Results GADD45B Expression Denotes More Aggressive Disease in MM Given their key role in oncogenesis, we sought to investigate the downstream mechanisms mediating NF-κB survival signaling in MM. Because this signaling involves the induction of antiapoptotic NF-κB target genes, and we had previously identified the GADD45-family gene, GADD45B, as a transcriptional target of NF-κB encoding a potent and selective inhibitor of the JNK MAPK pathway and, therefore, of apoptosis (De Smaele et al., 2001; Papa et al., 2004), we investigated the involvement of this gene in MM. GADD45B was markedly upregulated in monoclonal CD138+ PCs from MM patients compared with monoclonal PCs from patients with monoclonal gammopathy of undetermined significance (MGUS), a premalignant condition (Kyle and Rajkumar, 2009), or healthy polyclonal PCs (Figure 1A), thus establishing a correlation between GADD45B mRNA expression and PC malignancy. Strikingly, when MM patients were stratified at diagnosis on the basis of the GADD45B mRNA expression in CD138+ cells, the cohort of patients expressing high levels of GADD45B exhibited dramatically shorter progression-free survival and significantly shorter overall survival (OS) than the cohort of patients expressing low levels of GADD45B, despite both groups of patients having been treated with the same velcade/melphalan/prednisone protocol (Palumbo et al., 2010) (Figures 1B and 1C). A similar correlation of GADD45B expression with poor clinical outcome was observed using two independent gene expression data sets of newly diagnosed MM patients, thus providing external validation of our findings (Broyl et al., 2010; Dickens et al., 2010) (Figures S1A and S1B available online). Collectively, these results establish a strong correlation between GADD45B expression and disease progression in MM and identify GADD45β as a hallmark of more aggressive disease. Constitutive NF-κB Activity Promotes the Survival of MM Cells by Inhibiting JNK Signaling GADD45β inhibits apoptosis by suppressing JNK signaling. It mediates this function by binding to the JNK kinase MKK7 and blocking its enzymatic activity by engaging the kinase catalytic pocket (Papa et al., 2004, 2007). This activity of GADD45β on the JNK pathway and the elevated GADD45B mRNA levels observed in monoclonal PCs from MM patients (Figure 1A) prompted us to investigate whether GADD45β mediated an antiapoptotic crosstalk between the NF-κB and JNK pathways in MM cells. We reasoned that cell extrinsic stimulation, as well as intrinsic oncogenic signals and NF-κB-pathway mutations, could result in the activation of signaling pathways beyond the IKK/NF-κB cascade, such as the JNK pathway (Davies and Tournier, 2012), which we and others have previously shown can trigger apoptosis in a manner that can be suppressed by NF-κB (De Smaele et al., 2001; Tang et al., 2001; Papa et al., 2004). Indeed, at least in principle, this mechanism could provide a basis for the addiction of MM cells to NF-κB for survival. In line with this hypothesis, in each of five heterogeneous MM cell lines and a B lymphoblastoid cell line, the silencing of the NF-κB subunit RelA (DiDonato et al., 2012) triggered JNK activation and apoptosis (Figures S2A–S2C), and blocking this activation with the JNK inhibitor SP600125 effectively protected NF-κB/RelA-silenced MM cells from cell death (Figures S2D and S2E). Similarly, the small hairpin RNA (shRNA)-mediated silencing of MKK7 rescued MM cells from the cytotoxic effects of each of two pharmacological inhibitors of IKKβ/NF-κB, and the protective effect of this silencing was virtually complete (Figure S2F). Collectively, these data indicate that constitutive NF-κB activity promotes the survival of MM cells by inhibiting MKK7/JNK signaling. GADD45β as a Pivotal Survival Factor Downstream of NF-κB and a Potential Therapeutic Target in MM To investigate the possible contribution of GADD45β to MM pathogenesis, we examined whether GADD45β mediated the NF-κB-dependent survival function and inhibition of JNK signaling in MM cells. As with PCs from patients (Figure 1A), GADD45B was expressed at high levels in MM cell lines compared with most other cancer cell lines tested (Figure S2G, discussed below). Importantly, this high GADD45B expression in MM cells markedly diminished upon the silencing of RelA, thus demonstrating its dependence on constitutive NF-κB activity (Figure 2A; Figure S2C). Moreover, similar to the effects of RelA-targeting hairpins (Figures S2A–S2C), the introduction of GADD45β-specific shRNAs, but not of MKK7-specific or of nonspecific shRNAs, induced potent JNK activation and apoptosis in all but two of the MM cell lines tested, namely, the RPMI-8226 and KMM-1 cell lines, which exhibited almost undetectable levels of GADD45β and significantly lower levels of MKK7 than those GADD45β-dependent MM cell lines (Figures 2B and 2C; Figures S2G–S2K, further discussed below). Similar results were observed using additional GADD45β-targeting hairpins, thus confirming the gene-silencing efficiency and specificity of the shRNAs used (Figures S2H, S2J, S2L, and S2M). Strikingly, the extent of JNK activation induced by the silencing of GADD45β in sensitive MM cell lines was similar to that observed with 12-O-tetradecanoylphorbol-13-acetate (TPA)/ionomycin stimulation, which potently induces JNK (Figure 2C; Figures S2I and S2J). By contrast, GADD45β downregulation had no effect on IKK/NF-κB, ERK, or p38 activity. As seen with the inhibition of IKKβ/NF-κB (Figures S2D and S2E), both the silencing of JNK1 and the treatment with SP600125 effectively reversed apoptosis in GADD45β-silenced MM cells (Figures 2D and 2E; Figures S2N–S2P). Hence, GADD45β promotes the survival of MM cells by inhibiting JNK-mediated apoptosis. Collectively, these findings identify GADD45β as an essential NF-κB-regulated survival factor and selective MKK7/JNK-axis inhibitor and, therefore, as a potential therapeutic target in MM. Development of D-Tetrapeptide Inhibitors of the GADD45β/MKK7 Complex Given the essential antiapoptotic role of GADD45β in MM and our previous results showing that GADD45β suppresses JNK signaling and apoptosis by blocking MKK7 via direct physical interaction (De Smaele et al., 2001; Papa et al., 2004, 2007), we aimed to develop selective inhibitors of this protein-protein interaction in order to induce cytotoxic JNK signaling in MM cells. We screened a simplified combinatorial library of 20,736 L-tetrapeptides to select compounds capable of disrupting the GADD45β/MKK7 complex (Figure S3A and Table S1). Iterative deconvolution of this library in ELISA competition assays, followed by secondary screening and optimization of the resulting hits, yielded two acetylated L-tetrapeptides of similar structure, namely, Ac-LTP1 and Ac-LTP2, which disrupted the GADD45β/MKK7 complex, in vitro, with remarkable half-maximal inhibitory concentration (IC50) values in the subnanomolar range (Figure 3A; Figure S3B and Table S2), in line with the top-end potencies of other hits isolated from similar peptide library screens (Houghten et al., 1999). Next, we examined whether these two L-tetrapeptides retained in vitro potency upon synthesis in the D configuration, a strategy used successfully in some cases to render small peptides resistant to proteolysis (Zhou et al., 2002; Nickl et al., 2010). Strikingly, as shown in Figures 3A and 3B, the D-enantiomers of Ac-LTP1 and Ac-LTP2 (termed Ac-DTP1 and Ac-DTP2, respectively) displayed no loss of activity in vitro and, unlike their L counterparts, were highly stable in human serum, even after prolonged incubation. Coimmunoprecipitation assays confirmed the potent and specific inhibitory effect of Ac-DTP1 and Ac-DTP2 on GADD45β/MKK7 complex stability in vitro (Figure 3C). By contrast, control D-tetrapeptides had no such effect on stability, thus demonstrating the specificity of the inhibitory activities of Ac-DTP1 and Ac-DTP2. Crucially, the disruption of the complex by these two active D-peptides completely reversed the GADD45β-mediated inhibition of MKK7, fully restoring the kinase catalytic activity (Figure 3D, top). Importantly, none of Ac-DTP1, Ac-DTP2, or any of the control D-tetrapeptides affected MKK7 activity in the absence of GADD45β (Figure 3D, bottom). Collectively, these findings identify a downstream, drug-targetable module in the NF-κB pathway and confirm the potential of our pharmacological approach for inducing cytotoxic MKK7/JNK signaling in cells that rely on GADD45β for restraining MKK7 activation. To verify the therapeutic potential of GADD45β/MKK7 inhibitors, we aimed to improve their cell penetration. We replaced the N-terminal acetyl group of Ac-DTP1 and Ac-DTP2 with a benzyloxycarbonyl [Z] group, thereby generating z-DTP1 and z-DTP2, respectively (Table S3), which retained high activity and stability in vitro (Figures S4A and S4B) and, importantly, also exhibited potent cytotoxic activity across a panel of genetically heterogeneous MM cell lines (Figures 4A and 4B; Figures S4C–S4F and Table S3). Significantly, z-DTP1 and z-DTP2, but not a control D-tetrapeptide, induced potent and dose-dependent toxicity in all of the MM cell lines tested, except the two expressing nearly undetectable levels of GADD45B and low levels of MKK7 (further discussed below), exhibiting IC50 values in the sensitive MM cell lines in the low nanomolar to low micromolar range (Figures 4A and 4B; Figures S2G, S4E, and S4F). Consistent with the protective mechanism mediated by GADD45β (Figure S2G), the z-DTP1- and z-DTP2-afforded killing of sensitive MM cells was due to the induction of apoptosis (Figure 4C; Figures S4G and S4H). Importantly, both active D-peptides retained potent and cancer-selective activity in primary PCs from MM patients (Figure 4D, left). Crucially, these D-peptides also exhibited an apparently complete lack of toxicity to normal cells, even when used at very high concentrations (i.e., 100 μM; Figure 4D, right; Figure S4I). Hence, D-tetrapeptide antagonists of the GADD45β/MKK7 complex show exceptionally high activity and cancer cell specificity in terms of apoptosis induction in MM cells, without displaying any apparent toxicity to normal cells. The Development of DTP3: A GADD45β/MKK7 Inhibitor with Improved Bioavailability To improve the bioavailability of D-peptides in vivo, while retaining high cellular activity and specificity toward the GADD45β/MKK7 complex, we used a chemical optimization strategy based on structure-activity relationship and pharmacophore analyses (Figures S5A–S5C and Table S4). By combining these methods, we developed DTP3, a D-tripeptide with a molecular weight of 525 Da (Figure 5A), which retained all the main characteristics of the parental D-tetrapeptides in terms of bioactivity and specificity, including subnanomolar activity and high stability in vitro and potent and selective capacity to kill MM cells via apoptosis (Figures S5D–S5H), while exhibiting a superior pharmacokinetic profile compared with the parent molecules (discussed below). We further evaluated the modality and specificity of the binding of DTP3 to the GADD45β/MKK7 complex. Circular dichroism (CD) studies demonstrated that DTP3, but not a control scrambled (SCRB) D-tripeptide, can physically interact with and cause a significant conformational change of the structure of this complex, as well as of the isolated MKK7 protein, in a dose-dependent manner (Figure S5I). By contrast, DTP3 had no effect on the CD signal of GADD45β (Figure S5I). MALDI-TOF mass spectrometry confirmed the ability of DTP3 to bind to MKK7, but not to GADD45β (Figure S5J). Spectrofluorimetric analyses yielded similar results and, furthermore, established the 1:1 stoichiometry and low equilibrium dissociation constant (KD) value of the DTP3 interaction with MKK7 (Figure 5B; Figure S5K), thus underscoring the strength and specificity of this interaction. Conversely, the SCRB D-tripeptide had no impact on the fluorescence emission spectrum of MKK7, thus reaffirming the specificity of the effects of DTP3. Computational analyses predicted the presence of a pocket of MKK7, also found on the GADD45β/MKK7 complex, that can bind to DTP3 (Figures S5L–S5T). Collectively, these results demonstrate that DTP3 has the capacity to physically interact with MKK7, both in isolation and within the complex with GADD45β, and support a model whereby, upon binding to MKK7, DTP3 dissociates the GADD45β/MKK7 complex via an allosteric mechanism, potentially involving a conformational rearrangement of the kinase. The High Target Specificity of DTP3 for the GADD45β/MKK7 Complex In a panel of tumor cell lines of different tissues of origin, the sensitivity to DTP3-induced killing correlated with a very high degree of significance with the mRNA levels of GADD45B (Figure 5C; Figures S2G and S5U), which our data have shown is a pivotal NF-κB-regulated inhibitor of MKK7/JNK signaling and apoptosis in MM cells (Figures 2B–2E; Figures S2H–S2J and S2N–S2P). Significantly, DTP3 displayed potent and selective activity in both MM and non-MM cell lines exhibiting high levels of GADD45B expression, whereas it was completely inactive, even at high micromolar concentrations, in tumor cell lines featuring low GADD45B expression (Figures S2G and S5F–S5H). In the context of MM, the RPMI-8226 and KMM-1 cell lines, which express very low levels of GADD45β and low levels of MKK7 and are consequently unaffected by GADD45β-targeting hairpins, were completely refractory to DTP3-induced killing (Figure 2B; Figures S2G–S2I, S2K, and S5F–S5H). As seen with GADD45β-silencing shRNAs (Figure 2C; Figures S2H and S2I), treatment with DTP3 effectively activated JNK but not p38, ERK, or IKK/NF-κB signaling in sensitive, GADD45β-dependent MM cell lines, whereas it did not affect any of these pathways in DTP3-resistant, GADD45β-independent MM cell lines, such as RPMI-8226 (Figure 5D; Figure S5V; see also Figure S4H, z-DTP2). Strikingly, the magnitude of the effects of DTP3 on the JNK pathway was similar to that observed with tumor necrosis factor α (TNFα), a potent inducer of JNK (Figure 5D; Figures S4H and S5V, TPA/ionomycin). By contrast, a negative control D-peptide had no effect on JNK signaling (Figure S5V). As expected from the in vitro data (Figure 5B; Figure S5I, middle, bottom; Figure S5J, left), these activating effects of DTP3 on the JNK pathway hinged upon the ability of this D-tripeptide to bind to MKK7 in cells, as shown by DTP3 pull-down assays (Figure 5E). These effects were not observed with the SCRB control peptide, nor were they observed with a pull-down control in the absence of peptide, thus demonstrating the specificity of the binding of DTP3 to endogenous MKK7. Importantly, knocking down the GADD45β target kinase, MKK7, or its downstream effector, JNK1, completely abolished the cytotoxic activity of DTP3 in sensitive MM cell lines (Figure 5F; Figures S5W and S5X), thus excluding any off-target toxicity of DTP3 in cells. DTP3 also lacked any off-target effect when profiled in kinase assays against a panel of 142 human kinases (Table S5). Hence, bioactive D-peptides selectively induce apoptosis in MM cells with functional MKK7 and elevated GADD45β expression by activating JNK signaling via MKK7. These findings also correlate the pharmacological activity of DTP3 with a tumor cell dependence on GADD45β for survival and establish the exceptionally high target specificity of this D-tripeptide for the GADD45β/MKK7 complex in cells. The Potent and Cancer-Selective Activity of DTP3 in MM Cells from Patients To verify the suitability of DTP3 for the treatment of human diseases, we tested its activity and selectivity in MM PCs from patients. As seen with the active D-tetrapeptides (Figure 4D, left), DTP3 effectively killed primary MM PCs at low nanomolar concentrations (Figures 6A and 6B). Importantly, it retained potent and selective cytotoxic activity in these cells upon stimulation with interleukin-6 or insulin growth factor 1 or coculture with bone marrow stromal cells (BMSCs) (Figure 6C; Figures S6A and S6B), which promote the survival of MM cells (Hideshima et al., 2005). In order to compare the efficacy and cancer cell specificity of DTP3 with the corresponding parameters of bortezomib, the current gold-standard treatment for MM (Chen et al., 2011), we defined an “in vitro therapeutic index” (Figure 6A, bottom). As shown in Figure 6A (top), DTP3 had a similar IC50 value to bortezomib in primary MM PCs but, importantly, unlike bortezomib, which barely discriminated between malignant and normal cells, had no toxicity to normal cells. Strikingly, because of this cancer cell selective target specificity, DTP3 had an in vitro therapeutic index that was greater, by more than two orders of magnitude, than that of bortezomib (Figure 6A, bottom). DTP3, in fact, could also distinguish between PCs from MM and Waldenström’s macroglobulinemia (WM) patients (Kyle and Rajkumar, 2009), in line with the GADD45B expression levels in these cells, whereas bortezomib could not (Figures 6A and 6B; Figure S6C). Similarly, DTP3 displayed far more potent activity in primary MM PCs and far less toxicity to normal cells than the IKKβ inhibitor, PS-1145 (Figures 6C and 6D; Figures S6A and S6B; note the different concentrations of DTP3 and PS-1145 used). Because the standard MM treatment consists of combination therapy, and nearly all patients will relapse and/or develop drug resistance at some point, we evaluated the potential of DTP3 to operate in these settings. As shown in Figures S6D and S6E, DTP3 displayed synergistic activity with bortezomib in two different MM cell lines, exhibiting a combination index of 0.21 in U266 cells and of 0.56 in KMS-12 cells, suggesting that it could find indication in the clinic in combination with bortezomib (Chou, 2006). Importantly, DTP3 also retained full therapeutic efficacy in MM cell lines that were resistant to conventional MM treatments, such as dexamethasone, bortezomib, and lenalidomide (Bjorklund et al., 2011, 2014, Rückrich et al., 2009) (Figures S6F and S6G). Together, these results provide compelling evidence of the high therapeutic potential of DTP3 in MM patients. The Potent Therapeutic Efficacy and Excellent Tolerability of DTP3 against MM In Vivo DTP3 showed high aqueous solubility and very high stability in human serum, owing to its resistance to proteolysis, with a good pharmacokinetic profile and excellent in vivo tolerability, suitable for a therapeutic purpose (Figure 7A; Figure S5D and Table S6). Remarkably, in a plasmacytoma model, treatment with DTP3 at the dose of 14.5 mg/kg/day virtually eradicated established subcutaneous myeloma xenografts in mice, in the absence of any apparent side effects (Figures 7B and 7C; Figure S7A and Table S6). Similar results were obtained in a second plasmacytoma model, generated using a different MM cell line (Figures S7C and S7D). At the experimental end point, on day 28, all the control mice had developed large local tumors, whereas all the mice in the DTP3-treated cohort had shown a dramatic shrinkage of the tumors (Figures 7B and 7C; Figures S7C and S7D). This therapeutic effect of DTP3 was due to the potent and tumor-selective induction of JNK activation and apoptosis (Figures 7D and 7E; Figure S7B), as shown by the appearance of phosphorylated JNK, as early as 24 hr of the onset of treatment with DTP3, but not with PBS, followed by the appearance, starting on day 3, of caspase-3 and PARP-1 proteolysis products. Coincident with these events, apoptotic cells became evident in the tumor tissue at day 3 (Figure 7D; Figure S7B). As well as the tumor-ablative effects of DTP3 (Figure 7B; Figure S7A), the extent of this JNK-associated, tumor cell apoptosis markedly increased in magnitude over time (Figure 7D; Figure S7B). Importantly, DTP3 retained potent anticancer activity in an orthotopic xenograft model of MM, which more faithfully recapitulates the human disease. All the control mice developed severe limb paralysis and died within 32 days of treatment start, resulting in a median OS of 26 days (Figure 7F). Strikingly, DTP3 administration over a period of 8 weeks, at a dose of 29 mg/kg/day, extended the median OS of the mice past the experimental end point on day 161, without producing any apparent side effect, thus demonstrating the potent therapeutic efficacy of DTP3 against MM, in vivo, and the excellent tolerability of this agent at doses that achieve full therapeutic efficacy. Collectively, together with the data in primary MM cells (Figures 6A–6D; Figures S6A and S6B), these results underscore the potency, safety, and cancer cell specificity of the pharmacological approach targeting the GADD45β/MKK7 complex in MM and identify DTP3 as a therapeutic selectively inhibiting the NF-κB survival pathway in cancer (Figure 8).