of July 17, 2018.This information is current as

Adriamycin Nephropathy in MiceinGlomerular and Tubulointerstitial Damage

Complement Activation Contributes to Both

Marina Botto and H. Terence CookDaniel Turnberg, Margarita Lewis, Jill Moss, Yuanyuan Xu,

http://www.jimmunol.org/content/177/6/4094doi: 10.4049/jimmunol.177.6.4094

2006; 177:4094-4102; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/177/6/4094.full#ref-list-1

, 5 of which you can access for free at: cites 24 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2006 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on July 17, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Complement Activation Contributes to Both Glomerular andTubulointerstitial Damage in Adriamycin Nephropathy inMice1

Daniel Turnberg,* Margarita Lewis,* Jill Moss,† Yuanyuan Xu,‡ Marina Botto,* andH. Terence Cook2†

Adriamycin nephropathy is a model of focal segmental glomerulosclerosis, characterized by proteinuria and progressive glomer-ulosclerosis and tubulointerstitial damage. In this study, we examined the role of complement in the etiology of adriamycinnephropathy in mice. We used mice deficient in C1q, factor D, C3, and CD59, and compared them with strain-matched controls.C3 deposition occurred in the glomeruli of wild-type mice as early as 48 h following a single i.v. injection of adriamycin. C3-deficient mice developed significantly less proteinuria and less podocyte injury at day 3 postadriamycin than controls, suggestingthat complement is important in mediating the early podocyte injury. At later time points, C3-deficient mice were protected fromglomerulosclerosis, tubulointerstitial injury, and renal dysfunction. Factor D-deficient mice were also protected from renal disease,confirming the importance of alternative pathway activation in this model. In contrast, C1q-deficient mice developed similardisease to controls, indicating that the complement cascade was not activated via the classical pathway. CD59-deficient mice, whichlack adequate control of C5b-9 formation, developed significantly worse histological and functional markers of renal disease thancontrols. Interestingly, although more C9 deposited in glomeruli of CD59-deficient mice than controls, in neither group wastubulointerstitial C9 staining apparent. We have demonstrated for the first time that alternative pathway activation of complementplays an important role in mediating the initial glomerular damage in this in vivo model of focal segmental glomerulosclerosis.Lack of CD59, which regulates the membrane attack complex, led to greater glomerular and tubulointerstitial injury. TheJournal of Immunology, 2006, 177: 4094–4102.

A driamycin (also called doxorubicin) nephropathy is a ro-dent model of focal and segmental glomerulosclerosis(1, 2). It is characterized by rapid onset of glomerular

podocyte damage and glomerular proteinuria, which progresses tosegmental glomerular sclerosis. Accompanying this there is earlytubulointerstitial damage with subsequent tubular atrophy, accu-mulation of myofibroblasts around damaged tubules, and intersti-tial fibrosis. It is of particular interest as it is thought to provide ananimal model of the way in which glomerular proteinuria is asso-ciated with progressive tubulointerstitial scarring, which is thoughtto be a final common pathway in human glomerular disease.

There is increasing evidence that the complement system is im-portant in mediating renal injury in proteinuric diseases. In adria-mycin nephropathy, it has been shown recently that C6-deficientrats, which are unable to form the terminal membrane attack com-plex (MAC)3 of the complement system, are protected from peri-

tubular myofibroblast accumulation and interstitial extracellularmatrix deposition (3). These results have led to the suggestion thatintraluminal activation of the complement cascade leading to thegeneration of the MAC is the principal mediator of progressivetubulointerstitial injury in proteinuric renal diseases whatever thecause of the glomerular injury. However, the role of other com-ponents of the complement cascade in mediating the glomerularand tubulointerstitial changes of adriamycin nephropathy has notbeen defined. We therefore studied adriamycin nephropathy inmice, which allowed us to take advantage of mouse strains withtargeted deletions of other components of the complement system.

Adriamycin nephropathy has only been characterized usingmice of a BALB/c genetic background, and indeed, other com-monly used strains such as C57BL/6 are resistant to disease (1, 2).We therefore backcrossed mice lacking C3, C1q, factor D, or thecomplement regulatory protein CD59a onto the BALB/c geneticbackground and examined the course of the disease in these ani-mals. CD59 is the only membrane-bound factor that prevents theformation of C5b-9 (4, 5). In contrast to humans, mice have twoisoforms of CD59: CD59a, which is widely expressed, and CD59b,which is predominantly limited to the testes (6). We have shownpreviously that mice lacking CD59a are more susceptible to glo-merular immune complex injury (7) and to tubular injury and in-terstitial inflammation following ischemia-reperfusion injury (8).Although there is preliminary evidence that complement compo-nents, especially C3, are deposited in areas of tubulointerstitialinjury in mice with adriamycin nephropathy, the functional role ofthe complement system in the pathogenesis of this injury has notbeen studied previously (9).

In this study, we demonstrate for the first time that the absenceof C3 protects from the initial glomerular injury in adriamycin

*Rheumatology Section, Faculty of Medicine, Imperial College, Hammersmith Cam-pus, London, United Kingdom; †Department of Histopathology, Faculty of Medicine,Imperial College, Hammersmith Campus, London, United Kingdom; and ‡Division ofClinical Immunology and Rheumatology, Department of Medicine, University of Al-abama, Birmingham, AL 35294

Received for publication April 12, 2005. Accepted for publication June 8, 2006.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by Wellcome Trust Grant (071467). D.T. was a recipientof a fellowship from the National Kidney Research Fund.2 Address correspondence and reprint requests to Dr. H. Terence Cook, Departmentof Histopathology, Faculty of Medicine, Hammersmith Campus, Imperial College,Du Cane Road, London, W12 0NN, U.K. E-mail address: t.cook@imperial.ac.uk3 Abbreviations used in this paper: MAC, membrane attack complex; HPF, high-powered field.

The Journal of Immunology

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

nephropathy. We have also shown that complement-mediated in-jury occurs via activation of the alternative, but not the classicalpathway in this model. In addition, inhibition of the membraneattack pathway by its regulatory protein CD59 is important in con-trolling this type of complement-dependent glomerular injury andsubsequent tubulointerstitial injury.

Materials and MethodsAnimals

CD59-deficient (mCd59a�/�), C1q-deficient (C1qa�/�), and C3-deficient(C3�/�), and factor D (Cfd�/�) mice were generated, as described previ-ously (10–13). All gene-targeted animals were then backcrossed for sixgenerations onto a BALB/c genetic background and compared with strain-and age-matched controls. To ensure that C3�/� BALB/c (N6) mice wereof the same H2 haplotype as wild-type BALB/c mice, H2 haplotype waschecked for by PCR. Only female mice were used. Mice were kept in aspecific pathogen-free environment, and experiments were performed ac-cording to institutional guidelines.

Experimental protocol

To induce adriamycin nephropathy, a single i.v. injection of doxorubicin(adriamycin; Pharmacia & Upjohn) diluted 1/2 with 0.9% saline was ad-ministered into the tail vein. For experiments with C3�/�, Cfd �/�, andC1qa�/� mice, a dose of 10 mg/kg adriamycin was used. A lower dose of9 mg/kg was administered in experiments using mCd59a�/� mice, as pre-liminary experiments demonstrated that mCd59a�/� mice became rapidlyunwell with large doses. Mice were killed at 8 wk postadriamycin injectionunless they became ill at earlier time points. In addition, to ascertain thetime course of renal complement deposition, some wild-type BALB/c micewere killed at earlier time points.

Histological analysis

Kidneys were fixed for 24 h in 4% buffered formal saline, transferred to70% ethanol, and embedded in paraffin. Sections were then stained withperiodic acid-Schiff. All analyses were performed blinded to the sampleidentity. Tubulointerstitial damage was assessed by ranking of sectionstaking into account tubular necrosis, tubular dilatation, and cast formationin randomized sections. Glomerulosclerosis was assessed by grading thearea of periodic acid-Schiff-positive sclerosis for each glomerulus, as fol-lows: grade 0 � no sclerosis; grade 1 � 0–25%; grade 2 � 25–50%; grade3 � 50–75%; and grade 4 � 75–100%. The mean glomerular sclerosisscore for 50 glomeruli was then calculated. Paraffin-fixed sections werealso stained with picrosirius red for assessment of collagen deposition. Themean area of positive staining was quantified using Image-Pro Plus soft-ware (Media Cybernetics) of images captured from a Photonic ScienceColor Coolview digital camera (Photonic Science).

Immunofluorescence

Kidneys were snap frozen in isopentane and stored at �70°C. Frozen sec-tions were cut at a thickness of 5 �m. An observer without knowledge ofthe sample identity performed all the quantitative immunofluorescenceanalyses. For C3 staining, a goat anti-mouse C3 FITC-conjugated Ab(Valeant Pharmaceuticals) was used. For C9 staining, a rabbit anti-rat C9primary Ab (1:100) (cross-reactive with mouse C9 (14)) and a FITC-con-jugated mouse anti-rabbit IgG secondary (1:50) (Sigma-Aldrich) wereused. For the type IV collagen staining, a polyclonal rabbit anti-human typeIV collagen primary Ab (1:100) (Research Diagnostics) and a FITC-la-beled mouse anti-rabbit IgG secondary (1:50) (Sigma-Aldrich; F-4151)were used. All incubations were performed for 1 h at room temperature,and all Abs were diluted in PBS. A negative control (PBS) was used ineach experiment to ensure there was no binding of the secondary Ab to thekidney section. In quantitative immunofluorescence studies, to exclude ar-tifacts due to variable decay of the fluorochrome, all sections from oneexperiment were stained and analyzed at the same time. Sections wereexamined at �400 magnification using an Olympus BX4 fluorescence mi-croscope (Olympus Optical). A Photonic Science Color Coolview digitalcamera (Photonic Science) was attached to the microscope and, using Im-age-Pro Plus software (Media Cybernetics), images were captured for anal-ysis. For glomerular analysis 20 glomeruli were examined in each section,and for the tubulointerstitium 10 high-powered field (HPF) of the cortico-medullary area were assessed for each section. In each case, the meanfluorescence intensity was recorded, with results expressed in arbitraryfluorescence units.

Immunohistochemistry

For � smooth muscle actin staining, paraffin-embedded, formalin-fixed sec-tions were used. These were rehydrated and deparaffinized through gradedalcohols. An Ag retrieval step was then performed by heating slides in 0.01M citrate buffer (pH 6) in a microwave oven at 600 W for 20 min. Sectionswere blocked with 5% normal rabbit serum, and a primary monoclonalmouse anti-human smooth muscle actin (clone 1A4; Sigma-Aldrich) (1/50dilution) was then applied. The sections were then processed with theDAKO Envision AP kit (DakoCytomation), following the manufacturer’sinstructions. The kit uses Fast Red as the chromogenic substrate. The quan-tity of smooth muscle actin on renal sections was assessed by rankingsections according to the degree of Fast Red staining. Apoptosis was as-sessed by TUNEL staining using the In Situ Cell Death Detection Kit, POD(Roche Applied Science), following the manufacturer’s instructions, andquantified by counting the number of apoptotic nuclei per HPF averagedover 20 HPF per section.

Electron microscopy

Specimens (1 mm3) of renal cortex were fixed in 3% glutaraldehyde in 0.1M cacodylate buffer for 4 h at 4°C. The samples were then postfixed in 2%aqueous osmium tetroxide, block stained in 2% aqueous uranyl acetate,dehydrated through ascending grades of ethanol, and embedded in Spur’sresin at 70°C overnight. Ultrathin sections were collected on 400 meshcopper grids and stained with 1% aqueous uranyl acetate and Reynold’slead citrate.

Renal function assessment

Serum urea was measured using an Olympus AU600 analyzer (OlympusDiagnostics). Proteinuria was measured by dipstick of urine (Hema-Com-bistix; Bayer) daily for the first week and then weekly. In addition, micewere placed in metabolic cages for 24-h collection of urine to allow moreaccurate assessment of urine albumin concentration. The albumin concen-tration was measured by radial immunodiffusion. Samples and standards(mouse albumin) (Sigma-Aldrich) were placed in wells (4 �l/well) in 1.2%agarose in PBS containing rabbit anti-mouse albumin (Biogenesis). Gelswere dried and stained with Coomassie blue. Albumin concentration wascalculated with reference to a standard curve.

FIGURE 1. Glomerular C3 staining in adriamycin nephropathy. By48 h postadriamycin injection, C3 deposition had occurred in the glomeruliof BALB/c mice (B). The intensity of the staining increased by 5 days (C)and reached a maximum by day 10 (D). There was no glomerular C3deposition in control mice, except the usual murine Bowman’s capsulestaining (A).

4095The Journal of Immunology

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

Statistical analysis

All values described in the text and figures are expressed as median andrange for n observations. Statistical analysis was conducted using Graph-Pad Prism 3.02 (GraphPad). Data were analyzed using a Mann-Whitney Utest for all comparisons, and a p value of �0.05 was considered to besignificant. For survival analysis, a Kaplan-Meier statistic was used and ap value of �0.05 was also considered significant.

ResultsInitial dose-ranging experiments using between 7 and 11 mg/kgadriamycin in wild-type BALB/c female mice suggested that adose of 9–10 mg/kg adriamycin produced significant irreversibleinjury characterized by glomerulosclerosis and tubular injury at 2wk postinjection, followed by interstitial inflammation and fibrosisat later time points (4–6 wk; data not shown). Higher doses pro-duced much more severe injury that developed more rapidly, while

lower doses produced only mild reversible histological abnormal-ities. However, in all experiments, independently of the dose ad-ministered, significant proteinuria was observed (���� on dip-stick) from weeks 1 to 6.

Glomerular C3 deposition occurs by 48 h after adriamycinadministration

To investigate the role of the complement system in mediating theinitial glomerular injury following adriamycin injection, we killedBALB/c mice at various time points to study glomerular comple-ment deposition. Immunofluorescence microscopy revealed that inunmanipulated BALB/c mice kidneys, C3 was present around tu-bules and along Bowman’s capsule, but no C3 was detected inglomeruli (Fig. 1A). However, as early as 48 h postadriamycinadministration, there was sparse glomerular C3 staining (Fig. 1B).This increased in intensity by day 5 (Fig. 1C). At 10 days, bywhich point significant glomerulosclerosis and tubulointerstitialinjury had occurred, intense glomerular C3 staining was present,particularly within more severely diseased glomeruli (Fig. 1D). Inaddition, there was increased C3 deposition in areas of most severetubulointerstitial damage.

C3-deficient mice were protected from early glomerular injury

Guided by these findings, we decided to investigate the role ofcomplement activation in the pathogenesis of adriamycin nephrop-athy by studying the effect of injecting 9 mg/kg adriamycin in

FIGURE 2. Initial glomerular injury following adriamycin administra-tion in C3�/� and control animals. C3�/� mice were protected from initialproteinuria compared with wild-type mice, but albuminuria concentrationsreached a similar maximum value by day 10 postadriamycin administration(A). Electron micrographs at day 3 following adriamycin administrationshow extensive foot process effacement in wild-type animals (B), whilepodocytes were relatively well preserved in C3�/� mice (C).

FIGURE 3. Progressive renal damage following adriamycin adminis-tration in C3�/� and control animals. Quantitative analysis of periodicacid-Schiff-stained sections demonstrated that C3�/� animals were signif-icantly protected from glomerulosclerosis (A) and tubulointerstitial injury(B) compared with matched controls. In addition, C3�/� mice developedless renal dysfunction (measured by serum urea) than strain-matched con-trols (C). The error bars denote SEM.

4096 COMPLEMENT AND ADRIAMYCIN NEPHROPATHY

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

C3�/� animals compared with BALB/c controls. Small amountsof albumin could be detected in the urine of wild-type animals asearly as day 2 postadriamycin injection and reached a maximumlevel by day 6. In contrast, the onset of albuminuria was markedlydelayed in C3�/� mice, but achieved a similar upper limit to thewild-type mice by day 10 after the administration of adriamycin(Fig. 2A). Consistent with these findings, electron microscopy ofglomeruli at day 3 after adriamycin showed that there was betterpreservation of podocyte foot processes in C3�/� mice than wild-type animals (Fig. 2, B and C).

Absence of C3 reduced progressive renal injury

Four weeks postadriamycin injection, C3�/� mice were signifi-cantly protected from chronic glomerular damage (median sclero-sis score 0.8; range 0.3–1.1; n � 8 for C3�/� vs 2.0; 0.8–3.6; n �8 for wild type, p � 0.007; Fig. 3A) and tubulointerstitial injury(median rank of tubular injury 6; range 3–9; n � 8 for wild typevs 2; 1–5; n � 8 for C3�/�, p � 0.004; Fig. 3B). Consistent withthese structural differences, C3�/� animals were protected fromrenal impairment when compared with wild-type controls (medianurea 7.2 mmol/L 5.7–9.1; n � 8 for C3�/� vs 12.8; 10.3–60.0; n �8 for controls, p � 0.0002; Fig. 3C). Histological changes showedsevere, often global sclerosis in wild-type animals (Fig. 4A) withonly relatively minor, segmental abnormalities in C3�/� animals

(Fig. 4B). Concomitantly, wild-type animals displayed severe tu-bular damage typified by loss of the epithelial brush border, flat-tening of tubular cells, and tubular cast formation (Fig. 4C). Incontrast, tubular injury was mild in C3�/� mice (Fig. 4D). In wild-type animals, C3 was deposited especially in areas of most severerenal damage, but also in a peritubular distribution (Fig. 4E). NoC3 staining was seen in C3-deficient mice (Fig. 4F).

Complement activation by the alternative pathway caused renalinjury

C3 activation can occur by the classical, alternative, or lectin path-ways. To dissect out the role of alternative pathway activation, weused mice deficient in the alternative pathway component factor D.Cfd�/� mice were protected from early proteinuria in a similarfashion to C3�/� mice (Fig. 5A), suggesting that alternative path-way activation mediated the early complement-induced podocytedamage. Three weeks after adriamycin administration, Cfd�/�

mice were significantly protected from both glomerular injury(median injury score 3.2; range 0.8–3.5; n � 6 for controls vs 0;0.0–0.8; n � 5 for Cfd�/� mice, p � 0.02; Fig. 5B) and tubulo-interstitial damage (median injury rank 9; range 6–11; for controlsvs 3.5; 1–7; for Cfd�/� mice, p � 0.009; Fig. 5C). In addition,renal function was preserved in Cfd�/� mice as compared withmatched controls (median urea 9.4 mmol/L; 8.6–12.9; for controls

FIGURE 4. Periodic acid-Schiff stain-ing and C3 immunofluorescence of renalsections following adriamycin injection.Glomerular damage 4 wk postadriamycininjection was characterized by segmentalor global sclerosis in wild-type animals(A) compared with either normal appear-ances or mild segmental sclerosis (B) inC3�/� mice. Tubulointerstitial damagein the wild-type animals (C) was exem-plified by tubular cell thinning, loss ofbrush border epithelium, and tubular castformation, while C3�/� mice (D) wereprotected to a significant extent fromthese changes. In wild-type animals, C3was deposited especially in areas of mostsevere renal damage, but also in a peri-tubular distribution (E). C3-deficient an-imals were used as negative controls (F).

4097The Journal of Immunology

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

vs 4.8; 4.2–5.6; for Cfd�/� mice, p � 0.036; Fig. 5D). To confirmthat the absence of factor D protected Cfd�/� mice from renalinjury by preventing C3 activation, we stained renal sections forC3. No C3 immunofluorescence staining was apparent in eitherglomerular or tubulointerstitial compartments in the Cfd�/� mice(Fig. 5F).

Classical pathway activation of complement was not involved inthe pathogenesis of adriamycin nephropathy

To explore possible classical pathway involvement, we examinedadriamycin-induced nephropathy in C1q-deficient mice. The levelsof proteinuria were similar in C1qa�/� and wild-type animals dur-ing the entire period of observation (Fig. 6A). Structural injury 4wk postadriamycin administration was not significantly differentbetween C1qa�/� mice and BALB/c controls in terms of either

glomerulosclerosis (median injury score 2.8; range 0.4–3.2; n � 6for C1qa�/� vs 1.9; 1.1–2.7; n � 7 for controls, p � 0.48; Fig. 6B)or tubulointerstitial injury (median rank of tubular injury 7; range2–12; n � 6 for C1qa�/� vs 6; 1–11; n � 7 for controls, p � 0.76;Fig. 6C). In addition, there were no functional differences in termsof serum urea between the two groups of mice (median urea 12.5mmol/L; 6.6–32.2; n � 6 for C1qa�/� vs 8.4; 7.4–23.1; n � 7 forcontrols, p � 0.6; Fig. 6D).

The terminal complement pathway played an important role inmediating renal damage

To determine the role of the membrane attack pathway in thepathogenesis of adriamycin-induced renal damage, we usedmCd59a�/� mice that lack the major regulator of MAC formation.Despite a similar degree of proteinuria (Fig. 7A), mCd59a�/� mice

FIGURE 5. Markers of renal injury after adriamycin injection in Cfd�/� and control mice. Cfd�/� mice developed less proteinuria measured by dipstickthan wild-type animals after adriamycin administration (A). Renal sections from Cfd�/� animals displayed significantly less glomerulosclerosis (B) andtubulointerstitial injury (C) compared with matched controls. In addition, Cfd�/� mice were protected from renal dysfunction (measured by serum urea)compared with controls (D). Wild-type mouse kidneys showed significant glomerular C3 staining (E). In contrast, C3 staining was notably absent fromCfd�/� mouse kidneys (F). The error bars denote SEM.

4098 COMPLEMENT AND ADRIAMYCIN NEPHROPATHY

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

developed significantly more glomerulosclerosis (median injuryscore 2.8; range 1.1–2.7; n � 8 for mCd59a�/� vs 0.8; 0.1–2.0;n � 9 for controls, p � 0.0078; Fig. 7, B, E, and F) and tubulo-interstitial injury than wild-type animals (median rank of tubularinjury 8; range 4–10; n � 8 for mCd59a�/� vs 3; 1–6; n � 9 forcontrols, p � 0.03; Fig. 7C). As a functional correlate of thesestructural differences, serum urea was significantly greater in themCd59a�/� mice than strain-matched controls (median urea 23.6mmol/L; 19.2–28.0; n � 8 for mCd59a�/� vs 10.3; 8.0–14.3; n �9 for controls, p � 0.0006; Fig. 7D). Interestingly, although glo-meruli from mCd59a�/� mice showed strikingly more C9 depo-sition (as a marker of MAC) than wild-type animals (median C9staining 47.0 arbitrary fluorescence units; 31.2–77.4; n � 8 formCd59a�/� vs 27.1; 7.5–35.2; n � 9 for controls; p � 0.0006),there was very little tubulointerstitial C9 staining in either group.Importantly, mortality in mCd59a�/� mice was significantlygreater than wild-type animals following adriamycin administra-tion (4 of 8 mCd59a�/� mice died compared with 0 of 9 controls;p � 0.01), and therefore, in this series of experiments mice werekilled 3 wk following adriamycin dosing.

To characterize further the development of interstitial fibrosis,we stained sections for smooth muscle actin (as a marker of myo-fibroblast transformation) and with picrosirius red (to detect typesI and III collagen deposition). There was greater picrosirius red(median area of staining 0.02; range 0.005–0.08; n � 8 formCd59a�/� vs 0.008; 0.001–0.03; n � 9 for controls, p � 0.04;Fig. 8A) and smooth muscle actin staining (median score 5; range4–8; n � 8 for mCd59a�/� vs 4; 1–7; n � 9 for controls, p �0.02; Fig. 8B) in mCd59a�/� animals compared with wild-typeanimals. In addition, renal sections of mCd59a�/� mice displayeda significantly greater number of TUNEL-positive apoptotic cellsthan controls (median number of apoptotic cells per HPF 8.3;range 4.7–11; n � 8 for mCd59a�/� vs 3.3; 1.7–7.6; n � 9 forcontrols, p � 0.0003; Fig. 9).

DiscussionThe model of adriamycin nephropathy in rodents has been studiedas a model of proteinuric renal disease in which there are progres-

sive glomerulosclerosis and tubulointerstitial injury. In particular,it has been used to elucidate the role of proteinuria in leading toprogressive tubulointerstitial scarring. In this study, we examinedthe role of the complement system in this model in mice. Weconfirmed previous descriptions on the time course of the renaldisease in this model in BALB/c mice (15) and found that after asingle i.v. dose of adriamycin there was rapid onset of albuminuria,which was first detectable by day 2. Renal sections demonstratedfocal and segmental glomerulosclerosis and subsequent, progres-sive tubulointerstitial scarring. These changes were accompaniedby increased deposition of C3 in glomeruli. We therefore studiedthe course of the model in C3-deficient mice.

Mice lacking C3 showed a reduction in severity of the disease.The earliest manifestation of this was a delay in the onset of de-tectable albuminuria accompanied by a reduction in podocytedamage as assessed by effacement of foot processes. Subsequently,mice without C3 showed less glomerulosclerosis and less tubulo-interstitial scarring and, functionally, this was reflected by betterrenal function with a lower serum urea. The classical pathway ofcomplement activation was not involved in initiating this comple-ment-dependent injury, as the absence of C1q did not alter diseaseseverity. To investigate whether complement-mediated damageensued from alternative pathway activation, we studied mice de-ficient in the alternative pathway component factor D. We dem-onstrated that Cfd�/� mice were protected from early proteinuria,and from later glomerular and tubulointerstitial injury. This wasalso manifested functionally as an improvement in renal functionas measured by serum urea. It would have been interesting to studymice deficient in factor B. However, the factor B gene lies onchromosome 17 in the same region as the MHC, and therefore,backcrossing factor B-deficient mice (that are H2b) onto a BALB/cbackground (H2d) would have meant comparing mice with differ-ent H2 haplotypes, which could have confounded the resultsthereby obtained.

We hypothesize that adriamycin causes initial toxic injury to thepodocyte, and then the damaged cells are able to activate the al-ternative pathway possibly because of reduced surface expressionof complement-inhibitory molecules. The deposition of C3 then

FIGURE 6. Markers of renal injuryafter adriamycin injection in C1qa�/�

and control mice. At 3 wk after adria-mycin administration, there were nodifferences in proteinuria measured bydipstick between C1qa�/� and wild-type mice (A). Additionally, no signif-icant differences in either glomerular(B) or tubulointerstitial injury (C) or inserum urea (D) were demonstrated be-tween C1qa�/� animals and controls.

4099The Journal of Immunology

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

exacerbates podocyte injury, leading to earlier onset of proteinuriaand subsequently more severe glomerular injury and sclerosis.Once activated, the complement system may mediate renal damagethrough either the leukocyte chemoattractant effect of C5a or viaMAC. To investigate further the role of the membrane attack path-way, we used mCd59a�/� mice and demonstrated that the lack ofCD59, and therefore unregulated MAC deposition, significantlyexacerbated renal disease in terms of glomerulosclerosis, tubulo-interstitial injury, and serum urea. We have shown previously ex-acerbation of glomerular injury in nephrotoxic nephritis in micelacking CD59 (16), and the present results emphasize further theimportant role that CD59 plays in controlling glomerular injurydue to complement activation.

In parallel with the effects on glomerular injury, we found thattubulointerstitial injury was reduced in mice lacking C3 and ex-acerbated in mice without CD59, implying an important role forthe MAC. Mice lacking CD59 showed increased tubulointerstitialapoptosis, as demonstrated by TUNEL staining, increased inter-stitial collagen deposition, and increased interstitial myofibroblastsidentified by � smooth muscle actin staining. The pathogenicmechanisms that link proteinuric renal disease and progressive tu-bulointerstitial fibrosis are of considerable interest and have beenreviewed recently (17). There is a clear relationship between thedegree of proteinuria and the severity of interstitial disease in ro-

dent models, such as protein-overload proteinuria and puromycinaminonucleoside nephritis. Studies using these models have sug-gested that the proteinuria itself could induce the progressive in-terstitial injury, and several hypotheses have been proposed as po-tential mechanisms of proteinuria-induced tubulointerstitial injury.These include direct tubulotoxicity of high protein concentrations(18) or that specific proteins such as growth factors, lipoproteins(19, 20), transferrin (21), or activated complement components(22, 23) may be damaging. Recent studies have emphasized therole of the MAC. Rangan et al. (3) examined adriamycin nephrop-athy in C6-deficient rats that were unable to form MAC. Theyfound that MAC deposition occurred predominantly on the luminalsurface of proximal tubules in wild-type C6-sufficient rats. Despitea similar level of proteinuria and glomerular injury, C6-deficientrats had fewer peritubular myofibroblasts than controls. These datasuggest that proteinuria-induced intraluminal formation of C5b-9is responsible for myofibroblast formation, and thereby contributesto progressive interstitial damage and renal failure in proteinuricglomerular disease. An alternative hypothesis for the tubulointer-stitial injury in focal and segmental glomerulosclerosis is thatpodocyte injury causes adhesion of the glomerular tuft to Bow-man’s capsule, followed by accumulation of misdirected filtratearound the origin of the proximal tubules with subsequent occlu-sion and tubular atrophy (24). These two hypotheses of tubular

FIGURE 7. Markers of renal injuryafter adriamycin injection in mCd59a�/�

and control mice. At 3 wk after adriamy-cin administration, there were no differ-ences in proteinuria between mCd59a�/�

and wild-type mice (A). Renal sectionsfrom mCd59a�/� animals displayed sig-nificantly greater glomerulosclerosis (B)and tubulointerstitial injury (C) com-pared with matched controls. In addition,mCd59a�/� mice developed greater re-nal dysfunction (measured by serumurea) than controls (D). The wild-typemouse kidneys showed minor tubular in-jury (E). In contrast, mCd59a�/� mousekidneys displayed severe tubular damageand cast formation (F). The error bars de-note SEM.

4100 COMPLEMENT AND ADRIAMYCIN NEPHROPATHY

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

damage due to proteinuria and misdirected filtration with subse-quent nephron loss are not mutually exclusive.

Our study does not allow us to determine which mechanismleads to tubulointerstitial scarring in adriamycin nephropathy inmice. However, it is notable that, although we found markedlyincreased staining for C9 in the glomeruli of mice lacking CD59,we found very little C9 staining associated with tubules in wild-type or CD59-deficient mice, which argues against a major role oftubular MAC deposition in causing direct tubular injury, but rathersuggests that in the absence of CD59 there is increased glomerularMAC leading to increased glomerular scarring, and this then leadsto tubulointerstitial scarring by a process not dependent on tubularcomplement activation.

Clearly, there are significant differences between our model andthe model of adriamycin nephropathy in the rat described by Ran-gan et al. (3), in which tubular MAC was detected. In their model,no C3 was found in glomeruli, whereas we observed clear C3deposition that increased with time. The reason for these differ-ences is not clear and may reflect either a species difference ordifferences in the dose of adriamycin and severity of the model.They found only a mild glomerulosclerosis, but no difference inthe progression of glomerulosclerosis in rats lacking C6, suggest-ing that MAC did not play a role in the glomerular lesions, whereas

our results in CD59-deficient mice clearly show that if MACactivity is uncontrolled, then this leads to increasedglomerulosclerosis.

In conclusion, we have demonstrated an important functionalrole of the complement system in mediating injury in adriamycinnephropathy. We have shown that lack of C3 reduces early glo-merular injury and proteinuria and ameliorates subsequent glomer-ular and tubulointerstitial scarring with preservation of renal func-tion. We have demonstrated for the first time that lack of afunctional alternative pathway markedly reduces both histologicaland functional markers of renal injury in this model. We have alsoshown that CD59, by limiting deposition of the MAC, protects theglomerulus from scarring. CD59 also limits tubulointerstitial scar-ring, although whether this is indirect via the effect on glomerularscarring or a specific role for tubular CD59 is unclear.

AcknowledgmentsWe thank all of the staff in the animal facility for their technical assistance.We thank Ian Shore (Charing Cross Hospital, London, U.K.) for his helpwith the electron microscopy. We express our appreciation to YouhongZhang, Gregory A. Skibinski, and Mark A. McCrory for their technicalassistance.

FIGURE 8. Picrosirius red and � smooth muscle ac-tin staining of renal sections following adriamycin in-jection. mCd59a�/� mice developed significantly moreinterstitial fibrosis as measured by picrosirius red stain-ing (A) and more smooth muscle actin deposition (B)than wild-type animals. Normal smooth muscle actinstaining of blood vessels is indicated by filled arrows.The error bars denote SEM.

4101The Journal of Immunology

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

DisclosuresThe authors have no financial conflict of interest.

References1. Fogo, A. B. 2003. Animal models of FSGS: lessons for pathogenesis and treat-

ment. Semin. Nephrol. 23: 161–171.2. Wang, Y., P. Y. Wang, Y. C. Tay, and D. C. H. Harris. 2000. Progressive adria-

mycin nephropathy in mice: sequence of histologic and immunohistochemicalevents. Kidney Int. 58: 1797–1804.

3. Rangan, G. K., J. W. Pippin, and W. G. Couser. 2004. C5b-9 regulates peritubularmyofibroblast accumulation in experimental focal segmental glomerulosclerosis.Kidney Int. 66: 1838–1848.

4. Davies, A., and P. J. Lachmann. 1993. Membrane defense against complementlysis: the structure and biological properties of CD59. Immunol. Res. 12:258–275.

5. Meri, S., B. P. Morgan, A. Davies, R. H. Daniels, M. G. Olavesen, H. Waldmann,and P. J. Lachmann. 2005. Human protectin (CD59), an 18,000–20,000 MWcomplement lysis restricting factor, inhibits C5b-8 catalyzed insertion of C9 intolipid bylayers. Immunology 71: 1–9.

6. Qin, X., T. Miwa, H. Aktas, M. Gao, C. Lee, Y. M. Qian, C. C. Morton,A. Shahsafaei, W. C. Song, and J. A. Halperin. 2001. Genomic structure, func-tional comparison, and tissue distribution of mouse Cd59a and CD59b. Mamm.Genome 12: 582–589.

7. Turnberg, D., M. Botto, J. Warren, B. P. Morgan, M. J. Walport, and H. T. Cook.2003. CD59a deficiency exacerbates accelerated nephrotoxic nephritis in mice.J. Am. Soc. Nephrol. 14: 2271–2279.

8. Turnberg, D., M. Botto, M. Lewis, W. Zhou, S. H. Sacks, B. P. Morgan,M. J. Walport, and H. T. Cook. 2004. CD59a deficiency exacerbates ischemia-reperfusion injury in mice. Am. J. Pathol. 165: 825–832.

9. Abe, K., T. Springall, S. H. Sacks, and N. S. Sheerin. 2001. The role of com-plement in the regulation of renal fibrogenesis in adriamycin nephropathy. Mol.Immunol. 38: 77 (Abstr.).

10. Holt, D. S., M. Botto, A. E. Bygrave, S. M. Hanna, M. J. Walport, andB. P. Morgan. 2001. Targeted deletion of the CD59 gene causes spontaneousintravascular hemolysis and hemoglobinuria. Blood 98: 442–449.

11. Botto, M., C. Dell’Agnola, A. Bygrave, E. M. Thompson, H. T. Cook, F. Petry,M. Loos, P. P. Pandolfi, and M. J. Walport. 1998. Homozygous C1q deficiencycauses glomerulonephritis associated with multiple apoptotic bodies. Nat. Genet.19: 56–59.

12. Wessels, M. R., P. Butko, H. B. Warren, A. L. Lage, and M. C. Carroll. 1995.Studies of group B streptococcal infection in mice deficient in complement com-ponent C3 or C4 demonstrate an essential role for complement in both innate andacquired immunity. Proc. Natl. Acad. Sci. USA 92: 11490–11494.

13. Xu, Y., M. Ma, G. C. Ippolito, H. W. Schroeder, Jr., M. C. Carroll, andJ. E. Volanakis. 2001. Complement activation in factor D-deficient mice. Proc.Natl. Acad. Sci. USA 98: 14577–14582.

14. Matsuo, S., H. Nishikage, F. Yoshida, A. Nomura, S. J. Piddlesden, andB. P. Morgan. 1994. Role of CD59 in experimental glomerulonephritis in rats.Kidney Int. 46: 191–200.

15. Wang, Y., Y. P. Wang, Y. C. Tay, and D. C. Harris. 2000. Progressive adriamycinnephropathy in mice: sequence of histologic and immunohistochemical events.Kidney Int. 58: 1797–1804.

16. Turnberg, D., M. Botto, J. Warren, B. P. Morgan, M. J. Walport, and H. T. Cook.2003. CD59a deficiency exacerbates accelerated nephrotoxic nephritis in mice.J. Am. Soc. Nephrol. 14: 2271–2279.

17. Zandi-Nejad, K., A. A. Eddy, R. J. Glassock, and B. M. Brenner. 2004. Why isproteinuria an ominous biomarker of progressive kidney disease? Kidney Int.66(Suppl. 92): S76–S89.

18. Ledingham, J. G. 1990. Tubular toxicity of filtered proteins. Am. J. Nephrol.10(Suppl. 1): 52–57.

19. Thomas, M. E., and G. F. Schreiner. 1993. Contribution of proteinuria to pro-gressive renal injury: consequences of tubular uptake of fatty acid bearing albu-min. Am. J. Nephrol. 13: 385–398.

20. Kees-Folts, D., J. L. Sadow, and G. F. Schreiner. 1994. Tubular catabolism ofalbumin is associated with release of an inflammatory lipid. Kidney Int. 45:1697–1709.

21. Alfrey, A. C., D. H. Froment, and W. S. Hammond. 1989. Role of iron in thetubulo-interstitial injury in nephrotoxic serum nephritis. Kidney Int. 36: 753–759.

22. Biancone, L., S. David, V. Della Pietra, G. Montrucchio, V. Cambi, andG. Camussi. 1994. Alternative pathway activation of complement by culturedhuman proximal tubular epithelial cells. Kidney Int. 45: 451–460.

23. David, S., L. Biancone, C. Caserta, B. Bussolati, V. Cambi, and G. Camussi.1997. Alternative pathway activation induces proinflammatory activity in prox-imal tubular epithelial cells. Nephrol. Dial. Transplant. 12: 51–56.

24. Kriz, W. 2003. The pathogenesis of ‘classic’ focal segmental glomeruloscle-rosis: lessons from animal models. Nephrol. Dial. Transplant. 18(Suppl. 6):vi39 –vi44.

FIGURE 9. TUNEL staining of renal sections following adriamycin ad-ministration. There were significantly more TUNEL-positive cells seen inthe tubulointerstitium of mCd59a�/� animals compared with matchedcontrols.

4102 COMPLEMENT AND ADRIAMYCIN NEPHROPATHY

by guest on July 17, 2018http://w

ww

.jimm

unol.org/D

ownloaded from