) winged bean genotypes. Scores for individual genotypes are plotted along the first two principal components (PC1 and PC2), which summarize the major axes of variation in the trait dataset. PIS and PTIS classes are indicated by red and blue symbols, respectively, revealing their separation in multivariate space

Genome-wide identification of PEBP genes

A comprehensive genome-wide search identified 10 PEBP genes in the winged bean genome. The predicted PEBP genes ranged from 255 bp (PT04_G20189_FT2) to 525 bp (PT03_G18217_CEN1), encoding proteins of 84–174 amino acids. Their molecular weights ranged from 9.49 kDa (PT04_G20189_FT2) to 19.66 kDa (PT04_G24383_TFL1), and isoelectric points (pI) varied between 4.52 (PT04_G20189_FT2) and 9.90 (PT09_G43917_TFL2). Most PEBP proteins were hydrophilic, as indicated by negative GRAVY values, whereas the protein products of PT03_G14584_MFT and PT04_G23169_FT4 exhibited positive GRAVY values and were therefore predicted to be hydrophobic (Table 4).

Table 4.

List of identified PEBP genes in winged bean

S. No	Gene name	Base pair	Intron	Exon	Amino acids	pI	Molecular weight	Aliphatic Index	GRAVY
1	PT03_G14584_MFT	522	3	4	173	9.10	18.9660	86.13	0.042
2	PT03_G18217_CEN1	525	3	4	174	9.22	19.6486	73.79	−0.320
3	PT04_G20188_FT1	384	3	4	128	5.32	18.1134	83.01	−0.193
4	PT04_G20189_FT2	255	1	2	84	4.52	9.4929	103.00	−0.106
5	PT04_G23167_FT3	267	1	2	88	5.67	9.8260	71.82	−0.439
6	PT04_G23169_FT4	363	3	4	120	6.71	13.4104	103.00	0.007
7	PT04_G24383_TFL1	522	3	4	173	9.04	19.6696	76.01	−0.378
8	PT05_G27155_CEN2	522	3	4	173	9.67	19.6125	78.79	−0.317
9	PT09_G43917_TFL2	429	1	2	143	9.90	15.8803	83.08	−0.234
10	PT09_G44365_FT5	504	3	4	168	6.06	19.3900	82.14	−0.299

Prediction of conserved domains and motifs

All 10 PEBP protein sequences identified in the winged bean genome exhibited the characteristic conserved PEBP domain, validating their classification within the PEBP gene family (Fig. 3). The domain structures showed a high degree of conservation across all members, consistent with the presence of the phosphatidylethanolamine-binding core region typical of this protein family. Motif analysis identified five distinct conserved motifs shared among the PEBP proteins (Fig. 4a). Notably, five proteins (PT03_G14584_MFT, PT03_G18217_CEN1, PT04_G24383_TFL1, PT05_G27155_CEN2, and PT09_G44365_FT5) contained all five motifs, indicating complete retention of the canonical PEBP motif architecture. In contrast, two proteins, PT04_G20188_FT1 and PT09_G43917_TFL2, showed partial conservation, incorporating motifs (1, 2, 3, 5) and (2, 3, 4, 5), respectively. The remaining three proteins (PT04_G20189_FT2, PT04_G23167_FT3, and PT04_G23169_FT4) possessed only three conserved motifs (1, 2, 5), suggesting possible divergence in functional specialization. Complementing these observations, sequence logo representations revealed highly conserved residues within each motif, particularly motifs 1 and 3, which correspond to the ligand-binding and structural stabilization regions of PEBP proteins (Fig. 4b).

Fig. 3.

Predicted transmembrane helices in winged bean PEBP proteins. Hydropathy-based analysis identified possible transmembrane regions in the amino acid sequences of MFT-, FT-, and TFL1/CEN-like PEBP genes. Horizontal bars show the full protein length (amino acids) on the x-axis. Green marks predicted transmembrane helices. Error bars show prediction confidence

Fig. 4.

Conserved motif architecture and sequence conservation of winged bean PEBP proteins. a The five most significant motifs identified by MEME are shown as colored boxes from the N to the C terminus along each PEBP protein, with E-values listed to the left. Consensus sequences for all motifs, given in the key, show strong conservation across MFT-, FT-, TFL1-, and CEN-like subfamilies. b Sequence logos of these motifs display information content and residue frequency at each position, highlighting highly conserved amino acids that define the core PEBP domain

Phylogenetic analysis of PEBP genes

The phylogenetic analysis clustered the 10 winged bean PEBP proteins into three major subfamilies: FT-like, TFL1/CEN-like, and MFT-like (Fig. 5). Five proteins (PT04_G20188_FT1, PT04_G20189_FT2, PT04_G23167_FT3, PT04_G23169_FT4, and PT09_G44365_FT5) grouped closely with the FT-like clade of C. cajan and A. thaliana, indicating potential roles in floral promotion. Four proteins (PT04_G24383_TFL1, PT09_G43917_TFL2, PT03_G18217_CEN1, and PT05_G27155_CEN2) clustered within the TFL1/CEN-like subclades and are likely involved in repression of flowering or maintenance of floral meristem identity. The remaining protein, PT03_G14584_MFT, showed high sequence similarity to A. thaliana MFT and was classified as an MFT-like member, potentially functioning in seed development and germination regulation.

Fig. 5.

Phylogenetic relationships of winged bean PEBP proteins with homologs from Arabidopsis and Cajanus cajan. A radial neighbour-joining tree was constructed from full-length amino acid sequences of FT-like, TFL1/CEN-like, and MFT-like PEBP proteins from winged bean (highlighted in bold), Arabidopsis, and Cajanus cajan. FT clade members are shaded yellow, TFL1/CEN-like members are shaded pink, and MFT-like members are shaded blue, delineating the three major PEBP subfamilies

Extended comparative phylogenetic analysis of winged bean PEBP genes with those from other legumes, including G. max (21), C. cajan (21), V. radiata (7), M. truncatula (10), and V. unguiculata (12), as well as model species such as O. sativa (14) and A. thaliana (6), resolved three well-supported clades corresponding to the FT, TFL1, and MFT subgroups (Fig. 6). The MFT subgroup was represented by a single copy in A. thaliana, M. truncatula, G. max, and C. cajan; two copies in V. unguiculata and O. sativa; and was absent in V. radiata, possibly due to incomplete genome annotation or gene loss.

Fig. 6.

Comparative phylogeny of winged bean PEBP proteins with PEBP homologs from legumes and model crops rice and Arabidopsis. A radial neighbour-joining tree was generated using full-length amino acid sequences of FT-, TFL1/CEN-, and MFT-like PEBP proteins. These sequences are from winged bean (highlighted in magenta), Arabidopsis, and several legume species (Glycine max, Vigna radiata, Vigna unguiculata, Cajanus cajan, and Medicago truncatula). FT-like proteins are shaded yellow. TFL1/CEN-like proteins are shaded green. MFT-like proteins are shaded pink

Within the TFL1-like subgroup, the phylogeny included nine members from G. max, four each from C. cajan, V. unguiculata, V. radiata, and O. sativa, along with AtTFL1 and AtACT/CEN from A. thaliana. The expansion of TFL1-like genes in G. max reflects gene duplication events associated with its paleopolyploid genome. The FT-like clade was the largest and most diverse, comprising AtFT and AtTSF from A. thaliana, together with several orthologs: eight each from C. cajan and O. sativa, eleven from G. max, six from M. truncatula, seven from V. unguiculata, and three from V. radiata. The broad distribution and copy-number variation within the FT cluster indicate extensive diversification and subfunctionalization across species, highlighting the evolutionary complexity and distinct regulatory roles of PEBP genes in legumes.

Gene structure analysis

The exon–intron structure analysis of the 10 predicted PEBP genes in winged bean indicated that most PEBP genes exhibited a conserved exon–intron organization comprising four exons and three introns, including PT03_G14584_MFT, PT04_G20188_FT1, PT04_G23169_FT4, PT09_G44365_FT5, PT04_G24383_TFL1, PT03_G18217_CEN1, and PT05_G27155_CEN2. The remaining three genes—PT04_G20189_FT2, PT04_G23167_FT3, and PT09_G43917_TFL2—displayed a simpler structure with two exons and one intron (Fig. 7). Comparative visualization revealed that exon–intron boundaries were highly conserved within subfamilies, particularly among FT-like and TFL1-like genes, suggesting evolutionary conservation of gene architecture associated with functional stability. In contrast, minor structural variations observed within certain FT-like genes may reflect divergent regulatory evolution or differences in splicing potential related to their roles in floral induction.

Fig. 7.

Exon–intron organization of winged bean PEBP genes. Gene structures of MFT-, CEN-, FT-, and TFL1-like family members are shown in the 5′–3′ orientation. Coding regions are shown as yellow boxes, introns as black lines, and untranslated upstream/downstream regions as blue boxes, with the scale bar indicating gene length in kilobases

Determination of the relative time of divergence

The analysis of evolutionary relationships and relative divergence times among the winged bean PEBP genes (Fig. 8) indicated that PT03_G14584_MFT represents the most ancestral lineage within the PEBP family, from which all other members subsequently evolved. The bifurcation between the two major PEBP subgroups, FT-like and TFL1/CEN-like, was estimated to have occurred approximately 0.36 million years ago (Mya), following divergence from their common ancestral MFT-like gene. Subsequent diversification within these subgroups occurred at varying timescales. The FT-like lineage further differentiated roughly 0.18 Mya ago, leading to two internal clusters: (PT04_G20188_FT1, PT04_G23169_FT4, and PT04_G20189_FT2), which diverged from one another around 0.08 Mya, and a distinct clade formed by PT04_G23167_FT3 and PT09_G44365_FT5, with a more recent split estimated near 0.02 Mya. Similarly, members of the TFL1/CEN-like clade showed more recent divergence, with PT05_G27155_CEN2 and PT09_G43917_TFL2 diverging approximately 0.06 Mya and diverging from PT03_G18217_CEN1 at about 0.08 Mya. This cluster as a whole differentiated from other TFL1-like members around 0.12 Mya, consistent with gradual lineage expansion within the PEBP family. These results support a sequential model of PEBP gene evolution in winged bean, with an ancestral MFT-like gene giving rise to the FT-like and TFL1/CEN-like lineages over evolutionary timescales, indicative of functional specialization.

Fig. 8.

Time-calibrated phylogeny of PEBP genes in winged bean. Evolutionary relationships and relative divergence times among the 10 Psophocarpus tetragonolobus PEBP proteins were inferred using the RelTime-ML method in MEGA v10.0.5 with the Tamura–Nei substitution model. AtMFT served as the outgroup. The tree indicates that PT03_G14584_MFT forms the ancestral lineage. FT-like (FT1–FT5) and TFL1/CEN-like (TFL1–TFL2, CEN1–CEN2) clades subsequently diverged from it. Branch lengths and node values represent relative divergence times (Mya)

Expression analysis of PEBP genes

The relative expression profiles of PEBP genes, quantified in IIAB-PIS1 (photo-insensitive) and AKWB1 (photosensitive) under LD conditions (Fig. 9), showed that FT2 had the highest transcript abundance during the reproductive stage in both genotypes. This suggests FT2 may be the potential candidate gene for florigen synthesis and floral induction. FT1 and FT4 expression were higher during the reproductive phase of IIAB-PIS1. FT4 also moderately increased in AKWB1, indicating that these FT-like homologs are regulated differently in response to photoperiodic cues. TFL1 was expressed throughout the reproductive phase in both genotypes. CEN2 transcripts appeared only in the photosensitive AKWB-1. The simultaneous expression of TFL1 and CEN2 during flower initiation matches the indeterminate growth pattern of winged bean: floral meristem and shoot apices develop together. These patterns suggest that FT2 is a major floral activator in both genotypes, whereas FT1, FT4, TFL1, and CEN2 may modulate flowering and meristem maintenance in genotype-specific ways.

Fig. 9.

Expression profiles of PEBP genes in IIAB-PISI (photoperiod-insensitive) and AKWB1 (photoperiod-sensitive) genotypes of winged bean under long-day (LD) conditions. Relative transcript abundance for (A) FT1, (B) FT2, (C) FT4, (D) TFL1, (E) TFL2, (F) CEN1, (G) CEN2, and (H) MFT genes was quantified by RT-qPCR. Expression levels were normalised to actin. Data are presented as mean ± SE of three biological replicates. FT2, identified as the predominant florigen candidate due to its notably higher expression compared to other FT-like genes, is highlighted

Discussion

Winged bean (Psophocarpus tetragonolobus L.) is a nutrient-dense, underutilized tropical legume valued for its protein-rich seeds, tender pods, foliage, and tubers. The seeds contain about 30–35% protein, with a favourable amino acid profile, balanced levels of unsaturated fatty acids, and substantial micronutrient content. The tubers provide an additional high-protein carbohydrate source. Moreover, it forms efficient symbioses with diverse rhizobial strains, making it a strong nitrogen fixer suited to low-input, resilient farming systems. These attributes make the crop a promising candidate for addressing dietary protein gaps and nutritional deficiencies [24, 25]. The species has a strict SD requirement for both floral induction and tuberization. Under LD conditions, floral initiation and tuberization are suppressed or significantly delayed. This effect is especially pronounced when temperatures are suboptimal [38–41]. This physiological sensitivity restricts cultivation beyond equatorial regions and impedes multi-location evaluation and systematic breeding. Identifying and deploying photoperiod-insensitive or day-neutral genotypes would address this constraint, enable broader ecological adaptation, promote more stable reproductive development across latitudes, and facilitate the establishment of off-season nurseries to accelerate genetic improvement.

In this study, 76 (94%) of the 81 winged bean accessions evaluated under both SD and LD photoperiods for three consecutive years failed to initiate flowering under LD conditions. This finding is consistent with previous reports and reinforces that a strong SD requirement has been a major ecological and breeding constraint [25, 42]. Only five genotypes exhibited photoperiod insensitivity, and just one genotype (IIAB-PIS1) completed reproductive development under LD conditions. Although the small number of LD-responsive accessions restricts the potential for broad statistical generalization, this rarity is consistent with the evolutionary distribution of photoperiod adaptation in SD tropical legumes, where adaptation to LD conditions typically arises from infrequent allelic changes in flowering regulators [43, 44]. Flowering metrics for each accession were obtained from multi-year replicated trials, which strengthens the reliability of phenotypic classification. The combination of a non-significant p-value (0.085) and a large effect size (Cohen’s d ≈ 1.02) is likely attributable to limited statistical power rather than a lack of biological effect, as effect size more accurately reflects the magnitude of response in studies with small sample sizes [45, 46]. Notably, five accessions displayed exceptionally low PSI (0.017–0.074) and minimal differences in days to flowering (ΔDTF), revealing cryptic standing variation toward day neutrality, a feature largely undocumented in winged bean but well characterized and widely exploited in soybean and common bean [41, 47, 48].

The close clustering of four of these accessions in k-means space, along with their uniform weak responsiveness to photoperiod, suggests a shared genetic basis for LD flowering. This pattern may involve convergent changes at key photoperiod loci analogous to the E-series genes in soybean and major photoperiod loci in common bean and other legumes [28, 47, 48]. The rarity of such genotypes is consistent with strong historical selection for SD behaviour in equatorial environments. In these regions, strict coupling of flowering with shortening days may have conferred fitness advantages by synchronizing reproduction and tuberization [39, 49]. This rarity also highlights the importance of systematically screening broader global germplasm. Such screening should include landraces from marginal or subtropical environments where selection for relaxed photoperiod control may have occurred. This is similar to photoperiod-insensitive selections such as ‘Urizun’ identified at higher latitudes [50, 51]. A clear physiological distinction was observed between genotypes capable of floral initiation under LD and those capable of completing reproductive development. The difference between accessions that merely initiated flowering under LD and the single genotype (IIAB-PIS1) that proceeded to full pod and seed set under LD–high temperature conditions indicates that photoperiodic control of floral transition and temperature-dependent regulation of reproductive success are only partially overlapping modules [39, 40]. Extensive research on food legumes has shown that high or supra-optimal temperatures disproportionately impair reproductive processes, including pollen development, stigma and ovule function, fertilization, and early embryo growth. In contrast, vegetative growth or floral initiation is less affected. Consequently, flower abortion and pod sterility are common outcomes [52–55]. The complete sterility of four LD-flowering accessions under non-inductive, warmer LD regimes is consistent with this broader pattern of thermo-sensitivity in legumes. Even mild heat stress during flowering can drastically reduce seed set despite morphologically normal flowers [54, 56].

In contrast, IIAB-PIS1 showed unique dual insensitivity. It successfully initiated and completed reproductive development under LD and elevated temperatures. The combination of minimal PSI, ΔDTF in IIAB-PIS1, and continued pod and seed development under LD makes it genuinely photo-thermo-insensitive. This trait is rare in tropical legumes, where adaptation to photoperiod and heat is often decoupled [54, 57, 58]. Its distinct PCA cluster suggests that it carries unique alleles conferring reproductive thermoresilience, similar to combined photo- and thermotolerant lines used to breed resilient cultivars in other legumes, such as chickpea and common bean [54]. IIAB-PIS1, therefore, represents an exceptionally valuable breeding donor for expanding winged bean cultivation into higher latitudes or warmer off-season windows, supporting global efforts to develop stable, climate-resilient grain legume varieties [48, 54].

The PEBP genes play a crucial role in regulating photoperiodic flowering by integrating environmental cues with endogenous developmental signals through the opposing activities of FT-like floral activators and TFL1/CEN-like repressor proteins. In this study, we identified 10 PEBP genes through genome-wide analysis, all of which carry the characteristic conserved PEBP domain. Strong conservation of core motifs, particularly those overlapping ligand-binding and structural residues, reflects the tight structural constraints described in A. thaliana and multiple crops, where changes in just a few key amino acids can switch proteins from floral activators to repressors within the FT/TFL1 module [59, 60]. Several FT-like proteins in winged bean retained only a subset of the full motif complement, consistent with functional divergence and subfunctionalization within the FT lineage, as reported in cereals, legumes, and perennials, where distinct FT paralogs partition roles in photoperiod response, flowering, tuberization, and other developmental processes [61, 62].

Maximum Likelihood phylogenetic analysis clustered the 10 winged bean PEBP proteins into the three canonical subfamilies: FT-like, TFL1/CEN-like, and MFT-like. Five proteins grouped within the FT-like clade alongside A. thaliana and C. cajan orthologs, indicating potential florigenic functions. Four proteins clustered with TFL1/CEN-like sequences, suggesting roles in floral repression or meristem maintenance. A single PEBP protein formed the MFT-like subgroup, reflecting its conserved function in seed-related regulation. The clustering of PEBP proteins into these subfamilies closely matches the conserved three-clade architecture reported across angiosperms [5–7]. This also supports the hypothesis that an ancestral MFT-like gene duplicated early in angiosperm evolution, generating FT- and TFL1-like lineages [5, 7]. The modest copy number of PEBP genes in winged bean, compared with the marked FT/TFL1 expansion seen in soybean and other polyploid legumes, suggests a relatively compact but flexible PEBP network. Limited FT paralog diversification may still suffice to generate the observed variation in photoperiod responses [7]. Divergence patterns indicate that FT- and TFL1-like genes are separated downstream of an ancestral MFT-like gene. They also highlight more recent radiation of FT paralogs, consistent with the evolutionary models proposed by Karlgren et al. (2011) [5]. This supports the idea that recurrent FT diversification is a major driver of fine-scale tuning of flowering time and seasonal adaptation [28].

Expression profiling of PEBP genes under LD revealed that FT2 dominates during reproductive stages in both sensitive and insensitive genotypes, confirming its central role in integrating photoperiod and temperature signals [28, 60, 61]. Enhanced expression of FT1 and FT4 specifically in the photo-thermo-insensitive IIAB-PIS1 suggests that coordinated upregulation of multiple FT-like paralogs may collectively raise florigenic output above the threshold required to trigger floral transition under non-inductive LD conditions, functionally analogous to FT alleles or promoter haplotypes conferring LD flowering in otherwise SD or intermediate-response backgrounds in legumes and perennial species [29, 62].

The concurrent expression of TFL1 in both genotypes, together with the exclusive activation of CEN2 in the photosensitive accession, highlights the antagonistic FT–TFL1/CEN regulatory module that balances floral induction and meristem maintenance in indeterminate species [7, 30]. Elevated or genotype-specific expression of CEN-like repressors has been associated with delayed or unstable floral transition, prolonged vegetative growth, and altered inflorescence architecture in several crops, which may explain the stronger LD-mediated inhibition of flowering in the photosensitive genotype despite detectable FT2 expression [9]. In contrast, IIAB-PIS1 appears to shift the FT:TFL1/CEN equilibrium toward floral promotion under LD through additive activation of FT1 and FT4 and potentially reduced CEN activity. While our data point to a central role for PEBP genes, the distinctive dual insensitivity of IIAB-PIS1 to both photoperiod and temperature suggests the involvement of a broader genetic network extending beyond PEBP genes. Photoperiod perception via phytochromes (phyB, phyC) and cryptochromes (cry1/2) regulates CONSTANS-like transcription factors, which directly activate FT-like florigen genes. In grasses and legumes, phyC interacts with components of the evening complex, such as ELF3, to regulate PEBP paralogs under LD conditions, potentially derepressing FT1/FT4 in insensitive genotypes [63, 64]. Circadian clock outputs, including ELF3 and LUX, fine-tune the timing of CO/FT expression peaks in the afternoon, synchronizing PEBP expression with dusk; in sensitive genotypes, misalignment of these rhythms may suppress FT upregulation under LD through enhanced TFL1/CEN repression [63]. Temperature-responsive pathways, notably PIF4 activation at elevated temperatures, can further promote FT/TSF transcription independently of photoperiod, and in IIAB-PIS1, this mechanism may act synergistically with FT2 dominance to enhance florigenic flux and override repression during reproductive development [64], thereby enabling not only floral initiation but stable reproductive progression and improved thermotolerance [54, 55]. However, interpretation of these regulatory dynamics is limited by the single late-afternoon sampling, which precludes resolution of diurnal and circadian expression patterns that often differ among PEBP paralogs; because FT typically peaks in the late afternoon under inductive LD conditions while TFL1/CEN genes may display distinct phase relationships, single-time-point measurements cannot distinguish changes in transcript abundance from genotype-specific alterations in rhythmic phase or amplitude, thereby limiting mechanistic inference. From a breeding perspective, the identification of a winged bean genotype combining photoperiod and thermo-insensitivity offers a rare opportunity to overcome the strict SD requirement of the crop. Introgression of the IIAB-PIS1 haplotype into diverse genetic backgrounds, supported by genomic-assisted selection targeting key FT and TFL1/CEN loci, could facilitate development of dual-season, broad-latitude cultivars analogous to strategies deployed in soybean, chickpea, and common bean. Given the exceptional nutritional value and multipurpose utility of winged bean, such adaptable cultivars would substantially enhance its commercial potential in subtropical and temperate regions. Future work should prioritize functional dissection of PEBP paralogs through allelic and promoter analyses, regulatory interaction studies, genome editing, and comprehensive photo-thermal phenotyping across controlled and multi-location environments to define flowering thresholds and enable strategic allele stacking that decouples photoperiod sensitivity from thermal susceptibility. Nonetheless, deployment of IIAB-PIS1 alleles may be complicated by linkage drag and pleiotropic effects of PEBP loci on plant architecture, yield, and stress responses, as documented for soybean E-loci and pigeonpea flowering QTLs [65, 66]. Differential expression of FT1 and FT4, along with diagnostic SNPs in promoter or coding regions of FT genes, including FT2, could serve as robust markers for marker-assisted selection, and KASP-based assays targeting these polymorphisms would enable precise foreground and background selection to accelerate introgression while minimizing linkage drag, similar to photoperiod-related MAS strategies employed in soybean [66].

Conclusion

This study integrates multi-year phenotyping with genome-wide analyses to elucidate the physiological and molecular basis of flowering regulation in winged bean. Field evaluation of 81 accessions identified a single photo-thermo-insensitive genotype, IIAB-PIS1, capable of flowering and completing reproductive development under LD conditions. This genotype represents a valuable genetic resource for expanding the crop’s adaptive range. Genome-wide analysis of the PEBP gene family revealed 10 members distributed across the FT-like, TFL1/CEN-like, and MFT-like clades, all exhibiting strong structural conservation alongside evidence of recent subfamily diversification. Expression profiling confirmed FT2 as the primary florigen, whereas genotype-specific induction of FT1, FT4, TFL1, and CEN2 indicates distinct regulatory pathways governing floral transition and meristem identity. Collectively, these findings establish a conceptual framework linking photoperiodic and thermal responses to PEBP-mediated flowering regulation in winged bean. The discovery of a naturally occurring photo-thermo-insensitive (PTIS) genotype and the identification of key regulatory genes offer strategic opportunities for molecular breeding to develop photoperiod-insensitive, seasonally adaptable, and climate-resilient cultivars. Although LD-adapted genotypes are rare within the evaluated germplasm, the identification of the naturally occurring photo-thermo-insensitive IIAB-PIS1 genotype, together with the characterization of major regulatory genes, offers clear strategic entry points for breeding programmes targeting enhanced seasonal flexibility and climate resilience.

Authors’ contributions

BKS conceived the project and wrote the manuscript. BKS and KUT designed the experiments. KUT, MN, AG, DKS and SUDHIR K performed the experiments. AG, DKS and SAURAV K analysed the data. All the authors read and approved the manuscript.

Funding

This study was conducted at the ICAR-Indian Institute of Agricultural Biotechnology, Ranchi 834 003, Jharkhand, India. However, the work received no specific funding.

Data availability

The complete winged bean genome dataset used in the study is accessible at NCBI Accession No. GCA_053477545.1. The gene IDs listed in the manuscript are similar to those in the whole-genome data.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with all ethical guidelines and principles and was approved by the Institute's competent authority. All participants provided informed consent before participating in the study.

Consent for publication

All authors have agreed to the publication of this manuscript. The authors declare that they have no competing interests.

Competing interests

The authors declare no competing interests.