How Does HIV Cause AIDS?

Robin A. Weiss

Many questions have been posed about acquired immunodeficiency syndrome (AIDS)pathogenesis. Is human immunodeficiency virus (HIV) both necessary and sufficient tocause AIDS? Is AIDS essentially an autoimmune disease, triggering apoptosis, or is virusinfection the cause of T helper lymphocyte depletion? What is the significance of HIVtropism and the role of macrophages and dendritic cells in AIDS? Is there viral latency andwhy is there usually a long period between infection and AIDS? Is HIV variation a crucialaspect of its pathogenesis and, if so, do virulent strains emerge? Although this articleprovides few definitive answers, it aims to focus commentary on salient points. Overall, itis increasingly evident that both the tropism and burden of HIV infection correlate closelywith the manifestations of disease.

Acquired immunodeficiency syndrome(AIDS) is the end-stage disease of humanimmunodeficiency virus (HIV) infection.The key to understanding its pathogenesislies in elucidating the course of infectionand the virus-host relation in the yearspreceding terminal illness. Figure 1 depictsthe progression from initial infection toAIDS. Ever since its first description AIDShas been related to depletion of CD4C, Thelper lymphocytes in the blood (1). Wenow know that other cells, especially tissuemacrophages, become infected by HIV andthat there is a considerable viral load in thelymph nodes. HIV infection may causedisease by one or more of a variety of directand indirect means. However, early in in-fection, the cellular and humoral immuneresponses to HIV appear to be effective inlimiting viral replication in peripheralblood cells.

Why, then, does AIDS finally develop,and why does it take a long and variabletime to do so? There is no dearth of hypoth-eses to explain HIV pathogenesis. By andlarge they are not mutually exclusive butrequire critical examination before beingweighted in importance, supported, or dis-carded. Moreover, there are lessons to belearned from lentivirus infections of ani-mals that raise further questions regardinghuman disease. Why, for instance, do somehorses permanently recover from equineinfectious anemia when the virus evolvesimmune escape variants as readily as HIV?Can the wasting syndrome and brain dis-ease of sheep infected with visna-maedivirus be equated to human AIDS withoutCD4 depletion?

This article is not intended to provide adetailed description of HIV pathogenesis.A comprehensive review has recently ap-peared (2). Here I wish to highlight theThe author is at the Chester Beatty Laboratories,Institute of Cancer Research, Fulham Road, LondonSW3 6JB, United Kingdom.

mune deficiencypersons who tesHIV-2 even by Ereaction (PCR) i

ral genomes. Thomust have a diffedoes not negateglobal pandemictence of hepatithepatitis B virumAIDS is not a wefined; an outlyingbefore AIDS w

emerging consensus on progressive HIV doctors greatly (infection and to pose some of the questions In any illnessthat remain to be answered about HIV manifest in severinfection and AIDS in the hope that this expected to inflmay help to sharpen our thinking and to disease develops.focus further research on important topics. retroviral pathogi

virus-type I (HTIs HIV Sufficient to Cause AIDS? leukemia or trot

fewer than 6%Our lack of understanding of how HIV lifelong infectioncauses AIDS has led some commentators to Epstein-Barr vinquery whether HIV infection is sufficient to doubt the role ofcause AIDS or whether HIV may be essen- therefore, that s(tially harmless in the absence of other placed on the irncofactors (3). Indeed, one retrovirologist factors for AIDEhas concluded that HIV infection is too infections (3), Xsilent to be the etiologic agent of AIDS at HIV-infected p(all (4). However, the overwhelming view is Proponents of tthat HIV infection is active enough to be penetrans is a potdirectly pathogenic, as I shall argue here, ed the most fussand that the epidemiological evidence for a with AIDS remacausal relation between HIV and AIDS is tomegalovirus (9compelling (5). infections (3) art

Some cases of severe unexplained im- Thus there is no

Seroconvesn D

Infon Minor or no symptoms ARC/AIDS

E1000E

EWC PBL

S 8 so- \~~~~~~~~~~~~~~ V have come to light int negative for HIV-1 andsensitive polymerase chainmethods for amplifying vi-ese rare and sporadic caseserent etiology (6), but thatthe role of HIV in theany more than the exis-

tis C virus casts doubt onIs as a pathogen. Besides,ell-known old disease rede-ig HIV+ case observed longras recognized puzzled his7)-that takes years to becomere form, cofactors would beluence the rate at which. The other known humanyen, human T cell leukemiaFLV-I), causes adult T cellpical spastic paraparesis inof infected people duringn. Diseases associated withus are even rarer, but few'the virus. It seems curious,o much emphasis has beennportance of infectious co-S such as drugs and otherwhen more than 50% ofeople progress to AIDS.the idea that Mycoplasmatential cofactor have elicit-in the lay press but its linkains controversial (8). Cy-)) and other persistent virusre just as likely candidates.strong indication that the

h Fig. 1. Schematic courseof HIV infection.

AAntbodbs to HIV Env

~ Antibodiesto HIV p24

IemlVa

It

4-8 wees Up to 12 ye 2-3 years

SCIENCE * VOL. 260 * 28 MAY 1993 1273

lk.Lt:b

development of AIDS necessarily involvesa specific, infecting microorganism besidesHIV, although in theoretical model systemsthe cumulative exposure to infections ac-celerates the rate of progression to AIDS(10) and may exacerbate disease late in thecourse of HIV infection.

The various opportunistic infections andopportunistic neoplasms that occur inAIDS are bound to depend on prior orcurrent exposure to the proximate agent,and certain markers of AIDS may indeeddepend on independent factors; in otherwords, HIV acts as the cofactor and not theunderlying cause of that symptom. For ex-ample, the risk of Kaposi's sarcoma (KS) inWestern countries is much greater amongHIV+ male homosexuals than amongHIV+ intravenous drug users or hemophil-iacs (11). That and other observations ar-gue for a specific, sexually transmitted eti-ologic agent for KS in which immune sup-pression (both in AIDS and in transplantpatients) is the dominant cofactor for sub-sequent disease.

Emphasis on cofactors should not ob-scure the fact that population groups withdifferent lifestyles have similar rates of pro-gression to AIDS. Figure 2 shows that thedevelopment of AIDS in homosexual andhemophilic men in the West is indistin-guishable, with a mean time from HIVseroconversion to disease development ofapproximately 9 years. It is thought thatAIDS occurs more rapidly in Africa al-though insufficient cohort data are availableto be sure (12). Perinatally infected infantsalso appear to have a faster rate of progres-sion. Figure 2 also indicates there is a delayin the onset of AIDS in hemophilic pa-tients who acquired HIV at a younger agerelative to older individuals. The lag periodis followed by a similar rate of progression toAIDS which may result from younger per-sons having greater CD4 lymphocyte re-serves and precursors for renewal.

It is often assumed that there is some-thing special about long-term survivors ofHIV infection, and conversely, about thosewho succumb to AIDS relatively soon afterinfection. Yet the rates of progression de-picted in Fig. 2 are also consistent with astochastic, random occurrence ofAIDS afterHIV infection. "Long-term" survival couldbe pure luck. Nevertheless, I would expectgenetic, behavioral, and environmental fac-tors to influence the rate of progression toAIDS among individuals even if HIV is thesole etiologic pathogen. Predisposing anddelaying factors are likely to relate both tothe infecting HIV strain (see below) and tohost traits. For instance, some combinationsofmajor histocompatibility complex (MHC)antigens predispose to earlier AIDS (13);MHC-linked genetic determinants may bean important cofactor in HIV pathogenesis.1274

0

12Thw (year)

Fig. 2. Proportion of individuals surviving with-out AIDS plotted with data combined fromseveral U.S. and United Kingdom cohorts (5,62). 0, HIV+ homosexual men; V, HIV+ hemo-philic men who were >20-25 years of age attime of infection; A, HIV+ hemophilic men whowere <20 years of age at time of infection; and5, HIV- hemophiliacs and homosexuals.

Is AIDS an Autoimmune Disease?

A virus can damage its target tissue in atleast four ways: (i) Infection can be cyto-pathic; in other words, the virus kills thecells it infects. (ii) The immune response toinfected cells leads to their destruction; forexample, the pathogenesis of lymphocyticchoriomeningitis virus in mice and of hepa-titis B virus in humans is largely a conse-quence of specific cytotoxic T cells destroy-ing virus-infected cells. (iii) The function ofinfected cells may be affected without celldeath; in the immune system this couldresult in aberrant signaling via cytokines andcell-to-cell contacts. (iv) The virus maytrigger autoimmune destruction through mo-lecular mimicry of cellular antigens to whichthe host is normally tolerant. In HIV infec-tion the first three mechanisms act in con-cert to cause AIDS. Although the effect ofHIV on the immune system resembles au-toimmune disease, it is driven by persistent,active viral expression.

Given that HIV infects and perturbs cellsof the immune system, it is hardly surprisingthat some autoimmune signs are frequentlyseen in the course ofinfection (2, 14). AIDShas been likened to a chronic graft-versus-host type of disease (15). The fact that theHIV envelope binds to CD4 and might havesmall regions of similarity to MHC antigenhas led to the idea that the viral envelopestimulates the T cell receptor complex in amanner similar to allogeneic (geneticallynonidentical) MHC. In this respect it mightact like a superantigen to activate or deleteT cells in an antigen-independent manner(15), and those patients with inappropriateMHC haplotypes and T cell receptor reper-toires would progress more rapidly to AIDS.As activated lymphocytes are more permis-sive to HIV replication, alloactivation

SCIENCE * VOL. 260 * 28 MAY 1993

would act synergistically with more directcytopathic effects of the virus. However, theevidence for alloactivation is weak.

Another intriguing finding with HIV-infected asymptomatic persons is that thelymphocytes are primed to undergo pro-grammed cell death (apoptosis) when stim-ulated in culture (16). Apoptosis in thethymus is a natural process that eliminatesautoreactive T lymphocytes to establishself-tolerance. Ameisen and Capron (16)have postulated that apoptosis is a markerof immune cell dysfunction; HIV infectionleads to early priming of lymphocytes forsuicide upon further stimulation. They ar-gue that if apoptosis also occurs in vivo to ahigher degree than normal, it could ac-count for helper T cell depletion. However,a major fraction of peripheral blood lym-phocytes (PBLs) from HIV+ persons thatundergo apoptosis surprisingly appear to beuninfected CD8' lymphocytes (17), andapoptosis may be triggered by HIV associ-ated with antigen-presenting cells. Theoverall significance of the sensitivity of Tlymphocytes to apoptosis ex vivo has yet tobe clarified.

Which Cell Types and TissuesBecome Infected by HIV in Vivo?

The CD4 differentiation antigen of the Tlymphocyte serves as the major cell surfacereceptor for HIV, which helps to explainthe depletion of CD4' T helper lympho-cytes in AIDS (18). Even during theasymptomatic phase, many HIV+ individ-uals show a selective defect in T cell re-sponse to recall antigens (19). Memory Thelper cells (which are poised to respond tosecondary exposure to antigen) may bepreferentially infected over naive cells, anddifferent T helper subsets respond in com-plex ways to different stimuli at differentstages of HIV infection (20). The othermajor cell type that becomes infected byHIV is the macrophage (2, 21). Tissuessuch as the lung and brain harbor HIVmainly in macrophage-type cells (alveolarmacrophages and microglia, respectively).

Both T lymphocytes and macrophagesbecome infected via CD4, as shown by theability of monoclonal antibodies to CD4 toblock HIV entry into these cells in culture(2, 21). However, Fc and complementreceptors may also be involved in macro-phage infection by opsonized virus, as mostof the plasma virus particles in seropositiveindividuals will have bound antibodies.Several CD4- cell types can be infected byHIV in vitro, including epithelial and en-dothelial cells as well as astroglial, oligo-dendroglial, and neuronal cells from thecentral nervous system (CNS) in whichgalactocerebroside acts as an HIV receptor.Infection of these cell types in vivo remains

-0gnawl

controversial (2, 21). Blood dendritic cellsand their counterparts in the skin andmucous membranes, the Langerhans cells,have been reported to support HIV replica-tion (22), but this has been questioned bySteinman's group (23), which failed toconfirm HIV replication in dendritic cells.

It has long been known from electronmicroscope and immunofluorescence studies(24) that HIV is found in massive amountsin the lymph nodes, even in the asymptom-atic phase of infection. In addition to helperT lymphocytes and macrophages, virus par-ticles are frequently associated with folliculardendritic cells (FDC), which have a distinctlineage from blood dendritic cells. Thesefindings have recently been confirmed bymolecular techniques (25). Thus active HIVreplication is evident in lymph nodes at allstages of infection.

As proviral DNA is not detectable inFDC (25), it seems likely that the FDCentrap HIV particles and activate B lym-phocytes, as previously suggested (26).Similarly, antigen-presenting cells derivedfrom blood dendritic cells may capture andpresent HIV particles to clusters of helper Tcells in lymph nodes, whereupon viral rep-lication in the lymphocytes leads to theirdepletion in situ by viral cytopathicity andapoptosis (23, 27).

Is There a Latent State ofHIV Infection?

It is likely that some T cells harbor genu-inely latent HIV genomes, that is, DNAproviruses that do not express viral RNA.However, viral latency does not occur in allinfected cells, as is the case with varicella-zoster virus infection during the period be-tween chicken pox and shingles. RestingCD4' T cells in culture become abortivelyinfected, resulting in arrested, incompleteproviruses, unless the cells are activatedwithin a few days of infection (28). In vivo,however, latently infected PBL can be de-tected that carry complete, integrated pro-virus (29). These are probably memory cellsthat may have become infected while in anactive state and that became quiescent be-fore virus replication could be completed.In contrast, macrophages can become pro-ductively infected by HIV in a mature,nonproliferating, but immunologically ac-tive state (30). Notwithstanding the truelatency of HIV in individual resting lym-phocytes, the infection is not latent duringthe asymptomatic phase. Actively ex-pressed HIV is found in lymph nodes andother lymphoid organs and in tissue macro-phages at all stages of infection, indicatinga much higher level of virus activity than incirculating T cells (24, 25). Plasma viremiaalso varies, showing transient small peaksand troughs (Fig. 1). In a recent study, in

which quantitative PCR was used, viremiawas detected at all stages with highest virusparticle titers before seroconversion and inAIDS-related complex (ARC) and AIDS(31). It will be important to determinewhether the immune system is failing toclear virus replication or whether ongoinginfection is regenerated from latent provi-ruses as T cells become activated.

The common usage of the term asymp-tomatic for patients without opportunisticinfections, dementia, or severe weight lossdoes not mean that there are never signs ofinfection or relatively minor symptoms suchas diarrhea and night sweats. Peripheralgeneralized lymphadenopathy syndrome(PGL or LAS) is a feature of HIV infectionthat was recognized early in the AIDS epi-demic; hence the former name lymphaden-opathy virus (LAV) for the first HIV isolate,which was made from a lymph node biopsy(32). Thus, there are signs of HIV activitythroughout the course of the infection.

What Causes T Helper CellDepletion?

This question has generated much debateand a plethora of explanations. The precisemechanism by which HIV kills cells maynot matter in understanding how HIV caus-es AIDS, but it may be important fordetermining therapeutic strategy. Directkilling could be due to a cytolytic effect ofthe virus and to immune attack on virus-infected cells. Indirect killing of uninfectedcells could be due to adsorption of shedgpl20, cell fusion, interference with Thelper and dendritic cell function, or in-duction of T suppressor cells.

HIV can cause direct cytopathic effectsin activated CD4+ T cells in culture, eitherin single cells (33) or by syncytium induc-tion (18, 21). Syncytia soon die in cultureand would be expected to have an evenshorter half-life in the blood, although theyare sometimes seen in the brain and othertissues (2, 14). By incorporating nonin-fected cells into syncytia, a single gpl6O-expressing cell can eliminate many unin-fected CD4+ cells, the so-called bystandereffect. HIV-expressing cells will also bekilled by HIV-specific cytotoxic T cell re-sponses, which are the normal mechanismfor eliminating virus-infected cells by cell-mediated immunity. Antibody-dependent,complement-mediated cytotoxicity andother humoral immune effects may alsohelp to remove HIV-infected cells. Thenthere is the apoptotic phenomenon alreadyalluded to, although its significance in vivois not clear.T cell precursors in the thymus or pe-

ripheral pools may be infected with HIVand therefore fail to proliferate and replen-ish the mature T helper lymphocyte popu-

SCIENCE * VOL. 260 * 28 MAY 1993

lation. Defective antigen presentation (seebelow) will also inhibit T cell proliferation.During the asymptomatic period there is afairly steady decline in numbers of circulat-ing T helper lymphocytes (Fig. 1) oftenfollowed by a steeper decline in ARC andAIDS. Cell replenishment does not matchcell loss.

Among the CD4+ T lymphocyte sub-sets, HIV appears to infect and affect thefunction of memory cells over naive cells(20). Recent evidence also indicates thatthe effects of HIV on T cell responsesduring the asymptomatic stage reflects de-pletion or anergy of TH1 (helping cellularimmunity) in proportion to TH2 (helpinghumoral immunity) as also seen in manyparasitic infections (27). This may help toexplain the persistent activation of B cellsand the perturbation of T cell responses toantigen-presenting cells. The depletion ofCD4+ T cells by HIV may affect the overallhomeostasis of T cell populations, as re-cently reviewed by Stanley and Fauci (34).

What Role Do Macrophages andDendritic Cells Play in AIDS?

Infected macrophages could be importantreservoirs outside the blood and as carriersofHIV to different organs (the Trojan horsemetaphor). These nonproliferating, maturecells can sustain HIV production in vitrofor a considerable time without being killedby the virus. Cytokine secretion by infectedmacrophages is aberrant (35), which canlead to a cascade of secondary effects thatare likely to be important in the wastingsyndrome ("slim disease") and CNS diseaseseen in many AIDS patients. Indeed, thewasting and CNS attributes of humanAIDS closely resemble the visna-maedi syn-drome in sheep. Visna-maedi virus andequine infectious anemia virus are mac-rophage-tropic lentiviruses that do not in-duce severe depletion ofCD4+ T cells (36).In a recent report (37) the wasting syn-drome categorized by > 10% weight loss wasnot correlated with falling CD4 T cellcounts, but with interferon-mediated acti-vation of macrophages as measured by uri-nary neopterin levels. Of course, the sepa-ration of T cell immune deficiency fromwasting and CNS disease is a simplisticview that ignores the intricate relation be-tween antigen-presenting cells and effectorT lymphocytes. Nonetheless, it may beuseful to think of the enteropathic, wast-ing, and brain diseases as being linked tomacrophage infection and distinct from thesevere immunodeficiency caused by T help-er cell depletion.

Macrophages act as antigen-presentingcells particularly to memory T cells in theperiphery. This interaction does not appearto be significantly impaired during the

1275

m.,.A 111

asymptomatic phase of HIV infection. Incontrast, Knight's group and others (22,27) argue that the function of dendriticcells and macrophages in bringing and pre-senting antigen to naive T cells in thelymph nodes is affected early in the courseof HIV infection. Antigen-presenting cellswould serve both as a reservoir of HIV andas an effector of immune dysfunction, in-ducing anergy ofTHl and activation ofTH2lymphocytes (27). Patients may survive ontheir memory T cell responses, but whenthese cells eventually become depleted, thelack of recruitment of new memory cellsthat should normally have occurredthrough the interaction of antigen-present-ing cells with naive T cells could contributeto immune deficiency.

Is Genetic Variation Importantin Pathogenesis?

Lentiviruses are notoriously variable in ge-nome sequence. Within one infected host,millions of genetic subtypes exist. This"swarm" ofgenomes is called a quasi-species(38), although I prefer the old-fashionedterm "population polymorphism." Thus thegenetic diversity of HIV in vivo is vast, andthe population contains a large proportionof defective genomes. As soon as HIV isgrown in culture there is bound to beselection for a smaller number of subtypes.On this account, Wain-Hobson and col-leagues have stated that "to culture is todisturb" (38). While agreeing that this isundoubtedly the case, I would also say that"to culture is to discern"-the replication-competent viruses from genomic "noise."

There are three main aspects of geneticvariation of HIV that could influencepathogenesis. (i) The development of tro-pisms for efficient replication in differentcell types will affect the qualitative at-tributes of the syndrome-immune defi-ciency, wasting, enteropathic disease, andCNS disease. (ii) Antigenic variation canlead to escape from specific humoral andcell-mediated immune controls. (iii)Strains that differ in virulence may affectthe pace and severity of disease.

In my opinion, too much emphasis hasbeen placed on antigenic variation in iso-lation from considerations of viral titer, celltropism, and virulence. For example, it iswidely held that variation in the V3 loopsequence in gpl20 is driven by escape fromimmune responses, which has been seen invitro and in vivo (39). However, the V3loop also contributes to the determinationof lymphocyte versus macrophage tropismand, among T lymphotropic strains, to theemergence of highly cytopathic, syncytium-inducing variants (2, 14, 40). In primaryHIV infection, before seroconversion,there is uniformity of genome sequences

1276

within an infected individual (41), proba-bly due to a small founder population andselection of the fastest growing virus. Wedo not yet know to what extent the ensuingvariation is random for an expanding pop-ulation, driven by immune escape, anddriven by adaptation to new cell types.

The source of variation is the infidelity ofreverse transcriptase, which has no editingmechanism for transcriptional errors (38).Once the quasi-species has established adegree of variability, however, genetic re-combination (another attribute of retrovi-ruses) may allow new combinations of mu-tations to be selected. For instance, onevariant showing, for example, zidovudineresistance in the pol gene could recombinewith one showing escape from neutralizationin the env gene. Recombination may beparticularly rapid with syncytium-inducingstrains as cell fusion will permit reassortmentof RNA genomes derived from differentproviruses. Perhaps that is why these strainsbecome resistant to zidovudine more rapidly.Recombination as a source of genetic varia-tion merits further investigation.

What Is HIV Tropism?

Distinct substrains of HIV within the infect-ed body differ in their tropisms for differentcell types. Thus HIV isolates from the CNSincluding the cerebrospinal fluid (CSF) pref-erentially grow in macrophages in culture,whereas HIV isolates obtained from periph-eral blood lymphocytes (PBL) that havebeen stimulated by phytohemagglutinin(PHA) and interleukin-2 (IL-2) propagatebest in the same cell type (2). Such virusesare called macrophage-tropic and T cell-tropic, respectively. In reality, there is con-fusion about the terminology of HIV tro-pism, which, despite elegant molecular anal-ysis of virus strains and recombinants, hastended to obscure our understanding of theway HIV propagates in vivo.

First, there is actually a broad spectrumof the relative efficiency of HIV replicationbetween T lymphocytes and macrophagesthat is influenced by the cell type used forinitial culture (42). Second, there is confu-sion between the capacity of HIV to growin immortalized or leukemic T cell lines andreplication in primary PHA + IL-2-stimu-lated PBL; only a minority of HIV isolates,which are adapted in the laboratory, growin T cell lines, whereas nearly all isolatesincluding macrophage-tropic strains growin PBL. Third, primary isolates from PBL ofasymptomatic persons tend to replicateslowly and attain lower final titers in cul-ture, whereas isolates from AIDS patientsoften grow rapidly to high titers; they havebeen named slow-low and rapid-high strains(43). The latter are the viruses that mostquickly adapt to growth in T cell lines.

SCIENCE * VOL. 260 * 28 MAY 1993

However, both slow-low and rapid-highviruses grow to high titer in highly activat-ed primary T lymphocytes, as in mixtures ofcord-blood lymphocytes or PBL from twodonors, when allogeneic activation occurs(43). Fourth, HIV isolates can be classifiedas syncytium-inducing (SI) and non-syncy-tium-inducing (NSI). SI isolates are ob-tained more frequently but not inevitably inpatients with AIDS and ARC (44). Rapid-high isolates often have the SI phenotypeor acquire it during early passage in vitro.The SI/NSI classification is now gaininggeneral acceptance.

Nearly all types ofHIV isolate propagatewell in primary, stimulated PBL (42, 43).One real distinction is that most NSI andmacrophage-tropic strains do not replicatein established T cell lines. Conversely, Tcell line-adapted, SI isolates replicate poor-ly in macrophages, although Lazdins et al.(30) report that this restriction can beovercome by cultivating macrophages withtransforming growth factor 1.

Thus, the differential cellular tropisms ofHIV are not as hard and fast as might bethought. The rapid-high SI virus strainsoften presage AIDS, appearing before thefinal, rapid depletion of CD4 lymphocytesin the circulation (44). Whether SI virusesare primarily a cause or a consequence ofimmune suppression, it is likely that theyexacerbate AIDS once they take hold as themajor virus population (45).

Animal models of lentivirus infectionmay help to elucidate viral tropism andpathogenesis, and I have already alluded tovisna-maedi disease. The use of severe com-bined immunodeficient mice transplantedwith human lymphoid cells or organs (Hu-SCID mice) may help to illuminate HIVpathogenesis (46), although I once queriedwhether Hu-SCID mice are more than ex-pensive, furry culture flasks for PBL. It isnow clear that these mice allow lymphoiddevelopment from implants such as humanfetal thymus, and the system is susceptibleto infection and depletion by HIV. How-ever, the Hu-SCID model resembles pri-mary, pre-seroconversion HIV infectionrather than AIDS.

Simian immunodeficiency virus (SIV)infection of macaques provides an animalmodel that can aid our thinking on patho-genesis in relation to viral tropism, al-though there are differences, such as V3being invariant in SIV (47). Narayan'sgroup (48) demonstrated that if SIV recov-ered from a molecular clone with a Tcell-tropic phenotype was inoculated intothe brains of macaques, it did not result ininfection. If, however, the virus was inoc-ulated intravenously, infection of lympho-cytes occurred. When macrophage-tropicvariants appeared 3 months later, they wereable to replicate in the brain upon inocula-

- X~~~~~~~~~~~~~~~~~~--....1------------------------------aamum 1 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~r0:0lt~:::Xs

tion into naive monkeys. Thus tropismvariants analyzed and cloned in vitro dohave significance in vivo.

As Daniel et al. (49) have recentlyshown, specific deletion of nef or othergenes from SIV dramatically affects viralreplication in vivo and protects the animalsagainst challenge with a virulent strain.Studies of individual viral genes will help usto elucidate the process of pathogenesis, butit may be a misconception to think of oneor the other of the HIV or SIV genes as a"pathogene," in the same sense as theoncogenes of acutely transforming retrovi-ruses. Deletion of any viral gene that isrequired for efficient replication in vivo willsurely lead to virus attenuation.

Does Antigenic DiversityCause AIDS?

Considerable antigenic variation is evidentin all lentivirus infections. With HIV,there is evidence for selection of escapemutants from neutralizing antibodies (39)and possibly for escape from CTL as well(50). Antigenic variation presents a consid-erable problem for vaccine development asit will be essential to induce broadly cross-protective immunity against the divergingstrains as HIV spreads across the world.Antigenic variation is also postulated toaccount for HIV pathogenesis. Initially, thehost raises excellent immune responses toHIV, and that leads to the long asymptom-atic period upon seroconversion. Antigenicvariation permits escape from immune re-sponses, and the resurgence of sequentialvirus variants could lead to AIDS.An interesting analysis of antigenic vari-

ation as a contributor to pathogenesiscomes from the mathematical modeling ofNowak and May, based on antigenic diver-sity thresholds (51). Basically, they arguethat the generation of antigenic variantscauses an asymmetric interaction betweenimmunological and viral diversity. Whileeach virus strain impairs most immune re-sponses by diminishing MHC-II-restrictedT helper cells, strain-specific immune re-sponses can only control specific virus vari-ants. Eventually a diversity threshold isreached when the immune system fails tocontrol the virus population. According tothis model, HIV infection is an evolution-ary process that determines the time scalefrom infection to disease.

Although Nowak and May can fit theirmodel mathematically for all outcomes(acute disease, delayed disease, and no dis-ease), I am wary of accepting this Panglos-sian view in which perturbations of group-and strain-specific immune responses yieldquite different pathogenic consequences.The genetic diversity and load of SIV inasymptomatic African monkeys appears to

be similar to humans (52), yet only thelatter develop AIDS. Chimpanzees can beinfected with HIV-1 and immune escapemutants occur, but they have not yet de-veloped disease. I therefore suspect thatsomething besides a threshold of antigenicdiversity leads to pathogenesis. Diseasemight be a consequence of sheer viral bur-den, host immunogenetics, or possibly allo-stimulation or APC dysfunction becausePBL from infected chimpanzees do not ex-hibit apoptosis ex vivo (16, 53). This mayalso be true of other lentivirus infections;asymptomatically infected sheep from Ger-many introduced visna-maedi virus into therelatively inbred Icelandic sheep with dev-astating effects. These considerations re-quire discussion in terms of the evolution ofvirulence and tolerance in host-virus rela-tions (51, 54).

Do HIV Strains Ddffer in Virulence?

Until recently, variation in virulence hasbeen largely an ignored parameter in thedebate about HIV pathogenesis. Yet almostall pathogens produce virulent and morebenign variants. With HIV and relatedlentiviruses, the quasispecies are so diversethat different grades of virulence could begenerated within one individual.

The clearest evidence for lentivirus vari-ants that heritably differ in their pathogeniceffect comes from experimental SIV infec-tions. The Pbj 14 strain of SIV killsmacaques in 10 to 14 days by causing amassive lymphoid infiltration of the gutthat results in hemorrhagic diarrhea. Whengiven rehydration therapy, however, themonkeys can survive to develop more con-ventional AIDS 3 to 4 months later. Thepoint of the argument is not that the acutedisease differs from AIDS but that it is areproducible property of the virus, even ofan infectious molecular clone (55). Othermolecular clones of SIV also differ in viru-lence for inducing AIDS (49).

There are a few cases of fatal primaryinfections of HIV (56), but the virus hasbeen insufficiently characterized to saywhether this was a result of virulent strainsor exceptionally susceptible hosts. Con-versely, a cluster of long-surviving personsinfected by HIV from a single source hasbeen reported in Australia (57), raising thepossibility that this virus is relativelyapathogenic, or slow to cause disease.HIV-2 is thought to have a slower timecourse to AIDS than HIV-1 (2), and re-cently, SIV-like genomes were described inLiberians who maintain a very low load ofunculturable virus and show no symptomsof disease (58).

I have already discussed the variation indisease spectrum that may reflect the bal-ance of viral tropisms within the quasispe-SCIENCE * VOL. 260 * 28 MAY 1993

cies. It seems clear that variants can differin virulence and in the type of disease theyare capable of inducing. Roos et al. (59)found that NSI and, more seldom, SI virus-es occur in primary infections from trans-mitters with both types of virus; their pre-liminary evidence suggests that the recipi-ents with SI viruses may show a more rapiddecline in CD4+ T cells.

One may ask whether and how the clockis reset each time HIV transmits from oneperson to another. The transmission ofvirulent subtypes of HIV, and of drug-resistant variants, will require further study.Given that we have only witnessed thebeginning of the HIV pandemic, and thatthe scope for viral variation within andbetween humans is enormous, novel patho-genic HIV variants might emerge (60).One may also hope that when the rate oftransmission slows, less virulent strains willpredominate, as has been suggested forHIV-2 (54).

Synthesis

I have outlined several possible contribut-ing factors in HIV pathogenesis: (i) otherinfections activate HIV by stimulating thelymphocytes that harbor the proviruses; (ii)the HIV envelope itself nonspecifically ac-tivates lymphocytes in a manner similar toan allogeneic reaction-this renders thecells sensitive to apoptosis; (iii) antigen-presenting cells (macrophages and dendriticcells) produce aberrant cytokine signals,causing changes in T cell responses includ-ing a switch from TH1 to TH2 activity; (iv)memory T lymphocytes are not replenishedas fast as they are killed by HIV or immuneresponses to HIV antigens; (v) HIV devel-ops sequential escape mutants to keep onestep ahead of the immune response; (vi)cumulative antigenic diversity in the evo-lutionary dynamics of HIV overcomes theimmune system; and (vii) virulent cyto-pathic HIV variants emerge. We need todevise experiments that would allow us todistinguish between some of the models. Acommon feature among them is that persis-tent HIV replication occurs. It generatesconditions that promote its continuedgrowth; in other words, HIV eventuallybecomes its own opportunistic infection.

Almost all the models involve unbal-anced immune activation as a prelude toimmune collapse. An exception may be thedown-regulation of lymphocytes by affectedantigen-presenting cells (22), but even here,differential effects on naive and memory cellscould lead to simultaneous activation andanergy in different T cell subsets (27). I havealso postulated that severe immunosuppres-sion resulting from a deficiency in T cellsand the wasting/CNS disease may be twodistinct attributes of AIDS. If that is the

1277

case, distinctive therapeutic approachesmight be sought. And if immune alloactiva-tion by HIV is a significant factor, "thera-peutic vaccination" with envelope-based im-munogens might exacerbate the disease. Anexcellent review by Angela McLean on "thebalance of power between HIV and theimmune system" has just appeared (61), inwhich concepts similar to those reviewedhere are placed within the context of asimple mathematical model.

In conclusion, the foregoing discussionillustrates the complexity of HIV infectionand the many ideas on how the virus causesAIDS. There are enthusiasts for each of thedifferent aspects, and it is too early in theepidemic and in research to be dogmaticabout mechanism. Indeed many of themodels of HIV pathogenesis are not mutu-ally exclusive. We are like the blind menwho encountered an elephant: Each of ushad a different image of AIDS because wecould not grasp the whole. The emergingview, however, is that a progressive HIVburden involving first activation and even-tually destruction of the immune system iswhat lies behind AIDS.

REFERENCES AND NOTES

1. M. S. Gottlieb et al., N. Engi. J. Med. 305, 1425(1981).

2. J. A. Levy, Microbiol. Rev. 57,183 (1993).3. R. S. Root-Bernstein, Rethinking AIDS: The Tragic

Cost of Premature Consensus (Free Press, Mac-millan, New York, 1993).

4. P. H. Duesberg, Proc. Natl. Acad. Sci. U.S.A. 86,755 (1989); ibid. 88, 1575 (1991); Lancet341, 957(1993).

5. R. A. Weiss and H. W. Jaffe, Nature 345, 659(1990); M. S. Ascher, H. W. Sheppard, W. Winkel-stein, Jr., E. Vittinghoff, ibid. 362,103 (1993); M. T.Schechter et al., Lancet 341, 658 (1993).

6. A. S. Fauci, N. Engi. J. Med. 328, 429 (1993).7. G. Corbitt, A. S. Bailey, G. Williams, Lancet 336,

51 (1990).8. R. Y. H. Wang etal., ibid. 340, 1312 (1992); V. L.

Katseni et al., ibid. 341, 271 (1993).9. A. Webster etal., Clin. Exp. Immunol. 88, 6 (1992).

10. A. R. McLean and M. A. Nowak, J. Theor. Biol.155, 69 (1991).

11. V. Beral, T. A. Peterman, R. L. Berkelman, H. W.Jaffe, Lancet 335, 123 (1990); V. Beral, H. W.Jaffe, R. A. Weiss, Eds., Cancer, HIV and AIDS(Cold Spring Harbor Laboratory, Cold Spring Har-bor, NY, 1991).

12. K. deCock, B. Soro, I. M. Coulibaly, S. P. Lucas, J.Am. Med. Assoc. 268, 1581 (1992); D. Mulderreports 11% annual mortality in HIV+ adults inUganda (VIII International AIDS Conference, ab-stract PoC 4021, 1992). However, C. Gilks pointsout (personal communication) that the actual rateof immune impairment in Africans infected withHIV-1 may be no faster than in Western countriesbut that patients without access to treatment maydie earlier as a result of their first episode ofopportunistic infection.

13. C. M. Steel et al., Lancet i, 1185 (1988); R. A.Kaslow et al., ibid. 335, 927 (1990); P. U. Camer-on et al., Hum. Immunol. 29, 282 (1990); P.Simmonds et al., Lancet 338, 1159 (1991); F.Hentges, A. Hoffmann, F. Oliviera de Araujo, R.Hemmer, Clin. Exp. Immunol. 88, 237 (1992).

14. F. Miedema, Immunodefic. Rev. 3, 173 (1992); G.Janossy, B. Autran, F. Miedema, Eds.,Immunodeficiency in HIV Infection and AIDS(Karger, Basel, Switzerland, 1992).

1278

15. G. M. Shearer, N. Engi. J. Med. 308, 223 (1983);H. Golding etal., J. Exp. Med. 167, 914 (1988); J.Clin. Invest. 83,1430 (1989); L. Imberti, A. Sottini,A. Bettinardi, M. Puoti, D. Primi, Science 254, 860(1991); M. Clerici et al., Eur. J. Immunol. 21, 665(1991); A. G. Dalgleish et al., Lancet 339, 824(1992); J. Habeshaw, E. Hounsell, A. G. Dalgleish,Immunol. Today 13, 207 (1992); H. W. Sheppardand M. S. Ascher, J. Acquired Immune Defic.Syndr. 5, 143 (1992).

16. H. Groux etal., J. Exp. Med. 175, 331 (1992); J.-C.Ameisen and A. Capron, Immunol. Today 12,102(1991); M. L. Gougeon etal., CR. Acad. Sci. 312,529 (1991).

17. L. Meyaard et al., Science 257, 217 (1992).18. D. Klatzmann et al., ibid. 225, 59 (1984);

Nature 312, 767 (1984); A. G. Dalgleish et al.,ibid., p. 763.

19. M. Clerici et al., Nature 339, 383 (1989); F.Miedema, M. Tersmette, R. A. W. van her, Immu-nol. Today 11, 299 (1990); G. M. Shearer and M.Clerici, AIDS 5, 245 (1991).

20. C. J. M. van Noesel et al., J. Clin. Invest. 86, 293(1990); S. M. Schnittman et al., Proc. Natl. Acad.Sci. U.S.A. 87, 6058 (1990); A. Cayota, F. Vuiller,D. Scott-Algeira, V. Feuille, G. Dighiero, Clin. Exp.Immunol. 91, 241 (1993).

21. R. A. Weiss, in The Retroviridae, J. A. Levy, Ed.(Plenum, New York, 1993), vol. 2, pp. 1-108.

22. S. Patterson and S. C. Knight, J. Gen. Virol. 68,1177 (1987); E. Langhoff et al., Proc. Natl. Acad.Sci. U.S.A. 88, 7998 (1991); S. E. Macetonia etal.,Immunology 75, 576 (1992).

23. P. U. Cameron et al., Clin. Exp. Immunol. 88, 226(1992); Science 257, 383 (1992). These authorsthink that the HIV replication previously reported(22) was due to T cell and macrophage contam-ination of dendritic cell preparations, whereas S.Knight and S. Patterson argue (personal commu-nication) that the more mature, veiled dendriticcells that are permissive to HIV express some Tcell markers and were possibly removed by therosetting procedure employed by Cameron et al.

24. J. A. Armstrong and R. Horne, Lancet ii, 370(1984); K. Tenner-Tacz, P. Racz, M. Dietrich, P.Kem, ibid. i, 105 (1985); K. Tenner-Racz et al.,Am. J. Pathol. 123, 9 (1986).

25. G. Pantaleo et al., Nature 362, 355 (1993); J.Embretson et al., ibid., p. 359.

26. J. H. Humphrey, Lancet ii, 643 (1984); J. A.Armstrong, R. L. Dawkins, R. Horne, Immunol.Today 6, 121 (1985).

27. M. Clerici et al., J. Clin. Invest. 91, 759 (1993); M.Clerici and G. M. Shearer, Immunol. Today 14,107 (1993); L. Meyaard, H. Schuitemaker, F.Miedema, ibid., p. 161.

28. J. Zack, A. M. Haislip, P. Krogstad, I. S. Y. Chen,J. Virol. 65, 2102 (1991); M. I. Bukrinsky, T. L.Stanwick, M. P. Dempsey, M. Stevenson, Science254, 423 (1992).

29. S. M. Schnittman et al., Science 245, 305 (1989);J. E. Brinchmann, J. Albert, F. Vartdal, J. Virol. 65,2019 (1991); A. Chapel, A. Bensussan, E. Vilmer,D. Dormont, ibid. 66, 3966 (1992).

30. S. Gartner et al., Science 233, 215 (1986); D. D.Ho, T. R. Rota, M. S. Hirsch, J. Clin. Invest. 77,1712 (1986); A. Valentin, A. V. Gergerfelt, S.Matsuda, K. Nilsson, B. Asjo, AIDS 4, 751 (1991);J. B. Weinberg, T. J. Matthews, B. R. Cullen, M. H.Malim, J. Exp. Med. 174, 1477 (1991); J. K.Lazdins et al., AIDS Res. Hum. Retroviruses 8,505 (1992).

31. D. D. Ho, T. Moudgil, M. Alam, N. Engi. J. Med.321, 1622 (1989); R. W. Coombs et al., ibid., p.1626; M. Piatak, Jr., et al., Science 259, 1749(1993). Piatak et al. detect high titers of plasmavirions, of which approximately 1:60,000 are cul-turable. Previously, significant plasma titers ofinfectious HIV had been reported by D. D. Ho, T.Moudgil, M. Alam, N. EngI. J. Med. 321, 1621(1989); R. W. Coombs et al., ibid., p. 1626.

32. F. Barr6-Sinoussi et al., Science 220, 868 (1983);J. C. Gluckman, ibid. 259,1809 (1993).

33. M. Somasundaran and H. L. Robinson, J. Virol.61, 3114 (1987); I. Hirsch, D. Salsun, B.Brichacek, J. C. Chermann, Virology 186, 647

SCIENCE * VOL. 260 * 28 MAY 1993

(1992); Bergeron, N. Sullivan, J. Sodroski, J. Virol.66, 2389 (1992).

34. S. K. Stanley and A. S. Fauci, J. Acquired ImmuneDefic. Syndr. 6, 142 (1993).

35. D. Giulian, K. Vaca, C. A. Noonan, Science 250,1593 (1990); A. S. Fauci, S. M. Schnittman, G.Poli, S. Koenig, G. Pantaleo, Ann. Intern. Med.114, 678 (1991).

36. M. D. Gorrell, M. R. Brandon, D. Sheffer, R. J.Adams, 0. Narayan, J. Virol. 66, 2679 (1992); D.C. Sellon, S. T. Perry, L. Coggins, F. J. Fuller, ibid.,p. 5906.

37. R. Zangerle, G. Reibnegger, H. Wachter, D.Fuchs, AIDS 7, 175 (1993).

38. A. Meyerhans et al., Cell 58, 901 (1989).39. J. A. McKeating etal., AIDS3, 777 (1989); T. V. W.

Wolfs et al., Virology 185, 195 (1991); M. Aren-drup et al., J. Gen. Virol. 74, 855 (1993).

40. H. Schuitemaker et al., J. Virol. 65, 356 (1991); R.A. M. Fouchier et al., ibid. 66, 3183 (1992).

41. T. McNearney et al., Proc. Nati. Acad. Sci. U.S.A.89, 10247 (1992); T. F. W. Wolfs, G. Zwart, M.Bakker, J. Goudsmit, Virology 189, 103 (1992); P.Kasper et al., AIDS Res. Hum. Retroviruses 9, 153(1993).

42. H. E. Gendelman et al., AIDS 4, 221 (1990); C.Cheng-Mayer, D. Seto, J. A. Levy, Virology 181,288 (1991); A. Valentin, thesis, Karolinska Insti-tute, Stockholm, Sweden (1992); M. Collin, thesis,University of Oxford (1992).

43. B. Asjo etal., Lancet ii, 234 (1986); E. M. Fenyo etal., J. Virol. 62, 4414 (1988); W.-P. Tsai etal., AIDSRes. Hum. Retroviruses 8, 1749 (1992).

44. M. Tersmette et al., J. Virol. 63, 2118 (1989); H.Schuitemaker et al., ibid. 66, 1354 (1992); R. I.Connor, H. Mohri, Y. Cao, D. D. Ho, ibid. 67,1772(1993); M. Koot et al., Ann. Intern. Med., in press.

45. Although SI viruses appear before end-stageAIDS, it may be necessary for the host to dipbelow a threshold of immunocompetence in orderfor the virus to escape immune control. Alterna-tively, B. Asjo has pointed out (personal commu-nication) that SI variants may be adaptations to ahost in which the usual target cells for NSI viruseshave already been lost.

46. J. M. McCune, Cell 64, 351 (1991); D. E. Mosier,R. J. Gulizia, P. D. Macisaac, B. E. Torbett, J. A.Levy, Science 260, 689 (1993). Mosier et al.report that two "macrophage-tropic" virus strains,HIV-1 strain SF162 and HIV-2 strain UC1, rapidlydepleted CD4+ T lymphocytes in Hu-SCID mice,whereas the "T cell line-tropic" HIV-1 strain SF33did not. However, it would be wrong to assumethat macrophage infection leads to T cell deple-tion, as the Hu-SCID mice are devoid of humanmacrophages. It is an example of the confusionconveyed by tropism nomenclature as the so-called macrophage-tropic strains replicate just aswell in primary PBL. What is odd is that some ofthe HIV-1 strains selected to propagate in estab-lished T cell lines do not grow in human lympho-cytes in SCID mice.

47. J. Overbaugh et al., J. Virol. 65, 7025 (1991); N.Almond, A. Jenkins, A. B. Heath, P. Kitchin, J.Gen. Virol. 74, 865 (1993).

48. D. P. Sharma et al., J. Virol. 66, 3550 (1992).49. M. D. Daniel, F. Kirchhoff, S. C. Czajak, P. K.

Sehgal, R. C. Desrosiers, Science 258, 1938(1992).

50. R. E. Phillips et al., Nature 354, 453 (1991).51. M. A. Nowak eta/., Science254, 963 (1991); M A.

Nowak and R. M. May, AIDS 7, 53 (1993);J. Theor. Biol. 159, 329 (1992).

52. M. Baier, M. T. Dittmar, K. Cichutek, R. Kurth,Proc. Natl. Acad. Sci. U.S.A. 88, 8126 (1991); S.Hartung et al., J. Virol. 66, 2143 (1992).

53. M. Watanabe etal., J. Virol. 65, 3344 (1991); H. E.Gentelman et al., ibid., p. 3853.

54. P. W. Ewald, Hum. Nat. 2, 1 (1991); Science 257,10 (1992).

55. S. Dewhurst, J. E. Embretson, D. C. Anderson, J.1. Mullins, P. N. Fultz, Nature 345, 636 (1990); F. J.Novembre et al., J. Virol. 67, 2466 (1993).

56. S. Oka et al., VIII International AIDS Conference,abstract PoC 4752 (1992); D. J. Jeffries, personalcommunication.

.:. -.52...1 .11

FAR04 1~'

57. J. Learmont et al., Lancet 340, 863 (1992).58. F. Gao et al., Nature 358, 495 (1992).59. M. T. L. Roos et al., J. Infect. Dis. 165, 427 (1992).60. J. J. Holland, Ed., Curr. Top. Microbiol. Immunol.

176,1 (1992); S. Wain-Hobson, in The Evolution-ary Biology of Viruses, S. Morse, Ed., in press.

61. A. R. McLean, Trends Microbiol. 1, 9 (1993).

62. S. C. Darby etal., Brit. Med. J. 298, 1064 (1989);R. J. Biggar, AIDS 4, 1059 (1990); G. W. Ruther-ford et al., Brit. Med. J. 301, 1183 (1990); A. N.Phillips et al., J. Am. Med. Assoc. 268, 2662(1992); A. N. Phillips, C. A. Sabin, M. Befill, G.Janossy, C. A. Lee, Lancet 341, 1024 (1993).

63. AMthough this is a personal view, am grateful for

critical discussions and comments to: B. Asjo, P.Brown, P. R. Clapham, A. G. Dalgleish, E. M. Fenyo,D. J. Jeffries, M. 0. McClure, A. R. McLean, J. P.Moore, F. Miedema, M. Nowak, S. Patterson, A. J.Pinching, H. Schuitemaker, T. F. Schulz, S. Wain-Hobson, and J. N. Weber. Supported by the MedicalResearch Council.

Scientific and Social Issues of

Human Immunodeficiency VirusVaccine Development

tors to pursue several vaccine tracks simul-taneously in hope of the rapid developmentof a successful preventive HIV vaccine (9,10) (Fig. 1).

Scientific Problems of HIVPreventive Vaccine Development

Barton F. HaynesDevelopment of a preventive immunogen for human immunodeficiency virus (HIV) infec-tion is a national priority. The complexities associated with HIV host-virus interactions,coupled with the rapid progression of the HIV epidemic worldwide, have necessitatedlowering expectations for an HIV vaccine that is 100 percent effective and have raisedimportant scientific and nonscientific issues regarding development and use of preventiveand therapeutic HIV vaccines.

HIV infection is preventable (1, 2). Inspite of this, HIV is spreading worldwide atan alarming rate, and projections of themagnitude of the pandemic by the year 2000are staggering (3). The development of apreventive HIV vaccine (an immunogenadministered to HIV-uninfected individualsto prevent infection) is a national priority.Efforts have also begun to develop therapeu-tic HIV vaccines, whereby HIV-infectedindividuals would be treated with immuno-gens designed to boost salutary anti-HIVimmune responses, decrease virus-infectedcells, and either eradicate HIV or prolongthe time until development of acquired im-munodeficiency syndrome (AIDS) (4-6).

HIV Preventive VaccineDevelopment

knowledge of pathogenic mechanisms orcorrelates of protective immunity (such asfor the development of vaccines for small-pox or polio) (8), the emergent nature ofthe HIV pandemic, coupled with a plethoraof critical unknowns, has forced investiga-

Although more is known about HIV thanalmost any other infectious agent, scientificquestions remain unanswered that are crit-ical to development of an HIV preventivevaccine.

Optimal requirements for a preventive vac-cine. A successful preventive HIV vaccineshould be safe and effective for the preven-tion or quick eradication of initial HIVinfection by multiple HIV strains, regard-less of HIV exposure by mucosal or paren-teral routes (9, 11-17). It is important toemphasize, however, that most vaccinesprevent disease, not infection. Thus, asuccessful HIV vaccine may not preventestablishment of infection but still mayprevent the development of AIDS. For the

Tracks For Vaccine DevelopmentTracks

Traditional Vaccines - Sequential TracksDevelop a Killed orAttenuated Vaccine

Know Correlates ofUnderstand Immunity or Infectious Develop aPathogenesis Agent Structure Vaccine

* * t

Outcome Examples

Protection SmPallox

Protection Hepoanis B

The difficult scientific issues before us un-derlie the fact that, as yet, there is nopreventive HIV vaccine on the near hori-zon with clear prospects for clinical use.What has been developed are (i) promisingexperimental immunogens and (ii) clearideas of what the central questions are thatshould be asked in ongoing and plannedhuman clinical trials (7). Whereas tradi-tional non-HIV vaccine developmenttracks have led to successful killed or atten-uated immunogens in spite of lack of

The author is the Frederic M. Hanes Professor ofMedicine at the Duke University School of Medicineand is director of basic research at the Duke Centerfor AIDS Research, Durham, NC 27710. He serves asco-chair of the National Academy of Sciences Instituteof Medicine Roundtable for the Development of Drugsand Vaccines Against AIDS.

HIV Vaccine - Five Simultaneous TracksUnderstand PathogenesisUnderstand Correlates

of immunityDevelop Effective

ImmunogensDevelop Effective

Adjuvant FormulationsEstablish Adminitative andCommunity Infrastructurefor Efficacy Testing Sites

Fig. 1. Approaches to vaccine development. Traditional vaccines either use successful approacheswithout knowledge of pathogenesis or correlates of immunity (such as with the development of thesmallpox and polio vaccines) or proceed in sequential tracks of understanding aspects ofpathogenesis, correlates of immunity, or infectious agent structure before development of an

effective immunogen (such as with the hepatitis B vaccine). In contrast, HIV vaccine developmentis proceeding along several simultaneous tracks to maximize the chances of rapidly developing a

successful preventive vaccine.

SCIENCE * VOL. 260 * 28 MAY 1993 1279