Comprehensive Cellular Dissection of the Bone Marrow Microenvironment in Primary Myelofibrosis

Primary myelofibrosis (PMF) is a severe subtype of myeloproliferative neoplasm (MPN) characterized by progressive bone marrow (BM) fibrosis and hematopoietic insufficiency, reflecting profound pathology of the BM microenvironment. However, a comprehensive understanding of the stromal cell changes that drive BM remodeling in PMF remains lacking.

We performed lineage tracing and single-cell RNA sequencing (scRNA-seq) of PMF bone marrow stromal cells to comprehensively understand the maladaptive fibrosis process. PMF was induced using two established murine transplantation models: THPO overexpression (TOE) and the clinically relevant MPLW515L mutation. LepR and Gli1 are both putative markers of mesenchymal stromal cell (MSC) populations that give rise to myofibroblasts. To assess their relative contributions to BM fibrosis, we performed a head-to-head lineage tracing comparison using Lepr-Cre; tdTomato and Gli1-CreER; tdTomato mice. Our steady-state analysis in young adult animals (6-7 weeks) demonstrated that LepR- and Gli1-lineage cells occupy largely distinct areas of long bones aside from limited overlap in the metaphysis. LepR-lineage cells were distributed uniformly throughout the bone marrow whereas Gli1-lineage cells were sparse and primarily localized near the growth plate. Under PMF conditions, LepR ⁺ cells contributed to the majority of myofibroblasts (>80%) while Gli1 ⁺ cells contributed to a minority of myofibroblasts (<5%) in the central marrow. Accordingly, Gli1 knockout from either the stromal compartment or the hematopoietic compartment did not significantly alter the course of PMF development. We next performed scRNA-seq on control and PMF stromal cells from Lepr-Cre; tdTomato mice, which allowed us to retain the relevant lineage information and collect an unbiased representation of all BM cells. After excluding hematopoietic cells, three distinct stromal cell lineages (MSCs, endothelial cells, and glial cells) were apparent from the aggregate data. MSCs exclusively expressed tdTomato, prompting us to divide them into Lepr1, Lepr2, Lepr3, Lepr4, Lepr cycling, and pericyte subpopulations. Compared with controls, PMF mice had significant expansion of Lepr3, Lepr4, pericytes, and glial cells. Gene expression analysis showed that Lepr3 MSCs were osteolineage-fated cells (high Alpl, Postn, Bglap) whereas Lepr4 MSCs were enriched for matricellular genes including Timp1, Fbln2, and Sdc4, underscoring how fibrosis involves dynamic cell-matrix interactions. Importantly, as a whole, LepR-MSCs upregulated extracellular matrix proteins (e.g. Col1a1, Col3a1) and downregulated key niche factor genes (e.g. Cxcl12, Scf), consistent with reprogramming toward a myofibroblast fate that drives BM fibrosis. Concurrently, neurovascular changes manifested with endothelial cell upregulation of arteriolar-signature genes, accompanied by an expansion of pericytes and Sox10 ⁺ glial cells. Cells within the neurovascular unit regulate an array of functional responses in various organs, and these changes in PMF may coordinate processes such as angiogenesis, osteosclerosis, and hematopoietic stem cell (HSC) mobilization. Ligand-receptor interaction analysis revealed a dramatic increase in cell-cell signaling among the various stromal cell populations in PMF, with pericytes and glial cells serving as active signaling hubs. Differential intercellular information flow identified familiar fibrosis pathways such as PDGF and TGFb as well as novel pathways such as NOTCH, MIF, and GRN. Notably, Hedgehog signaling was not appreciated.

In summary, we performed a comprehensive cellular dissection of the transformed BM microenvironment in PMF using rigorous cell lineage tracing and unbiased scRNA-seq. Our analysis firmly establishes LepR ⁺ MSCs as the primary reservoir for myofibroblasts in the bone marrow and provides single-cell transcriptional profiling data that enable a global understanding of complex cellular crosstalk in the PMF niche.