Genome-scale DNA methylome and transcriptome profiling of midgut of Bombyx mori infected with BmCPV

Abstract

DNA methylation, as well as histone modifications, is an important regulatory mechanism for altered gene expressions. Our previous study has shown that Bombyx mori cytoplasmic polyhedrosis virus (BmCPV) infection could change the level of trimethylation of lysine 9 of histone 3 (H3K9me3) and acetylation of lysine 9 of histone 3 (H3K9ac), thus regulating the mRNAs expressions in the midgut of silkworm, B. mori. However, the correlation between genome-scale DNA methylome and transcriptome remains underexplored. In this study, whole genome bisulfite sequencing (WGBS) was performed on the midgut of BmCPV-infected silkworms at 48 h and 96 h post infection, and corresponding midguts of uninfected silkworms. Above analysis will contribute to further understanding how BmCPV regulate gene expression through epigenetic modification at the genome-wide level.

Background & Summary

Bombyx mori, belonging to Lepidoptera, is one of the important economic insects. It has made important contributions to agriculture, textile and pharmaceutical industry. However, silkworm diseases cause damage to the growth and development of silkworm, resulting in huge economic losses. Among them, B. mori cytoplasmic polyhedosis virus (BmCPV) is one of the main pathogens. BmCPV is a double stranded RNA virus (dsRNA virus) belonging to the family Reoviridae and mainly infects silkworm midgut cells^1,2.

With the development of high-throughput sequencing, some researchers have explored the effects of BmCPV infection on gene expression at the genome-wide level. The microarray analysis has found 285 differentially expressed genes (DEGs) in BmCPV-infected midgut³. And the number of up-regulated genes in the BmCPV-infected midgut increased with the severity of BmCPV infection⁴. There were significant differences in DEGs between samples with persistent infection and those with pathogenic infection⁵. Besides, silkworm varieties could have influence on gene expression patterns in midguts of BmCPV-infected silkworm. The number of up-regulated genes in sensitive variety LAN 5 was more than that in resistant variety Ou 17, while the results about down-regulated genes was opposite⁶. Moreover, BmCPV infection could alter the expression profiles of miRNA⁷, circRNAs⁸ and lncRNAs⁹.

The pioneering opening of chromatin and its accessible state set the stage for initial genome-wide expression in eukaryotes, a process meticulously orchestrated by epigenetic regulatory mechanisms^10,11. DNA methylation, the first epigenetic modification discovered in eukaryote, was formed by covalent binding of the Methyl group to 5-carbon of cytosine^12–14. DNA methylation usually play vital roles in competitive binding with transcription factors to promoter regions^15–19, X chromosome inactivation in female animals^20–23, gene reprogramming^24–28, targeting repeats in genes^29,30 and alternative splicing^31,32. Previous studies have found that viral infection also could change DNA methylation levels in the hosts. Epstein-Barr virus (EBV)³³, Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)³⁴ and hepatitis B virus (HBV)³⁵ could increase DNA methylation levels in hosts’ genomes, further regulating gene transcript. As for silkworm pathogens, B. mori nucleopolyhedrovirus (BmNPV) and BmCPV were found to regulate DNA methylation. It was found that DNA methylation levels positively correlated with BmNPV proliferation³⁶. Inactivation of DNA methyltransferase could induce apoptosis of BmNPV-infected cells, so as to inhibiting virus proliferation³⁶. It was reported that the CpG methylation level in midguts were much higher than that in silk glands in the infected silkworm with BmCPV³⁷. These differentially methylated genes (DMGs) might be involved in RNA transport, RNA degradation, nucleotide excision repair, DNA replication and so on³⁷. However, the association between the DNA methylation level and the infection process of BmCPV remains unclear.

Also, Histone modifications are one of epigenetic modifications that alter gene expression patterns^38,39. Our previous study has reported the correlations between the trimethylation of lysine 9 of histone 3 (H3K9me3) / acetylation of lysine 9 of histone 3 (H3K9ac) and gene expressions⁴⁰. These intersection genes regulated by H3K9me3 were enriched in insect hormone biosynthesis and MAPK signaling, while the intersection genes regulated by H3K9ac were enriched in glycolysis/gluconeogenesis and TGF-beta signaling⁴⁰. Given that DNA methylation is also closely associated with gene expressions, we aim to employ whole-genome bisulfite sequencing (WGBS) for BmCPV-infected silkworm midguts at 48 h post-infection(hpi) and 96 hpi (CPV48 and CPV96) and corresponding midguts of uninfected silkworms (GUT48 and GUT96) to identify DMGs (Fig. 1). With integration of WGBS and RNA-seq datasets, we screened DEGs regulated by DNA methylation (Fig. 1). This study will advance our understanding of how BmCPV orchestrates genome-wide regulation of transcriptional priming and infection progression through epigenetic modifications.

Fig. 1.

Experimental design. Midgut tissues were collected from BmCPV-infected silkworms (CPV48 and CPV96 groups) and corresponding control silkworms (GUT48 and GUT96 group). These samples were subjected to WGBS and RNA-seq, and the resulting datasets were used for subsequent integrated analysis.

Methods

Silkworm strain, BmCPV infection, and Sample collection

Detailed experimental design, BmCPV infection, and sample collection have been described in our previous study^40,41. Briefly, silkworms (strain: Dazao) on the first day of the fifth instar were infected with BmCPV. At 48 hpi (Group CPV48) and 96 hpi (Group CPV96), midguts from 8 randomly selected silkworms were pooled, respectively. Midgut tissues from 8 age-matched healthy silkworms without viral infection served as controls (Group GUT48 and Group GUT96). Each group contained two biological replicates. Since BmCPV has specific tropism for the midgut of silkworm, and its infection causes B. mori cytoplasmic polyhedrosis, the midgut was selected for high-throughput sequencing in this study^1,2.

WGBS library preparation

Genomic DNA was extracted by phenol-chloroform method, and degradation and contamination were assessed by agarose gels electrophoresis. DNA purity was checked using NanoPhotometer®spectrophotometer (IMPLEN, USA). DNA concentration was measured using Qubit® DNA Assay Kit in Qubit® 2.0 Flurometer (Life Technologies, USA).

After quality assurance, the genomic DNA was fragmented into 200–300 bp with Covaris S220 (30 s/30 s for 10 times). These short sequences were treated with bisulfite using EZ DNA Methylation-Gold^TM Kit (Zymo Research, USA) to turn unmethylated cytosine (C) into Uridine (U) and remain methylated cytosine(C) unchanged. DNA libraries were constructed through Adaptase-mediated adapter ligation to single-stranded DNA (ssDNA) fragments, followed by PCR amplification. Library quality was initially assessed by quantification using a Qubit 2.0 at a dilution of 1 ng/μL. The insert size distribution was verified using an Agilent 2100 Bioanalyzer. DNA libraries that met the size requirements were subjected to precise quantification of effective concentration (>2 nM) by qPCR to ensure sequencing-ready quality.

WGBS data analysis

WGBS were performed on an Illumina HiSeq platform (Novogene, China). Read alignments were conducted using bismark software (version 0.16.3, https://www.bioinformatics.bbsrc.ac.uk /projects/bismark/). Then clean reads were obtained after trimming reads with low quality, including reads contaminated with adapters, reads with either end having a base quality score below 3 or containing N nucleotides, reads whose average base quality in the window falls below 15, reads shorter than 36 nucleotides (nt) and reads failing to form proper paired-end pairs.

After trimming, clean reads were mapped to B. mori genome (silkDB3.0, https://silkdb.bioinfotoolkits.net/main/species-info/-1). Uniquely aligned reads were cross-referenced with the native genomic sequence to systematically determine the methylation status of all cytosine loci. The average coverage depths of all cytosine loci across all samples exceeded 7.0 × (Table 1). Based on the three canonical DNA methylation contexts (CG, CHG, and CHH; H = A/T/C), we systematically quantified context-specific coverage (defined as the number of reads supporting methylation calls) for each cytosine context (CpG, CHG, CHH) (Table 1). Notably, CpG contexts exhibited a significantly higher average methylation levels relative to the CHG and CHH contexts (Table 1).

Table 1.

The proportion of methylated C sites in different sequence environments.

Sample	C covg Mean	C (Mb)	CG (Mb)	CHG (Mb)	CHH (Mb)	Mean C	Mean CG	Mean CHG	Mean CHH
GUT48-1	9.5	1653.9	351.2	265.7	1036.9	0.51%	1.17%	0.32%	0.33%
GUT48-2	8.9	1548.0	329.6	249.2	969.3	0.46%	0.96%	0.32%	0.32%
GUT96-1	7.9	1381.1	294.4	222.4	864.2	0.49%	1.09%	0.32%	0.33%
GUT96-2	8.5	1478.1	316.6	239.2	922.2	0.45%	0.88%	0.33%	0.34%
CPV48-1	8.0	1387.6	292.5	221.7	873.5	0.46%	0.91%	0.33%	0.34%
CPV48-2	8.4	1457.8	313.1	236.3	908.4	0.48%	1.06%	0.32%	0.33%
CPV96-1	7.4	1283.8	271.2	205.8	806.8	0.49%	1.16%	0.31%	0.31%
CPV96-2	7.1	1230.7	261.6	198.2	771.0	0.50%	1.17%	0.32%	0.32%
Summary (Median [IQR])	8.2 [7.4–8.5]	1422.7 [1381.1–1478.1]	303.8 [271.2–316.6]	229.4 [205.8–239.2]	891.0 [806.8–922.2]	0.485% [0.46%–0.49%]	1.075% [0.91%–1.16%]	0.32% [0.31%–0.32%]	0.32% [0.31%–0.32%]

All continuous variables are summarized as median with interquartile range (IQR), representing the 25th to 75th percentiles across all biological samples.

(1) C covgMean: Average coverage depth at all cytosine sites.

(2) C (Mb): Total bases (megabases) aligned to genomic cytosine positions.

(3) CG/CHG/CHH (Mb): Bases mapped to cytosines in CG/CHG/CHH contexts (megabases).

(4) MeanC (%): Average methylation level of all cytosine sites.

(5) MeanCG/CHG/CHH (%): Average methylation level in CG/CHG/CHH contexts.

Furthermore, methylation calling was performed via binomial distribution analysis incorporating three key parameters: methylated cytosine counts (mC), total coverage (mC + unmethylated cytosine umC), and bisulfite non-conversion rate (r). Genomic positions with statistically significant methylation (FDR-adjusted p < 0.05) were designated as methylated loci. For regional methylation quantification, the genome was partitioned into consecutive 10-kb bins using a sliding window approach, and methylation levels for each bin were calculated by quantifying the total number of methylated and unmethylated reads within each genomic interval, defined as follows:

$$\mathit{ML}\left(\mathrm{C}\right)=\frac{\mathit{reads}\left(\mathrm{mC}\right)}{\mathit{reads}\left(\mathrm{mC}\right)+\mathit{reads}\left(\mathrm{um}\mathrm{C}\right)}$$

Where, ML(C) represented methylation level, mC represented the number of methylated C, umC represented the number of unmethylated C.

To ensure the reliability of the experimental results and the rationality of sample selection, we calculated the methylation levels within each sequence context in each 2 Kbp bins and performed Pearson correlation analysis⁴² and principal component analysis (PCA). The correlations coefficients (R2) of methylated CpG (mCG) between two samples within the same group were all above 0.9 (Fig. 2A). Also, PCA plots based on CG methylation further showed clear separation of samples from different time points (Fig. 2B), whereas those for CHG and CHH methylation exhibited indistinct separation (Supplementary Fig. 1). In addition, heatmaps were generated to compare the consistency of mCG (Fig. 2C), mCHG (Supplementary Fig. 2A) and mCHH (Supplementary Fig. 2B) methylation-level distributions across functional regions in all samples.

Fig. 2.

Correlation and clustering heatmap of CG methylation levels across all samples. (A) Pearson correlations analysis of CG methylation levels in all samples. (B) PCA plot of CG methylation levels across all samples. (C) clustering heatmap of CG methylation levels across all samples. CGI, CpG islands; CGI shore, CpG island flanking regions (±2 kb); 5′ UTR, 5′ untranslated regions; 3′ UTR, 3′ untranslated regions; Repeat, repeated regions.

To Investigate potential effects of BmCPV infection on genome-wide DNA methylation in B. mori, we statistically analyzed context-specific methylation levels of cytosines across functional genomic regions⁴³. As shown in Fig. 3, after merging biological replicates, methylation levels were computed across 20 equally sized bins within each functional region using the formula: mC / (mC + umC). The results were visualized, and no statistical tests were performed for the analyses presented in Fig. 3. Methylation levels in the CG contexts were ranked from highest to lowest as follows: 3′ untranslated region (3′ UTR), exons, 5′ UTR, and introns, while CpG islands (CGIs) and repeated regions displayed lower methylation levels (Fig. 3). Subsequently, differentially methylated regions (DMRs) were identified using the DSS software, which implements a dispersion shrinkage method for estimating dispersion parameter from Gamma-Poisson or Beta-Binomial distributions^44–46. Based on the genomic distribution of DMRs, DMR-associated genes were defined as those satisfying either of the following criteria: (1) DMRs located within their transcriptional units (from transcription start site (TSS) to transcription end site (TES)), or (2) DMRs present in proximal promoter regions (defined as the region 2-kb upstream of TSS). A Wald test was performed at each cytosine site, and DMRs were identified based on the p-values. The analysis parameters in DSS software were set as follows: smoothing = TRUE, smoothing.span = 200, delta = 0, p.threshold = 1e-05, minlen = 50, minCG = 3, dis.merge = 100, and pct.sig = 0.5. Specifically, a genomic region was designated as a DMR only if it met all of the following criteria: (1) more than 50% of the CpG sites within the region had a p-value < 1e-05, (2) The region contained at least 3 CpG sites, and (3) the length of the region was at least 50 bp. Additionally, adjacent DMRs separated by less than 100 bp were merged into a single region.

Fig. 3.

DNA methylation levels in different genomic functional regions. (A–D) DNA methylation levels in different genomic functional regions in the GUT96 vs GUT48 (A), CPV48 vs GUT48 (B), CPV96 vs GUT96 (C), and CPV96 vs CPV48 (D) comparison groups, respectively. The x-axis denotes various genomic functional elements; y-axis indicates methylation levels.

Based on the genomic distribution of DMRs, DMR-associated genes were defined as those whose gene body regions (from TSS to TES) or promoter regions (2 kb upstream of the TSS) overlapped with DMRs. Given the critical role of promoters in gene transcription, hyper- and hypo-promoter DMRs were categorized for each group and summarized in Table 2. Also, the counts of hyper- and hypo-promoter DMRs and the median effect sizes (Δβ) were determined (Table 2).

Table 2.

Counts and Median Δβ of promoter DMRs.

Group	Counts of hyper promoter DMRs	Median Δβ (Hyper)	Counts of hypo promoter DMRs	Median Δβ (Hypo)
GUT 96 vs GUT48	13 + 0.273	58	−0.216
CPV 48 vs GUT48	35 + 0.228	95	−0.233
CPV 96 vs GUT 96	176 + 0.259	63	−0.288
CPV 96 vs CPV 48	105 + 0.255	33	−0.272

Detailed software parameter settings are provided in Table 3.

Table 3.

All primers used in this manuscript.

Primer	Sequence (5′ to 3′)	Application
BMSK0010150 in GUT96 vs GUT48 group
10150-MF	TTTCGCGAGGTTATTTATTAGTTCG	MSP
10150-MR	ATACCTTTTACAATAAACGTTCCGTT	MSP
10150-UF	TTTTGTGAGGTTATTTATTAGTTTGG	MSP
10150-UR	ATACCTTTTACAATAAACATTCCATT	MSP
BMSK0012101 in CPV48 vs GUT48 group
12101-MF	GGCGGGATTTGTTAATAGTTATTC	MSP
12101-MR	AACTCTACAAAAAATTCGACGTACA	MSP
12101-UF	GGGTGGGATTTGTTAATAGTTATTT	MSP
12101-UR	AACTCTACAAAAAATTCAACATACA	MSP
BMSK0007114 in CPV96 vs GUT96 group
07114-MF	TGAATATTATATTAGTTTGTCGCGG	MSP
07114-MR	TTAAACAAATAACCTTAACTACGCA	MSP
07114-UF	TTTTGAATATTATATTAGTTTGTTGTGG	MSP
07114-UR	AACTTCTTAAACAAATAACCTTAACTACAC	MSP
BMSK0006886 in CPV96 vs CPV48 group
06886-MF	ATAAATAAAAGGAATTCGTTCGACG	MSP
06886-MR	GCTAAAACTCGAAAATAACTAAACCGAT	MSP
06886-UF	TAAATAAAAGGAATTTGTTTGATGG	MSP
06886-UR	CTCACTAAAACTCAAAAATAACTAAACCAA	MSP
BMSK0000862 in CPV48 vs GUT48 group
00862M-F	AAATTATATTAATTATTAAGGGAGCG	MSP
00862M-R	AAAAAAATTCATAAAACGATAACGA	MSP
00862U-F	AATTATATTAATTATTAAGGGAGTG	MSP
00862U-R	AAAAAAATTCATAAAACAATAACAA	MSP
00862-qPCR-F	GGTGGGCTCCGCTGATGAAT	qPCR
00862-qPCR-R	AGCGTACGGCTGGTAGACCT	qPCR
BMSK0011815 in CPV 96 vs GUT96 group
11815M-F	TTGATTATTTTAATAGGTTAATTTACG	MSP
11815M-R	AATACAAATACACGAACTCCTCGT	MSP
11815U-F	TGATTATTTTAATAGGTTAATTTATG	MSP
11815U-R	CAATACAAATACACAAACTCCTCAT	MSP
11815-qPCR-F	GCCAAGATGGACGGGTCGAG	qPCR
11815qPCR-R	AGGATGGTGGTGGGCCTCTT	qPCR
BMSK0008917 in CPV 96 vs CPV48 group
008917M-F	AGGGATGGTAGATTAGAAGGAGC	MSP
008917M-R	CTATCAAAATAATCTAAAAATAACGT	MSP
008917U-F	TAAGGGATGGTAGATTAGAAGGAGT	MSP
008917U-R	CTATCAAAATAATCTAAAAATAACAT	MSP
08917-qPCR-F	GGTGGGCTCCGCTGATGAAT	qPCR
08917-qPCR-R	AGCGTACGGCTGGTAGACCT	qPCR

Data Records

WGBS data have been deposited in NCBI Sequence Read Archive (SRA) with accession numbers SRR33787510⁴⁷, SRR33787511⁴⁸, SRR33787512⁴⁹, SRR33787513⁵⁰, SRR33787514⁵¹, SRR33787515⁵², SRR33787516⁵³ and SRR33787517⁵⁴. And RNA-Seq data are available in the NCBI database with accession numbers SRR28625435⁵⁵, SRR28625436⁵⁶, SRR28625437⁵⁷, SRR28625438⁵⁸, SRR28625439⁵⁹, SRR28625440⁶⁰, SRR28625441⁶¹, SRR28625442⁶².

The methylation bigWig tracks were deposited in the Gene Expression Omnibus (GEO) under accession number GSE313872⁶³. This dataset is explicitly linked to the raw sequencing data in the NCBI SRA under BioProject accession numbers PRJNA1270926 and PRJNA1099186, respectively. In addition, the per-sample methylation calls (including methylated/unmethylated counts and beta-values at single-C resolution) and DMR BED files are available in the figshare repository⁶⁴.

Data Overview

DNA methylation usually plays vital roles in gene transcription regulation. To explore the relationship between the transcriptional level and DNA methylation levels, we conducted a correlation analysis between DMR-associated genes and DEGs. Correlated genes were defined as those with an absolute value of log2FoldChange of RNA-seq FPKM was greater than 1, and the absolute value of methylation difference of the corresponding DMR was greater than 0.2.

Technical Validation

Quality control

Raw sequencing data were processed to remove low-quality reads, yielding clean reads. FastQC (v0.11.5) was used to assess the quality of clean reads. Over 97% of bases had a Q-score ≥ 20 (sequencing error rate < 1%), and more than 90% of bases had a Q-score ≥ 30 (error rate < 0.1%) (Table 4).These metrics confirmed the high quality of the sequencing data. In addition, bisulfite (BS) conversion efficiency was estimated by calculating the non-conversion rate at unmethylated cytosines across the genome, using the formula: BS conversion = 1 – (Percentage of C Reads / Total Reads). BS conversion rates for cytosines in three sequence contexts (CG, CHG, and CHH) are summarized for each sample in Table 4. As shown, conversion rates exceeded 99% across all samples and contexts, consistent with the typically low genome-wide DNA methylation levels observed in insects. Furthermore, methylated cytosines (mCs) and unmethylated cytosines(umCs) were quantified and summarized in Table 5. The number of umCs is more than that of mCs (Table 5). Histograms of mCs distribution across all samples were generated (Supplementary Fig. 3), which showed that the percent of methylated cytosines at each level was low than 20%. All these results showed the low methylation levels in B. mori.

Table 4.

WGBS FastQC-general statistics.

Sample	Raw Reads	Clean Reads	Clean ratio (%)	Q20 (%)	Q30 (%)	GC content (%)	BS Conversion Rate(%)
							Total C	CG	CHG	CHH
GUT48-1	67405163	66424359	90.75	97.14	91.27	19.09	99.659	99.660	99.660	99.655
GUT48-2	56598346	55727499	90.58	97.01	91.00	19.24	99.672	99.671	99.671	99.669
GUT96-1	54082888	53042209	90.20	97.25	91.41	19.25	99.673	99.682	99.682	99.660
GUT96-2	55102764	54163567	90.50	97.51	92.08	19.34	99.659	99.655	99.655	99.655
CPV48-1	53549181	52002038	88.36	96.66	90.35	19.05	99.652	99.650	99.650	99.647
CPV48-2	60648766	59043544	88.68	96.74	90.50	19.43	99.666	99.666	99.666	99.662
CPV96-1	54181255	52861623	89.17	97.11	91.20	19.03	99.677	99.684	99.684	99.669
CPV96-2	51818336	51038995	90.74	97.57	92.26	19.16	99.656	99.667	99.667	99.652

Table 5.

Counts of methylated cytosines(mCs) and unmethylated cytosines(umCs).

Sample	mC counts	umC counts
GUT48-1	419836	173808457
GUT48-2	357880	173870413
GUT96-1	394321	173833972
GUT96-2	289646	173938647
CPV48-1	280174	173948119
CPV48-2	383240	173845053
CPV96-1	378396	173849897
CPV96-2	372372	173855921

Bismark software (version 0.16.3, https://www.bioinformatics.bbsrc.ac.uk /projects/bismark/) was then used with the parameter -X 700 --dovetail⁶⁵ to align bisulfite-treated sequencing reads to the B. mori reference genome (SilkDB 3.0; https://silkdb.bioinfotoolkits.net/main/species-info/-1). Both of reference genome and clean reads were transformed into their bisulfite-converted version (C to T, G to A). Subsequently, bisulfite-converted clean reads were aligned to similarly converted reference genome. Uniquely aligned reads, determined by aligning positive and negative strands to the same regions in reference genomes, were used to infer the methylation status of all cytosine positions. Reads that aligned to the same regions were classified as duplicated reads. The proportion of uniquely aligned reads exceed 69% across all samples, with duplicate reads accounting for approximately 30% of the total aligned reads (Table 6), indicating that the data were adequate for downstream analyses⁶⁶. After excluding duplicate reads, statistical analysis of sequencing depth and coverage was performed to ensure the rigor and reliability of methylation calling. The data showed an average depth of ~20× across all cytosines, with more than 70% of the genome covered at ≥10× , a result supporting high-confidence methylation analysis (Table 1).

Table 6.

Mapping rate of clean reads to reference genomes.

Samples	Total reads	Mapped reads	Mapping rate(%)	Duplication rate(%)
GUT48-1	66424359	49838196	75.03	30.75
GUT48-2	55727499	44972091	80.70	28.09
GUT96-1	53042209	40068084	75.54	29.00
GUT96-2	54163567	44294965	81.78	30.36
CPV48-1	52002038	41409222	79.63	29.96
CPV48-2	59043544	42121664	71.34	28.31
CPV96-1	52861623	36543239	69.13	27.15
CPV96-2	51038995	35681361	69.91	28.53

Validations of DMRs and promoter-DMRs associated-DEGs by Methylation-specific PCR (MSP) and quantitative Real-time polymerase chain reaction (qPCR)

For DMRs validation, BMSK0010150 in the GUT96 vs GUT48 group, BMSK0012101 in the CPV48 vs GUT48 group, BMSK0007114 in the CPV96 vs GUT96 group and BMSK0006886 in the CPV96 vs CPV48 group were randomly selected to be verified by MSP. Candidate promoter region were identified using Integrative Genomics Viewer (IGV), and CpG islands were predicted with stringent criteria (GC content > 50%, observed/expected CpG ratio > 0.6, and length > 100 bp) using MethPrimer software. Specific primers (Table 3) were designed to amplify the derived sequences (100–250 bp in length) from both methylated and unmethylated bisulfite-converted DNA. After treatment, extracted genomic DNA was bisulfite-treated, followed by PCR amplification and Sanger sequencing. As shown in Supplementary Fig. 4A-D, the the MSP validation results were consistent with the those obtained by WGBS.

To validate the association between DEGs and promoter-DMRs identified by high-throughput sequencing, reverse transcription-quantitative PCR (RT-qPCR) and MSP were performed. BMSK0000862 in the CPV48 vs GUT48 group, BMSK0011815 in the CPV96 vs GUT96 group, BMSK0008917 in the CPV96 vs CPV48 group were randomly selected to be verified by MSP. Specific primers were shown in Table 3. Extracted DNA was bisulfite-treated, followed by PCR amplification and Sanger sequencing. As shown in Fig. 4A–C, the validation results were consistent with the those detected by WGBS.

Fig. 4.

Validation of DEGs associated with promoter DMRs. (A–C) MSP validation of partial promoter regions of BMSK0000862 in the GUT48 vs CPV48 group (A), BMSK0011815 in the CPV 96 vs GUT96 group (B), BMSK0008917 in the CPV 96 vs CPV48 group (C). The upper panel presents a positional distribution of DMR identified by WGBS. The lower panel shows the validation result obtained by MSP, where open and filled circles denote unmethylated and methylated cytosines, respectively. (D–F) qPCR validations of transcripts of BMSK0000862 (D), BMSK0011815 (E), BMSK0008917 (F). * p<0.05, ** p<0.01, *** p<0.001.

Meanwhile, specific primers (Table 3) were designed to detect the relative expression levels of BMSK0000862 in the CPV 48 vs GUT48 group, BMSK0011815 in the CPV 96 vs GUT96 group and BMSK0008917 in the CPV 96 vs CPV48 group via RT-qPCR. The results confirmed that the RT-qPCR data were consistent with the high-throughput sequencing data (Fig. 4D–F).

Data availability

WGBS data have been deposited in NCBI Sequence Read Archive (SAR) under BioProject PRJNA1270926 with accession numbers SRR33787510⁴⁷, SRR33787511⁴⁸, SRR33787512⁴⁹, SRR33787513⁵⁰, SRR33787514⁵¹, SRR33787515⁵², SRR33787516⁵³ and SRR33787517⁵⁴. The RNA-Seq data have been deposited in the NCBI SAR under BioProject PRJNA1099186 with accession numbers SRR28625435⁵⁵, SRR28625436⁵⁶, SRR28625437⁵⁷, SRR28625438⁵⁸, SRR28625439⁵⁹, SRR28625440⁶⁰, SRR28625441⁶¹, SRR28625442⁶². The methylation bigWig tracks have been deposited to the Gene Expression Omnibus (GEO) under the accession number GSE313872⁶³. These data are linked to the raw sequencing data in the NCBI SRA under BioProject accession numbers PRJNA1270926 and PRJNA1099186. In addition, sample-specific methylation calls (including methylated/unmethylated counts and single-C resolution beta-values) and DMR BED files are available in the figshare repository⁶⁴.

Code availability

Each methylation call file also includes binomial test p-values assessing whether methylation at each site is significantly above background within that sample. These values are provided for completeness but are not used in downstream differential analysis. No custom code was generated or applied in the analysis of the genomic data presented. All software tools and their corresponding parameter settings are detailed in Table 7.

Table 7.

Summary of software tools used in the current study.

Analysis	Software(version)	Parameters
Trim	fastp 0.20.0	--cut_front_window_size = 1 --cut_front_mean_quality = 3 --cut_tail_window_size = 1 --cut_tail_mean_quality = 3 --cut_right_window_size = 4 --cut_right_mean_quality = 15 --length_required = 36 --trim_front1 = 10 --trim_front2 = 10
QC	fastqc_v0.11.5
Mapping	Bismark(0.16.3)	--score_min L, 0, -0.2, -X 700 --dovetail
Deduplicate	Bismark(deduplicate_bismark)	--paired --samtools_path
mC calling	Bismark(bismark_methylation_extractor)	--multicore 4 --paired-end --no_overlap -ignore 5 --ignore_r2 5
DMR analysis	DSS(DSS_2.12.0)	smoothing.span = 200, delta = 0, p.threshold = 1e-05, minlen = 50, minCG = 3, dis.merge = 100, pct.sig = 0.5

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41597-026-06922-z.