Inhibition of mitochondrial function by efavirenz increases lipid content in hepatic cells

Inhibition of Mitochondrial Function by EfavirenzIncreases Lipid Content in Hepatic Cells

Ana Blas-Garcıa,1,2 Nadezda Apostolova,1 Daniel Ballesteros,1 Daniel Monleon,3 Jose M. Morales,4

Milagros Rocha,1,2 Victor M. Victor,1,2 and Juan V. Esplugues1,2

Efavirenz (EFV) is a non-nucleoside reverse transcriptase inhibitor (NNRTI) widely usedin human immunodeficiency virus (HIV) infection therapy. It has been associated withhepatotoxic effects and alterations in lipid and body fat composition. Given the impor-tance of the liver in lipid regulation, we have evaluated the effects of clinically used con-centrations of EFV on the mitochondria and lipid metabolism of human hepatic cellsin vitro. Mitochondrial function was rapidly undermined by EFV to an extent that variedwith the concentration employed; in particular, respiration and intracellular adenosine tri-phosphate (ATP) levels were reduced whereas reactive oxygen species (ROS) productionincreased. Results in isolated mitochondria suggest that the mechanism responsible forthese actions was a specific inhibition of complex I of the respiratory chain. The reductionin energy production triggered a compensatory mechanism mediated by the enzyme aden-osine monophosphate–activated protein kinase (AMPK), the master switch of cellular bio-energetics. Fluorescence and nuclear magnetic resonance demonstrated a rapidintracellular increase of neutral lipids, usually in the form of droplets. This was preventedby the AMPK inhibitor compound C and by removal of fatty acids from the culture me-dium. These effects were not reproduced by Nevirapine, another NNRTI. EFV is clinicallycoadministered with two nucleoside reverse transcriptase inhibitors. Evaluation of one ofthe most common combination, EFV/Lamivudine/Abacavir, revealed that the effects ofEFV on ROS production were enhanced. Conclusion: Clinical concentrations of EFVinduce bioenergetic stress in hepatic cells by acutely inhibiting mitochondrial function.This new mechanism of mitochondrial interference leads to an accumulation of lipids inthe cytoplasm that is mediated by activation of AMPK. (HEPATOLOGY 2010;00:000-000)

Continuous administration of the drugsincluded under the term highly active antire-troviral therapy has made acquired immune

deficiency syndrome a chronic rather than terminal ill-ness. The initial development of these drugs was par-ticularly rapid and focused on clinical efficacy—reduc-tion of mortality—before all other considerations.

However, as the disease has come under control, therehas been a growing emphasis on the long-term adverseeffects induced by this therapy. Efavirenz (EFV) is anon-nucleoside reverse transcriptase inhibitor(NNRTI) widely used in initial therapy for human im-munodeficiency virus (HIV) infection. It is adminis-tered to adults in a single daily dose of 600 mg, which

Abbreviations: 3TC, lamivudine; ABC, abacavir; AMPK, adenosine monophosphate–activated protein kinase; ATP, adenosine triphosphate; DCFH-DA, 20,70-dichlorodihydrofluorescein diacetate; EFV, Efavirenz; HIV, human immunodeficiency virus; HR-MAS, high resolution magic angle spectroscopy; NNRTIs, non-nucleoside reverse transcriptase inhibitors; NRTI, nucleoside reverse transcriptase inhibitors; NVP, nevirapine; P-AMPK, phosphorylated adenosine monophosphate–activated protein kinase; ROS, reactive oxygen species; SEM, standard error of the mean.From the 1Departamento de Farmacologıa, Facultad de Medicina, Universidad de Valencia-CIBERehd; 2Fundacion Hospital Universitario Dr. Peset; 3Fundacion

Investigacion Hospital Clınico Universitario; and 4Unidad Central Investigacion, Facultad de Medicina, Universidad de Valencia, Valencia, Spain.Received July 2, 2009; accepted February 23, 2010.Supported by grants PI081325 and PI070091 from FIS (Fondo de Investigacion Sanitaria, Spain), grant ACOMP09/009 from Generalitat Valenciana, and by

an unrestricted grant from Abbott Laboratories. V.M.V. and M.R. are recipients of FIS contracts (CP07/00171 and CA07/00366, respectively), and D.M. is arecipient of a Ramon y Cajal contract.Address reprint requests to: Juan V. Esplugues, Ph.D., M.D., Departamento de Farmacologıa, Facultad de Medicina, Universidad de Valencia. Avda. Blasco

Ibanez 15-17, 46010 Valencia, Spain. E-mail: [email protected]; fax: 34-963983879.Copyright VC 2010 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.23647Potential conflict of interest: Nothing to report.

1

leads to therapeutic plasma concentrations of up to3.17 to 12.67 lM.1,2 Current practice guidelines rec-ommend the use of EFV with two nucleoside reversetranscriptase inhibitors (NRTIs). More than 20 poten-tial combinations exist, and coadministration withLamivudine (3TC) and Abacavir (ABC) is one of themost common.3 Although considered to be a safedrug, EFV-based regimens have been associated withlipid disturbances,4-6 psychiatric symptoms, and hepa-totoxicity.7-9 Studies of the clinical manifestations ofthese effects have revealed that some forms of thesetoxicities resemble disorders induced by mitochondrialdysfunction,10 but their molecular and cellular mecha-nisms remain largely unknown. Mitochondrial dys-function is considered an important mechanism ofdrug-induced liver injury, and takes place throughpathways such as the inhibition of the respiratorychain, fatty acid beta-oxidation, and increased produc-tion of reactive oxygen species (ROS). These condi-tions may favor the appearance of nonalcoholic fattyliver disease and, in inflammatory conditions, nonalco-holic steatohepatitis.11 The current study shows thatclinical concentrations of EFV induce a condition ofbioenergetic stress in hepatic cells by inhibiting mito-chondrial function through an acute mechanism thatis independent of mitochondrial DNA replication.This leads to the accumulation of lipids in the cyto-plasm through a mechanism mediated by activation ofAMPK. Coadministration of EFV with 3TC and ABCmodifies some of the mitochondrial effects of EFV.

Materials and Methods

General, Reagents, and Drugs. Unless stated other-wise, experiments were performed in human hepato-blastoma Hep3B cells (ATCC HB-8064), cultured asdescribed previously.12 All reagents used for culturewere obtained from Gibco (Invitrogen, Carlsbad, CA).Clinically available preparations of the NNRTI EFV(Sustiva 600 mg, Bristol-Myers Squibb, Princeton, NJ)and Nevirapine (NVP, Viramune 200 mg, BoehringerIngelheim, Ingelheim, Germany) were dissolved inmethanol (3 mg/mL) or HCl 0.06 M (1 mg/mL),respectively, and insoluble substances were removed byfiltration. The purity and stability of these solutions(98%-100%) were evaluated by high-pressure liquidchromatography and compared with those of controlsolutions (Sequoia Research Products, Pangbourne,UK). ABC and 3TC (Sequoia Research Products) weredissolved in water. In control experiments, cells weretreated with maximal amounts of 5.3 lL/mL methanol

or 13.33 lL/mL HCl 0.06 M. Fluorescent probeswere acquired from Molecular Probes (Invitrogen,Carlsbad, CA), except for Hoechst 33342, which wassupplied by Sigma-Aldrich Chemicals (Steinheim, Ger-many), as were all the remaining chemicals.Animal studies were in accordance with institutional

guidelines for the care and use of laboratory animals.Male Sprague-Dawley rats were supplied by CharlesRiver Laboratories (Barcelona, Spain). Human liver tis-sue was obtained from biopsies from patients (3women, 4 men) that had undergone surgical resectionof liver tumors (Hospital ‘‘La Fe,’’ Valencia, Spain).Experiments were approved by the local ethicscommittee.Electrochemical Measurement of Oxygen (O2)

Consumption. Cells (3-5 � 106) or liver samples (20-35 mg) were placed in gas-tight chambers containing1 mL respiration buffer (Hank’s balanced salt solutionfor Hep3B and Krebs solution for human tissue) andagitated at 37�C. O2 consumption was evaluated witha Clark-type O2 electrode (Rank Brothers, Bottisham,UK) as described previously,13,14 and measurementswere recorded using the Duo.18 data acquisition de-vice (WPI, Stevenage, UK). Recordings were initiatedimmediately after addition of EFV (5-100 lM), NVP(10-50 lM), or their respective solvents. To study theeffect of prolonged exposure, some cells were treated withEFV (10 lM) for 4 hours before evaluating their O2 con-sumption. To assess the potential reversibility of EFV-induced inhibition, Hep3B cells were incubated for 1hour with EFV 25 lM. The EFV-containing mediumwas subsequently removed, and the cells were incubatedfor a further 1 hour before analyzing O2 consumption.Previous experiments demonstrated that O2 consump-tion was not modified by the solvents employed withEFV and NVP. Rotenone (10 lM) and sodium cyanide(1 mM), respective inhibitors of complex I and IV of theelectron transport chain, were employed as positive con-trols and to confirm the mitochondrial origin of O2 con-sumption (95%-99%). In several experiments, the low-est concentration of EFV (10 lM) was coadministeredwith ABC and 3TC at concentrations (10 lM) similarto those clinically present.Isolation of Liver Mitochondria and Analysis of

Complexes in Mitochondria O2 Consumption. Livermitochondria were obtained from fresh rat livers,15

and their O2 consumption in 1 mL of incubationbuffer (145 mM KCl, 30 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 5 mM KH2PO4,3 mM MgCl2, 0.1 mM ethylene glycogen tetra-aceticacid, 0.1% albumin, pH 7.4) was evaluated as described.Complex I-linked (2.5 mM glutamate/2.5 mM malate)

2 BLAS-GARCIA ET AL. HEPATOLOGY, Month 2010

or complex II-linked (5 mM succinate/2 lM rotenone)were employed as substrates. Assays were performed inthe absence (state 4—resting) and presence (state 3—phosphorylation) of 500 lM adenosine diphosphate.Mitochondrial proteins were measured employing thebicinchoninic acid (BCA) protein assay kit (PierceChemicals, Boulder, CO).ROS Production, Intracellular Adenosine Tripho-

phate Concentration and Glucose Uptake. ROS pro-duction was analyzed in cells seeded in a black 96-wellplate.13 The fluorescent probe DCFH-DA (20,70-dichlorodihydrofluorescein diacetate, 2.5 lM) wasadded for 30 minutes, cells were washed with Hank’sbalanced salt solution before addition of EFV, NVP(10-50 lM), or the combination of EFVþ3TCþABC(10 lM each one), and fluorescence was detected at 5-minute intervals over a 1-hour period using a Fluoros-kan (Thermo Labsystems, Thermo Scientific, Rock-ford, IL). High concentrations of rotenone (100 lM)or exogenous hydrogen peroxide (H2O2, 100 lM)were used as a positive control. The adenosine triphos-phate (ATP) concentration (nmol/mg protein) in cellsincubated (1 hour) with EFV, NVP (10-50 lM), or acombination of EFVþ3TCþABC (10 lM each one)was determined using an ATP Bioluminescence AssayKit HSII (Roche, Mannheim, Germany) and a Fluo-roskan microplate reader. Protein concentrations weredetermined with the BCA protein assay kit. Glucoseconcentration in the cell medium incubated with EFV(10 and 25 lM, 4 hours) was detected spectrophoto-metrically as an indicator of cellular glucose uptake. Inthis assay, NAD was reduced to reduced nicotinamideadenine dinucleotide phosphate after several coupledreactions. Its absorbance was measured at 340 nm andwas equivalent to the glucose concentration of the sam-ple. Reactions were performed in a 96-well plate usinga Multiskan plate-reader spectrophotometer (ThermoLabsystems, Thermo Scientific, Rockford, IL).Western Blotting. Cells were incubated with EFV

(10-50 lM, 1, 4, or 8 hours), NVP (10-50 lM,4 hours), or rotenone (10 lM, 4 hours) and were sub-sequently collected in ice-cold phosphate-buffered sa-line and centrifuged. Total protein extracts wereobtained by lysing pellets with PhosphoSafe ExtractionReagent (Novagen, Calbiochem, La Jolla, CA) and aprotease-inhibitor cocktail (Roche, Mannheim, Ger-many). Human liver samples (20-35 mg) incubatedwith EFV (25 lM, 4 hours) were treated with anextraction buffer (Tris-HCl 66 mM pH 7.5, ethyleneglycogen tetra-acetic acid 1 mM, Na O-Vanadate 1mM, NaF 1 mM, and protease inhibitor 10�) andthen homogenized, treated with 10% NP-40, soni-

cated, and centrifuged to obtain the final proteinextracts. Protein concentrations were determined withthe bicinchoninic acid (BCA) protein assay kit. A25-lg protein sample was resolved by sodium dodecylsulfate polyacrylamide gel electrophoresis, transferredto nitrocellulose membranes, and probed with anti-AMPK alpha 1þalpha 2 (phospho T172) polyclonalAb, anti-GLUT1 (glucose transporter 1) polyclonal Ab(Abcam, Cambridge, UK), and anti-actin polylonal Ab(Sigma-Aldrich, Steinheim, Germany). Horseradishperoxidase–coupled secondary antibody (peroxidase-labeled anti-rabbit immunoglobulin G, Vector Labo-ratories, Peterborough, UK) was detected using theenhanced chemiluminescence advanced system (Amer-sham Pharmacia Biotech, UK) or SuperSignal West-Femto (Pierce Chemicals, Boulder, CO) and visualizedwith a digital luminescent image analyzer (FUJIFILMLAS3000, Fujifilm, Japan). Densitometric analyseswere performed using ImageQuant software V4.0.Analysis of Cellular Lipids. Cells were incubated

for 24 hours with EFV or NVP (10-50 lM) in culturemedium supplemented with 1% bovine serum albu-min fatty acid free, 50 lM L-carnitine, and 0.1 mMpalmitic acid. Nuclei were stained with 1 lM Hoechstfor the last 30 minutes of treatment, after which cellswere washed with phosphate-buffered saline and lipiddroplets were stained with 0.5 lM Nile red in Hank’sbalanced salt solution at room temperature for 10minutes.16 Fluorescence was detected using an invertedmicroscope (IX81, Olympus, Hamburg, Germany)and ScanR Acquisition or Analysis software for staticcytometry.High-resolution magic angle spectroscopy (HR-

MAS spectroscopy) was employed as described previ-ously17 to evaluate the lipid profile of cells incubatedwith EFV (10 or 25 lM, 4 hours) in the same me-dium, with and without palmitic acid (0.1 mM). Insome experiments, the AMPK inhibitor compoundC (6-[4-(2-Piperidin-1-ylethoxy)-phenyl)]-3-pyridin-4-ylpyrazolo[1,5-a] pyrimidine, Calbiochem, La Jolla,CA) was added 30 minutes before EFV and main-tained throughout the 4-hour incubation period. Cellswere subsequently centrifuged for 5 minutes at 5000rpm. Forty microliters of the resulting pellet wereintroduced into a 4-mm ZrO2 rotor fitted with a50 lL cylindrical insert, and D2O (approximately10 lL) was added to the sample for field locking pur-poses. The rotor was then transferred to the NMRprobe, which had been cooled at 10�C to minimizesample degradation.17 The entire HR-MAS study wasperformed at this temperature, having been initiatedwhen the temperature inside the probe reached the

HEPATOLOGY, Vol. 000, No. 000, 2010 BLAS-GARCIA ET AL. 3

equilibrium condition (approximately 5 minutes). ABruker Cooling Unit controlled the temperature bycooling the bearing air flowing into the probe. TheHR-MAS spectra were recorded on a Bruker Avance600 spectrometer operating at a frequency of 600.13MHz and equipped with a 4-mm triple-resonanceHR-MAS probe. Samples were spun for 15 minutes at5000 Hz to keep the rotation sidebands out of the ac-quisition window, and one-dimensional proton spectrawith water pre-saturation were acquired for each sam-ple. Data were processed using the spectrometer soft-ware Topspin 1.3 (Bruker Biospin GmbH, Germany).The peak areas were calculated by deconvolution ofthe region of interest with in-house MATLAB soft-ware. Peaks were fitted to a Voight-shape, and calcu-lated areas were normalized with respect to globalspectral intensity.Data Analysis. Values are mean 6 standard error

of the mean (SEM) of 3-8 experiments. Statisticalanalysis was performed by one-way analysis of variancefollowed by a Newman-Keuls test for unpaired samples(Graph Pad Software V3.02, La Jolla, CA). Signifi-cance was *P < 0.05, **P < 0.01, and ***P < 0.001.

Results

Electrochemical Measurement of O2 Consump-tion. Figure 1A shows representative traces of respiringHep3B cells in control conditions and the acute inhib-itory effect of EFV (10 and 25 lM) on the rate of O2

consumption after its addition to the gas-tight cham-bers, illustrated by the slope of the curve. Figure1Brepresents the concentration-dependent reduction inO2 consumption produced by EFV (5-100 lM). Themaximal inhibitory effect of EFV was obtained with50 lM and did not differ from that of rotenone 10lM (31.25% 6 5.55% of control, n ¼ 3, P <0.001). Doubling the concentration of EFV to 100lM did not increase the inhibition of respiration.Unless stated otherwise, 15, 25, and 50 lM of EFVwere used in all remaining experiments. Incubation for4 hours did not augment the inhibitory effect of EFV(10 lM), because levels of O2 consumption were simi-lar to those following acute administration of the drug(n ¼ 3, 69.30% 6 3.04% versus 73.45% 6 7.02%,respectively). Respiration of Hep3B cells was restoredafter removal of EFV from the medium, which sug-gests that the effects observed were reversible andrelated to the presence of drug (67.31% 6 8.60% ver-sus 110.13% 6 7.34% without or with removal of25 lM of EFV; each value calculated versus its respec-tive control, n ¼ 5). Furthermore, the effects were spe-

cific for EFV, because the presence of NVP (10, 25,and 50 lM) in the gas-tight chambers did not modifythe rate of O2 consumption by Hep3B cells (Fig. 1C).EFV Impairs Mitochondrial O2 Consumption by

Inhibiting Complex I. Figure 2 shows the effects ofEFV on mitochondrial complex I activity in isolatedmitochondria of rats. Figure 2A, B, C (traces), and D(rate of O2 consumption) illustrate that the mitochon-dria incubated with EFV respired poorly in the pres-ence of complex I substrates malate and glutamate.Indeed, the inhibition of respiration with EFV 25 and50 lM did not differ from that induced by the specificcomplex I inhibitor rotenone (2 lM). The inhibitoryeffects of EFV were absent when succinate (5 mM), acomplex II electron donor, was added to bypass com-plex I–dependent respiration. The mitochondria

Fig. 1. Efavirenz (EFV) acutely inhibits O2 consumption in Hep3Bcells. (A) Representative traces showing the rate of O2 consumption inthe absence (solvent, solid line) or presence of EFV (10 and 25 lM)in a Clark-type O2 electrode. (B) Inhibitory effects of various concen-trations of EFV. (C) Effects of solvent (white bar) or various concentra-tions of Nevirapine (NVP). Data represent the mean 6 SEM ofbetween three and five experiments. **P < 0.01, ***P < 0.001 ver-sus solvent.


exhibited O2 consumption rates similar to those ofcontrols, thus suggesting that complex I was the maintarget of EFV.Parameters of Mitochondrial Function. Incubation

with EFV produced a significant and concentration-de-pendent increase in the fluorescence of DCFH-DA,indicating an augmented production of ROS (Fig.

3A). This effect was rapid, was maintained throughoutthe 1-hour period evaluated, and was not reproducedwhen cells were treated with NVP (Fig. 3B). Figure3C shows that 1 hour incubation with EFV induced asignificant and concentration-dependent decrease in in-tracellular ATP, whereas incubation with NVP did notalter ATP levels (Fig. 3D).

Fig. 2. EFV inhibition of cellular O2 con-sumption is attributable to a decrease incomplex I–dependent respiration. (A) Repre-sentative traces showing the rate of O2 con-sumption by isolated mitochondria from ratliver in the absence (solvent) or presence ofvarious concentrations of EFV, or (B) in thepresence of complex I substrates malate (2.5mM) and glutamate (2.5 mM). (C) When com-plex II substrate succinate (5 mM þ 2 lM ro-tenone) was added to bypass complex I–dependent respiration, the inhibitory effects ofEFV were absent, and mitochondria exhibitedO2 consumption rates similar to those of con-trols, thus suggesting that complex I was themain target of EFV. (D) O2 consumption rate inthe different conditions described (D). Datarepresent the mean 6 SEM of four experi-ments. Differences between basal O2 con-sumption (without addition of substrates) ineach EFV treatment group versus the respec-tive solvent condition are shown as **P <0.01 and ***P < 0.001. Differences betweenO2 consumption after addition of malate andglutamate in each EFV treatment group versusthe respective solvent condition are shown as#P < 0.05 and ###P < 0.001.

Fig. 3. Short-term (1 hour) exposure toEFV induces an increase in ROS produc-tion and decrease in intracellular ATP lev-els in Hep3B cells. (A) Effects of EFV and(B) NVP on total ROS production assessedafter incubation (30 minutes) with the flu-orescent probe 20,70-dichlorodihydrofluor-escein diacetate (DCFH-DA, 2.5 lM). Theeffects of the solvent are shown in thewhite bar. Measurements were obtainedduring a 1-hour period (n ¼ 8 and n ¼4, respectively). (C) Effects of EFV and (D)NVP on intracellular ATP levels. Data aremean 6 SEM of at least five experiments.**P < 0.01, ***P < 0.001 versussolvent.


Activation of AMPK. Figure 4 shows representativewestern blot analysis of phosphorylated AMPK (P-AMPK), the active form of the enzyme, of cells incu-bated with EFV. Densitometric analysis revealed thatEFV induced a concentration-dependent and time-de-pendent increase in P-AMPK. EFV 50 lM produceda significant phosphorylation of AMPK during a 1-hour incubation period. The effects of EFV 25 lMwere statistically significant after 4 hours, but not after1 hour, whereas those of 10 lM reached significanceonly after 8 hours (Fig. 4A, B, and C). The levels ofP-AMPK were also significantly (P < 0.01) increasedby incubation (4 hours) with rotenone (10 lM,281.28% 6 53.59% of control, n ¼ 4) but not withNVP (Fig. 4D). Incubation (4 hours) with EFV 10 or25 lM did not modify the concentration of glucose inthe medium (114.10% 6 24.04% and 105.74% 621.58% of control, n ¼ 3) or the expression ofGLUT-1 (111.70% 6 19.95% and 136.40% 6

37.20% of control, n ¼ 9), which suggests that theglucose metabolism targets of P-AMPK were notaffected.O2 Consumption and Activation of AMPK in

Human Hepatic Tissue. The effects of EFV werereproduced in primary hepatic cells. Figure 5A showsthat the inhibitory effect of EFV (10 and 25 lM) onthe rate of O2 consumption was similar to that exertedon Hep3B cells. Similarly, densitometric analysis ofblots from human tissue revealed that incubation withEFV (25 lM, 4 hours) induced a significant increasein P-AMPK (Fig. 5B).Changes in Intracellular Lipid Content. Cells

incubated with EFV for 24 hours exhibited a signifi-cant and concentration-dependent accumulation ofneutral lipids, primarily in lipid droplets.18 The quan-tification of cellular lipid accumulation by staticcytometry is represented in Fig. 6B. This increase wasnot observed when cells were incubated with NVP

Fig. 4. EFV increases proteinexpression of phosporylated AMPK(P-AMPK). Hep3B cells were incu-bated without (solvent) or with EFVfor 1 hour (A), 4 hours (B), and 8hours (C). Effects on proteinexpression of phosphorylated AMPK(P-AMPK) in Hep3B cells incubatedwithout (solvent) or with NVP for 4hours (D). The protein level of P-AMPK was examined by westernblot. The histogram representsmean 6 SEM of the densitometricscans for the P-AMPK protein bandof at least seven experiments, nor-malized by comparison with beta-actin and expressed as a percent-age of control. *P < 0.05, **P <0.01 versus solvent (white bars).


(Fig. 6C). HR-MAS spectroscopy was employed toevaluate rapid changes and the nature of the lipidsinvolved. The water-suppressed NMR spectra fromHep3B cells showed narrow line widths and adequatesignal-to-noise ratios with well-resolved spin–spin mul-tiplicities (Fig. 7A), and were similar in all the cell cul-tures measured. Fatty acid signals were dominant, aris-ing from both saturated (–CH2CH2CH2- at 1.3 ppmand –CH2CH3 at 0.9 ppm) and unsaturated fatty acidmoieties (-CH¼CH- at 5.4 ppm and CH¼CH- CH2

at 2.0 ppm). Signals from choline-containing com-pounds, typically associated with phospholipids, weresubstantially weaker that those from fatty acids buthigher than those from other metabolites such as lac-tate, glucose, and amino acids. Incubation of cells withEFV induced significant changes in the spectral patterncaused by a moderate but highly reproducible increasein total fatty acids. These changes are quantified in

Fig. 7B. Other signals related to lipid components,such as choline-containing compounds or unsaturatedfatty acids, were not affected by EFV, thus suggestingthat the changes observed were attributable to anincrease in saturated fatty acid moieties and not to analteration of the metabolism of membrane lipids. Thelack of significant changes in the levels of glucose orlactate suggests that the activation of glycolysis was notimplicated in any of the alterations of metabolic pa-rameters. The changes in fatty acids induced by EFV(10 and 25 lM) in the NMR spectra were notobserved when the medium contained the inhibitor ofAMPK compound C, in which case lipid levels weresimilar to those of controls. In addition, lipids in basalconditions were not significantly modified when cellswere incubated in a medium without palmitic acid,with no increase being observed with either of the twodoses of EFV employed (Fig. 7C).Coadministration with NRTIs. The effects of

10 lM EFV, 3TC, and ABC on respiration and mito-chondrial function are shown in Fig. 8. ABC, but not3TC, decreased O2 consumption and intracellularATP values to levels not significantly different fromthose induced by EFV alone. Combination of thethree drugs did not enhance the inhibitory action ofEFV on either parameter (Fig. 8A and C). Neither ofthe two NRTIs evaluated increased ROS production,but their presence significantly exacerbated the effectsof EFV (Fig. 8B).

Discussion

This study demonstrates that EFV induces an im-mediate and dose-dependent reduction in the respira-tion of both Hep3B cells and human hepatic tissue.This reduction reached statistical significance with aconcentration of 10 lM and was maximal with50 lM, approximately halving O2 consumption in thecase of the higher dose. This is a considerable diminu-tion, similar to that obtained with other inhibitors ofmitochondrial oxidative phosphorylation.13 The effectsof EFV observed in our experiments were reversible,but bearing in mind the prolonged plasmatic half-lifeand long-term daily administration of this drug,19 ourresults draw attention to the fact that patients arepotentially exposed to sustained mitochondrial dys-function. NNRTIs induce more liver toxicity thanother antiretrovirals, and up to 10% of HIV patientstreated with EFV exhibit increases in liver enzymesthat sometimes require discontinuation of therapy.9

Furthermore, the risk of hepatotoxicity is much greaterwhen HIV coexists with the HBV or HCV infection,8

Fig. 5. EFV inhibits O2 consumption and increases P-AMPK expres-sion in human hepatic tissue. (A) The histogram represents the rate ofO2 consumption in liver samples in the absence (solvent) and pres-ence of EFV in six experiments. (B) The protein levels of phosphoryl-ated AMPK after 4 hours’ incubation were examined by western blot.The histogram represents mean 6 SEM of the densitometric scans forthe P-AMPK protein band of three experiments, normalized by compar-ison with beta-actin, and expressed as a percentage of control. *P <0.05 versus solvent (white bars).


both of which are characterized by increases in media-tors known to undermine mitochondrial function.20

Therefore, we can speculate that low doses of EFVinduce levels of dysfunction below those necessary togenerate direct damage, whereas higher doses or thepresence of stimuli that further compromise the mito-chondria exacerbate these effects to a point thatbecomes clinically relevant. Of the concentrationsstudied, only that of 10 lM is within usual therapeuticplasma levels, but given the high rate of interindividualvariability in the pharmacokinetics of EFV, we believethat the higher doses employed in our experiments arealso relevant. In fact, clinical studies report suprathera-peutic plasma concentrations of EFV (up to 30-50 lM) in as many as 20% of HIV-1–infectedpatients,21,22 and the relationship between plasma con-centration and the adverse effects of EFV, in particularthose related to the liver23 and central nervous sys-tem,7 is well established. Our results with isolated mi-tochondria from rat livers point to inhibition of com-

plex I as the mechanism responsible for the effects ofEFV on cell respiration. The rapidness of the actionsdescribed in our experiments rules out any interferencewith mitochondrial DNA replication, which until nowwas considered to be the principal mechanism of themitochondrial toxicity of antiretroviral drugs, becausea much longer time frame is necessary for its effects tobe manifested.24

If not too intense or prolonged, a reduced cell respi-ration is not in itself a sign of mitochondrial malfunc-tion, and could be considered a form of cellular adap-tation to changes in environmental factors or oxygenavailability/redistribution.14,25 However, in light of ourprevious observations of a decreased mitochondrialmembrane potential with similar concentrations ofEFV,26 the increased ROS production and the drop inATP levels detected in the current study point to a cer-tain level of dysfunction within the respiratory chainthat compromises the functioning of the mitochondria.This is a novel area of research that has been the focus

Fig. 6. EFV increases intracellular lipid content in Hep3B cells. Cells were incubated for 24 hours in culture medium supplemented with 1%bovine serum albumin fatty acid free, 50 lM L-carnitine, and 0.1 mM palmitic acid, and stained with the fluorescent probe Nile red. Experimentswere performed in the absence (solvent) or presence of EFV or NVP. Fluorescence was detected by static cytometry using an inverted microscope.(A) Representative histograms and images and (B) fluorescence intensity values with EFV. (C) Fluorescence intensity values with NVP. Data repre-sent the mean 6 SEM of at least five experiments. **P < 0.01, ***P < 0.001 versus solvent (white bars).


of little attention, but one recent study of HIV-1–infected patients undergoing EFV-based treatment hasreported an increased lymphocyte mitochondrial depo-larization and other signs of dysfunction unrelated to

mitochondrial DNA replication.27 Nevertheless, thetwo mechanisms may coexist, and the effects of EFVcould feasibly be exacerbated in cells in which thenumber of mitochondria has been reduced by long-term coadministration of NRTIs. Moreover, prelimi-nary reports have hinted that some NRTIs also havethe ability to directly inhibit mitochondria,28 therebyimplying that specific combinations of the drugs used

Fig. 7. Nuclear magnetic resonance (NMR) analysis of metabolicchanges in Hep3B cells after 4 hours’ incubation in the absence (sol-vent) or presence of EFV. (A) Fatty acid region of single-pulse 1H HR-MAS NMR water-suppressed spectra of 40 lL of Hep3B cells in basalconditions (solvent, continuous line) and after incubation with EFV(10 lM, discontinuous line). Labels indicate the spectral position ofrelevant metabolites. Key: 1, CH3-(CH2)n lipids; 2, CH3 groups of Va-line, Isoleucine and Leucine; 3, (CH2)n lipids; 4, CH3 lactate; 5, CH3Alanine; 6, CH2-CH2-CO lipids; 7, CH¼CH-CH2-CH2 lipids; 8 CH Glu-tamine; 9, CH2-CH2-CO lipids; 10, CH Glutamate. (B) Bar graphs rep-resent HR-MAS measurements of relevant metabolites in Hep3B cellsin basal conditions (solvent, white) or incubated with 10 lM (grey) or25 lM (black) of EFV. Key: FA, fatty acids; UFA, unsaturated fattyacids; Cre, creatine; Cho, free choline; Gluc, glucose; Lac, lactate. (C)EFV-induced increases in fatty acid content were absent in Hep3Bcells incubated in medium without fatty acids or with the AMPK inhibi-tor compound C. The graph shows the variation of average fatty acidrelative content and standard error in the absence (solvent) or pres-ence of EFV (10 and 25 lM) in Hep3B cells, as measured by HR-MAS in three different situations: cells grown with palmitic acid in themedium, cells grown with palmitic acid and compound C in the me-dium, and cells grown without palmitic acid in the medium. Data rep-resent the mean 6 SEM of three to five experiments. **P < 0.01versus solvent (white bars).

Fig. 8. Effects of coadministration of EFV and two NRTIs (3TC andABC) on EFV-induced actions on mitochondrial function. (A) Rate ofO2 consumption in a Clark-type O2 electrode. (B) Effects on total ROSproduction assessed after incubation (30 minutes) with the fluorescentprobe 20,70-dichlorodihydrofluorescein diacetate (DCFH-DA) (2.5 lM).Measurements were obtained every 5 minutes during a 1-hour period.(C) Effects on intracellular ATP levels. Data are mean 6 SEM of 5-6experiments. *P < 0.05, **P < 0.01, ***P < 0.001 versus solvent(white bars). #P < 0.05 versus 3TC, þP < 0.05 versus ABC.


in highly active antiretroviral therapy have a particu-larly profound impact on mitochondrial function. Inour study, clinically relevant concentrations of ABC,but not of 3TC, significantly inhibited respiration andATP levels without interfering with the production ofROS. Their combined use with EFV did not result infurther synergistic reduction of respiration or ATP lev-els but did potentiate the effects of EFV on ROS pro-duction. The interpretation and clinical implicationsof these initial results, including the mitochondrial tar-gets involved, is a complex issue that demands furtherevaluation.Cells maintain ATP levels within narrow limits, and

the AMPK cascade is emerging as one important sen-sor of energy status. Its activation switches off ATP-consuming processes whereas switching on catabolicpathways that facilitate the generation of ATP, includ-ing processes related to lipid and glucose metabo-lism.29 The fact that EFV produces rapid increases inthe expression of P-AMPK in human cells and tissuecan be interpreted as a response to mitochondrial dys-function and reduction of ATP. Furthermore, there isprevious evidence that EFV modifies the metabolismof glucose and lipogenic pathways in a way that iscompatible with the activation of AMPK.30 The abilityto switch on glycolysis is an important factor in thecapacity of cells to survive the metabolic stress causedby intense mitochondrial malfunction.25 However, inthe current study, the analysis of glucose uptake andexpression of the glucose transporter GLUT-1 suggestthat there was no such increase in anaerobic glycolysisafter 4 hours’ incubation with EFV. This is somewhatparadoxical but could be a consequence of the highlyglycolytic nature of Hep3B, which leaves little roomfor an improvement in this pathway.31 Mitochondriaalso use fatty acids as fuel to generate ATP. Indeed, anAMPK-mediated increase in fatty acid uptake andbeta-oxidation has been described after energy depriva-tion, whereas activation of AMPK has been related toa decrease in hepatic steatosis.32 Nevertheless, giventhe oxidative capacity of the mitochondria inhibitedby EFV, it is possible that lipids entering the cell accu-mulate over time within the cytoplasm. This is sup-ported by our finding that 24-hour exposure to EFVproduced an increase in the levels of neutral lipids,which are present in lipid droplets. A more in-depthanalysis employing the high-resolution technique nu-clear magnetic resonance confirmed the rise in lipidlevels observed with static cytometry and the absenceof an elevated anaerobic glycolysis, evident in the lackof alterations in the levels of glucose or other metabo-lites such as lactate. The changes in intracellular lipids

were moderate but highly reproducible and becamesignificant after only 4 hours’ incubation with EFV.Furthermore, the chemical nature of the lipids whoselevels increased suggests that they did not originatefrom membranes, a reservoir of lipids for mitochon-drial use.33 Therefore, it is of relevance that removalof palmitic acid, the only fatty acid present in our cul-ture media, prevented the increase of intracellular lip-ids produced by EFV; this suggests that extracellularlipids are the source of the increase. Impairment ofmitochondrial function may initially cause microvesic-ular steatosis, which, if prolonged, results in moresevere forms of hepatic damage.34,35 Indeed, drugsthat inhibit mitochondrial respiration and impair ATPsynthesis are associated with hepatic steatosis and stea-tohepatitis, which are histologically indistinguishablefrom nonalcoholic fatty liver disease.36 The lipidchanges reported here seem to be related to AMPKactivation, and subsequently to the energetic imbalancethat produced AMPK phosphorylation in the firstplace, because they were not observed when thisenzyme was inhibited with compound C. Finally,NVP had no effect on respiration, ROS generation, in-tracellular ATP levels, expression of P-AMPK, or levelsof neutral lipids in Hep3B cells. The differencesbetween our results with NVP and EFV support exist-ing clinical evidence that the degree and mechanismsof the hepatotoxicity produced by both drugs are spe-cific to each one, and are not shared by NNRTIs as apharmacological group.8 In conclusion, the currentstudy demonstrates that clinical concentrations of EFVinduce bioenergetic stress in hepatic cells by acutely in-hibiting complex I of the respiratory chain. This newmechanism of mitochondrial interference leads to arapid accumulation of lipids in the cytoplasm that ismediated by activation of AMPK. Given that treat-ment with EFV is for life, these effects could easilyaccumulate and increase the liver toxicity induced bycoinfections and other drugs.

Acknowledgment: The authors thank Dr. Jose Este-ban Peris for assisting with the high-pressure liquidchromatography analysis of the NNRTI solutions, Dr.Manuel de Juan for providing liver biopsies, Prof. JuanSastre for his scientific revision of the manuscript, andBrian Normanly for his English language editing.

References1. Staszewski S, Morales-Ramirez J, Tashima KT, Rachlis A, Skiest D,

Stanford J, et al. Efavirenz plus zidovudine and lamivudine, efavirenzplus indinavir, and indinavir plus zidovudine and lamivudine in thetreatment of HIV-1 infection in adults. Study 006 Team. N Engl JMed 1999;341:1865-1873.


2. Starr SE, Fletcher CV, Spector SA, Yong FH, Fenton T, Brundage RC,et al. Combination therapy with efavirenz, nelfinavir, and nucleosidereverse-transcriptase inhibitors in children infected with human immu-nodeficiency virus type 1. Pediatric AIDS Clinical Trials Group 382Team. N Engl J Med 1999;341:1874-1881.

3. Hammer SM, Eron JJ Jr, Reiss P, Schooley RT, Thompson MA,Walmsley S, et al. Antiretroviral treatment of adult HIV infection:2008 recommendations of the International AIDS Society-USA panel.JAMA 2008;300:555-570.

4. Manfredi R, Calza L, Chiodo F. An extremely different dysmetabolicprofile between the two available nonnucleoside reverse transcriptaseinhibitors: efavirenz and nevirapine. J Acquir Immune Defic Syndr2005;38:236-238.

5. Perez-Molina JA, Domingo P, Martınez E, Moreno S. The role of efa-virenz compared with protease inhibitors in the body fat changes asso-ciated with highly active antiretroviral therapy. J Antimicrob Chemother2008;62:234-245.

6. Best BM, Goicoechea M. Efavirenz-still first-line king? Expert OpinDrug Metab Toxicol 2008;4:965-972.

7. Gutierrez F, Navarro A, Padilla S, Anton R, Masia M, Borras J, et al.Prediction of neuropsychiatric adverse events associated with long-termefavirenz therapy, using plasma drug level monitoring. Clin Infect Dis2005;41:1648-1653.

8. Sulkowski MS, Thomas DL, Metha SH, Chaisson RE, Moore RD.Hepatotoxicity associated with efavitenz-containing antiretroviral ther-apy: role of hepatitis C and B infections. HEPATOLOGY 2002;35:182-189.

9. Bruck S, Witte S, Brust J, Schuster D, Mosthaf F, Procaccianti F, et al.Hepatotoxicity in patients prescribed Efavirenz or Nevirapine. Eur JMed Res 2008;13:343-348.

10. Sato N. Central role of mitochondria in metabolic regulation of liverpathophysiology. J Gastroenterol Hepatol 2007;22:S1-S6.

11. Pessayre D, Mansouri A, Haouzi D, Fromenty B. Hepatotoxicity dueto mitochondrial dysfunction. Cell Biol Toxicol 1999;15:367-373.

12. Sanjuan-Pla A, Cervera AM, Apostolova N, Garcıa-Bou R, Vıctor VM,Murphy MP, et al. A targeted antioxidant reveals the importance of mi-tochondrial reactive oxygen species in the hypoxic signaling of HIF-1a.FEBS Lett 2005;579:2669-2674.

13. Esplugues JV, Rocha M, Nunez C, Bosca I, Ibiza S, Herance JR, et al.Complex I dysfunction and tolerance to nitroglycerin: an approachbase don mitochondrial-targeted antioxidants. Circ Res 2006;99:1067-1075.

14. Victor VM, Nunez C, D’Ocon P, Taylor CT, Esplugues JV, MoncadaS. Regulation of oxygen distribution in tissues by endothelial nitric ox-ide. Circ Res 2009;104:1178-1183.

15. Gomez J, Caro P, Naudı A, Portero-Otin M, Pamplona R, Barja G.Effect of 8.5% and 25% caloric restriction on mitochondrial free radi-cal production and oxidative stress in rat liver. Biogerontology 2007;8:555-566.

16. Greenspan P, Mayer EP, Fowler SD. Nile red: a selective fluorescentstain for intracellular lipid droplets. J Cell Biol 1985;100:965-973.

17. Lindon JC, Holmes E, Nicholson JK. Metabonomics in pharmaceuticalR&D. FEBS J 2007;274:1140-1151.

18. Beckman M. Great balls of fat. Science 2006;311:1232-1234.19. Vrouenraets SM, Wit FW, van Tongeren J, Lange JM. Efavirenz: a

review. Expert Opin Pharmacother 2007;8:851-871.

20. Beltran B, Quintero M, Garcıa-Zaragoza E, O’Connor E, EspluguesJV, Moncada S. Inhibition of mitochondrial respiration by endogenousnitric oxide: a critical step in Fas signalling. Proc Natl Acad Sci U S A2002;99:8892-8897.

21. Marzolini C, Telenti A, Decosterd LA, Greub G, Biollaz J, Buclin T. Efa-virenz plasma levels can predict treatment failure and central nervous sys-tem side effects in HIV-1-infected patients. AIDS 2001;15:71-75.

22. Burger D, van der Heiden I, la Porte C, van der Ende M, GroeneveldP, Richter C, et al. Interpatient variability in the pharmacokinetics ofthe HIV non-nucleoside reverse transcriptase inhibitor efavirenz: theeffect of gender, race, and CYP2B6 polymorphism. Br J Clin Pharma-col 2006;61:148-154.

23. Kappelhoff BS, van Leth F, Robinson PA, MacGregor TR, Baraldi E,Montella F, et al. 2NN Study Group: are adverse events of nevirapineand efavirenz related to plasma concentrations? Antivir Ther 2005;10:489-498.

24. Walker UA, Setzer B, Venhoff N. Increased long-term mitochondrialtoxicity in combinations of nucleoside analogue reverse-transcriptaseinhibitors. AIDS 2002;16:2165-2173.

25. Moncada S, Erusalimsky JD. Does nitric oxide modulate mitochondrialenergy generation and apoptosis? Nat Rev Mol Cell Biol 2002;3:214-220.

26. Apostolova N, Blas-Garcıa A, Ballesteros D, Gonzalez Y, Moran A,Gomez-Sucerquia J, et al. Clinical concentrations of efavirenz reducecellular proliferation and viability in several human cell lines. AntivTher 2008;13(Suppl 4):A23.

27. Karamchand L, Dawood H, Chuturgoon AA. Lymphocyte mitochon-drial depolarization and apoptosis in HIV-1 infected HAART patients.J Acquir Immune Defic Syndr 2008;48:381-388.

28. Modica-Napolitano JS. AZT causes tissue-specific inhibition of mito-chondrial bioenergetic function. Biochem Biophys Res Commun 1993;194:170-177.

29. Hardie DG. AMP-activated/SNF1 protein kinases: conserved guardiansof cellular energy. Nat Rev Mol Cell Biol 2007;8:774-785.

30. El Hadri K, Glorian M, Monsempes C, Dieudonne M, Pecquery R, Giu-dicelli Y, et al. In vitro suppression of the lipogenic pathway by the nonnu-cleoside reverse transcriptase inhibitor Efavirenz in 3T3 and humanpreadipocytes or adipocytes. J Biol Chem 2004;279:15130-15141.

31. Kim JS, Ahn KJ, Kim JA, Kim HM, Lee JD, Lee JM, et al. Role of re-active oxygen species-mediated mitochondrial dysregulation in 3-bro-mopyruvate induced cell death in hepatoma death: ROS-mediated celldeath by 3-BrPA. J Bioenerg Biomembr 2008;40:607-618.

32. Viollet B, Foretz M, Guigas B, Horman S, Dentin R, Bertrand L,et al. Activation of AMP-activated protein kinase in the liver: a newstrategy for the management of metabolic hepatic disorders. J Physiol2006;574:41-53.

33. Barba I, Cabalas ME, Arus C. The relationship between nuclear mag-netic resonance-visible lipids, lipid droplets, and cell proliferation incultured C6 cells. Cancer Res 1999;59:1861-1868.

34. Berson A, De Beco V, Letteron P, Robin MA, Moreau C, El Kahwaji J,et al. Steatohepatitis-inducing drugs cause mitochondrial dysfunctionand lipid peroxidation in rat hepatocytes. Gastroenterology 1998;114:764-774.

35. Fromenty B, Pessayre D. Inhibiton of mitochondrial beta-oxidation asa mechanism of hepatotoxicity. Pharmacol Ther 1995;67:101-154.

36. Browning JD, Horton JD. Molecular mediators of hepatic steatosis andliver injury. J Clin Invest 2004;114:147-152.


Inhibition of mitochondrial function by efavirenz increases lipid content in hepatic cells

Documents

Transcript of Inhibition of mitochondrial function by efavirenz increases lipid content in hepatic cells