Induction of Human Immunodeficiency Virus Type 1 (HIV-1)-Specific T-Cell Responses in HIV Vaccine Trial Participants Who Subsequently Acquire HIV-1 Infection

Abstract

Candidate human immunodeficiency virus type 1 (HIV-1) vaccines designed to elicit T-cell immunity in HIV-1-uninfected persons are under investigation in phase I to III clinical trials. Little is known about how these vaccines impact the immunologic response postinfection in persons who break through despite vaccination. Here, we describe the first comprehensive characterization of HIV-specific T-cell immunity in vaccine study participants following breakthrough HIV-1 infection in comparison to 16 nonvaccinated subjects with primary HIV-1 infection. Whereas none of the 16 breakthrough infections possessed vaccine-induced HIV-1-specific T-cell responses preinfection, 85% of vaccinees and 86% of nonvaccinees with primary HIV-1 infection developed HIV-specific T-cell responses postinfection. Breakthrough subjects' T cells recognized 43 unique HIV-1 T-cell epitopes, of which 8 are newly described, and 25% were present in the vaccine. The frequencies of gamma interferon (IFN-γ)-secreting cells recognizing epitopes within gene products that were and were not encoded by the vaccine were not different (P = 0.64), which suggests that responses were not anamnestic. Epitopes within Nef and Gag proteins were the most commonly recognized in both vaccinated and nonvaccinated infected subjects. One individual controlled viral replication without antiretroviral therapy and, notably, mounted a novel HIV-specific HLA-C14-restricted Gag LYNTVATL-specific T-cell response. Longitudinally, HIV-specific T cells in this individual were able to secrete IFN-γ and tumor necrosis factor alpha, as well as proliferate and degranulate in response to their cognate antigenic peptides up to 5 years postinfection. In conclusion, a vaccinee's ability to mount an HIV-specific T-cell response postinfection is not compromised by previous immunization, since the CD8⁺ T-cell responses postinfection are similar to those seen in vaccine-naïve individuals. Finding an individual who is controlling infection highlights the importance of comprehensive studies of breakthrough infections in vaccine trials to determine whether host genetics/immune responses and/or viral characteristics are responsible for controlling viral replication.

The human immunodeficiency virus (HIV) pandemic is of staggering proportions, with nearly 40 million individuals infected (1). This has created an urgent need to develop a safe, efficacious vaccine as the single prevention strategy to benefit the largest number of persons. In the absence of established correlates of protection, most experts predict that both humoral and cellular immunity will be required for sustained immunity against HIV type 1 (HIV-1). Successful vaccines against other viral infections rely on induction of protective levels of antiviral antibodies (10). Likewise, the first generation of HIV vaccine candidates, notably, recombinant monomeric envelope protein subunit vaccines, was designed to elicit antibody responses. Studies of individuals who participated in these trials but subsequently became infected with HIV (breakthrough infections), therefore, have focused on characterization of their HIV-specific neutralizing antibody responses prior to infection and on the ability of these antibodies to neutralize viruses infecting vaccine recipients (5, 12, 17, 23). Unfortunately, these studies revealed that nearly all vaccine-induced antibodies by this approach lack potency and breadth of reactivity against transmitting strains (discussed in reference 8) and ultimately fail to provide HIV protection.

While new strategies for the design of vaccines which induce effective anti-HIV-l antibodies are still being identified, the emphasis has shifted to the design of vaccines that will induce antiviral CD8⁺ T-cell immunity. Although HIV-specific CD8⁺ T cells alone are unlikely to prevent HIV infection, they may have an important impact in rapidly controlling viral replication following infection. A large effort has centered upon the development of recombinant attenuated poxviruses as a safe approach in humans to induce HIV-1-specific T-cell immunity. Recombinant canarypox HIV vaccines, the most extensively tested to date, induce CD8⁺ T-cell responses at various frequencies (9 to 70%) depending upon the construct tested and the method of detection (4, 11, 15, 18).

Here we investigate the characteristics of the T cells present during early infection in individuals previously enrolled in HIV vaccine trials evaluating recombinant canarypox candidate vaccines, contrasting their responses with those of nonvaccinated subjects who have recently acquired HIV-1 infection. The vaccine constructs contained full-length gag (HIV-1 LAI) and env gp120 (HIV-1 MN) coding regions, with some constructs also containing parts of pol and nef. Several individuals also received a recombinant gp120 SF-2 protein boost (Chiron) or a recombinant gp120 MN protein boost (Vaxgen Inc.). Vaccinees with breakthrough infections described in this report took part in phase I/II HIV vaccine trials where the net cumulative cytotoxic T-lymphocyte (CTL) response rates were between 33 and 50%, as assessed by the chromium release assay (4, 18). We determined whether any of the T-cell responses detected during early infection in these individuals represented anamnestic responses induced by immunization prior to infection. Furthermore, we characterized longitudinally the function of HIV-specific T cells postinfection in the only volunteer controlling infection in the absence of antiretroviral therapy (ART). These are the first comprehensive data describing HIV-specific T-cell responses pre- and postinfection in participants who became HIV-infected in HIV vaccine trials of vaccine candidates designed to elicit CD8⁺ T-cell immunity.

MATERIALS AND METHODS

Study population.

Volunteers enrolled in the National Institutes of Health (NIH)-sponsored HIV Vaccine Trials Network protocol 402 were evaluated in this study. Each individual had participated in phase I or II trials examining safety and immunogenicity of recombinant canarypox vectors (Sanofi Pasteur) with or without a recombinant gp120 protein vaccine boost (Vaxgen Inc., Chiron Corp.). Results from these studies and the description of the candidate vaccines have been previously published (3, 4, 11, 18). Each study participant became infected as a direct result of high-risk behavior. The institutional review boards at each site approved the trials, and each subject gave informed consent for participation in the studies.

The time of infection was estimated using an algorithm based on the participant's last negative and first positive HIV-1 Western blot assay or plasma RNA reverse transcriptase PCR, acute retroviral symptoms, and exposure history (17). When participants did not report exposure or seroconversion symptoms, the date was assigned as the mid-point between the last negative and first positive serological or DNA test. Plasma HIV-1 RNA was determined by quantitative reverse transcription-PCR assay (Roche Molecular Systems, Branchburg, N.J.) (14), which has a sensitivity of 50 copies/ml. Peripheral blood CD4⁺ and CD8⁺ T-cell counts were determined by flow cytometry with consensus methods (14) and expressed as cells per microliter. HLA typing was performed at the Puget Sound Blood Center by sequence-specific primer PCR as previously described (7).

The comparison population consisted of 16 patients with newly diagnosed HIV-1 infection who were followed longitudinally at the University of Washington Primary Infection Clinic (9). Primary infection was documented by signs and symptoms consistent with an acute retroviral syndrome (26, 27). The duration of infection in these subjects was designated from the date of onset of acute symptoms.

Peptides.

Peptides for the gamma interferon (IFN-γ) ELISpot were either synthesized by Fred Hutchinson Cancer Research Center Shared Resources Facility (Seattle, WA) or supplied by the NIH AIDS Research and Reference Reagent Program (Bethesda, MD). Peptides were 15 amino acids in length, overlapping by 11 amino acids, and were based on the proteins encoded by the HxB2 clade B HIV sequence, except for Env (based on MN) and Nef (based on BRU). The IFN-γ ELISpot assays were performed by stimulating peripheral blood mononuclear cells (PBMC) with pools of 50 peptides (each peptide at a final concentration of 1 μg/ml). Experiments were performed to define the 15-mer recognized within a peptide pool using peptide matrices and, whenever possible, the 8- to 11-mer optimal peptide as previously described (25).

IFN-γ ELISpot.

The ELISpot assay methodology has been described in detail elsewhere (25). In brief, cryopreserved PBMC were thawed, washed, and rested overnight in R10 (RPMI 1640 [GibcoBRL, New York] containing 10% fetal calf serum [Gemini Bioproducts, California], 2 mM l-glutamine [GibcoBRL], 100 U/ml penicillin G, 100 μg/ml streptomycin sulfate) before use in ELISpot assays. Cells were resuspended in R10 and aliquoted at 1 × 10⁵ to 2 × 10⁵ cells/well. Peptide antigens were added at a final concentration of 1 to 2 μg/ml, R10 alone provided a negative control, and 1 μg/ml phytohemagglutinin (PHA-P) served as a positive control. Due to limited numbers of cells, experimental and control testing was performed in duplicate only. Spots were counted using the CTL Analyzer and software (Cellular Technology Ltd., Cleveland, OH). For functional avidity measurements, a standard IFN-γ ELISpot was performed with the indicated peptides at twofold dilutions ranging from 2 μg/ml to 5.12 × 10⁻⁶ μg/ml. The regression curve was drawn with the sigmoidal fit tool in the Origin 6.0 software.

CFSE labeling and polychromatic flow cytometry.

PBMC were labeled with 1.2 μM carboxyl fluorescein succinimidyl ester (CFSE; Molecular Probes, Inc., Eugene, Oreg.) for 8 min in the dark at 37°C, and free CFSE was quenched with 100% 4°C fetal calf serum for 1 min. Labeled cells were washed before in vitro culture with 2 μg/ml of HIV-1 peptides for 5 days. As a positive control, the cells were stimulated with 30 ng/ml anti-CD3 and 1 μg/ml anti-CD28 monoclonal antibodies. On day 5 following CFSE labeling, cells were washed in R10 and rested overnight. The next day cells were washed and stimulated for 6 h in the presence of brefeldin A (10 μg/ml) with either staphylococcal enterotoxin B (1 μg/ml; Sigma) as a positive control, HIV-1 peptides (1 μg/ml each peptide/sample), or no peptide as a negative control. CD107a-PECy5 was added during stimulation. Intracellular staining was performed with standard techniques (BD CFC protocol), using previously titrated antibody reagents. GolgiStop (BD) was used in addition to brefeldin A in panels with CD107a. CD3-allophycocyanin-Cy7, CD8-peridinin chlorophyll a-Cy5.5, IFN-γ-phycoerythrin-Cy7, interleukin-2 (IL-2)-phycoerythrin, tumor necrosis factor alpha (TNF-α)-allophycocyanin, and CD107a-phycoerythrin-Cy5 monoclonal antibodies were supplied by BD. Data acquisition was performed on an LSRII flow cytometer (BD), collecting 100,000 to 200,000 lymphocyte-gated or 10,000 to 40,000 CD8⁺ events per sample, and analyzed with Flowjo software (Tree Star Inc., Ashland, OR). Positive responses for the ex vivo cytokine secretion assay were designated when the percentage of bright cytokine-positive CD8⁺ T cells was twice that of the negative control and at least 0.05% (background subtracted).

Statistical analysis.

The two-sided exact Wilcoxon rank-sum test was used to assess differences in the breadth and magnitude of HIV-1-specific T-cell responses between vaccine-primed and vaccine-naïve individuals during early infection. All statistical computations were performed using Jmp (version 5.0; SAS Institute) or SAS (version 8.2; SAS Institute) software.

RESULTS

Study participants.

Sixteen HIV-1 vaccine trial participants between the ages of 22 and 43 years (median, 32 years) were diagnosed with HIV-1 infection after study enrollment; 13 were vaccine recipients, and 3 were placebo recipients (Table 1). The vaccinees received a recombinant canarypox candidate vaccine, with or without a recombinant gp120 protein boost. Ten of the 13 completed their vaccination series (a total of four to five immunizations) before acquiring HIV-1 infection between November 1996 and May 2002 as a result of high-risk behavior. The predominant risk factor for HIV-1 acquisition was men having high-risk sexual contact with men (MSM). Two participants were women reporting low-risk sexual activities at enrollment of the vaccine trial. The median length of time between last vaccination and estimated time of infection was 249 days (Table 1; see Materials and Methods for the algorithm used to estimate time of infection). Median plasma HIV-1 RNA and CD4 counts in these individuals after HIV-1 infection (sampled on assay date) (Table 2) were not statistically significant between vaccinees (4.78 log₁₀ copies/ml; 552 cells/mm³) and placebo recipients (4.27 log₁₀ copies/ml; 657 cells/mm³) (P = 0.4 and 0.1, respectively; Wilcoxon rank sum test). A recent publication provides further details of the subjects' clinical courses (20).

TABLE 1.

Demographic profile and treatment of study population

Participant	Treatment	Immunogena	No. of vaccinations prior to infection date	Interval between last treatment and HIV acquisition (days)b	Gender	Race	Agec (yr)	Risk factord
014A-003	Placebo	ALVAC/saline	4	67	Male	Other	30	MSM
014A-006	Placebo	ALVAC/saline	3e	34	Male	White	25	MSM
402FIE	Placebo	ALVAC/saline	4	255	Male	White	31	MSM
014A-001	Vaccine	vCP205/gp120	4	376	Female	Other	33	Low risk
014A-004	Vaccine	vCP205	4	242	Male	White	35	MSM
014A-005	Vaccine	vCP205/gp120	4	392	Male	White	43	MSM
014A-008	Vaccine	vCP205	4	105	Male	Other	41	IDU
014A-009	Vaccine	vCP205	4	530	Male	White	25	MSM
014A-011	Vaccine	vCP205	3e	13	Male	White	42	MSM
40287N	Vaccine	vCP125/gp120	4	1,176	Female	Black	22	Low risk
402I03	Vaccine	vCP65/gp120	4	577	Male	White	32	MSM
402I15	Vaccine	vCP205/gp120	5	273	Male	White	34	MSM
403-006	Vaccine	vCP205/gp160	4	1,589	Male	White	38	MSM
2x0226	Vaccine	vCP205/gp120	4	215	Male	Other	31	MSM
203-033	Vaccine	vCP1452	4	157	Male	White	27	MSM
203-157	Vaccine	vCP1452/gp120	1e	14	Male	Black	27	Bisexual

^aRecombinant canarypox vectors are as follows: vCP125 expresses HIV-1 MN gp160 (Pasteur Merieux/Connaught); vCP65 expresses rabies virus glycoprotein (Pasteur Merieux/Connaught); vCP205 expresses HIV-1 envelope gp120 (strain MN) linked to the transmembrane portion of HIV-1 gp41 (strain LAI) and the HIV-1 LAI genes encoding the entire Gag protein and a portion of the Pol sequence, sufficient to evoke the protease portion (Pasteur Merieux/Connaught); vCP1452 expresses HIV-1 envelope gp120 (strain MN) linked to the transmembrane portion of HIV-1 gp41 (strain LAI), the HIV-1 LAI gene encoding the entire Gag protein, a portion of the pol gene encoding the protease, and a synthetic polynucleotide encompassing several known human CTL epitopes from the nef and pol gene products. It also contains sequences encoding the E3L and K3L vaccinia virus proteins into the C6 site (vCP1452) (Virogenetics Corp., Troy, NY, and Sanofi Pasteur, S.A., Marcy L'Etoile, France).

^bSee Materials and Methods for the algorithm used for estimating time of HIV-1 infection.

^cAge at enrollment of parent vaccine protocol.

^dMSM, men having sex with men; IDU, injection drug user. All volunteers reported HIV high-risk behavior prior to enrollment, except for those designated.

^eParticipant did not complete vaccination regimen.

TABLE 2.

HIV-1 epitopes recognized by study participant CD8⁺ T cells before and after HIV-1 infection

Participant identifier	Vaccine treatment	Class I HLA type	Days postinfection	Viral load RNA (copies/ml)	HIV-1 epitopei	IFN-γ SFC/ 106 PBMC
						Before	After
014A-003	Placebo	A*0101, A*2902; B*1501, B*5701	58	1,525	Gag TSTLQEQIGW	<20	125
014A-006	Placebo	A*0101; B*0801, B*1401	59	1,749,235	Env IVELLGRRGWEVLKY	<20	1,790
					Env RQGLERALL	<20	2,835
					Env ERYLKDQQL	<20	2,350
					Nef FLKEKGGL	<20	695
					Nef AAVDLSHFL	<20	2,545
402FIEa	Placebo	A3, A32; B38, B64	60	2,069	Gag RLRPGGKKK	<20	137
					Gag VDRFYKTLRAEQASQ	<20	690
					Env MHEDIISLW	<20	280
					Vpr GLGQHIYETYGDTWAe	<20	260
014A-001	vCP205/gp120	A*0202, A*8001; B*1801, B*5301	64	NAb	Gag QAISPRTLNAW	<20	103
					Env RIRQGLERA	<20	373
					Nef YPLTFGWCY	<20	113
014A-004	vCP205	A*0201, A*1101; B*4002, B*5101	46	<50	Pol TAFTIPSI	<20	210
					Gag CTERQANFL	129	70
					Gag SLYNTVATLh	47	149
					Nef AVDLSHFLK	74	740
014A-005	vCP205/gp120	A*2501, A*3002; B*0702, B*1801	71	65,775	Gag QAISPRTLNAW	<20	358
					Gag ETINEEAAEW	<20	363
					Env TENFNMWKNNMVEQM	<20	585
					Env IPRRIRQGL	<20	943
					Nef FPVTPQVPLR	<20	558
					Tat CCFHCQVC	<20	408
					Pol QKQGQGQWTYQIYQE	<20	165
					Pol KIQNFRVYY	<20	435
										Vpr EAVRHFPRIWLHGLG	<20	490
014A-008	vCP205	A*0201, A*0301; B*4501, B*5301	106	130,483	Gag RLRPGGKKK	<20	220
014A-009	vCP205	A1, A2; B8, B38	87	12,413	Gag EIYKRWII	<20	290
					Env MHEDIISLW	<20	300
					Env RRGWEVLKY	<20	115
					Env RQGLERALL	<20	240
					Nef IHSQRRQDILDLWIY	<20	125
014A-011	vCP205	A*0201; B*31012, B*1501	122	<400c	None
40287N	vCP125/gp120	A1, A33; B8, B44	318	17,392	Gag GGKKKYKL	<20	570
					Gag EIYKRWII	<20	150
					Vpr EAVRHFPRIWLHGLG	<20	160
					Vpu EYRKILRQRKIDRLI	<20	120
402I03f	vCP65/gp120	A3, A33; B15 (63), B27	242	2,109	Gag IRLRPGGKK	<20	3,960
					Gag KRWIILGLNK	<20	5,310
					Gag GLNKIVRMY	<20	570
					Gag MMQRGNFRNQRKIVKe	<20	2,510
					Env SFNCGGEFF	<20	750
					Env LQRAGRAILHIPTRI	<20	190
402I15	vCP205	A1, A10 (26); B8, B17 (57)	92	41,742d	Gag QMVHQAISPRTLNAWe	<20	300
					Rev DEELIRTVRLIKLLYe	<20	330
					Nef FLKEKGGL	<20	480
403-006	vCP205/gp160	A2, A2; B62, B35	141	3,069	Gag PPIPVGDIY	<20	95
					Vif DAKLVITTY	<20	95
2x0226	vCP205/gp120	NAb	30	NAb	Vif DAKLVITTY	<20	2,943
					Env PIPIHYCAPAGFAIL	<20	153
203-033	vCP1452	A*2501, A*6601; B*0702, B*1801	39	165,509d	Noneg
203-157	vCP1452/gp120	A*3001, A*3002; B*4201/02, B*4403/26/30	91	32,511d	Pol YPGIKVRQL	NDj	490
					Pol KIQNFRVYY		365
					Pol DDTVLEEMSLPGRWK		215

^aSome responses mapped in this individual at 143 dpi (viral load of <400 and participant on ART).

^bNA, data not available.

^cOn ART at time of mapping.

^dNo viral load was obtained on the visit date of sample collection. Thus, the closest viral load reported for 402I15 was 3.3 months post-sample date; for 203-033, it was 0.4 months post-sample date; for 203-157, it was 0.7 months post-sample date.

^eBy matrix only. The individual peptide was not tested due to insufficient cells.

^fResponses to Vif and Vpr peptide pools were not tested in this individual. Pol pools 9 and 10 were also positive, but an exact peptide(s) was not identified due to insufficient cells.

^gOnly a weak PHA response (587 SFC/10⁶ PBMC) was detected at this time point.

^hSee later figures and Discussion.

ⁱThe vaccine contained the epitopes shown in bold.

^jND, not determined.

Characterization of HIV-specific T-cell responses early in infection in previous vaccine recipients.

HIV-1-specific T-cell responses were measured by IFN-γ ELISpot as close to the estimated date of infection as possible. Of note, volunteers were typically followed every 3 to 6 months after completion of immunizations according to the study protocol, which did not permit us to necessarily capture events during acute infection. HIV-specific T cells were defined as recognizing 15-mers, unless there were sufficient cells to define the optimal 8- to 11-mer epitopic response. All assays were performed on cryopreserved PBMC isolated at times when the volunteers were not receiving ART, with the exception of 402FIE and 014A-011. Mean viability was 85% (median, 87%; range, 64 to 100%), and mean PHA response was 9,477 spot-forming cells (SFC)/10⁶ PBMC (median, 9,940 SFC/10⁶ PBMC; range, 587 SFC/10⁶ PBMC to too numerous to count). The peptides recognized by PBMC from each individual are summarized in Table 2, and their positions throughout the encoded gene products are shown in Fig. 1A.

FIG. 1.

A. Peptide sequences and amino acid locations within each protein of HIV peptides recognized by CD8⁺ T cells during early infection in individuals who have participated in previous canarypox trials of potential HIV vaccines. B. Comparison of epitopes recognized by HIV-specific T cells early in infection in vaccine-naïve individuals from the primary infection cohort (blue bars) with HIV peptides recognized by T cells from vaccinated individuals (red bars) or placebos (green bars). Open boxes indicate newly defined epitopes.

We detected postinfection HIV-specific T-cell responses in all but two participants by IFN-γ ELISpot. One of these individuals (014A-011) was receiving ART and had an undetectable viral load, which may account for the lack of responses detected in this individual. PBMC from the other individual were viable (88% viability), but the PHA response was weak (587 SFC/10⁶ PBMC), which suggests that there were fewer T cells than would be expected. The number of HIV peptides recognized by IFN-γ-secreting CD8⁺ T cells ranged from 1 to 5 (median, 4) among the placebos and from 0 to 9 (median, 3) among the vaccine recipients. Of the 11 vaccinees who mounted detectable T-cell responses, 9 possessed T cells that recognized HIV-1 epitopes that were present in the vaccines they received, but 10 of these individuals also possessed HIV-1-specific T cells recognizing epitopes that were not present in the vaccines (Table 2). The magnitude of the HIV-specific T-cell responses in the vaccinees ranged from 70 to 5,310 SFC/10⁶ PBMC and was not correlated with plasma HIV-1 RNA viral load (r = 0.357, P = 0.26) measured at the same time point. It was not possible to assess the correlation of T-cell response magnitude and viral load in the placebo group due to the small sample size (n = 3).

Forty-four unique HIV peptides were recognized in the 16 participants (13 vaccine and 3 placebo recipients), and of these, 8 have not been described previously (shown in Table 2). CD8⁺ (not CD4⁺) T cells mediated all responses when subsets were examined by IFN-γ intracellular staining using flow cytometry (data not shown). Nef and Gag proteins were the most commonly recognized among HIV-1 proteins by T cells in these individuals (Fig. 1A). Six of 16 individuals recognized at least one Nef epitope and 6 unique Nef epitopes were identified, whereas 11 of 16 individuals recognized at least 1 Gag epitope, with 15 unique epitopes identified. The frequency of IFN-γ-secreting cells was not greater for epitopes within the vaccine than for those outside the vaccine (P = 0.64; Wilcoxon rank sum test). There was insufficient power to determine whether any single epitope or HLA allele was overrepresented in this cohort (see Table 2 for epitopes and HLA typing).

Comparison of early T-cell responses detected in vaccinees to those in vaccine-naïve individuals.

HIV-specific T-cell responses identified in the 13 vaccinated individuals were compared to responses seen during early infection in 16 study participants enrolled in a longitudinal primary infection cohort (9) who had never received an HIV-1 vaccine. These individuals were selected based on availability of cryopreserved PBMC from leukapheresis performed during primary infection. They were comparable in age, ethnicity, gender, risk behavior, and date of infection to the individuals in our study (data not shown). Of the 35 unique peptides identified in the vaccine recipients (compared to the acutely infected cohort), 12 corresponded to epitopes identified in the primary infection group (9), 16 were epitopes previously described in chronic infection (1, 16), and 7 are newly described (Fig. 1B). Only two of the seven newly described epitopes were contained in the vaccines that the subjects received. Between the two populations of relatively small sample size, no statistical differences were found in the number of epitopes (P = 0.13) or in the number of HIV-1 proteins (P = 0.18) recognized by T cells. Similarly, no significant difference was found between the frequencies of cells expressing IFN-γ in vaccinees versus placebo recipients (P = 0.13; Wilcoxon rank sum test), suggesting that the responses observed in vaccinees were unlikely to be anamnestic.

Characterization of preseroconversion HIV-specific T-cell responses.

Cryopreserved PBMC were assayed preinfection, at the closest time before infection as possible, to determine whether any of the responses seen in early infection were anamnestic responses induced by immunization prior to infection. Mean viability was 87% (median, 87%; range, 72 to 97%), and mean PHA response was 4,548 SFC/10⁶ PBMC (median, 1,942 SFC/10⁶ PBMC; range, 245 to 19,585 SFC/10⁶ PBMC). Only one individual (014A-004) possessed measurable preseroconversion HIV-1-specific T-cell responses after four immunizations with vCP205 (Table 2). Interestingly, this individual never received ART and is the only subject among these vaccine trial study participants with breakthrough infection who is controlling viral replication in the absence of ART (Fig. 2A). Three of the four HIV-specific responses identified in this individual postseroconversion were present before seroconversion, albeit at very low frequencies (below 150 SFC/10⁶ PBMC). These T-cell responses were boosted after seroconversion (Fig. 2B), suggesting that they could represent memory responses that were induced upon immunization. However, a Nef AVDLSHFLK-specific T-cell response was detected prior to seroconversion, despite the fact that Nef was not present in the vaccine the subject received. T cells specific for this epitope were restricted by HLA-A*11, and their presence preseroconversion was confirmed by major histocompatibility complex (MHC)-peptide tetramer binding (0.51% of CD3⁺ CD8⁺ T cells [data not shown]). We looked for the presence of proviral DNA at the last immunization date to determine whether this individual was already infected when the HIV-specific T cells were first detected. Using limiting-dilution PCR as previously described (29), we confirmed that this individual was indeed infected at a very low level (1.32 ± 0.66 copies per million PBMC), despite being HIV-1 seronegative by Western blotting and negative by standard plasma viral RNA assays. No provirus was detected using this method 8 months prior to the last immunization date, although the number of cells that were available for assay was limited (T. Zhu et al., unpublished data).

FIG. 2.

A. Longitudinal plasma viral load (RNA copies/ml) and CD4 counts (cells/μl) in individual 014-004. B. HIV-specific T-cell responses, measured by IFN-γ ELISpot, before and after seroconversion in 014-004.

IFN-γ-secreting T cells specific for the well-characterized HLA-A2-restricted SLYNTVATL (SL9) peptide were detected in volunteer 014A-004 by the IFN-γ ELISpot assay (Table 2). PBMC from this individual obtained at postseroconversion time points, however, failed to bind the HLA-A2/SL9 peptide tetramer. Of note, T cells from these same time points had substantial reactivity to the SL9 peptide by IFN-γ ELISpot (Fig. 2B and data not shown). Further characterization of the optimal epitope recognized by these T cells revealed it to be an 8-mer, LYNTVATL (LL8) (Fig. 3A), which is a single amino acid short of the A2-restricted SL9 epitope. The newly defined LL8 epitope was restricted by HLA-C14 (Fig. 3B), not previously shown to restrict HIV-1 epitopes, and T cells recognizing this epitope were of high avidity (50% effective concentration, 0.26 nM) (Fig. 3C).

FIG. 3.

A. Characterization of the optimal epitope, LL8, recognized by T cells from 014-004. B. Identification of HLA-C14 as the restricting MHC molecule for LL8-specific T cells in 014-004. C. Peptide titrations of the Gag-LL8 epitope. Standard IFN-γ ELISpot was performed with the indicated peptides at twofold dilutions ranging from 2 μg/ml to 5.12 × 10⁻⁶ μg/ml. The regression curve was drawn with the sigmoidal fit tool in the Origin 6.0 software. Data shown are background subtracted.

In this subject who was controlling HIV-1 infection, we examined additional antiviral functions of the epitope-specific CD8⁺ T cells defined at day 46 postinfection and at three additional time points over 5 years after infection (Fig. 4). Similar quantities of CD8⁺ T cells secreted TNF-α and IFN-γ when pulsed with their cognate peptide (Gag p17 LYNTVATL, Gag p2p7p1p6 CTERQANFL, RT TAFTIPSI, and Nef AVDLSHFLK), whereas IL-2-secreting cells were either absent or just at the threshold level of detection by the ex vivo assay over the course of infection. At all time points, the majority of this subject's HIV-specific T cells were able to degranulate as measured by CD107a expression and proliferate as measured by CFSE^lo staining (Fig. 4E). Of note, ex vivo IL-2 expression of epitope-specific CD8⁺ T cells did not correlate with their ability to proliferate over 5 days of peptide stimulation. Most notably, the relative magnitudes of IFN-γ-secreting CD8⁺ T cells did change over time, with the numbers of those recognizing LL8, AK9, and TI8 diminishing while the CL9-specific cells increased, as observed prior to seroconversion.

FIG. 4.

A. Gating scheme for ex vivo intracellular cytokine staining assay. B. Representative example of ex vivo intracellular cytokine and CD107a expression data. PBMC were stimulated for 6 h with no antigen, LL8 peptide, or staphylococcal enterotoxin B (SEB). Data are gated on CD3⁺ CD8⁺ live lymphocytes. C. Postinfection cytokine secretion and degranulation (as measured by CD107a expression) profiles of epitope-specific T-cell responses assessed longitudinally in 014-004. Background-subtracted percent secretion of IFN-γ, TNF-α, and IL-2 and percent expression of CD107a in CD8⁺ T cells are shown for each epitope at 409, 751, and 1,838 dpi. Only CD8⁺ T-cell responses that were 2 times the background and at least 0.05% are reported. D. Representative example of CFSE proliferation data. Each row shows data from cultures stimulated for 5 days with no antigen or with either LL8 peptide or anti-CD3/anti-CD28 monoclonal antibody. Cultures were split into three aliquots and restimulated for 6 h with no antigen, LL8 peptide, or SEB before staining as for the ex vivo intracellular cytokine staining assay. Data are gated on CD3⁺ CD8⁺ live lymphocytes. E. Postinfection proliferative ability (measured as percentage of CD8⁺ T cells that are CFSE^lo) of epitope-specific T cells measured longitudinally in 014-004. The percentage of proliferating cells that secrete IFN-γ, TNF-α, or IL-2 and the percentage that express CD107a in response to cognate peptide are also shown.

DISCUSSION

The new generations of candidate HIV vaccines currently being evaluated in clinical trials aim primarily to induce HIV-specific CD8⁺ T-cell-mediated immunity. Specific T cells can only recognize HIV once the virus has infected cells or through cross-presentation (22). These vaccines are, therefore, unlikely to prevent infection itself but may aid in the efficient elimination of virus-infected cells, thus empowering the individual to control the virus. Although a recent study detailed responses from a single vaccinated individual pre- and postinfection (6), this study details the first thorough evaluation of pre- and postinfection HIV-specific T-cell responses in 16 participants in phase I/II canarypox prime-boost vaccine trials, 13 of whom received active immunogen.

We found no significant evidence of anamnestic responses to the vaccine epitopes versus the virus with which they were infected; however, since these vaccines have low immunogenicity and therefore have not moved forward into HIV Vaccine Trials Network (HVTN) phase IIb trials, this is perhaps not surprising. All of the individuals who received active immunogen were nonresponsive to the immunization regimen by our detection system, and their inability to mount a vaccine response may explain why no anamnestic response was observed.

We identified one participant who controlled his infection without antiretroviral therapy for 5 years. This control could be related to his HLA type, and this finding highlights the importance of comprehensive study of breakthrough infections in vaccine trials to determine whether it is the host genetics, characteristics of the initial transmitted virus, and/or immune responses which are controlling viral replication.

Eleven of the 13 individuals who received the vaccine had taken part in trials of canarypox prime-boost regimens where the net cumulative CTL response rates were between 33 and 50%, as assessed by the chromium release assay (4, 18). The other two vaccine recipients had taken part in a phase II trial where the net cumulative CTL response rate, as assessed by IFN-γ ELISpot, was only 9%. Sera from 9 of the 16 study participants had been assessed for neutralizing antibodies during the vaccination series, and of these, 6 had detectable antibody titers against the vaccine strain (neutralization titers ranged from 12 to 1,941 [data not shown]). No vaccine-induced HIV-specific CTL were detected using the ELISpot assay (20). It is possible that vaccine-induced responses were present at frequencies that were too low to detect by this assay, but the fact that the mean frequency of responses recognizing epitopes within the vaccine was no greater than that outside the vaccine suggests that there were no recall responses present prior to infection. Since none of the individuals in this study possessed detectable CTL prior to infection, it is interesting to speculate that the lack of infection in individuals with CTL responses is due to their HIV-specific vaccine-induced T-cell responses. However, the HIV-1 seroincidence rate among vaccine recipients was comparable to that in the HIVNET Vaccine Preparedness Study (1.38 per 100 person-years) (20, 28); due to the small sample sizes and differences in immunogens, reliable estimates of efficacy for these canarypox vaccines in reducing HIV-1 acquisition cannot be derived.

That the postinfection T-cell responses were similar in specificity, breadth, and magnitude between the acute infection group and the vaccine recipients indicates that an individual's ability to raise HIV-specific T-cell immunity during a viral challenge is not affected by having seen HIV in the context of vaccination. However, this may be because the vaccines in this study did not induce strong responses. It remains to be determined whether infection-induced responses will be disrupted by more immunogenic vaccines. Of note, Nef and Gag were recognized most frequently in the vaccinated individuals, which is in agreement with recent data from other primary infection cohorts, as is the fact that all proteins were targeted by HIV-specific T cells (1, 9). Differences between the specific responses observed in our primary infection group and those described in this study may be explained by the average time point for PBMC collection, which was 113 days postinfection (dpi) in our study compared to 30 dpi for the acute infection cohort. Furthermore, the numbers of individuals characterized for T-cell responses at the epitopic level against all HIV proteins during acute infection (1, 2, 9) are far more limited than for those characterized during chronic infection. This, together with the diversity of MHC haplotypes among individuals, means that the database of the complete repertoire of epitopes recognized during primary infection is not sufficiently robust to make distinctions between epitopes recognized during acute versus chronic infection at this time. Of the 44 epitopes recognized, 31 have been previously defined as restricted by 21 different HLA molecules. The number of epitopes/alleles was too small to be able to determine if any single epitope/HLA allele was overrepresented in these breakthrough infections.

014A-004 is the only individual who is controlling viral replication in the absence of ART. He never reported signs and symptoms consistent with an acute retroviral syndrome and has, to date, maintained extraordinarily low viral loads after infection. He possessed no neutralizing antibody preseroconversion (reciprocal titer giving 90% inhibition was 12 at day 46 before infection) and low levels soon after seroconversion (reciprocal titer giving 90% inhibition ranged from 107 to 117 between 38 and 87 days after infection). This individual has been tested for a number of genes that might explain his ability to control viral replication. His MHC haplotype is HLA-A2/A11/B51/B61/C2/C14. Interestingly, HLA-C14 has been associated with nonprogression (19), and this individual has a new HLA-C14-restricted response that has not been previously described. This response is of high avidity, which may explain his ability to control viral replication, since high-avidity CTL responses have been shown to be more protective than low-avidity responses (13). The LYNTVATL (LL8), HLA-C14-restricted epitope is a single amino acid short of SLYNTVATL (SL9), which has been described as a very common HLA-A2-restricted epitope recognized by T cells from chronically infected individuals. SL9 is not normally seen during acute infection (16) and, indeed, although this individual possesses HLA-A2, he does not recognize SL9 during acute infection. Not only is the LL8 epitope novel, but also, to our knowledge, this is the first description of any HLA-C14-restricted HIV epitope. Since HLA-A2 is present at a very high frequency (25%) in the Caucasian population, it is interesting to speculate that some LL8/C14-restricted responses may have been erroneously attributed to SL9/A2 if these persons possess both HLA-A2 and HLA-C14.

Notably, HIV-specific T cells present in this individual were able to proliferate at all time points measured. Maintenance of proliferative ability in HIV-specific CD8⁺ T cells has been shown to be associated with nonprogressive disease (24). Other than the presence of HLA-C14, this individual does not possess other known receptor polymorphisms or mutations in CCR5, CCR2, SDF-1, or DC-SIGN that would explain his ability to control viral replication. He is heterozygous for the CCR5 promoter P1 haplotype (21). It is possible that his ability to control viral replication and his delay in time to seroconversion were related to vaccine-induced immune responses, although there are insufficient immunologic data at present to substantiate this. Alternatively, this individual may have been infected with a less fit viral strain. Further studies are under way to examine this possibility.

Immunological and virological data from another vaccinated individual (202-T07) who received the same vaccine as 014A-004 but also became HIV infected have recently been reported (6). This individual possessed an HLA-B27-restricted response induced by vaccination. The authors inferred that he also possessed at least two other Gag-specific T-cell responses induced by vaccination, although there were insufficient cells to characterize these at the epitopic level. The authors postulated that this individual may have progressed to disease more quickly after vaccination than would be predicted for an HLA-B27⁺ individual. 014A-004 differs in a few respects from 202-T07 in that he possessed T-cell responses to Gag, Pol, and Nef prior to seroconversion, whereas 202-T07 possessed only Gag-specific responses. In addition, since 014A-004 was actually infected at a low level, it is likely that antigen was more persistent in this individual than in 202-T07. This raises the intriguing possibility that low-level persistence of antigen may be important for inducing immune responses that are capable of controlling viremia.

In conclusion, these data provide the first comprehensive analysis of the HIV-specific T-cell response in infected individuals who have previously been inoculated with vaccines designed to induce CD8⁺ T-cell immunity against HIV. Of note, we show that an individual's ability to mount an HIV-specific T-cell immune response is not compromised by HIV vaccination. Although the canarypox prime-boost vaccine strategies, as tested in the initial phase I/II trials, did not progress into HVTN efficacy trials, new vaccines that induce higher frequencies of CD8⁺ CTL are currently being evaluated. It will be important to perform similar studies as those described here on responses from breakthrough individuals to determine whether vaccine-induced HIV-specific T-cell responses will aid in control of HIV infection.