Genotype-Selective Combination Therapies for Melanoma Identified by High Throughput Drug Screening

Matthew A. Held, Casey G. Langdon, James T. Platt, Tisheeka Graham-Steed, Zongzhi Liu, Ashok Chakraborty, Antonella Bacchiocchi, Andrew Koo, Jonathan W. Haskins, Marcus W. Bosenberg, and David F. Stern

Supplementary Table and Figure Legends

Table S1. Agents used in single-agent HTS and their major targets.

Table S2. Melanoma cell lines and mutation status. “Low” PTEN protein status refers to less than half normal PTEN protein detected. “Null” refers to PTEN deleted lines and absent PTEN expression.

Table S3. Concentrations required to inhibit 50% of cell growth (GI50) in the single-agent

screen. Cell lines (columns) are organized by genotype.

Table S4. Single agents with significant genotype-selective efficacy.

Table S5. Agents and concentrations chosen for combinatorial HTS (cHTS).

Table S6. Partial list of drug pairs showing mutant BRAF- or RAS-selectivity across

multiple drug concentrations.

Table S7. Combinations with simvastatin that are selective for the mutant RAS or BRAF

group of melanomas.

Table S8. Most effective genotype-selective combinations for mutant RAS or mutant

BRAF melanomas, or combinations effective regardless of genotype. Table S9. Combination index (CI) values for lapatinib or bosutinib with MK-2206, and

flavopiridol or 17-DMAG with simvastatin confirming synergy in mutant BRAF or RAS melanomas, respectively.

Figure S1. Single agent concentration-effect curves, related to Figure 1.

(A) Drugs with sigmoidal concentration-effect curves reaching a growth inhibitory

plateau below maximum concentration tested. Dashed red lines mark asymptotes

where the growth inhibitory limit was reached and brackets mark the approximate

concentration range of the plateau. (B) Single agents duplicated on different drug stock

plates run in different experiments. (C) Drugs with similar targets or mechanisms of

action used in the single agent screen showing similar concentration-effect curves. (D)

Concentration-effect curves of agents demonstrating greater drug efficacy for a

genotypic group. Curves fit to median GI values for each group. Asterisks indicate

significance at indicated concentration, p<0.05; Kruskal-Wallis test. See also Table S4.

Figure S2. Drug class and concentration choices for combinatorial drug

screening, related to Figure 2.

(A) Frequencies of drug classes used in combinatorial high-throughput screening

(cHTS). (B) Methodology for choosing representative concentrations of agents for use in

cHTS: GI50 values for melanoma cell lines (where those values were reached) were calculated and medians were determined subsequent to exclusion of outliers by Tukey’s

interquartile outlier limit test. Median (consensus) concentrations representing the GI50,

GI25 and GI10 effect levels were identified for each drug chosen for cHTS and designated as “high (H)”, “medium (M)” and “low (L)” consensus concentrations, respectively. These

values are listed in Table S5.

Figure S3, Mutant BRAF and mutant RAS selective drug pairs, related to Figure 3.

(A) Representative genotype-selective drug pairs identified for mutant

BRAF melanomas, (B) mutant RAS melanomas, or (C) a non-genotype selective

combination chosen for validation of cytotoxicity. Drug concentrations and drug number

identifiers are indicated in bottom right of each panel. Number positions within each bar

indicate the percent growth inhibition (x-axis) of that drug at the indicated concentration.

Length of red bars indicates the additional percent growth inhibition achieved from synergy as calculated by Bliss additivism, and the rightmost end of the red bar indicates the total percent growth inhibition achieved. The length of green bars indicates antagonism, i.e. the loss of expected percent growth inhibition with the drug concentration combination. The intersection of the yellow and green bar indicates the total percent growth inhibition achieved in these cases. All cell lines analyzed in cHTS are shown for complete visual comparison. (D) Additional drug pairs among the most selective efficacy for the mutant BRAF group (bosutinib and MK-2206) and (E) the mutant RAS group (simvastatin and 17-DMAG). The highest concentrations of agents were used, as listed in Table S5. (F) Flow cytometry analysis of the mutant RAS- selective combination of simvastatin (2.5µM) and 17-DMAG (0.02µM), or the non- genotype selective drug pair SAHA/vorinostat (1.0µM) and flavopiridol (0.15µM) in the representative mutant BRAF line YUMAC and mutant NRAS line YUGASP. Numbers in the lower right quadrant correspond to early apoptotic cells, while numbers in the upper right and left quadrants correspond to late apoptotic or necrotic cells, respectively. The sum of these three quadrants equals the total cytotoxicity achieved. Data are from same experiment shown in Figure 4B.

Figure S4. Analysis of drug combinations by target class, related to Figure 4D.

(A) ErbB inhibitors afatinib/BIBW2992 and gefitinib in paired combination with the AKT inhibitors MK-2206 or GSK690693 induce greater effects in both a vemurafenib sensitive mutant BRAF line (YUMAC) and a primary vemurafenib-resistant mutant

BRAF line (YUKSI), while matched concentration combinations of these drugs are relatively less effective in the mutant NRAS or HRAS melanoma lines (YUGASP and

YUAME, respectively). (B) Combination of the statins lovastatin or atorvastatin/Lipitor with the pan-CDK inhibitors flavopiridol or AT7519 are more growth inhibitory at lower concentrations in mutant RAS melanomas relative to mutant BRAF melanomas, although efficacies (max growth inhibition achieved) are similar. All treatments were for

72 hours followed by growth inhibition analysis using CellTiter-Glo ATP-detection reagent.

Figure S5. Mutant BRAF lines selected for vemurafenib resistance are re- sensitized to vemurafenib with concurrent treatment of lapatinib and MK-2206, related to Figure 5.

(A) YULAC parental line shows reduced colony size and number when treated with lapatinib and MK-2206 combined at reduced concentrations (1.5µM for each) relative to maximum concentrations used in cHTS (5.5µM each). (B) Flow cytometry demonstrating marked resistance to vemurafenib up to 10µM in the in vitro acquired

resistant lines YULAC-R and YUCOT-R. (C) Representative 2-D clonogenic (i), soft

agar (ii), and flow cytometry (iii) assessments of lapatinib and MK-2206 at reduced concentrations in the YULAC-R line. D=DMSO vehicle; L=lapatinib (1.5µM); M=MK-

2206 (1.5µM); L/M=lapatinib and MK-2206 (1.5µM each). (D) Vemurafenib (V) alone or

in dual and triple agent combinations at 10µM with lapatinib and MK-2206 in YULAC-R.

Concentrations of agents and roman numerals above panels are same as in (C). (E)

Similar efficacy was seen in a second acquired resistant line YUCOT-R. (F) Single,

dual, or triple agent combinations of vemurafenib, lapatinib, and MK-2206 were

ineffective on the mutant NRAS line YUGASP as observed with flow cytometry and

clonogenic assays.

Figure S6. Immunoblot analyses of acquired and primary vemurafenib-resistant

lines after combinatorial treatments, related to Figure 5.

(A) MAPK and PI3K pathway activity at 1 hour following vemurafenib, lapatinib, and MK-

2206 as single, dual, or triple agent treatments, in the primary vemurafenib-resistant line

YUKSI. Vemurafenib and lapatinib depleted p-ERK and p-EGFR levels, respectively,

within 1 hour. MK-2206 increased p-EGFR and p-ERK levels within 24 hours (Figure

5F), but these changes were not initially observed, despite complete loss of p-AKT. (B)

Equivalent increase in p-ERK and p-EGFR were seen in a second primary vemurafenib-

resistant line YUKOLI within 24 hours. (C) 1 and 24 hour treatments with these agents

in the acquired vemurafenib-resistant line YUCOT-R, (D) the parental YULAC line, and

(E) the parental YUCOT line at 24 hours. Sustained suppression of p-AKT levels was observed in parental lines following vemurafenib treatment, whereas in acquired resistant lines (Figure 5G for YULAC-R) p-AKT levels did not change markedly.

Figure S7. Signaling and biological effects of simvastatin and flavopiridol in vitro and in vivo, related to Figure 6.

(A) p-ERK and p-AKT levels are effectively suppressed in mutant NRAS melanoma

YUTICA, and (B) mutant HRAS melanoma YUAME with simvastatin and flavopiridol combined. (C), (D) Anti-clonogenic effects of simvastatin/flavopiridol on YUTICA and

YUAME, respectively. (E) Representative flow cytometry indicating substantial reduction in cytotoxicity of simvastatin and flavopiridol alone or combined in the vemurafenib- resistant line YUKSI. (F) Simvastatin and flavopiridol combination modestly reduced colony formation and colony size in YUKSI and YULAC-R BRAF mutant lines. (G)

Mouse weight over time with mock treatment or simvastatin at 10mg/kg/day and flavopiridol at 4mg/kg/qod alone or in combination. (H) Representative formalin-fixed tumor sections stained with hematoxylin and eosin from tumors in mock treated group or simvastatin and flavopiridol combination treatment group. Mitotic index, ***p<0.0001 by Kruskal-Wallis ANOVA, significant difference between mock treatment group and combination group by Dunn’s multiple comparison test; ns=not significant.