Functional Multi-Omics Reveals Genetic and Pharmacologic Regulation of Surface CD38 in Multiple Myeloma

Abstract

Background: CD38 is a surface ectoenzyme expressed at high levels on myeloma plasma cells and is the target for the monoclonal antibodies (mAbs) daratumumab and isatuximab. These antibodies have multiple mechanisms of action, primarily involving recruiting and modulating components of the immune system, but they may also carry direct anti-tumor effects. CD38 density on tumor cells is an important determinant of mAb efficacy while CD38 is lost after mAb treatment. Several small molecules have been found to increase tumor surface CD38, with the goal of boosting mAb efficacy in a co-treatment strategy. However, we do not yet have a broad global sense of the transcriptional or post-transcriptional networks that most strongly impact CD38 expression. There may be alternate strategies to even more potently increase CD38 expression that have not yet been identified. Furthermore, prior clinical studies showed that CD38 downregulation after daratumumab treatment was accompanied by increases in the complement inhibitors CD55 and CD59. Are there other features of myeloma surface remodeling driven by CD38 downregulation? Here we sought to extend our currently limited insight into CD38 surface expression by using a multi-omics approach. Methods: Genome-wide CRISPR interference screening was performed in RPMI-8226 cells stably expressing the dCas9-KRAB fusion protein. Cells were grown for 14 days after library transduction, flow-sorted on the top and bottom 25% of CD38 surface expression, and sgRNA's deep sequenced. Antibody-dependent cellular cytoxicity assays were performed with NK92-CD16 cells. Cell surface proteomics was performed using N-glycoprotein cell surface capture in triplicate. Phosphoproteomics was performed used immobilized metal affinity chromatography in triplicate. Murine studies were performed in NSG mice under approved IACUC-approved institutional protocols. Results: A genome-wide CRISPR-interference screen in RPMI-8226 cells demonstrated that transcriptional and epigenetic factors played the most prominent role in surface CD38 regulation (Fig. 1A). One of the genes that when knocked down led to greatest surface CD38 increase was RARA. This finding supports the promise of all-trans retinoic acid (ATRA), which leads to RARA degradation, as a potent agent to induce CD38 upregulation. Validation of additional screen hits TLE3 and HEXIM1 also illustrated that these negative regulatory transcription factors suppress CD38 expression at baseline (not shown). We found the transcription factor SPI1 to be a prominent positive regulator of CD38. SPI1 knockdown led to daratumumab resistance both in vitro and in vivo, similar to the resistance observed after CD38 knockdown (not shown). Analysis of myeloma patient ATAC-seq data, assessing transcription factor motifs present at the CD38 locus, combined with a predictive machine learning model, further identified XBP1 as one of the most potent transcriptional regulators of CD38 (Fig. 1B). We next used "antigen escape profiling" - knockdown of CD38 followed by unbiased cell surface proteomics - to mimic surface alterations in the context of CD38 loss. We found minimal changes in other cell surface proteins beyond CD38 (Fig. 1C), indicating the CD38 loss alone is not sufficient to remodel the myeloma surfaceome. This finding also supports the hypothesis that complement or other immune system interactions are necessary to lead to other myeloma surface protein alterations. In a parallel analysis of pharmacologic regulation, we also used cell surface proteomics integrated with RNA-seq to demonstrate that ATRA leads to few other surface protein changes beyond CD38 (not shown). In contrast, other molecules, such as azacytidine and panobinostat, led to broader changes across many more surface proteins, showing a lack of specificity when driving CD38 upregulation. Finally, unbiased phosphoproteomics revealed partial inhibition of the MAP kinase pathway after daratumumab binding (Fig. 1D). This result may comprise a direct anti-proliferative effect of anti-CD38 therapeutic antibody engagement in myeloma. Conclusions: Our work provides a resource to design strategies to enhance efficacy of CD38-targeting immunotherapies in myeloma. Our approach also outlines a broad multi-omic strategy to evaluate surface and transcriptional regulation of other key immunotherapeutic targets in hematologic malignancies. Figure 1 Figure 1. Disclosures Choudhry: Genentech: Current Employment, Current equity holder in publicly-traded company. Ramkumar: Senti Biosciences: Current Employment, Current holder of individual stocks in a privately-held company.

Targeting high-risk multiple myeloma genotypes with optimized anti-CD70 CAR-T cells

“…These findings encouraged us to search for other epigenetic and transcriptional regulators of CD70 in MM. We used a strategy previously described by our group ( 41 ), analyzing both primary MM sample ATAC-seq data ( 42 ) and ENCODE ChIP-seq data, to together predict transcription factors (TFs) that may bind to the CD70 locus ( Fig. S6B ).…”

“…We then took this list of 84 transcription factors and correlated TF expression with CD70 expression across all 776 patient tumor samples in CoMMpass. Using these correlations, we then built a machine learning classifier via eXtreme Gradient Boosting (XGBoost) ( 41 ). This model enabled us to determine the relative contribution of expression of each TF with respect to CD70 expression.…”

“…We developed an XGBoost model (Extreme Gradient Boosting) using the xgboost package in Python. The code used for our analysis is available at our repository here (https://github.com/BonellPatinoE/CD70-predictor-in-Myeloma.git) and the approach is similar to that we previously used for analysis of CD38 and CD46 regulators ( 41, 55 ) in MM except here applied to CD70 . All transcription factors determined to be co-expressed were utilized together to build the model.…”

“…We also checked the potential motif binding in the CD70 promoter regions. We predicted the motif binding in the identified ATAC peak regions from MM patients Myeloma.git) and the approach is similar to that we previously used for analysis of CD38 and CD46 regulators(41,55) in MM except here applied to CD70. All transcription factors determined to be co-expressed were utilized together to build the model.…”

“…Myeloma.git) and the approach is similar to that we previously used for analysis of CD38 and CD46 regulators(41,55) in MM except here applied to CD70. All transcription factors determined to be co-expressed were utilized together to build the model.…”