CD4 responses to conserved HIV-1 T helper epitopes show both negative and positive associations with virus load in chronically infected subjects

Abstract

Characterization of immune responses to immunodominant CD4 epitopes in HIV-1 that are associated with control of HIV infection could be used to strengthen the efficacy of polyepitope HIV vaccines. We measured both the proliferative and the CD4 interferon (IFN)-γ and interleukin (IL)-2 cytokine responses specific for 11 previously identified HIV-1 T helper epitopes in 10 HIV-infected non-progressors (LTNPs) (infected for a median of 15 years with a stable CD4 count of >500 cells × 10⁶/l), and seven slow progressors (SPs) (infected for a median of 15 years with a CD4 count that had declined to <500 cells × 10⁶/l). Both groups were antiretroviral treatment-naive at the time of evaluation. The median virus load of SP group was higher than that of the LTNP group (P = 0·0002). The CD4 response to a peptide pool representing all potential CD4 Gag epitopes and to Gag p24 protein was also studied. Compared to SPs, LTNPs had higher numbers of Gag-specific IFN-γ⁺IL-2⁺ CD4s (P = 0·0059). The Gag-specific cytokine and proliferative responses correlated inversely with virus load (P = 0·03 and 0·0002, respectively), highlighting the potential importance of this response in immunity to HIV. A direct correlation was noted between proliferation and the Gag-specific IL-2 (P = 0·0053) rather than IFN-γ response (P = 0·1336), demonstrating that the proliferation assay reflected the IL-2 rather than the IFN-γ secreting capacity of CD4 cells. Several subjects with diverse class II DRB1 alleles responded, confirming the 11 selected peptides to be both antigenic and conserved. CD4 cytokine responses to one Gag and two conserved Pol peptides correlated negatively with virus load. The cytokine response to two additional Pol peptides correlated positively with virus load. The data indicate that there is not an absolute correlation between the CD4 immune response to conserved and broadly antigenic helper T cell epitopes in HIV non-progression.

INTRODUCTION

The definition of epitopes towards which the cellular immune response is focused in HIV infection provides vital information on the virus-specific response that is essential for the development of vaccine strategies for the treatment of HIV infection. Numerous T cell epitopes have been documented in HIV-encoded proteins that are recognized by CD8 cells in HIV-1 infected subjects [1–3]. However, much less is known of the characteristics of CD4 epitopes in HIV. This reflects, in part, the difficulty in identifying CD4 T helper responses in HIV-1-infected subjects, due to either dysfunction or deletion of specific CD4 T cells, and also due to the lower frequencies of CD4 T cell responses compared to CD8 T cell responses. However, studies of virus specific CD4 T cell responses in HIV-infected long-term non-progressors (LTNPs) and patients on highly active antiretroviral therapy (HAART) with suppressed viral load have identified a number of epitopes, mainly within Gag p24, to which the CD4 T cell response is directed [4–6].

Previous work on peptides that bind to various HLA Class II molecules has revealed that several different Class II molecules are associated with largely overlapping peptide-binding repertoires and can be grouped into broad supertypes [7]. In keeping with this information, Wilson et al. identified 11 supermotif-containing peptides in HIV-1 Gag and Pol proteins that bind to multiple HLA DRB1 alleles and are highly conserved across various clades of HIV-1. These peptides were shown to be immunogenic in a subset of tested patients, all of whom were on HAART with a median duration of infection of 7 years [8]. However, the importance of CD4 responses to these peptides in the control of HIV infection is currently unknown.

In this paper we correlated CD4 responses to these peptides with non-progression by analysing the specific CD4 cytokine and proliferative responses of a cohort of treatment-naive long-term non-progressors and slow progressors, infected with HIV-1 for a median of 15 or more years. All the patients selected for the study therefore had a relatively benign disease course, but included subjects with a range of virus loads and CD4 count, thereby enabling the immunological correlates associated with the control of infection to be identified. Recent work on such patients from our group has emphasized the potential importance of Gag-specific CD4 responses in control of HIV infection [9]. In particular, we demonstrated that the magnitude of the Gag-specific CD4 response, defined by cells that secreted both interferon (IFN)-γ and interleukin (IL)-2, was higher in non-progressors than in slow progressors and correlated negatively with virus load, positively with CD4 count and positively with the magnitude of the specific CD8 IFN-γ response [9]. These observations are consistent with other emerging evidence highlighting the importance of specific CD4 cells in immunity to HIV [10–12] in terms of governing both the magnitude [13] and possibly both cytolytic [14] and cytokine function [12] of specific CD8 responses. In addition, specific CD4 cells may curtail virus spread through the secretion of antiviral factors [6]. In this paper we confirm that the 11 supermotif peptides containing T helper epitopes identified by Wilson et al. [8] are recognized widely by CD4 cells in chronically infected subjects, but such responses do not necessarily correlate with non-progression.

MATERIALS AND METHODS

Samples were taken from members of a previously well-characterized HIV-infected cohort, established in 1995, of 165 long-term HIV-1 infected volunteers attending clinics in London, UK, who had been enrolled into a nested case–control study of the biological and behavioural correlates of non-progression in HIV-1 infection [15]. In the original cohort 46 subjects were defined as LTNPs based on a stable CD4 T cell count >500 cells/µl at 8 years of infection; 92 as slow progressors based on a CD4 T cell count <500 cells/µl at 8 years of infection, and 27 as rapid progressors because they had developed AIDS within 5 years of infection. We identified a subgroup of six LTNP and six SP patients from the original cohort and an additional four LTNPs and one SP who fulfilled the above criteria for inclusion in this study. At the time of sample evaluation, LTNPs (n = 10) were defined as individuals who had been HIV infected for at least 13 years (median duration of 15·9 years) and had remained asymptomatic with a CD4 cell count >500/µl; and slow progressors (n = 10) as individuals who had been HIV infected for at least 13 years (median duration of 16 years) but whose CD4 cell count had declined to 500 cells/µl at the time of study. The definition of LTNPs and SPs used in this study was thus based on immunological stability at the defined time-point of 16 years following infection which has been adopted widely in other studies [16]. All LTNPs were Caucasian. The SP group consisted of six Caucasians and one black Caribbean. All 17 subjects included in this study were antiretroviral-naive at the time of sample evaluation. In all 17 patients, viral loads and absolute CD4 counts were determined within 3 months of blood sampling, of whom nine were analysed at the same time, three within the first month, four within the second month and one within the third month. The median virus load (VL) in the 10 LTNPs was 2014 copies/ml (range = 50–437 389) and in the seven SPs was 99 000 copies/ml (range = 50 559–365 000) (P = 0·0002). One LTNP had a VL of 437 000 copies/ml, another 103 000 copies/ml but the remaining eight LTNP had viral loads of <10 000 copies/ml with five LTNPs having a VL of <400 copies/ml. Thus, the LTNPs studied had a wide range of VL at evaluation. This differs from some other studies that have incorporated a low VL as an added criterion for defining LTNPs [14]. The median CD4 count (cells/µl) in the LTNPs was 720 (414–2100) and in the SPs was 227 (range 157–379) (P = 0·0001). However, despite these differences, as only SPs who had not yet received ART at 15 years following infection were selected for inclusion, these individuals, despite their progressive disease, represent a eligible group with also a relatively benign disease course, and may not therefore differ substantially from the LTNPs. The study was approved by the local ethics committee (no. 99–075) and written, informed consent was obtained from all subjects.

Monoclonal antibodies and reagents

Anti-IFN-γ-FITC clone B27, anti-IL-2-PE clone MQ1–17H12, CD3-APC clone HIT3a, CD8-cychrome clone RPA-T8 and isotype-matched controls were all purchased from Pharmingen, Oxford, UK.

Intracytoplasmic cytokine staining (ICS) assay

Peripheral blood mononuclear cels (PBMC) were isolated from heparinized blood by density gradient centrifugation on Lymphoprep (Nycomed, Oslo, Norway), within 24 h of venipuncture. Freeze-thawed PBMC (viability > 90%) were plated at 1 × 10⁶ per ml with 0·5 µg/ml each of anti-CD28 and CD49d co-stimulatory antibodies (Becton Dickinson, Cowley, UK) and the following stimuli for 20 h at 37°C. (a) A pool of 122, 15mer HIV-1 Gag peptides that overlapped by 10aa covering the Gag region of HIV-1 HXB2 strain (NIH AIDS Research and Reference Reagent Program, MD, USA) was used at a final concentration of 2 µg/ml for each peptide; (b) 11 15-mer peptides encoding conserved HIV-1 CD4 T cell epitopes (Table 1) identified by Wilson et al. [8] were synthesized by EU Programme EVA/MRC Centralized Facility for AIDS Reagents, NIBSC, UK and tested individually at a final concentration of 10 µg/ml; (c) tetanus toxoid (Pasteur Merieux, MSD, Berks, UK) was used at a final concentration of 100 ng/ml; and (d) staphylococcal enterotoxin B at 1 µg/ml (Sigma, Poole, UK). Protein transport inhibitor, Brefeldin A (Sigma), was added at 5 µg/ml for the last 16 h of the stimulation. Stimulated cells were washed once, permeabilized with BD Permeabilizing solution (Becton Dickinson) and stained as described previously [9] with a cocktail of fluorochrome conjugated antibodies specific for CD3/CD8/IFN-γ/IL-2 for 40 min at 4°C. Stained samples were washed three times in cold PBS + 2% BSA. For each analysis 100 000–400 000 events were acquired using Cellquest software (Becton Dickinson) on a FACSCalibur flow cytometer (Becton Dickinson). The lymphocyte gate was defined manually on forward- and side-scatter characteristics for each sample. CD4 and CD8^- T cells were defined as the CD3⁺CD8⁺ and CD3⁺ CD8^– subsets, respectively, within the live lymphocyte gate; gating CD4 T cells using CD3⁺CD4⁺ expression gave comparable results (not shown). The net percentage of cytokine positive cells within the CD4 and the CD8 compartment was determined by subtracting background staining due to fluorochrome matched isotype control antibodies as well as cytokine production in response to co-stimulatory antibodies. Background frequencies of isotype controls did not exceed 0·10% and frequencies of cells stimulated with co-stimulatory antibodies without antigen did not exceed 0·10%. The mean coefficient of variation for the detected cytokine responses ranged from 10% to 38% (not shown), suggesting good reproducibility of positive responses that are similar to reproducibility found in other work [17]. In six HIV negative healthy controls no IFN-γ or IL-2 responses to the Gag peptide pool were found above the limit of detection (0·00%) of the assay (not shown).

Table 1.

Selected peptides encoding conserved HIV T helper epitopes and patient recognition

Peptide	Region of protein	Amino acid sequence	Parental protein	Patient recognition*	Projected recognition**
Gag1	Gag171–185	QGQMVHQAISPRTLN	Gag p24	71	77
Gag2	Gag294–308	GEIYKRWIILGLNKI	Gag p24	71	90
Gag3	Gag298–312	KRWIILGLNKIVRMY	Gag p24	82	95
Pol1	Pol303–317	FRKYTAFTIPSINNE	Pol reverse transcriptase	82	77
Pol2	Pol335–349	SPAIFQSSMTKILEP	Pol reverse transcriptase	88	89
Pol3	Pol596–610	WEFVNTPPLVKLWYQ	Pol reverse transcriptase	94	90
Pol4	Pol711–725	EKVYLAWVPAHKGIG	Pol reverse transcriptase	71	84
Pol5	Pol712–726	KVYLAWVPAHKGIGG	Pol reverse transcriptase	82	84
Pol6	Pol758–772	HSNWRAMASDFNLPP	Pol integrase	71	77
Pol7	Pol915–929	KTAVQMAVFIHNFKR	Pol integrase	71	84
Pol8	Pol956–970	QKQITKIQNFRVYYR	Pol integrase	88	93

Table shows abbreviations, sequence and parental protein for peptides encoding conserved HIV-1 T helper epitopes, in addition to recognition of peptides by patients.

^*Recognition of peptides based on detection of either IFN-γ or IL-2 positive CD4 T cells by flow cytometry.

^**Projected population coverage determined by considering the phenotypic frequencies of the HLA Class II haplotypes that bind the corresponding peptide, determined by Wilson et al. [8].

Proliferation assay

Freshly isolated PBMCs (2 × 10⁵) were cultured in RPMI-1640 with 10% AB serum in triplicate in 96-well round-bottomed plates (Greiner, Stonehouse, UK) with the following stimuli: (a) baculovirus-derived recombinant HIV-1 Gag p24 protein (Protein Sciences, Meriden, USA) at a final concentration of 5 µg/ml; (b) baculovirus-derived recombinant HIV-1 gp160 protein (Protein Sciences, Meriden, USA) at a final concentration of 5 µg/ml; (c) control, baculovirus protein (Protein Sciences, Meriden, USA) at a final concentration of 1 µg/ml; (d) 11 HIV-1 15mer peptides listed in Table 1 at a final concentration of 10 µg/ml; (e) CMV lysate (Biowhittaker, Wokingham, UK) at 1 : 100 final concentration; (f) tetanus toxoid (Pasteur Merieux) at a final concentration of 100 ng/ml; and (g) PHA (Sigma, Poole, UK) at a final concentration of 5 µg/ml. Cells were cultured in a final volume of 200 µl for 6 days at 37°C and 5% CO₂. For the last 6 h, cells were pulsed with 1 µCi tritiated thymidine (ICN, Irvine, CA, USA) per well and processed through a TopCount NXT^TM Microplate scintillation and luminescence counter (Packard, Pangbourne, Berks, UK). The mean count per minute (cpm) uptake of tritiated thymidine of triplicate wells was determined. Stimulation index (SI) was calculated as the ratio of the mean cpm of the wells containing test antigen divided by the mean cpm of the wells containing medium in the absence of test antigen. To determine p24- and Gp120-protein specific proliferation, background due to control baculovirus protein was used to determine SI rather than the mean cpm of wells containing medium in absence of test antigen.

Statistical analysis

Statistical analysis was performed using instat software and GraphPad Prism (both from GraphPad Software Inc, San Diego, CA, USA). Differences between the LTNP and SP groups were analysed by Mann–Whitney U-tests and Fisher's exact test and two-tailed P-values represented. Correlation of immune response with disease markers was calculated using Spearman's rank test.

RESULTS

Selected supermotif peptides bearing HIV-specific T helper epitopes

The 11 DR-binding supermotif peptides identified by Wilson et al. [8] are shown in Table 1 and include three peptides from Gag p24 protein, designated Gag1, 2 and 3; and eight peptides from Pol, designated Pol1 to Pol8. In order to determine if these peptides are broadly recognized by subjects in our cohort, it was necessary to establish that the HLA DR types of the selected patients were similar to those of an unselected wider Caucasian population [8]. Table 2 shows this to be the case for several HLA DR allele frequencies, except for DRB1*0301 and DRB1*0401 alleles, which were slightly under-represented in our patients.

Table 2.

Phenotypic frequencies of HLA types in study population

		Phenotypic frequency (%) of DR type in group
DR type	Representative allele	European Caucasians*	Patient population	Patient numbers
DR1	DRB1*0101	19·3	22	4
DR2	DRB1*1501 1	27·6	28	5
DR3	DRB1*0301	24·7	11	2
DR4	DRB1*0401,DRB1*0405	29·3	11	2
DR6	DRB1*1302	26·5	17	3
DR7	DRB1*0701	25·5	22	4
DR8	DRB1*0802	5·4	11	2
DR9	DRB1*0901	2·0	6	1
DR11	DRB1*1101	18·2	11	2
DR12	DRB1*1201	3·4	6	1
DR51	DRB5*0101	32·2	n.d.	–
DR52	DRB3*0101	39·7	n.d.	–
DR53	DRB4*0101	59·6	n.d.	–

^*Phenotypic frequencies of alleles in Caucasian population from Wilson et al. [8].

Widely detected CD4 cytokine responses to supermotif peptides in chronic HIV infection

Figure 1 shows the magnitude of the IFN-γ, IL-2 and IFN-γ⁺IL-2⁺ CD4 cytokine responses in all subjects tested to the 11 supermotif peptides. Also shown are responses to total Gag, TT and SEB. For the range of values and significant differences between LTNP and SPs see Tables 3 and 4. A CD4 cytokine response specific to all 11 peptides and total Gag was detected (Fig. 1). The magnitude of the HIV-specific IFN-γ response was generally higher than the corresponding IL-2 and IFN-γ⁺IL-2⁺ response. For example, the median IFN-γ response to the Gag pool was 0·06% compared to a median of 0·04% IL-2 and a median of 0·03% IFN-γ⁺IL-2⁺ response to the same stimulus. These data are consistent with others who have used global stimulation methods [18,19] to show progression to disease to be associated with loss of CD4s that secrete IL-2, but not CD4 cells that secrete IFN-γ in the absence of IL-2. In contrast to the total Gag response, the magnitude of the TT specific CD4 IL-2 response was higher than the analogous IFN-γ response, indicative of a predominantly central memory TT compared to an effector HIV-specific response [20].

Fig. 1.

CD4 cytokine response to total Gag and selected HIV-1 peptides in LTNPs and SPs. Percentages of CD4 T cells staining positive for (a) IFN-γ, (b) IL-2 and (c) simultaneously for IFN-γ and IL-2 following stimulation with selected HIV-1 peptides (10 µg/ml), Gag peptide pool (2 µg/ml), TT (100 ng/ml) and SEB (1 µg/ml) with anti-CD49d and CD28 MoAb (0·5 µg/ml each) for 20 h in 17 patients: 10 LTNPs and seven SPs. Each symbol represents a patient. Horizontal lines represent median values. All values have been corrected for isotype control and background due to co-stimulation without antigen. Numerical data based on responding patients are summarized.

Table 3.

Correlation of cytokine responses with CD4 T cell count and disease stage

	Correlation with CD4 count		LTNPs versus SPs % cytokine positive cells
CD4 T cell response	r-value	P-value	LTNP n = 10	SP n = 7	LTNP versus SP (P-value)
Gag total IL-2+	0·53	0·0203	0·04 (0·00–0·10)	0·00 (0·00–0·04)	0·0204
Gag IFN-γ+IL-2+	0·54	0·0253	0·02 (0·00–0·05)	0·00 (0·00–0·03)	0·0204
Pol3 total IL-2	−0·69	0·0023	0·01 (0·00–0·05)	0·07 (0·01–0·17)	0·0068
Pol8 total IFN-γ+	0·50	0·0417	0·02 (0·00–0·04)	0·00 (0·00–0·05)	0·0431
SEB total IL-2	0·6176	0·0082	1·30 (0·5–4·08)	0·46 (0·06–1·52)	0·0068
SEB IFN-γ+IL-2+	0·5445	0·0238	0·18 (0·03–0·76)	0·00 (0·00–0·13)	0·0031
SEB single IL-2	0·68	0·0028	1·14 (0·47–3·57)	0·46 (0·06–1·48)	0·0046

Percentages of the following antigen specific CD4 subsets: total IFN-γ⁺, total IL-2⁺, IFN-γ⁺IL-2^–, IL-2⁺IFN-γ^– and IFN-γ⁺IL-2⁺ of 17 HIV⁺ subjects were correlated with their CD4 T cell count using Spearman's rank correlation. Analysis included responses to all supermotif peptides. Only significant associations (P < 0·05) and r-values are shown. For each significant correlation the median percentage of cytokine positive cells and the range in brackets in LTNPs and SPs is shown. Difference between LTNPs and SPs was tested by non-parametric statistics. Statistically significant differences between LTNPs and SPs (P = <0·05) are highlighted in bold type.

Table 4.

Correlation of absolute numbers and percentages of specific cytokine producing cells with virus load and disease stage

	Correlation of absolute nos of cells with VL		Correlation of cell frequencies (%) with VL		LTNPs versusSPs (Absolute numbers of cytokine positive cells)
CD4 T cell response	r-value	P-value	r-value	P-value	LTNP n = 10	SP n = 7	(P-value)
Gag IFN-γ+IL-2+	−0·51	0·0308	−0·53	0·0241	17·4 (0·0–42·0)	0·0 (0·0–6·0)	0·0059
Gag1 single IFN-γ+	−0·50	0·0413	n.s.	n.s.	6·1 (0·0–32·4)	0·0 (0·0–11·4)	0·1613
Pol2 single IFN-γ+	−0·48	0·0494	n.s.	n.s.	4·5 (0·0–39·0)	0·0 (0·0–7·6)	0·2295
Pol3 total IL-2+	0·60	0·0103	0·79	0·0002	8·8 (0·0–33·0)	13·7 (9·1–35·9)	0·1613
Pol5 IFN-γ+IL-2+	0·51	0·0354	0·50	0·0395	0·0 (0·0–24·1)	3·8 (0·0–17·2)	0·6009
Pol6 single IFN-γ+	−0·52	0·0340	n.s.	n.s.	4·5 (0·0–24·3)	2·3 (0·0–7·6)	0·4747
SEB total IL-2+	−0·71	0·0014	−0·55	0·0224	986 (401–3178)	116 (9–359)	0·0068
SEB IFN-γ+IL-2+	−0·76	0·0004	−0·71	0·0013	95 (24–678)	0 (0–26)	0·0031
SEB single IL-2+	–0·78	0·0002	–0·56	0·0199	887 (377–2758)	107 (9–349)	0·0046

The absolute number and percentage of the following antigen-specific CD4 subsets: total IFN-γ⁺, total IL-2⁺, IFN-γ⁺IL-2^–, IL-2⁺IFN-γ^–and IFN-γ⁺IL-2⁺of 17 HIV⁺subjects were correlated with their virus load using Spearman's rank correlation. Analysis included responses to all supermotif peptides. Only significant associations (P = <0·05) and r-values are shown. For each significant correlation the median number of cytokine positive cells and the range in brackets in LTNPs and SPs is shown. Difference between LTNPs and SPs was tested by non-parametric statistics. Statistically significant differences between LTNPs and SPs (P < 0·05) are highlighted in bold type.

With the exception of Gag2, greater than 60% of individuals responded to the supermotif peptides by IFN-γ secretion. Greater than 55% of patients had an IL-2 response and with the exception of Gag2 and Pol6 greater than 40% of subjects had a specific IFN-γ⁺ IL-2⁺ response. A total of 71–94% of patients had a peptide specific IFN-γ or IL-2 CD4 response (Table 1) which is similar to the projected response towards each of the 11 supermotif peptides in a Caucasian population based on the DR-binding studies of Wilson et al. [8]. In addition, we found the percentage of subjects who responded to the three Gag peptides to be 13–15% higher when calculated as a percentage of subjects who responded to total Gag, rather than all HIV⁺ donors (Gag1: 84%versus 72%; Gag2: 62%versus 47%; Gag3: 76 versus 64%, respectively). These data indicate the 11 selected peptides to be broadly antigenic in the HIV⁺ patients studied.

Analysis of the proliferative response to supermotif peptides

Significant proliferative responses were noted to most antigens tested (SI > 3) and to the conserved peptides. More patients responded to p24- than Gp160-protein, in accordance with other reports [21,22]. The number of subjects who responded to the supermotif peptides by proliferation was lower than the ICS assay. Mean cpm counts in wells with no antigen ranged from 126 to 1758 with a median of 403; for p24, counts ranged from 135 to 27594 with a median of 1270; for Gag3 peptide, counts ranged from 98 to 3008, with a median of 636; and for Pol3 peptide, counts ranged from 101 to 2923 with a medium of 433.

Only 17–33% subjects tested were responders by the proliferation assay (Fig. 2) compared to 71–94% by the ICS assay (Fig. 1, Table 1). Only LTNPs were noted to have a specific proliferative response, whereas both LTNPs and SPs produced IFN-γ upon peptide stimulation. In a similar pattern to that noted for cytokine responses, the percentage of subjects who responded to the three Gag peptides was 8–50% higher when calculated as a percentage of subjects who responded to the p24 protein rather than all HIV⁺ donors (Gag1: 24%versus 16%; Gag2: 50%versus 24%; Gag3: 77 versus 24%, respectively). This indicates that the Gag epitopes represented in the three supermotif Gag peptides are probably represented in the naturally processed homologous protein.

Fig. 2.

Proliferation to Gag p24 and selected HIV-1 peptides in LTNPs and SPs. Proliferation calculated as stimulation index (SI) to antigens and the selected peptides is shown. Responses include both LTNPs and SPs, and four HIV negative healthy controls for p24 (HIV-p24). Each circle represents a patient. Horizontal bars represent median values. Numerical data based on patient responses are summarized.

Correlation of specific CD4 responses with markers of disease progression

The two groups of patients studied differed significantly in their virus load and CD4 count (this is highlighted in Fig. 3 showing a negative correlation between patients’ virus load and CD4 count).

Fig. 3.

Correlation of patients virus load with CD4 count. LTNPs are represented by triangle symbols and SPs by circle symbols.

In analysing the correlation of specific responses with markers of disease progression, we first addressed the issue of how the ICS assay correlated with the proliferation assay. By measuring the frequency of cells that were simultaneously stained with fluorochrome labelled anti-IFN-γ and anti-IL-2 antibodies, we identified five subsets of CD4 cells: total IL-2⁺ cells, total IFN-γ⁺ cells, IFN-γ⁺IL-2^– cells, IFN-γ^–IL-2⁺ cells and IFN-γ⁺IL-2⁺ cells. In Fig. 4 we show the correlation of the total Gag-specific cytokine response versus proliferation. The cytokine subsets that correlated directly with proliferation included all CD4 cells that stained for IL-2 (P = 0·0053, Fig. 4c), and the IFN-γ⁺IL-2⁺ subset (P = 0·0006, Fig. 4e) but not the subsets that comprised all IFN-γ⁺ cells (Fig. 4a, P = 0·1336) or cells that were IFN-γ⁺IL-2^– (Fig. 4b, P = 0·4693). Thus the ICS assay measured both the IL-2 and IFN-γ CD4 response, whereas the proliferation assay was a measure of the IL-2 rather than the IFN-γ secreting capacity of CD4 cells. This correlation between cell subsets identified by ICS and proliferation has not been reported before and as such is novel.

Fig. 4.

Correlation of Gag-specific CD4 T cell response measured by ICS assay versus proliferation. The absolute number of specific cells measured by the ICS assay was calculated by multiplying the percentage CD4-specific response with the patients’ CD4 count. The proliferative response to Gag p24 protein of the same subjects is shown as p24 SI. P- and r-values shown in each figure are determined by Spearman's rank correlation. The solid line represents a regression line. The following subsets of CD4 cells identified by the ICS assay were correlated with proliferation: (a) total IFN-γ⁺ cells, (b) IFN-γ⁺IL-2^– cells, (c) total IL-2⁺ cells, (d) IL-2⁺IFN-γ^– cells and (e) IFN-γ⁺IL-2⁺ cells.

Tables 3 and 4 summarize the correlation of cytokine responses to the supermotif peptides with patients’ virus load and CD4 count. Only those responses that correlated significantly (P = <0·05) with either CD4 count or VL are listed. In analysing specific cytokine responses, differences in CD4 numbers between patients was taken into account by correlating the absolute number, in addition to the percentage of cytokine-producing cells with viral load. The cytokine response to total Gag (Tables 3 and 4) and the p24-protein specific proliferative response (not shown) both correlated inversely with virus load, and positively with CD4 count. Loss of both these responses was associated with progression to disease, as indicated by higher responses in LTNPs compared to SPs (Tables 3 and 4). Of the Gag-specific cytokine subsets measured by the ICS assay, only the IFN-γ⁺IL-2⁺ CD4 subset correlated negatively with virus load, confirming our previous observations [9] and highlighting the importance of this subset as a marker of disease progression. The potential importance of IL-2 positive cells in HIV immunity is reflected by the negative correlation with virus load of the total number of Vβ⁺ IL-2⁺ CD4 cells measured in response to SEB stimulation (Table 4).

The only peptide that induced higher cytokine responses in LTNPs compared to SPs was Pol8 (P = 0·0431, Table 3) but this distinction was lost when absolute numbers of cytokine positive cells was calculated. No significant differences were noted between LTNPs and SPs in the proliferative response to any of the supermotif peptides. Cytokine responses to Gag1, Pol2 and Pol6 (Table 4) correlated negatively with virus load when using absolute numbers but not with percentages of responding cells. In contrast, cytokine responses to Pol3 and Pol5 showed a positive correlation with virus load by both absolute numbers and cell percentages (Table 4), with Pol3 responses also showing an associated negative correlation with CD4 count (Table 3). The correlation of these HIV-1 supermotif peptides with markers of disease progression in a patient cohort has not been undertaken before.

DISCUSSION

The identification of conserved T helper cell epitopes in HIV that are recognized by memory cells in the course of natural infection has significant implications for vaccine design. Recent work by Wilson et al. [8], who used a mathematical programme to determine peptide sequences in HIV-1 that are capable of binding multiple HLA DRB1 alleles, identified 11 supermotif peptides that were immunogenic in patients on HAART treatment. In this paper we demonstrate these peptides to be antigenic in treatment-naive subjects infected with HIV-1 for 15 or more years, and the percentage of recognition matched very closely the predicted population coverage estimated by Wilson et al. [8]. We are therefore able to confirm that these peptides are indeed ‘conserved’, in that they are recognized by subjects with differing DR alleles. In addition, the observation that the number of responders to the three Gag p24 supermotif peptides is higher when calculated as a percentage of patients who respond to the recombinant Gag p24 protein suggests that these peptides encode naturally processed and presented T helper epitopes.

The magnitude of CD4 responses to the 11 supermotif peptides was low, with median values in responders of approximately 0·03% cytokine positive CD4 T cells. This agrees well with the range of specific CD4 responses noted in chronic HIV infection. For example, Gag-specific CD4 T cell IFN-γ responses directed against multiple HIV Gag epitopes have been noted to be in the region of 0·10% (range 0·01–1%) [23–26], while CD4 IFN-γ responses to all potential epitopes represented in several HIV encoded proteins was noted to be around 0·74% (range 0·2–2·94%) [27]. Assuming that each of the 11 supermotif peptides encode a single rather than multiple epitopes, a median response of 0·03% specific cells (range 0·01–0·13%) is therefore in the expected range. This is highlighted by a recent comprehensive analysis of specific T cell responses directed against the entire HIV-1 genome where mean CD8 IFN-γ response to individual targeted peptides was 0·043% (range 0·004–0·2%), while the median total HIV-specific response was 0·4% (range 0·03–2·6%), with the highest responses noted in chronic untreated non-progressive infection [28]. In contrast to HIV specific CD8 responses, specific CD4 responses can be an order of magnitude lower. Betts et al. [27] showed that whereas the mean total HIV-specific CD4 response in chronic HIV infection was 0·74%, the same subjects had a mean specific CD8 IFN-γ response of 6·31%. Differences in the magnitude of HIV-specific CD4 versus CD8 responses despite persisting antigen are in keeping with emerging data, using the sensitive ELISPOT and ICS assays, showing low ex-vivo frequency of specific CD4 cells to be a characteristic of even self-resolving chronic virus infections that are not associated with a significant loss in CD4 count [29–31].

Several mechanisms may contribute to the characteristic, low ex-vivo specific CD4 memory response. Starting with a smaller burst size than the CD8 response [32], functional heterogeneity of CD4 cells with only a subset of effector cells contributing to long-term memory [32], associated with profound heterogeneity in triggering requirements [33], may all be contributing factors. In particular, effector Th1 cells defined by IFN-γ secretion do not develop into long-term memory cells as efficiently as CD4 T cells that do not secrete IFN-γ[34]. This is reflected in the CD4 responses of human subjects who have cleared a hepatitis C virus (HCV) infection. Godkin et al. [30] showed that the ex-vivo CD4 IFN-γ response to a core 15-mer HCV peptide declined from 0·012% to <0·005% over a 24-month period, whereas the response of in vitro-expanded cells to the same peptide rose from 0·1% to 1% over the same period; suggesting that that ex vivo and cultured responses probably detect different populations of memory cells. In addition to CD4 heterogeneity, the magnitude and specificity of the CD4 response may be influenced by disease stage. In the CD8 compartment there is evidence that the epitopes recognized in the acute and chronic stages of HIV infection differ [28,35]. The precise role of each of the above factors in governing responses to the supermotif peptides remains to be established.

In this paper, we provide an explanation as to why CD4 proliferative and cytokine responses do not strictly correlate, through novel work in correlating specific CD4 T cells identified by the production of IL-2 with those detected by proliferation. In keeping with the well-recognized role of IL-2 in T cell expansion and proliferation, we demonstrate a direct linear correlation between proliferation and the magnitude of those CD4 subsets that secrete IL-2 rather than effector cells that secrete only IFN-γ. Thus the two assays detect functionally distinct CD4 subsets. Whereas IFN-γ secretion is a measure of effector memory cells, CD4 memory cells defined by IL-2 secretion would include both central memory cells and a subset of effectors that secrete both IL-2 and IFN-γ[20]. However, even by this token, the number of subjects who responded to the 11 supermotif peptides by IL-2 secretion was 41–88% compared to 13–33% by the proliferation assay. This difference probably reflects the well-accepted notion that the ICS assay is more sensitive than the proliferation assay [23,25,36], that in turn is probably linked to the different requirements of the two assays. Cell proliferation is dependent on autologous APC, which may be impaired in HIV infection [37,38], in addition to possible fluctuations in culture conditions during this multi-day assay. These limitations are overcome in the ICS assay where co-stimulatory antibodies are included in the activation phase and the duration is much shorter. Lastly, the magnitude of the proliferative response has been shown previously to correlate negatively with virus load [22]. This may explain why many of our patients with detectable virus loads did not respond by proliferation, whereas Wilson et al. [8], who studied patients on HAART, noted up to 80% responders by the same assay to the same supermotif peptides. The virus load of the treatment-naive subjects in this study was higher than in the patients on HAART included in the Wilson study.

The identification of epitope specific memory CD4 T cells that play a protective role in HIV infection is a major goal, and while correlation of responses with disease markers is an indication of function it is by no means definitive. Thus, despite several lines of evidence, especially in animal models highlighting the importance of specific CD8 cells in virus clearance [39,40], both positive, negative or, indeed, no correlation of specific CD8 cells with virus load of HIV infected subjects has been noted [27,28,41]. Epitope specific responses that correlate positively with virus load are therefore not necessarily involved in promoting rather than curtailing virus spread and may simply reflect a higher response to increasing antigen concentration, as suggested in some studies [25,42]. However, it would be reasonable to expect a response that is associated with control of HIV infection to be higher in non-progressors than progressors. While this was noted to be the case for the total Gag-specific cytokine and Gag p24-specific proliferative CD4 responses, none of the supermotif peptides induced such a response. However, it remains possible that protective responses to peptides such as these may be found in other cohorts, for example exposed uninfected individuals, in whom responses to peptides not seen in HIV infected individuals have been noted [43]. In summary, the data indicate that despite their broadly antigenic nature, the importance of CD4 responses specific for the HIV CD4 epitopes identified by Wilson et al. in the control of HIV infection remains equivocal.