Proteasome Inhibitor-Induced Modulation Reveals the Spliceosome As a Specific Therapeutic Vulnerability in Multiple Myeloma
Total Page：16
File Type：pdf, Size：1020Kb
ARTICLE https://doi.org/10.1038/s41467-020-15521-4 OPEN Proteasome inhibitor-induced modulation reveals the spliceosome as a specific therapeutic vulnerability in multiple myeloma Hector H. Huang 1, Ian D. Ferguson1, Alexis M. Thornton2, Prabhakar Bastola 1, Christine Lam1, Yu-Hsiu T. Lin 1, Priya Choudhry 1, Margarette C. Mariano1, Makeba D. Marcoulis1, Chin Fen Teo3, Julia Malato4, Paul J. Phojanakong4, Thomas G. Martin III4,5, Jeffrey L. Wolf4,5, Sandy W. Wong4,5, Nina Shah4,5, ✉ Byron Hann4, Angela N. Brooks 2 & Arun P. Wiita 1,4 1234567890():,; Enhancing the efficacy of proteasome inhibitors (PI) is a central goal in myeloma therapy. We proposed that signaling-level responses after PI may reveal new mechanisms of action that can be therapeutically exploited. Unbiased phosphoproteomics after treatment with the PI carfilzomib surprisingly demonstrates the most prominent phosphorylation changes on splicing related proteins. Spliceosome modulation is invisible to RNA or protein abundance alone. Transcriptome analysis after PI demonstrates broad-scale intron retention, suggestive of spliceosome interference, as well as specific alternative splicing of protein homeostasis machinery components. These findings lead us to evaluate direct spliceosome inhibition in myeloma, which synergizes with carfilzomib and shows potent anti-tumor activity. Functional genomics and exome sequencing further support the spliceosome as a specific vulnerability in myeloma. Our results propose splicing interference as an unrecognized modality of PI mechanism, reveal additional modes of spliceosome modulation, and suggest spliceosome targeting as a promising therapeutic strategy in myeloma. 1 Department of Laboratory Medicine, University of California, San Francisco, CA, USA. 2 Department of Biomolecular Engineering, University of California, Santa Cruz, CA, USA. 3 Department of Physiology, University of California, San Francisco, CA, USA. 4 Helen Diller Family Comprehensive Cancer Center, ✉ University of California, San Francisco, CA, USA. 5 Department of Medicine, University of California, San Francisco, CA, USA. email: [email protected] NATURE COMMUNICATIONS | (2020) 11:1931 | https://doi.org/10.1038/s41467-020-15521-4 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15521-4 ultiple myeloma (MM) is a clonal malignancy of plasma 24 h for analysis based on our prior results demonstrating how Mcells with no known cure. Like normal plasma cells, the proteomic response to PI evolves over many hours9. This is in myeloma cells produce and secrete extremely high contrast with most phosphoproteomic studies, examining direct amounts of immunoglobulin. This unique function may be kinase effects on a timescale of minutes14. Here we instead con- exploited by therapeutically inhibiting the proteasome using the sider the indirect effects on phosphorylation induced by PI Food and Drug Administration-approved proteasome inhibitors exposure. Using label-free quantification of immobilized metal (PIs) bortezomib, carfilzomib (Cfz), and ixazomib. Proteotoxic affinity chromatography-isolated phosphopeptides, we found that stress caused by these first-line therapeutic agents has been pro- altered phosphorylation signatures were most prominent 24 h posed to induce the apoptotic function of the unfolded protein after treatment (Fig. 1 and Supplementary Fig. 1). In total, we response (UPR)1, leading to plasma cell death while largely quantified 5791 phosphosites in at least one technical replicate of sparing normal tissues2,3. However, despite the appealing sim- the timecourse, with >99% of phosphosites representing Ser or plicity of this mechanism, the canonical UPR is not always Thr phosphorylation events, as expected using this enrichment strongly induced in myeloma cells by PIs4 and is unlikely to be technique. the sole mode of PI cytotoxicity in MM. Indeed, many additional At each timepoint, we simultaneously performed single-end mechanisms of action of PIs have also been proposed, ranging RNA-seq for gene expression (Supplementary Data 2). Figure 1b from nuclear factor (NF)-κB inhibition to immune micro- environment effects to aberrant recycling of cytosolic amino 5,6 acids . a 1.Lysis IMAC Identifying the full range of PI mechanisms of action remains MM.1S 2.Reduce/ enrich phospho relevant given that acquired PI resistance is clinically widespread Add Cfz alkylate P 7,8 P but its origins remain unclear . Finding new methods to spe- P Harvest 0 h P cifically target PI-resistant disease, or molecules to synergize with ... 3.Trypsinize ... 8 h PIs to avoid resistance by driving deeper remissions, remains a (25 °C, 22h) Desalt ... 16 h long-standing goal. As one approach to achieving this goal, we peptides 24 h and others have studied the response of malignant plasma cells to 9–11 PIs using both gene expression and proteomic methods . P Notably, one of the most prominent features of the cellular mRNA Phospho- P P 12 peptides P P response to PIs is the activation of the heat shock response . This cDNA P mechanism leads to significant induction of cytosolic protein- folding chaperones, possibly to assist in protein refolding and decrease in unfolded protein stress. We and others9,12,13 have therefore proposed targeting mediators of the heat shock response RNA-seq LC-MS/MS as potential combination therapies with PIs. m/z However, one unresolved question is whether proteasome inhibition may carry additional effects on plasma cells that are b Log [8h,16h,24h]/0h not revealed by mRNA or protein abundance analysis alone. We Phospho RNA- 2 seq hypothesized that additional modalities of response, and thereby Rep1 Rep2 new myeloma-relevant therapeutic targets, may be revealed by 8 16 24 8 16 24 8 16 24 –6–4 –2 0462 studying the signaling-level response to PIs with unbiased mass RPS6 spectrometry-based phosphoproteomics. The large majority of RPS6 Phospho RNA- seq therapy-relevant investigations using this technique have focused EIF4E Rep1 Rep2 14 -BP1 on elucidating the effects of kinase inhibitors . However, we TRA2B 8 16 24 8 16 24 8 16 24 reasoned that a significant cellular perturbation such as protea- SRRM1 some inhibition would likely also indirectly perturb kinase and SRSF2 ZRANB2 phosphatase signaling in a broad fashion. SRSF2 TRA2A Here we use unbiased phosphoproteomics to quantify >5000 SRSF7 THRAP3 phosphopeptides in myeloma cells exposed to the irreversible PI SRSF2 Cfz. Surprisingly, we find the greatest increases in phosphoryla- SRRM1 SRSF7 ZRANB2 tion occur in proteins associated with the spliceosome machinery. SRSF5 SON A link between these processes is invisible at the gene expression TRA2A level. We further evaluate this link from a mechanistic and HNRNPD therapeutic perspective, finding that PIs lead to specific disrup- ZRANB2 PRPF4B tion of normal splicing. We suggest interference of splicing as an HNRNPF additional mechanism of action of PIs not previously explored. SRSF1 SON Inhibition of splicing has recently become a promising ther- CDK12 apeutic strategy in other hematologic malignancies15. Our results RPS6 reveal an intersection of cellular stress and the splicing machin- EIF4E -BP1 ery, which may have broad relevance in biology. Furthermore, we propose the spliceosome as a new and potentially selective ther- Fig. 1 Unbiased phosphoproteomic timecourse analysis of MM.1S cells apeutic target in myeloma. treated with the PI carfilzomib (Cfz). a Workflow of timecourse treatment of MM.1S cells with Cfz. Cells were allotted for both RNA-seq analysis and LC-MS/MS. b Downregulated (left panel) and upregulated (right panel) Results log2-transformed phosphopeptide MS1 intensities for two technical PI treatment leads to splicing factor phosphorylation.Wefirst replicates of proteins with unchanged transcript levels (RNA-seq). Labels used unbiased phosphoproteomics to examine the signaling-level highlight dephosphorylation of RPS6 and EIF4EBP1 on the left and response of MM.1s MM cells to Cfz. We chose timepoints across phosphorylation of splicing-related proteins on the right. 2 NATURE COMMUNICATIONS | (2020) 11:1931 | https://doi.org/10.1038/s41467-020-15521-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15521-4 ARTICLE shows 58 upregulated (red) and 75 downregulated (blue) corrected Gene Ontology (GO) analysis confirms that all of the phosphopeptides from proteins with largely unchanged RNA top enriched biological processes involve RNA splicing regulation transcript abundance as detected by unsupervised hierarchical and mRNA processing (Fig. 2e and Supplementary Fig. 1b). This clustering. We found decreased phosphorylation of the transla- signaling response is much weaker at 10 nM Cfz, with only 25 tion factor EIF4E-BP1 as well as RPS6 (Fig. 1b), as expected upon upregulated phosphosites and none that are splicing related. PI-induced cellular stress9. While other downregulated phospho- These results suggest a strong dose–response effect of phosphor- peptides did not suggest any highly enriched biological function, ylation changes after proteasome inhibition, both across splicing among upregulated phosphosites we were surprised to find that factors and the broader proteome. 14 of the 58 were present on proteins related to pre-mRNA To differentiate changes at the signaling level to changes at the splicing. These primarily included the heterogeneous ribonucleo- protein level, unenriched peptides were also analyzed (Supple- protein (HNRNP) family as well as SRSF splicing