Among the HTLV-1 regulatory and accessory genes, the tax gene is thought to play a central role in leukemogenesis since it immortalizes T-lymphocytes in vitro, and induces various cancers in transgenic animals 67. However, an enigma is that Tax expression is not detected in about 60% of leukemia cases 4. Three mechanisms for inactivating Tax expression in ATL cells have been described: 1) genetic changes (nonsense mutation, deletion, and insertion) of the tax gene 48, 2) deletion of the 5' long terminal repeat (LTR) containing the viral promoter 910, and 3) DNA methylation of the 5 'LTR leading to promoter inactivation 1112. One possible scenario is that since Tax is the major target of cytotoxic T-lymphocytes (CTL) in vivo 13, these mechanisms to disrupt or decrease Tax expression facilitate the escape of ATL cells from host CTL. Interestingly, analyses of HTLV-1 provirus in ATL cells showed that the 3' LTR was not deleted and remained unmethylated. In addition, the pX region (located between env and the 3' LTR) encoding the regulatory and accessory genes also is maintained. Detailed analyses of defective proviruses lacking the 5' LTR revealed that all cases maintained the HBZ gene and the 3' LTR. In one ATL case, the polyadenylation site of the HBZ gene was deleted 10; however, this HBZ gene utilized a downstream cellular polyadenylation signal for transcription. Taken together, these findings suggest that HBZ gene transcription is indispensable for the development of ATL.

Two transcripts have been reported that encode the HBZ gene (Figure 1); one is spliced (sHBZ) and the other is unspliced (usHBZ) 1415. The spliced transcript of the HBZ gene was first reported by Satou et al. 5, followed by subsequent reports that additionally identified a second minor spliced transcript 1415. Furthermore, while transcripts of the spliced HBZ gene were detected in all ATL cell lines and cells freshly isolated from ATL patients, the tax transcript was undetectable in some cell lines and most ATL cases 5. Prior to this study, the transcription of HTLV-1 viral genes in ATL patients was deemed to be undetectable. Therefore, the HBZ gene is recognized as the first gene that is uniformly expressed in the leukemic cells of all ATL patients.

Both sHBZ and usHBZ have TATA-less promoters. sHBZ has multiple transcriptional initiation sites in the U5 and R regions of the 3' LTR, whereas the usHBZ gene initiates within the tax gene. It has been reported that Sp1 is critical for many TATA-less promoters. Consistent with this, the transcription of sHBZ gene is dependent on Sp1 16.

Expression of the sHBZ gene which was detected not only in ATL cells but also in T-cells of asymptomatic carriers, appears to be correlated with provirus load 5. Quantitative analyses of HBZ gene transcripts were reported by two groups 1718. The sHBZ gene transcripts were found to be four times higher than the usHBZ gene transcripts in both ATL patients and HTLV-1 carriers 17. In addition, the half-life of the sHBZ protein isoform is longer than that of the usHBZ isoform 16. Together, the data are consistent with Western blot analyses of HBZ protein in ATL cell lines that detected only sHBZ protein 19, further supporting the significance of sHBZ protein.

It has been reported that HBZ transcription is correlated with provirus load 1718. As described later, transcription of the HBZ gene is dependent on the basal transcription factor, Sp1 16. Therefore, it is conceivable that the HBZ transcripts are proportional to provirus load. More importantly, Saito et al. reported the correlation between the levels of HBZ gene transcripts and severity of HTLV-1 associated myelopathy/tropical spastic paraparesis (HAM/TSP), suggesting that HBZ gene expression might contribute to the pathogenesis of HAM/TSP 18.

Anti-sense transcription of HTLV-1 was first reported in 1989 20. A decade later, the viral protein was detected in HTLV-1-transformed cell lines and further identified as a binding protein to CREB2 by the yeast two-hybrid method. This viral protein bound to CREB2 through its bZIP domain 21, and was designated as the HTLV-1 bZIP factor (HBZ). 5' rapid amplification of cDNA ends identified two different HBZ transcripts: spliced and unspliced forms (Figure 1) 51415. The promoter regions for the spliced and unspliced HBZ transcripts were identified, and both promoters are TATA-less. Sp1 has been identified as a critical transcription factor for the expression of the sHBZ gene 16. The Tax-response element (TRE) motif, which is present in the U3 region of the LTR, functions as an enhancer of viral sense gene transcription. Tax forms a complex with CREB and p300/CBP, resulting in marked activation of viral gene transcription. The TRE in the 3' LTR also functions to enhance transcription of the HBZ antisense transcripts 1622. However, the enhancing activity for anti-sense transcription is relatively weak when compared with sense transcription 16. This is consistent with the finding that transcription of the HBZ gene is relatively constant in ATL cases regardless of the variable expression levels of the tax gene 18.

The HBZ protein contains three domains: activation, central and bZIP (Figure 2) 2123. HBZ binds to host factors with bZIP domains, which include c-Jun, JunB, JunD, CREB2 and CREB 2425. In addition, HBZ can bind to the p65 subunit of NF-κB 26. The HBZ protein is localized in the nucleus with a speckled pattern 27. Three regions are associated with nuclear localization: two regions rich in basic amino acids and a DNA binding domain (Figure 2). In addition, the integrity of the HBZ amino acid sequence is necessary for the speckled distribution in the nucleus. HBZ is localized in heterochromatin consistent with its association with transcriptional inhibition 23 In addition, HBZ has been shown to sequester JunB into nuclear bodies, thus suppressing JunB-dependent transcriptional activity 27.

The difference between the sHBZ and the usHBZ isoforms is only a few amino acids at the N-terminus (Figure 1) 15. However, there are notably distinct characteristics. The half-life of sHBZ is much longer than that of usHBZ 16. In addition, the sHBZ mRNA is more predominant than usHBZ mRNA 17; thus, the protein level of sHBZ is much higher than that of usHBZ.

Growth-promoting activity was observed only in a T-cell line expressing sHBZ, but not in usHBZ-expressing T-cells 16. Furthermore, HBZ RNA was shown to have growth promoting activity 5. The difference between sHBZ and usHBZ lies with the presence of the first exon. This region corresponds to the Rex-responsive element (RxRE) in the R region of 3'LTR (Figure 1). RxRE forms a tight stem-loop structure, which is recognized by Rex to facilitate the nuclear export of viral RxRE-containing RNAs. The opposite strand of spliced HBZ RNA forms a different stem-loop structure, which might interact with host factors to induce the proliferation of ATL cells.

Biological differences between sHBZ and usHBZ proteins have also been demonstrated. usHBZ protein can induce the degradation of c-Jun in a ubiquitination-independent manner 28. usHBZ protein directly interacts with both the 26 S proteasome and c-Jun, which results in the delivery of c-Jun to the proteasome. It has been reported that this activity of sHBZ is much weaker than that of usHBZ. However, inhibition of AP-1 mediated transcription by sHBZ was much stronger than that of usHBZ 16. For the sHBZ protein, in addition to its higher protein level, its action to inhibit DNA binding by c-Jun or to sequester c-Jun in nuclear bodies might represent predominant mechanisms of transcriptional suppression.

In vitro studies investigating HBZ functions include both over-expression studies and those evaluating HBZ in the context of an infectious viral molecular clone. Initial studies utilized yeast two-hybrid analysis to show an interaction between HBZ bZIP binding domain and the bZIP transcription factor CREB2 (ATF-4) (see Figure 3). It was further shown that this interaction abolished the binding of CREB2 to the TRE in the HTLV-1 promoter and the cyclic AMP response element (CRE) in cellular promoters, consistent with the observations of HBZ dose-dependent down-regulation of Tax-mediated viral transcription 2129. Other cellular proteins including CREB and p300/CBP interact with HBZ and contribute to the down-regulation of Tax-dependent viral transcription 3031. However, the interaction of HBZ with p300/CBP is via two HBZ amino terminal motifs (not the HBZ bZIP domain) and the p300/CBP KIX domain 31. HBZ, via its bZIP domain, also interacts with Jun family members including JunB, c-Jun, and JunD, thereby modulating their transcriptional activity 2425. Like CREB, HBZ decreases the DNA binding activity of JunB and c-Jun, thus disrupting basal transcription of both HTLV-1 and cellular promoters via attenuation of AP-1 activation (Fos/Jun dimers) 2432. Additional AP-1 transcriptional repression is explained by HBZ-mediated reduction in c-Jun stability via the proteasome-dependent pathway, and sequestration of JunB by HBZ within nuclear bodies 2732. In contrast to JunB and c-Jun, the interaction of HBZ with JunD stimulates its transcriptional activity and results in the activation of JunD-dependent cellular genes including human telomerase reverse transcriptase (hTERT) 33. The significance of this finding is that the activation of telomerase is a critical late event in tumor progression and that HBZ is the only viral protein expressed in all ATL cells. Thus, the activation of telomerase by HBZ may contribute to the development and maintenance of leukemic cells.

It has been proposed that a highly regulated pattern of HTLV-1 gene expression is critical for virus-mediated T-lymphocyte immortalization/transformation and disease pathogenesis 34. One study utilized real-time RT-PCR to determine the kinetics of viral gene expression in cells transiently transfected with an HTLV-1 proviral plasmid and in human T-lymphocytes newly infected by virus. The HTLV-1 gene expression profiles revealed that all sense and antisense transcripts increased over time and then plateaued to stable levels. Gag/pol, tax/rex, and env mRNAs were detected first and at the highest levels, whereas expression of the accessory genes, including the anti-sense HBZ, was at significantly lower levels than tax/rex 35. Arnold et al. evaluated the functional role of HBZ in the context of an infectious molecular clone and, like other HTLV-1 accessory gene products, determined that the protein was dispensable for viral-induced immortalization of primary human T-lymphocytes 19. However, a significant increase in p19 Gag production was observed in cell clones expressing HBZ defective proviruses, a finding consistent with the conclusion that in stable cell lines the loss of HBZ function results in increased Tax-mediated viral gene expression. Although the inhibition of Tax-mediated gene expression is a reported function of HBZ, the fact that HBZ is expressed in ATL cells lacking tax transcripts suggests that HBZ may have additional functions or activities. Satou et al. reported that repression of HBZ expression in ATL cell lines by shRNA resulted in a significant decrease in cell proliferation 5. Moreover, shRNA repression of HBZ expression in established HTLV-1-transformed cell lines and newly immortalized T-lymphocytes also significantly suppressed T-lymphocyte proliferation 19. Stable expression of HBZ enhanced IL-2-independent survival of Kit-225 and increased Jurkat cell proliferative capacity [5 and Green unpublished]. Introduction of mutations that either abrogated HBZ protein expression or disrupted the HBZ mRNA without affecting the protein coding sequence indicated that the HBZ RNA, specifically a stem loop structure near the amino terminus of the gene, promoted T-cell proliferation; this contrasts with the finding that the HBZ protein inhibited Tax-mediated transactivation 5. Thus, these findings led to the conclusion that the HBZ gene has a bimodal function in two different molecular forms. Microarray results and follow up analyses indicated that transcription of the E2F1 gene and its downstream targets were up-regulated in HBZ-expressing cells, suggesting a role for E2F1 in the enhanced proliferation mediated by HBZ 5. The HBZ RNA, not the HBZ protein, was shown to be responsible for the up-regulation of E2F1.

In addition to binding transcription factors containing a bZIP domain, HBZ has been shown to bind the NF-κB subunit, p65, and inhibit NF-κB activation. Interestingly, this inhibition is selective for the classical NF-κB activation pathway, as the alternate NF-κB pathway is not inhibited 26. Both the activation and the bZIP domains are important for the binding of HBZ with the Rel homology domain of p65 (Figure 2). HBZ not only inhibits DNA binding by p65, but also promotes the degradation of p65. As a mechanism of enhanced p65 degradation, HBZ increases the expression of E3 ubiquitin ligase, PDLIM2, resulting in ubiquitination and degradation of p65. Thus, HBZ can suppress the classical NF-κB pathway using two distinct mechanisms. Other viruses including EBV, African swine fever virus, hepatitis C virus, and human herpesvirus-8 have also been found to target p65 and inhibit the classical NF-κB pathway 36.

The classical and alternative NF-κB pathways have distinct regulatory functions. Accumulating evidence suggests that the alternative NF-κB pathway is more critical than the classical pathway in several cancers 37. The two NF-κB pathways differentially control genes with anti-apoptotic functions in lymphoma cell lines 38. Interestingly, HTLV-1 Tax is known to activate both the classical and the alternative NF-κB pathways, and it has been reported that the alternative pathway is critical to cellular transformation by Tax 39. Potentially because HBZ selectively suppresses the classical NF-κB pathway, in cells that express both Tax and HBZ, Tax activity appears to predominantly activate NF-κB through the alternative pathway.

Recently, PDLIM2 has been reported to suppress Tax-mediated tumorigenicity by promoting degradation of Tax 40. HBZ enhances the expression of PDLIM2 which should increase the degradation of Tax. Therefore, a complex scenario could exist in which HBZ suppresses Tax expression at the level of transcription and by enhanced degradation.

HTLV-1 Tax plays a central role in the immortalization of T-lymphocytes in cell culture and in the early stages of leukemogenesis in infected patients. The observation that HBZ down-modulates Tax function while also promoting cellular proliferation suggests complex regulations that could influence viral latency, persistence, and disease. Although disruption of HBZ protein in an infectious proviral clone had no effect on the ability of the virus to immortalize T-lymphocytes in tissue culture, the loss of HBZ did result in a significant reduction of proviral load and an attenuated antibody response against viral proteins in a rabbit model of infection 29. As early as two weeks post infection, the proviral load was reduced 5–50 fold, indicating that the HBZ protein was important very early in the infection process. Kinetic analysis of viral gene expression in PBMCs from newly infected rabbits revealed that tax/rex and gag/pol mRNAs were expressed at the highest levels immediately after infection and then progressively declined over time, eventually stabilizing at low levels 35. Conversely, HBZ was expressed at a low level early after infection, and continued to increase before reaching a plateau, which was in direct correlation with proviral load levels in infected rabbit PBMCs. These results revealed an inverse correlation between tax/rex and HBZ mRNA expression over time, which provided important evidence linking HBZ expression to proviral load and the survival of the virus infected cell in an infected host.

Transgenic mice expressing HBZ under the control of the mouse CD4 promoter/enhancer displayed an increase in CD4⁺ splenic T-lymphocytes suggesting that HBZ promotes proliferation of CD4⁺ T-lymphocytes in vivo 5. One transplant tumorigenicity study indicated that knockdown of HBZ in a transformed T-cell line significantly reduced tumor formation and organ infiltration in NOD/SCID^γchain-/- mice 19. Taken together, these animal model studies further support a role for HBZ in the T-lymphocyte proliferation.

The HBZ gene is expressed in ATL cells from all patients and its expression level is correlated with provirus load. In many ATL cases, HBZ is the only viral gene that is expressed. Furthermore, the HBZ gene has growth-promoting activity in vivo and in vitro, indicating that the HBZ gene likely is critical for ATL cells even at the late stage of leukemogenesis. There are two scenarios, not necessarily mutually exclusive, for the requirement of the HBZ gene in ATL induction. First, at the early stage of infection, both tax and HBZ genes are needed for the proliferation and maintenance of infected cells. Since Tax is the major target of CTL, cells without or with low Tax expression can evade immune surveillance. The finding that defective proviruses without a 5' LTR are generated after integration support this scenario. Second, only the HBZ gene is essential for maintaining oncogenesis while Tax contributes to the process that initiates and promotes cell growth with subsequently causes genetic instability. In either scenario, the HBZ gene is critical for oncogenesis by HTLV-1. In addition, since HBZ promotes proliferation of HTLV-1 infected T-cells, increased HTLV-1 infected cells should be implicated in pathogenesis of HAM/TSP. Since the HBZ gene is the sole viral gene expressed in ATL cells, it is an ideal target for therapy of not only ATL, but also HAM/TSP.

Since the discovery of HTLV-1, there have been significant advances to our understanding of virus biology, immunology, and oncogenesis. However, the precise mechanism of oncogenesis by HTLV-1 remains to be determined. Recent intensive research on the antisense HBZ gene has yielded new important insight into the disease process: the HBZ gene appears to be the only viral gene that is constantly expressed in HTLV-1-infected cells and ATL cells. The data is consistent with HBZ playing a critical role in the proliferation of newly infected cells as well as in transformed ATL. Therefore, therapy targeted against HBZ might provide a promising new approach to the treatment of ATL as well as HAM/TSP. Some aspects of the HTLV-1 antisense transcript function may be conserved in other retroviruses since similar antisense RNAs have also been reported for HIV-1 41.