Although human T cell lymphotropic/leukemia virus type I (HTLV-I) is the etiologic agent of adult T cell leukemia (ATL) and HTLV-I-associated myelopathy/tropical spastic paraparesis (HAM/TSP), the host and viral factors involved in HTLV-I-associated oncogenesis and neuropathogenesis remain to be fully defined (Yoshida et al., 1982; Bhagavati et al., 1988; Jacobson et al., 1988; Franchini et al., 1993; Yoshida, 1994). Only a small proportion (< 5%) of HTLV-I-infected individuals will develop HAM/TSP or ATL, while greater than 95% will remain asymptomatic for life (Hollsberg and Hafler, 1993). Previous studies have demonstrated that CD4+ T lymphocytes represent the primary cellular target for HTLV-I infection in the peripheral blood (PB), and Tax expression in this cellular compartment leads to activation of nuclear factor-κB (NF-κB)-, serum response factor (SRF)- and activating transcription factor/cyclic AMP responsive element binding protein (ATF/CREB)-responsive genes and accelerates cell cycle progression which ultimately leads to ATL (Ballard et al., 1988; Fujii et al., 1988, 1991, 1992; Leung and Nabel, 1988; Hirai et al., 1994; Kwok et al., 1996; Uchiyama, 1997). In addition, during the course of HAM/TSP, HTLV-I-infected CD4+ T cells have been shown to spontaneously proliferate in vitro and express several proinflammatory cytokines and adhesion molecules, allowing these cells to traffic from the PB into the central nervous system (CNS) (Jacobson et al., 1988; Dhawan et al., 1993; Furuya et al., 1997; Mogensen and Paludan, 2001).
Previous observations have indicated that the proviral DNA load directly correlated with the frequency of Tax-specific effector/memory CD8+ T cells in patients with HAM/TSP (Nagai et al., 2001a), suggesting that the Tax-specific cytotoxic T lymphocyte (CTL) response is being driven by the proviral DNA load. Recently, however, the CD8+ T cell population has been demonstrated to represent an additional viral reservoir in vivo (Nagai et al., 2001b). Furthermore, utilizing a combination of BrdU or Ki-67 (specific for proliferating cell nuclear antigen) and HLA-A*0201/Tax(11-19) tetramer staining and flow cytometry, CD8+ T cells were found to be the predominant expanding HTLV-I-infected cell population during spontaneous proliferation in vitro (Sakai et al., 2001). Together, these observations suggest that not only is the Tax-specific CTL response driven by the relative level of proviral DNA load, but expansion of this cell population may also contribute significantly to the total proviral DNA load present in the PB and CNS during HTLV-I-induced neurologic disease (Nagai et al., 2001b; Sakai et al., 2001). The condition that develops then, may be a positive autoregulatory loop, where proviral DNA load drives the expansion of HTLV-I-specific CD8+ T cells, a significant proportion of which may be infected, thus increasing the proviral load. Antigen-specific lysing of infected cells, including CD8+ T cells, coupled with the expansion of HTLV-I-specific CD8+ CTLs may allow only relatively small changes in the proviral DNA load to occur during the progression of HAM/TSP, while promoting the development of an increasingly aggressive immune response against HTLV-I. Although transcriptional regulation and viral gene expression of the HTLV-I proviral genome has been extensively studied in CD4+ T cells, further examination of the regulatory mechanisms governing viral gene expression within CD8+ T cells may lead to a greater understanding of the role that HTLV-I+ CD8+ T cells play in the PB and CNS during the progression of HTLV-induced neurologic disease.
Within the CNS, CD4+ T cells represent at least one cell type where studies have confirmed the presence of proviral DNA by PCR-in situ hybridization and expression of viral gene products by in situ hybridization (Hara et al., 1994; Moritoyo et al., 1996). The other cell type that has been found to be infected with HTLV-I in the CNS is the resident astrocyte population (Lehky et al., 1995). Viral gene expression within the CNS not only promotes activated T cell and monocyte migration into the CNS as a result of proinflammatory cytokine and chemokine expression, but also targets these infected cells for HTLV-I-specific CD8+ CTL-mediated lysis (Umehara et al., 1994b, 1996; Bertini et al., 1995; Mogensen and Paludan, 2001). HTLV-I infection and Tax expression has been shown to cause the overexpression of TNF-α in astrocytes, resulting in the upregulation of MHC class I, proinflammatory cytokines and adhesion molecules, reduction in glutamate uptake, oligodendrocyte and myelin damage, and neurotoxicity (Selmaj and Raine, 1988; Mendez et al., 1997; Szymocha et al., 2000a,b). Although controversial, other potential cell types harboring HTLV-I proviral DNA in the CNS may include macrophages, microglial cells, oligodendrocytes, and neurons. However, HTLV-I proviral DNA or viral RNA has not yet been conclusively demonstrated in these cell types utilizing PCR-in situ hybridization and in situ hybridization, respectively (Hara et al., 1994; Kuroda et al., 1994; Lehky et al., 1995; Moritoyo et al., 1996). Interestingly, HTLV-I infection of PB monocytes and dendritic cells has been shown in vivo and in vitro (Koralnik et al., 1992; Macatonia et al., 1992; Koyanagi et al., 1993; Makino et al., 1999), suggesting that perivascular macrophages and microglial cells derived from PB monocytes may also be infected. To date, there has been no conclusive evidence indicating that HTLV-I-infected cells of the monocytic lineage are present in the CNS. However, several indirect lines of evidence suggest that cells of the monocyte/macrophage lineage within the CNS, including microglial cells, may be infected with the virus. First, CNS macrophages and perivascular microglial cells are continuously replenished from PB monocytes, where at least a small proportion of the monocyte population is infected in patients with HAM/TSP (Koyanagi et al., 1993; Santambrogio et al., 2001). Second, microglial cells are susceptible to infection with HTLV-I in vitro (Hoffman et al., 1992), suggesting that if these cells come into contact with infiltrating HTLV-I+ CD4+ T cells or HTLV-I+ astrocytes, cell–cell contact could result in transmission of HTLV-I to this resident CNS cell population. Third, in situ hybridization coupled with immunohistochemical analysis have shown that HTLV-I-specific RNA expression co-localized mainly to CD4+ T cells or GFAP+ astrocytes (Lehky et al., 1995; Moritoyo et al., 1996). However, a significant amount of probe hybridized to cell types that remained unidentifiable. This observation, coupled with the decreased level of sensitivity of such a dual-labeling procedure leaves open the possibility that other cell types within the CNS could also be infected with HTLV-I and expressing low levels of viral RNA, making them difficult to detect by in situ hybridization. Finally, the relative levels of proviral DNA, viral gene expression, and the relative proportions of CD4+ T cells, CD8+ T cells, and activated macrophages and microglial cells in and around the sites of CNS lesions fluctuate with duration of disease (Umehara et al., 1993; Kubota et al., 1994; Moritoyo et al., 1996). Therefore, the numbers of HTLV-I+ cells in the CNS would also be expected to change during this transition. Previous studies that have examined proviral DNA content and viral gene expression within the CNS have focused on HAM/TSP patients with a longer duration of disease, where there is a lower overall level of viral gene expression and lower numbers of macrophages, microglial cells, and CD4+ T cells, compared to patients with a shorter duration of disease (Umehara et al., 1993; Kubota et al., 1994; Lehky et al., 1995; Moritoyo et al., 1996). Collectively, these studies suggest that the ability to detect proviral DNA and viral gene expression within the CNS may vary with the duration of disease. Accordingly, examination of the proviral DNA and viral RNA levels within the CNS of HAM/TSP patients with a shorter duration of disease is necessary to determine what role CNS macrophages and microglial cells may play in the pathogenesis of HAM/TSP.
Whether or not cells of monocyte/macrophage lineage within the CNS are infected with HTLV-I, their role in the pathogenesis of HAM/TSP cannot be overlooked, as it has already been shown that macrophage and microglial cell activation in the CNS closely correlated with HTLV-I proviral DNA load in the CNS and progression of neurologic disease (Abe et al., 1999). Furthermore, infected dendritic cells from HAM/TSP patients exhibit an increased stimulatory capacity with respect to activation of both CD4+ and CD8+ T cells (Makino et al., 1999). This has profound implications with regard to the generation of CD8+ CTL responses directed against both viral and possibly even self antigens within CNS tissues. Whether through cross presentation of viral antigens derived from phagocytosed viral particles or apoptotic cells infected with HTLV-I, or by direct infection of the dendritic cell population by viral particles or infected cells, processing and presentation of immunodominant viral epitopes on the cell surface in the context of MHC class I represents a prime mechanism for the activation of HTLV-I-specific CD8+ CTLs. Hyperactive T cell stimulation mediated by infected dendritic cells in the lymph nodes (LN) and spleen, and infected macrophages and microglial cells in the CNS as a result of aberrant expression of MHC class I, MHC class II, and B7-like costimulatory molecules, as well as chemokine and pro-inflammatory cytokine production, may precipitate the onset of neurologic disease.
Viral gene expression within infected cells, cytokine dysregulation, anti-HTLV-I antibody production, and infiltration of Tax (11–19)-specific CD8+ CTLs and activated macrophages into the CNS has been postulated to play a critical role in the pathogenesis of HAM/TSP (Levin et al., 1997a, 1998). With respect to bone marrow (BM) from individuals with HAM/TSP and likely a subset of asymptomatic carriers, previous studies utilizing PCR-in situ hybridization analyses have demonstrated that a vast majority of cells within the BM are positive for HTLV-I proviral DNA, but negative for viral RNA expression, suggesting the presence of an extensive latent infection (Levin et al., 1997b). These results contrast those obtained from BM taken from individuals with ATL, where no HTLV-I proviral DNA+ CD34+ hematopoietic progenitor cells were detected (Nagafuji et al., 1993). Additional in vitro analyses and studies performed in reconstituted SCID-hu mice have also demonstrated that BM and PB CD34+ hematopoietic progenitor cells are susceptible to HTLV-I infection, and that the proviral genome is maintained after in vitro and in vivo differentiation down multiple hematopoietic lineages, including the T lymphocyte and monocyte/macrophage lineages (Feuer et al., 1996). Normal trafficking of CD4+ and CD8+ T cells through the BM compartment is a likely mechanism allowing HTLV-I to gain access to the hematopoietic progenitor cell population. Cumulatively, these results strongly suggest that the interaction of HTLV-I-infected T cells with CD34+ progenitor cells in the BM or PB compartments during the course of HAM/TSP likely leads to extensive viral invasion of the progenitor cell population including the lymphoid and myeloid progenitor cell compartments. However, HTLV-I-specific in situ hybridization and immunohistochemical analyses suggest that viral gene expression in CD34+ progenitor cells and cells of the monocyte/macrophage lineage in BM, PB, or CNS compartments is far more restricted than in productively infected CD4+ and CD8+ T cells. Therefore, HTLV-I infection of CD34+ progenitor cells may serve as a latently infected reservoir during the course of HAM/TSP. As CD34+ progenitor cells in the BM differentiate and migrate into the PB, viral gene expression may be initiated and maintained at a regulated, low level by several key cell type- and differentiation stage-specific transcription factors that interact with the viral promoter. Consequently, HTLV-I genomic activation and viral gene expression in infected cells (including CD4+ and CD8+ T cells, B cells, monocytes/macrophages, dendritic cells, and microglial cells) may provide an additional and continuous supply of infected cells that lead to an increase in PB proviral load and viral gene expression, resulting in immune dysregulation including the generation of cytotoxic Tax-specific CD8+ T cells, anti-HTLV-I antibodies, and neurotoxic cytokines involved in disruption of myelin-producing cells and neurodegeneration (Fig. 1).
VIRAL TRANSMISSION AND PRODUCTIVE T CELL INFECTION
HTLV-I can be transmitted from an infected individual to an uninfected individual by sexual contact, exposure to contaminated blood products, and from mother to child through either intrauterine contact or breast-feeding. The initial route of infection is hypothesized to allow the virus to access a unique subset of target cells within the host. T cell migration into peripheral tissues is limited in the absence of inflammation, resulting in a greater number of CD4+ and CD8+ T cells in the PB compared to peripheral tissues and mucosal membranes. In contrast, dendritic cells and macrophages represent only ∼ 2% of PB mononuclear cells (PBMCs) present in the PB, but are found in great abundance in the peripheral tissues and mucosal membranes. Therefore, exposure of the PB compartment to virus may facilitate the infection of a large number of CD4+ and CD8+ T cells and a relatively small number of dendritic cells and macrophages. In contrast, exposure of the mucosa to HTLV-I may facilitate the infection of mostly dendritic cells and macrophages and only a relatively small number of CD4+ and CD8+ T cells. The ability of HTLV-I to stimulate CD4+ and CD8+ T cells to enter the cell cycle and promote high levels of viral gene expression are believed to be key factors regulating productive infection within these cell populations. However, dendritic cells and macrophages are in a post-mitotic phase of their life cycle, and as a consequence, may be capable of promoting only a relatively low level of viral gene expression.
The current understanding of the natural history of HTLV-I infection derives predominantly from events that have been observed to take place within the CD4+ T cell population. Several reasons for this focus likely include: (1) observations that the vast majority (> 95%) of leukemic cells found in patients with ATL were CD4+ T lymphocytes; (2) initial reports that the CD4+ T cell compartment acted as the only major viral reservoir in vivo; and (3) the presence of CD4+ viral RNA+ T cells within the CNS of patients with HAM/TSP (Popovic et al., 1983; Richardson et al., 1990; Moritoyo et al., 1996). Therefore, most of the molecular pathogenesis of HTLV-I infection discussed in this and subsequent sections is based on observations that have focused on events that have taken place within the CD4+ T cell population.
The transmission of infected cells is believed to be required for infection of the new host to occur, since there is little, if any, cell-free virus found in plasma (Wodarz et al., 1999). Penetration of the host cell plasma membrane by HTLV-I is followed by reverse transcription and uncoating of the core, followed by integration of the proviral genome. Integration appears to occur at a random location within the host cell genome, although a preference for A/T-rich sequences has been suggested (Chou et al., 1996, Leclercq et al., 2000a,b). As with other retroviruses, the integrated HTLV-I proviral genome is flanked by non-coding long terminal repeat (LTR) sequences comprised of three regions, U3, R, and U5 (Fig. 2), which contain information essential for the regulation of reverse transcription, integration, transcription, and replication. Evidence suggests that the LTR sequences of some retroviruses play a role in tissue- and cell type-specific viral tropism, and may also be involved in determining the course of disease associated with infection (Chen et al., 1984; Corboy et al., 1992; Gonzalez-Dunia et al., 1992; Ross et al., 2001). For example, when transgenic mice were generated utilizing the LTRs from either T cell-tropic human immunodeficiency virus type 1 (HIV-1) strains or CNS-derived HIV-1 strains, expression of the reporter gene within the nervous system was detected only in mice transgenic for CNS-derived LTRs (Corboy et al., 1992). In addition, complementary studies performed to address the role of the retroviral LTR in cell type-specific viral gene expression utilized transgenic mice containing a β-gal transgene driven by an HTLV-I LTR isolated from a patient with HAM/TSP. The resultant LTR-directed expression occurred primarily within the CNS (Gonzalez-Dunia et al., 1992). Furthermore, several HTLV-I proviruses isolated from ATL or HAM/TSP patients as well as HTLV-I-infected asymptomatic carriers have been isolated and sequenced (Yoshida et al., 1982; Josephs et al., 1984; Jacobson et al., 1988; Gessian et al., 1991; Komurian et al., 1991; Nerurkar et al., 1993). Although there was 95% homology between the sequenced clones, a number of the differences in nucleotide sequence between the clones lie in and around the U3 region of the LTR, which contains the viral promoter (Paine et al., 1991; Ratner et al., 1991). Although some evidence has suggested that nucleotide sequence variation within the LTR may not yet correlate in an obvious manner with disease phenotype (Daenke et al., 1990), the precise interactions of cellular factors, the HTLV-I-encoded trans-activator protein Tax, and specific LTR sequences that play a role in determining the pathogenic course after infection are not yet fully defined and must continue to be examined. The emerging notion that retroviral sequence variation within the viral promoter, even at the single nucleotide level, can dramatically affect recruitment of activating or repressing transcription factors and determine the outcome of viral gene expression within infected cells, has recently been described (Ross et al., 2001). These results have suggested that sequence variation within the retroviral LTR may have great impact on viral gene expression within specific cell populations including the monocyte/macrophage lineage.
Studies of the early events that occur during natural HTLV-I infection have been hindered by the lack of an animal model that can mimic the pathogenic features found in HTLV-I-infected humans. Reports utilizing mouse, rat, and rabbit models of HTLV-I infection have shown some similarities to events observed in humans (Miwa et al., 1997; Albrecht et al., 1998; Kasai et al., 1999; Robek and Ratner, 1999; Lairmore et al., 2000; Ratner et al., 2000). More recently, two groups have described evidence derived from experimentally infected squirrel monkeys that point to a two step replication strategy for HTLV-I in vivo (Kazanji et al., 2000; Mortreux et al., 2001). During initial infection of susceptible cells within the host, evidence suggests that there is a transient phase of viral replication that is mediated by reverse transcription, accompanied by viral gene expression (Mortreux et al., 2001). Thereafter, viral replication proceeds via persistent clonal expansion of infected T cells harboring the integrated viral genome (Mortreux et al., 2001). During this later phase of viral replication, viral gene expression was silent (Kazanji et al., 2000). Squirrel monkeys infected intravenously with HTLV-I-transformed monkey cells were examined with respect to proviral DNA load at various time points up to 26 months after infection. Proviral DNA was detected early after infection in PBMCs and BM, and at later times in the LN, spleen, and spinal cord (Kazanji et al., 2000). A two-step replication strategy was suggested by the low mutation rate observed during HTLV-I replication (7 × 10−6 mutations per base pair per replication cycle), compared to the mutation rate observed during HIV-1 replication (3 × 10−5) (Mansky, 2000). However, even with this relatively low mutation rate (four-fold lower than HIV-1), this would not be able to fully account for the low level of genetic drift detected in cloned HTLV-I sequences, suggesting that replication of the integrated provirus during cellular DNA replication, rather than during the process of reverse transcription, is the primary mode of HTLV-I proviral DNA amplification (Mansky, 2000; Mortreux et al., 2001).
Viral gene expression within CD4+ T cells has been reported to be at least five-fold greater than that observed in CD8+ T cells, suggesting that viral gene expression may be more efficient in CD4+ T cells than in CD8+ T cells (Newbound et al., 1996). However, recent results have demonstrated the predominant expansion of CD8+ T cells during spontaneous proliferation in vitro, a process dependent on Tax-mediated trans-activation of the NF-κB pathway (Robek and Ratner, 1999; Sakai et al., 2001). Studies performed utilizing the HTLV-I ACH molecular clone in a transgenic mouse model have shown that HTLV-I can immortalize both primary CD4+ and CD8+ T cells. This activity was dependent on the ability of Tax to trans-activate the NF-κB pathway, while trans-activation through the ATF/CREB pathway was dispensable (Robek and Ratner, 1999). However, the molecular clone encoding the M47 Tax mutant (which is incapable of trans-activating the ATF/CREB pathway) exhibited preferential immortalization of CD8+ T cells instead of the preferential immortalization of CD4+ T cells observed with the wild type ACH molecular clone. This event might be explained by the relative levels of NF-κB and ATF/CREB factors present within each cell population (Newbound et al., 1996; Robek and Ratner, 1999; Ratner et al., 2000).
Activation of infected T cells occurs through several independent pathways, including protein kinase A (PKA), protein kinase C (PKC), mitogen-activated protein kinase (MAPK), and calcium-dependent signaling pathways (reviewed in Hollsberg, 1999). Basal activation of the HTLV-I LTR by cellular transcription factors results in the expression of low levels of viral proteins, including Tax and p12I. These viral proteins mimic naturally occurring stimulation through the T cell antigen receptor (TCR), the T cell costimulatory receptor CD28, and the interleukin-2 receptor (IL-2R) complex, allowing for increased trans-activation of the HTLV-I LTR and key cellular promoters driving the expression of genes that lead to a constitutively activated and cycling HTLV-I proviral DNA+ T cell population. HTLV-I is not only capable of activating infected T cells, but also can activate uninfected T cells via cell–cell contact between an infected and uninfected cell. Activation of resting, uninfected T cells has been shown to be dependent on the expression of leukocyte function-associated antigen-1 (LFA-1), LFA-3, intracellular adhesion molecule-1 (ICAM-1), and CD2 adhesion molecules (Buckle et al., 1996).
Although most observations concerning productive HTLV-I infection have focused on the CD4+ T cell population, far less is known about the effects HTLV-I infection has on the CD8+ T cell population, and the regulation of viral gene expression within infected CD8+ T cells. Although recent evidence has suggested that a significant proportion of the CD8+ T cell population in PB and CNS may be infected with HTLV-I during HAM/TSP (Nagai et al., 2001b; Sakai et al., 2001), further investigations will be required to determine the role of the CD8+ T cell population during the pathogenesis of HTLV-I-associated diseases.
The genome of HTLV-I is that of a complex, type C retrovirus. Like all replication competent retroviruses, HTLV-I contains the structural genes group-specific antigen ( gag), protease ( pro)/polymerase ( pol), and envelope (env). In addition, the genome is comprised of the critically important pX region and the uniquely structured LTRs at the ends of the viral genome (Fig. 2). Although the envelope glycoproteins encoded by the env gene determines the cellular tropism of a retrovirus, it is the viral promoter consisting of the U3 region of the LTR that in part determines whether that cell will be permissive for viral replication, including viral gene expression, viral particle assembly and egress. The pX region of HTLV-I lies near the 3′ region of the env gene and extends into the U3 region of the 3′ LTR. Within the pX region are four overlapping open reading frames encoding six functionally distinct proteins (Kanamori et al., 1990; Koralnik et al., 1993; Wagner and Green, 1993; Kubota et al., 1996; Derse et al., 1997; Lairmore et al., 2000).
HTLV-I LTR-driven gene expression within target cell populations is dependent on cis-acting enhancer sequences comprised of three 21 bp imperfectly repeated elements collectively designated as Tax-responsive element 1 (TRE-1) located within the U3 region of the LTR at positions −251 to −231 (promoter distal (pd), repeat I), −203 to −183 (promoter central (pc), repeat II), and −103 to −83 (promoter proximal (pp), repeat III) relative to the transcriptional start site (Brady et al., 1987; Tillmann et al., 1994; Wessner et al., 1995, 1997; Barnhart et al., 1997; Wessner and Wigdahl, 1997; reviewed in Yao and Wigdahl, 2000) (Fig. 3). ATF/CREB transcription factors have been shown to bind to each of the 21 bp repeats, while activator protein-1 (AP-1) has been shown to bind specifically to the pc repeat, and Sp1/Sp3 factors have been shown to bind specifically to the pp repeat (Wessner et al., 1995, 1997; Wessner and Wigdahl, 1997) (Fig. 4). Activator protein-2 (AP-2) has also been shown to interact with each of the TRE-1 21 bp repeats (Muchardt et al., 1992). In addition, HTLV-I transcription is also dependent on a second enhancer element designated as Tax-responsive element 2 (TRE-2) located between the pc and pp repeats from −163 to −117 (Marriott et al., 1990; Tanimura et al., 1993). Numerous transcription factors have been shown by footprinting and electrophoretic mobility shift analyses to bind to TRE-2 Sp1/Sp3 including Ets factors (Ets-1, Ets-2, Elf-1, p-ets, and TIF-1), and c-Myb (Bosselut et al., 1990, 1992; Clark et al., 1993; Colgin and Nyborg, 1998; Torgeman et al., 1999). NF-κB, AP-2, PU.1, NHF-3, and PEA-3 transcription factors have been proposed to also interact with this region based on sequence analysis (Datta et al., 2000), but further experimentation will be required to determine if these cellular factors physically interact with this region. Basal transcription driven by the HTLV-I viral promoter is regulated in a complex and not yet fully defined manner by the activational and/or developmental status of the infected cell.
HTLV-I basal transcription is greatly enhanced by the virus-encoded regulatory protein Tax, which mediates trans-activation of the HTLV-I LTR by interacting with cellular transcription factors that bind to the three 21 bp repeats within TRE-1, in addition to TRE-2 (Nyborg et al., 1988; Bosselut et al., 1990, 1992; Nyborg and Dynan, 1990; Muchardt et al., 1992; Zhao and Giam, 1992; Clark et al., 1993; Franklin et al., 1993; Baranger et al., 1995; Brauweiler et al., 1995; Fujii et al., 1995; Barnhart et al., 1997; Lenzmeier and Nyborg, 1997; Reddy et al., 1997; Yan et al., 1998; Goren et al., 1999; Van Orden et al., 1999; Mick et al., 2000; Scoggin et al., 2001). Initially, it was thought that Tax did not interact directly with Tax-responsive sequences within the LTR, but instead trans-activated viral transcription by interacting with cellular transcription factors that directly interact with the TREs (Baranger et al., 1995; Perini et al., 1995; Giebler et al., 1997). However, latest evidence has demonstrated that Tax can also bind to a very restricted G/C-rich region that flanks the core ATF/CREB binding site in the center of each of the 21-bp repeats comprising TRE-1. The region of the Tax protein that interacts with these flanking sequences lies within the CREB-binding protein (CBP) binding domain (Kimzey and Dynan, 1998, 1999). Thus, the current model of Tax-mediated trans-activation involves the interaction of Tax with conserved basic regions of cellular basic leucine zipper (bZIP) transcription factors, enhancing bZIP dimerization and DNA binding activity (Perini et al., 1995). In addition to the protein–protein contacts that Tax makes with bZIPs, Tax also makes protein-DNA contacts with sequences flanking the core bZIP binding site, resulting in a Tax-bZIP complex with higher relative affinity for specific target sites, than observed in the absence of Tax (Perini et al., 1995). A similar interaction between Tax and cellular transcription factors that bind to TRE-2 may also facilitate the direct interaction of Tax with DNA in this region. At present, it is not known precisely whether Tax interacts directly or indirectly through cellular transcription factors with TRE-2.
The individual 21 bp repeats comprising TRE-1 contain three completely conserved domains designated A, B, and C. Mutational analyses coupled with transient transfection analyses have demonstrated that the Tax responsiveness of the 21 bp repeats requires domains A and B or B and C. Domain B, the most pivotal of the three with respect to Tax responsiveness, is comprised of the first five base pairs of the cyclic AMP response element (CRE), TGACGTCA (Seeler et al., 1993). Although the three repeats are similar to one another with respect to nucleotide sequence, mutagenesis of the HTLV-I LTR has demonstrated that the 21 bp repeats are not functionally equivalent (Tillmann et al., 1994, Wessner et al., 1995, 1997; Barnhart et al., 1997; Wessner and Wigdahl, 1997; Yao and Wigdahl, 2000; Yao and Wigdahl, unpublished observations). Several cellular proteins of the ATF/CREB family, members of the AP-1 transcriptional complex, and the transcription factors Sp1 and Sp3 have been demonstrated to interact with TRE-1 (Fujii et al., 1995; Wessner et al., 1995, 1997; Wessner and Wigdahl, 1997) (Fig. 4). A number of studies have demonstrated that Tax increases the binding of selected ATF/CREB family members to the 21 bp repeats (Zhao and Giam, 1992; Giebler et al., 1997; Lenzmeier et al., 1998; Yao and Wigdahl, unpublished observations). Specifically, Tax interacts directly with CREB in a homodimeric or heterodimeric form with other ATF/CREB factors and promotes the stabilization of protein–protein dimerization and binding of these factors to the repeats. Studies performed to elucidate the regions of the CREB protein with which Tax interacts have mapped this region of CREB to the highly conserved DNA binding domain and reveal that an altered DNA-binding specificity of CREB occurs when complexed with Tax (Adya et al., 1994; Yin et al., 1995; Mick et al., 2000). Therefore, the enhanced recruitment of ATF/CREB factors to the TRE-1 21 bp repeats by Tax allows for Tax to be directly associated with DNA, and the transcriptional activation domain of Tax to be exposed. Additional studies have demonstrated that Tax can facilitate dimerization and DNA binding of a wide variety of bZIP family transcription factors to responsive cellular promoters (Wagner and Green, 1993; Adya et al., 1994; Baranger et al., 1995).
Although the interaction of Tax with ATF/CREB factors is a pivotal event in viral trans-activation, studies have demonstrated that sequences other than the CRE within the 21 bp repeats are critical for Tax-mediated trans-activation (Marriott et al., 1990; Tanimura et al., 1993). Therefore, transcription factors other than ATF/CREB family members appear to be involved in Tax-mediated trans-activation. AP-1 family proteins (c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunB, and JunD) bind specifically to TRE-1 at the pc repeat. AP-1 proteins are basic leucine zipper transcription factors comprised of Jun–Jun homodimers or Fos–Jun heterodimers, and recognize a 12-O-tetradecanoylphorbol 13-acetate (TPA)-responsive element (TRE)-like binding site that is nearly identical with the CRE-like binding site within the pc repeat. Molecular interactions between ATF/CREB and AP-1 transcription factors have been observed with respect to cellular promoters containing CREs and TREs (Sassone-Corsi et al., 1990; Masquilier and Sassone-Corsi, 1992), suggesting that the binding of AP-1 and ATF/CREB to the pc repeat may be a mutually exclusive event. The DNA binding activity of AP-1 is increased during T cell activation and monocytic differentiation in a PKC- and MAPK-dependent manner. Interestingly, the AP-1 family members that make up the AP-1 complex on the pc repeat have been found to be cell-type specific (Wessner et al., 1995, 1997; Barnhart et al., 1997; Yao and Wigdahl, 2000; Grant and Wigdahl, unpublished observations). Therefore, competition between AP-1 and ATF/CREB with respect to binding to TRE-1 elements may have important effects on the regulation of basal and Tax-mediated trans-activation.
Ets family transcription factors Ets-1, Ets-2, Elf-1, p-ets/TIF-1, and possibly PU.1 bind overlapping regions within TRE-2 (Bosselut et al., 1990; Clark et al., 1993; Gitlin et al., 1993). Ets-1 and Ets-2 have been shown to enhance both basal and Tax-mediated trans-activation of the LTR (Bosselut et al., 1990; Gitlin et al., 1993). Ets-1-mediated trans-activation was dependent on sequences spanning both TRE-1 and TRE-2, and a direct interaction between Ets-1 and cellular transcription factors that interact with TRE-1 (Gitlin et al., 1993). Possible candidates may include ATF/CREB, AP-1, and Sp1 (Gitlin et al., 1993). However, these studies were performed in NIH3T3, HeLa, or CV-1 cells. As a result, these studies did not take into account the facts that Ets-1 expression and DNA binding activity were decreased during T cell activation while Ets-2 expression was induced (Bhat et al., 1990). Elf-1 has been shown to interact with TRE-2 in T cells and its expression remains unchanged during T cell activation (Clark et al., 1993). PU.1 expression has been shown to be enhanced during monocytic differentiation (DeKoter et al., 1998; Fisher and Scott, 1998). Therefore, it is likely that Ets-mediated basal and Tax mediated trans-activation is a dynamic process, utilizing various Ets family members in a cell type- as well as activation and differentiation stage-specific manner.
c-Myb binds to six unique sites within the U3 and R regions of the LTR and is activated by signaling through the IL-2R and may provide another link between T cell activation and induction of LTR activation (Bosselut et al., 1992). However, c-Myb is downregulated during monocytic differentiation (Lee et al., 1987), suggesting that LTR transcriptional regulation may be cell-type specific. More recently, c-Myb and Tax were shown to compete for binding to the KIX domain of CBP (Colgin and Nyborg, 1998). c-Myb, AP-1, serine-133 phosphorylated CREB, and Tax are all able to bind to the KIX domain of CBP, and competition between these cellular transcription factors and Tax for limiting amounts of CBP may lead to dysregulated cellular gene expression. A more recent report has also demonstrated an interaction between Sp1 and p300/CBP (Billon et al., 1999).Tax not only enhances dimerization and DNA binding activity of ATF/CREB, but also facilitates the recruitment of the co-activator CBP in the absence of PKA-mediated phosphorylation of serine-133 on CREB. CBP and its homologue p300 exhibit intrinsic histone acetyltransferase (HAT) activity responsible for chromatin remodeling around the engaged promoter. This is complemented by the HAT activity of p300/CBP-associated factor (P/CAF). p300/CBP also functions by facilitating the recruitment of general transcription factors to the transcriptional initiation site (Colgin and Nyborg, 1998). Although a direct interaction between Tax and ATF/CREB has been studied extensively, a direct interaction between Tax and c-Myb or AP-1 has not yet been reported.
A direct interaction between Tax and Sp1 has been observed (Trejo et al., 1997), however, the role of Tax in regulating Sp1 binding to the LTR with respect to CREB is unknown at this time. Since Sp1/Sp3 and ATF/CREB binding sites on the pp repeat partially overlap, one would expect binding to be mutually exclusive. EMS competition analyses have shown that in the presence of Tax, CREB exhibits a higher affinity for the pp repeat than in the absence of Tax. This allows Tax-CREB to remain bound to the LTR even in the presence of Sp1 levels sufficient to displace CREB in the absence of Tax (reviewed in Yao and Wigdahl, 2000; Yao and Wigdahl, unpublished observations). In light of the competition between cellular transcription factors and Tax for limiting amounts of CBP in the infected cell, a possible mechanism for differential recruitment of cellular transcription factors to the LTR during the transition from basal transcriptional activation to Tax-mediated trans-activation is presented in Figure 5. After T cell activation, a transient phase of basal transcription mediated by cellular transcription factors allows for a low level of viral gene expression. This may be mediated through the interaction of c-Myb, AP-1, Sp1/Sp3, and ATF/CREB with the LTR and the KIX domain of CBP. Subsequent synthesis of Tax could shift p300/CBP and P/CAF recruitment away from c-Myb, Sp1/Sp3, and AP-1 towards ATF/CREB, possibly altering the structure of the LTR transcription initiation site, promoting increased levels and/or increased efficiency of viral gene expression. This process may also affect the regulation of cellular genes containing c-Myb-, AP-1-, Sp1/Sp3-, and ATF/CREB-responsive promoters.
A LTR region within the R/U5 region has been described, that when deleted, resulted in a 10-fold decrease in basal activation of the viral promoter (Kashanchi et al., 1993). This downstream regulatory element (DRE-1, +202 to +246) was bound by a 37 kDa cellular protein, later designated as YB-1 (Kashanchi et al., 1994) (Fig. 3). YB-1 is activated through signaling mediated by IL-2R engagement, providing yet another link between T cell activation and basal transcriptional activation of the LTR. Although results suggest that Tax may not directly alter the expression of YB-1, YB-1 has been shown to be upregulated in HTLV-I-immortalized T cells (Kashanchi et al., 1994; Harhaj et al., 1999). The function of the DRE-1 element in vivo is not yet known, but it may act either as a transcriptional regulatory element or as an RNA element. Interestingly, a repressive element located at +209 to +226 within DRE-1 exhibited a high affinity for an unidentified 70.5 kDa protein. This unidentified protein was found to interact with a 40.5 kDa protein that reacted with an anti-CREB antibody (Xu et al., 1994). This repressive element may function to help maintain the latent state of the proviral genome, the function of which may be counteracted by YB-1 during T cell activation, thus turning a repressive element into an enhancer element (Xu et al., 1994). Whether the CREB factor that binds to this repressive element is the transcriptional activator CREB-1, the transcriptional repressor CREB-2, or another member of the ATF/CREB family of transcription factors remains to be determined.
The balance between repressive and activating host and viral factors may indeed provide a mechanism to explain the initiation and maintenance of the latent state of the HTLV-I proviral genome within infected cells. From the evidence provided above, it is clear that the processes of T cell activation and monocytic differentiation, as well as cytokine– or cell–cell contact-mediated activation of potential cellular targets, may have a profound effect on regulating the level of viral gene expression, which could in turn lead to Tax-mediated dysregulation of cellular gene expression in specific cell populations within the immune and/or central nervous system.
VIRAL MECHANISMS REGULATING HTLV-I INFECTION IN TARGET CELLS
Encoded within the HTLV-I genome are several unique multifunctional regulatory and accessory proteins that are integrally involved in modulating viral gene expression and the activational status of the infected cell. The most extensively studied protein encoded by the pX region of the HTLV-I genome is p40IV (Tax), a 40 kDa protein encoded within the fourth pX open reading frame. Tax is a multifunctional oncoprotein, involved in transcriptional regulation, cell cycle control, and transformation (Fig. 2). It is believed that many of the pathogenic consequences associated with the progressive diseases ATL and HAM/TSP, as well as the many HTLV-I-associated disorders, are associated with a functional Tax protein. Tax localizes primarily to the nucleus in most cell types, but has been shown to shuttle between the nucleus and cytoplasm through the use of its nuclear localization sequence (NLS) located at the N-terminus of Tax between amino acids 2 and 59, and nuclear export sequence (NES) located near the dimerization domain between amino acids 188 and 202 (Burton et al., 2000, and Alefantis and Wigdahl, unpublished observations). Differences in subcellular localization of Tax may relate to its function within targeted cell populations. Tax localization in the cytoplasm of transfected cell lines results in the constitutive localization of NF-κB to the nucleus, whereas nuclear localization of Tax is required for Tax-mediated trans-activation of the LTR and responsive cellular promoters (Burton et al., 2000). The M22 and M47 Tax mutants have been indispensable with respect to dissecting the mechanisms of Tax-mediated trans-activation of the viral LTR and responsive cellular promoters and HTLV-I-induced transformation. The ability of Tax to trans-activate the LTR and cellular promoters is also modulated by phosphorylation of Tax at specific serine/threonine and tyrosine residues (Saggioro et al., 1994). Nuclear localization of Tax plays a pivotal role in the trans-activating function of this protein, while cytoplasmic localization may be an indication that Tax is being targeted for secretion.
Recent studies have indicated that Tax may act as an extracellular cytokine. Indeed, accumulating evidence suggests that extracellular Tax can act by binding to an unknown cell surface receptor(s) on target cells, and induce expression of a number of proinflammatory cytokines. Extracellular Tax has been shown to induce the production of interleukin-1β (IL-1β), IL-6, tumor necrosis factor-α (TNF-α) in macrophages, and microglial cells in a dose-dependent manner (Dhib-Jalbut et al., 1994). Additional studies will be required to determine the mechanism of extracellular Tax-mediated cytokine expression. With respect to HIV-1-induced neurologic disease, release of potentially neurotoxic viral gene products into the CNS by HTLV-I-infected cells may be a critical determinant in mediating neuroinflammation and bystander damage within the CNS.
p27III (Rex) has been shown to increase the nuclear export and cytoplasmic localization of nonspliced and singly spliced viral RNA and to stabilize interleukin-2 receptor α (IL-2Rα) mRNA (Kanamori et al., 1990) (Fig. 2). Rex contains a nuclear localization sequence (NLS) and a nuclear export sequence (NES), both of which are crucial for the proper functioning of Rex and productive viral replication (Kanamori et al., 1990; Kubota et al., 1996). Functionally, Rex has been shown to be similar to the HIV-1 Rev protein. The ability of Rex to selectively promote the nuclear export of unspliced and singly spliced viral RNAs has been shown to be dependent on the presence of a Rex-responsive element (RRE), not found in multiply-spliced viral RNAs or cellular RNA molecules (Kanamori et al., 1990; Kubota et al., 1996). p21I, a naturally occurring, truncated form of p27III Rex, has been shown to repress certain functions of full-length Rex, including its ability to shuttle between the nucleus and cytoplasm. Furthermore, the amount of Rex shown to localize to the nucleus was increased in the presence of p21I (Kubota et al., 1996). It has been hypothesized that p21I (which contains a NES, but not a NLS) was able to compete with Rex for unspliced and singly spliced viral RNA, and facilitate the accumulation of Rex in the nucleus (Kubota et al., 1996).
p12I has been shown to be a small (99 amino acid), multifunctional, weakly oncogenic, hydrophobic protein containing four SH3-binding motifs (PXXP), that localizes to cellular endomembranes (Fig. 2). One of the first functions for p12I to be elucidated was its capacity to bind to the 16 kDa subunit of the vacuolar H+ ATPase (Franchini et al., 1993). Since then, p12I has been shown to bind to the immature forms of the IL-2R β and γc chains, resulting in a reduction in IL-2Rβγc cell surface expression (Mulloy et al., 1996). In vivo, after an initial burst of proliferation, IL-2 limits the continued proliferation of activated T cells by downregulating the IL-2Rγc chain and STAT-mediated transcriptional activation of anti-apoptotic Bcl-2 family proteins, and increases susceptibility to apoptotic cell death (Li et al., 2001). Although p12I targets the cytoplasmic domains of the IL-2Rβγc and alters its cell surface expression, p12I has been observed to enhance activation of the Jak/STAT pathway (Nicot et al., 2001). Constitutive activation of the Jak/STAT pathway has been observed in IL-2-independent but not in IL-2-dependent HTLV-I-infected T cells, suggesting that p12I plays a role in decreasing the IL-2 requirement for clonal expansion of HTLV-I-infected T cells, and may provide an escape mechanism from IL-2-mediated activation-induced cell death (Migone et al., 1995; Li et al., 2001; Nicot et al., 2001; Waldmann et al., 2001). Because Tax trans-activates both IL-2 and the IL-2Rα genes in an NF-κB-dependent manner, while p12I downregulates the expression of IL-2R β and γc chains, the temporal expression pattern of p12I relative to Tax and Rex expression in vivo may be important in regulating the IL-2 status of HTLV-I-infected T cells (Mulloy et al., 1996; Hollsberg, 1999).
Investigations performed by Derse et al. (1997) have provided evidence that p12I is dispensable for viral infectivity, replication, gene expression, and immortalization of primary HTLV-I-infected T cells in vitro. However, in a more recent report, p12I has been found to play an important role during the early stages of infection in resting primary lymphocytes, possibly by activating resting T cells via its interaction with the IL-2R complex (Albrecht et al., 2000). Additional studies have also shown another mechanism whereby HTLV-I may be able to activate newly targeted resting T cells. p12I can stimulate calcium-dependent signaling pathways by binding to calnexin and calreticulin in the ER, leading to a 20-fold increase in NFAT-mediated trans-activation (Ding et al., 2001). p12I was also shown to interact directly with the free immature form of the MHC class I heavy chain molecule in the ER. This interaction results in a failure of the MHC class I heavy chain to associate with β2 microglobulin, the retrotranslocation of the MHC class I heavy chain into the cytosol, and its subsequent degradation by the proteosome (Johnson et al., 2001). However, p12I does not target cell surface MHC class I molecules for degradation; instead it prevents newly synthesized molecules from assembling complexes with β2 microglobulin and antigenic peptide within the ER and trafficking to the cell surface. Some of the functions just described for p12I parallel those associated with HIV-1 Nef.
Intriguingly, two natural variants of p12I have been reported, one found mainly in patients with HAM/TSP (R88, which is less stable) and the other mainly found in asymptomatic carriers and patients with ATL (K88, which is more stable). The less stable R88 variant has been shown to be ubiquitinated, and may differ in its ability to modulate the immune response (Johnson et al., 2001). A functional correlation between the activity of p12I and disease progression in asymptomatic carriers may provide valuable information with respect to the pathogenesis of both HAM/TSP and ATL. The effects of p12I on MHC class I cell surface expression have been shown to be antagonistic to that of Tax, which upregulates cell surface expression of MHC class I. Therefore, the balance between Tax-mediated upregulation and p12I-mediated downregulation of MHC class I on the cell surface may determine the fate of that cell during infection. Furthermore, it is possible that during HAM/TSP, infected activated CD4+ and CD8+ T cells and monocytes traffic to sites of inflammation within the CNS where the levels of proinflammatory cytokines are elevated. IFN-γ-and/or TNF-α-mediated upregulation of MHC class I expression could override the effects of p12I and tip the scales in favor of MHC class I upregulation on the surface of the infected cell, thus labeling that cell as a target for HTLV-I-specific CTL lysis. During viral infection, p12I may function by facilitating escape from immune surveillance activity, leading to unchallenged clonal expansion of infected T cells in the PB and CNS, and a subsequent overall increase in proviral DNA load.
p13II, an 87 amino acid protein, has been shown to primarily localize to the mitochondria and interact with several cellular proteins (Hou et al., 2000) (Fig. 2). Localization of this protein to the mitochondria has been shown to induce changes in mitochondrial morphology from a normal string-like dispersed network, to spherical clusters. Disruption of the mitochondrial inner membrane potential has been shown in cells expressing p13II, an early characteristic of cells committed to apoptotic cell death, however, cytochrome c is not released into cytosol (Ciminale et al., 1999). Although the function of p13II in the context of infection with HTLV-I remains to be determined, it does appear that p13II causes mild swelling of the mitochondria, while preserving the integrity of the outer mitochondrial membrane. The selective advantage of such a function for p13II remains to be determined. Further examination of the interaction of p13II with cellular regulators of apoptosis or other viral proteins may elucidate the potential role(s) of p13II during infection.
p30II, a 241 amino acid protein, has been shown to contain the p13II coding sequence at its C-terminus and localizes to the nucleus of transfected cells (Ciminale et al., 1999; Bartoe et al., 2000) (Fig. 2). The latter observation has suggested that p30II may act as a transcription factor, and like Tax, regulate viral and cellular gene expression. Indeed, a trans-activation domain(s) has been mapped to amino acids 62–220 of the protein (Zhang et al., 2000). Furthermore, the central region of p30II also contains a NLS and serine- and threonine-rich regions, with the latter exhibiting distant homology with the cellular transcription factors Oct-1, Oct-2, Pit-1, and POU-M1 (Ciminale et al., 1999). The effects of p30II in transfected 293 cells were dose-dependent and differentially affected transcriptional activation of the HTLV-I LTR and cellular CRE-containing promoters. Low amounts of p30II are able to trans-activate the LTR, either in the absence or presence of Tax. Increasing levels of p30II begin to repress activation of the LTR. However, both low and high levels of p30II repressed activation of CRE-containing promoters, whether or not Tax was present (Zhang et al., 2000). The differential activity of p30II with respect to regulating transcription from the LTR and CRE-containing cellular promoters suggests that during infection, p30II may facilitate trans-activation of the LTR while downregulating cellular gene expression (Zhang et al., 2000). Recent observations have demonstrated that p30II can also bind to the KIX domain of p300/CBP (Zhang et al., 2000), suggesting that competition between p30II, Tax, and cellular transcription factors for binding to the KIX domain of p300/CBP may dramatically affect the trans-activational status of the HTLV-I LTR and responsive cellular promoters. Further studies need to be performed to examine the expression levels of p30II in vivo and to determine if a direct interaction between Tax and p30II occurs within the nucleus. A functional interaction between Tax and p30II has not yet been described, but such an interaction may aid in the preferential recruitment of Tax and p30II to the HTLV-I LTR. Alternatively, an interaction between Tax and p30II may interfere with the ability of Tax to recruit CBP and general transcription factors to the LTR and responsive cellular promoters (Fig. 5). Recent observations made utilizing rabbits experimentally infected with the HTLV-I ACH molecular clone demonstrated that p13II and/or p30II were important for maintaining high proviral DNA load in vivo (Bartoe et al., 2000). Given the functions of p30II described above, these results indicate that p30II may have an important role in promoting greater levels of viral gene expression, possibly at the expense of cellular gene transcription during infection.
PRIMARY IMMUNE RESPONSE AND SURVEILLANCE ACTIVITY
It has been proposed that the initial route of viral entry in a majority of individuals suffering from HAM/TSP involves exposure to HTLV-I within the PB compartment (Osame et al., 1990) (Fig. 6). Primary infection based in the PB results primarily in the infection of the CD4+ and CD8+ T cell compartments. Alternatively, infection of the mucosa likely leads to the infection of a different subset of target cells, including dendritic cells and macrophages, with relatively low numbers of CD4+ and CD8+ T cells being infected. Recent evidence has demonstrated that antigen exposure within the gut (a representative mucosal surface) was first detected by the dendritic cell population, which proceeded to express TGF-β and promote an immunosuppressive Th2/Th3 type response (Viney et al., 1998; Akbari et al., 2001). This immunosuppressive environment supported by infected dendritic cells within the mucosa and regional LNs may inhibit the activation and clonal expansion of antigen specific T cell populations, ultimately resulting in a low level of mitotic HTLV-I amplification within the CD4+ and CD8+ T cell compartments. In contrast, primary infection of the PB compartment involving the CD4+ and CD8+ T cell populations may result in a higher level of viral gene expression than that found after primary infection of the mucosa. This higher level of viral gene expression may be a direct result of the ability of HTLV-I to stimulate T lymphocyte subsets to enter the cell cycle, thus promoting mitotic HTLV-I amplification, which could subsequently lead to higher proviral DNA load and viral gene expression. Furthermore, the expansion of the infected CD4+ and CD8+ T cell populations likely results in infectious HTLV-I amplification within the monocytic and dendritic cell compartments as well as resident cells within the BM and CNS. The resultant level of viral gene expression within a given cellular compartment represents one of the principle factors driving the expansion of the HTLV-I-specific immune response characterized by anti-HTLV-I antibody production, Tax-specific CD8+ CTL expansion, and proinflammatory cytokine production (reviewed in Jacobson, 1996; Bangham, 2000). Additional host, environmental, and viral factors may also play a role in predetermining the outcome of HTLV-I infection within a given immune compartment.
In general, the etiology of HAM/TSP and the role of HTLV-I in this pathogenic process can be divided into three general phases: primary infection, clinical latency, and symptomatic HAM/TSP. Although the exact sequence of events that occur during the initial cell–virus interaction are not known in detail, it is likely that initial virus infection involves viral invasion of several major cellular compartments that include CD4+ and CD8+ T lymphocytes, dendritic cells, monocytes, and B cells (Richardson et al., 1990; Macatonia et al., 1992; Koyanagi et al., 1993; Nagai et al., 2001) that reside in the PB and lymphoid organs. As described above, an initial, transient phase of reverse transcription of viral RNA is followed by a persistent phase of clonal expansion within the CD4+ and CD8+ T cell populations (Mansky, 2000; Mortreux et al., 2001). Compared to the productive infection observed in CD4+ and CD8+ T cells, HTLV-I infection of cells of the monocyte/macrophage lineage and dendritic cells is likely to resemble a low level, persistent viral infection, involving a relatively small number of cells with very little viral gene expression and low amounts of infectious virus production, a result that is likely attributable to their post-mitotic status and relatively short lifetime (Banchereau and Steinman, 1998; Valledor et al., 1998). It is this differential capacity for clonal expansion and viral gene expression between T lymphocytes and dendritic cells and macrophages that may determine the level of viral gene expression that drives the immune response generated against HTLV-I.
Although dendritic cells may only be capable of initiating a relatively low level of viral gene expression, recent evidence has suggested that HTLV-I-infected dendritic cells from HAM/TSP patients exhibit an enhance capacity to stimulate antigen-specific CD4+ and CD8+ T cell activation (Makino et al., 1999). Dendritic cells are the most potent antigen presenting cells in the immune system, as indicated by their ability to: (1) efficiently capture antigen in the periphery; (2) migrate from the site of antigen capture to the secondary lymphoid organs while undergoing maturation, and efficient processing and presentation of antigen on the cell surface in the context of MHC class I and II molecules; (3) display unprocessed antigens in their native conformations on the cell surface for extended periods; (4) express high levels of co-stimulatory molecules and cytokines; and (5) stimulate resting, naïve, and memory T cells and B cells (Banchereau and Steinman, 1998). Two major lineages of dendritic cells have been described in humans, each with partially overlapping phenotypic and functional characteristics. Myeloid dendritic cells (DC1s) are localized at the site of pathogen entry, mainly in the skin and mucosal tissues, whereas lymphoid dendritic cells (DC2s) are thought to be localized primarily within the PB (Liu et al., 2001). After antigen exposure, both subsets are able to migrate into the T cell zones of lymphoid organs, where stimulation of antigen-specific CD4+ and CD8+ T cells takes place. Studies have shown that there are numerous phenotypic as well as functional differences that distinguish these two dendritic cell subsets. The most notable was the different lineages that give rise to each subset (myeloid vs. lymphoid). This can be observed by the differential array of cell surface molecules and cytokines that each subset expresses. DC1s are CD11b+ and CD11c+, whereas DC2s are CD11b− and CD11c− (Pulendran et al., 2001a). Furthermore, DC1s express high levels of IL-12 and promote Th1 responses, while DC2s express high levels of IFN-α and promote Th2 responses (Pulendran et al., 2001a). DC1s appear to be better stimulators of CD4+ and CD8+ T cells than DC2s. Interestingly, the dendritic cells observed by Makino et al. (1999) to have hyperstimulatory activity towards autologous CD4+ and CD8+ T cells in vitro were monocyte-derived dendritic cells from patients with HAM/TSP. Furthermore, the Th1-type cytokines IL-1β, interferon-γ (IFN-γ), and TNF-α were overexpressed in asymptomatic carriers and patients with HAM/TSP, while the Th2/Th3-type cytokine TGF-β was overexpressed in patients with ATL (Tendler et al., 1991). Based on these findings and the fact that a Th1 type response predominates during HAM/TSP, while a Th2/Th3 type response predominates during ATL, we propose that DC1s (including monocyte-derived dendritic cells) may be preferentially infected by HTLV-I in patients that progress to HAM/TSP, while DC2s may be preferentially infected by HTLV-I in patients that develop ATL. Several events may lead to stimulation of a Th1 or a Th2/Th3 T cell response besides the subset of dendritic cells that are first encountered by antigen, including the nature of the pathogen, pattern recognition receptors expressed on the detecting dendritic cell population, the cytokine microenvironment, and site of antigen exposure (Akbari et al., 2001; Pulendran et al., 2001b).
Since it has been proposed that a majority of patients with HAM/TSP have been exposed to HTLV-I via the PB route, one might expect that DC2s (found in the PB) would be preferentially infected over DC1s (found in the skin and mucosal surfaces). However, infection of PB monocytes and subsequent differentiation into DC1s, may provide a mechanism for the induction of the strong Th1 type immune response observed during HAM/TSP. Furthermore, it has been proposed that what determines the type of T cell response induced by dendritic cells depends not so much on the ontogeny of the antigen presenting dendritic cell, but on the nature of the dendritic cell activating stimulus (Grabbe et al., 2000). Whether the initial route of infection results in the preferential infection of one particular subset of dendritic cells over another in vivo, and the consequences of such an event remain unknown.
Dendritic cell progenitors residing in the BM colonize tissues via the PB as non-dividing dendritic cell precursors. The abundance of immature dendritic cells in the peripheral tissues (e.g., skin and mucosa) has suggested that these cells may be the primary target of HTLV-I infection when the virus is introduced through the mucosal route (Banchereau and Steinman, 1998; Grabbe et al., 2000; Liu et al., 2001). Since dendritic cells are rarely found in the PB (< 1% of PBMCs), CD4+ and CD8+ T cells as well as monocytes may be the primary targets of HTLV-I infection when introduced through the PB route. Therefore, differential target cell selection by HTLV-I as a result of the site of primary infection may have an effect on the efficiency of viral gene expression, viral transmission, and subsequent HTLV-I-specific immune response. This, compounded by host, environmental, and viral factors, may determine which individuals will remain asymptomatic, and which will progress to HAM/TSP or ATL (Fig. 6).
HTLV-I may enter the CNS during primary HTLV-I infection, but this likely occurs without noticeable pathogenic consequences that lead to clinical signs or symptoms. p12I, based on its role in downregulating MHC class I expression, may be an important factor in regulating the spread of HTLV-I within the thymus, LN, spleen, and CNS as well as the PB without triggering an HTLV-I-specific CTL response and promoting the escape of a small number of HTLV-I-infected cells from immune surveillance during primary infection and clinical latency. Following the initial stages of HTLV-I infection, which primarily involves invasion of the PB and secondary lymphoid organs, a long and variable period of clinical latency follows, which may last from several months to more than a decade. This may be described as a period when the host immune response is quite effective in limiting the number of proviral DNA+, viral RNA+ cells within the PB (Manns et al., 1999). However, the presence of proviral DNA+ cells within the CNS during primary infection or clinical latency has not been examined, probably due to the difficulty in detecting the low number of infected cells likely to be present within these tissues and the limited availability of these tissues for experimentation. Further examination with more sensitive detection methods will be required to determine when HTLV-I first enters the CNS during the early stages of infection. As a consequence of the strong CD8+ CTL and antibody responses directed against viral antigens during primary infection, infected cell populations that can either maintain the proviral genome with minimal viral gene expression, or actively subvert the immune response via the actions of p12I, may be able to escape immune surveillance and persist throughout clinical latency. The development of an animal model that can accurately mimic the pathogenic events observed during human infection will be of critical importance in resolving the molecular events associated with early viral infection.
PROGRESSIVE VIRAL INVASION OF THE BONE MARROW
During the long and variable period of clinical latency, when the host immune response ablates the spread of HTLV-I limited to the PB, LN, and spleen, a number of HTLV-I-specific pathogenic processes may occur which set the stage for the overt neurologic dysfunction associated with HAM/TSP. Based on the normal trafficking of CD4+ and CD8+ T cells between the PB and lymphoid tissues, including the LN, spleen, thymus, and BM, HTLV-I-infected T cells may enter the BM which may lead to a progressive viral invasion of the hematopoietic stem cell system over the course of infection. The ability of T cells to traffic to the BM would likely be mediated by expression of the chemokine SDF-1 by BM stromal cells and its chemoattractive effects on CXCR4-expressing T cells (Fig. 7). SDF-1 is produced by BM and thymic stromal cells as well as stromal cells within secondary lymphoid organs, which allows CD34+ progenitors to migrate to the BM, and T cells to traffic between the PB and primary and secondary lymphoid organs (Aiuti et al., 1997; Mohle et al., 1998; Moser and Loetscher, 2001). Migration of T cells into the BM is also mediated by the cell surface adhesion molecules α4β1 and LFA-1 binding to VCAM-1 expressed constitutively on the BM microvascular endothelial cell lining (Berlin-Rufenach et al., 1999). In support of the theory that virus-infected CD4+ and CD8+ T cells traffic into the BM and infect resident cells, PCR-in situ hybridization analyses have suggested that BM from HAM/TSP patients contains large numbers of cells that are HTLV-I proviral DNA+, but viral RNA− (Levin et al., 1997b). This latently infected cell population likely includes CD34+ progenitor cells, which morphologically resemble large lymphocytes. Additional analyses have demonstrated that CD34+ hematopoietic progenitor cells are susceptible to HTLV-I infection in vitro, and that the proviral genome is maintained after in vitro and in vivo differentiation into multiple hematopoietic lineages, including cells which ultimately differentiate into T lymphocytes, B lymphocytes, dendritic cells and cells of the monocyte/macrophage lineage present in PB, LN, and CNS (Feuer et al., 1996). Furthermore, recent observations utilizing the experimentally (intravenously) infected squirrel monkey model demonstrated that early after infection, HTLV-I proviral DNA was present in the BM as well as in PBMCs, and the presence of proviral DNA within these two compartments persisted for the duration of the experiment (at least 26 months) (Kazanji et al., 2000; Mortreux et al., 2001). Cumulatively, these results strongly suggest that the interaction between HTLV-I-infected T cells and CD34+ progenitor cells in the BM compartment during the course of clinical latency and HAM/TSP leads to extensive viral invasion into the hematopoietic progenitor cell population. These cells will eventually initiate their specific differentiation program as they migrate out of the BM and enter the PB. Once in the PB, latently infected cells would disseminate into their respective anatomical niches. Furthermore, some CD34+ progenitor cells will give rise to common lymphoid progenitors, which will in turn give rise to cells that will differentiate into mature, antigen-specific CD4+ and CD8+ T cells. These cells will then have the ability to circulate between the PB and lymphoid organs, including the BM, ensuring that viral infection of the hematopoietic progenitor cell population progresses during the course of clinical latency and symptomatic disease.
The most pathologic consequence resulting from viral invasion and establishment of a latent infection in the BM is the persistent seeding of the periphery with HTLV-I-infected cells. CD34+ hematopoietic progenitors, while present in the BM as uncommitted, cycling progenitor cells, remain silent with respect to viral gene expression (Levin et al., 1997b). Recent observations have suggested a possible mechanism for this observation. EMS analyses utilizing nuclear extracts derived from BM hematopoietic progenitor cells from healthy donors, and radiolabeled double-stranded oligonucleotide probes corresponding to each of the TRE-1 21 bp repeats, showed no detectable levels of DNA-protein complex formation at the mobilities characteristic for ATF/CREB, AP-1, or Sp1/Sp3. This observation is in striking contrast to that observed when utilizing monocyte or lymphocyte nuclear extracts, where distinct super-shifted complexes corresponding to ATF/CREB, AP-1, and Sp1/Sp3 were detected (Grant and Wigdahl, unpublished observations). These results suggested that in the absence of viral gene expression, this latently infected stem cell population would be able to escape immune surveillance while continuing to seed the periphery with HTLV-I-infected cells. As a result, escape from immune surveillance by HTLV-I-infected BM progenitor cells would not be due to the immunologically privileged nature of this cell population within the BM, or a lack of MHC class I expression. Rather, escape would simply be based on the lack of active viral gene expression. A recent report has shown that the overexpression of foreign genes by CD34+ BM cells transduced with a retroviral vector expressing enhanced green fluorescent protein (eGFP) triggered a strong CTL and antibody response in rhesus macaques which was specific for various epitopes of eGFP. This was followed by deletion of eGFP-expressing cells in the BM (Rosenzweig et al., 2001). The absence of viral gene expression within the BM provides a plausible mechanism of escape from immune surveillance, allowing the establishment of a latent reservoir of HTLV-I-infected hematopoietic progenitor cells, with profound consequences relevant to the outcome of HTLV-I infection. Viral gene expression can be observed in proviral DNA+ cells in the PB, suggesting that viral gene expression was initiated at some point during differentiation and migration of CD34+ progenitor cells from the BM into the PB. These observations provide a mechanism linking viral latency and escape from immune surveillance activity to the differentiation status of the infected cell.
To further emphasize the importance of a latent infection within the BM, recent evidence has shown that clearance of the latently infected BM progenitor cell population has a profound impact on the outcome of HTLV-I infection. Previous studies by Kawa et al. (1998) reported the complete eradication of HTLV-I infection in an asymptomatic carrier by myelo-lymphoid-ablative chemotherapy followed by allogeneic BM transplantation. The patient described in this report was a 15-year-old boy with congenital red cell anemia, initially exposed to HTLV-I as a consequence of several blood transfusions given by the father who was later determined to be HTLV-I+ after developing neurologic symptoms indicative of HAM/TSP (Kawa et al., 1998). Although the pediatric patient showed no signs of HTLV-I-associated neurologic disease, there was evidence of proviral DNA in the PB and BM before transplantation. Based on this evidence, it is likely that this patient would have eventually progressed to symptomatic HAM/TSP within a few years. The patient received myelo-lymphoid-ablative chemotherapy followed by a BM transplant from a sister who was HTLV-I seronegative. Gradually, all signs of proviral DNA in the PB and BM disappeared within 320 days post-transplant, and remained HTLV-I negative for at least 60 months (Kawa et al., 1998). These results suggest that novel therapeutic strategies that clear the BM and PB of infected cells or perhaps prevent the initiation of viral gene expression in differentiating BM progenitor cells and target cells within the PB, secondary lymphoid organs, and CNS may prove to be effective in treating HAM/TSP. Furthermore, these observations suggest that viral invasion of the CNS may not occur during primary infection.
BM HEMATOPOIETIC PROGENITOR CELL DIFFERENTIATION AND ACTIVATION OF VIRAL GENE EXPRESSION
The ability of HTLV-I-infected CD34+ hematopoietic progenitor cells to maintain intact proviral DNA after differentiating down several pathways, the presence of proviral DNA+ cells within the BM of patients with HAM/TSP but only certain asymptomatic carriers, and the elevated numbers of HTLV-I-infected monocytes, dendritic cells, CD4+ and CD8+ T cells, and B cells detected in the PB of HAM/TSP patients compared to asymptomatic carriers, suggests that viral invasion of the BM is established at some point during the course of clinical latency, a process which likely progresses with duration of infection (Kira et al., 1991; Levin et al., 1997b) (Fig. 1). This is followed by the differentiation and migration of latently infected CD34+ progenitor cells into the PB, resulting in an elevated proviral load and an heightened HTLV-I-specific immune response associated with the pathogenesis of HAM/TSP (Levin et al., 1997b).
As an HTLV-I-infected individual progresses through clinical latency to HAM/TSP, the increase in proviral load in the PB may be caused by a number of different mechanisms: amplification of virus-infected T cells through clonal expansion (mitotic HTLV-I amplification); cell–cell contact between an infected and uninfected cell (infectious HTLV-I amplification); and seeding of the PB with latently infected CD34+ progenitor cell derivatives (progenitor HTLV-I amplification) (Levin et al., 1997b; Wodarz et al., 1999) (Fig. 8). Clinical latency is characterized as a state that involves an effective host immune response that results in ablation of viral activity and spread. This takes into consideration both mitotic HTLV-I amplification and infectious HTLV-I amplification of HTLV-I within the PB. However, during the later stages of clinical latency, when the BM has become saturated with HTLV-I-infected proviral DNA+ viral RNA− progenitor cells, the differentiation and migration of the HTLV-I+ hematopoietic progenitor cells into the PB may lead to an increase in proviral DNA load, viral gene expression, and trigger an enhanced, persistent, and increasingly detrimental immune response from that point forward (Fig. 1). The overall level of proviral DNA in each cellular compartment may be dependent on the impact of HTLV-I on the development of a given cellular compartment. The CD34+ hematopoietic progenitor cell population likely represents an abundant and renewable source of HTLV-I-infected cells including CD4+ and CD8+ T cells, monocytes/macrophages, microglial cells, dendritic cells, and B cells. These latently infected CD34+ progenitors are able to escape from immune surveillance activity, while continuing to seed the periphery with infected cells. Therefore, regardless of the intensity of the HTLV-I-specific immune response, it is highly unlikely that complete clearance of virus-infected cells by normal host defense mechanisms will be achieved.
Although CD34+ progenitor cells are actively cycling and transcriptionally active (as shown by NF-κB nuclear localization and DNA binding activity), the proviral genome is likely to remain in a transcriptionally inactive state (as shown by the absence of ATF/CREB, AP-1, and Sp1/Sp3 DNA binding activity) (Grant and Wigdahl, unpublished observations). However, viral gene expression can be detected in monocytes, as well as CD4+ and CD8+ T cells in the PB (Koyanagi et al., 1993; Cho et al., 1995). Viral gene expression in dendritic cells from HTLV-I-infected patients has not been reported, although it is likely that the level of viral gene expression is similar to that observed in PB monocytes. This implies that viral gene expression is turned on at some point during the differentiation program as the infected progenitor cells migrate from the BM into the PB. During this time, activation of developmentally regulated transcription factors capable of inducing basal activation of the HTLV-I LTR takes place allowing for a low level of viral gene expression. This leads to the accumulation of Tax in the nucleus, and Tax-mediated trans-activation of viral gene expression.
Since the expression of many of the cellular transcription factors that interact with the LTR are induced or repressed during the activation and differentiation processes, the levels of basal and Tax-mediated transcriptional activation of the viral promoter are likely to be cell type- and differentiation stage-specific. During the process of developmental commitment and subsequent differentiation, external factors initiate signaling cascades that regulate the availability of a variety of transcription factor families. These transcription factors act in concert with cis-acting promoter elements to determine the expression of cell type-specific genes. In monocytes, for example, transcription factors such as PU.1, AP-1, C/EBP, c-Myb, Sp1, AML1, NF-Y, and several others have been shown to either regulate the expression of differentiation-specific genes or directly contribute to the process of cellular differentiation (Valledor et al., 1998). For instance, c-Myb, which has been shown to activate basal transcription and inhibit Tax-mediated trans-activation of the LTR, becomes downregulated during monocytic differentiation (Lee et al., 1987; Colgin and Nyborg, 1998). However, this gene has been shown to be upregulated during T cell activation through the IL-2R signaling pathway (Bosselut et al., 1992). The Fos and Jun proteins that form the AP-1 transcriptional complex on the TRE-1 pc repeat, are cell type-specific, and are upregulated in a PKC-dependent manner during monocytic differentiation and T cell activation. In CD4+ T cells, AP-1 was predominantly composed of JunD, while in monocytes, AP-1 was composed of Fra-1, Fra-2, JunB, and JunD, and in astrocytes, AP-1 is composed of c-Jun and c-Fos (Mollinedo et al., 1993; Fujii et al., 1995; Wessner et al., 1997; Iwai et al., 2001; Grant and Wigdahl, unpublished observations). Sp1 has been shown to be a transcriptional activator, while Sp3 has been shown to be a repressor of transcriptional activation (reviewed in Black et al., 2001). During monocytic differentiation, the Sp1 activator/Sp3 repressor ratio increases, allowing Sp1 to preferentially interact with responsive cellular promoters and the LTR (MacAlister and Wigdahl, unpublished observations). An interesting feature of both AP-1 and Sp1/Sp3 is their ability to directly compete with ATF/CREB factors for binding to the TRE-1 pc and pp repeats, respectively (Sassone-Corsi et al., 1990; Masquilier and Sassone-Corsi, 1992; Yao and Wigdahl, unpublished observations). Therefore, viral gene expression may be regulated in such a way as to promote a low level of viral gene expression during monocytic differentiation, while being enhanced during T cell activation. Cell type- and differentiation stage-specific expression of transcription factors like c-Myb, AP-1, and Sp1/Sp3, may dramatically influence LTR-mediated viral gene expression during the differentiation of CD34+ progenitor cell in the BM, PB, and CNS. Differentiation stage-specific regulation of viral gene expression within cells of the monocyte/macrophage lineage is suggested by the induction or repression of several key transcription factors that are able to interact with the LTR (Fig. 9). A further level of complexity is added by the increasing number of reports demonstrating Tax-mediated trans-activation of cellular transcription factors such as ATF-4 (CREB-2), c-Fos, Fra-1, c-Jun, JunD, c-Rel, and YB-1 that regulate proliferation and differentiation of a number of cell types (Valledor et al., 1998; Harhaj et al., 1999; Neumann et al., 2000), suggesting the possibility that HTLV-I infection in differentiating progenitor cells may affect the developmental program of a number of cell lineages that rely on precise temporal regulation of these factors. This possibility is supported by the phenotypic and functional abnormalities observed in HTLV-I-infected T lymphocyte progenitors and PB monocytes from asymptomatic carriers or patients with HAM/TSP (discussed below).
Monocytes are a very dynamic cell lineage that retain the ability to differentiate into macrophages, dendritic cells, or microglial cells after transmigration from the PB into tissues (Randolph et al., 1998; Santambrogio et al., 2001). PB monocytes from patients with HAM/TSP exhibited no phenotypic abnormalities in their ability to differentiate into dendritic cells in vitro, compared to PB monocytes from healthy controls (Makino et al., 1999). However, HTLV-I-infected monocyte-derived dendritic cells did exhibit an enhanced capacity to stimulate autologous, naïve CD4+ and CD8+ T cells in vitro (Makino et al., 1999). In striking contrast, PB monocytes isolated from patients with ATL could not properly differentiate into dendritic cells in vitro, as shown by reduced MHC class I, CD86, CD14, and CD1a expression, decreased uptake of exogenous antigen, and decreased ability to stimulate autologous CD4+ and CD8+ T cells, compared to dendritic cells differentiated in vitro from PB monocytes isolated from healthy controls (Makino et al., 2000). Although HTLV-I-infected monocyte-derived dendritic cells could not infect CD4+ or CD8+ T cells during in vitro co-culture, viral invasion into this antigen presenting cell population, and resulting hyperstimulatory activity displayed by infected monocyte-derived dendritic cells in HAM/TSP patients suggests that these events may be a major contributor to the initiation and maintenance of the Tax-specific CD8+ effector/memory T cell population observed in the PB and CNS of patients with HAM/TSP (Makino et al., 1999; Nagai et al., 2001a).
BM and PB T cell colony-forming cells (T-CFCs, pre-thymic T lymphocyte progenitors) from asymptomatic carriers and HAM/TSP patients are able to spontaneously proliferate in vitro in the absence of exogenous growth factors such as phytohemaglutinin (PHA) and interleukin 2 (IL-2), whereas proliferation of T-CFCs from healthy control patients requires the presence of these growth factors (Lunardi-Iskandar et al., 1993). Interestingly, in addition to IL-2, autologous, but not allogeneic adherent cells were required for spontaneous proliferation of T-CFCs in vitro, suggesting that in HTLV-I-infected individuals, the hyperstimulatory activity reported by Makino et al. (1999) in monocyte-derived dendritic cells from HAM/TSP patients can promote spontaneous proliferation of HTLV-I-infected T cells. In addition to the abnormal proliferative properties, HTLV-I-infected T-CFCs also exhibited abnormal differentiation properties in vitro (Lunardi-Iskandar et al., 1993). Direct infection of the T-CFC population may occur through PB exposure to virus or infected cells, or differentiation of infected BM progenitor cells (Lunardi-Iskandar et al., 1993). Pre-thymic T lymphocyte progenitors in the PB could carry the virus into the thymus, thereby infecting the immature, uninfected T cell population, or possibly infecting thymic stromal cells or thymic dendritic cells, thus having adverse affects on both the positive and negative selection processes. Infected T cells may also infect the dendritic cell and B cell populations within the LN and spleen. Entry of virus-infected cells into the primary and secondary lymphoid organs, whether latent or productive, may represent one of the most effective mechanisms of viral transmission to susceptible target cells studied to date, as these regions bring several cell types into close proximity for extended periods of time, ensuring progressive spread of virus.
It is not clear at this time whether HTLV-I infection of the BM progenitor cell population leads to an altered differentiation pattern down any particular lineage through the activation or repression of cellular genes that regulate lineage commitment, resulting in elevated or reduced cell numbers in the periphery. However, it does appear that HTLV-I infection within the monocytic and T lymphocytic lineages has a profound effect on the phenotypic and functional characteristics of these differentiated cell types.
TRANSITION FROM NORMAL IMMUNE SURVEILLANCE TO NEUROINFLAMMATION
As a result of the pathogenic processes described in the preceding sections, clinical latency may culminate in an extensive latent HTLV-I infection of the BM progenitor cell population, which could subsequently lead to infiltration of HTLV-I-infected cells into the PB, CNS, and possibly other target organs (Fig. 7). The mechanism of progenitor HTLV-I amplification, coupled with the steady-state mitotic HTLV-I amplification and infectious HTLV-I amplification processes already taking place, may cause the proviral DNA load and viral gene expression to increase, resulting in the generation of an intense HTLV-I-specific immune response. The equilibrium proviral DNA load during clinical latency is primarily determined by the effectiveness of the host immune response (Wodarz et al., 1999). Ineffective immune responses against HTLV-I fail to efficiently control the expansion of virus-infected cells, eventually leading to viral invasion of the CNS and HAM/TSP. Alternatively, effective immune responses against HTLV-I are able to efficiently control the expansion of virus-infected cells, possibly preventing viral invasion of the CNS and the onset of neurologic disease. Furthermore, effective immune responses by long-term asymptomatic carriers may be able to keep the expansion of HTLV-I+ CD4+ and CD8+ T cells low enough to prevent viral infection of the BM. Thus, the effectiveness of the immune response against HTLV-I may in fact be a critical factor preventing the spread of viral infection to the BM and CNS, separating individuals that remain long-term asymptomatic and individuals that develop neurologic disease (Wodarz et al., 1999).
Host genetic factors, the most well studied being human leukocyte antigen (HLA) class I haplotype, are believed to make a critical difference in determining whether an individual will develop an effective immune response to HTLV-I infection. The effectiveness of the HTLV-I-specific immune response, especially the CD8+ CTL response has been shown to be key to controlling the proviral DNA load, a major risk factor for HAM/TSP. Previous observations have indicated that the expression of HLA-A*0201, HLA-Cw*08 alleles was associated with a lower proviral DNA load and a lower risk of developing HAM/TSP, while expression of HLA-B*5401 was associated with higher proviral DNA load and an increased risk of developing HAM/TSP in HTLV-I-infected patients (Jeffery et al., 1999, 2000). Immunodominant Tax peptides, such as Tax (11–19), are believed to make strong, stable interactions with the peptide binding groove of HLA-A*0201 and HLA-Cw*08 molecules (Garboczi et al., 1996; Ding et al., 1999). Stable interactions facilitate the expression of peptide-MHC class I molecules on the surface of an infected cell and recognition by antigen-specific CD8+ CTLs. The stronger the interaction between peptide and MHC molecule, the longer the peptide will remain bound in the peptide binding groove, which may allow for efficient immune surveillance and targeting of HTLV-I-infected cells. Several HLA class I alleles have been shown to predispose HTLV-I-infected patients to develop ATL, including HLA-A*26, HLA-B*4002, HLA-B*4006, and HLA-B*4801 alleles (Yashiki et al., 2001). These MHC class I alleles do not form stable complexes with immunodominant Tax peptides, and thus were not able to stimulate strong Tax-specific CD8+ CTL responses. These observations suggest that the relative affinity of immunodominant epitopes of viral proteins for the HLA peptide binding groove acts as a critical determinant with respect to controlling the proviral DNA load and viral gene expression. A common theme therefore emerges between HAM/TSP and ATL concerning the efficiency of the immune response with respect to controlling proviral DNA load. Although the efficiency with which the immune response controls the proviral DNA load may be largely determined by HLA class I expression, other host, environmental, or viral factors that manifest during primary infection likely play a decisive role in the differential pathogenesis of these two important HTLV-I-associated disorders. With respect to HAM/TSP pathogenesis, disruption of the equilibrium between proviral DNA load and immune response by the introduction of an additional source of virus-infected cells into the periphery could lead to a gradual increase in proviral DNA load and viral gene expression, resulting in the generation of an increasingly aggressive and detrimental Tax-specific CD8+ CTL response.
The ability of activated T cells to traffic into the CNS across the blood-brain barrier is exacerbated by the heightened attachment and trans-migration activity of HTLV-I-infected CD4+ T cells observed in patients with HAM/TSP (Hickey et al., 1991; Furuya et al., 1997). A recent study has shown that close contact between brain endothelial cells and HTLV-I-infected CD4+ T cells resulted in increased T cell production of TNF-α accompanied by massive and rapid budding of viral particles (Romero et al., 2000). Similar events likely govern the entry of infected CD8+ T cells into the CNS. Overexpression of TNF-α is thought to increase the paracellular permeability of the brain endothelial monolayer and facilitate trans-migration of activated T cells (as well as monocytes) across the blood-brain barrier, while the viral particles were shown to be taken up into vesicles by brain endothelial cells and apparently released from the basolateral surface (Romero et al., 2000). In addition, HTLV-I-infected T cells were also observed fusing to brain endothelial cells (Romero et al., 2000). Therefore, there appears to be at least three mechanisms that allow for HTLV-I entry into the CNS: paracellular trans-migration of HTLV-I-infected cells through the blood-brain barrier; absorption and release of viral particles from vesicles within brain endothelial cells; and direct infection of brain endothelial cells. The first of these three processes is probably facilitated by the overexpression of α4β1 and α5β1 integrins, as well as matrix metalloproteinase (MMP)-2 and MMP-9 in HTLV-I-infected T cells from HAM/TSP patients (Dhawan et al., 1993; Umehara et al., 1998).
Once in the CNS, HTLV-I+ CD4+ and CD8+ T cells may take part in a number of events leading to viral infection of resident CNS cell populations, activation of astrocytes and microglial cells, induction of proinflammatory cytokine and chemokine synthesis, recruitment of inflammatory infiltrates into the CNS, blood-brainbarrier disruption, dysregulation of oligodendrocyte homeostasis, demyelination, and axonal degradation (Fig. 10). The blood-brain barrier is surrounded by processes extending from the body of astrocytes, suggesting that these cells will come into contact with cells infiltrating into the CNS, and may become infected with HTLV-I during this trafficking process. This has been supported by several studies indicating that astrocytes can be infected with HTLV-I in vitro and in vivo (Lehky et al., 1995; Szymocha et al., 2000b). Astrocytes are believed to act as weak antigen presenting cells, capable of inducing T cell apoptosis, suppressing T cell and microglial cell activation, and providing a structural scaffold for neurons (Xiao and Link, 1999). In addition, astrocytes are a major source of nerve growth factor and protect neurons from toxic injury by taking up excess glutamate from the local environment (Rothstein et al., 1993; Xiao and Link, 1999). HTLV-I Tax expression in infected T cells or infected astrocytes in vitro, results in a corresponding increase in proinflammatory cytokine production by astrocytes, including TNF-α, IL-1α, and IL-12 (Szymocha et al., 2000a,b). MMP-2 and MMP-9 expression was also upregulated in astrocytes by Tax (Szymocha et al., 2000b). Infection in this cell population may lead to alterations in the structural integrity of the blood–brain-barrier, decreased uptake of extracellular glutamate, and increased infiltration of virus infected cells into the CNS and neurotoxicity.
Microglial cells represent an abundant and unique cell population in the CNS that is also susceptible to HTLV-I infection (Hoffman et al., 1992). These cells are derived from BM progenitors of myeloid lineage. In adults, two subsets of microglial cells exist in different regions of the CNS. Perivascular microglial cells are continually replenished from BM-derived PB monocytes and remain closely associated with the blood-brain barrier, whereas parenchymal microglial cells (resting microglial cells) are a relatively stable population that is not replenished by PB monocytes, but remain in an immature state located throughout the interior of the CNS (Santambrogio et al., 2001). This subset of microglial cells may be induced to further differentiate within the CNS and acquire some of the characteristics of immature dendritic cells or macrophages in the presence of GM-CSF and M-CSF, respectively (Santambrogio et al., 2001). These cells not only play a role in maintaining the structural integrity of the CNS, but also as key regulators of the immune response by acting as local antigen presenting cells and by secreting a vast array of proinflammatory and chemoattractant cytokines. PB monocytes may also be able to differentiate into perivascular dendritic cells after trafficking across the blood-brain barrier (Randolph et al., 1998; Aloisi et al., 2000; Santambrogio et al., 2001).
One of the most controversial issues concerning the pathogenesis of neurodegenerative disorders of autoimmune or viral origin concerns the initiation of the immune response directed against CNS-associated antigens. The CNS is believed to be a relatively immunologically privileged site, maintained by the endothelial monolayer and basement membrane of the blood–brain-barrier, the absence of lymphatic vessels and dendritic cells, and the presence of an immunosuppressive microenvironment (Aloisi et al., 2000). The integrity of the blood-brain barrier prohibits the entry of naïve T cells into the CNS (Wekerle et al., 1987; Hickey et al., 1991). In the absence of dendritic cells, the role of the antigen presenting cell falls upon microglial cells and astrocytes. Activation of these cells, however, does not result in their migration out of the CNS towards the draining LN or spleen. Rather, these cells are able to secrete a wide array of proinflammatory cytokines and chemokines that promote the recruitment of antigen-specific CD4+ and CD8+ T cells from the circulation that have been activated in the secondary lymphoid organs (Aloisi et al., 2000). Therefore, CNS-associated antigens have to be able to gain access to the LN or spleen in order to initiate a T cell response. This may occur by any of a number of mechanisms, such as antigen coming into contact with monocyte-derived dendritic cells localized to the perivascular areas and meninges and stroma of the choroid plexus (Aloisi et al., 2000). These cells could encounter antigen, migrate out of the CNS and into the cervical LN or spleen and stimulate an antigen-specific immune response. A second and more likely possibility with respect to HTLV-I infection is that viral gene expression within the dendritic cell compartment found in PB, and especially the LN and spleen, stimulate the activation of naïve and reactivation of memory HTLV-I-specific CD4+ and CD8+ T cells. Furthermore, given the hyperstimulatory activity of infected dendritic cells in HAM/TSP patients, the likelihood of coincidental activation of autoreactive or cross-reactive CD4+ and CD8+ T cells becomes an increasing risk. Support for the coincidental activation of CD4+ and CD8+ T cells specific for self-antigens or specific for viral antigens that cross-react with self-antigens comes from the wide variety of autoimmune-like HTLV-I-associated disorders that have been recently described (reviewed in Uchiyama, 1997). In light of recent evidence demonstrating the presence of a viral reservoir within the CD8+ T cell population, including Tax-specific CD8+ CTLs (Nagai et al., 2001b), the possibility emerges that traditional T cell-dendritic cell interactions within the secondary lymphoid organs may not be necessary to initiate CTL activation and proliferation in vivo. Support for this hypothesis comes from the observation that CD8+ T cells in vitro preferentially expand during spontaneous proliferation of PBMCs from HAM/TSP patients (Sakai et al., 2001). Furthermore, HTLV-I infection of anergic autoreactive T cell populations may reactivate these cells and lead to the induction of autoimmunity. Trafficking of these HTLV-I infected autoreactive T cells to sites of inflammation may allow viral invasion into regions where that specific self antigen is expressed, such as within the synovial membranes as in the case of HTLV-I-associated arthropathy.
Failure of the immune response to control the overall proviral DNA load and subsequent viral invasion of the BM, followed by seeding of infected dendritic cells into the PB, LN, and spleen, may lead to an enhanced and persistent stimulation of both naïve and memory HTLV-I-specific CD8+ CTLs. In addition, HTLV-I+ monocytes and CD4+ T cells could traffic into the CNS resulting in infection of resident cells, providing additional targets for the HTLV-I-specific CTLs (Fig. 7). This process would ultimately result in a continuous expansion in the size and efficiency of a Tax-specific CD8+ CTL effector/memory compartment during clinical latency and HAM/TSP (Nagai et al., 2001a). Studies have shown that during natural infection, a significant number of CD8+ T cells become infected, suggesting that HTLV-I-specific CTLs could target themselves for destruction, leading to an impaired antiviral CTL response (Hanon et al., 2000; Nagai et al., 2001b; Sakai et al., 2001).
In essence, throughout clinical latency, multiple rounds of HTLV-I activation and immune surveillance by an ever increasingly effective CTL compartment could ultimately lead to a chronic state of neuroinflammation. Clearly, one of the key regulators of clinical latency centers on the regulation of viral gene expression in HTLV-I-infected BM progenitor cells as they differentiate into dendritic cells and monocytes, macrophages and microglial cells present in the PB, LN, spleen, and CNS. HTLV-I infection and viral gene expression within the dendritic cell population may be one of the critical features of HTLV-I infection that leads to the high frequency of Tax-specific CD8+ CTLs detected in the PB and cerebrospinal fluid (CSF) in patients with HAM/TSP (Greten et al., 1998). Furthermore, the proportion of HTLV-I+ dendritic cells present in the PB and secondary lymphoid organs may be proportional to the total proviral DNA load in the PB, providing a mechanism behind the correlation of proviral DNA load and the frequency of effector/memory Tax-specific CD8+ T cell response (Nagai et al., 2001a). Alternatively, expansion of HTLV-I+ Tax-specific CD8+ CTLs may directly contribute to the proviral DNA load. HTLV-I infection and viral gene expression within the monocytic lineage may become important in enhancing the ability of these cells to enter the CNS by trans-migrating across the blood-brain barrier. Although adhesion molecule and chemokine receptor expression has not been examined in HTLV-I-infected monocytes, the chemokine macrophage inflammatory protein-1α (MIP-1α) produced by HTLV-I-infected CD4+ T cells, acts as a potent activation and migration factor for PB monocytes, and a potent migration factor for activated T cells (Bertini et al., 1995; Baggiolini, 1998; Kaufmann et al., 2001; Weber et al., 2001). Furthermore, PB monocytes express CCR2, the receptor for the chemokine monocyte chemotactic protein-1 (MCP-1), expressed by activated macrophages (Fantuzzi et al., 1999). These two chemokine–chemokine receptor interactions may facilitate the trafficking of activated T cells and monocytes to site of inflammation within the CNS (Fig. 7). Within the CNS, monocyte-derived macrophages or microglial cells may play a critical role in the initiation of neurotoxic cytokine production as a result of direct infection of these cells with virus, Tax, or a result of HTLV-I+ CD4+ and/or CD8+ T cell communication. Since a number of studies have suggested that infected monocytes and dendritic cells do not contain readily detectable quantities of HTLV-I RNA and protein but do appear to harbor large quantities of proviral DNA (Macatonia et al., 1992; Koyanagi et al., 1993), induction of antibody and CTL responses by dendritic cells or generation of aberrant cytokine production by macrophages or microglial cells based on HTLV-I infection may be accomplished as a consequence of some form of regulated low level gene expression, as discussed below.
HAM/TSP involves the transition from normal immune surveillance to a state of neuroinflammation indicated by events that have taken place in the PB and BM during primary infection and clinical latency. Consequently, infiltration of activated HTLV-I-infected CD4+ and CD8+ T cells across the blood-brain barrier and into the CNS is driven by the expansion of proviral DNA+ cells and egress of infected progenitor cells from the BM (Figs. 8 and 10). As a result, further infection of astrocytes and microglial cells may take place. Meanwhile, events in the PB and secondary lymphoid organs, including migration of dendritic cells to the T cell zones of the LN and spleen, results in the reactivation of a highly effective CD8+ CTL memory compartment and newly induced effector CD8+ T cells, and their subsequent migration throughout the PB and CNS. Consequently, HTLV-I genomic activation and regulated low level viral gene expression within infected astrocytes, and monocyte-derived macrophages and microglial cells within the CNS, neurotoxic cytokine production coupled with CTL-mediated lysis of infected cells may precipitate neuronal degeneration and dysregulation of myelin-producing cells and other cell populations within the CNS (Selmaj and Raine, 1988; Rothstein et al., 1993; Umehara et al., 1993, 1994b, 1996, 1998; Bertini et al., 1995; Cook et al., 1995; Jacobson, 1996; Levin et al., 1997a; Mendez et al., 1997; Greten et al., 1998; Abe et al., 1999; Wakamatsu et al., 1999; Nagai et al., 2001a) (Fig. 10).
Additional damage may also be caused by the recognition and disruption of uninfected neurons by cross-reactive anti-HTLV-I antibodies (Levin et al., 1998). CNS dysfunction may also be caused directly or indirectly by HTLV-I-induced dysregulation of astrocytic, macrophage, microglial, or neuronal function mediated by the multifunctional Tax protein. Recent studies have shown that Tax may be released from an infected cell and act as an extracellular cytokine on neighboring cells in the CNS. Extracellular Tax has been shown to induce NF-κB nuclear localization, and the expression of the immunoglobulin κ light chain, IL-2Rα, IL-1β, IL-6, TNF-α, and TNF-β (Lindholm et al., 1990; Lindholm et al., 1992; Marriott et al., 1992; Dhib-Jalbut et al., 1994). Possible mechanisms responsible for the cytokine-like effects of Tax may include extracellular Tax inducing a signaling cascade by binding to a specific cell surface receptor, or by the internalization and nuclear localization of extracellular Tax. Possible cytotoxic activities of extracellular Tax on neurons and glial cells remains to be determined. The effects of intracellular Tax on host cell cytokine production has been well documented (Table 1).
Table 1. Cytokine Expression Induced by HTLV-I Tax
EMERGING MODEL OF HTLV-INDUCED NEUROLOGIC DISEASE AND DIRECTIONS FOR FUTURE RESEARCH
HAM/TSP patients have a unique immunologic profile associated with a state of immune stimulation. This is reflected in the high antibody titer to HTLV-I proteins, increased levels of activated T cells, and the oligoclonal expansion of Tax (11–19)-specific CD8+ CTL in the PB and, relevant to CNS disease, in CSF. Additionally, HAM/TSP patients exhibit elevated levels of inflammatory cytokines and an increase in proviral DNA load in both the PB and CSF as compared to ATL patients and asymptomatic carriers (Ohbo et al., 1991; Kira et al., 1992; Kuroda and Matsui, 1993; Kuroda et al., 1993; Nakamura et al., 1993; Umehara et al., 1994a; Fox et al., 1996; Nagai et al., 1998; Furuya et al., 1999). Clearly, the immunologic response to HTLV-I infection in HAM/TSP patients is significantly different from that of ATL patients or asymptomatic carriers. Recent results have indicated that the route of primary infection may determine the type of immune response and, ultimately, the disease pathogenesis associated with HTLV-I infection (Osame et al., 1990; Kawa et al., 1998). In general, primary infection can occur either through the PB or the mucosa. It has been hypothesized that a primary infection based in the PB primarily results in the infection of the CD4+ and CD8+ T cell compartments. Conversely, infection of mucosal surfaces likely leads to the infection of a different subset of target cells, with relatively low numbers of CD4+ and CD8+ T cells being infected. Primary infection of the PB compartment involving the CD4+ and CD8+ T cell populations may result in a higher level of viral gene expression than that found after primary infection of the mucosa. The resulting relative levels of viral gene expression drive the expansion of the HTLV-I immune response. The greater frequency of T cells circulating within the PB compared to the mucosa, suggests that of the population of cells targeted by HTLV-I, a higher level of mitotic and infectious HTLV-I amplification and thus viral gene expression will be possible in the PB. Therefore, PB exposure likely results in the induction of a strong immune response against HTLV-I, most notably, the CD8+ CTL response directed against immunodominant epitopes of Tax.
HAM/TSP patients have an increased proviral DNA load in the PB as compared to ATL patients and asymptomatic carriers (Renjifo et al., 1996; Nagai et al., 1998). Levin et al. (1997) have demonstrated that the BM of HAM/TSP patients contains high levels of proviral DNA. PCR-in situ hybridization using a Tax-specific probe has demonstrated that greater than 95% of BM cells from HAM/TSP patients are proviral DNA positive. However, in situ hybridization for Tax RNA in BM from HAM/TSP patients demonstrates that BM cells are essentially negative for viral RNA (Levin et al., 1997b). These observations have suggested that there is an extensive latent HTLV-I infection of the BM of HAM/TSP patients. As previously described, it is hypothesized that PB infection leads to the infection of predominantly CD4+ and CD8+ T cells. These cell populations routinely traffic into the BM as part of normal immune surveillance. Therefore, the CD34+ progenitor population is likely to become infected by the normal trafficking of HTLV-I-infected T cells into the BM. However, the mechanisms involved in BM progenitor cell infection, including HTLV-I+ T cell trafficking into the BM, the mode of transmission (cell-associated vs. cell-free infection) to BM resident cells, and the initial effects viral transmission has on BM resident cells remain to be ascertained. Morphologically, proviral DNA+ cells within the BM resembled CD34+ hematopoietic progenitors, which may appear as large lymphocytes (Levin et al., 1997b). PCR-in situ hybridization coupled with immunohistochemical analysis will be required to definitively confirm the phenotype of the infected cell population(s) within the BM.
Up to this point, we have described a situation in which an individual becomes infected with HTLV-I through the PB, which leads to the infection primarily of CD4+ and CD8+ T cells. Infected cells can then traffic into the BM, where a latent infection of CD34+ progenitor cells can be established. Results have indicated that BM progenitor cells are capable of harboring integrated HTLV-I proviral DNA (Feuer et al., 1996; Levin et al., 1997b), and it has also been demonstrated that HTLV-I-infected CD34+ progenitor cells maintain their proviral genome throughout the process of differentiation (Feuer et al., 1996). This includes differentiation into primitive progenitors, erythroid progenitors, and myeloid progenitors. Furthermore, there is little or no transcription from the HTLV-I LTR in latently infected BM progenitors, as demonstrated by the lack of HTLV-I-specific RNA or protein in these cells (Levin et al., 1997b). One of the immunologic characteristics of HAM/TSP is a high level of Tax (11–19)-specific CD8+ CTL in both the PB and CSF. The production of this CTL compartment likely occurs by the presentation of the immunodominant Tax peptide in the context of MHC class I on the surface of infected antigen presenting cells. However, infected antigen presenting cells, including macrophages and dendritic cells, are likely to be present in HTLV-I-infected individuals other than those who develop HAM/TSP. This likely represents a mechanism by which Tax (11–19)-specific CD8+ CTLs serve to control HTLV-I infection by mediating the lysis of virus-infected cells. We hypothesize that the Tax (11–19)-specific CD8+ CTL compartment is hyperstimulated in HAM/TSP patients by an infected dendritic cell compartment, thus contributing to the neuropathogenesis associated with the disease.
Latently infected BM progenitor cells express little or no virus-specific RNA or protein. However, these latently infected cells are capable of differentiating into multiple cell types, including antigen presenting cells such as monocytes/macrophages and dendritic cells. We hypothesize that the differentiation of infected CD34+ BM progenitor cells into infected antigen presenting cells may lead to changes in the abundance or activity of cellular transcription factors critical to the regulation of viral gene expression as well as cellular function. The changes in transcription factor expression may then lead to a regulated low level of basal viral gene expression or regulated low level Tax-mediated trans-activation of the viral LTR. This would then allow for the production of viral-specific peptides and antigen presentation of the immunodominant Tax (11–19) epitope in the context of MHC class I on the surface of antigen presenting cells, such as dendritic cells or macrophages, derived from the infected BM progenitor cells. HAM/TSP patients may have a large, latently infected population of CD34+ BM progenitor cells, which are capable of continuously differentiating into infected antigen presenting cells. These antigen presenting cells can then continuously present virus-specific peptides to naïve CD8+ T cells, leading to an overproduction of Tax (11–19)-specific CTL. These CTLs are known to migrate into the CNS, where they mediate a variety of effects contributing to the pathogenesis associated with HAM/TSP, including lysis of target cells, cytokine dysregulation, and neuroinflammation (Greten et al., 1998). The overstimulation of the Tax (11–19)-specific CTL compartment could ultimately lead to a shift from normal immune surveillance involved in clearing the primary infection and maintaining the asymptomatic state, to a highly inflammatory disease that centers on the overproduction of Tax (11–19)-specific CD8+ CTLs and trans-migration of a fraction of these cells into the CNS. Furthermore, entry into the CNS of HTLV-I+ CD8+ T cells, including Tax (11–19)-specific CTLs, may help explain the progressive nature of HAM/TSP in patients with longer duration of disease.
With respect to viral invasion of the BM in HAM/TSP patients and some asymptomatic carriers, the evidence collected to date suggests that viral gene expression has a profound effect on the functional characteristics of the infected cell population. It is almost certain that viral gene expression is regulated in a cell type- and differentiation stage-specific manner, but future studies must now determine at what point along the differentiation pathway viral gene expression within CD34+ progenitor cells may be turned on, and what effects will this have on the differentiation process. Examination of transcriptional regulation of the LTR during the differentiation process may facilitate the design of therapeutic strategies to prevent the initiation of viral gene expression within the PB and CNS. Finally, further examination of the presence of proviral DNA and viral gene expression within the PB and CNS throughout the duration of clinical latency and neurologic disease with respect to cellular phenotype, may provide a better understanding of the disease process and facilitate novel therapeutic strategies to control disease progression.
Great strides have been made in the development of new experimentally infected animal models that exhibit many of the pathological features associated with HAM/TSP and ATL. These and other novel models will need to be exploited in order to fully understand the early events that take place after initial exposure of the host to HTLV-I and the kinetics of viral gene expression with respect to different cell types and stages of development.