In a mixed subtype epidemic, the HIV-1 Gag-specific T-cell response is biased towards the infecting subtype

Objectives: Southwest Tanzania is affected by an HIV-1 epidemic consisting of subtypes A, C, and D, and their recombinant forms. This study was designed to assess whether the Gag- and Nef-specific T-cell response is biased towards recognizing the infecting subtype.

Methods: The infecting subtypes were characterized with a Multi-hybridization assay that discriminates between subtypes A, C and D. The interferon-γ ELISPOT assay was used to detect the Gag- and Nef-specific T-cell responses in freshly isolated peripheral blood mononuclear cells in 56 seropositive patients. To study the HIV-specific T-cell responses, isolate-based Gag and Nef peptide sets representative of the locally occurring subtypes were used. The results were analysed at the total protein and single peptide level.

Results: In the study population, 35% were infected with a pure C subtype, 24% and 23% with ACD or AC recombinant forms, respectively. The total magnitude (P < 0.01) and breadth (P < 0.01) of the Gag-specific T-cell response detected with the subtype C-Gag peptide set was significantly greater than that detected with either the subtype A-Gag or D-Gag peptide sets. No significant difference was observed in the Nef-specific response. In 85% of responses targeting the most immunodominant Gag epitopes with subtype-specific sequence differences, the best recognized epitope variant corresponded to the infecting subtype.

Conclusions: The Gag-specific T-cell response had a preference for recognizing peptides related to the infecting subtype.

Introduction

The fine mapping of T-cell epitopes and the identification of immunodominant regions of HIV-1 proteins is integral to vaccine design and the optimization of assays for assessing vaccine immunogenicity. CD8 T cells constitute a major component of the cellular arm of the immune response and have been directly implicated in the control of the viraemia of primary HIV-1 infection, the establishment of long-term AIDS-free survival, and the steady-state viral load ^[1-3]. Hence, many of the current vaccine strategies are based, at least in part, on the induction of HIV-specific CD8 T cells ^[4]. Technology to rapidly and accurately characterize CD8 T-cell responses in HIV-1 seropositive individuals has improved dramatically in recent years. Detection of T-cell immune responses by measuring interferon-gamma (IFNγ) release with enzyme linked immunospot (ELISPOT) assay or by flow cytometry, when combined with overlapping pooled peptide technology, has allowed efficient quantification and mapping of HIV-1-specific CD8 T-cell responses ^[5-8]. In particular, responses targeting the Gag protein are of special interest as these have regularly been associated with control of HIV replication ^[9-12].

The genetic variability of HIV-1 is a major obstacle for both vaccine design and an impediment to immunological screening methods ^[13-15]. At least 9 subtypes and 16 circulating recombinant forms are currently described (http://www.hiv.lanl.gov/content/hivdb/CRFs/CRFs.html). As a further complication, intra-subtype variability can account for protein differences of up to 25%, depending on the protein considered. Even within an HIV-1 infected individual, the HIV mutant spectrum, or 'quasispecies' can reach a divergence of 10% ^[16,17]. However, the extent to which this genetic diversity impacts T-cell epitope recognition is not completely understood.

In East Africa subtypes A, C and D and their recombinant forms co-circulate, with a geographic asymmetry in the distribution of these subtypes: subtype A predominates in Kenya, subtype C in Tanzania and subtype D in Uganda ^[18-21]. In order to determine whether the predominance of subtype C in Southwest Tanzania is associated with a subtype C biased pattern of T-cell recognition on a population level, we have examined the infecting HIV strains and the HIV-specific T-cell responses in 56 seropositive individuals belonging to a high-risk cohort. Using primary isolate based Gag and Nef overlapping peptide sets, representative of the locally occurring HIV subtypes A, C and D, we were able to analyse the HIV-specific T-cell responses with respect to the magnitude and breadth of subtype-specific recognition.

Methods

Study subjects

The fifty-six individuals of this study are part of a larger high-risk HIV cohort of female bar workers enrolled for a prospective study of HIV-1 super-infection in the Mbeya region of Southwest Tanzania. Between September and December 2000, 600 women were recruited after giving informed consent and each participant provided blood samples at enrolment and every 3 months for a period of up to 4 years. During the study, all participants received health care that included treatment of all acute infectious diseases, screening and treatment of sexually transmitted diseases, and since 2003, cotrimoxazole prophylaxis for opportunistic infections for women with CD4 T-cell counts < 200/μl. Since 2005 antiretroviral treatment is available for all participants with AIDS defining symptoms or CD4 cell counts below < 200/μl. During the course of this study all individuals were anti-retroviral naive. HIV-1 status was determined using two diagnostic HIV ELISA tests (Enzygnost Anti HIV1/2 Plus, Dade Behring, Liederbach, Germany; and Determine HIV 1/2, Abbott, Wiesbaden, Germany). Discordant results were resolved using a Western Blot assay (Genelabs Diagnostics, Geneva, Switzerland). In April 2003, 56 of the HIV-1 positive participants were enrolled into the HISIS-CTL sub-study. The study was reviewed and approved by the ethic committees of all partners in compliance with national guidelines and institutional policies.

Synthetic peptides and peptide matrices

The six sets of overlapping peptides consisted of 15-mer peptides overlapping by 11 amino acids covering the entire Gag and Nef protein sequence of primary isolates 90CF402 (subtype A-Gag, AAB38823), DU422 (subtype C-Gag, CAD62240) 98UG57143 (subtype D-Gag, AF484514), 92UG037 (subtype A-Nef, AAC97549), DU151 (subtype C-Nef, AAL05314) and 94UG114 (subtype D-Nef, AAC97574). Peptides were synthesized using Fmoc chemistry and standard solid-phase techniques with free amino termini. All peptides were > 80% pure as determined by HPLC, mass spectrophotometry, amino acid analysis and N-terminal sequencing. Peptides were synthesized at the Natural and Medical Sciences Institute (University of Tuebingen, Germany), the Henry M. Jackson Foundation (Rockville, Maryland, USA), and by Anaspec Incorporated (San Jose, California, USA). The Gag peptide sets were closely related to HIV-1 isolates from the Mbeya Region (data not shown). Initial screening for T-cell responses was performed using the peptides in a matrix format. Subtype A-, C-, and D-Gag peptide sets were pooled in an 11 × 11 format, the A- and D-Nef peptide sets were pooled in a 7 × 7 format and the C-nef peptide set was pooled in 10 × 5 format.

ELISPOT assays

Freshly isolated peripheral blood mononuclear cells (PBMC) were screened for HIV-specific T-cell responses by stimulation with overlapping peptide pools representing Gag and Nef from isolates of subtypes A, C and D. ELISPOT assays were performed as previously described ^[23]. Confirmation of individual peptide responses was performed 3 months after the initial screening with peptide matrices. Responses with at least 70 SFC/1 × 10⁶ PBMC and at least three times the negative control were scored as positive.

Genetic characterization of the infecting HIV subtype

The infecting HIV-1 subtype was determined by the Multi-region Hybridization Assay (MHA_acd), a method that can distinguish the HIV subtypes A, C and D in five genome regions (gag, gag/pol, vpu, env and gp41) using subtype-specific florescent probes in a real-time PCR format. The detailed methodology is described elsewhere ^[21]. In previous studies, this method has proven a high accuracy in determining the subtype in discriminating pure from mixed or recombinant subtypes as well as dually infected individuals ^[20-22].

Results

HIV-1 subtypes

Previous epidemiological surveys of the HIV-1 epidemic in the Mbeya region, and the HISIS cohort in particular, have demonstrated the predominance of subtype C and subtype A/C containing recombinant forms ^[20,21]. Molecular characterization of the infecting HIV-1 strains of the 56 individuals from the HISIS-CTL cohort with the Multi-Hybridization assay revealed that 35% of these were infected with a pure C subtype, 24% and 23% with ACD or AC recombinant forms respectively. Another 6% were infected with a CD recombinant form. Subjects infected with a pure Subtype A accounted only for 6%. For Gag, 61% of infecting strains were typed to be pure subtype C, 19% AC and 3.5%CD, while only 14% and 2% were typed to be a pure subtype A or D respectively (data not shown).

Magnitude of immune responses

A peptide matrix-based IFNγ Elispot assay was used to screen for HIV-1 Gag- and Nef-specific T-cell responses. To accurately analyse T-cell responses in this mixed subtype epidemic, isolate-based peptide sets, representing subtypes A, C and D were used. Ninety-six percent (n = 54) of the HISIS-CTL cohort responded to at least one Gag peptide, while 93% (n = 53) responded to at least one Nef peptide; Fig. 1a and b summarize the magnitude of T-cell responses against subtype A, C and D Gag and Nef peptide sets respectively. A significantly greater magnitude of the Gag-specific T-cell response was detected with the subtype C-Gag peptides [median = 1190 spot forming units (SFU)/1 × 10⁶ PBMC] compared with either the A-Gag (median = 480 SFU/1 × 10⁶ PBMC) or D-Gag (median = 380 SFU/1 × 10⁶ PBMC) peptides. In contrast there was no difference in the magnitude of the Nef-specific T-cell responses between the subtype A, C and D peptide sets.

Fig. 1

Breadth of immune responses

Next we analysed the breadth of subtype A-, C- and D-Gag peptide pool responses (Fig. 2a). Each peptide set was subdivided into 11 pools containing 11 consecutive peptides. The breadth of detected Gag-specific T-cell responses was significantly greater with the C-Gag peptide set compared to the A- and D-Gag peptide set (P = 0.0003, Kruskal-Wallis test) with a median of three pool responses detected with the C-Gag peptides and a median of two and one pool responses detected with the D- and A-Gag peptides, respectively. Epitope responses that were detected in two consecutive peptide pools [such as responses against the epitope TPQDLNTML included in both peptides C-gag44 (pool 4) and C-gag45 (pool 5)] were counted as one pool response for this analysis. Further analysis underscores the superiority of the subtype C-Gag peptides for detecting T-cell responses in this cohort (Fig. 2b). Of the 180 Gag-specific responses detected, 149 (83%) were detected with the C-Gag peptides, 103 (57%) with the D-Gag peptides and 91 (51%) with the A-Gag peptides. Subtype A- and D-peptide sets detected 31 responses not detected by the C-peptides. Of note, 56 (31%) of the responses were detected with all three peptide sets and may represent broadly cross-reactive T-cell responses. These results indicate that in this mixed subtype exposed cohort both the magnitude and breadth of T-cell responses against the Gag protein are biased towards the dominant circulating subtype.

Fig. 2

Hot spots of immune recognition

A composite frequency distribution of the response to each individual peptide from both Gag and Nef was constructed (Fig. 3a and Nef data not shown). The p24 protein of Gag was targeted with the highest frequency (91% of subjects), followed by p17 (56%) and the p15 proteins (15%). The breadth of Gag recognition for each individual was a median of three epitopes (a single peptide or pair of adjacent overlapping peptides), with a range of 0-10 epitopes (Fig. 3b). As reported by others ^{[6,7,11,23,24]}, regions with hot spots of immune recognition within the gag gene product were detected: amino acids 10-47 (45% of subjects); amino acids 143-197 (64%) and amino acids 291-321 (46%). The vast majority of responses against Nef were detected in the central conserved region of the protein, which is in agreement with previous findings ^[6,7,11,24] with two hot spots at Nef-amino acid 78-108 (78% of subjects) and amino acids 125-159 (71%). Interestingly, Nef-specific responses from many individuals were highly focused on these two protein regions. Hot spots of immune recognition always contained epitopes presented by HLA class I alleles frequently expressed within the studied cohort (Table 1) and hot spots especially within Nef and Gag p24 were relatively well conserved.

Fig. 3

Table 1

Epitope recognition-point mutation versus frame of epitope effect

To determine the reason(s) for disparate magnitude of cross-recognition among the different peptide sets, the peptides responsible for the nine most frequently detected Gag-specific responses (cut-off value 12.5%) were analysed for differences in amino acid composition and alignment (Fig. 3c). While differences in the amino acid sequence between subtype-specific peptide variants were the most common reason for disparate peptide recognition by T cells, in some cases the position of a minimal T-cell epitope within a 15-mer peptide also had a profound effect on peptide recognition by T cells ('frame-of-epitope' effect; the effect of the position of a minimal 9-mer or 10-mer epitope within a 15-mer peptide on recognition by T cells), a finding supported by two recent studies ^[25,26]. The median magnitude of individual peptide responses to the subtype A-, C- and D- peptide variants of the nine most frequently detected Gag-specific peptide responses and the underlying reason(s) for the difference in recognition of the three subtype variants is shown in Fig. 3c. Recognition of subtype-specific peptide variants was equal only when the amino acid sequence and the epitopes position within the peptide was identical (Fig. 3c, Gag36/37 of subtype A and D). In seven of the nine peptide responses, differences in peptide variant recognition were associated with sequence variation of the targeted epitope sequence. For six of these seven peptide responses, it was the subtype C-consensus epitope sequence that was recognized best (Table 1), among these were the three most immunodominant peptides Gag45, Gag74 and Gag7 (Table 1). In only one of these seven peptide responses (Gag5), it was the subtype A and D epitope consensus sequence that was recognized best-predominantly by subjects infected with an subtype A Gag-bearing strain (Fig. 4). In two of the nine peptide responses the frame-of-epitope effect prevented conclusions about the effect of epitope point mutations on T cell recognition (Fig. 3c, Gag35 and Gag36/37).

Fig. 4

Within Gag p24, the effect of epitope sequence variation on T-cell recognition could only be resolved for subtype A and D peptide variants, because of a three amino acid deletion at the beginning of Gag p24 from primary isolate Du422 (C-Gag peptide set). Despite this obstacle, we were able to deduce the effect of subtype C-specific sequence variation on T-cell recognition in four of the six peptide responses. For three of these responses the subtype C epitope variants were identical to those included in peptides D-Gag45 (TPQDLNTML), A-Gag72 (assumed B8101 epitope: EPFRDYVDRF), A-Gag74 (YVDRFFKTL). The epitope sequence within peptide D-Gag88 (GPSHKARVL) is not included in the subtype C Gag peptide sequence, but is identical with the subtype C consensus sequence (Table 1). Therefore in all comparable immunodominant peptides, except for recognition of the peptide Gag5, it was always the epitope variant related to subtype C that was recognized best, indicating that the Gag-specific T-cell response is biased towards the predominant subtype C in this cohort.

Epitope recognition versus HIV subtype

To examine whether the preferential recognition of subtype-specific epitope variants is associated with the gag genotype of the infecting HIV strain, we compared the MHA gag results of the infecting HIV strain with the recognition of subtype-specific epitope variants for subjects responding to the nine most frequently recognized epitopes (Table 1 and Fig. 4). To exclude the above described 'frame-of-epitope effect' we compared only epitope variants of identical position within the peptides and therefore had to exclude responses to Gag35 and Gag36/37 (Table 1, Fig. 3c). Responses to Gag20 were excluded, because the targeted epitope sequence lacks subtype-specificity (Table 1). The remaining immunodominant Gag epitopes within peptides Gag5, 7, 45, 72, 74 and 88 were recognized 55 times. In 47 of 55 responses (85%), the best recognized epitope variant corresponded to the infecting subtype, whereas in only 3 of 55 responses (5%) there was no association of the best recognized epitope variant with the infecting subtype (Fig. 4, circles). In 5 of 55 responses (9.1%) no clear attribution could be made because the infecting strain was expressing a recombinant form of Gag.

Discussion

The question whether T cells are capable of cross-recognizing genetically diverse virus isolates representing multiple subtypes is key for HIV vaccine development and has been addressed in several studies ^[28,29]. It would seem rational to assume that the subtype-specificity of the T-cell response would be biased towards the subtype of infecting virus, since the amino acid content of viral proteins is in part a function of genetic subtype, and T-cell specificity is dependent upon the amino acid sequence of the targeted epitopes. However, most studies of HIV-1-specific T-cell responses have failed to reveal a general subtype-specificity of the response.

The data presented here shows that the Gag-specific T-cell response of HIV infected individuals exposed to a mixed-subtype HIV-1 epidemic has a high preference for the subtype an individual is infected with. In this cohort subtype C and C-containing recombinant forms of HIV were responsible for most infections and the subtype C-gag peptide set was superior to the A- or D-Gag peptides for detecting T-cell responses in terms of both magnitude (more than 2.4-fold greater median) and breadth (at least 40% more responses detected). In contrast to Gag, there was a lack of preferential recognition of subtype C Nef peptides. This may be explained in part by the complete conservation among the three subtypes within the most frequently targeted epitopes within peptides Nef17/18 (amino acids 65-82) and Nef33/34 (amino acids 129-146).

In order to exclude other reasons for the superiority of the subtype C Gag peptides, we carefully analysed the nine most frequently detected peptide responses for the previously described 'frame-of-epitope' effect, in which the position of a minimal epitope (8-11mer) within a 15-mer peptide can have a profound effect on the magnitude of the T-cell response detected ^[25,26]. In our study the C-Gag peptide set was offset relative to both the A-Gag and D-Gag peptide sets due to a three amino acid deletion at the beginning of Gag p24. While several examples of disparate peptide recognition could be attributed to offsetting of the peptides in the absence of amino acid difference, careful examination of the data clearly indicates that the subtype C related epitope sequences are the most superior (Table 1).

Specifically for the two immunodominant peptide responses (Gag35 and Gag36/37) the effect of amino acid point mutations could not be resolved and these were therefore excluded from our analysis. We also excluded responses to peptide Gag20, because the targeted epitope sequence (SLYNTVATL) lacks subtype-specificity for subtype C (Table 1). In our study population the subtype C peptide variant of Gag20 was recognized best by individuals infected with either subtypes A, C or D. For the remaining six of nine most immunodominant peptides within Gag, a strong correlation (85% of cases) between the infecting subtype and the corresponding subtype-specific epitope response was observed. In conjunction with the fact that more than 80% of Gag subtypes were either subtype C or a C containing recombinant form, the preferential recognition of subtype C related epitope sequences can be attributed to subtype-specific sequence differences within the targeted epitopes.

The results presented here stand in contrast to most other studies that show little or no subtype-specificity of T-cell recognition of HIV-1. Several reasons may explain this discrepancy. Most other studies have focused on comparing the response to full-length HIV-1 proteins either expressed in cells infected with vaccinia constructs ^[30-38] or as bulk pools of synthetic peptides ^[36,39], neither of these approaches has allowed deduction of T-cell recognition of individual variant epitopes. Other studies have used minimal peptides assessing a limited number of epitopes, and thus have not captured the complete breadth of the T-cell response against a given protein ^[28,29]. Using Gag and Nef derived overlapping peptides from isolates of multiple subtypes we have been able to assess T-cell responses at the individual peptide level across two highly immunogenic proteins. This approach allows a much more precise discrimination of a potential subtype-specific immune recognition. Our data are also supported by three recent studies that have implied that the subtype specificity of the T-cell response may be biased towards the infecting subtype ^[40-42].

In concordance with other studies particular regions within Gag and Nef were recognized by a high frequency of study subjects ^[6,7,11,23]. Nonetheless, the pattern of immunodominant regions within Gag and Nef is comparable not only across disparate HLA backgrounds, but also among HIV-1 subtypes. We show that for many of the immunodominant Gag epitopes with subtype-specific sequence differences, recognition is linked to the infecting subtype and that therefore the preferential recognition of subtype C related epitope variants appears to be linked to the predominance of subtype C in this mixed-subtype epidemic. The optimal HIV-1 vaccine may therefore not be one based on a single subtype, but either one based on multiple subtypes or one that could contain epitope variants from multiple subtypes that generate the most broadly cross-reactive response.

Acknowledgements

The authors thank the excellent staff at the Mbeya Medical Research Programme that conducted the HISIS study, especially Vera Kleinfeldt, Frowin Nichombe, Weston Assisya and Clemence Konkamkula, and all of the HISIS participants. Furthermore, we thank Dr Francine McCutchan of the provision of some of genetic subtype data through MHA analysis and Dr Patricia D'Souza (Division of AIDS, NIH), Dr Christian Brander (Massachusetts General Hospital), Steve Cate and Dr William Hildebrand (University of Oklahoma Health Science Center) for facilitating the HLA typing of the specimens.

Sponsorship: This work was supported by the European Commission, DG XII, INCO-DC, (grant ICA-CT-2002-10048) and by a cooperative agreement between the Henry M. Jackson Foundation for the Advancement of Military Medicine and the US Department of Defense. The views and opinions expressed herein do not necessarily reflect those of the US Army or the Department of Defense.