AIDS RESEARCH AND HUMAN RETROVIRUSESVolume 19, Number 11, 2003, pp. 957–965© Mary Ann Liebert, Inc.

Assessing Human Alloimmunization as a Strategy forInducing HIV Type 1 Neutralizing Anti-HLA Responses

JONATHAN G. LEITH,1 DAVID A. CLARK,2,3 THOMAS J. MATTHEWS,4 KENNETH L. ROSENTHAL,2

MARK A. LUSCHER,5 BRIAN H. BARBER,1 and KELLY S. MACDONALD1,6,7

ABSTRACT

Xenovaccination of rhesus macaques with human HLA Class I and II proteins has been demonstrated to elicitprotective immunity against challenge with SIV grown in human cells. To determine if alloimmunization inhumans could lead to protective immunity against HIV-1, we prospectively followed a small group of womenreceiving whole-cell alloimmunization in the form of leukocyte immunotherapy for recurrent spontaneousabortion. Whole-cell vaccine recipients and their respective partners (referred to as donors) provided pre-and postimmune blood samples for analysis. Study participants were HLA typed by sequence-specific PCRand antibodies specific for HLA Class I and II antigens were measured in recipient plasma. To determine ifanti-HLA antibody responses detected in recipient plasma samples were capable of neutralizing HIV-1 in vitro,we grew laboratory strain HIV-1IIIB and primary isolate HIV-1301660 in donor-derived CD41 T lymphocytes.The ability of purified whole IgG from responding patients to neutralizing infectivity of the respective donor-derived virus was then assayed in vitro. All donor–recipient pairs were determined to be HLA discordant forat least one Class I and one Class II locus. Two of seven female recipients in total made strong anti-HLA an-tibody responses specific to the HLA haplotype of the male donor in response to the alloimmunization regi-men. For one recipient, IgG antibodies specific for donor HLA Class I and II antigens were able to neutral-ize both HIV-1IIIB and a primary isolate HIV-1301660. In addition polyclonal anti-HLA class II antibodiesagainst a single determinant (DR4) of this donor were also neutralizing. In contrast, the other recipient ex-hibiting antibodies only against donor HLA Class I antigens did not neutralize HIV-1IIIB. Using samples froma small number of women undergoing leukocyte immunotherapy, we have demonstrated for the first timethat allele-specific anti-HLA antibodies elicited through human alloimmunization are capable of neutralizingHIV-1 in vitro.

957

INTRODUCTION

CONVENTION AL VACCINATION attempts to invoke protectiveimmune responses against human immunodeficiency virus

type-1 (HIV-1) have focused on immunization against virus-encoded proteins. However, novel approaches based on host-encoded proteins incorporated into the viral envelope (such ashuman leukocyte antigens, HLA) are also potential targets forstudy. This concept is based on the fact that HLA Class I and

II proteins as well as some other host proteins are selectivelyincorporated into the envelope of the budding HIV-1 virion.1,2

This phenomenon is also seen in other retroviruses includingfeline leukemia virus3 and simian immunodeficiency virus(SIV).4,5 While the underlying mechanisms are incompletelyunderstood,6 the expression of the HLA determinants on thesurface of the HIV-1 envelope could be expected to render thisvirus susceptible to recognition by anti-HLA antibodies. In aremarkable finding nearly 10 years ago, James Stott and col-

1Department of Immunology, University of Toronto, Toronto, Canada.2Department of Pathology, McMaster University, Hamilton, Canada.3Departments of Medicine, and Obstetrics and Gynecology, McMaster University, Hamilton, Canada.4Trimeris Inc., Durham, North Carolina.5Division of Infectious Diseases, University Health Network, Toronto, Canada.6Mount Sinai Hospital, Toronto, Canada.7Department of Medicine, University of Toronto, Toronto, Canada.

leagues demonstrated that macaques immunized with humancells alone exhibited sterilizing immunity to intravenous chal-lenge with SIV grown in the same human cells.7 This effectwas strongly correlated with xenogeneic antibody responsesagainst cellular proteins, an observation confirmed by subse-quent studies.8 Furthermore, the elicitation of sterile immunityagainst SIV challenge could be replicated by xenogeneic im-munization with either purified human Class I9 or Class II10

HLA protein. It should be noted, however, that these were allprimate studies of xenogeneic not allogeneic anti-HLA im-mune responses and if this approach were to be utilized as avaccine strategy, protective allogeneic immune responsesagainst virus transmitted from human to human would be re-quired.

In humans, we have previously reported that allogeneic im-mune responses against HLA determinants may account for re-duced susceptibility to HIV-1 infection in some individuals. Ina prospective study of perinatal HIV-1 transmission, Class IHLA discordance between HIV-1-infected mothers and theirchildren was independently associated with a highly significantreduction in perinatal HIV-1 transmission.11 We hypothesizethat the functional basis for this protection from infection couldbe related to infant allogeneic responses against maternal HLAdeterminants expressed on incoming viruses or infected mater-nal cells. Despite this supporting epidemiological evidence, un-til recently no suitable opportunity has been available to directlystudy the virus-neutralizing activity of allogeneic anti-HLA im-mune responses in humans.

If allogeneic anti-HLA immune responses can be demon-strated to neutralize HIV-1, it has been proposed that alloimmu-nization against the human HLA determinants could form a pre-ventative vaccine approach for HIV-1.12,13 As HLA have bothwell-defined and stable polymorphisms, we reason that such astrategy might prove efficacious in contrast to conventional ap-proaches to vaccine design based on virus-derived immunogens.Indeed, because HLA antigens are under distinct genetic controlsand stable evolutionary selective pressures, it is likely that thevirus would be incapable of completely escaping from immuneresponses directed at these host-derived proteins. Accordingly,with the HLA antigens present on the surface of HIV-1 in con-centrations comparable to the env gene-product gp120,4 onecould foresee vaccinating against several alleles at once, whichcould yield protection against a large majority of HIV-1 isolatesexpressing one or more of those alleles.

The unusual instance of leukocyte immunotherapy of womenfor recurrent spontaneous abortion14 provides a rare opportu-nity to study in a controlled way the development of anti-HLAimmune responses against specific determinants and their ca-pacity to neutralize in vitro HIV-1 bearing the same allogeneicHLA. We prospectively followed a small group of women re-ceiving a single well-defined allogeneic immunization in theform of leukocyte immunotherapy for recurrent spontaneousabortion. This regimen provided us with the unique opportu-nity to assess the magnitude, specificity, and kinetics of theserological anti-HLA immune response to systemic alloimmu-nization. In this paper we report on the induction of humoralresponses against cellular immunogens with defined HLAspecificities and demonstrate that IgG antibodies from such analloimmunized individual can neutralize HIV-1 in vitro, in-cluding a primary isolate.

MATERIALS AND METHODS

Patients and samples

Female patients (recipients) and their male partners (donors)were recruited to the study through the Reproductive Immu-nology clinic at McMaster Medical Centre, McMaster Univer-sity, Hamilton, Ontario. Recipients had been scheduled to re-ceive leukocyte immunotherapy for the treatment of recurrentspontaneous abortion (RSA)14 and consented to provide bloodsamples pre- and postimmunization to study the induction ofalloreactive responses and the ability of such responses to neu-tralize HIV in vitro. Although the primary end point for RSAis pregnancy, we recognized that the immunization of recipi-ents with donor-derived leukocytes represents a controlledmodel of allovaccination.

At the initial screening visit, all study participants were testedfor serological evidence of HIV-1 and -2, T cell leukemia virus(HTLV), hepatitis B and C, VDRL, and cytomegalovirus(CMV) infection. Female recipients were also tested for liverand renal function, rheumatoid factor, antinuclear antibodies,anticardiolipin, antileukocyte antibodies, and antierythrocyteantibodies. All patients provided informed consent prior to en-rollment in the study. Female recipients satisfying the abovecriteria were immunized with a total of 4 3 108 irradiated den-sity purified peripheral blood mononuclear cells (PBMC) iso-lated from their partners’ blood. This immunization was per-formed at a single time point with a single intravenous injectionand three intradermal injections of 108 cells. Blood sampleswere obtained from study participants at monthly visits start-ing at the immunization visit and continuing up to 3 monthspostimmunization. Approval for this study was provided by theinstitutional review boards at McMaster University and the Uni-versity of Toronto.

Human leukocyte antigen typing

To determine the relative HLA discordance between eachdonor and recipient pair, HLA typing was conducted in all studypatients using DNA isolated from blood samples (Puregene,Gentra, Minneapolis, MN) provided at the initial screening visit.The complete molecular HLA-typing of Class I (A, B, and C)and Class II (DR and DQ) alleles for each individual was con-ducted using sequence-specific primer (SSP) PCR amplifica-tion (Pel-Freez, Milwaukee, WI).

Anti-HLA antibody detection

Anti-HLA Class I and II antibodies were assessed usingFlowPRA beads (OneLambda, CA) according to the manufac-turer’s instructions. Briefly, each FlowPRA bead preparationcontains 20 different beads precoated with either Class I (Cat.#FL1-30) or Class II HLA (Cat. #FL2-30) antigens derived fromB lymphoblastoid cell lines. Pooled in this manner, the Flow-PRA bead preparation represents all major HLA Class I or IIantigens. These pooled beads are then stained with each plasmasample. Immunoglobulin G antibody specific for HLA Class Ior II antigens detected by binding to FlowPRA beads was detected by staining with a fluorescein-conjugated polyclonalmouse antihuman IgG followed by flow-cytometric analysisconducted using a FACSCalibur instrument (BD Biosciences,

LEITH ET AL.958

San Jose CA). Data were analyzed using Cell Quest (BD Bio-sciences, San Jose, CA) and FlowJo (Treestar, San Carlos, CA)software. To identify the allele specificity of anti-HLA anti-bodies detected by FlowPRA beads, recipient plasma sampleswere also tested by PRAStat enzyme-linked immunosorbent as-say (ELISA) (Sangstat, Fremont, CA). The PRAStat ELISAsystem utilizes defined HLA antigens immobilized on individ-ual wells of a 96-well plate. Approximately 10 different definedClass I and II HLA determinants were present in each well,with common determinants present in multiple wells. Plasmasamples were diluted 1:100 and then aliquotted into 96-wellplates coated with HLA antigens of defined specificity. Anti-HLA antibody specificity was determined by comparing the pat-tern of reactivity obtained by ELISA with the HLA antigenspresent in each well as defined by the manufacturer. Data wereanalyzed by proprietary SoftSTAT software. PRAStat ELISAdata were complemented by FlowPRA specific tests that usedan approach similar to FlowPRA beads except defined HLA-coated beads are stained with specific amounts of an FL-2-de-tectable fluorochrome (OneLambda, CA). Antibodies were in-cubated with pooled beads representing most known HLA ClassI (Cat. #FL1SP) or Class II (Cat. #FL2SP) antigens and thendetected with a mouse-antihuman IgG fluorescein isothio-cyanate (FITC) and analyzed using Cell Quest software on aFACSCalibur flow cytometer.

Control antibodies

An anti-HIV-1 polyclonal antiserum (Serum 2, Cat. #1983,NIH AIDS Research and Reference Reagent Program) reportedto neutralize HIV-1IIIB at .1:380 dilution was used in dilutionsfrom 1:60 to 1:3840. Monoclonal anti-HLA antibodies withspecificity for framework determinants of HLA Class I (W6/32,ATCC #HB-95) or Class II (44H10, kindly provided by Dr.Michelle Letarte) were used as controls in neutralization assays.In addition, as negative controls, whole immunoglobulin G(IgG) samples from each respective male donor was purifiedfrom plasma samples in the same manner as the recipient sam-ples described below. All donor samples were devoid of reac-tivity for HLA antigens when measured using the FlowPRAbeads (data not shown).

Recipient IgG samples

To ensure that any anti-HIV-1 activity measured in the sam-ples assayed could be attributed to antibodies generated in therecipient, we prepared whole IgG from each plasma sampleprior to inclusion in the neutralization assays. Briefly, wholeIgG was purified from 750-ml plasma samples by individual 1 ml ImmunopurePlus Protein G columns (Pierce Scientific,Rockford, IL), each of which was capable of binding up to 20 mg of IgG. Individual columns were equilibrated with a pro-prietary neutral pH binding buffer. Each sample was diluted 1:1in binding buffer and applied to the column. Each column wasthen washed three times in binding buffer and the IgG elutedby application of 4 ml elution buffer, pH 2.3. After elution eachpurified IgG sample was immediately neutralized with 103PBS, pH 7.0. Dialysis bags were prepared by boiling twice for15 min in distilled water and then equilibrated in PBS for 1 hr.Samples were then dialyzed over 48 hr with one change of PBS.Finally, each sample was concentrated to the original sample

volume of 750 ml (,8 mg/ml), tested for purity by sodium do-decyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE),and then retested for anti-HLA antibodies by FlowPRA (in com-parison to source plasma samples; data not shown) prior to in-clusion in the neutralization assays.

Virus stocks

Donor-passaged stocks of HIV-1 were prepared by growingprimary isolate HIV-1301660 (from the NIH AIDS Research andReference Reagent Program) or lab-adapted HIV-1IIIB in anti-CD3/anti-CD28-stimulated donor-derived CD41 T lympho-cytes. The viruses are referenced according to the stock strainand donor identifier: from donor 1 (301660/d1, IIIB/d1) anddonor 2 (IIIB/d2), respectively. All stocks were aliquotted andstored at 2152°C until use.

Virus neutralization assays

The ability of IgG purified from postimmunization recipientplasma samples to neutralize HIV-1 in vitro was assessed us-ing a modified version of the multinuclear activation of galac-tosidase indicator (MAGI) cell line (cMAGI) (NIH AIDS Research and Reference Reagent Program).15,16 MAGI cells ex-press CXCR4 and are stably transfected with CD4 and CCR5,and the HIV-1 LTR driving b-galactosidase expression. Con-trol monoclonal antibodies against HLA Class I (W6/32) andHLA Class II (44H10) were used in concentrations rangingfrom 200 to 0.002 mg/ml. Control anti-HVI-1 polyclonal anti-serum was tested in serial dilutions from 1:60 to 1:3840. Puri-fied whole IgG samples were tested in duplicate in 4-fold ser-ial dilutions from 1:6 to 1:384. Briefly, cMAGI target cellscultured in Dulbecco’s modified Eagle’s medium (DMEM) plus10% fetal bovine serum were trypsinized, washed, and thenaliquoted into 96-well flat-bottom plates (5000 cells/well) andallowed to adhere overnight at 37°C. The following day, donor-passaged HIV-1 was preincubated with antibody test samplesand controls for 1 hr at room temperature and then applied tothe monolayer of cMAGI cells. Following at 48-hr incubation,the antibody/virus-containing medium was removed from thecMAGI cells. The monolayers were then washed three times inwarm PBS and recultured in 100 ml of complete DMEM. Fol-lowing a further 48-hr incubation, medium samples were re-moved, diluted 10–1000 fold, and tested for p24 concentration(Organon Technika, Durham, NC) according to the manufac-turer’s instructions.

RESULTS

HLA class I and II discordance between donors and recipients

Although unlikely in an outbred population, we wanted tobe sure that for each donor to recipient immunization, thePBMC immunogen was truly allogeneic. Therefore all individ-uals entered into the study were HLA-typed at the HLA ClassI and II loci using sequence-specific primer (SSP)-PCR (Pel-Freez Clinical Systems, Brown Deer, WI). As shown in Table1, all donors were discordant for at least two HLA Class I alleles and on HLA Class II allele from their respective recipients.

HIV-1 NEUTRALIZATION BY ALLOIMMUNE ANTI-HLA SERA 959

Detection of anti-HLA antibodies in recipient plasma

Since all donor–recipient pairs had discordant HLA alleles,we then examined the recipient humoral immune responses tothe PBMC immunization regimen. We first assessed the rela-tive strength of the anti-Class I and II HLA antibodies presentin plasma in pre- and postimmunization visits using FlowPRAbeads. As seen in Figure 1a and b, prior to immunization, allrecipient women had low to undetectable anti-HLA Class I andII antibodies. In recipients 1 and 2, postimmunization anti-ClassI responses were significantly boosted (Fig. 1a). In both indi-viduals, these responses were sustained for at least 3 monthspostimmunization. The peak responses (12 and 4 weeks, re-

spectively) gave mean fluorescence intensities comparable tomultiparous serum containing anti-Class I antibodies (data notshown). Interestingly, only one of seven recipients (1) madeanti-Class II antibody responses against the whole-cell im-munogen. These responses peaked 12 weeks postimmunization(Fig. 1b).

Specificity of anti-HLA antibody response

To determine if the detected antibody responses were spe-cific to the HLA determinants expressed on the whole-cell im-munogen (rather than an autoimmune response against recipi-ent self-HLA determinants), we analyzed recipient plasma

LEITH ET AL.960

TABLE 1. HLA ALLELIC DISCORDANCE BETWEEN VACCINE DONORS AND RECIPIENTS PAIRS

Donor/ Discordant Discordantrecipient HLA Class I allelesa HLA Class II alleles

1 A*3101, A*2402, B*4001 DRB1*04011, DRB1*15011

2 A*0201, A*2402, B*1524, B*1501, DRB1*15011Cw*0501

3 A*0201, B*1801, B*2701, NDCw*0202, Cw*1203

4 A*1101, B*4402, B*5501, DRB1*04011, DRB1*1401,Cw*0303, Cw*0501 DQB1*03011

5 A*0201, A*1101, B*4701, DRB1*0701, DQB1*0201Cw*0602

6 A*2601, A*3101, B*3501, DRB1*04011, DQB1*0201, DQB1*0302Cw*0701, Cw*0401

7 A*3101, A*0201, B*5101, DQB1*0304Cw*1502

aAllelic discordance is shown as those HLA alleles present in the donor but not the recipient genotype.

FIG. 1. Anti-HLA antibodies elicited by allovaccination regimen. Recipient plasma samples were tested for the presence ofanti-HLA Class I (a) and Class II (b) antibodies using beads coated with HLA Class I and II molecules, respectively. Reactiv-ity of antibodies to the beads was detected by an antihuman IgG-FITC and is shown as mean fluorescence intensity (MFI). Sam-ples were tested neat at least twice. Representative data are shown.

samples using two separate assays capable of identifying thefine specificity of anti-HLA antibodies. The results for both as-says are summarized in Table 2 and the relevant HLA allelespresent in the donor, but not recipient haplotypes are includedfor comparison. Using the PRAStat ELISA system, which mea-sures only the specificity of anti-Class I antibodies, we detectedanti-HLA-A19 and anti-HLA-B62 antibodies in postimmu-nization samples taken from recipients 1 and 2, respectively.These anti-HLA Class I antibody responses were confirmed us-ing the FlowPRA specific assay, which measures specificity ofboth anti-HLA Class I and II antibodies. In addition, weak an-tibody reactivity against HLA-A24 detected in recipient 1 wasdetected, which was not observed using the PRAStat ELISAsystem. Experiments conducted using the FlowPRA specifictest also detected a strong monomorphic anti-HLA Class II re-sponse against DR4 in recipient 1. In all cases, both anti-HLAClass I and anti-HLA Class II antibody responses were specificto HLA antigens present in the donor haplotype, but not in thatof the recipient. Thus, the antibody response elicited by whole-cell alloimmunization appears to be specific to immunizingHLA determinants.

Neutralization of HIV-1

To determine if alloimmunization could elicit protective im-munity against HIV-1, we focused our subsequent experimentson the recipients who responded to the whole-cell vaccinationby producing anti-HLA Class I and/or anti HLA Class II anti-bodies (recipients 1 and 2). The ability of these antibodies toneutralize HIV-1 grown in the respective (donors 1 and 2)donor-partners’ CD41 T lymphocytes was assessed in com-parison to monoclonal antibodies W6/32 (anti-HLA Class I) and44H10 (anti-HLA Class II). Both monoclonal antibodies werefound to be capable of neutralizing donor 1- and 2-derivedviruses. With respect to recipient derived samples, whole IgGpurified from plasma samples was preincubated with donor-de-rived virus prior to addition to the cMAGI target cell line. Asshown in Figure 2a, 2-week, and to a lesser extent 4-week post-immunization samples from recipient 1 were able to neutralizedonor 1-derived IIIB/d1. Furthermore, these same samples con-taining both anti-HLA Class I and anti HLA Class II antibod-

ies were also able to neutralize primary isolate 301660/d1 (Fig.2d). This neutralization, maximal at a 1:6 dilution of purifiedIgG, was sustained to 1:24 dilutions for both laboratory(IIIB/d1) and primary isolates (301660/d1). In contrast, post-immunization IgG samples from recipient 2 known to containantidonor anti-HLA Class I antibodies but not anti-HLA ClassII antibodies did not neutralize donor 2-derived laboratory iso-late IIIB/d2 in vitro (Fig. 3a), insufficient donor CD41 T lym-phocytes were available from donor 2 to grow and test addi-tional primary isolates.

DISCUSSION

In a fortunate opportunity to study alloimmunization inwomen undergoing leukocyte immunotherapy where a varietyof recipient and donor samples could be obtained, we have dem-onstrated that following a single immunization with allogeneicPBMC, two of seven vaccine recipients mounted a strong anti-HLA antibody response. These responses were specific to theHLA type of the donor, did not induce autoantibodies againstHLA, and were detectable at 4 weeks postimmunization andsustained at up to 12 weeks postimmunization. Further, in oneof two responding patients, recipient IgG containing both anti-HLA Class I and II antibody was observed to have neutraliza-tion capability against allogeneic (donor-derived) HIV-1 invitro. This neutralizing effect was demonstrable against bothlaboratory strain HIV-1IIIB and against primary CCR5-utilizingisolate HIV-1301660. This result underscores a fundamental ad-vantage of an immunogen based on host-encoded determinantssuch as HLA. The Achilles heel of HIV-1 vaccines that rely onantibody-mediated neutralization is the tremendous variation invirus-encoded proteins and to date, these vaccines have notyielded responses capable of inducing cross-neutralizing anti-body against primary isolates of HIV-1.17 In contrast, the abil-ity to neutralize primary isolates of HIV-1 independent of thevariation in their envelope protein would overcome a prime hur-dle in HIV-1 vaccine design. Thus, a vaccine target that is poly-morphic and thus alloimmunogenic, that is naturally inducedsafely in a number of settings such as pregnancy, and is rela-

HIV-1 NEUTRALIZATION BY ALLOIMMUNE ANTI-HLA SERA 961

TABLE 2. SPECIFICITY OF ANTI-HLA ANTIBODIES

Recipient PRAStat b FlowPRA c FlowPRA Donorantibody Visit a HLA Class I HLA Class I HLA Class II HLAd

Recipient 1 12 weeks A19 A24 and A19 DR4 A19(A*3101)

A24(A*2402-07)

DR4(DRB*04011)

Recipient 2 4 weeks B62 B62 N/A B62 (B*1501)

aRefers to weeks postimmunization—peak responses were tested.bSpecificity tested using PRAStat ELISA system.cSpecificity tested using FlowPRA Class I or Class II specific beads.dRelevant donor HLA alleles shown as serological typing with molecular typing determined by SSP PCR typing shown in

parentheses.

tively stable in a population over time such as HLA, providesa rationale to further investigate its efficacy and safety as animmunogen.

To the best of our knowledge, this provides the first demon-stration that deliberate human alloimmunization (rather thanxenoimmunization) can induce antibodies capable of neutraliz-ing HIV-1. Thus these data are conceptually consistent withprevious reports demonstrating that xenovaccination with eitherhuman cells or purified HLA proteins can elicit protective im-munity in rhesus macaques against challenge with SIV grownin human cells.4,8–10 In most cases the macaque protective ef-fect was strongly correlated with the presence of anti-HLA an-tibodies, in particular those with specificity to alleles of theHLA Class II locus. Our data likewise show a correlation ofanti-HLA-Class II antibodies since we show that IgG contain-ing anti-HLA Class I and II, or IgG containing antibody againsta single HLA Class II determinant (DR4) can both neutralizeHIV-1. Taken together with previous reports demonstrating thatexpression of HLA Class II on the virus is required for infec-tivity of HIV-1,18,19 these data suggest that antibodies binding

to HLA Class II antigens may inhibit or antagonize importantevents requisite to efficient virus–cell fusion. Additionally, itmay be the result of the density of expression of HLA Class IImolecules on the viral envelope since the virus cannot easilydevelop mechanisms to down-regulate HLA Class II expres-sion on its surface in contrast to Nef-mediated down-regulationof HLA Class I expression on chronically HIV-1-infectedcells.20 However, as our in vitro measures of neutralization donot measure additional mechanisms of virus control such ascomplement deposition or opsonization, it is also possible thatonly antibodies capable of steric hindrance would have ademonstrable neutralizing effect in this assay. Accordingly, itremains possible that anti-HLA Class I antibodies induced bythe vaccination regimen might afford a protective effect in vivoin the presence of complement or other leukocyte effector sub-sets or may well have neutralizing activity against virions gen-erated from acutely infected cells that have not yet down-reg-ulated their class I HLA.

Another potential contributing determinant to the neutraliz-ing effect observed in our study is the relative breadth of the

LEITH ET AL.962

FIG. 2. Neutralizing capacity of antibodies elicited in recipient 1. Laboratory strain HIV-1IIIB (a–c) and primary isolate HIV-1301660 (d–f) grown in donor-derived CD41 T lymphocytes were tested for their respective ability to infect cMAGI target cellsin the presence of (a and d) purified IgG (,8 mg/ml) from allovaccine recipients and autologous donors, (b and e) polyclonalantiserum against HIV-1, and (c and f) monoclonal antibodies W6/32 (anti-HLA Class I [closed squares]) and 44H10 (anti-HLAClass II [closed diamonds]). Anti-HLA-DR4 polyclonal antiserum (open circles) and recipient 1 IgG from week 12 (open dia-monds) were tested against a low titer passage of primary isolate HIV-1301660 (g). Each experiment was conducted two or threetimes giving the same result. Representative data are shown as means of duplicates.

anti-HLA Class I and II response made by recipient 1 (Table2). Unlike the nonneutralizing monomorphic response againstthe Class I determinant HLA-B62 made by recipient 2, recipi-ent 1 made broad antibody responses against HLA-A24, A19,and DR4. This latter response may more similarly reflect thebroad anti-HLA reactivity and neutralization (in this system)exhibited by the framework monoclonal antibodies W6/32 and44H10 (Figs. 2c and 2f and 3c). Since there are examples ofnaturally induced broadly reactive and monomorphic anti-HLAantibodies in multiparous women and in transfusion recipients,a systemic study of their neutralizing capacity against a varietyof HIV-1 isolates grown in HLA-defined cells might furtherelucidate this issue. In support of this was our demonstrationof neutralizing activity of IgG from multiparous serum con-taining anti-DR4 antibodies against our DR4-expressing donor1-derived virus. Unfortunately, instances of leukocyte im-munotherapy amenable to detailed study are sufficiently rarethat it is unlikely that this setting could be employed for a large-scale systemic evaluation of anti-HLA antibody with respect tobreadth and specificity.

If breadth and specificity (including the importance of anti-HLA Class II responses) appear to be contributors to the neu-tralizing effect, then both of these factors are dependent on thedegree of mismatching between the recipient and donor HLA

haplotypes and on the relative expression of HLA Class I andII molecules on the donor whole-cell immunogen. Thus, theability of a recipient to respond to donor-derived HLA deter-minants is a function of the allelic polymorphism among indi-viduals at the HLA loci, responsiveness. As a result, it is in-herently limited to the non-self-determinants recognized by therecipient immune system (see Table 1). Accordingly, we werenot surprised to see such high variability in this model of im-munization, which involved only a single immunization. Futurestudies of alloimmunization would likely maximize the breadthof alloresponsiveness by administering defined HLA antigensin the form of either conventional adjuvanted recombinant pro-tein or viral-vector/naked DNA immunogens encoding multi-ple HLA genes and involving booster doses. We have previ-ously demonstrated that DNA immunization of macaques canbe an efficient method for the generation of anti-HLA anti-bodies without breaking tolerance.21 Because of the relativeconservation of certain common HLA alleles in human popu-lations, such a vector-based approach could be used to createcocktails of immunogens and would not depend on HLA typ-ing individuals since one could approach complete populationcoverage using only five or six different HLA alleles.22

It is also important to note that alloimmunization may pro-voke protective mechanisms distinct from the elicitation of anti-

HIV-1 NEUTRALIZATION BY ALLOIMMUNE ANTI-HLA SERA 963

FIG. 2. (continued)

HLA antibodies. These mechanisms may involve adaptive immune responses against other host-encoded determinantspresent on the viral envelope or involved in viral binding, orinnate immune responses. There is no evidence to date that thistype of immunization breaks tolerance against largely non-polymorphic determinants involved in binding such as CD4,CCR5, or other coreceptors in normal hosts but these were notmeasured directly in this study. With respect to the potentialadditional benefit of innate immune responses in the setting ofalloimmunization, a recent study by Wang et al.23 demonstratedthat leukocyte immunotherapy caused down-regulation ofCCR5 expression on CD41 T lymphocytes and increased se-cretion of CD8-derived anti-HIV factors. Those authors did notmeasure antibody responses against the donor HLA, but takentogether with our observations, these data highlight the fact thatinnate immune responses elicited by alloimmunization inde-pendent of anti-HLA antibodies may also contribute to protec-tion in vivo. Previous studies of xenovaccination in macaquesdemonstrated the critical importance of anti-HLA antibodies inprotection from intravenous challenge with pathogenic SIV.This study now provides the first evidence that human vacci-

nation with allogeneic PBMC can elicit allele-specific anti-HLA antibody, which represent the most likely mediator of theIgG neutralizing activity against both laboratory and primaryisolates of HIV-1. In particular, alloantibodies with anti-HLAClass II specificity appear to be the dominant contributor to theobserved in vitro HIV-1-neutralizing effect. Approaches to ad-minister HLA antigens using conventional vaccination modal-ities may prove useful as alternatives or adjuncts to HIV anti-gen-specific prophylactic vaccine regimens.

ACKNOWLEDGMENTS

J.G.L. is a recipient of a Canadian Institutes of Health Re-search graduate studentship. K.S.M. is the recipient of a careerscientist award from the Ontario HIV Treatment Network. Thiswork was supported by the Positive Action Fund, Ministry ofHealth, Ontario. We gratefully acknowledge the expert techni-cal assistance of Donna Davison (Trimeris, Inc.) and JustinManual (McMaster Medical Centre).

LEITH ET AL.964

FIG. 3. Neutralizing capacity of antibodies elicited in recipient 2. HIV-1IIIB grown in donor-derived CD41 T lymphocytes wastested for its ability to infect cMAGI target cells in the presence of (a) purified IgG (,8 mg/ml) from allovaccine recipients andautologous donors, (b) polyclonal antiserum against HIV-1, and (c) monoclonal antibodies W6/32 (anti-HLA Class I) and 44H10(anti-HLA Class II). Each experiment was conducted two or three times giving the same result. Representative data are shownas means of duplicates.

REFERENCES

1. Hoxie JA, Fitzharris TP, Youngbar PR, Matthews DM, RackowskiJL, and Radka SF: Nonrandom association of cellular antigens withHTLV-III virions. Hum Immunol 1987;18(1):39–52.

2. Castilletti C, Capobianchi MR, Fais S, Abbate I, Ficociello B,Ameglio F, Cordiali Fei P, Santini SM, and Dianzani F: HIV type1 grown on interferon gamma-treated U937 cells shows selectiveincrease in virion-associated intercellular adhesion molecule 1 andHLA-DR and enhanced infectivity for CD4-negative cells. AIDSRes Hum Retroviruses 1995;11(5):547–553.

3. Azocar J and Essex M: Incorporation of HIA antigens into the en-velope of RNA tumor viruses grown in human cells. Cancer Res1979;39(9):3388–3391.

4. Arthur LO, Bess JW Jr, Sowder RC 2nd, Benveniste RE, MannDL, Chermann JC, and Henderson LE: Cellular proteins bound toimmunodeficiency viruses: Implications for pathogenesis and vac-cines [see comments]. Science 1992;258(5090):1935–1938.

5. Henderson LE, Sowder R, Copeland TD, Oroszlan S, Arthur LO,Robey WG, and Fischinger PJ: Direct identification of class II his-tocompatibility DR proteins in preparations of human T-cell lym-photropic virus type III. J Virol 1987;61:(2):629–632.

6. Tremblay M, Fortin J-F, and Cantin R: The acquisition of host-encoded proteins by nascent HIV-1. Immunol Today 1998;19:346–351.

7. Stott EJ, Chan WL, Mills KH, Page M, Taffs F, Cranage M, Green-away P, and Kitchin P: Preliminary report: Protection of cynomol-gus macaques against simian immunodeficiency virus by fixed infected-cell vaccine [see comments]. Lancet 1990;336(8730):1538–1541.

8. Stott EJ: Anti-cell antibody in macaques [letter] [see comments].Nature 1991;353(6343):393.

9. Chan WL, Rodgers A, Grief C, Almond N, Ellis S, Flanagan B,Silvera P, Bootman J, Stott J, Kent K, et al.: Immunization withclass I human histocompatibility leukocyte antigen can protect macaques against challenge infection with SIVmac-32H. AIDS1995;9(3):223–228.

10. Arthur LO, Bess JW Jr, Urban RG, Stronringer JL, Morton WR,Mann DL, Henderson LE, and Benveniste RE: Macaques immu-nized with HLA-DR are protected from challenge with simian im-munodeficiency virus. J Virol 1995;69(5):3117–3124.

11. MacDonald KS, Embree J, Njenga S, Nagelkerke NJ, Ngatia I, Mo-hammed Z, Barber BH, Ndinya-Achola J, Bwayo J, and PlummerFA: Mother-child class 1 HLA concordance increases perinatal human immunodeficiency virus type 1 transmission. J Infect Dis1998;177(3):551–556.

12. Shearer GM, Clerici M, and Dalgleish A: Alloimmunization as anAIDS vaccine? Science 1993;262(5131):161–162.

13. Kiprov DD, Sheppard HW, and Hanson CV: Alloimmunization toprevent AIDS? Science 1994;263(5148):737–738.

14. Coulam CB, Clark DA, Beer AE, Kutteh WH, Silver R, Kwak J, andStephenson M: Current clinical options for diagnosis and treatmentof recurrent spontaneous abortion. Clinical Guidelines Recommen-dation Committee for Diagnosis and Treatment of Recurrent Spon-taneous Abortion. Am J Reprod Immunol 1997;38(2):57–74.

15. Chackerian B, Long EM, Luciw PA, and Overbaugh J: Human im-munodeficiency virus type 1 coreceptors participate in postentrystages in the virus replication cycle and function in simian immu-nodeficiency virus infection. J Virol 1997;71(5):3932–3939.

16. Kimpton J and Emennan M: Detection of replication-competentand pseudotyped human immunodeficiency virus with a sensitivecell line on the basis of activation of an integrated beta-galactosi-dase gene. J Virol 1992;66(4):2232–2239.

17. Sattentau QJ: Neutralization of HIV-1 by antibody. Curr Opin Im-munol 1996;8(4):540–545.

18. Cantin R, Fortin JF, Lamontagne G, and Tremblay M: The pres-ence of host-derived HLA-DR1 on human immunodeficiency virustype 1 increases viral infectivity. J Virol 1997;71(3):1922–1930.

19. Cantin R, Fortin JF, Lamontagne G, and Tremblay M: The acqui-sition of host-derived major histocompatibility complex class IIglycoproteins by human immunodeficiency virus type 1 acceler-ates the process of virus entry and infection in human T-lymphoidcells. Blood 1997;90(3):1091–1100.

20. Collins KL, Chen BK, Kalams SA, Walker BD, and Baltimore D:HIV-1 Nef protein protects infected primary cells against killingby cytotoxic T lymphocytes. Nature 1998;391(6665):397–401.

21. Dela Cruz CS, MacDonald KS, and Barber BH: Anti-major histo-compatibility complex antibody responses in macaques via intra-dermal DNA immunizations. Vaccine 2000;18(27):3152–3165.

22. Gulukota K and DeLisi C: HLA allele selection for designing pep-tide vaccines. Genet Anal 1996;13(3):81–86.

23. Wang Y, Tao L, Mitchell E, Bravery C, Berlingieri P, ArmstrongP, Vaughan R, Underwood J, and Lehner T: Allo-immunizationelicits CD81 T cell-derived chemokines, HIV suppressor factorsand resistance to HIV infection in women. Nat Med 1999;5(9):1004–1009.

Address reprint requests to:Kelly S. MacDonald

Department of MicrobiologyRoom 1484

Mount Sinai Hospital600 University Avenue

Toronto, Ontario, Canada M5G 1X5

E-mail: kmacdonald@mtsinai.on.ca

HIV-1 NEUTRALIZATION BY ALLOIMMUNE ANTI-HLA SERA 965