Translation initiation factor eIF4F modifies the dexamethasone response in multiple myeloma

Significance

Multiple myeloma (MM) is a cancer that develops in the bone marrow and remains incurable to this day. It is a cancer type that shows hallmarks of deregulated protein synthesis control. To uncover new vulnerabilities in this disease, we performed a focused RNAi screen to identify components of the translation apparatus that, when depleted, would sensitize tumor cells to dexamethasone (DEX), a component of frontline therapy in this cancer. We found that suppression of eukaryotic initiation factor 4F, a heterotrimeric complex required for cap-dependent translation initiation, is a modifier of the DEX response in MM. Our efforts uncover a previously unidentified vulnerability in MM that should be explored clinically.

Abstract

Enhanced protein synthesis capacity is associated with increased tumor cell survival, proliferation, and resistance to chemotherapy. Cancers like multiple myeloma (MM), which display elevated activity in key translation regulatory nodes, such as the PI3K/mammalian target of rapamycin and MYC-eukaryotic initiation factor (eIF) 4E pathways, are predicted to be particularly sensitive to therapeutic strategies that target this process. To identify novel vulnerabilities in MM, we undertook a focused RNAi screen in which components of the translation apparatus were targeted. Our screen was designed to identify synthetic lethal relationships between translation factors or regulators and dexamethasone (DEX), a corticosteroid used as frontline therapy in this disease. We find that suppression of all three subunits of the eIF4F cap-binding complex synergizes with DEX in MM to induce cell death. Using a suite of small molecules that target various activities of eIF4F, we observed that cell survival and DEX resistance are attenuated upon eIF4F inhibition in MM cell lines and primary human samples. Levels of MYC and myeloid cell leukemia 1, two known eIF4F-responsive transcripts and key survival factors in MM, were reduced upon eIF4F inhibition, and their independent suppression also synergized with DEX. Inhibition of eIF4F in MM exerts pleotropic effects unraveling a unique therapeutic opportunity.

Multiple myeloma (MM) is a bone marrow-derived malignancy of plasma cells that typically produce a monoclonal immunoglobulin (Ig). It is the second most frequent hematological neoplasm in adults, with ∼20,000 newly diagnosed cases annually in the United States. Corticosteroids [prednisone and dexamethasone (DEX)] are common to all treatment regimens, usually in combination with an alkylating agent (e.g., melphalan) or more recently introduced agents, such as proteasome inhibitors (bortezomib and carfilzomib) and immunomodulatory agents (thalidomide, lenalidomide, and pomalidomide). Despite the single-agent activity of these novel treatment options, enhanced clinical activity is observed when they are used in combination with DEX.

The MM genomic landscape includes recurrent DNA translocations involving mostly the IgH locus, chromosomal gains and losses, and a significant number of mutations in genes involved in translation and its regulation (1). In nearly 50% of patients analyzed, mutations were documented in the following: DIS3, a component of the exosome and predicted to increase mRNA content; FAM46C, a gene product functionally related to a regulator of protein synthesis; XBP1, an unfolded response protein linked to translational control; LRRK2, a eukaryotic initiation factor (eIF) 4E binding protein kinase; and, less frequently, eIF3B, rpL10, and rpS6KA1 (1). In addition, several signaling pathways are aberrantly activated in MM. These signaling pathways include the PI3K/mammalian target of rapamycin (mTOR), NF-κB, Ras, Raf, MAPK, and Janus kinase pathways, all of which promote proliferation, evasion of apoptosis, and resistance to therapy (2, 3). Among these signaling pathways, the PI3K/mTOR and MAPK pathways intimately impinge on the translation pathway, linking this energetically demanding process to intra- and extracellular proliferation and survival cues (4). As well, overexpression of MYC, a master regulator of ribosome biogenesis and translation, is frequently observed in MM (5–7) and is associated with a poor prognosis (7, 8). Deregulated translational control is therefore a central feature of MM, with perturbations occurring at the level of core components and regulatory pathways.

One of the best-studied MYC effectors and a downstream target of the PI3K/mTOR pathway implicated in translational control is eIF4F (4, 5, 9). eIF4F is a heterotrimeric complex consisting of: (i) eIF4E, a cap-binding protein; (ii) eIF4A, an RNA helicase implicated in remodeling mRNA structure; and (iii) eIF4G, a large scaffolding protein. The association of eIF4E with the eIF4F complex is regulated by mTOR, with mTOR activation leading to stimulation of eIF4F formation. Elevated PI3K/mTOR signaling flux or increased MYC levels exert profound activating effects on cap-dependent translation, and consequently on cellular proliferation and survival (10).

The PI3K/mTOR axis is being probed as a therapeutic target in MM (11). MM cells are sensitive to mTOR inhibition by rapamycin-related molecules (12), but the presence of an mTOR-S6K-insulin receptor substrate-1 negative-feedback loop, which reactivates PI3K and AKT upon mTOR inhibition, diminishes the efficacy of rapalogs (13). Second-generation mTOR complex (mTORC) kinase inhibitors (KIs) (e.g., OSI-027) and dual-specificity PI3K/mTOR KIs [e.g., (NVP)-BEZ235] avoid PI3K/AKT activation via the aforementioned feedback loop and show higher activity against MM cells than rapalogs (11, 14–16). Rapalogs and PI3K/mTOR inhibitors have also been shown to enhance the cytotoxic effects of DEX (14, 17, 18), which is an effect that can also be obtained by sequestering eIF4E from the eIF4F complex by overexpressing an inhibitory binding partner (18). Herein, we describe the results of a focused RNAi-based screen aimed at identifying druggable targets among components of the translation apparatus to identify DEX synthetic lethal partners in MM. Our results define eIF4F as a target for the treatment of MM, identify a small-molecule inhibitor with low nanomolar potency against human MM cells, and demonstrate that inhibition of translation is synthetic lethal with DEX when applied to MM tumor cells.

Results

RNAi-Based Synthetic Lethal Screen Identifies Modifiers of DEX Sensitivity in MM.

Given the profound deregulation at the level of translation that has been documented in MM (Introduction), we chose to perform an RNAi-based screen targeting this pathway to identify vulnerabilities for potential therapeutic intervention. Because DEX is used as frontline therapy in this disease, we designed the screen to identify potential modifiers of the DEX response (Fig. 1A). Among five human MM cell lines tested for their ability to be infected by our modified pGmP lentivirus, JJN-3, and KMS-11 were the most efficiently and reproducibly infected, with rates attaining 50–60% following one round of infection (Fig. S1A). Whereas KMS-11 cells were sensitive to DEX (IC₅₀ = 50 nM), JJN-3 cells were quite resistant to concentrations as high as 3 μM (Fig. S1B), as previously reported (19). Upon exposure to DEX, JJN-3 cells activated expression of the DEX-responsive glucocorticoid-induced leucine zipper (GILZ) (20), indicating that DEX resistance in JJN-3 cells is not due to defective glucocorticoid receptor (GR) signaling (Fig. S1C). GILZ induction is mediated by DEX in our setting because it is blocked by the antagonist RU-486 (Fig. S1C). These results establish JJN-3 cells as a DEX-resistant cell line with a functional GR/GILZ signaling axis.

Fig. 1.

DEX-dependent synthetic lethal RNAi screen in the JJN-3 cell line. (A) Schematic outline of the DEX-dependent synthetic lethal screen performed in this study. A diagram of the pGmP vector and a time line of the infection and cell-harvesting schedule are presented. Puro, puromycin; SIN, self-inactivating. (B) Categories of genes evaluated in the RNAi screen. The number of genes targeted in each category is indicated in parentheses. DHX, DEAH-box. (C) Pooled synthetic lethal shRNA screen in JJN-3 cells showing changes in overall representation of 1,099 informative shRNAs during 12 d of culture. Depletion of shRNAs was calculated as the relative abundance in vehicle (DMSO)-treated cells divided by the relative abundance in 100 nM DEX-treated cells. Values are plotted as the average of triplicate values for each representation. The locations of the shRNAs against MCL1, firefly luciferase (FLuc), and the scrambled (Scr) controls are indicated. The horizontal dotted line represents three SDs from the population mean. (D) Pie chart illustrating the number and validation outcome of shRNAs identified in the DEX-dependent RNAi screen.

We generated a custom, sequence-verified and arrayed, miR30-based shRNA library targeting amino acyl-tRNA synthetases, large and small ribosomal proteins, initiation factors, elongation factors, termination factors, RNA helicases, and components known to regulate protein synthesis (Fig. 1B and Dataset S1). The library consisted of 1,534 unique shRNAs targeting 268 genes (including controls; SI Text 1 and Fig. S2) cloned into pGmP (Fig. 1A and Dataset S1). The library was transduced into JJN-3 cells and cells cultured for 12 d in the presence of vehicle or DEX. Changes in shRNA representation were determined by deep sequencing of shRNA sense strands amplified from genomic DNA at day 0 [time (T) 0] and day 12 (T12). Each experimental time point was performed on three independent biological replicates, and correlation coefficients between replicates ranged from 0.6 to 0.96 (Dataset S1 and Table S1).

In total, 1,099 shRNAs (72% of the library) were detected above background in all three biological replicates at T0. At T12, we identified 18 potential DEX-dependent synthetic lethal candidates for which the representation of two independent shRNAs was reduced in the presence of DEX (Fig. 1 C and D and Dataset S2). Among these candidates, seven encoded RNA helicases, 10 coded for core components of the translation apparatus, and one encoded a translational regulator (GCN1L1) (Fig. 1D). During the analysis process, we recognized that suppression of all three eIF4F subunits synergized with DEX and focused our validation efforts on the eIF4F cap-binding protein complex (Fig. 1D).

eIF4F Is a Genetic Modifier of the DEX Response in JJN-3 Cells.

Individual testing of shRNAs targeting the three eIF4F subunits confirmed their synthetic lethal relationship with DEX (Fig. 2A). We compared the extent of shRNA depletion in the presence of DEX with the knockdown efficiency of the shRNAs toward their targets and found that the more potent the shRNA, the greater was the sensitization to DEX (Fig. 2 A–C). There are two cellular isoforms of eIF4A and eIF4G that can be incorporated into the eIF4F complex with some potential nonoverlapping roles in vivo (21–23), yet only shRNAs to eIF4A1 and eIF4G1 were identified in the primary screen as being synthetic lethal with DEX (Dataset S2). Testing of shRNAs to eIF4A2 and eIF4G3 indicated that suppression of eIF4A2 or eIF4G3 is tolerated in JJN-3 cells and is not synthetic lethal with DEX (Fig. S3). Taken together, our data implicate eIF4E, eIF4A1, and eIF4G1, but not eIF4A2 or eIF4G3, as potential modifiers of DEX sensitivity in JJN-3 cells.

Fig. 2.

Suppression of eIF4F subunits sensitizes JJN-3 cells to DEX-induced cell death. (A) Changes in the representation of JJN-3 cells infected with the indicated pGmP-shRNA constructs in the presence of vehicle (DMSO, Upper) or 100 nM DEX (Lower) over the course of 12 d (n = 3). Error bars represent ± SEM. (B) Immunoblots of whole-cell lysates of puromycin-selected JJN-3 cells infected with the indicated shRNAs. (C) Box plot comparing eIF4F subunit knockdown efficiency with DEX sensitization in JJN-3 cells.

Pharmacological Targeting of eIF4F Sensitizes MM Cells to DEX.

We sought to determine if we could phenocopy the sensitization of JJN-3 cells to DEX obtained by RNAi-mediated suppression of eIF4F using small-molecule inhibitors. Efforts toward “drugging” the eIF4F complex have identified a number of compounds capable of blocking eIF4E–cap interaction, interdicting eIF4E–eIF4G association, or inhibiting eIF4A-mediated helicase activity (4) (Fig. 3A). JJN-3 cells are exquisitely sensitive to the eIF4A helicase inhibitor silvestrol, displaying an IC₅₀ <10 nM (Fig. 3B and Fig. S4A). In comparison, the IC₅₀ of a second unrelated eIF4A inhibitor, hippuristanol (Hipp), was ∼100 nM (Fig. 3B). The eIF4E–eIF4G interaction inhibitors, 4E1RCat (IC₅₀ of ∼6 μM) and 4EGI-1 (IC₅₀ of ∼20 μM), were the least active of the compounds tested (Fig. 3B). For comparison, we assessed JJN-3 sensitivity to the mTORC KI, OSI-027, and the dual-specificity PI3K/mTOR KI, NVP-BEZ235, and found that JJN-3 cells were more sensitive to NVP-BEZ235 (IC₅₀ of ∼20 nM, compared with the IC₅₀ for OSI-027 of ∼500 nM), none of which displayed the potency obtained with silvestrol (Fig. S4B).

Fig. 3.

Pharmacological inhibition of eIF4F synergizes with DEX in JJN-3 cells. (A) Schematic diagram illustrating the sites of eIF4F inhibition by compounds used in this study. 4EGI-1 and 4E1RCat inhibit eIF4E–eIF4G interaction, Hipp prevents eIF4A from binding RNA, and silvestrol leads to depletion of eIF4A from the eIF4F complex (4). (B) Pharmacological inhibition of eIF4F in MM. Cells were exposed to increasing concentrations of compound for 48 h, and viability was determined (n = 3). Error bars represent ± SEM. (C) DEX-dependent synergy with silvestrol (Upper) or NVP-BEZ235 (Lower) in JJN-3 cells. CI, combination index. (D) Primary human MM cells are sensitive to silvestrol (Sil). Primary tumor cells were exposed ex vivo to the indicated concentrations of silvestrol for 24 or 48 h, after which point cell viability was determined by flow cytometry as indicated in Experimental Procedures. Presented is the percentage of the CD138⁺ cell population surviving following drug exposure relative to DMSO controls. (E) Cells from patients whose tumor was refractory to DEX are resensitized by silvestrol ex vivo. Primary tumors cells were exposed ex vivo to the indicated concentrations of DEX (nanomolar), silvestrol (nanomolar), or a combination of both for 48 h, at which point cell viability was determined by flow cytometry. Presented is the percentage of the cell population surviving following drug exposure relative to DMSO controls (n = 3). Error bars represent ± SEM.

A series of extended titrations combining eIF4F inhibitors with DEX revealed significant synergy (combination index value lower than 0.25) for the DEX/silvestrol combination at concentrations of silvestrol as low as 0.4 nM and over a 2.5-log₁₀ concentration range (Fig. 3C). NVP-BEZ235 also synergized with DEX, but not as profoundly as silvestrol over the concentrations tested (Fig. 3C). Strong synergy was also observed between Hipp and DEX at concentrations of Hipp ≥25 nM, moderate synergy was noted with OSI-027 and DEX over a wide concentration range, and 4E1RCat and 4EGI-1 synergized with DEX at concentrations between 50 and ∼750 nM (Fig. S5). These results were not a peculiarity of JJN-3 cells, because we reproduced our findings in a different MM cell line, KMS-11 (SI Text 2 and Fig. S6). Taken together, these results indicate that silvestrol, an inhibitor of eIF4F helicase activity, shows nanomolar potency as a single agent and synergizes with DEX in MM cells.

We also assessed silvestrol’s activity against primary patient-derived myeloma samples and observed significant depletion of CD138⁺ plasma cells following 48 h of exposure to 50 nM silvestrol ex vivo (Fig. 3D). We also assessed what effect silvestrol and DEX would have against primary tumor cells that no longer responded to DEX (Fig. 3E). Our results indicate that silvestrol as a single agent selectively affects CD138⁺ cell viability and can resensitize DEX-resistant CD138⁺ cells to DEX ex vivo. These results were not due to a general cytotoxic property of silvestrol, because the CD138⁻ cell population remained unaltered upon drug exposure (Fig. 3E). Three nontransformed human cell lines, as well as normal peripheral blood mononuclear cells, were found to be relatively resistant to silvestrol and DEX, with little to no DEX/silvestrol synergy noted (Fig. S7). Taken together, these results identify silvestrol as a potent single agent against MM cells and as being capable of resensitizing MM cells to DEX.

Targeting eIF4F Acutely Curtails Production of Myeloid Cell Leukemia 1 and MYC in MM.

The translation of mRNAs with elevated secondary structure within their 5′ UTRs is sensitive to alterations in eIF4F activity (4, 24). Accordingly, silvestrol has been documented to preferentially inhibit the translation of mRNAs with elevated 5′ UTR secondary structure (25, 26). We therefore investigated the silvestrol response of several known eIF4F-responsive transcripts [MYC, myeloid cell leukemia 1 (MCL1), BCL2, and BCLX_L] encoding products key to tumor cell survival (16, 27, 28). A significant reduction in MCL1 and MYC protein levels was observed in JJN-3 cells exposed to a short-term pulse (5 h) of silvestrol, Hipp, and 4E1RCat (Fig. 4A). This response correlated with a reduced association of MCL1 and MYC mRNA with polysomes in silvestrol-treated cells (Fig. 4B), with little change in mRNA levels (Fig. S8A). The DEX/silvestrol combination did not reduce MCL1 protein levels further beyond those observed with silvestrol alone (Fig. S8B, compare lanes 3 and 4), indicating that DEX, per se, is not influencing MCL1 levels. Cell death was not observed in JJN-3 cells within the experimental time period and at the concentrations of silvestrol used (Fig. S8C). In contrast, BCLX_L and BCL2 protein levels changed modestly at the highest tested concentrations of silvestrol and Hipp (Fig. S8D).

Fig. 4.

Silvestrol inhibits production of MCL1 and MYC in MM cells. (A) Western blot monitoring MCL1 and MYC levels following acute treatment of JJN-3 cells with eIF4F inhibitors. Cells were incubated with the indicated concentrations of silvestrol and Hipp, 4E1RCat (25 μM), or 4EGI-1 (50 μM) for 5 h, at which point extracts were prepared, fractionated on SDS/PAGE, and transferred to PVDF membranes. Immunoblots were probed with antibodies to the indicated proteins. (B) Polysome profile analysis of JJN-3 cells exposed to vehicle or 20 nM silvestrol for 1 h. Shown below the polysomes are the RT-quantitative PCR results demonstrating the distribution of MCL1 and MYC mRNA throughout the polysome gradients. (C) Expression levels of MCL1 in JJN-3 cells infected with lentiviral gene ontology (LEGO) or LEGO-MCL1 lentivirus. Following infection, cells were sorted by FACS. Forty-eight hours later, the infected population was lysed, fractionated by SDS/PAGE, and transferred to PVDF membranes. Western blots were probed with antibodies to the indicated proteins. (D) Change in representation of JJN-3 cells infected with LEGO or LEGO-MCL1 and grown in the presence of the indicated concentrations of silvestrol (n = 3). Error bars represent ± SEM. (E) Ectopic expression of MCL1 blunts DEX/silvestrol synergy.

MCL1 is essential for survival of MM cells, and its overexpression is associated with relapse and a shortened survival period (29). Indeed, elevating MCL1 levels in the DEX-responsive KMS-11 cell line can increase resistance to DEX (Fig. S9). Ectopic overexpression of MCL1 (Fig. 4C) also altered the sensitivity of MM cells to silvestrol by increasing the IC₅₀ twofold (Fig. 4D) and blunted the DEX/silvestrol synergy (Fig. 4E). Because IL-6 can induce expression of MCL1 in MM and favor MM cell survival (30) in some cases, we tested whether IL-6–induced MCL1 levels could still be blunted by silvestrol. Exposure of the IL-6–responsive MM.1S cell line to paracrine IL-6 increased MCL1 levels, and this response was still inhibited by silvestrol (Fig. S10). We also assessed if suppression of MYC could synergize with DEX in JJN-3 cells. Inhibition of MYC production using the BET(bromo and extra terminal)-bromodomain inhibitor JQ1 (Fig. 5 A and B) was associated with significant synergy with DEX (Fig. 5C). Taken together, these results identify MCL1 and MYC as eIF4F-responsive targets whose levels can influence sensitivity to DEX.

Fig. 5.

Suppression of MYC synergizes with DEX in JJN-3 cells. (A) JQ1 affects MYC production in JJN-3 cells. JJN-3 cells were exposed to vehicle or the indicated concentrations of JQ1 for 16 h, lysed, prepared for Western blotting, and probed with the indicated antibodies. (B) Response of JJN-3 cells to JQ1. Cells were incubated in the presence of JQ1 for 48 h, and viability was assessed (n = 3). Error bars represent ± SEM. (C) Synergy between DEX and JQ1.

Discussion

Translation, as well as several upstream regulatory nodes, is often usurped in MM (Introduction). Herein, we report on a focused shRNA screen that identified several vulnerabilities that could be therapeutically explored in this disease (Fig. 1). It is striking that among the targets identified as being synthetic lethal with DEX, suppression of all three subunits of the eIF4F complex was among the “hits” (Fig. 1). We independently validated these results by pharmacological targeting of eIF4F in MM in the presence of DEX. One difference we observed between the shRNA-based validation and experiments involving small-molecule inhibitors was that long-term (12 d) RNAi-mediated suppression of eIF4F subunits in JJN-3 cells was not as toxic as small-molecule–mediated inhibition of eIF4F on a shorter time scale (∼48 h). Although we cannot formally rule out potential off-target drug effects contributing to drug toxicity, our finding may also be a consequence of incomplete suppression of eIF4F subunits by the shRNAs used. It is interesting that we observed synergy between eIF4A1 or eIF4G1 suppression and DEX, but not when eIF4A2 or eIF4G3 was suppressed (Fig. 1 and Fig. S3). Although we cannot rule out insufficient knockdown of eIF4A2 and eIF4G3 levels as being responsible for this lack of response, the results are consistent with previous data alluding to functional differences between the eIF4A and eIF4G isoforms (21–23).

Previous studies have characterized an addiction of MM cells to PI3K/mTOR signaling (3) and MYC (16), both of which are pathways intimately linked to translational control. A previous report documented the ability of 50 μM 4EGI-1 to trigger apoptosis in several MM cells as a single agent and demonstrated a reduction in MCL1 and BCLX_L, but not BCL2, levels (31). We also find that 4EGI-1 is effective against MM cells; however, among the compounds tested, it was the least potent, displaying an IC₅₀ of 20 μM against JJN-3 cells (Fig. 3B). In contrast, blocking eIF4F helicase activity was a more effective means of sensitizing JJN-3 cells to DEX. An advantage of using silvestrol over mTOR or PI3K/TOR KIs is that elevated eIF4E levels can lead to PI3K/TOR KI resistance (32), and by inhibiting eIF4A, one targets downstream of this resistance node.

We found that silvestrol was quite potent as a single agent against all MM cells and primary MM samples tested (Fig. 3 and Fig. S6 C and D). Silvestrol was also capable of increasing the sensitivity of JJN-3 and KMS-11 cells to DEX. Although the molecular basis of DEX resistance in the clinic is not completely understood, several studies indicate that a source of acquired resistance in leukemic cells is due to low or defective GR (33–36). In addition, long-term DEX treatment can lead to acquired resistance through mechanisms that may involve epigenetic events. Although we have not measured the absolute number of GRs in JJN-3 cells, the GR/GILZ signaling axis appears intact and functional (Fig. S1C). In a clinical setting of DEX resistance, silvestrol resensitized MM cells to DEX ex vivo (Fig. 3E), although we do not have a molecular understanding of the basis of DEX resistance in this case. The clinical challenge will be to determine the optimal setting in which to use silvestrol (or a related compound) to take advantage of its single-agent potency and ability to augment DEX effectiveness.

Although large-scale characterization of changes in the translatome would be required to identify all silvestrol-responsive transcripts, our studies identify MCL1 and MYC as being important to silvestrol’s mechanism of action. Nonetheless, we cannot rule out the contribution from other silvestrol-responsive transcripts. The fact that silvestrol inhibits MCL1 and MYC production and that individual suppression of MCL1 or MYC expression synergizes with DEX (Fig. 5 and Figs. S2A and S6) indicate that silvestrol is exerting its sensitization effect through altering levels of these (and possibly other) translational targets. We note that MCL1 has also been implicated as a modifier of glucocorticoid-induced cell death in acute lymphoblastic leukemia (37). The ability of silvestrol to affect several biological processes simultaneously is a key distinguishing feature of inhibiting eIF4F activity. In sum, our results demonstrate that targeting the eIF4F/eIF4A translational node is an effective approach to curtail survival of MM cells, silvestrol is a potent single agent against MM, and silvestrol (or related compounds) could prove to be an attractive adjunct to DEX therapy.

Experimental Procedures

Cell Lines and Primary Myeloma Samples.

JJN-3, KMS-11, RPMI8226, U266B1, INA-6, MM.1S, MM.1R, and OPM1 cell lines were maintained in RPMI supplemented with 10% (vol/vol) FBS, penicillin/streptomycin, and glutamine. BJ, IMR90, and W138 cell lines were grown in DMEM supplemented with 10% (vol/vol) FBS, penicillin/streptomycin, and glutamine. Cells were routinely split 1:3 every 2–3 d and discarded after >3 wk in culture. The 293T/17 cells were maintained in DMEM supplemented with 10% (vol/vol) FBS, penicillin/streptomycin, and glutamine.

For apoptosis assays of primary patient samples, bone marrow samples from patients with MM were harvested following a McGill University Health Centre Institutional Review Board-approved informed consent protocol, and mononuclear cells were plated in Iscove’s modified Dulbecco’s medium supplemented with 15% (vol/vol) FCS in the presence of vehicle alone or the indicated concentrations of DEX and/or silvestrol. Following 24–48 h of incubation, cells were double-stained with anti–CD138-Cy5 and Annexin V-phycoerythrin (BD Pharmingen) and were analyzed for apoptosis by flow cytometry (FACSCalibur; Becton Dickinson).

shRNA Library Design and Synthetic Lethal RNAi Screen.

Information regarding the construction of the human shRNA library targeting the translation apparatus, establishment of parameters for the synthetic lethal RNAi screen, and analysis of deep sequencing data is provided in SI Experimental Procedures.

Western Blots.

Western blots were performed as previously described (26). Details are provided in SI Experimental Procedures.

In Vitro Fitness Assay and Median Effect Analysis.

Median effect analysis was performed essentially as described (26). Details are provided in SI Experimental Procedures.