A DRACH Observed/Expected Metric Reveals the Evolutionary Landscape of Epitranscriptomic Regulation Across Bilateria

ABSTRACT

N ⁶‐methyladenosine (m⁶A) RNA modification is a dynamic post‐transcriptional regulatory mechanism that rapidly modulates gene expression by influencing RNA stability, export, splicing, translation, and decay. In bilaterians, m⁶A is predominantly deposited at the DRACH consensus motif, which comprises 18 sequence variants. Although the presence of a DRACH motif does not indicate whether a transcript is currently, or has ever been, methylated, it reflects its potential for m⁶A‐mediated regulation. Here, we introduce a new metric, DRACH observed/expected (DRACH o/e), to quantify the relative enrichment of DRACH motifs independently of sequence composition. We applied this approach to characterize the abundance, distribution, and genomic organization of DRACH motifs in the Pacific oyster Magallana gigas, a representative lophotrochozoan species in which m⁶A regulation remains poorly characterized. These analyses were extended to eight additional species spanning major bilaterian lineages and exhibiting contrasting ecological traits. Comparative analysis revealed strong conservation of the m⁶A regulatory machinery together with conserved and lineage‐specific patterns of DRACH motif distribution and functional enrichment. This work introduces a robust framework for estimating the potential for transcript m⁶A methylation while overcoming the inherent variability of epitranscriptomes, thereby paving the way to explore comparative analyses of its potential phylogenetic and/or ecological drivers.

Epitranscriptomic marks are highly variable across life stages and tissues, complicating comparisons between species. Using m⁶A consensus motifs, we developed a genomic indicator reflecting transcript methylability, the DRACH o/e score, that we used to investigate epitranscriptomic features and associated functions across nine bilaterian species. Our results reveal highly conserved m⁶A machinery and that both ecological influence and phylogeny strongly shape genome DRACH organization and function, questioning evolutionary outcomes of epitranscriptomes.

1. Introduction

N ⁶‐methyladenosine (or m⁶A) is one of the most prevalent modifications among the ∼300 epitranscriptomic modifications in eukaryotic RNAs [1]. It corresponds to the methylation of the nitrogen atom at the sixth position of adenines within RNAs. This modification is catalyzed by the m⁶A writer complex, whose catalytic core consists of the METTL3–METTL14 heterodimer that transfers a methyl group from the methyl donor S‐adenosylmethionine to adenosine. The methyltransferase complex (MTC) also includes several regulatory subunits, such as Wilms’ tumor 1‐associating protein (WTAP) and RBM15/15B, which contribute to substrate recognition and promote methylation of the conserved DRACH consensus motif (D = A/G/U, R = A/G, H = A/C/U) [2−4].

The MTC is conserved in a wide range of eukaryotic lineages, including plants, fungi, insects, mollusks, and mammals [5−8], where m⁶A modifications occur in all classes of RNA [4]. Although m⁶A‐RNA decoration has been reported in all eukaryotes examined to date, some organisms use alternative methyltransferases, resulting in different m⁶A landscapes. For example, METTL16 is a solitary methyltransferase identified in mammals [9], and other species such as oysters [6] and E. coli [10], and preferentially methylates the MAT2A transcript and U6 snRNA at the UAC A GAGAA consensus sequence [9].

m⁶A‐decorated RNAs recruit reader proteins that mediate distinct biological functions depending on their subcellular localization and/or binding specificity. These readers regulate multiple aspects of RNA metabolism, including pre‐mRNA splicing and stability in the nucleus, as well as mRNA export, translation, and decay in the cytoplasm [11−13]. Their recruitment is strongly influenced by the localization of the m⁶A marks. For instance, m⁶A located in the 5′ untranslated region (5′UTR) promotes cap‐independent translation through the recruitment of the eukaryotic initiation factor 3 [14]. Likewise, m⁶A deposition near intron/exon junctions regulates alternative splicing through the recruitment of splicing factors mediated by m⁶A‐binding proteins [13, 15−17]. In addition, m⁶A residues within the 3′ untranslated region (3′UTR) are required for proper subcellular localization of specific transcripts, as demonstrated for neuronal mRNAs in mice [18].

m⁶A modification is reversible by the oxidation of N ⁶‐methyladenosines by eraser proteins, including FTO and ALKBH5 in mammals [19, 20]. Consequently, m⁶A levels fluctuate dynamically across developmental stages, cell type, and physiological context, whereas the underlying DRACH sequence content remains unchanged [21]. This dynamic epitranscriptomic regulation has not only been documented in mammals, but also in drosophila, oysters, yeast, and plants [7, 21−27], and it enables rapid adaptation of gene expression from pre‐existing transcripts during environmental stress responses or fast developmental processes [2, 11]. Indeed, epitranscriptomic regulation is essential to induce key developmental transitions from undifferentiated to differentiated cells [28], maternal to zygotic transition [29, 30], or metamorphosis [31]. Consequently, due to the transient nature of RNA methylation, actual m⁶A profiles appear highly context‐, cell‐type–, and developmental stage‐dependent, and, therefore, provide only a glimpse of a continuously changing epitranscriptome. To overcome this limitation, we sought to investigate the evolutionary potential for m⁶A regulation by focusing on the distribution of DRACH motifs rather than on transient methylation profiles.

Previous studies have shown that DRACH sites are highly conserved between humans and mice [2], whereas m⁶A peaks are not [32]. Similar comparisons indicate a host‐specific conservation of DRACH sites within influenza viruses, suggesting motif conservation is under ecological selective pressure [33]. Moreover, by analogy with CpG dinucleotides—which become progressively depleted through methylation‐induced mutations despite being ancestrally CpG‐rich—m⁶A sites might undergo a similar evolutionary process, given that highly m⁶A‐methylated regions are correlated with elevated single nucleotide polymorphism density [21]. Nevertheless, the extent to which DRACH motif distribution has been conserved or diversified across bilaterian evolution, and what this may reveal about the evolutionary history of epitranscriptomic regulation, remains largely unknown. Progress toward answering these questions has been limited by the scarcity of data from phylogenetically distant lineages, such as the lophotrochozoans, which encompass remarkable biodiversity but remain poorly characterized, especially when compared with the extensive literature on epitranscriptomic regulation in ecdysozoans, its sister clade [7, 34, 35].

The Pacific oyster (Magallana gigas) is a bivalve model that exhibits a highly conserved epitranscriptomic pathway—encompassing m⁶A machinery, distribution patterns, and biological roles [36]—and highly conserved functional outcomes, notably in developmental processes [24, 37], stress responses [36], and splicing regulation [37]. We, therefore, took advantage of this unique experimental framework to investigate the evolution of epitranscriptomic regulation across Bilateria. To this end, we developed a new metric, the DRACH observed/expected (DRACH o/e), that quantifies the relative abundance of DRACH motifs independently of sequence composition. We first used this metric to characterize the abundance and genomic distribution of DRACH motifs in the M. gigas genome, allowing the identification of two populations: DRACH‐enriched genes (DRACHen) and DRACH‐depleted genes (DRACHdep), and then examined their genomic organization and putative functions. To investigate whether variation in DRACH enrichment primarily reflects phylogenetic history or is instead associated with ecological traits, we extended our study to eight additional species representing the major bilaterian clades (cnidarians, insects, mollusks, and chordates), including contrasting reproductive strategies (e.g., K vs. r).

2. Materials and Methods

2.1. Motif Counting

DRACH motifs and their submotifs were counted using an in‐house Python script (DOI: 10.5281/zenodo.18388105) that computes the location and the abundance of each of the 18 submotifs within the transcriptomes or genome (in FASTA format) of M. gigas and the eight other studied species (Table 1). The same methodology was used to count DRCCH, DRTCH, and DRGCH motifs, considered as nonmethylable control motifs. Distribution differences were tested using a Kruskal−Wallis statistical test coupled with a Dunn's paired comparison post‐hoc test with Benjamini−Hochberg correction.

TABLE 1.

Summary table of species examined in this study, including their abbreviations and genome accession number (Genome_ID).

Species	Abbreviation	Genome_ID
Acropora millepora	A. mil	GCF_013753865.1
Apis mellifera	A. mel	GCF_003254395.2
Ciona intestinalis	C. int	GCF_000224145.3
Culex pipiens pallens	C. pip	GCF_016801865.2
Drosophila melanogaster	D. mel	GCF_000001215.4
Homo sapiens	H. sap	GCF_000001405.40
Magallana gigas	M. gig	GCF_963853765.1
Mytilus trossulus	M. tro	GCF_036588685.1
Ostrea edulis	O. edu	GCF_947568905.1

2.2. Species Selection

Because we questioned the potential link between reproductive strategy and DRACH content, species with the highest (r‐strategy) and lowest (K‐strategy) π_s scores were shortlisted [38]. We next selected species representing both types of reproductive strategies within the same clade to avoid any potential bias linked to phylogenetic proximity. To test the effect of phylogeny on DRACH distribution, species were also chosen based on their phylogenetic position within the largest possible range of major bilaterian groups. Finally, species were selected depending on the availability of the most complete genome sequence, chromosome assembly, and annotation. Based on these criteria, nine species among mollusks, insects, and deuterostomes (chordates) were retained, and a cnidarian was used as an outgroup (Table 1).

2.3. Genic Versus Intergenic Regions

DRACH motifs in intergenic regions were counted by (1) determining the coordinates of nongenic regions from the GFF gene category (GCF_963853765.1_genomic.gff from Sanger institute [39]), (2) extracting their fasta sequences using the get‐fasta tool from bedtools [40], and (3) determining the length and number of DRACH motifs in each region based solely on nonsoft‐masked regions. DRACHs in genic regions were determined from the transcriptome. For genic and intergenic regions, both the DRACH o/e score and the z‐score were calculated by taking into account the GC content in each region.

2.4. DRACH o/e Metric Determination

To normalize and enable comparison of the amount of each 5‐mer within an organism and between species, a DRACH o/e score was calculated for each transcript by dividing the number of DRACH sites by the expected number under a random distribution. DRACH expected takes into account the DRACH probability (or P(DRACH)), which was calculated according to the GC% of the considered transcriptome and the length of the transcript:

$$DRACHo/e=\frac{\mathrm{nb}\mathrm{DRAC}{\mathrm{H}}_{\mathrm{observed}}}{\mathrm{nb}\mathrm{DRAC}{\mathrm{H}}_{\mathrm{expected}}}$$

$$DRACHo/e=\frac{\mathrm{nb}\mathrm{DRAC}{\mathrm{H}}_{\mathrm{observed}}}{\mathrm{P}\left(\mathrm{DRACH}\right)\times length-4}$$

With

$$\begin{array}{ccc}\hfill \mathrm{P}\left(\mathrm{DRACH}\right)& =& \left[\mathrm{P}\left(\mathrm{A}\right)+\mathrm{P}\left(\mathrm{C}\right)+\mathrm{P}\left(\mathrm{G}\right)\right]\times \left[\mathrm{P}\left(\mathrm{A}\right)+\mathrm{P}\left(\mathrm{C}\right)\right]\hfill \\ & & \times \mathrm{P}\left(\mathrm{A}\right)\times \mathrm{P}\left(\mathrm{C}\right)\times \left[\mathrm{P}\left(\mathrm{A}\right)+\mathrm{P}\left(\mathrm{C}\right)+\mathrm{P}\left(\mathrm{T}\right)\right]\hfill \end{array}$$

DRACH o/e efficiency to standardize data was tested (Figure S1). The distribution was further transformed and centered around zero by computing a z‐score from DRACH o/e:

$${\mathrm{Z}}_{\mathrm{score}}=\frac{\mathrm{DRACH}\mathrm{o}/\mathrm{e}-\mathrm{mean}\left(\mathrm{DRACH}\mathrm{o}/\mathrm{e}\right)}{\mathrm{std}\left(\mathrm{DRACH}\mathrm{o}/\mathrm{e}\right)}$$

Across all considered species, the z‐score distribution follows a Gaussian‐like shape with slightly heavy‐tailed properties. We defined +2 and −2 as the upper and lower bounds of the distribution, with sites falling above or below these limits classified as DRACH‐enriched and DRACH‐depleted genes (DRACHen and DRACHdep), respectively. These thresholds encompass 95.4% of the general population, leaving approximately 2.3%–2.5% of values in each tail. Graphs were generated using the ggplot2 package (version 3.3.6) in RStudio software (version 4.3.3).

A similar methodology has been applied to DRCCH, DRGCH, and DRTCH motifs (Figure 4A).

FIGURE 4.

DRACH motifs demonstrate conservation across phylogeny. (A) DRACH, DRCCH, DRGCH, and DRTCH observed/expected mean score in each one of the nine species’ transcriptomes. (B) Correlation plot of the 18 DRACH submotifs abundance ratios between species (a similar plot for the three random motifs have been generated; Figure S4B−F). For better readability, colors correspond to the major clades: cnidaria in yellow, chordates in green, insects in purple, and mollusks in blue. The correlation plot and the corresponding Pearson's correlation score are identically colored for each plot, in mirror based on the diagonal. (C) Correlation scores (from C and Figure S4C−F) were used to compare the evolutive conservation of the methylable DRACH motifs compared to nonmethylable DRCCH, DRGCH, and DRTCH. Statistical significance was tested using the Kruskal−Wallis test (p < 0.05) with a Dunn's multiple comparison post‐hoc test using Benjamini−Hochberg correction. (D) Phylogenetic tree comparing the edginess values of DRACHen (blue), DRACHdep (red), and other (gray) transcripts of the nine studied species. Kruskal−Wallis test with post‐hoc paired comparison using Benjamini−Hochberg correction. Species displaying significantly different distributions are written in red. Abbreviations: A. mil, Acropora millepora; C. int, Ciona intestinalis; H. sap, Homo sapiens; A. mel, Apis mellifera; C. pip, Culex pipiens; D. mel, Drosophila melanogaster; M. tro, Mytilus trossulus; O. edu, Ostrea edulis; M. gig, Magallana gigas.

2.5. DRACHen and DRACHdep Location on the Chromosomes

The chromosomal position of DRACHen and DRACHdep was determined on the reference genome and visualized using the CircosPlot and ChromoMap (version 0.4.1) R packages (R 4.3.3) (Figure 2C and Figure S2A−C).

FIGURE 2.

DRACH motifs are enriched in genic regions and in the vicinity of chromosome ends. (A) DRACH observed/expected (DRACH o/e) metric in genic (pink) (i.e., transcriptome) or intergenic (blue) regions. Wilcoxon p‐value = 2.2e‐16. (B) Z‐score distribution in the genic region calculated from DRACH o/e value. The Gaussian distribution is a heavy‐tailed, non‐normal distribution. Transcripts with a score above 2 and below −2 are considered extremes and were split into DRACH‐enriched (DRACHen) and DRACH‐depleted (DRACHdep) gene populations, respectively. (C) Genomic distribution of both gene populations in Magallana gigas chromosomes. The exterior circle represents the chromosomes, the intermediate blue circle represents the gene location, and the inner circle displays the position of DRACHen (in blue) and DRACHdep genes (in red). (D) Edginess score between DRACHen, DRACHdep, and other transcripts. Edginess scores close to 0 reflect a gene close to a chromosome edge, while a score close to 1 corresponds to a gene located in the center of the chromosome. Distribution differences have been tested with Kruskal−Wallis test (p < 2.2e‐16) and paired comparison post‐hoc with Benjamini−Hochberg correction (for DRACHen vs. DRACHdep and DRACHen vs. other, p < 2.2e‐16; DRACHdep vs. other, p = 0.058). (E, F): DRACH profiles within DRACHen and DRACHdep mRNAs (E) and ncRNAs (F). (G) GO enrichment analysis of the DRACHen and DRACHdep gene populations.

DRACHen and DRACHdep proximity to chromosome ends was assessed in all species by assigning a normalized positional value—called edginess—to each gene, to obtain a score ranging between 0 (for a gene at either edge of the chromosome) and 1 (for a gene at the center of the chromosome). Gene positions were approximated as the center of the gene ((start + end)/2). Edginess was further calculated as the gene position relative to the chromosome end, using the formula:

$$edginess=\frac{min(gen{e}_{pos},ch{r}_{length}-gen{e}_{pos})}{ch{r}_{length}/2}$$

Edginess of DRACHen, DRACHdep, and the remaining fraction (called others) was plotted using the ggplot2 R package (version 4.0.0) [41]. Statistical tests were performed using Kruskal−Wallis tests followed by post‐hoc pairwise comparison using the Wilcoxon rank‐sum test with Benjamini−Hochberg adjustment.

2.6. DRACH Position Within the Transcripts and Functional Enrichment in DRACHen and DRACHdep Populations

DRACH positions in the mRNAs and ncRNAs of DRACHen and DRACHdep genes were assessed using the Guitar package (version 2.6.0) [42] using default options and the gtf file from the M. gigas genome assembly [39].

After extracting DRACHen and DRACHdep subpopulations, a GO Enrichment analysis was computed using the GOEnrichment module at useGalaxy.org (version 2.0.1) [43]. The whole transcriptome was used as the background dataset.

2.7. DRACH, DRCCH, DRGCH, and DRTCH Frequency

Each DRXCH (with X being A, C, G, T) motif encompasses 18 submotifs. The number of each submotif was counted in the genome and the transcriptomes of the nine considered species, and their frequencies were compared across species transcriptomes.

2.8. m⁶A Machinery Conservation and Protein Domain Architecture Visualization

Protein sequences for m⁶A machinery components (METTL3, METTL14, WTAP, RBM15, HAKAI, ZC3H13, VIRMA, METTL16, ALKBH5, YTHDC1, YTHDC2, YTHDF2, IGF2BP2, HNRNPA2B1) were retrieved from the NCBI Protein database or OrthoDB for the nine species considered. When orthologous sequences were not available for a given species, closely related species were used as substitutes (Table S1A,B).

Conserved protein domains were identified using the NCBI Conserved Domain Database (CDD) with default parameters. Domain annotations included MT‐A70 (methyltransferase‐70 kDa subunit), WTAP (Wilm's tumor associated protein), RRM (RNA recognition motif), SPOC (Spen paralog and ortholog C‐terminal), RING finger (Really Interesting New Gene finger), zf‐CCCH (zinc finger Cys₃His), Methyltransf (methyltransferase), 2OG‐FeII_Oxy (2‐oxoglutarate/Fe(II)‐dependent oxygenase), YTH (YT521‐B homology), R3H (RXXXH), HrpA (Human replication protein A), DEXHc (DEAD/DEAH box helicase), HA2 (helicase associated), OB (oligonucleotide/oligosaccharide‐binding fold), and KH (K homology) domains.

Protein domain architecture figures were generated using custom scripts. Proteins were aligned based on conserved domain positions to facilitate cross‐species comparisons (Figure 3). Domain coordinates for all species and proteins are detailed in Tables S1C‐1 and S1C‐2. Pairwise sequence identity versus human orthologs was computed for full‐length proteins and concatenated domains using EMBOSS Needle (BLOSUM62, gap open = 10, gap extend = 0.5), excluding gaps (Figure S3).

FIGURE 3.

Protein domain architecture of m⁶A methylation machinery components across metazoan species. Schematic representation of conserved protein domains in m⁶A writer proteins (left panel: METTL3, METTL14, WTAP, RBM15, HAKAI, METTL16, VIRMA, and ZC3H13), eraser protein (right panel, top: ALKBH5), and reader proteins (right panel, bottom: YTHDC1, YTHDC2, YTHDF2, IGF2BP2, and hnRNPA2B1). Protein sequences were analyzed from nine species representing key metazoan lineages: Homo sapiens, Ciona intestinalis, Apis mellifera (or Apis laboriosa, Colletes gigas), Culex pipiens (or Culex quinquefasciatus), Drosophila melanogaster, Mytilus trossulus (or Mytilus edulis, Mytilus coruscus, Mytilus californianus, Mytilus galloprovincialis), Ostrea edulis (or Saccostrea cuccullata), Magallana gigas, and Acropora millepora (or Acropora palmata, Acropora digitifera, Acropora muricata, Acropora cervicornis). Species names in blue italics indicate substitute species used when the ortholog was not identified in the primary species. Dashed outlines represent species where no ortholog was detected (notably ALKBH5 in the three analyzed insects, and METTL16 in C. intestinalis). Gray bars indicate full protein length; colored boxes represent conserved domains identified using the NCBI Conserved Domain Database (CDD). All proteins are drawn to scale (0.5 px/aa), with a 50 aa scale bar aligned to the start of the human protein for each alignment; VIRMA and ZC3H13 are shown at reduced scale (0.20 px/aa, 100 aa scale bar) and truncated at 1500 aa due to their large size. Domain color coding: Writers (green shades)—MT‐A70, WTAP, RRM, SPOC, RING‐Hakai, zf‐Hakai, Methyltransf_10, VIR_N, zf‐CCCH, and FT Rho; Eraser (red)—2OG‐FeII_Oxy; Readers (blue shades)—YTH, R3H, HrpA, DEXHc, HA2, OB, RRM, and KH.

2.9. Functional KOG Enrichment

To determine and compare the enriched functions in the DRACHen and DRACHdep, the genes from the two populations in the nine species were blasted and annotated. Only the best hit was kept, and GO (Gene Ontology) and KOG (euKaryotic Orthologous Groups) terms were associated. To assess whether biological functions were associated with the methylability of their related transcripts, a functional summary and meta‐analysis of gene DRACH abundance data were performed using the KOGMWU R package (version 1.2), ranking the z‐score using a delta rank [44]. This analysis relies on rank‐based tests for enrichment of KOG classes with DRACHen or ‐dep genes ranked on a continuous measure and one‐tailed Fisher's exact test for over‐representation of each KOG class among significant genes. Delta rank is the difference between the mean rank of genes belonging to a KOG class and the mean rank of all other genes, thereby reflecting whether biological functions are associated with DRACH enrichment or depletion across the examined species. Therefore, a high delta rank is associated with a DRACH‐rich gene set, and conversely a low delta rank corresponds to a DRACH‐poor gene set.

2.10. Correlation Plots

Correlation plots were produced using the GGally (version 2.3.0) extension package from ggplot2. Depending on the considered correlation, input data were either the submotifs frequency, the amino‐acid content in the proteomes, or the mean delta rank associated with each KOG category in each species. Correlation scores were extracted to compare DRACH, DRCCH, DRGCH, and DRTCH submotif frequency correlations in the nine species on the one hand (Figure S4A−D and Figure 4C), and to compare KOG enrichment correlation scores in DRACH versus DRCCH context on the other hand (Figure S5).

3. Results

3.1. DRACH Motifs in M. gigas Genome and Transcriptome

The 18 submotifs composing the DR A CH consensus sequence (where the A carries the methylation) were identified in silico and mapped across the M. gigas genome and transcriptome. To assess the relevance of DRACH methylability, the results were compared with those obtained for three nonmethylable 5‐mer motifs: DRCCH, DRTCH, and DRGCH.

The results showed a significant enrichment of DRACH motifs relative to DRCCH and DRGCH motifs in the oyster genome, but not relative to DRTCH (Figure 1A,B). Submotif frequencies were not uniform, with a marked overrepresentation of AAACA—and its corresponding 5‐mer variants (AACCA, AATCA, and AAGCA)—compared with the other submotifs (Figure 1A). However, all DRXCH 5‐mers were evenly distributed across chromosomes (Figure 1C). A similar pattern was observed in the M. gigas transcriptome, with AAACA and its counterparts also being overrepresented (Figure 1D). As observed in the genome, DRACH motifs were significantly more abundant than DRCCH and DRGCH in the transcriptome (Figure 1E). Interestingly, the number of DRTCH motifs in the transcriptome is reduced relative to the genome, as it does not differ significantly from either DRACH or DRCCH, placing it in an intermediate position. We, therefore, examined motif frequencies in genic and intergenic regions, and found that DRACH motifs were more abundant than expected in both genic (7,256,837 vs. 5,283,357) and intergenic regions (15,639,329 vs. 11,181,786). The same pattern was detected for DRTCH motifs, whereas DRCCH and DRGCH motifs were found less abundant than expected in both genic and intergenic regions (Figure 1F).

FIGURE 1.

DRACH motifs are specifically conserved in the Magallana gigas genome and transcriptome. Abundance of the 18 submotifs of potentially methylable DRACH sites and DRCCH, DRGCH, and DRTCH random 5‐mers in the genome (A) and in the transcriptome (D) of the Pacific oyster. There are significantly more DRACH in the genome (B) and in the transcriptome (E) of M. gigas than DRCCH and DRGCH. No significant difference has been observed between DRACH and DRTCH. Statistical differences were assessed using a Kruskal−Wallis test with a p‐value < 0.05. Dunn's post‐hoc test was performed between each pair with Benjamini−Hochberg correction. (C) Motifs are evenly distributed on the chromosomes. (F) Motif abundance in the genic (pink) and intergenic (blue) regions.

Overall, A/T rich motifs and submotifs were more abundant than G/C rich ones, consistent with the low GC content of the oyster genome (33.5%). This compositional bias may partly explain the lower frequencies of DRCCH and DRGCH relative to DRACH and DRTCH. In the genome, the DRTCH motif is complementary to half of the DRACH sequences on the opposite DNA strand (AGACT, AGACA, AGACC, TGACA, TGACC, TGACT, GGACA, GGACC, GGACT). In the transcriptome, however, this symmetry is absent because RNA transcripts are single‐stranded. This likely explains why DRTCH motifs are more represented in the genome than in the transcriptome. These findings highlight the need for a normalized metric that enables unbiased estimation of DRACH motif abundance in transcriptomes regardless of the GC content.

3.2. DRACH o/e Metric Across the Oyster Genome

To overcome GC‐content–related biases and enable comparisons across genomic contexts and species, we developed the novel DRACH o/e metric, which normalizes motif abundance according to sequence length and GC content. This normalization strategy was inspired by approaches widely used in epigenetic studies, in which the DNA CpG o/e ratio serves as a proxy for DNA methylability and long‐term methylation patterns across species [45−48]. The DRACH o/e score was calculated as the ratio between the observed number of DRACH motifs and the number expected under a random distribution, and normalized by transcript length and the GC content of the genomic region under consideration (transcriptome or intergenic regions). This approach provided robust normalization across sequences with different nucleotide compositions (Figure S1A−F).

Comparison of the resulting normalized DRACH content between genic and intergenic regions revealed a genome‐wide enrichment of DRACH motifs in the oyster genome, reaching 16% and 9%, respectively. This corresponds to DRACH o/e peaks of approximately 1.16 and 1.09 in genic and intergenic regions, respectively (Figure 2A), consistent with the raw counts reported above (Figure 1F). Those differences support the hypothesis that DRACH motifs are evolutionarily conserved because of their role in m⁶A‐mediated RNA regulation.

To further characterize DRACH enrichment within genes, DRACH o/e values from genic regions were standardized by calculating z‐scores, thereby centering the distribution around zero. High z‐scores identify transcripts with elevated DRACH o/e values (DRACH‐rich), whereas low z‐scores correspond to transcripts with low DRACH o/e values and consequently reduced DRACH motif density (Figure 2B). This transformation generated a Gaussian‐like distribution with heavy tails. Although the distribution deviated from strict normality, empirical z‐score thresholds were used to define extreme DRACH‐enriched and DRACH‐depleted transcript populations, enabling the statistical identification of genes with markedly divergent DRACH content. Accordingly, transcripts with z‐scores > 2 were classified as DRACH‐enriched (DRACHen; 1685 genes), whereas those with z‐scores < –2 were classified as DRACH‐depleted (DRACHdep; 986 genes).

These two transcript populations were subsequently used to explore the potential functional significance of DRACH motif abundance. Chromosomal mapping revealed that DRACHen and DRACHdep genes are not randomly distributed across the genome, with DRACHen genes preferentially located near the ends of several chromosomes (Figure 2C). This pattern was specific to DRACH motifs and was not observed for the nonmethylable DRCCH sequence, suggesting that methylation potential influences the genomic distribution of these motifs. Consistent with this interpretation, the distribution of genes according to their DRCCH content z‐score was broader, and both DRCCH‐enriched and DRCCH‐depleted genes were randomly distributed throughout the genome (Figure S2A−C).

This nonrandom chromosomal distribution was quantified using an “edginess” index, which measures gene proximity to chromosome ends (ranging from 0 at the 5′ or 3′ ends to 1 at the chromosome center). DRACHen genes exhibited significantly lower edginess values than those of both DRACHdep genes and genes with intermediate z‐scores, indicating preferential localization near chromosome ends. In contrast, no significant difference was observed between DRACHdep genes and genes with intermediate DRACH content (Figure 2D). Similar patterns were observed when DRACH motifs were analyzed in 1 kb genomic windows (Figure S2D).

We next examined the distribution of DRACH motifs within transcripts. In DRACHen mRNAs, motifs were predominantly located within the coding sequence (CDS) and in the 3′UTR, whereas their relative abundance within the 3′UTR was reduced in DRACHdep transcripts (Figure 2E). In contrast, noncoding RNAs (ncRNAs) displayed similar intratranscript distributions of DRACH motifs in both DRACHen and DRACHdep groups. However, upstream and downstream flanking regions were enriched in DRACH motifs around DRACHdep genes but not around DRACHen genes (Figure 2F). The apparent enrichment observed 1 kb upstream and downstream of DRACHdep transcripts is likely attributable to the relatively low number of DRACH motifs within these genes.

The preferential localization of DRACH motifs within coding regions and 3′UTRs of DRACHen transcripts suggests a targeted deposition of m⁶A marks, thereby promoting the recruitment of distinct reader proteins and regulating specific biological processes. GO enrichment analysis indicated that DRACHen genes were associated with gene expression regulation, including epigenetic processes (e.g., “histone H4K12 and H4K14 acetylation,” “heterochromatin organization involved in chromatin silencing,” RNA and DNA polymerase, and ribosomes), neuronal functions (e.g., anterograde axonal transport of mitochondrion, negative regulation of axon extension involved in regeneration, anterograde dendritic transport of neurotransmitter receptor complex/of messenger ribonucleoprotein complex, positive regulation of voltage‐gated sodium channel activity, associative learning, positive regulation of transmission of nerve impulse), cell cycle (e.g., regulation of mitotic spindle elongation, first cell‐cycle pseudocleavage, spindle localization, establishment of meiotic spindle orientation, aster), and cytoskeleton (e.g., microtubule sliding, plus‐end–directed vesicle transport along microtubule, aster, microtubule lateral binding, actin cap, dense core granule, cytoskeletal transport). In contrast, DRACHdep genes displayed only a few enriched functions relative to background and were limited to molecular function categories mainly related to transport activity (Figure 2G).

3.3. Conservation of m⁶A Machinery Proteins Across Metazoans

To further understand the epitranscriptomic regulation in an evolutionary and ecological perspective, we extended our analyses to eight additional species, spanning bilaterian phylogeny. Species were selected according to (i) the availability of a chromosomal‐level genomic assembly, (ii) their phylogenetic position, and (iii) their reproductive strategy. The selection comprises two deuterostomes (Ciona intestinalis and Homo sapiens), six protostomes including three ecdysozoans (the insects Apis mellifera, Culex pipiens, and Drosophila melanogaster), and three mollusks (Mytilus trossulus, Ostrea edulis, and M. gigas), together with the cnidaria Acropora millepora as an outgroup.

To evaluate the evolutionary conservation of the m⁶A epitranscriptomic machinery, we first compared the domain architecture and their functions, then the full‐length protein sequences across the nine examined species (Figure 3 and Figure S3).

Full‐length protein analysis revealed that the core m⁶A writer components METTL3 and METTL14 were the most highly conserved, with sequence identities ranging from 55% to 69% and 63% to 70%, respectively. The remaining writer‐complex subunits were less conserved: WTAP (39%–54%), RBM15 (39%–47%), and HAKAI (38%–48%), whereas the MACOM cofactors VIRMA (32%–40%) and ZC3H13 (29%–37%) were the most divergent writer components. VIRMA was not detected in A. mellifera. The eraser ALKBH5 exhibited moderate conservation (57%–60%) in species in which it was identified but was absent from all three insects examined (A. mellifera, C. pipiens, and D. melanogaster). Among readers, YTHDC1 (41%–49%), YTHDF2 (46%–56%), and YTHDC2 (39%–57%; not detected in D. melanogaster) displayed moderate conservation, whereas IGF2BP2 was the least conserved overall (33%–48%), reaching its lowest identity in A. millepora (33%) (Figure S3A).

Despite variation in protein length, domain architecture was markedly conserved among orthologs. METTL3 and METTL14 each contained a single MT‐A70 domain. METTL16 (Methyltransf_10) was not detected in C. intestinalis. The MACOM components WTAP, RBM15 (three RRM domains and one SPOC domain), and HAKAI (RING‐Hakai and zf‐Hakai domains) displayed fully conserved architecture across all nine species. VIRMA (VIR_N domain) was identified in every species except Apis mellifera. ZC3H13 showed greater divergence: both the zf‐CCCH and FT Rho domains were conserved in vertebrates, lophotrochozoans, and Acropora cervicornis, were not detected in C. intestinalis, D. melanogaster, or C. pipiens, and were reduced to a single zf‐CCCH domain in Apis laboriosa. The eraser ALKBH5 (2OG‐FeII_Oxy domain) was absent from the three insects examined, supporting its widespread loss within the Endopterygota, although not throughout all insects [49]. Among readers, YTHDC1, YTHDC2, and YTHDF2 all retained a conserved C‐terminal YTH despite variation in protein length. YTHDC2 displayed the most complex organization (R3H, HrpA, DEXHc, HA2, OB, YTH), all of which were conserved wherever the ortholog was identified. IGF2BP2 (two RRM and four KH domains) lacked one or both RRM domains in several invertebrates, notably Acropora muricata. In contrast, hnRNPA2B1 consistently displayed two RRM domains.

Comparison at the domain level revealed substantially higher conservation than that of full‐length sequence analysis, indicating stronger evolutionary constraints on functional domains. The METTL3 MT‐A70 domain was the most conserved (79%–89%), followed by METTL14 MT‐A70 (69%–81%), the YTH domains of YTHDC1 (61%–70%), YTHDF2 (70%–81%), and YTHDC2 (43%–56%), and the ALKBH5 2OG‐FeII_Oxy domain (66%–77%). HAKAI (58%–72%) and hnRNPA2B1 (45%–65%) displayed moderate conservation, while the regulatory MACOM cofactors VIRMA (VIR_N: 34%–49%) and ZC3H13 (zf‐CCCH and FT Rho: 37%–47%, when detected) remained comparatively divergent. IGF2BP2 was the least conserved (38%–52%) (Figure 3 and Figure S3B). Given the overall conservation of the m⁶A machinery in metazoans, we next investigated the epitranscriptomic features using the DRACH o/e metric across these species.

3.4. DRACH o/e, DRACHen, and DRACHdep Distribution Across Phylogeny

As observed in oysters, comparison of the frequencies of methylable DRACH motifs with those of the nonmethylable DRCCH, DRGCH, and DRTCH motifs across species transcriptomes revealed that the DRACH motif was the most prevalent in all three mollusks, as well as C. pipiens and C. intestinalis, whereas DRGCH predominated in humans, honeybees, drosophila, and corals (Figure 4). The abundance of each DRACH submotif was quantified and normalized. Pairwise comparisons of normalized DRACH submotif abundance revealed strong correlations between species (Pearson's r = 0.660−0.994; Figure 4B). In contrast, similar analyses performed on the nonmethylable DRCCH, DRTCH, and DRGCH motifs yielded weaker correlations (DRCCH: −0.211 to 0.987; DRGCH: 0.144–0.984; DRTCH: −0.007 to 0.992) (Figure 4B,C and Figure S4A−D), suggesting stronger evolutionary constraints on methylable motifs than on their nonmethylable counterparts. Based on their DRACH submotif composition, species clustered according to phylogenetic relationships, indicating that evolutionarily related taxa share similar DRACH distribution patterns. Despite substantial variation in DRACH submotif composition, amino acid composition was highly conserved across the proteomes of the nine species (Pearson r > 0.9 for all species pairs), suggesting that variation in submotif composition is unlikely to result from codon‐driven amino acid usage (Figure S4E). This is further supported by the higher correlation scores for DRACH submotif frequencies than for those of DRCCH and DRTCH motifs. Although these differences were not significant, the lowest pairwise correlation observed for DRGCH (0.144) remained substantially lower than that for DRACH (0.660) (Figure 4C and Figure S4C).

To explore whether DRACH enrichment follows a phylogenetically conserved genomic organization, we calculated the DRACH o/e values, associated z‐scores, and identified DRACHen and DRACHdep genes (Figure S4F,G), and compared their chromosomal distribution across species (Figure 4D). Differences between DRACHen and DRACHdep populations were quantified using the edginess metric. As in M. gigas, other mollusks displayed distinct patterns of gene localization. Within Ostreida (M. gigas and O. edulis), DRACHen genes were located significantly closer to chromosome ends than both DRACHdep genes and the genomic background, the latter being evenly distributed along chromosomes. In the mussel M. trossulus, DRACHen genes were likewise enriched near chromosome ends, whereas DRACHdep genes were preferentially localized toward chromosome centers. An opposite pattern was observed in H. sapiens, where DRACHen genes were preferentially located near chromosome centers and DRACHdep genes near chromosome ends. In the cnidarian A. millepora, DRACHen distribution did not differ from background genes, whereas DRACHdep genes were significantly closer to chromosome centers. By contrast, no significant chromosomal distribution bias was detected in insects or in the chordate C. intestinalis.

Collectively, these results indicate that the evolutionary conservation of DRACH motifs is shaped by clade‐specific and/or functional constraints rather than by stochastic sequence variation alone. They further suggest that the genomic distribution of DRACH‐rich and DRACH‐poor genes follows a phylogenetically structured pattern. Although the inference of the ancestral state would require a broader taxonomic sampling, the DRACH motif distribution appears to be strongly clade‐specific, with particularly divergent patterns observed between insects and mollusks.

3.5. Functional Significance and Evolutionary Conservation of DRACH Enrichment

Although DRACH motif abundance is highly conserved across metazoan phylogeny, the genomic distribution of DRACH‐rich and DRACH‐poor genes is clade‐specific. This raises the question of whether DRACH enrichment is associated with conserved biological functions across species. To address this, we performed a KOG enrichment analysis to identify functional categories associated with DRACH abundance. The analysis included 21 functional categories, and statistical significance was assessed using delta rank values derived from DRACH o/e z‐scores (see Methods), with positive delta rank values corresponding to DRACH‐rich gene sets and negative values to DRACH‐depleted gene sets.

The resulting KOG enrichment heatmap revealed substantial variation in the relationship between DRACH content and gene function across species (Figure 5A). Nevertheless, several functional categories exhibited consistent patterns. For example, genes related to cytoskeleton organization were consistently enriched in DRACH‐rich genes, whereas the “inorganic ion transport and metabolism” category is associated with DRACH‐poor genes in all examined species. Other functional categories showed clear clade‐specific associations. In mollusks, the “RNA processing and modification,” “translation, ribosomal structure and biogenesis,” and “replication, recombination and repair” were predominantly enriched among DRACHen transcripts, whereas these same functional categories were preferentially associated with DRACHdep genes in the mosquito. Interestingly, some functional categories exhibited similar DRACH enrichment patterns in phylogenetically distant species while differing among closely related taxa. For example, “cell cycle control, cell division, and chromosome partitioning” show significant enrichment in DRACH‐rich transcripts in A. mellifera, M. trossulus, and M. gigas.

FIGURE 5.

Functional conservation of genes according to their DRACH content through evolution. (A) KOG enrichment heatmap showing the 21 KOG categories and their respective enrichment in each species according to the delta rank value, determined by transcript z‐score. The delta rank value is a ranking of the z‐score value from highest to lowest. A high delta rank corresponds to a DRACH‐rich transcript. (B) Correlation test of the delta rank values between species. Abbreviations: A. mil, Acropora millepora; C. int, Ciona intestinalis; H. sap, Homo sapiens; A. mel, Apis mellifera; C. pip, Culex pipiens; D. mel, Drosophila melanogaster; M. tro, Mytilus trossulus; O. edu, Ostrea edulis; M. gig, Magallana gigas.

Correlation analysis of delta rank scores further supported these observations (Figure 5B). Molluscan species exhibited strong correlations (Pearson's r = 0.838–0.876 (***)), whereas correlations drop sharply between mollusks and insects (D. melanogaster: 0.353< r < 0.434; A. mellifera: 0.515< r < 0.651). C. pipiens appeared as a clear outlier, with correlation coefficients close to zero or negative (–0.456< r < 0.107). Outside mollusks, H. sapiens showed the greatest similarity to C. intestinalis and D. melanogaster regarding the functional distribution of DRACHen genes. A. millepora displays moderate correlations with most species, except C. pipiens and H. sapiens (0.438< r < 0.576). As a control, a similar analysis was performed using nonmethylable DRCCH motifs. Correlation coefficients obtained for DRACH were significantly higher than those obtained for DRCCH (Figure S5), providing additional evidence that methylable motifs are subject to stronger evolutionary constraints.

4. Discussion

Early studies of m⁶A RNA methylation identified the RRACH motif (R = A/G, H = A/C/U) as the consensus sequence [2, 50]. However, this definition emerged primarily from studies in mammals, and subsequent analyses revealed that many m⁶A peaks occur outside RRACH but always within the broader DRACH context (D = A/G/U, R = A/G, H = A/C/U) [4, 51]. We, therefore, adopted DRACH as the reference consensus for cross‐species comparisons, particularly in nonvertebrate taxa, as it provides a more comprehensive representation and a higher accuracy for the inclusion of methylable sites.

Because m⁶A is a highly variable, transient, and dynamically regulated modification, and because epitranscriptomic landscapes vary according to species [6, 34, 52, 53], developmental stage [6, 34], and tissue [21], direct comparisons of epitranscriptomes across phylogenetically distant organisms are inherently challenging. In mammals, methylations are deposited stochastically on accessible DRACH motifs during transcription [54−56]. During splicing, steric hindrance from exon‐junction complex occupancy prevents METTL3/METTL14 MTC from methylating sites near exon boundaries. Consequently, longer exons, which are predominantly lying near stop codons, accumulate more m⁶A marks, giving rise to the characteristic enrichment of m⁶A around the stop codon in mature mRNAs [50, 57]. Therefore, the methylation status of DRACH motifs is highly dynamic throughout a transcript lifetime, and experimental snapshots of the actual methylation profiles may fail to capture the full spectrum of biological outcomes of RNA m⁶A‐methylation. To circumvent the intrinsic variability of m⁶A marks, we investigated whether the abundance of DRACH motifs within nucleic acid sequences could provide insights into the evolution of the m⁶A‐mediated regulatory mechanisms. We hypothesized that transcripts containing more DRACH motifs have a greater probability of undergoing m⁶A‐mediated regulation than poorly methylable ones. Accordingly, we propose that DRACH abundance reflects the intrinsic methylation potential, or methylability, of a transcript. To address the limitations and biases arising from differences in GC content and transcript length among genomes and transcriptomes, we developed a new indicator—the DRACH observed over expected, or DRACH o/e score—which normalizes DRACH abundance across sequences and enables meaningful comparisons between genomic contexts and species. A DRACH o/e value greater than 1 indicates enrichment of DRACH motifs relative to expectation, whereas values below 1 indicate depletion.

The Pacific oyster primarily served as a lophotrochozoan model for the study of epitranscriptomic regulation in distant groups [6, 24, 37]. In the oyster, in contrast to nonmethylable motifs whose frequency is lower than or equivalent to what is expected, DRACH motifs are overrepresented across the genome and particularly within the gene regions, suggesting that these motifs are actively conserved. We further demonstrated that mRNAs (coding transcripts) exhibit distinct DRACHen/DRACHdep profiles, unlike ncRNAs. Together with the previously reported role of m⁶A marks during splicing in the oyster [37], and also consistent with their distinct functional outcomes depending on their localization on mammalian transcripts [14, 18], this further supports an evolutionarily conserved role for m⁶A‐methylation in post‐transcriptional and/or translational regulation. Conversely, the uniform distribution of DRACH motifs in DRACHen and DRACHdep ncRNAs is likely associated with the absence of compartmentalization compared to mRNAs (5′UTR, CDS, 3′UTR …). It may also reflect that transcript methylability has been selectively shaped primarily in protein‐coding genes, whose expression is subject to stronger, more direct selective pressure than ncRNAs. In the oyster, DRACH‐rich genes were physically localized near chromosome ends and associated with processes requiring rapid and tightly regulatory responses, such as cell‐cycle, gene expression including epigenetic regulation, and neuronal functions. In contrast, DRACH‐poor genes were less strongly associated with specific functions (only five GO terms were found significantly enriched), suggesting these DRACHdep population differs only modestly from the transcriptomic background. This is consistent with prior observations that housekeeping transcripts generally display lower m⁶A‐methylation than inducible transcripts [12, 58]. Additional support arises from functional studies of the m⁶A machinery in mammals. For example, deletion of METTL14 in mice alters inducible pathways like apoptosis, inflammation, and PPAR/AMPK signaling [59], and METTL3 KO in human cells affect histone modifications and MAPK pathway activity [60]. Likewise, ALKBH5 inactivation impacts neuronal functions and immune response [61]. Similar observations have been reported in drosophila [7, 53] and oysters [24, 37], collectively supporting the idea that transcripts with high methylation potential preferentially participate in inducible biological processes. Consistently, Golomb et al. demonstrated that tissue‐specific gene expression correlates with the density of promoter cis‐regulatory elements by introducing the tau index, which ranges from 0 for housekeeping genes to 1 for tissue‐specific genes, and serves as a quantitative measure of transcriptional specificity [62]. Comparing tau indices with DRACH o/e values may, therefore, provide a useful framework for investigating the contribution of epitranscriptomic regulation to tissue‐specific gene expression, particularly given the highly dynamic nature of m⁶A deposition and the scarcity of functional epitranscriptomic studies in most nonmodel organisms.

To investigate the potential evolutionary constraints acting on DRACH motifs, we expanded our analysis to include eight additional metazoan species. Species were selected from related clades while encompassing distinct ecological traits to try and disentangle the respective contributions of phylogenetic history and ecological drivers. Protein structure analyses revealed a strong conservation of the m⁶A‐machinery across the analyzed species. The writer complex, comprising the catalytic METTL3‐METTL14 core together with the regulatory MACOM cofactors WTAP, RBM15, HAKAI, VIRMA, and ZC3H13, was retrieved across all major metazoan lineages, supporting the conservation of an m⁶A regulatory system, as previously anticipated [6]. Although the catalytic writers and most cofactors displayed full domain conservation, ZC3H13 showed lineage‐specific divergence, with both its zf‐CCCH and FT Rho domains being undetectable in C. intestinalis and the dipterans D. melanogaster and C. pipiens. Likewise, the m⁶A readers YTHDC1, YTHDC2, YTHDF2, IGF2BP2, and hnRNPA2B1 were broadly conserved across phyla. In contrast, the eraser ALKBH5 was not detected in the insects analyzed here, though it is present in other insect species [49]. This observation nevertheless suggests a particular evolutionary history for the ALKBH‐family demethylases in insects [49]. Collectively, these results demonstrate a deep conservation of the m⁶A regulatory machinery throughout metazoan evolution, thereby providing a robust framework for comparative analyses of m⁶A target sequences. We, therefore, used the DRACH consensus motif to investigate the evolutionary diversification of potential m⁶A target sites across genomes and transcriptomes.

Comparative analysis using the DRACH o/e metric revealed distinct chromosomal distributions of DRACH‐rich genes among major metazoan clades. DRACHen genes were preferentially localized near chromosome ends in mollusks, localized toward chromosome centers in humans, and showed no significant positional bias in ecdysozoans. These clade‐specific edginess patterns suggest that genome organization associated with m⁶A regulation has diverged during bilaterian evolution. Therefore, our observations suggest that m⁶A‐RNA methylation contributes not only to RNA processing but also possibly to higher‐order chromatin organization. This is consistent with previous observations that genes with CpG‐rich promoters, whose expression is regulated by DNA‐5mC methylation, have more m⁶A‐methylated transcripts than other genes [21]. This crosstalk between m⁶A‐RNA and 5mC‐DNA is well documented [22, 51, 63] and may provide a potential feedback loop of epitranscriptomic regulation of chromatin structure. For instance, in mammals, m⁶A mediates Xist lncRNA recruitment of proteins involved in X‐chromosome compaction and silencing through DNA methylation and histone modifications [64−67]. Recent evidence further indicates that m⁶A‐RNA can also modulate DNA methylation by cotranscriptionally binding readers that recruit DNA and/or histone modifying enzymes [63, 68, 69] and/or chromatin‐conformation regulators such as Polycomb complexes, thereby influencing the compaction and transcriptional state of adjacent chromatin regions [51, 63, 70, 71]. Alternatively, or in addition, m⁶A‐mediated regulation of chromatin structure may include epitranscriptomic regulation of transposable element activity [72−74].

Analyses of transcriptome‐wide motif abundance further demonstrated that DRACH is not universally the predominant motif across metazoans. In mollusks and C. intestinalis, DRACH motifs were consistently enriched relative to expectation, whereas the nonmethylable control motifs were not. By contrast, in insects and humans, DRGCH and DRCCH motifs were more abundant despite lacking methylation potential. Although DRACH motifs are highly conserved at the sequence level across species, their methylation status—and consequently their biological functions—remain highly context‐dependent, varying with developmental stage, tissue, physiological state, and the abundance of reader proteins [6, 21, 34, 75, 76]. Overall, phylogenetic proximity was associated with greater conservation of DRACH enrichment patterns, although this is not a strict rule. While some functional categories, such as cytoskeleton organization, exhibited consistent DRACH enrichment across species, others displayed marked clade‐ or species‐specific patterns. For example, RNA processing, translation, or cell cycle control show differential DRACH enrichment between mollusks and insects, highlighting that functional deployment is shaped by additional evolutionary or ecological drivers beyond phylogeny alone. The distinct position of C. pipiens and the moderate correlations observed for A. millepora further suggest that ecological characteristics, environmental constraints, or life‐history traits may also contribute to shaping DRACH distribution. To examine the potential influence of ecological traits, we compared species exhibiting contrasting reproductive strategies within the same phylogenetic groups. An r‐reproductive strategy is characterized by high fecundity, numerous offspring, and elevated mortality (C. intestinalis, M. trossulus, C. pipiens, M. gigas, D. melanogaster), whereas K‐reproductive strategy species have a longer lifespan and produce fewer, higher‐quality offspring [30] (O. edulis, H. sapiens, A. mellifera, A. millepora). For example, within oysters, O. edulis follows a K‐selected strategy, whereas M. gigas is a typical r‐selected species [77]. However, none of the DRACH‐related features examined, including chromosomal distribution, functional enrichment, or motif abundance, showed any association with reproductive strategy. At present, no obvious selective pressure can explain the evolutionary maintenance of DRACHen and DRACHdep genomic organization. Instead, it can be hypothesized that the observed clustering of nucleotides may rather rely on mechanisms such as DNA repair after meiosis, as is the case for G and C enrichment in mammals [78]. Nevertheless, because m⁶A methylation has well‐established functional consequences, it is likely that at least part of the observed DRACH organization reflects selective pressures whose molecular basis remains to be determined. Overall, our results support the existence of partially clade‐specific DRACH organization, although broader phylogenetic inclusion beyond the scope of the present work would be required to establish its evolutionary origin. Future studies should, therefore, extend this framework to more distantly related eukaryotes, including fungi and plants, incorporate additional ecological and life‐history variables, and investigate species that have secondarily lost components of the m⁶A machinery, such as C. elegans [79].

This exploratory study establishes a comparative framework for investigating epitranscriptomic regulation across evolutionary scales, while overcoming the variable and dynamic nature of RNA m⁶A marks. Our findings broaden the understanding while raising new questions about the evo‐devo implications of m⁶A‐regulation. Although phylogenetic relatedness partly explains the conservation of DRACH enrichment within functional categories, it does not fully account for the observed patterns, emphasizing the likely contribution of additional ecological and evolutionary factors. By combining comparative analyses of DRACH motif distribution with a comprehensive characterization of m⁶A machinery proteins across nine representative metazoan species, this work provides a methodological framework for investigating the evolutionary constraints associated with, and potentially driven by, RNA methylation.

Author Contributions

Natacha Clairet: Investigation, writing – original draft, writing – review and editing. Hélène Auger, Jérémy Le Luyer: Investigation, writing – review and editing. Maximilian Haeussler: Investigation, writing – review. Guillaume Rivière: Funding acquisition, project administration, investigation, writing – review.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material: nyas70413‐sup‐0001‐SuppMat.docx

Supplementary Material: nyas70413‐sup‐0002‐SuppMat.docx

Supplementary Material: nyas70413‐sup‐0003‐FigureS1.tif

Supplementary Material: nyas70413‐sup‐0004‐FigureS2.tif

Supplementary Material: nyas70413‐sup‐0005‐FigureS3.tif

Supplementary Material: nyas70413‐sup‐0006‐FigureS4.tif

Supplementary Material: nyas70413‐sup‐0007‐FigureS5.tif

Supplementary Material: nyas70413‐sup‐0008‐Table.tif

Supplementary Material: nyas70413‐sup‐0009‐Table.tif

Supplementary Material: nyas70413‐sup‐0010‐Table.tif