Metaphyseal Stromal Progenitor Cells Are Skewed Via Wnt-Activation Towards Osteogenesis in Progressed Myelofibrosis

Bone remodeling occurs actively alongside bone marrow (BM) fibrosis in the context of myeloproliferative neoplasms (MPN). Osteosclerosis in MPN is characterized by para-trabecular apposition of new bone and increase in bone density. Yet, the pathogenesis of osteosclerosis is largely unknown and it remains an unanswered question how the 1) peri-vascular, and 2) endosteal stromal niche contribute to the formation of new bone.

Here, we employed genetic fate tracing and high-resolution imaging combined with single-cell RNA sequencing (scRNAseq) to map out the BM stromal niche in steady-state, after transplant of control or thrombopoietin (ThPO)-overexpressing hematopoietic stem and progenitor cells (HSPCs) which robustly and with defined kinetics results in BM fibrosis/osteosclerosis. High-resolution confocal imaging revealed that PdgfrbCreERt2-derived tdTomato+ cells are abundant throughout the whole marrow, growth plate and bone. Gli1CreERt2;tdTomato+ have a distinct localization specifically at the growth plate, in the trabecular region of the metaphysis and around the central artery in the diaphysis. Grem1CreERt2;tdTomato+ cells are the least abundant, specifically located at the growth plate.

Integrated data sets provided a high resolution of the BM stroma representing Cxcl12-abundant reticular (CAR) cells, fibroblasts (FBs), osteo-lineage cells (OLCs), osteoblasts (OBs), (pre)chondrocytes, Schwann cells and endothelial subpopulations. We next asked how these different subsets contribute to BM fibrosis and osteosclerosis. CAR cells were functionally reprogrammed in fibrosis: they lost their hematopoiesis-supporting capacity, gained a pro-fibrotic phenotype, produced extra-cellular matrix (ECM) and were less frequent compared to control conditions. BM-resident FBs rather gained a pro-inflammatory phenotype, comparable to inflammatory fibroblasts described in solid organ fibrosis, and were enriched in fibrosis but did not show a "pro-fibrotic" expression profile. Strikingly, the major expansion of tdTomato+ stromal cells occurred at the metaphysis, overlapping with increased reticulin deposition in this region.

Deconvolution revealed that mainly OLCs, pre-chondrocytes and chondrocytes but also FBs reside in the metaphysis. Cell trajectory analysis suggested that a subset of OLCs acts as a mesenchymal precursor reservoir for OBs, chondrocytes and FBs. In line with our hypothesis, only metaphyseal, but not diaphyseal, PdgfrbCreERt2;tdTomato+ cells from steady-state mice harbored CFU-F capacity and gave rise to colonies in vitro. Receptor-ligand interaction demonstrated that OLCs function as important information-hubs to CAR cells after BM transplantation compared to steady-state bone. This pro-regenerative phase is characterized by upregulation of adipogenesis-related genes, potentially prompting the BM niche to recover hematopoiesis. In response to a fibrotic cue, OLCs expand, lose cellular crosstalk reflected by decreased receptor-ligand pairs, downregulate adipogenic signatures but upregulate osteogenic signatures, thus skewing the BM stroma towards osteogenesis. Time-course imaging of the stepwise invasion of Gli1CreERt2;tdTomato+ cells into the BM from the growth plate additionally highlighted the active bone remodeling occurring at the chondrocyte-OLC border during fibrosis. µCT imaging revealed increased ossification specifically at the metaphyseal region. Pathways analysis demonstrated that metaphyseal cells remain active even in progressed fibrosis and show enriched Wnt, PI3K and ECM receptor signaling. Interestingly, Wnt pathway inhibitors were downregulated in pro-fibrotic-CAR cells, whereas Wnt signaling was increased in OLCs as mesenchymal progenitor cells upon fibrotic transformation.

In summary, we provide evidence that active bone remodeling is co-occurring with the fibrotic transformation with a skewing of stromal-cell fate towards osteogenesis rather than adipogenesis. Our analysis highlights the functional differences of metaphyseal and diaphyseal macro-niches within bone and postulates that a metaphyseal stromal progenitor is activated in an injury-specific manner, being an attractive cellular target.

No relevant conflicts of interest to declare.