Human T-lymphotropic virus type 1 (HTLV-1) infection is associated with the clonal expansion and transformation of mature T lymphocytes. While the mechanisms involved are incompletely understood the viral regulatory protein Tax plays a central role in these processes. Recent studies employing genomic and proteomic approaches have demonstrated the marked complexity of gene deregulation associated with Tax expression and confirmed the remarkable pleiotropism of this protein as evidenced by the numerous Tax–cellular protein interactions in infected cells. In this review, we summarize the role of Tax in the deregulation of selected cellular-signaling pathways. Specifically, this has focused on the influence and interaction of Tax with the AP-1 and NF-AT transcription factors, PDZ domain-containing proteins, Rho-GTPases, and the Janus kinase/signal transducer and activator of transcription and transforming growth factor-β-signaling pathways. In addition to identifying the deregulation of events within these pathways, attempts have been made to highlight differences between HTLV-1 and -2, which may relate to differences in their pathogenic properties.
The human T-lymphotropic viruses type 1 (HTLV-1) and type 2 (HTLV-2) are members of a family of transforming mammalian retroviruses with a similar genomic organization and common modes of transmission. HTLV-1 is endemic in a number of well-defined geographical regions and it is estimated that as many as 20 million individuals are infected worldwide (Yoshida, 2001; Matsuoka, 2003). While the vast majority of infected individuals remain clinically asymptomatic, some 2–5% will develop either a severe neurological or lymphoproliferative disorder. The latter, adult T-cell leukemia/lymphoma (ATLL) is a group of mature T-cell malignancies with distinct clinical presentations (Uchiyama et al., 1977; Takatsuki et al., 1985). ATLL generally occurs in individuals infected around the time of birth, and presents only after prolonged incubation periods ranging from 20 to 60 years which is consistent with an age-related accumulation of leukemogenic events and a multistep process of transformation (Yoshida, 2001; Matsuoka, 2003). HTLV-2 infection is also associated with the development of chronic neurological diseases (Araujo and Hall, 2004); however, there is no evidence that the virus causes significant lymphoproliferative disorders and certainly no disease equivalent to ATLL has been associated with infection suggesting that the oncogenic potential of HTLV-2 may be less than that of HTLV-1.
While the pathogenesis of ATLL remains incompletely understood the viral regulatory protein Tax appears to play a central role (Yoshida, 2001; Matsuoka, 2003; Azran et al., 2004; Jeang et al., 2004). Tax has been shown to transform primary lymphocytes and rat fibroblasts in vitro (Grassmann et al., 1992) and transgenic mice models have confirmed the oncogenic properties of the protein in vivo in that these have developed a range of malignant disorders including mesenchymal tumors, neurofibromas and large granular lymphocytic leukemia (Hinrichs et al., 1987; Nerenberg et al., 1987; Grossman et al., 1995). The transformation properties of Tax are a consequence of the ability of the protein to deregulate the transcription of genes and signaling pathways involved in cellular proliferation, cell cycle control and apoptosis. The complexity of gene expression deregulation has recently been highlighted in a series of studies involving transcriptional profiling using microarray analysis (Harhaj et al., 1999; De La Fuenta et al., 2000; Ng et al., 2001; Pise-Masison et al., 2002) and protein profiling employing a range of proteomic approaches (Wu et al., 2004). In the most recent transcriptome analysis, the expression of >7000 genes in both HTLV-1-transformed (IL-2 independent) and immortalized (IL-2 dependent) cell lines and in activated normal PBMCs were evaluated (Pise-Masison et al., 2002). In all more than 760 genes were found to be deregulated on average by twofold or greater in three of the five HTLV-1-transformed cell lines analysed. Tax functions primarily through protein–protein interactions. The modulation and deregulation of cellular signaling by Tax involves a range of pathways, including both direct and indirect interactions with a range of transcription factors such as CREB/ATF, NF-κB, AP-1, SRF and NFAT; and distinct cellular-signaling pathways, involving PDZ domain-containing proteins, Rho-GTPases and the Janus kinase (JAK)/signal transducer and activator of transcription (STAT), and transforming growth factor-β (TGF-β) pathways. Activation of CREB/ATF and NF-κB-dependent gene expression plays central roles in both viral replication and cellular transformation and proliferation. In brief, the Tax-CREB/ATF interaction is important in the regulation of virus transcription, and persistent and constitutive activation of NF-κB is central to the initiation and maintenance of the malignant phenotype in ATLL (Yoshida, 2001). NF-κB activation by Tax results in the upregulation of expression of a large number of cellular genes involved in cell proliferation, including a number of cytokines and their corresponding receptor genes. The CREB/ATF and NF-κB pathways are described in detail in accompanying reviews (see Kashanchi and Brady, and Sun and Yamaoka respectively, this issue). In this report, we will summarize the role and interaction of Tax and where relevant other viral proteins in the deregulation of the other noted signaling pathways. In addition, we will attempt to highlight differences in the deregulation observed between the HTLV-1 and -2 Tax proteins (Tax1, Tax2), which in turn may contribute to differences in the pathogenic properties of these viruses, a topic which is described in more detail by Feuer and Green (this issue).
AP-1 is a group of dimeric transcription factor complexes composed of members of the Fos (c-Fos, FosB, Fra-1 and Fra-2) and Jun (c-Jun, JunB, and JunD) families which play a central role in the proliferation and transformation of T lymphocytes, prevention of apoptosis and cytokine production (Shaulian and Karin, 2002). In unstimulated T cells, the basal level of AP-1 proteins is low, but T-cell activation results in a rapid induction of jun and fos genes (Behrens et al., 2001). Formation of the AP-1 transcription factors complexes follows activation by the serum responsive factor (SRF) in response to various external signaling events including serum, lysophosphatidic acid, lipopolysaccharide, 12-O-tetradecanoylphorbol-13-acetate (TPA), stimulation and the activity of cytokines and tumor necrosis factor-α (TNF-α). SRF acts through a SRF-responsive element (SRE) (Chai and Tarnawski, 2002), which contains two binding sites; a CArG box [CC(A/T)6GG] and an upstream Ets box (GGA(A/T)). Following binding to the CArG box, SRF protein interacts with the ternary complex factors (TCFs) (Elk-1, SAP-1), which then bind to the Ets box. In addition, SRF requires the CBP/p300 and p/CAF coactivators for transcriptional activation (Shuh and Derse, 2000). SRE was originally identified as serum inducible element in the c-fos promoter, and it has been shown to regulate transcriptional activation by serum, TPA and mitogens of several immediate early genes including egr-1 and egr-2 (Chai and Tarnawski, 2002; Buchwalter et al., 2004). HTLV-1-infected T-cell lines express a number of AP-1 proteins (Fujii et al., 1991, 1995b, 2000), and c-fos, fra-1, c-jun, and junD have been shown to be activated by Tax at the level of transcription (Fujii et al., 1988; Nagata et al., 1989; Sakamoto et al., 1992; Tsuchiya et al., 1993; Iwakura et al., 1995). Tax1 activates some of these immediate early genes by interacting with SRF (Fujii et al., 1992a, 1992b, 1995a) and with TCFs, CBP/p300 and P/CAF (Fujii et al., 1992a, 1992b; Shuh and Derse, 2000). The constitutive activation of these genes in infected T cells independent of external signals is believed to contribute to the initial steps in the transformation process in ATLL (Alexandre et al., 1991; Fujii et al., 1992a, 1992b; Tsuchiya et al., 1993).
In the induction of AP-1 gene expression, Tax1 activates transcription with considerable specificity (Iwai et al., 2001). A prototype AP-1-site from the collagenase gene, even containing multiple copies of the AP-1-binding site, did not respond to Tax1, although electro-mobility shift assays demonstrated significant binding to this element in HTLV-1-infected and Tax-expressing T cells. In contrast, only two copies of IL-8 AP-1-site cloned upstream of IL-8 minimal promoter responded to Tax1 (Iwai et al., 2001). HTLV-2 Tax, Tax2 also activated gene expression from the IL-8 AP-1-Luc reporter equivalent to Tax1 in Jurkat cells. The specificity observed would suggest that Tax1 may only activate a limited number of genes containing AP-1-binding sites; however the mechanism(s) of the specificity of Tax1 to the IL-8 AP-1-site is unclear and require further investigation.
Tax1 mutant studies have suggested that Tax1 activation through AP-1-site utilizes both the CREB/ATF/SRF and NF-κB pathways, since neither TaxM22 (defective for the NF-κB pathway) nor TaxM47 (defective for the CREB/ATF/SRF pathway) activated an AP-1 reporter construct (Iwai et al., 2001). Interestingly, coexpression of M22 and M47 exhibited activity to half of wild-type Tax1. Thus, two independent activities of Tax might be required for the activation through AP-1-site. M47 is active on the c-fos promoter through the SRE, and as such optimal activation through the AP-1-site might also require expression of the c-fos gene. AP-1 activity can also be regulated at the post-transcriptional level by the activation of c-Jun N-terminal kinase (JNK) (Davis, 2000). JNK phosphorylates c-Jun increasing its DNA-binding activity (Behrens et al., 1999). Tax contributes to this pathway by constitutively activating JNK (Xu et al., 1996; Jin et al., 1997).
The full biological consequences of AP-1 activation by Tax1 in the HTLV-1 life cycle have not yet been fully clarified. An infectious molecular clone of HTLV-1 carrying a Tax M47 mutant defective for AP-1 activation was found to be fully active in the transformation of T cells (Robek and Ratner, 1999). In addition, AP-1 activation by Tax1 was found to be dispensable for promotion of cell growth and inhibition of apoptosis in a T-cell line, as indicated by the IL-2-independent growth induction by Tax M47 (Iwai et al., 2001). Additional genes regulated by Tax1 through AP-1-site include TGF-β, fra-1, TR3/nur77 and TIMP-1 (Kim et al., 1990, 1991; Tsuchiya et al., 1993; Uchijima et al., 1994; Liu et al., 1999). With the exception of TGF-β to be discussed below, these genes do not have an immediately obvious role in the HTLV-1 life cycle. Thus, it is highly likely that as yet unknown AP-1 target genes may be found to play critical roles in HTLV-1 infection. It should be noted that AP-1 proteins such as c-Jun and c-Fos can activate the transcription through the 21 bp repeat in HTLV-1 LTR (Jeang et al., 1991; Fujii et al., 1995a). Thus under certain conditions, AP-1 could also play a role or contribute to pathways involved in viral gene expression.
As constitutive activation of AP-1 is observed in all the fresh primary uncultured ATL cells and in all ATL-derived cell lines (Mori et al., 2000), it would seem that activation is certainly important in the development of ATL. However, uncultured ATL cells do not or only express very low amounts of Tax (Mori et al., 2000). Thus, it is possible that the activation of AP-1 in ATL may not always be due to Tax, but could under certain conditions be mediated by genetic and/or epigenetic changes in leukemic cells. Activated AP-1 DNA binding activity in fresh ATL cells contained JunD, and JunD mRNA is highly upregulated in fresh ATL cells compared to normal PBMCs (Mori et al., 2000). Thus, the constitutive activation of AP-1 in fresh ATL cells could in part be explained by induction of the junD transcript.
The complementary strand of the HTLV-1 genome encodes HBZ (HTLV-1-encoded basic leucine zipper protein) (Gaudray et al., 2002). HBZ was originally identified in a ‘two hybrid’ screening using CREB-2 as ‘bait’, and is a nuclear basic leucine zipper protein, resembling both AP-1 and CREB/ATF. HBZ has both positive and negative effects on AP-1-dependent transcription and Tax activation of HTLV-1 LTR. Cotransfection experiments showed that HBZ through the formation of a CREB-2 heterocomplex, inhibited Tax-dependent activation of the LTR (Basbous et al., 2003). HBZ also inhibited the transcriptional activation of HTLV-1 LTR by c-Jun, through heterocomplex formation with and an associated degradation of c-Jun (Basbous et al., 2003; Matsumoto et al., 2004). In contrast, HBZ could be shown under certain conditions to stimulate JunB- or JunD-mediated activation of AP-1 (Basbous et al., 2003; Thebault et al., 2004). Thus, the importance of HBZ is unclear. However, it is interesting that an open reading frame encoding HBZ has not been found in HTLV-2, indicating that HBZ could make a distinct contribution(s) to HTLV-1 pathogenesis.
NFAT represents a family of enhancer-binding proteins that are involved in the regulation of the expression of number of genes encoding cytokines, cell surface receptors and transcription factors (Crabtree and Clipstone, 1994; Rao, 1994; Jain et al., 1995; Park et al., 1995; Ruff and Leach, 1995; Loh et al., 1996). The NFAT family consists of at least five structurally related proteins which are differentially expressed in different cell types, and whose activity is regulated by a calcium-dependent phosphatase, calcineurin (Shibasaki et al., 1996). As shall be detailed below, a number of studies have now clearly documented the involvement of NFAT in Tax activation of the promoters of the cytokines IL-2, and IL-13, the Fas ligand (FasL) and interferon regulatory factor 4 (IRF4).
Transient transfection studies have clearly shown that Tax together with TPA or ionomycin activates the expression of IL-2 gene through NFAT in the Jurkat T-cell line (Good et al., 1996, 1997). Moreover, the transcriptional activation of NFAT by Tax also appears to be tissue specific, in that Tax can activate NFAT in T-cell lines but not in the nonlymphoid cell lines F9 and Cos (Good et al., 1996, 1997). This tissue specificity may be due to the relatively low abundance of NFATp protein in nonlymphoid cells, since Tax1 together with exogenous NFATp efficiently activated NFAT-dependent transcription in Cos cells (Rivera et al., 1998). Tax induces transcription of the IL-2 gene through binding of transcription factors to the CD28-responsive element (CD28RE), to which NFAT predominantly binds. Tax1 mutants deficient in the NF-κB pathway (M22) and inactive in CREB (M47) exhibited only weak activation of NFAT-dependent transcription, but together with TPA and/or ionomycin synergistically activated this to high levels (Good et al., 1996, 1997; Rivera et al., 1998). The constitutive expression of functional IL-2 receptors (IL-2R) through NF-κB with activation of IL-2 gene expression involving NFAT has led to the hypothesis that the aberrant activation of an IL-2/IL-2R autocrine loop by Tax1 contributes to the proliferation of both infected and transformed leukemic cells in vivo (Cross et al., 1987; Maruyama et al., 1987). However, it should be noted that relatively few HTLV-1-infected T-cell lines with the exception of HUT102, express significant levels of IL-2 (Arya et al., 1984; Volkman et al., 1985) and thus how this relates to infection in vivo is unclear. HTLV-2 Tax, Tax2 can also activate IL-2 gene expression through NFAT, and interestingly the activity of Tax2 would appear to be much greater than Tax1 (Fujii, M et al., submitted for publication). HTLV-2-infected T-cell lines possess constitutively active NFAT and constitutively express both IL-2 mRNA and protein. Moreover, cyclosporine A (an inhibitor of NFAT) as well as anti-IL-2 receptor antibodies were found to inhibit the proliferation of HTLV-2-infected T-cell lines.
A recent report has also shown that there is a consistent upregulation of the IL-13 gene and increased expression of this cytokine in HTLV-1-infected cells and in cell lines derived from patients with ATLL. This upregulation has been reported to be mediated by Tax1 transactivation of NFATp with possibly the additive effect of a putative AP-1-binding site (Waldele et al., 2004). IL-13 has both proliferative and antiapoptotic properties and has been associated with leukemogenesis in Hodgkin's lymphoma. Neither the CREB pathway-deficient mutant M47 or the NF-κB activation-deficient mutant M22 were able to fully activate the IL-13 promotor. However, together they resulted in full activation which resembles the effects noted previously on the CD28RE and in relation to this it has been shown that CD28 signaling does contribute to IL-13 transcription (Minty et al., 1993, 1997). IL-13 production in turn can lead to IL-13 receptor (IL-13Rα1) expression (Gauchat et al., 1997) and the interaction of IL-13 with its receptor and resultant autocrine stimulation results in the activation of an antiapoptotic phosphatidylinositol-3-kinase-dependent-signaling cascade (Wright et al., 1999) producing an inhibition of apoptosis.
The IRFs are a family of transcription factors involved in hematopoietic development, cytokine signaling and cell growth. Of the nine members IRF-4 expression is specifically restricted to lymphoid and myeloid cells. In T lymphocytes, IRF-4 is induced upon T-cell activation which can be induced by TCR crosslinking, treatment with TPA/ionomycin and ConA or by anti-CD3/CD28 treatment. Recently, it has been shown that IRF-4 expression occurs in HTLV-1-infected cells and in leukemic cells in ATLL and that this like IL-2 is Tax dependent and involves activation of both NF-κB and NFAT pathways with the latter involving interactions of NFATp with the CD28RE (Sharma et al., 2002). In addition, overexpression of NFATp in Jurkat cells synergizes with NIK to activate the CD28RE, suggesting that both may be involved in IRF4/CD28RE transactivation. Overexpression of IRF-4 has been shown to transform Rat fibroblasts in vitro (Tsuboi et al., 2000) and as such in HTLV-1 infected cells this might be expected to contribute to the development of leukemogenesis. This is supported by the finding that IRF-4 expression appears to occur exclusively in ATLL cells but does not occur of lymphocytes in patients with HTLV-1-associated neurological disease (Mamane et al., 2002a). Several studies have identified direct downstream targets of IRF-4 including IL-4 (Rengarajan et al., 2002) and cyclin B1 (Mamane et al., 2002b) suggesting that the enhancement of IRF-4 expression by Tax in these pathways might possibly influence additional signaling events in infected cells. More recently it has been demonstrated that IRF-4 can specifically activate transcription of the IL-15 receptor (IL-15R) promoter, which is also known to be activated through the NF-κB pathway. Thus Tax coactivation of IRF-4 and NF-κB and upregulation of IL-15 and IL-15R may allow proliferation of infected and transformed cells in an autocrine loop fashion (Mariner et al., 2002).
HTLV-1-infected cells constitutively express FasL and it has been shown that Tax activation of this involves several transcription pathways including NFAT (Rivera et al., 1998). As was noted previously the M22 and M47 mutants had markedly reduced FasL transactivation activity; however, both mutants could synergize with mitogens, TPA and ionomycin to produce high levels of transactivation (Rivera et al., 1998). While it could be shown that NF-κB and AP-1 sites did not appear to be essential for FasL activation it remains possible that they may play some cooperative role with NFAT and/or that Tax may also induce other cofactors which contribute to NFAT function. At present the biological consequences of the increased expression of FasL remain unclear.
In addition to Tax, another HTLV-1-regulatory gene product p12I activates the expression of IL-2 through NFAT in Jurkat cells (Albrecht et al., 2002; Ding et al., 2002, 2003). Activation of NFAT by p12I requires simultaneous treatment with TPA. TPA activates the Ras/MAPK pathway, and U0126, an inhibitor of MEK-1 (MAPKK) abrogated the synergistic activation of NFAT by p12I and TPA, indicating that Ras/MAPK pathway is involved in the observed activity of p12I. AP-1 is a major downstream target for the Ras/MAPK pathway, and it forms a ternary complex with NFAT on the NFAT element in the IL-2 promoter to mediate the transcriptional activation. A dominant-negative mutant of AP-1 (A-Fos) completely abrogated the p12I-induced activation of NFAT as well as that of TPA/ionomycin stimulation (Albrecht et al., 2002). Thus, AP-1 activation would appear to be essential in the activation of NFAT by p121.
The roles of HTLV-1 p12I gene in the infectivity and the transformation of T cells in vitro and in vivo have also been studied by employing an infectious molecular clone of HTLV-1 (Albrecht et al., 2000). A molecular clone with inactivated p121 was found to be infectious for activated PBMCs treated with PHA and IL-2 and the infectivity was not only equivalent to wild-type virus, but also resulted in the efficient immortalization of primary T cells (Derse et al., 1997). In contrast the p12I mutant virus had markedly reduced infectivity for quiescent PBMCs (Albrecht et al., 2000). In vivo, the mutant virus produced reduced amounts of viral antigen and was associated with reduced anti-HTLV-1 antibody responses (Collins et al., 1998). These results suggest that HTLV-1 p12I, by activating host T cells during the early stages of infection, supports viral infectivity and this would be consistent with the activation of NFAT pathway and IL-2 production.
PDZ domain-containing proteins
The PDZ (PSD-95/Discs Large/Z0-1) domain-containing proteins include hDlg, a human homolog of the Drosophila discs large tumor suppressor protein, hScrib, human homolog of the Drosophila scribble tumor suppressor protein, MAGI-1, -2 and -3 (for ‘membrane-associated guanylate kinase with inverted orientation’), and MUPP1, a multi-PDZ protein. PDZ proteins are primarily localized at the membrane–cytoskeleton interfaces of cell–cell contact and form signaling complexes at the inner surface of the membrane to regulate cell growth, polarity, and adhesion in response to cell contact (Craven and Bredt, 1998; Fanning and Anderson, 1999). In addition, PDZ domain-containing proteins possess numerous other binding motifs including SH3, guanylate kinase-like, pleckskin and protein tyrosine phosphatase all of which highlight the possible involvement of these proteins in a range of other signaling processes (Panting and Phillips, 1995; Lee et al., 1997; Craven and Bredt, 1998; Bezprozvanny and Maximov, 2001). In Drosophila embryos, mutation of dlg or scrib causes cellular hyperproliferation and loss of cell polarity in epithelial tissues such as the imaginal discs, (Perrimon, 1988; Woods et al., 1996; Goode and Perrimon, 1997; Bilder et al., 2000; Bilder and Perrimon, 2000) demonstrating that both the structural and signaling functions of the PDZ proteins are essential for their antitumor activities.
A number of PDZ domain-containing proteins have now been identified as interacting with HTLV-1 Tax as well as the oncoproteins of several DNA tumor viruses including the E6 protein of high-risk papillomaviruses and the E4 oncoprotein (E4ORF1) of group D adenovirus (Kiyono et al., 1997; Lee et al., 1997, 2000; Rousset et al., 1998; Gardiol et al., 1999; Suzuki et al., 1999; Glaunsinger et al., 2000; Nakagawa and Huibregtse, 2000; Pim et al., 2000; Thomas et al., 2002; Lee and Laimins, 2004). Both Tax and the high-risk HPV E6 proteins bind to the PDZ proteins through a PDZ-binding motif (PBM) (X-(T/S)-X-V, where X is any amino acid) located at their extreme carboxy terminus (Songyang et al., 1997; Jelen et al., 2003). In contrast, none of the low-risk HPV E6 proteins possess such domains. Mutation of the second or fourth conserved amino acid residue ((T/S) or V) in this motif has been shown to compromise the transforming activity of E6, demonstrating that this PDZ domain-binding motif plays a critical role in E6-induced oncogenesis (Kiyono et al., 1997). Employment of the yeast two-hybrid system using Tax as ‘bait’ demonstrated the direct interaction of Tax with a number of PDZ domain-containing proteins including hDlg (Rousset et al., 1998; Suzuki et al., 1999). hDlg is a signaling molecule downstream of the Wnt/Frizzle pathway (Woods and Bryant, 1991) and is known to interact with the C-terminus of the tumor suppressor protein, APC (Matsumine et al., 1996). Notably the C-terminus of Tax containing the PBM and APC was found to compete in the interaction with hDlg (Suzuki et al., 1999). Tax-expression resulted in hyperphosphorylation of hDlg and the employment of Tax mutants demonstrated that Tax binding to hDlg was responsible for the induction of hyperphosphorylation. These two effects of Tax, the competitive binding with APC and the hyperphosphorylation of hDlg appear to be central to the ability of Tax to induce cell growth. BrdU incorporation assays demonstrated that binding of Tax to hDlg is able to perturb the cytostatic effect of hDlg and promotes abnormal proliferation of cells. Thus, while all of the functions of hDlg have not yet been fully elucidated, hDlg clearly acts as a tumor suppressor and this is supported by observations that overexpression of hDlg in the 3T3 fibroblast cell line resulted in the cell cycle arrest at G0/G1 (Ishidate et al., 2000), and this arrest could be abrogated by Tax1 expression (Suzuki et al., 1999).
The interaction of HPV E6 with the PDZ domain-containing proteins results in targeted degradation via a proteasome-mediated pathway and in the case of hDlg, this could be shown to correlate with the ability of E6 to transform rodent cells (Gardiol et al., 1999). More recently (Massimi et al., 2004) in a more direct study of HPV-infected cells derived from cervical tumors found that E6 tended to target the nuclear pools of the PDZ proteins rather than the membrane forms. Morever, it could be shown that whereas hDlg, MAGI-1 and MUPPI efficiently suppressed transformation, this was not observed for MAGI-2 and MAGI-3, suggesting that the former but not the latter function as tumor suppressors. In contrast, Tax1 did not induce the degradation of hDlg, but rather resulted in its translocation from a detergent-soluble fraction to a detergent-insoluble fraction (Hirata et al., 2004), suggesting that Tax1 may inactivate hDlg function by alteration of its intracellular localization.
In contrast to HTLV-1 Tax, HTLV-2 Tax (Tax2) does not contain a PBM and it has been proposed this may in part account for differences in the transforming properties of the two proteins (Endo et al., 2002; Hirata et al., 2004). While HTLV-2 infection has not been consistently associated with the development of malignant lymphoproliferative diseases in man, the virus can effectively transform human T lymphocytes in vitro (Chen et al., 1983). However, in contrast to HTLV-1 which preferentially transforms CD4+ T lymphocytes, HTLV-2 preferentially transforms CD8+ T cells (Miyoshi et al., 1981; Yamamoto et al., 1982; Chen et al., 1983; Ijichi et al., 1992; Wang et al., 2000), and it has been shown that establishment and maintenance of transformation by Tax2 is similar to Tax1 in that this involves both the NF-κB and CREB pathways (Green and Chen, 2001). Tax1 and Tax2 have more than 75% amino-acid identity, their C-terminal regions are quite divergent, and as noted a PBM is present in Tax1, but not in Tax 2 (Rousset et al., 1998; Suzuki et al., 1999; Hirata et al., 2004). Using assays of colony formation in soft agar, Tax1 has been found to transform the Rat-1 fibroblast cell line more efficiently than Tax2 (Endo et al., 2002), and this difference is related to the Tax1 PBM (Hirata et al., 2004). PBM is a mobile motif, since an artificial chimeric protein of full-length Tax2 with a 10 amino-acid peptide containing the Tax1 PBM induced an equivalent number of transformed colonies to Tax1 (Hirata et al., 2004). HTLV-1 and -2 are believed to have originated from cross-species transmission of simian T-cell leukemia virus type 1 (STLV-1) and STLV type 2 (STLV-2), respectively. Whereas, STLV-1 is associated with the development of leukemia and lymphoma in non-human primate species STLV-2 is not (Hubbard et al., 1993; Voevodin et al., 1996). Interestingly, STLV-1 Tax1 has a PBM, whereas STLV-2 Tax2 does not, which suggest that the PBM is evolutionarily conserved within the STLVs/HTLVs.
Both the number and size of the transformed colonies of Rat-1/Tax1 cells were greater than those of Rat-1/Tax2 cells, but the expression of Tax1 PBM was only correlated with the number but not the colony size (Hirata et al., 2004). These results suggest that the Tax1 PBM may have a selective role in the initiation of anchorage-independent growth of Rat-1 cells in soft agar but not the subsequent rate of growth. Rat-1 cells are immortalized but not transformed and are usually arrested at G0/G1 in the cell cycle in soft agar. Thus it is possible that Tax1 activity through an interaction involving the PBM may overcome cell cycle arrest. The activation of NF-κB by Tax1 has also been shown to be essential in the transformation of Rat-1 cells (Yamaoka et al., 1996, 1998; Matsumoto et al., 1997). It could be shown that the inactivation of NEMO/IKKγ abrogated the transformed phenotype of the Rat-1/Tax1 cells, but that this could be rescued by re-expression of wild-type NEMO/IKKγ (Yamaoka et al., 1998). However, deletion of the PBM was found to only minimally affect the transcriptional activity of Tax1 in the NF-κB pathway (Hirata et al., 2004; Ohashi et al., 2004) suggesting that activity involving the PBM may utilize a distinct mechanism to augment the transforming activity of Tax1. Reporter assays with Tax1 mutants have shown that the PBM does not influence either the CREB or AP-1 pathways. Thus, the mechanisms of the PBM and PDZ domain-containing proteins in the observed transformation of Rat-1 cells remains unresolved.
It is interesting to observe that HTLV-1 envelope glycoprotein gp21 through a PBM in the C-terminus of the protein can also interact with hDlg (Blot et al., 2004). Recombinant HTLV-1 containing a mutant gp21 with deletion of the PBM exhibited reduced cell–cell fusion activity in Jurkat cells. The HTLV-2 envelope protein also contains a PBM, and this also interacted with Dlg, and as such this is also likely to be involved in the regulation of HTLV-2 envelope-mediated cell fusion as well. However, it is unclear if the interaction of Tax with hDlg could negatively influence infectivity by reducing the levels available for efficient cell to cell spread of virus.
In addition to hDlg, Tax1 has also been reported to interact with other PDZ domain-containing proteins including TIP-1 (Rousset et al., 1998; Fabre et al., 2000), the IL-16 precursor protein (Wilson et al., 2003) and MAGI-3 (Ohashi et al., 2004). The biological significance of the TIP-1 interaction is unclear. In contrast, the precursor of IL-16, pro-IL-16 contains three PDZ domains and expression is associated with cell cycle arrest at G0/G1. Tax was found to interact directly with pro-IL-16 through its PBM and the first PDZ domain and that this negated the cell cycle arrest effect of pro-IL-16. Thus, it is possible that this interaction could play a role in the deregulation of cell proliferation by Tax1. Recently, it has also been shown that Tax1 transformation of Rat-1 cells in vitro is associated with both the induction of and a subsequent direct interaction with MAGI-3, a protein with multiple PDZ domains (Ohashi et al., 2004). As with other PDZ-containing proteins this interaction was dependent on the PBM in the C-terminal and like hDlg did not interact with Tax2. Colocalization studies also demonstrated that MAGI-3 altered the subcellular localization of Tax1 from the nucleus to the cytoplasm and perinuclear region. While it would be shown that all human T-cell lines including HTLV-1-infected lines expressed MAGI-3 mRNA a significant upregulation of MAGI-3 in HTLV-1 infected relative to uninfected cells was not observed. Thus, it is unclear what if any the role of the Tax1-MAGI-3 interaction may play in viral replication or viral transformation.
Rho family GTPases
A recent study by Wu et al. (2004) employing proteomic approaches to identify Tax-binding proteins in a HTLV-1-infected T-cell line (C8166) has identified direct interactions of Tax with several small GTPases, and GST ‘pull-down’ assays and Western blotting analysis confirmed that Tax directly interacted with RhoA, Racl and Cdc42. Classically Rho GTPases are activated by a range of transmembrane receptors with a routing of signals to effector proteins which are involved in a wide variety of cellular processes including cytoskeleton organization and transcriptional activation (Arndt et al., 2000). The small GTPases directly interact with and regulate the activities of proteins in the actin cytoskeleton. The activation of Cdc42, Rac and Rho in fibroblasts causes the formation of filopodia, membrane ruffling, and stress fiber formation, respectively (Stam et al., 1998; del Pozo et al., 1999; Maruta et al., 1999; Schmitz et al., 2000). A number of early studies have also shown that Tax can bind also directly with a number of cytoskeleton proteins including α-internexin and cytokeratin (Salvetti et al., 1993; Trihn et al., 1997; Reddy et al., 1998) and in their proteomic analysis (Wu et al., 2004) identified interactions with several other proteins including gelsolin and actin. Thus, the direct effects of Tax binding with components of the cytoskeleton and the secondary influences arising from interactions with the Rho GTPases would be expected to have profound effects on cytoskeleton organization and dynamics. This could certainly contribute to the loss of polarity of transformed cells, and also alter their migration and invasive properties. Indeed, this may play a role in characteristic aggressive cutaneous and visceral infiltration of leukemic cells which is observed in ATLL. (Takatsuki et al., 1985).
It is also likely that the interaction of Tax1 with the GTPases could result in significant transcriptional reprogramming. Specifically, the binding of Tax to the active forms of the small GTPases might be expected to modulate or influence the activity of secondary messengers. Such an interaction with Tax might be expected to influence JNK signaling as JNK components are regulated to a large extent by small GTPases. As shall be discussed later, several studies have shown that constitutive activation of JNK might promote IL-2-independent growth of HTLV-1-infected T cells (Xu et al., 1996; Jin et al., 1997). Thus, the involvement of Tax in this setting could well-affect JNK downstream targets such as DNA-binding transcription factors and MEKKs. In addition, on the basis of their proteomic analysis (Wu et al., 2004) it was proposed that Tax might also modulate the upstream activity of effecter proteins in combination with small GTPases, which in turn could influence downstream-signaling cascades, such as those involving JNK, p38, MEKKs and NF-κB (Montaner et al., 1998; Jeang, 2001). In this regard, it has been previously shown that an upstream protein, G-protein pathway suppressor 2, directly interacts with Tax and as a result modulates the activation of the JNK pathway (Jin et al., 1997). The observation that a number of oncogenes encode exchange factors of the small GTPases raises the possibility of additional mechanisms whereby Tax could indirectly influence the process of transformation. Thus, while the exact influence of Tax on small GTPase function still remains speculative, specific coexpression studies should readily reveal the biological significance of these interactions.
JAK/STAT pathway is one of the major cytokine-signaling pathways regulating T-lymphocyte function (Leonard and O'Shea, 1998; Imada and Leonard, 2000; Lin and Leonard, 2000). Specifically the STAT proteins are a family of transcription factors essential for cytokine-regulated processes including growth and proliferation by the activation of downstream genes (Williams, 2000; Bromberg, 2001). The STATs are activated by the JAKs, a group of receptor-associated enzymes with tyrosine phosphorylation activity. Whereas, JAK and STAT proteins are normally unphosphorylated and inactive in quiescent lymphocytes, JAK1, JAK3, STAT3 and STAT5 are activated in normal T lymphocytes in response to IL-2. Similarly, HTLV-1-transformed cells typically show constitutive tyrosine phosphorylation of JAK3 and STAT5 (Xu et al., 1995; Migone et al., 1995; Takemoto et al., 1997; Mulloy et al., 1998) and lymphocytes obtained from HTLV-1-infected patients display tyrosine-phosphorylation of JAK3, STAT5, and STAT3, but not JAK2 and JAK1. HTLV-1-immortalized cells are initially IL-2 dependent but can acquire independence from exogenous IL-2 support following prolonged periods in culture (Sun et al., 1999). It has been shown that this IL-2 independence corresponds with a gradual constitutive activation of JAK/STAT proteins in all HTLV-1 fully transformed T-cell lines and as such the activation of the JAK/STAT pathway would appear to play a major role in the IL-2-independent growth and proliferation of some HTLV-1-infected cells. This is indirectly supported by studies (Mohapatra et al., 2003) employing roscovotine, an inhibitor of cyclin-dependent kinases where it could be shown that this induced apoptosis in the HTLV-1-transformed cell line MT-2. Roscovotine was found to inhibit the tyrosine phosphorylation and activation of STATs suggesting an essential role for the latter in ensuring cell survival.
In a study of leukemic cells of ATLL patients, it could be shown that eight of 12 exhibited constitutive activation of JAK/STAT pathway on the basis of either DNA-binding assays or tyrosine phosphorylation of JAK and STAT proteins, and this activation was associated with augmented cell cycle progression (Takemoto et al., 1997). These results support the concept that constitutively activated JAK/STAT pathway is involved in ATLL development. However, the results contrast sharply with those of Zhang et al. (1999), which failed to demonstrate constitutive phosphorylation of JAK3 and STAT5 in leukemic cells from ATLL patients and who reported that this only occurred after stimulation with IL-2. As such in some cases of ATLL constitutive activation of the JAK/STAT pathway may not occur and instead this might only result following IL-2 activation. Thus, it would appear that constitutive activation of the JAK/STAT pathway may not be a consistent feature in ATLL. Recently, Fung et al. (2005) reported that STAT5-inducible gene expression is correlated with cell growth of a T-cell line immortalized by Herpes saimiri vector encoding Tax, and microarray analysis revealed that a number of STAT5-inducible genes including IL-5, IL-9, IL-13 were uniquely upregulated by IL-2 in the presence of Tax. These results are consistent with the findings that Tax directly activates transcription and expression of STAT1 and STAT5 genes (Nakamura et al., 1999). Thus IL-13 is upregulated in HTLV-1 infection through both NFAT and JAK/STAT activation and as previously noted this cytokine would appear to play an important role in lymphocyte proliferation and transformation.
In contrast to HTLV-1, transformation of T-cell lines by HTLV-2 does not appear to involve activation of the JAK/STAT pathway (Mulloy et al., 1998). In an analysis of six IL-2-independent HTLV-2-transformed cells lines using a range of assays there was clearly no evidence of activation. Thus, it would appear that there may be significant differences in the relative importance of JAK/STAT activation in transformation of T lymphocytes by the two viruses and additional studies are certainly required to establish the significance and the molecular mechanism(s) of activation of the JAK/STAT pathway in the development of ATLL.
TGF-β is a member of a family of growth factors, which control a wide range of cellular functions including cell cycle control, proliferation, differentiation and immune responses (Shi and Massague, 2003). TGF-β inhibits the growth of a wide range of cells including activated T lymphocytes and secondary to the negative effects on cell growth, is associated with cellular transformation. TGF-β initiates signaling by binding of TGF-β to the cell surface type II TGF-β receptor (TβRII), which recruits type I TGF-β (TβI) and the propagation of the intracellular signal to the nucleus occurs through phosphorylation of the Smad proteins. In all, there are eight Smad proteins, representing three functional classes; the receptor-regulated Smad (R-Smad), the co-mediator-Smad (Co-Smad) and the inhibitory-Smad (I-Smad). R-Smads (Smad1, 2, 3, 5 and 8) are directly phosphorylated and activated by the type I receptor kinases and undergo homotrimerization and formation of complexes with the Co-Smad, Smad4. The activated Smad complexes are translocated to the nucleus and in conjunction with other nuclear cofactors, regulate the transcription of target genes. The I-Smad, Smad6 and Smad7, negatively regulate TGF-β signaling by competing with R-Smads for receptor or the Co-Smad interaction and by targeting the receptors for degradation.
In ATL cells, constitutive AP-1 activity results in the production of TGF-β and this can be readily detected in the sera of infected individuals (Niitsu et al., 1988; Kim et al., 1990). TGF-β as noted can readily inhibit the growth of normal T lymphocytes but paradoxically HTLV-1-infected cells are resistant to this inhibition and in fact this can lead to increased virus production in vitro (Nagai et al., 1995). This finding may be partially related to recent studies by Jones et al. (2005) who demonstrated that TGF-β could induce the expression of GLUT-1, the major HTLV-1 receptor in quiescent T lymphocytes and enhanced the binding of HTLV-1 virions. Thus, this cytokine would clearly contribute to viral spread as a result of an increase in cell susceptibility to infection. The resistance to TGF-β inhibition was first observed by Hollsberg et al. (1994) who demonstrated that HTLV-1-infected T-cell lines derived from patients with HTLV-1-associated myelopathy by single cell cloning were resistant to growth suppression by the cytokine. Subsequent studies have not only confirmed the resistance of growth inhibition in HTLV-1 infection but have demonstrated that this is related to Tax expression. Mori et al. (2000) demonstrated that whereas Tax would clearly inhibit Smad-dependent TGF-β signaling this was not due to direct interactions of Tax and Smad proteins but instead involved the interference of Tax by binding CBP/p300 thus preventing its recruitment into transcription complexes on the TGF-β response elements. In contrast, Lee et al. (2002) reported a direct interaction of Tax with Smad2, Smad3 and Smad4 proteins both in vitro and in vivo and demonstrated that Tax could inhibit the interaction of Smad3–Smad4 and DNA binding. Thus, Tax may inhibit TGF-β signaling not only by directly interacting with Smad proteins but also by competing with the Smad CBP/p300 interaction. Subsequently, Arnulf et al. (2002) provided evidence that Tax can inhibit TGF-β signaling by reducing Smad3 DNA-binding activity. Tax was found not to bind Smad3 but instead induced JNK activity with c-Jun phosphorylation resulting in Smad3/c-Jun complex formation and a repression of Smad3 function. The Tax-induced TGF-β inhibition of infected cells would be expected to contribute to the proliferation of both infected and transformed cells. Moreover, the inhibition of normal lymphocyte function by TGF-β would be expected to decrease immune surveillance during infection, which would indirectly contribute to virus persistence and the survival of infected and transformed cells.
A series of studies employing genomic approaches to investigate gene expression both in HTLV-1-infected and -transformed cells have not only highlighted the marked complexity of deregulation associated with the activity of viral regulatory protein Tax but have permitted the specific identification of numerous signaling pathways involved in transformation. Similarly proteomic approaches have identified many novel new targets of Tax and allowed correlations with the findings of the genomic analyses. In this review, we have focused on the deregulation of cellular-signaling processes associated with the interaction of Tax with AP-1 and NFAT transcriptional activation, PDZ domain-containing proteins, Rho GTPases and the TGF-β and JAK/STAT-signaling pathways. As a result of space limitations other pathways which would also appear to be deregulated could not be discussed in detail. In particular, studies have suggested that the P13K/Akt pathway is activated during HTLV-1 infection (Cereseto et al., 1999; Liu et al., 2001; Fung et al., 2005) and may contribute to transformation events. The present review has also highlighted specific differences between HTLV-1 and -2 Tax proteins with respect to involvement in these pathways. Specifically, the absence of a PBM in Tax2, which would prevent interactions with PDZ domain-containing proteins and the lack of constitutive JAK/STAT activation in HTLV-2-infected cell lines suggest there are important differences in their pathogenic properties. Further studies should determine the biological implications of these observations. On a cautionary note as it is clear that many of the Tax-related events noted in the signaling pathways are based on in vitro studies and in some cases employing overexpression systems, the results obtained will ultimately require confirmation of their physiological significance and their relevance to events in vivo.