Functional Genomic Characterization of Endoplasmic Reticulum-Associated Dependencies in Multiple Myeloma - Biologic and Therapeutic Implications

Background: Endoplasmic reticulum (ER)-associated degradation (ERAD) for unfolded proteins represents an important biological vulnerability for multiple myeloma (MM) cells, given the high proteostatic stress in these cells due to immunoglobulin secretion. This helps explain the efficacy of proteasome inhibitors in MM compared to most other neoplasms.

Aim: To systematically investigate ER-associated genes that mediate the essential role of ERAD in MM.

Methods: We integrated results from (i) genome-scale studies for CRISPR-gene knockout in 19 MM cell lines (Avana library) and CRISPR-activation (Calabrese library) in 5 MM lines, to define ER-associated dependencies and drivers of increased MM cell growth; (ii) publicly-available CRISPR-based studies for genes regulating ERAD of substrates located in the luminal (ERAD-L), membrane (ERAD-M) or cytosolic (ERAD-C) side of the ER in non-MM models; (iii) functional characterization of candidate genes in MM models in vitro and in vivo.

Results: While ER function in general is globally required for all normal and malignant cells, a surprisingly large proportion of ER-related genes are not "pan-essential" genes (e.g. CERES scores <-0.4 are for <80% of human cell lines): in fact CRISPR-mediated disruption of many ER genes elicits no growth disadvantage for most non-MM tumor cell lines. Notably, though, several of these latter non-"pan- essential" genes are recurrently essential for MM cell lines, or even preferential dependencies for MM compared to other cancers. By overlaying the patterns of CRISPR essentiality in MM lines vs. the functional modules of ER function, we observe that recurrent or preferential ER-associated dependencies for MM cells include genes involved in (a) ER membrane protein complexes mediating dislocation of misfolded ER proteins to the cytosol (e.g. HERPUD1, SEL1L, AUP1 for ERAD-L substrates; and AMFR and RNF139 for ERAD-M and -C substrates,) and associated ER-specific E2/E3 enzymes (SYVN1, UBE2J1, UBE2J2, and UBE2G2); (b) upstream of the ER-to-cytosol dislocation complexes, several enzymes (e.g. DPM1, DPM3, ALG3, ALG9, ALG12, ALG6, ALG8) required for N-glycan-dependent surveillance of quality control for luminal ER glycoproteins; (c) chaperones (e.g. DNAJB11, DNAJBC3) for BiP complexes with misfolded proteins; (d) the known ER stress-sensor IRE1a (ERN1) and its downstream transcription factor XBP1. CRISPR-activation of nearly all these recurrent/MM-preferential ER-associated genes led to no significant growth advantage for MM cell lines. Several of the MM-preferential ER-associated genes (HERPUD1, SEL1L, SYVN1, UBE2J1) are expressed at significantly higher levels in MM vs. non-MM cell lines (CCLE) or patient derived samples (CoMMpass study vs. TCGA) and are proximal to binding sites of key transcription factors for plasma cell identity such as IRF4 and PRDM1. By contrast, CRISPR-mediated LOF of such genes in MM cells decreases their growth in vitro (e.g. competition assays vs. CRISPR LOF of non-essential genes not expressed in MM cells) and in vivo for MM cells growing in a bone marrow-like scaffold system with a "humanized" mesenchymal bone marrow stromal cell compartment. Knockout of MM-preferential ER-associated genes induces markers of ER stress/unfolded protein response (e.g. BiP induction by UBE2J1 or HERPUD1 LOF); and sensitizes MM cells to proteasome inhibition. Accordingly, pharmacological inhibition (LS-102) of SYVN1 (Synovolin-1) an E3 ligase which helps remove unfolded proteins from the cytosol into the ER was more effective at a lower drug concentration in MM compared to non-MM-cell lines.

Conclusions: Our results provide functional insights into the role of ERAD in MM cells and nominate specific mediators which are not broadly essential across all cancers, but are recurrent or even preferentially essential for MM. Our study indicates a previously underappreciated role for the ER-to-cytosol dislocon complex that includes HERPUD1, SEL1L, SYVN1 and UBE2J1; distinct enzymes of the N-glycan-dependent ER glycoproteins quality control system; and ER chaperones in MM. These findings provide a rationale for development of agents selectively targeting these key regulators of ER protein folding processes in MM.