Evaluation of in Vitro Mitochondrial Toxicity Assays and

Article pubs.acs.org/crt

Cite This: Chem. Res. Toxicol. 2019, 32, 156−167

Evaluation of in Vitro Mitochondrial Toxicity Assays and Physicochemical Properties for Prediction of Organ Toxicity Using 228 Pharmaceutical Drugs Payal Rana,*,† Michael D. Aleo,† Mark Gosink,† and Yvonne Will† †

Drug Safety Research & Development, Pfizer, Eastern Point Road, Groton, Connecticut 06340, United States

Downloaded via UNIV OF WINNIPEG on January 25, 2019 at 12:13:03 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Mitochondrial toxicity has been shown to contribute to a variety of organ toxicities such as liver, cardiac, and kidney. In the past decades, two highthroughput applicable screening assays (isolated rat liver mitochondria; glucosegalactose grown HepG2 cells) to assess mitochondrial toxicity have been deployed in many pharmaceutical companies, and numerous publications have demonstrated its usefulness for mechanistic investigations. However, only two publications have demonstrated the utility of these screens as a predictor of human drug-induced liver injury. In the present study, we screened 73 hepatotoxicants, 46 cardiotoxicants, 49 nephrotoxicants, and 60 compounds not known to cause human organ toxicity for their effects on mitochondrial function(s) in the assays mentioned above. Predictive performance was evaluated using specificity and sensitivity of the assays for predicting organ toxicity. Our results show that the predictive performance of the mitochondrial assays are superior for hepatotoxicity as compared to cardiotoxicity and nephrotoxicity (sensitivity 63% vs 33% and 28% with similar specificity of 93%), when the analysis was done at 100* Cmax (drug concentration in human plasma level). We further explored the association of mitochondrial toxicity with physicochemical properties such as calculated log partition coefficient (cLogP), topological polar surface area, ionization status, and molecular weight of the drugs and found that cLogP was most significantly associated mitochondrial toxicity. Since these assays are amenable to higher throughput, we recommend that chemists use these assays to perform structure activity relationship early in the drug discovery process, when chemical matter is abundant. This assures that compounds that lack the propensity to cause mitochondrial dysfunction (and associated organ toxicity) will move forward into animals and humans.

■

INTRODUCTION Drug-induced mitochondrial toxicity contributes to toxicities of many organs, such as the liver, heart, kidney, skeletal muscle, and brain.1 In addition, it has been shown that mitochondrial toxicity at least in part contributed to the attrition of phenformin, troglitazone, nefazodone, and cerivastatin.2,3 The knowledge that many drug classes (antidiabetic, antilipidemic, antivirals, antibiotics, and antidepressants) can exhibit mitochondrial liabilities has led to the development of highthroughput applicable mitochondrial assays4,5 that can be positioned early in the drug discovery screening process. These assays have thus far mostly been used for mechanistic evaluations of numerous drug classes.6−9 Currently, few published studies have used mitochondrial toxicity assessment as a predictor of human liver injury.10−12 Porceddu et al. tested 87 drugs known to cause hepatotoxicity and 37 drugs not reported to cause hepatotoxicity and reported a greater than 90% positive predictive value for human drug-induced liver injury using a multiparameter assay conducted in mouse liver mitochondria.11 Aleo et al.,12 used 24 Most-DILI-, 28 LessDILI-, and 20 No-DILI-concern drugs annotated in the United States National Center for Toxicological Research Liver Toxicity Knowledge Base (NCTR-LTKB) and demonstrated © 2018 American Chemical Society

mitochondrial toxicity was generally correlated across human DILI concern categories, such as death or black box warnings when combined with inhibition of the bile salt export protein.12 In this study, we utilized two routine assays for detection of mitochondrial toxicants deployed within Pfizer to advance our understanding of (1) predictivity toward major organ toxicity, other than hepatotoxicity, and (2) to investigate the physicochemical attributes that contribute to positive findings in these assays. The first assay utilizes cells grown in two types of media, namely, (a) high glucose- and (b) galactosecontaining media.13−15 Cells grown in high glucose-containing medium use glycolysis for adenosine triphosphate (ATP) generation and are resistant to mitochondrial insult. In contrast, cells grown in galactose-containing medium rely almost exclusively on mitochondria for their ATP production and, hence, are very sensitive to mitochondrial insult. The second assay, called Respiratory Screening Technology (RST), measures mitochondrial respiration in the form of oxygen Received: August 28, 2018 Published: December 10, 2018 156

DOI: 10.1021/acs.chemrestox.8b00246 Chem. Res. Toxicol. 2019, 32, 156−167

Article

Chemical Research in Toxicology

All compounds were submitted to the two mitochondrial assays described above, and IC50 values were generated. A drug was considered positive if the RST IC50 value was less than the Cmax value and was considered negative if the RST IC50 value was higher than the Cmax value. If the Cmax values were higher than the concentration range tested in the assay, compounds were classified as “No Test”. The analysis was conducted at 1*, 10*, 30*, and 100* Cmax concentration values. Figure 1 shows the drug distribution of the hepatotoxic (32%), cardiotoxic (20.2%), nephrotoxic (21.5%), and no

consumption in freshly isolated rat liver mitochondria using a time-resolved fluorescent oxygen-sensitive probe.15−18 In the present study, we screened 73 hepatotoxicants, 46 cardiotoxicants, 49 nephrotoxicants, and 60 compounds not known to cause human organ toxicity. It was our desire to advance the usage of mitochondrial toxicity assays as a tool for predicting organ toxicity and understand the physicochemical property space that drives these liabilities. We addressed the following three questions: (A) What percentage of compounds in each organ toxicity class tested positive in the mitochondrial assays? (B) What is the predictive value toward particular organ toxicity (specificity/sensitivity)? (C) Do compounds that tested positive in the mitochondrial assays occupy a different physicochemical property space than those that tested negative?

■

EXPERIMENTAL PROCEDURES

Mitochondrial Toxicity in Vitro Assays. Methods for measurements of mitochondrial respiration in isolated rat liver mitochondria4,11,16−18 and assessment of mitochondrial toxicity using glucose/ galactose model5,13,14,37,52 have been previously published. Statistical Analysis. Concentration−response plots and IC50 values for each compound were generated either in Graph Pad Prism 5 using a nonlinear regression analysis or using System Integrated Global High Throughput Screening (SiGHTS), which is Pfizer’s internal proprietary data analysis software using a nonlinear regression analysis. Statistical analysis was performed using the twoway analysis of variance (ANOVA) between groups in Graph Pad Prism 5. P value less than 0.05 was considered statistically significant. Data Sources. All compounds selected for three categories of organ toxicity (hepatic, cardiac, and renal) were either reported in previous publications19 or extracted from the United States Food and Drug Administration’s Adverse Event Reporting System (FAERS) database with subsequent medline search for confirmation. Drug adverse event counts were extracted from the Oracle Health Science’s Empirica Signal database. The Empirica database included all reports up to 2014 Q1 from the FAERS database. Counts supplied at the preferred term (PT) of the database were aggregated into their respective System Order Class (SOC) level term. Multiple counts originating from the same report were normalized to a single count to mitigate double reporting. SOC level counts for all the drugs were scored using the SDgamma method described in Johnson et al.20 Drug scores for cardiac disorders, hepatobiliary disorders, and renal disorders were manually inspected and used to assign each drug for its respective organ toxicity category. The assignment of organ toxicity has been done to the best of our abilities. The maximum total drug plasma concentration (Cmax) was extracted from the literature21 and through medline searches to the best of our abilities. These compounds represent diverse target pharmacology and belong to multiple indication and therapeutic areas. Compounds that had a Cmax value greater than 100 μM were not included in the study. Random Forest Model. Ten physicochemical properties as well as our in vitro assay IC50 values were used to build the random forest model. This includes values of RST inhibition assay, RST uncoupling assay, HepG2 glucose assay, HepG2 galactose assay, IC50 ratio of glucose/galactose assay, ionization, Cmax, CLogP, topological polar surface area (TPSA), and molecular weight. Models were trained using caret package in R (version 3.5.1) using random forest “rf” method with ntree = 500 and importance = TRUE. For each toxicity group, 75% of the samples were used for training with the remaining 25% used for testing. The importance of each descriptor was determined using the “varimpplot” function on the model generated by the caret package using random forest “rf” method.

Figure 1. Drug distribution of hepatotoxic, cardiotoxic, nephrotoxic, and no organ tox compound set (N = 228).

organtox (26.3%) compound set. Here, we wanted to make sure that the drug distribution of each class is equivalent. Figure 2A shows that at 1*Cmax, 11% of the compounds known to cause human liver injury tested positive in the mitochondrial assays. At 10*Cmax, 20.5% of compounds were positive, and 21.9% of compounds were considered a “No Test”. At 30* and 100*Cmax, the percentage of compounds that were positive increased to 24.7% and 34.2%, respectively. The percentage of “No Test” compounds also increased further at 30* and 100*Cmax (35.6% and 45.2%, respectively). Figure 2B shows data for compounds known to cause cardiotoxicity in humans. At 1*Cmax, 6.5% of compounds tested positive. Similar to the compounds causing hepatotoxicity, the percentage of compounds testing positive and “No test” compounds increased with higher Cmax normalizations. The same trend was observed for the nephrotoxicants (Figure 2C). Figure 2D shows the results for the compounds not known to cause human organ toxicity. There were no positive compounds at 1*Cmax, 10*Cmax, or 30*Cmax. Only 5% of compounds were positive at the 100*Cmax. Similar to the other drug classes, the percentage of “No Test” compounds increased with higher Cmax normalizations. What Is the Predictive Value toward Particular Organ Toxicity (Specificity/Sensitivity)? Using this classification, we proceeded to perform statistical analysis of sensitivity (True positive/ (True positive + False negative)) and specificity (True negative/ (True negative+ False positive)). Table 1A demonstrates the overall predictivity toward the particular

■

RESULTS What Percentage of Compounds in Each Organ System Tested Positive in the Mitochondrial Assays? 157


Article


Figure 2. Drug distribution of (A) hepatotoxic, (B) cardiotoxic, (C) nephrotoxic, and (D) no organ tox compounds at 1*Cmax, 10*Cmax, 30*Cmax, and 100*Cmax concentration value. A drug was considered positive if the RST IC50 value was less than the Cmax value and was considered negative if the RST IC50 value was higher than the Cmax value. If the Cmax values were higher than the concentration range tested in the assay, compounds were classified as no test.

organ toxicity at 1*Cmax, 10*Cmax, 30*Cmax, and 100*Cmax normalized mitochondrial assay, whereas Table 1B indicates the numbers of compounds included from each organ toxicity group for sensitivity and specificity analysis (“No Test” compounds were not included in the sensitivity/

specificity calculations). The analysis suggests that the overall specificity was 93% and above, indicative of a very low (7%) false positive rate. The highest sensitivity for all the drug classes was at 100*Cmax. The sensitivity was 63%, 33%, and 28% for hepatotoxicants, cardiotoxicants, and nephrotoxicants, 158


Article

Chemical Research in Toxicology Table 1. (A) The Truth Table Containing Statistical Analysis of Sensitivitya and Specificityb Demonstrating the Overall Predictivity Towards the Particular Organ Toxicity at 1*Cmax, 10*Cmax, 30*Cmax, and 100*Cmax Normalized Mitochondrial Assay. (B) The Table Indicating Numbers of Compounds Included from Each Organ Toxicity Group for Sensitivity and Specificity Analysis

Table 2A. Hepatotoxic Compounds Were Positive in the Mitochondrial Assay Including Cmax Values, Reported Mitochondrial Impairment, and References Cmax (μM)

reported mitochondrial impairment

alpidem

0.208

amiodarone

4.650

inhibits complex I; uncouples OXPHOS inhibits complex I

benzbromarone

4.339

inhibits OXPHOS

chlorpromazine

0.941

inhibits complex I

clomipramine

0.191

inhibits OXPHOS

diclofenac

4.20

diflunisal

8.816

inhibits complex I; uncouples OXPHOS uncouples OXPHOS

flutamide

0.362

inhibits complex I

hexachlorophene

5.407

uncouples OXPHOS

compound

(A) Cmax normalized mitochondrial assay hepatotoxic cardiotoxic nephrotoxic

sensitivity specificity sensitivity specificity sensitivity specificity

1x

10x

30x

100x

11% 100% 7% 100% 0% 100% (B)

26% 100% 19% 100% 10% 100%

38% 100% 22% 100% 12% 100%

63% 93% 33% 93% 28% 93%

number of compounds included for specificity & sensitivity analysis Cmax normalized mitochondrial assay hepatotoxic no organtox cardiotoxic no organtox nephrotoxic no organtox a

1x

10x

30x

100x

73 60 46 60 49 60

57 52 42 52 30 52

47 48 36 48 25 48

40 43 33 43 18 43

ketaconazole

12.507

lovastatin

0.025

nimesulide

21.083

inhibits complex I; uncouples OXPHOS inhibits complex I; uncouples OXPHOS uncouples OXPHOS

nitrofurantoin

4.199

uncouples OXPHOS

phenformin

0.487

inhibits OXPHOS

rosiglitazone

1.044

simvastain

0.020

inhibits complex I; uncouples OXPHOS inhibits complex I

sitaxsentan

30.556

sulindac

19.00

b

True positive/ (True positive + False negative). True negative/ (True negative+ False positive).

respectively. Tables 2A (hepatotoxic), 2B (cardiotoxic), 2C (nephrotoxic), and 2D (no organtox) indicate compounds that were positive in the mitochondrial assays including Cmax values and reported mitochondrial impairment. We further demonstrate new evidence of compounds in each category with reported mitochondrial impairment in one of our assays. Tables 3A (hepatotoxic), 3B (cardiotoxic), 3C (nephrotoxic), and 3D (no organtox) indicate compounds that were negative in the mitochondrial assay and were reported to show mitochondrial impairment via mechanism other than mitochondrial respiration in state II and state III. The Supporting Information Table includes the in vitro results of all 228 commercial compounds with Cmax values including HepG2 glucose and galactose cell viability assays results and RST inhibitory and uncoupling assays results. We further examined the use of Random Forest modeling to classify the hepatotoxic, cardiotoxic, and nephrotoxic compounds from the no organtox compounds. We utilized both the physicochemical properties for the compounds as well as the assay data to build the random forest model. For each class of compounds, the ability of the Random Forest model to distinguish between the toxic compounds and the no organtox compounds was evaluated by examining the area under the curve (AUC) for each receiver operating characteristic (ROC) curve (Figure 3, left panels). AUC values of 0.826 for hepatotox, 0.839 for cardiotox, and 0.767 for nephrotox predictions indicate that good predictivity was achieved by each model. Since there were unequal numbers of no organtox to the various toxic compounds, we examined whether down sampling to equal class numbers would improve the predictivity; however, we observed little to no improvement. We also examined the importance of the various factors to each

tamoxifen

0.409

troglitazone

6.387

beclomethasone bicalutamide danazol ethynodiol diacetate ezlopitant procarbazine retinoic acid rifabutin ritolukast triflurin

0.0001 1.970 0.074 0.029

inhibits complex I; uncouples OXPHOS induces mitochondrial permeability transition pore and ROS inhibits and uncouples OXPHOS inhibits OXPHOS; induces mitochondrial permeability transition pore inhibits OXPHOS inhibits OXPHOS inhibits OXPHOS inhibits OXPHOS

3.214 3.127 1.331 0.685 6.543 0.005

uncouples OXPHOS uncouples OXPHOS inhibits OXPHOS uncouples OXPHOS uncouples OXPHOS uncouples OXPHOS

references for mitochondrial impairment Porceddu et al., 201211 Porceddu et al., 201211 Felser et al., 201238 Nadanaciva et al., 201239 Higgins and Pilkington, 201040 Ghosh et al.,41 Nadanaciva et al., 201342 Boelsterli et al., 200643 Cammer and Moor, 197244 Porceddu et al., 201211 Porceddu et al., 201211 Porceddu et al., 201211 Porceddu et al., 201211 Bridges et al., 201445 Nadanaciva et al., 201239 Nadanaciva et al., 201218 Kenna et al., 201546 Seo et al., 200747 Taquet et al., 200048 Nadanaciva et al., 201239 new new new new

evidence evidence evidence evidence

new new new new new new

evidence evidence evidence evidence evidence evidence

predictive model (Figure 3, right panels). For the models, we observed that maximal exposure Cmax was most important to hepatotoxic prediction, cLogP was most important to cardiotox prediction, and TPSA was most important to nephrotox predictions. Do Compounds That Tested Positive in Mitochondrial Assays Occupy a Different Physicochemical Property Space than Those That Tested Negative? Next we analyzed the physicochemical properties of the four 159


Article

Chemical Research in Toxicology Table 2B. Cardiotoxic Compounds Were Positive in the Mitochondrial Assay including Cmax Values, Reported Mitochondrial Impairment and References compound

Cmax (μM)

doxorubicin

0.037

inhibition of OXPHOS

4.830

uncouples OXPHOS

imipramine

0.802

menadione

12.206

inhibits complex I; uncouples OXPHOS uncouples OXPHOS

sertraline thioridazine

0.039 0.578

zoniporide aripiprazole axitinib clofilium crizotinib tenidap

6.310 2.240 0.194 1.00 13.805 8.437

references for mitochondrial impairment


erlotinib

Table 3A. Hepatotoxic Compounds Were Negative in the Mitochondrial Assay and Were Reported to Show Mitochondrial Impairment via Mechanism Other than Mitochondrial Respiration in State II and State III compound

Porceddu et al., 201211 Porceddu et al., 201211 Porceddu et al., 201211 Pandian et al., 200349 Li et al., 201250 de Faria et al., 201551

inhibits complex I induce mitochondrial permeability transition pore inhibits complex I inhibits OXPHOS inhibits OXPHOS inhibits OXPHOS uncouples OXPHOS inhibits & uncouples OXPHOS

chenodeoxycholic acid etodolac etoposide gemfibrozil

Rana et al., 201152 new evidence new evidence new evidence new evidence new evidence

phenytoin pirprofen

compound

Cmax (μM)

pentamidine isethionate

0.900

ritonovir tenofovir paranonylphenol polymyxin B

inhibits and uncouples OXPHOS inhibits OXPHOS inhibits OXPHOS

references for mitochondrial impairment Moreno, 199653

5.586

uncouples OXPHOS

Porceddu et al., 201211 Ramamoorthy et al., 201454 new evidence

1.00

inhibits OXPHOS

new evidence

15.271 0.926

compound


amitriptyline

inhibition of OXPHOS in pig brain mitochondria at high concentrations (>100 μmol/L) inhibition of mitochondrial respiration, permeability transition pore and depolarization at high concentrations inhibition of OXPHOS complex I and II in pig brain mitochondria inhibition of complex I at high concentrations

bupivacaine citalopram haloperidol propranolol

Table 2D. No Organtox Compounds Were Positive in the Mitochondrial Assay including Cmax Values, Reported Mitochondrial Impairment, and References compound

Cmax (μM)


atrovstatin

0.448

inhibits OXPHOS

miconazole

0.024

inhibits OXPHOS

benztropine mesylate clemastine darifenacin

0.586

uncoupling OXPHOS

pimozide protriptyline tolterodine

0.002 0.028 0.006

0.060 0.475

uncoupling OXPHOS inhibition and uncoupling of OXPHOS inhibits OXPHOS uncoupling OXPHOS uncoupling OXPHOS

inhibition of mitochondrial oxidation of fatty acids at 0.25, 0.5, 1, and 2 mM concentrations induces permeability transition pore in rat liver mitochondria uncouples oxidative phosphorylation in isolated mitochondria induces cytochrome c release and swelling of isolated mitochondria inhibition of complex I in isolated mitochondria at (29.5% inhibition at 500 μM concentrations) mitochondrial dysfunction induced by bio activated phenytoin metabolite inhibition of mitochondrial fatty acid oxidation

Le Dinh et al., 199856 Rolo et a., 200157 Mahmud et al., 199658 Park and Kim, 200559 Nadanaciva et al., 201218 Santos et al., 200860 Begriche et l., 201161

Table 3B. Cardiotoxic Compounds Were Negative in the Mitochondrial Assay and Were Reported to Show Mitochondrial Impairment via Mechanism Other than Mitochondrial Respiration in State II and State III

Table 2C. Nephrotoxic Compounds Were Positive in the Mitochondrial Assay Including Cmax Values, Reported Mitochondrial Impairment and References reported mitochondrial impairment


amineptine

references for mitochondrial impairment

inhibition of OXPHOS in pig brain mitochondria at high concentrations (>100 μmol/L)

references for mitochondrial impairment Hroudova and Fisar, 201262 Irwin et al., 200263 Hroudova and Fisar, 201064 ModicaNapolitano et al., 200365 Hroudova and Fisar, 201262

Table 3C. Nephroxic Compounds Were Negative in the Mitochondrial Assay and Were Reported to Show Mitochondrial Impairment via Mechanism Other than Mitochondrial Respiration in State II and State III

Porceddu et al., 2012 11 Porceddu et al., 201211 Rodriguez and Acosta, 199655 new evidence

compound aristolochic acid

new evidence new evidence

indomethacin new evidence new evidence new evidence

ochratoxin a

compound sets, which included ionization state, cLogP, TPSA, and molecular weight. Previously, there have been reports of association of high lipophilicity and low polarity with toxic outcomes.22,23 Figure 4 shows the contribution of acids, bases, neutrals, and zwitterions. The hepatotoxicants data set (n = 73) contained 34% acids, 27% bases, 27% neutrals, and 11%

paraquat tacrolimus

160

reported mitochondrial impairment induces mitochondrial permeability transition pore in isolated kidney mitochondria uncouples OXPHOS at low concentrations and inhibits OXPHOS at high concentration in rat liver mitochondria uncouples and inhibits OXPHOS at various concentrations in rat liver mitochondria induces calcium dependent membrane depolarization, uncoupling and swelling inhibits OXPHOS in isolated rat kidney mitochondria

references for mitochondrial impairment Qi et al., 200766 Jacob et al., 200167 Wei et al., 198568 Costantini et al., 199569 Simon et al., 200370


Article


Figure 3. ROC curves and Factor Importance for Random Forest models. Physicochemical properties and assay results were used to build predictive models of (A) hepatotoxicity, (B) cardiotoxicity, or (C) nephrotoxicity. The models were evaluated by the AUC for each ROC curve for each toxicity category (left panels). Both unweighted classes (solid line) and equally weighted classes (dotted line) were evaluated. The importance that each factor played in building the model is shown in the right panel.

zwitterions. The cardiotoxicants data set (n = 49) contained 7% acids, 74% bases, 17% neutrals, and less than 1% zwitterions. The nephrotoxicants data set (n = 46) contained

22% acids, 20% bases, 35% neutrals, and 22% zwitterions. The no organtox data set (n = 60) contained 13% acids, 62% bases, 22% neutrals, and 3% zwitterions. 161


Article


mitochondrial assays (mean cLogP > 3, 3.87) compared to those compounds that tested negative in the mitochondrial assays (mean cLogP < 3, 1.59) as shown in Figure 5A. In contrast, as shown in Figure 5B,C, there was no statistically significant difference between the groups with respect to TPSA and molecular weight. Next, we evaluated lipophilicity within different ionization of the compound set between compounds that tested positive or negative in the mitochondrial assays (Figure 6). We observed that, regardless of the ionization state, compounds that tested positive in the mitochondrial assays had statistically significant (P < 0.05) higher cLogP values than those compounds that tested negative in the mitochondrial assays (Figure 6) except zwitterion. Moreover, the distribution of compounds that tested positive in the mitochondrial assays across each ionization state was 42% acids, 35% neutral, 28% base, and 16% zwitterions had higher cLogP values (Figure 6). We also evaluated lipophilicity (cLogP) within the different organ toxicity data sets to see if there were differences between different organ toxicities that could help predict future organ toxicity (Figure 7). There was a statistically significant (P < 0.0001) difference in cLogP for compounds that tested positive in the mitochondrial assays (mean cLogP > 3, 4.36, n = 29) compared to those compounds that tested negative in the mitochondrial assays (mean cLogP < 3, 1.54, n = 44) in hepatotoxic compounds as shown in Figure 7A. Similarly, there was a statistically significant (P < 0.0001) difference in cLogP for compounds that tested positive in the mitochondrial assays (mean cLogP > 3, 4.76, n = 8) compared to those compounds that tested negative in the mitochondrial assays (mean cLogP < 3, 1.23, n = 52) in the no organtox compounds as shown in Figure 7D. In contrast, as shown in Figure 7B,C, there was no statistically significant difference between the groups with respect to cLogP in the cardiotoxic and nephrotoxic data set.

Figure 4. Drug distribution of ionization across (A) hepatotoxic, (B) cardiotoxic, (C) nephrotoxic, and (D) no organ toxic categories.

We took a detailed look at physicochemical properties of compounds that were positive in the mitochondria assays and compared them to the compounds that were negative in the assays. Here (Figures 5−7), a compound is considered

■

DISCUSSION The high rate of attrition and immense cost of drug development have urged the pharmaceutical industry to move safety assessments much earlier into the drug discovery process when chemical matter is available and structure activity relationship (SAR) determination is feasible. At Pfizer, we have been using various in silico tools, such as structural alerts, physicochemical properties,24,25 and the results of in vitro cellbased toxicity assays26,27 to move the best compound forward into animal exploratory toxicity studies. It is well-established that existing animal models are not highly predictive of human drug-induced liver injury. Olson and colleagues showed that preclinical animals (data combined from rodents, dogs, and monkeys) can only identify approximately half of hepatotoxic drugs in humans,28 which were further substantiated in a more recent analysis.29,30 Therefore, various attempts have been made to develop an in vitro testing paradigm that is capable of identifying hepatotoxic drugs that were previously missed by animal testing.31,32 Advancements have been made in linking particular mechanisms to organ toxicity, such as in the case of mitochondrial toxicity and liver injury. Whereas most studies have been of mostly mechanistic nature, two studies have examined larger compound sets to evaluate predictivity toward human liver injury. Porceddu et al.11 utilized a high-throughput screening platform using isolated mouse liver mitochondria and measured mitochondrial (swelling), inner membrane permeabilization (transmembrane potential), outer membrane

Figure 5. Physicochemical properties (A) cLogP, (B) TPSA, and (C) molecular weight of all mitochondrial assay positive and negative compounds. P < 0.001 compares values of mitochondrial assay negative compounds against positive compounds.

mitochondrial positive if it has an IC50 value of less than 100 nmol/mg mitochondrial protein in the RST inhibitory or RST uncoupling assay or an HepG2 glucose/galactose ratio greater than 3. There was a statistically significant (P < 0.0001) difference in cLogP for compounds that tested positive in the 162


Article


Figure 6. Physicochemical property cLogP distribution in different ionization (A) acid, (B) neutral, (C) base, and (D) zwitter of all mitochondrial assay positive and negative compounds. P < 0.05 compares values of mitochondrial assay negative compounds against positive compounds.

permeabilization (cytochrome c release), and alteration of mitochondrial respiration driven by succinate or malate/ glutamate, all indicators of mitochondrial toxicity.11 The authors tested 87 compounds with documented hepatotoxicity and 37 without reported clinical hepatotoxicity. The authors reported a high positive predictive value (82%). Our predictive value is lower (63%), but our analysis was different and compared to a more equal completely nontoxic compound set, whereas Porceddu et al. tested less non-toxic compounds, and some of them were reported to be toxic to other organs. The study by Aleo et al.12 showed a strong association of mitochondrial toxicity, when normalized to exposure toward compounds exhibiting severe drug-induced liver injury as well as compounds with black box warnings.12 The authors reported that this correlation was strengthened significantly when bile salt export pump (BSEP) inhibition was also measured and included in the prediction model. This emphasizes the fact that organ toxicity is most likely multifactorial and that no one mechanistic assay can predict the organ toxicity by 100%. Considering that, our sensitivity of

63% is probably more realistic than the high predictivity, for example, reported by Porceddu et al.12 In general, the usage of in vitro assays to predict human outcome of organ toxicity is difficult because of the multiple and complex mechanisms of toxicity due to chronic exposure, complex pharmacokinetic profiles of test compounds in each organ, and multifaceted physicochemical conditions. Therefore, our approach here is to combine the in vitro assays with physicochemical properties while taking exposure (Cmax concentrations) into account. Highly lipophilic compounds may accumulate largely in liver (and mitochondria) during the first pass metabolism; therefore, we observed that highly lipophilic compounds show liver toxicity. After the liver is exposed to higher concentration of a drug, the systemic concentration of a drug reduces by the time it reaches other organs such as heart and kidney. The highest predictivity for all three organ toxicity was at 100*Cmax accompanied by highest predictivity in liver. This was in agreement with the study by Xu et al.,33 where they found the best predictivity of orally dosed drugs to cause liver toxicity at 100*Cmax concentrations combined with in vitro 163


Article


al., where doxorubicin shows toxicity in both functional and structural in vitro assay end points. There are significant numbers of publications promoting the use of primary cultures of renal cells and cell lines for druginduced nephrotoxicity research. A fit for purpose cell-based assay would be of great value for drug discovery programs, as early preanimal toxicity studies screening, and for structure− activity relationship evaluation. However, the ability to use these in vitro models to reproduce the effect observed in a functioning kidney remains questionable, and it is clear that simple single-readout assays in immortalized cell lines are unlikely to provide useful data for kidney toxicity. Furthermore, nephrotoxic compounds evaluated here consisted of different physicochemical properties such a higher TPSA and lower cLogP. Because of the higher polar surface area, they have higher systemic concentration that may lead to higher chance to reach to proximal tubule cells of kidney. Moreover, because of lower lipophilicity, they have lower accumulation in the liver leading to higher accumulation into other organs such as kidney. Additionally, accumulation in the kidney is facilitated by active solute carrier transporter, which is necessary for drug exposure in the kidney. This makes us believe that nephrotoxic class of compounds may not be sensitive to the simple in vitro assays noted here but may have to include more physiologically relevant aspects such as flow, transporter expression, etc.36 Here, we have used HepG2 liver cell line for the glucose/ galactose assay and have used the rat isolated liver mitochondria for RST assay. We recognize that we have used the liver in vitro models to evaluate all three drug classes. However, previously we have utilized hepatic, cardiac, and kidney-derived cell lines for evaluation of organ toxicity, and our results suggested that specific organ toxicity cannot be accurately predicted using organ-specific cell line approach.19 In our opinion, selected organ toxicity potentially results from compound accumulation in a particular tissue, cell types within the organs, metabolism, and off-target effects. We have also evaluated selected compounds from hepatotoxic and cardiotoxic data set and tested in mitochondria isolated from liver and heart, respectively. We observed that most of the mitochondrial toxicants (there are few exceptions) impaired the respiration of heart and liver mitochondria in the similar fashion without sensitivity toward one particular organ (data not shown). We also understand that, depending upon the source of the mitochondria from particular organ, they utilize different substrates for energy production. For example, liver and heart mitochondria use pyruvate and fatty acid, respectively, as their primary substrates. Previously, we have evaluated an in vitro model using induced pluripotent stem cell (iPSC)-derived cardiomyocytes growing on different substrates such as glucose, galactose, and fatty acids. We concluded that, regardless of substrate sources, complex I mitochondrial inhibitor rotenone impaired respiration and cell viability in both galactose and fatty acids growing cardiomyocytes.37 Hughes et al. studied physicochemical properties associated with adverse events observed during in vivo toleration studies.23 These authors concluded that compounds with both high lipophilicity and low polarity (cLogP > 3 and TPSA < 75 Å2) had significantly increased toxic outcomes (6:1) in in vivo studies, in comparison to compounds that had both lower lipophilicity and higher polarity (cLogP < 3 and TPSA > 75 Å2). Green et al. also published similar study, where they demonstrated a correlation between the physicochemical

Figure 7. Physicochemical property cLogP distribution of four data sets (A) hepatotoxic, (B) cardiotoxic, (C) nephrotoxic, and (D) no organtox of all mitochondrial assay positive and negative compounds. P < 0.0001 compares values of mitochondrial assay negative compounds against positive compounds.

mitochondrial high content imaging models.33 They used a scaling factor of sixfold to account for population Cmax variability from the average therapeutic Cmax to account for patient genetic (such as metabolic enzymes and transporter) and epigenetic (such as age and pre-existing conditions) factors that may affect drug clearance. Another sixfold uncertainty factor was implemented to account for higher drug exposure to the liver via liver portal vein.34 Finally, threefold uncertainty factor was added to account for drug− drug or drug−diet interactions concluding that liver exposure of the drug could be 100 times the Cmax concentration, which may explain higher hepatic damage compared to other organ system. Cardiotoxicity has both functional and structural components, and therefore a more holistic approach that combine these two elements needs to be incorporated in the risk assessment.35 Clements et al.35 suggested that incorporating functional in vitro assay end points such as electrophysiology, calcium transients, contractility, metabolic activity, and cell movement/beating and structural in vitro assay end points such as nuclear area, cytoplasmic calcium concentrations, mitochondrial count/area, and cell viability need to be included for the cardiotoxicity risk assessment.35 Mexiletine, a class 1b anti-arrhythmic drug, acts by blocking sodium channels. On the one hand, Mexiletine was negative in our assays, and Clements et al. also observed in their assessment that this compound was negative in the in vitro assay that measured structural end points, whereas it was positive in the in vitro assays that measured functional end point such as sodium influx and beating. On the other hand, an antitumor drug, doxorubicin, is associated with congestive heart failure and a decrease in left ventricular ejection fraction due to changes in contractility. Doxorubicin is positive in mitochondrial respiration assay; this finding is supported by Clements et 164



■

properties of compounds and the likelihood of seeing an adverse finding in an in vivo toxicology study.22 They demonstrated that lipophilicity (cLogP) and TPSA were key factors of risk and suggested that compounds with cLogP > 3 and TPSA < 75 Å2 were more likely to have a finding at a Cmax < 10 μM than compounds with cLogP < 3 and TPSA > 75 Å2.22 Thus, correlation of higher lipophilicity and low polarity with toxic outcomes in animal studies are wellestablished. We performed random forest modeling on a number of physicochemical properties and assay values to determine if good sensitivity could be achieved for all compounds in each category. Using this approach, we were able to achieve sensitivity values near 80% while maintaining near 80% specificity (Figure 3). While examination of the variable importance to the random forest models suggests that the physicochemical properties play a bigger role in predictivity than do the assay data, the large number of “No Test” or missing values likely play a role in their apparent lack of importance. In our opinion, the physicochemical properties of the compounds are useful for understanding structure toxicity relationships for identification of new chemical series for a biological target. However, IC50 results from mitochondrial toxicity assays may contribute in determination of the safety margin. Moreover, compounds those were positive in mitochondrial assay had a different physicochemical space in terms of lipophilicity (cLogP > 3), and hepatotoxic compound set is the most sensitive to lipophilicity (cLogP) in ability to distinguish mitochondrial positive compounds from negative. We further demonstrated that mitochondrial positive compounds were more prone to be in acid and neutral ionization states. Therefore, using a combination of calculated physicochemical properties and dose-response values in highthroughput in vitro mitochondrial assay may lead to designing and selecting compounds with fewer safety concerns. In conclusion, the RST and the glucose/galactose assays are high-throughput assays and require minimal experimental efforts, cost, and physical compound matter. It can be effectively used to help prioritize compounds that advance into exploratory toxicity study and in prediction of organ toxicity. In addition, readily calculated physicochemical properties can be used to segregate compounds that are of high risk for in vitro testing (e.g., mitochondrial assays described herein) or help design compounds that are in lower risk chemical space. Once synthesized, a simple highthroughput mitochondrial toxicity assessment can further enhance the ability of drug discovery teams to select compounds with less potential for clinical organ toxicity.

■

Article

AUTHOR INFORMATION

Corresponding Author

*Phone: 860 715 6154. E-mail: payal.m.rana@pfizer.com. ORCID

Payal Rana: 0000-0001-7147-4711 Michael D. Aleo: 0000-0003-1549-2629 Author Contributions

P.R. was responsible for the hypothesis generation, predictive analysis, manuscript outline, and generation of figures and tables. P.R. and Y.W. were responsible for creating the primary manuscript. M.G. was responsible for generating receiver operating characteristic curves and factor importance for random forest models. M.A. was responsible for providing literature scholarship. All authors reviewed, edited, and refined the final manuscript and have given approval to the final version. Funding

Funding was from an internal grant from the Drug Safety Research and Development at Pfizer Inc., Worldwide Research and Development. Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS The authors thank the following individuals for their assistance in the preparation of this article: L. Marroquin and R. Swiss. ABBREVIATIONS ATCC, American Type Culture Collection; BSEP, Bile Salt Export Pump; cLogP, calculated Log Partition Coefficient; DMEM, Dulbecco’s modified Eagle’s medium; FDA, Food and Drug Administration; FAERS, FDA’s Adverse Event Reporting System; HEPES, N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid; NCTR-LTKB, National Center for Toxicological Research-Liver Toxicity Knowledge Base; RST, Respiratory Screening Technology; SAR, Structure Activity Relationship; TPSA, Topological Polar Surface Area

■

REFERENCES

(1) Will, Y., and Dykens, J. (2014) Mitochondrial toxicity assessment in industry–a decade of technology development and insight. Expert Opin. Drug Metab. Toxicol. 10 (8), 1061−7. (2) Dykens, J. A., Jamieson, J. D., Marroquin, L. D., Nadanaciva, S., Xu, J. J., Dunn, M. C., Smith, A. R., and Will, Y. (2008) In vitro assessment of mitochondrial dysfunction and cytotoxicity of nefazodone, trazodone, and buspirone. Toxicol. Sci. 103 (2), 335−45. (3) Dykens, J. A., and Will, Y. (2007) The significance of mitochondrial toxicity testing in drug development. Drug Discovery Today 12 (17−18), 777−85. (4) Hynes, J., Carey, C., and Will, Y. (2016) Fluorescence-Based Microplate Assays for In Vitro Assessment of Mitochondrial Toxicity, Metabolic Perturbation, and Cellular Oxygenation. Curr. Protoc Toxicol 70, 2.16.1−2.16.30. (5) Swiss, R., and Will, Y. Assessment of mitochondrial toxicity in HepG2 cells cultured in high-glucose- or galactose-containing media. Curr. Protoc Toxicol 2011. DOI: 10.1002/0471140856.tx0220s49 Chapter 2, Unit2.20. (6) Dykens, J. A., Marroquin, L. D., and Will, Y. (2007) Strategies to reduce late-stage drug attrition due to mitochondrial toxicity. Expert Rev. Mol. Diagn. 7 (2), 161−75. (7) Dykens, J. A., and Will, Y. (2007) The significance of mitochondrial toxicity testing in drug development. Drug Discovery Today 12 (17−18), 777−85.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemrestox.8b00246. This file contains the 228 compound data set used in this publication including drug names, Cmax values, organ toxicity class, RST in vitro assay results, glucose/ galactose in vitro assay results and physical chemical properties (XLSX) 165


Article

Chemical Research in Toxicology (8) Nadanaciva, S., Dykens, J. A., Bernal, A., Capaldi, R. A., and Will, Y. (2007) Mitochondrial impairment by PPAR agonists and statins identified via immunocaptured OXPHOS complex activities and respiration. Toxicol. Appl. Pharmacol. 223 (3), 277−87. (9) Nadanaciva, S., and Will, Y. (2009) The role of mitochondrial dysfunction and drug safety. IDrugs 12 (11), 706−10. (10) Thompson, R. A., Isin, E. M., Li, Y., Weidolf, L., Page, K., Wilson, I., Swallow, S., Middleton, B., Stahl, S., Foster, A. J., Dolgos, H., Weaver, R., and Kenna, J. G. (2012) In vitro approach to assess the potential for risk of idiosyncratic adverse reactions caused by candidate drugs. Chem. Res. Toxicol. 25 (8), 1616−32. (11) Porceddu, M., Buron, N., Roussel, C., Labbe, G., Fromenty, B., and Borgne-Sanchez, A. (2012) Prediction of liver injury induced by chemicals in human with a multiparametric assay on isolated mouse liver mitochondria. Toxicol. Sci. 129 (2), 332−45. (12) Aleo, M. D., Luo, Y., Swiss, R., Bonin, P. D., Potter, D. M., and Will, Y. (2014) Human drug-induced liver injury severity is highly associated with dual inhibition of liver mitochondrial function and bile salt export pump. Hepatology 60 (3), 1015−22. (13) Gohil, V. M., Sheth, S. A., Nilsson, R., Wojtovich, A. P., Lee, J. H., Perocchi, F., Chen, W., Clish, C. B., Ayata, C., Brookes, P. S., and Mootha, V. K. (2010) Nutrient-sensitized screening for drugs that shift energy metabolism from mitochondrial respiration to glycolysis. Nat. Biotechnol. 28 (3), 249−55. (14) Marroquin, L. D., Hynes, J., Dykens, J. A., Jamieson, J. D., and Will, Y. (2007) Circumventing the Crabtree effect: replacing media glucose with galactose increases susceptibility of HepG2 cells to mitochondrial toxicants. Toxicol. Sci. 97 (2), 539−47. (15) Dykens, J. A., Marroquin, L. D., and Will, Y. (2007) Strategies to reduce late-stage drug attrition due to mitochondrial toxicity. Expert Rev. Mol. Diagn. 7 (2), 161−75. (16) Hynes, J., Marroquin, L. D., Ogurtsov, V. I., Christiansen, K. N., Stevens, G. J., Papkovsky, D. B., and Will, Y. (2006) Investigation of drug-induced mitochondrial toxicity using fluorescence-based oxygensensitive probes. Toxicol. Sci. 92 (1), 186−200. (17) Hynes, J., Swiss, R. L., and Will, Y. (2012) High-throughput analysis of mitochondrial oxygen consumption. Methods Mol. Biol. 810, 59−72. (18) Nadanaciva, S., Dykens, J. A., Bernal, A., Capaldi, R. A., and Will, Y. (2007) Mitochondrial impairment by PPAR agonists and statins identified via immunocaptured OXPHOS complex activities and respiration. Toxicol. Appl. Pharmacol. 223 (3), 277−87. (19) Lin, Z., and Will, Y. (2012) Evaluation of drugs with specific organ toxicities in organ-specific cell lines. Toxicol. Sci. 126 (1), 114− 27. (20) Johnson, K., Guo, C., Gosink, M., Wang, V., and Hauben, M. (2012) Multinomial modeling and an evaluation of common datamining algorithms for identifying signals of disproportionate reporting in pharmacovigilance databases. Bioinformatics 28 (23), 3123−30. (21) Schulz, M., Iwersen-Bergmann, S., Andresen, H., and Schmoldt, A. (2012) Therapeutic and toxic blood concentrations of nearly 1,000 drugs and other xenobiotics. Crit Care 16 (4), R136. (22) Greene, N., Aleo, M. D., Louise-May, S., Price, D. A., and Will, Y. (2010) Using an in vitro cytotoxicity assay to aid in compound selection for in vivo safety studies. Bioorg. Med. Chem. Lett. 20 (17), 5308−12. (23) Hughes, J. D., Blagg, J., Price, D. A., Bailey, S., Decrescenzo, G. A., Devraj, R. V., Ellsworth, E., Fobian, Y. M., Gibbs, M. E., Gilles, R. W., Greene, N., Huang, E., Krieger-Burke, T., Loesel, J., Wager, T., Whiteley, L., and Zhang, Y. (2008) Physiochemical drug properties associated with in vivo toxicological outcomes. Bioorg. Med. Chem. Lett. 18 (17), 4872−5. (24) Wager, T. T., Chandrasekaran, R. Y., Hou, X., Troutman, M. D., Verhoest, P. R., Villalobos, A., and Will, Y. (2010) Defining desirable central nervous system drug space through the alignment of molecular properties, in vitro ADME, and safety attributes. ACS Chem. Neurosci. 1 (6), 420−34. (25) Wager, T. T., Hou, X., Verhoest, P. R., and Villalobos, A. (2010) Moving beyond rules: the development of a central nervous

system multiparameter optimization (CNS MPO) approach to enable alignment of druglike properties. ACS Chem. Neurosci. 1 (6), 435−49. (26) Nadanaciva, S., and Will, Y. (2011) New insights in druginduced mitochondrial toxicity. Curr. Pharm. Des. 17 (20), 2100−12. (27) Nadanaciva, S., and Will, Y. (2011) Investigating mitochondrial dysfunction to increase drug safety in the pharmaceutical industry. Curr. Drug Targets 12 (6), 774−82. (28) Olson, H., Betton, G., Robinson, D., Thomas, K., Monro, A., Kolaja, G., Lilly, P., Sanders, J., Sipes, G., Bracken, W., Dorato, M., Van Deun, K., Smith, P., Berger, B., and Heller, A. (2000) Concordance of the toxicity of pharmaceuticals in humans and in animals. Regul. Toxicol. Pharmacol. 32 (1), 56−67. (29) Butler, L. D., Guzzie-Peck, P., Hartke, J., Bogdanffy, M. S., Will, Y., Diaz, D., Mortimer-Cassen, E., Derzi, M., Greene, N., and DeGeorge, J. J. (2017) Current nonclinical testing paradigms in support of safe clinical trials: An IQ Consortium DruSafe perspective. Regul. Toxicol. Pharmacol. 87, S1−s15. (30) Monticello, T. M., Jones, T. W., Dambach, D. M., Potter, D. M., Bolt, M. W., Liu, M., Keller, D. A., Hart, T. K., and Kadambi, V. J. (2017) Current nonclinical testing paradigm enables safe entry to First-In-Human clinical trials: The IQ consortium nonclinical to clinical translational database. Toxicol. Appl. Pharmacol. 334, 100− 109. (31) Thompson, R. A., Isin, E. M., Li, Y., Weidolf, L., Page, K., Wilson, I., Swallow, S., Middleton, B., Stahl, S., Foster, A. J., Dolgos, H., Weaver, R., and Kenna, J. G. (2012) In vitro approach to assess the potential for risk of idiosyncratic adverse reactions caused by candidate drugs. Chem. Res. Toxicol. 25 (8), 1616−32. (32) Eakins, J., Bauch, C., Woodhouse, H., Park, B., Bevan, S., Dilworth, C., and Walker, P. (2016) A combined in vitro approach to improve the prediction of mitochondrial toxicants. Toxicol. In Vitro 34, 161−70. (33) Xu, J. J., Henstock, P. V., Dunn, M. C., Smith, A. R., Chabot, J. R., and de Graaf, D. (2008) Cellular imaging predictions of clinical drug-induced liver injury. Toxicol. Sci. 105 (1), 97−105. (34) Ito, K., Chiba, K., Horikawa, M., Ishigami, M., Mizuno, N., Aoki, J., Gotoh, Y., Iwatsubo, T., Kanamitsu, S., Kato, M., Kawahara, I., Niinuma, K., Nishino, A., Sato, N., Tsukamoto, Y., Ueda, K., Itoh, T., and Sugiyama, Y. (2002) Which concentration of the inhibitor should be used to predict in vivo drug interactions from in vitro data? AAPS PharmSci 4 (4), No. E25. (35) Clements, M., Millar, V., Williams, A. S., and Kalinka, S. (2015) Bridging Functional and Structural Cardiotoxicity Assays Using Human Embryonic Stem Cell-Derived Cardiomyocytes for a More Comprehensive Risk Assessment. Toxicol. Sci. 148 (1), 241−60. (36) Huang, J. X., Blaskovich, M. A., and Cooper, M. A. (2014) Celland biomarker-based assays for predicting nephrotoxicity. Expert Opin. Drug Metab. Toxicol. 10 (12), 1621−35. (37) Rana, P., Anson, B., Engle, S., and Will, Y. (2012) Characterization of human-induced pluripotent stem cell-derived cardiomyocytes: bioenergetics and utilization in safety screening. Toxicol. Sci. 130 (1), 117−31. (38) Felser, A., Lindinger, P. W., Schnell, D., Kratschmar, D. V., Odermatt, A., Mies, S., Jeno, P., and Krahenbuhl, S. (2014) Hepatocellular toxicity of benzbromarone: effects on mitochondrial function and structure. Toxicology 324, 136−46. (39) Nadanaciva, S., Rana, P., Beeson, G. C., Chen, D., Ferrick, D. A., Beeson, C. C., and Will, Y. (2012) Assessment of drug-induced mitochondrial dysfunction via altered cellular respiration and acidification measured in a 96-well platform. J. Bioenerg. Biomembr. 44 (4), 421−37. (40) Higgins, S. C., and Pilkington, G. J. (2010) The in vitro effects of tricyclic drugs and dexamethasone on cellular respiration of malignant glioma. Anticancer Res. 30 (2), 391−7. (41) Ghosh, R., Goswami, S. K., Feitoza, L. F., Hammock, B., and Gomes, A. V. (2016) Diclofenac induces proteasome and mitochondrial dysfunction in murine cardiomyocytes and hearts. Int. J. Cardiol. 223, 923−935. 166


Article

Chemical Research in Toxicology (42) Nadanaciva, S., Aleo, M. D., Strock, C. J., Stedman, D. B., Wang, H., and Will, Y. (2013) Toxicity assessments of nonsteroidal anti-inflammatory drugs in isolated mitochondria, rat hepatocytes, and zebrafish show good concordance across chemical classes. Toxicol. Appl. Pharmacol. 272 (2), 272−80. (43) Boelsterli, U. A., Ho, H. K., Zhou, S., and Leow, K. Y. (2006) Bioactivation and hepatotoxicity of nitroaromatic drugs. Curr. Drug Metab. 7 (7), 715−727. (44) Cammer, W., and Moore, C. L. (1972) The effect of hexachlorophene on the respiration of brain and liver mitochondria. Biochem. Biophys. Res. Commun. 46 (5), 1887−94. (45) Bridges, H. R., Jones, A. J., Pollak, M. N., and Hirst, J. (2014) Effects of metformin and other biguanides on oxidative phosphorylation in mitochondria. Biochem. J. 462 (3), 475−87. (46) Kenna, J. G., Stahl, S. H., Eakins, J. A., Foster, A. J., Andersson, L. C., Bergare, J., Billger, M., Elebring, M., Elmore, C. S., and Thompson, R. A. (2015) Multiple compound-related adverse properties contribute to liver injury caused by endothelin receptor antagonists. J. Pharmacol. Exp. Ther. 352 (2), 281−90. (47) Seo, S. K., Lee, H. C., Woo, S. H., Jin, H. O., Yoo, D. H., Lee, S. J., An, S., Choe, T. B., Park, M. J., Hong, S. I., Park, I. C., and Rhee, C. H. (2007) Sulindac-derived reactive oxygen species induce apoptosis of human multiple myeloma cells via p38 mitogen activated protein kinase-induced mitochondrial dysfunction. Apoptosis 12 (1), 195− 209. (48) Tuquet, C., Dupont, J., Mesneau, A., and Roussaux, J. (2000) Effects of tamoxifen on the electron transport chain of isolated rat liver mitochondria. Cell Biol. Toxicol. 16 (4), 207−19. (49) Pandian, R. P., Kutala, V. K., Parinandi, N. L., Zweier, J. L., and Kuppusamy, P. (2003) Measurement of oxygen consumption in mouse aortic endothelial cells using a microparticulate oximetry probe. Arch. Biochem. Biophys. 420 (1), 169−175. (50) Li, Y., Couch, L., Higuchi, M., Fang, J. L., and Guo, L. (2012) Mitochondrial dysfunction induced by sertraline, an antidepressant agent. Toxicol. Sci. 127 (2), 582−91. (51) de Faria, P. A., Bettanin, F., Cunha, R. L., Paredes-Gamero, E. J., Homem-de-Mello, P., Nantes, I. L., and Rodrigues, T. (2015) Cytotoxicity of phenothiazine derivatives associated with mitochondrial dysfunction: a structure-activity investigation. Toxicology 330, 44−54. (52) Rana, P., Nadanaciva, S., and Will, Y. (2011) Mitochondrial membrane potential measurement of H9c2 cells grown in highglucose and galactose-containing media does not provide additional predictivity towards mitochondrial assessment. Toxicol. In Vitro 25 (2), 580−7. (53) Moreno, S. N. (1996) Pentamidine is an uncoupler of oxidative phosphorylation in rat liver mitochondria. Arch. Biochem. Biophys. 326 (1), 15−20. (54) Ramamoorthy, H., Abraham, P., and Isaac, B. (2014) Mitochondrial dysfunction and electron transport chain complex defect in a rat model of tenofovir disoproxil fumarate nephrotoxicity. J. Biochem. Mol. Toxicol. 28 (6), 246−55. (55) Rodriguez, R. J., and Acosta, D., Jr. (1996) Inhibition of mitochondrial function in isolated rate liver mitochondria by azole antifungals. J. Biochem. Toxicol. 11 (3), 127−31. (56) Le Dinh, T., Freneaux, E., Labbe, G., Letteron, P., Degott, C., Geneve, J., Berson, A., Larrey, D., and Pessayre, D. (1988) Amineptine, a tricyclic antidepressant, inhibits the mitochondrial oxidation of fatty acids and produces microvesicular steatosis of the liver in mice. J. Pharmacol Exp Ther 247 (2), 745−50. (57) Rolo, A. P., Oliveira, P. J., Moreno, A. J., and Palmeira, C. M. (2001) Chenodeoxycholate is a potent inducer of the permeability transition pore in rat liver mitochondria. Biosci. Rep. 21 (1), 73−80. (58) Mahmud, T., Rafi, S. S., Scott, D. L., Wrigglesworth, J. M., and Bjarnason, I. (1996) Nonsteroidal antiinflammatory drugs and uncoupling of mitochondrial oxidative phosphorylation. Arthritis Rheum. 39 (12), 1998−2003.

(59) Park, J. H., and Kim, T. H. (2005) Release of cytochrome c from isolated mitochondria by etoposide. J. Biochem Mol. Biol. 38 (5), 619−23. (60) Santos, N. A., Medina, W. S., Martins, N. M., Mingatto, F. E., Curti, C., and Santos, A. C. (2008) Aromatic antiepileptic drugs and mitochondrial toxicity: effects on mitochondria isolated from rat liver. Toxicol. In Vitro 22 (5), 1143−52. (61) Begriche, K., Massart, J., Robin, M. A., Borgne-Sanchez, A., and Fromenty, B. (2011) Drug-induced toxicity on mitochondria and lipid metabolism: mechanistic diversity and deleterious consequences for the liver. J. Hepatol. 54 (4), 773−94. (62) Hroudova, J., and Fisar, Z. (2012) In vitro inhibition of mitochondrial respiratory rate by antidepressants. Toxicol. Lett. 213 (3), 345−52. (63) Irwin, W., Fontaine, E., Agnolucci, L., Penzo, D., Betto, R., Bortolotto, S., Reggiani, C., Salviati, G., and Bernardi, P. (2002) Bupivacaine myotoxicity is mediated by mitochondria. J. Biol. Chem. 277 (14), 12221−7. (64) Hroudova, J., and Fisar, Z. (2010) Activities of respiratory chain complexes and citrate synthase influenced by pharmacologically different antidepressants and mood stabilizers. Neuro Endocrinol Lett. 31 (3), 336−342. (65) Modica-Napolitano, J. S., Lagace, C. J., Brennan, W. A., and Aprille, J. R. (2003) Differential effects of typical and atypical neuroleptics on mitochondrial function in vitro. Arch. Pharmacal Res. 26 (11), 951−9. (66) Qi, X., Cai, Y., Gong, L., Liu, L., Chen, F., Xiao, Y., Wu, X., Li, Y., Xue, X., and Ren, J. (2007) Role of mitochondrial permeability transition in human renal tubular epithelial cell death induced by aristolochic acid. Toxicol. Appl. Pharmacol. 222 (1), 105−10. (67) Jacob, M., Bjarnason, I., Rafi, S., Wrigglesworth, J., and Simpson, R. J. (2001) A study of the effects of indometacin on liver mitochondria from rats, mice and humans. Aliment. Pharmacol. Ther. 15 (11), 1837−42. (68) Wei, Y. H., Lu, C. Y., Lin, T. N., and Wei, R. D. (1985) Effect of ochratoxin A on rat liver mitochondrial respiration and oxidative phosphorylation. Toxicology 36 (2−3), 119−30. (69) Costantini, P., Petronilli, V., Colonna, R., and Bernardi, P. (1995) On the effects of paraquat on isolated mitochondria. Evidence that paraquat causes opening of the cyclosporin A-sensitive permeability transition pore synergistically with nitric oxide. Toxicology 99 (1−2), 77−88. (70) Simon, N., Morin, C., Urien, S., Tillement, J. P., and Bruguerolle, B. (2003) Tacrolimus and sirolimus decrease oxidative phosphorylation of isolated rat kidney mitochondria. Br. J. Pharmacol. 138 (2), 369−76.

167


Evaluation of in Vitro Mitochondrial Toxicity Assays and

Recommend Documents