Kidney International, Vol. 36 (1989), pp. 1069—1077

Classical pathway of complement activation in normal anddiseased human glomeruli

JORG ZWIRNER, ERICH FELBER, VOLKER HERZOG, GERT RIETHMULLER,and HELMUT E. FEUCHT

Institut fur Immunologie, Inst itut für Zeilbiologie und Medizinische Klinik Innenstadt, Universität München, Munich, Federal Republic ofGermany

Classical pathway of complement activation in normal and diseasedhuman glomeruli. Monoclonal antibodies (mAb) reactive against com-plement components involved in the classical activation pathway wereapplied in an indirect immunoperoxidase technique for the histologicalstudy of normal and diseased human renal tissues. Prominent stainingwith antibodies against the C4d fragment was seen in all glomeruli andsome renal arteriolar walls. The C4d staining was mesangial with lightmicroscopy, whereas the subendothelial site of the glomerutar base-ment membrane (GBM) also appeared to be positive in immunoelectronmicroscopy. In similar localization, albeit with distinctly weaker inten-sity, 1gM and C4 binding protein (C4bp) were detected. In kidneybiopsies from patients with various types of glomerulonephritis, C4dreactive antibodies stained the glomerular structures in a strong, diffuseor granular pattern in contrast to the more segmental distribution andweaker staining intensity in normal kidney specimens. Increasedamounts of C4d, occasionally also of C4b, were paralleled in diseasedkidney tissues by distinct deposits of 1gM and/or IgG in the presence ofC4bp. This study suggests that the C4d fragment in normal humanglomeruli is indicative of a continuous, local complement activation viathe classical pathway induced by the physiological deposition of 1gM-containing immune complexes.

Complement activation in glomeruli is supposed to result intissue damage by formation of the membrane attack complex,enhancement of phagocytosis and promotion of the inflamma-tory response by the release of anaphylatoxins [1—31. Comple-ment can be activated either via the classical pathway byantibody-coated targets, by immune complexes (ICs) or via thealternative pathway in the absence of antibody [4, 5]. Bothpathways result in C3 activation and covalent binding of C3bfragments to target structures via their internal thiolester 16, 7].The characteristic step in classical pathway activation is thecovalent binding of reactive C4b fragments to activating orclosely neighboring structures on ICs or antibody-coated tar-gets. Regulatory proteins such as complement receptor type 1,decay accelerating factor, glycoprotein 45-70, C4bp, factor H orfactor I control the complement cascade by cleavage of acti-vated complement components 18].

Recently, the demonstration of complement components in

Received for publication August 23, 1988and in revised form May 15, 1989Accepted for publication July 6, 1989

© 1989 by the International Society of Nephrology

diseased human kidneys has been extended to C3d fragmentand C5b-9 complex deposition [9, 10]. However, the detectionof these complement activation products also in non-diseasedkidney tissue without the simultaneous deposition of ICs [11]raised new questions concerning the mechanism of complementactivation and its function in mediating tissue damage. Inaddition, the presence of both isotypic components of comple-ment C4, C4A and C4B, in normal human glomeruli and somerenal interstitial arterioles has been observed [12]. In thepresent study, we were interested in further characterization oforigin and function of the glomerular C4 compound. We applieda panel of mono- and polyclonal antibodies on cryostat sectionsof normal and diseased human kidney tissues. We demonstratethat: 1) the glomerular C4 compound can be characterized asthe C4d fragment that is covalently bound in the mesangium andalong the glomerular basement membrane; and 2) low amountsof 1gM and C4bp parallel glomerular C4d deposition. Thehypothesis of a continuous, local complement activation via theclassical pathway induced by physiological 1gM-complex dep-osition in normal human glomeruli is corroborated by the studyof diseased kidney tissue with glomerulonephritis.

Methods

Monoclonal antibodies against complement component C4

Monoclonal antibodies (mAb) were raised in female BALB/cmice by immunization with affinity-purified human C4 as de-scribed in detail elsewhere [13]. The reactivity of the antibodieswas determined in a C4-ELISA and by Western-blot analysis.Thus, antibody M4a1 is reactive with the C4a (a1—) fragment,M4c 1 with the p-chain, M4c3 with the y-chain and M4d1 withthe C4d (a2—) portion of complement C4 [13]. The reactivity ofadditional antibodies M4c2 (p-chain), M4d2 and M4d3 (a2—portion) is shown below.

Commercially purchased antibodies

The following antibodies have been purchased and applied inindirect immunoperoxidase staining and/or immunoblotting:two different rabbit antisera to human C4 (Dakopatts, Ham-burg, FRG; Calbiochem-Behring, Frankfurt, FRG), goat anti-serum to human C4 (Atlantic Antibodies, Scarborough, USA),mAb to C4d, mAb to C4c and mAb to C4bp (Cytotech, SanDiego, California, USA), polyclonal antiserum to C4bp (Calbi-

1069

1070 Zwirner et a!: C4d in human glomeruli

ochem), two mAbs reactive against human IgG and 1gM,respectively (Dianova, Hamburg, FRG), mAb to 1gM (Nordic,Hamburg, FRG), polyclonal antisera against human IgG or 1gM(Dakopatts). Peroxidase coupled antisera to rabbit or mouse Ig(Dakopatts) were applied as second antibodies in indirectimmunoperoxidase and immunoblotting reactions.

Isolation of C4 and its cleavage fragments byimmunoprecipitation

Native C4 and its split product C4d were isolated using amodified immunomatrix as described in detail elsewhere [13,14]. In short, native C4 protein was precipitated in humanEDTA-plasma with rabbit antiserum against human C4 (Dako-patts) coupled to protein A-Sepharose (PAS) beads (Pharmacia,Freiburg, FRG). C4 protein was eluted with 50 mi dieth-ylamine (Sigma) and neutralized with 0.5 M NaH2PO4. Contam-inating immunoglobulins were removed by PAS and the purityof C4 was assessed by SDS-PAGE. The C4d fragment wasgenerated in human serum by incubation with heat aggregatedIgG [15]. C4d was likewise immunoprecipitated in activatedserum by a sandwich complex consisting of PAS beads, rabbitantiserum to mouse Ig (Dakopatts) and ascites fluid of thea2-specific mAb M4d 1. The C4a fragment was generated bycleaving affinity-purified C4 with the C1 component of Cl. TheCl fraction was isolated by affinity chromatography accordingto the method of Assimeh, Bing and Painter [161.

SDS-PAGE and immunoblotting

SDS-PAGE was performed according to the method ofLaemmli [17]. To reduce proteins, samples were boiled forthree minutes in the presence of 2% 2-ME (Sigma). Afterwards,the proteins were blotted onto nitrocellulose paper (Schleicherand Schuell, Dassel, FRG) in 25 mM Tris/HCI, ph 9.0, 192 mMglycine, 20% (vol/vol) methanol for two hours at 30 V at roomtemperature [18]. The nitrocellulose paper was washed andincubated in a first step with adequately diluted antisera orhybridoma supernatants. Then peroxidase conjugated secondantibody, either rabbit anti-mouse 1g. swine anti-rabbit 1g. orrabbit anti-goat Ig (Dakopatts) was added. The peroxidaseactivity was assayed with nitroprusside-Na (Sigma), o-dianisi-dine (Sigma) and H202 (Merck, Darmstadt, FRG).

Tissue specimens

Morphologically normal, adult renal tissue was obtained fromunaffected parts of 14 nephrectomies performed for localizedrenal carcinoma. Samples were immediately (2 to 10 mm afterorgan resection) snap-frozen in liquid nitrogen and stored at—80°C until preparation of cryostat sections. Intraoperativekidney biopsies were taken from two children, four months andfive years old, respectively, who suffered from obstructiveureteral stenosis. One renal tissue probe was obtained from a22-week-old embryo due to legal abortion. All these samplesappeared normal on light microscopic examination and, withthe exception of one nephrectomy sample, did not demonstrateIC deposition on immunofluorescence examination. Percutane-ous kidney biopsies from 40 adult patients suffering fromglomerulonephritis were obtained between January and April1986. All biopsies were evaluated and characterized by directimmunofluorescence carried out with FITC-labeled polyclonal

antisera to human IgG, 1gM, IgA, C3, C4, albumin and fibrin-ogen [191. These immunohistochemical results will not bedescribed in this paper.

Immunoperoxidase microscopyImmunoperoxidase staining was performed with modifica-

tions as described in detail elsewhere [20]. In brief, 5 mcryostat sections were cut using a Frigocut Mod. 2700 (Re-ichert-Jung, NuBloch, FRG). Sections were air-dried for fivehours followed by a three minute fixation in cold acetone. Theywere then incubated with appropriately diluted first antibodyfor 45 minutes. After a washing step in PBS the peroxidase-conjugated second antibody was incubated for another 30minutes. After final washing in PBS the sections were stainedwith 3-amino-9-ethylcarbazole (Sigma), DMSO (Merck) andH202 in PBS. Counterstaining was performed in hemalaun(Merck), As a control, sections were incubated with non-immune rabbit IgG and with supernatants from irrelevantmouse hybi-idomas.

Immunoelectron microscopyNormal parts from three nephrectomies were examined.

Whole glomeruli were isolated from freshly obtained ice-cooledrenal cortex by subsequent sieving through meshes of differentpore size [21]. Glomeruli were fixed at room temperature in 0.1M cacodylate buffer pH 7.4 containing 2% formaldehyde and0.1% glutaraldehyde for five minutes. Longer fixation periodand increased concentration of fixative resulted in loss ofantigenicity. Glomeruli were washed in PBS and 'quenched' in1 mM NH4CI followed by a single freezing-thawing step. Alter-natively, small renal cortex specimens were immediately fixedfor five minutes (see above), washed, 'quenched' and cut into20 m sections. Both preparations were then incubated withoptimal dilutions of mAbs reactive against C4d (M4d1) and C4cfragment (M4c1) for 20 hours at room temperature. Afterwashing in PBS overnight, sections were incubated with per-oxidase-conjugated rabbit antiserum to mouse Ig (Dakopatts)for another 20-hour incubation period. After a final washingstep of 24 hours, peroxidase activity was revealed by reactionwith diaminobenzidine (Sigma) and H202 [22]. Sections werepostfixed with 1% unbuffered 0S04, dehydrated and flat embed-ded in Epon. Ultrathin sections were cut using a ReichertUltracut E and examined with a Siemens Elmiskop I or 102(Siemens, München, FRG).

Elution of complement fragments from tissue sectionsElution experiments were performed applying; a) 0.05 M

citrate buffer pH 3.3; b) 0.5 M, 1.054 or 1.5 M NaCI solutions onsections from normal and diseased kidney tissue. The sectionswere incubated for 30 minutes at room temperature with a finalwashing step in PBS, followed by the described indirect immu-noperoxidase staining procedure.

Results

Reactivity of antibodies against C4Reactivity of mAbs against the a-, /3- and y-chain of C4 and

the cleavage fragment C4d (a2) is demonstrated in Figure 1A.Antibodies M4dl, M4d2, M4d3 as well as a commercial mAbanti-C4d reacted with the a-chain and the C4d fragment.

Zwirner et a!: C4d in human glomeruli 1071

S

a3, a4, /3, y) and C4a (a1) is demonstrated in Figure lB. MAbM4dl reacted with whole C4 and with C4b. MAb M4a1 reactedwith whole C4 and the C4a-fragment (* represents a bandproduced by contaminating C4c' (/3, a1, a3, a4, y). The polyva-lent goat anti-C4 also reacted with C4a.

Binding sites of mAbs on the C4 molecule are schematicallyillustrated in Figure 2.

Immunoperoxidase microscopyImmunohistological staining results are schematically dem-

onstrated in Figure 3.

Normal kidney tissue

Fig. 1. Reactivity of different anti-C'4 antibodies determined by West-ern blot analysis. (A) A mixture of immunoprecipitated C4 and C4d wasanalyzed in SDS-PAGE (10%) under reducing conditions. After blot-ting, identical lanes were incubated with rabbit antiserum to C4(Dakopatts) or the different C4 specific mAbs (lane 1: polyvalent goatanti-C4; 2: polyvalent rabbit anti-C4; 3: M4d1; 4: M4d2; 5: M4d3; 6:commercial mAb anti-C4d; 7: M4c1; 8: M4c2; 9: commercial mAbanti-C4c; 10: M4c3). (B) Immunoprecipitated C4 was cleaved by Cisand separated on a 3.5 - 20% SDS-PAGE under non-reducing condi-tions. After blotting the lanes were incubated with: lane I: polyvalentgoat anti-C4; lane 2: M4d1; lane 3: M4a1 [* represents C4c' (/3, a, a3,a4,

Antibodies M4c1, M4c2 and a commercial mAb anti-C4c re-acted with the /3-chain. Antibody M4c3 reacted exclusively withthe y-chain. It was shown by cross-blocking experiments thata2- and /3-specific mAbs reacted with different epitopes (datanot shown). Two commercial polyvalent antisera against C4(goat anti-C4; rabbit anti-C4) reacted with the a-, /3- andy.chain; they did not, however, recognize the C4d fragment.The reactivity of mAbs against the cleavage fragments C4b (a2,

C4 reactive antibodies: Application of /3- or y-chain specificmAbs and of commercially available antisera to human C4failed to stain 13 out of 14 normal human kidney sections (Fig.4C). In addition, no C4a antigen was detected with mAb M4a1.Exclusively C4d fragment specific mAbs (M4d1, M4d2, M4d3,commercially available mAb anti-C4d) stained all human gb-meruli and some renal arteriolar walls in the examined normalkidney tissues (Fig. 4A). Occasionally, there was predominantstaining of arteriolar parts of the vascular pole. The pattern ofpositive gbomerular staining reaction could be described asgranular and segmental in mesangial and capillary areas.Whether the positive staining reaction of arteriolar walls be-longed to vascular poles of adjacent glomeruli could not beassessed. Other vascular elements as well as constituents of thetubular system were, however, consistently negative as de-scribed previously [12].

C4bp reactive antibodies. In 12 out of 14 cases antigens ofC4bp were observed in similar glomerular localization to C4dfragment deposition applying poly- and monoclonal anti-C4bpantibodies (Fig. 4B). Compared to the C4d deposition, thestaining intensity for C4bp was clearly weaker. Comparativestudy with mono- and polyclonal antibodies against C4bprevealed lower sensitivity of the latter in antigen detection. Intwo cases only scarce glomerular distribution of C4bp occurredthat could not be detected by the polyclonal reagent.

1gM reactive antibodies. Application of two mAbs resulted insimilar staining patterns. Nine out of 14 tissue probes displayeda weak to moderate 1gM antigen deposition comparable to C4d

M4c1,2

Li

M4a1 S

I• 13

M4d1,2,3

ti r-t ' I

Va, A a3 a ' jI

a4

SM4c3

I IFig. 2. Schematic diagram of the molecular structure of human C4protein. Arrows indicate the epitopes recognized by representativemAbs with reactivity against C4 (* represents the internal thiolesterbond which mediates covalent linkage of the C4b/d fragment to targetmolecules).

• i in —

—a

— C4d

—a

1 2 3 4 5 6 7 8 9 10B

kD

200

150" *

97

68

43.

29

12

1072 Zwirner et a!: G4d in human glomeruli

demonstrated in kidney sections from two children, fourmonths and five years old.

Diseased human kidney tissue

Fig. 3. Co-deposition of 4d and C4b with C4bp, 1gM and iga intissue specimens from normal human kidneys (A) and from variousglomerulonephritides (B-E). Range of staining intensity: 0 absent, 1+moderate, 2+ moderate to strong, 3+ strong to 4+ very strong; therewas intermediate rating if indicated ('/2+ weakly positive, etc.) Ciphersunder the columns indicate the number of cases or the age of thepatient. I indicates the range of variability when 2 cases or more aresummarized. Abbreviations are: SLE, systemic lupus erythematosus;MPGN, membranoproliferative glomerulonephritis; MGN, mesangialproliferative glomerulonephritis; GPN, Goodpasture nephropathy;RPGN, rapidly progressive glomerulonephritis; FGS, focal glomerularsclerosis; MN, membranous nephropathy; MCNS, minimal-changenephrotic syndrome; SHN, SchOnlein-Henoch nephropathy. Symbolsare: (1111) C4d; (fl) C4bp; () C4b; (Lfl 1gM; (D) IgG.

and C4bp in its glomerular localization. The amount of antigenwas similar to C4bp or less (Fig. 4D). When the polyclonalantiserum against 1gM was used, only four sections could bejudged as positive. Hereby, evaluation was complicated by adistinct staining background that was not seen with the mono-clonal reagents. It should be noted that the direct immunofiuo-rescence examination of 14 normal kidney probes had consis-tently been negative for 1gM with the exception of one section,being modestly 1gM positive. In this case we were able todemonstrate increased deposition of C4d, C4bp, 1gM and eventraces of C4b applying the mAbs. The 22-week-old embryonicrenal tissue displayed negative histological staining results.

Interestingly, distinct glomerular deposition of C4d and to aless extent of 1gM and C4bp antigens, respectively, could be

Biopsies from 40 patients with different types of glomerulo-nephritis were examined. The following diagnoses (based onclinical, immunomorphological and histopathological findings)had been raised: minimal-change nephrotic syndrome (MCNS,N = 2), membranoproliferative glomerulonephritis type I(MPGN, N = 1), SLE (systemic lupus erythematosus) nephrop-athy (N = 7), focal glomerular sclerosis (FGS, N = 5),IgA-nephritis (N = 13), mesangial proliferative glomerulone-phritis (MGN, N = 5), Schönlein-Henoch nephropathy (SHN,N = 2), membranous nephropathy (MN, N = 1), rapidlyprogressive glomerulonephritis (RPGN, N = 2), and Goodpas-ture nephropathy (GPN) (N = 2). Sections were investigated fordeposition of C4c (by mAb M4c1), C4d (mAb M4d1), C4bp(mAb, Cytotech) and concomitant IgM/G (mAbs purchasedfrom Dianova) by indirect immunoperoxidase microscopy. C4aantigen could never be detected as assessed by mAb M4a1.Positive staining of mAb M4d1 in the absence of C4a and C4cindicated deposition of C4d. In contrast, the presence of C4cand C4d in the absence of C4a strongly suggested deposition ofC4b.

IgA nephritis (N = 13). Eight cases displayed modest C4ddeposition comparable to normal kidney tissue. All of thesewere accompanied by C4bp deposits and in six by low amountsof 1gM, and one biopsy additionally displayed IgG antigens. Infour patients, C4b and increased quantities of C4d, respec-tively, were detected in the presence of distinct C4bp and 1gMdeposits. Finally in one case increased quantities of C4d and1gM were paralleled by only weak staining for C4bp in theabsence of C4b.

SLE nephropathy (N = 7; Fig. 5A, B, C). In all examinedcases the presence of C4b was accompanied by deposition ofIgG and/or 1gM and distinct amounts of C4bp.

For the immunohistological characterization of MGN (Fig. 5D, E, F), GPN, RPGN, FGS, MN (Fig. 6A, B, C), MCNS,SHN and MPGN (Fig. 6D, E, F), refer to Figure 3. It isnoteworthy that C4b deposition was accompanied by demon-stration of 1gM and/or IgG regardless of diagnosis, stainingintensity and antigen localization. This is not surprising sinceICs are thought to be causative for the activation of the classicalcomplement pathway [23]. In addition, C4b always coincidedwith increased amounts of C4bp. In only one case (MGN)increased C4bp deposits were demonstrated in the absence ofC4b antigen.

Elution studies

To examine whether the complement components C4b andC4d would bind covalently or non-covalently to tissue struc-tures, elution studies were performed. Treatment of unfixednormal kidney sections with 1,5 M NaCI solution or citratebuffer of pH 3.3 resulted in only slight diminution of the C4dstaining intensity. Three cases of glomerulonephritis with dis-tinct deposition of C4b were equally treated and displayed aslight decrease of antigen reactivity.

A

4+

3 3+S05 2+C

1+j2 3 8 I 5 4

years months

Normal kidney

4+

3÷

J4JMdA2+

2 5 2 1 1 2 1 1 1 2 2 1

gA nophropathy SLE nophropathy MPGN

D E4+

3-1-

2+

11J}JdJ Sdi1 1 1 2 1 1 2 2 2 31 1 1

MGN UrN SHN MCNS FGS MN RPGN

Zwirner et a!: C4d in human glomeruli 1073

Fig. 4. Staining of normal human kidney sections. (A) Segmental and granular glomerular distribution of C4d (M4d1). (B) Weakly positive stainingreaction for C4bp (mAb), note the segmentally labeled glomerular capillary wall. (C) Absence of C4b antigen (M4c1). (D) Weak but distinctlypositive labeling for 1gM (mAb).

Fig. 5. Renal biopsies from one SLE patient (upper series) and one patient with mesangial proliferative g!omerulonephritis (lower series). Sectionsfrom SLE nephropathy displayed distinct deposition of C4d (A, M4dl), C4bp (B, mAb) and C4b (C, M4c1). The patient with MGN had strongmesangial deposits of C4d (D, M4dl), but normal levels of C4bp (E, mAb) in the absence of C4b (F, M4c1).

a

I. S •—

I •S•t S.

IS

d S

I.

S

-iS

4

•I'. •V1V '. •

A flt C

E' •1.

I -4.- I.,1'•

-S

4

I-'I

1/'

4

-I,

-

'C.

Zn—- — I

.4

;51% •) •C

I—. •L'

.2 "Sr

i

1

S

r E

S'

1%

-I

- C

p

p

1074 Zwirner et a!: c4d in human glomeruli

Fig. 6. Biopsies from patients with membranous glomerulonephritis (upper series) and membranoproliferative glomerulonephriris (lower series).Thepatient with MN displayed strong labeling for C4d (A, M4dl), moderate deposits of C4bp (B, mAb) in the absence of C4b (C, M4c1). In MPGN,distinct deposits of C4d (D, M4d1) and C4bp (E, mAb) were paralleled by low amounts of C4b (F, M4c1).

Immunoelectron microscopy C4d antigen was found to be predominantly located at theNormal parts from three nephrectomies were examined by subendothelial site of the glomerular basement membrane

immunoelectron microscopic examination. With this technique, (GBM) showing a gradient of decreasing intensity towards theepithelial site (Fig. 7). A mesangial staining pattern, as apparentin immunoperoxidase light microscopy, could not be seen inimmunoelectron microscopy.

Discussion

In the present study, the previously described [121 glomerularC4 compound could be characterized as the C4d cleavagefragment by the use of mAbs on cryostat sections of normalkidney tissue. Thus, the commonly encountered negative un-munohistological staining results with polyvalent antiseraagainst C4 can be explained by their failure to recognize the C4d(a2) fragment (Fig. 1). Unaffected parts from kidney tissueobtained during tumor nephrectomies have been widely used toinvestigate the distribution of normal renal antigens, includingcomplement components also in other studies [11, 24]. Sincethe classical pathway of complement activation is generallyinduced by ICs [5, 23, 251, the question appeared whether C4ddeposition in normal glomeruli was also connected with ICformation. For the immunohistochemical demonstration of ICswe therefore applied mAbs with specificity for 1gM or IgG incomparison to conventionally used polyclonal antisera. No IgGantigens were detected irrespective of the applied antibodies.However, a surprisingly clear difference in sensitivity wasfound in the detection of 1gM by either mono- or polyclonalreagents (11 vs. 4 positive staining reactions). Benefits of the

Fig. 7. Deposition of 4d in immunoelectron microscopy. Stainingintroduction of mabs into immunohistochemistry are well rec-

reaction with monoclonal M4d1 occurred predominantly along the ognized [26]. The main obstacle of the polyclonal antiserumsubendothelial site (arrow) of the glomerular basement membrane. used was a distinct staining background, that could only be

Zwirner et a!: C4d in human glomeruli 1075

abolished using very high antibody dilutions. Unspecific stain-ing reactions might indeed be causative for a remarkable lack ofsensitivity of conventional antisera in comparison to mAbs. Theoccurrence of Ig deposits in normal human kidney tissue iscontroversially discussed. In general, glomerular ICs are con-sidered as being pathologic [19]. However, by the application offluoresceinated (FITC) antisera, glomerular immune deposits(1gM and/or IgG) have been detected in patients with no clinicalor light microscopic evidence of glomerulonephritis in up to39% of cases [27, 281. Differences in antigen specificity of theantisera could very well explain the divergent results. In mice.an ubiquitous glomerular presence of ICs (IgG and/or 1gM) hasbeen demonstrated in animals with high and permanent expo-sure to foreign infective agents [291. In contrast, animals grownin a germ-free environment showed a much lower incidence ofdeposits. Furthermore, previous experimental observations inmice suggest that trapping and deposition of preformed, p1] -neutral ICs occurs only temporarily in glomeruli [30, 31]. Inaddition, the binding of Clq, that is associated with ICs, tolaminin [32] and fibronectin [331, which are both constituents ofbasement membranes, is well established and could provide thebasis for retention of ICs at the GBM. Thus, complementdeposition via activation of the classical pathway in glomeruliand renal arteriolar walls possibly takes place as a result of thetransient glomerular deposition and clearance of low amountsof antigen-antibody complexes formed during natural antibodyresponse in vivo [29]. This deposition may either occur in themesangium or along the GBM. Glomerular C4d deposits mostlikely represent stable C4d-IgM complexes [5] as well ascovalently bound C4d fragments in close vicinity to the activat-ing immune deposits. The preponderance of C4d in comparisonto local 1gM deposits could thus be explained, and also by arather slow turnover of such C4d molecules that are covalentlybound to glomerular structures. The reasons why the indirectimmunoperoxidase techniques show predominantly mesangialC4d with light microscopy whereas subendothelial C4d appearsin electron microscopy, are not entirely clear. Several technicaldifferences in the processing of the tissue samples could beconsidered, for example, long incubation periods and possiblediffusion of the DAB product [34] out of the mesangium withimmunoelectron microscopy.

An alternative explanation for the glomerular deposition ofC4d could be the passive absorption of circulating C4d gener-ated during systemic complement activation. However, a re-cent study investigating the binding specificities of C3d innormal glomeruli made passive absorption of C3d highly un-likely but favored more specific interactions with the GBM [24].Also in our study, treatment of unfixed normal kidney sectionswith low pH or high salt concentration resulted in only a slightdiminution of glomerular C4d antigen. We conclude that themajority of C4d is covalently bound to the glomerular basementmembrane and thus indicates complement activation via theclassical pathway in close proximity. Furthermore, our hypoth-esis of a local, IC-mediated complement activation is corrobo-rated by the detection of glomerular C4bp similar in localizationto C4d. C4bp binds to C4b as a cofactor for cleavage of C4b intoC4d and C4c by factor I [35, 36]. We suggest that C4bpaccumulates locally at the site of complement consumption and,after cleavage of the C4b-C4bp complex, disappears slowly dueto its high molecular wt (550 kD). It should be noted, that a very

similar histologic picture, that is, association of C4d and C4bpwith 1gM (in the absence of C4a and C4c), is observed in humanlymphoid germinal centres, where ICs are retained on folliculardentritic cells [13].

Despite the simultaneous detection of 1gM in 9 out of 14probes of normal kidney tissue, a direct, spontaneous activationof the classical complement pathway along the GBM indepen-dently from prior Ig attachment can not be completely ex-cluded. Cardiolipin from mitochondrial membranes, vascularendothelium, DNA, RNA, the myclin basic protein, retrovi-ruses and other agents have been shown to represent antibody-independent activators of the classical complement pathway[37, 38]. Yet, the subendothelial localization of C4d as shown inimmunoelectron microscopy clearly corresponds to the suben-dothelial electron-dense deposits in glomeruli of mice afterinjection of ICs [30].

The investigation of diseased kidney tissue provided addi-tional arguments for the IC-mediated C4d deposition in normalhuman glomeruli. In all eight cases of SLE nephropathy andMPGN, abundant deposition of C4d and C4b, respectively, wasaccompanied and presumably caused by ICs via activation ofthe classical pathway. In addition, glomerular appearance ofC4b fragment, as judged by the simultaneous detection of C4cand C4d antigens, was invariably paralleled by an increase inglomerular 1gM and/or IgG. Presence of C4b might thereforeindicate the very acute phase of complement activation inducedby IC deposition in close proximity (Fig. 5A-C). On thecontrary, abundant glomerular C4d in the absence of C4bantigen suggests classical complement activation returning tophysiological levels (Fig. 6A-C).

As would be predicted, the amount of C4bp detected indiseased kidney tissue varied with the appearance of C4bantigen. Taken together, higher quantities of C4d were paral-leled by increased 1gM and/or IgG deposition. The amount ofC4bp, however, closely followed the appearance of C4b. TheC4b-C4bp complex can therefore be regarded as a histologicalmarker of disease activity whenever the classical pathway ofcomplement activation is involved. Its significance for theevaluation of complement activation in the fluid phase has beenstated elsewhere [39]. Definite Ig-induced classical complementactivation has recently been observed in idiopathic membra-nous glomerulonephritis also by other investigators [40].

Several authors have already reported complement activationin normal human kidney tissue [9, 11]. However, complementproteins C3d, factor H and C5b-9 complex were localized to allrenal connective matrices, that is, arteriolar media, glomerularand tubular basement membranes and mesangial matrix. Ultra-structurally, cell membranes shed from adjacent cells could beidentified as the most probable target structure. Vascular C5b-9deposits were recently correlated to the development of arte-riosclerosis [41, 42] and a similar process, that is, activation bycell debris, was suggested for the kidney. In contrast to C3d andC5b-9, C4d deposition representing the classical pathway ofcomplement activation, is strictly confined to glomerular andfew renal arteriolar structures and coincides with 1gM-com-plexes. Therefore, from our data a clear discrimination betweencomplement activation via the alternative and the classicalpathway operating in different renal compartments appears.Detection of C4d in glomeruli of healthy subjects, including twochildren, supports the idea that fixation of 1gM-complexes to

1076 Zwirner el a!: C4d in human glomeruli

glomerular and certain arteriolar basement membranes may bea common route for the catabolism of antigen-IgM complexesformed during natural antibody response in vivo.

Finally, it should be noted that the classical pathway ofcomplement activation is a powerful tool for the elimination ofICs [231. It seems quite likely therefore, that the glomerular C4dfragments represent the remnant of a continuous physiologicalclearance mechanism. This would also contribute to understandthe propensity to renal IC injury in persons with inheritedcomplement deficiencies [2, 43].

Acknowledgments

This work was supported by the Deutsche Forschungsgemeinschaft(SFB 217: C8), in part by the Wilhelm Sander Stiftung and theFriedrich-Baur-Stiftung.

We are indebted to Dr. M. Prosinger (Urologische Klinik, UniversitätMUnchen) and Dr. K. Devens (Kinclerchirurgie, Universitflt München)for providing organ specimens. We are grateful to Dr. G. Thoenes, Dr.E. Held (Medizinische Klinik, Universität München) and Dr. M. Gokel(Pathologisches Institut, UniversitSt Munchen) for providing the diag-noses and tissue specimens from patients with glomerulonephritis.

Reprint requests to Dr. H. E. Feucht, Medizinische Klinik Innens-tadi, Universität München, Ziemssenstraj3e 1, D-8000 Miinchen 2,Federal Republic of Germany.

References

1. VERROUST PJ, WILSON CB, COOPER NR, EDGINGTON TS, DIXONFJ: Glomerular complement components in human glomerulone-phritis. J C/in Invest 53:77—84, 1974

2. MALLICK NP, EYRES K, ACHESON EJ, GOLDBY FS, JEAKINS J,LAwLER W, LUCAS S, TAYLER G, WILLIAMS G: Complement andglomerular disease—a natural history study. Nephron 27:67—73,1981

3. COUSER, WG: Mechanisms of glomerular injury in immune-com-plex disease. Kidney mt 28:569—583, 1985

4. LAW SK, LICHTENBERG NA, HOLCOMBE FA, LEVINE RP: Inter-action between the labile binding sites of the fourth (C4) and fifth(CS) human complement protein and erythrocyte cell membranes. JImmunol 125:634—639, 1980

5. CAMPBELL RD, DODDS AW, PORTER RR: The binding of humancomplement component C4 to antigen-antibody aggregates. Bio-chem J 189:67—80, 1980

6. TACK BF, HARRISON RA, JANATOVA J, THOMAS ML, PRAHL JW:Evidence for presence of an internal thiolester bond in thirdcomponent of human complement. Proc Nat! Acad Sci (USA)77:5764—5768, 1980

7. Yu CY, CAMPBELL RD: Definitive RFLPs to distinguish betweenthe human complement C4A!C4B isotypes and the major Rodgers!Chido determinants: Application to the study of C4 null alleles.immunogenetics 25:383—390, 1987

8. REID KBM, BENTLEY DR, CAMPBELL RD, CHUNG LP, SIM RB,KRISTENSEN T, TACK BF: Complement system proteins whichinteract with C3b or C4b. immuno! Today 7:230—234, 1986

9. FALK RF, DALMASSO AP, KIM Y, TSAI CH, SCHEINMAN JI,GEWURZ H, MICHAEL AF: Neoantigen of the polymerized ninthcomponent of complement. Characterization of a monoclonal anti-body and immunohistochemical localization in renal disease. J C/inInvest 72:560—573, 1983

10. Dot T, KANAT5U F, SUEHIRO F, NAGAI H, Y05HIDA H, HAMASH-IMA Y: Clinicopathological study of patients with mesangial iso-lated C3d deposition in various glomerular diseases. Nephron46:188—193, 1987

11. HINGLAIS N, KAZATCHKINE MD, BHAKDI S, APPAY M-D, MAN-DET CH, GROSSETETE J, BARIETY J: Immunohistochemical study ofthe C5b-9 complex of complement in human kidneys. Kidney mt30:399—410, 1986

12. FEUCHT HE, JUNG MCH, GOKEL Mi, RIETHMULLER G, ZWIRNERJ, BRASE A, HELD E, O'NEILL GJ: Detection of both isotypes of

complement C4, C4A and C4B, in normal human glomeruli. Kidneymt 30:932—936, 1986

13. ZwIRNER J, FELBER E, SCMMIDT P. RIETHMULLER G, FEUCHTHE: Complement activation in human lymphoid germinal centres.Immunology 66:270—277, 1989

14. SCHNEIDER C, NEWMAN RA, SUTHERLAND DR, ASSAR U,GREAVES MF: A one-step purification of membrane proteins usinga high-efficiency immunomatrix. J Biol Chem 257:10766—10769,1982

15. ZICCARDI Ri, COOPER NR: Demonstration and quantitation ofactivation of the first component of complement in human serum. JExp Med 147:385—395, 1978

16. ASSIMEH SN, BINO DH, PAINTER RH: A simple method for theisolation of the subcomponents of the first component of comple-ment by affinity chromatography. J Immune! 113:225—234, 1974

17. LAEMMLI UK: Cleavage of structural proteins during the assemblyof the head of bacteriophage T4. Nature 227:680—685, 1970

18. TSANG VCW, PERALTA JM, SIMONS AR: Enzyme-linked immuno-electrotransfer blot techniques (EITB) for studying the specificitiesof antigens and antibodies separated by gel electrophoresis. MethEnzymo!92:377—391, 1983

19. THOENES G: The immunohistology of glomerulonephritis. Distinc-tive marks and variability, in Current Topics in Pathology, "Gb-merulonephritis" (vol 61), Berlin-Heidelberg, Springer-Verlag,1976

20. PAPE CR, RIEBER EP, EISENBURG J, HOFFMANN R, BALCH CM,PAUMGARTNER G, RIETHMULLER G: Involvement of the cytotoxic!suppressor T-cell subset in liver tissue injury of patients with acuteand chronic liver disease. Gastroentero!ogy 85:657—662, 1983

21. STRIKER GE, KILLEN PD, FARIN FM: Human glomerular cells invitro: Isolation and characterization. Transplant Proc 12:88—99,1980

22. GRAHAM RC, KARNOVSKY Mi: The early stages of absorption ofinjected horseradish peroxidase in the proximal tubules of mousekidneys. Ultrastructural cytochemistry by a new technique. JHistochem Cytochem 14:291—302, 1966

23. SCHIFFERLI JA, NG YC, PETERS DK: The role of complement andits receptor in the elimination of immune complexes. N Eng! J Med

315:488—495, 198624. LElvo I, ENUVALL E: C3d fragment of complement interacts with

laminin and binds to basement membranes of glomerulus andtrophoblast. J Cell Biol 103:1091—1100, 1986

25. DOEKES U, VAN Es LA, DAHA MR: Influence of aggregate size onthe binding and activation of the first component of human com-plement by soluble IgG aggregates. Immunology 45:705—713, 1982

26. GATTER KC, CORDELL JL, FALLINI B, GHOSH AK, HERYET A,NASH JRG, PULFORD KA, M0IR DJ, ERBER WN, STEIN H:Monoclonal antibodies in diagnostic pathology: Techniques andapplications. J Biol Resp Mod 2:369—395, 1983

27. LARSEN S: Glomerular immune deposits in kidneys from patientswith no clinical or light microscopic evidence of glomerulonephri-tis. Acta Path Microbiol Scand Sect A 87:313—319, 1979

28. VELOSA J, MILLER K, MICHAEL AF: Immunopathology of theend-stage kidney. Am J Pathol 84:149—160, 1976

29. MARKHAM JV JR, SUTHERLAND JC, MARDINEY MR: The ubiqui-tous occurrence of immune complex localization in the renalglomeruli of normal mice. Lab Invest 29:111—120, 1973

30. MANNIK M, AGODOA LYC, DAVID KA: Rearrangement of immunecomplexes in gbomeruli leads to persistence and development ofelectron-dense deposits. J Exp Med 157:1516—1527, 1983

31. HAAKENSTAD AO, STRIKER GE, MANNIK M: The disappearancekinetics and glomerular deposition of small-latticed soluble immunecomplexes. Immunology 47:407—414, 1982

32. BOHNSACK iF, TENNER AJ, LAURIE GW, KLEINMAN ilK, MARTINGR, BROWN EJ: The CIq subunit of the first component ofcomplement binds to laminin: A mechanism for the deposition andretention of immune complexes in basement membrane. Proc NailAcad Sci USA 82:3824—3828, 1985

33. SORVILLO J, GIGLI I, PEARLSTEIN E: Fibronectin binding to Clqassociated with antigen-antibody complexes in EDTA-treatedplasma. ScandJ Immunol 23:153—160, 1986

34. KERJASCHKI D, SAWADA H, FARQUHAR MG: Immunoelectron

Zwirner et a!: C4d in human glomeruli 1077

microscopy in kidney research: Some contributions and limitations.Kidney mt 30:229—245, 1986

35. FUJITA T, GIGu I, NUSSENZWEIG V: Human C4-binding protein.II. Role in proteolysis of C4b-inactivator. J Exp Med 148:1044—1051, 1978

36. SCHARFSTEIN J, FERREIRA A, GIGLI I, NUSSENZWEIG V: HumanC4-binding protein. I. Isolation and characterization. J Exp Med148:207—222, 1978

37. Kovcsovics T, TSCHOPP J, KRESS A, ISLIKER H: Antibody-independent activation of Cl, the first component of complement,by cardiolipin. J Immunol 135:2695—2700, 1985

38. LINDER E: Binding of Clq and complement activation by vascularendothelium. J Immunol 126:648—658, 1981

39. ITo S, FUJITA T, TAMURA N: Determination of C4b C4bp complexformed by the activation of classical complement pathway using an

enzyme-linked immunosorbent assay. J immunol Meth 105:145—150, 1987

40. KusuNoKi Y, ITAMI N, TOCHIMURA H, TAKEKOSHI Y, NAGASAWA5, YOSHIKI T: Glomerular deposition of C4 cleavage fragment(C4d) and C4-binding protein in idiopathic membranous glomeru-lonephritis. Nephron 51:17—19, 1989

41. SCHAFER H, MATHEY D, HuGo F, BHAKDI S: Deposition of theterminal CSb-9 complement complex in infarcted areas of humanmyocardium. J Immunol 137:1945—1949, 1986

42. VLAICU R, NICULESCU F, Rus HG, CRISTEA A: Immunohisto-chemical localization of the terminal C5b-9 complex of comple-ment in human aortic fibrous plaque. Atherosclerosis 57:163—177,1985

43. AGNELL0 V: Lupus diseases associated with hereditary and ac-quired deficiencies of complement. Springer Semin Immunopathol9: 161—178, 1986