Induction of Gas6 protein in CCl4-induced rat liver injury and anti-apoptotic effect on hepatic stellate cells


  • Potential conflict of interest: Nothing to report.


The protein product of the growth arrest–specific gene 6 (Gas6) is a secreted ligand for tyrosine kinase receptors, among which Axl is the most widely distributed and displays the highest affinity for Gas6. The Gas6/Axl signaling pathway has been increasingly implicated in growth and survival processes occurring during development and tissue repair. In liver, after an acute or chronic injury, repair involves macrophages and hepatic stellate cells (HSC) activated into myofibroblastic cells (HSC/MFB), which produce cytokines and matrix proteins. We investigated the expression and the role of Gas6 and its receptor Axl in liver repair. Three days after CCl4-induced liver injury in the rat, we detected the expression of Gas6 in ED1-positive macrophages as well as in desmin-positive HSC, which accumulated in injured areas. Axl, the high-affinity receptor for Gas6, was detected in macrophages, HSC, and HSC/MFB. In vitro, expression of γ-carboxylated Gas6 was strongly induced in HSC along with their transformation into myofibroblasts, and it exerted an anti-apoptotic effect on both HSC and HSC/MFB mediated by the Axl/PI3-kinase/Akt pathway. In conclusion, Gas6 is a survival factor for these cells and we suggest that Gas6, secreted by macrophages and HSC/MFB in vivo after liver injury, promotes HSC and HSC/MFB survival and might support transient HSC/MFB accumulation during liver healing. (HEPATOLOGY 2006;44:228-239.)

The growth arrest–specific gene 6 protein (Gas6) is a protein related to protein S, which is γ-carboxylated on glutamic acid residues of its N-terminal sequence.1 Gas6 was identified as the ligand for three structurally related tyrosine kinase receptors (Axl, Mer, and Sky), characterized by an extra-cellular domain with motifs similar to those found in cell adhesion molecules.2 Binding of Gas6 to its receptors is dependent on a vitamin K–dependent NH2-terminal γ-carboxylation. The Axl receptor3 (also called ARK in mouse and Tyro7/UFO in rat) is the most widely distributed and displays the highest affinity for Gas6; c-Mer (Eyk, Nyk, Tyro 12)4 has a lower affinity for Gas6, and Sky (Rse/Tyro3)5 is preferentially expressed in neurons. Gas6 supports haematopoietic stem cell growth6 and promotes fibroblast and endothelial cell survival.7–9 Gas6/Axl signaling induces accumulation of mesangial cells in kidney fibrosis,10, 11 vascular smooth muscle cells in response to intimal vascular injury,12 and cardiac fibroblasts9 during the wound healing process. Thus, this pathway has been increasingly implicated in growth and survival processes during development, regeneration, and tissue repair.

The liver is an organ with an almost unlimited capacity to regenerate in response to toxic damage or surgical resection. Regeneration generally originates from the division of mature hepatocytes and biliary cells13 or from oval precursor cells when the replicative capacity of remnant hepatocytes is severely impaired.14 Acute CCl4 injury is the most widely used model in rodents to mimic acute human toxic liver injury and regeneration from hepatocytes.15 CCl4 induces centrilobular necrosis, which is rapidly replaced by inflammatory cells and, within a few days, by new hepatocytes issued from the division of mature hepatocytes surrounding the lesion. Accumulation of hepatic stellate cells (HSC) in the injured areas and their transformation into myofibroblasts (HSC/MFB), which produce cytokines and matrix proteins, are important steps in the repair process. This is supported by the defective regeneration observed in Foxf −/+ mice associated with a reduced HSC activation, a diminished induction of type I collagen, and a delayed induction of MCP-1.16 Pathways controlling HSC/MFB transformation and survival are therefore important issues in liver regeneration,17 and their understanding might provide new options in the treatment of hepatocellular insufficiency.

We investigated the expression and the role of Gas6 and of its receptor Axl in liver repair after acute CCl4-induced liver injury. We showed that the increase in the number of activated macrophages and HSC/MFB in liver necrotic areas was associated with an induction of Gas6 expression. In vitro, Gas6 expression was strongly induced in HSC along with their activation into HSC/MFB, and Gas6 exerted an anti-apoptotic effect on these cells. We conclude that Gas6 is a survival factor of HSC and HSC/MFB that might contribute to the accumulation of HSC/MFB involved in liver healing.


Gas6, growth arrest–specific protein-6; CCl4, carbon tetrachloride; HSC, hepatic stellate cells; HSC/MFB, hepatic stellate cell–derived myofibroblasts; MCP-1, monocyte chemoattractant protein-1; α-SMA, alpha-smooth muscle actin; VKORC1, vitamin K epoxide reductase complex subunit 1; 15-d-PGJ2, 15-deoxy-δ12, 14-prostaglandin J2; PCR, polymerase chain reaction.

Materials and Methods

Animal Model of Liver Injury.

Male Wistar rats (R. Janvier Animal Center, Le Genest-ST-Isle, France) received a unique injection of 3 mL/kg body weight diluted CCl4 1/1 in olive oil or the same volume of vehicle. Rats were killed 72 hours after the injection when full activation of HSC occurred.18 Liver was divided into three parts: one part was fixed in buffered formalin for histopathological study, the second was snap-frozen for immunofluorescence studies, and the last part was lysed19 for RNA isolation.20 Animal manipulations were performed according to the recommendations of the French ethical committee and under the supervision of authorized investigators.

Histological Study.

Hematoxylin-eosin staining was performed on 4-μm-thick sections of paraffin-embedded liver and immunohistochemistry on frozen sections (5 μm) treated as previously described.20 Primary antibodies were mouse monoclonal anti-desmin (1:100) (DakoCytomotion), mouse monoclonal anti-alpha-smooth muscle actin (α-SMA) (1:1,000) (Sigma-Aldrich, Saint Quentin Fallavier, France), mouse monoclonal anti-ED1 (1:50) (Serotec, Cergy Saint Christophe, France), goat polyclonal anti-Gas6 (1:10) (R&D Systems, Lille, France) or goat polyclonal anti-Axl (1:10) (R&D Systems). Sections were then incubated for 30 minutes with fluorescein isothiocyanate–conjugated donkey anti-mouse IgG or Cy3-conjugated donkey anti-goat IgG (1:200) (Jackson Immunoresearch, Marseille, France).

Isolation and Primary Culture of Rat Hepatic Stellate Cells.

Nonparenchymal cells were isolated from untreated and CCl4-treated rats, hepatic stellate cells (HSC) from untreated rats according to Vrochides et al.21 Briefly, after perfusion with an isotonic saline solution, the liver was perfused with Gey Balanced Salt Solution (Eurobio, Courtaboeuf, France) containing 0.425 g/L collagenase B (0.572 U/mg), 0.016 g/L DNAse I (Roche Diagnostics, Meylan, France), and 20 mmol/L hydroxy-ethyl-piperazine-ethane sulfonate adjusted to pH 7.4. Dissociated cells were filtrated on a nylon (102 mesh) and hepatocytes pelleted at 50 gAV for 2 minutes at 4°C. The nonparenchymal cells in the supernatant were separated on a 13% Nycodenz gradient at 900 gAVfor 30 minutes. Cells at the top of the gradient were collected and layered under a 50% and a 35% step-Percoll gradient and centrifuged at 900 gAVfor 30 minutes. Purified HSC on the top of the 35% layer were seeded at 4.5 × 104 cells/cm2 in DMEM containing 20% fetal calf serum (FCS). This procedure yielded 8 to 12.106cells; 70% were identified as HSC by vitamin A autofluorescence. One day after plating, cultures contained more than 95% HSC identified by their morphology, and the presence of lipid droplets. HSC were cultured in DMEM supplemented with 16% FCS. Cell morphology was examined under phase contrast microscopy and by immunocytochemistry after fixation in pre-cooled methanol for 7 minutes. Primary antibodies were rabbit anti-glial fibrillary acidic protein (GFAP) (1:100) (DakoCytomotion), mouse monoclonal anti-desmin (1:100) (DakoCytomotion), and mouse monoclonal anti–α-SMA (1:400) (Sigma-Aldrich).

RNA Analysis.

Total RNA was isolated from HSC lysed after 4 or 8 days in culture. Unless stated, cells were cultured in serum-deprived medium 3 days before lysis in Rneasy Mini kit lysis buffer (Qiagen, Courtaboeuf, France). RNA (2 μg) was reverse-transcribed from a random hexamer using a first-strand synthesis kit (Fermentas, Life Sciences). Gas6 and Axl transcript levels were measured by quantitative polymerase chain reaction (PCR) using a LightCycler FastStart DNA Master Plus SYBR green I kit (Roche Applied Science), using external standards and primers as previously described.20 Primers for gamma-glutamyl carboxylase (Ggcx) and vitamin K epoxide reductase complex subunit 1 (VKORC1) were Ggcx forward (TGTGAAAAAGCTGGATGCTG) and reverse (GTCTGGAGGCATCGAAGAAG), which map to positions 720 to 739 and 884 to 904 of the rat sequence (GI: 13929061) and VKORC1 forward (TGTCTGTCGCTGGTTCTCTG) and reverse (CATGTGCTAAGGCAAAGCAA), which map to positions 357 to 376 and 538 to 557 of the rat sequence (GI: 45827740).

Protein Analysis.

HSC were cultured for 1 or 5 days in DMEM with 16% FCS followed by a 3-day culture in serum-free medium. Protein samples from concentrated conditioned medium (100 μg) and cell lysate (60 μg) were size-fractionated on a 10% (Gas6) or 7.5% (Axl) sodium dodecyl sulfate polyacrylamide gel and analyzed by Western blotting as previously described,20 using goat polyclonal anti-Gas6 (sc-1935) (1:500) or anti-Axl (sc-1096) (1:500) antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) in the presence or absence of a fivefold excess of their corresponding blocking peptide (Santa Cruz). Mouse monoclonal anti-Gla antibody (M3B 1:100, Pr. J. Stenflo, Lund University, Sweden) was used to detect γ-carboxylated protein secreted by HSC/MFB either in the absence of vitamin K and presence of 5 μmol/L of warfarin or in the presence of 5 μmol/L vitamin K during the 3 day-serum deprivation. Proteins reacting with the primary antibodies were detected with horseradish peroxidase–conjugated rabbit anti-goat or donkey anti-mouse IgG (1/10,000) using the Enhanced Chemiluminescence System (ECL+, Amersham Biosciences, Orsay, France).

For analysis of Akt, ERK, and p65 nuclear factor-kappaB (NF-κB) phosphorylation, cell lysates were spun down at 14,000 gav for 10 minutes, and proteins (50 μg) were separated on 10% acrylamide gel and analyzed by Western blotting using antibodies directed against phospho-Akt (Ser473), phospho-ERK1/2 (Thr202/Tyr204), phospho-nuclear factor kappaB (NF-κB) p65 (Ser536) (1/1,000) (Cell Signaling, Danvers, MA). Blotted proteins were analyzed as indicated.

Induction of Apoptosis.

HSC were plated at 4.5 × 104 cells/cm2 in 35-mm dishes and cultured in DMEM with 16% FCS for one (HSC) or 6 days (HSC/MFB). After washing, apoptosis was induced either by adding 5 or 10 μmol/L 15-deoxy-Δ12, 14-prostaglandin J2 (15-d-PGJ2) (Cayman Chemical, Massy, France), a known inducer of myofibroblast apoptosis,22 for 11 or 16 hours, or by a 3-day serum starvation. Vitamin K (5 μmol/L) (Sigma-Aldrich), mouse recombinant Gas6 (400 ng/mL) (R&D Systems), 10 μmol/L LY294002 (Sigma-Aldrich), or 20 μmol/L PD98059 (Sigma-Aldrich) were added as indicated.

Caspase-3 Assay.

Non-adherent and adherent cells were lysed and caspase-3 activity measured using Ac-DEVD-AFC (BIOMOL Research Labs), as described by Li et al.23 The fluorescence was monitored at 400 nm for 7 hours during the linear range of the reaction.

Statistical Analysis.

Results (mean ± 1 SEM) were analyzed with the Wilcoxon matched pairs signed rank test.


In Vivo Expression of Gas6 and Axl.

Acute CCl4 administration caused massive centrilobular injury, which was still observed 3 days after treatment (Fig. 1A). At this stage, an accumulation of desmin-expressing HSC was noted around the central veins in healing necrotic areas (Fig. 1B). Cells exhibiting a weak alpha smooth muscle actin (α-SMA) signal around strongly labeled central veins demonstrated transformation of HSC into myofibroblasts (Fig. 1C).

Figure 1.

Gas6 and desmin immunolocalization on liver sections. Hematoxylin-eosin staining shows hepatocellular necrosis around the central veins (CV) 3 days after a single CCl4 injection (A). Accumulation of desmin-positive hepatic stellate cells (HSC) in the damaged areas (green fluorescence) (B). Accumulation of α-SMA–positive HSC around CV (C). In control liver, cells expressing Gas6 (red fluorescence) are observed all over the lobule (D). Desmin-positive cells are dispersed in the lobule (E). Merge of desmin and Gas6 signals shows that desmin-positive cells do not express Gas6 (F). In CCl4-treated animals, Gas6 (G) and desmin labeling (H) are concentrated around CV. By merging both signals of the framed areas, Gas6 was localized in desmin-positive cells (yellow signal) (I). Original magnifications are ×100 (A, B), ×400 (C, G, H), and ×400 with a 3× electronic zoom (D, E, F, I). α-SMA, alpha smooth muscle actin; Gas6, growth arrest-specific protein-6.

In control livers, Gas6 immunolabeling was observed all over the lobule (Fig. 1D), as observed for desmin labeling (Fig. 1E). The merge of both signals did not show any colocalization of these two proteins (Fig. 1F), indicating that quiescent HSC do not produce Gas6 in normal liver. In sections from CCl4-treated animals, Gas6 (Fig.1G) and desmin (Fig. 1H) labeling was observed in the damaged area around the central veins, and fusion of both signals showed Gas6 expression in desmin-positive HSC (Fig. 1I). As observed in control livers (Fig. 1F), Gas6 labeling also occurred in desmin-negative cells (Fig. 1I), indicating that HSC are not the only Gas6-producing cells in CCl4- injured liver.

Macrophages also participate in tissue repair15 and might be the other Gas6-producing cells. Macrophages, identified by ED1 staining, appeared mostly at the periphery of the lobule (Fig. 2A). At a higher magnification, ED1 (Fig. 2B) and Gas6 staining (Fig. 2C) displayed a similar distribution. Fusion of both signals showed that resident macrophages are the only cells producing Gas6 in control liver (Fig. 2D). CCl4 treatment induced an increase in the number of centrilobular ED1-positive macrophages (Fig. 2E-F), which accumulated in centrilobular areas and expressed Gas6 (Fig. 2G-H). In summary, in normal liver, Gas6 was only expressed in ED1-positive macrophages (Kupffer cells), and, after an acute CCl4-induced injury, it was detected in HSC/MFB and macrophages that accumulated in damaged areas.

Figure 2.

Gas6 and ED1 immunolocalization on liver sections. In control sections, ED1-positive cells (green fluorescence) are localized at the periphery of the lobule (A, B). Gas6-positive cells (red fluorescence) display an identical distribution (C). Merging ED1 and Gas6 labeling shows a localization of Gas6 in ED1-positive cells as revealed by a yellow signal (D). In liver from CCl4-treated rats, ED1 expression reveals an accumulation of positive macrophages (green fluorescence) in the injured areas around central veins (CV) (E, F). Gas6 labeling is also concentrated around CV (G). Merging ED1 and Gas6 signals framed in F and G yields a yellow signal revealing Gas6 expression in macrophages (H). Original magnifications are ×100 (A, E), ×400 (F, G) and ×400 with a 3× electronic zoom (B, C, D, H).

To identify liver cells that could be targets for Gas6, we studied the expression of the Gas6 receptor Axl. In control livers, cells expressing Axl (Fig. 3A) and desmin (Fig. 3B) were dispersed throughout the lobule. Merging both signals (Fig. 3C) showed Axl expression in HSC as well as in other cells that were identified as macrophages by ED1 and Axl co-localization (Supplementary Fig. 1; supplementary material for this article can be found on the HEPATOLOGY website: Three days after injury, the number of Axl-positive cells dramatically increased in damaged centrilobular areas (Fig. 3D). In these areas, Axl-positive cells corresponded to HSC and macrophages, as shown by desmin (Fig. 3E-F) and ED1 double immunostaining (Supplementary Fig. 1), respectively. Expression of Axl receptor by HSC and macrophages was confirmed by immunocytochemistry on isolated liver nonparenchymal cells. Axl labeling colocalized with desmin (Fig. 3G) or ED1 (Fig. 3H) in cells purified from control rats. In cells isolated from CCl4-treated rats, Axl colocalized with both desmin (not shown) and α-SMA (Fig. 3I) or ED1 (Fig. 3J). Axl expression in α-SMA–positive nonparenchymal cells isolated from acute CCl4-treated livers confirmed the persistence of Axl in HSC that underwent a myofibroblastic transformation. In conclusion, macrophages and quiescent or myofibroblastic HSC express the receptor Axl and appear as putative targets for Gas6, which could exert autocrine or paracrine effects on these cells.

Figure 3.

Axl and desmin immunolocalization. In control liver sections, Axl labeling (red fluorescence) (A) and desmin labeling (green fluorescence) (B). Colocalization of Axl and desmin can be visualized (yellow color) by merging both signals (C). After CCl4 treatment, we note an increase in the number of Axl-positive cells (red fluorescence) (D) and desmin-positive cells (green fluorescence) (E). Comparison of D and E reveals that all desmin-positive cells are also Axl-positive (see arrows). Merge of both signals yielded a faint yellow color that confirms the colocalization (F). Arrowheads in D, E, and F point desmin-negative cells that express Axl. Axl (Ga) and desmin labeling (Gb) colocalized in liver nonparenchymal cells isolated from control rats, as shown by the yellow color in the merged picture (Gc). Localization of Axl (Ha) in ED1-positive cells (Hb) was also observed in merge pictures (Hc). In cells from CCl4-treated animals, Axl expression (Ia) was found in α-SMA–positive cells (Ib) as shown by the yellow merged color (Ic) and also in macrophages as revealed by Axl (Ja) and ED-1 signals (Jb) merged in (Jc). Original magnifications are ×400 (G H, I, J) and ×400 with a 2.5× electronic zoom (A, B, C, D, E, F).

Expression of Gas6 and Axl in HSC Primary Culture.

Gas6 expression was further examined in vitro on isolated HSC cultured. Two days after plating, these cells exhibited typical stellate morphology of quiescent HSC with numerous cytoplasmic extensions and accumulation of intracellular vitamin A droplets (Supplementary Fig. 2A); desmin (Supplementary Fig. 2C) and GFAP (Supplementary Fig. 2E) expression were detected in more than 95% of the cells. After 8 days of culture, the cells remained desmin positive (Supplementary Fig. 2D) and underwent a myofibroblastic transformation characterized by a spindle-shaped morphology, the loss of lipid droplets (Supplementary Fig. 2B), a strong decrease in GFAP staining (not shown), and the expression of α-SMA (Supplementary Fig. 2F).

Gas6 mRNA level was low in cells plated for 4 days, including a 3-day serum starvation before sampling, and it increased up to more than 15-fold over this basal level at day 8 (Fig. 4A). Cells maintained in serum-containing medium exhibited lower Gas6 mRNA levels at day 4 and at day 8, representing 30% and 10% of the values obtained in serum-free medium, respectively (data not shown). In both conditions, transformation of HSC into HSC/MFB strongly induces Gas6 mRNA expression. Western blot analysis of Gas6 secreted during 3 days in serum-free medium (Fig. 4B) showed a single band of 80 kd in medium collected at day 8 that was not detected in medium collected at day 4. This 80-kd band was at the expected size for Gas624 and was efficiently pushed out by an excess of the Gas6 peptide used to raise the antibody. No Gas6 protein was detected in cell lysates (data not shown). These data reveal an induction of Gas6 expression and secretion in HSC along with their myofibroblastic transformation, an expression potentiated by serum deprivation. The ability of HSC/MFB to produce active γ-carboxylated Gas6 was demonstrated by the detection of an 80-kd band using a Gla-domain specific antibody in medium of cells incubated with vitamin K, a cofactor essential for Gas6 carboxylation. This band was not observed when cells were cultured without vitamin K (Fig. 4C). These data were in accordance with the detection in 1-day cultured HSC of Ggcx and VKORC1 mRNA coding for two enzymes required for γ-carboxylation (Fig. 4D). The presence of these two mRNAs was revealed by the detection of the expected 201 bp VKORC1 and the 185 bp Ggcx amplicons on a gel loaded with the 40 cycle amplification products of the real time PCR. The quantitative analysis showed a 6.71 ± 1.71 (Ggcx) and 7.17 ± 1.87 (VKORC1) times higher level of these mRNAs in HSC/MFB than in HSC.

Figure 4.

Gas 6 and Axl expression in cultured HSC. (A) mRNA quantitation: RNA was extracted from HSC cultured for 4 or 8 days (D4 and D8) with a 3-day serum starvation before harvesting. Results are expressed as fold increase over basal value obtained at day 4 (878.7 ± 233.8 Gas6 mRNA copies/ng total RNA and 242.4 ± 29.0 Axl mRNA copies/ng total RNA). Results are expressed as the mean ± SEM of three different RNA preparations. (B) Protein expression: Western blot analysis of Gas6 in conditioned medium and Axl in cell lysate was performed as described in Materials and Methods section at D4 (lane 1) and D8 (lane 2, 3). In lane 3, the antibodies were pre-incubated with an excess of Gas6 or Axl blocking peptide, respectively (typical experiment out of three). (C) Protein γ-carboxylation: γ-carboxylated proteins were analyzed using M3B antibodies in 3-day serum starved conditioned medium of HSC/MFB (D8) with (+VK) or without (-VK) vitamin K. (D) VKORC1 and Ggcx mRNA expression: at the end of the 40 cycle quantitative PCR, amplicons were analyzed on a 2.5% agarose gel and revealed by SYBR green staining. Controls (−) are PCR performed on RNA samples that have not been reverse transcribed and M is a ladder of DNA molecular weight markers. HSC, hepatic stellate cells; PCR, polymerase chain reaction.

Messenger RNA expression of the Gas6 receptor Axl decreased by 45% from day 4 to day 8, in cultured HSC (Fig. 4A). Analysis of Axl protein in HSC lysate indicated two bands of 140 and 120 kd at day 4, which decreased at day 8 (Fig. 4B). Both signals disappeared in the presence of an excess of Axl blocking peptide (Fig. 4B). These two bands correspond to the fully processed and partially glycosylated Axl receptor, respectively.25 The expression of the Gas6 receptor c-mer mRNA was undetectable in quiescent HSC and extremely low after 8 days of culture (less than 10 copies/ng of RNA), and the c-mer protein was undetectable on Western blot (data not shown). In summary, Axl is expressed in HSC, and its expression is down-regulated along with their myofibroblastic transformation. This decrease was observed in absence of vitamin K, necessary to Gas6 binding on Axl, indicating that Axl down-regulation was not mediated by Gas6.

Anti-Apoptotic Effects of Gas6.

One day after plating, HSC apoptosis was induced by 5 μmol/L 15-d-PGJ2, an endogenous prostanoid26 induced in injured liver and known to trigger apoptosis of portal myofibroblasts.23 At this stage of culture, HSC were still in the quiescent phenotype as shown by the accumulation of lipid droplets in their cytoplasm (Fig. 5A). Sixteen hours after addition of 15-d-PGJ2, HSC had reduced cytoplasmic extensions and a rounded refringent shape (Fig. 5C). At 16 hours, most of the cells were apoptotic, displaying a small nucleus, with condensed chromatin after DAPI staining (Fig. 5D) as compared with control cultures (Fig. 5B). These cytoplasmic and nuclear alterations were largely prevented by the addition of exogenous recombinant Gas6 (Fig. 5E-F). Six days after plating, HSC that underwent a transformation into HSC/MFB (Fig. 6Aa) were exposed to 10 μmol/L 15-d-PGJ2 for 16 hours. This treatment induced cell death as shown by a decreased number of adherent cells that were small, round, and highly refringent (Fig. 6Ac). Most of the remaining cells also displayed nuclear condensation (Fig. 6Ad) not observed in control experiments (Fig. 6Ab), and this effect was largely inhibited by adding recombinant Gas6 (Fig. 6Ae-Af). Anti-apoptotic effects of Gas6 were confirmed by monitoring caspase-3 activity, an effector of apoptosis. As shown in Fig. 6B, 15-d-PGJ2 induced a 10-fold increase in caspase-3 activity, which was reduced by 40% in the presence of Gas6.

Figure 5.

Effect of Gas6 on 15-d-PGJ2–induced apoptosis in quiescent HSC. One day after plating, cells were cultured for 16 hours in DMEM with 16% FCS (A, B) or in serum-deprived medium containing 5 μmol/L 15-d-PGJ2 (C to F) without (C, D) or with 400 ng/mL recombinant Gas6 (E, F). Cell morphology was observed under phase contrast microscopy (A, C, E) (×320). Nuclear morphology was visualized under fluorescent microscopy after DAPI staining (B, D, F) (×400 with a 1.5× electronic zoom). Gas6, growth arrest-specific protein-6.

Figure 6.

Effect of Gas6 on 15-d-PGJ2–induced apoptosis in cultured HSC/MFB. Six days after plating, HSC/MFB were cultured in DMEM supplemented with 16% FCS (Aa-Ab) or in serum-free DMEM containing 10 μmol/L 15-d-PGJ2 (Ac to f) with 400 ng/mL recombinant Gas6 (Ae-Af). (A) Cell and nuclear morphology were examined after 16 hours under phase contrast microscopy (a, c, e) (×320) and after DAPI staining under fluorescent microscopy (b, d, f) (×400 with a 1.5× electronic zoom). (B) Caspase-3 activity was measured after 11 hours. Results are the mean ± SEM of eight independent experiments (*P ≤ .05 versus 16% FCS and #P ≤ .05 versus 0% FCS + 15-d-PGJ2). HSC/MFB, hepatic stellate cell-derived myofibroblasts.

To verify whether Gas6 synthesized and secreted by HSC/MFB could exert an autocrine control on HSC/MFB apoptosis, 6 day-plated HSC/MFB were cultured for 3 days in serum deprived conditions with or without vitamin K. Serum deprivation induced cell apoptosis, as shown by the presence of numerous highly refringent rounded cells (Fig. 7Ac) with condensed chromatin (Fig. 7Ad) as compared with cells cultured in presence of serum (Fig. 7Aa-Ab). Serum deprivation-induced apoptosis was largely prevented by the addition of vitamin K in the medium (Fig. 7Ae-Af). Caspase-3 activity showed a significant 2.5-fold increase after serum deprivation (Fig. 7B), an effect that was reduced by 25% in the presence of vitamin K (Fig. 7B). Inhibition of vitamin K protection by anti-Gas6 antibodies was not successful, probably because of an insufficient level of apoptotic protection by secreted Gas6 and of an unexpected inhibitory effect of these antibodies on caspase-3 activity (data not shown).

Figure 7.

Effect of vitamin K on serum deprivation–induced apoptosis in cultured HSC/MFB. Six days after plating, HSC/MFB were cultured for a further 3 days in DMEM containing 16% FCS (Aa, Ab) or in serum-free medium (Ac to Af) without (Ac, Ad) or with 5 μmol/L vitamin K (Ae, Af). (A) Cell and nuclear morphology were examined under phase contrast microscopy (a, c, e) (×100) and after DAPI staining under fluorescent microscopy (b, d, f) (×400). (B) Caspase-3 activity. Results are the mean ± SEM of 11 independent experiments (*P ≤.05 versus 0% FCS and #P ≤.05 versus 0% FCS + vitamin K). HSC/MFB, hepatic stellate cell-derived myofibroblasts; DMEM, Dulbecco's minimum essential medium; FCS, fetal calf serum.

Intracellular Signaling in Gas6 Survival Effects.

The phosphorylation of Akt and ERK, two transduction factors involved in anti-apoptotic processes,27 was analyzed in HSC. Cells cultured in serum displayed a high level of phosphorylated Akt and ERK, which disappeared in serum-free conditions (Fig. 8A). Addition of Gas6 induced Akt and ERK phosphorylation, which was observed within 5 minutes and therefore was independent of protein synthesis. We also noticed a concomitant increase in NF-κB (p65) phosphorylation, a downstream target of Akt (Fig. 8A). Actin signal showed that equal amounts of protein were loaded on the gel. To further establish the role of Akt and ERK pathways in Gas6 survival effect, the LY294002 PI3-kinase inhibitor or the PD98059 MEK inhibitor was added to 15-d-PGJ2–treated cells. The aspect of untreated or 15-d-PGJ2–treated cells (Fig. 8Ba-Bb) was not affected by LY294002 addition (data not shown). However, Gas6 protection against 15-d-PGJ2–induced apoptosis (Fig. 8Bc-Bd) was not observed in presence of LY294002 as shown by a persistent alteration in cell morphology (Fig. 8Be) and nuclei DAPI staining (Fig. 8Bf). In contrast, PD98059 treatment, which slightly potentiated 15-d-PGJ2 effects, did not alter Gas6 survival effect on 15-d-PGJ2–treated cells (Supplementary Fig. 3). Altogether, these data showed that Gas6-induced ERK, Akt, and NF-κB phosphorylation in HSC and that the Gas6 survival effect is Akt-mediated and ERK-independent.

Figure 8.

Gas6 signaling pathway. (A) Phosphorylation of Akt, ERK, and p65 (NF-κB) were assessed on cell lysates by Western blotting using corresponding antibodies. Actin served as a loading control. (B) Effect of LY294002 on Gas6 protection in 15-d-PGJ2–induced apoptosis in cultured HSC/MFB. Six days after plating, HSC/MFB were cultured for a further 16 hours in serum-free DMEM containing 10 μmol/L 15-d-PGJ2 (a to f) with 400 ng/mL recombinant Gas6 (c to f) and 10 μmol/L LY294002 (Be, Bf). Cell morphology was observed under phase contrast microscopy (a, c, e) (×320). Nuclear morphology was visualized under fluorescent microscopy after DAPI staining (b, d, f) (×400 with a 1.5× electronic zoom). HSC/MFB, hepatic stellate cell-derived myofibroblasts.


In this study we investigated expression of Gas6 and of its receptor Axl in the liver and their ability to modulate HSC and HSC/MFB survival during liver repair after acute injury. Three days after CCl4-induced necrosis, macrophages and HSC/MFB accumulated in the damaged areas as previously reported.15, 18 In normal liver, Gas6 is expressed in sinusoidal resident macrophages and not in quiescent perisinusoidal HSC. After CCl4 administration, Gas6 was detected in ED1-positive macrophages (Kupffer and inflammatory macrophages) as well in HSC/MFB. Gas6 expression in HSC/MFB appears to be linked to their myofibroblastic phenotype, as supported by the induction of Gas6 in cultured HSC along with their transformation. Axl, the high-affinity receptor for Gas6, was found in macrophages and in quiescent HSC, where its level decreased after their transformation into HSC/MFB. In summary, these data provide evidence that Gas6 secreted by macrophages and HSC/MFB may exert autocrine and paracrine effects on HSC, HSC/MFB, and macrophages in damaged liver areas.

In search for an anti-apoptotic effect of Gas6 on HSC and HSC/MFB, active recombinant Gas6 was added to these cells exposed to apoptotic stimuli. Gas6 protects HSC and HSC/MFB against apoptosis, as previously reported for other cell types.9, 28 We also show a vitamin K–dependent anti-apoptotic effect on HSC/MFB, which revealed the production of a vitamin K-dependent survival factor by HSC/MFB, consistent with an autocrine effect of γ-carboxylated Gas6 secreted by these cells. Recombinant Gas6 induced the phosphorylation of Akt/protein kinase B, a downstream component of Axl/PI3-kinase pathway,29, 30 and NF-κB, a known inducer of the synthesis of anti-apoptotic proteins such as Mcl-1 or bcl-2.31 Inhibition of Gas6 anti-apoptotic effect on HSC/MFB by the PI3-kinase inhibitor confirmed the role of Akt signaling in Gas6-induced survival effect, and such a result is in accordance with bcl-2 induction in endothelial cells treated with Gas632 and the protective role of NF-κB against apoptosis in HSC/MFB.33 Serum-induced ERK phosphorylation was not sufficient to exert a mitogenic effect on cultured HSC/MFB assessed by cellular counting or by measuring DNA synthesis (data not shown). Independently of Gas6, HSC/MFB had a very poor proliferative capacity in vitro as previously documented,34 even when they were cultured in the presence of serum. In addition, Gas6 was reported as an anti-apoptotic protein without a direct mitogenic effect in other cell types.8

HSC/MFB, the main matrix-producing cells in liver, are highly sensitive to apoptosis in culture35 and in vivo.36 The increase of Fas-Ligand (CD95L) synthesis, the strong decrease in bcl-2 and bcl-xl during the course of HSC activation, as well as the high content of Fas-receptor (CD95) and pro-apoptotic factors in HSC/MFB, have been proposed as mechanisms underlying apoptotic sensitivity of these cells and their rapid elimination after tissue repair.36, 37 The decrease in Axl content that we observed in association with HSC myofibroblastic transformation provides another clue to explain the short half-life of HSC/MFB. Unfortunately, we cannot explore more deeply the Axl decrease in HSC/MFB because of their short half-life in culture after transformation and the parallel rapid outgrowth of desmin-negative fibroblasts in the dish as previously reported.38 Nevertheless, Axl decrease is probably not due to a downregulation by Gas6, because it was also observed in cells cultured in the absence of vitamin K, a situation that prevents binding of Gas6 to its receptor.

After acute liver injury, activation of Axl by Gas6 could provide a survival stimulus for HSC/MFB and macrophages exposed to an increased level of tumor necrosis factor alpha (TNF-α),15-d-PGJ2, and Fas-ligand reported in different models of liver injury.26, 39 Activation of PI3-kinase and Akt also inhibits Fas receptor-mediated cell death in hepatocytes.40, 41 Therefore, under the context of a constitutive Fas receptor expression in HSC/MFB,35 Gas6 could protect these cells and macrophages against apoptosis in vivo during tissue repair. It may favor a transient accumulation of these cytokine- and matrix-producing cells, which are necessary to the reorganization of liver architecture and hepatocyte proliferation. Gas6 involvement in tissue repair has been documented in injured rat carotid where Axl and Gas6 expression was related to neointima formation.12 Another role for Gas6 could be its involvement in clearance of dying cells that is carried out by macrophages and HSC/MFB42 and precedes regeneration. Recognition of dying cells involves surface receptors on phagocytes and extracellular ligands with binding sites on both damaged cells and phagocytes.43 Gas6, one of these ligands, binds to phosphatidylserine exposed at the surface of dying cells through its NH2-terminal Gla domain, and to its receptors on macrophages and HSC through its COOH domain.44 This role of Gas6 was confirmed in mice invalidated for one of the Gas6 receptors, which develop an inability to clear apoptotic cells.45

Pathways controlling HSC transformation into MFB and survival are important issues in tissue repair and fibrosis.46, 47 Gas6 induction in response to acute injury and its positive effect on HSC/MFB survival should promote a transient accumulation of cytokine-and matrix-producing cells and support tissue repair. Interestingly, the progressive decrease in Axl content of HSC/MFB should prevent these cells from a sustained survival effect leading to a fibroproliferative scar.


The authors thank Pr. J. Stenflo (Lund University, Sweden) for the generous gift of M3B antibodies, Pr. J. Foucrier (University Paris XII) and Dr. S. Lotersztajn (Inserm U581) for helpful discussions.