α2β1 Integrin Promotes Chemoresistance against Doxorubicin in Cancer Cells through Extracellular Signal-regulated Kinase (ERK)*

Background: Mechanisms by which β1 integrins regulate resistance of cancer cells to chemotherapy. Results: α2β1 integrin but not fibronectin-binding integrins inhibits doxorubicin-induced apoptosis of leukemic T cells via JNK inhibition and Mcl-1 up-regulation in an MAPK/ERK-dependent pathway. Conclusion: α2β1 integrin/ERK pathway promotes chemoresistance of cancer cells. Significance: Activation of this pathway can contribute to the appearance of the drug resistance phenotype. The role and the mechanisms by which β1 integrins regulate the survival and chemoresistance of T cell acute lymphoblastic leukemia (T-ALL) still are poorly addressed. In this study, we demonstrate in T-ALL cell lines and primary blasts, that engagement of α2β1 integrin with its ligand collagen I (ColI), reduces doxorubicin-induced apoptosis, whereas fibronectin (Fn) had no effect. ColI but not Fn inhibited doxorubicin-induced mitochondrial depolarization, cytochrome c release, and activation of caspase-9 and -3. ColI but not Fn also prevented doxorubicin from down-regulating the levels of the prosurvival Bcl-2 protein family member Mcl-1. The effect of ColI on Mcl-1 occurred through the inhibition of doxorubicin-induced activation of c-Jun N-terminal kinase (JNK). Mcl-1 knockdown experiments showed that the maintenance of Mcl-1 levels is essential for ColI-mediated T-ALL cell survival. Furthermore, activation of MAPK/ERK, but not PI3K/AKT, is required for ColI-mediated inhibition of doxorubicin-induced JNK activation and apoptosis and for ColI-mediated maintenance of Mcl-1 levels. Thus, our study identifies α2β1 integrin as an important survival pathway in drug-induced apoptosis of T-ALL cells and suggests that its activation can contribute to the generation of drug resistance.

Integrins are ␣/␤ heterodimeric membrane proteins that mediate cell adhesion to the surrounding extracellular matrix (ECM). 4 In addition to their anchoring function, integrins induce several intracellular signals that modulate cell behavior (1). 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 (2). We and others (3,4) have shown that integrins can also protect the cells against cytokine withdrawal, activation-induced cell death in T lymphocytes (5), and ligation of death receptors in endothelial cells (6). Integrin-mediated cell survival has been associated with the activation of the two major cell survival pathways, the phosphatidylinositol 3-kinase (PI3K)/serine/threonine kinase (AKT) and the MAPK/ERK pathways (1).
Growing evidence suggests that tumor cell interactions with ECM of their microenvironment are important for tumor resistance against chemotherapy-induced apoptosis. However, the mechanisms by which integrins promote chemoresistance in cancer cells are still not fully understood. The ␣4␤1 integrin, which binds fibronectin (Fn) protects hematopoietic malignancies, including myeloma (7) B cell leukemia (8), and myeloid leukemia cell lines (9) from the apoptotic effects of melphalan and Ara-c, and from radiation-induced apoptosis (10).
Malignant T cells such as T cell acute lymphoblastic leukemia (T-ALL) cell lines express several ␤1 integrins, which serve as receptors for collagens, fibronectin and laminins (11). T-ALL is a hematopoietic malignancy, which also grows in the bone marrow that is rich in ECM (12). We have previously demonstrated that collagen type I (ColI) inhibited Fas-induced apoptosis of the Jurkat T-ALL cell line (13). ColI was also shown to protect Jurkat cells from serum starvation-induced apoptosis (14). In contrast, Fn has been shown to induce apoptosis of some leukemic T cell lines (15). Whether ECM proteins regu-late the response of T-ALL cells to chemotherapy is currently unclear. Resistance of cancer cells to apoptosis is a major hurdle in anti-cancer therapies, and understanding how these cells escape apoptosis is likely to lead to new therapeutical avenues.
Chemotherapy-induced apoptosis involves the activation of the mitochondrial death pathway (16), which is regulated tightly by the balance between pro-and antiapoptotic Bcl-2 family proteins (17,18). Activation of Bcl-2 proapoptotic proteins lead to the permeabilization of the mitochondria, and to the release into the cytosol of apoptogenic factors such as cytochrome c. Cytochrome c then participates in the activation of caspase-9, which, in turn, activates executioner caspases.
In this study, we investigated the regulation of doxorubicininduced apoptosis of T-ALL cells by ␤1 integrin signaling. Doxorubicin is a drug widely used in anti-cancer therapy, including in T cell malignancies. We show that engagement of ␣2␤1 integrin by its ligand ColI, inhibited doxorubicin-induced apoptosis of T-ALL cells by inhibiting activation of the c-Jun N-terminal kinase (JNK). This resulted in the maintenance of the prosurvival Mcl-1 levels. The protective effect of ColI is mediated through the activation of the MAPK/ERK survival pathway. In contrast to ColI, Fn, previously shown to be a weak inducer of MAPK/ERK in Jurkat cells (13,19), had no effect on doxorubicin-induced JNK activation, did not maintain Mcl-1 levels and thereby did not protect T-ALL cells from doxorubicin-induced apoptosis. Our study demonstrates an important survival role for ␣2␤1 integrin and its ligand ColI in drug-induced apoptosis of T-ALL cells and suggests that its activation can contribute to the generation of drug resistance.
Cell Culture and Primary T-ALL Blasts-The human T-ALL cell lines Jurkat (E6.1) and HSB-2 were obtained from ATCC (Manassas, VA) and were maintained in RPMI 1640 medium supplemented with 10% of fetal bovine serum (FBS), 2 mmol/ liter glutamine, and 100 units/ml penicillin and streptomycin (complete medium). T-ALL patients were diagnosed and treated at Hôpital Saint-Louis (Paris, France). Informed consent was obtained from the patients or relatives in accordance with the Declaration of Helsinki, and the study was approved by the Hôpital Saint-Louis and Institut Universitaire d'Hematologie Institutional Review Board. The study was carried out with cryopreserved leukemic cells from the bone marrow of patients at diagnosis. Two patients were diagnosed as stage III (cortical immature T-ALL) and one patient as stage IV (mature T-ALL). The expression of ␣2 and ␤1 integrin chains on these T-ALL blasts was determined by the use of phycoerythrin-conjugated anti-␣2 (CD49b) and allophycocyaninconjugated anti-␤1 (CD29) specific antibodies. Samples were analyzed by Canto II flow cytometer (BD Biosciences).
Determination of Apoptosis and Clonogenic Survival-T-ALL cells were resuspended at 1 ϫ 10 6 /ml in RPMI 1640 medium containing 2.5% serum. The cells were then seeded in 24-well plates (5 ϫ 10 5 /well) and activated or not for 4 h with 100 g/ml of ECM (ColI, Fn). The cell cultures were then treated with doxorubicin at 250 ng/ml for Jurkat and HSB-2 T cell lines, and at 600 ng/ml for T-ALL primary blasts. After 16 to 24 h of drug treatment, apoptosis was determined by annexin V staining and flow cytometry analysis using the FACSCalibur cytometer (BD Biosciences) as we described previously (13). Apoptosis also was determined by a cell death detection ELISA kit measuring DNA fragmentation (Roche Applied Science) as we described previously (13).
For clonogenic survival assays, the cells were activated or not with ECMp for 4 h and then treated with doxorubicin for 24 h. The cells were then washed and seeded in complete RPMI medium containing 1% methylcellulose (StemCell Technologies, Vancouver, BC) at 1 ϫ 10 4 cells/ml. After 21 days, colonies with Ͼ40 cells were counted.
Co-culture of Jurkat Cells with Bone Marrow Stromal Cells-Human bone marrow-derived mesenchymal stem cells (MSCs) were a generous gift from Dr. Nicholas Pineault (Hema-Québec, Québec, Canada) and were described previously (20). The co-culture of Jurkat cells with MSCs and drug treatment was carried out as described previously (21,22). MSCs at passage 2 or 3 were seeded in 24 well-plates for 24 h to form a monolayer. Jurkat cells (5 ϫ 10 5 in 500 l of RPMI medium) were kept in suspension or co-cultured with the monolayer of MSCs for 24 h. The cultures were then treated with doxorubicin (250 ng/ml) for 24 h. Jurkat cells were separated from MSCs by pipetting with ice-cold PBS. This treatment did not affect the MSC monolayer nor did it result in the detachment of MSCs. Apoptosis of Jurkat cells in suspension or cocultured with MSCs was then evaluated as described above.
Mitochondrial Depolarization and Cytochrome c Release-Loss of the mitochondrial membrane potential (⌬⌿ m ) was measured by staining the cells with DioC 6 (3) (Molecular Probes, Eugene, OR) and flow cytometry analysis as described previously (24). Cytochrome c release was determined by immunoblot analysis of cytosolic fractions with an anti-cytochrome c antibody (clone 7H8.2C12; BD Pharmingen) as we described previously (25). The purity of the cytosolic fractions was verified by immunoblot analysis using an antibody against ␣2␤1 Integrin in T-ALL Survival the mitochondrial cytochrome c oxidase (antibody 20E8-C12, Molecular Probes) as we described previously (25). We found that only the unbroken cell fraction, which contains the mitochondria and not the cytosolic fraction, contains cytochrome c oxidase (data not shown).
RNA Interference-Silencer-validated siRNA-specific for Mcl-1 (siRNA ID 120644; sense, 5Ј-GGACUUUUAUACCU-GUUAUtt-3Ј; antisense, 5Ј-AUAACAGGUAUAAAAGUC-Ctg-3Ј) and Silencer negative control siRNA were from Ambion and were used in our previous study (23). A second Silencer-validated siRNA sequence targeting Mcl-1 (siRNA ID 4170; sense, 5Ј-CCAGUAUACUUCUUAGAAAtt-3Ј; antisense, 5Ј-UUUCUAAGAAGUAUACUGGga-3Ј) and its Silencer negative control sequence were also from Ambion. 5 ϫ 10 6 Jurkat cells were transfected with 200 nmol/liter of Mcl-1 specific siRNA or negative control siRNA using the nucleofector method from Amaxa according to the manufacturer's instructions. Jurkat cells were mixed with siRNA in the T cell nucleofector solution (solution 5) and transfected using the C016 program. The cells were then cultured in RPMI supplemented with 10% of fetal bovine serum for 6 h. Viable cells were recovered by Ficoll-hypaque density gradient centrifugation and used in subsequent apoptosis experiments. The inhibition of Mcl-1 protein expression was assessed by immunoblot analysis using Mcl-1-specific antibody.
Statistical Analysis-Statistical analysis was performed by the Student's t test. Results with p Ͻ 0.05 were considered significant.

Collagen Type I-␣2␤1 Integrin Inhibits Doxorubicin-induced
Apoptosis-The effects of ColI and Fn, two major matrix proteins of the bone marrow microenvironment, on doxorubicininduced apoptosis were examined in two established T-ALL cell lines, Jurkat and HSB-2. We found that preactivation of Jurkat and HSB-2 cells for 3-4 h with ColI but not with Fn inhibited doxorubicin-induced apoptosis; the percentage of annexin V-positive cells was decreased by 30% (Fig. 1a), and DNA fragmentation was inhibited by ϳ38% (Fig. 1b). Preactivation of the cells for Ͼ4 h with ColI but not with Fn also inhibited doxorubicin-induced apoptosis, whereas treatment of the cells with poly-L-lysine; a non-integrin binding ligand, did not modulate doxorubicin-induced apoptosis (data not shown).
We then tested whether the combination of both matrix proteins would lead to a synergistic cell survival. Preactivation of the cells simultaneously with Fn and ColI only slightly decreased doxorubicin-induced apoptosis in comparison with cells preactivated with ColI alone; however, this did not reach statistical significance (Fig. 1c). Changing the concentrations of ColI or Fn or varying the time of cellular preactivation did not lead to a further decrease in doxorubicin-induced apoptosis of cells preactivated with both ColI and Fn in comparison with cells preactivated only with ColI (data not shown).
Several studies have previously shown that ␣2␤1 integrin is the major ColI receptor expressed on T cells, including T cell lines (5,26,27). Herein, we found that blocking anti-␣2 and anti-␤1 integrin antibodies but not control isotypic antibody significantly block the ability of ColI to reduce doxorubicininduced apoptosis (Fig. 1d). The blocking antibodies had no effect on doxorubicin-induced apoptosis (data not shown). Together these results indicate that the protective effect of ColI involves ␣2␤1 integrin.
Bone marrow-derived MSCs, which are the producers of ECM in the hematopoeitic microenvironment, previously have been shown to protect B cell and myeloid leukemia from chemotherapy (21,22). Therefore, we evaluated the role of MSCs and of ␣2␤1 integrin in the protection of Jurkat cells from doxorubicin-induced apoptosis. In agreement (28), we found that coculture of Jurkat cells with MSCs resulted in a reduction of doxorubicin-induced apoptosis compared with cells grown in suspension (Fig. 1e). In addition, we found that anti-␣2 and anti-␤1 integrin blocking antibodies but not control antibodies also abolished the protective effect of MSCs on Jurkat cells (Fig. 1e).
We then carried out clonogenic survival assays to assess whether ColI promoted long term survival. As shown in Fig. 1f, treatment with doxorubicin led to the formation of very few colonies in both T cell lines. However, treatment with doxorubicin in the presence of ColI but not of Fn led to a significantly higher number of colonies in Jurkat and HSB-2 T cell lines. Taken together, these results indicate that ColI via ␣2␤1 integrin could constitute a major pathway contributing to T-ALL drug resistance.
ColI Inhibits Doxorubicin-induced Mitochondria Depolarization, Cytochrome c Release, and Caspase Activation and Maintains Mcl-1 Protein Levels-The mitochondrial death pathway plays a critical role in drug-induced apoptosis including in doxorubicin-induced apoptosis of T-ALL cell lines (29,30). Thus, we examined whether ColI modulates this apoptotic pathway. Treatment of Jurkat cells with doxorubicin induces the loss of mitochondrial membrane potential, which is reduced by ColI but not by Fn (Fig. 2a). Similarly, ColI but not Fn also reduces doxorubicin-induced cytochrome c release (Fig. 2b). We then examined the regulation of caspase activation. Doxorubicin induces the activation of both caspase-9 and caspase-3 as determined by the reduction in the levels of procaspase forms and in the appearance of caspase-9 and caspase-3 active fragments (Fig. 2c). The presence of ColI reduces the capacity of doxorubicin to activate both caspases, whereas Fn, which had no effect on doxorubicin-induced apoptosis, did not modulate doxorubicin-induced caspase activation (Fig. 2c). ColI but not Fn also reduced doxorubicin-induced caspase activation in HSB-2 cells (supplemental Fig. S1). These results indicate that the prosurvival effect of ColI occurs at the level of the mitochondria.
The balance between pro-and anti-apoptotic Bcl-2 proteins regulates the mitochondrial death pathway and some studies have suggested that integrins can regulate Bcl-2 prosurvival proteins (31,32). Accordingly, we studied the regulation of the three major Bcl-2 prosurvival proteins. Treatment of Jurkat cells with doxorubicin in the presence or absence of ColI had no effect on the protein levels of Bcl-2 and Bcl-x L (Fig. 2d). However, doxorubicin treatment dramatically decreased the levels of Mcl-1, and ColI but not Fn, restored those levels (Fig. 2d).
ColI restored Mcl-1 levels up to 80% in doxorubicin-treated cells (average of three independent experiments). The maintenance of Mcl-1 protein levels by ColI/␣2␤1 integrin signaling ␣2␤1 Integrin in T-ALL Survival also occurs in HSB-2 cells (supplemental Fig. S2). Together, these results suggest that ColI can inhibit doxorubicin-induced mitochondrial signaling and apoptosis by maintaining the levels of Mcl-1.
To test this possibility, we performed an Mcl-1 knockdown by RNA interference and tested the ability of ColI to protect against doxorubicin-induced apoptosis. Transfection of Jurkat cells with Mcl-1-specific siRNA (ID 120644) but not with control siRNA drastically reduced Mcl-1 protein levels (reduction of 85 to 90%; average of three experiments) ( Fig. 2e; left panel). As expected, ColI restored Mcl-1 levels upon doxorubicin treatment in control siRNA-transfected cells but not in Mcl-1 siRNA-transfected cells. In addition, Mcl-1 siRNA enhances doxorubicin-induced mitochondrial membrane depolarization, and more importantly, the protective effect of ColI observed in control siRNA-transfected cells is abrogated in Mcl-1 siRNA-transfected cells (Fig. 2e, right panel). For more specificity, we used a second siRNA sequence targeting Mcl-1 (ID 4170). We found that this sequence also drastically reduces the levels of Mcl-1 and abolishes the protective effect of ColI (supplemental Fig. S3). Similar findings were obtained when apoptosis was measured by annexin V binding (data not shown). Together, these results demonstrate that the protective effect of ColI on doxorubicin-induced mitochondrial membrane depolarization and apoptosis occurs through the maintenance of Mcl-1 protein levels. It is noteworthy that knockdown of Mcl-1 by itself led to an increase in mitochondrial depolarization, which is in line with Mcl-1 as a major regulator of mitochondrial integrity and survival of leukemic cells.
ColI Inhibits Doxorubicin-induced JNK Activation-Previous studies have shown that JNK is involved in doxorubicin-induced apoptosis of T-ALL cells (29,33) and that Mcl-1 stability and levels can be regulated by JNK (34). Thus, we considered the possibility that doxorubicin reduces Mcl-1 levels through a mechanism involving JNK, and that ColI could restore Mcl-1 levels by blocking JNK activation. In agreement, we found that treatment of T-ALL cell lines with doxorubicin increases the phosphorylation of JNK (Fig. 3a), and the JNK inhibitor SP600125 reduces doxorubicin-induced apoptosis by ϳ50 -60% and also reduces doxorubicin-induced caspase-9 activation (supplemental Fig. S4). We then examined the role of JNK in the regulation of Mcl-1 protein levels in doxorubicin-treated cells. The results show that the JNK inhibitor SP600125 inhibits the ability of doxorubicin to down-regulate Mcl-1 protein levels in Jurkat and HSB-2 cells (Fig. 3b, left and right panel, respectively) indicating that JNK activation is essential for the downmodulating effect of doxorubicin on Mcl-1 levels. Finally, we found that ColI but not Fn inhibited the ability of doxorubicin to activate JNK in T-ALL cell lines (Fig. 3c). Together, these results indicate that doxorubicin reduces Mcl-1 protein levels through JNK and ColI could restore these levels by preventing activation of JNK.
ColI-mediated Cell Survival Is MAPK/ERK-dependent-Integrin-mediated signaling results in many cell types into the activation of the PI3K/AKT and MAPK/ERK survival pathways. We have previously shown in T cell lines that engagement of ␣2␤1 integrin with ColI activated the MAPK/ERK but not the PI3K/AKT pathway (13,19). Herein, we found that treatment of Jurkat cells with doxorubicin had no effect on ERK phosphorylation, and ColI induced a significant increase in ERK phosphorylation both in the absence and in the presence of doxorubicin (Fig. 4a). In contrast, doxorubicin or ColI did not regulate AKT, which is constitutively phosphorylated in Jurkat cells (Fig. 4b). These results suggest that the protective effect of ColI could be mediated via the activation of the MAPK/ERK survival pathway.
To test this possibility, we examined the effect of the MEK-1/ERK inhibitor U0126 on ColI-mediated cell survival. Treatment of T-ALL cell lines with U0126 but not with LY294002 (PI3 kinase/AKT inhibitor) abolished the ability of ColI to protect the cells from doxorubicin-induced apoptosis (Fig. 4c). As a control, treatment of the cells with U0126 abolished ColI-induced ERK phosphorylation, and treatment of the cells with LY294002 abolished AKT phosphorylation (data not shown). Because chemical inhibitors could possess off-target effects, we studied the implication of ERK and AKT using a genetic approach. Expression in Jurkat cells of a dominant-negative form of MEK-1 (DN-MEK-1) but not of AKT (DN-AKT) partially reduced the protective effect of ColI. We found that ColI reduced doxorubicin-induced apoptosis by ϳ40% in control transfected cells and only by 15% in DN-MEK-1-transfected cells (Fig. 4d). As we found previously (13,25,35), expression of DN-AKT and DN-MEK-1 also respectively diminished phosphorylations of AKT and ERK (data not shown). Together, these data confirm the implication of MAPK/ERK in ColI-mediated cell survival. We then assessed whether the effects of ColI on the maintenance of Mcl-1 levels and on the inhibition of JNK activation (see Figs. 2d and 3c) also were dependent on MAPK/ERK. The results show that the MEK-1 inhibitor but not the PI3K/AKT inhibitor reduced the ability of ColI to restore Mcl-1 levels in doxorubicin-treated cells (Fig. 4e) and abolished the ability of ColI to inhibit doxorubicin-induced JNK activation (Fig. 4f). Together, these results indicate that ␣2␤1 integrin promotes resistance to doxorubicin by activating the MAPK/ERK signaling pathway.
ColI Protects T-ALL Blasts from Doxorubicin-induced Apoptosis-To assess whether our findings could have a clinical significance, we examined the expression and the function of ␣2␤1 integrin in primary T-ALL blasts isolated from the bone marrow. Three different patient samples were obtained and analyzed in our study. The three samples expressed significant levels of ␣2 integrin chain. Between 49 to 58% of the total cells expressed ␣2 integrin; although with different MFI (high levels of ␣2 integrin were detected on samples 1 and 2, whereas sample 3 expresses lower levels). However, all samples expressed high levels of the ␤1 integrin chain (Fig. 5a). Having established the expression of ␣2␤1 integrin, we tested whether ColI would protect these primary T cell blasts from doxorubicin-induced apoptosis. Preactivation of all three primary T-ALL blasts samples with ColI reduced doxorubicin-induced apoptosis, whereas Fn had no significant effect (Fig. 5b). Doxorubicininduced apoptosis was reduced by 27% in sample 1, 25% in sample 2, and by 15% in sample 3. Finally, the MEK-1 inhibitor U0126 but not the PI3K/AKT inhibitor LY294002 also abolished ColI-mediated cell survival in primary T-ALL blasts (Fig.  5c). Together, these results demonstrate that the protective effect of ColI is not restricted only to T-ALL cell lines and that the ColI-␣2␤1 interaction could be an important cell survival pathway in the resistance of T-ALL toward doxorubicin-induced apoptosis.

DISCUSSION
The tumor microenvironment is recognized as a critical factor promoting tumor progression and survival. The bone marrow, which is rich in ECM proteins, is the growth site for the development of hematological malignancies (12). However, the role of ␤1 integrin signaling in the resistance of malignant T cells to apoptosis is still poorly addressed. In this study, we demonstrate that ColI through its receptor ␣2␤1 integrin inhibits apoptosis of T-ALL cells that is induced by doxorubicin, a chemotherapeutic drug widely used in anti-cancer therapy.  In contrast to ColI, Fn did not protect T-ALL cells from doxorubicin-induced apoptosis despite the fact that malignant T cells express Fn-binding integrins ␣4␤1 and ␣5␤1 (11). However, Fn, through ␣4␤1 integrin, has been shown to reduce drug-induced apoptosis in other hematological maligancies such as myeloma and myeloid leukemia (7,9). This suggests that hematological tumors could respond differently to their tissue microenvironment depending on the integrin expression profile and on the signaling events active in the cells. The fact that Fn does not protect T-ALL cells from doxorubicin-induced apoptosis is reminiscent to previous studies. We have previously demonstrated that in contrast to ColI, Fn did not protect Jurkat cells from Fas-induced apoptosis (13). Engagement of ␣4␤1 integrin with its ligand VCAM-1 has also been shown to enhance TCR-induced apoptosis of antigen-specific T cell clones (36), and Fn via its ␣5␤1 integrin receptor, can induce significant apoptosis in some myeloid and leukemia T cell lines (15). Together, these studies indicate that ␣2␤1 integrin could constitute the major ␤1 integrin survival pathway operating in T-ALL cells.
ColI/␣2␤1 integrin interaction not only inhibited doxorubicin-induced apoptosis but also promoted clonogenic growth of T-ALL cells. Taken together with our previous study that has shown that ␣2␤1 integrin reduces chemotherapy-induced apoptosis in breast cancer cells (25), our results strongly suggest that ␣2␤1 integrin can be an important pathway contributing to the development of drug resistance in T-ALL and other cancer cells.
In our study, we used soluble ColI to ligate ␣2␤1 integrin because T-ALL cell lines attach poorly to immobilized ECM. We and others (13,14,19,27,37) have reported that in Jurkat cells, soluble ColI binds to ␣2␤1 integrin and induces intracellular signaling. Soluble ECM ligands also induce intracellular signaling and modulate cell behavior in solid tumors. Soluble ColI has been shown to influence TGF␤ receptor signaling and gene expression in breast cancer cells (38,39), and soluble integrin ligands such as ColI also rescue neuroblastoma cells from apoptosis under nonadherent conditions (40). ECM remodeling and degradation by metalloproteinases, which is associated with cancer growth and invasion, is likely to lead to the release of ECM components that can act as "soluble" pep-

␣2␤1 Integrin in T-ALL Survival
tides within the tumor microenvironment. T cell malignancies are also associated with metalloproteinases and ECM remodeling as well as dissemination (41)(42)(43)(44). Thus, our study suggests that the ColI released from tissue ECM can bind to ␣2␤1 integrin and provides T-ALL cells with a survival advantage against the cytotoxic effect of doxorubicin. In addition to soluble ColI, we found that co-culture of Jurkat cells with MSCs protected them from doxorubicin-induced apoptosis, which is in agreement with a previous report (28). Moreover, our results showed that the protective effect of MSCs also involves ␣2␤1 integrin. Interestingly, the study of Guo et al. (28) reported that the effect of MSCs on Jurkat cell resistance to doxorubicin involves Notch1 signaling. Together, these studies suggest that MSCs are likely to activate several signaling pathways to support T-ALL resistance to chemotherapy. Although it remains to be tested, it is also possible that a cross-talk exists between Notch1 and ␣2␤1 integrin pathways in T-ALL chemoresistance. Because MSCs are known to produce and to express at their cell surface various ECM proteins, including type I collagen (␣2␤1 integrin ligand) (45), our study indicates that interactions of T-ALL cells with MSCs via ␣2␤1 integrin can constitute one important signaling pathway in their resistance against doxorubicin-induced apoptosis.
The mitochondrial death pathway plays a critical role in drug-induced apoptosis including in doxorubicin-induced apoptosis (29,30). Our results demonstrate that the protective effect of ColI occurs at the level of the mitochondria. We found that ColI reduces mitochondrial membrane depolarization, cytochrome c release, activation of caspase-9 and caspase-3, and maintained the levels of prosurvival Mcl-1 levels in doxorubicin-treated cells. ColI signaling had no effect on the basal levels of Bcl-x L , Bcl-2, and Mcl-1, but it prevented doxorubicin from downregulating Mcl-1 protein levels. In agreement with our results, doxorubicin-induced apoptosis in leukemic cell lines has been associated recently with the down-modulation of Mcl-1 levels and the subsequent activation of the mitochondrial death pathway (30). Knockdown experiments showed that the maintenance of Mcl-1 levels by ColI is necessary for the inhibition of doxorubicin-induced mitochondrial membrane depolarization and apoptosis. Together, these results indicate that ColI/␣2␤1 integrin signaling regulates doxorubicin-induced apoptosis of T-ALL cells by maintaining Mcl-1 levels, which contributes to the protection of mitochondria. Interestingly, Mcl-1 knockdown in T-ALL cells led to an increase in basal cell apoptosis. This effect has also been observed in different tumor cells, including melanoma (23) and myeloma (46), further emphasizing the importance of Mcl-1 in regulating mitochondrial integrity and cell survival.
We also demonstrated that the maintenance of Mcl-1 levels by ColI signaling in doxorubicin-treated cells could be attributed to the inhibition of doxorubicin-induced JNK activation. Indeed, we found that doxorubicin-induced down-regulation of Mcl-1 levels is mediated via JNK and that ColI inhibited doxorubicin-induced JNK activation. Our results showed that JNK is important for doxorubicin-induced caspase activation and apoptosis in T-ALL cells, which is in line with previous studies (29,33). In addition, JNK has also been involved in the degradation of Mcl-1 through its phosphorylation and ubiquitination, a pathway that has recently been shown to play a key role in the sensitization of breast cancer cells to TRAIL by antimicrotubules agents (47) and in the synergy between the Bcl-2 pan-inhibitor ABT-737 and retinamide in the apoptosis of lymphoblastic leukemia cells (48). Thus, our study showed that inhibiting activation of JNK, which led to the maintenance of Mcl-1 levels, is one important mechanism accounting for ␣2␤1 integrin-mediated doxorubicin resistance. Our results demonstrated that the prosurvival effect of ColI is mediated through the activation of the MAPK/ERK but not the PI3K/AKT survival pathway. In agreement with our previous report (13), we found that ColI increases the phosphorylation of ERK but not AKT. We also show that activation of MAPK/ERK is required for the inhibitory effect of ColI on doxorubicin-induced JNK activation. In agreement with our results, activation of MAPK/ERK has been reported to negatively affect the activation of the JNK pathway in leukemic cell lines treated with the proteasome inhibitor bortezomib (49).
We found that Fn, which did not activate the MAPK/ERK pathway (13,19), had no effect on apoptosis, Mcl-1 levels, or JNK activation. This suggests that the differential ability of ␤1 integrins to regulate doxorubicin-induced apoptosis could be due, at least in part, to their differential ability to activate the MAPK/ERK survival pathway. These results further support the role of ColI/␣2␤1 integrin signaling pathway in the resistance of T-ALL cells to drug-induced apoptosis and suggest that targeting ␣2␤1 integrin/MAPK/ERK pathway can be beneficial for the treatment of T-ALL.
Our study reported for the first time that ␣2␤1 integrin can be expressed on primary T-leukemic blasts obtained from patients with T-ALL. We showed that ColI but not Fn protected T-ALL blasts from doxorubicin-induced apoptosis. These results indicate that the prosurvival function of ␣2␤1 is not restricted to T cell lines but could have a clinical implication as well. Analysis of a larger number of samples will be necessary to establish if ␣2␤1 integrin expression and function correlate with drug resistance and patient relapse. In support of our findings, Cleaver et al. (50) recently reported in three different cohorts that ␣2␤1 integrin (VLA-2) mRNA expression levels correlated with the resistance of pediatric T-ALL to the treatment with glucocorticoïds, suggesting that ␣2␤1 can represent an important survival pathway contributing to drug resistance of T-ALL cells. Because of the limited availability of T-ALL samples, we did not examine in details the mechanisms by which ColI protected primary T-ALL blasts from doxorubicininduced apoptosis. However, the fact that JNK has been involved in doxorubicin-induced apoptosis of T-ALL blasts (29) and that we found that the MEK-1/ERK inhibitor abrogated the protective effect of ColI in primary T-ALL blasts, argues in favor of the possibility that ColI-mediated protection in primary T cell blasts, occurs through the same mechanism demonstrated in T-ALL cell lines.
Thus, our study has unraveled a ␤1 integrin survival signaling pathway operating in T-ALL cells in which ligation of ␣2␤1 integrin with ColI inhibits doxorubicin-induced apoptosis (depicted in Fig. 6). Doxorubicin activates JNK, which induces the down-modulation of Mcl-1 levels, thus allowing the activation of the mitochondrial death pathway (cytochrome c release and caspase activation). Ligation of ␣2␤1 integrin with ColI activates the MAPK/ERK survival pathway, which inhibits doxorubicin-induced JNK activation, leading to the maintenance of Mcl-1 levels. In turn, Mcl-1 protects the cells from doxorubicin-induced mitochondrial cell death, thus promoting doxorubicin resistance. The mechanisms by which ColI inhibits doxorubicin-induced JNK activation are not clear, but a previous study reported that in Jurkat T cells, doxorubicin-induced JNK activation is dependent on caspase-2 and PKC␦ activities (29). Thus, it is tempting to speculate that the inhibitory effect of ColI could be at the level of either caspase-2 or PKC␦. These studies are currently underway in our laboratory.
Further elucidation of the mechanisms by which ␣2␤1 integrin signaling regulates drug-induced apoptosis in T-ALL cells and other cancer cells is likely to lead to new therapeutic avenues.