Single-Cell Transcriptomics of Bone Marrow Mesenchymal Stromal Cells Identifies Altered CLCF1 Signaling in DNMT3A-Mutated Clonal Hematopoiesis

Clonal hematopoiesis of indeterminate potential (CHIP) is characterized by the presence of somatic mutations in hematopoietic stem and progenitor cells (HSPC) and is considered a precursor to myeloid neoplasia. HSPC in DNMT3A and TET2 mutant CHIP show a proliferative advantage over time, but the underlying mechanisms have not been completely unraveled. We have previously shown that mesenchymal stromal cells (MSC) from individuals harboring a DNMT3A mutation instruct young healthy HSPC and influence their fate in vitro, pointing to a potential role for the bone marrow (BM) microenvironment in progression to myeloid neoplasia.

To gain further insight into the BM composition and cell extrinsic mechanisms contributing to clonal expansion in CHIP, we performed single-cell analysis on BM cells from 12 CHIP individuals harboring DNMT3A (DNMT3A^mut, VAF 4.5-25.5%) and/or TET2 mutations (TET2^mut, VAF 6.8-32.2%) and compared their transcriptomic profile to 3 healthy non-CHIP controls. scRNASeq was performed on sorted hematopoietic (CD45+ CD34+) and non-hematopoietic (CD45- CD271+ and CD45- CD31+) cells. For additional functional studies, mononuclear cells were enriched for CD34+ cells using magnetic cell sorting, and MSCs were isolated by plastic adherence.

In an unsupervised embedding of 57020 HSPC we annotated 10 clusters. Surprisingly, there were no significant differences in the cell type frequencies between CHIP and non-CHIP samples. However, we found an upregulated systemic inflammatory response mediated by TNFα, and activation of TGFβ signaling and hypoxia programs across all cell types in CHIP samples regardless of their genotype. Interestingly, only HSPC from DNMT3A^mut samples showed decreased oxidative phosphorylation and altered cycling capacity. In silico cell cycle analysis revealed gradual changes along the myeloid differentiation trajectory in DNMT3A^mut samples. While there were no differences in the HSC/MPP compartment, myeloid progenitors (CMP/GMP/MonoP) had fewer cells in G1 phase and a higher percentage in S and G2/M compared to non-CHIP samples.

To determine the contribution of extrinsic signaling to changes in cell cycle behavior of DNMT3A^mut HSPC, we analyzed 1801 MSCs from the same samples. We identified 4 clusters representing multipotent MSC, adipocyte-primed, osteogenic, and osteo-chondrogenic MSCs. Subset frequencies revealed a higher fraction of multipotent MSC in DNMT3A^mut CHIP BM compared to non-CHIP, and an imbalance between the adipogenic and osteogenic lineages. Overall, we identified 286 differentially expressed genes (DEG) in DNMT3A^mut CHIP-MSC, characterized by an upregulation of genes involved in inflammation (e.g. FOSL, KLF4, SGK1; 3.4-, 3.1-, 2.4-fold increase, respectively) and BM niche dynamics (e.g. IGF1, CD44, CXCL2; 5.6-, 2.6-, 2.6-fold increase, respectively), as well as downregulation of adhesion molecules (CADM3, CADM1, 7.1-, 2.1-fold decrease, respectively), collagen genes (COL8A1, COL14A1; 2.3-, 2-fold decrease, respectively), and the chemokine CXCL14 (2.9-fold decrease). To analyze the crosstalk with HSPC, we focused on the surfactome of MSCs at the single-cell level, which revealed a distinct pattern in CHIP-MSC. Cardiotrophin like cytokine factor 1 (CLCF1), which modulates hematopoiesis by promoting differentiation into the myeloid lineage, emerged as one of the top (4.4-fold increase) upregulated genes in the MSC surfactome of DNMT3A^mut donors. Strikingly, by using our previously described 42-hour in vitro co-culture set-up with DNMT3A^mut CHIP-MSC and healthy HSPC, we again identified CLCF1 signaling as one of the 50 ligand-receptor pairs shown to be differentially regulated in direct co-culture in vitro. The reduced cycling capacity was recapitulated in healthy HSPC after co-culture with DNMT3A^mut CHIP-MSC, along with a skewed HSPC compartment, further validated by increased clonogenicity and similar trends in HSPC subpopulations after 4-day culture in vitro.

In summary, our data recapitulates the well-known inflammatory state in CHIP, both in the hematopoietic and non-hematopoietic BM compartments. In addition, we show that the BM niche in CHIP exhibits profound alterations that influence HSPC and identify CLCF1 signaling as a putative mechanism underlying hematopoietic remodeling. Proteomic analysis of the BM secretome in CHIP and functional studies validating the identified targets are ongoing.