Compromising mitochondrial function with the antiretroviral drug efavirenz induces cell survival-promoting autophagy


  • Nadezda Apostolova,

    Corresponding author
    1. Departamento de Farmacología, Facultad de Medicina, Universidad de Valencia, Valencia, Spain
    2. CIBERehd, Valencia, Spain
    • Facultad de Medicina, Universidad de Valencia, Avda. Blasco Ibáñez n.15-17, 46010 Valencia, Spain
    Search for more papers by this author
    • fax: (+34) 963983879

  • Leysa J. Gomez-Sucerquia,

    1. Departamento de Farmacología, Facultad de Medicina, Universidad de Valencia, Valencia, Spain
    2. Hospital Universitario Dr. Peset, Valencia, Spain
    Search for more papers by this author
  • Anna Gortat,

    1. Departamento de Farmacología, Facultad de Medicina, Universidad de Valencia, Valencia, Spain
    Search for more papers by this author
  • Ana Blas-Garcia,

    1. Departamento de Farmacología, Facultad de Medicina, Universidad de Valencia, Valencia, Spain
    2. Hospital Universitario Dr. Peset, Valencia, Spain
    Search for more papers by this author
  • Juan V. Esplugues

    1. Departamento de Farmacología, Facultad de Medicina, Universidad de Valencia, Valencia, Spain
    2. CIBERehd, Valencia, Spain
    3. Hospital Universitario Dr. Peset, Valencia, Spain
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • Supported by grants PI081325 and CIBER CD06/04/0071 (Ministerio de Sanidad y Consumo, Spain), PROMETEO/2010/060 and ACOMP/2010/207 (GeneralitatValenciana, Spain) and an unrestricted grant from Abbott Laboratories.


Hepatotoxicity is a very common side effect associated with the pharmacological treatment of human immunodeficiency virus (HIV) infection and its pathogenesis is poorly understood. Efavirenz (EFV) is the most widely used nonnucleoside reverse transcriptase inhibitor administered for the control of HIV and some of its toxic effects in hepatic cells have been recently shown to display features of mitochondrial dysfunction. Here we studied the activation of autophagy and, in particular, mitophagy, the main mitochondrial turnover mechanism, in human hepatic cells treated with clinically relevant concentrations of this drug. EFV-treated cells had altered mitochondria, characterized by a relative increase in mitochondrial mass and defective morphology. This was followed by induction of autophagy as shown by the presence of autophagic vacuoles and the presence of the specific autophagic marker proteins microtubule-associated protein 1A/1B light chain 3 and Beclin-1. Importantly, whereas moderate levels of EFV activated autophagy, higher concentrations led to blockage in the autophagic flux, a condition that promotes “autophagic stress” and produces severe cellular damage. Finally, pharmacological inhibition of autophagy exacerbated the deleterious effect of EFV on cell survival/proliferation promoting apoptosis, which suggests that autophagy acts as an adaptive mechanism of cell survival. Conclusion: Clinical concentrations of EFV induce autophagy and, in particular, mitophagy in hepatic cells. Activation of this process promotes cell survival, but exceeding a certain threshold of mitochondrial dysfunction is associated with an autophagic overload or stress. This effect could be involved in the EFV-associated hepatotoxicity and may constitute a new mechanism implicated in the genesis of drug-induced liver damage. (HEPATOLOGY 2011;)

Highly active antiretroviral therapy (HAART), also known as combined antiretroviral therapy (cART), has rendered human acquired immunodeficiency syndrome (AIDS) a chronic rather than mortal illness. However, there is increasing concern about its adverse effects and, in particular, the extent of liver damage related to this medication. Significant drug-induced hepatotoxicity has been identified in 8.5%-23% of HAART patients, leading up to a third of the therapy discontinuations, and this can be underreported because 50% of patients with increased liver enzymes are asymptomatic.1, 2 Mitochondrial toxicity is a major mechanism of this liver injury, but it has been generally attributed to one component of this multidrug therapy: nucleoside analog reverse transcriptase inhibitors (NRTI), which inhibit mitochondrial DNA (mtDNA) polymerase gamma (Pol-γ), the enzyme responsible for mtDNA replication.3 HAART regimens usually comprise two NRTI plus either a boosted protease inhibitor or a nonnucleoside reverse transcriptase inhibitor (NNRTI).4 NNRTI does not inhibit Pol-γ, but some of the toxic effects display features of mitochondrial dysfunction.5, 6 Efavirenz (EFV), the most widely used NNRTI, is generally considered safe, although there is growing concern about its relation to psychiatric symptoms, lipid and metabolic disorders, and hepatotoxicity, with between 1%-8% of patients exhibiting raised liver function test results.7-10 The molecular mechanisms responsible for these effects remain largely unknown, although there is evidence that EFV reduces cellular proliferation and triggers apoptosis in vitro.11, 12 We recently reported similar deleterious effects in human hepatic cells involving mitochondrial and metabolic alterations that led to accumulation of lipids.13, 14 EFV induced a major bioenergetic change manifested by reduced mitochondrial respiration with specific inhibition at Complex I, decreased adenosine triphosphate (ATP) production, and mitochondrial membrane potential (ΔΨm), and increased reactive oxygen species generation. Mitochondrial damage/dysfunction is one of the main inducers of macroautophagy (also called autophagy), which is a mechanism of mitochondrial quality control and a general, controlled cytoprotective response. This evolutionarily conserved, degradative process functions in all eukaryotic cells, under basal conditions, enabling physiological turnover of cellular compartments, and upon induction by a long list of stimuli. When autophagic sequestration selectively involves mitochondria, this process is denoted mitophagy.15

Here we report that clinically relevant concentrations of EFV induce autophagy and, in particular, mitophagy in human hepatic cells. We provide evidence that this process promotes cell survival, but exceeding a certain threshold of mitochondrial dysfunction is associated with an autophagic overload or stress. This complex effect could be involved in EFV-related hepatic toxicity and may constitute a new mechanism implicated in the genesis of drug-generated liver damage.


AIDS, acquired immunodeficiency syndrome; ΔΨm, mitochondrial transmembrane potential; EFV, Efavirenz; HAART, highly active antiretroviral therapy; HIV, human immunodeficiency virus; LC3, microtubule-associated protein 1A/1B light chain 3; 3MA, 3-methyladenine; MPT, mitochondrial permeability transition; NAO, 10-N-nonyl acridine orange; NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; Pol-γ, DNA polymerase gamma; PI, propidium iodide; STS, staurosporine; TEM, transmission electron microscopy.

Materials and Methods

Reagents and Drugs

Unless stated otherwise, chemical reagents and fluorochromes were purchased from Sigma-Aldrich (Steinheim, Germany). Efavirenz (Sustiva 600 mg, Bristol-Myers Squibb) was acquired in its clinically available form and dissolved in methanol (3 mg/mL) once insoluble substances had been removed by filtration. The purity (98%-100%) and stability were evaluated by high-performance liquid chromatography (HPLC) and compared with a control solution of EFV (Sequoia Research Products, Pangbourne, UK). The employed range of EFV (10, 25, and 50 μM) is clinically relevant and was chosen considering the important interindividual variability in its pharmacokinetics.16 Although the therapeutic plasma levels of EFV are believed to be 3.17-12.67 μM, as many as 20% of patients exhibit higher levels, with values of 30-50 μM being documented.17-19 0.5% methanol was employed in all EFV treatments and vehicle control experiments, versus which statistical analysis was performed. In most experiments the vehicle-treated were compared to untreated cells and no significant differences in any of the parameters were detected.

Cell Culture and Gene Expression

We used Hep3B cells (American Type Culture Collection [ATCC] HB-8064), which despite constituting a transformed cell line, is considered metabolically competent and, unlike other human hepatoma cell lines, such as HepG2, has an active cytochrome P450 system. Confirmatory experiments were performed in primary human hepatocytes and for gene overexpression we used the human cervical carcinoma cell line HeLa (ATCC CCL-2), as these cells also possess a high mitochondrial content and are frequently employed for transfection (details in Supporting Material).

Western Blotting (WB)

WB was performed using whole-cell protein extracts as described.13 Primary antibodies: anti-Beclin (Abcam), anti-microtubule-associated protein 1A/1B light chain 3 (LC3), and anti-actin (both from Sigma-Aldrich, Steinheim, Germany), all at 1:1,000, and a secondary antibody peroxidase-labeled antirabbit IgG (Vector Laboratories, Burlingame, CA) at 1:5,000.

Fluorescence Microscopy and Static Cytometry

Fluorescence was visualized using a fluorescence microscope (IX81, Olympus, Hamburg, Germany). “CellR” software v. 2.8 was employed to capture individual images and the fluorescent signal was quantified using static cytometry software “ScanR” v. 2.03.2 (Olympus). Following treatment and incubation with fluorochromes, cells were washed in Hank's balanced salt solution (HBSS) and life-cell images were recorded. Nuclei were stained with the fluorochrome Hoechst 33342 (1 μM) (last 30 minutes of the treatment).

Mitochondrial Morphology and Mitochondrial Mass.

Mitochondria were visualized and mitochondrial mass was monitored in Hep3B cells treated with EFV (6 hours) using the fluorescent dye 10-N-nonyl acridine orange (NAO) 0.5 μM, which specifically binds to cardiolipin independent of ΔΨm.20 We also used stably transfected HeLa cells expressing the red fluorescent protein mtdsRed tagged for mitochondrial localization and specifically designed for the fluorescent labeling of these organelles (details in Supporting Material).


LC3 expression and localization were studied using HeLa cells stably expressing LC3-GFP, treated with EFV (24 or 48 hours) (details in Supporting Material).

Lysosomal Content.

Lysosomes were stained with the fluorescent dye Lysotracker Green 0.1 μM (last 30 minutes of the treatment) in EFV-treated HeLa cells (24 hours).

Cell Proliferation and Survival/Apoptosis.

For cell proliferation/survival studies, Hep3B, primary hepatocytes, or HeLa cells stably expressing mtdsRed were allowed to proliferate exponentially (48-well plates) for 24 hours in the presence of EFV. To study the role of autophagy, cells were cotreated with 2.5 mM 3-methyladenine (3MA), a specific inhibitor of autophagosome formation, for 1 hour prior to EFV treatment and during the entire treatment period (24 hours). Cells were counted according to Hoechst fluorescence (25 images/well). Apoptosis was studied in Hep3B cells as bivariate Annexin V/PI analysis (apoptosis detection kit, Abcam). Following treatment (24 hours), the medium was replaced with HBSS containing 0.9 μL/well of AnnexinV-fluorescein (to detect phosphatidyl serine exteriorization) and incubated (30 minutes), after which 0.3 μL/well of the chromatin-detecting dye propidium iodide (PI) was added (5 minutes) to label dead or damaged cells. The protein kinase inhibitor staurosporine (STS) was employed as a positive proapoptotic control.

Transmission Electron Microscopy (TEM)

Hep3B (5 × 104/chamber), primary hepatocytes (105/chamber), or HeLa cells (3 × 104/chamber) were seeded in 4-well Lab-Tek chamber slides (Nalge Nunc International, Naperville, IL). After treatment, cells were fixed in 3.5% glutaraldehyde (1 hour, 37°C), postfixed in 2% OsO4 (1 hour, room temperature), and stained with 2% uranyl acetate in the dark (2 hours, 4°C). Finally, cells were rinsed in sodium phosphate buffer (0.1M, pH 7.2), dehydrated in ethanol, and infiltrated overnight in araldite (Durcupan, Fluka, Buchs, Switzerland). Following polymerization, embedded cultures were detached from the chamber slide and glued to araldite blocks. Serial semithin (1.5 μm) sections were cut with an Ultracut UC-6 (Leica, Heidelberg, Germany), mounted onto slides, stained with 1% toluidine blue, and glued (Super Glue, Loctite) to araldite blocks and detached from the glass slide by repeated freezing (in liquid nitrogen) and thawing. Ultracut-prepared ultrathin (0.07 μm) sections were stained with lead citrate. Finally, photomicrographs were obtained with a TEM (FEI Tecnai Spirit G2) using a digital camera (Morada, Soft Imaging System, Olympus).

Confocal Microscopy

Stably transfected HeLa LC3-GFP and mtdsRed cells were treated with EFV (24 hours) and Lysotracker Green or Red 0.1 μM (Molecular Probes, Invitrogen, Eugene, OR) added for the last 30 minutes of the treatment to stain the lysosomes. After washing with HBSS, life-cell images were acquired with a Leica TCS-SP2 confocal laser scanning unit with argon and helium-neon laser beams and attached to a Leica DM-IRBE inverted microscope. Images were captured at 63× magnification with HCX PL APO 63.0 × 1.32 oil UV objective. The excitation wavelength used for mtdsRed and Lysotracker Red was 543 nm, 488 nm in the case of LC3-GFP and Lysotracker Green, and the emission apertures for fluorescence detection were 560-700 nm and 502-539 nm, respectively. Images were analyzed with LCS Lite software and overlapping of the red and the green fluorescent signal was quantified with the program ImageJ. The Colocalization Colormap Plugin was used to calculate the Correlation Index (Icorr).

Presentation of Data and Statistical Analysis

Data were analyzed using GraphPad Prism v. 3 software with one-way analysis of variance (ANOVA), followed by Newman-Keuls multiple comparison test or by Student's t test. All values are mean ± standard error of the mean (SEM) and statistical significance was: *P < 0.05, **P < 0.01, and ***P < 0.001.


Alteration of Mitochondrial Morphology and Increase in Mitochondrial Mass.

Taking into consideration recently published evidence concerning EFV-induced mitochondrial dysfunction in hepatic cells, we delved more deeply by assessing mitochondrial mass and morphology. Fluorescence microscopy in NAO-stained Hep3B and primary hepatocytes treated with EFV revealed considerable alterations of the mitochondrial signal, which were concentration-dependent and visible as early as 6 hours after EFV 50 μM treatment. Although the mitochondrial net spread over the entire cytoplasm in control (untreated) cells, EFV 50 μM treatment produced a localized and compacted mitochondrial signal (Figs. 1A, 8B). Similar modifications were obtained in Hep3B cells stained with another mitochondrial stain Mitotracker Green (data not shown). To further analyze these effects, we treated HeLa cells stably expressing mtdsRed with increasing concentrations of EFV for periods of up to 48 hours. Alterations of mitochondrial size and shape similar to those appearing in hepatic cells were detected (results not shown). Moreover, quantification of the red mitochondrial signal (mtdsRed) using static cytometry revealed a concentration-dependent increase in the relative mitochondrial mass (Fig. 1B) that was statistically significant as early as at 6 hours treatment with EFV 50 μM. At 24 hours, EFV 25 μM also reached statistical significance. Mean red fluorescence values at 48 hours treatment did not differ from those at 24 hours.

Figure 1.

Mitochondrial mass in EFV-treated cells. (A) Fluorescence microscopy images of Hep3B cells treated with EFV 50 μM for 6 hours showing nuclei (Hoechst-stained, blue) and mitochondria (NAO-stained, green). (B) Static cytometry of mean red fluorescence in stably transfected HeLa mtdsRed cells; data shown as mean ± SEM, n = 6 (1 hour), 4-5 (6 hours), 9 (24 hours), and 3-4 (48 hours). Data were analyzed by one-way ANOVA multiple comparison test followed by Newman-Keuls test, *P < 0.05, ***P < 0.001, versus vehicle.

Induction of Severe Damage in Mitochondria.

TEM images of Hep3B, primary hepatocytes, and HeLa cells (Figs. 2, 8A) revealed that 24-hour treatment with EFV produced concentration-dependent mitochondrial damage. In control cells mitochondria were smooth, with distinct cristae and complete membranes. Cells treated with 10 μM displayed mitochondria that were generally normal and only occasionally altered, whereas 25 μM-exposed cells exhibited a severely damaged mitochondrial ultrastructure with aberrant cristae and decreased cristae number. Some of the damaged mitochondria had a swollen appearance and there was a clear change in their shape. Although control cells had a higher percentage of rod-shaped mitochondria, exposure to EFV produced irregular or round structures. Furthermore, we observed a significant augmentation in mitochondrial size, accompanied by a concentration-dependent reduction in the number of mitochondria. When using EFV 50 μM, a large number of mitochondria did not have visible cristae, and many showed alterations of the outer membrane, including surface whorls. In addition, their internal structure was hypercondensed and obscured by an electron-dense matrix. Of note, in the case of both EFV 25 μM and 50 μM, we also found evidence of autophagic degradation of mitochondria, manifested in double-membrane vacuolar structures that contained mitochondria. Moreover, careful examination of the TEM images revealed that endoplasmatic reticulum (ER) appeared to be wrapped around the mitochondria, possibly in order to generate a membrane that would be later incorporated into the autophagic vacuoles.

Figure 2.

Representative images of TEM of Hep3B (upper panel) and HeLa cells (lower panel) treated with vehicle or EFV (10, 25, or 50 μM) for 24 hours. Magnification 26,500×.

Triggering of Autophagy.

Several experimental approaches confirmed the activation of autophagy suggested by TEM imaging. Using WB, we studied the expression of two autophagic protein markers, Beclin-1 and LC3. Following translation, the unprocessed form of LC3 (proLC3) is proteolytically cleaved, resulting in the LC3-I form (18 kDa). Upon activation of autophagy, LC3-I is cleaved at its C-terminus, the free C-terminal glycine is modified by lipidation to LC3-II (16 kDa), which relocalizes to newly-formed vesicles. The conversion of LC3-I to LC3-II is considered a major hallmark of autophagy and commonly interpreted as an autophagic indicator.21, 22 In EFV-treated Hep3B cells, both LC3-II and Beclin-1 expression were enhanced (Fig. 3A,B). As a positive control, we employed cells exposed to nutrient deprivation (cultured in HBSS). LC3-II expression was augmented at 8 hours in a concentration-dependent manner, and this increase was maintained at 24 hours. An enhanced signal for Beclin-1 was only detected after 24 hours of EFV exposure; nevertheless, at 8 hours the positive control also failed to induce Beclin-1 up-regulation. LC3 activation was also detected in primary hepatocytes treated with EFV for 24 hours (Fig. 8C,D). In addition, a concentration-dependent increase in the presence of LC3-II-characteristic punctae was detected by fluorescence microscopy in EFV-treated HeLa cells stably expressing LC3-GFP (Fig. 3C,D), even with the lowest concentration employed (10 μM), detected at 24 hours and maintained at 48 hours. Interestingly, when overall mean green fluorescence was evaluated, EFV 50 μM-treated HeLa LC3-GFP cells exhibited a significant increase, which was particularly evident at 48 hours (Fig. 3E). The activation of autophagy shown by fluorescence microscopy was further confirmed by confocal microscopy with HeLa cells stably expressing LC3-GFP stained with the lysosomal fluorescent marker Lysotracker Red. While control cells showed a disperse LC3-GFP signal, LC3-II-specific punctae were present with EFV 25 and 50 μM (24 hours) and only occasionally in those treated with 10 μM (Fig. 4A). Importantly, EFV induced substantial overlapping of the green (LC3-GFP) and the red signal (Lysotracker Red), thus suggesting the formation of autophagolysomes. Analysis of the two signals, displayed as Icorr in Fig. 4B, revealed statistically significant colocalization in cells treated with 25 and 50 μM of EFV, whereas the value of 10 μM-treated did not differ from that of vehicle-treated cells.

Figure 3.

Analysis of autophagic markers. (A) Representative WB images of LC3 and Beclin-1 in Hep3B cells treated with vehicle or EFV or exposed to nutrient deprivation (n.d.) for 24 hours. (B) Summary of densitometry data of protein expression in 8-hour (upper panel) and 24-hour-treated (lower panel) Hep3B cells; results (mean ± SEM, n = 4-6) are expressed as protein expression in relation to that of untreated cells in each individual experiment, which was considered 100%. Statistical analysis was performed by Student's t test (*P < 0.05, **P < 0.01, versus vehicle); the control (n.d.) was analyzed separately (#P < 0.05, ##P < 0.01, versus untreated cells). (C-E) Analysis of LC3 expression by fluorescence microscopy in stably transfected HeLa LC3-GFP cells treated with EFV for 24 and 48 hours. (C) Representative fluorescence images after 24 hours treatment with vehicle and EFV 50 μM (green: LC3-GFP and blue: Hoechst). (D) Analysis of the presence of green (LC3-II) punctae; data shown as mean ± SEM, n = 4. (E) Overall mean LC3-GFP fluorescence; data shown as mean ± SEM, n = 4, the mean green fluorescence recorded in untreated cells was considered 100%. Data were analyzed by one-way ANOVA multiple comparison test followed by Newman-Keuls test, *P < 0.05, **P < 0.01, versus vehicle and by Student's t test, #P < 0.05, 24 hours versus 48 hours treatment.

Figure 4.

Confocal microscopy of stably transfected HeLa LC3-GFP cells, treated with vehicle or EFV (10, 25, or 50 μM) for 24 hours. (A) Representative fluorescence images show LC3-GFP (green) and lysosomes stained with Lysotracker Red (red). (B) Overlapping analysis of the green and the red signal expressed as Icorr; data shown as mean ± SEM were analyzed by Student's t test (n = 3-6, *P < 0.05, versus vehicle).

Moderate Concentrations of EFV Induce Mitochondrial Degradation by Autophagy, Whereas High Concentrations of EFV Induce Autophagic Stress.

To further study mitochondrial degradation by autophagy, additional confocal microscopy experiments were performed in which HeLa cells stably expressing mtdsRed protein were treated with EFV (24 hours). Lysosomes were stained with Lysotracker Green and colocalization of the two signals was assessed. As expected, little or no overlapping of mitochondrial and lysosomal signals was observed in control cells, whereas EFV led to increased positive colocalization (Fig. 5). To our surprise, the concentration-effect curve seemed hormetic, as EFV 50 μM-treated cells showed less overlapping (Fig. 5). This result indicated a possible blockage of the autophagic flux by EFV 50 μM. Similarly, static cytometry experiments in EFV-treated HeLa cells (24 hours) revealed a major increase in mean Lysotracker Green fluorescence with 50 μM, whereas no changes were detected with 10 μM or 25 μM (Fig. 6A). To confirm these results, we monitored the autophagic flux by studying LC3 expression in both primary hepatic and Hep3B cells in the presence of Bafilomycin A1, a vacuolar-type ATPase inhibitor that impairs lysosomal function by inhibiting its Na+H+ pump. In the presence of this compound, accumulation of LC3-II positive autophagosomes would be evidence of an efficient autophagic flux, whereas the lack of such an increase would point to a defect or delay in this process prior to degradation at the lysosome.23 Our WB experiments showed that cotreatment with 20 nM Bafilomycin A1 led to LC3 accumulation in cells treated with EFV 10 μM and 25 μM (24 hours) in a way similar to that observed in control cells. However, in the presence of EFV 50 μM, exposure to Bafilomycin A1 did not induce such an increase (Figs. 6B, 8C). This confirmed a defect in the progression/resolution of autophagy, a condition also known as “autophagic stress,” in cells treated with EFV 50 μM.

Figure 5.

Confocal microscopy of stably transfected HeLa mtdsRed cells, treated with vehicle or EFV (10, 25, or 50 μM) for 24 hours. (A) Representative fluorescence images show mtdsRed signal (red) and lysosomes stained with Lysotracker Green (green). (B) Overlapping analysis of the green and the red signal expressed as Icorr; data shown as mean ± SEM were analyzed by Student's t test (n = 4-6, *P < 0.05 versus vehicle).

Figure 6.

Analysis of the autophagic flux. (A) Summary of the overall mean Lysotracker Green fluorescence (mean ± SEM, n = 4, *P < 0.05 versus vehicle) in HeLa cells treated with EFV for 24 hours. (B) WB of LC3 expression after 24-hour EFV treatment of Hep3B cells in the presence or absence of 20 nM Bafilomycin A1 (representative image).

EFV-Induced Autophagy Promotes Cell Survival.

Autophagy is an adaptive, cell survival-promoting mechanism. However, it is also considered a cell death-inducing condition that, if prolonged, can lead to what is known as “nonapoptotic type II programmed cell death.” To study whether the autophagic activation in our model promotes or compromises cell survival, we treated HeLa cells stably expressing mtdsRed with 3MA, a class III PI3K inhibitor often applied as a suppressor of autophagosomal formation.24 Previous reports have shown that EFV exerts an inhibitory effect on cell viability and proliferation in both Hep3B and HeLa, with higher concentrations of this drug promoting apoptosis.13 Our experiments revealed that inhibition of autophagy worsened the damaging effect of EFV, suggesting that autophagy plays a cell survival-promoting role. Static cytometry showed that exposure to EFV (24 hours) produced a concentration-dependent cell number reduction (92.35 ± 3.50% and 43.04 ± 2.74% in EFV 25 μM and 50 μM, respectively, versus 100% in untreated cells). Importantly, this reduction was more pronounced in the presence of 3MA (76.84 ± 5.22% and 30.36 ± 2.11% in EFV 25 μM and 50 μM, respectively, versus 100% in 3MA-treated controls) (Fig. 7A). When we studied the mitochondrial signal by means of mtdsRed fluorescence, cells treated with EFV 25 μM in the presence of 3MA showed higher mean fluorescence values than those in which autophagy was not inhibited. However, in the case of EFV 50 μM the increase in the red signal was modest and without statistical significance. This provides further confirmation that EFV 50 μM leads to a blockage of the autophagic pathway in our model. Finally, no significant changes were detected with the lowest EFV concentration (10 μM) in the presence of 3MA (Fig. 7A). Similarly, incubation with 3MA alone did not affect cell number or mean mtdsRed fluorescence (data not shown). A similar effect of 3MA regarding cell survival was observed in Hep3B (Fig. 7B) and primary human hepatocytes (Fig. 8E). Moreover, we performed Bivariate Annexin V/PI analysis to address the induction of apoptotic cell death in Hep3B cells subjected to EFV in the presence of 3MA. The presence of four cellular subpopulations was evaluated by static cytometry: vital (double negative), apoptotic (Annexin V+/PI), late apoptotic/necrotic (Annexin V+/PI+), and damaged cells (Annexin V/PI+) cells. As displayed in Fig. 7B. cotreatment with 3MA enhances the apoptotic effect of EFV but it does not interfere with the action of the common apoptotic inducer STS, thus suggesting a specific role of autophagy in the EFV-induced effect.

Figure 7.

EFV effect in the presence of 3MA studied by static cytometry. Cells were treated with vehicle, EFV (10, 25, or 50 μM) or STS 1 μM for 24 hours, in the absence or presence of 2.5 mM 3MA. (A) Cell number and mean mtdsRed fluorescence in HeLa cells stably expressing mtdsRed. (B) Representative histograms showing bivariate Annexin V/PI analysis (upper panel) and summary of cell number and apoptosis activation (lower panels) in Hep3B cells. Data are shown as mean ± SEM, n = 6-9. Statistical significance: *P < 0.05, **P < 0.01 ***P < 0.001 versus vehicle in the same experiment (one-way ANOVA and Newman-Keuls multiple comparison test); #P < 0.05, ##P < 0.01 versus the corresponding value in the experiment performed in the absence of 3MA (Student's t test). The positive control for apoptosis, STS 1 μM was statistically analyzed separately by Student's t test and it was always significant ***P < 0.001 (not shown).

Figure 8.

Mitochondrial alterations and autophagy in primary human hepatocytes treated with EFV for 24 hours. (A) Representative TEM images of cells treated with EFV 50 μM. Magnification 26,500×. (B) Representative fluorescence images of cells treated with EFV 50 μM (nuclei: blue-Hoechst; mitochondria: green-NAO). (C) WB showing LC3 expression in the presence or absence of 20 nM Bafilomycin A1 (representative image). (D) Summary of the densitometry data of the WB results for basal LC3-II expression (in the absence of Bafilomycin A1). Data shown as mean ± SEM, n = 3, statistical significance: *P < 0.05 versus vehicle (Student's t test). (E) Cell number in EFV-treated cells cotreated with 2.5 mM 3MA. Data shown as mean ± SEM, n = 4-5. Statistical significance: *P < 0.05, ***P < 0.001 versus vehicle in the same experiment (one-way ANOVA and Newman-Keuls multiple comparison test); #P < 0.05, versus the corresponding value in the experiment performed in the absence of 3MA (Student's t test).


Autophagy is a cellular self-digestion process crucial for cell differentiation and survival.25 All eukaryotic cells rely on constitutive autophagy to carry out the basal elimination of damaged organelles. In addition, this function is induced by a rapidly growing list of conditions including starvation, amino acid deprivation, radiation, ER stress, proteasome inhibition, cytokines, chemicals, hypoxia, and intracellular pathogens. Autophagy has been implicated in a variety of important physiopathological processes, such as neurodegeneration, cancer, viral infections, inflammatory disorders, and liver disease.26 The mitochondrion is one of the organelles that can become targets for autophagic degradation in a process known as mitophagy, which is specifically induced by nutrient deprivation, reduced ATP generation, mitochondrial membrane depolarization, triggering of the mitochondria permeability transition (MPT), and oxidative stress.27 In fact, compelling evidence has emerged indicating that the removal of mitochondria is a highly regulated and organelle-specific process, and mitophagic signaling has only very recently come to light.15

To our knowledge, the present study is the first to address the relationship between NNRTI-induced toxicity and induction of autophagy. We have documented the induction of autophagy and, in particular, mitophagy in hepatic cells treated with EFV, the most commonly used NNRTI. Nevirapine, the other NNRTI, was not evaluated, as previous studies in this model have shown that it lacks a direct mitochondrial effect.14

Autophagy was assessed using several approaches. We employed TEM to study mitochondrial morphology and to detect the presence of autophagic vacuoles, as this continues to be the most sensitive and widely employed technique for these purposes.23 We also studied LC3-II, the only protein known to be specifically localized to autophagic structures throughout the entire autophagic process, from the phagophore to the lysosomal degradation.28 Nevertheless, it is important to point out that increases in LC3-II levels have been associated not only with an enhanced autophagosome synthesis but also with a reduced autophagosome turnover. This is relevant to our results because, whereas moderate EFV concentrations (10 and 25 μM) triggered a normal autophagic flux, the highest concentration (50 μM), which produced severe mitochondrial damage, was associated with a delayed or an inhibited autophagic flux. Such an effect may be due to a reduced fusion between compartments and/or impaired lysosomal proteolysis. Interestingly, this may also explain the increased mitochondrial mass we observed in cells treated with the same concentration of EFV, because an impaired mitochondrial clearance can result in an apparently enhanced mass of these organelles. In connection with this, it is relevant to stress that this increase in the mitochondrial mass occurs in the absence of true mitochondrial biogenesis, as shown by the lack of changes in the mtDNA/nDNA ratio in EFV-treated Hep3B cells.13

Autophagy is related to cell death, but this relationship is still not well understood. Stress or injury signals can activate both autophagy and cell death pathways in which the role of the former can vary depending on the context.25, 29, 30 It is important to underline that the highest concentration of EFV employed in the present study has been reported to induce apoptotic cell death in Hep3B.13 How this is related to the autophagic stress that we describe herein is not fully known, but we can speculate that both phenomena are associated. Importantly, pharmacological inhibition of autophagy enhances the proapoptotic action of EFV. A complex relationship between autophagy and apoptosis has been suggested for several xenobiotics that induced both processes (imiquimod in basal cell carcinoma31 or oridonin in HeLa cells32) and, of note, in both cases the inhibition of autophagy promoted apoptosis which is in keeping with our results.

Our understanding of the role of autophagy in liver pathophysiology, especially regarding drug-induced hepatotoxicity, is limited.33, 34 However, sequestration of several subcellular compartments has been documented in hepatocytes under different conditions. Autophagy may play a role in three important aspects of hepatic physiopathology: organelle turnover, balance of nutrients and energy, and removal of misfolded/damaged proteins,33 and has been recently implicated in conditions such as liver ischemia-reperfusion injury, alcohol-related liver damage, hepatitis B/C infection, hepatocellular carcinoma, and nonalcoholic liver disease.33, 34 Interestingly, hepatocytes were an early model for mitophagy following MPT and loss of ΔΨm. Recent data suggest that autophagy facilitates cell survival in various conditions of liver injury, including drug toxicity34; mitophagy was found to reduce hepatotoxicity and steatosis associated with acute ethanol exposure,35 confer resistance to injury from menadione-induced oxidative stress,36 and promote survival of HepG2 cells against ginsenoside Rk1-induced apoptosis.37 Failure of this adaptive mechanism may lead to autophagic cell death. Our results add weight to this hypothesis, because the mitochondrial degradation detected in our model occurs as a rescue mechanism that promotes hepatic cell survival, as shown by the fact that its pharmacological inhibition leads to increased EFV-induced cell damage. Nevertheless, when a massive autophagic response is induced the degradation capacity of the cell is exceeded, and “autophagic stress” is produced.

Finally, there is growing evidence of a complex role of autophagy in viral infections including HIV38 and HBV/HCV,34 which is of special relevance in the light of our results. Hepatitis coinfections are very common among HIV patients and greatly enhance the hepatic toxicity of EFV.1, 2 In addition, there is evidence of autophagy induced by several protease inhibitors.39, 40, 41 Moreover, HIV patients usually receive concurrent medications that may be potentially hepatotoxic.1 All of this provides a picture of autophagic signaling/induction in which complex interactions take place between EFV and concomitant conditions which may ultimately influence liver function. This hypothesis could have major therapeutic importance and deserves further study.

In conclusion, our results reinforce the idea that compromising mitochondrial function induces autophagy and provide evidence that this process promotes cell survival in hepatic cells. We observed that crossing a threshold of mitochondrial dysfunction is associated with autophagic overload or autophagic stress, which severely limits the viability of cells. This complex effect could be involved in the hepatic toxicity associated not only with EFV but also with other drugs that interfere with mitochondrial function and, thus, may constitute a new mechanism implicated in hepatic damage.


We thank Mario Soriano Navarro (“Centro de Investigación Principe Felipe,” Valencia) for assistance with TEM and Brian Normanly for English language editing.