Potential conflict of interest: Nothing to report.
The cellular basis of liver regeneration has been intensely investigated for many years. However, the mechanisms initiating hepatocyte “plasticity” and priming for proliferation are not yet fully clear. We investigated alterations in gene expression patterns during the first 72 hours of C57BL/6N mouse hepatocyte culture on collagen monolayers (CM), which display a high basal frequency of proliferation in the absence of cytokines. Although many metabolic genes were down-regulated, genes related to mitogen-activated protein kinase (MAPK) signaling and cell cycle were up-regulated. The latter genes showed an overrepresentation of transcription factor binding sites (TFBS) for ETF (TEA domain family member 2), E2F1 (E2F transcription factor 1), and SP-1 (Sp1 transcription factor) (P < 0.001), all depending on MAPK signaling. Time-dependent increase of ERK1/2 phosphorylation occurred during the first 48 hours (and beyond) in the absence of cytokines, accompanied by an enhanced bromodeoxyuridine labeling index of 20%. The MEK inhibitor PD98059 blunted these effects indicating MAPK signaling as major trigger for this cytokine-independent proliferative response. In line with these in vitro findings, liver tissue of mice challenged with CCl4 displayed hepatocytes with intense p-ERK1/2 staining and nuclear SP-1 and E2F1 expression. Furthermore, differentially expressed genes in mice after partial hepatectomy contained overrepresented TFBS for ETF, E2F1, and SP-1 and displayed increased expression of E2F1. Conclusion: Cultivation of murine hepatocytes on CM primes cells for proliferation through cytokine-independent activation of MAPK signaling. The transcription factors ETF, E2F1, and SP-1 seem to play a pronounced role in mediating proliferation-dependent differential gene expression. Similar events, but on a shorter time-scale, occur very early after liver damage in vivo. (HEPATOLOGY 2010;.)
In normal liver, hepatocytes remain in the quiescent G0 phase. Loss of liver mass by partial hepatectomy (PHx) or hepatotoxic chemicals (e.g., CCl4) induces a complex regeneration process restoring the original liver mass within 5-7 days. Within minutes after partial hepatectomy, signals occur that prime hepatocytes for proliferation. A broad range of cytokines are released and transcription factors are activated (summarized in recent reviews1-4). In addition to cytokine signaling, the extracellular matrix (ECM) plays an important role. Loss of cell-matrix contacts activated c-fos5 and sensitized rat hepatocytes for cytokine induced proliferation.6In vivo, degradation of ECM in an experimental mouse model resulted in hepatocyte proliferation.7 Also, loss of integrin-linked kinase signaling increased hepatocyte proliferation8 and enhanced liver regeneration.9
Hepatocyte cultures are well established and often used in studies on drug metabolism, enzyme induction and toxicity.10, 11 The latter represent applications where a robust hepatocyte physiology, such as a high metabolizing capacity, is required. Recently, we have shown that collagen sandwich cultures preserve polarized mouse hepatocyte morphology and are superior to collagen monolayers (CM), which rapidly and spontaneously lose polarity and transdifferentiate.12, 13 However, CM cultures are resistant to apoptotic stimuli and display an activation of survival pathways. Therefore, we speculated that this type of cultivation may represent an in vitro model for hepatocyte priming and entry into G1/S phase in the absence of cytokine stimulation. This assumption is supported by a higher basal extracellular signal-regulated kinase 1/2 (ERK1/2) and Akt/protein kinase B (PKB) phosphorylation compared to collagen sandwich (CS) cultured hepatocytes.12
In the present study, we explored gene expression profiles during the first hours of mouse hepatocyte CM culture and identified signatures describing loss of metabolic functions, increased cell-matrix interactions and transformation into a proliferative state. TF binding sites for up-regulated genes were characterized and identified the transcription factors (TFs) ETF (TEA domain family member 2), E2F1 (E2F transcription factor 1), and SP-1 (Sp1 transcription factor) as most important regulators, particularly for genes of MAPK signaling pathways. Similar regulatory events, though more rapidly, occurred in vivo after CCl4 intoxication and after PHx reflecting findings observed in culture on a different time scale. This supports the idea that cytokine-independent MAPK activation contributes to the proliferative response of mouse hepatocytes during regenerative processes.
Male C57BL/6N mice (8-12 weeks old) were allowed food and water adlibitum. All experiments were approved by the ethics committees. Mice were intoxicated with 1.6 g CCl4/kg body weight, administered intraperitoneally, dissolved in 0.5 mL olive oil. At selected time points, mice were killed by heart puncture and liver sections were snap frozen in liquid nitrogen.
Hepatocyte Isolation and Culture.
Hepatocytes were isolated according to a modified method of Seglen11, 14 and cultured in Williams E medium (Lonza Verviers, Verviers, Belgium), supplemented with 10−7 M dexamethasone and penicillin/streptomycin (all from Sigma-Aldrich, Taufkirchen, Germany), 2 mM glutamine (Roth, Karlsruhe, Germany) and 10% fetal bovine serum (FBS) in six-well plates (TPP, Trasadingen, Switzerland). After attachment, medium was changed to medium without FBS. Fifty μM PD98059 (Sigma-Aldrich) was added 1 hour before stimulation with 20 ng/mL murine HGF (R&D Systems GmbH, Wiesbaden-Nordenstadt, Germany) and 100 nM insulin (Sigma-Aldrich). Then, cells were cultured for 3 days with inhibitor, cytokines, and BrdU. A wound-like spacing between confluent cultured hepatocytes was generated with an Ibidi culture insert (Applied Biophysics, New York, NY). The device was removed from confluent cultured hepatocytes after 1 day, followed by 3 days culture in the presence of BrdU with and without HGF/insulin.
Hepatocyte Proliferation Assays and BrdU Incorporation.
DNA synthesis of hepatocytes was measured by the incorporation of BrdU into nuclei. Briefly, following preculture for 24 hours in the presence of 150 μg/mL BrdU, hepatocytes were incubated with 20 ng/mL HGF and 100 nM insulin for 3 days. Then, cells were washed with PBS, fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton-X 100, followed by 10 minutes incubation with 2 N HCl for DNA denaturation. The slides were blocked with 3% BSA/0.1% Tween-20 in PBS for 1 hour at RT and incubated with rat anti-BrdU antibody (Serotec, Oxford, UK; 1:25) for 1 hour at RT. Detection was achieved using a biotinylated anti-rat (IgG) antibody (Vector Lab, Burlingame, CA; 1:250), followed by peroxidase conjugated streptavidin (Jackson ImmunoResearch, Baltimore, MD; 1:500). Staining was visualized using DAB Substrate Kit (Vector Lab). Mayer's hemalaum solution (Merck KGaA, Darmstadt, Germany; dilution 1:5) was used for counterstaining. Proliferation was expressed as percentage of BrdU positive nuclei, of at least 400 nuclei in four regions per slide.
Liver sections (4 μm) were deparaffinised. Sections were incubated in peroxidase blocking reagent (Dako Deutschland GmbH, Hamburg, Germany) for 1 hour and subsequently with phospho-Erk (mouse IgG2α, sc-7383) or SP-1 antibodies (rabbit polyclonal, sc-14027), both from Santa Cruz Biotechnology (Heidelberg, Germany) overnight at 4°C. EnVision peroxidase (Dako Deutschland GmbH) was applied for 1 hour at RT. E2F1 was stained with polyclonal rabbit anti-E2F1 antibody (Atlas Antibodies, Stockholm, Sweden) overnight followed by incubations with a biotinylated anti-rabbit antibody (AP132B, Chemicon) and POD-extravidin (E2886, Sigma, Taufkirchen, Germany) for 60 and 30 minutes, respectively. Staining was performed in 0.1 M Tris buffer (pH 7.6), composed of 0.05% diaminobenzidine, 2% aminotriazole, and 0.033% H2O2. Counterstaining was performed as described above.
Further Applied Methods.
RNA isolation, quantitative real-time PCR, preprocessing of microarray data and the subsequent Multivariate Data analysis, promoter analysis, chromatin immunoprecipitation, and preparation of immunoblots are described in the Supporting Information.
Isolation of Hepatocytes Alters Gene Expression Within the First 24 Hours.
To understand how isolation and subsequent cultivation of primary mouse hepatocytes on collagen monolayers (CM) alters the physiology of hepatocytes, we studied alterations in gene expression patterns using microarrays. Two different experimental designs were performed in different laboratories to obtain robust data. The “Leipzig” experiment (Supporting Fig. 1) included fresh liver tissue, freshly isolated hepatocytes, hepatocytes after attachment to CM and after cultivation periods between 1 and 2 days. Samples were analyzed using an Affymetrix platform. The second study (“Freiburg” experiment) included cultivation periods of up to 6 days and a cDNA platform for analysis.
Alterations in the gene expression pattern between liver and freshly isolated hepatocytes were small (Fig. 1A). The most prominent changes (>1.5-fold) occurred between attachment of hepatocytes on CM and cultivation from day zero to day one. Between days one and two, alterations in gene expression were less pronounced.
The observed changes in gene expression were visualized by principal component analysis (PCA). Hepatocytes in both data sets change their expression pattern in a similar “direction” (Fig. 1B; dashed arrows) indicating that cultivation on CM induces equal adaptation of hepatocytes, regardless of interlaboratory differences in array platform usage and analysis.
Detailed analysis of the four-fold overexpressed genes of both platforms using ANOVA (Supporting Table 1) revealed that the overall pattern of altered gene expression was reproducible between both laboratories and that alterations in gene expression patterns can be attributed mainly to culture time on CM. Thus, cultivation of primary mouse hepatocytes on CM causes major changes of hepatocyte physiology within the first 24 hours.
Table 1. Transcription factor binding sites (TFBS) overrepresented in differentially expressed genes during the first 24 hours of cultivation on collagen monolayers and after partial hepatectomy (raw data from Otu and coworkers17), identified by the PRIMA algorithm.
TF binding site (ID)
P Value (Bonferroni Adjusted)
Frequency in Set
Direction of Regulation
Frequency in set: the frequency of TFBS within the examined set; frequency ratio: frequency in set divided by frequency in background set.
First 24 hours of cultivation
5.92 × 10–17
9.56 × 10–14
5.89 × 10–08
3.16 × 10–03
9.78 × 10–14
2.21 × 10–05
After partial hepatectomy
8.03 × 10–11
4.97 × 10–09
1.38 × 10–07
1.67 × 10–06
8.71 × 10–06
CM Culture Conditions Decrease Expression of Metabolic Genes and Induces Cell Cycle and Cytoskeleton Associated Genes.
Gene set enrichment analysis based on ‘KEGG Pathway’ definitions (http://www.genome.jp/kegg/) was performed and identified down-regulation of genes related to metabolism on both platforms (Supporting Table 2). In more detail, almost 50% of genes related to fatty acid metabolism (P = 4.2 × 10−16), 30% of genes related to bile acid biosynthesis (P = 2.2 × 10−8) and approximately 50% of genes related to ketone bodies metabolism (P = 6.6 × 10−6) were down-regulated. In contrast, genes related to regulation of the actin cytoskeleton (P = 9.76 × 10−10) or involved in cell-cell communication, like focal adhesions, tight junctions and adherence junctions were up-regulated. Additionally, 16% of genes related to cell cycle (P = 1.9 × 10−6), 18% related to ribosomes (P = 3.0 × 10−5) and 10% related to MAPK signaling pathway (P = 4.5 × 10−4) were up-regulated (Supporting Table 3). Up-regulation of three selected genes CDK4, CDK6 and rRAS2, representing members of either the MAPK pathway or the cell cycle, was validated with quantitative rtPCR (Fig. 2D). These results show that murine hepatocytes cultured in CM for 1 day lose many of their metabolic functions, but increase gene expression signatures related to cell cycle control and cell-cell/cell-matrix interactions.
Overrepresentation of ETF, E2F1, and SP-1 Transcription Factor Binding Sites in Differentially Expressed Genes.
We were interested whether changes in gene expression are related to the action of specific transcription factors (TF). Using PRIMA (PRomoter Integration in Microarray Analysis),15 we identified ETF, E2F1, and SP-1 binding sites enriched in differentially expressed genes of CM cultured mouse hepatocytes (Table 1; for a complete list, see Supporting Tables 4 and 5). In order to validate this insilico approach, FUNCTIONAL ANNOTATION OF THE MAMMALIAN GENOME (FANTOM 4; http://fantom.gsc.riken.jp/4/) was used, a database containing experimental data (e.g., ChIP-chip data) of mouse and human genomes, in a database query (Fig. 2A). As listed in Supporting Table 3, E2F1 and SP-1 are ChIP-chip proven regulators of most of the up-regulated genes downstream of the MAPK pathway. In addition, E2F1 and SP-1 are regulators for all overexpressed “cell cycle”-related genes (Fig. 2B). To confirm that this validation is applicable to our data, PNCR2 was chosen as one overexpressed gene with an E2F1 TFBS (Supporting Table 5). Chromatin immunoprecipitation (ChIP) demonstrated E2F1 binding close to transcriptional start site of PNCR2 (Fig. 2C), as predicted by the insilico approach.
Finally, changes in expression of E2F1 and SP-1 during CM culture were quantified using qRT-PCR. Expression of these TFs was significantly up-regulated in cultured murine hepatocytes in the absence of cytokine stimulation within 48 hours (Fig. 3A). Furthermore, nuclear TFs were isolated from hepatocytes after cultivation for 4 hours and 24 hours. Within that cultivation period a selected nuclear enrichment of E2F1 and SP-1 compared to some other TFs (especially some STAT factors involved in cytokine signaling) could be demonstrated (Fig. 3B). Based on these data, we conclude that changes in gene expression occurring during the first 2 days of mouse hepatocyte CM culture in the absence of cytokines are mediated by E2F1, SP-1 and ETF transcription factors.
CM Culture Primes Hepatocytes for Proliferation.
E2F1 is controlled by MAP kinase signaling.16 Therefore, we asked whether the Ras-ERK1/2 pathway is activated during hepatocyte CM culture without cytokine stimulation. Indeed, we observed that the amount of phosphorylated ERK1/2 started to increase after 48 hours and reached a maximum after 72 hours of cultivation (Fig. 3C).
Next we sought to determine whether spontaneously activated MAP-kinase signaling leads to increased hepatocyte proliferation by analyzing BrdU incorporation in unstimulated mouse hepatocytes on CM. Subconfluent cultures (125,000 hepatocytes/9 cm2) showed a basal BrdU labeling index (LI) of 20% (Fig. 4A,C), which is clearly enhanced compared to in situ conditions in normal mouse liver. The initial 4 hours cultivation of hepatocytes in the presence of 10% FCS does not induce a proliferative state, because the LI in the presence and absence of FCS was equal (Supporting Fig. 3). In order to test whether BrdU incorporation can be enhanced further by the presence of cytokines, 20 ng/mL HGF was added in combination with 0.1 μM insulin. This induced a rapid and transient activation of the MAPK pathway as indicated by rapid ERK1/2 phosphorylation (Supporting Fig. 2) and a further increase of the LI to 80% (Fig. 4B).
Blunting MAP kinase signaling with MEK inhibitor PD98059 blunted HGF stimulated and basal BrdU incorporation (Fig. 4B), indicating the requirement of this pathway for induced and spontaneous hepatocyte proliferation in CM culture.
Of note, hepatocyte proliferation was much lower in confluent hepatocyte CM cultures (LI < 6%; Fig. 4D), demonstrating that the state of confluency is critical for proliferation. However, after generating an empty space in a confluent layer by “wounding,” hepatocytes again started to incorporate BrdU (Fig. 5C) preferentially at the “wound” margin, although hepatocytes distant from the “wound” also were able to incorporate BrdU to some extent (Fig. 5B).
We conclude that cultivation of murine hepatocytes on CM for 48 hours induces a proliferative cell state, characterized by an activated MAP kinase pathway. MAP kinase downstream signaling induces cell cycle associated genes and a LI of 20% in subconfluent cultures. In confluent cultures proliferation events are prevented by high cell densities, probably through contact inhibition.
Activation of ERK1/2 Signaling Is an Early Event in CCl4 Intoxication and Precedes SP-1 and ETF Activation.
Next, we were interested whether activation of the MAPK pathway and overrepresentation of binding sites for SP-1, ETF and E2F in differentially expressed genes observed in CM cultured mouse hepatocytes may be relevant for in vivo models of hepatocyte priming.
Intoxication with CCl4 transiently increased p-ERK1/2 levels in mice. Elevated levels were found after 20 minutes (Fig. 6A), which decreased again after 60 minutes (Fig. 6B). In addition, E2F1 levels remained up-regulated both in the cytoplasm and in the nucleus until 1 hour after CCl4 challenge (Fig. 6D) and decreased within the next 6 hours (Fig. 6E). SP-1 expression showed a similar time course (Fig. 6G,H). To the best of our knowledge, these data show for the first time that CCl4 intoxication results in a rapid activation of the MAPK pathway in vivo, which then triggers up-regulation of transcription factors E2F1 and SP-1, thus very rapidly directing murine hepatocytes into a proliferative state. The mechanisms seem to have much in common with hepatocyte culture on CM, but occur on a much shorter time-scale.
Partial Hepatectomy Results in an Overrepresentation of E2F, SP-1 and ETF Transcription Factors.
Liver regeneration can be induced by PHx. This prompted us to explore, whether the TFBS for E2F, SP-1 and ETF are also overrepresented in differentially expressed genes under these conditions. Published gene expression data17 were used for a promoter analysis using PRIMA. Indeed, differentially expressed genes following PHx also showed an overrepresentation of the three TFBS (Table 1). In line, elevated levels of E2F1 in liver after PHx were detected by immunohistochemistry (Fig. 7).
Cultured primary hepatocytes represent a valuable tool for studying molecular mechanisms of hepatocyte physiology, including liver regeneration and the response of the liver to cytokines.11, 12, 18 Here, we show by a genome-wide analysis that primary mouse hepatocytes change their gene expression pattern within the first 24 hours after cultivation on CM. Changes comprised the down-regulation of many metabolic genes and the up-regulation of cytoskeleton-associated and cell-cell interaction-associated genes (Supporting Tables 2 and 3). Similar data have been reported before for primary rat hepatocytes19, 20 cultured on CM. However, whereas prominent up-regulation of cell cycle genes indicative of a preproliferative state was found herein for murine hepatocytes, no such alterations was reported for rat hepatocytes. In line with this, the BrdU labeling index (LI) in unstimulated rat hepatocytes was very low21, 22 but reached approximately 20% in murine hepatocytes (Fig. 4A). Likewise, cytokine stimulation increased the LI in rat hepatocytes by approximately 30%, whereas murine hepatocytes reached a LI of 70%-80%. These differences demonstrate that murine hepatocytes show a higher basal proliferative potential than rat hepatocytes, which is reflected also in their gene expression pattern. The lower basal proliferation of rat hepatocytes might be due to the presence of a growth factor dependent restriction point at the second third of the G1 phase in rat hepatocytes.23
The high proliferative potential of unstimulated murine hepatocytes cultivated on CM is also reflected in elevated levels of p-ERK1/2 that last for several days which are sensitive to inhibition of the MAPK pathway. Although BrdU incorporation is indicative only for progression of S-phase, a certain number of murine hepatocytes complete the cell cycle and undergo cell division, as proven by time lapse video microscopy (data not shown). ERK activation is sensitive to inhibition of MAPK pathway, but the reason for this continuous up-regulation of MAPK signaling is yet unknown. Again, there is a strong contrast to rat hepatocytes, where elevated levels of p-ERK1/2 are found only transiently within 24 hours in response to EGF.18
As shown experimentally for PNCR2 (Fig. 2C) and MAT2A24 as well as by insilico analysis of existing ChIP-chip data (Fig. 2A,B). binding sites of the TFs ETF, E2F1, and SP-1 are highly overrepresented in the set of genes up-regulated during cultivation on CM. Interestingly, all three transcription factors are partially involved in proliferation processes. ETF is known as a critical determinant of proliferation.25 E2F transcription factors are well known as critical regulators of genes required for appropriate progression through the cell cycle.26, 27
SP-1 can enhance both proliferation28 as well as cell cycle arrest,29 depending on the specific context. As shown recently, it interacts with E2F1 thereby stimulating S-phase entry and cell proliferation.29 Thus, in the context of cultivation of murine hepatocytes on CM, it is likely that all three TFs contribute to a pre-proliferative state, even in the absence of cytokines. Because these TFs are themselves up-regulated in cultured murine hepatocytes (Fig. 3A), they may be part of a positive feed-back loop contributing to the continuously increasing levels of p-ERK1/2.
To our surprise, increases in MAPK signaling, expression of E2F1 and SP-1 and differential gene expression as described herein for in vitro cultures could also be demonstrated in vivo in two different models of liver injury. Interestingly, however, all events occurred on a shorter time-scale. To the best of our knowledge, we first demonstrated that intoxication with CCl4 results in rapid phosphorylation of ERK1/2 within 20 minutes. Because CCl4-dependent activation of MAPK pathway is accompanied by up-regulation of E2F1 and SP-1, these obvious cytokine-independent events seem to be involved in triggering the early proliferative response of hepatocytes after CCl4 injury. In this model we were able in identifying further similarities between the in vivo and in vitro situation. For example, in the present work, down-regulation of genes related to amino acid metabolism was found in vitro in line, with the reported reduction of hepatic protein synthesis within 30 minutes after CCl4 intoxication.30 The decrease in vitro in genes related to oxidative phosphorylation fits with decreased levels of ATP following CCl4 intoxication.31 Furthermore, Ca2+ homeostasis is disturbed within minutes in vivo,31 due to changes in calcium binding and membrane transport.32 Altered Ca2+ levels may also induce proliferation in vitro.33
Concerning the second model, PHx, we could show by an insilico approach that the differential gene expression induced by PHx is also characterized by an overrepresentation of TFBS for the same TFs—ETF, E2F, and SP-1—indicating that such cytokine-independent events contribute to the initiation of proliferation under conditions where the integrity of parenchyma remains intact. Recent work34 showing that nuclear E2F1 increases in rat hepatocytes after one-third PHx corroborates the importance of this TF, independent of the species studied.
In conclusion, we have shown that cytokine independent events associated with cell-separation and cultivation prime murine hepatocytes for proliferation by activation of MAPK signaling. The transcription factors E2F1, ETF, and SP-1 seem to be key players in orchestrating the alterations in gene expression during the early culture period. Similar processes occur under in vivo conditions indicating that murine hepatocytes cultivated on CM represent a valuable system to study proliferation and regeneration processes.
Alexander Bauer, involved in the early work, passed away in 2009. We miss him as a competent, innovative, and valued colleague and friend. We would like to thank J. Amin and E. Bedawy for technical support, C. Kern, S. Lagoutte, D. Mahn, S. Mac Nelly, F. Struck, and especially K. Heise for excellent technical assistance, J. Böttger for providing tissue after PHx as well as K. Krohn and P. Süptitz for microarray analysis.