Hepatitis B virus regulatory HBx protein binding to DDB1 is required but is not sufficient for maximal HBV replication

Abstract

Robust hepatitis B virus (HBV) replication is stimulated by the regulatory HBx protein. HBx binds the cellular protein DDB1; however, the importance of this interaction for HBV replication remains unknown. We tested whether HBx binding to DDB1 was required for HBV replication using a plasmid based replication assay in HepG2 cells. Three DDB1 binding-deficient HBx point mutants (HBx⁶⁹, HBx^90/91, HBx^R96E) failed to restore wildtype levels of replication from an HBx-deficient plasmid, which established the importance of the HBx-DDB1 interaction for maximal HBV replication. Analysis of overlapping HBx truncation mutants revealed that both the HBx-DDB1 binding domain and the carboxyl region are required for maximal HBV replication both in vitro and in vivo, suggesting the HBx-DDB1 interaction recruits regulatory functions critical for replication. Finally we demonstrate that HBx localizes to the Cul4A-DDB1 complex, and discuss the possible implications for models of HBV replication.

Introduction

Hepatitis B virus (HBV) infection is a serious health problem worldwide, with greater than 350 million people chronically infected and at risk for developing serious liver disease, including cirrhosis, fibrosis, and hepatocellular carcinoma (Beasley, 1988; Seeger et al., 2007). HBV is a 3.2-kb, partially double-stranded DNA virus that has four overlapping reading frames, which encode seven viral proteins: three surface antigens, a polymerase, two core proteins, and HBx. The HBx protein is the sole HBV regulatory protein and it has multiple functions both in vitro and in vivo, including transactivation of cellular and viral promoters (Spandau and Lee, 1988; Twu and Schloemer, 1987), activation of signaling pathways (Benn and Schneider, 1994; Cross et al., 1993), alteration of cell cycle progression (Benn and Schneider, 1995; Gearhart and Bouchard, 2010; Hodgson et al., 2008; Koike et al., 1994; Madden and Slagle, 2001), induction or prevention of apoptosis [reviewed in (Bouchard and Schneider, 2004)], and inhibition of cellular DNA repair (Becker et al., 1998; Groisman et al., 1999; Madden et al., 2000; Prost et al., 1998). HBx also acts as a tumor promoter in transgenic mice (Madden et al., 2001; Terradillos et al., 1997). HBx localizes to both the cytoplasm and the nucleus, where it presumably has different functions [reviewed in (Bouchard and Schneider, 2004)]. With the establishment of a plasmid-based HBV replication assay, it is now known that HBx is required for maximal virus replication (Bouchard et al., 2002; Keasler et al., 2007; Leupin et al., 2005; Tang et al., 2005) although the mechanism by which HBx facilitates HBV replication remain unclear.

HBx interacts with several cellular proteins and may mediate its role in virus replication through these interactions. The most well-characterized HBx binding partner is the damage-specific DNA binding protein 1 (DDB1) (Lee et al., 1995; Sitterlin et al., 1997; Lin-Marq et al., 2001) [reviewed in (Keasler and Slagle, 2008)]. DDB1 is a highly conserved, multi-functional protein expressed in both the nucleus and cytoplasm (Liu et al., 2000). The interaction between HBx and DDB1 is conserved among the HBx proteins from all mammalian hepadnaviruses (Sitterlin et al., 1997), suggesting an important role for this interaction in virus replication. Further, interaction of the woodchuck hepatitis virus (WHV) X protein (WHx) with DDB1 is critical for WHV replication in woodchucks (Sitterlin et al., 2000). The minimal DDB1 binding domain on HBx has been identified by several laboratories to be amino acids 88–100 (HBx^88–100) (Fig. 1A).

Figure 1. HBx point mutants that do not bind DDB1 fail to restore HBx-deficient replication.

(A) Schematic representation of HBx protein showing minimum domain required for binding to DDB1 [adapted from (Keasler and Slagle, 2008)]. (B) Schematic of wildtype and point mutant HBx proteins. DDB1 binding determined previously (Becker et al., 1998; Lin-Marq et al., 2001). (C) Quantitation of capsid-associated viral DNA as described in Materials and Methods. Mean copy number from cells transfected with pHBV was set to 100% and compared to others. Error bars (SE) from three independent experiments. Statistical significance compared to pHBV is noted by an asterisk (p<0.05). (D) Western blot detection of wildtype and point mutant HBx proteins with rabbit anti-HBx.

DDB1 functions as an adaptor protein for the Cul4A E3 ubiquitin ligase complex (Angers et al., 2006; Higa et al., 2006; Shiyanov et al., 1999). DDB1 recruits DDB1 Cullin Associated Factors (DCAFs), which in turn recruit substrates to the DDB1-Cul4A complex for subsequent ubiquitination and degradation by the proteosome (Angers et al., 2006; He et al., 2006; Higa et al., 2006). In this manner, DDB1 plays important roles in diverse cellular processes, such as DNA synthesis, gene expression, cell division, and apoptosis. The DCAFs have in common a 16-aa DDB1-binding WD40 (DWD) motif that is characterized not by amino acid sequence, but rather by the biophysical/biochemical characteristics of those amino acids (He et al., 2006). The minimal DDB1-binding domain of HBx (aa 88–100) contains a sequence similar to that of DWD motifs in cellular DCAFs (Keasler and Slagle, 2008; Li et al., 2010). A recent high resolution crystal structure demonstrated the direct interaction between DDB1 and an HBx^88–100 peptide (Li et al., 2010). The specific region on DDB1 to which HBx binds is shared with at least 79 other DCAFs (He et al., 2006). This raises the possibility that HBx binding to DDB1 may benefit virus replication by displacing DCAFs and thereby altering the spectrum of DCAFs and their substrates recruited to the Cul4A complex.

In the present study, we used the plasmid-based HBV replication assay in cultured HepG2 cells and in hydrodyanamically injected mice to investigate the contribution of the HBx-DDB1 interaction to HBV replication. Three HBx point mutant proteins that no longer bind DDB1 were unable to restore HBx-deficient replication, demonstrating that the HBx-DDB1 interaction is required for maximal HBV replication. However, further analysis of HBx truncation mutants revealed that HBx-DDB1 binding is not sufficient to restore HBx-deficient replication, and that an additional function(s) residing in the carboxyl half of the HBx protein is essential for maximal replication.

Results

HBx interaction with DDB1 is required for virus replication

The importance of the HBx-DDB1 interaction has been suggested by the observation that an HBx point mutant that no longer binds to DDB1 (HBx^R96E) was unable to restore HBx-deficient pHBVΔX replication (Leupin et al., 2005). In the present study, the same plasmid-based HBV replication assay that requires HBx expression for maximal virus replication was used to investigate the HBx-DDB1-mediated requirements for HBV replication. Human liver HepG2 cells were transfected with a plasmid encoding a greater-than-genome length (129%) HBV genome (pHBV) (Melegari et al., 1998) or an identical plasmid encoding HBV with a point mutation in the X open reading frame that prevents HBx expression (pHBVΔX) (Scaglioni et al., 1997). Cells receiving pHBVΔX were additionally co-transfected with plasmids encoding HBx proteins that either bind DDB1 (e.g., HBx, HBx⁷) or do not (HBx⁶⁹, HBx^90/91, HBx^R96E) (Fig. 1B). Quantitation of capsid-associated DNA revealed that HBx-deficient replication from pHBVΔX was reduced by approximately 60% compared to wildtype pHBV (Fig. 1C). Replication from pHBVΔX was restored to wildtype pHBV levels by co-transfection of a second plasmid encoding HBx^WT, as reported previously (Bouchard et al., 2002; Keasler et al., 2007; Keasler et al., 2009; Kumar et al., 2011; Leupin et al., 2005; Tang et al., 2005) and also by the HBx⁷ point mutant that retained DDB1-binding (Fig. 1C). Importantly, the three HBx point mutants that do not bind DDB1 failed to restore pHBVΔX replication. We have previously shown that the amount of HBx required for maximal replication in the HepG2 assay is well below the limit of detection by our sensitive western blot assay (Keasler et al., 2009). Therefore, expression of these HBx mutants at detectable levels (Fig. 1D) indicates that sufficient protein was present to restore HBVΔX replication. No conclusions can be drawn from this experiment regarding the relative levels of the HBx and mutant proteins, as we are not certain the polyclonal anti-HBx serum reacts equally with all HBx proteins. Together, these results confirm the previous finding that HBx^R96E is unable to restore pHBVΔX (Leupin et al., 2005), and extends that result to include two additional HBx mutants that lack DDB1-binding (HBx⁶⁹, HBx^90/91). We conclude that the interaction of HBx with DDB1 is required for maximal virus replication in this assay.

HBx interaction with DDB1 is not sufficient for virus replication

DDB1 functions as an adaptor protein that recruits DCAFs to the Cul4A E3 ligase complex for ubiquitination (Angers et al., 2006; He et al., 2006; Higa et al., 2006; Shiyanov et al., 1999). HBx binds to a site on DDB1 (Li et al., 2010) to which at least 79 DCAFs bind (He et al., 2006), and displaces at least two DCAFs (Bontron et al., 2002; Li et al., 2010), presumably altering the spectrum of DCAF-recruited substrates that are ubiquitinated and degraded. These observations lead to the question: Is HBx binding to DDB1 sufficient for HBV replication?

The minimal domain of HBx required for binding to DDB1 contains aa88–100 (Fig. 1A), and the smallest trunctation mutant used in this study, HBx^55–101, contains the full DDB1 binding site. To determine whether HBx binding to DDB1 is sufficient to restore replication, we repeated the wildtype (pHBV) and HBx-deficient (pHBVΔX) replication assay and tested the ability of co-transfected pSI-X^WT, pSI-X^55–101, or a negative control vector alone (pSI) to restore pHBVΔX replication. Although HBx^WT restored HBx-deficient replication, HBx^55–101 was unable to increase HBV replication above the levels of pHBVΔX alone (Fig. 2A), even though this mutant is able to bind DDB1 (Becker et al., 1998).

Figure 2. HBx^55–101 is not sufficient to rescue pHBVΔX replication.

(A) Quantitation of capsid-associated viral DNA. Mean copy number from cells transfected with pHBV was set to 100% and compared to others. Error bars (SEM) are from three independent experiments. Statistical significance compared to pHBV is noted by an asterisk (p<0.05). (B) Mice were hydrodynamically injected with plasmid DNA and viremia measured on day 4 post-injection at the peak in viremia. The number of mice per injection group is indicated (error bars shown are SEM), and statistical significance is designated with asterisks. A total of 22 mice were used for this experiment. (C) Representative western blot detection of wild type and truncation mutant HBx proteins using an anti-HA monoclonal antibody. HBx-specific bands in each lane are indicated by stars; other bands, apparent also in the pSI control (lane 1) are considered non-specific.

To confirm this result in vivo, we repeated this experiment using the hydrodynamic injection model for HBV replication. Groups of ICR mice were injected with pHBV or pHBVΔX, and mice receiving pHBVΔX were additionally co-injected with pSI-X, pSI-X^55–101, or empty pSI as a negative control plasmid. Mice were sacrificed at 4-days post injection, a time point that coincides with the peak in viremia (Keasler et al., 2007; Yang et al., 2002), and capsid-associated DNA was quantitated. Mice injected with pHBVΔX had 95% lower viremia than those injected with pHBV (Fig. 2B), which confirmed our previous results (Keasler et al., 2007; Keasler et al., 2009). Reduced viremia in pHBVΔX-injected mice was restored to maximal levels by co-injection of pSI-X, but not by the pSI negative control plasmid. Importantly, mice co-injected with pHBVΔX and pSI-X^55–101 had viremia at a level similar to miice injected with pHBVΔX alone (Fig. 2B). These data confirm those obtained in transfected HepG2 cells and demonstrate that DDB1-binding alone is not sufficient to rescue HBx-deficient replication. We conclude that, in addition to the DDB1-binding domain, other regions of HBx are required for maximal virus replication in this assay.

Rescue of HBVΔX replication requires the HBx carboxyl terminal domain

We next sought to define the HBx domains that act in concert with the DDB1 binding domain to restore HBx-deficient replication. We first tested HBx truncations that retained either the amino (aa1–101) or the carboxyl (aa43–154) half of HBx for their ability to restore HBx-deficient replication in HepG2 cells (Fig. 3A). Although both truncations included the full DDB1-binding domain, HBx-deficient replication was restored to maximal levels by the HBx^43–154 carboxyl half of HBx, but not by the amino half HBx^1–101 truncation mutant (Fig. 3B). Expression of HBx^1–101 and HBX^43–154 was confirmed by western blot (Fig. 2C, lanes 3 and 5). These results indicate that the amino terminal 42 aa of HBx was dispensable for replication in this assay, and that the carboxyl portion of HBx (aa43–154) was sufficient for HBV replication in HepG2 cells. The observation that both HBx^1–101 and HBx^43–154 contained the DDB1-binding domain (Fig. 3A), yet only HBx^43–154 restored HBx-deficient replication, further supports the conclusion that the DDB1-binding region alone is not sufficient to restore HBx-deficient replication.

Figure 3. HBx^43–154 is sufficient to restore pHBVΔX replication.

(A) Schematic representation of HBx protein and truncation mutants. Gray box indications the region of HBx known to bind DDB1 (see Fig. 1A). “+” indicates DDB1 binding, as reported previously (Becker et al., 1998). (B) Quantitation of capsid-associated viral DNA. Mean copy number from cells transfected with pHBV was set to 100% and compared to others. Error bars (SEM) are from three independent experiments. Statistical significance compared to pHBV is noted by an asterisk (p<0.05).

We next sought to demonstrate whether the HBx carboxyl-terminal domain downstream of the DDB1 binding domain, i.e., HBx^101–154 (Fig. 4A) could restore pHBVΔX replication. While HBx^WT was able to restore HBx-deficient replication, truncation mutants HBx^55–101 and HBx^101–154 failed to do so either when transfected singly or co-transfected (Fig. 4B). These results indicate that both the DDB1-binding domain and the carboxyl half of HBx are required for HBV replication and that these two function in cis to potentiate HBV replication.

Figure 4. HBx^101–154 does not restore HBx-deficient replication.

(A) Schematic representation of HBx protein and truncation mutants. (B) Quantitation of capsid-associated viral DNA. Mean copy number from cells transfected with pHBV was set to 100% and compared to others. Statistical significance compared to pHBV is noted by an asterisk (p<0.05). (C) HBx transactivation of HBV ENH1-luciferase. Normalized transactivation from cells receiving pSI was set to 1.0 and compared to fold transactivation in the presence of wild type or mutant HBx. Error bars (SEM) are from three to five independent experiments. Asterisks indicate statistical significance compared to pSI-X.

HBx transactivation of HBV Enhancer 1

HBx is well known as a broadly-acting transcriptional coactivator that augments the expression of both viral and cellular genes (Spandau and Lee, 1988; Twu and Schloemer, 1987) [reviewed in (Bouchard and Schneider, 2004)]. Our results above suggest that, in addition to DDB1 binding, the carboxyl portion of HBx also functions to restore HBx-deficient replication in HepG2 cells. Therefore, we next tested the ability of select HBx truncation mutants to transactivate an HBV Enhancer 1 (ENH1) luciferase reporter. HepG2 cells were co-transfected with plasmids encoding luciferase reporters and constructs expressing full length or truncated HBx. Transactivation was then measured using the dual luciferase assay (see Materials and Methods). Expression of HBx^WT led to a modest but reproducible 1.5-fold transactivation of ENH1-Luc expression (Fig. 4C), a result consistent with other studies reporting a 2- to 4-fold transactivation of ENH1 luciferase by HBx (Cha et al., 2009; Tang et al., 2005). The truncation mutant HBx^43–154 transactivated ENH1 luciferase to the same level as full length HBx. However, when the DDB1-binding domain was removed from the carboxyl terminus (e.g., HBx^101–154) transactivation of ENH1-Luc was significantly reduced (Fig. 4C). HBx^1–101, which lacks the carboxyl half of HBx, also transactivated ENH1 to levels significantly less than HBx^WT. These results demonstrate that the DDB1-binding domain, and/or some other function(s) associated with that region of HBx was required for this transactivation function.

HBx localizes to the Cul4A-DDB1 complex

DDB1 is an adaptor protein for the Cul4A E3 ligase complex, which ubiquitinates protein substrates to promote their degradation (Angers et al., 2006; He et al., 2006; Higa et al., 2006; Shiyanov et al., 1999). A recent study by Li et al. demonstrated that HBx expressed as a GFP fusion protein (GFP-HBx) forms a complex with HA-DDB1 and myc-Cul4A (Li et al., 2010). We next determined whether HBx (without a GFP tag) similarly is found in the Cul4A-DDB1 complex. HBx was detected by IP/western blot from cells transfected with pSI-X (Fig. 5). Analysis of the same IP by western blot using anti-DDB1 demonstrated that endogenous DDB1 was also pulled down by anti-HBx (Fig. 5, lane 2), a result that confirms previous reports of the interaction between these two proteins (Lee et al., 1995; Sitterlin et al., 1997). In addition, Cul 4A was detected by anti-myc IP/western from cells transfected with a plasmid encoding myc-tagged Cul4A (Fig. 5, lane 3), and endogenous DDB1 was similarly detected by western blot, confirming reports that DDB1 binds to Cul4A (Angers et al., 2006; Shiyanov et al., 1999). Finally, cells co-transfected with plasmids expressing HBx and myc-Cul4A were analayzed by IP/western, which revealed myc-Cul4A co-precipitating endogenous DDB1 and HBx (Fig. 5, lane 4). Though small amounts of DDB1 were precipitated by the myc-specific antibody, presumedly through a complex formed between DDB1 and endogenous c-myc (Fig. 5, lane 1), co-transfection of myc-tagged Cul4 without or with HBx substantially increased the amount DDB1 precipitated by the anti-myc antibody (Fig. 5, lanes 3 and 4). We conclude that HBx can be found in the Cul4A-DDB1 complex, in agreement with the previous study that utilized GFP-HBx (Li et al., 2010).

Figure 5. HBx interacts with the Cul4A-DDB1 complex.

Representative co-IP/western blot for wild type HBx with the DDB1/Cul4A complex. HepG2 cells were co-transfected with the plasmids shown, and at 48 hr post-transfection cells were lysed for IP using the indicated antibody. Western blots for myc-Cul4A, DDB1, or HBx. * indicates non-specific background band.

DDB1 stabilizes HBx steady-state levels

The DDB1-Cul4A complex recruits substrate proteins for ubiquitination, which targets them for proteosome degradation. Although HBx can reside in the DDB1-Cul4A complex, there is only indirect evidence that DDB1 stabilizes HBx expression rather than promoting its degradation (Bergametti et al., 2002; Bontron et al., 2002). We used liver tissue from conditional DDB1 knockout mice to directly examine steady-state levels of HBx in the presence and absence of DDB1. Mice with floxed DDB1 alleles [DDB1^F/F (Yamaji et al., 2010)] were crossed with mice transgenic for HBx under control of the alpha-1-anti-trypsin regulatory region [ATX (Lee et al., 1990)] to create DDB1^F/F; ATX mice. We further crossed the DDB1^F/F;ATX with mice transgenic for the Cre recombinase under control of the liver-specific albumin promoter to promote excision of the DDB1 gene (DDB1^F/F;Alb-Cre^+/−, Fig. 6A) in the ATX mice. DDB1^F/F; ATX and DDB1^F/F; Alb-Cre^+/−; ATX mice were sacrificed and their livers removed to prepare lysates that were examined for DDB1 and HBx protein expression. Western blot analysis of liver lysates for DDB1 expression demonstrated similar DDB1 levels in ATX (Fig. 6B, lane 2) and DDB1^F/F;ATX livers (Fig. 6B, lanes 3 and 4). However, livers from DDB1^F/F;ATX mice that additionally expressed Cre (DDB1^F/F;Alb-Cre^+/−;ATX) showed a greater than 90% reduction of DDB1 steady-state levels, normalized to the tubulin loading control (Fig. 6B, lanes 5 and 6). Immunohistochemical analysis of paraffin-embedded liver tissue revealed DDB1-positive hepatocytes interspersed among DDB1-negative cells, indicating that the Cre-driven deletion of DDB1^F/F was not complete in these 6 week animals (Fig. 6C). The same liver lysates were then tested by IP/western blot for HBx. The supernatants from the first IP were saved and new anti-HBx antibody was added to recover any HBx not retrieved in the first IP. Subsequent western blot analysis for both rounds of IP revealed that levels of HBx were highest when DDB1 was present (Fig. 6D, lanes 2, 3 and 4), while a reduction in DDB1 levels was associated with significantly lower HBx levels (Fig. 6D, lanes 5 and 6). These results demonstrate that HBx steady-state levels are increased in the presence of DDB1, a result that suggests that HBx is stabilized by DDB1. These results further indicate that HBx is not a target for DDB1-Cul4A-mediated degradation, but may instead be functioning like a DCAF.

Figure 6. DDB1-binding stabilizes HBx expression.

(A) Schematic representation of the floxed DDB1 locus (DDB1F/F) before and after excision by the Cre recombinase protein, which is under the control of the Albumin promoter (Alb-Cre+/−). (B) Western blot analysis of DDB1 and tubulin loading control in liver lysates. Numbers below each lane indicate results of densitometer scanning for DDB1 normalized to tubulin for that lane. (C) Immunostaining for DDB1 protein of liver sections of 6-wk old mice. (D) Sequential IP/western blot detection HBx expression in liver lysates.

Discussion

The function(s) of HBx in virus replication is poorly understood, although a previous study indicates that HBx binding to cellular DDB1 is critical for the HBx-mediation potentiation of HBV replication (Leupin et al., 2005). The goal of the present study was to further investigate the contribution of the HBx-DDB1 interaction to HBV replication using plasmid-based replication assays in HepG2 cells and in hydrodynamically injected mice. The analysis of four different HBx point mutants firmly establishes that DDB1-binding to HBx is critical for maximal HBV replication (Fig. 1). While the HBx-DDB1 interaction is required, we now report that this binding was not sufficient for virus replication and that additional regions of HBx residing in the carboxyl termimus, but not in the amino terminus, are required. We show that the carboxyl portion of HBx is able to transactivate the HBV ENH1 Luceriferase reporter, and that removal of the DDB1-binding domain from the carboxyl region results in a loss of transactivation function. Thus, there is a strong correlation between the restoration of HBV replication and HBx transactivation of the HBV ENH1, suggesting these activities are mechanistically related in this assay. We demonstrate that HBx is present in the Cul4A-DDB1 E3 ligase complex, and that endogenous DDB1 is required to stabilize HBx in mouse liver. These results suggest a model in which HBx uses the interaction with DDB1 to redirect DDB1 function in order to benefit HBV replication.

Previous conclusions regarding the importance of HBx-DDB1 in HBV replication were derived from the analysis of the HBx^R96E point mutant protein that can no longer bind DDB1 (Leupin et al., 2005). It remained possible that the inability of HBx^R96E to restore HBx-deficient replication was due to the loss of an HBx function independent of DDB1-binding. Therefore, our finding that two additional DDB1-binding defective HBx mutant proteins are similarly unable to restore HBx-deficient replication firmly establishes the importance of the HBx-DDB1 interaction to HBV replication. The present study used HBx subtype adw2, whereas the study by Leupin et al. utilized HBx subtype ayw (Leupin et al., 2005). Since HBx subtypes differ by 5.8% in amino acid sequence (Keasler et al., 2007), certain residues within HBx^WT can be varied without affecting the protein’s ability to restore HBx-deficient replication. Our results also highlight the importance of other HBx residues that when mutated no longer bind DDB1 and are unable to support HBV replication. We note that the mutation in HBx⁶⁹ resides outside of the domain on HBx required for binding to DDB1. This mutant was created by converting a conserved cysteine (Kidd-Ljunggren et al., 1995) to a leucine at aa position 69 (Becker et al., 1998), a change predicted to alter the conformation of that HBx mutant protein. Other HBx point mutant proteins with changes in aa-68, -110, -139, and -139 that also reside outside the DDB1 binding domain have a similar loss of DDB1 binding activity (Sitterlin et al., 1997). The most likely explanation is that the mutant proteins have an altered conformation that interferes with the ability of HBx to bind DDB1. We conclude that the HBx interaction with DDB1 is critical for virus replication in HepG2 cells.

The analysis of overlapping HBx deletion mutants representing the entire HBx protein revealed that a function(s) residing in the carboxyl half of HBx is additionally critical for virus replication in HepG2 cells (Fig. 3). Interestingly, previous studies have identified transcription factor binding sites in the carboxyl terminus of HBx, including sites for CREB (Barnabas et al., 1997), TFIIB (Lin et al., 1997), RPB5 (Lin et al., 1997), CEBPα (Choi et al., 1999), and TBP (Qadri et al., 1995). Consistent with this, crystallography data predicts that the carboxyl portion of HBx projects out from the binding pocket on DDB1 (Li et al., 2010), which would make it available for interactions with transcription factors. Indeed, the importance of the HBx-DDB1 interaction in transactivation was first suggested by a study in which HBx mutants that did not bind DDB1 also did not transactivate an AP-1 reporter gene (Sitterlin et al., 1997). In addition, the HBx-DDB1 interaction was required for coactivation of HBV mRNA, although the HBV ENH1 was not specifically examined in that study (Leupin et al., 2005). Our present results further this notion by suggesting that both DDB1-binding and the carboxyl portion of HBx are required for transactivation of the HBV ENH1 in HepG2 cells (Fig. 4C). Consistent with this, other studies have demonstrated transactivation of HBV ENH1 luciferase by HBx^51–154 (XpLUC) (Tang et al., 2005) and HBV ENH1 CAT by HBx^57–148 (pHECx2CAT) (Murakami et al., 1994), although DDB1-binding was not examined in those studies. Finally, the HBx^R96E mutant used in the present study failed to transactivate HBV ENH1 luciferase expression, providing futher support of the importance of HBx-DDB1 binding to the transactivation of HBV ENH1 (Li et al., 2010).

While HBx transactivation of HBV ENH1 luciferase in the present study was modest, it was reproducible and in agreement with previous reports in which HBx expression caused a 2- to 4-fold increase in ENH1 activation (Cha et al., 2009; Tang et al., 2005). We note that low levels of transactivation in transient transfection assays may reflect the differentiated state of the cells. A comparison of HBV enhancer activity in transfected cells versus mouse liver revealed 100-fold higher levels of activity in vivo (Du et al., 2008). We conclude that the modest effect of HBx observed in cultured cells may represent a very significant effect in vivo. This idea is supported by our observation that viremia is reduced by 95–99% in mice hydrodynamically-injected with pHBVΔX compared to pHBV, while HBx-deficient replication is reduced by 50% in HepG2 cells [Figure 3, (Keasler et al., 2007)].

Available data suggest several possible models for HBx-DDB1 function in HBV replication. During replication, the HBV partially double-stranded DNA (~dsDNA) genome is converted, in the nucleus, to cccDNA, which serves as the template for the transcription of viral mRNAs [reviewed in (Seeger et al., 2007)]. The HBx-DDB1 cccDNA model (Fig. 7A) predict that DDB1-binding is required for HBx activation of viral transcription in the nucleus, and that the carboxyl portion of HBx is required for its interaction with cellular transcription factors. There is considerable support for this model. Although HBx localizes to both the nucleus and the cytoplasm, nuclear-localized HBx is responsible for maximal HBV replication (Keasler et al., 2009) and for transactivation of HBV ENH1 (Doria et al., 1995). It is now clear that HBx-DDB1 binding is required for maximal virus replication (Fig. 1) (Leupin et al., 2005; Li et al., 2010), and that HBx regulates the level of HBV mRNA (Keasler et al., 2007; Keasler et al., 2009; Melegari et al., 2005; Tang et al., 2005). The model further suggests that in addition to HBx-DDB1 binding, maximal HBV replication requires a function from the carboxyl portion of HBx (Fig. 3B) (Murakami et al., 1994; Tang et al., 2005). Indeed, previous studies have demonstrated that several transcription factors bind to the carboxyl portion of HBx (Choi et al., 1999; Lin et al., 1997). In addition, an X-ray crystallography study revealed that HBx-DDB1 binding results in the carboxyl region of HBx protruding outward from DDB1 where it is accessible to bind transcription factors (Li et al., 2010). This cccDNA model further predicts that HBx will localize to the cccDNA complex, and this was demonstrated using a cccDNA-specific chromatin immunoprecipitation (ChIP)-based assay (Belloni et al., 2009). It is generally accepted that HBx does not bind dsDNA directly (Rossner, 1992). However, DDB1 is a well-known DNA-binding protein (Abramic et al., 1991; Hwang and Chu, 1993). We propose that HBx recruitment to cccDNA could be facilitated by its interaction with DDB1. This model may help explain the promiscuity of HBx transactivation, if DDB1 scanning for damaged DNA brings HBx into contact with many different promoters.

Figure 7. Models for the role of HBx-DDB1 in virus replication.

(A) HBx-DDB1 cccDNA model. During HBV replication, the partially double-stranded DNA genome is repaired to form cccDNA. This model predicts that HBx (which does not bind DNA) is tethered to the cccDNA via DDB1, and that the carboxyl portion of HBx is needed to recruit transcription factors (TFs) for the transcription of viral mRNAs, as described in the text. (B) HBx-DDB1 Displacement/Recruitment model. DDB1 functions as an adaptor for the CUL4A E3 ubiquitin ligase complex, where it recruits DDB1-CUL4 Accessory Factors (DCAFs), which bind substrate proteins that are ubiquinated and degraded by the proteosome. This model proposes that during HBV replication, the HBx interaction with DDB1 may either displace one or more DCAFs to block the degradation of DCAF substrates that benefit HBV replication, and/or recruit other substrates either for protection from degradation (positive factor) or for degradation (negative factor).

Additional models for HBx-DDB1 function in virus replication, that are not mutually exclusive of the cccDNA model, arise from our demonstration that HBx can localize to the DDB1-Cul4A complex (Fig. 5). DDB1 is an adaptor protein for the Cul4A E3 ligase adaptor, which recruits DCAFs that bring in various substrates for ubiquitination and proteosome degradation (Angers et al., 2006; He et al., 2006; Higa et al., 2006; Shiyanov et al., 1999). Interestingly, the region of DDB1 to which HBx binds (Li et al., 2010) is shared by at least 79 DCAFs (He et al., 2006). Several lines of evidence suggest that HBx is a DCAF. First, it shares a DWD motif with the known DCAFs (Keasler and Slagle, 2008; Li et al., 2010). Second, HBx binding to DDB1 can displace the DCAF DDB2 (Bontron et al., 2002) and woodchuck hepatitis WHx binding to DDB1 can displace DCAF9 (Li et al., 2010). Finally, data present in the present study demonstrate that HBx is actually stabilized by DDB1, and unlikely to be a substrate of the DDB1 E3 ligase activity (Fig. 6). We conclude that HBx shares many features of DCAFs.

If HBx acts as a DCAF during HBV replication, we envision that it could act in either of two ways. First, HBx may act by simply displacing existing DDB1-associated DCAFs. In this DCAF Displacement Model (Fig. 7B), the DDB1-binding domain of HBx would be sufficient to displace other DCAFs and result in restoration of HBVΔX replication. However, we show several HBx truncation mutants that retain the ability to bind DDB1 but are unable to restore HBx-deficient replication (Figs. 2, 3, 4). A related model, designated the DCAF Recruitment Model, predicts that HBx behaves as a DCAF by binding DDB1 and recruiting a substrate(s) for degradation. HBx could recruit either a “negative factor” substrate whose degradation would benefit virus replication, or a “positive factor” substrate that needs to be protected from degradation. The latter possibility is consistent with HBx being stabilized by endogenous DDB1 (Fig. 6) and is supported by the observation that proteosome inhibitor MG132 restores HBx-deficient virus replication but has no effect on wildtype HBV replication (Zhang et al., 2004).

Several viruses encode proteins that bind DDB1-Cul4A, presumably as a mechanism to promote virus replication. The Paramyxovirus SV5 regulatory V protein binds DDB1 (Lin et al., 1998), redirecting the DDB1-Cul4A E3 ligase to ubiquitinate and degrade STAT1, thereby inactivating the host interferon antiviral response (Precious et al., 2005). The HIV1 regulatory Vpr protein also binds DDB1 and redirects Cul4A ligase activity to induce cell cycle arrest (Le Rouzic et al., 2007; Schrofelbauer et al., 2007). Murine gamma herpesvirus 68 encodes the M2 latency protein, which interacts with the Cul4A-DDB1^COP9 complex deregulating DNA repair by inhibiting repair signal transduction (Liang et al., 2006). Although there is evidence that HBx can inactivate the host interferon response (Jiang and Tang, 2011; Kumar et al., 2011; Wang et al., 2010; Wei et al., 2010), it remains to be determined whether this occurs in a DDB1-dependent manner. Similarly, HBx is reported to enhance the progression of G0 hepatocytes into G1, both in regenerating liver (Hodgson et al., 2008) and in primary hepatocytes in culture (Bouchard et al., 2001; Gearhart and Bouchard, 2010), although a role for DDB1 in this process has not been investigated. A correlation has been noted between HBx-DDB1 binding and the inhibition of DNA repair. Notably, HBx⁶⁹ and HBx^90/91, which do not bind DDB1, do not inhibit DNA repair (Becker et al., 1998). Additional studies are needed to establish the functions provided by the HBx-DDB1 interaction that contribute to virus replication.

Materials and Methods

Plasmids and cloning

A plasmid carrying a greater-than-unit-length (129%) HBV genome (payw1.2; subtype ayw) and the same plasmid with a stop codon at amino acid 7 of HBx that prevents expression of HBx (McClain et al., 2007) (payw1.2*7) were previously described (Melegari et al., 1998; Scaglioni et al., 1997), and are referred to here as pHBV and pHBVΔX, respectively. A plasmid encoding a heat-stable, secretable form of alkaline phosphatase (pSI-SEAP) was used as a negative control (Keasler et al., 2007). A plasmid encoding myc-tagged Cul4A has been described (Li et al., 2010). A reporter plasmid in which firefly luciferase is under control of the HBV Enhancer I (pXStNc-Luc), referred to in this study as ENH1-Luc, was obtained from Dr. Shinako Takada (MD Anderson Cancer Center, Houston, TX). A plasmid encoding Renilla luciferase, pSI-RLuc, was used for normalization and was described previously (Kumar et al., 2011). pSI plasmids encoding HBx point mutants included HBx⁷, HBx⁶⁹, and HBx^90/91 were described previously (Becker et al., 1998). The plasmid HBx^R96E (Lin-Marq et al., 2001) was a kind gift of Dr. Michel Strubin and was re-cloned into the same pSI vector. HBx truncations were created by PCR cloning from pSI-X [subtype adw2; (Keasler et al., 2007; Keasler et al., 2009)] into the Mlu I and Not I sites of the pSI vector (Promega). Truncation mutants included HBx^1–101, HBx^43–154, HBx^101–154, and HBx^55–101, with numbers indicating the amino acids present in the truncated protein. In addition, PCR amplification was used to add two N-terminal HA epitope-tags (YPYDVPDYA) (Helliwell et al., 2001) and a C-terminal hexahistidine tag to the HBx truncation constructs. A full-length HBx with a N-terminal HA-tag was generated similarly. Plasmid constructs were confirmed by DNA sequencing (Lone Star Labs, Houston, TX and Epoch LifeSciences, Missouri City, TX).

Cell culture and transfections

HepG2 cells obtained from the ATCC were maintained in Eagle’s media as described previously (Keasler et al., 2007) and used at an early passage. Cells were plated at 1 ×10⁵ cells per well in 6-well plates for replication assays, at 5 ×10⁴ cells per well in 24-well plates for luciferase assays, or at 5 ×10⁵ cells per 60-mm dish for western blots. Cells were allowed to adhere overnight before being transfected using TransIT-LT1 reagent (Mirus Bio Corporation) according to the manufacturer’s directions. For replication assays, 1 ×10⁶ untransfected HepG2 cells were added to each well 2 hr post-transfection in order to create a confluent monolayer needed for HBx-dependent replication (Keasler et al., 2007).

Purification of capsid-associated viral DNA

Capsid-associated DNA was extracted as described previously (Keasler et al., 2007; Keasler et al., 2009), with modifications as reported (Kumar et al., 2011). Three days after transfection, capsid-associated viral DNA was isolated from HepG2 cell lysates using the QIAamp MinElute Virus Spin kit (Qiagen).

Real-time PCR detection of HBV DNA

Capsid-associated DNA was quantitated using TaqMan real-time PCR, as previously described (Keasler et al., 2007; Keasler et al., 2009; Kumar et al., 2011). Triplicate samples were analyzed in duplicate PCR wells using 5 μl of isolated DNA. Copy number was determined from standard curve (10⁷-10⁰) as described (Keasler et al., 2007; Keasler et al., 2009). All samples were compared as a percentage of the copy number measured in cells transfected with wild type pHBV. Results were confirmed in three independent experiments.

Immunoprecipitation (IP) and western blot

To detect wildtype and mutant HBx, transfected HEK 293T cells from four 60-mm plates were pooled and lysed in 400 μL phosphate buffered saline (PBS) supplemented with 1% sodium dodecelsulfate (SDS) and 0.1 mg/mL DNase I. Samples were resolved by SDS polyacrylamide gel electrophoresis (PAGE) and transferred onto a nitrocellulose membrane (Schleicher & Schuell Bioscience). Membranes were fixed with 0.5% gluteraldehyde (Sigma-Aldrich) in PBS for 15 min, washed in PBS for 30 min, and then blocked with 5% non-fat dry milk in PBS (BLOTTO) for 15 minutes. Full-length HBx point mutant proteins were detected by a rabbit sera raised against HBx from HBV subtype adw2, diluted 1:1000 in 0.5% BLOTTO and 0.5% Tween 20 (BLOTTO-T). The HA-tagged HBx truncation proteins were detected using an anti-HA monoclonal antibody (Covance) diluted 1:1000 in 0.5% BLOTTO-T. Membranes were incubated in primary antibody overnight at 4°C, and then with alkaline phosphatase-conjugated secondary antibodies, diluted 1:3000 in 0.5% BLOTTO-T, for 2 hr at room temperature. Blots were developed with a colorimetric substrate [(50 mM Tris base, 3 mM MgCl₂, 0.1 mg/ml nitro-blue tetrazolium choloride (NBT) and 0.5 mg/mL 5-bromo-4-chloro-3′-indolyphosphate p-toluidine salt (BCIP)].

The myc-Cul4A complex was detected by first using IP with mouse anti-myc (Santa Cruz; 1:1000), and then separating proteins by 10% SDS-PAGE, and analyzing by western blot using the primary antibodies mouse anti-myc (Santa Cruz; 1:1000), mouse anti-DDB1 (Zymed; 1:1000), and rabbit anti-HBx (1:1000). All immunoblots of IP samples were detected using the SuperSignal® West Femto Substrate kit (Pierce).

Liver extracts were prepared by homogenizing liver in extraction buffer (50mM Tris-HCl, 100mM NaCl, 1mM EDTA, 1% NP-40, and 1% aprotinin). For the detection of DDB1, a total of 50 μg protein from each liver was separated by 10% SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was incubated with either rabbit anti-DDB1 (Zymed; 1:1000) or anti-tubulin (Santa Cruz; 1:1000). Membranes were then incubated with horseradish peroxidase conjugated secondary antibody (Pierce; 1:1000). HBx was detected by IP from 3 mg liver lysate using rabbit anti-HBx serum (1:1000). Sequential IP was done by incubating the remaining IP supernatant with additional rabbit anti-HBx serum (1:1000). Samples were separated by 15% SDS-PAGE, as previously described (Madden et al., 2000), and analyzed by western blot using the primary rabbit anti-HBx serum (1:1000) and secondary anti-rabbit antibody (Pierce; 1:1000). Bound antibody was detected using the SuperSignal West Femto kit (Pierce). To compare levels of DDB1, western blots for DDB1 and for tubulin were quantitated by densitometry (Molecular Dynamics), analyzed by ImageQuant5.2 software, and DDB1 was normalized to tubulin within the same sample.

Luciferase assay

HepG2 cells were cotransfected with ENH1-Luc, pSI-Rluc, and combinations of pSI-X, pSI-X mutants, or the pSI vector control, as indicated. Cells were harvested 24 hr post-transfection and lysed with Passive Lysis Buffer (Promega). Firefly and Renilla luciferase activities were determined using the Dual-Luciferase Reporter Assay (Promega) according to manufacturer’s directions. Readings were normalized to Renilla luciferase within a sample group. Activation of the luciferase construct in the absence of HBx was set to 1.0 and fold activation in the presence of HBx or HBx mutants was calculated.

Transgenic mice and hydrodynamic injections

Approval for all experiments involving mice was obtained from the Institutional Animal Care and Use Committee at Baylor College of Medicine. Mice harboring two floxed alleles of DDB1 (DDB1^F/F) (Yamaji et al., 2010) were crossed with ATX mice, which express HBx under control of the α-1-antitrypsin regulatory region (Lee et al., 1990). These double-transgenic mice were then crossed with mice transgenic for the Cre recombinase under the liver-specific control of the albumin promoter (Alb-Cre^+/−) on a DDB1^F/F background. Resulting progeny were genotyped by PCR analysis of high molecular weight DNA from tail clippings (Transnetyx).

Hydrodynamic injections were performed on outbred Crl:CD-1 (ICR) mice at 10–13 weeks of age as previously described (Keasler et al., 2007; Keasler et al., 2009). Briefly, mice were injected with plasmid DNA diluted in phosphate-buffered saline to a volume equivalent to 8% of the total body weight per animal. Each injected mouse received a total of 18 μg plasmid DNA: 9 μg of HBV plasmid DNA (either pHBV or pHBVΔX), 5 μg of plasmid DNA encoding heat-stable, secretable alkaline phosphatase (pSI-SEAP), and 4 μg of pSI-X, pSI-X^55–101, or control plasmid. Negative control animals received 18 μg plasmid DNA: 5 μg pSI-SEAP and 13 μg of promoterless pSEAP2-Basic plasmid DNA. Mice were sacrificed on day 4 post-injection at the peak of viremia (Keasler et al., 2007; Yang et al., 2002), and blood was collected and livers harvested. HBV copy number in the serum was determined by real time PCR quantitation of capsid-associated viral DNA, as previously described (Keasler et al., 2007; Keasler et al., 2009). 10μl serum was heated to 65°C for 30 min to inactivate endogenouse phosphatase, and then diluted 1:25 in saline for measurement of secretable alkaline phosphatase as a serum marker of transfection efficiency, as described (Keasler et al., 2007).

Immunostaining

Formalin fixed, paraffin-embedded mouse liver samples were examined for DDB1 expression. Deparaffinized liver sections were first treated with Target Retrieval Solution (Dako) at 121°C for 5 min. Following blocking steps, sections were incubated with primary anti-DDB1 (Bethyl Labs, 1:100), secondary antibody (goat anti-rabbit; Vector, 1:200), and Avidin-biotin complex (Vectastain Elite ABC; Vector).

Quantitation and Statistical analysis

All results were confirmed in at least three independent experiments. Statistical significance was determined using the Student’s T-test (Microsoft Excel software package). Error bars shown in figures indicate standard error of the mean, unless otherwise indicated. Statistical significance assigned for p< 0.05.

Conclusions

The present study examined the requirements of the HBx-DDB1 interaction in the context of HBV replication. After establishing that HBx-binding to DDB1 is critical for maximal HBV replication, we went on to show that HBx-DDB1 binding alone is not sufficient to restore HBx-deficient replication. Our results demonstrate that additional regions of HBx residing in the carboxyl portion are required for virus replication. While the plasmid assays used do not represent all steps in authentic virus infection, they do allow us to examine HBx functions in the context of HBV replication. Our results are further supported by previous studies showing the carboxyl portion of HBx contains transactivation function, and we additionally demonstrate that the region of HBx that spans the DDB1-binding domain is required for transcriptional activation of HBV ENH 1. Together, our study brings together the role of HBx as a transcriptional activator and as a DDB1-binding protein, and proposes testable models for role(s) of HBx-DDB1 in HBV replication. Although the precise function(s) of DDB1 usurped by HBx and its importance to virus replication remains to be identified, the observation that HBx binds to DDB1 may help to explain the diversity of functions proposed for HBx.

Highlights.

• We studied the requirement of HBx binding to DDB1 in HBV replication.

• HBx binding to DDB1 was required but was not sufficient for virus replication.

• HBx transactivation function from the COOH portion of HBx is also needed.

• The two functions must be provided on the same protein.