Cartilage Oligomeric Matrix Protein Protects Cells against Death by Elevating Members of the IAP Family of Survival Proteins*

Cartilage oligomeric matrix protein (COMP) is a component of cartilage, synovium, ligament, and tendon, yet its normal function is largely unknown. To identify its function we have expressed it in 293 and HeLa cell lines and in primary human chondrocytes. We find that COMP protects these cells against death, either in the presence or absence of tumor necrosis factor α and is able to block activation of caspase 3, a critical effector caspase. This effect appears to be mediated by the IAP (inhibitor of apoptosis protein) family of anti-apoptotic proteins because the levels of XIAP, survivin, cIAP1 and cIAP2 are significantly elevated in the COMP-expressing cells and down-regulation of survivin and XIAP protein levels by small interfering RNAs blocks the ability of COMP to enhance survival. The mRNAs for most of the IAP family members were not increased by COMP, indicating that a translational/post-translational mechanism was involved in their induction. However, in both HeLa cells and chondrocytes, COMP induced survivin mRNA by 5-fold. Thus survivin is the first gene identified to be up-regulated transcriptionally by COMP. The carboxyl-terminal half of the protein comprising the type 3 repeats and the RGD sequence (CaCTD domain) was sufficient to promote survival and to elevate the IAPs. Further, an RGD peptide was able to block the prosurvival effect of COMP and the induction of XIAP and survivin, indicating that survival is likely mediated through integrin signaling. These data point to a new role for COMP in protecting cells against death.

pentamer of identical subunits where each monomer is linked to its neighbor via a disulfide bond located at the very amino terminus of the protein (7)(8)(9). Interestingly, the amino-terminal domain of each monomer is also ␣-helical, and each helix also interacts with its neighbor via a coiled-coil structure (9,10). This interaction results in the formation of a hydrophobic pore within this oligomerization region, which is large enough to bind small aliphatic compounds such as retinol or vitamin D (11), although the role of this binding is not yet clear.
The central domain of COMP contains both epidermal growth factor-like repeats (type 2 repeats) and a calcium-binding region (type 3 repeats), which are present in all the thrombospondins (1,2). Missense mutations in the calcium-binding region of COMP gene are responsible for at least two forms of skeletal dysplasias in humans, multiple epiphysial dysplasia (MED) and pseudoachondroplasia (PSACH) (12)(13)(14)(15). As a corollary to these genetic studies, it has been demonstrated that COMP is subjected to cleavage via extracellular proteases in the cartilage of patients suffering from various forms of arthritis (16 -19). Some of the resulting fragments of COMP that are produced from this cleavage appear to be stable and can be easily identified in both the synovival fluid and serum of these patients (16 -19), yet it is not known what role these fragments may play, if any, in the pathology of arthritis.
In cartilage, COMP is expressed in both the developing and mature tissue (7,20,21). In adult cartilage, COMP has been shown to be located primarily in the interterritorial matrix between chondrocytes (22). As such it has been proposed that COMP may directly interact with cells. COMP has been shown to aid in the attachment of cells to surfaces (23,24) and does so via an interaction with integrins (24). A COMP mutation in the type 3 repeat, which occurs in 30% of patients with PSACH, abolishes the ability of COMP to aid in attachment (24). It has also been shown that COMP supports the migration of cells in vitro (23). In terms of the extracellular matrix protein, it is presumed that COMP may play a role in the structural integrity of cartilage via its interaction with other extracellular matrix proteins such as the collagens and fibronectin. However, it is not yet known whether COMP plays any role in maintaining the stability of cartilage. Additional studies have indicated that COMP can inhibit cell proliferation while enhancing in vitro chondrogenesis (21). Finally, expression of COMP in the presence of BMP2 stimulation can lead to a late increase in apoptosis (21). This latter finding is consistent with the fact that the mutant COMP proteins in multiple epiphysial dysplasia and PSACH appear to mediate apoptosis of chondrocytes in cartilage in vivo (25)(26)(27).
To better understand the function of COMP, we began transient expression studies in both transformed and nontransformed human cells with the goal of assessing its effects on cell viability. We find that COMP is a potent suppressor of apoptosis in both primary human chondrocytes and transformed cells and that it accomplishes this function by inducing the IAP family of survival proteins.

EXPERIMENTAL PROCEDURES
Cell Culture, Transfections, Cell Counts, and Flow Cytometry-HeLa and 293 cells were obtained from the ATCC, whereas human chondrocytes were isolated from patients undergoing total knee arthroplasty. The knee samples were provided by National Disease Research Interchange (NDRI) (Philadelphia, PA). The average age was 62 years old (range 52-71 years). To isolate chondrocytes, areas of undamaged cartilage were shaved off the femoral condyles or the tibial plateau. The shavings were washed three times in PBS and then digested with 0.1% trypsin-EDTA in serumfree culture medium for 1 h at 37°C under agitation. After washing in PBS, the cartilage pieces were minced and then digested with collagenase 2 (Sigma) at 50 g/ml in serum-free medium overnight at 37°C under agitation. The chondrocyte suspension was filtered through a 40-m cell strainer, washed in PBS, and plated at 1.5 ϫ 10 5 cells/cm 2 in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (Atlanta Biologicals, Lawrenceville, GA) plus antibiotic-antimycotic solution. The cells were allowed to adhere, spread, and proliferate for 7-10 days. Chondrocytes were used for experiments between passage 1 and 4, during which time cartilage marker genes were expressed. The actinomycin D was purchased from Sigma-Aldrich, and the TNF␣ was from R & D Systems, Inc. (Merrisville, NC).
Monolayer cultures were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin. The cultures were incubated in a humidified atmosphere at 37°C and 5% CO 2 . The medium was changed every 3 days. All of the transfection experiments were initiated on 50% confluent monolayer cultures. Plasmids (20 g) were transfected either by the calcium phosphate procedure for 293 cells or by Amaxa nucleofection for human chondrocytes and HeLa cells (Amaxa, Gaithersburg, MD) according to the manufacturer's protocol. The human COMP cDNA expression plasmid was obtained from Origene Technologies (Rockville, MD). The empty vector was pSec2B (Invitrogen).
The cells were rinsed with PBS, trypsinized, resuspended in 10 ml of growth medium, and counted in the presence of trypan blue using a hemocytometer. For cell cycle analysis (flow cytometry), the cells were rinsed once in chilled PBS then trypsinized and resuspended in 10 ml of Dulbecco's modified Eagle's medium plus 10% fetal bovine serum. The cells were pelleted and resuspended in 70% ethanol. The cells were kept on ice for 10 min, pelleted, and then treated with RNase A (180 g; Sigma) for 30 min at room temperature. Propidium iodide (Sigma) was added to a final concentration of 75 mg/ml. Cell cycle analysis was performed on a Coulter Profile 2 flow cytometer. For the TUNEL assay, monolayer cells were fixed in paraformaldehyde and then processed as per the manufacturer's recommendation (DeadEnd Tunel System, Promega, Madison, WI).
The RGD (cycloRGDFV) and control (cycloRADFV) peptides were purchased from Bachem and used at a concentration of 1 mM. The RGD is a potent inhibitor of integrin signaling, whereas the control peptide is not (40). siRNA for human COMP was purchased from Ambion, whereas the siRNAs for survivin and XIAP were obtained from Cell Signaling Technology (Danvers, MA). The COMP, XIAP, and survivin siRNAs (100 nM) were transfected into 293 cells, HeLa cells, and chondrocytes as per each manufacturer's recommendation.
RNA Isolation and RT/PCR Analysis-Total RNA was isolated from cells by the TRIzol method (Invitrogen). Oligo(dT) was used as the primer in the reverse transcription reaction which was followed by a PCR with the appropriate primers. For the RT/PCRs, one mg of total RNA from the cells was used in each reaction. All of the RNA samples were DNase I-treated prior to the PCRs. Additionally, as a control, PCR done in the absence of reverse transcriptase was negative for any ethidium bromide-stained bands (data not shown).
Protein Analysis and Immunoblotting-The cells were lysed on ice in 0.1% Nonidet P-40, 10 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 15 mM NaCl, 0.2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and a protease inhibitor mixture (Roche Applied Science). The nuclei were pelleted by centrifugation at 800 ϫ g for 5 min, and the supernatant fraction was termed "cytosol." The nuclei were resuspended in extraction buffer consisting of 0.5 M NaCl, 20 mM Hepes, pH 7.9, 20% glycerol, 0.2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (Roche Applied Science) and incubated on ice for 10 min. The nuclei were then spun at 14,000 ϫ g for 5 min to pellet the residual nuclear material, and the supernatant fraction was termed "nuclear extract." For analysis of secreted COMP proteins, the cells were cultured in serum-free medium for 24 h. The medium was harvested, and the cellular debris was removed by centrifugation. Aliquots of the medium were added to two volumes of ice-cold acetone, and the proteins were precipitated at Ϫ20°C, as by Jordan-Sciutto et al. (28). The protein precipitate was pelleted by centrifugation at 14,000 ϫ g for 10 min at 4°C. The protein precipitate was then boiled in SDS-PAGE sample buffer in either the presence or absence of the reducing agent ␤-mercaptoethanol. Additionally, an enzyme-linked immunosorbent assay plate reader assay kit (Serotec, Seattle, WA) for COMP was used to determine COMP protein levels secreted into the medium. For generation of conditioned medium, 293 cells were transfected with the COMP plasmid, and 24 h post-transfection the cells were washed, and serum free medium was added for an additional 48 h. COMP concentrations were determined by enzyme-linked immunosorbent assay.
For immunoblotting, the proteins were electrophoretically resolved by SDS-PAGE (40 g of protein/lane) and transferred onto nitrocellulose membranes. The blots were then washed in TBST buffer (10 mM Tris, pH 8, 150 mM NaCl, 0.05% Tween 20), blocked with 5% bovine serum albumin in TBST for 1 h at room temperature, and incubated with the primary antibodies overnight at 4°C in TBST. Anti-human COMP antibody (Sero-tec, Raleigh, NC) was used at a concentration of 1:500. The blot was then washed three times, 10 min each, in TBST and then incubated with a 1:2,500 dilution of secondary antibody conjugated to either alkaline phosphatase (Protoblot System) or to horseradish peroxidase (Chemiluminescence) for 1 h at room temperature in TBST. The blots were then processed using either the Protoblot system (Promega) using 5-bromo-4-chloro-3-indolyl-phosphate/nitro blue tetrazoleum (BCIP/NBT) as a color developer or using the Supersignal West Pico Chemiluminescent Substrate System (Pierce) and exposed to x-ray film.

COMP Enhances Cell Survival in 293
Cells-It has been previously demonstrated that expression of COMP, as either a wild type or mutant form, can affect the level of apoptosis both in vivo and in vitro (21,(25)(26)(27). To explore an effect of COMP on apoptosis, we began transient transfection experiments using a COMP expression plasmid in 293 cells. These cells represent an excellent cell line to test the biological effects of COMP because they are effectively null for COMP expression (they express no detectable COMP by either RT/PCR analysis of RNA or by immunoblotting of cell extracts or conditioned medium). Also, very high transfection efficiencies can be obtained in 293 cells by the calcium phosphate method (50 -80%, based on 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) staining), which makes them very useful in transient expression studies. 293 cells were transfected with a human COMP expression vector and then monitored for expression of the COMP protein. Immunoblots of cytosolic and secreted proteins from the cells showed detectable levels of the protein (Fig. 1A). When these same analyses were performed in the absence of ␤-mercaptoethanol in the Laemmli sample buffer, the COMP protein migrated as a very large molecular complex (Ͼ500 kDa), indicating that it had likely formed a pentamer in solution (Fig. 1B). When the concentration of COMP was assessed in 293 cell medium following the transfection, the levels were on average about 3.5 g/ml. Given that its pentameric molecular mass is 524 kDa, it would suggest an 8 nM concentration of COMP in the media.
To determine whether the expression of COMP affected the cells in any way, total viable adherent cell numbers were assessed at 24 h post-transfection. From Fig.  1C, it is clear that the calcium phosphate-mediated transfection led to a decrease in viable cell numbers, a phenomena that is well known (29 -32). This toxicity appears independent of the plasmid used to transfect the cells in that transfection of a number of vectors leads to a significant decrease in cell number (Fig. 1C). However, as seen in Fig. 1C, COMP expression in these cells resulted in an increase in viable cell number (4 -7-fold) at 24 h post-transfection compared with the vector controls. We next assessed whether the increase in cell numbers following COMP expression was due to a reduction in apoptosis. As seen in Fig. 1D, there is a 3-fold decrease in apoptosis in the COMP expressing cultures at 24 h post-transfection, as assessed by either flow cytometry analysis, evident as cells with less than a 2 N DNA content, or by TUNEL assay. These data indicate that the increase in cell number following COMP expression is due at least in part to a decrease in apoptosis.
As a control for these studies, a COMP siRNA obtained from Ambion was cotransfected with the COMP expression plasmid. As seen in the Western blot ( Fig. 2A), the COMP siRNA reduces the level of COMP protein in these cells compared with the cells transfected with the control siRNA. Correspondingly, the COMP siRNA blocks the ability of the COMP expression plasmid to enhance cell survival as measured either by total viable cells (Fig. 2B) or by a percentage of apoptotic cells (Fig.  2C). These data indicate that simple transfection of the COMP expression plasmid into cells is not sufficient to aid in survival 293 cells were transfected with the vector control (pSec2B) or COMP expression plasmid (25 g DNA/plate), and 24 h later the medium was collected, and the cells were harvested. The media and cytosolic extracts were assessed for COMP protein expression by immunoblotting with a COMP-specific antibody. The data show detectable expression of COMP in both the cell extracts and in the media. B, immunoblot. The proteins from the media in A were subjected to electrophoresis in the presence or absence of the reducing agent ␤-mercaptoethanol (betaME) and then immunoblotted for COMP as in A. The positions of the monomer and pentamer of COMP are shown. The data indicate that COMP forms a pentamer in solution. C, viable cell numbers. 293 cells transfected as in A with the indicated control plasmids (pSec2B, pcDNA3, and pcDNA3-LacZ) or the human COMP-expressing plasmid were seeded at 500,000 cells/10-cm plate in monolayer. At 24 h the cells were trypsinized and processed for viable cell counts by trypan blue exclusion. Shown are the total viable cell numbers/plate. The data indicate that expression of COMP elevates the number of viable cells. D, apoptosis. 293 cells were transfected with a vector (pSec2B) control and COMP expression plasmids as in A. At 24 h post-transfection, the percentage of apoptosis was assayed by both flow cytometry and TUNEL assay. The data show that COMP expression leads to a significant reduction (3-4-fold) in the level of apoptosis. For C and D the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown.
but that production of the COMP protein is required. Thus ectopic expression of COMP in these cells appears to aid in cell survival. To determine whether the survival effect is specific for COMP, an additional secreted protein was expressed in these 293 cells (prostate-specific antigen). We found that prostatespecific antigen was not able to protect the cells against apoptosis when compared with cell expressing COMP (data not shown).
COMP Induces the IAP Family of Prosurvival Proteins in 293 Cells-To more thoroughly explore the nature of the apoptosis inhibited by COMP, we examined the levels of a number of active caspases, the mediators of the apoptotic response. We assessed active effector caspases 3, 6, and 7, the TNF␣-activated caspase 8, and the stress-responsive/endoplasmic reticulum caspases 10 and 12. As seen in the immunoblot (Fig. 3A), active caspase 3 is present in the control-transfected cells, yet its level is reduced in the COMP-expressing cells. Further, levels of procaspase 3 (uncleaved) are slightly increased in the COMP-expressing cells. An examination of the other caspases (6,7,8,9,10, and 12) did not reveal any COMP-mediated changes following transfection (data not shown). Therefore, these data show that COMP inhibits the activation of caspase 3.
To identify the mechanism by which COMP blocks the activation of caspase 3 and protects cells from apoptosis, we examined the levels of a number of apoptotic and antiapoptotic proteins. It would be expected that COMP would affect the levels of some of these pro-and antiapoptotic factors, thereby mediating an effect on cell survival. These factors comprise mitochondrial (cytochromeC/Bcl2) and nonmitochondrial (Akt) mediators of apoptosis. As seen in Fig.  3B, a number of these proteins are not affected by COMP, which includes Akt, Akt1, phosphoAkt(Ser), cytochrome C, Bcl2, phosphoBcl2(Ser), BclXL, and Mcl1. These data indicate that the block to caspase 3 activation and apoptosis is not through the regulation of these factors.
The IAP family of survival proteins was then examined because they act to directly block caspase 3 activation (33, At 24 h the medium was collected and was assessed for COMP protein by immunoblotting with a COMP-specific antibody. The data show that COMP protein levels were decreased by about 5-fold the COMP siRNA as assessed by densitometry of the protein bands. B, cell survival/siRNA. 293 cells were transfected as in A and were seeded at 500,000 cells/10-cm plate in monolayer. At 24 h the cells were trypsinized and processed for viable cell counts by trypan blue exclusion. Shown are the total viable cell numbers/plate, which indicates that although expression of COMP elevates the number of viable cells, the COMP siRNA blocks this effect. C, apoptosis/siRNA. 293 cells were transfected as in A were processed at 24 h post-transfection for flow cytometry, and the percentage of apoptosis was determined. The data show that although COMP expression leads to a significant reduction (3-4-fold) in the level of apoptosis, the COMP siRNA blocks this effect. For B and C the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown. 34). The levels of XIAP, cIAP1, cIAP2, and survivin were significantly increased in the COMP-expressing cells (Fig.  3C, left panels). To determine whether the increase in levels of these proteins was due to an increase in their cognate mRNAs, RT/PCR analysis was performed. As seen in Fig. 3C (right panels), the levels of their corresponding mRNAs were not affected by COMP. Because it has been demonstrated that NF-B transcriptionally induces the IAP family of proteins (33,34), we examined levels of NF-B in the control and COMP-expressing cells. We found no increase in NF-B or phosphoNF-B in the COMP-expressing cells (data not shown), which is consistent with the fact that the IAP transcripts are not up-regulated by COMP. These data would indicate that the elevation in cIAP1, cIAP2, survivin, and XIAP protein levels are likely regulated at the translational/ post-translational level.
XIAP and Survivin Induction Mediates the Survival Effect of COMP in 293 Cells-To determine whether these IAP family members mediate the survival effect of COMP in 293 cells, siRNA experiments were performed in an attempt to reduce their levels. XIAP and survivin were chosen as targets because they are the most potent inhibitors of caspase 3 (33). Following transfection of the XIAP and survivin siRNAs, their cellular protein levels were reduced by 3-4-fold compared with the control siRNA plasmid (Fig. 4, A and B). When the cells were analyzed for levels of apoptosis, as shown in Fig. 4C, COMP was not able to optimally enhance cell survival in the presence of either the XIAP or survivin siRNAs. The survivin siRNA was not as effective as the XIAP siRNA in blocking the antiapoptotic effect of COMP, suggesting differences in the way these two proteins mediate cell survival. These data indicate that COMP mediates cell survival at least in part through an elevation the XIAP and survivin proteins.
COMP Expression in 293 and HeLa Cells Leads to Resistance to TNF␣-From the data above it appears that expression of COMP reduces the level of transfectionmediated cell death in 293 cells. It was next important to determine whether induction of cell death by the addition of an apoptotic-inducing agent could be blocked by COMP. The cells were therefore transfected with COMP and then treated with TNF␣ and actinomycin D (ActD). As seen in Fig. 5A, TNF␣/ ActD treatment significantly reduced the number of viable cells by nearly 3-fold in the control transfected condition. However, TNF␣/ ActD treatment was not able to reduce viable cell numbers when COMP was expressed. Further, as shown in Fig. 5B, COMP was effective in blocking the increase in apoptosis induced by TNF␣/ActD. These data indicate that induction of cell death by an apoptotic-inducing agent in 293 cells could be blocked by COMP expression.
Although 293 cells provide an excellent cell type to test protein function, they are relatively insensitive to the apoptotic effects of TNF␣ because they require actinomycin D treatment to achieve a high level of apoptosis. To test whether COMP can block the apoptotic effects of TNF␣ alone, HeLa cells were employed because they are known to be sensitive to low doses of TNF␣ (35,36). When HeLa cells were Amaxa transfected with COMP, expression was evident as seen in the Western blot in Fig. 6A. When the HeLa cells were treated with TNF␣, they were found to be very sensitive to its toxic effects. A dose of 10 ng/ml led to an increase in apoptosis, as shown in Fig. 6B. However, TNF␣ had no effect on apoptosis in the cells expressing COMP (Fig. 6B). These data indicate that COMP blocks the death-inducing effects of TNF␣ in HeLa cells.
When the IAP family of proteins was examined in these cells, it appeared that XIAP and survivin protein levels were induced by COMP, whereas cIAP1 and cIAP2 were unaffected (Fig. 6C). Interestingly when the levels of mRNA were also examined, survivin was found to be up-regulated by 3.5-fold as shown in Fig. 6D. This is the first demonstration of a gene up-regulated by COMP and suggests that COMP signaling mediates tran-  (Fig. 1A). A, immunoblot. Cell extracts from the transfected cells were generated at 24 h post-transfection and were analyzed by immunoblot using antibodies that recognize active and procaspase 3 or actin. The data show a dramatic reduction in the level of active caspase 3 in the extracts of the COMP-expressing cells and a slight increase in the level of procaspase 3, but no change was evident in the actin control. B, immunoblot. The protein extracts from cells transfected as in A were used in immunoblot assays with antibodies to cytochrome c, Akt, phospho-Akt(Ser), Akt1, Bcl2, phosphoBcl2(Ser), BclXL, and Mcl-1. The data show no change in the levels of these proteins following COMP expression. C, immunoblot/RT/PCR. Left panels, the protein extracts were also used in immunoblot assays with antibodies to cIAP1, cIAP2, XIAP, survivin, and ␤-actin. The data show that the protein levels of cIAP1, cIAP2, XIAP, and survivin are up-regulated by COMP expression in 293 cells. Right panels, total RNA was isolated also from cells, and 1 g was used in RT/PCRs with primers for the indicated genes. The data show that the mRNA levels for these genes are not affected by COMP. scriptional activation. As mentioned above, NF-B is known to transcriptionally induce the IAP family of proteins; however, we found no increase in NF-B in the COMP-expressing HeLa cells (data not shown), indicating that the up-regulation of survivin was not mediated by this factor.
COMP Expression in Human Chondrocytes Leads to Increased Survival-Because expression of COMP aids in survival of 293 and HeLa cells, two transformed cell lines, it was important to determine whether this effect was seen in human chondrocytes, cells that are normally exposed to COMP. To determine whether this is the case, human articular chondrocytes were used in transfection studies. These chondrocytes in culture display a number of well known cartilage markers. As shown in Fig. 7A, they express aggrecan, the "a" and "b" forms of collagen type 2(␣1), collagen type 9(␣1), and collagen type 11(␣1). These chondrocytes were transfected with a COMP expression plasmid using the Amaxa nucleofector technique, which results in an 85-90% transfection efficiency. As seen in Fig. 7B, following transfection COMP is overexpressed in these chondrocytes as detected by Western blot of proteins in the medium. It is important to note that COMP is present in the control transfected lane in Fig. 7B; however, the blot was underdeveloped to show enhanced expression of COMP in the extracts of the COMP transfected cells.
When viable cells were examined it was found that COMP had a very potent effect on increasing their number (Fig. 7C) when compared with cells transfected with the vector control. Also, as shown in Fig.  7D (left panels, "untreated"), at 2 days post-transfection COMP had reduced the level of apoptosis in these cells, indicating that it is acting to block apoptosis in human chondrocytes.
It was next important to determine whether induction of chondrocyte death by the addition of an apoptotic-inducing agent could be blocked by COMP expression. The cells were therefore transfected with COMP and then treated with TNF␣ (10 ng/ml) and actinomycin D (0.2 g/ml). As seen in Fig. 7D, TNF␣/ActD treatment increased the level of apoptosis in the control transfected condition by nearly 3-fold. However, expression of COMP effectively blocked the ability of TNF␣/ActD to induce death. As shown in Fig. 7D, in the presence of COMP, TNF␣/ActD treatment did not lead to a significant increase in apoptosis. These data indicate that induction of chondrocyte death by an apoptotic-inducing agent could be blocked by COMP expression.
To more accurately determine the effective prosurvival dose of COMP, serum-free conditioned medium was generated and used at varying concentrations to treat pSec2B transfected cells. As shown in Fig. 7E, COMP was added simultaneously with Amaxa transfection, and apoptosis was monitored at 24 h postaddition/transfection. From the data it is clear that a dose range of 0.03-0.08 g/ml was effective at blocking apoptosis. Additionally, COMP at varying concentrations was added either 24 h prior to (Fig. 7F), or together with (Fig. 7G) TNF␣/Act D addition. In both cases COMP at a concentration of 0.3-0.8 . XIAP protein levels were assessed by immunoblotting followed by densitometry of the bands. The data indicate a 3-fold reduction in XIAP protein levels by the XIAP siRNA. B, 293 cells were transfected with the COMP-expressing plasmid with either a control siRNA (COMPϩControl siRNA) or with a survivin siRNA (COMPϩSurvivin siRNA). Survivin protein levels were assessed by immunoblotting followed by densitometry of the bands. The data indicate a 4-fold reduction in survivin protein levels by the survivin siRNA. C, apoptosis. The percentage of apoptotic cells was determined following transfection of cells with a vector control or COMP plasmid along with a control siRNA, the XIAP siRNA, or the survivin siRNA. Shown are the percent apoptotic cells determined by flow cytometry at 2 days post-transfection. The data show that both XIAP siRNA and survivin siRNA are effective at blocking the ability of COMP to affect cell survival. However, reduction of the survivin protein was not as effective in blocking the antiapoptotic effect of COMP compared with the XIAP siRNA. For A, B, and C, the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown. g/ml was effective at blocking apoptosis. These data suggest that a dose of 50 -150 picomolar is efficient in blocking apoptotis and that the cells do not need to be pretreated with COMP to affect survival.
Induction of the IAP Family of Prosurvival Proteins in Human Chondrocytes-Because COMP blocks the activation of caspase 3 in 293 cells, this critical effector caspase was also examined in the human chondrocytes. As shown in Fig. 8A, it was found that COMP was able to reduce caspase 3 activation considerably in human chondrocytes. These data would indicate that COMP promotes chondrocyte survival by blocking caspase 3 activation in a fashion similar to that in 293 cells.
To identify the mechanism by which COMP protects chondrocytes from apoptosis, we examined the levels of the IAP family of survival proteins. As shown in the immunoblots in Fig.  8B (left panels), XIAP, cIAP1, cIAP2, and survivin protein levels were induced by COMP. To determine whether the increase in levels of these proteins was due to an increase in the transcription of these genes, RT/PCR analysis was performed. The levels of their corresponding mRNAs (Fig. 8B, right panels) are not affected by COMP with the exception of survivin, whose mRNA levels were increased by COMP by nearly 4-fold (Fig.  8C). As seen with 293 cells, these data would indicate that the elevation in cIAP1, cIAP2, and XIAP protein levels are likely mediated at the post-translational level. However, the data would also indicate that survivin is transcriptionally up-regulated by COMP as it is in HeLa cells. As was the case for HeLa cells, we found no up-regulation in NF-B by COMP (data not shown), indicating that the transcriptional elevation of survivin is not mediated by this factor.

XIAP and Survivin Mediate the Survival Effect of COMP in
Human Chondrocytes-To determine whether these IAP family members mediate the survival effect of COMP in human chondrocytes, siRNA experiments were performed in an attempt to reduce their levels. XIAP and survivin were chosen as a target because their protein levels were increased significantly following COMP expression. As seen in Fig. 8 (D and E), following transfection of the XIAP and survivin siRNAs, their cellular protein levels were reduced by 5-and 2.5-fold, respectively, compared with the control siRNA plasmid. The chondrocytes were then transfected with either the COMP expression plasmid or a vector control, in the presence of a control siRNA, the XIAP siRNA, or the survivin siRNA. COMP was not effective in reducing the level of apoptosis in the presence of either the survivin or XIAP siRNA (Fig. 8F). These data indicate that COMP mediates cell survival at least in part through an elevation the survivin and XIAP proteins in human chondrocytes.
The CaCTD Domain of COMP Mediates Survival and Elevates the IAP Proteins-Because COMP is composed of multiple domains, it is possible that some of these domains would be responsible for its prosurvival function. As outlined in Fig. 9A, three deletion mutants were generated that span the COMP protein. Each mutant was Myc-tagged and cloned into a secretory expression plasmid (pSec2B). Chondrocytes were then Amaxa transfected with the mutant expression plasmids, and the proteins were expressed and detected in the culture media (Fig. 9B). When the chondrocytes, transfected as in Fig. 9B were treated with TNF␣/ActD, it was clear that only the CaCTD domain was effective in blocking apoptosis, whereas the CTD and NT domains were not (Fig. 9C).
To next examine whether the CaCTD domain could function to elevate the level of XIAP and survivin, immunoblots of the IAP proteins were assessed following the transfections. As evident in Fig. 9D, the CaCTD domain along with full-length COMP was effective at inducing XIAP and survivin but not the NT or CTD domains. In total, these data show that the CaCTD can induce both the IAPs and cell survival in human chondrocytes. It should be noted that we have recently examined a new deletion mutant of COMP that spans the NT domain and the type 2-EGF-like repeats. This mutant was not effective at either inducing the IAP proteins or in protecting against cells death. 3 Addition of an RGD Peptide to Chondrocytes Blocks the Ability of COMP to Protect against Cell Death-From the data above it appears that the sequence containing the type 3 repeats in COMP retains the survival function and the ability to induce the IAP proteins. Although the type 3 repeats are a well described calcium binding region (1,2), they also contain an RGD sequence that mediates cell attachment to surfaces through integrin binding (24). Given the positive role of integrins in mediating cell survival (37)(38)(39), it was important to determine whether the antiapoptotic function of COMP was mediated through integrin signaling. To determine whether this is the case, a short RGD peptide capable of blocking binding to integrins (40), or a control peptide, was added to chondrocytes overexpressing COMP. The cells were then monitored for survival. COMP and the CaCTD domain protect against death in the presence of the control peptide (Fig. 10A). However, the addition of the RGD peptide to cells blocks the prosurvival function of COMP and CaCTD. These data would indicate that integrin signaling mediates the survival effect of COMP.
It would be expected that the RGD peptide would block the induction of the IAP proteins by COMP and the CaCTD domain. To test this, the cells overexpressing COMP or the CaCTD domain were treated with either the control or RGD peptide, and the levels of XIAP and survivin were examined in cell extracts. As seen in Fig. 10B, the RGD peptide blocks the ability of COMP or the CaCTD domain to induce both XIAP and survivin. These data indicate that blocking integrin signaling interferes with the ability of COMP to induce the XIAP and survivin proteins and to affect cell survival.

DISCUSSION
Here we have identified a novel antiapoptotic function for the extracellular matrix protein COMP. This antiapoptotic effect is not cell type-specific and was found to occur in 293 cells, HeLa cells, and primary human chondrocytes. Given that COMP is expressed at sites other than connective tissue, such as vascular smooth muscle, synovium, skin, bone, pancreas, and liver (23,(41)(42)(43)(44)(45), this antiapoptotic effect may have broad implications for cell survival throughout the body. Interestingly, whereas the wild type COMP protein is able to block apoptosis, mutant COMP proteins found in the human diseases multiple epiphysial dysplasia and PSACH have been found to promote apoptosis (25)(26)(27). These data indicate that COMP can play opposite roles in the regulation of cell survival, depending on whether it is in a wild type or mutant form.
The prosurvival effect of COMP occurs in either the presence or absence of TNF␣ in all of the cells examined, indicating The HeLa cells transfected as in A were seeded at 1 ϫ 10 6 cells/10-cm plate in monolayer and then treated with or without 10 ng/ml TNF␣. The cells were cultured for an additional 72 h, and the percentage of apoptotic cells was assessed. The data show that TNF␣ treatment leads to an increase in the percent apoptotic cells in the control transfected cells. However, TNF␣ treatment does not significantly affect the level of apoptosis in the COMP-expressing cells. C, immunoblot/RT/PCR. Right panels, the protein extracts were also used in immunoblot assays with antibodies to cIAP1, cIAP2, XIAP, survivin, and ␤-actin. The data show that the protein levels of XIAP and survivin are up-regulated by COMP in HeLa cells. Left panels, total RNA was also isolated from the cells, and 1 g was used in RT/PCRs with primers for the indicated genes. The data show that the mRNA levels for cIAP1 and cIAP2 are downregulated by COMP, whereas XIAP is unaffected. However, the survivin mRNA was strongly up-regulated by COMP. D, survivin RNA levels were assessed by RT/PCR at varying cycle numbers followed by densitometry of the ethidium bromide-stained bands. The data indicate a 3.5-fold induction of survivin mRNA levels by COMP in HeLa cells. For B and D the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown.
that the survival effect may not be directed at the mechanisms surrounding a death receptor. Perhaps consistent with this notion, it was found that COMP signaling blocked the activation of caspase 3, a critical downstream effector caspase, whose action is required for induction of apoptosis by a variety of inducing agents (46, 47). We could not detect changes in the activation of any other caspase by COMP and so conclude that the mechanism must be directed to caspase 3. Examination of a large number of intracellular factors that mediate apoptosis and act through caspase 3 (e.g. Akt, Bcl2, cytochrome c, etc) revealed no alterations by COMP. However, when the IAP family of proteins was examined (XIAP, survivin, cIAP1, cIAP2), a striking up-regulation in their levels was identified. The IAP proteins are known to function to directly block the activation of caspase 3 (33,47), and as such they are potent anti-apoptotic factors. Although the IAP protein levels showed significant elevation in response to COMP in 293 cells, HeLa cells and human chondrocytes, the corresponding mRNAs for most of these IAPs, were not up-regulated, indicating that the increase in protein levels is likely due to translational/post-translational effects. The fact that transcripts for the IAP genes contain internal ribosome entry site sequences suggests that they are regulated translationally (33). Additionally it is known that the IAP proteins are modulated post-translationally via breakdown by the proteasome (33,48). In initial experiments it appears that COMP may regulate IAP stability because we find that proteasome activity levels are significantly down-regulated in the COMP expressing 293 cells and human chondrocytes. 3 Interestingly, the transcript levels for survivin were elevated by COMP in both HeLa cells and human chondrocytes. Survivin therefore appears to be the first gene identified that is induced by the action of COMP. Thus, depending on the cell type, COMP has distinct mechanisms for elevating the levels of the IAP family of proteins. The survivin gene does not appear to be regulated by pNF-B because this inflammatory mediator protein was not affected by COMP. Therefore a separate mechanism must exist for the transcriptional control of survivin by COMP.
That the IAPs are central to the survival effect of COMP was shown by siRNA experiments. siRNAs directed against either XIAP or survivin were effective in reducing their protein levels. In the presence of either the XIAP or survivin siRNA, COMP was not able to efficiently aid in cell survival, indicating that a reduction in at least one member of the IAP family abolished the survival effect of COMP. Thus it appears that the elevation of the IAP proteins is crucial for the prosurvival effect of COMP but that optimal cell survival may depend on the induction of multiple IAP proteins.
As mentioned above, the COMP mutants in multiple epiphysial dysplasia and PSACH are known to affect apoptosis of chondrocytes (25)(26)(27). One likely mechanism is that these COMP mutants induce apoptosis through an endoplasmic reticulum stress response because they are not efficiently secreted and are retained within the endoplasmic reticulum. A question that remains is whether or not these COMP mutants are able to induce the IAP proteins once they are secreted. It may be that the mutants are defective in this regard.
In addition to the effect on survival, COMP has been demonstrated to have a number of biological functions. It can inhibit cell proliferation (21), enhance attachment of cells to surfaces (23,24), regulate motility (23), and regulate apoptosis (Refs. 21 and 25-27 and data presented here). A central question is whether COMP can accomplish these diverse functions through the binding and activation of a single receptor. Although no unique receptor has been identified by which COMP acts, data suggest that the RGD sequence in the type 3 repeats can interact with integrins to mediate cell attachment (24). COMP exerting its prosurvival effects through integrin binding would be consistent with a role of integrins in cell survival. It is widely known that integrin signaling, mediated via cell attachment, is a requisite for survival of adherent cells (37)(38)(39). In this regard, the data presented here show that the CaCTD domain of COMP, which contains the RGD sequence, can protect cells against death. Impor- Human chondrocytes were Amaxa transfected with either a vector control (pSec2B) or the COMP-expressing plasmid. Cell extracts from the transfected cells were generated at 24 h post-transfection and were analyzed by immunoblot using antibodies that recognize active caspase 3 or ␤-actin. The data show a dramatic reduction in the level of active caspase 3 in the extracts of the COMP-expressing cells. The actin control shows no change. B, immunoblot, RT/PCR. Three days after transfection, the cells were harvested, and protein extracts and total RNA were generated. Left panels, the protein extracts were used in immunoblot assays with the indicated antibodies; cIAP1, cIAP2, XIAP, and survivin. The data show that the protein levels of cIAP1, cIAP2, XIAP, and survivin are up-regulated by COMP in these cells. Right panels, RT/PCR. Total RNA was isolated also from cells, and 1 g was used in RT/PCRs with primers for the indicated genes. The data show that the mRNA levels for these genes are not affected by COMP with the exception of survivin. C, survivin RNA levels were assessed by RT/PCR at varying cycle numbers followed by densitometry of the ethidium bromide-stained bands. The data indicate a 4-fold induction of survivin mRNA levels by the COMP in chondrocytes. D and E, human chondrocytes were Amaxa transfected with either a vector control or the COMP-expressing plasmid as in Fig. 7A with either a control siRNA (COMPϩcontrol siRNA), an XIAP siRNA (COMPϩXIAPsiRNA), or a survivin siRNA (COMPϩsurvivin siRNA). XIAP and survivin protein levels were assessed by immunoblotting followed by densitometry of the bands as shown. The data indicate a 4-fold reduction in XIAP protein levels and a 2.4-fold reduction in survivin levels. F, apoptosis. The percentage of apoptotic cells was assessed following transfection of cells with a vector control or COMP plasmid along with a control siRNA, XIAP siRNA, or survivin siRNA. Shown are the total viable cell numbers at 3 days post-transfection. The data show that both the XIAP and survivin siRNAs block the ability of COMP to increase cell survival, whereas the control siRNA had no effect. For C-F the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown.
tantly, when an RGD peptide was added to cells expressing either COMP or the CaCTD domain, it completely blocked the ability of these proteins to enhance cell survival. A control peptide had no such effect. These data implicate the RGD sequence within COMP in mediating the induction of both the IAPs and the antiapoptotic response. Thus it is possible that COMP mediates cell attachment and cell survival through the same integrin signaling pathway.
That integrin-induced signaling is linked to the IAP proteins is confirmed by a recent study. It has been demonstrated that under certain conditions, integrin signaling induces NF-B, which in turn upregulates the IAP proteins leading to cell survival (49). However, here we see no activation of NF-B and no transcriptional induction of most of the IAP family members, indicating that a translational/posttranslational mechanism may be involved in the elevation of the IAPs. Although the IAPs have been shown to be regulated by the proteasome, our initial experiments using proteasome inhibitors to test whether COMP mediates IAP upregulation are so far inconclusive. However, we do find that total proteasome activity drops significantly in cells expressing COMP (data not shown). The interesting exception to the above findings for IAP induction is that of the survivin gene, which is transcriptionally regulated by COMP in both HeLa cells and chondrocytes. We believe this is the first gene demonstrated to be transcriptionally regulated by COMP. Because NF-B is likely not involved in the transcriptional induction, it is likely that other transcription factors mediate this process.
Based on the idea that cartilage extracellular matrix proteins may play an important role in maintaining chondrocyte viability, we have shown for the first time that COMP plays a critical function in chondrocyte survival. COMP accomplishes this in a very novel way, by the induction of members of the IAP family of survival proteins, which are very effective at blocking the activation of caspase 3 and thereby blocking apoptosis. Yet the induction of most of these IAP proteins appears to be by a translational/post-translational mechanism. Finally, it is well recognized that COMP undergoes cleavage in patients suffering from various forms of arthritis and that the carboxyl-terminal half of COMP, containing the CaCTD domain, can be identified both FIGURE 9. The CaCTD domain of COMP retains the survival function and the ability to induce the IAP proteins. A, deletion mutants of COMP were generated, Myc-tagged, and cloned into the pSec2B expression vector that targets the expressed proteins for secretion. Next to each mutant are the amino acid residues spanned. B, immunoblots. Each plasmid was individually transfected into chondrocytes, the media were harvested, and cell extracts were generated at 24 h post-transfection. The data show an immunoblot of the secreted COMP deletion mutants using the anti-Myc tag antibody. C, apoptosis. Human chondrocytes transfected with the full-length and COMP mutants were treated with or without 10 ng/ml TNF␣ and 0.2 g/ml ActD. The cells were cultured for an additional 72 h, and the percentage of apoptotic cells were assessed by flow cytometry. The data show that TNF␣/ActD treatment leads to an increase in the number of apoptotic cells and that both full-length COMP and the CaCTD domain of COMP mediate cell survival, whereas the CTD and NT domains do not. D, immunoblots. Cell extracts from the cells transfected with the mutants as in A above were generated at 72 h post-transfection and were analyzed by immunoblot using antibodies that recognize the IAP family members and ␤-actin. The data show that both full-length COMP and the CaCTD domain induce XIAP and survivin, whereas the NT and CTD domains do not. The actin control shows no change. For C, the data are given as the means Ϯ S.D. The statistical significance with the indicated p values is shown.

FIGURE 10. An RGD peptide is able to block survival and IAP induction by COMP or the CaCTD domain.
Human chondrocytes were Amaxa transfected with either a vector control, full-length COMP, CTD or CaCTD expressing plasmids. A, apoptosis. At 24 h post-transfection, the cells were incubated with either an RGD peptide (cycloRGDFV, 1 mM) or a control peptide (cycloRADFV, 1 mM) for an additional 24 h at which time the percentage of apoptotic cells were assessed by flow cytometry. The data show that the RGD peptide completely blocks the ability of COMP or the CaCTD domain to mediate survival, whereas the control peptide had no effect on COMP-mediated survival. B, immunoblotting. XIAP and survivin protein levels were assessed by immunoblotting in the cells transfected and treated as in A. The data show that the RGD peptide, but not the control peptide, blocks the ability of both COMP and the CaCTD domain to induce XIAP and survivin. For A the data are given as the means Ϯ S.D. The statistical significance with the indicated p values are shown. in serum and in the synovial fluid of these patients (50 -52). The data presented here suggest that during the process of cartilage breakdown, some fragments of COMP still retain their antiapoptotic function. These fragments may therefore be bioactive and reduce the level of apoptosis within osteoarthritis cartilage. Further, COMP fragments in serum may also affect tissues at sites distal from cartilage.