Repeated exposure to human immunodeficiency virus (HIV) does not always result in seroconversion. Understanding the conditions that permit or protect against progressive infection with HIV is important for vaccine development. Nineteen subjects at risk for HIV infection were CCR-5 genotyped and screened for virus-specific memory cytotoxic T lymphocytes (CTL). None had the Δ32CCR-5/Δ32CCR-5 genotype associated with HIV resistance. HIV-specific CTL were detected in 7 (41.1%) of 17 exposed uninfected subjects versus 0 of 14 seronegative subjects with no HIV risk factors (P = .006 χ2 test). Recognition of virus by CTL in exposed uninfected subjects was major histocompatibility complex class I-restricted and multispecific, and specificity could change with time. Activity could persist up to 34 months after the last virus exposure. The presence of HIV-specific CTL in a greater proportion of seronegative HIV-exposed versus unexposed subjects supports the notion that in some cases, virus exposure induces HIV immunity without seroconversion or disease progression.
As observed for other human pathogens, individuals differ in their susceptibility to and outcome after infection with the human immunodeficiency virus (HIV). Some persons appear to be resistant to this virus. Elucidation of the mechanisms underlying their resistance has implications for the development of effective prophylactic vaccines for HIV.
Resistance to HIV infection in exposed uninfected persons (EUs) has been observed in sex partners of HIV-infected persons [1–8], infants born to HIV-infected mothers [9–12], health care workers occupationally exposed to HIV-contaminated body fluids [13, 14], and commercial sex workers [15, 16]. Resistance may result from either inherited or acquired factors. Homozygosity for a 32-bp deletion mutation at the CCR-5 locus (Δ32CCR-5) is a genetically acquired HIV resistance factor [17–21]. The wild type allele at this locus encodes a functional coreceptor for entry of macrophage-tropic HIV into CD4 cells [22–27]. The Δ32CCR-5 protein is not expressed at the cell surface [17, 18]. People homozygous for the deleted allele are substantially, although not absolutely, resistant to HIV infection in vivo [17–19, 28, 29]. About 1% of Caucasians are homozygous for Δ32CCR-5 [17–19, 21]. Studies in highly exposed seronegative Caucasian cohorts suggest that the Δ32CCR-5/Δ32CCR-5 genotype accounts for some but not all instances of HIV resistance despite high levels of exposure [5, 18, 19].
Acquired immunity may be another mechanism that plays a role in resistance to HIV infection. T helper type 1 (Th1) responses to HIV, such as the secretion of interleukin-2 (IL-2) after stimulation with HIV peptides, have been observed more frequently in EUs than in those at low risk for HIV exposure [1, 3, 5, 10, 13, 14]. HIV-specific IgA antibody in the mucosa has also been observed in a larger proportion of EUs than in those at low risk for infection with HIV . HIV-specific cytotoxic T lymphocytes (CTL) [4, 9, 11, 12, 14, 15] and suppression of HIV replication by CD8 T cells [30, 31] are other manifestations of immunity to HIV reported in some EUs.
We report here on a cohort of 19 EUs, 14 of whom have been in stable relationships with HIV-infected sex partners.
Nineteen HIV EUs were studied. Characteristics of the EU study population, including the length of the relationship of each EU with the seropositive partner and the last date of exposure, when applicable, are shown in table 1. Documentation of HIV seronegativity was done by ELISA and confirmed by Western blotting. For EUs for whom a last date of exposure to HIV was known, serologic HIV testing was done at least within 1 month of the last exposure date as well as ∼3 months after the first negative serologic HIV test. For EUs with no known last date of exposure (either because exposure is ongoing or because they are gay men not involved in stable relationships with partners of known HIV status), HIV testing was done at least within 1 month of the time at which samples for CTL analysis were obtained as well as ∼3 months after this first negative serologic HIV test. All serologic tests done on EUs were negative. After the last CTL analysis reported here, cell samples were obtained from EUs to screen for HIV-1 DNA proviral sequences by a qualitative polymerase chain reaction (PCR) assay (Amplicor, HIV-1 PCR-EIA; Roche Diagnostics, Mississauga, Canada). In no case was it possible to amplify HIV proviral sequences from the peripheral blood mononuclear cells (PBMC) of any of the EUs in this cohort. Fourteen seronegative persons with no risk factors for HIV infection were also studied as controls. Control subjects were genotyped for CCR-5 and screened for HIV-specific CTL activity. Eleven of these controls were screened for HIV-specific CTL activity on three occasions, 1 was tested on two occasions, and 2 were screened on one occasion.
Genomic DNA was prepared from fresh whole blood, fresh or cryopreserved PBMC, or Epstein-Barr virus (EBV)—transformed B cell lines. Fresh or thawed frozen samples were pelleted by centrifugation at 400 g for 10 min. Cell pellets were osmotically lysed by sucrose—Triton X-100 buffer (0.32 M sucrose, 10 mM Tris, pH 7.5, 5 mM MgCl2, 1% Triton X-100). Nuclei were pelleted by centrifugation and then lysed by incubation at 56°C for 60 min in a buffer containing 50 mM KCl, 10 mM Tris, pH 8.3, 2.5 mM MgCl2, 0.5% Tween 20, 0.1 mg/mL gelatin, and 200 mg/mL proteinase K. After this incubation, proteinase K was inactivated at 95°C for 10 min. Genomic DNA samples were centrifuged at 13,000 g for 1 min to clear supernatant of debris and were stored frozen at −20°C.
CCR-5 genotyping was done by a modified version of the method described by Samson et al. . Briefly, 5′-CCTGGCTGTCGTC-CATGCTG-3′ and 5′-CTGATCTAGAGCCATGTGCACAACT-CT-3′ forward and reverse oligonucleotide primers, respectively, were used to amplify a region of genomic DNA that spanned the 32-bp deletion differentiating the Δ32CCR-5 allele from its wild type counterpart at the CCR-5 locus. The PCR reaction was carried out in 30 μL of 10 mM Tris-HCl, pH 8.0, 50 mM KCl, 0.75 mM MgCl2, 0.2 mM each dATP, dCTP, dTTP, and dGTP, 0.01% gelatin, 5% dimethyl sulfoxide, 200 ng of DNA, 60 ng of each oligonucleotide primer, and 1.5 U of Taq DNA polymerase. The PCR conditions were 93°C for 2 min 30 s, followed by 30 cycles of 93°C for 1 min, 60°C for 1 min, and 72°C for 1 min and a 6-min extension at 72°C. The resulting product was digested with 10 U of EcoRI at 37°C for 60 min. Digested amplification products were loaded onto a 2% agarose gel and electrophoresed for 1 h at 90 V Amplified DNA bands were visualized by ethidium bromide staining and sized by comparison with a 1-kb ladder standard.
Assay for Cytotoxic T Lymphocyte Activity
Preparation of effectors
Effector cells were HIV antigen-driven cell lines generated from freshly isolated or frozen PBMC by stimulation with fixed autologous PBMC expressing HIV gene products. PBMC were obtained by ficoll-hypaque density centrifugation of heparinized blood. Stimulators were prepared by infecting PBMC with vaccinia virus vectors encoding HIV Env, Gag, reverse transcriptase (RT), and Nef at an MOI of 5:1 vaccinia virus to PBMC. After a 16-h incubation at 37°C in 5% CO2, stimulators were fixed for 45 s with 0.1 mL of 0.05% glutaraldehyde (Sigma, St. Louis). Fixation was stopped by the addition of 0.1 mL of 2 M L-lysine (Sigma). Stimulators infected separately with the four vaccinia virus constructs were washed three times and pooled at a ratio of 1:1:1:1. Uninfected PBMC and autologous stimulators were then mixed at a ratio of between 5:1 and 10:1 and cultured in RPMI 1640 (Gibco BRL, Burlington, Canada), supplemented with 15% fetal calf serum (Gibco BRL), 2 mMl-glutamine (ICN Biomedicals, Costa Mesa, CA), 50 IU/mL penicillin (ICN), 50 μg/mL streptomycin (ICN), 50 μM 2-mercaptoethanol (Sigma), and 200 U/mL IL-2 (Chiron, Emeryville, CA) (R-15-200). Twice a week, cell cultures were adjusted to 5 × 105 cells/mL in fresh R-15-200. Cell lines were tested for CTL activity between days 14 and 21 of culture.
Vaccinia virus—HIV constructs
Five vaccinia virus—HIV constructs were used to infect autologous EBV-B cell lines for use as targets; four of these were also used to infect autologous PBMC for use as stimulators. vMN462 expressing the full-length gp160 Env of HIV-1MN, vDK1 expressing Gag of the HXB2 subclone of HIV-1, vCF21 expressing all but the last 22 residues of RT of the HXB2 subclone of HIV-1, and vSC8 expressing the LacZ gene of Escherichia coli (a non-HIV gene product used as a negative specificity control) were obtained from the AIDS Research and Reference Reagent Program (Division of AIDS, National Institute of Allergy and Infectious Diseases, NIH, Bethesda, MD). vP1218 expressing Nef of HIV-1MN was a gift of E. Paoletti (Virogenetics, Troy, NY). Each of these constructs was added to PBMC (to be used as stimulators) or EBV-B cells (to be used as targets) at a vaccinia virus-to-cell ratio of 5:1 for 30 min with occasional shaking. EBV-B cells to be used as targets were then labeled with 100 μCi of [51Cr]NaCrO4 (New England Nuclear, Boston) per 5 × 105 EBV-B cells and incubated at 37°C in 5% CO2 for 16 h. Labeled targets were washed three times before assay.
Cytolytic activity of antigen-driven effectors present in bulk cultures established from PBMC was measured in a 4-h 51Cr-release assay. Targets were autologous EBV-B cells or EBV-B cell lines unmatched or matched with effectors for single HLA class I alleles. EBV-B cells were generated for each subject by use of B95-8 marmoset line supernatant . Assays were done in triplicate in round-bottomed 96-well microtiter plates at effector to target (E:T) ratios of 10:1, 5:1, and 2.5:1 or as indicated for individual experiments. Chromium release was measured in a gamma counter (5500B; Beckman Instruments, Fullerton, CA). The percentage of specific lysis was determined as follows: 100 × (experimental release − spontaneous release)/(maximal release − spontaneous release). For each target, spontaneous release was determined from wells containing medium only, and maximal release was calculated from wells containing 1% Triton X-100. Spontaneous release ranged from 15% to 30% of maximal release. CTL activity was defined as present if the percentage of specific lysis of test wells was >10% above background lysis (cytotoxicity directed at vSC8, a non—HIV antigen—expressing target) at at least two E:T ratios or in more than one experiment. Vaccinia virus-specific cytolysis also present in bulk cultures was reduced by adding to replicate cultures, for each E:T, a 30-fold excess of unlabeled vSC8-infected targets. All of the results reported for CTL activity of antigen-stimulated bulk cultures are those corresponding to wells seeded with cold targets to inhibit recognition of non-HIV specificities.
PBMC were typed for major histocompatibility complex (MHC) class I antigens by standard complement-mediated cytotoxicity assay with 72-well plates purchased from the Canadian Red Cross Society (Ottawa).
The χ2 test was used for comparison of sample proportions. P <.05 was considered significant.
The expected size of EcoRI-digested fragments amplified from the genomic DNA of persons homozygous for the CCR-5 locus wild type allele determined by use of the primer set described in Methods was 332 and 403 bp. Fragment sizes expected from heterozygotes were 332, 371, and 403 bp. Fragment sizes expected from homozygotes for the deletion mutant were 332 and 371 bp. Of the 19 HIV EUs genotyped for CCR-5, 15 (79%) were homozygous wild type and 4 (21%) were heterozygous at this locus. None were Δ32CCR-5 homozygous (table 2). No differences in allele or genotype frequencies were detected between these 19 EUs and 52 HIV-infected patients (table 2) or between these 19 EUs and 33 seronegative persons with no risk factors for HIV infection (not shown) by the χ2 test. Furthermore, no differences in the proportion of CCR-5 heterozygotes were detected between the 14 EUs who were in HIV infection status—discordant couples and their HIV-seropositive partners, whether all subjects or only Caucasians were included in analyses (table 2).
HIV-specific memory CTL activity
To detect potentially low-frequency CTL clones specific for HIV in EUs, an in vitro stimulation step with autologous cells expressing HIV-1 antigens was used to amplify HIV-specific memory CTL. Bulk cultures prepared from HIV-infected persons by this method display HIV-specific CTL responses (figure 1A, 1B), whereas those prepared from HIV-seronegative subjects do not (figure 1C, 1D). HIV-specific CTL activity was measured in bulk cultures established from the PBMC of 17 EUs and 14 uninfected controls at low risk for HIV infection. Figures 1E and 1F show examples of HIV-specific CTL activity detected in 2 EUs. Table 3 summarizes the specificity of HIV-specific CTL derived from the 17 EUs and 14 low-risk seronegative controls screened by this method. Seven (41.1%) of 17 EUs, compared with none of the low-risk seronegative controls, exhibited above-background levels of CTL activity directed at HIV. A significantly greater proportion of EUs versus low-risk seronegative persons had HIV-specific memory CTL (P = .006, χ2 test).
Antigen-driven bulk cultures derived from 3 EUs were screened for MHC class I antigen restriction by use of targets matched or not for MHC alleles expressed by bulk CTL effectors (figure 2). CTL from all 3 EUs lysed MHC-matched but not unmatched targets. Serologic MHC class I typing of EU subject 8 (EU8) showed that this subject was HLA-A2, B14, B62, Cw3, Cw5. HIV-specific CTL activity amplified from the PBMC of EU8 recognized Gag presented on autologous and HLA-A2— and HLA-B62—matched targets but not on MHC class I—unmatched targets (figure 2).
PBMC from 3 EUs were tested several times after their last exposure to HIV to assess persistence of HIV-specific CTL activity. EU6's last exposure to HIV was in July 1995. In September 1995, EU6 had HIV RT— and Nef—specific CTL activity. In January 1996, 6 months after the last exposure, it was still possible to amplify RT- and Nef-specific CTL activity from EU6's PBMC. In addition, reactivity to HIV Gag was also detected at this time (figure 3A). EU18 is a health care worker who was exposed four times to HIV-infected body fluids by needlestick injuries. The last exposure occurred in September 1995. This subject was tested three times: in September 1995, 18 months later in March 1997, and 34 months later in July 1998. In September 1995, EU18 had HIV-specific CTL to HIV Env, Gag, RT, and Nef. Eighteen months later, this subject had above-background CTL activity to Gag, RT, and Nef. Thirty-four months after the final exposure, EU18 had above-background CTL activity to Nef at an E:T ratio of 10:1 (figure 3B). EU15 is a gay man with a history of high-risk behavior. As of September 1995, EU15 was involved in a stable relationship with a seronegative partner and was not exposed to HIV during the observation period. In September 1995, EU15 had HIV-specific CTL to HIV Nef. This HIV-specific reactivity broadened by November 1995 to recognize HIV Env and Nef. When the subject was retested in December 1995, the same HIV gene products were recognized (figure 3C).
This report shows that a significant proportion of HIV-exposed seronegative subjects have HIV-specific precursor CTL that can be expanded in vitro under conditions of antigen-specific stimulation. This activity can persist after the last HIV exposure for up to 34 months. During this time, the specificity of the response can change. Patterns of reactivity varied from person to person, with recognition of up to four HIV gene products by bulk cultures derived from EUs.
Despite exposure to HIV, in no case can persistent HIV seronegativity in the 19 subjects described herein be explained by homozygosity for Δ32CCR-5, a known genetic resistance factor [17–21]. The product of this mutant allele is a nonfunctional form of the coreceptor for HIV entry into CD4 cells of the monocyte/macrophage lineage [17, 18, 28]. Macrophage-tropic viruses are the principal form of virus involved in person-to-person transmission . The Δ32CCR-5 allele is found in ∼10% of Caucasians [17–19, 21]. People expressing the homozygous phenotype are substantially, although not absolutely, resistant to HIV [17–19, 28, 29], and CD4 cells from these persons are resistant to infection with macrophage-tropic HIV [2, 18, 28, 34]. As many highly exposed seronegative persons, including members of the cohort described here, are not Δ32CCR-5 homozygotes, other factors likely play a role in resistance to HIV [5, 18, 19, 34].
Hoffman et al.  found a higher frequency of Δ32CCR-5 heterozygotes in seronegative partners of HIV infection status-discordant couples compared with their HIV-positive partners. This effect was seen in heterosexual but not in homosexual couples. The results imply that heterozygosity at this locus may confer partial protection from infection through a heterosexual route. Although the sample sizes are too small for a definitive conclusion, comparison of the proportion of CCR-5 heterozygotes in the EU cohort described here with that in 52 HIV-infected subjects or with that in their HIV-infected partners revealed no significant proportional differences.
The presence of memory HIV-specific MHC class I—restricted CTL in these persons implies that they were exposed to live replication-competent HIV. Induction of CTL likely results from infection of CD4 cells by HIV followed by one or more cycles of replication to produce immunogenic viral epitope—MHC antigen combinations. Defective virus particles or virus debris would not be efficient at inducing HIV-specific CTL . Therefore, maintenance of seronegativity in EUs with HIV-specific CTL implies that, although they were exposed to live virus, they were able to control viral replication and clear virus before establishment of a progressive infection characteristic of those who undergo HIV seroconversion.
The interest in immune mechanisms potentially mediating resistance arises from the implications this would have for vaccine development. There is no evidence that systemic antibodies to HIV mediate protection. On the contrary, presence of serum anti-HIV antibodies is diagnostic for HIV infection. Cellular immune responses to HIV are often observed in the absence of infection [1, 3–5, 8–15]. Although proof is lacking as to whether HIV-specific memory CTL activity in these persons protects them from seroconversion on subsequent exposure, information from several sources supports this possibility.
In mice, as in humans, infectious pathogens frequently induce either a predominant cell-mediated or humoral immune response [36, 37]. The nature of the dominant response depends on the type of T helper cell induced. Th1 cells secrete IL-2 and interferon-γ and support the differentiation delayed-type hypersensitivity, CTL, and macrophage activation . Th2 cells produce IL-4, IL-5, and IL-10 and provide help for B cell differentiation into antibody-secreting cells . There is a tendency for either cell-mediated or antibody responses to predominate in any particular immune response . This tendency is thought to result from cross-regulation through the cytokine network, such that Th1 cells and their products inhibit induction of Th2 cells and vice versa . In several animal and human disease models, one or the other pole of the T helper response is associated with resistance or susceptibility to infection or disease [40–46]. The genetic background of the host and dose and route of antigen delivery influence which type of immune response will predominate . In macaques, low-dose immunization with a simian immunodeficiency virus vaccine results in induction of cellular immune responses without humoral immunity. Macaques immunized this way are protected from challenge with infectious virus .
HIV-specific CTL may be particularly well-suited to controlling virus spread. In vitro studies on HIV-specific CTL clones revealed that HIV-specific CTL recognize and kill targets infected with replication-competent virus before mature virus can bud from the host cell and infect new targets . Once CTL clones have been induced to kill HIV-infected targets through MHC-restricted recognition of HIV epitopes, they also secrete factors that inhibit the replication of HIV in a non—MHC-restricted manner .
Although 7 of 17 EUs had detectable HIV-specific CTL activity, 10 subjects did not. Several reasons may explain why no HIV-specific CTL activity was detected in these persons. Several steps must be completed successfully before HIV either establishes a progressive infection or induces a CTL response without seroconversion in its host. HIV must pass through mucosal and epithelial barriers, enter CD4 cells, replicate itself, and present viral antigens to the immune system. HIV entry may be aborted at any of these steps, such that the host, despite exposure to HIV through high-risk behavior, never encounters the signals that induce differentiation of virus-specific CTL. Another possibility is that HIV-specific CTL have been induced but are directed at epitopes not present on the virus isolates used to amplify and screen for CTL activity. Although this is a possibility, others have shown that HIV-specific CTL can tolerate up to 3 amino acid changes within a recognized HIV epitope [49, 50]. This permits a degree of cross-isolate and cross-clade recognition of HIV isolates by HIV-specific CTL.
In summary, the study of HIV-exposed seronegative persons is important because such persons hold the key to what constitutes protection from HIV infection. Identification of and replication of acquired immunity factors that protect against HIV infection is the principal goal for vaccine development. Data presented here show that HIV-exposed seronegative subjects can be primed through their high-risk behavior to develop HIV-specific CTL activity in the absence of seroconversion. These persons seem to have encountered HIV as a limited infection and as a result have been able to clear it from the circulation. These memory CTL responses can be multispecific, can change over time, and can persist for up to 34 months after the last exposure.
We thank Famane Chung for performance of many of the CTL assays, Rick Pilon and Sharon Cassol (Ottawa General Hospital, Ottawa) for HIV quantitation, Galit Alter for CCR-5 genotyping, and Nadine Ayoub for technical assistance. We are indebted to Jason Szabo, Julian Falutz, Erwin Goldberg, Howard Turner, Vihn Kim Nguyen, and Philip Joliot for referring EU patients to this study. We also thank the members of the HIV EU cohort and their partners, whose cooperation has made this study possible.