PRIMA-1^Met induces myeloma cell death independent of p53 by impairing the GSH/ROS balance

TP53 is the most frequently mutated gene in cancers, and the mutations are associated with resistance to therapy in numerous cancers, including multiple myeloma (MM).^1,2  MM remains an incurable plasma cell malignancy, although treatments for the disease have progressed in the last decade.³  It is well established that at diagnosis, patients with a deletion of the short arm of chromosome 17 (del17p), which overlaps the TP53 locus (17p13), have been shown to have a shorter survival time, irrespective of the treatment regimens.^3-7  Patients with del17p frequently (>30%) harbored a mutation in the remaining allele.⁸  Thus, there is an obvious need for compounds that bypass the defective p53 pathway, such as molecules targeting antiapoptotic molecules (eg, ABT-737 and ABT-199), which act downstream of p53.^9,10  An alternative strategy is to target the overexpressed p53 mutant protein by directly modifying its conformation to restore its proapoptotic transcriptional functions.¹¹  Indeed, molecules that can reactivate cell death in p53 mutant cells in a p53-dependent manner have been selected on the basis of their ability to either kill the cells (phenotypic screening) or bind to mutated p53 and restore a functional p53 conformation (biochemical screening).^11-14  Among them, PRIMA-1^Met (APR-246) was isolated according to its ability to restore apoptosis in SAOS2-His-273 cells in a p53-dependent manner and was shown to bind to p53.^11,15,16  PRIMA-1^Met, recently evaluated in a phase 1 study, was shown to be effective in vivo in 2 patients with a mutation in the core (V173M) or tetramerization domain (A355V). After treatment, expression of the p53 target genes Bax, Noxa, and Puma was increased in tumor cells.¹⁷ 

At a molecular level, p53 reactivation and induction of massive apoptosis (PRIMA-1^Met) induced reactive oxygen species (ROS) production and requires an oxidative environment to reactivate p53, although the precise mechanism remains elusive.¹⁶  PRIMA-1^Met may have other targets, such as oxidosqualene cyclase, which was identified using a docking-inverse approach.¹⁸  Several reports have described a p53-independent ability of PRIMA-1^Met to induce tumor apoptosis. Indeed, in AML, APR-246 was found to be efficient irrespective of TP53 mutational status.¹⁹  Russo et al also have reported that PRIMA-1^Met kills cell lines that lack p53 expression.²⁰  Very recently, Saha et al reported that PRIMA-1^Met kills MM cells via Noxa induction in a p53-independent manner.²¹ 

In the present work, we evaluated the efficacy of PRIMA-1^Met across a collection of 27 human myeloma cell lines (HMCLs) representative of the heterogeneity of myeloma, as well as in 23 independent primary MM samples characterized for del17p. This HMCL collection was representative of TP53 abnormalities found in patients (chromosome 17p deletion, different point mutations, and exon deletion) and allowed us to provide an accurate preclinical evaluation.^9,22-24  The efficacy of PRIMA-1^Met was compared with that of nutlin3a, which reactivates the p53 pathway and induces cell death only in TP53^wt HMCLs and primary cells.²⁴  We show that PRIMA-1^Met, in contrast to nutlin3a, killed HMCLs and primary myeloma cells independent of TP53 status and p53 expression in a GSH/ROS-dependent manner.

All HMCLs used in this article have been extensively characterized.^9,23-26  The human stromal cell line HS5 was purchased from ATCC (LGC Standards SARL, Molsheim, France). After informed consent, blood or bone marrow samples from patients with MM were collected at the Department of Hematology at the University Hospital of Nantes or at the Intergroupe Francophone du Myélome (ethical approval DC-2011-1399). The study was conducted in accordance with the Declaration of Helsinki. Plasma cells were obtained after gradient density centrifugation, using Ficoll-Hypaque and purification with CD138 immunomagnetic beads (Stemcell Technologies, Le Plessis Robinson, France). Purified cells were cultured for 24 hours in RPMI1640 containing 5% fetal calf serum and 3 ng/mL interleukin 6. Deletion of the short arm of chromosome 17 was assessed by fluorescence in situ hybridization.⁸ 

Nutlin3a, N-acetyl-l-cysteine (NAC), glutathione reduced monoethyl ester (GSH-MEE), and l-buthionine sulphoximine (BSO) were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France). PRIMA-1^Met, small interfering RNA (siRNA), and antibodies against caspase 2, caspase 3, γ-glutamate-cysteine synthetase modifier subunit (GCLM), and glutathione synthetase (GSS) were purchased from Santa Cruz Biotechnology (Clinisciences, Nanterre, France). Anti-Apo2.7-PE, anti-CD138-PE, control immunoglobulin G (IgG1)-phycoerythrin (PE) monoclonal antibodies (mAbs) were purchased from BD Biosciences (Le Pont de Claix, France), and anti-DR5-PE was purchased from eBioscience. Anti-p53, anti-Noxa, anti-Bax, anti-p21, and anti-actin antibodies were purchased from Oncogene Science (Life Technologies, Paris, France), Alexis Biotech (Enzo Life Sciences, Villeurbanne, France), Immunotech Beckman Coulter (Marseilles, France), Cell Signaling (Ozyme, St Quentin en Yvelines, France), and Millipore Bioscience Research Reagents (Molsheim, France), respectively. Glutathione (GSH) content was determined using a GSH/glutathione disulfide (GSSG) colorimetric detection kit (Arbor Assays, Euromedex, Strasbourg, France).

The evaluation of cell death in HMCLs was performed by flow cytometry, using Apo2.7 staining.²⁷  Cell death in primary myeloma cells was measured via the loss of CD138 staining.²⁴ 

Endogenous ROS production was assessed using CellROX oxidative stress reagent (Molecular Probes, Life Technologies, Saint-Aubin, France). CellROX Green reagent (5 μM) was added to the cell culture for 30 to 45 minutes at 37°C. After 3 washes in phosphate-buffered saline, fluorescence was analyzed on FACSCalibur (Cytocell, Structure Fédérative de Recherche Bonamy).

Stably modified shp53 myeloma cell lines were obtained via lentiviral cell transduction, as previously described.²⁴  siRNA experiments also were performed as described previously.²⁸ 

Female severe combined immunodeficiency (SCID) beige 7-week-old mice were purchased from Charles River (L'Arbresle, France). Mice were bred and housed in the Experimental Therapeutic Unit (UTE, Structure Fédérative de Recherche Bonamy) under animal care license 44565. Tumors were generated by implanting 10 × 10⁶ cells in 100 μL phosphate-buffered saline subcutaneously in the right flank above the hind leg.²⁹  The tumor volume was recorded in 3 dimensions, using a digital caliper and calculated by measuring length × width × depth.

Statistical analyses were performed using the Kruskal-Wallis, analysis of variance, Fisher exact, Mann-Whitney, and Wilcoxon matched-pairs signed-rank tests.

The present HMCL collection consisted of 9 TP53^wt HMCLs, 12 TP53^mut HMCLs, 2 TP53^wt+mut, and 4 TP53^neg HMCLs. TP53^neg HMCLs did not express p53 protein because they either lacked p53 expression at the mRNA level (JJN3, KMS11) or had a premature stop codon (NAN1) or a lack of intron splicing (L363; supplemental Table 1, available on the Blood Web site).^23,24  Thus, 4 HMCLs harbored a hot-spot mutated 175, 248, 273, or 282 codon. PRIMA-1^Met LD₅₀ values were determined by incubating myeloma cells for 72 hours with a serial dilution of PRIMA-1^Met (1-200 μM) in culture media. A cell death assay was performed, using Apo2.7 staining.^24,27  The LD₅₀ values ranged from 3 up to more than 200 μM (median, 37 μM; supplemental Table 1). The median values for TP53^wt, TP53^mut, TP53^wt+mut, and TP53^neg HMCLs were, respectively, 22, 37, 49, and 38 μM, with no significant differences between the HMCLs (Figure 1A; P = .86). Myeloma cells lacking p53 expression, such as JJN3 and KMS11 (full length as well as all isoforms; data not shown), were as sensitive as TP53^wt or TP53^mut cells. Moreover, the shTP53 HMCLs, either TP53^wt (XG6, NCI-H929) or TP53^mut (XG5) myeloma cells, displayed no differential sensitivity to PRIMA-1^Met (Figure 1B), confirming that PRIMA-1^Met-induced myeloma cell death was not dependent on p53 expression. In contrast, the LD₅₀ values for nutlin3a in shTP53 XG6 and NCI-H929 TP53^wt HMCLs increased from 4 μM up to more than 10 μM (P < .05; supplemental Figure 1A), as described previously.²⁴  Myeloma-specific genomic heterogeneity characterized by recurrent 14q32 translocation and the overexpression of the CCND1, MAF, or MMSET genes was not significantly associated with the PRIMA-1^Met LD₅₀ values (Figure 1C; P = .55).

Despite a lack of correlation between sensitivity of myeloma cells to PRIMA-1^Met and p53 expression, using flow cytometry or western blotting, we assessed whether PRIMA-1^Met induced the expression of p53 target genes p21, Noxa, and DR5. We performed experiments using 3 HMCLs that expressed either a wild-type protein (XG6) or a TP53^R175H mutant protein previously reported to be reactivated by PRIMA-1^Met (OPM2) or no p53 protein (JJN3).¹⁶  Cells were treated with PRIMA-1^Met (LD₅₀ value) or nutlin3a (LD₅₀ value or 10 μM) as a control of the p53 pathway mobilization. PRIMA-1^Met did not induce p21 expression but did induce strong expression of Noxa in HMCLs, regardless of the p53 expression or status (Figure 2A). In contrast, nutlin3a increased p53, p21, and Noxa expression in XG6^wt and NCI-H929^wt cells (supplemental Figure 1B). Of note, the expression of p53, either mutated or wild-type, became undetectable after PRIMA-1^Met treatment. PRIMA-1^Met induced apoptosis, as revealed by the cleavage of caspases 2 and 3, and PARP. We previously reported that activation of a functional p53 pathway increased DR5 expression, as p53 directly bound to the DR5/TNFRSF10B gene.²⁴  PRIMA-1^Met did not increase DR5 expression; the median fold increases were 0.99 ± 0.09, 0.76 ± 0.25, and 0.89 ± 0.17 for XG6^wt, OPM2^R275H, and JJN3^neg, respectively (Figure 2B). As expected, nutlin3a increased the expression of DR5 in XG6^wt cells only (1.3 ± 0.1). These results confirmed that PRIMA-1^Met-induced myeloma cell death occurred independent of p53.

Because PRIMA-1^Met was reported to induce ROS, we assessed endogenous ROS production, using CellRox staining. XG6, OPM2, and JJN3 cells were incubated with 20 μM PRIMA-1^Met, with or without the ROS scavenger NAC.¹⁶  Indeed, the mean fluorescence intensity was increased in cells treated with PRIMA-1^Met, and this increase was inhibited by NAC (Figure 3A). The addition of NAC (5 mM) also inhibited Noxa increase, decrease in p53 expression, cleavage of both PARP and caspases, and cell death (Figures 2B and 3B). GSH, GSH-MEE (5 mM), also inhibited PRIMA-1^Met-induced cell death in all sensitive HMCLs (P < .001; Figure 3B). In contrast, nutlin3a did not increase ROS production, and NAC did not impair nutlin3a-induced cell death in 3 TP53^wt HMCLs (supplemental Figure 1C-D). The increase in Noxa expression was at the mRNA level: PRIMA-1^Met increased the median expression of NOXA mRNA 3.8-fold in OPM2^R175H cells (P = .026), and NAC fully prevented it (P = .022; Figure 3C). Silencing of Noxa inhibited PRIMA-1^Met-induced cell death by 89% ± 21% (P = .022) in OPM2^R175H and 36% ± 9% in LP1^E286K cells (P = .034), respectively (Figure 3D-E).

Because PRIMA-1^Met induced ROS production, we assessed whether endogenous GSH depletion by BSO, an irreversible inhibitor of γ-glutamyl cysteine-synthase (γ-GCS), would favor PRIMA-1^Met-induced cell death. Neither suboptimal concentration of PRIMA-1^Met (10 μM) nor BSO (0.5 mM) alone induced any significant cell death (median, 4% and 6%, respectively) in 27 HMCLs, but they had significant synergy in all but XG7^wt and JIM3^R273S (median value, 93%; P < .0001; Figure 4A and supplemental Figure 2A). However, the synergy was recovered in XG7^wt and JIM3^R273S cells by increasing the doses of PRIMA-1^Met (Figure 4B). The efficacy of the combination was not related to TP53 status (Figure 4C; P = .78), as confirmed in JJN3^neg cells in which BSO increased PRIMA-1^Met-induced caspase 3 activation and Noxa expression (supplemental Figure 2B). Addition of NAC also inhibited cell death induced by the combination of PRIMA-1^Met with BSO by more than 90% in XG6, OPM2, and JJN3 (Figure 4D). Of note, the addition of BSO increased ROS production induced by PRIMA-1^Met, as illustrated in OPM2 cells (Figure 4E). In contrast, BSO did not synergize with nutlin3a in TP53^wt HMCLs (supplemental Figure 1E), nor did it overcome resistance to nutlin3a in OPM2^R175H or JJN3^neg cells (data not shown).

Because depletion of GSH induced by BSO synergized with PRIMA-1^Met, we assessed whether PRIMA-1^Met impaired GSH content. As shown in Figure 4F, PRIMA-1^Met induced depletion in GSH content (GSH+GSSG). The decrease in GSH content induced by PRIMA-1^Met was 52% ± 10%, 75% ± 8%, and 100% ± 5% in OPM2, JJN3, and XG6 cells, respectively. As expected, BSO decreased GSH content by 80% ± 4%, 81% ± 5%, and 73% ± 2% in XG6, OPM2, and JJN3, respectively (Figure 4F). Combination of PRIMA-1^Met with BSO induced a total depletion in GSH content. In contrast, GSH content was slightly decreased (30% ± 10%) in nutlin3a-treated XG6 cells (supplemental Figure 1F).

Synthesis of GSH successively involves γ-GCS and GSS enzymes.³⁰  PRIMA-1^Met did not significantly modify expression of GCLM (regulatory subunit of γ-GCS) or GSS, indicating that GSH depletion was not caused by disappearance of any enzymes (supplemental Figure 3A). Expression of GSS, but not that of γGCS (GCLM or GCLC), was correlated with the LD₅₀ PRIMA-1^Met values (Figure 4G). Using siGSS RNA transfection, we showed that decrease in GSS expression in OPM2 cells increased their sensitivity to 25 μM PRIMA-1^Met by twofold: cell death was 18% ± 4% and 39% ± 5% in siControl and siGSS cells, respectively (P = .026; Figure 4H), and GSH content was 31 ± 10 and 20 ± 6 μM in siControl and siGSS cells, respectively (P = .03).

Because the microenvironment can protect tumor cells, we assessed whether stromal cells could prevent PRIMA-1^Met-induced myeloma cell death. Indeed, HS5 cells inhibited cell death induced by PRIMA-1^Met by 50% in the 3 cell lines (P = .03), but they were unable to prevent PRIMA-1^Met+BSO-induced cell death (supplemental Figure 3B). This inhibition was associated with a decrease in ROS production, in Noxa induction, and in GSH depletion (supplemental Figure 3C). Indeed, the mean global increase in GSH content induced by stromal cells was 1.23 ± 1.5, 2.0 ± 2.7 (P < .05), and 1.36 ± 0.5 in control, PRIMA-1^Met, and PRIMA-1^Met+BSO-treated cell lines (supplemental Figure 3D). In contrast, stromal cells did not protect XG6 from nutlin3a-induced cell death (supplemental Figure 3G).

The effect of PRIMA-1^Met was further assessed using primary cells (25 samples from 23 consecutive patients). Cell death was assessed by the loss of CD138 expression, and modulation of DR5 expression was used as a surrogate for the activation of the p53 pathway, as previously described and illustrated in Figure 5A.²⁴  PRIMA-1^Met, which significantly reduced myeloma survival (P = .0067 Wilcoxon matched-pairs signed rank test), induced a median cell death of 55% (Figure 5B and Table 1). The sensitivity of myeloma cells taken from PCL or MM patients was not significantly different (P = .34). Although NAC (n = 9) or GSH-MEE (n = 6) inhibited cell death (P = .009 and P = .036, respectively), the addition of BSO increased PRIMA-1^Met-induced cell death (87.5% of median cell death vs 32% with PRIMA-1^Met alone; P = .037; n = 6; and 0% with BSO alone). PRIMA-1^Met was more efficient than nutlin3a at inducing myeloma cell death (55% vs 7%, respectively; Figure 5B). The presence of a hemi-deletion of the 17p chromosome (del17p) negatively impairs overall patient survival, regardless of treatment, and TP53 mutations were found in samples harboring a del17p.⁸  Samples displayed a 17p deletion in a minor (<50%, considered as del17p−) or major (>50%, considered as del17p+) fraction of cells. Cell death induced by PRIMA-1^Met was not different in samples with or without del17p (P = .24, Mann-Whitney test; Figure 5C). PRIMA-1^Met induced a decrease in DR5 expression in primary samples (median fold increase, 0.92; n = 18; P = .0166; Figure 5D), whether harboring a del17p or not (median fold increase, 0.93 [n = 5; P = .0625], and median fold increase, 0.93 [n = 11; P = .0942], respectively; Figure 5E-F). Moreover, no correlation was found between the proportion of cells with del17p and the modulation in DR5 expression (P = .23, Spearman test; Figure 5G). In contrast, nutlin3a induced a significant increase in DR5 expression (median value, 1.25; n = 16; P < .001), which was significant in samples without del17p (median fold increase, 1.38; n = 11; P = .001), but not in samples with del17p (median fold increase, 1.10; n = 5; P = .31; Figure 5D-F). The increase in DR5 expression induced by nutlin3a was negatively correlated with the proportion of del17p within the samples (P < .001; Figure 5H). Finally, the increase in DR5 expression induced by nutlin-3a, but not by PRIMA-1^Met, was significantly higher in samples without del17p (Figure 5I). Thus, in primary myeloma cells, PRIMA-1^Met, in contrast to nutlin3a, did not increase DR5 expression and, instead, induced cell death irrespective of del17p status in synergy with BSO.

We assessed PRIMA-1^Met efficiency at preventing myeloma growth, both alone and in combination with BSO. SCID-beige mice bearing JJN3 tumor cells received either no treatment (control), PRIMA-1^Met (18 mg/kg, intravenous injection), BSO (10 mM, drinking water), or a combination of BSO and PRIMA-1^Met. PRIMA-1^Met (20 mg/kg) was previously shown to be efficient in preventing SAOS2 growth in vivo.¹⁵  BSO (30 mM) was evaluated in combination with melphalan.³¹  Mice were treated with 10 mM BSO in drinking water because this concentration was shown to reduce GSH content in the liver.³²  Treatments were performed daily for 4 days, stopped for 2 days, and then performed again for another 4 days (as indicated by arrows in Figure 6A). Mice were then killed at day 16 because control and BSO-treated tumors exceeded the authorized tumor load. Body weight was not significantly affected by any treatments: The variation of weight between day 5 and day 15 was 102.2% ± 3.0%, 102.4% ± 3.3%, 99.0% ± 2.2%, and 98.8% ± 2.7% for control, BSO-treated, PRIMA-1^Met-treated, and PRIMA-1^Met+BSO-treated mice, respectively (Figure 6B). The presence of BSO did not significantly modify tumor growth (P > .05, 2-way analysis of variance test). In contrast, PRIMA-1^Met significantly impaired tumor growth (P < .001), and its combination with BSO further inhibited tumor growth (P < .05). Tumor volume increased by 6.2-, 5.0-, 2.5-, and 1.05-fold in control, BSO-treated, PRIMA-1^Met-treated, and PRIMA-1^Met+BSO-treated mice, respectively. Tumor size tended to increase during treatment with PRIMA-1^Met; the mean tumor size was 93.0 ± 7.4 mm³ and 228.8 ± 67.3 mm³ at day 6 and day 16, respectively (P = .18, Wilcoxon matched-pairs signed rank test). In contrast, in the presence of PRIMA-1^Met and BSO, the tumors failed to increase in size; mean tumor volume was 91.4 ± 5.9 mm³ and 106 ± 30.6 mm³ at day 6 and day 16, respectively (P = .7). In another set of mice, tumors were removed after 4 days of treatment to analyze expression of caspase 3 (Figure 6C). The level of procaspase 3 was decreased in tumors from mice treated with PRIMA-1^Met±BSO (Figure 6D).

In this study, we assessed the efficacy of PRIMA-1^Met, a p53 targeted drug selected for its ability to reactivate mutant p53 proteins, in inducing apoptosis in myeloma cell lines (n = 27) and primary myeloma cells (n = 23), both in vitro and in vivo. The TP53 status of HMCLs and primary cells was assessed by direct sequencing of reverse transcription-polymerase chain reaction products and by the presence of deletion of the short arm of chromosome 17, del17p, as previously described.²⁴  We showed that PRIMA-1^Met induced myeloma cell death, irrespective of TP53 status. Moreover, HMCLs lacking p53 expression (as well as p53 isoforms, data not shown), such as JJN3 or KMS11, were as sensitive as TP53^mut or TP53^wt HMCLs. In primary cells, we showed that 10 μM PRIMA-1^Met induced cell death in primary samples (median cell death, 50%) without significant differences between samples with or without del17p, and PRIMA-1^Met, in contrast to nutlin3a, did not induce any increase in DR5 expression. Using 3 HMCLs expressing either a wild-type p53 protein (XG6) or a R175H mutant protein harboring a mutation that could be reactivated by PRIMA-1^Met (OPM2), or expressing no p53 protein (JJN3), we showed that PRIMA-1^Met did not reactivate p53 target genes such as CDKN1A or TNFRSF10B but, rather, increased expression of NOXA and induced production of ROS. In the presence of the ROS scavengers NAC or GSH-MEE, cell death, Noxa expression, and ROS production induced by PRIMA-1^Met were prevented. In contrast, NAC did not prevent cell death or Noxa increase induced by nutlin3a. We show that PRIMA-1^Met, but not nutlin3a, induced the depletion of GSH content. Moreover, irreversible inhibition of γGCS by BSO synergized with PRIMA-1^Met in cell lines, in primary samples, and in vivo. Involvement of GSH metabolism was further confirmed in GSS-silenced cells, which became more sensitive to PRIMA-1^Met. BSO did not synergize with nutlin3a or RITA (another MDM2/p53 inhibitor, data not shown), which confirms that the mechanism of cell death induced by PRIMA-1^Met is different from that induced by p53 stabilization.

Similar data showing the ability of PRIMA-1^Met to kill tumor cells independent of p53 were recently reported in several tumor models, including myeloma.^19-21  Of note, PRIMA-1^Met was isolated from the chemical library by a phenotypic screening that assessed the alternative sensitivity of Saos-2-His-273 cells carrying tetracycline-regulated mutant p53.³³  In contrast, nutlin3a was isolated in a biochemical screen via its direct binding to MDM2.³⁴  These different screening strategies may, at least in part, explain why PRIMA-1^Met has activities beyond binding to p53. Although PRIMA-1^Met was not specifically linked to p53, the drug was able to kill myeloma cells such as OPM2 or JJN3, which are resistant to the alkylating drugs melphalan and bendamustin.²⁶  Of major interest, Peng et al recently reported that PRIMA-1^Met targeted the selenoprotein thioredoxin reductase 1 and converted it into a NADPH oxidase enzyme.³⁵  Our data show that PRIMA-1^Met induced a depletion of GSH content (GSH+GSSG), suggesting an inhibition of GSH synthesis. Moreover, silencing of GSS increased sensitivity of OPM2 cells to PRIMA-1^Met, confirming that GSH plays a central role.

Because NAC or GSH-MEE can, either indirectly via γGCS or directly, increase intracellular GSH content, their ability to block PRIMA-1^Met-induced cell death could be related to an increase in GSH content. Indeed, PRIMA-1^Met-induced cell death was fully prevented by NAC or GSH-MEE, as well as partly by another ROS scavenger, L-ascorbic acid (50% of inhibition, data not shown). However, because NAC overcame the BSO-PRIMA-1^Met synergy and because BSO is an irreversible inhibitor or γGCS, NAC cannot only act as a source of GSH. This suggests that beyond their ability to neutralize ROS and regenerate GSH, NAC or GSH-MEE can directly interact with PRIMA-1^Met to neutralize it, as shown by Lambert et al.¹⁶  The depletion of GSH by BSO did not induce cell death and did not induce ROS production. Thus, PRIMA-1^Met-induced GSH depletion is necessary but not sufficient to induce cell death. Indeed, because PRIMA-1^Met can convert the antioxidant activity of thioredoxin into a prooxidant activity and because thioredoxin reductase and GSH peroxidase are essential to neutralize ROS, it appears that PRIMA-1^Met-treated cells do not have any functional detoxifying metabolism and cannot face ROS.

The mechanism of increase in Noxa expression induced by PRIMA-1^Met remains unclear. Recently, Saha et al reported that PRIMA-1^Met-induced cell death was mediated by Noxa, via p73 in myeloma cell lines.²¹  Although we also showed that Noxa silencing in OPM2 or LP1 cells decreased cell death induced by PRIMA-1^Met, we failed to find any increase in p73 expression (data not shown) or in p53/p73 target genes such as p21. P73 is not constitutively expressed in myeloma cells because of its silencing by DNA methylation, and it seems unlikely that PRIMA-1^Met inhibits DNA methylation within 24 hours in cell lines or in primary cells, which do not proliferate.³⁶  Myeloma cells are sensitive to ROS, and ROS production is involved in melphalan or bendamustine-induced myeloma cell death. Indeed, although NAC and GSH partly prevent melphalan-induced tumor death, BSO synergizes with melphalan.^26,37-39  The melphalan plus BSO combination was also assessed in phase 1 studies, which reported that the toxicity was acceptable.^40,41  However, no phase 3 study is available to determine the ultimate efficacy of that combination. We show in the JJN3-SCID xenograft model that 18 mg/kg PRIMA-1^Met was enough to inhibit tumor growth and that addition of BSO induced tumor regression. PRIMA-1^Met (100 mg/kg) was also shown to reduce tumor growth in the RPMI8226 xenograft model and to synergize with dexamethasone or doxorubicine.²¹  In both RPMI8226 and JJN3 xenograft models, mouse body weight remained stable, excluding the possibility of major toxicity from PRIMA-1^Met. Thus, evaluating the combined results, it appears that PRIMA-1^Met is effective against myeloma cells, both in vitro and in vivo, and that its efficacy is increased when it is used in combination with myeloma-approved drugs or BSO.

In JJN3-SCID mice, administration of PRIMA-1^Met was performed to mimic the reported assessment of PRIMA-1^Met/APR-246 in patients who received APR-246 once per day for 4 consecutive days over the course of 1 or 2 weeks.¹⁷  This phase 1 study, which involved patients with prostate cancer and hematologic malignancies, including myeloma, established that the maximum tolerated dose is 60 mg/kg. The authors reported that the tumor load of 2 patients with TP53^mut tumors decreased during treatment. Our in vivo experiments demonstrated that BSO significantly increased PRIMA-1^Met efficiency.

In summary, PRIMA-1^Met, alone or in combination with BSO, is potent against myeloma cells in vitro and in vivo, independent of p53 expression and TP53 status.

We thank Tumorothèque Institut Régional du Cancer Nantes-Atlantique (Centre Hospitalo-Universitaire and Institut de Cancérologie de l'Ouest, Nantes, France) for providing us with purified myeloma cells.

B.T. was supported by a grant from Institut National du Cancer. This work was supported by a grant from Fondation Françaisepour la Recherche Contre le Myélome et les Gammapathies Monoclonales.

Contribution: B.T., G.D., and S.M. performed experiments and participated in the design of the study; L.L. and C.G. provided purified myeloma samples and performed chromosomal analyses; S.M.-L. and T.O. performed in vivo experiments; P.M. and S.L.G. provided myeloma primary samples, participated in the design of the study, and reviewed the paper; M.A. participated in the design of the study and reviewed the paper; and C.P.-D. designed the study and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Catherine Pellat-Deceunynck, INSERM, UMR 892, Centre de Recherches en Cancérologie Nantes Angers, IRS-UN, 8, quai Moncousu, Nantes, BP70721 F-44007 France; e-mail: catherine.pellat-deceunynck@inserm.fr.