Potential conflict of interest: Nothing to report.
Cell adhesion to the extracellular matrix (ECM) plays vital roles in both morphogenesis and regulation of gene expression in cells of adult organisms. How intracellular, cytoskeletal, and signaling factors connect and communicate with the ECM is a fundamental question. Using a cDNA microarray analysis, we identified phosphatidylinositol 4,5-bisphosphate (PI[4,5]P2) phosphatase mRNA as being up-regulated in hepatocytes cultured on a basement membrane matrix, Engelbreth-Holm-Swarm (EHS) gel, which led to the finding that the PI(4,5)P2 levels of hepatocytes decreased on EHS gel. These changes in hepatocytes on EHS gel were accompanied by promotion of actin depolymerization and differentiated phenotypes of the hepatocytes. Treatment with PI(4,5)P2 or a phospholipase C inhibitor, U73122, resulted in decreased mRNA expressions of albumin and hepatocyte nuclear factor 4 (HNF-4) in hepatocytes. In contrast, actin-disrupting agent gelsolin increased mRNA expressions of albumin and HNF-4. In conclusion, organization of the actin cytoskeleton via PI(4,5)P2 is involved in the regulation of hepatocyte differentiation by the ECM. (HEPATOLOGY 2006;44:140–151.)
Adhesive interaction between cells and their surrounding extracellular matrix (ECM) plays vital roles in tissue morphogenesis, cell migration, proliferation, and the differentiation of a variety of cell types.1 Despite biochemical, cell biological, and genetic advances, how intracellular cytoskeleton and signaling proteins connect and communicate with ECM remains a fundamental question in cell biology. Rat hepatocytes in primary culture are used for studying the mechanism by which differentiation of the liver is regulated.2, 3 When isolated rat hepatocytes are cultured on type I collagen–coated dishes, the cells rapidly change their normal appearance, become flattened, and dramatically reduce liver-specific gene transcription.4 However, when hepatocytes are cultured on Engelbreth-Holm-Swarm sarcoma (EHS) gels, they appear to retain normal cell polarity and structure, and the products of liver-specific genes such as albumin continue to be secreted for prolonged periods of culture.5–7 Thus, the ECM may influence hepatocyte differentiation via the assembly of endocytic components. However, the precise mechanism by which ECM, intracellular molecules associated with the organization of cytoskeletal filaments, and hepatocyte function interact is poorly understood.
Actin is a highly conserved protein that is essential for a variety of cellular processes in eukaryotes. The actin cytoskeleton is fundamental for various motile and morphogenic processes in cells. In migrating cells a specialized network of actin filaments (F-actin) assembles at the leading edges. This network is polar, with the plus ends of actin near the plasma membrane and the minus ends farthest from the leading edge.8 The structure and dynamics of the actin cytoskeleton are regulated by a wide array of actin-binding proteins whose activities are controlled by various signal transduction pathways. The most thoroughly characterized regulators of actin-binding proteins are the small GTPases of the Rho family,9 which regulate actin dynamics by either directly interacting with other regulators of actin dynamics or regulating the activities of these proteins through phosphorylation pathways.
In addition to regulation by GTPases and the protein phosphorylation/dephosphorylation pathways, the dynamics of the actin cytoskeleton are regulated by phosphorylated phosphatidylinositol (PI).9 PI is sequentially phosphorylated by a PI 4-kinase to produce phosphatidylinositol 4-phosphate (PIP) and by a PI(4)P 5-kinase to generate phosphatidylinositol 4,5-bisphosphate (PI[4,5]P2). PI(4,5)P2 is a substrate for phospholipase C, yielding inositol 1,4,5-triphosphate (IP3) and diacylglycerol. Recent studies have shown that certain membrane phospholipids, especially PI(4,5)P2 and phosphatidylinositol 3,4,5-trisphosphate (PI[3,4,5]P3), regulate F-actin assembly in cells and cell extracts. PI(4,5)P2, which plays a critical role not only in generating second messengers such as IP3 and diacylglycerol but also in modulating a variety of cellular functions including cytoskeletal organization and membrane trafficking, appears to be a general regulator of actin polymerization at the plasma membrane or at membrane microdomains.9–11
In the present study, to address the question of which genes are influenced in hepatocytes in response to EHS gel, a model of the basement membrane matrix, we used cDNA microarray technology. We observed a remarkable level of PI(4,5)P2 phosphatase mRNA in hepatocytes cultured on EHS gel, resulting in a decline of PI(4,5)P2 and promotion of actin depolymerization in the cells. The organization of actin cytoskeleton via PI(4,5)P2 could be involved in the regulation of hepatocyte differentiation by ECM.
Adult rat hepatocytes were isolated from 8- to 12-week-old male Wistar rats by an in situ 0.05% collagenase perfusion method as previously reported.12 Viability of the hepatocytes was tested using trypan blue dye exclusion and found to always be above 85%. The freshly isolated hepatocytes were plated on 100-mm tissue culture dishes at a concentration of 5 × 105 cells/mL in Williams' medium E (Gibco, Carlsbad, CA) supplemented with 10−7 mol/L insulin (Biosourse, Camarillo, CA), 10−6 mol/L dexamethasone (Wako, Osaka, Japan), 1% penicillin and streptomycin (Gibco BRL, Rockville, MD), and 10% heat-inactivated fetal bovine serum (Wako). Hepatocytes were cultured on dishes coated with thin EHS, laminin, type IV collagen (all from Beckton Dickinsin Labware, Bedford, MA), or type I collagen (Iwaki/Asahi Techno Glass, Tokyo, Japan) or on plastic dishes coated with EHS gel, laminin gel, laminin/entactin complex gel, type IV collagen gel (all from Beckton Dickinsin Labware), or type I collagen gel (Iwaki/Asahi Techno Glass). The culture medium was changed after 4 hours and every 24 hours thereafter with fresh medium composed of Williams' medium E supplemented with 10−9 M insulin, 10−8 mol/L dexamethasone, 1% penicillin and streptomycin, and 5% heat-inactivated fetal bovine serum. Hepatocytes were treated with PI(4,5)P2 (EMD Biosciences, Darmstadt, Germany), U73122 (Biosourse), gelsolin (Cytoskeleton, Denver, CO), and rabbit polyclonal antibodies (Abs) M-106 and H-101 against integrin β subunits 1 and 4, respectively (Santa Cruz Biotechnology, Santa Cruz, CA), which are reactive for rat integrins β1 and β4.
An atlas glass array was used according to the manufacturer's instructions to compare mRNA expression patterns in cells cultured on type I collagen with those cultured on EHS gel. Total RNA was isolated from cells cultured on type I collagen–coated dishes and on EHS gel using ISOGEN (Nippon Gene, Tokyo, Japan) according to the manufacturer's instructions, and poly(A) RNA was purified using an Oligotex-dT30 mRNA purification kit (Takara, Osaka, Japan). One microgram of highly purified mRNA was reverse-transcribed and labeled with 5′ Cy3- or Cy5-labeled random 9-mers (Amersham Biosciences, Buckinghamshire, U.K.) using an Atlas PowerScript Fluorescent Labeling Kit (Clontech, Palo Alto, CA). The paired reactions were combined and purified with a Quantum Prep PCR Kleen Spin column (Bio-Rad, Hercules, CA). The fluorescently labeled probe was then applied to a microarray glass (Atlas Glass Rat 3.8 I Microarray; Clontech) containing 3,757 supplied clones and hybridized at 50°C for 24 hours. The microarray was scanned at a resolution of 10 μm in order to detect Cy3 and Cy5 fluorescence. After scanning the signal, expression analyses were performed using GenePix 4000A and GenePix 3.0 software (Axon Instruments, Union City, CA).
Northern Blot Analysis and Reverse-Transcription Polymerase Chain Reaction Analysis.
Total RNA (30-50 μg) was subjected to electrophoresis on a 1% agarose-formaldehyde gel and transferred to positive-charged nylon membranes (GeneScreen Plus; PerkinElmer Life Sciences, Boston, MA). The membranes were hybridized with cDNA probes for rat albumin, PI(4,5)P2 phosphatase, PI(4)P 5-kinase, β-actin (Rat Actin Beta Primer Set Kit; Maxim Biotech, South San Francisco, CA), cofilin, LIM kinase, GAPDH (rat GAPDH Primer Set Kit; Maxim Biotech), cytochrome P450 (CYP) 2B3, CYP1A1, and hepatocyte nuclear factor 4 (HNF-4), each of which was created by the PCR method using a digoxigenin (DIG) luminescent labeling kit (Roche Diagnostics, Mannheim, Germany). Total RNA (3 μg) from hepatocytes was reverse-transcribed by Bca PLUS reverse transcriptase (Takarabio, Shiga, Japan) according to the manufacturer's instructions. Thirty cycles of PCR amplification were performed at 94°C for 1 min, 59°C for 1 min, and 72°C for 1 min. The primers are listed in Table 1.
Hepatocytes cultured on type I collagen or EHS gel were fixed with 4% paraformaldehyde for 20 minutes and permeabilized with 0.1% Triton X-100 in PBS for 5 minutes. For F-actin staining, the cells were stained with 1.0 U/100 μL FITC-conjugated phalloidin (Molecular Probes, Eugene, OR) for 30 minutes. For β1 and β4 integrin staining, cells were resuspended in 1% PBS-BSA containing rabbit polyclonal Abs M-106 and H-101 or a control isotype, IgG2b (Echelon, Salt Lake City, UT), and stained with FITC-conjugated goat antirabbit IgG2b (Southern Biotechnology, Birmingham, AL). Cells were examined by fluorescence microscopy.
Rat hepatocytes cultured in each set of culture conditions were washed and resuspended in 0.5% PBS-BSA. The cells were fixed with 4% paraformaldehyde for 20 minutes at room temperature. The cells were resuspended in PBS containing 0.1% saponin and anti-PI(4,5)P2 Ab (Echelon) or control IgG2b for 30 minutes, stained with the FITC-conjugated goat antimouse IgG2b (Southern Biotechnology), and observed by fluorescence microscopy. According to the technical specifications, the cross-reactivity of this anti-PI(4,5)P2 with 1-palmitoyl-oleoylphosphatidic acid is less than 5%, that with phosphatidylinositol phosphate is less than 0.1%, and that with phosphatidylinositol, phosphatidylcholine, phosphatidylserine, phosphatidylethanolamine, cardiolipin, phosphatidyl glycerol, cholesterol, and 1,2-diacylglycerol is less than 0.2% each.13
Morphology and Function of Hepatocytes.
Hepatocytes cultured on EHS gel retained their spherical shape (Fig. 1A) and showed a strong hybridization signal for albumin mRNA expression (Fig. 1B-C) compared with those on type I collagen, as previously reported.14, 15 The composition of the EHS gel resembles that of the basement membrane in containing laminin (60%), type IV collagen (30%), heparan sulfate proteoglycan (3%), and entactin (1%).16 Which single component in the EHS substrata interacts with hepatocytes to affect the expression of specific genes missing from the plastic or type I collagen substratum is unclear. Hepatocytes cultured on laminin/entactin complex gel or laminin gel retained spherical-shaped cells on the EHS gel (Fig. 1A). Hepatocytes on laminin-coated dishes exhibited slight spreading, and cells on the type IV collagen gel or coat became flattened. As shown in Fig. 1B, hepatocytes cultured on laminin gel showed albumin mRNA to a lesser extent than did hepatocytes on the EHS gel. Hepatocytes cultured on the type IV collagen gel, the type IV collagen coat, or the laminin coat showed little albumin mRNA. On the other hand, hepatocytes on the laminin/entactin complex gel showed levels of albumin mRNA similar to those on the EHS gel. These findings suggest that several matrix components including laminin and entactin may play a pivotal role in a coordinated way in regulating hepatocyte functions.
Next, to elucidate whether the state of the matrix affects hepatocyte function, we compared albumin mRNA of hepatocytes cultured on EHS gel, EHS coat, type I collagen gel, or type I collagen coat. Hepatocytes cultured on EHS gel or EHS coat showed comparable levels of albumin mRNA (Fig. 1C). However, hepatocytes cultured on type I collagen gel expressed albumin mRNA to a greater extent than did cells on type I collagen coat, as previously reported.12 On the basis of these results, it is likely the effects of EHS on hepatocytes are not influenced by the state of the matrix unlike with type I collagen.
It is known that many interactions between cells and ECM are mediated by the integrin family of cell surface receptors,17 and the signals through integrins may be able to modify the phosphatidylinositol pathway or modulate F-actin assembly. We investigated whether the ability of EHS gel to affect cell differentiation and the phosphatidylinositol pathway was mediated by one or more integrins. Because it was reported that MDA-MB-231 cell adhesion to laminin is cooperatively controlled by β1 and β4 integrins18 and also we could show the expressions of β1 and β4 integrins in cultured hepatocytes by immunocytochemistry and reverse-transcription polymerase chain reaction (RT-PCR), shown in Fig. 1D, we determined the mRNA expression of albumin, PI(4,5)P2 phosphatase, and β-actin of hepatocytes cultured on EHS gel or type I collagen in the absence or presence of single or multiple blocking Abs against integrin β subunits 1 and 4 (Fig. 1E). In the presence of EHS gel, the mRNA expression of albumin or PI(4,5)P2 phosphatase was down-regulated when β4 integrin was blocked by single-antibody treatment, but not when β1 integrin was blocked. In the presence of type I collagen, Abs against integrin β subunits 1 and 4 did not affect mRNA expression of albumin and PI(4,5)P2 phosphatase. Treatment with an Ab against β4 integrin up-regulated β-actin mRNA in hepatocytes cultured on EHS gel or type I collagen. Treatment with the β1 integrin Ab did not alter the morphology of hepatocytes on EHS, but hepatocytes treated with the β4 integrin Ab exhibited slight spreading (Fig. 1F). Neither Ab affected the morphology of hepatocytes on type I collagen. These findings suggest that the interactions between hepatocytes and the EHS gel that control cell differentiation, the phosphatidylinositol pathway, and F-actin assembly may be mediated mainly by β4 integrin.
Gene Expression Profiling of EHS Gel-Modulated Genes.
cDNA microarray analysis was performed to identify genes that are associated with hepatocyte differentiation induced by the EHS gel. Genes whose expression changed more than twofold were considered positive. Three independent experiments were conducted to compare hepatocytes cultured on collagen with those cultured on EHS gel. Duplicate analyses were conducted on a single mRNA preparation from hepatocytes. After searching for genes that were up- or down-regulated in at least two of the three experiments, we found 84 genes up-regulated and 273 genes down-regulated by EHS gel on day 1 and 236 genes up-regulated and 115 genes down-regulated on day 3 (Fig. 2A).
Genes were classified on the basis of the biological function of the encoded protein, using a modified version of a previously established classification scheme.19 The classification scheme was composed of seven major functional categories, each of which contained several minor functional categories. Genes observed in the present study were placed in a single major class if a function of the encoded protein had been well established. As shown in Fig. 2B, genes involved in cell signaling/cell communication and metabolism were more selectively induced by EHS gel on days 1 and 3, as compared with the gene expression profiling of all the 3,757 genes. Genes involved in gene/protein expression were more selectively suppressed by EHS gel on days 1 and 3, and genes involved in cell structure/motility were more selectively suppressed by EHS gel on day 3. We concentrated on 21 genes in involved in cell signaling/cell communication that were up-regulated on day 1 and 6 genes involved in cell structure/motility down-regulated on day 3; the results are summarized in Table 2. After careful examination of the 27 genes, we found two important genes, PI(4,5)P2 phosphatase and cofilin, associated with actin polymerization and depolymerization, respectively, that were involved in the up-regulation and down-regulation of all 27 genes. Therefore, we focused on the roles of PI(4,5)P2 on hepatocyte differentiation induced by EHS gel.
Table 2. Genes Whose Expression Was Altered More Than 2-fold
NOTE. Rat hepatocytes were cultured on EHS gel or type 1 collagen. mRNAs were isolated 1 day or 3 days after inoculation, reverse-transcribed into fluorescent-labeled cDNA, and hybrized to a glass microarray containing 3,757 genes. There were 21 upregulated and 6 downregulated genes that showed a more than 2-fold change in expression ratio (EHS gel/collagen) in at least two of the three experiments.
To confirm the mRNA expression observed by microarray, Northern blot analysis was performed (Fig. 3). mRNA expression of PI(4,5)P2 phosphatase was up-regulated by EHS gel in a time-dependent manner. In contrast, mRNA expression of cofilin was depressed in hepatocytes on EHS gel during the culture period, although this expression was up-regulated in hepatocytes on type I collagen. mRNA expression of PI(4)P 5-kinase, by which PI(4)P is converted to PI(4,5)P2, was decreased, and mRNA expression of LIM kinase, by which cofilin is phosphorylated and inactivated, was increased by EHS gel compared with type I collagen. These results suggested that PI(4,5)P2 might be decreased in hepatocytes cultured on EHS gel and that actin depolymerization might be promoted in hepatocytes cultured on EHS gel. In fact, mRNA expression of β-actin was enhanced in hepatocytes cultured on type I collagen compared with those on EHS gel, which was contrary to the mRNA expression of hepatocyte-specific genes, including albumin, CYP2B3, and CYP1A1 (Fig. 3).
Decrease in PI(4,5)P2 Staining of Hepatocytes on EHS Gel.
In the Northern blot analysis, induction of PI(4,5)P2 phosphatase mRNA and reduction of PI(4)P 5-kinase mRNA were observed in hepatocytes cultured on EHS gel. These findings suggested that EHS gel could affect the level of PI(4,5)P2 in cultured hepatocytes. PI(4,5)P2 can be stained in cells with an anti-PI(4,5)P2 Ab.13, 20 As shown in Fig. 4, hepatocytes cultured on EHS gel exhibited a decreased level of PI(4,5)P2 staining compared with cells on collagen.
Effects of PI(4,5)P2 on Morphology and Actin Cytoskeleton of Hepatocytes.
PI(4,5)P2 regulates actin organization by modulating the activities of various actin regulatory proteins, and activates actin polymerization in response to extracellular stimuli.10, 21–24 As shown in Fig. 5, hepatocytes cultured on type I collagen assumed a flattened and extended shape. In contrast, cells cultured on EHS gel exhibited minimal spreading and retained a spherical shape as previously reported.14, 15 However, changes in the shape of hepatocytes on EHS gel were observed in the presence of PI(4,5)P2 or U73122, a phospholipase C inhibitor. Treatment with these agents resulted in the cells being flattened and spreading on the EHS gel such that they resembled cells cultured on collagen. On the other hand, treatment with gelsolin resulted in hepatocytes cultured on collagen that were cuboidal in shape, although treatment with PI(4,5)P2 or U73122 had little effect on the morphology of the hepatocytes cultured on collagen.
To ascertain the functional consequences of PI(4,5)P2 and cofilin on the actin cytoskeleton, F-actin was visualized with FITC-conjugated phalloidin (Fig. 6). F-actin was totally suppressed in hepatocytes cultured on EHS gel but not in hepatocytes cultured on type I collagen. In the presence of PI(4,5)P2 or U73122, F-actin increased in hepatocytes cultured on both EHS gel and collagen. In contrast, treatment with cofilin or gelsolin suppressed F-actin in hepatocytes cultured on collagen. These results indicated that hepatocytes cultured on EHS gel showed marked actin depolymerization and that PI(4,5)P2 and cofilin promoted actin polymerization and depolymerization in cells, respectively.
Effects of Actin Cytoskeleton on Hepatocyte-Specific Gene Expression.
To clarify the interaction between the actin cytoskeleton and hepatocyte function in response to ECM, mRNA expression in hepatocytes of albumin and HNF-4, a liver-enriched transcription factor, was determined (Fig. 7). Regulation of hepatic genes was implicated in actin remodeling. Northern blot analysis revealed that hepatocytes on EHS gel showed a higher level of albumin mRNA than did cells on collagen, as previously reported.15 Treatment with PI(4,5)P2 or U73122 resulted in decreased expression of albumin mRNA in hepatocytes on EHS gel. In contrast, albumin mRNA in hepatocytes on collagen increased in the presence of gelsolin. The alterations in expression of albumin mRNA were synchronized with those of HNF-4 mRNA. Expression of β-actin mRNA was negatively correlated with that of albumin and HNF-4 mRNA. Treatment with PI(4,5)P2 or U73122 increased β-actin mRNA levels in hepatocytes on EHS gel, and treatment with gelsolin decreased β-actin mRNA levels in hepatocytes on collagen. When hepatocytes were incubated in the presence of gelsolin after 24 hours of treatment with PI(4,5)P2, both hepatocytes on EHS gel and hepatocytes on collagen showed increased expression of albumin and HNF-4 mRNA and decreased levels of phalloidin-stained intracellular F-actin (data not shown).
We and other investigators have reported that ECM to which hepatocytes were attached modulated liver-specific genes and transcription factors in a coordinated manner, probably through morphological changes. We previously reported that EHS gel, a basement membrane matrix, is the most effective matrix for maintaining hepatocytes in cell culture with a differentiated phenotype.15 The cells cultured on EHS gel formed small aggregates of round cells, expressed less actin mRNA, and maintained abundant liver-specific gene expression throughout the culture period, whereas on collagen, they formed a cobblestone arrangement of well-spread cells, higher actin mRNA expression, and rapidly lost liver-specific mRNA such as albumin and HNF-4. A number of studies have indicated the importance of overall cell morphology in regulating tissue-specific gene expression. Ben-Ze'ev et al. reported that the organization of the cytoskeleton, which is dictated by the extent of cell-cell and cell-matrix interaction, was intimately associated with mechanisms that regulated tissue-specific gene expression.25 Similar results have been reported for hepatocytes; for example, Yuasa et al. found that hepatocytes proceed through the cell cycle with their differentiated functions suppressed on conversion from a spheroid to a monolayer culture.26 Recent studies have documented the importance of the Rho GTPases, which were originally isolated during the search for proteins homologous to the Ras proto-oncoprotein, for the heterogeneity of the actin cytoskeleton. The Rho GTPases are important regulators of the actin cytoskeleton and the morphological heterogeneity of cells27 and are involved in a number of vital cellular processes including transcriptional regulation via the c-Jun N-terminal kinase signaling pathway, induction of apoptosis, and cell-cycle control.28 However, the precise mechanisms by which the ECM regulates the organization of the actin cytoskeleton in hepatocytes and consequently determines the hepatocyte phenotype remain unclear.
The present cDNA microarray study revealed that cell signaling/cell communication–associated genes were up-regulated and that gene/protein expression–associated genes were down-regulated by EHS gel. These genes included PI(4,5)P2 phosphatase and cofilin, both of which are known to be associated with actin-binding proteins.9 The structure and dynamics of the actin cytoskeleton are regulated by an array of actin-binding proteins that interact with monomers or filaments. A number of recent studies have indicated that phosphoinositide signaling plays a pivotal role in the regulation of the actin cytoskeleton. The various phosphorylated phosphoinositides control the activity and localizations of numerous actin-binding proteins and of other proteins that regulate the actin cytoskeleton. Actin-binding proteins are the most commonly affected by PI(4,5)P2.9 PI(4,5)P2 induces actin assembly in Xenopus extracts, suggesting that this phospholipid can regulate actin dynamics in vivo.29 PI(4)P 5-kinase overexpression dramatically increases actin polymerization in CV1 cells.24 Overexpressing PI(4,5)P2 phosphatase, which causes a reduction of PI(4,5)P2 levels, inhibits actin polymerization in COS-7 cells.30 In the present study, attachment to EHS gel caused an increase in PI(4,5)P2 phosphatase mRNA and a decrease in PI(4)P 5-kinase mRNA in cultured rat hepatocytes and resulted in a reduction of PI(4,5)P2 levels in hepatocytes. These findings indicated that inositol lipid signaling in response to EHS gel influenced the remodeling of the actin cytoskeleton by changing PI(4,5)P2 levels in hepatocytes. To our knowledge, this is the first study to show that PI(4,5)P2 phosphatase mRNA was regulated through the β4 integrin receptor, although it has been demonstrated that the small GTPase Rac1-dependent pathway mediated by integrin receptors involves recruitment of PI(4)P 5-kinase to form PI(4,5)P231 and that α3β1 integrin is important for the attachment of hepatocytes to laminin and collagen and also for the expression of albumin mRNA.32
Indeed, actin polymerization was suppressed to a greater degree in hepatocytes cultured on EHS gel than in those cultured on type I collagen (Fig. 6). Moreover, treatment with PI(4,5)P2 or U73122, which can induce expression of PI(4,5)P2 by inhibiting phospholipase C, promoted actin polymerization in both hepatocytes cultured on EHS gel and hepatocytes cultured on type I collagen. Gelsolin is the most potent actin filament-severing protein and is likely to be inhibited by PI(4,5)P2.33 Cofilin enhances F-actin turnover by depolymerizing filaments from its minus ends34 and by severing F-actin.35 These F-actin-disrupting agents inhibited the localization of actin filaments in hepatocytes on type I collagen (Fig. 6). The reorganizations of the actin cytoskeleton was related to changes in cell shape induced by PI(4,5)P2, U73122, or gelsolin (Fig. 5). Bhadriraju et al. demonstrated that disrupting F-actin assembly reduced cell spreading,36 which is consistent with our results.
To assess whether cytoskeletal F-actin assembly plays a role in differential cell function, we examined the effects of actin-binding proteins on the expression of hepatocyte-specific genes (Fig. 7). PI(4,5)P2 and U73122 decreased the expressions of albumin and HNF-4 mRNA in hepatocytes cultured on EHS gel. In contrast, gelsolin increased albumin and HNF-4 mRNA in hepatocytes cultured on type I collagen. Hepatocytes treated with gelsolin after pretreatment with PI(4,5)P2, which finally exhibited actin depolymerization (data not shown), showed increased expression of albumin and HNF-4 mRNA. These findings indicate that cells exhibiting actin depolymerization undergo differentiation, whereas cells in which actin polymerization is enhanced undergo dedifferentiation. In general, higher levels of differentiation-specific functions are exhibited by cells prevented from extending by modulating cell-ECM binding.37, 38 Our results show that similar control of hepatocyte shape and function can be achieved by altering actin assembly. However, the precise mechanisms by which actin polymerization or depolymerization influences hepatocyte differentiated functions are still unknown. The changes in cellular levels of albumin mRNA, which were regulated by actin-binding proteins such as PI(4,5)P2 and gelsolin, corresponded with those of HNF-4 (Fig. 7). HNF-4 is a major transcription factor for various liver-specific genes, including those regulating protein synthesis and lipid, glucose, and drug metabolism.39 Cytoskeletal organization is likely to determine and maintain the differentiated phenotype of hepatocytes by regulating the liver-specific transcription factor HNF-4. Further studies are needed to define the signal transduction pathway from actin reorganization that regulates liver-specific transcription factors.
In summary, we have demonstrated a link between ECM, actin-binding proteins, the actin cytoskeleton, and hepatocyte differentiation (Fig. 8). Our results showed that cell adhesion to ECM induces changes in the assembly of actin filaments through an actin-binding protein, PI(4,5)P2. The changes in cytoskeletal filament assembly that occurred following cell attachment to ECM and/or alterations in the pattern of actin-binding proteins played crucial roles in determining the differentiated phenotype of hepatocytes.
The authors thank Dr. Yoshiko Banno (Department of Biochemistry, Gifu, Japan) for her advice on PI(4,5)P2 and PI(4,5)P2 staining.