Potential conflict of interest: Nothing to report.
Adiponectin is an adipocyte-derived, antidiabetic, antiatherogenic adipocytokine that is present in serum as 3 isoforms. Decreased plasma adiponectin levels are closely associated with the severity of nonalcoholic fatty liver diseases. This study was designed to elucidate a role of adiponectin and its mediator adenosine monophosphate–activated protein kinase (AMPK) on proliferation of activated hepatic stellate cells (HSCs), the key cells promoting fibrosis. Immortalized human HSC line hTERT and primary rat HSCs were stimulated with platelet-derived growth factor (PDGF) with or without pretreatment with AMPK activator 5-aminoimidazole-4-carboxamide-1-4-ribofuranoside (AICAR), metformin, or high molecular weight (HMW) adiponectin. HMW adiponectin dose-dependently suppressed PDGF-induced HSC proliferation. Adenoviral transduction with dominant-negative AMPK (DN-AMPK) abolished the suppressive effect of adiponectin in HSCs. AICAR, metformin, or transduction of constitutively active AMPK attenuated PDGF-induced [3H]thymidine incorporation, which was abolished by either a chemical AMPK inhibitor or transduction of DN-AMPK, consistent with an antiproliferative effect of AMPK. The suppressive effect of AMPK on HSC proliferation is mediated through multiple mechanisms, including (1) an inhibition of the AKT pathway, (2) inhibition of NADPH oxidase–dependent reactive oxygen species (ROS) production via induction of antioxidant enzymes, and (3) an increase in the expression of the cyclin-dependent kinase inhibitors p27kip1 and p21cip1. Conclusion: Adiponectin inhibits HSC proliferation via activation of AMPK. AMPK activation by AICAR or metformin inhibits HSC proliferation via suppression of ROS production and subsequent inhibition of AKT pathway. Thus, adiponectin and AMPK inhibit HSC proliferation and hepatic fibrosis via multiple molecular mechanisms. (HEPATOLOGY 2008;47:677–685.)
Hepatic fibrosis is a wound-healing response to chronic liver injury, and hepatic stellate cells (HSCs) play a crucial role in this fibrotic response.1 Nonalcoholic fatty liver disease (NAFLD) is a major cause of liver morbidity and is highly associated with obesity and insulin resistance.2 In addition, obesity is an independent risk factor for the development of fibrosis in NAFLD,3 alcoholic liver disease,4 and chronic hepatitis C.5
Adiponectin is an adipocyte-specific adipocytokine. Its plasma levels are reduced in obese individuals, and the reduced adiponectin levels correlate with insulin resistance and diabetes as well as coronary artery disease.6 Adiponectin exists mainly in plasma as 3 forms of full-length adiponectin: a low molecular weight (LMW) trimer, a medium molecular weight hexamer, and a high molecular weight (HMW) multimer.7, 8 The level of HMW adiponectin is a more sensitive marker of insulin resistance than that of total adiponectin. For example, increases in the ratio of HMW to total adiponectin in plasma, but not total adiponectin levels, correlate with improvement in insulin sensitivity during treatment with an insulin-sensitizing drug in both mice and human diabetic patients.9
The insulin-sensitizing effect and anti-atherosclerosis effect of adiponectin are mediated mainly by the activation of adenosine monophosphate (AMP) activated protein kinase (AMPK) and peroxisome proliferator–activated receptor-α (PPAR-α).6 AMPK plays a key role in the regulation of energy homeostasis and acts as a “metabolic sensor” to regulate adenosine triphosphate (ATP) concentrations.10 AMPK is activated by various stresses including exercise, calorie deprivation, hypoxia, ischemia, and oxidative stress, all of which increase the ATP to AMP ratio. Once activated, AMPK stimulates ATP, generating cellular events such as glucose uptake and lipid oxidation to produce energy, while turning off energy-consuming processes such as glucose and lipid production to restore energy balance. Recent studies have reported that AMPK suppresses proliferation in vascular smooth muscle cells11, 12 and cancer cells.13
Decreased adiponectin levels are closely associated with degree of hepatic steatosis, necroinflammation, and fibrosis in NAFLD.14 Adiponectin-null mice develop more extensive carbon tetrachloride-induced hepatic fibrosis than wild-type mice, suggesting that adiponectin also regulates liver fibrogenesis.15 In addition, adiponectin inhibits proliferation of vascular smooth muscle cells16 and myelomonocytic lineage cells.17 These experimental and epidemiological findings imply that adiponectin has a suppressive effect on liver fibrogenesis; however, the mechanism by which adiponectin inhibits HSC proliferation is unclear. Accordingly, the purpose of the present study was to evaluate the role of adiponectin and its intracellular mediator AMPK on the regulation of HSC proliferation. Our results show that adiponectin, especially the HMW form, inhibits HSC proliferation via AMPK. The mechanisms by which activation of AMPK suppresses HSC proliferation include inhibition of AKT pathway, suppression of reactive oxygen species (ROS) production via up-regulation of antioxidant genes, and regulation of genes that control the cell cycle.
5-Aminoimidazole-4-carboxamide-1-4-ri-bofuranoside (AICAR), metformin (1,1-dimethyl-biguanide), diphenyleneiodonium chloride (DPI), glutathione, and WY-14643 were obtained from Sigma-Aldrich (St. Louis, MO). AMPK inhibitor Compound C (6-[4-(2-piperidin-1-yl-ethoxy)-phenyl)]-3-pyridin-4-yl-pyrrazolo[1,5-a]-pyrimidine) was obtained from EMD Biosciences (La Jolla, CA). PDGF-BB was obtained from Roche (Mannheim, Germany). HMW and LMW adiponectin isolated via gel filtration chromatography9 was a kind gift from Dr. L. Shapiro (Columbia University, New York, NY).
Cell Cultures and Treatments.
The immortalized human HSC line hTERT was used as described previously.18, 19 Microarray and reverse-transcription polymerase chain reaction (RT-PCR) analysis of hTERT HSCs showed that messenger RNA (mRNA) expression patterns are similar to those in activated primary human HSCs, and hTERT HSCs exhibit morphological and functional characteristics of activated HSCs.19 Primary HSCs were isolated from male Sprague-Dawley rats as described previously and were used for experiments between day 10 and day 14.
The adenoviral vectors encoding constitutively active mutant of AMPKα1 (CA-AMPK)20 and dominant-negative mutant of AMPKα2 (DN-AMPK)21 were a kind gift from Dr. K. Walsh (Boston University, Boston, MA). Control adenovirus expressing green fluorescent protein (GFP) has been described previously.18 Cells were transduced with each adenovirus at a multiplicity of infection of 200 in order to achieve transduction rates of >80%.
[3H]Thymidine Incorporation Assay.
DNA synthesis was estimated as the amount of methyl-[3H]thymidine as described previously.18 HSCs were incubated with 1 μCi/mL [3H]thymidine (Amersham, Piscataway, NJ) for 18 hours followed by trichloroacetic precipitation, lysis, and measurement in a scintillation counter.
Fluorescence-Activated Cell Sorting Analysis.
Twenty-four hours after treatment with PDGF, cells were harvested by scraping and fixed with cold ethanol (50%) in phosphate-buffered saline for 1 hour. Cells were then washed with phosphate-buffered saline and treated with 0.5 mg/mL RNase A (Qiagen, Valencia, CA) for 1 hour at 37°C. Cells were incubated with 20 μg/mL propidium iodide (Sigma-Aldrich) at 4°C in the dark, and cell cycle state was assessed via flow cytometry using a fluorescence-activated cell sorting (FACS) instrument (LSRII; BD Biosciences, San Jose, CA).
Western Blot Analysis and Immunoprecipitation.
Electrophoresis of protein extracts and subsequent blotting were performed as described previously. Blots were incubated with antibodies against phospho-FoxO1 (Ser-256), phospho-AKT (Ser-473), AKT, Phospho-AMPK (Thr172), Phospho-p70S6K, α-AMPK, p27kip1, phospho-p53 (Ser-15), p53 antibodies (all from Cell Signaling Technology, Beverly, MA), FoxO1, p21cip1 (both from Santa Cruz Biotechnology, Santa Cruz, CA), and β-actin (both Simgma-Aldrich) overnight at 4°C.
Determination of Intracellular ROS Production in HSCs.
ROS production was measured using 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester (DCF-DA; Molecular Probes, Eugene, OR). DCF-DA–associated ROS were measured in a time course over 30 minutes using a multiwell plate reader (BMG Optima, Durham, NC).
Quantitative Real-Time RT-PCR.
Extracted RNA from hTERT HSCs was reverse-transcribed (First-Strand cDNA Synthesis Kit, Amersham), and quantitative real-time PCR with the probe–primers sets (Applied Biosystems, Foster City, CA) was performed using Taqman analysis (ABI Prism 7000 Sequence Detection System, Applied Biosystems). The changes were normalized to 18S ribosomal RNA (Hs99999901_s1).
Results are expressed as the mean ± standard deviation (SD). The results were analyzed using an unpaired Student t test. A P value of less than 0.05 was considered statistically significant.
AICAR and Metformin Activate AMPK and Inhibit PDGF-Stimulated Proliferation of HSCs.
Because insulin-sensitizing and anti-atherosclerosis effects of adiponectin are mainly mediated by the activation of AMPK and PPAR-α,6 the effect of agents that activate either AMPK or PPAR-α on proliferation was evaluated in hTERT HSCs. Treatment with the AMPK activators AICAR and metformin for 2 hours dose-dependently increased the Thr-172 phosphorylation of α-AMPK, which is required for AMPK activation10 (Fig. 1A). Pretreatment with AICAR or metformin inhibited PDGF (20 ng/mL)-stimulated DNA synthesis in a dose-dependent manner as measured via [3H]thymidine incorporation (Fig. 1B). Pretreatment with the AMPK inhibitor Compound C significantly abolished the inhibitory effect of AICAR or metformin on HSC proliferation. In contrast, the PPAR-α activator WY-14643 did not inhibit HSC proliferation at the concentration known to only activate PPAR-α22 (Fig. 1C).
To further investigate the mechanisms by which activation of AMPK inhibits proliferation, FACS analyses were performed to examine the cell cycle. PDGF induced cells to accumulate in the S/G2/M phase of the cell cycle. Pretreatment with AICAR resulted in an increase in a number of cells in the G0/G1 phase even after treatment with PDGF (Fig. 1D-E). These results suggest that AMPK inhibits proliferation by inducing cell cycle arrest in the G1 phase.
To further support the role of AMPK in the antiproliferative effect of AICAR or metformin, we used adenoviral vectors that express dominant-negative AMPK (DN-AMPK) and constitutively active AMPK (CA-AMPK). DN-AMPK significantly abolished the suppressive effect of AICAR or metformin on HSC proliferation (Fig. 2A), whereas transduction of CA-AMPK also inhibited thymidine incorporation (Fig. 2B). To confirm that the antiproliferative effect of AMPK in hTERT HSCs can be reproduced in primary HSCs, we evaluated the effects of AICAR, metformin, or CA-AMPK in primary rat HSCs and in primary human HSCs.23 Pretreatment with AICAR or metformin significantly suppressed PDGF-induced proliferation, and the transduction of DN-AMPK abolished these suppressive effects in activated primary rat HSCs (Fig. 2C) and activated primary human HSCs (Supplementary Fig. 1A). In addition, transduction of CA-AMPK inhibited thymidine incorporation in activated primary rat HSCs (Fig. 2D) and activated primary human HSCs (Supplementary Fig. 1B). These results indicate that AMPK activation is a key process for an inhibitory effect of AICAR or metformin on HSC proliferation.
HMW Adiponectin Inhibits PDGF-Stimulated HSC Proliferation via AMPK Activation.
We evaluated the effect of HMW and LMW adiponectin on AMPK activation and HSC proliferation. Treatment with HMW adiponectin for 2 hours dose-dependently phosphorylated AMPK (Fig. 3A) and suppressed proliferation (Fig. 3B). Transduction of DN-AMPK abolished the suppressive effect of HMW adiponectin in hTERT HSCs, suggesting that the suppressive effect of adiponectin is mediated by AMPK (Fig. 3B). A similar suppressive effect of HMW adiponectin on proliferation was also observed in primary rat HSCs (Fig. 3C). In contrast, LMW adiponectin did not activate AMPK nor inhibit proliferation at the physiological concentration in plasma (2–20 μg/mL),24 whereas the higher dose (60 μg/mL) of LMW adiponectin weakly inhibited proliferation (Fig. 3D–F).
AICAR, Metformin, and HMW Adiponectin Inhibit the Phosphatidylinositol 3-Kinase/AKT Pathway via AMPK in HSCs.
Because AICAR inhibits cell proliferation, and because the phosphatidylinositol 3-kinase (PI3K)/AKT pathway is one of the most important pathways regulating proliferation in HSCs,25 we examined the effect of AMPK activation on the PI3K/AKT pathway. FoxO118 and mTOR/p70S6K,26 both downstream targets of PI3K/AKT pathway, are important mediators that regulate HSC proliferation. It has been reported that AMPK inhibits the mTOR pathway and is associated with the inhibition of cell growth and protein synthesis.27 PDGF phosphorylated FoxO1 and p70S6K in parallel with AKT in hTERT HSCs (Fig. 4A). AICAR and metformin significantly inhibited phosphorylation of p70S6K, which was reversed by an AMPK inhibitor. AICAR or metformin also inhibited PDGF-induced phosphorylation of AKT and FoxO1. The transduction of CA-AMPK inhibited AKT phosphorylation, whereas transduction of DN-AMPK abolished the suppressive effect of AMPK on AKT phosphorylation (Fig. 4B–C). In contrast, pretreatment with HMW adiponectin suppressed PDGF-induced AKT phosphorylation, and this suppressive effect was abolished by DN-AMPK (Fig. 4D). Thus, inhibition of the AKT pathway is one mechanism by which AMPK negatively regulates HSC proliferation.
AICAR or Metformin Inhibit PDGF-Induced ROS Production in hTERT HSCs.
NADPH oxidase–derived ROS are important intracellular mediators in mitogen-induced growth stimuli in HSCs.28, 29 Therefore, we focused on ROS as mediators of AKT activation. DCF-DA fluorescence significantly increased after incubation with PDGF, suggesting that ROS were produced in hTERT HSCs (Fig. 5A–B). Pretreatment with the NADPH oxidase inhibitor DPI prevented PDGF-induced intracellular ROS production, consistent with NADPH oxidase as the source of ROS in PDGF-treated HSCs.29 Pretreatment with metformin or AICAR also reduced PDGF-induced ROS production. Because AMPK inhibits NADPH oxidase–derived ROS production, we next examined whether AKT activation is NADPH oxidase–dependent. Indeed, PDGF-induced phosphorylation of AKT was inhibited by the NADPH oxidase inhibitor DPI and antioxidant glutathione, suggesting that PDGF-induced AKT signaling is mediated by NADPH oxidase–derived ROS (Fig. 5C). These results suggest that AMPK inhibits NADPH oxidase–derived ROS production and subsequent activation of AKT.
AMPK Up-regulates mRNA Expression of PGC-1α and Antioxidant Enzymes.
We next evaluated the mRNA expression of several genes via real-time TaqMan RT-PCR. Peroxisome proliferator–activated response-γ coactivator-1α (PGC-1α) has been reported to be up-regulated through an AMPK-related mechanism.30 Indeed, PGC-1α mRNA was significantly increased by either AICAR, metformin, or CA-AMPK, whereas DN-AMPK abolished the effect of AMPK or metformin (Fig. 5D). Because PGC-1α regulates ROS metabolism via induction of mitochondrial31–33 and nonmitochondrial34 antioxidant enzyme genes, we also evaluated mRNA expression of antioxidant enzymes. Either AICAR, metformin, or CA-AMPK up-regulated mRNA expression of SOD2 and catalase, which were abolished by DN-AMPK (Fig. 5D). On the other hand, SOD1 and glutathione peroxidase 1 were not up-regulated by AMPK.
AMPK Induces the Expression of Cyclin-Dependent Kinase Inhibitors and Phosphorylation of p53 at Ser-15.
Because AICAR inhibits cell proliferation by arresting cell cycle at G1/S phase (Fig. 1D–E), we also examined the expression of cyclin-dependent kinase (cdk) inhibitors, which inhibit the progression of cell cycle at G1/S phase. AICAR or metformin increased the mRNA expression of p21cip1 and p27kip1, which was abolished by DN-AMPK (Fig. 6A). Transduction of CA-AMPK also increased protein expression of p21cip1, p27kip1, and p53 (Fig. 6B). In addition, transduction of CA-AMPK increased p53 phosphorylation at Ser-15 (Fig. 6B).
The present study demonstrates that adiponectin, especially the high molecular weight (HMW) form, suppresses PDGF-induced HSC proliferation. HMW adiponectin also activates AMPK, and transduction of DN-AMPK abolished the antiproliferative effect of HMW adiponectin, indicating that activation of AMPK is crucial in the antiproliferative effect of HMW adiponectin in HSCs. Although the biological activities of each adiponectin isoform are unresolved, several observations indicate that decreases in the ratio of HMW to total adiponectin and in the level of plasma HMW adiponectin correlate more closely with insulin resistance and type 2 diabetes than total adiponectin levels.9, 35 Only HMW adiponectin induces activation of AMPK in hepatocytes,8 and HMW adiponectin but not LMW adiponectin protects from apoptosis via AMPK activation in human umbilical vein endothelial cells (HUVEC).36 The present study demonstrates that LMW adiponectin could only activate AMPK and inhibit HSC proliferation at concentrations higher than physiological plasma levels. Consistent with these observations, our data demonstrate that HMW adiponectin has more biological activity than LMW adiponectin in HSCs.
Although the beneficial effect of metformin in a mouse model of NAFLD has been demonstrated,37 an in vivo study to evaluate the effect of AMPK activators on HSCs in liver fibrosis is difficult to design and interpret, because the effect of AMPK activators cannot be attributed only to hepatic stellate cells, but also to major effects on hepatocytes. Metformin has shown mixed results in human NAFLD (see review by Angelico et al.38). Although the Bugianesi 2005 trial showed a significant improvement in steatosis, necroinflammation, and liver fibrosis in 17 metformin-treated cases,39 other trials did not demonstrate improved hepatic fibrosis. However, these results are difficult to interpret because of a limited number of cases with histological assessment. Adequately powered, randomized, placebo-controlled studies of NASH with histologic evaluation are still required.
The antiproliferative effect of AMPK on HSCs appears to be mediated through multiple mechanisms, including (1) inhibition of the AKT pathway, (2) inhibition of NADPH oxidase–dependent ROS production via induction of antioxidant enzymes, and (3) an increase in the expression of cdk inhibitors p27kip1 and p21cip1. The PI3K/AKT pathway is activated by PDGF and is an important intracellular mediator of growth signals in HSCs.25 In HSCs, activation of AMPK not only inhibits mTOR/p70S6K pathway as documented previously in other cells,27 but also attenuates the upstream AKT activity. Because FoxO1 is phosphorylated and inhibited by AKT, and FoxO1 inhibits HSC proliferation via FoxO1-dependent gene expression of p27kip1 and SOD2,18 inhibiting not only mTOR/p70S6K pathway but also upstream AKT activity more effectively produces an antiproliferative effect.
The mechanism by which AMPK inhibits PDGF-induced activation of the AKT pathway may also be mediated by blocking intracellular ROS production. ROS derived from NADPH oxidase acts as a second messenger for mitogen-induced proliferation in HSCs.28, 29 The present study demonstrates that activation of AMPK inhibits NADPH-derived intracellular ROS production in PDGF-treated HSCs. This inhibitory effect of AMPK on ROS production is consistent with previous reports observed in neutrophils40 and HUVEC.33 We also demonstrated that the NADPH oxidase inhibitor DPI abolished PDGF-induced ROS production and phosphorylation of AKT. These results indicate that activation of AMPK inhibits NADPH oxidase–dependent ROS production, therefore inhibiting AKT activation in HSCs.
We hypothesize that PGC-1α is the AMPK-induced mediator that inhibits ROS. PGC-1α acts as a coactivator of several transcriptional factors and has been implicated in energy homeostasis and glucose metabolism.41 Recent reports also suggest that AMPK induces PGC-1α expression30 and that PGC-1α stimulates mitochondrial biogenesis, including induction of mitochondrial antioxidant enzymes.31–33 Consistent with these reports, the present study demonstrated that AICAR, metformin, or transduction of CA-AMPK increased the mRNA levels of PGC-1α as well as the mitochondrial antioxidant enzyme SOD2. In addition, activation of AMPK induced—though to a lesser extent—nonmitochondrial antioxidant enzyme catalase, which is reported to be up-regulated by PGC-1α.34 Collectively, the present study implies that AMPK attenuates PDGF-induced intracellular ROS production and subsequent AKT activation, at least in part, via up-regulation of antioxidant enzymes. Further study will be needed to determine the role of PGC-1α in HSCs.
The present study demonstrated that AMPK also controls HSC proliferation via induction of cdk inhibitors p21cip1 and p27kip1. The increased expression of p21cip1 is possibly mediated by a p53-dependent mechanism. This study also demonstrated that AMPK phosphorylates p53 at Ser-15 and increased p53 protein levels, consistent with previous observations.12, 42, 43 p53 phosphorylation at Ser-15 induces accumulation and activation of this protein,44 resulting in up-regulation of p21cip1 transcription. AMPK is also reported to phosphorylate p27kip1 at Thr-198, thereby increasing p27 stability.45 As we demonstrated that AMPK increases p27kip1 mRNA levels, increased expression of p27kip1 might be mediated by both posttranscriptional and transcriptional regulation. Because FoxO1 regulates expressions of p27kip1 and SOD2 genes in HSCs,18 AMPK inhibition of the AKT pathway may, in part, participate in the up-regulation of p27kip1 and SOD2 genes. Because PGC-1α coactivates FoxO1,46 it is also possible that it coactivates and up-regulates the expression of these genes. These effects would be cumulative, resulting in higher levels of the cdk inhibitors.
Low adiponectin levels may play a pivotal role in the progression of NAFLD. Recent human studies14, 47 indicate that low adiponectin levels correlate with the severity of NAFLD from simple steatosis to fibrosis and cirrhosis. The present study suggests that adiponectin per se inhibits HSC proliferation via AMPK and that a decrease in HMW adiponectin would enhance the progression of fibrosis in obese patients with NAFLD.
In conclusion, the present study indicates that AMPK signaling is essential for the antiproliferative activities of HMW adiponectin on cultured activated HSCs. The identification of an antiproliferative action of HMW adiponectin, as reported in this study, is consistent with an antifibrotic effect on liver disease for this adipocytokine. Understanding the roles of HMW adiponectin and AMPK in fibrogenesis may provide new insight into the pathophysiology and therapeutic strategies for liver fibrosis, especially in patients with NAFLD.