Tissue Inhibitor of Metalloproteinase-3 Induces a Fas-associated Death Domain-dependent Type II Apoptotic Pathway*

Tissue inhibitors of metalloproteinases (TIMPs) are important regulators of matrix metalloproteinase (MMP) and adamalysin metalloproteinase activity. We previously reported that overexpression of TIMP-3 inhibits MMPs and induces apoptotic cell death in a variety of cell types and demonstrated that apoptosis is mediated through the N terminus of TIMP-3, which harbors the MMP inhibitory domain. However, little is known about the mechanisms underlying TIMP-3-induced apoptosis. Here we demonstrate that overexpression of TIMP-3 induced activation of initiator caspase-8 and -9 and promoted caspase-mediated cleavage of the death substrates poly(ADP-ribose) polymerase and focal adhesion kinase. Furthermore, TIMP-3 induced mitochondrial activation as demonstrated by loss of mitochondrial membrane potential and release of cytochrome c . Intervention studies demonstrated that overexpression of Bcl-2, the anti-apoptotic mitochondrial membrane protein, or CrmA, a viral serpin inhibitor of caspase-8, completely inhibited TIMP-3-induced apoptosis. Fur-thermore, function as amplifiers of the caspase cascade. To test whether TIMP-3 induces a similar type II apoptotic mechanism in HeLa cells, we used an adenovirus carrying CrmA. Infection of HeLa cells with rAd/CrmA (but not the control virus) inhibited TIMP-3-induced activation of caspase-9 and cleavage of FAK and PARP death substrates. These data are consistent with the hypothesis that TIMP-3 induces a type II apoptotic pathway initiated by caspase-8.

Tissue inhibitors of metalloproteinases (TIMPs) 1 are a family of four secreted proteins that collectively regulate the activity of the matrix metalloproteinases (MMPs) and at least some of the adamalysin proteases (1,2). TIMPs are thought to fold into a similar two-domain structure, with each domain folded into three loops constrained by three disulfide bonds (3)(4)(5)(6). The N-terminal domain contains a highly conserved CXC motif deemed responsible for MMP inhibition, whereas the smaller C-terminal domain confers specific functions such as the ability of TIMP-1 to interact with pro-MMP-9 and of TIMP-2 to interact with pro-MMP-2 (7)(8)(9)(10)(11)(12)(13).
TIMPs regulate MMP activity. Thus, TIMPs play important roles in regulating cellular functions that are dependent on matrix composition such as invasion, migration, differentiation, and proliferation. Several additional functions have also been attributed to individual TIMPs such as regulation of angiogenesis, activation of pro-MMP-2, and mitogenesis (14 -18).
Using adenovirus-mediated gene transfer, we recently demonstrated that overexpression of TIMP-3 in human vascular smooth muscle cells and cancer cell cultures reduces cell migration and promotes apoptotic cell death, the latter being evoked through an unknown mechanism (19,20). Furthermore, this unique phenotype induced by TIMP-3 translates to a beneficial reduction in vascular neointima formation in a human model of late vein graft failure (21). The pro-death domain of TIMP-3 has recently been localized to the N terminus, the region associated with MMP inhibitory activity (22); and it has been proposed, at least in colon cancer lines, that TIMP-3 promotes death through stabilization of tumor necrosis factor-␣ (TNF-␣) receptors on the cell surface, leading to increased susceptibility to apoptosis (23). Interestingly, deficiency of TIMP-3 in homozygous knockout mice results in enhanced apoptosis during mammary gland involution (24), suggesting that the physiological levels of TIMP-3 play an important role in regulating apoptosis during a number of physiological and pathological processes. It is therefore important to define the mechanism(s) through which TIMP-3 regulates apoptotic mediators when overexpressed.
Apoptotic cell death can be induced by a variety of stimuli, including death ligands such as TNF-␣, Fas ligand, and TRAIL; DNA-damaging agents; and loss of matrix attachment in adherence-dependent cells. Many of these stimuli activate the apoptotic cascade through the action of two distinct initiator caspases, caspase-8 and -9. Caspase-9 is an important regulator of downstream effector caspases (caspase-3, -6, and -7) in response to signals generated by the mitochondria (25,26). Its activation requires the formation of a cytosolic complex with apoptosis protease activating factor, ATP, and cytochrome c released from the mitochondria (26). Caspase-8 is typically recruited to activated death receptors via interactions with adaptor proteins such as FADD, where they form part of the death-inducing signaling complex (27). Two caspase-8-initiated apoptotic signaling pathways have been described (28). A strong activation of caspase-8 in response to death receptor activation can directly activate downstream effector caspases (type I pathway), whereas a weaker activation of caspase-8 can initiate mitochondrial activation, resulting in mitochondrial * This work was supported in part by the British Heart Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed. membrane depolarization, release of cytochrome c into the cytosol, and subsequent activation of caspase-9 and effector caspases (type II pathway) (28). During a type II pathway, the mitochondria act as amplifiers, resulting in additional caspase-8 activation (28).
Here we investigated the cellular mechanism activated by overexpression of TIMP-3 in primary cultures of both vascular smooth muscle cells (VSMCs) and a HeLa carcinoma cell line. We show that TIMP-3 induces a FADD-dependent type II apoptotic signaling pathway.

Materials
Human embryonic kidney 293 cells were purchased from Microbix (Toronto, Canada), and HeLa cells were from the European Collection of Animal Cell Cultures (Salisbury, United Kingdom). All chemicals were purchased from Sigma (Poole, UK) and were of the highest quality available. Culture media and additives were obtained from Invitrogen (Paisley, Scotland). Rabbit anti-human TIMP-3 polyclonal antibody was purchase from Chemicon International, Inc. (Harrow, UK). Recombinant human TNF-␣ and rabbit anti-cleaved poly(ADP-ribose) polymerase (PARP) (p85) antibody were purchased from Roche Molecular Biochemicals (Lewes, UK). Anti-CrmA, anti-Bcl-2, anti-focal adhesion kinase (FAK), and anti-cytochrome c antibodies and caspase substrates and inhibitors were obtained from CN Biosciences (Nottingham, UK). Anti-TNF-␣ antibody was purchased from R & D Systems. Mito-Tracker ® Orange CMTMRos was obtained from Molecular Probes, Inc. (Leiden, The Netherlands).

Methods
Cell Culture-Human embryonic kidney 293 cells were maintained in minimal essential medium supplemented with 100 units/ml penicillin, 100 g/ml streptomycin, and 10% (v/v) fetal calf serum. Rat smooth muscle cells (SMCs) were prepared from thoracic aortas as described previously (29). Rat SMCs and HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. All cells were maintained at 37°C in an atmosphere of 95% air and 5% carbon dioxide.
Adenoviral Constructs-The adenoviral construct rAd/TIMP-3 has been described previously (30). The control adenoviral construct rAd/␤galactosidase was a gift from Dr. G. W. G. Wilkinson (University of Wales College of Medicine, Heath Park, Cardiff, UK). rAd/DN-FADD was a gift from Dr. C. Trautwein (Department of Gastroenterology and Hepatology, Mediziniche Hochschule Hannover, Hannover, Germany). The rAd/CrmA adenoviral construct was a gift from Dr. G. Nabel, and the rAd/Bcl-2 adenoviral construct was a gift from Dr. J. Uney.
Adenoviral Infection-Primary rat aortic SMCs (passages 2-4) and HeLa cells were cultured in six-well plates or on sterile glass coverslips until 80% confluent. An accurate cell number was determined preinfection by trypsinization of three wells and counting using a Neubauer hemocytometer. The remaining wells were infected at 300 (rat SMCs) and 100 (HeLa cells) plaque-forming units (pfu)/cell in 1 ml of fresh complete medium for 2 h. The medium was then replaced with 2 ml of fresh complete medium and left for the required length of time prior to analysis. For adenoviral co-infections, cells were pre-infected with either the control adenovirus or the test adenovirus at 50 pfu/cell for 2 h. Cells were then incubated in growth medium for an additional 2 h and subsequently infected with the second adenovirus at 100 pfu/ cell for an additional 2 h, after which the medium was replaced with fresh complete medium.
Western Blotting-Total cell lysates were prepared in SDS lysis buffer (50 mM Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, and 1ϫ Complete protease inhibitor mixture (Roche Molecular Biochemicals)). Protein concentration was determined using a Bio-Rad micro-BCA protein assay according to the manufacturer's instructions. Proteins (150 g) were separated by SDS-PAGE under reducing conditions and transferred to Bio-Rad polyvinylidene difluoride nylon membranes. Membranes were blocked in Tris-buffered saline/Tween (200 mM Tris-HCl (pH 7.4), 137 mM NaCl, and 0.2% Tween 20) containing 5% skim milk powder, followed by incubation with primary antibody. Following washing with Tris-buffered saline/Tween, blots were incubated with horseradish peroxidase-conjugated secondary antibody for 60 min, and immunoreactive proteins were visualized using the enhanced chemiluminescence system (ECL, Amersham Biosciences).
In Situ End Labeling of DNA-Cells were grown on glass coverslips and fixed in ice-cold methanol 72 h post-infection. After air drying, cells were washed twice with 1ϫ TE buffer (10 mM Tris-HCl (pH 8.0) and 1 mM EDTA) and incubated in labeling mixture (50 mM Tris-HCl (pH 7.2), 10 mM MgSO 4 , 0.1 mM dithiothreitol, 0.01 mM dATP, 0.01 mM dCTP, 0.01 mM dGTP, 0.01 mM biotin-dUTP, and 8 units/ml DNA polymerase I (Klenow)) for 15 min at room temperature. Cells were rinsed in 1ϫ TE buffer, and endogenous peroxidase activity was reduced by incubation in 2% H 2 O 2 for 5 min. After further washing, biotin was labeled with ExtrAvidin TM -peroxidase conjugate (1:200). Incubation with diaminobenzidine and subsequent counterstaining with hematoxylin were carried out to distinguish between positive apoptotic nuclei containing nicked DNA and non-apoptotic nuclei.
Cell Death Enzyme-linked Immunosorbent Assay-Apoptotic cell death was quantified using a photometric enzyme-linked immunosorbent assay for the detection of cytoplasmic histo-associated DNA fragments (Roche Molecular Biochemicals). Briefly, cell lysates were incubated with a biotin-conjugated anti-histone antibody and a peroxidaseconjugated anti-DNA antibody for 2 h. Complexes were captured on streptavidin-coated microtiter plates and quantified using ABTS colorimetric substrate.
Measurement of Mitochondrial Membrane Potential-Mitochondrial membrane potential (⌿ m ) was measured using the MitoTracker ® Orange CMTMRos fluorescent dye (31). Briefly, cells were incubated with MitoTracker ® Orange CMTMRos (150 nM) for 30 min in culture medium at 37°C and 5% CO 2 . As a positive control for ⌿ m loss, cells were incubated with 2 M staurosporine for 2 h. Cells were washed once with phosphate-buffered saline, collected by centrifugation at 200 ϫ g, and fixed in 2 ml of 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) for 10 min at 4°C. Fixed cells were analyzed by flow cytometry (BD PharMingen FACScan) using CellQuest acquisition software.
Preparation of Cytosolic Extracts-Cells were collected by trypsinization and centrifuged at 200 ϫ g for 5 min at 4°C. Cells were then washed twice with ice-cold phosphate-buffered saline (pH 7.4), followed by an additional centrifugation at 200 ϫ g for 5 min. The cell pellet was resuspended in 600 l of extraction buffer containing 220 mM mannitol, 68 mM sucrose, 50 mM PIPES-KOH (pH 7.4), 50 mM KCl, 5 mM EGTA, 2 mM MgCl 2 , 1 mM dithiothreitol, and Complete protease inhibitor mixture. After 30 min of incubation on ice, cells were homogenized with a glass Dounce homogenizer and a small clearance pestle until the majority of the cells had been disrupted. Cell homogenates were spun at 14,000 ϫ g for 15 min, and supernatants were removed and analyzed by Western blotting for cytochrome c.

TIMP-3 Overexpression Induces Caspase
Activation-In agreement with previously published data (19,21,22), infection of rat VSMCs and HeLa cells with a replication-deficient adenovirus expressing human TIMP-3 (rAd/TIMP-3; 300 and 100 pfu/cell, respectively), but not with the control adenovirus (rAd/␤-galactosidase), induced apoptotic cell death 48 -72 h post-infection (data not shown) (22,30). TIMP-3-induced apoptosis was characterized by cell shrinkage, cell rounding, membrane blebbing, and cell detachment from the substratum. In situ end labeling staining of rAd/TIMP-3-infected cells revealed intense brown nuclear staining indicative of fragmented DNA, with nuclear condensation and fragmentation (data not shown). Cells expressing TIMP-3 also exhibited phosphatidylserine externalization as indicated by strong annexin V staining without uptake of propidium iodide (data not shown).
Caspase-mediated cleavage of death substrates is an important event during the commitment to and execution of apoptotic death. Cleavage of PARP and FAK has previously been reported in numerous cell types undergoing apoptosis (32,33). We therefore analyzed PARP and FAK cleavage status in TIMP-overexpressing cells. Infection of HeLa cells with rAd/ TIMP-3 (but not with rAd/␤-galactosidase) resulted in the formation of two prominent lower molecular mass FAK fragments between 70 and 80 kDa and the appearance of an 85-kDa fragment of PARP detected with an anti-cleaved PARP antibody ( Fig. 2A). Cleaved forms of FAK and PARP were first detectable 32-36 h after rAd/TIMP-3 infection and increased through the time course of the experiment ( Fig. 2A). As expected, overexpression of TIMP-3 preceded the appearance of PARP and FAK cleavage by ϳ20 h (Fig. 2A). The time course of PARP and FAK cleavage mirrored that of initiator caspase-8 and -9 activation, suggesting that TIMP-3 induces caspase-mediated cleavage of FAK and PARP. To test this hypothesis, we incubated HeLa cells overexpressing TIMP-3 with either the pan-caspase inhibitor benzyloxycarbonyl-VAD-fluoromethyl ketone or the calpain peptide inhibitor N-acetyl-Leu-Leu-norleucinal. Incubation with benzyloxycarbonyl-VAD-fluoromethyl ketone (but not with N-acetyl-Leu-Leu-norleucinal) completely inhibited the formation of the lower molecular mass forms of FAK and the p85 PARP fragment, demonstrating caspase-mediated cleavage of these proteins in response to TIMP-3 overexpression (Fig. 2B).
TIMP-3 Induces Mitochondrial Activation-Activation of the mitochondria and release of cytochrome c into the cytosol have been shown to participate in activation of caspase-9 (26). As we observed activation of caspase-9 in response to TIMP-3, we first measured changes in mitochondrial membrane potential (⌬⌿ m ) using CMTMRos and release of cytochrome c into the cytosol in rat VSMCs and HeLa cells infected with rAd/TIMP-3. rAd/␤galactosidase-infected rat VSMCs and HeLa cells showed polarized ⌿ m as indicated by a high level of CMTMRos fluorescence (Fig. 3A). As expected, treatment with 2 M staurosporine, a potent stimulus for mitochondrial activation, resulted in a marked reduction in CMTMRos fluorescence in both rat SMCs and HeLa cells, demonstrating a reduction in ⌿ m (Fig. 3A). Infection with rAd/TIMP-3 resulted in a similar reduction in ⌿ m (Fig. 3A). Interestingly, TIMP-3 overexpression resulted in a larger reduction in CMTMRos fluorescence than that evoked by staurosporine. To confirm these observations, efflux of cytochrome c into the cytosol, another indicator of mitochondrial activation, was evaluated (28,34). Low levels of cytochrome c were detectable in cytosolic extracts from HeLa cells (but not from rat VSMCs) infected with the control virus rAd/␤-galactosidase (Fig. 3B). However, infection of both VSMCs and HeLa cells with rAd/TIMP-3 dramatically increased the levels of cytosolic cytochrome c (Fig. 3B).

FIG. 3. TIMP-3 induces mitochondrial activation. HeLa cells and rat
VSMCs were infected with either the control adenovirus or rAd/TIMP-3 at 100 or 600 pfu/cell, respectively. A, 72 h postinfection, ⌬⌿ m was measured by CMTM-Ros staining and flow cytometry. The percentage of cells with low ⌿ m is seen as a shift to weaker CMTMRos fluorescence and was detected using CellQuest software. Uninfected cells were stimulated with 2 M staurosporine for 2 h as a positive control. B, cytosolic extracts were analyzed for cytochrome c by Western blotting.

TIMP-3-induced Apoptosis
bers is thought to regulate such functions as permeability transition pore formation and release of apoptosis inducing factor and cytochrome c into the cytosol and ultimately activation of caspase-9 (26). To determine whether mitochondrial activation is an important step in the apoptotic cascade initiated by TIMP-3, we used an adenoviral vector to up-regulate Bcl-2. Infection of HeLa cells with rAd/Bcl-2 (but not with rAd/␤-galactosidase) strongly elevated levels of Bcl-2 (Fig. 4A). For coexpression studies, we pretreated cells with rAd/Bcl-2 prior to rAd/TIMP-3 infection to allow sufficient inhibitor expression to precede expression of TIMP-3. Pre-infection with rAd/␤-galactosidase prior to co-infection with rAd/TIMP-3 failed to rescue cleavage of PARP and FAK (Fig. 4B). Preinfection of HeLa cells with rAd/Bcl-2 prior to subsequent infection 3 h later with rAd/TIMP-3 blocked TIMP-3-induced cleavage of PARP and FAK death substrates (Fig. 4B). Additionally, overexpression of Bcl-2 inhibited activation of caspase-9, implying that the mitochondria function as regulators of caspase-9 activation in response to TIMP-3 (Fig. 4B). Furthermore, Bcl-2 co-overexpression also inhibited activation of caspase-8, suggesting that TIMP-3-induced caspase-8 activation occurs downstream of mitochondrial activation (Fig.  4B).
Although we have demonstrated activation of caspase-8 following adenovirus-mediated overexpression of TIMP-3, we further assessed the direct involvement of this caspase by cooverexpression of CrmA, a viral serpin inhibitor of caspase-8 (37), using adenovirus-mediated gene transfer. As expected, infection of HeLa cells with rAd/CrmA (but not with rAd/␤-galactosidase) resulted in high level expression of CrmA (Fig.  5A). Pre-infection of HeLa cells with rAd/CrmA (but not with rAd/␤-galactosidase) blocked cleavage of PARP and FAK induced by TIMP-3 and activation of caspase-8 (Fig. 5B). Furthermore, CrmA expression also inhibited activation of caspase-9 (Fig. 5B), implying that TIMP-3-induced caspase-9 activation occurs downstream of caspase-9.
Dominant-negative FADD Blocks TIMP-3-induced Apoptosis-Caspase-8 is typically activated in response to engagement of death receptors, which contain cytoplasmic death domains (38,39). Caspase-8 is recruited to activated death receptors through interactions with the adaptor protein FADD, which contains a C-terminal death domain and an N-terminal death effector domain and which is responsible for recruitment and activation of caspase-8 (27). To test whether TIMP-3 induces death receptor-initiated apoptosis through caspase-8, we used an adenoviral vector expressing a death effector domaindeleted dominant-negative mutant of FADD (rAd/DN-FADD) (40). Infection of cells with rAd/DN-FADD resulted in high levels of DN-FADD expression (Fig. 6A) without affecting cell morphology and viability (Fig. 6D) or cleavage of PARP and FAK (Fig. 6, C and D). As expected, infection of HeLa cells with rAd/DN-FADD (but not with the control virus) completely blocked the cell detachment and membrane blebbing (Fig. 6D) and cleavage of PARP and FAK (Fig. 6B)  induced apoptosis (including cell shrinkage and detachment and membrane blebbing), but also completely blocked cleavage of PARP and FAK death substrates (Fig. 6, B and D). In addition, DN-FADD blocked both the basal and TIMP-3-induced apoptotic cell deaths as measured by a cell death nucleosome enzyme-linked immunosorbent assay (Fig. 6E), indicating that death receptor engagement is the initiating signal during TIMP-3-induced apoptosis.
Analysis of Candidate Receptors-To define the death receptor(s) responsible for the transmission of the apoptotic stimuli, we investigated the expression of known death receptors in response to TIMP-3 overexpression. It has been hypothesized that stabilization of TNF-␣ receptors on the cell surface of colon cancer cells is involved in the apoptotic cascade initiated by TIMP-3 (23). Additionally, TIMP-3 is a potent inhibitor of TNF-␣-converting enzyme, a cell-surface metalloprotease involved in the processing of cell-surface TNF-␣ (2). Consistent with this, infection of VSMCs or HeLa cells with rAd/TIMP-3 (but not with rAd/␤-galactosidase) resulted in increased cell-surface levels of TNF-␣ (Fig. 7A), presumably through inhibition of TNF-␣-converting enzyme-mediated TNF-␣ shedding from the cell surface. To test whether this increase in cell-surface TNF-␣ is responsible for TIMP-3-induced apoptosis, we used an antibody that neutralizes the biological actions of TNF-␣. As expected, cleavage of PARP and FAK death substrates in re-sponse to recombinant TNF-␣ was almost completely blocked by co-incubation with 5 g/ml TNF-␣-neutralizing antibody (Fig. 7B), demonstrating the efficacy of the antibody. However, incubation with the TNF-␣-neutralizing antibody had no effect on TIMP-3-induced death substrate cleavage, indicating that TIMP-3-induced apoptosis occurs independently of TNF-␣ (Fig.  7B). Furthermore, co-incubation with recombinant soluble Fas/Fc or soluble TRAIL receptor also had no effect on TIMP-3-induced apoptosis (data not shown).

DISCUSSION
Imbalances between the levels of TIMPs and MMPs have been implicated in the pathogenesis of many diseases such as rheumatoid arthritis, tumor cell invasion and metastasis, atherosclerosis, and fibrosis. Although imbalances in net proteolytic activity are a major factor in the progression of these diseases, the additional functions of TIMPs, including the regulation of apoptosis, are also likely to be important. The processes regulating cell death are complex and involve intricate interplay between the extracellular microenvironment, the cell membrane, and intracellular organelles. MMPs, ADAMs, and their endogenous inhibitors (TIMPs) are attracting increasing attention as mediators of apoptotic stimuli through mechanisms including control of matrix remodeling and anoikis and Although there is no direct evidence in the literature for regulation of apoptosis by physiological levels of TIMP-3, there are ample data to suggest that dysregulation of TIMP-3 levels, either through increased endogenous expression or ectopic overexpression or through knockout, leads to direct effects on apoptotic cell death. For example, endogenous TIMP-3 levels are high in shoulder regions of advanced atherosclerotic plaques, the regions at which increased apoptosis of VSMCs leads to plaque destabilization and rupture (42)(43)(44). More direct evidence is observed in the rare inherited disorder Sorsby's fundus dystrophy, a degenerative disease of the retina characterized by thickening of the Bruch's membrane and apoptotic death of the retinal epithelial pigment cells (45)(46)(47). In Sorsby's fundus dystrophy, mutations in the coding region of TIMP-3, which, interestingly, all lead to an alteration in the number of cysteine residues, have been identified (45,46), although a direct link between altered TIMP-3 activity and/or distribution has not been documented. Induction of apoptotic cell death by TIMP-3 overexpression is well documented and has been shown in vascular cells and cancer cells in vitro and in vivo (19,23,30,48). Beneficial effects of TIMP-3 in these scenarios include modulation of both cell migration and invasion through MMP blockade, reduced angiogenesis, and elevated rates of apoptosis, the latter through unknown mechanisms. Smith et al. (23) have reported that TIMP-3 induces apoptosis of DLD carcinoma cells via stabilization of cell-surface TNF-␣ receptors. Interestingly, recent evidence has demonstrated, at least in the mouse, that the absence of TIMP-3 leads to apoptotic cell death in involuting breast tissue (24). There is therefore an increasing need to understand the precise mechanism(s) that TIMP-3 regulates during physiological processes as well as the signaling pathways activated through TIMP-3 overexpression. The latter is especially important for the potential exploitation of TIMP-3 as an effective gene-based pro-death therapy for vascular disease (21) and cancer (48,49). Modulation of apoptosis by other metalloproteinase inhibitors has been described (50 -52). Mitsiades et al. (53) have described the induction of Fas-mediated apoptosis in Ewing's sarcoma cell lines by synthetic metalloproteinase inhibitors. Additionally, TIMP-4 has recently been shown to promote apoptosis of transformed cardiac fibroblasts (54).
A large body of evidence demonstrates the major role played by the caspases in the execution of apoptotic cell death in response to numerous signals (55). The caspase cascade can be activated via the activity of two families of initiator caspases. Members of the caspase-8 family are typically recruited to activated death receptors via interactions with adaptor proteins such as FADD, where they form part of the death-inducing signaling complex (27). The caspase-9 family (caspase-9 and -2) is thought to be important in activating the downstream effector caspases in response to signals generated by the mitochondria (25,26). Caspase-9 activation requires the formation of a cytosolic complex with APAF, ATP, and cytochrome c released from the mitochondria (26). Here we have demonstrated that TIMP-3 increased caspase activity and caspase-mediated cleavage of the death substrates FAK and PARP. Depolarization of the mitochondrial membrane and release of cytochrome c also suggest that the mitochondria play an important role in regulating apoptosis induced by TIMP-3. Interestingly, TIMP-3 induced activation of both initiator caspase-8 and -9, suggesting that these caspases may play a role in the initiation of apoptosis by TIMP-3. To identify the initiating signal and the apical caspase activated by TIMP-3, we employed recombinant adenoviral vectors expressing Bcl-2, a negative regulator of apoptosis-specific mitochondrial functions (56,57); CrmA, a cowpox serpin inhibitor of caspase-8; and DN-FADD, a death receptor adaptor protein involved in activation of caspase-8 (27). Overexpression of Bcl-2 completely inhibited activation of caspase-9, demonstrating that mitochondrial activation is responsible for activation of caspase-9 during TIMP-3-induced apoptosis. TIMP-3 overexpression also resulted in a drop in mitochondrial membrane potential and a release of cytochrome c into the cytosol. The importance of this mitochondrial activation in the progression of TIMP-3-induced apoptosis is demonstrated by the complete inhibition of death substrate cleavage by overexpression of Bcl-2. Expression of Bcl-2 also inhibited activation of caspase-8, suggesting that the majority of caspase-8 activation in response to TIMP-3 occurs downstream of mitochondrial activation. Inhibition of caspase-8 activation by Bcl-2 has been described previously in Jurkat cells that undergo a type II apoptotic pathway in response to the Fas ligand (28). In cells that undergo a type II apoptotic pathway, death receptor engagement initially triggers only a small amount of caspase-8 activation, leading to activation of the mitochondria, which in turn results in cleavage of caspase-9, downstream effector caspases such as caspase-3, and ultimately more caspase-8. In this type of pathway, caspase-8 is the apical caspase, and the mitochondria function as amplifiers of the caspase cascade. To test whether TIMP-3 induces a similar type II apoptotic mechanism in HeLa cells, we used an adenovirus carrying CrmA. Infection of HeLa cells with rAd/CrmA (but not the control virus) inhibited TIMP-3-induced activation of caspase-9 and cleavage of FAK and PARP death substrates. These data are consistent with the hypothesis that TIMP-3 induces a type II apoptotic pathway initiated by caspase-8.
Taken together, the data suggest that TIMP-3 induces apoptosis via a death receptor-mediated mechanism. To confirm this, we used a dominant-negative mutant of FADD, a protein involved in recruitment of caspase-8 to the death-inducing signaling complex, where it is subsequently activated. Apoptosis induced by death ligands such as TNF-␣ and Fas ligand have previously been shown to be blocked by overexpression of DN-FADD mutants that are unable to recruit caspase-8 (58 -60). We have shown that infection of HeLa cells with rAd/DN-FADD completely blocked TNF-␣-induced cleavage of FAK and PARP. Furthermore, expression of DN-FADD also completely inhibited TIMP-3-induced cleavage of FAK and PARP and the morphological changes associated with apoptosis such as membrane blebbing and cell detachment, demonstrating that TIMP-3 induces FADD-dependent apoptosis. Taken together, these data strongly suggest that TIMP-3 initiates a type II apoptotic pathway via a FADD-sensitive death receptor.
Clearly, the identity of the death receptor that mediates TIMP-3-induced apoptosis is of great interest. Previous studies have suggested that TNF receptor-1 plays a role in TIMP-3induced apoptosis of DLD carcinoma cells (23). Here we have reported that TIMP-3 overexpression resulted in increased levels of cell-surface TNF-␣ in both VSMCs and HeLa cells. However, neutralization of this TNF-␣ had no effect on TIMP-3induced apoptosis. This probably reflects differences in death receptor and death ligand expression. Furthermore, incubation with recombinant soluble Fas/Fc and soluble TRAIL receptors, which inhibit Fas ligand-and TRAIL-mediated apoptosis by acting as decoy receptors (41), also failed to block TIMP-3induced apoptosis. These data demonstrate that TIMP-3-induced apoptosis in HeLa cells and VSMCs occurs independently of TNF-␣, Fas ligand, or TRAIL. However, this does not rule out the possibility of ligand-independent actions of TNF receptor-1 or Fas or TRAIL receptors.
In summary, this study has highlighted the involvement of a FADD-dependent type II apoptotic pathway in the induction of apoptosis by TIMP-3. These findings are likely to have important implications for our understanding of the normal physiological roles of TIMP-3, the involvement of TIMP-3 in diseases such as Sorsby's fundus dystrophy, and the future development of TIMP-3-based gene therapies.