Impact of constitutional TET2 haploinsufficiency on molecular and clinical phenotype in humans

Abstract

Clonal hematopoiesis driven by somatic heterozygous TET2 loss is linked to malignant degeneration via consequent aberrant DNA methylation, and possibly to cardiovascular disease via increased cytokine and chemokine expression as reported in mice. Here, we discover a germline TET2 mutation in a lymphoma family. We observe neither unusual predisposition to atherosclerosis nor abnormal pro-inflammatory cytokine or chemokine expression. The latter finding is confirmed in cells from three additional unrelated TET2 germline mutation carriers. The TET2 defect elevates blood DNA methylation levels, especially at active enhancers and cell-type specific regulatory regions with binding sequences of master transcription factors involved in hematopoiesis. The regions display reduced methylation relative to all open chromatin regions in four DNMT3A germline mutation carriers, potentially due to TET2-mediated oxidation. Our findings provide insight into the interplay between epigenetic modulators and transcription factor activity in hematological neoplasia, but do not confirm the putative role of TET2 in atherosclerosis.

Fig. 1

TET2 c.4500delA causes loss of the protein. a Pedigree (details modified for confidentiality) showing TET2 c.4500delA (TET2delA) statuses; mutation carriers are marked with +/− and wild-type individuals with +/+. Solid square (male) or circle (female) depicts patients with lymphoma. b Electropherograms of TET2delA-mutated and wild-type sample. c (Upper section) Somatic frameshifts or pathogenic substitutions along TET2 in hematopoietic and lymphoid tissue malignancies from COSMIC database (release v81). (Lower section) Protein domain structure of TET2 from Hu et al.. The C-terminal catalytic domain of TET2 comprises Cys-rich N-terminal (Cys-N) and C-terminal (Cys-C) subdomains, and a double-stranded β helix (DSBH) domain. Locations of germline frameshift mutations identified in this study and previously (Schaub et al.) are shown. d–f Cartoon representation of the catalytic domain of human TET2 complexed with methylated DNA (PDB entry 4nm6). Overall structure of the wild-type protein (d). DNA is colored with blue, flanking Cys-rich domains are colored with light and dark orange, the DSBH core of the protein is colored with cyan and pink, the latter depicting the sequence lost due to the TET2delA mutation. The predicted loss includes residues involved in formation of the DSBH domain such as H1912 involved in coordination of Zn3-ion (e), as well as residues R1896 and S1898 critical for binding the substrate N-oxalyglycine (NOG) (f), H1881 involved in coordination of Fe-ion (f), and Y1902 and H1904 interacting with 5-methylcytosine. Proper folding of the mutant protein lacking these critical amino acids appears highly unlikely based on the structure. N-oxalyglycine, 5-methylcytosine, and residues mentioned above are shown with stick representation. An iron and three zinc ions are shown as brown and gray balls, respectively. g Representative western blot of TET2 from lymphoblastoid cells of three TET2delA mutation carriers and three wild-type individuals of the family. The graph shows the quantification of TET2 protein normalized to the vinculin. Bars represent the mean ± s.d. of TET2 intensities

Fig. 2

Heterozygous TET2 loss causes increased DNA methylation. a Count of hypermethylated (dark red bars) and hypomethylated (light gray bars) cytosines per million tested CpGs in TET2delA carriers (on the left side) and wild-type controls (on the right side). Mutation carriers display elevated hypermethylation levels in blood as compared to non-carriers. Each sample was compared to the set of five baseline control samples in order to derive counts of hyper- and hypomethylated CpGs. Methylation profiling of the patients Ly1 and Ly2 was performed from two separate DNA samples extracted in years 2007 and 2009, others were sampled once. b–f Density plots of differentially methylated CpGs from whole-genome bisulfite sequencing (WGBS). b–d Pairwise comparison of each TET2delA carrier and age-matched wild-type family member. e Group-wise comparison of three TET2delA carriers to three age-matched non-carriers of the family. f Pairwise comparison of TET2X carrier and age-matched wild-type individuals from the Northern Finland Intellectual Disability Cohort. In b–f, X- and Y-axis represent the methylation fraction in TET2 mutation carriers and non-carriers, respectively

Fig. 3

Hypermethylation caused by TET2 loss is enriched at active enhancer regions. a, b Regions with different chromatin marks in 24 different subtypes of primary human blood cells (each subtype represented by a dot) were available from the Roadmap Epigenomics project. These were compared to the hyper- and hypomethylated sites detected in blood whole-genome bisulfite sequencing (WGBS) of three cancer-free TET2delA carriers relative to three age-matched non-carriers. a The hypermethylated sites in TET2delA carriers were characterized by frequent presence of normally H3K4me1-marked chromatin in primary blood cells. In addition, hypermethylation was enriched strongly at DNaseI hypersensitive and H3K27ac-marked chromatin, suggesting that methylation was increased in enhancer regions that are normally active. b The hypomethylated sites in TET2delA carriers were enriched at normally H3K9me3-marked regions that typically represent transcriptionally repressed heterochromatin. Odds ratios and p-values from the Fisher’s exact test implemented in LOLA R package. c Chromatin immunoprecipitation with an anti-H3K27ac antibody in lymphoblastoid cells from TET2delA carriers (Ly9, Ly11, and Ly14) was compared to that from wild-type individuals (Ly8 and Ly10). The difference in the enrichment of active H3K27ac-marked chromatin at CpG islands is displayed as a function of increasing methylation at CpG islands in whole blood of the TET2delA carriers as measured by WGBS. Negative correlation is stronger at CpG islands outside transcription start sites (TSS). Boxplots show the median, and the first and third quartiles

Fig. 4

TET2- and DNMT3A-mediated methylation changes are enhanced at open chromatin with master transcription factor-binding sequences (TFBSs). a Methylation is significantly increased in TET2delA carriers at lineage-specific open chromatin regions of monocytes, granulocyte/macrophage progenitors (GMPs), common lymphoid progenitors (CLPs), B cells, and megakaryocyte/erythroid progenitors (MEPs). Percentages represent average methylation differences at the respective open chromatin regions between mutation carriers (Ly1, Ly2, Ly9, Ly11, and Ly14) and 10 controls (Ly8, Ly10, Ly12, Ly13, HLRCC_N7, and controls 1–5) from targeted bisulfite sequencing. False discovery rate (FDR)-adjusted p-values are from two-sided Wilcoxon rank sum test. b Schematic of the human hematopoietic cell lineage hierarchy showing the 10 cell types analyzed in a. Black arrows depict emergence of lineages that showed significant enrichment of methylation at indicated TFBSs at lineage-specific open chromatin regions in the TET2delA carriers. c Count of hypermethylated (dark red bars) and hypomethylated (light gray bars) cytosines per million tested CpGs in DNMT3A mutation carriers (+/−) and age-matched controls (+/+) from the Northern Finland Intellectual Disability Cohort. Each sample was compared to the same set of five baseline controls as in Fig. 2a. d Magnitudes of the methylation changes at master TFBSs located in cell-specific open chromatin regions correlate in TET2 and DNMT3A mutation carriers, albeit these methylation changes occur toward different directions. X- and Y-axis represent average methylation change in mutation carriers as compared to controls at open chromatin regions with master TFBS relative to all open chromatin regions in each of the 10 cell types. Each dot of a particular color represents one of the 10 cell types, respectively. Pearson’s product-moment correlation coefficient (r) and p-value (p) are calculated for the relative ratios. Methylation values from TET2delA carriers were compared to the 10 controls as in a, and those from DNMT3A mutation carriers to the four controls as in c. RUNX and GATA represent binding sequences of RUNX1/2/3 and GATA1/2/3, respectively

Fig. 5

Single-cell RNA-sequencing analysis from peripheral blood. a The main cell types were present in similar numbers between cancer-free TET2delA carriers (Ly9, Ly11, and Ly14; right) and wild-type individuals (Ly8, Ly10, and Ly13; left). Colors represent blood cells with similar expression profiles in K-means (K = 10) clustering. Each point represents a cell in the coordinates specified by the two t-SNE (t-distributed stochastic neighbor embedding) components. b TET2 expression in each of identified cell types represented with unique molecular identifier (UMI) counts. Boxplot shows the mean ± standard deviation. TET2delA carriers are marked with +/− and wild-type individuals with +/+. c CXCR4 was the most significantly increased transcript in natural killer (NK) cells and CD8+ T cells [log2 fold change (FC) 2.06], CD4+ T cells (FC 1.12), and all cell types compiled (FC 0.92), in cancer-free mutation carriers (right) as compared to non-carriers (left). d TSC22D3 expression was significantly increased in NK cells and CD8+ T cells (FC 1.14) and monocytes (FC 1.1) of mutation carriers

Fig. 6

Inflammasome-mediated cytokine response and plasma IL-8 are not elevated in constitutional heterozygous TET2 loss. a–c Monocyte-derived macrophages from TET2delA mutation carriers and controls were a, b primed for 6 h with lipopolysaccharides (LPS) and interferon-gamma (IFN-gamma) followed by NLRP3 inflammasome activation with ATP, cholesterol crystals (CHC), or monosodium urate crystals (MSU), or c treated with LPS and IFN-gamma ± trichostatin A (TSA) for the depicted times. d, e Monocytes from TET2del4 and TET2X mutation carriers and controls and g macrophages cultured from normal human donors and transfected with control or TET2 siRNAs (SN) were stimulated with LPS + IFN-gamma + ATP as in a and b. a, d, e, g Secretion of mature interleukin-1 beta (IL-1B) and IL-18 was measured by enzyme-linked immunosorbent assay (ELISA) from cell culture supernatants, b cellular caspase-1 activity was detected by flow cytometry (MFI, median fluorescence intensity), and c mRNA expression was analyzed by quantitative PCR (AU, arbitrary units). f Plasma levels of the CXC chemokine IL-8 were measured by ELISA from TET2delA mutation carriers and healthy volunteers. Individual data points from each donor are shown with mean values ± s.e.m. f, g Statistical analysis was performed with Wilcoxon rank sum test