AXL/AKT axis mediated-resistance to BRAF inhibitor depends on PTEN status in melanoma

Abstract

Multiple genetic mutations within melanoma not only cause lesion-specific responses to targeted therapy but also alter the molecular route of resistance to that therapy. Inactivation of PTEN occurs in up to 30% of melanomas, frequently with a concurrent activating BRAF mutation. PTEN loss regulates both acquired and intrinsic drug resistance. Here we show that AXL/AKT axis mediated-resistance to BRAF inhibitor (BRAFi) depends upon PTEN status in melanoma. Hyperactivation of both ERK and AKT pathways was associated with BRAFi resistance in melanoma with wildtype PTEN. The PTEN-impaired melanoma cells required only the ERK resistance mechanism. Moreover, we identified AXL as a key upstream effector of AKT pathway-associated resistance to BRAFi in melanoma with wildtype PTEN, but not in melanoma with impaired PTEN. Notably, we confirmed that blocking AXL by shRNA and a small molecular inhibitor could rescue the sensitivity of resistant melanoma cells with wildtype PTEN to BRAFi and inhibit their growth in vitro and in vivo. Our study has uncovered a mechanism by which PTEN status contributes to acquired resistance to BRAFi and offers a rational strategy to guide clinical testing in pre-identified subsets of patients who relapse during treatment with BRAFi. The identified protein AXL represents a promising therapeutic target for BRAF mutant melanoma patients with wildtype PTEN.

Figure 1.. The melanoma models of acquired resistance to BRAF inhibitor in human melanoma with/without wildtype PTEN.

(A) The Western blotting showed the protein level of PTEN in 6 BRAF^V600E human melanoma cell lines with various PTEN genetic background, the protein level of tubulin as loading control. (B) The values of 50%-inhibitory-concentration (IC50) of 6 pairs parental and resistant human melanoma cells were determined using CCK8 assays. Error bars denote s.d. for biological three repeats. Results are statistically significant between parental and resistant groups by Student’s t-test (* p<0.05, **p<0.01). (C) The cell viability curves of 6 pairs parental and resistant human melanoma cells were determined using CCK8 assays. Error bars denote s.d. for biological three repeats.

Figure 2.. Oncogene addiction resulted by both hyperactivated ERK and AKT pathways is dependents on PTEN status in BRAF inhibitor-resistant melanoma.

(A and B) WM88 cells with wildtype PTEN or WLH6215 cells with impaired PTEN and their PLX4032-resistant counterparts were treated with 2μM PLX4032 for the indicated time (A) or indicated the concentration of PLX4032 for 2 hours (B), and the effects on the activation of ERK½ and AKT were determined by immunoblotting for p-ERK and p-AKT. The protein level of total ERK, total AKT and tubulin are as loading controls. (C) WM9, WM858, sk-mel 28 and A375sm cells and their PLX4032-resistant counterparts were treated with indicated concentration of PLX4032 for 2 hours, and the effects on the activation of ERK½ and AKT were determined by immunoblotting for p-ERK and protein p-AKT levels. The protein level of total ERK, total AKT and tubulin are as loading controls.

Figure 3.. Synergistic growth inhibition of combination with AKT, MEK, and BRAF inhibitors also is dependents on PTEN status in BRAF inhibitor-resistant melanoma.

(A) Western blotting showed indicated protein levels from WM88R and WLH6215R human BRAF inhibitor-resistant melanoma cells treated for 2 hours with 2.0μmol/L PLX4032 (+), 10μmol/L MEK inhibitor PD98059 (+) or 200 nmol/L AKT inhibitor MK2206 (+) and DMSO (−) control. The protein level of total ERK, total AKT and tubulin are as loading controls. (B) Cell viability curves of WM88R or WLH6215R human BRAF inhibitor-resistant melanoma cells in response to PLX4032 or combination of PLX4032 and PD98059, MK2206 or both were determined using CCK8 assay. Error bars denote s.d. for biological three repeats. (C) Graphs of combination indexes (CI) for WM88R or WLH6215R human BRAF inhibitor-resistant melanoma cells treated with the combination of PLX4032 and MK2206 or PD98059 were determined using the Chou and Talalay method.

Figure 4.. Impaired PTEN in BRAF ^V600E mutated melanoma influences the resistance to BRAF inhibitor.

(A) The protein level of PTEN in A375sm (C), A375sm/PTEN KO (PTEN KO) and its BRAF inhibitor-resistant counterpart A375sm/PTEN KO/R (PTEN KO/R) or WLH6215 (C), WLH6215/PTEN-M (PTEN-M) cells and its BRAF inhibitor-resistant counterpart WLH6215/PTEN-M/R (PTEN-M/R) was determined using western blot. The protein level of tubulin is as loading controls. (B) IC50 values of A375sm/PTEN KO and WLH6215/PTEN-M parental and it’s BRAF inhibitor-resistant cells. Error bars denote s.d. for biological three repeats. Results are statistically significant between parental and resistant groups by Student’s t-test (*p<0.05, **p<0.01). (C) Cell viability curve of A375sm/PTEN KO and WLH6215/PTEN-M parental and their BRAF inhibitor-resistant cells. Error bars denote s.d. for biological three repeats. (D) A375sm/PTEN KO and WLH6215/PTEN-M cells and their BRAF inhibitor-resistant counterparts were treated with indicated concentration of PLX4032 for 2 hours, and the effects on MAPK or PI3K/AKT signaling pathway were determined by immunoblotting for p-ERK and p-AKT levels. The protein level of total ERK, total AKT and tubulin are as loading controls.

Figure 5.. AXL protein mediates acquired-resistance to PLX4032 via activating PI3K/AKT pathway in melanoma with wild-type PTEN.

(A) Whole-cell extracts from WM88 and WM88R or WLH6215 and WLH6215R were incubated with the RTK antibody arrays, and phosphorylated proteins were determined by subsequent incubation with anti-phosphotyrosine horseradish peroxidase (each RTK spotted in duplicate, positive controls in corners). (B) The expression of AXL and p-AXL were confirmed using western blot with indicated parental (P), BRAF inhibitor-resistant (R) human melanoma cells. The protein level of tubulin is a loading control. (C) Western blotting (stripping the same western blots in figure 4A) showed the expression level of AXL and p-AXL in cells with PTEN knockout (PTEN KO) or mutant PTEN (PTEN-M) and their counterpart resistant cells (PTEN KO/R or PTEN-M/R). The protein level of tubulin is a loading control. C, the cells with empty vector as control. (D) Knockdown of AXL by shRNAs rescued the sensitivity of BRAF inhibitor-resistant WM88R cells to BRAF inhibitor and decreased the activation of AKT. The left panel showed the expression level of AXL, pAKT and indicated proteins using western blot. The middle panel showed the IC50 values of BRAF inhibitor for the WM88R with/without shRNA for AXL. The right panel showed the cell viability of cells stable expressing shRNAs for AXL in the treatment of BRAF inhibitor determined by CCK8 assay. WM88R: p, parent cells; shC, stable expressing shRNA control; shAXL1, shAXL2, and shAXL3, stable expressing different three shRNA constructs. Error bars denote s.d. for biological repeats. Results are statistically significant for all shAXL groups vs control by one-way ANOVA (*p < 0.05, **p < 0.01). (E) Knockdown of AXL by shRNA rescued the sensitivity of BRAF inhibitor-resistant A375smR cells to BRAF inhibitor and decreased the activation of AKT. The left panel showed the expression level of AXL, pAKT and indicated proteins using western blot. The middle panel showed the IC50 values of BRAF inhibitor for the A375smR with/without shRNA for AXL. The right panel showed the cell viability of cells stable expressing shRNAs for AXL in the treatment of BRAF inhibitor determined by CCK8 assay. A375smR: p, parent cells; shC, stable expressing shRNA control; shAXL1, shAXL2, and shAXL3, stable expressing different three shRNA constructs. Error bars denote s.d. for biological repeats. Results are statistically significant for all shAXL groups vs control by one-way ANOVA (*p < 0.05, **p < 0.01).

Figure 6.. (A) AXL promotes resistance to BRAF inhibitor in PTEN wildtype melanoma cells.

(Top panel) The western blotting showed that the protein level of AXL in WM88 cells transfected with empty vector (EV), ALX wildtype (WT) or AXL kinase-dead mutant form (DN). (bottom panel) Cell viability curves of WM88 cells transfected with empty vector (EV), expressing AXL wildtype (WT) or AXL kinase-dead mutant form (DN) plasmid in response to PLX4032 were determined using CCK8 assay. Error bars denote s.d. for biological three repeats. (B) The constitutively active Akt abrogates the inhibition of PLX-4032 resistance mediated by shAXL. (Top panel) The western blotting showed that the protein level of AKT in WM88R/shAXL cells transfected with empty vector (EV), expressing Myr-AKT1 (AKT1), Myr-AKT2 (AKT2) or Myr-AKT3 (AKT3) plasmid. (bottom panel) Cell viability curves of WM88R/shAXL cells transfected with empty vector (EV), expressing Myr-AKT1 (AKT1), Myr-AKT2 (AKT2) or Myr-AKT3 (AKT3) plasmid in response to PLX4032 were determined using CCK8 assay. Error bars denote s.d. for biological three repeats. (C, D, and E) A pharmacological inhibitor R428 of AXL decreases the phosphorylation of AKT and restores the sensitivity of WM88R and A375smR cells to BRAF inhibitor with synergistic inhibition. (C) The expression of indicated proteins in WM88R and A375smR cells treated with 2 μM AXL inhibitor R428 and 2 μM BRAF inhibitor PLX4032 for 2 hours was determined by western blot. (D) The cell viability of WM88R and A375smR treated with BRAF inhibitor, or combination of BRAF inhibitor and AXL inhibitor R428 was determined by CCK8 assay. Error bars denote s.d. for biological three repeats. (E) Combination indexes (CI) for WM88R or A375smR cells treated with the combination of PLX4032 and R428 were determined using the Chou and Talalay method.

Figure 7.. Inhibition of AXL by a pharmacological inhibitor R428 blocks tumor growth in vivo.

(A, left panel) The growth curves of resistant WM88R tumor cells treated with mock (WM88R Mock, n=10), BRAF inhibitor (WM88R PLX 4032, n=7), AXL inhibitor R428 (WMR R428, n=8) or combination of AXL inhibitor R428 with PLX4032 (WM88R PLX4032+R428, n=7). (A, right panel) The growth curves of parent WM88 tumor cells treated with mock (WM88 Mock, n=6), BRAF inhibitor (WM88 PLX 4032, n=6), AXL inhibitor R428 (WM88 R428, n=7). Each point represents mean tumor volume ± s.e.m. ND, no statistical difference; *p<0.05; ** p<0.01; *** p<0.001. (B, left panel) The growth curves of resistant A375smR tumor cells treated with mock (375smR Mock, n=10), BRAF inhibitor (375smR PLX 4032, n=8), AXL inhibitor R428 (375smR R428, n=10) or combination of AXL inhibitor R428 with PLX4032 (375smR PLX4032+R428, n=8). (B, right panel) The growth curves of parent A375sm tumor cells treated with mock (A375sm Mock, n=6), BRAF inhibitor (A375sm PLX 4032, n=6), AXL inhibitor R428 (A375sm R428, n=7). Each point represents mean tumor volume ± s.e.m. ND, no statistical difference; *p<0.05; ** p<0.01; *** p<0.001.