The tumor stroma and its cellular components are known to play an important role in tumor response to treatment. Here, we report a novel resistance mechanism in melanoma that is elicited by BRAF inhibitor (BRAFi)–induced noncanonical activation of nuclear β-catenin signaling in cancer-associated fibroblasts (CAF). Treatment with BRAFi leads to an expanded CAF population with increased β-catenin nuclear accumulation and enhanced biological properties. This CAF subpopulation is essential for melanoma cells to proliferate and acquire resistance to BRAFi/MEK inhibitors (MEKi). Mechanistically, BRAFi induces BRAF-CRAF heterodimerization and subsequent activation of ERK signaling in CAFs, leading to inactivation of the β-catenin destruction complex. RNA-seq identified periostin (POSTN) as a major downstream effector of β-catenin in CAFs. POSTN compensates for the loss of β-catenin in CAFs and mediates melanoma cell BRAFi/MEKi resistance. In melanoma cells, POSTN activates phosphoinositide 3-kinase (PI3K)/AKT signaling and subsequently reactivates the ERK pathway that was inhibited by BRAFi/MEKi. Collectively, these data underscore the role of BRAFi-induced CAF reprogramming in matrix remodeling and therapeutic escape of melanoma cells.

Significance:

β-Catenin activation in cancer-associated fibroblasts in response to BRAF inhibitors stimulates POSTN secretion to promote resistance in cancer cells, revealing POSTN as a potential matrix target in cancer therapy.

Resistance to targeted therapies is a persistent hurdle in malignant melanoma treatment and can occur through multiple mechanisms. However, current studies of melanoma drug resistance mainly focus on intrinsic molecular alterations and have largely overlooked extrinsic resistance mechanisms. Tumors are complex tissues consisting of mixed populations of cancer cells and stromal cells embedded in the dense and stiff extracellular matrix (ECM; ref. 1). These heterogeneous cell populations produce a myriad of cell–cell and cell–ECM interactions, contributing to tumor cell adaptation and, eventually, tumor recurrence (2). An increasing number of studies report that treatment-induced changes can occur in multiple stromal cell compartments and lead to the development of drug resistance phenotypes (3, 4). It was also reported that treatment-relapsed melanomas converge on a shared resistance phenotype characterized by remodeled ECM and increased tumor stiffness (5).

One notorious key component of the tumor stroma is cancer-associated fibroblasts (CAF). Although many other stromal cell types and tumor cells can produce ECM proteins, CAFs appear to be the major player in the stroma that synthesize, secrete, assemble, and modify the ECM composition, organization, and stiffness (6). Therefore, CAFs could be a rich and complex source for melanoma cells to rely on in case of a therapeutic attack. However, CAFs are not a monolithic cell population and include a dynamic collection of fibroblast subsets with different characteristics and functions (7, 8). For example, the Tuveson group identified two distinct CAF subpopulations in the pancreatic tumor microenvironment, including a myofibroblast-like CAF population, termed “myCAFs,” and an inflammatory CAF population, termed “iCAFs” (9). As such, a better understanding of heterogeneous CAF populations is urgently needed to solve the mystery of melanoma drug resistance.

Previously, we reported that β-catenin, a molecule driving normal and pathologic responses in fibroblasts during wound healing, fibrosis, and keloid pathogenesis, is essential for the biological properties of dermal fibroblasts (10). We demonstrated that targeted β-catenin ablation in CAFs led to suppressed melanoma growth in vitro and in vivo (11). In this study, we have investigated how BRAF inhibitor (BRAFi) stimulates CAFs through nuclear β-catenin activation to remodel the ECM microenvironment, eliciting alternative signaling in BRAF-mutant melanoma cells to bypass BRAFi/MEKi inhibition. We found that BRAFi leads to BRAF and CRAF dimerization and increased nuclear β-catenin accumulation in CAFs. We identified matricellular protein periostin (POSTN) as an important downstream effector of β-catenin in CAFs. POSTN is uniquely expressed in CAFs, and BRAFi can upregulate POSTN production in CAFs but not in melanoma cells. Recombinant POSTN can compensate for the loss of β-catenin in CAFs and contribute to melanoma cell BRAFi/MEKi resistance. Importantly, we determined that POSTN activates PI3K/AKT signaling in melanoma cells, which reactivates ERK signaling under BRAFi inhibition. Together, our data reveal a novel BRAFi-stimulated, CAF-mediated, targeted drug resistance pathway in melanoma and POSTN as an exciting therapeutic candidate to overcome melanoma resistance to BRAFi/MEKi.

Human melanoma samples

Human melanoma tissue sections were provided by the University of Cincinnati Biorepository and the Center for Rare Melanomas at the University of Colorado. All melanoma tissue samples were previously obtained by both institutions with written informed consent from the patients, approved by the Institutional Review Board, and not collected specifically for this study. The studies were conducted in accordance with recognized ethical guidelines (e.g., Declaration of Helsinki, CIOMS, Belmont Report, U.S. Common Rule).

Cell lines

Human melanoma cell lines, including A375 (Cat# CRL-1619, RRID:CVCL_0132) and SK-MEL-24 (Cat# HTB-71, RRID:CVCL_0599), were purchased from ATCC. Human melanoma cell lines, including 1205Lu (1205 Lu-01–0001, RRID:CVCL_5239), WM278 (WM278–01–0001, RRID:CVCL_6473), and WM1366 (WM1366–01–0001, RRID:CVCL_6789), were purchased from Rockland Immunochemicals. Four CAF cell lines (DT01056P1, 224350P1, DT01027P1, and DT01028P1) isolated from surgically excised human melanoma tissues were purchased from Asterand Bioscience and designated as M56, M50, M27, and M28, respectively. CAF M77 and M80 were isolated and provided by Dr. Guangyong Peng at the Saint Louis University School of Medicine (St. Louis, MO). Human dermal fibroblasts were provided by Dr. Ana Luisa Kadekaro from the Department of Dermatology at the University of Cincinnati (Cincinnati, OH). For melanoma cell lines, they were authenticated by either ATCC or Rockland Immunochemicals and routinely validated in the laboratory based on the expression of melanoma proteoglycan antigens and surface molecules, including GP100, MITF, Melan-A, and tyrosinase. For CAFs and normal human dermal fibroblasts, we authenticated them by evaluating their unique morphologic characteristics and the expression of specific skin fibroblast markers, including TE7, vimentin, and α-smooth muscle actin (αSMA; ref. 12). All cell lines were passaged less than 30 times and maintained in DMEM supplemented with 10% (v/v) fetal bovine serum (FBS), 1% (v/v) 10,000 U/mL penicillin, and 10,000 U/mL streptomycin in a humidified incubator at 37°C with 5% CO2. Cell culture reagents were purchased from Thermo Fisher Scientific unless otherwise stated. Experimental procedures involving biosafety issues were carried out under the University of Cincinnati Institutional Biosafety Committee protocol 16–08–17–01.

Lentiviral shRNA

To stably silence β-catenin expression, we transduced CAFs, including M27, M50, and M56, with inducible lentiviruses expressing shRNA that specifically target β-catenin expression (GE Dharmacon, cat# V3SH7675–02EG1499; ref. 10). An inducible lentivirus expressing a nontarget shRNA (GE Dharmacon, cat# VSC6572) was used to generate control CAFs. The lentiviral vector carries genes encoding a puromycin-resistant protein and green fluorescent protein (GFP). Thus, transduced CAFs can be selected by puromycin. The inducible lentiviral shRNA vector, which utilizes the Tet-On inducible system, only allows the expression of either β-catenin–targeting shRNA or nontarget shRNA when cells are treated with doxycycline (Sigma, 33429). Upon induction, GFP expression is simultaneously induced in CAFs so that the cells expressing shRNA can be visually tracked by green fluorescence. To distinguish melanoma cells from CAFs, nontarget shRNA expressing red fluorescent protein (RFP; GE Dharmacon, cat# VSC6573) was used to tag melanoma cells with red fluorescence.

Three-dimensional drug resistance assay

To form cell spheroids, 5,000 RFP-tagged A375 or SK-MEL-24 cells were mixed with 5,000 GFP-tagged CAFs in one well of a U-bottomed 96-well plate (Thermo Fisher Scientific). Spheroids were formed by centrifuging the plate at 1,000 rpm for 10 minutes followed by incubation at 37°C in a humidified incubator with 5% CO2 overnight. For the drug resistance assay, spheroids were cultured using standard culture medium or medium containing 500 nmol/L PLX4032 or 10 nmol/L GDC0973 for 72 hours before they were collected for cell number counting. For POSTN rescue assay, 250 ng/mL recombinant POSTN protein (R&D Systems, 3548-F2–050) were added to the spheroids containing β-catenin-deficient or bcat-GFP/M27 or bcat-GFP/M50 while the spheroids were also treated with 500 nmol/L PLX4032 or a combination of 500 nmol/L PLX4032 and 10 nmol/L GDC0973. To count RFP-positive (RFP+) melanoma cell number in the spheroids, 12 spheroids from each group were collected and digested using 2 mg/mL collagenase IV (Therma Fisher Scientific, 17104019) for one hour with stirring to generate a single-cell suspension. A Countess II Automated Cell Counter (Thermo Fisher Scientific) was used to quantify melanoma cell numbers based on red fluorescence. The average number of melanoma cells in each spheroid was calculated by dividing the total cell number by 12.

Tumor xenografting experiments

All animal experiments were performed in accordance with the protocol approved by the Institutional Animal Care and Use Committee of the University of Cincinnati (10–09–23–01). The human melanoma xenograft model was established using a combination of BRAF-mutant human melanoma cell lines, including A375 and SK-Mel-24, and CAF cell lines, including M27 and M50. To induce melanomas in mice, a mix of 1 × 106 A375 cells and 1 × 106 uninduced GFP/M50 or bcat-GFP/M50; or 1 × 106 SK-Mel-24 cells and 1 × 106 uninduced GFP/M27 or bcat-GFP/M27 in 50 μL of growth factor–reduced Matrigel/PBS (1:1; BD Biosciences) was injected intradermally into the flanks of 4- to 6-week-old NOD.Cg-Prkdcscid/J mice (IMSR Cat# JAX:001303, RRID:IMSR_JAX:001303), respectively. The mice were observed daily for tumor appearance. To induce β-catenin ablation, all mice were fed a doxycycline diet (Bio-Serv, F-4096) for four weeks after the tumors reached a volume size of approximately 50 cubic millimeters. The mice in each group received either corn oil or drug treatment as indicated (vemurafenib/PLX4032, 50 mg/kg every other day; cobimetinib/GDC0973, 20 mg/kg every other day). Meanwhile, the tumor size was measured and recorded every other day until the endpoint, when the tumor size exceeded 20% of body size. After the mice were euthanized, the tumors were harvested for various analyses.

POSTN assays

For the 2D recombinant POSTN assay, A375 or SK-MEL-24 human melanoma cells were seeded in 6-well plates coated with 2.5 ng/μL POSTN. Melanoma cells cultured in the wells without coated POSTN were used as a control. The cells were then treated with 500 nmol/L PLX4032 and/or 10 nmol/L GDC0973 for 72 hours before they were collected for cell counting. Images of 2D-cultured RFP+ melanoma cells were captured using a Cytation 1 cell imaging multi-mode reader (BioTek Instruments).

For POSTN treatment, Western blotting, A375, or SK-MEL-24 were cultured in 10-cm dishes coated with or without POSTN. For BRAFi treatment, 500 nmol/L PLX4032 was added. Two days later, 1.25 μg of POSTN was added to each dish coated with POSTN. For Akt inhibitor treatment, 50 μmol/L LY294002 (Selleck Chemicals, S1105) was added to the dishes as indicated. Following another two-day incubation, the cells were collected for ERK, pERK, Akt, and pAKT Western blotting.

For three-dimensional (3D) POSTN assays, 8,000 melanoma cells were seeded into one well of a U-bottomed 96-well plate to generate the spheroids as described above. The spheroids were cultured using normal culture medium or medium containing 500 nmol/L PLX4032 and/or 10 nmol/L GDC0973 with or without POSTN at a concentration of 250 ng/mL as indicated for 72 hours before they were collected for cell number counting. Images of the spheroids were captured using a Cytation 1 cell imaging multi-mode reader.

Statistical analysis

All quantitative results were obtained from a minimum of three independent experiments. All data were analyzed using the GraphPad Prism 9 software package (GraphPad Prism, RRID:SCR_002798) and expressed as the mean ± SEM. Differences between means were determined by Student t tests and were considered statistically significant at P < 0.05.

Additional materials and methods

Additional materials and methods used in this study can be found in Supplementary File S2.

Data availability

The data generated in this study are available within the article and its supplementary data files. RNA sequencing (RNA-seq) data generated in this study are publicly available in Gene Expression Omnibus (GEO) at GSE121186.

A subset of CAFs with increased nuclear β-catenin accumulation is expanded in BRAFi/MEKi–treated melanoma

To understand the roles of CAFs in melanoma resistance to BRAFi and MEKi, we performed a paired-analysis of melanoma tissue samples collected from patients carrying BRAFV600-mutated melanomas before and after BRAFi/MEKi therapy (Supplementary Table S1). CAFs in melanomas were identified using a known CAF marker, αSMA (13). Notably, increased numbers of αSMA–positive (αSMA+) CAFs expressed nuclear β-catenin after BRAFi/MEKi treatment (Fig. 1AD). As shown in Fig. 1E and F, not only the total number of αSMA+ CAFs was increased in melanomas isolated from the same patients after BRAFi/MEKi treatment, the percentage of αSMA+ cells expressing nuclear β-catenin was increased from 17% ± 2% to 41% ± 2%.

Figure 1.

BRAFi drives increased nuclear β-catenin in CAFs. A–D, Human melanoma sections [collected from eight patients with melanoma before (A) and after (B) BRAFi/MEKi therapy] were coimmunostained for αSMA (green) and β-catenin expression (red). Nuclei were stained blue with DAPI. C and D are close-up pictures of the area circled by white boxes in A and B, respectively. Yellow triangles in D indicate αSMA+ fibroblasts expressing nuclear β-catenin. Scale bar, 100 μm. E, The scattered plot represents the numbers of αSMA+ cells per mm2 counted in pre- and posttreated melanoma samples. F, Percentage of αSMA+ cells expressing nuclear β-catenin in pre- and posttreated melanoma samples. Each data point represents the quantification result of one picture. At least five pictures were taken per tumor section. n ≥ 40. G–L, Immunostaining of β-catenin in M50 treated with DMSO or indicated inhibitors for 72 hours. M, Quantification and comparison of relative fluorescence intensities of nuclear β-catenin in M50 treated with DMSO or indicated BRAFi/MEKi for 72 hours. The fluorescence intensity of nuclear β-catenin in DMSO-treated M50 was used as the control and set at one for comparison. Each data point represents the relative fluorescent intensities of nuclear β-catenin in one treated cell. n = 30. N, Nuclear and cytoplasmic fractions of M50 treated with DMSO or different concentrations of PLX4032. Lamin B was used as a nuclear loading control and tubulin as a cytoplasmic loading control. O and P, Quantification of fluorescence intensities of nuclear and cytoplasmic β-catenin Western blot bands. n = 3. For all graphs, data are represented as mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ns, not significant.

Figure 1.

BRAFi drives increased nuclear β-catenin in CAFs. A–D, Human melanoma sections [collected from eight patients with melanoma before (A) and after (B) BRAFi/MEKi therapy] were coimmunostained for αSMA (green) and β-catenin expression (red). Nuclei were stained blue with DAPI. C and D are close-up pictures of the area circled by white boxes in A and B, respectively. Yellow triangles in D indicate αSMA+ fibroblasts expressing nuclear β-catenin. Scale bar, 100 μm. E, The scattered plot represents the numbers of αSMA+ cells per mm2 counted in pre- and posttreated melanoma samples. F, Percentage of αSMA+ cells expressing nuclear β-catenin in pre- and posttreated melanoma samples. Each data point represents the quantification result of one picture. At least five pictures were taken per tumor section. n ≥ 40. G–L, Immunostaining of β-catenin in M50 treated with DMSO or indicated inhibitors for 72 hours. M, Quantification and comparison of relative fluorescence intensities of nuclear β-catenin in M50 treated with DMSO or indicated BRAFi/MEKi for 72 hours. The fluorescence intensity of nuclear β-catenin in DMSO-treated M50 was used as the control and set at one for comparison. Each data point represents the relative fluorescent intensities of nuclear β-catenin in one treated cell. n = 30. N, Nuclear and cytoplasmic fractions of M50 treated with DMSO or different concentrations of PLX4032. Lamin B was used as a nuclear loading control and tubulin as a cytoplasmic loading control. O and P, Quantification of fluorescence intensities of nuclear and cytoplasmic β-catenin Western blot bands. n = 3. For all graphs, data are represented as mean ± SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ns, not significant.

Close modal

BRAFi drives increased nuclear β-catenin in CAFs

To understand whether increased nuclear β-catenin in CAFs is driven by either BRAFi or MEKi, we obtained six CAF cell lines isolated from surgically excised human melanoma tissues (M27, M28, M50, M56, M77, and M80). To ensure that CAFs do not carry BRAF and NRAS mutations, we performed testing on M27, M28, M50, and M56 using Sanger sequencing. None of the tested CAF cell lines carry mutations in BRAF or NRAS genes (Supplementary File 3).

As expected, the CAF cell lines express increased levels of known CAF markers, including αSMA, FAP, and FSP-1 (also known as S100A4), compared with normal human dermal fibroblasts (NF; Supplementary Fig. S1A–S1C). Quantitative analysis revealed that M27 and M50 express the highest levels of these markers (Supplementary Table S2). In addition, all CAF cell lines exhibit pronounced filamentous actin (F-actin) expression (Supplementary Fig. S1D), which is known to be involved in migration and cell division (14). As shown by the scratch assay, all six CAF cell lines exhibited increased migratory abilities compared with normal fibroblasts (Supplementary Fig. S1E). Collagen gel contraction assays, as a measure of the ability to remodel the ECM, showed that all six CAF cell lines had enhanced variable ability to contract the gel, among which, M27 and M50 showed the strongest abilities (Supplementary Fig. S1F).

Next, we investigated whether BRAFi or MEKi led to increased nuclear β-catenin in CAFs. Interestingly, all three BRAF inhibitors (PLX4032, GSK2118436, and LGX818) led to increased nuclear β-catenin in M50 compared with DMSO-treated cells (Fig. 1GJ) while the nuclear β-catenin levels in MEKi-treated M50 (GDC0973 and GSK1120212) remained unchanged (Fig. 1KL). This revelation suggests that increased nuclear β-catenin in CAFs is driven by BRAFi but not MEKi (Fig. 1M). To confirm this finding, the nuclear and cytoplasmic fractions of M50 treated with either DMSO or different concentrations of PLX4032 were extracted for Western blot analysis. As shown in Fig. 1N, PLX4032-treated M50 showed increased amounts of nuclear β-catenin in a concentration-dependent manner (Fig. 1O), while the levels of cytoplasmic β-catenin in M50 remained unchanged under PLX4032 treatment (Fig. 1P). M27 and M56 showed a similar β-catenin expression pattern upon PLX4032 treatment (Supplementary Fig. S2).

PLX4032 stimulates the biological properties of CAFs

Seventy-two hours after PLX4032 treatment, M50 not only showed an increase in nuclear β-catenin (Supplementary Fig. S3A) but also exhibited a more pronounced F-actin expression (Supplementary Fig. S3B) and increased levels of focal adhesion protein paxillin (Supplementary Fig. S3C) and phosphorylated myosin light chain 2 (p-MLC2; Supplementary Fig. S3D). We next asked whether PLX4032 can enhance the biological functions of CAFs. Compared with M50 treated with DMSO (vehicle), the gel-contracting ability of M50 was increased by PLX4032 treatment (61.4% ± 2.9% vs. 37.6% ± 2.1%; Supplementary Fig. S3E). Moreover, confocal reflection microscopy (CRM) revealed increased fiber connectivity in the gel embedded with PLX4032-treated M50 (Supplementary Fig. S3F). Connectivity denotes the extent of line connections in a network, indicating the complexity of collagen fibers in the gel. In addition, the migratory ability of M50 was significantly increased by PLX4032 (87.8% ± 1.6%) compared with M50 treated with DMSO (68.0% ± 1.5%; Supplementary Fig. S3G). The stiffness of the gels embedded with M50 increased substantially from 1,038 ± 70 Pa to 5,863 ± 74 Pa upon PLX4032 treatment (Supplementary Fig. S3H). M27 and M56 showed similar phenotypes under PLX4032 treatment. The data collectively reveal that PLX4032 enhances cytoskeletal dynamics and matrix remodeling in CAFs.

β-Catenin depletion eliminates the ability of CAFs to respond to PLX4032 stimulation

To understand whether β-catenin is essential for CAFs to respond to BRAFi, β-catenin expression in M27 and M50 was silenced using a doxycycline-inducible shRNA coexpressing GFP, as we have reported previously (10). M50 and M27 cells carrying β-catenin shRNA are designated as bcat-GFP/M50 and bcat-GFP/M27. Control GFP/M50 and GFP/M27 were generated by transducing the cells with a GFP-tagged nontargeting lentiviral shRNA. Western blotting showed that the expression of β-catenin was ablated efficiently in bcat-GFP/M50 (Supplementary Fig. S4A and S4B).

Depleting β-catenin reduced the viability of M50 cells as indicated by cell number counting (Supplementary Fig. S4C) and the MTT assay (Supplementary Fig. S4D). Suppressed M50 expansion was mainly caused by decreased cell proliferation but not by increased cell apoptosis. As shown in Supplementary Fig. S4E–S4G, 90% ± 1% of GFP/M50 cells were Ki67-positive (Ki67+), while only 72% ± 2% of bcat-GFP/M50 cells were Ki67+. No significant difference in the numbers of TUNEL-positive (TUNEL+) cells was detected between GFP/M50 and bcat-GFP/M50 (Supplementary Fig. S4H–S4J). β-Catenin deficiency also weakened the ability of M50 cells to contract a collagen gel (Supplementary Fig. S4K and S4L). Similarly, the density of collagen fibers in the gel embedded with bcat-GFP/M50 was lower than that of the gel embedded with GFP/M50 (Supplementary Fig. S4M). Quantification in Supplementary Fig. S4N shows that the connectivity of collagen fibers in the gel embedded with bcat-GFP/M50 was significantly less than that of the gel embedded with GFP/M50 (0.54 ± 0.07 vs. 4.64 ± 0.23). β-Catenin deficiency also reduced the ability of M50 to migrate as shown by the scratch assay (Supplementary Fig. S4O and S4P). The stiffness of the gel embedded with bcat-GFP/M50 was 122.0 ± 9.3 Pa compared with 1,007 ± 32.2 Pa for the gel embedded with GFP/M50 (Supplementary Fig. S4Q).

Upon β-catenin depletion (Supplementary Fig. S5A), M50 not only exhibited reduced levels of F-actin (Supplementary Fig. S5B), paxillin (Supplementary Fig. S5C), and p-MLC2 (Supplementary Fig. S5D), but they also no longer responded to PLX4032 stimulation. PLX4032 failed to enhance the gel contracting ability of bcat-GFP/M50 (Supplementary Fig. S5E). CRM did not detect increased fiber connectivity in the gel embedded with PLX4032-treated bcat-GFP/M50 (Supplementary Fig. S5F). The migratory ability of bcat-GFP/M50 was not improved upon PLX4032 treatment (Supplementary Fig. S5G). The data demonstrate that β-catenin activity is essential for CAFs to respond to PLX4032. The data also suggest that depleting β-catenin could be an effective way to eliminate this unique subset of BRAFi-stimulated CAFs.

BRAFi-stimulated CAFs contribute to melanoma cell resistance to BRAFi and MEKi

To understand whether this subset of CAFs is relevant to BRAF-mutant melanoma cell resistance to BRAFi and MEKi, we designed a multicolor 3D coculture system consisting of human melanoma cells and CAFs for a drug resistance assay. Mitomycin-treated CAFs (designated as mito-GFP/M27 and mito-GFP/M50) were used as nonfunctional CAF controls.

As shown in Fig. 2A and B, after a 72-hour coculture, there were 29% ± 4% more A375 cells in A375+GFP/M50 spheroids compared with A375+mito-GFP/M50 spheroids. However, no significant difference in A375 numbers was observed between A375+bcat-GFP/M50 and A375+mito-GFP/M50. This indicates that β-catenin depletion deprived M50 cells of their ability to support A375 growth. Because BRAFV600E mutation leads to abnormal activation of MAPK/ERK signaling (15), we examined the level of pERK in A375 in 3D cocultured spheroids. M50 cells in the spheroids were identified using a fibroblast-specific antibody, TE7 (12). As shown in Fig. 2C, only 42% ± 2% of A375 expressed pERK when cocultured with mito-GFP/M50, while 56% ± 2% of A375 expressed pERK when cocultured with GFP/M50. β-Catenin ablation in M50 cells had adverse effects on ERK signaling in A375 (39% ± 1% pERK–positive cells).

Figure 2.

BRAFi-stimulated CAFs require β-catenin to contribute to melanoma cell resistance to BRAFi and MEKi. A, Top, fluorescence images of spheroids consisting of A375 (red) and M50 (green). Bottom, images using an anti-TE7 antibody (green) and an anti-pERK antibody (red) with DAPI counterstaining (blue). Scale bar, 200 μm. B, The bar graph shows the mean A375 number per spheroid. n = 3. For each group per experiment, 12 spheroids were collected for cell counting, and the number was divided by 12 to obtain the mean A375 number. C, The scatter graph shows the percentages of A375 cells (TE7-) expressing pERK in each spheroid type. Each data point represents the quantified percentage in each spheroid. n ≥ 10. D, Representative images of cocultured spheroids treated with PLX4032 for 72 hours. The same experimental procedure was performed as described above in DMSO-treated groups. E, A bar graph shows the mean A375 numbers per spheroid from three independent experiments. F, The scatter graph shows the percentages of A375 cells (TE7-) expressing pERK in each spheroid type. Each data point represents the quantified percentage in each spheroid. n ≥ 10. In all graphs, the data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Figure 2.

BRAFi-stimulated CAFs require β-catenin to contribute to melanoma cell resistance to BRAFi and MEKi. A, Top, fluorescence images of spheroids consisting of A375 (red) and M50 (green). Bottom, images using an anti-TE7 antibody (green) and an anti-pERK antibody (red) with DAPI counterstaining (blue). Scale bar, 200 μm. B, The bar graph shows the mean A375 number per spheroid. n = 3. For each group per experiment, 12 spheroids were collected for cell counting, and the number was divided by 12 to obtain the mean A375 number. C, The scatter graph shows the percentages of A375 cells (TE7-) expressing pERK in each spheroid type. Each data point represents the quantified percentage in each spheroid. n ≥ 10. D, Representative images of cocultured spheroids treated with PLX4032 for 72 hours. The same experimental procedure was performed as described above in DMSO-treated groups. E, A bar graph shows the mean A375 numbers per spheroid from three independent experiments. F, The scatter graph shows the percentages of A375 cells (TE7-) expressing pERK in each spheroid type. Each data point represents the quantified percentage in each spheroid. n ≥ 10. In all graphs, the data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Close modal

When the spheroids were treated with PLX4032 for 72 hours (Fig. 2D and E), there were significantly more melanoma cells in A375+GFP/M50 spheroids than in A375+mito-GFP/M50 spheroids (119% ± 8% more A375 cells). This shows that M50 cells contributed to the resistance of A375 against PLX4032. β-Catenin ablation significantly reduced the ability of M50 cells to elicit the resistance of A375 to PLX4032. The number of A375 in A375+bcat-GFP/M50 spheroids was only 60% ± 1% of that in A375+GFP/M50 spheroids. In line with these findings, β-catenin ablation in GFP/M50 reduced the level of pERK to 19% ± 2% in A375 when treated with PLX4032 (Fig. 2F). These results were validated using another BRAFV600E melanoma cell line, SK-Mel-24, and CAF cell line, M27 (Supplementary Fig. S6). The data demonstrate that there is a β-catenin–dependent cross-talk between melanoma cells and CAFs that allows melanoma cells to bypass the inhibition of BRAFi on MAPK/ERK signaling.

We evaluated whether β-catenin is equally important for CAFs to contribute to melanoma resistance to BRAFi/MEKi. As shown in Supplementary Fig. S7, under PLX4032 and GDC0973 treatment, the numbers of A375 in the spheroids cocultured with either M50 or M27 cells were significantly higher than those cocultured with mitomycin-treated M50, mitomycin-treated M27, β-catenin–deficient M50 or β-catenin–deficient M27.

Depleting β-catenin in BRAFi-stimulated CAFs sensitizes BRAF-mutant melanoma cells to BRAFi in vivo

To investigate whether β-catenin depletion in CAFs has the potential to improve the responsiveness of BRAF-mutant melanoma cells to BRAFi in vivo, we established a human melanoma xenograft model using a combination of A375 cells with GFP/M50 or bcat-GFP/M50. On day 28, in immunodeficient NOD.Cg-Prkdcscid/J mice fed with corn oil, the size and weight of A375+bcat-GFP/M50 tumors (518.6 ± 36.8 mm3; 0.57 ± 0.04g; n = 10) were significantly smaller and lighter than those of control A375+GFP/M50 tumors (1,027 ± 113 mm3; 1.03 ± 0.10g; n = 10) as shown in Fig. 3AC. This confirms that β-catenin activity is important for CAFs to support melanoma growth. Similarly, in mice treated with PLX4032 for 28 days, β-catenin–deficient tumors (A375+bcat-GFP/M50) were also smaller and lighter (222.6 ± 14.2 mm3; 0.26 ± 0.02g; n = 11) than those of control A375 + GFP/M50 tumors (549.5 ± 26.6 mm3; 0.54 ± 0.02g; n = 11). These results show that ablating β-catenin in CAFs has the potential to increase the responsiveness of BRAF-mutant melanoma cells to BRAFi.

Figure 3.

Ablating β-catenin in BRAFi-stimulated CAFs sensitizes BRAF-mutant melanoma cells to BRAFi in vivo. A, Representative pictures of melanoma xenografts derived from indicated cell mixtures after administering corn oil or PLX4032 for 28 days. B and C, Comparative analysis of the volumes (B) and weights (C) of melanoma xenografts between four experimental groups. Box and whisker plot representing volumes of melanoma xenografts at each indicated time point. Graph shows tumor weight comparisons among four groups at day 28. Each data point represents the tumor weight of one tumor. n ≥ 10. D, Images show αSMA staining of melanoma tissues collected from each group (nuclei DAPI/blue). E, Graph shows the percentages of αSMA+ fibroblasts in total cells from tumors derived from indicated groups. Each data point represents the fibroblast percentage in one 40× field. n ≥ 15. F, Images show collagen staining of melanoma tissue sections in indicated groups (red). G, Images show GP-100 (melanoma cells, green) and Ki67 (red) staining of melanoma tissue sections in indicated groups. H, Graph shows the percentages of GP100+ melanoma cells that are Ki67+ among the groups. Each data point represents the percentage of Ki67+ melanoma cells in one 40× field. n ≥ 15. I, Images show GP-100 and CC3+ (red) staining of melanoma tissues sections of indicated groups. J, Graph shows the percentages of GP100+ melanoma cells that are CC3+ among the groups. Each data point represents the percentage of CC3+ melanoma cells in one 40× field. n ≥ 15. K, Images show GP-100 (green) and pERK (red) staining of melanoma tissues sections in indicated groups. L, The scatter graph shows the percentages of GP100+ melanoma cells that are pERK+ among the groups. Each data point represents the percentage of pERK+ melanoma cells in one 40× field. n ≥ 15. For all staining pictures, the scale bar represents 100 μm. In all graphs, data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Figure 3.

Ablating β-catenin in BRAFi-stimulated CAFs sensitizes BRAF-mutant melanoma cells to BRAFi in vivo. A, Representative pictures of melanoma xenografts derived from indicated cell mixtures after administering corn oil or PLX4032 for 28 days. B and C, Comparative analysis of the volumes (B) and weights (C) of melanoma xenografts between four experimental groups. Box and whisker plot representing volumes of melanoma xenografts at each indicated time point. Graph shows tumor weight comparisons among four groups at day 28. Each data point represents the tumor weight of one tumor. n ≥ 10. D, Images show αSMA staining of melanoma tissues collected from each group (nuclei DAPI/blue). E, Graph shows the percentages of αSMA+ fibroblasts in total cells from tumors derived from indicated groups. Each data point represents the fibroblast percentage in one 40× field. n ≥ 15. F, Images show collagen staining of melanoma tissue sections in indicated groups (red). G, Images show GP-100 (melanoma cells, green) and Ki67 (red) staining of melanoma tissue sections in indicated groups. H, Graph shows the percentages of GP100+ melanoma cells that are Ki67+ among the groups. Each data point represents the percentage of Ki67+ melanoma cells in one 40× field. n ≥ 15. I, Images show GP-100 and CC3+ (red) staining of melanoma tissues sections of indicated groups. J, Graph shows the percentages of GP100+ melanoma cells that are CC3+ among the groups. Each data point represents the percentage of CC3+ melanoma cells in one 40× field. n ≥ 15. K, Images show GP-100 (green) and pERK (red) staining of melanoma tissues sections in indicated groups. L, The scatter graph shows the percentages of GP100+ melanoma cells that are pERK+ among the groups. Each data point represents the percentage of pERK+ melanoma cells in one 40× field. n ≥ 15. For all staining pictures, the scale bar represents 100 μm. In all graphs, data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Close modal

As we observed increased numbers of fibroblasts and fibroblasts expressing nuclear β-catenin in post-BRAFi/MEKi treated human melanomas (Fig. 1), we asked how β-catenin depletion affects the number of M50 in melanoma xenografts. As shown by αSMA immunostaining in Fig. 3D and E, in the tumors derived from the mixture of A375+GFP/M50 without PLX4032 treatment, 35±2% of the cells are αSMA+ fibroblasts. In A375+bcat-GFP/M50 tumors, the percentage of αSMA+ fibroblasts was reduced to 11% ± 1%. This reduction is consistent with the reduction in melanoma size and weight. In PLX4032-treated melanomas, the percentage of αSMA+ fibroblasts in A375+GFP/M50 tumors was increased to 48% ± 2% and conversely reduced to 10% ± 1% in A375+bcat-GFP/M50 tumors. The numbers of αSMA+ fibroblasts expressing nuclear β-catenin in A375+GFP/M50 tumors with and without BRAFi (Supplementary Fig. S8A–S8D) showed the same trend (Supplementary Fig. S8E) that was discovered in human melanoma samples (Fig. 1F). CAFs are known to be an important source of fibrillar collagen. PLX4032 treatment promoted collagen production in the melanomas containing with normal CAFs (Fig. 3F). However, a significant reduction of collagen content was observed in the tumors derived from the mixture of A375+bcat-GFP/M50 with or without PLX4032 treatment.

Next, we investigated the proliferation and apoptosis of melanoma cells (labeled by GP100). As shown in Fig. 3G and H, in A375+GFP/M50 melanomas without PLX4032 treatment, 25% ± 1% of melanoma cells were Ki67+. Depleting β-catenin in M50 (bcat-GFP/M50) resulted in a decrease in the number of Ki67+ melanoma cells (15% ± 2%). When A375 tumors carrying GFP/M50 were treated with PLX4032, 16% ± 1.7% of melanoma cells were Ki67+. However, in PLX4032-treated A375+bcat-GFP/M50 tumors, the percentages of Ki67+ melanoma cells were reduced to 11% ± 1%. Cleaved caspase-3 (CC3) is a critical executioner of apoptosis. As shown in Fig. 3I and J, in mice fed with corn oil, A375+GFP/M50 tumors had 7% ± 1% CC3-positive (CC3+) melanoma cells. This percentage increased to 15% ± 2% when β-catenin was depleted in CAFs. In mice treated with PLX4032, the percentages of melanoma cells expressing CC3 increased to 23% ± 2% in A375+bcat-GFP/M50 tumors, which was significantly higher than the number of CC3+ melanoma cells in A375+GFP/M50 tumors (10% ± 2%). The level of pERK in melanoma cells displayed the same trend as Ki67 (Fig. 3K and L). The data show that β-catenin activity is important for CAFs to contribute to melanoma cell proliferation, viability and BRAFi resistance.

As combined BRAF and MEK inhibition has shown improved clinical benefits for individuals with advanced melanoma, we carried out melanoma xenografts exposed to a combination of BRAFi/MEKi treatment. Smaller tumor size, diminished proliferation, and increased number of apoptotic cells were observed in melanomas carrying β-catenin–deficient CAFs compared with CAFs carrying wild-type β-catenin under BRAFi/MEKi treatment (Supplementary Fig. S9). These findings underscore the importance of BRAFi-induced β-catenin signaling in CAFs in melanoma drug resistance.

β-Catenin nuclear translocation is driven by paradoxical activation of RAF kinases in CAFs

β-Catenin is known to be phosphorylated by GSK3β and subsequently ubiquitinated and targeted for proteasomal degradation (16). As shown in Fig. 4A, in PLX4032-treated M50 cells, the level of phosphorylated GSK3β (p-GSK3β, marked for GSK3β degradation) displayed a two-fold increase while the amount of p-β-catenin (Ser33/37/Thr41, marked for β-catenin degradation) decreased significantly. Interestingly, pERK, which is known to have the ability to phosphorylate GSK3β, increased approximately 6-fold after M50 cells were treated with PLX4032.

Figure 4.

BRAFi-induced BRAF/CRAF dimerization drives β-catenin nuclear accumulation in CAFs. A, Western blotting of p-β-catenin (Ser33/37/Thr41), β-catenin, p-GSK3β (Ser9), ERK, and pERK levels in M50 treated with DMSO or PLX4032 for two hours. Charts show quantification of fluorescence intensities of indicated protein bands after DMSO or PLX4032 treatment. n = 3. B, Western blotting of BRAF, CRAF, ERK, pERK, and p-GSK3β in M50, in which BRAF, CRAF, or a combination of BRAF and CRAF were silenced, treated with DMSO or PLX4032, respectively. C, Chart shows fluorescent intensities of p-GSK3β Western blot bands. n = 3. D, Representative images show β-catenin immunostaining of PLX4032-treated M50 with silenced BRAF or CRAF or BRAF/CRAF expression (indicated by yellow triangles). E, Graph shows relative nuclear β-catenin expressions in M50 treated with PLX4032 after BRAF, CRAF, or a combination of BRAF and CRAF were silenced by siRNAs. The expression level of nuclear β-catenin was normalized by the mean nuclear β-catenin expression in M50 transfected by scramble siRNA after a 72-hour PLX4032 treatment. n ≥ 100. F, Western blotting of BRAF and CRAF in coimmunoprecipitated M50 samples treated with DMSO or PLX4032 for 72 hours using IgG or an anti-BRAF antibody. G, Chart shows fluorescence intensities of CRAF bands from indicated coimmunoprecipitation samples. n = 3. H, Single or double silencing of BRAF and CRAF partially deprives the ability of CAFs to promote A375 resistance to PLX4032. Chart shows the numbers of A375 in the spheroids cocultured with M50 with silenced BRAF, CRAF, or BRAF/CRAF expression after a 72-hour PLX4032 treatment. n = 3. For all graphs, the data are shown as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Figure 4.

BRAFi-induced BRAF/CRAF dimerization drives β-catenin nuclear accumulation in CAFs. A, Western blotting of p-β-catenin (Ser33/37/Thr41), β-catenin, p-GSK3β (Ser9), ERK, and pERK levels in M50 treated with DMSO or PLX4032 for two hours. Charts show quantification of fluorescence intensities of indicated protein bands after DMSO or PLX4032 treatment. n = 3. B, Western blotting of BRAF, CRAF, ERK, pERK, and p-GSK3β in M50, in which BRAF, CRAF, or a combination of BRAF and CRAF were silenced, treated with DMSO or PLX4032, respectively. C, Chart shows fluorescent intensities of p-GSK3β Western blot bands. n = 3. D, Representative images show β-catenin immunostaining of PLX4032-treated M50 with silenced BRAF or CRAF or BRAF/CRAF expression (indicated by yellow triangles). E, Graph shows relative nuclear β-catenin expressions in M50 treated with PLX4032 after BRAF, CRAF, or a combination of BRAF and CRAF were silenced by siRNAs. The expression level of nuclear β-catenin was normalized by the mean nuclear β-catenin expression in M50 transfected by scramble siRNA after a 72-hour PLX4032 treatment. n ≥ 100. F, Western blotting of BRAF and CRAF in coimmunoprecipitated M50 samples treated with DMSO or PLX4032 for 72 hours using IgG or an anti-BRAF antibody. G, Chart shows fluorescence intensities of CRAF bands from indicated coimmunoprecipitation samples. n = 3. H, Single or double silencing of BRAF and CRAF partially deprives the ability of CAFs to promote A375 resistance to PLX4032. Chart shows the numbers of A375 in the spheroids cocultured with M50 with silenced BRAF, CRAF, or BRAF/CRAF expression after a 72-hour PLX4032 treatment. n = 3. For all graphs, the data are shown as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Close modal

It has been shown that ATP-competitive RAF inhibitors can bind to and activate wild-type BRAF and consequently activate downstream signaling cascades (17–21). We hypothesized that increased nuclear β-catenin in BRAF wild-type CAFs is driven by BRAFi-induced BRAF and CRAF heterodimerization and activation of the downstream ERK signaling pathway. The expression of BRAF, CRAF, or a combination of BRAF and CRAF was silenced in M50 and M27 by siRNAs. As shown in Fig. 4B and Supplementary Fig. S10A, the expression of BRAF and/or CRAF was successfully knocked down in M27 and M50. Ablation of either BRAF or CRAF was able to reduce PLX4032-induced increases in pERK and p-GSK3β levels, indicating that BRAF and CRAF activities are required for PLX4032-induced nuclear β-catenin accumulation in M50 cells. Depleting both BRAF and CRAF blocked the pERK, and p-GSK3β increases in M27 and M50 (Fig. 4C; Supplementary Fig. S10B). The importance of BRAF and CRAF was further underscored by the fact that both BRAF and CRAF silencing suppressed BRAFi-induced β-catenin nuclear translocation, and ablation of both BRAF and CRAF was sufficient to diminish nuclear β-catenin accumulation in M50 induced by PLX4032 (Fig. 4D and E; Supplementary Fig. S10C). Coimmunoprecipitation was performed to determine whether BRAFi leads to BRAF/CRAF dimerization in CAFs. The binding of BRAF to CRAF could be observed in M50 cells in the absence of any inhibitor (Fig. 4F). PLX4032 treatment led to a 2-fold increase of BRAF/CRAF dimer formation (Fig. 4G). Three-dimensional coculture assays also showed that both BRAF and CRAF are required by CAFs to contribute to melanoma cell resistance to PLX4032 (Fig. 4H). Altogether, the above findings establish that BRAFi drives the activation of RAF kinases and elicits nuclear β-catenin signaling in CAFs.

To understand whether the MAPK/ERK pathway is a major pathway downstream of RAF activation that contributes to β-catenin nuclear accumulation in CAFs, we used MEK inhibitors, GDC0973 and GSK1120212, and ERK1/2 inhibitor, SCH772984, to treat CAFs under PLX4032 treatment (22). PLX4032-induced β-catenin nuclear accumulation in CAFs was significantly blocked by GDC0973, GSK1120212, and SCH772984 (Supplementary Figs. S11 and S12A and S12B). Furthermore, the increase of p-GSK3β induced by PLX4032 was also inhibited by SCH772984 (Supplementary Fig. S12C and S12D). Altogether, the abovementioned findings establish that BRAFi-induced nuclear β-catenin signaling is mainly mediated via RAF kinase activation and MEK/ERK signaling.

POSTN is a downstream effector of β-catenin signaling in CAFs

To obtain a global picture of the underlying mechanisms by which β-catenin regulates the properties of CAFs and to identify β-catenin–regulated genes that are involved in CAF-elicited melanoma BRAFi/MEKi resistance, we isolated wild-type CAFs and β-catenin–deficient CAFs (bcat/CAFs) from our BRAF-mutant melanoma-CAF mouse model (11) for RNA-seq. NFs isolated from C57BL/6J mouse skin were used as a control. The genes that were expressed in CAFs two-fold above or below their respective levels in both NFs and bcat/CAFs were identified. This yielded a list of 1,790 genes (Supplementary Fig. S13A). β-Catenin ablation in CAFs dramatically altered the gene expression pattern, which now resembled that of normal fibroblasts. The data suggest that β-catenin is a major driver of the identity of CAFs, and depleting β-catenin reverts CAFs into a normal fibroblast-like state. Using Venn diagrams (23), we compared genes that were upregulated or downregulated at least 5-fold in either NF or bcat/CAFs in comparison with CAFs. In these 1,054 filtered genes, a 79.9% overlap between significantly changed genes in CAFs (NFs as a control) and bcat/CAFs (CAFs as a control) was observed (Supplementary Fig. S13B). This further confirms that β-catenin-deficient CAFs phenocopied NFs. The genes that were reduced at least 2-fold in bcat/CAFs compared with CAFs underwent a functional annotation clustering analysis using DAVID Bioinformatics Resources 6.8 (Supplementary Fig. S13C; ref. 24). Most of the differentially expressed genes fall into the categories of ECM production, focal adhesion, paracrine signaling, intracellular signaling, transcription regulation, and cytoskeleton remodeling.

Based on this list of genes, we identified POSTN as a top downstream effector of β-catenin that may mediate CAF-induced melanoma drug resistance. To evaluate the significance of POSTN for BRAFi resistance, we performed paired analyses of POSTN expression in melanoma tissues isolated from the same melanoma patients before and after BRAFi/MEKi therapy (Fig. 5A and B). We found that most POSTN-expressing cells were also positive for TE7 expression (12), suggesting that POSTN was mainly produced by fibroblasts. Furthermore, the percentage of POSTN-expressing fibroblasts in TE7+ fibroblasts increased from 36% ± 2% (pretreatment) to 47% ± 2% (posttreatment). In vitro experiments confirmed that POSTN is highly expressed in CAF cell lines, but not in any melanoma cell line (Fig. 5C). In addition, POSTN expression was upregulated by PLX4032 in CAFs, while β-catenin depletion blocked PLX4032-induced POSTN upregulation (Fig. 5D). We also measured the POSTN concentration in culture media conditioned by A375, M27, and M50 using enzyme-linked immunosorbent assay (ELISA). The ELISA data are consistent with the expression data. As shown in Supplementary Fig. S14, A375 secreted little POSTN into the conditioned medium, but both M27 and M50 produced a high amount of POSTN. While PLX4032 further stimulated the production of POSTN in M27 and M50, β-catenin ablation deprived their abilities to produce POSTN.

Figure 5.

POSTN is a direct downstream effector of nuclear β-catenin signaling in CAFs. A, Fibroblasts and POSTN expression in human melanomas pre- and post-BRAFi/MEKi treatment were visualized by coimmunostaining using an anti-TE7 antibody (green) and an anti-POSTN antibody (red). Scale bar, 100 μm. B, The scatter plot represents the percentage of TE7+ cells expressing POSTN in pre- and posttreated melanoma samples. Each data point represents the percentage of TE7+ fibroblasts that are POSTN+ in each 40× field. n ≥ 40. C, Western blotting of POSTN expression levels in indicated melanoma cells and CAFs. Chart shows fluorescence intensities of POSTN bands. n = 3. D, Graph shows qPCR data of POSTN expression in A375, GFP/M50, and bcat-GFP/M50 treated with DMSO or PLX4032. n = 4. E, Illustration of the promoter region of POSTN that contains a potential LEF/TCF–binding site. F, Bar graph shows signals obtained from chromatin immunoprecipitation normalized by signals obtained from the input sample. Axin 2 was used as a positive control. Three different primer sets targeting the TCF/LEF–binding sites in the POSTN promoter region (POSTN1, POSTN2, and POSTN3) and three negative control primers targeting the 3′UTR, exon 6, and the 5′UTR were used. G, Western blotting of POSTN expression in M50 transfected with scramble siRNA or POSTN siRNA. Graph shows fluorescence intensities of POSTN bands. n = 3. H and I, Graph shows A375 numbers in cocultured spheroids of indicated groups under PLX4032 treatment (H) or PLX4032 and GDC0973 treatment (I). n = 3. J and K, Graph shows A375 numbers in cocultured spheroids of indicated groups with or without recombinant POSTN under PLX4032 treatment (J) or PLX4032 and GDC0973 treatment (K). n = 3. For all graphs, the data are shown as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001. ns, not significant.

Figure 5.

POSTN is a direct downstream effector of nuclear β-catenin signaling in CAFs. A, Fibroblasts and POSTN expression in human melanomas pre- and post-BRAFi/MEKi treatment were visualized by coimmunostaining using an anti-TE7 antibody (green) and an anti-POSTN antibody (red). Scale bar, 100 μm. B, The scatter plot represents the percentage of TE7+ cells expressing POSTN in pre- and posttreated melanoma samples. Each data point represents the percentage of TE7+ fibroblasts that are POSTN+ in each 40× field. n ≥ 40. C, Western blotting of POSTN expression levels in indicated melanoma cells and CAFs. Chart shows fluorescence intensities of POSTN bands. n = 3. D, Graph shows qPCR data of POSTN expression in A375, GFP/M50, and bcat-GFP/M50 treated with DMSO or PLX4032. n = 4. E, Illustration of the promoter region of POSTN that contains a potential LEF/TCF–binding site. F, Bar graph shows signals obtained from chromatin immunoprecipitation normalized by signals obtained from the input sample. Axin 2 was used as a positive control. Three different primer sets targeting the TCF/LEF–binding sites in the POSTN promoter region (POSTN1, POSTN2, and POSTN3) and three negative control primers targeting the 3′UTR, exon 6, and the 5′UTR were used. G, Western blotting of POSTN expression in M50 transfected with scramble siRNA or POSTN siRNA. Graph shows fluorescence intensities of POSTN bands. n = 3. H and I, Graph shows A375 numbers in cocultured spheroids of indicated groups under PLX4032 treatment (H) or PLX4032 and GDC0973 treatment (I). n = 3. J and K, Graph shows A375 numbers in cocultured spheroids of indicated groups with or without recombinant POSTN under PLX4032 treatment (J) or PLX4032 and GDC0973 treatment (K). n = 3. For all graphs, the data are shown as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001. ns, not significant.

Close modal

POSTN is a direct β-catenin target in CAFs

We identified an evolutionarily conserved T-cell factor/lymphoid enhancer-binding factor (TCF/LEF)-binding site in the proximal promoter region of the POSTN gene (chr13:37599016–37599023, human reference sequence GRCh37, Supplementary Fig. S15) using PROMO (25, 26). Guided by this observation, chromatin immunoprecipitation was performed to investigate whether POSTN is a direct β-catenin target in CAFs (Fig. 5E). As shown in Fig. 5F, significant binding of β-catenin was found in the POSTN promoter region that contains the TCF/LEF consensus sequence (Supplementary File S4). We then used siRNAs to silence POSTN expression in M50 CAFs, and POSTN siRNA#2 showed the highest depletion efficiency (Supplementary Fig. S16A). The ablation of POSTN in M50 by all three siRNAs did not show any inhibitory effects on M50 survival (Supplementary Fig. S16B). Therefore, siRNA#2 was used to silence POSTN expression in M27 and M50. Drug resistance assays using PLX4032 or a combination of PLX4032 + GDC0973 showed that fewer melanoma cells (A375 and SK-MEL-24) survived treatment in the spheroids cocultured with POSTN-deficient M50 and M27 cells compared with control CAFs (Fig. 5H and I; Supplementary Fig. S16C–S16F), indicating that POSTN is required for CAF-mediated drug resistance to BRAFi/MEKi.

To determine whether POSTN can compensate for the loss of β-catenin in CAFs, we added recombinant POSTN to the spheroids formed by melanoma cells and β-catenin–deficient CAFs, including bcat-GFP/M50 and bcat-GFP/M27, under PLX4032 or combined PLX4032 + GDC0973 treatment (Fig. 5J and K; Supplementary Fig. S17). The data show that recombinant POSTN, at least partially, rescues the loss of β-catenin in CAFs to promote melanoma resistance to BRAFi and MEKi.

POSTN reactivates ERK signaling in BRAF-mutant melanoma cells to bypass BRAFi and MEKi inhibition

A375 and SK-MEL-24 cells grew better and had increased resistance to PLX4032 when cultured on plates coated with recombinant POSTN (Fig. 6AC; Supplementary Fig. S18A). We confirmed these findings using melanoma spheroids (Fig. 6DF; Supplementary Fig. S18B). Recombinant POSTN induced an increase in spheroid size. Furthermore, recombinant POSTN also induced A375 resistance to MEKi. As shown in Fig. 6G, A375 became less sensitive to GDC0973 and to the combination of PLX4032 + GDC0973 when POSTN was added.

Figure 6.

POSTN activates PI3K/AKT signaling in BRAF-mutant melanoma cells to bypass BRAFi and MEKi inhibition. A, Fluorescent images show RFP-tagged A375 cells in collagen-coated dishes and collagen + POSTN–coated dishes with or without PLX4032 treatment. B and C, Graph shows relative A375 numbers in regular and POSTN-coated dishes under DMSO (B) or PLX4032 (C) treatment. Each data point represents the relative cell number in each 4× field normalized by the mean A375 number in No POSTN group. n = 10. D, Representative images of RFP-tagged A375 spheroids cultured with and without POSTN under DMSO or PLX4032 treatment. Scale bar, 200 μm. E and F, Graph shows the sizes of the spheroids cultured with or without POSTN under DMSO (E) or PLX4032 treatment (F). Each data point represents the size of one spheroid from indicated groups. n = 24. G, Graph shows the numbers of A375 cultured with and without POSTN when treated with PLX4032, GDC0973, or a combination of PLX4032 + GDC0973. n = 3. H, Western blotting of ERK and pERK levels in A375 treated with and without POSTN. Chart shows the fluorescent intensities of ERK and pERK bands. n = 3. I, Western blotting of AKT and pAKT levels in A375 treated with and without POSTN. Chart shows fluorescent intensities of AKT and pAKT bands. n = 3. J, Western blotting of AKT, pAKT, ERK, and pERK levels in A375 treated with PLX4032, POSTN, and LY294002 as indicated. K and L, Charts show fluorescence intensities of pERK bands (K) and pAKT bands (L) under indicated conditions. n = 3. For all graphs, the data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Figure 6.

POSTN activates PI3K/AKT signaling in BRAF-mutant melanoma cells to bypass BRAFi and MEKi inhibition. A, Fluorescent images show RFP-tagged A375 cells in collagen-coated dishes and collagen + POSTN–coated dishes with or without PLX4032 treatment. B and C, Graph shows relative A375 numbers in regular and POSTN-coated dishes under DMSO (B) or PLX4032 (C) treatment. Each data point represents the relative cell number in each 4× field normalized by the mean A375 number in No POSTN group. n = 10. D, Representative images of RFP-tagged A375 spheroids cultured with and without POSTN under DMSO or PLX4032 treatment. Scale bar, 200 μm. E and F, Graph shows the sizes of the spheroids cultured with or without POSTN under DMSO (E) or PLX4032 treatment (F). Each data point represents the size of one spheroid from indicated groups. n = 24. G, Graph shows the numbers of A375 cultured with and without POSTN when treated with PLX4032, GDC0973, or a combination of PLX4032 + GDC0973. n = 3. H, Western blotting of ERK and pERK levels in A375 treated with and without POSTN. Chart shows the fluorescent intensities of ERK and pERK bands. n = 3. I, Western blotting of AKT and pAKT levels in A375 treated with and without POSTN. Chart shows fluorescent intensities of AKT and pAKT bands. n = 3. J, Western blotting of AKT, pAKT, ERK, and pERK levels in A375 treated with PLX4032, POSTN, and LY294002 as indicated. K and L, Charts show fluorescence intensities of pERK bands (K) and pAKT bands (L) under indicated conditions. n = 3. For all graphs, the data are represented as mean ± SEM. *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ns, not significant.

Close modal

POSTN is an ECM protein that establishes contact and communication between the ECM and the cell by binding to integrin receptors (27). POSTN activates the integrin/AKT/MAPK pathway in cancer cells to accelerate cell proliferation (28). Given that the addition of POSTN promoted melanoma cell proliferation and drug resistance, we investigated the phosphatidylinositol 3 kinase (PI3K)/AKT pathway and the ERK pathway in A375 cells treated with POSTN. A roughly 2-fold increase was observed in the expression of pERK and pAKT after POSTN treatment (Fig. 6H and I). To confirm whether POSTN induces the reactivation of ERK signaling in BRAFi-treated A375 and whether this alternative pathway is PI3K/AKT-dependent, we assessed the activation of the AKT and ERK pathways in A375 and SK-MEL-24 treated with POSTN and PLX4032 (Fig. 6JL; Supplementary Fig. S18C–S18E). In A375 and SK-MEL-24 treated with PLX4032, pERK was significantly inhibited. The addition of POSTN increased the level of pERK that was inhibited by PLX4032 and also led to a significant increase in pAKT expression in PLX4032-treated A375 and SK-MEL-24 cells. LY294002 is a highly selective PI3K inhibitor (29) and inhibits AKT phosphorylation by PI3K in A375. LY294002 blocked POSTN-induced reactivation of ERK signaling in melanoma cells under PLX4032 treatment. The data show that POSTN reactivates ERK signaling in melanoma cells, which is inhibited by BRAFi and MEKi.

Melanomas harbor a significant number of stromal cell populations. However, for the longest time, the main thrust in cancer therapy has been to exploit “weaknesses” of cancer cells themselves, such as oncogene addiction, high proliferation rates, or specific mutations in oncogenes and tumor suppressor genes. Such efforts have resulted in the development of BRAFi. Nevertheless, the development of drug resistance after the initial positive response is the biggest hurdle to achieving the ultimate success of BRAF-targeted cancer therapies. The data reported here highlight that the occurrence of drug resistance is not a simple response by merely melanoma cells and can substantially be affected by complex interactions among tumor cells, stromal cells, the ECM, and targeted therapy drugs.

Our study demonstrates that CAFs play important roles in supporting melanoma cell survival and proliferation under targeted therapy. CAFs are not a static population of fibroblasts and possess significant heterogeneity in characteristics and functions, which are in fact a direct reflection of their interactions with the surrounding environment (8). Six CAF cell lines used in our study showed clear heterogeneity, displaying distinct levels of αSMA, F-actin, paxillin, and p-MLC2 expression. In addition, these CAF cell lines possess strong but variable abilities to migrate and contract collagen gels. We discovered that BRAFi led to an expanded αSMA+ CAF population with increased nuclear β-catenin accumulation. These αSMA+/nuclear β-catenin+ CAFs have enhanced biological properties, especially in their ability to remodel the cytoskeleton and the ECM. The data suggest that CAFs in BRAFi/MEKi–treated melanoma develop hyperactivated ECM remodeling capabilities. The findings reveal that BRAFi can elicit adaptive transcriptional and cellular activities in CAFs and equip them with specific phenotypes, which allow them to better support melanoma cells under the pressure of therapeutic agents.

Conventionally, nuclear β-catenin accumulation is associated with canonical Wnt signaling (30). Without Wnt ligands, cytoplasmic β-catenin is phosphorylated at its N-terminus by GSK3β in a destruction complex for degradation via the ubiquitin/proteasome pathway. When Wnt ligands activate the Wnt pathway, GSK3β is phosphorylated, and β-catenin is stabilized and translocated into the nucleus. In our study, we found no evidence of increased production of Wnt ligands by either melanoma cells or stromal cell populations. BRAFi, such as PLX4032, selectively interferes with the kinase domain of BRAF molecules carrying a V600E mutation and inhibits its activity (31). However, such ATP-competitive BRAFi has paradoxical side effects on wild-type RAF molecules (32). BRAFi can induce kinase domain dimerization among the wild-type members of the RAF family, for example, BRAF/CRAF dimers (20, 21), which is a critical event in inducing RAF activation (33). So far, most of the observed “paradoxical” side effects of BRAFi have been reported to occur in tumor cells (34, 35). The fact that we observed CRAF/BRAF dimerization in CAFs in response to BRAFi provides evidence that BRAFi can elicit paradoxical effects in stromal cell types carrying wildtype BRAF as well. Intriguingly, BRAFi-induced BRAF/CRAF dimerization is the driving force that leads to the nuclear translocation of β-catenin in CAFs. By eliminating CRAF and BRAF in CAFs, nuclear β-catenin translocation can no longer occur upon BRAFi treatment.

We discovered that this translocation of β-catenin is mainly mediated by the MEK-ERK signaling pathway, which is activated by BRAF/CRAF dimerization. Using MEKi and ERKi in combination with BRAFi significantly inhibited BRAFi-induced nuclear β-catenin translocation in CAFs. However, we cannot exclude the existence of other pathways downstream of BRAF/CRAF dimerization that contribute to the activation of MEK-ERK signaling. As such, further research on this matter is undoubtedly needed.

The findings in this report could reasonably explain why a CAF population with increased nuclear β-catenin is expanded in melanoma after BRAFi/MEKi treatment. Nuclear β-catenin binds the members of the TCF/LEF family and activates target gene expression (36). The importance of the resulting β-catenin transcriptional activities has been demonstrated in fibroblast activation, fibrosis, and tissue repair (37). As such, BRAFi-induced β-catenin nuclear activation in CAFs has a plethora of consequences beyond the impact of the canonical function of RAS–RAF–MEK–ERK activation and drives a program in CAFs that enhances (i) their biological functions; (ii) their ability to contract collagen and stiffen the ECM; and (iii) their ability to protect melanoma cells from the toxic effects of BRAFi. Our transcriptome analysis supports this finding and shows that a loss of β-catenin completely reverses the CAF transcriptome into a transcriptome that resembles that of normal fibroblasts. This astonishing transcriptomic reversal of CAFs into normal fibroblasts reveals a previously unappreciated but pivotal role of β-catenin in CAF functions. The changes are of such an extent that β-catenin can be regarded as a master regulator of the CAF phenotype that is modulated by BRAFi. In addition, there is a possibility that β-catenin could eventually be a biomarker for BRAFi-stimulated CAFs, although more clinical validation and tests are required.

Our analysis has focused on a top β-catenin target gene in CAFs, POSTN, which is known as a protein that is linked to human malignant melanoma (28, 38–40). POSTN may have multiple effects on melanoma cells, directly interacting with integrin receptors on the cell membrane or indirectly remodeling the ECM structure. In our study, we discovered that POSTN directly induces an alternative PI3K/Akt pathway in melanoma cells and leads to the activation of ERK signaling, allowing melanoma cells to continue to proliferate despite the dual inhibition by BRAFi and MEKi (Fig. 7). The data indicate that POSTN could be an interesting target to disrupt the tumor microenvironment and improve targeted therapy or even immunotherapy. In fact, multiple approaches have been explored to suppress POSTN functions (41), including targeting POSTN mRNA with siRNAs (42–44), and targeting POSTN protein with anti-POSTN antibodies (45, 46) or DNA aptamers (47). However, a combination of POSTN targeting and currently available targeted therapies or immunotherapies has not been tested and may be worth further investigation.

Figure 7.

Model of BRAFi-induced CAF reprogramming in matrix remodeling and melanoma BRAFi/MEKi resistance. BRAFi blocks BRAF/MEK/ERK signaling in BRAFV600E-mutant melanoma cells, leading to cell death. On the other hand, BRAFi stimulates CAFs and elicits the formation of BRAF/CRAF dimers and nuclear β-catenin accumulation. Nuclear β-catenin signaling reprograms CAFs and leads to the production of matricellular protein POSTN. POSTN activates alterative PI3K/AKT signaling in BRAF-mutant melanoma cells and subsequently ERK activation, which allows melanoma cells to bypass BRAFi/MEKi.

Figure 7.

Model of BRAFi-induced CAF reprogramming in matrix remodeling and melanoma BRAFi/MEKi resistance. BRAFi blocks BRAF/MEK/ERK signaling in BRAFV600E-mutant melanoma cells, leading to cell death. On the other hand, BRAFi stimulates CAFs and elicits the formation of BRAF/CRAF dimers and nuclear β-catenin accumulation. Nuclear β-catenin signaling reprograms CAFs and leads to the production of matricellular protein POSTN. POSTN activates alterative PI3K/AKT signaling in BRAF-mutant melanoma cells and subsequently ERK activation, which allows melanoma cells to bypass BRAFi/MEKi.

Close modal

Despite the frequent development of acquired resistance, BRAFi/MEKi combination therapy represents the standard of care for patients carrying metastatic BRAF-mutated melanoma (48). Theoretically, BRAFi/MEKi therapy has the edge over BRAFi single-agent therapy because it can effectively block the activation of CAFs, thereby inhibiting the microenvironment-initiated drug resistance mechanism. Nevertheless, drug penetration and distribution within solid tumors present a long-standing hurdle for effective cancer therapies. It is likely that not all fibroblasts in the stroma are exposed in the same way to BRAFi and MEKi due to drug penetration issues. As such, some CAFs may only experience BRAFi treatment and initiate BRAF/CRAF dimerization and MEK/ERK/nuclear β-catenin activation, thereby contributing to BRAF-mutant melanoma resistance to BRAFi/MEKi. Consequently, in BRAFi/MEKi–treated human melanoma samples, we observe an increase of CAFs with nuclear β-catenin compared with the naïve tumor.

In conclusion, we present novel insights into the biology of CAFs, their adaptive response to targeted therapy drugs, and how CAFs and CAF-derived signals may be targeted to improve current therapies. We demonstrated that β-catenin is a master regulator of the BRAFi-stimulated CAF phenotype. Identification of its downstream effectors offers a novel angle to attack an important cancer-supporting cell type on which tumor cells rely. Taken together, the data reveal a signaling loop among BRAFi, CAFs, the matrix, and tumor cells that drives the development of acquired resistance in melanoma.

Y. Zhang reports grants from the Cincinnati Cancer Center, grants from the Harry J Lloyd Trust, and grants from The Center for Clinical and Translational Science and Training (CCTST) during the conduct of the study. No disclosures were reported by the other authors.

T. Liu: Conceptualization, data curation, validation, investigation, writing–original draft. L. Zhou: Data curation, formal analysis, validation, investigation. Y. Xiao: Formal analysis, validation. T. Andl: Data curation, writing–original draft, writing–review and editing. Y. Zhang: Conceptualization, resources, data curation, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.

This work was supported by grants from the Cincinnati Cancer Center (YZ), the Harry J Lloyd Trust (YZ), and the CCTST Pilot Translational Research & Innovative Core (YZ).

We thank the University of Cincinnati Biorepository and the Melanoma Biorepository at the University of Colorado Anschutz Medical Campus for providing human melanoma tissue sections. We thank Dr. Jie Liu at the Cincinnati Children's Hospital Medical Center Genetics Laboratory for analyzing CAF sequencing data. We thank Drs. Kentaro Iwasawa and Takanori Takebe at the Cincinnati Children's Hospital Medical Center for helping with AFM. We thank Ryan Dickerson from the Educational Technology Division at the University of Central Florida for assisting with illustration in Fig 7. We also thank Ashlee Harris for carefully reading and editing the manuscript.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Pietras
K
,
Ostman
A
. 
Hallmarks of cancer: interactions with the tumor stroma
.
Exp Cell Res
2010
;
316
:
1324
31
.
2.
Johansson
A
,
Ganss
R
. 
Remodeling of tumor stroma and response to therapy
.
Cancers (Basel)
2012
;
4
:
340
53
.
3.
Fedorenko
IV
,
Wargo
JA
,
Flaherty
KT
,
Messina
JL
,
Smalley
KSM
. 
BRAF inhibition generates a host-tumor niche that mediates therapeutic escape
.
J Invest Dermatol
2015
;
135
:
3115
24
.
4.
Reddy
SM
,
Reuben
A
,
Wargo
JA
. 
Influences of BRAF inhibitors on the immune microenvironment and the rationale for combined molecular and immune targeted therapy
.
Curr Oncol Rep
2016
;
18
:
42
.
5.
Long
JE
,
Wongchenko
MJ
,
Nickles
D
,
Chung
WJ
,
Wang
BE
,
Riegler
J
, et al
Therapeutic resistance and susceptibility is shaped by cooperative multi-compartment tumor adaptation
.
Cell Death Differ
2019
;
26
:
2416
29
.
6.
Erdogan
B
,
Webb
DJ
. 
Cancer-associated fibroblasts modulate growth factor signaling and extracellular matrix remodeling to regulate tumor metastasis
.
Biochem Soc Trans
2017
;
45
:
229
36
.
7.
Nurmik
M
,
Ullmann
P
,
Rodriguez
F
,
Haan
S
,
Letellier
E
. 
In search of definitions: Cancer-associated fibroblasts and their markers
.
Int J Cancer
2020
;
146
:
895
905
.
8.
Sahai
E
,
Astsaturov
I
,
Cukierman
E
,
DeNardo
DG
,
Egeblad
M
,
Evans
RM
, et al
A framework for advancing our understanding of cancer-associated fibroblasts
.
Nat Rev Cancer
2020
;
20
:
174
86
.
9.
Ohlund
D
,
Handly-Santana
A
,
Biffi
G
,
Elyada
E
,
Almeida
AS
,
Ponz-Sarvise
M
, et al
Distinct populations of inflammatory fibroblasts and myofibroblasts in pancreatic cancer
.
J Exp Med
2017
;
214
:
579
96
.
10.
Liu
T
,
Zhou
L
,
Yang
K
,
Iwasawa
K
,
Kadekaro
AL
,
Takebe
T
, et al
The beta-catenin/YAP signaling axis is a key regulator of melanoma-associated fibroblasts
.
Signal Transduct Target Ther
2019
;
4
:
63
.
11.
Zhou
L
,
Yang
K
,
Dunaway
S
,
Abdel-Malek
Z
,
Andl
T
,
Kadekaro
AL
, et al
Suppression of MAPK signaling in BRAF-activated PTEN-deficient melanoma by blocking beta-catenin signaling in cancer-associated fibroblasts
.
Pigment Cell Melanoma Res
2018
;
31
:
297
307
.
12.
Goodpaster
T
,
Legesse-Miller
A
,
Hameed
MR
,
Aisner
SC
,
Randolph-Habecker
J
,
Coller
HA
. 
An immunohistochemical method for identifying fibroblasts in formalin-fixed, paraffin-embedded tissue
.
J Histochem Cytochem
2008
;
56
:
347
58
.
13.
Choi
SY
,
Sung
R
,
Lee
SJ
,
Lee
TG
,
Kim
N
,
Yoon
SM
, et al
Podoplanin, alpha-smooth muscle actin or S100A4 expressing cancer-associated fibroblasts are associated with different prognosis in colorectal cancers
.
J Korean Med Sci
2013
;
28
:
1293
301
.
14.
Murrell
MP
,
Gardel
ML
. 
F-actin buckling coordinates contractility and severing in a biomimetic actomyosin cortex
.
Proc Natl Acad Sci U S A
2012
;
109
:
20820
5
.
15.
Cheng
Y
,
Zhang
G
,
Li
G
. 
Targeting MAPK pathway in melanoma therapy
.
Cancer Metastasis Rev
2013
;
32
:
567
84
.
16.
Clevers
H
,
Nusse
R
. 
Wnt/beta-catenin signaling and disease
.
Cell
2012
;
149
:
1192
205
.
17.
Heidorn
SJ
,
Milagre
C
,
Whittaker
S
,
Nourry
A
,
Niculescu-Duvas
I
,
Dhomen
N
, et al
Kinase-dead BRAF and oncogenic RAS cooperate to drive tumor progression through CRAF
.
Cell
2010
;
140
:
209
21
.
18.
Dumaz
N
,
Hayward
R
,
Martin
J
,
Ogilvie
L
,
Hedley
D
,
Curtin
JA
, et al
In melanoma, RAS mutations are accompanied by switching signaling from BRAF to CRAF and disrupted cyclic AMP signaling
.
Cancer Res
2006
;
66
:
9483
91
.
19.
Montagut
C
,
Sharma
SV
,
Shioda
T
,
McDermott
U
,
Ulman
M
,
Ulkus
LE
, et al
Elevated CRAF as a potential mechanism of acquired resistance to BRAF inhibition in melanoma
.
Cancer Res
2008
;
68
:
4853
61
.
20.
Poulikakos
PI
,
Zhang
C
,
Bollag
G
,
Shokat
KM
,
Rosen
N
. 
RAF inhibitors transactivate RAF dimers and ERK signalling in cells with wild-type BRAF
.
Nature
2010
;
464
:
427
30
.
21.
Hatzivassiliou
G
,
Song
K
,
Yen
I
,
Brandhuber
BJ
,
Anderson
DJ
,
Alvarado
R
, et al
RAF inhibitors prime wild-type RAF to activate the MAPK pathway and enhance growth
.
Nature
2010
;
464
:
431
5
.
22.
Wong
DJ
,
Robert
L
,
Atefi
MS
,
Lassen
A
,
Avarappatt
G
,
Cerniglia
M
, et al
Antitumor activity of the ERK inhibitor SCH772984 [corrected] against BRAF mutant, NRAS mutant and wild-type melanoma
.
Mol Cancer
2014
;
13
:
194
.
23.
Hulsen
T
,
de Vlieg
J
,
Alkema
W
. 
BioVenn - a web application for the comparison and visualization of biological lists using area-proportional Venn diagrams
.
BMC Genomics
2008
;
9
:
488
.
24.
Dennis
G
 Jr
,
Sherman
BT
,
Hosack
DA
,
Yang
J
,
Gao
W
,
Lane
HC
, et al
DAVID: Database for Annotation, Visualization, and Integrated Discovery
.
Genome Biol
2003
;
4
:
P3
.
25.
Farre
D
,
Roset
R
,
Huerta
M
,
Adsuara
JE
,
Rosello
L
,
Alba
MM
, et al
Identification of patterns in biological sequences at the ALGGEN server: PROMO and MALGEN
.
Nucleic Acids Res
2003
;
31
:
3651
3
.
26.
Messeguer
X
,
Escudero
R
,
Farre
D
,
Nunez
O
,
Martinez
J
,
Alba
MM
. 
PROMO: detection of known transcription regulatory elements using species-tailored searches
.
Bioinformatics
2002
;
18
:
333
4
.
27.
Kii
I
. 
Periostin Functions as a Scaffold for Assembly of Extracellular Proteins
.
Adv Exp Med Biol
2019
;
1132
:
23
32
.
28.
Kotobuki
Y
,
Yang
L
,
Serada
S
,
Tanemura
A
,
Yang
F
,
Nomura
S
, et al
Periostin accelerates human malignant melanoma progression by modifying the melanoma microenvironment
.
Pigment Cell Melanoma Res
2014
;
27
:
630
9
.
29.
Vlahos
CJ
,
Matter
WF
,
Hui
KY
,
Brown
RF
. 
A specific inhibitor of phosphatidylinositol 3-kinase, 2-(4-morpholinyl)-8-phenyl-4H-1-benzopyran-4-one (LY294002)
.
J Biol Chem
1994
;
269
:
5241
8
.
30.
Miller
JR
,
Moon
RT
. 
Signal transduction through beta-catenin and specification of cell fate during embryogenesis
.
Genes Dev
1996
;
10
:
2527
39
.
31.
Garbe
C
,
Abusaif
S
,
Eigentler
TK
.
Vemurafenib. In: Martens UM, editor
.
Small Molecules in Oncology
:
Springer Berlin Heidelberg
; 
2014
.
p.
215
25
.
32.
Gibney
GT
,
Messina
JL
,
Fedorenko
IV
,
Sondak
VK
,
Smalley
KS
. 
Paradoxical oncogenesis–the long-term effects of BRAF inhibition in melanoma
.
Nat Rev Clin Oncol
2013
;
10
:
390
9
.
33.
Rajakulendran
T
,
Sahmi
M
,
Lefrancois
M
,
Sicheri
F
,
Therrien
M
. 
A dimerization-dependent mechanism drives RAF catalytic activation
.
Nature
2009
;
461
:
542
5
.
34.
Holderfield
M
,
Nagel
TE
,
Stuart
DD
. 
Mechanism and consequences of RAF kinase activation by small-molecule inhibitors
.
Br J Cancer
2014
;
111
:
640
5
.
35.
Belum
VR
,
Fischer
A
,
Choi
JN
,
Lacouture
ME
. 
Dermatological adverse events from BRAF inhibitors: a growing problem
.
Curr Oncol Rep
2013
;
15
:
249
59
.
36.
Cadigan
KM
,
Waterman
ML
. 
TCF/LEFs and Wnt signaling in the nucleus
.
Cold Spring Harb Perspect Biol
2012
;
4
:
a007906
.
37.
Valenta
T
,
Hausmann
G
,
Basler
K
. 
The many faces and functions of beta-catenin
.
EMBO J
2012
;
31
:
2714
36
.
38.
Tilman
G
,
Mattiussi
M
,
Brasseur
F
,
van Baren
N
,
Decottignies
A
. 
Human periostin gene expression in normal tissues, tumors and melanoma: evidences for periostin production by both stromal and melanoma cells
.
Mol Cancer
2007
;
6
:
80
.
39.
Fukuda
K
,
Sugihara
E
,
Ohta
S
,
Izuhara
K
,
Funakoshi
T
,
Amagai
M
, et al
Periostin is a key niche component for wound metastasis of melanoma
.
PLoS One
2015
;
10
:
e0129704
.
40.
Ohno
F
,
Nakahara
T
,
Kido-Nakahara
M
,
Ito
T
,
Nunomura
S
,
Izuhara
K
, et al
Periostin links skin inflammation to melanoma progression in humans and mice
.
Int J Mol Sci
2019
;
20
:
169
.
41.
Kudo
A
. 
Clinical applications targeting periostin
.
Adv Exp Med Biol
2019
;
1132
:
207
10
.
42.
Tomaru
A
,
Kobayashi
T
,
Hinneh
JA
,
Tonto
PB
,
D'Alessandro-Gabazza
CN
,
Fujimoto
H
, et al
Oligonucleotide-targeting periostin ameliorates pulmonary fibrosis
.
Gene Ther
2017
;
24
:
706
16
.
43.
Ye
D
,
Zhou
C
,
Wang
S
,
Deng
H
,
Shen
Z
. 
Tumor suppression effect of targeting periostin with siRNA in a nude mouse model of human laryngeal squamous cell carcinoma
.
J Clin Lab Anal
2019
;
33
:
e22622
.
44.
Nakama
T
,
Yoshida
S
,
Ishikawa
K
,
Kubo
Y
,
Kobayashi
Y
,
Zhou
Y
, et al
Therapeutic Effect of Novel Single-Stranded RNAi Agent Targeting Periostin in Eyes with Retinal Neovascularization
.
Mol Ther Nucleic Acids
2017
;
6
:
279
89
.
45.
Kyutoku
M
,
Taniyama
Y
,
Katsuragi
N
,
Shimizu
H
,
Kunugiza
Y
,
Iekushi
K
, et al
Role of periostin in cancer progression and metastasis: inhibition of breast cancer progression and metastasis by anti-periostin antibody in a murine model
.
Int J Mol Med
2011
;
28
:
181
6
.
46.
Zhu
M
,
Saxton
RE
,
Ramos
L
,
Chang
DD
,
Karlan
BY
,
Gasson
JC
, et al
Neutralizing monoclonal antibody to periostin inhibits ovarian tumor growth and metastasis
.
Mol Cancer Ther
2011
;
10
:
1500
8
.
47.
Lee
YJ
,
Kim
IS
,
Park
SA
,
Kim
Y
,
Lee
JE
,
Noh
DY
, et al
Periostin-binding DNA aptamer inhibits breast cancer growth and metastasis
.
Mol Ther
2013
;
21
:
1004
13
.
48.
Stagno
A
,
Vari
S
,
Annovazzi
A
,
Anelli
V
,
Russillo
M
,
Cognetti
F
, et al
Case 
Report: Rechallenge With BRAF and MEK Inhibitors in Metastatic Melanoma: A Further Therapeutic Option in Salvage Setting?
Front Oncol
2021
;
11
:
993
.