The p97 Inhibitor CB-5083 Is a Unique Disrupter of Protein Homeostasis in Models of Multiple Myeloma

Inhibition of the β5 peptidase of the 20S proteasome with bortezomib provided clinical validation for targeting protein homeostasis and transformed the standard of care in both multiple myeloma and mantle cell lymphoma (1, 2). The recent clinical successes of the second-generation proteasome inhibitors (PI), carfilzomib and ixazomib, have built upon this therapeutic strategy and confirmed the clinical susceptibility of multiple myeloma to inhibitors of the ubiquitin proteasome system (UPS; ref. 3). The dependency of multiple myeloma on protein homeostasis pathways is owing to the fact that they retain the characteristic biology of normal plasma cells, namely synthesis and secretion of large amounts of immunoglobulins (4). However, multiple myeloma remains an incurable disease, even with the introduction of new therapeutic modalities over the past two years, and hence, there is a need to develop new agents intervening in the myeloma cell's protein homeostasis pathways at different points compared with PIs.

Recently, inhibition of p97 has emerged as a new approach for targeting protein homeostasis in tumor cells (5, 6). Also known as valosin-containing protein, p97 is an essential and conserved member of the AAA family of adenosine triphosphatases (AAA ATPases) and is a key regulator of protein homeostasis. p97 generates mechanical force via ATP hydrolysis, and this force is used to extract proteins from macromolecular complexes and organelles. p97 is involved in multiple biological processes, including protein homeostasis, endoplasmic reticulum–associated degradation (ERAD), autophagy, chromatin remodeling, and Golgi reassembly through its interaction with cofactors (7). Reducing p97 levels with siRNA causes ER stress and activates apoptosis through the unfolded protein response (UPR), a pathway that acts both to resolve unfolded protein stress and to trigger cell death when the buildup of such unfolded proteins becomes irresolvable (5, 8, 9).

CB-5083 is a potent and highly selective inhibitor of the D2 ATPase domain of p97 (10, 11). Inhibition of p97 by CB-5083 activates the UPR and subsequently induces apoptosis in a wide variety of hematologic and solid tumor cell lines. CB-5083 was shown to be the first inhibitor of p97 to demonstrate significant tumor growth inhibition in vivo and was also active in solid tumor models where PIs were inactive (10). As CB-5083 disrupts protein homeostasis upstream of the proteasome, we wanted to assess whether targeting p97 would have different effects from PIs in various in vitro and in vivo models of multiple myeloma. Our studies described here suggest there is a therapeutic rationale for targeting p97 in multiple myeloma and that inhibition of p97 is biologically distinct from targeting the proteasome.

Bortezomib- and carfilzomib-resistant AMO-1 cell lines were generated as described previously (12, 13). Cancer cell lines were obtained from ATCC or Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH between 2011 and 2015 and were cultured according to the manufacturer's instructions. Fresh bone marrow samples from adult patients with multiple myeloma were extracted at the respective hospital centers during the patient's regular treatment schemes, following clinical practice at the center. These samples were treated with CB-5083, ex vivo, and viability of the CD38/CB138-negative and positive populations were analyzed with the ExviTech platform (14). Lenalidomide, bortezomib, carfilzomib, and ixazomib were obtained from Selleck Chemicals. Thapsigargin and dexamethasone were obtained from Sigma.

Cells were cultured in clear bottomed, tissue culture–treated 384-well plates and treated with compound in well duplicates. For immunofluorescence staining, paraformaldehyde (4% final) was added to plates after 8 hours of treatment. Cells were blocked in PBS with 1% BSA, 0.3% Triton X-100 and Hoechst (1:10,000) for 1 hour and then incubated in primary antibodies at 4°C for 16 hours. Cells were washed three times in PBS, and secondary antibodies were added for 2 hours at 25°C. Cells were washed four times in PBS and imaged with an automated wide field fluorescence microscope (Cell Insight, Thermo Fisher Scientific). Automated image analysis was written in MATLAB (MathWorks) to count nuclei, mask cellular compartments, and measure fluorescence intensities within cellular compartments.

Cells or ex vivo lysates were lysed with RIPA buffer supplemented with protease inhibitors (Roche Applied Science) and phosphatase inhibitors (Sigma). Lysates were cleared and protein was quantified by Pierce BCA Protein Assay Kit. Western blotting was performed using the Novex NuPAGE SDS-PAGE Gel system. Briefly, 10 μg of protein was resolved on 4% to 12% Bis-Tris gradient gels and then transferred onto nitrocellulose membranes. Membranes were blocked for 1 hour at room temperature in TBS + 0.1% Tween (TBST) with 5% nonfat dry milk. Membranes were probed overnight at 4°C with primary antibodies (Supplementary Table S1). Membranes were washed three times in TBST and then incubated with goat anti-rabbit or goat anti-mouse secondary antibodies (Supplementary Table S1) for 1 hour at room temperature. Membranes were then washed three times with TBST and developed with SuperSignal West Dura Chemiluminescent Substrate. Images were taken using a Bio-Rad Gel Doc Imager system, exported as TIF files and cropped using NIH ImageJ.

Viability assays conducted in myeloma cell lines, stably transduced with firefly luciferase, were carried out as described previously (15, 16). For coculture assays, short-term ex vivo cultures of CD138⁻ fractions from bone marrow aspirates were cultured for three passages. HS5, HS27A, or patient-derived cultures were seeded into assay plates 16 hours prior to addition of multiple myeloma cells. Cocultures were grown for 24 hours, and then after drug addition, multiple myeloma cell viability was monitored using luciferase as described above. GI₅₀ values were derived from dose–response curves plotted and fitted using Prism (GraphPad). CellTiter-Glo assays were conducted as described previously (10).

Mean gene expression values were calculated for all solid tumor–derived cell lines versus multiple myeloma cell lines using the CCLE (https://portals.broadinstitute.org/ccle/home) gene expression data. Difference in gene expression between the two cell groups was calculated using a Student t test. Genes overexpressed in multiple myeloma with a P >0.001 were analyzed using ENRICHR (http://amp.pharm.mssm.edu/Enrichr/).

Multiple myeloma cells were treated with CB-5083, bortezomib, or with DMSO as a control. RNA samples were converted into cDNA libraries using the Illumina TruSeq Stranded mRNA Sample Preparation Kit (Illumina # RS-122-2103). The pipeline mRNAv8-RSEM (EA-Quintiles) was used to analyze RNA sequencing (RNA-seq) data. Total RNA extraction, purification, and data analysis were performed as described previously (10). The RNA-seq data have been deposited in NCBI's Gene Expression Omnibus (17) and are accessible through GEO Series accession number GSE101923 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE101923). For gene ontology analysis, samples were processed as described previously (10).

All experiments involving animals were approved by an institutional Animal Care and Use Committee. Female athymic nude mice and SCID beige mice, 5 to 8 weeks of age and weighing approximately 20 to 25 g were purchased from Envigo. Cancer cell line xenografts were established by implanting 1 × 10⁶ to 20 × 10⁶ cells subcutaneously or intravenously. For the ortho-metastatic models, biweekly luciferase imaging was conducted using the Xenogen IVIS Spectrum Imaging System and Living Image software. Overall tumor burden was both quantified in photons/second and visualized by heatmap representations of radiance (photons/sec/cm²/sr). CB-5083, lenalidomide, ixazomib, and the CB-5083 vehicle (0.5% methyl cellulose in water) were administered by oral gavage, bortezomib and carfilzomib were dosed by tail vein injection, and dexamethasone was administered intraperitoneally. Tumor volume and body weights were measured twice weekly and daily, respectively, throughout the study. Vk*MYC mice work and SPEP analysis was performed as described previously (18).

Tumors were excised, flash frozen, and stored at −60°C to −90°C. Following thawing, tissues were processed as described previously (10). Determination of cleaved PARP and CHOP accumulation in tumor lysates was performed using commercially available MesoScale Kits (MSD). Determination of K48-polyubiquitin accumulation in tumor lysates was performed by developing a MesoScale assay with a K48-Ubiquitin antibody (Millipore). Total protein concentrations were determined with a BCA assay (BCA Protein Assay #23225, Thermo Fisher Scientific) and then normalized to 2 mg/mL. Aliquots were added to the ELISA/MSD plate, and the remaining assay procedures were carried out according to the manufacturer's protocol. Determination of circulating lambda light chain in mouse plasma was performed using a commercially available kit (Biovendor).

For in vivo studies, statistical analysis was performed using two-way ANOVA on delta tumor values in Prism (GraphPad).

To characterize the potency of cell death following exposure to CB-5083, 18 human multiple myeloma cell lines were treated with CB-5083. Following 72-hour exposure, CB-5083 showed GI₅₀s ranging from 96 to 1,152 nmol/L, with a median GI₅₀ of 326 nmol/L (Fig. 1A). CB-5083 was less potent in immortalized or primary patient-derived bone marrow stromal cultures compared with multiple myeloma cell lines (Figs. 1A and 6D). CB-5083 has potent antitumor activity in several solid tumor cell lines (10); however, multiple myeloma cell lines showed significantly greater overall sensitivity to CB-5083 than solid tumor cell lines (Fig. 1B). In an attempt to understand the difference in sensitivity of solid tumor versus multiple myeloma cell lines, we analyzed publicly available genomic data (CCLE) via ENRICHR (http://amp.pharm.mssm.edu/Enrichr/). GO Biological Process showed the myeloma cells to have significant enrichment of ER/UPR–related pathways compared with the solid tumor cells (Fig. 1C). Frequently, the levels of a drug target can influence sensitivity to its inhibitor. Therefore, we analyzed p97 protein levels in a subset of three multiple myeloma and three solid tumor cell lines (Supplementary Fig. S1A) and found no correlation between p97 levels and CB-5083 sensitivity. We next analyzed the kinetics of cell death in this subset of cell lines. The multiple myeloma cell lines MM.1S, AMO-1, and RPMI8226 died far more rapidly than the colon tumor cell lines HCT116 and DLD-1 and the pancreatic cancer cell line, CFPAC-1 (Fig. 1D). These data were consistent with the induction of apoptosis that peaked between 8 and 16 hours in the multiple myeloma cell lines versus 36 to 48 hours in the colon and pancreatic tumor cell lines (Fig. 1E). This rapid kinetic of cell death was confirmed using an independent viability assay (time-lapse bioluminescence; Supplementary Fig. S1B and S1C). In conclusion, we found that multiple myeloma cells grown in the presence of CB-5083 are highly sensitive, die rapidly, and have a baseline gene expression profile that is distinct from less sensitive solid tumor cell lines.

p97 knockdown or inhibition with small molecules has been shown to induce endoplasmic reticulum (ER) stress and activate all three arms of the UPR (5, 19). CB-5083 caused a strong induction of the UPR as measured by the phosphorylation of PERK and accumulation of sXBP1 and BiP (Fig. 2A; Supplementary Fig. S2). CB-5083 treatment led to a robust induction of the transcription factor CCAAT/enhancer-binding protein homologous protein (CHOP), which was higher than the maximum response observed with bortezomib and similar to what was seen with the ER stress agent thapsigargin.

K48-linked polyubiquitin accumulation, a hallmark of inhibition of protein degradation (20, 21), was followed as a pathway marker for p97 inhibition (Fig. 2A and B). CB-5083, like bortezomib, induced a rapid accumulation of K48-linked polyubiquitinated proteins in cultured cells and xenografted multiple myeloma tumors. In RPMI8226 xenografts, a single dose of CB-5083 led to plasma levels of CB-5083 that ranged from 2 to 20 μmol/L over a period of 24 hours, sufficient to maintain significant K48 polyubiquitin accumulation (Fig. 2B). This dose was sufficient to activate the UPR, and CHOP protein expression increased up to 2.7-fold compared with baseline levels. Consequently, apoptosis was activated, as demonstrated by an increase in cleaved PARP (cPARP) levels (Fig. 2C).

The target engagement and pathway protein modulations seen in multiple myeloma cell lines in vitro and in vivo in response to CB-5083 treatment were consistent with the cellular responses to p97 inhibition, with differences noted when comparing with bortezomib.

To evaluate the similarities or differences in protein homeostasis disruption via PIs versus p97 inhibition, we compared the relative sensitivity of 18 multiple myeloma cell lines to bortezomib, carfilzomib, and CB-5083 by correlating GI₅₀ values across the cell line panel for all pairwise comparisons of inhibitors. The correlation for relative sensitivities of the multiple myeloma cell lines to bortezomib and carfilzomib was significant (P = 0.0001), reflecting the similar mechanism of action of both drugs (Fig. 3C). However, the relative response and sensitivity across the multiple myeloma lines were not significantly correlated between CB-5083 and either PI (P = 0.613 and 0.118, respectively; Fig. 3A and B).

Next, we analyzed gene expression in response to either bortezomib or CB-5083 in RPMI8226, AMO-1, and MM.1S cell lines. In these experiments, cross-comparisons of ranked gene lists associated with each agent were clearly distinct (Supplementary Fig. S3A–S3B). Although transcripts that were most strongly induced in response to CB-5083 were also induced by bortezomib (Supplementary Fig. S3A), transcripts that were most strongly induced in response to bortezomib showed a mixed profile in response to CB-5083 (Supplementary Fig. S3B). More significant differences were noted upon analysis of ranked gene identities for bortezomib-induced transcripts, which indicated that many of the heat shock protein family members and chaperones from the Bag family typically induced by bortezomib (22) were decreased in response to CB-5083 (Fig. 3D). Using RNA-seq data, gene ontology analysis was performed to identify the pathways that were altered after bortezomib or CB-5083 treatment. In both RPMI8226 (Fig. 3E) and AMO-1 (Fig. 3F) cell lines, the gene ontology analysis showed both compounds induced similar pathways, but CB-5083 responses were predominantly focused on ER stress and UPR, whereas bortezomib responses were focused more on general protein stress. Altogether, these cell viability and transcriptomic results suggest that the mechanism of action of CB-5083 and the transcriptional response it produced has some unique features compared with PIs.

We next investigated whether the blockade of both p97 and the proteasome could lead to enhanced cytotoxicity. RPMI8226 cells were treated with a range of CB-5083 concentrations in the presence of various dose levels of bortezomib (Fig. 4A) or carfilzomib (Fig. 4B). When PIs were combined with CB-5083, a lower GI₅₀ of CB-5083 was observed, suggesting enhanced cell killing activity. These data were confirmed in three additional multiple myeloma cell lines (Supplementary Fig. S4A–S4C). The combination of CB-5083 with PIs was then studied in vivo. After a single dose of CB-5083 combined with a single dose of bortezomib, the combination arm showed a much more dramatic effect on the induction of K48-ubiquitin and cPARP in comparison with each single agent alone (Fig. 4C and D). The sustained K48-ubiquitin and cPARP accumulation translated into significantly improved antitumor activity (Fig. 4E and F). The combination of CB-5083 and bortezomib in the AMO-1 model was highly active (TGI 108%– Supplementary Fig. S4D), and when represented in a waterfall plot, the individual data showed an increased response rate with the combination when compared with single-agent bortezomib or CB-5083 (Fig. 4E). In RPMI8226 tumors, the combination of suboptimal doses of CB-5083 and carfilzomib showed superiority to both single agents (P = 0.0004; Fig. 4F). In summary, when CB-5083 was combined with PIs, enhanced antitumor effects were observed.

Upon exposure to PIs, mammalian cells increase expression of multiple proteasome subunits, which elevate proteasome content and promote survival (22, 23). This process is driven by the transcription factor Nrf1/NFE2L1. It was recently demonstrated that p97 is required for this Nrf1-mediated proteasome bounce-back response owing to its activity in retro-translocating Nrf1 from the ER membrane to the cytosol (24). As this proteasome bounce-back has been described to limit the ability of PIs to kill myeloma cells (25), blocking such a compensatory response through p97 inhibition may provide a mechanism to enhance the efficacy of PIs by coadministration with a p97 inhibitor, such as CB-5083. Therefore, we treated RPMI8226 and AMO-1 cells with either CB-5083, bortezomib, or the combination of both agents (CB-5083 treatment 1 hour prior to bortezomib; Fig. 5A). Treatment with bortezomib for 1 hour was not long enough to stabilize Nrf1 (Fig. 5A). At later time points, a low molecular weight band of Nrf1 accumulated in cells treated with bortezomib, whereas after CB-5083 treatment, a higher molecular weight band of Nrf1 accumulated. In the combination experiment of both agents, most of the Nrf1 was found as the higher molecular weight band. The low molecular weight band represents a cleaved and active form of Nrf1 that is translocated to the nucleus, where it activates many proteasome subunit genes (PSM; see later Fig. 5C). CB-5083 treatment leads to the accumulation of an uncleaved higher molecular weight form of Nrf1, which corresponds to Nrf1 that is retained in the ER membrane and cannot translocate to the nucleus (Fig. 5C and D). The CB-5083–mediated retention of Nrf1 in the ER was observed in additional multiple myeloma cell lines (Supplementary Fig. S5A) and the dominant CB-5083 phenotype in combination with bortezomib was recapitulated in vivo in RPMI8226 xenografts (Fig. 5B). In A549 cells, CB-5083 prevented the bortezomib-induced translocation of Nrf1 to the nucleus in a dose-dependent manner (Fig. 5C and D).

Consistent with a block in Nrf1 activation, bortezomib-induced expression of proteasome genes PSMA6, PSMC6, and PSMD11 is prevented by cotreatment with CB-5083 (Supplementary Fig. S5B–S5D). In conclusion, combining CB-5083 with bortezomib prevented the activation of Nrf1 and the downstream activation of proteasome subunit expression.

CB-5083 is a protein homeostasis disruptor with a differentiated mechanism of action compared with the PIs and has the potential to be active in multiple myeloma cells resistant to PIs. Therefore, CB-5083 was tested against AMO-1 cell line clones that are resistant to bortezomib or carfilzomib (12). AMO1-BtzR has a bortezomib GI₅₀ >100-fold higher than AMO1-WT. A similar shift in sensitivity was seen in the AMO1-CfzR cell line with respect to carfilzomib (Fig. 6B and C). Interestingly, each clone possessed significant cross-resistance to the alternative PI. However, CB-5083 potency was similar in the wild-type AMO-1 cells and the AMO1-BtzR and AMO1-CfzR clones (Fig. 6A).

One potential mechanism of resistance to PIs is through interaction of multiple myeloma cells with the supportive bone marrow stromal compartment (BMSC; refs. 26, 27). To assess the influence of BMSCs on CB-5083's anti-myeloma activity, compartment-specific bioluminescence (CS-BLI; ref. 15) assays were performed with MM.1S cells cocultured in the presence of the BMSC cell lines HS5 and HS27A or primary multiple myeloma patient-derived BMSC culture, BM-61. In these assays, the MM.1S GI₅₀ was 0.53 μmol/L and was unaffected by the presence or absence of BMSCs (Fig. 6D). Furthermore, the GI₅₀ against the MM.1S cell compartment was 3- and 6-fold lower than the immortalized and primary BMSC compartment GI₅₀ alone (1.58 μmol/L for HS5 and 3.18 μmol/L for BM-61). In contrast, carfilzomib demonstrated a marked reduction in potency and decreased anti-myeloma activity in the context of all BMSCs, with its GI₅₀ shifting 1.7- to 2.4-fold, from 21 nmol/L to 36–51 nmol/L in coculture with BMSCs (Fig. 6E).

Extending the above observations from CS-BLI into patient-derived samples, CB-5083 activity was assessed in bone marrow aspirates from multiple myeloma patients. Twenty-three samples were assessed by flow to measure apoptosis in CD138⁺ and CD138⁻ cells, 14 from newly diagnosed patients and 9 from patients that had been previously treated with bortezomib. In the newly diagnosed samples, CB-5083 activity varied widely with GI₅₀s ranging from 0.47 to 36.3 μmol/L. However, CB-5083 had GI₅₀s below 1.6 μmol/L in all but one previously treated patient samples (Fig. 6F). In the same bone marrow sample, analysis of CB-5083 activity in CD138⁻ cells showed lower sensitivity compared with the CD138⁺ cell population in 16 of 18 samples, where both types were analyzed (Fig. 6G). These findings suggest that CB-5083 has more cytotoxic activity against multiple myeloma cells compared with the normal cell population of bone marrow.

In conclusion, CB-5083 was shown to be active against PI adapted cell line models, retained potent anti-myeloma activity in BMSC coculture models, and had potent activity in ex vivo previously treated multiple myeloma patient samples.

Previously, CB-5083 was shown to be active in a variety of solid tumor models (10). We used an ortho-metastatic mouse model of multiple myeloma to assess efficacy of CB-5083. Mice were injected intravenously with MM.1S (Fig. 7A) or RPMI8226 cells (Supplementary Fig. S6A) that were modified to express luciferase. Oral treatment with CB-5083 at a dose of 60 mg/kg on a 4 days on/3 days off schedule inhibited tumor progression to an extent that compared favorably (P < 0.0001) to intravenous bortezomib over a 3-week dosing cycle (Fig. 7B). CB-5083 demonstrated a similar level of efficacy in the disseminated RPMI8226 tumor model, which again, compared favorably with an effective regimen of intraperitoneal carfilzomib (Supplementary Fig. S6B).

In a subcutaneous RPMI8226 model, CB-5083 was compared with bortezomib, carfilzomib, and ixazomib. In this model, all agents inhibited RPMI8226 growth with CB-5083 showing the highest TGI (Fig. 7C). After three cycles of dosing with CB-5083, tumors were allowed to regrow and were rechallenged with CB-5083 once they reached 700 mm³ (Supplementary Fig. S6C). CB-5083 was again able to induce 50% regression of these tumors.

Next, CB-5083 activity was investigated in combination with dexamethasone and lenalidomide, two active anti–multiple myeloma agents (Fig. 7D). Suboptimal doses of CB-5083 combined with suboptimal doses of dexamethasone and lenalidomide led to regression of tumors, demonstrating the superiority of the triple combination arm (P < 0.0001). These findings were consistent with the level of lambda light chains quantified in the plasma of the tumor-bearing mice (Supplementary Fig. S6D).

Finally, CB-5083 antitumor activity was evaluated in the Vk*MYC model, a transgenic mouse model that reproduces the pathogenesis, biology, and clinical features of multiple myeloma and is predictive of clinical activity (18, 28). Prior to dosing, baseline serum M-spike was determined by serum protein electrophoresis (SPEP) in individual mice. CB-5083 was administered orally for 2 weekly cycles in C57/BL6 Vk*MYC mice. On day 14, SPEP analysis of serum showed a 55% decrease in M-spike when compared with predose levels (Fig. 7E). Taken together, CB-5083 showed potent antitumor activity in a wide variety of clinically relevant in vivo multiple myeloma models.

It has been well documented that myeloma cells are exquisitely sensitive to perturbation of protein homeostasis, owing to their high rates of immunoglobulin synthesis and secretion (4). Pharmacologic inhibition of the chymotryptic site of the proteasome exhibits potent clinical activity in myeloma, Waldenström macroglobulinemia, and amyloidosis (1, 29, 30). These observations suggest that the activity of proteasome inhibition in multiple myeloma may represent an effect against the intrinsic biology of plasma cells rather than the specific oncogenic events harbored by the multiple myeloma cells.

The current study examines the phenotypic consequences of pharmacologic inhibition of p97/VCP using CB-5083 in preclinical models of multiple myeloma. p97 plays important roles in protein homeostasis upstream of the proteasome, and CB-5083 has recently been described as a novel anticancer agent targeting p97. CB-5083 appears to be quite selective for the ERAD-related functions of p97, with only minor (if any) effects on its other cellular functions. This relative selectivity of CB-5083 for the ERAD-related functions of p97 renders it an attractive option for the treatment of multiple myeloma, which exhibits pronounced responsiveness upon pharmacologic perturbation of ERAD. As p97 was demonstrated to be involved in the regulation of other oncogenic pathways [including NFκB (31) or HIF1α (32)], further study of the non-ER functions of p97 that may be affected by CB-5083 and may have been missed by the gene ontology analysis is warranted.

On the basis of these considerations, we set out to characterize the response to CB-5083 in models of multiple myeloma. We observed that submicromolar concentrations of CB-5083 were capable of inducing a significant decrease in viability for the majority of multiple myeloma cell lines tested. The observed GI₅₀ values for multiple myeloma cell lines were significantly lower than those observed for non–multiple myeloma cell lines tested in this work or previously (10). The exquisite sensitivity of multiple myeloma cells to CB-5083 could be explained by a much higher level of ER stress and UPR at baseline, in comparison with solid tumor cell lines. In viability assays, there was little correlation between the sensitivity of multiple myeloma cell lines to CB-5083 and their sensitivity to PIs. The anti–multiple myeloma activity of CB-5083 was not significantly attenuated by in vitro interactions of multiple myeloma cells with BMSCs. This can be viewed as a favorable phenotypic result, given BMSCs confer resistance to anti–multiple myeloma agents via cell adhesion–mediated drug resistance (15, 26). The preclinical anti–multiple myeloma activity of CB-5083 was associated with potent induction of canonical markers of the UPR, including phosphorylation of PERK, accumulation of sXBP1 and BiP, and induction of CHOP, as well as accumulation of K48-linked polyubiquitin, a hallmark of inhibition of protein degradation. Despite some of the similarities in gene expression changes seen in response to CB-5083 and PIs, CB-5083 did exhibit some distinct features. The transcriptional profile of CB-5083–treated multiple myeloma cells lacked the pronounced upregulation of heat shock genes or proteasome subunits, which has been reported to reflect compensatory responses mounted by multiple myeloma cells in the context of 20S proteasome inhibition (22, 33). These distinct events for CB-5083 versus PIs may be related to the differential impact of these classes of drugs on posttranslational processing, intracellular localization, and transcriptional activity of Nrf1. Unlike bortezomib or carfilzomib, CB-5083 treatment is associated with inhibition of Nrf1 cleavage and its retention in the ER membrane, thereby preventing its nuclear translocation and transcription of proteasome subunit genes. Notably, Nrf1-mediated induction of proteasome gene expression is thought to limit the cytotoxic effects of the UPR in myeloma cells in response to PIs (24, 25, 34, 35).

CB-5083 enhanced the anti–multiple myeloma activity of bortezomib both in vitro and in vivo and was active in bortezomib resistance models as well as primary multiple myeloma cells derived from patients who had been treated with bortezomib-based regimens. The observation that CB-5083 decreases the myeloma tumor burden in the Vk*MYC genetically engineered mouse model of myeloma is also notable, because it suggests that p97 inhibition may have utility at different stages of the disease. This model is considered to capture the behavior of multiple myeloma in its earlier stages, whereas xenografts of human myeloma cell lines are considered to be reflective of the molecular features of advanced multiple myeloma (18, 28).

An important area of investigation relates to the characterization of molecular markers that may identify myeloma cells with pronounced responses to CB-5083 monotherapy. Indeed, some of the myeloma cell lines tested in our studies exhibited more potent and/or rapid responses to CB-5083 compared with other cell lines in the panel. Our current study was not designed to identify such candidate markers, but it is reasonable to hypothesize that these markers may prove to be quite distinct from those associated with potent responses to PIs given that the quantitative patterns of response of multiple myeloma cells to these agents differ substantially from the observed pattern for CB-5083. In a separate recent study from our group on the impact of CB-5083 on a large panel of solid tumor cell lines, mutational activation of Ras/Raf/MAPK signaling exhibited a trend for a higher degree of responsiveness to CB-5083 (10). Multiple myeloma also exhibits frequent activation of Ras/Raf/MAPK signaling through mutations in diverse components of this pathway (36) and/or cell nonautonomous activation in the context of tumor microenvironment interactions (26, 37). We were able to demonstrate on a smaller panel of multiple myeloma cells listed in Anderson and colleagues' article (AMO-1, RPMI8226, OPM2, MM.1S, and MM.1R) that there was a strong correlation between multiple myeloma tumor responses and phospho/total ERK ratio [unpublished analysis (10)]. We envision that validation of this hypothesis in studies of larger cell line panels or in samples from patients in clinical trials of CB-5083 may represent a path toward confirming or refuting whether CB-5083 might play an important role in the treatment of multiple myeloma patients with mutations in the Ras/Raf/MAPK pathway.

In summary, CB-5083 has demonstrated robust preclinical activity via a novel and distinctive mechanism of action. Currently, clinical trials are ongoing in patients with multiple myeloma who have exhausted other available therapies.

B.T. Aftab is the director at Atara Biotherapeutics, Inc. and reports receiving commercial research grants from Amgen, Cleave Biosciences, CytomX Therapeutics, Omniox, Inc., and Sanofi Oncology. S. Djakovic has ownership interest (including patents) in Cleave Biosciences. Z.Y. Wu is a senior research associate at Cleave Biosciences. A.P. Wiita reports receiving a commercial research grant from Cleave Biosciences. L. Shawver is an employee at and has ownership interest (including patents) in Cleave Biosciences. C.S. Mitsiades is the director at Takeda and reports receiving commercial research grants from Abbvie, Janssen/Johnson & Johnson, Novartis, Ono Pharmaceutical, and TEVA. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R. Le Moigne, B.T. Aftab, S. Djakovic, E. Valle, F.M. Yakes, H.-J. Zhou, C.S. Mitsiades, D.J. Anderson, M. Rolfe

Development of methodology: R. Le Moigne, B.T. Aftab, S. Djakovic, E. Dhimolea, E. Valle, Z.Y. Wu, C. Driessen, D. Wustrow, D.J. Anderson

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Le Moigne, B.T. Aftab, S. Djakovic, E. Dhimolea, E. Valle, M. Murnane, E.M. King, F. Soriano, M.-K. Menon, Z.Y. Wu, S.T. Wong, G.J. Lee, C. Lam, J. Wang, M. Chesi, P.L. Bergsagel, M. Kraus, C. Driessen, S. Kiss von Soly, L. Shawver, T.G. Martin III

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R. Le Moigne, B.T. Aftab, S. Djakovic, E. Valle, M. Murnane, M.-K. Menon, Z.Y. Wu, S.T. Wong, G.J. Lee, B. Yao, A.P. Wiita, J. Rice, F.M. Yakes, T.G. Martin III, C.S. Mitsiades, D.J. Anderson, M. Rolfe

Writing, review, and/or revision of the manuscript: R. Le Moigne, B.T. Aftab, Z.Y. Wu, G.J. Lee, C. Driessen, D. Wustrow, L. Shawver, H.-J. Zhou, T.G. Martin III, C.S. Mitsiades, D.J. Anderson, M. Rolfe

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Murnane, S.T. Wong, J. Wang, C. Driessen, D. Wustrow, J.L. Wolf

Study supervision: R. Le Moigne, B.T. Aftab, S.T. Wong, F.M. Yakes, H.-J. Zhou, T.G. Martin III, C.S. Mitsiades, D.J. Anderson, M. Rolfe

We thank Drs. Suzanne Trudel, Joan Levy, Jesse Vargas, and Daniel Auclair for scientific discussion and contributions to work presented in this manuscript.

This work was funded in part by the Multiple Myeloma Research Foundation (to B.T. Aftab), The Myeloma Fund of the Silicon Valley Community Foundation (to B.T. Aftab), and The Stephen and Nancy Grand Multiple Myeloma Translational Initiative (to B.T. Aftab). C. Driessen was supported by a grant from Oncosuisse.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.