Intracellular activation of gelatinase A (72-kDa type IV collagenase) by normal fibroblasts

Abstract

Normal fibroblasts cultured as monolayers secrete matrix metalloproteinases (MMP), including gelatinase A (72-kDa type IV collagenase) as inactive zymogens. Previously we found that normal fibroblasts cultured in a type I collagen lattice (dermal equivalent) secrete active gelatinase A. Here we show that the activation of progelatinase A occurs within the cell and that the activator copurifies with Golgi membranes. Cell extracts of fibroblasts cultured in collagen lattices contain active 62-kDa gelatinase A at least 4–6 h before active enzyme is detected in the culture medium. Pulse–chase experiments confirm these results. The activator is membrane-bound and localizes to the Golgi-enriched fraction. Highly purified plasma membranes from lattice cultures are unable to convert gelatinase A from the zymogen to its active form. The activator may be a metalloproteinase because EDTA prevents activation of exogenous proenzyme by membrane fractions. Membrane-type MMP1, the enzyme thought to be responsible for activation of gelatinase A on the plasma membrane of tumor cells, shows no significant change in either mRNA or protein levels during lattice culture. Intracellular levels of gelatinase A mRNA and protein increase during the culture period, and tissue inhibitor of metalloproteinases concentration does not change. Because of the greater availability of tissue inhibitor of metalloproteinases-free proenzyme as a substrate for the activator, it is possible that membrane-type MMP1 is the activating enzyme. In that case, malignant transformation may involve a change in the localization of the activator to the plasma membrane.

The enzymes in the matrix metalloproteinase (MMP) family are implicated in multiple physiological and pathological processes related to extracellular matrix turnover, such as wound healing, angiogenesis, tumor invasion, and metastasis (1–5). Gelatinase A (72-kDa type IV collagenase; MMP2) is thought to be especially important in tumor invasion and metastasis because of its ability to degrade type IV collagen (6–8). Like all MMPs, gelatinase A is produced as a zymogen, and therefore its activation is a subject of major importance. In contrast to most members of this family of secreted enzymes, it cannot be activated by serine proteinases such as plasmin or by catalytic quantities of other MMPs (9, 10). A possible mechanism of activation in tumor cells has been proposed based on data showing activation of progelatinase A at the external cell membrane of these cells (11–13); a new subfamily of the MMPs—the membrane-type MMPs (MT-MMPs)—has been proposed as the activators (14–16).

It has been assumed that normal fibroblasts activate gelatinase A using the same mechanism as tumor cells. In this paper, we show that normal fibroblasts use a different route for proenzyme activation. Normal dermal fibroblasts grown on plastic in monolayer culture secrete progelatinase A, and not the active proteinase, into the culture medium. When cultured in a lattice of type I collagen, however, they secrete active gelatinase A as well as the proenzyme. We previously found that this activation is mediated by the α₂β₁ integrin receptor (1). The results presented here show that, in normal fibroblasts, integrin-mediated activation of progelatinase A occurs intracellularly. Active enzyme can be detected in cell extracts at least 4–6 h before it is seen in the medium. The activator is not present in plasma membranes but localizes to the Golgi membrane fraction.

MATERIALS AND METHODS

Cell Culture.

Human fibroblast cell strains were established from normal adult skin by outgrowth from explants as described (17). Cultures were maintained with DMEM/F12 containing a mixture of 5% newborn calf serum and 3% bovine embryonic fluid (Sigma). To exclude interference resulting from latent and active gelatinase A existing in serum and embryonic fluid, experiments were performed in serum-free DMEM/F12 containing insulin transferrin selenium (ITS) culture supplement (Collaborative Biomedical Products). ITS medium has been shown to be an adequate serum substitute for several days of serum-free culture in the type I collagen lattice (1).

Fibroblast-Populated Collagen Lattices.

Lattices were prepared using sterile type I collagen from rat tail tendons as described (1). Collagen concentration varied in the lattice from 0.5 to 1 mg/ml. Lattice mixtures were made in pH-adjusted DMEM/F12 such that the culture conditions were identical to the control monolayer plates, and the mixtures were kept at 4°C to maintain solubility of the collagen. After the addition of cells, the collagen mixture was thoroughly mixed and pipetted in 1- to 2-ml aliquots into 35-mm bacteriological dishes that had been precoated with BSA to prevent fibroblast adhesion. Gelation began immediately and was completed in less than 2 min at 37°C, trapping the cells within the lattice. After 1 h, lattices were detached from the sides of the dish by gentle tapping.

Zymogram Analysis of Proenzyme Activation.

Latent and active gelatinases were detected by zymogram analysis using SDS/polyacrylamide gels copolymerized with 2.5% gelatin as described (18). These enzymes are dissociated from tissue inhibitor of metalloproteinases by the presence of SDS during electrophoresis. Removal of SDS after electrophoresis allows the proenzymes to renature in an active or partially active conformation. After a suitable incubation period, the enzymes degrade the polymerized substrate, permitting the visualization of both proenzyme and smaller, active forms as clear bands against the blue background of stained, undigested gelatin (19).

Western Immunoblotting Analysis.

Samples of conditioned medium were concentrated ≈4-fold by lyophilization. Concentrated media samples (20 μl) or 5 μg of protein from cell lysates was electrophoresed on 10% or 12% SDS/polyacrylamide gels and electrically transferred to a polyvinylidene difluoride membrane (Millipore) following the product directions. After blockage of nonspecific binding sites with 5% nonfat milk, blots were incubated for 1 h at room temperature with primary antibody. After extensive washing, blots were incubated with alkaline phosphatase-conjugated second antibodies and developed with 5-bromo-4-chloro-3-indoylphosphate p-toluidine salt/nitroblue tetrazolium. Polyclonal antibody raised in chickens against gelatinase A was affinity purified using a column of gelatinase A coupled to Affi-Gel 10 (Bio-Rad). A monoclonal anti-peptide antibody (gift of Gregory Goldberg, Washington University) was used for Western blotting of MT-MMP1.

Preparation of Cell Lysate.

Cell extracts for Western immunoblots were prepared in several different manners. Gelled collagen lattices containing embedded cells were centrifuged at low speed to remove as much water as possible, then suspended in a large volume of ice-cold PBS, mixed well, and centrifuged again. The wash procedure was performed five to six times. The entire lattice was then mixed with either Triton X-100 lysis buffer, hypotonic buffer, or isotonic-buffered sucrose and forced several times through a syringe with a 21-gauge needle. Collagen debris was removed by pelleting in a microcentrifuge. All buffers contained protease inhibitors at the following concentrations: 10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride or Pefabloc (Boehringer Mannheim), 1 μg/ml leupeptin, 10 μg/ml pepstatin, and either 10 μg/ml aprotinin or 20 μg/ml benzamidine. The detergent-containing lysis buffer was buffered with 100 mM Tris and contained 2% Triton X-100. Hypotonic buffer consisted of 10 mM Tris (pH 7.5) and 10 mM NaCl. Isotonic-buffered sucrose contained 0.25 M sucrose buffered with 10 mM Hepes (pH 7.6).

In some experiments, cells were digested from collagen lattices using 1200 units/ml clostridial collagenase type V (Sigma); 5% serum was added to inhibit proteinases, and the cells were washed repeatedly with PBS before homogenization in a Dounce homogenizer or in a syringe with a 21-gauge needle. Monolayer cell extracts were made using cells either scraped or trypsinized from the dish.

Metabolic Labeling and Purification of Progelatinase A.

[³⁵S]methionine-labeled progelatinase A was prepared from culture medium of the p2AHT7211A cell line (20, 21) (a gift from Gregory Goldberg, Washington University). The proenzyme was purified using Procion red–agarose (Sigma) chromatography followed by affinity chromatography on gelatin–agarose as described (22).

Subcellular Fractionation and Assays of Enzymatic Markers.

Cells were digested from collagen lattices after 24 h at 37°C using clostridial collagenase type V (Sigma) at 100 units/mg collagen. The digestion was complete after about 15 minutes with shaking. The cells were harvested, washed several times in culture medium containing 10% newborn bovine serum, counted, and suspended in STM buffer [0.25 M sucrose/5 mM Tris, pH 7.5/1 mM MgCl₂/10 μg/ml pepstatin/10 μg/ml aprotinin/1 μg/ml leupeptin/1 mM Pefabloc (Boehringer Mannheim)] at 10⁷ cells/ml. The cell suspension was homogenized in ice with up and down strokes in a loose-fitting Dounce homogenizer until microscopic inspection showed ≈80% broken cells. The homogenate was centrifuged at 1000 × g, and the supernatant was removed and saved on ice. The pellet, consisting of debris and unbroken cells, was resuspended in a small quantity of STM buffer and rehomogenized with 30 strokes of the Dounce homogenizer. After centrifugation at 1000 × g, the pellet was discarded, and the supernatant was combined with the first. This was considered the whole cell homogenate. The homogenate was then centrifuged at 1500 × g (3500 rpm) in a Sorvall-refrigerated centrifuge for 30 min. The supernatant was ultracentrifuged at 100,000 × g for 1 h, and the pellet was labeled “intracellular membranes.” The supernatant was labeled “cytosol.” The pellet from the 1500 × g, 30-minute centrifugation was washed in ice-cold STM buffer, resuspended in a small volume of 0.25 M sucrose buffered with 10 mM Tris (pH 7.5), and homogenized with three strokes of the loose-fitting Dounce homogenizer. A small aliquot was taken and labeled “pre-Percoll membranes.” The remainder was layered on top of a mixture of 16.5 ml of 0.31 M sucrose/12.1 mM Tris (pH 7.5) and 3.5 ml of Percoll (Sigma) (23) and centrifuged at 12,000 × g for 15 minutes. Plasma membranes were recovered as a white band near the top of the gradient and washed in 0.15 M NaCl/10 mM Tris (pH 7.5). The washed pellet was resuspended in 0.05 M Tris (pH 7.5)/0.01 M CaCl₂ and labeled “Percoll plasma membranes.” In some experiments, the intracellular membranes were further fractionated by layering on discontinuous sucrose gradients and centrifuging at 250,000 × g in a Beckman SW-40 rotor for 2 h.

Purity of fractions was monitored using standard enzyme assays as markers. Present only in plasma membranes, 5′-nucleotidase was assayed spectrophotometrically using 5′-AMP as a substrate (24). Microsomal NADH–cytochrome c reductase, also assayed spectrophotometrically, was used as the marker for the endoplasmic reticulum (25). The marker enzyme for the Golgi fraction, galactosyl transferase, was measured by the transfer of [³H]galactose from UDP galactose to ovomucoid (26).

Total Cellular RNA Preparation and Northern Blot Analysis.

Cells were cultured in collagen lattices for 3, 6, and 24 h. Total cellular RNA was extracted with the Trizol Kit (GIBCO/BRL). Total RNA (20 or 30 μg) was denatured at 95°C for 3 minutes and separated on formaldehyde-1% agarose gels, passively transferred to nylon membranes, and fixed at 80°C. Northern blot analysis was performed with cDNA probes labeled with the Genius I system (Boehringer Mannheim) and detected by chemiluminescence (Genius VII, Boehringer Mannheim). cDNA probes were a gift of Gregory Goldberg (Washington University). Quantitation was performed by direct densitometric measurement of the bands seen on x-ray films using the National Institutes of Health image 1.52 program.

Immunoprecipitation.

Fibroblasts in monolayer culture were pulsed with 10 μCi/ml (1 Ci = 37 GBq) trans-labeled [³⁵S]methionine for 20–30 min. They were removed from the flasks by trypsinization, washed extensively with cold PBS, and incorporated into collagen lattices with nonradioactive medium as described above. At various time points, conditioned medium was collected, and the lattices digested with bacterial collagenase to release the cells. Cell pellets were lysed, and the protein concentration in cell lysates was determined with either the Quantigold (27, 28) or Bradford (29) protein assay methods.

Protein-adjusted aliquots of cell lysates were precleared twice by mixing for at least 4 h at 4°C with 25 μl of recombinant protein A–Sepharose beads (Zymed) that had been previously washed with 0.05 M Tris (pH 7.5) containing IB buffer (40 mM EDTA/0.1% Triton X-100/2 mM Pefabloc/20 μg/ml pepstatin/20 μg/ml aprotinin/2 μg/ml leupeptin). Affinity-purified chicken anti-gelatinase A polyclonal antibody (20 μg/ml) was mixed with protein A beads previously conjugated with 30 μl/ml goat anti-chicken immunoglobulin G–alkaline phosphatase (Sigma) and blocked with serum. The beads containing both primary and secondary antibody were mixed with the precleared samples and agitated at 4°C overnight. The beads containing the immunocomplexed ³⁵S-labeled gelatinase were washed twice with IB buffer containing 0.5% BSA, once with IB buffer containing 1 M NaCl, and once with IB buffer. The immune complexes were removed from the beads by boiling in SDS sample buffer; samples were electrophoresed on SDS/PAGE gels that were soaked in Fluoro-Hance (Research Products International) for 45 minutes, dried, and subjected to autoradiography.

RESULTS

Active Gelatinase A Is Found in Cell Extracts Before It Appears in Culture Medium.

As reported, active gelatinase A can be detected by zymography in cell culture medium within 24 h after normal dermal fibroblasts are embedded in a type I collagen lattice. Cycloheximide treatment prevents activation (36), indicating a requirement for protein synthesis to achieve proenzyme activation. In the time course experiment shown in Fig. 1, active gelatinase A was clearly detected as a second clear band ≈10 kDa below the proenzyme band in the 2-h cell extract but was barely detectable even in 6-h culture medium. Because fibroblasts exhibit minimal cell division when cultured in the lattice (30, 31), the accumulation of activity in the extracts reflects an actual increase in intracellular gelatinase A protein. Northern blot analysis showed a nearly 2-fold increase in gelatinase A mRNA after 24 h of lattice culture (Fig. 2a). Western blotting using affinity-purified antibody to gelatinase A confirmed the zymographic results (Fig. 2b), indicating that the increased mRNA resulted in a higher intracellular concentration of the enzyme. Fibroblasts grown in monolayer culture also were extracted but never showed intracellular 62-kDa active enzyme. Zymography is a semi-quantitative assay. Fig. 3 shows that, when zymograms were adjusted so that the clear areas representing gelatin degradation were approximately equal in both the culture medium and extracts, activation was still detected in cell extracts but not in the culture medium.

Figure 1.

Time course of gelatinase activation in conditioned medium and cell extracts. Aliquots containing 10⁶ fibroblasts were incorporated into 2-ml collagen lattices at time 0. At the indicated times, culture medium was collected, and the lattices were washed and extracted as described. Equal volumes of conditioned medium and equal amounts of protein from cell extracts were then subjected to gelatin zymgography.

Figure 2.

(A) mRNA for gelatinase A as shown by Northern blot analysis and quantitated by normalizing to 28s RNA. (B) Time course of appearance of 62-kDa gelatinase A in cell extracts. Western immunoblots of cell extracts using affinity-purified gelatinase A antibody were performed as described. The arrow indicates 62-kDa active enzyme. Lanes: A, monolayer control extract 24 h in ITS serum-free medium; B, monolayer control extract 70 h in ITS serum-free medium; C, 0 time extract of trypsinized cells; D, 20-min lattice extract; E, 1-h lattice extract; F, 3-h lattice extract; G, 5-h lattice extract; H, 24-h lattice extract; and I, 48-h lattice extract.

Figure 3.

Normalized zymograms of conditioned medium and cell extracts from lattices containing 2 × 10⁶ cells. The zymograms were adjusted so that the amount of substrate digestion by progelatinase A was approximately equivalent in the culture medium and corresponding cell extract. The cells always contained active enzyme before being seen in culture medium. CM, culture medium; Ext, whole cell extract.

Figs. 1 and 3 show extracts made using the washed collagen lattice with cells embedded. Extrapolating from the proposed mechanism of progelatinase A activation by tumor cells, it seemed possible that gelatinase A might actually be secreted as a proenzyme, activated at the cell surface, and then bound to the collagen fibers in the lattice. To address this concern, fibroblasts were digested out of the lattice at various time points using bacterial collagenase, washed extensively, and then homogenized. To detect any artifacts that might be created by the detergent Triton X-100, both lattices and fibroblasts digested free from lattices were homogenized using hypotonic-buffered saline. As shown in Fig. 4, all results were essentially identical in the ratio of active enzyme to proenzyme and were unaffected by any residual bacterial collagenase, which accounts for the heavier bands of gelatin digestion in the figure. Cells removed from the lattice and homogenized using isotonic sucrose produced identical zymograms (not shown).

Figure 4.

Comparison of zymograms from cell lysates made before and after digestion from collagen lattices with clostridial collagenase. Twenty four-hour lattice cultures were extracted either with Triton X-100 lysis buffer or a hypotonic buffer containing proteinase inhibitors. Identical lattices were digested with bacterial collagenase, and the washed cell pellets were extracted with either buffer. Bands above the 72-kDa gelatinase A result from residual clostridial collagenase.

Pulse–Chase Experiments Show Newly Synthesized Gelatinase A in Cell Extracts Before Appearance in Culture Medium.

Fibroblasts from monolayer culture were starved for 20 minutes in cysteine and methionine-free DMEM/F12, then pulsed with 5 μCi/ml trans-labeled [³⁵S]methionine. The cells were harvested, extensively washed, and mixed into collagen lattices using unlabeled medium. At the time points indicated in Fig. 5, both cells and medium were harvested for immunoprecipitation, followed by SDS/PAGE and autoradiography. Immunoprecipitable active gelatinase A was seen in the 4- and 24-h cell extracts before it was detected in the medium.

Figure 5.

Autoradiogram of immunoprecipitation experiment showing activation of metabolically ³⁵S-labeled gelatinase A in collagen lattices with time. Pulse–chase was performed as described. Ext, cell extract; Med, conditioned medium.

Several experiments were performed to determine if the delay of appearance of active gelatinase A in the medium was a function of binding to the collagen lattice. First, exogenous [³⁵S]methionine-labeled progelatinase A was added to monolayer and lattice cultures containing equal amounts of cells. As seen in Table 1, after extracting and washing, monolayers and lattices retained an equal but small amount of the added radioactivity; these results weigh against the collagen in the lattices being a “sink” for gelatinase. In addition, in experiments in which cell-free lattices were incubated with labeled enzyme to determine binding to the collagen fibers, radioactivity from both active and progelatinase A bound equally to the fibers (Table 1). The higher extent of binding in the cell-free lattices probably reflects the fact that these noncontracted gels were quite highly hydrated; radioactivity was removed with each wash. Further washing in some experiments reduced the radioactivity bound to the lattice pellet to 6% (not shown). These experiments indicate that helical collagen does not significantly bind gelatinase, in either its pro- or active form.

Table 1.

Binding of ³⁵S-labeled gelatinase A

Experiment	Form of Gel A	Substrate	Radioactivity bound, %
1	Proenzyme	Fibroblast monolayer	4
	Proenzyme	24-h lattice	3.8
2	Proenzyme	Cell-free collagen gel	9
	Active enzyme	Cell-free collagen gel	10

In experiment 1, ≈50,000–70,000 cpm (0.5–1 μg) of ³⁵S-labeled gelatinase A was added to equal numbers of cells in monolayer or in lattice culture. After 24 h, the monolayer and contracted lattice were washed extensively, and the cells were extracted with lysis buffer. Supernatants, washes, and extracts were all counted separately in a scintillation counter. The percentage of radioactivity bound represents the remaining counts after six washes and extraction. In experiment 2, the labeled enzyme was incorporated into cell-free collagen lattices and incubated for 2 h at 37°C. Medium was removed, and the lattices were washed three to five times with cold PBS. Medium, washes, and lattice were counted separately. Counts bound remained with the lattice pellet. Gel A, gelatinase A. 

The Activator of Gelatinase A Copurifies with Golgi Membranes.

Initial experiments to determine the location of the gelatinase A activator consisted of incubating either 24-h conditioned medium or washed, 24-h contracted lattices overnight at 37°C with exogenously supplied ³⁵S-labeled progelatinase A. In neither case was any active gelatinase A seen upon autoradiography (not shown). To localize activator activity within the cell, subcellular fractionation experiments were undertaken. Enzymatic markers were used as a measure of the proportion of different membranes within any fraction. The marker for plasma membranes was 5′-nucleotidase (24). For endoplasmic reticulum, NADH cytochrome c reductase activity was measured (25). Galactosyl transferase was the marker enzyme for Golgi membranes (26). Table 2 lists the activities of marker enzymes in the fractions used to activate ³⁵S-labeled progelatinase A in Fig. 6. As shown in Table 2, after centrifugation in Percoll, a plasma membrane fraction was obtained that was essentially free of either Golgi or endoplasmic reticulum membranes. This fraction contained approximately twice the concentration of plasma membrane marker enzyme as other fractions but was unable to activate exogenously supplied ³⁵S-labeled gelatinase A. These experiments clearly show that the activator induced through α₂β₁ integrin is not in the plasma membrane. Table 2 shows that we were unable completely to separate endoplasmic reticulum from Golgi membranes by this method. Nonetheless, as seen in Fig. 6, the fraction that was most enriched in the Golgi marker enzyme galactosyl transferase (intracellular membranes) was best able to activate progelatinase A. Zymograms of cell fractions from early time points showed a concentration of active gelatinase A in the Golgi-enriched fractions (not shown). The extent of gelatinase A activation roughly parallels the galactosyl transferase activity in the various fractions with the exception of the cytosol, which contains galactosyl transferase activity apparently detached from the membrane during the extraction process. It should be emphasized that activator activity never was seen in the soluble fraction and always was associated with membranes. Fractionation experiments using other methods, such as sucrose density gradients, also showed that the membrane fractions most enriched in the Golgi marker enzyme galactosyl transferase contained the activator. Experiments such as these clearly show that the activator in normal fibroblasts is associated with intracellular membranes and not plasma membranes. The activation by the Golgi-enriched membranes was completely inhibited by 10 mM EDTA but not by serine or the sulfhydryl proteinase inhibitors tested (data not shown). It is thus most likely a metalloproteinase.

Table 2.

Marker enzyme activities in the activating fractions of the experiment shown in Fig. 5

Fraction	Marker enzymes
	Plasma membrane, 5′-Nucleotidase, μMol Pi/30 min	Endoplasmic reticulum, NADH cytochrome c reductase, μmol NADH/30 min	Golgi membranes, UDP galactosyl transferase, nmol galactosyl × 105/16 h
Crude plasma membranes	0.216	46	9.8
Percoll-purified plasma membranes	0.487	0	1.3
Golgi-enriched membranes	0.299	28	12.8
Cytosol	0.085	0	9.3

Figure 6.

Activation of exogenous ³⁵S-labeled progelatinase A by subcellular fractions. Samples were normalized to 5′-nucleotidase activity so that the highly purified Percoll plasma membrane fraction contained twice the amount of plasma membrane marker enzyme activity as the other membranous fractions. The rationale was that, if the activator were bound to the plasma membrane, the extent of activation would be greatest in the Percoll plasma membrane fraction. In the case of cytosol, where there was no 5′-nucleotidase activity, the maximum volume was loaded. Samples were incubated for 18 h at room temperature before electrophoresis and autoradiography. Lanes: A, proenzyme alone; B, proenzyme plus Golgi-enriched membranes; C, proenzyme plus Percoll-purified plasma membranes; D, proenzyme plus crude plasma membranes; E, proenzyme plus cytosol; and F, molecular mass markers.

Role of MT-MMP.

The question of whether the activator in integrin-mediated progelatinase A activation is MT-MMP1 was investigated using Northern blotting analysis for mRNA and Western immunoblot analysis for MT-MMP1 protein. MT-MMP1 mRNA showed a small increase with time in the collagen lattice (Fig. 7A), but a corresponding increase in immunoreactive protein was not found in cell extracts from lattices compared with monolayer extracts (Fig. 7B). The possible role of MT-MMP1 in the activation of progelatinase A by normal cells therefore is still unclear.

Figure 7.

(A) Northern blot analysis of mRNA for MT-MMP1 after 24 h in lattice culture. (B) Western immunoblot of MT-MMP1 protein in a cell extract from a 24-h lattice culture. Equal amounts of protein were used from the monolayer control extract and the lattice extract. The lower immunostaining band is seen in other publications but is not identified (16).

DISCUSSION

The results presented in this study show that normal fibroblasts in a collagen matrix activate gelatinase A intracellularly. Zymography and immunoblotting of cell extracts showed that active 62-kDa gelatinase A was present 4–6 h before the activity could be detected in the culture medium. Pulse–chase labeling of immunoprecipitated enzyme confirmed these findings. Conceivably, the presence of active gelatinase A in cell extracts could involve an activator on the outside surface of the plasma membrane. This possibility was eliminated by the finding that highly purified plasma membranes were unable to activate exogenously labeled progelatinase A. An intracellular membrane copurifying with the galactosyl transferase activity specific to Golgi membranes accomplished extensive activation. Although the identity of the activator was not definitively established, it appears to be a membrane-bound proteinase, inhibitable by EDTA. It is interesting to note that Salamonsen et al. (32) showed immunohistochemical localization of gelatinase A over the Golgi complex in stromal fibroblasts from endometrium.

Figs. 1 and 2 clearly show an increase with time in intracellular gelatinase A concentration of cells cultured in type I collagen lattices. These results from zymography were confirmed by Western immunoblot analysis. Furthermore, mRNA for gelatinase A also was increased in collagen lattices. A relationship between the increase in enzyme production and activation is not clear. However, we have shown that TIMP2 mRNA concentration does not change in lattice culture (1), so it is possible that mass action plays a role, making more TIMP2-free progelatinase A available to react with the activator.

A candidate for the activating enzyme might be MT-MMP1 (14). This enzyme is found in normal fibroblasts at low levels. However, Northern blot analysis of MT-MMP1 mRNA showed only a slight increase after cells were cultured in collagen lattices, and immunoblots showed no significant increase in MT-MMP1 protein. It is certainly possible that no increase in MT-MMP1 is necessary to produce active enzyme when substrate proenzyme concentration is increased (see above). It is also possible, in view of the requirement for protein synthesis for activation of gelatinase A in lattice culture, that an activator of MT-MMP1 proenzyme is induced. Still another possibility is that a completely different proteinase processes gelatinase A in normal cells. We presently are investigating possible roles for the other MT-MMPs.

There have been a number of studies using transformed cell lines and cell lines derived from metastatic cancers that show gelatinase A localized at the cell membrane (11, 13, 33). In addition, transformed cells are capable of activating exogenously supplied, labeled gelatinase A (34). In similar experiments using normal fibroblasts in collagen lattices, we were not able to show activation of ³⁵S-labeled progelatinase A (not shown). Likewise, MT-MMP1 has been shown to be located on the plasma membrane in cancer cell lines (14, 16) and has been proposed as a “docking agent” for progelatinase A (35). In this paper, we investigated integrin-mediated activation of progelatinase A in normal fibroblasts cultured in collagen lattices. We have shown that an EDTA-inhibitable activator that converts 72-kDa progelatinase A to its 62-kDa active form is not present in plasma membranes but colocalizes with Golgi-rich intracellular membranes in these cells. If future experiments show that the gelatinase A activator in normal fibroblasts is indeed MT-MMP1, then there is the intriguing possibility that malignant transformation involves a change in the localization of MT-MMP to the plasma membrane. Studies are underway to explore this possibility.

ABBREVIATIONS

MMP

matrix metalloproteinase

MT-MMP

membrane-type MMP

ITS

insulin transferrin selenium