Inhibition of Focal Adhesion Kinase by PF-562,271 Inhibits the Growth and Metastasis of Pancreatic Cancer Concomitant with Altering the Tumor Microenvironment

Pancreatic ductal adenocarcinoma (PDA) remains the fourth leading cause of cancer-related deaths in the United States (1). Current cytotoxic therapies (gemcitabine, 5-fluorouracil, and oxaliplatin) and anti–epidermal growth factor (EGF) therapy (erlotinib) provide only modest improvement in overall survival (2–6).

PDA is characterized by aggressive locoregional invasion and early metastasis, indicating a central role for signaling pathways that regulate cell migration. In addition, PDA typically evokes a desmoplastic reaction characterized by significant invasion and proliferation of stromal cancer–associated fibroblasts (myofibroblasts and stellate cells) and deposition of extracellular matrix (ECM) proteins (7).

Focal adhesion kinase (FAK), a nonreceptor cytoplasmic protein tyrosine kinase, is a key regulator of signals from the ECM mediated by integrins and growth factor receptors (GFR; ref. 8). FAK has been implicated in the regulation of a variety of cellular signaling pathways that control cell proliferation, cell-cycle progression, migration, apoptosis, and cell survival (9–12). Expression of FAK is reported to be upregulated in many tumor types including colon, breast, prostate, thyroid, head and neck, liver, and esophageal (12–17), and elevated expression of FAK correlates with poor survival rates (17, 18). In PDA, it has been reported that increased FAK expression correlates with tumor size (19). Gene silencing of FAK by RNA interference resulted in suppression of anoikis resistance and metastasis in a preclinical model of PDA (20). Several small-molecule inhibitors of FAK have been described, most notably PF-562,271, a potent small-molecule inhibitor of FAK, and the related tyrosine kinase PYK2 (21). PF-562,271 is reported to be efficacious in inhibiting the proliferation of tumors in both xenograft and transgenic mouse models (22, 23)

Because of the prominent role of the microenvironment in the progression and metastasis of PDA, we undertook to determine the role of FAK/PYK2 catalytic activity in the proliferation and migration of tumor cells, and stroma-associated cells (macrophages and pancreatic stellate cells), the latter cells that populate the tumor microenvironment and contribute to the desmoplastic response of PDA. We show that chemotactic [insulin-like growth factor (IGF)-I and EGF] and haptotactic [collagen-I (Col-I)] migration and invasion of human PDA cells were inhibited by concentrations of PF-562,271 that effectively block FAK catalytic activity. In addition, PF-562,271 inhibited the migration of primary mouse macrophages and both immortalized human stellate cells and tumor-associated fibroblasts to tumor cell conditioned media. In a preclinical mouse model, in which tumor-derived or -established cancer cells are implanted into the pancreas of immunocompromised mice, treatment with PF-562,271 or gemcitabine significantly reduced the rate of tumor growth, local tumor invasion, and metastasis. Interestingly, only PF-562,271 led to a significant decrease in tumor-associated macrophages and cancer-associated fibroblasts. The decrease in these cells correlated with the onset of treatment with PF-562,271 and correlated with a decrease in tumor cell proliferation and an observed reduction in tumor volume. These observations suggest that in the setting of an orthotopic tumor xenograft, treatment with PF-562,271 inhibits PDA cell growth either directly or through a mechanism that disrupts the recruitment of cancer-associated fibroblasts and monocytic cells to the tumor.

The human PDA cell line L3.6pl was kindly provided by I.J. Fidler (The University of Texas MD Anderson Cancer Center, Houston, TX; ref. 24). Human pancreatic stellate cells (HPSC; ref. 25) were obtained from Rosa Hwang (The University of Texas MD Anderson Cancer Center, Houston, TX; July 2009). MPanc-96 and BxPC-3 cells were obtained from the American Type Culture Collection (August 2005) and maintained in Dulbecco's Modified Eagle's Medium (HPSCs and MPanc-96) or RPMI (BxPC-3) supplemented with 10% FBS and antibiotics. Mouse macrophages were obtained in 2007 and prepared as described previously (26). All cell lines were expanded, aliquoted, and frozen upon initial receipt and then new cells were thawed, propagated, and used for experiments every 6 months. MPanc-96 and BxPC-3 were authenticated before purchase by the American Type Culture Collection with cytochrome c oxidase subunit 1 analysis, DNA profiling, cytogenetic analysis, flow cytometry, and immunocytochemistry. L3.6pl cells, HPSCs, and MAD08-608 cells and mouse fibroblasts (described later) were authenticated in 2010 by the University of Virginia Biomolecular Research Facility with DNA profiling, cytogenetic analysis, flow cytometry, and immunocytochemistry.

The MAD08-608 tumor was derived from remnant tissue from a surgical pathology specimen of a patient with PDA in 2008 that was obtained under an Institutional Review Board–approved protocol. Tumor tissue (1-mm³ cubes) was affixed to the pancreas of athymic nude mice. After 8 weeks, tumors were passaged into a second generation (F₂) of mice. Tumors from F₂ mice were then either digested with 0.8 mg/mL collagenase to establish a cell line or passaged into a third generation (F₃) of mice for further experiments. The MAD08-608 fibroblasts were isolated by preferential trypsinization from an in vitro coculture of tumor cells and fibroblasts. Fibroblast cultures were more than 90% pure as determined by staining with anti-vimentin.

PF-562,271 (N-methyl-N-{3-[({2-[(2-oxo-2,3-dihydro-1H-indol-5-yl)amino]-5-(trifluoromethyl)pyrimidin-4-yl}amino)methyl]pyridin-2-yl}methanesulfonamide), generously provided by Pfizer, Inc., is a potent ATP-competitive reversible inhibitor selective for recombinant FAK and PYK2 kinase with an IC₅₀ of 1.5 and 14 nmol/L, respectively (21, 22). Animals were treated with PF-562,271 at 33 mg/kg body weight twice per day as described previously (22, 23). Gemcitabine (2′,2′-difluoro-2′-deoxycytidine) was given intraperitoneally at a dose of 10 mg/kg twice per week as described previously (27).

Cultured cells were lysed and subjected to Western blotting for FAK, PYK2, phospho-FAK (pY397), and phospho-PYK2 (Y402) as described previously (28). Human phospho-receptor tyrosine kinase (pRTK) arrays (R&D Systems) were used to evaluate the pRTK signature in PDA cell lines per the manufacturer's protocol.

Cell migration and invasion assays were conducted by Transwell migration chambers (BD Biosciences) as previously described (26–28). Conditioned media were prepared from subconfluent MPanc-96 or MAD08-608 cells by harvesting growth medium (containing 0.5% FBS) for two 24-hour periods. These conditioned media were centrifuged to remove cellular debris and concentrated 3-fold with Amicon Ultra-15 3K centrifugal filter devices (Millipore). Mock conditioned media were prepared by concentrating 0.5% FBS in a similar fashion. For proliferation assays, a total of 1.5 × 10⁵ cells were plated in 96-well plates in 1% FBS-containing media, and cell number was measured by the CyQuant Assay (Invitrogen).

Six-week-old male athymic nude mice (NCI, Fredericksburg, MD) were maintained in accordance with Institutional Animal Care and Use Committee standards. Mice were anesthetized, a skin incision was made on the left flank, and the pancreas was exteriorized. For MPanc-96 experiments, 50 μL of cell suspension was inoculated into the pancreatic parenchyma. For MAD08-608 experiments, 1-mm³ fragments were affixed to the pancreas with a 6-0 polypropylene suture (Ethicon). Mice were randomized into 1 of 4 treatment groups: PF-562,271 (33 mg/kg in 200-μL vehicle by gastric lavage twice per day) and gemcitabine (10 mg/kg intraperitoneally twice per week) beginning on day 7, and treatment continued until the mice were sacrificed on day 21 (MPanc-96) or day 35 (MAD08-608). At necropsy, retroperitoneal tumor invasion was assessed [confirmed by hematoxylin and eosin (H&E)], tumors were excised and measured, and tumor volume was calculated. Livers were excised and examined, along with the peritoneal and mesenteric surfaces, for the presence of surface metastases.

Small-animal MRI was carried out on a 7.0T ClinScan MRI system (Siemens/Bruker; Siemens Corp). MRI data were collected by a True FISP sequence with a 30-mm mouse whole-body coil and respiratory gating (SA Instruments, Inc.). Images were processed and reviewed on PACS (Kodak Carestream). Volumetric analysis was conducted by multiplying the slice thickness (0.5 mm) by the sum of the area of the tumor visible (mm²) in each slice.

Paraffin-embedded sections of tumors were stained with H&E according to standard methods. Immunohistochemical staining was carried out with antibodies to human vimentin (Epitomics), human and mouse vimentin (Abcam), FAK (Santa Cruz Biotechnology), Ki67 (Cell Signaling), F4/80 (AbD Serotec), alpha smooth muscle actin (α-SMA; Abcam), and PARP (Santa Cruz Biotechnology). CD31 (BD Pharmingen) staining was done on frozen sections of tissue embedded in optimum cutting temperature compound. Negative controls were obtained by omitting the primary antibody. Colon tissue from normal mice was used as a positive control.

Statistical analyses were carried out by GraphPad Prism (GraphPad Software, Inc.) Nonlinear regression was used to evaluate the inhibitory effect of PF-562,271. One-way ANOVA was used to determine significance of difference among multiple arms of treatment in vivo, and the Bonferroni multiple comparison test was used to define significance between groups. The Student t test was used to determine the significance of difference in the mean values and the Fisher exact test to calculate the significance of difference between categorical variables. Significance was determined with 95% confidence interval.

FAK and the related protein PYK2 are well-studied regulators of cell adhesion signaling and migration (29). To better understand the role of FAK/PYK2 signaling in both PDA and stromal cell migration, we examined MPanc-96 and MAD08-608 PDA cells, macrophages, and HPSCs (25). Orthotopic implantation of either MPanc-96 or MAD08-608 cells in the mouse pancreas resulted in the rapid growth of tumors and the maintenance of pancreatic cancer architecture (Fig. 1A). The original patient MAD08-608 tumor (F₀) readily stained with a human-specific vimentin antibody, confirming the presence of human-derived stromal tissue. F₁ tumors propagated in mice failed to stain with a human-specific anti-vimentin antibody but showed robust staining with an antibody that recognized both mouse and human vimentin (Fig. 1A), indicating that the microenvironment of human tumors growing orthotopically in the mouse pancreas was reconstituted by stroma of mouse origin. Immunohistochemical staining of normal pancreas and PDA with an antibody to FAK revealed strong staining for FAK throughout the PDA sections, with staining most intense in the malignant ductal cells (Fig. 1B). In contrast, weaker staining of normal pancreatic tissue for FAK expression was observed (Fig. 1B). Using Western immunoblotting, FAK and PYK2 expression was readily observed in both MPanc-96 and MAD08-608 cells, as well as several other PDA cell lines and cultured HPSCs (Fig. 1C). In all cases, FAK and PYK2 were activated (Fig. 1C), as evidenced by tyrosine phosphorylation of FAK (Y397) and PYK (Y402).

To evaluate the effects of PF-562,271 on FAK activity, MPanc-96 and MAD08-608 cells were incubated, and PF-562,271 and FAK activity was assessed by immunoblotting extracts with an antibody to FAK phospho-Y397. As shown in Fig. 2A, the addition of PF-562,271 reduced the phosphorylation of FAK in both cell lines. Maximal inhibition of FAK Y397 phosphorylation was observed at 0.1 to 0.3 μmol/L PF-562,271, consistent with the earlier published observations (22, 28). These experiments confirm that FAK is active in PDA and is inhibited by treatment with PF-562,271.

The activation of GFR signaling pathways plays an important role in PDA growth, migration, invasion, and metastasis. Both MPanc-96 and MAD08-608 cells express multiple activated GFRs as detected by pRTK arrays. As shown in Fig. 2B, 9 of the 42 selected RTKs exhibited signals above background, strongest of which were EGF receptor (EGFR), ErbB2 (MAD08-608), FGFR3, insulin-R (InsR), IGF-IR, and hepatocyte growth factor receptor (HGFR; c-Met) and EphA2. Western blots of extracts prepared from MAD08-608 and MPanc-96 cells confirmed the expression of EGFR, ErbB2, fibroblast growth factor receptor, and HGFR (data not shown).

Because activated EGF and insulin/IGF family receptors were expressed in both MPanc-96 and MAD08-608 cells, we evaluated the effects of PF-562,271 on growth factor–mediated cell migration. In MPanc-96 cells, IGF-I (100 ng/mL) significantly stimulated cell migration (Fig. 3A) and treatment of MPanc-96 cells with 0.1 μmol/L PF-562,271 resulted in a significant decrease in migration compared with IGF-I stimulation alone (Fig. 3A). EGF also stimulated migration, but, interestingly, the observed migration was insensitive to the addition of PF-562,271.

Because of the increased deposition of ECM proteins (e.g., Col-I) in PDA, we evaluated the ability of PF-562,271 to inhibit haptotactic migration. Col-I stimulated migration of both MPanc-96 (Fig. 3A) and MAD08-608 cells (Supplementary Data). Pretreatment of either cell line with PF-562,271 (0.1 μmol/L) led to a significant reduction in haptotactic migration (Fig. 3A). Perineural, lymphovascular, and extrapancreatic invasion are common histologic characteristics of PDA and associated with worse prognosis. To assess the invasive activities of MPanc-96 and MAD08-608 cells, assays were conducted as previously but with Matrigel inserts. Serum stimulated invasion in both cell types, and pretreatment with 0.1 μmol/L of PF-562,271 significantly inhibited invasion (Fig. 3A and Supplementary Data). These data illustrate that migration and invasion of MPanc-96 and MAD08-608 cells are dependent to some degree on the catalytic activity of FAK and/or PYK2.

Tumor cells often produce cellular factors that contribute to the recruitment of host monocytes/macrophages and fibroblasts/myofibroblasts. To test whether MPanc-96 and MAD08-608 cells produced factors chemotactic for stromal cells, conditioned medium from tumor cells was used to stimulate the migration of mouse macrophages and either HPSCs or tumor-associated mouse fibroblasts (Fig. 3B). Tumor cell conditioned medium readily stimulated the migration of both macrophages and HPSCs, and this migration was significantly inhibited by treatment with PF-562,271. These observations support a role of FAK and PYK2 in cell migration of both tumor cells and cells that comprise the tumor microenvironment.

We evaluated the effects of PF-562,271 on in vitro cell proliferation. Growth assays were conducted in the presence of serum-containing media supplemented with either PF-562,271, gemcitabine (a known inhibitor of PDA cell proliferation), or both. Treatment of both MPanc-96 and MAD08-608 cells with increasing concentrations of gemcitabine (1, 10, and 100 ng/mL) resulted in the dose-dependent inhibition of cell growth with approximately 50% inhibition achieved at 100 ng/mL (Fig. 3C and Supplementary Data). The addition of PF-562,271 at concentrations up to 10 times the dose required to inhibit catalytic activity and migration/invasion (1 μmol/L) had no effect on cell proliferation, either alone or in combination with gemcitabine. These results clearly indicated that PF-562,271 does not effectively block the growth of PDA cells under the culture conditions used in these experiments and does not enhance the inhibition of growth by gemcitabine.

Because PF-562,271 inhibited tumor and stromal cell migration in vitro, we tested the efficacy of this inhibitor in an orthotopic mouse model. Following injection (MPanc-96) or implantation (MAD08-608) into the mouse pancreas, mice were treated with vehicle, PF-562,271 (33 mg/kg, twice daily, a dose that inhibited growth of several different tumor types in subcutaneous models; ref. 22), gemcitabine (10 mg/kg, twice weekly, a dose selected to reduce tumor growth by approximately 50%; ref. 27), or PF-562,271 plus gemcitabine. Tumor growth was assessed by MRI at weekly intervals (Fig. 4). As shown in Fig. 4B, 2 weeks posttreatment tumor volumes in mice treated with PF-562,271 were 46% ± 8% smaller than control tumors and gemcitabine-treated mice were 53% ± 7% smaller than control mice. Tumors in mice treated with gemcitabine plus PF-562,271 were significantly smaller than control tumors; however, they were not significantly different from tumors receiving either drug alone.

Parallel analysis of mice bearing MAD08-608 tumors (Fig. 4C) showed that treatment with PF-562,271 or gemcitabine significantly reduced tumor size relative to control mice (59% ± 15% and 60% ± 11% reduction, respectively). Tumors in mice treated with gemcitabine plus PF-562,271 were significantly smaller than control tumors; however, they were not significantly different from tumors receiving either drug alone.

At necropsy, tumors were evaluated for invasion into the retroperitoneum, and mouse livers and peritoneum were excised and evaluated for the presence of surface metastases. Treatment with PF-562,271 reduced the number of MPanc-96 tumor–bearing mice with detectable retroperitoneal invasion and liver and peritoneal metastasis (Table 1). Mice bearing MAD08-608 tumors showed a reduced incidence of peritoneal metastasis upon treatment with PF-562,271.

To evaluate the effects of PF-562,271, gemcitabine, and combination treatment on tumor cell proliferation, apoptosis, angiogenesis, macrophage infiltration, and fibroblast infiltration, representative sections from individual tumors from mice treated for 2 weeks (MPanc-96) or 4 weeks (MAD08-608) were stained with antibodies for Ki67, cleaved PARP, CD31, F4/80, and α-SMA, respectively. The analysis of sections from MPanc-96 and MAD08-608 tumors revealed no significant differences in apoptosis (cleaved PARP staining) or angiogenesis (CD31 staining) among any of the treatment groups (data not shown). Treatment of mice with PF-562,271 in the presence or absence of gemcitabine led to a significant decrease in tumor cell proliferation (Ki67-positive cells) in MPanc-96 (29%, P < 0.01 and 23%, P < 0.05, respectively) and MAD08-608 (47%, P < 0.001 and 41%, P < 0.01, respectively) tumors relative to control (data not shown). In mice bearing MPanc-96 (Fig. 5A) and MAD08-608 tumors (Supplementary Data), significantly fewer F4/80-staining cells (macrophages) were observed in tumors from mice treated with PF-562,271 than in control mice or mice treated with gemcitabine. Interestingly, mice treated with gemcitabine had the same (MAD08-608) or significantly more (MPanc-96) macrophages present in their tumors than in control, although this difference was blunted in mice treated with both agents.

The presence of α-SMA–positive cells (cancer-associated fibroblasts) was also significantly reduced in mice treated with PF-562,271 in a pattern similar to that seen with the macrophages (Fig. 5B and Supplementary Data). Treatment of mice with gemcitabine also led to no change or a slight increase in fibroblasts within the tumor compared with control, whereas treatment with PF-562,271 yielded significantly fewer fibroblasts as a single agent than control (Fig. 5B and Supplementary Data).

To further evaluate the effects of PF-562,271 on timing of macrophage and fibroblast infiltration during the early phase of treatment, MPanc-96 mice were sacrificed at weeks 1 and 2 and tumor sections stained for F4/80 and α-SMA. Normal murine pancreas had no detectable F4/80 or α-SMA staining (week 0; data not shown). F4/80-positive cells (macrophages; Fig. 5C) and α-SMA–positive cells (fibroblasts; Fig. 5D) were readily observed after 1 week of tumor growth. The number of F4/80- and α-SMA–positive cells increased between weeks 1 and 2 in control mice harboring either MPanc-96 or MAD08-608 tumors, and in MAD08-608 mice, an additional increase was observed after 3 weeks of tumor growth (Fig. 5C and D and Supplementary Data). Treatment of mice with PF-562,271 resulted in a reduction in the number of both F4/80- and α-SMA–positive cells (Fig. 5C and D and Supplementary Data) compared with control mice. These observations support the idea that treatment with the PF-562,271 influenced the recruitment or persistence of macrophages and fibroblasts within the tumor microenvironment.

PDA is a dismal disease with a poor prognosis and limited therapeutic options. Tumor cell migration and invasion are critical steps in tumor progression and metastasis, both occurring at early stages in PDA. An additional hallmark of PDA is the intense desmoplastic, fibrotic response, characterized by copious collagen deposition and abundant infiltrating macrophages and cancer-associated fibroblasts (7). In this study, we use the FAK/PYK2 inhibitor PF-562,271 to examine the efficacy of PF-562,271 treatment in chemotactic and haptotactic migration of PDA cells in vitro, the migration of macrophages and stromal cells to factors produced by tumor cells, and tumor formation and metastasis following orthotopic implantation of tumor cells in immunocompromised mice. We show that the treatment of MPanc-96 cells, a well-characterized PDA cell line, or MAD08-608, a recently derived tumor, with PF-562,271 inhibited cell migration to IGF-I and collagen but failed to block cell proliferation in culture. Stromal cell migration and migration of macrophages and fibroblasts to tumor cell conditioned medium were also blocked by PF-562,271. Treatment of mice bearing MPanc-96 or MAD08-608 tumors with PF-562,271 inhibited tumor growth and reduced retroperitoneal invasion and peritoneal and liver metastases. Importantly, we show that tumors from animals treated with PF-562,271 showed a significant decrease in the number of tumor-associated F4/80-staining (macrophages) and α-SMA–staining (cancer-associated fibroblasts) cells and a significant decrease in tumor cell proliferation. In contrast, we observed no difference in the number of F4/80- or α-SMA–staining cells in tumors from mice treated with tumor-inhibitory doses of gemcitabine. These data lead us to propose that PF-562,271 acts to inhibit tumor proliferation either directly by inhibiting tumor cell proliferation or indirectly by blocking the recruitment and/or proliferation of stromal cells to the tumor microenvironment.

FAK upregulation in solid organ malignancies and the inverse correlation between FAK levels and overall survival have been shown previously (12–17). Using immunohistochemistry, FAK was readily detectable in the malignant ductal cells and to a lesser degree the surrounding stromal cells. Activated FAK and PYK2 were present in both human tumor cells and immortalized human stellate cells as well as in mouse macrophages (26). FAK plays a significant role in the regulation of adhesion turnover and migration (10, 12, 30). Consistent with this function of FAK, treatment of PDA cells with PF-562,271 blocked migration mediated by IGF-I and collagen and invasion in response to serum. Paradoxically, PF-562,271 failed to block the migration of MPanc-96 cells to EGF, indicating that these cells may express an alternative pathway(s) to regulate cell migration. This latter observation is consistent with recent findings that PF-562,271 fails to block the migration of a number of other non-PDA cell lines (J. Slack-Davis and J.T. Parsons; unpublished data) underscoring the variability of pathways available to cancer cells to mediate migration. Of note is the observation that PF-562,271 efficiently blocked the serum-induced invasion of both MPanc-96 and MAD08-608 cells. Indeed, other studies have shown that PF-562,271 consistently inhibits the invasive properties of non-PDA cell lines (J. Slack-Davis and J.T. Parsons, unpublished data), supporting the notion that FAK and PYK2 are important for events leading to the invasive properties of tumor cells. The observations that PF-562,271 failed to inhibit the growth in culture of MPanc-96 and MAD08-608 cells are consistent with a growing body of data indicating that PF-562,271 is not an effective inhibitor of cancer cell proliferation when cells are grown in vitro on rigid substrates such as plastic (ref. 28; J. Slack-Davis, R.W. Tilghman, and J.T. Parsons, unpublished data). These observations are in contrast to published reports, as well as data presented herein that show that in vivo PF-562,271 mediates significant tumor inhibition, suggesting that the importance of the FAK/PYK2 pathway(s) in tumor cell proliferation in vitro and in vivo is distinct.

A role of FAK in macrophage and fibroblast migration is well established (26, 31). Deletion of FAK in primary bone marrow macrophages leads to altered adhesion dynamics and impaired chemotaxis. Deletion of FAK impairs the recruitment of monocytes to sites of inflammation. In addition, knockdown of PYK2 expression in primary macrophages resulted in a diminution of invasive capacity, indicating a possible role of this FAK-related tyrosine kinase in the inflammatory response. Similarly, deletion of FAK, knockdown of FAK, or inhibition of FAK catalytic activity results in the impaired migration of fibroblasts (29, 31). Given these observations, it is not surprising that treatment with PF-562,271 could possibly restrict the recruitment of macrophages and fibroblasts to sites of tumor growth in the pancreas.

Using an orthotopic model of tumor growth that closely mimics the natural progression of PDA, we showed that treatment of mice with PF-562,271 inhibited tumor growth. Interestingly, the addition of gemcitabine revealed little increase in tumor inhibition. The lack of inhibition of in vitro proliferation by PF-562,271, but significant inhibition of tumor cell proliferation in vivo, suggested to us that PF-562,271 may inhibit tumor growth in mice via a direct effect on the tumor as well as indirect mechanisms affecting the microenvironment. Using immunohistochemistry, we found that CD31 staining of tumors was not affected by PF-562,271, suggesting that reduced tumor growth was not the result of inhibition of angiogenesis. In contrast, F4/80 and α-SMA staining was significantly reduced after PF-562,271 treatment but not in mice treated with gemcitabine. The decreased staining is consistent with the possibility that PF-562,271 either directly inhibits the recruitment of monocytic cells and cancer-associated fibroblasts or inhibits the tumor-dependent production of cytokines necessary for the recruitment of such cells. The fact that tumors from mice treated with gemcitabine (and having tumors similar in size to PF-562,271–treated animals) have F4/80 and α-SMA staining similar to control tumors argues that the reduction in staining is not due simply to a decrease in tumor size. Thus, we conclude from these observations that PF-562,271 treatment has an effect on the tumor microenvironment that suppresses tumor growth and metastasis. Interestingly, recent data show the importance of cancer-associated fibroblasts in pancreatic tumor progression and chemotherapy resistance (32).

The efficacy of PF-562,271 has been shown in a number of preclinical studies. Roberts and colleagues showed that administration of PF-562,271 inhibited the growth of several tumor cell lines engrafted subcutaneously in immunocompromised mice (22). Slack-Davis and colleagues reported that PF-562,271 inhibited the outgrowth of castrate resistant prostate cancer in a genetically engineered murine model of prostate cancer (23). Recently, Bagi and colleagues reported that the combination of sunitinib and PF-562,271 inhibited both angiogenesis and tumor aggressiveness, exhibiting greater effects than the relevant single agent (21). This treatment not only blocked tumor growth but also impacted the ability of the tumor to recover upon withdrawal of the therapy.

Targeting the tumor cell–microenvironment interaction represents a novel approach to cancer therapy. Given recent evidence that cancer-associated fibroblasts promote tumor progression and our results in this study, further investigation of FAK inhibition and other therapies targeting the tumor microenvironment is warranted in the treatment of PDA.

No potential conflicts of interest were disclosed.

This work was supported by NIH T32 HL007849 (J.B. Stokes), American Cancer Society MRSG-0700201CCE (T.W. Bauer), UVA Cancer Center grant P30 CA44579 (T.W. Bauer), CA 40042 (J.T. Parsons), and Pfizer, Inc. (J.T. Parsons).

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.