Tissue Inhibitor of Metalloproteinase-1 Protects Human Breast Epithelial Cells Against Intrinsic Apoptotic Cell Death via the Focal Adhesion Kinase/Phosphatidylinositol 3-Kinase and MAPK Signaling Pathway*

Tissue inhibitor of metalloproteinase (TIMP-1) is a natural protease inhibitor of matrix metalloproteinases (MMPs). Recent studies revealed a novel function of TIMP-1 as a potent inhibitor of apoptosis in mammalian cells. However, the mechanisms by which TIMP-1 exerts its anti-apoptotic effect are not understood. Here we show that TIMP-1 activates cell survival signaling pathways involving focal adhesion kinase, phosphatidylinositol 3-kinase, and ERKs in human breast epithelial cells to TIMP-1. TIMP-1-activated cell survival signaling down-regulates caspase-mediated classical apoptotic pathways induced by a variety of stimuli including anoikis, staurosporine exposure, and growth factor withdrawal. Consistently, down-regulation of TIMP-1 expression greatly enhances apoptotic cell death. In a previous study, substitution of the second amino acid residue threonine for glycine in TIMP-1, which confers selective MMP inhibition, was shown to obliterate its anti-apoptotic activity in activated hepatic stellate cells suggesting that the anti-apoptotic activity of TIMP-1 is dependent on MMP inhibition. Here we show that the same mutant inhibits apoptosis of human breast epithelial cells, suggesting different mechanisms of TIMP-1 regulation of apoptosis depending on cell types. Neither TIMP-2 nor a synthetic MMP inhibitor protects breast epithelial cells from intrinsic apoptotic cell death. Furthermore, TIMP-1 enhances cell survival in the presence of the synthetic MMP inhibitor. Taken together, the present study unveils some of the mechanisms mediating the anti-apoptotic effects of TIMP-1 in human breast epithelial cells through TIMP-1-specific signal transduction pathways.


Tissue inhibitor of metalloproteinase (TIMP-1) is a natural protease inhibitor of matrix metalloproteinases (MMPs)
. Recent studies revealed a novel function of TIMP-1 as a potent inhibitor of apoptosis in mammalian cells. However, the mechanisms by which TIMP-1 exerts its anti-apoptotic effect are not understood. Here we show that TIMP-1 activates cell survival signaling pathways involving focal adhesion kinase, phosphatidylinositol 3-kinase, and ERKs in human breast epithelial cells to TIMP-1. TIMP-1-activated cell survival signaling down-regulates caspase-mediated classical apoptotic pathways induced by a variety of stimuli including anoikis, staurosporine exposure, and growth factor withdrawal. Consistently, down-regulation of TIMP-1 expression greatly enhances apoptotic cell death. In a previous study, substitution of the second amino acid residue threonine for glycine in TIMP-1, which confers selective MMP inhibition, was shown to obliterate its anti-apoptotic activity in activated hepatic stellate cells suggesting that the anti-apoptotic activity of TIMP-1 is dependent on MMP inhibition. Here we show that the same mutant inhibits apoptosis of human breast epithelial cells, suggesting different mechanisms of TIMP-1 regulation of apoptosis depending on cell types. Neither TIMP-2 nor a synthetic MMP inhibitor protects breast epithelial cells from intrinsic apoptotic cell death. Furthermore, TIMP-1 enhances cell survival in the presence of the synthetic MMP inhibitor. Taken together, the present study unveils some of the mechanisms mediating the anti-apoptotic effects of TIMP-1 in human breast epithelial cells through TIMP-1-specific signal transduction pathways.
Cell interactions with extracellular matrix (ECM) 1 greatly influence cell survival, and removal of anchorage-dependent cells from their association with the ECM results in apoptotic cell death, known as anoikis (1,2). Cell-ECM interaction-mediated signal transduction is regulated in part by the composition and integrity of the ECM and actions of its components on specific cell adhesion receptors (3)(4)(5)(6). Integrity and turnover of the ECM are in part regulated by matrix metalloproteinases (MMPs), a family of zinc-dependent endopeptidases (7)(8)(9) known to accomplish the degradation of ECM components. Four members of the tissue inhibitor of metalloproteinase family (TIMP-1 to -4) have been identified as natural inhibitors of MMPs.
Previous studies in our laboratory (10) showed that bcl-2 overexpression is associated with enhanced levels of TIMP-1 expression in human breast epithelial cells, suggesting a role for TIMP-1 in apoptosis. Indeed, apoptosis studies showed that TIMP-1 protects against a variety of apoptotic stimuli including anoikis, hydrogen peroxide, x-ray irradiation, and adriamycin treatment (10). Furthermore, TIMP-1 inhibition of apoptosis in human breast epithelial cells involves focal adhesion kinase (FAK)-mediated cell survival signaling rather than regulation of cell-ECM interactions via MMP activity.
During the past several years, investigators have shown the role of TIMPs in apoptosis regulation. TIMP-1 inhibits apoptosis in many cell types including activated hepatic stellate cells (11,12), TIMP-1-negative Burkitt's lymphoma cell lines (13,14), human breast epithelial cells (10), and mammary epithelial cell in transgenic mice (15), whereas TIMP-3 enhances apoptosis in retinal pigment epithelial and MCF-7 cells (16,17), human DLD colon carcinoma cells (18,19), rat vascular smooth muscle cells (20,21), melanoma SK-Mel-5 and A2058 cells (22), and cancer cell lines such as HeLa (21) and HT1080 cells (23). TIMP-3 induction of human colon carcinoma cell apoptosis is mediated by protecting tumor necrosis factor-␣ receptor on the cell surface against MMP-mediated cleavage (18,19). In contrast, a recent study (24) showed increased epithelial cell apoptosis in TIMP-3-deficient mice during mammary gland involution, suggesting an anti-apoptotic activity of TIMP-3. Both pro-and anti-apoptotic activities of TIMP-2 and TIMP-4 have been reported depending on the cell type. For example, TIMP-2 and TIMP-4 enhance apoptosis in human T lymphocytes and cardiac fibroblasts, respectively (25,26), whereas TIMP-2 enhances cell survival of metanephric mesenchymes (27) and B16F10 melanoma cells (28), and TIMP-4 protects human breast cancer cells, as well as mammary tumor xenografts, from apoptotic cell death (29). Although it has become clear that TIMPs are critical determinants for cell survival, the signaling mechanisms by which TIMPs control cell survival in a cell type-specific manner remain undefined. Also, it is still controversial whether the anti-apoptotic effect of TIMP-1 depends on its ability to inhibit MMP activity. In Burkitt's lymphoma cell lines TIMP-1 inhibition of apoptosis was shown to be independent of its inhibitory activity (13,14). However, a mutant TIMP-1 devoid of inhibitory activity against certain MMPs had no anti-apoptotic effect in hepatic stellate cells (11).
In this study, we further examined the TIMP-1-mediated cell survival pathways and its significance in TIMP-1 regulation of apoptosis using breast epithelial MCF10A cells engineered to express high or low levels of TIMP-1. We also investigated whether TIMP-1 inhibition of apoptosis in these cells was dependent or independent of MMP inhibitory activity.
Plasmid Constructs and Transfection-Antisense TIMP-1 construct (AS TIMP-1) was generated by EcoRI digestion and re-ligation from its sense construct containing the human full-length TIMP-1 cDNA and the neomycin resistance gene, under control of the long terminal repeats of the Moloney murine sarcoma virus (kindly provided by Dr. M. Johnson at Northwestern University, Chicago, IL). To construct a TIMP-1/FLAG fusion cDNA, BamHI sites were introduced to both 5Ј and 3Ј ends of the human full-length TIMP-1 cDNA fragment by the polymerase chain reaction using primers TIMP-1AA BamHI, 5Ј-AAG GAT CCA TGG CCC CCT TTG AGC CCC TGG-3Ј, and TIMP1-207AA BamHIA, 5Ј-AAG GAT CCG GCT ATC TGG GAC CGC AGG GAC-3Ј. The fragment was digested with BamHI and cloned into the BamHI site in the p3XFLAG-CMV-14 vector (Sigma). To substitute Thr-2 with Gly at the mature TIMP-1 protein (after cleavage of the signal peptide), site-directed mutagenesis was performed using primers T1T2G, 5Ј-CCA GCA GGG CCT GCG GGT GTG TCC CAC CCC A-3Ј, and T1T2GAA, 5Ј-TGG GGT GGG ACA CAC CCG CAG GCC CTG CTG G-3Ј, and a TIMP-1/FLAG construct as a template as instructed by the manufacturer (Stratagene). DNA sequencing analysis confirmed the fidelity of the constructs. Hereafter, the wild type TIMP-1/FLAG and T2G mutated TIMP-1/FLAG constructs are referred to as WT TIMP-1/FLAG and T2G TIMP-1/FLAG. Generation of TIMP-1 overexpressing MCF10A clones was described previously (10). AS TIMP-1 construct was transfected into MCF10A cells as previously described (10), and WT TIMP-1/FLAG, T2G TIMP-1/FLAG, and FLAG control constructs were transfected, respectively, into both MCF10A cells and human embryonic kidney 293 cells using LipofectAMINE2000 (Invitrogen) according to the manufacturer's protocol. The cells were subjected to 400 g/ml G418 antibiotic selection for 14 days, and at least six colonies from each transfection were isolated for further analysis.
Immunoblot Analysis-Cell lysates were obtained by lysing the cell monolayer with SDS lysis buffer (2% SDS, 125 mM Tris-HCl, pH 6.8, and 20% glycerol). The lysates were boiled for 5 min and then clarified by a 20-min centrifugation at 4°C. Protein concentration was measured using the BCA protein assay reagent (Pierce). Equal amount of protein samples in SDS sample buffer (1% SDS, 62.5 mM Tris-HCl, pH 6.8, 10% glycerol, 5% ␤-mercaptoethanol, and 0.05% bromphenol blue) were boiled for 5 min and subjected to reducing SDS-PAGE. After electrophoresis, the proteins were transferred to a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk in T-TBS (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.02% NaN 3 , and 0.2% Tween 20) for 1 h at room temperature. The membranes were incubated with T-TBS containing 5% milk and in the appropriate antibodies. After three washes with T-TBS, the blot was incubated with the appropriate horseradish peroxidase-conjugated secondary antibodies. The antigen was detected using the Western blot Chemiluminescence Reagent Plus (PerkinElmer Life Sciences), according to the manufacturer's instruction.
Apoptosis Induction-Apoptosis was induced by culturing cells in serum-free medium (growth factor withdrawal) or by treatment with staurosporine. Anoikis (apoptosis induced by loss of cell adhesion) was induced by culturing cells in polyHEMA-coated 6-well plates, which prevents cell adhesion (34).
Evaluation of TIMP-1 Effects on Cell Survival Using MTT Assay-Cell number was determined by an indirect colorimetric immunoassay (MTT assay). In brief, the cells (4 ϫ 10 3 cells/well) were plated in a 96-well culture plate overnight followed by culture in serum-free medium for 48 h in the absence or presence of 500 ng/ml TIMP-1, 500 ng/ml TIMP-2, 10 M PD98059, 10 M SB202474, 200 nM wortmannin, 50 nM LY294002, 50 nM LY303511, 5 M BB-94, or vehicle (Me 2 SO). MTT (0.5 mg/ml) was then added, and the plates were incubated for 4 h at 37°C. The cellular formazan was extracted with acidic isopropanol, and the absorbance of the converted dye was measured at a wavelength of 570 nm, with background subtraction at 650 nm, using a Bio-Rad Benchmark microplate reader (Bio-Rad).
Purification of FLAG-tagged TIMP-1 Proteins-Human embryonic kidney 293 cells that stably expressed WT TIMP-1/FLAG or T2G TIMP-1/FLAG, were cultured in phenol red-free conditioned medium for 48 h. The conditioned medium was clarified by centrifugation and concentrated 5-fold using Centricon YM-10 (Millipore Corp.). Twenty ml of the conditioned medium were incubated with 120 l of EZview Red anti-FLAG M2 affinity gel (Sigma) at 4°C. After the resin was collected and washed with TBS, the proteins were eluted with TBS buffer containing 3ϫ FLAG peptide (150 g/ml). Proteins were then analyzed by reducing SDS-PAGE and stained with Simply Blue SafeStain (Invitrogen). The fidelity of the T 3 G TIMP-1 mutation protein was confirmed by N-terminal microsequencing of affinity purified T2G TIMP-1/FLAG (Proseq, Inc., Boxford, MA). The concentrations of wild type TIMP-1/ FLAG and T2G TIMP-1/FLAG were estimated by densitometry after immunoblot analysis and by titration with a MMP-9 solution with known concentration.
MMP Inhibition Studies-The inhibitory activity of WT TIMP-1/ FLAG, T2G TIMP-1/FLAG, and TIMP-1 produced by the vaccinia expression system was tested with MMP-2, MMP-9, MMP-3, and MMP-7. The active enzymes (MMP-2 and MMP-9) were obtained by incubating pro-MMP-2 and pro-MMP-9 with 1 mM p-aminophenylmercuric acetate dissolved in 200 mM Tris at 37°C in Buffer A (50 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , 0.01% Brij-35, pH 7.5). The activation reaction was monitored with the fluorescence-quenched peptide MOCAcPLGLA 2 pr(Dnp)AR-NH 2 , as will be described below. p-Aminophenylmercuric acetate was removed by dialysis against collagenase buffer (50 mM Tris, 150 mM NaCl, 5 mM CaCl 2 , 0.02% Brij-35, pH 7.5) at 4°C. The enzyme concentrations were determined by active-site titration with TIMP-1 solutions of known concentration. The concentration of MMP-3 and MMP-7 was determined by titration with TIMP-1 and TIMP-2, respectively. The enzymatic assays were carried out at 25°C, using a Photon Technology International spectrofluorometer, equipped with the RadioMaster TM and FeliX TM hardware and software, respectively. Excitation and emission band passes of 1 and 3 nm, respectively, were used. Fluorescence measurements were taken every 4 s. MMP-2, MMP-9, and MMP-7 enzymatic activities were monitored with the fluorogenic synthetic substrate MOCAcPLGLA 2 pr(Dnp)AR-NH 2 , at excitation and emission wavelengths of 328 and 393 nm, respectively. MOCAcRPKPVE(Nva)WRK(Dnp)NH 2 was the substrate used to monitor MMP-3 activity at ex ϭ 325 and em ϭ 393 nm. All fluorogenic substrates were obtained from Peptides International Inc., Louisville, KY. MMP inhibition was monitored after incubating the enzyme (0.2-0.7 nM) with increasing concentrations of the inhibitor at 37°C in Buffer A for at least 1 h. The remaining enzymatic activity was monitored with the appropriate fluorogenic substrate (5-7 M) at 25°C. The initial rates of the enzyme reaction with the fluorogenic substrate were determined by linear regression analysis of the fluorescence versus time traces using FeliX TM . These rates were analyzed according to a competitive model of inhibition (35) using the program SCIENTIST (Micro-Math Scientific Software, Salt Lake City, UT), yielding K i (app) values, because TIMP-1 has been shown to be a tight, slow binding inhibitor of the MMPs (32,36), and the association and dissociation rate constants (k on and k off , respectively) were not determined.

Down-regulation of TIMP-1 Expression in MCF10A Cells and
Its Effects on Apoptosis-Our previous study (10) showed that TIMP-1 overexpression protects MCF10A cells from apoptosis induced by a variety of stimuli. To further investigate TIMP-1 regulation of human breast epithelial cell survival, we generated MCF10A clones in which endogenous TIMP-1 expression is down-regulated by an antisense construct (AS TIMP-1 MCF10A). Immunoblot analysis confirmed significant downregulation of TIMP-1 in AS TIMP-1 MCF10A cells and upregulation of TIMP-1 in TIMP-1 overexpressing MCF10A clone 29 (T29) cells, when compared with the neo-control MCF10A cells (Fig. 1, A and B). To test the consequences of different levels of TIMP-1 expression on cell death, we compared the percentage of cell survival of neo-MCF10A (Neo), AS TIMP-1 MCF10A (AS), and TIMP-1 overexpressing MCF10A clone 29 (T29) cells, following loss of cell adhesion. As shown in Fig. 1C, only ϳ25% of AS TIMP-1 MCF10A cells remained viable after 18 h of anoikis, whereas ϳ80% of control vector-transfected and ϳ95% of T29 MCF10A cells were viable. After 48 h, almost none of AS TIMP-1 MCF10A and only ϳ15% neo-MCF10A cells survived, whereas ϳ85% of T29 MCF10A cells remained viable. These results further confirmed our previous finding (10) that TIMP-1 levels are critical determinants for apoptosis sensitivity in human breast epithelial cells.
We next examined whether exogenously added TIMP-1 could protect AS TIMP-1 MCF10A cells from cell death. We also tested TIMP-2, a close homologue of TIMP-1, for its effect on apoptosis. As shown in Fig. 1D, TIMP-1 significantly enhanced AS TIMP-1 MCF10A cell survival (ϳ5-fold increase) following growth factor withdrawal, whereas TIMP-2 had little effect.
Compared with AS TIMP-1 MCF10A cells, TIMP-1 effect on neo-MCF10A cell survival was less drastic (ϳ2-fold). This is likely to be because of the fact that the neo-MCF10A cells express higher levels of endogenous TIMP-1 than AS TIMP-1 MCF10A cells (Fig. 1, A and B). Considering that both TIMP-1 and TIMP-2 inhibit MMP activity, these results suggest that inhibition of MMP activity is unlikely to be a key mechanism for the anti-apoptotic effects of TIMP-1.

TIMP-1 Activation of PI 3-Kinase and MAPK Survival Signaling Pathways Are Critical for Its Apoptosis Regulation-Our
previous study demonstrated that TIMP-1 overexpression results in phosphorylation of FAK regardless of cell adhesion. FAK associates with a number of signaling molecules and activates the survival signaling cascades, which include PI 3-kinase and MAPK family members. To investigate the significance of TIMP-1-mediated FAK phosphorylation, we examined whether the anti-apoptotic action of TIMP-1 is mediated via the activation of PI 3-kinase and/or MAPK signaling pathways. As shown in Fig. 3, A and C, the levels of active ERKs and PI 3-kinase are significantly higher in T29 MCF10A cells compared with neo-MCF10A cells. In contrast, TIMP-1 overexpression had little effect on p38 activation in these cells, whereas it slightly enhanced the levels of active JNK1/2. To exclude the possibility that the increased levels of active PI 3-kinase and ERKs resulted from clonal selection and/or long term cell culture with stable TIMP-1 expression, AS TIMP-1 MCF10A cells were treated with recombinant TIMP-1 and then examined for ERKs and PI 3-kinase activation. As shown in Fig. 3, B and D, TIMP-1 effectively activated ERKs and PI 3-kinase in 10 min, whereas the peak level of phospho-ERK is at 30 min after TIMP-1 treatment.
To evaluate the role for PI 3-kinase in TIMP-1 inhibition of cell death, we examined the ability of TIMP-1 to enhance cell survival in the presence and absence of wortmannin or LY294002, inhibitors of PI 3-kinase (39). As shown in Fig. 4A, TIMP-1-mediated cell survival was drastically reduced by wortmannin and LY294002 but not LY303511 (a negative control for LY294002) in T29 MCF10A cells. Similarly, wortmannin and LY294002, but not LY303511, abrogated the protective effect of exogenously added TIMP-1 in AS TIMP-1 MCF10A cells without evidence of toxicity (Fig. 4B). We also examined the effect of PD98059 (an inhibitor of MAPK kinase) on TIMP-1 inhibition of cell death. Inhibition of MAPK pathway reduced T29 MCF10A cell survival following growth factor withdrawal but not in the presence of SB202474 (a negative control for PD98059) (Fig. 4A). Similarly, exogenous TIMP-1-mediated AS MCF10A cell survival was significantly reduced in the presence of PD98059 but not in the presence of SB202474 (Fig. 4B). It should be mentioned that TIMP-1 has little mitogenic activity in MCF10A cells, although it effectively activates MAPK and PI 3-kinase. Surprisingly, overexpression of TIMP-1 reduced growth rate in these cells (data not shown). Taken together, these results indicate that TIMP-1-enhanced cell survival results from activation of signaling molecules including PI 3-kinase and ERKs that are critical for apoptosis inhibition.
The Antiapoptotic Effect of TIMP-1 in Human Breast Epithelial Cells Is Independent of MMP Inhibition-A previous study in Burkitt's lymphoma cell lines showed that inhibition of apoptosis by TIMP-1 was independent of its MMP inhibitory activity (10,14). However, a recent study in hepatic stellate cells showed that a T2G TIMP-1 mutant, shown previously to be practically non-inhibitory toward MMP-1, MMP-2, and MMP-3 (K i in the M range) (40), lost its ability to inhibit apoptosis (11). To examine whether MMP inhibition was involved in TIMP-1 anti-apoptotic activity, we tested the T2G TIMP-1 mutant in MCF10A cells. In addition, we tested the anti-apoptotic activity of the broad spectrum synthetic MMP inhibitor BB-94 and the natural MMP inhibitor TIMP-2. To this end, we generated MCF10A clones expressing WT or T2G TIMP-1/FLAG proteins, and the expression of the recombinant proteins was confirmed by immunoblot analysis (Fig. 5, A and  B). When the effect of WT and T2G TIMP-1 on cell survival was examined after 48 h of anoikis induction, 60 -70% of both WT and T2G TIMP-1/FLAG expressing MCF10A cells remained viable. In contrast, only ϳ25% of control vector-transfected MCF10A cells remained viable under the same conditions (Fig.  6). Consistently, DEVDase activity following anoikis and staurosporine treatment was greatly inhibited in both WT and T2G TIMP-1/FLAG MCF10A cells (Fig. 7, A and B). Thus, unlike the results with the hepatic stellate cells (11), breast epithelial MCF10A cells are protected against apoptosis by a TIMP-1 protein bearing a Gly substitution at Thr-2. Because our apoptosis data disagreed with the apoptosis study using the hepatic stellate cells (11), we examined the inhibitory capacity of an affinity purified T2G TIMP-1/FLAG against various MMPs. Previously (40), the T2G TIMP-1 mutant was shown to be inactive against MMP-1, MMP-2, and MMP-3. Here we show that the T2G TIMP-1/FLAG is an effective inhibitor of MMP-2, MMP-9, and MMP-7 (K i (app) in the nano-or picomolar range) ( Table I) when compared with both the wild type TIMP-1 or the WT TIMP-1/FLAG, thus indicating that the incorporation of the FLAG epitope at the C terminus had no effect on inhibitory activity. In contrast, the T2G TIMP-1/FLAG was a very poor inhibitor of MMP-3 (K i (app) ϭ 0.486 ϩ 0.045 M). Thus, in our hands, the T2G TIMP-1 mutant shows selective MMP inhibition exhibiting high affinity for gelatinases and MMP-7 and low affinity for MMP-3, in disagreement and agreement, respectively, with the previous study (40). Because the T2G TIMP-1 mutant exhibits selective MMP inhibition, an involvement of MMP inhibition in the anti-apoptotic effects of TIMP-1 cannot be ruled out by using this mutant only. Thus, to further investigate whether MMP inhibition has any anti-apoptotic effect in MCF10A cells, we cultured AS TIMP-1 MCF10A cells in serumfree medium for 48 h in the presence of TIMP-1, TIMP-2, or BB-94. These studies showed that, whereas TIMP-1 almost completely prevented AS TIMP-1 MCF10A cell death, as expected, BB-94 and TIMP-2 had little effect on survival (Fig. 8). Furthermore, in the presence of BB-94, exogenously added TIMP-1 proteins, but not TIMP-2, effectively prevented MCF10A cell death. Collectively, these studies suggest that the anti-apoptotic effect of TIMP-1 in breast epithelial cells is unlikely to result from MMP inhibition. DISCUSSION TIMP-1 is well known for its ability to inhibit MMP activity, thereby inhibiting tumor growth and metastasis. However, recent studies show that TIMP-1 expression is associated with resistance to apoptosis in a variety of cell types (10,13,41). Consistently, our present study shows that the apoptotic potential of cultured non-malignant human breast epithelial cells inversely correlates with TIMP-1 expression levels, and administration of TIMP-1 reverses apoptotic sensitivity, demonstrating the anti-apoptotic activity of the TIMP-1 protein. These results raise important basic questions regarding the biological role of TIMP-1 in breast epithelial cells. For example, in breast cancer, TIMP-1 expression has been associated with poor prognosis, an unexpected finding considering that TIMP-1 is a known MMP inhibitor (42)(43)(44). Our study provides new clues about TIMP-1 function in breast neoplasia, exerting potential oncogenic activity in breast epithelium through apoptosis inhibition. TIMP-1-enhanced cell survival in response to intrinsic apoptotic stimuli appears to be epithelial cell-specific, because it had no anti-apoptotic effect on either murine embryonic fibroblasts or normal rat kidney fibroblast cells (data not shown). Tissue-specific regulation of apoptosis by TIMP was also suggested by studies with TIMP-4. Systemic TIMP-4 gene delivery inhibits apoptosis and enhances mammary tumorigenesis in mice (29), whereas TIMP-4 induces apoptosis in cardiac fibroblasts (26). Anti-apoptotic activity of TIMP-4 was suggested to be independent of its inhibition of the enzymatic activities of MMPs, because a synthetic MMP inhibitor had no effect on TIMP-4 inhibition of apoptosis in breast cancer cells in mice and in vitro (29).
When the anti-apoptotic activity of TIMP-1 was compared with bcl-2 activity, TIMP-1 overexpression was as effective as bcl-2 overexpression in protecting MCF10A cells against cell death induced by hydrogen peroxide, adriamycin, and anoikis (10). TIMP-1 effectively inhibited apoptosis induced by staurosporine, which rapidly decreases the transmembrane potential of the mitochondria, resulting in activation of the intrinsic caspase cascade within 2 to 4 h. The ability of TIMP-1 to protect breast epithelial cells against apoptosis induced by agents working through different mechanisms suggests that TIMP-1 may regulate the apoptosis commitment step, a common pathway regulated by members of the bcl-2 family. Consistently, our previous and present studies have indicated that FAK, PI 3-kinase, and MAPK pathways are critical for TIMP-1 inhibition of apoptosis. It was demonstrated previously that constitutively activated forms of FAK mediates cell survival signaling, protecting cells against anoikis (45) and free radicalinduced cell death (46). A well studied signaling pathway mediated by FAK involves PI 3-kinase leading to regulation of the bcl-2 family members. Phosphorylated FAK binds to PI 3-kinase and activates its activity, which in turn activates Akt (47,48). Akt phosphorylates bad, a pro-apoptotic member of the bcl-2 family, at the Ser-136 residue (49). The phosphorylated form of bad no longer interacts with bcl-2 or bcl-X L (antiapoptotic members of the bcl-2 family), resulting in bcl-2 and bcl-X L activation. Similarly, ERKs, MAPK family members, phosphorylate the bad protein at the Ser-112 residue, resulting in bcl-2 and bcl-X L activation (50). Akt and ERKs also phosphorylate transcription factors, translation factors, and other kinases (51,52), which is critical for regulation of gene expression, protein synthesis, and cell cycle progression, thereby regulating cell proliferation and survival. The present study provides the basis to further investigate TIMP-1-mediated survival signaling pathways critical for the apoptosis commitment step.
Although 150 -200 ng/ml of rTIMP-1 was required for maximum protection of MCF10A cells from apoptotic cell death (data not shown), 500 ng/ml of rTIMP-1 was used in this study to ensure its maximum effects. Previous studies (13,53) reported varying concentrations (5-250 ng/ml) of rTIMP-1 as necessary for apoptosis inhibition. Because the sources of rTIMP-1 proteins are different, and direct comparison of biological activities of these rTIMP-1 proteins is not available, it is difficult to assess the significance of the differences in rTIMP-1 concentration necessary for apoptosis inhibition. However, it should be mentioned that the TIMP-1 protein concentration necessary for the regulation of apoptosis may be determined not only by the biological activity of rTIMP-1 itself but also by expression levels of TIMP-1 interacting protein(s) on the cell surface and/or expression levels of endogenous MMPs. The anti-apoptotic activity of exogenously added TIMP-1 was shown previously (14,54) to be abolished by a neutralizing antibody against TIMP-1. We also found that incubation of rTIMP-1 with this antibody significantly reduced its anti-apoptotic activity (data not shown). However, it was unsuccessful to test the neutralizing activity of this antibody on endogenously expressed TIMP-1 in MCF10A cells, likely because of complications associated with its nonspecific interactions with other proteins besides TIMP-1.
Here we showed that among the MAPK family members, ERKs are effectively activated in MCF10A cells following treatment with recombinant TIMP-1 proteins for 10 min, whereas the same treatment had little effect on p38 kinase and JNK. It should be mentioned that both TIMP-1 and TIMP-2 can activate ERKs in MCF10A cells (data not shown), although only TIMP-1, and not TIMP-2, inhibits apoptosis. These results suggest that ERKs are not solely responsible for TIMP-1 inhibition of apoptosis, although ERKs activation is required for TIMP-1 inhibition of apoptosis as evidenced by studies using PD98059 (Fig. 4). In stable TIMP-1 transfectants, the basal levels of JNK1 decreased with slightly higher levels of active JNK1 (Fig. 3A). TIMP-1 activation of MAPK family members appears to vary, depending on cell-types. TIMP-1 activates p38 kinase and JNK in human leukemic cell line UT-7 (55,56), whereas it activates ERKs in human osteosarcoma cell line MG-63 (57) as in MCF10A cells. These kinases were shown to function cooperatively or independently in various biological processes, such as cytoskeletal regulation, cell motility, cell proliferation, and apoptosis (51,52,58). Whereas ERKs mediate mostly the cell-survival pathway, JNKs and p38 often function as pro-apoptotic signaling molecules. Apoptotic and antiapoptotic stimuli often activate both pro-and anti-apoptotic signaling pathways simultaneously, and the cell fate is determined by the balance and strength of those pathways in a cell  type-specific manner. At present, it is unclear how TIMP-1 differentially activates MAPK kinase members and how TIMP-1-activated FAK/PI 3-kinase/Akt and MAPK (ERKs, JNKs, and p38) pathways are integrated to agonize or antagonize the apoptotic process in a cell type-specific manner.
TIMP-1 is a classical inhibitor of metalloproteinases, thus it is logical to assume that some of its biological effects may be mediated by inhibition of enzymatic activity. Indeed, MMPs are known to affect cell behavior by cleaving a vast array of substrates, which are not limited to ECM components and include various growth factors and their receptors, cytokines, and cell adhesion molecules (59). TIMP regulation of apoptosis through MMP inhibition is supported by the studies with TIMP-3. TIMP-3 enhances extrinsic cell death by inhibiting the shedding of the cell surface death receptors mediated by tumor necrosis factor-␣ converting enzyme/a distintegrin and metalloprotease-17 (18,19,60,61). TIMP-regulation of apoptosis through MMP inhibition was also suggested by studies with T2G TIMP-1 mutant, because this mutant, reported to lack MMP inhibitory activity (11), failed to protect hepatic stellate cells against apoptosis. Although anti-MMP activity may play a role in mediating the pro-and anti-apoptotic effects of some TIMPs, this effect cannot be generalized. First, even though TIMPs are closely related proteins, each one elicits different biological activities. Second, the biological effects of TIMPs may vary according to the tissue or cell type. Indeed, our data suggest that in breast epithelial cells inhibition of MMP activity may not be involved in the anti-apoptotic effects of TIMP-1, whereas in hepatic stellate cells MMP inhibition appears to be essential (11). These different results may be ascribed to cell type differences involving different signaling pathways and/or differences in MMP expression profiles, if MMP activity are involved in apoptosis. TIMPs are known as general MMP inhibitors; however they exhibit significant selectivity in their specificity toward the various members of the MMP family. For example, TIMP-2 is an efficient inhibitor of both secreted and membrane type-MMPs, whereas TIMP-1 is a poor inhibitor of the latter (62,63). TIMP-1 and TIMP-3 inhibit a distintegrin and metalloproteases more effectively than TIMP-2 (64). Likewise, mutations at critical residues of TIMP-1 inhibitory domain produce selective MMP inhibition (40). We showed that the T2G TIMP-1 is an efficient inhibitor of gelatinases when compared with the wild type inhibitor but is a very weak MMP-3 inhibitor. Because of this selectivity and the limited number of MMPs tested, we cannot conclude, based on this mutant alone, that inhibition of apoptosis by TIMP-1 involves MMP inhibition, as it was claimed to be dependent of MMP inhibition in a previous study (11). However, our data using BB94 and TIMP-2, which lack any anti-apoptotic activity, strongly suggest that TIMP-1 regulation of cell survival may be mediated by yet undefined signaling pathways rather than through their anti-proteolytic activity in breast epithelial cells.