Nuclear FAK Controls Chemokine Transcription, Tregs, and Evasion of Anti-tumor Immunity

1First described more than a decade ago, regulatory T cells (Tregs) have become recognized as a core component of the immuno-suppressive armory utilized by many tumors to keep the anti-tumor activity of antigen-primed CD8+ T cells at bay. Increased Treg numbers has been associated with poorer survival in ovarian, gastrointestinal, and esophageal cancer. Indeed, the ratio of CD8+ T cells/Tregs correlates with poor prognosis, shifting the balance from anti-tumor immunity toward tumor tolerance. Through secreting a range of chemokines and cytokines, cancer cells can promote the recruitment of Tregs into tumors and can also facilitate their peripheral expansion and retention. Thus, Tregs can act as a barrier to effective immune-based therapy aimed at activation of a CD8+ T cell anti-tumor immune response. However, the specific signals within tumor cells that stimulate elevated intra-tumoral Tregs, giving rise to tumor tolerance, remain elusive.

2FAK is a tyrosine kinase that regulates diverse cellular functions, including adhesion, migration, invasion, polarity, proliferation, and survival. Using targeted gene deletion in mouse skin, we have previously shown a requirement for fak in tumor initiation and progression to malignant disease. FAK is also required for mammary tumor progression, intestinal tumorigenesis, and the androgen-independent formation of neuroendocrine carcinoma in a mouse model of prostate cancer. Expression of FAK is elevated in a number of tumor types (reviewed in ), and FAK inhibitors are being developed as potential cancer therapeutics. Many of FAK’s functions in cancer are via its role in signaling downstream of integrins and growth factor receptors at the plasma membrane. FAK also contains putative nuclear localization sequences (NLS) within the F2 lobe of its FERM domain and can localize to the nucleus upon receipt of cellular stress, where it binds to p53. However, the extent of FAK’s nuclear functions remains largely unknown. Here, we report a function for nuclear FAK in regulating transcription of inflammatory cytokines and chemokines, in turn promoting an immuno-suppressive, pro-tumorigenic microenvironment. This is mediated by recruitment and expansion of Tregs via FAK-regulated chemokine/cytokine networks, and we have found an important role for Ccl5 and TGFβ2. Therefore, FAK controls the tumor environment, and suppressing FAK activity, including via a clinically relevant FAK inhibitor, may be therapeutically beneficial by triggering immune-mediated tumor regression.

3We used a syngeneic model of SCC in which the fak gene had been deleted by Cre-lox recombination and mutant tumor cell lines generated. We monitored tumor growth following injection of 1 × 106 FAK-deficient cells (FAK−/−) or FAK-deficient cells that re-expressed wild-type FAK (FAK-WT) at comparable levels to endogenous FAK in both CD-1 nude and FVB (syngeneic) host mouse strains. In CD-1 nude mice, SCC FAK−/− tumor growth was characterized by a modest growth delay (Figure 1A) as reported previously. By contrast, in FVB mice, SCC FAK−/− tumor growth was characterized by an initial period of growth in the first 7 days followed by complete regression by day 21 (Figure 1B). Thus, FAK expression is required for the survival and growth of SCC tumors in FVB mice with a functional adaptive immune system.

4To characterize the role of adaptive immunity in FAK−/− SCC tumor regression, we used antibody-mediated T cell depletion in animals bearing FAK−/− tumors (Figures 1C and S1). Depletion of CD4+ T cells had no effect on tumor growth. In contrast, depletion of CD8+ T cells, either alone or in combination with CD4+ T cells, restored SCC FAK−/− tumor growth. This implies that cytotoxic CD8+ T cells were responsible for regression of FAK−/− tumors (Figure 1C) but does not exclude an accessory role for CD4+ T cells. T cell depletion in mice bearing SCC FAK-WT tumors (Figure 1D) revealed that: (1) depletion of CD8+ T cells, either alone or in combination with CD4+ T cells, caused a significant increase in tumor growth when compared to isotype-treated controls at day 14, and (2) depletion of CD4+ T cells alone caused regression of FAK-WT SCC tumors by day 21. This implied that FAK-expressing tumors were also under negative pressure from the immune system and that cells from the CD4+ T cell compartment play a role in protecting FAK-WT tumors from immune-mediated regression (reason discussed later; Figure 3).

5Next, we re-challenged mice with 1 × 106 SCC FAK-WT cells after regression of primary FAK−/− SCC tumors, following 7 days of tumor-free survival after the tumors had regressed (Figure 1E, top and middle graphs). Neither FAK-deficient nor FAK-expressing SCC cells were able to grow after the mice had been pre-challenged with SCC FAK−/− cells. As controls, SCC FAK-WT and FAK−/− cells were injected at day 28 into mice with no pre-challenge, and these grew as expected (Figure 1E, bottom). This implies that, following FAK−/− SCC tumor regression, host mice remain immunized against further tumor challenge because immunological memory had been established. It is possible that either broad immunization against SCCs may have occurred or, more likely, that the FAK−/− and FAK-WT SCCs shared common antigen(s) that are expressed irrespective of FAK status. We conclude that FAK enables SCC cancer cells to suppress an adaptive immune response rather than to circumvent it through evading recognition per se. SCC FAK−/− cells in which a FAK kinase-deficient mutant was re-expressed (SCC FAK-KD) initially grew and then regressed with kinetics that were only modestly delayed when compared to FAK−/− cells, indicating that immune suppression depends on FAK kinase activity (Figure 1F).

6We next investigated the nature of the T cell response within tumors derived from all three SCC cell lines using FACS analysis on disaggregated tumor tissue taken at day 7. We did not observe a significant change in the percentage of total CD4+ T cells (Figures 2A and S2 and Table S2) or the percentage of CD4+ T cells that expressed the activation marker CD69 (Figure 2B). In contrast, we did observe a significant increase in the proportion of effector CD4+CD44hiCD62Llow T cells in SCC FAK−/− and FAK-KD tumors when compared to FAK-WT tumors (Figures 2C and S2 and Table S2). Analysis of tumor-infiltrating CD8+ T cells revealed a significant increase in SCC FAK−/− and SCC FAK-KD tumors when compared to SCC FAK-WT tumors (Figures 2D and S2 and Table S2), indicative of a heightened cytotoxic anti-tumor immune response. Staining with the activation marker CD69 identified the presence of CD8+CD69+ T cells in all tumors (Figure 2E). Further analysis revealed an increase in percentage of effector CD8+CD44hiCD62Llow T cells in SCC FAK−/− and SCC FAK-KD tumors when compared to SCC FAK-WT tumors (Figures 2F and S2 and Table S2), especially when effector CD8+ T cell numbers were normalized to account for the observed changes in total CD8+ T cells and presented as a “fold change” (Figure 2G). However, while SCC FAK−/− and SCC FAK-KD tumors had increased effector CD8+ T cells, there were activated CD8+ T cells present in all of the SCC tumors, raising the question of why SCC FAK-WT tumors do not succumb to the cytotoxic CD8+ T cell response.

7It is now established that not only the quantity of tumor-infiltrating CD8+ T cells is important, but also their “quality.” Tumor-induced T cell exhaustion has been reported in a number of tumor types, including melanoma and ovarian cancer, and is characterized by expression of co-inhibitory surface receptors, including programmed death receptor 1 (PD-1), lymphocyte-activation gene 3 (LAG-3), and T cell immunoglobulin mucin-3 (Tim-3), either alone or in combination. Analysis of these markers on antigen-primed CD8+CD44hi T cells infiltrating SCC FAK-WT, FAK−/−, and FAK-KD tumors revealed increased surface expression of PD-1, LAG-3, and Tim-3 in CD8+CD44hi T cells present in SCC FAK-WT tumors (Figures 2H–2J). Together, our data imply that antigen-primed CD8+CD44hi T cells infiltrating SCC FAK-WT tumors exhibit a heightened state of exhaustion indicative of a dysfunctional T cell response. Linked to their exhausted state, there was also evidence of decreased proliferation of CD8+ T cells isolated from SCC FAK-WT tumors (judged by Ki-67 staining in Figure 2K).

8Histological staining of tumor sections taken at day 7 revealed that: (1) CD8+ T cells are present throughout all tumors, and (2) while CD8+ T cells infiltrating SCC FAK-WT tumors appear predominantly as individual cells, CD8+ T cells infiltrating SCC FAK−/− and FAK-KD tumors are clustered (Figure 2L). Thus, the ability of SCC FAK-WT tumors to evade the anti-tumor immune response is not due to limited CD8+ T cell penetration into these tumors.

9Macrophages, myeloid-derived suppressor cells (MDSC), and Tregs with intrinsic immuno-suppressive capabilities can promote tumor development by inhibiting cytotoxic CD8+ T cell activity in mouse and humans. Flow cytometric analysis revealed no differences in macrophage or MDSC populations that correlated with tumor regression (Figures 3A, 3B, S3, and S4 and Table S2), although this does not rule out an accessory role for these cells in eventual tumor clearance. However, we did find a significantly greater number of CD4+FoxP3+CD25+ Tregs in SCC FAK-WT tumors (Figures 3C and S4 and Table S2) when compared with FAK−/− and FAK-KD tumors (Figure 3C). Tregs have been associated with the development of CD8+ T cell exhaustion and may therefore be linked to the CD8+ T cell exhaustion that we observed in SCC FAK-WT tumors (Figures 2H–2J). We next calculated the ratio of CD8+ T cells to Tregs (Figure 3D), as this has been reported to correlate with poor prognosis in a number of tumor types. We found a substantially lower CD8+ T cell to Treg ratio in SCC FAK-WT tumors when compared to SCC FAK−/− and SCC FAK-KD tumors, which correlated with outcome in terms of tumor tolerance versus immune-mediated tumor regression.

10We next examined SCC FAK-WT tumor growth in animals treated with an anti-CD25 antibody to deplete Tregs (Figure 3E). Depletion of CD25+ cells led to regression of SCC FAK-WT tumors. Therefore, FAK-dependent Tregs are required for the growth of FAK-WT-expressing tumors by creating an immuno-suppressive environment that impairs cytotoxic CD8+ T cell activity. This role of CD4+ Tregs is the likely reason for effects of the CD4-depleting antibody in promoting regression of SCC FAK-WT tumors (Figure 1D). We note that high Treg levels have been reported in a number of solid tumor types and that elevated Tregs are linked to poor clinical outcome.

11We demonstrated that Tregs derived from SCC FAK-WT tumors expressed the transcription factor (TF) Helios (Figure S5A), indicative of thymic origin. Thus, we hypothesized that FAK may drive the recruitment and expansion of the intra-tumoral Tregs by influencing the availability of secreted factors.

12To address how FAK activity in SCC cancer cells promotes elevated intra-tumoral Tregs, we next analyzed global transcriptional profiles of SCC FAK-WT and SCC FAK−/− cells using Affymetrix GeneChip microarrays (Figure 4A). FAK expression resulted in the upregulation of 498 genes and the downregulation of 598 genes (p < 0.01). The upregulated transcript set in SCC FAK-WT cells was associated with a number of processes, including cell migration, receptor binding, secretion, wounding, and ovulation (Figure 4B, top). Analysis of this gene set revealed the chemokine ligand group of genes to be significantly overrepresented (Figure 4B, bottom), which is interesting given that a number of these chemokines and cytokines mediate both Treg recruitment to tumors and induction of peripheral Tregs within tumors.

13To establish which chemokines and cytokines were regulated by FAK and to address whether the FAK-dependent transcriptional profile was linked to chemokine receptor expression on tumor-infiltrating Tregs, we performed quantitative (q)RT-PCR array analysis. Comparison of chemokine/cytokine transcript levels between SCC FAK-WT and SCC FAK−/− cells revealed a subset of ligands increased >2-fold in SCC FAK-WT cells (Figure 4C). Several of these (Ccl1, Ccl5, Ccl7, Cxcl10) have roles in Treg recruitment (green arrowheads, Figure 4C), while one (Tgfb2) has a reported role in peripheral induction and expansion of Tregs (red arrowhead, Figure 4C). To complement this, comparison of Tregs isolated from the thymus of normal FVB mice with those isolated directly from SCC FAK-WT tumors revealed a chemokine receptor switch (Figure 4D). We found increased expression of the cognate receptors for five of the six chemokine ligands upregulated in SCC FAK-WT cells (Figure 4C). These receptor changes may represent a switch from lymphoid homing receptors, including Ccr7 and Cxcr4, toward expression of memory/effector-type chemokine receptors, including Ccr2, Ccr5, Ccr8, and Cxcr6, involved in recruitment to non-lymphoid tissues and sites of inflammation. Network analysis of the relationship between FAK-dependent chemokine ligand expression in SCC cells and tumor-infiltrating Treg chemokine receptor expression revealed the existence of a FAK-dependent paracrine signaling axis between cancer cells and intra-tumoral Tregs based on chemokine ligand-receptor interactions (Figure 4E). Furthermore, (q)RT-PCR analysis of Ccl5, Cxcl10, and Tgfb2 demonstrated that their expression was dependent on FAK kinase activity (Figure 4F). We note that disruption of the Ccl5/Ccr5 axis in a model of pancreatic adenocarcinoma results in reduced intra-tumoral Tregs and slows tumor growth, implying that FAK-dependent regulation of this paracrine signaling axis may be more generally important. Thus, FAK activity regulates the expression of a subset of chemokines that can specifically mediate crosstalk between tumor cells and tumor-infiltrating Tregs. This likely has importance in recruitment and retention of CD4+FoxP3+CD25+ Tregs into SCC FAK-WT tumors.

14The finding that the Tregs enriched in SCC FAK-WT tumors were likely recruited into SCC FAK-WT tumors led us to consider a potential role for Ccl5 that has been implicated in the recruitment and expansion of CD4+FoxP3+CD25+ Tregs, via the paracrine signaling axis that we identified. We found that efficient knockdown of Ccl5 using two independent shRNA hairpins (P1 and P2, Figure 5A) resulted in SCC FAK-WT shRNA-Ccl5 tumor regression by days 21–27 (Figure 5B). We measured the absolute number of Tregs in SCC FAK-WT shRNA-Ccl5 tumors at day 7 and found that there was a substantial reduction in both Ccl5-depleted tumors when compared with empty vector control SCC FAK-WT pLKO tumors (Figure 5C).

15Expanding on these findings, shRNA-mediated knockdown of Tgfb2 expression in SCC FAK-WT cells also influenced tumor growth (Figures S5B and S5C). Partial knockdown of TGFβ2 had complex effects, which resulted in one of two outcomes. One group (Figure S5C, dashed blue line), grew more rapidly and ulcerated, leading to removal from study at day 14. In the other group that did not display such frank ulceration, we observed tumor regression by day 27 (Figure S5C, dashed red line). Analysis of Treg levels in SCC FAK-WT shRNA-TGFβ2 tumors at day 7 (regardless of initial growth characteristics) revealed that TGFβ2 knockdown was also associated with a reduction in CD4+FoxP3+CD25+ Tregs (Figure S5D). Therefore, while the effects of reducing TGFβ2 expression are more complicated than for Ccl5, FAK-dependent TGFβ2 expression does contribute to elevated CD4+FoxP3+CD25+ Tregs in SCC FAK-WT tumors; and in the subset of mice bearing tumors that were able to complete the study, TGFβ2 knockdown also caused tumor regression.

16Our findings that FAK regulated the transcription of cytokines and chemokines (including Ccl5 and TGFβ2) that were associated with elevated intra-tumoral Tregs and tumor tolerance led us to consider a possible role for nuclear FAK in regulating the transcription of these genes. Based on previous reports, which identified putative NLSs within the FERM domain of FAK, we constructed an optimally nuclear targeting-impaired mutant FAK by replacing two arginines (positions 177 and 178) and four lysines (positions 190, 191, 216, and 218) with alanines (termed FAK-NLS). Western blotting of cytoplasmic and nuclear fractions confirmed that the FAK-NLS mutant was indeed defective in nuclear localization (Figure 5D). Subsequent (q)RT-PCR analysis of Ccl5 and Tgfb2 expression in SCC cells expressing only FAK-NLS revealed that FAK nuclear localization was required for transcription of these genes (Figures 5E and S5E, respectively). Thus, nuclear FAK drives the transcription of Ccl5 and TGFβ2, which are required for recruitment and expansion of immuno-suppressive Tregs into SCC tumors, altering the balance between CD8+ T cells and Tregs in favor of tumor tolerance. In support of this, growth of SCC FAK-NLS tumor cells was similar to that of SCC FAK−/−, with ultimate tumor regression (Figure 5F). This confirmed that it was nuclear FAK that afforded protection from the anti-tumor immune response. Western blotting of cytoplasmic and nuclear fractions from SCC FAK-KD showed that the kinase-deficient mutant was able to localize to the nucleus, so we conclude that the immune modulatory effects of FAK are dependent on FAK kinase activity in the nucleus (Figure 5G).

17We next examined nuclear FAK levels in primary skin keratinocytes, the normal cellular counterparts of the SCC cells used here, and did not find detectable nuclear FAK (Figure 5H). Thus, abundant nuclear localization, and therefore the capacity to exert regulatory control over chemokine and cytokine expression, is likely a feature of oncogenic transformation in skin keratinocytes. This suggests that the nuclear functions of FAK that we have identified—namely, regulating transcription of chemokine/cytokine networks—may be associated with the cancerous state when FAK is highly expressed.

18Having established an important role for the nuclear FAK-dependent transcription of Ccl5 in mediating recruitment and expansion of intra-tumoral Tregs, we wanted to determine how nuclear FAK could exert control over Ccl5 transcription. Using sucrose gradients, we fractionated the nuclei of SCC FAK-WT cells and demonstrated that nuclear FAK was present in the chromatin-containing fraction (Figure 6A). Transcriptional regulation of Ccl5 is mediated predominantly through six short regulatory elements contained within a region of the Ccl5 promoter spanning ∼300 base pairs. These regulatory elements contain binding sites for a number of TFs, including AP-1, C/EBP, IRF-1, NF-κB, and TATA box-binding protein (TBP), which is part of the transcription factor IID complex (TFIID). Using FAK immunoprecipitation and quantitative label-free mass spectrometry, we identified FAK binding partners in purified nuclear extracts and contextualized these by mapping onto a network of proteins associated with predicted Ccl5 TFs (constructed in silico; Figure 6B). This integrative approach identified a subset of Ccl5 TFs and regulators of these that interact with FAK in SCC cell nuclei (Figures 6C, S6 and Table S1). Interaction network analysis of this protein subset revealed nuclear FAK binding partners with roles in multiple transcriptional pathways, including regulators of AP-1, C/EBP, IRF-1/-7, NF-κB/Rel, and TFIID. Thus, we identified nuclear FAK binding partners that can interact, directly or indirectly, with five of the six main regulatory elements reported to control transcription of Ccl5 in multiple cell types. Given that our interaction network was somewhat dominated by proteins associated with the TFIID pathway, including three TBP-associated factors (TAFs) (Figures 6C and S6), we used co-immunoprecipitation to confirm the interaction of nuclear FAK with one of these, TAF9, a core component of the TFIID complex (Figure 6D). Our data show that FAK binds to core components of the transcriptional machinery, many of which are known to be located on the promoter of genes undergoing active transcription and that are known or predicted to regulate Ccl5. Therefore, in SCC cells, nuclear FAK associates with chromatin and is physically linked to a network of TFs and their regulators known to modulate Ccl5 expression.

19Therapeutic targeting of FAK kinase activity using small-molecule inhibitors will inhibit FAK signaling not only in tumor cells, but also potentially in multiple host cell types. To complement expression of the FAK-KD mutant protein in the cancer cells and investigate whether a FAK inhibitor could induce immune-mediated regression of SCC tumors, we used the FAK/Pyk2 kinase inhibitor VS-4718, which is currently in clinical development. Mice were treated with VS-4718 at 75 mg/kg for 24 hr prior to injection of 1 × 106 FAK-WT or FAK−/− SCC tumor cells and twice daily thereafter. This resulted in VS-4718-induced regression of SCC FAK-WT tumors by day 24 (Figure 7A). Following cessation of VS-4718 treatment, no tumor regrowth was observed (data not shown). SCC FAK−/− tumor growth and clearance was not greatly affected by VS-4718 treatment, suggesting that the anti-tumor effects of VS-4718 can be explained by FAK inhibition in tumor cells. Activity of VS-4718 was confirmed using an ELISA to measure FAK autophosphorylation on tyrosine-397 in tumor lysates from mice treated with 75 mg/kg VS-4718 (Figure S7). Regression of VS-4718-treated SCC tumors was not accompanied by loss of cell viability at day 7, as measured by FACS using a viability stain following tumor disaggregation (Figure 7B). There was a significant but small increase in leukocytes in VS-4718-treated SCC FAK-WT tumors (Figure 7C) and a significant increase in total CD4+ T cells (Figures 7D and S2 and Table S2) and effector CD4+CD44hiCD62Llow T cells (Figures 7E and S2 and Table S2). A significant increase in CD8+ T cells was also evident in SCC FAK-WT VS-4718-treated tumors (Figures 7F and S2 and Table S2), although there was no change in effector CD8+CD44hiCD62Llow T cells (Figures 7G and S2 and Table S2). Crucially, there was a significant reduction in CD4+CD25+FoxP3+ Treg cells in VS-4718-treated SCC FAK-WT tumors, which was similar to that observed in vehicle and VS-4718-treated SCC FAK−/− tumors (Figures 7H and S4 and Table S2).

20Thus, VS-4718 promoted robust anti-tumor activity, with similar immune cell changes to that observed upon FAK deletion or expression of a kinase-deficient form of FAK. Furthermore, anti-tumor efficacy of VS-4718 was also dependent on CD8+ T cells, and SCC FAK-WT tumors treated with VS-4718 on a CD8+ T cell-depleted background exhibited a growth delay but did not undergo tumor regression (Figure 7I). We conclude that the FAK kinase inhibitor targets mechanisms of immune suppression and may therefore represent a form of effective “immuno-modulatory” therapy that reduces Tregs in the tumor environment. Importantly, the FAK kinase inhibitor does not affect the cytotoxic function of antigen-primed CD8+ T cells. We also found that VS-4718 treatment that was initiated 5 days post-inoculation of 1 × 106 SCC FAK-WT cells, when these had already formed palpable tumors of ∼50 mm3, led to complete tumor regression (Figure 7J).

21We show that nuclear FAK in SCC cancer cells drives the transcription of chemokines and cytokines, including Ccl5 and TGFβ2, which promote the formation of an immuno-suppressive, pro-tumorigenic microenvironment. This is dependent on FAK kinase activity, and expression of a catalytically inactive mutant FAK protein (FAK-KD) or treatment with a small-molecule inhibitor causes tumor regression. This is effective even when tumors are already established, raising the exciting possibility that targeting of FAK kinase activity may have immune-mediated anti-tumor efficacy in patients. We established that nuclear FAK is associated with chromatin and interacts with a number of TFs and transcriptional regulators, including components of the TFIID complex, that are linked to regulation of Ccl5 expression. Our data imply that FAK interacts with core transcriptional machinery to influence gene transcription and promote tumor immune escape.

22Historically, FAK has been recognized as an adhesion-related non-receptor protein tyrosine kinase that clusters at focal adhesion (FA) structures and regulates cancer-associated processes, including adhesion, migration, invasion, survival, and proliferation (reviewed in ). FAK was also found to translocate to the nucleus, leading to the idea of nuclear functions for FAK within the nucleus. Our data show that, at least in cancer cells, FAK regulates inflammatory transcriptional programs associated with generation and maintenance of a pro-tumorigenic and immuno-suppressive microenvironment. FAK associates with chromatin, and in the context of Ccl5 expression, it interacts with a number of TFs, and regulators of TFs, that bind regulatory elements in the Ccl5 promoter. Our data imply that FAK exists in complexes with a number of TAF proteins, including TAF9 and TAF12, key components of the core promoter complex TFIID that serves to initiate transcription by driving recruitment of chromatin remodeling complexes, coactivators, and RNA polymerase II to the promoter. Therefore, FAK interacts with components of the core transcriptional machinery in order to drive transcription of chemokines and cytokines that contribute to recruitment of Tregs into the tumor environment, promoting immunological tolerance and permitting tumor growth.

23Recently, nuclear accumulation of active FAK (phosphorylated on Tyr-397) within tumor cells of patients with colorectal cancer was reported to correlate with poor prognosis, highlighting the need to understand the nature of FAK’s role within the nucleus. Studies using endothelial cells, muscle cells, and fibroblasts have previously reported low steady-state levels of nuclear FAK that are substantially increased in response to cellular stress. Our work implies that oncogenic stress is another route to inducing high levels of nuclear FAK and that this, in turn, can influence transcriptional programs, such as the chemokine and cytokine networks that control the tumor microenvironment.

24A number of therapeutic strategies targeting components of the immuno-suppressive tumor microenvironment are currently being tested, with the aim of restoring anti-tumor immunity by releasing the break on CD8+ T cell cytotoxic activity. In pre-clinical models of cancer, targeting Tregs has shown anti-tumor efficacy, either alone or when used in combination with agents that enhance CD8+ T cell activation. A clinical study combining agents targeting cytotoxic-T-lymphocyte-associated antigen 4 (CTLA-4), which is thought to influence Treg function, and PD-1, which blocks signals that inhibit T cell function, has reported impressive responses in patients with advanced melanoma. However, this combination of checkpoint blockade antibodies elicits substantial side effects in >50% of patients, highlighting the need to find alternative combinations with improved tolerability. We have shown that targeting FAK kinase activity has the potential to modulate intra-tumoral Treg levels, resulting in robust CD8+ T cell anti-tumor immunity, while others have reported previously that FAK kinase inhibitors block monocyte/macrophage and cancer-associated fibroblast recruitment into tumors by virtue of FAK’s role in regulating their migration. Taken together, these findings suggest that targeting the pleiotropic cellular functions of FAK may have a broad impact on the immuno-suppressive tumor microenvironment, differentiating these agents from many therapeutic approaches that target single immune cell populations.

25Targeting a molecular pathway that is upregulated in cancer cells may provide tumor specificity and help to overcome some of the potential issues with severe autoimmunity when modulating immune cell populations. FAK inhibitors, such as VS-4718, are in clinical development. VS-4718 is currently in a phase I dose escalation clinical trial in patients with solid tumors (www.clinicaltrials.gov NCT01849744). Our findings provide good rationale for pre-clinical and clinical testing of FAK kinase inhibitors alongside agents that stimulate CD8+ T cell activity, such as the checkpoint blockade therapies that target PD-1 and CTLA-4, which are both in clinical development.

26Experiments involving animals were carried out in accordance with the UKCCCR guidelines by approved protocol (HO PL 60/4248). Brief experimental procedures are listed here. For details, please see the Supplemental Experimental Procedures.

27Mutations were introduced into FAK-WT at R177A, R178A, K190A, K191A, K216A, and K218A using PCR-based site-directed mutagenesis.

28Isolation and generation of the FAK SCC cell model is described in. Keratinocyte cultures were prepared as detailed in.

29To prepare whole-cell lysates, cells were washed in cold PBS and lysed in RIPA buffer. Cytoplasmic and nuclear extracts were prepared as described in. Lysates were resolved by gel electrophoresis, transferred to nitrocellulose, and probed with respective antibodies.

30Cells were injected into both flanks of either CD-1 nude mice or FVB mice and tumor growth measured twice-weekly. Animals were sacrificed when tumors reached maximum allowed size or when signs of ulceration were evident. For treatment with VS-4718, drug was prepared in 0.5% carboxymethyl cellulose + 0.1% Tween 80 and mice treated at 75 mg/kg BID by gavage. No signs of toxicity were observed.

31SCC FAK−/− cells were injected into the left flank of FVB mice. Following tumor regression, mice were left for 7 days before being challenged with SCC FAK-WT or FAK−/− cells injected into the right flank. Tumor growth was measured twice-weekly. Control groups were injected into both flanks at day 28 using mice that had not been pre-challenged with SCC FAK−/− cells.

32T cell depletion was achieved following IP injection of 150 μg of depleting antibody into female age-matched FVB mice for 3 consecutive days and was maintained by further IP injection at 3 day intervals until the study was terminated. SCC FAK-WT or FAK−/− cells were injected into both flanks 6 days after initial antibody treatment and tumor growth measured. The extent of T cell depletion was determined at the end of the study using FACS (Figure S1).

33Tumors established following injection of SCC cells into both flanks of an FVB mouse were removed at day 7. Tumor tissue was processed to obtain single cell suspension for staining and subsequent FACS analysis (antibodies listed in Table S2).

34RNA was analyzed using the GeneChip Mouse Genome 430 2.0 Array. Normalized data for differentially expressed genes were median centered and clustered using Cluster 3.0 and Java TreeView. Functional enrichment analysis was performed using ToppGene.

35RNA prepared from SCC cells was analyzed using the mouse cytokine and chemokine RT2 Profiler PCR Array and that from isolated Tregs was analyzed using the mouse chemokine and receptor array. Relative gene expression (2−ΔCt) values were log transformed, median centered, and subjected to hierarchical clustering as for microarray analysis. An interactome of chemokine ligands and receptors was constructed using the IUPHAR/BPS Guide to Pharmacology database and curated from the literature, onto which expression data for detected genes were mapped and visualized using Cytoscape. Expression of selected cytokine and chemokine genes was assessed by standard quantitative RT-PCR.

36Cells were subject to two rounds of lentiviral infection prior to selection with puromycin. shRNA constructs used were part of the pLKO lentiviral TRC library.

37Nuclei were prepared as described but with a reduced concentration (0.05%) of NP-40 in nuclei buffer B. Soluble chromatin was prepared as described and fractionated on a sucrose step gradient to separate soluble and chromatin-associated nuclear proteins. DNA was recovered from fractions and subjected to agarose gel electrophoresis. Protein was purified using TCA precipitation. Samples were analyzed by SDS-PAGE and blotted using anti FAK, HP1α, and histone H3 antibodies.

38FAK nuclear protein complexes were subjected to on-bead proteolytic digestion, desalting, and liquid chromatography-tandem mass spectrometry, as described. For interaction network analysis, Ccl5 transcription factors were extracted from the DECODE database and used to seed a network of 1,000 transcription factor-related proteins using the GeneMANIA plugin in Cytoscape. Proteins specifically isolated in nuclear FAK protein complexes were mapped onto the interactome, and those with physical or predicted direct or indirect interactions with Ccl5 transcription factors were analyzed using the NetworkAnalyzer plugin in Cytoscape.

39Tumors were removed 7 days post-implantation and frozen by submersing in liquid nitrogen. Tumor sections were cut, processed and stained. They were imaged using an Olympus FV1000 confocal microscope.

40A.S. and M.C.F. devised and oversaw the project. A.S., T.L., B.S., A.B., S.M.A., R.J.B.N., and M.C.F. designed the experiments with contributions from E.M., J.A.P., V.G.B., and N.G. A.S., T.L., B.S., A.B., R.C.M., and A.v.K. performed experiments with contributions from L.G.-C., M.C., M.M., and J.E.R. A.S., T.L., B.S., A.B., R.C.M., A.v.K., and A.H.S. analyzed the data. A.B. and A.H.S. performed bioinformatic analysis. A.S. and M.C.F. wrote the manuscript with contributions from T.L., B.S., and A.B.; all authors commented on and approved the final version. We consider that A.S. and T.L. made equal contributions and that B.S. and A.B. made equal contributions.