Potential conflict of interest: Nothing to report.
Adiponectin limits the development of liver fibrosis and activates adenosine monophosphate–activated protein kinase (AMPK). AMPK is a sensor of the cellular energy status, but its possible modulation of the fibrogenic properties of hepatic stellate cells (HSCs) has not been established. In this study, we investigated the role of AMPK activation in the biology of activated human HSCs. A time-dependent activation of AMPK was observed in response to a number of stimuli, including globular adiponectin, 5-aminoimidazole-4-carboxamide-1-beta-4-ribofuranoside (AICAR), or metformin. All these compounds significantly inhibited platelet-derived growth factor (PDGF)-stimulated proliferation and migration of human HSCs and reduced the secretion of monocyte chemoattractant protein-1. In addition, AICAR limited the secretion of type I procollagen. Knockdown of AMPK by gene silencing increased the mitogenic effects of PDGF, confirming the negative modulation exerted by this pathway on HSCs. AMPK activation did not reduce PDGF-dependent activation of extracellular signal-regulated kinase (ERK) or Akt at early time points, whereas a marked inhibition was observed 24 hours after addition of PDGF, reflecting a block in cell cycle progression. In contrast, AICAR blocked short-term phosphorylation of ribosomal S6 kinase (p70S6K) and 4E binding protein-1 (4EBP1), 2 downstream effectors of the mammalian target of rapamycin (mTOR) pathway, by PDGF. The ability of interleukin-a (IL-1) to activate nuclear factor kappa B (NF-κB) was also reduced by AICAR. Conclusion: Activation of AMPK negatively modulates the activated phenotype of HSCs. (HEPATOLOGY 2007.)
Nonalcoholic fatty liver disease, the most common chronic liver disease in western countries, is considered the hepatic manifestation of the metabolic syndrome.1 The more severe form, nonalcoholic steatohepatitis, is associated with hepatocyte damage, chronic inflammation, and fibrosis, and may progress to liver cirrhosis and its complications. Despite the clear role of insulin resistance in the progression of fibrosis during nonalcoholic steatohepatitis, the molecular mechanisms involved in these conditions are still unclear. Adiponectin, a hormone secreted by adipose tissue, is closely and directly correlated with insulin sensitivity.2 In humans, plasma adiponectin concentrations markedly decrease in conditions of obesity and type 2 diabetes and are inversely correlated with the presence of the metabolic syndrome.3 In addition, administration of adiponectin ameliorates liver damage during experimental steatohepatitis and reduces the development of fibrosis.4, 5 Adiponectin circulates in different forms, including multimeric, high-molecular-weight complexes, and a truncated form comprising the C-terminal domain, referred to as “globular” adiponectin, which binds both adiponectin receptors and recapitulates the metabolic effects elicited by this molecule.6, 7 One of the downstream effectors of adiponectin is the adenosine monophosphate (AMP)-activated protein kinase (AMPK),8 a fuel-sensing enzyme capable of regulating cellular metabolism in response to different stimuli, such as stress conditions, exercise, and adipocyte-derived hormones.9 Pharmacological agents, such as 5-aminoimidazole-4-carboxamide-1-beta-4-ribofuranoside (AICAR) and metformin, a widely used antidiabetic drug, can also activate AMPK.10, 11 Once activated, AMPK switches on catabolic pathways, leading to adenosine triphosphate generation, and inactivates adenosine triphosphate–consuming processes not essential for short-term cell survival.9
Hepatic stellate cells (HSCs) play a pivotal role in the development of liver fibrosis.12 Stimuli such as liver injury activate and transdifferentiate HSC from vitamin A–storing pericytes to myofibroblast-like cells. Once activated, human HSC become proliferative, proinflammatory and profibrogenic through increased responsiveness to several soluble mediators.12 Previous studies have demonstrated that in rat HSCs, adiponectin inhibits proliferation, migration, and expression of fibrogenic genes, and it may induce apoptosis of activated cells.5, 13 However, little is known regarding the possible role of AMPK activation in the biology of HSCs and its possible relevance to the pathogenesis of liver fibrosis. This study was undertaken to establish the actions of AMPK on the cellular phenotype associated with HSC activation. Our results indicate that AMPK inhibits several profibrogenic actions, including cell proliferation and migration, chemokine secretion, and collagen production, in human HSCs.
Phosphorylation-specific, polyclonal antibodies against AMPK, extracellular signal-regulated kinase (ERK), Akt, p70S6K, 4EBP1, tuberous sclerosis complex 2 (TSC2) and IκBα, and polyclonal antibodies against AMPK were purchased from Cell Signaling Technology (Beverly, MA). Monoclonal antibodies against alpha-smooth muscle actin and beta-actin were from Sigma (St. Louis, MO). Human recombinant adiponectin, human recombinant platelet-derived growth factor (PDGF)-BB, and human recombinant interleukin-1β (IL-1β) were purchased from PeproTech (Rocky Hill, NJ). The AMPK activator AICAR was purchased from Calbiochem (San Diego, CA). Methyl-[3H]thymidine ([3H]TdR) was from New England Nuclear (Milan, Italy). Unless otherwise indicated, all other agents were of analytical grade and were purchased from Sigma.
Isolation and Culture of Human HSCs.
Human HSCs were isolated from wedge sections of normal human liver unsuitable for transplantation, by combined digestion with collagenase and pronase, followed by centrifugation on Stractan gradients. Procedures for cell isolation and characterization have been extensively described elsewhere.14 Cells were cultured on uncoated plastic dishes and used after complete transition toward a myofibroblast-like phenotype. Activated HSCs were subcultured by trypsinization and used between the 3rd and the 7th passages. Analysis of cell toxicity was performed by trypan blue exclusion.
RNA Isolation and Quantitative Reverse-Transcription Polymerase Chain Reaction.
Isolation of total RNA from cultured HSCs was performed using silica membrane filters (Macherey-Nagel, Duren, Germany). Four hundred nanograms RNA from each sample were reverse transcribed with Taqman Reverse Trancriptase Reagents (Applied Biosystems, Foster City, CA). The profile of the 1-step, reverse-transcription reaction was 10 minutes at 25°C, 30 minutes at 48°C, and 5 minutes at 95°C, in a final volume of 80 μL. Twenty-five nanograms complementary DNA for each sample were analyzed in duplicate by quantitative polymerase chain reaction, to measure AdipoR1 and AdipoR2 gene expression, using an ABI 7700 Sequence Detection System (Applied Biosystems) under the following conditions: 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute. 6-carboxy-fluorescein (FAM)-labeled probes and primers for AdipoR1, AdipoR2, and for the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase were purchased as “Assays-On-Demand” (Applied Biosystems) and used as specified by the manufacturer.
DNA synthesis was measured as the amount of [3H]TdR incorporated into trichloroacetic acid–precipitable material, as described elsewhere.15
Cell migration was measured in Modified Boyden Chambers equipped with 8-μm-porosity polyvinylpyrrolidone-free polycarbonate filters, precoated with collagen (20 μg/mL human type I collagen for 30 minutes at 37°C), as described.16
Measurement of Monocyte Chemoattractant Protein-1 and Procollagen Type I Secretion.
Confluent HSCs in 24-well plates were deprived of serum for 24 hours. After replacement with 500 μL fresh serum-free medium, cells were treated with different agonists at the indicated doses and time points. At the end of the incubation, conditioned medium was collected and stored at −20°C until assayed. Monocyte chemoattractant protein-1 (MCP-1) concentration in the medium was measured by enzyme-linked immunosorbent assay (ELISA) (Biosource, Camarillo, CA). Type I procollagen was detected using a C-peptide enzyme immunoassay kit (Takara Bio Inc., Otsu, Shiga, Japan).
Western Blot Analysis.
Confluent, serum-starved HSCs were treated with the appropriate stimuli, quickly placed on ice, and washed with ice-cold phosphate-buffered saline. Procedures for preparation of cell lysates, sodium dodecyl sulfate polyacrylamide gel electrophoresis analysis and Western blotting have been described elsewhere.17
Transfection of human HSCs was performed by nucleofection technology (Amaxa, Koln, Germany) as described,18, 19 using 100 nM non-targeting small interfering RNA (siRNA) (Dharmacon Inc., Lafayette, CO; catalog #D-001210-01) or siRNA targeting the α subunits of AMPK (Santa Cruz Biotechnology Inc., catalog #sc-45312). Transfection efficiency was monitored adding 1.5 μg of a plasmid encoding for green fluorescent protein before nucleofection. Fluorescence was consistently observed 24 to 48 hours after the end of transfection. The efficiency of silencing was evaluated by immunoblotting for AMPK.
Unless otherwise indicated, data presented as bar graphs are the mean of 3 independent experiments. Auto-luminograms are representative of at least 3 experiments with similar results. Statistical analysis was performed by Student t test. P values ≤0.05 were considered significant.
Adiponectin Activates AMPK in Human HSCs.
Two adiponectin receptors, AdipoR1 and AdipoR2, have been identified,6 but their expression patterns in human HSC have not been previously investigated. Using quantitative reverse transcription polymerase chain reaction, expression of both AdipoR1 and AdipoR2 was found in transdifferentiated HSCs, with a predominance of AdipoR1 expression (Fig. 1A,B). We next evaluated the effects of recombinant adiponectin on the biological actions of activated HSCs using the “globular” form, which has been shown to bind both AdipoR1 and AdipoR2 with high affinity.6 As expected, incubation of serum-starved HSCs with PDGF induced an increase in the uptake of [3H]TdR, an index of cell proliferation. Preincubation with increasing concentrations of adiponectin resulted in a marked and dose-dependent inhibition of PDGF-induced DNA synthesis (Fig. 1C).
PDGF also stimulates migration of cells implicated in wound repair, an important characteristic of the fibrogenic response. Also in this case, the marked increase in HSC chemotaxis induced by PDGF was significantly inhibited by incubation with adiponectin (Fig. 1D). Adiponectin also determined a diminished secretion of the proinflammatory factor MCP-1 in response to IL-1β, which potently induces secretion of this chemokine (Fig. 1E). However, incubation with adiponectin for 24 hours did not significantly modify the expression of alpha-smooth muscle actin (Fig. 1F).
Because activation of adiponectin receptors has been shown to increase AMPK activity in different systems, we evaluated the ability of globular adiponectin to phosphorylate AMPK on activation-specific residues in human HSC (Fig. 2). When these cells were exposed to adiponectin, an increase in AMPK phosphorylation was evident after 30 minutes of incubation. An additional, late peak of AMPK activation was also observed 24 hours after addition of adiponectin.
The Activated Phenotype of HSCs Is Modulated by Direct AMPK Activation.
Because adiponectin triggers AMPK-independent signals,20, 21 we next established whether direct activation of AMPK is sufficient to modulate the activated phenotype of human HSCs. To do this, HSCs were exposed to AICAR, a widely used AMPK activator with antidiabetic properties. Preliminary dose–response experiments showed that AICAR, at a concentration of 1 mM, consistently and significantly activated AMPK without any signs of cell toxicity (Table 1). Thus, this concentration was used in all subsequent experiments. As expected, AICAR increased AMPK phosphorylation on activation-specific residues (Fig. 3A) with a time-course similar to that observed in cells incubated with globular adiponectin. We next analyzed the effects of AICAR on HSC proliferation and migration in response to PDGF. Similarly to data obtained with adiponectin, the increase in DNA synthesis caused by PDGF was markedly and significantly inhibited by preincubation with AICAR (Fig. 3B). Time-course experiments varying the times of exposure to AICAR before addition of PDGF demonstrated that a short-term addition (10 minutes) is sufficient to provide inhibition of PDGF's effects (data not shown). Similarly, the marked increase in cell migration induced by PDGF was significantly reduced after incubation with AICAR (Fig. 3C).
Table 1. Viability of Cells Exposed to AMPK Activators
1 mM AICAR
10 mM metformin
20 mM metformin
Serum-deprived HSCs were either left untreated or incubated with the indicated concentration of AMPK activators for 48 hours. At the end of incubation, cell viability was measured by trypan blue exclusion (mean ± SD of 3 determinations).
Viable cells (% of total)
93.4 ± 1.9
93.1 ± 3.7
93.0 ± 2.9
92.8 ± 4.0
HSC transdifferentiation is associated with complex changes in cell function that involve secretion of extracellular matrix components and different cytokines. To evaluate the effects of AMPK activation on these features, secretion of type I procollagen in the conditioned medium was measured in the presence or absence of AICAR. Procollagen accumulated in the conditioned medium in a time-dependent fashion and were significantly lower after a 48-hour incubation with AICAR (Fig. 4A).
To explore the possible interference of AMPK with the proinflammatory activity of HSCs, we measured MCP-1 secretion in unstimulated conditions and after exposure to IL-1. As previously reported, IL-1 markedly increased MCP-1 secretion, at both times tested. Preincubation for 10 minutes with 1 mM AICAR markedly inhibited this effect (Fig. 4B). Similar to what was observed with adiponectin, AICAR did not modify the expression of α-smooth muscle actin (data not shown).
Metformin Mimics the Effects of AICAR on HSCs.
To confirm the effects of AMPK on activated HSCs, we used the antidiabetic drug metformin as an alternative strategy to activate AMPK. Metformin effectively induced AMPK phosphorylation (Fig. 5A), without any toxic effects on HSCs (Table 1). In addition, PDGF-induced DNA synthesis and cell migration were significantly inhibited by exposure to this drug (Fig. 5B,C). Similarly, metformin, at all concentrations tested, significantly inhibited the increase in MCP-1 secretion stimulated by IL-1β (Fig. 5D). These results indicate that metformin mimics the effects of AICAR on the activated phenotype of HSCs and confirm the involvement of AMPK in this setting.
AMPK Silencing Increases PDGF-Induced Proliferation of HSCs.
To substantiate that AMPK negatively modulates the transdifferentiated phenotype of human HSCs, using a nonpharmacological approach, we analyzed the effects of AMPK depletion using gene silencing. AMPK-targeting siRNA significantly decreased endogenous protein levels, with a mean reduction of approximately 50%, when compared with cells treated with nontargeting siRNA (Fig. 6A,B). Knockdown of AMPK by RNA interference was associated with a significant increase in PDGF-induced cell proliferation (Fig. 6C), confirming that AMPK is a negative modulator of the effects of PDGF on activated HSCs.
AMPK Activation Regulates Intracellular Signaling Pathways in HSCs.
We next explored some of the molecular mechanisms mediating the inhibitory effects of AMPK on the activated phenotype of HSCs. Activation of nuclear factor kappa B (NF-κB) contributes to the expression of proinflammatory chemokines such as MCP-1.17 Exposure of HSCs to IL-1 resulted in a marked increase in IκBα phosphorylation, an index of NF-κB activation, whereas preincubation with AICAR reduced the effect of IL-1 (Fig. 7A). To investigate the mechanisms responsible for the effect of AMPK activators on PDGF's actions, we first analyzed the activation of ERK and Akt, 2 critical pathways for PDGF-dependent proliferation and migration.16, 22 As shown in Fig. 7B, PDGF induced a marked increase in ERK activation after 10 minutes, but AICAR did not affect this pathway even when added 24 hours before PDGF. Similarly, AICAR had no inhibitory effects on Akt phosphorylation induced by short-term incubation with PDGF. At late time points after addition of growth factors, activation of ERK and Akt is related to progression of the cell cycle.23 When HSCs were exposed to AICAR for 10 minutes and then stimulated with PDGF for 24 hours, a reduction in the activation of both ERK and Akt was observed. Thus, these results indicate that AMPK does not interfere with the early signaling pathways leading to ERK or Akt activation and confirm the role of AMPK activators in blocking cell growth (Fig. 7C).
Because recent evidence indicates that AMPK may regulate cell growth acting through the mammalian target of rapamycin (mTOR) pathway, we analyzed the effect of AICAR on mTOR's downstream effectors, ribosomal S6 kinase (p70S6K) and eukaryotic initiation factor 4EBP1. A 10-minute exposure of HSCs to PDGF augmented the phosphorylation of p70S6Kand 4EBP1. Preincubation of cells with AICAR for different time points resulted in a reduction of the stimulatory effect of PDGF, reaching a maximum level of inhibition after 24 hours (Fig. 7B). Recent studies demonstrate that the gene products of TSC1 and TSC2 exert an inhibitory effect on the mTOR cascade.24 Growth factors, by acting through Akt, phosphorylate TSC2, thus decreasing its ability to inhibit the phosphorylation of p70S6K and 4EBP1.24 As expected, PDGF induced an increase in TSC2 phosphorylation on Thr1462 residue, without any changes in cells preincubated with AICAR (Fig. 7B).
Accumulating evidence shows a causal relation between obesity, insulin resistance, and progression of chronic liver diseases, including nonalcoholic steatohepatitis. In the current study, we demonstrate for the first time that activation of AMPK in response to adiponectin or to antidiabetic drugs negatively modulates the myofibroblastic phenotype of human HSCs, and identify AMPK as a point of integration of signals deriving from fibrogenic and anti-fibrogenic cytokines. AMPK controls hepatic glucose and lipid metabolism, reduces gluconeogenesis and de novo lipogenesis, and stimulates fatty acid oxidation.11 The beneficial effects of AMPK activation have been recently demonstrated in several models of the metabolic syndrome, where AMPK activation in response to hormones or drugs ameliorates liver steatosis and glucose metabolism.25 At a cellular level, AMPK monitors cellular energy status, acting as a master switch to reduce adenosine triphosphate–consuming pathways, and coordinates cell cycle progression with the availability of energy sources.26 The processes accompanying HSC activation, and particularly replicative cell division, are energetically demanding and need sufficient metabolic resources.26 Because AMPK activity is repressed by a large availability of nutrients, this pathway may be viewed as a mechanism linking the excess of substrates typical of the metabolic syndrome with a faster development of fibrosis.
In physiological conditions, adiponectin is one of the major hormonal systems responsible for AMPK activation. In this study, the actions of adiponectin on human HSCs were found to be qualitatively similar to those previously observed in rodent HSCs.5, 13 In fact, globular adiponectin caused a marked reduction in HSC growth, motility, and chemokine secretion, and consistently activated AMPK, demonstrating that the adiponectin–AMPK axis is functional in these cells. In contrast, adiponectin did not modify alpha-smooth muscle actin expression, at variance with data reported in rat HSC.13 Differences in the species, experimental protocol (recombinant versus adenovirally expressed adiponectin), length of incubation (24 versus 48 hours), or activation status may possibly explain this discrepancy.
Although AMPK is an established downstream effector of adiponectin receptor activation, its role in the regulation of fibrogenesis had not been clarified. AICAR mimics the presence of high levels of AMP inside the cells, and therefore reproduces a situation of limited nutrient availability. This drug resulted in a negative modulation of cell proliferation and migration in response to PDGF, and inhibited the secretion of MCP-1 and collagen. This latter effect was evident only after prolonged incubation, suggesting that the effects of AICAR are mediated by other soluble mediators acting in an autocrine fashion.
Because AICAR has been shown to regulate other AMP-sensitive enzymes,11 we employed additional strategies to establish a role for AMPK in the setting of HSC biology. In addition to the classical activation pathway, AMPK can be stimulated by the antidiabetic drug metformin.27 We found that incubation of HSCs with metformin was also effective in increasing AMPK activation, and reproduced the biological actions of AICAR. Of note, the effects of metformin were in general less marked than those elicited by AICAR and were evident at higher concentrations than those reached after therapeutic doses of this drug, usually in the micromolar range. It should be considered that most of the metabolic effects of metformin are mediated by inhibition of hepatic gluconeogenesis, and that the liver is exposed to much higher metformin concentrations via portal inflow. In addition, the limited effectiveness of metformin in cultured cells as compared with its in vivo administration has been widely recognized and could be related to the possibility that metformin does not efficiently accumulate in vitro. Nonetheless, the range of metformin concentrations used in the current study is similar to that employed in other studies conducted in myofibroblasts or endothelial cells.28, 29 To provide further support to the role of AMPK, we down-regulated its intracellular levels by gene silencing. This molecular approach was accompanied by increased cell proliferation in response to PDGF, in accordance with the observation that activation of AMPK exerts a down-regulatory action on the fibrogenic properties of HSCs.
This study also confirms in human myofibroblastic HSCs the expression of both AdipoR1 and AdipoR2, which was recently reported in rat HSCs.13 Interestingly, a very recent study investigating the effects of deletion of AdipoR1 and AdipoR2 indicated that AMPK activation is predominantly linked to AdipoR1 activation.21 It is of interest that in HSCs, expression of AdipoR1 was higher than that of AdipoR2, at least at the messenger RNA level. In addition, globular adiponectin, which has a higher affinity for AdipoR1,6 effectively modulated HSC biology. These data support a major role played by AMPK as a downstream effector of adiponectin's actions in these cells. Further studies are needed to establish the relative involvement of each receptor and the possible contribution of AMPK-independent pathways.
We also provided novel information on the intracellular signaling pathways regulated by AMPK in this cell type. In cells exposed to AICAR, the ability of IL-1 to activate NF-κB was reduced, providing a possible mechanism for the less effective induction of MCP-1 secretion.30 This observation is in agreement with data recently reported in endothelial cells,28 although further studies are needed to clarify the mechanisms of NF-κB inhibition and the actual contribution of this pathway to the inhibitory effects of AMPK activators on MCP-1 expression.
In cells exposed to PDGF, ERK and Akt activation are necessary for mitogenesis and chemotaxis, and in some systems AMPK has been shown to decrease the activation of these kinases.31 However, our data indicate that AICAR does not inhibit ERK or Akt when HSCs are exposed to PDGF for 10 minutes, demonstrating that AMPK does not interfere with early postreceptor signaling pathways leading to activation of these kinases. In contrast, when PDGF stimulation was protracted to 24 hours, AICAR reduced the activity of both ERK and Akt. Because a late peak of activation of these kinases occurs in the G2/M phase,23 these results confirm a role for AMPK in blocking cell cycle progression.32 We therefore extended our investigation to the signaling systems dependent on activation of mTOR, recently identified as a major target of AMPK.33 mTOR can be directly inactivated by AMPK through phosphorylation of Thr2446, or indirectly, by phosphorylation-mediated activation of TSC2, on Thr1227 and Ser1345 residues. In fact, the TSC1/TSC2 complex integrates signals originating from AMPK and growth factor receptors, these latter inducing dissociation of the complex, and mTOR activation.24 We found that in HSCs, PDGF activates the mTOR pathway, as indicated by increased TSC2 phosphorylation on inhibitory residues,33 and by phosphorylation of p70S6K and 4EBP1, 2 downstream targets of mTOR. When cells were exposed to AICAR, activation of mTOR was markedly reduced, indicating that the inhibitory action of AMPK is exerted, at least partially, at this level. The role played by mTOR is further supported by 2 additional lines of evidence. First, AICAR did not modify TSC2 phosphorylation, suggesting that the pathways upstream of TSC2 are unaffected by AMPK. Second, Akt activation was slightly increased in HSCs exposed to AICAR, indicating that AMPK activation relieves the feedback inhibition of the PI3K-Akt pathway mediated by mTOR.24 These results identify NF-κB and the mTOR pathway as molecular mediators of the inhibitory action of AMPK on the activated HSC phenotype.
Taken together, data from this study provide evidence that AMPK is a novel modulator of the activated phenotype of HSCs, affecting several functions relevant to the hepatic wound-healing response, including inflammation, proliferation, migration, and expression of extracellular matrix components. These results, and the described role of the mTOR pathway, indicate that AMPK integrates signals from cytokine receptors with those provided by nutrient availability and sensing of the energy status in HSCs. The use of drugs activating AMPK, which have a proven beneficial action on liver glucose and lipid metabolism, may have an additional rationale in their antifibrogenic properties.
We thank Wanda Delogu and Nadia Navari for expert technical assistance.