In vivo T lymphocyte dynamics in humans and the impact of human T-lymphotropic virus 1 infection

Abstract

Human T-lymphotropic virus type 1 (HTLV-1) is a persistent CD4⁺ T-lymphotropic retrovirus. Most HTLV-1-infected individuals remain asymptomatic, but a proportion develop adult T cell leukemia or inflammatory disease. It is not fully understood how HTLV-1 persists despite a strong immune response or what determines the risk of HTLV-1-associated diseases. Until recently, it has been difficult to quantify lymphocyte kinetics in humans in vivo. Here, we used deuterated glucose labeling to quantify in vivo lymphocyte dynamics in HTLV-1-infected individuals. We then used these results to address four questions. (i) What is the impact of HTLV-1 infection on lymphocyte dynamics? (ii) How does HTLV-1 persist? (iii) What is the extent of HTLV-1 expression in vivo? (iv) What features of lymphocyte kinetics are associated with HTLV-1-associated myelopathy/tropical spastic paraparesis? We found that CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ T lymphocyte proliferation was elevated in HTLV-1-infected subjects compared with controls, with an extra 10¹² lymphocytes produced per year in an HTLV-1-infected subject. The in vivo proliferation rate of CD4⁺CD45RO⁺ cells also correlated with ex vivo viral expression. Finally, the inflammatory disease HTLV-1-associated myelopathy/tropical spastic paraparesis was associated with significantly increased CD4⁺CD45RO⁺ cell proliferation. We suggest that there is persistent viral gene expression in vivo, which is necessary for the maintenance of the proviral load and determines HTLV-1-associated myelopathy/tropical spastic paraparesis risk.

Human T-lymphotropic virus type 1 (HTLV-1) is an exogenous retrovirus that persistently infects 20–30 million people worldwide. It is the etiological agent of adult T cell leukemia and a range of inflammatory diseases including HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP), a chronic disease of the central nervous system. The majority of HTLV-1-infected individuals remain lifelong asymptomatic carriers (ACs) of the virus.

HTLV-1 persists despite the large HTLV-1-specific CD8⁺ cytotoxic T lymphocyte (CTL) response that is observed in most HTLV-1-infected individuals (1, 2). The observation that cell-free virus, viral protein, and viral mRNA are usually undetectable in the blood (3, 4) suggests that HTLV-1 is largely transcriptionally inactive, at least in the periphery. The viral proteins HBZ, Rex, and p30^II (5–7), as well as epigenetic modifications (8), have been shown to regulate HTLV-1 gene expression and could promote this viral latency. However, the observation of HTLV-1-specific CTL responses (1, 9, 10) indicates that there was certainly viral protein expression in the past and that there is probably a degree of ongoing viral expression. Although HTLV-1 reverse transcriptase has an error rate comparable with other retroviruses (11), the HTLV-1 genome is remarkably stable (12), indicating that very few rounds of replication via reverse transcription have occurred. Thus, it appears that HTLV-1 favors a mode of persistence in which most, but maybe not all, infected cells are silent at any one time and in which most new infected cells are produced by division of infected cells rather than de novo infectious events. Two questions arise from these observations. First, is there detectable persistent HTLV-1 expression in vivo? Second, if so, is the expression a mere epiphenomenon or does it contribute to the persistence of the virus in the chronic phase?

We consider two possible models of HTLV-1 persistence (Table 1). Both models allow for a degree of viral expression, and in both cases HTLV-1 propagates mainly by mitotic division. The main difference between the models is whether viral expression increases or decreases proviral load. In the first model, “passive clonal expansion of infected cells,” viral silence favors viral propagation because it enables infected cells to evade immune surveillance. In the second model, “active division of virus-expressing cells,” viral expression is essential for persistence because it drives infected cell division; several mechanisms by which the viral proteins Tax and p12^I could increase infected cell division have been identified including up-regulation of cellular genes involved in proliferation (13), deregulation of cell cycle checkpoints (14), and reduction in the threshold of T cell activation (15). Thus, in the first model, viral expression leads to a decrease in proviral load, whereas in the second, it leads to an increase; by extension, the first model predicts that clinical interventions that increase viral expression would be beneficial, by decreasing proviral load, whereas the second model would predict the converse. Distinguishing between the two models of persistence in humans in vivo has proved difficult: most observations such as undetectable levels of viral protein or viral genome stability are compatible with both models. One way to distinguish between the two models is to investigate lymphocyte kinetics.

Table 1.

Summary of the key differences between the two models of HTLV-1 persistence

Key features	Model 1: Passive proliferation of latently infected cells	Model 2: Active proliferation of Tax-expressing cells
HTLV-1 persistence	Latency is essential for persistence.	Tax-driven proliferation is essential for persistence.
Tax expression	Tax expression in vivo is an epiphenomenon. Tax expression ex vivo is an artefact.	Tax expression ex vivo is correlated with HTLV-1 gene expression in vivo.
Production of new infected cells	Latent integrated proviruses are “passively” replicated when the host cell divides. There is no selective proliferation of infected cells.	Tax protein drives infected cell proliferation. There is selective proliferation of infected cells.
HTLV-1-specific CTL	CTL have few targets and thus have minimal impact on proviral load.	CTL inhibit the key pathway of viral persistence and thus have a significant impact on proviral load.
HAM/TSP occurrence	HAM/TSP is associated with reactivation of HTLV-1 following a long period of viral and clinical latency.	HAM/TSP is associated with a high continuous rate of Tax expression.
Impact of increased Tax expression (other factors unchanged)	Proviral load would decrease (as previously latent cells would be exposed to immune surveillance).	Proviral load would increase (as Tax-driven proliferation would exceed CTL lysis).

Both models allow for a degree of viral expression, and in both cases HTLV-1 propagates mainly by mitotic division. The main difference between the models is whether viral expression increases or decreases proviral load.

Recently, a technique has been developed that permits the quantification of lymphocyte turnover in humans in vivo by using nonradioactive isotopes to label dividing cells (16). This technique has been used to measure lymphocyte dynamics in HIV-1 and acute EBV infection (17–19). The aim of this project was to quantify T lymphocyte dynamics in HTLV-1-infected individuals in vivo by using deuterium labeling and then to use these results to address four questions. (i) What is the impact of HTLV-1 infection on lymphocyte dynamics? (ii) How does HTLV-1 persist? (iii) What is the extent of HTLV-1 expression in vivo? (iv) What features of lymphocyte kinetics are associated with HAM/TSP?

Results

Quantification of T Lymphocyte Kinetics in HTLV-1 Infection.

T lymphocyte kinetics in vivo were investigated by using deuterated glucose labeling (16). Subjects were infused with deuterium-labeled glucose for 24 h. Blood samples were taken at days 0, 3, 4, 7, and/or 10, and the level of deuterium incorporated into the DNA of four T lymphocyte subpopulations (CD4⁺CD45RA⁺, CD4⁺CD45RO⁺, CD8⁺CD45RA⁺, and CD8⁺CD45RO⁺) was measured. Deuterium is incorporated into DNA on cell division, so the magnitude of labeling reflects the amount of cell proliferation that occurred during the labeling period. The loss of label over subsequent time points reflects the loss of labeled cells from the blood. This could be due to the death of labeled cells, emigration of labeled cells out of the bloodstream, or a change of cell surface phenotype. In both humans (18) and sheep (20, 21), rapid equilibration (<1 day) of labeled cells between blood and lymphoid tissues has been described. We therefore assumed that by day 3, i.e., 2 days after the end of labeling, equilibrium had been reached between blood and lymph so that there was no subsequent net loss of label due to circulation of labeled cells out of the bloodstream. We further assumed that, because HTLV-1 is a chronic infection, phenotype switching (CD45RA⁺ to CD45RO⁺ and vice versa) was negligible over the 10-day time period of the experiment. That is, we assumed that the loss of label was primarily due to labeled cell death.

By fitting a mathematical model that we had previously developed (22) and applied (17, 23–25), the average rate of cell proliferation and the average rate of labeled cell death were estimated for each of the four T lymphocyte subpopulations. The results showed that the median proliferation rate of CD45RO⁺ cells was of the order of 3% per day (i.e., 3% of the population divides every day); the median proliferation rate of CD45RA⁺ cells was of the order of 0.4% per day. The death rates of recently proliferated CD45RO⁺ and CD45RA⁺ cells were ≈8% per day and ≈4% per day, respectively. Results are shown in Fig. 1, supporting information (SI) Fig. 5, and SI Table 2.

Fig. 1.

Deuterium enrichment in HTLV-1-infected subjects: CD45RO⁺ cells. The labeled fraction, F, and the theoretical fit of the model to the data are shown for each of the 14 subjects. The first eight graphs show labeling in ACs, and the last six graphs show labeling in HAM/TSP patients. Black diamonds, labeled fraction in CD8⁺ T cells; solid line, theoretical fit; gray squares, labeled fraction in CD4⁺ cells; dashed line, theoretical fit.

In HTLV-1-infected subjects, as in uninfected subjects (24, 25) and subjects with acute infectious mononucleosis (17), the rate of proliferation of CD45RA⁺ cells was significantly lower than the rate of proliferation of CD45RO⁺ cells (P = 0.003 for CD4⁺CD45RA⁺ vs. CD4⁺CD45RO⁺; P < 0.001 for CD8⁺CD45RA⁺ vs. CD8⁺CD45RO⁺ paired sign two-tailed test). The deuterium enrichment in the CD4⁺CD45RA⁺ and CD8⁺CD45RA⁺ subpopulations was close to background levels, so it was difficult to analyze these subpopulations confidently. We therefore concentrated on the CD45RO⁺ subpopulations for further analysis.

T Lymphocyte Turnover in HTLV-1-Infected Individuals Compared with Controls.

CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ T lymphocyte kinetics were compared between 14 HTLV-1-infected subjects and 15 controls that had been previously studied by using the same protocol (24, 25). The median proliferation rate of CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ cells in HTLV-1-infected subjects was 2.5% per day and 3.6% per day, respectively; the corresponding rates in controls were 2.0% per day and 1.3% per day, respectively. Both CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ T lymphocyte proliferation rates tended to be higher in infected subjects compared with controls (P = 0.07 and 0.046, respectively, by Mann–Whitney two-tailed test). There was a trend for the death rates of recently proliferated CD4⁺CD45RO⁺ and CD8⁺CD45RO⁺ cells to be higher in infected subjects than in controls, but this was not statistically significant.

Association Between Tax Expression and Cell Division.

CD4⁺CD45RO⁺ T lymphocytes are the main cell type infected by HTLV-1 in vivo (26). If Tax drives infected-cell division and ex vivo Tax expression is correlated with in vivo Tax expression then a positive association between the rate of CD4⁺CD45RO⁺ cell proliferation and Tax expression ex vivo would be expected. The relationship between ex vivo Tax expression and CD4⁺CD45RO⁺ T lymphocyte proliferation in vivo was examined in two ways: between subjects and within subjects.

Between subjects.

The proportion of CD4⁺ lymphocytes expressing Tax was measured after 18 h ex vivo culture following CD8⁺ cell depletion in each of the HTLV-1-infected individuals. There was a significant positive correlation between Tax expression ex vivo and CD4⁺CD45RO⁺ T cell proliferation in vivo (Fig. 2) (P = 0.016 by Spearman rank correlation two-tailed test).

Fig. 2.

The proliferation rate of CD4⁺CD45RO⁺ T cells in vivo was correlated with Tax expression ex vivo. HTLV-1-infected individuals whose CD4⁺ cells express high levels of Tax ex vivo tended to have higher rates CD4⁺CD45RO⁺ T lymphcyte proliferation in vivo (P = 0.016 by Spearman rank correlation two-tailed test).

Within subjects.

To characterize further the relationship between Tax expression and CD4⁺CD45RO⁺ T lymphocyte proliferation, the rate of proliferation in vivo of cells that express Tax ex vivo was estimated. Two subjects whose T lymphocyte kinetics had previously been estimated (L02-HAY and L07-TBI) were infused with deuterated glucose for a second time, in each case >6 months after the first infusion. Blood samples were taken after infusion, and peripheral blood mononuclear cells were cultured for 12 h in the absence of CD8⁺ cells and then sorted into Tax⁺ CD4⁺CD45RO⁺ and Tax⁻ CD4⁺CD45RO⁺ subpopulations. The level of deuterium enrichment in each of these populations was measured, and the parameters of Tax⁺ and Tax⁻ CD4⁺CD45RO⁺ T cell kinetics were estimated (Fig. 3). In each case, the proliferation rate of Tax⁺ CD4⁺CD45RO⁺ cells was greater than the proliferation rate of Tax⁻ CD4⁺CD45RO⁺ cells from the same person (L02 Tax⁺ proliferation rate, p = 7% per day; Tax⁻, p = 4% per day; L07 Tax⁺, p = 13% per day; Tax⁻, p = 3% per day). Thus, we found a striking increase in the proliferation rate of CD4⁺CD45RO⁺ cells that express Tax ex vivo compared with cells from the same subject that did not express Tax, despite the fact that these figures probably underestimate the kinetics of Tax⁺ cells in vivo (see next section).

Fig. 3.

Tax⁺ cells proliferate more rapidly than Tax⁻ cells from the same individual. Deuterium enrichment was measured in Tax⁺ and Tax⁻ CD4⁺CD45RO⁺ cells in two individuals (L02-HAY and L07-TBI). Gray circles, enrichment in Tax⁺ CD4⁺CD45RO⁺ cells; dashed line, theoretical fit; black diamonds, enrichment in Tax⁻ CD4⁺CD45RO⁺ cells; solid line, theoretical fit. (Note the different y axis scale in the two parts of the figure.)

Compatibility with Rapid CTL Lysis of Tax-Expressing Cells.

It has previously been estimated that CTLs kill Tax-expressing cells at a rate of ≈100% per day (and therefore that, at steady state, Tax-expressing cells proliferate at a rate of ≈100% per day) (27). We wished to know whether this rapid rate of turnover was compatible with the measured kinetics of cells expressing Tax ex vivo. The population of cells that express Tax ex vivo will consist of cells that expressed Tax in vivo during the labeling period and survived and of cells that did not express Tax in vivo but subsequently expressed Tax ex vivo. Because the Tax-specific CTL response and Tax-induced toxicity are postulated to rapidly kill Tax-expressing cells in vivo, many cells that expressed Tax during the labeling period are likely to have died before labeling is measured ex vivo. If we postulate that cells that expressed Tax in vivo proliferated and died at a rate of 100% per day (as observed ex vivo) and that cells that did not express Tax in vivo proliferated at the average rate of uninfected CD4⁺CD45RO⁺ cells (2% per day), then we would expect that the rate of proliferation of cells that express Tax ex vivo would be ≈7% per day (see Methods), which is similar to the rate observed in the two subjects studied (7% per day and 13% per day, respectively).

Proportion of Infected Cells that Express Tax.

Interestingly, although there was a trend for subjects with a high proviral load to have a high rate of CD4⁺CD45RO⁺ cell proliferation, this correlation was not significant (P = 0.08 by Spearman rank correlation two-tailed test) and was clearly weaker than the correlation between Tax expression and CD4⁺CD45RO⁺ cell proliferation (P = 0.016 by Spearman rank correlation two-tailed test). This observation is consistent with the hypothesis that the majority of provirus-positive cells do not express viral protein during 24 h in vivo. To estimate the actual proportion of provirus-positive cells that express Tax within 24 h, we used the measured CD4⁺CD45RO⁺ cell proliferation rates (see Methods). It was found that the proportion of provirus-positive cells that expressed Tax in vivo in 1 day was up to ≈3% in HAM/TSP patients and up to ≈0.03% in ACs. Even if the assumptions underlying this calculation were greatly relaxed (see Methods), we found that the fraction was, at most, 36% per day in HAM/TSP patients and 0.4% per day in ACs. This indicates that a minority of infected cells express Tax in 1 day in vivo, especially in ACs.

HAM/TSP and T Cell Proliferation.

The rate of CD4⁺CD45RO⁺ T lymphocyte proliferation was significantly higher in HAM/TSP patients than in ACs (Fig. 4) (P = 0.013 by Mann–Whitney two-tailed test). In the cases where there were HAM/TSP patients and ACs with a similar proviral load (e.g., L05-HT and L09-TAU; L12-HBZ and L06-TAU; L10-HBO and L13-TBX) the HAM/TSP patient always had a higher rate of CD4⁺CD45RO⁺ cell proliferation than the ACs.

Fig. 4.

CD4⁺CD45RO⁺ T lymphocytes proliferate more rapidly in HAM/TSP patients than in ACs. The proliferation rates of CD4⁺CD45RO⁺ T lympocytes in ACs (black diamonds) and HAM/TSP patients (gray squares) are shown. The median proliferation rate (2.2% per day for ACs; 3.5% for HAM/TSPs) is depicted by a dashed line. HAM/TSP patients have significantly higher proliferation rates than ACs (P = 0.01 by Mann–Whitney two-tailed test).

Discussion

We have quantified, in HTLV-1-infected subjects, the average proliferation rate and the labeled cell (i.e., recently proliferated cell) death rate in four lymphocyte subpopulations: CD4⁺CD45RO⁺, CD4⁺CD45RA⁺, CD8⁺CD45RO⁺, and CD8⁺CD45RA⁺ T lymphocytes.

To date, it has been impossible to quantify the level of HTLV-1 expression in vivo. In particular, it has been argued (3) that a serum factor suppresses expression in vivo but that removal of this factor ex vivo allows expression, leading to the conclusion that ex vivo expression is a meaningless artifact. Here, we have provided evidence for continuous low-level HTLV-1 expression in vivo. In particular, the proliferation rate of CD8⁺CD45RO⁺ T lymphocytes in vivo was significantly greater in HTLV-1-infected subjects than in uninfected controls: this is consistent with persistent immune activation in HTLV-1-infected individuals. Unless people with more activated T cells are more likely to acquire HTLV-1 infection, there must be some level of continuous viral expression to increase immune system dynamics. Additionally, we found an association between Tax expression ex vivo and CD4⁺CD45RO⁺ T cell proliferation in vivo, both between subjects and within subjects. Tax, an early viral protein, initiates viral transcription and thus Tax expression is indicative of expression of at least Tax and possibly other viral proteins. The simplest explanation for the observed correlation between Tax expression ex vivo and CD4⁺CD45RO⁺ T cell proliferation is that Tax expression ex vivo correlates with viral expression in vivo (although the two may differ in magnitude) and that viral expression increases CD4⁺CD45RO⁺ cell proliferation. Viral expression could increase CD4⁺CD45RO⁺ T cell proliferation either by the mitogenic effects of Tax and p12^I on virus-expressing CD4⁺CD45RO⁺ cells, by antigenic stimulation of HTLV-1-specific CD4⁺CD45RO⁺ cells, or by bystander activation. The fact that cells that express Tax ex vivo have proliferated more rapidly than cells that do not within the same subject suggests that the first explanation is the most likely. We conclude that our in vivo lymphocyte kinetics data show that HTLV-1 genes are continuously expressed in vivo and that the level of in vivo expression is correlated with Tax expression ex vivo.

The conclusions that HTLV-1 genes are continuously expressed in vivo and that Tax expression ex vivo is correlated with Tax expression in vivo are important for distinguishing between the two models of HTLV-1 persistence. First, the conclusion that Tax expression after an 18-h ex vivo culture is correlated with in vivo Tax expression indicates that Tax expression ex vivo is a physiologically meaningful measure. Three previous findings based on ex vivo Tax expression support the second model: (i) a high probability of Tax expression by an infected cell was correlated with a high proviral load (28); (ii) CTL activity was strongly negatively correlated with proviral load (27); and (iii) infected cells from HAM/TSP patients had high probabilities of Tax expression (28, 29). The kinetics data reported here provide the link between these ex vivo results and the in vivo system and thus support the second model. This is the strongest evidence to date that HTLV-1 expression increases proviral load in humans in vivo. Second, in agreement with previous data (1, 9, 10), the results show that at least some HTLV-1 genes are expressed in vivo. Upon synthesising viral protein, an infected cell becomes exposed to both the HTLV-1-specific CTL response and the toxic effects of Tax, leading to an increased probability of cell death. Even if only a few provirus-positive cells express Tax and subsequently die, this will rapidly deplete the number of provirus-positive cells if the silently infected cells are not replaced. For instance, even if only 1% of provirus-positive cells express Tax per day then, if these cells are not replaced, the proviral load would drop 40-fold in one year (e^−0.01×365 ≈ 1/40). Because proviral load is typically stable within an individual over time (30), the provirus-positive cells must be replaced. Simple, passive replication as hypothesised in the first model would increase all CD4⁺ cells equally and could not selectively increase the proportion of provirus-positive cells. Passive proliferation is also difficult to reconcile with the observation of large infected T cell clones. Infected CD4⁺ cell clones (i.e., cells with the same provirus integration site) are consistently detected in all HTLV-1-infected subjects, and these clones are frequently very large (1/1,500–1/300 peripheral blood mononuclear cells) (31, 32). Even if the provirus was integrated into the DNA of a CD4⁺ cell that subsequently met its cognate antigen and clonally expanded, passive proliferation could not explain such large clone sizes because antigen-specific CD4⁺ cell clones are typically very small (33) and do not reach frequencies of 1 cell in 10,000 even during acute infectious mononucleosis (34).

Therefore, on the basis of these data, we reject the first model and favor the second model. We propose that, upon starting to express viral protein, an infected cell rapidly divides because of the promitotic effect of Tax and possibly p12^I; some of this clone will be killed by CTL or the toxic or proapoptotic effects of Tax, but some cells will survive, Tax will be down-regulated [in part at least by Rex and p30^II (5, 7)], and the surviving cells will contribute to an increase in proviral load. This hypothesis, which is consistent with our observation of a correlation between CD4⁺CD45RO⁺ cell proliferation and Tax expression, also explains the origin of oligoclonal HTLV-1 integration sites (31, 32), the observation that the rate of Tax expression in an infected cell is a significant predictor of proviral load (28), and the observation that HTLV-1-specific CTL are a significant determinant of proviral load (27). In this study, we use the early viral protein Tax as an indicator of viral expression. Other HTLV-1 gene products are likely to contribute to HTLV-1 persistence in vivo, including p12^I (15, 35), p30^II (7), Rex (5), and HBZ (6); indeed in the rabbit model Tax, p12^I and p30^II are essential for efficient viral infectivity (36–39).

Whereas T lymphocyte turnover rates were higher in HTLV-1-infected subjects than in uninfected controls, the effect was small compared with both HIV-1 and acute EBV infection. For instance, median CD8⁺CD45RO⁺ proliferation rates were 3-fold higher in HTLV-1-infected subjects than in uninfected controls but are 9-fold higher in acute EBV infection (17) and 8-fold higher in chronic HIV-1 infection (19) (the latter figure is for the whole CD8⁺ T cell population, not just the CD45RO⁺ subpopulation). This is not unexpected because both HIV-1 and acute EBV are thought to be unusually dynamic infections. The increase in T lymphocyte kinetics in HTLV-1 infection appears to be relatively small because, although most provirus-positive cells are transcriptionally competent (40), only a minority, probably <10%, express Tax in 1 day. This conclusion is also consistent with the finding that proviral load, unlike Tax expression, was not significantly correlated with CD4⁺CD45RO⁺ T cell proliferation.

Although the differences in proliferation rates between HTLV-1-infected subjects and controls were small, this difference has a large cumulative effect over the lifetime of the patient. Assuming that there are 3 × 10¹¹ CD4⁺CD45RO⁺ T lymphocytes and 10¹¹ CD8⁺CD45RO⁺ T lymphocytes in an average adult, we find (see Methods) that an extra 10¹² CD45RO⁺ T lymphocytes are produced in 1 year in an HTLV-1-infected subject compared with an uninfected control. This increase, which would be maintained every year of infection, could have considerable implications for immune senescence, the probability of acquiring transforming mutations, and the maintenance of the T cell repertoire.

We have shown that a minority of infected cells express Tax in 1 day. That is, at any given time, only a small proportion of infected cells are vulnerable to the CTL response. How can this be reconciled with the apparent importance of CTL in determining the proviral load of HTLV-1 (27, 41–43)? We hypothesise that expression of viral proteins is necessary to drive cell division and maintain proviral load in the chronic phase of infection. Any factor that kills virus-expressing cells will therefore block this pathway and is likely to have a large effect on proviral load.

Finally, we demonstrate that HAM/TSP patients have a phenotype of lymphocyte kinetics that is distinct from that of ACs, with a significantly increased proliferation of CD4⁺CD45RO⁺ cells. This is consistent with reports showing that HAM/TSP is associated with an increased rate of ex vivo Tax expression (28, 29).

In summary, we find that CD45RO⁺ T cell proliferation is elevated in HTLV-1 infection, that Tax expression is correlated with CD4⁺CD45RO⁺ T cell proliferation, and that HAM/TSP is associated with abnormally rapid CD4⁺CD45RO⁺ T cell proliferation. We suggest that HTLV-1 persists not because it is latent but because continual, low-level viral protein expression actively increases infected cell division. Continual viral expression and excess CD4⁺CD45RO⁺ cell proliferation may increase the probability of inflammation and cell migration, thus increasing the risk of HAM/TSP and other HTLV-1-associated inflammatory diseases. Increased CD4⁺CD45RO⁺ cell proliferation would also increase the probability of acquiring mutations that could increase the risk of malignant transformation and the development of adult T cell leukaemia. HTLV-1's pathogenic potential thus appears intimately linked to its strategy of persistence and impact on T lymphocyte kinetics.

Methods

A full description of the methods is provided as SI Methods.

Subjects.

All patients attended the HTLV-1 clinic at St. Mary's Hospital, London and gave written informed consent. The study was approved by the Local Research Ethics Committee of St. Mary's Hospital Trust.

Measurement of Cell Turnover.

Cell turnover was calculated from the quantitative incorporation of deuterium into DNA of dividing cells. Briefly, labeling consisted of a 24-h i.v. infusion of [6,6-²H₂]glucose. Blood samples were taken at subsequent timepoints, the cell subpopulations of interest were FACS-sorted, and deuterium enrichment in DNA was quantified (16). The resulting data were analyzed mathematically to quantify the kinetics of the lymphocyte subpopulations (22, 24).

Proviral Load.

Proviral load was measured by real time quantitative PCR (27).

Tax Expression.

A blood sample was taken during the course of the experiment, peripheral blood mononuclear cells were isolated, and CD8⁺ cells were depleted. The CD8⁻ fraction was cultured for 18 h at 37°C, and the cells were washed, fixed, and surface stained for CD4. The cells were washed again, permeabilized, and stained intracellularly for Tax protein (40).

Estimating the Proportion of Cells that Express Tax.

The estimates of CD4⁺CD45RO⁺ cell proliferation rate were used to infer the proportion of provirus-positive cells that express Tax in vivo in 1 day. Initially, we assumed that Tax⁺ and Tax⁻ cells divide at a rate of 1 day⁻¹ and 0.023 day⁻¹, respectively (see SI Methods). Both of these assumptions were later relaxed to ensure that our results were robust.

Predicted Kinetics of Cells that Express Tax ex Vivo.

It has been estimated (27) that cells that express Tax ex vivo are lysed by CTL at a rate of 100% per day. We wished to investigate whether such fast turnover rates were compatible with the observed proliferation rates of cells that express Tax ex vivo. If Tax-expressing cells do die at a rate of 100% per day in vivo, then 5% (e^−1×3 = 0.05) of cells that express Tax during the labeling period will survive to be detected at day 3. Therefore, the majority of cells that express Tax ex vivo will not have expressed Tax during the labeling phase. If we postulate that cells that did not express Tax proliferated at the average rate of uninfected CD4⁺CD45RO⁺ cells (2% per day), we would expect that the rate of proliferation of cells that express Tax ex vivo would be 0.05 × 100% per day + 0.95 × 2% per day = 7% per day.

Statistical Tests.

All statistical tests used for hypothesis testing were nonparametric and appropriate for the sample size analyzed, all P values quoted are two-tailed.

Abbreviations

AC

asymptomatic carrier

CTL

cytotoxic T lymphocyte

HTLV-1

human T-lymphotropic virus type 1

HAM/TSP

HTLV-1-associated myelopathy/tropical spastic paraparesis.