Human T-cell leukemia virus type 1 (HTLV-1) is the causative retrovirus of adult T-cell leukemia (ATL) and HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP). HTLV-1-specific T-cell responses elicit antitumor and antiviral effects in experimental models, and are considered to be one of the most important determinants of the disease manifestation, since they are activated in HAM/TSP but not in ATL patients. The combination of low T-cell responses and elevated HTLV-1 proviral loads are features of ATL, and are also observed in a subpopulation of HTLV-1 carriers at the asymptomatic stage, suggesting that these features may be underlying risk factors. These risks may potentially be reduced by vaccination to activate HTLV-1-specific T-cell responses. HAM/TSP and ATL patients also differ in their levels of HTLV-1 mRNA expression, which are generally low in vivo but slightly higher in HAM/TSP patients. Our recent study indicated that viral expression in HTLV-1-infected T-cells is suppressed by stromal cells in culture through type-I IFNs. The suppression was reversible after isolation from the stromal cells, mimicking a long-standing puzzling phenomenon in HTLV-1 infection where the viral expression is very low in vivo and rapidly induced in vitro. Collectively, HTLV-1 is controlled by both acquired and innate immunity in vivo: HTLV-1-specific T-cells survey infected cells, and IFNs suppress viral expression. Both effects would contribute to a reduction in viral pathogenesis, although they may potentially influence or conflict with one another. The presence of double control systems for HTLV-1 infection provides a new concept for understanding the pathogenesis of HTLV-1-mediated malignant and inflammatory diseases. (Cancer Sci 2011; 102: 670–676)
It has been three decades since the discovery of human T-cell leukemia virus type 1 (HTLV-1) as the causative retrovirus of adult T-cell leukemia (ATL).(1,2) ATL develops during middle age or later mainly in a small portion of vertically HTLV-1-infected populations.(3,4) HTLV-1 also causes HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP) in another small population of infected individuals.(5,6) Some other inflammatory diseases such as uveitis and arthritis are also associated with HTLV-1 infection.(7,8) New therapeutic approaches such as hematopoietic stem cell transplantation (HSCT),(9,10) an antibody therapy targeting CCR4,(11) and antiviral therapy with interferon-alpha and zidovudin(12) partly improved the prognosis of ATL. However, ATL still shows high mortality, and HAM/TSP remains to be an intractable disease.
Enormous amounts of research findings have been accumulated regarding the virus-mediated pathogenesis. HTLV-1 Tax, a virus-encoded regulatory gene product, mediates cell activation, proliferation and resistance to apoptosis by transactivation through NF-κB, cAMP response element binding protein (CREB) and serum response factor (SRF), and by inactivation of tumor suppressors,(13–15) which would be involved in leukemogenesis and inflammation in HTLV-1 infection. Another minus-strand HTLV-1-encoded gene product, HTLV-1 basic leucine zipper factor (HBZ), is continuously expressed in infected cells in vivo regardless of the disease and may also be involved in the growth ability of infected cells.(16)
However, many unsolved questions still remain regarding the pathogenesis of HTLV-1 infection, for example, how the same virus causes totally different diseases such as ATL and HAM/TSP, why only small portions of HTLV-1-infected populations develop diseases, and why it takes more than 40 years to develop ATL. The answers to these questions would provide hints for predicting disease risks as well as aiding the development of prophylactic and therapeutic strategies.
HTLV-1-specific T-cell responses that contribute to antiviral and antitumor surveillance could be one of the most important determinants of the diseases. In fact, HTLV-1-specific T-cells are activated in HAM/TSP but not in ATL.(17–19) Oral HTLV-1 infection induces T-cell tolerance to HTLV-1 and increased proviral loads,(20,21) consistent with the epidemiological finding that vertical HTLV-1 infection is one of the risk factors for ATL.(3) Therefore, the individual status of HTLV-1-specific T-cell responses is expected to be an indicator of risk for ATL.(22) Although the pathological significance of HTLV-1-specific T-cells in HAM/TSP remains controversial,(23,24) advantages for HLA-A02-positive individuals in protection against HAM/TSP have been reported, and interpreted through the association of this HLA with strong CTL responses to a major epitope of HTLV-1 Tax.(25)
Elevation of proviral loads is also a risk factor for ATL. Given the fact that HTLV-1-specific CTLs have antiviral effects, these CTLs are likely to be one of the determinants of proviral loads.(26) However, proviral loads are also increased in HAM/TSP patients, and the correlations between proviral loads and HTLV-1-specific T-cell responses vary among studies,(27,28) suggesting the presence of additional factors for determining individual proviral loads.
Another curious finding in HTLV-1 infection is the scarcity of viral antigen expression in the peripheral blood, although the viral mRNA is barely expressed.(29) The transcription of HTLV-1 is mainly regulated by CRE-like repeats in the HTLV-1 LTR.(30) Involvement of inducible cAMP early repressor (ICER) and transducers of regulated CREB 2 (TORC2) in the inhibition of HTLV-1 transactivation has been suggested.(31,32) However, the mechanism involved in suppressing viral expression only in vivo has remained obscure. It is a paradox that HTLV-1 Tax contributes to the pathogenesis while Tax protein is undetectable in vivo. Expression of HBZ in the absence of Tax may partly explain the growth advantage of infected cells,(33) but not all of HTLV-1-mediated leukemogenesis. In addition, it does not make sense that Tax-specific T-cell responses are maintained if Tax is not expressed in vivo. The paradox will remain until the state of viral expression and the mechanisms for suppressing HTLV-1 expression in vivo are clarified.
We recently found that innate immune responses, especially type-I interferons (IFNs), suppress HTLV-1 expression.(34) This integrates the issue of viral expression and the host defense system against HTLV-1, which includes innate immunity as well as acquired immunity. The presence of double control systems explains some of the paradox in persistent HTLV-1 infection, and adds new aspects to the pathogenesis of HTLV-1-mediated diseases.
Control of HTLV-1 by HTLV-1-specific T-cell responses
Antitumor surveillance by HTLV-1-specific T-cells.
CD8+ HTLV-1-specific CTL responses are found in many HAM/TSP patients and asymptomatic carriers (AC), but rarely in ATL patients.(17–19,35,36) These CTLs kill HTLV-1-infected cells in vitro, and mainly recognize HTLV-1 Tax.(18,37) The HTLV-1 envelope is also a popular target, especially for CD4+ CTLs.(38) Other viral antigens, including polymerase,(39) ROF (p12) and TOF (p30/p13),(40) and HBZ,(41) have also been shown to be targets of CTLs. Elimination of CD8+ cells among PBMCs from HAM/TSP patients induces HTLV-1 expression during subsequent cell culture,(42) clearly indicating that CD8+ HTLV-1-specific CTLs contribute to the control of HTLV-1-infected cells.
A series of animal model experiments indicated that HTLV-1-specific T-cell responses limit the expansion of HTLV-1-infected cells in vivo. Oral HTLV-1 infection induced insufficiency of HTLV-1-specific T-cell responses in rats, and the HTLV-1 proviral loads were inversely correlated with HTLV-1-specific T-cell responses.(21) Re-immunization of these rats with mitomycin C-treated HTLV-1-infected cells restored HTLV-1-specific T-cell responses and reduced the proviral loads(43) (Fig. 1). In another rat model of HTLV-1-induced tumors, the otherwise fatal HTLV-1-infected lymphomas in T-cell-deficient rats were eradicated by transfer of T-cells from syngeneic rats that had been vaccinated with a Tax-encoding DNA or peptides corresponding to a major epitope for Tax-specific CTLs.(44,45)
Recent clinical reports have indicated that HTLV-1-carrying recipients after liver transplantation developed ATL under the administration of immunosuppressants.(46,47) In contrast, Tax-specific CTL responses were strongly activated in some ATL patients who obtained complete remission after HSCT, but were not observed in the same patients before transplantation.(48) These findings suggest that HTLV-1-specific T-cells, including Tax-specific CTLs, play important roles in antitumor surveillance against HTLV-1 leukemogenesis.
Insufficient HTLV-1-specific T-cell responses as a potential risk for ATL.
Most HTLV-1-infected individuals are asymptomatic, and only about 5% develop ATL and <1% develop HAM/TSP.(3,49) The epidemiological risk factors for ATL include vertical transmission and increases in the number of abnormal lymphocytes or HTLV-1 proviral loads.(3,50,51) HTLV-1 proviral loads are also elevated in HAM/TSP patients.(52)
Immunological studies have suggested that insufficiency in host T-cell responses against HTLV-1 might be another risk factor for ATL.(22) A small-scale survey measuring Tax protein-specific IFN-γ production revealed a wide variety in the strengths of HTLV-1-specific T-cell responses among HTLV-1 carriers.(53) The combinations of HTLV-1-specific T-cell responses and proviral loads categorize HTLV-1 carriers into the following four groups: (i) low proviral loads with HTLV-1-specific T-cell responses; (ii) elevated proviral loads with HTLV-1-specific T-cell responses; (iii) low proviral loads with low T-cell responses; and (iv) elevated proviral loads with low T-cell responses (Fig. 2).
Regarding these groups, ATL patients exhibit elevated proviral loads with low T-cell responses, while many, but not all, HAM/TSP patients show elevated proviral loads with high HTLV-1-specific T-cell responses. ACs are found in all four categories. It is noteworthy that small subgroups of ACs and smoldering ATL patients share a common feature with ATL patients. This indicates that the insufficiency of HTLV-1-specific T-cell responses is not merely the result of malignancy but is an underlying problem before the stage without apparent lymphoproliferation. Further follow-up studies are required to clarify whether the extent of the combination of elevated proviral loads with low T-cell responses could be a diagnostic indicator for risk of ATL.
Dissociation between proviral loads and T-cell responses.
Although HTLV-1-specific T-cells have the potential to control infected cells, there are no clear correlations between proviral loads and HTLV-1-specific T-cell responses among HTLV-1-infected individuals. This is not surprising because both the proviral loads and T-cell responses are high in HAM/TSP patients. The proviral loads may be negatively correlated with T-cell responses only within an individual but not among individuals. Several other reports have indicated various findings concerning this issue. For example, a study measuring IFN-γ-producing CD8+ HTLV-1-specific CTLs indicated a positive correlation with proviral loads in HAM/TSP patients but not in ACs,(28) while a study evaluating CD8+ CTL function by ex vivo clearance of infected cells showed negative correlations with low proviral loads within an AC or a HAM/TSP group,(42) and another study indicated an association of higher frequency of tetramer-binding Tax-specific CTLs with low proviral loads in ACs.(27) Such inconsistent results suggest the presence of certain other determinants of proviral loads in addition to HTLV-1-specific CTLs.
The HTLV-1 proviral loads reflect the number of infected cells in the peripheral blood. Expansion of HTLV-1-infected cells in vivo occurs through both de novo infection and proliferation of infected cells.(54) The number of CD4+ FoxP3+ cells,(55) the frequency of iNKT cells,(56) or MHC-I favorable for HBZ-specific T-cell responses(41) have been suggested to influence HTLV-1 proviral loads.
In HTLV-1-infected rats, however, the proviral loads are inversely correlated with HTLV-1-specific T-cell responses.(21) One reason for the discrepancy between humans and rats may be the genetic heterogeneity in humans. It appears that, under the homogeneous genetic background in the experimental rat system, the influence of insufficient HTLV-1-specific T-cell responses may appear more clearly than in humans, allowing de novo infection and proliferation of HTLV-1-infected cells in vivo. The dissociation of proviral loads and HTLV-1-specific T-cell responses in humans suggests that additional determinants of proviral loads may vary genetically among individuals. As described in the next section, we suppose that innate immunity could be a candidate for this effect.
Control of HTLV-1 by innate immunity
Status of HTLV-1 expression in vivo.
Since HTLV-1-specific antibodies and T-cells are maintained in HTLV-1-infected individuals, viral expression must occur somewhere in vivo. This notion is further supported by the emergence of Tax-specific CTL responses in HTLV-1-uninfected donor-derived hematopoietic systems reconstituted in recipient ATL patients after HSCT.(48,57) However, HTLV-1 mRNA but not viral proteins are detectable in PBMCs freshly isolated from HTLV-1-infected individuals. The levels of HTLV-1 mRNA are higher in HAM/TSP patients than in ACs,(58) but viral proteins are still undetectable. Only a few reports have indicated HTLV-1 protein expression in situ.(59)
HTLV-1 expression in ATL cells immediately after isolation from the peripheral blood is very low, and becomes significantly induced after culture for some hours in vitro.(60,61) This phenomenon is observed in about one half of ATL patients regardless of the disease severity.(62) Viral induction after in vitro culture does not occur in the other one half of ATL patients, probably because of genetic and epigenetic changes in the viral genome.(63–65) Rapid induction of viral expression after in vitro culture has also been observed in PBMCs from HAM/TSP patients and ACs,(66) indicating that there must be a common mechanism for transiently suppressing HTLV-1 expression in vivo regardless of the diseases.
Suppression of HTLV-1 expression by type-I IFN responses.
Recently, we found that type-I IFN responses are involved in the suppression of HTLV-1 expression.(34) When HTLV-1-infected T-cell line cells were co-cultured with stromal cells such as epithelial cells and fibroblasts, HTLV-1 mRNA and proteins were markedly decreased in HTLV-1-infected cells. Similarly, induction of HTLV-1 expression in cultures of primary ATL cells was also suppressed by co-culture with stromal cells. Type-I IFNs were involved in the stromal cell-mediated suppression of HTLV-1 expression, because it was partly neutralized by anti-IFN-α/β receptor antibodies. Since efficient HTLV-1 expression is dependent on transactivation of its own LTR by Tax protein,(30,67) limitation of this protein below a certain level will lead to the maintenance of HTLV-1 expression at low levels. Stromal cells reduced viral expression via type-I IFNs, but did not reduce cell growth and even supported it by unknown mechanisms.(34,68)
It has been reported that plasmacytoid dendritic cells (pDCs), a major producer of type-I IFNs, are susceptible to HTLV-1 infection.(69,70) In ATL patients, pDCs are decreased in number and also lack the ability to produce IFN-α.(69) A recent report indicated that pDCs generate type-I IFNs mainly through TLR7 recognition of HTLV-1 RNA.(71) The precise mechanisms of the HTLV-1-mediated IFN responses remain to be clarified.
In addition to recombinant IFN-α and IFN-β, recombinant IFN-γ was also capable of reducing HTLV-1 expression to lesser extents in HTLV-1-infected cell lines.(34,72) Participation of type-II IFN-producing cells other than stromal cells in HTLV-1 suppression in vivo is also conceivable.
Potential involvement of type-I IFNs in HTLV-1 suppression in vivo.
In in vitro experiments, co-cultured stromal cells suppressed viral expression in HTLV-1-infected cells. Interestingly, when infected cells were re-isolated from the co-cultures, viral expression was restored to the original level over the following 48 h (Fig. 3).(34) This observation shows a striking similarity to the rapid induction of HTLV-1 expression in freshly isolated ATL cells after culture in vitro.
Involvement of type-I IFN responses in the suppression of HTLV-1 expression in vivo was confirmed using interferon regulatory factor-7-KO mice, which are deficient in most type-I IFN responses. Viral expression in HTLV-1-infected cells was significantly suppressed when the infected cells were intraperitoneally injected into WT mice but not into interferon regulatory factor-7-KO mice.(34)
It is speculated that the levels of viral expression in HTLV-1-infected lymphocytes may differ among various tissues depending upon the strength of IFN responses. Thus far, there is little information regarding HTLV-1 expression in various tissues. In transgenic mice with an HTLV-1 LTR-driven construct of the pX gene, expression of the transgene was only observed in limited organs including the central nervous system, eyes, salivary glands and joints.(73) It is intriguing that all of these tissues are involved in human inflammatory diseases related to HTLV-1 infection. Such coincidences suggest the involvement of HTLV-1 gene expression in the pathogenesis of these inflammatory diseases.
Double control of HTLV-1 by innate and acquire immunity
Relationship between acquired and innate immune control in HTLV-1 infection.
At the primary infection, type-I IFNs generally play a critical role in limiting viral replication, and have positive effects on antigen presentation by activating DCs, inducing type-II IFN, and upregulating MHC-I, which subsequently augments T-cell responses.(74) However, the role of type-I IFNs in the chronic phase of viral infection may not always be positive. In HIV-1 infection, type-I IFNs may be a progressive factor for the disease by accelerating T-cell exhaustion.(75)
Suppression of HTLV-1 expression by type-I IFNs may reduce the efficacy of T-cell-mediated surveillance against HTLV-1-infected cells, because T-cells require viral proteins for recognition. On the contrary, if the IFN-mediated suppressive system is insufficient, HTLV-1-specific T-cell responses will be activated in response to viral antigens.
The relationship between innate and acquired immunity may also differ among tissues. In tissues with strong IFN responses, viral expression in the infected cells would be suppressed and CTLs would ignore these cells. However, in tissues with weak IFN responses, infected cells would express viral antigens to be recognized by CTLs (Fig. 3). These presumptions can explain the status of HTLV-1-infected cells in vivo, which comprises a large reservoir of infected cells without viral expression and a low-efficiency surveillance system by CTLs that can only work on limited occasions.
Potential relationship between disease manifestation and innate and acquired host immunity in HTLV-1 infection.
Although suppression of HTLV-1 expression may partly interfere with the efficacy of T-cell immunity, it may contribute to a slowing down of the Tax-mediated pathogenesis, tumorigenesis and inflammation (Fig. 4). In a rat model, shRNA-mediated suppression of Tax in HTLV-1-transformed cells rendered these cells resistant to Tax-specific CTLs but also reduced their ability for tumorigenesis in vivo.(76) Continuous suppression of HTLV-1 expression in humans may have a similar decelerating effect against Tax-mediated tumorigenesis. This might be a reason why it takes so long for ATL to develop. So long as the viral expression is well controlled, the viral pathogenesis may not be apparent until malignant cell clones finally come through the process of clonal evolution in the infected cell reservoir. Without proper T-cell responses, the emergence of such clones may occur earlier, because they would have more chance to survive.
HAM/TSP patients show elevated levels of viral expression for an unknown reason. Increased levels of inflammatory cytokines could be either a cause or a result of this phenomenon. The involvement of HTLV-1 proviral integration sites in transcription units in elevated viral expression has also been suggested.(77) An experimental rat model of HAM/TSP using a certain WKAH strain exhibits increased Tax mRNA expression in the spinal cord without T-cell infiltration,(78) suggesting that viral expression is a primary event while T-cell responses are not. Further studies revealed that this particular rat strain contains mutations in the promoter region of the IL-12 receptor, which potentially lead to reduced IFN-γ production in the spinal cord.(72) The associations of genetic factors related to the IFN system with HAM/TSP patients have remained obscure. Very recently, a gene expression profiling study indicated that expression of suppressor of cytokine signaling 1 (SOCS1) is upregulated in HAM/TSP patients and ACs, and is positively correlated with high HTLV-1 mRNA loads.(79)
HTLV-1 is controlled by both acquired and innate immunity. HTLV-1-specific T-cells contribute to antitumor surveillance, and type-I IFNs contribute to silencing viral expression. The presence of the double control systems with partial conflicts would explain some of the puzzles in HTLV-1 infection, such as the transient suppression of viral expression in vivo, apparently reciprocal occurrence of ATL and HAM/TSP, inconsistent correlations of proviral loads with T-cell responses, and a long incubation period.
Insufficient T-cell responses are regarded as a risk factor for ATL, and vaccines that augment HTLV-1-specific T-cell responses would be beneficial in reducing the risk in a subpopulation of HTLV-1 carriers exhibiting insufficient T-cell responses and elevated proviral loads.
Innate immune responses in HTLV-1 infection should be further investigated, because they could be another important determinant of disease manifestation and represent therapeutic targets in HTLV-1-related diseases.
We thank Dr Jun Okamura (National Kyushu Cancer Center) for his invaluable advices and enormous efforts to co-ordinate basic and clinical investigators on HTLV-1 research in Japan. The authors have no conflicting financial interests. This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan, and the Ministry of Health, Labour, and Welfare of Japan.