Nephrotoxic serum (NTS) was prepared by immunizing male sheep with a lysate of rat glomeruli, isolated by differential sieving as described previously (19). The sheep serum was heat inactivated at 56°C, absorbed with rat red blood cells and serum proteins, and then filter sterilized. As reported elsewhere (20), NTS has moderate reactivity to type IV collagen and laminin, and substantial reactivity to glomerular cell membrane proteins, particularly β1 integrin and its accompanying α chain.

The protocol for inducing crescentic glomerulonephritis in mice was that of Morley and Wheeler (21), modified for our NTS to avoid rapid lethality from acute renal failure while ensuring adequate tissue injury for analysis. CD1 mice weighing ∼30 g were pre-immunized by subcutaneous injection of 200 μg normal sheep IgG in Freund's complete adjuvant. After 5 d, experimental mice were injected intravenously with 50 μl NTS on three consecutive days and controls were injected with 50 μl normal sheep serum (NSS) on the same schedule. At various intervals from 12 h to 3 wk after the first dose of NTS, groups of 6 mice were killed by ether inhalation after an overnight collection of urine in single animal metabolism cages. During this collection, mice were allowed free access to water but not food. At the time of death, serum was obtained by cardiac puncture, and portions of kidney were snap frozen for immunohistology and RNA extraction, or fixed in 4% paraformaldehyde for in situ hybridization or in buffered formalin for routine histology.

Experiments were performed to modulate disease by the administration of mediators that neutralize or modify chemokine function. Thus in separate experiments, groups of mice (n = 6) received daily intravenous doses of either the RANTES antagonist, MetRANTES (22) or hamster anti-MCP1/JE (23) 30 min before administration of 50 μl of either NTS or NSS. MetRANTES was administered at either 16 or 32 μg per mouse per day, and anti–MCP-1/JE was given at 5 μg per mouse per day. Controls included mice given sterile PBS or 5 μg normal hamster immunoglobulin (Jackson ImmunoResearch Labs., West Grove, PA), respectively. The blocking reagents were given to mice in the stated doses from day 0 to day 6. Mice were then killed on day 7, after 24 h in metabolic cages, and kidneys were removed for morphological and immunohistological analyses, as described below.

Kidney halves were fixed overnight at 4°C in 10% neutral buffered formalin (Fisher), embedded in paraffin, sectioned at 3 μm and stained with hematoxylin and eosin following standard techniques. Sections from NTS mice were examined for glomerular hypercellularity and necrosis, mesangial thickening, formation of glomerular crescents, interstitial infiltrates, and development of glomerular and interstitial fibrosis, as compared to NSS mice. A crescent score was obtained by counting the number of glomeruli showing cellular crescents in at least 100 glomeruli per mouse.

Urine protein excretion was measured on timed overnight specimens collected at intervals from 3 d to 3 wk from individual mice in metabolism cages and assayed by the sulphosalicylic method (19). Blood urea nitrogen was measured on 200-μl serum samples by the Boston University Medical Center clinical laboratories using an auto-analyzer.

Kidneys from treated and untreated mice were excised, rolled in Tissue Tek OCT compound (Cryoform, IEC, Needham, MA) snap frozen in liquid nitrogen and stored at −70°C. 4-μm cryosections were cut onto microscope slides, air dried, and then fixed in acetone at 4°C for 20 min. Fixed sections were stained with rat monoclonal antibodies directed against Thy1 (T cells), GR-1 (granulocytes) (both PharMingen, San Diego, CA) and MOMA-2 (mononuclear phagocytes) (Biosource International, Camarillo, CA) using an avidin/biotin staining method. All incubations were carried out under humidified conditions and slides were washed twice between steps for 5 min each in 0.1 M PBS. First, endogenous peroxidase was blocked by incubation for 20 min in methanol containing 0.3% hydrogen peroxide. Nonspecific staining due to crossreaction with endogenous avidin or biotin was then blocked by incubation with avidin solution followed by biotin solution, both for 20 min (Vector, Burlingame, CA). Thereafter, sections were overlaid with 20% normal rabbit serum in PBS for 15 min and then incubated overnight at 4°C with mAbs specific for either Thy1, MOMA-2, or GR-1 (all at 1/10 in 10% normal mouse serum in PBS) (PharMingen). Bound monoclonal antibody was visualized by incubation with biotinylated rabbit anti–rat immunoglobulin diluted 1/200 in 10% normal mouse serum in PBS, and then streptavidin peroxidase complex (prepared according to manufacturer's instructions) for 30 min each (both Dako, Carpinteria, CA). Finally, slides were flooded with peroxidase substrate solution (400 mg diaminobenzidine in 10 ml of PBS, containing 0.01% hydrogen peroxide) for 10 min. Control slides were included where monoclonal antibody, biotinylated anti-rat immunoglobulin or streptavidin complex were selectively omitted. All slides were counterstained with hematoxylin. For each antibody the number of positively stained cells was determined in 15 high-power fields chosen at random per section (40× magnification; total area 1.8 mm²) and the average number of cells/field was calculated for each mouse.

Total RNA from kidneys obtained from NTS- and NSS-treated mice was isolated using the guanidinium thiocyanate/acid phenol procedure (24). RNA was extracted from PBS-treated mice at the same time points as NTS and NSS mice to determine basal levels of chemokines. Northern blots (25) were prepared with 20 μg of total RNA, purified as indicated above, fractionated in a 1.5% agarose/formaldehyde gel and blotted onto a nylon membrane (Genescreen, DuPont). Membranes were probed using ³²P-labeled probes for RANTES (26), MCP-1/JE (27), MIP-1α (28), and TCA-3 (29) applied in 50% formamide hybridization solution at 42°C for 18 h. Blots were washed in 2× SSC/1%SDS at 45°C and exposed at −70°C on Kodak XAR5 film (Sigma).

Samples of frozen kidney from NTS- and NSS-treated mice were sectioned at 4 μm, washed in PBS, fixed in acetone, and incubated with polyclonal rabbit anti– mouse type I collagen (Chemicon International, Temecula, CA). The primary antibody was detected with goat anti–rabbit IgGFITC (Jackson ImmunoResearch Labs.). Both the primary and secondary antibodies were diluted to their optimal working concentration in 2% sheep IgG in PBS. The effectiveness of this blocking procedure was confirmed by a negative staining reaction when a non-specific normal rabbit serum was substituted for the primary antibody. Tissues were also stained for sheep and mouse IgG and mouse C3 using various commercially available FITCconjugated antisera. Sections were imaged using NIH image 1.56, with 10 randomly chosen fields per section (at a magnification of ×20) being screened to calculate the mean area stained for each section.

These studies used a 600-bp cDNA probe that codes for the COOH terminus and a small part of the untranslated region of the rat α1 type I collagen gene (30) as previously described (31). The cDNA was inserted into and propagated in a pBluescript vector (Stratagene) carrying flanking T3 and T7 RNA polymerase promoters. Once linearized, either of the flanking RNA polymerase promoters could be employed to create radioactively labeled sense or anti-sense riboprobes by a runoff transcription method (25). Runoff transcription products were hydrolyzed in 0.01 M DTT, 0.08 M NaHCO₃, 0.12 M NaCO₃ for 60 min at 60°C and neutralized in 0.01 M DTT, 0.2 M Na acetate, 0.17 M acetic acid. The product was recovered by salt– ethanol precipitation and Sephadex G-50 chromatography (Pharmacia, Piscataway, NJ). The fractions of interest were pooled, the volume adjusted to give a final specific activity of 3 × 10⁶ cpm/ ml in 0.3 M DTT, and stored at −70°C until used. Paraffin sections of 4-μm thickness were cut onto Superfrost^®/Plus slides (Fisher Scientific, Pittsburgh, PA), dewaxed, and hydrated. Prehybridization, hybridization, and post-hybridization procedures were carried out exactly as previously described (31). Finally, the sections were coated with Kodak NTB-2 emulsion (Eastman Kodak Co., Rochester, NY) and sealed in light-tight boxes at 4°C for 7–14 d. The slides were developed with Kodak D-19 (Eastman Kodak Co.), counterstained with Mayer's hematoxylin and mounted using a water soluble mountant. Specificity was evaluated by use of the sense riboprobe or by introduction of an RNAse incubation step during the prehybridization. Sections were examined and photographed with a Nikon Optiphot microscope equipped for dark-field illumination at 100× magnification and bright-field illumination at 100× and 400× magnification (Nikon, Garden City, NY).

The significance of differences between experimental groups was calculated by the Wilcoxon MannWhitney test using Xlstat in Microsoft Excel 5.

The severity of glomerulonephritis was dose dependent and varied from mild glomerular hypercellularity to anuric acute renal failure and early death from extensive glomerular necrosis and crescent formation. There were no differences in histological appearance of kidneys from NSS mice compared with mice given PBS only (Fig. 1 A). 12 h after the first injection of NTS, proteinaceous casts were present in renal tubules, coincident with the onset of proteinuria. After 24 h, leukocytes were evident within the interstitium and by day 3 prominent perivascular infiltrates were present (Fig. 1 B). At this point the glomeruli were clearly hypercellular with focal areas of necrosis. On day 5 some glomeruli contained crescents and interstitial and perivascular infiltrates were pronounced. On day 7, 55% of glomeruli exhibited prominent cellular crescents (Fig. 1 C) and marked abnormalities were present within the tubules and interstitium. These included severely dilated tubules with flattened or denuded epithelia and expansion of the interstitium due to edema, interstitial infiltrates, and the onset of fibrosis (Fig. 1 D). On day 14, the fibrotic process was more diffuse throughout the interstitium and had extended to involve some of the glomerular crescents, apparently by invasion through Bowman's capsule.

Mice injected with NTS were significantly proteinuric by day 3 and became progressively more so throughout the course of the study (Fig. 2 A). Blood urea nitrogen levels rose rapidly during the acute phase of immunologic injury but then improved towards baseline values over the course of 10 d (Fig. 2 B). However, concomitant with the development of progressive interstitial and glomerular fibrosis, blood urea nitrogen levels rose progressively from day 14.

Over the course of the disease, both the interstitium and crescents became increasingly fibrotic as demonstrated by Masson trichrome staining (not shown) and deposition of type I collagen (Fig. 3). Most strikingly, a rapid induction of cells expressing mRNA for α1 type I collagen was seen by in situ hybridization (ISH) in perivascular, periglomerular and intervening regions of cortex within 3 d of the final injection of NTS (Fig. 3, A and C). These cells differed from those that constitutively express α1(I) collagen in the perivascular region of normal and control mice and they progressively increased in number, intensity of signal, and extent over the course of disease. Immunofluorescent staining showed deposition of collagen protein in corresponding areas (Fig. 3, B and D). Initially, glomeruli were negative for type I collagen by IF and ISH, but with time (around day 14), α1(I) mRNApositive cells encircling the glomeruli appeared to infiltrate the Bowman's capsule of some glomeruli and were followed by positive IF staining of the crescents for type I collagen (data not shown).

Infiltrating lymphocytes, neutrophils and mononuclear phagocytes were identified at different time points during the development of crescentic GN by immunohistochemical staining of frozen sections. (Fig. 4). Increases in both neutrophils and mononuclear phagocytes, but not lymphocytes, were observed 12 h after the first injection of NTS. However, by day 3, all three cell types were increased in kidneys from NTS mice when compared with those from NSS mice at the same time period. Maximal numbers of neutrophils were observed at day 3 of disease, whereas lymphocyte numbers continued to be raised even at day 14. Peak macrophage infiltration was seen at day 7 with a decline in numbers occurring by day 14.

Expression of chemokines involved in the recruitment of leukocytes to sites of inflammation was determined by Northern analysis of RNA isolated throughout the development of crescentic GN (Fig. 5). Therefore, mRNA expression for RANTES (CD4⁺ T cell and eosinophil chemoattractant) (32); MCP-1/JE (monocyte chemoattractant) (33); MIP-1α (monocyte and eosinophil chemoattractant) (28); and TCA3 (neutrophil and monocyte chemoattractant) (23) was assessed. There was no expression of either MCP-1/JE or RANTES at 12 or 24 h after the induction of disease, but from day 3 onwards NTS treated mice showed strong expression of these chemokines, whereas NSS- or PBS-treated mice did not. There was no detectable expression of TCA-3 in either NTS- or NSS-treated mice (data not shown). Low expression of MIP-1α was detected in all mice, with no discernible difference between mice from NTS or NSS groups.

MetRANTES was used to neutralize the effect of this chemokine in order to define the functional role that RANTES plays in the development of crescentic GN. Day 7 was chosen to determine the effect of blocking RANTES as by this time there is a significant infiltration by leukocytes, an increase in proteinuria, and the beginning of the development of fibrosis, as marked by collagen I deposition. Thus, 16 μg MetRANTES was administered daily until day 6, when mice were housed in metabolic cages for 24 h before killing on day 7. Four disease parameters were then assessed for the effect of MetRANTES: 24 h proteinuria, leukocyte infiltration, glomerular crescent formation and collagen deposition. First, MetRANTES was seen to induce a 36% decrease in excretion of protein (31 ± 5 mg/24 h in treated versus 52 ± 7 mg/24 h in untreated NTS mice, P <0.05) (Fig. 6 A). In conjunction with this there was a 52% decrease in numbers of T cells and a 40% decline in mononuclear phagocyte numbers within sections (P <0.005). However, there was no difference in numbers of neutrophils after MetRANTES treatment (Fig. 6 B). A 16% decrease in glomerular crescent formation was counted in H&E stained sections (not significant), although no other changes in morphological abnormalities was seen to occur (Fig. 6 C). There was no difference in the deposition of type I collagen after administration of MetRANTES (Fig. 6 D). This experiment was repeated with a higher dose of MetRANTES (32 μg/mouse/d) without additional benefit (data not shown).

A neutralizing monoclonal antibody specific for MCP-1/JE was used to assess the contribution that this chemokine makes to the development of crescentic GN. Experiments were conducted exactly as described above for MetRANTES, with 5 μg of anti–MCP-1/JE or 5 μg hamster immunoglobulin G (hIgG) being administered daily from day 0 to day 6 to NTS or NSS mice. Hamster Ig had no effect upon the development of disease. Proteinuria was seen to diminish by 25% upon administration of anti–MCP-1/JE (P <0.01) (Fig. 7 A). Immunohistochemical phenotyping revealed that neutralizing MCP-1/JE resulted in a 47% decrease in macrophages within both glomerular and the interstitial areas (P <0.05). Similarly, there was a decrease in numbers of lymphocytes (P <0.05). Interestingly, neutralizing MCP1/JE enhanced the increase in numbers of neutrophils, by 37% compared to untreated NTS mice (P <0.01)(Fig. 7 B). Interestingly there was a striking decrease (by 73.5%) in the number of glomerular crescents on day 7 in treated mice compared to untreated NTS mice (P <0.001) (Fig. 7 C). Concomitant with this there was a 65% decrease in the deposition of type I collagen in sections from anti–MCP-1–treated NTS mice compared to untreated NTS mice (P <0.005) (Fig. 7 D).