High degree of inter-clade cross-reactivity of HIV-1-specific T cell responses at the single peptide level

Objectives: To determine HIV-1-specific T cell responses in clade B infected individuals recognizing the clade A, B and C consensus sequences in order to assess the degree of inter-clade cross-reactivity of these immune responses at the single epitope level.

Methods: HIV-1-specific T cell responses were assessed cross-sectionally in 27 chronically HIV-1-infected individuals from a population infected mainly with clade B viral strains, using an interferon-γ Elispot assay with a total of 1230 overlapping peptides spanning the entire amino acid sequence of the clade A, B and C 2001 consensus sequences.

Results: No significant difference was observed between the total magnitude or breadth of T cell responses recognizing either the clade A, B or C consensus sequences. However, at the single peptide level, 194 T cell responses were identified that recognized only one of the three different clade-specific peptide variants (A: B: C, 34: 105: 55), 125 T cell responses recognized two of the three peptide variants (AB: AC: BC, 71: 15: 39) and 166 T cell responses (34%) were cross-reactive with all three different peptide variants. Peptides recognized in all three consensus sequence variants had a significantly lower entropy (P < 0.0001) and a significantly higher inter-clade homology (P < 0.0001).

Conclusions: Viral epitopes within regions of low HIV-1 clade B diversity and high inter-clade homology can be recognized in the clade A, B and C variants and indicate a wide degree of cross-isolate and cross-clade recognition by HIV-1-specific T cells. These regions may therefore be of particular relevance for the design of HIV-1 vaccines.

Introduction

HIV-1-specific CD4 and CD8 T cell responses represent a crucial element of the adaptive immune response against HIV-1 infection ^[1-3]. In acute infection, the first emergence of HIV-1-specific CD8 T cells leads to a dramatic decline of initial peak viremia to viral set point and the rapid selection of viral escape mutations in HIV-1-specific CD8 T cell epitopes ^[4,5]. In addition, strong HIV-1-specific CD4 and CD8 T cell responses have been detected in individuals who spontaneously control HIV-1 replication for prolonged time periods ^[6,7]. Prophylactic and therapeutic HIV-1-specific vaccine candidates aiming at eliciting potent HIV-1-specific T cell responses are therefore increasingly being tested in pre-clinical and clinical trials ^[8].

A major obstacle to the successful design of HIV-1 vaccines is the high genetic variability of HIV-1, which allows the virus to rapidly adapt to immune pressure and leads to an immense heterogeneity of circulating viral strains ^[9]. Thus, a potentially successful and widely applicable HIV-1-specific vaccine will need to stimulate immune responses that cover a broad range of viral variants ^[10]. As a first step in the effort to design such a vaccine, it is important to define the cross-reactivity of HIV-1 specific T-cells. A number of studies have shown that bulk HIV-1-specific T cells clearly have the ability to cross-recognize different HIV-1 strains ^[11,12]. However, as most studies have employed peptide pools or recombinant vaccinia virus constructs expressing selected HIV gene sequences, the breadth of cross-reactive T cell responses has not been defined ^[13-15]. In addition, it is currently not known which viral regions are preferentially recognized by cross-reactive HIV-1-specific T cells.

Clade-specific HIV-1 consensus sequences reflect viral amino acid sequences that are most frequently found in the circulating viral species of a distinct geographic region. Clade B viruses are predominantly found in Western Europe, North America and Australia. In contrast, viral clade A and C species are found mainly in eastern Africa and sub-Saharan Africa, respectively ^[16]. Here, we comprehensively determined HIV-1-specific T cell responses recognizing the clade A, B and C consensus sequences on the single peptide level in individuals from a clade B infected population and assessed the degree of inter-clade cross-reactivity of these immune responses.

Patients and methods

Study population

This study characterized HIV-1-specific CD8 T cell responses in 27 chronically HIV-1 infected individuals. The clinical and demographic characteristics of these study persons were as follows: median age, 40 years (range, 29-51 years); median time since infection, 4 years (range 2-19 years); median HIV-1 RNA, 20 600 copies/ml plasma (range, 119-133 860 copies/ml plasma); median CD4 cell count, 458 × 10⁶ cells/l (range, 213-1489 × 10⁶ cells/l). All individuals were anti-retroviral-therapy naive at the time of study. Study subjects were recruited from the Massachusetts General Hospital and the Fenway Community Heath Care Center in Boston and were considered clade B infected. According to the Los Alamos HIV-1 sequence database, more than 99% of infected individuals in North American population are infected with HIV-1 clade B (http://hiv-web.lanl.gov). The study was approved by the respective institutional review boards and was conducted in accordance with human experimentation guidelines of the Massachusetts General Hospital.

Synthetic HIV-1-peptides

Four-hundred and ten synthetic 15-20-amino acid long peptides, overlapping by 10 amino acids and spanning the entire HIV-1 clade A, B or C 2001 consensus sequence (http://hiv-web.lanl.gov/), respectively, were synthesized at the MGH Peptide Core Facility on an automated peptide synthesizer using Fmoc technology. All peptides were synthesized at the same time and using the same reagents. Except for a few cases of insertion or residue deletions between clades, corresponding peptides from the different consensus sequences were always of the same length and spanned identical regions.

Elispot assays

Elispot assays were carried out as described previously ^[17]. Briefly, peripheral blood mononuclear cells (PBMC) isolated by Ficoll-Hypaque (Sigma, St. Louis, Missouri, USA) density gradient centrifugation were plated in 96-well polyvinylidene plates that had been precoated with 0.5 g/ml of an anti-human interferon-γ monoclonal antibody (Mabtech, Stockholm, Sweden). PBMC were added at a concentration of 100 000 cells/well in a volume of 100 μl of RPMI 1640 medium supplemented with fetal calf serum (10%), Hepes buffer (10 mM), L-glutamine (2 mM) and penicillin-streptomycin (50 U/ml). Corresponding clade A, B and C peptides were combined in pools of three peptides and tested individually when the three-peptide pool gave a positive response. The final concentration of the peptides in every single well was 14 μg/ml. Plates were incubated overnight at 37°C, 5% CO₂ and developed the next day as described elsewhere ^[17]. Wells containing PBMC and medium with phytohemagglutinin or without any peptide were used as positive or negative controls and run in triplicate on each plate. To calculate the number of specific T cells, the number of spots in the negative control wells was subtracted from the counted number of spots in each well. Responses were considered positive if there were > 50 spot-forming cells (SFC)/1 × 10⁶ PBMC after subtracting background and had at least three times the mean number of SFC of the three control wells.

Calculation of peptide variability

A Shannon entropy score was calculated as described previously ^[17] for each position in the 2001 alignment of clade B sequences published in the Los Alamos National Laboratory HIV Database (http://hiv-web.lanl.gov). Entropy is a measure of the amino acid variability at a given position that takes into account both the number of possible amino acids allowed and their frequency. An average entropy score for all positions in each of the 410 overlapping peptides was determined to provide a single value that characterizes the overall variation of each peptide.

Statistical analysis

Results are given as means or medians with ranges. Statistical analysis was based on Student t tests, Wilcoxon rank sum test, or a multiparametric ANOVA test, as appropriate; a P < 0.05 was considered significant.

Results

Previous studies showed that bulk HIV-1-specific T cell responses were detected using a wide variety of heterogeneous viral strains ^{[11,14,15,18]}. However, in most of these studies, T cell stimulation was performed using HIV-1-specific peptide pools or B-cell lines infected with vaccinia viruses expressing selected HIV-1 proteins, which did not allow for determining to what degree cross-reactivity of HIV-1-specific T cells occurs on the single epitope level. Here, we extended these studies and used a total of 1230 single peptides spanning the entire amino acid sequence of the clade A, B and C HIV-1 consensus sequences to comparatively analyze cross-reactive HIV-1-specific T cell responses in 27 individuals from a clade B infected population.

Figure 1a summarizes the total magnitude of T cell responses reactive against the clade A, B or C consensus sequences, respectively in 27 individuals screened. Nine hundred of the 1230 peptides were not recognized by HIV-1-specific T cell responses in any of the study subjects (314 clade A, 276 clade B, 310 clade C). Interestingly, despite the fact that all study individuals were derived from a population with circulating clade B viral strains, no significant differences were seen between the total magnitude of T cells recognizing the clade A, B or C consensus sequence. In addition, the total number of clade A, B or C peptides recognized in this study population was not statistically different (Fig. 1b), although there was a trend for a preferential recognition of clade B peptides as one might have expected. Thus, these data indicate that all three clade-specific consensus sequences can be broadly and strongly recognized by HIV-1-specific T cells in HIV-1 clade B infected individuals.

Fig. 1

We next analyzed the recognition of each of the individual 410 overlapping peptides from the three consensus sequences to determine inter-clade cross-recognition of HIV-1-specific T cells at the single peptide level (total of 1230 peptides). Figure 2a shows data from a representative study individual FW027: overall, consensus clade B based peptides were most frequently recognized in this individual. However, we also found a considerable number of peptides that were recognized in all three different consensus sequences, although the amino acid composition of the corresponding clade A, B and C peptides differed by up to 27.8% (Table 1). Interestingly, we also detected HIV-1-specific T cell responses that recognized peptides derived only from the clade A consensus sequence (peptides number 42, 267 and 274), the clade C consensus sequence (peptide number 260) or both (peptides number 11, 266), but not from the clade B consensus sequence. These data demonstrate that at the single peptide level, T cell responses with different degrees of cross-reactivity to the three clade-specific consensus sequences can be detected, although the total magnitude and breadth of responses cross-reactive with these three clade-specific consensus sequences were not statistically different on the level of the entire HIV-1 proteome. Thus, this study allowed for the classification of immune responses in seven different categories: (i) responses to clade A sequence peptides only; (ii) responses to clade B sequence peptides only; (iii) responses to clade C sequence peptides only; (iv) responses cross-reactive to clade A and B sequence peptides; (v) responses cross-reactive to clade A and C sequence peptides; (vi) responses cross-reactive to clade B and C sequence peptides; (vii) responses to all three consensus sequence peptides.

Fig. 2

Table 1

We subsequently analyzed the HIV-1-specific T cell responses from all 27 study subjects with chronic HIV-1 infection at the single peptide level. The comprehensive characterization of HIV-1-specific T cell responses using all 1230 overlapping peptides spanning the entire expressed HIV-1 clade consensus sequences of clade A, B and C generated 33 210 data points on the recognition of these individual peptides. One-hundred and ninety-four T cell responses were directed against only one of the three clade A, B or C variants, respectively, including 34 responses to the clade A variant peptide only, 105 responses to the clade B variant peptide only and 55 responses to the clade C variant peptide only. One-hundred and twenty-five T cell responses were detected that were cross-reactive to two of the three clade specific variant peptides (AB: AC: BC, 71: 15: 39). Finally, 166 T cell responses were observed to be cross-reactive to all three consensus sequence peptides at the same time (Fig. 2b). Taken together, these data show that in a clade B infected population, a substantial proportion of HIV-1-specific T cells can be identified that are cross-reactive against the clade A, B and C consensus sequence on the single peptide level.

Recent data indicate that viral clade B peptides with low intra-clade entropy, defined as the degree of sequence variability within published clade B sequences, are more likely to be recognized by T cells in HIV-1 infected persons ^[19-21]. To more closely characterize the viral regions that were preferentially cross-recognized by HIV-1-specific T cells, we classified all peptides according to recognition of one, two or three consensus sequence variants and then determined the median clade B Shannon entropy scores of the peptides in these three different groups. As shown in Fig. 3a, we observed a significantly lower entropy of viral peptides recognized in all three consensus sequence variants, compared to peptides exclusively recognized in one or two clade-specific variants. This suggests that peptides that are more conserved within the published clade B sequences are more likely to be cross-recognized between different HIV-1 clades. In addition, the degree of inter-clade homology, defined as the degree of sequence similarity between the different clade A, B and C consensus sequences, was highest in the subset of peptides recognized in all three peptide variants and significantly exceeded the inter-clade peptide homology of peptides recognized in only one or two consensus sequences (Fig. 3b and c). The median inter-clade amino acid variation between peptides recognized in all three consensus sequence variants reached 5.6%, while corresponding peptides recognized in only two or one consensus sequence variants differed by a considerably higher amino acid number between the different clades (Table 2). Taken together, these data show that cross-recognition of viral peptides derived from the clade A, B and C consensus sequences by T cells occurs preferentially in viral regions with low intra-clade entropy and high inter-clade homology. This suggests that regions conserved within clade B are similarly conserved across clades A and C, and are more likely to be recognized by clade B-specific T cells.

Fig. 3

Table 2

Discussion

HIV-1 specific T cell responses are considered a crucial element of protective immunity against HIV-1, and numerous HIV-1 vaccine approaches aim at the induction of HIV-1-specific cellular immune responses. The extraordinary genetic diversity of HIV-1 strains represents a major problem for the development of HIV-1 vaccines that could be used in regions with a wide heterogeneity of circulating viral species ^[9]. One way to overcome this obstacle is to design HIV-1 immunogens able to induce broadly cross-reactive cellular immune responses that can recognize a widespread selection of different viral quasispecies. A number of recent studies have clearly shown that T cell cross-recognition of genetically diverse viral strains is possible. However, in these studies, cross-recognition was either assessed for a limited number of selected epitopes ^[12,22,23] or not on the single epitope level at all ^[11,14,18], but using pools of overlapping peptides or cells infected with vaccinia constructs expressing entire HIV-1 proteins. Here, we extended these studies and analyzed T cell cross-reactivity against the entire clade A, B and C consensus sequence at the single epitope level in 27 individuals from a region with a high clade B prevalence. Our data demonstrate that in these individuals, a significant proportion (34%) of detected HIV-1-specific T cell responses showed cross-reactivity against peptides derived from all three consensus sequences. In addition, broadly cross-reactive T cell responses were preferentially directed against the conserved viral regions with low intra-clade diversity and high inter-clade homology.

In this study, we assessed HIV-1 specific immune responses in 27 individuals from a geographic region with a high prevalence of clade B viral strains (> 99% according to the Los Alamos database). However, the actual clade of the virus that had originally infected these study persons was not determined. Interestingly, in these study subjects, the clade of the infecting virus could not be assumed based on a preferential recognition of one of the three clade-specific consensus sequences, as no difference was found between the total magnitude of T cells recognizing either one of the different consensus sequences tested. Furthermore, the total number of recognized clade B peptides in each individual just slightly exceeded the corresponding number of recognized peptides from the clade A or C sequence. Thus, using a comprehensive approach for the screening of T cell responses, these data show that HIV-1-specific T cell mediated inter-clade cross-recognition is more widespread than previously anticipated. Future studies will be necessary to determine the cross-reactivity of HIV-1-specific T cells in individuals with differential ethnic background and infected with non-clade B viruses, as these viral regions may be of particular value for the design of HIV-1 vaccines. However, all HIV-1-specific T cells responses described in this study occurred during natural infection, and these immune responses can apparently contain viral replication only to a limited extent in most infected individuals ^[19,24]. Future vaccine strategies may therefore rely on eliciting broadly cross-reactive HIV-1-specific T cells with higher anti-viral activity than those observed during the natural disease process.

Our results also demonstrate that within single study individuals, some HIV-1 peptides can be exclusively recognized in the clade B sequence variant, while others were uniquely recognized in the clade A or C sequence variant, likely reflecting the overall sequence diversity within clade B. Moreover, we identified viral regions with low intra-clade diversity and simultaneous high inter-clade homology that were preferentially recognized by T cells among all three different clade-specific consensus sequences. Although the entropy scores of peptides from the clade A and C consensus sequence were not calculated, it is highly likely that the peptides recognized from these sequences also represent very conserved regions of the viral genome. Indeed, we observed a highly significant correlation between the entropy of clade B peptides and the degree of inter-clade sequence homology (A versus B, r = -0.6; P < 0.0001. B versus C, r = -0.56; P < 0.0001). Overall, these data suggest that within the clade A, B and C sequences, corresponding viral regions share a similar degree of viral diversity, possibly due to structural constraints that prevent sequence mutations in specific parts of the viral genome. Some of these cross-recognized viral peptides were completely identical between the clade A, B and C sequence and their cross-recognition is therefore evident. Yet, we also observed cross-clade recognition of peptides with considerable differences in their amino acid composition, suggesting that T cell receptors of HIV-1-specific cells can tolerate some degree of amino acids substitutions in their target epitopes without a total loss of epitope recognition ^[25,26].

Taken together, these data indicate that the majority of detectable HIV-1-specific T cell responses mounted during chronic infection can be cross-reactive against the clade A, B and C consensus sequence. These broadly cross-reactive T cells preferentially recognize conserved viral regions with low intra-clade diversity and high inter-clade homology. These viral regions may be of particular value for the design of broadly applicable HIV-1 vaccines.

Note: Xu G. Yu and Mathias Lichterfeld contributed equally to this work.