the complement system: importance response to viral …the complement system is composed of a series...

MICROBIOLOGICAL REVIEWS, Mar. 1982, p. 71-85 Vol. 46, No. 10146-0749/82/010071-15$02.00/0

The Complement System: Its Importance in the HostResponse to Viral Infection

ROBERT L. HIRSCHHoward Hughes Medical Institute, Department of Neurology, The Johns Hopkins University School of

Medicine, Baltimore, Maryland 21205

INTRODUCTION ................................................................ 71THE COMPLEMENT SYSTEM: COMPONENTS AND ACTIVATION ..... ............ 71

Activation of the Classical Pathway ............. .................................. 71Activation of the Alternative Pathway ............ ................................. 72Terminal Lytic Events........................................................... 73

INTERACTION OF COMPLEMENT WITH VIRUS AND VIRUS-INFECTEDCELLS IN VITRO....................................................... 73

Antibody-Dependent, Complement-Enhancing Neutralization of Viruses ..... .......... 73Virolysis ................................................................. 73Complement-Enhancing Neutralization Without Virolysis ...... .................... 74

Antibody-Independent Activation of Complement by Virus........................... 75Antibody-Dependent and -Independent Interactions of Complement with

Virus-Infected Cells ....................................................... 77ROLE OF COMPLEMENT IN VIRAL INFECTIONS IN VIVO........................ 79ROLE OF THE COMPLEMENT SYSTEM IN VIRAL INFECTIONS OF HUMANS ...... 81SUMMARY ................................................................. 83LITERATURE CITED ............................................................ 83

INTRODUCTIONThe complement system is composed of a

series of serum proteins that function in a vari-ety of specific and nonspecific immune defensemechanisms. Upon interaction with activators,the complement system initiates or amplifiesevents that might normally occur at low levels.Activation of the complement system can beviewed as a series of enzymatic effects whichresult in limited proteolytic events, generation offragments, and biologically significant occur-rences, which in some manner modify the acti-vating substance or its environment. The acti-vating substance or particle, as the case may be,is either directly affected by the activation or isprepared for other arms of the immune re-sponse.

It is not the purpose of this review to detail thespecific proteins and regulatory mechanisms in-volved in the activation of the complement sys-tem, but rather to review what is known aboutthe interaction of the complement system withviruses and the role of the complement system inhost defense against viral infection. Nonethe-less, a brief summary of the salient features ofthe complement system is a requisite for acritical discussion of the interaction of the com-plement system with viruses.

THE COMPLEMENT SYSTEM:COMPONENTS AND ACTIVATION

The complement system can be activated viatwo mechanisms, the classical and alternative

pathways. Although these pathways share manycommon features and the biological result ofactivation of each pathway can be the same, thepathways are quite different.

Activation of the Classical PathwayThe activation of the classical pathway is most

often initiated by the binding of the first compo-nent of complement (Cl) to antigen-antibodycomplexes; however, other molecules, includingpolyanions, viral proteins, and lipid A of lipo-polysaccharides, may also mediate classicalpathway activation in the absence of antibody.Cl exists in the serum as a calcium-dependentcomplex of three subunits Clq, Clr, and Cls. Insituations involving antigen-antibody complex-es, Clq binds to the Fc region of immunoglob-ulin in the complexes. This binding initiates theclassical complement pathway cascade (Fig. 1).After the union of Clq to the antigen-antibodycomplex, Clr activates Cls to become activatedCls (Cls; by convention, a superscript bar des-ignates an active enzyme of the classical path-way) (reviewed in reference 20). This protease(Cls) cleaves C4 to yield a small-molecular-weight fragment, C4a, and the major cleavagefragment, C4b. The presence of C4b allows Clsto cleave C2, the major cleavage product ofwhich is C2a. C2a binds to C4b in the presenceof magnesium, and the C4b2a complex thusformed is known as the classical pathway C3convertase. The C3 convertase allows for subse-quent cleavage of C3 into C3a and C3b, and the

71

on July 18, 2020 by guesthttp://m

mbr.asm

.org/D

ownloaded from

http://mmbr.asm.org/

MICROBIOL. REV.

CLASSICAL PATHWAY

Ab, Clqrs

Cl qrs - C4, C2 t C4bW,2a (C3 convertase)

C3'-C3a

Biologically SignificantFragments & Events

Anaphylatoxin

Opsonin

C4b2a3b (C5 convertae)

C5b

+C5a Anaphylatoxin, Chemotactic Factor

! C6, 7,8,9 (Membrane Attck) + LYSIS

C5 - o C5a

C3bBbPC3b (C5 convertase)

C3 lb, C3a

C3bBbP (C3 convertase)

ALTERNATIVE PATHWAY

FIG. 1. Schematic representation of the complement system showing the two pathways by which microorga-nisms initiate complement activation and generate biologically important events. Binding of immune complexesto Clq via the Fc portion of an immunoglobulin molecule activates the classical pathway. Activated Cl (Clqrs)then activates C4 and C2. The bimolecular complex (C4b) in turn cleaves C3 into two fragments, C3a and C3b.C3b can either form part of the enzyme which cleaves C5 and initiates the membrane attack complex or it canbind to microbial surfaces and act as an opsonin. When tested in vivo, with the use of purified complementcomponents, activation of the ACP does not require immunoglobulin. The binding of C3b (generated either bythe inefficient interaction of C3, B, and D in the fluid phase or by action of the classical pathway C3 convertase)to a microbial surface initiates the ACP. The fixed C3b binds factor B and, unless stabilized by properdin (P),undergoes intrinsic decay. When stabilized, however, the C3bBb complex is an efficient enzyme and cleavesfurther C3 molecules (amplification loop). When an additional C3b binds to the C3bBbP complex, it acquires C5cleaving capability and generates the membrane attack complex.

activation of the terminal lytic mechanism of thecomplement system (see below). The classicalpathway C3 convertase is under at least threeforms of control. First, the enzyme is unstable at37°C; second, C3b inactivator (I) in the presenceof a cofactor, C4b binding protein, degrades C4bin the fluid phase, causing accelerated decay ofthe convertase; and finally, C4b binding proteinin the presence of I removes C4b from thesurface to which it is bound (17).

Activation of the Alternative PathwayThe alternative or properdin pathway of the

complement system is essentially a bypass sys-tem for the classical pathway, and its activationprobably does not require (but may be enhancedby) the presence of immunoglobulins. For thisreason and because of the fact that bacterial cellsurfaces and viruses are among the activators ofthe alternative complement pathway (ACP), it is

likely that the ACP evolved as a defense mecha-nism prior to the classical pathway. An essentialfeature of the ACP is the ACP amplification C3convertase. Upon binding of C3b, which is gen-erated by either the classical pathway or low-grade fluid-phase ACP activation , to an activat-ing surface, factor B binds to C3b to form C3bBcomplexes (Fig. 1). Factor D cleaves factor B toform C3bBb, which is the ACP C3 convertase.This process occurs at low levels in the fluidphase, but is amplified and controlled when C3bbinds to the surface of a particle. Properdin (P),like C3 nephritic factor (C3NeF, an anti-C3bBbglobulin present in the serum of patients withglomerulonephritis), stabilizes the complex suchthat further cleavage of C3 (amplification) canoccur via the ACP C3 convertase (C3bBbP orC3bBbC3NeF). The C3bBb enzyme is con-trolled by four mechanisms. The complex isunstable, and therefore intrinsic decay occurs

72 HIRSCH


mbr.asm

.org/D

ownloaded from


COMPLEMENT SYSTEM IN RESPONSE TO VIRAL INFECTION

readily at 37°C. H, a serum glycoprotein, canbind to C3b and impair the binding of B; H canincrease the rate of Bb dissociation from C3b;and H can increase the susceptibility of C3b tothe action of I. The action of H and I on theC3bB complex depends on its ability to bind toC3b, which is directly related to the levels ofsialic acid on the particle surface. It is likely thatsurfaces with low levels of sialic acid are able toprotect, in some manner, the ACP C3 conver-tase from degradation by H and I. Thus, theability of a particle to sustain ACP activation isinversely related to the amount of surface sialicacid (16). When H does bind to the C3bBcomplex, C3b is exposed and the cleavage of theconvertase by I is enhanced, resulting in disso-ciation of the complex to C3bi (cleaved andinactive form of C3b) and Bbi (inactive Bb,unable to hydrolyze C3 and C5) (16).

Terminal Lytic EventsAs outlined above, C3 can be cleaved by

either the classical C3 convertase (C4b2a) or bythe ACP C3 convertase (C3bBb). The C4b2a3benzyme is formed when C3 is cleaved via theclassical pathway. This enzyme cleaves C5 intoC5a and C5b. C6 binds to and stabilizes C5b(otherwise, dissociation ofC5b from the C4b2a3benzyme occurs). The C5b6 complex then reactswith C7 to form a trimolecular complex, C5b67.Upon interaction of this complex with C8 and C9(for example, on the surface of an erythrocyte),lysis occurs. The mechanism of the lytic attackthrough the classical pathway has been reviewedelsewhere (34, 35). The initiation of the lyticmechanism through the ACP occurs when theACP C3 convertase, upon stabilization by pro-perdin and binding of an additional C3b, ac-quires the ability to cleave C5. The remainingsteps in the lytic mechanism are presumably thesame as those in the classical pathway (55).An important event of biological significance

associated with activation of the complementsystem is the generation of the membrane attackcomplex and subsequent lysis of lipid bilayers(membranes). However, the generation of sever-al complement fragments as a result of comple-ment activation is also of great biological impor-tance (Fig. 1). The C3a and C5a fragments caninduce histamine release, and C5a can also pro-mote leukocyte chemotaxis. C3b and, to a lesserextent, C4b are opsonins and, when bound tothe surface of a microbe, can facilitate thebinding of microbes to lymphocytes or phago-cytic cells. The Bb fragment of factor B of thealternative pathway can induce macrophagespreading and inhibit macrophage mobility,whereas the Ba fragment can induced chemotax-is of polymorphonuclear leukocytes. Other re-cent reviews have detailed the biological activi-

ties of complement components and fragment(27, 53).INTERACTION OF COMPLEMENT WITHVIRUS AND VIRUS-INFECTED CELLS IN

VITROOver the past 50 years, scattered reports in

the literature have indicated that the comple-ment system, either independently or in thepresence of antibody, plays an important role inhost defense against virus infection. Douglasand Smith showed that a heat-labile componentof rabbit serum enhanced the interaction ofvaccinia virus with the cellular elements ofblood (14). Mueller demonstrated that freshguinea pig serum enhanced the activity of im-mune serum against Rous sarcoma virus andsuggested that this enhanced activity was due tocomplement (40). Morgan (39) and Whitman (67)showed that unheated rabbit serum, containingantibody to western equine encephalomyelitisvirus, had a higher neutralizing antibody titerthan serum heated at 56°C for 30 min. Further-more, addition of unheated fresh serum fromseveral species increased neutralizing antibodytiters. These were the first of many studiesindicating that a heat-labile serum factor wasrequired for or enhanced antibody-mediated vi-rus neutralization.

Antibody-Dependent, Complement-EnhancingNeutralization of Viruses

The neutralization of virus by antibody is amultifaceted reaction which is dependent uponthe virus, the class and concentration ofimmunoglobulin, and the host cell under study(11). Antibody may prevent viral adsorption to acell surface by covering critical sites, by causingaggregation, or by altering the charge of a virus.Antibody may also influence infection of a cellafter virus adsorption to a cell surface by inhibit-ing virus penetration or uncoating of viral nucle-ic acid, or both (11). Antibody may also enhanceuptake and phagocytic destruction of virus by acell. Since antibody may influence viral infectiv-ity by several mechanisms, the complementsystem may also act to enhance neutralization inseveral ways.

Virolysis. One mechanism by which comple-ment has been shown to enhance antibody-mediated virus neutralization is by virolysis,actual destruction of virion integrity. This phe-nomenon has been demonstrated for severaldifferent classes of enveloped viruses. Electronmicroscopic evidence for virolysis was first pre-sented by Berry and Almeida, using avian infec-tious bronchitis virus and a rabbit anti-avianinfectious bronchitis virus serum (5). Stollardemonstrated that an antibody-coated togavirus,

73VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


74 HIRSCH

Sindbis virus, when exposed to a complementsource, was lysed, as determined by the releaseof the radioactively labeled viral ribonucleic acid(RNA) (60). Biochemical and electron micro-scopic studies showed that antibody and com-plement also lysed an arenavirus, lymphocyticchoriomeningitis virus (LCMV) (66). Studies onthe neutralization of equine arteritis virus (EAV)by late (immunoglobulin G, IgG) antibody raisedin several species demonstrated that neutraliza-tion was dependent upon complement in thatonly partial neutralization occurred after a pro-longed incubation in the absence of complement(50). The release of labeled RNA from antibody-coated EAV occurred upon addition of equine orguinea pig complement (52), indicating that theenhanced neutralization by late antiserum pro-ceeded through virolysis. However, other evi-dence indicated that complement interacted withIgG-sensitized EAV to induce structural alter-ations in the absence of virolysis. Complement-antibody-neutralized EAV complexes weretreated with trypsin, and 32% of original EAVinfectivity was recovered. Subsequently, 89% ofthe recovered infectivity was destroyed by treat-ment with ribonuclease, suggesting that a frac-tion of the EAV suffered structural damage yetwas still infectious after treatment with comple-ment. Control, non-complement-treated, sensi-tized EAV, when treated wth anti-IgG and sub-sequently treated with trypsin, were resistant toribonuclease and fully reneutralizable with anti-IgG serum (51). Thus, it appears as if comple-ment, through the action of the terminal mem-brane attack complex, can lyse antibody-coatedenveloped virions as well as cause major alter-ations in the virion envelope in the absence ofcomplete virolysis.Complement-enhancing ineutralization without

virolysis. The requirement for complement inantibody-mediated neutralization depends uponthe class of immunoglobulin used to sensitizevirus. For example, in studies with herpesvir-uses, early antiserum, containing IgM antiviralimmunoglobulin, requires complement for anysignificant neutralization of herpes simplex virus(HSV), whereas late antiserum, containing IgGantibody, does not require complement (59, 70);complement may, however, enhance IgG-medi-ated neutralization (69).

In the studies of avian infectious bronchitisvirus cited above (5), rabbit antiserum causedvirolysis. However, unheated homologous(fowl) antiserum was shown to induce greaterneutralization than heated serum in the absenceof typical complement lesions on the surface ofvirions. The enhanced neutralization of avianinfectious bronchitis virus in the presence ofcomplement correlated with electron microscop-ic findings: virions treated with unheated serum

demonstrated a protein halo of larger diameterthan did virions treated with heated serum.Thus, these studies indicate that complementcan enhance neutralization by contributing extraprotein on the virion surface.

Studies in other systems have been performedto evaluate other mechanisms by which comple-ment could enhance neutralization of virus in theabsence of virolysis. Serum deficient in eitherearly or late complement components was usedto evaluate the requirements for complement inviral neutralization. It was shown that C4-defi-cient guinea pig serum did not neutralize IgM-HSV complexes, whereas C5- or C6-deficientsera did neutralize the IgM-HSV complexes(12). (Table 1 indicates several types of sera orserum treatments used to evaluate the contribu-tions of the classical and alternative pathways inbiological phenomena.) These studies indicatedthat completion of the terminal lytic phase of thecomplement system was not always required forcomplement-induced neutralization of antibody-virus complexes. Furthermore, when functional-ly purified complement components were used,sequential addition of Cl, C4, C2, and C3 toIgM-HSV complexes caused sequential de-creases in infectivity. Thus, studies with theIgM-HSV infectious complexes showed that vir-olysis was not a prerequisite for neutralizationinduced by complement (12). Similar studieswith Newcastle disease virus also showed thatvirolysis was not required for complement-medi-ated neutralization (33).More recent studies have also demonstrated

that the late components of complement (C5 orC6) are not required for enhancement of IgG-mediated neutralization of virus (31). Studiesusing sera deficient in the early components ofcomplement (Clr, C4, C2, or C3) showed thatCl through C3 were responsible for enhance-ment of IgG-mediated neutralization of bothvaccinia virus and vesicular stomatis virus(VSV) (31). Detailed studies have evaluated theinteraction of VSV with the complement sys-tem. Activation of the classical pathway by VSVresulted in significant viral neutralization in 50%fresh, but not heat-inactivated, human serum(37, 38, 61). Antibody in human serum wasinitially thought not to be required for neutral-ization in that immunoglobulin isolated fromhuman serum failed to neutralize VSV. In addi-tion, serum absorbed with VSV-infected or un-infected cells, to remove antibody, still pos-sessed neutralizing activity, and C3b wasdeposited on the surface of the virus whenserum from a patient with agammaglobulinemiawas tested. However, it was noted that theserum from the patient with agammaglobulin-emia did not neutralize the virus, and this sug-gested that immunoglobulin may have been im-

MICROBIOL. REV.


mbr.asm

.org/D

ownloaded from


COMPLEMENT SYSTEM IN RESPONSE TO VIRAL INFECTION 75

TABLE 1. Common methods for evaluating contributions of complement to biological phenomenaMaterial Effect

Serum treatment or sourceHeat (56°C, 30 min) Destroys hemolytic complement activityEGTAa_Mg2+ Selective chelation of Ca2e inhibits the classical pathwayC4D-GPS Serum from guinea pigs with an inherited deficiency of C4; isolates the ACPC2D-HS Serum from patients with inherited deficiency of C2; same use as C4D-GPSC6D-HS, C7D-HS, C8D-HS Serum from patients with inherited deficiencies of C6, C7, or C8; used to

eliminate the membrane attack complexAnimal sources and treatmentB1O.D2 old line Mice with an inherited deficiency of C5 and identical to B1O.D2 new-line Mice

except for this traitC4-deficient guinea pigs Guinea pigs with an inherited deficiency of C4COVF A protein from cobra venom which temporarily depletes animals of C3

through C9 by virtue of the fact that COVF is analogous to mammalianC3b but is insensitive to C3b inactivator

a Ethylene glycol-bis(P-aminoethyl ether)-N,N-tetraacetic acid.

portant for neutralization. The ACP was notrequired but may have enhanced neutralization(factor B-depleted serum possessed lower neu-tralizing activity), and the terminal complementcomponents (C5 through C8) were not requiredfor neutralization (C5-deficient serum and C6-,C7-, or C8-depleted serum retained significantneutralizing capacity). Clq-deficient, C4-deplet-ed, or C2-deficient serum did not support neu-tralization of VSV, and upon addition of themissing component to these sera, neutralizingactivity was restored. In addition, direct evi-dence for activation of the classical pathway waspresented (38), suggesting that neutralizationoccurred via antibody-independent activation ofthe classical pathway. However, it was alsodemonstrated that addition of purified Cl, C2,C3, C4, and C5 did not neutralize VSV, furthersuggesting that an additional serum factor wasrequired for neutralization (38). Recent studieshave in fact shown that anti-VSV IgM present inall sera tested was the additional factor requiredfor neutralization of VSV (4).A number of mechanisms may be envisioned

as contributing to the complement enhancementof antibody neutralization of enveloped viruses.As noted above, the interaction of antibody-coated virions with the full complement cascademay result in virolysis. As shown for EAV,structural damage may also occur in which criti-cal sites of the virus are destroyed. Complementcomponents may also increase the amount ofprotein on the surface of the virion, thus inter-fering with the binding of viruses to target cellsor altering the surface charge of the virus. Final-ly, in situations in which neutralization occursafter virus penetration of a target cell, comple-ment may enhance the uptake and destruction ofvirus within the cell.The effect of complement on the neutraliza-

tion of nonenveloped viruses has also been

studied. Early studies on poliovirus (WS strain)showed that neutralizing antibody titers werenot influenced by the presence of complement,in that heating of serum at 56°C for 30 min failedto decrease antibody titers (32). More recentstudies using polyoma virus have shown that theearly complement components (Cl through C3)and not C5 through C7 enhanced neutralizationof virus-antibody complexes (41). Furthermore,cleavage of C3 to C3b seemed to be necessaryfor enhanced neutralization, as the presence ofC2 was required. Since the sedimentation rate oflabeled virus was shown to increase upon addi-tion of complement, the enhancement of poly-oma virus neutralization by complement seemedto proceed through agglutination of the virus-antibody-complement complexes.

Antibody-Independent Activation ofComplement by Virus

The demonstration by Douglas and Smith (14)of a heat-labile, calcium-dependent, virus-inhibi-tory serum factor went unnoticed until the late1940s, when Ginsberg and Horsfall reported thatserum obtained from humans, guinea pigs, orrabbits could neutralize mumps virus, Newcas-tle disease virus (NDV), and influenza A and Bviruses (18). The factors responsible for neutral-ization were inactivated upon heating (56°C for30 min) and upon prolonged storage at 4°C. Thevirus-serum component combination was calci-um dependent, did not undergo spontaneousdissociation, and was only partially reversibleby heating mixture for 24 h or by removal ofcalcium. The authors did not think that thisserum component was related to hemolytic com-plement. At the same time that this report ap-peared, Howitt made a similar observation thatNDV was neutralized by normal serum (26). Thecharacteristics of this serum factor were similar

VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


MICROBIOL. REV.

to those reported by Ginsberg and Horsfall (18),and this factor was speculated by Howitt to beassociated with serum complement activity. Inretrospect, the neutralization may have beendue to the presence of natural antibody (64), andthe neutralizing activity may have been depen-dent on complement (e.g., antibody-mediated,complement-requiring neutralization).

Regardless of the nature of the factor respon-sible for the observed neutralization of virus bynormal sera, the results inspired Pillemer and hisco-workers to investigate the influence of theproperdin system (ACP) on virus infectivity(63). At this time, Pillemer and his colleagueshad observed that the properdin system wascapable of causing the lysis of certain erythro-cytes, bacteria, and protozoa. It was shown that50%o fresh human serum inhibited both infectiv-ity and hemagglutination of NDV. The activitywas heat labile, and properdin was required(e.g., zymosan-treated serum was ineffective ininhibiting infectivity or hemagglutination). Themaximum inhibition of infectivity (100-fold re-duction) produced by properdin occurred at theconcentration range for properdin observed infresh serum. Although at this time in the historyof the complement system there were only fourrecognized components, sera depleted of any ofthese components eliminated the inhibitory ef-fect. Magnesium was also shown to be requiredfor inhibition of virus infectivity. In contrast toearlier work in which calcium was shown to berequired for virus inhibition (18), these studiesshowed that only magnesium was required forinhibition of hemagglutination activity. In addi-tion, calcium did not enhance the virus inhibi-tion observed in the presence of magnesium.Finally, in comparison to properdin levels infresh serum incubated with normal allantoicfluid, the levels of properdin were significantlyreduced in serum incubated with allantoic fluidcontaining NDV. Thus, these first detailed stud-ies showed that complement could interact withvirus in the absence of detectable antiviral anti-body and cause virus inactivation. Although it islikely that much of the antiviral activity wascomplement mediated and independent of anti-viral antibody, it is possible that natural anti-body to host membrane components of chickcells, in the presence of complement, may havecontributed to and enhanced the reported ac-tions of the properdin pathway on NDV. Itshould also be noted that a similar report on theinteraction of fresh human serum with T2 bacte-riophage suggested that the properdin systemwas involved in neutralization of this bacterialpathogen (62).

Several recent studies have now confirmedand extended the work reported by Wedgewoodand colleagues (63). After incubation of either

LCMV or NDV in 50% fresh human serum, itwas shown that, in the absence of detectableantibody, significant reduction in virus infectiv-ity occurred (64). Several sera and serum treat-ments were employed to determine the mecha-nism by which NDV and LCMV wereinactivated. In the case of NDV, significantinactivation of virus could be demonstrated inC4-depleted human serum, indicating that NDVwas inactivated by the action of the ACP. How-ever, factor B-depleted human serum could alsocause significant reduction of infectivity, indi-cating that the classical pathway, in the absenceof antiviral antibody, could also be utilized.Furthermore, sera from four patients with agam-maglobulinemia were also effective in reducingNDV infectivity. However, the ability of freshhuman serum to inactivate NDV was dependentupon the cells in which the virus was grown. Ifgrown in chicken embryo fibroblasts, the abovecharacteristics for serum inactivation of NDVheld true. If, however, the virus was grown in ahuman cell line, HeLa cells, fresh serum wasincapable of causing virus inactivation. Thisresult suggested that antibody to host cell com-ponents may have contributed to the observedinactivation of NDV grown in chick cells; how-ever, sera absorbed with chick cells continued toinactivate virus (64). Thus, these studies andthose concerning VSV mentioned above (61)suggest that the passage history of virus maydetermine its ability to be inactivated by comple-ment, and this factor may play a role in deter-mining viral virulence. The case of serum inacti-vation of LCMV was quite different (64). Inthese studies host modifications of the virus alsoplayed a significant role in determining the ex-tent of inactivation; however, it was shown thatthe presence of natural antibody to the host cellcomponents determined the degree of viral inac-tivation. That is, LCMV assimilated host cellantigens during the production of virus, andantibodies directed against these antigens deter-mined the degree of virus inactivation (64).

In addition to studies of NDV, in which indi-rect evidence (viral inactivation) for comple-ment activation was presented, studies withother viruses have directly shown that the classi-cal pathway can be activated independently ofdetectable antiviral or anti-host cell antibody. Itwas initially observed that a number of differentRNA tumor viruses could be lysed by freshhuman serum (65), but not by serum from otherspecies (guinea pig, feline, murine, or simian).The assay used employed the release of theRNA-dependent deoxyribonucleic acid poly-merase from virions after incubation of virus in50% human serum. Normal fresh human serum,but not C2-deficient, C4-deficient, C2-depleted,or C4-depleted human serum or heated serum

76 HIRSCH


mbr.asm

.org/D

ownloaded from



(56°C for 30 min), caused release of (RNA-dependent deoxyribonucleic acid polymerase),or significant reductions in infectivity or both(65). Although no antibody to virus was detectedin the serum tested, a requirement for heat-labileantibody-dependent virolysis could not be com-pletely ruled out; however, serum absorbed withcells expressing retrovirus antigen and serumfrom patients with agammaglobulemia were alsoeffective in causing virolysis. Finally, heatedhuman serum was unable to sensitize virions forlysis by fresh guinea pig complement. Since onlyhuman serum and not serum from nonprimateslysed oncomaviruses, these studies suggestedthat complement may play a role in limitingtransmission of type C retroviruses in humans(65). Studies by others, however, using similarmethods, i.e., release ofRNA-dependent deoxy-ribonucleic acid polymerase from murine leuke-mia virus (MuLV) and primate retroviruses,showed that fresh sera from cats and rabbits, aswell as humans and Old and New World mon-keys, lysed type C retroviruses (56).

Further studies were performed to detail theinteraction of human complement with retrovir-uses (3, 10). These studies showed that Clqbound directly to the envelope of oncornavir-uses in the absence of immunoglobulins (10) andthat complement components were deposited onthe virion surface. In addition to the observedactivation of the classical pathway by retrovir-uses, reduced lysis of MuLv was observed infactor B-depleted serum, suggesting that theACP amplification C3 convertase loop may havebeen required for effective virolysis. Subsequentstudies show that the ACP could be activated byMuLV, but activation of the ACP required ap-proximately 5,000 times more virions than foractivation of the classical pathway (3). Thenature of the Clq receptor for MuLV was deter-mined to be the pl5E virion protein (3). Subse-quent studies showed that the inability of serumfrom guinea pigs to lyse MuLV was not due to adefect in the binding of guinea pig Clq to virusbut rather to the inability of guinea pig Cls to beactivated (2).

In addition to NDV and MuLV, Sindbis virushas recently been shown to activate both theclassical complement pathway and the ACP inthe absence of detectable antibody. In theseinvestigations it was observed that gradient-purified Sindbis virus (as little as 10 ,ug/ml),propagated in either hamster or chicken fibro-blasts, consumed C3 when incubated in normalhuman serum or ethylene glycol-bis(3-amin-oethyl ether)-N,N-tetraacetic acid-magnesium-treated human serum, as well as in normalguinea pig serum and C4-deficient guinea pigserum (Fig. 2) (25). These results indicated thatSindbis virus activated, at least, the ACP. Fur-

ther studies showed that C4 was also consumedin normal human serum, indicating that theclassical pathway was also activated by Sindbisvirus. The observed activation required little orno antibody in that C3 and C4 were consumed inserum from three patients with agammaglobulin-emia, in serum absorbed with infected or unin-fected host cells in which the virus was pro-duced, and in serum containing no detectableneutralizing antibody to Sindbis virus. In con-trast to the studies with MuLV, guinea pigserum supported C3 consumption by both theclassical pathway and the ACP, and unlike thestudies of MuLV and VSV, activation of bothpathways was evoked at similar concentrationsof virus.

In studies of C3 consumption by Sindbis vi-rus, it was shown that, as with the other virusesdiscussed above (NDV and VSV), the host cellof origin influenced the ability of Sindbis virus toactivate complement. No differences in classicalpathway activation by Sindbis virus grown inchick or hamster cells were observed; when theACP was evaluated, virus grown in chick cellsdid consume more C3 than did virus grown inhamster cells (Fig. 2) (25). Recent studies havedemonstrated that the differential activation ofthe ACP by Sindbis virus propagated in differenthost cells can be related to the host-determinedsialic acid content of the virus (24a; R. L.Hirsch, D. E. Griffin, and J. A. Winkelstein,Fed. Proc. 40:966, 1981). Virus grown in amosquito cell line and lacking detectable sialicacid activated the ACP much better than didvirus grown in baby hamster kidney cells andcontaining large amounts of sialic acid. Further-more, the hamster virus, when treated withneuraminidase to remove sialic acid, became amuch more effective activator of the ACP (Table2). These results are in agreement with those ofFearon (16), who has shown that high levels ofsialic acid on a particle favor the interaction ofHwith the surface-bound C3bBb enzyme. Thisinteraction results in dissociation and inactiva-tion by I of the C3bBb complex, effectivelyinhibiting the ACP. Thus, the antibody-indepen-dent activation of the ACP by Sindbis virus isinfluenced by host origin of the virus. Since viralsialic acid is host determined, these studiessuggest that the activity of the complementsystem and its effects on viral pathogenesisdepend, in part, upon the genetic constitution ofthe infected host and the host's ability to modifythe infecting virus.

Antibody-Dependent and -IndependentInteraction of Complement with Virus-Infected

CellsIn addition to the ability of the complement

system to interact with free virus in the presence

VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


MICROBIOL. REV.

Normal GPS60r

501F O-O BHK-SV

*~~~*CEF-SV40-

301-

20 F101

%If

1 3 10 30 100 300 1000

,ug Viral protein/mlFIG. 2. Activation of complement by purified Sindbis virus (SV). SV produced in hamster (BHK) and chick

(CEF) cells consumes similar amounts of C3 in normal guinea pig serum (GPS); however, in C4-deficientGPS(C4D-GPS), when the ACP is isolated, virus made in CEF consumes significantly more C3 (P < 0.05, at 1mg/ml). This difference may be due to differences in the host-determined sialic acid levels ofthe virions producedby these cells. (Reprinted from reference 25 with permission from the Williams & Wilkins Co., Baltimore, Md.)

or absence of antibody, the complement systemcan also interact with viral antigens expressedon the surface of cells. The interaction of com-plement and antibody with infected cells mayresult in cell lysis. The conditions necessary forcomplement-mediated lysis of infected cellshave recently been reviewed in detail (45, 46).However, important features will be summa-rized here.

Lysis of virus-infected cells by complementand antibody requires that sufficient viral anti-gen be expressed on the surface of the cell and

TABLE 2. Effect of virion sialic acid content onactivation of the ACP0

Sialic % C3acid Viral consump-

Source of Sindbis virus content protein tion by(nmollmg 6L/n)virus inOf /m)C4D-GPS

protein) (± SD)

Baby hamster kidney 10.2 800 49 ± 6cells 280 40 ± 9

80 34±1Baby hamster kidney 2.0 800 76 ± 12

cells (virus treated 280 67 ± 10with neuraminidase) 80 59 ± 12

Mosquito cells <2.0 800 75 ± 4280 69 ± 480 57 ± 3

a Purified Sidbis virus, from the indicated hostsource, was incubated in 10% C4-deficient guinea pigserum (C4D-GPS) (to isolate the alternative pathway),and after 30 min at 37°C, the amount of C3 consumedwas ascertained by a standard hemolytic assay. (Datafrom three individual experiments; 24a.)

that the antigen be present in a form and config-uration recognizable by antibody (6). An ex-tremely large number of antibody moleculesneed to be bound to the surface of an infectedcell prior to lysis. This could be a result of one ormore factors. For example, the subclass ofimmunoglobulin which binds to the viral anti-gens may be inefficient in activating comple-ment, or the antigens expressed on the surfaceof the cell may not bind to complement-fixingantibody with high efficiency. The requirementfor sufficient expression of viral antigen hasbeen demonstrated with cells infected with dif-ferent strains of measles virus (30). Acutelyinfected cells bound two to three times moreantibody and were more susceptible to comple-ment-mediated, antibody-dependent lysis thanwere persistently infected cells.The specificity of complement-mediated, anti-

body-dependent lysis of infected cells has alsobeen addressed. It was demonstrated that, afteran individual was vaccinated with mumps virus,antibody developed which, in the presence ofcomplement, could lyse mumps virus-infectedcells (49). Evidence for the specificity of thisphenomenon has also been presented for mea-sles virus vaccination (45).

Further studies with the measles model forcomplement-mediated, antibody-dependent ly-sis of infected cells have shown that lysis occursthrough F(Ab')2-mediated activation of theACP. Participation of the classical pathway wasnot necessary for lysis (29) Studies using varioustreatments of serum to eliminate classical path-way activity have shown that there was noreduction in the efficiency of complement-medi-

0

E

cC0u

0V)

C4D-GPS

78 HIRSCH


mbr.asm

.org/D

ownloaded from



ated, antibody-dependent lysis of infected cells.However, removal of factor B or properdincompletely inhibited lysis of measles-infectedcells, and upon addition of these componentsback to the serum, lytic activity was recovered(46). The requirement of magnesium, but notcalcium, for effective lysis also implied that theACP was responsible for the lysis of infectedcells (29). More recently, lysis of antibody-sensitized, measles-infected cells has been ac-complished by adding to the cells all of thepurified components of the ACP (58). Underthese conditions, it was demonstrated that acti-vation of the ACP was initiated by the virus-infected cells, in the absence of antibody. In-creased deposition of C3b and cell lysis occurredonly upon addition of divalent antibody (57).Although ACP-mediated, antibody-dependent

lysis of infected cells has now been shown tooccur in several virus systems (45), studies byothers suggest that antibody may not always benecessary. Evidence obtained with Rous sarco-ma virus-transformed quail tumor cells has sug-gested that the ACP, in the absence of detect-able antibody, can cause the lysis of Roussarcoma virus-transformed cells. The dependen-cy of the ACP was demonstrated by showingthat the serum reactivity was heat labile, re-quired magnesium, and was not present in zy-mosan- or inulin-treated sera (68). More rigorousstudies will be necessary to prove the antibody-independent nature of this phenomenon.More recently, another complement-mediated

lytic mechanism has been described in vitro.Bovine polymorphonuclear leukocytes (PMN)have been shown to mediate, in the presence ofcomplement and in the absence of detectableantibody, lysis of herpesvirus-infected bovinekidney cells. PMN were much more effective inlysing target cells than were macrophages or

lymphocytes (19). Complement-facilitated, anti-body-dependent cellular cytotoxicity was orig-inally thought to be the mechanism of thisphenomenon (54); however, PMN from anti-body-positive animals were subsequently shownto be no more effective than PMN from nonim-mune animals (19). The mechanism of this com-plement-mediated cytotoxic phenomenon is notknown, but it is possible, since PMN contactwith target cells is required for lysis, that thevirus-infected cells activate complement and fa-cilitate the binding of PMN, by virtue of theirC3b receptors, to the infected cells.

ROLE OF COMPLEMENT IN VIRALINFECTIONS IN VIVO

Recent studies suggest that the several mecha-nisms discussed above by which the comple-ment system can interact with virus or virus-

infected cells in vitro may be of importance invivo during viral infections. Both beneficial andimmunopathological roles for complement havebeen described in a limited number of animalmodels.The first demonstration for a role of comple-

ment in vivo in the host response to virusinfection was shown by Oldstone and Dixon(42). In these studies, the susceptibility of differ-ent mouse strains to LCMV infection was stud-ied. It was shown that B10.D2 old-line mice,which lack serum C5, were more resistant tochallenge with LCMV than the congenic, C5normal, B1O.D2 new-line mice. Since C5-nega-tive mice were more resistant to LCMV chal-lenge than C5-positive mice, these early studiessuggested that terminal complement compo-nents may play an immunopathological role invirus infections. Mice of other strains testedwhich were C5 negative (e.g., SWR/J) werehighly susceptible, but since other genetic fac-tors may also influence susceptibility to LCMV,conclusions like those made in the congenicB1O.D2 mouse system may not be valid. Howev-er, it is interesting to note that the levels of virusin the blood, brain, kidneys, and livers of 3-month-old C5-negative mice (B1O.D2 old lineand SWR/J) persistently infected with LCMV atbirth were higher than those in the organs of C5-positive mice (C3H/HeJ) (42). Thus, these earlystudies suggested that complement may play adual role in the host response to viral infection.Further studies performed with the LCMV mod-el showed that mice depleted of serum hemolyticcomplement levels by treatment with cobra ven-om factor (COVF is cobra C3b, is insensitive tomammalian I, and causes temporary depletion ofserum C3 levels) were more resistant to LCMVchallenge (44). These studies further suggestedthat complement played an immunopathologicalrole in the normal animal. Complement was alsoimplicated in the pathology of LCMV-inducedglomerulonephritis, as it was demonstrated thatcomplement, virus, and antibody were deposit-ed in glomerular-capillary basement membrane(7, 44).Recent studies have shown that the comple-

ment system plays an important role in the hostresponse to viral infection (Table 3). Two stud-ies showed that, in mice depleted of complementby COVF treatment, more severe infectionswith Sindbis virus (22) or influenza virus (21)occurred. In the influenza virus model, B1O.D2old-line and new-line mice were also studied,and in B1O.D2 old-line mice, as well as in theCOVF-treated mice, greater amounts of influen-za virus were present in the lungs after infectionand a higher mortality rate was observed (21). Inthe case of Sindbis virus infection of mice, threeimportant observations were made (22). Firstly,

VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


MICROBIOL. REV.

TABLE 3. Studies on the role of complement inviral infections of mice

Virus Effect of complement depletion Reference

LCMV Dose of virus causing 50% 44mortality increased

Influenza A Increased levels of virus 21in lung and increasedmortality

Sindbis Prolonged viremia; 22, 23increased levels of virusin CNS; later deaths

Rabies Increased levels of virus 36in CNS

despite the fact that there were no significantdifferences between COVF and control mice inlocal replication of virus in muscle after subcuta-neous inoculation, the viremia was prolonged inthe COVF-treated mice (Fig. 3). Secondly, atthe peak of virus replication in the brain, 1,000-fold more virus was present in the brains of theCOVF-treated mice. However, despite the1,000-fold greater levels of virus in the centralnervous system (CNS) of treated mice, overallmortality (30%) was the same in both groups.Thirdly, the COVF-treated animals died signifi-cantly later (approximately 2 days) than controlinfected mice. The later deaths coincided withthe return of normal serum C3 levels in thepreviously decomplemented mice. Thus, thesestudies suggested a dual role for complement inSindbis virus infection of mice. Early in theinfection, during the period of viremia whenthere is not yet any detectable antibody, comple-ment in some manner controls viremia. Thus, inthe COVF-treated animals, viremia persists andmakes infection of target tissues (in this case theCNS) a more likely event. This could accountfor the 1,000-fold greater levels of virus in thebrain of the treated animals. The later deaths inthe COVF-treated mice may indicate that thecomplement system in normal mice plays animmunopathological role in the normal animalby killing virus-infected cells in the brain, elabo-rating chemotactic factors, or promoting themononuclear inflammatory response or by acombination of these. In the case of COVF-treated mice, when C3 levels return to normalthese immunopathological events are manifestedby later deaths.

Further studies were performed to determinewhether the major difference seen in the patho-genesis of Sindbis virus infection in normal andcomplement-depleted mice (i.e., 1,000-fold in-crease in brain virus titers) was due to a lack ofcomplement-mediated effector mechanisms op-erative in the CNS in the COVF-treated mice orto a lack of effective complement interactionwith blood-borne virus early in the infection. In

studies designed to determine whether comple-ment depletion directly influenced the replica-tion of Sindbis virus in the CNS, animals weredepleted of complement by COVF treatmentand were inoculated intracerebrally with twostrains of Sindbis virus which varied in theirvirulence (23). As observed for subcutaneouslyinoculated animals (22), there was no differencein percent mortality after intracerebral inocula-tion of Sindbis virus in complement-depletedanimals and control mice. However, unlike theresults observed after subcutaneous inoculation,there were no differences in levels of virus in theCNS after intracerebral inoculation (23). In oth-er experiments, animals were inoculated subcu-taneously; however, COVF was not adminis-tered until after the viremia had ended. Thesestudies also showed no differences in the growthof the virus in the CNS between control andCOVF-treated animals and therefore suggestthat complement-mediated defense mechanismsoperate during the period of viremia to limitsubsequent infection of the CNS (Fig. 4). Directinvestigations were performed to determinewhether the complement system interacted withvirus and influenced viremia. It was shown fortwo strains of Sindbis virus that, in mice deplet-ed of complement by COVF treatment, a clear-ance defect was evident after intracardiac inocu-lation of virus. Thus, these studies suggestedthat the complement system interacts with virus,during the viremia, and influences the clearanceof virus from, or the inactivation of virus in, theblood stream in the absence of antibody (23).The mechanism(s) by which the complementsystem accomplishes this has not yet been re-solved. However, as mentioned above, we haveshown that Sindbis virus can activate both theACP and the classical complement pathway inthe absence of detectable antibody (25); thus,the complement system may neutralize virus ordeposit C3b on the virion surface and influenceits uptake by the reticuloendothelial cell system.

Further studies in the Sindbis virus modelhave shown the importance of the terminal com-plement components (C5 through C9) in recov-ery from infection. Infection of 2-week-oldB1O.D2 old-line mice with Sindbis virus resultedin approximately 80% mortality, whereas only20% of B10.D2 new-line mice succumbed tofatal Sindbis virus infection. There were nodifferences in the viremia in these mice which,unlike COVF-treated mice, possess normal C3levels; however, there were significantly greateramounts of virus present in the brains of B1O.D2old-line mice 5 days after infection (24). Thesestudies suggested that there may have been apreviously undetected role for complement inlimiting virus growth in the CNS itself whenseeding of the brain (viremia) is held constant

80 HIRSCH


mbr.asm

.org/D

ownloaded from



5'r

FEET

(o (4)

~ (s~~~ ~ ~ ~

4

w

Uf)

(I)

0-

0

3

2

2 NI (2)

5 BLOOD

4 -

IAIN

I \I

/ /%I/

11 IV, (7\

aIi ~~~~~~~~~~~(4)0 0

(,

I a I Ii

8Ii 2 4 6SMDAYS AFTEft INFECTION

(6)

(4) (7)&

BLOOD BRAIN

(7)\

L L__~~~~~1-i41 II

(7) (7) (7)I I It 2 3 4 5 6 8 lo

VirusCoVF CoVF

DAY AFTER INFECTION

FIG. 4. Levels of Sindbis virus in brains of normal) and COVF-treated mice (---) mice after sub-

cutaneous inoculation of virus. When COVF is givenafter the viremia has ended, no differences in level ofvirus in the brain are observed. Numbers in parenthe-ses indicate numbers of animals in each group as-sessed at the indicated times. (Reprinted from refer-ence 23 with permission from the University ofChicago Press, Chicago, Ill.)

and contributions of the terminal complemencomponents are definitively eliminated. Thus,these studies implicate either the generation ofchemotactic factors (C5a) or the terminal lyticcomponents of complement (C6 through C9) incontrolling some aspects of virus infection.One other in vivo study has implicated the

complement system as being important in viralinfections. Mice treated with COVF and inocu-lated intracerebrally with rabies virus demon-strated delayed clearance of virus from the CNS(36). These results are comparable to thosereported for Sindbis virus infection in C5-defi-cient mice.

ROLE OF THE COMPLEMENT SYSTEM INVIRAL INFECTIONS OF HUMANS

Although it has been clearly demonstrated inanimal models that the complement system canplay both beneficial and immunopathologicalroles in the host response to viral infection,limited data are available concerning the role of

FIG. 3. Levels of Sindbis virus in feet, bloods, andbrains of 12-day-old normal (-) and COVF-treat-ed(---) mice inoculated subcutaneously with virus. Aprolonged viremia and greater levels of virus werenoted in the brains of treated mice. Numbers in

parentheses indicate number of animals in each groupassessed at the indicated times. (Reprinted from refer-ence 22 with permission from the Williams & WiLkinsCo., Baltimore, Md.)

VOL. 46, 1982

I

S

7

6

5

4-

3-

6

2

2

r

I -

-

1-

I

I

7


mbr.asm

.org/D

ownloaded from


82 HIRSCH

complement in viral infections of humans. Thereare two major reasons for this lack of informa-tion. Firstly, although patients with inheriteddeficiencies of complement components are de-scribed in the literature, these patients usuallypresent with recurrent or chronic bacterial infec-tions, and problems with viral infections are notusually described. Secondly, there are only lim-ited studies which can be done in humans thatrelate to the role of complement in the pathogen-esis of infectious diseases.Most investigations on the role of the comple-

ment system in viral infections of humans haveconcentrated on changes in serum complementprofiles during infection. There is a major prob-lem with interpretation of these studies. Whenalterations of serum complement componentsare found, it is difficult to discriminate betweenconsumption of complement components, acti-vation of complement by virus-antibody com-plexes, changes in synthesis or catabolism ofcomplement components, or extravasation ofcomplement components as a result of the virusinfection. Despite these limitations, some usefulinformation has been collected.

In a group of 50 children with uncomplicatedacute measles virus infections, it was found that48% of patients had reduced serum Clq levels,12% had reduced C4 levels, 26% had reduced C3levels, and 16% had reduced C5 levels as mea-sured immunochemically by radial immunodiffu-sion. Ten of the patients showed complementprofiles indicative of activation of the classicalpathway (decreased Clq, C4, and C3 levels),whereas seven showed decreased C3 levels withnormal C4 and Clq levels, which the authorssuggested as evidence of activation of the ACP(9). However, factor B and C7 levels in thesepatients were normal, suggesting that the ACPand terminal complement components were notutilized. The decreased C3 levels, therefore, inpatients not showing decreased Clq and C4levels may indicate an effect of measles virus,known to infect mononuclear cells, on C3 syn-thesis in these patients. Eleven of the patientsdemonstrated reduced levels of serum Clq only.Despite this pattern, immune complexes, whichwould consume Cl, were not detected by a Clqprecipitation test (9).

Charlesworth and colleagues have also stud-ied serum complement profiles in 34 patientswith infectious mononucleosis (8). When thevalues for all of the patients were pooled, meanvalues for all complement components mea-sured (Clq, Cis, Cl inhibitor, C4, C3, factor B,properdin, C5, and C7) were not significantlydifferent from values in normal patients. Howev-er, in three patients with complicated cases (onesevere hemolytic anemia, one severe arthralgia-myalgia, and one proliferative glomerulonephri-

tis), reduced C3 levels were observed. Two haddecreased Clq and C4 as well as reduction in theterminal complement components C5 and C7.The patient with glomerulonephritis had mark-edly reduced C3 levels and was positive forC3NeF but had normal Clq and C4 levels. Inthese complicated cases the decreased levels ofC3 may be related to the presence of C3NeF(shown to be increased in these studies) and itsability to enhance activation of the ACP (C3NeFinhibits the ability of H and I to interact withfactor B and C3 on the surface of activatingparticles). However, serum properdin levelswere not reported for these three patients withcomplications, and this detail would be helpfulin determining the mechanism of the decreasedC3 levels. In the uncomplicated cases, comple-ment levels may be influenced by direct infec-tion of macrophages by Epstein-Barr virus, thecausative agent of infectious mononucleosis.Thus, the reported decreases in C4 and C3 couldbe due to a direct inhibitory effect of the infec-tion on synthesis of complement components.Two studies have examined complement lev-

els during acute viral hepatitis infections ofchildren. In one study decreased hemolytic C3and C4 levels were found in the acute phase ofdisease, with levels returning to normal duringconvalescence (1). Since peak immune complexlevels were also detected at the time of de-creased C3 and C4 levels, these studies suggestactivation of the classical pathway, with thereservation that liver damage may have had amajor influence on serum levels of C3 and C4. Ina second study of acute viral hepatitis, de-creased levels of hemolytic Cl, C4, C2, C3, andC5 were observed as well as decreased levels ofimmunochemically reactive Cls, C4, factor B,and C3. Clq levels were normal (15). In onepatient, who received infusion of fresh plasmacontaining normal levels of complement compo-nents, C4 levels were followed relative to anoth-er patient, with a-1-antitrypsin deficiency, whoalso received fresh plasma. Within 8 h afterplasma transfusion, there was no detectable C4in the serum of the hepatitis patient, whereas C4levels in the control patient remained at 30 to40%6 up to 70 h after transfusion. These studiesalso showed decreased prothrombin levels in thehepatitis patient, suggesting that, in part, thedecreased complement levels were related toliver damage. However, there was also rapidcatabolism of C4 and C3 in patients receivingfresh plasma, suggesting that complement con-sumption as well as decreased synthesis contrib-uted to the depressed complement levels. It wasalso noted that low levels ofcomplement persist-ed in patients who died, probably reflectingsevere liver damage (15).

In a study of the complement system in pa-

MICROBIOL. REV.


mbr.asm

.org/D

ownloaded from



tients infected with Junin virus, an arenavirus,and presenting with Argentine hemorrhagic fe-ver, total hemolytic complement levels werereduced by 68% in 10 patients with moderate orsevere disease (13). In these patients C2, C3,and C5 levels, as determined by imunochemicaltechniques, were 12 to 60% below normal,whereas C4 levels were 160%o of normal levels.Unlike the mouse model of an arenavirus(LCMV) infection discussed previously (43),these results do not support the involvement ofimmune complexes in the pathogenesis of Ar-gentine hemorrhagic fever (no decreased Cl orC4), but complement activation may play a rolein some other respect.

All of the studies outlined above on the effectof viral infections on human complement pro-files were performed on patients who were in-fected with viruses that either infect mononucle-ar cells (e.g., measles, Epstein-Barr virus) orhave cytopathic effects on hepatic cells (e.g.,hepatitis virus). Thus, it is not possible to delin-eate with any certainty from these studies therole that complement plays in viral infections ofhumans. There are, however, a few suggestivenotes which have been made in case reports ofpatients with inherited deficiencies of particularcomplement components. For example, in thestudy of kindred in which a member had acomplete C3 deficiency, the brother of the C3-deficient patient, the son of two heterozygotes,died at 21 months of age with documented St.Louis encephalitis (47). Although the death oc-curred prior to the recognition of the C3-defi-cient patient, it is tempting to speculate that thebrother who died from viral encephalitis mayalso have been C3 deficient. Similarly, the find-ings at autopsy in a patient with an inheriteddeficiency of C4 and systemic lupus erythemato-sus showed intranuclear inclusion bodies in thelung, and cytomegalovirus was isolated fromboth liver and lungs (28).Another interesting association exists be-

tween systemic lupus erythematosus, comple-ment deficiencies, and C-type viruses (retrovir-uses). An association between systemic lupuserythematosus and deficiency of the early com-plement components has been observed. Fur-thermore, expression of C-type viruses has beenreported in kidney tissue from systemic lupuserythematosus patients (48). Since C-type virus-es are lysed by the action of the classical path-way (65), it is conceivable that a deficiency inany of the early complement components couldincrease the yet unknown risks of C-type virusexpression in humans.

SUMMARYMany lines of evidence indicate that the com-

plement system plays an important role in the

pathogenesis of viral infections. Studies in vitrohave shown that complement, in either the pres-ence or the absence of specific antibody, canneutralize virus by one or more mechanisms.Furthermore, both the classical pathway and theACP have been shown to participate in virusneutralization. In addition, the ACP can causelysis of antibody-coated virus-infected cells,suggesting that, in vivo, complement may actnot only on free virus but also on infected cells.Studies in animal models have demonstratedthat complement does in fact play a significantrole in viral infections and that the action of thecomplement system may be beneficial or harm-ful. Although the mechanisms by which thecomplement system exerts influence in vivo areunknown, studies have demonstrated that com-plement either independently or in the presenceof antibody can exert beneficial effects at sites oflocal viral replication or during the period ofviremia. Furthermore, the presence of comple-ment-containing immune complexes in tissue,and the delayed death and increased doses ofvirus required for lethal infection in comple-ment-depleted animals, suggest that the comple-ment system can also exert immunopathologicaleffects during viral infections.

ACKNOWLEDGMENTSI thank Diane Griffin, Richard T. Johnson, and Jerry

Winkelstein for their constructive criticism and review of thismanuscript. I also thank Linda Keily for typing this manu-script innumerable times.R.L.H. is a Research Associate of the Howard Hughes

Medical Institute. This work was supported in part by PublicHealth Service grant NS15749 from the National Institute ofNeurological and Communicative Disorders and Stroke.

LITERATURE CITED

1. Baer, G. M., and J. A. Walker. 1977. Studies of anoutbreak of acute hepatitis A: complement level fluctua-tion. J. Med. Virol. 1:1-7.

2. Bartholomew, R. M., and A. F. Esser. 1978. Differences inactivation of human and guinea pig complement by retro-viruses. J. Immunol. 121:1748-1751.

3. Bartholmew, R. M., A. F. Esser, and H. J. Muller-Eberhard. 1978. Lysis of oncornaviruses by human se-rum, isolation of the viral complement (Cl) receptor andidentification as p15E. J. Exp. Med. 147:844-853.

4. Beebe, D. P., and N. R. Cooper. 1981. Neutralization ofvesicular stomatitis virus (VSV) by human complementrequires a natural IgM antibody present in human serum.J. Immunol. 126:1562-1568.

5. Berry, D. M., and J. D. Almelda. 1968. The morphologicaland biological effects of various antisera on avian infec-tious bronchitis virus. J. Gen. Virol. 3:97-102.

6. Brier, A. M., C. Wohlenberg, J. Rosenthal, M. Mage, andA. L. Notkhns. 1971. Inhibition or enhancement of immu-nological injury of virus-infected cells. Proc. Natl. Acad.Sci. U.S.A. 68:3073-3077.

7. Buchmeier, M. J., and M. B. A. Oldstone. 1978. Virus-induced immune complex disease: identification of specif-ic viral antigens and antibodies deposited in complexesduring chronic lymphocytic choriomeningitis virus infec-tion. J. Immunol. 120:1297-1304.

VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


84 HIRSCH

8. Charesworth, J. A., B. A. Pussell, L. P. Roy, S. Lawrence,and M. R. Robertson. 1977. The complement system ininfectious mononucleosis. Aust. 14Z. J. Med. 7:23-26.

9. Charlesworth, J. A., B. A. Pussell, L. P. Roy, M. R.Robertson, and J. Beveridge. 1976. Measles infection:involvement of the complement system. Clin. Exp. Im-munol. 24:401-406.

10. Cooper, N. R., F. C. Jensen, R. M. Welsh, and M. B. A.Oldstone. 1976. Lysis of RNA tumor viruses by humanserum: direct antibody independent triggering of the clas-sical complement pathway. J. Exp. Med. 144:970-984.

11. Daniels, C. A. 1975. Mechanisms of viral neutralization, p.79-97. In A. L. Notkins (ed.), Viral immunology andimmunopathology. Academic Press, Inc., New York.

12. Dankls, C. A., T. Borsos, H. J. Rapp, R. Synderman, andA. L. Notklns. 1970. Neutralization of sensitized virus bypurified components of complement. Proc. Natl. Acad.Sci. U.S.A. 65:528-535.

13. deBracco, M. M. E., M. T. Rimoldi, P. M. Cossio, A.Rabinovich, J. I. Maiztegui, G. Carballal, and R. M.Arana. 1978. Argentine hemorrhagic fever: alterations ofthe complement system and anti-Junin virus humoralresponse. N. Engl. J. Med. 299:216-221.

14. Douglas, S. R., and W. Smith. 1930. A study of vaccinalimmunity in rabbits by means of in vitro methods. J. Exp.Pathol. 11:96-11l1.

15. Dupuy, C., and J. M. Dupuy. 1977. Complement studies insevere viral hepatitis of childhood. Pathol. Biol. 25:553-558.

16. Fearon, D. T. 1979. Activation of the alternative comple-ment pathway. Crit. Rev. Immunol. 1:1-32.

17. Fujita, T., I. Gigli, and V. Nussenzweig. 1978. Human C4binding protein. II. Role in proteolysis of C4b by C3binactivator. J. Exp. Med. 148:1044-1051.

18. Ginsberg, H. S., and F. L. Horsfall. 1949. A labile compo-nent of normal serum combines with various viruses.Neutralization of infectivity and inhibition of hemaggluti-nation by the component. J. Exp. Med. 90:475-495.

19. Grewal, A. S., B. T. Rouse, and L. A. Babluk. 1980.Mechanism of recovery from viral infections: destructionof infected cells by neutrophils and complement. J. Im-munol. 124:312-319.

20. Hadding, U. 1977. The reactivity of the complementsystem. Monogr. Allergy 12:36-45.

21. Hkks, J. T., F. A. Ennis, E. Kim, and M. Verbonitz. 1978.The importance of an intact complement pathway inrecovery from a primary viral infection. Influenza indecomplemented and in C5-deficient mice. J. Immunol.121:1437-1445.

22. Hh-sch, R. L., D. E. Griffn, and J. A. Winke}stein. 1978.The effect of complement depletion on the course ofSindbis virus infection in mice. J. Immunol. 121:1276-1278.

23. Hirsch, R. L., D. E. Griffin, and J. A. WinkeLstein. 1980.The role of complement in viral infections. II. The clear-ance of Sindbis virus from the bloodstream and centralnervous system of mice depleted of complement. J. In-fect. Dis. 141:212-217.

24. HIrsch, R. L., D. E. Griffin, and J. A. Winkelstein. 1980.The role of complement in viral infections: participation ofterminal complement components (C5 to C9) in recoveryof mice from Sindbis virus infection. Infect. Immun.30:899-901.

24a.Hirsch, R. L., D. E. Griffin, and J. A. Winkeistein. 1981.Host modification of Sindbis virus sialic acid contentinfluences alternative complement pathway activation andvirus clearance. J. Immunol. 127:1740-1743.

25. Hirsch, R. L., J. A. Winkeistein, and D. E. Griffin. 1980.The role of complement in viral infections. III. Activationof the classical and alternative complement pathways bySindbis virus. J. Immunol. 124:2507-2510.

26. Howltt, B. F. 1950. A nonspecific heat-labile factor in theserum neutralization test for Newcastle disease virus. J.Immunol. 64:73-84.

27. Hugli, T. E., and H. J. Muller-Eberhard. 1978. Anaphyla-

toxins: C3. and Cs.. Adv. Immunol. 26:1-51.28. Jackson, C. G., H. D. Ochs, and R. J. Wedgewopd. 1979.

Immune response of a patient with deficiency of the fourthcomponent of complement and systemic lupus erythema-tosus. N. Engl. J. Med. 300:1124-1129.

29. Joseph, B. S., N. R. Cooper, and M. B. A. Oldstone. 1975.Immunologic injury of cultured cells infected with measlesvirus. I. Role of IgG antibody and the alternative comple-ment pathway. J. Exp. Med. 141:761-774.

30. Joseph, B. S., L. H. Perrin, and M. B. A. Oldstone. 1976.Measurement of virus antigens on the surfaces of HeLacells persistently infected with wild type and vaccinestrains of measles virus by radioimmune assay. J. Gen.Virol. 30:329-337.

31. Leddy, J. P., R. L. Slmons, and R. G. Douglas. 1977.Effect of selective complement deficiency on the rate ofneutralization of enveloped viruses by human sera. J.Immunol. 118:28-34.

32. Ledlnko, N., and J. L. MeLick. 1953. Poliomyelitis virus-es in tissue culture. V. Reaction of virus and antibody;variables of the quantitative neutralization test. Am. J.Hyg. 58:223-247.

33. Llnscott, W. D., and W. E. Levlnson. 1969. Complementcomponents required for virus neutralization by earlyimmunoglobulin antibody. Proc. Natl. Acad. Sci. U.S.A.64:s20-S26.

34. Mayer, M. M. 1977. The cytolytic attack mechanism ofcomplement. Monogr. Allergy 12:1-12.

35. Mayer, M. M. 1981. Membrane damage by complement.Johns Hopkins Med. J. 148:243-258.

36. Mlle, A., H. C. Morse, J. A. Winkeistein, and N.Nathanson. 1978. The role of antibody in recovery fromexperimental rabies. I. Effect of depletion of B and Tcells. J. Immunol. 121:321-326.

37. Mi, B. J., D. P. Beebe, and N. R. Cooper. 1979.Antibody independent neutralization of vesicular stomati-tis virus by human complement. II. Formation of VSV-lipoprotein complexes in human serum and complement-dependent viral lysis. J. Immunol. 123:2518-2524.

38. MIlls, B. J., and N. R. Cooper. 1978. Antibody indepen-dent neutralization of vesicular stomatitis virus by humancomplement. I. Complement requirements. J. Immunol.121:1549-1557.

39. Morga, I. M. 1945. Quantitative study of the neutraliza-tion of Western equine encephalomyelitis virus by itsantiserum and the effect of complement. J. Immunol.50:359-371.

40. MueLer, J. H. 1931. The effect of alexin in virus-antivirusmixtures. J. Immunol. 20:17-23.

41. OhRtone, M. B. A.? N. R. Cooper, and D. L. Larson. 1974.Formation and biologic role of polyoma virus-antibodycomplexes. A critical role for complement. J. Exp. Med.140:549-565.

42. Oldsone, M. B. A., and F. J. Dixon. 1968. Susceptibilityof different mouse strains to lymphocytic choriomeningitisvirus. J. Immunol. 100:355-357.

43. 01_tone, M. B. A., and F. J. Dixon. 1969. Pathogenesis ofchronic disease associated with persistent lymphocyticchoriomeningitis viral infection. I. Relationship of anti-body production in neonatally infected mice. J. Exp. Med.1391h483-499.

44. Oldtone, M. B. A., and F. J. Dixon. 1971. Acute viralinfection: tissue injury mediated by anti-viral antibodythrough a complement effector system. J. Immunol.107:1274-1280.

45. Oldtone, M. B. A., and P. W. Lampet. 1979. Antibodymediated complement dependent lysis of virus infectedcells. Springer Semin. Immunopathol. 2:261-283.

46. Oldtone, M. B. A., J. G. P. Sbsons, and R. S. Fujnmai.1980. Action of antibody and complement in regulatingvirus infection, p. 599-621. In M. Fougereau and J.Dausset (ed.), Immunology 80. Academic Press, Inc.,New York.

47. Osofsky, S. G., B. H. Thompson, T. F. Lint, and H.Gewurz. 1977. Hereditary deficiency of the third compo-

MICROBIOL. REV.


mbr.asm

.org/D

ownloaded from



nent of complement in a child with fever, skin rash, andarthralgias: response to transfusion of whole blood. J.Pediatr. 90:180-186.

48. Panem, S., N. G. Ordonez, W. H. Kirstein, A. Katz, and B.H. Spargo. 1976. C-type virus expression in systemiclupus erythematosous. N. Engl. J. Med. 295:470-475.

49. Perrin, L. H., B. S. Joseph, N. R. Cooper, and M. B. A.Oldstone. 1976. Mechanism of injury of virus infected cellsby antiviral antibody and complement: participation ofIgG, Fab12 and the alternative complement pathway. J.Exp. Med. 143:1027-1041.

50. Radwan, A. I., and D. Burger. 1973. The complementrequiring neutralization of equine arteritis virus by lateantisera. Virology 51:71-77.

51. Radwan, A. I., and D. Burger. 1973. The role of sensitiz-ing antibody in the neutralization of equine arteritis virusby complement or anti-IgG serum. Virology 53:366-371.

52. Radwan, A. I., D. Burger, and W. C. Davis. 1973. Thefate of sensitized equine arteritis virus following neutral-ization by complement or anti-IgG serum. Virology53:372-378.

53. Rother, K. 1977. Biological functions of the complementsystem. Monogr. Allergy 12:90-100.

54. Rouse, B. T., A. S. Grewal, and L. A. Babluk. 1977.Complement enhances antiviral antibody dependent cellcytotoxicity. Nature (London) 266:456-458.

55. Schreiber, R. D., and H. J. Muller-Eberhard. 1978. As-sembly of the cytolytic alternative pathway of comple-ment from 11 isolated plasma proteins. J. Exp. Med.148:1722-1727.

56. Sherwin, S. A., R. E. Benveniste, and G. J. Todaro. 1978.Complement mediated lysis of type C virus: effect ofprimate and human sera on various retroviruses. Int. J.Cancer 21:6-11.

57. Sissons, J. G. P., M. B. A. Oldstone, and R. D. Schreiber.1980. Antibody independent activation of the alternativecomplement pathway by measles virus infected cells.Proc. Natl. Acad. Sci. U.S.A. 77:559-562.

58. Sssons, J. G. P., R. D. Schreiber, L. H. Perrin, N. R.Cooper, H. J. Muller-Eberhard, and M. B. A. Oldstone.1979. Lysis of measles virus infected cells by the purifiedcytolytic alternative complement pathway and antibody.J. Exp. Med. 150:445-454.

59. Snyder, D. B., A. C. Myrup, and S. K. Dutta. 1981.

Complement requirement for virus neutralization by anti-body and reduced serum complement levels associatedwith experimental equine herpesvirus infection. InfectImmun. 31:636-640.

60. Stollar, V. 1975. Immune lysis of Sindbis virus. Virology66:620-624.

61. Thiry, L., J. Cogniaux-Le Clerc, J. Content, and L. Tack.1978. Factors which influence inactivation of vesicularstomatitis virus by fresh human serum. Virology 87:384-393.

62. van Vunakis, H., J. L. Barlow, and L. Levine. 1956.Neutralization of bacteriophage by the properdin system.Proc. Nati. Acad. Sci. U.S.A. 42:391-394.

63. Wedgewood, R. J., H. S. Ginsberg, and L. Pillemer. 1956.The properdin system and immunity. VI. The inactivationof Newcastle disease virus by the properdin system. J.Exp. Med. 104:707-725.

64. Welsh, R. M. 1977. Host cell modification of lymphocyticchoriomenigitis virus and Newcastle disease virus alteringviral inactivation by human complement. J. Immunol.118:348-354.

65. Welsh, R. M., N. R. Cooper, F. C. Jensen, and M. B. A.Oldstone. 1975. Human serum lyses RNA tumor viruses.Nature (London) 257:612-614.

66. Welsh, R. M., P. W. Lampert, P. A. Burner, and M. B. A.Oldstone. 1976. Antibody-complement interactions withpurified lymphocytic chorimeningitis virus. Virology73:59-71.

67. Whitman, L. 1947. The neutralization of Western equineencephalomyelitis virus by human convalescent serum.The influence of host labile substances in serum on theneutralization index. J. Immunol. 56:97-108.

68. Yoshikawa, Y., K. Yamanouchi, M. Hishlyama, and K.Kobune. 1978. Lysis of RSV-transformed Japanese quailcells by a factor from normal quail serum. Int. J. Cancer21:658-666.

69. Yoshino, K., M. Hashimoto, and K. Shinkai. 1977. Studieson the neutralization of herpes simplex virus. VIII. Signif-icance of viral sensitization for inactivation by comple-ment. Microbiol. Immunol. 21:231-241.

70. Yoshino, K., and S. Tanipchi. 1965. Studies on theneutralization of herpes simplex virus. I. Appearance ofneutralizing antibodies having different grades of comple-ment requirement. Virology 26:61-72.

VOL. 46, 1982


mbr.asm

.org/D

ownloaded from


the complement system: importance response to viral …the complement system is composed of a series...

Documents