ERK2 but not ERK1 plays a key role in hepatocyte replication: An RNAi-mediated ERK2 knockdown approach in wild-type and ERK1 null hepatocytes

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

The mitogen-activated protein kinases (MAPKs) ERK1 and ERK2 have been implicated in various physiological events, and specific targeting of these MAPKs could affect cell proliferation in many cell types. First, to evaluate the potential specific roles of these two MAPKs, we analyzed the mitogenic response in regenerating liver after partial hepatectomy (PH) and in primary culture of hepatocytes isolated from ERK1-deficient mice. We show that ERK1 knockout and wild-type (wt) cells replicate with the same kinetics after PH in liver, in vivo, and in primary cultures of hepatocytes, in vitro. Indeed, Cyclin D1 and Cdk1 appear to be expressed concomitantly in knockout and wt cells, highlighting that hepatocytes progress in the cell cycle independently of the presence of ERK1. Second, we specifically abolished ERK2 expression by RNA interference in mouse and rat hepatocytes. We investigated whether small interfering RNA (siRNA) targeting ERK2 could specifically inhibit its expression and interfere with the process of replication. In ERK1-deficient hepatocytes, silencing ERK2 expression by RNA interference and ERK2 activation by U0126 clearly demonstrate that DNA replication is regulated by an ERK2-dependent mechanism. Furthermore, in rat wt hepatocytes, whereas ERK2 targeting inhibits late G1 and S phase progression, ERK1 silencing is devoid of any effect on cell proliferation, indicating that ERK1 cannot rescue ERK2 deficiency. Conclusion: Our results emphasize the importance of the MAPK cascade in hepatocyte replication and allow us to conclude that ERK2 is the key form involved in this regulation, in vivo and in vitro. (HEPATOLOGY 2007;45:1035–1045.)

The Ras-dependent mitogen-activated protein kinase (MAPK) signaling cascade participates in control of cell fate, proliferation, and survival in various mammalian organs, including the liver. In adult rodent hepatocytes and regenerating liver, the MAPK MEK/ERK cascade plays a key role in regulating G1 phase progression and consequently proliferation.1, 2 The ability to control precisely the order and timing of events is determinant for cell cycle regulation.3, 4

According to our previous data, the first part of G1 is devoted to growth factor–dependent MEK/ERK-morphogenic events, whereas the mitogenic signal occurs in mid-G1 phase. We showed that the sequential control mechanism of hepatocyte morphology and S phase entry by growth factor could involve successive activation of MEK2-ERK2 and then MEK1/MEK2-ERK1/ERK2 isoforms.3 The process in early G1 in relation to cytoskeletal reorganization induces hepatocyte spreading, making them permissive to DNA replication. In late G1 phase, MEK1/MEK2-ERK1/ERK2 activation is associated with accumulation of Cyclin D1 and mitogen-dependent progression of hepatocytes to S phase. The central role of ERK2 in regulation was highlighted by the observation that ERK2 was preferentially activated in early and mid-late G1 phase. However, although ERK1 was slightly expressed and phosphorylated in early G1, a gradual recruitment of this phosphorylated isoform appeared all along G1 phase progression.

Few studies have been performed concerning the specific role of ERK1 and ERK2 isoforms. However, the recent ablation of these 2 genes in mice has provided an excellent means to study the specific and shared functions of these 2 isoforms. ERK2 mutant mice embryos die in utero before day 8.5 because of defects in trophoblast and placental development and in mesoderm differentiation.5–7 Conversely, ERK1 knockout mice are viable, fertile, and of normal size.8, 9 Thus, ERK1 is apparently dispensable and ERK2 may compensate for its loss. However, in ERK1−/− thymocytes, maturation and proliferation in response to specific inducers were severely reduced.8 The ERK1 isoform has been shown to be specifically required for in vitro and in vivo adipogenesis.10 Moreover, mice lacking ERK1 have an enhanced synaptic plasticity facilitating striatal-mediated learning and memory.11 Finally, recent work done on NIH3T3 cells has revealed that ablation of ERK1 was accompanied by an increase of ERK2 signaling, and consequently by enhanced growth, suggesting the importance of ERK2 in controlling cell proliferation. Conversely, ERK1 overexpression in these cells induced a decrease in ERK2 activation accompanied by a reduction in oncogenic Ras-mediated proliferation, which suggests that ERK1 could antagonize ERK2 activity.12

The capacity of the livers of these ERK1-deficient mice to regenerate, in response to partial hepatectomy (PH), has not been studied. The consequences of this deficient gene expression on cell cycle progression or checkpoint control in G1 phase and G1/S transition have not been investigated. To evaluate the potential specific roles of ERK2 compared with ERK1 in proliferating hepatocytes, we analyzed the mitogenic response in liver after PH and in primary culture of hepatocytes obtained from ERK1-deficient mice. Liver regeneration triggered by PH is a well-established in vivo model to dissect normal hepatocyte growth control mechanisms.1, 13–17 Indeed, removal of 70% of the liver mass induces a synchronized growth response in most hepatocytes from the remaining tissue. Primary culture of hepatocytes is a powerful model to understand the mechanisms that regulate proliferation. Indeed, in normal liver, hepatocytes can remain quiescent for very long periods. However, after tissue disruption during cell isolation, hepatocytes can enter the G1 phase and undergo, depending on the culture conditions, at least 1 round of division with or without growth factor addition in vitro.18–22

In this study, the specific role of ERK1 in hepatocyte proliferation has been fully investigated in ERK1 knockout mice. Data now emerging from the literature suggest that ERK1/ERK2 could have distinct roles to contribute to a regulated cell cycle and apoptosis progression. We wanted to learn whether loss of ERK1, despite its lack of apparent effects on normal liver development, could impair the proliferative response of differentiated hepatocytes in vivo and in vitro. Then, ERK2 activation and expression were silenced using a specific MEK1/2 inhibitor, U0126, and RNA interference, respectively. We analyzed the consequence of ERK1 knockout and ERK2 knockdown on cell cycle progression. We extended our study to normal rat hepatocytes, a well-established and controlled model to study events implied in the progression of the cell cycle in G1 and S phase, using siRNA targeting both ERK1 and ERK2 isoforms.

Abbreviations

BrdU, bromodeoxyuridine; EGF, epidermal growth factor; FBS, fetal bovine serum; MAPK, mitogen-activated protein kinase; MLC, myosin light chain; PBS, phosphate-buffered saline; PH, partial hepatectomy; siRNA, small interfering RNA; TBS, Tris-buffered saline; wt, wild type.

Materials and Methods

Chemicals.

Bromodeoxyuridine (BrdU) and [methyl-3H]thymidine (5 Ci/mmol) were from Amersham Corp. (Buckinghamshire, UK); insulin I-5500 was from Sigma (Saint Quentin Fallavier, France); recombinant human EGF and MEK inhibitor U0126 were from Promega (Charbonnières, France).

Animals.

Wild-type (wt) mice (weighing approximately 20 g) and male Sprague-Dawley rats (weighing approximately 200 g) were from Charles River France (L'Arbresie, France). ERK1−/− mice were obtained from Gilles Pages (Nice, France). Animals were given food and water ad libitum, and experiments were carried out in accordance with French laws and regulations. For PH, 70% of the liver was removed, which is known to induce a synchronized growth response that involves almost only hepatocytes during the first wave of replication. At different times after PH, animals were killed; the livers were harvested, immediately frozen in liquid nitrogen, and stored at −80°C until analysis.

Cell Culture.

Hepatocytes were isolated from mouse and rat livers by the two-step perfusion procedure using 0.025% collagenase (Boehringer-Ingelheim, Gagny, France) buffered with 0.1 M Hepes (pH 7.4) as described.18 They were plated at a density of 5 × 104 (50,000) cells/cm2 in 35-mm diameter dishes in 2 ml minimal essential medium/medium 199 [3:1 (vol/vol)] containing penicillin (100 IU/ml), streptomycin (100 μg/ml), insulin (5 μg/ml), glutamine 2 mM, and bovine serum albumin (1 mg/ml). The medium was supplemented or not with 10% fetal bovine serum (FBS) for 4 hours as indicated in the figure legends. Four hours after plating, the medium was replaced with basal medium without any FBS and renewed every day. Epidermal growth factor (EGF) stimulations were performed at 50 ng/ml. At the indicated times, U0126, dissolved in DMSO, was added at the final concentration of 50 μM. All control cultures containing DMSO at final concentration of 0.5% were changed at the same time as treated cells.

Immunoblotting Analysis.

For harvesting, cultures were rinsed with phosphate-buffered saline (PBS) and lysed in the homogenization buffer (60 mM β-glycerophosphate, 15 mM p-nitrophenylphosphate, 25 mM MOPS pH 7.2, 15 mM EGTA, 15 mM MgCl2, 2 mM DTT, 1 mM vanadate, 1 mM NaF, 1 mM phenylphosphate, 100 μM benzamidin, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 10 μg/ml soybean trypsin inhibitor). The amount of total protein was determined using the Bio-Rad Protein assay (Bio-Rad, Hercules, CA). After sodium dodecyl sulfate polyacrylamide gel electrophoresis, proteins were transferred onto nitrocellulose membranes using a Trans-Blot TM Cell apparatus (Bio-Rad) for 1 hour at 400 mA in Tris buffer 25 mM, glycine 192 mM, ethanol 20%. The amount of protein loaded was checked with Ponceau dye. Subsequently, filters were rinsed in Tris-buffered saline (TBS pH 7.4), blocked with 5% nonfat dry milk in TBS at room temperature, and incubated overnight at 4°C with primary antibodies diluted in the same buffer. Anti–phospho-ERK1/ERK2 was a mouse monoclonal antibody directed to a synthetic phosphotyrosine peptide corresponding to residues 196 to 209 of human p44 MAPK (New England Biolabs, Beverly, MA). Polyclonal antibodies against ERK1 (sc-94) and ERK2 (sc-154) were from Santa Cruz Biotechnology (Santa Cruz, CA). The anti-cyclin D1 antibody was from Neomarkers (Westinghouse, CA). The anti-Cdk1 is a polyclonal antiserum specifically directed against the C-terminal part of human p34.18 Anti–myosin light chain (MLC) antibody was from Sigma (St. Quentin Fallavier, France). After 3 washes in TBS, membranes were incubated in 5% nonfat dry milk in TBS for 1 hour and with horseradish peroxidase–conjugated secondary antibody for 1 hour at room temperature. After 3 washes in TBS, proteins were detected according to the SuperSignal Ultra Chemiluminescent Substrate procedure (Pierce, Rockford, IL).

Densitometric analysis of the bands was performed using Quantity One Software developed by Bio-Rad (Hercules, CA).

[3H]Thymidine Incorporation.

The rate of DNA synthesis was measured in primary cultures, by adding 2 μCi [methyl-3H]thymidine (5 Ci/mmol) for given periods before cell harvesting as indicated. Cells were washed twice in PBS, scraped from the petri dish, and aliquoted for protein content determination and [3H]thymidine counting after precipitation and washing in trichloroacetic acid. Results are expressed as a percentage of control [methyl-3H]thymidine incorporation.

In Vivo BrDU Labeling.

After PH, in vivo incorporation of the thymidine analog BrdU was used as an index of cell proliferation. BrdU was dissolved in PBS and given intraperitoneally (5 mg/100 g body weight) for 1 hour before killing. BrdU-positive cells were detected by using a cell proliferation kit (Amersham, Orsay, France). Percentages of BrdU-labeled hepatocytes were determined in each condition.

Design and Transfection of ERK1-Specific and ERK2-Specific siRNAs.

Small interfering RNA (siRNA) sequences against ERK1 and ERK2 were designed using criteria previously described.23 Sequences, with two 3′ deoxythymidine overhands, were: siERK2 (duplex 1), 5′-GUG CUG UGU CUU CAA GAG C-3′; siERK2 (duplex 2), 5′-UCA CAA GAG GAU UGA AGU U-3′; siERK-1 (duplex 1), 5′-UGA CCA CAU CUG CUA CUU C-3′; siERK1 (duplex 2), 5′-CUG GCU UUC UGA CCG AGU A-3′; and a control siRNA differing from the siERK2 (duplex 1) by 3 nucleotides. Sequences were purchased from Eurogentec. Absence of matching of these sequences with any other mouse and rat gene was checked via the NCBI standard nucleotide-nucleotide BLAST program. Transfection of siRNAs was performed 24 hours after plating using the Transfectin lipid reagent (Bio-Rad Laboratories, France). Hepatocytes, cultured in 6-well plates, were incubated with the transfection mix containing 100 nM siRNA and 5 μl Transfectin in 1 ml OptiMEM (Invitrogen, France), in agreement with the manufacturer's instructions. After 3 hours' incubation, the transfection medium was removed and cells were changed in serum-free medium. Stimulations by FBS or EGF were done 48 hours after plating.

All experiments were done at least 3 times.

Results

Kinetics of Cyclin D1 and Cdk1 Expressions and BrdU Incorporation After PH of WT and ERK1−/− Livers In Vivo.

ERK1 and ERK2 have been shown to be implicated both in liver regeneration and in hepatocyte proliferation in vitro. To discriminate the role of these 2 kinases in these processes, we analyzed the kinetics of liver regeneration in wt and ERK1 knockout animals.

We first assessed kinetics of ERK protein expressions at different times after PH in ERK1 knockout livers compared with wt livers. Livers from wt mice showed sustained ERK1 and ERK2 expression throughout the analyzed period (Fig. 1A). As expected, no ERK1 expression was detected in livers from ERK1 knockout mice. The patterns of ERK1 and/or ERK2 expressions did not exhibit any detectable variation within the interval analyzed (12–96 hours). Next, we studied the cell cycle progression after PH and analyzed the expression of two markers: cyclin D1, which is induced in late G1 phase, and Cdk1, a cell cycle–modulated kinase expressed in S phase in regenerating livers and proliferating hepatocytes.16 As expected, cyclin D1 protein expression appeared 24 hours after PH and was expressed up to 72 hours in wt liver. In ERK1−/− animals, a similar kinetic pattern was found for cyclin D1. Moreover, Cdk1 was also concomitantly expressed 48 hours after PH, in wt and knockout mice, indicating that hepatocytes could progress in late G1 and S phase in absence of ERK1. These results indicate that, in vivo, cell cycle replication was independent of the presence of ERK1.

Figure 1.

Regeneration in wt and ERK1−/− mouse livers. (A) Kinetics of ERK1/ERK2, cyclin D1, and Cdk1 expressions in wt (+/+) and ERK1 knockout (−/−) livers at the indicated times after partial hepatectomy (PH). (B) BrdU incorporation 72 hours after PH. (C) Quantification of BrdU incorporation at the indicated times after PH.

Furthermore, we analyzed DNA replication, in wt and ERK1−/− regenerating livers by measuring BrdU incorporation in vivo (Fig. 1B,C). No significant differences in BrdU incorporation were seen between wt and ERK1−/− regenerating livers as observed at 48, 72, 96, and 120 hours after PH. Moreover, no difference in the percentage of diploid cells versus tetraploid hepatocytes could be detected.

DNA Replication in ERK1 Knockout and WT Hepatocytes In Vitro.

To further investigate the effect of the absence of ERK1 on hepatocyte DNA replication, we analyzed [methyl-3H]thymidine incorporation in primary cultures of ERK1-deficient hepatocytes. DNA synthesis was analyzed in basal condition or in medium supplemented by FBS, EGF, or both (Fig. 2).

Figure 2.

Time course of [methyl-3H]thymidine incorporation into DNA in mouse hepatocytes isolated from wt (A) and ERK1−/− (B) livers. Primary cultures of hepatocytes were stimulated or not (control) with FCS or EGF, alone or together. [Methyl-3H]thymidine incorporation was analyzed at the indicated times after seeding.

In wt hepatocytes, DNA replication clearly occurred in the absence of EGF and FBS treatment (Fig. 2A). DNA replication peaked 72 hours after seeding and decreased slowly thereafter. The kinetics of replication were not greatly influenced by EGF or FBS treatment, and only a slight increase in thymidine incorporation was noted in the presence of EGF. We did not detect any significant differences in the kinetics of replication between wt and ERK1 knockout cells (Fig. 2B). As noted in wt cells, a slight increase in thymidine incorporation occurred in EGF-stimulated ERK1−/− hepatocytes. By showing that cyclin D1 and Cdk1 started to be expressed concomitantly in wt and knockout hepatocytes, we confirmed that ERK1 knockout hepatocytes could progress in late G1 and S phase (result not shown). All these experiments, in vivo and in vitro, ascertain that ERK1 is dispensable for full DNA replication of ERK1−/− hepatocytes.

ERK2 Inhibition and Silencing in ERK1 Knockout Hepatocytes.

To further understand the role of ERK2, we first inhibited ERK2 activation with the specific MEK1/MEK2 inhibitor U0126 and, second, silenced ERK2 expression by RNAi in ERK1 knockout hepatocytes.

Initially, we found that DNA replication of ERK1−/− hepatocytes was almost totally inhibited in presence of U0126 (Fig. 3A). This inhibition was obtained both in FBS-stimulated and EGF-stimulated cultures. In control experiments, ERK2 phosphorylation was completely abolished with the MEK inhibitor, and no variation in the amount of ERK2 (loading control) could be noted 10 and 24 hours after EGF stimulation (Fig. 3B). Moreover, ERK2 inhibition with U0126 completely inhibited ERK1−/− hepatocyte spreading but not adhesion to the support (data not shown) as described for primary culture of wt rat hepatocytes.3 A reversion experiment allowed us to demonstrate that blockade of cell spreading by MEK inhibition was not toxic because those hepatocytes could spread within 6 to 8 hours when U0126 was withdrawn, the cells undertaking a morphological appearance close to that of the control culture.

Figure 3.

Inhibition of ERK2 activation/expression and DNA replication with the specific MEK inhibitor U0126 and by RNA interference. (A,B) Hepatocytes from ERK1−/− knockout mouse livers were cultured with EGF or FBS in the presence or absence of U0126 (50 μM). The solvent (DMSO) was added in the control experiment. (A) [Methyl-3H]thymidine incorporation into DNA between 24 and 48 hours after seeding in presence or absence of U0126. (B) Western blotting of phospho-ERK2 (top panel) after 10 and 24 hours of U0126 treatment in presence of EGF. The blot was reprobed with an anti-ERK2 antibody (lower panel). (C,D) Mouse hepatocytes from ERK1−/− knockout livers were transfected with a siERK2 or a control siRNA, 24 hours after seeding. (C) ERK2, Cdk1 expressions, and ERK2 activation (P-ERK) were analyzed 48 and 72 hours after transfection. Myosin light chain (MLC) was used as a loading control. (D) Time course of [methyl-3H]thymidine incorporation into DNA in ERK1−/− primary cultures of hepatocytes, transfected by a siERK2 or by a control siRNA and analyzed at the indicated times after transfection.

Second, inhibition of ERK2 gene expression by siRNA was performed. This strategy of sequence-specific gene silencing at the posttranscriptional level was performed in primary cultures of ERK1 knockout hepatocytes.

Because this technique has not still been fully applied in hepatocytes, we first optimized our protocol. We used a rhodamine-labeled siRNA and observed that, in optimal experiments, when cells were transfected for 5 hours with Transfectin, more than 90% of the hepatocytes appeared positive for rhodamine (result not shown).

ERK2 silencing was performed in ERK1 knockout hepatocytes transfected with the double-stranded ERK2 RNA duplex, 20 hours after seeding. Endogenous ERK2 expression and DNA replication were analyzed after transfection at the indicated times (Fig. 3C, D). At 48 and 72 hours after transfection, the siERK2 inhibited protein expression by 80% and 95%, respectively, compared with control transfection experiments (Fig. 3C). ERK2 phosphorylation was also slightly to highly inhibited, 48 and 72 hours after transfection, respectively. ERK2 knockdown correlated with inhibition of Cdk1 expression, and DNA replication was also strongly inhibited in these cells (Fig. 3D), confirming results obtained with chemical inhibitors showing that ERK2 plays a major role in ERK1−/− hepatocyte replication.

ERK2 and ERK1 Silencing in WT Rat Hepatocytes.

We then attempted to extend our results to wt rat hepatocytes. Indeed, contrary to mice, rat hepatocytes do not proliferate spontaneously in vitro and must be stimulated by a growth factor to enter the S phase and proliferate. We previously demonstrated the existence of a growth factor–dependent restriction point located at the second third of the G1 phase at which hepatocytes stay blocked if not stimulated by a growth factor. Addition of growth factor (i.e., EGF) in the medium allows cells to go over this MEK/ERK–dependent control point and to reach late G1 and S phases.1 Consequently, primary culture of rat hepatocytes is a well-established controlled model to study events implied in the progression of the cell cycle in G1 and S phase.

First, we silenced ERK2 using 2 siRNAs: siERK2(1) and siERK2(2) (Fig. 4). We obtained an 85% to 90 % decrease of the expression level of the isoform with both siRNAs (Fig. 4A). This inhibition correlated with a strong decrease in ERK2 phosphorylation. ERK1 protein expression was not affected by ERK2 silencing, showing the high specificity of this inhibition for ERK2. Moreover, silencing ERK2 was accompanied by an increase in ERK1 phosphorylation. The decrease in ERK2 expression/phosphorylation correlated with a decrease of cyclin D1 and Cdk1 expressions. Furthermore, in these wt hepatocytes, inhibition of DNA replication (60%) was obtained, indicating that ERK2 alone could be responsible for the replicating property of the pathway (Fig. 4B).

Figure 4.

ERK2 silencing in wt rat hepatocytes. (A) ERK1/ERK2, cyclin D1, Cdk1 expressions, and ERK1/ERK2 activation (P-ERK) in wt rat hepatocytes transfected with siERK2 (duplex 1 and 2), or siControl 24 hours after seeding. Cells were stimulated by EGF 48 hours after seeding. Expression levels were analyzed at the indicated times after transfection. MLC is the loading control. (B) Time course of [methyl-3H]thymidine incorporation into DNA in wt rat hepatocytes transfected with siERK2 (duplex 1 and 2) or with siControl and analyzed at the indicated times after transfection.

In the same way, we targeted ERK1, using 2 siRNA duplexes: siERK1(1) and siERK1(2) (Fig. 5). Our results show that in the presence of specific siRNAs, ERK1 expression/phosphorylation decreased by 85% to 90%, without affecting ERK2 expression as compared with control cells (Fig. 5A). As previously reported for phospho-ERK1 in ERK2-inhibited cells, a significant increase of ERK2 phosphorylation was noted in ERK1-inhibited cells. Contrary to ERK2 silencing, ERK1 knockdown had only a slight effect on cyclin D1 expression. Moreover, DNA replication was not affected by the knockdown of ERK1, showing that hepatocytes did progress in S phase (Fig. 5B). However, a partial decrease of Cdk1 expression in ERK1 knockdown cells was observed, which suggests that ERK1 could nevertheless take part to the cell cycle progression in G1 phase but would not be essential, as replication occurred when it was silenced. Our results demonstrate that specific ERK1 knockdown and knockout do not affect the progression of wt hepatocytes in S phase, showing that ERK1 is dispensable for full DNA replication.

Figure 5.

ERK1 silencing in wt rat hepatocytes: (A) ERK1/ERK2, cyclin D1, Cdk1 expressions, and ERK1/ERK2 activation (P-ERK) in wt rat hepatocytes transfected with siERK1 (duplex 1 and 2), or siControl, 24 hours after seeding. Cells were stimulated by EGF 48 hours after seeding. Expression levels were analyzed at the indicated times after transfection. MLC is the loading control. (B) Time course of [methyl-3H]thymidine incorporation into DNA in wt rat hepatocytes transfected with siERK1 (duplex 1 and 2) or with siControl and analyzed at the indicated times after transfection.

As observed previously, silencing ERK2 in rat hepatocytes induced an important decrease in DNA synthesis. However, DNA replication was not totally abolished. This result could be explained by the involvement of ERK1 isoform, which could partially compensate the inhibition of ERK2 but not sufficiently to recover a normal proliferating rate. Consequently, we targeted ERK1, besides ERK2, in rat hepatocytes, to analyze whether an inhibition of ERK1, in an ERK2-silenced context, could totally abolish the DNA synthesis (Fig. 6).

Figure 6.

Dual silencing of ERK2 and ERK1 in wt rat hepatocytes. (A) ERK1/ERK2, cyclin D1, Cdk1 expressions, and ERK1/ERK2 activation (P-ERK) in wt rat hepatocytes transfected with siERK2 alone, siERK1 alone, both siERK2 and siERK1, or siControl, 24 hours after seeding. Cells were stimulated by EGF 48 hours after seeding. Expression levels were analyzed 72 hours after transfection. (B) Densitometric quantification of ERK1 and ERK2 expressions and (C) quantification of ERK phosphorylation related to total ERK protein were done on 3 independent experiments. (D) Time course of [methyl-3H]thymidine incorporation into DNA in wt rat hepatocytes transfected with siERK2 alone, siERK1 alone, both siERK2 and siERK1, or siControl and analyzed 72 hours after transfection.

Dual knockdown of ERK1 and ERK2 induced a strong decrease of the expression level of both isoforms, which was quantified as being close to 90% to 95% (Fig. 6A,B). When both isoforms were inhibited, cyclin D1 and Cdk1 expressions decreased in the same way as in ERK2 knockdown cells. Finally, we did not obtain a higher decrease of [3H]thymidine incorporation when both isoforms were silenced compared with ERK2-silenced cells (Fig. 6D). As in Fig. 5A, we could note a decrease in Cdk1 expression in ERK1 knockdown cells, however, not sufficiently to interfere with DNA replication. All these results argue that ERK1 cannot compensate ERK2 proliferating function.

Moreover, we confirmed that ERK1 and ERK2 silencing were accompanied by an increase of ERK2 and ERK1 phosphorylation by 70% and 60%, respectively (Fig. 6C). Although dual inhibition led to a strong decrease of both protein expressions, we noted an increase in the phosphorylation rate of the remaining proteins: an 11-fold increase for phospho-ERK2/ERK2 and phospho-ERK1/ERK1 ratios, in comparison with control siRNA transfected cells. Our results show that up-phosphorylation of the remaining proteins is not sufficient to overcome the decrease of protein synthesis in siRNA-transfected cells because DNA replication was highly inhibited in ERK1/ERK2–silenced hepatocytes as in ERK2 knockdown cells.

Discussion

In this study, we focused on the specific and critical role of ERKs in hepatocyte proliferation, using two experimental models: ERK1 knockout in mouse hepatocytes, ERK1 and ERK2 knockdown in rat hepatocytes. Whereas ERK1 knockout and knockdown have no effect on mouse and rat hepatocyte proliferation, respectively, specific ERK2 knockdown in both species abolishes cell cycle progression and defines ERK2 as a pivotal mediator of hepatocyte cell cycle progression both in vitro and in vivo.

Complementary reports regarding the role of the MEK/ERK signaling pathway in hepatocyte cell cycle progression may be found in the literature. As already reported by others and us, the MAPK/ERK cascade is involved in the regulation of G1 phase progression during rat liver regeneration in vivo and in proliferating hepatocytes in vitro.1, 24–26 Some reports have suggested that sustained activation of ERK inhibits hepatocyte DNA replication and that transient activation of this pathway stimulates DNA synthesis.21, 27–29 Leu et al.2 reported impaired hepatocyte DNA synthesis after hepatectomy with defects in ERK regulation in IGFBP1-deficient mice. In this work, we found this MAPK pathway is involved in the regulation of hepatocyte proliferation and we demonstrate that ERK2 is responsible for the MAPK proliferative response.

We first showed that ERK1−/− liver regenerates with the same kinetic as wt liver. In mice, hepatocyte proliferation in vitro occurs in the absence of growth factor in knockout as well as in wt hepatocytes. Targeting MEK with the specific chemical inhibitor, U0126, in ERK1 knockout hepatocytes abolished DNA replication. This confirms the importance of the MAPK ERK1/ERK2 pathway in the regulation of hepatocyte proliferation in mice, similar to that previous described in the rat.1 Second, we emphasized the role of ERK2 by showing that in ERK1 knockout hepatocytes, ERK2 silencing by RNA interference strongly abolishes DNA replication. Finally, we confirmed and reinforced these data in wt rat hepatocytes, which represent a well-known model to study the major events that regulate the progression of hepatocytes in G1 phase. ERK2 silencing abolishes late G1 and S phase progression, leading to a decrease of DNA replication. These data were confirmed in wt mouse hepatocytes, where ERK2 silencing also strongly decreased replication (data not shown). In wt mouse hepatocytes, siRNA interference inhibited ERK2 expression by 80% to 85% and this inhibition correlated with a decrease of cyclin D1 and cdk1 expressions. On the whole, by targeting ERK2 in rat and mouse wt hepatocytes, strong inhibition of DNA replication was obtained, indicating that ERK2 alone could support the major replicating property of the pathway.

We noticed an increase of ERK1 phosphorylation level in ERK2-silenced rat and mouse hepatocytes. This could be due to loss of competition between ERK1 and ERK2 for their binding and activation by MEKs, as reported: Mazzucchelli et al.11 showed that ERK1 knockout can be also accompanied by an increase in ERK2 signaling. In NIH 3T3 fibroblasts, however, ERK1 knockdown resulted in a significant increase in ERK2 activation profile, and loss of ERK2 only marginally affected ERK1 phosphorylation.12 Despite this hyperactivity, ERK1 is not able to counterbalance the lack of ERK2 with regard to its proliferating function in hepatocytes. Moreover, the dual silencing of ERK1 and ERK2 in rat hepatocytes did not result in a more pronounced reduction of proliferation compared with ERK2-silenced cells. Interestingly, remaining ERK1 and ERK2 proteins showed an 11-fold increase of their phosphorylation levels, confirming, first, that even up-phosphorylated ERK1 is not able to compensate for the drop of ERK2 expression, and second, that ERK2, despite its hyper-phosphorylation, needs to be expressed at a minimal threshold to play its proliferating role. All together, these results highlight that ERKs' function depends not only on their level of phosphorylation but also on their level of expression.

Little is known about the specific roles that ERK1 (p44) and ERK2 (p42) proteins play. Both isoforms have long been considered interchangeable in their functions, notably because of their: (1) sequence similarity, 75% at the amino acid level; (2) activation by same stimuli; (3) similar spatiotemporal regulations. Our results and other recent evidence suggest that each kinase protein could regulate different cellular events. The specific implication of ERK2 in cell proliferation, in comparison with ERK1, was ever observed in pre-adipocytes, whereas ERK1 was shown to be a key molecule of the differentiating process leading to mature adipocytes.10 Transfection of embryo fibroblast with a dominant-negative ERK1 was correlated with growth arrest.30 Moreover, recently, ERK1 ablation in mouse embryo fibroblasts and NIH3T3 cells by gene targeting and RNAi resulted in an enhancement of ERK2-dependent signaling and in a proliferating advantage. In these NIH 3T3 cells, knockdown of ERK2 almost completely abolishes normal and Ras-dependent cell proliferation.12 Although we did also observe an increase of phosphorylated ERK2 in ERK1 knockdown hepatocytes, we did not detect any increase of cell proliferation.

We demonstrated that ERK2 alone can support the replicative response because RNAi-mediated ERK2 knockdown is sufficient to inhibit cyclin D1, Cdk1 expressions, and DNA replication in ERK1 null and wt hepatocytes. From the literature, it is known that ERK2 is the main phosphorylated isoform in a wide range of proliferating cancer cell types, including cells from kidney, lung, and liver. During posthepatectomy progression and in proliferating hepatocytes in vitro, ERK2 is activated in early G1 phase in direct relation with a growth factor–induced morphogenic effect, whereas in mid late G1 phase, the 2 isoforms ERK1 and 2 are activated and seem to be involved in late G1 phase progression of proliferating hepatocytes.1, 3 Our conclusions can be extended to the MEK/ERKs morphogenic effects in EGF-stimulated rat hepatocytes, emphasizing the specific role of ERK2 in this particular regulation. In fact, U0126 MEK inhibitor treatment completely inhibits ERK1−/− hepatocytes spreading but not adhesion to the support, as demonstrated previously for rat hepatocytes (results not shown).3 Fassett et al.31 revealed important insights into the role of the MEK/ERK pathway and cell shape in hepatocyte cell cycle progression. They confirmed that the role of MEK in spreading is apparent when hepatocytes are dependent on MEK to produce their own matrix on which to spread. Furthermore, in cells adherent to a permissive extracellular matrix, cyclin D1 expression and progression into S-phase require MEKs activation. We previously provided information concerning the mechanism by which the growth factor can temporally control morphogenic and mitogenic signals during G1 phase in relation to cytoskeletal reorganization and integrin B1 and cyclin D1 up-regulation in a MEK-ERK–dependent manner.

In hepatocytes, as in many other cell types, up-regulation of cyclin D1 in mid-late G1 is indicative of G1/S transition and mitogenic response. Cyclin D1, which was shown to be regulated positively by the p42/p44 MAPK pathway,32 and Cdk1 were also reported up-regulated in regenerating liver in vivo after PH. The mechanism regulating cyclin D1 expression could be mainly ERK2/p70S6K dependent in rat hepatocytes, showing molecular cross-talk between the MEK/ERK and Pi3K pathways.33 In the absence of growth factors, cyclin D1 overexpression results in hepatocyte DNA synthesis and leads to activation of downstream biochemical events, including cyclin A– and E-associated kinases activation and Cdk1 induction.34 Furthermore, an exogenous expression of cyclin D1 restored DNA synthesis in MEK-inhibited cells, suggesting that the major role for MEK in hepatocyte cell cycle progression is to increase cyclin D1 expression.15 However, in knockout mice, lack of cyclin D1 might transiently delay entry into S phase but was not sufficient to inhibit the response of hepatocytes to mitogenic stimuli.35 Under our experimental conditions, overexpression of cyclin D1 could never allow non–growth factor–stimulated rat hepatocytes to progress in S phase, indicating that G1 arrest was not strictly attributable to the loss of cyclin D1 alone (results not shown).

Our results point out that the absence of ERK1 influences neither the expression pattern of cyclin D1 nor DNA synthesis in proliferating hepatocytes. However, Cdk1 was less expressed in ERK1 knockdown cells, suggesting that ERK1 isoform, although dispensable for hepatocyte proliferation, could interfere with Cdk1 regulation. Furthermore, we confirmed that proliferation is not delayed in ERK1−/− hepatocytes undergoing cyclin D1 (result not shown) and DNA synthesis at the same times as it is observed in wt cells. Our study demonstrates that ERK1 is apparently dispensable for hepatocyte replication, both in regenerating liver and in primary culture. These results are in accordance with Pagès' report8 showing that knockout ERK1−/− mice are viable, fertile, and of normal size.8 However, ERK1 could have a specific role in thymocyte development. In fact, in p44−/− thymocytes, proliferation is severely reduced in response to activation with a monoclonal antibody in presence of phorbol myristate acetate. Moreover, preliminary results obtained in ERK1 and ERK2 knockdown hepatocytes have shown a differential effect of the inhibition of each ERK protein kinase on caspase-3 activity. Whereas ERK2 silencing has no or little effect on caspase-3 activity, ERK1 inhibition did induce an important decrease, suggesting that ERK1 could play a negative role in hepatocyte survival. Moreover, cultured-ERK1 knockout hepatocytes could progress in apoptosis more slowly than wt hepatocytes (results not shown). Works are now in progress studying apoptotic engagement in ERK1-silenced hepatocytes. In this way, keratinocytes coming from ERK1−/− mice were recently shown to be more resistant to apoptotic signals.36 Whereas ERK2 seems to have a positive role in controlling cell proliferation, ERK1 could probably affect the overall signaling output of fibroblast NIH 3T3 cells by antagonizing ERK2 activity.12 Conversely, silencing ERK1 expression alone is sufficient to significantly decrease tumor cell viability.37

A recent study suggested a model in which the specific activation of either MEK1 or MEK2 is part of the cellular system that enables the mitogenic switch between inhibition or enhancement of cell proliferation.38 Our work and recent data from the literature underline the possible existence of specific functions for each ERK1 and ERK2 kinase.12 We are now working to define specific substrates, notably transcription factors, that ERK1 and ERK2 might phosphorylate to better understand shared and specific roles within the balance between proliferation, apoptosis, and cell survival.

In liver disease, irregular proliferation of hepatocytes is an important factor favoring hepatocarcinogenesis. Silencing gene expression by RNAi would therefore offer promising therapeutic perspectives. Silencing ERK1/ERK2 expression using RNA interference led to suppression of ovarian tumor cell proliferation.39 The specific ERK2 siRNA duplex might be an effective tool for preventing hepatocyte proliferation. Ongoing improvement of in vivo nucleic acid delivery technologies could enable RNAi to be used therapeutically in the near future, one step closer to meeting the challenge of gene therapy–based approaches in which a supply of siRNA could be introduced intracellularly to interfere with hepatocyte proliferation.

Acknowledgements

We thank C. Ribault for technical assistance and Dr. C. Brahimi-Horn for critical reading of the manuscript.

Ancillary