Ovarian cancer ascites fluid contains matrix proteins that can impact tumor growth via integrin receptor binding. In human ovarian tumor tissue arrays, we find that activation of the cytoplasmic focal adhesion (FAK) tyrosine kinase parallels increased tumor stage, β5 integrin, and osteopontin matrix staining. Elevated osteopontin, β5 integrin, and FAK mRNA levels are associated with decreased serous ovarian cancer patient survival. FAK remains active within ovarian cancer cells grown as spheroids, and anchorage-independent growth analyses of seven ovarian carcinoma cell lines identified sensitive (HEY, OVCAR8) and resistant (SKOV3-IP, OVCAR10) cells to 0.1 μmol/L FAK inhibitor (VS-4718, formerly PND-1186) treatment. VS-4718 promoted HEY and OVCAR8 G0–G1 cell-cycle arrest followed by cell death, whereas growth of SKOV3-IP and OVCAR10 cells was resistant to 1.0 μmol/L VS-4718. In HEY cells, genetic or pharmacological FAK inhibition prevented tumor growth in mice with corresponding reductions in β5 integrin and osteopontin expression. β5 knockdown reduced HEY cell growth in soft agar, tumor growth in mice, and both FAK Y397 phosphorylation and osteopontin expression in spheroids. FAK inhibitor–resistant (SKOV3-IP, OVCAR10) cells exhibited anchorage-independent Akt S473 phosphorylation, and expression of membrane-targeted and active Akt in sensitive cells (HEY, OVCAR8) increased growth but did not create a FAK inhibitor–resistant phenotype. These results link osteopontin, β5 integrin, and FAK in promoting ovarian tumor progression. β5 integrin expression may serve as a biomarker for serous ovarian carcinoma cells that possess active FAK signaling. Mol Cancer Ther; 13(8); 2050–61. ©2014 AACR.
This article is featured in Highlights of This Issue, p. 1977
Ovarian cancer is the fifth leading cause of cancer-related death in women in the United States (1). Initial tumor spread is by an exfoliative mechanism whereby cells dissociate from a primary site and can proliferate in an anchorage-independent manner as clumps of aggregated cells termed spheroids within the peritoneal space (2). Anchorage-independent growth is a hallmark of cell transformation and is connected to elevated tumorigenic potential (3).
In addition to being a sign of advanced disease, ascites contains growth factors and soluble matrix proteins that can enhance ovarian spheroid growth (4). Matrix proteins such as fibronectin, vitronectin, and osteopontin are ligands for integrin receptors and are present in high levels within ascites (5). Osteopontin is also a potential diagnostic blood biomarker for ovarian cancer (6, 7). Matrix proteins can become integrated within tumor spheroids to provide a structural scaffold as well as promote signals regulating tumor growth and survival (8, 9). Transmembrane integrin receptors bind matrix proteins and integrin α5β1 binding to fibronectin is linked to ovarian tumor metastasis in mouse models (10). However, clinical trials of an anti-α5β1 antibody did not show activity as a single agent in patients with platinum-resistant ovarian cancer (11). This may be due to signals from multiple β-integrin receptors for various matrix proteins that may require co-inhibition to prevent refractory ovarian tumor growth in vivo.
Integrin β integrin subunits activate a common set of cytoplasmic tyrosine kinases, and targeting this proximal linkage may be an effective means to block signals from multiple integrin receptors (12). The cytoplasmic focal adhesion (FAK) tyrosine kinase is recruited and activated by β1, β3, and β5 integrin subunits. These β integrins can pair with the αv integrin subunit, and together, signals are generated that modulate tumor survival and growth (13). FAK gene amplification occurs in about 24% of serous ovarian cancer, and elevated FAK mRNA levels are associated with decreased overall patient survival (12). Although canonically known as a cell adhesion–activated kinase, FAK inhibition does not prevent the proliferation of tumor cells normally cultured on plastic (14, 15). However, increased tumor apoptosis occurs upon pharmacological FAK inhibition in mouse xenograft tumor models (16–18), and submicromolar concentrations can trigger apoptosis of tumor cell lines when cultured under anchorage-independent conditions (12, 19). Completed phase I trials of PF-00562271 FAK inhibitor revealed a subset of patients with stable disease (20), but molecular mechanisms driving tumor cell sensitivity or resistance to FAK inhibitors remain incomplete. Here, we show that FAK, β5 integrin, and osteopontin comprise a signaling axis promoting serous ovarian carcinoma tumor growth.
Materials and Methods
Antibodies and reagents
PND-1186 (ref. 21; renamed VS-4718 by Verastem Inc.) was from Poniard Inc. and PF-271 was synthesized as described (17). Compounds were dissolved in DMSO. Supplementary Table S1 contains antibody, plasmid, and probe sets used in this study. Additional materials and methods, including details of cell cycle, apoptosis, and real-time quantitative PCR analyses, are described in Supplementary Materials and Methods.
Supplementary Table S2 lists source, culture conditions, and selective DNA sequencing information for the cells used. Human ovarian cancer cell lines IGROV1, SKOV3, and SKOV3-IP were from J. Chien in 2008 (Mayo Clinic, Rochester, MN). OVCAR3, OVCAR8, and OVCAR10 cells were from D. Connolly in 2011 (cells generated at Fox Chase) and HEY cells were from S. Howell in 2011 (UCSD). BT474 cells were from ATCC (2008) and maintained in low passages (<3 months). For other cells, no authentication was performed by the authors. ID8-IP, IGROV1-IP, and SKOV3-IP cells were generated by intraperitoneal injection into nude mice in 2012 as described (12, 22). IGROV1, IGROV1-IP, SKOV3, SKVO3-IP, and HEY cells were cultured in DMEM; OVCAR3, OVCAR8, OVCAR10, and BT474 cells were culture in RPMI. All cell media were supplemented with 10% FBS, 0.1 nmol/L nonessential amino acids, 2 mmol/L glutamine, 100 U/mL penicillin, and 100 μg/mL streptomycin. Cell lines were propagated adherently on plastic and replated on low-binding poly 2-hydroxyethyl methacrylate (poly-HEMA, Corning)-coated plates for experimental anchorage-independent analyses.
DNA and retroviral constructs
shRNA targeting human FAK and a scrambled (Scr) control in pLentiLox 3.7-Puro were created as described (23). Lentiviral-transduced cells were selected by growth in puromycin; clones were isolated by single-cell sorting and characterized by anti-FAK immunoblotting. Three clones were pooled, expanded, and stored frozen as Scr- or FAK shRNA-expressing HEY cells. GFP-tagged FAK wild-type (WT) and FAK kinase-dead (KD) from the murine cDNA were cloned into the lentiviral vector pCDH1-MCS1-EF1-Puro (System Biosciences), selected for growth in puromycin, sorted via flow cytometry for GFP expression, and used as a pooled population. HEY cells were transduced with lentiviral shRNAs targeting human β5 integrin or Scr shRNA (Mission, Sigma). HEY and OVCAR8 cells were stably transduced with a myristoylated and membrane-targeted form of Akt (Addgene) via retrovirus produced by 293 cell transfection (23).
Cells were plated under adherent (0.5 × 104 cells, tissue culture–treated) and nonadherent conditions (25 × 104 cells, poly-HEMA–coated) in 6-well plates in 2 mL growth media. After 72 hours, all cells were collected by limited trypsin–EDTA treatment, a single-cell suspension was prepared, and the viable (trypan blue–negative) total cell number determined by ViCell XR counting (Beckman). For soft agar assays, 0.2 × 104 cells per well were plated in 0.3% agar in 0.2 mL growth media as described (12). After 7 days, colonies were stained with crystal violet, imaged in phase contrast, and enumerated. All experimental points were performed in triplicate and repeated at least 2 times.
For surface integrin expression, cells were trypsinized and incubated with primary antibodies to integrins (106 cells/μg antibody) for 60 minutes on ice and washed in cold PBS. Allophycocyanin (APC)-conjugated goat anti-mouse IgG was used as secondary antibody, and flow cytometry analyses (FACS Calibur) performed using FlowJo software. Mouse IgG was the negative control. For cell-cycle analyses, cells were collected as a single-cell suspension by limited trypsin treatment and fixed in 70% ethanol. Cells were incubated in 100 μL of PBS containing DNAse-free RNAse (100 μg/mL, Qiagen), and after 45 minutes, propidium iodide (PI; 5 μg/mL) was added before flow cytometry. For cell apoptosis analyses, cells were stained using APC-conjugated Annexin V and 7-amino-actinomycin (7-AAD; BD Pharmingen) and analyzed within 1 hour by flow cytometry.
Protein extracts and immunoblotting
Cell lysis buffer (1% Triton X100, 1% sodium deoxycholic acid, 0.1% SDS, 50 mmol/L HEPES, pH 7.4, 150 mmol/L NaCl, 10% glycerol, 1.5 mmol/L MgCl2, 1 mmol/L EGTA, 10 mmol/L sodium pyrophosphate, 100 mmol/L NaF, 1 mmol/L sodium orthovanadate, 10 μg/mL leupeptin, 10 μg/mL aprotinin) was used to extract proteins from cultured cells and tumors as described (12). For conditioned media analyses, cells were cultured in serum-free OptiMEM (Life Technologies) for 24 hours, media collected, filtered (0.45 μm), and concentrated using centrifugal filtration (Millipore).
Paraffin-embedded normal ovarian and ovarian tumor tissue arrays were deparaffinized, rehydrated, processed for antigen retrieval, and peroxidase quenched as described (12). OV811, OV807, OV1502, and OV8010 (US Biomax) slides were used for β5 staining and OV811 used for FAK, pY397 FAK, and osteopontin. Tissues were blocked (PBS with 5% normal goat serum, 0.5% BSA, and 0.1% Triton X-100) for 45 minutes at room temperature (RT) and incubated with anti-pY397 FAK (1:100), anti-FAK (1:100), anti-β5 integrin (1:50), anti-osteopontin (1:500) in blocking buffer overnight. Biotinylated goat-anti-(rabbit/mouse) IgG or rabbit-anti-goat IgG (1:300), Vectastain ABC Elite, and diaminobenzidine were used to visualize antibody binding. Slides were counterstained with hematoxylin. Images were captured using an upright microscope (Olympus BX43) with color camera (Olympus SC100). Staining intensity scoring was blinded.
Frozen tumors were thin sectioned (7 μm) using a cryostat (Leica), mounted onto glass slides, fixed with acetone, permeabilized (PBS with 0.1% Triton) for 1 minute, and blocked (PBS with 8% goat serum) for 2 hours at RT. Sections were incubated in anti-αvβ5 integrin (1:200) in PBS with 2% goat serum overnight, washed, and incubated with goat-anti rabbit Alexa Fluor-647 with Hoechst 33342 to visualize nuclei. Images were acquired using a spinning disc confocal microscope (IX81; Olympus), OrcaER CCD camera (Hamamatsu), pseudo-colored, and merged using Adobe Photoshop.
Three-dimensional spheroid imaging
Tumor spheroid staining was performed as described (24), with some modifications. Spheroids were fixed and permeabilized for 3 hours at 4°C in PBS containing 4% PFA and 1% Triton X-100 with gentle rocking and then blocked in PBST (0.1% Triton X-100 in PBS) containing 3% BSA and 8% goat serum overnight at 4°C. Primary anti-pY397FAK (1:100) and anti-osteopontin (1:500) in PBST were incubated at 4°C for 24 hours followed by Alexa Fluor–conjugated secondary antibodies and Hoechst 33342 for 90 min at RT. Spheroids were mounted onto glass slides in 15 μL of PBS with 30 μL of Vectashield and images acquired using a Nikon Eclipse C1 confocal microscope (EZ-C1 3.50 imaging software).
Mouse tumor studies
Eight-week-old female nude (nu/nu) mice (UCSD breeding colony) were housed in pathogen-free conditions. Tumor cells were washed in PBS, injected (2 × 106 cells in 100 μL of PBS) subcutaneously into right and left flanks of nude mice, and tumor volume (length × width2/2) determined by Vernier caliper measurements over 24 days. Orthotopic tumor growth was initiated by surgical implantation (0.4 × 106 cells in 7 μL of growth factor–depleted Matrigel) within the bursal region surrounding one ovary as described (12). Primary tumor weight was determined following euthanasia upon dissection. Fluorescent images of the intra-abdominal cavity and internal organs were acquired using an OV100 Small Animal Imaging System (Olympus). Blood was collected by heart puncture following euthanasia, samples were centrifuged, and serum was stored at −80°C. The UCSD Institutional Animal Care and Use Committee approved all mouse procedures.
The Kaplan–Meier plotter (25) was used to query gene expression and survival data from Gene Expression Omnibus and The Cancer Genome Atlas (Affymetrix HG- U133A, HG-U133A 2.0, HG-U133 Plus 2.0, and U95Av2 microarrays). Probes used are listed in Supplementary Table S1. Query parameters were overall survival, split patients by median, auto-select best cut-off, and follow-up threshold of 10 years. Restriction analyses were stage (all), histology (serous), grade (all), optimal debulk (all), and chemotherapy treatments (all). A total of 1,038 patient samples were analyzed and HRs and log-rank P significance were calculated via website interface.
Differences between groups were determined using one-way ANOVA with Tukey post hoc analyses (Prism). Differences between pairs of data were determined using an unpaired 2-tailed Student t test (Prism). Differences between β5 integrin in normal ovary, stage I, and stages II–IV was determined using the Kruskal–Wallis test. P < 0.05 was considered significant.
Osteopontin, β5 integrin, and FAK levels correlate with serous ovarian cancer patient survival
Whereas integrins αv and β1 can promote ovarian carcinoma growth, elevated β3 integrin expression may inhibit tumor progression (26, 27). Although increased β5 integrin levels are part of an unfavorable ovarian cancer gene signature (28), limited IHC analyses detected αvβ5 reactivity only in ovarian tumors of low malignant potential (29). Therefore, connections between αvβ5 integrin and ovarian tumor progression remain unclear. We evaluated the importance of β5, αv, β3, and β1 integrin mRNA levels in a large annotated database of ovarian cancer patient samples (Fig. 1). Kaplan–Meier analyses showed that elevated β5, αv, and β1 integrin levels are significantly associated with decreased patient survival (Fig. 1A). In contrast, β3 integrin levels were not associated with patient survival differences (Fig. 1A). Expression of matrix ligands for αvβ5 integrins such as osteopontin and a downstream target of αvβ5 signaling such as FAK were also significantly associated with decreased patient survival (Fig. 1A).
Increased β5 integrin staining in stage II–IV serous ovarian tumors
As determined by tumor staining, increased FAK, pY397 FAK, and osteopontin levels correlate with a poor ovarian cancer patient prognosis (6, 30, 31). Staining of tumor tissue array serial sections with antibodies to osteopontin, FAK, FAK pY397, and β5 integrin revealed parallel increases as a function of tumor stage (Fig. 1B and Supplementary Fig. S1A). Specificity of FAK pY397 staining was confirmed by analyses of ID8-IP ovarian tumors from mice treated with vehicle or PF-271 FAK inhibitor (Supplementary Fig. S1B). Additional tumor tissue array staining analyses revealed no difference between β5 integrin levels in normal ovary tissue and stage I serous tumors (Fig. 1C). However, analyses of advanced stages II–IV tumors that present foci of dissemination showed significantly increased β5 integrin staining compared with stage I tumors that are confined to the ovary (Fig. 1C, P < 0.05). Together with the mRNA array analyses, these results support the hypothesis that osteopontin, αvβ5 integrin, and FAK activity may function as a signaling axis promoting ovarian tumor progression. Moreover, β5 integrin expression may serve as a biomarker for serous ovarian carcinoma cells that possess active FAK.
Identification of FAK inhibitor–sensitive and -resistant ovarian cancer cells
Analyses of seven ovarian carcinoma cell lines in anchorage-independent growth assays identified sensitive (HEY, OVCAR8) and resistant (SKOV3-IP, OVCAR10) cells to 0.1 μmol/L FAK inhibitor (VS-4718) addition (Fig. 2A). SKOV3-IP and OVCAR10 cells remained resistant with up to 1.0 μmol/L VS-4718 for 72 hours, whereas OVCAR3, ID8-IP, and IGROV1-IP cells exhibited an intermediate growth-inhibitory response. Flow cytometric analyses were performed to determine whether VS-4718 (1 μmol/L, 72 hours) triggered cell death (7-AAD staining and Annexin V binding) and/or alterations in cell-cycle progression in sensitive (HEY, OVCAR8) or resistant (SKOV3-IP, OVCAR10) cells. Early (Annexin V–positive) and late (Annexin V and 7-AAD–positive cells) OVCAR8 apoptotic cells were detected as well as OVCAR8 cells with G0–G1 block, and decreased S-phase cell-cycle percentage upon VS-4718 treatment (Supplementary Fig. S2). HEY cells did not exhibit changes in apoptosis, but VS-4718 blocked HEY cell-cycle progression (Supplementary Fig. S2). Treatment of OVCAR10- or SKOV3-IP–resistant cells with 1 μmol/L VS-4718 did not alter cell-cycle progression or promote cell death (Supplementary Fig. S2). Thus, in sensitive cells, FAK inhibitor treatment promotes G0–G1 cell-cycle arrest followed by cell death.
Previous studies implicated the PI3K/Akt kinase pathway as a downstream target of FAK in ovarian tumor cells (31, 32). Akt activation is common in high-grade, late-stage serous ovarian tumors (33). To gain insights into molecular targets altered by FAK inhibitor treatment, immunoblotting analyses were performed on lysates of sensitive (HEY, OVCAR8) and resistant (OVCAR10, SKOV3-IP) cells grown in suspension for 72 hours in the presence or absence of 1 μmol/L VS-4718 (Fig. 2B). VS-4718 prevented FAK Y397 phosphorylation in SKOV3-IP, HEY, and OVCAR8 cells, whereas FAK Y397 phosphorylation was already low in OVCAR10 cells. Resistant OVCAR10 and SKOV3-IP cells had high Akt S473 phosphorylation and no changes in β5 integrin levels upon VS-4718 addition (Fig. 2B). In contrast, Akt S473 phosphorylation was not detected and β5 integrin levels were decreased in VS-4718–treated sensitive HEY and OVCAR8 cells, compared with controls. These results suggest that FAK inhibitor–resistant cells may contain genetic alterations promoting Akt S473 phosphorylation and that FAK activation may be part of a signaling loop controlling β5 integrin levels in sensitive cells.
FAK activity regulates β5 integrin expression and anchorage-independent cell growth
Intraperitoneal (IP) growth of murine ID8 ovarian carcinoma cells followed by in vitro culture resulted in the isolation of aggressive cells, named ID8-IP (12). Compared with parental ID8 cells, FAK Y397 phosphorylation (pY397 FAK), β5 integrin, and osteopontin levels are elevated in ID8-IP cells under anchorage-independent conditions (Fig. 3A). In both ID8-IP and HEY cells, 1 μmol/L VS-4718 treatment selectively lowers pY397 FAK, β5 integrin, and osteopontin levels (Fig. 3B–D). To confirm that this was due to FAK inactivation, HEY cells were transduced with scrambled (Scr) or FAK shRNA to knockdown FAK expression about 90% (Fig. 3E). GFP-tagged FAK-WT or -KD were stably re-expressed in HEY FAK shRNA cells at equivalent levels (Fig. 3E and F). GFP-FAK-WT cells exhibited elevated pY397 FAK compared with GFP-FAK-KD cells (Fig. 3F).
To determine whether loss of FAK expression or activity altered HEY cell growth, analyses were performed under adherent, suspended, and soft agar conditions (Fig. 3G–I). No growth differences were noted when cells were grown on plastic (Fig. 3G), but FAK knockdown reduced growth in suspension and soft agar (Fig. 3H and I). This was rescued by GFP-FAK-WT but not GFP-FAK-KD re-expression. Correspondingly, FAK knockdown reduced HEY growth as subcutaneous tumors and this was rescued by GFP-FAK-WT but not GFP-FAK-KD re-expression (Fig. 4A and B). GFP-FAK WT also promoted orthotopic HEY tumor growth and spontaneous peritoneal metastasis that was significantly reduced in HEY cells expressing GFP-FAK-KD (Fig. 4C and D). These results show that FAK activity is important for anchorage-independent and ovarian tumor growth.
Analyses of HEY tumors showed reduced pY397 FAK, osteopontin, and αvβ5 integrin levels in GFP-FAK-KD compared with GFP-FAK-WT tumors (Fig. 4E and F). Immunoblotting ID8-IP tumor lysates showed that oral FAK inhibitor administration reduced pY397 FAK, osteopontin, and β5 integrin levels compared with vehicle control-treated mice (Supplementary Fig. S3). Interestingly, qPCR revealed no changes in β5 integrin mRNA levels upon genetic or pharmacological FAK inhibition in HEY cells (Supplementary Fig. S4). Together, these results show that the inhibition of FAK activity in HEY cells decreases tumor growth with a corresponding reduction in β5 integrin protein levels that occurs independently of changes in β5 integrin mRNA expression.
β5 integrin promotes HEY ovarian tumor growth
To determine whether FAK and β5 integrin comprise a signaling axis promoting ovarian tumor growth, 2 independent lentiviral shRNAs were used to stably knockdown HEY β5 integrin expression (Fig. 5A). Flow cytometric analyses showed that αvβ5 integrin was reduced about 10-fold on the surface of HEY cells (Fig. 5B). HEY β5 integrin knockdown did not result in compensatory increases in αvβ3 or β1 integrin surface expression (Supplementary Fig. S5). β5 knockdown minimally affected the growth of HEY cells in adherent conditions compared with the scrambled control (Fig. 5C). In contrast, β5 knockdown significantly reduced HEY growth in soft agar (Fig. 5D). This was associated with decreased FAK Y397 phosphorylation and osteopontin expression as determined by immunofluorescent staining of spheroids (Fig. 5E). When injected orthotopically into the ovarian bursa space, HEY β5 integrin knockdown cells resulted in decreased tumor size after 21 days and reduced serum levels of cleaved (25 kDa) human osteopontin (Fig. 5F and G). Together, these results show that the FAK–β5 integrin signaling axis promotes HEY tumor growth and that osteopontin may serve as a secreted ligand in this pathway.
Partial phenotypic rescue by activated Akt expression
FAK inhibitor–resistant SKOV3-IP and OVCAR10 cells exhibited elevated Akt S473 phosphorylation, an indirect marker of Akt activation in anchorage-independent conditions (Fig. 2B). One possible explanation is that SKOV3-IP cells contain activating mutations in PIK3CA (Supplementary Table S2) and this may bypass effects of upstream FAK inhibition. Moreover, studies have shown that inhibition of mTOR, a downstream target of Akt, prevents SKOV3 and OVCAR10 cell growth (34). To determine whether Akt activation is sufficient to bypass FAK inhibition, membrane-targeted myristoylated Akt (Akt*) was stably expressed in sensitive HEY and OVCAR8 cells (Fig. 6). Immunoblotting of lysates showed that Akt* remained highly phosphorylated at S473 and T308 in the presence of 1 μmol/L VS-4718 treatment of cells for 72 hours in suspension (Fig. 6A). Although Akt* remained active, FAK Y397 phosphorylation was equally reduced by VS-4718 addition in control vector (CTRL) and Akt*-expressing HEY and OVCAR8 cells (Fig. 6A). These results are consistent with Akt being downstream of FAK.
To test the effects of Akt* on suspended cell growth, CTRL and Akt*-expressing HEY and OVCAR8 cells were grown in suspension in the presence or absence of 1 μmol/L VS-4718 treatment for 72 hours (Fig. 6B). CTRL HEY and OVCAR8 cells remained highly sensitive to FAK inhibitor addition (70% growth inhibition), and surprisingly, Akt*-expressing cells showed about 50% growth inhibition to VS-4718. Although this was significantly higher than CTRL cells, Akt* did not completely suppress HEY and OVCAR8 sensitivity to VS-4718 growth inhibition (Fig. 6B). When resistant OVCAR10 cells were treated with VS-4718 in combination with wortmannin (a PI3K inhibitor), anchorage-independent growth and Akt S473 phosphorylation were decreased (Fig. 6C). Finally, when analyzing β5 integrin surface expression, there was a significant reduction in CTRL HEY cells upon VS-4718 addition and this reduction was not observed in HEY Akt* cells (Fig. 6D). These results support the conclusion that FAK to Akt signaling is important for maintenance of β5 integrin surface expression. However, ovarian tumor growth resistance to FAK inhibitor treatment likely involves multiple pathways in addition to Akt activation.
The high mortality rate in ovarian cancer is partially due to its unusual mechanism of dissemination. Cells are shed from the primary tumor into the peritoneal cavity where tumor growth occurs in an anchorage-independent manner as clumps of aggregated cells termed spheroids (2). Under these conditions, interactions between integrins and matrix proteins promote cell survival and proliferation. We find that pharmacologic and genetic inhibition of FAK decreases ovarian carcinoma β5 integrin and osteopontin levels in tumors. This role for FAK activity is distinct from the canonical linkage of matrix–integrin binding leading to FAK activation (32). FAK inhibition or β5 knockdown reduced ovarian tumor cell growth under anchorage-independent conditions with corresponding decreases in orthotopic tumor growth. High osteopontin, β5 integrin, and FAK mRNA levels are associated with decreased survival of patients with serous ovarian cancer and IHC analyses confirmed that protein levels correlate with increasing serous ovarian tumor stage. Our results support a model whereby FAK inhibition disrupts autocrine or paracrine signaling regulating β5 integrin and osteopontin levels in ovarian carcinoma cells. β5 integrin expression may serve as a biomarker for serous ovarian carcinoma cells that possess active FAK signaling. Moreover, reduction of β5 integrin levels may serve as an indicator of FAK inhibitor effectiveness in ovarian cancer.
Notably, genetic and pharmacological FAK inhibition prevented anchorage-independent but not adherent ovarian cancer cell growth. Moreover, we identified cells as either sensitive (HEY, OVCAR8) or resistant (SKOV3-IP, OVCAR10) to VS-4718 treatment. Despite the fact that several drugs have low efficacy on tumor cells cultured as spheroids (35, 36), nanomolar concentrations of VS-4718 prevented sensitive ovarian cancer cell growth as spheroids by triggering cell-cycle blockage and apoptosis. As FAK inhibitors are being tested in clinical trials, it is important to identify molecular drivers of potential resistance as a means to select patients that may preferentially benefit from FAK inhibitor treatment.
Analysis of mutation frequency, copy number, or gene expression changes revealed that about 45% of serous ovarian cancer contain some type of alteration that would activate PI3K and RAS signaling pathways (37). Interestingly, studies have found that pharmacologic FAK inhibition (PF-271, 40 mg/kg) decreased tumor volume in a KRAS G12D mouse model of non–small cell lung carcinoma and human lung tumor cell xenografts (38). In addition, VS-4718 (PND-1186) FAK inhibition was effective in preventing MDA-MB-231 (KRAS G13D and BRAF G464V) breast carcinoma orthotopic tumor growth and metastasis (18). Sequencing of the HEY-A8 ovarian tumor subclone reveals KRAS G12D and BRAF G464E activating mutations (Supplementary Table S2; ref. 39), and HEY-A8 cells are responsive to pharmacologic FAK inhibition (Verastem, VS-6063; ref. 31). Because HEY cells are sensitive to VS-4718 FAK inhibition, these studies support the notion that KRAS and BRAF oncogenic mutations do not confer a FAK inhibitor–resistant phenotype.
Furthermore, it is known that PI3K and Akt activation can be downstream targets of FAK signaling in ovarian cancer (31, 32). Sequencing of SKOV3 and IGROV1 ovarian tumor cells has revealed activating mutations in PIK3CA (Supplementary Table S2; ref. 39). Although combined PI3K and FAK inhibition had additive effects in preventing OVCAR10 anchorage-independent growth, expression of activated Akt was not sufficient to generate a FAK inhibitor–resistant phenotype in HEY or OVCAR8 cells. Together, these results support the notion that FAK signaling impacts a growth-promoting pathway distinct from that activated by oncogenic mutations in KRAS, BRAF, and PIK3CA.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: I. Tancioni, D.D. Schlaepfer
Development of methodology: I. Tancioni, D.D. Schlaepfer
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Tancioni, S. Uryu, F.J. Sulzmaier, N.R. Shah, C. Lawson, N.L.G. Miller, C. Jean, X.L. Chen, K.K. Ward
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Tancioni, F.J. Sulzmaier, N.R. Shah, C. Lawson, N.L.G. Miller, C. Jean, K.K. Ward, D.D. Schlaepfer
Writing, review, and or revision of the manuscript: I. Tancioni, F.J. Sulzmaier, N.R. Shah, C. Lawson, N.L.G. Miller, C. Jean, K.K. Ward, D.D. Schlaepfer
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Uryu, N.L.G. Miller, X.L. Chen
Study supervision: I. Tancioni, D.D. Schlaepfer
The authors thank David Tarin for providing guidance and expertise in tumor pathology.
This work was supported by NIH grant CA102310 and a grant from “Nine Girls Ask?” to D.D. Schlaepfer. I. Tancioni was supported by a grant from Susan G. Komen for the Cure (KG111237). F.J. Sulzmaier was supported by NIH training grant (T32-CA121938). C. Lawson was supported in part by an Ovarian Cancer Research Fund fellowship (258835). C. Jean was supported by an American Heart Association fellowship (12POST11760014). N.L.G. Miller was supported by a National Research Service Award (1F32CA159558). N.R. Shah and K.K. Ward are fellows of the UCSD Reproductive Medicine Gynecologic Oncology program.
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.