Single Cell Transcriptomic Analysis of the Multiple Myeloma Bone Marrow Identifies a Unique Inflammatory Stromal Cell Population Associated with TNF Signaling

Background: In multiple myeloma, tumor cell survival, disease progression and therapy response are influenced by signals derived from the non-malignant bone marrow niche. This notwithstanding, a detailed in-vivo definition of the cells that define the multiple myeloma niche is lacking. Mesenchymal stromal cells are important niche constituents. Recent progress made with single cell transciptomics suggests that mesenchymal stromal cells are a dynamic population of cells that can exist as several subsets with functionally distinct activation and differentiation profiles.

Aim: To identify mesenchymal stromal cell subsets specific for the multiple myeloma bone marrow niche, by comparing stromal cells from myeloma patients to non-cancer controls.

Methods: The non-hematopoietic bone marrow niche was isolated from viably frozen bone marrow aspirates from 10 newly diagnosed multiple myeloma patients (6 hyperdiploid, 3 t(11;14) and 1 with deletion of 17p) and 2 non-cancer controls using high speed cell sorting. The purified cells were analyzed by 10X Genomics single cell sequencing directly post-thawing, without prior cell culture. From 10 multiple myeloma patients we generated single cell transcriptomes with an average read-depth of 20,000 reads per cell of in total 12,000 niche cells and from the 2 non-cancer controls a total of 3,500 niche cells. Transcriptomes were pooled and subjected to clustering analyses using the Seurat package for R to identify genetically distinct clusters of niche cells and changes in these clusters associated specifically with multiple myeloma.

Results: The bioinformatical analyses generated 10 distinct clusters of niche cells, all of which were present in both non-cancer and multiple myeloma bone marrow. One of these clusters contained CDH5+ endothelial cells while the remaining 9 clusters were subsets of CXCL12+LEPR+ mesenchymal stromal cells. Because samples were taken from the central marrow by aspiration, peripheral endosteal or neuronal lineage cells were not represented in these clusters. Gene Set Enrichment Analysis (GSEA) of the stromal cell clusters from myeloma versus non-cancer controls revealed two significantly altered pathways: TNF signaling via NF-kB and Inflammatory response. Detailed analyses of the individual stromal cell clusters identified two clusters that were responsible for the inflammatory changes identified by GSEA. Both clusters were present in all myeloma patients, constituted on average 20% of total stromal cells and were defined by transcription of the inflammatory chemokines CXCL2, CXCL3 and CXCL8 the cytokine IL6. All these transcripts were absent from the equivalent clusters in control bone marrow.

The presence of inflammatory stroma in the multiple myeloma niche indicates either the appearance of a novel stromal cell subset, or activation of pre-existing stromal cells. GSEA analyses suggested inflammatory signaling, and to functionally confirm this, we tested whether activation of stromal cells would induce the inflammatory stromal phenotype. Stimulation of primary human stromal cells in vitro with recombinant TNF was sufficient to induce transcription of CXCL2, CXCL3 and CXCL8, recapitulating the inflammatory transcriptome. Moreover, manual removal of these TNF target genes from the in-silico clustering analyses led to a merging of the inflammatory clusters with non-inflammatory clusters. This indicates that the major distinguishing feature of the myeloma-specific stromal cells are genes induced upon stromal cell activation.

Conclusion: Through single cell transcriptomic analyses we have identified the presence of activated inflammatory stromal cells associated with TNF signaling in the multiple myeloma stromal niche. These inflammatory stromal cells are reminiscent of pathogenic cancer-associated fibroblasts found in solid tumors, where these cells create a pro-tumorigenic niche that favors tumor survival and proliferation while simultaneously inhibiting anti-cancer immunity. These findings represent the first description of myeloma-specific stromal cell subsets, and provide novel cellular targets for interventions aimed at disrupting the pro-tumorigenic microenvironment in multiple myeloma.