The Requirement for Granulocyte-Macrophage Colony- Stimulating Factor and Granulocyte Colony-Stimulating Factor in Leukocyte-Mediated Immune Glomerular Injury

Home , Granulocyte

J Am Soc Nephrol 13: 350–358, 2002 The Requirement for Granulocyte-Macrophage Colony- Stimulating Factor and Granulocyte Colony-Stimulating Factor in Leukocyte-Mediated Immune Glomerular Injury

A. RICHARD KITCHING,* XIAO RU HUANG,* AMANDA L. TURNER,* PETER G. TIPPING,* ASHLEY R. DUNN,† and STEPHEN R. HOLDSWORTH* *Centre for Inflammatory Diseases, Monash University, Department of Medicine, Monash Medical Centre, Clayton, Australia; †Ludwig Institute for Cancer Research, Royal Melbourne Hospital, Parkville, Victoria, Australia.

Abstract. Proliferative glomerulonephritis in humans is char- ologous-phase anti-GBM glomerulonephritis compared with acterized by the presence of leukocytes in glomeruli. Granu- genetically normal (CSF WT) mice, with reduced proteinuria locyte-macrophage colony-stimulating factor (GM-CSF) and and glomerular neutrophil numbers. However, only GM-CSF granulocyte colony-stimulating factor (G-CSF) can potentially Ϫ/Ϫ mice were protected from crescentic glomerular injury in stimulate or affect T cell, macrophage, and neutrophil function. the autologous phase, whereas G-CSF Ϫ/Ϫ mice were not To define the roles of GM-CSF and G-CSF in leukocyte- protected and in fact had increased numbers of T cells in mediated glomerulonephritis, glomerular injury was studied in glomeruli. Humoral responses to the nephritogenic antigen mice genetically deficient in either GM-CSF (GM-CSF Ϫ/Ϫ were unaltered by deficiency of either GM-CSF or G-CSF, but mice) or G-CSF (G-CSF Ϫ/Ϫ mice). Two models of glomer- glomerular T cell and macrophage numbers, as well as dermal ulonephritis were studied: neutrophil-mediated heterologous- delayed-type hypersensitivity to the nephritogenic antigen, phase anti–glomerular basement membrane (GBM) glomeru- were reduced in GM-CSF Ϫ/Ϫ mice. These studies demon- lonephritis and T cell/macrophage-mediated crescentic strate that endogenous GM-CSF plays a role in experimental autologous-phase anti-GBM glomerulonephritis. Both GM- glomerulonephritis in both the autologous and heterologous CSF Ϫ/Ϫ and G-CSF Ϫ/Ϫ mice were protected from heter- phases of injury.

Granulocyte-macrophage colony-stimulating factor (GM-CSF) cell function (8), and, via its secretion by the T helper 1 (Th1) and granulocyte colony-stimulating factor (G-CSF) are growth subset, the full expression of dermal delayed-type hypersensi- factors and cytokines that are classically described as acting on tivity (DTH) responses (9). hemopoietic cells, particularly granulocytes and monocytes. The most aggressive forms of glomerulonephritis (GN) They are involved in the normal expression of innate and show significant glomerular leukocytic infiltration (10). A cognate immune responses and have been considered to be number of studies in experimental models of GN have con- required for the normal development of hemopoietic progenitor firmed that infiltrating glomerular leukocytes are major effec- cells (1,2). The generation of mice genetically deficient in tors of glomerular injury in proliferative GN (11–13). Prolif- either G-CSF (G-CSF Ϫ/Ϫ mice) or GM-CSF (GM-CSF Ϫ/Ϫ erative forms of human GN show a variety of immune effectors mice) have facilitated studies of their roles in hematopoiesis and histologic outcomes with variable involvement of leuko- and in immune and inflammatory injury. G-CSF Ϫ/Ϫ mice cyte subsets (14). Underpinning many of these differences is show abnormal numbers of bone marrow and circulating gran- the Th1 or T helper 2 predominance of the systemic nephrito- ulocytes (3) and deficient granulocyte defense to infective genic immune response (14), with evidence emerging that Th1 Ϫ Ϫ microorganisms (4). GM-CSF / mice have normal numbers responses are important in the development of severe prolifer- of circulating leukocytes and bone marrow hemopoietic pro- ative and crescentic GN (14–17). genitors (5). GM-CSF has proinflammatory effects on granu- Neutrophils have been implicated in a number of forms of locytes (1) and is required for normal function of at least some proliferative human GN, including anti-neutrophil cytoplasmic populations of dendritic cells (6,7), monocyte activation (1), T antibody–associated crescentic GN (18,19). The best-characterized experimental model of neutrophil-mediated immune glomerular injury is heterologous-phase anti–glomerular base- Received November 3, 2000. Accepted September 6, 2001. ment membrane (GBM) GN (or nephrotoxic serum nephritis). Correspondence to Prof. Stephen R Holdsworth, Monash University, Depart- In this model, GN is induced by administering a heterologous ment of Medicine, Monash Medical Centre, 246 Clayton Road, Clayton, globulin directed against the GBM. Glomerular injury is asso- Victoria 3168, Australia. Phone: 61-3-9594-5525; Fax: 61-3-9594-6437; E- mail: [email protected] ciated with neutrophil accumulation in glomeruli (11) and is induced predominantly by neutrophil degranulation with the 1046-6673/1302-0350 Journal of the American Society of Nephrology release of proteolytic enzymes and reactive oxygen species Copyright © 2002 by the American Society of Nephrology (20,21). There is little known involvement for macrophages in J Am Soc Nephrol 13: 350–358, 2002 Requirement for GM-CSF and G-CSF in GN 351

this form of acute glomerular injury in murine anti-GBM GN ANOVA, followed by Tukey’s multiple comparison test for paired (11), although macrophages may have a role in the heterolo- comparisons (GraphPad Prism, GraphPad Software Inc., San Diego, gous phase in rat models (22). CA). DTH responses have been shown to be important in experimental crescentic GN (12). The presence of DTH effectors in Assessment of Glomerular Crescent Formation human crescentic GN, whether or not glomerular antibody Kidney tissue was fixed in Bouin fixative and embedded in paraf- deposition is present, implies that findings in experimental fin, and 3-␮m tissue sections were cut and stained with periodic models are relevant to human disease (23,24). Th1-predomi- acid–Schiff. A glomerulus was considered to exhibit crescent forma- nant nephritogenic immune responses result in strong glomer- tion if two or more layers of cells were observed in Bowman space. A minimum of 50 glomeruli was assessed to determine the crescent ular DTH with T cell–directed macrophage recruitment and score for each animal. the development of crescentic GN (15,25). The experimental model most widely used to study these events is the autologous phase of experimental anti-GBM antibody–induced GN. In this Glomerular T Cell, Macrophage, and Neutrophil model it is the host (autologous) immune response to the Accumulation deposited heterologous Ig (the heterologous anti-GBM anti- Kidney tissue was fixed in periodate lysine paraformaldehyde for 4 h, washed in 7% sucrose solution, and then frozen in liquid nitrogen. body) in glomeruli that induces immune injury (26). ϩ Tissue sections (6 ␮m) were stained to demonstrate CD3 cells, In human glomerulonephritis, evidence that GM-CSF may neutrophils, and macrophages by use of a three-layer immunoperox- play a functional role in inducing injury in GN comes from idase technique, as described elsewhere (12,30). The primary mono- observations in humans in which GM-CSF has been found to clonal antibodies were KT3 anti-mouse CD3 (American Type Culture be increased in human renal biopsies of patients with prolifer- Collection, Manassas, VA), the anti-mouse macrophage monoclonal ative GN (27). In experimental GN, injection of tubular epi- FA/11 antibody (a gift of Dr G.L. Koch, MRC Laboratory of Molec- thelial cells transfected with the gene for GM-CSF under the ular Biology, Cambridge, UK) (31), and RB6–8C5 (anti Gr-1, which renal capsule of lupus-prone MRL/lpr mice induces inflamma- recognizes neutrophils, a gift from DNAX Research Institute, Palo tory renal injury (28). G-CSF and GM-CSF have the potential Alto, CA). A minimum of 20 equatorially sectioned glomeruli was to play important roles in the effector functions of inflamma- assessed per animal, and results were expressed as cells per glomer- tion induced by granulocytes, T cells, and monocytes. They are ular cross section (c/gcs). therefore potential therapeutic targets in injurious inflammatory diseases. Because of evidence of participation of granu- Glomerular Deposition of Sheep Globulin, Mouse Ig, locytes and mononuclear leukocytes in several forms of pro- and C3 liferative GN, we sought to determine the roles of G-CSF and For detection of sheep globulin, immunofluorescence was per- ␮ GM-CSF in models of leukocyte-induced glomerular injury in formed on 4- m cryostat-cut periodate lysine paraformaldehyde– fixed tissue, detected by use of FITC-rabbit anti-sheep Ig (Silenus, which either granulocytes or mononuclear cells are known to Hawthorn, Victoria, Australia) at dilutions of 1:250 and/or 1:1000. be the major effectors of injury. For detection of autologous antibody and C3 in glomeruli, tissue was embedded in Optimal Cutting Temperature Compound, frozen in liquid nitrogen, and stored at Ϫ70°C. Mouse Ig was detected by use Materials and Methods of FITC-sheep anti-mouse Ig (Silenus, Hawthorn, Victoria, Australia) Experimental Design at dilutions of 1:100 and 1:10,000 and C3 by use of FITC-goat Anti-mouse GBM globulin was prepared from serum (adsorbed anti-mouse C3 (Cappel, Durham, NC) at dilutions of 1:100 and twice with mouse red blood cells then precipitated with ammonium 1:5000. Fluorescence intensity was assessed semiquantitatively (0 to sulfate) of a sheep immunized against homogenized and sonicated 3ϩ). Sections in which only some glomeruli were positive were murine renal cortex in Freund’s complete adjuvant (Sigma Chemical graded as 0.5 (i.e. ϩ/Ϫ). Co., St. Louis, MO) and, later, Freund’s incomplete adjuvant. Mice ϫ with a genetic deletion of either GM-CSF or G-CSF on a C57BL/6 Proteinuria and Serum Creatinine 129Ola background were bred at Monash University (Melbourne, Australia) in an specific pathogen-free facility. For studies in heter- Urinary protein concentrations were determined by a modified ologous injury, male GM-CSF Ϫ/Ϫ, G-CSF Ϫ/Ϫ, and genetically Bradford method on timed urine collections. For studies in heterolo- normal C57BL/6 ϫ 129Ola mice, 8 to 10 wk of age, were injected gous injury, mice were placed on metabolic cages 8 h after intrave- with 4 mg of sheep anti-mouse GBM globulin. In one experiment, nous injection of sheep anti-mouse GBM globulin. Urine was col- glomerular neutrophil accumulation was studied at1h(n ϭ 8 each lected for 16 h and results converted to provide a value for 24 h. For group), the time of maximum influx in this model (29). In a further studies in autologous GN, urine was collected for 24 h from days 9 to experiment, mice were placed in metabolic cages 8 h after injection, 10 of GN. Serum creatinine concentrations at the completion of and urine was collected for 16 h for measurement of urinary protein experiments in autologous phase GN (day 10) were measured by the excretion (n ϭ 6 each group). For studies in autologous injury (n ϭ alkaline picric acid method with the use of an autoanalyzer. 7 each group), mice were sensitized by subcutaneous injection of a total of 100 ␮g of sheep globulin in 50 ␮l of Freund’s complete Skin DTH to Sheep Globulin adjuvant in divided doses in each flank 10 d before challenge with 2.5 In the experimental groups where GN was induced and studied at mg of sheep anti-mouse GBM globulin. Renal injury was studied after day 10, mice were challenged 24 h before the end of the experiment a further 10 d. Results are expressed as the mean Ϯ SEM. The by intradermal injection of sheep globulin (100 ␮gin50␮lof significance of differences between groups was determined by phosphate-buffered saline) into the plantar surface of a hind foot. An 352 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 350–358, 2002 equivalent dose of an irrelevant antigen (horse globulin) in an equivalent volume of phosphate-buffered saline was injected in the opposite foot pad as a control. DTH was assessed 24 h later by measuring the difference between the sheep– and horse globulin– injected foot pads in each mouse by use of a micrometer (Mitutoyo Corporation, Japan).

Titers of Serum Anti-Sheep Globulin Ig Titers of mouse anti-sheep globulin Ig were measured by enzyme- linked immunosorbent assay on serum collected at the end of experiments in autologous injury, as described elsewhere (16). Plates were coated with 10 ␮g/ml normal sheep globulin, washed, then blocked with 1% bovine serum albumin in phosphate-buffered saline. Plates were washed, then incubated with mouse serum at the dilutions of 1:100, 1:1000, and 1:10,000, and incubated for1hat37°C. Bound mouse Ig was detected with horseradish peroxidase conjugated sheep anti-mouse Ig (Amersham, Little Chalfont, UK, 1:2000); 0.1 M 2,2'- azino-di-3-ethylbenzthiazoline sulfonate (Boehringer Mannheim, Germany) substrate solution was added and the absorbance read at Figure 1. Heterologous injury in colony-stimulating factor (CSF) wild 405 nm. Sera from nonimmunized mice were tested to provide neg- type (WT), granulocyte–CSF (G–CSF Ϫ/Ϫ), and granulocyte-macro- ative controls. phage–CSF (GM–CSF Ϫ/Ϫ) mice showing a (A) reduced glomerular neutrophil influx 1 h after injection of sheep anti-mouse glomerular basement membrane (GBM) globulin as well as (B) decreased urinary Peripheral Blood White Blood Cell and Neutrophil Ϫ Ϫ Ϫ Ϫ Counts protein excretion at 24 h in both G-CSF / and GM-CSF / mice, compared with CSF WT mice (*P Ͻ 0.05, **P Ͻ 0.01 versus The effects of the administration of heterologous antigen on pe- CSF WT mice). The horizontal dotted line in panel B represents the ripheral blood neutrophils were studied by use of mice 9 to 11 wk of mean value for normal mice without glomerulonephritis (GN). age. Mice were bled at baseline or 1 h after intravenous injection of 4 mg sheep anti-mouse GBM, to determine total leukocyte count and the circulating neutrophil count. For baseline studies, six GM-CSF Ϫ Ϫ Ϫ Ϫ / mice, four G-CSF / mice, and six CSF wild type (WT) mice significant and similar attenuation of glomerular neutrophil were used; for studies 1 h after administration of sheep anti-mouse numbers (G-CSF Ϫ/Ϫ, 0.61 Ϯ 0.12 c/gcs; GM-CSF Ϫ/Ϫ, 0.62 GBM, eight GM-CSF Ϫ/Ϫ mice, five G-CSF Ϫ/Ϫ mice, and seven Ϯ 0.08 c/gcs, both P Ͻ 0.05). There was no significant influx CSF WT mice were used. One hundred microliters of blood was ϩ obtained from each mouse and added to 50 ␮l of 3.3% tri-sodium of T cells or macrophages into glomeruli at 1 h (CD3 cells: Ϯ Ϫ Ϫ Ϯ citrate. A 100 ␮l volume of blood/tri-sodium citrate mix was trans- CSF WT, 0.10 0.03 c/gcs; G-CSF / ,0 0 c/gcs; ferred to a tube and red blood cells lysed (Q prep, Coulter Electronics, GM-CSF Ϫ/Ϫ, 0.10 Ϯ 0.01 c/gcs and macrophages: CSF WT, Sydney, Australia) and total white blood cell count determined via a 0.10 Ϯ 0.04 c/gcs; G-CSF Ϫ/Ϫ, 0.05 Ϯ 0.01 c/gcs; GM-CSF ϩ hemocytometer. Subsequently, the percentage and absolute number of Ϫ/Ϫ,0Ϯ 0 c/gcs) or at 24 h in this lesion (CD3 cells: CSF neutrophils present in the sample were determined by flow cytometry WT, 0.17 Ϯ 0.03 c/gcs; G-CSF Ϫ/Ϫ, 0.06 Ϯ 0.03 c/gcs; (Mo-flo cytometer, Fort Collins, CO), gating on the granulocyte GM-CSF Ϫ/Ϫ, 0.13 Ϯ 0.03 c/gcs and macrophages: CSF WT, population via forward and 90° scatter. 0.14 Ϯ 0.04 c/gcs; G-CSF Ϫ/Ϫ, 0.03 Ϯ 0.02 c/gcs; GM-CSF Ϫ/Ϫ, 0.06 Ϯ 0.03 c/gcs), consistent with an important role for Results neutrophils in this lesion in mice (11). Both G-CSF Ϫ/Ϫ Mice and GM-CSF Ϫ/Ϫ Mice Are There was no difference in the deposition of sheep glob- Protected from Neutrophil-Mediated, Heterologous- ulin in glomeruli among the three groups of mice (1:250 Phase GN dilution average score 2ϩ [fluorescence intensity], 1:1000, CSF WT mice developed significant proteinuria 24 h after ϩ for all groups), which shows that the absence of either intravenous injection of 4 mg sheep anti-mouse GBM globulin G-CSF or GM-CSF did not alter the deposition of the (9.5 Ϯ 0.7 mg/24 h; normal, 0.8 Ϯ 0.2 mg/24 h, Figure 1B). disease-initiating globulin. The potential for the disease to Histologic examination of kidneys at the time of peak neutro- be affected by differences in circulating neutrophil numbers phil accumulation in this model (1 h) (29) revealed exudative was assessed. The total white blood cell and circulating proliferative GN characterized by a significant neutrophil in- neutrophil count, both before and after injection of sheep flux in CSF WT mice (2.5 Ϯ 0.7 c/gcs; mice without GN, 0.1 anti-mouse GBM, was reduced in G-CSF Ϫ/Ϫ mice com- Ϯ 0.01 c/gcs; Figures 1A and 2A). Both G-CSF Ϫ/Ϫ and pared with CSF WT mice and GM-CSF Ϫ/Ϫ mice (Table 1). GM-CSF Ϫ/Ϫ mice were significantly protected from heter- These findings are consistent with data published elsewhere ologous-phase glomerular injury (Figures 1 and 2, B and C). on the phenotype of this strain of mouse (3). However, there Proteinuria in G-CSF Ϫ/Ϫ mice was 3.5 Ϯ 0.3 mg/24 h, and was no difference in total white blood cell count or circu- that in GM-CSF Ϫ/Ϫ mice was 3.7 Ϯ 0.5 mg/24 h (both P Ͻ lating neutrophil count between GM-CSF Ϫ/Ϫ mice and 0.01, Figure 1B). These reductions were associated with a CSF WT mice, either before or after injection. J Am Soc Nephrol 13: 350–358, 2002 Requirement for GM-CSF and G-CSF in GN 353

Table 1. Circulating total white blood cell (WBC) counts at an antibody dilution of 1:100. Fluorescence intensity had and neutrophil counts in mice before and 1 h after fallen to ϩ/Ϫ (i.e., only some glomeruli in each section were injection with sheep anti-mouse glomerular positive) in each group at a dilution of 1:10,000 for autologous basement membrane globulin, showing leucopenia antibody and 1:5000 for C3. and neutropenia in G-CSFϪ/Ϫ micea CSF WT mice developed renal impairment (serum creatinine 44 Ϯ 5 ␮mol/L, Figure 4A) and heavy proteinuria (9.4 Ϯ Total WBC Neutrophils Group State 6 6 Ϫ Ϫ (ϫ10 /ml) (ϫ10 /ml) 1.2 mg/24 h, Figure 4B). G-CSF / mice with autologous, crescentic GN were not protected from functional renal injury. CSF WT Preinjection 9.6 Ϯ 2.8 1.8 Ϯ 0.7 Their degree of proteinuria and renal impairment was similar to Postinjection 7.5 Ϯ 2.6 2.7 Ϯ 1.2 that of CSF WT mice (Figure 4). However, consistent with the GM-CSFϪ/Ϫ Preinjection 9.8 Ϯ 1.8 2.3 Ϯ 0.6 significant reduction in glomerular crescent formation and Postinjection 8.0 Ϯ 2.8 2.8 Ϯ 1.0 glomerular T cell and macrophage infiltrate, GM-CSF Ϫ/Ϫ G-CSFϪ/Ϫ Preinjection 1.9 Ϯ 1.2 0.04 Ϯ 0.01 mice were significantly protected from the development of Postinjection 2.2 Ϯ 0.5 0.25 Ϯ 0.07 renal impairment (serum creatinine 20 Ϯ 1 ␮mol/L, P Ͻ 0.01, Ϫ Ϫ a Figure 4A). Proteinuria in GM-CSF / mice was reduced by G-CSF, granulocyte colony-stimulating factor; GM, ϳ Ϯ Ͻ granulocyte-macrophage; WT, wild-type; Ϫ/Ϫ mice, knockout 40% (5.6 0.4 mg/24 h, P 0.01, Figure 4B). mice. Results are expressed as the mean Ϯ SEM. Systemic Immune Responses to Sheep Globulin In the final 24 h of experiments in autologous-phase (cres- Only GM-CSF Ϫ/Ϫ Mice Are Protected from centic) injury, mice were challenged subcutaneously with Crescentic Glomerulonephritis in the Autologous Phase sheep and horse globulin. The difference between the two sites of Injury of challenge served as a measure of dermal DTH. Control mice Control animals develop a severe proliferative crescentic developed significant antigen-specific skin DTH (Figure 5). GN (51 Ϯ 4% of glomeruli affected [Figures 2D and 3A]). Although the degree of DTH was unchanged in G-CSF Ϫ/Ϫ This was associated with a significant influx of inflammatory mice, dermal DTH was significantly reduced in the absence of cells, including CD3ϩ T cells, and macrophages and a minor GM-CSF. degree of neutrophil accumulation (Figure 3, B through D). In the autologous phase of injury, all mice developed mea- G-CSF Ϫ/Ϫ mice developed similar numbers of glomerular surable titers of sheep globulin specific antibodies in serum crescents and were not protected from autologous injury (Fig- drawn at day 10 of GN (Figure 6). However, there were no ures 2F and 3A), which demonstrates that G-CSF is not re- differences among CSF WT, G-CSF Ϫ/Ϫ, and GM-CSF Ϫ/Ϫ quired for the development of crescentic GN (53 Ϯ 4% of mice: all groups developed similar titers of antigen-specific glomeruli affected). However, GM-CSF Ϫ/Ϫ mice were sig- circulating antibodies (Figure 6). nificantly protected from severe crescentic glomerular injury, glomerular crescents being present in only 13 Ϯ 2% of glo- Discussion meruli (Figure 2E and 3A). In this autologous form of injury, These studies show that both neutrophil-associated glomer- where injury is driven by an adaptive nephritogenic immune ular injury and T cell/macrophage-mediated crescentic glomer- response directed against sheep globulin, there was no differ- ular injury are mediated by endogenous GM-CSF Ϫ/Ϫ, ence in the deposition of the nephritogenic antigen (sheep whereas G-CSF is important in neutrophil-mediated injury globulin) in glomeruli among the three groups of mice (1:1000 (possibly by virtue of its effects on the development of neu- dilution average score ϩ for all groups). trophils) and has no pathogenetic role in experimental crescen- Reductions in crescentic injury observed in GM-CSF Ϫ/Ϫ tic GN. mice with GN were paralleled by changes in the T cell and G-CSF is known to be important in the generation of gran- macrophage infiltrate in glomeruli. Although 1.4 Ϯ 0.1 CD3ϩ ulocyte stem cell progenitors. Mice with a targeted disruption cells per glomerular cross-section were observed in CSF WT of the gene for this cytokine have a known predisposition to mice with GN, GM-CSF Ϫ/Ϫ mice had one-third of this bacterial infection (4). However, this deficiency has variable number in glomeruli (0.5 Ϯ 0.10 c/gcs, Figure 3B) A similar effects on neutrophil recruitment and function in different reduction was observed in glomerular macrophage infiltrate models of inflammation. Zhan et al. (4) reported increased (CSF WT, 3.3 Ϯ 0.4 c/gcs and GM-CSF Ϫ/Ϫ 1.1 Ϯ 0.05 mortality to and exacerbated in vivo proliferation of Listeria c/gcs; Figures 2H and 3C). The modest increase in CD3ϩ cells monocytogenes in G-CSF Ϫ/Ϫ mice, and Metcalf et al. (32) in glomeruli of G-CSF Ϫ/Ϫ mice reached statistical signifi- observed that the intraperitoneal injection of casein (containing cance (P Ͻ 0.05, Figure 3B), but other indices of injury were bacteria) into G-CSF Ϫ/Ϫ mice did not affect the subsequent not significantly increased. neutrophil exudation observed. The model of heterologous In contrast to the alterations observed in the cellular infiltrate anti-GBM antibody–induced GN used in those studies is asso- in glomeruli, the deposition of humoral reactants (autologous ciated with a significant glomerular neutrophil influx. The antibody and C3) was not affected by deficiency of either accumulating neutrophils have been shown to be responsible G-CSF or GM-CSF. When examined at two dilutions, average for the associated glomerular injury (11,21). In these studies, semiquantitative fluorescence intensity was 3ϩ in each group G-CSF Ϫ/Ϫ mice were found to be neutropenic, as reported 354 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 350–358, 2002

Figure 2. Histologic features of injury in CST WT, GM-CSF Ϫ/Ϫ, and G-CSF Ϫ/Ϫ mice with heterologous- and autologous-phase anti-GBM GN. (A through C) Heterologous injury: (A) genetically normal mice (CSF WT) developed proliferative GN with neutrophils in glomeruli (brown reaction product). Mice deficient in either GM-CSF (B) or G-CSF (C) had fewer neutrophils in their glomeruli. (D through F) Autologous-phase anti-GBM GN: (D) CSF WT mice developed severe proliferative and crescentic GN. GM-CSF Ϫ/Ϫ mice had a lesser degree of injury (E), whereas renal injury in G-CSF Ϫ/Ϫ mice was as severe as that of CSF WT mice (F), with substantial glomerular crescent formation. (G through I) Macrophage infiltration in autologous phase anti-GBM GN: CSF WT mice with GN had significant numbers of FA/11-positive macrophages (brown reaction product) within glomeruli (G). Numbers were significantly reduced in GM-CSF Ϫ/Ϫ mice (H), but glomerular macrophage numbers (arrowheads) were unchanged in G-CSF Ϫ/Ϫ mice compared with CSF WT mice with GN (I). (A through C and G through I) Immunoperoxidase with hematoxylin counterstain. (D through F) periodic acid–Schiff stain. Magnification: ϫ200. elsewhere (3). These animals also showed a markedly reduced gous injury, despite having normal circulating neutrophil num- glomerular influx of neutrophils and, consequently, significant bers. Thus deficiency of either G-CSF or GM-CSF gave protection from the development of glomerular injury. How- similar levels of protection in this model of acute antibody- ever GM-CSF Ϫ/Ϫ mice were also protected from heterolo- induced neutrophil-mediated glomerular injury. Although there J Am Soc Nephrol 13: 350–358, 2002 Requirement for GM-CSF and G-CSF in GN 355

Figure 4. Functional indices of injury in mice with autologous-phase anti-GBM GN. (A) Serum creatinine, showing impaired renal function in CSF WT mice with GN, and significantly lower serum creatinine in GM-CSF Ϫ/Ϫ mice (**P Ͻ 0.01 versus CSF WT mice). (B) Proteinuria, showing a reduction in GM-CSF Ϫ/Ϫ mice compared with CSF WT mice (**P Ͻ 0.01 versus CSF WT mice). G-CSF Ϫ/Ϫ mice were not protected from functional renal injury in this model. The horizontal dotted lines represent mean values for normal mice without GN.

antigen as well as prevention of the accumulation of glomerular DTH effectors and injury in this model (25,33). G-CSF Ϫ/Ϫ mice developed normal dermal DTH and titers of circulating antibody to the nephritogenic antigen. There was no deficiency in glomerular DTH, because significant numbers of ϩ Figure 3. Glomerular injury in autologous (crescentic) anti-GBM GN, CD3 cells and macrophages were observed in the glomeruli ϩ of injected animals. There was a statistically significant in- showing (A) reduced glomerular crescent formation, (B) CD3 cells, ϩ and (C) macrophages in glomeruli of mice lacking endogenous GM- crease in CD3 cells within glomeruli in G-CSF Ϫ/Ϫ mice, CSF but no reduction in glomerular crescent formation in G-CSF Ϫ/Ϫ which suggests that endogenous G-CSF may in fact play a mice and, in fact, an increase in intraglomerular T cells (*P Ͻ 0.05, protective role in the development of this lesion. However, **P Ͻ 0.01 versus CSF WT mice for panels A through C). (D) The glomerular crescent formation, renal impairment, and protein- low numbers of neutrophils in glomeruli were not different in the uria were not increased in the absence of G-CSF, which sug- absence of either G-CSF or GM-CSF. gests that the increase in glomerular T cells, although statistically significant, was not biologically relevant. Glomerular was only a minor degree of glomerular macrophage infiltrate at both 1 and 24 h in this murine model, it is difficult to exclude a role for GM-CSF in promoting macrophage-mediated proteinuria as measured from 8 to 24 h after injection of sheep anti-mouse GBM globulin. The autologous model of anti-GBM injury involves quite different effector systems and is dependent on the active generation of DTH to the nephritogenic antigen (12,25). Although this model of injury is associated with development of both circulating antibodies and DTH to the inducing antigen, glomerular injury is dependent only on glomerular DTH effectors (25,33). Mice with agammaglobulinemia consequent to disrup- Figure 5. Dermal delayed-type hypersensitivity responses (DTH) to ␮ tion of the -chain Ig gene developed equivalent injury to that sheep globulin (the nephritogenic antigen) at day 10 of the autolo- observed in animals with an intact humoral response (33). gous-phase of injury, showing antigen specific swelling in CSF WT ϩ Depletion of CD4 cells in the effector phase of injury is mice, unchanged responses in G-CSFϪ/Ϫ , and reduced DTH re- associated with attenuation of skin DTH to the nephritogenic sponses in GM-CSFϪ/Ϫ mice (P Ͻ 0.05 versus CSF WT mice). 356 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 350–358, 2002

are the predominant GM-CSF–producing leukocyte sub population. These cells are essential for the development of GN by their recruitment and activation of macrophages. The normal levels of antibody seen in GM-CSF Ϫ/Ϫ mice would suggest normal T cell help, at least for humoral responses. Thus the beneficial effects of GM-CSF deficiency in GN would appear to be predominantly via diminished T cell–induced macrophage activation. The significant protection afforded to GM-CSF Ϫ/Ϫ mice with autologous GN is consistent with a role for GM-CSF as an important T cell–derived activator of macrophages. The incomplete protection observed in terms of crescent formation suggests partially redundant roles for several cytokines in the Figure 6. Circulating mouse anti-sheep globulin titers in CSF WT, activation of macrophages, including interferon-␥ (25,34) and G-CSFϪ/Ϫ, and GM-CSFϪ/Ϫ mice with autologous-phase injury at macrophage migration inhibitory factor (35,36). Although au- day 10, showing similar levels of serum anti-sheep globulin antibody tologous antibody has little role in glomerular crescent forma- titers in the three groups of mice. tion in this model (33), it has the capacity to induce complement dependent proteinuric injury in the autologous phase of injury (25,37). GM-CSF plays little role in the generation of neutrophil accumulation is unlikely to be a major contributor to humoral responses in this model. the development of crescentic GN in autologous-phase injury. Augmented expression of GM-CSF has been observed in The numbers of glomerular neutrophils were much lower in human proliferative GN in association with leukocyte infiltra- autologous-phase injury than those seen in the heterologous tion (27). The demonstration that mesangial cells have the model. Injury in G-CSF Ϫ/Ϫ mice was similar to that seen in capacity to produce GM-CSF illustrates the potential for local controls, despite their inability to fully develop a neutrophil- production of GM-CSF to contribute to injury by activating dependent (heterologous-phase) form of injury. In contrast to macrophages (38). However, these studies do not address a CSF WT and G-CSF Ϫ/Ϫ mice, GM-CSF Ϫ/Ϫ mice were potential role for GM-CSF in macrophage activation by intrin- unable to mount a normal dermal DTH response to the nephri- sic glomerular cells or in response to Fc␥ receptor engagement. togenic antigen. When autologous-phase anti-GBM GN was These results suggest that both the G-CSF and GM-CSF are induced in these animals, they showed a reduced accumulation important in the generation of leukocyte-induced GN. G-CSF of CD3ϩ T cells and macrophages. There was significant and GM-CSF are both required for normal neutrophil recruit- protection from the development of crescent formation and ment and glomerular localization, whereas G-CSF does not impairment of renal function in GM-CSF Ϫ/Ϫ mice, and appear to be required for cell-mediated immune glomerular proteinuria was reduced but not to normal levels. The levels of injury. In contrast, GM-CSF is required (potentially at several circulating antibody to the immunizing antigen were unaf- points in the immune response in the generation of DTH) for fected by deficiency of GM-CSF, consistent with results else- the development of crescentic GN. Targeting G-CSF as a where in studies of these mice (9). therapeutic potential modality, although theoretically possible, GM-CSF–deficient animals are protected from collagen- would be of more limited therapeutic benefit and may only be induced arthritis, where similar alterations in the injurious effective in those forms of GN in which humoral immunity immune response were observed (9). Levels of antibody, in- associated with neutrophil influx is a predominant immune cluding titers of IgG subclasses, were unaffected, whereas effector mechanism. In addition, concerns exist about attenu- DTH was reduced. Moreover, injection of GM-CSF systemi- ating any potential anti-inflammatory effects of G-CSF, given cally (in collagen arthritis) (9) and locally (by the inoculation our observations of increased glomerular T cell accumulation of GM-CSF–transfected tubular epithelial cells under the cap- in G-CSF Ϫ/Ϫ mice, allied with in vitro studies that have sule of the kidney and MRL/lpr-prone mice) (28) both exac- shown that G-CSF reduces interferon-␥, tumor necrosis factor, erbated cell mediated immune injury in the relevant organs. and inerleukin-1 production (39,40). The results of these stud- These observations are consistent with a role for GM-CSF in ies suggest that targeting of GM-CSF could potentially be of directing cell-mediated immune injury. benefit in all forms of leukocyte-associated GN. The protective effects of GM-CSF deficiency could be acting at several levels. It is known that antigen presentation is Acknowledgments affected by GM-CSF deficiency (8); thus, systemic T cell These studies were supported by grants from the National Health responsiveness may be deficient in these animals. GM-CSF is and Medical Research Council of Australia (NH&MRC) and the produced by T cells to activate monocytes; therefore, deficient Australian Kidney Foundation. Dr. Tipping is an NH&MRC Senior T cell production of GM-CSF is likely to be relevant to the Research Fellow. The DNAX Research Institute, Palo Alto, Califor- diminution of injury observed. GM-CSF may be deficient in nia, is thanked for permission to use the RB6-8C5 monoclonal anti- the target organ where its role as a proinflammatory cytokine body. The assistance of Janelle Sharkey and Paul Hutchinson is may be important in the development of local injury. T cells acknowledged. J Am Soc Nephrol 13: 350–358, 2002 Requirement for GM-CSF and G-CSF in GN 357

References 17. Kitching AR, Tipping PG, Timoshanko JR, Holdsworth SR: 1. Gasson JC: Molecular physiology of granulocyte-macrophage Endogenous IL-10 regulates Th1 responses that induce crescen- stimulating factor. Blood 77: 1131–1145, 1991 tic glomerulonephritis. Kidney Int 57: 518–525, 2000 2. Demetri GD, Griffin JD: Granulocyte colony-stimulating factor 18. Suh KS, Kim BK, Kim KH: Crescentic glomerulonephritis: A and its receptor. Blood 78: 2791–2808, 1991 clinicopathologic analysis of 17 cases with emphasis on glomer- 3. Lieschke GJ, Grail D, Hodgson G, Metcalf D, Stanley E, Cheers ular and interstitial neutrophil infiltration. J Korean Med Sci 14: C, Fowler KJ, Basu S, Zhan YF, Dunn AR: Mice lacking 293–298, 1999 granulocyte colony-stimulating factor have chronic neutropenia, 19. Cockwell P, Brooks CJ, Adu D, Savage COS: Interleukin-8: A granulocyte and macrophage progenitor cell deficiency, and im- pathogenetic role in antineutrophil cytoplasmic autoantibody- paired neutrophil mobilization. Blood 84: 1737–1746, 1994 associated glomerulonephritis. Kidney Int 55: 852–863, 1999 4. Zhan Y, Lieschke GJ, Grail D, Dunn AR, Cheers C: Essential 20. Boyce NW, Holdsworth SR: Hydroxyl radical mediation of roles for granulocyte-macrophage colony-stimulating factor immune renal injury by desferrioxamine. Kidney Int 30: 813– (GM-CSF) and G-CSF in the sustained hematopoietic response 817, 1986 of Listeria monocytogenes-infected mice. Blood 91: 863–869, 21. Schrijver G, Schalkwijk J, Robben JC, Assmann KJ, Koene RA: 1998 Antiglomerular basement membrane nephritis in beige mice. 5. Stanley E, Lieschke GJ, Grail D, Metcalf D, Hodgson G, Gall Deficiency of leukocytic neutral proteinases prevents the induc- JA, Maher DW, Cebon J, Sinickas V, Dunn AR: Granulocyte/ tion of albuminuria in the heterologous phase. J Exp Med 169: macrophage colony-stimulating factor-deficient mice show no 1435–1448, 1989 major perturbation of hematopoiesis but develop a characteristic 22. Shigematsu H: Glomerular events during the initial phase of rat pulmonary pathology. Proc Natl Acad Sci USA 91: 5592–5596, Masugi nephritis. Virchows Arch B Cell Pathol 5: 187–200, 1970 1994 23. Neale TJ, Tipping PG, Carson SD, Holdsworth SR: Participation 6. Inaba K, Inaba M, Romani N, Aya H, Deguchi S, Ikehara S, of cell-mediated immunity in deposition of fibrin in glomerulo- Muramatsu S, Steinman RM: Generation of large numbers of nephritis. Lancet 2: 421–424, 1988 dendritic cells from mouse bone marrow cultures supplemented 24. Cunningham MA, Huang XR, Dowling JP, Tipping PG, Hold- with granulocyte/macrophage-colony stimulating factor. J Exp sworth SR: Prominence of cell-mediated immunity effectors in Med 184: 2185–2196, 1992 “pauci-immune” glomerulonephritis. J Am Soc Nephrol 10: 499– 7. Saunders D, Lucas K, Ismaili J, Wu L, Maraskovsky E, Dunn A, 506, 1999 Shortman K: Dendritic cell development in culture from thymic 25. Huang XR, Tipping PG, Shuo L, Holdsworth SR: Th1 respon- precursor cells in the absence of granulocyte/macrophage colo- siveness to nephritogenic antigens determines susceptibility to ny-stimulating factor. J Exp Med 184: 2185–2196, 1996 crescentic glomerulonephritis in mice. Kidney Int 51: 94–103, 8. Wada H, Noguchi Y, Marino MW, Dunn AR, Old LJ: T cell 1997 functions in granulocyte/macrophage colony-stimulating factor 26. Unanue ER, Dixon FJ: Experimental glomerulonephritis. VI. The deficient mice. Proc Natl Acad Sci USA 94: 12557–12561, 1997 autologous phase of nephrotoxic serum nephritis. J Exp Med 121: 9. Campbell IK, Rich MJ, Bischof RJ, Dunn AR, Grail D, Hamilton 715–725, 1965 JA: Protection from collagen-induced arthritis in granulocyte- 27. Matsuda M, Shikata K, Makino H, Sugimoto H, Ota Z: Glomer- macrophage colony-stimulating factor-deficient mice. J Immunol ular expression of macrophage colony-stimulating factor and 161: 3639–3644, 1998 granulocyte-macrophage colony-stimulating factor in patients 10. Glassock RJ, Cohen AH, Adler SG: Primary glomerular diseases. with various forms of glomerulonephritis. Lab Invest 75: 403– In: Brenner and Rector’s the Kidney, 5th Ed., edited by Brenner 412, 1996 BM, Philadelphia, WB Saunders, 1996, pp 1392–1497 28. Naito T, Yokoyama H, Moore KJ, Dranoff G, Mulligan RC, 11. Schrijver G, Bogman MJ, Assmann KJ, de Waal RM, Robben HC, van Gasteren H, Koene RA: Anti-GBM nephritis in the Kelley VR: Macrophage growth factors introduced into the kid- mouse: Role of granulocytes in the heterologous phase. Kidney ney initiate renal injury. Mol Med 2: 297–312, 1996 Int 38: 86–95, 1990 29. Tipping PG, Huang XR, Berndt MC, Holdsworth SR: A role for 12. Huang XR, Holdsworth SR, Tipping PG: Evidence for delayed P selectin in complement-independent neutrophil- mediated glo- type hypersensitivity mechanisms in glomerular crescent forma- merular injury. Kidney Int 46: 79–88, 1994 tion. Kidney Int 46: 69–78, 1994 30. Tipping PG, Huang XR, Berndt MC, Holdsworth SR: P-selectin 13. Huang XR, Tipping PG, Apostolopoulos J, Oettinger C, D’Souza directs T lymphocyte-mediated injury in delayed-type hypersen- M, Milton G, Holdsworth SR: Mechanisms of T cell-induced sitivity responses: Studies in glomerulonephritis and cutaneous glomerular injury in anti-glomerular basement membrane delayed-type hypersensitivity: Eur J Immunol 26: 454–460, (GBM) glomerulonephritis in rats. Clin Exp Immunol 109: 134– 1996 142, 1997 31. Smith MJ, Koch GL: Differential expression of murine macro- 14. Holdsworth SR, Kitching AR, Tipping PG: Th1 and Th2 T phage surface glycoprotein antigens in intracellular membranes. helper cell subsets affect patterns of injury and outcomes in J Cell Sci 87: 113–119, 1987 glomerulonephritis. Kidney Int 55: 1198–1216, 1999 32. Metcalf D, Robb L, Dunn AR, Mifsud S, Di RL: Role of 15. Kitching AR, Tipping PG, Holdsworth SR: IL-12 directs severe granulocyte-macrophage colony-stimulating factor and granulo- renal injury, crescent formation and Th1 responses in murine cyte colony-stimulating factor in the development of an acute glomerulonephritis: Eur J Immunol 29: 1–10, 1999 neutrophil inflammatory response in mice. Blood 88: 3755–3764, 16. Tipping PG, Kitching AR, Huang XR, Mutch DA, Holdsworth 1996 SR: Immune modulation with interleukin-4 and interleukin-10 33. Li S, Holdsworth SR, Tipping PG: Antibody independent cres- prevents crescent formation and glomerular injury in experimen- centic glomerulonephritis in ␮ chain deficient mice. Kidney Int tal glomerulonephritis. Eur J Immunol 27: 530–537, 1997 51: 672–678, 1997 358 Journal of the American Society of Nephrology J Am Soc Nephrol 13: 350–358, 2002

34. Kitching AR, Holdsworth SR, Tipping PG: IFN-␥ mediates ment membrane glomerulonephritis. J Am Soc Nephrol 8: 1101– crescent formation and cell-mediated immune injury in murine 1108, 1997 glomerulonephritis. J Am Soc Nephrol 10: 752–759, 1999 38. Budde K, Coleman DL, Lacy J, Sterzel RB: Rat mesangial cells 35. Boyce NW, Tipping PG, Holdsworth SR: Lymphokine (MIF) produce granulocyte-macrophage colony-stimulating factor. production by glomerular T-lymphocytes in experimental glo- Am J Physiol 257: F1065–F1078, 1989 merulonephritis. Kidney Int 30: 673–677, 1986 39. Boneberg EM, Hareng L, Gantner F, Wendel A, Hartung T: 36. Lan HY, Bacher M, Yang N, Mu W, Nikolic-Paterson DJ, Metz C, Human monocytes express functional receptors for granulocyte Meinhardt A, Bucala R, Atkins RC: The pathogenic role of macro- colony-stimulating factor that mediate suppression of monokines phage migration inhibitory factor in immunologically induced kid- and interferon-gamma. Blood 95: 270–276, 2000 ney disease in the rat. J Exp Med 185: 1455–1465, 1997 40. Hartung T: Anti-inflammatory effects of granulocyte colony- 37. Huang XR, Holdsworth SR, Tipping PG: Th2 responses induce stimulating factor. Curr Opin Hematol 5: 221–225, 1998 humorally mediated injury in experimental anti-glomerular base-

Access to UpToDate on-line is available for additional clinical information at http://www.jasn.org/