Mechanisms underlying the antifibrotic properties of noninhibitory PAI-1 (PAI-1R) in experimental nephritis

Abstract

Administration of a mutant, noninhibitory PAI-1 (PAI-1R), reduces disease in experimental glomerulonephritis. Here we investigated the importance of vitronectin (Vn) binding, PAI-1 stability and protease binding in this therapeutic effect using a panel of PAI-1 mutants differing in half-life, protease binding, and Vn binding. PAI-1R binds Vn normally but does not inhibit proteases. PAI-1AK has a complete defect in Vn binding but retains full inhibitory activity, with a short half-life similar to wild-type (wt)-PAI-1. Mutant 14-lb is identical to wt-PAI-1 but with a longer half-life. PAI-1K has defective Vn binding, inhibits proteases normally, and has a long half-life. In vitro wt-PAI-1 dramatically inhibited degradation of mesangial cell ECM while the AK mutant had much less effect. Mutants 14-1b and PAI-1K, like wt-PAI-1, inhibited matrix degradation but PAI-1R failed to reverse this inhibition although PAI-1R reversed the wt-PAI-1-induced inhibition of ECM degradation in a plasmin-, time-, and dose-dependent manner. Thus the ability of PAI-1 to inhibit ECM degradation is dependent both on its antiproteinase activity and on maintaining an active conformation achieved either by Vn binding or mutation to a stable form. Administration of these PAI-1 mutants to nephritic rats confirmed the in vitro data; only PAI-1R showed therapeutic effects. PAI-1K did not bind to nephritic kidney, indicating that Vn binding is essential to the therapeutic action of PAI-1R. The ability of PAI-1R to remain bound to Vn even in a high-protease environment is very likely the key to its therapeutic efficacy. Furthermore, because both PAI-1R and 14-1b bound to the nephritic kidney in the same pattern and differ only in their ability to bind proteases, lack of protease inhibition is also keyed to PAI-1R's therapeutic action.

increasing evidence indicates that the plasminogen activator (PA)/plasminogen/plasmin system plays a dominant role in renal ECM degradation (2, 11, 12, 15, 28, 32). Plasmin, produced by the action of uPA or tPA on plasminogen, can directly degrade the matrix proteins fibronectin, laminin, proteoglycan, and type IV collagen as well as fibrin (26, 27, 29). Plasmin also converts inactive matrix metalloproteinases to active forms that degrade collagenous proteins (25, 38). PAI-1, as the principal inhibitor of uPA and tPA, constitutes a powerful, negative regulatory system to control the formation and activity of plasmin. Decreased PA and plasmin and increased PAI-1 have been reported in many experimental and human glomerular diseases characterized by mesangial matrix accumulation (7). The importance of PAI-1 in normal glomerular mesangial matrix turnover was clearly shown when matrix degradation increased fourfold after a monoclonal antibody to PAI-1 was added to cultured mesangial cells (1). That PAI-1 facilitates ECM accumulation in vivo was shown in the bleomycin model of pulmonary fibrosis when decreased fibrosis was seen in PAI-1-deficient and increased fibrosis was seen in PAI-1-overexpressing mice (8). Treatment of PAI-1 null mice with the inhibitor of plasmin formation, tranexamic acid, reversed the protective effect of PAI-1 deficiency, indicating that decreased disease in PAI-1 null mice was due to increased plasmin activity (13). Similar results have recently been shown in the kidney (15, 31). Thus it has been thought that the increased PAI-1 levels seen in fibrotic diseases limit plasmin generation and thereby promote persistence of fibrin deposits and accumulation of fibrotic matrix.

PAI-1 is synthesized and released in an active conformation with an inherently brief half-life in vivo (14). The structural characteristics of PAI-1 endow the protein with a strong tendency to assume a noninhibitory and thermodynamically stable latent conformation (30). However, this transition to the inactive latent form is slowed by binding to vitronectin (Vn), a glycoprotein of 70 kDa present in both plasma and the ECM. Binding to Vn stabilizes active PAI-1 and localizes it to its site of action. Expansion of the main β-sheet of PAI-1, through either the latency transition or cleavage of the reactive center loop, results in disruptions of the complex and the release of inactive PAI-1 and unliganded Vn (5, 23, 33, 34, 42). Therefore, Vn acts as a protein cofactor to determine the efficiency of PAI-1's protease-inhibitory ability. It is possible that interfering with PAI-1:Vn binding in vivo may diminish the profibrotic activity of PAI-1.

PAI-1 also has plasmin-independent actions. PAI-1:Vn binding directly blocks the interactions of Vn with cell-surface integrins required for cellular motility (35) and with the receptor for cell-bound uPA (uPAR) (4). Whether these actions play a role in fibrosis is unknown.

A mutant PAI-1 (PAI-1R) was designed with no proteinase-inhibitory activity but retaining Vn binding (37). Wild-type (wt)-PAI-1 loses affinity for Vn upon binding a protease and is then targeted to an endocytic, degradative pathway. Since the PAI-1R mutant does not inhibit proteases and retains Vn binding, it may remain bound to Vn longer than native PAI-1, thereby effectively competing with wt-PAI-1 for Vn binding sites if both of them are present in vivo. Investigating the therapeutic potential of this mutant, we have shown that PAI-1R administration to nephritic rats increases glomerular plasmin generation and reduces disease (17). Using the plasminogen/plasmin blocker tranexamic acid, we have also shown that the therapeutic effect of PAI-1R is largely plasmin mediated (15). The present study examines the effect of wild-type and mutant PAI-1 molecules on ECM degradation in vitro and in vivo, using the anti-thy-1 model of glomerulonephritis, to determine the importance of stability, Vn binding, and protease binding in the therapeutic action of PAI-IR.

MATERIALS AND METHODS

Study 1: Effects of PAI-1 Mutants on ECM Degradation by Mesangial Cells In Vitro

Materials.

Unless otherwise indicated, materials, chemicals, and culture media were purchased from Sigma-Aldrich (St. Louis, MO).

Proteins.

All mutant forms were constructed using the Transformer site-directed mutagenesis kit (Clontech) according to the manufacturer's instructions and isolated as described (21). Preparations having significant detectable endotoxin levels following phenyl-Sepharose chromatography were further purified by polymyxin B chromatography (Detoxi-Gel, Pierce). Final endotoxin levels for all proteins used were below the FDA limit for parenteral drugs of 5 endotoxin units/Kg. The PAI-1 proteins used were as follows: 14-1b, a form of PAI-1 mutant that contains the four-point mutations N150H-K154T-Q319L-M354I, has been shown to be essentially identical to wt-PAI-1 with respect to inhibitory activity, as well as binding to Vn, heparin, and low-density lipoprotein receptor-related protein but has a half-life nearly 75-fold longer than wt-PAI-1 at 37°C (145 vs. 2 h) (3, 19, 36); PAI-1K, constructed on the 141-1b background and containing an additional mutation of Gln123 to Lys, which has previously been shown to have a specific defect in Vn binding but retains full inhibitory activity (24); PAI-1R, constructed on the wt-PAI-1 background and containing a double mutation (Thr333 to Arg and Ala335 to Arg) which binds Vn normally but has no inhibitory activity toward any proteinase and which like 14-1b PAI-1 does not convert to the latent conformation (37); and PAI-1AK, containing a double mutation (Gln123 to Lys and Arg101 to Ala) constructed on the wt-PAI-1 background, has a complete defect in Vn binding but retains full inhibitory activity and has a short half-life similar to wt-PAI-1 (39). wt-PAI-1 was produced as an active form in Escherichia coli. The active form of PAI-1 was chromatographically separated from the latent form as described (21, 35).

Degradation of the matrices by mesangial cells.

Rat mesangial cell matrices were obtained as described previously (15). In brief, mesangial cells were seeded into six-well plates at 2 × 10⁵ cells/well. The cultures received ascorbic acid (25 μg/ml) the following day and daily thereafter. The medium was changed twice weekly. l-[^2,3-3H]proline was added to the culture medium at 0.5 μCi/ml on days 3, 7, and 10 after seeding. Two weeks after seeding of the cells, cultures were washed with PBS and the cells were lysed by the addition of 2 ml 2.5 mM NH₄OH, 0.1% Triton X-100 for 2 min. The matrices were then washed extensively with distilled water and kept covered with distilled water under sterile conditions at 4°C until further use.

Wells containing labeled matrices were washed three times with 2 ml of serum-free RPMI-1640 medium immediately before addition of mesangial cells in 15% FCS RPMI-1640 medium. Plates were then incubated for 24 h to allow the mesangial cells to attach to the matrices and to recover from the plating procedures. After 24 h, the medium was carefully removed. Cells were washed three times with 2 ml of serum-free RPMI-1640 medium (to remove proteolytic enzyme inhibitors potentially present in the serum) and incubated for 72 h in 2 ml of serum-free RPMI-1640 containing 0.2% lactalbumin hydrolysate (RPMI-LH) (22). Exogenously added agents were dissolved in RPMI-LH at the concentrations indicated below.

At the end of incubation, culture supernatant radioactivity was measured in a scintillation counter to quantitate digested matrix. The undigested matrix remaining in the culture dish was digested with 2 N NaOH, and radioactivity was measured. The sum of the supernatant and residual undigested matrix counts was 100%. The percentage of matrix degradation during the incubation time was the supernatant counts divided by the total counts × 100. Background values obtained with medium (RPMI-LH) in the absence of cells were subtracted from these values. A plasmin-specific chromogenic substrate, Chromozym PL (Roche Molecular Biochemicals, Indianapolis, IN) was used to measure plasmin activity (17). This substance is specifically cleaved by plasmin into a residual peptide and 4-nitraniline, which can be detected spectrophotometrically. Eighty microliters of cultured supernatant and 20 μl 3 mM Chromozyme PL (Diapharma Group, West Chester, OH) were added per well. The absorbance was measured at 405 nm three times over a 2-h interval. The increase in absorbance, corresponding to plasmin activity, was calculated. The standard linear curve was generated with serial dilutions of porcine plasmin. Results were expressed as 10⁻⁴ U/ml. The plasminogen dependence of mesangial cell ECM degradation in serum-free medium was assessed in experiments with and without added plasminogen (4 μg/ml).

Zymography.

To further determine plasmin activity in cultured supernatant, 30 μl of each supernatant was separated by a 4–16% Tris-glycine gel with blue-stained β-casein incorporated as a substrate for plasmin (Novex, San Diego, CA). The gels were incubated in the Novex zymogram renaturing buffer twice for 30 min at room temperature and then incubated at 37°C overnight in Novex zymogram developing buffer. The gel was photographed by a Bio-Rad GS-700 imaging densitometer (Bio-Rad Laboratories, Hercules, CA). Porcine plasmin was loaded as a control. Plasmin activity is easily characterized as clear bands against a dark blue background where plasmin had digested the substrate.

Study 2: Time Course of Disappearance of Recombinant PAI-1 Proteins from Nephritic Glomeruli

Animals.

Experiments in vivo were performed using male Sprague-Dawley rats (180–200 g) obtained from the SASCO colony of Charles River Laboratories (Wilmington, MA). Animal housing and care were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The animal experiments were approved by the Animal Care Committee of the University of Utah. Glomerulonephritis was induced by tail vein injection of 1.75 mg/kg of the monoclonal anti-thy-1 antibody OX-7 (NCCC; Biovest International, Minneapolis, MN). OX-7 binds to a thy-1-like epitope on the surface of MCs, causing complement-dependent cell lysis followed by fibrotic tissue repair. Normal control animals were injected with the same volume of PBS (17).

Experimental design.

Eight groups of two nephritic rats received an intravenous injection of 14-1b or PAI-1K (1 mg/kg body wt) 24 h after disease induction with anti-thy-1 antibody OX-7 (1.75 mg/kg body wt) as described (17). Groups were killed at eight time points from 10 min to 18 h. Renal cortical tissue was used for immunostaining of human PAI-1.

Immunofluorescent staining.

Indirect immunofluorescence was performed on 3-μm cryostat sections. Polyclonal goat anti-human PAI-1 antibody (1:100, American Diagnostica, Greenwich, CT), with no cross-reactivity with rat PAI-1 (17), was applied at 4°C overnight. FITC-conjugated donkey anti-goat immunoglobulin (at 1:200 dilution, Jackson ImmunoReseach Laboratories, West Grove, PA) was used as the secondary antibody at room temperature for 2 h. Control slides treated with PBS or nonimmunized normal serum instead of primary antibodies showed no staining.

Study 3. In Vivo Therapeutic Efficacy of Mutant PAI-1 Proteins

Experimental design.

Eight male Sprague-Dawley rats (160–180 g) were assigned to each of the following five groups: normal controls, disease controls, and diseased animals treated with PAI-1R or 14-1b or PAI-1K. PAI-1R, 14-1b, and PAI-1K were blindly administrated intravenously by tail-vein injection twice daily from day 1 (d1) after disease induction to d5 at a dose of 1 mg/kg body wt (17). Control rats received an equal volume of PBS. Animals were placed in metabolic cages for 24-h urine collection from d5 to d6 and were killed at d6. Twenty-four-hour urinary protein excretion was measured by the Bradford method (Bio-Rad Protein Assay, Bio-Rad Laboratories). The code was broken at the end of the analyses.

Rats were killed and blood was collected into 5% EDTA on d6. Plasma was separated by centrifugation, snap-frozen. and stored at −70°C. The kidneys were perfused with 30 ml of cold PBS and harvested. Renal cortical tissue was snap-frozen for immunofluorescent staining or fixed on 10% neutral-buffered formalin for periodic acid Schiff (PAS) staining. Glomeruli isolated by sieving were resuspended at 5,000 glomeruli·ml⁻¹·well⁻¹ in RPMI supplemented with 0.1 U/ml insulin, 100 U/ml penicillin, 100 μg/ml streptomycin, and 25 mmol/l HEPES buffer. After a 48-h incubation at 37°C/5% CO₂, the supernatant was harvested and stored at −70°C until analysis of glomerular production of transforming growth factor (TGF)-β1 by ELISA as described previously (17).

Plasma levels of active rat PAI-1.

Active rat PAI-1 was measured using a commercially available kit (Molecular Innovations, Southfield, MI).

Light microscopy.

All microscopic examinations were performed in a blinded fashion on 3-μm sections of paraffin-embedded tissues stained with PAS. Glomerular matrix expansion was evaluated in 30 glomeruli from each rat, where the percentage of mesangial matrix occupying each glomerulus was rated on a 0–4 scale where 0 = 0%, 1 = 25%, 2 = 50%, 3 = 75%, or 4 = 100% as described previously (16).

Immunofluorescent staining for matrix proteins and macrophages.

Monoclonal mouse FN-EDA⁺ (Harlan Sera-Lab, Belton, UK), goat anti-human type I collagen, and goat anti-human type III collagen (Southern Biotechnology Associates, Birmingham, AL) were used as the primary antibodies for detection of ECM components in glomeruli at d6. FITC-conjugated rat F(ab′)₂ anti-mouse IgG (H+l; Jackson ImmunoResearch Laboratories) and FITC-conjugated rabbit anti-goat IgG (Dako) were used as the secondary antibodies. For immunostaining of fibrinogen/fibrin, FITC-conjugated rabbit anti-human fibrinogen/fibrin (Dako) was used directly. For the determination of monocyte/macrophage or leukocyte infiltration into glomeruli, FITC-conjugated mouse anti-rat ED-1 antibody (Serotec, Oxford, UK) and FITC-conjugated mouse anti-rat CD45 antibodies were used. Intraglomerular deposition of these ECM components was quantified by scoring 20 randomly selected glomeruli/section on a 0–4 scale as described above. The number of monocyte/macrophage cells per glomerulus was counted in 20 glomeruli selected randomly per section.

RNA preparation and Northern hybridization.

Total RNA was extracted immediately from freshly isolated glomeruli by a guanidinium isothiocyanate method using TRIzol Reagent (GIBCO BRL, Gaithersburg, MD) according to the manufacturer's instructions. RNA from each group was pooled, and Northern analysis was performed as previously described (17). Three blots per probe were performed. Autoradiographic films were scanned on a Bio-Rad GS-700 imaging densitometer (Bio-Rad Laboratories). For quantitative densitometric measurements of Northern blots, all the signals were normalized compared with GAPDH levels used for equal loading.

Statistics and Calculation of Percent Reduction in Disease Severity

Data are expressed as means ± SE. The significance of differences in the measured values between groups was analyzed by the Student-Newman-Keuls multiple comparisons test. P < 0.05 was considered statistically significant. In vitro, triplicate wells were analyzed for each experiment, and each experiment was performed independently at least three times. In vivo, the disease-induced increase in a variable was defined as the mean value for the disease control group minus the mean value of the normal control group. The percent reduction in disease severity in a PAI-1R-treated group was calculated as follows: [1 − (PAI-1R-treated group mean − normal control group mean)/(disease control group mean − normal control group mean)] × 100.

RESULTS

Study 1: Effect of PAI-1 Mutants on ECM Degradation by Mesangial Cells In Vitro

Effect of Vn binding affinity of PAI-1 on ECM degradation.

To study ECM degradation under well-defined conditions, we developed an in vitro system utilizing mesangial cells cultured on extracellular matrices produced by mesangial cells themselves, which contain similar matrix components to those seen in vivo (15). Vn was also included in the ECM. Utilizing this system, we previously confirmed that ECM degradation by cultured rat mesangial cells is mediated by plasmin and addition of exogenous wt-PAI-1 in the presence of exogenous plasminogen reduced ECM degradation in a dose-dependent manner (15), whereas the AK mutant, with decreased Vn binding and a short half-life similar to wt-PAI-1, inhibits matrix degradation only ∼20% as much as wt-PAI-1 after 72-h incubation (Fig. 1). There was no significant change in the concentration of wt-PAI-1 and AK mutant in cultured supernant after 72-h incubation (data not shown).

Fig. 1.

Effects of wild-type (wt)-plasminogen activator I-1 (PAI-1) and PAI-1AK on ECM degradation. Addition of wt-PAI-1 at 1.5 × 10⁻⁷ M in the presence of exogenous plasminogen for 72 h reduced ECM degradation. The same amount of PAI-1AK mutant at 1.5 × 10⁻⁷ M, with decreased Vn binding and a short half-life similar to wt-PAI-1, inhibits ECM degradation only ∼20% as much as wt-PAI-1 after 72-h incubation. *P < 0.05 vs. control. #P < 0.05 vs. wt-PAI-1 alone treated for 72 h.

Effect of protease binding ability of PAI-1 on ECM degradation.

PAI-1K binds Vn poorly but retains wild-type inhibitory activity and has a half-life of ∼145 h in vitro (24). As shown in Fig. 2, this mutant, like fully functional wt-PAI-1, significantly inhibited plasmin activity and ECM degradation after 72-h incubation. However, addition of PAI-1R at the same concentration had no effect on PAI-1K's inhibition.

Fig. 2.

Effects of 14-1b or PAI-1K alone on ECM degradation and effects of PAI-1 mutant (PAI-1R) on the inhibition of 14-1b or PAI-1K. A: addition of 14-1b or PAI-1K alone at 1.5 × 10⁻⁷ M for 72 h produces reduced ECM degradation by 78 and 76%, respectively. When both PAI-1R and 14-1b or PAI-1K was added at 1.5 × 10⁻⁷ M for 72 h, the inhibition of ECM degradation seen with 14-1b or PAI-1K was not reversed by PAI-1R. *P < 0.001 vs. control. Consistently, plasmin activity in cultured supernatant after 72-h treatment was determined by a chromogenic substrate assay (B) and 4–16% zymogram (C). *P < 0.001 vs. control.

The 14-1b mutant is identical to wt-PAI-1 with respect to inhibitory activity, as well as binding to Vn but has the same long half-life as that of PAI-1K (3). Addition of 1.5 × 10⁻⁷ M 14-1b reduced plasmin activity and ECM degradation as well as wt-PAI-1 (Fig. 2). Similarly, addition of PAI-1R at 1.5 × 10⁻⁷ M for 72 h was unable to reverse inhibition of 14-1b (P > 0.05) (Fig. 2).

Study 2: Time Course of Disappearance of Recombinant PAI-1 Proteins from Nephritic Glomeruli

To determine whether the mechanisms of PAI-1's action identified above in vitro operate in vivo, we compared the effect of 14-1b, PAI-1K with PAI-IR in the anti-thy-1 nephritis model of renal fibrosis. We have reported previously that PAI-1R binds to nephritic glomeruli and remains there for at least 8 h (17). The pattern for injected 14-1b, as shown in Fig. 3A, was very similar to that seen for PAI-1R (17). In contrast, the mutant PAI-1K, with diminished Vn binding, was not deposited in glomeruli (Fig. 3B). Tubular staining suggests that PAI-1K is filtered by the kidney.

Fig. 3.

Time course of disappearance of injected 14-1b (A) and PAI-1K (B) from OX-7-induced nephritic glomeruli. The staining antibody (Ab) was specific for human PAI-1 and did not stain rat PAI-1. No staining was seen in normal and nephritic kidney without injection of PAI-1 mutants. Original magnification ×400.

Study 3: In Vivo Therapeutic Efficacy of Mutant PAI-1 Proteins

Effects of mutant PAI-1 proteins on plasma levels of active rat PAI-1.

Active rat PAI-1 levels were measured in rat plasma to ensure that treatment with PAI-1 mutants did not alter plasma levels of active PAI-1, an effect that might have negative in vivo consequences. The data shown in Table 1 indicate that there are no significant differences in active rat PAI-1 in the plasma in response to injection of PAI-1 mutant molecules.

Table 1.

Effect of mutant PAI-1s on plasma levels of active rat PAI-1

Group	NC	DC	PAI-1R	14-1b	PAI-1K
Active PAI-1, ng/ml	3.32±0.93	3.2±0.99	4.26±1.40	3.30±0.90	3.97±0.89
P value vs. NC		0.44	0.11	0.72	0.60

Values are means ± SE, except for P values. PAI-1R, 14-1b, and PAI-1K represent PAI-1R-, 141b-, or PAI-1K-treated nephritic rats, respectively. NC, normal control; DC, disease control.

Effects of mutant PAI-1 proteins on urinary protein excretion in anti-thy-1 nephritis.

Twenty-four-hour urinary protein excretion was measured from d5 to d6 (Fig. 4). Disease induced a significant increase in urinary protein excretion compared with normal rats, which was significantly reduced only by PAI-1R treatment. Both 14-1b and PAI-1K had no effect on urinary protein excretion in nephritic rats.

Fig. 4.

Effects of PAI-1 mutants on urinary protein excretion in anti-thy-1 nephritis. PAI-1R, 14-1b and PAI-1K represent PAI-1R-, 14-1b-, or PAI-1K-treated nephritic rats, respectively. *P < 0.05 vs. normal control (NC). #P < 0.05 vs. disease control (DC).

PAS staining.

Representative glomeruli stained with PAS are shown in Fig. 5. The glomeruli from the disease control rats showed marked accumulation of ECM expressed as PAS-positive material at d6 compared with glomeruli from normal control rats (Fig. 5). Among the three PAI-1 mutants, only treatment of nephritic rats with the PAI-1R resulted in significantly less mesangial ECM accumulation in glomeruli (Fig. 5). Figure 5 also includes a graphical representation of the PAS matrix score for each group. The PAS score increased from 0.58 ± 0.02 in normal control rats to 2.70 ± 0.13 in disease control rats as a result of nephritis. PAI-1R treatment significantly decreased matrix accumulation (P < 0.02) from 2.70 ± 0.13 in the disease control group to 1.85 ± 0.22. This is a 39% reduction in the disease-induced increase in the PAS staining score. 14-1b and PAI-1K did not either decrease or further increase matrix accumulation over that seen in diseased glomeruli.

Fig. 5.

Glomerular histology. Representative photomicrographs of glomeruli from NC, DC, PAI-1R-, 14-1b-, and PAI-1K-treated nephritic rats at day 6 (d6) are shown. Graphic representation of PAS staining scores is also shown (bottom right). *P < 0.05 vs. NC. #P < 0.05 vs. DC.

Immunofluorescent staining.

The results of the semiquantitative analysis of immunofluorescent staining for ECM proteins are shown in Fig. 6. The specific staining for pathological glomerular proteins FN-EDA+, collagen I, collagen III, and fibrinogen/fibrin was dramatically increased in disease control animals respectively compared with normal control animals (Fig. 6, A–D). Only PAI-1R reduced the accumulation of specific matrix components while 14-1b and PAI-1K had no effect.

Fig. 6.

Immunofluorescent staining scores of glomerular ECM proteins at d6 (A–D). PAI-1R, 14-1b, and PAI-1K represent PAI-1R-, 14-1b-, or PAI-1K-treated nephritic rats, respectively.*P < 0.05 vs. NC. #P < 0.05 vs. DC.

Effects of mutant PAI-1 proteins on inflammatory cell infiltration in anti-thy-1 nephritis.

The number of monocytes/macrophages and leucocytes was determined in kidney sections from all rats in each group (Fig. 7). Nephritic glomeruli from disease control rats had higher numbers of monocytes/macrophages and leukocytes than did glomeruli from normal control rats. Among the three treatments, only PAI-1R significantly reduced infiltration of monocytes/macrophages and leukocytes into the glomerulus. PAI-1R decreased the disease-induced infiltration of these inflammatory cells by 66 and 69%, respectively.

Fig. 7.

Number of monocytes/macrophages (A) and leucocytes (B) infiltrating nephritic glomeruli at d6. PAI-1R, 14-1b, and PAI-1K represent PAI-1R-, 14-1b-, or PAI-1K-treated nephritic rats, respectively. *P < 0.05 vs. NC. #P < 0.05 vs. DC.

Effects of mutant PAI-1 proteins on TGF-β1 production in glomeruli.

Nephritic glomeruli from disease control rats produced significantly higher TGF-β1 levels than did normal glomeruli. Among the three treatments, only PAI-1R significantly reduced glomerular TGF-β1 content by 42% compared with disease control (Fig. 8).

Fig. 8.

Effects of PAI-1 mutants on glomerular production of transforming growth factor (TGF)-β1 at d6. PAI-1R, 14-1b, and PAI-1K represent PAI-1R-,14-1b-, or PAI-1K-treated nephritic rats, respectively. *P < 0.05 vs. NC. #P < 0.05 vs. DC.

Effect of mutant PAI-1 proteins on glomerular mRNA levels of TGF-β1, PAI-1, FN-EDA+, and type I collagen in anti-thy-1 nephritis.

As shown in Fig. 9, A and B, glomerular mRNA analysis revealed the dramatic increases in disease control rats compared with normal control rats. Among the three treatments, only PAI-1R treatment significantly reduced the levels of FN-EDA+ and type I collagen mRNAs by 11 (P < 0.05) and 47% (P < 0.05), respectively, but did not affect the overexpression of TGF-β1 and PAI-1 mRNAs significantly. Interestingly, the treatment with 14-1b did not affect the disease-induced overexpression of TGF-β1, PAI-1, and FN-EDA+ but further increased the overexpression of type I collagen mRNA by 54% (P < 0.05) compared with the disease group. The treatment with PAI-1K significantly further increased the disease-induced overexpression of type I collagen and FN-EDA+ mRNAs by 101 (P < 0.05) and 90% (P < 0.05), respectively. In addition, PAI-1K treatment also slightly further increased the diseased-induced overexpression of TGF-β1 and PAI-1 mRNAs.

Fig. 9.

Effects of PAI-1 mutants on glomerular mRNA expression in anti-thy-1 nephritis at d6. A: representative Northern blot. B: relative levels of glomerular mRNA expression of TGF-β1, PAI-1, FN-EDA+, and type I collagen shown graphically. PAI-1R, 14-1b, and PAI-1K represent PAI-1R-, 14-1b-, or PAI-1K-treated nephritic rats, respectively.*P < 0.05 vs. NC. #P < 0.05 vs. DC.

DISCUSSION

The present study further demonstrates that Vn binding stabilizes wt-PAI-1 in the active conformation maintaining wt-PAI-1's inhibition, since wt-PAI-1 incubated with uPA at 37°C lost the ability to bind uPA after 30 min in the absence of Vn but kept binding to uPA at least for 3 h in the presence of Vn (seen in Supplemental Fig. 1; all supplementary material for this article is available on the journal web site); wt-PAI-1 added in the culture system containing Vn inhibited matrix degradation for at least 72 h (15) (Fig. 1). In contrast, decreased Vn binding generated by mutagenesis as seen with the AK mutant in Fig. 1 decreases PAI-1's inhibition of matrix degradation. These results are consistent with the in vivo report that animals deficient in Vn had reduced plasma PAI-1 levels following endotoxin administration (41). Effective Vn binding appears to be a key factor in keeping PAI-1 in the antiproteolytic form. Interfering with PAI-1:Vn binding may diminish the antiprotease activity of PAI-1. Indeed, when both PAI-1R and wt-PAI-1 were added in vitro, the inhibition of ECM degradation seen with wt-PAI-1 was successfully reversed by PAI-1R, although addition of PAI-1R alone had no effect on matrix degradation (15). These data suggest that PAI-1R successfully competes with wt-PAI-1 for Vn binding sites to disable wt-PAI-1's antiprotease activity and Vn binding is essential to the therapeutic action of PAI-1R.

As a stable, active form of PAI-1 generated by mutagenesis, both PAI-1K and 14-1b dramatically inhibited plasmin activity and matrix degradation in cultured mesangial cells, and PAI-1R was not able to override the actions of these mutants. These results suggest that PAI-1s' antiproteinase activity is critical for the inhibition of ECM degradation. In other words, maintaining an active conformation of PAI-1, whether by Vn binding or mutation to a stable form like 14-1b, as well as maintaining the ability to bind proteases like PAI-1K are important for this inhibition.

Since 14-b inhibited matrix degradation and PAI-1R did not, since PAI-1R restored plasmin activity and matrix degradation inhibited by wt-PAI-1 and 14-1b did not (Fig. 2) and since these two molecules differ only in their ability to bind proteases, the stable conformation and the ability to bind Vn but lack of protease binding should be essential to PAI-1R's action to enhance plasmin-dependent matrix degradation.

The mechanisms underlying enhancing matrix degradation of PAI-1R identified in vitro are further confirmed in vivo in the anti-thy-1 nephritis. Injected PAI-1R significantly and repeatedly reduced pathological ECM accumulation through targeting Vn in nephritic glomeruli. That PAI-1R treatment enhances glomerular plasmin activity is suggested by the fact that it reduced fibrin deposits (Fig. 5D), a finding consistent with our previous report that the action of PAI-1R to reduce ECM accumulation was reversed when both PAI-1R and the inhibitor of plasmin formation, tranexamic acid, were given (15).

In vivo, PAI-1K never reached the fibrotic glomerulus because it did not bind Vn so it could not act there. It also could not affect the plasma PAI-1 level since it was filtered quickly by the kidney (Fig. 3B). These data suggest that Vn binding targets PAI-1R to fibrotic kidneys and that a mutant with diminished Vn binding, regardless of protease binding, would not be a therapeutic target.

It may have been expected that the 14-1b mutants would worsen disease by binding to nephritic glomeruli and enhancing the contributing protease-inhibitory activity of endogenous, native PAI-1, but they had no effect on fibrin deposition or matrix accumulation (Fig. 5). This may be because the large disease-induced increases in endogenous PAI-1 maximally inhibit degradation. Interestingly, 14-1b treatment further increased the overexpression of glomerular type I collagen mRNA in nephritic rats. PAI-1K also had a similar effect on glomerular protein mRNA expression but not matrix accumulation in this disease model. The details of this mechanism remain unclear. However, PAI-1's ability to stimulate cell infiltration or the recruitment of macrophages and myofibroblasts and signal directly to regulate TGF-β expression (7, 18) provides a possible mechanistic path to the mRNA overexpression of matrix proteins, independently of plasmin generation and action.

It has been thought that PAI-1 regulates cell migration in the ECM by blocking the interaction between integrins and the RGD integrin-binding site in Vn (6, 20, 35). Therefore, independently of its antiproteolytic activity, PAI-1 may block cell adhesion and migration for the time it is bound to Vn. As a preliminary step to investigate this possibility, the number of monocytes/macrophages and leucocytes was determined in kidney sections from all rats in each group (Fig. 6). Nephritic glomeruli from disease-injected control rats had higher numbers of monocytes/macrophages and leukocytes than did glomeruli from normal control rats. Although there is great variability in both the extent and characteristics of these infiltrates over the spectrum of renal diseases and considerable controversy concerning the contribution of these cells to disease, it is generally accepted that macrophages or leukocytes, in particular, contribute to disease by release of numerous growth factors (9, 10). Fewer inflammatory cells should reduce release of these factors. Repeatedly, PAI-1R significantly reduced infiltration of monocytes/macrophages and leukocytes into the glomerulus, and glomerular TGF-β1 content was subsequently reduced by 42% compared with disease control glomeruli (Fig. 7). The mechanism responsible for the PAI-1R effect on cell infiltration, however, is unclear at this time. Arguing against the hypothesis that PAI-1R inhibits Vn-mediated cell migration is the fact that 14-1b, which did not decrease cell infiltration (Fig. 6), had similar kidney-binding half-lives (Fig. 3), suggesting they bound to glomerular Vn for about the same amount of time. Thus PAI-1R may affect cell migration through a mechanism not involving RGD sites on Vn. Previous data from our laboratory indicate that part of the PAI-1R-induced decrease in inflammatory cell migration is plasmin dependent and part is plasmin independent (15). The increased glomerular fibrin seen during disease may act as a chemoattractant for inflammatory cells (40). Degradation of this fibrin through PAI-1R-induced increases in plasmin would be expected to decrease infiltration through decreased concentrations of chemoattractants. Cleary, further work is necessary to understand the mechanisms by which PAI-1R reduces cell infiltration.

In summary, both in vitro and in vivo data suggest that effective Vn binding and subsequent stabilizing the antiprotease activity are critical for the inhibitory proteolytic action of wt-PAI-1. Effective Vn binding but lack of protease inhibition are essential to the ability of PAI-1R mutants to remain bound to Vn at the site of injury, thereby disabling wt-PAI-1's activity, which is very likely the key to its therapeutic efficacy. However, it deserves further investigation whether the therapeutic potential of PAI-1R mutants is better than that of small molecules or neutralizing antibodies that directly block the inhibitory activity of wt-PAI-1 without Vn targeting.

GRANTS

The work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-60508 (N. A. Noble), DK-49374 (W. A. Border), and DK-43609 (W. A. Border).