Dataset on DNA methylation and gene expression changes induced by 5-aza-2′-deoxycytidine in Syrian golden hamster fetal cell cultures

Abstract

The Syrian hamster (SH) is an animal model used in virology, toxicology, and carcinogenesis, where a better understanding of epigenetic mechanisms is required. Finding genetic loci regulated by DNA methylation may assist in the development of DNA methylation-based in vitro assays for the identification of carcinogens. This dataset informs on the regulation of gene expression by DNA methylation. Primary cultures of SH male fetal cells (sex determined by differences in kdm5 loci on the X and Y chromosome) were exposed for 7 days to the carcinogen benzo[a]pyrene (20 µM) from which a morphologically transformed colony was collected and reseeded. The colony bypassed senescence and sustained growth. After 210 days of culture, the cells were collected and divided in 16 aliquots to create 4 experimental groups to test the effects of the DNA methylation inhibitor 5-aza-2′-deoxycytidine (5adC). The experiment was initiated 24 h after cell seeding in 10 cm plates. The groups are naïve cells (N), cells exposed for 48 h to either 0.05% DMSO as vehicle (V), or to 5adC at 1 µM and 5 µM. DNA and RNA libraries were sequenced on an Illumina NextSeq 500. Gene expression was analysed by RNAseq and differentially methylated DNA regions (DMRs: clusters of 200 base pairs (bp), read depth >20, q< 0.05, methylation difference >|25%|) were identified by reduce representation bisulfite sequencing (RRBS). Global genome DNA methylation was similar between the N (mean±SD, 47.3%±0.02) and V groups (47.3%±0.01). Although 5adC reduced methylation, the reduction was larger in the 1 µM (39.2%±0.002) than in the 5 µM group (44.3%±0.01). 5adC induced a total of 612 and 190 DMRs by 1 µM and 5 µM, among which 79 and 23 were in the promoter regions (±3,000 bp from the transcription start site), respectively. 5adC induced a total of 1,170 and 1,797 differentially expressed genes (DEGs) by 1 µM and 5 µM, respectively. The 5 µM treatment induced statistically significant toxicity (% cell viability: group N 97%±8, V 98.8%±1.3, 1 µM 97.3%±0.5, 5 µM 93.8%±1.5), which perhaps reduced cell division and daughter cell numbers with inherited changes in methylation, but increased number of DEGs due to both toxicity and methylation changes. As usually observed in the literature, a small portion of DEGs (4% and 4% at 1 µM and 5 µM, respectively) are associated with DMRs in their promoters. These promoter DMRs by themselves are sufficient among other epigenetic marks to induce DEGs. The dataset provides the genomic coordinates of the DMRs and an opportunity to further examine their roles in distal putative promoters or enhancers (yet to be described in the SH) in contributing to gene expression changes, senescence bypass and sustained proliferation as essential carcinogenic events (see companion paper [1]). Finally, this experiment confirms the possibility in future experiments to use 5adC as a positive control for effects on DNA methylation in cells derived from SH.

Specifications Table

Subject	Biological Sciences. Genetics: Epigenetics
Specific subject area	DNA methylation is a DNA modification and an epigenetic mark that modify the chromatin structure and functions (transcription, replication, repair, stability).
Type of data	Gene expression (RNA-Seq) data, epigenetic (DNA methylation status) data from reduced representation bisulfite sequencing, described in three figures and one table. Four figures describing the sex determination assay (genomic location, melt curves, gel electrophoresis).
How the data were acquired	RNA-seq and RRBS using an Illumina NextSeq 500 sequencer. The Kdm5 DNA sequences were from the mouse genome (assembly GRCm380) using the primer design and search tool BiSearch [2]. The Kdm5c sequence homology between the Mus musculus and Mesocricetus auratus genomes was confirmed by nBLAST (NCBI). The alignments were created with Clustal Omega version 1.2.4. qPCR and melt curve analyses were performed on a CFX-96 thermocycler. Melt curve images were retrieved using Biorad CFX Maestro 1.1 version 4.1.2433.1219 (Biorad, Mississauga ON). The gel was imaged on a GelDoc XR imaging station (Biorad).
Data format	FASTQ (raw data for all samples), bedGraph (RRBS data), genes X counts table (RNA-seq data)
Description of data collection	The dataset was acquired from a cell culture experiment. The DNA and RNA were extracted and analysed by next generation sequencing on an Illumina NextSeq 500 sequencer. Differentially expressed genes (DEGs) were identified by RNA-seq, and differentially methylated DNA regions (DMRs) were identified by reduced representation bisulfite sequencing (RRBS). RNA-seq data was normalized by library size using DESeq2. RRBS data was analyzed using MethylKit.
Data source location	Hazard Identification Division. Environmental Health Science and Research Bureau. Environmental and Radiation Health Sciences Directorate. Healthy Environments and Consumer Safety Branch. Health Canada Sir Frederick G Banting Research Centre, AL:2203B 251 Sir Frederick Banting Driveway, Building 22 Ottawa, ON, Canada, K1A 0K9
Data accessibility	Repository name, data identification number, and direct URL to data: The nucleotide sequences of raw reads and processed data were submitted to NCBI's Gene Expression Omnibus and are available under the accession number GSE218911. All the next generation sequencing data (RNAseq and RRBS) for the main Toxicology paper and the current paper were deposited in a publicly open NCBI super series GEO submission GSE220238 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE220238). Meier MJ, Cummings-Lorbetskie C, Rowan-Carroll A, Desaulniers D: Dataset on DNA methylation and gene expression changes induced by 5-aza-2′-deoxycytidine in Syrian hamster fetal cell cultures. Mendeley Data 2022; https://data.mendeley.com/datasets/xfdjyd24hb.
Related research article	Desaulniers, D., Cummings-Lorbetskie, C., Leingartner, K., Meier, M. J., Pickles, J. C., and Yauk, C. L. DNA methylation changes from primary cultures through senescence-bypass in Syrian hamster fetal cells initially exposed to benzo[a]pyrene. Toxicology (2023) 10.1016/j.tox.2023.153451.

Value of the Data

• •Mesocricetus auratus is an important animal model in toxicology, carcinogenesis [1] and virology [3,4]. Its genome is poorly annotated and this RRBS dataset contributes to the description of the DNA methylome in this species.
• •Scientists working with SH in the field of epigenetics, carcinogenesis, toxicology, virology, and fetal development can benefit from these data originating from primary cultures of fetal cells.
• •The location and description of DMRs sensitive to 5adC associated with DEGs, can be used for further bioinformatic analyses to inform gene annotations and examine potential distal promoters and enhancer regulatory regions during carcinogenesis.
• •This dataset informs on the regulation of gene expression by DNA methylation. In combination with the data in the companion paper describing the chronology of DNA methylation changes from primary fetal cell cultures to beyond senescence-bypass, collectively these data may assist in the development of DNA methylation-based in vitro assay for the identification of carcinogens.
• •These data are useful confirming the possibility in future experiments to use 5adC as a positive control for effects on DNA methylation in cells derived from SH.
• •The regulation of imprinted genes and many cellular functions are sex dependent. To avoid data interpretation biased based on sex, the rapid qPCR method for sex determination included here can be used with any qPCR instruments and SH cell preparations derived from embryos, fetuses, or cell lines.

1. Objective

We tested the effects of the DNA methylation inhibitor 5adC to establish the gene list of those for which the expression could be altered by DNA methylation changes. This data was collected to compare with DMR bearing genes identified in the companion paper which does not include gene expression analyses. A method had to be developed to identify the sex of the cell preparations for the samples in the current and in the companion paper to avoid DMRs misinterpretation associated with the silencing of the X chromosome or imprinted genes.

2. Data Description

The raw sequencing output for each sample in this experiment for both RNA-seq and RRBS is reported in Table 1. The raw sequencing data in the form of FASTQ files, as well as the processed bedGraph data for RRBS and count matrix for RNA-seq, are available at the Gene Expression Omnibus under Series GSE218911. A summary of RNA-seq data is presented in Fig. 1, showing a PCA (Fig. 1A) in which samples cluster strongly based on 5adC exposure, and a heatmap showing the top 50 DEGs, where this experimental group-level clustering is also observed (Fig. 1B). Fig. 1C shows a strong gene expression response to 5adC treatment relative to vehicle controls.

Table 1.

Sample metadata of high-throughput sequencing for RNA-seq and RRBS.

Sample ID	Chemical Treatment	Dose 5adC (µM)	RNA-seq % Aligned1	RNA-seq Aligned2	RRBS % mCpG3	RRBS % aligned4	RRBS Aligned5
Naive_rep1	Naive	0	93.5	19.4	43.0%	61.5%	23.9
Naive_rep2	Naive	0	93.4	18.1	41.9%	61.5%	29.6
Naive_rep3	Naive	0	93.5	17.0	39.1%	61.8%	39.5
Naive_rep4	Naive	0	93.5	18.1	41.2%	60.8%	27.8
Vehicle_rep1	0.05% DMSO	0	93.2	21.5	41.8%	61.7%	35.5
Vehicle_rep2	0.05% DMSO	0	93.0	19.0	40.1%	61.7%	32.3
Vehicle_rep3	0.05% DMSO	0	89.8	24.2	42.0%	61.5%	27.0
Vehicle_rep4	0.05% DMSO	0	93.2	27.5	41.1%	61.5%	19.7
1_um_5adC_rep_1	5adC	1	92.7	16.4	34.1%	62.5%	22.3
1_um_5adC_rep_2	5adC	1	92.4	16.6	34.0%	62.1%	19.3
1_um_5adC_rep_3	5adC	1	92.6	17.6	34.4%	62.4%	30.1
1_um_5adC_rep_4	5adC	1	92.5	21.2	34.2%	62.6%	28.6
5_um_5adC_rep_1	5adC	5	91.2	26.6	39.1%	62.7%	27.7
5_um_5adC_rep_2	5adC	5	91.3	21.4	37.7%	62.6%	26.3
5_um_5adC_rep_3	5adC	5	91.6	19.4	38.1%	63.2%	32.1
5_um_5adC_rep_4	5adC	5	91.6	18.6	39.5%	62.9%	33.3

¹Percentage of reads successfully aligned to the reference genome from RNA-seq data.

²Number of reads in millions successfully aligned to the reference genome from RNA-seq data.

³Percentage of methylated CpG sites detected by RRBS per sample.

⁴Percentage of reads successfully aligned to the reference genome from RRBS data.

⁵Number of reads in millions successfully aligned to the reference genome from RRBS data.

Fig. 1.

(A) Heatmap, (B) PCA, and (C) volcano plot for RNA-seq data analysis for differentially expressed genes (DEGs) in SH cells exposed to two concentrations of 5adC.

The overlap between DEGs and DMRs is shown in Fig. 2 and available in tabular format on Mendeley Data [5]. A Venn diagram (Fig. 2A) shows that there are 122 genes in common that were both differentially expressed and differentially methylated. Fig. 2B shows that these were primarily demethylated regions that were up regulated in their gene expression.

Fig. 2.

(A) Overlap between RRBS and RNA-seq. (B) Top DEGs that were also differentially methylated.

Fig. 3 shows a pathway-level analysis in which the 122 genes in common were tested for enrichment against a background set of genes that were detected in both sequencing assays. Fig. 3A shows the most enriched pathways, while Figs. 3B and 3C show the extent of differential expression and methylation, respectively, for the set of genes that were found to be enriched.

Fig. 3.

Pathway analysis of DEGs that were also differentially methylated. (A) Pathways found to be enriched when considering the overlap between DMRs and DEGs relative to the background set of genes observed in both assays. Extent of changes in (B) expression and (C) methylation for the genes found in enriched pathways (shown on the x-axis, with their respective GO biological pathways on the y-axis).

The current qPCR sex determination assay was adapted from a mouse blastocysts sex determination assay [6]. The development of the qPCR assay specific to the SH was possible because of the sequence homology in the lysine (k) demethylase 5 (Kdm5) loci between the X and Y chromosome (Fig. 4), between species (97.5%), and in primer binding sites (100%) (Fig. 5). With just one primer pair, the qPCR assay simultaneously targets the genes of two isoforms of Kdm5, with the c (Kdm5c) and d (Kdm5d) loci located on the X and Y chromosome, respectively (Fig. 4). These enzymes demethylate di- or trimethylated histone 3 (H3) (H3K4me2/me3) but not the monomethylated H3K4me1 [7]. Amplification of both Kdm5c and d genes identifies a XY males, whereas amplification of the single Kdm5c gene identifies XX females (Fig. 4.). The amplicons from X- and Y-linked loci differ in size by 29 base pairs (331bp for Kdm5c and 302 for Kdm5d) and can be discriminated through melt curve analyses (Fig. 6) and/or by gel electrophoresis (Fig. 7).

Fig. 4.

The sequence of the X and Y linked amplicons were retrieved from the mouse genome (assembly GRCm380) using the Primer Design and Search tool BiSearch [2] with the kdmc/d primer pair [6] and default settings. The amplicons for the mouse X and Y chromosomes were aligned with Clustal Omega multiple sequence alignment tool version 1.2.4. (European Bioinformatics Institute, https://www.ebi.ac.uk/Tools/msa/clustalo/) *:matched nucleotides between the X and Y sequence. Bases matching the primer sequences are underlined.

Fig. 5.

Alignment of X amplicon sequences from the Mus musculus (Mus mus) the Mesocricetus auratus (Mes aur) genome. The Basic Local Alignment Search Tool (blastn, National Institute of Health) was used with default settings to search the Mes aur genome (BCM Maur 2.0 reference, Annotation Release 103) for alignments with the mouse X amplicon from Fig. 4. The blastn returned one result with coordinates NW_024429188.1:8317358-8317698 from Mes aur isolate SY011 unplaced genomic scaffold, BCM_Maur_2.0 Super-Scaffold_100003. In the aligned region displayed below (Clustal Omega), Mus mus and Mes aur genomes are 90.32% identical and match completely in the region where qPCR primers bind. *:matched nucleotides between the Mus mus and Mes aur. Kdm5c transcripts. Bases matching the primer sequences are underlined. Genome sequencing projects of Mes aur have not included the Y chromosome so an equivalent genome search could not be performed with the mouse Y amplicon.

Fig. 6.

Examples of melt curves of amplicons generated from SH DNA samples of 5 females (red) and 5 males (blue) analyzed in triplicate (15 curves per group). The shoulder peak at 81.5°C represents the smaller Y-chromosome amplicon while the larger peak at 85°C represents the larger X-chromosome amplicon. The smaller amplicon binds less Sybr green so it fluoresces at lower intensity relative to the X-chromosome amplicon. The flat black curve was generated by the “no template control” sample.

Fig. 7.

The amplicons generated using SH DNA were confirmed of the appropriate size and numbers by sex by electrophoresis on a 2% agarose gel stained with 1µg/mL ethidium bromide. Liver DNA from pregnant dams (lanes 1 to 5) generated amplicons from both X chromosomes resolved by one band, in contrast in male breeders (lanes 6 to 10) amplicons of different size are generated by the X and Y chromosome and are resolved by two bands. Lane 11 contained the PCR reaction from the “no template” control sample. The gel was imaged on a GelDoc XR imaging station (Biorad) with fluorescent trans-illumination.

3. Experimental Design, Materials and Methods

3.1. Cell Culture and Experimental Set-up

Primary cultures of SH fetal cells were exposed for 7 days to the carcinogen benzo[a]pyrene (20 µM) from which a morphologically transformed colony was collected and reseeded. The colony bypassed senescence and sustained growth (this clone was named as HC26d1, details in companion paper [1]). This clone was identified as a male clone using our methodology, this is relevant given that epigenetic marks can differ between sex. After 210 days of culture, the cells were collected and divided in 16 aliquots to create 4 experimental groups. The experiment was initiated 24 h after seeding aliquots in 10 cm plates. The groups are naïve cells (N), cells exposed for 48 h to either 0.05% DMSO as vehicle (V), or to the DNA methylation inhibitor 5adC at 1 and 5 µM. DNA and RNA libraries were sequenced on an Illumina NextSeq 500 and reduced representation bisulfite sequencing (RRBS) was used to identify DMRs (clusters of 200 base pairs (bp), read depth >20, q< 0.05, methylation difference >|25%|). DNA integrity was verified by 0.8% agarose electrophoresis. RRBS libraries were prepared from 100 ng genomic DNA using the Premium RRBS kit from Diagenode (Cat. # C02030032).

DNA isolation and preparation of libraries are described in the companion paper [1].

3.2. RNA Isolation and RNA-seq Library Preparation

Gene expression was analysed by RNA-seq. SHF cell pellets were thawed from frozen then homogenized on QIAshredder spin columns (Cat. # 79654, Qiagen, Mississauga, Canada) before total RNA was Isolated using the RNEasy mini kit (Cat. # 74104, Qiagen) using the manufacturer instructions that included the on-column genomic DNA digestion step. The RNA integrity numbers (RIN) of purified RNA samples exceeded 9.7 (except for the reference RNA with RIN 8.1) as determined using an Agilent 4200 Tapestation System (Mississauga, ON,Canada) with RNA screen tape (Cat. # 5067-5576). RNA-seq libraries were prepared using the Illumina TruSeq Stranded mRNA kit (Cat. # 20020594) according to the manufacturer's protocol starting with 1 μg total RNA. Briefly, the purified RNAs was fragmented and converted to cDNA with reverse transcriptase. The resulting cDNAs were converted to double stranded cDNAs and subjected to end-repair, A-tailing, and then ligation to dual-Indexed adapters (Illumina Cat. # 20019792). The constructed libraries were amplified using 15 cycles of PCR then quality controlled for size and purity using a DNA 1000 chip on a Bioanalyzer 2100 instrument (Agilent). The bioanalyzer trace was gated between 200 to 600bp to estimate library size. Libraries were quantified by qPCR on a CFX 96 machine (Biorad, Mississauga, ON) with Kapa polymerase, library quantification primers, and control oligos (Roche Diagnostics, Laval, QC) following Roche instructions. Libraries were then diluted to 4nM and pooled.

3.3. Sequencing

Pooled libraries were sequenced on an Illumina NextSeq 500 using a 75-cycle flow cell (high output kit v2.5). Following sequencing, FASTQ files were generated from the BCL files using bcl2fastq (v2.20.0.422).

3.4. Analysis of Transcriptomic Data

FASTQ files were trimmed with fastp v. 0.20.0 [8] with the parameters ‘cut_front –cut_front_window_size 1 –cut_front_mean_quality 3 –cut_tail –cut_tail_window_size 1 –cut_tail_mean_quality 3 –cut_right –cut_right_window_size 4 –cut_right_mean_quality 15 –length_required 36’ and then aligned to the Mesocricetus auratus genome (MesAur1.0, GCA_000349665.1, downloaded from ensembl.org) using STAR version v. 2.7.4a. Read counts were estimated using the rsem-calculate-expression function from RSEM v. 1.3.1 [9]. DESeq2 [10] v. 1.38.1 was used to model DEGs for each chemical treatment group versus the vehicle control group. We used ashr [11] to shrink log fold changes. Filtering of DEGs was based on a cutoff of 0.05 for the false discovery rate (FDR) adjusted Wald test p-value, and DEGs had a minimum linear fold change of 1.5. Cook's cutoff was not used for filtering; however, genes were filtered for false positives using the criteria in the Omics Data Analysis Framework for Regulatory Application (R-ODAF) [12]. Briefly, the R-ODAF criteria filters genes to include only those where 75% of at least one experimental group were above 1 CPM, and removes genes with spurious spikes in which (max - median) of counts were less than (sum of counts)/(number of replicates + 1).

3.5. Analysis of Bisulfite Sequencing Data

FASTQ files were trimmed using TrimGalore v. 0.5.0 using the parameters –illumina –rrbs (https://github.com/FelixKrueger/TrimGalore). Alignments of reads were generated using bismark [13] v. 0.22.1 and bowtie2 [14] v2-2.3.5.1. MethylKit count files were generated using R (https://www.R-project.org/) v. 3.6.1 and MethylKit v1.10.0 [15].

3.6. Statistical Analyses

Global genome DNA methylation and cell viability data were analysed with the software SigmaPlot (v13.0). The data passed the normality (Shapiro-Wilk) and equal variance (Brown-Forsythe) tests, and then a one-way ANOVA followed by the all pairwise Holm-Sidak test were conducted. P<0.05 identified statistically significant differences.

Overlap between DEGs and DMRs was determined using a custom R script [5].

3.7. QPCR Sex Determination Assay

Genomic DNA was isolated using the DNEasy blood and tissue kits (Qiagen, Mississauga, Ontario) following the supplied methods including the RNAse treatment step. qPCR was performed on a CFX-96 thermocycler (Biorad, Mississauga ON) using SSO Advanced universal SYBR Green 2x supermix (Bio-rad), 20-50ng template DNA, and 0.5µM primers with sequence CTGAAGCTTTTGGCTTTGAG and CCACTGCCAAATTCTTTGG [6] (Integrated DNA Technologies, Coralville, Iowa). The amplification protocol started with a 98°C hold for 3 minutes followed by 40 cycles of melting/annealing/extension at 98°C for 10 seconds, 55°C for 30 seconds, and then 72°C for 30 seconds. Melt curve analysis immediately following PCR with sequential 5-second holds at temperatures incremented by 0.5°C from 65°C to 95°C. Melt curves identified samples as female or male origin by displaying single or double peaks respectively, which were confirmed by electrophoresis through 2% agarose gel (Gibco, Toronto, Ontario) with 1µg/mL ethidium bromide (Sigma, Toronto, Ontario).

The Y amplicon appears as a shoulder peak (Fig. 6). To generate larger Y amplicon peaks, an alternative reverse primer was designed to include 2 additional Y-specific bases at the 3’ end (sequence CCACTGCCAAATTCTTTGGAG). While the redesigned reverse primer amplified larger Y-chromosome peaks, the peak heights were more variable, and this assay based on the alternative primer was abandoned.

Ethics Statement

The samples were produced from SH through the work described in the companion paper [1], which include the statement that animal procedures and housing conditions were approved by the Animal Care Committee of Health Canada (approval number HCO-ACC 2017-014) and conformed to the Guidelines of the Canadian Council on Animal Care.

CRediT authorship contribution statement

Matthew J. Meier: Writing – original draft, Methodology, Software, Formal analysis, Data curation, Visualization, Writing – review & editing. Cathy Cummings-Lorbetskie: Writing – original draft, Methodology, Investigation, Resources, Data curation, Writing – review & editing. Andrea Rowan-Carroll: Writing – original draft, Methodology, Investigation, Resources, Data curation, Writing – review & editing. Daniel Desaulniers: Writing – original draft, Methodology, Investigation, Resources, Data curation, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data Availability

• Dataset on DNA methylation and gene expression changes induced by 5-aza-2'-deoxycytidine in Syrian hamster fetal cell cultures. (Original data) (Mendeley Data).