Functional Analysis of Ground Squirrel Hepatitis Virus Enhancer II

Abstract

We have characterized a major regulatory element of ground squirrel hepatitis virus (GSHV) located within a 90-nucleotide fragment of the core promoter upstream sequences and have compared its organization to that of woodchuck hepatitis virus (WHV) enhancer II (We2). The GSHV element (Ge2) stimulates transcription from the viral core promoter and heterologous promoters in an orientation-independent manner but displays a lower level of activity than We2 in transient transfection assays in human hepatoma cells. The general organization of Ge2 into binding sites for the liver-enriched HNF-1 and HNF-4 proteins and for ubiquitous factors of the NF1 and Oct families was found to be mostly conserved with respect to the homologous We2 region. Accordingly, transactivation by HNF-1 and HNF-4 plays an essential role in the liver-specific transcriptional activity of both the GSHV and WHV core promoters. Distinctive features of the GSHV enhancer consist of its ability to bind C/EBP family factors in a central motif that overlaps with one of the two HNF-4 sites and its differential binding affinities for HNF-4.

Hepadnaviruses are small enveloped DNA viruses that replicate in the host liver through reverse transcription of an RNA intermediate called pregenome and that cause acute and chronic hepatitis. They include the human hepatitis B virus (HBV) and related viruses infecting several rodents and birds. Chronic infections with HBV, woodchuck hepatitis virus (WHV), and ground squirrel hepatitis virus (GSHV) have been causally related to the development of primary liver cancer in the hosts (reviewed in reference 2). It is therefore important to elucidate the mechanisms that control the expression of viral genes and the hepadnavirus life cycle. A key event in hepadnavirus infection is the production of the 3.5-kb pregenome RNA, which also encodes the core antigen and the polymerase (reviewed in reference 12).

While hepadnavirus genomes have similar genetic organization and transcription patterns, they show marked evolutionary divergence in the enhancer sequences that modulate transcription from the viral promoters. The activity of the basal core/pregenomic promoter of HBV is controlled in a liver-specific manner by its coordinated interactions with two enhancers, distally located enhancer I (EnI) and enhancer II (EnII), the latter of which overlaps with the core promoter upstream regulatory sequences (28, 33, 43). In sharp contrast, a single enhancer, located immediately upstream of the core promoter, was characterized in the duck hepatitis B virus (6, 22). Recent studies of the WHV transcription control sequences similarly identified a single autonomous regulatory element (We2) which also maps to the core promoter upstream sequences (11, 38). It has been shown that WHV sequences homologous to HBV EnI display inefficient intrinsic activity in transient transfection assays (7, 37) but that they are essential for the establishment of an active enhancer complex in conjunction with We2 (9, 11). The tissue specificity of hepadnavirus core promoters and enhancers was found to be determined by different combinations of binding motifs for liver-enriched and ubiquitous factors. We have recently shown that the activity of WHV EnII is primarily controlled by the liver-enriched HNF-1 and HNF-4 proteins, although members of the NF1 and Oct families of transcription factors also bind in a central region (11). The GSHV regulatory sequences have so far received little attention. In a recent study, Ueda and coworkers observed qualitatively similar properties of the GSHV and WHV enhancer elements, although significant differences in the levels of EnII activity between the two related viruses were noted (37). Here we report a detailed analysis of GSHV sequences homologous to WHV EnII, showing that the GSHV element (Ge2) shares with We2 a common structural and functional organization but exhibits distinctive nuclear factor binding ability and reduced transcriptional activation properties.

Alignment of the GSHV and WHV core promoter sequences reveals 85% nucleotide homology in this region, as well as over the entire genome (Fig. 1A). In the basal core promoter sequences, a variant TATA box/initiator element and two distinct transcription initiation sites for the precore and core/pregenomic transcripts are well conserved among all mammalian hepadnaviruses (3, 8, 11, 23, 44). The proximal HNF-1 site of WHV (nucleotides 1830 to 1845; numbering according to Girones et al. [14]) is perfectly conserved in GSHV sequences. In contrast, some sequence divergence in the upstream region (nucleotides 1766 to 1810) might alter protein-DNA interactions previously identified in the We2 element, notably at binding motifs for the HNF-4, NF1, and octamer families of transcription factors (11, 38).

FIG. 1.

(A) Nucleotide sequence analysis of the GSHV and WHV core promoter regions. Only nucleotides that differ between the two sequences are represented in the WHV sequence. Numbering is as in reference 14. Initiation sites of precore and pregenomic RNAs are indicated by arrows, and the translation initiation codon for precore protein is indicated by an oval. The partially conserved basal core promoter element is boxed. WHV sequences known to bind HNF-1, HNF-4, and NF1 are shaded. The C/EBP-binding motif in GSHV is overlined by a black bar. Oct-binding sequences containing either a canonical octamer motif (overlapping with the NF1b site) or noncanonical motifs (overlapping with the GSHV C/EBP or WNF1a site) are boxed. DR1 is one of the two direct viral repeats involved in the replication process. (B) Transcriptional activation capacities of the We2 and Ge2 elements. At the left is a schematic representation of the constructs used in transient transfections of HepG2 cells. LUC, luciferase gene; N-myc2, minimal N-myc2 promoter. Numbers indicate positions relative to the transcription initiation site (solid arrows). The black box indicates the noncoding region; the grey box indicates the coding region. promC, WHV or GSHV sequences used in direct orientation as promoter sequences. The simian virus 40 bidirectional polyadenylation signal (SV40pA) was placed upstream of the inserted viral sequences (insert; broken arrows) to arrest transcripts putatively arising from cryptic-vector or viral insert promoters. The right side shows mean values of luciferase activity from at least three independent experiments. Numbers indicate fold activation relative to the WHV basal core promoter (lanes 1 to 3) or N-myc2 promoter (lanes 4 to 6).

Transcriptional activation potency of GSHV upstream core promoter sequences.

We previously reported that the We2 element is a potent, liver-specific transcriptional activator, acting both on the homologous WHV core promoter and on heterologous promoters in an orientation-independent manner (11). The activity of the corresponding GSHV sequences was assessed in transient transfections of HepG2 cells, by using as a reporter the firefly luciferase gene. Culture and transient transfection of human hepatoma cell line HepG2 were performed as described previously (41). Semiconfluent HepG2 cells were transfected by the calcium phosphate coprecipitation method using 13 μg of luciferase construct and 2 μg of β-galactosidase expression vector pCH110 per duplicate 6-cm-diameter dish. The constructs used in transient assays are represented in Fig. 1B. Homologous GSHV and WHV sequences known to encompass the viral core promoter were either inserted immediately upstream of the luciferase gene in the same orientation or fused in reverse orientation to a chimeric N-myc2 promoter/luciferase construct by a previously described cloning strategy (11). The data shown in Fig. 1B indicate that GSHV sequences corresponding in position to the We2 element are functionally homologous to the latter, activating transcription from the viral basal core promoter or from the promoter of the N-myc2 oncogene. In both cases however, the GSHV construct was less active than its WHV counterpart, the observed differences ranging from 2.3-fold with the N-myc2 promoter to 2.8-fold with the core promoter (compare lines 5 and 6 and lines 2 and 3). Similar results were obtained by using as a reporter the luciferase gene driven by the HSV thymidine kinase minimal promoter (data not shown). Our data are compatible with a recent report in which GSHV EnII was found to be fourfold less active than its WHV counterpart in N-myc2 promoter stimulation (37).

HNF-1 and HNF-4 binding sites in the GSHV core promoter.

DNase I footprinting analysis of the GSHV core promoter region was performed as described previously (11). The GSHV and control WHV probes used in DNase I footprinting were generated by PCR amplification of cloned GSHV (27) and WHV type 8 (WHV8) genomes (14) from positions 1696 to 1960. As shown in Fig. 2A, a large domain extending from positions 1755 to 1892 on the plus strand appeared similarly protected, to a first approximation, in GSHV and WHV sequences by proteins from mouse liver nuclear extract. HNF-1 and HNF-4 recombinant proteins both induced footprinting over one and two sites, respectively, on the GSHV probe (Fig. 2A, lanes 3 and 4), as previously shown for the WHV probe (11). Conversely, protection of these sites by mouse liver nuclear proteins was specifically abolished when double-stranded oligonucleotides representing the WHV HNF-1 and HNF-4a sites (WHNF1 and WHNF4a) were included as competitors in the reaction (Fig. 2B, lanes 5 and 6). Thus, liver-enriched factors HNF-1 and HNF-4 can bind at one and two sites, respectively, within the GSHV element, as previously shown for the homologous We2 sequence.

FIG. 2.

Comparative DNase I footprinting analysis of the core promoter regions of GSHV and WHV. (A) The DNA fragment (nucleotides 1696 to 1960) of WHV (plus strand; We2+) or GSHV (Ge2+) 5′-end-labeled at nucleotide 1696 was incubated in the presence of bovine serum albumin, mouse liver nuclear extract (NE), the His-tagged HNF-1 DNA-binding domain, or HNF-4 protein, as indicated. The positions of transcription factor binding motifs are indicated (Fig. 1A). (B) The corresponding regions of WHV and GSHV (minus strand) were analyzed with mouse liver nuclear extracts and with the indicated double-stranded oligonucleotides (50 ng) as competitors. The WHNF1, WHNF4a, and WNF1a oligonucleotides have been described previously (11), and Gola was a 20-bp GSHV oligonucleotide spanning nucleotides 1788 to 1808. Chemical cleavage products of the viral probes at A and G residues were used as size markers. The sequence ladder has to be read as C plus T for minus-strand analysis to refer to Fig. 1A where only the plus strand is represented. The positions of protein-binding motifs and of the footprints induced by mouse liver nuclear extract are indicated as bars. DNase I-hypersensitive sites are diagrammed as arrows.

While the HNF-1 site is strictly conserved between GSHV and WHV sequences, both GSHV HNF-4 sites (GHNF4a and GHNF4b) display slight nucleotide divergences from the corresponding WHNF-4 motifs (Fig. 3A). To compare the relative protein-binding affinities of the HNF-4a and -4b sites between GSHV and WHV, double-stranded oligonucleotides corresponding to all four sites were synthesized and their abilities to compete for the interaction of HNF-4 with the GHNF4b site were analyzed by electrophoretic mobility shift assay (EMSA), by using mouse liver nuclear extracts as described previously (11). As shown in Fig. 3B, the unlabeled WHNF4a oligonucleotide competed as efficiently as the GHNF4b oligonucleotide in binding the HNF-4 protein to the GHNF4b probe but WHNF4b and GHNF4a did not compete as efficiently as WHNF4a. In a quantitative analysis, the molar excess of unlabeled WHNF4b or GHNF4a oligonucleotides required to reduce binding to the labeled probe to 50% of its original value was estimated to be threefold and sevenfold higher, respectively, than the required molar excess of the WHNF4a oligonucleotide. Similar results were obtained with recombinant His-tagged HNF-4 protein (data not shown). These results strongly suggest that the HNF-4 binding sites of Ge2 and We2 display marked differences in binding affinities, with apparent decreasing affinity order as follows: WHNF4a = GHNF4b > WHNF4b > GHNF4a. Strikingly, the low-affinity WHNF4b and GHNF4a sites differ from the high-affinity sites by one nucleotide exchange (G→A) at the second position of either direct repeat. An identical mutation is found in the distal HNF-4 site of HBV EnII (16) (Fig. 3A). This site was recently shown to bind primarily HNF-4, while other members of the nuclear receptor superfamily, including COUP-TF1, ARP1, RXR, and PPAR, can interact with the proximal HNF-4 site of HBV EnII (25). Further studies are required to evaluate the possible contribution of these nuclear receptors to the activity of the GSHV and WHV core promoters.

FIG. 3.

HNF-4 binding by Ge2 and We2. (A) The HNF-4 consensus binding site (29) was aligned with WHNF4a and -b sites, with putative HNF-4-binding sites found in the GSHV core promoter, and with the proximal and distal HNF-4 sites of the HBV core promoter. +, plus strand; −, minus strand. Arrows indicate repeats. The top line gives the nucleotide most frequently found at each position. Less-frequent nucleotides are shown underneath, with the lowercase letters representing nucleotides at 10 to 20% of the sites. In the WHV and GSHV sequences, nucleotides are underlined when they match the consensus, not underlined when they are represented in 10 to 20% of the previously characterized sites, and indicated by a double dot when they are present in less than 10% of the sites. (B) Relative binding activities of the GSHV and WHV HNF-4 sites. EMSAs were performed with mouse liver nuclear extracts and the end-labeled GHNF4b oligonucleotide (0.5 ng). Assay mixtures contained either no competitor (lanes 1 and 14) or a 20-, 60-, or 200-fold molar excesses of the unlabeled oligonucleotides representing the HNF-4-binding sites in We2 and Ge2. The bound complexes were resolved in 6% polyacrylamide gels and quantified with a PhosphorImager (Molecular Dynamics). F, free probe.

A C/EBP binding site in Ge2 that partially overlaps with the low-affinity HNF-4 site.

We previously reported that the We2 region lying between the HNF-4a and HNF-4b sites harbors two variant NF1 sites [termed NF1(1*) sites], in which inverted repeats are separated by only one nucleotide (11). These WHV NF1a and NF1b motifs overlap with canonical and noncanonical Oct motifs (18, 32) (Fig. 1A). They constitute an extended, composite binding area for members of the NF1 and octamer families of transcription factors, although the in vitro footprinting activity appears to be essentially contributed by NF1 binding (11). Comparative DNase I footprinting analysis of the GSHV and WHV core promoters revealed a distinct protection pattern in this region (Fig. 2). A strong DNase I-hypersensitive site and a short unprotected region were specifically observed on both the plus- and minus-strand GSHV probes around position 1780 (Fig. 2B, lane 4, and data not shown). Immediately adjacent to this DNase I-sensitive region, a footprint coinciding with the NF1b motif was efficiently abolished by the WHV NF1a oligonucleotide (WNF1a), used as a competitor (Fig. 2B, lane 7). As the NF1b motif of GSHV differs from the WHV NF1a and NF1b sites at a single position and as WNF1a and WNF1b also differ at this position, this competition strongly suggests that NF1 binding is responsible for the footprint observed over the GNF1b sequence.

In contrast, the WNF1a oligonucleotide competitor did not significantly affect the DNase I digestion pattern between the NF1b and HNF-4a sites, in the region corresponding to the NF1a site in We2 (Fig. 2B, lane 7). Moreover, a protected subdomain extending into the HNF-4a motif up to position 1810 emerged distinctively on both strands upon competition with the WHNF4a oligonucleotide (Fig. 2B, lane 6, and data not shown). To better explore the protein-binding pattern in this domain, we designed GSHV-specific oligonucleotide Gola (5′ CTGGGCATGATGCAAAAGGAC 3′, positions 1788 to 1808; Fig. 1A). Gola used as a competitor in a DNase I footprinting analysis of the GSHV probe was found to restore sensitivity to DNase I digestion in the corresponding region (Fig. 4, lanes 4 to 6; Fig. 2B, lane 8). In contrast to the corresponding WNF1a oligonucleotide, Gola was unable to compete for the binding of octamer family members in an EMSA (data not shown). An examination of Gola sequences revealed a potential C/EBP binding site (ATGATGCAAA). Accordingly, an oligonucleotide matching the consensus C/EBP binding site with directly abutted GCAAT inverted repeats, but not the WNF1a oligonucleotide, competed for protection against DNase I digestion over the Gola region with stoichiometry similar to that of the Gola oligonucleotide itself (Fig. 4, lanes 7 to 9 and 10 to 12). Furthermore, bacterially expressed C/EBPα -β, and -δ all induced a protected subdomain extending over precisely the same region (lanes 15 to 17). This footprint was also specifically induced with mouse liver nuclear extract preliminarily heated at 80°C, a treatment known to preserve C/EBP protein-binding capacity (Fig. 4, lane 18), and protection over this subdomain was abolished by using Gola as a competitor (lane 19), while the corresponding WHV sequence (WNF1a) had no effect (lane 20). Under these conditions, no significant protein binding was observed on any other area of the GSHV core promoter region or on the WHV probe (data not shown).

FIG. 4.

DNase I footprinting analysis of the GSHV core promoter region, with focus on the region extending from nucleotides 1770 to 1830 (plus-strand probe). Ranges of 2 to 20 ng of competitor oligonucleotides were included in the upper panel. Further increasing the amount of competitors resulted in nonspecific cross-competition of C/EBP binding. C/EBP proteins were purified from a bacterial expression system (42). When indicated, mouse liver nuclear extract (NE) was subjected to a 5-min treatment at 80°C before incubation with the probe. Other symbols and terminology are as defined in the legend for Fig. 2.

These data demonstrate that the Ge2 element harbors a C/EBP binding site which overlaps with the distal part of the low-affinity HNF-4a motif. This situation is reminiscent of the organization of HBV EnII, in which box-α contains overlapping binding motifs for a C/EBP-like protein and for HNF-4 (16, 45).

Substantial sequence divergence between the Oct-binding motifs in We2 and the corresponding GSHV region prompted us to analyze the ability of Oct proteins to bind Ge2 in the central region. We first compared the binding capacities of two oligonucleotides spanning either the WHV classical Oct motif (Woct; positions 1770 to 1790) (11) or the corresponding Ge2 sequence (Goct; 5′ TGGCATGCTAAGCGACAGCTG 3′) in an EMSA. Incubation of the Woct probe with mouse liver nuclear extracts yielded one major retarded complex (designated complex A), and the unlabeled Woct oligonucleotide competed effectively for specific protein binding (Fig. 5, lanes 1 and 2). Identical results were obtained with unlabeled Goct and H2Boct (a strong Oct site of the histone H2B promoter [5]) as the competitors (data not shown). In contrast, protein interactions with the Goct probe were detected as a band with a mobility identical to that of complex A together with a faster-migrating complex (Fig. 5). In competition experiments, a 100-fold excess of the unlabeled Woct oligonucleotide competed for binding of complex A but not of the faster-migrating complex (Fig. 5, lanes 2 and 4), while a similar excess of the unlabeled Goct competitor prevented any detectable protein binding, suggesting specific DNA-protein interactions (data not shown). The associated factor(s) was not investigated further. Thus, the Oct-binding capacity of the WNF1b site is retained in the corresponding GSHV sequence, in spite of substitutions in the consensus octamer motif (39). The Oct-binding capacities of the regions spanning the WHV NF1a site and the GSHV C/EBP site, which harbor noncanonical Oct-binding motifs (nc1; positions 1790 to 1810; Fig. 1A) were next investigated. Strikingly, incubation of the Wnc1 and Gnc1 probes with mouse nuclear extracts yielded similar binding patterns (Fig. 5, lanes 5 and 7). Two complexes were generated: a minor one with mobility identical to that of complex A and a prominent, faster-migrating doublet designated complex B. The Woct oligomer competed effectively for complex A formation and to a lesser extent for complex B formation (Fig. 5, lanes 6 and 8), and a similar excess of the unlabeled Wnc1 and Gnc1 oligonucleotides prevented formation of both complexes when incubated with the homologous labeled probes (results not shown). These data strongly suggest that the different Oct-binding sequences of the e2 central region exhibit overlapping binding specificities but might preferentially associate with distinct members of the octamer family (26). Furthermore, the binding specificities of Wnc1 and Gnc1 sites are not markedly different, in spite of significant nucleotide changes in the Ge2 region corresponding to the noncanonical Wnc1 Oct motif, predicted to interact suboptimally with both POUs and POUhd units of the Oct-binding domain (32). This may be attributed to the emergence of a distinct, more POU-specific Oct motif that is shifted by three nucleotides in the Ge2 region overlapping with the C/EBP binding site (Fig. 1A). Collectively, the results presented here underscore the conservation of a peculiar binding pattern with multiple, intricate motifs in the central e2 region, although Ge2 differs from We2 by the loss of one NF1 site and by its specific ability to bind C/EBP factors.

FIG. 5.

EMSA of Oct protein binding to the central region of Ge2 and We2. The different oligonucleotide probes named above each lane were derived from GSHV and WHV sequences that overlap with NF1 or C/EBP binding sites, as shown in Fig. 1A. A 100-fold molar excess of unlabeled Woct competitor was included when indicated (+). The specific DNA-protein interaction detected by the Woct probe is indicated as complex A. Faster-migrating complexes formed with the Goct probe are designated by a star, and those formed with Wnc1 and Gnc1 are indicated as complex B.

HNF-1 and HNF-4 sites are essential for Ge2 activity.

The functional relevance of each of the characterized DNA-protein interactions in the Ge2 element was then assessed by introducing mutations in the GSHV core promoter/luciferase construct and by analyzing luciferase expression in transiently transfected HepG2 cells. The enzymatic inverse PCR method (31) for site-directed mutagenesis was used to introduce mutations in selected motifs. Remarkably, mutation of the GHNF1 site, which eliminated factor binding, resulted in a 100-fold reduction of Ge2 activity (Fig. 6, lines 1 and 2). Moreover, Ge2 activity was reduced by 10-fold when mutations were introduced into either of the GHNF4 sites and by 100-fold when both GHNF4 sites were mutated (lines 2 to 5), and mutations in all three HNF-1 and HNF-4 sites further decreased luciferase activity to barely detectable levels (line 6). Thus, as previously shown for We2, cooperative interactions between liver-enriched transcription factors HNF-1 and HNF-4 bound at three sites in Ge2 sequences are crucial for transcriptional activation mediated by the GSHV element. We found a major difference in the relative impact of mutations introduced into the HNF-4 binding sites between the related WHV and GSHV core promoters. Whereas the data presented here demonstrate synergistic interactions between the two Ge2 HNF-4 sites, partial redundancy of the We2 HNF-4 sites was inferred from a mutational analysis showing, for each site, a moderate decrease of the WHV core promoter activity (30 to 50% of wild type) when activity was assayed either in the context of the full-length genome (11) or in luciferase constructs (9a). This suggests different interplays between activators bound to cis regulatory elements and the basal machinery at the core promoters of GSHV and WHV.

FIG. 6.

Impact of mutations in different protein-binding sites on the capacity of Ge2 to activate the core promoter. At the left is a schematic representation of the GSHV core promoter region and an alignment of the viral sequences inserted in constructs. Transcription initiation sites are indicated by arrows. Sequences of the basal core promoter include a TATA box/initiator element (box with gradient shading) surrounded by two stretches of 20 to 25 bp conserved with HBV (grey ovals) and a second putative initiator element (white box). Transcription factor binding sites shown in Fig. 1A are represented as shown in the top line and mutated sites are indicated in black. Mutations that abolished DNA-protein interactions were as follows: HNF-1, ATCCAGTATTCGA; HNF-4a, GGACGTTTCGACT; HNF-4b, TGAAGTTTCTACC; C/EBP-Oct a, TGGGCTAGATGGTTA; NF1-Oct b, CAAGGCATGCTTTG (mutated positions are underlined). At the upper right is a schematic representation of the constructs used in transient transfections of HepG2 cells (see Fig. 1B for more details). At the lower right, the bar diagram depicts the mean (± standard error of the mean) luciferase activities of Ge2 mutants relative to that of the wild type, obtained from more than three independent transfection experiments. The activity of the wild-type core promoter construct was arbitrarily set at 100.

In contrast, mutations at the NF1b-Oct site or at the C/EBP-Oct site (Fig. 6, lines 7 and 8) did not significantly modify Ge2 activity compared to that of the wild type, and slightly enhanced luciferase expression. Combining mutations at NF1b-Oct and C/EBP-Oct sites yielded a similar result (Fig. 6, line 9). Therefore, neither the NF1-Oct nor the C/EBP-Oct motifs in the central region of Ge2 were found to have any significant influence on the level of transcription from the nucleocapsid promoter in transient assays. Noticeably, similar data were obtained in mutational analysis of the corresponding We2 NF1-Oct sites (11).

In this study, we show that liver-specific transcriptional activity of the GSHV core promoter is mediated by cooperative interactions of HNF-1 and HNF-4 proteins with one and two cis-acting sites within the Ge2 element. Thus, the general organization and functional importance of the major nuclear factor recognition sites in Ge2 are shown to be roughly conserved between GSHV and WHV (11). Accordingly, the core promoter could be swapped between GSHV and WHV genomes without major impact on viral transcriptional activity in transient transfection assays (9a). The strong HNF-1-binding site at nucleotides −78 to −66 of the precore mRNA start site plays a predominant role in the activity of GSHV and WHV core promoters, as also found for duck hepatitis B virus (21). Transcription factors of the HNF-1 homeoprotein family are pleiotropic activators of liver-specific genes (36). Interestingly, mutations in the HBV core promoter that create an HNF-1 binding site are associated with enhanced replication in patients with severe liver disease (15). The presence of two binding motifs for the HNF-4 orphan receptor (29), which might also be recognized by other members of the nuclear hormone receptor superfamily, is also crucial for Ge2 and We2 activities. However, marked differences in HNF-4-binding affinities between the GSHV and WHV sites, as well as in the synergistic interplay between the two HNF-4 sites, might account for the lower level of activity of Ge2 in transfected hepatoma cells.

Moreover, distinct changes in the pattern of protein binding to a centrally positioned region which exhibits significant nucleotide divergence could be conclusively demonstrated. Whereas this region of WHV EnII was previously shown to constitute a composite binding area for NF1 and Oct proteins (11), our present data indicate that a single NF1-binding site was retained in the corresponding GSHV region, which evolved binding specificity for members of the C/EBP family of transcription factors (20), as well as reduced binding affinity for Oct proteins. Although these changes are expected to affect the interactions of the core promoter upstream sequences with the basal transcription machinery, a clear definition of the contribution of this region to Ge2 and We2 activity appeared to be refractory to standard analysis by transient transfection assays. This may arise as a consequence of atypical chromatin packaging of transfected DNA (19) or of a defect in the ability of established human hepatoma cell lines to express some factors required for WHV transcription (7). However, in recent studies in vivo, mutation of the canonical Oct site of WHV enhancer II had no effect on the virus ability to infect and replicate in woodchucks, whereas virus with mutations in the HNF-1 site was apparently unable to grow (41a). Interestingly, the C/EBP-binding site of Ge2 did not confer significant responsiveness to expression of C/EBPα, -β, or -δ or sensitivity to the expression of the CHOP protein, a natural inhibitor of C/EBPs (9a). Features differentiating We2 and Ge2 elements might be indirectly linked to transcription, residing instead in architectural aspects (34). Indeed, differences in the patterns of DNase I-hypersensitive sites between We2 and Ge2 in the central region might reflect in differential accessibility and distortion of the DNA molecule assembled in nucleoprotein complexes. Oct binding to DNA through the POU domain can induce marked bending associated with DNase I-hypersensitive sites (24, 40). Thus, the loss of a composite NF1-Oct binding site as well as differential binding affinities of the Oct sites in Ge2 compared to We2 may plausibly result in differential DNA bending and altered interactions with the basal transcription machinery.

The high oncogenicity of WHV correlates with insertional mutagenesis of myc oncogenes (10, 13), and recent studies have designated the e2 element as the major cis-acting viral element candidate for myc gene activation (11, 37). By contrast, the related GSHV is less oncogenic and appears to be inefficient at carrying out selected integration events, a difference attributed to viral determinants rather than host factors (17, 35). Whether WHV and GSHV genomes differ in their abilities to integrate into appropriate loci or to cis activate the myc promoters is presently unknown. In either case, it is tempting to speculate that the differences between the major enhancer elements of WHV and GSHV noted in this study might be relevant. Compelling support for this hypothesis stems from earlier studies of slow-transforming retroviruses, demonstrating that subtle alterations in the protein-binding capacity of the enhancer may result in dramatic differences in the latent period of tumor onset (30). In particular, mutations in one of the two NF1 sites of the Moloney murine leukemia virus markedly decreased the oncogenic potential of this virus. It has been shown that NF1 interacts with histone H3 (1), and it might counteract repressive chromatin structures as reported for a simian virus 40 replication origin (4). Thus, the loss of one NF1 site in Ge2 might be of little consequence for transcription of the episomal genome but might have a critical impact on the enhancer activity in the context of integrated viral sequences during hepatocarcinogenesis. Differential binding affinities for HNF-4 and/or other nuclear receptors and for octamer proteins might have similar effects. Studies of chimeric viruses in which the core promoter/e2 element has been swapped between GSHV and WHV are in progress to delineate the contribution of this viral determinant in the process of myc gene activation in whole-animal hosts.