Introduction

Mung bean (Vigna radiata) is a globally important pulse crop, valued for its high nutritional content and agronomic versatility. It is a rich source of easily digestible protein, essential macro- and micronutrients—including phosphorus, magnesium, potassium, zinc, iron, copper, manganese, and vitamins A, B2, and B6—along with dietary fiber and various bioactive compounds beneficial to human health1,2. Consequently, mung bean serves as a vital cash crop for smallholder farmers and plays a crucial role in combating global hunger and malnutrition, particularly in South and Southeast Asia.

Globally, mung bean is cultivated on approximately 7 million hectares, with Asia (mainly India, Myanmar, China, Pakistan, and Thailand) accounting for about 90% of production, averaging 721 kg/ha3. India produces about1.6 million tonnes (Mt), followed closely by Myanmar at 1.59 Mt4. As a legume, mung bean enhances soil fertility by fixing atmospheric nitrogen through root-associated rhizobacteria, thereby supporting sustainable agroecosystems5. Its short growth cycle also makes it suitable for intercropping and double-cropping systems6.

Mung bean productivity is increasingly limited by multiple abiotic stresses, notably drought, salinity, flooding, and heat stress. Under ongoing climate change, rising global temperatures pose a major threat to mung bean yield, as heat stress affects both vegetative and reproductive development7,8,9. During the vegetative phase, high temperatures impair key physiological processes such as photosynthesis, respiration, transpiration, and stomatal conductance, while damaging the photosynthetic apparatus9,10. More critically, heat stress during the reproductive phase drastically reduces productivity by inducing excessive flower drop and impairing pollen formation, viability, germination, and stigma receptivity, ultimately resulting in malformed pods and seeds and thus reduced yield2,9,10.

Although substantial research has elucidated the physiological and biochemical mechanisms underlying heat tolerance, as well as its effects on seed quality traits such as protein and nutrient composition2,10, the metabolomic basis of heat stress response in mung bean seeds remains largely uncharacterized. Metabolomics provides a powerful means of profiling the complete set of metabolites within a biological system at a given time, enabling insights into the metabolic adjustments underpinning stress adaptation11,12,13. This approach has been pivotal in uncovering key heat-responsive metabolites and pathways in several crops, including rice (Oryza sativa L.)11,14,15, maize (Zea mays L.)12,16,17,18, wheat (Triticum aestivum)19,20,21,22, sorghum (Sorghum bicolor)23, Medicago sativa13, soybean (Glycine max L.)24,25 and chickpea (Cicer arietinum L.)26,27.

Despite these advances, metabolomic studies aimed at deciphering the mechanisms of heat tolerance in mung bean are scarce. Therefore, this study employs an untargeted metabolomics approach to characterize the seed metabolite profiles of two contrasting mung bean genotypes—one heat-tolerant and one heat-sensitive—under controlled heat stress (42/30 °C) conditions, providing novel insights into the metabolic basis of heat stress tolerance in this important crop.

Material and methods

Plant material and growth conditions

Two contrasting mung bean genotypes—PI425243 (heat-tolerant; HT) and PI223002 (heat-sensitive; HS)2—were cultivated under controlled environment conditions in a greenhouse. Plants were grown under two temperature regimes: control conditions (32 °C day/25 °C night; 12 h each) and heat stress conditions (42 °C day/30 °C night; 12 h each).”Seeds were sown in 20 cm diameter pots filled with potting soil (“Fafard® 3B Mix/Metro-Mix® 830; SUNGRO Horticulture, Agawam, MA, USA”). Three biological replicates were used for each genotype, with each replicate consisting of three plants. At maturity, three seeds per plant were randomly harvested for subsequent metabolomic analysis.

In the growth chamber, photosynthetically active radiation (400–700 nm) was maintained at 600 μmolm⁻2 s⁻1 using cool fluorescent lamps, with a 12 h photoperiod and an average relative humidity of 60%2. Plants were watered regularly to maintain field capacity (moist but without runoff) and fertilized every 7–14 days with ½ teaspoon of Miracle-Gro (24-8-16) per 4.5 L of water2,9. Temperature data were recorded using a HOBO® data logger (Onset Computer Corporation, USA), as shown in Fig. S1.

Metabolite extraction

Physiologically mature mung bean seeds were thawed on ice, and 100 mg of each sample was transferred into 2 mL microcentrifuge tubes. Metabolites were extracted with 800 µL of 80% methanol and ground at 65 Hz for 180 s. The homogenized samples were vortexed and sonicated for 30 min at 4 °C (for details see Jha et al.26).

Samples were stored at − 20 °C for 1 h, vortexed for 30 s, incubated at 4 °C for 30 min, centrifuged at 12,000 rpm for 15 min at 4 °C, and the supernatant was collected. The supernatant was stored at − 20 °C for 1 h before centrifuging at 12,000 rpm for 15 min at 4 °C26. Finally, 200 µL of the resulting supernatant and 5 µL of DL-o-chlorophenylalanine (0.14 mg/mL) were transferred into vials for LC–MS analysis28,29.

Ultra-high-performance liquid chromatography–mass spectrometry (UHPLC–MS) analysis

Metabolite profiling was performed using an ACQUITY Ultra-Performance Liquid Chromatography (UPLC) system (Waters, Milford, MA, USA) coupled to a Q Exactive™ Orbitrap high-resolution mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) equipped with a heated electrospray ionization (HESI) source. This platform enabled high-resolution, high-mass-accuracy detection of a broad range of primary and secondary metabolites.

Chromatographic separation was achieved on an ACQUITY UPLC HSS T3 reversed-phase column (100 × 2.1 mm, 1.8 μm particle size), which is optimized for the retention and resolution of both polar and moderately non-polar metabolites. The column temperature was maintained at 40 °C to ensure reproducible retention times and optimal peak shapes, while the autosampler temperature was set at 4 °C to prevent metabolite degradation during analysis.

A binary solvent system was employed, consisting of 0.05% (v/v) formic acid in ultrapure water as mobile phase A and acetonitrile as mobile phase B. Metabolites were eluted using a linear gradient program designed to maximize metabolome coverage: 0–1 min, 5% B; 1–12 min, 5–95% B; 12–13.5 min, 95% B; 13.5–13.6 min, 95–5% B; and 13.6–16 min, 5% B for column re-equilibration. The flow rate was maintained at 0.3 mL min⁻1, ensuring efficient separation with minimal backpressure. This chromatographic method was adapted from previously optimized protocols (Jha et al.26).

Mass spectrometry conditions

Mass spectrometric data acquisition was carried out in both positive (ESI⁺) and negative (ESI⁻) ionization modes to enhance metabolite coverage, as different chemical classes preferentially ionize under different polarity conditions. The mass spectrometer was operated in full-scan mode with high mass accuracy, allowing reliable metabolite detection and annotation.

For ESI⁺ mode, the ion source parameters were set as follows: heater temperature 300 °C, sheath gas flow 45 arbitrary units (arb), auxiliary gas flow 15 arb, sweep gas flow 1 arb, spray voltage 3.0 kV, capillary temperature 350 °C, and S-lens RF level 30%.

For ESI⁻ mode, the corresponding parameters were: heater temperature 300 °C, sheath gas flow 45 arb, auxiliary gas flow 15 arb, sweep gas flow 1 arb, spray voltage 3.2 kV, capillary temperature 350 °C, and S-lens RF level 60%. These optimized conditions ensured efficient ionization, reduced in-source fragmentation, and improved sensitivity across diverse metabolite classes26.

Together, this UHPLC–MS workflow provided high chromatographic resolution and accurate mass detection, enabling robust, reproducible, and comprehensive metabolomic profiling suitable for downstream statistical and biological interpretation.

Statistical analysis

Raw data acquisition and alignment were performed in Compound Discoverer (v3.0, Thermo) using m/z values and retention times. Data from both ionization modes were merged and imported into SIMCA-P (v14.1) for multivariate analysis27. Principal component analysis (PCA) was first used as an unsupervised method for visualizing sample clustering and detecting outliers30. Supervised models, including partial least squares discriminant analysis (PLS-DA)31 and orthogonal PLS-DA (OPLS-DA), were then applied to identify potential biomarkers. Candidate metabolites were screened using a variable importance in projection (VIP)32 score > 1.5 and a fold-change (FC) threshold|log₂FC > 1. Metabolites with log₂FC ≥ 1 (upregulated) or ≤ − 1 (downregulated) and P < 0.05 were considered significantly differentially expressed.

Metabolite identification

Key metabolites were identified by matching precise mass and MS/MS fragmentation data against publicly available databases, including the Human Metabolome Database (HMDB; www.hmdb.ca), ChemSpider (www.chemspider.com), and MassBank (www.massbank.jp). When necessary, metabolite identities were confirmed using authentic standards based on retention times and MS/MS fragmentation patterns26.

Cluster analysis

Hierarchical clustering analysis (HCA) was performed using the complete linkage algorithm in Cluster 3.0 (Stanford University), and visualized using the pheatmap “R package (version 1.0.12, Raivo Kolde).The HCA was based on metabolite ratios from two experimental conditions, highlighting significantly altered metabolites. In the heatmaps, red and green represent metabolite levels higher or lower than the mean, respectively.

Metabolite correlation network and pathway enrichment

A metabolite correlation network was constructed using KEGG pathway data33,34 integrated with the MetaboAnalyst platform to explore functional associations among significantly altered metabolites (P < 0.05). Key metabolites were annotated through HMDB and KEGG to identify related pathways, enzymes, and biological functions. Enriched pathways were visualized using dot plots and network diagrams to illustrate the interconnected roles of crucial biomarkers involved in the heat stress response26.

Results and discussion

Phenotypic performance of contrasting genotypes under control and heat stress

Under control condition (32 °C day/25 °C night), the heat-tolerant mung bean genotype PI425243 exhibited significantly superior reproductive performance compared with the heat-sensitive genotype PI223002, producing higher numbers of effective pods per plant (EPPP; 25 vs. 21), seeds per plant (SPP; 150 vs. 138), and seed yield per plant (SYPP; 6.6 g vs. 6.1 g) (Fig. 1). These differences, though moderate under optimal conditions, indicate inherent genotypic variation in yield-related traits. Additional morpho-physiological parameters are presented in Table 1.

Fig. 1
Fig. 1
Full size image

Yield parameters of mung bean genotypes PI425243 (heat tolerant) and PI223002 (heat sensitive) under non stress (NS) and heat stress (HS) conditions. Performance under NS (32°/25 °C): (a) effective pods/plant, (b) total seeds/plant, and (c) seed yield per plant. Performance under HS (42°/30 °C);(d) effective pods/plant, (e) total seeds/plant, and (f) seed yield/plant. Values represent means + SE (n = 5). Different lowercase letters above bars indicate significant differences between genotypes within a treatment at P < 0.05 according to Tukey’s Honest Significant Difference (HSD) test.

Table 1 Mean value of morpho-physiological traits of contrasting genotypes under non stress and heat stress condition.
Full size table

Exposure to heat stress (42 °C day/30 °C night) markedly reduced reproductive output in both genotypes; however, PI425243 maintained significantly higher productivity than PI223002. Under heat stress, PI425243 produced 10 EPPP and 40 SPP, whereas PI223002 produced only 6 EPPP and 24 SPP (Fig. 1). Correspondingly, SYPP was significantly higher in PI425243 (4.6 g) than in PI223002 (2.9 g), confirming its superior heat tolerance. These results are consistent with earlier reports demonstrating that elevated temperatures severely impair reproductive development and yield in mung bean and other grain legumes, with tolerant genotypes showing greater yield stability under stress2,10.

Multivariate metabolomic analysis reveals genotype-specific metabolic reprogramming

To investigate global metabolic variations among samples, principal component analysis (PCA) was first performed using data acquired from both ionization modes. PCA is an unsupervised multivariate technique that highlights intrinsic variation and enables visualization of overall metabolic patterns. PCA revealed a clear separation between PI425243 and PI223002 under control and heat stress conditions (Fig. 2a, d), indicating distinct metabolite profiles associated with genotype and thermal regime. To enhance group discrimination and reduce nonspecific variation, supervised models including PLS-DA and OPLS-DA were employed. Both models showed strong separation between genotypes under control (Fig. 2b, e) and heat stress (Fig. 2c, f), reflecting robust genotype-dependent metabolic reprogramming in response to heat.

Fig. 2
Fig. 2
Full size image

Multivariate analysis of metabolites under non-stress and heat stress conditions in mung bean genotypes. (a, d) PCA score scatter plot under non-stress and heat stress; (b, e) PLS-DA score scatter plot non-stress and heat stress; (c, f) OPLS-DA score scatter plot non-stress and heat stress.

Metabolites contributing most strongly to genotype discrimination were identified using a VIP score > 1.5. Score and loading plots (Fig. 3a–c under non stress) and (Fig. 3b–d under heat stress) highlighted key metabolites responsible for separation, suggesting that specific secondary metabolites play central roles in conferring heat tolerance. Similar study has been reported in chickpea seed metabolomics assessed under heat stress condition26.

Fig. 3
Fig. 3
Full size image

Distribution and significance of metabolites in mung bean genotypes.(a) VIP values (VIP > 1.5) under non-stress; (b) VIP values (VIP > 1.5) under heat stress; (c) PLS-DA loading plot under non stress, significant metabolites (VIP > 1.5) highlighted in red boxes; (d) PLS-DA loading plot under heat stress, significant metabolites (VIP > 1.5) highlighted in red boxes.

Differential metabolite accumulation under control (non stress) and heat stress

Single-variable analysis using volcano plots revealed substantial differences in metabolite abundance between the two genotypes. Under control conditions, 87 metabolites were differentially accumulated in PI425243 relative to PI223002, with 55 upregulated and 32 downregulated (Fig. 4a). Under heat stress, 68 metabolites showed differential accumulation, including 39 upregulated and 29 downregulated metabolites in PI425243 (Fig. 4b).

Fig. 4
Fig. 4
Full size image

Volcano plots of differentially expressed metabolites in PI425243 vs PI 223002. (a) Non-stress condition: red, upregulated; green, downregulated; gray, no change. (b) Heat-stress condition: red, upregulated; green, downregulated; gray, no change.

Notably, under control conditions, PI425243 showed significantly higher accumulation of several phenolic acids and flavonoids, including hydrocinnamic acid [Fold change (FC) = 6.03), 5-hydroxyferulic acid (FC = 11.75), luteolin 7-malonylglucoside (FC = 4.28), quercetin 3-(6″-malonyl-glucoside) (FC = 5.89), and kaempferol 3-(2G-apiosylrobinobioside) (FC = 8.19) (Tables 2 and S1). In contrast, metabolites such as diosmin (FC = − 9.92), naringin (FC = − 6.63), rhoifolin (FC = − 7.25), pinocembrin 7-rhamnosylglucoside (FC = − 3.96), and rubraflavone D (FC = − 4.57) were more abundant in the heat-sensitive genotype PI223002, particularly under heat stress. Similarly, metabolomics analysis resulted in high accumulation of various flavonoids in chickpea seeds26.

Table 2 Selected differentially expressed metabolite in mung bean seeds of PI425243 compared to PI223002 under non stress conditions.
Full size table

Under heat stress, 68 metabolites were differentially accumulated, with 39 upregulated and 29 downregulated in PI425243. Notable upregulated metabolites included hydrocinnamic acid (FC = 4.46), 5-hydroxyferulic acid (FC = 13.1), quercetin (FC = 2.3), myricetin (FC = 4.2), and kaempferol 3-(2G-apiosylrobinobioside) (FC = 8.66). The most significantly downregulated metabolites were oleanolic acid (FC = − 2.04), naringin (FC = − 4.8), diosmin (FC = − 9.1), rhoifolin (FC = − 4.9), and goshonoside F7 (FC = − 5.7) (Tables 3 and S2).

Table 3 Selected differentially expressed metabolite in mung bean seeds of PI425243 compared to PI223002 under heat stress conditions.
Full size table

Hierarchical clustering analysis further confirmed clear genotype-specific metabolic signatures under heat stress (Fig. 5). PI425243 clustered separately from PI223002 and was characterized by higher accumulation of metabolites including hydrocinnamic acid, 5-hydroxyferulic acid, quercetin, myricetin, kaempferol 3-(2G-apiosylrobinobioside), and hesperetin 7-neohesperidoside with known antioxidant and stress-protective functions, suggesting their potential as biochemical markers for heat tolerance.

Fig. 5
Fig. 5
Full size image

Heat map and hierarchical cluster analysis of significant metabolites under heat stress. PI425243 and PI223002 were clearly separated, with PI425243 showing higher accumulation of potential biomarker metabolites, including hydrocinnamic acid, 5-hydroxyferulic acid, quercetin, myricetin, kaempferol 3-(2G-apiosylrobinobioside), and hesperetin 7-neohesperidoside.

Flavonoids as central mediators of heat stress tolerance

Flavonoids constitute a major class of plant secondary metabolites with well-established roles in mitigating oxidative stress through ROS scavenging35,36,37. In the present study, heat tolerance in PI425243 was strongly associated with increased accumulation of flavonols such as kaempferol, quercetin, and myricetin under heat stress. These compounds are known to interact with ABA and auxin signaling pathways, regulate stomatal behavior, and enhance antioxidant enzyme activities, thereby maintaining cellular redox homeostasis under stress.

Kaempferol and quercetin have been widely implicated in tolerance to drought, salinity, and heat stress across multiple crop species, including chickpea, rice, maize, and tomato38,39,40,41. Their enhanced accumulation in PI425243 likely contributes to improved protection of photosynthetic machinery and reproductive tissues under high temperature. Similarly, increased myricetin accumulation may further strengthen antioxidant capacity, as reported previously in heat-tolerant rice, maize and tomato genotypes42,43,44.

Interestingly, naringin and diosmin—flavonoids often associated with stress protection45,46,47—were more abundant in the heat-sensitive genotype PI223002 under heat stress. This suggests that their accumulation alone may not be sufficient to confer heat tolerance and may instead reflect a stress-induced metabolic imbalance rather than an adaptive response.

Phenolic acids as metabolic signatures of heat stress response in mungbean seeds

Hydroxycinnamic acids (HCAs), including p-coumaric, caffeic, ferulic, and sinapic acids, along with their conjugates such as chlorogenic acid, are central intermediates in plant metabolism48.

Hydroxycinnamic acids, including hydrocinnamic acid and ferulic acid derivatives, were significantly enriched in PI425243 under heat stress. These compounds are central intermediates in phenylpropanoid metabolism and play key roles in lignin biosynthesis, membrane stabilization, and antioxidant defense49,50.

Ferulic acid , a phenolic acid derived from cinnamic acid, plays diverse roles in plant responses to both biotic and abiotic stresses51. It is well-established as a protective compound against heat stress52. Ferulic acid and its derivatives have been shown to enhance heat tolerance by upregulating antioxidant enzymes, increasing osmolyte accumulation, and preserving photosynthetic efficiency51. The elevated levels of 5-hydroxyferulic acid observed in PI425243 likely contribute to enhanced ROS detoxification and structural integrity under heat stress, consistent with reports in maize, barley (Hordeum vulgare), tomato (Solanum lycopersicum), and blueberry (Vaccinium corymbosum)51,52,53,54.

Terpenoids exhibit genotype-specific responses to heat stress

Terpenoids, also known as isoprenoids, represent the largest class of specialized plant metabolites55 and play pivotal roles in plant growth, survival, and defense56. More broadly, terpenoids have been implicated in enhancing tolerance to various stresses, including heat57,58. For example, in tomato, exposure to 37 °C increased emissions of several monoterpenes, such as 2-carene, α-phellandrene, limonene, and β-phellandrene, whereas elevated α-pinene was detected only at temperatures ≥ 46 °C57. The monoterpene (E)-β-ocimene was observed exclusively at ≥ 46 °C57. Furthermore, heat-tolerant tomato genotypes, such as IIHR-2841, exhibited a two-fold induction of terpenoid synthase genes under heat stress, with the enhanced expression of β-caryophyllene synthase (TPS12) and β-phyllandrene synthase (TPS20) linked to heat tolerance59. Heat stress responses also vary across species. For instance, at 49 °C, Amaranthus cruentus, A. hybridus, Solanum aethiopicum, Telfairia occidentalis, and Vigna unguiculata exhibited species-specific terpenoid emissions, with higher constitutive isoprenoid emission correlating with greater heat tolerance, whereas less resistant species displayed stronger stress-induced terpenoid induction60.

In the present study, oleanolic acid, a pentacyclic triterpenoid, accumulated to higher levels in the heat-sensitive genotype PI223002 under heat stress. While terpenoids have been associated with stress tolerance in several species, this contrasting pattern suggests that terpenoid-mediated heat adaptation in mung bean may involve complex, genotype-specific regulatory mechanisms rather than a simple positive correlation with tolerance.

Pathway-level insights into heat stress adaptation

KEGG pathway enrichment analysis revealed that differentially accumulated metabolites were significantly associated with starch and sucrose metabolism, tyrosine metabolism under control condition and starch and sucrose metabolism and terpenoid biosynthesis under heat stress. Disruption of starch biosynthesis under heat stress is well documented and often leads to reduced assimilate availability during grain filling61,62,63,64. The enrichment of carbohydrate metabolism pathways in PI425243 suggests improved metabolic flexibility and energy homeostasis under heat stress.

Tyrosine metabolism and steroid hormone biosynthesis

Heat stress induces extensive metabolic and hormonal reprogramming, with tyrosine metabolism and steroid hormone biosynthesis emerging as key pathways associated with thermotolerance. Tyrosine-derived metabolites, including the thermomemory compound salidroside65, contribute to heat stress acclimation and oxidative stress mitigation, as demonstrated across diverse plant systems, while elevated tyrosine levels may directly scavenge reactive oxygen species66. Although caffeine biosynthesis is limited to a few plant species, evidence from rice suggests a potential role in abiotic stress tolerance through Ca2⁺-dependent protein kinase signaling67,68. In parallel, brassinosteroids play a central role in heat stress tolerance by activating HsfA1d-mediated transcription, enhancing ethylene biosynthesis, and strengthening cellular defense mechanisms69,70. Consistent improvements in growth, yield, and antioxidant capacity following exogenous brassinosteroid application across crops further support the involvement of steroid hormone signaling in thermotolerance71,72,73. Collectively, these findings highlight tyrosine-associated metabolic pathways and steroid hormone biosynthesis as plausible molecular mechanisms underlying heat tolerance in mung bean.

Implications for heat tolerance breeding in mung bean

Collectively, these results demonstrate that heat tolerance in mung bean is associated with coordinated metabolic reprogramming, particularly involving flavonoids, phenolic acids, and hormone-related pathways (see Fig. 6). The consistent enrichment of kaempferol, quercetin, myricetin, hydrocinnamic acid, and 5-hydroxyferulic acid in the heat-tolerant genotype PI425243 highlights their potential as metabolic biomarkers for heat tolerance. These findings provide valuable biochemical insights that can be integrated with genetic and genomic approaches to accelerate the development of heat-resilient mung bean cultivars.

Fig. 6
Fig. 6
Full size image

Experimental workflow and comparative metabolomic profiling of heat-tolerant and heat-sensitive mung bean genotypes under heat stress. Plants were grown under heat stress conditions (42°/30 °C), comparing the heat-tolerant genotype PI 425243 with the heat-sensitive genotype PI 2230023. Heat stress resulted in normal pod and grain development in the tolerant genotype, whereas the sensitive genotype exhibited shriveled pods and grains. Grains were subjected to metabolite extraction followed by qualitative and quantitative metabolite analysis. Multivariate and univariate analyses identified heat-responsive differential metabolites, visualized through VIP score plots, volcano plots, and heat maps, highlighting metabolic differences associated with heat tolerance.

Conclusion

The increasing frequency and intensity of heat stress driven by climate change pose a significant threat to mungbean productivity and global food security. This study provides metabolomics-based evidence that seed traits are highly responsive to heat stress and identifies specific metabolites—hydroxycinnamic acid, 5-hydroxyferulic acid, quercetin, myricetin, kaempferol 3-(2G-apiosylrobinobioside), and hesperetin 7-neohesperidoside—as key biochemical signatures associated with heat tolerance in mungbean. In addition, the involvement of starch and sugar metabolism, tyrosine metabolism, caffeine metabolism, and steroid hormone biosynthesis highlights coordinated metabolic adjustments underlying thermal resilience. Collectively, these findings document seed metabolic signatures associated with heat stress tolerance, while acknowledging that functional validation is required to establish causal or mechanistic links. The identified metabolites and pathways nonetheless offer promising biomarkers for seed-based selection and metabolite-assisted breeding of climate-resilient mungbean.