Unique Impact of RB Loss on Hepatic Proliferation

The retinoblastoma (RB) tumor suppressor pathway is disrupted at high frequency in hepatocellular carcinoma. However, the mechanisms through which RB modulates physiological responses in the liver remain poorly defined. Despite the well established role of RB in cell cycle control, the deletion of RB had no impact on the kinetics of cell cycle entry or the restoration of quiescence during the course of liver regeneration. Although these findings indicated compensatory effects from the RB-related proteins p107 and p130, even the dual deletion of RB with p107 or p130 failed to deregulate hepatic proliferation. Furthermore, although these findings suggested a modest role for the RB-pathway in the context of proliferative control, RB loss had striking effects on response to the genotoxic hepatocarcinogen diethylnitrosamine. With diethylnitrosamine, RB deletion resulted in inappropriate cell cycle entry that facilitated secondary genetic damage and further uncoupling of DNA replication with mitotic entry. Analysis of the mechanism underlying the differential impact of RB status on liver biology revealed that, while liver regeneration is associated with the conventional induction of cyclin D1 expression, the RB-dependent cell cycle entry, occurring with diethylnitrosamine treatment, was independent of cyclin D1 levels and associated with the specific induction of E2F1. Combined, these studies demonstrate that RB loss has disparate effects on the response to unique tumorigenic stresses, which is reflective of distinct mechanisms of cell cycle entry.

The retinoblastoma (RB) tumor suppressor pathway is disrupted at high frequency in hepatocellular carcinoma. However, the mechanisms through which RB modulates physiological responses in the liver remain poorly defined. Despite the well established role of RB in cell cycle control, the deletion of RB had no impact on the kinetics of cell cycle entry or the restoration of quiescence during the course of liver regeneration. Although these findings indicated compensatory effects from the RB-related proteins p107 and p130, even the dual deletion of RB with p107 or p130 failed to deregulate hepatic proliferation. Furthermore, although these findings suggested a modest role for the RB-pathway in the context of proliferative control, RB loss had striking effects on response to the genotoxic hepatocarcinogen diethylnitrosamine. With diethylnitrosamine, RB deletion resulted in inappropriate cell cycle entry that facilitated secondary genetic damage and further uncoupling of DNA replication with mitotic entry. Analysis of the mechanism underlying the differential impact of RB status on liver biology revealed that, while liver regeneration is associated with the conventional induction of cyclin D1 expression, the RB-dependent cell cycle entry, occurring with diethylnitrosamine treatment, was independent of cyclin D1 levels and associated with the specific induction of E2F1. Combined, these studies demonstrate that RB loss has disparate effects on the response to unique tumorigenic stresses, which is reflective of distinct mechanisms of cell cycle entry.
Liver cancer is a major worldwide health concern. More than 500,000 cases are diagnosed throughout the world annually, and nearly 680,000 deaths were attributed to liver cancer in 2007. In addition, there is a significant mortality rate associated with the disease, with a 5-year survival rate of only 11% in developed countries (1)(2)(3). Furthermore, there are few successful interventions for advanced disease, and while new therapeutics are being deployed in the clinic, their overall efficacy has been limited. Most cases of liver cancer can be traced to specific etiological events (2,4). For example, the majority of liver cancer cases in Eastern Asia are associated with hepatitis B infection, whereas in Western Europe, hepatitis C infection represents a likely primary causal event (2)(3)(4)(5). In addition to viral infection, it is appreciated that certain environmental agents can predispose to the development of liver cancer. Such toxins can represent genotoxic agents such as Aflatoxin B1 or vinylchloride exposures (2,4). Conversely, agents that are not primarily associated with DNA damage have also been implicated in liver tumorigenesis. Such nongenotoxic carcinogens can lead to chronic liver damage, steatosis, or deregulation of hepatic proliferation (4,6). The causal impact of these agents has been directly addressed utilizing animal models. Critically, the progression of liver cancer likely represents a complex relationship between environmental stresses and genetic events associated with tumor development (2,3,7,8).
Loss of the retinoblastoma (RB) tumor suppressor pathway represents a relatively common event in liver cancer (2,7,8). Specifically, the region of the Rb locus (13q14) is subject to loss of heterozygosity at relatively high frequency in liver cancer and histological loss of RB protein expression occurs in a subset of tumors (8 -11). RB can be functionally inactivated through a variety of mechanisms, including deregulated phosphorylation and direct sequestration by oncoproteins. In liver cancer, cyclin D1 deregulation or loss of the p16ink4a tumor suppressor is hypothesized to contribute to RB inactivation (11)(12)(13). In addition, the oncoprotein gankyrin and specific hepatitis virus-encoded proteins (e.g. NS5B and NS5A) have been shown to promote the functional inactivation of RB (3,12,14,15). Thus, through these combined mechanisms, it has been postulated that the loss of RB function is a relatively common event in liver cancer and has been associated with tumor development in addition to overall poor prognosis of the disease.
RB functions as a cell cycle regulatory factor to modulate proliferation in response to mitogenic and anti-mitogenic signals (16 -18). In quiescent cells, RB is hypophosphorylated and assembles protein complexes that repress the activity of genes regulated by the E2F family of transcription factors. Mitogenic signaling induces the expression of cyclin D1, which leads to the activation of CDK4/6 complexes that mediate RB phosphorylation. This phosphorylation disrupts the association of RB with E2F proteins, thereby stimulating the expression of E2F target genes that encompass a wide range of factors critical for progression through S-phase and mitosis (19). In contrast with such mitogenic signaling, agents that inhibit proliferation lead to the dephosphorylation of RB and the repression of E2F-mediated gene expression. Thus, RB is a critical integrator of proliferative and anti-proliferative stresses.
Typically, RB loss is viewed as promoting unchecked proliferation in tissue and contributing specifically to the initiation of tumor development. However, studies in multiple tumor tis-sues and mouse models have indicated that the functional consequence of RB loss is strongly modulated by both the genetic and tissue context (16,17,20,21). In part, this context specificity is mediated by the RB-related proteins p107 and p130 that can compensate for RB loss and serve to restrict E2F activity. In the liver, deletion of RB leads to a relatively modest effect on cell cycle control but demonstrates a significant effect on tumor susceptibility (22). Here, we explored the requirement for RB function in multiple facets of liver biology and define both critical RB-independent and -dependent aspects of liver physiology that are of profound importance in considering the mechanisms through which RB loss predisposes to tumor development.

MATERIALS AND METHODS
Animals-The generation and genotyping of the Rb f/f and Rb f/f ;albcre ϩ mice have been discussed previously (22). To directly address the function of the other pocket protein members, RB/p107 and RB/p130, double-null mice were created. All experiments were conducted in accordance with of the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by Thomas Jefferson University.
Animal Treatments-8-week-old Rb f/f and Rb f/f ;albcre ϩ and double-null RB/p107 or RB/p130 littermates were administered a single interperitoneal injection of a 10% solution of carbon tetrachloride (CCl 4 ) dissolved in corn oil (Sigma). Mice were then sacrificed at 0, 24, 48, 60, and 168 h post-CCl 4 for Rb f/f and Rb f/f ;albcre ϩ mice. Double-null RB/p107 or p130 mice were sacrificed 0, 48, and 168 h post-treatment. For acute diethylnitrosamine (DEN) 3 (Sigma) studies, 14-day-old littermates were given a single intraperitoneal injection, dissolved in a 0.9% saline at a dose of 20 mg/kg body weight. Mice were then euthanized at 24 and 72 h post-DEN injection. All mice were given a single intraperitoneal injection of 150 mg/kg 5-bromo-2-deoxyuridine (BrdUrd; Sigma) dissolved in 0.9% saline 1 h before sacrifice to label S-phase hepatocytes.
PCR Genotyping and Detection of Recombination-DNA extraction and genotyping was preformed as described previously for Rb f/f and Rb f/f ;albcre ϩ mice (22). Primers used to detect p107 and p130 were commercially available. Genomic PCR steps were configured to manufactures protocol. Extraction of RNA and real-time PCR analyses were performed using standard approaches.
Statistical Analysis and Quantitation Data-Slides were scored in a blind fashion for at least 600 hepatocytes from several different fields for each staining. For histological assessment, a veterinary pathologist scored slides in a blind fashion. Statistical significance between groups was calculated by using a two-tailed unpaired t test. Differences were classified as statistically significant at p Ͻ 0.05.

RESULTS
Based on the central role of the RB tumor suppressor in cell cycle control, it was anticipated that RB loss would have a significant impact on the kinetics of liver regeneration. To directly investigate the impact of RB loss on regeneration, we utilized Rb f/f and Rb f/f ;albcre ϩ mouse strains (22). These mice harbor liver-specific deletion at the Rb locus, which can be clearly visualized in liver tissue by genomic PCR. Furthermore, RB protein levels were ablated in livers of Rb f/f ;albcre ϩ mice, as assessed by immunoprecipitation analyses (Fig. 1A). In this context, CCl 4 administration leads to massive liver damage that precipitates a highly synchronous regenerative process to replace liver mass within 168 h (24,25). Mice were treated and then sacrificed at 0, 24, 48, 60, and 168 h. Livers were weighed, and the relative liver weight was expressed as a percentage of total body weight. The regenerative process is associated with a transient increase in liver weight. While there was no significant difference in weight between RBproficient and -deficient livers treated with CCl 4 , the RBdeficient livers tended toward larger weight (Fig. 1A). Such a finding suggested that RB might have a relevant impact on liver proliferation in response to CCl 4 -induced damage. Furthermore, qualitative histological examination of the livers showed that there was minimal difference in the kinetics of repair to liver injury. Specifically, at 24 h both Rb f/f and Rb f/f ; albcre ϩ livers exhibit necrotic regions in proximity to portal veins, whereas at 48 h, necrotic regions encompass ϳ50% of liver area; however by 168 h, a complete recovery was observed in both Rb f/f and Rb f/f ;albcre ϩ livers (Fig. 1B). Kinetic analyses of DNA synthesis, as determined by BrdUrd incorporation, indicated that there was an equivalent entry of RB-proficient and -deficient cells into S-phase. Moreover,

Effect of RB Loss on Hepatic Proliferation under Stresses
the kinetics of re-entry into quiescence was virtually identical (Fig. 1C). Under these conditions, mitotic figures were readily apparent at 48 h, and mitotic progression was further noted (Fig. 1D). Again, RB status had minimal effect on these markers during regeneration. Together, these data indicate that regeneration as a result of CCl 4 -induced damage in the mouse liver occurs independently of RB status as well as the re-establishment of quiescence.
In conjunction with these findings, we investigated the control of RB/E2F target genes and possible compensatory mechanisms that would mask the effect of RB deficiency. In untreated RB-proficient and -deficient livers, E2F target genes (MCM7 and PCNA) are largely undetectable at the RNA and protein level (Fig. 2, A and B, lanes 1, 2, 7, and 8). This finding indicates that the adult liver is largely quiescent and that compensatory mechanisms are mediating repression of these genes in the absence of RB. At 48 h post-CCl 4 administration, there is an induction of cyclin D1 expression and CDK2 activity (Fig.  2B, lanes 3, 4, 9, and 10). In relation to these findings, RBrelated proteins were analyzed. It is believed that p130 is a key mediator of compensation with RB loss; however, in this context p130 RNA and protein levels remain relatively constant. In contrast, p107 RNA and protein levels accumulate and dissipate similar to other E2F target genes (Fig. 2, C and D, lanes 3, 4,  9, and 10). This finding is consistent with the established regulation of p107 expression via E2F (26). Together, these data suggest that modulation of proliferation in this context is not dependent on RB status; rather, cyclin D1 accumulation in conjunction with CDK2 suppress the activity of all pocket proteins inducing proliferation, with attenuation of this CDK activity and p107/p130 mediating cell cycle exit.
In multiple systems, the deletion of p107 and/or p130 will synergize with RB loss to deregulate cell cycle control (27). Therefore, to uncover potential roles of the other pocket proteins (p107 and p130) in the re-establishment of quiescence, mice harboring the combined loss of Rb/p107 and Rb/p130 were created. As shown in Fig. 3, A and B, these mice exhibited p107 or p130 deletion in addition to liver-specific RB deletion, as noted by Cre recombinase. At 8 weeks of age, mice were administered CCl 4 to induce liver injury. Immediately after sacrifice, livers were excised and weighed. No apparent difference in liver weight at any time point was detected between the double-null mice, and all mouse livers had returned to control weights by 168 h (data not shown). Additionally, induction of DNA synthesis was analyzed by BrdUrd incorporation. At 48 h, an induction of BrdUrd incorporation was observed regardless of pocket protein status; however, by 168 h, BrdUrd levels had returned to control levels in all conditions (Fig. 3C). In conjunction with the establishment of quiescence, RB/E2F target gene expression returned to control levels by 168 h post-injection as well (Fig. 3D). Combined, these data suggest that loss of p107 or p130 in conjunction with RB ablation does not influence the regeneration process in the mouse liver. Thus, highly redundant mechanisms can constrain E2F activity in the liver and mediate quiescence in vivo.
RB has been implicated in mediating DNA damage checkpoints in cell culture (28 -30) and has been shown to function as an important inhibitor of DEN-mediated liver tumorigenesis (23). To investigate the general effect of RB loss on the response of the liver to the genotoxic hepatocarcinogens, the acute response of DEN was monitored. 14-Day-old mice received a single intraperitoneal injection of DEN and were sacrificed at 24 and 72 h post-injection. Immediately after sacrifice, liver weight was determined and expressed as a percentage of total body weight. DEN exposure had little overall influence on liver weight in RB-proficient and -deficient mice (data not shown). Furthermore, histological analyses revealed little qualitative difference between conditions (data not shown). Because there is proliferation occurring in neonatal liver tissue, potential impacts on checkpoint responses in liver tissue were identified by BrdUrd incorporation. Interestingly, RB-proficient livers exhibited minimal BrdUrd incorporation (ϳ3%) in either the absence or presence of DEN. However, RB-deficient livers displayed a greater than 4-fold increase in BrdUrd incorporation (Fig. 4A). At 72 h post-DEN, BrdUrd levels in all conditions had decreased to the levels of control animals. Because these effects on BrdUrd incorporation could reflect repair synthesis associated with DNA damage, Ki67 staining was performed. These data showed that consistent with the BrdUrd incorporation, an aberrant proliferative response was occurring (Fig. 4A). In keeping with these findings, the induction of a number of RB/E2F target genes was also observed (Fig. 4C, compare lanes  3 and 4 with 7 and 8). To determine whether the liver cells were capable of progressing into mitosis, phosphohistone H3 serine 10 reactivity was examined. Consistent with such genotoxic agents inducing a "checkpoint," the percent of phosphohistone H3 serine 10-positive cells in RB-proficient livers was diminished; however, RBdeficient livers exhibited an increase in positive cells (Fig. 4D). Interestingly, in contrast with CCl 4 treatment, we observed a total lack of mitotic figures in DEN-treated livers sections irrespective of RB status. In conjunction with these findings, livers deficient for RB and DEN treated while accumulating CDK1 show a lack of cyclin B1 induction (Fig. 4, C and D). Thus, RB deficiency yields a unique impact on genotoxic carcinogens by facilitating cell cycle entry yet yields a failure to produce a productive mitosis, due at least in part to the lack of cyclin B1 accumulation. It has been shown that checkpoint deregulation can elicit additional secondary damage (e.g. strand breaks) driven by inappropriate DNA replication. Thus, ␥H2AX reactivity was determined by immunofluorescence analyses of tissues sections and immunoblot analyses of liver extracts, to determine the impact of RB status on this marker of DNA strand breaks. These analyses demonstrated that RB-deficient livers exhibited a significant increased in ␥H2AX versus RB-proficient littermates (Fig. 4B). Combined, these findings indicate that loss of RB has a profound impact on the cell cycle response to the genotoxic hepatocarcinogen DEN that influences DNA damage burden. To define how RB status impacts the response to diverse stresses, a detailed analyses of the mechanisms through which DEN and CCl 4 impinge on cell cycle control was determined. In mice exposed to CCl 4 , cell cycle entry was associated with the induction cyclin D1 levels, the corresponding loss of p130, and accumulation of E2F target genes (Fig. 5A, lanes 5, 6, 11, and  12). These facets of signaling occurred irrespective of RB status and are consistent with the well proscribed mechanisms of cell cycle control. In contrast, DEN-treated livers displayed a response that was highly dependent on RB status. Although DEN treatment had little impact on the levels of cyclin D1 or p130, there was a significant increase in E2F1 protein specifically in RB-deficient livers (Fig. 5A, compare lanes 3 and 4 with  9 and 10). Correspondingly, the induction of E2F target gene expression was specifically stimulated in RB-deficient livers FIGURE 3. p107 and p130 status has no effect on regeneration in the presence of RB loss. A, p130 status was determined by PCR analysis in addition to Rb ablation noted by Cre recombinase. B, p107 status was determined by PCR analysis in addition to Rb ablation noted by Cre recombinase. C, BrdUrd incorporation was detected in both Rb/p107-and Rb/p130-null livers, counted, and then expressed as a percent of total hepatocytes. D, liver tissue samples of equal total protein were separated by SDS-PAGE. Levels of RB/E2F targets (MCM7 and PCNA) were then analyzed in both RB/p130 and RB/p107 wild type (WT) and null conditions, compared with a CCl 4 positive control. Nuclear lysates of RB/p130 wild type and null conditions are on top and RB/p107 wild type and null conditions are below, in both cases Lamin B was used as a loading control. KO, knock-out. All error bars represent the S.D. between experimental samples. treated with DEN (Fig. 5B, compare lanes 3 and 4 with 9 and  10). Thus, these data indicate that discrete mechanisms are engaged by DNA damage versus regeneration signals to mediate cell cycle entry in the liver.

DISCUSSION
It is well established that RB status impacts the interplay between mitogenic and anti-mitogenic signals (16,17,29). Thus, it was suspected that loss of RB would yield highly significant effects on all facets of proliferative signaling in liver tissue. However, there was a striking specificity to the impact of RB status in response to exposure to diverse agents, suggesting that loss of RB may be particularly relevant in the context of specific etiological agents for hepatocellular carcinoma.
The capacity and appropriate regulation of liver regenerative processes is a critical facet in the liver and is associated with liver failure in the face of chronic damage. Numerous studies have suggested that the RB pathway plays a crucial role in regulating the kinetics of liver regeneration and the re-establishment of quiescence (31,32). However, the results described here clearly show that loss of the RB gene has a relatively modest effect on the regeneration process induced by CCl 4 . Particularly, the kinetics of entry and exit from the cell cycle are indistinguishable with loss of RB. Presumably, this observation is due to the potent mitogenic effect of cyclin D1 to drive the cell cycle irrespective of RB status or the presence of other pocket proteins. In addition, it has been previously shown that deficiency of cyclin D1 abrogates the proliferation response following exposure to hepatic mitogens (33). Interestingly, because quiescence occurred regardless of RB status, it suggests those mechanisms that restrain proliferation in the liver are also RBindependent. While p107 or p130 deletion can illustrate that these factors compensate for RB loss, the dual loss of RB/p107 or RB/p130 did not deregulate these cell cycle transitions. These findings suggest that either a single pocket protein can mediate quiescence or that alternative mechanisms are suppressing cell cycle progression in these models. Unfortunately, the simultaneous ablation of p107 and p130 in the mouse leads to neonatal lethality precluding the analyses of complete loss of pocket protein function (27). However, such findings suggest that the specific role of RB in liver cancer is likely not associated with the suppression of aberrant proliferation.
In contrast with regeneration, it is hypothesized that the initiation of liver tumorigenesis with genotoxic carcinogens is dependent on the mutations induced by such agents (2,4,8). It is well established from cell culture models that loss of RB can affect DNA damage responses leading to an uncoupling of S-phase checkpoints (28,29). For the in vivo analyses, we utilized the potent hepatocarcinogen DEN. This exposure yielded relatively modest induction of DNA damage (e.g. ␥H2AX), and we were incapable of detecting ATM activation in this model. However, we found that DEN exhibits diametrically distinct effects in RB-proficient versus RB-deficient liver tissue. In RB-deficient cells DEN has a paradoxical mitogenic activity. This effect of DEN is not due to solicitation of a regenerative response, as RB loss does not lead to enhanced cell death following DEN exposure. Furthermore, the resultant DEN-induced cell cycle is both aberrant and associated with molecular events distinct from regeneration. Specifically, the cell cycle initiated by DEN exposure in RB-deficient livers does not progress to a productive mitosis. This finding, suggests that certain checkpoints are still operative in the absence of RB, as indicated by the failure to accumulate cyclin B1. The fate of these cells that fail to progress into mitosis remains uncertain, but such hepatocytes could harbor replication induced mutations that contribute to tumorigenesis. The mechanism through which DEN leads to cell cycle entry is distinct from a regenerative response as clearly evidenced by the failure to induce cyclin D1. Rather, DEN treatment results in the stimulation of E2F1 expression, which is most profound in the RB-deficient liver. Studies from multiple laboratories have shown that DNA damage can induce FIGURE 4. RB loss deregulates the response to the genotoxic carcinogen DEN. A, BrdUrd levels were detected at 24 h post-treatment, counted, and then expressed as a percentage of BrdUrd-positive hepatocytes. Ki67 levels were also monitored. Positive hepatocytes for Ki67 at 24-hour post-treatment were expressed. B, levels of ␥H2AX was detected in all conditions and expressed as a percentage of total positive hepatocytes. Representative images were taken of ␥H2AX IF staining in DEN-treated conditions. In addition, liver tissue samples of equal total protein were separated by SDS-PAGE, and levels of ␥H2AX were analyzed. Two samples were present per condition, and lamin B was utilized as a loading control. C, equal total protein was separated by SDS-PAGE. Levels of CDK1, MCM7, PCNA, and cyclin A were analyzed. Two samples were present per condition, and lamin B was utilized as a loading control. D, formalin fixed paraffin embedded sections were stained for phosphohistone H3 serine 10 and counted. Plot shows total positive hepatocytes for phosphohistone H3 serine 10 as a percentage of total counted hepatocytes. Images at 24 h post-treatment were taken to show typical mitotic figures. In addition, liver tissue samples of equal total protein were separated by SDS-PAGE and levels of cyclin B1 were analyzed. Two samples were present per condition, and lamin B was utilized as a loading control. E2F1 by stabilizing the protein (34,35). Our findings suggest that RB-status is critical for limiting the accumulation of this protein, and limiting the induction of E2F-mediated transcription following DNA damage. While additional studies will be required to specifically define the requirement of E2F1, our findings demonstrate that in contrast with regenerative signals, DNA damage engages a very specific and novel pathway that is critically dependent on RB.
Combined, these studies illustrate the complexity and context dependence through which RB elaborates cell cycle control in the liver. Furthermore, these analyses reinforce the concept that discrete extrinsic signals will be modified by distinct genetic events in specifying the basis for cell cycle deregulation and tumor susceptibly.