2971
Journal of Cell Science 112, 2971-2981 (1999) Printed in Great Britain © The Company of Biologists Limited 1999 JCS4655
Regulation of the hepatocyte cell cycle by type I collagen matrix: role of cyclin D1 Linda K. Hansen1,2,* and Jeffrey H. Albrecht2,3,4 1Department
of Laboratory Medicine and Pathology, and 2Cancer Center, and 3Department of Medicine, University of Minnesota, Minneapolis, MN 55455, USA 4Division of Gastroenterology, Hennepin County Medical Center, Minneapolis, MN 55414, USA *Author for correspondence (e-mail:
[email protected])
Accepted 8 July 1999; published on WWW 12 August 1999
SUMMARY Rat hepatocytes adherent to a rigid film of type I collagen will spread and enter S phase, while those attached to collagen gel or a dried collagen substrate remain round and quiescent. The current studies were initiated to determine the mechanism by which these different substrates differentially influence cell cycle progression. Cyclin D1 mRNA and protein expression and associated kinase activity was low on dried collagen relative to collagen film. In contrast, cyclin E and cdk2 protein levels were similar on the two substrates. Although cyclin E and cdk2 were present, cells on dried collagen lacked cdk2 kinase activity. p27 protein levels did not differ between dried collagen and film, but more p27 was associated with cdk2 in cells on
dried collagen than those on collagen film. Cyclin D1 expression on collagen film was inhibited by cytochalasin D and exoenzyme C3, suggesting a role for the GTPbinding protein, Rho, in regulating cyclin D1 expression. Cyclin D1 over-expression induced hepatocytes into S phase in the absence of cell shape change on dried collagen or collagen gel. These results demonstrate a novel, substrate-dependent mechanism for cyclin D1 expression in hepatocytes, and also demonstrate that cyclin D1 overexpression allows shape-independent S phase entry.
INTRODUCTION
This active enzymatic complex in turn phosphorylates a number of regulatory proteins important in cell cycle progression. For example, cyclin D1, expressed in mid-G1, forms a complex with cdk4. Once activated, this cyclin D1/cdk4 complex phosphorylates ligands such as the retinoblastoma (Rb) protein (Sherr, 1994), which regulates the activity of transcription factors. Thus, formation of an active cyclin/cdk complex initiates a cascade of molecular events necessary for proper cell cycle progression. Many lines of evidence suggest that the adhesion dependent signaling event in late G1 may be related to cyclin expression and kinase activity. Cyclin A expression in NRK fibroblasts occurs only in adherent, not suspended, cells, while cyclin D1 appears to be regulated by adhesion in NIH3T3 and Rat1 fibroblasts (Guadagno et al., 1993; Zhu et al., 1996; Resnitzky, 1997). Transformation of NRK fibroblasts with human cyclin A gene bypasses the adhesion requirement at the G1/S border and allows fibroblast proliferation in suspension (Guadagno et al., 1993). However, these studies compared only suspended to fully adherent, spread cells. Growth arrest also occurs on certain adhesive substrates, such as a gel composed of type I collagen (Santhosh and Sudhakaran, 1994) or certain adhesive peptides (Hansen et al., 1994), suggesting that adhesion, per se, may not be sufficient for growth activation. Several studies demonstrate that the physical structure of the
It has long been recognized that cells require adhesion to a solid substratum in order to proliferate. Non-transformed cells cultured in suspension in the presence of growth factors fail to undergo DNA synthesis (MacPherson and Montagnier, 1964; Benecke et al., 1978), and the loss of ‘anchorage-dependence’ is a hallmark of cell transformation (Stoker et al., 1968; Shin et al., 1975). Several studies have mapped the adhesiondependent step to mid- to late-G1 (Otsuka and Moskowitz, 1975; Guadagno and Assoian, 1991; Han et al., 1993), but the mechanism by which adhesion regulates the G1/S phase transition remains unclear. The transition from G1 to S phase is an important regulatory point, or checkpoint, in the cell cycle. A well-ordered sequence of molecular events must occur for proper transit through this checkpoint, and inhibition of these events leads to growth arrest at a point in late G1 called the restriction point (R). Members of the cyclin family of proteins, along with cyclin-dependent kinases (cdk), act as important regulators of this and other cell cycle arrest points. The expression of specific cyclin genes is regulated in a cellcycle dependent manner, and when expressed at sufficient levels, cyclins associate with a specific cdk. Complexed cdks are then capable of being phosphorylated and thus activated.
Key words: Cell cycle, Cyclin D1, Type I collagen, Hepatocyte, Extracellular matrix
2972 L. K. Hansen and J. H. Albrecht ECM regulates both cell morphology and function. The ability of a matrix to support cell spreading correlates with its ability to promote cell growth (Folkman and Moscona, 1978; Watt et al., 1988; Ingber, 1990; Hansen et al., 1994; Singhvi et al., 1994). In contrast, adhesion to substrates that do not permit cell spreading correlates with DNA synthesis inhibition (Hansen et al., 1994; Singhvi et al., 1994). The physical structure of a type I collagen substrate, for example, can be altered to create a rigid film, which promotes cell spreading and growth (Mooney et al., 1992a), or a malleable gel, which maintains a rounded morphology and differentiated phenotype (Santhosh and Sudhakaran, 1994). In addition, cellular response to native versus denatured type I collagen substrates may differ (Morton et al., 1994). The molecular mechanism by which physical structure of the ECM regulates cell growth and function has not been elucidated. Hepatocytes provide an ideal primary cell system with which to study adhesion-dependent cell cycle progression. These epithelial cells normally exist in the liver in a resting, or G0, state and they rapidly respond to in vivo signals, such as partial hepatectomy or liver damage, with a burst of synchronous proliferation. This process of liver regeneration is accompanied by a well controlled pattern of cyclin expression and activity (Lu et al., 1992; Albrecht et al., 1993; Loyer et al., 1994). Furthermore, regulation of hepatocyte phenotype by collagen structure may be important in in vivo conditions such as liver fibrosis, in which both the amount and structure of type I collagen are altered (Scott et al., 1994; Ricard-Blum et al., 1996). Different type I collagen substrates are known to induce different phenotypes of isolated hepatocytes. The goal of this study is to characterize isolated rat hepatocyte cell cycle progression on different type I collagen substrates, and to determine the role of cyclin expression and activity on the substrates which differ in their ability to promote progression into S phase. MATERIALS AND METHODS Reagents Molecular biology reagents were generously provided by the following investigators: cyclin D1 cDNA probe by Dr Steven Reed; retinoblastoma-glutathione-S-transferase (Rb-GST) by Dr Jean Wang; and β-gal adenovirus and 293 cells by Dr Howard Towle. Hepatocyte culture Primary rat hepatocytes were obtained by collagenase perfusion of adult Lewis rat liver (Seglen, 1976), followed by purification through a Percoll gradient (Sigma Chemical Company, St Louis, MO). Only cells from harvests yielding >90% viability were used. Hepatocytes were plated at sub-confluent density (10-12,000 cells/cm2) in serum-free William’s medium E (Gibco, Grand Island, NY) with the following additives as previously described (Hansen et al., 1994): epidermal growth factor (10 ng/ml, Collaborative Research, Bedford, MA), insulin (20 mU/ml, Sigma), dexamethasone (5 nM, Sigma), sodium pyruvate (1 mM, Gibco), ascorbic acid (50 µg/ml, Gibco), and penicillin/streptomycin (100 U/ml, Irvine Scientific, Santa Ana, CA). Cultures were refed daily. Cultures were always performed in the presence of growth factors unless otherwise stated. For growth factor-free conditions, both EGF and insulin were excluded. Collagen substrate preparation Type I collagen (‘Vitrogen 100’, Collagen Biomaterials, Palo Alto,
CA) was coated onto non-adhesive Petri dishes using three different techniques. The first substrate is prepared by coating dishes with collagen diluted in a basic, carbonate buffer (pH 9.4) as previously described (Mooney et al., 1992a). Collagen was coated at 1 µg/cm2 (5-9 µg/ml depending on the plate size) with an approximate 65% coating efficiency, yielding about 0.65 µg/cm2. Plates are incubated in this solution at 4°C overnight. This preparation is believed to promote adsorption of single collagen molecules onto the plastic surface, creating a thin, rigid film (Ingber, 1990). This preparation is referred as ‘collagen film’ throughout this report. The second coating method consists of diluting collagen to approximately 0.4 mg/ml in 0.1% acetic acid, adding a thin layer onto a Petri dish (e.g. 10 ml/100 mm dish), and drying overnight at 55°C (Berry et al., 1991). The coating efficiency was determined to be approximately 26%, yielding a final coating density of 13 µg/cm2. This substrate is referred to as ‘dried collagen’ throughout this report. Finally, to create a ‘gel’, collagen was diluted 1:7 in phosphate buffered saline (PBS) or Williams E medium without growth factors, added to a Petri dish at a volume equivalent to the film or dried collagen preparations, and incubated at 37°C for one hour. After coating, all plates were washed twice in PBS and incubated at least 20 minutes in 1% bovine serum albumin (BSA) in Williams E medium to block any non-coated sites. Cell spreading analysis Hepatocytes cultured on the collagen substrates were rinsed twice with PBS, then fixed for 10 minutes in 1% glutaraldehyde, dehydrated with two methanol washes, and stained with Coomassie blue. Average cell area was obtained using computerized image analysis, consisting of a Nikon Diaphot microscope with camera hooked up to a Compac 486 computer with Optronics Engineering video imaging system and BioScan (Optimas, Inc.) software. This system allows projected cell area measurements of multiple cells simultaneously. 50-100 cells were measured for each condition. Data is presented as mean µm2 ± s.d. DNA synthesis DNA synthesis is measured by adding [3H]thymidine at 10 µCi/ml (specific activity = 50 Ci/mmol, ICN Biomedicals, Costa Mesa, CA) to 96-well plate cultures 52 hours after plating, a time corresponding with early S phase under these conditions. After a 16 hour incubation, [3H]thymidine incorporation into newly synthesized DNA was determined by harvesting cell lysates onto filter paper using a cell harvester (Brandel, Gaithersburg, MD) and quantitation by scintillation counting. Counts were normalized to cell number determined in parallel plates using the CyQuant Assay (Molecular Probes, Eugene OR), assessed prior to the onset of DNA replication. Albumin secretion Cell culture medium was changed after 48 hours in culture with fresh medium. 24 hours later, medium samples were collected and frozen. Frozen samples were thawed and analyzed using a competitive ELISA (Peshwa et al., 1994). At the time of medium collection, adherent cells were trypsinized and counted on a Coulter Z2 particle analyzer (Beckman Coulter, Inc., Fullerton, CA). RNA analysis Total cellular RNA was obtained using a guanidinium isothiocyanate lysis solution, purified, and analyzed by northern blot analysis (20 µg/lane) (Albrecht et al., 1995). The EcoRI/HindIII fragment of human cyclin D1 cDNA (Lew et al., 1991) was used to generate probe by random priming. Autoradiography with a phosphoimager was used for analysis. Each experiment includes RNA from freshly isolated cells as a baseline control (t=0 hours). Protein isolation, western blot, and kinase assays Cultured hepatocytes were removed from the plates with collagenase and scraping, then cells were homogenized in a Tween-20 lysis buffer
Hepatocyte growth regulation by collagen 2973 (Albrecht et al., 1998). Western blot was performed using 10 µg of cellular protein per lane as previously described (Albrecht et al., 1998). Antibodies used for westerns include anti-rat cyclin E (SC-481, Santa Cruz Biotechnology, Santa Cruz, CA), anti-mouse cyclin A (SC-596, Santa Cruz), anti-p27 (Transduction Laboratories, Lexington, KY), anti-cdk2 and cdk4 (Santa Cruz), and anti-human cyclin D1 (rabbit polyclonal, UBI, Lake Placid, NY, and clones DCS6 and DCS11, NeoMarkers, Union City, CA). Immunoprecipitation/kinase assays were performed as described (Albrecht et al., 1998) using 50 µg and 200 µg of cellular protein, respectively, for cdk2 and cyclin D1. DCS-11 and anti-human cdk2 (SC-163, Santa Cruz Biotechnology) were used for the kinase assays. Adenovirus transfection ADV-βgal (β-galactosidase adenovirus), pACCMV.pLpA, and pJM17 were generously provided by Dr Howard Towle (Becker et al., 1994; Kaytor et al., 1997). A recombinant cyclin D1 adenovirus (ADV-D1) was constructed by ligating the EcoRI/HindIII fragment of human cyclin D1 (generously provided by Dr Steven Reed) into pACCMV.pLpA. This recombinant plasmid was co-transfected with pJM17 into 293 cells as previously described (Becker et al., 1994). The resulting recombinant virus was purified by isolating agar plaques, which were then amplified in 293 cells. Southern blot analysis was used to verify the presence of the cyclin D1 gene. Once a clone was chosen and amplified, transfection of primary rat hepatocytes was performed. Rat hepatocytes were plated for three hours on collagen-coated plates, followed by a two hour incubation with the recombinant adenovirus. The adenovirus titer used was approximately TCID50/cell = 8 (TCID50 = ‘tissue culture infective dose’, or the viral dose which induces cell lysis in at least 50% of 293 cell cultures). Adenovirus containing the β-galactosidase gene was used at the equivalent titer as a control. After the two hour incubation, the media containing the adenovirus was removed and replaced with virus-free medium.
RESULTS Hepatocyte morphology on film, gel and dried collagen Type I collagen substrates can be manipulated into different forms. Three types of collagen preparations were used in these studies, resulting in a thin film, denatured and dried film, and a polymerized gel. Incubation of extracellular matrix molecules in a high pH carbonate buffer at 4°C is believed to precipitate the protein monomers on the plastic dish, creating a thin, rigid film (Ingber, 1990) (referred to as ‘film’ throughout this report). Preparation of the ‘dried’ collagen substrate involves dilution in dilute acetic acid followed by drying at 55°C, well above the melting temperature of collagen (~40°C)
Fig. 1. Hepatocytes spread more extensively on a film of type I collagen than on dried collagen or collagen gel. Primary rat hepatocytes were cultured on collagen film (‘film’), dried collagen (‘dried’), or collagen gel (‘gel’) for 48 hours, followed by fixation with glutaraldehyde and photography using phase contrast microscopy (×25).
(Linsenmayer, 1981), which results in a denatured substrate. In contrast, dilution of collagen in a neutral pH solution and incubation at 37°C creates a polymerized, gel-like substrate (‘gel’). Most comparisons in this report focus on collagen film and dried collagen since the dried and gel substrates consistently produced similar results. Data obtained from hepatocytes on the collagen gel is included in some figures to illustrate this point. Cell morphology was compared on each substrate. When plated at a low cell density (10,000 cells/cm2) for 48 hours, hepatocytes cultured on a thin film of type I collagen spread extensively, as reported previously (Mooney et al., 1992a). In contrast, hepatocytes on either dried collagen or collagen gel demonstrated little spreading after 48 hours in culture (Fig. 1). Instead, these cells remained rounded or slightly oblong in shape with light-refractile edges. While the rounded morphology on collagen gel has been described previously (Santhosh and Sudhakaran, 1994), hepatocytes on the dried substrate appeared more rounded than in previous reports (Ben-Ze’ev et al., 1988), which is likely due to heat denaturation during the drying procedure. The difference in spreading seen between collagen film and dried collagen was detectable as early as three hours after cell plating (Table 1), as determined by computerized image analysis, and the difference was maintained for the duration of all experiments. DNA synthesis and differentiated function on collagen substrates A correlation between cell spreading and DNA synthesis has been observed in a variety of cell types (Folkman and Moscona, 1978; Ingber, 1990; Hansen et al., 1994). To determine if this correlation is also observed in hepatocytes on each collagen substrate, [3H]thymidine was added to hepatocyte cultures 52-68 hours after plating to measure DNA synthesis. Previous studies of rat hepatocytes on growthpermissive matrices indicate that the cells enter S phase around 48 hours after plating, with peak DNA synthesis occurring at approximately 72 hours (Loyer et al., 1996; L. K. Hansen, unpublished data). In the presence of growth factors, cells on the collagen film exhibited a significantly greater level of DNA synthesis than cells on dried collagen or collagen gel (Fig. 2). The difference in DNA synthesis was not due to a difference in amount of collagen in each substrate, since varying the amount of collagen coated up to 60-fold higher or lower did not alter the difference between collagen film and dried collagen (data not shown). Hepatocytes cultured on a collagen gel are reported to
2974 L. K. Hansen and J. H. Albrecht µm2/cell) Table 1. Cell area on collagen substrates (µ Time (hours) after plating 3 24
Film
Dried
1342±195 1986±173
277±43 1115±239
Hepatocytes spread more readily on collagen film than dried collagen substrate. Hepatocytes were cultured on collagen film or dried collagen in the presence of growth factors for 3 or 24 hours, followed by fixation in 1% glutaraldehyde, and staining with Coomassie brilliant blue. Projected cell area was obtained using computerized image analysis.
possess a highly differentiated phenotype (Santhosh and Sudhakaran, 1994), and cell spreading in vitro is correlated with diminished differentiated function (Mooney et al., 1992a,b). To determine the effects of each collagen substrate on differentiated function, the secretion rate of albumin, a protein synthesized and secreted by differentiated hepatocytes, was measured between 48 and 72 hours after plating. Albumin secretion was maintained on each collagen substrate out to at least 72 hours; however, cells on collagen film secrete albumin at a lower rate that those on dried collagen or gel (Fig. 2). Thus, these rounded cells on gel or dried collagen not only maintained viability but they also maintained a higher degree of differentiated function than spread cells on collagen film.
DNA SYNTHESIS (% FILM)
15
DNA synth alb secretion
100 80
10
60 40
5
20 0
Film
Dried
Gel
ALBUMIN SECRETION (ng/ml/1000 cells/24 hr)
Cyclin expression and activity The inability of hepatocytes on dried collagen or collagen gel to enter S phase suggested a substrate-dependent growth arrest in G0/G1. Because progression through this phase is regulated by cyclins and their associated kinases, it was hypothesized that growth arrest occurs on dried collagen due to lack of expression of one or more of the G1 cyclins. The D-type cyclins (cyclin D1, D2, and D3) are the first cyclins to be expressed, and expression of cyclin D1 is dramatically upregulated in the prereplicative phase of liver regeneration in vivo and in cultured primary hepatocytes in vitro (Albrecht et al., 1993, 1995). To determine if cyclin D1 expression differs on the
different substrates, cyclin D1 mRNA expression was compared in hepatocytes cultured in the presence or absence of growth factors (EGF and insulin) for 52 and 72 hours, on either collagen film or dried collagen. Freshly isolated hepatocytes were used as the 0 hour timepoint. Because the data obtained to this point were similar on dried collagen and collagen gel, subsequent analyses were limited to comparison between collagen film and dried collagen, unless otherwise noted. Northern blot analysis demonstrated little cyclin D1 mRNA expressed on collagen film in the absence of growth factors, confirming the growth factor dependence of its expression as shown previously in other cell types (Grana and Reddy, 1995). Cyclin D1 was significantly up-regulated on collagen film in the presence of growth factors, yet its mRNA was barely detectable in cells cultured on dried collagen even in the presence of growth factors (Fig. 3A). To confirm that cyclin D1 protein expression was also inhibited on dried collagen, equal amount of cell lysates from hepatocytes cultured on dried collagen or film for 24, 48, and 72 hours were probed with anti-cyclin D1 antibody (UBI). Expression of cyclin D1 protein was significantly up-regulated on collagen film by 48 hours after plating (Fig. 3B), consistent with entry into S phase around 48-52 hours after plating. In contrast, significantly less cyclin D1 protein was seen on dried collagen at any timepoint examined. Similarly, cyclin D1 protein expression was also low in cells cultured on collagen gel (data not shown). The data presented here are the first to link ECM conditions to cyclin D1 mRNA and protein expression. In order to determine if the inhibition of gene expression on dried collagen was specific for cyclin D1 rather than a non-
A
B
0
ECM Condition Fig. 2. DNA synthesis is greater, but albumin secretion is lower, on collagen film than on dried collagen or collagen gel. Hepatocytes were cultured on collagen film, dried collagen, or collagen gel in the presence of growth factors. [3H]thymidine was added from 52-68 hours after plating, followed by cell lysis, harvest onto glass fiber paper, and scintillation counting. Data was normalized to attached cell number, and presented as a percentage of DNA synthesis observed on collagen film (± s.d.). Albumin secretion was measured from 48-72 hours using a sandwich ELISA and normalized to attached cell number.
Fig. 3. Cyclin D1 mRNA and protein expression is diminished on dried collagen relative to collagen film. (A) Total RNA was isolated from hepatocytes cultured in the presence or absence of growth factors for 52 or 72 hours on collagen film or dried collagen, or from freshly isolated hepatocytes (t=0). Northern analysis of equal RNA samples (20 µg RNA per lane) was performed using a human cyclin D1 cDNA probe. (B) Cell lysates were obtained from hepatocytes cultured on collagen film or dried collagen with growth factors for 24, 48, and 72 hours in culture. Equal protein samples (10 µg/lane) were separated by SDS-PAGE and probed by western analysis with anti-cyclin D1 antibody.
Hepatocyte growth regulation by collagen 2975 Fig. 4. Expression of several cell cycle regulatory proteins does not differ in hepatocytes cultured on collagen film or dried collagen. (A) Protein was isolated from hepatocytes cultured on collagen film or dried collagen for 24, 52, or 72 hours, and analyzed by western analysis using antibodies against cdk4, cdk2, cyclin E, and p27. (B) Protein was isolated from hepatocytes cultured on collagen film or dried collagen (dr) with (+) or without (−) growth factors (GF) for 72 hours, and probed by western analysis with anti-cyclin A antibody.
A
specific down-regulation of many genes, the expression of other G1 phase proteins was examined. Results of western blot analysis are shown in Fig. 4A,B. Cyclin E, cdk4, and cdk2 were all expressed at similar levels on collagen film and dried collagen (Fig. 4A), demonstrating that a general downregulation of gene expression was not occurring. p27, an inhibitor of G1 cyclin/cdk complexes (Polyak et al., 1994a), was also expressed similarly on collagen film and dried collagen, with perhaps a slight decline by 72 hours on dried collagen (Fig. 4A). However, cyclin A protein was substantially diminished on dried collagen relative to film (Fig. 4B). Thus, both cyclin D1 and A expression appeared to be inhibited by adhesion to dried collagen, while other G1 regulatory proteins were unaffected. These results are in sharp contrast with studies of smooth muscle cells (Koyama et al., 1996), in which p27 was upregulated but cyclin D1 protein expression was unaffected on polymerized collagen gel. Thus, substrate-dependent cyclin expression may be uniquely regulated in hepatocytes. To determine whether the low level of cyclin D1 protein observed on dried collagen (Fig. 3B) was sufficient to generate detectable cyclin D1/cdk4 kinase activity, kinase assays were performed by immunoprecipitating cell lysates from the different culture conditions with anti-cyclin D1 antibody. Immunoprecipitates were incubated with a genetically engineered form of the cyclin D1/cdk4 target, Rb (Rb-GST), and the level of phosphorylation of the Rb-GST substrate was determined (Matsushime et al., 1994; Connell-Crowley and Harper, 1995). Hepatocytes on collagen film expressed abundant cyclin D1/cdk4 activity, while kinase activity was inhibited by attachment to dried collagen (Fig. 5A). The kinase activity associated with cdk2, which associates with both cyclin E and A, was determined on each substrate condition in a kinase assay using anti-cdk2 immunoprecipitates and histone H1 as the enzyme substrate. Fig. 5B demonstrates that, although both cyclin E and cdk2 were present on either collagen substrate, little cdk2-associated kinase activity was detected in cells cultured on dried collagen. In contrast, cdk2associated kinase activity was abundant in cells cultured on collagen film. The lack of cdk2-associated kinase activity on dried collagen in spite of the presence of both cyclin E and cdk2 suggested a role for cdk inhibitors. Such inhibitors can bind to cyclin/cdk complexes and inhibit their activity. To test this possibility, p27 was examined for its association with cdk2 complexes by immunoprecipitating cell lysates with anti-p27 antibody (Santa Cruz), followed by anti-cdk2 western blot. While p27 levels were not higher on dried collagen than collagen film (Fig. 4A),
B
more p27 was associated with cdk2 in cells cultured on dried collagen than on film (Fig. 6A). These data support the hypothesis that adhesion to dried collagen inhibits cdk2associated kinase activity by promoting greater association of cdk2-complexes with p27, rather than inhibiting cdk2 expression. This hypothesis is also supported by studies suggesting that cyclin D1/cdk4 complexes serve as a reservoir for p27 (Poon et al., 1995). If this is the case, there should be is a greater pool of free p27 available to associate with cyclin E/cdk2 and cyclin A/cdk2 complexes, thus inhibiting their activity, under conditions of minimal cyclin D1 expression (i.e. dried collagen). On collagen film, sufficient cyclin D1/cdk4 complexes would be formed to redistribute the pool of p27 from cyclin E- and cyclin A-cdk2 complexes to cyclin D1/cdk4, thus allowing cdk2 complex activity. Since p27 may be less effective at inhibiting cyclin D complexes than cyclin E complexes (Polyak et al., 1994b), this redistribution of p27 may be permissive for G1 progression. To determine if more p27 was associated with cyclin D1 on collagen film, cyclin D1 was immunoprecipitated from cells adherent to film or dried collagen, and the immunoprecipitates were probed for p27
A
B
Fig. 5. Cyclin/cdk kinase activity is absent in cells cultured on dried collagen. (A) Immunoprecipitates were prepared from cell lysates using an anti-cyclin D1 antibody. The precipitates were incubated with Rb-GST substrate and [γ-32P]ATP, and analyzed by autoradiography. (B) Immunoprecipitates were prepared from cell lysates using an anti-cdk2 antibody. The precipitates were incubated with histone H1 substrate and [γ-32P]ATP, and analyzed by autoradiography.
2976 L. K. Hansen and J. H. Albrecht expression in hepatocytes occurs through an actin-dependent mechanism. Soluble mediators of the actin cytoskeleton also appear to be involved in cell growth, and these make ideal candidates for upstream regulators of ECM-dependent cyclin D1 expression. The Rho small GTPase family of proteins, which includes RhoA, rac, and cdc42, regulates changes in actin structure involved in cell spreading, filapodia formation, and membrane ruffling, respectively (Ridley and Hall, 1992; Ridley et al., 1992). Inhibition of Rho also results in G1 arrest (Yamamoto et al., 1993; Udagawa and McIntyre, 1996). It is possible that in primary rat hepatocytes, Rho family proteins may be required for cyclin D1 expression and their activation in hepatocytes is substrate dependent. To begin to test this hypothesis, the expression of cyclin D1 protein on collagen film was examined in the presence of C3 exoenzyme (Calbiochem), a specific inhibitor of RhoA. The concentration used (2 µg/ml) was sufficient to inhibit DNA synthesis in hepatocytes by 60% (data not shown). Cyclin D1 protein expression was suppressed in hepatocytes cultured in the presence of C3 exoenzyme (Fig. 7). These results support the hypothesis that mediators of the actin cytoskeleton regulate cyclin D1 expression in primary hepatocytes. Fig. 6. p27 association differs on collagen film and dried collagen. (A) Hepatocytes were cultured for 52 hours on the different substrates, at which time immunoprecipitation was performed using anti-p27 antibody. The immunoprecipitates were separated by SDSPAGE and analyzed by western blot analysis with anti-cdk2 antibody. (B) Hepatocytes were cultured for 52 hours on the different substrates, at which time immunoprecipitation was performed using anti-cyclin D1 antibody. The immunoprecipitates were separated by SDS-PAGE and analyzed by western blot analysis with anti-p27 antibody.
(Fig. 6B). Abundant p27 was associated with cyclin D1 on collagen film, but little is detected in cyclin D1 immunoprecipitates on the dried collagen substrate. Since the overall level of p27 does not differ between collagen film and dried collagen, these data suggest that the down-regulation of cyclin D1 on dried collagen not only prevents cyclin D1/cdk4dependent events, but also is responsible for inhibiting cdk2 complexes via the redistribution of p27.
Cyclin D1 over-expression The data presented thus far suggest that cyclin D1 is the ratelimiting step in hepatocyte cell cycle arrest on dried collagen. In order to test this possibility directly, the cyclin D1 gene was transfected into primary hepatocytes cultured on collagen film or dried collagen, and subsequent cell cycle progression was determined. A recombinant adenovirus vector was formed by inserting the human cyclin D1 gene driven by a CMV promoter into the adenovirus vector, to create the ADV-D1 virus. Adenovirus is a highly efficient vector for the introduction of foreign genes into hepatocytes (Gomez-Foix et al., 1992). Hepatocytes were plated on the different substrates for 3 hours, after which virus was added for a two hour period at the indicated viral titers. Virus was then removed and replaced with fresh, virus-free medium. Adenovirus containing a nuclear-localizing variant of the β-galactosidase gene was used as a negative control. To confirm the ability of ADV-D1-transfected hepatocytes
Role of actin cytoskeleton The lack of cyclin D1 mRNA or protein in hepatocytes cultured on dried collagen suggests that a substrate-dependent signaling mechanism lies upstream of cyclin D1 expression. Because of the correlation observed between cyclin D1 expression and cell morphology (Fig. 1A), it was hypothesized that the putative substrate-dependent regulator of cyclin D1 may also regulate the cytoskeleton. In fact, it has been shown that actin polymerization is required for substrate-dependent cyclin D1 expression in fibroblasts (Bohmer et al., 1996). To test this hypothesis in hepatocytes, cyclin D1 protein expression was examined in hepatocytes on collagen film cultured in the presence of different inhibitors of the actin cytoskeleton. Hepatocytes were cultured on collagen film in the presence of cytochalasin D at 1 µg/ml. Fig. 7 demonstrates that little if any cyclin D1 protein was detectable in hepatocytes cultured for 52 hours on collagen film in the presence of GFs and cytochalasin D (CD). These data indicate that ECM regulation of cyclin
Fig. 7. Cyclin D1 protein expression in hepatocytes is inhibited by mediators of the actin cytoskeleton. Hepatocytes were cultured for 52 hours on collagen film in the presence or absence (‘no inh’) of cytochalasin D (CD, 1 µg/ml) or C3 exoenzyme (C3, 2 µg/ml). Cell lysates were examined for cyclin D1 protein by western blot analysis.
Hepatocyte growth regulation by collagen 2977
A
C
B
D
to express the exogenous cyclin D1 gene, western blot analysis was performed using an antibody specific for human cyclin D1 (clone DCS-6, Oncogene Science). Significant levels of human cyclin D1 protein were observed only in the ADV-D1transfected cells after 52 hours in culture (49 hours after virus addition; Fig. 8A). No human cyclin D1 was observed in the non-transfected or β-gal-transfected cells. To test the ability of exogenous cyclin D1 expression to alter cell cycle progression, the expression of PCNA (proliferating cell nuclear antigen), an S phase-specific protein, was examined. PCNA was abundantly expressed in proliferating hepatocytes cultured on collagen film, while cells on dried collagen, either non-transfected or transfected with β-gal, demonstrated little to no PCNA expression, in accordance with their arrest in G1 phase. In contrast, cyclin D1-transfected hepatocytes on dried collagen expressed abundant PCNA, demonstrating that cyclin D1 overexpression can drive G1-arrested hepatocytes on dried collagen into S phase (Fig. 8A). The transfected cyclin D1 also promoted kinase activity, as determined by Rb-GST kinase assays using anti-human cyclin D1 immunoprecipitates (Fig. 8B). In addition, cdk2 kinase activity was restored in the presence of exogenous cyclin D1 (Fig. 8B). DNA synthesis was also enhanced on dried collagen and collagen gel in ADV-D1-transfected hepatocytes compared to β-gal-transfected cells (Fig. 8C). Finally, the association of p27 was examined following transfection of cells on dried collagen with ADV-βgal or ADV-D1. Significantly greater levels of p27 were associated with cyclin D1 in transfected cells, suggesting a shift in the distribution of p27 from cdk2 complexes to cyclin D1 complexes (Fig. 8D). These data
Fig. 8. ADV-D1 transfected hepatocytes overexpress human cyclin D1 and progress into S phase. (A) Hepatocytes were transfected with adenovirus containing the gene coding for either β-galactosidase (‘ADV-βgal’) or human cyclin D1 (‘ADV-D1’), and cultured on collagen film or dried collagen for 52 hours. Cell lysates were probed by western blot analysis for expression of human cyclin D1 (hu D1) and PCNA, a marker of S phase. (B) Hepatocytes were cultured and transfected as in 8A, lysed after 52 hours in culture, and anti-cyclin D1 immunoprecipitates were analyzed for cyclin D/cdk4 kinase activity, and anti-cdk2 immunoprecipitates were analyzed for cyclin E/cdk2 kinase activity using RbGST or histone H1 (HH1), respectively, as substrates. (C) Hepatocytes were plated onto collagen film, dried collagen, or collagen gel, and transfected with either ADV-βgal or ADV-D1. [3H]Thymidine uptake was measured from 52-68 hours after plating. (D) Hepatocytes were cultured and transfected on dried collagen as in 8A, lysed after 52 hours in culture, and anti-cyclin D1 immunoprecipitates were examined for association with p27 by running the samples on SDS-PAGE and probing the western blot with anti-p27 antibody.
support the hypothesis that cyclin D1 is indeed the rate-limiting step in hepatocyte growth arrest on dried collagen and collagen gel, and downstream cell cycle events that are normally absent on these substrates can be restored by the over-expression of cyclin D1. Because of the consistent correlation observed between cell spreading and DNA synthesis in many cell types, the morphology of ADV-D1-transfected cells on dried collagen was examined. While these cells were clearly entering S phase (Fig. 8), their morphology appeared similar to that seen in either non- or control-transfected cells on dried collagen which remain arrested in G1 (Fig. 9). Adenovirus transfection using either βgal or cyclin D1 vector at this titer had no effect on hepatocyte morphology on collagen film (data not shown). Thus, cyclin D1 has no effect on cell spreading, and its overexpression abrogates the spreading requirement for DNA synthesis normally observed in vitro. DISCUSSION The data presented here are the first to demonstrate regulation of cyclin D1 mRNA and protein levels by the type of substrate to which cells adhere. Primary rat hepatocytes adherent to a substrate of dried, denatured type I collagen or collagen gel are unable to express cyclin D1, even in the presence of growth factors. Thus, adhesion per se is insufficient for cyclin D1 expression and cell cycle progression, and the physical structure of an ECM substrate can dramatically alter the cellular growth response at the level of cyclin expression. In
2978 L. K. Hansen and J. H. Albrecht
Fig. 9. Hepatocyte spreading does not occur following cyclin D1 overexpression. Hepatocytes were cultured on collagen film without transfection or on dried collagen followed by transfection with either ADV-βgal or ADV-D1. After 48 hours, cells were fixed in 1% glutaraldehyde and methanol, and photographs were taken at ×60 magnification using Varel optics on an Axiovert 25 microscope (Zeiss).
addition, the ability of transfected cyclin D1 to overcome G1 arrest in hepatocytes on dried collagen or collagen gel occurs in the absence of cell shape changes, and suggests that expression of cyclin D1 is the rate limiting step in this substrate-dependent growth arrest. Cyclin D1 appears to play a critical role in regulating hepatocyte proliferation in vivo. Previous studies show that cyclin D1 is upregulated during G1 phase in regenerating rodent liver after partial hepatectomy, in proliferating hepatocytes in culture, and in regenerating human liver (Lu et al., 1992; Albrecht et al., 1993, 1995; Loyer et al., 1996). Furthermore, deregulation of cyclin D1 can lead to abnormal growth and tumorigenesis. The cyclin D1 gene is amplified in many human tumors, including hepatoma, and cyclin D1 protein is over-expressed in these tumors (Zhang et al., 1993). Cyclin D1 is thus implicated as a crucial regulator of normal hepatocyte growth control. Several studies have previously demonstrated the lack of cyclin expression (D1 or A, depending on the cell type) in suspended cells (Guadagno et al., 1993; Shulze et al., 1996; Zhu et al., 1996; Resnitzky, 1997), indicating a requirement for adhesion in cyclin expression. The data presented here build on those previous studies by comparing cell cycle progression on different types of substrates. Changes in hepatocyte morphology and function on different type I collagen substrata have been well-characterized (Ben-Ze’ev et al., 1988; Ezzell et al., 1993; Santhosh and Sudhakaran, 1994). Such differences have also been described for other cell types (Lee et al., 1984; Koyama et al., 1996). The data presented in this paper are the first to demonstrate regulation of cyclin D1 simply by the physical structure of the ECM, and its expression correlates with cell morphology. Thus, it is clear that the manner in which type I collagen is presented to the cell can have a profound impact on the cellular response.
The mechanism of cyclin D1 expression and G1-S phase regulation by ECM substrates appears to be distinct in hepatocytes from that of other cell types studied thus far. The lack of cyclin D1 expression on dried collagen differs from results obtained by Koyama et al. (1996) in studies of smooth muscle cells on collagen substrates. In these studies, smooth muscle cells cultured on polymerized collagen (collagen gel) demonstrated normal expression of cyclin D1 and cyclin E, but diminished cyclin E kinase activity compared to that on monomeric collagen. In addition, they demonstrate upregulation of p27 protein in smooth muscle cells on polymerized collagen which they hypothesize is the mechanism by which cyclin E/cdk2 activity is inhibited. p27 also appears to play a role in inhibiting cell cycle events in hepatocytes on dried collagen, but clearly by a different mechanism. Since the level of p27 does not differ in hepatocytes on the two substrates by 52 hours, its enhanced association with cdk2 complexes on dried collagen may be facilitated by the lack of cyclin D1/cdk4 complexes, since these complexes have been postulated to serve as a pool for p27 binding. If cyclin D1 is present at high levels, cyclin D1/cdk4 complexes may preferentially bind up p27, thus diminishing p27 association with cdk2 complexes. p27 has been postulated to be a less efficient inhibitor of cyclin D/cdk4 complexes than cyclin E/cdk2 complexes (Polyak et al., 1994b). Thus, the level of cyclin D1 expression may regulate the inhibitory activity of p27. This hypothesis is supported by the enhanced cdk2-associated kinase activity seen in cyclin D1-transfected hepatocytes on dried collagen and enhanced association between p27 and cyclin D1 in ADV-D1 transfected cells (Fig. 8B and D). In addition, ADVD1 transfection of hepatocytes arrested by growth factor depletion also induces DNA synthesis (Albrecht and Hansen, 1999), demonstrating that cyclin D1 overexpression can
Hepatocyte growth regulation by collagen 2979 overcome both growth factor-dependent, as well as substratedependent, signals. The ability of cyclin D1-transfected hepatocytes on dried collagen to enter S phase without spreading is of interest considering the correlation between cell spreading and DNA synthesis. The correlation between spreading and DNA synthesis suggests that changes in the cytoskeleton are required for entry into S phase. Indeed, cytochalasin D, an inhibitor of actin polymerization, inhibits DNA synthesis in hepatocytes (data not shown) and cyclin D1 mRNA expression in fibroblasts (Bohmer et al., 1996) and hepatocytes (Fig. 6). Certain soluble mediators of the actin cytoskeleton also appear to be involved in cell growth. C3 exoenzyme, a specific inhibitor of RhoA, also inhibits cyclin A, but not cyclin D1, expression in fibroblasts (Udagawa and McIntyre, 1996), and over-expression of active rac and rho mutants in 3T3 fibroblasts activates the human cyclin D1 promoter (Westwick et al., 1997). In the present study, cytochalasin D and C3 exoenzyme each inhibit the expression of cyclin D1 protein, suggesting that actin polymerization and/or activation of its upstream mediators are in some way necessary for cyclin D1 expression. The ability of cyclin D1 overexpression to overcome growth arrest in the absence of spreading suggests that spreading per se in the presence of sufficient cyclin D1 may not be required for DNA synthesis. Rather, effectors of actin polymerization may affect an event upstream of cyclin D1 which is necessary for cyclin D1 expression. It appears that activated Rho is required for cyclin D1 expression in primary rat hepatocytes, and it may remain inactive in hepatocytes on dried collagen. While this apparent shape-related growth requirement remains to be elucidated, it is clear that this requirement is negated by overexpression of cyclin D1 in the absence of cell shape changes. The mechanism by which alteration of the physical structure of the type I collagen substrate leads to different cellular responses is not clear. One possibility is that different integrin receptors recognize the different forms of collagen, each transmitting different intracellular signals. Some cell types possess two different integrin receptors for type I collagen: α1β1 and α2β1 integrins. Smooth muscle cells utilize the α1β1 integrin to bind to native type I collagen, while they bind to heat-denatured type I collagen with both α1β1 and α2β1 integrins (Yamamoto and Yamamoto, 1994). In contrast, hepatocytes possess only one integrin receptor for type I collagen, the α1β1 integrin (Gullberg et al., 1990, 1992). α2β1 is not normally expressed (Gullberg et al., 1990; Volpes et al., 1991; Couvelard et al., 1998), although it is up-regulated in vivo under certain diseased states (Garcia-Monzon et al., 1992). Preliminary data suggest that the α1β1 integrin binds to each of the three forms of collagen substrate in the current study (L. Hansen, unpublished data). The lack of α2β1 was confirmed in our system by the inability of anti-α2 blocking antibody to inhibit hepatocyte adhesion to either form of collagen (data not shown). Thus, the difference in morphologic and proliferative response seen in hepatocytes on different collagen substrates does not appear to be mediated by distinct collagen-binding integrin receptors. Previous studies demonstrate inhibition of hepatocyte adhesion to denatured, but not native, type I collagen with the hexapeptide GRGDTP (Gullberg et al., 1992). The authors suggest that the denatured form of collagen exposes an RGD
binding site to which hepatocytes adhere via α5β1 integrin. Alternatively, other ECM proteins may also contribute to the morphological differences seen on the different collagen substrates. While cell culture experiments described in this paper were performed under serum-free conditions and thus in the absence of serum proteins, the hepatocytes do secrete and deposit ECM proteins such as fibronectin. Thus, the secreted fibronectin could serve as a bridge, with cells binding solely to the fibronectin which in turn binds to the collagen, as suggested for chondrocyte attachment to denatured type II collagen (Tuckwell et al., 1994) and endothelial cell attachment to denatured, but not native, type I collagen (Norris et al., 1990). However, hepatocytes are known to proliferate on a thin film of fibronectin (Mooney et al., 1992a; Hansen et al., 1994). Thus, if the signal to proliferate was dependent solely on the type of integrin receptor bound, proliferation would be expected to occur if hepatocytes bind to denatured collagen via an α5β1-Fn bridge. Alternatively, the difference in cellular response may be related to a difference in mechanical properties of the substrate. Upon adhesion, cells generate forces that pull on the substrate via cytoskeletal linkages to focal contacts (Ingber, 1991; Ingber et al., 1993). It has been shown recently that mechanical loads enhance integrin-dependent tyrosine phosphorylation of cytoskeletal-associated proteins (Schmidt et al., 1998), suggesting that substrates which impart different mechanical loads on the cell may induce different signaling events. Rigid substrates, such as the collagen film, resist these cellular forces, and thus may impart a greater mechanical stress on the integrin complex, whereas malleable substrates do not resist such forces. Thus, greater mechanical stress may be exerted on integrins in hepatocytes adherent to collagen film than to collagen gel or dried collagen, resulting in the induction of different intracellular signals. The exact nature of the mechanosensitive signals remains to be elucidated. In spite of these remaining questions, it is clear that these substratedependent signals are required, along with growth factordependent signals, for the proper induction of cyclin D1 expression and cell cycle progression, and this requirement can be overcome by the over-expression of cyclin D1. The authors thank Brenda Reiland, Lisa Jungers, and Kristine Groehler for excellent technical assistance; and Drs Jim McCarthy and Ted Oegema for helpful discussions. This research was funded by grants from the American Cancer Society (IRG-13-37), National Science Foundation (MCB9509600), and Minnesota Medical Foundation to L.K.H., and from NIH (1R01-DK-54921) and American Liver Foundation to J.H.A.
REFERENCES Albrecht, J. H., Hoffman, J. S., Kren, B. T. and Steer, C. J. (1993). Cyclin and cyclin-dependent kinase 1 mRNA expression in models of regenerating liver and human liver diseases. Am. J. Phys. 265, G857-G864. Albrecht, J. H., Hu, M. Y. and Cerra, F. B. (1995). Distinct patterns of cyclin D1 regulation in models of liver regeneration and human liver. Biochem. Biophys. Res. Commun. 209, 648-655. Albrecht, J. H., Poon, R. Y. C., Ahonen, C. L., Rieland, B. M., Deng, C. and Crary, G. S. (1998). Involvement of p21 and p27 in the regulation of CDK activity and cell cycle progression in the regenerating liver. Oncogene 16, 2141-2150. Albrecht, J. H. and Hansen, L. K. (1999). Cyclin D1 promotes mitogenindependent cell cycle progression in hepatocytes. Cell Growth Differ. 10, 397-404.
2980 L. K. Hansen and J. H. Albrecht Becker, T. C., Noel, R. J., Coats, W. S., Gomez-Foix, A. M., Alam, T., Gerard, R. D. and Newgard, C. B. (1994). Use of recombinant adenovirus for metabolic engineering of mammalian cells. Meth. Cell Biol. 43, 161189. Ben-Ze’ev, A., Robinson, G. S. and Farmer, S. R. (1988). Cell-cell and cellmatrix interactions differentially regulate the expression of hepatic and cytoskeletal genes in primary cultures of rat hepatocytes. Proc. Nat. Acad. Sci. USA 85, 2161-2165. Benecke, B.-J., Ben-Ze’ev, A. and Penman, S. (1978). The control of mRNA production, translation and turnover in suspended and reattached anchoragedependent fibroblasts. Cell 14, 931-939. Berry, M. N., Edwards, A. M. and Barritt, G. J. (1991). Isolated hepatocytes: Preparation, properties and applications. In Laboratory Techniques in Biochemistry and Molecular Biology (ed. R. H. Burdon and P. H. v. Knippenberg), pp. 265-354. Elsevier, Amsterdam. Bohmer, R.-M., Scharf, E. and Assoian, R. K. (1996). Cytoskeletal integrity is required throughout the mitogen stimulation phase of the cell cycle and mediates the anchorage-dependent expression of cyclin D1. Mol. Biol. Cell 7, 101-111. Connell-Crowley, L. and Harper, W. (1995). Bacterial expression and in vitro assay of cell cycle kinases. In Cell Cycle – Materials and Methods (ed. M. Pagano), pp. 157-168. Springer, New York. Couvelard, A., Bringuier, A.-F., Dauge, M.-C., Nejjari, M., Darai, E., Benifla, J.-L., Feldmann, G., Henin, D. and Scozaec, J. Y. (1998). Expression of integrins during liver organogenesis in humans. Hepatology 26, 839-847. Ezzell, R. M., Toner, M., Hendricks, K., Dunn, J. C. Y., Tompkins, R. G. and Yarmush, M. L. (1993). Effect of collagen gel configuration on the cytoskeleton in cultured rat hepatocytes. Exp. Cell Res. 208, 442-452. Folkman, J. and Moscona, A. (1978). Role of cell shape in growth control. Nature 346, 760-763. Garcia-Monzon, C., Moreno-Otero, R., Garcia-Buey, L., Garcia-Sanchez, A., Campanero, M. R. and Sanchez-Madrid, F. (1992). Intrahepatic upregulated expression of extracellular matrix protein receptors in chronic active hepatitis type B. Gastroenterology 102, 255-262. Gomez-Foix, A. M., Coats, W. S., Baque, S., Alam, T., Gerard, R. D. and Newgard, C. B. (1992). Adenovirus-mediated transfer of the muscle glycogen phosphorylase gene into hepatocytes confers altered regulation of glycogen metabolism. J. Biol. Chem. 267, 25129-25134. Grana, X. and Reddy, E. P. (1995). Cell cycle control in mammalian cells: Role of cyclins, cyclin dependent kinases (CDKs), growth suppressor genes, and cyclin-dependent kinase inhibitors (CKIs). Oncogene 11, 211-219. Guadagno, T. M. and Assoian, R. K. (1991). G1/S control of anchorageindependent growth in the fibroblast cell cycle. J. Cell Biol. 115, 1419-1425. Guadagno, T. M., Ohtsubo, M., Roberts, J. M. and Assoian, R. K. (1993). A link between cyclin A expression and adhesion-dependent cell cycle progression. Science 262, 1572-1575. Gullberg, D., Turner, D. C., Terracio, L., Borg, T. K. and Rubin, K. (1990). Different β1-integrin collagen receptors on rat hepatocytes and cardiac fibroblasts. Exp. Cell Res. 190, 254-264. Gullberg, D., Gehlsen, K. R., Turner, D. C., Ahlen, K., Zijenah, L. S., Barnes, M. J. and Rubin, K. (1992). Analysis of α1β1, α2β1, and α3β1 integrins in cell-collagen interactions: identification of conformation dependent α1β1 binding sites in collagen type I. EMBO J. 11, 3865-3873. Han, E. K., Guadagno, T. M., Dalton, S. L. and Assoian, R. K. (1993). A cell cycle and mutational analysis of anchorage-independent growth: cell adhesion and TGF-beta 1 control G1/S transit specifically. J. Cell Biol. 122, 461-471. Hansen, L. K., Mooney, D. J., Vacanti, J. P. and Ingber, D. E. (1994). Integrin binding and cell spreading on extracellular matrix act at different points in the cell cycle to promote hepatocyte growth. Mol. Biol. Cell 5, 967975. Ingber, D. E. (1990). Fibronectin controls capillary endothelial cell growth by modulating cell shape. Proc. Nat. Acad. Sci. USA 87, 3579-3583. Ingber, D. E. (1991). Integrins as mechanochemical transducers. Curr. Opin. Cell Biol. 3, 841-848. Ingber, D. E., Karp, S., Plopper, G., Hansen, L. K. and Mooney, D. J. (1993). Mechanochemical transduction across extracellular matrix and through the cytoskeleton. In Physical Forces and the Mammalian Cell (ed. J. F. a. C. L. Ives), pp. 61-79. Academic Press, San Diego. Kaytor, E. N., Shih, H. and Towle, H. C. (1997). Carbohydrate regulation of hepatic gene expression. Evidence against a role for the upstream stimulatory factor. J. Biol. Chem. 272, 7525-7531. Koyama, H., Raines, E. W., Bornfeldt, K. E., Roberts, J. M. and Ross, R.
(1996). Fibrillar collagen inhibits arterial smooth muscle proliferation through regulation of cdk2 inhibitors. Cell 87, 1069-1078. Lee, E. Y.-H., Parry, G. and Bissell, M. J. (1984). Modulation of secreted proteins of mouse mammary epithelial cells by the collagenous substrata. J. Cell Biol. 107, 1589-1598. Lew, D. J., Dulic, V. and Reed, S. I. (1991). Isolation of three novel human cyclins by rescue of G1 cyclin (Cln) function in yeast. Cell 66, 1197-1206. Linsenmayer, T. F. (1981). Collagen In Cell Biology of Extracellular Matrix (ed. E. D. Hay), pp. 5-37. Plenum Press, New York. Loyer, P., Glaise, D., Cariou, S., Baffet, G., Meijer, L. and GuguenGuillouzo, C. (1994). Expression and activation of cdks (1 and 2) and cyclins in the cell cycle progression during liver regeneration. J. Biol. Chem. 269, 2491-2500. Loyer, P., Cariou, S., Glaise, D., Bilodeau, M., Baffet, G. and GuguenGuillouzo, C. (1996). Growth factor dependence of progression through G1 and S phases of adult rat hepatocytes in vitro. J. Biol. Chem. 271, 1148411492. Lu, X. P., Koch, K. S., Lew, D. J., Dulic, V., Pines, J., Reed, S. I., Hunter, T. and Leffert, H. L. (1992). Induction of cyclin mRNA and cyclinassociated histone H1 kinase during liver regeneration. J. Biol. Chem. 267, 2841-2844. MacPherson, I. and Montagnier, L. (1964). Agar suspension culture for selective assay of cells transformed by polyoma virus. Virology 23, 291-294. Matsushime, H., Quelle, D. E., Shurtleff, S. A., Shibuya, M., Sherr, C. J. and Kato, J. Y. (1994). D-type cyclin-dependent kinase activity in mammalian cells. Mol. Cell. Biol. 14, 2066-2076. Mooney, D. J., Hansen, L. K., Vacanti, J. P., Langer, R., Farmer, S. R. and Ingber, D. E. (1992a). Switching from differentiation to growth in hepatocytes: Control by extracellular matrix. J. Cell. Physiol. 151, 497-505. Mooney, D. J., Hansen, L. K., Vacanti, J. P., Langer, R. and Ingber, D. E. (1992b). Induction of hepatocyte differentiation by the extracellular matrix and an RGD-containing synthetic peptide. Mat. Res. Soc. Symp. Proc. 252, 199-204. Morton, L. F., Peachey, A. R., Zijenah, L. S., Goodall, A. H., Humphries, M. J. and Barnes, M. J. (1994). Conformation-dependent platelet adhesion to collagen involving integrin alpha 2 beta 1-mediated and other mechanisms: multiple alpha 2 beta 1-recognition sites in collagen type I. Biochem. J. 299, 791-797. Norris, W. D., Steele, J. G., Johnson, G. and Underwood, P. A. (1990). Serum enhancement of human endothelial cell attachment to and spreading on collagens I and IV does not require serum fibronectin or vitronectin. J. Cell Sci. 95, 255-262. Otsuka, H. and Moskowitz, M. (1975). Arrest of 3T3 cells in G1 phase in suspension culture. J. Cell. Physiol. 87, 213-220. Peshwa, M. V., Wu, F. J., Follstad, B. D., Cerra, F. B. and Hu, W.-S. (1994). Kinetics of hepatocyte spheroid formation. Biotechnol. Prog. 10, 460-466. Polyak, K., Kato, J. Y., Soloman, M. J., Sherr, C. J., Massague, J., Roberts, J. M. and Koff, A. (1994a). p27KIP1, a cyclin-Cdk inhibitor, links transforming growth factor-β and contact inhibition to cell cycle arrest. Genes Dev. 8, 9-22. Polyak, K., Lee, M. H., Erdjument-Bromage, H., Koff, A., Roberts, J. M., Tempst, P. and Massague, J. (1994b). Cloning of p27KIP1, a cyclindependent kinase inhibitor and a potential mediator of extracellular antimitogenic signals. Cell 78, 59-66. Poon, R. Y., Toyoshima, H. and Hunter, T. (1995). Redistribution of the CDK inhibitor p27 between different cyclin/CDK complexes in the mouse fibroblast cell cycle and in cells arrested with lovastatin or ultraviolet irradiation. Mol. Biol. Cell 6, 1197-1213. Resnitzky, D. (1997). Ectopic expression of cyclin D1 but not cyclin E induces anchorage-independent cell cycle progression. Mol. Cell. Biol. 17, 56405647. Ricard-Blum, S., Bresson-Hadni, S. S., Guerret, S., Grenard, P., Volle, P. J., Risteli, L., Grimaud, J. A. and Vuitton, D. A. (1996). Mechanism of collagen network stabilization in human irreversible granulomatous liver fibrosis. Gastroenterology 111, 172-82. Ridley, A. and Hall, A. (1992). The small GTP-binding protein rho regulates the assembly of focal adhesions and actin stress fibers in response to growth factors. Cell 70, 389-399. Ridley, A., Paterson, H. F., Johnston, C. L., Diekmann, D. and Hall, A. (1992). The small GTP-binding protein rac regulates growth factor-induced membrane ruffling. Cell 70, 401-410. Santhosh, A. and Sudhakaran, P. R. (1994). Influence of collagen gel substrata on certain biochemical activities of hepatocytes in primary culture. Mol. Cell. Biochem. 137, 127-133.
Hepatocyte growth regulation by collagen 2981 Schmidt, C., Pommerenke, H., Durr, F., Nebe, B. and Rychly, J. (1998). Mechanical stressing of integrin receptors induces enhanced tyrosine phosphorylation of cytoskeletally anchored proteins. J. Biol. Chem. 273, 5081-5085. Scott, J. E., Bosworth, T. R., Cribb, A. M. and Gressner, A. M. (1994). The chemical morphology of extracellular matrix in experimental rat liver fibrosis resembles that of normal developing connective tissue. Virchows Archiv. 424, 89-98. Seglen, P. O. (1976). Preparation of isolated rat liver cells. Meth. Cell Biol. 13, 29-83. Sherr, C. J. (1994). G1 phase progression: Cycling on cue. Cell 79, 551-555. Shin, S., Freedman, V. H., Risser, R. and Pollack, R. (1975). Tumorigenicity of virus-transformed cells in nude mice is correlated specifically with anchorage-independent growth in vitro. Proc. Nat. Acad. Sci. USA 72, 44354439. Shulze, A., Zerfass-Thome, K., Berges, J., Middendorp, S., Jansen-Durr, P. and Henglein, B. (1996). Anchorage-dependent transcription of the cyclin A gene. Mol. Cell. Biol. 16, 4632-4638. Singhvi, R., Kumar, A., Lopez, G. P., Stephanopoulos, G. N., Wang, D. I. C., Whitesides, G. M. and Ingber, D. E. (1994). Engineering cell shape and function. Science 264, 696-698. Stoker, M., O’Neill, C., Berryman, S. and Waxman, V. (1968). Anchorage and growth regulation in normal and virus transformed cells. Int. J. Cancer 3, 683-693. Tuckwell, D. S., Ayad, S., Grant, M. E., Takigawa, M. and Humphries, M. J. (1994). Conformation dependence of integrin-type II collagen binding. Inability of collagen peptides to support α2β1 binding, and mediation of adhesion to denatured collagen by a novel α5β1-fibronectin bridge. J. Cell Sci. 107, 993-1005.
Udagawa, T. and McIntyre, B. W. (1996). ADP-ribosylation of the G protein rho inhibits integrin regulation of tumor cell growth. J. Biol. Chem. 271, 12542-12548. Volpes, R., Oord, J. J. c. d. and Desmet, V. J. (1991). Distribution of the VLA family of integrins in normal and pathological human liver tissue. Gastroenterology 01, 200-206. Watt, F. M., Jordan, P. W. and O’Neill, C. H. (1988). Cell shape controls terminal differentiation of human epidermal keratinocytes. Proc. Nat. Acad. Sci. USA 85, 5576-5580. Westwick, J. K., Lambert, Q. T., Clark, G. J., Symons, M., Aelst, L. V., Pestell, R. G. and Der, C. J. (1997). Rac regulation of transformation, gene expression, and actin organization by multiple, PAK-independent pathways. Mol. Cel. Biol. 17, 1324-1335. Yamamoto, K. and Yamamoto, M. (1994). Cell adhesion receptors for native and denatured type I collagens and fibronectin in rabbit arterial smooth muscle cells in culture. Exp. Cell Res. 214, 258-263. Yamamoto, M., Marui, N., Sakai, T., Morii, N., Kozaki, S., Ikai, K., Imamura, S. and Narumiya, S. (1993). ADP-ribosylation of the rhoA gene product by botulinum C3 exoenzyme causes Swiss 3T3 cells to accumulate in the G1 phase of the cell cycle. Oncogene 8, 1449-1455. Zhang, Y. J., Jiang, W., Chen, C. J., Lee, C. S., Kahn, S. M., Santella, R. M. and Weinstein, I. B. (1993). Amplification and overexpression of cyclin D1 in human hepatocellular carcinoma. Biochem. Biophys. Res. Commun. 196, 1010-1016. Zhu, X., Ohtsubo, M., Bohmer, R. M., Roberts, J. M. and Assoian, R. K. (1996). Adhesion-dependent cell cycle progression linked to the expression of cyclin D1, activation of cyclin E-cdk2, and phosphorylation of the retinoblastoma protein. J. Cell Biol. 133, 391-403.
