SGM Journals
RNA Viruses

A functional CD8+ cell assay reveals individual variation in CD8+ cell antiviral efficacy and explains differences in human T-lymphotropic virus type 1 proviral load

Becca Asquith, Angelina J. Mosley, Anna Barfield, Sara E. F. Marshall, Adrian Heaps, Peter Goon, Emmanuel Hanon, Yuetsu Tanaka, Graham P. Taylor, Charles R. M. Bangham

  • 1Department of Immunology, Imperial College, London, UK
  • 2Department of Genito-Urinary Medicine and Communicable Diseases, Imperial College, London, UK
  • 3GlaxoSmithKline Biologicals, Belgium
  • 4Department of Immunology, Graduate School and Faculty of Medicine, University of the Ryukyus, Japan
  • Correspondence
    Becca Asquith
    b.asquith{at}imperial.ac.uk
    Charles R. M. Bangham
    c.bangham{at}imperial.ac.uk

    • Journal of General Virology 2005; 86(5):1515–1523 · https://doi.org/10.1099/vir.0.80766-0

    View at publisher PubMed

    Abstract

    The CD8+ lymphocyte response is a main component of host immunity, yet it is difficult to quantify its contribution to the control of persistent viruses. Consequently, it remains controversial as to whether CD8+ cells have a biologically significant impact on viral burden and disease progression in infections such as human immunodeficiency virus-1 and human T-lymphotropic virus type I (HTLV-I). Experiments to ascertain the impact of CD8+ cells on viral burden based on CD8+ cell frequency or specificity alone give inconsistent results. Here, an alternative approach was developed that directly quantifies the impact of CD8+ lymphocytes on HTLV-I proviral burden by measuring the rate at which HTLV-I-infected CD4+ cells were cleared by autologous CD8+ cells ex vivo. It was demonstrated that CD8+ cells reduced the lifespan of infected CD4+ cells to 1 day, considerably shorter than the 30 day lifespan of uninfected cells in vivo. Furthermore, it was shown that HTLV-I-infected individuals vary considerably in the rate at which their CD8+ cells clear infected cells, and that this was a significant predictor of their HTLV-I proviral load. Forty to 50 % of between-individual variation in HTLV-I proviral load was explained by variation in the rate at which CD8+ cells cleared infected cells. This novel approach demonstrates that CD8+ cells are a major determinant of HTLV-I proviral load. This assay is applicable to quantifying the CD8+ cell response to other viruses and malignancies and may be of particular importance in assessing vaccines.

    • Supplementary material is available in JGV Online.

    • ↵†These authors contributed equally to this work.

    INTRODUCTION

    Human T-lymphotropic virus type I (HTLV-I) is a persistent retrovirus that infects 10–20 million people worldwide. The majority of infected individuals remain lifelong, asymptomatic carriers (ACs) of the virus. However, 2–3 % of infected individuals develop a progressive inflammation of the central nervous system called HTLV-I-associated myelopathy/tropical spastic paraparesis (HAM/TSP; Osame et al., 1986). HTLV-I infection is also associated with adult T-cell leukaemia (ATL) and a range of other inflammatory diseases (Nagai & Osame, 2003). It is not known why some individuals develop HAM/TSP and others remain asymptomatic. One factor thought to be associated with an increased risk of disease is a high viral burden. Viral burden, measured as the proviral load (number of HTLV-I proviruses per 100 PBMC) is approximately constant within an individual over time (Matsuzaki et al., 2001). However, between-individuals, proviral load varies considerably from below the detection limit of 0·001 up to 40 % of PBMC infected (Nagai et al., 1998). The prevalence of HAM/TSP increases strikingly once the proviral load exceeds 1 % (Nagai et al., 1998). The identification of the determinants of an individual's proviral load is clearly important to understanding why some individuals remain healthy and others develop HAM/TSP.

    One possible determinant of viral burden is the host cellular immune response to HTLV-I. In most infected individuals, the HTLV-I-specific CD8+ lymphocyte response is large and chronically activated (Biddison et al., 1997; Daenke et al., 1996). The role of this response is controversial. Some evidence indicates that a high frequency of HTLV-I-specific CD8+ cells is pathogenic and causes the inflammation associated with HAM/TSP (Bieganowska et al., 1999; Greten et al., 1998; Hoger et al., 1997; Jacobson, 2002); conversely, other data suggest that the HTLV-I-specific CD8+ cell response reduces proviral load and is associated with a lower risk of disease (Bangham, 2000; Jeffery et al., 1999, 2000; Niewiesk et al., 1994). These contradictory results may reflect the limitations of traditional measures of CD8+ cell function: it is increasingly being realized that the size of the virus-specific CD8+ cell response is not the only factor determining its impact on the virus in vivo (Rowland-Jones et al., 2001; van Baalen et al., 2002; Yang, 2003; Yang et al., 2003; Zhang et al., 2003). Other CD8+ cell attributes such as T-cell receptor avidity, specificity and cell maturation state, as well as target cell attributes including efficiency of epitope processing and presentation and phenotype, will all affect the ability of CD8+ cells to control a viral infection. Simply measuring specific CD8+ cell frequency by tetramer analysis, or by using additional functional measures such as cytokine production, fails to account for this complexity. Chromium release assays do not accurately reflect the complexity of the in vivo cytotoxic response, both because of the non-physiological nature of the target cells and because of the usual requirement for prolonged culture in vitro. Even if all key CD8+ cell attributes (frequency, avidity, phenotype etc.) are measured simultaneously, it is still not possible to synthesize this data into a single index of CD8+ cell antiviral efficacy. Thus, although great advances have been made in techniques to measure CD8+ cell function, all current assays measure individual attributes in isolation and therefore cannot capture the overall impact of the CD8+ lymphocyte response on long-term viral load.

    The failure of current assays of the CD8+ cell response is demonstrated by the use of specific CD8+ cell frequency as a measure of in vivo CD8+ cell function in both HTLV-I and HIV-1. It has been argued that, if the specific CD8+ T-cell response is important in controlling persistent viral infection, individuals with a large virus-specific CD8+ cell frequency would be expected to have a low viral burden; that is, a negative correlation between virus-specific CD8+ cell frequency and viral burden would be observed. However, in HIV-1, virus-specific CD8+ cell frequency as measured by tetramers and/or IFNγ ELISpot has shown positive, negative and zero correlations with HIV-1 viral load (Betts et al., 2001; Ogg et al., 1998). In HTLV-I, similarly divergent results have been obtained (Goon et al., 2004; Kubota et al., 2000).

    The aim of this study was to develop a composite measure of CD8+ T-cell mediated antiviral efficacy that directly measures the combined impact of key determinants of CD8+ cell function, including virus-specific cell frequency, cytolytic ability, scope of epitope recognition and efficiency of antigen presentation. We achieved this by measuring the rate at which naturally, endogenously infected cells were cleared by autologous CD8+ cells ex vivo. Having developed such an assay we use it to answer two questions: (i) how important is the CD8+ cell response in controlling HTLV-I proviral burden, and (ii) what is the relationship between CD8+ cell efficacy and HAM/TSP?

    METHODS

    Patients.

    All patients attended the HTLV-I clinic at St Mary's Hospital, London and gave informed consent. The study was approved by the Local Research Ethics Committee of St Mary's Hospital NHS Trust. HTLV-I infection was confirmed by the presence of antibodies to HTLV-I Gag and Env antigens in sera by Western blot (HTLV blot 2·4; Genelabs). Diagnosis of HAM/TSP was made following World Health Organization criteria. None of the patients was receiving antiviral or immunosuppressive therapy at the time of the study. Patient data are summarized in Table 1⇓.

    Table 1.

    Subject data and CD8+ cell antiviral efficacy estimates

    Disease duration is measured in years since first diagnosis. CD8+ cell antiviral efficacy is defined as the proportion of Tax+CD4+ cells cleared per CD8+ cell per day. Negative estimates of antiviral efficacy were likely to be due to noise contributing in such a way that near zero clearance rates appeared negative. A-C, Afro-Caribbean; Cauc, caucasian; nd, not done; Dis. Dur., disease duration.

    CD8+ cell assay.

    CD8+ cells were positively selected from thawed cryopreserved PBMC using magnetic microbeads (Miltenyi Biotec). The CD8+ and CD8− fractions were then washed twice, resuspended in standard culture medium (RPMI 1640 supplemented with 10 % fetal calf serum, 2 mM l-glutamine, 100 IU penicillin ml−1 and 100 μg streptomycin ml−1) and aliquotted (total volume 1 ml) into 5 ml round-bottomed, vented capped tubes at three to six different CD8+ : CD8− ratios (lower, including and higher than the subject's normal ratio) depending on cell numbers available. No mitogens, cytokines or artificial peptides were added. When required, concanamycin A (CMA; Sigma) was added at a final concentration of 20 nM. After 18 h culture at 37 °C, 5 % CO2, the cells were washed in PBS, fixed for 20 min at room temperature in 2 % paraformaldehyde (pH 7·4; Sigma), washed, then surface stained for CD4 and CD8 antigens by incubation at room temperature for 20 min in PBS/7 % normal goat serum with relevant mAbs (15 μg PC5-conjugated anti-CD4 and ECD-conjugated anti-CD8 ml−1; Beckman Coulter). The cells were washed once and stained intracellularly for the HTLV-I early protein Tax (Lee et al., 1989), as described by Hanon et al. (2000a), then analysed by flow cytometry on a Coulter EPICS XL. Thirty thousand events were routinely collected. All assays were done in duplicate.

    The resulting data were analysed using non-linear regression. The antiviral efficacy, i.e. the rate at which Tax+CD4+ cells were cleared, was estimated in each subject using the following model:

    Figure image not available in archive
    where y is the proportion of CD4+ cells expressing Tax (i.e Tax+CD4+ cells/CD4+ cells), c is the rate of increase of Tax expression, ε is the CD8+ cell-mediated antiviral efficacy and z is the proportion of lymphocytes that are CD8+. This model was solved analytically and fitted to the data using non-linear least squares regression, providing an estimate of the antiviral efficacy (ε) in each individual. This model does not assume that CD8+ cells reduce the proportion of Tax-expressing cells, since ε is free to be negative or zero as well as positive. At time zero the proportion of CD4+ cells expressing Tax is usually below the detection limit. In the absence of CD8+ cells it then increases over time. We model this empirically using a constant rate of Tax expression (c); more complex Tax expression dynamics are discussed below.

    Quantification of HTLV-I proviral load.

    Cells were available to measure proviral load in 16 subjects. The HTLV-I proviral load and β-globin copy number of each sample was quantified using real-time quantitative PCR (primers as in Kwok et al., 1988; Seiki et al., 1985) SYBR Green 1 dye incorporation in a Roche LightCycler. SYBR Green 1 incorporation was detected at 85 °C at the end of each of 40 amplification cycles. Standard curves were generated for both PCRs using genomic DNA from C10-PBLs carrying a single HTLV-I provirus. Sample copy numbers were estimated by interpolation from the standard curves. Proviral load was expressed as percentage PBMC infected.

    Saturation.

    In all cases model 1 fitted the data well. However, in four of our data series there were no surviving Tax+CD4+ cells following incubation with any but the lowest frequency of CD8+ cells, either because too many CD8+ cells were added and/or because the rate of Tax+CD4+ cell clearance was high in these subjects. These estimates of antiviral efficacy were therefore excessively dependent on a single data point and were excluded from the study.

    Effect of model choice.

    We considered two alternative models to see the effect of the choice of model on the results obtained. The first was a simple linear model (2) (derived from a finite difference equation with time step Δt=18 h) where target cells were assumed to be at a constant level in the absence of CD8+ cells and to decline instantaneously in the presence of CD8+ cells:

    Figure image not available in archive

    The second alternative model was a simple exponential decay model (3) in which target cells were assumed to be at a constant level in the absence of CD8+ cells and to decline exponentially in the presence of CD8+ cells:

    Figure image not available in archive

    Antiviral efficacy estimates generated by each model were strongly correlated with estimates from the basic model 1 (basic model and linear model P<10−14; basic model and exponential model P<10−14; Spearman's rank correlation two-tailed test) despite the simplifying assumptions in the linear and exponential models.

    Effect of viral protein expression dynamics.

    The assumption, inherent in the basic model 1, that expression of the viral protein Tax increases at a constant rate over time is clearly an over simplification. We measured Tax expression in CD4+ cells over 18 h in eight of our subjects (five HAM/TSP and three ACs) and found no systematic difference in the time course of Tax expression between HAM/TSP patients and ACs, or between subjects with a high or low proviral load. We estimated the antiviral efficacy in five of these subjects – once with the basic model 1 and once taking detailed dynamics of Tax expression into account. The results from the two methods were perfectly correlated, i.e. the rank order of the CD8+ cell antiviral efficacy was identical. Between-individual variation in the dynamics of Tax expression was therefore unlikely to systematically affect our results.

    Supporting information.

    Experimental data showing the rate of clearance of autologous and allogenic target cells and an estimate of the number of infected cells cleared per HTLV-I-specific CD8+ cell per day are included in the Supplementary material in JGV Online.

    RESULTS

    Developing a composite measure of CD8+ cell-mediated antiviral efficacy

    The CD8+ cell-mediated ‘antiviral efficacy’ was defined as the rate of clearance of HTLV-I-infected Tax-expressing CD4+ cells by autologous CD8+ cells ex vivo. This CD8+ cell antiviral efficacy is calculated for the CD8+ cell population as a whole, reflecting both the frequency and cytotoxicity of HTLV-I-specific CD8+ cells.

    CD4+ cells are the predominant infected cell population in HTLV-I (Richardson et al., 1997; Yamano et al., 2004). In endogenously infected cells, the early viral protein Tax is usually undetectable immediately ex vivo, but increases after short-term culture (Hanon et al., 2000a). We quantified the CD8+ cell-mediated antiviral efficacy by measuring the proportion of CD4+ cells expressing Tax, after exposure to different numbers of autologous CD8+ cells. To do this, PBMC were separated by CD8-positive selection into CD8+ and CD8− cell fractions. Neither cell population was artificially expanded or stimulated ex vivo, and no exogenous cytokines or peptides were added. The cell populations were immediately recombined in different proportions, to include CD8+ : CD8− ratios above, below and at the physiological ratio for that individual. Cells were then co-cultured for 18 h, after which the proportion of Tax+CD4+ cells was determined by flow cytometry. A co-culture period of 18 h was chosen to provide sufficient time for infected cells to express Tax (which normally peaks at 6–12 h) and for Tax-expressing cells to be lysed. All assays were performed in duplicate.

    The resulting data – the proportion of Tax+CD4+ cells as a function of CD8+ cell frequency – was analysed mathematically. The technique used was similar to that used to analyse BrdU and deuterated glucose lymphocyte labelling data (Debacq et al., 2002; Mohri et al., 1998). That is, a simple mathematical model was formulated – in this case reflecting the expression of Tax by infected cells and the clearance of Tax-expressing cells by CD8+ cells – and fitted to the data using non-linear regression. In this way we estimated the CD8+ cell antiviral efficacy, that is the rate of CD8+ cell-mediated clearance of Tax-expressing cells. An example experiment is shown in Fig. 1⇓. Since we estimated infected CD4+ cell survival as a function of CD8+ cell frequency, we automatically excluded CD8+ cell-independent death of infected cells (e.g. natural cell death).

    Figure image not available in archive
    Fig. 1.

    CD8+ antiviral efficacy assay: an example. The figure shows an example of the antiviral efficacy assay. The proportion of CD4+ lymphocytes that were Tax+ following 18 h co-culture with different proportions of CD8+ lymphocytes was measured. The model (equation 1, Methods) was fitted to this data and in this way the rate of clearance of Tax+CD4+ cells per day per CD8+ cell (antiviral efficacy) was estimated. This was repeated in the same subject. Repeat 1, □ observed data; dashed line, best theoretical fit. Repeat 2, ▴ observed data; solid line, best theoretical fit.

    It was found that the frequency and mean fluorescence intensity of Tax expression in CD4+ lymphocytes both decreased following co-culture with increasing numbers of CD8+ cells (Fig. 2⇓). Previous evidence (Hanon et al., 2000a, b) indicated this decrease was principally mediated by perforin-dependent cell lysis. To confirm this, we measured the change in antiviral efficacy in the presence and absence of 20 nM of the perforin inhibitor, CMA (Kataoka et al., 1996). In samples from two unrelated subjects, treatment with CMA reduced the rate at which Tax+ cells were cleared by CD8+ cells in each case by 80 % (data not shown), indicating that the major CD8+ cell-mediated antiviral pathway was perforin-dependent. We also compared the rate at which autologous and allogeneic Tax-expressing CD4+ target cells were cleared by CD8+ cells from a range of HTLV-I-seropositive subjects. Since antiviral efficacy, as defined here, is a measure of specific clearance of HTLV-I-infected cells, any non-HTLV-I-associated allogeneic killing does not contribute to the result. We found that HTLV-I-infected allogeneic target cells were cleared at a much lower rate than autologous target cells, and that the rate of clearance decreased as the extent of HLA class I mismatch increased (see Supplementary material in JGV Online, Table S1). These results indicate that the dominant mechanism of reduction in Tax expression by CD8+ cells is perforin-dependent and MHC class I-restricted; consistent with classical CD8+ cell-mediated lysis of virally infected cells. However, our assay is not just restricted to perforin-dependent lysis: it will capture all CD8+ cell-mediated antiviral effects that result in a reduction of the number of Tax+ cells in 18 h, e.g. Fas-induced apoptosis. CD8+ cell-mediated antiviral effects that occur over a longer time, such as IFNγ suppression of virus replication, will not be captured in this assay.

    Figure image not available in archive
    Fig. 2.

    Reduction in Tax expression with increasing CD8+ cell frequency. Frequency and intensity of Tax expression in CD4+ cells decreases as a function of increasing CD8+ cell frequency during 18 h co-culture. Example data from subject HBE (see Table 1) are shown. CD8+ cell frequency: thin line, low (2·9 %); medium line, intermediate (17·6 %); thick line, high (33·9 %). Frequency of Tax+CD4+ cells (as a proportion of CD4+ cells): 3·2 % (low CD8), 0·9 % (intermediate CD8), 0·2 % (high CD8). Mean fluorescence intensity of Tax staining in Tax+CD4+ cells: 43·3 (low CD8), 33·4 (intermediate CD8), 26·7 (high CD8).

    Quantifying antiviral efficacy in an HTLV-I-infected cohort

    This novel assay was applied to 23 HTLV-I-seropositive individuals, of whom nine were ACs and 14 had HAM/TSP. The results are summarized in Table 1⇑. All assays were done in duplicate and the results expressed as the mean. The agreement between duplicate assays was good (Pearson's correlation coefficient r=0·94, P<1×10−11 two-tailed test) and satisfied the reproducibility criteria of Bland and Altman (Bland & Altman, 1986) with over 95 % of the difference between the two repeats lying within ±2 standard deviations of zero. The results obtained were not dependent on the exact form of the model used to analyse the data (see Methods). The inter-individual variation in CD8+ antiviral efficacy was large, ranging from −1·3 to 42·7 % Tax+CD4+ cells cleared per CD8+ cell per day (negative estimates were likely to be due to noise contributing in such a way that near zero clearance rates appeared negative). No significant difference in the efficacy of the CD8+ cell response between HAM/TSP patients and ACs was identified (Fig. 3⇓, Mann–Whitney two-tailed U-test P=0·6).

    Figure image not available in archive
    Fig. 3.

    Antiviral efficacy of the CD8+ cell response in 23 HTLV-I-seropositive individuals. The CD8+ cell antiviral efficacy was measured in 23 HTLV-I-seropositive individuals, of whom nine were ACs (▴) and 14 had HAM/TSP (□). There was no significant difference in antiviral efficacy between the two groups.

    Do CD8+ cells control proviral load in vivo?

    Using data from seven ACs and nine HAM/TSP patients, we calculated the correlation between the CD8+ cell antiviral efficacy and proviral load. In both groups, a significant negative correlation was observed (Spearman's rank correlation two-tailed test; AC: rs=−0·76 P=0·03; HAM/TSP: rs=−0·68 P=0·04). That is, individuals with a high rate of clearance of infected cells tended to have a low proviral load (Fig. 4⇓).

    Figure image not available in archive
    Fig. 4.

    Proviral load plotted against antiviral efficacy for seven ACs and nine HAM/TSP patients. A negative correlation between proviral load and antiviral efficacy was observed in both ACs and HAM/TSP patients (P=0·03 and 0·04, respectively; Spearman's rank correlation two-tailed test). Furthermore, for a given rate of clearance the proviral load was lower in ACs than in HAM/TSP patients (P=0·03; permutation two-tailed test). Eight AC data points from seven individuals are illustrated. One subject (HS – see Table 1) appears twice because blood samples from two different time points with two different proviral loads were available. Including HS twice did not alter our conclusions. We found a significant negative correlation between proviral load and antiviral efficacy if only the first point or only the second point was included [excluding first HS point: rs=−1, P<0·01; excluding second HS point: rs=−0·78, P=0·036]. We also found that at a given rate of clearance proviral load was lower in ACs compared with HAM/TSP patients (P=0·03), regardless of which HS point we included.

    A negative correlation between proviral load and CD8+ cell antiviral efficacy does not necessarily imply that the antiviral efficacy determines proviral load; antiviral efficacy may be correlated with another variable that decreases proviral load, or a high proviral load may decrease the CD8+ cell efficacy, e.g. by functional inactivation of CD8+ cells (Oxenius et al., 2002). However, it seems unlikely that a high proviral load per se reduces the antiviral efficacy because HAM/TSP patients consistently have higher proviral loads than ACs at the same antiviral efficacy. Furthermore, CD8+ cells efficiently reduce Tax expression ex vivo (Hanon et al., 2000a, b; Figs 1 and 2⇑⇑) and have been shown to reduce viral load in a number of other virus infections. We conclude that the observed negative correlation between antiviral efficacy and proviral load results from reduction of the proviral load in vivo by CD8+ cells.

    To quantify the importance of the CD8+ cell-mediated antiviral response in determining the proviral load, we used the coefficient of determination (r2) (Howell, 1992) and a simple empirical model of the relationship between proviral load and antiviral efficacy. The proviral load appeared to fall exponentially with increasing CD8+ cell efficacy, i.e. log[proviral load] plotted against the antiviral efficacy gave an approximately straight line. Using this relationship we found that r2=0·43 in the ACs and r2=0·5 in the HAM/TSP patients. These figures indicate that, within both patient groups, 40–50 % of the observed inter-individual variation in proviral load could be accounted for by variation in the rate at which CD8+ cells cleared HTLV-I-infected cells. This indicates that variation in CD8+ cell antiviral efficacy is the largest single determinant of inter-individual variation in HTLV-I proviral load within each clinical group.

    Relationship between CD8+ cell antiviral efficacy, proviral load and disease

    While there was no significant difference in antiviral efficacy between the HAM/TSP patients and the ACs, a relationship between CD8+ cell efficacy, proviral load and disease status was demonstrated: at any given antiviral efficacy, subjects with HAM/TSP had a significantly higher proviral load than ACs (P=0·03 two-tailed permutation test, Fig. 4⇑). This observation implies that one or more factor(s) associated with the increased proviral load in HAM/TSP patients is also specifically associated with the disease HAM/TSP. Thus, within this cohort, antiviral efficacy and proviral load are more accurate predictors of the disease status of HTLV-I-seropositive individuals than proviral load alone.

    Estimating the number of infected cells cleared by an HTLV-I-specific CD8+ cell

    Multiplying an individual's antiviral efficacy by the frequency of CD8+ cells gives the rate at which Tax+CD4+ cells are cleared per day (i.e. the death rate of Tax+ cells). We applied this to the full cohort of 23 patients, and found that the median death rate of Tax+CD4+ cells was 4·25 % per hour (102 % per day). Importantly, this is the death rate of Tax+CD4+cells due to the presence of CD8+ cells. Background cell death due, for example, to cytokine deprivation ex vivo is excluded from this estimate. A death rate of 102 % per day is equivalent to a lifespan for a Tax+CD4+ cell of approximately 1 day, which is considerably shorter than the 30-day natural lifespan of CD4+ cells in vivo in uninfected individuals (Macallan et al., 2003).

    The antiviral efficacy provides a measure of the rate at which an average CD8+ cell clears Tax-expressing cells. These data can be used to make order of magnitude estimates of the absolute number of cells cleared by a single HTLV-I-specific CD8+ lymphocyte in vivo. We assumed that the ex vivo rates of Tax expression and clearance reflected in vivo kinetics and estimated the frequency of HTLV-I-specific CD8+ cells from class I tetramer binding assays and ELISpot assays (see Supplementary material in JGV Online for details of calculation). The results show that a single HTLV-I-specific CD8+ cell clears approximately five infected cells per day. The total number of infected CD4+ cells cleared per day by the CD8+ lymphocyte response is of the order of 2×109 cells. This figure is similar to the estimates in HIV-1 infection (Perelson et al., 1996). HTLV-I infection is not associated with the severe immune suppression seen in HIV-1 infection, suggesting that the fast turnover of infected CD4+ cells per se is insufficient to cause marked immune deficiency.

    DISCUSSION

    The biological significance of the CD8+ cell response in persistent viral infections such as HTLV-I and HIV-1 remains controversial and there are many conflicting data. In particular, in HTLV-I it is not known if CD8+ cells play a biologically significant role in controlling proviral load or even whether a strong CD8+ cell response protects from or causes HAM/TSP. Part of the difficulty lies in the current assays of the CD8+ lymphocyte response (Yang, 2003). The efficacy of the CD8+ lymphocyte response is a function of many factors, including the frequency of the virus-specific CD8+ cells, the number and identity of epitopes recognized, the lytic capacity of specific cells, their activation status and phenotype and also the efficiency with which target cells express and present antigen. Techniques exist to assay these attributes separately. However, it is difficult to determine the relative importance of each attribute and to integrate these data to ascertain the composite effectiveness of the CD8+ lymphocyte response. Furthermore, these techniques are often highly non-physiological. For example, exogenous addition of artificial peptides is common in chromium release, ELISpot and LYSISpot (Snyder et al., 2003) assays, and does not allow for natural processing and presentation of viral proteins in the context of productive infection.

    The CD8+ cell assay described here has the advantage that it measures clearance of natural target cells (a subject's HTLV-I-infected CD4+ lymphocytes) by the relevant effectors (autologous CD8+ cells), and naturally processed viral peptides are presented in the appropriate context. The assay excludes the effects of any redundant CD8+ cell clones directed against viral epitopes that have undergone mutation and are no longer present in the viral genome; such CD8+ cell clones remain detectable by tetramer staining, LYSISpot or ELISpot. The assay also avoids a bias towards immunodominant epitopes, which are not necessarily the most important epitopes for the control of viral infection in vivo. Thus, in contrast to previous measures of the CD8+ response, this approach provides a physiological index of the overall impact of CD8+ cells on virally infected cells and allows direct comparison of the efficacy of the CD8+ cell response between individuals.

    Inevitable disadvantages of this assay are that peripheral blood cells may not be representative of the majority of lymphocytes and that ex vivo cell behaviour may not reflect in vivo behaviour. Both these differences will be manifested as a reduced correlation between proviral load and CD8+ cell efficacy, i.e. our analysis will err on the side of caution. We have developed this assay in the context of HTLV-I infection but it could be adapted to measure the efficacy of the CD8+ lymphocyte response to a range of targets – allogeneic tissue, leukaemic cells or alternative pathogen-infected cells. It could be particularly important in assessing the efficacy of vaccines designed to elicit a protective CD8+ cell response (Pantaleo & Koup, 2004).

    The most commonly used measure of the ‘strength’ of the antiviral response is the frequency of virus-specific cells. Previous studies (Kubota et al., 2000; Wodarz et al., 2001) have failed to find a consistent relationship between proviral load and HTLV-I-specific CD8+ cell frequency in both AC and HAM/TSP subject groups. When correlations have been observed they have tended to be positive, which has been interpreted as indicating that the CD8+ cell response passively reflects, rather than controls, proviral load. In contrast, using the assay presented here, we found a significant negative correlation between proviral load and CD8+ cell antiviral efficacy in both ACs and HAM/TSP patients. Indeed, antiviral efficacy is a much better predictor of proviral load than HTLV-I-specific CD8+ cell frequency. Quantifying this relationship, we found that within the AC and HAM/TSP groups 40–50 % of the between-individual variation in proviral load was attributable to variation in the antiviral efficacy. That is, the rate at which infected cells were cleared by CD8+ cells was a major determinant of long-term viral burden. Other possible determinants include the antibody response, the strain of infecting virus and the route of infection.

    We also explored the relationship between CD8+ cell antiviral efficacy and clinical outcome. We found no difference in antiviral efficacy between HAM/TSP patients and ACs. However, there was no significant difference in proviral load between the HAM/TSP and AC groups in our cohort, and as the protective effect of the CD8+ response is hypothesized to operate through a reduction in proviral load (Asquith & Bangham, 2000; Jeffery et al., 1999, 2000; Niewiesk et al., 1994; Vine et al., 2004), our study did not have the power to detect such a protective effect. We suggest that, given that a low CD8+ cell antiviral efficacy is significantly associated with a high proviral load and given that, in large population studies, a high viral load is associated with HAM/TSP (Nagai et al., 1998), then it is likely that in a larger patient cohort we would observe an association between low CD8+ cell antiviral efficacy and HAM/TSP. Interestingly, we did find an unexpected relationship between CD8+ cell antiviral efficacy and clinical status, in that there was a clear separation of the antiviral efficacy–proviral load curves between the HAM/TSP subjects and the ACs (Fig. 4⇑). So, at any given antiviral efficacy, HAM/TSP patients had a significantly higher proviral load than ACs. Therefore, within our patient group the combination of proviral load and antiviral efficacy together was a significantly better predictor of disease status than proviral load alone. This suggests that there is an extra factor in addition to CD8+ cell antiviral efficacy that determines an individual's proviral load and, moreover, that this factor differs systematically between HAM/TSP patients and ACs.

    Several observations have led to the suggestion that the CD8+ lymphocyte response has negligible benefit in HTLV-I infection: proviral loads can be extremely high, suggesting a breakdown of immune control; there appears to be little presentation of viral peptide by infected cells in vivo; and individuals with a high proviral load tend to have a high HTLV-I-specific CD8+ cell frequency (Bieganowska et al., 1999; Kubota et al., 2000). However, these arguments are based on indirect inference. The quantification of CD8+ cell antiviral efficacy that we have presented here indicates that CD8+ cells play a major role in the control of HTLV-I proviral load.

    Acknowledgments

    We are very grateful to Paul Klenerman, Keith Gould and Andrew George for helpful discussions and comments on the text. We also thank the patients of St Mary's Hospital for taking part in this study. This work was supported by the Wellcome Trust (UK) and the Leverhulme Trust. The authors have no conflicting financial interests.

    References

    1. Asquith, B. & Bangham, C. R. (2000). The role of cytotoxic T lymphocytes in human T-cell lymphotropic virus type 1 infection. J Theor Biol 207, 65–79.
    2. Bangham, C. R. (2000). The immune response to HTLV-I. Curr Opin Immunol 12, 397–402.
    3. Betts, M. R., Ambrozak, D. R., Douek, D. C., Bonhoeffer, S., Brenchley, J. M., Casazza, J. P., Koup, R. A. & Picker, L. J. (2001). Analysis of total human immunodeficiency virus (HIV)-specific CD4+ and CD8+ T-cell responses: relationship to viral load in untreated HIV infection. J Virol 75, 11983–11991.
    4. Biddison, W. E., Kubota, R., Kawanishi, T., Taub, D. D., Cruikshank, W. W., Center, D. M., Connor, E. W., Utz, U. & Jacobson, S. (1997). Human T cell leukemia virus type I (HTLV-I)-specific CD8+ CTL clones from patients with HTLV-I-associated neurologic disease secrete proinflammatory cytokines, chemokines, and matrix metalloproteinase. J Immunol 159, 2018–2025.
    5. Bieganowska, K., Hollsberg, P., Buckle, G. J. & 11 other authors (1999). Direct analysis of viral-specific CD8+ T cells with soluble HLA-A2/Tax11-19 tetramer complexes in patients with human T cell lymphotropic virus-associated myelopathy. J Immunol 162, 1765–1771.
    6. Bland, J. M. & Altman, D. G. (1986). Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1, 307–310.
    7. Daenke, S., Kermode, A. G., Hall, S. E., Taylor, G., Weber, J., Nightingale, S. & Bangham, C. R. (1996). High activated and memory cytotoxic T-cell responses to HTLV-1 in healthy carriers and patients with tropical spastic paraparesis. Virology 217, 139–146.
    8. Debacq, C., Asquith, B., Kerkhofs, P., Portetelle, D., Burny, A., Kettmann, R. & Willems, L. (2002). Increased cell proliferation, but not reduced cell death, induces lymphocytosis in bovine leukemia virus-infected sheep. Proc Natl Acad Sci U S A 99, 10048–10053.
    9. Goon, P. K., Biancardi, A., Fast, N. & 8 other authors (2004). Human T cell lymphotropic virus (HTLV) type-1-specific CD8+ T cells: frequency and immunodominance hierarchy. J Infect Dis 189, 2294–2298.
    10. Greten, T. F., Slansky, J. E., Kubota, R., Soldan, S. S., Jaffee, E. M., Leist, T. P., Pardoll, D. M., Jacobson, S. & Schneck, J. P. (1998). Direct visualization of antigen-specific T cells: HTLV-1 Tax11-19- specific CD8+ T cells are activated in peripheral blood and accumulate in cerebrospinal fluid from HAM/TSP patients. Proc Natl Acad Sci U S A 95, 7568–7573.
    11. Hanon, E., Hall, S., Taylor, G. P. & 7 other authors (2000a). Abundant tax protein expression in CD4+ T cells infected with human T-cell lymphotropic virus type I (HTLV-I) is prevented by cytotoxic T lymphocytes. Blood 95, 1386–1392.
    12. Hanon, E., Stinchcombe, J. C., Saito, M., Asquith, B. E., Taylor, G. P., Tanaka, Y., Weber, J. N., Griffiths, G. M. & Bangham, C. R. (2000b). Fratricide among CD8+ T lymphocytes naturally infected with human T cell lymphotropic virus type I. Immunity 13, 657–664.
    13. Hoger, T. A., Jacobson, S., Kawanishi, T., Kato, T., Nishioka, K. & Yamamoto, K. (1997). Accumulation of human T lymphotropic virus (HTLV)-I-specific T cell clones in HTLV-I-associated myelopathy/tropical spastic paraparesis patients. J Immunol 159, 2042–2048.
    14. Howell, D. C. (1992). Statistical Methods for Psychology, 3rd edn. Belmont: Duxbury Press.
    15. Jacobson, S. (2002). Immunopathogenesis of human T cell lymphotropic virus type I-associated neurologic disease. J Infect Dis 186, S187–S192.
    16. Jeffery, K. J., Usuku, K., Hall, S. E. & 14 other authors (1999). HLA alleles determine human T-lymphotropic virus-I (HTLV-I) proviral load and the risk of HTLV-I-associated myelopathy. Proc Natl Acad Sci U S A 96, 3848–3853.
    17. Jeffery, K. J., Siddiqui, A. A., Bunce, M. & 8 other authors (2000). The influence of HLA class I alleles and heterozygosity on the outcome of human T cell lymphotropic virus type I infection. J Immunol 165, 7278–7284.
    18. Kataoka, T., Shinohara, N., Takayama, H., Takaku, K., Kondo, S., Yonehara, S. & Nagai, K. (1996). Concanamycin A, a powerful tool for characterization and estimation of contribution of perforin- and Fas-based lytic pathways in cell-mediated cytotoxicity. J Immunol 156, 3678–3686.
    19. Kubota, R., Kawanishi, T., Matsubara, H., Manns, A. & Jacobson, S. (2000). HTLV-I specific IFN-γ+ CD8+ lymphocytes correlate with the proviral load in peripheral blood of infected individuals. J Neuroimmunol 102, 208–215.
    20. Kwok, S., Ehrlich, G., Poiesz, B., Kalish, R. & Sninsky, J. J. (1988). Enzymatic amplification of HTLV-I viral sequences from peripheral blood mononuclear cells and infected tissues. Blood 72, 1117–1123.
    21. Lee, B., Tanaka, Y. & Tozawa, H. (1989). Monoclonal antibody defining tax protein of human T-cell leukemia virus type-I. Tohoku J Exp Med 157, 1–11.
    22. Macallan, D. C., Asquith, B., Irvine, A. J. & 7 other authors (2003). Measurement and modeling of human T cell kinetics. Eur J Immunol 33, 2316–2326.
    23. Matsuzaki, T., Nakagawa, M., Nagai, M. & 7 other authors (2001). HTLV-I proviral load correlates with progression of motor disability in HAM/TSP: analysis of 239 HAM/TSP patients including 64 patients followed up for 10 years. J Neurovirol 7, 228–234.
    24. Mohri, H., Bonhoeffer, S., Monard, S., Perelson, A. S. & Ho, D. D. (1998). Rapid turnover of T lymphocytes in SIV-infected rhesus macaques. Science 279, 1223–1227.
    25. Nagai, M. & Osame, M. (2003). Human T-cell lymphotropic virus type I and neurological diseases. J Neurovirol 9, 228–235.
    26. Nagai, M., Usuku, K., Matsumoto, W. & 8 other authors (1998). Analysis of HTLV-I proviral load in 202 HAM/TSP patients and 243 asymptomatic HTLV-I carriers: high proviral load strongly predisposes to HAM/TSP. J Neurovirol 4, 586–593.
    27. Niewiesk, S., Daenke, S., Parker, C. E., Taylor, G., Weber, J., Nightingale, S. & Bangham, C. R. (1994). The transactivator gene of human T-cell leukemia virus type I is more variable within and between healthy carriers than patients with tropical spastic paraparesis. J Virol 68, 6778–6781.
    28. Ogg, G. S., Jin, X., Bonhoeffer, S. & 12 other authors (1998). Quantitation of HIV-1-specific cytotoxic T lymphocytes and plasma load of viral RNA. Science 279, 2103–2106.
    29. Osame, M., Usuku, K., Izumo, S., Ijichi, N., Amitani, H., Igata, A., Matsumoto, M. & Tara, M. (1986). HTLV-I associated myelopathy, a new clinical entity. Lancet 1, 1031–1032.
    30. Oxenius, A., Sewell, A. K., Dawson, S. J. & 9 other authors (2002). Functional discrepancies in HIV-specific CD8+ T-lymphocyte populations are related to plasma virus load. J Clin Immunol 22, 363–374.
    31. Pantaleo, G. & Koup, R. A. (2004). Correlates of immune protection in HIV-1 infection: what we know, what we don't know, what we should know. Nat Med 10, 806–810.
    32. Perelson, A. S., Neumann, A. U., Markowitz, M., Leonard, J. M. & Ho, D. D. (1996). HIV-1 dynamics in vivo: virion clearance rate, infected cell life-span, and viral generation time. Science 271, 1582–1586.
    33. Richardson, J. H., Hollsberg, P., Windhagen, A., Child, L. A., Hafler, D. A. & Lever, A. M. (1997). Variable immortalizing potential and frequent virus latency in blood-derived T-cell clones infected with human T-cell leukemia virus type I. Blood 89, 3303–3314.
    34. Rowland-Jones, S. L., Pinheiro, S., Kaul, R. & 9 other authors (2001). How important is the ‘quality’ of the cytotoxic T lymphocyte (CTL) response in protection against HIV infection? Immunol Lett 79, 15–20.
    35. Seiki, M., Hikikoshi, A., Taniguchi, T. & Yoshida, M. (1985). Expression of the pX gene of HTLV-I: general splicing mechanism in the HTLV family. Science 228, 1532–1534.
    36. Snyder, J. E., Bowers, W. J., Livingstone, A. M., Lee, F. E., Federoff, H. J. & Mosmann, T. R. (2003). Measuring the frequency of mouse and human cytotoxic T cells by the Lysispot assay: independent regulation of cytokine secretion and short-term killing. Nat Med 9, 231–235.
    37. van Baalen, C. A., Guillon, C., van Baalen, M., Verschuren, E. J., Boers, P. H., Osterhaus, A. D. & Gruters, R. A. (2002). Impact of antigen expression kinetics on the effectiveness of HIV-specific cytotoxic T lymphocytes. Eur J Immunol 32, 2644–2652.
    38. Vine, A. M., Heaps, A. G., Kaftantzi, L. & 10 other authors (2004). The role of CTLs in persistent viral infection: cytolytic gene expression in CD8+ lymphocytes distinguishes between individuals with a high or low proviral load of human T cell lymphotropic virus type 1. J Immunol 173, 5121–5129.
    39. Wodarz, D., Hall, S. E., Usuku, K., Osame, M., Ogg, G. S., McMichael, A. J., Nowak, M. A. & Bangham, C. R. (2001). Cytotoxic T-cell abundance and virus load in human immunodeficiency virus type 1 and human T-cell leukaemia virus type 1. Proc R Soc Lond B Biol Sci 268, 1215–1221.
    40. Yamano, Y., Cohen, C. J., Takenouchi, N., Yao, K., Tomaru, U., Li, H. C., Reiter, Y. & Jacobson, S. (2004). Increased expression of human T lymphocyte virus type I (HTLV-I) Tax11-19 peptide-human histocompatibility leukocyte antigen A*201 complexes on CD4+ CD25+ T cells detected by peptide-specific, major histocompatibility complex-restricted antibodies in patients with HTLV-I-associated neurologic disease. J Exp Med 199, 1367–1377.
    41. Yang, O. O. (2003). Will we be able to ‘spot’ an effective HIV-1 vaccine? Trends Immunol 24, 67–72.
    42. Yang, O. O., Sarkis, P. T., Trocha, A., Kalams, S. A., Johnson, R. P. & Walker, B. D. (2003). Impacts of avidity and specificity on the antiviral efficiency of HIV-1-specific CTL. J Immunol 171, 3718–3724.
    43. Zhang, D., Shankar, P., Xu, Z. & 7 other authors (2003). Most antiviral CD8 T cells during chronic viral infection do not express high levels of perforin and are not directly cytotoxic. Blood 101, 226–235.