Preparation and characterization of recombinant tissue inhibitor of metalloproteinase 4 (TIMP-4).

TIMP-4, a novel human tissue inhibitor of metalloproteinase, was identified and cloned (Greene, J., Wang, M., Raymond, L. A., Liu, Y. E., Rosen, C., and Shi, Y. E. (1996) J. Biol. Chem. 271, 30375-30380). In this report, the production and characterization of recombinant TIMP-4 (rTIMP4p) are described. rTIMP4p, expressed in baculovirus-infected insect cells, was purified to homogeneity by a combination of cation exchange, hydrophobic, and size-exclusion chromatographies. The purified protein migrated as a single 23-kDa band in SDS-polyacrylamide gel electrophoresis and in Western blot using a specific anti-TIMP-4 antibody. Inhibition of matrix metalloproteinase (MMP) activities by rTIMP4p was demonstrated in five MMPs. Enzymatic kinetic studies revealed IC50 values (concentration at 50% inhibition) of 19, 3, 45, 8, and 83 nM for MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9, respectively. Purified rTIMP4p demonstrated a strong inhibitory effect on the invasion of human breast cancer cells across reconstituted basement membranes. Thus, TIMP-4 is a new enzymatic inhibitor in MMP-mediated extracellular matrix degradation and may have therapeutic potential in treating cancer malignant progression.

MMPs 1 and their inhibitors TIMPs play a critical role in ECM homeostasis. Controlled remodeling of the ECM is an essential aspect of normal development, and deregulated remodeling has been indicated to have a role in the etiology of diseases such as arthritis, periodontal disease, and cancer metastasis (1)(2)(3)(4)(5). Four mammalian TIMPs have been identified so far: TIMP-1 (6), TIMP-2 (7), TIMP-3 (8 -11), and the recently cloned TIMP-4 (12,41). The proteins are classified based on structural similarity to each other, as well as their ability to inhibit metalloproteinases.
TIMPs are secreted multifunctional proteins that have anti-MMP activity as well as erythroid-potentiating and cell growth-promoting activities. The stimulating effect on cell growth was initially recognized when TIMP-1 and TIMP-2 were identified as having erythroid-potentiating activities (14,15). It is now clear through several recent reports that TIMP-1 and TIMP-2 are mitogenic for non-erythroid cells, including normal keratinocytes (16), fibroblasts (17), lung adenocarcinoma cells (18), and melanoma cells (18). The involvement of TIMPs in the activation of pro-MMP has also been demonstrated (19). In addition, the recent evidence indicates that the TIMP family may be involved in steroidogenesis of rat testis and ovary indicating the potential role of TIMP in the reproduction (20).
The most widely appreciated biological function of the TIMPs is their role in the inhibition of cell invasions in vitro (21)(22)(23)(24) and tumorigenesis (25)(26)(27)(28)(29) and metastasis in vivo (25)(26)(27)(28)(29)(30)(31). Since the net MMP activity is the result of the balance between activated enzyme levels and TIMP levels, an increase in the amount of TIMPs relative to MMPs could function to block tumor cell invasion and metastasis. The tumor-suppressing activity of TIMP on primary tumor growth may be in part due to its anti-angiogenic activity. In fact, both TIMP-1 (32) and TIMP-2 (33,34) have been demonstrated to have an antiangiogenic activity, and such inhibition of angiogenesis is mediated by inhibition of both endothelial cell proliferation (34) and migration (32). The underlying molecular mechanism for the tumor suppressing activities of TIMPs, nevertheless, is thought to depend on their anti-MMP activities.
We had recently cloned and characterized a human TIMP-4 (12). Transfection of TIMP-4 into human breast cancer cells inhibited the invasion potential of the cells in the in vitro invasion assay (13). When injected orthotopically into nude mice, TIMP-4 transfectants were significantly inhibited in their tumor growth and axillary lymph node and lung metastasis as compared with controls (13). These results suggest the therapeutic potential of TIMP-4 in treating cancer malignant progression. These results suggest an important role of TIMP-4 in inhibiting primary tumor growth and progression leading to invasion and metastasis. In the present study, we have produced and purified rTIMP4p from baculovirus infected cells. rTIMP4p was shown to inhibit MMP activity and tumor cell invasion across reconstituted basement membrane.

MATERIALS AND METHODS
Reagents-Restriction enzymes were obtained from Boehringer Mannhem. Chromatography supplies were purchased from PerSeptive. All the other reagents are listed as indicated below.
Preparation of rTIMP4p from Baculovirus-The coding sequence of TIMP-4 is amplified using a standard polymerase chain reaction ap-proach with the primers corresponding to the 5Ј and 3Ј sequences of the cDNA (5Ј primer: GCT AGT GGA TCC CTG CAG CTG CGC CCC GGC  G; 3Ј primer: CGG CTT CTA GAA GGG CTG AAC GAT GTC AAC). The amplified fragment was gel-purified and digested with BamHI and XbaI. To construct the recombinant baculovirus expression vector, the purified TIMP-4 polymerase chain reaction fragment was ligated into pA2-GP vector, which was derived from pVL94 (35). The resulting pA2-GP/TIMP4 vector was transfected into HB101 cells, and positive clones were identified using polymerase chain reaction screening and restriction enzyme analysis. The DNA sequence was confirmed by automatic DNA sequencing of both strands. A recombinant virus was produced and purified. Sf9 insect cells were infected with recombinant baculovirus in EXCEL401 serum-free medium (JRH Scientific) supplemented with 1% pencillin/streptomycin (Life Technologies, Inc.) and 1% fetal bovine serum (Life Technologies, Inc.). A 5-L Bioreactor was harvested 70 h post-infection, and cell viability was estimated to be 80%. The bioreactor supernatant was clarified using a continuous flow centrifuge (18,000 ϫ g). The harvesting and subsequent chromatography steps were carried out at 4 -8°C.
Purification of rTIMP4p-As we previously described (13,36), an MMP-2-mediated gelatin degradation assay was used to monitor the anti-MMP activity of rTIMP4p during the purification. The clarified supernatant was directly loaded onto a strong cation-exchange column (POROS HS50 from PerSeptive Biosystems; column dimensions, 3 ϫ 10 cm) at a linear flow rate of 800 cm/h. The column was previously equilibrated with 50 mM sodium acetate, 100 mM NaCl, pH 5.8, for 10 column volumes. The bound proteins were eluted using the step elution of 200 mM NaCl, 400 mM NaCl, 600 mM NaCl, 1 M NaCl, and 2 M NaCl (in the same equilibration buffer as before). The 600 mM eluted fractions were found containing anti-MMP-2 activity. The active fractions were pooled and diluted with 50 mM sodium acetate, pH 5.8, to a conductivity of six millisiemens.
A weak cation-exchange column (POROS CM20 from PerSeptive Biosystems; column dimensions, 2 ϫ 7 cm) was equilibrated with 10 column volumes of 50 mM Tris-HCl, 100 mM NaCl, pH 7.5. The pooled active fractions from the strong cation-exchange column were loaded onto the column at a linear flow rate of 840 cm/h. The bound proteins were first washed with the equilibration buffer and then eluted by a gradient elution using the equilibration buffer and the elution buffer containing 50 mM Tris-HCl and 1 M NaCl, pH 7.5. The gradient elution was conducted from 0.1 M to 1 M NaCl within a 10-bed volume. The active fractions were pooled, and NaCl was added to raise the conductivity to 200 millisiemens.
A moderate hydrophobic interaction chromatography column (PO-ROS PE50 from PerSeptive Biosystems; column dimensions, 2 ϫ 10 cm) was equilibrated with 10 column volumes of buffer containing 50 mM sodium acetate and 4 M NaCl, pH 5.8. The pooled fractions from the weak cation-exchange column were loaded onto the column at the flow rate of 100 ml/45 min. The bound proteins were eluted using 50 mM sodium acetate and 100 mM NaCl, pH 5.8. The eluted active fractions from the column were pooled together.
The pooled fractions from the hydrophobic column were loaded onto a size-exclusion column (Superdex S-200 from Pharmacia Biotech Inc.; column dimensions, 2.5 ϫ 90 cm) with a flow rate of 20 ml/30 min. The was previously equilibrated using 50 mM sodium acetate and 100 mM NaCl, pH 5.8. The sizing fractions were analyzed by SDS-PAGE and Coomassie Blue staining, and the relevant fractions corresponding to a size of 23 kDa were pooled and tested for purity and anti-MMP activity.
Preparation of Anti-TIMP-4 Antibody-A peptide sequence corresponding to amino acids 207-225 of human TIMP-4 (12) was synthesized by an ABI 431A peptide synthesizer. Peptide synthesis reagents were from Advanced Chemtech, Louisville, KY. The purified peptide was conjugated to keyhole limpet hemocyanin (Sigma) via 1-ethyl-3-(3dimethylaminopropyl)carbodiimide (Pierce). New Zealand rabbits were immunized with the carrier-hapten conjugate in Freund's complete adjuvant (Pierce) followed by Freund's incomplete adjuvant at the recommended intervals. Animals were anesthetized and exsanguinated, and antibodies were separated from serum by ion exchange. For final purification, a TIMP-4 peptide affinity column was made by conjugating 20 mg of TIMP-4 peptide to 5 ml of AminoLink resin (Pierce Chemical Co.) using sodium cyanoborohydride (Sigma).
Gelatin Degradation Assay-Inhibition of enzymatic activity by rTIMP4p was assayed by measuring degradation of [ 3 H]gelatin as we previously described (13).
In Vitro Invasion Assay-Inhibition of breast cancer cell invasions by purified rTIMP4p was evaluated in the Matrigel invasion assay with reconstituted basement membrane as we previously described (13).

RESULTS
Expression and Purification of rTIMP4p-rTIMP4p was produced in Sf9 insect cells using the baculovirus expression system. A pVL94-based transfection vector PA2-GP/TIMP4 was constructed to generate the recombinant virus which was subsequently used to infect Sf9 cells. The optimal yield of rTIMP4p was obtained from the conditioned medium of the infected cells at 70 h post-infection. The best purification of rTIMP4p was achieved by a 4-step chromatography including a strong cation chromatography, a weak cation chromatography, a hydrophobic interaction chromatography, and a size-exclusion column. The rTIMP4p eluted from a size exclusion column was stored in the buffer containing 50 mM sodium acetate and 100 mM NaCl, pH 5.8. When analyzed by SDS-PAGE, this preparation showed a single band at molecular mass of 23 kDa (Fig. 1A), which is consistent with the predicted molecular mass based on the calculation from the protein sequence (12,41). The purified 23-kDa protein was confirmed as TIMP-4 by Western blot using a specific anti-TIMP-4 antibody (Fig. 1B). The consistency of calculated molecular mass and the actual molecular mass of purified rTIMP4p suggests that there was no posttranslational glycosylation for rTIMP4p. This is in agreement with the absence of the glycosylation site for TIMP-4 (41). In addition, rTIMP4p gave a negative result in the glycosylation test using modified periodic acid-Schiff method (42) (data not shown). The specific activity of the recovery of rTIMP4p is summarized in Table I. The yield of purified rTIMP4p was approximately 1.7 mg/2 ϫ 10 7 cells.
Characterization of Anti-MMP Activities of rTIMP4p-The inhibitory activity of rTIMP4p on MMPs was analyzed by a soluble gelatin degradation assay. As shown in Fig. 2, when MMP-2 and MMP-9 were incubated with purified rTIMP4p at the mole ratio of 1 to 2, the gelatinolytic activities were inhibited 88% for MMP-2 and 66% for MMP-9, respectively. A similar pattern with higher magnitude of inhibition was also observed for TIMP-2, suggesting that TIMP-4 may be more specific for MMP-2 in a manner similar to TIMP-2.
Kinetic analysis of the inhibition of MMPs by rTIMP4p was performed in a continuous fluorometric assay with a quenched fluorescent peptide substrate. The inhibition kinetics of rTIMP4p were analyzed against human MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9. The MMPs were incubated with different concentrations of rTIMP4p. As shown in Fig. 3, the inhibitor concentrations that reached to 50% inhibition of MMP activities (IC 50 ) were determined to be 19, 3, 45, 8, and 83 nM for MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9, respectively. Therefore, TIMP-4 is a potent inhibitor of all five tested MMPs, and it has preference for MMP-2 and MMP-7.

Inhibition of Invasion Potential of Human Breast Cancer Cells-Previously, we demonstrated that transfection of TIMP-4 cDNA into human breast cancer cells inhibited tumor cell invasion cross-reconstituted basement membrane (Matrigel) (13). The effect of purified rTIMP4p on the invasion of MDA-MB-435 human breast cancer cells was investigated.
MDA-MB-435 cells were moderately invasive. At the end of a 24-h incubation, about 10% of MDA-MB-435 cells had crossed the Matrigel barrier. A significant reduction in invasive potential was noted when rTIMP4p was added at two different concentrations. The percentages of invaded cells were 1.5% for the cells treated with 10 nM rTIMP4p and 0.6% for the cells treated with 100 nM rTIMP4p, respectively. To facilitate the comparison of the relative invasiveness between controls and rTIMP4p-treated cells in this study, all values were normalized to the percent invasion of control MDA-MB-435 cells which were taken as 100% (Fig. 4).
To rule out the possibility that the different invasion potentials between the control cells and rTIMP4p-treated cells are due to the potential inhibitory effect of TIMP-4 on cell growth, we conducted growth rate experiments to determine whether addition of rTIMP4p affects the growth of MDA-MB-435 cells. When the cells were treated with 20, 50, and 100 rTIMP4p in the Dulbecco's modified Eagle's medium containing 5% fetal calf serum (changing the fresh medium and rTIMP4p every 2 days) for 7 days, no significant differences in growth rate were observed between the control and rTIMP4p-treated cells (data not shown). These results are consistent with our previous report on the similar growth rates of the control MDA-MB-435 cells and TIMP-4-transfected cells (13). DISCUSSION Active recombinant TIMP-4 protein is required for characterization of its biochemical activity against MMPs and biological functions in inhibiting tumor growth and metastasis. Proteins of eukaryotic cells expressed in Escherichia coli are often generated as inactive, insoluble aggregates known as inclusion bodies and therefore require in vitro complicated refolding. In the present study, we expressed, purified, and characterized recombinant TIMP-4 protein prepared from baculovirusinfected insect Sf9 cells. The identity of rTIMP4p was confirmed by several criteria. First, as expected, the purified protein had a molecular mass of 23 kDa in SDS-PAGE, which is in close agreement with the calculated molecular mass of the 22.5-kDa protein, based on the mature protein of 195 amino acids after removal of the signal sequence (12). Second, the purified protein possessed a metalloproteinase inhibitory activity against MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9. Third, the purified protein can be recognized immunochemically by an affinity-purified specific anti-TIMP-4 polyclonal antibody. Fourth, the purified protein inhibited tumor cell invasion in the Matrigel invasion assay, and a similar effect was also reported for other TIMPs (21,22).
We demonstrated here that human recombinant TIMP-4 can effectively inhibit human MMP-1, MMP-2, MMP-3, MMP-7, and MMP-9, with the IC 50 values of 19, 3, 45, 8, and 83 nM, respectively. This relatively higher potency of TIMP-4 against MMP-2 than other MMP suggests that TIMP-4, like TIMP-2, is more specific for MMP-2. In fact, the predicted structure of the TIMP-4 shares 37% sequence identity with TIMP-1 and 51% identity with TIMP-2 (12). In addition, we also demonstrated a high affinity interaction between TIMP-4 and the C domain of MMP-2 and showed that TIMP-4 bound both full-length Conditioned media (5 liters) from the recombinant baculovirus-infected Sf9 cells were collected and subjected to different chromatographies as described under "Materials and Methods." The activity of rTIMP4p was determined in the gelatin degradation assay. Protein present in the different steps of purification was estimated by Bio-Rad protein assay using bovine serum albumin as a standard. a One unit activity of rTIMP4p is defined as 60% inhibition of the enzymatic activity of recombinant MMP-2 (0.3 g/ml) in the soluble gelatin degradation assay containing 20 l of pooled active fraction.
b Specific activity was calculated by dividing total activity by total protein.

FIG. 2. Inhibition of MMP-2 and MMP-9 by rTIMP4p in a soluble gelatin degradation assay.
MMP-2 (100 ng) and MMP-9 (120 ng) were incubated either with TIMP-4 or TIMP-2 at the mole ratio of 1 to 2 (MMP versus TIMP). The TIMP-mediated inhibition of enzymatic activity was expressed as percentage of the corresponding MMP activities without added inhibitors. The basal level enzymatic activity without the inhibitor was taken as 100%. The number represents the mean Ϯ S.E. of three tests.
MMP-2 and the C domain of MMP-2 in a manner similar to TIMP-2 (43). Binding of MMP-2 to TIMP-4 was of high affinity with an apparent K d of 1.7 ϫ 10 Ϫ7 M but sightly weaker than that to TIMP-2 (apparent K d of 6.6 ϫ 10 Ϫ8 M) (43). These K d differences are in agreement with the relatively more potent inhibitory effect of TIMP-2 on MMP-2 than that of TIMP-4 (Fig.  2). The overall sequence identity between TIMP-4 and other TIMPs suggests that TIMP-4 may inhibit MMPs through a similar mechanism by forming a strong noncovalent complex with a 1:1 stoichiometry (44). Although the inhibitory activity of TIMP is distributed throughout the molecule, the N-terminal regions of the TIMP family are highly conserved and thus may contribute to the inhibitory activities, and the C-terminal regions are divergent and may enhance the selectivity to the target enzymes (41,45). A more detailed structural comparison indicated that TIMP-4 shares a relatively high identity with TIMP-2 particularly in the loops of 4 and 5 within the Cterminal domain (41). Thus, it is possible that TIMP-4 and TIMP-2 may share similar mechanistic and functional properties based on the sequence identity, similar enzymatic kinetics, and the high affinity binding to MMP-2 (43).
Augmented MMP activities are associated with the metastatic phenotype of carcinomas, especially breast cancer (46 -49). The down-regulation of MMPs may occur at the levels of transcriptional regulation of the genes, activation of secreted proenzymes, and through interaction with TIMPs. The clinical importance of MMPs during the tumor progression emphasizes the need to effectively block MMPs and the subsequent tumor cell invasion. The inhibitory effect of TIMPs on MMP activity leads one to expect that an increase in the amount of TIMPs relative to MMPs could function to block tumor cell invasion and metastasis. Indeed, tumor cell invasion and metastasis can be inhibited by up-regulation of TIMP expression or by an exogenous supply of TIMPs (23, 24, 28 -31). Alternatively, down-regulation of TIMP-1 and TIMP-2 have also been reported to contribute significantly to the tumorigenic and invasive potentials of the cells (25)(26)(27). These results suggest that an inhibitory activity of TIMPs play an important role in inhibiting tumor cell malignant progression leading to invasion and metastasis.  4. Inhibition of cell invasion by rTIMP4p. The bottom wells of the invasion chamber were filled with Dulbecco's modified Eagle's medium containing 10% serum. MDA-MB-435 cells were seeded at a density of 50,000 cells/ml/well with or without rTIMP4p at the indicated final concentrations, and cell invasions were analyzed 24 h later as described previously (13). The invasion potential of the cells incubated without rTIMP4p was expressed as 100%. The number represents the mean Ϯ S.E. of three cultures. In this study, we demonstrated an inhibitory effect of the purified rTIMP4p on the invasion of human breast cancer cells, which is consistent with our previous studies on the inhibition of cell invasion on the TIMP-4-transfected cells compared with the TIMP-4 negative control cells (13). In the experimental Matrigel invasion assay, approximately 95% inhibition of invasion potential was achieved when the breast cancer cells were treated with 100 nM rTIMP4p. Similar inhibitory effects with much less magnitude were also reported for TIMP-1 (21) and TIMP-2 (22) on different tumor cells. The almost complete suppression of invasion potential of breast cancer cells by rTIMP4p suggests that the major matrix degradation proteinases required for the invasion of breast cancer cells in the Matrigel invasion assay are MMPs, and their enzymatic activities can be inhibited effectively by TIMP-4. The inhibition of breast cancer cell invasion by both an exogenous supply of rTIMP4p and the endogenous expressed TIMP-4 suggest that the TIMP-4-mediated anti-invasion activity could be physiologically or pathologically relevant in the tumor microenvironment.
Using in situ hybridization analysis, we have demonstrated a stromal expression of TIMP-4 mRNA in the fibroblasts surrounding the breast carcinomas. 2 The expression of TIMP-4 in the stroma adjunct to the breast carcinomas may indicate one of the host responses to try to balance the local tissue degradation due to the tumor cell invasion. Therefore, availability of the excess TIMP-4 relative to MMP (either by exogenous supply or endogenous expression) would create a microenvironment in the tumoral-stromal interface where the MMP-mediated ECM degradation and the subsequent tumor cell invasion can be inhibited by TIMP-4. While we are aware that the Matrigel in vitro invasion assay may not be an accurate predictor of breast cancer cell invasion as it occurs in vivo, we have recently demonstrated the TIMP-4-mediated anti-tumor and anti-metastasis activities of TIMP-4 transfected breast cancer cells in the animal model (13). These results support a role for MMPs and the inhibitor TIMP-4 in breast cancer cell invasion. Therefore, the potential therapeutic value of TIMP-4 for controlling cancer progression warrants further investigation.