Targeting the PREX2/RAC1/PI3Kβ Signaling Axis Confers Sensitivity to Clinically Relevant Therapeutic Approaches in Melanoma

Abstract

Metastatic melanoma remains a major clinical challenge. Large-scale genomic sequencing of melanoma has identified bona fide activating mutations in RAC1, which are associated with resistance to BRAF-targeting therapies. Targeting the RAC1-GTPase pathway, including the upstream activator PREX2 and the downstream effector PI3Kβ, could be a potential strategy for overcoming therapeutic resistance, limiting melanoma recurrence, and suppressing metastatic progression. Here, we used genetically engineered mouse models and patient-derived BRAFV600E-driven melanoma cell lines to dissect the role of PREX2 in melanomagenesis and response to therapy. Although PREX2 was dispensable for the initiation and progression of melanoma, its loss conferred sensitivity to clinically relevant therapeutics targeting the MAPK pathway. Importantly, genetic and pharmacologic targeting of PI3Kβ phenocopied PREX2 deficiency, sensitizing model systems to therapy. These data reveal a druggable PREX2/RAC1/PI3Kβ signaling axis in BRAF-mutant melanoma that could be exploited clinically. Significance: Cotargeting the MAPK and the PREX2/RAC1/PI3Kβ pathways has remarkable efficacy and outperforms monotherapy MAPK inhibition in BRAF-mutant melanoma, supporting the potential of this combination therapy for treating metastatic melanoma.

Figure 1.

PREX2 ablation does not impact melanoma progression in vivo. A, Schematic of RTK and G-protein–coupled receptor (GPCR) signaling in melanoma. Arrows, activation; blunt-ended lines, inhibition; circled P, phosphorylation. B, Histograms of mutation frequencies of select commonly mutated genes in total (left) and BRAF-mutant (right) skin cutaneous melanoma (SKCM) cohorts from TCGA PanCancer study, accessed via cBioPortal on June 23, 2023. C, Progression-free survival (PFS) of PREX2 mutant vs. wild-type cases in the curated cohort of patients with SKCM from the TCGA PanCancer cohort, censored at 5 years. Left, all cases (n = 381; PREX2 wild-type = 290; PREX2 mutant = 91); right, BRAF-mutant cases (n = 207; PREX2 wild-type = 155; PREX2 mutant = 52). D, Top, schematic timeline of melanoma development (melanocytic dysplasia, naevus formation, and progression to melanoma) with common clinically relevant mutations indicated. Bottom, schematics of transgenic alleles used to generate melanoma GEMMs. Exons are boxed, with exon numbers indicated. Asterisks, location of mutations. Filled triangles, loxP sites; MG, minigene encoding exons 15–18 of wild-type Braf with a 5′ splice acceptor; neo^R, neomycin phosphotransferase gene; GT1, gene trap vector containing the β-galactosidase/neomycin resistance fusion gene (β-geo); Prex2^gd, Prex2^E22A,N204A allele. E, Top left, overall survival of Tyr-Cre^ERT2Braf^V600E/+Prex2^+/+ (n = 23; median survival, 466 days) vs. Tyr-Cre^ERT2Braf^V600E/+Prex2^−/− (n = 35; median survival, 375 days) mice; P = 0.1333. Top right, overall survival of Tyr-Cre^ERT2Braf^V600E/+Pten^fl/+Prex2^+/+ (n = 16; median survival, 81 days) vs. Tyr-Cre^ERT2Braf^V600E/+Pten^fl/+Prex2^−/− (n = 38; median survival, 64 days) mice, P = 0.8286. Bottom left, overall survival of Tyr-Cre^ERT2Braf^V600E/+Trp53^fl/flPrex2^+/+ (n = 32; median survival, 112.5 days) vs. Tyr-Cre^ERT2Braf^V600E/+Trp53^fl/flPrex2^−/− (n = 17; median survival, 116 days) mice, P = 0.4488. Bottom right, overall survival of Tyr-Cre^ERT2Braf^V600E/+Pten^fl/+Prex2^+/+ (n = 12; median survival, 65 days) vs. Tyr-Cre^ERT2Braf^V600E/+Pten^fl/+Prex2^gd/gd (n = 13; median survival, 64 days) mice, P = 0.7544. P values calculated using the log-rank (Mantel–Cox) test. PIP₂, phosphatidylinositol-4,5-diphosphate; PIP₃, phosphatidylinositol-3,4,5-trisphosphate; RTK, receptor tyrosine kinase. [A and D, Created in BioRender. Campbell, A. (2025), https://BioRender.com/n12n755.]

Figure 2.

Functional loss of PREX2 function sensitizes to MAPK inhibition in vitro and in vivo. A, Left, Kaplan–Meier overall survival of mice with the indicated genotypes and treatments. BRAF PTEN + vehicle (n = 8; median survival, 12.5 days) vs. BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days), P = 0.0096; BRAF PTEN PREX2 + vehicle, (n = 7; median survival, 13 days) vs. BRAF PTEN PREX2 + AZD6244 (n = 8; median survival, 108.5 days), P = 0.00005; BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days) vs. BRAF PTEN PREX2 + AZD6244 (n = 8; median survival, 108.5 days), P = 0.0021 log-rank (Mantel–Cox) test. Middle, longitudinal growth of individual tumors from vehicle-treated (n = 8) vs. AZD6244-treated (n = 7) BRAF PTEN cohorts and vehicle-treated (n = 7) vs. AZD6244-treated (n = 8) BRAF PTEN PREX2 cohorts. Right, relative change in tumor volume in mice with the indicated genotypes over the first 7 days of indicated treatment. BRAF PTEN + vehicle (n = 8), BRAF PTEN + AZD6244 (n = 7), BRAF PTEN PREX2 + vehicle (n = 7), BRAF PTEN PREX2 + AZD6244 (n = 8). P values were calculated by one-way ANOVA corrected for multiple comparisons (Tukey). B, Left, Kaplan–Meier overall survival of mice with the indicated genotypes and treatments. BRAF P53 + vehicle (n = 7; median survival, 14 days) vs. BRAF P53 + AZD6244 (n = 9; median survival, 26 days), P = 0.0204; BRAF P53 PREX2 + vehicle (n = 5; median survival, 14 days) vs. BRAF P53 PREX2 + AZD6244 (n = 8; median survival, 39 days), P = 0.0010; BRAF P53 + AZD6244 (n = 8; median survival, 14.5 days) vs. BRAF P53 PREX2 + AZD6244 (n = 10; median survival, 35 days), P = 0.0034. P values calculated using the log-rank (Mantel–Cox) test. Middle, longitudinal growth of individual tumors from vehicle-treated (n = 5) vs. AZD6244-treated (n = 4) BRAF P53 cohorts and vehicle-treated (n = 7) vs. AZD6244-treated (n = 6) BRAF P53 PREX2 cohorts. Right, relative change in tumor volume over the first 7 days of indicated treatment. BRAF P53 + vehicle (n = 5), BRAF P53 + AZD6244 (n = 4), BRAF P53 PREX2 + vehicle (n = 5), BRAF P53 PREX2 + AZD6244 (n = 6); P values, one-way ANOVA corrected for multiple comparisons (Tukey). C, Left, Kaplan–Meier overall survival of BRAF PTEN and BRAF PTEN PREX2-GD cohorts treated with vehicle or AZD6244. BRAF PTEN + vehicle (n = 8; median survival, 12.5 days) vs. BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days), P = 0.0096; BRAF PTEN PREX2-GD + vehicle (n = 6; median survival, 11 days) vs. BRAF PTEN PREX2-GD + AZD6244 (n = 8; median survival, 100.5 days), P = 0.0002; BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days) vs. BRAF PTEN PREX2-GD⁺AZD6244 (n = 8; median survival, 100.5 days), P = 0.0414. P values calculated using the log-rank (Mantel–Cox) test. Middle, longitudinal growth of individual tumors from vehicle-treated (n = 8) vs. AZD6244-treated (n = 7) BRAF PTEN cohorts and vehicle-treated (n = 6) vs. AZD6244-treated (n = 8) BRAF PTEN PREX2-GD cohorts. Right, relative change in tumor volume over the first 7 days of indicated treatment. BRAF PTEN + vehicle (n = 8), BRAF PTEN + AZD6244 (n = 7), BRAF PTEN PREX2-GD + vehicle (n = 5), BRAF PTEN PREX2-GD + AZD6244 (n = 6). P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). Note that the same BRAF PTEN treatment datasets are represented in A and C. D, Top left, relative confluence of parental WM266.4 cells treated with indicated treatments over the 96-hour period. Representative of a minimum of three independent experiments and three technical replicates. Data, mean ± SEM (confluence relative to starting point). Top right, relative confluence of WM266.4 PREX2^KO pooled lines D2 and A4 treated with indicated treatments over the 96-hour period. Representative of a minimum of four independent experiments and three technical replicates. Data, mean ± SEM (confluence relative to starting point). Bottom left, relative confluence of WM266.4 PREX1^KO pooled lines B4 and A4 treated with indicated treatments over the 96-hour period. Representative of a minimum of three independent experiments and three technical replicates. Data, mean ± SEM (confluence relative to starting point). Bottom right, relative suppression of growth, difference in calculated AUC of DMSO treated vs. AZD6244-treated lines as indicated. Center line, mean. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey).

Figure 3.

Cotargeting of MEK1/2 and the RAC1 effector p110β suppresses growth of melanoma lines in vitro. A, Left, relative confluence of WM266.4 cells treated with indicated treatments over time. Representative of a minimum of five independent experiments and three technical replicates. Data, mean ± SEM (confluence relative to starting point). Right, relative change in confluence of WM266.4 cells over indicated 72-hour treatment. Center line, mean. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). B, Left, relative confluence of A375 cells treated with indicated treatments over time. Representative of three independent experiments with a minimum of three technical replicates. Data, mean ± SEM (confluence relative to starting point). Right, relative change in confluence of A375 cells over indicated 72-hour treatment. Center line, mean. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). C, Fluorescence labeled (LI-COR) immunoblotting for indicated activated components of the MAPK–PI3K–mTOR pathway in the indicated human melanoma cells treated with vehicle or the indicated targeted therapeutics. Co-labeled total protein detection for each phosphoprotein serves as a sample integrity control. The blots are representative of three repeated experiments. D, RPPA dataset comparing PTEN-deficient and PTEN-proficient melanoma cell lines treated with indicated treatments for 24 hours. Cell line names and treatments are indicated below the heatmap. Antigens, detected by RPPA antibodies, are listed vertically according to biological process/signaling pathway. Color intensity scale indicates high (red) and low (blue) log₂FC of RPPA intensity values relative to the relevant DMSO control. Results are representative of two technical and five biological replicates per condition.

Figure 4.

Genetic targeting of p110β and its interaction with RAC1 sensitizes to MEK1/2 inhibition in vitro and in vivo.A, Left, flow cytometry–based cell-cycle profiling of WM266.4 cells following indicated 24-hour treatment. Right, proportion of WM266.4 cells in G₁/S at 24 hours. n = 3 independent experiments. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). B, Left, flow cytometry–based cell-cycle profiling of A375 cells following indicated 24 hours treatment. Middle, proportion of A375 cells in G₁/S at 24 hours. Right, proportion of subdiploid A375 cells at 24 hours. n = 4 independent experiments. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). C, Left, Kaplan–Meier overall survival of BRAF PTEN mice treated with vehicle or AZD6244. BRAF PTEN + vehicle (n = 8; median survival, 12.5 days) vs. BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days), P = 0.0096; BRAF PTEN PIK3CB-mut + vehicle (n = 4; median survival, 11.5 days) vs. BRAF PTEN PIK3CB-mut + AZD6244 (n = 7; median survival, 83 days), P = 0.0082; BRAF PTEN + AZD6244 (n = 8; median survival, 74.5 days) vs. BRAF PTEN PIK3CB-mut + AZD6244 (n = 7; median survival, 83 days), P = 0.1643. P values calculated using the log-rank (Mantel–Cox) test. Middle, longitudinal growth of individual tumors from vehicle-treated (n = 8) vs. AZD6244-treated (n = 7) BRAF PTEN cohorts and vehicle-treated (n = 5) vs. AZD6244-treated (n = 7) BRAF PTEN PIK3CB-mut cohorts. Right, relative change in tumor volume over the first 7 days of indicated treatment. BRAF PTEN + vehicle (n = 8), BRAF PTEN + AZD6244 (n = 7), BRAF PTEN PIK3CB-mut + vehicle (n = 5), BRAF PTEN PIK3CB-mut + AZD6244 (n = 6); P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). Center line, mean. Note that the BRAF PTEN treatment datasets are reproduced from Fig. 2A and C. D, Left, relative confluence of WM266.4 PIK3CB^KO cells treated with indicated treatments over the 96-hour period. Representative of a minimum of three independent experiments and three technical replicates. Data, mean ± SEM (confluence relative to starting point). Right, relative change in confluence of WM266.4 PIK3CB^KO cells over the indicated 96-hour period. Center line, mean. P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey).

Figure 5.

Therapeutic targeting of MEK1/2 and the RAC1 effector p110β has marked therapeutic efficacy in melanoma in vivo. A, Left, Kaplan–Meier overall survival of mice harboring WM266.4 subcutaneous xenografts treated with vehicle (n = 5; median survival, 33 days), AZD6244 (n = 5; median survival, 49 days), AZD8186 (n = 5; median survival, 44 days), or AZD6244/AZD8186 (n = 5; median survival, 128 days). P values calculated using the log-rank (Mantel–Cox) test: vehicle vs. AZD6244, P = 0.2269; vehicle vs. AZD8186, P = 0.3091; vehicle vs. AZD6244/AZD8186, P = 0.0017 and AZD6244 vs. AZD6244/AZD8186, P = 0.0064. Middle, relative change in tumor volume of WM266.4 xenografts over the first 21 days of indicated treatment. Vehicle (n = 5), AZD6244 (n = 4), AZD8186 (n = 5), AZD6244/AZD8186 (n = 5). P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). Right, longitudinal growth of individual WM266.4 xenografts treated with vehicle (n = 5) vs. AZD6244 (n = 5), AZD8186 (n = 5), or AZD6244/AZD8186 (n = 5). B, Left, Kaplan–Meier overall survival of BRAF PTEN mice treated with vehicle (n = 8; median survival, 12.5 days), AZD6244 (n = 8; median survival, 74.5 days), AZD8186 (n = 8; median survival, 15 days), or AZD6244/AZD8186 (n = 8; median survival, 139 days). P values calculated using the log-rank (Mantel–Cox) test: vehicle vs. AZD6244, P = 0.009550; vehicle vs. AZD8186, P = 0.437704; vehicle vs. AZD6244/AZD8186, P = 0.000012 and AZD6244 vs. AZD6244/AZD8186, P = 0.004813. Middle, relative change in tumor volume of BRAF PTEN mice over the first 7 days of indicated treatment. Vehicle (n = 8), AZD6244 (n = 7), AZD8186 (n = 8), AZD6244/AZD8186 (n = 11). P values calculated by one-way ANOVA corrected for multiple comparisons (Tukey). Right, longitudinal growth of individual tumors from vehicle-treated (n = 8) vs. AZD6244-treated (n = 7), AZD8186-treated (n = 8), and AZD6244/AZD8186-treated (n = 11) BRAF PTEN cohorts. Note that the BRAF PTEN vehicle- and AZD6244-treated cohort data are also used in Figs. 2 and 3. C, Left, representative H&E staining and IHC against pERK1/2 (Thr202/Tyr204), pRPS6 (Ser235/Ser236), pRPS6 (Ser240/Ser244), cyclin D1, and p21 in BRAF PTEN mice treated with indicated treatment. Right, representative H&E staining and IHC against cyclin D1 and p21 in tumor sections from BRAF PTEN PREX2 mice treated with vehicle or AZD6244. D, Top, Kaplan–Meier overall survival of BRAF PTEN PREX2-GD mice treated with vehicle (n = 8; median survival, 11 days), AZD6244 (n = 8; median survival, 100.5 days), AZD8186 (n = 4; median survival, 7 days), or AZD6244/AZD8186 (n = 7; median survival, 87 days). P values calculated using the log-rank (Mantel–Cox) test: vehicle vs. AZD6244, P = 0.00008; vehicle vs. AZD8186, P = 0.73075; vehicle vs. AZD6244/AZD8186, P = 0.00328 and AZD6244 vs. AZD6244/AZD8186, P = 0.11574. Bottom, longitudinal growth of individual tumors from vehicle-treated (n = 7) vs. AZD6244-treated (n = 6), AZD8186-treated (n = 3), and AZD6244/AZD8186-treated (n = 7) BRAF PTEN PREX2-GD cohorts. NB: the BRAF PTEN PREX2-GD vehicle- and AZD6244-treated cohort data are also used in Fig. 2C. E, Schematic of proposed proproliferative relationship between p110β and PREX2 in melanoma in vivo. GPCR, G-protein–coupled receptor; PIP₃, phosphatidylinositol-3,4,5-trisphosphate; RTK, receptor tyrosine kinase. [E, Created in BioRender. Campbell, A. (2025), https://BioRender.com/a89p130.]