To gain more insight into the mechanisms underlying ibipinabant-induced myotoxicity, we used C2C12 murine myoblasts as a cell model. Already after 24 hours of exposure to increasing concentrations of ibipinabant, cell viability was significantly (P=1.61·10^-7) decreased to 73 ± 5% at the highest concentration tested (100 μM, Fig. 1A). After 48 hours of exposure only 33 ± 4% of the cells remained viable at this concentration (Fig. 1B). The validity of our model was confirmed by the potent inhibition of cell viability by the known mitochondrial toxicant etoposide. At the highest concentration of 100 μM 42 ± 6% cells remained viable after 24 hours (Fig. 1A), which further decreased to 7 ± 3% after 48 hours (Fig. 1B).

Next, we determined the effects of ibipinabant on mitochondrial function by measuring cellular ROS generation. A more than 2-fold increase could already be observed after 8 hours exposure with the highest ibipinabant concentration of 100 μM compared to the vehicle treated C2C12 myoblasts (Fig. 2A,B). We also investigated maximal mitochondrial ATP production capacity after permeabilisation of the C2C12 myoblasts (Fig. 2C). As indicated by the low IC₅₀ value, a rapid decrease of ATP production capacity was observed after 4 hours incubation with increasing concentrations of ibipinabant (Fig. 2C).

A possible explanation for a decrease in ATP production capacity, accompanied by an increased generation of cellular ROS, could be inhibition of one of the enzyme complexes of the respiratory chain (complexes I–V). To examine this assumption we measured the oxygen consumption in C2C12 myoblasts after 4 hours of pre-incubation with ibipinabant (1–100 μM). In intact cells we did not observe a statistically significant effect of ibipinabant (see Supplementary Fig. S1 online). To investigate whether one of the specific complexes might be affected by ibipinabant under maximal stimulation, we used complex specific substrates to measure oxygen consumption in permeabilised C2C12 myoblasts (see Supplementary Fig. S1 online). However, none of the specific complex-driven respiratory rates was different from control. To be able to exclude any influence of 4 hours ibipinabant treatment on these complexes we also determined their catalytic activity and in agreement with the absence of effects on oxygen consumption, no effects could be observed on the enzyme activities (see Supplementary Fig. S1 online).

Uncoupling of oxidative phosphorylation is one of the explanations for the absence of an effect of ibipinabant on the enzyme complexes of the respiratory chain with a decreased ATP production¹⁶. The respiratory chain and mitochondrial ATP production, which together are referred to as oxidative phosphorylation, are coupled through the mitochondrial membrane potential (Ψ_m ), providing the proton motive force needed for ATP production at mitochondrial complex V. Indeed, Ψ_m appeared to be increased after 4 hours of incubation with 100 μM ibipinabant (Fig. 3A). An increase in Ψ_m could already be observed at low ibipinabant concentrations (EC₅₀ 1.5 95%−CI: 0.7–3.1 μM) (Fig. 3B). Such an increased Ψ_m could be indicative of a decreased function of complex V in intact mitochondria, following from the decreased ATP production capacity. Together with the absence of an effect on the catalytic activity of complex V, under excess substrate conditions (see Supplementary Fig. S1 online), these findings point at a decreased ADP availability in the mitochondrial matrix.

To address this possibility, in silico off-target predictions were performed to search for similarities between protein-ligand binding sites with the structure-based pharmacophore method KRIPO, possibly identifying off-targets involved in mitochondrial ADP/ATP exchange¹⁷. Each protein-ligand binding site is defined via the ligand of the complex followed by translation of the properties of the surrounding amino acids of the proteins into pharmacophore features, which are stored in bit strings called fingerprints. The fingerprints allow fast computational comparison of all publically available structures in the Protein Data Bank (pdb)¹⁸. Since for CB1R no experimental structure is available, a homology model of the receptor was build. From the list of similar targets obtained by KRIPO, targets that could play a key role mitochondrial ADP/ATP exchange (adenine nucleotide translocase 1 (ANT1), voltage-dependent anion channel (VDAC), and mitochondrial complex V) were manually selected.

Of these three targets, an effect of ibipinabant on mitochondrial complex V could be excluded from our previous observations (see Supplementary Fig. S1 online). To investigate whether ibipinabant can actually fit into ANT1, the drug was docked into the binding pocket of the protein binding sites (Fig. 4). In the co-crystallised complex the binding site is occupied by the inhibitor carboxyatractyloside (Fig. 4B), a ten times more potent inhibitor compared to atractyloside, which was used in the experiments described below¹⁹. These docking runs suggested that binding of ibipinabant to ANT1 appears likely, and a possible binding mode is illustrated (Fig. 4C). Whether ibipinabant would fit into VDAC could not be determined, because reliable docking was not possible due to the structural properties of the channel, e.g. having a large pore allowing unselective transport of many compounds.

The in silico predictions were experimentally verified by measuring ADP uptake into isolated bovine heart mitochondria (Fig. 5). This resulted in a significant (p = 0.041) decrease of mitochondrial VDAC-dependent ADP import after direct ibipinabant application (Fig. 5B). A similar effect could be observed on the ANT-dependent mitochondrial ADP uptake (Fig. 5A). Such an effect on both transmembrane transporters could be explained by the formation of a complex spanning the inner and outer mitochondrial membrane, as described previously²⁰. The observations in intact bovine heart mitochondria (containing both ANT and VDAC), could therefore not distinguish which of these transporters contributed to the effects of ibipinabant on ATP production and Ψ_m.

To investigate whether the observed ibipinabant-induced decrease in mitochondrial ADP/ATP exchange was due to inhibition of either ANT or VDAC, C2C12 mitoplasts (mitochondria without outer membrane) were isolated to exclude any effects of VDAC, followed by determination of the maximal CI- and CII-specific respiratory rates (Fig. 6). Although CI-specific respiration was only moderately affected after direct ibipinabant addition, a significant (P=0.038) inhibition was found for CII-specific respiratory rates. These effects were also in line with the tendency towards a decreased routine respiration (see Supplementary Fig. S1 online), which also indicated that mitoplasts were more sensitive than intact mitochondria to ibipinabant-induced respiratory inhibition. It also provided a firm indication that ibipinabant mainly reduced respiration by inhibition of the ANT-mediated ADP/ATP exchange.

Finally, we compared the inhibitory effects of ibipinabant on mitochondrial ADP import with the closely related 3,4-diarylpyrazoline derivative, CB23 (Fig. 7B). Replacement of the chlorophenyl group of ibipinabant with a 1,1-dioxo-thiomorpholine group (Fig. 7A,B), completely abolished the inhibitory effect on VDAC- and ANT-dependent mitochondrial ADP import (Fig. 7C,D). Moreover, CB23 exhibited a significantly lower cytotoxic potential as compared to ibipinabant (IC₅₀ values: 188 vs 17 μM, P=0.034, for 95%-CI see Fig. 7E). These findings also support that the cytotoxic effects observed with ibipinabant are related to inhibition of the ANT-dependent ADP/ATP exchange.

Ibipinabant (S-SLV 319) was ordered at Cayman Chemical (Ann Arbor, MI). A racemic mixture of 3-(4-chlorophenyl)-4-phenyl-4,5-dihydro-1H-pyrazole-1-carboxamide derivate 23 (CB23) was kindly provided by Abbott Products GmbH (Hannover, Germany). Etoposide, 4,4′-diisothiocyanatostilbene-2,2′-disulfonicacid disodium salt hydrate (DIDS), and atractyloside sodium salt were from Sigma Aldrich (Zwijndrecht, The Netherlands).

C2C12 murine myoblasts (CRTL 1772) were obtained from American Type Culture Collection (Wesel, Germany). C2C12 cells were maintained in Dulbecco’s modified Eagle’s medium with Glutamax I formulation containing 25 mM glucose and 25 mM HEPES supplemented with 10% (v/v) fetal calf serum (MP Biomedicals, Santa Ana, CA) at 37 °C in a humidified atmosphere of 5% CO₂.

C2C12 myoblasts were seeded in 384-wells black/clear imaging plates at a density of 4,000 cells/well, 24 hours prior to ibipinabant treatment. A 1,000× concentrated serial √10-dilution in DMSO was made for ibipinabant and the positive control etoposide. Immediately before use, ibipinabant and etoposide stock solutions were diluted 100× in phosphate buffered saline (PBS) and 10× in the culture medium, resulting in a final DMSO concentration of 0.1% (v/v). Four replicates of each compound were tested up to maximum solubility (100 μM). After 24 h and 48 h exposure, nuclei were stained for cell viability analysis using Hoechst 33342. After 20 minutes of staining at 37 °C, fluorescence was imaged on a BD Pathway 855 high-throughput microscope (Becton Dickinson (BD) Bioscience, Breda, The Netherlands). Followed by the analysis of the number of nuclei using Cellprofiler³³.

C2C12 myoblasts were seeded in 384-wells black/clear imaging plates as described for analysis of cell death. After 8 hours of ibipinabant exposure cells were loaded with 10 μM CM-H₂DCFDA in HEPES-Tris (HT) buffer (132 mM NaCl, 10 mM HEPES, 4.2 mM KCl, 1 mM MgCl₂, 1 mM CaCl₂ and 25 mM D-glucose, adjusted to pH 7.4 with Tris) for 20 minutes at 37 °C. Next, the cells were washed twice with HT buffer and imaged on a BD Pathway 855 high-throughput microscope. CM-DCF images were processed using Image Pro Plus 6.3 software (Media Cybernetics, Silver Spring, MD). A mask of the images was made to separate the cellular pixels from the background; afterwards average CM-DCF intensity per cellular pixel was determined.

After 4 hours ibipinabant exposure, C2C12 cells were resuspended in PBS with 3% bovine serum albumin, to a final density of 10·10⁶ cells/mL. Next, mitochondrial ATP production from pyruvate was measured with 1.10⁵ cells per incubation in the presence of 20 μg digitonin/10⁵ cells in duplicate as described previously³⁴,³⁵,³⁶. All values were normalized to the activity of the mitochondrial matrix enzyme, citrate synthase, to correct for differences in mitochondrial mass. Citrate synthase was determined spectrophotometrically in duplicate as described before³⁵.

For measurement of the cellular respiration rate, 1.5·10⁶ C2C12 cells were resuspended in mitochondrial respiration medium MiR05 after 4 hours ibipinabant exposure, and transferred to the thermostated (37 °C) chamber of an Oxygraph-2k equipped with Datlab 5 recording and analysis software (Oroboros Instruments, Innsbruck, Austria). Mitochondrial respiration driven by the separate respiratory chain complexes was measured after digitonin permeabilisation (10 μg/1·10⁶ cells) of the cell membrane in MiR05 containing complex-specific substrates and 4 mM ADP³⁷. Glutamate (10 mM) plus malate (2 mM) were used as substrates for complex I (CI), succinate (10 mM) for CII, and ascorbate (2 mM) plus TMPD (0.5 mM) for CIV. CIII-driven respiration was measured in the presence of glycerophosphate (20 mM) and flavine adenine dinucleotide (10 μM). Rotenone (0.5 μM) and atpenin A5 (50 nM) were added to inhibit CI and CII, respectively. Cytochrome c (10 μM) was added to check the integrity of the mitochondrial outer membrane. Finally, antimycin A (2,5 μM) was added to completely block CIII-driven respiration. Mitoplasts were isolated from C2C12 myoblasts to determine CI- and CII-specific respiration. Briefly, pellets of 10·10⁶ cells were resuspended in 10 mM Tris-HCl and pottered, and sucrose was added (215 mM). The lysate was cleared of unbroken cells by centrifugation (10 minutes 600 g), after which the supernatant containing the mitochondria was pelleted at 14,000 g for 10 minutes. Next, mitochondria were resuspended in hypotonic buffer (2% essential fatty acid free BSA, 20 mM K₂HPO₄, pH 7.4) and allowed to swell for 30 minutes, as described previously³⁸, and subsequently pottered to remove the outer membrane. Mitoplasts were obtained from the lysate by centrifugation at 14,000 g for 15 minutes, followed by resuspension of the pellet in MiR05. At the start of the experiment DIDS (250 μM) was added, just after the addition of cytochrome c, to exclude any influence of VDAC, and followed by determination of the CI- and CII-specific respiration as described above. Respiratory rates were normalized to mg protein in the mitoplast fractions as determined in duplicate by a Biorad protein assay (Biorad), after three sequential freeze thaw cycles to disrupt the inner membrane. Finally, all respiratory rates were corrected for non-mitochondrial respiration by subtraction of the residual respiratory rates after rotenone or antimycin A addition.

After 4 hours ibipinabant exposure, C2C12 pellets of 20·10⁶ cells were snap frozen in liquid nitrogen and kept at −80 °C until use. For mitochondrial preparations, cells were resuspended in 10 mM Tris-HCl and pottered, and sucrose was added (215 mM). The lysate was cleared of unbroken cells by centrifugation (10 minutes 600 g) after which the supernatant containing the mitochondria was pelleted at 14,000 g for 10 minutes, resuspended in 10mM Tris-HCl (pH 7.6), and snap frozen in aliquots. Catalytic capacity of OXPHOS complexes was measured spectrophotometrically in duplicate, as described previously³⁹,⁴⁰,⁴¹. Complex I-IV values were normalized to the activity of the mitochondrial matrix enzyme, citrate synthase, to correct for differences in mitochondrial mass. Citrate synthase was determined spectrophotometrically in duplicate as described before³⁵. Complex V values were normalized to cytochrome c oxidase (COX, complex IV) activity.

C2C12 cells (5.0·10⁴) were seeded in 35 mm Fluorodishes (World Precision Instruments, Sarasota, FL), 24 hours prior to ibipinabant exposure. To measure mitochondrial membrane potential after 8 hours ibipinabant treatment, cells were loaded with 100 nM tetramethylrhodamine methyl ester (TMRM) for 25 minutes at 37 °C as described previously⁴². Next, cells were maintained in HT buffer, and images were obtained using a temperature-controlled chamber attached to a stage of an inverted microscope (Axiovert 200M, Carl Zeiss, Jena, Germany) equipped with a ×63, 1.25 NA Plan NeoFluar oil immersion objective. TMRM intensity analysis was performed on background corrected images, which were masked with a binarised image for mitochondrial morphology using Image Pro Plus 6.3 software.

Since no experimental structure of CB1R is available, a homology model was built with the automated protocol of Yasara (version 13.9.8, www.yasara.org)⁴³ and pdb-code 3v2y as template⁴⁴. The co-crystallised ligand ML5 of the pdb-entry 3v2y complex is used to define the binding site of CB1R for KRIPO and calculate a structure based fingerprint bit string¹⁷. This bit string was compared to the PDB (version March 2014)⁴⁵.

The structure of bovine mitochondrial ADP/ATP carrier isoform 1 (pdb-code 2c3e)⁴⁶ was used for docking of ibipinabant using Molecular Operating Environment (MOE) (version 2013.0802, Chemical Computing Group Inc, Montreal, Canada). The structure preparation and protonate 3D protocol of MOE was applied before the docking. Docking was performed using standard settings. Ibipinabant was prepared in MOE by assigning PEOE charges and energy minimization.

ADP transport across the mitochondrial membrane was measured in isolated bovine heart mitochondria, as described before⁴⁷,⁴⁸. Briefly, 10 uL (equal to 130 μg protein) mitochondrial pellet was resuspended in ADP import buffer (250 mM sucrose, 20 mM HEPES, pH 7.2, 10 mM KCl, 5 mM succinate, 3 mM KH₂PO₄, 1.5 mM MgCl₂, 1 mM EGTA, and 5 μM rotenone) with or without the adenine nucleotide translocase inhibitor atractyloside (50 μM) to inhibit ANT-dependent ADP uptake, or with or without DIDS (500 μM) to inhibit VDAC-dependent ADP uptake. [14C]ADP (1 μCi) was added to the mitochondrial suspension and incubated for 10 min on ice. After washing two times with ADP import buffer, the samples were resuspended in scintillant and quantified using a scintillation counter (PerkinElmer Tri-Carb^® 2900TR, Perkin Elmer, Waltham, MA). ANT- and VDAC-dependent ADP transport activity was determined by the difference in counts between samples that respectively were or were not pre-incubated with atractyloside or DIDS. ADP import was decreased to 13.8 ± 1.1% and to 8.4 ± 0.8% respectively with atractyloside or DIDS alone.

Curve-fitting and statistical analysis was performed using GraphPad prism 5.02 software (GraphPad Software Inc., San Diego, CA). Unless indicated otherwise, all results are presented as mean ± SEM. Differences between groups were tested using one-way ANOVA analysis with appropriate post-hoc tests, unless indicated otherwise.