Patient-Derived Models of Clonal Hematopoiesis Point to Cell-Intrinsic and -Extrinsic Contributions to Clonal Dominance and Lineage Imbalances

Clonal hematopoiesis (CH) likely results from both cell-intrinsic and -extrinsic mechanisms that award a competitive advantage to the clone. Cell-autonomous regulation occurs through mutations, frequently in epigenetic regulators (DNMT3A, ASXL1, TET2), affecting differentiation and self-renewal of hematopoietic stem cells (HSC). Bone marrow aging, tightly linked to CH, is characterized by inflammation, suggesting that cell-extrinsic pressures in CH might not only select the mutant clone, but also potentially disturb the entire CD34+ compartment. Using bone marrow (BM) samples from individuals with DNMT3A CH, we developed in vitro and patient-derived xenograft models to study the contribution of cell intrinsic and extrinsic pressures on clonal architecture and hematopoietic lineage imbalances. To characterize the hematopoietic stem and progenitor cell (HSPC) compartment of CH, we performed scRNAseq on CD34+ BM cells from two patients with non-R882 DNMT3A CH (p.G543C VAF = 29.82%; p.R771Q VAF = 20.56%) and age-/sex-matched controls (CTR). We found myeloid-biased hematopoiesis and a contracted HSC compartment in CH (p<0.0001). Flow cytometry showed a gradual contraction of the lymphoid progenitor compartment (CD34+CD10+) from young, to old donors, to CH CD34+ BM cells (median, Q1-Q3: young: 20.8%, 10.6-30.4%; old: 11.30%, 3.3-15.5%; CH: 5.6%, 2.4-9.3%; p<0.01). We performed Genotyping of Transcriptomes (GoT) to study the transcriptional differences between mutant (CH-M) and wild-type (CH-WT) HSPC from CH BM and non-CH CTR. Surprisingly, CH-WT and CH-M HSPC demonstrated subtle transcriptomic differences while profound changes in the transcriptome were present in CH-M and CH-WT compared to non-CH CTR. CH HSPCs have lower expression of CXCR4 (p<0.05), the CXCL12-binding receptor responsible for HSC maintenance, thus potentially explaining the contraction of the HSC compartment. More so, they showed upregulated expression of AP-1 complex genes (JUN, FOS, FOSB) pointing toward an inflammatory bone marrow microenvironment (BME). To identify transcriptomic signatures suggestive of inflammation in the BME, we first created an atlas of human BME by performing scRNAseq of the CD45-CD235a-CD71-SLAMF7- fraction of BM from young donors. We identified a heterogeneous population of cells including mesenchymal stromal cells (MSC), osteolineage cells, smooth muscle cells, fibroblasts, and endothelial cells. Since CH is strongly correlated with advanced age, we compared these cells with data generated from old donors. Compared to young MSC, old MSC presented upregulation of AP-1 complex genes and downregulation of CXCL12. Thus, the aged BME showed a similar inflammatory signature as CH hematopoietic cells and is likely implicated in defective CXCL12-CXCR4 axis signaling in CH. We sought to understand the ability of MSCs to modulate HSPCs, by co-culture of CD34+ cells from a healthy, old donor in the presence of MSCs from young donors, old donors, or donors with CH. Co-culture with CH-MSC favored expansion of the CD34+CD38+ progenitors at the expense of the HSC-enriched CD34+CD38- fraction compared to healthy old MSC (p<0.05). This suggests a superior support of progenitor cells, with loss of HSC by the CH BME. Among progenitors, pro-B cells (CD34+CD10+) are particularly lost during co-culture with CH-MSC (p<0.05). To study the cell-intrinsic mechanism of clonal dominance, we used a novel xenograft model. We engrafted DNMT3A CH-CD34+ into clodronate/busulfan-conditioned NSG mice. CH-CD34+ cells had higher engraftment than CTR (p<0.05, BM at 3 months). The clonal burden of DNMT3A CH was maintained at 3 months post-engraftment at the same size as baseline (VAF ~ 14%). Despite the murine BME, reconstituted hematopoiesis of CH cells showed a trend toward impaired lymphoid reconstitution(CD19+: mean, SD: CH: 15.23%+/-18.39, CTR: 30.22%+/-14.18). These findings suggest that clonal fitness and the differentiation imbalances of DNMT3A CH CD34+ cells are, at least partially, driven by intrinsic factors. In conclusion, our findings from scRNAseq analysis, in vitro and in vivo models of human-derived CH point towards both cell-intrinsic and cell-extrinsic factors that contribute to clonal behavior and lineage imbalances.