The liver exhibits an exquisitely controlled cell cycle, wherein hepatocytes are maintained in quiescence until stimulated to proliferate. The retinoblastoma tumor suppressor, pRB, plays a central role in proliferative control by inhibiting inappropriate cell cycle entry. In many cases, liver cancer arises due to aberrant cycles of proliferation, and correspondingly, pRB is functionally inactivated in the majority of hepatocellular carcinomas. Therefore, to determine how pRB loss may provide conditions permissive for deregulated hepatocyte proliferation, we investigated the consequence of somatic pRB inactivation in murine liver. We show that liver-specific pRB loss results in E2F target gene deregulation and elevated cell cycle progression during post-natal growth. However, in adult livers, E2F targets are repressed and hepatocytes become quiescent independent of pRB, suggesting that other factors may compensate for pRB loss. Therefore, to probe the consequences of acute pRB inactivation in livers of adult mice, we gave adenoviral-Cre by i.v. injection. We show that acute pRB loss is sufficient to elicit E2F target gene expression and cell cycle entry in adult liver, demonstrating a critical role for pRB in maintaining hepatocyte quiescence. Finally, we show that liver-specific pRB loss results in the development of nuclear pleomorphism associated with elevated ploidy that is evident in adult mice harboring both acute and chronic pRB loss. Together, these results show the crucial role played by pRB in maintaining hepatocyte quiescence and ploidy in adult liver in vivo and underscore the critical importance of delineating the consequences of acute pRB loss in adult animals.
- retinoblastoma tumor suppressor
- conditional knockout
- cell cycle control
The retinoblastoma tumor suppressor (Rb) plays a critical role in cellular proliferation control and inhibition of oncogenic transformation ( 1, 2). Germ line mutations in the Rb gene predispose individuals to bilateral retinoblastoma as well as osteosarcoma ( 3, 4). Somatic Rb inactivation contributes to the development of these tumors as well as to tumors in a number of other organs (e.g., bladder, lung, breast, and prostate; refs. 5– 8). Furthermore, inactivation of the Rb pathway, either by direct mutation of Rb or deregulation of upstream regulators of pRB, such as p16INK4a or cyclin D1, is believed to be an obligatory step in sporadic tumor formation that renders cells hyperproliferative and unresponsive to antimitogenic signals ( 9).
Rb encodes a nuclear phosphoprotein (pRB) that negatively regulates G1-S phase cell cycle progression. It is believed that the antiproliferative function of pRB is attributable to its role as a transcriptional corepressor. In early G1, pRB is hypophosphorylated and binds to members of the E2F family of transcription factors and recruits corepressors to actively attenuate the transcription of E2F-regulated genes ( 10). The E2F family of transcription factors are critical regulators of numerous genes including those important for cell cycle progression (e.g., cyclin E and cyclin A) and DNA replication [e.g., proliferating cell nuclear antigen (PCNA), MCM5, and MCM7]. In response to mitogenic signaling, the activities of cyclin D-CDK4/6 and cyclin E-CDK2 are induced, resulting in phosphorylation of pRB and disruption of pRB assembled repressor complexes. Consequently, E2F-responsive genes are subsequently expressed and cells progress into S phase ( 10, 11). Conversely, antimitogenic signaling results in diminished cyclin-CDK activities, maintenance of pRB in its hypophosphorylated/active form, and inhibition of cell cycle progression ( 12).
As the Rb pathway is targeted at high frequency in human cancer, there is considerable interest in understanding the consequences of pRB loss. The biochemical effects of pRB loss have been studied extensively in cell culture model systems. Analysis of mouse embryonic fibroblasts from mice harboring germ line Rb loss has shown that the expression of a number of pRB-E2F target genes is deregulated ( 13, 14). Furthermore, Rb−/− MEFs exhibit deregulated cell cycle kinetics ( 13) and are unable to elicit cell cycle arrest following exposure to a variety of antimitogenic signals (e.g., DNA damage and transforming growth factor-β; refs. 15– 17). However, recent studies have suggested that Rb harbors additional activities that are associated with the maintenance of quiescence/senescence and are apparent only when Rb is inactivated acutely. These aspects of Rb function are masked when pRB is chronically absent through functional compensation by the pRB-related pocket protein p107 ( 18).
The complex consequences of pRB loss in cell culture are mirrored by studies analyzing mice with targeted Rb deletion. Mice heterozygous for Rb are viable and develop pituitary and thyroid tumors early in life associated with loss of heterozygosity ( 19– 21). Inactivation of both Rb alleles in mice results in unscheduled cell proliferation, apoptosis, and widespread developmental defects, leading to embryonic death by day 14.5 ( 20, 22, 23). Whereas many of these events were suspected to be cell autonomous, it has subsequently been shown that many of the phenotypes arise due to defective extraembryonic development in Rb−/− embryos ( 24, 25). Additionally, in the Rb germ line–deficient animals, there is clear evidence for functional compensation by the other pocket proteins ( 26– 28). To circumvent some of the difficulties associated with development and compensation, conditional models of pRB loss in specific organs have been generated ( 29– 34). In these models, a spectrum of phenotypes associated with both full and heterozygous inactivation of Rb has been described.
The liver is an organ that is maintained in a quiescent state in adults but harbors remarkable proliferative capacity regulated by incompletely understood pathways ( 35, 36). Following acute or chronic liver damage, quiescent hepatocytes enter the cell cycle and divide to restore the functional capacity of the liver. Precise control over hepatocyte proliferation is critical for the suppression of tumorigenesis in the liver, as chronic liver damage and corresponding cycles of regeneration are known to fuel the development of hepatocellular carcinoma ( 37). There is clear evidence that CDK activities that function upstream (e.g., cyclin D) and downstream (e.g., cyclin E and cyclin A) of pRB are involved in regulating hepatocyte proliferation ( 38– 40). Correspondingly, the pRB pathway is disrupted at high frequency in human hepatocellular carcinoma ( 41– 43.
Despite evidence that pRB plays a central role in the regulation of hepatocyte proliferation and inhibition of hepatocellular carcinoma, the biological consequences of pRB loss in the liver have yet to be challenged. Therefore, here we investigated the effect of somatic pRB inactivation in murine liver. Liver-specific pRB ablation results in E2F target gene deregulation and elevated cell cycle progression during post-natal growth. However, compensation eventually occurs such that in adult livers, the expression of E2F targets is repressed and hepatocytes become quiescent independently of pRB. Furthermore, we show that acute pRB elimination in the livers of adult mice is sufficient to elicit E2F target gene expression and cell cycle entry. Finally, we show that liver-specific pRB loss results in rapid and dramatic pleomorphism with nuclear enlargement and elevated ploidy. Together, these results reveal the critical role played by pRB in maintaining hepatocyte quiescence and ploidy in vivo and underscore the critical importance of delineating the consequences of acute pRB loss in adult animals.
Materials and Methods
Mice. Mice harboring Rb alleles in which exon 19 is flanked by loxP sites (Rbf/f; ref. 44) were obtained from the National Cancer Institute. Albumin-Cre mice ( 45), hemizygous for the Alb-Cre transgene, were purchased from The Jackson Laboratory (Bar Harbor, ME). To generate liver-specific Rb conditional knockouts, Rbf/f and Alb-Cre mice were intercrossed to obtain mice homozygous for the floxed Rb locus and hemizygous for Alb-Cre (Rbf/f;Alb-Cre+). These mice were then interbred with Rbf/f mice to produce Rbf/f and Rbf/f;Alb-cre+ littermates at 1:1 ratio. Rbf/f;Alb-Cre+ mice were also intercrossed with nontransgenic FVB/N mice to generate Rbwt/f and Rbwt/f;Alb-Cre+ littermates at 1:1 ratio. Mice were housed in a pathogen-free animal facility under standard 12-hour light/12-hour dark cycle with ad libitum water and chow. All of the experiments were conducted using the highest standards for humane care in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the University of Cincinnati Institutional Animal Care and Use Committee.
Mouse treatments and tissue harvesting. All experiments were done using littermates. Bromodeoxyuridine (BrdUrd, 150 mg/kg, dissolved in 0.9% saline) was given by i.p. injection 1 hour before sacrifice. Livers were rapidly excised from euthanized mice, weighed, and the left lobe was fixed in 10% buffered formalin for 16 hours for histologic analysis. Remaining liver tissue was immediately frozen in liquid N2 and stored at −80°C until use. Liver imprints (touch preparations) were prepared by excising the liver and gently touching the sliced edge of the left lobe to a glass slide. This method allows single hepatocytes to attach in a nondisruptive manner so that their nuclei remain intact. Fixation and staining of touch preparations for centrosomes and centromeric DNA was then done as described below. For adenoviral injections, male mice were anesthetized with pentobarbital (60 mg/kg, i.p.) before i.v. delivery of 1 × 109 plaque-forming units (pfu) adenovirus via intrapenile injection. Adenovirus expressing cre-recombinase (Ad-CMV-Cre) was purchased from the University of Iowa Gene Transfer Core Facility (Iowa, IA). Control adenovirus expressing LacZ from the same promoter (Ad-CMV-LacZ) was obtained from the Vector Core Laboratory, University of North Carolina. Statistical analyses were done using unpaired two-tailed t test. Ps < 0.05 were considered significant.
PCR genotyping and detection of recombination. Genomic DNA was isolated by standard phenol/chloroform extraction. Primers used for detection of the Alb-Cre transgene were 5′-GCGGTCTGGCAGTAAAAACTATC-3′ (sense) and 5′-GTGAAACAGCATTGCTGTCACTT-3′ (antisense). To analyze Rb genotype and to detect the presence of Cre-mediated recombination, PCR analysis of Rb exon 19 was done ( 44, 46). Primers for Rb18 (5′- GGCGTGTGCATCAATG-3′) and Rb19E (5′-CTCAAGAGCTCAGACTCATGG-3′) yielded 283- and 235-bp products for the unrecombined-floxed and wild-type alleles, respectively. Cre-mediated recombination was detected using primers Rb18 and Rb212 (5′-GAAAGGAAAGTCAGGGACATTGGG-3′), yielding products of 746 and 260 bp for the unrecombined and recombined-floxed Rb alleles respectively. Amplification was done using 1 unit Taq polymerase (Promega, Madison, WI) and annealing temperature of 58°C for 35 cycles. Total liver RNA was extracted with Trizol (Invitrogen, Carlsbad, CA) and first-strand cDNA was synthesized from 1 μg of RNA using the Superscript II reverse transcription-PCR (RT-PCR) system (Invitrogen) following the manufacturer's recommendations. cDNA was subject to amplification using primers Rb exon 18 (5′-CCTTGAACCTGCTTGTCCTC-3′) and Rb exon 20 (5′-GAAGGCGTGCACAGAGTGTA-3′) at 54°C for 35 cycles.
Preparation of liver nuclear extracts and Western blotting. Frozen liver was dissociated in buffer A [25 mmol/L Tris (pH 7.5), 50 mmol/L KCl, 2 mM MgCl2, 1 mmol/L EDTA, and 1 mmol/L phenylmethylsulfonyl fluoride (PMSF)] and incubated for 10 minutes on ice. Nuclei were then released using a Dounce homogenizer and tight pestle. Following two washes in buffer A, nuclei were extracted on ice for 10 minutes with radioimmunoprecipitation assay buffer [150 mmol/L NaCl, 1.0% NP40, 0.5% deoxycholate, 0.1% SDS, 50 mmol/L Tris (pH 8.0), 50 mmol/L NaF, 13 mg/mL β-glycerophosphate, 120 μg/mL sodium vanadate supplemented with 1 mmol/L PMSF and protease inhibitors]. Lysates were centrifuged at 13,000 rpm and total nuclear protein concentration determined by Bio-Rad protein assay. Equal amounts of total nuclear protein were separated by PAGE and transferred to polyvinylidene difluoride membranes. Specific proteins were detected by standard Western blotting procedures. pRB was detected as previously described ( 47). Monoclonal antibodies to Brg1 (G-7), MCM7 (141.2), and PCNA (PC10) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal antibody to MCM5 (clone 33) was from BD Transduction Laboratories (San Diego, CA). Polyclonal antibodies recognizing cyclin E (M-20), p107 (C-18), and p130 (C-20) were from Santa Cruz Biotechnology.
Bromodeoxyuridine labeling and immunohistochemistry. Sections from the left lobe were fixed in 10% neutral buffered formalin, paraffin embedded, and cut into 5-μm sections. For immunohistochemical staining, sections were deparaffinized in xylenes and rehydrated through a graded series of ethanol/water solutions. For analysis of cyclin E expression, liver sections were initially boiled in Antigen Retrieval Solution (DakoCytomation, Cambridge, MA) using a microwave (5 minutes at 100% power followed by 20 minutes at 30% power). Sections were then cooled to room temperature before proceeding with staining. Endogenous peroxidase activity was quenched by treatment in 3% H2O2 in methanol (DakoCytomation). Samples were blocked for 30 minutes in 5% goat serum in PBS. Primary polyclonal anti–cyclin E antibody (Santa Cruz Biotechnology, M20) was diluted 1:100 in blocking buffer and incubated for 1 hour at room temperature. Following washing with PBS, biotinylated secondary antibody and streptavidin-peroxidase conjugate were applied according to the Vectastain Elite avidin-biotin complex kit instructions (Vector Laboratories, Burlingame, CA). Staining was developed in 3,3′diaminobenzidine solution (Vector Laboratories) for 1 to 2 minutes and quenched in H2O. Slides were counterstained with hematoxylin, dehydrated through graded ethanols and xylene, and mounted using permount (Sigma, St. Louis, MO). BrdUrd incorporation was analyzed using a BrdUrd detection kit (Zymed, South San Francisco, CA) exactly as recommended by the manufacturer. BrdUrd incorporation was scored blind and at least 600 hepatocytes per section were counted from several random fields. For histologic analysis, sections were stained with H&E using standard techniques. TUNEL assay for colorimetric apoptosis detection was done using the FragEL DNA Fragmentation Detection Kit exactly as described by the manufacturer (Calbiochem, San Diego, CA).
Fluorescence-activated cell sorting analyses. Flow cytometry of unfixed-propidium iodide (PI)–stained nuclei was done as previously described ( 48). Briefly, liver nuclei were prepared from frozen tissue as described above. Unfixed nuclei were resuspended in 0.5 mg/mL PI, 0.1% NP40, 40 μg/mL RNase A, 0.1% sodium citrate in PBS. Following incubation for 10 minutes at room temperature, samples were fluorescence-activated cell sorting acquired. Flow Cytometry data was acquired on a Coulter Epics XL (Beckman-Coulter, Miami, FL) with a 488-nm argon-ion laser. Acquisition and ploidy analysis were done with System II software (Beckman Coulter, Fullerton, CA). Doublet discrimination was done by gating on peak versus integral signals. The gates used to quantitate the number of events at each ploidy level were set to correspond to 2-32N DNA content (PI) and were not moved for all subsequent analysis. Histograms and zebra plots were prepared using FlowJo software (Tree Star, Inc., Ashland, OR).
Centrosome staining. Imprints from freshly excised livers were fixed in methanol for 20 minutes at −20°C. Cells were washed in PBS and permeabilized with 1% NP40 in PBS for 5 minutes at room temperature. Slides were blocked with 10% normal goat serum in PBS for 1 hour and probed with mouse anti–γ-tubulin antibody ( 49) for 1 hour at room temperature. After washing in PBS, antibody-antigen complexes were detected by incubation with rhodamine-conjugated goat anti-mouse immunoglobulin G antibody for 1 hour at room temperature. Cells were washed thrice in TBS and counterstained with 4′-6′-diamidino-2-phenylindole (DAPI) before mounting and visualization by fluorescence microscopy.
Centromeric DNA staining. Freshly excised livers were gently pressed onto glass slides. Liver cells were fixed using two changes of methanol/acetic acid (3:1) at −20°C for 20 minutes each. Cells were dried at room temperature and stored at −20°C until use. For centromeric DNA staining, fixed cells were dehydrated through graded ethanol solutions (70%, 80%, and 95%) and incubated with 2× SSC for 2 minutes at 73°C. Cells were again dehydrated through graded ethanol solutions and denatured in 70% formamide/2× SSC for 5 minutes at 73°C. Cells were dehydrated again and incubated with a Cy3-labeled pan-centromeric DNA probe exactly as described by the manufacturer (Cambio, Cambridge, United Kingdom). Cells were counterstained with DAPI before mounting and visualized using fluorescence microscopy.
Somatic elimination of pRB in murine hepatocytes. To investigate the action of pRB in liver, a model for tissue-specific inactivation of Rb was developed. In this model, mice transgenic for Cre under the control of the albumin promoter (Alb-Cre) were crossed to mice that harbor loxP sites flanking exon 19 of the Rb gene (Rbf/f). Cre-mediated excision of the region flanked by the loxP sites results in truncation of the RB protein that is functionally equivalent to a null allele ( 46, 50). Litters contained 50% control (Rbf/f) and 50% experimental (Rbf/f;Alb-Cre+) mice and littermates were used for all analyses. Mice were born at the expected ratios and histologic analysis of neonatal livers reflected no gross changes in either the hepatic or hematopoietic cells in the liver (data not shown).
We first evaluated whether Alb-Cre mediated recombination of the Rb locus was specific to the liver. Consistent with a restricted pattern of albumin promoter activity and Cre expression, PCR evaluation of recombination in DNA extracted from liver, spleen, lung, and kidney revealed recombination exclusively in the liver Fig. 1A ). In addition, recombination was not detected in DNA extracted from heart, skin, brain, thymus, bone marrow, ovary, or colon (data not shown). We next determined the temporal extent of recombination in postnatal mice. Consistent with previous findings ( 51), recombination was incomplete in neonatal mice but was complete by 8 weeks of age ( Fig. 1B, compare lanes 2 and 6). To determine the corresponding influence of recombination on the Rb transcript, evaluation of Rb mRNA by RT-PCR was done and showed predominance of the recombined message specific to the livers of adult Rbf/f;Alb-Cre+ mice ( Fig. 1C). Lastly, to examine the influence of Cre-mediated recombination on pRB protein levels, pRB was immunoprecipitated from nuclear lysates derived from Rbf/f and Rbf/f;Alb-Cre+ mice and immunoblotted for pRB. These results confirmed a significant reduction in pRB levels in livers from Rbf/f;Alb-Cre+ mice ( Fig. 1D). Together, these data show effective liver-specific ablation of pRB using Alb-Cre. Once validated, this model system was used to assess the consequence of somatic pRB loss in the liver.
pRB-E2F target genes are deregulated in livers of Rbf/f;Alb-Cre+Cre+ mice. Initially, the consequence of liver-specific pRB loss in 21-day-old mice was assessed. These mice were chosen because they are the youngest mice at which complete recombination was detected by PCR, and represent a point at which extramedullary hematopoiesis is complete (data not shown). Given the well-established role of pRB in repressing E2F activity and inhibiting S-phase entry, the effect of pRB loss on the expression of several pRB/E2F target genes was analyzed. Immunoblotting revealed elevated levels of cyclin E, MCM7, PCNA, and p107 proteins in nuclear lysates prepared from Rbf/f;Alb-Cre+ mice compared with littermate controls ( Fig 2A , compare lanes 5-8 with lanes 1-4). In contrast, levels of the pRB-related p130 pocket protein, which has not been shown to be regulated by E2F, were unchanged in Rbf/f;Alb-Cre+ mice. Furthermore, BrdUrd incorporation was ∼2-fold higher in hepatocytes from Rbf/f;Alb-Cre+ mice than controls ( Fig. 2B and C), demonstrating increased S-phase entry in pRB-deficient hepatocytes. Whereas the percentage of hepatocytes incorporating BrdUrd was increased in Rbf/f;Alb-Cre+ mice, there was no significant change in liver mass in these animals ( Fig. 2D). To delineate the basis for this finding, the apoptotic fraction in Rbf/f and Rbf/f;Alb-Cre+ livers was analyzed. TUNEL assays revealed no elevation in levels of apoptosis in Rbf/f;Alb-Cre+ livers (data not shown), demonstrating that the lack of increased liver mass in these animals was not due to removal of aberrantly proliferating hepatocytes. Combined, these data suggest that the increase in the number of hepatocytes in the S phase in pRB-deficient mice was not reflective of a gross deregulation of proliferation.
pRB is dispensable for repression of pRB/E2F target genes and inhibition of proliferation in adult mice. Adult hepatocytes have a long half-life (∼180 days in the mouse) and low basal proliferation rate; for example, mitosis is observed only in about 1 in 20,000 hepatocytes at any given time ( 52). Given the observed deregulation of pRB/E2F targets and increased cell cycle progression in young pRB-deficient mice, the consequences of prolonged pRB deficiency were assessed in adult mice. Interestingly, in the livers of 16-week-old mice, cyclin E, MCM5, MCM7, and PCNA expression was not detectable by immunoblot, even in mice lacking pRB ( Fig. 3A, lanes 3 and 4 ). Comparable with young mice, there was also no significant difference in the levels of apoptosis in the livers of adult Rbf/f and Rbf/f;Alb-Cre+ mice (data not shown). Additionally, BrdUrd was not incorporated (comparable with control mice) in hepatocytes of adult pRB-deficient mice (data not shown). Consistent with these observations, the liver mass of Rbf/f;Alb-Cre+ mice was not statistically different from Rbf/f mice ( Fig. 3B). These data show that despite the deregulated gene expression noted in young mice, pRB is not required for the repression of these genes or inhibition of proliferation and induction of quiescence in adult mice.
It is now clearly established that significant functional redundancy exists between pRB and the closely related pocket proteins p107 and p130 ( 18, 26, 27). Furthermore, recent evidence has suggested that functional compensation by p107 in pRB-deficient keratinocytes is critically dependent on the levels of p107 protein ( 33). Therefore, given the observation that E2F target gene expression and hepatocyte proliferation are repressed in both pRB-proficient and pRB-deficient adult livers, p107 and p130 protein levels were monitored in livers of adult Rbf/f and Rbf/f;Alb-Cre+ mice. In 16-week-old Rbf/f;Alb-Cre+ mice, the floxed Rb locus remains fully recombined ( Fig. 1B), confirming that pRB is absent in these livers. Immunoblot analysis of p107 and p130 in liver nuclear lysates prepared from 16-week-old mice showed significantly elevated levels of both of these proteins in Rbf/f;Alb-Cre+ mice ( Fig. 3C, compare lanes 1-2 with lanes 3-4). These data suggest that with respect to the ability to repress E2F-directed transcription and proliferation, p107 and p130 may coordinately compensate for pRB loss in the adult liver.
Acute pRB ablation in adult liver deregulates E2F target gene expression and promotes cell cycle entry. Although functional compensation by p107 and p130 may explain the lack of a proliferative phenotype in pRB-deficient livers of adult Rbf/f;Alb-Cre+ mice, our data do not rule out the possibility that pRB plays no role in maintaining the quiescence of adult mouse hepatocytes. To address this question, recombinant adenovirus encoding Cre (Ad-Cre) was used to elicit acute pRB loss in livers of adult Rbf/f mice. We initially assessed the ability of Ad-Cre (1 × 109 pfu, delivered by i.v. injection) to facilitate recombination at the floxed Rb locus in liver tissue. PCR analysis showed the ability of Ad-Cre to mediate complete recombination of the floxed Rb locus within 3 days post-infection ( Fig. 4A ). We next determined whether acute pRB loss in the liver was sufficient to deregulate the expression of E2F-responsive genes. Immunoblot analysis of the levels of several pRB-E2F targets that function in cell cycle control (p107, cyclin E, and cyclin A) and DNA replication (MCM5, MCM7, and PCNA) revealed elevated levels of each of these proteins in pRB-deficient livers ( Fig. 4B). In contrast, acute pRB ablation did not influence levels of p130 protein. These data show that acute pRB loss is sufficient to deregulate the expression of E2F target genes in the adult liver. To determine whether acute pRB loss was sufficient for cell cycle entry in adult hepatocytes in vivo, BrdUrd incorporation was analyzed. As reported previously ( 53), infection with control (Ad-LacZ) adenovirus led to a low level of hepatocyte DNA synthesis. However, compared with controls, Ad-Cre–injected mice displayed a ∼4-fold increase in the number of cells staining positive for BrdUrd ( Fig. 4C). Nonetheless, in spite of the increased BrdUrd incorporation, no significant increase in liver mass was observed in Rbf/f mice infected with Ad-Cre ( Fig. 4D). Furthermore, TUNEL staining showed that acute pRB loss did not result in increased levels of hepatocyte apoptosis (data not shown). Collectively, these data show that acute pRB loss in adult liver is sufficient to deregulate E2F target gene expression and to facilitate cell cycle entry but this does not result in loss of control over organ size.
pRB controls hepatocyte ploidy. Together, the preceding data show that pRB controls hepatocyte cell cycle entry, but that compensation can occur with chronic loss of pRB. Whereas such compensation dampens some facets associated with pRB loss, (e.g., maintenance of quiescence; ref. 18), clearly chronically deficient pRB-null cells harbor distinct phenotypes ( 13, 15). Therefore, to investigate possible deleterious consequences of pRB loss, histologic analysis of liver sections from 16-week-old Rbf/f;Alb-Cre+ animals was done. Whereas no induction of hyperplastic proliferation or tumorigenesis was observed, nuclear pleomorphism with dramatically enlarged nuclei was readily evident in pRB-deficient livers by H&E staining ( Fig. 5A ). Such changes in nuclear size have been associated with increased hepatocyte ploidy. To determine whether the large nuclei specifically arising in Rbf/f;Alb-Cre+ mice did reflect elevated ploidy, in situ hybridization using a pan-centromeric DNA probe was done to monitor relative chromosome number. As shown in Fig. 5B, the number of positively labeled centromeres was dramatically elevated in the inherently large nuclei. These analyses indicate that pRB loss facilitates the acquisition of enhanced ploidy. To quantitatively assess ploidy changes throughout the liver, flow cytometry was used to concurrently analyze liver nuclear size and DNA content. Analysis of DNA content (PI) in both 8- and 16-week-old animals showed a shift towards higher DNA levels in Rbf/f;Alb-Cre+ mice ( Fig. 6A, top ). In addition, analysis of DNA content and nuclear size (forward scatter) showed that the populations with higher DNA content in Rbf/f;Alb-Cre+ mice were coincident with the larger nuclei ( Fig. 6A, bottom). Quantitative analysis of the changes in hepatocyte ploidy showed populations of nuclei with significantly elevated 8- and 16N DNA content and significantly reduced nuclei with 2N DNA content, in both 8- and 16-week-old Rbf/f;Alb-Cre+ mice ( Fig. 6B). Furthermore, in 16-week-old Rbf/f;Alb-Cre+ animals, there was a small population of nuclei with 32N DNA.
It has been shown that Rb, with haploinsufficiency, functions to maintain genetic stability in mouse embryonic stem cells ( 54). Therefore, the influence of loss of a single Rb allele on hepatocyte ploidy was determined by generating mice containing one wild-type and one floxed Rb allele, either with (Rbwt/f;Alb-Cre+) or without (Rbwt/f) Alb-Cre expression. Flow cytometric comparison of nuclear DNA content and size in 12-week-old mice revealed no significant differences between Rbwt/f and Rbwt/Δexon19 mice ( Fig. 6C). In contrast to loss of both Rb alleles, these data show that chronic loss of a single allele of Rb is insufficient to influence the ploidy of adult hepatocytes. Furthermore, these findings also show that in contrast to previously reported data demonstrating the development of aneuploidy during prolonged high-level expression of Cre recombinase ( 55), long-term Cre expression in our animals is not associated with ploidy changes. Together, the above data show that pRB regulates hepatocyte ploidy in vivo, that a single Rb allele is sufficient for this function, and that the changes in ploidy seen in pRB-deficient mice were specific to pRB loss and not due to Cre expression.
We reasoned that changes in ploidy observed in Rbf/f;Alb-Cre+ mice could be due to aberrant replication or some facet of mitotic deregulation. pRB loss has been associated with centrosome hyperamplification ( 29) that could fuel aberrant ploidy in specific tissues. Therefore, centrosome number was analyzed in hepatocytes of Rbf/f and Rbf/f;Alb-Cre+ animals. To quantitate centrosomes, liver touch preparations were stained for the centrosome marker γ-tubulin, counterstained with DAPI, and visualized by immunofluorescence. Interestingly, in contrast to the effects of Cre-mediated pRB inactivation in mouse epidermis ( 29), we detected no centrosome hyperamplification in pRB-deficient hepatocytes, even in those cells with significantly larger nuclei ( Fig. 6D, left). Liver cells from both Rbf/f and Rbf/f;Alb-Cre+ animals predominantly contained a single centrosome (∼85%) and all remaining cells contained two centrosomes ( Fig. 6D, right). No cells were identified that contained more than two centrosomes, demonstrating that deregulation of centrosome duplication does not contribute to the ploidy changes in pRB-deficient hepatocytes.
Acute pRB loss results in rapid augmentation of hepatocyte ploidy. Having shown that long-term pRB deficiency is associated with significant elevations in hepatocyte ploidy, the kinetics of these changes was assessed. To address this, Ad-Cre was used to acutely ablate pRB in vivo. Specifically, Rbf/f mice were injected with Ad-Cre or Ad-LacZ and sacrificed 6 days later, at which time livers were excised and examined for changes in hepatocyte nuclear size and ploidy. First, liver sections from adenovirus-infected mice were stained with DAPI to visualize nuclei. Numerous large nuclei were evident in sections specifically from livers of Ad-Cre–infected mice ( Fig. 7A ). Furthermore, flow cytometric analysis showed the presence of nuclei with elevated DNA content and nuclear size in the livers of Ad-Cre–infected mice ( Fig. 7B). Quantitation of these changes revealed significantly increased numbers of nuclei with 16- and 32N DNA content in Ad-Cre–infected mice ( Fig. 7C). These data show that acute pRB loss results in the rapid acquisition (within 6 days) of elevated hepatocyte ploidy in vivo.
The development of knockout mice has provided insight into the roles of the retinoblastoma tumor suppressor in development and inhibition of tumorigenesis in mice ( 49). However, the embryonic lethality associated with germ line Rb deletion in mice has largely precluded study of the consequences of pRB loss in adult animals, a condition highly relevant to human cancer. Using an Rb conditional knockout, we show that pRB loss is sufficient to deregulate E2F target gene expression and cell cycle control in post-natal hepatocytes. However, despite this deregulated cell cycle progression, pRB loss is compensated for such that in adult mice pRB-deficient hepatocytes are biochemically normal and do not exhibit hyperplasia. Consistent with compensation within the animal, the acute elimination of pRB in adult liver is sufficient to deregulate E2F target gene expression and elicit cell cycle entry. As a consequence of pRB loss, both chronic and acute, there is a significant development of aberrant hepatocyte ploidy.
Studies using mice in which Rb is deleted in the germ line revealed that pRB loss results in deregulated development associated with extensive apoptosis and differentiation defects in many tissues ( 20, 22, 23). Rb-null cells from these embryos undergo complete ectopic cell division followed by apoptotic cell death ( 56). Furthermore, recent experiments showed that conditional pRB loss in mouse epidermis is associated with epithelial cell proliferation defects and hyperplasia ( 29, 33). In contrast, Rb-deficient cells in chimeric mice consisting of both Rb−/− and Rb+/+ cells survive but do not complete cytokinesis ( 56, 57), suggesting that these cells may undergo cell cycle arrest before entering mitosis. Our analysis in the livers of postnatal mice similarly shows that liver-specific pRB knockout results in a cell cycle defect, as evidenced by enhanced hepatocyte BrdUrd incorporation. However, the absence of evidence of mitotic figures (data not shown) or hyperplasia, coupled with the lack of significantly increased liver weight suggests that although pRB-deficient hepatocytes are able to ectopically enter the S phase, they do not progress through mitosis. Supporting this idea, the absence of elevated levels of apoptosis in pRB-deficient hepatocytes counters the possibility that organ mass is maintained by the removal of ectopically dividing cells. Together, these data show that in contrast to the epidermis, conditional pRB elimination in the liver, although sufficient to cause ectopic cell cycle entry, is not sufficient to result in aberrant cell division and hyperplasia in vivo.
Acute ablation of pRB in the adult mouse liver results in hepatocyte cell cycle entry and shows that a critical function of pRB is the maintenance of quiescence in adult hepatocytes. However, it is also clear that in livers in which pRB is chronically absent, the induction and maintenance of quiescence is largely unaffected, demonstrating that other factors must compensate for pRB in this regard. We found that levels of the pRB-related p130 protein were elevated in the livers of adult mice harboring chronic pRB loss. Importantly, such increases are specifically associated with conditions wherein there is evidence of compensation occurring in pRB-deficient livers (i.e., E2F target genes are not deregulated), strongly implicating a role for p130 in compensating for pRB in the liver. Although there is clear evidence that p130 and pRB have overlapping functions in development ( 26), this is the first demonstration that p130 may be involved in compensating for pRB loss in adult tissues. Additionally, levels of p107 protein were elevated in quiescent pRB-deficient adult livers, suggesting a compensatory role for p107. This idea is supported by a previous study showing that p107 compensates for pRB loss in mouse epidermis by initiating proliferative arrest in the prolonged absence of pRB ( 33). The mechanism through which p107 and p130 are up-regulated as a consequence of pRB loss is unclear. It is well established that p107 is an E2F target gene ( 58) and the levels of p107 RNA and protein are up-regulated in pRB-deficient cells ( 14, 18, 59). Indeed, p107 levels were up-regulated during the deregulated cell cycle progression in the pRB-deficient livers of young mice and following acute pRB inactivation in adult mice. However, how the levels of p107 remain elevated, whereas a number of other E2F target genes (e.g., MCM7 and PCNA) are repressed in quiescence is unclear. Similarly, the basis for the up-regulation of p130 is equally enigmatic. p130 is not an identified E2F target gene and levels of p130 protein were not elevated in livers exhibiting deregulated proliferation following pRB loss, suggesting the possibility that accumulation of this protein could be due to the regulated stability of p130 protein. Thus, although pRB loss clearly deregulates cell cycle entry and target gene expression in pRB-deficient livers of young animals, the accumulation of p107/p130 through processes not completely understood can preclude these effects.
Human cancer is primarily a sporadic disease arising in post-mitotic tissues of adults. To mechanistically dissect which function(s) of pRB are most relevant to its function as a tumor suppressor, it is therefore critical to understand the biochemical consequences of acute, somatic pRB inactivation in adult tissues. Here we show that acute pRB ablation in liver results in E2F target gene deregulation and ectopic cell cycle entry. Strikingly, we did not detect any evidence for cell division or hypercellularity in livers following acute pRB loss, suggesting that pRB loss alone is not sufficient to facilitate completion of mitosis. However, our data clearly show that pRB is responsible for maintaining the quiescence of post-mitotic hepatocytes in mice. A formal possibility is that the acute loss of pRB will lead to hyperplasia with aging or that even in the adult compensation will occur with time. Delineating these facets of pRB action will clearly forward our understanding of the role of pRB loss on tumor etiology in sporadic cancers.
Whereas the loss of pRB does not lead to detectable hyperplasia in the livers of adult mice, it does lead to a specific histologic aberration within the liver, the appearance of nuclear pleomorphism. This is in fact so striking that the genotype of mice can be determined by the histology of the liver. Here we show that the likely cause of the pleomorphism is the enhanced ploidy of pRB-deficient cells. We and others have found that pRB loss results in aberrant ploidy in cell culture ( 46, 60); however, although enlarged nuclei were detected in pRB-deficient hepatocytes in chimeric animals ( 61), aberrant ploidy has not heretofore been shown in an animal model of pRB loss. The mechanism underlying the change in ploidy is complicated in that two facets of cellular biology cooperate to induce changes in nuclear ploidy. First, increased ploidy reflects the failure of nuclei with replicated DNA to segregate their genetic material. This event is then followed by subsequent cycles of replication to increase the ploidy above 4N. As such, the generation of polyploid cells requires the uncoupling of DNA replication from mitosis ( 62). In the specific case of hepatocytes, it has been recently shown that the generation of polyploidy is indeed due to a loosening of the normally tight coupling of DNA replication and mitosis ( 63). Clearly, this coupling is further antagonized via the loss of pRB. Avni et al. recently showed that pRB blocks ectopic DNA synthesis in fibroblasts by inhibiting the firing of specific origins of replication under some conditions ( 64). Thus, it is feasible that this function of pRB is transiently lost following pRB ablation in the liver leading to endoreduplication and increased ploidy in hepatocytes the absence of cell division. An aspect of mitotic deregulation commonly manifested in tumors is centrosome hyperamplification. Centrosomes nucleate the microtubule array and as such play an important role in mitosis ( 65). Aberrant centrosome numbers are found in a variety of tumors and have been shown in conditional knockout of Rb in the epidermis ( 29). Here we failed to observe centrosome abnormalities associated with pRB loss, even in cells with the large (polyploid) nuclei, demonstrating that the development of aberrant ploidy is not associated with centrosome hyperamplification.
Together, our studies show that pRB in the liver has some interesting biological consequences that may contribute to liver tumorigenesis. Clearly, aberrant cycles of proliferation as occur during chronic liver damage (e.g., hepatitis infection or alcoholism) fuels hepatocellular carcinoma ( 37). As such, loss of pRB and the observed cell cycle deregulation in adult livers could represent a progression event in liver tumorigenesis. Given that compensation can occur, we speculate that the window of deregulated cell cycle progression associated with pRB loss is limited and thus provides an explanation for the lack of a hyperplastic phenotype associated with chronic pRB loss. However, the manifestation of increased ploidy represents a durable characteristic of pRB loss. Although polyploid hepatocytes are a common feature of normal liver, their prevalence increases in a variety of pathophysiologic conditions ( 62, 66). Furthermore, it is believed that genetic instability in these cells may provide a means to develop aneuploidy and therefore contribute to tumorigenesis ( 62). Thus, these studies provide the basis for delineating the action of pRB in the suppression of liver tumorigenesis and the maintenance of appropriate liver function.
Grant support: National Cancer Institute grant CA106471 (E.S. Knudsen) and training grant T32 CA 59268 (C.N. Mayhew) and NIEHS Core grant E30-ES-06096 (E.S. Knudsen).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Dr. David Witte (Department of Pathology, Cincinnati Childrens Hospital Medical Center) for evaluation of liver histology, Dr. Lique Coolen for assistance with microscopic image capture, the UC Comparative Pathology Core for histologic preparations, Erin Williams for assistance with immunohistochemistry techniques, Dr. Jeff Albrecht (Division of Gastroenterology, Hennepin County Medical Center, Minneapolis) for helpful discussions, Drs. Karen Knudsen and Christin Petre for critical reading of the article, and all the members of both of the Knudsen laboratories for helpful comments.
- Received November 24, 2004.
- Revision received March 16, 2005.
- Accepted March 17, 2005.
- ©2005 American Association for Cancer Research.