The Cdk inhibitor p21 is required for necrosis, but it inhibits apoptosis following toxin-induced liver injury.

Liver injury and repair were examined in wild type, p21Waf1/Cip1, and p27Kip1-deficient mice following carbon tetrachloride (CCl4) administration. In wild type liver, p21 expression is induced in a biphasic manner following injection of CCl4, with an early peak of p21 expression occurring in pericentral hepatocytes at 6 h, prior to evidence of injury, and a second peak succeeding regenerative proliferation. In contrast, p27 is present throughout the quiescent liver, but its expression decreases following CCl4 injection. Surprisingly, p21-deficient animals were resistant to CCl4-induced necrotic injury, indicating that rapid induction of p21 in pericentral hepatocytes following CCl4 injection contributes to subsequent necrosis. Expression of cytochrome P450 2E1, which plays an essential role in CCl4-induced necrotic injury, was not affected in p21-deficient mice. Although they had the least injury, p21-deficient mice had the highest levels of hepatic proliferation that correlated with increases in hyperphosphorylated retinoblastoma protein and Cyclin A gene expression. Increased replication in p21-deficient livers was counteracted by an increase in hepatocyte apoptosis as detected by caspase-3 activation. p21 plays distinct and opposing roles regulating hepatocyte survival during injury and subsequent repair, with early induction of p21 contributing to necrotic injury and later expression to cessation of proliferation and hepatocyte survival.

physiological roles of CKIs in tissue repair and remodeling. It has the unique ability to regenerate following tissue loss because of chemical or physical injury (reviewed in Refs. 8 and 9). Regulation of the regenerative process must include signals that stop cell proliferation, once the liver has been restored to the appropriate size. Carbon tetrachloride (CCl 4 ) has been widely used to study mechanisms of hepatic injury and repair following toxin-induced injury (10). It is metabolized in the centrilobular zone of the liver where the production of highly reactive trichloromethyl radicals leads to the necrotic death of pericentral hepatocytes (11,12). One to 2 days following CCl 4 injection, liver cells within the periportal and intermediate zones divide in response to the severe damage and cell death that occurs in the pericentral region.
Previously we detected a dramatic increase in expression of mRNA encoding the CKI p21 Waf1/Cip1 in mouse pericentral hepatocytes between 4 and 8 h after CCl 4 injection, long before any clear evidence of liver injury was apparent. A second peak of p21 mRNA expression was observed between 1 and 2 days post-injection in periportal hepatocytes that had divided to replace the dead cells. The two peaks of p21 mRNA expression observed following CCl 4 -induced injury were detected in both wild type and p53-deficient mice, indicating that p21 gene expression was not p53-dependent (13). In the current study, we have addressed the roles of p21 and the related CKI p27 during the course of CCl 4 -induced liver injury and regeneration using mice deficient for p21 or p27. These studies provide a better understanding of the molecular mechanisms regulating liver injury and repair following chemical injury, as toxins and liver cell necrosis play significant roles in a number of clinical liver diseases (reviewed in Ref. 14).

EXPERIMENTAL PROCEDURES
Animals-Conventionally raised viral antigen free wild type, p21Ϫ/Ϫ (15), and p27Ϫ/Ϫ (16) male mice at 8 -10 weeks of age were used in all experiments, unless otherwise indicated. At least three mice per genotype were examined per time point. Mice were fed a commercial diet and water ad libitum. To induce injury 10 l/g body weight of a 10% solution of carbon tetrachloride in corn oil or corn oil alone was injected intraperitoneally, and mice were sacrificed at the times indicated. The left lobe of the liver was fixed in 4% paraformaldehyde and paraffinembedded for sectioning. Other portions of liver were frozen in liquid nitrogen and kept at Ϫ70°C for preparation of protein lysates and total RNA.
Semi-quantitative and Quantitative PCR-Semi-quantitative RT-PCR was performed to examine expression of mouse p21, p27, and cytochrome P450 2E1 (CYP 2E1) as described (7). The following primer pairs and conditions were used for detecting expression of p21 (upstream primer, AGTGTGCCGTTGTCTCTTCG; downstream primer, ACACCAGAGTGCAAGACAGC; annealing temperature 62°C, 30 cycles; product 311 bp), p27 (upstream primer, TCGCAGAACTTCGAAG-AGG; downstream primer, TGACTCGCTTCTTCCATATCC; annealing temperature 62°C, 30 cycles; product 300 bp), and CYP 2E1 (upstream primer, AGTGTTCACACTGCACCTGG; downstream primer, CCTGG-AACACAGGAATGTCC; annealing temperature 57°C, 20 cycles; product 125 bp). Expression of mouse S16 ribosomal protein was examined as an internal control as described previously (13). For each combination of primers, the kinetics of PCR amplification were studied; the number of cycles corresponding to plateau was determined, and PCR was performed within the exponential range. Amplified products were separated on either 1.2 or 2% agarose gel and visualized with ethidium bromide staining.
Quantitative real time PCR was performed to measure cyclin A gene expression (upstream primer, AGTACCTGCCTTCACTCATTGCTG; downstream primer, TCTGGTGAAGGTCCACAAGACAAG) using the ABI PRISM 7700 (Applied Biosystems, Foster City, CA) and SYBR Green technology. Mouse S18 expression was examined as a control using the primers described (17). 96-Well optical plates with reaction mixtures were heated for 2 min at 50°C and 4 min at 95°C, followed by 40 cycles of PCR consisting of 15 s at 95°C and 20 s at 63°C. At the end of the run, samples were heated to 95°C with a ramp time of 10 s to construct dissociation curves to check that single PCR products were obtained. PCRs were also analyzed by gel electrophoresis to confirm that a single product of the expected size was amplified. Validation experiments were performed to demonstrate that efficiencies of target and reference amplifications were approximately equal. The comparative C T method for relative quantitation of gene expression described by Applied Biosystems was used to determine cyclin A expression levels. Experiments were carried out in triplicate for each data point. Sequence Detection Systems 1.7 software (Applied Biosystems, Foster City, CA) was used for analysis.
Assessment of Liver Injury-Levels of serum alanine aminotransferase (ALT) were measured to assess the degree of hepatic necrosis. Serum was collected from animals and analyzed for ALT activity using an in vitro diagnostic kit from Sigma.
The DeadEnd TM Fluorometric TUNEL System (Promega Corp., Madison, WI) was used to examine injury in the liver. Fluorescein-12-UTPlabeled DNA in liver sections was visualized using fluorescence microscopy.
To examine caspase-3 activation in situ, liver sections were pretreated for antigen recovery and accessibility by microwave heating in 0.01 M citrate (pH 6.0) solution, 4ϫ for 5 min. The Tyramide Signal Amplification kit (PerkinElmer Life Sciences) and cleaved caspase-3 antibody (Asp-175, Cell Signaling Technology, Beverly, MA) were used for immunohistochemistry.
Autoradiographic Analyses-For determination of hepatocyte proliferation, 2 Ci of methyl-[ 3 H]thymidine (Amersham Biosciences)/g body weight was injected intraperitoneally into three mice of each genotype 1 h prior to sacrifice. Tissue sections were dipped into NTB-2 emulsion and exposed for 17 days. Developed slides were counterstained with hematoxylin-eosin, and the number of hepatocytes incorporating [ 3 H]thymidine was determined by counting 3 H-labeled hepatocyte nuclei in 3 grids per field. At least 5 fields were counted on each slide, and 3 slides were analyzed for each sample.

RESULTS
Differential Regulation of p21 and p27 following Liver Injury-Expression of p21 and p27 protein was examined in the liver during the course of CCl 4 -induced liver injury and regeneration in wild type mice and mice lacking p21 or p27. Wild type, p21Ϫ/Ϫ, and p27Ϫ/Ϫ mice injected with CCl 4 or corn oil alone were sacrificed at 0, 3,6,9,12,24,36,48,72, and 168 h after injection. Total proteins were isolated and subjected to immunoblotting with antibodies against p21, p27, or ␤-actin (Fig. 1A). We were unable to detect p21 protein in wild type liver lysates, but its expression was induced by 3 h after CCl 4 injection. Two peaks of p21 protein were detected at 6 and 48 h in wild type mice. p53-independent transcriptional regulation plays a key role in the induction of p21 following CCl 4 injection (13), and the two peaks of p21 protein expression followed peak mRNA expression detected using RT-PCR (Fig. 1B).
Significant p27 protein expression was detected in wild type quiescent liver, but shortly after CCl 4 administration p27 levels decreased. Decreases in p27 protein levels after injury appear to be post-transcriptionally regulated, because p27 mRNA levels remain fairly constant during the course of regeneration (Fig. 1B). Interestingly, significant levels of p21 protein were detected in p27-deficient mice at the zero time point (Figs. 1A and 5B), although no increase in p21 RNA expression was detected in p27-deficient mice (Fig. 1B). Our earlier studies indicate that p21 accumulates in the livers of p27-deficient mice through a post-transcriptional mechanism and compensates for the absence of p27 in maintaining hepatocyte quiescence (7).
Differences in p21 expression were also detected at later time points in p27-deficient mice when compared with wild type FIG. 1. Expression of p21 and p27 in the livers of wild type mice and mice deficient for p21 or p27 following CCl 4 -induced injury. A, at the indicated time points after CCl 4 injection, immunoblot analyses were performed with antibodies against p21, p27, or ␤-actin to examine expression of p21 and p27 protein in wild type and knockout mice during the course of liver injury and subsequent regeneration. The genotypes of the mice are indicated on the left, and antibodies used are indicated on the right. Two peaks of p21 protein expression were induced following CCl 4 , whereas p27 levels decreased in wild type mice. To confirm equal loading and transfer of proteins, membranes were stained with Ponceau S (not shown) and probed with an antibody against ␤-actin. B, p21 and p27 mRNA expression was examined using RT-PCR with liver RNAs isolated at the indicated time points following CCl 4 injection. Two peaks of p21 mRNA expression were detected following CCl 4 injection, whereas p27 mRNA levels remained constant in wild type mice. Expression of the mouse S16 ribosomal protein gene was examined as an internal control. Typical ethidium bromide-stained agarose gels with bands corresponding to p21, p27, and mouse S16 ribosomal protein are shown. mice (Figs. 1A and 5B). The early peak of p21 protein expression detected in wild type animals was muted, and the later peak of p21 expression occurred at 36 h in p27-deficient mice versus 48 h in the wild type mice. These data indicate that hepatocytes deficient for p27 differentially regulate expression of p21 at several time points. In p27-deficient mice the pattern of p21 RNA expression was very similar to that of wild type mice, suggesting that the differences in p21 expression are regulated post-transcriptionally in p27-deficient hepatocytes. We found previously (7) that p21 protein expression was regulated at the level of protein stability in cultured primary hepatocytes.
Decreased Necrotic Injury in p21-deficient Mice-CCl 4 is a classical inducer of centrilobular necrosis. When hepatocytes undergo necrosis, ALT is released from hepatocytes into the bloodstream, resulting in abnormally high serum levels of this enzyme. To assess the level of hepatocyte necrosis, we measured serum ALT levels throughout the course of liver injury and repair (Fig. 2). Peak levels of serum ALT were detected at 18 h post-CCl 4 injection in both wild type and p27Ϫ/Ϫ mice. At this time point serum levels of ALT in p21-deficient mice were ϳ5-fold less than in the other two genotypes. Release of ALT was delayed in p21-deficient mice, with peak levels appearing at 24 h. However, cumulative serum ALT in p21Ϫ/Ϫ animals never reached the levels detected in wild type and p27 null animals. p27-deficient mice had higher levels of serum ALT at FIG. 3. Assessment of liver injury using the TUNEL assay and immunohistochemical localization of cleaved caspase-3. A, the TUNEL assay was performed with sections of liver tissue from wild type (WT), p21Ϫ/Ϫ, and p27Ϫ/Ϫ mice at 12, 18, and 24 h after CCl 4 injection. Injury occurs in pericentral hepatocytes surrounding the central veins. Examples of central veins (CV) and portal triads (PT) are labeled. Strong TUNEL labeling of pericentral hepatocytes was detected in wild type and p27Ϫ/Ϫ liver sections at 18 and 24 h but not in p21Ϫ/Ϫ sections. B, tissue sections from livers at 24 h post-CCl 4 injection were incubated with an antibody specific for cleaved caspase-3. Few cells in the wild type tissue sections were positive for cleaved caspase-3 expression, indicating that the strong TUNEL labeling shown for wild type sections in A represents necrosis and not apoptosis. Highest levels of apoptotic cells (cleaved caspase-3-positive) were detected in the p21Ϫ/Ϫ liver sections (arrows). Antibody binding was visualized using fluorescein isothiocyanate (green), and 4,6-diamidino-2-phenylindole was used to stain nuclei (blue). Size bar represents 100 m.
FIG. 2. p21-deficient mice have reduced serum ALT levels following injection of CCl 4 . Blood samples were drawn by cardiac puncture; plasma was separated by centrifugation, and ALT enzyme levels were measured as described. ALT levels are shown as units/liter Ϯ S.E. Reduced levels of serum ALT in p21-deficient mice are indicative of reduced necrotic liver injury. 24 h post-injection than wild type mice, and the p27-deficient mice also express higher levels of p21 at this time point (see Figs. 1A and 5B).
The decreased level of necrotic injury in p21-deficient mice following CCl 4 injection was verified using a variety of different techniques. The basic histological subunit of the liver is the lobule, a hexagonal structure consisting of six portal triads surrounding a central vein (reviewed in Ref. 18). Only hepatocytes surrounding the central vein are killed following the metabolism of CCl 4 . During the peak period of CCl 4 -induced injury, viable cells were detected by simple hematoxylin-eosin staining in the pericentral region of the p21Ϫ/Ϫ liver lobule but not in the wild type or p27Ϫ/Ϫ liver lobule (data not shown).
To assess further the level of injury, TUNEL assays were performed. At 18 and 24 h after CCl 4 injection, intense labeling of cells was detected in the pericentral regions of wild type and p27-deficient liver lobules but not in the livers of p21-deficient mice (Fig. 3A). It has been shown that the TUNEL assay fails to discriminate between apoptosis and necrosis in the liver (19).
To determine whether the TUNEL labeling indicated necrosis or apoptosis, immunohistochemistry was performed with parallel liver sections and antibodies specific for activated (cleaved) caspase-3. The vast majority of TUNEL-positive cells in wild type and p27-deficient livers were negative for cleaved caspase-3, indicating that the cells were necrotic. However, in p21-deficient livers, cells near the central vein were found to be undergoing apoptosis. Cleaved caspase-3 antibody reactivity at 24 h when peak levels of ALT are released in p21-deficent mice is shown in Fig. 3B.
The cytochrome P450 2E1 (CYP2E1) metabolizes CCl 4 and is constitutively expressed in pericentral hepatocytes. The importance of this enzyme in CCl 4 metabolism was demonstrated using CYP2E1-deficient mice, which are resistant to CCl 4induced injury (20). Decreased injury in response to CCl 4 in p21-deficient mice led us to ask if CYP2E1 was efficiently expressed in these animals. We examined CYP2E1 mRNA levels in wild type and p21Ϫ/Ϫ mice before treatment and 18 h after CCl 4 injection, when peak serum ALT was detected in wild type animals (Fig. 4). Wild type and p21-deficient mice expressed comparable levels of CYP2E1 mRNA at the zero time point. By 18 h post-CCl 4 injection, it became difficult to detect CYP2E1 mRNA in wild type mice, probably due to the high level of necrotic death of CYP2E1-expressing cells. However, CYP2E1 mRNA was still readily detected in the p21-deficient animals. Thus, decreased CYP2E1 expression is not responsible for the decrease in necrotic injury in the p21-deficient mice. Our ability to detect CYP2E1 mRNA in p21-deficient mice at 18 h post-CCl 4 injection is an additional indicator of increased cell viability in the centrilobular zone.
Increased Regenerative Proliferation and Expression of Cyclin A in Mice Lacking p21 following CCl 4 -induced Liver Injury-Expression of Cdk2, Cdk4, Cyclin A, Cyclin D1, Cyclin E, and PCNA was examined during the course of CCl 4 -induced regeneration in wild type, p21Ϫ/Ϫ, and p27Ϫ/Ϫ mice (Fig. 5A). Expression of most of these genes was increased during the proliferative phase of liver regeneration (36 -72 h) in these mice. A notable prolonged increase in Cyclin A expression was detected in the livers of p21Ϫ/Ϫ mice when compared with wild type controls and p27Ϫ/Ϫ mice.
In Fig. 5A relative amounts of protein expressed at different time points in each genotype are displayed. However, differences in experimental conditions preclude making comparisons regarding absolute levels of protein expressed at each time point between livers of mice of each genotype. To compare directly protein expression levels in the different genotypes, 50 g of liver protein from wild type and p21-and p27-deficient mice at 0, 6, 24, 36, and 48 h post-CCl 4 injection were loaded on the same gel, and immunoblotting was performed (Fig. 5B). Increased p21 protein levels were detected in the liver at the zero time point and 24 and 36 h post-CCl 4 injection in mice lacking p27, whereas reduced levels of p21 are expressed at 6 h in the p27-deficient animals. Increased Cdk and PCNA expression was also present at the zero time point in p27 null mice (see also Ref. 7). A substantial increase in Cyclin A levels was detected in p21Ϫ/Ϫ liver samples at 48 h post-CCl 4 injection.
To determine whether p21 or p27 are necessary for regulation of hepatocyte proliferation following CCl 4 -induced injury, [ 3 H]thymidine was injected into wild type, p21Ϫ/Ϫ, and p27Ϫ/Ϫ mice 1 h prior to the time of sacrifice at 24, 36, 48, 72, and 168 h following CCl 4 injection, and [ 3 H]thymidine incorporation was examined in liver sections (Fig. 6). Significant regenerative proliferation was not observed at the 24-and 168-h time points in any of the genotypes. In all genotypes of mice, peak proliferation was detected at 36 h post-CCl 4 injection. Mice deficient for p21 had the highest level of proliferation, with ϳ30% more hepatocytes incorporating [ 3 H]thymidine in the 1-h period in p21Ϫ/Ϫ mice than in wild type mice. Higher levels of cell proliferation were maintained in p21 null mice at 48 and 72 h post-CCl 4 injection.
The p27-deficient animals had lower levels of cell proliferation at all time points. This may be due in part to the growth inhibitory effect of increased expression of p21 in the livers of these mice (see Fig. 5B). Immunoprecipitations were performed with p21 antibodies and wild type and p27-deficient liver lysates, and immunoblotting was performed with antibodies against different Cdks and cyclins. We easily detected association of p21 with Cdk2, Cdk4, and cyclin D1 in p27-deficient liver lysates (Fig. 7). Because lower levels of p21 are expressed in wild type liver, longer exposure times were required to detect the same co-immunoprecipitated proteins leading to increased background levels. At 6 h post-CCl 4 injection, p21 was complexed with Cdk2, and the p21-Cdk2 complex was more abundant in wild type mice. However, no increase in proliferation was detected in p21-deficient mice at the 6-h time point (data not shown).
Prolonged Hepatocyte Proliferation Correlates with Increased Phosphorylation of pRb and Induction of Genes Required for S-phase Progression in p21-deficient Mice-Hyperphosphorylation of members of the retinoblastoma protein (pRb) family by Cdks leads to inhibition of the growth-suppressive functions of members of the pRb family, activation of the E2F family of transcription factors, and gene expression required for cell cycle progression. p21 is an inhibitor of pRb phosphorylation raising the possibility that p21-deficient mice might have increased FIG. 4. Expression of CYP2E1 is not reduced in p21-deficient mice. Expression of the mRNA encoding cytochrome P450 2E1 (CYP2E1) that plays an essential role in CCl 4 -induced necrosis was examined using RT-PCR with cDNA prepared from liver RNAs from male and female mice. Total RNA was isolated from untreated livers and livers at 18 h following CCl 4 injection, when peak levels of serum ALT are detected in wild type mice. At 18 h pericentral hepatocytes that express CYP2E1 in wild type mice have undergone necrosis, and CYP2E1 mRNA cannot be detected. Expression of the mouse S16 ribosomal protein gene was examined as an internal control. Typical ethidium bromide-stained agarose gels with bands corresponding to CYP2E1 and mouse S16 ribosomal protein mRNAs are shown. phosphorylation of members of the pRb family. We examined phosphorylation of pRb in the livers of p21Ϫ/Ϫ and wild type mice following CCl 4 administration (Fig. 8A). Hyperphosphorylated pRb levels were highest at the 48-h time point post-CCl 4 injection in wild type mice. Peak p21 protein induction is also detected at 48 h in wild type mice, which prevents the sustained proliferation detected in livers of p21-deficient mice. In p21deficient mice, prolonged hyperphosphorylation of pRb was detected from 24 to 72 h post-CCl 4 injection in the liver, indicating that p21 activity is important for inhibiting pRb hyperphosphorylation during liver regeneration. Increases in levels of the pRbrelated protein p107 were also detected in the p21Ϫ/Ϫ mouse livers between 36 and 72 h after CCl 4 injection.
Increased phosphorylation of members of the pRb family leads to the alleviation of repression and induction of genes required for S-phase progression such as Cyclin A (21,22). Because Cyclin A promoter activity is regulated by pRb (23), we examined Cyclin A mRNA expression levels using quantitative real time PCR. In p21-deficient animals the increased phosphorylation and inhibition of pRb repressive functions corresponded with increased Cyclin A mRNA (Fig. 8B) and protein expression (Fig. 5). The increased levels of Cyclin A may contribute to the prolonged proliferation observed in the p21Ϫ/Ϫ liver during regeneration. Increased Apoptosis Counteracts Increased Proliferation in p21-deficient Mice-Although the lowest levels of necrosis were detected in p21-deficient mice, these animals had highest levels of hepatocyte proliferation during regeneration. In many different systems, expression of p21 has been correlated with an inhibition of apoptosis (reviewed in Ref. 24). To determine whether apoptosis plays a role in regulating liver mass in p21-deficient mice, we used antibodies specific for activated caspase-3 to examine apoptosis in the livers of wild type and p21-deficient mice. Many hepatocytes were positive for activated caspase-3 in p21-deficient livers, but not in wild type livers, at 36 h post-CCl 4 administration, when proliferation is at its peak (Fig. 9). Diffuse green staining in wild type liver sections represents background that is specific for the injured region of the liver lobule. Results obtained using three different wild type and p21-deficient mice from three independent experiments are presented in Fig. 9. We do detect some variability in the numbers of cleaved caspase-3-positive cells in the p21-deficient animals, which may be due in part to differences in metabolic activity in the liver arising from variations in feeding, because the animals have free access to food. Still, in all cases the p21-deficient mice have a significant increase in apoptotic cell number. Thus, in the absence of p21 there is more proliferation at 36 h post-CCl 4 injection, but this proliferation is counteracted by an increase in programmed cell death. DISCUSSION The liver is the largest gland in the body and the site where drugs and chemicals are metabolized. Toxins produced during metabolism may give rise to injury, but the healthy liver is To examine the cell cycle regulatory proteins associated with p21 in wild type (A) and p27Ϫ/Ϫ mice (B) at different times following CCl 4 administration in the liver, p21 and its associated proteins were coimmunoprecipitated (IP) and subjected to immunoblotting. Filters were probed with antibodies against Cdk2, Cdk4, and cyclin D1. Coimmunoprecipitation of Cdk2 with p21 is visible at early time points, suggesting a functional role for p21 during its first peak of induction. The Cdk2 band migrates above the immunoglobulin light chain background band. Cdk4 and cyclin D1 also coimmunoprecipitate with p21 from 36 to 72 h post-CCl 4 injection.
capable of repairing itself. We explored the roles of p21 and p27 throughout the course of liver injury and repair following CCl 4induced liver injury, and we have determined the functions of the early and late peaks of p21 expression following CCl 4 in-jection. The early peak of p21 expression promotes necrotic injury, and this is one of the first reports that suggests a role for p21 in necrosis. The later peak of p21 induction serves as a cell cycle checkpoint and inhibitor of cell proliferation during regeneration. For the most part, p27-deficient mice responded like wild type mice, although increased p21 expression in p27deficient mice led to a slight increase in necrotic injury at 24 h (Fig. 2) and decreased proliferation at later time points (Fig. 6). Although animals of all three genotypes recovered from the liver injury, it appears that p21 plays a major role in regulating proliferation and cell viability following toxin-induced injury and subsequent regeneration.
Although p21 expression was not detected in the uninjured liver, high levels were expressed within 3-6 h specifically in pericentral hepatocytes after CCl 4 injection (13). In p21-deficient mice no significant aberrant proliferation was detected during the first 24 h post-CCl 4 injection. However, p21-deficient animals displayed a marked decrease in the level of necrotic injury, indicating that p21 expression promotes necrosis. Disruption of the p21 gene did not lead to altered expression of CYP2E1, which plays an essential role in CCl 4 -induced necrosis.
The mechanisms underlying the dependence of necrotic injury on induction of p21 remain unclear. The immediate early genes c-fos and c-myc are rapidly induced in the pericentral region before p21 is expressed following CCl 4 injection (13), and the expression of these genes with opposing growth regulatory functions in the pericentral region may set the stage for ensuing events that lead to necrosis. To try to explore further the mechanism underlying regulation of necrosis, we isolated primary hepatocytes from wild type and p21-deficient mice and treated them with CCl 4 . However, these isolated hepatocytes did not respond like hepatocytes within the three-dimensional structure of the adult liver, and p21 was not induced in the cultured wild type cells in response to CCl 4 . It appears that a number of signals are provided by the complex liver architecture, and this system is not recapitulated in vitro.
p21 expression decreased after 6 h post-CCl 4 injection and FIG. 8. Hyperphosphorylation of pRb and p107 and increased expression of cyclin A mRNA in p21-deficient mice correlate with prolonged proliferation. A, immunoblotting experiments demonstrate prolonged hyperphosphorylation of pRb in the regenerating livers of mice deficient for p21. In addition, increased levels of p107 are detected in p21Ϫ/Ϫ mouse livers after CCl 4 injection. B, quantitative real time PCR was used to examine Cyclin A mRNA expression in wild type (WT) and p21-deficient mice. Cyclin A mRNA expression levels were normalized to S18 RNA expression levels using the comparative C T method. Increased Cyclin A mRNA levels were detected in p21-deficient mice from 24 to 48 h after CCl 4 injection.
FIG. 9. Increased apoptosis in the livers of p21-deficient mice during the replicative phase of liver regeneration counteracts excessive proliferation. Paraffin-embedded liver sections from wild type (WT) and p21Ϫ/Ϫ mice at 36 h post-CCl 4 injection were stained with antibodies against activated caspase-3, and antibody binding was visualized using fluorescein isothiocyanate, and 4,6-diamidino-2-phenylindole was used to stain nuclei. Data from three wild type and three p21Ϫ/Ϫ mice are shown. Increased apoptosis of hepatocytes surrounding the central veins is observed in p21-deficient livers. Examples of central veins (CV) and portal triads (PT) are marked. Size bars represent 100 m. was expressed at low levels until it was induced again from 36 to 72 h, coincident with the major phase of regenerative proliferation that occurs following toxin-induced injury. The second peak of p21 expression is consistent with a role for p21 as a cell cycle checkpoint protein and an inhibitor of cell proliferation following the initial wave of hepatocyte replication. We examined cell proliferation in wild type, p21Ϫ/Ϫ, and p27Ϫ/Ϫ mice during the course of CCl 4 -induced injury and repair. In all cases peak proliferation as measured by tritiated thymidine incorporation was detected at 36 h post-CCl 4 injection. The highest numbers of proliferating cells were detected in p21Ϫ/Ϫ mice at all time points examined (Fig. 6). The extended time course of hepatocyte proliferation in animals lacking p21 at 48 and 72 h post-CCl 4 injection was not observed following partial hepatectomy, after which accelerated early peak proliferation, but no prolonged proliferation, was observed in p21Ϫ/Ϫ female mice (25). p21 expression was higher in p27Ϫ/Ϫ liver when compared with wild type livers at several time points examined, and p27-deficient animals had lowest levels of proliferation. We showed previously that p21 protein accumulates in the p27-deficient liver where it can inhibit hepatocyte proliferation (7). pRb and p107 have been shown to have overlapping functions in the developing liver (26). We observed increased levels of hyperphosphorylated pRb and increased expression of p107 post-CCl 4 injection in the livers of p21 null mice. Hyperphosphorylation of pRb abrogates its growth-suppressive functions and disrupts its ability to bind members of the E2F family, leading to activation of a number of genes required for DNA synthesis (reviewed in Ref. 27). Increased expression of Cyclin A would support the longer duration of DNA replication detected in p21Ϫ/Ϫ animals following CCl 4 injury. Recently it was reported that expression of another E2F target gene, the Cdk-activating phosphatase Cdc25A, is higher in p21-deficient livers, and this may also contribute to the increased cell proliferation detected in p21Ϫ/Ϫ mice (28).
We asked how liver mass is properly maintained in p21deficient mice, which had the lowest levels of liver injury but highest levels of cell proliferation. By performing immunohistochemistry with antibodies specific for activated caspase-3, we found that livers from the p21-deficient mice had significantly higher levels of apoptosis, particularly during the peak of proliferation. Thus apoptosis is important for control of liver size in the face of too much cell proliferation during the course of liver regeneration.
Many reports have indicated that p21 induction may inhibit apoptosis by a variety of mechanisms (reviewed in 24), and our data suggest an apoptosis inhibiting role for p21 in the liver. Increased E2F activity, which may occur as a consequence of disruption of the p21 gene, could contribute to increased apoptosis by a number of p53-dependent and independent mechanisms (reviewed in Refs. 29 -31). It was reported recently (32) that deregulation of E2F activity may directly contribute to the regulation of caspase expression. Increased apoptosis of hepatocytes during the proliferative phase of liver regeneration may also be regulated by environmental signals, as the liver is remodeled and there is an increased number of cells in an organ with restricted size as a consequence of p21 disruption.
We have shown here that p21 plays a central role in regulating both proliferation and cell viability in the toxin-injured liver. The importance of counteracting excessive proliferation with increased cell death has been recognized as a necessary mechanism for preventing the development of cancer (33). We are currently investigating the susceptibility of p21-deficient mice to developing liver tumors after treatment with specific carcinogens. Both decreased and increased p21 expression have been correlated with the development of hepatocellular carcinoma (34,35), perhaps attesting to the dual functions of p21 as a regulator of injury and proliferation. Because of its abilities to regulate both proliferation and cell death, modulation of p21 activity following liver injury may have therapeutic benefits in patients with liver disease.