An MLCK-dependent window in late G1 controls S phase entry of proliferating rodent hepatocytes via ERK-p70S6K pathway


  • Potential conflict of interest: Nothing to report.


We show that MLCK (myosin light chain kinase) plays a key role in cell cycle progression of hepatocytes: either chemical inhibitor ML7 or RNA interference led to blockade of cyclin D1 expression and DNA replication, providing evidence that MLCK regulated S phase entry. Conversely, inhibition of RhoK by specific inhibitor Y27632 or RhoK dominant-negative vector did not influence progression in late G1 and S phase entry. Inhibition of either MLCK or RhoK did not block ERK1/2 phosphorylation, whereas MLCK regulated ERK2-dependent p70S6K activation. In addition, DNA synthesis was reduced in hepatocytes treated with p70S6K siRNA, demonstrating the key role played by the kinase in S phase entry. Interestingly, after the G1/S transition, DNA replication in S phase was no longer dependent on MLCK activity. We strengthened this result by ex vivo experiments and evidenced an MLCK-dependent window in late G1 phase of regenerating liver after two-thirds partial hepatectomy. In conclusion, our results underline an MLCK-dependent restriction point in G1/S transition, occurring downstream of ERK2 through the regulation of p70S6K activation, and highlighting a new signaling pathway critical for hepatocyte proliferation. (HEPATOLOGY 2006;44:152–163.)

The G1 phase is a preparative step where cells temporally integrate complex signals from the microenvironment. At this time, the cells are able to switch between proliferation, differentiation or apoptosis, depending on extracellular matrix composition, cytokines, and cell–cell contacts. In the hepatocyte, a highly differentiated cell, cell cycle progression in vivo during liver regeneration is regulated by a sequential acquisition of complex signals depending on transduction pathway activation.1–3 Primary cultures of hepatocytes are a powerful model in studying the precise sequence of events that are necessary for cell progression from a G0-like state to S phase. The model mimics the physiological process of hepatic regeneration after liver injury or partial hepatectomy. In this context, we have already demonstrated that epidermal growth factor (EGF) possesses two distinct and complementary effects on G1 phase hepatocytes.4 First, the growth factor is a morphogen in early G1 phase by inducing controlled spreading of hepatocytes via integrin β1 regulation. Second, a mitogenic effect occurs in mid-late G1 phase and allows hepatocytes to progress through a restriction point located two thirds of the way through G1 phase.5–7 In addition, EGF promotes cell progression up to late G1.7 In hepatocytes, EGF activates specific transduction pathways. MEK and PI3K signaling cascades are essential for progression past the G1/S checkpoint and hepatocyte progression to S phase.8 They both control expression of cyclin D1, a key cell cycle protein that is upregulated in the pre-replicative phase of liver regeneration and in primary culture hepatocytes.4, 9, 10

MLCK and RhoK, two kinases linked to the contractile apparatus, are known to be involved in the regulation of a wide range of cellular functions, including proliferation. In recent years, extensive studies on RhoK have implicated this kinase in mitotic cell rounding,11 membrane blebbing,12 cancer cell migration,13 control of cell polarization,14 and differentiation.15 In the presence of either RhoK inhibitor or dominant-negative p160 RhoK, hepatocarcinoma cells failed to form lamellipodia and showed a lower invasiveness of infiltrative growths into the sinusoidal area at the tumor boundary in the liver.16 Conversely, MLCK is involved in axon guidance regulation,17 in apoptosis,18, 19 and in the control of endothelial permeability.20 Extended studies on relationships between signaling pathways and motility effectors identified Rho/RhoK–MEK cooperation,21 inhibition of RhoK expression by activated MEK,22, 23 and activation of MLCK by the MEK/ERK pathway.24–26 Moreover, Rho is required for G1 to S progression in mammary epithelial cells27 and RhoK is primordial for the progression of 3T3 fibroblasts to S phase.28 Little is known about the role of MLCK and RhoK in the cell cycle progression in hepatocytes. Studies in hepatocytes have demonstrated that several proteins involved in motility are upregulated after growth factor stimulation.29, 30 Furthermore, inhibition of MLCK was found to block DNA synthesis in hepatocytes cultured on high-density fibronectin.31

In the current study, the involvement of MLCK and RhoK in late G1 phase progression was specified, and we evidenced the pathways involved in MLCK-dependent regulation of G1/S transition in proliferating hepatocytes. This effect was enforced by the establishment of a precise window, in late G1, in which inhibition of MLCK blocked hepatocyte progression.


EGF, epidermal growth factor; cdk, cyclin-dependent kinase; ERK, extracellular-signal-regulated kinase; MLCK, myosin light chain kinase; PH, partial hepatectomy; RhoK, Rho kinase.

Materials and Methods


Sprague-Dawley male rats (150-200 g) were from Janvier (Saint Genest, France). Animals were given food and water ad libitum, and experiments were carried out in accordance with French laws and regulations.

Cell Cultures.

Hepatocytes were isolated from rat livers by a two-step perfusion procedure using 0.025% collagenase (Boehringer-Ingelheim, Gagny, France) buffered with 0.1 mol/L Hepes (pH 7.4) as previously described.32 ML7 (dissolved in DMSO) or Y27632 (dissolved in water) were added at the defined concentrations 1 hour before growth stimulation.


[α–32P]dCTP (3,000 Ci/mmol) and [methyl-3H] thymidine (5 Ci/mmol) were from Amersham Corporation (Buckinghamshire, England); ML7, Y27632, 14–22 amide (476485) and Ro318220 (557520) were from Calbiochem (La Jolla, CA).

Transfection Experiments.

Hepatocytes were transfected 24 hours after seeding with RhoK dominant-negative vector (a gift of K. Kaibuchi) or with cyclin D1 expression plasmid (pcDNA3V-Cyclin D1) using Lipofectamine 2000 (Invitrogen; Cergy Pontoise, France), as described by the manufacturers, and stimulated with EGF 24 hours later.

Immunoblotting Analysis.

Anti-cyclin D1 antibody was from Neomarkers (Westinghouse, CA). Anti-cdk1 is a polyclonal anti-serum specifically directed against the C-terminal part of human p34.6 Polyclonal antibodies directed against ERK1 (sc-94), ERK2 (sc-153), cyclin A2 (sc-596), cyclin E (sc-481), c-jun (sc-45), and c-fos (sc-52) were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-MLCK antibody was from Sigma (St. Quentin Fallavier, France). Antibody directed against p70S6K phosphorylated on residue Thr 389 (#9205), anti-p70S6K (#9430) and anti-phospho ERK 1/2 (#9106) were from Cell Signaling (Beverly, MA). Anti-PCNA antibody was from Dako (Trappes, France). Dominant-negative RhoK coupled with a myc-tag was detected using an anti-myc horseradish peroxidase antibody (Roche, Basel, Switzerland). Quantifications were performed from three independent experiments with Quantity One software (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 and precipitation with ice-cold trichloroacetic acid (5%).

Fluorescence Microscopy.

Hepatocytes were fixed in 4% formaldehyde and permeabilized. Cells were sequentially incubated with either anti-cyclin D1 or anti- phosphorylated MLC (Thr18/Ser19) antibody (Cell Signaling) for 1 hour at room temperature (10 μg/mL final) and secondary antibody coupled to FITC (Santa Cruz Biotechnology) for 1 hour at room temperature (1:50 dilution). Nuclei were stained with Hoechst dye for 10 minutes. Preparations were mounted and examined with an upright fluorescence microscope (Leica; DMRXA2 model) equipped with a 63× objective (1.32 Pl Apo/PH3). Images were captured with a Photometrics CoolSNAP HQ cooled-camera (Roper Scientific, Evry, France).

Northern Blotting and Quantitative RT-PCR.

Total RNA was extracted from hepatocytes at the indicated times using the SV total RNA isolation system (Promega; Charbonnières, France), and Northern blotting was performed as previously described.5

All quantitative RT-PCR assays were performed in triplicate, with standard dilution curves, using qPCR kit for SybrGreen1 (Eurogentec, Belgium) on ABI prism 7000 (PE-Biosystems). Expression data were normalized with endogenous GAPDH and compared with their respective controls. Primers for cyclin D1 and GAPDH were (Gene: Forward primer, Reverse primer): cyclin D1: 5′-GCACTTTCTTTCCAGAGTCATCAA-3′, 5′-CGATGTTCTGCTGGGCCT-3′; GAPDH: 5′-TGCCAAGTATGATGACATCAAGAAG-3′, 5′-TAGCCCAGGATGCCCTTTAGT-3′.

RNA Interference.

Two siRNAs directed against mouse/rat MLCK (da: AUUCAGCAUCAGGACCCAC, db: GUGGAACAGUGCUGGACAA), one against mouse/rat ERK2 (GUGCUGUGUCUUCAAGAGC), and one specific for p70S6K (GCAGAUGGAUGUGACAACG) were designed. 24 hours after plating, siRNAs were introduced into hepatocytes using Lipofectamine 2000 (Invitrogen) in OptiMEM (Invitrogen) with 5% fetal calf serum for 4 hours.


Involvement of MLCK and RhoK in Cyclin D1 Regulation.

First, we investigated the role of MLCK and RhoK inhibition in late G1 phase progression by examining cyclin D1 expression by immunofluorescent staining. In Fig. 1A, we show that cyclin D1, a marker of late G1,6, 33, 34 accumulated in the nucleus 24 hours after EGF stimulation. The importance of MLCK and RhoK in this process was investigated using specific inhibitors, ML7 and Y27632, respectively. Cyclin D1 accumulation was totally absent in ML7-treated cells, whereas the protein was normally detected after Y27632 treatment (Fig. 1A).

Figure 1.

Role of MLCK and RhoK in cyclin D1 regulation and cell shape. After enzymatic dissociation of liver, hepatocytes in primary culture progress in G1 phase independently of growth factor and block at a mitogen-dependent restriction point (RP) located 48 hours after seeding. At this time, hepatocytes were stimulated with epidermal growth factor (EGF). Cyclin D1 (A) and phosphorylated MLC (P-MLC) (B) staining 24 hours after hepatocyte stimulation by EGF in the presence or absence of Y27632 10 μmol/L or ML7 20 μmol/L. Bar, 20 μm.

To confirm specificity of RhoK and MLCK inhibitors, we looked at cell shape and localization of MLC phosphorylated on residue Thr18/Ser19 (P-MLC) in the presence of ML7 or Y27632. In unstimulated hepatocytes, P-MLC was diffused within the cytoplasm (Fig. 1B). In contrast, after growth factor stimulation, hepatocytes extended their cytoplasm, and P-MLC accumulated at the leading edge of the extension. In the presence of RhoK inhibitor, hepatocytes appeared flat and P-MLC accumulated at the cell periphery. In the presence of ML7, extensions induced by EGF were completely abrogated and P-MLC localization at the leading edge was abolished.

RhoK Does Not Interfere With Replication.

We determined the role of RhoK in hepatocyte cell cycle progression by analyzing DNA replication in RhoK-inhibited cells. DNA synthesis was quantified using [3H]thymidine incorporation, in the presence or absence of Y27632 (Fig. 2A). A concentration–response experiment with Y27632 was performed (left panel): whatever the concentration used, DNA replication occurred in the presence of RhoK inhibitor. At 50 μmol/L and 100 μmol/L, DNA synthesis was even higher than in EGF-stimulated cells. In the presence of 10 μmol/L Y27632 (right panel), replication induced by growth factor followed the same time-course as hepatocytes stimulated by EGF alone.

Figure 2.

RhoK is not involved in cell cycle progression. (A) Hepatocyte progression up to S phase was determined using [3H] methyl-thymidine incorporation in the presence of increasing concentrations of Y27632 (10, 50, 100 μmol/L) (left panel). Results were expressed as a percentage of control [3H] thymidine incorporation after EGF stimulation. A time-course was recorded using 10 μmol/L Y27632 at the indicated times up to 60 hours after EGF stimulation (right panel). (B-C) Twenty-four hours after seeding, hepatocytes were transfected with a RhoK dominant–negative vector and stimulated 24 hours later with EGF. RhoK, cdk1, and ERK1/2 protein expressions (B) and DNA replication (C) were analyzed at the indicated times after stimulation. (D) Western blot analysis of cyclin D1 expression at the indicated times after EGF stimulation in the presence or absence of Y27632 (10 μmol/L). A mix of ERK1 and ERK2 antibodies was used as loading control.

We strengthened these results by showing that a dominant-negative RhoK (DN RhoK) did not interfere with S phase entry (Fig. 2B), as evidenced by expression of cdk1, a marker of S phase progression.32, 35, 36 In addition, DN RhoK expression, visualized by its myc tag, did not influence replication (Fig. 2C). DNA replication was even quite high between 24 and 48 hours, as already observed in cells treated with Y27632. The two approaches provided evidence that RhoK is not involved in S phase progression of stimulated hepatocytes.

In hepatocytes as in other cells, upregulation of cyclin D1 has been related to late G1 progression. In 3T3 immortalized fibroblasts, Roovers and Assoian showed that RhoK inhibition led to a more rapid progression through G1 phase.28 To test the hypothesis of an accelerated G1 phase in RhoK-inhibited hepatocytes, we determined the time-course of cyclin D1 induction in cells treated with Y27632 (Fig. 2D). Western-blot analysis highlighted only a 2-hour faster induction in cyclin D1 expression, in comparison with hepatocytes treated with EGF alone. This result suggests that progression in late G1 could be only 2 hours faster in the presence of RhoK inhibitor.

MLCK Is Involved in Cell Cycle Progression.

We next determined the role of MLCK in the hepatocyte cell cycle. We first found that hepatocytes did not progress in S phase in the presence of MLCK inhibitor: 5 μmol/L and 10 μmol/L inhibited DNA replication by 40% and 60%, respectively, and 20 μmol/L decreased replication by 100% (Fig. 3A, left panel). At this optimal concentration (20 μmol/L), ML7 blocked DNA replication through the time-course analyzed (right panel), showing that DNA replication was totally blocked but not delayed by MLCK inhibition. Unlike with Y27632, results later than 48 hours after stimulation were not analyzed because MLCK inhibition led to apoptosis commitment at 60 hours (results not shown). At concentrations higher than those used in this study, ML7 could weakly inhibit PKA and PKC. We therefore performed control experiments in the presence of the chemical inhibitors PKA (14–22 amide) or PKC (Ro318220), at 20 nmol/L and 40 nmol/L, respectively. In both cases, hepatocyte proliferation was unaltered (unpublished data).

Figure 3.

MLCK regulates hepatocyte progression up to S phase. (A) Concentration-dependent inhibition of DNA replication by ML7. Hepatocytes were stimulated with EGF in the presence of increasing concentrations of ML7 (from 5 to 20 μmol/L) and [3H] methyl-thymidine incorporation was analyzed between 24 and 36 hours (left panel). Results were expressed as a percentage of control [3H] thymidine incorporation after EGF stimulation. On the right panel, time-course of DNA replication in the presence of ML7 (20 μmol/L), at the indicated times up to 48 hours. (B-D) Twenty-four-hours-old hepatocytes were transfected for 4 hours with either MLCK-specific (da and db) or control siRNA and stimulated with EGF 24 hours later. MLCK (B), cdk1, and cyclin E (C) expression were analyzed by immunoblotting, 24 and 48 hours, respectively, after stimulation. (D) DNA synthesis was determined in hepatocytes transfected with MLCK or control siRNA. Results are expressed as percentage of control (EGF + siRNA control = 100%).

We confirmed the involvement of MLCK in hepatocyte proliferation using RNA interference inhibition. With an optimized protocol, 80% to 90% of the hepatocytes were transfected by MLCK siRNA (duplex da or db) and stimulated by EGF. Results showed that both duplexes decreased MLCK expression by 70% (Fig. 3B), as compared with cells transfected with control siRNA. Loading control was not affected by the experimental procedure. Interestingly, in these conditions, MLCK inhibition blocked the expression of cdk1 and cyclin E by 60% to 70% (Fig. 3C).

To provide further evidence that MLCK is involved in the regulation of hepatocyte proliferation, we analyzed DNA replication in cells transfected with siRNA specific for MLCK (Fig. 3D). Results showed a decrease in DNA replication (75% and 50% with duplex da and db, respectively) when MLCK expression was inhibited with siRNA.

Thus we demonstrated a good correlation between expressions of cdk1, cyclin E, DNA replication, and MLCK inhibition by ML7 or siRNA, confirming the role of the kinase in S phase entry of hepatocytes.

Cyclin D1 Expression Is Mediated by an MLCK-Dependent Process.

As MLCK inhibition blocked hepatocyte entry in S phase, we wondered at which step, in late G1, MLCK controlled hepatocyte progression. We therefore analyzed cyclin D1 expression in MLCK-inhibited hepatocytes. We confirmed, first by Western blot analysis, that cyclin D1 expression was blocked when cells were treated with ML7 (Fig. 4A). In addition, we showed that cyclin E, cyclin A2 and cdk1 were downregulated in MLCK-inhibited cells, showing that MLCK inhibition blocked hepatocyte commitment to S phase.

Figure 4.

Cyclin D1 expression depends on MLCK. (A) Total cell lysates were collected 24 and 48 hours after stimulation, in the presence or absence of 20 μmol/L ML7, and immunoblotted using anti-cyclin D1, -cyclin E, -cyclin A2, -cdk1, and -ERK1/2 antibodies. (B-C) At the indicated times, cyclin D1 mRNA was analyzed by Northern blot (B) and quantitative RT-PCR (C; unstimulated cells at 24 hours were used as reference, fixed arbitrarily at 1). (D) Western blot analysis of c-jun and c-fos expressions in stimulated hepatocytes treated or not with MLCK inhibitor. (E, F) Forty-eight-hour-old hepatocytes were transfected with pcDNA cyclin D1 expression vector and stimulated with EGF, in the presence or absence of 20 μmol/L ML7. Twenty-four and 48 hours later, cyclin D1 and ERK1/2 were analyzed by Western blot (E), and DNA replication was determined by [3H] thymidine incorporation (F). EGF, epidermal growth factor.

In addition, cyclin D1 has been shown to be regulated at the mRNA and protein level in many cell types including hepatocytes. We therefore investigated the effect of ML7 on mRNA by Northern blot and quantitative RT-PCR (respectively Figs. 4B-C). As already observed with protein, cyclin D1 mRNA expression clearly increased after growth factor stimulation, whereas its expression was inhibited in the presence of ML7. These results were confirmed by real-time RT-PCR amplification: cyclin D1 mRNA expression diminished by 70% after ML7 treatment. This decrease was correlated with the lack of induction of immediate early genes c-Jun and c-Fos as early as 2 hours after ML7 addition to hepatocytes (Fig. 4D).

Cyclin D1 has been shown to be an important regulator of late G1 phase progression. To specify the interactions between MLCK, cyclin D1 expression, and late G1 phase progression, we rescued cyclin D1 inhibition by expressing exogenous cyclin D1 in MLCK-inhibited cells. In these conditions, cyclin D1 expression was analyzed 24 and 48 hours after transfection (Fig. 4E). Our results showed that expression of exogenous cyclin D1 in ML7-treated cells was not able to rescue DNA replication (Fig. 4F). Thus, in our experimental conditions, cyclin D1 alone did not allow S phase entry when MLCK was inhibited. Furthermore, in hepatocytes treated with the MAPK inhibitor U0126, another condition of G1/S arrest, overexpression of cyclin D1 was not able to induce progression in S phase (result not shown).

MLCK Is Required for ERK2-Dependent p70S6K Activation.

Recently, MLCK and RhoK have been implicated in sustained ERK activation.28 Moreover, in hepatocytes, MEK/ERK and PI3K mediate growth factor-related response and cyclin D1 regulation.8 First, we demonstrated that MLCK and RhoK were not involved in ERK activation in hepatocytes, in contrast to the results of Roovers and Assoian in 3T3 fibroblasts. No differences in the phosphorylation status of ERK1/2 were detected in the presence of ML7 or Y27632 (up to 24 hours; data not shown), showing that MLCK and RhoK were not involved in ERK phosphorylation induced by the growth factor (Fig. 5A).

Figure 5.

MLCK regulates ERK-dependent p70S6K phosphorylation. (A-B) Forty-eight-hour-old hepatocytes were pre-treated with RhoK inhibitor (10 μmol/L Y27632) or MLCK inhibitor (20 μmol/L ML7). Phosphorylations of ERK1/2 (A) and p70S6K (B) were investigated by immunoblotting, 0.5 and 2 hours after EGF stimulation. (C,E) Twenty-four-hour-old hepatocytes were transfected for 4 hours with either MLCK siRNA (da or db) (C) or ERK2 siRNA (E) and then stimulated with growth factor 48 hours later. Cell lysates were collected 30 minutes after stimulation and analyzed by Western blot against MLCK and phosphorylated p70S6K (Thr389) (C) or ERK1/2 and phosphorylated p70S6K (E). (D,F) Quantification of p70S6K phosphorylation was performed on three independent blots and averaged.

Interestingly, we found that the two kinases differentially regulated p70S6K phosphorylation. MLCK but not RhoK controlled p70S6K activation by EGF (Fig. 5B): MLCK inhibition completely abolished EGF-induced p70S6K phosphorylation on residue Thr389. We strengthened this result with siRNA specific for MLCK (Fig. 5C): in transfected hepatocytes, MLCK expression decreased by 54% and 60% with siRNA da and db respectively, in parallel with a decrease of 45% and 60% in p70S6K phosphorylation on residue Thr389 (Fig. 5D). These results led us to conclude that MLCK is involved in p70S6K phosphorylation induced by EGF.

Interestingly, p70S6K has been implicated in the regulation of hepatocyte proliferation and ERK has been reported to phosphorylate p70S6K in hepatocytes and cardiac muscle cells.8, 37 To investigate this signaling pathway further, we chose to inhibit ERK2 expression by RNA interference and then to analyze p70S6K phosphorylation (Fig. 5E). First, we obtained high knockdown specificity for ERK2 by this experimental approach: ERK2 expression was abolished by 80% (±15), whereas ERK1 level was not affected by siRNA treatment. In these conditions, inhibition of ERK2 reduced p70S6K phosphorylation by 78% (±8) (Fig. 5F), showing that p70S6K was located downstream of ERK2. Taken together, these results indicated an MLCK-dependent p70S6K activation that occurred downstream of ERK2 in hepatocytes.

p70S6K Is Essential for Hepatocyte Entry in S Phase.

In hepatocytes and in other models, inhibition of FRAP/mTOR was found to prevent DNA replication. To specify the involvement of p70S6K in S phase entry, we transfected hepatocytes with siRNA directed against p70S6K and analyzed DNA synthesis. Western blot showed a decrease of 85% in p70S6K expression in the presence of siRNA, contrary to cells treated with control-siRNA (Fig. 6A-B). In these conditions, in the presence of p70S6K-siRNA, DNA synthesis decreased by 45% and 60%, 24 to 48 hours and 48 to 72 hours, respectively, after growth factor stimulation (Fig. 6C). Thus, all these results demonstrated a link between MLCK-dependent regulation of p70S6K and cell cycle progression.

Figure 6.

p70S6K is essential for S phase entry. Twenty-four-hour-old hepatocytes were transfected for 4 hours with p70S6K- or control- siRNA and stimulated with EGF 24 hours later. (A) Western blot analysis of p70S6K and ERK1/2 expression 48 hours after EGF stimulation (72 hours after transfection). (B) p70S6K expression was quantified from three independent experiments and averaged. (C) [3H] methyl-thymidine incorporation was analyzed at the indicated times after EGF stimulation.

An MLCK-Dependent Window in Late G1 Phase.

Reversion experiments were performed to see whether MLCK-inhibited cells remained blocked at the restriction point or progressed in late G1 phase independently of MLCK inhibition. Hepatocytes were stimulated by EGF and treated or not with ML7 for 24 hours. The inhibitor was then removed and replication analyzed in the presence of the growth factor. In control experiments (EGF alone), DNA replication occurred classically 24 hours after growth factor stimulation. Hepatocytes treated with ML7 for 24 hours were able to undergo DNA synthesis 30 hours after stimulation, corresponding to an only 6-hour delay compared with control cells stimulated in the absence of ML7 (EGF alone) (Fig. 7A). Moreover, cyclin D1 protein in expression (Fig. 7B) was induced only 2 hours after ML7 removal, whereas its expression appeared long after GF stimulation (8 hours) in control cells (see Fig. 2D). These experiments strongly argue that cells continued to progress in late G1 phase even in the presence of ML7, and that MLCK could control the G1/S transition.

Figure 7.

MLCK inhibition blocks G1/S transition but not S phase progression. (A, B) DNA replication and cyclin D1 expression after ML7 treatment and drug removal: forty-eight-hour-old hepatocytes were cultured in basal conditions or stimulated with EGF in the presence or absence of ML7 (20 μmol/L). Inhibitor was removed 24 hours later, and hepatocytes were cultured in the presence of EGF alone. As control, cells were continuously treated with ML7. DNA replication was expressed as a percentage of control [3H] thymidine incorporation after EGF stimulation. (A) [3H] thymidine incorporation into DNA. (B) cyclin D1, cdk1, and ERK1/2 protein expression was analyzed before and after drug removal (R) at the indicated times. (C) Time-course of [3H] thymidine incorporation into DNA over two different periods of ML7 treatment after EGF stimulation: ML7 was added 24 hours (late G1) or 30 hours (S phase) after EGF stimulation. (D) Cyclin D1 expression was determined at the indicated times by Western blotting following addition of ML7 at 24 h (upper panel) or 30 hours (lower panel) after EGF stimulation.

To confirm the importance of MLCK in controlling the G1/S transition and to define a possible MLCK-dependent window in late G1 phase, we designed the following experiment: 48-hour-old hepatocytes were stimulated with growth factor and treated with ML7 at different times after stimulation either before (18 and 24 hours) or after (30 and 36 hours) G1/S transition (Fig. 7C and unpublished data). Interestingly, ML7 addition a few hours before the G1/S transition (18 and 24 hours) was sufficient to block progression to S phase, confirming the key role of MLCK before S phase entry. Surprisingly, cells inhibited within S phase (30 hours after GF stimulation) replicated DNA in the presence of MLCK inhibitor. Thus, we demonstrated that hepatocytes in S phase progressed further, independently of MLCK activation. Moreover, Western blot analysis of cyclin D1 expression (Fig. 7D) indicated that addition of ML7 in late G1 phase (24 hours) resulted in a rapid downregulation of cyclin D1 and blocked S phase entry. Conversely, ML7 added in S phase (30 hours) was not able to inhibit both cyclin D1 expression and DNA replication.

Thus, MLCK dynamically controlled cyclin D1 expression in the 6 hours preceding S phase entry, allowing us to define an MLCK checkpoint of the G1/S transition in proliferating hepatocytes.

Localization of an MLCK-Dependent Checkpoint in Regenerating Liver.

Liver regeneration triggered by partial hepatectomy (PH) is a well-established model in dissecting hepatocyte growth control mechanisms in vivo.

To determine whether MLCK activation was associated with a dependent window in late G1 in vivo, as observed in hepatocytes stimulated in vitro, we designed ex vivo experiments: hepatocytes were isolated from regenerating liver 11 and 16 hours after PH, corresponding to cells in late G1 and in S phase, respectively. Cells were then seeded in the presence or absence of either MLCK or RhoK inhibitor, and DNA synthesis was analyzed between 36 and 60 hours after seeding. We have already demonstrated that 11 hours after PH, hepatocytes cross a growth factor restriction point in vivo and are programmed to progress in S phase without external growth factor stimulation in vitro.5 First, we confirmed this growth factor independence 11 hours after PH, because hepatocytes proliferated in vitro in the absence of growth factor. Second, we extended the in vitro results to the ex vivo model: in the presence of ML7, hepatocyte replication was 60% abolished when the drug was added to hepatocytes isolated from regenerating liver 11 hours after PH (late G1) (Fig. 8A). Interestingly, MLCK inhibitor did not block DNA synthesis with hepatocytes isolated from regenerating liver 16 hours after PH (S phase) (Fig. 8.B). In accordance with in vitro experiments, RhoK inhibitor had no effect on DNA replication whatever the position of hepatocytes in the cell cycle (Fig. 8C–D). These results demonstrated that ex vivo, hepatocytes in S phase (16 hours) had crossed an MLCK-dependent checkpoint.

Figure 8.

MLCK controls G1/S transition in ex vivo experiments. DNA replication of hepatocytes isolated 11 hours (late G1) or 16 hours (S phase) after partial hepatectomy (PH) and cultured in the presence or absence of ML7 (20 μmol/L) (A–B) or Y27632 (10 μmol/L) (C–D). [3H] thymidine incorporation is expressed as a percentage of control (arbitrarily set at 100%). As control of in vivo cell cycle progression, cyclin D1 and cdk1 expressions were analyzed immediately after hepatocyte isolation.


In this study, we demonstrate that MLCK controls the G1/S transition and regulates the expression of cyclin D1, a major player of late G1 progression. Our results underline an MLCK-dependent restriction point in G1/S transition, occurring downstream of ERK2, through the regulation of p70S6K activation.

Here, we clearly identify an MLCK dependency in late G1 phase, whereas the kinase no longer plays a role in DNA replication later on in the S phase. We strengthened this result by ex vivo experiments: hepatocytes primed in vivo, after partial hepatectomy, showed the same MLCK dependency for entry in S phase when placed in vitro, demonstrating that MLCK has a key role in late G1 phase but not in S phase of regenerating liver. RhoK inhibition by chemical inhibitor or RhoK dominant-negative expression vector did not block DNA replication and cyclin D1 expression. Conversely, Hansen and Albrecht reported that cyclin D1 expression was inhibited by Rho inhibitor,38 suggesting that downstream targets of Rho other than RhoK could play a role in cyclin D1 regulation, probably through actin structure disorganization. Because of correlation observed between cyclin D1 and cell morphology, cyclin D1 expression and hepatocyte replication were hypothesized to occur through an actin-dependent mechanism.4, 6, 33 Here, we found that MLCK inhibition prevents cyclin D1 expression at the mRNA and protein levels. This process could occur independently of actin because actin cytoskeleton breakdown inhibits ERK phosphorylation in hepatocytes,31 whereas ML7 treatment did not in our study. In 3T3 cells, RhoK has also been reported to drive late G1 progression, leading to a more rapid progression in S phase in RhoK-inhibited cells in parallel with strikingly early expression of cyclin D1.28 In hepatocytes, RhoK inhibition led to a 2-hour faster expression of cyclin D1, suggesting that the progression in late G1 could be only 2 hours faster in the presence of RhoK inhibitor.

To enhance understanding of the role of MLCK in hepatocyte proliferation, we investigated the involvement of ERK and p70S6K in this process. MLCK was found to be downstream of ERK in Cos-7,24 leukocytes25 and MCF-7.26 Interestingly, our results show that an inhibition of MLCK or RhoK did not block ERK activation, whereas p70S6K phosphorylation was MLCK-dependent but not regulated by RhoK. The consequences for cell proliferation are dramatically different because inhibition of MLCK abolishes G1/S phase transition, whereas RhoK inhibition does not. In mouse myoblasts, inhibition of MLCK and RhoK led to cell cycle arrest uncoupled from differentiation. In this context, RhoK and MLCK play different roles in the myogenic program: signals regulated by MLCK are critical, because inhibition of MLCK suppressed MyoD expression, whereas inhibition of RhoK did not.39 However, in 3T3 cells, either MLCK or RhoK inhibition leads to an absence of ERK phosphorylation, and RhoK-inhibited cells proliferate in the absence of ERK activation, via an LIMK-dependent cascade40, 41 and the Rac/CDc42 pathway.42, 43

In this study, we show that in stimulated hepatocytes, p70S6K activation was partially to totally abolished in the presence of MLCK inhibitor or siRNA. This regulation occurred shortly after growth factor stimulation (30 minutes) and before any cytoskeleton defects. A lower phosphorylation of p70S6K could contribute to the decline in the proliferative response of hepatocytes.44, 45 We demonstrated that downregulation of p70S6K expression with siRNA significantly inhibits DNA synthesis. Furthermore, inhibition of ERK2 by siRNA decreased p70S6K phosphorylation, showing that p70S6K was located downstream of ERK2 in hepatocytes. Thus, p70S6K plays a key role in hepatocyte proliferation and, for the first time, we present data showing that MLCK regulates S phase entry by controlling ERK2-dependent phosphorylation of p70S6K. A possible mechanism by which MLCK regulates p70S6K could involve the formation of a kinase complex at the leading edge of the plasma membrane. MLCK and translocation of active ERK to newly forming focal adhesions may direct specificity toward appropriate downstream targets. In addition, activated p70S6K has been described to be part of a plasma membrane complex shown to bind to a subset of actin filaments involved in Rac-mediated signaling.46

In the presence of p70S6K-siRNA, we never found a decrease in cyclin D1 expression (result not shown), suggesting that MLCK could regulate hepatocyte progression in late G1 phase in two distinct ways: a mechanism involving p70S6K but independent of cyclin D1 regulation, and a cyclin D1–dependent mechanism probably regulated at the level of early gene expression, as suggested by c-jun and c-fos downregulation in the presence of MLCK inhibitor (Fig. 4.D). We found also that exogenous cyclin D1 does not overcome G1 phase arrest induced by ML7, indicating that G1 arrest is not strictly due to the loss of cyclin D1 alone, in our experimental conditions. In contrast, an overexpression of cyclin D1 has been reported to be sufficient to allow hepatocyte progression in S phase in vitro47 and in vivo.48 In hepatocytes seeded on fibronectin, Bhadriraju and Hansen (2004) demonstrated that exogenous expression of cyclin D1 was sufficient to overcome DNA synthesis inhibition by MLCK inhibitor ML9. Discrepancies between results could be linked to experimental conditions: ERK phosphorylation was completely inhibited in the presence of ML9 on high-density fibronectin, whereas ERK signaling was never blocked by inhibition of MLCK in our experiments.

Concerning the role of p70S6K in late G1 progression, the failure to progress through the cell cycle in S6-deficient liver, after partial hepatectomy, was linked to a block in cyclin E mRNA expression.49 Moreover, the impairment of proliferation did not seem to result from a lack of translational capacity, because there was no difference in the abundance or rate of accumulation of p21 or cyclin D1 in p70S6K-deficient versus wild-type liver cells. Consistent with these data, wild-type and constitutively active alleles of p70S6K were sufficient to activate cyclin E promoter in 3T3 fibroblasts.50

In summary, this study brings insights into cross-talk between MLCK activation and G1/S phase transition. An MLCK-dependent p70S6K regulation occurs after growth factor stimulation. This activation is located downstream of ERK2 in a narrow window in late G1 phase, whereas in S phase cell cycle progression becomes MLCK-independent. A precise location in the cell cycle appears determinant for the regulation of the ERK pathway because sequential checkpoints in early G1,4 late G1,6, 34 and G1/S transition (this study) control hepatocyte progression, making them permissive for DNA replication. We are now studying this process during liver regeneration, in relation to MLCK activation/inactivation by chemical inhibitor or RNA interference in vivo.


We want to thank S. Dutertre for imaging (IFR 140, plate forme de microscopie, Université de Rennes1) and C. Ribault for technical assistance. We thank Dr. K. Kaibuchi for giving us the RhoK negative vector.