Longitudinal analysis of CD8 T-cell responses to HIV and hepatitis C virus in a cohort of co-infected haemophiliacs

Objective: To investigate CD8 T-cell responses to HIV and hepatitis C virus (HCV) over time in a group of co-infected children with haemophilia to assess the influence of the virus infections on each other and on clinical outcome.

Design: The HIV and HCV CD8 T-cell response of HLA-A2 co-infected individuals in the cohort were analysed at two time points, looking at the frequency and phenotype of HIV-specific T cells and assessing overall responses to the two viruses.

Methods: Peripheral blood mononuclear cells (PBMC) from 72 HLA-A2 co-infected individuals were analysed using an HIV HLA-A2 tetramer and by IFN-γ ELISpot using a panel of HIV and HCV antigens. PBMC from a group of 26 HLA-A2 HIV mono-infected adults were also analysed as a comparison.

Results: We identified two distinct patterns of response: some patients had a limited response to either virus whilst others made responses to a range of HIV epitopes. HCV responses were detected only in those who made multiple responses to HIV epitopes (P<0.0001). HCV infection had an influence on the phenotype of HIV-specific CD8 T cells, with a reduction in relative perforin and CD57 expression. Lack of functional or tetramer-positive HIV-specific T cells was associated with a decline in absolute CD4 T-cell counts between the time points (up to 7 years; P = 0.005).

Conclusion: HCV infection has an impact on the phenotype of HIV-specific CD8 T cells. In this well-defined cohort, failure to maintain effective CD8 T-cell responses against HIV may contribute to disease progression.

Introduction

Hepatitis C virus (HCV) and HIV are both major causes of morbidity and mortality worldwide. As the two viruses share routes of transmission, co-infection is common. In the USA, 16-20% of HIV-infected individuals are HCV seropositive ^[1]. It is known that HIV infection can impact on the progression of HCV disease ^[2], and liver disease is now a significant cause of morbidity and mortality in HIV infected cohorts ^[3]. Interestingly, HCV infection can also impact on the progression of HIV and the response to antiretroviral therapy ^[4,5]. The mechanism behind this is unclear.

The two viruses also share common elements in terms of the key immune responses thought to be relevant to control. HIV is associated with generation of strong CD8 T-cell responses which arise early and are maintained during viral persistence ^[6]. CD8 T cells play a key role in control of viral infection, as evidenced by substantial work in the SIV model ^[7], and the emergence of escape mutations in SIV and HIV ^[8,9]. In the case of HCV, strong immune responses in both CD4 and CD8 T-cell compartments may be observed in acute disease and in resolvers, where it appears that memory CD4 cells play an essential role in protection, although typically in chronic infection such responses are weak or undetectable ^[10,11]. An inverse correlation with viral load has been described ^[12]. Viral variation and immune escape may play some part in this, but these effects are seen even if epitopes remain intact ^[13].

HIV-specific CD8 T cells have been found to possess an 'intermediate' memory phenotype ^[14] with loss of CD28 and some expression of perforin and CD57, a marker of senescence ^[15]. By comparison, HCV specific CD8 T cells which are still detectable in chronic disease are CD28 high and perforin low and may retain CCR7 expression ^[16]. This 'stunted' phenotype may be observed in cells of diverse specificities in HCV-positive patients, for example cytomegalovirus (CMV)-specific T cells and the overall CD8 T-cell population have been found to express slightly lower levels of perforin in the presence of HCV ^[17]. It is not clear whether this represents an excess loss of mature effector cells ^[18], a failure to generate such cells ^[19] or excess regulatory T-cell activity ^[20]. Co-infection with HIV and CMV is also common and some authors have noted that CMV-specific CD8 T cells retain strong perforin expression in this setting ^[21].

To understand better the potential immunological basis for the adverse clinical outcome associated with co-infection, we analysed a cohort of young subjects with haemophilia described in the Hemophilia Growth and Development Study ^[5,22]. An adverse effect of HCV on HIV disease progression had been defined previously in this cohort ^[5]. We studied the CD8 T-cell responses against HCV and HIV in a subset of HLA-A2 individuals, to measure the breadth, magnitude and phenotype of T-cell response to both viruses. HLA-A2 was selected as it is the most prevalent class I allele and restricts several immunodominant HIV epitopes, including SLYNTVATL to which 70% of HIV-positive A2 adults have a response ^[23].

Methods

Cohort and cells

The Hemophilia Growth and Development Study was a multicentre, US study that enrolled, from 1989 to 1990, 207 HIV-1 infected children and adolescents, aged 6-19 years, with haemophilia who were also HCV infected ^[22,24]. During a 7-year follow-up period, blood samples were taken every 6 months and lymphocytes frozen. For this study, two samples were selected from 72 individuals who were HLA-A2 (median time interval 4.75 years, range 1-7 years). The median CD4 cell count at the first time point was 437 × 10⁶ cells/l (range. 3 × 10⁶-1965 × 10⁶ cells/l), the median plasma HIV-1 RNA was 3079 copies/ml (range, 32-84 520 copies/ml) and the median plasma HCV RNA was 3.36 × 10⁶ copies/ml (range, 2 × 10⁵-80.1 × 10⁶ copies/ml). Clinical disease progression was defined as a decline in absolute CD4 cell count between the time points examined ^[22].

Frozen peripheral blood mononuclear cells (PBMC) from a cohort of 26 HLA-A2 HIV mono-infected adults who were enrolled on separate HIV research projects in Oxford, were also studied at single time points as a control group. The median CD4 cell count for this group was 558 × 10⁶ cells/l (range, 297-924 × 10⁶ cells/l) and the median plasma HIV-1 RNA was 473 copies/ml (range, 50-750 000 copies/ml).

Peptides and recombinant proteins

Peptides previously identified as HIV CD8 epitopes ^[25] were selected for testing according to the HLA type of each subject. Where sufficient cells were available, three peptides were tested for each class I allele and the A2 restricted peptide SLYNTVATL was always included. Also included were pools of peptides covering all of p24 (twenty-two 20-mers, overlapping by 10) and p17 (thirteen 15-mers, overlapping by 5) and recombinant p24 and p17 proteins (all obtained from the AIDS Directed Programme). In addition, two pools of peptides from HCV were made consisting of twenty-two 9-mer peptides previously identified as HLA-A restricted CD8 epitopes ^[26] from HCV (HCV-A pool) and 22 HLA-B restricted peptides (HCV-B pool) and a pool of recombinant proteins covering core, NS3, NS4 and NS5 of HCV.

Tetramers and FACS analysis

HLA-A2-SLYNTVATL tetramer was synthesised and staining performed as previously described ^[27] using CD4-allophycocyanin, CD8-peridinin chlorophyll protein, and either CD57-fluorescein isothiocyanate or perforin-fluorescein isothiocyanate (BD Biosciences, Mountain View, CA, USA), and analysed using a FACScalibur (BD Biosciences). Individuals were considered tetramer positive when the frequency of tetramer-positive cells was >0.05% of the total CD8 population. This cut-off is based on results from a number of previous studies ^[17,28,29].

ELISpot assay for IFN-γ secretion

This was performed as previously described using 100 000 cells per well ^[30].

Statistical methods

Correlations were performed by Spearman rank correlation and statistical significances were performed by Fisher's exact or Mann-Whitney test using Prism 3.0 software.

Results

Analysis of immune responses in co-infected patients using class I-peptide tetramers and ELISpot assays

Class I peptide tetramer analysis of CD8 T responses to the immunodominant SLYNTVATL epitope was performed. This epitope was chosen as it is commonly targeted by HLA-A2 HIV-positive individuals in chronic disease ^[31]. Of 64 A2 dually infected individuals tested, 52 (81%) had positive responses using this assay while 12 were consistently negative at both time points (Fig. 1a). Among the HIV mono-infected group, 15 of 26 (57.7%) were positive (Fig. 1a) and discordant results (positive at only one time point) were obtained in 11. Examples of such staining are shown in Fig. 1b and d. The overall magnitude of SLYNTVATL-specific responses was mean 0.72 ± 0.97% of total CD8 T cells, range <0.05 to 4.8%.

Fig. 1

We analysed the functionality of the SLYNTVATL-specific responses in this cohort using ex vivo interferon (IFN)-γ ELISpot assays. Data from the ELISpot assays of the 54 individuals who showed a positive phytohaemagglutinin response were analysed. Of these 21 were ELISpot positive for a response to this peptide. Among this group of 54 25.9% were on antiretroviral treatment at the first time point, 51.85% were on antiretroviral treatment at the second time point but only one was on a triple-drug therapy regimen that included a protease inhibitor.

An overall correlation between the ELISpot and tetramer assays was obtained. Firstly, all ELISpot-positive subjects had tetramer responses. However, 18 individuals (33%) showed clear tetramer responses without detectable ELISpot responses. In 10 of these, repeat assays detected responsiveness to other HIV peptides, suggesting that the lack of SLYNTVATL response was not an artefact. The maximum tetramer response in this group was 1.6% of total CD8 cells suggesting that the effect was not due to differences in detection sensitivity. Secondly, a linear correlation was found between the frequency of antigen-specific cells in the ELISpot compared to the tetramer assays (r = 0.3257, P = 0.009; Fig. 2a). Using these data, it was possible to calculate the percentage of tetramer-positive cells that secrete IFN-γ for each sample [(spots per 1 × 10⁶ PBMC/tetramer-positive cells per 1 × 10⁶ PBMC) × 100]. The fraction of tetramer-positive cells that produced detectable IFN-γ in ELISpot assays was usually less than 10% (mean, 2.9 ± 5%; n = 62) and this is comparable with the value for the HIV mono-infected control group (mean, 4.4 ± 3.3%; n = 14).

Fig. 2

To analyse the maximum breadth of the immune response against HIV and HCV using very limited cell numbers, we used an ELISpot approach based on pools of peptides as described in Methods. Of the 54 individuals in whom ELISpot data were obtained, 35 of 54 (64%) showed responses to HIV peptides. An example of a positive assay is shown in Fig. 2b, which illustrates the broad response to a diverse range of epitopes that has previously been described in chronic HIV infection ^[32]. The 19 non-responders showed responses only to phytohaemagglutinin at both time points.

HCV-specific immune responses were detected in 9 of 31 individuals in whom enough cells were available to screen the epitopes from both viruses. All nine had a response to the HCV HLA-A restricted peptide pool of which six also had a response to the HLA-B restricted peptide pool. As previously reported using a broader screening method ^[33], this study also detected immune responses against HCV much less frequently than those against HIV in co-infected individuals [9/31 (29%) versus 35/54 (64%)]. Interestingly, all cases where HCV responses were detected were associated with HIV immune responsiveness and all were associated with SLYNTVATL responsiveness (P = 0.02, Fisher's exact test). Furthermore, HCV responsiveness was associated with response to two or more HIV antigens (9/9 versus 4/22; P = 0.0001, Fisher's exact test, Table 1). The overall level of responsiveness to pooled HCV peptides was not significantly different from that in a group of matched HCV mono-infected patients from the same cohort (9/31 versus 3/20). It is possible that viral variation between HCV genotypes could account for some of the unresponsiveness to the HCV peptides, which were all derived from genotype I sequence. However, the bulk of patients in this cohort were genotype I so this is unlikely to be the reason for the general lack of responsiveness ^[13]. There was no difference in the clinical status of HCV infection in this group of nine when compared with the rest, as measured by alanine aminotransferase levels or viral loads. Co-infection with CMV in the HIV/HCV group (based on serological status) did not appear to influence the cellular response to HIV since the numbers of individuals positive for CMV antibodies were equally distributed among responders and non-responders, measured by either ELISpot or tetramer analysis.

Table 1

The phenotypes of the tetramer-positive cells were analysed by looking at expression of either perforin (Fig. 1b and c) or CD57 (Fig. 1d), markers of cytotoxic potential ^[34] and replicative senescence ^[15] respectively. In total, PBMC from 39 of 64 co-infected and 25 of 26 HIV mono-infected individuals were stained for the expression of perforin, and PBMC from the remaining 25 of 64 co-infected individuals were stained for CD57. Overall, in the dually infected individuals, tetramer+CD8 cells were perforin low (mean, 25.7 ± 19.7% compared to 46.7 ± 20.6% in the total CD8 population). Within any individual, the levels of perforin were significantly lower in the HIV-specific cells (mean of differences, -20%; P = 0.0001) compared to bulk CD8 T cells. This was in comparison to the 26 HIV mono-infected individuals where the levels of perforin were not significantly different overall between tetramer-positive CD8 cells and the total CD8 population (29.1 ± 17.1% versus 29.7 ± 17.5%) (Fig. 1a) or within individuals (mean of differences, -0.19%). When levels of CD8 T-cell perforin expression were compared between the SLYNTVATL responder and non-responder groups, no differences were apparent.

In parallel, staining of CD4 cells for perforin showed the overall levels to be significantly higher in the HIV mono-infected group (5.3 ± 5.8% of total CD4 cells were perforin-positive) than in the co-infected group (2.6 ± 3.9%; P = 0.0016, Fig. 1c). This compares to a previous study ^[35] in which this subset of CD4 cells were found to be increased in HIV infection, some as high as 60% (mean, 17% in chronic infection), in comparison to a group of healthy donors where the mean level was 2.2%.

CD57 expression on CD8 T cells has been reported as elevated in one study of HIV infection ^[15]. In the study reported here, the presence of both viruses seemed to have led to significantly reduced expression of CD57 on tetramer-positive cells (Fig. 1c) (33.5 ± 10.8% in HIV mono-infection versus 21.3 ± 12.5% in HIV/HCV dual infection; P = 0.0007). This difference was not observed in the total CD8 population where levels were similar in both groups (31.2 ± 13.2% and 34.2 ± 12.7%). Thus in the presence of HCV infection, a significant proportion of the HIV-specific tetramer-positive cells had relatively reduced levels of both perforin and, particularly, CD57 and so expressed a less 'mature' phenotype than similar cells in the HIV mono-infected group.

Relationship between immune responsiveness and disease progression

As the immunological assays demonstrated that a significant fraction of individuals lacked detectable immune responses measured by either the combined overlapping/optimal peptide approach or tetramer staining, we ascertained whether responsiveness was linked to disease progression as determined by decline in CD4 cell count. Analysis of the change in CD4 cell count between the two time points tested in those who responded to HIV antigens by ELISpot revealed that such responders had a significantly lower rate of progression. Amongst those whose CD4 cell count was maintained or increased over the study period, anti-HIV ELISpot responses were more readily detectable than in those in whom progression occurred (11/11 versus 24/43; P = 0.005; Table 2). A similar result was obtained for tetramer analysis (14/14 versus 38/50; P = 0.05). Figure 3 shows the relationship between the change in CD4 cell count over time and the presence of a (functional) response to SLYNTVATL. Again, a significant CD4 cell count decline in the non-responders is observed (P = 0.0333) and this association was also significant when looking at responders to any HIV/HCV antigen (P = 0.0398). Analysis of the CD4 cell count decline among the subgroup of 31 individuals who were tested for response to HCV antigens did not reveal any significant difference between those with and without a response to HCV. Similarly, when change in CD4 cell count was analysed against CMV status, the decline in CD4 cell count was not significantly different between those who were CMV-positive and those who were CMV-negative.

Table 2

Fig. 3

Conclusions

The data, derived from a longitudinal study of HIV infection from the pre-HAART era, provide important information in three areas; the relationship between immune responses to HCV and HIV, the impact of HCV on HIV-specific T-cells and the relationship between HIV-specific cells and disease progression.

Relationship between responses to HCV and HIV

Using two independent techniques, we investigated to what extent immune responses against HIV and HCV were detectable in co-infected subjects. Our data reinforces the findings of others ^[33] that in chronic infection, the immune response against HIV and HCV differ significantly in overall vigour.

The HCV-specific responses in this study were linked to the presence of strong anti-HIV responses. In the earlier cross-sectional investigation carried by Lauer et al. ^[33] on a more limited number of co-infected individual using the recombinant vaccinia approach, all individuals had a detectable immune response to HIV. However, these studies were performed on fresh cells and may therefore have greater sensitivity, as responsiveness may diminish after freezing. If only relatively strong responses were counted (>250 spots per million), these were detectable in 15 of 20 individuals, the other responses being just above background. However, Lauer et al. found that all five individuals who had detectable responses to HCV had robust HIV-specific responses, a result quite similar to that obtained in this analysis. A further important difference with the work reported here is that, in the previous work, the majority of individuals previously were on HAART, with low HIV viral loads. A more recent study by Kim et al. ^[36] revealed higher levels of HCV responses using a matrix of overlapping peptides, and an important relationship with CD4 cell count. In both of the above studies, effective HAART therapy (which has been reported to have little impact on HCV viral load) ^[37] was widely used, unlike the subjects studied here.

The tetramer staining used here revealed T-cell responses which were linked in most cases to functional responses in the ELISpot assays. Overall, T-cell functionality is intact, although the presence of a substantial minority of 'dysfunctional' CD8 T cells is intriguing. It is theoretically possible that this relates to viral variation within the SLYNTVATL epitope, although it should be noted that the same peptide is present in both tetramer and ELISpot. The mechanism behind the unresponsiveness is therefore not clear, and may be unrelated to the presence of HCV since the fractions of tetramer-positive cells producing IFN-γ were similar in both the dually-infected and mono-infected control group. The question of the functionality of HIV-specific CD8 T cells has been addressed before in various studies ^[34,38], although with discrepant results. Analysis of interleukin (IL)-2 secretion was not performed in this study but this may be of future interest - a subset of virus-specific CD8 cells producing both IFN-γ and IL-2 has been detected in HIV long term non-progressors (and in Epstein-Barr virus and CMV) but rarely in chronic HIV, suggesting a protective role ^[39].

Impact of HCV on the phenotype of HIV-specific T cells

This is the first study in which the phenotype of HIV-specific CD8 T cells in the context of HCV infection has been analysed. The phenotype of the tetramer-positive cells obtained showed some important variation from that of the HIV mono-infected control group. Analysis of the perforin content of HIV-specific cells showed a reduction in perforin when compared to CD8 T cells as a whole in the co-infected group. In the mono-infected group, the levels were similar in tetramer-positive and tetramer-negative cells whilst in the co-infected group the ratio was about 2: 3. The functional significance of this was not assessed but expression of perforin may be linked to maturation as well as killing.

The influence of HCV on levels of perforin in CD4 T cells was also detectable as levels of this subset of cells in the co-infected group were significantly lower than among the mono-infected subjects. Although the levels in the mono-infected group were not as high as those reported previously ^[35], with a range of 0-21%, this may reflect variations in cell preparation or staining. Since the detection of perforin is thought to indicate an end-stage differentiated cell, the lower levels seen in the presence of HCV may once more illustrate the influence of this virus on maturation of lymphocytes.

Due to limitations in cell numbers, further markers of cytotoxicity were not examined. However, it would be of interest to discover whether expression of additional molecules such as Fas, and TRAIL, etc. are altered in the presence of HCV. One defect of the study is that a group of age-matched HIV-positive, HCV-negative (HIV mono-infected) haemophiliac children were not available for comparison. In order to obtain control data we used an HLA-A2 matched group of otherwise unrelated adult patients. It is possible that differences in age and other factors influenced the comparison of phenotype.

A marker which is thought to be significant in the maturation of T cells is CD57. The CD57 phenotype is associated with replicative senescence such that cells expressing this marker lose their proliferative ability and are highly susceptible to activation-induced apoptosis ^[15], although they are still capable of producing IFN-γ. In this study, the proportion of tetramer-positive cells expressing CD57 was significantly lower in the co-infected group compared to the mono-infected subjects (21.3: 34.1%). This may reflect the pervasive influence of HCV on the maturation pathway of CD8 cells specific for other viruses, as described previously ^[17]. In co-infection with CMV and HCV, a more immature phenotype of the CMV-specific CD8 T-cell population - e.g., perforin low - is also seen. Indeed, other co-infections could also potentially impact on the results obtained. CMV is known to influence HIV progression and to have immunomodulatory effects ^[40] and other potentially confounding persistent co-infections might include hepatitis G virus and Epstein-Barr virus. The impact of HCV on phenotype and function of acute and persistent adult HIV infection is an important further question currently under study.

Relevance to clinical outcome

Importantly, we identified a link between the vigour of the immune response and clinical outcome, defined as a decline in absolute CD4 cell count between two time points examined. Lack of responsiveness in the ELISpot and tetramer assays was associated with a worse clinical outcome in terms of disease progression. This was despite no differences in initial CD4 cell count. Previous associations between the magnitude of CD8 T-cell responses and virus load have been largely cross-sectional ^[6], so this study provides important information regarding the potential role of HIV-specific T cells in co-infection, with implications for mono-infection.

Conclusions

This group of dually infected individuals is unusual in that it represents a paediatric cohort in whom disease was acquired horizontally and where HCV infection was virtually 100%. In summary, CD8 T-cell responses against HIV in HCV-HIV co-infected individuals are readily detected in the majority of this cohort by a variety of tests, but weak or absent in a substantial minority. We provide novel data to support the idea that HCV does have an impact on the maturation state of HIV-specific cytotoxic T lymphocytes and that cytotoxic T lymphocyte responses against HIV play an important role in protection against disease progression in such co-infected populations.

Acknowledgements

Financial support was received from the Wellcome Trust (G.C.H. and P.K.) and the National Institutes of Health, National Institute of Child Health and Human Development (HD41224), Bethesda, MD, USA. We thank Dr. Mary Carrington for providing the HLA class I genotypes for the Hemophilia Growth and Development Study participants.