Hepatocyte-specific deletion of Cdc42 results in delayed liver regeneration after partial hepatectomy in mice

Authors

  • Haixin Yuan,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    2. Graduate School of the Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Hong Zhang,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    2. Graduate School of the Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
    • These authors contributed equally to this study.

  • Xunwei Wu,

    1. Cutaneous Biology Research Center, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA
    Search for more papers by this author
  • Zhe Zhang,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
  • Dan Du,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    2. Graduate School of the Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
  • Wenchao Zhou,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    2. Graduate School of the Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
  • Shuhua Zhou,

    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    2. Graduate School of the Chinese Academy of Sciences, Shanghai, China
    Search for more papers by this author
  • Cord Brakebusch,

    1. Biomedical Institute, BRIC, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Zhengjun Chen

    Corresponding author
    1. State Key Laboratory of Molecular Biology, Institute of Biochemistry and Cell Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China
    • Shanghai Institute of Biochemistry and Cell Biology, 320 Yueyang Road, Shanghai 200031, China
    Search for more papers by this author
    • Fax: (86)-21-54921071.


  • Potential conflict of interest: Nothing to report.

Abstract

Cdc42, a member of the Rho guanosine triphosphatase (GTPase) family, plays important roles in the regulation of the cytoskeleton, cell proliferation, cell polarity, and cellular transport, but little is known about its specific function in mammalian liver. We investigated the function of Cdc42 in regulating liver regeneration. Using a mouse model with liver-specific knockout of Cdc42 (Cdc42LK), we studied liver regeneration after partial hepatectomy. Histological analysis, immunostaining, and western blot analysis were performed to characterize Cdc42LK livers and to explore the role of Cdc42 in liver regeneration. In control mouse livers, Cdc42 became activated between 3 and 24 hours after partial hepatectomy. Loss of Cdc42 led to a significant delay of liver recovery after partial hepatectomy, which was associated with reduced and delayed DNA synthesis indicated by 5-bromo-2′-deoxyuridine staining. Consistent with this, expression of cyclins D1, A, and E was markedly delayed or reduced in Cdc42LK livers during regeneration. As a potential effector of Cdc42, Rac1 activation was dramatically attenuated in Cdc42LK livers after partial hepatectomy, suggesting it is regulated in a Cdc42-dependent manner. Activation of certain proliferative signaling pathways, such as extracellular signal–regulated kinase, c-Jun N-terminal kinase, and p70S6 kinase pathways, was delayed in Cdc42LK livers. In addition, dilated bile canaliculi and excessive lipid accumulation were observed in mutant livers during liver regeneration, which may result from impaired cytoskeletal organization and intracellular trafficking in hepatocytes. Conclusion: Our results revealed important roles of Cdc42 in the regulation of proliferative signaling during liver regeneration. (HEPATOLOGY 2009:49:240-249.)

The liver is critical for regulating metabolic homeostasis, because it is responsible for the metabolism, synthesis, storage, and redistribution of nutrients, carbohydrates, fats, and vitamins. Although hepatocytes rarely divide under normal circumstances, the liver has a remarkable ability to regenerate after surgical removal or after viral or chemical injury.1 Partial hepatectomy (PH) in rodents is one of the most widely used models to study liver regeneration. After a two-thirds hepatectomy in which the large and median lobes of liver are surgically removed,2 the remaining liver grows rapidly, leading to the restoration of the original mass, structure and function within a few days. In response to PH, over 95% of mature hepatocytes synchronously exit the G0 phase and re-enter the cell cycle.3 Hepatocytes are the first to replicate, followed sequentially by biliary epithelial cells, Kupffer cells, stellate cells, and sinusoidal endothelial cells.1

Liver regeneration is a well-orchestrated process in which complex signaling pathways coordinate the progression of distinct stages, including the withdrawal of hepatocytes from quiescence (“priming phase”), cell cycle entry and progression, and cessation of cell division and the return of hepatocytes to quiescence. Cell cycle progression is driven by growth factors, especially by hepatocyte growth factor (HGF) and the epidermal growth factor receptor ligand family, which can activate various mitogenic signaling pathways, including extracellular signal–regulated kinase (ERK), c-Jun N-terminal kinase (JNK), phosphoinositide-3 kinase, and S6 kinase.3, 4 Signaling cascades triggered by cytokines and growth factors have been extensively studied because they seem to play central roles in liver regeneration after PH.5 Recently, it has been demonstrated that liver regeneration is also tightly regulated by physiological functions of the liver, such as lipid6, 7 and bile acid8 metabolism.

In this study, we addressed the role of Cdc42 in liver regeneration. Cdc42, a small GTPase of the Rho GTPase family, cycles between an inactive GDP-bound state and an active GTP-bound state. It is known to play important roles in multiple cell functions, such as organization of the cytoskeleton, cell proliferation, cell polarity, and cellular transport. Cdc42 (as well as the Rho GTPases RhoA and Rac1) is required for progression from G1 to S phase in Swiss 3T3 fibroblasts.9 This may result at least partly from its ability to induce cyclin D1 expression.10–12 Cdc42 may also regulate cell growth and proliferation by activating several downstream signaling events, such as the p21-activated kinase (PAK),13 JNK,14, 15 and p70S6 kinase (p70S6K)16, 17 cascades. In recent years, some reports have focused on the role of Cdc42 in the regulation of mammalian epithelial cell polarity.18 Based on its ability to modulate the actin cytoskeleton, Cdc42 is also essential for intracellular lipid transport.19

In a mouse model with liver-specific knockout of Cdc42 (Cdc42LK mouse), livers exhibited significant hepatomegaly.20 The liver weight/body weight ratio of Cdc42LK mice markedly increased as the mice grew older, whereas it was stable in controls. This observation seemed inconsistent with current knowledge of Cdc42's role in promoting cell proliferation. To further investigate the mechanism of how Cdc42 regulates hepatocyte proliferation and liver growth, we used the mouse model of liver regeneration. Contrary to expectations, we found that liver regeneration was significantly delayed in Cdc42LK mice after PH. Accordingly, a delay in DNA synthesis and cell cycle progression during liver regeneration in Cdc42LK mice was observed. Furthermore, we found that Cdc42 deficiency impaired timely activation of ERK, JNK, and p70S6K after PH. These results suggest that Cdc42 plays a critical role in liver regeneration after PH.

Abbreviations

ABCA1, adenosine triphosphate–binding cassette transporter-1; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BrdU, 5-bromo-2′-deoxyuridine; Cdc42LK mice, Cdc42 liver-specific knockout mice; CDK4, cyclin-dependent kinase 4; ERK, extracellular signal–regulated kinase; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GST-PBD, GST-PAK1–binding domain; HGF, hepatocyte growth factor; JNK, c-Jun N-terminal kinase; mRNA, messegner RNA; NPCs, nonparenchymal cells; p70S6K, p70S6 kinase; PAK, p21-activated kinase; PH, partial hepatectomy.

Materials and Methods

Animal Studies.

Cdc42LK mice were generated by crossing Cdc42 flox/flox mice with mice expressing Cre recombinase under control of a rat albumin promoter as described.20 Mice were maintained in alternating 12-hour light/dark cycles. Animals received humane care in compliance with the institutional guidelines of the Shanghai Institutes for Biological Sciences.

Six- or seven-week-old mice with matched sexes were used for PH. Mice were anesthetized with ether and subjected to approximately 70% liver resection of the left lateral and median lobes after a midventral laparatomy.2 The mortality of the animals following PH was less than 5%. One hour prior to sacrifice, a single dose of 5-bromo-2′-deoxyuridine (BrdU) (Sigma, St. Louis, MO) was injected intraperitoneally at 50 mg/kg animal weight (5 mg/mL in phosphate-buffered saline). Mice were reanesthetized for blood collection and sacrificed at several time points after hepatectomy. The liver/body weight ratios were measured. Liver specimens were then harvested and stored in liquid nitrogen or fixed in 4% paraformalin. Serum levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase, and bile acids were measured via standard automotive techniques using a Synchron CX5 Pro analyzer (Beckman Coulter, Fullerton, CA).

Histology and Immunohistochemistry.

Five-micrometer-thick paraffin sections were prepared from embedded tissues and analyzed via hematoxylin-eosin staining. For Oil Red O staining, 8-μm-thick frozen sections were prepared from frozen tissues, stained with Oil Red O (Sigma), and counterstained slightly with hematoxylin. Immunohistochemistry was performed as described.20 BrdU staining was performed with the BrdU Labeling and Detection Kit II (Roche, Germany) according to the manufacturer's instructions, and the sections were counterstained slightly with hematoxylin. Three samples and two visual fields per sample were surveyed to determine the average percentage of BrdU-positive cells. Nonparenchymal cells (NPCs) were identified by their smaller, irregularly shaped nuclei compared with the larger, circular nuclei of hepatocytes. Because total NPCs were too faint to be seen in sections stained slightly with hematoxylin, they were counted in serial sections with regular hematoxylin-eosin staining. Section images were taken with a digital camera (Nikon DXM1200F, Tokyo, Japan) mounted on a light microscope (Nikon Eclipse E600, Tokyo, Japan).

GST-PAK1–Binding Domain Pull-Down Assay.

For the GST-PAK1–binding domain (GST-PBD) pull-down assay, tissue samples were homogenized on ice in cell lysis buffer (50 mM Tris-HCl [pH 7.4], 200 mM NaCl, 5 mM MgCl2, 1% [vol/vol] Nonidet P40, and 10% [vol/vol] glycerol) with 1 mM phenylmethylsulfonyl fluoride and 2 mM Na3VO4. Equal amounts of protein lysates were incubated with 8 μg GST-PBD (amino acids 69 to 150 of murine PAK1) for 4 hours at 4°C and washed as reported.21 The bead pellet was suspended in 20 μL Laemmli sample buffer. Proteins were separated via 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and analyzed via western blotting.

Western Blotting.

Equal amounts of protein (50-80 μg/lane) were separated via sodium dodecyl sulfate–polyacrylamide gel electrophoresis and electro-blotted onto nitrocellulose transfer membrane (Schleicher & Schuell BioScience, Keene, NH). Immunodetection of proteins was performed using specific antibodies. Densitometric analysis was performed with Quantity One software (Bio-Rad, Hercules, CA).

Antibodies.

Rat anti-mouse F4/80 was purchased from AbD Serotec (Oxford, UK). Anti-MRP2 was a gift from Dr. Stieger Bruno (University Hospital, Zurich, Switzerland). Antibodies recognizing Cdc42, cyclin D1, cyclin E, cyclin A, and total ERK1/2 protein were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-Rac1 was from BD Biosciences (San Jose, CA). Anti-ABCA1 was from Abcam (Cambridge, UK). Antibodies to cyclin-dependent kinase 4 (CDK4), phospho-ERK1/2, phospho-/total JNK1/2, phospho-/total p70S6K, phospho-/total Akt, and phospho-/total mTOR were purchased from Cell Signaling Technology (Beverly, MA). Mouse anti–glyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody was purchased from KangChen Bio-tech (Shanghai, China). Anti–α-tubulin was from Sigma.

Statistical Analysis.

All data are expressed as the mean ± standard deviation. The statistical significance of differences between mean values (P < 0.05) was evaluated using the unpaired Student t test.

Results

Generation and Analysis of Cdc42LK Mice.

Cdc42LK mice were born at the expected Mendelian ratio, were fertile, and looked healthy. To test the efficiency of Cdc42 deletion in mutant mice, we isolated livers from 3- and 6-week-old mice. Compared with the control group, the level of Cdc42 protein in mutant mice was reduced by about 70% at the age of 3 weeks, and by 80% to 90% at the age of 6 weeks (Fig. 1A). Because Cdc42 was retained in some NPCs that did not express albumin, complete ablation of Cdc42 in the liver could not be achieved. We then chose 6- to 7-week-old mice for further studies. At this age, the body weight of male Cdc42LK mice was lower than that of male control mice, but the difference between Cdc42LK and control females was insignificant (Fig. 1B). However, the liver/body weight ratio was notably higher in both male and female Cdc42LK mice (Fig. 1C). We also observed elevated levels of serum ALT, alkaline phosphatase, and bile acids in Cdc42LK mice (Supplementary Fig. 1A-C). Furthermore, messenger RNA (mRNA) levels of the proinflammation factor tumor necrosis factor α increased in the livers of these mice (Supplementary Fig. 1D). Disarranged structure and dilated bile canaliculi were demonstrated by hematoxylin-eosin staining and immunostaining of multidrug resistance protein 2 (a canalicular transporter for bile salts22), which were described in detail later.

Figure 1.

Analysis of Cdc42LK mice. (A) Western blot analysis of Cdc42 in the livers of control (Ctrl) and Cdc42LK mice aged 3 weeks and 6 weeks; α-tubulin was used as a loading control. (B) Comparison of male or female body weight of control and Cdc42LK mice aged 6 weeks. (C) Comparison of the liver weight/body weight ratio of control and Cdc42LK mice aged 6 weeks. The graphs in (B) and (C) summarize average values of 5 to 9 mice. Error bars indicate standard deviations. *P < 0.05. **P < 0.005.

Cdc42 Activation After PH.

To determine whether Cdc42 became activated after PH, we performed a pull-down assay to measure its ability to bind to the PBD effector domain.21 In control mice, Cdc42 activity increased markedly at 3 hours and peaked at 6 hours after PH (Fig. 2). The activity level showed a small decrease at 24 hours, and then returned to the initial level at 48 hours. Although Cdc42 was retained in some NPCs in Cdc42LK livers, we did not detect Cdc42 activation in them using the same assay (Supplementary Fig. 2), demonstrating that the activation signal comes primarily from hepatocytes.

Figure 2.

Activation of Cdc42 during liver regeneration in control mice. (A) Activation of Cdc42 after PH determined by GST-PBD pull-down assay. As quantitative controls, GST-PBD fusion protein was reprobed on the same membrane, and total Cdc42 in protein lysates is shown. (B) Densitometric analysis is represented with the value of 0 hours fixed at 1.0. Values are expressed as the mean ± standard deviation of three independent experiments.

Delayed Recovery of Liver Mass in Cdc42LK Mice After PH.

We then performed a two-third hepatectomy on control and Cdc42LK mice. And liver recovery was monitored by comparing the liver weight/body weight ratio at each time point. Although initially the liver weight/body weight ratio of mutant mice was higher than that of control mice, this difference gradually decreased to insignificant levels during the late phase of regeneration, suggesting the relative slower growth rate of mutant mice during liver regeneration (Fig. 3A). Despite the difference of the liver/body weight ratio between the mice, the liver usually recovers to its pre-hepatectomy ratio within a given time after resection to meet the functional requirements of the body.23 To facilitate the comparison of liver recovery between control and Cdc42LK mice, the liver/body weight ratio at each time point after PH was normalized to the ones before hepatectomy (Fig. 3B). Significantly delayed liver regrowth could be observed since 48 hours after PH in Cdc42LK mice (Fig. 3B). At 168 hours, Cdc42LK mice recovered 67.0% of the original ratio, while control mice gained 90.6%. However, the livers of Cdc42LK mice could ultimately restore their original ratios at about 3 weeks after PH (data not shown), indicating that loss of Cdc42 was unable to completely block liver regeneration.

Figure 3.

Liver recovery after PH in control and Cdc42LK mice. (A) More rapid recovery of liver weight/body weight ratio in control mice after hepatectomy. (B) Normalized liver recovery curves during liver regeneration. Remnant liver weight/body weight ratios were recorded when animals were killed at each time point. To assay liver recovery, post-hepatectomy ratios were divided by the pre-hepatectomy ratio of control or Cdc42LK mice with sex-matched mice. Mean values were calculated based on four samples for each time point. Error bars indicate standard deviations. *P < 0.05. **P < 0.005 versus controls at the respective time points.

Histological and Biochemical Analysis During Liver Regeneration.

In hematoxylin-eosin—stained sections, no major abnormalities were observed in Cdc42LK livers, whereas disarranged structure of hepatic plates could be seen (Fig. 4A, panels A, C). In addition, dilated bile canaliculi were observed in Cdc42LK livers (Fig. 4A, panel c, inset), which were demonstrated by immunostaining of multidrug resistance protein 2. Although Cdc42LK livers displayed a disorganized lobular structure, no obvious degeneration or necrosis was observed during liver regeneration. However, more dilated bile canaliculi were seen in Cdc42LK liver sections (Fig. 4A, panels b, d). More NPCs were observed in mutant livers, which were indicated by their smaller, irregularly shaped nuclei in contrast to the larger, circular nuclei of hepatocytes. This was further confirmed by more Kupffer cells (the major NPCs in the liver) with F4/80 immunostaining in Cdc42LK livers (Fig. 4B).

Figure 4.

Histological analysis of mice livers during regeneration. (A) Hematoxylin-eosin stained liver sections at (a, c) 0 hours and (b, d) 48 hours after PH. Dilated bile canaliculi in Cdc42LK livers were apparent before hepatectomy and became more obvious at 48 hours revealed by multidrug resistance protein 2 staining (arrows in each inset). Low power scale bars: 100 μm, insets: 20 μm. (B) More Kupffer cells were observed in Cdc42LK livers revealed by F4/80 staining. Scale bars: 50 μm. (C) Excessive lipid accumulation in Cdc42LK livers at 48 hours after PH. Frozen sections at (a, d) 0 hours, (b, e) 24 hours, and (c, f) 48 hours were stained with Oil Red O. Most scale bars: 50 μm, insets: 20 μm. (D) Serum levels of ALT (left panel) and AST (right panel) during liver regeneration. Error bars indicate standard deviations. *P < 0.05.

Because Cdc42 is involved in intracellular lipid transport,19 we examined whether lipid metabolism during liver regeneration was affected by Cdc42 deficiency with Oil Red O staining. Lipid accumulation was excessively observed in Cdc42LK livers 48 hours after PH (Fig. 4C, panels c, f), though it was normal before PH and remained normal until 24 hours after PH compared with controls (Fig. 4C, panels a, b, d, e). Interestingly, the localization of an important transporter for cellular efflux of cholesterol and phospholipids, adenosine triphosphate–binding cassette transporter-1 (ABCA1), which can bind Cdc42 in vitro,24 was also impaired in Cdc42LK livers during liver regeneration. Juxtanuclear staining of ABCA1 in Cdc42LK livers was lost (Supplementary Fig. 3) although its membrane staining seemed unimpaired.

Apoptosis during liver regeneration was similar between control and Cdc42LK livers, with very few apoptotic hepatocytes detected (Supplementary Fig. 4). In addition, the ALT and AST levels were significantly higher in Cdc42LK mice before PH, suggesting impairment in the mutant hepatocytes (Fig. 4D). However, there was no obvious difference in the change curves between control and Cdc42LK mice, although ALT levels at 24 hours after PH were slightly higher in the control group (Fig. 4D). This indicated that global liver functions were comparable between control and mutant mice during liver regeneration.

DNA Synthesis During Liver Regeneration Is Attenuated and Delayed in Cdc42LK Mouse Livers.

DNA synthesis at various time points after PH was monitored using a BrdU incorporation assay. Before hepatectomy, DNA synthesis in Cdc42LK livers was higher than that in controls, which was consistent with the observed hepatomegaly in Cdc42LK livers. At 48 hours after PH, a peak of DNA synthesis was detected in proliferating hepatocytes in control livers (Fig. 5A,B). After that, DNA synthesis decreased rapidly, and returned to the basal level at 168 hours after PH. Compared with the control group, the level of DNA synthesis in Cdc42LK livers was much lower at 48 hours but decreased very slowly, resulting in higher DNA synthesis levels from 120 hours to 168 hours after PH (Fig. 5B). Regeneration of NPCs was similar in both control and Cdc42LK livers (Fig. 5C), although the levels were slightly higher in Cdc42LK livers during the initial and last stages. These results suggest that loss of Cdc42 attenuates the peak of hepatocyte DNA synthesis and delays the process of regeneration.

Figure 5.

DNA synthesis during liver regeneration. (A) Immunohistochemical staining for BrdU-positive cells. Scale bars: 50 μm. (B, C) Percentage of (B) BrdU-positive hepatocytes or (C) NPCs was quantified for each time point. Error bars indicate standard deviations. *P < 0.05. **P < 0.005 versus controls at the respective time points.

Delayed and Decreased Expression of Cyclins in Regenerating Cdc42LK Livers.

Because DNA replication was attenuated and delayed in the regenerating livers of Cdc42LK mice, we compared the expression of cyclin proteins in control and Cdc42LK mice. Consistent with the results from BrdU staining, cyclin D1 expression was higher in the Cdc42LK livers before PH. During liver regeneration, cyclin D1 levels started to increase at 24 hours and peaked at 48 hours in control mice. In Cdc42LK livers, however, cyclin D1 was not expressed until 48 hours, and it remained low until 120 hours (Fig. 6). The expression curve of cyclin A, as well as cyclin E, was similar between control and Cdc42LK mice during liver regeneration, but their levels were reduced in Cdc42LK livers, indicating an impaired transition from G1 to S phase (Fig. 6). The level of CDK4 expression was similar in both livers after PH, although it was much higher in Cdc42LK livers before PH (Fig. 6).

Figure 6.

Expression of cyclin D1, A, E, and CDK4 during liver regeneration. (A) Western blotting of cyclin D1, A, E, and CDK4 in whole-liver extracts prepared from control and Cdc42LK mice. GAPDH was used as a loading control. (B) Densitometric analysis of the results in (A). Values were normalized to that of GAPDH and are expressed as the mean ± standard deviation of three independent experiments. *P < 0.05 versus controls at the respective time points.

Impaired Proliferative Signaling in Cdc42-Deficient Regenerating Livers.

To further investigate the molecular mechanism by which Cdc42 regulates liver regeneration, we first examined the activation of Rac1 (a potential effector of Cdc42 belonging to the same Rho GTPase family25) during liver regeneration. Like Cdc42, Rac1 was markedly activated from 3 hours to 24 hours after PH in control mice (Fig. 7A). Interestingly, Cdc42 deficiency led to a significant decrease in Rac1 activation, although the expression levels of Rac1 were slightly increased in Cdc42LK mice (Fig. 7A). This indicates that Rac1 activation during liver regeneration is largely dependent on Cdc42 activity.

Figure 7.

Activation of signaling cascades in the regenerating liver of control and Cdc42LK mice. Western blot analyses of (A) GTP-bound Rac1 and total Rac1 protein; (B) phospho-/total ERK1/2; (C) phospho-/total JNK1/2 protein; and (D) phospho-/total p70S6K, phospho-/total Akt, and phospho-/total mTOR in control and Cdc42LK livers are shown at the indicated time points after PH.

Some signaling cascades involving Cdc42 and known to be important for regeneration were then studied. Activation of the primary proliferative kinase ERK1/2 was delayed in Cdc42LK mice after PH (Fig. 7B). Notably, ERK1/2 activation was stronger in control mice than in Cdc42LK mice before PH. Activation of JNK and p70S6K, two candidate effectors of Cdc42, were also delayed in the Cdc42LK group during liver regeneration (Fig. 7C,D). However, in contrast to ERK, JNK activation was stronger in Cdc42LK mice than in control mice before hepatectomy (Fig. 7C). Note that p70S6K was activated in an earlier stage than ERK and JNK. We then examined the activity of Akt26 and mTOR,27 which could regulate p70S6K activity and were known to be involved in liver regeneration. Akt activation after hepatectomy was not affected by loss of Cdc42 (Fig. 7D). However, a prolonged activation of mTOR was detected in Cdc42LK livers (Fig. 7D), which might partially contribute to delayed activation of p70S6K at 12 hours after PH. In addition, we also explored the mechanism that may compensate for delayed proliferative signals in the late stages of regeneration in Cdc42LK livers. It was found that HGF mRNA levels were significantly up-regulated between 24 and 48 hours after PH in mutant livers compared with the steady level in the controls (Supplementary Fig. 5). Together, these results show that Cdc42 regulates regenerative signaling through ERK, JNK, and p70S6K pathways.

Discussion

In the rat oval cell model of liver regeneration induced by 2-acetamidofluorene treatment and PH, Cdc42 RNA and protein levels were both up-regulated, indicating a possible role for Cdc42 in liver regeneration.28 However, the precise functions of Cdc42 during common liver regeneration remain to be clarified. We found that activation of Cdc42 occurred 3 to 6 hours after PH and persisted until about 24 hours after PH. Rac1 activation occurred within a similar time frame and was largely dependent on Cdc42 activity. Hence, Rac1 seems to be a downstream effector of Cdc42 in the liver, which is in agreement with previous findings in cell lines from other tissues.25, 29–31 In another study of hepatectomy in rats, activated RhoA was detected at 6 hours and 18 hours after PH,32 which is similar to what we found here. Taken together, these results indicate critical roles for Rho GTPase, Cdc42, Rac1, and RhoA in the cell cycle progression of hepatocytes.9

A perplexing consequence of Cdc42 deletion in mouse livers is the appearance of hepatomegaly, as well as signs of liver transformation and hepatocellular carcinoma that appeared later.20 These phenotypes might be caused by chronic liver injury or increased hepatocyte proliferation; and the possibility that Cdc42 deficiency directly promotes cell proliferation and liver growth had not been excluded.20 Based on our results here, the increased proliferation of hepatocytes in Cdc42LK livers might not be a direct consequence of Cdc42 deficiency. In fact, we observed a higher level of serum bile acids (Supplementary Fig. 1C) in Cdc42LK mice, which might be caused by an impairment of bile canaliculi and cell–cell junctions.20 Elevated bile acids could trigger an inflammatory response (Supplementary Fig. 1D) and act as endogenous carcinogens in the liver.33, 34 We also found that JNK activity was higher in quiescent Cdc42LK livers than in controls. JNK was possibly induced by a stress response to bile acids, and it could account for the increased cyclin D1 before PH (Fig. 6). However, ERK activity, which mediates primary proliferative signals, was reduced in Cdc42LK livers before PH. Moreover, studies using in vitro cultured hepatocytes revealed that JNK activation was also down-regulated in Cdc42-deficient primary hepatocytes in response to HGF or epidermal growth factor stimulation (data not shown), which was similar to the results of a study in primary mouse embryonic fibroblasts with Cdc42 deletion.30 These data suggest that Cdc42 deficiency actually impaired the innate ability of hepatocyte to proliferate. However, signaling induced by extracellular stress was increased in Cdc42-deficient livers, which could overcome the negative influence of Cdc42 deletion and lead to excessive liver growth. What is more, acute injury by PH seems to produce much more intense stress and growth signals that substitute for those in quiescent livers to restore original mass and physiological functions, in which Cdc42 might be indispensable for the proper regulation of signaling.

We then found that loss of Cdc42 delayed the recovery of the liver to its original mass after PH. This was associated with an attenuated peak of DNA synthesis at 48 hours as well as delayed or reduced expression of cyclins after PH. Previous studies have demonstrated a role for Cdc42 in promoting G1 progression by inducing cyclin D110–12 and cyclin E expression.35 We did not observe increased levels of apoptosis in Cdc42LK livers, suggesting that delayed liver recovery was not related to cell death. Interestingly, compared with the sharp decline of DNA synthesis after 48 hours in controls, there was only a minor decrease in DNA synthesis from 48 hours to 168 hours in Cdc42LK livers (although the level of synthesis was much lower than the peak level of controls). Apparently, liver recovery in Cdc42LK livers progressed at a slow rate, and full recovery was achieved in about 3 weeks. Cdc42, therefore, is not essential for liver regeneration, but it does play an important role in the process. In fact, because a complicated and redundant intracellular signaling network is active in liver regeneration, loss of an individual gene rarely leads to complete inhibition of the process.5

We developed a more detailed mechanistic understanding of how Cdc42 regulates proliferation in the liver by examining key proliferative signaling pathways that involve Cdc42. ERK1 and ERK2, two primary kinases known to transmit mitogenic signals, were activated much later in Cdc42LK livers than in control livers. The precise mechanism through which Cdc42 regulates ERK activation is not clear, but it might be through a PAK/Raf/MEK pathway.36 The Ras/Cdc42/JNK/c-Jun cascade has been implicated in regulating DNA synthesis in hepatocytes.37 In rats, inhibition of JNK activity suppressed cyclin D1 expression and reduced mitotic figures after PH.38 We found that loss of Cdc42 delayed the activation of JNK during liver regeneration, which is consistent with the delay we observed in cyclin D1 expression. Thus, our results reinforce the notion that a pathway involving Cdc42, JNK, and cyclin D1 may contribute sequentially to G1 progression in hepatocytes. Interestingly, HGF mRNA levels were significantly up-regulated between 24 and 48 hours after PH in mutant livers (Supplementary Fig. 5), whereas the levels seemed steady in controls. Note that both ERK and JNK were activated between 24 and 48 hours after PH in Cdc42LK mice. Up-regulated HGF may activate some extra cascades and partially contribute to these delayed proliferative signals in Cdc42LK mice.

p70S6K, which regulates the phosphorylation of the ribosomal S6 subunit, has been demonstrated to play a critical role in hepatocyte proliferation.39–41 Specific deletion of p70S6K in the mouse liver blocks hepatocyte proliferation after PH due to a failure to induce cyclin E expression.42 Our finding that p70S6K activation was delayed in Cdc42LK livers after PH confirms that p70S6K is regulated by Cdc42 during hepatocyte proliferation, which might be responsible for biosynthesis of early response proteins after PH.42 In addition, the possibility that these proliferative pathways were nonspecifically disturbed in Cdc42LK livers could be ruled out, because activation of Akt (also involved in liver regeneration and p70S6K activity regulation) was not affected in Cdc42LK livers. Interestingly, a prolonged activation of mTOR was found in mutant livers, which may contribute to the delayed p70S6K activation. Although mTOR is an important regulator for p70S6K activity,27, 43 activation of p70S6K in the early phase might need the help of Cdc42 to recruit it to the membrane.16 Thus, it is not surprising that little phosphorylated p70S6K could be detected in Cdc42LK mice up to 6 hours after PH. However, a prolonged activation of mTOR may activate some extra pathways and bypass the requirement of Cdc42, resulting in the delayed p70S6K activation. Taken together, these results suggest that Cdc42 participates in both early and late proliferative signal events for proper regulation of liver regeneration.

Because Cdc42 is an important regulator of actin cytoskeleton system, severe dilation of bile canaliculi (Fig. 4A) may be due to abnormal cytoskeleton organization in Cdc42LK livers. In fact, thickened actin filaments at pericanalicular regions were often observed in Cdc42LK hepatocytes (Supplementary Fig. 6), which was confirmed by an ultrastructure analysis.20 Cytoskeleton disarrangement in Cdc42LK livers may also lead to defects of cellular lipid transport and excessive lipid accumulation (Fig. 4C). Strikingly, Cdc42 signaling has been reported to mediate lipid transport in skin fibroblasts,19, 24 and our results suggest a similar role in hepatocytes. Moreover, it was also found that loss of Cdc42 in hepatocytes result in a lack of juxtanuclear localization of the lipid transporter ABCA1, which seemed to support earlier results that Cdc42 can bind and mediate ABCA1 signaling cascades.24 Although the mechanisms through which Cdc42 regulates hepatic lipid metabolism remain unclear, a plausible hypothesis is that loss of Cdc42 disrupts intracellular localization of ABCA1, which then leads to a defect in outward lipid transport. More studies are needed to address Cdc42's role in constructing cell and tissue architecture as well as in intracellular trafficking through the cytoskeleton during liver regeneration.

In conclusion, our results demonstrate an important role for Cdc42 in promoting hepatocyte proliferation and liver regeneration. Loss of Cdc42 activity after PH caused delayed liver regeneration, which was correlated with delayed or decreased cyclin expression and delayed activation of ERK, JNK, and p70S6K. We also show that Cdc42 is necessary for proper regulation of hepatocellular lipid metabolism during liver regeneration. This work may help expand our understanding of the hepatic response in acute liver diseases, and it may ultimately lead to novel therapeutic strategies for enhancing liver regeneration after hepatic injury.

Acknowledgements

We thank Dr. Stieger Bruno (University Hospital, Zurich, Switzerland) for providing excellent anti-MRP2.

Ancillary