Single-cell multiomics reveal divergent effects of DNMT3A- and TET2-mutant clonal hematopoiesis in inflammatory response

Abstract

DNMT3A and TET2 are epigenetic regulator genes commonly mutated in age-related clonal hematopoiesis (CH). Despite having opposed epigenetic functions, these mutations are associated with increased all-cause mortality and a low risk for progression to hematologic neoplasms. Although individual impacts on the epigenome have been described using different model systems, the phenotypic complexity in humans remains to be elucidated. Here, we make use of a natural inflammatory response occurring during coronavirus disease 2019 (COVID-19), to understand the association of these mutations with inflammatory morbidity (acute respiratory distress syndrome [ARDS]) and mortality. We demonstrate the age-independent, negative impact of DNMT3A mutant (DNMT3Amt) CH on COVID-19-related ARDS and mortality. Using single-cell proteogenomics we show that DNMT3A mutations involve myeloid and lymphoid lineage cells. Using single-cell multiomics sequencing, we identify cell-specific gene expression changes associated with DNMT3A mutations, along with significant epigenomic deregulation affecting enhancer accessibility, resulting in overexpression of interleukin-32 (IL-32), a proinflammatory cytokine that can result in inflammasome activation in monocytes and macrophages. Finally, we show with single-cell resolution that the loss of function of DNMT3A is directly associated with increased chromatin accessibility in mutant cells. Hence, we demonstrate the negative prognostic impact of DNMT3Amt CH on COVID-19-related ARDS and mortality. DNMT3Amt CH in the context of COVID-19, was associated with inflammatory transcriptional priming, resulting in overexpression of IL32. This overexpression was secondary to increased chromatic accessibility, specific to DNMT3Amt CH cells. DNMT3Amt CH can thus serve as a potential biomarker for adverse outcomes in COVID-19.

Graphical abstract

Figure 1.

Clinical features, prevalence of CH, and disease-related outcomes of 243 patients hospitalized with COVID-19 in the prevaccination era. (A) Heat map showing the spectrum of CH mutations, sex distribution, COVID-19–related complications, prevalence of CRS, and serum cytokines and inflammatory markers. (B) Bar plots comparing the prevalence of ARDS among patients with COVID-19 with TET2mt CH and DNMT3Amt CH. ARDS exclusively occurred in patients with COVID-19 with underlying DNMT3Amt CH but not TET2mt CH (Mann-Whitney U test, P = .007). (C) Box plots comparing the serum MCP-1 concentrations among patients with COVID-19 with TET2mt CH and DNMT3Amt CH at the time of hospitalization. There was an increase in serum MCP-1 concentration in patients with COVID-19 with underlying DNMT3Amt CH compared with those with TET2mt CH (Mann-Whitney U test, P = .014). (D) Kaplan-Meier plot showing the OS estimates for 243 patients with COVID-19, stratified by CH status. There was increased all-cause mortality among patients with COVID-19 with underlying CH (log-rank test, P < .001). (E) Kaplan-Meier plot showing the OS estimates for 218 patients with COVID-19, stratified by CH status (further stratified into TET2mt CH and DNMT3Amt CH). The increased all-cause mortality among patients with COVID-19 with underlying CH was mainly driven by DNMT3Amt CH (log-rank test, P < .001). This association remained consistent after adjusting for age at COVID-19 diagnosis: Hazard ratio (HR), 2.84 (95% CI, 1.16-6.94; P = .022). (F) Forest plot showing the HRs calculated using Cox models adjusted for age, sex, and various comorbidities. We observed significant increase in all-cause mortality in patients with COVID-19 with DNMT3Amt CH compared with those without CH mutations, even after adjusting for comorbidities. AKI, acute kidney injury; ALI, acute lung injury; CLOT, venous thromboembolism; CI, confidence interval; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; IMV, invasive mechanical ventilation; MODS, multiple organ dysfunction syndrome; NIV, noninvasive ventilation.

Figure 2.

DNA methylation changes across the genome in patients with TET2mt CH and DNMT3Amt CH. (A) Overview of patient cohorts from whom PBMCs were collected and analyzed in this study using DNA methylation (Illumina Methylation EPIC array), single-cell proteogenomics (Tapestri assay), scRNA-seq (10x Genomics), multiome (10x Genomics), and GoTChA modalities. Mutations and VAFs are shown above the figurines. Patient ages and modalities used are shown below the figurines. Sample identifier is shown inside the figurines. (B) Box plot showing the comparison of global DNA methylation between DNMT3Amt and TET2mt CH. There was no significant difference by Wilcoxon signed-rank test (P = .057). (C) Density plot demonstrating DNA methylation differences between TET2mt CH and DNMT3Amt CH, primarily affecting highly methylated CpGs (β > 0.75; Kolmogorov-Smirnov test, P < 2.2 × 10⁻¹⁶). (D) Circos plot showing the number, genomic location, and density of differentially methylated regions between TET2mt CH and DNMT3Amt CH. There was an increased number of hypomethylated sites in DNMT3Amt CH compared with TET2mt CH. (E) Functional annotation of the differentially methylated regions (with Δβ > 10% and P < .010) using the ENCODE Epigenomics Roadmap PBMC reference data. Hypermethylation of Enh and promoters (TssA, TssAFlnk) was more commonly observed in DNMT3Amt CH (compared with TET2mt CH), whereas the hypomethylation observed in DNMT3A CH was predominately found at actively transcribed states (Tx, TxWk). scATAC, single-cell ATAC; TssA, active transcription start site; TssAFlnk, Flanking bivalent transcription start sites.

Figure 3.

Identification of cell types carrying CH mutations by combined single-cell surface protein and genotype analysis. (A) Overview of COVID-19 cohorts with TET2mt CH, DNMT3Amt CH, TET2mt DNMT3Amt (comutant) CH, and without CH (CH⁻), analyzed using single-cell proteogenomics (Tapestri assay). Mutations and VAFs are shown above, and patient ages are shown below, the figurines. Sample identifier is shown inside the figurines. (B) Uniform manifold approximation and projection (UMAP) projections showing the distribution of 28 941 cells from single-cell proteogenomics analysis from 13 patient samples, colored by cell types. The bar below shows the proportion of cells in each cell type. (C) UMAP projections of the single-cell proteogenomics data showing only mutated cells, which are then stratified by TET2 and DNMT3A mutated cells, demonstrating the myeloid and lymphoid lineage restriction in TET2mt and DNMT3Amt cells, respectively. The bars below show the proportion of cells in each cell type. (D-E) Bar plots showing the proportion of cells in each cell type stratifying cells by sample/patient genotype (D) and by cell genotype (E). Although TET2 mutations had a clear myeloid lineage restriction bias, DNMT3A mutations were seen in myeloid and lymphoid lineages, respectively. cDC, classical dendritic cells; gdT, γδ T cells; Int, intermediate; MAIT, mucosal-associated invariant T cells; mono, monocytes; pDC, plasmacytoid dendritic cells; Treg, regulatory T cells; WT, wild type.

Figure 4.

Identification of biomarkers of COVID-19 severity associated with CH mutations. (A) Overview of COVID-19 cohorts without CH (CH⁻), TET2mt CH, and DNMT3Amt CH, analyzed using scRNA-seq. Mutations and VAFs are shown above, and patient ages are shown below, the figurines. Sample identifier is shown inside the figurines. (B) UMAP projections showing the distribution of 78 083 cells from scRNA-seq analysis from 24 patient samples, colored by cell types identified using SingleR. The bar below shows the proportion of cells in each cell type. (C) Proportion of cells in each cell type stratified by 5 conditions as shown in y-axis, in the scRNA-seq analysis. The healthy cohort (from Stephenson et al⁴⁰) is further stratified by age: <50 and >50 years, respectively. (D) Volcano plot showing significantly differentially expressed genes (adjusted P < .05, Wilcoxon rank sum test) in comparisons between TET2mt CH and DNMT3Amt CH in the context of COVID-19. Cells from each cell type were tested independently. Bars below the volcano plots show the proportion of genes per cell type that are downregulated and upregulated in the DNMT3Amt CH in each comparison. (E) Violin plots showing expression of IL-32 in cell types in which, patients with DNMT3Amt CH had significantly higher expression of IL-32 in comparison to those with TET2mt CH. Black dots show mean expression. ∗∗∗∗P ≤ .0001 (Wilcoxon rank sum test). (F) Survival analysis separating patients showing IL-32 high and low (upper tertile and lower quartiles respectively) in protein level profiles using the Olink assay. Inset plot shows the IL-32 Olink measurements reported as normalized protein expression (NPX) values, with a 1-unit increase equating to a doubling of the protein concentration. This analysis shows that patients with higher levels of IL-32 in their serum had worse OS. cDC, classical dendritic cells; gdT, γδ T cells; Int, intermediate; MAIT, mucosal-associated invariant T cells; Max, maximum; Min, minimum; Mono, monocytes; pDC, plasmacytoid dendritic cells; SD, standard deviation; Treg, regulatory T cells.

Figure 5.

Characterization of epigenetic deregulation in patients with CH with DNMT3A mutations. (A) Proportion of each cell type identified in the scRNA-seq data from the 10x multiome platform stratified by 5 conditions as shown in the y-axis. The healthy cohort is further stratified by age: <50 and >50 years, respectively. (B) Violin plots showing cell-type–specific changes in chromatin accessibility measured as the total number of cut sites (sum of TF-IDF–normalized cut site counts across all peaks; scATAC-seq data from multiome) in each cell type. Only cell types with >100 cells are shown. Black dots show the mean value. Not significant (ns), P > .05; ∗P ≤ .05; ∗∗P ≤ .01; ∗∗∗P ≤ .001; ∗∗∗∗P ≤ .0001 (Wilcoxon rank sum test). NK and CD4 T cells showed significant increase in chromatin accessibility in patients with DNMT3Amt CH compared with TET2mt CH in the context of COVID-19. (C) Coverage plot showing epigenomic dysregulation of IL-32 in DNMT3Amt CH compared to TET2mt CH, through multiomics analysis. The plot shows the coaccessible peaks with IL-32 transcription start site (TSS) in patients with TET2mt CH and those with DNMT3Amt CH, both with COVID-19 (coaccessibility score > 0.1; blue and red arcs), the chromatin accessibility signal per group of cells, IL-32 gene expression (violin plots; right), candidate cis-regulatory elements predicted by ENCODE (colored-coded bars), open chromatin peaks (gray bars), differentially accessible peaks that are more accessible in patients with DNMT3Amt CH than those with TET2mt CH in CD4 T cells (light blue bars), CpG sites hypomethylated in patients with DNMT3Amt CH compared with patients with TET2mt CH (dark blue bars), and CpG sites overlapping open chromatin regions (black bars) around the IL-32 gene locus. Labeled loci A and B are chr16:3123999-3124965 and chr16:3263558-3264913, respectively. These loci are regions in which patients with DNMT3Amt CH gained accessibility in CD4 T cells, overlapped with hypomethylated CpG sites and gained coaccessibility with IL-32 TSS. (D) Box plots showing methylation levels (β values) per patient in TET2mt CH and DNMT3Amt CH cohorts (both with COVID-19) at 3 hypomethylated CpG sites shown in panel C. The middle line represents the median; the lower and upper edges of the rectangle represent the first and third quartiles, respectively; and the lower and upper whiskers represent the interquartile range × 1.5. The groups were compared using Wilcoxon rank sum test. (E-F) Violin plots showing significant increase in chromatin accessibility in DNMT3Amt cells compared with DNMT3Awild-type cells as determined by GoTChA analysis. The data shown are the total number of cut sites (E) and the number of cut sites at loci A and B from panel C (F), in DNMT3A wild-type and DNMT3Amt cells from 2 samples (DNMT3Amt clonal cytopenias of undetermined significance) profiled using GoTChA. Red dots show the mean value. Mutation site in the DNMT3A gene is shown in the bottom of panel E. ns, P > .05; ∗P ≤ .05; ∗∗P ≤ .01; ∗∗∗P ≤ .001; ∗∗∗∗P ≤ .0001 (Wilcoxon rank sum test). cDC, classical dendritic cells; CTCF, CCCTC-binding factor; Dist., distal; gdT, γδ T cells; Int, intermediate; MAIT, mucosal-associated invariant T cells; Mono, monocytes; pDC, plasmacytoid dendritic cells; Prox., proximal; Treg, regulatory T cells.

Figure 5.

Characterization of epigenetic deregulation in patients with CH with DNMT3A mutations. (A) Proportion of each cell type identified in the scRNA-seq data from the 10x multiome platform stratified by 5 conditions as shown in the y-axis. The healthy cohort is further stratified by age: <50 and >50 years, respectively. (B) Violin plots showing cell-type–specific changes in chromatin accessibility measured as the total number of cut sites (sum of TF-IDF–normalized cut site counts across all peaks; scATAC-seq data from multiome) in each cell type. Only cell types with >100 cells are shown. Black dots show the mean value. Not significant (ns), P > .05; ∗P ≤ .05; ∗∗P ≤ .01; ∗∗∗P ≤ .001; ∗∗∗∗P ≤ .0001 (Wilcoxon rank sum test). NK and CD4 T cells showed significant increase in chromatin accessibility in patients with DNMT3Amt CH compared with TET2mt CH in the context of COVID-19. (C) Coverage plot showing epigenomic dysregulation of IL-32 in DNMT3Amt CH compared to TET2mt CH, through multiomics analysis. The plot shows the coaccessible peaks with IL-32 transcription start site (TSS) in patients with TET2mt CH and those with DNMT3Amt CH, both with COVID-19 (coaccessibility score > 0.1; blue and red arcs), the chromatin accessibility signal per group of cells, IL-32 gene expression (violin plots; right), candidate cis-regulatory elements predicted by ENCODE (colored-coded bars), open chromatin peaks (gray bars), differentially accessible peaks that are more accessible in patients with DNMT3Amt CH than those with TET2mt CH in CD4 T cells (light blue bars), CpG sites hypomethylated in patients with DNMT3Amt CH compared with patients with TET2mt CH (dark blue bars), and CpG sites overlapping open chromatin regions (black bars) around the IL-32 gene locus. Labeled loci A and B are chr16:3123999-3124965 and chr16:3263558-3264913, respectively. These loci are regions in which patients with DNMT3Amt CH gained accessibility in CD4 T cells, overlapped with hypomethylated CpG sites and gained coaccessibility with IL-32 TSS. (D) Box plots showing methylation levels (β values) per patient in TET2mt CH and DNMT3Amt CH cohorts (both with COVID-19) at 3 hypomethylated CpG sites shown in panel C. The middle line represents the median; the lower and upper edges of the rectangle represent the first and third quartiles, respectively; and the lower and upper whiskers represent the interquartile range × 1.5. The groups were compared using Wilcoxon rank sum test. (E-F) Violin plots showing significant increase in chromatin accessibility in DNMT3Amt cells compared with DNMT3Awild-type cells as determined by GoTChA analysis. The data shown are the total number of cut sites (E) and the number of cut sites at loci A and B from panel C (F), in DNMT3A wild-type and DNMT3Amt cells from 2 samples (DNMT3Amt clonal cytopenias of undetermined significance) profiled using GoTChA. Red dots show the mean value. Mutation site in the DNMT3A gene is shown in the bottom of panel E. ns, P > .05; ∗P ≤ .05; ∗∗P ≤ .01; ∗∗∗P ≤ .001; ∗∗∗∗P ≤ .0001 (Wilcoxon rank sum test). cDC, classical dendritic cells; CTCF, CCCTC-binding factor; Dist., distal; gdT, γδ T cells; Int, intermediate; MAIT, mucosal-associated invariant T cells; Mono, monocytes; pDC, plasmacytoid dendritic cells; Prox., proximal; Treg, regulatory T cells.