Oysters in transition: hermaphrodite oysters display unique DNA methylation patterns in gill tissue

Abstract

Background

European flat oysters (Ostrea edulis) are sequential hermaphrodites that alternate sex in response to environmental change. Epigenetics, including DNA methylation, are often involved in sex reversal through influencing gene transcription. Knowledge on the epigenetic mechanisms underlying sex reversal in hermaphrodite bivalves is limited to gonadal tissue and previous studies have only compared DNA methylomes of males and females. Therefore, the aim of this study is to assess whether sex-specific DNA methylation can be identified in somatic gill tissue of the flat oyster.

Results

By comparing whole-genome methylomes of 35 oysters of different sex phenotypes using nanopore sequencing, we demonstrate the presence of sex-specific DNA methylation patterns in somatic gill tissue. A total of 9,654 regions and 2,576 genes were differentially methylated between male, female, and hermaphrodite oysters. Functional analysis of differentially methylated genes indicated an association with energy homeostasis and metabolic processes, implying a remodeling of the energy balance.

Conclusions

This study is the first to characterize DNA methylomes of hermaphrodite oysters, providing new insights into the epigenetic mechanisms underlying sex reversal in a sequential hermaphrodite invertebrate. Additionally, this study characterizes sex-specific DNA methylation in somatic gill tissue, paving the way for non-lethal sex identification using epigenetic biomarkers.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-025-11736-1.

Introduction

Epigenetic mechanisms influence gene transcription without changing the DNA sequence itself [1]. DNA methylation, the covalent transfer a methyl group to a DNA nucleotide, is one of the most common epigenetic modifications in eukaryotes and is involved with various developmental and regulatory processes [2, 3]. In vertebrates, DNA methylation occurs primarily on CpG islands where de-methylation often results in gene silencing [4, 5]. Contrasting to vertebrates, invertebrates acquire a more mosaic pattern of DNA methylation, mainly occurring on gene bodies [6, 7]. Reduced gene body methylation has the potential to facilitate transcriptional opportunities by alternative splicing, for example through exon skipping or by providing access to alternative transcription start sites [7, 8]. Growth, development, and environmental conditions can initiate de novo DNA methylation or change its maintenance [4, 9]. By influencing gene transcription, DNA methylation forms the crucial ingredient in facilitating phenotypic plasticity, whereby organisms can change their phenotype in response to changing environments [8]. Understanding the epigenetic mechanisms that are at the core of the interplay between the environment and phenotypic plasticity is pivotal for understanding how species respond to or cope with environmental change.

One prevalent example of phenotypic plasticity is sex reversal, by which an animal is capable of changing its phenotypic sex without the need for a corroborative change in DNA. Sex reversal occurs in many lineages of the animal kingdom, such as in reptiles [10], amphibians [11], fish [12], and a wide variety of invertebrates [13–17]. The plasticity and frequency with which animals can reverse sex are partly determined by the underlying sex determination mechanisms, which can be roughly divided into genetic sex determination (GSD) and environmental sex determination (ESD). The ability to reverse sex can be advantageous in highly dynamic environments, where sex-related differences in fitness may favour transition into one sex under specific environmental conditions [18–21]. During sex reversal, epigenetic mechanisms play a pivotal role in facilitating sex differentiation by influencing gene expression [8, 22–24].

Oysters represent excellent candidates for studying epigenetic mechanisms underlying sex reversal since they are sequential hermaphrodites that can repeatedly alternate between male and female sex phenotypes [15, 25, 26]. A variety of environmental factors have been documented to initiate sex reversal in oysters by altering DNA methylation (e.g. acidification [27] and starvation [28]), in the gonads. Differential DNA methylation between phenotypic sexes, referred to as sex-specific DNA methylation, are suggested to lead to sex differentiation by influencing gene transcription [14, 28–31]. Generally, energetically demanding or restricting conditions lead to masculinization of gonadal tissue [28, 32]. The association between DNA methylation and sex reversal has been primarily studied in gonad tissue, where sex-specific DNA methylation and gene expression are expected to have a direct relation to gametogenesis [33, 31]. In the Pacific oyster (Magallana gigas), differential methylation and expression of several genes are shown to be involved with sex differentiation in the gonads [29–31, 34]. Interestingly, during the process of sex reversal, oysters often display intermediate sex stages where they simultaneously produce male and female reproductive tissue of different developmental stages [15]. These individuals are characterized as hermaphrodites that display either male or female biased gonad tissue, or represent gonadal tissue containing both sexes with equal ratios [15, 35]. The simultaneous occurrence of male, female, and hermaphrodite individuals within one population is suggested to represent a phenotypically plastic response to temporally dynamic environments [36]. Although sequential hermaphroditism is frequently observed in oysters and hermaphrodite individuals contribute largely to reproductive cohorts in wild populations [36], previous studies have been exclusively comparing sex-specific DNA methylation between males and females [29, 30, 37]. Hence, sex-specific DNA methylation of hermaphrodite individuals remains poorly understood [15, 30, 36–38]. However, understanding how DNA methylation may differ between male, female, and hermaphrodite individuals is crucial for understanding the role of DNA methylation during different stages of sex reversal in oysters [38].

Currently, sex reversal in oysters is understudied, since existing methods of sex-identification require sampling of gonad tissue and are therefore lethal. Notably, sex-specific DNA methylation is not restricted to gonadal tissue and accommodates sex-associated differences in somatic tissues (e.g. mantle and abductor muscle) and even whole-body methylomes of different invertebrates [14, 39, 40]. This highlights the relevance of characterizing differential DNA methylation in somatic tissues between different sex phenotypes in oysters [14]. Gill tissue, along with other somatic tissues, is often collected during biosecurity or monitoring surveys where the tissue can be sampled non-lethally, by using anaesthetics [41, 42]. As such, identifying sex-specific methylation in somatic tissue would pave the way for non-lethal alternatives of sex-identification.

Like many other bivalves, the European flat oyster (Ostrea edulis; Linnaeus, 1758) is a sequential hermaphrodite that can reverse sex within one reproductive season [15, 25]. It plays a critical role in the ecosystem by sustaining biodiversity through generating biogenic reef habitats [43–45]. The species is currently considered ‘threatened’ according to the Oslo-Paris Convention, due to ongoing habitat degradation, overfishing, and the spread of diseases [46–49]. Recently, its decline in population stocks initiated restoration efforts to restock degraded populations with hatchery-produced oysters [44, 50, 51]. However, male-biased skewed sex ratios limit reproductive output and are thought to hamper the upscaling of oyster culturing used for kick-starting oyster reef restoration [35, 52]. Identifying epigenetic mechanisms that underly sex reversal is crucial for a better understanding of the reproductive strategy of O. edulis and can facilitate in improving culturing strategies of the species [32, 38].

Therefore, this study aimed to assess whether sex-specific DNA methylation is present in somatic gill tissue, and to characterize the regional differences in sex-specific DNA methylation between male, female, and hermaphrodite flat oysters (O. edulis). By comparing gill methylomes using nanopore sequencing we show the presence of regional sex-specific DNA methylation across the genome. This study is the first to include DNA methylation of hermaphrodite individuals, providing new insights into the epigenetic mechanisms underlying sex reversal in a sequential hermaphrodite invertebrate. Additionally, this study highlights the potential of using somatic tissues to study sex reversal for non-lethal applications.

Results

Whole genome DNA methylomes obtained from gill tissue of flat oysters were compared between male, female, and hermaphrodite sex stages, with the aim of characterizing sex-specific DNA methylation. Oyster sex was identified using histology and five sex stages were distinguished: male (M), female (F), hermaphrodite predominantly male (HPM), hermaphrodite predominantly female (HPF), and hermaphrodite both sexes equally present (HBS). In total, whole-genome sequencing of 35 flat oyster individuals resulted in 111 million reads, with an average of 3.2 million reads per individual (Supplementary Table S1). Sequencing of all oysters was performed on 8 different 10.4 PromethION flow cells. Mapping of reads with minimal quality of 60 to the O. edulis reference genome (XbOstEdul1.1), resulted in an average of 12.2 Gb per individual with read lengths varying between 1.1 −14.6 Kb and an average mapping percentage of 86.9% (Supplemental Table S1; submitted to ENA under accession number PRJEB85091). No significant difference in number of sequenced base pairs was observed between the five sex phenotypes (one-way ANOVA, F statistic = 0.872, p-value = 0.49; Supplementary Fig S1A). After CpG calling using modkit, between 10.4 and 30.5 million CpGs per individual, with a minimum coverage of 3, were used for downstream analysis.

General methylation patterns

Analysis of differentially methylated loci (DML) resulted in 555,259 CpG loci that were differentially methylated between any of the five sex phenotypes, with minimal coverage of 3 for all individuals. These CpGs were used as input for PCA analysis representing all sites that were differentially methylated between any two of the five sex phenotypes (Fig. 1A). PC1 (13.1% variance explained) shows clear separation between the five sex phenotypes with a gradient from male and female, to the three hermaphrodite sex phenotypes (HPM, HPF, and HBS).

Fig. 1.

Comparative analysis of differentially methylated loci (DMLs) between five sex phenotypes of flat oysters (M, F, HPM, HPF, and HBS). A Principal component analysis of all differentially methylated CpGs between any two of the five sex phenotypes (555,259 sites). B Distribution over genomic location of CpGs (minimal coverage of 3), DMLs identified from comparisons between any two of the five sex phenotypes, and DMLs from pairwise comparisons between the five sex phenotypes

Differentially methylated loci and regions between sex phenotypes

Characterization of DMLs between all five sex stages resulted in a total of 1506 loci in which minimally two sex phenotypes showed significant differential methylation (BH corrected q-value < 0.01, ≥ 20% methylation difference; Supplementary Table S2). The pairwise comparison between HPM and HBS resulted in the highest number of DMLs (27,363 DMLs; 13,366 hypermethylated in HBS, and 13,997 hypermethylated in HPM; Supplementary Fig. S2A). Between M-HPF 4136 DMLs were identified, representing the lowest number of DMLs from all pairwise comparisons, followed by 4252 DMLs between F-HPF (Supplementary Fig. S2A).

The majority of DMLs were located in introns (56.6%), followed by intergenic regions (20.4%), exons (16.0%), and promoters (7.0%; Fig. 1B). The genomic distribution of significant DMLs from pairwise comparisons between sex phenotypes differed significantly from the genomic distribution of all CpGs (minimal coverage of 3 for all individuals; Pearson’s correlation: p-value < 0.05; Supplementary Table S3). Here, significant DMLs found between pairwise comparisons showed a positive association with intronic regions, and a negative association with intergenic regions.

Characterization of differentially methylated regions (DMRs) using a sliding-window approach resulted in a total of 21,160 significant DMRs from pairwise comparisons between any two of the five sex phenotypes (BH corrected q < 0.01, ≥ 20% methylation difference; Supplementary Table S4). Merging DMRs that were within 100 bp proximity of each other obtained a total of 9,651 DMRs that varied in length of 201 to 4001 base pairs (Supplementary Table S4). Generally, hermaphrodite sex phenotypes HPM and HBS showed higher numbers of DMRs than pairwise comparisons with or among males, females, or HPF. The highest number of DMRs was found between HPM and HBS (3369 DMRs, 1829 hypermethylated in HPM, and 1540 hypermethylated in HBS; Supplementary Fig. S2B). The lowest number of DMRs weas observed between females and HPF (164 DMRs; 84 hypermethylated in females, and 80 hypermethylated in HPF). Methylome comparison between males and females resulted in a total of 239 DMRs (125 hypermethylated in males, 114 hypermethylated in females). Two examples of DMRs between sex phenotypes have been visualized (Fig. 2A,B).

Fig. 2.

Differentially methylated regions (DMRs) and genes (DMGs) from gill methylome comparisons between five sex phenotypes (M, F, HPM, HPF, and HBS) in flat oysters. A DNA methylation profile of all 35 flat oyster individuals at the position of gene LOC125645858. Vertical bars represent methylation percentages per CpG site from 0–100%. Black boxes indicate DMR areas. B DNA methylation profile of all male and female oyster individuals at the position of gene LOC125647918. Vertical bars represent methylation percentages per CpG site from 0–100%. C Upset plot showing the number of overlapping and unique DMGs (n = 2,576) identified from pairwise comparisons between the five sex. D Hierarchical clustering of overlap in DMGs (n = 2,576) between the five sex phenotypes using Pearson’s correlation coefficient

Gene annotation of differential methylation

Identifying differentially methylated genes (DMGs), genes overlapping with DMRs, resulted in a total of 2,576 unique DMGs between pairwise comparisons of all five sex phenotypes (Fig. 2C). Of all these genes, 828 unique DMGs were identified between HPM and HBS. Males, HPM, and HBS showed to have the highest overlap in DMGs (Fig. 2). Clustering analysis also revealed highest similarity in DMGs between M-HBS, and HPM-HBS. Additionally, F-HPM, M-HPM, and HPM-HPF also showed high similarity in identified DMGs, but clustered as a separate group with unique DMGs (Fig. 2D). The DMGs with highest difference in average methylation percentage between pairwise comparisons were: MAM and LDL-receptor class A domain-containing protein 2-like (LOC125683627; HPM-HBS), UDP-GalNAc:beta-1,3-N-acetylgalactosaminyltransferase 2-like (LOC125659768; F-HPM and F-HPM), ATP-dependent DNA helicase Q5-like (LOC125679612; M-HBS), and sacsin-like (LOC125679280; HPM-HBS; Table 1). A complete overview of the top 10 most hypermethylated and hypomethylated DMGs identified for all pairwise comparisons can be found in the supplementary material (Supplementary Table S5).

Table 1.

Overview of the top most hypermethylated and hypomethylated DMGs for pairwise comparisons between all five sex phenotypes (M, F, HPM, HPF, and HBS)

Gene ID	Gene Description	Pairwise Comparison	Pseudochromosome	Length	Methylation (%)	Adjusted qvalue
LOC125652448	transport and Golgi organization protein 6 homolog	M-F	NC_079164.1	1301	-53.3680433178866	2.96208951245896E-052
LOC130051844	ester hydrolase C11orf54 homolog	M-F	NC_079164.1	301	37.2582417582418	1.61137197226376E-021
LOC125645560	proton-coupled zinc antiporter SLC30 A2-like	M-HPM	NC_079169.1	1701	-54.8091058670907	8.10987876182568E-037
LOC130050362	uncharacterized LOC130050362	M-HPM	NC_079172.1	301	47.4035563592526	6.92625857610277E-037
LOC125652448	transport and Golgi organization protein 6 homolog	M-HPF	NC_079164.1	401	-42.175698181644	1.27453457293131E-046
LOC130051844	ester hydrolase C11orf54 homolog	M-HPF	NC_079164.1	301	37.2582417582418	1.61137197226376E-021
LOC125679612	ATP-dependent DNA helicase Q5-like	M-HBS	NC_079165.1	901	-60.2998006773845	8.6947215528774E-042
LOC125670462	protocadherin Fat 4-like	M-HBS	NC_079167.1	201	52.0976094204441	1.67199085281488E-031
LOC125659768	UDP-GalNAc:beta-1,3-N-acetylgalactosaminyltransferase 2-like	F-HPM	NC_079172.1	301	-50.8465939360249	6.72229257813295E-043
LOC125655710	uncharacterized LOC125655710	F-HPM	NC_079170.1	201	61.8869420254375	4.9771803906424E-070
LOC125675975	uncharacterized LOC125675975	F-HPF	NC_079166.1	301	-40.6276149185132	1.85859901577164E-033
LOC125658752	arsenite methyltransferase-like	F-HPF	NC_079172.1	201	40.3920477137177	7.05313397251047E-032
LOC125675975	uncharacterized LOC125675975	F-HBS	NC_079166.1	301	-50.5105617059029	1.61397292093311E-031
LOC125679490	putative polypeptide N-acetylgalactosaminyltransferase 10	F-HBS	NC_079165.1	201	51.5837104072398	8.53880056942331E-040
LOC125660846	ATP-binding cassette sub-family B member 6-like	HPM-HPF	NC_079171.1	301	-47.2298409261188	2.49588780513731E-018
LOC125659768	UDP-GalNAc:beta-1,3-N-acetylgalactosaminyltransferase 2-like	HPM-HPF	NC_079172.1	201	56.2110240874834	1.77573266151024E-055
LOC125679280	sacsin-like	HPM-HBS	NC_079165.1	1201	-62.4605706501535	1.14454321578361E-048
LOC125683627	MAM and LDL-receptor class A domain-containing protein 2-like	HPM-HBS	NC_079169.1	601	62.6613682616021	4.54624205126091E-032
LOC125681288	uncharacterized LOC125681288	HPF-HBS	NC_079164.1	201	-51.4323730742718	1.01037750392091E-039
LOC125673062	uncharacterized LOC125673062	HPF-HBS	NC_079166.1	201	49.1092278719397	1.14601071580134E-032

In total, 125 genes were found to be differentially methylated between male and female oysters, of which 62 DMGs were hypermethylated in males, and 63 DMGs were hypermethylated in females (Supplementary Table S6). When compared to DMGs identified between other sex phenotypes, 15 DMGs were uniquely differentially methylated between males and females (Supplementary Table S6). DMGs with highest difference in methylation percentage between males and females were: tRNA (guanine-N(7)-)-methyltransferase non-catalytic subunit wdr4-like (LOC125645763), uncharacterized LOC125655458, uncharacterized LOC125673983, where males were hypermethylated compared to females. DMGs with highest difference in average methylation percentage and females showing hypermethylation compared to males were: uncharacterized LOC125661226, uncharacterized LOC125661253, and periodic tryptophan protein 1 homolog (LOC125674140). Other DMGs uniquely identified between male and female oysters were: cyclic AMP-dependent transcription factor ATF-2-like (LOC125646345), inhibitor of nuclear factor kappa-B kinase subunit alpha-like (LOC125668035), neuronal calcium sensor 2-like (LOC125669759), PR domain zinc finger protein 4-like (LOC125680421), and zinc finger protein 474-like (LOC130054705).

Functional enrichment analyses

Analysis of difference in average methylation percentage over all DMRs within a DMGs resulted in 21 DMGs with a difference in average methylation percentage > 45%. DMGs with highest difference in average methylation percentage were proton-coupled zinc antiporter SLC30 A2-like (LOC125645560; HPM-HBS), MAM and LDL-receptor class A domain-containing protein 2-like (LOC125683627; HPM-HBS), LOC125655710 (F-HPM), and probable nuclear hormone receptor HR3 (LOC125661857; M-HPM & F-HPM), ATP-dependent DNA helicase Q5-like (LOC125679612; Me-HBS), and protein ILRUN-like (LOC125645979; M-HBS; Fig. 3B). Differential methylation between HPM-HBS was observed in the lysosome membrane protein 2-like gene (LOC125645858), belonging to the superfamily of ABC transporters [53].

Fig. 3.

Functional annotation of differentially methylated genes (DMGs) identified from pairwise comparisons of the five sex phenotypes (M, F, HPM, HPF, and HBS). A Heatmap of Gene Ontology (GO) of DMGs (n = 2,576) enriched for Biological Processes (BP; n = 128). BP were clustered hierarchically with Pearson’s correlation coefficient. B Heatmap of DMGs with an average difference in methylation percentage > 45%. Positive differences in average methylation percentage indicate hypomethylation for the first group per pairwise comparison, negative differences in average methylation percentage indicate hypermethylation for the first group per pairwise comparison. C Scatter plot of enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways for DMGs identified between the five sex phenotypes

Functional enrichment analysis of Gene Ontology (GO) terms of annotated DMGs resulted in 128 significant enriched GO terms for Biological Processes (BP; Fig. 3A), 45 enriched GO terms for Cellular Components (CC; Supplementary Figure S4), and 64 enriched GO terms for Molecular Function (MF; Supplementary Figure S5). GO terms for Biological Processes were related to neuronal activity and locomotion (e.g. GO0061564: axon development, GO0048666: neuron development, GO0040011: locomotion, and GO0006935: chemotaxis), protein targeting and localization (e.g. GO0006605: protein targeting and GO0015031: protein transport) and cellular and metabolic processes (e.g. GO0030030: cell projection organization, GO0051641: cellular localization, GO0009117: nucleotide metabolic process, and GOGO0006793: phosphorus metabolic process). Processes related to stress response were enriched in DMGs found between males and HBS (GO0043087: regulation of GTPase activity, GO0033554: cellular response to stress, and GO0042221: response to chemical”).Additionally, DMGs found between males and HPF were enriched in the “semaphorin-plexin signalling pathway” (GO0071526, GO1902285, GO1902287). Lastly, unique GO terms found between HPM-HPF were related to meiotic recombination (GO0140527, GO0007131, and GO0035825).

Regarding CC, the largest group of DMGs showed enrichment for “intracellular anatomical structure” (GO0005622) and “organelle” (GO0043226; Supplementary Figure S3. DMGs enriched in MF were mostly related to pathways associated with nucleotide binding (e.g. GO0005488: binding, GO0005524: ATP binding, GO0030554: adenyl nucleotide binding, and GO0032559: adenyl ribonucleotide binding”; Supplementary Figure S4). Here, DMGs identified between HPM and HBS contributed mostly to these enriched pathways.

Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of DMGs revealed 25 enriched pathways, of which 2 significant (p_adjust < 0.05; Fig. 3C). The most significantly enriched pathways were involved in “Cytoskeleton in muscle cells” and “purine metabolism”. Enriched pathways were mostly related to energy homeostasis due to their association with lipid, amino acid, and carbohydrate metabolism (e.g. “Purine metabolism”, “Ascorbate and aldarate metabolism”, “Beta-Alanine metabolism”, “Inositol phosphate metabolism”, and “Nicotinate and nicotinamide metabolism”). In addition to metabolism related pathways, DMGs from pairwise comparisons also showed enrichment for processes related to innate immunity (e.g. “ABC transporter” and “Peroxisome”; Fig. 3C).

Discussion

In invertebrates, DNA methylation is thought to contribute to phenotypic plasticity by increasing transcriptional variation in various ways (e.g. alternative splicing, increasing mutation rate) [8]. The mosaic gene-body methylation is suggested to create transcriptional opportunities rather than activate or inhibit transcription through (de)methylation in the promoter region [8]. Sex reversal, a phenotypically plastic trait whereby animals can reverse sex in response to environmental change, is associated with DNA methylation where differences between sexes are thought to facilitate this reversal [11, 22, 29, 54–57]. Sex-specific DNA methylation has been widely reported in both vertebrates and invertebrates [22, 58–62]. In oysters, differential DNA methylation between male and female individuals is suggested to be associated with sex differentiation in the gonads, where epigenetic differences are linked to differential expression of genes involved in the sex determination pathway [28–30]. Although oysters are sequential hermaphrodites, knowledge on DNA methylation profiles of hermaphrodite individuals, undergoing a sex change, is limited. In this study, we aimed to characterize sex-specific DNA methylation of male, female and hermaphrodite European flat oysters (Ostrea edulis), by comparing gill methylomes obtained through nanopore sequencing. In general, differential methylation occurred more frequently in intragenic regions (promoters, introns, and exons) than in intergenic regions. These findings are consistent with results from other oyster species, where differential methylation is enriched in intergenic regions [63–65]. The most distinct DNA methylation profile was observed in hermaphrodite oysters, whereas males and females demonstrated more similar patterns of DNA methylation in their gills. Functional enrichment analysis indicated that differentially methylated genes (DMGs) were significantly enriched in pathways related to energy homeostasis and metabolism. Collectively, our data suggests that sequential hermaphrodites display an alternative state that represents the epigenetic reprogramming required for sex reversal. Additionally, the observed presence of sex-specific DNA methylation in gill tissue highlights the potential for non-lethal sex identification using somatic tissues.

Hermaphrodite individuals display an alternative epigenetic state in the gills

When oysters undergo sex reversal, gametogenesis of both sexes can occur within the same follicle [15]. In these hermaphrodites, discrepancies in gamete development most often lead to asynchronous spawning of male and female gametes [15]. In the gonads, the reversal of sex is accompanied by epigenetic reprogramming that facilitates cell growth, development, regulation, and restructuring, especially during early gametogenesis [30, 66]. In this study, whole-genome gill methylome comparisons between sex phenotypes demonstrated that hermaphrodites display the most distinct DNA methylation patterns when compared to males or females. Significantly enriched GO terms overlapping DMGs found between hermaphrodites and other sex phenotypes were related to neuronal activity and locomotion, and cell morphogenesis pathways. Additionally, the KEGG pathway related to “Cytoskeleton in muscle cells” was significantly enriched between hermaphrodite oysters. These findings suggest that hermaphrodite individuals undergoing sex reversal display an alternative state of epigenetic reprogramming when compared to males or females. It is possible that males and females, that have undergone the full process of sex reversal, represent an epigenetic state that matches the completion of sex reversal. Similarly, in a study of Sun et al. (2024), genome-wide DNA methylation patterns of gonadal tissue of Magallana gigas were accurately remodeled to the opposite sex after the completion of sex reversal [30].

In addition to the differential methylation patterns observed between male, female, and hermaphrodite individuals, numerous unique GO terms were identified between the different hermaphrodite sex phenotypes (HPM, HPF, and HBS). The majority of differentially methylated loci and regions (DMLs and DMRs) were identified between HPM and HBS. DMGs overlapping those regions showed significant enrichment for processes related to “cellular localization”, “intracellular signal transduction”, “phosphorylation” and “phosphate metabolism”. The differences in annotated GO terms identified across hermaphrodite sex phenotypes imply that the direction of sex reversal is also associated with differential DNA methylation. In M. gigas, a genome-wide increase in DNA methylation was observed when oysters reversed sex from female to male [30]. While further elucidation of epigenetic reprogramming in hermaphrodite phenotypes is necessary to validate this, the highest number of DMLs and DMRs found between HPM-HBS suggests that the direction of sex reversal is associated with different epigenetic states. This supports the hypothesis that although sex reversal is bidirectional, patterns of epigenetic reprogramming may be different depending on the direction of a sex reversal event.

Sex-specific DNA methylation of gill tissue is related to energy homeostasis and metabolism

In this study, functional enrichment analysis of DMGs resulted in highest significant enrichment for GO-terms related to metabolic or energy homeostasis processes (e.g. ATP binding, nucleotide binding, hydrolase activity, DNA damage sensor activity). Similar to our findings, differential methylation and expression between prespawning and postspawning oysters was enriched for metabolism and energy homeostasis pathways in gill tissue of M. gigas [67]. KEGG enrichment analysis of DMGs resulted in enriched pathways related to lipid, amino-acid, and carbohydrate metabolisms (e.g. “purine metabolism”, “ascorbate and aldarate metabolism”, “beta-Alanine metabolism”, “nicotinate and nicotinamide metabolism”). The purine metabolism plays an essential role in energy supply, where it is involved in synthesis, degradation, and recycling of nucleic acids [68]. In the mussel Mytilus coruscus, a reduction in energy consumption in the gills was observed after starvation exposure, where proteins related to purine metabolism and nicotinic and nicotinamide metabolism pathways were downregulated [68]. Additionally, the inositol phosphate metabolism pathway, enriched in DMGs identified between M-HSB and HPM-HBS, contributes to metabolic signaling, and cellular reprogramming and growth in oysters [69, 70]. Similarly, GO terms related to innate immunity, cell proliferation, and metabolic and developmental processes were enriched in transitionary transcripts of sex reversing wrasses [23]. Gametogenesis requires excessive amounts of energy due to enhanced cell growth, cellular restructuring, and reprogramming [71, 72]. Therefore, it is likely that oysters undergoing sex reversal require a remodeling of the energy balance in all their tissues, reflecting processes of cell restructuring and rebuilding. The sex-specific differences in genes related to energy homeostasis and metabolism observed in this study, could be explained by a remodeling of the energy balance due to changes in energy requirement as a consequence of sex reversal.

Changes in immunity and stress response as a consequence of sex-specific DNA methylation

In addition to changes in DNA methylation of genes related to energy homeostasis and metabolism, DMGs found between the different sex phenotypes were also enriched in pathways related to immunity, stress, and toxin response. ATP-binding cassette (ABC) transporters play a key role in the defensive mechanism against pathogens in bivalves [73, 74]. Furthermore, peroxisomes are crucial in toxin response mechanisms [75]. Lysosome membrane proteins are involved in the cellular defensive response and vital for the production of phagolysosomes [76]. Phagocytosis can have multiple functions in bivalves, namely enhanced innate immune response, or energy recycling by the reabsorption of damaged cells, which can be initiated by sex reversal [77]. Upregulation of the phagocytosis pathway was observed in gill proteomes after starvation exposure in the mussel M. coruscus [68]. One probable explanation of the observed differences in DNA methylation is that oysters of different sex phenotypes adopt different metabolic and immune response strategies depending on their energy requirements for reproduction. Notably, gametogenesis can lead to enhanced susceptibility to pathogens due to increased energetic demands required for reproduction [78]. During gametogenesis, reduced phagocytosis activity of hemocytes has also been reported in oysters [79]. Similarly, immunity and stress response pathways were found to be enriched in differentially expressed genes identified between pre- and postspawning oysters [67]. Postspawning oysters are reported to experience a decrease in global immunity defense and become more vulnerable to chemical stress, due to energy loss after spawning [67, 80]. However, this study only assessed the differences in sex-specific DNA methylation, therefore, statements on how sex-specific DNA methylation in gill tissue affects functionality of these genes remains speculative and should be interpreted with caution.

Somatic sexual dimorphism as a consequence of DNA methylation differences between male and female oysters

Although comparison of DNA methylation between male and female oysters yielded the lowest number of DMRs and DMGs, 125 DMGs could still be identified of which 15 were uniquely differentially methylated between males and females. Most of these unique DMGs were related to metabolic, signaling, or expression related processes. Guanine-N7-methyltransferases are involved in modifications of mRNA caps and required for mRNA processing and functioning [81]. Cyclic adenosine monophosphate (cAMP) activity is involved in regulating glycogen breakdown, and suggested to be related to the control of gonad maturation [82]. Another unique DMG found between male and female oysters was inhibitor of nuclear factor kappa-B kinase (IKK) subunit alpha-like. IKKs play a central role in cell signalling and are involved with innate immunity in oysters [83]. Besides the unique DMGs identified between males and females, other DMGs with high difference in methylation percentage between male and female oysters are associated with pathways that regulate somatic sexual dimorphism in other animals. For example, the serine/arginine repetitive matrix protein (SRRM) was observed to be expressed in the gonads and has been allocated as a candidate sex-identification marker in a shrimp species [84]. In Drosophila melanogaster, SRRM1 is involved in genitalia development and somatic sex determination [85, 86]. Additionally, Transport and Golgi organization protein 6 homolog (TANGO6) is associated with mating behaviour in seahorses, where differential expression in somatic brain tissue was observed during mating [87]. In zebrafish, sex-specific DNA methylation was observed in genes associated with the arsenic metabolism, which was suggested to lead to sex-specific behavioural changes [88]. Likewise, the arsenic metabolism appears to be gender-specific in rats [89]. Another interesting gene that was differentially methylated between males and females was prostaglandin E synthase 3-like. Prostaglandins are involved with ovary development and steroidogenesis in other bivalves [90]. DMGs related to the semaphorin-plexin signaling pathway were solely enriched between M and HPF. Semaphorin proteins are involved in immune system regulation and are crucial for nervous system development [91]. Additionally, semaphorin proteins were discovered to be involved in the development of neurons that secrete gonadotropin-releasing hormones in vertebrates, which are essential for reproductive development [91]. The ECM-receptor interaction pathway was solely enriched between F-HPF. Extracellular matrix (ECM)-receptor interactions were shown to be enriched in skin transcriptomes of female fish compared to males upon exposure to sex steroids [92]. Lastly, MAM and LDL-receptor class A domain-containing protein 2-like are involved in ovarian development in clams [90]. These examples indicate that even though sex reversal mostly impacts gonadal tissue, somatic tissues are not left unchanged and sex reversal can induce sex-specific processes of cell restructuring or reprogramming in somatic tissues accordingly. Sexual dimorphism is not as profound in oysters as in some other animals, however, indicators of sexual dimorphism are present (e.g. differences in growth or innate immunity [67]). Based on the findings of this study, we hypothesize that differential DNA methylation between male and female oysters is associated with sexual dimorphism in somatic tissues as a consequence of sex reversal. However, understanding the link between sex-specific DNA methylation and sexual dimorphism requires further elucidation of the effects of sex-specific DNA methylation on gene functioning.

Tissue specificity of sex-specific DNA methylation in oyster gills

The occurrence of sex-specific DNA methylation has been reported in various animals and tissue types [14, 61, 93–95]. However, literature is inconsistent on the existence of tissue specificity for sex-specific DNA methylation as some studies clearly show altered sex-specific DNA methylation in different tissue types [14, 29], whereas other studies demonstrate consistent sex-specific whole-body methylomes [93, 96]. In a study of Teng et al. (2025), similar DMGs were identified in somatic muscle and gonadal tissue between male and female oysters. The diacylglycerol kinase delta (Dgkd) gene was consistently hypomethylated in somatic muscle and gonadal tissue of female oysters. Additionally, genes involved in Rho protein signal transduction were found to be consistently differentially methylated in both muscle and gonadal tissue [40]. In oyster gonads, multiple genes (e.g. Foxl2, Dmrt1, Dgkd), were discovered to be involved in sex differentiation, whereas these genes where not found to be differentially methylated in gill tissue in this study [28–31, 40]. Additionally, DNA methylation in gonad tissue was often found to be hypermethylated among males compared to females [29, 30, 40]. In our study, males or females did not show consistent hyper or hypomethylation for DMRs and DMGs. Although the enrichment for GO terms related to metabolism and energy homeostasis pathways do not overlap with DNA methylation patterns found for gonad tissue in other oysters, and may therefore be indicative of functional differences between tissues, a direct comparison of DNA methylation patterns between gill and gonadal tissue was not performed in this study and requires further elucidation.

In Crassostrea virginica, differences in sex-specific gene expression were identified between mantle and gonad transcriptomes [14]. Here, differentially expressed genes found between male and female mantle tissue were associated with cell proliferation, apoptosis, and lipid transportation. Based on their findings, the authors suggest that sex-specific gene expression in mantle tissue has a specific function that could be potentially related to environmental sensing or signalling in C. virginica [14]. The observed enrichment in energy homeostasis and metabolism related pathways in sex-specific DMGs identified in this study could be indicative of functional processes related to sex reversal in the gills. However, to further elucidate potential functional differences in sex-specific DNA methylation of different tissues, more elaborate methylome comparisons between various tissues are required. Additionally, GO and KEGG enrichment analysis are both largely dependent on available gene annotations, and so our discussion on the occurrence of functional sex-specific DNA methylation in the gills is limited to currently available reference genomes and gene annotations.

Causality dilemma of the role of somatic DNA methylation in sex reversal

Our findings clearly show the presence of sex-specific DNA methylation in gill tissue of O. edulis, however, it remains unknown where in the process of sex reversal this gill tissue DNA methylation plays a role. Sex steroids are known to not only influence sex differentiation in reproductive tissues, but can also affect cell fates in somatic tissues all throughout the body [92]. The concept of altered whole-body methylomes or transcriptomes in response to treatment with endocrine disrupting chemicals is widely observed in fish [97–99]. Interestingly, DNA methylation can also affect sex differentiation vice versa [100]. Based on this, it could be hypothesized that sex-specific DNA methylation of the gills occurs prior to gonadal sex differentiation, and plays a more active part in sex reversal. Environmental stressors like starvation have been demonstrated to affect energy homeostasis and metabolism pathways in the gills of bivalves through influencing gene expression [28, 67]. Hence, it could be inferred that environmental stressors induce changes in gill DNA methylation, where a remodeling of the energy balance could subsequently induce sex differentiation in the gonads. In this case, gill DNA methylation would represent a more active player in sex reversal. Although gill and gonadal samples were collected simultaneously in this study, we do not know whether the phenotypic sex represented by the gonads is similarly represented by the DNA methylation state of the gills. Furthermore, to better understand differences in sex-specific DNA methylation between different tissues, the effect of sex-specific DNA methylation on gene expression and corresponding functionality should be elucidated.

Non-lethal sex identification using epigenetic markers

Sex identification is extremely relevant for understanding the process and dynamics of sex reversal in sequential hermaphrodites. Currently, sex identification in oysters is complicated by the dependency on gonadal tissue, of which sampling is lethal. With the use of anesthetics, some somatic tissues can be sampled non-lethally (e.g. abductor muscle, gill, and mantle tissue). Therefore, the use of molecular markers, that rely on somatic tissues for sex identification, provide a promising non-lethal alternative. Epigenetic markers are currently applied to assess sex and age, or response to pollutants in various marine species [101–103]. In our study, comparison of gill methylomes between male and female oysters resulted in the characterization of 125 DMGs. Differential methylation on these genes could provide the basis for generating epigenetic markers that can assess the current sex phenotype of an oyster. Additionally, the distinct DNA methylation profile of hermaphrodite oysters would allow for a higher resolution of sex identification, whereby hermaphrodites could be distinguished from male or female individuals. Collectively, our findings indicate the potential of using somatic gill tissue for non-lethal sex identification utilizing epigenetic markers that can predict the current sex phenotype of an oyster.

Conclusion

In conclusion, this study is the first to characterize sex-specific DNA methylation in somatic gill tissue of a sequential hermaphrodite invertebrate. Our findings suggest that hermaphrodite oysters display distinct differences in gill DNA methylation patterns compared to male and female individuals. Annotation of differentially methylated genes resulted in significant enrichment of pathways associated with energy homeostasis, metabolism, and innate immunity. Collectively, our data suggests that sequential hermaphrodites display an alternative state that represents the epigenetic reprogramming required for sex reversal in a somatic tissue. Additionally, the characterized differentially methylated loci and regions between sex phenotypes from somatic gill tissue provide a basis for non-lethal alternatives to sex identification with epigenetic markers. Taken together, this study contributes to the understanding of epigenetic mechanisms underlying sex reversal in different sex phenotypes of the flat oyster, and shows the relevance of studying sex-specific DNA methylation in somatic tissues in relation to sex reversal.

Materials and methods

Sampling site and collection

In this study, DNA methylomes were obtained from whole-genome sequencing of gill tissue of 35 European flat oysters (Ostrea edulis) of different sex phenotypes. Oysters were collected by dredging in April and September 2023, and April 2024 in Grevelingenmeer, the Netherlands (53°30′40″N, 08°06′20″ E). Individuals of similar shell length (74.1 ± 7.45 mm; Supplementary FiS1B) were selected and gill tissue was collected and preserved with RNAlater or snap-frozen using liquid nitrogen. All gill tissue samples were stored at −80 °C until further use. Oysters were collected in the context of the national monitoring for mollusc diseases in the Netherlands. Samples from gill, mantle, and digestive diverticulum tissue were tested for the presence of the haplosporidian Bonamia ostreae, using species-specific real time PCR as described in Kamermans et al. (2023). Two oysters (nr. 3 and 17) tested positive for an infection of B. ostreae. We did not correct for this in our analysis, as other physiological differences between individuals were not assessed nor included in this analysis.

Histological sex identification

Cross sections of tissue representing both reproductive, respiratory, and digestive organs were embedded in paraffin. Sections (± 5 µm) were stained with haematoxylin–eosin for histological observation. Each oyster was categorized a sex stage and gonad development index, according to the study of da Silva et al. (2009) [15]. The five sex stages are as follows: male (M), female (F), hermaphrodite predominantly male (HPM), hermaphrodite predominantly female (HPF), or hermaphrodite both sexes equally present (HBS). The gonad development index was considered as follows: Inactive or resting gonad (0), early gametogenesis (1), advanced gametogenesis (2), mature (3), partially spawned gonad (4), reabsorbing gonad (5). Oysters that displayed no clear gonad tissue and were therefore categorized to be in the inactive or resting gonad phase (0) were not used for analysis. In total, 35 oysters were used for DNA methylome sequencing; 9 M, 8 F, 5 HPM, 9 HPF, and 4 HBS oysters. An overview of all histological images can be found in the supplementary material (Supplementary Figure S5).

DNA extraction

Genomic DNA was extracted from oyster gill tissue using the Qiagen Gentra Kit for genomic DNA or Monarch HMW DNA extraction kit for tissue, according to the manufacturer’s instructions (Qiagen catalog no. 158667; NEB catalog no. T3060). Sample input varied from 20 to 30 mg of frozen gill tissue and was grinded using liquid nitrogen. Long DNA (> 20 Kb) was sheared using g-TUBES (Covaris, USA) in 1 min centrifugation steps of 8, 10, 12, 14 (x1000G) respectively. Short reads were eliminated using the SRE XS PacBio kit, following the manufacturer’s instructions (Pacific Biosciences, Menlo Park, CA, USA). DNA was quantified using the HS Qubit assay (Life Technologies, ThermoFisher Scientific). Quality and length of DNA was visualized using agarose gel electrophoresis.

Library preparation & sequencing

To prepare samples for whole-genome sequencing using the Oxford Nanopore Platform (ONT, Oxford, UK), short-read eliminated samples were used for end-prep and DNA repair, followed by barcode and adapter ligation according to the native barcoding protocol (SQK-NBD114-24). Up to a maximum of five oysters were simultaneously sequenced using PromethtION R10.4.1 flowcells (FLO-PRO114M) on a PromethION-24 device, using MinKNOW software versions: 23.04.5 (July 2023), 23.07.12 (December 2023), 24.02.19 (May 2024), and 24.06.15 (December 2024). A minimum read quality threshold of 7 was applied.

Bioinformatics

Raw read processing

Raw sequenced reads (POD5) were basecalled using Dorado v0.4.1 [104], allowing for simultaneous super accuracy basecalling (SUP) of canonical basepairs using model dna_r10.4.1_e8.2_400bps_sup@4.2.0, and 5 mC modifications using the –modified-bases 5 mC flag for the detection of modified bases. The reads were mapped to the reference genome xbOstEdul1.1 (NCBI RefSeq Assembly: GCF_947568905.1 [105]) with minimap2 [106] as a built-in tool within the Dorado basecaller function. After mapping, the reads were demultiplexed using the demux function of Dorado. Mapped BAM files were then filtered on a minimal mapping quality of 60 and all unmapped reads or secondary alignments were removed using Samtools v1.17 [107]. After filtering, BAM files were sorted and indexed using Samtools. Descriptive statistics for BAM files were generated using NanoPack2 v1.12.0 [108].

CpG calling

The pileup function of Modkit v0.2.1 (https://github.com/nanoporetech/modkit) was used to call CpGs from the mapped BAM files to generate BEDmethyl output files, using the xbOstEdul1.1 reference genome [105]. For some CpG sites more than half of the reads were identified as Ndel, indicating the read contains a deletion on this site compared to the reference genome. Reads where the number of Ndels was ≥ 60% of the total coverage for the corresponding CpG site were removed from the dataset.

Differential methylation analysis

Characterization of differentially methylated loci and regions

BEDmethyl files generated by Modkit were used as input for methylKit v.0.99.2 in R v.4.3.3 [109, 110]. Methylation percentages were calculated for every CpG site by dividing the sum of methylated CpGs by the total number of CpGs covered. Methylation calls were filtered to a minimum required coverage of 3 reads per CpG site for all individuals. Methylation call distributions between samples were normalized using the normalizeCoverage function to reduce the potential bias in varying coverage among individuals. For pairwise comparisons between sex phenotypes, data was reorganized to include only the individuals pertaining the selected sex phenotypes, using the reorganize function. CpG sites of individuals from two sex phenotypes were then merged using the unite function. To correct for a potential effect of maturation development stage on DNA methylation, maturation stage was added as a covariate.

Using a logistic regression, differentially methylated loci (DMLs) were identified for each pairwise comparison, using the calculateDiffMeth function in methylKit. P-values were adjusted for multiple testing using the Benjamin-Hochberg correction. DMLs were defined as loci with a minimal methylation difference of 20% between the two compared sex phenotypes, and a q-value < 0.01 (BH corrected). For every pairwise comparison, the number of hypermethylated and hypomethylated DMLs were determined whereby a hypermethylated DML was defined as a CpG site where the treatment group was more methylated than the control group. A hypomethylated DML was defined as a CpG locus where the treatment group was less methylated than the control group.

Differentially methylated regions (DMRs) were identified using a sliding-window approach of 200 bp window slides and 100 bp intervals, using the tileMethylCounts function from methylKit, with maturation stage added as a covariate. Only windows containing ≥ 5 CG loci (cov.bases), a minimal difference in methylation percentage of 20%, and BH-adjusted q-value < 0.01 were considered as DMRs. Hypermethylated DMRs were defined as DMRs where the treatment group was hypermethylated compared to the control group. Hypomethylated DMRs were defined as DMRs where the treatment group was hypomethylated compared to the control group. Differentially methylated loci and regions were visualized from BED files using IGV v2.19.1 [111].

To identify the distribution of DMRs over genomic features (promoter, exon, intron, intergenic region), the position of genomic features was extracted from the reference genome using readTranscriptFeatures from methylKit. Subsequently DMRs were annotated to these genomic features using annotateWithGeneParts function from the genomation package [112]. Promoters were defined as regions 1000 bp upstream or downstream of Transcription Start Sites (TSS). To test if there was an association between genomic location of DMRs and the genomic location of all CpGs with minimal coverage of 3 for all individuals, chi-square contingency pairwise tests were used.

Identifying differentially methylated genes

To identify the differentially methylated genes (DMGs), genes overlapping with DMRs, the DMRs within 100 bp proximity of one another were first merged together. Subsequently, gene coordinates were extracted from the reference annotation (GCF_947568905.1_xbOstEdul1.1_genomic.gtf.gz [105]) using the gene function of GenomicFeatures [113]. Genetic description and GeneID were added using getGenes from mygene [114]. Genes were considered statistically differentially methylated if they contained at least one DMR with minimal methylation percentage of 20% and BH-adjusted q-value < 0.01.

To analyse functional enrichment of annotated DMGs, Gene Ontology (GO) analysis was performed using TopGO in Bioconductor [115]. Using the Gene Annotation File (GAF) the reference genome (GCF_947568905.1-RS_2023_05_gene_ontology.gaf), GO terms were annotated to geneIDs and separated based on GO term categories: Biological Processes (BP), Cellular Components (CC), or Molecular Function (MF). Go terms with p < 0.001 were considered significantly enriched by DMR-overlapping genes. Heatmaps were generated using the pheatmap package in R [116]. The upset plot was generating using UpSetR [117].

Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis was performed using the KEGGrest [118] package in R.Enrichment pathways were imported for Ostrea edulis (organism code: “oed”; RefSeq Assembly: GCF_947568905.1) using keggLink. Enrichment analysis was performed using the enrichKEGG function from the clusterProfiler package [119]. A Benjamini–Hochberg adjustment of p-values was applied to correct for False Discovery Rate (FDR).

General statistical analysis

To check for differences in average shell length and average sequenced Gb per sex phenotype, one-way ANOVAs were performed using the stats package from R [120]. With Shapiro–Wilk’s and Levene’s test, deviations from normality and homogeneity of variance of the residuals were validated using the car package v3.1.3 [121].

Abbreviations

M

Male

F

Female

HPM

Hermaphrodite predominantly male

HPF

Hermaphrodite predominantly female

HBS

Hermaphrodite both sexes equally present

DML

Differentially methylated loci

DMR

Differentially methylated region

DMG

Differentially methylated gene

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

FDR

False discovery rate

Author’s contributions

SV, HJM, and RN conceptualized the study. SV carried out the methodology and sequencing. Data analysis was performed by SV and HJM. Writing of the manuscript was done by SV, and review & editing of the manuscript was carried out by SV, ME, HJM, PK, AM, and RN. RN was involved in supervision of the research. SV and PK acquired funding.

Funding

Whole genome sequencing was made possible with financial support of the Centre of Genetic Resources (CGN) of Wageningen University and Research (project no/WOT-03–001-069).

Data availability

All processed sequencing data generated in this study (BAM format) have been submitted to the European Nucleotide Archive (ENA; https://www.ebi.ac.uk/ena/) under project number PRJEB85091 with accession number ERP168521.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.