Induction of autophagy by depolarization of mitochondria

ABSTRACT

Mitochondrial dysfunction plays a crucial role in the macroautophagy/autophagy cascade. In a recently published study Sun et al. described the induction of autophagy by the membranophilic triphenylphosphonium (TPP)-based cation 10-(6′-ubiquinonyl) decyltriphenylphosphonium (MitoQ) in HepG2 cells (Sun C, et al. “MitoQ regulates autophagy by inducing a pseudo-mitochondrial membrane potential [PMMP]”, Autophagy 2017, 13:730-738.). Sun et al. suggested that MitoQ adsorbed to the inner mitochondrial membrane with its cationic moiety remaining in the intermembrane space, adding a large number of positive charges and establishing a “pseudo-mitochondrial membrane potential,” which blocked the ATP synthase. Here we argue that the suggested mechanism for generation of the “pseudo-mitochondrial membrane potential” is physically implausible and contradicts earlier findings on the electrophoretic displacements of membranophilic cations within and through phospholipid membranes. We provide evidence that TPP-cations dissipated the mitochondrial membrane potential in HepG2 cells and that the induction of autophagy in carcinoma cells by TPP-cations correlated with the uncoupling of oxidative phosphorylation. The mild uncoupling of oxidative phosphorylation by various mitochondria-targeted penetrating cations may contribute to their reported therapeutic effects via inducing both autophagy and mitochondria-selective mitophagy.

Mitochondria could be considered both as the “victims” of autophagy and as important regulators of signaling pathways that eventually cause autophagy. Hence, mitochondria contribute to prosurvival functions of autophagy such as hormesis, an adaptive response to toxins and other stressors. Furthermore, activation of autophagy can lead to the removal of damaged organelles (including mitochondria) and protein aggregates (such as polyglutamine proteins that cause Huntington disease). In addition, autophagy can provide substrates for fasting cells, thereby interfering with apoptosis and protecting against tissue damage. One of the well-known examples of hormesis is ischemic preconditioning (IPC) where short episodes of ischemia protect various organs against a later ischemic insult. The IPC-activated autophagy in heart, brain, kidney, and liver is due to the decrease in mitochondrial ATP production, whereas the inhibition of autophagy ameliorates the protection of these organs.¹ A decrease in the cellular ATP:AMP ratio may induce autophagy via the AMP-activated protein kinase (AMPK) acting either as the negative regulator of MTOR (mechanistic target of rapamycin kinase) or directly by phosphorylating the kinase ULK1 that is involved in the initiation of autophagy.²

Recently, Sun et al. have reported³ that autophagy in HepG2 cells could be stimulated by 10-(6′-ubiquinonyl) decyltriphenylphosphonium (MitoQ), a mitochondria-targeted antioxidant.⁴ Sun et al. showed that the addition of 1 μM MitoQ leads to the decrease in ATP:AMP ratio, stimulates AMPK and inhibits MTOR. To explain the diminishing of the ATP:AMP ratio, the authors suggested that the ATP synthesis was suppressed by “pseudo-mitochondrial membrane potential”. Sun et al. suggested that “MitoQ is adsorbed to the inner mitochondrial membrane; however, its cationic moiety remains in the intermembrane space, resulting in the addition of a large number of positive charges…. Furthermore, the positive charges in MitoQ replaced protons in the establishment of the electrochemical gradient across the inner membrane…. This distinct mitochondrial membrane potential (MMP) … was designated “pseudo-MMP” (PMMP)” (quoted from ref.³).

We think that the suggested accumulation of MitoQ molecules in the outer leaflet of the inner mitochondrial membrane with the cationic moieties of triphenylphosphonium (TPP⁺) facing the intermembrane space is physically implausible. The electric field of >200,000 V/cm over mitochondrial membranes (negative inside) would push the TPP⁺ cations towards the inner leaflet and, ultimately, inside the mitochondrial matrix. Application of potential to voltage-clamped phospholipid bilayers leads to the translocation of membranophilic cations from the positively charged monolayer to the negatively charged monolayer in a voltage-dependent manner with characteristic times in the range from 10⁻¹ sec (at temperatures above the phase transition of lipids)⁵ to 10–100 sec (at temperatures below the phase transition).⁶

The transmembrane electrophoresis of TPP⁺ and other “penetrating ions,” as driven by the transmembrane difference in electric potential (membrane potential, Δψ), was demonstrated by one of us (VPS) in the end of 1960s.^7,8 Later, with phospholipid bilayers, it was shown that the transmembrane concentration gradients of MitoQ, its plastoquinone-containing analog 10-(6′-plastoquinonyl) decyltriphenylphosphonium (SkQ1), and dodecyltriphenylphosphonium (C₁₂TPP), the analog of the cationic residue of MitoQ and SkQ1, causes the appearance of a Nerstian electric potential (approx. 60 mV per 10-fold gradient of a monovalent cation).⁹ It was also shown that membranophilic cations can diminish membrane potential not only by their own redistribution, but also by facilitating the transmembrane transport of free fatty acid (FFA) anions.¹⁰ The transport of FFA anions across the inner mitochondrial membrane is the rate-limiting step of the FFA-mediated uncoupling proton flux (see the scheme in Figure 1A). A FFA-dependent induction of the transmembrane proton flow by membranophilic cations was directly demonstrated in FFA-containing phospholipid bilayers, liposomes, isolated mitochondria, and yeast cells.¹⁰

Figure 1.

Membrane depolarization as a trigger of autophagy in mitochondria. (A) Scheme of the protonophoric action of membranophilic cations (adapted from ref.¹⁰ with permission). Membranophilic cations (R-TPP) facilitate the transmembrane translocation of anionic free fatty acids (FFA) by forming complexes with them. The symbols ⊖ and ⊕ correspond to the anionic and cationic groups of FFA and R-TPP, respectively. The letter “H” within a circle indicates the protonated COOH headpiece of a fatty acid. (B) Dissipation of Δψ by MitoQ in HepG2 cells. The conditions of the experiments were exactly the same as in Sun et al.³ Cells were incubated for 60 min at 37°C with 1 μM MitoQ and then for 30 min with 10 μM CCCP (Sigma-Aldrich, C2759). MitoQ was synthesized as described in ref.⁴ To measure Δψ the cells were stained with 100 nM TMRM (Thermo Scientific, T668) for 30 min. Fluorescence was analyzed using a Beckman Coulter FC500 flow cytometer (Beckman Coulter Inc., USA). Unimodal distribution close to Gaussian was observed in all samples and the X_mean ±SD (n = 3) values are plotted. a.u., arbitrary units. (C) Dissipation of Δψ by C₁₂TPP, SkQ1 and FCCP in RKO cells. Cells were incubated with C₁₂TPP, SkQ1 and FCCP (Sigma-Aldrich, C2920) for 60 min at 37°C and analyzed as in Figure 1B. C₁₂TPP and SkQ1 were synthesized as described in ref.⁹ (D) LC3 clustering in RKO cells in response to the SkQ1 addition. To follow LC3 clustering a GFP-LC3 fusion construct was expressed in RKO cells using lentivirus transfection. Cells were treated with SkQ1 for 24 h and analyzed using a fluorescence Axiovert 200 microscope (Carl Zeiss, Germany). Typical images are shown. Scale bar: 15 μm. (E) Quantification of the LC3 clustering in RKO cells in response to the addition of SkQ1 and C₁₂TPP. Cells were treated as in Figure 1D. GFP-LC3 clusters were counted in 100 cells/probe using automated counting software Kalaimoscope (Transinsight, Germany). Data are presented as X_mean ±SD; n = 3; *P < 0.05 (ANOVA) for treated vs control cells. (F) LC3 lipidation in RKO cells in response to the addition of SkQ1 and FCCP. Сells were incubated as in Figure 1D and subjected to immunoblot analysis using anti-LC3 antibody (Cell Signaling Technology, 2775) and anti-GAPDH antibody (as a loading control; Santa Cruz Biotechnology, sc-25778).

The Δψ−dependent accumulation of MitoQ (and the other TPP-containing compounds) in the mitochondrial matrix was shown in many works from different labs, as reviewed elsewhere.^4,9,11 Therefore, the addition of membranophilic cations should lead to the diminishing of membrane potential, but not to its increase.

To measure the “pseudo-mitochondrial membrane potential” Sun et al.³ have used the potential-sensitive cyanine dye 5,5',6,6'-tetrachloro-1,1',3,3'-tetraethylbenzimidazolylcarbocyanine iodide (JC-1). This green fluorescent cationic dye accumulates in mitochondria due to the membrane potential  and forms J-aggregates with red fluorescence. In the study of Sun et al.,³ MitoQ did not affect the red fluorescence of JC-1, which speaks against the ability of MitoQ to generate an additional “pseudo-mitochondrial membrane potential.” At the same time, Sun et al.³ showed that the green shift of JC-1 fluorescence, as caused by protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP), was smaller in the presence of MitoQ than in its absence. The authors attributed this difference to the contribution of “pseudo-mitochondrial potential” induced by MitoQ in the presence of CCCP. This set of intrinsically ambiguous observations, however, can be explained in several other ways. JC-1 was shown to form large aggregates that do not dissociate after depolarization of the membrane.¹² This property of JC-1 leads to some well-known artifacts such as mosaic heterogeneity of Δψ along a single mitochondrion.¹³ The earlier attempt to analyze the effects of MitoQ on Δψ by using JC-1 resulted in observations of either depolarization or hyperpolarization of mitochondria depending on the cell line.¹⁴ In addition, MitoQ and JC-1 can compete for ABCB1/Pgp170, a P-glycoprotein multidrug efflux pump. TPP⁺-containing cations are the substrates of ABCB1 and inhibit the efflux of other cationic substrates.^15,16 In addition, JC-1 is a substrate of ABCB1 as well,¹⁷ so that MitoQ, by blocking ABCB1, could stimulate the accumulation of JC-1 in the cells with active multidrug efflux pumps (such as HepG2) and promote the formation of JC-1 aggregates insensitive to CCCP. All these effects could additionally interfere with the shown ability of CCCP to induce proton release from lysosomes and acidify the cytoplasm.¹⁸ The named factors could account for the intrinsic discrepancies in the work of Sun et al.³

To make a long story short, we have repeated the experiment shown in Figure 3A of Sun et al.³ To avoid pitfalls of the JC-1-based assay, we used tetramethylrhodamine methyl ester (TMRM) as a fluorescent cation for measuring mitochondrial membrane potential.¹⁹ As shown in Figure 1B, MitoQ, C₁₂TPP, and CCCP dissipated Δψ in HepG2 cells; no particular interference between MitoQ and CCCP was observed.

That said, we agree with Sun et al.³ that MitoQ (and other TPP⁺-containing compounds) can induce autophagy. However, alternatively to Sun et al.,³ we attribute the effects to the partial uncoupling of oxidative phosphorylation. In support of our suggestion, we demonstrate that mitochondrial depolarization in response to the addition of the membranophilic cations C₁₂TPP and SkQ1, as well as the uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) correlated with the activation of autophagy in RKO carcinoma cells (Figure 1C-F). The membrane potential of mitochondria was measured by TMRM (Figure 1C). The autophagy was followed by the accumulation of MAP1LC3B/LC3B (microtubule associated protein 1 light chain 3 beta) clusters (Figure 1D,E) and LC3 lipidation (LC3-II formation) (Figure 1F). The data in Figure 1B-F suggest that the autophagy in the experiments of Sun et al.³ could be also induced by depolarization of mitochondria in response to the addition of MitoQ.

In our experiments, membranophilic cations C₁₂TPP and SkQ1 appeared more effective in dissipating Δψ than the dedicated protonophoric uncoupler FCCP (Figure 1B,C). These data are in agreement with earlier observations where C₁₂TPP, already at nanomolar concentrations, induces a measurable decrease in mitochondrial membrane potential in endothelial cells, being slightly more effective than SkQ1.²⁰ The high efficiency of C₁₂TPP and SkQ1 in dissipating the membrane potential could be caused by their selective accumulation in mitochondria, which is driven by the cumulative impact from the membrane potential at the plasma membrane and at the inner mitochondrial membrane.⁷

In the experiments presented in Figure 1F, the high doses of FCCP induced excessive accumulation of LC3-II (as compared to the effects of C₁₂TPP and SkQ1), which was presumably due to the inhibition of autophagic flux. Earlier it was shown that high doses of protonophoric uncouplers induce proton release from lysosomes and appear to hinder the lysosomal degradation of autophagosomal cargo in both yeast and mammalian cells.¹⁸ In contrast, membranophilic cations are unlikely to interfere with the autophagic flux. These cations selectively accumulate in mitochondria; their concentrations in the other cellular membranes are much lower, so that the artifacts inherent to traditional uncouplers are minimized.

Membranophilic cations can be considered as self-limiting or “mild” uncouplers because the increase in their concentration should result in lowering the mitochondrial membrane potential that drives their accumulation in mitochondria. The mild uncoupling is largely considered as a promising therapeutic approach because it minimizes the production of reactive oxygen species without notably decreasing the ATP yield.²¹ Induction of autophagy (and mitophagy) in response to the decrease in membrane potential could additionally contribute to the therapeutic effects of the membranophilic cation, which are observed with the models of kidney and brain ischemia/reperfusion injuries^22,23 as well as with the model of systemic inflammatory response syndrome.^20,24

Funding Statement

This work was supported by the Russian Scientific Foundation [grant number 17-14-01314].

Disclosure of Potential Conflicts of Interest

The authors report no conflicts of interest.