- Split View
-
Views
-
Cite
Cite
Nianyu Li, Elisa Oquendo, Roderick A. Capaldi, J. Paul Robinson, Yudong D. He, Hisham K. Hamadeh, Cynthia A. Afshari, Ruth Lightfoot-Dunn, Padma Kumar Narayanan, A Systematic Assessment of Mitochondrial Function Identified Novel Signatures for Drug-Induced Mitochondrial Disruption in Cells, Toxicological Sciences, Volume 142, Issue 1, November 2014, Pages 261–273, https://doi.org/10.1093/toxsci/kfu176
- Share Icon Share
Abstract
Mitochondrial perturbation has been recognized as a contributing factor to various drug-induced organ toxicities. To address this issue, we developed a high-throughput flow cytometry-based mitochondrial signaling assay to systematically investigate mitochondrial/cellular parameters known to be directly impacted by mitochondrial dysfunction: mitochondrial membrane potential (MMP), mitochondrial reactive oxygen species (ROS), intracellular reduced glutathione (GSH) level, and cell viability. Modulation of these parameters by a training set of compounds, comprised of established mitochondrial poisons and 60 marketed drugs (30nM to 1mM), was tested in HL-60 cells (a human pro-myelocytic leukemia cell line) cultured in either glucose-supplemented (GSM) or glucose-free (containing galactose/glutamine; GFM) RPMI-1640 media. Post-hoc bio-informatic analyses of IC50 or EC50 values for all parameters tested revealed that MMP depolarization in HL-60 cells cultured in GSM was the most reliable parameter for determining mitochondrial dysfunction in these cells. Disruptors of mitochondrial function depolarized MMP at concentrations lower than those that caused loss of cell viability, especially in cells cultured in GSM; cellular GSH levels correlated more closely to loss of viability in vitro. Some mitochondrial respiratory chain inhibitors increased mitochondrial ROS generation; however, measuring an increase in ROS alone was not sufficient to identify mitochondrial disruptors. Furthermore, hierarchical cluster analysis of all measured parameters provided confirmation that MMP depletion, without loss of cell viability, was the key signature for identifying mitochondrial disruptors. Subsequent classification of compounds based on ratios of IC50s of cell viability:MMP determined that this parameter is the most critical indicator of mitochondrial health in cells and provides a powerful tool to predict whether novel small molecule entities possess this liability.
Mitochondria play critical roles in cellular homeostasis by generating ATP for energy and regulating cell death (Mayer and Oberbauer, 2003). The importance of mitochondrial function in human health cannot be overstated due to the fact that patients with inherited mutations of mitochondrial DNA or nuclear genes encoding critical mitochondrial components demonstrated a variety of disorders (Atkuri et al., 2009; Schapira, 2006). Different target organs, such as the peripheral and central nervous system, skeletal muscle, heart, liver, kidney, gastrointestinal, and respiratory system, were impacted in these patients (DiMauro and Schon, 2003; Finsterer, 2007). Emerging evidence also implicates a role for mitochondrial disruption in drug-induced toxicity (Chan et al., 2005; Pereira et al., 2009). For example, mitochondrial perturbation is considered as a major contributing factor to the development of hepatic, cardiac, and skeletal muscle toxicities caused by administration of drugs that either carry black box label warnings or have been withdrawn from the market, such as troglitazone, tolcapone, and cerivastatin (Bova et al., 2005; Dykens and Will, 2007; Haasio et al., 2001; Kaufmann et al., 2006; Tirmenstein et al., 2002).
Given the relevance of mitochondrial perturbation in drug-induced toxicity in vivo, pharmaceutical companies are developing models to detect mitochondrial dysfunction in an effort to reduce the enormous costs that are associated with late stage drug attrition (Pereira et al., 2009). Several in vitro methods have been developed to estimate drug-induced mitochondrial perturbation (Marroquin et al., 2007; Rana et al., 2011). Biochemical assays using isolated mitochondria, either from animal organs or cultured cells, identify a drug's direct effect on mitochondria (Hynes et al., 2006; Porceddu et al., 2012). Cellular mitochondrial assays provide the opportunity to integrate a more complex cellular signaling environment that can impact mitochondrial function (Pereira et al., 2009). Additionally, measuring mitochondrial function in intact cells more closely mimics the complex state of internal cellular biochemical signaling interactions that are found in vivo. However, using intact cell models is fraught with challenges because mitochondrial perturbation quite often leads to cell death, making it difficult to distinguish cytotoxic compounds from those that disrupt mitochondria.
One of the in vitro techniques developed to identify drugs disrupting mitochondria is to use cells grown in glucose-free media (supplemented with galactose/glutamine; GFM). This method deprives cells from energy derived by glycolysis and forces them to survive solely on ATP generated via oxidative phosphorylation (OXPHOS) in the mitochondria, which leads to increased sensitivity to drugs disrupting mitochondrial function (Hakan et al., 2010; Xu et al., 2005). Comparison of viability IC50s, for many drugs that disrupted mitochondrial function in cells grown in both GFM and glucose-supplied media (GSM), revealed that cells are at least 3-fold or more susceptible to cytotoxicity in GFM as compared with GSM (Marroquin et al., 2007). Cytotoxicity, not mitochondrial activity, was the primary readout in these studies, and therefore drugs primarily disrupting mitochondrial function may not be efficiently differentiated from those that impact other cellular metabolic pathways (i.e., false positives) (Gohil et al., 2010). For instance, cells cultured in GFM may grow slower than those cultured in GSM, which will add more variables to the assay. One such example was C2C12 murine myoblast cells, cultured in GFM, that showed decreased growth rate and failed to differentiate into myotubes, which prevented researchers from effectively utilizing this model to study mitochondrial function (Elkalaf et al., 2013).
Monitoring mitochondrial membrane potential (MMP) of intact cells is a direct measurement of mitochondrial health. The mitochondrial respiratory chain transfers protons across the mitochondrial inner membrane to store energy for ATP generation. The resulting proton gradient generates an electrical gradient termed MMP (Chen, 1988), which contributes to 75% of eukaryotic cellular energy. The remainder is stored in the form of a chemical gradient (∼25%), i.e., pH gradient (Chen, 1988). Maintaining MMP is the fundamental requirement for healthy mitochondrial function. Numerous publications have demonstrated that disrupting mitochondrial function by either blockade of mitochondrial respiratory chain or interference with mitochondria-related death pathways leads to MMP loss (Chen, 1988; Choi and Lee, 2011; Isenberg and Klaunig, 2000; Kim et al., 2007; You and Park, 2010). Therefore, MMP is regarded as a key indicator of mitochondrial health in cells (Green and Van, 2011).
In addition to MMP, mitochondrial oxidative stress is also used to determine mitochondrial health. Mitochondria are the major source of ROS production in cells (Green and Van, 2011; Wallace, 1999). Results from many in vitro and in vivo studies on drugs suggested that mitochondrial disruption leads to increased mitochondrial reactive oxygen species (ROS) production, which subsequently contributed to the drug-induced toxicity (Lenaz, 2001). Under normal physiological condition, mitochondrial ROS are neutralized by antioxidants, such as reduced glutathione (GSH) in cells, where GSH is transformed to its oxidized form GSSG to negate the effects of ROS (Schafer and Buettner, 2001). However, prolonged exposure to elevated mitochondrial ROS can result in the depletion of GSH. Several recent studies showed that cellular GSH level is lower in patients with genetically inherited mitochondrial diseases compared with those of healthy individuals, suggesting that oxidative stress endpoints may also be key indicators of mitochondrial health (Atkuri et al., 2009).
In the present study, our goal was to identify signatures of mitochondrial dysfunction in cultured cells. We explored the potential use of these signatures to identify drugs that disrupt mitochondrial function. To achieve these goals, we developed a multi-parametric flow cytometry assay to directly monitor key endpoints associated with mitochondrial function in cells, including MMP, mitochondrial ROS production, intracellular GSH, and cell viability. A small set of established mitochondrial poisons and 60 marketed drugs, known to carry a range of effects on mitochondria, was selected for assay qualification. Our results indicate that when cells are cultured in GSM; disruption of MMP alone did not induce cell death, as long as cellular ATP generation via glycolysis was kept intact. Hierarchical cluster analysis of all measured endpoints further confirmed that specific MMP depolarization without the loss of cell viability was the key signature for mitochondrial disruptors. Therefore, specific depolarization of the MMP can be used to identify chemical entities targeting the mitochondrial respiratory chain.
MATERIALS AND METHODS
Cell culture
All cell culture media and supplements were purchased from Life Technology (Grand Island, NY). The human pro-myelocytic leukemia cell line, HL-60, was obtained from ATCC (Manassas, VA). Cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), 2-mM L-glutamine, and 2-mM penicillin-streptomycin.
In some experiments, HL-60 cells were cultured in glucose-free RPMI-1640 supplied with 10-mM galactose, 10% dialyzed fetal bovine serum, 2-mM L-glutamine, and 2-mM penicillin-streptomycin for at least 1 week prior to assays.
Measurement of MMP, mitochondrial ROS, cellular GSH, and cell viability with flow cytometry
HL-60 cells were treated with vehicle or test articles for 6–24 h in 96-well plates. After treatment, cells were pelleted by centrifugation at 250 × g for 2 min. Cells were then re-suspended in 100-μl PBS containing fluorescent dyes. To measure MMP, HL-60 cells were treated with 5-μM JC-1 (Life Technology) for 10 min at 37°C. To measure cell viability, cellular GSH, and mitochondrial ROS production, HL-60 cells were simultaneously treated with 0.033-μg/ml calcein-AM, 10-μM monobromobimane (mBBr), and 10-μM MitoSox (Life Technology) for 10 min at 37°C. After fluorescent dye-treatment, cells were pelleted at 300 g, supernatant removed, and re-suspended in 100-μl PBS. Details on the fluorescent dyes used in these assays: Calcein-AM is a cell-permeant dye that is used commonly to determine cell viability. In live cells, calcein-AM is converted to green-fluorescent calcein after acetoxymethyl ester hydrolysis by intracellular esterases (Grieshaber et al., 2010). Calcein was excited by a 488-nm argon laser and fluorescence was detected at 525 nm. JC-1 is a cationic carbocyanine dye that accumulates in healthy mitochondria. The dye exists as a monomer at low concentrations and yields green fluorescence (Cossarizza et al., 1993). At higher concentrations, the dye forms J-aggregates that emit orange fluorescence. JC-1 was excited by a 488 laser, the green fluorescence was detected at 525 nm, and the orange fluorescence was detected at ∼585 nm. mBBr, agent of choice for measuring GSH in human cells (Hedley and Chow, 1994), reacts non-enzymatically (and more readily) with GSH than with protein sulphydryls in cells and emits blue fluorescence. mBBR was excited by a 405-nm violet laser and the resulting emissions were detected at 450 nm. MitoSOX selectively targets mitochondria and is rapidly oxidized by ROS to an oxidized product, the red fluorescent product was excited by a 640-nm red laser and the fluorescence was detected at 675 nm (Mukhopadhyay et al., 2007). Fluorescence measurements were made either on a Becton Dickinson FACSCalibur (BD Biosciences, San Jose, CA) or a Becton Dickinson LSRII (BD Biosciences) to detect fluorescence intensity in single cells (Cossarizza et al., 2009) using vendor-specified band pass filters. For cell viability, cellular GSH, and MMP, “healthy” cells were defined by gates set up based on vehicle (0.1% dimethyl sulfoxide (DMSO))-treated cells (Figs. 1B, C, and E). After exposure, the decreased percentage of “healthy” cells was used to estimate effects of compounds. The mean fluorescence intensity of MitoSox was used to measure mitochondrial ROS production.
Measurement of cellular ATP content
HL-60 cells were plated at 1 × 106 cells/ml in 96-well plates. After treatment with test articles, cellular ATP content was measured by CellTiter-Glo kit (Promega, Madison, WI) following manufacturer's instructions. Percent inhibition was calculated by normalizing to 0.1% DMSO-treated cells (Vehicle control).
Data analysis
Half maximal inhibitory concentration (IC50; for MMP, cell viability, GSH, and cellular ATP generation) and half maximal effective concentration (EC50; for mitochondrial ROS) for all test articles were obtained by constructing a 10-point concentration-response curve (0–1mM; n = 2), utilizing GraphPad Prism for Windows version 5.0 (GraphPad Software, San Diego, CA). The resultant dendrogram captured similarities among the above-mentioned parameters based on the similarity score defined as 1-correlation.
RESULTS
Mitochondrial Respiratory Chain Inhibitors Decrease MMP While Maintaining Cell Viability in Glucose-Supplemented Culture
To measure drug-induced mitochondrial effects in cells, a flow cytometry-based multi-parametric assay was developed (Fig. 1). HL-60 cells, which had previously been shown to be sensitive to a variety of mitochondrial respiratory chain inhibitors, were used in the assay (Gonzalez-Aragon et al., 2007; Li et al., 2003; Sweet and Singh, 1995; Thompson et al., 1988). Three parameters closely associated with mitochondrial function in cells: MMP, mitochondrial ROS, and intracellular GSH level, were chosen. We used JC-1 to measure MMP, MitoSox to measure mitochondrial ROS, and mBBr to measure cellular GSH level. In addition, cell viability was monitored by calcein-AM staining. The signal intensities of MitoSox, mBBr, and calcein fluorescence were measured simultaneously. JC-1 fluorescence was measured in a separate flow cytometry assay due to the spectral overlap of JC-1 with calcein-AM and MitoSox. Healthy cells—viable cells with high intracellular GSH and MMP—were defined by “electronic” gates set up based on vehicle-treated cells (Fig. 1). The percent decrease in “healthy” cells was used to estimate concentration-dependent effects on parameters of interest in this study following exposure to various test articles. The mean fluorescence intensity of MitoSox was used to measure mitochondrial ROS production (Fig. 1). As an example, Figure 1 shows the concentration-dependent effect of the mitochondrial complex III inhibitor, antimycin A, which triggered loss of MMP, increased mitochondrial ROS production, and decreased GSH content and cell viability (Fig. 1).
To obtain a “signature” of mitochondrial disruptors, we first screened a set of compounds, which included several established mitochondrial respiratory chain disruptors—a complex I inhibitor: rotenone; a complex II inhibitor: thenoyltrifluoroacetone (TTFA); complex III inhibitors: antimycin A and myxothiazol; an ATP synthase inhibitor: oligomycin; and an uncoupler: trifluorocarbonylcyanide phenylhydrazone (FCCP). HL-60 cells were cultured in standard RPMI-1640 media (GSM) and exposed to mitochondrial disruptors for 6 h. As expected, all mitochondrial respiratory chain disruptors tested in the assay triggered loss of MMP in a concentration-dependent manner (Fig. 2), confirming that disruption of the mitochondrial respiratory chain abolished mitochondrial function. However, a decrease in MMP in HL-60 cells exposed to these compounds did not necessarily lead to overt cytotoxicity. On the contrary, viability was still maintained even if MMP was significantly decreased. For instance, over 90% of cells depolarized MMP while remaining viable when treated with 10–100μM of antimycin (Fig. 2D). When concentration-response curves of these compounds were plotted for cell viability and MMP and compared with each other, it clearly demonstrated that HL-60 cells were much more sensitive to disruption of MMP than loss of cell viability (Fig. 2). To quantify the different potencies of these compounds on MMP as opposed to cell viability, the concept of ratios between IC50s (cell viability vs. MMP) was introduced (Table 1). For all mitochondrial respiratory chain disruptors that we tested, ratios of IC50s between cell viability and MMP were at least 10 or higher (Table 1). For compounds such as myxothiazol, antimycin A, and FCCP, the ratios could reach over 100. These observations clearly indicated that mitochondrial respiratory chain disruptors, employed in this study, deplete MMP at concentrations that do not trigger cytotoxicity.
Mitochondrial Poisons Disrupt MMP
Compound name . | MMP IC50 (μM) . | Cell viability IC50 (μM) . | Cell viability IC50/MMP IC50 . |
---|---|---|---|
Rotenone | 20.3 | >1000 | 49.3 |
TTFA | 16.9 | 285 | 16.9 |
Myxothiazol | 1.2 | 183.4 | 152.8 |
Antimycin A | 2.1 | 290.6 | 138.4 |
Oligomycin | 15.6 | 181.8 | 11.7 |
FCCP | 2.3 | 293.2 | 127.5 |
Compound name . | MMP IC50 (μM) . | Cell viability IC50 (μM) . | Cell viability IC50/MMP IC50 . |
---|---|---|---|
Rotenone | 20.3 | >1000 | 49.3 |
TTFA | 16.9 | 285 | 16.9 |
Myxothiazol | 1.2 | 183.4 | 152.8 |
Antimycin A | 2.1 | 290.6 | 138.4 |
Oligomycin | 15.6 | 181.8 | 11.7 |
FCCP | 2.3 | 293.2 | 127.5 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Concentration-response curves of these compounds were generated for both MMP and cell viability in Figure 1. IC50 values of compounds on MMP, cell viability, and ratio of cell viability IC50/MMP IC50 are indicated. Results are means of two independent experiments.
Compound name . | MMP IC50 (μM) . | Cell viability IC50 (μM) . | Cell viability IC50/MMP IC50 . |
---|---|---|---|
Rotenone | 20.3 | >1000 | 49.3 |
TTFA | 16.9 | 285 | 16.9 |
Myxothiazol | 1.2 | 183.4 | 152.8 |
Antimycin A | 2.1 | 290.6 | 138.4 |
Oligomycin | 15.6 | 181.8 | 11.7 |
FCCP | 2.3 | 293.2 | 127.5 |
Compound name . | MMP IC50 (μM) . | Cell viability IC50 (μM) . | Cell viability IC50/MMP IC50 . |
---|---|---|---|
Rotenone | 20.3 | >1000 | 49.3 |
TTFA | 16.9 | 285 | 16.9 |
Myxothiazol | 1.2 | 183.4 | 152.8 |
Antimycin A | 2.1 | 290.6 | 138.4 |
Oligomycin | 15.6 | 181.8 | 11.7 |
FCCP | 2.3 | 293.2 | 127.5 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Concentration-response curves of these compounds were generated for both MMP and cell viability in Figure 1. IC50 values of compounds on MMP, cell viability, and ratio of cell viability IC50/MMP IC50 are indicated. Results are means of two independent experiments.
Glycolytic Pathways Are Critical for Maintaining Cell Survival During Disruption of Mitochondrial Respiratory Function
Extensive observations have been made, in the literature, that cells with impaired mitochondrial function preferentially utilize glycolysis over oxidative phosphorylation (Marroquin et al., 2007; Rodriguez-Enriquez et al., 2001) for energy. This is consistent with observations in Supplementary Data that the IC50 values of cellular ATP depletion induced by mitochondrial disruptors were significantly higher than those obtained for MMP disruption. As shown, the IC50 values of cellular ATP generation correlated more with loss of cell viability and were at least 10-fold or higher than IC50 values of MMP (Supplementary Data). These results emphasize that depletion of cellular ATP by itself cannot be used as the sole biomarker of mitochondrial function in glucose containing media.
A well-described approach in forcing in vitro cell cultures to spare glucose and use mitochondrial oxidative phosphorylation, to generate ATP, is to propagate them in GFM (Hakan et al., 2010; Marroquin et al., 2007; Xu et al., 2005). Because many mitochondrial disruptors that we tested did not trigger cell death at 6 h, exposure to these agents was extended to 24 h to observe for a possible overlap between MMP and cell viability. In GFM, disruption of mitochondrial function directly led to loss of cell viability (Fig. 3). The IC50 values for various mitochondrial disruptors in the cell viability and MMP assays were almost identical at 24 h post treatment (Fig. 3 and Supplementary Data). In comparison for these same compounds, the IC50s for MMP were much lower than those for cell viability at 24 h when HL-60 cells were cultured in GSM (Supplementary Data). This provides further proof that cells cultured in GSM can withstand impairment of mitochondrial function and preserve viability by maintaining ATP levels via glycolysis.
Disrupting MMP While Maintaining Cell Viability (in GSM) Identifies Compounds that Perturb Mitochondrial Function
Collectively, these data point to an important finding that disruption of mitochondrial function (triggering loss of MMP) can occur in the absence of cytotoxicity (in GSM) due to the fact that anaerobic glycolytic mechanisms can provide cellular ATP sufficient for cell survival. We further investigated whether cytotoxic compounds acting primarily on non-mitochondrial targets had similar effects on MMP and cell viability, given that mitochondria also play an important role in cell death pathways. To address this question, HL-60 cells were treated with the following cytotoxic compounds: terfenedine, paroxetine, menadione, and camptothecin. All the compounds mentioned above have documented cytotoxic effects on various cell types, but do not target mitochondrial respiratory chain components. The results indicated that these compounds, unlike mitochondrial respiratory chain disruptors, did not trigger loss of MMP, rather they induced cytotoxicity and loss of MMP with similar potency (Fig. 4 and Supplementary Data). These results indicate that cytotoxic compounds that primarily target non-mitochondrial pathways will not preferentially depolarize MMP.
Mitochondrial Respiratory Chain Uncouplers and Inhibitors (of Mitochondrial Respiratory Chain Complexes I, II, and III) Increase Mitochondrial ROS Production
When mitochondrial respiratory chain is disrupted, excess ROS is produced (Drose and Brandt, 2008; Schapira, 2006). Major sites for ROS formation reside in mitochondrial complexes I and III (Drose and Brandt, 2008). Inhibition of mitochondrial complex II does not trigger ROS production directly from complex II; rather, it triggers complex III-dependent ROS production (Chen et al., 2007; Drose and Brandt, 2008). In agreement with previous findings, we demonstrated that mitochondrial complex I, II, and III inhibitors (rotenone, TTFA, and antimycin A/myxothiazol) and a respiratory chain uncoupler (FCCP) triggered mitochondrial ROS production in HL-60 cells when cultured in GSM (Table 2). Conversely, a complex V (ATP synthase) inhibitor-oligomycin had mild effects on mitochondrial ROS generation (≤20% increase) at concentrations that depolarized MMP (Table 2). For complex I/II/III inhibitors, EC50s of ROS generation were comparable to IC50s of MMP depolarization; MMP depolarization and ROS generation were at least 10-fold lower than respective IC50s for cell viability (Table 1). These findings demonstrate a correlation between initial disruption of mitochondrial function and subsequent mitochondrial oxidative stress responses. Peak mitochondrial ROS production induced by rotenone was approximately 2-fold above basal level (ROS production from 0.1% DMSO-treated cells; Table 2). Blocking electron transfer at complex II by TTFA, or at complex III by antimycin A, generated more than a 3-fold increase in ROS above basal level (Table 2). Complex III inhibitor myxothiazol, which blocks electron transfer further downstream of the antimycin A-binding site in complex III (Drose and Brandt, 2008), increased basal ROS production by 2-fold (Table 2). The magnitude of increases in ROS production observed in these cell-based studies are strikingly similar to a recent report using sub-mitochondrial particles, which showed a 2–3-fold mitochondrial ROS increase induced by rotenone, myxothiazol, TTFA, or antimycin A treatment (Drose and Brandt, 2008). These results further confirmed the specificity of mitochondrial ROS detection by MitoSox in intact cells.
Effects of Classical Mitochondrial Disruptors on Mitochondrial ROS Production
Compound name . | Induction of mitochondrial-ROS production Mito-SOX EC50 (μM) . | Mito-SOX max response (fold increase vs. 0.1% DMSO[vehicle]-treated cells) . |
---|---|---|
Rotenone | 10.5 | 2.4 |
TTFA | 18.0 | 3.4 |
Myxothiazol | 3.3 | 2.3 |
Antimycin A | 17.7 | 3.6 |
Oligomycin | >1000 | 1.2 |
FCCP | 4.5 | 2.9 |
Compound name . | Induction of mitochondrial-ROS production Mito-SOX EC50 (μM) . | Mito-SOX max response (fold increase vs. 0.1% DMSO[vehicle]-treated cells) . |
---|---|---|
Rotenone | 10.5 | 2.4 |
TTFA | 18.0 | 3.4 |
Myxothiazol | 3.3 | 2.3 |
Antimycin A | 17.7 | 3.6 |
Oligomycin | >1000 | 1.2 |
FCCP | 4.5 | 2.9 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Effects of compounds on mitochondrial ROS production were measured by MitoSox. EC50s and max fold increase of mitochondrial ROS are shown. Results are means of two independent experiments.
Compound name . | Induction of mitochondrial-ROS production Mito-SOX EC50 (μM) . | Mito-SOX max response (fold increase vs. 0.1% DMSO[vehicle]-treated cells) . |
---|---|---|
Rotenone | 10.5 | 2.4 |
TTFA | 18.0 | 3.4 |
Myxothiazol | 3.3 | 2.3 |
Antimycin A | 17.7 | 3.6 |
Oligomycin | >1000 | 1.2 |
FCCP | 4.5 | 2.9 |
Compound name . | Induction of mitochondrial-ROS production Mito-SOX EC50 (μM) . | Mito-SOX max response (fold increase vs. 0.1% DMSO[vehicle]-treated cells) . |
---|---|---|
Rotenone | 10.5 | 2.4 |
TTFA | 18.0 | 3.4 |
Myxothiazol | 3.3 | 2.3 |
Antimycin A | 17.7 | 3.6 |
Oligomycin | >1000 | 1.2 |
FCCP | 4.5 | 2.9 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Effects of compounds on mitochondrial ROS production were measured by MitoSox. EC50s and max fold increase of mitochondrial ROS are shown. Results are means of two independent experiments.
Disruption of Mitochondrial Function does not Directly Cause Loss of Cellular Glutathione Levels In Vitro
One would assume that cellular GSH levels would decrease as a result of an increase in oxidative stress due to mitochondrial respiration disruption via complexes I–III, as reported in patients with mitochondrial diseases (Atkuri et al., 2009). However, in vitro exposure for 6 h to mitochondrial respiratory chain disruptors: rotenone, antimycin A, and oligomycin, did not decrease intracellular GSH at equivalent concentrations that disrupted MMP (Table 3). These compounds depleted cellular GSH primarily due to cytotoxic effects at very high concentrations (Table 3).
Effects of Classical Mitochondrial Disruptors on Cellular GSH Level
Compound name . | Cellular GSH IC50 (μM) . |
---|---|
Rotenone | 518.3 |
TTFA | >1000 |
Myxothiazol | >1000 |
Antimycin A | 227.8 |
Oligomycin | 160.9 |
FCCP | >1000 |
Compound name . | Cellular GSH IC50 (μM) . |
---|---|
Rotenone | 518.3 |
TTFA | >1000 |
Myxothiazol | >1000 |
Antimycin A | 227.8 |
Oligomycin | 160.9 |
FCCP | >1000 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Effects of compounds on cellular GSH content were measured by mBBr staining. Cellular GSH IC50s are shown. Results are means of two independent experiments.
Compound name . | Cellular GSH IC50 (μM) . |
---|---|
Rotenone | 518.3 |
TTFA | >1000 |
Myxothiazol | >1000 |
Antimycin A | 227.8 |
Oligomycin | 160.9 |
FCCP | >1000 |
Compound name . | Cellular GSH IC50 (μM) . |
---|---|
Rotenone | 518.3 |
TTFA | >1000 |
Myxothiazol | >1000 |
Antimycin A | 227.8 |
Oligomycin | 160.9 |
FCCP | >1000 |
Notes. HL-60 cells were cultured in RPMI-1640 (GSM) and treated with either vehicle (0.1% DMSO) or various compounds for 6 h. Effects of compounds on cellular GSH content were measured by mBBr staining. Cellular GSH IC50s are shown. Results are means of two independent experiments.
Identifying Drugs that Disrupt Mitochondrial Function
To further characterize key signatures of mitochondrial disruption, 60 marketed drugs from a variety of therapeutic areas were selected and tested in HL-60 cells cultured in GSM. Some of these drugs (e.g., amiodarone and troglitazone) are known to disrupt mitochondrial function. Effects of these compounds on MMP, cell viability, mitochondrial ROS production, and cellular GSH level were determined, and IC50s or EC50s were calculated (Supplementary Data). A hierarchical clustering analysis of the data demonstrated that IC50s between GSH depletion and loss of cell viability were highly correlated, which further confirmed that loss of cell viability occurred concomitantly with the loss of GSH in vitro (Fig. 5). However, overall profiles for loss of MMP or increase in mitochondrial ROS production are quite different to cell viability resulting in relatively low correlations between either MMP IC50s or mitochondrial ROS EC50s to viability IC50s (Fig. 5).
To find possible associations between marketed drugs and classical mitochondrial disruptors, an unsupervised K-mean-based hierarchical clustering approach was used to group all compounds based on the similarity of their assay profiles (Fig. 5). This unsupervised analysis grouped compounds into nine clusters, as indicated in Figure 5. Clusters 1–4 are comprised of 12 compounds, all of which have viability:MMP ratios >3. However, discernible differences were evident among these clusters vis-à-vis MMP depolarization and ROS production. Compounds grouped in clusters 1, 2, and 4 not only depolarized MMP but also induced mitochondrial ROS production in cells. Cluster 3 contained compounds that solely depolarized MMP, but did not elicit ROS production. Additionally, the compounds in cluster 4 had lower potencies on MMP (IC50 ≥ 100μM) compared with those in clusters 1 and 2 (IC50 ≤ 50μM). It is critical to point out that all 12 compounds included in these four clusters are either established mitochondrial disruptors or drugs that are reported to disrupt mitochondrial function either in vitro or in vivo: amiodarone, nelfinavir, etoposide, troglitazone, flutamide, and tolcapone (Duan et al., 2003; Fau et al., 1994; Felber and Brand, 1982; Haasio et al., 2002; Lee et al., 1999; Narayanan et al., 2003; Spaniol et al., 2001; Tirmenstein et al., 2002).
Compounds in clusters 5–9, had viability:MMP ratios ≤3 (Fig. 5); however, adriamycin, in cluster 5, induced mitochondrial ROS production, without significantly affecting cell viability at 6 h. Compounds in clusters 6 and 7 disrupted MMP and decreased cell viability at similar potencies. The difference between compounds in cluster 6 compared with those in cluster 7 was potency on various cellular endpoints. IC50s of MMP/viability/GSH inhibition or EC50s of ROS induction for most compounds in cluster 6 were less than 50μM, whereas those for compounds in cluster 7 were generally higher than 50μM. Clusters 8 and 9 included drugs that had little effect in any of the four parameters measured.
DISCUSSION
There is a large body of evidence suggesting that disruption of mitochondrial energy production contributes to drug-induced toxicity (Chan et al., 2005; Dykens and Will, 2007; Wallace, 1999). In this study, we introduced a multi-parametric mitochondrial function assay to explore endpoints closely associated with mitochondrial activity in cells. After systematically screening a panel of established mitochondrial toxicants and cytotoxic compounds, we conclude that the most sensitive marker for mitochondrial disruption was the loss of MMP at non-cytotoxic concentrations (as reflected by the IC50 ratio between viability and MMP) for cells grown in GSM.
It is also identified that the underlying mechanism for mitochondrial disruptors to deplete MMP in the absence of cell death was by replenishing cellular ATP level via glycolysis. In theory, the loss of MMP imposes two types of threats to cell survival: firstly, the immediate loss of ATP generation from mitochondrial OXPHOS (Tsujimoto, 1997), and secondly, the release of pro-apoptotic proteins. It has been a long-standing observation that cells can have elevated rates of glycolysis when cultured in media supplemented with glucose, also known as Crabtree effect (Marroquin et al., 2007; Rodriguez-Enriquez et al., 2001). For instance, glycolysis was reported to account for over 90% of the ATP generated in HL-60 cells cultured in RPMI-1640 containing ∼11-mM glucose (Hakan et al., 2010). Glycolysis can be further upregulated when cells were treated with mitochondrial respiratory chain inhibitors or cultured in anaerobic conditions (Hakan et al., 2010; Xu et al., 2005). In addition, many cells (including HL-60 cells) could be chronically treated with ethidium bromide to deplete their mitochondria DNA. These so-called ρ° cells were completely deficient of OXPHOS. However, these cells can still survive and replicate using ATP derived solely from glycolysis, further lending support to the hypothesis that glycolysis alone is sufficient to support energy needs for cell survival in vitro (Appleby et al., 1999; Pelicano et al., 2006). In agreement with these findings, our results clearly demonstrated that glycolysis was the major mechanism for cells to evade cell death upon mitochondrial OXPHOS inhibition. Furthermore, this mechanism can be used to identify compounds that disrupt mitochondrial function in vitro. Diminishing glycolysis by culturing cells in glucose-free media drastically sensitized HL-60 cells to OXPHOS inhibitors. When cultured in GFM, MMP depolarization following exposure to mitochondrial toxicants was accompanied by immediate cytotoxicity in HL-60 cells. In contrast, HL-60 cells were able to survive in GSM despite complete depolarization of MMP. Another threat to cell survival via mitochondrial disruption is the release of many pro-apoptotic proteins from mitochondria. Mitochondria play an essential role in apoptosis by harboring many proteins that are responsible for initiation and amplification of apoptotic signals, such as cytochrome c, Smac/Diablo, AIF, endonuclease G, and serine protease HtrA2/OMI (Fuchs and Steller, 2011; Saelens et al., 2004; Wang, 2001). Among these proteins, cytochrome c is a component of the mitochondrial respiratory chain (Fuchs and Steller, 2011), AIF was found to be closely associated with OXPHOS function as well (Perciavalle et al., 2012; Pospisilik et al., 2007; Vahsen et al., 2004). During apoptosis, release of these proteins from the mitochondria results in the loss of important components of respiratory chain. Not surprisingly, mitochondrial depolarization is commonly observed during apoptosis and widely recognized as an important apoptotic biomarker (Green and Van, 2011; Ly et al., 2003). However, less clear is whether mitochondrial depolarization is destined to induce cell death when cellular ATP production can be compensated by glycolysis. By systematically testing six classical inhibitors that deplete MMP via disruption of various mitochondrial respiratory chain components, our data strongly suggested that mitochondrial depolarization does not necessarily induce cell death as long as the glycolysis is intact. These results are in agreement with an earlier finding that dissipation of MMP by the mitochondrial respiratory chain un-coupler, CCCP, did not spontaneously induce apoptosis (Finucane et al., 1999). Mechanisms by which loss of MMP that do not necessarily lead to apoptosis are still not well understood. However, one can speculate that it could relate to maintenance of a healthy population of mitochondria—an essential pre-requisite for cellular homeostasis. In most cases, elimination of damaged mitochondria (either due to loss of MMP or generation of ROS) via autophagic delivery to lysosomes may not only limit mitochondrial DNA mutations but could also circumvent cell death mechanisms. (Narendra et al., 2008; Wang and Klionsky, 2011).
Several marketed drugs also had high IC50 ratios between viability and MMP. A non-supervised K-means clustering approach grouped these drugs with classical mitochondrial disruptors, further supporting the observation that the IC50 ratio between viability and MMP was the key feature for identifying mitochondrial disruptors. This ratiometric approach also revealed that marketed drugs had lower viability:MMP IC50 ratios (≤10-fold) compared with established mitochondrial disruptors (≥10-fold), which is a reflection of the fact that most marketed drugs target mitochondria through their “off-target” activities. For instance, amiodarone was developed as an anti-arrhythmic agent. However, several studies suggested that the drug caused liver injury due to its inhibitory effect on mitochondrial respiratory chain complexes I and II (Fromenty et al., 1990; Papiris et al., 2010; Spaniol et al., 2001). In agreement with these observations, amiodarone had a viability:MMP IC50 ratio around 3.5 and depolarized MMP at 20.6μM, suggesting specific disruption of OXPHOS. The antiretroviral protease inhibitor, nelfinavir, was reported to inhibit MMP in isolated mouse liver mitochondria (Porceddu et al., 2012). Similar results were observed in our assays, providing the first piece of evidence that this drug could deplete MMP in cells. Other drugs identified by our assays included weak mitochondrial complex I inhibitors troglitazone and flutamide (Bova et al., 2005; Fau et al., 1994; Kashimshetty et al., 2009; Narayanan et al., 2003), weak respiratory chain uncoupler tolcapone (Korlipara et al., 2004), and mitochondrial permeability transition inducer etoposide (Custodio et al., 2001; Hande, 1998; Robertson et al., 2000). In summary, our results indicated that comparison of IC50s of viability:MMP can successfully identify drugs disrupting mitochondrial function.
Mitochondrial ROS production alone was insufficient to predict mitochondrial disruption. Mitochondrial respiratory chain uncouplers or complex I, II, and III inhibitors tended to increase basal mitochondrial ROS production. These compounds include FCCP, rotenone, TTFA, antimycin A, myxothiazol, amiodarone, troglitazone, and flutamide. Some mitochondrial disruptors did not trigger mitochondrial ROS production and those included oligomycin, nelfinavir, and etoposide. The association between mitochondrial ROS production and mitochondrial disruption became even more unclear in the case of adriamycin and idarubicin, both of which belong to the anthracycline antibiotics family known to induce robust ROS production via the formation of semiquinone free radicals mediated by mitochondrial NADH dehydrogenase (Liu et al., 2001; Malisza and Hasinoff, 1996; Winterbourn et al., 1985). Adriamycin induced an 8-fold increase in mitochondrial ROS production in HL-60 cells at 6 h. However, such profound increases in mitochondrial ROS production did not result in concomitant MMP depolarization at similar concentrations. One possible explanation is that the antioxidants in mitochondria may be able to counter-balance the excessive ROS production for a short period of time after which sustained ROS production will compromise cell viability; adriamycin triggered severe cytotoxicity at 24 h (IC50 less than 1μM, data not shown). Idarubicin induced mitochondrial ROS production at an even lower EC50 compared with adriamycin and triggered severe cytotoxicity at 6 h and was grouped with many cytotoxic compounds by the K-means clustering analysis in group 6.
Although many of the drugs tested in this study increased mitochondrial ROS production, it was surprising to find that they did not induce a decrease in total cellular GSH content. GSH is known to be one of the major endogenous antioxidants in cells (Scholz et al., 1997). During oxidative stress, GSH is transformed to the oxidized form (GSSG) to counter-balance increased amounts of cellular ROS (Schafer and Buettner, 2001). GSH is lower in patients with genetically inherited mitochondrial diseases compared with healthy individuals, ostensibly due to the prolonged exposure to oxidative stress caused by mitochondrial dysfunction (Atkuri et al., 2009). However, a specific decrease of GSH was not observed in HL-60 cells treated with known mitochondrial disruptors in our assays. One possible explanation is that the mitochondria-generated ROS was not sufficient to influence overall redox homeostasis of cells, at least over a short period of time. Cells in mitochondrial deficient patients were exposed to mitochondrial oxidative stress for their life time, events which are hard to recapitulate in vitro.
Tamoxifen and ketoconazole were found to inhibit OXPHOS activity in isolated mitochondria (Cardoso et al., 2001; Hynes et al., 2006; Rodriguez and Acosta, 1996). However, both compounds are also known to have strong cytotoxic effects on cells. Tamoxifen was known to induce cell death through non-estrogen receptor mediated mechanisms (Gelmann, 1996). Ketoconazole was also found to have cytotoxic effects on cells via mechanisms yet to be identified (Gelmann, 1996; Rochlitz et al., 1988). In our assay, both tamoxifen and ketoconazole triggered loss of cell viability and MMP with similar potency, suggesting that the primary effect of these compounds on cells was cytotoxicity.
HL-60 cells are human acute leukemia cells that have been previously shown to be sensitive to mitochondrial complex I, II, III, IV, and V inhibitors (Gonzalez-Aragon et al., 2007; Li et al., 2003; Sweet and Singh, 1995; Thompson et al., 1988). As a suspension cell line, HL-60 cells were suitable to high-throughput flow cytometry assays and could be adapted to glucose free culture (GFM). Given that many tumor cells or primary cultured cells maintain glycolysis capability, one would expect that mitochondrial disruptors could depolarize MMP at non-cytotoxic concentrations in other cells cultured in GSM. Although not systematically studied, data from other groups and from our lab indicated that antimycin A can depolarize MMP at non-cytotoxic concentrations in a variety of cell lines cultured in GSM, including Jurkat T cells, KMS-12-BM, Toledo cells (data not shown), and H9C2 rat cardiomyoblast cells (Rana et al., 2011).
Similar to many other in vitro assays, HL-60 cells do not express the complete set of drug metabolizing enzymes; therefore, drugs disrupting mitochondrial function via their metabolites can go undetected. For instance, acetaminophen induces GSH depletion and mitochondrial damage associated with its liver toxicity by forming reactive intermediate N-acetyl-p-benzoquinone imine (Nelson, 1990). However, both GSH and mitochondrial effects of acetaminophen were not observed in HL-60 cells, most likely due to the lack of corresponding metabolic enzymes. In addition, many compounds bind to serum proteins (HL-60 cells were cultured in media containing 10% FBS) and can thus limit the amount of free drug available to cells in culture media (Smith et al., 2010). Therefore, it is anticipated that potencies of many drugs to affect mitochondrial function are likely lower in this assay compared to assays performed under serum-free conditions. It should also be borne in mind that when administered in vivo, all drugs will be exposed to proteins present in the body. Results from serum-containing cultures will closely resemble in vivo situation than those from serum-free assays.
The primary goal of the current study was to identify a sensitive and reliable marker of drug-induced mitochondrial dysfunction, while not trying to bin mitochondrial disruptors based on signaling mechanisms. It must also be borne in mind that current experimental design was out-of-scope for investigation of whether MMP depolarization was either due to chemicals directly targeting mitochondria or secondary to other primary mechanisms of cytotoxicity. As indicated in Table 2, all classical mitochondrial disruptors had ratios of IC50s between cell viability and MMP at least 10 or higher, although their inhibition sites were known to be different.
In conclusion, cytomics-based multi-parametric mitochondrial function assays (CMSAs) can be used as a useful tool to identify small molecule drug candidates that potentially disrupt mitochondrial function in cells. We demonstrated that specific disruption of MMP while maintaining cell viability in HL-60 cells cultured in GSM can provide valuable information about effects of drugs on mitochondrial function in cells. Mitochondrial ROS could further differentiate mitochondrial disruptors based on their specific inhibition mechanisms. These results also demonstrate that caution needs to be taken when measuring a test article's effect on MMP in a cell-based system. Glycolysis is critical to the survival of cells when OXPHOS is blocked. Therefore, it will be difficult to distinguish specific mitochondrial disruptors from cytotoxic compounds when cells either have impaired glycolytic mechanisms or are cultured in GFM. The CMSA platform measures endpoints directly associated with mitochondrial function in cells. When combined with data from other mitochondria function assays, such as oxygen consumption, differential cytotoxicity measurements between GSM and GFM cultures, biochemical assays from isolated mitochondria, the CMSA platform can help deliver comprehensive assessments of drug effects on mitochondria.
FUNDING
This work was supported by Amgen Inc.
We are very grateful to Paul Acton and Stacie Clark for their critical review and editing of this manuscript. We would like to thank Yusheng Qu, Hugo Vargas, Jeff Lawrence, and Chuck Qualls for their helpful advice and support. We also thank Hema Ingle for her excellent Information Systems support.
REFERENCE
Comments