Proteasome inhibitors (PI) are a cornerstone in the treatment of multiple myeloma (MM).Despite high initial response rates, nearly all MM patients relapse, marking the incurable nature of the disease and the emerging clin. challenge of combating PI resistance.This has created a need for drug discovery approaches that identify new drug cocktails of PIs, such as bortezomib/VELCADE (Btz; Millennium Pharmaceuticals, Inc., Cambridge, MA, USA), and other classes of FDA-approved or investigational new drugs.With these objectives in mind, we set out in this study to develop a high-throughput drug screening (HTS) platform to identify chem. structures that selectively kill or re-sensitize PI-resistant MM cells to PIs.We established a cell-based drug screening assay that incorporated isogenic pairs of Btz-sensitive (BzS) and Btz-resistant (BzR) mouse and human MM cells, which have been described previously and exhibit an approx. 5-7-fold difference in sensitivity to Btz (Supplementary Figure S1).Each plate of library compounds was tested against a plate containing BzS cells, one containing BzR cells, and a third containing BzR cells in the presence of Btz.Cell viability was used as the assay read-out to measure the effects of screened compounds on cell survival and proliferation (Figure 1a and Supplementary Figure S2).It is important to note that the inclusion of the three cell groups allowed us to calculate the relative effects of drugs, and thus identify the structures with selective activity against the resistant cells as single agents or those with the ability to restore Btz sensitivity, rather than those with general cytotoxicity in vitro (Figures 1a and b).To demonstrate the utility of this HTS assay in practice, we conducted a pilot round of screening using the NCI Diversity Set II (NCI Developmental Therapeutics Program) of ∼1600 small mols. chosen for their core structural diversity and favorably predicted drug-like qualities.In total, our primary screening identified 12 compounds with activity against any of the treatment groups (Supplementary Table S1).Of greater significance, four of the hits showed reproducibly greater activity against BzR cells as single agents or restored sensitivity to Btz in BzR cells when co-treated with Btz.We elected to further pursue compound NSC622608, named Velcade Re-sensitizing Compound 2 (VRC2) due to its unknown and potentially novel mol. target/mechanism of action and its ability to synergize with Btz and reverse the resistance phenotype in vitro (Figure 1c).We investigated the activity of VRC2 in combination with the other PIs that have been approved for MM (carfilzomib) or are in clin. development (MLN2238).Comparable to what we observed with Btz, VRC2 restored sensitivity to these next-generation PIs in a panel of mouse and human BzR cell lines (Figure 1d).This synergy appeared to be specific for PIs, as VRC2 failed to enhance the sensitivity of MM cells to other classical MM agents including, the glucocorticoid dexamethasone (Figure 1d).To identify the potential mol. targets of VRC2, we used a combination of data mining and chem. genomics approaches.First, we screened the NCBI PubChem Bioassay Database for published drug screening results that reported VRC2 as a pos. hit.We found one report of VRC2 activity against wild-type and mutant forms of the murine double minute 2 (MDM2) E3 ligase, a known regulator of TP53 (p53) activity (PubChem BioAssay identifiers AID: 1442 and 1444 from the Penn Center for Mol. Discovery; Reference 'A' and 'B').Our genomic studies, in which we used kinetic gene expression profiling (GEP) to characterize the gene networks that were induced or repressed following VRC2 treatment, further implicated the p53 pathway as a mechanism of VRC2 action.In human MM.1S BzR cells, kinetic VRC2 pathway anal. showed a strong p53 activation signature (z=2.078; P=2.39 × 10-8 by Fisher's Exact Test) (Ingenuity Pathway Anal.), which was driven by the upregulation of MT1H, HMOX1 and ANXA2 and downregulation of POLD2, MCM5, MCM4, MCM3, MCM2, KIAA0101 and CCNA2 following VRC2 treatment (Figure 2a and Supplementary Figure S3).Notably, this p53 activation signature was absent in U266 BzR cells, which express an inactivating p53 mutation at codon 161 (Figure 2a).Consistent with the predicted effects of an MDM2 inhibitor, VRC2 increased the expression of p53 in a dose- and time-dependent manner in wild-type p53-expressing MM cell lines (Figure 2b and Supplementary Figure S4A), and induced the expression of p53 target genes P21, PUMA and NOXA (Figure 2c and Supplementary Figure S4B).These mechanistic findings suggested that invoking the p53 pathway by means of MDM2 inhibition is a promising mol. strategy to overcome PI resistance in MM cells.Indeed, when we combined Btz with Nutlin3a, a quintessential MDM2 inhibitor in clin. development, we detected statistically significant synergy in two resistant clones derived from wild-type p53 expressing MM.1S BzR cells (Figure 2d and Supplementary Figure S5A).Interestingly, we found that the combination of carfilzomib and Nutlin3a was highly synergistic and significantly more robust than the combination with Btz (Figure 2d and Supplementary Figure S5A).While the addition of Nutlin3a to Btz treatment increased the percentage of apoptotic cells by 8-10% (P=0.05, N=4), the combination of Nutlin3a and carfilzomib increased apoptosis by 35-45% (P=0.004, N=4; Figure 2e).Similar results were observed using resistant mouse cell models 595 BzR (Supplementary Figure S5B) and 589 BzR (data not shown), both of which express wild-type p53.In mutant p53-expressing U266 BzR cells, synergy between Nutlin3a and PIs was also evident, albeit requiring higher concentrations of Nutlin3a (Figure 2f), and, consistent with the wild-type p53 models, the synergy was more pronounced with carfilzomib than Btz.In this study, we have described and implemented a cell-based HTS platform for discovering the drugs and mol. mechanisms that specifically target treatment-resistant MM cells.The cell-based approach is advantageous for two reasons.First, given the multiple known and unknown mechanisms by which cells acquire resistance to PIs, a cell-based approach removes mol. bias.Second, the use of isogenic BzS and BzR cell models provides a relative assessment of cytotoxicity and the opportunity to identify compounds that selectively target PI-resistant populations.As proof of concept, we conducted a small-scale drug screen from which we identified VRC2, a compound with the ability to restore PI sensitivity to resistant MM cells.While VRC2 is not an ideal candidate for further development due to characteristics that predict unfavorable physiochem. properties (i.e., furan and diathiazol toxicophores, a high number of hydrogen bond acceptors at 13, and predicted high clearance risk), in our study, it effectively served as a mol. probe to demonstrate the utility of coupling HTS with other approaches such as chem. genomics to pinpoint mol. mechanisms and therapies that are already approved for clin. use or are in human trials.Our investigation into the mechanism of VRC2 implicated MDM2 inhibition and p53 pathway activation as mol. strategies for targeting PI-resistant cells, findings that directed us toward the more clin. advanced MDM2 inhibitor, Nutlin3a.MDM2 inhibition has been shown by others to enhance the activity of Btz, 13 which supports the ability of our method to identify bona fide druggable mechanisms for overcoming PI resistance.Our study further demonstrates that Nutlin3a not only augments the cytotoxic activity of Btz in Btz naive cells, but can also restore sensitivity to Btz once MM cells have acquired therapeutic resistance.It is also noteworthy that we found the combination of Nutlin3a/carfilzomib to be substantially more synergistic than Nutlin3a/Btz, findings that require further investigation but carry potentially significant translational implications.The mechanistic basis for this difference is not clear, however, it may be explained by the known differences in binding kinetics between the two PIs-inhibition of the 26S proteasome by Btz is reversible, whereas carfilzomib binds irreversibly.In summary, this proof-of-concept work validates the utility of our HTS approach for discovering lead compounds and uncovering mol. mechanisms for targeting PI-resistant MM, and provides a platform and the impetus for larger-scale screening efforts.
