Clonal evolution of CD8⁺ T-cell expansions in HIV-infected patients on long-term HAART

Abstract

HIV-1 continually replicates in spite of long-term highly active anti-retroviral therapy (HAART) and therefore, it is conceivable that the low level, persistent viral activity could continue to stimulate the hosts immune system despite remaining below the detection limit of the current assays. In this study, we performed a longitudinal analysis of the CD8⁺ T-cell receptor Vβ repertoire in HAART-treated and untreated HIV patients. HAART-mediated control of viremia, for up to 18 months, did not prevent similar perturbations within the CD8⁺ Vβ repertoire in both study groups as defined by CDR3 spectratyping. Oligoclonal Vβ expansions, with new dominant CDR3 lengths, were observed throughout the study period. Our findings are compatible with antigen-driven CD8⁺ immune responses to bursts of replication from a continuously changing viral reservoir, regardless of HAART-mediated suppression of HIV-1 viremia.

Introduction

CD8⁺ T cells play an important role in the immune response to HIV infection. Vast expansions of CD8⁺ T cells are seen during Primary HIV Infection (PHI) [1–3]. These expansions have been attributed to stimulation by viral antigens and correlate with subsequent immunological control of viremia, as the appearance of HIV-specific CD8⁺ cytotoxic T lymphocytes (CTL) coincides with the decline of initial viraemia [4–6]. This massive expansion of HIV-specific CD8⁺ T-cell clones during PHI is often transient and may disappear through mechanisms such as ‘clonal exhaustion’ [7]. However, circulating HIV-specific CD8⁺ T cells are detectable in chronically infected patients [8], indicating continued HIV-1 antigenic stimulation.

Early control of viremia can prevent the loss of HIV-specific CD4⁺ T cells [9] and reportedly decreases CD4⁺ T-cell repertoire disruptions [10,11]. This has led to the advocacy of very early initiation of highly active anti-retroviral therapy (HAART).

Although HAART, in most cases, dramatically and rapidly reduces viral replication to undetectable levels in the plasma [12,13], a reservoir of replication-competent HIV-1 persists even after 2 years of ‘successful’ therapy [14–17]. Indeed, one study [16] demonstrates the presence of ongoing HIV replication, and even genetic sequence evolution, in a group of PHI patients with undetectable levels of plasma viremia on triple therapy for 20–35 months.

With this in mind, we undertook a detailed longitudinal study to evaluate whether we could detect ongoing alterations in the CD8⁺ T cells in a cohort of HAART-treated PHI patients by analysing the hypervariable CDR3 (complementarity determining region 3) area of their Vβ chains using CDR3 spectratyping [18,19]. TCR diversity is generated through T-cell receptor gene rearrangements and the introduction of untemplated nucleotides. A number of T-cell repertoire studies in HIV-infected patients have used CDR3 spectratyping to analyse the entire repertoire or selected elements based on the variable family regions [10,20–22]. While vital for the detection of large or dominant T-cell expansions, this has been at the expense of resolution, as minor alterations in the T-cell pool are lost in the background of the assay. Restricting CDR3 spectratyping to a single Joining region removes the other 12 Joining region families, which are all superimposed in the previous assays. This gives a heightened sensitivity to smaller fluctuations and movement within the T-cell compartment [23,24]. Although this measures only an estimated 6% of the T-cell repertoire directly [23], it has recently been shown that such results can be extrapolated to reflect the whole repertoire [25].

To date, no controlled longitudinal study on the CD8⁺ T-cell compartment has been performed in PHI patients. Here, we have analysed and compared, over a period of up to 18 months, the CD8⁺ T-cell repertoire in PHI patients initiated on HAART during their acute seroconversion syndrome with similar patients declining HAART. Our results demonstrate that the CD8⁺ T-cell repertoire remains highly perturbed, with severe biases and vast oligoclonal expansions even in those subjects with HAART-mediated viral suppression for up to 18 months. The impact of antigenic stimulation by HIV quasispecies during anti-retroviral therapy on the dynamics of the CD8⁺ T-cell compartment is shown.

Materials and methods

Patients and controls

Fresh peripheral blood from 10 HIV-1-infected individuals was studied, at regular intervals, for up to 18 months. All patients gave informed consent and all protocols were approved by the Research Ethics Committee at the Chelsea and Westminster Hospital. Six PHI patients were immediately placed on triple combination therapy (696, 699, 116, 972, 403 and 618), namely indinavir (3 × 800 mg/day), 3TC (2 × 150 mg/day) and AZT (2 × 250 mg/day). In this group, viremia was rapidly suppressed below detection (< 50 RNA copies/ml) (data not shown). Four patients within the first year of infection, including one PHI patient, declined therapy (691, 374, 244 and 196). Untreated patients had high viral loads throughout the study period, except for patient 244. All patients were infected through sexual intercourse. Three uninfected individuals were followed in a similar fashion to act as a control group.

CD8⁺ T-cell isolation and purity

Peripheral blood mononuclear cells (PBMC) were obtained by density centrifugation (Histopaque, Sigma, Poole, UK) and CD8⁺ T cells were isolated using anti-CD8 antibody-coated magnetic beads (Miltenyi, Harrow, UK). Purities of the selected T cells were > 98% as assessed by flow cytometry.

DNA preparation

DNA was extracted using Tri-reagent (Sigma) as per the manufacturer's recommendations. Integrity of the DNA was confirmed by a beta-actin PCR.

CDR3 spectratyping

The DNA preparations were used as templates in a previously described [23,24], semi-quantitative PCR procedure. Briefly, a 99 cycle linear PCR reaction was conducted using a Jβ1·6 specific primer with a 5 U Stoffel fragment, 2·5 mm MgCl₂, 200 µm each of dNTP and 1 × Stoffel Buffer (Perkin-Elmer Applied Biosystems, Warrington, UK). Products from the first PCR were used as templates for 24 subsequent, concurrent, 35-cycle reactions employing a nested Jβ1·6 primer and one each of previously validated [26] Vβ-specific primers in a 5 U Stoffel fragment, 2·5 mm MgCl₂, 200 µm each of dNTP and 1·25 × Stoffel Buffer. The products generated from each set of PCRs were labelled in a final, 15 cycle linear PCR with a flourochrome-conjugated internal Jβ1·6 primer; relevant PCR controls were employed as previously described [23]. These reactions resulted in clusters of CDR3 PCR products (termed spectratypes). These labelled PCR products were sized and quantified on an automated DNA sequencer (ABI Prism 310, Perkin-Elmer Applied Biosystems) using POP-4 polymer and ROX-500 as internal size standards.

Analysis of CDR3 spectratypes

Analysis of CDR3 spectratypes was conducted using Genescan and Genotyper software (Perkin Elmer Applied Biosystems). The sum of the fluorescence intensities generated within each CDR3 spectratype was calculated to determine the relative abundance of the individual Vβ families at each time point measured. As has been previously defined [18,19,23,24], if one PCR product constituted at least 40% of a spectratype, that Vβ family was considered ‘perturbed’. Vβ families depleted to levels below detection were termed ‘non-detectable’ and the rest were termed ‘non-perturbed’. The variation of the relative abundance was determined by the highest and lowest amounts of CDR3 PCR products over time. Variance was assessed using Microsoft Excel 5 (nΣχ² – [Σχ]²)/n(n – 1).

Results

CD8⁺ TCRVβ skewing persists despite controlled viremia

Figure 1 illustrates the type of CDR3 spectratypes obtained using this technique; the size of the PCR product is represented on the x-axis and the amount of product on the y-axis, as detected by the amount of incorporated labelled primer. Figure 2 represents over 1000 CDR3 spectratypes, showing the states, either ‘non-perturbed’, ‘perturbed’ or ‘non-detectable’, of all the CDR3 spectratypes from each patient in this study. A highly perturbed CD8⁺ T-cell repertoire is detected in all patients, with a considerable number of expanded Vβ families. In patients where an early time point of CDR3 spectratyping could be performed, i.e. when high viremia was present during PHI and before initiation of HAART, several CDR3 spectratypes were non-detectable. A number of these initially non-detectable CDR3 spectratypes became detectable at later time points in both HAART-treated and -untreated patients. The reverse was also observed, that initially detected either non-perturbed or perturbed CDR3 spectratypes became non-detectable. A large proportion of the CDR3 spectratypes also alternated between non-perturbed and perturbed between consecutive bleeds in both the HAART-treated and the untreated patients (Fig. 2). Moreover, new dominant CDR3 products emerged within many spectratypes between consecutive bleeds (Fig. 1, left panel).

Fig. 1.

Representative CDR3 spectratype patterns over time of a PHI patient and a control subject. DNA from purified CD8⁺ T cells from Patient 691 (left panel) and control subject IC 1 (right panel) was subjected to PCR amplification using Vß1-Jß1·6 primers as described in Materials and methods. The CDR3 length (in nucleotides) is represented along the x-axis and the fluorescence intensity in arbitrary units along the y-axis.

Fig. 2.

Longitudinal representation of CD8⁺ Vß perturbations in treated and untreated PHI patients. CDR3 spectratyping of the 24 Vß families was performed on purified CD8⁺ T cells (as described in Materials and methods) at the indicated time points (weeks after symptomatic PHI). Patients 696, 699, 116, 972, 403 and 618 received HAART, while patients 691, 374, 244 and 196 remained untreated. Vß families with non-perturbed CDR3 profiles are represented by white squares, perturbed CDR3 profiles by black squares (i.e. when one CDR3 peak constitutes at least 40% of the DNA in a CDR3 spectratype) and non-detectable CDR3 profiles by grey squares.

Since persistent clonal CD8⁺ T-cell expansions have been identified in healthy individuals [27,28], we also performed the assay on three non-HIV-infected controls (IC 1, IC 2 and IC 3). The overall results within the control group revealed a non-perturbed repertoire with the vast majority of the CDR3 spectratypes being non-perturbed. Figure 1 (right panel) depicts serial CDR3 spectratypes from a control subject (IC 1) to illustrate the stability between consecutive bleeds.

Thus, the CD8⁺ repertoire remained highly variable and skewed towards oligoclonally-expanded CDR3 spectratypes in both untreated HIV-infected patients with detectable viremia and in HAART-treated HIV patients with controlled viral replication for up to 90 weeks (Fig. 2).

CD8⁺ TCRVβ composition fluctuates over time in treated and untreated HIV patients, but not in controls

The continuous changes in CDR3 spectratype patterns observed in all HIV patients, regardless of therapy, viremia level or time of sampling during the first year of infection within the JB 1·6 family, suggest a highly dynamic CD8⁺ T-cell compartment. In order to determine whether major changes in the whole CD8⁺ T-cell compartment were occurring, we analysed the relative abundance of each of the 24 Vβ families within the JB 1·6 family over time. As the sum of intensities of fluorescence of the PCR products generated within the CDR3 spectratypes represents the amount of DNA, the highest and lowest amount of PCR products, over time, for each Vβ family was plotted as a percentile histogram to gauge the variation of abundance. The variation in the composition within each Vβ family is presented in Fig. 3.

Fig. 3.

Range of relative Vβ abundance within the CD8⁺ T-cell compartment over the study period. The sum of intensities of fluorescence generated within each Vβ CDR3 spectratype was calculated to determine the relative abundance of the individual Vβ families within the CD8⁺ T-cell subset. The variation in the relative abundance of the 24 Vβ families is determined by the highest and lowest amounts of the PCR products detected over all measured time points. IC 1, IC 2 and IC 3 are three healthy controls where CDR3 spectratyping was performed at four time points over 6, 9 and 12 months, respectively. Three untreated HIV patients (244, 196 and 374) and six HAART-treated HIV patients (972, 116, 403, 618, 696 and 699) are depicted. The variation for the HIV-infected patients is based on the time points indicated in Fig. 2.

Within the HIV-negative control group, the range of variation within any particular Vβ family over a period of up to 12 months remained remarkably small, both within and between individuals (Fig. 3). In the majority of the untreated HIV-infected patients with uncontrolled viremia (244, 196 and 374), a substantial degree of variation within given Vβ families was observed (Fig. 3). Interestingly, an equally fluctuating Vβ composition was also seen in the HAART-treated group (972, 116, 403, 618, 696 and 699).

As there was a high degree of variation in individual Vβ families in both the treated and untreated group of HIV-infected patients, we also determined the relative abundance of all the Vβ families for their average variance over time. As shown in Fig. 4, control subjects all exhibit a low Vβ variance (range: 2–5) within the JB 1·6 family, while the variance for both the treated and untreated HIV patients is much higher (range: 8–19). Obviously, a concern was that some of the more substantial Vβ variations may have occurred at early time points and then levelled off over time. However, no difference was observed when comparing the average variance during the first and last 6 months of the study in the two HAART-treated PHI patients enrolled for the longest period (696 and 699) (data not shown). Therefore, the JB 1·6/Vβ variance was consistently elevated in both treated and untreated HIV-infected patients as compared with uninfected controls (Fig. 4).

Fig. 4.

Vß variance over time within each subject. The average level of variance within each subject was calculated over time using MS Excel as described in Material and Methods. The average variance for the healthy controls (left panel), the untreated HIV patients (middle panel) and the HAART-treated HIV patients (right panel) is presented.

Sequential waves of expansion and retraction of CD8⁺ TCRVβ families in HIV-infected patients

The relative contribution of each of the 24 Vβ families within the CD8⁺ T-cell compartment was obtained by calculating the surface area under each CDR3 spectratype at each time point and dividing it by the combined area under all the peaks at that particular time point.

The results from one untreated individual (691) and one HAART-treated patient with controlled viraemia (699) are illustrated in Fig. 5(a,b). Remarkable fluctuations in the relative contribution of several Vβ families are obvious in both subjects. During a 20-week follow-up period starting at PHI in patient 691 (Fig. 5a), five Vβs were initially highly expanded, decreased over a few weeks and subsequently reappeared as dominant. The reverse pattern was seen for other Vβs. Interestingly, certain Vβ families appeared synchronized in their waves of appearance and disappearance over time. Patient 699 (Fig. 5b), who was initiated on anti-retroviral therapy during the PHI syndrome, showed similar patterns of expansions and retractions of some Vβ families, despite undetectable viremia for more than 18 months. In both subjects 691 and 699, approximately 50% of the Vβ families showed patterns of ‘waves’. Such expansions and retractions of peripheral Vβ families were not observed in the healthy control group (data not shown).

Fig. 5.

Sequential waves of expansion and retraction of CD8⁺ Vß families in treated and untreated HIV patients. The abundance of individual peripheral Vß families, at given time points, is presented as the relative composition (%) of the peripheral CD8⁺ Vß repertoire and calculated as described in Methods and in Fig. 3. The relative Vß composition for (a) Patient 691 (untreated) and (b) Patient 699 (HAART-treated). The various Vßs have been grouped according to their patterns of fluctuations over time.

Discussion

Here, we describe ongoing oligoclonal expansions in the CD8⁺ T-cell compartment as measured by CDR3 spectratyping in patients during PHI, and a period of up to 18 months after infection. Undetectable viraemia over long periods of time did not preclude the persistence of these expansions. Interestingly, the degree of expansions observed in the HAART- treated PHI group appeared similar to those observed in the untreated group of HIV-infected patients with uncontrolled viremia (Fig. 3). The mechanisms responsible for the persistent CD8⁺ T-cell expansions in the group of PHI patients after a long period of HAART-controlled viremia is unclear. In view of the recent findings that replication-competent virus is active during, and can be recovered after, long-term HAART [14–17], one possible explanation may be persistent low level antigenic stimulation.

In untreated HIV-infected patients with chronic infection, expansions of CD8⁺ T cells have been found to be HIV-specific [29], and an inverse correlation between the frequency of HIV-specific CTL and plasma viremia has been demonstrated by the use of MHC Class I MHC tetramers [30], suggesting that CTL responses are important in controlling virus replication. Emerging data also indicate that HIV-specific CTL responses actively induce virus mutants, which are sometimes associated with evasion of the CTL response [31–35]. The continuous changes in CDR3 spectratype patterns and the appearance of new dominant CDR3 lengths among the Vβ families observed in both the HAART-treated and the untreated HIV-infected patients, further supports the proposition that the persistent CD8⁺ T-cell expansions are antigen driven in both cases. Unfortunately, we do not have data as to the antigen specificity of these T-cell expansions or sequence evolution of the CTL epitopes which would strengthen this argument.

Recently, a new model for viral decay during HAART was proposed where, in addition to half-lives of viral reservoirs, HIV replication occurs in multiple bursts associated with immune response episodes to antigen [36]. A fluctuating Vβ composition was indeed maintained in all the HAART-treated PHI patients (Figs 3 and 4), suggesting that even low levels of viral replication may be able to sustain a high degree of perturbation in the CD8⁺ T-cell repertoire. Although viral reservoirs are believed to decline over time with successful HAART [14–17], and even more rapidly when Interleukin-2 (IL-2) is given intermittently [37], the new model mentioned above [35] proposes that bursts of viral replication may actually be sustained in a steady state, driving (and being driven by) T-cell expansions. Even patients on long-term HAART and IL-2, with undetectable replication-competent virus, show a strong rebound in plasma viremia following discontinuation of HAART [38], which in speed and magnitude resembles that observed during PHI [39]. Such findings, together with the continuous CD8⁺ T-cell expansions in the presence of long-term controlled plasma viremia, suggest the persistence of the virus in other reservoirs.

Replication-competent HIV found after long-term HAART [14–17] may be due to the lack of access of drugs to residual compartments, or to the insufficient potency of the anti-viral regimen. The recent discovery of covert virus replication [40] and genetic sequence evolution [16] in patients on long-term HAART suggests that some degree of selective pressure takes place even during HAART. The waves of clonal expansions and retractions of selected Vβ families within the CD8⁺ T-cell compartment shown here may precisely reflect the stimulation by HIV quasispecies during HAART. In other words, during viral containment, low levels of proliferating HIV-specific CD8⁺ T cells may positively select new viral epitopes, which would emerge from long-lived cells and possibly form sanctuaries which are inaccessible to therapy.

While some reports suggest that the initially expanded CD8⁺ T cells are transient and disappear within 6 months after PHI [3,7], one longitudinal study in monkeys shows that CD8⁺ T-cell expansions following primary simian immunodeficiency virus infection persist [41]. Our data in the untreated group of HIV patients (Figs 2 and 3) confirm that similar CD8⁺ T-cell expansions also persist following PHI in humans. Massive clonal expansion of CD8⁺ T cells is also an important feature of acute infectious mononucleosis (AIM) [42], where up to 40% of CD8⁺ T cells can be specific for a single EBV epitope [43]. In contrast to HIV infection, however, the same EBV-specific T-cell clones present during AIM are found at high frequencies one year later [44].

This is the first longitudinal study to demonstrate a highly dynamic CD8⁺ T-cell compartment in PHI patients after long-term HAART. The persistent CD8⁺ T-cell expansions have important implications for cell-mediated immunity emerging during HAART, as well as for the development of more efficient treatment regimens in order to eradicate residual viral replication.