Myelodysplastic Syndrome associated TET2 mutations affect NK cell function and genome methylation

Abstract

Myelodysplastic syndromes (MDS) are clonal hematopoietic disorders, representing high risk of progression to acute myeloid leukaemia, and frequently associated to somatic mutations, notably in the epigenetic regulator TET2. Natural Killer (NK) cells play a role in the anti-leukemic immune response via their cytolytic activity. Here we show that patients with MDS clones harbouring mutations in the TET2 gene are characterised by phenotypic defects in their circulating NK cells. Remarkably, NK cells and MDS clones from the same patient share the TET2 genotype, and the NK cells are characterised by increased methylation of genomic DNA and reduced expression of Killer Immunoglobulin-like receptors (KIR), perforin, and TNF-α. In vitro inhibition of TET2 in NK cells of healthy donors reduces their cytotoxicity, supporting its critical role in NK cell function. Conversely, NK cells from patients treated with azacytidine (#NCT02985190; https://clinicaltrials.gov/ ) show increased KIR and cytolytic protein expression, and IFN-γ production. Altogether, our findings show that, in addition to their oncogenic consequences in the myeloid cell subsets, TET2 mutations contribute to repressing NK-cell function in MDS patients.

Fig. 1. TET2/IDH mutations in MDS/CMML patients lead to the reduction of KIR and Perforin expression in NK cells.

a Percentages of blood NK cells expressing activating (NKp30, NKp46, DNAM-1, NKG2D) and inhibitory (CD96, CD85j, KIR2D, NKG2A) receptors, and maturation/activation markers (KLRG1, CD69, CD57) were measured by flow cytometry in TET2/IDH^WT (n = 13, n = 11 for NKG2A) and TET2/IDH^MUT (n = 19, n = 15 for NKG2A) MDS/CMML patients. Statistics were calculated with the nonparametric Mann–Whitney test, two-sided, KIR2D ***p = 0.0005, NKG2A *p = 0.0246. b Percentages of KIR2D⁺ BM NK cells measured by flow cytometry in TET2/IDH^WT (n = 12) and TET2/IDH^MUT (n = 7) MDS/CMML patients. Statistics were calculated with the nonparametric Mann–Whitney test two-sided *p = 0.013. c and d Specific expression of KIR2DL1, KIR2DL2/DL3, and KIR3DL1/DL2 in blood NK cells detected by flow cytometry in TET2/IDH^WT (n = 11) and TET2/IDH^MUT (n = 15) MDS/CMML patients. One representative example is shown in c. Statistics were calculated with the nonparametric Mann–Whitney test, KIR2DL1 **p = 0.0092, two-sided KIR2DL2/DL3 **p = 0.0077. e Intracellular perforin and granzyme B expression in blood NK cells measured by flow cytometry in TET2/IDH^WT (n = 19) and TET2/IDH^MUT (n = 22) MDS/CMML patients. Statistics were calculated with the nonparametric Mann–Whitney test *p = 0.0337. f Receiver operating characteristic (ROC) curve depicting the relationship of true TET2 mutation presence (sensitivity) and false TET2 mutation presence (100%-specificity) for a KIR2D expression threshold at 25% in blood NK cells (p < 0.0001) quantified by flow cytometry in TET2/IDH^WT (n = 19) and TET2^MUT (n = 17) MDS/CMML patients. g KIR2D expression on blood NK cells of the TET2/IDH^WT (n = 19), TET2^MUT (n = 17) and IDH^MUT (n = 5) patients showed in the ROC curve. The horizontal black bar indicates the threshold at 25% KIR2D⁺ NK cells. Statistics were calculated with the nonparametric Mann–Whitney test, TET2/IDH^WT vs. TET2^MUT: ****p < 0.0001; TET2/IDH^WT vs. IDH^MUT: *p = 0.0152. For all the analysis, data are presented as medians and interquartile ranges. Source data are provided as a Source Data file.

Fig. 2. TET2 mutations identified in bulk cells of MDS patients at diagnosis are also observed in NK cells and correlate with KIR2D expression.

a The percentage of Variant Allele Frequency (VAF) for the mutations of TET2 observed in the white mononuclear cells (WMC) and in sorted NK/T cells at diagnosis of TET2^MUT MDS/CMML patients (n = 10) was evaluated by NGS analysis (with a range from 1 to 4 mutations per patient). Data are represented as box-and-whisker plots (minimum VAF, 25% percentile, median, 75% percentile, and maximum VAF respectively for WMC: 0%, 11.25%, 32.5%, 42.5%, 49%; for NK cells: 0%, 3.5%, 14%, 35.5%, 54%; and for T cells: 0%, 0%, 0%, 0%, 7%). Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance. ****p < 0.0001. b VAF (%) of the different mutations detected in the WMC at diagnosis (blue) and in sorted NK cells (red) in 4 patients (MUT08, MUT22, MUT19, MUT31; see Supplementary Table 6 for more information). c Correlation curve between the percentage of KIR2D⁺ NK cells and the TET2 VAF (%) in blood NK cells from the MDS/CMML patients analyzed in (a). Linear regression was calculated, r = 0.88, p < 0.0001. d The VAF percentage in the WMC and in sorted KIR+ and KIR− NK cells at diagnosis from TET2^MUT MDS/CMML patients (n = 5, range from 1 to 4 mutations per patient) was evaluated by NGS analysis. Data are represented as box-and-whisker plots (minimum VAF, 25% percentile, median, 75% percentile, and maximum VAF respectively for KIR− NK cells: 6.1%, 8.3%, 18.2%, 36.5%, 42.1%; and KIR+ NK cells: 0%, 1.2%, 7.3%, 13%, 48%). Statistics were calculated with a two-sided nonparametric Wilcoxon matched-pairs signed rank test. *p = 0.0186. Source data are provided as a Source Data file.

Fig. 3. TET2 regulates KIR expression on NK cells through its binding onto and regulating methylation of the KIR locus.

a TET2 mutational landscape in the 13 TET2^MUT patients with KIR2D expression below 25% established with the St Jude protein paint software (https://proteinpaint.stjude.org/). Mutations were classified as Missense (blue), Frameshift (red), and Nonsense (orange). Numbers associated with each mutation designed patients (Supplementary Data 1). b Percentage of KIR2D⁺ NK cells after 5 days of in vitro treatment with 1 µM DMOG (n = 10). DMSO alone was used as a control. One dot represents one PBMC sample. Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance. **p = 0.0059. Data are presented as median and interquartile range. c Fold enrichment of sequences specific for the Cis-regulatory element (CRE), KIR2DL1 promoter, and KIR2DL2/3 promoter analyzed in sorted NK cells by ChIP-qPCR with TET2, H3 or H3K18 specific mAbs or IgG isotype control. Means ± SD is shown (n = 3). Each dot represents one independent experiment. d Luciferase activity in HEK293T cells co-transfected with a TET2 full-length plasmid or an empty plasmid as control, and the luciferase-reporter construct containing the region (−147 + 60) of the KIR2DL1 promoter. Means ± SD is shown (n = 4). Each dot represents one independent experiment. Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance. *p = 0.0286. e Jurkat cells transfected with the KIR2DL1 promoter-luciferase reporter plasmid were treated with 500 µM of L-AA for 16 h and analyzed by detecting luminescence signal. Means ± SD is shown (n = 4). Each dot represents one independent experiment. Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance. *p = 0.0286. f RRBS DNA methylation profiles of the extended KIR locus of NK cells with low or high KIR2D expression. Each graph represents the DNA methylation profile of sorted NK cells from blood samples of 5 patients; vertical bars represent the percentage of DNA methylation at the CpG position. CRE and KIR genes were highlighted in green and gray respectively. g and h IGV (Integrative Genomics Viewer) view of CpG read signals corresponding to DNA methylation in NK cells, based on the high (TET2^WT/MUTKIR^HIGH, in blue bars) and low (TET2^MUTKIR^LOW, in red bars) KIR expression. Genome profiles at the CRE region (g) and the KIR2DL1 gene (h) loci showed variation in the DNA methylation pattern between the two groups of patient NK cells. Red peaks/boxes show significant differences in the DNA methylation levels at particular CpG positions of these specific loci. Source data are provided as a Source Data file.

Fig. 4. Loss of TET2 leads to DNA hypermethylation and decreases key gene expression for NK cell function.

a Circos plots showing whole-genome CpG methylation status in patient MUT14 characterized by the absence of TET2 mutations in NK cells and a high expression of KIR2D, and in patient MUT22 with a VAF of 50% for the TET2 mutation [NM_001127208:exon11:c,4669_4672del:p.V1557fs] and a very low expression of KIR2D in NK cells. b Volcano plots and heatmaps showed the overall increase of global DNA methylation levels reported after RRBS analysis, in the TET2^MUTKIR^LOW (n = 3) vs. TET2^WT/MUTKIR^HIGH (n = 2) NK cells. Heatmaps depicted supervised clustering of the significantly modified sited genes between patients’ subgroups. Blue dots/bars show the hypomethylated CpG/genes whereas red dots/bars show the hypermethylated ones (methylation difference ≥20%, unadjusted p-value ≤ 0.05). Top panel shows the differentially methylated CpG sites. Bottom panel shows the differentially methylated genes. c GO-enrichment analysis on the differentially methylated genes in the TET2^MUTKIR^LOW and TET2^WT/MUTKIR^HIGH NK cells. Percentages of methylated CpG sites were calculated in gene bodies and 10 kb upstream or downstream of the gene of interest in TET2^MUTKIR^LOW and TET2^WT/MUTKIR^HIGH NK cells (in red and blue, respectively) and aggregated across all genes of a given KEGG pathway for each sample. Pathways of interest shown are the cytokine–cytokine receptor interactions (KEGG reference hsa04060), the JAK-STAT signaling pathway (hsa04630), the TNF signaling pathway (hsa04668), and the NK cell-mediated cytotoxicity pathway (hsa04650). Source data are provided as a Source Data file.

Fig. 5. Hypermethylation of particular CpG sites represses NK cell gene expression and function.

a Methylation profiles, established after RRBS analyses, depicted with IVG at the IFNG (upper panel), TNF (middle panel) and PRF1 (lower panel) loci, were shown in NK cells from patients segregated on the high (TET2^WT/MUTKIR^HIGH, in blue bars) and low (TET2^MUTKIR^LOW, in red bars) KIR expression. Red peaks/boxes show significant differences in the DNA methylation levels of these specific loci. b The regulatory activity of specific CpG sites was analyzed in the HEK293T cell line transfected with a luciferase-reporter plasmid including methylated or non-methylated genes’ regulatory regions. Relative luciferase activities of the in vitro methylated regions were compared to their non-methylated counterpart (FC = Methylated/Putative Promoter). Means ± SD is shown (n = 4). Data were analyzed using the one-way Friedman test followed by a Dunn’s test. *p = 0.0433. c Quantification of the KIR2DL1, TNF, and IFNG transcripts by RT-qPCR on KIR2D⁻ NK cells sorted from TET2/IDH^WT (n = 11, in blue) and TET2^MUT (n = 10, in red) patients. Medians and interquartile are shown. Statistics were calculated with the two-sided Mann–Whitney test, **p = 0.0036. d Fold changes of NK cells killing activity against the NK-sensitive cell line K562 previously treated for 5 days with 1 µM DMOG (n = 10). Each dot represents one independent experiment. DMSO alone was used as a control. Means ± SD is shown. Data were analyzed using the one-way Friedman test followed by a Dunn’s test. *p = 0.0417. Source data are provided as a Source Data file.

Fig. 6. Hypomethylating agents normalize the NK cell phenotype of MDS patients.

a Evaluation of KIR2D surface expression on NK cells of TET2/IDH^MUT patients after treatment with azacitidine (AZA, n = 12), decitabine (DAC, n = 12), acid ascorbic (AA, n = 7), and DAC+AA (n = 7). Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance, DAC: ***p = 0.001, DAC+AA: *p = 0.0156. b NK cells were isolated from patients’ PBMC before and after 3 cycles of treatment with AZA, and cultured overnight at 100U/ml of IL-2. Subsequently, cells were cultured with PMA-Ionomycin for 6 h. The frequency of responding cells in terms of IFN-γ was assessed by flow cytometry (n = 7). c KIR2D, perforin, and granzyme B expression were measured by flow cytometry in the blood of MDS patients (n = 9) before (in blue) and after (in red) 6 cycles of AZA treatment. Nonparametric two-sided Wilcoxon matched-pairs signed rank test was used to determine statistical significance, KIR2D: *p = 0.0195, Perforin: *p = 0.0273, Granzyme B: *p = 0.0273. Data are presented as medians and interquartile ranges. Source data are provided as a Source Data file.