Indeed, four E box motifs have been described in the LTR of HIV-1; two are located 11 base pairs (bp) upstream and 6 bp downstream of the TATA sequence 45. These two palindromic sequence motifs (CAGATG and CAGCTG) have been referred to as the 5' E box and the 3' E box, respectively 4. More recently, the presence of a unique E box sequence (CATATG) in the LTR of HTLV-1, located 28 bp upstream the TATA box, has been reported 6. The close proximity of these E boxes to the TATA region appears to be specific to the LTRs of HIV-1 and HTLV-1. In the 5' LTR of bovine leukemia virus, (an HTLV-1 related retrovirus), three E box motifs overlapping the cyclic AMP responsive elements (CREs) have been shown to be involved in transcriptional repression of BLV basal gene expression 7.

Disruptive mutagenesis experiments have underlined the functional importance of the TATA sequences and the flanking E boxes of the HIV-1 LTR, and more particularly of the 3' E box in regulating basal and Tat-induced gene expression 4. Thus, it has been observed that one natural clone of HIV-1 carrying two mutated E boxes exhibited a high LTR basal activity in U937 cells 8. More interestingly, experiments have been performed to characterize the bHLH proteins that bind to these E boxes in order to determine their role in the regulation of proviral transcription. Gel retardation analysis demonstrated that the specific binding of E box proteins (as either E47 homodimers or HEB homodimers or HEB/E47 heterodimers), and AP-4, a bHLH-ZIP protein, to the 3' E box of HIV-1 LTR 45. Recently, the binding of AP-4 (and possibly other bHLH proteins) to the 3' E box was found to exclude the binding of TATA-binding proteins (TBP) to the TATA box and to inhibit the LTR-mediated transcription of the HIV-1 provirus in vitro 49. Consequently, E boxes in the LTR may account for a modulation of viral replication, and even for the establishment and maintenance of latency in HIV-1 infected cells. The presence of an E box motif in the LTR of HTLV-1 and the observation that E47 was repressing both basal and Tax-induced LTR activity suggest that this bHLH protein indeed favors HTLV-1 proviral latency, possibly by interfering with the binding of TBP to the proximal TATA element 10. The functional significance of the HTLV-1 putative E box has not been assessed yet. However, the above observations suggest that these E boxes are able to mediate the effect of bHLH proteins on the LTR activity of human retroviruses. Of interest, the overlap of E box elements and CREs in the BLV LTR has been proposed as a strategy to allow better silencing of viral transcription. In this model, suppression of viral gene expression has been shown to contribute not only to the impairment of immune surveillance, but also to the onset and progression of lymphoid tumours in BLV-infected sheep 11.

Collectively, these observations support that E box motifs in the LTRs of HIV-1 and HTLV-1 might represent important mediators of proviral transcription, by allowing the binding of bHLH proteins that might interfere with the transcriptional complex recruited at the TATA element. Consequently, some of these motifs would contribute to post-integration latency by turning off proviral expression. Furthermore, epigenetic mechanisms could also be implicated in the long-term suppression of viral expression, as E47 is known to bring chromatin-remodeling complex to specific promoters and therefore could induce epigenetic changes in proviral genome 12. In return, that inhibition could be relieved by the action of tissue-specific class II bHLH proteins, such as TAL1 (T-cell Acute Leukemia 1), providing a way to exit proviral latency (Figure 1). Indeed, the binding of E47/HEB heterodimers to the HIV-1 3' E box is abrogated by the over-expression of a tissue-specific class II bHLH factor, TAL1, a functional inhibitor of E proteins 13. Similarly, the over-expression of TAL1 is also able to counteract the E47 protein-mediated inhibition of the HTLV-1 LTR 10. The regulation of proviral transcription during the early steps of T lymphocyte maturation might, therefore, be an important event contributing to the pathogenesis of this retroviral infection 1415. As an alternative to this scenario, one can imagine that the tissue-specificity of class II bHLH proteins may be linked to the differential ability of retroviruses to establish post-integration latency. For example, HIV-1 is known to establish latency in mature, but not in immature thymocytes 16. Such a difference may be linked to the expression of TAL1 that is restricted to immature thymocytes. Clearly, much more experimental work is needed to assess the functional significance of these E boxes; however, the observations that have been devoted to delineate their intervention in the control of proviral latency are suggestive of their critical contributions.

As outlined above, the functionality of the E box elements needs further experimental evidence. Nevertheless, one can already address their evolutionary significance through the analysis of their distribution and conservation in LTRs of retroviruses that infect individuals, populations and species. Indeed such an analysis might provide valuable information on their involvement as mediators of proviral latency and on the evolutionary significance of the mechanisms highlighted above. At the scale of one to a few individuals, nearly all HIV-1 quasispecies exhibit intact E boxes (85 to 100% of intact E boxes) 81718. At a larger phylogenetic scale, E box mutants are strictly restricted to HIV-1 subgroups E, F and G (most virological studies have been performed with subtype B isolates) 19. Interestingly, isolates of subtypes E and G while accounting for less than 9% of all HIV-1 infections were found to display a two- to three-fold higher LTR basal activity than those of subtype B, consistent with a role of the intact E box in decreasing viral expression 1920. Thus, the presence of intact E boxes in the HIV-1 LTR might account for a mechanism that regulates HIV-1 gene expression, as well as proviral latency.

Considering that HTLV-1 replication occurs mainly by clonal expansion of infected cells (rather than by reverse transcription), this retrovirus is expected to lack intra-individual sequence variations. However, variations have been observed at the population level. On the basis of phylogenetic analyses, 6 groups of HTLV-1 (HTLV-1a to f) have been identified 21. As the early diverged HTLV-1c sequences lack the putative E box 6, the ancestral state of HTLV-1 should be characterized by the absence of a TATA-neighbouring putative E box thus arguing for an active proviral transcription and reverse transcription. Of note, this putative E box (CANNTG) can only be found in the LTR sequences of HTLV-1a and HTLV-1e (as deduced from the analysis of alignments 6). HTLV-1a, often referred to as the Cosmopolitan clade, represents by far the most widely distributed strain in HTLV-1 infected persons. It is striking to observe that an overwhelming majority (97%) of HTLV-1a isolates harbour this putative E box (as deduced from the analysis of alignments 22), suggesting a selection for this motif during evolution (Figure 2). Such an observation might suggest that the presence of an E box might have contributed to the spread of the retrovirus, possibly by favouring the transcriptional repression of viral genes and thus facilitating virus- escape from the host immune system. Indeed, such a strategy might be critical to proviral maintenance via clonal expansion which necessarily precedes leukemia development.