Landscape of Targeted Anti-Cancer Drug Synergies in Melanoma Identifies a Novel BRAF-VEGFR/PDGFR Combination Treatment

Abstract

A newer generation of anti-cancer drugs targeting underlying somatic genetic driver events have resulted in high single-agent or single-pathway response rates in selected patients, but few patients achieve complete responses and a sizeable fraction of patients relapse within a year. Thus, there is a pressing need for identification of combinations of targeted agents which induce more complete responses and prevent disease progression. We describe the results of a combination screen of an unprecedented scale in mammalian cells performed using a collection of targeted, clinically tractable agents across a large panel of melanoma cell lines. We find that even the most synergistic drug pairs are effective only in a discrete number of cell lines, underlying a strong context dependency for synergy, with strong, widespread synergies often corresponding to non-specific or off-target drug effects such as multidrug resistance protein 1 (MDR1) transporter inhibition. We identified drugs sensitizing cell lines that are BRAFV600E mutant but intrinsically resistant to BRAF inhibitor PLX4720, including the vascular endothelial growth factor receptor/kinase insert domain receptor (VEGFR/KDR) and platelet derived growth factor receptor (PDGFR) family inhibitor cediranib. The combination of cediranib and PLX4720 induced apoptosis in vitro and tumor regression in animal models. This synergistic interaction is likely due to engagement of multiple receptor tyrosine kinases (RTKs), demonstrating the potential of drug- rather than gene-specific combination discovery approaches. Patients with elevated biopsy KDR expression showed decreased progression free survival in trials of mitogen-activated protein kinase (MAPK) kinase pathway inhibitors. Thus, high-throughput unbiased screening of targeted drug combinations, with appropriate library selection and mechanistic follow-up, can yield clinically-actionable drug combinations.

Fig 1. Ultra high-throughput screen to identify synergistic combinations in melanoma cells.

(A) Summary of clinical development stage of 108 drugs included in the combination drug panel. (B) Example of raw UHTS data generated, demonstrating cell count information collected from DAPI channel and apoptosis data from cleaved PARP immunofluorescence; positive control of HSP90 inhibitor 17-AAG treatment is shown. (C) Summary matrix of combinatorial drug data, with each point representing the effect of one of 5,778 combinations at the standard drug concentration, as the median effect of the drug combination across all 36 melanoma cell lines on the relative cell count (left) and the calculated Bliss synergy score for that combination (right). (D) Histogram of number of cell lines a given drug combination showed synergy. Peak number of synergies were seen in one cell line, indicating many synergies are private. (E) As in (C), showing median effect of the drug combination (at standard concentration) on the relative cPARP positive proportion (left) and the calculated Bliss synergy for that cPARP level (right). (F) Graphical representation of drug combinations (drug pairs connected by an edge) that showed a significant unexpectedly high cPARP over a predicted level at the given cell count. Node size indicates the number of drug pairs that the given drug appears with other drugs on the “unexpectedly apoptotic” list. Edge color indicates the drug pair concentration (standard or low) where the elevated cPARP was found; edge pattern indicates whether the elevated cPARP was found in the setting of low cell count or normal cell count (> 80% control), with elevated cPARP in the setting of normal viability potentially representing “slow” death kinetics for that combination.

Fig 2. Cytotoxic potentiation by a MDR inhibitor.

(A) Screening results showing the effects at both concentrations of vincristine and lapatinib, individually, and as a combination, showing strong synergy across most melanoma cell lines as indicated by high Bliss synergy scores. (B) Confirmation of the synergistic effect of the combination of lapatinib (5μM) in A375 (n = 19) but not WM451Lu (n = 14) cells. Error bars represent s.d. of measurement replicates. (C) Representative flow cytometry data showing lapatinib potentiates G₂/M shift of A375 cell population consistent with increased vincristine effect. (D) Log₂ relative expression of the given multi-drug resistance transporter in A375 versus WM451Lu cells, showing increased MDR1 expression in A375 cells. Error bars represent s.d. of measurement replicates (n = 9). (E) Calcein dye flux experiments showing increased fluorescence intensity (indicating decreased MDR flux) in the presence of lapatinib or control MDR inhibitor verapamil (both at 5μM); quantitation of cell grey values shown at left. Error bars represent s.d. of measurement replicates (n = 4, > 200 cells per replicate). (F) MDR1 knockdown by siRNA causes a synergistic effect on cell viability in the presence of 5nM vincristine. Error bars represent s.d. of measurement replicates (n = 7). (G) Overexpression of MDR1 (compared to GFP control) in WM451Lu decreases sensitivity to vincristine, an effect reversible with 5μM lapatinib. Error bars represent s.d. of measurement replicates (n = 4)

Fig 3. Cediranib and PLX4720 synergize in intrinsically PLX4720-resistant melanoma cells.

(A) Effect of PLX4720 (PLX4720) across melanoma cells in the primary screen, demonstrating some BRAF mutant melanomas display intrinsic resistance to BRAF inhibition. Resistance (defined as > 50% viability at > 5μM PLX4720) in these lines was confirmed in secondary assays (below). (B) Cediranib displayed synergistic effects with PLX4720 in several intrinsically resistant lines in the primary screen; these effects were verified in standard growth assays in secondary screens (below, with 2μM cediranib). BRAF inhibitor-sensitive lines UACC62 and SkMel28 showed no synergy at any PLX4720 dose, while resistant ISTMel1 and RPMI7051 lines showed strong synergy with cediranib (Bliss > 30%). Error bars represent s.d. of measurement replicates (n = 3–4). (C) Long-term crystal violet-stained colony growth assays confirmed sustained synergy between PLX4720 (PLX) and cediranib (ced) in resistant ISTMel1 and RPMI7951 lines but not PLX4720 sensitive lines UACC62 and SkMel28. (D) Annexin V assays showed that cediranib induced apoptosis when combined with PLX4720 in resistant ISTMel1 cells, as compared to sensitive UACC62, and as confirmed by Western blotting to cleaved PARP (below). Error bars represent s.d. of measurement replicates (n = 3). (E) Western blotting confirmed expression of cediranib targets PDGFRβ, and variably, KDR and PDGFRα in PLX4720-insensitive ISTMel1 and RPMI7951 cells, with weaker or absent expression in PLX4720 sensitive UACC62 and SkMel28 cells. Twenty-four hour treatment with PLX4720 moderately suppressed KDR and induced PDGFRβ expression, a pattern also seen in vivo (S9A Fig). (F) Western blotting confirmed on-target suppression of PDGFRα and PDGFRβ phosphorylation (latter after immunoprecipitation and blotting to phospho-tyrosine given low abundance). KDR phosphorylation was too weak to be observed by Western routinely in culture but was seen in vivo (see S9A Fig). (G) RNAi-mediated knockdown of KDR (left) with pools or individual siRNAs showed synergy to ~30%; synergy was proportional to KDR knockdown (see S7A Fig). Error bars represent s.d. of measurement replicates (n = 6). At right, average Bliss values over multiple replicates showed increasing synergy with simultaneous knockdown of multiple cediranib targets KDR and PDGFRβ, and, more weakly, PDGFRα. Error bars represent s.d. of measurement replicates (n = 5–8).

Fig 4. Cediranib synergizes with PLX4720 in vivo.

(A) ISTMel1 xenografts were generated in immunodeficient mice (n = 8 per group), which were treated with PLX4720 or control chow and given cediranib or water by oral gavage. ISTMel1 xenografts showed an initial response to PLX4720 treatment alone after approximately two weeks, with some additional but non-significant response to cediranib treatment. Values are shown as mean +/- S.E.M. ranges, with significant differences in mean tumor size of both PLX4720 and PLX4720 and cediranib-treated mice compared to control treatment by ANOVA. However, the combination significantly delayed progression (defined as > 500 mm³ size) beyond this initial response (right, p < 0.002 between PLX4720 and PLX4720 + cediranib arms by log-rank test). (B) In contrast to ISTMel1 xenografts, RPMI7951 xenografts showed a significant difference by ANOVA test in initial response to PLX4720 versus PLX4720 and cediranib treatment after three weeks, and, right, showed a prolonged delay of progression (defined as tumor > 250 mm³) of completely PLX4720-resistant tumors (p < 0.0001 between PLX4720 and PLX4720 and cediranib arms by log-rank test). (C) KDR staining of tumor biopsies from patients entering clinical trials for BRAF with or without MEK inhibitors demonstrated some had strong membrane KDR staining (top) throughout the tumor (inset showing membrane staining, at arrowhead), while others were negative for KDR staining (bottom) except for expected endothelial staining. (D) Comparison of progression-free survival of patients with (n = 6) and without (n = 10) membrane KDR staining showed a significant reduction in PFS (9.3 vs. 3.8 months, p < 0.01 by Student’s t-test) if the patient’s biopsy expressed KDR.