Blood DNA methylation and COVID-19 outcomes

Abstract

Background: There are no prior reports that compare differentially methylated regions of DNA in blood samples from COVID-19 patients to samples collected before the SARS-CoV-2 pandemic using a shared epigenotyping platform. We performed a genome-wide analysis of circulating blood DNA CpG methylation using the Infinium Human MethylationEPIC BeadChip on 124 blood samples from hospitalized COVID-19-positive and COVID-19-negative patients and compared these data with previously reported data from 39 healthy individuals collected before the pandemic. Prospective outcome measures such as COVID-19-GRAM risk-score and mortality were combined with methylation data.

Results: Global mean methylation levels did not differ between COVID-19 patients and healthy pre-pandemic controls. About 75% of acute illness-associated differentially methylated regions were located near gene promoter regions and were hypo-methylated in comparison with healthy pre-pandemic controls. Gene ontology analyses revealed terms associated with the immune response to viral infections and leukocyte activation; and disease ontology analyses revealed a predominance of autoimmune disorders. Among COVID-19-positive patients, worse outcomes were associated with a prevailing hyper-methylated status. Recursive feature elimination identified 77 differentially methylated positions predictive of COVID-19 severity measured by the GRAM-risk score.

Conclusion: Our data contribute to the awareness that DNA methylation may influence the expression of genes that regulate COVID-19 progression and represent a targetable process in that setting.

Keywords: Acute respiratory distress syndrome (ARDS); COVID-19; DNA methylation; Gene expression; Mortality; Outcomes.

Fig. 1

Diagram of the entire cohort involved in study: Notice that while the hospitalized patients’ cohort contributed 128 patients, only 124 were part of the analyses due to inadequate quality of 4 samples; see diagram and details in the text

Fig. 2

Differential SARS-CoV-2 DNA methylation between blood samples from patients on hospital admission for COVID-19 compared to blood samples from healthy controls before the COVID-19 pandemic. A A box and whisker plot depicts the difference in mean global methylation level (y-axis) between COVID-19 patients and healthy controls (x-axis). Each black dot represents the mean methylation level of each participant. These results indicate that global mean methylation levels do not distinguish COVID-19 patients from healthy pre-pandemic controls. B A Manhattan plot of DNA methylation regions shows the distribution of SARS-CoV-2-associated significantly differentially methylated regions (DMRs) across the genome by chromosome number. Hyper-methylated regions are displayed with a positive log10 (p value), and hypo-methylated regions are displayed with a negative log10 (p value). DMRs were ascertained as regions having at least 5 consecutive CpGs where > 75% of the CpGs in the region had an FDR p value < 0.05, and all were either hyper-methylated or hypo-methylated. This approach identified 1505 DMRs, that are displayed above and below the blue lines. Dots alternate colors to depict a change in chromosome. Sex chromosomes were excluded from analysis. These results indicate that 1505 DNA regions are differentially methylated within days of SARS-CoV-2 infection. C A pie chart showing the percent distribution of DMRs to standard genomic features. 5′UTR = 5′ untranslated region 3′UTR = 3′ untranslated region. In keeping with the known role of DNA methylation in regulation of gene expression, a preponderance of DMRs are in gene promoter regions. D Bar graphs of the top ten gene ontological (GO) biological processes related to the COVID-19 differentially methylated genes, ordered by statistical significance. The X-axis indicates the number of COVID-19 DMR-associated genes that contribute to each GO term. Bar color indicates the FDR P-value using a Fischer test. These results indicate that the observed DMRs occur in genes that participate in leukocyte activation and immune responses. E Bar Graph of the top 10 disease ontological (DO) processes related to the COVID-19-associated differentially methylated genes, ordered by statistical significance. The X-axis indicates the number of COVID-19 DMR-associated genes contributing to each GO term. Bar color indicates the FDR P-value using a Fischer test. These results indicate that the observed DMRs occur in genes that participate in the pathogenesis of inflammatory and leukocyte disorders

Fig. 3

DMRs in blood samples from COVID-19 patients on hospital admission are distinct from patients with non-COVID-19 respiratory illness in genes that participate in virus-related pathways and disorders. A Box and whisker plot depicts the difference in mean global methylation level (y-axis) between COVID-19 and non-COVID-19 respiratory ill patients (1 and 0, respectively; x-axis). Each black dot represents the mean methylation level of each participant. These results indicate that global mean methylation levels do not distinguish COVID-19 from non-COVID-19 respiratory ill patients. B Circos plot depicts genomic distribution of differentially methylated regions (DMRs) across the human genome. (Outer ring) Each chromosome is shown as a different color. The relative chromosome size is represented by the arc bar length. (Inner rings) Hyper-methylated DMRs are shown in red and hypo-methylated regions are shown in blue. Sex chromosomes were omitted from the analysis. These results indicate that 254 DNA differentially methylated regions distinguish SARS-Cov-2 infection from non-COVID-19 respiratory illness. C Bar Graph of the top ten disease ontological (DO) biological processes related to the SARS-CoV-2-associted differentially methylated genes, ordered by statistical significance. The X-axis indicates the number of SARS-CoV-2 DMR-associated genes that contribute to each DO term. Bar color indicates the FDR p value using a Fischer test. These results indicate that the observed DMRs occur in genes that participate in inflammatory and host-defense processes. D Bar Graph of the top ten gene ontological (GO) processes related to the SARS-CoV-2-associated differentially methylated genes, ordered by statistical significance. The X-axis indicates the number of SARS-CoV-2 DMR-associated genes that contribute to each GO term. BAR color indicates the FDR P-value by using a Fischer test. These results indicate that the observed DMRs occur in genes that participate in the pathology of influenza, other viral infections and inflammatory disorders

Fig. 4

Overlap of COVID-19 DMR-associated genes in blood. A Venn diagram of the overlap of COVID-19 DMR-associated genes identified by comparison of DMRs between COVID-19 patients and healthy pre-pandemic controls, and DMRs between COVID-19 and non-COVID-19 respiratory illness patients on admission. Asterisks indicate overlap that is significant at p value < 0.001. Twenty-five of the 47 overlapping genes with DMRs encode proteins that participate in leukocyte viral defense, inflammation and immune responses. B Ontology analysis of the 47 overlapping genes with DMRs indicate a role in viral defense mechanisms. C Relative positions of COVID-19-associated DMRs in the promoter region of OAS2 (C-1) and IFI27 (C-2) with a schematic depicted for each gene. The relative positions of probes measuring methylation levels of CpG sites annotated to each gene with their genomic 5′-3′ positions are provided (inset panel; x-axis) versus the -log10 of the p value (y-axis). The p value < 0.05 is displayed as a black dashed line. Probes residing in a COVID-19-associated DMR are shown as hypo-methylation (blue dots) and hyper-methylation (red dots). Probes not meeting a p value < 0.05 at the individual CpG level are shown as hollow dots. These results indicate that the DMRs comprise a cluster of differentially methylated positions within days of SARS-CoV-2 infection

Fig. 5

Transcriptional expression of prototypical interferon stimulated genes (ISGs) -IFI27 and OAS2- correlates with methylation status of their gene promoter regions. RNA from circulating leukocytes obtained from the same COVID-19 positive and negative patients presented in Fig. 4 was used to interrogate expression level of two ISGs. A OAS2 and B IFI27 expression levels are significantly higher in hospitalized patients with COVID-19, which correlates with their gene promoter regions predominant hypomethylation. GAPDH was used as a reference gene; see methods for details. **; p value < 0.01

Fig. 6

DNA methylation is associated with COVID-19 outcomes. A Volcano plot shows genes associated with dichotomized GRAM-risk scores, either hyper-methylated (red) or hypo-methylated (blue). B DNA methylation levels at 77 differentially methylated positions (DMPs) correlate with disease severity in COVID-19 patients. DMRs (N = 19) associated with the GRAM-score were identified in COVID-19 patients (N = 100). DMRs were ascertained as regions with at least 3 consecutive CpGs where > 75% of the CpGs in the region had a FDR p value < 0.05 and all were either hyper-methylated or hypo-methylated. DNA methylation levels of the DMPs (N = 145) residing in the DMRs were subjected to recursive feature elimination to identify CpGs that best distinguish GRAM-score risk. Shown is a hierarchical cluster using the DNA methylation data from the 77 DMPS (see Additional file 1: Table S8), that are shown as a heatmap of the M-values. Low GRAM-score risk (gray) and high GRAM-score risk (black) are indicated. These results indicate that DNA methylation levels at these 77 DMPs may be useful as biomarkers of the severity of COVID-19 patients. (see Additional file 1: Table S6-1 and S6-2)