Single Cell Multi-Omic Analysis of Neoplastic Plasma Cells across the Disease Spectrum Identifies Novel Pathobiologic Mediators and Potential Therapeutic Targets in Multiple Myeloma (MM)

Background: A challenge to developing a full understanding of MM pathobiology using bulk profiling techniques is the large degree of intra-tumoral heterogeneity from minor subclones that, despite their low representation, nonetheless contribute to therapy resistance and disease progression. Moreover, our current therapies tend to be targeted against pathways common to both neoplastic and normal plasma cells (PCs), indicating a need to define novel tumor biology modifiers that could potentially serve to improve risk stratification and as therapeutic targets. Methods: To overcome these limitations, we simultaneously evaluated individual PC single-cell transcriptome and B-cell receptor variable (VDJ) sequences from freshly enriched bone marrow biopsies. These included 33 patients with precursor conditions, 17 each with newly diagnosed symptomatic or relapsed/refractory disease (RRMM), and 3 controls using the 10x Genomics platform. A total of ~320,000 PCs were analyzed, yielding median per sample cell counts of 4,626 and ~300 million total reads. Identical immunoglobulin VDJ sequences defined monoclonal PCs, while diverse VDJ sequences at low frequencies defined polyclonal cells, allowing each patient to serve as their own control. Results: The percentage of monoclonal PCs in active MM was >84%, while precursor stages were more variable (40-100% in smoldering MM; 0.3-82% in monoclonal gammopathy of undetermined significance (MGUS)). Though we could not identify monoclonal PCs in two MGUS cases, we identified them by scRNA-seq in four MGUS patients whose bone marrow showed no clonal cells by standard clinical multiparametric flow cytometry. In differential gene expression analysis where samples were grouped based on the diagnosis, 501 dysregulated genes were identified in monoclonal PCs from symptomatic MM compared to polyclonal PCs. These included genes previously identified as disease modifiers, including Hepatocyte growth factor, Lactate dehydrogenase A, Dickkopf WNT signaling pathway inhibitor 1, CD27, and Lysosomal associated membrane protein family member 5, providing validation for our approach. Next, we identified 279 genes that were commonly dysregulated in monoclonal PCs from symptomatic vs. asymptomatic MM. We also performed unsupervised hierarchical clustering of the top 1000 most variable genes and defined three diagnostic categories. Significant ontology terms characterizing these categories broadly involved antigen processing and recognition in Group 1 (enriched in precursor stages); transcription and translation control in Group 2 (enriched in active MM); and cell division and DNA metabolism in Group 3 (enriched with RRMM) ( A). We focused on dysregulated targets that: showed differential expression in monoclonal vs. polyclonal PCs; expression in most patients within a diagnostic category; a correlation with outcomes in the CoMMpassstudy; and novelty in MM. Among these were Mitotic arrest deficient 2 like 1 (MAD2L1), a component of the mitotic spindle assembly checkpoint, and Methionine adenosyltransferase 2A (MAT2A), which catalyzes the production of S-adenosylmethionine, a key methyl donor in multiple cellular processes, from methionine and ATP. Notably, MAD2L1 inhibition with M2I1, and MAT2A targeting with the allosteric inhibitor AG-270, both produced a dose- and time-dependent inhibition in cell proliferation and reduction in myeloma cell line viability. Importantly, the effects of AG-270 ( B) were associated with induction of markers of apoptosis, including Caspase cleavage, and a reduction of Histone H3 dimethylation. Moreover, they occurred at clinically relevant concentrations based on the known pharmacokinetics of this drug in Phase I testing. Conclusions: Our combined single-cell transcriptomics and VDJ sequencing allowed a greater understanding of heterogeneity among neoplastic PCs compared to each patient's own polyclonal PCs. These analyses can identify putative novel mediators of disease pathobiology that impact prognosis, and that can serve as potential new therapeutic targets using either novel agents, or drug repurposed from other therapeutic areas, setting the stage for their translation to the clinic.