Program of Cardiovascular and Metabolic Disorders, Duke-NUS Graduate Medical School, Singapore
Sarah W. Stedman Nutrition and Metabolism Center, Departments of Medicine and Pharmacology and Cancer Biology, Duke University Medical Center, Durham, NC
Address reprint requests to: Paul M. Yen, M.D., Duke-NUS Graduate Medical School, Laboratory of Hormonal Regulation, CVMD Program, 8 College Road, Singapore 018987. E-mail: email@example.com; fax: +65-6516-7396.
Potential conflict of interest: Nothing to report.
This work was supported by Duke-NUS Graduate Medical School Faculty Funds (to P.M.Y.) sponsored by the Ministry of Health, Ministry of Education, and Ministry of Trade, Singapore, and A*StaR and a National Institutes of Health grant (PO1DK58398; to C.B.N., L.M. acknowledges “Formacion de Personal Investigador” (FPI) and the reference BES-2009-027637 in the Spanish Ministry project with reference SAF2011-23031.
Caffeine is one of the world's most consumed drugs. Recently, several studies showed that its consumption is associated with lower risk for nonalcoholic fatty liver disease (NAFLD), an obesity-related condition that recently has become the major cause of liver disease worldwide. Although caffeine is known to stimulate hepatic fat oxidation, its mechanism of action on lipid metabolism is still not clear. Here, we show that caffeine surprisingly is a potent stimulator of hepatic autophagic flux. Using genetic, pharmacological, and metabolomic approaches, we demonstrate that caffeine reduces intrahepatic lipid content and stimulates β-oxidation in hepatic cells and liver by an autophagy-lysosomal pathway. Furthermore, caffeine-induced autophagy involved down-regulation of mammalian target of rapamycin signaling and alteration in hepatic amino acids and sphingolipid levels. In mice fed a high-fat diet, caffeine markedly reduces hepatosteatosis and concomitantly increases autophagy and lipid uptake in lysosomes. Conclusion: These results provide novel insight into caffeine's lipolytic actions through autophagy in mammalian liver and its potential beneficial effects in NAFLD. (Hepatology 2014;59:1366-1380)
lipidation of microtubule-associated protein light-chain 3
magnetic resonance imaging
mammalian target of rapamycin
nonalcoholic fatty liver disease
normal chow diet
National Institutes of Health
Oil Red O
primary mouse hepatocytes
quantitative polymerase chain reaction
respiratory exchange ratio
red fluorescent protein
standard error of the mean
small interfering RNA
transmission electron microscope
tandem fluorescent-tagged LC3
Caffeine is one of the most widely consumed drugs in the world. Although its effect on whole-body metabolism and fat oxidation has been well documented in both animals and humans,[1-3] little is known about its direct action on the liver.
The liver is the major site for fatty acid oxidation (FAO) in mammals. Decreased turnover of hepatic lipid droplets can lead to the development of fatty liver disease in humans. Recently, the rapid rise in the prevalence of obesity and diabetes in the general population has contributed to a parallel increase in nonalcoholic fatty liver disease (NAFLD) in many parts of the world. Currently, it is estimated that up to 46% of the adult U.S. population may have hepatosteatosis. Presently, there are no effective drug therapies for NAFLD, currently considered a risk factor for type II diabetes. Recently, several studies have shown that caffeine intake in humans and animals is inversely correlated with severity of NAFLD and type II diabetes,[7-11] but the mechanism for this action is not known.
To gain insight into the association between caffeine consumption and NAFLD in humans, we studied the mechanism for caffeine induction of lipolysis in the liver. In this study, we focused on a recently discovered nonclassical pathway of lypolysis: the autophagy-lysosomal pathway.[12-14] The autophagic process involves sequestration of cytoplasmic contents in double-membrane autophagosomes (a process that embodies lipidation of microtubule-associated protein light-chain 3 [LC3]), followed by fusion with lysosomes to form autolysosomes, and subsequent degradation of the cytosolic components by acid proteases, hydrolases, and lipases within the autolysosomes.
Our results show that caffeine increases lipid droplet turnover, fat oxidation, and oxidative phosphorylation in hepatic cells by the autophagy-mediated pathway. Using both genetic and pharmacological inhibitors of autophagy, we directly link caffeine-induced autophagy with oxidative lipid metabolism both in vitro and in vivo. Furthermore, using a rodent model of NAFLD, we demonstrate that autophagy is associated with caffeine-induced hepatic fat clearance in vivo, thereby explaining the antisteatotic action of this widely consumed drug.
Materials and Methods
Caffeine and acridine orange (AO) were purchased from Sigma-Aldrich (St Louis, MO). [Correction added on March 13, 2014, after first online publication: previous sentence changed from ““;Caffeine, in the form of acridine orange (AO), was purchased”; to “;Caffeine and acridine orange (AO) were purchased”;.] Antibodies were from Cell Signaling Technology (Danvers, MA). Culture media and serum were purchased from Invitrogen (Madison, WI). Green fluorescent protein/red fluorescent protein (GFP-RFP)-LC3 tandem fluorescent-tagged (tf-LC3) and enhanced GFP-LC3 (plasmid 21073; Addgene, Cambridge, MA) plasmids were gifts from Prof. T. Yoshimori (Osaka University, Osaka, Japan).[16, 17]
Male mice (C57Bl/6) between ages 8 and 10 weeks were purchased from The Jackson Laboratory (Bar Harbor, ME). Studies were conducted in accord with the principles and procedures outlined in the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals and were approved by the institutional animal care and use committee at the Duke-National University of Singapore Graduate Medical School.
HepG2 cells maintained at 37°C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. For lipid treatment, an oleic acid (OA; product no.: O1008; Sigma-Aldrich) and palmitic acid (PA) combination was diluted in DMEM containing 2% (w/v) bovine serum albumin (product no.: A9576; Sigma-Aldrich). Primary mouse hepatocytes (PMHs) were isolated and cultured using standard protocols.
RNA Isolation and Real-Time Polymerase Chain Reaction
Total RNA was isolated and quantitative polymerase chain reaction (qPCR) performed using the QuantiTect SYBR Green PCR Kit (Qiagen, Valencia, CA) in accord with the manufacturer's instructions. Actin levels were taken for normalization, and fold change was calculated using 2-ddCt. Primer sequence is available on request.
AO, LysoTracker Red DND-99, and BODIPY 493/503 Staining
Cells were grown on glass coverslips and treated with caffeine for the required period. Thereafter, cells were incubated with either 1 μg/mL of AO (Sigma-Aldrich) or 100 nM of LysoTracker Red (Molecular Probes, Eugene, OR) for 15-30 minutes at 37°C, followed by three phosphate-buffered saline (PBS) washes, and then immediately observed under a fluorescence microscope. For double labeling, 4% formaldehyde-fixed cells were stained for 15 minuets at 1 μg/mL of BODIPY 493/503 (Invitrogen), washed with PBS three times, and observed using an LSM710 Carl Zeiss confocal microscope (Carl Zeiss AG, Jena, Germany).
Cells or tissue samples were lysed using mammalian lysis buffer (Sigma-Aldrich), and immunobloting was performed as per the manufacturer's guidelines (Bio-Rad, Hercules, CA). Densitometry analysis was performed using ImageJ software (NIH, Bethesda, MD).
Immunofluorescences (IFs) were performed according to standard protocols. For autophagic flux analysis, tandem RFP/GFP-tagged LC3 plasmid (a kind gift from Dr. T. Yoshimori, Osaka University) was transfected into HepG2 cells with Lipofectamine 2000 transfection reagent (Invitrogen). Image colocalization was done using ZEN software (Carl Zeiss AG).
Cells were seeded onto four-chambered coverglass (Lab-Tek Chambered Coverglass System; Nalgene-Nunc, Rochester, NY) at a density of 2 × 104 cells/mL (14,000 cells/well). Images were acquired using the Olympus EM208S transmission electron microscope (TEM; Olympus, Tokyo, Japan).
Autophagy Inhibition and Intracellular Fat Measurement In Vitro
ATG5 knockdown was carried out using Stealth small interfering RNA (siRNA) duplex oligoribonucleotides targeting human ATG5 (Invitrogen). After 24 hours, cells were subjected to caffeine (1.5 mM) and OA (0.5 mM) cotreatment, and changes in intracellular lipid content after 48 hours of treatment were assessed by measuring Oil Red O (ORO) absorbance at 520 nm. In brief, cells were washed once with PBS and fixed with 10% formaldehyde for 1 hour. After fixation, cells were washed twice with dH20 and incubated in 60% isopropanol for 5 minutes. Cells were dried and stained with ORO solution (working solution: 0.5 g of ORO powder dissolved in 60% isopropanol) for 10 minutes at room temperature. Cells were washed four times in dH20, images were acquired, and then cells were dried. To quantify ORO content levels, isopropanol was added to each sample, after shaking at room temperature for 10 minutes, followed by absorbance measurement at 520 nm on a spectrophotometer.
To assess the effect of late autophagic inhibition using chloroquine (CQ), cells were subjected to OA alone, OA + caffeine, or OA + caffeine + CQ for 72 hours and ORO absorbance at 520 nm was measured.
Acute Caffeine, CQ Administration, and Ketogenesis Measurements
Caffeine (30 mg/kg body weight [b.w.]) was injected intraperitoneally (IP) daily for 3 days in male mice (C57Bl/6) fed on a normal chow diet (NCD), and tissues were collected for western blotting, real-time PCR analysis, and lipidomics. To assess the effect of autophagy inhibition on caffeine-induced fat oxidation in vivo, CQ (60 mg/kg b.w. IP for 3 days) was coadministered with caffeine, and beta-hydroxybutyrate (β-HB) levels were assessed using a commercially available colorimetric kit (Cayman Chemical, Ann Arbor, MI).
High-Fat Diet and Chronic Caffeine Treatment
Male mice (C57Bl6) were fed an NCD or a high-fat diet (HFD) (D12492; Research Diets, Inc., New Brunswick, NJ) for 4 weeks, at which point half of the HFD-fed mice began receiving caffeine (0.05% w/v) in drinking water for the next 4 weeks. The b.w. of mice in each group was recorded every week until sacrifice. Fat mass was analyzed on the day of sacrifice using magnetic resonance imaging (MRI). Tissue was collected for electron microscopy, western blotting, real-time PCR analysis, histology, and triglyceride (TG) estimation.
Histological Analysis of Intrahepatic Lipid Content
Paraformaldehyde-fixed livers (4%) were sectioned at 10 μm in a cryostat. Sections were stained with ORO (Sigma-Aldrich) dissolved in 70% isopropyl alcohol.
Seahorse XF-24 Metabolic Flux Analysis
Oxygen consumption was measured at 37°C using an XF24 extracellular analyzer (Seahorse Bioscience Inc., North Billerica, MA). HepG2 cells (40,000) were seeded in 24-well plates and transected with either ATG5 or negative siRNA. Forty-eight hours after transfection, fat (0.5 mM of 2:1 OA and PA) was added with or without 1.5 mM of caffeine overnight. Cells were changed to unbuffered DMEM and incubated at 37°C in a non-CO2 incubator for 1 hour. All injection reagents were adjusted to pH 7.4 on the day of the assay. Every point represents an average from five different wells.
Acylcarnitines, amino acids, and sphingolipids were measured in liver extracts by previously described mass spectrometry (MS)-based methods([18, 19]). Nucleotides (nts) were measured by a method modified from a previously reported liquid chromatography/tandem MS method (please refer to the Supporting Materials). Analysis employed tandem MS with a Quattro Micro instrument (Waters Corporation, Milford, MA).
Body Composition and Indirect Calorimetry
Mice were weighed and their body composition was measured by using an EchoMRI-100 (Echo Medical Systems). Respiratory exchange ratio (RER) and food intake were simultaneously measured for each mouse after a 2-day acclimatization period by using the Oxymax/Comprehensive Lab Animal Monitoring System (Columbus Instruments, Columbus, OH).
Individual culture and animal experiments were performed in duplicate or triplicate and repeated three times using matched controls, and the data were pooled. Results are expressed as either standard deviation (SD) or mean ± standard error of the mean (SEM). The statistical significance of differences (P < 0.05) was assessed by t test.
Caffeine Induced Lipid Clearance Is Associated With a Coordinate Increase in Autophagy in Hepatic Cells
To investigate the role of autophagy in caffeine-induced lipid turnover, we treated HepG2 cells with OA and examined the relationship between lipid clearance and autophagosome formation. Caffeine-induced lipid clearance (Fig. 1A) was temporally coupled with an increase in LC3 expression without large differences in diacylglycerol acyltransferase 1 (DGAT1), adipose TG lipase (ATGL), or hepatic lipase (LIPC) levels (Fig. 1B). These results indicated that caffeine probably does not affect TG packaging or the classical lypolysis pathway in these cells. Next, punctate IF staining of ectopic GFP-LC3 (Fig. 1C,D) and endogenous LC3-II (Fig. 1E) confirmed coupling of autophagic induction and lipid clearance after caffeine treatment
Furthermore, we found that caffeine, at concentrations as low as 0.05 mM (Fig. 2A) and, at times, as early as 6 hours (Fig. 2B) induced autophagosome formation in HepG2 cells. Caffeine-induced autophagy in HepG2 was associated with increase in proautophagic proteins, such as ATG7, ATG5, Beclin, and down-regulation of mammalian target of rapamycin (mTOR; Fig. 2C). A similar increase in LC3-II levels and down-regulation of mTOR signaling was noted in primary mouse hepatocytes (Fig. 2D,E).
We employed four different methods to assess autophagic flux. First, tf-LC3 staining demonstrated autophagic flux. In this assay, GFP tagged to LC3 detects only autophagosomes, whereas RFP detects both autophagosomes and -lysosomes. The overlay of green and red fluorescence creates yellow dots and identifies autophagosomes, relative to autolysosomes (remaining red dots), in merged images. We observed that caffeine increased both autophagosome (yellow dots) and -lysosome (remaining red dots) formation in merged images (Fig. 3A). Second, we assessed the protein levels of p62 in cells after caffeine treatment. p62 accumulates when autophagy is inhibited and is decreased when there is autophagic flux. Immunoblotting of p62 in caffeine-treated cells showed significant reduction in its levels after 24-hour treatment, indicating increased autophagic flux (Fig. 3B). Third, we observed an increase in LC3-II levels in cells treated with caffeine, followed by CQ treatment versus caffeine alone, suggesting that caffeine indeed induced autophagic flux (Fig. 3C). Last, electron microscopic (EM) evaluation of caffeine-treated cells showed increased autophagic induction (Fig. 3D). Collectively, these results demonstrate strong proautophagic actions by caffeine.
Autophagy Mediates Caffeine-Induced Oxidative Metabolism in Hepatic Cells
To determine whether autophagy induced by caffeine is directly involved in reducing intracellular lipid content, we used ATG5 siRNA to block autophagosome formation and measured its effect on reduction of intracellular lipid by ORO colorimetric absorbance. ATG5 knockdown (KD) not only reduced LC3-II levels (Fig. 4A), but also significantly decreased caffeine-mediated reduction of intracellular lipid (Fig. 4B), strongly suggesting involvement of autophagy in reduction of intracellular lipid. Furthermore, ATG5 KD also abolished caffeine-induced increase in β-HB, suggesting that autophagy is essential for caffeine-induced fat oxidation and clearance (Fig. 4C,D). Because FAO leads to increased mitochondrial electron transport and oxygen consumption, we performed seahorse metabolic analysis in HepG2 cells loaded with fat in the presence or absence of caffeine. Our results showed that fat loading itself significantly diminishes spare respiratory activity, compared to non-fat-loaded control HepG2 cells (Supporting Fig. 1A,B). Surprisingly, caffeine treatment could rescue and increase the spare respiratory capacity in fat-loaded cells. However, this effect of caffeine was dependent on autophagy, because ATG5 KD significantly impaired caffeine-induced respiration (Supporting Fig. 1A,B), suggesting a direct link between autophagy and mitochondrial spare respiratory capacity. However, no significant changes were observed in adenosine triphosphate (ATP) production in either group.
Furthermore, colocalization of LC3 with BODIPY 493 (a lipophilic dye; Invitrogen) in 16-hour caffeine-treated cells confirmed a direct interaction between autophagosomes and intracellular lipids (Fig. 4E and Supporting Fig. 2A).
Next, we assessed lysosomal activity using the lysosomotropic agent, AO, which showed increase yellow-orange fluorescence characteristic of lysosomal acidification in caffeine-treated cells (Fig. 5A). Because lysosomes are the sites of autophagy-induced degradation of macromolecules, including lipids, we assessed the effects of lysosomal inhibition on reduction of intracellular lipid by caffeine in HepG2 cells. Similar to the effect of ATG5 KD, lysosomal inhibition by CQ also blocked lipid clearance by caffeine (Fig. 5B,C). Confocal microscopy also showed colocalization of lipid droplets within the lysosomal compartment (LysoTracker Red stained [Molecular Probes]; Fig. 5D and Supporting Fig 2B) after 16-hour caffeine treatment. EM images confirmed these findings by showing lipid deposits residing within the autolysosomes in caffeine-treated cells (Fig. 5E). Taken together, these findings demonstrate the involvement of the autophagy-lysosomal pathway in reduction of intracellular lipid by caffeine.
Caffeine Increases Hepatic Lipolysis and Overall Oxidative Metabolism In Vivo
To study the metabolic activity of caffeine in live animals, we examined the RER before and after caffeine treatment (30 mg/kg b.w. for 24 hours) in mice housed in metabolic cages. Caffeine treatment dramatically lowered dark-cycle RER (Supporting Fig. 3A). Caffeine showed a mild, but not significant, suppressive effect on food intake (Supporting Fig. 3B). These findings demonstrate an increase in FAO in vivo after caffeine treatment. We performed metabolic profiling of hepatic acylcarnitines in untreated and caffeine-treated mice (30 mg/kg b.w. for 3 days) and observed increases in short and medium acylcarnitines (SCACs and MCACs; Fig. 6A,B), consistent with increased flux through the β-oxidation pathway in the liver. Similarly, we observed increased long-chain and very-long-chain acylcarnitines (LCACs and VLCACs; Fig. 6C,D), suggesting increased hepatic lipolysis. Further support for increased oxidative flux came from nt metabolomics analysis that showed concomitant increases in hepatic nicotinamide adenine dinucleotide, ATP, adenosine monophosphate, and adenosine diphosphate levels (Fig. 6E and Supporting Fig. 4). These results suggest that an increased rate of ATP synthesis is linked to a higher rate of utilization, perhaps to provide energy for autophagic processes (Fig. 6E). Analysis of hepatic amino acid levels showed a general decrease in caffeine-treated mice (Supporting Fig. 5A). Sphingolipids were relatively unchanged, with the exception of sphinganine and sphingosine, which were elevated in caffeine-treated mouse liver (Supporting Fig. 5B-E).
Autophagy/Lysosomal Activity Mediates Caffeine-Induced Fat Oxidation In Vivo
To understand the contribution of autophagy in caffeine-induced hepatic β-oxidation in vivo, we examined its effect on hepatic autophagy. Lower hepatic levels of p62 in conjunction with increased LC3-II levels in caffeine-treated mice demonstrated increased autophagic flux (Fig. 7A,B). Additionally, caffeine-treated mice exhibited down-regulation of mTOR signaling showing reduced levels of p-mTOR, p-p70S6K, and p-4E-BP1 (Fig. 7A,B). We coadministered CQ in caffeine-treated mice and measured serum β-HB (ketogenesis endproduct) levels after 3 days of treatment. CQ is a well-known autophagy inhibitor used in vivo, and the purpose of using it in this study was to show that caffeine-induced β-oxidation in vivo requires autophagy. Caffeine treatment increased serum β-HB levels; however, in contrast to untreated mice, the caffeine-mediated increase in ketogenesis was abrogated after CQ administration (Fig. 7C,D), suggesting that the induction of fatty acid β-oxidation by caffeine was dependent upon the autophagy-lysosomal pathway. Of note, CQ had no suppressive effect on caffeine-induced acetyl coenzyme A carboxylase p-(ACC)/ACC protein ratio and carnitine palmitoyl transferase 1 alpha (Cpt-1α) protein level (Fig. 7E). These data suggest that the inhibitory effect of CQ on caffeine-induced β-oxidation occurs specifically through inhibition of autophagy and not by downstream pathways affecting β-oxidation. Therefore, similar to our results in vitro or in vivo, data also show that autophagy is essential for caffeine-stimulated β-oxidation.
Caffeine Induction of Hepatic Autophagy Is Associated With Decreased Hepatic Steatosis and Increased Lipid Accumulation in Autolysosomes
Given the epidemiological evidence linking caffeine intake and NAFLD,[10, 11] we investigated the effect of caffeine on autophagy and hepatosteatosis in an animal model of diet-induced obesity. Mice were fed with either an NCD or HFD for 4 weeks. Mice on HFD were then continued to feed on the same diet in the absence or presence of caffeine for another 4 weeks. No significant differences in food or water intake were observed in any of the animal groups (data not shown). At the end of the study, caffeine-treated mice inhibited HFD-induced weight gain (Fig. 8A). MRI measurements showed that caffeine-treated mice on HFD had significantly lower percent fat mass and higher percent lean mass than mice on HFD alone when normalized to their body weights (Fig. 8B). Hepatic lipid content, measured by an enzymatic kit as well as by ORO staining, was significantly increased in mice fed HFD, compared to mice fed NCD. Interestingly, caffeine-treated mice exhibited little or no intrahepatic lipid accumulation, even while on HFD (Fig. 8C,D). LC3-II immunoblotting showed significant induction of autophagy in caffeine-treated HFD mice (Fig. 8E). We also saw a concomitant up-regulation of lysosomal lipase and β-oxidative genes (Supporting Fig. 6A) in caffeine-treated HFD mice, compared to mice on HFD alone. These results reveal a combinatorial up-regulation of autophagic, lipolytic, and FAO pathways in the liver after chronic caffeine treatment. The loss of hepatic fat and increase in autophagy was evident on EM images that showed virtually no cytosolic lipid in liver sections of caffeine-treated mice on HFD, compared to mice on HFD alone (Fig. 8F and Supporting Fig. 6B,C). Instead, lipid was found almost exclusively within the autolysosomal compartment and often was surrounded by double membranes, suggestive of engulfed lipid-laden autophagosomes. Collectively, our results strongly suggest that the inverse clinical association between caffeine intake and NAFLD progression is the result of caffeine's proautophagic action and subsequent reduction of intrahepatic lipid through stimulation of lipid oxidation.
Here, we have identified and characterized a novel mechanism of coordinated lipid mobilization and metabolism in mammalian liver mediated by caffeine-induced autophagy. In particular, we showed that caffeine effectively decreases intracellular lipids and increases autophagic flux in several human hepatic cells and in vivo. The lowest dose at which we detected induction of autophagy in vitro was 50 μM, which is similar to the maximum plasma concentration of 46 μM measured after drinking four cups of coffee or the caffeine equivalent of 5 mg/kg b.w. Likewise, the dose of caffeine used in mice (30 mg/kg b.w.), roughly converted to approximately 2.43 mg/kg in humans, is equivalent to the amount of caffeine found in two to three cups of coffee. Therefore, our observations of caffeine-induced hepatic autophagy occurred at serum caffeine concentrations achieved by moderate-to-heavy coffee drinkers.
We used genetic (ATG5 siRNA) and pharmacological (CQ) inhibitors of autophagic and lysosomal function to demonstrate that the autophagy-lysosomal pathway is essential for reduction of intracellular fat and stimulation of oxidative phosphorylation by caffeine (Figs. 4 and 5). These results were further corroborated by electron micrographs showing colocalized lipid within the autophagosomal and lysosomal compartments, indicating ingestion of cytosolic lipids by autophagosomes and their subsequent delivery to lysosomes. Although our data suggest direct ingestion of lipid droplets within autophagosomes as a likely cause of lipid degradation by caffeine-induced autophagy, we cannot rule out other indirect effects of autophagy that may promote lipid clearance. To demonstrate that autophagy is essential for FAO in vivo, we treated mice with caffeine in the presence or absence of CQ to establish a direct link between caffeine-induced autophagy and hepatic fatty acid β-oxidation (Fig. 7). Taken together, these studies demonstrate the importance of autophagy and lysosomal degradation in promoting fatty acid β-oxidation and maintaining intrahepatic lipid homeostasis.
In this study, we employed multiple metabolic approaches to understand caffeine's action on lipid, nt, and amino acid metabolism in vivo. Metabolomic analyses of hepatic acylcarnitines and nts from mice treated with caffeine supported the notion that caffeine increases delivery and flow of fatty acids to the mitochondria and promotes fatty acid β-oxidation pathway and oxidative phosphorylation (Fig. 6). Caffeine also had suppressive effects on intrahepatic amino acid levels, which play an important role in autophagy regulation in the liver. Along with amino acids, sphingolipids have also been implicated in autophagy regulation. Our results show that caffeine decreased levels of ceramide and increased sphingosine levels.
Caffeine induction of autophagy, both in vitro and in vivo, involved inhibition of mTOR signaling (Fig. 5). Caffeine previously was shown to be a direct inhibitor of TOR signaling in yeast[28, 29] and mammalian cells, and its ability to inhibit mTOR activity, both in vitro and in vivo, is most likely the major mechanism for autophagy induction. Although autophagic induction by caffeine, in some cells, is associated with mitochondrial depolarization and cell death, at the dose we used in our experiments on HepG2 cells and primary hepatocytes we did not observe these effects (data not shown).
In summary, we have demonstrated that caffeine has a potent effect in lowering levels of hepatic lipids by activation of autophagy in cell culture and in vivo. This mobilization and hydrolysis of TGs to free fatty acids (FFAs) through the autophagy-lysosomal pathway led to increased delivery of FFAs to the mitochondria. This, in turn, increased the acylcarnitine flux, β-oxidation of fatty acids, and oxidative phosphorylation (Supporting Fig. 7). These findings are especially noteworthy, because it is possible that pharmacological[31, 32] or hormonal stimulation of autophagy may be useful therapeutic strategies for liver diseases such as NAFLD.[33-35] Because there still are no approved drug therapies for NAFLD, understanding the mechanistic basis of action of natural dietary products such as caffeine offers further insight into developing drugs for the prevention and treatment of NAFLD.