Focal adhesion kinase signaling – tumor vulnerabilities and clinical opportunities

Focal adhesion kinase (FAK; encoded by as PTK2) is a cytoplasmic protein tyrosine kinase that is recruited to and activated at sites of integrin clustering and actomyosin tension generation (McLean et al., 2005; Parsons, 2003; Schaller, 2010). Integrins do not possess intrinsic catalytic activity, yet integrin-initiated signals impact cell division, movement and survival, and promote tumor progression (Hamidi and Ivaska, 2018; Pang et al., 2023). The discovery of highly tyrosine-phosphorylated proteins at cell adhesion sites [where transmembrane integrin receptors link the extracellular matrix (ECM) to the intracellular actin cytoskeleton] (Burridge, 2017) provided initial support for the notion of integrin-associated tyrosine kinase activity. Indeed, various kinases are clustered at cell adhesion sites (Cooper and Giancotti, 2019); among these, phosphorylation events by the non-receptor protein tyrosine kinases FAK and Src relay integrin signal transmission within cells.

Loss of FAK expression or mutational inhibition of FAK activity results in embryonic lethality in mice at gastrulation (Ilic et al., 1995). FAK-null fibroblasts exhibit static cell adhesions, loss of cell polarity and slow directional cell movement in response to integrin ligands or soluble growth factor stimuli (Mitra et al., 2005; Sieg et al., 2000; Tomar and Schlaepfer, 2009). Although integrin–matrix binding is required for normal adhesion-dependent cell proliferation, this does not necessarily require FAK expression or activity (Lim et al., 2010a). Rather, FAK is associated with survival signaling and the suppression of anoikis, a type of programmed cell death after a cell detaches from ECM (Frisch et al., 1996).

In this Review, we discuss the various roles for FAK in normal and tumor cells, with an emphasis on cell phenotypes associated with the inhibition of FAK activity. We highlight recent studies regarding biomechanical and conformational-induced FAK activation, roles for FAK in the nucleus and developments in targeting kinase-independent FAK functions. Although FAK can be activated by alternative splicing (Toutant et al., 2002), gains in the FAK gene locus (PTK2) are also associated with FAK activation in both mouse and human tumors (Diaz Osterman et al., 2019). Elevated FAK signaling can also occur in parallel with oncogene activation, and FAK signaling is associated with adaptive tumor resistance to chemotherapy. As FAK functions as a master regulator of chemotherapy resistance, we discuss the tumor-intrinsic and -extrinsic effects of small-molecule FAK inhibitors on sensitizing cells to chemotherapies, reducing tumor fibrosis and counteracting immune evasion. Additionally, we will highlight ongoing FAK inhibitor phase II and III clinical trials in high- and low-grade serous ovarian cancers, respectively, that are incorporating combinatorial approaches to address unmet clinical needs.

FAK is a 120 kDa protein comprising an N-terminal band 4.1, ezrin, radixin, moesin homology (FERM) domain, a kinase domain, an unstructured region containing proline-rich motifs and a C-terminal focal adhesion-targeting (FAT) domain (Parsons, 2003) (Fig. 1). FAK can form higher-order structures via dimerization (FERM–FERM and FAT–FAT domain interactions). Within FAK, the FERM domain also makes intramolecular regulatory contacts with the FAK kinase domain to stabilize an inactive (‘closed’) conformation (Le Coq et al., 2022). FERM and FAT domains, as well as proline-rich regions, also bind to various structural and signaling proteins that contribute to FAK scaffolding function at adhesions (Schaller, 2010).

FERM domains are present in over 50 different proteins and generally comprise three lobes termed F1, F2 and F3 (Frame et al., 2010). For FAK, activation requires the release of FERM-mediated inhibitory constraints followed by induced protein dimerization. For the FAK FERM F1 lobe, point mutations or N-terminal truncations result in elevated FAK tyrosine (Y) phosphorylation at Y397 in cells (Cohen and Guan, 2005). The FERM F1 lobe blocks access to Y397 (Lietha et al., 2007), a key regulatory site that lies within a short linker region between the FERM and kinase domains (Marlowe et al., 2019).

As discussed above, FAK exists in a closed conformation that is inactive due to multiple steric constraints (Fig. 2), another of which is mediated by FAK FERM F2 lobe intramolecular binding to the FAK kinase domain (Lietha et al., 2007). This conformational restraint is released by the FAK FERM F2 lobe binding to phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂], which also brings FAK to membranes (Cai et al., 2008; Goni et al., 2014) or endosomes (Alanko and Ivaska, 2016; Takahashi et al., 2021) (Figs 1 and 2). FAK can become ‘primed’ by first acquiring an ‘open’ conformation, and becomes fully active upon Y576 and Y577 kinase domain phosphorylation.

Alternatively, ezrin protein binding to the FAK FERM F1 lobe (Poullet et al., 2001) or a rise in intracellular pH (Choi et al., 2013) increases FAK Y397 phosphorylation (and FAK activity), likely by causing similar conformational changes. Cryogenic electron microscopy analyses between auto-inhibited and membrane-bound FAK have revealed conformational changes and oligomeric FAK assembly on membranes (Acebron et al., 2020), supporting the notion that FAK signaling is likely mediated by multimeric protein complexes.

Whereas FAK is ubiquitously expressed in mice and in humans, the FAK-related Pyk2 protein tyrosine kinase (encoded by PTK2B) is co-expressed with FAK in distinct cells and tissues (Gil-Henn et al., 2024). Pyk2 possesses similar domain structures and regulatory phosphorylation sites to FAK and binds many of the same adaptor proteins (Fig. 1). However, mouse knockout studies have revealed that Pyk2 does not rescue the motility defects of FAK-null cells in part due to reduced focal adhesion localization and lack of binding to talin family proteins (hereafter talin) compared to FAK (Klingbeil et al., 2001; Lawson et al., 2012). Nevertheless, interpretations of FAK-null phenotypes can be complicated by potential effects of Pyk2 signaling (Sieg et al., 1998; Weis et al., 2008). Although elevated Pyk2 expression and tyrosine phosphorylation can enhance carcinoma, glioblastoma and chronic lymphocytic leukemia tumor growth (Gil-Henn et al., 2024; Sbrana et al., 2023), it remains unclear at a fundamental level how Pyk2-generated signals are distinct from those induced by FAK (Thomas et al., 2019).

FAK expression, FAK Y397 phosphorylation and intrinsic FAK activity are essential for mouse embryogenesis (Heim et al., 2018; Ilic et al., 1995; Lim et al., 2010a). During cell adhesion to matrix proteins, following FAK Y397 cis- or trans-phosphorylation, full catalytic FAK activation occurs upon Src family tyrosine kinase-mediated phosphorylation of FAK at Y576 and Y577 in the kinase activation loop (Lietha et al., 2007). Src-family kinases contain SH2 and SH3 domains, which bind to phosphorylated FAK at Y397 and to a nearby FAK proline-rich motif, respectively. This stabilizes the Src–FAK signaling complex (Fig. 2) (Marlowe et al., 2019; Schaller, 2010), with both proteins bound in open and kinase-active conformations. Notably, many functions of a FAK–Src complex are likely to be Src specific, as Src inhibition prevents the tyrosine phosphorylation of several FAK-bound proteins (Dawson et al., 2021; Mitra and Schlaepfer, 2006; Sulzmaier et al., 2014), as well as FAK Y925 phosphorylation, which acts as a signaling linkage to the mitogen-activated protein kinase cascade through Grb2 (Mitra and Schlaepfer, 2006; Mitra et al., 2006b). FAK Y861 is also phosphorylated by Src, although the function of this phosphorylation remains unclear.

FAK–Src signaling at integrin clusters is reinforced by tension-based FAK activation (discussed below), or alternatively, FAK and Src can be differentially activated by growth factor stimuli (Sulzmaier et al., 2014) (Fig. 3). For example, cytoplasmic FAK is rapidly recruited to endothelial cell–cell contacts upon vascular endothelial growth factor (VEGF) stimulation, which results in cell–cell junction breakdown and increased vascular permeability, which is crucial for the initiation of angiogenesis, achieved in part through vascular endothelial cadherin and β-catenin tyrosine phosphorylation (Chen et al., 2012; Jean et al., 2014). Thus, the role of Src in VEGF signaling is different from FAK (Kim et al., 2009). Additionally, as β-catenin is a central component of the Wingless and Int-1 (Wnt) signaling cascade, FAK-specific activation of β-catenin and Wnt signaling also represents a major pathway supporting neoplastic transformation (Fig. 3) (Chen et al., 2018; Diaz Osterman et al., 2019; Gao et al., 2015; Shang et al., 2019; Worthmuller and Ruegg, 2020). Finally, as revealed by FAK activity-dependent biosensors, temporal differences in FAK and Src activation are observed at adhesions (Wu et al., 2016), further supporting co-operative and overlapping, but non-redundant roles for these kinases.

Advances in the integrin signaling field have revealed that adhesion proteins, such as talin and vinculin, are part of a flexible force transduction unit where integrins and the actin cytoskeleton sense and transmit changes in mechanical forces (Carisey et al., 2013; Grashoff et al., 2010). FAK does not bind directly to integrins and is recruited to newly forming adhesion sites in part through the FAK C-terminal FAT domain binding to paxillin and talin (Kadare et al., 2015). Although FAK FAT binding to talin occurs independently of paxillin (Lawson et al., 2012) (Fig. 1), and is not essential for initial FAK Y397 phosphorylation at nascent adhesions, prevention of FAK and talin binding disrupts force-activated FAK signaling affecting cell migration (Zhou et al., 2021). In addition, reduced FAK–talin binding attenuates Yes-associated protein (YAP; also known as YAP1) nuclear localization and transcriptional activity, providing a signaling linkage between tension and gene transcription (Holland et al., 2024). FAK activity has been shown to increase in a manner that is proportional to the substrate rigidity (0.2 to 40 kPa) in experiments using a phosphorylation biosensor (Seong et al., 2013). Interestingly, FAK-associated mechanosensitive signaling proteins, such as p130Cas (also known as BCAR1) and downstream Rac GTPase activation also transduce matrix stiffness cues into changes in cell cycle progression (Bae et al., 2014). Together, these studies support the notion that FAK signaling can facilitate or regulate crosstalk between tensional, motility and proliferative regulatory signaling pathways.

Using atomic force microscopy and fluorescently tagged FAK proteins, one study revealed that applied tensile force to cells results in stepwise conformational changes in FAK domains (Bauer et al., 2019). Current models propose that FAK protein unfolding and clustering at nascent adhesions leads to FAK activation (Le Coq et al., 2022). However, in cells with mature adhesions, applied tension precedes full FAK activation (Li et al., 2023). These findings are consistent with earlier studies showing that FAK Y397 phosphorylation is greater when breast carcinoma cells are on stiff matrices than when on soft matrices, with stiff matrix enhancing integrin to FAK signaling and tumor progression (Levental et al., 2009). Taken together, FAK activation can precede or follow tensional changes in cells in nascent and mature adhesions, respectively. Moreover, tensional FAK activation is associated with endothelial, cardiac muscle, bone and tumor cells pathologies (as reviewed in Urciuoli and Peruzzi, 2020).

Key cell survival functions of FAK are also mediated via protein interactions with the FAK FERM domain, which contains a nuclear localization sequence in the FAK FERM F2 lobe (Lim et al., 2008) (Fig. 1). Notably, FAK FERM-mediated nuclear localization plays important roles during development and in tumor progression (Frame et al., 2010). During early mouse development, FAK loss is lethal and is associated with p53 tumor suppressor activation, with induction of p21CIP-dependent mesenchymal cell cycle arrest (Ilic et al., 1995). p21CIP1 (also known as CDKN1A) is a cyclin-dependent kinase inhibitor, and primary fibroblast proliferation in the absence of FAK is enabled by p21CIP1 knockout; in these cells, FAK FERM nuclear localization functions as a binding scaffold for p53 and the Mdm2 ubiquitin E3 ligase in the nucleus, resulting in p53 ubiquitylation and degradation (Lim et al., 2008). This nuclear FAK FERM scaffolding function does not require intrinsic FAK activity (Lim et al., 2010a) and is conserved in Pyk2 (Lim et al., 2010b). FAK-FERM binding to different ubiquitin E3 ligases can also modulate levels of other target proteins via proteasomal regulation (Canel et al., 2017, 2023; Jeong et al., 2019; Lim et al., 2012) (Fig. 3). Although FAK contains nuclear import and export consensus motifs, the temporal regulation of nuclear-, cytoplasmic- and adhesion-associated FAK in cells remains unresolved.

FAK signaling (both scaffold and kinase activity) connects to a variety of different downstream intracellular pathways (Fig. 3). In this section, we highlight the broad range of cellular commonalities and phenotypes arising from FAK activation and/or FAK catalytic inhibition.

Both FAK and Pyk2 function as hubs for signaling networks that promote an invasive tumor cell phenotype. FAK is canonically known to promote cell motility at focal adhesions (Mitra et al., 2005) and Pyk2 signaling supports formation of invadopodia, protease-containing cell projections that create stromal escape conduits for tumor cells (Genna and Gil-Henn, 2018; Genna et al., 2018; Mierke et al., 2017). However, in breast carcinoma cells, FAK signaling also enhances matrix metalloproteinase (Hsia et al., 2003; Wu et al., 2005) and urokinase plasminogen activator expression (Mitra et al., 2006a), which both act in the tumor microenvironment to facilitate matrix degradation and cell invasion.

Interestingly, in melanoma cells, preventing FAK activity or inhibiting paxillin binding to FAK reduces cell migration and prevents cell-associated proteolytic activity, but paradoxically increases invadopodia structure formation (Mousson et al., 2021). This reciprocal regulation has also been observed upon FAK depletion in breast carcinoma cells, where FAK knockdown cells exhibit decreased tyrosine-phosphorylated protein abundance at focal adhesions, with increased protein tyrosine phosphorylation at invadopodia (Chan et al., 2009). As such, it is tempting to speculate that upon FAK loss, this reciprocal regulation of adhesion and invadopodia by tyrosine phosphorylation in tumor cells involves Pyk2.

A key regulatory point of tumor metastasis is the ability of tumor cells to survive anoikis following loss of a supportive ECM environment. Facilitating such survival, membrane-targeted FAK promotes cell transformation and anchorage-independent growth, which is dependent on FAK Y397 phosphorylation and kinase activity (Frisch et al., 1996). FAK also activates other signaling pathways, such as phosphoinositide 3-kinase (PI3K) and Wnt signaling pathways, to facilitate cell survival (Zouq et al., 2009). Notably, genetic or pharmacological FAK inhibition selectively promotes tumor cell apoptosis only in anchorage-independent conditions (Tancioni et al., 2014; Tanjoni et al., 2010). Interestingly, FAK activation can also occur in a tumor-intrinsic manner upon cellular adaptive resistance to chemotherapy (Diaz Osterman et al., 2019; Kessler et al., 2019; Taylor and Schlaepfer, 2018). Although the molecular mechanism(s) of FAK activation in response to chemotherapy stress are not known, stress-induced FAK activation supports cell survival (Shapiro et al., 2014), a cancer stem cell-like phenotype (Kolev et al., 2017), and chemotherapy resistance (Diaz Osterman et al., 2019) (Fig. 4).

As well as its cytoplasmic functions, nuclear FAK is also implicated in cancer cell survival and chemotherapy resistance. In vascular smooth muscle and endothelial cells, pharmacological FAK inhibition is associated with FAK nuclear accumulation (Jeong et al., 2022, 2021; Murphy et al., 2023). However, antibodies against phosphorylated FAK Y397 stain both cytoplasmic and nucleoli structures in human breast tumors (Tancioni et al., 2015). Accordingly, wild-type but not kinase-inactive K454R FAK biochemically co-fractionates with nucleoli in breast carcinoma cells grown under anchorage-independent conditions (Tancioni et al., 2015). Together, these results support the notion that FAK nuclear localization is not necessarily associated with inhibition or activation of FAK Y397 phosphorylation. Moreover, as genetically inactive (FAK K454R) FAK can become localized to the nucleus, the processes of FAK activation or nuclear localization are likely independently regulated.

One target of nuclear FAK is the p53 tumor suppressor protein, and this linkage extends to squamous cell carcinoma (SCC) tumors (Pifer et al., 2023). In a mouse SCC model, nuclear FAK activity was found to promote chemokine transcription and subsequent alterations in the tumor-immune microenvironment (Serrels et al., 2015, 2017), contrasting with the absence of FAK in the nuclei of non-transformed keratinocytes. Although specific phosphorylation target(s) of nuclear FAK have not been identified, FAK-dependent changes in chromatin accessibility are associated with altered transcription factor binding to gene regulatory sites (Griffith et al., 2021). Taken together, the control of nuclear FAK distribution is complex and regulatory mechanisms might differ in normal versus transformed cells. Moreover, nuclear FAK signaling in tumors likely has important roles in promoting anoikis and chemotherapy resistance in a manner that does not directly involve signaling from adhesions.

There is a strong link between FAK, VEGF signaling and the regulation of angiogenesis during development and tumor progression (Roy-Luzarraga and Hodivala-Dilke, 2016; Stone et al., 2014). For example, inhibition of FAK activity in breast cancer cells reduces VEGF expression and prevents angiogenesis (Mitra et al., 2006b). Additionally, as FAK inactivation is lethal during development and is associated with blood vessel morphogenesis defects (Lim et al., 2010a), conditional models of FAK inactivation have been developed. In blood vessel endothelial cells (ECs) of adult mice, conditional EC FAK loss supported normal tumor growth and angiogenesis (Weis et al., 2008), whereas in a related EC FAK knockout model, FAK loss resulted in partial inhibitory effects on tumor size and neovascularization, albeit using less-aggressive implanted tumor cells (Tavora et al., 2010).

One explanation for these results is that EC FAK and Pyk2 share overlapping functions. Pyk2 is expressed in FAK-null ECs (Weis et al., 2008), and like FAK, Pyk2 is activated by VEGF and promotes EC sprouting (Shen et al., 2011). Mechanistically, intrinsic FAK activity and FAK Y397 phosphorylation support EC angiogenesis (Pedrosa et al., 2019). FAK Y397 is homologous to Pyk2 Y402 and both serve as Src SH2-binding sites when phosphorylated. Distinguishing EC FAK from Pyk2 function in vivo will likely require the creation of a double FAK and Pyk2 conditional knockout mouse model.

Meanwhile, the use of an EC-specific inducible FAK kinase-dead (KD) knock-in mouse (EC-FAK-KD) has revealed a new role for FAK in facilitating VEGF-stimulated vascular permeability (Chen et al., 2012). VEGF promotes tension-independent FAK activation, FAK localization to EC cell–cell junctions and binding of the FAK FERM domain to the vascular endothelial cadherin (VE-cadherin) cytoplasmic tail. Additionally, direct FAK phosphorylation of β-catenin at Y142 facilitates VE-cadherin–β-catenin dissociation and EC junctional breakdown (Chen et al., 2012). In the EC-FAK-KD model and in breast tumor-bearing mice treated with an oral FAK inhibitor, orthotopic melanoma metastasis was attenuated without effects on tumor growth (Jean et al., 2014). As tumor cell spread through blood vessels requires cell intravasation and extravasation across EC barriers, these results are consistent with the role of EC FAK activity in controlling barrier function. Further, a recent study confirmed that FAK loss in ECs prevents pancreatic tumor metastasis in the presence of the chemotherapeutic drug gemcitabine, without affecting tumor size (Roy-Luzarraga et al., 2022). Taken together, FAK connections to VEGF signaling are both tumor extrinsic and EC intrinsic (Fig. 4).

Despite limited effects of EC FAK knockout or FAK inhibitor treatment on tumor growth, genetic loss of FAK in ECs can sensitize tumors to DNA-damaging therapies (Tavora et al., 2014). This finding was extended to a different EC-FAK-KD model using melanoma tumors (Newport et al., 2022). Additionally, low EC FAK Y397 phosphorylation levels within human breast tumors (i.e. reduced EC-associated protective signals) have been correlated with tumor chemosensitivity and increased patient survival (Roy-Luzarraga et al., 2020). Although it is hypothesized that a FAK protective signal is in part due to changes in EC cytokine production, this link remains incompletely defined.

In pancreatic ductal adenocarcinoma (PDAC), cancer-associated fibroblasts (CAFs) play key roles in tumor progression and chemoresistance, and form a physical ECM barrier limiting immune cell infiltration (Qin et al., 2024). Elevated FAK Y397 phosphorylation in CAFs is a poor prognostic marker for disease-free PDAC patient survival (Zaghdoudi et al., 2020). Expression of the FAK-KD mutant in CAFs reduced PDAC tumor fibrosis with decreased collagen matrix production (Zaghdoudi et al., 2020). Notably, development of resistance to FAK inhibition over time has been shown to be associated with the reversal of an activated CAF phenotype in a mouse PDAC model (Jiang et al., 2020). In a breast cancer model, FAK inactivation in CAFs reduced tumor cell metastasis but not tumor size (Wu et al., 2020), which was associated with alterations in CAF exosome production that impacted tumor cell migration.

In a BRAF-mutated melanoma mouse model, adaptive resistance alterations of CAFs leads to matrix production, remodeling and subsequent FAK activation within tumor cells to promote melanoma survival (Hirata et al., 2015). In addition, in breast cancer, chemotherapy-induced stromal collagen IV upregulation promotes tumor FAK activation and an invasive phenotype (Fatherree et al., 2022). Together, these studies point to a potential reciprocal signaling loop occurring with CAF FAK activation, increased collagen matrix production, and stromal stiffening leading to tensional activation of FAK in tumor cells. However, this type of stromal stiffness linkage to FAK may be tumor type specific.

Early studies evaluating FAK levels in human tumor cell lines noted that several lung, breast and colon carcinoma cells exhibited FAK gene (PTK2) gains (Agochiya et al., 1999). The PTK2 gene, in the Chr8q24.3 locus, is frequently part of a DNA amplification region in breast, uterine, cervical and ovarian female cancers (Kaveh et al., 2016). Genome-wide association studies identified the Chr8q24 region as a high-grade serous ovarian cancer (HGSOC) susceptibility locus that included MYC at Chr8q24.2 (Goode et al., 2010). Notably, over 70% of HGSOC tumors contain gains at both the PTK2 and MYC loci (Diaz Osterman et al., 2019), and gains in Chr8q24 are associated with poor patient survival (Gorringe et al., 2010).

Additionally, PTK2 gene copy number gains parallel increased PTK2 mRNA and FAK protein levels in HGSOC tumors, and elevated PTK2 mRNA levels are associated with decreased overall patient survival (Diaz Osterman et al., 2019). Proteomic analyses have revealed that FAK protein and FAK Y397 phosphorylation are elevated in HGSOC tumors relative to surrounding stromal tissue (Zhang et al., 2016). Interestingly, proliferative serous tubal intraepithelial carcinoma, a precursor of HGSOC, exhibits gains in Chr8q24 (Wang et al., 2024). Additionally, Chr8q24.3-associated circulating tumor DNA is detected in over 75% of plasma blood samples from individuals with HGSOC at the time of diagnosis (Paracchini et al., 2021). Together, these results support the notion that Chr8q24.3 gains and elevated FAK expression occur early in HGSOC tumor initiation and might be considered a potential biomarker associated with aggressive tumors.

ID8 mouse ovarian tumor cells are a highly used syngeneic but slow-growing tumor model (Roby et al., 2000). To select more aggressive cells, ID8 cells were intraperitoneally injected into C57Bl6 mice, re-collected by peritoneal wash after 40 days, and grown ex vivo under anchorage-independent conditions (Ward et al., 2013). Exome DNA sequencing of the recovered cells revealed that there were thousands of nucleotide changes compared to parental cells, but less than 1% of these changes were located in exons and none were predicted to be oncogenic- or tumor suppressor-associated changes (Diaz Osterman et al., 2019). Instead, comparisons revealed gains or losses across murine chromosomes, and the loci for murine Kras, Myc and Ptk2 (FAK) genes exhibited two-to-three-fold gains compared to levels in parental ID8 cells. These in vivo and unbiased selected cells were renamed KMF to denote the Kras, Myc and FAK gene gains that also co-occur in HGSOC tumors (Diaz Osterman et al., 2019). Interestingly, spontaneous gains in regions of Chr15D1-4, the mouse equivalent of Chr8q24, were also detected in fallopian tube-derived tumor cells from genetically engineered mouse models (Maniati et al., 2020). Together, gains in the loci for Myc and FAK, common in murine and human ovarian tumors, are multi-factorial contributors to ovarian tumorigenesis.

Despite the above findings, genetic inactivation of FAK in KMF cells did not alter growth in adherent culture (Diaz Osterman et al., 2019). Instead, comparison of KMF FAK-null and FAK-reconstituted cells showed that FAK expression and activity promotes anchorage-independent cell survival, Wnt–β-catenin signaling and elevated intrinsic resistance to cisplatin chemotherapy (Diaz Osterman et al., 2019). Although these results point to an oncogenic role for FAK, the signals promoting FAK activation in KMF cells and in HGSOC tumors remain incompletely defined. One possibility is that FAK may be activated by alternative splicing. Alternatively spliced FAK transcripts (termed Box 6, Box 7 and Box 28, named for the number of inserted residues around the FAK Y397 site) are present in neuronal tissue transcripts and these FAK isoforms exhibit elevated FAK Y397 phosphorylation compared to wild-type FAK (Toutant et al., 2002). In pancreatic and breast neuroendocrine tumors, a high frequency of FAK mRNAs with alternative splicing are detected and these are associated with elevated FAK Y397 phosphorylation and more aggressive tumors (Xie et al., 2023). In colorectal cancer, FAK Box 6 expression is associated with increased tumor metastasis (Devaud et al., 2019). Thus, increased FAK expression associated with gene gains or alternatively spliced FAK isoforms can be drivers of tumor initiation and progression.

Several different small-molecule ATP-competitive inhibitors to FAK have been developed that exhibit high specificity, can be delivered orally (and in pill form to patients) and exhibit on-target FAK Y397 phosphorylation reduction in tumor and stromal cells (as reviewed by Spallarossa et al., 2022) (Fig. 5). As discussed above, increases in FAK Y397 phosphorylation might not always be reflective of ‘active FAK’. Nevertheless, loss of FAK Y397 phosphorylation parallels reduction in intrinsic FAK activity. FAK inhibitors are cell permeable and act to reduce normal and tumor adherent cell movement, but do not prevent cell proliferation at concentrations of 1 µM and below (Mitra et al., 2005; Slack-Davis et al., 2007).

It is important to note that some of the FAK inhibitors currently being tested in clinical trials exhibit >50-fold selectivity for FAK versus Pyk2. Ifebemtinib (InxMed Inc.) and narmafotinib (Amplia Therapeutics) are selective for FAK, whereas defactinib (Verastem Inc.) equally inhibits both FAK and Pyk2 kinase activities. Small-molecule Pyk2-specific inhibitors have been created, but not characterized in cells (Farand et al., 2016). Neither ifebemtinib nor defactinib prevent adherent tumor cell growth; however, both can negatively impact tumorsphere anchorage-independent cell proliferation and survival (Diaz Osterman et al., 2019; Laszlo et al., 2019; Tanjoni et al., 2010). Despite differences in targeting FAK or FAK–Pyk2, similar phenotypic findings have been observed to date with these inhibitors in cultured cell or mouse tumors (Dawson et al., 2021). In general, FAK and FAK–Pyk2 inhibitors are not cytotoxic drugs.

In addition to targeting tumors, the disruption of tensional signaling using small-molecule FAK inhibitors has revealed potential translational opportunities in promoting tissue regeneration and limiting scar formation. In mice, pharmacological FAK inhibition accelerates wound healing with reduced collagen deposition and fewer scar forming myofibroblasts (Ma et al., 2018). FAK inhibition also enhances skin graft healing in large mammals (Chen et al., 2021) with decreased fibrosis and contracture phenotypes (Chen et al., 2022). In atopic dermatitis, mechanical scratching is both causal for inflammatory skin ulceration, and associated with increased epidermal FAK Y397 phosphorylation (Jia et al., 2023). Pharmacological FAK inhibition delivered in a hydrogel dressing (also containing nanoparticle reactive-oxygen scavengers) reduces dermatitis-associated inflammation and epithelial barrier damage (Jia et al., 2023). Together, these studies highlight the potential clinical promise of inhibiting FAK as a mechano-therapy target.

Recent advancements in chemical FAK inhibitor design have resulted in the generation of proteolysis targeting chimera (PROTAC) FAK degraders (Cromm et al., 2018). FAK PROTACs are chemical molecules designed to post-translationally target FAK by linking ATP-competitive FAK inhibitors with a chemical E3 ligase-binding moiety, such as pomalidomide, to recruit cereblon E3 ligase or a chemical structure to promote von Hippel–Lindau (VHL) E3 ligase binding (Fig. 5) (as reviewed by Chirnomas et al., 2023; Spallarossa et al., 2022). PROTACs are dependent on E3 ligase expression in target cells for target protein degradation and the chemical destruction of FAK should block both kinase and scaffolding signaling activities. In general, although early FAK PROTAC compound development identified molecules that inhibit tumor cell motility and invasion at nanomolar concentrations (Cromm et al., 2018), other FAK PROTAC compounds showed that FAK knockdown had no effect on tumor cell growth in vitro (Gao et al., 2020a; Popow et al., 2019). As FAK suppresses anoikis, this is best measured by anchorage-independent growth assays. Additionally, few of the published FAK PROTAC studies measured effects on Pyk2, which might function in parallel to FAK (as discussed above). Moreover, extended FAK PROTAC administration in mice has shown that these compounds might accumulate in tissues, with associated toxicity (Gao et al., 2020b).

However, recent studies have shown that FAK PROTAC addition to tumor cells in vitro at sub-micromolar concentrations recapitulates many of the FAK inhibitor-associated cell phenotypes (Hou et al., 2023; Law et al., 2021). Looking toward the future, FAK PROTAC modifications are still required to optimize oral delivery, cell penetration, pharmacokinetics and off-target toxicity (Koide et al., 2023). As genetic tumor models have revealed important differences resulting from FAK loss versus inactivation of FAK activity (Dawson et al., 2021), FAK PROTAC compounds hold great promise as tools to decipher consequences of FAK or FAK–Pyk2 inhibition versus loss of expression.

FAK activation can occur downstream of oncogenes and in response to chemotherapy or environmental stress, which makes FAK an attractive target for several different cancers. In PDAC mouse models, oral FAK inhibitor administration enhances the cytotoxic effects of gemcitabine and increases responsiveness to combined anti-checkpoint receptor immunotherapy (Jiang et al., 2016; Osipov et al., 2021). FAK inhibition is also associated with anti-fibrotic effects on the pancreatic tumor stroma, changes that also enhance the effects of radiation therapy (Lander et al., 2022). Notably, in human PDAC tumors, FAK Y397 phosphorylation increases with disease progression and is greater in stromal compartments compared to tumor cells (Murphy et al., 2021; Zaghdoudi et al., 2020). In models of KRAS G12C oncogenic-driven non-small cell lung carcinoma, orally delivered FAK inhibitors radio-sensitize tumors (Tang et al., 2016), synergize with inhibitors of KRAS G12C (Zhang et al., 2021) and boost checkpoint inhibitor immune responses (Qiao et al., 2024) via tumor intrinsic and extrinsic mechanisms.

This type of intrinsic–extrinsic multi-factorial FAK inhibition has also been observed in ovarian cancer models whereby gains in FAK expression, FAK Y397 phosphorylation and FAK activity enhance intrinsic cisplatin and taxane resistance in part through transcriptomic changes (Diaz Osterman et al., 2019; Kang et al., 2013). In primary ovarian tumors, changes in extrinsic stromal matrix composition and tissue stiffness are associated with tumor FAK activation and cisplatin chemoresistance (Pietila et al., 2021). Moreover, tumor intrinsic FAK inhibition in mouse ovarian or SCC models results in enhanced T and B cell recruitment, potentiated activity of immune checkpoint antibodies and formation of tertiary lymphoid structures, which are known markers of immune activation (Canel et al., 2020; Ozmadenci et al., 2022; Serrels et al., 2015, 2017). In a Ras- and p53-mutated mouse PDAC model, loss of tumor FAK results in increased antigen processing and presentation that might impact immune responses (Canel et al., 2023). In summary, these studies show that FAK activation can impact (and be activated by) the tumor microenvironment and that FAK inhibition uncovers several tumor vulnerabilities altering immune recognition and cell survival.

Over the past 10 years, several early phase clinical trials have tested small-molecule FAK inhibitors from different companies (Dawson et al., 2021; Sulzmaier et al., 2014). These trials revealed that daily dosing for extended periods of time is safe and associated with low levels of manageable adverse events, and that small-molecule FAK inhibitors in general had good pharmacokinetic properties (de Jong et al., 2014; Jones et al., 2015; Wang-Gillam et al., 2022). However, in early phase II trials for malignant pleural mesothelioma and PDAC, the clinical endpoints were not met (Aung et al., 2022; Fennell et al., 2019). In unselected mesothelioma patients, defactinib (as maintenance after primary standard-of-care chemotherapy) was terminated early due to lack of efficacy (Fennell et al., 2019). Conversely, in a clinical trial of meningioma patients with somatic loss of the neurofibromatosis type II (NF2) gene, FAK inhibitor monotherapy extended progression-free survival at six months in grade 1–3 tumors compared to historical controls (Brastianos et al., 2022). This trial was testing the hypothesis that NF2 loss creates a synthetic lethal relationship with FAK inhibition (Shapiro et al., 2014). Additionally, partial responses to defactinib monotherapy were reported in meningioma patients with NF2 mutations as part of the NCI-MATCH trial (NCT04439331).

Despite limited effectiveness of defactinib monotherapy, several phase II trials are testing the role of defactinib FAK inhibition as part of combination chemotherapies (Table 1). The combination of defactinib with the RAF/MEK clamp avutometinib (VS-6766) was tested in patients with KRAS mutant lung or ovarian cancers (NCT03875820). Early phase II results in low-grade serous ovarian cancer (LGSOC) showed that a defactinib and avutometinib combination elicited an overall response rate (ORR) of 46% in 11 of 26 evaluable patients, which was 2–8-fold greater than the historical ORR in this pre-treated patient population.

LGSOC is a rare subtype that is histologically, molecularly and clinically distinct from HGSOC. Molecularly, LGSOC is characterized by chemoresistance, estrogen and progesterone receptor positivity, activation of the MAPK pathway and wild-type p53 expression (Cobb and Gershenson, 2023). Verastem was given ‘breakthrough therapy’ designation by the FDA in 2021 for the defactinib and avutometinib combination, which allows for expedited review of drugs for serious life-threatening conditions. Preliminary results for the Raf and MEK Program (RAMP201) phase II trial (NCT04625270) in LGSOC, testing the combination of defactinib and avutometinib, reveal an ORR of 45% (13/29) and tumor shrinkage in 86% of evaluable patients receiving combination treatment. A confirmatory international phase III (RAMP301, NCT06072781) trial has recently opened to recruitment, with patient enrollment coinciding with an expected application for FDA approval.

In 2019, InxMed Inc. purchased the rights to the FAK inhibitor BI853520 (from Boehringer Ingelheim) and renamed this compound ifebemtinib (InxMed, IN10018). A single arm phase I-II trial was subsequently started, testing ifebemtinib with pegylated doxorubicin (Doxil) in recurrent platinum-resistant HGSOC (NCT05551507) (Table 1), the most lethal gynecologic malignancy in the United States, with a historical response rate of less than 20%. As of April 2023, 54 evaluable patients showed one complete response, 21 partial responses, and 20 patients with stable disease for an ORR of 46% (Zhang et al., 2024). Currently, a randomized, double-blind phase II trial is evaluating ifebemtinib and Doxil versus placebo and Doxil in platinum-resistant recurrent HGSOC (NCT06014528).

Cells are exposed to a variety of stimuli on a constant basis, including mitogenic and chemotactic factors, death signals and chemical insults, all delivered against a backdrop of various biomechanical alterations. In consideration of this complexity, it is believed that intermediary proteins function to contextualize and integrate discrete extracellular signals to elicit appropriate intracellular responses. FAK functions in this manner as a signal integration hub and as a master regulator of drug resistance. This involves FAK kinase-dependent and -independent signals, as well as FAK localization to focal adhesions, cell–cell contacts, endosomal membranes and in the nucleus. Herein, we have touched on how FAK plays an important signal integration role and ultimately functions to guide cellular behavior.

The ability of small-molecule FAK inhibitors to block kinase activity, yet permit scaffolding functions, might explain their relatively limited toxicity in vivo. The regulation of FAK activity represents an actionable target in a broad set of pathological conditions, ranging from tumors to stromal tissues, which includes targeting CAFs and ECs as well as altering tensional pathologies, such as wound healing. By contrast, FAK inhibitors have limited or possibly stimulatory effects on tumor-responsive immune cells. This range of targets, and limited toxicity of FAK inhibitors, might present a powerful tool to influence the physiological response to other therapeutic approaches. Indeed, after more than 30 years since the discovery of FAK, a clinical approval might finally be on the near-term horizon.

We thank Jonathan Pachter (Verastem) and Zaiqi Wang (InxMed) for reviewing and Daniel Lietha for comments.