Distinct Pathogenesis of Clonal Hematopoiesis Revealed By Single Cell RNA Sequencing Integrated with Highly Sensitive Genotyping Method

Background

Recent evidence suggests that age-related clonal hematopoiesis (CH) might represent the earliest precursor of myeloid neoplasms. Although the exact mechanism of clonal selection that shapes CH is still to be elucidated, both cell intrinsic and non-cell intrinsic effects of mutations, including the interplay between mutated cells and the bone marrow environment, are thought to play important roles, which are best studied using single-cell sequencing analysis of both mutations and gene expression.

Methods

We performed single-cell sequencing of hematopoietic stem and progenitors (HSPCs) from BM of the 16 patients with CH along with 16 control patients without CH identified by screening otherwise healthy individuals who received hip joint replacement, using a novel platform that enables simultaneous detection of gene mutations and expression based on the Fluidigm C1-HT system. Sequence data were analyzed with Seurat (Stuart et al Cell 2019) with integration of genotyping information. Cells were clustered and each cluster was assigned by marker-gene expressions for major cell-types in HSPCs, including hematopoietic stem cell (HSC)-like and erythroid progenitors. Cells were grouped by their genotypes and pathway analysis were performed.

Results

In total, we identified 35 subjects who had CH-related mutations, including those affecting DNMT3A, TET2, ASXL1, SF3B1, PPM1D, IDH1, GNB1 and TP53, of which 11 had more than one CH-related mutation. Most of these mutations showed a low variant allele frequency (VAF) ≤ 0.05. However, clones having double mutations of DNMT3A/TET2 or those having biallelic TET2 mutations tended to show a higher VAF as high as 0.4, suggesting an enhanced clonal advantage for clones having multiple mutations.

Using our novel single-cell platform, we analyzed 3,767 cells from control patients without CH and 1,474 mutated cells and 7,234 wild-type (WT) cells from patients with CH. By targeting both genomic DNA and RNA, we successfully obtained a sufficient number of single-cell reads for genes whose expression was too low to evaluate by only targeting RNA, such as TET2 and DNMT3A. Although some clones having a high-VAF mutation caused a skewed clustering to be detected as a CH clone, many clones with low-VAF mutations did not make distinct clusters, indicating the importance of genotyping at a single cell level to identify and characterize mutated cells.

Simultaneous detection of genotype and expression allowed us to see the effect of CH-mutations on cell phenotype and differentiation. For example, cells having compound TET2/DNMT3A mutations were significantly enriched in the erythroid cluster, while another clone with double TET2 mutations were more enriched in the HSC-like cluster, compared to cells from individuals without CH (WT^cont). These are in line with the previous findings of TET2/DNMT3A double knockout mice or TET2 knockout mice, respectively.

In another case with an IDH1 mutation, IDH1-mutated (MUT^IDH1) cells less contributed to the HSC-like fraction, showing an enhancement of cell proliferation-signature, compared to WT (WT^IDH1) cells in the same patient. Strikingly, compared to WT^cont cells, WT^IDH1 cells were significantly enriched in the HSC-like fraction and showed an enhanced expression of cytokine-related pathway genes, which was in line with a finding seen in mouse cells treated with 2-hydroxy-glutalate, an mutant IDH-related oncometabolite. Similarly, when compared to WT^cont cells, WT cells from patients with DNMT3A- (WT^DNMT3A) or TET2- (WT^TET2) mutated CH significantly showed an enhanced cell proliferation. HSC-like WT^TET2 cells also showed aberrant IFN-response signatures compared to corresponding WT^cont cells, which was confirmed in competitive transplantation of Tet2 heterozygous knockout (hKO) and WT cells in a mouse model; HSPCs of WT competitors transplanted with Tet2-hKO cells showed a significant enhancement of IFN-response signatures compared to those transplanted with WT cells. Intriguingly, monocytes of Tet2-hKO donors showed aberrant expression of S100a8/a9, which might contribute to the non-cell intrinsic effect of Tet2-hKO cells.

Conclusions

In CH, not only mutated cells but also surrounding WT cells show an aberrant gene expression phenotype, suggesting the presence of non-cell autonomous phenotype or an altered bone marrow environment that favors the positive selection of CH-clones.