Mucosal complement deposition in inflammatory bowel disease

TS HALSTENSEN, P BRANDTZAEG. Mucosa! complement deposition in inflam, matory bowel disease. Can J Gastroenterol 1993; 7 (2) :91-101. Mucosa I deposition of activated complement and immunoglobulin (lg) in inflammatory bowel disease (IBD) was examined by indirect twoand three-colour immunoAuoresccnce staming applied on sections of ethanol-fixed or fro::cn tissue specimens from pancnti; with ulcerative colitis or Crohn's disease. Monoclonal antibodies ( rnAhs) to lgG subclasses and neoepicopes of activated C3b or terminal complement complex (TCC) were used m combination with rabbit antiserum to various complement components (Clq, C3c, C3<lg, C4c). Acttvated C3b was found on the luminal face of the surface epithelium in the most affected ulcerative colitis specimens from 91 % of23 patients, together with cycolytic TCC in 8.1 % Similar deposition was observed in 50% of 18 patients with Crohn's disease. However, co-deposition of the lgGI subclass and complement components involved in the cla,~ical activation pathway (Clq and C4c) was seen only in ulcerative colitis and in complement components involved in the classical activation pathway (Clq and C4c). Moreover, m ulcerative colitis the:se epithelial immune complexes often co-localized with a previously identified Mr 40 kDa putative autoantigen (mAb 7EuH12) Addmonal type Ill immune reaction might take place in both diseases because evidence of continuous vascular complement activation ha:s been seen in ~ubmucosal blood vessels. The results demonstrated that local complement acuvation takes place in lBD lesions. While epithelial deposition of lgGI and activated complement suggested an autoimmune attack in ulcerattve col1t1s, the ah~ence of IgGl, Clq and C4c in Crohn's disease was rather consistent with the alternative activation pathway.

T HE LESIONS t)l• INR.!\MMATORY bowel disease (IBD) are infiltrated with lymphocytes, macrophages, 1111munoglobuli n (Ig)-pro<lucing plasma cells ( mainly of the comple ment activating lgG l subclass) aml a !>tnktng acc umula tion of ncutrophils (1 -, ).These featu res suggest that local complement acn vatton takes part m the pathogenesis.Complement system: Comp lement is a complex multifacto n al defence system composed of at least 26 solu ble and me mbra ne-bound prote tns wit h the potenttal of formtng an au1vared cascade.Such acttvau on may he induced by complexing lgG or lgM a nti bodies with luminal l)r epithe lial a n t 1gem (classical pathway) (Figure 1), or by direct activation of C3 by components of the loca l m1c rohiota (alterna t1vc pathway) ( 4 ).Both pathways induce CJ cleavage a nd me rge m the terminal ac tivation step with the assembly of the te rminal comple ment complex (TCC}, ie, CSb-9 (5)(6)(7).Classical activation pathway: The classical pathway components comist of the C l complex (C l qrs), C2 a nd C 4. This pathway b activated by bmding of C I qrs with the Fe portion oflgG or lgM bound to anngcn.lgM has a supe rior acti vation potential a nd of the lgG subclasses, 11-,,03 seems to be more potent than lgG l .T he complement-activating po ten tial of IgG2 1s somewh at controvcr:;ial, but may J epenJ on antigen denstty (8).lgG 4 (9) , lgA ( 10) and lgE ( 11 ) Jo not .ictivatehuman complement by th l' classrrn l pa thway.Ac-posantes du complement (Clq, C3c, C3dg, C4c).Le C3b active a ete observe sur L a face luminale de l'epithelium de surface des specimens de colite ulcereuse les plus touches de 91 % des 23 patients, le complexe de complement terminal cytolytique etant present chez 8, l %.Un depot similaire a ete observe chez 50 % des 18 patients atteints de maladie de Crohn.Cependant, le depot conjoint d'une sous-classe d'lgGl et de composantes du complement associees a la voie d'activation classiques (Clq et C4c).De plus, clans la colite ulcereuse, ces complexes immuns epitheliaux se trouvaient localises avec un autoantigene Mt 40kDa possiblemcnt deja identifie (mAb 7El2Hl2).Une autre reaction immunitaire de type Ill peut avoir lieu clans les deux maladies parce que I' on observe une activation continue du complement vasculaire clans Jes vaisseaux sanguins sous-muqueux.Nos resultats ont demontre que !'activation du complement local a lieu clans des lesions de la maladie intestinale inflammatoire.Bien que les depots epitheliaux d'lgG 1 et !'activation du complement suggerent une attaque autoimmune dans la colite ulcereuse, l'absence d'lgGl, de Clq et de C4c dans la maladie de Crohn concorde plutot avec la voie d'activation alternative.tivated C l catalyzes the assembly of the classical CJ convertasc, C4b2a (1 2).Complement activation may be controlled at this stage by the C4b-binding protein (C4bp) or facror l in plasma and by complement receptor 1, decay-ac-celerating factor and membrane cofactor protein on cell membranes ( 7,13).Alternative activation pathway: The alternative pathway is activated by certa in po lysaccharides, vi ruses, fungi, bacteria, cell organelles and aggregated r Cl~I pathway J § ernative pathw~ C3b-likc molecule, C3( J-IZO), which may associate with factor B. The C3(1-120)B complex is activated by facto r D anJ factor B is cleaved into Ba and Bb.The product C3(Hzo)Bb is the alternative path way by which C3 convertase cleaves CJ; the anaphylatoxin C3a is released and C3b bin<ls covalently to the target with its very reactive thioester bone.I.Regard less whether C3b is generated by the classical or alternative pathway, it may bind another facto r B, thereby activating the alternative amplification loop, o r bind to factor H and be cleaved by facto r l, which terminates further activation (Figure 2) (1 6,17).It is the nature of the surface that determines whether C3b preferentially binds to factor B or to the inh ibitory factor H. Factor P is not essential for the alternative pathway but stabilizes the alternative CJ con• vertase, resulting in mo re efficient activation.

CB CB
The two activation pathways are not strictly separated because antibodies and immune complexes may also activate the alternative pathway (18), and agents (such as cardiolipin, C-reactive protein and bacterial surface antigens) may activate the classical pathway under certain circumstances (18)(19)(20)(21)(22). Terminal activation pathway: When target-bound CJ convertase (C4bC2a orC3bBb) binds its own product (C3h), it becomes the CS convertase, which initiates terminal pathway activation by generating CSa and CSh.Whether surface-bound or in the fluid phase, CSb reacts with C6, C7, CB and C9 to form TCC. When C7 binds to CSb-6 on biolog ica l membranes, it exposes hydrophobic regions and becomes inserted into the lipid bilayer.Thereafter CB and C9 bind to C5b-C7.T he lync activity of the CSb-8 complex is li mited to erythrocytes, wh ereas ki lling oi nucleated cells requ ires at least one C9 withi n the me mbra ne-boun<l TCC (CSb-9[m]) ( 23) .Two me m brane molecules have been shown to mhibit TCC formation on cell me mbranes: the homologous restrictton fac tor (24 ), also called CB-binding protein (C8bp) (25 ), and the Mr 20 kDa ho mo logous restriction factor (C DS9 ) ( 26,27 ).

C3 b inactivation: C3b
bo und to either comple me nt rece ptor I or 2 (CRI or C R2) or ~crum fauor Hbecomes partly degraded by sc rum factorl, and a small Mr 2 kDa fragment C3f is released from the a chain (Figure 2) (31 -33).The in ac ti v ated produ c t (iC 3b) is degraded to a small Mr 40 kDa fragment, C3dg, and a major Mr 150 kDa fragment, C3c.ThL• latte r 1s released to the flu id phase whe rea~ C3Jg (w h ich cn n ta 111 s rhc C3b thioester bond) re mains covale ntly bound to the target surface 0 4 ).C3d g may he further c leaved by trypsin to C3J (Mr 30 kDa ) a nd C3g (M r lO kDE) in vitro (35).Neoepitopes in the complement system: The structural alterat ion of complement compone nt~ induced during activation exposes new an tigen ic determinants (eg, mo noclonal ant1bod1es [mAbs)) to neoepitops; t hese react selccttvely w1th the ac ti vat ed components and can therefore be used for 1mmunoh istoc hc mi stry o n direc tly fixeJ or frozen u ssuc to reveal m situ complement activation.
The mAb bH6 reacts with a ncocpitope in the C3c part of C3b o r with its inacti vated form iC3b (36).Terminal activation may be disclosed by mAb aEll react ing with a C 9 ncoepitopc cxpo~ed during C9 polymerization in membrnnc-mcmporatcd TCC (C5b-9[ml) or when C9 bmds to C8 m the soluble S protein-  bound form ot TCC (SC5b-9) (37 ).These two forms of TCC, th erefore, cannot he d im nguished by rnAb aEll.Prev ious 1mmunoh1sroch emical sludies of complement deposition in IBD were based on an tiscra to native complement components applied o n tissue specimens that con ta med unpredic ta ble a mo unts of ext racellula r scrum proteins (38,39).In fact, there may be quite high concent rat ion s of inte rstitial na t ive C3 in infla med tissue wit h inc reased vascular permeability (40 ).
The purpose of the authors' previous stud ies (41 -45) has bee n to trace the site and nature of mucosa!comple ment activation in IBO.In addition , we reexamined the tissue d1stnbution of the previo usly identified Mr 40 kDa putati ve a urnanc1gen in ulcerative colitis (46.47) by using t hl.' mAh 7E1 2l l 12 for 1mmunofluorcscenu: l wo-and threecolour seaming.

Tissu e and immunological reage nts:
T h e clinical ma t e ri al consist ed of colonic b iopsy o r resection specimens fro m 39 patie nts with ulcerative colitis, 18 with C rohn\ colitis, anJ 26 controls (mainly h1stolog1cally normal mucosa fro m patients with colonic carc moma).Heal samples from 14 patie nts with C rohn's ileitis and six contro ls were a lso included .
Because infla med tissue contains more vessels than normal tissue, vasc ular TCC and C3c positivity was scored on a sem14uanti tative scale and related to the numher of blood vessels in the actual section.Each specimen was given a vascular T CC a nd C3c percentage score as detailed elsewhere ( 42).With regard to epithelial deposits, a semiquantitative score was applied for each immunological marker ranging from no epithe lia l staining (-) to m -te nse stammg on many cpithclial cell, (3+ ).Histopathological examination was perfo rmed to grade each specimen with regard to inflammation fro m negative(-) to intense (3+ ).Three-colour immunofluorescence staining: Selected coloni c and ilea! specimens fro m five patie nts with Crohn's disease and five with ulcerative coli tis havin g apical compl e ment deposition (sec he low) were subj ecteJ 10 three colour staining in which mAb to human lgG l (murine lgGl) was combined with mAb bH6 to C3h {IgG2a) or mAh aEll to T CC (lgG2a) and mixed wirh rabbit antiserum to C3c or C4c.Seco nd a ry reage nts were biotinylated and FITC-conjugaced subclass-specific goat anti mouse lgG2a and l gG 1 ( Southern Biotec hn o logy, Alabama) followed by 7-amino+ mcthylcoumarin -3-acetic acid (AMCA)-conjugated goat anttrabb1t IgG (1:20, Vector Laboratories, loca• tion) in combinatio n with S trep• tavidin -T cxas Red (0.0025 g/L; Bethesda Research Lahornto n es, Gaithersburg, Maryland) .
Becau,.e colon-spec1f1c autrnmtthodies have been reported in IBO (54-60), the authors examined prewashed tissue specimens from 11 ulcerative colttis and 19 Crohn's colttts patients for co-deposition ot complement components ,md lgG subclasses.Patients with ulcerative colttis (63'Yo) :.howed ,elective binding of lgG l apically on colonJC.ep1rhel1al cdls; C3b and TCC nftcn wen.• present wtthin these lgG l deposits (Figure 9).Crohn's d1sem,e, four of whom t1lso sta111cd for TCC on the apical face oi the ep1theltum.Er1rhcl1al complemem deposits were not seen in 96% of 26 colonic controls, nor m six 1leal con trols.EpiLhclial depo~1uon of lgG wa~ not ohserved in any uintrol.DISCUSSION Vascular complement activation: Vasc.ular complement deposits arc usually regarded as a sign of immune complex form,\tlon, often associated with vas• culi t1s (61).llowever, the authors ohserved no ce llul ar 1nftltration or f1brino1d necrosis a long with TCC or C3h, although thromhosed suhmucosal ,•cssl•ls post rive for TCC and C3b werl seen in tw11 patients (42).The vascular TCC was apparently 111 a membram-hounJ form hecause 1t wa, rctamed after extensive prewashing of the tissue specimens, suggesting that it represented the lytic, memhrane-incorporated C5b-9 (m).However, there was a striking co-localization of TCC and S protein, suggesting that it rcprescntcJ the soluble S prorem-associated form of TCC ( 42).Nevertheless, the possibility for soluble SC5b-9 (passively retaineJ from the circulation) was apparently excluded by the extensive washing procedure applied, which removed most 1. ltffusible molecules of comparable sm~ (eg, lgM, Mr 1000 kDa) from the tissue.Another possibility was an S protem-v1tronectin-me<liated bmdmg ofSC5b-9 to S protcm receptors on fihroblasrs or/and smooth muscle cells in the vascular walls (62), but this apparently was contrad1cced by the parallel staining for TCC and C3d, suggesting that both early and late phase complement activation haJ occurred in situ.Moreover, the vascular immune deposits in ulcerative colitis, and especially in Crohn's disease, contained significantly more C3b than control colon.
Because breakdown of C3b 1s a rapid process, the presence of both C3b and TCC suggested that vascular complement activation is a continuow, process m IBD.Although the vascular immune deposits did not generally stain for lg (63 ), segmental co-deposition of lgG I, Clq and C4c in C3b-and TCC-pos1nve vessels observed in two samples from Crohn's colitis suggested antibody-mediated classical activation (12).
Hemorrhagic necrosis induced by a local Shwartzman's reaction depends on both C5a and cumour necrosis factor (64 ).Complement activation might initiate this reaction as C5a stimulates macrophages to tumour necrosis factor transcription (65).Such activity could induce thrombosis of submucosal vessels in IBD, and larger vessels arc probably involved in Crohn's disease because of the transmural mtlammanon; perhaps this explains why vascular thrombosis and intestinal infarcnon seem to be a feature of Crohn's disease (66).The diffuse C3b pos1t1vity observed in the mucin layer of patients with Crohn's ileitlS might reflect mcreased local synthesis and external Epithelial lgG 1 and complement deposition: The epithelial deposition of lgG 1 along with activated complement (C4c, C3b and TCC) in active ulcerative colitis could reflect bindmg of autoantibodies to brush border-associated ant1gen(s) and complementmediated epithelial attack (43,44).This was further supported by the frequent co-localization of the epithelial immune complexes and the Mr 40 kDa colo111c protein, which has previously been proposed to be an autoantigcn in ulcerative colitis (45)(46)(47).
The expression of chis antigen, also observed in ilcal mucin (45), might cxplam 'backwash ileitis' (68) and ilea! pouch1tis (69), commonly seen in ulcerative col1us patients.The colonic distribution of the Mr 40 kDa antigen, from mainly goblet cell pos1tiv1ry in the proximal colon to mcrcasing cntcrocyte expression in the distal direction, further supports its putative role as an autoantigen m ulcerative colitis.Moreover, the mAb to the Mr 40 kDa antigen and the classical scrum anucolon antibodies (60) produce a similar staining pattern in human and rat colon, and ulcerative colitis patients have scrum antihod1es to both the Mr 40 kDa antigen (70) and rat colon extracts ( 54); this suggests that the Mr 40 kDa protein may be associated with the antigen for the classical anticolon antibodies.
Although epithelial cells arc partly protected against autologous complement attack by CD59 (71,72), this survival mechanism may not be sufficient 111 IBO mucosa.The cntcrocytes will probably also try to escape lgGmedtated complement attack hy rapid sheddmg of damaged plasma mcmhrancs (73 ).This event may involve loss of the epithelial brush borJer in ulcerative colins; the putative consequence 1s reduced surface area and hampered absorptive function.Epithelial damage, m addition, will increase leakage of autoantibod1es and complement components into the colomc lumen where transport takes place distally.Because the putative autoanngen 1s located at the apical surface, epithelium located downstream may he subjected to an lgGl-mcdiated complement attack contmuously.Perhaps this explains why ulcerative colitis is typically a continuum 111 distal direction.In add ition, autnant1bod1es may complex with mucm-associated antigen and cause luminal complement activation.The anaphylatoxms (C4a, C3a, C5a) produced may contribute to   (44).However, lack of co-deposition of lgG and clas~ical complement activation pathway componcnrs (C lq and C4c) suggests that complement wa~ acnvatcd by the alternative pathway in Crohn's Jisease.
Subcpithclial deposition uf TCC, S protein and C3J, was mainly seen directly beneath or adjacent tn epithelial destruction in ulcerative colins (43), but also under intact epithelium in Crohn's disease (44).Although this might be a secondary phenomenon, soluble immune complexes generated in the basement membrane :one could have induced subcpithclial u>mplcmcnc activation.The po~sibtlity that the C3b-positive globular elements nbserved in mucosa with Cmhn's colitis represented bacteria, such as chose previously identified hy electron microscopy in this bion ( 74, 75) llr chc isolated wall-deficient sphcroblast~ of Mycobacterium /laramberculusis ( 7 6), needs further clucidawm.Consequences of mucosal complement activation: Rcgardlcs~ of the mode of complement activation and the nature of the target attacked, generation of the anaphylatoxins (CJa, C4a and CSa) causes inflammation and release of other inflammatory mediators such as leukotrienes and prostaglandms (77-83).The elevated levels of these substances found in the ulcerative colitis lesion (84)(85)(86) could therefore be secondary to complement activation.Both leukotriene 84 and anaphylatoxin CSa are highly chemotactic for granulocytes and arc probably involved in the massive mucosa!mobilization of such cells seen in ulcerative colitis; lysosomal enzymes and toxic oxygen radicals liberated from neutrophils may also attack the epithelium.The intraluminal and crypt staining for TCC and C3d in ulcerative colitis might have been induced by autoantibodies reacting with the Mr 40 kDa autoantigen in mucin, shedding of attacked cell-membranes, as well as being induced by the alternative activation pathway, perhaps secondary to epithelial damage (the latter may dominate in Crohn's disease).
Expression of the Mr 40 kDa autoantigen by the colonic surface epithelium was often decreased in the surface epithelium in active ulcerative colitis anJ in patients with inactive disease of long duration.This suggests that lgG 1med iated comp,lement attack might exert a selective pressure on the epithelium, thereby allowing cells with low expression of this antigen co survive.One of the consequences could be positive selection of dysplastic epithelium in ulcerative colitis.Such a putative mechanism would explain the increasing risk for developing colonic carcinoma after long disease duration.This possibility is te~table and shoul<l be exploreJ in the future.