Defining the Differentiation States of Multiple Myeloma at Single Cell Resolution Reveals Opportunities for Immunotherapy

Introduction: Despite recent advances in the treatment of multiple myeloma, responses may be short-lived and therapeutic resistance develops almost invariably. Non-genetic cellular plasticity and dedifferentiation have recently emerged as a basis for therapeutic resistance in cancer as cells acquire transcriptional states which no longer depend on the drug target. Therefore, a better understanding of plasticity and adaptive state changes in myeloma cells is critical to develop effective therapeutic approaches that can overcome drug resistance. Here we show that cellular plasticity, though frequently invoked as a basis for therapeutic resistance in cancer, can also lead to new therapeutic opportunities.

Methods: To define transcriptional states in myeloma at a single cell level, we performed fluorescence activated cell sorting and full-length single-cell RNA sequencing. We assayed a total 6000 CD38+CD138+ plasma cells and CD45+ immune cells from the bone marrow of 8 patients with relapsed and refractory multiple myeloma (RRMM) before and after immuno-modulatory treatment on a clinical trial with elotuzumab, pomalidomide, bortezomib and dexamethasone (Elo-PVD; NCT02718833) and 2 healthy donors. Surface expression of selected markers was validated by flow cytometry.

Results: Assessing pre-treatment samples, we discovered that the transcriptional states of single myeloma cells are highly distinct between individual patients, despite the presence of the same established genomic classifiers, such as t(11;14). Furthermore, distinct transcriptional states co-exist within individual patients, indicating there is substantial inter- and intra-individual heterogeneity. Transcriptional states diverge from normal plasma cells towards more immature cells, of the B lymphoid lineage, suggesting a substantial cellular plasticity. Notably, we detected co-expression of myeloid and lymphoid developmental programs in the same single cells. Interestingly, these altered differentiation states were associated with up-regulation of potential immunotherapeutic targets, such as CD20, CD19, and CD33, indicating that this plasticity may result in novel therapeutic vulnerabilities.

To define gene-regulatory relationships, we identified a shared core regulatory network present in malignant and normal plasma cells with the active transcription factors XBP1, ATF4, and CREB3, suggesting that myeloma cells retain lineage-specific regulons. However, we further identified patient-specific regulons not detected in any of the mature immune cell populations assayed, such as TEAD4, ELF3 and SNAI1, illustrating an aberrant and promiscuous activation of transcriptional regulators in myeloma cells. Consistent with this finding, we observed an increased number of expressed genes in myeloma cells compared to normal plasma cells as well as an increase in single cell transcriptional entropy, measures that have been linked to cell potency in normal development and cancer. Comparison of pre- and post-treatment samples interestingly revealed a further increase in transcriptional diversity and signatures associated with stemness and developmental potential following treatment.

Conclusions: In conclusion, we find that higher transcriptional diversity and activation of alternate gene regulatory programs facilitate the emergence of altered transcriptional states. Interestingly, these altered states are associated with up-regulation of putative immune-therapeutic targets in myeloma cells, thus providing novel therapeutic vulnerabilities.