Plasminogen-Mediated Activation and Release of Hepatocyte Growth Factor from Extracellular Matrix

Materials

L-ascorbic acid, heparin, heparin-agarose, guanidine-HCl, leupeptin, ε-aminocaproic acid, PMSF, and aprotinin were purchased from Sigma Chemical Company (St. Louis, MO). Human plasmin and α₂-antiplasmin were obtained from Calbiochem (La Jolla, CA). Human Glu-plasminogen was purchased from American Diagnostica Inc. (Greenwich, CT). Dulbecco's modified Eagle's media (DMEM) and FBS were purchased from Invitrogen (Carlsbad, CA). Penicillin-streptomycin-amphotericin B mixture was purchased from Cambrex (East Rutherford, NJ). Carrier-free human HGF, neutralizing anti-human HGF antibody, and normal goat IgG were purchased from R&D Systems (Minneapolis, MN). Goat anti-human HGFα antibody that was raised against a peptide mapping to the N terminus of human HGFα was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). This antibody cross-reacts with murine HGF. Rabbit anti-goat horseradish peroxidase–conjugated antibody was purchased from Bio-Rad (Hercules, CA).

Cells

NIH3T3 cells and Madin Darby canine kidney (MDCK) cells were obtained from the American Type Culture Collection (Rockville, MD) and grown to subconfluence in DMEM supplemented with 10% FBS at 37°C in a 5% CO₂, humidified environment. Subclones of MDCK cells were selected that form tightly aggregated colonies when grown in the absence of HGF. Primary cultures of mouse lung fibroblasts were prepared from three strains of mice. WT C57BL/6 mice were purchased from Charles River Laboratories, Inc. (Portage, MI). PAI-1^−/− mice were originally obtained from Dr. Peter Carmeliet (University of Leuven, Leuven, Belgium) and have been back-crossed with C57BL/6 mice for at least eight generations. C57BL/6 mice genetically deficient in uPA (uPA^−/−) were purchased from Jackson Laboratories (Bar Harbor, ME). Lung fibroblasts were isolated and cultured by a previously described technique (31). Briefly, mice were killed by sodium pentobarbital overdose, exsanguinated, and the lungs lavaged with PBS. After removing major airways and pulmonary blood vessels, the lungs were minced with scissors and the pieces placed into 100-cm² dishes in DMEM containing 15% FBS and penicillin-streptomycin-amphotericin B. The cells that grew from the tissue fragments were maintained in media at 37°C in a 5% CO₂ incubator and were serially passed a total of five times to yield pure populations of lung fibroblasts, as previously described (31). The fibroblasts were used after the fifth passage.

¹²⁵I Labeling of HGF

Carrier-free HGF was radioiodinated by modifications of the procedure of Lyon and colleagues (25). Briefly, 5 μg of recombinant human HGF was adsorbed to a 150-μl suspension of heparin-agarose in 0.5 M NaCl, 20 mM sodium phosphate, pH 7.0. Na^l25I (0.25 mCi) was added, followed by 50 μl of 0.1% (wt/vol) chloramine-T in 0.5 M NaCl, 20 mM sodium phosphate, pH 7.0. After 1 min, an additional 50 μl of chloramine-T solution was added. One minute later, the gel suspension was transferred to a small disposable column and washed extensively with 0.5 M NaCl, 20 mM sodium phosphate, pH 7.0, to remove all free ^l25I. Bound ^l25I-HGF (1,200–3,000 dpm/fmol) was then step-eluted with a small volume of 1.5 M NaCl, 20 mM sodium phosphate, 2 mg/ml BSA, pH 7.0, and stored at −20°C until used. The integrity of the ^l25I-HGF was confirmed by SDS-PAGE and autoradiography.

Preparation of ECM-Coated Culture Plates

ECMs were generated by NIH3T3 cells using the method of Cukierman (32). Briefly, NIH3T3 cells were added to 24-well plates and incubated for 5–7 d past confluence in DMEM, 10% FCS, antibiotic/antimycotic, and 50 μg/ml L-ascorbic acid. Cells and soluble proteins were removed by incubation for 5 min in 0.5% Triton X-100 plus 20 mM NH₄OH at room temperature, followed by four washes in PBS. For the experiments testing the effects of added plasminogen, matrices were also incubated for 2 h at 37°C with 5 mM PMSF to minimize residual serine protease activity within the matrices.

Binding and Release of ¹²⁵I-HGF from NIH3T3-Derived ECM

ECMs were incubated with 10 or 50 ng/ml ¹²⁵I-HGF for 2 h at 4°C in PBS. After extensive washing, the labeled matrices were immediately used for experiments. To detect inhibition of ^l25I-HGF binding by soluble heparin, ECMs were incubated for 2 h at 4°C with 10 ng/ml of ¹²⁵I-HGF and various concentrations of heparin. Unbound ¹²⁵I-HGF was removed by repeated washing, after which the ECMs were solubilized with 5 M guanidine HCl and radioactivity was measured using a γ counter (PerkinElmer, Boston, MA). To measure release of HGF from ECM, serum-free DMEM containing various reagents or cell types were added to wells containing ^l25I-HGF–labeled matrices. After incubation for various times at 37°C, media were collected and their radioactivity measured in a γ counter. The amount of radioactivity released from ECM when incubated with DMEM alone was subtracted as background.

Analysis of Released HGF

The molecular weights of the radiolabeled HGF that was released from ECM were analyzed by SDS-PAGE run under reducing conditions followed by autoradiography. To measure HGF bioactivity, an MDCK scatter assay was used (33). One hundred MDCK cells were added to each well of a 96-well plate and incubated for 4 d in DMEM with 10% FBS. The attached colonies were washed, and media from the HGF-release experiments were added. DMEM alone or containing 50 ng/ml recombinant HGF were added to a set of wells as negative and positive control conditions, respectively. To verify that the scatter activity observed in conditioned media was caused by HGF, 10 μg/ml of neutralizing anti-human HGF antibody was used. As an isotype-matched control antibody, normal goat IgG was used. After overnight incubation at 37°C, the cells were stained with 0.5% crystal violet in 50% ethanol for 10 min. Photographic images were digitally captured (Spot RT Slider camera and software; Diagnostic Instruments, Sterling Heights, MI) and the percentage of scattered colonies to total colonies in each well was determined. A colony was counted as “scattered” if the colony contained more than ten cells and half of them had lost contact with their neighbors and exhibited a fibroblastic phenotype (33).

Activation of Endogenously Produced HGF by NIH3T3 Cells or Mouse Lung Fibroblasts

NIH3T3 cells (1 × 10⁵ cells/well) were plated on 24-well tissue culture plates and incubated in DMEM, 10% FBS, 50 μg/ml ascorbic acid, with or without 20 μg/ml leupeptin, for 7 d. Media were changed every 2–3 d. To evaluate the activation state of HGF, conditioned media were collected and analyzed by SDS-PAGE followed by Western blot. To test the effects of plasminogen and plasmin on HGF activation, cells were cultured for 7 d with leupeptin, after which the cells were washed with PBS. Serum-free DMEM was added containing various concentrations of plasminogen or plasmin and with or without 20 μg/ml α₂-antiplasmin or 10 μg/ml aprotinin. After 24 h, media were collected and 20 μl of 50% slurry of heparin-agarose was added to concentrate the HGF.

To evaluate the ability of primary lung fibroblasts from WT, uPA^−/−, and PAI-1^−/− mice to activate HGF, cells (5 × 10⁵ cells/well) were plated onto 6-well tissue culture plates and incubated in DMEM, 15% FBS, 50 μg/ml ascorbic acid, with 20 μg/ml leupeptin, for 7 d. The cells were washed with PBS, and media were changed to serum-free DMEM containing 10 μg/ml plasminogen or 370 mU/ml plasmin (a concentration that is equivalent to 10 μg/ml plasminogen after full conversion). After 12 h, media were collected and added to heparin-agarose. The media-heparin-agarose mixtures were incubated at 4°C for 12 h with gentle rolling. After centrifugation and washing the heparin-agarose three times with PBS, SDS sample buffer (2% SDS, 50 mM dithiothreitol, 10% glycerol, and 0.1% bromphenol blue in 62.5 mM Tris HCl, pH 7.4) was added and the mixture placed in a 100°C water bath for 5 min. The samples were separated by SDS-PAGE, and the protein bands were transferred to a polyvinylidene fluoride membrane (Immobilon; Millipore, Bedford, MA). Blots were incubated with goat primary antibodies against HGFα diluted 1:500 in blocking solution for 1 h at room temperature, followed by incubation for 1 h at room temperature with rabbit anti-goat horseradish peroxidase–conjugated secondary antibody at a 1:5,000 dilution. Proteins were visualized after incubation of the blots in Western blot chemiluminescent substrate (Pierce Biotechnology, Rockford, IL), according to the manufacturer's protocol, and exposure to Hyperfilm (Amersham Bioscience, Buckinghamshire, UK). The densities of the pro-HGF and α-chain HGF were compared by densitometric analysis using Scion Image software (Scion Corporation, Frederick, MD).

Statistical Analysis

All values are shown as means ± SEM. The significance of differences between groups was assessed using ANOVA followed by Tukey's multiple comparison test for pairwise comparisons. Analyses were performed using Prism software (GraphPad Software, San Diego, CA), accepting a P < 0.05 as statistically significant.

¹²⁵I-HGF Binding to ECM and Release by Plasmin

To study the binding and release of HGF, ECMs were generated by NIH3T3 cells, after which the cells and soluble proteins were removed by treatment with Triton X-100 and NH₄OH. The cell-free ECMs were then labeled with ¹²⁵I-HGF. To demonstrate that HGF was binding to ECM via heparan sulfate moieties, soluble heparin was added to the wells during the labeling period to compete with ECM binding sites. Figure 1A demonstrates that heparin was able to inhibit ¹²⁵I-HGF binding to the ECM at concentrations known to block HGF binding to heparan sulfate (34). NaCl (2 M) was also able to elute ¹²⁵I-HGF from the ECM (data not shown). To assess the release of radiolabel from the ECM, the amount of ¹²⁵I-HGF in the media was measured at various times. Incubation with DMEM alone led to the release of a small amount of radioactivity (Figure 1B). This release was considered background and subtracted from the results of subsequent experiments. Treatment of the labeled matrices with 10 and 100 mU/ml plasmin resulted in an increased release of bound radiolabel into the media in a dose- and time-dependent fashion (Figure 1B).

Release of ¹²⁵I-HGF from ECM by NIH3T3 Cells

To determine if cells were able to release HGF from ECM, fresh NIH3T3 cells were added to ¹²⁵I-HGF–labeled cell-free matrices in the presence of various concentrations of plasminogen. The amount of radiolabel released into the media after 6 h of incubation was measured. The addition of plasminogen or cells alone released small amounts of radiolabel from the ECM (Figure 2A); however, when NIH3T3 cells (which are known to produce uPA [35–37]) and plasminogen were combined, the release of radioactivity from the matrices increased in a plasminogen dose–dependent fashion. Extending the incubation time of cells plus 10 μg/ml plasminogen to 24 h caused release of greater than 90% of the bound radioactivity (data not shown).

To further assess the role of plasminogen activation in the release of HGF, ε-aminocaproic acid was used to inhibit plasminogen activation and plasmin activity (38). Plasminogen and NIH3T3 cells were added to ¹²⁵I-HGF–labeled matrices in the absence or presence of ε-aminocaproic acid, and the release of radiolabel was measured after a 6-h incubation (Figure 2B). As shown previously, plasminogen or cells alone caused a small release of radiolabel into the media. ε-Aminocaproic acid inhibited the plasminogen-mediated release, but did not affect the amount of growth factor released by cells alone. Importantly, ε-aminocaproic acid inhibited the synergistic effect of cells and plasminogen in a dose-dependent fashion.

Characterization of the HGF Released from ECM by NIH3T3 Cells

To determine if the radioactivity released into the media by the NIH3T3 cells represented intact HGF rather than degraded peptides, supernatants from matrices exposed for 5 h to DMEM with or without 10 μg/ml plasminogen and/or 2 × 10⁵ NIH3T3 cells were separated by SDS-PAGE under reducing conditions. The resulting bands were visualized by autoradiography (Figure 3A). A control sample containing the stock ¹²⁵I-HGF that was used to label the matrices was included (Figure 3A, leftmost lane). The radioactivity in the stock ¹²⁵I-HGF migrated as bands of molecular weight appropriate for intact pro-HGF (∼ 92 kD) and the active, cleaved α-chain and β-chain forms (∼ 69 kD and 34 kD, respectively). The majority of this stock ¹²⁵I-HGF was in the cleaved, active conformation that migrates as α- and β-chains. Exposure of ¹²⁵I-HGF–labeled ECM to DMEM alone revealed no radioactive bands in the media after incubation for 5 h. ECM exposed to plasminogen or 3T3 cells alone released radioactive material that migrated in bands similar to the stock ¹²⁵I-HGF. In accordance with our previous results, the intensities of the bands were greatest from ECM that was exposed to both cells and plasminogen.

To determine if the released HGF retained biological activity, an MDCK scatter assay was used. Matrices were prepared, labeled with HGF, and then incubated with 10 μg/ml plasminogen and/or 2 × 10⁵ NIH3T3 cells. After 5 h, the media were transferred to wells containing compact colonies of MDCK cells, and the amount of colony scattering measured after overnight incubation. DMEM and recombinant HGF (50 ng/ml) were used as negative and positive controls, respectively (Figure 3B). Media conditioned by exposure to HGF-labeled ECM for 5 h contained no detectable scatter activity. Plasminogen or cells alone released small amounts of scatter activity; however, the combination of cells and plasminogen had a synergistic effect on the release of scatter activity in condition media. The scatter activity generated by fibroblasts and plasminogen was inhibited by 10 μg/ml of an anti-HGF neutralizing antibody.

Release of ¹²⁵I-HGF from ECM by Mouse Lung Fibroblasts

To further evaluate the role of plasminogen activation on the release of ¹²⁵I-HGF from ECM, fibroblasts were isolated from the lungs of WT, uPA^−/−, and PAI-1^−/− mice, and 5 × 10⁴ cells were plated onto radiolabeled matrices in the absence (control) or presence of 10 μg/ml plasminogen. After 10-h incubation, media were collected, and the amount of radioactivity was measured (Figure 4). Fibroblasts from WT mice released an increased amount of radiolabel above background (P = 0.03). The level was increased further when plasminogen was present in the culture environment (P < 0.001). Cells from uPA^−/− mice also released a small amount of radiolabel, but the further addition of plasminogen had no incremental effect (P > 0.05). Finally, fibroblasts from PAI-1^−/− mice caused a small release of radiolabel (P < 0.002) that was greatly accentuated with the addition of plasminogen (P < 0.001).

Activation of Endogenously Generated HGF

Commercially available sources of recombinant HGF are not well suited for studying processes involved in its activation because a high proportion of the HGF has already been cleaved to the active disulfide-linked heterodimer. An alternative approach was therefore taken, in which NIH3T3 cells or mouse lung fibroblasts were allowed to generate their own HGF in a manner that maintained the majority of HGF in the single-chain, pro-HGF conformation. This was accomplished by adding to the culture media the protease inhibitor leupeptin, which blocks an enzyme, HGF activator, which is present in bovine plasma and is activated by thrombin during FBS processing (39). The amount of HGF in the pro- and active form could then be assessed by Western analysis using an antibody that recognizes an epitope on the α-chain. Under reducing conditions, the larger pro-HGF can be easily distinguished from the smaller cleaved α-chain. If leupeptin is not added to the serum-containing media used during NIH3T3 monolayer formation, the majority of HGF is in the activated form, as indicated by the dominance of the α-chain band (Figure 5A). On the other hand, after 7 d in culture in the presence of leupeptin, the majority of the HGF is in the pro-HGF form.

To determine if NIH3T3 cells could activate pro-HGF using plasminogen, 7-d monolayers were washed and incubated in serum-free, leupeptin-free media containing various concentrations of reagent plasminogen or plasmin. After 24 h, conditioned media were collected and equal volumes were analyzed by SDS PAGE and immunoblotting (Figure 5B). In the absence of added plasminogen or plasmin, the majority of the HGF released into the media was in the pro-HGF form. Addition of plasminogen or plasmin caused activation of pro-HGF in a dose-dependent manner, as indicated by the accumulation of α-chain. Inclusion of aprotinin (inhibitor of serine proteases) or α₂-antiplasmin (inhibitor of plasmin, but not plasminogen activators at the concentration used [40–42]) blocked the effect of plasminogen or plasmin on HGF activation.

To evaluate the role of uPA in HGF activation, 5 × 10⁵ lung fibroblasts from WT, uPA^−/−, and PAI-1^−/− mice were added to wells and grown for 7 d in media containing serum and leupeptin. When cultured for another 24 h in serum-free and leupeptin-free media, the HGF generated by all 3 genotypes was found to be partially activated (Figure 6A, representative of duplicate experiments). The addition of plasmin caused the majority of the pro-HGF to be activated, regardless of genotype. When plasminogen was added to the fibroblast cultures, the increase in pro-HGF activation by PAI-1^−/− cells was greater than the increase induced by plasminogen in the WT fibroblast cultures. In fact, the level of activation of HGF by PAI-1^−/− fibroblasts rivaled that achieved by the addition of plasmin. In marked contrast, there was no plasminogen effect on uPA^−/− fibroblasts. In particular, addition of plasminogen to uPA^−/− fibroblasts caused no increase in pro-HGF activation above that which was induced by cells incubated in media alone. We note that the combined densities of the bands in the WT genotype lanes were less than those in the uPA^−/− and PAI-1^−/− lanes. This is because, in this experiment, the monolayer cell density of the WT fibroblasts was less than that of the other two genotypes, and equal volumes of conditioned media were analyzed. To semiquantify the results, the ratios of densities of the α-chain HGF to pro-HGF are shown in Figure 6B.