Pan-genomic and transcriptomic analyses reveal subfunctionalization of CBP/p300-like histone acetyltransferases in soybean seed development

Abstract

Background

Soybean is a crucial global source of protein and oil. The CBP/p300 histone acetyltransferases (HACs) are key transcriptional regulators, yet their diversity and functions in soybean remain unexplored at a pan-genomic level.

Results

Here, we constructed a pan-genomic resource for the HAC gene family across 29 wild, landrace, and cultivated soybean accessions, identifying 142 HAC genes. These genes are confined to chromosomes 7, 8, 15, and 19, indicating strong evolutionary constraints. Phylogenetic analysis divided HACs into five subgroups with distinct domain architectures: Group 4–5 retain full CBP/p300 domains, whereas Group 1–3 show progressive domain loss. Pan-transcriptomic analyses revealed an expression dichotomy: Group 3–5 are broadly expressed, while Group 1–2 exhibit endosperm-specific expression during early seed development, suggesting specialized roles in nutrient transfer and embryogenesis. Notably, elite cultivars (e.g., Wm82, ZH13) have lost Group 2 homologs preserved in wild soybeans, highlighting domestication-driven erosion of epigenetic diversity. Co-expression network analysis prioritized Wm82-HAC1 (Group 1) as a candidate gene coordinating nutrient metabolism and seed maturation pathways.

Conclusion

Our study provides the first comprehensive panorama of epigenetic regulators in the soybean pan-genome. Our findings reveal how subfunctionalization and domestication help shape the HAC regulatory network in soybean, highlighting wild germplasm as a valuable reservoir for recovering lost alleles (Group2 homologs) and identifying Wm82-HAC1 (Group 1) as a prime target for precision breeding of seed traits.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12870-026-09449-y.

Introduction

Histone acetylation, a highly conserved epigenetic feature catalyzed by histone acetyltransferases (HATs), orchestrates chromatin dynamics and gene expression in eukaryotes [1]. While traditionally linked to transcriptional activation, emerging evidence demonstrates that histone acetylation is also integral to gene repression mechanisms [2], or even the result rather than the cause of active transcription [3, 4].

Central to this regulation are CBP and its close paralog p300, which were first identified as transcriptional co-activators that bind to the transcription factor CREB and the adenovirus E1A, respectively, but later found to be endowed with intrinsic acetyltransferase activity [5–7]. The mechanisms by which CBP/p300 regulate transcription can be broadly categorized into three modes of action: (i) Histone acetylation to promote chromatin remodeling, allowing transcription machinery access; (ii) Non-histone protein acetylation to directly regulate transcription factors, chromatin associated proteins and transcription co-factors; (iii) Scaffolding function to bridge transcription factors to the basal transcriptional machinery [8, 9]. CBP/p300 have been evidenced to be central regulators in metazoan organisms and critical for organ development and morphogenesis in humans, mice, flies and worms [7, 8, 10]. Their indispensability is underscored by severe developmental disorders (e.g., Rubinstein-Taybi syndrome) [11, 12] and oncogenesis upon dysfunction in human [7, 13].

In plants, CBP/p300-like genes have been characterized in Arabidopsis thaliana including HAC1, HAC2, HAC4, HAC5 and HAC12 [14]. Of these gene products, HAC1, HAC5 and HAC12 were revealed to possess HAT activity [15]. HAC1 and HAC5 were uncovered to be plant-specific MEDIATOR subunits [9]. Moreover, CBP/p300-like genes were reported to be implicated in multiple developmental processes as well as response to various environmental cues in Arabidopsis, such as heat stress response [16], flowering time [17, 18], ethylene signaling [19], salicylic acid-triggered immunity [20] and leaf senescence [21]. Nevertheless, the genetic diversity and functional landscape of CBP/p300-like genes in major crops, particularly within a pan-genomic context, remain unexplored.

Soybean, a globally vital legume providing plant-based protein and oil [22], is a diploidized paleotetraploid legume crop (2n = 40) that originated from the domestication of Glycine soja (wild soybean) in East Asia approximately 6,000–9,000 years ago [23–26]. This complex genome architecture, coupled with extensive natural variation across wild and cultivated accessions, necessitates a pan-genome analysis to fully capture gene family diversity. Given the high conservation and regulatory roles of CBP/p300 in metazoan organisms [8–10, 27], identifying CBP/p300-like genes within the soybean pan-genome might provide novel information for the regulatory mechanism underlying seed development and stress tolerance, ultimately facilitating the development of improved soybean varieties.

Here, we leverage the soybean pan-genome and pan-transcriptome data spanning 29 accessions to: (i) Systematically identify and characterize CBP/p300-like histone acetyltransferases (HACs); (ii) Decipher their domain architecture, gene structure, conserved motifs, cis-regulatory elements, association with transposable elements, and phylogenetic relationships; (iii) Analyze the expression patterns and co-expression network of HACs via pan-transcriptome data across multiple tissues; (iv). Prioritize candidate HACs with potential roles in seed development and stress adaptation. This study provides the first pan-genomic resource for epigenetic regulators in soybean, highlighting wild germplasm as a reservoir for recovering lost seed development alleles and nominating Wm82-HAC1 for precision breeding.

Results

Pan-genomic catalog of soybean HACs reveals evolutionary constraints

Advances in genome sequencing have fueled the rise of pan-genomics, overcoming the constraints of single reference genomes to reveal the complete genetic diversity within species. The increasing availability of crop pangenomes (e.g., maize, rice, wheat, soybean) provides deep evolutionary and functional insights into gene families, enabling robust discovery of key alleles for important agronomic traits and enhancing crop breeding efforts. To obtain the complete profile of CBP/300-like genes across soybean accessions, the published genome assemblies and annotations of 29 soybean accessions, including 4 wild soybeans, 9 landraces, and 16 cultivars, were collected (Supplementary Table 1) [28]. To identify CBP/p300-like genes across the genome (termed HAC in this study), we performed a BLAST search of the entire genome assembly with Arabidopsis thaliana HAC (AtHAC) protein sequences as the reference, which enabled the discovery of previously unknown HAC sequences in soybean [14, 17]. A total of 142 HAC genes were identified across the 29 soybean accessions. Except for the cultivars SoyL04, Wm82 and ZH13, which contain four homologous HAC genes, all other accessions possess five HAC genes. This indicates a generally conserved HAC gene number across the soybean pangenome, with gene loss detected in a small subset of cultivated accessions. (Figure. 1A and Supplementary Table 1). Moreover, across all examined accessions, HAC genes are confined to chromosomes 7, 8, 15, and 19, with no observed distribution on other chromosomes (Figure. 1B). Synteny analysis of representative wild (SoyW01), landrace (SoyL03), and cultivated (Wm82) genomes confirmed that all HAC loci reside within conserved syntenic blocks, with no evidence of inter-chromosomal translocation (Supplementary Fig. 1). For instance, HAC loci on chromosome 19 maintain collinear gene order despite one copy loss in Wm82. Likewise, HAC loci on chromosomes 7, 8, and 15 show conserved synteny, suggesting genomic stability since soybean polyploidization. The combination of chromosomal confinement and syntenic conservation implies purifying selection maintains HAC genomic positions, potentially due to essential regulatory contexts. However, functional studies are required to confirm the selective forces involved.

Fig. 1.

Identification of HAC genes in the soybean pangenome. A Number of HAC genes in 29 accessions. B Number of HAC genes across chromosomes in 29 accessions

Phylogenetic diversification mirrors functional specialization

To investigate the evolutionary status and potential function of HAC genes in soybean, the 142 identified HAC candidate genes were further validated by InterPro (https://www.ebi.ac.uk/interpro) for the presence of a complete HAT domain. Phylogenetic analysis of 137 soybean HAC proteins together with five Arabidopsis HACs (HAC1, HAC2, HAC4, HAC5, HAC12) resolved five distinct groups (Group1-5; Fig. 2 and Supplementary Table 1) [14, 17]. Clearly, the phylogenetic relationships between soybean and Arabidopsis HACs are complex and do not reflect simple one-to-one orthology. AtHAC2 formed a well-supported clade exclusively with soybean Group2 sequences, suggesting a close evolutionary relationship that may indicate orthology or conserved functional specialization. In contrast, the remaining Arabidopsis HACs (AtHAC1, AtHAC4, AtHAC5, AtHAC12) clustered together in distinct branches. This pattern suggests that these Arabidopsis HACs are more closely related to each other than to any individual soybean group, implying that much of the HAC gene family diversification in soybean occurred after the divergence of the Arabidopsis and soybean lineages (Fig. 2 and Supplementary Table 1). Remarkably, single-copy HACs in the cultivated varieties ZH13 and Wm82 were maintained in all groups except Group2, revealing domestication-driven selection purged Group 2 in elite cultivars (Figs. 1 and 2). Moreover, HAC genes from the same chromosome often cluster together phylogenetically, suggesting a potential link between genomic location and evolutionary relationship. For example, most Group5 members are located on chromosome 8, while Group4 members are predominantly found on chromosome 7. However, an exception is W05-HAC4-2, which resides on chromosome 8 but clusters within Group4 based on sequence homology. This observation raises interesting questions about the evolutionary history of this locus, and further analysis incorporating genomic synteny and regulatory data would be needed to fully understand the relationship between chromosomal context and HAC gene evolution.

Fig. 2.

Phylogenetic classification of soybean and Arabidopsis HAC proteins. The complete amino acid sequences of 137 soybean and five Arabidopsis HAC proteins were included. The HAC proteins in soybean were classified into five phylogenetic groups marked with different colors. The phylogenetic tree was constructed using MEGA X software via the neighbor-joining algorithm and then beautified using the iTOL web tool (https://itol.embl.de/)

To investigate sequence variations of HACs among the soybean accessions, the protein length and domain architecture of HAC genes in the 5 groups were then assessed. The amino acid length of HACs ranged from 250 to 1807 residues across accessions, with an average of 1467 amino acids. The CBP/p300 family of HAT proteins, comprised of large multi-domain proteins, have been reported in mammals and plants (previously identified only in Arabidopsis thaliana), but is absent in fungi [14]. This phylogenetic distribution suggests lineage-specific innovation essential for advanced epigenetic regulation in multicellular eukaryotes. Arabidopsis HAC proteins maintain near-identical domain architectures, with HAC2 representing the sole exception through its loss of the N-terminal TAZ zinc finger 14, 17]. In soybean, HACs exhibit group-specific domain architectures while maintaining intra-group conservation (Fig. 3A and Supplementary Table 2). Consistent with Arabidopsis, Group 4–5 retain the canonical CBP/p300 structure, including one Cys-rich HAT domain, one PHD finger, two TAZ zinc fingers, and two ZZ zinc fingers. However, Group 1–2 lack the N-terminal TAZ zinc finger, whereas Group 3 possesses only one ZZ zinc finger. These group-specific domain reductions suggest distinct evolutionary trajectories of the HAC family in soybean compared to Arabidopsis.

Fig. 3.

Domain architecture, motif composition and gene structure of HAC genes in soybean. A The approximate location of predicted protein domains is marked by boxes of different colors. HAT, Histone acetyltransferase; PHD, plant homeodomain zinc finger; ZnF (TAZ), transcriptional adaptor zinc finger; ZnF (ZZ), zinc finger present in dystrophin and CBP. B Schematic representation of conserved motifs in HAC proteins. The motif patterns were predicted using the MEME database (https://meme-suite.org/meme/tools/meme) and were sorted according to the order in the phylogenetic tree shown in Fig. 2. Boxes of different colors represent different motifs. The black lines represent the non-conserved sequences. C Gene structures of the HAC genes. Yellow boxes: exons; green boxes: UTR; spaces between the boxes: introns. The scale bar of bottom demonstrates the length of protein (A-B) and gene (C), respectively

Group-specific motif patterns and gene structures underpin HAC groups

To further elucidate HAC evolutionary dynamics in soybean, we analyzed motif composition and gene structure across groups. MEME analysis identified 10 conserved motifs across soybean HACs. While motif composition was homogeneous within groups, distinct motif signatures distinguished the five groups (Fig. 3B), implying functional diversification. Notably, Group 3–5 share identical C-terminal motif arrangements (all 10 motifs conserved), but exhibit N-terminal divergence. Specifically, Motif 7 exclusively localizes to the N-terminal in Group 5, while Motif 5 is Group 3-specific in N-terminal regions. Moreover, Group 1–2 display unique motif distributions in both N- and C-terminal, differing fundamentally from Group 3–5. These group-specific motif patterns likely underpin functional subfunctionalization of HACs. Furthermore, phylogenetically related HACs shared conserved motif patterns and exon-intron architectures (Fig. 3C). While Group4-5 maintained highly conserved gene structures, Group3 exhibited notably longer introns based on the available genome annotations. Groups 1–2 appeared to lack untranslated regions (UTRs) in the current annotations, suggesting possible truncated transcript structures or incomplete annotation. However, UTR predictions from genome annotations are computational inferences that may not fully capture true transcript architecture. Experimental validation (e.g., 5’ RACE) would be necessary to definitively determine the UTR boundaries and confirm the absence of UTRs in Groups 1-2 [29]. These structural variations, if validated, could contribute to transcriptional divergence, as non-coding regions often harbor regulatory elements influencing gene expression.

Cis-regulatory and transposable element analyses of soybean HACs

To decode functional divergence among soybean HACs, we systematically mapped the cis-regulatory elements (CREs) within 2 kb promoter regions, which revealed three functionally partitioned CRE modules (Fig. 4A and Supplementary Table 3). The identified CREs spanned environmental responses (light, drought, temperature, wound), hormone signaling (ABA, GA, auxin, SA, MeJA), and development processes (palisade mesophyll cell differentiation, circadian regulation, meristem expression, zein metabolism). Furthermore, these CREs exhibited both conserved and group-specific features. Apparently, while light-responsive elements were universally enriched and wound-responsive elements were systematically depleted across all groups, Group1-2 uniquely harbored zein metabolism, X60K and MYBHv1 elements. Particularly, zeins are seed storage proteins in maize that function as a nitrogen reserve during endosperm development to support germination [30, 31]. This regulatory landscape underscores Group1-2 may specialize in seed nutrient allocation.

Fig. 4.

Cis-regulatory and transposable element analyses of soybean HAC genes. A Cis-regulatory elements (CREs) distribution patterns in 2 kb potential HAC promoters. The heatmap displays the number of different CREs in the potential promoter region of HACs. CREs were categorized into three functional groups: environmental response (light, drought, temperature, wound), hormone signaling (ABA, GA, auxin, SA, MeJA), and developmental processes (palisade mesophyll differentiation, circadian regulation, meristem expression, zein metabolism). B Transposable elements (TEs) associations with HAC genes. The heatmap shows the number of different types of TEs identified in the 2 kb upstream to 2 kb downstream region of HACs

Transposable elements (TEs) serve as fundamental catalysts of genome evolution, driving structural variation, reshaping regulatory landscapes, and furnishing raw material for adaptive evolution [32, 33]. To elucidate their impact on HAC gene family diversification in soybean, we systematically analyzed TEs distribution within 2 kb flanking regions (including gene bodies) of all HAC loci identified in this study. Our analysis revealed that all the 137 HACs in the pangenome were associated with TEs (Fig. 4B). DNA transposons, particularly PIF-Harbinger, hAT, and CACTA, accounted for 4.7% of TEs identified within or close to HAC loci, while retrotransposons (Helitron, Copia, Gypsy, SINE, and LINE) comprised approximately 40.9%. Notably, TE distributions varied significantly among HAC groups. For example, DNA transposons (CACTA) were markedly enriched in Group3-5 and Copia/Gypsy retrotransposons showed preferential enrichment in Group 3, while Group 1–2 overall contains relatively few typical TEs. These results demonstrate group-specific TE associations within the HAC gene family. Nevertheless, the presence of TEs near HAC genes does not constitute evidence that TEs mediated gene family expansion. The observation that most soybean accessions maintain five HAC copies (Fig. 1A) points to other evolutionary mechanisms, including whole-genome duplication (WGD), segmental duplication, or unequal recombination, as more likely drivers of copy number variation [34, 35]. Comprehensive synteny analysis and duplication dating will be necessary to resolve the relative contributions of these processes.

Expression dichotomy defines housekeeping and seed-specific HAC functions

While gene expression patterns provide valuable functional clues, tissue-specific regulation of HAC genes has not been comprehensively characterized at pangenome scale in soybean. To address this, we leveraged published pantranscriptome data spanning 74 tissue samples from 29 soybean accessions [28, 36]. Expression-based hierarchical clustering of HAC genes revealed distinct regulatory architectures among the 5 evolutionary groups (Fig. 5 and Supplementary Table 4). Overall, HAC genes exhibited highly divergent, group-specific expression patterns. Genes in Group3-5 showed strong, constitutive expression across most examined tissues, with particularly elevated expression in reproductive tissues, such as flowers, early-stage seed coats and endosperm, suggesting their involvement in core cellular processes critical for plant growth and reproduction. In contrast, Group1-2 displayed pronounced tissue specificity, with uniquely high expression restricted to the endosperm during early seed developmental stages, including the globular, heart, cotyledon, and early maturation phases. This endosperm-specific expression pattern suggests their potential roles in early seed formation processes, as the endosperm is known to function not only in nutrient transfer but also in developmental signaling that coordinates embryo patterning and growth [37–39]. This expression dichotomy of HAC genes reflects evolutionary subfunctionalization, where ancestral HAC gene duplication and divergence partitioned functions into housekeeping (Group3-5) and seed-specific (Group1-2) regulatory programs. These findings highlight Group1-2 as prime targets for seed quality improvement, while Group3-5 may modulate overall plant growth. Notably, all HAC homologs in Group1 and Group2 are located on chromasome19. Group1 contains single HAC homologs from the cultivated varieties Wm82 (Wm82-HAC1) and ZH13 (ZH13-HAC1), whereas Group2 lacks corresponding homologs in both cultivars. Furthermore, all four wild soybean accessions retain conserved HAC representation in both groups (SoyW01-HAC1, SoyW02-HAC1, SoyW03-HAC1 and W05-HAC1 in Group 1; SoyW01-HAC2, SoyW02-HAC2, SoyW03-HAC2, W05-HAC2 in Group 2). This differential gene retention pattern suggests subfunctionalization followed by group-specific gene loss during soybean domestication. In light of the endosperm-specific expression pattern of Group1 and Group2 homologs, further analysis was conducted to compared the expression levels of these genes in early maturation stage endosperm among the 29 accessions. We found that these homologs exhibit varied expression levels across accessions (Fig. 6 and Supplementary Table 4). Overall, TPM values range from approximately 12 to 24 for Group 1 HAC genes and from 2 to 20 for Group 2 HAC genes, representing roughly 2‑fold and 10‑fold variation between the lowest and highest expressing varieties, respectively. To quantify this variation, we calculated the coefficient of variation (CV) for Group 1 (HAC1) and Group 2 (HAC2) genes. Group 1 showed a CV of 11.5% across 29 accessions, indicating relatively stable expression. In contrast, Group 2 exhibited a CV of 36.4% across 26 accessions, revealing substantially higher inter‑accession variability. This inter-accession expression variation may reflect underlying regulatory diversity among soybean germplasm. However, expression data alone do not establish a genetic basis for regulatory variation, and further analysis is required to explore the causes of this expression divergence.

Fig. 5.

Pantranscriptome analyses of HAC genes in soybean. Heatmap of the developmental expression patterns of HAC genes from the five groups in 74 soybean tissues. Gene expression values were normalized using log₂(TPM + 1) transformation

Fig. 6.

Expression levels of Group 1 (A) and Group 2 (B) homologs in the early maturation stage endosperm across accessions

Co-expression networks nominate Wm82-HAC1 as a seed development hub

Genes participating in shared biological pathways often exhibit coordinated expression patterns. To systematically characterize HAC gene regulatory networks across tissues, we performed weighted gene co-expression network analysis (WGCNA) on 29,150 filtered genes from the Wm82 genome. Hierarchical clustering identified 44 distinct co-expression modules (designated M1-M44), with 11,830 genes remaining unassigned (designated M-gray) and module sizes ranging from 31 to 2366 genes (Fig. 7A). Crucially, module M12 specifically highlighted co-expressed partners of the endosperm-specific Wm82-HAC1, a key candidate for early seed development regulation. Within M12, 511 genes were co-expressed with Wm82-HAC1, among which 3, 7, 7, 14 and 15 genes are involved in response to nutrient, seed maturation, glycosylation, anatomical structure maturation and developmental maturation, respectively (Fig. 7B and Supplementary Table 5). This co-expression clustering positions Wm82-HAC1 as a potential coordinator of seed development pathways, nominating it as a prime epigenetic target for precision breeding of improved seed nutritional and maturation traits.

Fig. 7.

Topological overlap matrix plots for the gene co-expression network in Wm82. A Hierarchical cluster tree showing co-expression modules. The major tree branches constitute 44 modules labeled by different colors. Assigned modules are colored on the bottom; gray genes are not assigned to any module as listed. B Functional annotation of genes co-expressed with Wm82-HAC1

Materials and methods

Identification of HAC genes in soybean pangenome

To identify soybean HAC genes, BLASTP searches (version 2.16.0+) were performed using Arabidopsis HAC protein sequences as queries against the soybean protein datasets of all 29 accessions download from Soyomics (https://ngdc.cncb.ac.cn/soyomics/download)28. The following cutoff thresholds were applied: E-value ≤ 1e-5; sequence identity ≥ 45%; alignment length ≥ 150 amino acids. Initial candidates were further filtered by requiring the presence of the conserved CBP/p300 histone acetyltransferase domain validated using InterProScan.

Phylogenetic analysis of HAC members in soybean and Arabidopsis

The amino acid sequences of soybean and Arabidopsis HAC genes were retrieved and aligned using MEGA11 with ClustalW model. Phylogenetic analysis were conducted by Neighbor-Joining algorithm method. Branch support was calculated using 1000 bootstrap replicates and the Shimodaira-Hasegawa-like approximate likelihood ratio test. Tree annotation was performed with evolview (https://www.evolgenius.info/evolview/).

Analysis of gene structures, conserved motifs and domains

Conserved motif analysis was performed using MEME (https://www.omicsclass.com/article/432) (accessed on 20 July 2025). The analysis was run with the following parameters: maximum number of motifs to find: 10, minimum motif width: 6, maximum motif width: 10. The conserved domain analysis were performed by Web CD-Search Tool (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi). The results of motif and domain compositions, as well as gene structures were visualized with TBtools [40].

Identification of the cis-regulatory elements of soybean HAC genes

Cis-regulatory element (CRE) analysis was performed using the PlantCARE database (https://bioinformatics.psb.ugent.be/webtools/plantcare/html/) [41]. For each HAC gene, 2 kb of genomic sequence upstream of the translation start site was extracted from the genome assemblies of all 29 accessions and submitted to PlantCARE for CREs prediction.

TE identification

Transposable element (TE) annotations for all 29 soybean accessions were retrieved from Soyomics (https://ngdc.cncb.ac.cn/soyomics/download). Only intact TEs were retained for downstream analyses. TEs were overlapped with HAC genes using the bedtools program (v2.30.0) [42].

Pantranscriptome profiling of HACs in soybean

TPM values for each gene and tissue across 29 soybean accessions were downloaded from SoyMD (https://yanglab.hzau.edu.cn/SoyMD/#/). The expression heatmap was generated using R package pheatmap. The tissue sample dendrogram was generated using hierarchical clustering with average linkage.

Co-expression analysis and module identification in Wm82

RNA-seq data for soybean cultivar Williams 82 (Wm82) were downloaded from the NCBI Sequence Read Archive (SRA) under accessions PRJNA246783, PRJNA197251, PRJNA238493, PRJNA246314, and PRJNA246315. The datasets comprised 278 samples covering 50 tissues and developmental stages. Genes with normalized expression level > 2 in at least one tissue were retained for analysis. Gene modules and networks were analyzed using the R package WGCNA (version 1.72) with the following parameters: power = softPower, TOMType = “unsigned”, minModuleSize = 30, reassignThreshold = 0, mergeCutHeight = 0.25, saveTOMs = FALSE. Genes that co-expressed with HACs were retrieved for GO enrichment analysis.

Synteny analysis

Inter-genomic synteny between SoyW01 (wild), SoyL03 (landrace) and Wm82 (cultivated) was analyzed via MCScanX with default parameters, and the results were visualized by TBtools.

Coefficient of variation (CV) calculation

The coefficient of variation (CV) was calculated as the ratio of the standard deviation (SD) to the mean (µ), expressed as a percentage:

×100%

For HAC genes from Group 1 (HAC1) and Group 2 (HAC2), CV values were computed across accessions with TPM values in the early maturation stage endosperm samples (n = 29 for Group 1; n = 26 for Group 2).

Discussion

In this study, we conducted the first soybean pan-genomic analysis of epigenetic regulators, revealing how domestication shaped HAC regulatory networks via subfunctionalization. Wm82-HAC1 is highlighted as a potential target for seed improvement, with wild HAC alleles offering a route to recover genetic diversity. Future directions include identifying HAC acetylation targets in embryogenesis and developing epigenetic breeding strategies. This work extends Arabidopsis foundations [43] to soybean, providing a key resource and testable hypotheses. Empirical validation is the essential next step toward practical application.

Evolutionary subfunctionalization of HACs shapes seed trait regulation in soybean

Our pan-genomic analysis reveals that functional divergence of HAC genes via subfunctionalization may underpin their specialized roles in soybean seed development. The preservation of full-length domains in Group4-5 versus domain loss in Group1-3 mirrors evolutionary trajectories observed in other epigenetic regulators and domesticated gene families [14, 44–46] (Fig. 3A). Such structural simplification likely accompanies functional specialization: broad constitutive expression of Group4-5 supports their role in core cellular processes, while endosperm-specific expression of Group1-2 aligns with their enrichment in seed-specific regulatory elements (Figs. 4A and 5). This dichotomy reflects an adaptive trade-off between pleiotropic functions and tissue-precision regulation, a hallmark of gene family evolution after duplication [46].

In animals, the N-terminal TAZ zinc finger domains of CBP/p300 proteins mediate interactions with transcription factors such as CREB and p53, while ZZ-type zinc fingers contribute to substrate recognition and chromatin binding [7, 8]. The loss of these domains in Group1-3 soybean HACs, particularly the absence of TAZ fingers in Group1-2, suggests these proteins may have evolved alternative protein interaction specificities or adapted to function in distinct chromatin contexts. Similar domain simplification has been documented in other plant epigenetic regulators. For example, certain plant-specific histone deacetylases lack domains present in their animal orthologs, reflecting lineage-specific adaptation [44]. Future biochemical studies examining the interaction partners and chromatin binding preferences of representative HAC proteins from each group would test whether these structural variations indeed translate into functional specialization.

The tissue-specific expression patterns, with Group1-2 exclusively expressed in early endosperm, raise intriguing questions about the regulatory mechanisms underlying this specificity. Our promoter analysis revealed enrichment of zein metabolism elements in Group1-2 (Fig. 4A), which in maize are associated with seed storage protein gene regulation during endosperm development [30, 31]. This suggests that cis-regulatory element co-option may have contributed to the acquisition of seed-specific functions by these HAC paralogs. Such regulatory neofunctionalization following gene duplication has been documented in other plant gene families, including MADS-box transcription factors involved in flower development [46, 47]. The observation that all Group1-2 members reside on chromosome 19 (Fig. 1B) further suggests that chromosomal context, perhaps including local chromatin environments or higher-order genomic architecture, may influence their coordinated regulation.

Domestication-driven erosion of epigenetic diversity

The loss of Group 2 homologs in cultivated soybeans (Wm82 and ZH13; Fig. 2) starkly contrasts with their retention in wild accessions. This parallels findings in classic domestication genes (e.g., tb1 in maize, sh4 in rice), where domestication selected against “wild-adaptive” redundant alleles under cultivation [48]. We propose that Group 2 HAC genes likely regulated energetically costly processes (e.g., embryonic patterning) dispensable in domesticated environments, while retention of Group 1 (e.g., Wm82-HAC1) reflects its essential role in nutrient transfer, a trait prioritized during selection for seed size [49]. Wild soybeans thus serve as genetic reservoirs for “re-domesticating” seed traits via reintroducing lost HAC alleles.

The retention of Group 2 homologs in all four wild soybean accessions examined suggests these genes confer fitness advantages in natural environments. Wild soybeans typically produce smaller seeds with harder seed coats and greater dormancy compared to cultivars, traits that facilitate persistence in variable natural habitats but are counterproductive in agricultural settings [25, 26]. We hypothesize that Group 2 HACs may contribute to these wild-type seed characteristics by regulating genes involved in dormancy establishment or seed coat development. Testing this hypothesis through comparative epigenomic profiling of near-isogenic lines differing in Group 2 presence could reveal the regulatory networks controlled by these lost alleles.

HAC networks as targets for epigenetic breeding

Wm82-HAC1 (Group1) emerges as a candidate for precision breeding. Its endosperm-specific expression, co-expression with seed maturation and nutrient metabolism genes, and selective retention in cultivars position it as a key regulator of seed quality. This specialization may reflect heightened epigenetic complexity required for soybean’s nutrient-rich seed development. We hypothesize that modulating Wm82-HAC1 could enhance carbon partitioning to seeds, akin to OsglHAT1 overexpression boosting rice yield [50]. Future efforts should validate its function via gene editing and explore whether reintroducing wild Group 2 alleles improves yield without compromising other agronomic traits.

Conclusions

This study provides the first pan-genomic overview of epigenetic regulators in soybean, revealing how domestication and subfunctionalization have shaped the HAC gene family. We show that HAC genes are confined to specific chromosomes and divided into five evolutionary groups with distinct expression patterns: Groups 1–2 are specifically expressed in endosperm during early seed development, while Groups 3–5 are broadly active across tissues. Domesticated soybeans have lost Group 2 HAC homologs retained in wild relatives, highlighting a reduction in epigenetic diversity during cultivation. Through co-expression analysis, we identify Wm82-HAC1 as a potential regulator linked to seed maturation and nutrient metabolism, making it a promising target for precision breeding. Our findings underscore the value of wild soybean germplasm in recovering lost genetic diversity and advancing epigenetic improvement of seed traits.

Abbreviations

HAT

Histone acetyltransferase

HAC

CBP/p300-like histone acetyltransferase

AtHAC

Arobidopsis thaliana HAC

TE

Transposable element

UTR

Untranslated region

PHD finger

Plant Homeodomain finger

TAZ finger

Transcription adapter zinc finger

ZZ finger

ZZ-type zinc finger

CRE

Cis-regulatory elements

Wm82

Williams 82

ZH13

Zhonghuang 13

SoyW

Soybean Wildtype

SoyL

Soybean Landrace

SoyC

Soybean Cultivar

Authors' contributions

Conceptualization, Y.C. and Z.D.; data curation, C.W., D.H. and Y.C.; original draft preparation, Y.C. and C.W.; supervision, Z.D. and Y.C.. All authors have read and agreed to the submitted version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (32201799) and Sichuan Provincial Engineering and Technology Research Center for Innovative Development of Bamboo Fiber Nutrition (PETC202508).

Data availability

The data used to support the conclusions of this study were contained in the Materials and methods of this article. All materials are available through corresponding authors upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.