Abstract
Introduction: Targeting BCMA has shown promising activity in multiple myeloma (MM). However, durability of responses has been somewhat disappointing. To determine potential mechanisms of resistance to BCMA CAR T cells, we applied Spatiotemporal Genomic Analysis (SaGA) to isolate myeloma cells that are engaged by a CAR T cell but not killed in culture and performed RNAseq on the isolated cells.
Methods: To perform SAGA, Dendra2 was introduced into myeloma cell lines (KMS12PE, KMS18, KMS26, KMS27 and RPMI8226). Cells were cultured with BCMA- or CD19-CAR T cells (1:1) and live imaged for up to 20 h. Killing was measure by Annexin V staining. Cells that were engaged for 16 hours and alive, were photoconverted and sorted for RNAseq. Differentially expressed genes were compared to untreated cells. Gene sets from these data as well as from differentially expressed genes on day 28 post CAR T infusion (Dhodapkar, BCD, 2022) were then used to find pre-treatment cells, indicating pre-existing cells with a resistant gene signature.
Results: Initial studies demonstrated that the average time of engagement that led to cell death in the 5 cell lines ranged from 5.48-9.42 h. However, in each line a proportion of cells survived until the end of the incubation (mean 24.7%, range 3.8-42.9%) and the average engagement of these cells was significantly longer in 4/5 cell lines, suggesting this was not lack of engagement but lack of killing. To assure this was due to antigen specific engagement, we repeated the experiment with CD19 CAR T cells in KMS18 and KMS12PE and the total time of engagement of live cells was reduced from 12.49 and 14.63 h to 5.39 and 2.74 h respectively. Importantly with BCMA CAR T the average and total time of engagement was nearly identical for all cells, indicating that once engaged, the T cell did not disengage. In contrast for CD19 CAR T the average time of engagement was less than 30 min, suggesting the interactions were transient. To determine if engagement time could be influenced by resistance, we created KMS18 cells that lacked CD95 and over-expressed the Granzyme B inhibitor Serpin B9. When exposed to BCMA CAR T, the viability at the end of 16 hours increased from 28.8% to 88.3%. This was associated with an increase in the proportion of cells engaged for 16 hours (53.3% vs. 83.0%). We then performed 4 independent isolations of photoconverted cells from KMS12PE and KMS26 cells for RNAseq analysis. We confirmed there was no T cell contamination. We determined that in KMS12PE and KMS26 there were 1101 commonly upregulated genes (padj<0.01 and FC>2) and top gene ontology hits included “integrin signaling” “WNT signaling” and “Inflammation by chemokine and cytokine signaling.” The latter was driven by upregulation of IFNγ-responsive genes (e.g., CXCL10, GBP1, GBP2, GBP4, IL6). Expression of IFNg receptors was confirmed by flow cytometry and IFNginduction of genes validated by qPCR and/or flow cytometry. There were only 3 commonly down regulated genes and only 1 coding. This was due to the low number of significantly down-regulated genes in KMS26, therefore we focused on KMS12PE. The down-regulated genes were associated with cell death (e.g., TNFRSF10B, PMAIP1, BBC3) and unfolded protein response (e.g., ERN1, ATF4, DDIT3). Downregulation of these genes was previously reported in residual cells 28 days following BCMA CAR T infusion (Dhodapkar, BCD, 2022) and during in vitro selection of cells resistant to BCMA T cell engagers (Lee, ASH, 2024). In these studies the selection time was 4 weeks, in this present study it was 16 h, suggesting these cells were pre-existing. To determine if similar cells existed in patients pre-CAR T infusion, we re-analyzed the scRNAseq comparing Days 0 and 28 post-infusion cells, determined the top 100 down-regulated genes and if there were cells at day 0 where gene expression correlated (>0.8). We found that a subset of Day 0 cells in each patient had a day 28-like expression pattern. We performed a similar analysis using the top 100 down-regulated genes from the resistant KMS12PE (correlation >0.85) and the findings reflected the day 28 signature.Conclusions: These data indicate that MM cells that are resistant to BCMA-targeted therapies are likely to be pre-existing in patients and have a gene expression pattern associated with decreased UPR and cell death signaling. These cells can be detected by single cell analysis and could be used to inform treatment decisions.
