Potential conflict of interest: Nothing to report.
Glucagon-like peptide 1 (GLP-1) is a naturally occurring peptide secreted by the L cells of the small intestine. GLP-1 functions as an incretin and stimulates glucose-mediated insulin production by pancreatic β cells. In this study, we demonstrate that exendin-4/GLP-1 has a cognate receptor on human hepatocytes and that exendin-4 has a direct effect on the reduction of hepatic steatosis in the absence of insulin. Both glucagon-like peptide 1 receptor (GLP/R) messenger RNA and protein were detected on primary human hepatocytes, and receptor was internalized in the presence of GLP-1. Exendin-4 increased the phosphorylation of 3-phosphoinositide-dependent kinase-1 (PDK-1), AKT, and protein kinase C ζ (PKC-ζ) in HepG2 and Huh7 cells. Small interfering RNA against GLP-1R abolished the effects on PDK-1 and PKC-ζ. Treatment with exendin-4 quantitatively reduced triglyceride stores compared with control-treated cells. Conclusion: This is the first report that the G protein–coupled receptor GLP-1R is present on human hepatocytes. Furthermore, it appears that exendin-4 has the same beneficial effects in vitro as those seen in our previously published in vivo study in ob/ob mice, directly reducing hepatocyte steatosis. Future use for human nonalcoholic fatty liver disease, either in combination with dietary manipulation or other pharmacotherapy, may be a significant advance in treatment of this common form of liver disease. (HEPATOLOGY 2010)
Glucagon-like peptide 1 (GLP-1) is a peptide product of the L cells of the small intestine and proximal colon and has been the subject of considerable laboratory research over the past two decades. Although the primary function of GLP-1 is to serve as an incretin in β cells of the mammalian pancreas, the functioning peptide is quickly cleaved by dipeptidyl peptidase IV, rendering the peptide functionally inactive.1-3 The principle pleotropic effects of GLP-1 include enhanced satiety, delayed gastric emptying,4, 5 and increased lower gastrointestinal motility.1, 6 GLP-1 binds to its cognate receptor, glucagon-like peptide 1 receptor (GLP-1R), a G protein–coupled receptor (GPCR) that has been found in many tissues, including the brain and pancreas.4, 7 However, it is unknown whether GLP-1 has a functioning receptor on hepatocytes. Mice that lack GLP-1R (DIRKO) do not seem to have marked hepatic metabolic changes.8-12 exendin-4 is a 39–amino acid agonist of GLP-1R that is derived from the saliva of the Gila monster (Heloderma suspectum). At present, exendin-4 is being used to augment insulin production in patients with type 2 diabetes.13 We recently reported that exendin-4 significantly reduced hepatic steatosis found in ob/ob mice, but we did not elucidate a cellular mechanism, nor did we determine whether the beneficial effects were the result of direct action on hepatocytes or nonhepatic effectors.14
Nonalcoholic fatty liver disease (NAFLD) is strongly associated with other clinical features of the metabolic syndrome, including obesity, type 2 diabetes mellitus, hypertension, and dyslipidemia. Insulin resistance is a central feature of the metabolic syndrome. In particular, hepatocyte insulin resistance—in part related to impaired insulin signal transduction—may be a key problem in the development of hepatocyte steatosis. In the present study, we positively identified GLP-1R not only in the transformed hepatocyte cell lines Huh7 and HepG2, but also in primary human hepatocytes. We have also demonstrated, as with other GPCRs, that GLP-1R internalizes on binding to its ligand.3 GLP-1 or exendin-4 can activate key signaling molecules downstream of insulin receptor substrate (IRS)-2. Furthermore, in the absence of insulin, we demonstrated a significant loss of triglycerides (TGs) from steatotic hepatocytes following exendin-4 treatment. To our knowledge, this is the first study that convincingly demonstrates GLP-1R on hepatocytes and provides a signaling mechanism whereby GLP-1 proteins can independently reduce hepatocyte TG accumulation.
GLP-1, glucagon-like peptide 1; GLP-1R, glucagon-like peptide 1 receptor; GPCR, G protein–coupled receptor; IRS, insulin receptor substrate; NAFLD, nonalcoholic fatty liver disease; SE, standard error; siRNA, small interfering RNA; TG, triglyceride.
Materials and Methods
HepG2 and Huh7 cells were purchased from American Type Culture Collection (Manassas, VA) and cultured using Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) with 10% fetal bovine serum (Hyclone, Logan, UT). Cells were treated with 10 nM GLP-1 or 10 nM exendin-4 (Sigma, St. Louis, MO) for varying time intervals from 5 minutes to 12 hours in accordance with published reports.15, 16
Primary Hepatocyte Culture.
Primary hepatocytes were purchased from Lonza (Allendale, NJ) and were grown to confluence in medium (HMM CC-3197 with HMM single quots CC-4192) on collagen-coated plates (BD Biosciences, Bedford, MA), at a density of 0.15 mL cells/0.5 mL medium. RNA and protein were subsequently extracted in the absence of insulin.
Real-Time Polymerase Chain Reaction.
Total RNA was extracted from Huh7 and human hepatocytes using TRIzol reagent (Invitrogen). Real-time polymerase chain reaction was performed using the following primers for GLP-1R: forward, 5′-TTG GGG TGA ACT TCC TCA TC-3′; reverse, 5′-CTT GGC AAG TCT GCA TTT GA-3′.
Immunoblot Detection of the GLP-1R/Exendin-4 Signaling Pathway.
Lysates from Huh7 and HepG2 cells were prepared after treatment with exendin-4 or GLP-1 for 5, 15, 30, 60, 90, 180, and 360 minutes. Equal amounts of protein were resolved on sodium dodecyl sulfate–polyacrylamide gel electrophoresis,17 transblotted, and subjected to immunodetection using the primary antibody for GLP-1R (ab39072 [1:500], Abcam), the phosphorylated and total species of 3-phosphoinositide-dependent kinase-1 (PDK-1), AKT, and protein kinase C ζ (PKC-ζ), β-Actin served as a loading control.17
Subcellular Fraction Analysis.
Huh7 cells were treated with exendin-4 for 30 minutes and 1 hour. Cells treated with preimmune serum served as controls. Cytosolic, membrane, and nuclear fractions (20 μg) were separated according to the manufacturer's instructions (Biovision #K270), resolved on sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and subjected to immunodetection against the GLP-1R antibody.
Cell surface expression assays were performed as described by Xu et al.18 Briefly, 35-mm3 collagen-coated dishes (BD#354459) were used to plate an equal number of Huh7 cells. The cells were treated with GLP-1 or exendin-4 for 4, 10, and 30 minutes. After cells were formalin-fixed, cells were treated with 3% nonfat dry milk in phosphate-buffered saline and subsequently incubated with primary antibody (anti-GLP-1R [1:500]) for 1 hour, followed by secondary antibody [1:1,000]. Antibody-receptor binding was detected using the Supersignal ELISA Pico enhanced chemiluminescence reagent (Pierce, Rockford, IL). The luminescence, which corresponds to the amount of receptor on the cell surface, was determined by way of a TD 20/20 luminometer (Turner Designs, Sunnyvale, CA). Control cells were treated with preimmune serum.
Immunofluorescence and Confocal Microscopy.
To visualize GLP-1R, Huh7 cells were grown on chamber slides and treated with GLP-1 or exendin-4 for 4 minutes, 15 minutes, 30 minutes, and 1 hour, and routine immunostaining was performed. Briefly, the cells were fixed with paraformaldehyde, permeabilized with 0.5% Triton X-100 + 0.08% saponin in H+ at 25° for 40 minutes and then incubated with 50 μL of rhodamine phalloidin diluted 1:60 at 25° for 45 minutes. Cells were blocked with 2% bovine serum albumin for 1 hour at 25°, followed by incubation with primary antibody (GLP-1R [1:200]) overnight at 4°C. After washing, cells were incubated with secondary antibody (anti-rabbit fluorescein isothiocyanate). Fluorescence and confocal microscopy were performed.
Oil Red O Staining and TG Assay.
Huh7 and HepG2 cells were exposed to medium containing 1% free fatty acid–free bovine serum albumin, and fat-loaded with 400 μM of palmitic and oleic acid (Sigma). After 12 hours, cells were treated with 20 nM of exendin-4 for 6 hours and stained with Oil Red O (Polyscience, Niles, IL) to visualize TG accumulation. TG quantification assay was performed according to the manufacturer's instructions (Biovision #K622-100). These experiments were conducted in the absence of insulin.
Methionine-Choline–Deficient Media Treatment and Flow Cytometric Analysis.
After serum starvation for 24 hours, HepG2 cells were exposed to either control media or methionine-choline–deficient medium (Gibco) supplemented with 10% fetal bovine serum as described.19 These experiments were also conducted under insulin-free conditions. Cells were treated with exendin-4 for up to 24 hours and stained with Nile Red (MP Biomedical, Solon, Ohio) at a concentration of 0.5 μg/mL and incubated for 15 minutes at 37°C as described.20 Flow cytometry was performed. Briefly, cells were resuspended in phosphate-buffered saline plus 0.5% bovine serum albumin and analyzed on a FACSCalibur flow cytometer (Becton Dickinson, San Jose, CA) on the FL3 channel (670LP filter with 488 nm excitation) and FL4 channel (660/20 filter with 633 nm excitation).
siRNA Against GLP-1R.
Three primer pair sequences for siRNA–GLP-1R and negative control (Stealth Negative) were purchased from Invitrogen as shown in Table 1. Huh7 cells were transfected using Lipofectamine RNAiMAX reagent (Invitrogen) following the manufacturer's reverse transfection protocol. Cells were plated at 50% confluency and transfected with the siRNA sequences at 30 nM and maintained for 48 hours. GLP-1R knockdown was confirmed by way of immunoblot analysis. Cell lysates were prepared and subjected to immunoblot analysis for GLP-1R, PDK1, AKT, and PKC-ζ.
Table 1. Primer Pairs for siRNA Experiments
All data are presented as the mean ± standard error (SE). Statistical analysis was performed using Graphpad Instat 3 software (http://www.graphpad.com). Groups were compared using parametric tests (paired Student t test or one-way analysis of variance with posttest following statistical standards). P < 0.05 was considered statistically significant.
Confirmation of GLP-1R on Huh7 Cells and Human Hepatocytes.
Western blot analysis revealed the presence of GLP-1R in Huh7 cells and primary human hepatocytes (Fig. 1A).
Bioluminescence Confirms Specific GLP-1R Localization.
As shown in Fig. 1B, there was a multifold increase of GLP-1R in Huh7 cells compared with preimmune serum-treated controls (P < 0.05).
GLP-1R Is Internalized in Huh7 Cells.
GLP-1R is internalized on stimulation by GLP-1 or exendin-4 (Fig. 2). This was first demonstrated by way of cell surface expression analysis (bioluminescence assay) (Fig. 2A). We then confirmed the microscopic findings by way of subcellular fractionation (Fig. 2B). This demonstrated that following GLP-1R exposure to its agonist, the membrane-bound fraction was reduced. Upon stimulation with either GLP-1 or exendin-4, there was a decrease in the amount of receptor seen on the cell membrane under confocal microcopy (Fig. 2C). These data suggest that there is loss of the receptor from the cell membrane.
Fluorescence and Confocal Microscopy Confirm Bioluminescence and Subcellular Fractionation Results.
Both confocal and fluorescent imaging confirmed that GLP-1R is internalized. Fig. 2C (left panel) shows untreated cells in which GLP-1R (in green) is seen lining the cell membrane. On treatment with GLP-1 or exendin-4, the receptor (Fig. 2C, right panel) was detected primarily in the cytoplasm rather than on the plasma membrane (yellow arrows). These data support the detection of internalization of the receptor by way of bioluminescence assay, which was also confirmed by subcellular fractionation analysis.
Exendin-4 Reduces TG Stores in Huh7 and HepG2 Cells.
To determine whether a physiologic endpoint of putative GLP-1 receptor signaling could be achieved, we used several approaches to explore whether there was a significant reduction in the cellular TG content following exendin-4 treatment. As seen on Oil Red O staining (Fig. 3A), following engorgement of Huh7 cells with palmitate and oleate, exendin-4 greatly reduced TG stores; this was further corroborated by TG quantitation (Fig. 3B). The reduction in cellular lipid content (both neutral and polar lipids) by exendin-4 was also confirmed using flow cytometry, with Nile Red staining in cells rendered steatotic by methionine-choline–deficient media (Fig. 3C).
Exendin-4 Increases Phosphorylation of PDK-1, AKT, and PKC-ζ Proteins.
Exendin-4 resulted in a significant increase in phosphorylation at 60 minutes of PDK-1, and AKT (Fig. 4) (P < 0.05,). The phosphorylation of PKC-ζ was significantly increased at 30, 60, and 90 minutes (P < 0.05) (Fig. 4). siRNA against GLP-1R (Supporting Fig. 1) was used to abolish effects seen in Huh7 cells treated with exendin-4. The knockdown of GLP-1R abolished the effects for PDK-1 and PKC-ζ (P < 0.05 [n = 3]) (Fig. 5), but not AKT (data not shown).
A key problem facing biologists and clinicians is a plausible molecular basis for metabolic syndrome and its hepatic complications. It is widely believed that NAFLD is a component of this epidemic and is the most common reason patients see gastroenterologists in developed countries. Although we have published intriguing findings in which the long-acting GLP-1 agonist, exendin-4, significantly reduced hepatic TG stores in the livers of ob/ob mice, we did not provide a molecular mechanism for how GLP-1 proteins mediate this beneficial effect.14 Furthermore, there was a lack of evidence—particularly with regard to human liver—as to whether GLP-1Rs are present, specifically on hepatocytes, and whether they are biologically active, although a recent study demonstrated the presence of GLP-1R on cholangiocytes.21
In the present study, we provide a direct molecular explanation for the effects of GLP-1 or a long-acting homologue, exendin-4, in steatotic liver cells. Our data strongly suggest that as in other mammalian tissues, GLP-1R is present in human hepatocytes. These data are corroborated not only by conventional analysis (real-time polymerase chain reaction, immunoblotting) but also by bioluminescence, which also demonstrates internalization of GLP-1R. These data are supported by confocal microscopy and subcellular fractionation findings that suggest that the receptor is internalized. Studies are ongoing to directly measure ligand–receptor interactions, which we recognize gauge more specific properties than the antibody-receptor analyses in our study. On the other hand, the physiologic data indicating a direct reduction of cellular TG is a strong corollary to the receptor work in the present work.
GLP-1R is a member of the seven-transmembrane family of GPCRs,22 the signaling and functioning capabilities of which have been well defined. Widmann et al.3 have demonstrated that GLP-1R is internalized on stimulation with its agonist and recycles back to the plasma membrane after several hours following endocytosis. They have also reported that the receptor after endocytosis is partly internalized into an endosomal compartment such as endoplasmic reticulum, desensitized or recycled back to the plasma membrane.23 However, other target organelles for internalization cannot be excluded. Several mechanisms of internalization have been proposed, and β-arrestin-1 may be an important adapter protein for several GPCRs.24, 25 Sonoda et al.26 suggested a role of β-arrestin-1 in receptor signaling but not in trafficking, a hypothesis consistent with a report by Syme et al.27 Internalization was measured by the loss of surface expression of the receptor (as in our study) and has also been seen in response to partial GLP-1R agonists.28 Although the confocal data are convincing and corroborate with our blots from the membrane and nuclear fractions, the immunoblot data regarding transfer of GLP-1R to the cytoplasmic fraction is not as robust as visualized in the confocal data. Future work to clarify the internalization results will need to be performed but are beyond the scope of the present study. New techniques may be feasible; for example, self-labeling protein tags that are covalently linked with fluorophores and selectively label the specific pool of GPCRs present at the plasma membrane without labeling any of the internal pools. Thus, a nonpermeable labeled substrate will label only the plasma membrane–bound GPCR proteins. This selective labeling approach may significantly reduce the signal intensity obtained by the confocal microscopic examination of cells performed with exendin-4, as we have demonstrated here. Although we have demonstrated the hepatic GLP-1 receptor can be internalized, the data cannot quantify the degree to which exendin-4 induces this process. Although we recognize that much of the work performed in this study was in transformed malignant hepatocyte cell lines (primarily Huh7 cells), the identification of GLP-1R was also performed in primary human hepatocytes. We suspect that one of the reasons that some (but not all) previous studies have not identified GLP-1R is the availability of better quality antibodies against the receptor and both the purity and availability of viable human hepatocytes for in vitro experimentation.
These data are exciting from a clinical and translational perspective, because they offer a plausible explanation as to why GLP-1 or GLP-like proteins may be beneficial in the treatment of the metabolic syndrome and NAFLD in particular. Importantly, these data indicate a direct effect of GLP-1 protein, as opposed to an indirect or pleotropic effect. As has been recently reported, patients undergoing bariatric surgery are found to have higher circulating levels of GLP-1 with significant histological improvement in their livers,29-32 especially those who undergo ileal transposition.31 In the present study, we provide evidence for direct cellular effects of GLP-1 proteins by potentiating hepatocyte steatosis in vitro by supplementing Huh7 cells with palmitic and oleic acids and gauging the reduction of steatosis by Oil Red O staining and supportive TG quantification. Flow cytometric analysis demonstrated that methionine-choline–deficiency increased cellular neutral lipid content, which was significantly decreased by exendin-4 treatment. Interestingly, polar lipid content was only slightly altered by methionine-choline–deficiency but was dramatically decreased by exendin-4 treatment, which also warrants further investigation.
Treatment with exendin-4 at concentrations seen in either treated diabetic patients33 or at levels of GLP-1 seen in postbariatric surgery patients34, 35 results in decreased hepatic TG content. These data clearly underscore that GLP-1 has a direct, independent, and novel action on steatotic hepatocytes.
Our study also provides a molecular mechanism to explain the signal effectors of GLP-1 in its potential role in hepatocyte TG reduction. A key signaling effector for insulin signaling downstream from IRS-1 is AKT. Based on our data, we have outlined a proposed molecular pathway whereby GLP-1 or homologs intersect the insulin signaling pathway in hepatocytes (Fig. 6), because this and interrelated pathways in hepatocytes have emerged as critical for the molecular basis of the emergence of hepatocyte insulin resistance. It has been widely reported that AKT phosphorylation is markedly diminished in steatotic hepatocytes.36 In this study, we show that GLP-1 ligands increase not only the phosphorylation status of AKT but other key molecules downstream. Our signaling studies are noteworthy because they confirm that exendin-4 not only activated AKT, but also resulted in robust phosphorylation of both PDK-1 and PKC-ζ. However, we failed to knock down AKT phosphorylation by siGLP-1R, although we were successful in doing so against PKD-1 and PKC-ζ. These data provide a plausible mechanism by which exendin-4 may bypass AKT activation in patients with hepatic insulin resistance.
PDK-1 activates PKC-ζ; moreover, PKC-ζ appears to have a significant role in exendin-4–mediated lypolysis in rat adipocytes. Studies by Arnes et al.37 in the rat liver showed that GLP-1 significantly increased Glut 2 messenger RNA levels, increasing lipolysis. In addition, knockout studies of IRS-1 and IRS-2 in rat hepatocytes by Sajan et al.38 demonstrated that both appear to activate the AKT pathway, but that only IRS-2 appears to activate the PKC-ζ. Our data suggest that GLP-1R activates the same pathway as IRS-2, which may account for our failure to knock down AKT phosphorylation and our ability to significantly knock down PDK-1 and PKC-ζ phosphorylation. What is apparent from our data is that more than one pathway related to insulin signal transduction can act to execute an action of insulin, but in this case such an action (reduction in TG store in liver cells) was executed by GLP-1 proteins. The siRNA studies knocking out GLP-1R demonstrate a novel insulin action of GLP-1 proteins by up-regulating key elements of the hepatocyte insulin signaling pathway (Fig. 6).
Future cellular analysis should focus on GLP-1 proteins which serve as insulin sensitizing agents in hepatocytes as opposed to an incretin effect seen in pancreatic β cells. These cellular data may reveal a definitive role for a higher dose distribution of GLP -1 analogues to reduce hepatic steatosis—particularly in patients with type 2 diabetes mellitus—and raises the possibility that such agents may, in combination, be safely administered to reduce hepatic TG stores in NAFLD.