Neratinib, a pan ERBB/HER inhibitor, restores sensitivity of PTEN-null, BRAFV600E melanoma to BRAF/MEK inhibition

Abstract

Introduction: Approximately 50% of melanomas harbor an activating BRAFV600E mutation. Standard of care involves a combination of inhibitors targeting mutant BRAF and MEK1/2, the substrate for BRAF in the MAPK pathway. PTEN loss-of-function mutations occur in ~40% of BRAFV600E melanomas, resulting in increased PI3K/AKT activity that enhances resistance to BRAF/MEK combination inhibitor therapy.

Methods: To compare the response of PTEN null to PTEN wild-type cells in an isogenic background, CRISPR/Cas9 was used to knock out PTEN in a melanoma cell line that harbors a BRAFV600E mutation. RNA sequencing, functional kinome analysis, and drug synergy screening were employed in the context of BRAF/MEK inhibition.

Results: RNA sequencing and functional kinome analysis revealed that the loss of PTEN led to an induction of FOXD3 and an increase in expression of the FOXD3 target gene, ERBB3/HER3. Inhibition of BRAF and MEK1/2 in PTEN null, BRAFV600E cells dramatically induced the expression of ERBB3/HER3 relative to wild-type cells. A synergy screen of epigenetic modifiers and kinase inhibitors in combination with BRAFi/MEKi revealed that the pan ERBB/HER inhibitor, neratinib, could reverse the resistance observed in PTEN null, BRAFV600E cells.

Conclusions: The findings indicate that PTEN null BRAFV600E melanoma exhibits increased reliance on ERBB/HER signaling when treated with clinically approved BRAFi/MEKi combinations. Future studies are warranted to test neratinib reversal of BRAFi/MEKi resistance in patient melanomas expressing ERBB3/HER3 in combination with its dimerization partner ERBB2/HER2.

Keywords: ERBB3; drug synergy; kinase inhibitor; melanoma; resistance.

Figure 1

Knockdown of PTEN in BRAFV600E melanoma cell lines. (A) A375 and (B) SK-MEL-181 cells expressing non-targeting (scrambled control) shRNA or shRNA targeting PTEN were generated by lentiviral infection and puromycin selection. Polyclonal populations were harvested and immunoblotting performed using the indicated antibodies. ERK2 was used as a loading control. Full-length blots are presented in Supplementary Figure S9 . (C) A375 and (D) SK-MEL-181 cells were treated with DMSO (control) or the indicated dose of dabrafenib/trametinib for 96 h prior to viability measurement using CellTiter-Glo. Dabrafenib (1×) and trametinib (0.1×) were used at 10:1 dose ratio. Mean +/− SD, n = 6. Indicated p-values were calculated using the extra sum of squares F-test in Prism.

Figure 2

CRISPR-mediated PTEN loss decreases sensitivity of BRAFV600E melanoma cells to combined dabrafenib/trametinib inhibition. (A) A375 cells were transfected with a single-guide RNA pool targeting human PTEN and selected with puromycin. Single-cell clones were isolated, harvested, and immunoblotting was performed with the indicated antibodies. ERK2 was used as a loading control. Full-length blots are presented in Supplementary Figure S8 . (B) A375 wild-type, PTEN null 5 (KO5), and PTEN null 11 (KO11) cell lines were harvested for immunoblotting with the indicated antibodies. ERK2 was used as a loading control. Full-length blots are presented in Supplementary Figure S10 . (C) A375 wild-type, KO5, and KO11 cell lines were grown on glass cover slips, fixed, and immunofluorescent staining was performed for PTEN, actin (Phalloidin), and Hoechst 33342 (nuclear marker). Scale bar, 20 µm. (D) A375 wild-type, KO5, and KO11 cell lines were treated with DMSO (control) or the indicated dose of dabrafenib/trametinib for 96 h prior to viability measurement using CellTiter-Glo. Dabrafenib (1×) and trametinib (0.1×) were used at a 10:1 dose ratio. Graphical representation of a single experiment. Bar plot of delta IC₅₀ values calculated by comparing fold change in IC₅₀ in the wild-type vs. each PTEN KO cell line. Indicated p-values were calculated using two-tailed t-tests, n = 7. (E) Six-day cell growth curves across A375 wild-type, KO5, and KO11 cell lines treated with DMSO (control) or 10/1 nM dabrafenib/trametinib. Mean +/− SD, n = 6. Indicated p-values were calculated using two-tailed t-tests. (F) One-week crystal violet assays in wild-type, KO5, and KO11 cell lines. Cells were dosed with DMSO (control) or 10/1 nM dabrafenib/trametinib. Bar plot represents quantification of crystal violet stain, mean +/− SD, n = 3. Indicated p-values were calculated using two-tailed t-tests. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3

PTEN loss alters the transcriptomic landscape of BRAFV600E melanoma. (A) Principal component analysis (PCA) of A375 wild-type, PTEN KO5, PTEN KO11, wild-type PIK3CA (PI3K) overexpressing, and mutant PIK3CA-H1047R (H1047R) overexpressing transcriptomes from three mRNA-seq biological replicates. Component 1 (dab/tram treatment) accounts for the largest possible variance (34%) in the data set. (B) K-means clustering (set to 4) of the top 2,000 genes (variable expression) was performed using iDEP0.96. Clusters and gene pattern associations are graphically illustrated at the right. (C-E) Pathway enrichment from the clusters shown in (B) was performed (GO Biological Process Pathway) using iDEP0.96, and the top 10 most significant pathways are shown for clusters 1, 3, and 4. (F, G) DESeq2 was used to identify differentially expressed genes in KO5 (F) or KO11 (G) cells versus wild-type A375 cells at baseline (DMSO). Kinases were annotated and log2 fold change in DESeq2 normalized counts and the −log10 adjusted P-values used for the volcano plot. (H) ERBB3/HER3, FOXD3, and ERBB2/HER2 log₂ read counts from BRAFV600E/K mutant, PTEN unaltered, or BRAFV600E/K mutant, PTEN altered skin cutaneous melanoma patient samples accessed from TCGA database (cbioportal.org). Indicated p-values were calculated in GraphPad Prism using an unpaired, two-sample Student’s t-test. *P < 0.05, **P < 0.01.

Figure 4

Dabrafenib/trametinib treatment enhances the SOX10-FOXD3-ERBB3/HER3 signaling axis PTEN null melanoma. (A) SOX10 log₂ read counts from BRAFV600E/K mutant, PTEN unaltered, or BRAFV600E/K mutant, PTEN altered skin cutaneous melanoma patient samples accessed from TCGA database (cbioportal.org). Indicated p-value was calculated in GraphPad Prism using an unpaired, two-sample Student’s t-test. *P < 0.05. (B-F) DESeq2 was used to identify differentially expressed genes in wild-type (B) or KO5 (C) cells, 1 day after 100 nM/10 nM dabrafenib/trametinib treatment, or in wild-type (D), KO5 (E), or KO11 (F) cells 7 days after treatment with 100 nM/10 nM dabrafenib treatment versus baseline (DMSO). Transcription factors were annotated and log2 fold change in DESeq2-normalized counts and the −log10 adjusted P-values used for the volcano plot. (E) A375 wild-type, KO5, and KO11 cell lines were treated with DMSO (0) or with 100/10 nM dabrafenib/trametinib for 1 or 7 days. Cells were harvested, and immunoblotting for the indicated proteins in the HER/FOXD3 and ERK/AKT axes was performed. FOXD3 has a predicted molecular weight of 60 kDa. A non-specific ~75-kDa band is marked with an asterisk. ERK2 was used as a loading control. Full length blots are presented in Supplementary Figure S11 . (F) A375 wild-type, KO5, and KO11 cells were treated with DMSO (0) or with 100 nM/10 nM dabrafenib/trametinib for 1 or 7 days. Cells were harvested and lysates used for MIB/MS kinome profiling. The heat map displays the normalized MIB binding values (log₂ label-free quantification (LFQ) intensities) for the indicated kinases. (G) Venn diagram of the average log₂ fold change of MIB binding in PTEN KO (clones 5 and 11) vs. wild type for top 25 kinases with increased binding at 7 days of 100/10 nM dabrafenib/trametinib treatment (two biological replicates). The common kinases with increased functional enrichment in PTEN null cell lines in response to dabrafenib/trametinib treatment are highlighted in red.

Figure 5

Drug synergy screen against dabrafenib and trametinib in BRAFV600E, PTEN-null melanoma cells. (A) Schematic of the synergy screening approach. A375 wild-type cells or PTEN KO clones were plated in 384-well plates and treated 96 h with 6 × 6 concentration matrix of library compound in combination with dabrafenib and trametinib. Cell growth was measured with CellTiter-Glo, and drug synergy was measured using the SynergyFinder package. (B) Synergistic compound rankings using Bliss scoring for dabrafenib/trametinib screen in the KO11 cell line. Synergy scores denote the percent inhibition observed following combination treatment, which exceeded the expected growth inhibition as calculated by the bliss independence model. (C) 96-h dose–response curves from the complete drug synergy screen show growth inhibition across all doses of the library compound for top synergistic hits, afatinib and neratinib in the KO11 cell line. (D) 3D drug interaction landscapes produced using SynergyFinder between dabrafenib/trametinib and neratinib/afatinib or JQ1 (-) in the KO11 cell line following 96-h drug treatment. (E) 96-h dose–response curves show synergistic growth inhibition between dabrafenib/trametinib and GSK690693 (AKT inhibitor) only at the 10-μM dose in PTEN null cell lines.

Figure 6

Neratinib synergizes with dabrafenib/trametinib inhibition to reverse resistance in BRAFV600E, PTEN null cells. (A) A375 wild-type, KO5, and KO11 cell lines were treated for 96 h with the indicated dose of dabrafenib/trametinib (dab/tram), 100 nM neratinib, or all three drugs (triple combination) prior to viability measurement using CellTiter-Glo. Dabrafenib (1×) and trametinib (0.1×) were used at a 10:1 dose ratio with or without 100 nM neratinib. Indicated p-values calculated using the extra sum of squares F-test in Prism. (B) 9-day cell growth curves across A375 wild-type, KO5, and KO11 cell lines treated with DMSO (control), 10/1 nM dabrafenib/trametinib, 100 nM neratinib, or all three drugs (triple combination). Cells were fixed if confluent before 9 days elapsed. Mean +/− SD, n = 6. Indicated p-values were calculated using two-tailed t-tests. (C) Four-week crystal violet assays in wild-type, KO5, and KO11 cell lines. Cells were dosed with DMSO, 10/1 nM dabrafenib/trametinib, 100 nM neratinib, or all three drugs (triple combination). (D) Bar plot represents quantification of crystal violet stain in (c), mean +/− SD, n = 3. Indicated p-values were calculated using two-tailed t-tests. (E-G) Senescence-associated beta-galactosidase staining in A375 wild-type (E), KO5 (F), and KO11 (G) cell lines following 96-h drug treatment. Cells were dosed with DMSO, 10 nM/1 nM dabrafenib/trametinib, 100 nM neratinib, or the triple combination. Nuclei were counter-stained with Hoechst 33342. (H) Bar plot represents quantification of percent of beta-galactosidase positive cells treated and stained as in (E-G). At least 200 cells were counted in biological triplicate. (I, J) 3D spheroids with A375 wild-type, KO5, and KO11 cell lines treated 7 days with DMSO, 10/1 nM dabrafenib/trametinib, 100 nM neratinib, or all three drugs (triple combination). Brightfield images were taken at 7 days. Cell viability was measured at endpoint with CellTiter-Glo 3D reagent. Mean +/− SD, n = 3. Indicated p-values were calculated using two-tailed t-tests. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Figure 7

A putative model of heightened SOX10-FOXD3-ERBB3/HER3 signaling in PTEN-deficient BRAFV600E melanoma and intervention with neratinib. Melanoma cells exhibit adaptive and intrinsic resistance to targeted BRAF/MEK inhibition that is now the standard of care for BRAFV600E/K melanoma. Previous studies have shown that ERK regulates SOX10, limiting its ability to transactivate FOXD3 (left panel). BRAFi/MEKi suppresses the MAPK pathway and enables SOX10 and FOXD3 to drive ERBB3/HER3 expression, which can contribute to resistance (center panel). Here, we provide evidence that PTEN loss seems to amplify the disrupted feedback leading to hyperactive ERBB3/HER3-AKT signaling. Drug screening identified synergy between BRAFi/MEKi and neratinib (an EGFR/ERBB inhibitor), suggesting that the triple combination of dabrafenib/trametinib/neratinib could suppress the heightened signaling through ERBB3/HER3 and prolong BRAFi/MEKi responsiveness in PTEN deficient, BRAFV600E melanoma. Created using BioRender.