Glutamate metabolic correlation analysis reveals CnP5CS1 contributes to 2-acetyl-1-pyrroline accumulation in aromatic coconut Authors: Hao Ding, Xiwei Sun, Jinyao Yin, Dan Luo, Cuinan Shi, Xiaojun Shen, Xiaomei Liu, Jing Li, Amjad Iqbal, Yaodong Yang Published: 2026 ABSTRACT AbstractFlavor quality, a key sensory attribute of coconut, has consistently been a central breeding objective throughout long-term domestication and varietal improvement efforts. Developing high-aroma varieties requires a clear understanding of their underlying molecular genetic mechanisms. However, research on the metabolic regulatory enzymes involved remains limited, particularly those linked to 2-acetyl-1-pyrroline (2AP), a volatile compound that primarily contributes to the unique scent of aromatic coconuts. We developed contrasting populations and systematically evaluated the role of CnP5CS in 2AP accumulation by examining enzyme activity, metabolic flux, population-level genetic variation, and transcriptional regulatory networks. In the aromatic coconut population, the selected genomic regions were enriched in pathways associated with amino acid metabolism and stress responses. Conspicuously, glutamate (Glu) and its derivatives showed significant correlations within the differentiated populations. The Glu metabolic enzyme P5CS was subjected to strong purifying selection, and haplotype-phenotype association analysis further identified the dominant CnP5CS1 allele genotype. Moreover, we established metabolic marker indicators to assess relative 2AP levels, based on the metabolic profiles of CnP5CS and the substrates and products of its catalyzed reactions. The Y1H assay identified the key transcription factor CnYAB2, which exhibited a strongly correlated expression pattern with CnP5CS1 and major markers of 2AP metabolism. The identification of CnP5CS1 offers a novel perspective on the genetic regulation of 2AP metabolism in aromatic coconuts and establishes a theoretical foundation for developing molecular markers to support the breeding of high-aroma varieties.Keywords: Coconut, Aroma, P5CS, Association analysis, Transcription factors Introduction The coconut (Cocos nucifera L.), the most economically important woody oil crop in tropical regions, serves as both a food and energy resource and is widely cultivated because of its high overall utilization value (DebMandal & Mandal, 2011; Lima et al., 2015). China has cultivated coconuts for more than 2000 years. Among its growing regions, Hainan Province plays a central role in the industry due to its unique geographic and climatic conditions, making coconut production vital to local economic development and livelihoods. The natural flavor of raw coconut fruit is a primary concern for farmers, plantations, and processing companies, as is the case with many other crops (Cai et al., 2024; Li et al., 2025; Liu et al., 2025). As a critical sensory trait, the enhancement of flavor has long been a central focus in coconut breeding (Xu et al., 2025; Zhou et al., 2023). In-depth research on the aroma compound profile of coconuts holds scientific significance for improving flavor quality and developing distinctive aromatic varieties. The aromatic coconut, a mutant variety of dwarf coconut with high-quality fruit and a distinctive taro-like aroma, was first identified by Thai scientists. It has since been widely cultivated across Southeast Asia and has become highly popular among consumers (Luckanatinvong et al., 2018; Phonphoem et al., 2022; Yang et al., 2020). Chinese researchers began systematic studies on coconuts in the 1970s. Over the following three decades, these efforts culminated in the establishment of an aromatic germplasm nursery, creating a valuable platform for the selection and breeding of coconuts with superior traits. Research by Luckanatinvong et al. (2018) has shown that the signature taro-like note in aromatic coconuts comes from 2-acetyl-1-pyrroline (2AP), an odorant that likewise produces the characteristic smell of aromatic rice. The 2AP abundance in aromatic versus non-aromatic coconut varieties was compared via GC–MS, using ten replicates for each category. The results showed that 2AP was present in all aromatic coconuts, but absent in all non-aromatic ones (Saensuk et al., 2016). Current research indicates that 2AP biosynthesis is primarily linked to the glutamate and proline metabolic pathway, involving both enzymatic and non-enzymatic processes (Imran et al., 2023; Romanczyk et al., 1995; Schieberle, 1990). Within the enzymatic pathway, betaine aldehyde dehydrogenase (BADH2) functions as the central enzyme that controls 2AP production (Bradbury et al., 2008). GAB-ald spontaneously cyclizes into Δ1-pyrroline, the precursor of 2AP, resulting in 2AP formation when BADH2 is mutated and loses its catalytic activity. In the non-enzymatic pathway, certain amino acids are converted into the key 2AP precursor, pyrroline-5-carboxylate (P5C). Glutamate is transformed into P5C through spontaneous cyclization catalyzed by Δ1-pyrroline-5-carboxylate synthetase (P5CS), which is then further converted into 2AP (Huang et al., 2007; Huang et al., 2008; Imran et al., 2023). Glutamate represents a primary driver in the formation of 2AP. Romanczyk et al. (1995) were the first to directly demonstrate, using 13C and 15N isotope labeling experiments, that glutamate and proline serve as the direct nitrogen source precursors for 2AP biosynthesis. Isotope tracing is an effective approach for investigating the metabolic precursors of 2AP. Huang et al. (2008), using 15N-labeled glutamine, established that the nitrogen in 2AP originates from this amino acid. Later, it was shown that no-tillage practices significantly upregulate crucial glutamate-metabolism genes, resulting in higher levels of glutamate and, consequently, a pronounced increase in the 2AP content of fragrant rice (Huang et al., 2024). P5CS is a bifunctional enzyme with both glutamyl kinase and glutamate-5-semialdehyde dehydrogenase activities, catalyzing the ATP-dependent phosphorylation of glutamate and the NADPH-dependent redox reaction, respectively (Hu et al., 1992; Rai & Penna, 2013). As a key enzyme in plant glutamate and proline metabolism, P5CS converts glutamate into P5C, playing a vital role in plant stress responses. Its catalytic product, P5C, also serves as an essential substrate for 2AP synthesis. A study comparing P5CS gene expression, P5C content, and P5CS enzyme activity in callus tissues of aromatic and non-aromatic rice varieties revealed that upregulation of OsP5CS expression increases P5C levels, thereby promoting 2AP production (Huang et al., 2008). Furthermore, experiments involving OsP5CS1 overexpression confirmed a significant increase in 2AP in transgenic plants, providing direct evidence to clarify the molecular mechanism of 2AP biosynthesis (Kaikavoosi et al., 2015). The development of high-aroma coconut varieties requires a deep understanding of the molecular and genetic mechanisms underlying aroma metabolism, especially regarding the key aroma compound 2AP. However, most research on related crops has focused on physiological traits, while studies on the metabolic regulatory enzymes involved in aroma formation remain limited, significantly hindering the molecular breeding of high-aroma varieties. The lack of fundamental research has resulted in inconsistent quality of coconut seedlings in the market, making it difficult for raw fruit to meet demand and significantly hindering industry development. In our preliminary work, genome-wide association analysis revealed that 2AP metabolism in aromatic coconuts is regulated through multiple pathways linked to glutamate homeostasis, highlighting the non-enzymatic P5C pathway as a critical branch. In this study, we aim to identify the P5CS gene family in coconut and systematically investigate the key role of CnP5CS in flavor formation. We examine this from the perspectives of enzyme activity, metabolic flux, population genetic variation, and transcription factor regulation, with the goal of providing both theoretical and practical foundations for breeding high-aroma coconut varieties. Materials and methods Material collection and 2AP determination The study utilized a genetically diverse collection of aromatic coconut accessions exhibiting uniform tree vigor. All materials were planted at the coconut plantations of the Coconut Research Institute, Chinese Academy of Tropical Agricultural Sciences (19°32′ N, 110°46′ E), where the aromatic coconut variety was first introduced from Thailand in 1998. Harvested leaf tissues were promptly snap-frozen in liquid nitrogen and maintained at −80 °C until further analysis. For sample pretreatment, a portion of the leaves was ground into powder under liquid nitrogen, and 1 g of this powder was mixed with 2 mL of anhydrous ethanol. The mixture was sealed and subjected to ultrasonic extraction at 80 °C for 240 min using an ultrasonic cleaner. After filtration through a 0.22 μm organic syringe filter into a headspace sample vial, the extract was subjected to gas chromatography–mass spectrometry (GC–MS) for quantification of 2AP. The GC operating conditions were as detailed below: helium (He) carrier gas (purity >99.999%), a flow rate of 1 mL.min−1, and injection volume of 1 mL. The temperature protocol commenced with an initial isothermal step at 45 °C for 1 min, followed by a linear ramp at 5 °C.min−1 to 100 °C. Subsequently, the temperature was rapidly increased at 300 °C.min−1 to a final temperature of 280 °C. The following MS parameters were applied: electron ionization (EI) at 70 eV, an ion source temperature maintained at 230 °C, and a full-scan acquisition over the m/z range of 20–280. To investigate the genomic underpinnings of 2AP aroma biosynthesis, we conducted a selective sweep analysis using resequencing data from 168 aromatic coconut germplasms cultivated in plantations. Regions under natural selection in the coconut genome were identified using inter-population nucleotide diversity (θπ ratio) and the fixation index (FST) (Danecek et al., 2011; Vatsiou et al., 2016). Population π and FST values were calculated with VCFtools, and the results were visualized using R (Yin et al., 2021). Gene annotation and expression analysis The protein sequences of the target genes were retrieved from the coconut reference genome (GCA_008124465.1_ASM812446v1) and uploaded to KOBAS (http://kobas.cbi.pku.edu.cn) for functional annotation and enrichment analysis, using Arabidopsis thaliana as the reference species. After evaluating significance, the candidate genes were systematically classified into biological process, cellular component, and molecular function categories, and mapped to metabolic and signaling pathways in the KEGG database, providing a basis for further analysis (Ashburner et al., 2000; Kanehisa et al., 2023). RNA-seq data for relevant genes were extracted from different 2AP-producing germplasm resources and visualized as heatmaps using the TBtools software plugin, with expression quantified by FPKM (Chen et al., 2020). Determination of glutamate and its derivatives Mature aromatic coconut (AL, rich in 2AP) and non-aromatic coconut (NAL, lacking 2AP) plants were selected, and leaf samples were collected at three developmental stages: apical new leaves (young tissue), middle leaves (mature tissue), and basal old leaves (senescent tissue). The harvested biological specimens were promptly snap-frozen in sterile liquid nitrogen and maintained at a constant −80 °C for comprehensive subsequent evaluation. Metabolites were extracted using 80% methanol (v/v) containing 0.1% formic acid, with ultrasonication-assisted cell disruption. The extracts were then centrifuged at 12,000 rpm for 15 min at 4 °C. The supernatant was collected, filtered through a 0.22 μm organic membrane, and analyzed using ultra-performance liquid chromatography-high-resolution mass spectrometry. Metabolite relative abundances were semi-quantitatively compared based on mass spectrometry peak areas (Gika et al., 2014). A principal component analysis (PCA) model was constructed using metabolite content data to assess overall metabolic differences between the AL and NAL populations. Spearman and Pearson correlation analyses were conducted to determine correlation coefficients (R) and P-values between metabolites. To account for multiple comparisons, the false discovery rate (FDR) method was applied for adjustment, thereby enhancing the statistical reliability of the findings (Tang et al., 2023). Identification of the P5CS family and analysis of physicochemical properties We retrieved the coding sequences (CDSs), protein sequences, and genomic annotation information of the P5CS (Δ1-pyrroline-5-carboxylate synthetase) gene family from the NCBI database. The test species included coconut (Cocos nucifera L., GCA_008124465.1), Arabidopsis thaliana L. (GCF_000001735.2), rice (Oryza sativa L., GCF_001433935.1), grape (Vitis vinifera L., GCA_000003745.2), peach (Prunus persica L.) Batsch, GCF_000346465.2), oil palm (Elaeis guineensis Jacq., GCF_000442705.1), date palm (Phoenix dactylifera L., GCF_009389715.1), and areca (Areca catechu L., GCA_021397845.1). In Arabidopsis thaliana, the P5CS gene family encodes the rate-limiting enzyme responsible for Δ1-pyrroline-5-carboxylate (P5C) biosynthesis and comprises two partially divergent members, P5CS1 and P5CS2 (Székely et al., 2008). The protein sequences of AtP5CS1 and AtP5CS2 were used as queries to identify homologous sequences in coconut and other species. Final identification of P5CS gene family members was completed by confirming the presence of the unique PLN02418 domain, which is characteristic of these proteins. This conserved domain served as the criterion for establishing the final set of family members. The identified P5CS genes in coconut and other species were then named based on homology analysis with those of Arabidopsis thaliana. To characterize the physicochemical properties of P5CS family members, we performed a comprehensive analysis using several bioinformatics tools. The ExPASy website (https://www.expasy.org/) was used to predict molecular weight (MW), theoretical isoelectric point (pI), instability index (II), aliphatic index, and grand average of hydropathicity (GRAVY) for each protein (Gasteiger et al., 2003). WoLF PSORT (https://wolfpsort.hgc.jp/) and CELLO (http://cello.life.nctu.edu.tw/) were used to cross-validate protein subcellular localization predictions (Horton et al., 2007). The STRING database (https://string-db.org/) was employed to identify recurrent neighboring genes and construct the protein–protein interaction network (Szklarczyk et al., 2023). Evolutionary analysis of the P5CS family The phylogenetic tree was constructed with the Neighbor-Joining algorithm in MEGA11 (https://www.megasoftware.net/), following multiple sequence alignment (Saitou & Nei, 1987; Tamura et al., 2021). The Poisson correction model was applied for nucleotide substitution. Missing data were treated by pairwise deletion, and the node support was measured via bootstrapping (1000 resampled datasets). Additionally, MEME (https://meme-suite.org/meme/) was used to analyze conserved motifs, and NCBI's Batch CDD tool (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) was employed to predict conserved domains (Bailey et al., 2015). The generated phylogenetic tree was visualized and refined using ITOL (https://itol.embl.de/), and TBtools was used to integrate and display phylogenetic relationships alongside motifs, conserved domains, and gene structure information (Chen et al., 2020; Letunic & Bork, 2021). To assess the evolutionary selection pressure acting on members of the P5CS family, the synonymous (Ks) and nonsynonymous (Ka) substitution rates, along with the Ka/Ks ratio, were computed using the Simple Ka/Ks Calculator plugin in TBtools (Hurst, 2002). The Ka/Ks ratio serves as an indicator of selection type: values exceeding 1.0 imply positive selection, those below 1.0 denote purifying selection, and a value of 1.0 signifies neutral evolution. Functional analysis of CnP5CS1 and Cis-acting element analysis AlphaFold3 was used to model the three-dimensional structure of the P5CS protein and predict potential binding sites for small-molecule ligands (Jumper et al., 2021). To identify potential genetic polymorphisms influencing phenotypic variation, we used the R package geneHapR to conduct haplotype-phenotype association analysis in the genomic region surrounding the CnP5CS1 gene (Zhang et al., 2023). TBtools was used to extract the 2000-bp promoter sequences upstream of the transcription start sites for each gene, based on genome annotation and full-length sequences. The PlantCARE online tool (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/) was then employed to systematically scan these promoter regions and identify key cis-regulatory elements (CREs), including light-responsive and hormone-responsive elements (Lescot et al., 2002). We used the PlantTFDB database (https://planttfdb.gao-lab.org/) to predict transcription factor families that could bind to these CREs and constructed a preliminary regulatory network (Jin et al., 2017). R software was then used to visualize the CREs and their associated transcription factors, while TBtools was employed to map the chromosomal distribution of CREs across the genome. Expression and purification of CnP5CS1-his fusion protein BL21(DE3) competent cells were transformed with the recombinant plasmid, heat shocked, allowed to recover, spread onto antibiotic-containing LB plates, and incubated upside down overnight. The next day, a single colony was picked and inoculated into 3 mL of LB liquid medium for cultivation, followed by induction with IPTG for 2–3 h. Samples were collected, centrifuged, and analyzed by SDS-PAGE to assess protein expression. The strain showing high expression was selected and subcultured at a 1:100 dilution into 350 mL of LB medium. The culture was grown at 37 °C until the OD600 reached approximately 0.6, and then induced with 0.3 mM IPTG and incubated overnight at 20 °C with shaking at 200 rpm. Cells were harvested, washed with PBS, resuspended in buffer, and disrupted twice using a pressure homogenizer. The lysate was centrifuged to separate the supernatant and inclusion body pellet. The supernatant was incubated with activated affinity resin at 4 °C overnight. The resin was then washed and eluted with elution buffer to obtain the purified soluble protein. The inclusion body pellet was washed and sonicated multiple times, then dissolved in lysis buffer containing 8 M urea and DTT. After centrifugation, the supernatant was incubated with resin at room temperature for 2–4 h, washed, and eluted with urea-containing elution buffer. The eluted protein was renatured through gradient dialysis, and the final purified protein was subjected to Coomassie staining for quality control. Determination of P5CS enzyme activity, P5C, and Glu content in aromatic coconuts P5CS enzyme activity in aromatic coconuts was measured using the Plant P5CS ELISA Detection Kit (Beijing Solarbio Science & Technology Co., Ltd.), and glutamate (Glu) content was determined using the Plant Glutamate Assay Kit (Suzhou Grace Biotechnology Co., Ltd.). P5CS activity was quantified by monitoring the change in NADPH absorbance at 340 nm. Similarly, the relative Glu level was determined by measuring the dynamic increase in the characteristic 340 nm absorption peak, based on the catalytic reaction of glutamate dehydrogenase (GDH). Pyrroline-5-carboxylate (P5C) content was determined following the standard colorimetric method. Briefly, 0.1 g of fresh sample was weighed, mixed with 1 mL of 3% sulfosalicylic acid, ground into a slurry, and centrifuged at 12,000 rpm for 15 min. A 375 μL aliquot of the supernatant was combined with 500 μL of 10% trichloroacetic acid and 125 μL of 2-aminobenzaldehyde, mixed thoroughly, and incubated at room temperature in the dark for 30 min. After the reaction, the mixture was centrifuged at 8000 rpm for 10 min to remove precipitates, and 300 μL of the supernatant was transferred to a 96-well plate for absorbance measurement at 440 nm. The absorbance values were then used to calculate the P5C concentration in the sample based on the standard curve and the molar extinction coefficient. Y1H library preparation and bait vector construction A yeast one-hybrid system (Y1H) was utilized to analyze adult leaves, carpellate flowers, male flowers, and 9-month post-anthesis endosperm of Wenye 4, a fragrant coconut variety (Fu et al., 2023). Total RNA was extracted using TRIzol reagent (Invitrogen, 15596018) and reverse-transcribed into cDNA with MMLV reverse transcriptase (Invitrogen, 28025013), followed by purification and fragmentation for downstream applications. To construct the Y1H cDNA library, purified cDNA fragments were ligated into the pGADT7 vector, and the resulting homologous recombination products were purified using Acryl Carrier (Solarbio, SA1020) through low-temperature precipitation. Electroporation was used to introduce the recombinant vector into Escherichia coli DH5α competent cells. After recovery in SOC medium, a library bacterial suspension was prepared by adding 20% glycerol. A 10 μL aliquot of the library was diluted 10,000-fold and spread onto an LB plate containing ampicillin (100 μg/mL), followed by incubation at 37 °C for 16 h to count colonies. A total of twenty-four colonies were randomly selected for colony PCR, and the size distribution of the inserted fragments was analyzed by 1.5% agarose gel electrophoresis to assess library quality. The bait vector pHIS2-P5CS1-pro was constructed by cloning the CnP5CS1 promoter sequence (P5CS1-pro) into the pHIS2 vector using EcoR I/Sac I restriction sites. Following transformation of the recombinant plasmid into E. coli competent cells, positive clones were selected for PCR verification and sequencing confirmation. The verified bait plasmid was transformed into the Y187 yeast strain using the lithium acetate method, and transformants were plated on SD/−Trp medium for selection. To determine the optimal concentration of 3-amino-1,2,4-triazole (3AT) for screening, yeast strains carrying the bait plasmid were spotted onto SD/−Trp/-His medium containing a 0–100 mM gradient of 3AT. The interaction system composed of pHIS2-p53 and pGAD53m served as the internal positive standard. An optimal 3AT concentration of 30 mM was identified by monitoring growth after three days, effectively suppressing background growth while allowing detection of positive interactions. Screening of P5CS1 promoter-interacting proteins High-quality plasmid DNA was extracted from an LB-Amp library culture (OD600 = 1) using the High Pure Maxi Plasmid Kit (Tiangen, DP116). The library plasmid and bait vector pHIS2-P5CS1-pro were co-transformed into the Y187 yeast strain, and the transformants were plated on SD/−Trp/−Leu/-His medium supplemented with 30 mM 3AT, then incubated at 30 °C for 3–5 days. Positive clones were purified through secondary streaking, and single colonies were selected and cultured in SD/−Trp/−Leu liquid medium for expansion. Plasmids were then extracted, and the inserted fragments were analyzed by sequencing. Random clones from positive monoclonal colonies were subjected to colony PCR amplification, and the purified products were sent for sequencing. The resulting sequences were mapped to the coconut reference CDS to identify the coding genes of candidate proteins that interact with the bait sequence P5CS1-pro. The corresponding monoclonal colonies of these interacting proteins were then inoculated onto SD/−Trp/−Leu, SD/−Trp/−Leu/-His, and SD/−Trp/−Leu/-His plates supplemented with 30 mM 3AT for retransformation validation of positive clones. Gene function annotation and expression analysis The target protein sequences were retrieved from the coconut reference genome and submitted to KOBAS (http://kobas.cbi.pku.edu.cn) for functional annotation and enrichment analysis, using A. thaliana as the reference species (Bu et al., 2021; Xie et al., 2011). The candidate genes were systematically categorized into different classes based on significance evaluation, and then mapped to metabolic and signaling pathways in the KEGG database, providing a foundation for further analysis. RNA-seq data of relevant genes from coconut accessions with varying 2AP levels were visualized using the heatmap plugin in TBtools. Results Genomic selective sweep analysis of aromatic coconut populations Integrative selective sweep analysis (θπ ratio and FST) of 168 aromatic coconut accessions identified 96 candidate genes within the top 5% of overlapping regions exhibiting strong selection signatures. Accessions were grouped by 2-acetyl-1-pyrroline (2AP) content into high- and low-aroma types (Table S1; Fig. 1A-B). GO enrichment revealed strong overrepresentation of genes involved in light signaling, stress responses, hormone pathways, development, and lipid metabolism. Functional categorization implicated membrane and photosynthetic components, kinase-mediated signaling, redox regulation, and energy metabolism. Consistently, KEGG analysis highlighted pathways associated with environmental adaptation, hormone signaling (ABA, cytokinin), and lipid metabolism (Fig. 1C). Genomic selection sweep analysis of high/low aroma populations. (A) Manhattan plot displaying the distribution of π ratios across the genome (π_low-aroma/π_high-aroma). Each point represents a 100 kb sliding window, with the vertical axis indicating the π ratio and the horizontal axis showing the physical position of the window on the chromosome. (B) Genome-wide distribution of Fixation Index (FST). The height of the Y-axis corresponds to the FST value, with the X-axis position represents the genomic coordinates. Chromosomes are alternately colored for distinction. (C) Functional enrichment bar plot of selected genes. Candidate genes were obtained from the intersection of the top 5% genomic regions identified by both the π ratio and Fst selection pressure detection metrics, including Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). (D) KEGG pathway enrichment bubble plot of candidate genes. (E) A Sankey diagram visualizes the genes contained in each pathway classification. (F) 2AP differential germplasm gene expression profiles. Use colour to represent expression values and perform clustering using a distance matrix. KEGG pathway analysis indicated that the top 5% overlapping genes were predominantly enriched in amino acid metabolism (including glycine, serine and threonine metabolism; lysine biosynthesis; cysteine and methionine metabolism; branched-chain amino acid degradation; and β-alanine metabolism), energy metabolism (TCA cycle, 2-oxocarboxylic acid metabolism, oxidative phosphorylation, pentose and glucuronate interconversions, and lipid metabolism), and signaling/stress-response pathways (plant hormone signal transduction, MAPK signaling, and glutathione metabolism) (Fig. 1D; Table S2). Core genes under selection were significantly enriched in pathways governing amino acid homeostasis, stress adaptation, metabolism, development, and hormone and light-mediated signaling (Fig. 1E; Table S3). Expression profiling revealed marked transcriptional divergence of selected genes among aromatic germplasms (AL1, AL2, and AL3) differing in 2- 2 AP content (Fig. 1F; Table S4). Amino acid metabolism-related genes, particularly the aspartate semialdehyde dehydrogenase-coding gene ASADH, showed a positive correlation trend with 2 AP content. ASADH is a key enzyme in aspartate metabolism, serving as a common precursor node for the biosynthesis of multiple amino acids. It may work together with other amino acid metabolism-related genes to maintain amino acid homeostasis, which could potentially influence the metabolic flux or precursor availability for 2AP synthesis. The genomic selection results reflect the evolutionary strategies of the population in adapting to environmental stress, optimizing resource allocation, and regulating growth and development. The differences in characteristic aroma content among aromatic coconut accessions may be influenced by the synergistic effects of multiple pathways, including amino acid metabolism and stress response. Analysis of glutamate metabolic characteristics in aromatic coconuts 2AP is one of the volatile organic compounds in aromatic coconuts and a key volatile responsible for their aroma. Amino acids serve as important precursors for aroma components, particularly 2AP synthesis. To investigate the contribution of amino acid homeostasis to the key aroma components of aromatic coconuts, we collected apical new leaves, middle leaves, and basal old leaves from mature aromatic coconuts (AL, containing 2AP) and non-aromatic coconuts (NAL, lacking 2AP), and constructed amino acid metabolic profiles. Differences were observed in glutamate metabolism between aromatic coconuts and non-aromatic coconuts. The PCA matrix constructed based on the metabolic profiles of glutamate and its derivatives (glutamine and pyroglutamate) effectively distinguished the aromatic and non-aromatic germplasms into two separate confidence intervals, indicating that these compounds could serve as characteristic biomarkers differentiating the two groups (Fig. 2A; Table S5). The metabolic levels of glutamate and its derivatives showed significant correlations. Specifically, glutamate and glutamine exhibited a positive correlation at P = 9.5e-05 (R = 0.79), while glutamate and pyroglutamate showed a positive correlation at P = 0.00013 (R = 0.78) (Fig. 2E-F). Analysis of 2AP-related amino acid metabolism. (A) Principal component analysis (PCA) matrix based on the metabolic profiles of glutamic acid and its derivatives (glutamine and pyroglutamic acid). The materials were selected from the apical new leaves (ALT, NALT), middle-aged leaves (ALM, NALM), and basal old leaves (ALB, NALB) of mature aromatic coconut (AL) and non-aromatic coconut (NAL). (B—D) Spatial distribution characteristics of amino acids in coconut leaves. (B) Glutamic acid. (C) Glutamine. (D) Pyroglutamic acid. (E-F) A correlation scatter plot reflecting the linear correlation between amino acids. The correlation coefficient (R) represents the relationship between two random variables, and the P-value is used to assess statistical significance. (E) Glutamate vs. glutamine scatter plot. (F) Glutamate vs. pyroglutamine scatter plot. The metabolite content exhibits a distinct spatial gradient in coconut palms, with higher metabolic levels in the apical young leaves and lower levels in the basal old leaves (Fig. 2B-D). The elevated levels of glutamate and its derivatives in the apical young leaves reflect high metabolic activity, likely associated with demands for cell division, elongation, and organ development. In contrast, the basal old leaves undergo physiological functional decline, reduced photosynthetic capacity, and enter senescence, leading to decreased levels of glutamate and its derivatives and a near-halt in metabolic activity. In aromatic coconuts, the content of glutamate and its derivatives in the apical young leaves is significantly higher than in non-aromatic coconuts, while the opposite trend is observed in the middle leaves, with no significant differences in the old leaves. The differential expression of glutamate metabolism in apical young leaves and middle leaves may reflect their distinct roles in the source-sink relationship. The glutamate metabolites are maintained at high levels, driven by the strong sink demand of the apical young leaves. The additional resource competition from the 2AP biosynthesis pathway further enhances metabolite accumulation in the young leaves of aromatic coconuts. As a source, the middle leaves of aromatic coconuts divert part of their glutamate to the 2AP synthesis pathway, resulting in a relatively lower precursor pool content. These findings reveal the potential synergistic regulatory mechanism of glutamate homeostasis in the source-sink allocation during aroma formation in coconut palms. Identification and physicochemical characterization of P5CS members P5CS (Δ1-pyrroline-5-carboxylate synthetase) is involved in glutamate metabolism and exhibits catalytic activity promoting the biosynthesis of P5C (Δ1-pyrroline-5-carboxylate) from glutamate (Fig. 3A). In this study, we systematically identified P5CS family in various crops through BLAST homology searches and PFAM domain analysis, using the protein sequences of two P5CS members (AtP5CS1 and AtP5CS2) from Arabidopsis thaliana as query sequences. The results revealed that two P5CS homologous genes were screened from the coconut (Cocos nucifera) genome, designated as CnP5CS1 and CnP5CS2. Similarly, rice (Oryza sativa), grape (Vitis vinifera), peach (Prunus persica), and oil palm (Elaeis guineensis) each harbored two P5CS homologous copies, while date palm (Phoenix dactylifera) possessed three copies, and areca palm (Areca catechu) contained only one copy (Table S6). Identification of P5CS family members. (A) Ontology enrichment of P5CS interaction network gene sets. (B) Physicochemical property analysis of P5CS members. Includes Protein length (aa), Molecular Weight (kD), Theoretical pI, Instability Index, and Aliphatic Index. (C) The ratio of nonsynonymous (Ka) to synonymous (Ks) substitution rates (Ka/Ks) between CnP5CS and homologous genes in other monocot and dicot species. (D) Subcellular localization prediction. (E-H) Phylogenetic and structural analysis of the P5CS family. (E) Phylogenetic tree. (F) Motif types. (G) Composition of conserved domains. (H) Gene structure. Analysis of the physicochemical properties of the identified P5CS-encoded proteins revealed that family members share similar characteristics (Fig. 3B, Table S6). P5CS members contain 612 to 768 amino acids (aa) (with CnP5CS1 having 645 aa and CnP5CS2 containing 768 aa), with molecular weights ranging from 66.53 to 83.99 kDa and theoretical pI between 5.67 and 8.14. Most P5CS proteins exhibit weak hydrophilicity, with grand average of hydropathicity (GRAVY) scores below 0. The instability index (II) values (23.35–36.26) indicate moderate stability, while the aliphatic index (100.97–113.03) suggests high thermostability, reflecting the preservation of stable hydrophobic core structures during evolution. To evaluate the selective pressures acting on P5CS genes, we calculated the ratio of nonsynonymous (Ka) to synonymous (Ks) substitution rates (Ka/Ks) between CnP5CS genes and the homologs in other species (Fig. 3C). The results showed that the Ka/Ks ratios were predominantly clustered within a low range of 0.07 to 0.37, all significantly below 1, indicating that P5CS genes have undergone strong purifying selection during evolution with highly conserved protein functions. Subcellular localization predictions indicated that P5CS gene products localize to chloroplasts (chlo) and cytoplasm (cyto) (Fig. 3D), implying their potential functional similarity in growth, development, and stress response processes. Phylogenetic and structural analysis of the P5CS family To summarize the molecular evolution and phylogenetic characteristics of the P5CS family, we constructed an phylogenetic tree based on amino acid sequences. The results showed that the P5CS members of dicotyledonous and monocotyledonous plants formed distinct branches (Fig. 3E). In the dicot branch, P5CS1 and P5CS2 from peach, Arabidopsis, and grape cluster together by species, showing high sequence similarity and strong evolutionary conservation. Similarly, in monocots, rice P5CS1 and P5CS2 exhibit a comparable clustering pattern within monocotyledonous plants. Notably, P5CS1 and P5CS2 from Arecaceae plants (coconut, oil palm, date palm, and areca nut) formed separate branches, suggesting more pronounced functional divergence. All dicotyledonous P5CS genes retained all 10 motifs intact with consistent arrangements, while monocotyledonous P5CS genes from coconut and date palm exhibited deletions in motif9 or motif10 (Fig. 3F). Domain analysis revealed that P5CS genes from all species contained the family-characteristic PLN02418 or PLN02418 superfamily domains (Fig. 3G). Additionally, the exon length and number of P5CS family were highly conserved, but the introns of P5CS genes in coconut, oil palm, date palm, and areca were generally longer, reflecting the unique evolutionary trajectory of Arecaceae (Fig. 3H). These results highlight the conservation of the P5CS family during evolution and the distinctive divergence features specific to Arecaceae. CnP5CS1 enzyme activity assay and correlation analysis with metabolites P5CS is a bifunctional enzyme that catalyzes the two-step conversion of glutamate (Glu) to P5C, a key intermediate in 2AP biosynthesis. We successfully constructed the three-dimensional structural model of the CnP5CS1 enzyme and identified its potential active binding sites using AlphaFold3-based protein structure prediction technology. The P5CS enzyme binds ATP via its glutamate kinase domain (GK domain) to catalyze the phosphorylation of Glu into glutamyl 5-phosphate, followed by the consumption of NADPH via its glutamyl-5-semialdehyde dehydrogenase domain (GPR domain) to catalyze glutamyl 5-phosphate into glutamate 5-semialdehyde, which spontaneously cyclizes to form P5C (Fig. 4A). To investigate the in vitro activity of the CnP5CS enzyme, we constructed an E. coli expression system for heterologous expression and protein purification of the coconut P5CS1 gene (Fig. 4B). Small-scale expression tests revealed a prominently expressed band at approximately 75 kDa, corresponding to the expected size of the target protein. Subsequently, the culture was scaled up and protein expression was induced. SDS-PAGE analysis of the cell lysate after disruption showed that the target protein was predominantly found in the insoluble pellet (inclusion bodies). The pellet fraction was therefore used for purification. The purified protein migrated as a single band with the expected molecular weight, providing reliable material for subsequent enzyme activity assays. The activity assay demonstrated that the CnP5CS1 protein possesses in vitro catalytic activity. Moreover, the enzyme activity increased with increasing protein concentration and gradually approached saturation at higher enzyme concentrations (Fig. 4C). Analysis of CnP5CS1 variants and enzyme activity. (A) Predicted 3D structural model of CnP5CS1, with the glutamate kinase domain (GK domain) and glutamyl-5-semialdehyde dehydrogenase domain (GPR domain) labeled. The GK domain binds ATP to catalyze the phosphorylation of glutamate (Glu) into glutamyl-5-phosphate (Glu-5-P), followed by the GPR domain utilizing NADPH to reduce Glu-5-P into glutamyl-5-semialdehyde (GSA), which spontaneously cyclizes to form P5C. Green dashed boxes indicate the active binding regions. (B) SDS-PAGE analysis of expression and purification of CnP5CS1-His fusion protein. The gel sections from left to right are small-scale expression test (lanes 1–2), lysis product analysis (lanes 3–4), and purified product (lane 5). M: protein molecular weight marker (kDa); 1: uninduced whole cell lysate; 2: IPTG-induced whole cell lysate; 3: supernatant of induced cell lysate; 4: pellet of induced cell lysate; 5: Ni-NTA affinity chromatography purified elution product (P5CS-His). (C) In vitro enzyme activity assay of CnP5CS1. CK represents the reaction system without enzyme protein, while 1–5 represent reaction systems supplemented with enzyme protein at concentrations of 2, 4, 8, 12, and 16 μg/mL, respectively. Each treatment was performed with three biological replicates. (D) The expression pattern of CnP5CS1 in germplasms with different 2AP contents. (E-G) The scatter plot reflects the linear correlation of metabolites in the CnP5CS1-mediated catalytic reaction. The correlation coefficient (R) represents the relationship, and the P-value is used to assess statistical significance. (E) A scatter plot of the correlation between CnP5CS1 expression level and 2AP content. (F) A scatter plot of the correlation between P5CS enzyme activity and Glu substrate level. (G) A scatter plot of the correlation between P5CS activity and P5C metabolite level. (H) A principal component analysis (PCA) scatter plot based on P5CS enzyme activity, glutamate, and P5C metabolic levels. (I) The distribution of π ratios in the region near CnP5CS1 in the genome. (J) Haplotype block linkage disequilibrium analysis. (K) Comparative analysis of 2AP content among different haplotypes. (L) qRT-PCR analysis of CnP5CS1 expression levels in different haplotypes. Three biological replicates and three technical replicates were performed for each haplotype. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) CnP5CS1 was upregulated in high-2AP germplasms and exhibited differential expression among the three germplasms AL1, AL2, and AL3 with varying 2AP contents (Fig. 4D). Correlation analysis further confirmed a significant positive correlation (R = 0.87) between CnP5CS1 expression levels and 2AP content at P = 0.0025 (Fig. 4E). Subsequently, we conducted metabolic correlation analyses between the P5CS enzyme and its substrate Glu as well as its product P5C (Figs. 4F-G). The results demonstrated an extremely strong and highly significant correlation (R = 0.73, P = 4.7e-05) between P5CS enzyme activity and Glu substrate levels, as well as a significant positive correlation (R = 0.45, P = 0.026) between P5CS and P5C. Changes in P5CS enzyme activity effectively reflect the strength of upstream and downstream metabolic fluxes. The P5CS enzyme, along with Glu and P5C, can serve as characteristic markers for detecting the relative metabolic levels of 2AP, the key aromatic compound in aromatic coconuts (Fig. 4H). The selective pressure analysis revealed that CnP5CS1 may be under positive selection pressure, as evidenced by a significantly higher θπ ratio in its chromosomal region compared to adjacent regions (Fig. 4I). The haplotype block constructed based on CnP5CS1 exhibited linkage disequilibrium, a feature confirmed by pairwise SNP correlations (Fig. 4J). To explore genetic variations influencing 2AP content in this region, we extracted relevant SNPs from the linkage region of the association analysis population. The haplotype-phenotype association analysis was conducted by integrating SNP genotypes and 2AP phenotypic data to identify superior allelic variants (Fig. 4K). The results showed that the HapGA genotype at locus S4_3269672 had significantly higher 2AP content than HapGG, with a statistically significant difference at P = 0.032. qRT-PCR analysis results showed that the expression levels of CnP5CS1 differed among germplasms with different haplotypes, with the dominant haplotype exhibiting higher expression (Fig. 4L). The strong correlation between CnP5CS1 and 2AP (along with its key metabolic precursors), combined with the identification of advantageous haplotypes, provides critical evidence for elucidating the molecular mechanisms underlying aroma formation in coconut. Prediction of cis-acting elements and transcription factors in the P5CS gene family Gene expression is modulated by transcription factors through their interaction with cis-regulatory elements located in promoter regions. The P5CS promoter contains a variety of responsive elements, and these predominantly participate in diverse biological functions, including stress adaptation (80 elements), hormone signaling (96 elements), and plant growth and development (336 elements) (Fig. 5A-C). Multiple light-responsive elements, including A-box, G-box, and I-box, as well as circadian elements, are involved in plant growth and development. In terms of hormone response, the promoter region includes ABRE (ABA-responsive element), CGTCA-motif/TGACG-motif (JA-responsive element), TCA-element (SA-responsive element), AuxRR-core (auxin-responsive element), and GARE-motif (GA-responsive element). Moreover, stress-responsive elements including ARE (antioxidant response element), MBS (drought-responsive MYB binding site), TC-rich repeats, and WUN-motif (wound-responsive element) play crucial roles in responding to both abiotic and biotic stresses. P5CS may be finely regulated by multiple transcription factors, thereby participating in the coordinated regulatory network integrating plant growth/development and stress responses. Transcription factor prediction revealed that CnP5CS1 is controlled by a complex and highly specific transcriptional regulatory network, showing significant enrichment in transcription factor families such as ERF (31), bZIP (30), bHLH (28), NAC (14), Dof (12), MYB (11), and WRKY (11) (Fig. 5D). CnP5CS1 may be finely regulated by floral organ development and cell proliferation pathways through its interaction with transcription factors, consistent with the 2AP aroma accumulation pattern during coconut maturation. Analysis of cis-acting elements and transcription factors in the P5CS gene family. (A) Distribution matrix diagram of response elements in the P5CS promoter region. (B) Statistical analysis of the involvement of cis-acting elements in plant growth and development, phytohormone response, and stress adaptation processes. (C) Genomic distribution of cis-acting elements. (D) Transcription factor regulatory network diagram. Y1H identifies proteins interacting with the CnP5CS1 promoter To further investigate the role of CnP5CS1 in aromatic coconut, we constructed a yeast one-hybrid (Y1H) cDNA library using Gateway technology to screen for proteins interacting with the CnP5CS1 promoter sequence. The promoter sequence CnP5CS1-pro was cloned into the pHIS2 vector at the EcoRI/BamHI sites to construct the bait plasmid, with sequencing and agarose gel electrophoresis confirming its consistency with the reference sequence (Fig. 6A). A screening concentration of 30 mM 3-amino-1,2,4-triazole (3AT) was selected for the Y1H assay, as stringent autoactivation tests confirmed that 30 mM 3AT completely suppressed background growth, indicating no activation of the HIS3 reporter gene (Fig. 6B). Large-scale Y1H screening yielded 96 positive single clones (Fig. 6C). After gene amplification, sequencing, and comparison with genomic databases, 32 candidate genes were ultimately identified. These included 4 genes related to transcriptional regulation and RNA processing, 3 genes associated with protein homeostasis, 3 genes involved in stress response and signal transduction, 4 metabolic enzyme-coding genes, 3 genes related to membrane transport and cytoskeleton, 2 genes regulating ribosomal translation, 2 calcium signaling-related genes. Validation experiments confirmed that all candidate gene single clones could grow normally on SD -Trp/−Leu, SD -Trp/−Leu/-His, and SD -Trp/−Leu/-His +30 mM 3AT plates (Fig. 6D). Yeast one-hybrid (Y1H) assay identifies the transcription factors regulating CnP5CS1. (A) Detection of the bait vector. The lanes from left to right are the bait vector plasmid, the bait vector plasmid digested with EcoRI and SacI, and the KB DNA ladder. (B) In the autoactivation assay of the Y1H system, a concentration gradient of 3-amino-1,2,4-triazole (3AT) was set at 0, 10, 20, 30, 40, 50, 75, and 100 mM. 30 mM 3AT completely inhibited background yeast growth, with pHIS2-p53 + pGAD53m serving as the positive control. (C) Y1H screening plate. (D) Screening of positive clones. (E) The correlation heatmap of the key metabolites and candidate genes. The heatmap displays the correlation matrix between characteristic markers (P5CS, P5C, Glu, 2AP) of the 2AP biosynthesis pathway in aromatic coconuts and candidate genes screened through Y1H assay. Blue indicates positive correlations, and yellow represents negative correlations. (F) Molecular docking model for the sequence-specific binding of CnYAB2 and CnP5CS1-pro. (G) Tissue expression profile of CnP5CS1. (H) Tissue expression profile of CnYAB2. (I) A correlation scatter plot reflecting the linear correlation between the tissue expression of CnP5CS1 and CnYAB2. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) We constructed a correlation matrix involving P5CS, P5C, Glu, 2AP, and candidate genes based on the characteristic marker contents in the 2AP pathway of aromatic coconut (Fig. 6E). The results showed that nine genes were correlated with the metabolic levels of the characteristic markers, with YAB2 exhibiting significant positive correlations with multiple markers under significance thresholds. The molecular docking results showed that CnYAB2 can specifically bind to the CnP5CS1-pro sequence through three key amino acid residues, Tyr-116, Trp-147, and His-136, forming a stable protein-DNA complex (Fig. 6F). Next, we investigated the tissue expression patterns of CnP5CS1 and CnYAB2, both of which are highly expressed in female flowers and 10-month-old developing fruits (Figs. 6G-H). Their tissue expression showed a significant positive correlation (R = 0.67, P = 0.00032), and this spatiotemporal co-expression pattern provides compelling evidence for their synergistic regulatory roles in 2AP biosynthesis (Fig. 6I). Discussion The compound 2AP is the characteristic aroma of aromatic coconut varieties, and identifying key enzymes involved in its metabolic regulation and elucidating the underlying molecular genetic mechanisms are crucial for improving flavor quality in coconuts. Using aromatic rice as experimental material, researchers generated Osp5cs1 knockout mutants and OsP5CS1 overexpression lines (Hinge et al., 2016; Kaikavoosi et al., 2015). The Osp5cs1 mutation caused a significant decrease in 2AP accumulation, whereas overexpression of OsP5CS1 led to a substantial increase in 2AP levels compared with wild-type plants. This study marks the first investigation into the potential regulatory role of CnP5CS1 from the tropical woody economic tree coconut in 2AP metabolism. The expression level of CnP5CS1 was significantly upregulated in aromatic coconut germplasms with high 2AP content, and its expression showed a significantly positive correlation with 2AP levels (P = 0.021). Furthermore, subsequent haplotype-phenotype association analysis identified superior allelic variants contributing to 2AP accumulation and provided precise genetic markers for molecular breeding targeting coconut flavor. The characteristic aroma compound 2AP in aromatic coconuts is closely linked to glutamate and proline metabolic pathways (Saensuk et al., 2016; Renuka et al., 2022; Schieberle et al., 1990). Our previous genome-wide association study (GWAS) also confirmed several glutamate and proline homeostasis pathways that contribute to 2AP biosynthesis in aromatic coconuts (Ding et al., 2025). In this study, we further found that regions selected in high and low aroma populations were significantly enriched in amino acid metabolism, suggesting that this non-enzymatic, amino acid-driven pathway may be a key route for 2AP biosynthesis in aromatic coconuts. CnP5CS1 is likely the core enzyme in this pathway, catalyzing the conversion of glutamate to P5C. Upregulation of P5CS can elevate P5C levels, thereby promoting 2AP production (Huang et al., 2008). To deeply elucidate the role of CnP5CS1 in the coconut 2AP metabolic network, we systematically analyzed its gene expression levels, enzymatic catalytic activity, and the metabolic correlations between its substrate (glutamate), product (P5C), and 2AP content. The results demonstrate that CnP5CS1 expression, glutamate levels, and P5C content all exhibit strong positive correlations with 2AP content. Changes in CnP5CS1 activity effectively reflect the flux of upstream and downstream metabolism, making CnP5CS1, glutamate, and P5C reliable markers for monitoring 2AP metabolic levels in aromatic coconuts. Transcription factors (TFs) act as key molecular switches in plant responses to abiotic stress. By specifically recognizing and binding to promoter regions of target genes, they activate or repress transcription, triggering broad physiological and biochemical adaptations (Dhatterwal et al., 2024). To further explore the specific regulatory mechanisms of 2AP metabolism in aromatic coconut, we constructed a Y1H library and identified CnYAB2, a member of the YABBY transcription factor family. YABBY proteins are central to establishing polarity in lateral organs, shaping leaf and floral morphology, and regulating secondary metabolism, including the synthesis of aromatic compounds, such as floral scents and terpenes (Jara-Cornejo et al., 2024; Zhang et al., 2020). CnYAB2 shows positive correlations with multiple 2AP signature markers, suggesting that it may bind the CnP5CS1 promoter to regulate gene expression, thereby contributing synergistically to 2AP biosynthesis in coconut. In rice, the transcription factor OsBZIP60 specifically binds to and acts as a positive regulator enhancing OsP5CS1 expression, consequently increasing 2AP content (Bao et al., 2023). Differences in regulatory factors among species suggest species-specific regulatory mechanisms for the 2AP synthesis pathway. The identification of CnYAB2 provides important clues for understanding the distinct regulatory network underlying aroma formation in coconut. Using gene editing and other approaches to investigate its transcriptional activation function and regulatory role can point the way toward further elucidating the metabolic regulatory mechanism of 2AP aroma in aromatic coconut. The identification of CnP5CS1 and the interacting TFs provides a theoretical basis for elucidating the genetic mechanism of characteristic aroma metabolism in aromatic coconut. Meanwhile, the correlation analysis of gene expression, enzyme activity, and metabolites clarifies the multi-level regulatory pathways involved in the synthesis of characteristic aroma. Our study enhances the understanding of the genetic regulatory mechanisms of 2AP biosynthesis and holds significant application value for the targeted breeding of coconut varieties with aromatic quality. Conclusion We analyzed 2AP variation in an association population to uncover the role of P5CS in coconut aroma. Selected genomic regions reflect adaptation to stress and resource allocation, with aroma differences linked to amino acid metabolism and stress-response pathways. Glutamate metabolism and its derivatives correlate strongly with 2AP, showing spatial gradients across leaf development. The enzyme CnP5CS1, along with its substrate Glu and product P5C, serves as a key metabolic marker, while haplotype analysis identified the high-2AP allele HapGA. Yeast one-hybrid assays revealed the transcription factor CnYAB2, whose expression aligns with CnP5CS1 and 2AP markers, suggesting regulatory control via promoter binding. These findings provide integrated genetic and metabolic insights into 2AP biosynthesis in aromatic coconuts. CRediT authorship contribution statement Hao Ding: Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Xiwei Sun: Investigation, Formal analysis, Data curation, Conceptualization. Jinyao Yin: Investigation, Data curation. Dan Luo: Visualization, Investigation. Cuinan Shi: Visualization, Software, Data curation. Xiaojun Shen: Investigation, Formal analysis. Xiaomei Liu: Visualization, Software, Data curation. Jing Li: Investigation, Data curation. Amjad Iqbal: Data curation. Yaodong Yang: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Formal analysis. Ethics declaration The author(s) declare(s) that the study does not involve humans nor animal subjects. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements This work was supported by the Hainan Provincial Natural Science Foundation of China (325QN462), the National Key Laboratory for Tropical Crop Breeding (No. NKLTCB-ZX06), the Fundamental Scientific Research Funds of the Chinese Academy of Tropical Agricultural Sciences (CATAS-Nos. 1630152023007), and the National Natural Science Foundation of China (Grant No. 32071805). Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochms.2026.100452. Supplementary data Supplementary material Data availability The original contributions presented in the study are publicly available. This data can be found at https://ngdc.cncb.ac.cn/gvm/getProjectDetail?Project¼GVM000816. References Associated Data Supplementary Materials Supplementary material Data Availability Statement The original contributions presented in the study are publicly available. This data can be found at https://ngdc.cncb.ac.cn/gvm/getProjectDetail?Project¼GVM000816.