complement, immune complexes and systemic lupus erythematosus

British Journal of Rheumatology 1996^5:5-23

MICHAEL MASON PRIZE ESSAY 1995

COMPLEMENT, IMMUNE COMPLEXES AND SYSTEMIC LUPUSERYTHEMATOSUS*

K. A- DA VIESRheumatology Unit, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road,

London W12 ONN

COMPLEMENT has two main roles as part of the humaninnate immune system. Historically, the first of these tobe defined was that of defence against infection withpyogenic bacteria. However, it is now evident thatcomplement proteins, particularly those of the classicalpathway, play a role in the processing of immunecomplexes (ICs) and in protecting the body against thedevelopment of diseases such as systemic lupus erythe-matosus (SLE), which are thought to be mediated byICs. The study of individuals with inherited deficienciesof individual complement proteins has provided manyinsights into the functions of these proteins in vivo. Themost convincing evidence that complement is importantin protection against IC disease follows from the obser-vation that the prevalence of certain diseases in whichICs are a characteristic feature, particularly SLE, isgreatly increased in individuals who are geneticallydeficient in classical pathway complement proteins [1].

In this essay, I plan to discuss the results of the workI have performed during the last 5 yr in the Rheumato-logy Unit at the Royal Postgraduate Medical School(RPMS). During this time we have explored specificallythe interactions between ICs and the proteins of thecomplement system and their cellular receptors inthe pathogenesis of SLE. After a brief discussion of thebackground to the work, I describe some of the key invivo and in vitro experiments and clinical studies wehave performed, which have advanced our understand-ing in this area, and at the end of the dissertation Iattempt to address some of the implications of ourobservations for the management of patients with SLE.

IMMUNE COMPLEX PROCESSINGWhat was the state of knowledge of the role of

complement and complement receptors in the process-ing of ICs in the late 1980s, when I first becameinvolved in research in this area? It was well knownthat complement reacted with ICs to inhibit immuneprecipitation, solubilize immune aggregates and pro-mote IC binding to erythrocyte complement receptortype 1 (CR1). These reactions were thought to preventlocal accumulation of ICs in tissues outside the mono-nuclear phagocytic system by producing soluble com-plexes which are removed by simple diffusion awayfrom the site of formation and by binding to erythro-

Presented at the British Society for Rheumatology AnnualMeeting, April 1995, in Glasgow

Submitted 20 December 1994; revised version accepted IS June1995.

cyte CR1. This receptor decreases the phlogistic poten-tial of ICs by: (i) acting as a cofactor to the alternativepathway control protein, Factor I, which facilitates thefurther breakdown of the most important complementprotein C3; and (ii) transporting ICs through thecirculation to the fixed mononuclear phagocytic sys-tem. Tissue macrophages in the liver and spleen bearboth immunoglobulin Fc-, and complement receptorstype 3 and 4 (CR3 and CR4), and opsonized complexescan interact with both these groups of receptors. It hadlong been mooted that there may be a primary oracquired defect in mononuclear phagocytic function inSLE which predisposes to the development of diseaseby impairment of complex clearance. This idea thatabnormal function of the reticulocndothelial systemmight result in failure of IC processing stemmed fromearly experimental work in animals by Biozzi andcolleagues using colloidal carbon particles [2], and byHaakenstadt and Mannik [3], who demonstrated thatIC injection into rabbits resulted in saturable hepaticuptake, followed by spillover into other organs.Whether or not this so-called 'reticuloendothelial satura-tion' occurs in humans and is a contributory factorin the development of disease is not clear. Muchexperimental effort has been devoted to addressing thequestion of whether there is indeed a fundamentalabnormality of mononuclear phagocytic system func-tion in SLE related primarily to IC clearance mediatedby Fc receptors [4], or whether the primary problem isone of defective IC delivery to the mononuclear phago-cytic system secondary to hypocomplementaemiaand/or low levels of erythrocyte CR1, as suggested byPeter Lachmann in 1987 [5]. The experimental workwhich I describe in the first part of this essay attemptsspecifically to resolve this question.

A number of different model systems had beenemployed to address this problem previously. Earlystudies used erythrocytes coated with IgG or IgM, andmore recent studies have employed either aggregatedimmunoglobulin or soluble ICs [4,6-8]. Erythrocytescoated with IgG are cleared in the spleen (Fc receptordependent) [4,9], while IgM-coated cells show transientretention in the liver mediated by reversible binding tocomplement receptors [10]. The clearance of IgG-coated erythrocytes has been specifically studied inSLE patients. Frank and colleagues [11, 12] demon-strated a correlation between clearance rate, diseaseactivity and levels of circulating ICs in patients withSLE, but a number of other similar studies have failedto show any such direct correlations [13-15]. Mark

© 1996 British Society for Rheumatology

Downloaded from https://academic.oup.com/rheumatology/article-abstract/35/1/5/1782225by gueston 10 April 2018

BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 1

Walport showed in 1985 that splenic blood flow is animportant factor affecting the clearance of IgG-coatedred cells [16], and this may, of course, influence disease-associated clearance of these cells.

More recently, Lobatto and colleagues [6] assessedmononuclear phagocyte function in different diseases,using radiolabellcd soluble aggregates of IgG. Theseaggregates were predominantly cleared from the circu-lation in the liver and spleen. Significant differenceswere seen between normal subjects and patients withSLE. In particular, the liver/spleen uptake ratios werehigher in the patients, due to reduced splenic uptake ofthe aggregates. Halma and colleagues [17] analysedclearance of aggregated IgG in 22 patients with SLEand 12 normal volunteers, demonstrating reducedbinding of aggregates to red cell CRl in the patientgroup, with a faster initial elimination rate. In thisstudy, the major factor influencing the aggregate clear-ance rate was the serum IgG concentration.

Soluble IC clearance had been studied by Schifferliand co-workers [8] in our unit using mI-labelled tet-anus toxoid/anti-tetanus toxoid complexes [8]. Eithernative complexes, or complexes pre-opsonized in vitrowith autologous serum, were injected into normalvolunteers, and into 15 patients with IC disease orhypocomplementaemia. Immune complexes bound toerythrocyte CRl receptors in a complement-dependentmanner, and CRl number correlated with the level ofuptake. In subjects with low CRl numbers and hypo-complementaemia, there was a very rapid initial disap-pearance of IC, which was attributed at that time topossible deposition of complexes outside the reticulo-endothelial system, at other sites where they might bepotentially harmful, although in this early work nodirect imaging of the sites of IC clearance was per-formed. Two critical questions remained unanswered atthis time: (1) to what extent are the results of theseclearance studies using pre-formed ICs applicable tothe situation which exists physiologically, in which thecomplexes are actually formed in vivo in the presenceof complement proteins, which have profound effectson their physicochemical properties (discussed in detailin a recent review written in collaboration with Pro-fessor Schifferli [18]); and (2) where are soluble ICsreally cleared in humans?

One of the criticisms of all the studies of IC process-ing described above is that they were all performedusing large ICs prepared in vitro, in the absence ofcomplement, and may not therefore be physiological.There is conflicting evidence regarding the binding ofICs formed in the presence of complement to erythro-cyte CRl. Varga and colleagues [19] demonstrated thatbovine serum albumin (BSA)-anti-BSA complexesformed in the presence of serum failed to bind toerythrocytes. However, others have shown that thesuccessive infusion of human dsDNA antibodies anddsDNA into monkeys and rabbits leads to the rapidformation of ICs capable of binding to red cell CRl[20]. In human IC disease, it is inevitable that com-plexes will be formed in vivo, either in the circulation,or at a site of inflammation and autoantigen presen-

tation, in the presence of complement, and in the firstproject upon which I embarked when I started as anArthritis and Rheumatism Council junior fellow inMark Walport's laboratory at the Hammersmith, Iattempted to address, for the first time, the fate of ICsformed in vivo in man.

How did we approach this problem? We needed tofind a clinical situation in which it was likely thatcomplexes would be formed in the circulation, ideallyas a consequence of planned therapeutic interventionof some sort. The classical models of in vivo ICformation in animals involved the induction of serumsickness and IC-mediated nephritis [21]. The adminis-tration of an exogenous antigen, in the form of a drug,to a pre-immune individual could clearly producesimilar effects in man, as we demonstrated in a patientwho had been pre-sensitized, and then received i.v.streptokinase therapy [22]. A model was thereforerequired in which 'safe' complexes were formed, ideallyin a situation in which their formation and clearancecould be monitored in some way. It was Dr Epenetos,from the Department of Clinical Oncology here at theRPMS, who provided us with the model we needed. Hehad recently started treating his ovarian cancer patientswith a new form of radioimmunotherapy. Treatmentcomprised the following: under general anaesthetic,mouse IgGl monoclonal anti-tumour antibody (anti-human milk fat globule 1, HMFG1), at a dose of10-12 mg, labelled with 13II, was injected i.p. via a rigidcatheter (of the type designed for peritoneal dialysis)under direct laparoscopic vision on day 0. Twodoses (15-18 mg) of '"I-labelled antibody to mouse

DAYO131 l-labelledanti-HMFG1

DAYS 1 & 21 2 5 l-labelledHAMA

FIG. 1.—Protocol for the administration of radioimmunotherapy(HAMA — human anti-mouse antibody).


DA VIES: COMPLEMENT, IMMUNE COMPLEXES AND SLE

120 -

100-

6 0 -

6 0 -

4 0 -

2 0 -

o -

\

I0

^ >

<^

I10

:^fr—-

I20

— •

30

o•

A

a—

40

C3

C4

CH50

C3a

I50

-a

—A

I60

- 3

- 2

- 1

- 0

1)C3a

I

time (minutes)

FKJ. 2.—Fluid-phase complement activation after IC formationin vivo.

immunoglobulin were administered over 15-20 min byi.v. infusion into a peripheral vein, 24 h after the first(anti-tumour) antibody (Fig. 1) given in order toaccelerate the clearance of the first antibody and reducepotential radiotoxkaty from anti-HMFGl antibody thatdid not bind to the tumour. This model facilitated theanalysis of the formation and clearance of ICs formedin the circulation from the two antibody species.

Three patients (patients 1, 2 and 3) aged 43, 44 and42 yr were studied. All three were systemically entirelywell at the onset of therapy, but had advanced ovariancarcinoma with evidence of peritoneal spread. Treat-ment was given as part of a trial of radioimmunother-apy in ovarian cancer approved by the HammersmithHospital ethics committee. Sequential samples of venousblood were obtained from all three patients beforetherapy and at 5 min intervals for 1 h after startinginfusion of the human anti-mouse antibody. Ad-ditional samples were obtained 12 and 24 h later, and at1-2 week follow-up visits. We monitored the patientexternally with a gamma counter over the heart andabdomen, in order to obtain an idea of their main siteof localization.

The use of this model system, in which ICs wereformed in vivo between an antigen and an antibodylabelled with two different radioisotopes, one at a highspecific activity, facilitated the detailed characterizationboth of the kinetics of complex formation and disposal,and the way in which these complexes interacted withCRl on erythrocytes. An acquired reduction in erythro-cyte CRl numbers has been described by a number ofdifferent groups in patients with SLE [23, 24, 25], andit had been postulated (i) that such a reduction mayoccur as a consequence of IC ligation by CR1, transportto the Hver and spleen, and proteolytic removal of boththe complex and part or all of the receptor as aconsequence of this interaction between the IC-E-CR1complex and the cells of the fixed macrophage system,and (ii) that this acquired reduction in receptor numbermay be one of the factors contributing to abnormal ICprocessing in patients with this condition [5]. In theradioimmunotherapy model, in which relatively largequantities of ICs were formed in vivo, two of the keyquestions which I set out to address were whethererythrocyte CRl are indeed implicated in the clearanceof complexes, and whether a fall in receptor numberscould be demonstrated consequent upon IC formationand clearance. Systemic complement activation, anddeposition of the complement fragments C4d, C3dg andiC3b on the erythrocyte surface, have also been demon-strated in patients with SLE by ourselves and others[26-28], and in the radioimmunotherapy model wesought evidence for similar changes, occurring moreacutely, following IC formation.

A full account of this work may be found in ourpaper published in 1990 in the Journal of Immunology[29]. Our results confirmed that systemic complementactivation, and deposition of C4 and C3 fragments onerythrocytes, did indeed occur (Figs 2 and 3). It wasalso demonstrated by sucrose gradient analysis that thecomplexes formed comprised both antibody species(Fig. 4), and that the clearance of the first antibody wasgreatly accelerated as a consequence of IC formation

2000 - ,Day 1

u 1000 -

120 40

Day 2

• CR1o C3dg• C4d

I60 50

I100

Time after anti-mouse IgG (mins)

Fie. 3.—Changes in erythrocyte CRl, C3dg and C4d following in vivo formation of ICs in the radioimmunotherapy model.


BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 1

19S 7S

40000-

30000-

g- 20000-

10000-

30000 - i

time 0

I20000 -

I r

time 5 min

I l

10000 -

o -J

15000 -

I I I I I I

time 20 mins

I10000 -

5000 -

0 -Iro

T5

I I10 15Fraction

I20

I25

FIG. 4.—Sucrose density-gradient profiles from patient 3, followinghuman anti-mouse antibody injection.

[30]. From studies on the patients' red cells, it was clearthat a fraction of the complexes bound to erythrocyteCR1 (Fig. 5). We also showed that these in vivo -formedcomplexes were cleared mainly in the liver, although arelatively crude method of external monitoring wasemployed. Perhaps the most exciting finding in thisstudy was the observation that in all the subjectsstudied, a fall of up to 30% of the initial value in CR1could be demonstrated following infusion of the secondantibody, and IC formation and clearance (Fig. 3). Inparallel in vitro studies, employing the same radio-labelled antigen/antibody system, we were able todemonstrate analogous complement deposition on redcells, and fluid phase complement activation, but nofall in erythrocyte CR1. This would strongly suggestthat it is indeed the process of IC removal from theerythrocyte by cells of the fixed macrophage system

that is responsible for CR1 loss, and that this is notmerely a consequence of E-CR1 ligation by C3-opsonized complexes. In the in vivo studies, only asmall number of complexes could be demonstrated ona red cell at any one time (1-3 complexes per cell) andit is, at first sight, difficult to imagine how the processof IC clearance could result in the loss of a significantnumber of CR1, as we demonstrated. It should beremembered, however, that the processes of IC for-mation and clearance in vivo are dynamic ones, and.redcells binding complement and IC on their surface areable to interact sequentially with cells of the fixedmacrophage system, with stripping, and subsequentre-binding of other complexes. It is also important tonote that the distribution of CR1 is not homogeneouson the erythrocyte surface [31]. It is clear from electronmicroscopy studies that the CR1 are clustered in smallgroups of between 3 and 12 molecules. All red cellsexhibit this clustering, though to a variable degree: thecells bearing the highest CR1 number are generallythose having the largest clusters, and are probably themost efficient at binding ICs. Cosio and colleagues [32]analysed the binding of ICs to red cells using animmunofluorescence technique. It was observed thatonly some erythrocytes are capable of binding ICs, andas in the radioimmunotherapy model system onlyrelatively few ICs bind per erythrocyte, suggesting thatthe binding was restricted to defined areas on the cellsurface. It is therefore likely that the ligation andsubsequent stripping of a large C3b-opsonized anti-body-antibody complex, as in this model, would resultin the loss of multiple CR1.

Further evidence for the loss of CR1 from red cellsby proteolytic mechanisms comes from the work ofBarbosa and colleagues, who suggested that CR1 aredegraded in vivo to leave only 'stumps' consisting oftransmembrane and intraceUular domains of themolecule, to which he developed a specific monoclonalantibody [33].

30—|

20-

&10 -

o — RBCI I I I0 30 60 90

Time after injection of anti-mouse IgG (min)

I120

FIG. 5.—Clearance of l3lI-labelled antigen following injection ofhuman anti-mouse antibody, TCA and polyethylene glycol (PEG)precipitation. Erythrocyte antigen binding is shown on an expandedscale on the insert graph.


DA VIES: COMPLEMENT, IMMUNE COMPLEXES AND SLE

Low CR1 numbers have been described on erythro-cytes of patients with a variety of diseases otherthan SLE, including paroxysmal nocturnal haemo-globinuria, autoimmune haemolytic anaemias, AIDSand lepromatous leprosy. The mechanisms of CR1reduction in AIDS patients have recently been studiedin detail. Patients with HIV infection have a progress-ive loss of CR1 on red cells when the infection pro-gresses to AIDS [34-36], and genetic studies haveconfirmed that the defect is acquired [37]. Thesepatients do not have a loss of glycosylphosphatidyl-inositol (GPI)-anchored proteins, and the pattern ofCR1 fragments on erythrocytes is different from thaton normal ageing erythrocytes, indicating that in AIDSthe reduced number of CR1 is due to enhanced proteo-lytic cleavage [38]. AH these observations support thehypothesis that the reduction of erythrocyte CR1 in' IC diseases is indeed related to proteolytic loss of thereceptor as a consequence of IC processing.

STUDIES IN PATIENTS WITH SLEOur radioimmunotherapy study strongly suggested

that the main site of IC clearance in man was the liver.In the next phase of our work, we developed a modelusing radiolabelled pre-formed ICs, of high specificactivity, with a view to determining definitively the sitesof soluble complex clearance in man. We also plannedto use this model to examine further the clearance ofICs in patients with SLE, and to analyse the ways inwhich the kinetics and sites of complex clearance wereinfluenced by low serum complement levels, and lowlevels of erythrocyte CR1 in the disease. These studieswere the first to be performed in man in which the sitesof clearance of soluble ICs were directly visualizedusing an imaging technique, and were reported by ourgroup in the Journal of Clinical Investigation [40]. Thecomplexes used comprised hepatitis B antigen (HBsAg)and a polyclonal anti-hepatitis B reagent. These modelcomplexes comprise components specifically preparedand licensed for use in humans (the antigen is thematerial used to prepare the vaccine, and the antibodyis an approved preparation from the Swiss Red Crosswhich is normally used for passive immunization).Radiolabelling with I23I (an isotope with a short half-life, which has an emission spectrum ideal for imagingstudies) was performed using the #-bromosuccinamidemethod [39]. Twenty-six subjects were studied—eightmales and 18 females. Twelve normal volunteers wererecruited from the laboratory staff, aged from 23 to59 yr, six female and six male. The patients were 10subjects with SLE, aged from 21 to 56. Nine werefemale and one male. All fulfilled the revised ACRcriteria for this disease. A 'butterfly' (19G) i.v. cannulawas inserted into the subject's left antecubital fossa forinjection of radiolabelled complexes, and a similar linein the right arm for blood sampling. Five millilitres ofIC were injected as a bolus over 20 s at time 0, with thepatient positioned under a gamma camera (IGE 400Ton line to an MDS A2 computer). Dynamic imagingwas performed (serial 20 s frames) for 50 min, followedby 5 min posterior and anterior static images at 1,4 and

24 h. Four subjects underwent dynamic scanning for2 h, and whole-body scanning was also performed intwo cases, using a Starcam (IGE) gamma camera.

During scanning, sequential blood samples weretaken for measurement of whole-blood 1BI, trichloro-acetic acid (TCA) precipitation of protein-bound ac-tivity, estimation of erythrocyte-bound l73l andcomplement assays. Immune complexes were charac-terized by . co-precipitation with staphylococcalprotein A and sucrose density-gradient centrifugation.Erythrocyte CR1 numbers were measured using aradioligand binding assay with an anti-CRl mono-clonal antibody [41]. The quantification of IC uptakein the liver and spleen was performed from radioac-tivity count rates in specific 'regions of interest' (ROI)drawn around each organ on anterior and posteriorimages. Both SLE patients and controls exhibiteduptake of labelled ICs in the liver and spleen, withrapid clearance from the blood (Fig. 6). Scanning overthe lungs, kidneys and other organs revealed no specificuptake above blood pool level in either group (Fig. 7).The initial rate of IC clearance from the bloodwas significantly faster in the SLE group (median*i/2 = 2.15 min, range 1.3-6.6 min) than in the normalcontrols (median tl/2 = 5.15 min, range 3.6-14)

100

8 ^

£ 60 -

'jc _

20 -

0 -

100 —

eo —

•§ 60 —

£ 40 —a?

20 —

o—'

Normal subject

10 20 30I

40I

50

SLE patient

I0

1 110

1

time

120

after

1 130

injection

1 140

(mins)

1 150

Fio. 6.—Immune complex clearance kinetics in a normal subject anda patient with SLE.


10 BRITISH JOURNAL OF RHEUMATOLOGY VOL. 35 NO. 1

FIG. 7.—Whole-body IC scan performed at 2 h in a normal subject(left: anterior; right: posterior).

(£/ = 12.5, P < 0.002) (Fig. 8A). The rapid initial re-moval of complexes from the blood in the patients withSLE corresponded with their localization in the liver.

The median time at which 90% of maximum hepaticuptake was reached in the SLE group was 9.0 min(range 4.3-18 min), compared with 16.0 min(13-23 min) in the normal subjects (U = 8.0,P< 0.002) (Fig. 8B). At 10 min, between 27.3 and67.5% (median 40.7%) of injected complexes could bedetected in the liver in the normal subjects, comparedwith 43.0-79.6% of injected IC (median 56.3%) in thepatient group ([/ = 27, P < 0.05).

In both normal subjects and in the SLE group, a fallin hepatic activity was observed between 30 min and 2h after injection. This was more marked in the patients,as demonstrated by comparison of the ratio of hepaticcounts measured at 40 and 60 min in the two groups—median ratio 1.24 (range 0.88-1.44) in the studiesperformed in patients with SLE, compared with 0.75(0.48-0.95) in the controls (U = 2, P < 0.02). Thesedata are shown in Fig. 9.

The measurement of whole-blood, protein-boundand red cell-bound activity in sequential samples fol-lowing injection of labelled IC indicated that there wassignificant release of protein-bound radioactivity fromthe liver in the patient group occurring after 30-40 min.In Fig. 10, in which the patterns of IC handling in anormal subject and a typical SLE patient are com-pared, there is a clear rise in the whole-blood andTCA-precipitable activity between 25 and 60 min. TheTCA precipitation data indicated that a significantlygreater proportion of the activity detected in the bloodat 1 h was protein bound in the SLE patients (median86.7%, range 77.0-96.7) than in the normal controls(median 69.7%, range 52.0-79.1) (U = 2.0, P < 0.02).Precipitation of this material using Sepharose-staphylococcal protein A showed that the [mI]HBsAgremained complexed with IgG. The size of this materialwas estimated by sucrose density-gradient centrifu-gation, which showed it to be composed primarily ofmaterial of between 35 and 50S—intermediate betweenantigen and the injected ICs (Fig. lla and c).

(A)

15 —

1 0 -

8Sm2.uO

o

.E 5 -

co

0—'

Controls SLE patients

(B) 30—I

U

8-

oen

p

io —

o—'

iControls SLE patients

FIG. 8.—(A) Half-time of initial IC clearance in normals and patients with SLE. (B) Tune at which 90% hepatic uptake of ICs was achievedin patients and normal subjects.


DA VIES: COMPLEMENT, IMMUNE COMPLEXES AND SLE 11

CO

1

1.6 —I

1.2 —

0.8 -

t 0.4 ~oo

"•§rro.o —

I1

u E

ControlsI

SLE patients

Fio. 9.—Ratio of hepatic activity measured at 40 min: 60 min afterinjection in patients and normal subjects.

There was significantly reduced uptake of ICs by thespleen during the first hour amongst the patients(median 9.03% of injected ICs, range 4.05-23.7%)compared with the normal controls (median 23.9%,range 17.9-30.7%) (U = 4, P < 0.02). The ability of thespleen to retain ICs following initial uptake intothe organ was also abnormal in the SLE patients. Theactivity remaining in the spleen at 24 h was expressedas a percentage of the maximum organ uptakemeasured during the first hour after IC injection.Amongst the SLE patients, median uptake was 39%maximum (range 24-52%) compared with normal sub-jects—median 65.5% (range 58-73%) (t/ = 0,P < 0.002) (Fig. 12a, uptake at 1 h as a percentage ofinjected dose; Fig. 12b, 24 h splenic uptake as a per-centage of the maximum counts in the organ).

The major differences in the processing of ICs be-tween the normal subjects and patients with SLE can

30000 - lNormal

E8.

20000 -

10000 -

o - 1

r0

be summarized as follows: (i) initial clearance of ICswas more rapid in patients than in controls, as com-plexes localized rapidly in the liver; (ii) IC release fromthe liver occurred subsequently in patients; (iii) therewas significantly reduced splenic complex uptake in thepatient group. The various different factors whichmight explain these differences were therefore exam-ined. The median erythrocyte CR1 number measuredin the normal control group was 950 (range 467-1218).The median value for the SLE patients was significantlylower at 482, range 78-969 (U = 19, P <0.02). In allsubjects, there was a very close linear correlationbetween the maximum binding in vivo of IC to erythro-cytes and the CR1 number (r^ = 0.96). CR1 numberson erythrocytes were also correlated with the rate ofclearance of ICs by the liver (Fig. 13).

In the patients with SLE, there was a close corre-lation between the levels of both C4 and C3 in theplasma and the tm of IC clearance—C4: r^ = 0.997;C3: i-jp = 0.811. The correlation for C4 is illustrated inFig. 14. These data suggest that hypocomplementaemiamight play an important role in determining the abnor-mal IC kinetics amongst the patients with SLE. Weconsidered the possibility that this link was related tothe IC size, which we measured by sucrose density-gra-dient analysis. Figure 12 shows typical data. A shift inpeak complex size was seen in the 3 and 5 min samplesobtained from a normal subject, which was not ob-served in the samples from an SLE patient.

These imaging studies provided direct evidence thata reduction in plasma complement levels and erythro-cyte CR1 profoundly affects in vivo processing of ICs.Hypocomplementaemia was associated with morerapid uptake of complexes into the liver, with sub-sequent release into the circulation of IC of intermedi-ate size. As we have previously discussed, it had beennoted in earlier studies using l25I-labelled tetanus tox-oid/anti-tetanus toxoid complexes [8] that there wasmore rapid initial clearance of ICs from the circulation

30000 - i

20000 -

10000 -

0 ->

SLE patient

I I I I I I I10 20 30 40 50 60 70

Time after injection (mirts)

T10

r20

r30

r40 50

r60

i70

Time after injection (mins)

Fio. 10.—Radioactivity measured in blood after IC injection. TCA-precipitable, whole-blood and erythrocyte-bound activity is compared ina normal and an SLE patient.



in patients with SLE, or Clq deficiency. This wasattributed to 'trapping' of complexes outside thereticuloendothelial system, but in these earlier studiesdirect imaging was not available. In our imagingexperiments, we saw no evidence of 'trapping' outsidethe liver and spleen. The markedly increased rapidity ofinitial clearance from the circulation was due to morerapid hepatic uptake. There was then release of ICs

from the liver following this initial uptake phase. Wedo not know the true significance of this observation,but clearly material of this sort could have phlogisticpotential in itself, or serve to modulate autoantibodyproduction. It is tempting to speculate that ligation ofboth Fc and complement receptors is required forefficient processing of complexes, a hypothesis that weare actively investigating.

(a)100 —i

19S 7S

80 -

§ 60 -EI 40-3Eae 20 -

o -

(b)100 —i

80 -mc

8E3

E 4 0 -3s8 20 —

0 —'

r0

(c)100 —l

80 Hto

8 M -E3

I 40-|3E5? 2 0 - 1

0 —'

I6

I T i l l8 10 12 14 16

Normal

- t - o- t - 3 tnln

• • - t - 5 mlns

1 i i i i r i i2 4 6 8 10 12 14 16

i i i i i i r i i0 2 4 6 8 10 12 14 16

Fraction

100 - i

80 -

60 -

40 -

20 -

0 —'

100 —i

80 -

60 -

40 -

20 -

0 —'* A g a l o n e

I I I I T I I I I0 2 4 6 8 10 12 14 16

Fraction

FIG. 11.—(a) SDG profiles generated by labelled Ag alone and by the IC preparation before injection; (b) analysis of plasma samples from apatient and normal at 0, 3 and 5 min; (c) profiles at 0, 3 and SO min, showing material of intermediate size at 50 min.



[a]

3 5 -

3 0 -

1 2 5 -

•g 2 0 -

S" 1 5 -

1 0 -

5 -

0 -

®

I3

[b]

80 -

70 -

60 -

50 •

40 -

30 •

20 -

10 •

0 -

IControls

* •

ISLE patients

•

• • • •i

I IControls SLE patients

Fio. 12.—Reduced splenic uptake and retention of ICi in ourpmtients with SLE.

re -0.62 p-0.002£>P

2 0 -

1 0 -

o —I

200I

400I

600I I

1400BOO 1000 1200

erythrocyte CR1 number

Fio. 13.—Correlation between time of 90% liver uptake and E-CR1.

STUDIES IN A C2-DEFICIENT PATIENTFrom the above studies, it appeared that com-

plement levels affected both the sites and kinetics of ICclearance in SLE. Our next step was to use the samemodel system to study IC processing in a patient withhomozygous C2 deficiency and SLE, exploring theeffects on IC processing of the repletion of normalcomplement activity with fresh frozen plasma (FFP).

7 —

6 -

5 -

4 —

_•

|O

CO

.I

.I

.I

o —I0

rsp= 0.997 .

•

> I i I i i i i i i i i20 40 60 80 100 120

plasma C4 (% NHS)

Fio. 14.—Correlation between initial IC clearance rate and plasmaC4 in patients with SLE.

1st IC clearancestudy: day 0

Colloid scan - 24 hours

4x3uFFPdays 5 - 8

2nd IC clearancestudy: days 8/9

Colloid scan - 24 hours

Fio. 15.—Protocol for IC clearance studies in the C2-deficientpatient

This work has recently been reported in the Journal ofImmunology [43]. The patient described in the study hasbeen reported previously and has homozygous C2deficiency [42]. She is under the care of Dr KristianErlendsson, University Hospital, Rekjavik, Iceland. DrErlendsson kindly accompanied her on her visit tothe Hammersmith Hospital for the performance of thescans, and also supervised her therapy with FFP. Shewas 53 yr old at the time the study was performed. Heroriginal presentation in 1982 to the Department ofInternal Medicine and Clinical Immunology, Univer-sity Hospital, Rekjavik, was with a butterfly facial rash,severe Raynaud's, a symmetrical polyarthritis andalopoecia. The patient's symptoms initially improvedwith hydroxychloroquine and prednisolone, but in1988 she relapsed with a severe photosensitive rash anddigital gangrene. At this stage, therapy with regularinfusions of FFP (3 units administered daily for 4 days,every 6-8 weeks) was instituted. Each treatment re-sulted in marked improvement in her general well-being for up to 2 months, with rapid resolution on eachoccasion of her rash and digital vasculitic lesions.



A |C2-deficientpatient|100—1

60 —

60 —

40 '

2 0 -

o—'

— • — Ivor— • — btoodpoolNo splenic uptake

I I I I 1 I I0 10 20 30 40 50 60

time after injection (mlns)

lOO—i

80 —

88a*

40 —

20 -

I I I I I I0 10 20 30 40 50

time after Injection (mins)60

FIG. 16.—Kinetics of IC disposal in the C2-deficient patient beforetreatment (A) and in a normal subject (B).

Initiation of this treatment regime facilitated cessationof all drug therapy. At the time she was studied at theHammersmith Hospital, she was well, with no rash orother specific manifestations of her disease. She was

receiving FFP at ~8 week intervals, and was trans-ferred to our care for her first IC study 4 days beforeshe was due to receive a further series of infusions. Theprotocol for the study was as shown in Fig. 15. Controlsubjects were healthy volunteers from the laboratory,one male and one female, aged 34 and 32 yr, respect-ively.

In the IC scanning study performed before treatmentwith FFP therapy, there was rapid uptake of labelledcomplexes into the liver [tg^, (time for 90% up-t a k e ^ 13.6 min], corresponding to their rapid clear-ance from the circulation—tl/2 = 6.8 min (Fig. 16). Nouptake in the spleen was detected (Fig. 17). Between 30and 60 min, there was release of the tracer from theliver. This had been demonstrated in the previousstudies performed in lupus patients to be due to releaseof labelled ICs of intermediate size back into thecirculation. We did not observe this phenomenon in acontrol subject studied with the same labelled com-plexes on the same day. Values in the control subjectfor tm and r ^ were 12.5 and 26 min, respectively(Fig. 16B). A maximum of 21% of injected IC localizedin the spleen in the control subject.

The second IC scanning study was performed on day8, immediately after the completion of the last infusionof FFP. The observed pattern of complex uptake andprocessing was markedly different in this second study,resembling that observed in the control subject. /1/2 ofIC clearance was 9.8 min and tK^ was 27 min (Fig. 18).Twenty per cent of injected complexes localized to thespleen (Fig. 19), and there was no release between 30and 60 min.

We addressed the questions of whether there was ageneralized defect in the function of the mononuclearphagocytic system, or a reduction in splenic blood flowin the patient (as has been described in SLE pre-viously), which might explain the impaired splenicuptake of ICs. This was done by performing T c -colloid scans on days 2 and 9. The spleen/liver uptakeratio was 0.131 in the pre-treatment study and 0.132 inthe second study. Both values are within the normalrange for the tracer used. The two scans obtained were

FIG. 17.—Immune complex scan in the C2-deficient patient, before therapy (left: anterior, right posterior).



100 —i

80

•p 60120

liverblood poolspleen

I I I I I I0 10 20 30 40 50

lime after Injection (mlns)60

FKJ. 18.—Kinetics of IC processing in the C2-deficient patient aftertreatment with FFP.

almost identical in appearance and the computer-generated livenspleen profile curves were very similar(not shown).

Immune complex size was estimated prior to theirinjection and in plasma obtained from the patient at 3and 5min. In the C2-deficient patient, there was noobvious change in size subsequent to injection(Fig. 20A), while in the normal control, the majority ofthe labelled complexes detected in the circulation at5 min were smaller than those injected—demonstratedby a shift in the gradient peak in Fig. 22. FollowingFFP therapy, and the restoration of serum comp-lement, a similar shift could be demonstrated between0 and 5 min in the patient (Fig. 20C).

The binding of ICs to erythrocytes was measured inserial blood samples after IC injection. The clearanceof protein-bound tracer from the circulation, in thestudies performed before and after FFP therapy inthe patient, is shown in Fig. 21. Less than 2% ofthe injected complexes were bound to erythrocyte CR1at 2 min in the pre-treatment study, while 72% ofprotein-bound activity was measured on erythrocytesat 2 min in the study performed after normalization ofserum complement levels.

There was markedly reduced uptake and retention ofcomplexes in the spleen in the SLE patients, and thepre- and post-treatment studies in the Icelandic C2-deficient patient demonstrated for the first time inhumans that the uptake of ICs in the spleen is acomplement-dependent phenomenon. Similar obser-vations were made in the 1970s by Pepys and colleagues[44 46], and complement has been demonstrated in anumber of model systems to be important for thelocalization of certain antigens to follicular dendriticcells, and within splenic lymphoid follicles. In guinea-pigs, the spleen has also been shown to be of majorimportance for the clearance of complement-opsonizedbacteria [47]. Our observations relating to the com-plement-dependent nature of splenic IC processing inhumans have both immunopathological and clinicalimplications. Failure to localize ICs and their antigensin the spleen in SLE may have a bearing on theproduction of antibodies in this condition, either to selfor exogenous antigens. Efficient and safe processing ofcertain bacterial or viral antigens, or autoantigensderived, for example, from apoptotic leucocytes, maydepend on their uptake in the spleen, localization to theappropriate antigen-presenting cells and the rapid pro-duction of antibody. It is tempting to speculate thatpersistence of either exogenous or autoantigens as aresult of impaired clearance in the reticuloendothelialsystem might result in the presentation of antigen to'non-professional' antigen-presenting cells elsewhere inthe immune system, resulting in an abnormal cellularor humoral immune response. Alternatively, the 'plant-ing' of a persistent antigen, for example in the kidney,might then result in the in situ formation of potentiallyharmful ICs, as has been demonstrated in variousexperimental models of glomerulonephritis [48]. Ispeculated above that there may be a primary abnor-mality in Fc-mediated reticuloendothelial function inSLE, as has been suggested by Kimberly and hisco-workers, who performed many of the originalstudies of antibody-coated erythrocyte clearance. Theobservations that the handling of sulphur-colloid (pre-FFP therapy) was completely normal in our IcelandicC2-deficient patient, and that the abnormalities of both

FIG. 19.—Immune complex scan after FFP therapy in C2 deficiency flefu anterior; right posterior).



(A) Normal subject (B) C2-deficiency : pre- FFP (C) C2-deficiency : post- FFP

§ »i5

t - zto5 minutes 80 —

t - zero5 minute*

12Fraction nunib#f

r \ i r0 4 6 12

Fraction rtumbar

r i i i i0 4 e 12 16

Fraction number

Fio. 20.—IC size estimation by sucrose deniity-gradient centrifugation in the C2-deficient patient before and after treatment with fresh frozenplasma (FFP) (B and Q , and in a normal control subject (A).

the kinetics and sites of IC clearance seen pre-treatmentwere totally correctable by normalizing her classicalpathway complement function with FFP, would seemto mitigate against there being a primary abnormalityof reticuloendothelial function, at least in this patient.

CLINICAL IMPLICATIONSSLE and infection

What are the clinical implications of the observationsmade in our studies of in vivo IC processing in SLE andC2 deficiency? Firstly, we might consider to whatextent our observations may be extrapolated to theclearance of ICs comprising antibody and bacteria.Infection is a major cause of morbidity in patients withSLE, and in two recent large studies of mortality inSLE, sepsis was implicated as a major cause of death[49, 50]. More specifically, there are many reports in theliterature of fatal overwhelming infections with bothStreptococcus pneumoniae and Neisseria meningitidis, inpatients with SLE [51]. While steroids and immunosup-pressive therapy may be contributory factors in thesecases, it is possible that functional hyposplenism, re-lated to defective splenic clearance of bacterial ICs,may also also of major importance. It is particularlyinteresting in this context that SLE patients arc suscep-tible to a very similar spectrum of infections by bac-terial pathogens, as patients who have had a surgicalsplenectomy, notably pneumococci. Pathogenic S.pneumoniae are encapsulated, and are resistant to directcomplement-mediated opsonization and phagocytosis[52]. The pneumococcal cell wall can bind opsonicallyactive C3b and iC3b, but there is evidence that thecapsule masks polymorph access to the bound com-plement compo-nents. It is tempting to speculate that theprocessing of these paniculate pathogens in the spleenis critical both for their removal from the circulation inthe event of bacteraemia, and for the development of anappropriate antibody response. Hypccomplementaemiawill result in defective opsonization, with both a re-duction in the local phagocytosis and delivery to themain site of processing of paniculate antigens, the

(A) Before FFP infusion

50000 —

<D 40000 —

jjj 30000 —Q.

f 20000

10000 —

o—'

Protein-boundactivityRed celt-boundactivity

r i0 .10 20

(B) After FFP infusion

100000 —i

« 800003

30 40 50

I60

&

60000 —

40000 •

20000

o—'

Protein-boundactivityRed cefl-boundactivity

I I I I I I I0 10 20 30 40 50 60

Time after injection (mins)

FIG. 21.—Protein- and red cell-bound activity following injection ofIC; C2-deficient patient before (A) and after FFP (B).

spleen. Genetic deficiency of plasma complement pro-teins (of the dassicaL alternative and terminal pathways[1, 53, 54]), or of leucocyte complement receptorsCR3 and CR4 (as in 'leucocyte adhesion deficiency',


DAVIES: COMPLEMENT, IMMUNE COMPLEXES AND SLE 17

Fio. 22.—The binding of bacterial complexes comprising the IgM anti-lipid A antibody E5 and Salmonella minnesola to monocytes is visualizedwith a FTTC-conjugated anti-/j reagent (from Seelen et al. [62], with permission).

LAD [55]) arc strongly associated with the developmentof severe infections, and we suggest that the acquiredhypocomplementaemia of SLE may also put patients atrisk of serious infection. Do we have any direct evid-ence for this suggestion? We have recently identified astrong association between hypocomplementaemia inSLE and the presence of antibodies to the first com-plement component, Clq [56]. The latter are found inaround 25-30% of patients with SLE [57], and we arecurrently studying both the diagnostic utility and po-tential pathogenetic role of this autoantibody in SLE.Our preliminary findings were presented at the recentBritish Society for Rheumatology Annual Meeting[58]. One alarming, though perhaps not entirely sur-prising, association which we identified with anti-Clqantibodies and low complement levels was with suddenseptic death in patients with SLE. Of the four patientswith SLE from our unit who have died over the last 5 yrfrom infection, one had homozygous deficiency of Clq(reported recently in the Quarterly Journal of Medicine[59]), and the other three had persistently low com-plement levels and high-titre anti-Clq antibodies.

We have therefore recently suggested that patientswith SLE and persistently low complement levels betreated with prophylactic penicillin, and considered forpneumococcal and meningococcal vaccination [60],and this is now our unit policy.

It is worth noting that one of the four lupus patientswho died succumbed not from a pneumococcal orneisserial infection, but with septicaemia due to Pseu-domonas. Two research fellows working in our labor-atory for brief periods under my supervision—MarcoTonoli from Verona, and Marc Seelen from ProfessorDaha's laboratory in Leiden—have recently been ex-ploring the role of complement and both CR1 and CR3in the processing of Gram-negative organisms in arange of in vitro assay systems. Dr Tonoli demon-

strated that a monoclonal antibody HA-1 A, used in thetherapy of septic shock, could mediate complementfixation on red cells and binding to CR1 [61] and, more

BP An»-TNF FFP+ FFP

tt 1

i • i • i " i • r ' i • - i - i0 10 20 160 191 102 194 IBS 200 530

time after admission (houre)

Fio. 23.—Rises in endotoxin and terminal complement pathway ac-tivity were detected after FFP infusion in a C6-deficient patient withmeningococcal meningitis (from Lehner et al. [64], with permission).



recently, we have shown that complement, fixed byboth alternative and classical pathways, can directlymediate binding of Gram-negative bacteria to leuco-cyte CR3 [62], as shown in Fig. 22. It should not beforgotten that one of the main physiological roles ofthe complement system is in defence against infection,and our results would suggest that in hypocomplement-aemic states there will be increased susceptibility toboth Gram-positive and Gram-negative infection.

Is there a role for repletion of complement proteins in lupus?The second clinical question which we might con-

sider relates to the use of complement proteins in thetreatment of SLE. Should replacement therapy beconsidered in all deficient patients? Would it be appro-priate to use regular therapy, say with FFP, in allpatients who are chronically hypocomplementaemicwith SLE? Is there a role perhaps for soluble CR1 inpatients who have low levels on their erythrocytes?There are two major considerations to be taken intoaccount when trying to answer these questions, (i) Dowe know how often treatment should be given, andhow much would be needed? (ii) Are there any poten-tial risks to a therapeutic approach of this kind? OurIcelandic patient clearly benefited from treatment withFFP for some weeks, even though her classical path-way activity was normalized for only a few days. Mostcomplement proteins turn over at ~ 1 %/h, and infu-

sion of sufficient FFP to normalize C3, or C4 concen-trations, would only result in a similarly transientcorrection of the plasma level of the protein. In theC2-deficient case, we assume that correction ofhypocomplementaemia may have resulted in mobiliz-ation of ICs from sites of inflammation, notably theskin, and facilitated their effective processing by thereticuloendothelial system. It is possible that it thentakes some weeks for the further generation and depo-sition of ICs to occur and produce local inflammation.These mechanisms are poorly understood, and it wouldbe very difficult to estimate how frequently therapywould be indicated in any one patient with SLE andgenetic or acquired complement deficiency. To normal-ize classical pathway activity on a long-term basisrequires treatment with many units of plasma every fewdays. It is unlikely that it would be commercially viablefor a pharmaceutical company to produce a purified orrecombinant protein to be used in the treatment spe-cifically of complement-deficient patients, in view of therarity of these patients. The regular use of the largequantities of FFP required would be both prohibitivelyexpensive, and expose the patients to the risks ofacquiring retroviral, hepatitis C or other infections. Inaddition, there are other immunologically importantrisks of such therapy. It should not be forgotten thatcomplement, in addition to facilitating the processingof ICs, as we have demonstrated in the work described

Erythrocytes -1O8.'mT

une complexes

I IL-1 50u/mT

10 - I

8 -

o °929- 4

CO 2 -

T

TI T

CR1 per RBCPMN + E with varying CR1 numbers852 543 298 212 134 111

OC+ O

Q. a

FIG. 24.—E-CRl can protect the endothelium from PMN and IC-mediated damage (from Beynon et al. [66], with permission).


DAVIES: COMPLEMENT, IMMUNE COMPLEXES AND SLE 19

Immune complex Polymorphonuciearleucocyte

Opsonisedimmune complex

RBC bearingCRl clusters

FIG. 25.—Erythrocyte CRl can protect the endothelium from neutrophil and IC-mediated injury, by binding opsonized complexes and keepingthem in the central jet stream (from [68], with permission).

in this essay, also has a major role mediating theinflammatory process, both in autoimmune and infect-ive diseases. Complement deposition is a characteristicfeature of glomerulonephritis in SLE, and C3 productscan readily be demonstrated in inflamed skin, or in thesurface of erythrocytes in patients with SLE andhaemolytic anaemia. There is good evidence fromstudies performed in genetically C3-deficient dogs withnephritis that the disease is exacerbated by plasmainfusion and the normalization of complement levels[63]. We have also demonstrated in humans thatproviding a missing complement component may havepotentially adverse consequences. In a patient withhereditary C6 deficiency and meningococcal menin-gitis, we showed that infusion of FFP resulted in aclinical deterioration rather than an improvement, andthat this was associated with a rise in endotoxin levels[64] (Fig. 23). This patient also developed antibodies toC6, which resulted in a less marked response to laterinfusions of plasma. We have also recently demon-strated the development of antibodies to the missingcomponent in the serum in a Clq-deficient patienttreated on multiple occasions with FFP. Subsequenttherapy with plasma at a time of an exacerbation of hislupus resulted in the development of an acute serumsickness-like illness [59]. Clearly, the development ofantibodies in this way might also reduce the efficacyof repeated therapy, as well as putting the patient atrisk of further IC-mediated disease.

Should we consider the use of recombinant solubleCRl (sCRl) in SLE? This compound has been used ina number of experimental models of inflammation(recently reviewed by Fearon and colleagues [65]),including cardiac allo- and xenografts, IC-mediatedvasculitis, thermal trauma, and both myocardial andintestinal reperfusion systems. Early evidence suggeststhat sCRl may have a role in reducing inflammationacutely, and in protecting grafts, and a trial of its usein acute myocardial infarction in humans is under way.We have recently published work which suggests thatCRl on erythocytes may have a direct role in protect-ing the endothelium from IC and neutrophil-mediateddamage [66]. In these studies, a model system wasdeveloped by Huw Beynon in our laboratory, in whichhe assessed endothelial damage by growing- the cellsdirectly on a fluorescein isothiocyanate (FITC) matrix,and measuring the effect of different potentially phlo-gistic stimuli on the access of a radiolabelled anti-FITCantibody [67]. An increase in permeability in thissystem equates with greater cell damage. Figure 24,taken from Dr Beynon's recent paper in the Journal ofImmunology, demonstrates the protective effect of redcell CRl on endothelial cells cultured in the presenceof IL-1, neutrophils and ICs.

A detailed discussion of leucocyte-endothelial inter-actions in SLE is clearly beyond the scope of this essay,but we have reviewed the subject recently [68]. Wesuggest that one of the functions of CRl may be to



'capture' ICs in the circulation, and keep them in thefaster flowing central jet stream in the blood, awayfrom the endothelial surface where, in association withneutrophils, they have the potential to cause damage(Fig. 25). Physiologically, it is erythrocyte-bound CRlwhich subserve this function, but it is possible thatsCRl, administered in sufficient dose, might be able towork in a similar way. The most likely clinical situationin which one might consider this approach is in thecontext of an acute inflammatory exacerbation ofthe condition. However, we do not know whetherthe soluble form of CRl would be as effective aserythrocyte CRl which, as we have discussed pre-viously, is clustered, and has very high affinity for ICs,and many of the same questions which we raised in thecontext of repletion therapy for deficient plasma pro-teins—e.g. how much, and how often?—are difficult toanswer. The data we obtained in the radioimmunother-apy patients suggested that CRl may be rapidly proteo-lytically degraded when large numbers of ICs arepresented to the fixed macrophage system, and if asimilar effect were observed with sCRl, it is likely thatit would only be of transient benefit. It is perhaps alsoworth sounding a note of caution regarding the extra-polation of observations made in animals to the human

situation. There is a great deal of interest at present inthe use of pig organs in xenotransplantation, and thepossible role of sCRl or other complement-inhibitorymolecules in prolonging graft survival. We have re-cently studied the clearance of soluble model ICs(HBsAg:pig anti-HBsAg) in the pig (work reported atthe recent European Complement Workshop inSwitzerland [69]) and, to our surprise, we found that ICinitially localized primarily to the lungs, rather than tothe liver and spleen, as in man, and that this uptake washighly complement dependent, reflecting what wouldappear to be fundamentally different organization andfunction of the reticuloendothelial system in thisspecies.

CONCLUDING REMARKSOn the basis of the work described in this essay, is

it possible to formulate a unifying hypothesis to explainthe role of complement and CRl in pathological ICprocessing? An attempt is made in Fig. 26, in which wecompare 'physiological' IC processing with the situ-ation in SLE. We postulate that IC formation, eitherat sites of tissue injury or in the circulation, may causecomplement activation and the deposition of C3 frag-ments on erythrocytes, and the subsequent clearance of

IC PROCESSING IN SLE PHYSIOLOGICAL IC PROCESSING

!

Antigen release

Tissuedamage

tRelease of

immune complexesA

iLow affinitybinding and

reduced splenicuptake

Stimulation ofantibody production

JIMMUNE COMPLEX

FORMATION11

Systemic or localcomplement activation :Ucomplement H E-CR1

I*Poorly opsonlsed

immune complexes|

TDelivery to fixedM$ of MPS influid phase

J

Opsonisation with C3and lattice modfflcatton

Birxfing of complexto erythrocyte

CR1

Disposal ofantigen and ICs

Effective processingin the liver and

spleen

Delivery to fixedM^ofMPS

bound to E-CR1

Binding tocomplement and

Fc receptors

L JFta. 26.—Proposed schema for the immunopathogenesis of SLE.



these IC may result in a fall in erythrocyte CR1. Theseabnormalities may then result in defective furtherprocessing of complexes, as a consequence of pooropsonization, failure of modification of IC structureand non-physiological delivery to the fixed macrophagesystem, as a result of impaired binding to CR1. Suchdefective processing may, in itself, result in the persist-ence of potentially harmful complexes, causing tissuedamage, further systemic and local complement ac-tivation, and the development of a positive feedbackloop stimulating further antigen release and autoanti-body production. Data have also been presented thatsuggest that erythrocyte CR1 may have a specific roleto play in the protection of the endothelium from directneutrophil and IC-mediated damage, and the acquiredfall in CR1 numbers observed in SLE would clearlyresult in a reduction in the efficacy of such a protectivemechanism. We also suggest that hypocomplement-aemia, a reduction in erythrocyte CR1 and functionalhyposplenism may predispose certain patients withSLE to infection. We are currently involved in explor-ing these hypotheses further, both in humans and in thepig, and in a mouse model of Clq deficiency, which isbeing developed by Marina Botto in our laboratoryusing gene 'knockout' technology.

It remains our hope that the development of a fullerunderstanding of the relationships between com-plement proteins and receptors, ICs, and the manyother components of the cellular and humoral immunesystems in SLE, will one day enable us to devise morerational and effective treatments for this most fascinat-ing, but frustrating, rheumatic disease.

ACKNOWLEDGEMENTS

All the work described was supported by theArthritis and Rheumatism Council, who remain mostgenerous supporters of both myself and the RPMSRheumatology Unit. I would like to acknowledge theguidance and support of Professor Mark Walport, andam also grateful to Dr Huw Beynon, Dr Jurg Schifferli,Dr John Savill, Dr Mike Peters, Dr Chris Bunn, DrSozos Loizou, Dr Dorian Haskard and Mr PeterNorsworthy for their advice and practical assistance.Figures 6-14 appear with kind permission of the Editorat the Journal of Clinical Investigation, and Figures 2-5,16-21 and 24 with kind permission of the Editor of theJournal of Immunology.

REFERENCES

1. Davies KA, Schifferli JA, Walport MJ. Complementdeficiency and immune complex disease. Springer SeminImmunopathol 1994;15:397-^16.

2. Madi N, Paccaud J-P, Steiger G, Schifferli JA. Immuneadherence of nascent hepatitis B surface antigen-anti-body complexes in vivo in humans. Clin Exp Immunol1989;7&201-6.

3. Haakenstad AO, Mannik M. Saturation of the reucu-locndothelia) system with soluble immune complexes.J Immunol 1974;112:1939-48.

4. Frank MM, Lawley TJ, Hamburger MI, Brown EJ.Immunoglobulin G Fc receptor-mediated clearance inautoimmune diseases. Ann Intern Med 1983;9&206-18.

5. Lachmann PJ, Walport MJ. Deficiency of the effectormechanisms of the immune response and autoimmunity.In: Whelan J, ed. Autoimmunity and autoimmune disease.Ciba Foundation Symposium # 129. Chichesten Wiley,1987:149-71.

6. Lobatto S, Daha MR, Breedveld FC et al. Abnormalclearance of soluble aggregates of human immunoglobu-lin G in patients with systemic lupus erythematosus. ClinExp Immunol 1988;72:55-9.

7. Lobatto S, Daha MR, Voetman AA et al. Clearance ofsoluble aggregates of immunoblobulin G in healthyvolunteers and chimpanzees. Clin Exp Immunol1987;68:133.

8. Schifferli JA, Ng YC, Paccaud J-P, Walport MJ. The roleof hypocomplementaemia and low erythrocyte comple-ment receptor type 1 numbers in determining abnormalimmune complex clearance in humans. Clin Exp Immunol1989;75:329-35.

9. Kimberly RP, Ralph P. Endocytosis by the mononuclearphagocyte system and autoimmune disease. Am J Med1983;74:481-93.

10. Atkinson JP, Frank MM. Studies on the in vivo effects ofantibody. Interaction of IgM antibody and complementin the immune clearance and destruction of erythrocytesin man. J Clin Invest 1974^4:339-48.

11. Frank MM, Hamburger MI, Lawley TJ, Kimberly RP,Plotz PH. Defective reticuloendothehal system Fc-recep-tor function in systemic lupus erythematosus. N EnglJ Med 1979^00:518-23.

12. Hamburger MI, Lawley TJ, Kimberly RP, Plotz PH,Frank MM. A serial study of splenic reticuloendothelialsystem Fc-receptor function in systemic lupus erythe-matosus. Arthritis Rheum 1982;25:48.

13. Kimberly RP, Parris TM, Inman RD, McDougal JS.Dynamics of mononuclear phagocyte system Fc receptorfunction in systemic lupus erythematosus. Relation todisease activity and circulating immune complexes. ClinExp Immunol 1983;51:261-8.

14. Parris TM, Kimberly RP, Inman RD, McDougal S,Gibofsky A, Christian C. Defective Fc-receptor mediatedfunction of the mononuclear phagocyte system in lupusnephritis. Ann Intern Med 1982,-97:526.

15. Valentijn RM, van Overhagen H, Hazevoet HM et al.The value of complement and immune complex deter-minations in monitoring disease activity in patients withsystemic lupus erythematosus. Arthritis Rheum 1985;2&904-13.

16. Walport MJ, Peters AM, Elkon KB, Pusey C, LavenderJP, Hughes GRV. The splenic extraction ratio of anti-body-coated erythrocytes and its response to plasmaexchange and pulse methylprednisolone. Clin ExpImmunol 1985;«k465-73.

17. Halma C, Daha MR, van Furth R et al. Elimination ofsoluble mI-labelled aggregates of human immunoglobu-lin G in humans; the effect of splenectomy. Clin ExpImmunol 1989;77:62-6.

18. Davies KA, Walport MJ, Schifferli JA. Maintenance ofimmune complex solubility and immune adherence In:Rother, Hfinsch and Till, eds. The complement system.Heidelberg: Springer-Verlag, 1994.

19. Varga L, Thiry E, Fust G. BSA-anti-BSA immunecomplexes formed in the presence of serum do not bindto autologous red cells. Immunology 1988;64:381-6.

20. Edberg JC, Kujala GA, Taylor RP. Rapid immuneadherence reactivity of nascent, soluble antibody/DNAimmune complexes in the circulation. / Immunol 1987;139:1240-5.



21. Germuth FG. A comparative histologic and immuno-logic study in rabbits of induced hypcrsensitivity of theserum sickness type. J Exp Med 1953^7^57-64.

22. Davies KA, Matbieson P, Winearls CG, Rees AJ,Walport Ml. Serum sickness and acute renal failure afterstreptokinase therapy for myocardial infarction. ClinExp Immunol 199O;8O:83-8.

23. Walport MJ, Ross GD, Mackworth-Young C, WatsonJV, Hogg N, Lachmann PJ. Family studies of erythrocytecomplement receptor type 1 levels: reduced levels inpatients with SLE are acquired, not inherited. Clin ExpImmunol 1985^9:547-54.

24. Wilson JG, Wong WW, Schur PH, Fearon DT. Mode ofinheritance of decreased C3b receptors on erythrocytes ofpatients with systemic lupus erythematosus. N EnglJ Med 1982^07^81-6.

25. Iida K, Mornaghi R, Nussenzweig V. Complement re-ceptor (CR1) deficiency in erythrocytes from patientswith systemic lupus erythematosus. J Exp Med 1982;155:1427-38.

26. Giles CM, Davies KA, Walport MJ. In vivo and in vitroC4 molecules on red cells; a correlation of numbers ofmolecules and haemagglutination. Transfusion 1991;31:222-8.

27. Giles CM, Davies KA, Loizou S, Moulds JJ, WalportMJ. Quantification of IgG on erythrocytes of patientsand normals by radio-ligand assay. Transfus Med 1991;1:223-8.

28. Hammond A, Rudge AC, Loizou S, Bowcock SJ, Wal-port MJ. Reduced numbers of complement receptor type1 on erythrocytes are associated with increased levels ofanticardiolipin antibodies. Arthritis Rheum 1989;32:259-64.

29. Davies KA, Hird V, Stewart S et al. A study of in vivoimmune complex formation and clearance in man.J Immunol 1990;144:4613-20.

30. Stewart JSW, Sivalopenko GB, Hird V, Davies KA,Walport MJ, Ritter MA, Epenetos AA. Clearance of'"I-labclled murine monoclonal antibody from patients'blood by intravenous human anti-murine immunoglobu-lin antibody. Cancer Res 1990^0:563-7.

31. Paccaud J-P, Carpenticr J-L, Schifferli JA. Direct evi-dence for the clustered nature of complement receptorstype 1 on the erythrocyte membrane. J Immunol 1988;141:3889-94.

32. Cosio FG, Shen XP, Hebert LA. Immune complexesbind preferentially to specific subpopulations of humanerythrocytes. Clin Immunol Immtmopathol 1990;55:337-54.

33. Barbosa JE, Harrison RA, Backer PJ, Lachmann PJ. Anantipeptide antibody that recognizes a neoantigen in theCR1 stump remaining on E after proteolysis. Clin ExpImmunol 1992;87:144-9.

34. Tausk FA, McCutchan JA, Schrciber RD, Spechko P,Gigli I. Deficiency of erythrocyte C3b receptor (CR1) inAIDS and AIDS-rclated syndromes. Biosci Rep 1986;6:81-6.

35. Jouvin MH, Rozenbaum W, Russo R, Kazatchkine MD.Decreased expression of C3b/C4b receptor (CR1) onerythrocytes in AIDS and AIDS-related syndromes cor-relates with clinical sub-populations of patients with HIVinfection. AIDS 1987;l:89-94.

36. Spycher MO, Spath PJ. Quantification of complementreceptor 1 on E; follow up of HIV-1 infected patientswith AIDS-related complex and Walter-Red 5 undertreatment with intravenous immunoglobulin. Vox Sang1990;(suppl. ll):44-50.

37. Cohen JHM, Geffriaud C, Caudwell V, KazatchkineMD. Genetic analysis of CR1 (the C3b complementreceptor, CD35) expression on erythrocytes of HIV-infected individuals. AIDS 1989;3:397-9.

38. Pascual M, Danielsson C, Steiger G, Schifferli JA. Pro-teolytic cleavage of CR1 on human erythrocytes in vivo.Evidence for enhanced cleavage in AIDS. Ew J Immunol1994;24:702-8.

39. Reay P. Use of N-bromosuccinamide for the iodinationof proteins for radioimmunoassay. Ann Clin Biochem1982;19:129-33.

40. Davies KA, Peters AM, Beynon HLC, Walport MJ.Immune complex processing in systemic lupus erytbe-matosus—in vivo imaging and clearance studies. J ClinInvest 1992;90:2075-83.

41. Hogg N, Ross GD, Jones DB, Slusarenko M, WalportMJ, Lachmann PJ. Identification of an anti-monocytemonoclonal antibody that is specific for membranecomplement receptor type one (CR1). Eur J Immunol1984;14:236-43.

42. Steinsson K, Erlendsson K, Valdimarsson H. Successfulplasma infusion treatment of a patient with C2 deficiencyand systemic lupus erythematosus: clinical experienceover forty-five months. Arthritis Rheum 1989;32.-906-13.

43. Davies KA, Erlendsson K, Beynon HLC, Peters AM,Valdimarsson H, Walport MJ. Splenic uptake of immunecomplexes in man is complement dependent. / Immunol1993;151:3866-73.

44. Pepys MB. Role of complement in induction of antibodyproduction in vivo. Effect of cobra factor and otherC3-reactive agents on thymus-dependent and thymus-independent antibody responses. J Exp Med 1974;140:126^15.

45. Pryjma J, Humphrey JH. Prolonged C3 depletion bycobra venom factor in thymus-deprived mice and itsimplication for the role of C3 as an essential secondsignal for B-cell triggering. Immunology 1975^8:569-76.

46. Klaus GGB, Humphrey JH. The generation of memorycells. I. The role of C3 in the generation of B memorycells. Immunology 1977^3:31-40.

47. Hosea SW, Brown EJ, Hamburger MI, Frank MM.Opsonic requirements for intravascular clearance aftersplenectomy. N Engl J Med 1981;304:245-50.

48. Couser WG, Salant DJ. In-situ immune complex forma-tion and glomerular injury. Kidney Int 1980;17:l-13.

49. Cohen MG, Li EK. Mortality in systemic lupus erythe-matosus: active disease is the most important factor. AustN ZJ Med 1992^2:5-8.

50. Breban M, Meyer O, Bourgeois P, Palazzo E, Kahn MF.The actuarial survival rate in systemic lupus erythemato-sus: study of a 1976 cohort. Clin Rheumatol 1991;l(h283-8.

51. Mitchell SR, Nguyen PQ, Katz P. Increased risk ofneisserial infections in systemic lupus erythematosus.Semin Arthritis Rheum 1990^0:174-84.

52. Moffit MC, Frank MM. Complement resistance inmicrobes. Springer Semin Immunopathol 1994;15:327--44.

53. Vyse TJ, Spath PJ, Davies KA et al. Hereditary comp-lement factor I deficiencies: 4 case reports and reviewof hereditary complement factor I, factor H and C3deficiencies. Q J Med 1994;87:385-402.

54. Lachmann PJ. Complement deficiencies—genetic andacquired. In: Lachmann PJ, Peters DK, Rosen F, Wal-port MJ, eds. Clinical aspects of immunology, 5th edn.Oxford: Blackwell Scientific Publications, 1993:1287-304.



55. Davics KA, Toothill VJ, Savill J et al. A 19-year old boywith leucocyte adhesion deficiency. In vitro and in vivostudies of leucocyte function. Clin Exp Immunol 1991;84:223-31.

56. Davies KA, Norswrthy PJ, Loizou S, Walport MJ.Anti-Clq antibodies and SLE. Clin Exp Immunol 1994;97:28.

57. Wisnieski JJ, Jones SM. IgG autoantibody to collagen-like region of Clq in hypocomplementemic urticarialvasculitis syndrome, systemic lupus erythematosus, and 6other musculoskeletal or rheumatic diseases. J Rheumatol1992;19:884-8.

58. Davies KA, Norsworthy PJ, Athanassiou P, Mason PD,Loizou S, Walport MJ. Anti-Clq antibodies in SLE. BrJ Rheumatol 1994;33:96.

59. Bowness P, Davies KA, Norsworthy PJ et al.Hereditary Clq deficiency and systemic lupus erythe-matosus. Q J Med 1994;87:455-64.

60. Davies KA, Beynon HLC, Walport MJ. Consider pro-phylaxis in systemic lupus erythematosus. Br MedJ 1994^08:133.

61. Tonoli M, Davies KA, Norsworthy PJ, Cohen J, Wal-port MJ. The anti-lipid A antibody HA-1A binds torough Gram-negative bacteria, fixes complement andfacilitates binding to erythrocyte CR1 (CD35). Clin ExpImmunol 1993;92:232-8.

62. Seelen MA, Athanassiou P, Lynn WA et al. The anti-lipid A monoclonal antibody E5 binds to rough Gram-negative bacteria, fixes C3, and facilitates binding ofbacterial immune complexes to both erythrocytes andmonocytes. Immunology 1995;84:653-61.

63. Cork CL, Morris JM, Olson JL, Krakowka S, Swift AJ,Winkelstein JA. Membranoproliferative glomerulo-nephritis in dogs with a genetically-determined deficiencyof the third component of complement. Clin ImmunolImmunopathol 1991;6(h455-70.

64. Lehner PJ, Davies KA, Cope AP et al. Meningococcalsepticaemia in a C6-deficient patient and effects ofplasma infusion on lipopolysaccharide release. Lancet1992^40:1379-81.

65. Kalli KR, Hsu P, Fearon DT. Therapeutic uses ofrecombinant complement protein inhibitors. SpringerSemin Immunopathol 1994; 15:417-31.

66. Beynon HLC, Davies KA, Haskard DO, Walport MJ.Erythrocyte complement receptor type 1 and interactionsbetween immune complexes, neutrophils and endo-thelium. J Immunol 1994;153:3160-7.

67. Beynon HLC, Haskard DO, Davies KA, Haroutunian R,Walport MJ. Combinations of low concentrations ofcytokines and acute agonists synergise in increasing thepermeability of endothelial monolayers. Clin ExpImmunol 1993,-91:314-9.

68. Beynon HLC, Athanassiou P, Davies KA. The endo-thelium in SLE and Sjogren's syndrome. In: Savage COS,Pearson JD, eds. Immunological aspects of the vascularendothelium. Cambridge University Press, 1995:124-52.

69. Davics KA, Chapman PT, Norsworthy PJ, Jamar F,Athanassiou P, Keelan ETM, Harrison AA, Binns RM,Haskard DO, Walport MJ. Clearance pathways ofsoluble immune complexes in the pig. Insights intothe adaptive nature of antigen clearance in humans. /Immunol 1995; in press.


Documents

complement, immune complexes and systemic lupus erythematosus