Focal Adhesion Kinase Regulates the DNA Damage Response and Its Inhibition Radiosensitizes Mutant KRAS Lung Cancer

Non–small cell lung cancer (NSCLC) is associated with an aggressive course and poor prognosis. These features are due to the fact that this cancer type is often diagnosed at a stage not amenable to surgical resection. In addition, NSCLC is either endowed with or acquires resistance to available medical treatments. Accordingly, NSCLC is a leading cause of cancer-related deaths worldwide (1). In NSCLC several known oncogenes promote tumorigenesis and predict response to therapy. For instance, approximately 40% of NSCLC harbor mutations of KRAS, EGFR, ALK translocation, or amplification of c-MET (2, 3). While there are specific inhibitors of EGFR, ALK, or MET that cause clinically meaningful, but short-lived antitumor responses, mutant KRAS remains undruggable (4, 5). Thus, therapeutic options are still limited for patients with oncogenic KRAS NSCLC.

KRAS belongs to a family of small guanosine triphosphatases (GTPase). Tumor-associated mutations lock KRAS in a constitutively active state (i.e., oncogenic KRAS; refs. 6, 7). When bound to GTP, KRAS activates several critical cell proliferation and survival signals, which include the PI3K/mTOR, MEK1/2/ERK1/2, RHOA-Focal adhesion kinase (FAK), and TBK1 signaling networks (6, 8–10).

Oncogenic KRAS is not only sufficient to induce lung cancer, but also required for its maintenance in transgenic mice and in human lung cancer cells (11–14). However, progression to high-grade lung adenocarcinoma requires co-occurring mutations, such as the loss of TP53, CDKN2AB, Ataxia Telangiectasia Mutated (ATM), or LKB1, which allow bypass of tumor suppressive responses induced by inappropriate proliferative stimuli, DNA damage, or energetic stress (2, 3, 8, 11–13, 15–17). The loss of tumor suppressors implicated in DNA damage repair is relevant because oncogenic KRAS also stimulates the production of reactive oxygen species (ROS), which promote DNA damage and genomic instability (18, 19).

Because attempts to develop direct inhibitors of oncogenic KRAS have been unsuccessful (4), intense efforts have been spent on targeting critical components of its downstream signaling networks in preclinical models. Several PI3K, mTOR, MEK1/2, and FAK inhibitors (FAKi) are undergoing clinical testing, but they have not been approved for therapy of lung cancer (20–23). Thus, there is still an urgent need for the development of therapies that target oncogenic KRAS tumors.

Protein tyrosine kinase 2 (PTK2), also known as FAK, is a nonreceptor tyrosine kinase and a major mediator of integrin signaling. Upon autophosphorylation at Tyr397, FAK interacts with SRC protein kinase family members, initiating several signaling cascades that regulate cytoskeleton remodeling, cell migration and resistance to anoikis (24). FAK is amplified or overexpressed in several cancer types, including ovarian, colon, breast, and lung cancers (25, 26). Importantly, FAK inhibition is detrimental to breast and lung cancer cells: in this context, disruption of FAK is associated with alterations in the cytoskeleton or induction of senescence and activation of DNA damage pathways, respectively (8, 25). Furthermore, in lung cancer, mutant KRAS is a positive regulator of FAK (8). However, the mechanisms underlying senescence induction following FAK inhibition and the functional consequences of this event in cancer cells remain unexplored.

In this study, we characterized the effects of FAK suppression in a large panel of NSCLC cells representative of frequent cancer associated mutations. We found that FAK inhibition invariably impairs the viability of mutant KRAS NSCLC cells. In this genetic context, suppression or inhibition of FAK was accompanied by DNA damage. In addition, we demonstrate that FAK suppression synergizes with radiotherapy both in vivo and in vitro. We propose that combination therapy with FAK inhibition and ionizing radiation (IR) may lead to important clinical benefits in the treatment of NSCLC with oncogenic KRAS.

Human NSCLC cell lines and human bronchoalveolar cells were provided by Dr. John Minna (UT Southwestern Medical Center, Dallas, TX) and cultured as described (27, 28). A549 and H460 cells expressing vector control or LKB1 were previously described (17). All cells were mycoplasma free and were identified by DNA fingerprinting. Supplementary Table S1 shows the mutations that they harbor according to COSMIC: Catalogue of Somatic Mutations in Cancer (Cosmic; cancer.sanger.ac.uk). FAKi PF-562,271 and VS-4718 were obtained from Selleckchem and Verastem, Inc., respectively (29, 30). Inhibitors were added to mid-log phase cell cultures at the indicated concentrations. All other chemicals were purchased from Sigma-Aldrich.

Stable FAK mRNA knockdown was performed with lentiviral vectors containing shRNAs against FAK obtained from the RNAi Consortium (TRC) following procedure described previously (8). Inducible expression of FAK shRNA was performed with the GEPIR vector (31). See also Supplementary Materials and Methods.

We followed established procedures to ablate FAK (exon 4) using the following vectors: pCW57.1 (Addgene plasmid 50661) and pLX-sgRNA (Addgene plasmid 50662; ref. 32). We selected several single clones and assessed FAK editing by direct sequencing and by lack of FAK protein by Western blot (WB) analysis.

Cells (0.5×10⁴ to 1.5×10⁴) were plated in triplicate (24-well plates) 14 to 16 hours prior to exposure to pharmacologic inhibitors. At the indicated time points or 72 hours later in cell viability assays, the cells were fixed with 10% formalin and stained with 0.1% crystal violet (8).

WB analyses were performed as previously described (33). We used the following antibodies: FAK, Tyr397 phospho-FAK, Akt, Ser473 phospho-Akt, S6, Ser235/236 phospho-S6, Ser139 γ-H2AX (Cell Signaling Technology), H2AX (Bethyl), β-Tubulin and GAPDH (Santa Cruz Biotechnology), Vinculin and LKB1 (Cell Signaling Technology).

For colony formation assays in plastic, we plated 500 to 1,500 cells in 60-mm tissue culture dishes and scored colonies of >50 normal-appearing cells after 12 to 30 days. For soft-agar colony assays, we seeded 2,000 cells/well on semisolid agar medium in a 6-well plate in triplicate and after 14 to 21 days, colonies larger than 50 μm were counted using an inverted microscope (34).

Cells were allowed to adhere overnight. When indicated, cells were treated with PF-562,271 and/or IR (2 Gy). Analysis of the cell cycle and percentage of cells stained with propidium iodide (PI; Sigma-Aldrich) and γ-H2AX (Millipore) were performed following a standard procedure with a FC500 Beckman Coulter flow cytometer using the WinMDI V2.8 software (35, 36).

pMXs-Puro-GFP-Fak was obtained from Addgene (FAK-Plasmid #38194). Retroviral transductions were performed as described (37).

Gene expression profiles were obtained from exponentially growing HBEC cells transduced with pMXs-Puro-GFP or pMXs-Puro-GFP-Fak. Microarray results have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO series accession number GSE72470. See also Supplementary Data.

Clonogenic assays were performed as described previously (38). Surviving fractions (SF) were derived using the number of colonies formed after treatment, divided by the number of cells seeded multiplied by plating efficiency. Cells were plated in triplicate onto 60-mm dishes 14 to 16 hours prior to irradiation. We added PF-562,271 at indicated concentrations to the cells four hours before irradiation. Drug-containing medium was replaced with drug-free medium after 48 hours. Cells were irradiated with a Mark I cesium irradiator (1.31 Gy/min). We scored colonies of >50 normal-appearing cells 12 to 30 days after treatment and graphed the SF versus dose of IR (Gy). Do (relative dose of IR required for 37% lethality on a log-phase kill curve), Dq (inherent DNA repair capacity: dose (Gy) required to eliminate the survival curve shoulder) and dose enhancement ratios (DERs at LD₅₀ and LD₂₀) were calculated as described (39).

Immunofluorescence was performed as described previously using γ-H2AX (Millipore), TP53BP1 (Bethyl), and FAK (Abicam ab40794) antibodies (8).

Xenograft experiments using T2.2 FAK wild-type (FAK⁺) and T2.2 FAK-null (FAK⁻) H460 NSCLC cells were performed by subcutaneous inoculation of cells into 6-week-old female athymic nude mice. Mice with xenograft tumors of 300 mm³ (7 mice/group) were treated with IR. Mice were irradiated with five 4-Gy fractions every other day for 10 days using an X-RAD 320 irradiator (Precision X-Ray, Inc.) to deliver local irradiation to the flank or thigh of lead-shielded mice. Tumor volumes were calculated every other day using the formula: (length × width²)/2. All studies were performed according to the guidelines of the UT Southwestern Institutional Animal Care and Use Committee.

All data presented are the average ± standard deviations of experiments repeated three or more times. Significance was determined using two-tailed unpaired Student t tests or one-way ANOVA. Curve fitting for the radiosensitization experiments was performed using the linear-quadratic formula which has two components of cell killing: one is proportional to the dose of IR (aD) and the other is proportional to the square of the dose of IR (bD²): exp (aD+bD²) (38, 40).

The goal of our study was to shed light on the function of FAK in lung cancer cells and to leverage this knowledge to identify novel therapeutic opportunities. Pharmacologic inhibition of FAK with the small molecule inhibitor (FAKi) PF-562,271 led to a striking inhibition of cell viability in a panel of NSCLC cell lines (detailed genotype information is provided in Supplementary Table S1). We noted that the IC₅₀ of mutant KRAS NSCLC cell lines ranged between 2 and 4 μmol/L, while in wild-type KRAS cell lines the IC₅₀ was 8 μmol/L or higher (Fig. 1A; Supplementary Table S2). When used at a concentration between 2 and 4 μmol/L, PF-562,271 led to approximately a 50% reduction of auto-phosphorylation of FAK at Y397 (P-FAK, the active form of FAK), both in wild-type and in mutant KRAS NSCLC cells (Fig. 1B; Supplementary Fig. S1A). We obtained a similar inhibition of cell viability and P-FAK when treating NSCLC cells with VS-4718, a FAKi structurally distinct from PF-562,271 (Supplementary Fig. S1B–S1D).

Of note, pharmacokinetic studies have demonstrated that PF-562,271 and VS-4718 (formerly known as PND-1186) reach a concentration of 1 to 2 μmol/L in vivo (29, 41). Notably, PF-562,271 inhibits its target, causing antitumor effects in vivo in a mutant KRAS lung cancer model (8).

Integrin engagement with the extracellular matrix has been implicated in the activation of FAK; thus, we tested whether culturing NSCLC cells on collagen coated plates would affect their vulnerability to FAKi. However, culturing NSCLC cells on collagen-coated plates did not affect their sensitivities to FAKi (Supplementary Fig. S1E–S1G). Taken together, our findings suggest that pharmacologic inhibition of FAK may be of therapeutic value in mutant KRAS NSCLC.

To validate, with an alternative technique, the results obtained with FAKi, we inactivated FAK genetically in a panel of 21 NSCLC cells harboring mutant KRAS or wild-type KRAS, mutant EGFR, the EML4–ALK fusion gene or MET amplification and loss/mutations of the TP53, CDKN2A/B, or LKB1 tumor suppressors (detailed genotype information is provided in Supplementary Table S1). We found that silencing FAK invariably leads to loss of viability in mutant KRAS lung cancer cells (Fig. 1C). Notably, the degree of FAK silencing was comparable between wild-type and mutant KRAS NSCLC cells (Fig. 1C–D; Supplementary Fig. S1H). In contrast, the effect of FAK silencing on cell viability was not consistent in lung cancer cells carrying genotypes other than mutant KRAS. For instance, FAK silencing strikingly reduced the viability of H1975 (mutant EGFR) and H1993 (MET amplified) cells, but not of H1650 and HCC827 (mutant EGFR) or H920 cells (MET amplified). We also noticed that the vulnerability to FAK silencing was comparable between mutant KRAS cells that carry p53, CDKN2a, or LKB1 mutations (Fig. 1C; Supplementary Table S1). This result confirms and extends our previous observations in a smaller sample size of mutant KRAS NSCLC cells, providing further support to the notion that FAK is a therapeutic target in mutant KRAS NSCLC (8).

It is well known that small molecule inhibitors of protein tyrosine kinases as well as RNAi-mediated gene silencing, may lead to off-target effects. Thus, we ablated FAK by CRISPR/CAS9 gene editing in H460-mutant KRAS lung NSCLC cells, as a representative example of the NSCLC used in our studies. We carried out our analysis in two independent FAK-null (T2.2 FAK⁻ and T2.7 FAK⁻) and wild-type FAK (T2.2 FAK⁺ and T2.7 FAK⁺) H460 clones (Fig. 1E). We determined that, also in this setting, loss of FAK dramatically affects the proliferative capacity of H460 cells (Fig. 1F and G).

Next, we determined that FAK silencing impairs the ability of mutant KRAS H460, but not of wild-type KRAS H522 NSCLC cells, which we chose as representative examples, to grow in an anchorage-independent manner, a hallmark of oncogenic transformation and tumor aggressiveness (ref. 42; Fig. 2A). These cell lines are widely used in the literature for similar experiments. Furthermore, we determined that CRISPR/CAS9-mediated ablation of FAK significantly reduces the clonogenic ability of H460 cells when cultured on plastic dishes and in soft agar (Fig. 2B and C).

It is also noteworthy that the acute silencing of FAK with a retrovirus that expresses a doxycycline (doxy)-inducible FAK shRNA, which mimics the action of a pharmacologic agent, significantly impairs the clonogenic ability of H460 cells when grown on plastic (Fig. 2D).

Taken together, these data support the conclusion that FAK is required for the transforming ability of oncogenic KRAS, which could be targeted for therapeutic purposes in lung cancer.

Flow cytometry revealed that pharmacologic inhibition of FAK leads to a significant increase of the percentage of the G₂ phase of the cell cycle in mutant KRAS NSCLC cells, but not in wild-type KRAS H522 NSCLC cells (Fig. 3A and B). These data suggest that the effects of FAK suppression are mediated by a mechanism that utilizes a G₂ cell-cycle checkpoint.

To characterize the effect of FAK suppression in NSCLC cells, we assessed the status of several targets of KRAS (6). We found that pharmacologic inhibition of FAK (FAKi) or FAK gene silencing did not affect p-AKT, p-ERK, and p-S6 levels (Supplementary Fig. S2A–S2E). This observation suggests that inhibitory effects on these signaling pathways do not cause the effect of FAKi on cell proliferation and viability.

To gain insight into the cellular networks affected by FAK, we performed gene expression analysis of primary human bronchoalveolar epithelial cells (HBEC) immortalized by expression of hTERT and CDK4. These cells have been extensively studied to determine the effect of oncogenic mutations in respiratory epithelial cells (28). In addition, using this cellular system we previously demonstrated that mutant KRAS activates FAK (8). Therefore, we reasoned that these cells are an experimental system to assess the cellular networks regulated by FAK. To this end, we compared the transcriptome of HBEC cells expressing pMXs-Puro-GFP or pMXs-Puro-GFP-Fak. Database for Annotation, Visualization and Integrated Discovery-based (DAVID) functional enrichment analysis and Gene Set Enrichment Analysis (GSEA) between experimental groups revealed a significant upregulation of genes that regulate G₂–M DNA damage checkpoint, TGFβ signaling, and PKA signaling (Fig. 3C).

To address the relationship of FAK with DNA damage repair networks, we tested whether loss of FAK affects activation of γ-H2AX, an occurrence that was also previously reported to occur in breast cancer cells (25). Indeed, we found that silencing, pharmacologic inhibition, or ablation of FAK leads to upregulation of γ-H2AX in mutant KRAS NSCLC A549 and H460 cells but not in wild-type KRAS NSCLC H522 and H596 cells. Notably, there were no major differences in baseline γ-H2AX among the cell lines used for these experiments (Fig. 3D–F; Supplementary Fig. S2F). LKB1, which is lost in a significant percentage of mutant KRAS lung cancer, has been implicated in the regulation of the DNA damage response (43). To test whether this is the case also in lung cancer cells, we treated with FAKi PF-562,271 or IRs H460 (LKB1 mutant), A549 cells (LKB1 mutant) and their counterparts where LKB1 was reintroduced by retroviral transduction, H358 (wild-type LKB1) and HCC4017 (wild-type LKB1) cells (17). We found that PF-562,271 and 4 Gy of IR cause a comparable degree of upregulation of γ-H2AX. Furthermore, LKB1 did not influence the upregulation of γ-H2AX (Fig 3G and H; and Supplementary Fig. S3A–S3C). Taken together, our data support the conclusion that loss of FAK leads to a DNA damage response in mutant KRAS NSCLC cells.

Our results suggest that FAK is required for promoting DNA damage repair in oncogenic KRAS NSCLC cells. Therefore, we tested whether pharmacologic inhibition of FAK sensitized NSCLC cells expressing oncogenic KRAS to the antiproliferative effects of IR. We performed clonogenic survival assays with H460, H358, H522, and H596 cells as representative examples of mutant and wild-type KRAS NSCLC cells, respectively, exposed to increasing doses of IR (1–6 Gy). We chose H358 cells for this experiment because they are LKB1 wild-type, while H460 and A549 are LBK1 mutant, to rule out the contribution of LKB1 loss to radiosensitization. We administered 1 μmol/L PF-562,271, because we found that this concentration inhibits P-FAK in a comparable manner in the cells we used for this study (data not shown), 4 hours before exposure to IR. We limited the incubation time to 48 hours not to affect plating efficiency.

We found that pharmacologic inhibition of FAK with PF-562,271 to irradiated mutant KRAS H460 and H358 NSCLC cells resulted in profound changes in Dq as well as significant, but less dramatic, decreases in Do (Fig. 4A and B). Thus, we concluded that FAKi reduces inherent DNA repair capacity (Dq) as indicated by significant changes in the shoulder of the survival curves. In contrast, exposure of wild-type KRAS NSCLC cells (H522 or H596; Fig. 4C and D) to PF-562,271 had no significant effect on IR-induced lethality as noted by the survival curve and estimations of Dq and Do. Most importantly, the presence of FAKi increased dose enhancement ratios (DER) with values ranging between 2.1 and 1.9 at LD₅₀ levels, and between 1.7 and 1.4 at LD₂₀ levels, respectively (Supplementary Table S3).

Notably, the effect of pharmacologic inhibition of FAK was comparable to the dual PI3K/mTOR inhibitor BEZ-235, which is a known radiosensitizing agent (Fig. 4A; refs. 20, 44).

To confirm these findings with an independent approach, we silenced FAK. Because prolonged silencing of FAK severely impairs clonogenic growth of mutant KRAS NSCLC cell lines, we used the GEPIR retrovirus to express FAK shRNA1 in a doxycycline-regulated manner, obtaining results equivalent to FAKi (Fig. 4E and F). We found that inducible silencing of FAK had effects similar to PF-562,271 treatment in both mutant and wild-type KRAS NSCLC cells.

These data indicate that silencing or pharmacologic inhibition of FAK inhibits the DNA repair and/or augment DNA damage created by IR.

Our data suggest that inhibition of FAK facilitates the cytotoxic effects of IR by promoting increased or unresolved DNA damage. Thus, we examined whether PF-562,271 affects induction and repair of DNA breaks after IR exposure in H460 cells. We determined that resolution of γ-H2AX foci, a well-known marker of DNA double-strand break (DSB) damage and repair (when foci decrease), occurred rapidly after treatment with IR (2 Gy). In contrast, treatment with the FAK inhibitor PF-562,271 in combination with IR (2 Gy) led to a striking persistence of γ-H2AX foci at 24 hours after IR administration compared with exposure with IR alone. The effect of PF-562,271 in regard to inducing the persistence of γ-H2AX foci was comparable to the effects of the dual PI3K/mTOR inhibitor BEZ-235, a known radiosensitizer (Fig. 5A; refs. 20, 44). Inhibition of FAK in the absence of IR induces γ-H2AX foci slightly above the background level in this assay (Fig. 5A; Supplementary Fig. S4A). We obtained equivalent results in H358 and HCC4017 (LKB1 wild-type) cells (Supplementary Fig. S4B–S4E).

We also obtained equivalent results when we determined the induction and resolution of γ-H2AX and TP53BP1 foci, as readout of DNA damage, in CRISPR/CAS9 H460 T2.2 FAK⁺ and H460 T2.2 FAK⁻ cells (Fig. 5B; Supplementary Fig. S5A).

Next, we confirmed by flow cytometry that inhibition of FAK with PF-562,271 or in FAK null T2.2 H460 cells affects the percentage of γ-H2AX–positive cells after exposure to IR at every time point tested (Supplementary Fig. S5B and S5C). We found that mutant KRAS H460 cells treated with FAKi display a significant increase of the percentage of cells in the G₂ phase of the cell cycle, which was further increased 4 and 24 hours after combination treatment with FAKi and IR as compared to H460 cells treated with IR only (Fig. 5C).

Western blot analysis confirmed that PF-562,271 increases γ-H2AX and in combination of IR leads to further upregulation and persistence of γ-H2AX in H460 cells (Fig. 5D). Furthermore, we found that IR activates P-FAK, which persists for at least 8 hours after IR (Fig. 5D). In addition, cell fractionation experiments revealed that a portion of FAK resides in the cell nucleus where DNA repair takes place (Supplementary Fig. S5D). FAK-null H460 cells not only display an increased basal level of γ-H2AX and its further upregulation after IR treatment, as compared to wild-type FAK H460 cells, but also persistent activation of CHK2 after exposure to IR (Fig. 5E). We detected similar findings with respect to γ-H2AX in H460 cells that underwent FAK inhibition or CRISPR/CAS9-mediated ablation, with only minor differences in the kinetics of resolution of γ-H2AX activation at later time points (Fig. 5D and E; and Supplementary Figs. S4A, S5A–C, and S5E). In this regard, there were small incongruences between the intensities of the signals obtained by IF, WB, and flow cytometry, which may be due to the fact that WB detects total cellular γ-H2AX while IF detects γ-H2AX accumulated in foci, which are also more readily detected by flow cytometry. Because γ-H2AX is a marker of DNA damage when it is accumulated in nuclear foci, we conclude that the presence of nuclear foci is representative of ongoing DNA damage.

Taken together, these data suggest that inhibition/suppression of FAK results in persistent DNA damage in NSCLC cells because of inhibition of DNA repair or augmentation of damage by cell-cycle checkpoint abrogation, which occur without affecting the activation and recruitment to sites of DNA damage of the DNA damage–sensing machinery. These observations also suggest that IR therapy could be exploited to sensitize cancer cells to therapy with FAK inhibitors.

We tested the antitumor effects of FAK inactivation mediated by CRISP/CAS9 editing in combination with IR in H460 NSCLC xenografts, a well-established model of aggressive NSCLC and a representative example of the cells we used in tissue culture experiments.

We generated four cohorts of seven athymic nude mice bearing xenografts of H460 T2.2 FAK⁺ and H460 T2.2 FAK⁻ cells of 300-mm³ average size. Mice were either mock treated or treated with five 4-Gy fractions every other day for 10 days. We delivered local irradiation to the flank or thigh of lead-shielded mice using a fractionated dose to limit overall tissue toxicity and mimic the administration modality used in the clinic (45). Notably, this xenograft volume and IR dose is comparable with previous studies involving H460 xenografts (20, 46).

As expected, FAK ablation impaired xenograft growth compared to the parental H460 cells (Fig. 6A). The median survivals of mock-treated mice carrying H460 T2.2 FAK⁺ and FAK⁻ cells were 19 days and 24 days, respectively (P = 0.0467; Fig. 6B). IR treatment of H460 T2.2 FAK⁻ cells resulted in a greater than 75% reduction in xenograft volume as compared with H460 T2.2 FAK⁺ cells 30 days after the first dose of IR (P < 0.001; Fig. 6C and D). All but one of the irradiated mice carrying xenografts of H460 T2.2 FAK⁻ cells were alive 43 days after the initiation of IR treatment; in contrast, every mouse carrying xenografts of H460 T2.2 FAK⁺ cells was sacrificed between days 34 and 43 after radiation due to excessive tumor burden (Fig. 6E).

H460 T2.2 FAK⁻ cells resumed their growth about 40 days after the first dose of radiation was administered. Xenograft growth eventually reached the maximum size allowed in these experiments, and mice had to be sacrificed. By IF analysis of frozen sections we determined that T2.2 FAK⁻ cells remained negative for FAK at the experimental endpoint (Supplementary Fig. S5F). This finding suggests that FAK-independent mechanisms may mediate resistance to FAK inhibition. It is possible that other tyrosine kinases or other prosurvival networks may mediate resistance to FAK inhibition. IR treatment was well tolerated in xenograft bearing nude mice and we did not observe any drop in body weight or other signs of toxicity in both groups (data not shown).

These results indicate that the ablation of FAK leads to significant radiosensitizing effects, which in turn led to significant antitumor effects in vivo.

NSCLC remains a significant clinical challenge due to the fact that few medical treatments are effective in this disease. It is well known that NSCLC displays either primary or acquired resistance to chemotherapy and targeted therapy. The limitations of current therapies are evident in mutant KRAS NSCLC. For instance, direct inhibitors of oncogenic KRAS lack the specificity needed for their deployment in vivo (47–49). Furthermore, inhibition of the canonical KRAS signaling pathways MEK1/2, PI3K, and mTORC1/2 have not shown benefits in lung cancer patients that justify their FDA approval for clinical use (4). As a consequence, there has been an intense interest in the identification of novel therapeutic targets for mutant KRAS NSCLC.

The data presented in this article lead to several important conclusions: not only is FAK a targetable vulnerability of mutant KRAS lung cancer, but its inhibition also leads to significant radiosensitizing effects that can be exploited in a combination therapy regimen. This finding is of significance because of the availability of several FAKi in clinical development.

We previously reported a mutant KRAS–RHOA–FAK signaling axis in NSCLC and that both RHOA and FAK are requirements of high-grade NSCLC (8). Our analysis in a large panel of lung cancer cells confirms and extends our prior findings obtained with RNAi in a smaller sample set or NSCLC cells and in a genetically engineered mouse model of lung cancer (8). Our new set of data is of significance because we used specific FAKi PF-562,271 (the parental compound of Vs6063, also known as defactinib) and VS-4718. Both defactinib and VS-4718 are being evaluated in clinical trials in lung cancer and mesothelioma. Furthermore, we used shRNAs and gene editing, which are complementary and nonoverlapping methods to genetically inactivate FAK. Thus, we reason that by using these complementary approaches we adequately addressed the concern that the interpretation of RNAi experiments may be hampered by off-target effects (50). Notably, the results we obtained are internally consistent, indicating that the deleterious effects on cell proliferation, clonogenic capacity, and delayed xenograft grow in vivo are caused by FAK inhibition.

Even though we tested NSCLC cells carrying oncogenotypes other than mutant KRAS, we did not find antiproliferative responses consistently in wild-type KRAS NSCLC cells. Thus, we conclude that the dependency on FAK is a specific feature of mutant KRAS NSCLC cells. Accordingly, mutant KRAS status could be used as a biomarker for the enrolment of patients in clinical trials using FAK inhibitors.

Our data indicate that FAK is implicated in the response to DNA damage in mutant KRAS NSCLC cells. Pharmacologic inhibition of FAK protein, FAK gene silencing, or ablation results in activation and maintenance of a DNA damage response as demonstrated by the presence of γ-H2AX in Western blots and the recruitment of γ-H2AX and TP53BP1 to DNA damage foci. In this regard, it is also noteworthy that it was reported that FAK loss in murine breast cancer cells and primary fibroblast is associated with induction of replicative senescence, even though the mechanism underlying this outcome was not defined (25). In addition, FAK inhibition was implicated in mediating radiosensitization in several other settings, for instance, melanoma and head and neck cancers (51, 52).

In xenograft experiments with FAK-null lung cancer cells, we noted that the impairment of the growth of FAK-null xenografts declined over time. The observation that FAK-null xenografts stain negative for FAK at the experimental endpoint suggests that FAK-independent mechanisms mediate resistance to FAK inhibition. It is possible that other tyrosine kinases or other prosurvival networks may mediate this phenomenon. The identification of mechanisms that mediate resistance to FAK inhibition is the subject of ongoing investigations in our lab.

Our report also provides the first demonstration that the FAK blockade is an effective strategy to sensitize mutant KRAS NSCLC cancer cells to the deleterious effects of IR both in clonogenic assays and in vivo. In this regard, it is also possible that IR, by potentiating the therapeutic effects of FAKi, may delay or prevent the emergence of drug resistance.

These findings suggest that FAK blockade/ablation in combination with IR both inhibits inherent DNA repair [i.e., decreased shoulder (Dq) values] in oncogenic KRAS-driven NSCLC, as well as augments DSBs (noted by decreased Do values, representing the number of hits needed to cause death, high-dose enhancement ratios), consistent with enhanced formation of foci due to DNA DSBs. These interpretations are consistent with the observation that the DNA damage foci induced by IR persist in FAK-deficient cells. In this setting, decreased clonogenic capacity may be caused by growth arrest or impaired cell proliferation, inhibition of prosurvival stress responses or mitotic catastrophe. Indeed, we found that inhibition or loss of FAK leads to a G₂ cell-cycle arrest, which is further increased by IR treatment. Thus, it is likely that the effects of FAK inhibition/suppression are mediated by loss of G₂ checkpoint control that, in turn, leads to an inherent loss of DNA repair capacity. Alternatively, these effects could be caused by inhibition of a DSB repair protein (such as ATM, ATR, or DNA-PKcs) that affects DSB repair and cell-cycle checkpoint control. In this respect, it is noteworthy that PF-562,271 exposure phenocopies the effects of the dual PI3K/mTOR inhibitor BEZ-235, which was reported to be an inhibitor of ATM and DNA-PK (20, 44). In the future, it will be of interest to differentiate between these two possibilities.

We noted for the first time that IR activates FAK and that FAK depletion promotes the hyperactivation of checkpoint kinase CHK2, a known downstream target of ATM/ATR and G₂–M effectors of cell-cycle arrest or apoptosis in response to DNA damage.

In this respect, it is noteworthy that FAK was reported to interact in the nucleus with PIAS1, a SUMO E3 ligase that interacts with BRCA1 in the DNA damage response (53–55). Thus, it is tempting to speculate that the requirement for FAK in mutant KRAS NSCLC cells is due to the fact that in basal growth conditions FAK promotes the repair of the DNA damage caused by oncogenic KRAS activation and that the requirement for FAK is further magnified in cells exposed to IR. In addition, p53, CDKN2A, and ATM tumor suppressors, which are well-known players in the DNA damage response and are also frequently comutated with mutant KRAS, may collaborate with FAK either directly or indirectly in mediating DNA damage repair (2, 3, 56, 57). The results of the experiments with H460 and A549 cells with restored LKB1 lead to the conclusion that loss of LKB1, another tumor suppressor implicated in DNA damage response, does not determine the dependency on FAK, at least in this context (43).

The FAKi defactinib and VS-4718 are being tested in phase I/II clinical trials, including lung cancer (58). Therefore, our work provides preclinical data useful for the design of therapeutic protocols using this novel class of drugs. We propose that mutant KRAS should represent a biomarker for the enrollment of patients in clinical trials using FAKi in NSCLC. Furthermore, we suggest that the clinical testing of combined therapy using FAKi and IR should be prioritized. In the future, it will be of interest to determine the mechanistic underpinning of the role of FAK in DNA damage repair to optimize the use of FAKi in cancer patients, identify biomarkers that predict clinical response and test FAKi in other tumor types that harbor mutant KRAS.

J.V. Heymach reports receiving other commercial research support from AstraZeneca, Bayer, and GlaxoSmithKline; and is a consultant/advisory board member for Ariad, AstraZeneca, Boerhinger Ingelheim, Exelixis, Genentech, GlaxoSmithKline, Lilly, Novartis, and Synta. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K.-J. Tang, J.D. Constanzo, N. Venkateswaran, P.P. Scaglioni

Development of methodology: K.-J. Tang, J.D. Constanzo, N. Venkateswaran, M. Melegari, J.C. Morales, P.P. Scaglioni

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.-J. Tang, J.D. Constanzo, N. Venkateswaran, M. Ilcheva, F. Skoulidis, J.V. Heymach, P.P. Scaglioni

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.-J. Tang, J.D. Constanzo, N. Venkateswaran, M. Ilcheva, J.V. Heymach, P.P. Scaglioni

Writing, review, and/or revision of the manuscript: K.-J. Tang, J.D. Constanzo, N. Venkateswaran, M. Ilcheva, F. Skoulidis, J.V. Heymach, D.A. Boothman, P.P. Scaglioni

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.-J. Tang, N. Venkateswaran, M. Melegari, D.A. Boothman, P.P. Scaglioni

Study supervision: M. Melegari, D.A. Boothman, P.P. Scaglioni

The authors thank Dr. John Minna, Luc Girard, and Amit Das at the UT Southwestern Medical Center for providing unpublished data regarding oncogenic mutations of NSCLC cell lines. They thank Mahesh Padval and Jonathan Pachter at Verastem for providing VS-4718 and for helpful discussions.

This study was financially supported by CDMRP LCRP grant #LC110229, American Cancer Society Scholar Award 13-068-01-TBG, UT Southwestern Friends of the Comprehensive Cancer Center, Texas 4000 (PPS), #2012J5100031 Science and Technology Program of Guangzhou, China (K.-J. Tang), NCI #1F31CA180689-01 and NCI T32CA124334 (JDC), NCI R01 CA102972 (D.A. Boothman), Lung Cancer Moonshot Program; V Foundation Grant; Ford Petrin Donation; CCSG Program (J.V. Heymach), The University of Texas Southwestern Medical Center and The University of Texas MD Anderson Cancer Center Lung SPORE grant 5 P50 CA070907, NIH Cancer Center Grant CA016672 and 2P30 CA142543-06.

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.