Functional impairment of Tax-specific but not cytomegalovirus-specific CD8+ T lymphocytes in a minor population of asymptomatic human T-cell leukemia virus type 1-carriers

Background Human T-cell leukemia virus type 1 (HTLV-1) causes adult T-cell leukemia (ATL) and HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) in a small percentage of infected individuals. ATL is often associated with general immune suppression and an impaired HTLV-1-specific T-cell response, an important host defense system. We previously found that a small fraction of asymptomatic HTLV-1-carriers (AC) already showed impaired T-cell responses against the major target antigen, Tax. However, it is unclear whether the impaired HTLV-1 Tax-specific T-cell response in these individuals is an HTLV-1-specific phenomenon, or merely reflects general immune suppression. In this study, in order to characterize the impaired HTLV-1-specific T-cell response, we investigated the function of Tax-specific CD8+ T-cells in various clinical status of HTLV-1 infection. Results By using tetramers consisting of HLA-A*0201, -A*2402, or -A*1101, and corresponding Tax epitope peptides, we detected Tax-specific CD8+ T-cells in the peripheral blood from 87.0% of ACs (n = 20/23) and 100% of HAM/TSP patients (n = 18/18) tested. We also detected Tax-specific CD8+ T-cells in 38.1% of chronic type ATL (cATL) patients (n = 8/21), although its frequencies in peripheral blood CD8+ T cells were significantly lower than those of ACs or HAM/TSP patients. Tax-specific CD8+ T-cells detected in HAM/TSP patients proliferated well in culture and produced IFN-γ when stimulated with Tax peptides. However, such functions were severely impaired in the Tax-specific CD8+ T-cells detected in cATL patients. In ACs, the responses of Tax-specific CD8+ T-cells were retained in most cases. However, we found one AC sample whose Tax-specific CD8+ T-cells hardly produced IFN-γ, and failed to proliferate and express activation (CD69) and degranulation (CD107a) markers in response to Tax peptide. Importantly, the same AC sample contained cytomegalovirus (CMV) pp65-specific CD8+ T-cells that possessed functions upon CMV pp65 peptide stimulation. We further examined additional samples of two smoldering type ATL patients and found that they also showed dysfunctions of Tax-specific but not CMV-specific CD8+ T-cells. Conclusions These findings indicated that Tax-specific CD8+ T-cells were scarce and dysfunctional not only in ATL patients but also in a limited AC population, and that the dysfunction was selective for HTLV-1-specifc CD8+ T-cells in early stages.

HTLV-1-specific cytotoxic T-lymphocytes (CTLs) are thought to play a pivotal role in containing the proliferation of HTLV-1-infected T-cells [9,10]. Tax is known to be the dominant target antigen for HTLV-1-specific CTLs [10][11][12][13], and a high frequency of Tax-specific CTLs can be detected in HAM/TSP patients and some asymptomatic HTLV-1 carriers (ACs) [10][11][12][13][14]. However, ATL patients show general immune suppression [15], reduced frequency and dysfunction of Tax-specific CTLs [16,17]. Regulatory T cell (Treg)-like function of FoxP3 + ATL cells and diminished function of dendritic cells may be involved in the immune suppression in ATL patients [18,19], but the precise mechanism is not yet clarified. We previously demonstrated that a fraction of ACs also exhibit reduced T-cell responses against Tax protein [20]. These observations suggest that the reduced HTLV-1-specific T-cell response might be an underlying risk of ATL development, but not the result of ATL. However, it is unknown how the function of HTLV-1specific CD8 + T-cells becomes impaired in a small percentage of ACs and whether its dysfunction is specific for HTLV-1 antigen or due to general immune suppression.
During chronic stage of infection with several viruses, such as HIV and hepatitis C virus (HCV), virus-specific CTLs gradually lose their cytotoxic activity, the ability to proliferate and secrete a diverse profile of cytokines, ultimately leading to exhaustion, anergy or even deletion of these cells [21][22][23][24][25][26]. Programmed death-1 (PD-1), a negative regulator in the CD28 superfamily, has recently been shown to be highly expressed on virus-specific Tcells during many chronic viral infections [27][28][29]. It has also been reported that the interaction of PD-1 with PD-ligand 1 (PD-L1) negatively regulates cytokine production and proliferation of T-cells [30,31]. A previous report indicates that PD-1 is up-regulated on the dominant Tax-specific CTLs in ATL patients and ACs and that immune regulation through the PD-1/PD-L1 pathway may be involved in the dysfunction of HTLV-1-specific CTLs in ATL patients [32].
Studies on memory T-cell differentiation have shown that phenotype, function, and homeostasis of memory T-cells vary for different persistent virus infections [33]. Central memory T-cells (T CM ; CD45RA -CCR7 + ) are elicited by non-persisting virus that provide transient antigen stimulation, such as in Influenza virus infection. In contrast, effector memory T-cells (T EM ; CD45RA -CCR7 -) predominate when relatively high levels of antigen persist, such as in HIV infection. Terminally differentiated memory (T Diff ; CD45RA + CCR7 -) can be seen when antigen persists at a low level, such as in cytomegalovirus (CMV) infection. In HTLV-1 infection, it has been reported that dominant Tax-specific CTLs in HAM/TSP patients consist of T EM and T Diff compartments [34].
We previously identified some major epitopes recognized by HTLV-1-specific CTLs in infected individuals carrying HLA-A2, -A11, or -A24 [12,35,36]. These allowed us to monitor HTLV-1-specific CTLs and analyze their functions ex vivo, by using antigen/HLA tetrameric complexes. In this study, we demonstrate that IFN-γ production and proliferative capacity of tetramerbinding Tax-specific CD8 + T-cells were severely impaired not only in ATL patients but also in a minor population of asymptomatic HTLV-1 carriers (ACs). Importantly, the T-cell dysfunction at the asymptomatic stage was selective for HTLV-1 but not for CMV antigen. In addition, severely impaired HTLV-1-specific but not CMV-specific CD8 + T-cells responses were also observed in patients diagnosed as smoldering ATL, the clinical condition of which is close to that of AC. The dysfunction of HTLV-1-specific CD8 + T-cells in an early clinical stage implies HTLV-1-specific immune suppressive mechanism might be an underlying risk for ATL.
Impaired cell proliferation and IFN-g production of Taxspecific CD8 + T-cells in cATL but not HAM/TSP patients We next examined IFN-γ production and cell proliferation of Tax-specific CD8 + T-cells in HAM/TSP and cATL patients (Figure 2A). Intracellular IFN-γ staining showed that Tax-specific CD8 + T-cells in all HAM/TSP patients tested produced IFN-γ when stimulated with Tax peptide (Figure 2A). Tax-specific CD8 + T-cells in those HAM/TSP patients proliferated regardless of stimulation with Tax peptide ( Figure 2B). In contrast to HAM/TSP patients, IFN-γ production from Tax-specific CD8 + T-cells in a cATL patient was hardly detectable even when stimulated with Tax peptide (4.8%, Figure  2A). In the same donor, Tax 1 No. of samples with detectable Tax-specific CD8 + T-cells/total no. of samples carrying each HLA allele. When the frequency of tetramer + cells was more than 0.04% of CD8 + T-cells, the sample was regarded as detectable. 2 In case Tax-specific CD8 + T-cells was detectable by either tetramer in a sample carrying two of three HLA-A alleles above, the sample was regarded as positive.
could be detected in fresh blood (1.32%) and after 6 hrs incubation as shown in Figure 2A, but not after 6 dayculture, suggesting that Tax-specific CD8 + T-cells in this cATL patient had no proliferative capacity ( Figure  2B). We tested PBMC from four other cATL patients who had detectable Tax-specific CD8 + T-cells, but none of them showed proliferation of Tax-specific CD8 + Tcells by either the CFSE-based proliferation assay or 13day culture (Additional file 1). Collectively, these results indicate that Tax-specific CD8 + T-cells from most cATL patients are impaired in their capacities to proliferate and produce IFN-γ.

Diversity in the IFN-g production and cell proliferation of Tax-specific CD8 + T-cells in ACs
Our recent studies using the GST-Tax protein-based assay demonstrated that the extent of Tax-specific Tcell responses varied widely in ACs [20]. We then evaluated proliferation and/or IFN-γ production of tetramerbinding Tax-specific CD8 + T-cells in 14 ACs (   specific CD8 + T-cells in these ACs spontaneously proliferated without stimulation with Tax peptide, probably due to viral reactivation in HTLV-1-infected cells in vitro [38,39]. IFN-γ production was specifically detected for peptide stimulation, and 35.8-55.7% of Tax-specific CD8 + T-cells produced a good amount of IFN-γ (mean fluorescence intensity, MFI: 63.7-195.3) upon stimulation in the samples of #251, #313, and #360. In contrast, Tax-specific CD8 + T-cells in one AC (#287) did not proliferate in response to Tax peptide and showed a very weak IFN-γ response with low amounts of IFN-γ (MFI: 37.5) in a low percentage (11.1%) of Tax-specific CD8 + T-cells ( Figures 3A and 3B). In other ACs (#243 and #279), low frequency of IFN-γ + Tax-specific CD8 + T-cells was observed, but the levels of IFN-γ production (MFI: #243; 58.8, #279; 77.6) and the proliferative responses were comparable to other ACs (Table 2). Tax-specific CD8 + T-cells in #236 failed to proliferate but showed favorable IFN-γ production (MFI: 80.1) in 31.1% of the cells. Among AC samples tested, AC#287 carried higher proviral load (81 copies in 1000 PBMCs) than any other ACs (Table 2). Since Tax-specific CD8 + T-cells in #287 had severely impaired IFN-γ production and proliferative potential, we examined the relationship of the function of these T-cells with proviral loads. Both percentages of IFN-γ + and dividing Tax-specific CD8 + T-cells among CD8 + T-cells were likely to be inversely correlated with proviral loads although they were not statistically significant ( Figure 3C and 3D). Because of the limited availability of the samples, we focused mainly on two ACs (#287 and #313) in the studies hereafter.

Dysfunction of Tax-specific CD8 + T-cells and inefficient CD8 + cell-mediated HTLV-1 control in AC #287
To examine whether Tax-specific CD8 + T-cell responses were influenced by activation of antigen-presenting cells (APCs), PBMC from #313 (responder) and #287 (low responder) were stimulated with Tax peptide in the presence or absence of LPS, a potent activator of APCs such as dendritic cells (DCs) and monocytes/macrophages. In #313, the frequency of Tax-specific CD8 + Tcells increased from 1.11% to 6.47% or 4.07% at day 13, after stimulation with or without Tax peptide, respectively. The frequency of Tax-specific CD8 + T-cells in #313 further increased in the presence of Tax peptide and LPS (15.81%). In contrast to #313, the frequency of Tax-specific CD8 + T-cells in #287 decreased from 1.17% The number indicates percentage of CD4 + or CD8 + T cells in lymphocytes. 2 The number indicates percentages of tetramer + cells in CD8 + T-cells. Two numbers divided by a slash represent those detected by two different tetramers corresponding to two HLA alleles shown in the HLA column. 3 In case of a sample carrying two of three HLA-A alleles (A2, A11, or A24), Tax-specific CTLs predominantly detected by a tetramer were used. The number represents percentage of indicated cells in the tetramer-binding CD8 + T cells. 4 Evaluated by intracellular IFN-γ staining following 6 hours stimulation with corresponding Tax peptide. 5 Evaluated by CFSE intensities in labeled PBMC after 6 days incubation with corresponding Tax peptide stimulation. 6 The number represents percentage of indicated PD-1 + Tax-specific CD8 + T cells without culture. to 0.2% after stimulation with Tax peptide, and was not recovered by LPS stimulation ( Figure 4A). In addition, HTLV-1-infected cells have been reported to express C-C chemokine receptor type 4 (CCR4) and have FoxP3 + Treg-like function [18,40]. However, the proliferative ability of Tax-specific CD8 + T-cells in #287 was not restored even in the absence of CCR4 + infected cells (data not shown).
To further examine the function of Tax-specific CD8 + T-cells in #313 and #287, we observed the expression of CD69, an early activation marker transiently expressed on T lymphocytes that precedes cytokine secretion after antigenic stimulation, and CD107a, a marker of degranulation associated with cytotoxic activity in an antigenspecific manner [41]. CD69 was up-regulated on Taxspecific CD8 + T-cells in #313 when stimulated with Tax peptide, but not in #287, which was in agreement with their abilities to produce IFN-γ ( Figure 4B). In #313, 22.4% of Tax-specific CD8 + T-cells mobilized CD107a to the surface during a 6-hr culture with Tax peptide stimulation, while CD107a surface expression was detected on 4% of Tax-specific CD8 + T-cells in the culture without stimulation ( Figure 4C). However, no CD107a mobilization was detected on the surface of Tax-specific CD8 + T-cells in #287 with or without Tax peptide stimulation ( Figure 4C). These results indicate that HTLV-1-specific CD8 + T-cells in AC #287 did not properly activate upon antigen stimulation, and therefore failed to control HTLV-1-infected cells.
The Tax/HLA tetramers used in this study allow us to evaluate the functions of CD8 + T-cells only against an immunodominant epitope, Tax. We therefore compared HTLV-1 Gag p19 in the culture between whole and CD8 + cell-depleted PBMCs to examine the role of total HTLV-1-specific CD8 + T-cells including the dominant Tax-specific CD8 + T-cells, in suppression of HTLV-1 production from infected cells ( Figure 4D). As expected, depletion of CD8 + cells from PBMCs in #313 led to significantly higher HTLV-1 production compared to whole PBMCs (P = 0.0115). In contrast, HTLV-1 p19 production increased only a little in the culture of CD8 + cell-depleted PBMCs in #287 (P = 0.1563), indicating that HTLV-1-specific CD8 + T-cells other than the dominant Tax-specific CD8 + T-cells might have a reduced ability to control the infected cells in this donor. It is of note that HTLV-1-infected cells from both two donors carried intact HTLV-1 proviral genomic DNA because HTLV-1 p19 could be detected after 7 day-culture.
A previous report has shown that PD-1 was highly upregulated on Tax-specific CD8 + T-cells in ATL patients and ACs [32]. We therefore examined PD-1 expression on Tax-specific CD8 + T-cells in several AC samples, including #287. The frequency of PD-1 + Tax-specific CD8 + T-cells was very high in #309 (85.3%) and #313 (96%) ( Figure 5B and Table 2) while those Tax-specific CD8 + T-cells retained the proliferative and the cytokineproducing abilities ( Figure 3A and Table 2). In #287, the frequency of PD-1-expressing Tax-specific CD8 + T-cells (55.6%) was lower than #309 and #313, but higher than that of PD-1 + CMVpp65-specific CD8 + T-cells in the same donor ( Figure 5B). The levels of PD-1 expression showed a similar tendency to the frequency of PD-1 + Tcells. In addition, the blockade of PD-1/PD-ligand 1 (PD-L1) pathway did not restore the proliferative capacity of Tax-specific CD8 + T-cells in #287 (data not shown).

Conserved functions of CMV-specific CD8 + T-cells in #287
We next examined whether the impairment of proliferative capacity and effector functions observed in #287 CD8 + T-cells were specific for HTLV-1 antigens or the result of general immune suppression. PBMC from #287 contained CMVpp65-specific CD8 + T-cells (2.3% of CD8 + T-cells), as detected by tetramer staining. The frequency of CMVpp65-specific CD8 + T-cells increased from 2.3% to 66.0% following in vitro CMVpp65 peptide stimulation, but not without the peptide stimulation ( Figure 6A). Antigen-specific IFN-γ and CD69 expression were clearly detected in CMVpp65-specific CD8 + T-cells in #287 (Figures 6B and 6C). Furthermore, CMVpp65-specific CD8 + T-cells mobilized CD107a to the surface in response to CMVpp65 peptide ( Figure  6D). These results demonstrate that in #287, CMVpp65specific CD8 + T-cells, but not Tax-specific CD8 + Tcells, have proliferative potential and effector functions, such as cytotoxic activity and IFN-γ release, suggesting that the impaired CD8 + T-cell function in #287 was specific for HTLV-1.

Dysfunction of Tax-specific but not CMVpp65-specific CD8 + T-cells also in sATL patients
Finally, we extended the study to see whether patients with early stage ATL might exhibit similar dysfunction selective for HTLV-1-specific CD8 + T-cells. We found two smoldering ATL (sATL) patients (#110 and #353) (D) Whole PBMCs and CD8-depleted fractions in ACs (#287 and #313) were cultured for 7 days and HTLV-1 p19 in the supernatants were measured by HTLV-1 p19 ELISA. P value was determined by the unpaired t test. possessing 6.89% and 3.15% of tetramer-binding Taxspecific CD8 + T-cells, respectively. The sATL patient #353 carried 5% of abnormal lymphocytes (ably) with a normal range of lymphocyte number, whose status is very close to the borderline with ACs. Patient #110 carried 4% of abnormal lymphocytes with mild lymphocytosis. Tax-specific CD8 + T-cells of two sATL patients (#110 and #353) did not proliferate in response to Tax peptides as similarly observed in a cATL patient (#224) ( Figure 7A) and most other cATL patients (Figure 2A and Additional file 1). In contrast, CMVpp65-specific CD8 + T-cells in both sATL patients vigorously proliferated when stimulated with CMVpp65 peptides. CMVpp65-specific CD8 + T-cells in a cATL (#224) also proliferated, but to a lesser degree, which might reflect general immune suppression in this patient (Figure 7).

Discussion
In this study, we detected Tax-specific CD8 + T-cells in 87%, but not the rest of ACs tested, by using tetramers containing Tax major epitope-peptides presented by HLA-A*0201, A*1101, and A*2402. Tax-specific CD8 + T-cells were also detected in 38% of cATL patients, but at reduced frequencies and with severely impaired functions. Further analysis of Tax-specific CD8 + T-cells in 14 ACs indicated that they were functional in most of ACs tested except one (#287), whose Tax-specific CD8 + T-cells poorly responded to specific peptides. However, CMVpp65-specific CD8 + T-cells of this individual were fully functional. Similar T-cell dysfunction selective for HTLV-1, but not CMV, was also observed in sATL patients, one of which (#353) had no clinical symptoms but 5% abnormal lymphocytes. General immune suppression might partly account for the scarcity and/or the dysfunction of Tax-specific CD8 + T-cells in ATL patients, but not those in the AC or the sATL patients as they were selective for HTLV-1. These findings suggest that HTLV-1-specific immune suppression is undergoing in a minor group of ACs and an early stage of ATL.  The presence of tetramer-binding Tax-specific CD8 + T-cells in cATL patients, although at low frequencies, implies that they have encountered antigen during the chronic phase of ATL disease, suggesting that Tax may be expressed in vivo. This may be supported by a previous report showing that virus-specific CD8 + T-cells fails to acquire memory T-cell property of long-term antigen-independent persistence during chronic lymphocytic choriomeningitis virus (LCMV) infection [45]. However, there is no direct evidence that infected cells produce Tax in infected individuals. HTLV-1-specific Tcell responses in cATL patients are largely different from HAM/TSP patients. In HAM/TSP patients, Taxspecific CD8 + T-cells proliferated vigorously and a large population of them produced IFN-γ. In contrast, the function of Tax-specific CD8 + T-cells in cATL patients was profoundly suppressed, similarly to tumor infiltrating lymphocytes (TIL) [46]. In cATL patients, Tax-specific CD8 + T-cells that were detected before culture decreased in number to undetectable or very low levels after 6 days, regardless of peptide stimulation (data not shown). This is not likely to be due to TCR down-regulation, because TCRs on Tax and reappeared by day 6 in vitro [34]. Moreover, we could not observe any tetramer + CD8 + T-cells even in the 13-day culture (data not shown), suggesting these cells might have died during the culture. Severe dysfunction of Tax-specific CD8 + T-cells was observed not only in cATL patients, but also in an AC #287. Fresh PBMCs of #287 contained 1.17% tetramer + cells in the CD8 + T-cell fraction. However, none of these tetramer-positive T-cells proliferated in culture, with or without Tax peptide stimulation ( Figure 3B). Although a few populations of them (11.1%) produced a small amount of IFN-γ, they lacked degranulation activity for cytotoxicity or expression of CD69, an early activation marker, upon specific stimulation (Figures 3 and  4). Importantly, CMVpp65-specific CD8 + T-cells in the same donor were clearly activated, and exhibited these characteristics upon stimulation with pp65 peptides (Figure 6). These observations indicated that the impaired Tax-specific CD8 + T-cells function in #287 was not attributable to general immune suppression, but to an HTLV-1-specific phenomenon. In addition, CD8depletion study indicated that not only the dominant Tax-specific CD8 + T-cell function but also other HTLV-1-specific CD8 + T cell responses might be reduced in #287 ( Figure 4D). Since CMV-specific CD8 + T-cells responded well to the specific peptides, antigen-presenting cells in culture were not likely to be responsible for the selective suppression of Tax-specific CD8 + T-cells.

#313
In addition, it has been shown that HTLV-1-infected cells generally express CCR4 and have Treg-like function [18,40]. However, depletion of CCR4 + cells did not restore the proliferative ability of Tax-specific CD8 + Tcells (data not shown), indicating that suppression of the infected cells were not likely to be the major reason for the impaired Tax-specific CD8 + T-cell function in our culture system. These observations suggest that in #287, Tax-specific CD8 + T-cells themselves might lose their functions.
Many chronic viral infections affect the phenotype, function, and maintenance of memory T-cells [24,42,47,48]. T EM cells predominate in infections in which relatively high levels of antigen persist and continuous antigen stimulation are required for maintenance of T EM cells. As described in HAM/TSP patients [34], Tax-specific CD8 + T-cells in both ACs (#287 and #313) were primarily enriched in T EM memory pool in spite of the functionality of Tax-specific CD8 + T-cells ( Figure 5A), which may support continuous or periodical expression of viral antigen in vivo during an asymptomatic stage.
PD-1 is known to play a major role in regulating Tcell exhaustion during chronic infection. In this study, we could not obtain any data supporting the involvement of PD-1 in the dysfunction of Tax-specific CD8 + T-cells. However, we observed that Tax-specific CD8 + T-cells in some ACs showed IFN-γ production, but not  proliferative capacity (Table 2). This partially lacked function of Tax-specific CD8 + T-cells is similar to the features of T-cell exhaustion. Whether Tax-specific CD8 + T-cells are exhausted in HTLV-1 infection, and whether other molecules associated with T-cell exhaustion are involved in the impairment of Tax-specific CD8 + T-cell responses are necessary to be clarified because some inhibitory molecules such as T-cell immunoglobulin and mucin domain-containing protein-3 (TIM-3), lymphocyte activated gene-3 (LAG-3), and transcription factors including BLIMP-1 are also found to be associated with T-cell exhaustion [49]. The incidence of Tax-specific CD8 + T-cell detection was high (87.0%) in ACs. Given the fact that the incidence of Tax-specific CD8 + T-cells in HAM/TSP patients was 100%, a small fraction of ACs lacking detectable tetramer-binding cells might lack Tax-specific T-cell responses. Our previous study investigating GST-Tax protein-based T-cell responses supports this notion [20]. In the present study, even in ACs possessing Taxspecific CD8 + T-cells, at least one individual exhibited T-cell dysfunction selectively for HTLV-1. The incidence of tetramer-positive cells was reduced in ATL patients (38.1%), and the function of these cells was impaired in all the ATL patients even with detectable tetramer-binding Tax-specific CD8 + T-cells. Our findings suggest that HTLV-1-specific T-cell responses are selectively impaired in a small percentage of HTLV-1infected individuals in the asymptomatic stages, and the proportion of individuals with such characteristics increase as the stages proceed towards ATL. Strategies to reactivate HTLV-1-specific T-cells at early stages might contribute to a reduction in the immunological risk of ATL.

Conclusions
Tax-specific CD8 + T-cells were scarce and dysfunctional in a limited AC population and ATL patients, and the dysfunction of CD8 + T-cells was selective for HTLV-1 in early stages. These results implied the presence of some HTLV-1-specific T-cell suppressive mechanisms even in asymptomatic stages, which are not a result of general immune suppression in ATL but could be underlying conditions toward disease progression.

Samples
Blood samples from 64 HTLV-1-seropositive individuals were used in this study: 23 asymptomatic carriers (ACs), 18 HAM/TSP patients, 2 smoldering type ATL (sATL) patients, and 21 chronic type ATL (cATL) patients. All blood samples were obtained following written informed consent, and this study was reviewed and approved by the Institutional Review Board of the Tokyo Medical and Dental University.

Depletion of CD8 + cells and Detection of HTLV-1 p19
CD8 + cells were depleted from PBMCs by negative selection using 10-fold numbers of Dynabeads M-450 CD8 (Invitrogen, Carlsbad, CA), according to the manufacturer's instructions. The PBMCs were adjust to 1 × 10 6 cells/ml before depletion, and the resulting CD8 + celldepleted fractions were resuspended in medium with the same initial volume, irrespective of the remaining cell number. PBMCs (1 × 10 6 cells/ml) and CD8 + celldepleted PBMCs were cultured for 7 days. HTLV-1 p19 in the supernatants of those PBMCs were measured by HTLV p19 antigen ELISA (RETRO tek, Buffalo, NY).

Statistics
The Mann-Whitney U-test, the unpaired t test, and the Spearman rank correlation test were performed for statistical significance by using the Graphpad Prism software (Graphpad Software). In all cases, two-tailed P values less than 0.05 were considered significant.

Additional material
Additional file 1: Tax-specific CD8 + T-cells in cATL patients could not proliferate against Tax-peptide stimulation. (A) CFSE-labeled PBMCs were cultured with or without 100 nM Tax-peptide for 6 days. The number indicates the percentage of tetramer + cells in CD8 + T cells (Day 0) or the percentage of dividing (CFSE low ) cells in Tax-specific CD8 + T-cells (Day 6). In a cATL sample #54, CFSE-labeled PBMCs were cultured in the presence of mouse IgG for other experiment. (B) PBMCs (#224) and CCR4-depleted PBMCs (#280) were cultured for 13 days in the presence of 100 nM Tax-peptide. The number indicates the percentage of tetramer + cells in CD8 + T-cells.