Increased expression of focal adhesion kinase (FAK) was consistently observed in low- and high-grade astrocytomas and during glioblastoma progression after radiotherapy, but not in the more benign oligodendroglioma. In glioblastoma cell lines deficient for p53, p16INK4A, and p14ARF, FAK was inhibited in a dominant-negative manner by the focal adhesion targeting (FAT) domain, reducing invasion. In addition, caspase-3 activity was increased after serum withdrawal, or by cisplatin in the presence of serum, or upon loss of substrate attachment, and was in each case independent of PTEN status. Our results identify FAK as a potential target for anti-invasive strategies against infiltrating glioma cells.
The mean survival time for patients diagnosed with brain tumors arising from the glial lineage varies between 10 months for a high-grade, highly invasive GBM3 to 9.8 years for a low-grade oligodendroglioma (1). The invasion of healthy brain tissue by tumor cells proceeds along the ECM of white matter tracts and the perivascular space, which contains ligands for the integrin family of transmembrane receptors (2). FAK is required for integrin-dependent signaling and modulates cellular adhesion, migration (3), and survival (4). FAK is targeted to the focal adhesions by the COOH-terminal FAT domain, promoting phosphorylation of tyrosine-397 (Tyr397) which is essential for FAK activity (5). Expression of the FAT domain itself is sufficient to inhibit FAK (4). The dual-specificity phosphatase PTEN, which is inactivated in 25–35% of glioblastomas (6), also modulates integrin signaling and influences cell survival (7) and migration (8). To address the role of FAK in human brain tumors, we have investigated the expression of FAK in low- and high-grade tumors and tested the role of FAK by use of a dominant-negative FAT-domain construct in genetically well-characterized glioblastoma cell lines that differ in PTEN status (9).
Materials and Methods
Immunohistochemistry of Tumor Samples.
Formalin-fixed tissue samples were obtained from the University of Basel Department of Neuropathology. Immunohistochemistry was performed as described previously (10) using an overnight incubation with the polyclonal anti-FAK antibody at 4°C at a dilution of 1:50 (C20; Santa Cruz Biotechnology, Santa Cruz, CA). Bound antigens were detected using avidin-biotin-peroxidase (Vectastain, Elite kit; Vector Laboratories, Burlingame, CA), and the sections were weakly counterstained with hematoxylin.
Construction of FAT Expression Constructs.
The FAT domain of FAK was amplified from human fetal brain cDNA, beginning at Ser (840) in human FAK and using the oligonucleotides FAT840ser_s 5′-GACTCGGATCCATGTCTCGAGGCAGTATTGACAG-3′ and FAK3388_as. 5′-GGCCCGTCGACGTGTGGTCTCGTCTGCC-3′. Underlined are BamHI and SalI cleavage sites for the cloning of PCR products. To make pFATmycZeo, the FAT domain PCR product was subcloned into pcDNA_myc (11) and subsequently into pcDNA3.1/Zeo(+) (Invitrogen, San Diego, CA).
Cell Culture and Stable Cell Lines.
Cells were transfected with GenePORTER 2 (Gene Therapy Systems, San Diego, CA), and selected for 2 weeks in 800 μg/ml Zeocin (Invitrogen). Single clones were analyzed for FAT domain expression using a polyclonal anti-myc antibody (Upstate Biotechnology, Lake Placid, NY) and maintained in the presence of 250 μg/ml Zeocin. No apparent differences were noted between five independent clones of each cell line in terms of invasive behavior, sensitivity to apoptosis, and expression of the FAT domain. Adherent cells were cultured on 6-cm FN coated dishes (0.2 μg/cm2; Sigma Chemical Co., St. Louis, Mo). For suspension cultures, 6-cm plates were thinly coated with 0.7% agar in PBS and equilibrated for 1 h in growth media (DMEM/10% FCS) before adding 3 × 106 cells in 2 ml of growth media.
Western Blotting and Immunochemistry of Glioblastoma Cell Lines.
For analysis of FAK phosphorylation, cells were starved overnight in 0.5% FCS before plating onto FN-coated plates in 0.5% FCS for 30 min. Cells were harvested in ice-cold PBS and lysed on ice for 10 min in 1% Triton X-100, 150 mm NaCl, 50 mm Tris.HCl (pH 7.5), 2 mm EGTA, 2 mm EDTA, 5% glycerol, 500 μm Na3VO4, 50 μm PMSF, 10 μg/ml aprotinin, and 10 μg/ml leupeptin, and was analyzed by Western blotting using rabbit anti-FAK (pY397; Biosource International, Camarillo, CA). Total FAK was visualized by stripping the membranes for 60 min in 0.1 m glycine (pH 2.5) and using rabbit anti-FAK (A-17; Upstate Biotechnology). Westerns were developed using SuperSignal West Pico Chemiluminescent substrate (Pierce, Rockford, IL). To visualize focal adhesion proteins, 2.5 × 104 cells were cultured in Lab-Tek Chamber slides (Nalgene Nunc International, Naperville, IL), fixed for 5 min in 4% paraformaldehyde, washed in PBS, and blocked for 30 min in 5% goat serum, 1% BSA, and 0.01% Triton-X100/PBS. The following antibodies were used in the same buffer for 2 h at room temperature: rabbit anti-FAK (A-17), which recognizes an epitope within the NH2-terminal half of FAK (Upstate Biotechnology); MAbV284 anti-Vinculin (Upstate Biotechnology), rabbit anti-Myc tag (residues 409–420, Upstate Biotechnology). Secondary antibodies were BODIPY FL goat antimouse IgG or Cy3-conjugated goat antirabbit IgG (Molecular Probes, Eugene, CA).
Invasion and Apoptosis Assays.
Invasion was measured by a Matrigel invasion assay in a Boyden Chamber as described previously (12). Caspase-3 activity was determined by harvesting attached and nonattached cells and were assayed using a specific ELISA-based caspase-3 fluorometric assay according to the manufacturer’s instructions (Roche Diagnostics). Cisplatin (Platinol) was obtained from Bristol-Myers Squibb. BCNU (Bristol-Myers Squibb) was dissolved in alcohol and used immediately.
Semiquantitative Reverse Transcription-PCR.
Total RNA was harvested using Trizol, and cDNA was synthesized using Thermoscript reverse transcriptase according to the manufacturer’s instructions (Invitrogen, San Diego, CA). Primers for cIAP1, cIAP2, and XIAP were the same as used by Sonoda et al. (13), or for the NH2-terminal region of FAK from nucleotides 1108 to 1608. To ensure that amplification was linear, cDNA was serially diluted and PCR was done for 25 cycles.
Results and Discussion
Fifteen GBM (WHO grade IV), 2 anaplastic astrocytomas (WHO grade III), 5 low-grade fibrillary astrocytomas (WHO grade II), and 5 low-grade oligodendrogliomas (WHO grade II) were examined for FAK immunoreactivity. In 14 of 15 GBM (Fig. 1,A, top), 2 of 2 anaplastic astrocytomas, and 5 of 5 low-grade astrocytomas (Fig. 1,A, middle), FAK immunoreactivity was observed in all of the astrocytic cell compartments (cell soma and astrocytic processes) and was often highly localized at the cell membrane (Fig. 1,A, top). Although FAK expression was increased in higher-grade astrocytomas, FAK immunoreactivity was virtually absent from all 5 of the oligodendrogliomas examined (Fig. 1,A, bottom), a glioma subtype with a much better prognosis than tumors of the astrocytic lineage (14). Increased FAK expression was also seen in a recurrent glioblastoma after radiotherapy (Fig. 1,B, top) when compared with the original tumor (Fig. 1 B, bottom).
To investigate whether FAK was required for tumor cell invasion and survival, we used two well-defined glioblastoma cell lines, LN-401 and LN-229, both of which are deficient for p53 and p16INK4A/p14ARF. Importantly, only the LN-229 cell line has retained a functional PTEN protein (9). The FAT domain (Fig. 1,C) was stably expressed in each glioblastoma cell line, generating LN-401/FAT and LN-229/FAT cell lines. Mock-transfected clones that did not express the FAT domain were selected as controls. The LN-401/mock and LN-401/FAT cell lines expressed similar levels of the FN receptor α5β1 (data not shown), and the recruitment of endogenous FAK or the FAT domain to the focal adhesions was examined in a FN adhesion assay. Endogenous FAK and the FAT domain were distinguished using antibodies that recognized either an epitope in the NH2-terminal part of FAK or a myc epitope engineered onto the FAT domain (Fig. 1,C). In LN-401/mock cells, FAK colocalized with the focal adhesion protein vinculin (Fig. 1,D, mock/N-FAK and mock/vinculin), whereas in LN-401/FAT cells, FAK immunoreactivity was lost from the focal adhesions and was accompanied by a similar decrease in vinculin immunoreactivity (Fig. 1,D, FAT/N-FAK and FAT/vinculin). In contrast, FAT immunoreactivity was concentrated at the cell periphery (Fig. 1 D, FAT/myc), which demonstrated that FAK was excluded from the focal adhesions by direct competition with the FAT domain.
Because both cell lines expressed the FN receptor, and FAK integrates growth factor signaling by integrins (15), the activation of FAK in LN-401 and LN-229 cell lines was assayed by adhesion to FN in either the absence or presence of serum (Fig. 2,A, −, +, respectively). Using a specific antibody that recognized FAK phosphorylation at Tyr397 (Fig. 2,A, FAK pY397), we found that adhesion of LN-401/mock cells to FN in the absence of serum-induced robust FAK phosphorylation at Tyr397, which was not increased by the addition of serum (Fig. 2,A, LN-401, mock). In comparison, adhesion of LN-401/FAT cells to FN only weakly induced FAK phosphorylation at Tyr397, and although the addition of serum afforded some stimulation, the level of phosphorylation remained low (Fig. 2,A, LN-401, FAT). In LN-229/mock cells, which express PTEN, adhesion to FN in the absence of serum also induced robust phosphorylation of Tyr397 (Fig. 2,A, LN-229, mock); and, similar to the results obtained in LN-401 cells, Tyr397 phosphorylation LN-229/FAT cells was inhibited (Fig. 2 A, LN-229, FAT).
We next investigated whether the inhibition of FAK reduced invasion and sensitized the cells to pro-apoptotic stimuli. Using a Matrigel Boyden-chamber assay, loss of FAK activation in LN-401/FAT and LN-229/FAT cell lines was accompanied by a significant 60% decrease in invasive behavior over that seen in mock-transfected controls (Fig. 2,B). Activation of the integrin receptors by specific ligands in the ECM is a potent survival signal (16), and adhesion to FN protects both LN-401 and LN-229 cell lines from apoptosis on serum withdrawal (data not shown). To test whether the FAT domain might attenuate this effect, cells were adhered to FN and cultured for 2 days under reduced serum (0.5% FCS). In both LN-401/FAT and LN-229/FAT cell lines, specific caspase-3 activity was increased 2- to 3-fold above that seen in the mock-transfected controls (Fig. 2 C).
Whereas FAK has been reported to be a substrate for PTEN (8), in the experiments described here, FN-dependent phosphorylation of FAK was seen in glioma cell lines with differing PTEN status, which suggested that FAK was activated independently of PTEN. Consistent with this conclusion, transfection of PTEN into U87 MG cells, which do not express functional PTEN, failed to inhibit FAK phosphorylation on adhesion to FN, although specific phosphorylation of PKB/Akt at Ser473 was reduced (12). Furthermore, the phosphotyrosine content of adherent U87 MG cells was not affected by expression of PTEN (12). Similar to results obtained in LN-401 cells, however, expression of the FAT domain in U87 MG cells inhibited both FAK and total phosphotyrosine of adherent U87 MG cells, thereby inhibiting migration and increasing caspase-3 activity on the withdrawal of serum.4
Although FAK phosphorylation was not influenced by PTEN in glioma cells, PTEN is an important modulator of cell survival, negatively regulating the PKB/Akt kinase in the presence of serum and increasing the sensitivity of cells to cytotoxic stimuli (7). Therefore, we examined whether the pro-apoptotic effect of PTEN in the presence of either cisplatin or BCNU was enhanced by the FAT domain and whether expression of the FAT domain itself increased the cytotoxic effect of these agents. Cell lines were cultured in normal concentration of serum (10% FCS), incubated with either cisplatin or BCNU for 2 days, and assayed for caspase-3 activity. In the LN-401 cell lines, which do not express functional PTEN, cisplatin significantly increased caspase-3 activity in LN-401/FAT cells (Fig. 2,D, left, white bars) 1.6-fold above that seen in LN-401/mock cells (Fig. 2,D, left, black bars; relative caspase-3 activity was as follows: LN-401/mock, 9,248 ± 1,000; LN-401/FAT, 16,200 ± 1,230, mean ± SE; n = 9; U<U*, α = 0.05, Mann-Whitney test). BCNU failed to effectively stimulate caspase-3 activity in either cell line (Fig. 2,D). In LN-229/mock cells, which are wild-type for PTEN, the relative increase in caspase-3 activity induced by cisplatin was 2- to 3-fold higher than that seen in LN-401/mock cells (Fig. 2,D, right, black bars), and caspase-3 activity was increased a further 1.7 fold in LN-229/FAT cells (Fig. 2,D, right, white bars; relative caspase-3 activity was as follows: LN-229/mock, 14,976 ± 1579; LN-229/FAT, 25,075 ± 2,340, mean ± SE; n = 9; U<U*, α = 0.05, Mann-Whitney Test). Again, BCNU did not induce caspase-3 activity (Fig. 2 D). Antibodies that block integrin activation enhance the chemosensitivity of small cell lung cancer (17), which suggests that the FAT domain similarly abrogates the protective effect afforded by integrin and growth factor signaling against cytotoxic insults.
Loss of substrate attachment in normal cells leads to the induction of apoptosis, or anoikis (18). Tumor cells are generally resistant to anoikis, reflecting independence from integrin signaling. Introduction of functional PTEN into glioma cells is sufficient to induce anoikis by inhibiting PKB/Akt (19). To examine whether disruption of FAK signaling by dominant-negative FAT affects survival, tumor cells were cultured in suspension on a nonadherent agar substrate. Both LN-401/mock and LN-229/mock cells remained viable for at least 2 weeks when cultured in suspension in the presence of serum (10% FCS; data not shown). In contrast, expression of the FAT domain in either LN-401 or LN-229 cells induced a 3-fold increase in caspase-3 activity after 4 days in culture in the presence of serum (Fig. 3,A). Interestingly, caspase-3 activity was greatest in LN-401/FAT cells, which suggested that additional mutations in LN-229 cells negated the pro-apoptotic effect of PTEN. Overexpression of FAK has been shown to increase expression of the IAP gene family (cIAP1, cIAP2, and XIAP; Ref. 13). In LN-401/FAT cells, FAK RNA had decreased after 20 h in suspension (Fig. 3,B, FAK), and the levels of RNA encoding each of the IAP genes examined were consistently less in LN-401/FAT cells than in LN-401/mock cell lines (Fig. 3,B; cIAP1, cIAP2, and XIAP). Similar results were obtained in LN-229/FAT cells (data not shown). Further characterization of each cell line revealed elevated expression of the antiapoptotic protein Bcl-2 in both LN-229/mock and LN-229/FAT, but not in the LN-401 cell lines (data not shown). Under adherent conditions, there was a small difference in the levels of Bcl-2 in LN-229/mock and LN-229/FAT cells (Fig. 3,C, adherent). However, the level of Bcl-2 in LN-229/FAT cells decreased nearly 2-fold below that seen in LN-229/mock cells when cultured in suspension (Fig. 3 C, suspension). Detachment of epithelial cells from the ECM leads to a decrease in the level of Bcl-2, which is blocked by oncogenic ras (20), demonstrating that mutations in signaling proteins downstream of integrin receptors contribute to anoikis. Significantly, ras activation is also regulated by FAK (21). Therefore, the FAT domain induces anoikis in glioblastoma cells by decreasing FAK expression and activation and inhibiting antiapoptotic pathways.
In summary, the results presented here show that (a) astrocytomas of WHO grades II-IV exhibited high levels of FAK expression, whereas oligodendrogliomas, which have a more favorable prognosis than astrocytomas, displayed almost no FAK immunoreactivity; (b) FAK expression is increased during tumor progression as shown in a recurrent tumor following radiotherapy; (c) expression of the FAT domain in glioblastoma cell lines competed with endogenous FAK for localization to the focal adhesions, inhibiting FAK activation in a dominant-negative manner; and (d) the FAT domain was sufficient to inhibit invasion and sensitize glioblastoma cells to differing apoptotic stimuli irrespective of PTEN, p53, and p16INK4A/p14ARF status. Given that the majority of human brain tumors are incurable because of tumor cell infiltration and chemoresistance, these results establish FAK as a important mediator of survival and migration in these tumors.
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.
Supported by Grant NF374037-055167/1 from the Swiss National Science Foundation program in somatic gene therapy (to A. M.).
The abbreviations used are: GBM, glioblastoma multiforme; FAT, focal adhesion targeting (domain); FAK, focal adhesion kinase; ECM, extracellular matrix; BCNU, 1,3-bis(2-chloroethyl)-1-nitrosourea; FN, fibronectin; IAP, inhibitor of apoptosis.
G. Jones and A. Merlo, unpublished data.
We acknowledge the excellent technical assistance of Beatrice Dolder.