Premature Cell Cycle Entry Induced by Hepatitis B Virus Regulatory HBx Protein During Compensatory Liver Regeneration

Abstract

The cycles of cell death and compensatory regeneration that occur during chronic hepatitis B virus (HBV) infection are central to viral pathogenesis and are a risk factor for the development of liver cancer. The HBV genome encodes one regulatory protein, HBx, which is required for virus replication, although its precise role in replication and pathogenesis is unclear. Since HBx can induce the G0/G1 transition in cultured cells, the purpose of this study was to examine the effect of HBx during liver regeneration. Transgenic mice expressing HBx (ATX) and their wild type (WT) littermates were used in the partial hepatectomy (PH) model for compensatory regeneration. Liver tissues collected from ATX and WT mice at varying sacrifice time points after PH were examined for markers of cell cycle progression. When compared to WT liver tissues, ATX livers had evidence of premature cell cycle entry as assessed by several parameters (BrdU incorporation, PCNA- and mitotic-indices, and reduced steady-state p21 protein levels). However, HBx did not affect apoptosis, glycogen storage, or PH-induced steatosis. Together, these results demonstrate that HBx expression can induce cell cycle progression within the regenerating liver. Our data are consistent with a model in which HBx expression contributes to liver disease and cancer formation by affecting early steps in liver regeneration.

Introduction

Chronic infection with hepatitis B virus (HBV) affects over 350 million people worldwide and is a major risk factor for hepatocellular carcinoma (HCC) (1, 2). The HBV genome encodes one regulatory protein, HBx, which is required for virus replication in vivo (3-5). While the precise role of HBx in virus replication and pathogenesis is unclear, HBx likely has multiple functions in these processes. Studies in transfected cells revealed that HBx can transactivate viral and cellular promoters (6, 7), can activate cytoplasmic signaling pathways (8, 9), can be either pro- or anti-apoptotic [reviewed in (10)], can inhibit cellular DNA repair (11), and can behave as a tumor promoter in transgenic mice (12, 13) [reviewed in (14)].

Cycles of immune-mediated cell death and compensatory regeneration that accompany chronic HBV infection are thought to contribute to the development of liver disease (15), but the role of HBx in this process is unknown. Previous studies have demonstrated that in immortalized cells in culture, HBx induces cell cycle progression to overcome a serum-induced G0 block (16) and causes accelerated transit through cell cycle checkpoints (17) [reviewed in (18)]. If HBx is similarly able to alter cell cycle progression during compensatory regeneration, it could be contributing to viral pathogenesis over the decades of chronic hepatitis.

The purpose of our study was to examine the effect of HBx in vivo during the early steps of liver regeneration. We used transgenic mice expressing HBx (ATX) and their wild type (WT) littermates in the partial hepatectomy (PH) model for compensatory regeneration. In this model, surgical removal of 70% of the liver leads to carefully regulated signaling events causing remaining hepatocytes to undergo synchronized growth and cell division [reviewed in (19-21)]. Based on studies of HBx in cell culture (16, 17), we hypothesized that HBx expression would interfere with the regulation of early steps in liver regeneration. This function of HBx could contribute to the development of liver disease, particularly if it occurred in the context of exposure to environmental carcinogens.

Materials and Methods

Transgenic mice

Animal procedures were done in accordance with federal regulations for the care and treatment of laboratory animals. Transgenic mice expressing HBx under the control of the liver-specific α-1 anti-trypsin regulatory region (ATX mice) (22) and their non-transgenic (WT) littermates were used in this study. Genotypes were determined by PCR analysis of high molecular weight tail DNA using X-specific primers as described (23) and confirmed by immunoprecipitation (IP) and Western blot as described below.

Partial hepatectomy

All PH surgeries were performed in the morning hours following a 4 hour fast using 10-12 week old ATX and WT mice as previously described (24). Animal body weight was determined at both the time of PH and at sacrifice, and liver removed at PH and remnant liver masses were weighed. Representative portions of liver tissues were either flash-frozen in 2-methylbutane, or fixed in neutral buffered formalin and paraffin-embedded for histological analysis.

BrdU incorporation and detection

Mice were either injected with 0.1 mg 5-bromodeoxyuridine (BrdU) per gram of body weight 2 hours before sacrifice, or given 1 mg/ml BrdU in their drinking water for 60 hours before sacrifice. Fixed liver sections were stained with anti-BrdU antibody (Dako, 1:100) using the ARK kit (Dako). Liver BrdU labeling index was calculated for each animal in this study as the percent of BrdU-positive cells in 10 random fields each containing approximately 300 hepatocytes.

Histological analyses and IHC

Paraffin embedded liver sections were stained with hematoxylin and eosin (H&E) for histological analysis and for quantitation of mitotic cells. Mitotic cells were observed in high-powered fields (400X) and included those with condensed chromosomes, and those in late prophase, metaphase, or anaphase. Mitotic index was calculated as the percentage of mitotic cells in 10 random fields each containing approximately 100 hepatocytes for 3 ATX and 3 WT livers at each of six sacrifice time points. To examine cellular glycogen content, deparaffinized liver sections were treated with 0.5% periodic acid and stained with Schiff's reagent (PAS). To measure apoptosis, liver sections were stained by terminal deoxynucleotidyl transferase dUTP nick end labeling assay (TUNEL) using the Apop Tag Kit (Millipore) according to the manufacturer's protocol. TUNEL positive cells were quantitated in 10 random fields each containing approximately 300 hepatocytes for 3 ATX and 3 WT livers at each of four sacrifice time points. To optimize detection of proliferating cell nuclear antigen (PCNA), liver sections were treated with heat-induced epitope retrieval (HIER; Sigma). Following blocking steps, tissues were incubated with primary anti-PCNA (FL-261; Santa Cruz Biotechnology, 1:5000), secondary antibody (goat anti-rabbit; Vector, 1:200), and Avidin-biotin complex (ABC; Vector). PCNA labeling index was calculated in livers 0 hr and 24 hr after PH by quantitating positive cells in 10 random fields each containing approximately 300 hepatocytes for 3 ATX and 3 WT livers. Using the HIER system, a basal level of PCNA was detected in 0 hr samples and this was subtracted from each sample data point during analysis as previously described (25).

For the identification of neutral lipids, frozen tissue sections were stained with Oil Red O (ORO) dye. Severity of steatosis in liver sections was scored using the method of Brunt et al. (26) for 3 ATX and 3 WT mice at 0 hr and 28 hr after PH. Briefly, coded ORO-stained liver sections from each animal were scored for the percentage of hepatocytes that contained ORO-positive cytoplasmic vacuoles in 10 random fields per animal as described previously (23).

Detection of HBx and p21 by IP/Western blot and Western blot

Liver extracts from 2 to 3 animals of each genotype per sacrifice time point were prepared by homogenizing frozen liver tissue in extraction buffer containing 1% NP-40 and 1% aprotinin as described (13). HBx was detected by IP/Western blot as described (13). Western blot detection of p21 and tubulin control was by primary antibodies anti-p21 (sc-6246, Santa Cruz; 1:500), and anti-tubulin (sc-9104, Santa Cruz; 1:1000). For all Western blots, bound antibody was detected and quantitated as previously described (5). For each sacrifice time point examined, the normalized WT p21 value was set to 1.00 and compared to the normalized ATX p21 value.

Reverse transcription real time PCR

RNA was extracted from 10 mg liver tissue using the RNAqueous-4PCR kit (Applied Biosystems) and 2 μg RNA were reverse transcribed using the high capacity cDNA reverse transcription kit (Applied Biosystems). Real time PCR analysis was done using approximately 1 μg cDNA, Taqman Universal PCR Master Mix, and primer/probe sets for Serpina1a (Mm02748477_g1), p21 (Mm01303209_m1), 18S rRNA (4319413E), or HBx primers (5'-CCGGTCCGTGTGCACTTC-3', 5'-AGGATCTGATGGGCGTTCAC-3') and probe (5'-6FAM ACCTCTGCACGTTGCATGGAGACC-3') all from Applied Biosystems. Values for Serpina1a, p21, and HBx were normalized to 18S rRNA values. Copy number was calculated using 10 ^{-[(sample Ct - Y-intercept) / slope]}, where, based on optimal reaction efficiencies, the Y-intercept was assumed to be 38 and the slope assumed to be -3.3 as described by Applied Biosystems.

Statistical Analysis

The Student's t test was used to compare ATX vs. WT mice for the following: liver mass, BrdU, mitosis, PCNA labeling indices, RT-qPCR, and TUNEL, using the Microsoft Excel software package, and significance assigned for p < 0.05. Error bars are reported as SEM or SD, as indicated in figure legends. Mann Whitney rank sum test for significance (for ORO) was done using the SigmaStat software package.

Results

ATX transgenic mice and partial hepatectomy

ATX mice express HBx under the control of the liver-specific alpha-1-antitrypsin (A1AT) regulatory region (22). HBx expression continues throughout the lifespan of these mice at levels similar to that observed during chronic infection, as discussed previously (13). There is no observable pathology in the livers of untreated ATX or WT mice. However, ATX mice develop more tumors (vs. WT littermates) when exposed to carcinogenic insults such as diethylnitrosamine (13, 27) and hepatitis C virus (23).

In cell culture, HBx expression has been shown to induce G0/G1 progression (16, 17). To address whether HBx might similarly affect cell cycle progression in vivo, we performed 70% partial hepatectomies (PH) on ATX mice and their WT littermates. Seventy-seven mice were included in the initial study (Supplemental Table 1) and total body weight did not differ between 37 ATX and 40 WT mice at the time of surgery, indicating that HBx expression did not affect animal size (Fig 1A). Equivalent amounts of liver tissue were removed from animals of each genotype and represented approximately 70% of the total mass as predicted from body weight (Fig 1B). Finally, HBx protein expression was confirmed by IP/Western blot using livers collected at 0 hours (hr) and in regenerating livers (Fig 1C). These data demonstrate that any differences observed during regeneration will be due to the presence of HBx rather than to issues related to the surgery or animal physiology.

Figure 1.

Characteristics of mice used in the present study. A, body mass of mice (40 WT, 37 ATX) was determined prior to PH. B, liver mass removed during surgery was measured for the same WT and ATX mice. Data represent mean values mean ± SD. NS indicates not significant as determined by Student's t test. C, HBx was detected by IP/Western blot analysis of liver extracts as described in Materials and Methods. Top panel, representative results from 2 WT and 2 ATX livers removed at surgery. Bottom panel, representative results from 2 WT and 2 ATX livers harvested 40 hr post-PH.

Altered DNA synthesis following PH in ATX mice

Following PH, the liver progresses through a well described series of steps from priming, which prepares hepatocytes to move from G0 to G1 phase of the cell cycle, to DNA synthesis, and finally to cell division [reviewed in (19-21)]. In cell culture, HBx induces the G0/G1 transition (16, 17), so we examined the effect of HBx on the early stages of liver regeneration. To measure DNA synthesis, mice were injected with BrdU 2 hr prior to each of 8 sacrifice time points. BrdU incorporation was detected by antibody staining of fixed liver tissue and was observed as dark brown nuclei (Fig 2A). Calculation of the BrdU labeling index in WT livers revealed an increasing percentage of BrdU-positive cells beginning at 32 hr post-PH with a synchronous peak in DNA synthesis 40 hr after PH (Fig 2B). This result is in agreement with previously published data for non-transgenic mice (28, 29). In contrast, the BrdU labeling index was more complex in ATX livers. First, we observed a population of ATX hepatocytes that entered S phase earlier than WT hepatocytes (Fig 2B; 28 hr, p<0.02). In addition, a second peak in BrdU incorporation in ATX livers occurred between 40 and 44 hr, corresponding with the single 40 hr peak observed in WT livers (Fig 2B). These results demonstrate that HBx expression alters early steps in liver regeneration.

Figure 2.

Evidence of early DNA synthesis in ATX livers. BrdU that incorporated into newly synthesized DNA was detected by IHC staining. A, representative liver sections from 0, 28, and 40 hr after PH (200x). B, mean BrdU labeling indices from 3 to 7 mice per genotype per time point (see Supplemental Table 1). Error bars represent SEM; asterisk (*) denotes time points at which p<0.05, determined using the Student's t test.

Total hepatocyte division not affected by HBx

The 2-hr BrdU labeling method commonly used to measure DNA synthesis following PH identifies those hepatocytes undergoing DNA synthesis during a 2-hr window between BrdU administration and sacrifice (28, 30-33). By summing all the hepatocytes labeled by this method, it appears that the total number of hepatocytes that have undergone DNA synthesis in ATX and WT livers by 60 hr after PH is similar. To confirm this, we performed PH and administered BrdU continuously for 60 hr so the total number of hepatocytes undergoing DNA synthesis could be measured. An additional nine mice (3 ATX, 6 WT) were subjected to PH and given BrdU in their drinking water. Animals were sacrificed at 60 hr, BrdU incorporation was detected by antibody staining of fixed liver sections (Fig 3A), and the labeling index was determined for each animal. Using this approach, equivalent percentages of BrdU-positive hepatocytes were detected in both ATX and WT livers (p = 0.41), indicating similar levels of DNA synthesis were achieved in both groups by 60 hr post-PH (Fig 3B). Finally, we sought to confirm that ATX mice survived beyond 60 hr. Surgeries were performed on another eight mice (3 ATX, 5 WT), and animals were sacrificed at 8 days after PH. Approximately 100% of the liver mass removed at surgery had regenerated by 8 days in both WT and ATX mice (Fig 3C). This result is consistent with the 60-hr BrdU labeling data and indicates that, despite the effects of HBx on early regeneration (Fig 2B), the ATX mice were able to successfully regenerate their livers.

Figure 3.

Total regeneration is not affected by HBx. Mice were subjected to either sham operation (2 WT, 1 ATX) or PH (6 WT, 3 ATX). For 60 hr following surgery, mice were provided continuous access to drinking water containing BrdU. Fixed liver tissues collected at 60 hr were stained with a BrdU antibody and the BrdU labeling index determined for each animal. A, representative IHC staining after 60-hr labeling (200x). B, BrdU labeling index for 60-hr labeling after PH. C, percentage of removed liver mass that regenerated after 8 days for 5 WT and 3 ATX mice. Data shown are the mean value for each genotype ± SD; NS, not significant, as determined by the Student's t test.

Possible mechanisms for early cell cycle progression in ATX hepatocytes

Deregulated liver regeneration is hypothesized to precede the development of liver disease, so the observation of the complex pattern of DNA synthesis in ATX livers (Fig 2B) was intriguing. There are several possible explanations for this result, and we first considered the possibility that there are two populations of ATX hepatocytes.

The idea of two populations of ATX hepatocytes leads to several testable predictions. For example, the sub-population of ATX hepatocytes that enters S phase prematurely should also display markers of G1/S phase earlier than WT hepatocytes. This idea was tested by monitoring expression of PCNA, a commonly used marker for late G1/S phase. Previous PH studies in non-transgenic mice indicate PCNA protein expression begins to increase from baseline 24 hr post-PH, as the cells begin to progress from late G1 into S phase (25, 34, 35). Therefore, we measured the PCNA labeling index in 3 ATX and 3 WT livers. PCNA was detected in WT livers at 24 hr (Fig 4A), indicating that some hepatocytes are in late G1/S phase. At the same time point in ATX livers, the percentage of cells expressing detectable PCNA was significantly higher than that of WT livers (p<0.05). This result is consistent with the 2-hr BrdU data showing a subpopulation of ATX hepatocytes had entered S phase earlier than WT (Fig 2B).

Figure 4.

PCNA- and mitotic-indices in ATX (vs. WT) livers. A, increased percentage of PCNA-positive hepatocytes in ATX (versus WT) livers. Formalin-fixed liver tissue was stained with antibodies for PCNA and the total percentage of PCNA positive hepatocytes. Data shown are the average ± SEM; significance was determined by the Student's t test and * indicates p<0.05. B, formalin-fixed liver sections from WT and ATX mice were stained by H&E. Mitotic bodies indicated by white arrows. Magnification is 400x. C, mean mitotic indices for 3 WT livers (white bars) and 3 ATX livers (black bars). Data represent mean values ± SEM; significance determined by the Student's t test is shown as either * p ≤0.05, or ** p < 0.00001.

If ATX hepatocytes prematurely entering S phase continued through the cell cycle, we would expect to see a corresponding subpopulation of ATX hepatocytes reach mitosis sooner than those in WT livers. The possible early appearance of mitotic cells in ATX livers was examined by determining the mitotic index in fixed livers tissues at 6 sacrifice time points after PH (Fig 4B). Data from 3 WT livers revealed a peak in mitosis 48 hr post-PH (Fig 4C), as has been reported previously in PH models of non-transgenic mice (30, 36). In contrast, data from 3 ATX livers revealed a more complex pattern. There were significant increases in the number of mitotic cells in ATX livers at 40 and 44 hr relative to that found in WT livers at the same time points (Fig 4C). The peak in mitotic cells in the ATX livers occurred at 44 hr, about 4 hr earlier than the WT peak at 48 hr. When the time to progress through S and G2 phase into M phase is taken into consideration, the early appearance of mitotic cells in the ATX livers is consistent both with the early onset S phase subpopulation revealed in the 2-hr BrdU labeling experiment and with the increased PCNA detection in ATX livers at 24 hr. The mitosis data extends those results by demonstrating the subpopulation of ATX hepatocytes that enters S phase prematurely continues through the cell cycle.

We next investigated potential molecular mechanisms that might explain a premature entry into S phase. One important cell cycle regulator at the G1 phase checkpoint is the cyclin-dependent kinase inhibitor p21. Liver regeneration studies in p21 knock-out mice revealed an accelerated G1 and early onset of S phase evidenced by early increases in BrdU and PCNA after PH (31, 33). The effect of HBx on p21 has been studied in transfected cells, where HBx can repress (37), activate (38), or both repress and activate (39) p21. We considered the possibility that HBx expressed in ATX livers would deregulate p21 expression, leading to premature entry into S phase. Since p21 can be regulated at the transcriptional [reviewed in (40)] and post-translational levels (31, 41), we examined p21 mRNA by reverse transcription real time PCR (RT-qPCR) and p21 protein by Western blot. There were no significant differences in the p21 mRNA levels in ATX and WT livers at any time point examined (Fig 5A). In contrast, we found a trend towards lower p21 protein levels in ATX (vs. WT) livers (Fig 5B). The lack of statistical significance may be due either to variability among animals or that we are measuring an effect on p21 that occurs only in a subpopulation of hepatocytes. The generally decreased p21 levels are consistent with the observations of early increases in BrdU incorporation and PCNA levels in ATX (vs. WT) livers. Together, these results support the observation that HBx expression leads to a premature onset of cell cycle entry during liver regeneration. These data also raise the possibility that HBx can deregulate p21 at a post-transcriptional level.

Figure 5.

p21 mRNA and protein levels in ATX (vs. WT) livers. A, levels of p21 mRNA were measured by RT-qPCR and normalized to 18S rRNA expression. Each data point represents the average of 3 WT livers (white bars) and 3 ATX livers (black bars). Error bars represent SD. NS indicates not significant as determined by Student's t test. B, liver lysates from ATX and WT livers were examined by Western blot for p21 and tubulin. Each band, representing a different animal, was quantitated by densitometer scanning and p21 levels were normalized to tubulin levels. For each time point, the average of normalized WT samples was set to 1.00 and compared to ATX samples.

Other possible mechanisms for deregulated regeneration in ATX hepatocytes

We also considered other functions previously reported for HBx that might influence liver regeneration. These included possible effects of HBx on apoptosis [reviewed in (10)], glycogen storage (30), and fat accumulation (23, 30), as well as effects due to the presence of the ATX transgene. We began by investigating a possible effect of HBx on apoptosis.

Under certain conditions, HBx expression can induce and/or prevent apoptosis both in cell culture and in vivo [reviewed in (10)]. We considered that increased cell death in the ATX livers might contribute to the complex pattern of DNA synthesis (Fig 2B). To determine a possible effect of HBx on apoptosis, fixed liver sections were stained by TUNEL assay and TUNEL positive cells were quantitated in ATX and WT livers (3 mice per genotype per time point) at 0, 24, 28, 32, and 36 hr after PH. Apoptotic cells were rare (fewer than 0.06%), with no genotype-specific differences in the incidence of TUNEL positive cells (p>0.19, Supplemental Table 2). Therefore, we concluded that apoptosis was not increased in ATX (vs. WT) livers and likely does not explain the complex pattern of DNA synthesis (Fig 2B).

The ATX mice used in this study express HBx under the control of the liver-specific A1AT regulatory region. Murine A1AT (Serpina1a) is an acute phase response (APR) protein expected to be activated in response to PH (42). We considered that the presence of the human A1AT regulatory element in the ATX transgene might alter the activation of endogenous A1AT and disrupt compensatory regeneration. To address this, RT-qPCR was used to quantitate Serpina1a levels in 3 ATX and 3 WT livers at both 0 hr and 36 hr after PH. There were no differences between ATX and WT livers in the levels of endogenous Serpina1a in either resting liver (0 hr, p=0.087) or regenerating liver (36 hr, p=0.398) (Fig 6A). We also compared the Serpina1a levels in ATX livers at 0 hr vs. 36 hr and found no significant effect of HBx (p=0.677), indicating that the presence of the human A1AT regulatory element did not affect endogenous Serpina1a mRNA levels during regeneration.

Figure 6.

HBx transgene expression is stable after PH. A, levels of Serpina1a (A1AT) mRNA at 0 h and 36 h post-PH were measured by RT-qPCR and normalized to 18S rRNA expression. Each data point represents the average of 3 liver sections per genotype. Error bars represent SD. B, HBx mRNA levels at 0 h and 36 h after PH were quantitated by RT-qPCR and normalized to 18S rRNA expression. Each data point represents the average of 3 liver sections per genotype. Error bars represent SD. NS indicates not significant as determined by Student's t test. C, immunoprecipitation/Western blot detection of HBx expression in liver tissue at 0 hr and 36 hr after PH. Each lane is from an individual liver sample.

Although endogenous Serpina1a levels were not affected by the presence of the ATX transgene, activation of APR genes following PH might enhance HBx expression from the A1AT-driven transgene, and high levels of HBx could interfere with compensatory regeneration. A similar approach was used to evaluate whether HBx expression levels increase following PH. First, RT-qPCR analysis was used to measure HBx mRNA expression in WT and ATX livers at 0 hr and 36 hr after PH. HBx expression was detected exclusively in ATX livers, and there was no significant change in HBx mRNA expression in regenerating liver compared to that of resting liver (Fig 6B; p=0.77). Analysis by IP/Western showed there was no reproducible difference in HBx protein levels in the resting versus regenerating ATX livers (Fig 6C). We conclude that HBx protein and RNA levels expressed from the ATX transgene remained constant during liver regeneration.

In response to PH, glycogen is mobilized from the remnant liver to produce enough glucose to maintain homeostasis in the body. This mobilization transiently depletes the liver's glycogen content, which is restored to normal levels at later time points during regeneration. A recent study in another lineage of HBx transgenic mice concluded that severely impaired glycogen storage contributed to inhibited regeneration after PH (30). To examine this possibility in our model, fixed liver sections collected at 0, 24, and 48 hr after PH were examined for glycogen content by treating with periodic acid and staining with Schiff's reagent (PAS). At 0 hr, liver samples stained PAS-positive for glycogen with no obvious difference in staining patterns between ATX and WT (Supplemental Fig 1). By 24 hr, decreased PAS-positive staining intensity in livers of both genotypes compared to that of 0 hr livers indicated that glycogen had been mobilized from the hepatocytes. At 48 hr, glycogen stores had increased to similar levels in both ATX and WT hepatocytes. The PAS staining patterns we observed for WT and ATX livers were similar to that reported previously for non-transgenic livers (30); therefore, we concluded that glycogen storage was not impaired in the ATX livers.

Low levels of hepatic fat accumulation (i.e., steatosis) are normal during the first 24 hr after PH (30, 43). HBx expression can increase the severity of existing steatosis under certain experimental conditions (23, 30), and high levels of steatosis are an independent risk factor that can interfere with the liver's ability to regenerate (29, 34, 43). To examine a possible effect of HBx on PH-induced steatosis, frozen liver sections collected at 0 and 28 hr after PH were stained with oil red O (ORO) and scored for steatosis using the method described by Brunt et al. (26). No ORO positive cells were observed in 0 hr WT or ATX livers. There was no significant difference in steatosis severity between 3 ATX and 3 WT livers at 28 hr (p=0.10, Supplemental Fig 2). These results indicate that fat did not accumulate in the regenerating ATX livers to levels greater than that in WT livers, and is therefore an unlikely contributor to the complex pattern of DNA synthesis revealed in the 2-hr BrdU labeling experiment (Fig 2B).

Discussion

Chronic HBV infection is marked by cycles of cell death and regeneration, and is a risk factor for the development of HCC. In this study, the effect of HBx expression on the early steps of liver regeneration was examined using ATX and WT mice in the 70% PH model. We predicted that HBx might alter PH induced liver regeneration, given its reported ability to induce cell cycle progression in cultured cells [(16, 17), reviewed in (18)]. Our results indicate that HBx affects early steps in liver regeneration, leading to a complex pattern of DNA synthesis. Liver tissues collected at varying sacrifice time points after PH showed evidence of premature cell cycle entry as assessed by several parameters (BrdU incorporation, PCNA- and mitotic-index, and reduced steady-state p21 levels). There was no effect of HBx, however, on apoptosis, glycogen storage, or PH-induced steatosis. Together, these results demonstrate that HBx expression has the ability to induce cell cycle progression within the milieu of the regenerating liver.

Since the liver has many redundant pathways to facilitate regeneration (44), it is not surprising that the effects of HBx on regeneration were subtle. However, the expression of HBx in the current study caused a modest effect of premature cell cycle entry that was reproducible among animals of a given sacrifice time point. While no single assay yielded a striking effect of HBx, it is significant that the effect of HBx in each assay consistently supported our conclusion that HBx induced premature cell cycle entry. Our conclusions are also consistent with the results from cell culture experiments indicating HBx can induce proliferation in cultured cells (16, 17).

The results of our study are in contrast to two other studies which concluded that HBx inhibited hepatocyte regeneration (30, 45). It is important to note that our analysis focused on time points within 48 hr post-PH, revealing effects of HBx that were missed in the design of those other studies (30, 45). In addition, differences in genetic background of the mice used may be important. For example, the HBx transgenic mice used in the study by Wu, et al. were on a C57BL/6 genetic background, and the WT mice in that study had steatosis that was worsened by HBx expression (30). It is known that steatosis alone may interfere with liver regeneration (29, 43), providing an explanation for why the mice in that study did not survive past 72 hr. In contrast, the outbred ICR background of the ATX mice showed mild PH-induced steatosis that was not worsened by HBx expression (Supplemental Fig 2). A direct comparison of our study with the original report that HBx inhibits regeneration (45) is difficult because that study involved HBx hepatocytes transplanted into SCID mice, and PH performed on 30-day old HBx transgenic mice (compared to 10-12 week old mice used in the present study). Finally, the HBx lineage studied by Wu et al. developed HCC (30), whereas the ATX mice used in our study do not develop spontaneous tumors (22) indicating some other underlying differences between these transgenic mouse models.

Our results suggest that HBx may act by driving a subpopulation of hepatocytes to prematurely enter the cell cycle. It is uncertain why all ATX hepatocytes do not behave in a synchronous fashion, but several possibilities may be considered. It is possible that HBx expression levels vary from cell-to-cell, and a threshold level of HBx is required to overcome a restriction point. Consistent with this idea, limited IHC staining of ATX livers detected HBx in 26% of hepatocytes, with HBx-positive cells not restricted to a particular zone of the liver (data not shown). Another consideration is that not all hepatocytes are equal, and indeed hepatocyte function varies by location in the liver (46). Therefore, it is possible that HBx affects some hepatocytes but not others. This possibility is difficult to address experimentally, but we do note that BrdU-positive hepatocytes were distributed evenly throughout the liver, similarly to that of the BrdU-positive WT hepatocytes. Finally, it is intriguing to consider the possibility that HBx can prime G0 hepatocytes to move into G1, since ATX hepatocytes began entry into S phase (and M phase) about 4 hr earlier than hepatocytes from WT mice, consistent with the time needed to prime the liver after PH [4 hr, (20)].

There are several reasons why we believe the effect of HBx on cell cycle progression during liver regeneration is biologically relevant. Normal hepatocytes are typically non-dividing, and would have time to repair DNA adducts that form after exposure to carcinogens. However, if HBx-expressing cells are primed to enter the cell cycle, they may have less time to repair these lesions, leading to the formation of DNA mutations. This idea is supported by data showing the increased DNA mutation frequency in ATX livers after exposure to liver carcinogens (13, 47).

It is known that hepatocyte grown factor (HGF) and cytokines interleukin (IL)-6, and TNFα are necessary for the normal proliferative response during PH-induced liver regeneration [reviewed in (19-21)]. HGF is considered an initiator of liver regeneration [reviewed in (19-21)]. IL-6 acts by enhancing the mitogen-activated protein kinase (MAPK) signaling cascade, and TNFα participates through its ability to induce IL-6 [reviewed in (19-21)]. Although the possible effect of HBx on HGF, IL-6, and TNFα was not addressed in our study, it is established that HBx can activate MAPK signal transduction pathways [reviewed in (10, 14)], and can upregulate TNFα production in transfected hepatocytes (48). The idea that HBx might enhance IL-6-mediated cell proliferation is consistent with our observation that a subset of ATX hepatocytes enters S phase prematurely. HBx can also sensitize HepG2 cells to TNFα-mediated apoptotic killing (49); however, no effect of HBx on apoptosis was observed in our study.

In summary, the PH model allows a unique opportunity to examine the effect of HBx on hepatocyte regeneration in the setting of the intact liver. Although this approach was not used in the context of HBV replication, it may still provide clues about liver pathology during chronic infection, which is characterized by immune-mediated cycles of cell death and regeneration. We recognize that compensatory regeneration after 70% PH may differ from regeneration of individual virus-infected cells that are killed by the immune system. However, in the latter scenario we are unable to experimentally address a role for HBx. We hypothesize that the ability of HBx to alter cell cycle progression makes the cellular environment more favorable for HBV replication, but it is unclear at which exact step in replication HBx has a role. The results of our study indicate that HBx alters early steps in liver regeneration, causing hepatocytes to enter the cell cycle prematurely. Over decades of chronic HBV infection, these alterations could contribute to pathogenesis, especially in the presence of liver carcinogens.