Integrin α2β1 Inhibits Fas-mediated Apoptosis in T Lymphocytes by Protein Phosphatase 2A-dependent Activation of the MAPK/ERK Pathway

The mechanisms by which T lymphocytes escape apoptosis during their activation are still poorly defined. In this study, we elucidated the intracellular signaling pathways through which β1 integrins modulate Fas-mediated apoptosis in T lymphocytes. In experiments done in Jurkat T cells and activated peripheral blood T lymphocytes, engagement of α2β1 integrin with collagen type I (Coll I) was found to significantly reduce Fas-induced apoptosis and caspase-8 activation; Annexin V binding and DNA fragmentation were reduced by ∼42 and 38%, respectively. We demonstrated that the protective action of Coll I does not require new protein synthesis but was dependent on the activation of the MAPK/Erk pathway. Furthermore, we found that activation of protein phosphatase 2A (PP2A) by Coll I was required for both Coll I-mediated activation of Erk, and inhibition of Fas-induced caspase-8 activation and apoptosis. Other ligands of β1 integrins, fibronectin (Fbn), and laminin (Lam), did not sustain significant Erk activation and had no effect on Fas-induced apoptosis. Taken together, these results provide the first evidence of a PP2A-dependent activation of the MAPK/Erk pathway downstream of α2β1 integrin, which has a functional role in regulating Fas-mediated apoptosis in T lymphocytes. As such, this study emphasizes the potential importance that Coll I interactions may have on the control of T lymphocyte homeostasis and their persistence in chronic inflammatory diseases.

Integrins are ␣/␤ heterodimeric membrane proteins that mediate cell adhesion to the surrounding extracellular matrix (ECM), 1 and can elicit a wide variety of intracellular signals that modulate cell growth and proliferation (1,2). Normal epithelial and endothelial cells depend on integrin signals for cell cycle progression, and disruption of matrix attachment induces their apoptosis, a process termed anoikis (3). We and others (4 -7) have also shown that integrin signaling protects anchored cells from different forms of apoptosis such as serum withdrawal and chemotherapeutic agents. Focal adhesion kinase (FAK), integrin-linked kinase (ILK), and Src kinases have all been involved in connecting integrins to downstream survival signaling pathways such as the phosphatidylinositol 3-kinase (PI 3-kinase), the serine/threonine kinase Akt, and the MAPK/Erk pathways (8 -10). In addition to kinases, and although less studied, phosphatases have been involved in integrin-mediated signal transduction (11,12).
T lymphocytes express several ␤ 1 integrins, the most studied being ␣ 4 ␤ 1 and ␣ 5 ␤ 1 (13). Both of them mediate cell adhesion to the ECM protein (ECMp) fibronectin (Fbn), and to VCAM-1 in the case of ␣ 4 ␤ 1 integrin, and transduce costimulatory signals for T cell receptor (TCR)-mediated proliferation (14 -16). The signaling events associated with other ␤ 1 integrins, such as the collagen-binding integrins ␣ 1 ␤ 1 and ␣ 2 ␤ 1 that are expressed during late stages of T cell activation (17), and the cellular functions modulated by these integrins are still poorly defined.
TCR-mediated apoptosis (also known as activation-induced cell death (AICD)) plays an important role in the regulation of immune response and is largely mediated by the interaction of Fas ligand (Fas-L) with its receptor Fas (18,19). During activation, Fas-mediated apoptosis must be inhibited for the T lymphocytes to survive and accomplish their task. However, the mechanisms involved in such process are still poorly defined. Some evidence suggests that integrin signaling could modulate apoptosis in T lymphocytes. Recently, we have shown that collagen type I (Coll I) but not Fbn or laminin (Lam) protected Jurkat T cells from AICD (20). In intestinal CD4positive lymphocytes, activation of ␤ 1 integrins mediates proliferation and inhibition of AICD (21), whereas in antigenspecific T lymphocyte clones, AICD is enhanced by the coligation of the TCR and the ␣ 4 ␤ 1 and ␣ L ␤ 2 (LFA-1) integrins (22). Currently, the exact role of integrin signaling in T lymphocyte apoptosis and the molecular mechanisms by which integrins may accomplish this role are still unclear.
In this study, we investigated the regulation of Fas-mediated apoptosis of T lymphocytes by ␤ 1 integrin signaling. We show that engagement of ␣ 2 ␤ 1 integrin by its ligand Coll I inhibits Fas-induced apoptosis by activating the MAPK/Erk survival pathway in a protein phosphatase 2A (PP2A)-dependent manner. In contrast, Fbn and Lam failed to induce PP2A activity or sustain significant Erk activation, and had no effect on Fas-induced apoptosis. Taken together, these results demonstrate a role for ␣ 2 ␤ 1 integrin and its ligand Coll I in the regulation of Fas-mediated apoptosis in T cells, and underline a differential signaling and functions of the ␤ 1 integrins molecules in activated T lymphocytes. These findings may have an important impact on the modulation of immune response and development of inflammatory diseases.

EXPERIMENTAL PROCEDURES
Cell Culture and Preparation of T Lymphocytes-The human Jurkat T-cell line E6 -1 was obtained from ATCC and maintained in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mmol/liter glutamine and 100 units/ml penicillin and streptomycin (complete medium). Human peripheral blood T lymphocytes from healthy donors were isolated on Ficoll gradients and were then enriched and purified on human T lymphocyte enrichment columns from R&D Systems (Minneapolis, MN) according to the manufacturer's instructions. Staining with anti-CD3 antibody and flow cytometry (FACS) analysis indicated that more than 97% of the isolated cells were CD3-positive T cells.
Freshly isolated blood T lymphocytes are resistant to Fas-mediated apoptosis and do not express the integrins ␣ 1 ␤ 1 and ␣ 2 ␤ 1 . They become sensitive to this form of apoptosis and express both integrins only after in vitro activation with anti-TCR/CD3 or with PHA and subsequent culture in the presence of IL-2 for 6 to 8 days (17,23). Thus, the purified peripheral blood T lymphocytes were activated on day 1 with PHA for 24 h and then cultured in complete medium containing 50 units/ml of IL-2 for 6 days before being used in further experiments.
Cell Death and Determination of Apoptosis-Jurkat T cells or activated peripheral blood T lymphocytes were resuspended at 1 ϫ 10 6 /ml in RPMI 1640 medium containing 2.5% (medium A) or 5% (medium B) fetal calf serum, respectively. The cells were then seeded in 24-well plates (5 ϫ 10 5 /well) and cultured alone or in the presence of ECMp (Coll I, Fbn, Lam), or poly-L-lysine (PLL), added in their soluble forms at 20 g/ml. The cell cultures were then treated with anti-Fas mAb (CH11) as indicated in the figure legends. For activation-induced cell death experiments, 24-well plates were coated with anti-CD3 mAb (OKT3) (10 g/ml) alone or in combination with 20 g/ml of ECMp overnight at 4°C. The wells were then washed three times with PBS, and peripheral blood T lymphocytes were seeded (5 ϫ 10 5 /well) and cultured for 36 h. At the end of cultures, apoptosis was determined using a cell death detection ELISA kit measuring DNA fragmentation, according to the manufacturer's instructions (Roche Applied Science). Briefly, after stimulation, 10 4 cells were washed in PBS and lysed in 200 l of lysis buffer. After centrifugation, 20 l of supernatant were transferred to the streptavidin-coated microtiter plate (MTP) and then 80 l of immunoreagent were added to each well. Plates were incubated with shaking for 2 h at room temperature, washed, and then 100 l/well of ABTS substrate were added. The color development for photometric analysis was measured at 405 nm against ABTS solution as a blank by an ELISA reader. Apoptosis was also determined using the Annexin V-PE/7AAD detection kit from BD Pharmingen (San Diego, CA). After stimulation, the cells were washed in cold PBS, and 10 5 cells were incubated in 1ϫ buffer containing 5 l of Annexin V-PE and 5 l of 7-AAD for 15 min at room temperature, in the dark. The cells were then analyzed by flow cytometry using a FACScan (BD Biosciences).
Cell Surface Molecule Expression, CH11 Binding, and Flow Cytometry-The expression of CD3, ␣ 2 ␤ 1 , and ␣ 1 ␤ 1 integrins was determined by flow cytometry analysis using specific antibodies as previously described (20). CH11 binding was analyzed by determining the expression of Fas receptor by flow cytometry analysis. Briefly, 1 ϫ 10 6 cells in 0.1 ml of PBS containing 1% fetal calf serum and 0.2% sodium azide (binding buffer) were incubated with 10 g/ml of specific antibody (anti-CD3, anti-␣ 1 ␤ 1 , and anti-␣ 2 ␤ 1 ) or their isotype-matched control antibodies for 30 min at 4°C. For CH11 antibody binding experiments, 5 ϫ 10 5 cells resuspended in medium A were cultured alone or in the presence of 20 g/ml of Coll I for 3 h at 37°C. The cells were further cultured for 2 h with 50 ng/ml of CH11 or an isotype-matched control antibody. After incubation with the first antibody, the cells were washed three times with cold PBS, and incubated at 4°C with FITCconjugated goat anti-mouse IgG secondary antibody for CD3, ␣ 1 ␤ 1 , and ␣ 2 ␤ 1 staining, or with FITC-conjugated goat anti-mouse IgM secondary antibody for CH11 binding. Labeled cells were analyzed by flow cytometry using FACScan (BD Biosciences). The FITC-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories (West Grove, PA).
Immunoblotting, Caspase-8 Activation, and Erk and Akt Phosphorylation-The cells were resuspended at 1 ϫ 10 6 /ml in medium A and stimulated in 24-well plates (5 ϫ 10 5 /well) as indicated in the figure legends. At the end of the culture, the cells were harvested, washed in cold PBS, and cell lysates were prepared in radioimmune precipitation assay buffer containing protease and phosphatase inhibitors as we previously described (5). Caspase-8 activation was determined by immunoblot analysis using anti-caspase-8 antibody that recognizes the 18-kDa active form of caspase-8 (p18). Erk and Akt activation was determined by immunoblot analysis using specific antibodies that recognize the phosphorylated forms of human Erk1/2 or Akt. The blots were stripped and reprobed with control antibodies (anti-actin, anti-Erk2, or anti-Akt antibodies) as indicated to ensure equal loading. In all experiments, immunoblots were visualized using a horseradish peroxidase-conjugated secondary antibody followed by enhanced chemiluminescence detection (Pierce).
Plasmids and Transient Transfections-The plasmid encoding the dominant negative form of Mek-1 (DN-Mek-1) was a generous gift of Dr. Jean Charron (Laval University, Canada). The mutant cDNA encoding a kinase-dead domain form of Mek-1 was cloned in a pCMV-flag mammalian expression vector (Stratagene, La Jolla, CA). The plasmid encoding the green fluorescent protein (pGFP) was a generous gift of Dr. Kristiina Vuori (The Burnham Institute, La Jolla, CA). Briefly, 20 ϫ 10 6 cells in log-phase growth were harvested, washed, and resuspended in 400 l of RPMI 1640 medium containing 40 g of total plasmid DNA. Jurkat T cells were transfected by electroporation at 250 V and 960 F settings using a Bio-Rad electroporator. Two days after transfection, viable cells were recovered by Ficoll-hypaque density gradient centrifugation and used in subsequent apoptosis experiments as described above.
Phosphatase Assay-Phosphatase activity was determined using a non-radioactive serine/threonine phosphatase assay system (Promega, Madison, WI) according to the manufacturer's instructions. The assay is based on the detection of the amount of free phosphate generated in the reaction by measuring the absorbance of a molybdate-malachite greenphosphate complex. Briefly, Jurkat T cells were resuspended in medium A and stimulated with Coll I, Fbn, or Lam for various periods of time as indicated. After stimulation, cells were lysed in a phosphatase lysis buffer (20 mM HEPES pH 7.4; 10% glycerol, 0.1% Nonidet P-40, 1 mM EGTA, 30 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, leupeptin, and pepstatin at 2 g/ml each). 25 g of cell lysate were incubated in 96-well plates in the presence or absence of a peptide substrate RRA (pT) VA in a buffer appropriate for PP2A activity (50 nM imidazole, pH 7.2, 0.2 mM EGTA, 0.02% ␤-mercaptoethanol, 0.1 mg/ml bovine serum albumin) for 30 min at 30°C. After incubation, the molybdate complex dye was added and incubated for an additional 30 min at room temperature for color development. The reaction was read at 630 nm using a plate reader. Of note, the phosphopeptide used in this assay is a preferred substrate for PP2 enzymes and is a poor one for protein phosphatase 1 (PP1) because of its more stringent structural requirements (24).

Collagen Type I Inhibits Fas-mediated Apoptosis in T Lymphocytes-
The human T cell line Jurkat, similar to activated primary T lymphocytes expresses several members of the ␤ 1integrin subfamily (25) and is widely used for Fas-mediated apoptosis studies (20,26,27). Thus, to test the effect of ␤ 1 integrin signaling on Fas-mediated apoptosis, Jurkat T cells were preincubated with Coll I, Fbn, or Lam for 3 h and then treated with the agonistic anti-Fas antibody (CH11) for 12 h. Treatment with Coll I but not with Fbn, Lam or PLL, a nonintegrin binding ligand used as control, protected Jurkat T cells from CH11-induced apoptosis; the percentage of Annexin Vpositive cells was decreased by 50% (Fig. 1A) and 42% (Fig. 1B), and DNA fragmentation was inhibited by 38% (Fig. 1C). The strongest protection from Fas-induced apoptosis was observed after 3-4 h of pretreatment with Coll I. Fas-induced apoptosis was reduced by 40 and 20% after 3-4 h and 6 -24 h of pretreatment with Coll I respectively (data not shown). Flow cytometry analysis of CH11 binding demonstrated that preincubation of Jurkat T cells with Coll I did not alter the ability of CH11 to

FIG. 1. Coll I inhibits Fas-induced apoptosis in Jurkat T cells.
Jurkat T cells resuspended in medium A were cultured in the presence or absence of 20 g/ml of the indicated ECMp or PLL for 3 h, after which they were treated or not with 50 ng/ml of anti-Fas mAb (CH11). A, after 12 h of treatment with CH11, apoptosis was determined by Annexin V/7AAD double staining and flow cytometry analysis as described under "Experimental Procedures." The percentages of Annexin V-positive (lower right) and Annexin V-positive/7AAD-positive (upper right) cells are indicated. One representative experiment is shown. B, average percentages of Annexin V-positive cells from three independent experiments, like the one presented in panel A, are presented graphically with S.E. as indicated. C, apoptosis was also determined by DNA fragmentation analysis as described under "Experimental Procedures." The results are presented as mean of optical density values from three independent experiments, with S.E. as indicated. Statistical analysis was carried out using Student's t test: *, p Ͻ 0.01 between anti-Fas-treated PLL-and Coll I-stimulated samples.

FIG. 2. Coll I does not affect the binding of CH11 to Fas receptor. Jurkat T cells resuspended in medium A
were cultured in the presence or absence of 20 g/ml of Coll I for 3 h, after which they were treated with 50 ng/ml of CH11 or its isotype control antibody for another 2 h. The cells were then harvested, stained with FITC-conjugated anti-mouse IgM antibody, and analyzed by flow cytometry. Staining of non-treated or Coll I-treated cells after incubation with isotype control antibody (---) and with CH11 (-) is shown. The results are representative of three independent experiments. bind to Fas receptor (Fig. 2), indicating that the observed inhibition did not result from a reduced CH11 binding. The protective action of Coll I was totally abrogated in the presence of blocking anti-␣ 2 integrin antibodies (data not shown), consistent with our previous results and those by others that ␣ 2 ␤ 1 integrin is the functional receptor for Coll I on Jurkat T cells (20,25,28). Thus, Coll I induces signals through the ␣ 2 ␤ 1 integrin that modulate Fas-mediated apoptosis.
The modulation of Fas-mediated apoptosis by Coll I signaling also occurs in activated peripheral blood T lymphocytes. Flow cytometry analysis showed that after 6 days of in vitro activation, purified peripheral T lymphocytes express significant levels of ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins (Fig. 3A), and their preincubation with Coll I but not Fbn, Lam, or PLL inhibited their Fasmediated apoptosis (Fig. 3B). This Coll I-mediated inhibition was strongly inhibited by anti-␣ 2 and to a lesser extent by anti-␣ 1 integrin blocking antibodies (Fig. 3B), indicating that the protective effect of Coll I was mediated through its ␣ 2 ␤ 1 and ␣ 1 ␤ 1 integrin receptors. Since AICD in T lymphocytes is mainly mediated through Fas signaling, we examined the effect of Coll FIG. 3. Coll I inhibits Fas-and anti-CD3-induced apoptosis in activated primary T lymphocyte through ␣ 2 ␤ 1 and ␣ 2 ␤ 1 integrins. A, peripheral blood T lymphocytes were activated in vitro and the expression of ␣ 1 ␤ 1 and ␣ 2 ␤ 1 integrins was determined by FACS analysis using specific antibodies and FITC-conjugated anti-mouse IgG antibody as described under "Experimental Procedures." B, activated peripheral blood T lymphocytes resuspended in medium B were cultured in the presence or absence of the indicated ECMp or PLL for 3 h, after which they were treated with 0.5 g/ml of CH11 for 24 h. As indicated, the cells were pretreated with 10 g/ml of anti-␣ 1 or anti-␣ 2 integrin blocking antibodies or a mixture of both antibodies prior to their stimulation with Coll I. Apoptosis was determined by Annexin V/7AAD double staining and flow cytometry. C, activated peripheral blood T lymphocytes resuspended in medium B were stimulated with or without immobilized anti-CD3 mAb for 36 h in the presence or absence of 20 g/ml of the indicated ECMp. Apoptosis was determined by DNA fragmentation analysis as described under "Experimental Procedures." In both B and C, the results are presented as average values from three independent experiments done with T lymphocytes from different blood donors, with S.E. Statistical analysis was carried out using Student's t test: *, p Ͻ 0.03 between Fas-treated PLL or control and Coll I-stimulated samples.
I on anti-CD3-induced apoptosis of activated peripheral blood T lymphocytes. These cells underwent significant apoptosis when cultured in wells coated with anti-CD3 mAb, which was reduced by 38% in the presence of Coll I but not of Fbn, or Lam (Fig. 3C). Similar to the inhibition of Fas-induced apoptosis, Coll I-mediated protection from AICD is also blocked by anti-␣ 2 integrin blocking antibodies (data not shown). Together these results indicate that in activated primary T cells, Coll I equally reduces both AICD and Fas-induced apoptosis.
When effects of other apoptotic stimuli were evaluated, Coll I also inhibited TRAIL-induced apoptosis, a form of apoptosis that is mediated through death receptor and activation of the caspase-8 apoptotic pathway (29), to a similar degree as it did inhibit Fas-induced apoptosis (40% for Fas versus 35% for TRAIL) (Fig. 4). In contrast, Coll I did not protect Jurkat T cells from staurosporine-and etoposide-induced apoptosis (Fig. 4), two stimuli that activate the mitochondrial death pathway independently from death receptors (30,31). Treatment with Coll I for longer periods before the addition of staurosporine or etoposide also did not result in any protection from apoptosis (data not shown). These results indicate that in Jurkat T cells, Coll I selectively inhibits death receptor-mediated apoptosis rather than mitochondria-death pathway, and the observed protection does not result from a general inhibitory effect on cell death.
We next investigated whether new protein synthesis is required for the anti-apoptotic effect of Coll I. As shown in Fig. 5, Coll I-mediated protection from Fas-induced apoptosis occurred equally well in the presence or absence of cycloheximide, indicating that the protective effect of Coll I was independent from new protein synthesis.
The ␣ 2 ␤ 1 Integrin-mediated Protection Against Fas-induced Apoptosis Is MAPK/Erk-dependent-Some evidence indicates that activation of MAPK/Erk and PI 3-kinase/Akt pathways downstream of integrins can block various forms of apoptosis (8,9). Thus, to investigate the signaling mechanisms underlying the ␣ 2 ␤ 1 integrin-mediated protection against Fas-induced apoptosis, we analyzed the activation of these pathways by Coll I in Jurkat T cells. Stimulation of the cells with Coll I induced a significant and sustained increase in the phosphorylation levels of Erk1/2 starting at 2 h and lasting up to 24 h (Fig. 6A). Densitometry analysis showed that the increase in Erk1/2 phosphorylation peaked at 4 h, being 3-fold higher than control value (Fig. 6B). In contrast, AKT was found constitutively phosphorylated in Jurkat T cells, which is in agreement with a previous report (32), and treatment with Coll I did not significantly change the levels of phosphorylated Akt (Fig. 6C). No increase in Akt or Erk phosphorylation was detected in cells treated with Coll I for less than 2 h (data not shown).
To demonstrate the implication of MAPK/Erk pathway in the protective action of Coll I, we first ascertained that the stimulation of Erk phosphorylation by Coll I also occurred in the presence of the apoptotic trigger (CH11) (Fig. 7A). Next, we examined the effects of the Mek-1-specific inhibitor PD98059 on ␣ 2 ␤ 1 integrin-mediated cell survival. Treatment of Jurkat T cells with PD98059 abolished both the ability of Coll I to increase Erk phosphorylation (Fig. 7B) and to protect against Fas-induced apoptosis (Fig. 7C). Treatment of Jurkat T cells with PD98059 had no effect on cell viability, and had only a minor effect on the dose-dependent increase of CH11-induced Jurkat T cell apoptosis (Fig. 7C). These results are in agreement with previous reports, which demonstrated that PD98059 did not sensitize nor did it potentiate Fas-induced apoptosis in Jurkat T cells (27,33). Importantly, Coll I inhibited apoptosis induced with all tested amounts of CH11 and the presence of PD98059 reversed its effect (Fig. 7C). These results indicate that the effects of PD98059 on apoptosis are not on overall survival in response to Fas ligation but rather on Coll I-mediated survival. Interestingly, Coll I-mediated Erk phosphorylation (see Fig. 6B) and inhibition of Fas-induced apoptosis; which were both optimal after 3-4 h of pretreatment and sustained up to 24

FIG. 7. Coll I-mediated inhibition of Fas-induced apoptosis is Erkdependent.
A, increase of Erk1/2 phosphorylation by Coll I occurs in the presence of CH11. Jurkat T cells were stimulated or not with Coll I for 3 h, after which they were treated or not with 50 ng/ml of CH11 for the indicated periods of time. Cell lysates were prepared and Erk activation was determined by immunoblot analysis with antiphospho-Erk1/2 antibody as described above. B, treatment with Mek-1 inhibitor abrogates Coll I-induced increase in Erk phosphorylation. Jurkat T cells were pretreated or not with the Mek-1 inhibitor PD98059 (25 M) for 1 h, and Erk phosphorylation was determined after 3 h of Coll I stimulation by immunoblot analysis as described above. Similar results were obtained with purified activated T lymphocytes (data not shown). C, treatment with PD98059 abolishes the protective effect of Coll I. The cells were first treated or not with PD98059 (25 M) for 1 h before stimulation with Coll I and treatment with the indicated concentrations of CH11. After a 12-h treatment with CH11, apoptosis was determined by Annexin V/7AAD double staining and flow cytometry as described under "Experimental Procedures." The results are presented as average percentages of Annexin V-positive cells from three independent experiments with S.E. as indicated. Similar results were obtained when apoptosis was measured by DNA fragmentation assay (data not shown).

FIG. 6. Coll I modulates the phosphorylation of Erk1/2 but not of AKT.
A, Coll I induces an increase in Erk1/2 phosphorylation. Jurkat T cells resuspended in medium A were stimulated or not with 20 g/ml of Coll I for the indicated periods of time. Cell lysates were prepared and Erk activation was determined by immunoblot analysis with antiphospho-Erk1/2 antibody (top). The membrane was stripped and reprobed with anti-Erk2 antibody to confirm equal loading (bottom). B, densitometry quantification of relative increase in total Erk phosphorylation shown in the above blot. The results are expressed as fold increase of the ratio between total phospho-Erk-1/2 and Erk2. C, Coll I does not affect the level of AKT phosphorylation. Jurkat T cells were stimulated as described in A, and AKT activation was determined by immunoblot analysis with anti-phospho-AKT antibody (top). The membrane was stripped and reprobed with anti-AKT antibody to confirm equal loading (bottom). The results are representative of three different experiments.
against Fas-induced apoptosis. Together, these results demonstrate that ␣ 2 ␤ 1 integrin-mediated activation of the MAPK/Erk pathway is involved in the protective effects of Coll I against Fas-induced apoptosis.
Because Fbn and Lam had no effect on Fas-mediated apoptosis in Jurkat T cells (Fig. 1), we examined the capacity of these ECMp to increase Erk phosphorylation. The results in Fig. 9 demonstrate that Fbn and Lam induce only a slight and transient increase of Erk1/2 phosphorylation that is weaker than that induced by 2 h of stimulation with Coll I. Prolonged stimulations, up to 24 h, with either ECMp, or increasing their concentrations did not result in any further Erk activation, and interestingly, the phosphorylation levels of Akt were also not modulated by these ECMp (data not shown). These results strongly suggest that the differential ability of ␤ 1 integrin ligands to modulate Fas-mediated apoptosis in Jurkat T cells may be due to their differential ability to increase Erk phosphorylation.
Activation of Protein Phosphatase 2A Is Necessary for the Protective Action of Collagen Type I-Recently ␣ 2 ␤ 1 integrin has been shown to promote activation of the protein phosphatase type 2A (PP2A) in osteosarcoma cells (34). PP2A is one of the four major serine/threonine phosphatases that regulate diverse cellular functions such as cell division and transcription, and several reports have shown that PP2A can regulate either positively or negatively the activation of the MAPK/Erk pathway (35)(36)(37). Thus, we sought to investigate the role of PP2A in ␣ 2 ␤ 1 integrin-mediated Erk phosphorylation and inhibition of Fas-induced apoptosis. Since the strongest effects of Coll I on Erk phosphorylation and cell survival were observed after 3-4 h of pretreatment with Coll I, the role of PP2A was also investigated under these conditions. Jurkat T cells were found to express basal levels of active PP2A, and treatment with Coll I, used at the concentration inducing optimal Erk phosphorylation and inhibition of Fas-mediated apoptosis, induced a 2-fold increase in PP2A activity (Fig. 10A). The Coll I-induced increase in PP2A activity was detected after 30 min, peaked after 1 h of stimulation, and remained detectable after 3 h of stimulation. The increase in PP2A activity dropped to basal levels after 4 h of treatment with Coll I and remained so for up to 24 h (data not shown). Pretreatment of Jurkat T cells with 50 nM of okadaic Acid (OA), a selective inhibitor of PP2A when used at nanomolar concentrations (38), abolished both the basal level as well as the Coll I-induced increase in PP2A activity. Because OA, at high concentrations, can also inhibit protein phosphatase 1 (PP1), the phosphatase reactions were carried out in the presence of OA to distinguish between PP1 and PP2A. In phosphatase assays, OA inhibits PP2A at IC 50 ϭ 0.1 nM and PP1 at IC 50 ϭ 10 nM (39). As shown in Fig. 10B, the Coll I-induced increase in phosphatase activity is reduced by ϳ42 and 75% in the presence of 0.1 nM and 1 nM OA, respectively. These results indicate that the increase in phosphatase activity observed following Coll I stimulation was due to PP2A. Interestingly, neither Fbn nor Lam was able to induce any significant increase in PP2A activity (Fig. 10A).
To determine whether the observed increase in PP2A activity is necessary for the anti-apoptotic effect of Coll I, we have assessed the effects of nanomolar concentrations of OA on both Coll I-mediated increase of Erk phosphorylation and inhibition of Fas-induced apoptosis. As shown in Fig. 11, OA on its own has no effect on Erk phosphorylation (panel A, lane 2) and apoptosis (panel B). However, treatment of Jurkat T cells with OA abolished in a dose-dependent manner the ability of Coll I to increase Erk phosphorylation (Fig. 11A) and to protect the FIG. 11. Coll I-mediated increase in Erk phosphorylation and inhibition of Fas-induced apoptosis are PP2A-dependent. A, Jurkat T cells were pretreated with the indicated concentrations of OA or vehicle alone for 1 h, after which they were stimulated or not with 20 g/ml Coll I for 4 h. Cell lysates were prepared and Erk activation was determined by immunoblot analysis as described. The results are representative of three independent experiments. B, Jurkat T cells were pretreated for 1 h with the indicated concentrations of OA, followed by a 3 h stimulation with Coll I and treatment with the indicated concentrations of CH11. After 12 h of treatment with CH11, apoptosis was measured by DNA fragmentation analysis as described under "Experimental Procedures." The results are presented as mean of optical density values from three independent experiments, with S.E. C, Jurkat T cells were pretreated with OA as described in A and then stimulated with 20 ng/ml of PMA for 30 min. Erk activation was determined by immunoblot analysis as described. The results are representative of three independent experiments. D, Jurkat T cells were stimulated or not with PMA for 30 min before the addition of CH11. As indicated, the cells were pretreated or not with 50 nM of OA before PMA stimulation. Apoptosis was determined 12 h after the addition of CH11 by DNA fragmentation analysis as described under "Experimental Procedures." The results are presented as mean of optical density values from three independent experiments, with S.E. Similar results were also obtained when apoptosis was determined by Annexin V binding (data not shown). cells from Fas-induced apoptosis (Fig. 11B). Importantly, OA had no effect on Fas-induced apoptosis but reversed equally well the effect of Coll I at both low and high concentrations of CH11, indicating that the observed effects of OA are exerted specifically on Coll I-mediated cell survival (Fig. 11B). As a control, OA had no effect on PMA-mediated Erk phosphorylation (Fig. 11C) and inhibition of Fas-induced apoptosis (Fig.  11D), indicating the specificity of the OA treatment, since the effects of PMA on Erk phosphorylation are dependent on protein kinase C. Although Erk phosphorylation, and to a certain extent, cell survival, were sustained for up to 24 h with Coll I stimulation, we found that the observed increase in PP2A activity (between 30 min and 3 h) is necessary for Coll I-mediated Erk phosphorylation and cell survival for up to 8 h of stimulation with Coll I. However, it is not involved in the effects of Coll I on Erk phosphorylation and cell survival after 10 -24 h stimulation since these effects were not abolished by OA (data not shown). Together these results indicate that PP2A is involved in the initial protective effect of Coll I, which was the strongest, and not in the late protective action of Coll I.
To further support the role of PP2A/Erk signaling in the protective action of Coll I, we studied the regulation of caspase-8 activation. As shown in Fig. 12, treatment of Jurkat T cells with CH11 results in the generation of the p18 active form of caspase-8, which is reduced in the presence of Coll I but not PLL. Pretreatment of the cells with either PD98059 or OA abolished completely the ability of Coll I to reduce caspase-8 activation. Together these results indicate that ligation of ␣ 2 ␤ 1 integrin inhibits Fas-mediated apoptosis by activating the MAPK/Erk pathway through a mechanism involving activation of PP2A.

DISCUSSION
In this study, we analyzed the intracellular signaling by which ␤ 1 integrins modulate apoptosis in T lymphocytes, and showed that Coll I, but not Fbn or Lam, inhibits Fas-mediated apoptosis in Jurkat T cells and in primary human T lymphocytes through ␣ 2 ␤ 1 integrin signaling. We demonstrated that Coll I can activate the MAPK/Erk pathway by a mechanism that is dependent on the activity of PP2A, and that both PP2A and Erk activation have a functional role in ␣ 2 ␤ 1 integrinmediated protection against Fas-induced apoptosis in T lymphocytes.
Our results show that in contrast to Coll I, Fbn and Lam had no effect on PP2A activity, induced only a slight transient increase in Erk phosphorylation, and had no effect on Fasmediated apoptosis. Jurkat T cells attach to Lam and Fbn through ␣ 6 ␤ 1 and ␣ 4 ␤ 1 integrins, respectively (25,40). In agreement with our results, it has recently been reported that in Jurkat T cells, ligation of ␣ 4 ␤ 1 integrin induced only moderate levels of Erk1/2 phosphorylation (14). Together these results indicate that different members of ␤ 1 integrin family are connected to different signaling pathways in T lymphocytes, and may thus modulate different T cell functions in the course of immune response. The exact mechanisms by which members of the ␤ 1 integrin family connect to downstream signaling pathways and modulate T cell apoptosis are not fully elucidated. In this regard, our results strongly suggest that the differential ability of ␤ 1 integrins to modulate Fas-mediated apoptosis may be due, at least in part, to their ability to activate the MAPK/ Erk signaling pathway.
We also show that the PI 3-kinase/Akt pathway is not involved in ␣ 2 ␤ 1 integrin-mediated cell survival against Fasinduced apoptosis. Coll I had no effect on the levels of phosphorylated AKT, and PI 3-kinase inhibitors (wortmanin and LY294002) have no effect on the protective action of Coll I (data not shown). However, our results do not exclude the possible implication of other signaling mechanisms downstream of ␣ 2 ␤ 1 integrin that may cooperate with MAPK/Erk pathway in protecting T cells from Fas-mediated apoptosis.
Although the mechanisms involved in the regulation of this form of death in T cells are not fully defined, activation of the MAPK/Erk pathway has been implicated in the protective action of PMA, TCR/CD3 signaling and concanavalin-A (27,41,42). Our results are in agreement with these studies and indicate that ␣ 2 ␤ 1 integrin signaling is an important modulator of T cell survival. We also demonstrated that in Jurkat T cells, Coll I inhibited selectively death receptor-mediated apoptosis rather than the mitochondrial death pathway, which further supports the observed inhibition of caspase-8 processing by Coll I. Because Coll I-mediated protection is independent from new protein synthesis, it is likely that the observed inhibition of caspase-8 was due to post-translational mechanisms. This is in agreement with a report suggesting that in Jurkat T cells, activation of MAPK/Erk downstream of TCR/CD3 inhibited, by post-translational mechanisms, the processing of procaspase-8 downstream of DISC formation (27).
Using OA, a selective inhibitor of PP2A (38,39), we found that the optimal effects of Coll I on Erk phosphorylation and Fas-induced apoptosis require the activity of PP2A. The role of PP2A in Erk phosphorylation and in cell survival downstream of ␣ 2 ␤ 1 integrin is supported by the fact that Coll I-induced activation of PP2A precedes the increase of Erk phosphorylation, and that both OA and PD98059 had similar effects on Coll I-mediated increase in Erk phosphorylation, inhibition of caspase-8 and cell survival. Our findings concur with a recent report showing that ␣ 2 ␤ 1 integrin activates PP2A in osteosarcoma cells (34). However in these cells, activation of PP2A led to an attenuation of AKT phosphorylation. Such alteration in the level of AKT phosphorylation was not observed in our cell model, probably due to the different cell types and/or to the use of three-dimensional collagen in the case of osteosarcoma cells.
Several studies have shown that OA increases Erk activity in various cell lines (43)(44)(45), and have therefore concluded that PP2A is a negative regulator of MAPK/Erk signaling cascade. However, this seems not to be the case in Jurkat T cells. Their treatment with nanomolar amounts (50 nM) of OA inhibited PP2A activity but did not increase basal levels of Erk phosphorylation. In agreement with our results, it is only at high concentration of OA (1 M) that treatment of Jurkat T cells with OA resulted in a significant increase of Erk phosphoryla- FIG. 12. Coll I reduces Fas-induced caspase-8 activation in a PP2A/Erk-dependent manner. Jurkat T cells were stimulated for 3 h in the presence or absence of Coll I or PLL, after which they were treated with CH11 (50 ng/ml) for another 3 h. As indicated, the cells were pretreated with PD98059 or OA for 1 h prior to their stimulation with Coll I. Caspase-8 activation was determined by immunoblot analysis with anti-caspase antibody (Ab-1) that recognizes the 18 kDa active fragment of caspase-8 (p18) as described under "Experimental Procedures." The membrane was stripped and reprobed with anti-actin antibody to confirm equal loading. The results are representative of three independent experiments. tion (46,47). Furthermore, Laakko and Juliano have found that it takes up to 200 nM of OA in whole Jurkat T cell lysates to increase basal Erk phosphorylation (47), suggesting that the phosphatase involved in this process is different from PP2A since in phosphatase assays, PP2A is inhibited with an IC 50 of 0.1 nM OA (34,39). Together with ours, these results indicate that in Jurkat T cells, the phosphatase that may dephosphorylate Erk is probably different from PP2A since it is sensitive to high concentrations of OA. At high concentrations of OA, it is likely that other phosphatases, such as PP1, PP4, and PP5 that can also act as Erk phosphatases, are inhibited as well (36,38,39). PP2A is a multimeric protein with a catalytic protein (subunit C), a scaffolding protein (subunit A), which are expressed ubiquitously, and a regulatory protein (subunit B) (48). The subunit B is a diverse family of proteins expressed in a tissue-specific manner that determines the substrate specificity and cellular localization of the holoenzyme complex. Therefore, it is possible that the subunit(s) B expressed in Jurkat T cells may regulate PP2A activity in such that Erk1/2 may not be recognized as a substrate, which can be due to a differential subcellular localization of PP2A and Erk1/2.
On the other hand, genetic studies in Drosophila photoreceptor development demonstrated that PP2A had both negative and positive effects on multiple substrates in the MAPK pathway (49), suggesting that the summation of these effects could dictate whether PP2A results in the activation or inactivation of the MAPK/Erk pathway. Indeed, recent studies have indicated that by activating c-Raf-1, PP2A had a positive effect on the activation of MAPK/Erk signaling cascade in macrophages and in Caenorhabditis elegans (35,37). Furthermore, overexpression of PP2A regulatory subunit B␥ promoted neuronal differentiation by activating the MAPK pathway (50). Thus, our results provide another example of positive regulation of MAPK signaling by PP2A. A model by which PP2A can activate Raf-1 has recently been proposed (51). PP2A can promote Raf-1 activation by dephosphorylating its p-S259 that in turn will result in the dissociation of Raf-1 from 14 -3-3 proteins allowing Ras to fully activate Raf-1 (51). Interestingly, ␣ 2 ␤ 1 integrin signaling in Jurkat T cells can also lead to a rapid accumulation of activated p21Ras molecules (52). Together, with our results showing that ␣ 2 ␤ 1 integrin signaling does activate PP2A, it is likely that ␣ 2 ␤ 1 integrin signaling activates the MAPK/Erk in Jurkat T cells according to the proposed model (51). However, this remains to be shown and we cannot exclude that PP2A may also regulate additional steps in the ␣ 2 ␤ 1 integrin signaling cascade such as in integrin inside-out signaling (53).
PP2A has been involved in both cell growth and apoptosis (48). Our results indicating that PP2A is associated with cell survival and probably cell growth downstream of ␣ 2 ␤ 1 integrin signaling are further supported by a recent study showing the potent costimulatory effect that Coll I had on TCR-mediated proliferation of human effector T lymphocytes (54). These studies identify the ␣ 2 ␤ 1 /PP2A/Erk signaling complex as one important modulator of T cell survival and growth. Since PP2A is not involved in the late effects of Coll I on Erk phosphorylation (data not shown), additional signaling events are likely to be involved, which underscores the complexity of Coll I signaling in T lymphocytes.
AICD in T lymphocytes is mediated by activating the Fas/ Fas-L apoptotic pathway in response to TCR/CD3 complex signaling. We have previously shown that Coll I reduces anti-CD3-induced Fas-L expression in Jurkat T cells (20). Thus, with the results presented in this study, it appears that ␣ 2 ␤ 1 integrin signaling blocks AICD not only by interfering with the expression of Fas-L, but also by directly inhibiting the Fas signaling death pathway. The expression of ␣ 2 ␤ 1 integrin on T lymphocytes is associated with late stages of activation that occur in peripheral tissues (17). Thus, our results suggest that in vivo, Coll I can protect activated T cells from AICD in tissues rich in collagen, such as connective tissues of skin and in the course of chronic inflammation such as rheumatoid arthritis (RA). Indeed, RA synovial T lymphocytes do have increased expression of ␣2␤ 1 and ␣ 2 ␤ 1 integrins (55,56) and show resistance to Fas-mediated apoptosis (57). In this context, further elucidation of the signaling events activated by ␣ 2 ␤ 1 integrin in T lymphocytes is likely to provide novel insights into the role of ␣ 2 ␤ 1 integrin signaling in the regulation of immune response and in chronic inflammatory diseases.