Integrated Systems Level Examination of Proteasome Inhibitor Stress Recovery in Myeloma Cells Reveals Druggable Vulnerabilities Linked to Multiple Metabolic Processes

Proteasome inhibitors (PIs) form the backbone of multiple myeloma (MM) treatment regimens used at diagnosis and relapse, but their clinical benefit is limited by varying degrees of resistance. The mechanisms that protect MM cells (MMCs) from PI-induced cell death are only partly understood, and experimental resistance studies often rely on the prolonged exposure of MMCs to relatively low levels of PIs. However, patients receive PI doses that result in high but short plasma peaks. Given that each PI dose reduces a patient's MMC load until a plateau is reached, one can assume that each PI dose kills some MMCs by triggering overwhelming stress. This proposition implies that a proportion of MMCs not only survive but also resolve PI-induced stress. We hypothesised that the mechanisms of PI-induced stress recovery require a redistribution of cellular resources, and that this process triggers specific recovery-associated and potentially druggable vulnerabilities. To test this hypothesis, we obtained a global systems-level view of the cellular response to proteasome inhibition by performing a multi-omics 10-day time-course experiment. To mimic in vivo pharmacokinetics and anti-MM effects, RPMI-8226 MMCs were treated with a 1h pulse of carfilzomib (CFZ; 750nmol/L), which reduced the number of viable MMCs to a nadir of approximately 50% on day +2, followed by a return to pretreatment baseline levels of viable cells by day +6. Levels of ubiquitinated proteins as a readout of effective proteasome inhibition peaked on day +1 and returned to baseline levels on day +6. Samples for transcriptome analysis by RNA-seq, proteome analysis by a multiplexed tandem mass tag (TMT)-based approach, and global metabolomic profiling by ultrahigh performance liquid chromatography-tandem mass spectroscopy (LC-MS) were collected 2h prior to the CFZ pulse (day 0) and on days +1, +2, +4, +6, +8, and +10. The results demonstrate extensive and kinetically complex changes of the MMC transcriptome, proteome and metabolome that did not fully return to baseline by day +10. Conventional and advanced computational analyses of RNA-seq data (including biological homogeneity score-validated graph-based clustering of temporal patterns combined with gene enrichment and KEGG/GO pathway analysis) revealed different patterns of involvement of multiple and functionally diverse pathways. Perhaps most notably, RNA-seq data revealed extensive metabolic pathway responses while cells were dying and throughout recovery. Consistently, metabolomic profiling by LC-MS showed wide-ranging metabolite changes. These included rapid intracellular depletion of glucose, which was paralleled by the accumulation of lactate and pyruvate and lasted until day +8. Consistently, the major glucose uptake regulator TXNIP stood out as upregulated during recovery, which was accompanied by the predicted downregulation of the glucose transporter GLUT1. These findings were confirmed by real-time PCR and immunoblotting in RPMI-8226 and other MMC lines. Moreover, pharmacological inhibition of glucose and lactate metabolising enzymes and transmembrane transport enhanced MMC killing by CFZ, confirming profound and druggable changes in glucose and energy metabolism during PI stress recovery. We also observed prolonged depletion of a number of amino acids, including glutamine, which was paralleled by depletion of glutamate and TCA cycle intermediates. Moreover, we found RNA-seq and real-time PCR/immunoblot-based evidence for an amino acid response (AAR) during stress recovery that was driven by the amino acid sensing kinase GCN2 (EIF2AK4). The preclinical GCN2 inhibitor GCN2iB was effective at blocking the AAR and enhanced MMC killing in a panel of CFZ-treated MMC lines. This effect was particularly pronounced during the intermediate stages of recovery, in line with the kinetics of amino acid depletion and the AAR. Thus, our observations define the complex and prolonged cellular responses that characterise the recovery of MMCs from proteasome inhibition. They reveal wide-ranging metabolic changes that persist far beyond immediate stress survival and highlight the dependence of MMCs on the GCN2-driven AAR to overcome PI stress. The findings demonstrate that the mechanisms of PI recovery trigger druggable vulnerabilities, providing the basis for ongoing investigations into sequential therapeutic interventions to enhance responses to PIs.