Table 1.
Demographic information and SF-36 results for all ME/CFS cases (n = 33) and controls (n = 31) that participated in DNA methylation analysis and validation.
*—p<0.05, Student’s t-test, ME/CFS cases versus controls subjects. Data are shown as mean ± standard error of mean.
Fig 1.
Genomic DNA was isolated from ME/CFS cases (N = 13) and controls (N = 12) and analyzed using Illumina Infinium Human MethylationEPIC BeadChip microarrays and this was defined as “experimental cohort”. Following data analysis results were validated on all the samples from the “experimental”, as well as from a secondary cohort that included samples from ME/CFS cases (N = 8) and controls (N = 8) from Miami/Fort Lauderdale area, as well as ME/CFS cases (N = 12) and controls (N = 11) from Valencia (Spain), using pyrosequencing analysis.
Fig 2.
Principal component analysis of whole DNA methylomes allows differentiating between ME/CFS cases (N = 13, red circles) and matched controls (N = 12, blue circles).
Fig 3.
Unsupervised hierarchical clustering of 500 most variable CpG sites derived from samples distinguishes ME/CFS (N = 13) cases and controls (N = 12).
500 DMS above and below mean level are represented by blue and yellow across all cases and were used to cluster cases. As shown above, 12 of 13 ME/CFS cases (blue on color-coded bar above the dendrogram) cluster together (left dendrogram branch) and 12 of the 12 controls (blue color-coded bar) cluster together (right dendrogram branch), resulting in a divergence of these sub-phenotypes.
Fig 4.
Distribution of differentially methylated sites according to genic location and to CpG island relation.
(A) Distribution of DMS according to genic location. Relative percentages of DMS (both hypo-methylated and hyper-methylated CpG sites between ME/CFS cases and controls) are reported for each genic body location: gene regulatory elements (regulatory: TSS1500, TSS200, 5’ UTR, 3’ UTR), the coding regions of genes (gene body), as well as all Intergenic and 1st Exon gene body regions. (B) Distribution of DMS according to their location in relation to CpG islands. Relative percentages of DMS (both hypo-methylated and hyper-methylated) are reported. CpG islands with clustered CpG sites; 2 kb upstream and downstream of CpG islands (N, S Shores); 2 kb upstream and downstream of CpG shores (N, S Shelves).
Fig 5.
Distribution of differentially methylated promoter regions in ME/CFS according to gene biotype.
In RnBeads package, promoters are defined as regions 1500 bp upstream of TSS and 500 bp downstream of TSS. Criteria for DMPs were: FDR ≤ 0.1 and mean difference > 0.05. The distribution of these DMPs according to the gene biotype: protein-coding genes and regulatory RNA genes including short non-coding RNAs (miRNAs, small nuclear RNAs, small nucleolar RNAs), long non-coding RNAs (including antisense RNAs and intergenic RNAs) and pseudogenes.
Fig 6.
Canonical pathway analysis of genes with the DMPs of PBMCs isolated from ME/CFS cases.
Genes listed to the right belong to the respective pathways. Gene Ontology analysis was performed using the Ingenuity Pathway Analysis (IPA) on 307 genes with the DMPs.
Fig 7.
Bisulfite-conversion followed by pyrosequencing (PS) validation of DMPs identified by Illumina MethylationEPIC microarrays (EPIC).
Average methylation level of CpG sites within promoters of TABPB and ZBTB18 were assessed using pyrosequencing. Corresponding probes for TABPB: CpG1—cg04415168 (TSS1500), CpG2—cg10376053 (TSS1500), CpG3—cg14288848 (TSS1500), CpG4—cg17055704 (TSS1500), CpG5—cg14473643 (TSS1500), CpG6—cg14309283 (TSS1500). Corresponding probes for ZBTB18: CpG1—cg16399365 (TSS1500), CpG2—cg15896892 (TSS1500), CpG6—cg19698993 (TSS1500). CpG3, CpG4 and CpG5 are not printed on the Illumina MethylationEPIC microarrays and are located in ZBTB18 promoter between CpG2 and CpG6. * = FDR ≤ 0.05 for EPIC; * = p ≤ 0.05 for PS. Error bars represent the standard error of the mean.