Plasminogen Activator Inhibitor-1 Is a Significant Determinant of Renal Injury in Experimental Crescentic Glomerulonephritis

Home , Fibrin

J Am Soc Nephrol 14: 1487–1495, 2003 Plasminogen Activator Inhibitor-1 Is a Significant Determinant of Renal Injury in Experimental Crescentic Glomerulonephritis

A. RICHARD KITCHING,*† YAO Z. KONG,* XIAO RU HUANG,* PIERS DAVENPORT,* KRISTY L. EDGTTON,* PETER CARMELIET,† STEPHEN R. HOLDSWORTH,* and PETER G. TIPPING* *Centre for Inflammatory Diseases, Monash University, Department of Medicine, Monash Medical Centre, Clayton, Victoria, Australia; and †Centre for Transgene Technology and Gene Therapy, Flanders Interuniversity Institute for Biotechnology, Leuven, Belgium.

Abstract. Crescentic glomerulonephritis is characterized by press PAI-1 (PAI-1 tg). Compared with strain-matched genet- glomerular fibrin deposition, and experimental crescentic glo- ically normal animals, PAI-1 Ϫ/Ϫ mice with glomerulonephri- merulonephritis has been shown to be fibrin-dependent. Net tis developed fewer glomerular crescents, less glomerular fibrin deposition is a balance between activation of the coag- fibrin deposition, fewer infiltrating leukocytes, and less renal ulation system causing glomerular fibrin deposition and fibrin collagen accumulation at day 14 of disease. The reduction in removal by the plasminogen-plasmin (fibrinolytic) system. disease persisted at day 28, when injury had become more Plasminogen activator inhibitor-1 (PAI-1) inhibits fibrinolysis established. In contrast, mice overexpressing the PAI-1 gene by inhibiting plasminogen activators and has effects on leuko- (PAI-1 tg), that have basal plasma and renal PAI-1 levels cyte recruitment and matrix deposition. To test the hypothesis several times, normal developed increased glomerular crescent that the presence of PAI-1 and its levels were a determinant of formation, more glomerular fibrin deposition, increased num- injury in crescentic glomerulonephritis, accelerated anti-glo- bers of infiltrating leukocytes, and more renal collagen at both merular basement membrane glomerulonephritis was induced time points. These studies demonstrate that PAI-1 is a deter- in mice genetically deficient in PAI-1 (PAI-1 Ϫ/Ϫ), PAI-1 minant of glomerular fibrin deposition and renal injury in heterozygotes (PAI-1 ϩ/Ϫ), and mice engineered to overex- crescentic glomerulonephritis.

Crescentic glomerulonephritis (GN) is characterized by glo- by inhibiting plasmin generation and the direct effects of uPA. merular fibrin deposition (1,2). Glomerular fibrin deposition is PAI-1 has other effects on cell migration and leukocytes that a feature of rapidly progressive GN in both humans and in seem to favor recruitment of leukocytes to inflammatory le- experimental models (3,4) In experimental systems, this lesion sions (11). There is some evidence, including studies of in vivo is fibrin-dependent (5,6). Net glomerular fibrin deposition is a fibrotic interstitial renal injury (14), that PAI-1 has a pathoge- balance between activation of the coagulation system and netic role in renal disease. Recent reviews have highlighted fibrin deposition and fibrin removal by the plasminogen-plas- both the pathogenetic potential for PAI-1 in renal disease and min system (7,8). Plasminogen activator inhibitor-1 (PAI-1) is its potential role as a therapeutic target (10,11). a serine protease inhibitor that inhibits the activity of both Functional studies of the role of fibrin in crescentic GN have tissue type plasminogen activator (tPA - the predominant glo- shown that this fibrin deposition is an important mediator of merular plasminogen activator (9)) and urokinase type PA injury (4–6,15). The net amount of fibrin accumulation in (uPA) (reviewed in references 10 and 11). PAI-1 is upregulated glomeruli results from pathogenetic procoagulant effects early in the course of experimental immune renal injury driven by tissue factor (7,16) and the protective effects of the (12,13). Inhibition of tPA and uPA by PAI-1 leads to multiple plasminogen plasmin system (8,12). Studies in mice deficient events that favor deposition of fibrin and other matrix proteins, in components of this system have demonstrated a protective role for endogenous plasmin (generated by the conversion of plasminogen to plasmin by tPA) in experimental crescentic GN Received October 2, 2002. Accepted February 15, 2003. (8). Increased glomerular PAI-1 expression has been found in Correspondence to Dr. Peter G. Tipping, Monash University Department of both human and experimental crescentic GN (12,17–19). Medicine, Monash Medical Centre, 246 Clayton Road Clayton, Victoria 3168, Australia. Phone: 61-3-9594-5547; Fax: 61-3-9594-6495; E-mail: Given these findings, it is logical that molecules that inhibit [email protected] plasmin generation, such as PAI-1, would be pathogenetic in 1046-6673/1406-1487 crescentic GN. However, this hypothesis has not been tested in Journal of the American Society of Nephrology vivo. Copyright © 2003 by the American Society of Nephrology These studies address the hypothesis that in fibrin-associated DOI: 10.1097/01.ASN.0000065550.13931.00 immune glomerular injury, PAI-1 is an important determinant 1488 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 1487–1495, 2003

of injury, crescent formation, and matrix accumulation. They for the wild-type PAI-1 gene (22) and by detecting the presence of the use mice that have been genetically manipulated to be PAI-1– neomycin resistance gene as a marker for the mutant allele (20). The deficient (PAI-1 Ϫ/Ϫ mice) or PAI-1 overexpressing (PAI-1 tg absence of the PAI-1 was confirmed by measurement of PAI-1 protein mice) in which experimental crescentic GN, a fibrin associated in plasma via ELISA using monoclonal antibodies generated by form of immune glomerular injury, had been induced. In ad- immunizing genetically deficient mice as described previously (24). Results are expressed as ng of PAI-1 protein per ml of plasma. For dition to studying disease at 14 d, experimental crescentic GN measurement of PAI-1 in renal tissue by ELISA, tissue from mice was was studied at 28 d to determine whether any effect of PAI-1 homogenized and extracted as described previously by Lijnen et al. on fibrin/matrix deposition and injury could be overcome by (25) and results expressed as pg of PAI-1 per mg of wet weight. other inhibitors of plasmin.

Materials and Methods Assessment of Glomerular Crescent Formation and Experimental Design Glomerular Fibrin Deposition Mice with a genetic deletion of PAI-1 (20,21) on a C57BL/6 ϫ Kidney tissue was fixed in Bouin fixative and embedded in paraf- ␮ 129/SvJ background, backcrossed once onto a C57BL/6 background fin, and 3- m tissue sections were cut and stained with periodic (75% C57BL/6, 25% 129/SvJ) were bred in a specific pathogen-free acid-Schiff (PAS). A glomerulus was considered to exhibit crescent facility (Monash Medical Center, Melbourne, Australia). Mice of an formation if two or more layers of cells were observed in Bowman’s identical genetic background (i.e., 75% C57BL/6, 25% 129/SvJ) bred space. A minimum of 50 glomeruli was assessed to determine the from the backcross were used as controls. PAI-1 ϩ/Ϫ mice were crescent score for each animal. For detection of fibrin in glomeruli, generated by interbreeding PAI-1 Ϫ/Ϫ and PAI-1 ϩ/ϩ mice. PAI-1 tissue was embedded in Optimal Cutting Temperature Compound, Ϫ transgenic mice on a pure C57BL/6 background expressing a murine frozen in liquid nitrogen, and stored at 70°C. Immunofluorescence ␮ PAI-1 minigene under a CMV promoter were generously provided by was performed on 4- m cryostat cut tissue. Glomerular fibrin depo- Dr. D. Ginsburg (Howard Hughes Medical Institute, University of sition was assessed on a minimum of 30 glomeruli per mouse using Michigan Medical Center, Ann Arbor, Michigan). (22) PAI-1 tg mice FITC-goat anti-rabbit fibrin/fibrinogen serum (Nordic Immunological were bred in a specific pathogen-free facility (Monash Medical Cen- Laboratories, Berks, UK), which cross-reacts with mouse fibrin at a ϩ ter). Genetically normal pure C57BL/6 mice were used as controls. dilution of 1:100 and scored semiquantitatively (0 to 3 ) using two Male mice aged 8 to 12 wk were used for experiments. Anti-mouse different scales. First, the proportion of glomeruli staining positive for glomerular basement membrane (GBM) globulin was prepared as fibrin in each mouse was assessed. Second, the overall intensity of described previously (23) Mice were sensitized by subcutaneous fluorescence present in each mouse was scored. injection of a total of 2 mg of sheep globulin in 100 ␮l of Freund complete adjuvant in divided doses in each flank 10 d before chal- Glomerular Leukocyte Accumulation lenge with 5 mg of sheep anti-mouse GBM globulin. Renal injury was Kidney tissue was fixed in periodate lysine paraformaldehyde studied at two time points: 14 d and 28 d after challenge with sheep (PLP) for 4 h, washed in 7% sucrose solution, and then frozen in anti-mouse GBM globulin. Results are expressed as the mean Ϯ SEM. liquid nitrogen. Tissue sections (6 ␮m) were stained to demonstrate The significance of differences between groups was determined by CD3ϩ cells and macrophages using a three-layer immunoperoxidase ANOVA, followed by Bonferroni Multiple Comparison Test for technique, as described previously (26,27). The primary monoclonal paired comparisons (GraphPad Prism; GraphPad Software Inc., San antibodies were KT3, anti-mouse CD3, which recognizes T cells Diego, CA). For studies involving PAI-1 tg mice (two groups only) (American Type Culture Collection [ATCC], Manassas, VA), and the unpaired t test was used. M1/70, anti-CD11b, which recognizes macrophages and neutrophils For studies using mice genetically deficient in PAI-1, the following (ATCC). A minimum of 20 glomeruli were assessed per animal, and groups of mice were studied: results were expressed as cells per glomerular cross section (c/gcs). • PAI-1 ϩ/ϩ mice: baseline (n ϭ 6), GN day 14 (n ϭ 6), GN day 28 (n ϭ 6) • PAI-1 ϩ/Ϫ mice: baseline (n ϭ 6), GN day 14 (n ϭ 6), GN day 28 Assessment of Proteinuria and Serum Creatinine (n ϭ 6) Urinary protein concentrations were determined by the Bradford • PAI-1 Ϫ/Ϫ mice: baseline (n ϭ 7), GN day 14 (n ϭ 7), GN day 28 method on timed urine collections. Urine was collected for 24 h at (n ϭ 7) baseline and over the last 24 h of an experiment. Serum creatinine concentrations at the completion of experiments (day 14 or day 28) For studies using mice genetically engineered to overexpress were measured by the alkaline picric acid method using an PAI-1, the following groups of mice were studied: autoanalyser. • C57BL/6 mice: baseline (n ϭ 8), GN day 14 (n ϭ 8), GN day 28 (n ϭ 8) Renal Collagen and Matrix Accumulation • ϭ ϭ PAI-1 tg mice (C57BL/6): baseline (n 6), GN day 14 (n 6), Renal collagen content was assessed by determining total hy- ϭ GN day 28 (n 6) droxyproline according to the method of Bergam and Loxley (28). Samples were hydrolyzed in 6 N HCL by incubation at 110°C over- Assessment of PAI-1 Gentoype and Plasma and Renal night. The hydrolysate was neutralized with 2.5 M NaOH. Hydroly- PAI-1 Levels sates in isopropanol were oxidized by chloramine T then mixed with PCR based protocols were used to determine the genotype of PAI-1 p-dimethylsaminonbenzaldehyde (25 min, 60°C), and the absorbance ϩ/ϩ, PAI-1 ϩ/Ϫ , PAI-1 Ϫ/Ϫ, and PAI-1 tg mice. The PAI-1 was measured at 558 nm. Total collagen was calculated using the transgene was detected as described previously (22). PAI-1 Ϫ/Ϫ, assumption that collagen contains 12.7% hydroxyproline by weight. PAI-1 ϩ/Ϫ, and PAI-1 ϩ/ϩ mice were genotyped by using primers Results were expressed as ␮g/mg kidney wet weight. J Am Soc Nephrol 14: 1487–1495, 2003 PAI-1 in Crescentic GN 1489

Results PAI-1 Protein Levels are Undetectable in PAI-1 Ϫ/Ϫ Mice and Increased in PAI-1 tg Mice Plasma PAI-1 levels in normal (without GN) PAI-1 ϩ/ϩ mice (C57BL/6 ϫ 129Sv/J background) were 9.7 Ϯ 1.5 ng/ml (Table 1). PAI-1 protein was undetectable in all PAI-1 Ϫ/Ϫ mice. PAI-1 ϩ/Ϫ mice had intermediate levels of PAI-1 protein (reduced to 30 to 40% of normal). Normal C57BL/6 mice had PAI-1 levels falling between those of PAI-1 ϩ/Ϫ and PAI-1 ϩ/ϩ mice. Mice engineered to genetically overexpress PAI-1 (PAI-1 tg mice) produced significantly more PAI-1 than all other mice (72 Ϯ 8.1 ng/ml). Figure 1. Glomerular injury and fibrin deposition in plasminogen acti- Glomerulonephritis is Less Severe in Mice Lacking vator inhibitor-1 (PAI-1) ϩ/ϩ and PAI-1 Ϫ/Ϫ mice with glomerular PAI-1 nephritis (GN) 14 d after the initiation of injury. (A) Glomerulus from a Endogenous PAI-1 Mediates Glomerular Crescent For- PAI-1 ϩ/ϩ mouse without GN. (B) PAI-1 ϩ/ϩ mice developed prolif- mation and Fibrin Deposition. Genetically normal PAI-1 erative crescentic glomerular injury (mean % glomeruli affected, see ϩ/ϩ mice developed proliferative GN with glomerular cres- Figure 2A). (C) PAI-1 Ϫ/Ϫ mice developed proliferative GN that is ϩ ϩ cent formation and fibrin deposition 14 d after challenge with significantly less severe than PAI-1 / mice with only occasional sheep anti-mouse GBM globulin (Figures 1, B and E; Figure crescent formation. (D) Negative immunofluorescence for fibrin in a glomerulus from a PAI-1 ϩ/ϩ mouse without GN. (E) Significant fibrin 2). Glomerular crescent formation and fibrin deposition per- ϩ ϩ Ϫ Ϫ deposition in the glomerular tuft of a PAI-1 / mouse with GN. (F) sisted to day 28. PAI-1 / mice were significantly protected Reduced fibrin deposition in a glomerulus of a PAI-1 Ϫ/Ϫ mouse with from glomerular crescent formation at both the day 14 and day GN. (A through C, periodic acid-Schiff [PAS] stain, high power view; D 28 time points (Figures 1C and 2A). Renal PAI-1 protein levels through F, immunofluorescence, high power). progressively increased in PAI-1 ϩ/ϩ mice but remained undetectable in PAI-1 Ϫ/Ϫ mice (Table 2). A modest reduction in crescent formation was observed in PAI-1 ϩ/Ϫ mice at day 14, but not at day 28. Fibrin deposition was detected in most glomeruli of PAI-1 ϩ/ϩ mice, mainly in the glomerular tuft. However, fibrin deposition was absent in over half of glomeruli in PAI-1 Ϫ/Ϫ mice (Figures 1F and 2B). At the later time point (day 28), there had been no progressive increase in glomeruli affected in PAI-1 Ϫ/Ϫ mice. In addition to there being fewer glomeruli in which fibrin was detected, overall intensity of fluorescence was decreased in PAI-1–deficient mice (0 to 3ϩ, per mouse) (day 14: PAI-1 ϩ/ϩ 2.8 Ϯ 0.2, PAI-1 Ϫ/Ϫ 1.4 Ϯ 0.2; day 28: PAI-1 ϩ/ϩ 1.8 Ϯ 0.3, PAI-1 Ϫ/Ϫ mice all a score of 1).

Table 1. Plasma levels of PAI-1 protein measured by ELISA in mice without glomerulonephritis

a Plasma PAI-1 (ng/ml) Figure 2. Glomerular crescent formation and fibrin deposition in ϩ ϩ ϩ Ϫ Ϫ Ϫ Ϫ Ϫc PAI-1 / , PAI-1 / , and PAI-1 / mice with GN at 14 and PAI-1 / ND 28 d. Open bars represent PAI-1 ϩ/ϩ mice, gray bars represent PAI-1 ϩ Ϫ Ϯ b PAI-1 / 3.7 0.2 ϩ/Ϫ mice, and black bars represent PAI-1 Ϫ/Ϫ mice. Crescent PAI-1 ϩ/ϩ 9.7 Ϯ 1.5 formation (A) and fibrin deposition (B, expressed as the proportion of C57BL/6 4.3 Ϯ 0.2 glomeruli with immunofluorescent fibrin deposits present) are mark- C57BL/6 PAI-1 tg 72.7 Ϯ 8.1d edly reduced in mice that are genetically deficient in PAI-1. * P Ͻ 0.05 versus PAI-1 ϩ/ϩ mice, ** P Ͻ 0.001 versus either PAI-1 ϩ/ϩ a ϩ ϩ PAI-1, plasminogen activator inhibitor-1; / , wildtype mice; or ϩ/Ϫ mice at the same timepoint (ANOVA). ϩ/Ϫ, heterozyogous mice; Ϫ/Ϫ, knockout mice; ND not detected. b Results are expressed in ng of protein per ml of plasma, as the mean Ϯ SEM. c PAI-1 Ϫ/Ϫ, ϩ/Ϫ, and ϩ/ϩ mice had a mixed genetic background (C57BL/6 75%, 129Sv/J 25%). Leukocyte Recruitment Is Diminished in the Absence of d Significant at a level of P Ͻ 0.001 versus all other groups PAI-1. CD3ϩ T cells and macrophages were detected in (ANOVA, Bonferroni Multiple Comparison Test). glomeruli of genetically normal PAI-1 ϩ/ϩ mice (Figure 3). 1490 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 1487–1495, 2003

Table 2. Renal PAI-1 protein levels measured by ELISA in PAI-1 intact and deficient mice with GN

Baseline no GNa GN day 14 GN day 28

PAI-1 ϩ/ϩ 40.2 Ϯ 3.1b (8) 175.3 Ϯ 117.8 (7) 1000 Ϯ 173.2 (6) PAI-1 ϩ/Ϫ NP 166.2 Ϯ 81.0 (4) 189.4, 115.5 (2) PAI-1 Ϫ/Ϫ NP ND ND

a GN, glomerulonephritis; PAI-1, plasminogen activator inhibitor-1; ϩ/ϩ, wildtype mice; ϩ/Ϫ, heterozygous mice; Ϫ/Ϫ, knockout mice; NP, test not performed, ND, not detected. b Results are expressed in pg of PAI-1 per mg of renal tissue (wet weight), as the mean Ϯ SEM. Numbers in parentheses indicate number of mice tested.

Figure 3. Glomerular leukocyte accumulation in PAI-1 ϩ/ϩ, PAI-1 Figure 4. Functional renal injury in PAI-1ϩ/ϩ, PAI-1 ϩ/Ϫ, and ϩ/Ϫ, and PAI-1 Ϫ/Ϫ mice with GN at 14 and 28 d. Open bars PAI-1 Ϫ/Ϫ mice with GN at 14 and 28 d. Open bars represent PAI-1 represent PAI-1 ϩ/ϩ mice, gray bars represent PAI-1 ϩ/Ϫ mice, and ϩ/ϩ mice, gray bars represent PAI-1 ϩ/Ϫ mice, and black bars black bars represent PAI-1 Ϫ/Ϫ mice. T cells (A) and CD11bϩ cells represent PAI-1 Ϫ/Ϫ mice. (A) Proteinuria is reduced in PAI-1 Ϫ/Ϫ (B) are markedly reduced in glomeruli of mice that are genetically mice, though this reduction, compared with PAI-1 ϩ/ϩ mice, reached deficient in PAI-1. Dotted lines represent values for mice without GN. statistical significance only at 14 d. (B) There was a trend to decreased ** P Ͻ 0.001 versus either PAI-1 ϩ/ϩ or ϩ/Ϫ mice at the same time serum creatinine in PAI-1 Ϫ/Ϫ mice. The dotted line represents the point (ANOVA). mean urinary protein excretion (A) and serum creatinine (B) of mice without GN. There was no difference in these baseline values between groups. * P Ͻ 0.05 versus PAI-1 ϩ/ϩ mice and Ͻ 0.001 versus PAI-1 ϩ/Ϫ mice, ** P Ͻ 0.01 versus PAI-1 ϩ/Ϫ mice. *** P Ͻ 0.001 While leukocyte numbers were unchanged in PAI-1 heterozy- versus PAI-1 Ϫ/Ϫ and ϩ/ϩ mice. gotes, complete absence of PAI-1 resulted in a substantial reduction in both T cells and macrophages at both time points. Renal Collagen Content in GN in the Absence of Endog- Functional Indices of Renal Injury in the Absence of enous PAI-1. Genetic deletion of PAI-1 did not affect base- PAI-1. Baseline urinary protein and renal function (mea- line renal collagen levels (Table 3). By day 14 of disease, sured by serum creatinine) were similar in the three experi- PAI-1 ϩ/ϩ mice had developed a 100% increase in renal mental groups (urinary protein: 0.4 Ϯ 0.1 mg/d in all groups; collagen content, with levels remaining elevated at day 28. Ϫ Ϫ serum creatinine: PAI-1 ϩ/ϩ 17 Ϯ 0.4, PAI-1 ϩ/Ϫ 17 Ϯ 0.5, Renal collagen content in PAI-1 / mice was significantly ϩ ϩ PAI-1 Ϫ/Ϫ 19 Ϯ 0.4 ␮mol/L). PAI-1 ϩ/ϩ and PAI-1 ϩ/Ϫ less that that in PAI-1 / mice at day 14, but this decrease mice had developed significant proteinuria and renal impair- failed to reach statistical significance at day 28. Collagen ϩ Ϫ Ϫ Ϫ ment by day 14 of disease. Urinary protein excretion in PAI-1 accumulation in PAI-1 / mice was more than in PAI-1 / ϩ ϩ Ϫ/Ϫ mice with GN was significantly lower than the protein- mice but less than PAI-1 / mice at day 14 of GN. uria observed in either PAI-1 ϩ/ϩ or PAI-1 ϩ/Ϫ mice at day Glomerulonephritis is More Severe in Mice 14 and lower than proteinuria in PAI-1 ϩ/Ϫ mice at day 28 Overexpressing PAI-1 (Figure 4A). Serum creatinine levels in PAI-1 Ϫ/Ϫ mice at Overexpression of PAI-1 Enhances Glomerular Crescent either day 14 or day 28 were not different from baseline values Formation and Glomerular Fibrin Deposition. Normal (Figure 4B). C57BL/6 mice developed proliferative and crescentic GN 14 d J Am Soc Nephrol 14: 1487–1495, 2003 PAI-1 in Crescentic GN 1491

Table 3. Renal collagen content in PAI-1 intact and deficient mice with GN

Baseline GN GN no GNa day 14 day 28

PAI-1ϩ/ϩ 4.4 Ϯ 0.3b 9.0 Ϯ 1.0 7.7 Ϯ 0.3 PAI-1ϩ/Ϫ NP 7.8 Ϯ 0.8 10.0, 5.9c PAI-1Ϫ/Ϫ 4.6 Ϯ 0.3 6.4 Ϯ 0.4d 6.9 Ϯ 0.5

a GN, glomerulonephritis; PAI-1 plasminogen activator inhibitor- 1; ϩ/ϩ, wildtype mice; ϩ/Ϫ, heterozygous mice; Ϫ/Ϫ, knockout mice; NP, test not preformed. b Results are expressed in ␮g of collagen per mg of of renal tissue (wet weight), as the mean Ϯ SEM. c Results of samples for two animals. d P Ͻ 0.05 versus day 14 PAI-1 ϩ/ϩ values (ANOVA). after challenge with sheep anti-mouse GBM globulin, with glomerular fibrin deposition (Figures 5A, 5C, and 6). Overall, Figure 6. Glomerular crescent formation and fibrin deposition in disease in these mice was marginally less than disease in PAI-1 genetically normal C57BL/6 mice and PAI-1 overexpressing (PAI-1 ϩ/ϩ mice on a mixed 75% C57BL/6 ϫ 25% 129Sv/J back- tg) mice with GN at 14 and 28 d. Open bars represent genetically ground. Crescent formation and glomerular fibrin deposition normal C57BL/6 mice, and gray bars represent PAI-1 tg C57BL/6 persisted at day 28. Renal PAI-1 protein was increased in mice. Crescent formation (A) and fibrin deposition (B, expressed as PAI-1 tg mice, most evident at day 14 (Table 4). At day 14, the proportion of glomeruli with immunofluorescent fibrin deposits crescentic GN was more severe in PAI-1 tg mice compared present) are increased in mice that genetically overexpress PAI-1. * P Ͻ 0.005, ** P Ͻ 0.001 versus C57BL/6 mice at the same time point with normal C57BL/6 mice (Figures 5B, 5D, and 6). An (unpaired t test). increased proportion of glomeruli from PAI-1 tg mice was affected by crescent formation and more glomeruli exhibited fibrin deposition. This increased severity of disease persisted at day 28. Overall intensity of fibrin deposition was also increased in PAI-1 tg mice. C57BL/6 mice with GN had moderate fluorescence intensity for fibrin (0 to 3ϩ [mean Ϯ SEM]: day 14, 1.5 Ϯ 0.2; day 28, 0.6 Ϯ 0.1). Sections from all PAI-1 tg mice developed strong immunofluorescence for fibrin (all 3ϩ) at day 14, which was still moderate in intensity at day 28 (1.6 Ϯ 0.3). Leukocyte Recruitment Is Enhanced in PAI-1 tg Mice. Mice that overexpressed PAI-1 demonstrated increased recruitment of T cells and CD11bϩ leukocytes into glomeruli compared with genetically normal C57BL/6 mice with GN (Figure 7). This increased leukocyte influx in PAI-1 tg mice persisted at day 28 of disease. Functional Renal Injury Is Enhanced by PAI-1 Overex- pression. C57BL/6 mice with GN developed significant proteinuria, measured at both day 14 and day 28, with only minor renal impairment, consistent with the degree of glomerular crescent formation in these mice (Figure 8). As with glomerular crescent, formation, fibrin deposition, and leukocyte re- Figure 5. Glomerular injury and fibrin deposition in genetically normal cruitment, PAI-1 tg mice with GN developed increased urinary C57BL/6 mice and C57Bl/6 PAI-1 transgenic (PAI-1 tg) mice with GN protein excretion and renal impairment, evident at both day 14 14 d after the initiation of injury. (A) Significant glomerular injury with and day 28. some crescent formation was present in C57BL/6 mice with (mean % Renal Collagen Accumulation Is Accelerated in PAI-1 tg glomeruli affected see Figure 6A). (B) C57BL/6 PAI-1 tg mice developed Mice with GN. Normal C57BL/6 mice and PAI-1 tg mice accelerated crescentic GN compared with genetically normal mice with more frequent crescent formation. (C) Fibrin deposition in the glomerular had similar baseline renal collagen (Table 5), indicating that tuft of a C57BL/6 mouse with GN. (D) Increased glomerular fibrin transgenic overexpression of PAI-1 in the absence of patho- deposition in a glomerular from a PAI-1 tg mouse with GN. (A and B, logic stimuli does not lead to increased renal collagen content. PAS stain, high power view; C and D, immunofluorescence, high power, Compared with C57BL/6 mice, PAI-1 tg mice developed in- photographed at the same exposure). creased renal collagen accumulation at both 14 and 28 d. 1492 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 1487–1495, 2003

Table 4. Renal PAI-1 protein levels measured by ELISA in genetically normal C57BL/6 mice and in mice overexpressing PAI-1 (PAI-1 tg mice) with GN

Baseline no GNa GN day 14 GN day 28

C57BL/6 29.4 Ϯ 7.1 (8)b 150.5 Ϯ 75.2 (4) 382.4 Ϯ 93.5 (6) PAI-1 tg 961.4 Ϯ 117.8 (6)c 2480 Ϯ 677.3 (6)d 1127 Ϯ 527.5 (4)

a GN, glomerulonephritis; PAI-1 tg, plasminogen activator inhibitor-1 overexpressing transgenic mouse on a C57BL/6 background. bResults are expressed in pg of PAI per mg of renal tissue (wet weight), as the mean Ϯ SEM. Numbers in parentheses indicate number of mice tested. c P Ͻ 0.0001 versus baseline values for C57BL/6 mice (unpaired t test). d P ϭ 0.02 versus day 14 values for C57BL/6 mice (unpaired t test).

Figure 7. Glomerular leukocyte accumulation in genetically normal Figure 8. Functional renal injury in genetically normal C57BL/6 mice C57BL/6 mice and PAI-1 overexpressing (PAI-1 tg) mice with GN at and PAI-1 overexpressing (PAI-1 tg) C57BL/6 mice with GN at 14 14 and 28 d. Open bars represent genetically normal C57BL/6 mice, and 28 d. (A) Proteinuria is increased in PAI-1 tg mice. (B) In PAI-1 ϩ gray bars represent PAI-1 tg C57BL/6 mice. T cells (A) and CD11b tg mice, serum creatinine values were significantly increased com- cells (B) are present in increased numbers in glomeruli of mice that pared with normal values and C57BL/6 mice with GN. The dotted line genetically overexpress PAI-1. Dotted lines represent values for mice represents the mean urinary protein excretion (A) and serum creati- Ͻ Ͻ without GN. * P 0.005, ** P 0.001 versus C57BL/6 mice at the nine (B) of mice without GN. There was no difference in these same timepoint (unpaired t test). baseline values between groups. * P Ͻ 0.005, ** P Ͻ 0.001 versus C57BL/6 mice at the same time point (unpaired t test). Discussion PAI-1 has major potential pathogenetic effects in inflamma- crescentic GN. Findings of alterations in renal collagen content tory renal disease, potentially acting on several different me- in PAI-1 Ϫ/Ϫ and PAI-1 tg mice support recent studies impli- diator systems (10,11). By inhibiting the activity of plasmino- cating PAI-1 in renal fibrosis (14). Collectively, the current gen activators and therefore the generation of plasmin, it can studies suggest that PAI-1 is important in determining the inhibit fibrin removal, resulting in enhanced fibrin accumula- outcome of renal injury. They support the hypothesis that tion. In addition to its capacity to enhance net fibrin deposition, inhibition of PAI-1 is a potential strategy in the therapy of both PAI-1 has the capacity to contribute to the accumulation of rapidly progressive GN and other progressive renal diseases. collagens and other matrix proteins seen in progressive renal In the first section of this study, mice genetically deficient in disease. PAI-1 limits conversion of plasminogen to plasmin, PAI-1 were significantly protected from renal injury in exper- the active enzyme that may activate matrix metalloproteinases imental crescentic GN compared with genetically matched and act directly on matrix to limit collagen matrix accumula- PAI-1 ϩ/ϩ mice on a C57BL/6 ϫ 129Sv/J background. The tion. The increased PAI-1 levels observed in GN may result in relatively mild injury in PAI-1 Ϫ/Ϫ mice did not show a increased leukocyte recruitment in GN both directly (14) and significant “catch up” effect when anti-GBM GN was pro- via the chemotactic effects of fibrin (29). longed to 28 d. The protection afforded PAI-1 Ϫ/Ϫ mice was The alterations in glomerular crescent formation fibrin dep- most significant in crescent formation, fibrin deposition, leu- osition and leukocyte accumulation in PAI-1 Ϫ/Ϫ and PAI-1 tg kocyte recruitment, and the accumulation of collagen. A sec- mice demonstrate an important role for PAI-1 in experimental ondary finding in these studies was that PAI-1 ϩ/Ϫ mice, J Am Soc Nephrol 14: 1487–1495, 2003 PAI-1 in Crescentic GN 1493

Table 5. Renal collagen content in genetically normal showed decreased leukocyte recruitment (34). In contrast, in C57BL/6 mice and in mice overexpressing PAI-1 fibrin-mediated GN, plasminogen-deficient mice showed a (PAI-1 tg mice) with GN marked increase in fibrin deposition and an increase in glomerular leukocytes (8), while uPA Ϫ/Ϫ mice showed in- Baseline no GNa GN day 14 GN day 28 creased leukocytes without increased fibrin deposition (8). In C57BL/6 4.6 Ϯ 0.2b 7.1 Ϯ 0.4 7.7 Ϯ 0.7 obstructive uropathy, PAI-1 Ϫ/Ϫ mice have fewer leukocytes; PAI-1 tg 3.9 Ϯ 0.3 9.5 Ϯ 0.6c 10.1 Ϯ 0.7d in the same study, PAI-1 was chemotactic for macrophages ex vivo (14). The exact roles of fibrin, plasmin, and PAI-1 leuko- a GN, glomerulonephritis; PAI-1 tg, plasminogen activator cyte recruitment in the kidney and the mechanisms by which inhibitor-1 overexpressing transgenic mouse on a C57BL/6 they exert their effects remain to be fully elucidated. background. b Results are expressed in ␮g of collagen per mg of renal tissue PAI-1 is a major inhibitor of plasmin via its effects on tPA ␣ (wet weight), as the mean Ϯ SEM. and uPA. However, other inhibitors of plasmin such as 2– c P Ͻ 0.01 versus day 14 C57BL/6 values (unpaired t test). antiplasmin have potential roles in promoting net glomerular d P ϭ 0.04 versus day 28 C57BL/6 values (unpaired t test). ␣ fibrin deposition and injury. Studies in mice deficient in 2– antiplasmin do not support a prominent pathogenetic role for this molecule in immune glomerular injury (35). These (carrying one normal allele and one disrupted allele) showed a findings, together with the current and previous studies (8), modest degree of protection in only some parameters of dis- suggest that in immune glomerular injury, PAI-1, and tPA ease, more evident at day 14. However, no protection was (as well as plasminogen) are the critical components of the apparent at 28 d. These PAI-1 ϩ/Ϫ mice had baseline PAI-1 plasminogen–plasmin system that determine net glomerular plasma levels less than half of those of PAI-1 ϩ/ϩ mice. fibrin deposition. Assessment of functional plasma PAI-1 activity in PAI-1 ϩ/Ϫ The results of the current studies seem at first glance to be mice after endotoxin stimulation has shown significant in- discordant from those of Dewerchin et al. (36), who found that creases, but not to levels seen in PAI-1 ϩ/ϩ mice (20). In the renal fibrin deposition was not significantly reduced in the current studies, renal PAI-1 levels in PAI-1 ϩ/Ϫ mice were absence of PAI-1. However, in the model of acute LPS induced similar to those in PAI-1 ϩ/ϩ mice at day 14. The relative renal injury used in those studies, fibrin deposition was extra- contributions of plasma and tissue derived PAI-1 in immune glomerular. In the model used in our current studies, fibrin renal injury remain unclear. deposition is predominantly glomerular. Analysis of renal fi- The second section of these studies used mice that are brinolytic activity after endotoxin injection, as well as renal transgenic for PAI-1 and produce several times more PAI-1 tPA and uPA activity (by Dewerchin et al. (36)), showed than genetically normal C57BL/6 mice. Control mice in these increased renal fibrinolytic activity in PAI-1 Ϫ/Ϫ mice and studies also developed crescentic GN, though their degree of preservation of renal tPA activity in PAI-1 Ϫ/Ϫ mice (com- injury was less that the PAI-1ϩ/ϩ mice. There are many pared with PAI-1 ϩ/ϩ mice, when renal tPA activity fell to potential contributors to the variation in pattern and severity of undetectable levels). These results are consistent with PAI-1 injury in experimental GN. Several studies have shown that the having an important role in protective fibrinolytic activity in genetic background and the nature of the T cell response are murine “anti-GBM GN” as (1) tPA is the major PA in the important (30–32). For this reason, the current studies are glomerulus (9); (2) we have demonstrated tPA to be the more presented and considered as two separate studies with separate important protective PA in glomerular fibrin deposition in strain-matched control groups. However, it is interesting to murine crescentic GN (8); and (3) PAI-1 Ϫ/Ϫ mice have note that basal PAI-1 levels in C57BL/6 mice were less than preserved renal tPA activity and increased total renal fibrino- those in C57BL/6 ϫ 129Sv/J PAI-1 ϩ/ϩ mice (plasma and lytic activity (36). renal). C57BL/6 PAI-1 tg mice with increased renal PAI-1 The current studies do not address the relative importance of levels developed accelerated disease compared with C57BL/6 the potential sources of PAI-1 in crescentic GN. PAI-1 is mice when measured by a variety of parameters, including present in plasma and is produced by a variety of renal cells, crescent formation, fibrin deposition, glomerular leukocyte including mesangial cells (37,38), endothelial cells (38), and accumulation, proteinuria, renal impairment, and renal colla- tubular cells (39). It is likely that both tissue-derived and gen accumulation at both day 14 and day 28. plasma-derived PAI-1 are important in this disease, though it is This form of GN is fibrin-mediated, and fibrin is a major probable that tubular cell–derived PAI-1 is not particularly substrate of plasmin. Therefore many of the effects observed in important in the glomerular fibrin deposition observed in ex- this study are likely to be mediated by the effects of PAI-1 on perimental anti-GBM GN. net fibrin accumulation in the glomerulus. Fibrin is chemotac- There is evidence in human GN that expression of PAI-1 tic for leukocytes (29), and mice deficient in fibrinogen show may help determine disease severity. Lupus nephritis, WHO fewer leukocytes in glomeruli in this model (6). However, the classes III and IV, is more severe in humans homozygous for plasminogen-plasmin system has multiple potential roles in the PAI-1 5' gene promoter 4G polymorphism (4G/4G) (40), leukocyte accumulation in GN. Plasmin itself can be chemoat- which, compared with the 5G/5G phenotype, results in in- tractant for leukocytes (33) and in obstructive uropathy (pre- creased PAI-1 expression, particularly in response to inflam- sumably not mediated by fibrin) mice deficient in plasminogen matory stimuli. Our experimental data that disease severity is 1494 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 1487–1495, 2003 increased in PAI-1 tg mice and decreased in PAI-1 Ϫ/Ϫ mice 15. Thomson NM, Moran J, Simpson IJ, Peters DK: Defibrination (both strain-matched) support these human observational stud- with ancrod in nephrotoxic nephritis in rabbits. Kidney Int 10: ies. In summary, the current studies demonstrate that PAI-1 is 343–347, 1976 an important determinant of crescent formation, injury, fibrin 16. Tipping PG, Dowling JP, Holdsworth SR: Glomerular procoagu- deposition. and collagen accumulation in experimental cres- lant activity in human proliferative glomerulonephritis. J Clin centic GN. Invest 81: 119–125, 1988 17. Rondeau E, Mougenot B, Lacave R, Peraldi MN, Kruithof EK, Sraer JD: Plasminogen activator inhibitor 1 in renal fibrin de- Acknowledgments posits of human nephropathies. Clin Nephrol 33: 55–60, 1990 Dr. Ginsburg is thanked for providing the PAI-1 tg mice bred and 18. Moll S, Menoud PA, Fulpius T, Pastore Y, Takahashi S, Fossati used in these studies. The assistance of Ms. Janelle Sharkey and Ms. L, Vassalli JD, Sappino AP, Schifferli JA, Izui S: Induction of Alice Wright is acknowledged. These studies were supported by plasminogen activator inhibitor type 1 in murine lupus-like glo- grants from the National Health and Medical Research Council of merulonephritis. Kidney Int 48: 1459–1468, 1995 Australia (NH&MRC), the FWO, the Interuniversitary Attraction 19. Keeton M, Ahn C, Eguchi Y, Burlingame R, Loskutoff DJ: Poles, the Research Fund K.U.Leuven, and European Union. Expression of type 1 plasminogen activator inhibitor in renal tissue in murine lupus nephritis. Kidney Int 47: 148–157, 1995 20. Carmeliet P, Kieckens L, Schoonjans L, Ream B, van Nuffelen References A, Prendergast G, Cole M, Bronson R, Collen D, Mulligan RC: 1. Neale TJ, Tipping PG, Carson SD, Holdsworth SR: Participation Plasminogen activator inhibitor-1 gene-deficient mice. I. Gener- of cell-mediated immunity in deposition of fibrin in glomerulo- ation by homologous recombination and characterization. J Clin nephritis. Lancet 2: 421–424, 1988 Invest 92: 2746–2755, 1993 2. Kincaid-Smith P: Coagulation and renal disease. Kidney Int 2: 21. Carmeliet P, Stassen JM, Schoonjans L, Ream B, van den Oord 183–190, 1972 JJ, De Mol M, Mulligan RC, Collen D: Plasminogen activator 3. Vassalli P, McCluskey RT: Pathogenetic role of the coagulation inhibitor-1 gene-deficient mice. II. Effects on hemostasis, throm- process in rabbit Masugi nephritis. Am J Pathol 45: 653–677, bosis, and thrombolysis. J Clin Invest 92: 2756–2760, 1993 1964 22. Eitzman DT, McCoy RD, Zheng X, Fay WP, Shen T, Ginsburg 4. Holdsworth SR, Tipping PG: Macrophage-induced glomerular D, Simon RH: Bleomycin-induced pulmonary fibrosis in trans- fibrin deposition in experimental glomerulonephritis in the rab- genic mice that either lack or overexpress the murine plasmino- bit. J Clin Invest 76: 1367–1374, 1985 gen activator inhibitor-1 gene. J Clin Invest 97: 232–237, 1996 5. Naish P, Penn GB, Evans DJ, Peters DK: The effect of defibrina- 23. Kitching AR, Huang XR, Turner AL, Tipping PG, Dunn AR, tion on nephrotoxic serum nephritis in rabbits. Clin Sci 42: Holdsworth SR: The requirement for granulocyte-macrophage 643–646, 1972 colony-stimulating factor and granulocyte colony-stimulating 6. Drew AF, Tucker HL, Liu H, Witte DP, Degen JL, Tipping PG: factor in leukocyte-mediated immune glomerular injury. JAm Crescentic glomerulonephritis is diminished in fibrinogen-defi- Soc Nephrol 13: 350–358, 2002 cient mice. Am J Physiol Renal Physiol 281: F1157–F1163, 2001 24. Declerck PJ, Verstreken M, Collen D: Immunoassay of murine 7. Erlich JH, Holdsworth SR, Tipping PG: Tissue factor initiates t-PA, u-PA and PAI-1 using monoclonal antibodies raised in glomerular fibrin deposition and promotes major histocompati- gene-inactivated mice. Thromb Haemost 74: 1305–1309, 1995 bility complex class II expression in crescentic glomerulonephri- 25. Lijnen HR, Okada K, Matsuo O, Collen D, Dewerchin M: tis. Am J Pathol 150: 873–880, 1997 Alpha2-antiplasmin gene deficiency in mice is associated with 8. Kitching AR, Holdsworth SR, Ploplis VA, Plow EF, Collen D, enhanced fibrinolytic potential without overt bleeding. Blood Carmeliet P, Tipping PG: Plasminogen and plasminogen activators protect against renal injury in crescentic glomerulonephritis. 93:2274–2281, 1999 J Exp Med 185: 963–968, 1997 26. Tipping PG, Huang XR, Berndt MC, Holdsworth SR: A role for 9. Angles-Cano E, Rondeau E, Delarue F, Hagege J, Sultan Y, P selectin in complement-independent neutrophil- mediated glo- Sraer JD: Identification and cellular localization of plasminogen merular injury. Kidney Int 46: 79–88, 1994 activators from human glomeruli. Thromb Haemost 54: 688– 27. Huang XR, Holdsworth SR, Tipping PG: Evidence for delayed 692, 1985 type hypersensitivity mechanisms in glomerular crescent forma- 10. Rerolle JP, Hertig A, Nguyen G, Sraer JD, Rondeau EP: Plas- tion. Kidney Int 46: 69–78, 1994 minogen activator inhibitor type 1 is a potential target in renal 28. Bergman I, Loxley R: Two improved and simplified methods for fibrogenesis. Kidney Int 58: 1841–1850, 2000 the spectrophotometric determination of hyroxyproline. Anal 11. Eddy AA: Plasminogen activator inhibitor-1 and the kidney. Chem 35: 1961–1965, 1963 Am J Physiol Renal Physiol 283: F209–220, 2002 29. Holdsworth SR, Thomson NM, Glasgow EF, Atkins RC: The 12. Malliaros J, Holdsworth SR, Wojta J, Erlich J, Tipping PG: effect of defibrination on macrophage participation in rabbit Glomerular fibrinolytic activity in anti-GBM glomerulonephritis nephrotoxic nephritis: Studies using glomerular culture and elec- in rabbits. Kidney Int 44: 557–564, 1993 tron microscopy. Clin Exp Immunol 37: 38–43, 1979 13. Feng L, Tang WW, Loskutoff DJ, Wilson CB: Dysfunction of 30. Huang XR, Tipping PG, Shuo L, Holdsworth SR: Th1 responsive- glomerular fibrinolysis in experimental antiglomerular basement ness to nephritogenic antigens determines susceptibility to crescen- membrane antibody glomerulonephritis. J Am Soc Nephrol 3: tic glomerulonephritis in mice. Kidney Int 51: 94–103, 1997 1753–1764, 1993 31. Kalluri R, Danoff TM, Okada H, Neilson EG: Susceptibility to anti- 14. Oda T, Jung YO, Kim HS, Cai X, Lopez-Guisa JM, Ikeda Y, glomerular basement membrane disease and Goodpasture syndrome is Eddy AA: PAI-1 deficiency attenuates the fibrogenic response to linked to MHC class II genes and the emergence of T cell-mediated ureteral obstruction. Kidney Int 60: 587–596, 2001 immunity in mice. J Clin Invest 100: 2263–2275, 1997 J Am Soc Nephrol 14: 1487–1495, 2003 PAI-1 in Crescentic GN 1495

32. Kitching AR, Tipping PG, Holdsworth SR: IL-12 directs severe 37. Lacave R, Rondeau E, Ochi S, Delarue F, Schleuning WD, Sraer renal injury, crescent formation and Th1 responses in murine JD: Characterization of a plasminogen activator and its inhibitor glomerulonephritis. Eur J Immunol 29: 1–10, 1999 in human mesangial cells. Kidney Int 35: 806–811, 1989 33. Ploplis VA, French EL, Carmeliet P, Collen D, Plow EF: Plas- 38. Xu Y, Hagege J, Mougenot B, Sraer JD, Ronne E, Rondeau E: minogen deficiency differentially affects recruitment of inflam- Different expression of the plasminogen activation system in matory cell populations in mice. Blood 91: 2005–2009, 1998 renal thrombotic microangiopathy and the normal human kidney. 34. Edgtton KL, Carmeliet P, Kitching AR: Endogenous plasmin is Kidney Int 50: 2011–2019, 1996 not protective in unilateral ureteric ligation induced renal inter- 39. Duymelinck C, Dauwe SE, Nouwen EJ, De Broe ME, Verpooten stitial fibrosis. Nephrology 7: A 76, 2002 GA: Cholesterol feeding accentuates the cyclosporine-induced 35. Kitching AR, Turner AL, Lijnen HR: Endogenous ␣2-antiplas- elevation of renal plasminogen activator inhibitor type 1. Kidney min does not enhance fibrin deposition or injury in glomerulo- Int 51: 1818–1830, 1997 nephritis. Nephrology 7: A76, 2002 40. Wang AY, Poon P, Lai FM, Yu L, Choi PC, Lui SF, Li PK: 36. Dewerchin M, Collen D, Lijnen HR: Enhanced fibrinolytic Plasminogen activator inhibitor-1 gene polymorphism 4G/4G potential in mice with combined homozygous deficiency of alpha2-antiplasmin and PAI-1. Thromb Haemost 86: 640– genotype and lupus nephritis in Chinese patients. Kidney Int 59: 646, 2001 1520–1528, 2001