Bone Marrow Fibroblasts at the Single-Cell Resolution Level in Multiple Myeloma

Introduction
Fibroblasts play an important role in facilitating the development of multiple solid cancers, but their role in the multiple myeloma (MM) microenvironment is less well understood. Fibroblasts and platelets have complementary functions in normal wound healing: damage and mechanical stress trigger both cell types to secret cytokines and chemokines which facilitate tissue growth. In solid malignancies, mechanical stress also triggers cancer-associated fibroblasts to promote tumor growth through similar mechanisms. Characterization of fibroblasts in the MM microenvironment may provide important parallel insights into targetable pathways by which fibroblasts promote MM growth and development.

Methods
We used single-cell RNA sequencing to profile bone marrow stromal fibroblasts in healthy mice and in the 5TGM1 mouse MM model. Using Seurat, fibroblasts were divided into 10 subpopulations which were characterized based on gene expression and Gene Set Enrichment Analysis (GSEA). To validate the clinical relevance of these subpopulations, we analyzed a human bone marrow stromal data set from MM patients and controls, but the human data lacked identifiable fibroblasts. To uncover pathways by which MM cells might reprogram murine fibroblasts, we then used NicheNet to predict ligand-receptor interactions between fibroblasts and MM cells, identifying interactions which could account most fully for the differential gene expression in fibroblasts, across conditions.

Results
Among the stromal cells, fibroblasts were identified by their expression of canonical markers including Dcn, Col1a1, and Fn1. Fibroblasts were enriched in MM compared to controls (11.46% vs. 9.42% of stromal cells; p = 0.000018, post-hoc chi-square test with Bonferroni correction). The distributions among the 10 fibroblast subclusters were also different across the two conditions (p < 2.2 * 10^-16, chi-square test). Two subclusters were very substantially enhanced in the MM mice: fibroblasts-9 and -5. Fibroblasts-9 was composed of 66 of the 1630 MM fibroblasts and 9 of the 1339 healthy fibroblasts (4.05% vs. 0.67%; p < 0.000001, post-hoc chi-square test with Bonferroni correction). All 123 cells in fibroblasts-5 were found in the MM marrows. To probe the role of fibroblasts in shaping the MM microenvironment, we next sought to characterize these two subpopulations strongly associated with MM.

Compared with other fibroblasts, fibroblasts-9 highly expressed numerous platelet markers, including many genes coding for platelet cell-surface receptors. Among these were Mpl (encoding the thrombopoietin receptor); Itga2b and Itgb3 (encoding CD41 and CD61, the two components of glycoprotein IIb/IIIa, a major platelet receptor for fibrin); Gp9, Gp5, Gp1ba, and Gp1bb (encoding all components of the glycoprotein Ib/IX/V complex, the main platelet receptor for von Willebrand factor); Treml1 (encoding TLT-1, a platelet fibrin receptor); Clec1b (encoding CLEC-2, a receptor for podoplanin expressed on platelets and immune cells); and P2ry12 (encoding P2Y12, the major platelet receptor for ADP). The most significantly overexpressed pathways by GSEA were HALLMARK_COAGULATION and HALLMARK_ANGIOGENESIS. Fibroblasts-9 may thus represent a novel fibroblast subtype with platelet-like markers, enriched in the MM bone marrow.

Fibroblasts-5 was found exclusively in MM mice and is likely metabolically active, with significantly overexpressed HALLMARK_FATTY_ACID_METABOLISM, HALLMARK_OXIDATIVE_PHOSPHORYLATION, HALLMARK_MYC_TARGETS_V1, and HALLMARK_GLYCOLYSIS in GSEA analysis. Its role in the MM microenvironment, however, is not clear.

We used NicheNet to predict ligand-receptor interactions between MM cells and fibroblasts. Col1a1 on MM cells was predicted to interact with Sdc1 and several integrins expressed by fibroblasts and to regulate Serpina3n. Serpina3n was most highly expressed by fibroblasts-5 and fibroblasts-9, and its human ortholog SERPINA3 has numerous protumor effects in other malignancies, such as suppression of apoptosis.

Conclusions
We found substantial differences between murine bone marrow fibroblast subpopulations in health versus MM, identifying potentially therapeutically targetable constituents of the MM microenvironment. Fibroblasts and platelets mediate a progrowth response in wound healing, and platelet-like fibroblasts may influence MM development.