Modulation of NOXA and MCL-1 as a Strategy for Sensitizing Melanoma Cells to the BH3-Mimetic ABT-737

Metastatic melanoma is highly therapy-resistant and standard chemotherapy has failed in clinical trials. Targeted therapy with new drugs, such as the BRAF inhibitor PLX4720/4032/vemurafenib (1, 2), holds great promise but suffers from onset of resistance after about 7 months (3–7). Defective apoptotic pathways are thought to be one barrier to effective systemic treatment of melanoma. For example, the frequent inactivation of ARF subverts p53 activity (8) and certain antiapoptotic BCL-2 family proteins are upregulated (9, 10). Although BCL-2 in particular has attracted much attention, treatment targeting it (e.g., oblimersen; Genta) had only modest impact on survival in clinical trials (11, 12). BCL-2 is only one of several related apoptosis proteins with altered expression in melanoma, for example, BCL-XL and MCL-1 are elevated during melanoma progression (9, 10) and these might prove superior targets for melanoma therapy.

ABT-737 (Abbott Laboratories) is a BH3-mimetic with a similar binding profile to the proapoptotic BH3-only protein BAD; it strongly inhibits the activity of BCL-2, BCL-XL, and BCL-w but not MCL-1 or BFL-1/A1, for which it has negligible affinity (13, 14). ABT-737 is effective as a single agent in preclinical models of lymphomas and small-cell lung carcinomas (SCLC)—cancers with low MCL-1 expression (13, 15–17). It also sensitizes tumors to the effects of chemotherapy and radiotherapy (13, 15, 16, 18, 19).

Apoptosis is initiated by the release of apoptogenic mediators, such as cytochrome C, from the mitochondrion, controlled by proapoptotic proteins BAK and BAX (20). These are inhibited by BCL-2 and related proteins; 3 of them, MCL-1, BFL-1, and BCL-XL, specifically regulate BAK (21). In the presence of ABT-737, BCL-XL is inhibited, so BAK activity is determined by MCL-1 and BFL-1. There is clear evidence that MCL-1 is a major factor in sensitivity to ABT-737 (e.g., 14) but the role of BFL-1 is less clear. Evidence currently suggests that BFL-1 functions in a similar manner to MCL-1 (22) and its expression is correlated with sensitivity to ABT-737 (23, 24).

Prior studies suggest that drugs that repress MCL-1 or induce NOXA sensitized cells to ABT-737 in a variety of tumor models. For example, flavopiridol treatment leads to transcriptional repression of MCL-1 via upregulation of E2F1 (25). While flavopiridol showed efficacy against melanoma in vitro (26), it was ineffective for patients with metastatic melanoma as a single agent (27). Melphalan is an alkylating agent used for isolated limb perfusion treatment of metastatic melanoma and may be preferable to dacarbazine/temozolomide for melanomas with high O₆-alkylguaninetransferase activity (28). Melphalan has been observed to decrease MCL-1 levels in myeloma cells, sometimes resulting in cleavage of the protein (29).

In the present study, we investigated the extent to which MCL-1 and BFL-1, and their antagonist NOXA, affect sensitivity of melanoma cells to ABT-737 and thus whether combination therapy targeting MCL-1 and BFL-1 in addition to BCL-2, BCL-XL, and BCL-w has promise in the context of melanoma. The activity of MCL-1 and BFL-1 was manipulated by overexpression of NOXA, by short hairpin RNA (shRNA)-mediated knockdown of MCL-1 or by cytotoxic drugs. The contribution of BFL-1 to sensitivity to ABT-737 was examined by ectopic expression of NOXA variants with altered MCL-1 and BFL-1 affinities. The consequences for sensitivity to ABT-737 were assayed in 2- and 3-dimensional models in vitro and contrasted with observations in a tumor xenograft model.

The human melanoma cell lines IgR3, SKMel28, MelCV, MM200, MelRM, and Me4405 were grown as described (30) and authenticated using AmpFISTR profiling as described (31). C8161, 1205Lu, and WM793 (32) were grown as described (33); the latter were authenticated by STR fingerprinting (Genomics Facility, The Wistar Institute, Philadelphia, PA). MelRM/NOXA and MM200/NOXA clones matched the parent lines in microsatellite analyses of 12 loci conducted by the Australian Genome Research Facility Ltd, which also confirmed the purity. Unnecessary passaging was avoided; experiments with manipulated cell lines were carried out within 2 passages of reference stocks.

Cells were lysed in 10 mmol/L Tris-HCl, pH 8, 5 mmol/L EDTA, 0.1% Triton X-100 and protease inhibitor cocktail (Sigma P8340, per directions), 1 μL per 10⁶ cells. Tumors were homogenized and sonicated in 10 mmol/L Tris, pH 7.4, 10 mmol/L KCl, 1.5 mmol/L MgCl₂, and 0.5% SDS plus protease inhibitor cocktail. Protein was quantified by the Bradford method; 30 μg per lane was analyzed by reducing SDS-PAGE on 15% gels and electroblotted to polyvinylidene difluoride (PVDF) membrane (Millipore). Membranes were blocked in 5% blotto and probed with monoclonal antibodies against: NOXA (114C307, Abcam), HA-tag (HA.11 16B12, Covance), MCL-1 (clone 22, BD Pharmingen), BCL-2 (clone 100, BD Pharmingen), BCL-XL (2H12, BD Pharmingen), BFL-1 (51B2, from D.C. Huang), and α-tubulin (YL1/2, Abcam). Following washes with PBST, bound antibodies were detected with goat anti-mouse IgG-HRP or goat anti-rat IgG-HRP (Santa Cruz Biotechnology, sc-2005 and sc-2065) and enhanced chemiluminescence (ECL).

MM200, Me4405, MelRM, and/or C8161 cells were transduced with VSVG-packaged lentiviral shRNA constructs from the MISSION Library (Sigma-Aldrich; MCL-1: TRCN0000005514-18; BFL-1: TRCN0000033494-98; NOXA: TRCN0000150555, TRCN0000151311, TRCN0000153637, TRCN0000155570, TRCN0000155978), per manufacturer's directions.

Hemagglutinin (HA)-tagged human NOXA cDNA was inserted into retroviral vector pQC-X-IRES-CD4, a derivative of pQCXIX (Clontech) containing a human CD4 marker. The mNoxaB variants in pMIG retroviral vector carrying an EGFP marker have been described (14). Constructs were packaged with Phoenix amphotropic cells (34) and used to transduce melanoma cells. Transduced cells were sorted by flow cytometry, expanded and frozen as reference stocks. These remained more than 99% positive on retest. Single-cell clones were derived from pools by serial dilution and facilitated with 40% v/v conditioned medium from the parent cell line.

Proliferation inhibition assays were conducted as described (35, 36), on 4 or more occasions. Sensitization to ABT-737 by melphalan or flavopiridol, and vice versa, was assessed by proliferation inhibition assays testing all combinations of a range of ABT-737 with flavopiridol or melphalan concentrations, as shown, applied simultaneously.

Melanoma spheroids were prepared as described (37, 38). After embedding into a gel of bovine collagen I, spheroids were treated with ABT-737 and/or flavopiridol at the concentrations and times indicated. Termination of the experiments: Spheroids were washed 3 times in PBS before live-dead staining with 4 mmol/L calcein-AM and 2 mmol/L ethidium bromide (Invitrogen) for 1 hour at 37°C. Images were taken using a Nikon-300 inverted fluorescence microscope. For reculture, spheroids were plucked from the matrix with a pipette and plated in complete medium for expansion by 2D adherent growth. Spheroid sizes were quantified by image analysis using ImageJ software (NIH, Bethesda, MD).

The flanks of 6-week-old CB17 NOD/SCID male mice were injected subcutaneously with 1 × 10⁶ MM200 cells that overexpressed NOXA or had MCL-1 knocked down, in 100 μL complete medium. Controls were cells transduced with an ineffective MCL-1 knockdown (shRNA #14) with similar sensitivity to ABT-737 as untransduced cells. One week postengraftment, the mice were treated intraperitoneally with 75 mg/kg ABT-737 or vehicle only, daily for 21 days, 10 mice per group. Vehicle was 30% (v/v) propylene glycol, 5% (v/v) Tween-80, and 65% (v/v) of 5% (w/v) dextrose in water. The pH was adjusted to 1.0 with HCl. ABT-737 was dissolved in dimethyl sulfoxide at 125 mg/mL, diluted 10-fold with 9 volumes of vehicle, and then pH readjusted to 4.0 with 1 mol/L HEPES. Mice were weighed daily and tumor growth measured with digital calipers. Mice were sacrificed when tumors reached volume of 1 cm³ or became ulcerated. Experiments were approved by the University of Sydney Animal Ethics Committee, Camperdown, NSW, Australia.

Comparisons of IC₅₀ values of manipulated cell lines with parent or control lines were done by t tests. As drug resistance raw data are by nature positively skewed, tests were conducted on log-transformed data. Differences between xenograft or spheroid growth rates were detected as interactions between time and treatment factors in 2-way mixed ANOVA with time as a repeated measure.

The sensitivity of melanoma cell lines to ABT-737, versus other tumor lines, was tested by proliferation inhibition assay. Of the 9 melanoma lines examined, 8 were insensitive to ABT-737 compared with, for example, lymphoid cell lines (Table 1). However, overexpression of NOXA in 6 melanoma lines by retroviral transduction (Fig. 1A) dramatically sensitized all of them to ABT-737 (Fig. 1B and Table 1), in most cases ≥100-fold. The least sensitization, in the Me4405 line, was still about 10-fold. This line had higher endogenous NOXA and MCL-1 levels (Fig. 1A). The strong sensitization was observed in both transduced cell pools and in single-cell clones derived from them (Table 1) and cannot be attributed to effects of NOXA overexpression on growth rates (Supplementary Fig. S1) or nonspecific sensitivity to apoptotic stimuli other than ABT-737 such as serum withdrawal (Supplementary Fig. S1) or DNA-damaging agents cytarabine and mitomycin C (Supplementary Table S1). NOXA overexpression had no consistent effect on levels of other proapoptotic (e.g., PUMA) or antiapoptotic proteins, although lines with more MCL-1 tolerated more NOXA overexpression (Fig. 1A; Supplementary Fig. S2). In contrast to overexpression, NOXA knockdown modestly decreased sensitivity to ABT-737 (Table 1).

NOXA directly antagonizes MCL-1 and BFL-1; knockdown of MCL-1 in 2 melanoma cell lines (Fig. 1C) also sensitized to ABT-737 (up to 9-fold), although not as much as NOXA overexpression (Table 1). Incomplete MCL-1 knockdown is likely part of the reason. Second, MCL-1 knockdown was accompanied by reduction of NOXA levels (Fig. 1C), counteracting the knockdown. The most obvious mechanism for this is selection of cells with lower NOXA levels, given that MCL-1 knockdown promotes apoptosis. A third possibility is that NOXA has other actions contributing to sensitization to ABT-737, including inhibition of BFL-1 function. Indeed, BFL-1 knockdown (Fig. 1D) modestly sensitized MM200 cells to ABT-737 (up to 2.4-fold) but had little effect on Me4405 (Table 1), MelRM, or C8161 cells (not shown).

The relative importance of functional MCL-1 and BFL-1 inhibition is difficult to evaluate with incomplete knockdowns and so was assessed using Noxa-Bim chimeras containing Noxa BH3 domains in an inert Bim backbone; mNoxaB binds and inhibits MCL-1 only, whereas mNoxaB E74F inhibits both MCL-1 and BFL-1 (14). Overexpression of either variant (Fig. 1E) resulted in dramatic sensitization to ABT-737 (Table 1), as observed for wild-type NOXA. Thus, although these were the cell lines wherein BFL-1 knockdown had most effect, inhibition of MCL-1 alone appears to be sufficient for the dramatic sensitization to ABT-737 in this context.

Our 3D spheroid model mimics tumor architecture and microenvironment (37, 38) and is used for investigating drug effects on growth, invasion, and viability of melanoma cells, which mimic those of treatment in vivo (2, 39, 40).

Spheroids derived from sorted pools of MM200/NOXA cells were strongly sensitized to ABT-737 in 3-day assays, reflected in dose-dependent inhibition of proliferation and invasion as well as decreased viable cell staining (calcein-AM) and increased dead cell staining (ethidium bromide) compared with control spheroids (Fig. 1G). NOXA overexpression also did not affect growth or invasion of the matrix (Supplementary Fig. S3).

Because increased NOXA expression strongly sensitized melanoma cells to ABT-737, its combination with cytotoxic drugs that induce NOXA or repress MCL-1 may be an attractive therapeutic strategy. Drugs that have been reported to alter MCL-1 levels, melphalan (29) and flavopiridol (25), were therefore tested as sensitizers to ABT-737.

In our hands, melphalan yielded only modest and transient increases in NOXA levels in MM200 and Me4405 cells, which disappeared within 24 hours (Fig. 2A). Flavopiridol had greater and more lasting effects in both cell lines tested but they were strongly dose-dependent—lower doses increased NOXA levels whereas higher doses had the opposite effect (Fig. 2A). Thus, the net effects were time- and dose-dependent. Moreover, the changes in NOXA levels appeared to be mirrored by similar changes in MCL-1 levels, which would tend to counteract each other.

Nevertheless, flavopiridol sensitized both MM200 and Me4405 cells to ABT-737 by more than 12- and 8-fold, respectively (Table 2; Supplementary Fig. S4A). The sensitization of MM200 cells to ABT-737 was largely ablated by prior knockdown of NOXA, from 15- to 1.3-fold (Fig. 1F and Table 1 and Supplementary Table S2). Moreover, it occurred close to the IC₅₀ value for flavopiridol, consistent with the dose–response for NOXA induction (Fig. 2A). Flavopiridol might thus be an effective sensitizer at physiologically attainable concentrations. Melphalan was less effective in this respect (consistent with its lesser effect on NOXA and MCL-1 levels) but still, at concentrations near to its IC₅₀, sensitized Me4405 cells to ABT-737 by 3- to 4-fold (Table 2; Supplementary Fig. S4C).

The same data set showed the effects of ABT-737 on sensitivity of the cells to the cytotoxic drugs (Table 2; Supplementary Fig. S4B and S4D). Even at very high concentrations of ABT-737, there was little sensitization to either flavopiridol or melphalan. There can be little doubt that ABT-737 was inhibiting BCL-2, BCL-XL, and BCL-w, as the K_i for these proteins was reported as less than 1 nmol/L (13) and we observed that low concentrations of ABT-737 in complete medium were sufficient to kill melanoma cell lines that overexpressed NOXA (Fig. 1B and Table 1). Thus, reported cooperativity between ABT-737 and cytotoxic drugs can be due to sensitization to ABT-737 rather than the converse.

The 3D melanoma spheroids were also sensitized to ABT-737 by flavopiridol. Growth of the spheroids (increase in radius, Fig. 2B) was expressed relative to the controls (0 μmol/L ABT-737), analogous to cytotoxicity curves for flat cultures (Fig. 2C). In the presence of 80 to 160 nmol/L flavopiridol, there was obvious sensitization to ABT-737. This also manifested as reduced cell viability (Fig. 2F). At 320 nmol/L flavopiridol, there was no additional effect from ABT-737 on spheroid growth but still an enhancement of cell killing (Fig. 2F). The converse sensitization to flavopiridol by ABT-737 was notably weaker, as observed in 2D culture (Fig. 2D and E; compare 2D culture results in Supplementary Fig. S4 and Table 2). A concentration of 1 μmol/L ABT-737 was sufficient for this modest sensitization, consistent with on-target effect.

NOXA overexpression or MCL-1 knockdown markedly sensitized melanoma cells to ABT-737 in vitro, so it was of interest to ascertain whether the same occurs in vivo. MM200 cells in which MCL-1 was knocked down, or which overexpressed NOXA, were engrafted into CB17 NOD/SCID mice and treated with ABT-737 daily for 21 days. Surprisingly, ABT-737 had little effect on xenograft growth under this regimen, compared with vehicle-treated mice (Supplementary Fig. S5). Initial experiments with MM200/NOXA cells were carried out with transduced cell pools purified by flow cytometry (Supplementary Fig. S5A). It was anticipated that xenografts of a single-cell clone would exhibit greater sensitivity to ABT-737 simply because it harbored less heterogeneity than the pool but, again, there was no difference in rate of growth of the ABT-737- and vehicle-treated tumors (Fig. 3A).

Blood samples from treated mice showed thrombocytopenia (0.62 ± 0.07 × 10⁶/μL vs. 1.3 ± 0.11 × 10⁶/μL, P < 0.001) and leukocytosis (6.3 ± 0.43 × 10³/μL vs. 3.3 ± 0.24 × 10³/μL, P < 0.001; Fig. 3B and C), confirming uptake of ABT-737 into the circulation with cellular consequences similar in magnitude to those obtained by others using the same treatment regimen (41). The xenografts were always vascularized.

There was no evidence that insensitivity to ABT-737 in vivo reflected selection during engraftment for cells expressing less NOXA, as the levels in the tumors at time of sacrifice were not different from that of the injected cells, whether the tumors were derived from a single-cell clone (Fig. 3D) or even transduced pools (Supplementary Fig. S6). Explants grown from tumors retained sensitivity to ABT-737 in 2D cytotoxicity assays (Fig. 3E). Thus, neither the engraftment itself nor treatment with ABT-737 selected for stable resistance to ABT-737, suggesting that insufficient amounts reached the tumors in vivo.

The 3D spheroid model provides insight into the difference in response of NOXA-overexpressing melanoma cells to ABT-737 in vitro and in vivo. As noted above, spheroids derived from NOXA-transduced MM200 cell pools that were treated with ABT-737 showed a strong initial response (Fig. 1G). Similar results were obtained with spheroids derived from homogeneous single-cell clones of MM200/NOXA (Fig. 4) and C8161/NOXA cells (Supplementary Fig. S7). However, by 10 days of treatment with medium and drug replenishment every 3 days, resistant cells had nevertheless started to proliferate and invade the matrix and there was little impact on cell viability except at the highest ABT-737 concentration (16 μmol/L; Fig. 4B). Thus, the melanoma cells became resistant quickly. This was confirmed by expanding the spheroids in flat culture; in 2D cytotoxicity assays, cells from ABT-737–treated spheroids were resistant to ABT-737, whereas cells from untreated spheroids remained sensitive (Fig. 4D). Moreover, immunoblotting showed that the resistant cells had lost overexpression of NOXA (Fig. 4E).

To model the growth of the xenografts, spheroids were then allowed 7 days unfettered growth before treatment with ABT-737. It is striking that there was far less impact on proliferation or invasion of these larger spheroids, but a noticeable effect on viability in the periphery of the spheroids (Fig. 4C; Supplementary Fig. S7D), suggesting that diffusion of ABT-737 into larger melanoma masses may be limiting.

BH3-mimetics with specificity for particular members of the BCL-2 family, such as ABT-737, are invaluable tools for investigating their influence on response to chemotherapeutic drugs. Sensitivity to ABT-737 in a variety of tumor types is understood to be determined primarily by the activity of MCL-1 (14); ABT-737 has highest potency in tumors with low MCL-1 expression (42). Although melanoma cells are insensitive to ABT-737, lowering MCL-1 levels by shRNA knockdown markedly sensitized them, in agreement with other recent studies (24, 43, 44). Moreover, our results show that overexpression of the MCL-1 antagonist NOXA has an even more potent effect, sensitizing the cells more than 100-fold in most cases. NOXA overexpression has been shown to sensitize other cell types to ABT-737, including SCLCs (45) but dramatic effect on the IC₅₀ was not apparent in those studies, in part, due to a focus on the fraction of cells in apoptosis rather than inhibition of growth, which is arguably more relevant to treatment response.

The relative importance of BFL-1 in this regard has been considered only more recently. Like MCL-1, elevated BFL-1 can certainly confer potent resistance to ABT-737 in lymphomas (23, 24). Our results with the MCL-1–specific mNoxaB variant as well as MCL-1 and BFL-1 knockdowns indicate that inhibition of MCL-1 alone is sufficient for marked sensitization of melanoma cells to ABT-737. This suggests, conversely, that elevated expression or activity of MCL-1 will suffice for resistance to ABT-737. The results do not discount the possibility that BFL-1 activity can be a source of resistance to ABT-737 in melanoma but it is interesting that the 3E Noxa mutant, which binds BFL-1 but not MCL-1, was ineffective in sensitizing wild-type mouse embryo fibroblast (MEF) cells to ABT-737 or other drugs (14). It remains conceivable that NOXA has drug-sensitizing actions not mediated by either MCL-1 or BFL-1 but the simplest interpretation is that MCL-1 has greater influence over BAK in this context.

A number of other cancer types are sensitive to ABT-737 and there is evidence that its addition to chemotherapy regimens is beneficial (15), particularly in combination with antineoplastic drugs that alter the activity of MCL-1 and NOXA (46, 47). Two examples are melphalan and flavopiridol, which are both reported to repress MCL-1, for example, in lung carcinoma and myeloma cells (25, 29). In our hands, flavopiridol was more efficient at altering NOXA and MCL-1 levels in melanoma lines. It was striking that, in each cell line tested, doses around the IC₅₀ induced both proteins whereas higher concentrations repressed both. Such coordinated changes in NOXA and MCL-1 levels, not reported previously, would tend to counteract. Even so, at concentrations around its IC₅₀, flavopiridol markedly sensitized both melanoma lines tested to ABT-737, in 2D and 3D cell culture. Melphalan provided lesser sensitization, consistent with its lesser effects on NOXA and MCL-1 levels. Another recent study confirmed weak synergy between ABT-737 and the alkylating agent fotemustine on melanoma cells and that NOXA induction was crucial to the effect (44). Better results were obtained in that study with dacarbazine but their significance is unclear as the prodrug is not activated in vitro.

The scale of sensitization of the melanoma cell lines to ABT-737 by flavopiridol—8- to 15-fold—raises the possibility that even tumors normally unresponsive to ABT-737 might be amenable to treatment by a suitable combination with cytotoxic drugs. Moreover, the most efficient partner drug will not necessarily be among those normally used to treat the tumor—flavopiridol, for example, is not part of melanoma regimens. This point is illustrated by the strong synergy observed in vitro with colorectal carcinoma cells between ABT-737 and imiquimod (48), even though this drug, an immune modifier, has undefined actions in vitro. Resistance to ABT-737 developed rapidly in melanoma spheroids through loss of constitutive NOXA overexpression but that is less likely to be an obstacle when NOXA is induced transiently by cytotoxic drugs. Finally, although melanoma cells were sensitized to ABT-737 by cytotoxic drugs, the converse sensitization to cytotoxic drugs by ABT-737 was considerably weaker or absent, other than at unrealistically high concentrations of ABT-737 that likely have off-target effects. Reports of synergy between ABT-737 and cytotoxic drugs often do not distinguish between potentiation of one by the other and vice versa but they are distinct possibilities that should both be considered to fully encompass the potential of combination chemotherapy with BH3-mimetics.

Treatment of ABT-737–sensitive tumor lines in xenograft models has previously shown impressive results. In the case of 2 different SCLC lines, complete regression of the tumors was observed in more than three quarters of treated mice and tumors did not relapse when treatment was stopped (15–17). However, primary human SCLC xenografts were less responsive, as were other SCLC lines in vitro (16). Recent studies have reported synergy between ABT-737 or its relative ABT-263 and cytotoxic drugs in a variety of tumor xenograft models (47, 49–51), including melanomas (52, 53). The IC₅₀ values we observed for ABT-737 on NOXA-overexpressing melanoma cells in flat culture were comparable with those reported for ABT-737–sensitive SCLC and lymphoid malignancies (13). Yet, despite sensitivity to ABT-737 in vitro, the NOXA-transduced and MCL-1 knockdown melanoma lines did not respond as xenografts. This is not unprecedented—for example, hepatocellular carcinomas sensitive to ABT-737 in vitro also failed to respond as xenografts (18). Resistance to ABT-737 in spheroids of the NOXA-overexpressing melanoma cells developed rapidly in vitro. However, the obvious possibility that selection against the proapoptotic changes in the manipulated cells occurred in vivo during engraftment and treatment with ABT-737 was not supported by the evidence. Overexpression of NOXA was maintained at levels similar to the injected cells, in both ABT-737- and vehicle-treated xenografts, and explants of ABT-737–treated xenografts remained sensitive in vitro. A second possibility is that insufficient ABT-737 reached the xenografts. However, we detected thrombocytopenia and leukocytosis in the treated mice, consistent with other xenograft studies where ABT-737 was effective with the same regimen (54), so it likely reached circulation. While leukocytosis could be a manifestation of reaction to ABT-737 precipitate in the peritoneal cavity, thrombocytopenia is not consistent with that explanation. The subcutaneous melanoma xenografts were observed to be vascularized and subcutaneous SCLC xenografts in other studies responded to ABT-737 (16) but it is nevertheless possible that melanomas in other locations, such as metastases to the lymph nodes, lungs or liver, are more accessible to the compound, as it has been effective against model tumors in such locations, either as a single agent or a potentiator for other drugs (13, 15–18, 54).

A third, related, possibility is inadequate diffusion of ABT-737 into the melanoma xenografts (55), which could also be specific to the cell type. The 3D spheroids grown in vitro from melanoma cells overexpressing NOXA were less sensitized to ABT-737 than flat cultures, and resistance developed early with loss of NOXA expression. Moreover, ABT-737 was less effective when the spheroids were allowed to grow for 7 days prior to exposure, by which time they numbered more than 10⁵ cells. Xenografts consisted of an injected bolus of 10⁶ cells and were allowed 7-day growth before treatment with ABT-737 so the lack of response in vivo may reflect the insensitivity of large spheroids in vitro. If so, it does not represent an obstacle in principle. The MM200 cells tested may not be representative melanoma xenografts, as suggested by the responsiveness of A375 xenografts to ABT-737 plus temozolomide (52). In any case, development of more diffusible BH3 mimetics (56), which have improved oral bioavailability, charge balance, and metabolism properties, will likely obviate such problems.

In conclusion, our results indicate that addition of BH3-mimetics to current drug regimens or development of new drug combinations with BH3-mimetics are avenues worth exploring for melanoma therapy, particularly if dosing schedules can be developed to maintain the most favorable balance between MCL-1 and NOXA activity over time.

No potential conflicts of interest were disclosed.

Abbott Laboratories kindly provided ABT-737. The authors thank Dr. Meenhard Herlyn and Patricia Brafford, The Wistar Institute, for providing cell lines.

K.M. Lucas and N. Mohana-Kumaran are Cancer Institute New South Wales Scholar Award recipients. K.M. Lucas received an Australian Postgraduate Award. N. Mohana-Kumaran was supported by the Government of Malaysia. X.D. Zhang and J.D. Allen are Cancer Institute New South Wales Fellows. N.K. Haass is a recipient of the Cameron Fellowship from the Melanoma and Skin Cancer Research Institute, Australia. N.K. Haass is CIA on Project Grant RG 09–08 (Cancer Council New South Wales), Project Grant 570778 (Priority-driven collaborative cancer research scheme/Cancer Australia/Cure Cancer Australia Foundation), and Research Innovation Grant 08/RFG/1–27 (Cancer Institute New South Wales) and Project Grant 1003637 (National Health and Medical Research Council).

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.