Potential conflict of interest: Nothing to report.
Chronic hepatitis B virus (HBV) infection is a major risk factor for liver cancer development. HBV encodes the hepatitis B virus X (HBx) protein that promotes transcription of the viral episomal DNA genome by the host cell RNA polymerase II. Here we provide evidence that HBx accomplishes this task by a conserved and unusual mechanism. Thus, HBx strongly stimulates expression of transiently transfected reporter constructs, regardless of the enhancer and promoter sequences. This activity invariably requires HBx binding to the cellular UV-damaged DDB1 E3 ubiquitin ligase, suggesting a common mechanism. Unexpectedly, none of the reporters tested is stimulated by HBx when integrated into the chromosome, despite remaining responsive to their cognate activators. Likewise, HBx promotes gene expression from the natural HBV episomal template but not from a chromosomally integrated HBV construct. The same was observed with the HBx protein of woodchuck HBV. HBx does not affect nuclear plasmid copy number and functions independently of CpG dinucleotide methylation. Conclusion: We propose that HBx supports HBV gene expression by a conserved mechanism that acts specifically on episomal DNA templates independently of the nature of the cis-regulatory sequences. Because of its uncommon property and key role in viral transcription, HBx represents an attractive target for new antiviral therapies. (HEPATOLOGY 2012;56:2116–2124)
Chronic infection by hepatitis B virus (HBV) affects close to 400 million people worldwide and is a leading cause of hepatocellular carcinoma, one of the most common human cancers.1 HBV belongs to the hepadnavirus family of DNA viruses that also includes woodchuck hepatitis virus and ground squirrel hepatitis virus.2 HBV replicates its genome in a manner very analogous to retroviruses, by reverse transcription (RT) of an RNA intermediate into double-stranded DNA that serves as template for transcription by the host cell RNA Polymerase II (Pol II) (reviewed3). A distinctive feature of HBV, however, is that the viral DNA genome does not integrate into the chromosome of the newly infected cell but instead is maintained as a covalently closed circular DNA (cccDNA) molecule. The cccDNA is transcribed into four major RNA species encoding the viral proteins, including a more than full-length transcript termed the pregenomic RNA. The pregenomic RNA is then reverse-transcribed in the cytoplasm within newly formed viral particles. As the cccDNA does not replicate, a pool of 10-100 copies of the cccDNA is maintained by recycling of a small proportion of the newly synthesized viral DNA genomes into the nucleus.
HBV encodes the nonstructural hepatitis B virus X (HBx) protein that is conserved among mammalian hepadnaviruses, suggesting an important function. In support of this notion, the corresponding woodchuck hepatitis virus WHx protein is essential for the establishment of virus infection,4, 5 and there is evidence that the same is true for HBx.6, 7 Several studies suggest that a primary function of HBx in the HBV life cycle is to promote viral gene expression.8-10 Perhaps most compelling is the recent finding that primary human hepatocytes infected with HBx-deficient HBV particles show normal levels of cccDNA but essentially no viral gene expression.11
The underlying mechanism whereby HBx promotes viral messenger RNA (mRNA) synthesis remains elusive. In cell culture, HBx behaves as a pleiotropic transactivator capable of stimulating a variety of cellular and viral promoters.12, 13 Although typically modest, the transactivation activity of HBx is likely biologically relevant. It is conserved among the HBx proteins encoded by HBV, woodchuck hepatitis virus, and ground squirrel hepatitis virus.8 Furthermore, the ability of HBx to stimulate reporter gene expression and HBV replication correlate.10, 14 The current explanation for the pleiotropic transactivation effects of HBx is that the protein can interact with numerous cellular proteins and has functions in both the cytoplasm and the nucleus of cells. Thus, HBx has been proposed to activate diverse signal transduction pathways in the cytoplasm,12 whereas in the nucleus it is believed to function by way of direct interaction with transcription factors15, 16 (and references therein), components of the basal transcription machinery (reviewed12, 17), as well as DNA- and histone-modifying enzymes.18-20 That HBx may have so many activities is puzzling, especially because the HBx gene largely overlaps the polymerase gene on the viral genome, a situation that has likely limited its potential to evolve multiple functions. In the present study we provide an alternative explanation for the pleiotropic transactivation properties of HBx. Previous work has established that HBx and WHx bind to host cell protein UV-damaged DNA binding protein 1 (DDB1) and likely function as viral substrate-recruiting subunits of the DDB1-containing E3 ubiquitin ligase complex.14, 21 We show here that through its interaction with the E3 ligase, HBx up-regulates luciferase reporter and HBV gene expression by a mechanism that operates selectively on extrachromosomal DNA templates irrespective of the nature of the promoter sequences and cognate activators.
cccDNA, covalently closed circular DNA; DDB1, UV-damaged DNA binding protein 1; GFP, green fluorescent protein; HBV, hepatitis B virus; HBx, hepatitis B virus X.
Materials and Methods
GFP-HBx, GFP-HBx(R96E),22 GFP-SV5V,23 and the HBx(R96E)-DDB1 fusions14, 23 have been described and are all expressed at detectable levels.14, 22, 23 GFP-WHx was generated by amplifying the woodchuck WHx coding region by polymerase chain reaction (PCR) from a WHV genomic construct (OHVCGA prototype). The proteins were produced from the episomal vectors KEBOB-PL24 in Fig. 1A and EBS-PL24 in Figs. 1C, 2C, and 4C, and from the lentivirus vector pWPT9 in all the other figures.
Luciferase Reporter Constructs.
The promoter-free Firefly luciferase reporter plasmid (pGL3-Basic; no Pro in Fig. 2) and the derivatives containing the simian virus 40 (SV-40) promoter alone (pGL3-Promoter; SV40 Pro) or together with a SV40 enhancer located downstream of the luciferase gene (pGL3-Control; SV40 Pro + Enh) were from Promega. The reporter construct driven by the HBV Enhancer I and associated core promoter (HBV Enh I) was produced by replacing the SV40 promoter in vector pGL3-Promoter by the relevant region amplified from the HBV genomic construct (see below). The nuclear factor kappa B (NF-κB)-responsive reporter construct was generated the same way using a PCR-amplified fragment derived from plasmid pNF-κB-Luciferase (Stratagene). The plasmid containing the Firefly luciferase gene under the control of the human IFN-β promoter (IFN) was described previously.25 The tetracycline-responsive Firefly luciferase reporter construct pTRE2hyg (Tet-Luciferase) was purchased from ClonTech. The Renilla luciferase reporter construct used in Fig. 4C is driven by the cytomegalovirus major immediate early promoter.26 To allow for chromosomal integration, the HBV Enh I and NF-κB-responsive reporter constructs were cloned into the self-inactivating lentiviral vector pWPXL (http://tronolab.epfl.ch/), replacing the EF1α promoter and green fluorescent protein (GFP) marker.
HBV Integrative Construct.
The integrative HBV genomic construct bearing four consecutive point mutations in the 3′ redundant region was generated in a pBS-SK (Stratagene) backbone. It consists of a 1.2 unit-length HBV genome (payw*7) carrying a translational termination signal after codon 7 in the HBx gene,27 flanked by an upstream hygromycin-resistance gene derived from pTRE2hyg and a downstream GFP marker amplified from pEGFP-N1 (ClonTech).
Cell Culture, Transfection, Transduction, and Luciferase Assay.
The human hepatoma cell lines HepG2 (ATCC), HepG2tet-on,28 and derivatives were grown at 37°C in the presence of 5% CO2 in modified Eagle's medium (MEM) (Invitrogen or Sigma-Aldrich) supplemented with 100 U of penicillin/mL, 100 μg of streptomycin/mL, 2 mM L-glutamine, 1 mM sodium pyruvate, 1% nonessential amino acids, and 10% (vol/vol) fetal calf serum (Invitrogen or Sigma-Aldrich). Cells were transfected using FuGENE 6 or FuGENE HD (Roche) following the manufacturer's instructions. Transduction of cells and luciferase reporter gene assays are described in the Supporting Methods.
Establishment of HepG2 Cell Lines with an Integrated HBx or WHx Expression Construct, Reporter Gene, or HBV Genome.
The stable HepG2 clones expressing HBx and WHx from a tetracycline-inducible promoter (Fig. 1B) will be described elsewhere. The HepG2-derived cell lines containing a randomly integrated tetracycline-responsive Firefly luciferase gene (Fig. 4) or an HBV genomic construct (Fig. 6) were established as described in detail in the Supporting Methods.
Plasmid Quantification and RT-PCR Analysis of HBV mRNA.
Plasmid DNA extraction and quantification in Fig. 5B and HBV mRNA analysis in Fig. 6 were performed as described in the Supporting Methods.
HBx and WHx Up-regulate Transient Reporter Gene Expression by a Conserved Mechanism.
In most studies the stimulatory effect of HBx on transiently transfected reporter genes is modest, typically 2- to 4-fold. We found that, although HBx activity can vary considerably (see Supporting Methods), very strong activation was observed in some experiments. Thus, HBx could up-regulate over 35-fold the expression of a luciferase reporter gene driven by the HBV Enhancer I and associated core promoter in human hepatoma HepG2 cells, in which HBx enhances HBV replication8, 9, 27, 29 (Fig. 1A). HBx also exhibited activity when expressed from an HBV genomic plasmid or at very low levels from a chromosomally integrated construct (Supporting Fig. S1). The woodchuck WHx protein showed comparable transactivation potential, in accordance with previous studies (Fig. 1A).8 Activation by HBx and WHx decreased upon overexpression of the paramyxovirus SV5-V protein, which competitively inhibits HBx binding to DDB1,23 and this occurred only when HBx and WHx expression was low (Fig. 1B and data not shown). Furthermore, the HBx(R96E) point mutant that is impaired in its DDB1-binding ability14, 23 is essentially inactive in this assay (Fig. 1A). However, the mutant regains full activity when covalently fused to wildtype DDB1, a situation that forces interaction between the two proteins (Fig. 1C).23 This is not the case when mutations are introduced into DDB1 to block its incorporation into the E3 ligase complex, or further compromise the HBx-DDB1 interaction (Fig. 1C).14, 23 This suggests that HBx(R96E) is impaired solely in DDB1 binding and that HBx requires DDB1 to function as a subunit of the E3 ligase complex to carry out its stimulatory activity. We conclude that HBx and WHx can efficiently stimulate transient reporter gene activity and that they likely do so by a conserved mechanism involving the DDB1 E3 ligase.
HBx Functions Irrespective of Enhancer and Promoter Types.
We then examined whether HBx would exhibit the same strong activation potential on luciferase reporter constructs placed under control of other, unrelated promoter and enhancer elements. Figure 2 shows that this is indeed the case; HBx showed a similarly strong effect on expression of an SV40 promoter-driven construct, regardless of the presence or absence of a downstream SV40 enhancer (Fig. 2A), and on expression of an interferon-regulated promoter construct (Fig. 2A). HBx also increased activity of a synthetic NF-κB responsive promoter (Fig. 2B), and basal activity of a tetracycline-inducible promoter even in cells producing no tetracycline-regulated activator (Fig. 2C). In all cases, the DDB1-binding HBx(R96E) point mutant failed to transactivate, suggesting that the stimulatory function requires interaction of HBx with the DDB1 E3 ligase at all promoter types tested. This suggests that HBx functions by a common mechanism regardless of the nature of the cis-regulatory elements.
HBx Has No Effect on Reporter Genes Integrated into the Cellular Genome.
An obvious common feature of reporter constructs tested by transient transfection is the extrachromosomal nature of the DNA template. To examine whether HBx would exhibit similar pleiotropic transactivation properties on chromosomal reporter genes, we tested the effect of HBx on two reporter constructs driven, respectively, by the HBV Enhancer I and by the unrelated NF-κB responsive promoter, when randomly integrated into the host cell chromosomes. Surprisingly, whereas responding normally to HBx in a transient transfection assay, none of the two chromosomal reporter constructs was stimulated by HBx (Fig. 3A,B). This is not due to the integrated reporter genes becoming refractory to stimulation. Indeed, treatment with interleukin (IL)-1β, a cytokine known to activate the NF-κB pathway,30 stimulated the NF-κB reporter gene similarly (Fig. 3B), if not better (Fig. 3C), when the latter was integrated into the chromosome. Furthermore, HBx strongly synergized with IL-1β in up-regulating the transiently transfected but not the chromosomal NF-κB reporter construct (Fig. 3C). Thus, increasing expression of the integrated reporter gene does not restore its responsiveness to HBx. IL-1β had no effect on the HBV Enhancer I construct (Fig. 3A), indicating that HBV Enhancer I is not regulated by NF-κB. These results strongly argue against HBx acting through the NF-κB pathway. Instead, they suggest that HBx functions by an unusual mechanism that acts selectively on extrachromosomal DNA templates and independently of the nature of the cis-regulatory elements.
To provide further evidence for this possibility, we tested a third reporter gene for its responsiveness to HBx when integrated at various chromosomal locations. We chose the tetracycline-regulated promoter construct, which is up-regulated by HBx in a transient transfection assay (Fig. 2C). The reporter construct was randomly integrated into the chromosomes of an HepG2-derived cell line expressing the tetracycline-inducible transactivator.28 Four stable clones were selected that showed basal luciferase expression levels in the absence of tetracycline and HBx varying over a 250-fold range (Fig. 4A). Treatment with tetracycline led to an increase in luciferase gene expression in all cases (Fig. 4B). Remarkably, HBx showed a complete lack of activity in all four clones. Thus, HBx fails to stimulate a reporter gene integrated into the chromosome regardless of its chromosomal location and basal expression level.
To ensure that HBx retains stimulatory activities on extrachromosomal templates in these HepG2 clones, we cotransfected HBx together with a Renilla luciferase reporter construct whose expression can be distinguished from that of the stably integrated Firefly luciferase gene. HBx was indeed efficient at up-regulating the transiently transfected Renilla construct in the two clones tested, but had no effect on the chromosomal Firefly gene (Fig. 4C; Fig. S2). Thus, HBx promotes expression of an extrachromosomal reporter gene without having an effect on an integrated counterpart in the same cell.
HBx Does Not Affect Nuclear Plasmid Copy Number and Functions Independently of CpG Methylation.
A simple mechanism whereby HBx may specifically increase transiently transfected reporter gene activity is by promoting import of the transfected DNA into the nucleus, or by protecting the transfected DNA from degradation or methylation-mediated silencing.31 To address the first two possibilities, we measured the amount of nuclear plasmid recovered from cells expressing HBx or control proteins. To avoid interference with cytoplasmic DNA, we performed the experiment 5 days after reporter gene transfection, a time when HBx shows strong stimulatory effects and the transfected DNA has been eliminated from the cytoplasm.32 Figure 5 shows that under conditions in which it induced strong activation, HBx had no obvious effect on the amount of reporter DNA recovered (Fig. 5A,B). This argues against HBx acting on plasmid nuclear import or copy number. To address whether HBx would prevent methylation-mediated silencing of the transfected DNA, we examined the ability of HBx to increase activity of a luciferase reporter construct lacking CpG dinucleotides.33 HBx was similarly efficient at activating the CpG-free reporter gene, thus excluding that HBx functions by preventing plasmid DNA methylation (Fig. 5C).
HBx Promotes Transcription from the Natural HBV Episomal cccDNA but Not from a Chromosomally Integrated HBV Construct.
The ability of HBx to enhance expression from extrachromosomal DNA specifically is of special interest, because the HBV genomic template transcribed by RNA Pol II exists as an episomal, nonreplicating cccDNA in the infected cells.34 We therefore wished to assess whether HBx could discriminate between a chromosomally integrated HBV genome versus the extrachromosomal cccDNA, the natural viral template, in the same cells. For this purpose, we designed an integrative HBV genomic construct carrying a defective HBx gene and four consecutive point mutations in the 3′-terminal redundant region. As a result, the mRNAs transcribed from this construct will contain the mutations at their 3′ ends. However, during reverse transcription of the pregenomic RNA by the viral polymerase and its conversion into cccDNA, the mutations will be lost (Fig. 6B; Fig. S3). As a consequence, the mRNAs originating from the cccDNA can be distinguished from those transcribed from the chromosomal construct by RT-PCR using appropriate primers (Figs. S3, S4). We generated stable HepG2 cell lines containing the integrated HBV genomic construct and producing detectable amounts of cccDNA (Fig. S5). The cells were then transduced at high efficiency with wildtype HBx or the DDB1-binding defective HBx(R96E) mutant, the woodchuck WHx counterpart, or the paramyxovirus SV5-V protein that binds the DDB1 subunit of the E3 ligase the same way but lacks stimulatory activities (Fig. S1).14 As a further control, we transfected the Enhancer I-driven luciferase reporter construct, which is responsive to HBx and WHx in a transient assay (Fig. 1). As shown in Fig. 6A, HBx and WHx exhibited the expected stimulatory effect on transient luciferase expression, whereas HBx(R96E) and SV5-V were inactive. RT-PCR analyses using template-selective HBV primers revealed that HBx and WHx also potently up-regulated the expression of the cccDNA but had no noticeable effect on the expression of the integrated HBV genomic construct (Fig. 6B). The increase in cccDNA-derived mRNA levels in the presence of HBx and WHx was estimated by serial dilution to be in the range of 16-fold (Fig. S6B). Thus, HBx promotes HBV genome expression by a mechanism that is likely conserved among mammalian hepadnaviruses and that operates selectively on the natural episomal cccDNA template.
Recent work has demonstrated that a major role for HBx during HBV infection is to promote viral gene expression.11 Here we provide evidence that HBx exerts this function by an uncommon mechanism. We found that HBx can strongly up-regulate reporter gene expression and, unexpectedly, has activity only on episomal but not on chromosomally integrated templates. Because the same reporter constructs were used in both situations, these findings exclude that HBx acts on mRNA stability or translation efficiency, thus pointing to a transcriptional effect. Activation by HBx does not show any promoter specificity but invariably requires incorporation of HBx into the DDB1 E3 ligase complex. These findings make it unlikely that HBx functions by deregulating cellular transcription factors. Instead, they point to a common mechanism that operates independently of the mode of action of the activators and that involves some specific feature of the extrachromosomal DNA template. This is of interest because the HBV genomic template transcribed by RNA Pol II exists as an episomal entity in the infected cells.34 Indeed, we show that HBx promotes gene expression from the natural HBV cccDNA but not from a chromosomally integrated HBV construct.
The notion that HBx elicits pleiotropic transactivation effects by modulating, directly or indirectly, the activity of a number of unrelated transcription factors including NF-κB has been extensively documented (reviewed12, 13). However, most studies were conducted using transiently transfected reporter constructs. The observation that HBx induces expression of any transiently transfected DNA template, regardless of the promoter and enhancer sequences, suggests that data obtained in transient transfection assays should be interpreted with caution. For example, we found that activation of the NF-κB pathway up-regulates an NF-κB-responsive promoter construct both in transient transfections and when stably integrated into the cell chromosome. By contrast, HBx is effective only on the extrachromosomal reporter template, arguing against it acting through the NF-κB pathway. It may be prudent therefore to confirm the potential effect of HBx on the activity of specific transcription factors by either testing some known cellular target genes or by using chromosomally integrated reporter constructs.
How exactly HBx functions to specifically increase expression of extrachromosomal templates remains unknown. However, it likely does so by a conserved mechanism because woodchuck WHx also binds DDB1 and shows similar stimulatory abilities. HBx does not appear to have an effect on copy number or nuclear import of the plasmid DNA. This is consistent with previous work pointing to a nuclear function of HBx9, 35 and with its lack of effect on the amount of cccDNA in infected cells.11 We therefore envision two possible scenarios. One is that HBx acts directly on the DNA. Transiently transfected reporter plasmids36 and the HBV cccDNA37 are assembled into chromatin structures that differ from those of chromosomal genes. HBx may selectively bind extrachromosomal DNA templates because of their distinct chromatin organization. Once bound to the template, HBx may act like a cellular activator, by recruiting the basal transcription machinery or chromatin-modifying factors. Indeed, HBx has been proposed to promote HBV gene expression by recruiting the histone acetylases CBP/p300 and PCAF/GCN5 to the cccDNA.38 However, such a mechanism fails to explain why HBx stimulatory activity invariably requires HBx binding to the DDB1 E3 ubiquitin ligase machinery. Recent structural studies of the HBx-DDB1 complex strongly suggest that HBx functions as a substrate receptor to dock a yet unknown cellular factor to the DDB1 E3 ligase.14 Hence, were HBx to act directly on the DNA, we would favor a mechanism that involves ubiquitination of a component of the chromatin or basal transcription machinery.39
Another and perhaps more attractive possibility, which also relies on a E3 ligase substrate receptor function, is that HBx acts indirectly to counteract a cellular restriction factor by triggering its ubiquitin-mediated degradation, as shown recently for the Vpx protein of human immunodeficiency virus (HIV).40, 41 This factor may sense extrachromosomal DNA and silence its expression. Silencing, however, is unlikely to involve DNA methylation because HBx shows the same ability to up-regulate a reporter construct devoid of CpG dinucleotides (Fig. 5C). The factor may therefore function by reorganizing the chromatin into a repressed state, or by affecting the subnuclear localization of the transfected or viral DNA, which can in turn impact on their transcriptional activity.42 The identification of the HBx substrate(s) will likely provide key insights into the mechanism by which HBx mediates HBV gene expression.
We thank Chris E.P. Goldring for the HepG2tet-on cell line, Michael Rehli for the CpG-less reporter vector pCpGL, Joseph Curran for the Renilla reporter, Dominique Garcin for the IFN-responsive reporter, Patrick Salmon and Didier Trono for the self-inactivating lentiviral vector, and Walter Reith and Joseph Curran for critical reading of the article.