Transcriptional Landscape of the Microenvironment in Bone Marrow Fibrosis at Single Cell Level

Although the molecular alterations in hematopoietic cells which drive the development of myeloproliferative neoplasms (MPN) have been largely defined, reactive cellular alterations in the non-hematopoietic compartment remain rather obscure and have not been studied at single cell level.

We therefore profiled enriched non-hematopoietic bone marrow cells by scRNAseq in bone marrow (BM) fibrosis compared to healthy marrow. BM fibrosis was induced by transplantation of hematopoietic stem and progenitor cells (HSPCs) with overexpression of Thrombopoietin (ThPO) into lethally irradiated mice. As ThPO-overexpression robustly leads to reticulin fibrosis in all mice (100%), we were able to study 1) pre-fibrosis (5 weeks after transplantation; reticulin fibrosis grade 0) and 2) manifest bone marrow fibrosis (10 weeks after transplantation, reticulin grade 2-3). The analysis revealed a total of 8 distinct clusters: 1-4) subpopulations of mesenchymal stromal cells (MSC-1: adipogenic, MSC-2: osteogenic, MSC-3: transition, MSC-4: interferon^high), 5) osteoblastic lineage cells (OLCs), 6) arterial endothelial cells (ECs) and 7-8) Schwann cell precursors (SCP-1: non-myelinating SCPs; SCP-2: myelinating SCPs). Exposure to ThPO overexpressing HSPCs resulted in an overrepresentation of adipogenic MSCs at the expense of all other MSC subclusters. Differential gene expression analysis revealed a functional reprogramming of the "adipogenic" expanding MSCs with down-regulation of hematopoiesis-support and induction of a secretory phenotype including upregulation of various extracellular matrix (ECM) proteins driving fibrosis. Interestingly, only two MSC subclusters gained significant ECM expression indicating myofibroblast differentiation. Expansion of OLCs in BM fibrosis suggested a differentiation of the underrepresented MSC subpopulations into osteolineage cells which was confirmed by pseudotime analysis. Myelinating SCPs, highly expressing interleukin-33 (IL-33), showed the largest expansion in fibrosis. IL-33 is described to play a significant role in solid organ fibrosis by having both pro- and anti-fibrotic effects. Nerve injury triggers the expansion of myelinating and non-myelinating Schwann cells to promote repair, suggesting that mSCPs increase as compensatory and regenerative mechanism for the previously described MPN-induced sympathetic neuropathy. Dissection of cellular and molecular alterations in pre-fibrosis and manifest fibrosis demonstrated that only one MSC subpopulation was already significantly expanded in the pre-fibrotic phase, but only showed minor transcriptional changes. The upregulation of ECM proteins, osteogenesis as well as proinflammatory genes were hallmark features of manifest fibrosis. Interestingly, the overrepresentation of IL-33 expressing mSCPs was more pronounced in the pre-fibrotic phase, indicating that the expansion is a regenerative phenomenon failing in the stages of manifest fibrosis. Our findings were validated in the clinically relevant JAK2(V617F)-induced model of myelofibrosis.

In conclusion, we here identified two distinct MSC subsets that are pro-fibrotic and contribute to osteosclerosis in PMF. The functional reprogramming of these MSCs in the bone marrow niche was accompanied by expansion of mSCPs with regenerative capacities, most likely caused by neural damage and Schwann cell death triggered by mutant HSCs.