Article pubs.acs.org/crt
Toxicity of Carboxylic Acid-Containing Drugs: The Role of Acyl Migration and CoA Conjugation Investigated Toni Lassila,*,†,‡ Juho Hokkanen,§ Sanna-Mari Aatsinki,§ Sampo Mattila,† Miia Turpeinen,‡,∥ and Ari Tolonen§ †
Department of Chemistry, University of Oulu, P.O. Box 3000, 90014 Oulu, Finland Research Unit of Biomedicine, Department of Pharmacology and Toxicology, and Medical Research Center Oulu, University of Oulu, P.O. Box 5000, 90014 Oulu, Finland § Admescope Ltd., Typpitie 1, 90620 Oulu, Finland ∥ Administration Center, Oulu University Hospital, P.O. Box 10, 90029 OYS, Oulu, Finland ‡
S Supporting Information *
ABSTRACT: Many carboxylic acid-containing drugs are associated with idiosyncratic drug toxicity (IDT), which may be caused by reactive acyl glucuronide metabolites. The rate of acyl migration has been earlier suggested as a predictor of acyl glucuronide reactivity. Additionally, acyl Coenzyme A (CoA) conjugates are known to be reactive. Here, 13 drugs with a carboxylic acid moiety were incubated with human liver microsomes to produce acyl glucuronide conjugates for the determination of acyl glucuronide half-lives by acyl migration and with HepaRG cells to monitor the formation of acyl CoA conjugates, their further conjugate metabolites, and transacylation products with glutathione. Additionally, in vitro cytotoxicity and mitochondrial toxicity experiments were performed with HepaRG cells to compare the predictability of toxicity. Clearly, longer acyl glucuronide half-lives were observed for safe drugs compared to drugs that can cause IDT. Correlation between half-lives and toxicity classification increased when “relative half-lives,” taking into account the formation of isomeric AG-forms due to acyl migration and eliminating the effect of hydrolysis, were used instead of plain disappearance of the initial 1-O-β-AG-form. Correlation was improved further when a daily dose of the drug was taken into account. CoA and related conjugates were detected primarily for the drugs that have the capability to cause IDT, although some exceptions to this were observed. Cytotoxicity and mitochondrial toxicity did not correlate to drug safety. On the basis of the results, the short relative half-life of the acyl glucuronide (high acyl migration rate), high daily dose and detection of acyl CoA conjugates, or further metabolites derived from acyl CoA together seem to indicate that carboxylic acid-containing drugs have a higher probability to cause drug-induced liver injury (DILI).
■
INTRODUCTION Certain drugs can cause rare but serious idiosyncratic drug toxicity (IDT), and several drugs, such as zomepirac, benoxaprofen, and bromfenac, have been withdrawn from the market due to an unacceptable number of IDTs such as anaphylaxis and liver damage.1,2 The exact mechanism of IDT is not known, but it is believed that one reason may be electrophilic metabolites that react with biological macromolecules such as proteins or DNA, causing the toxicity.3 Drugs containing a carboxylic acid group are often metabolized to acyl glucuronide conjugates (AGs), which are known to be unstable through hydrolysis and acyl migration. Acyl glucuronides can be reactive by two mechanisms, i.e., the trans-acylation mechanism and the glycation mechanism. In the trans-acylation mechanism, glucuronic acid is directly replaced in a nucleophilic substitution reaction by a nucleophile. The glycation mechanism occurs via acyl migration in which the 1© XXXX American Chemical Society
O-β-acyl glucuronide is transformed to 2, 3, or 4-O-acyl glucuronides, where the alpha-hydroxy aldehyde group is exposed and can react with suitable nucleophiles.4 Both mechanisms have been shown to occur; for example, the lysine residues of human serum albumin have been shown to be reactive toward acyl glucuronides via trans-acylation and glycation reactions.5 The stability and half-life of 1-O-β-acyl glucuronide has been suggested as a measure of its reactivity and toxicity in earlier papers, utilizing several different matrices, such as buffers, microsomal incubation and plasma, for the experiments.6−9 Benet et al. found an excellent correlation between covalent binding to serum albumin and the degradation rate of acyl glucuronides for six drugs.9 Sawamura et al. found a correlation between the IDT tendency and halfReceived: July 28, 2015
A
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology Scheme 1. Metabolic Reactions and Reactivity of Acyl Glucuronides and CoA Conjugatesa
a
X = N, S, and O. The glycation reaction can occur for the 2-, 3-, and 4-isomers, but only the 3- and 4-isomers can stabilize further by the Amadori rearrangement.
lives of acyl glucuronides for 21 drugs.8 However, use of this assay has been criticized due to the lack of complete validation and comparison with other assay types.4 In addition to acyl glucuronides, carboxylic acid-containing drugs can be metabolized to acyl coenzyme A conjugates (CoA), which can be even more reactive than the corresponding acyl glucuronide. For example, the CoA conjugates of phenylpropionic acid, mefenamic acid, and naproxen were found to be 70, 80, and 100 times more reactive toward glutathione than the respective acyl glucuronides.10−12 The reaction between CoA or AG with glutathione will form an S-acyl-glutathione thioester (SG), which may be reactive as well and has been shown to react with Nacetylcysteine.13−15 Acyl CoA is formed through an acyl adenylate (adenosine monophosphate, AMP) intermediate, which can also be reactive.16 CoA and AMP conjugates can further react to form glycine or taurine amides or carnitine esters, which may interfere with β-oxidation or deplete the CoA and carnitine reserves of the cells and disrupt mitochondrial function.17 These reactions are summarized in Scheme 1. Often, the production of AGs is the limiting factor in their reactivity testing. In this work, we used a method introduced by
Chen et al. in which the AGs are produced in primary liver microsomal incubation with UDPGA and alamethicin, and the production of AGs is terminated by the addition of UDP.18 Acyl migration rate is then determined directly by a secondary incubation, without the need for synthesis of pure 1-O-β-AG. Because of the presence of liver microsomal enzymes, ester hydrolysis of the acyl glucuronide occurs in parallel with the acyl migration reaction, producing the initial free drug. Because of this, we used a secondary parameter called the “relative halflife”, in which the hydrolysis reaction has been separated from acyl migration. The chromatographic peak areas of all AG isomers, including 1-O-β-AG, were measured by UPLC/TOFMS, and the disappearance was calculated based on the relative fraction of 1-O-β-AG from the sum of all AG isomers at each time point, with an assumption that all AG isomers from the same drug have similar responses in UPLC/ESI-MS. In order to study CoA conjugation in a whole cell system, the test compounds were incubated with HepaRG cells, cell lysis was performed with cyclohexane, and CoA, AMP, taurine, glycine, carnitine, and glutathione conjugates were searched by UPLC/MS. Additionally, in vitro cytotoxicity and mitochondrial toxicity experiments were performed. The results from all of B
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology Chart 1. Structures of the Test Compounds
incubation. The secondary incubations were sampled at 0, 15, 30, 60, 120, 240, and 360 min, mixed with acetonitrile containing 2% formic acid, and the samples were analyzed using UPLC/TOF-MS to monitor the chemical stability of the AG formed in the microsomal incubation and the formation of isomeric AGs via acyl migration. HepaRG Incubations. Vials of cryopreserved HepaRG-cells (lot HPR116178-TA08) were thawed and suspended in 50 mL of HepaRG-medium (lot MIL60026) supplemented with thawing additives (lot 670035, Biopredic International). Cells were centrifuged (300 g, 2 min) and resuspended in InVitro GRO HI-medium (lot C03104A, Bioreclamation IVT). The cell density and viability were determined by the trypan blue exclusion method, and viability of the cells was 80−90%. Incubations were conducted in 48-well tissue culture plates. Study compound stock solutions were diluted to 200 μM in incubation medium (pH 7.4), which was then mixed with an equal volume of cell suspension with 2 million viable cells/mL, to have a total incubation volume of 300 μL, final test compound concentration of 100 μM, cell content of 1 million/mL, and final DMSO concentration of 0.5%. Three microliters of 100 mM GSH solution was added to the incubation medium, and the final GSH concentration was 1 mM. Forty microliter samples were collected at 0, 15, 30, 60, 120, 240, and 360 min time points from the well and suspended in an equal volume of cold acetonitrile with 0.5% phosphoric acid. Several time points were used to maximize the likelihood of observing potentially unstable and low abundance conjugates. The cells were
these experiments were correlated with the tendency of the 13 test drugs, whose structures and classification are presented in Chart 1, to cause IDTs.
■
EXPERIMENTAL PROCEDURES
Reagents and Materials. HPLC-grade acetonitrile was purchased from Merck (LiChrosolv GG, Darmstadt, Germany). Ammonium formate was purchased from BDH Laboratory Supplies (Poole, England). Laboratory water was distilled and purified with a Direct-Q water purifier (Millipore, Molsheim, France). The test compounds, glutathione, UDPGA, alamethicin, UDP, chlorpromazine, and carbonyl cyanide m-chlorophenyl hydrazone (CCCP) were purchased from Sigma-Aldrich (Helsinki, Finland). HepaRG cells (pooled) were acquired from Biopredic International (Rennes, France). Human liver microsomes (pooled M-class) were from Bioreclamation IVT (Brussels, Belgium). Human Liver Microsome Incubations. Stock solutions of the study compounds were prepared in DMSO. Compounds were incubated at 100 μM concentration (DMSO content 0.5%) with pooled human liver microsomes (1.0 mg/mL) in the presence of cofactor UDPGA (1 mM) and alamethicin (15 mg/mL) in 100 mM phosphate buffer (pH 6.5) for 40 min for the formation of AG conjugates. One hundred microliters of the preincubation solution was diluted to 500 μL with 100 mM phosphate buffer (pH 7.4) containing UDP (5 mM) to suppress the formation of AGs in the secondary C
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology lysed by the addition of 160 μL of cyclohexane and vortexed, and the water-phase was isolated and analyzed immediately by an UPLC/ PDA/QE-orbitrap-MS, to monitor the formation of AG-, GSH-, CoA-, AMP, glycine-, taurine-, and carnitine-conjugates. LC/MS Analyses. For the analysis of HepaRG-incubation samples, a Thermo Ultimate 3000 UHPLC with an autosampler, vacuum degasser, photodiode-array (PDA) detector, and column oven connected to a Q-Exactive orbitrap mass spectrometer was used. The analytical column used was a Waters BEH ShieldRP18 (2.1 × 50 mm, 1.7 μm, Waters Corp, Milford, MA, USA). The eluents were 2 mM ammonium formate (A, pH 6.5) and acetonitrile (B). An elution starting with an isocratic step at 5% B for 0.5 min, followed by a linear gradient to 90% B over the next 5.5 min, was employed. The flow rate was 0.5 mL/min, and the column oven temperature was 40 °C. The flow was directed to the MS through a PDA detector. The measurements were made with both positive and negative ionization polarities. The positive ionization mode of electrospray was used with a spray voltage of 2.0 kV and negative ionization mode with spray voltage of 3.0 kV. Sheath gas was set to 10, auxiliary gas to 4, and capillary temperature to 300 °C. The data-dependent-MS2 measurement mode was used, which performed a full mass spectral scan and triggered further MS/MS experiments for specified target ions. A fullscan was performed with a resolution of 35000 (full width at halfmaximum, at m/z 200), while an Automated Gain Control (AGC) target of a million ions, maximum injection time of 50 ms, and scan range of 100−1500 m/z were used. MS/MS was performed with a resolution of 17500, while the AGC target was a million ions, with a maximum injection time of 50 ms, isolation range of 4 m/z, a mass range starting from 60 m/z, and a combination of fragmentation energies (NCE stop) of 25, 35, and 48. The data were processed with Thermo Xcalibur 3.0.63 software. For analysis of acyl-glucuronide stability from HLM incubations, a Waters Acquity ultraperformance liquid chromatographic (UPLC) system (Waters Corp., Milford, MA, USA) with an autosampler, vacuum degasser, and column oven was used. The analytical column used was a Phenomenex Kinetex XB-C18 (2.1 × 50 mm, 1.7 μm, Phenomenex Inc., Torrance, CA, USA). The eluents were 2 mM ammonium formate (A, pH 6.5) and acetonitrile (B). An elution starting with an isocratic step at 1% B for 0.5 min, followed by linear gradient to 40% B over the next 3.0 min was employed, followed by a 1.5 min gradient to 90% B. The flow rate was 0.5 mL/min, and the column oven temperature was 35 °C. The flow was directed to MS through an Acquity photodiode-array detector. LC/TOF-MS data were acquired with a Waters LCT Premier XE-time-of-flight (Q-TOF) high-resolution mass spectrometer (Waters Corp., Milford, MA, USA) equipped with a LockSpray electrospray ion source. A positive ionization mode of electrospray was used with a capillary voltage of 3200 V and a cone voltage of 40 V. The mass range of m/z 100−1200 was acquired using “W-mode” flight tube optics at a mass resolution of about 12 000 (fwhm, at m/z 556). The desolvation temperature was 400 °C and source temperature 150 °C. Nitrogen was used as both the desolvation and nebulizer gas. Leucine enkephalin was used as the lock mass compound ([M + H]+ m/z 556.2771) for accurate mass measurements and was infused to the LockSpray ion source via a separate ionization probe using a syringe pump. The mass spectrometer and UPLC system were operated with MassLynx 4.1 software. Safety Categories. Categorizing was based on FDA labeling of the drugs (http://labels.fda.gov/) in the same way as Chen et al. and Pedersen et al.19,20 The FDA classifies the side effects of drugs to three different sections in the drug label. The most serious side effects are mentioned in the black box warning section, less serious side effects are listed in the warnings and precautions section, and even milder side effects in the adverse reactions section. If a drug label mentioned liver injury or other liver-related interaction in the boxed warning or warning and precautions section, it was categorized as a warning drug. Drugs with hepatic interactions mentioned in the adverse reactions category or drugs with no mention on hepatic interactions were categorized as safe. Drugs that have been withdrawn from the market
due to hepatotoxic or anaphylactic effects were categorized as withdrawn. Calculating Observed Half-Lives and Relative Half-Lives. The half-lives of 1-O-β acyl glucuronides were calculated with the following equation: T (1/2) = ln 2/K where K is the degradation constant. For the observed degradation half-lives of each 1-O-β-AG, K(obs) was calculated by fitting the following equation to the experimental results: A(1‐O‐β‐AG) = A(0)exp(− K (obs)t ) For the relative half-lives, K(rel) was calculated by fitting the equation: A(1‐O‐β‐AG)/A(AG, total) = A(0)exp(− K (rel)t) where A(1-O-β-AG) is the peak area of 1-O-β-acyl glucuronide, A(AG, total) is the peak area of all isomeric acyl glucuronides, including the 1O-β-AG isomer, A(0) is the peak area of the 1-O-β- acyl glucuronide at the beginning of the secondary incubation, and t is the time in the secondary incubation. Fitting was performed using GraphPad 5.04 software (GraphPad Software Inc.). Mitochondrial Toxicity and Cytotoxicity Experiments. Cryopreserved HepaRG-cells (lot HPR116178-TA08) were thawed, and cell viability was determined by the trypan blue exclusion method. Approximately 3.5 × 104 and 2 × 104 viable cells were plated onto 96well tissue culture plates (Greiner-Bio One) for cytotoxicity and mitochondrial toxicity assay studies, respectively. Three days after plating, for the cytotoxicity assay, the cells were exposed to the study compounds (1, 30, and 100 μM concentration) for 24 h diluted into InVitro GRO HI-medium (lot C03104A, Bioreclamation IVT). For the mitochondrial toxicity assay, the cells were exposed to the study compounds (1, 30, and 100 μM concentration) for 2 h diluted into DMEM medium (Sigma-Aldrich) containing 10 mM galactose (Sigma-Aldrich) as the sole carbohydrate source. The final DMSO concentration in both assays was 0.1%. Appropriate control compounds were used in both assays; chlorpromazine for cytotoxicity at 1−100 μM concentration and CCCP for mitochondrial toxicity at 0.1−10 μM concentration.21,22 Cell viability and cytotoxicity were assayed from the same well using the CellTiter-Glo assay (ATP from cells) and CytoTox-ONE (LDH leakage to medium) assays (Promega) according to the manufacturer’s instructions in a 96-well plate format. The results were analyzed with a multiwell plate reader (Infinite M1000 Pro, Tecan) for luminescence (ATP) and fluorescence (LDH; 560ex, 590em). Mitochondrial toxicity was assayed using the Mitochondrial ToxGlo assay (Promega) according to the manufacturer’s instructions in a 96-well plate format. The results were analyzed for fluorescence (cytotoxicity; 485ex, 520em) and luminescence (ATP).
■
RESULTS Formation of Acyl Glucuronides and the Extent of Total Acyl Migration. Acyl glucuronide metabolites were identified from the data based on an increase of 176.0321 Da (C6H8O6) with respect the parent compound given by LC/ TOF-MS analysis and cleavage of the corresponding unit from the molecular ion in MS/MS experiments. The amount of total acyl glucuronides relative to the parent compound in the HepaRG incubation determined by LC/UV showed no difference between safe and nonsafe drugs. Drugs from all categories produced varying amounts (0.03−2.9% from parent, assuming a similar UV response for the parent and formed AGs) of acyl glucuronides in the HepaRG incubation. Valproic acid does not have any UV response, and the extent of AG formation could not be determined; therefore, it is presented as D
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology Table 1. Acyl Migration, Formation of Acyl Glucuronides, Conjugates, and Cell Toxicitya half-lives (h)
AG migration at 6 hd
conjugates
cell toxicity
category
compound
Obsb
Relc
HLM
HepaRG
total AG formation, % of parente
SG
CoA/Tau/Gly/Car
Mitotox
Cytotox
BSEPif
safe safe safe safe warning warning warning warning warning warning warning withdrawn withdrawn
montelukast telmisartan repaglinide furosemide gemfibrozil valproic acid ibuprofen indomethacin naproxen tolmetin diclofenac zomepirac isoxepac
4.3 3.7 9.2 2.6 6.3 4.4 2.6 1.2 1.5 0.5 0.2 1.1 0.6
>40 32 18 3.1 > 40 30 2.4 2.2 1.4 0.4 0.3 1.4 0.6
0.0 11.0 22.6 66.2 0.0 11.7 77.4 76.6 59 97.8 100 91.5 100
0 2 30 35 7 0 69 29 38 13 22 22 15
1.0 1.0 2.9 0.4 2.8 NA 0.5 2.0 0.2 0.03 2.2 0.1 0.04
− + − − − + + + + − + − −
− + − − − + + − − + + + −
** * ** 0 0 0 0 * 0 0 ** 0 0
** * 0 0 0 0 0 0 0 * 0 0 0
** ** * 0 0 0 0 * 0 NA * NA NA
+ = conjugate detected; − = conjugate not detected. ** = toxic; ATP/viability 70% inhibition at 50 μM concentration); * = moderate inhibitor of BSEP (30−70% at 50 μM concentration); 0 = no BSEP inhibition (6 h), the sampling up to 6 h is adequate because the relevant information from these results is mainly to differentiate the compounds that have half-lives below 3 h or clearly above 6 h, as the threshold between reactive and safe acyl glucuronides appeared to be around these time points.8 Mitochondrial Toxicity and Cytotoxicity. Results from the mitochondrial toxicity and cytotoxicity experiments are presented in Table 1 as fraction viability/cytotoxicity compared to those of vehicle/toxicity control treatments. These experiments showed no correlation to the safety classification of the drugs. In the mitochondrial toxicity experiment, safe category drug repaglinide and warning category drug diclofenac were the most toxic, while safe category drug telmisartan and warning category drug indomethacin also showed some mitochondrial toxicity. Montelukast displayed severe cytotoxicity, which was also shown as mitochondrial toxicity. In the cytotoxicity experiment, in addition to montelukast, telmisartan and tolmetin also showed some toxicity. As no correlation was found between these toxicity experiments and the safety categories of the study compounds, an explanation for the observed mitochondrial toxicity and cytotoxicity was sought elsewhere. Interestingly, the toxicity results correlated very well with the BSEP inhibition data that were available from the literature for 10 out of 13 compounds studied.20,23 Montelukast and telmisartan, which showed both cytotoxicity and mitochondrial toxicity, are strong BSEP-inhibitors as described by Pedersen et al., and repaglinide, indomethacin, and diclofenac, all positive in the toxicity assays, are also classified as BSEP-inhibitors.20 Gemfibrozil, furosemide, valproic acid, ibuprofen, and naproxen, all negative in the toxicity assays, are poor BSEP-inhibitors. Relating Half-Lives, Acyl Migration, Formation of Acyl Glucuronides, and Cellular Toxicity to Daily Dose. F
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology Table 3. Detected CoA-Route Conjugates and GSH Trans-Acylation Products (SG)
a
category
compound
SG
AMP
CoA
taurine
glycine
carnitine
safe safe safe safe warning warning warning warning warning warning warning withdrawn withdrawn
montelukast telmisartan repaglinide furosemide gemfibrozil valproic acid ibuprofen indomethacin naproxen tolmetin diclofenac zomepirac isoxepac
−b +a − − − + + + + − + − −
− − − − − − − − − − − − −
− − − − − + + − − − − − −
− − − − − + + − − + + + −
− + − − − − − − − − − − −
− − − − − − + − − − − − −
+ = conjugate detected. b− = conjugate not detected.
Table 4. Formulas, m/z Values, and Retention Times (tR) of the Detected CoA-Route Conjugates and GSH Trans-Acylation Products positive ions
negative ions
conjugate
formula
m/z calc.
m/z obs.
mDa
m/z calc.
m/z obs.
mDa
tR
telmisartan-SG telmisartan-Gly valproic acid-CoA valproic acid-SG valproic acid-Tau ibuprofen-CoA ibuprofen-SG ibuprofen-Tau ibuprofen-Car indomethacin-SG naproxen-SG tolmetin-Tau diclofenac-SG diclofenac-Tau zomepirac-Tau
C43H45N7O7S C35H33N5O3 C29H50N7O17P3S C18H31N3O7S C10H21NO4S C34H52N7O17P3S C23H33N3O7S C15H23NO4S C20H32NO4 C29H31ClN4O9S C24H29N3O8S C17H20N2O5S C24H26Cl2N4O7S C16H16Cl2N2O4S C17H19ClN2O5S
804.3174 572.2656 894.2270 434.1955 252.1264 956.2426 496.2112 314.1421 350.2331 647.1573 520.1748 365.1166 585.0972 403.0281 399.0776
804.3157 572.2656 894.2271 434.1957 − 956.2425 496.2107 314.1422 350.2328 647.1563 520.1751 365.1175 585.0969 403.0269 −
1.7 0.0 −0.1 −0.2 − 0.1 0.5 −0.1 0.4 1.0 −0.3 −0.9 0.3 1.1 −
802.3028 570.2511 892.2124 432.1810 250.1119 954.2281 494.1966 312.1275 348.2186 645.1428 518.1603 363.1020 583.0826 401.0135 397.0630
802.3037 − 892.2135 432.1818 250.1117 954.2283 − 312.1275 − 645.1428 518.1603 363.1023 583.0821 401.0126 397.0634
−0.9 − −1.1 −0.8 0.1 0.6 − 0.0 − 0.0 0.0 −0.3 0.6 0.9 −0.3
3.33 4.24 2.58 2.56 2.35 3.20 3.17 3.23 3.35 3.26 2.76 2.81 3.07 3.38 3.00
conjugates. MS/MS data for the conjugates are presented in Table S1.
HepaRG incubations. The results for CoA-route conjugates AMP, CoA, taurine, glycine, and carnitine conjugates and GSHthioester are presented in Table 3. More detailed UPLC/-HRMS data for the detected conjugates are presented in Table 4. The acyl adenylate conjugate, i.e., intermediate in the formation of acyl-CoA-conjugate, was not detected for any of the compounds. Carnitine and glycine conjugates were rare and were detected only for one compound each, i.e., the carnitine conjugate for ibuprofen and glycine conjugate for telmisartan, and acyl CoA conjugates were detected for two compounds, i.e., ibuprofen and valproic acid. Taurine and glutathione thioester conjugates were most common and were detected for five and six compounds, respectively. With the exception of telmisartan, all of these conjugates were detected for compounds in the nonsafe category. However, no corresponding conjugates were detected for gemfibrozil and isoxepac, both in the nonsafe category. Overall, the abundance of the detected CoA-conjugates or conjugates derived from CoA was low, usually 0.01%−7% compared to corresponding total acyl glucuronide, as measured with MS and assuming a similar MS response. Highest relative levels were observed with ibuprofen and valproic acid CoAconjugates, and with ibuprofen, tolmetin and diclofenac taurine
■
DISCUSSION Safety Classification. Classifying drugs into different safety categories was based on the FDA labeling, similar to that in several recent publications, and was mainly similar to what was done by Sawamura et al.8,19,20 However, different safety categories were given to gemfibrozil and furosemide. Gemfibrozil was classified into the warning category instead of the safe category because abnormal liver function tests have been observed during gemfibrozil use, and it was mentioned in the more serious warnings and precautions section of the drug label. These symptoms have been mild and reversible when drug administration has been stopped, so it is debatable whether the drug should be classified as safe in this context. Sawamura et al. classified furosemide as a warning drug and according to the drug label, furosemide can cause numerous adverse reactions, including increased liver enzymes, anaphylactic reactions, and hematologic reactions.8 All of these reactions, however, are classified as milder adverse reactions, so furosemide was classified as a safe drug in this study. The slightly ambiguous classification of these two drugs is reflected in the results as the relative half-lives corrected with daily dose G
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology
much more to the degradation constant than acyl migration. Gemfibrozil was an exception in this case because it was classified as a warning drug, but its long observed half-life increased to an even longer relative half-life. Another exception was furosemide, which was classified as a safe drug, but it had only a small difference between its observed and relative halflives, both of which were only slightly longer than the half-lives of the nonsafe drugs. However, as discussed above, different safety categories were earlier suggested for these two compounds by Sawamura et al. (safe for gemfibrozil and warning for furosemide).8 Among the compounds with a large difference in direct (observed) and relative half-lives, transacylated GSH thioester or CoA-derived conjugates were detected only for telmisartan and valproic acid, but not for montelukast, repaglinide, and gemfibrozil. Additionally, the difference between observed and relative half-lives was greatest with safe drugs. These results would indicate that hydrolysis and the trans-acylation reaction do not directly correlate with each other. Otherwise, the higher amounts of hydrolysis and trans-acylation products would be expected for unsafe compounds, assuming that transacylated products are not themselves prone to hydrolysis. Determination of the relative half-lives of gemfibrozil and montelukast acyl glucuronides was not successful because of their high stability and low acyl migration, and in the calculation of dose-corrected relative half-lives, a value of 40 h was used. The degradation of the gemfibrozil and montelukast acyl glucuronides occurred solely through hydrolysis, and no acyl migration isomers were detected for these compounds in the microsomal incubation. For this reason, the calculated dose-corrected relative half-lives for these compounds are only estimates and may be actually much higher. Division of the relative half-lives by daily dose improved correlation to the safety categories mainly because drugs in the safe category had a much lower daily dose than drugs in the nonsafe categories. Montelukast and repaglinide in particular are used in very low doses. Valproic acid was classified as a warning drug, but its AG had relative half-lives similar to those of safe drugs. However, when the large dose of valproic acid was taken into account, it had values similar to those of other warning drugs. Gemfibrozil was classified as a warning drug, but it had a very stable AG. When its estimated relative half-life was divided with its high daily dose, the result was somewhere between other nonsafe and safe drugs. Furosemide, which had a low daily dose and relatively short relative half-life for a safe drug, similarly received a value that was between that of other safe and nonsafe drugs. Reactive Metabolites via Oxidative Reactions. Although the acyl migration rate of the observed acyl glucuronide conjugates did correlate well with the safety classification of their parent compounds, it is worth noting that in all cases it cannot be clearly shown that the known toxicity, or the metabolism-caused toxicity, of the model compounds would be directly related to this phenomenon. In addition, it was not possible to set up a suitable test group of compounds without oxidative reactive metabolites. Valproic acid was classified as a warning drug due to its well-known capability to cause liver injury.4 However, the toxicity of valproic acid is more likely mediated by 2-ene and 2,4-diene-valproate, which are formed by oxidative metabolism, and it is unlikely that its toxicity is caused by reactive acyl glucuronides.25 In a good correlation to this, valproic acid acyl glucuronide had a higher
of these two drugs were somewhat between the other safe and nonsafe drugs, and without the dose correction, these drugs had relative half-lives that were unusual in their safety classification. The classification was similar to that of Pedersen et al., with the exception that telmisartan was classified as a safe drug (instead of a warning drug).20 This was because elevated liver enzymes were mentioned in the adverse reactions section, but in the warnings and precautions section, there was only a recommendation to use lower doses with patients with impaired hepatic function and no mention of any liver damage. It is worth mentioning that the classification based on FDA labeling may be problematic, as the quality and completeness of the data and the reports on which they are based are not known. Thus, it may not necessarily give an accurate indication of the safety of the drug. This is reflected by the different classifications in this study and in the study by Sawamura et al. and Pedersen et al., even though the same classification method was used.8,20 Yet another issue is the acceptable level of idiosyncratic drug toxicity. Usually, it is very low, and for example, many NSAIDs, such as ibuprofen, are still being used despite the low risk of IDTs involved.24 Observed and Relative Half-Lives. Sawamura et al. found a good correlation between the tendency of a drug to cause IDT and acyl glucuronide degradation rate in potassium phosphate buffer (KPB), while the correlation was not as good in bovine serum albumin (BSA) solution or human plasma.8 This correlation was improved here using relative half-lives, meaning that the effect of hydrolysis has been calculated out of the degradation constant. This way, the relative half-life represents more accurately the rate of acyl migration, and the reactivity of the acyl glucuronide via the glycation mechanism is emphasized. The observed relative half-lives match well with the earlier studies performed in buffer solution without active esterase enzymes that could catalyze the hydrolysis, but these studies required synthesized acyl glucuronides. Here, the acyl glucuronides were produced with liver microsomal incubations and an immediate secondary incubation (with some active esterases) was performed as a stability test. The hydrolysis of the acyl glucuronide generates the parent drug, which is not intrinsically reactive, and the rate of the hydrolysis does not necessarily correlate with the reactivity of the acyl glucuronide. It is worth noting that the assumption that all observed disappearance is caused only by two mechanisms, i.e., hydrolysis and acyl migration, is not completely valid as trans-acylation of proteins may also occur. However, as the observed relative half-lives (neglecting hydrolysis) fit very well with the earlier results for synthesized acyl glucuronides in protein-free buffer,8 this suggests that the observed relative halflife is actually mechanistically affected mainly by acyl migration. Yet, covalent binding in the physiological concentration of the compound is more likely to occur via the glycation mechanism than via the trans-acylation reaction.5 The role of the transacylation reaction, however, was additionally studied here as a measurement of GSH thioesters in HepaRG cells, but it is not known whether the observed thioesters are formed via the CoA or AMP conjugates or acyl glucuronides. In either case, the detection of GSH thioesters indicates that the compound can react via the trans-acylation mechanism and the short relative half-life of the acyl glucuronide indicates that it can react via the glycation mechanism. However, due to the potential instability of the GSH thioesters, the lack of their detection does not indicate that the compound cannot undergo the trans-acylation reaction. For many of the safe drugs, hydrolysis contributed H
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology
which showed a correlation between the hydrolysis rate of acylCoA conjugates and the reaction rate of acyl-CoA conjugates with glutathione.30 In their study, CoA conjugate of ibuprofen was shown to be the most stable among the investigated CoA conjugates with respect to hydrolysis, being in agreement with the findings in this study. Ibuprofen was the only compound for which the carnitine ester was detected in addition to the glutathione thioester and taurine conjugates. The taurine conjugate of ibuprofen has been previously detected in human urine, while no glycine conjugate has been detected.31 In the study by Sidenius et al., the degradation rates of indomethacin CoA and tolmetin CoA were shown to be 3 and 9 times higher, respectively, than the degradation rate of ibuprofen CoA.30 In this study, the half-life of ibuprofen AG was 2 and 5 times longer than those of indomethacin AG and tolmetin AG, respectively, so there seems to be a correlation between the stability of AG and CoA-conjugates, although the number of examples for this conclusion is low. Similarly, the observation of valproic acid CoA conjugate shows that it is stable enough to be analyzed, which is in good agreement of a long half-life of valproic acid AG conjugate. For compounds other than ibuprofen and valproic acid, the CoA conjugate might have been hydrolyzed despite attempts to analyze the samples immediately and the use of acidification in sample preparation to prevent hydrolysis, as it is assumed that hydrolysis occurs much more slowly in low pH.32 However, the stabilities of the different conjugates or their behavior in different pH values were not determined. For indomethacin and naproxen, only the GSH thioester was detected but no CoA-related conjugates. The CoA conjugate might have reacted completely to the thioester, or it might not have formed at all, meaning that the observed GSH-thioester had formed from a reaction between acyl glucuronide or AMP conjugate and glutathione. For zomepirac, only the taurine conjugate was detected, but Olsen et al. were able to also detect the CoA, taurine, glycine, and glutathione thioester conjugates with rat hepatocytes in vitro and additionally carnitine ester in vivo with rats.33 Olsen used a higher concentration of zomepirac, and taurine was much more abundant than glycine or CoA conjugates, so it is possible that the glycine and CoA were not detected here because of their low abundance. This is supported by a recent study, in which no zomepirac CoA conjugate was found in incubations with human liver microsomes.34 However, it is noteworthy that no glutathione thioester was detected in our experiment but that it was detected in works by Olsen et al. and Grillo et al. by incubation with rat hepatocytes.33,35 It is possible that this may be a result of the differences between rat hepatocytes and HepaRG cells. Another possibility is that the glutathione thioester was hydrolyzed as it was shown to degrade rapidly, within a few minutes, during incubation with rat hepatocytes.35 The taurine conjugate and the GSH thioester were detected for diclofenac. As demonstrated by Grillo et al., the thioester is most likely not formed from the acyl glucuronide.35 Because the taurine conjugate is formed via reactive CoA or AMP intermediates, the observation suggests the existence of these intermediates, and it is possible that the glutathione thioester was formed via these intermediates as well. For tolmetin, only GSH thioester was detected here, while Olsen et al. detected also CoA, taurine, and carnitine conjugates in the rat liver homogenate and thioester in bile when tolmetin was administered in vivo.37 However, only trace amounts of the
half-life and especially higher relative half-life than other drugs in the warning category. However, because valproic acid is consumed in high doses, it had a dose-corrected relative half-life value similar to that of other nonsafe drugs. Another possibility is that its toxicity is mediated by reactive CoA-conjugates or by the disruption of β-oxidation. This is supported by the detection of CoA and taurine conjugates in the hepatocyte incubation. The detection of glutathione thioester also points to reactivity that is most likely occurring through CoA conjugation because of the high stability of the acyl-glucuronide conjugate. This is supported by a study with ibuprofen, in which the formation of ibuprofen glutathione thioester was significantly reduced by the inhibition of CoA formation but not by inhibition of acyl glucuronide formation.26 Diclofenac is known to have reactive metabolites other than acyl glucuronides, i.e., quinone imines and arene oxides that are generated by oxidative P450 metabolism. These are reactive and known to form conjugates with glutathione.27 Zomepirac and tolmetin have also been shown to generate glutathione adducts in vitro with human liver microsomes and in vivo in rats; most likely, the arene oxide reactive intermediate is formed on the pyrrole ring of these compounds.28 Indomethacin is metabolized to desmethyldeschlorobenzoylindometahcin, which forms a reactive metabolite in activated neutrophils by reaction with HOCl and is mediated by the myeloperoxidase system. This metabolite is a possible reason for the blood disorders, such as agranulocytosis or aplastic anemia, caused by this drug but does not fully explain the rare hepatic effects.29 Glutathione adducts were detected in this study in the incubation with HepaRG cells for diclofenac, zomepirac, and tolmetin (data not shown). The oxidative reactive metabolites as intermediates to these conjugates may contribute to the toxicity of these drugs, but their relative role compared to acyl glucuronides and CoA conjugates in causing idiosyncratic drug toxicity is difficult to estimate. The reactive metabolites of valproic acid, diclofenac, zomepirac, and tolmetin, which have the highest incidence of reported IDT, may play a role in their toxicity. A recent study, however, reported that oxidative metabolites of tolmetin, zomepirac, and ibuprofen do not contribute significantly to covalent binding to proteins compared to conjugative metabolites, but oxidative metabolites had a significant role for covalent binding of fenclozic acid, suprofen, and tienilic acid.34 CoA-Route Conjugates. The initial step of the acyl-CoAroute metabolism, i.e., formation of AMP conjugate, was not detected for any of the compounds studied, and it is possibly too unstable to be detected as further conjugates derived from it were observed. Acyl AMP has been earlier detected for mefenamic acid, and there is a significant difference in the reactivities of its acyl AMP and acyl CoA conjugates toward glycine, taurine, glutathione, and N-acetyl cysteine. CoA conjugates were more reactive toward thiol groups, but AMP conjugates were much more reactive toward amino acids, e.g., glycine.15,16 CoA conjugates were detected here only for valproic acid and ibuprofen. However, taurine or glycine conjugates, which require CoA or AMP conjugate as an intermediate, were detected for diclofenac, telmisartan, tolmetin, and zomepirac. It may be that in addition to reactions with taurine, glycine, or glutathione, the CoA conjugates of these compounds hydrolyzed or reacted with other nucleophiles that were present in the incubation. This is supported by a paper by Sidenius et al., I
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology tolmetin-CoA conjugate were detected in incubations with human liver microsomes in the recent study, so it is possible that more CoA conjugate is formed in rats compared to that in humans.34 Acyl-CoA-route conjugates were not detected for gemfibrozil and isoxepac. Isoxepac is a withdrawn drug, and gemfibrozil is classified as a warning drug. For all other drugs in these classes, acyl-CoA route conjugates were detected. The results for gemfibrozil are similar compared to those of montelukast and repaglinide, as for all of these compounds the half-life of acyl glucuronide was high, and no other conjugates were detected. On the basis of the discussion above, the CoA conjugates are more likely to be more stable for compounds that have stable acyl glucuronides and if these conjugates were formed, they should have been detectable. For isoxepac, it is likely that if any CoA was formed, it was quickly degraded, because its acyl glucuronide had a short half-life as well. Glycine conjugate was detected only for telmisartan, together with GSH thioester. This makes telmisartan an exception among the compounds classified as safe because it was the only drug for which these conjugates were detected. In general, the results were similar to those of valproic acid because relative half-lives for both AGs were long, and for both compounds, GSH thioester and acyl-CoA-derived conjugates were detected. The abundance of the CoA-derived conjugates was very low in most cases, and it was not possible to acquire confirmatory MS/MS data for most of the conjugates. Mass spectrometric response can vary a lot for different conjugate types, but it seems that all CoA-derived conjugates had much lower abundance compared to that of acyl glucuronides, which has been observed by others as well.33,36 The abundance of these conjugates was generally approximately 0.01−1% of the abundance of the corresponding acyl glucuronides, with the exception of valproic acid, whose CoA-conjugate had roughly about 7% abundance in comparison to that of the total acyl glucuronides (based on LC/MS peak area). Mitochondrial Toxicity and Cytotoxicity Experiments. Cytotoxicity was chosen as the routine assay for cell functions. Mitochondrial toxicity was also studied as CoA-conjugates are known to possibly interfere with mitochondrial function.17 HepaRG cells were chosen because those are metabolically active and cost-efficient. Also, glucuronidation and CoA-route metabolism have been shown to occur in the model. Cytotoxicity or mitochondrial toxicity experiments did not correlate with the given safety classification of the drugs, as many drugs in the safe category displayed significant toxicity, and most drugs in the nonsafe categories did not display any toxicity. The results, however, showed a clear correlation between mitochondrial toxicity and BSEP inhibition data obtained from the literature.20,23 All of the compounds that displayed mitochondrial toxicity were also BSEP inhibitors. The most effective BSEP inhibitors, montelukast and telmisartan, were also cytotoxic. BSEP is the canalicular efflux transporter responsible for carrying bile salts such as cholate and chenodeoxycholate conjugates from the hepatocyte into the bile. Bile acids are key components in cholesterol catabolism by metabolizing cholesterol into a more water-soluble molecule. BSEP inhibition by several pharmaceuticals has been shown to be well correlated with DILI.20,38 According to Aleo et al., compounds causing both mitochondrial dysfunction and BSEP inhibition account for the most severe DILI concerns. However, it is not known how IDT events relate to this information.39
Metabolites formed in these toxicity experiments were not analyzed, but as the same HepaRG cell lot in somewhat similar conditions was used, it can be assumed that the same metabolites as discussed above are formed in these experiments as in earlier experiments. A longer incubation time was used in the cytotoxicity assay (24 h) compared to the mitochondrial toxicity assay (2 h). More metabolites might form when longer incubation times are used, but usually, the reaction is complete in a few hours, and only small amounts of metabolites are formed at longer time points. Current knowledge suggests that BSEP expression is lower in HepaRG cells compared to fresh human hepatocytes. However, the canalicular transporter functionality and relative abundance compared to other transporters in these cells is close to the same level as in hepatocytes.40 HepaRG cells are thus a relevant model when compared to BSEP inhibition data obtained from the literature. The literature data were from a metabolically noncompetent system and demonstrate BSEP inhibition caused only by the drug itself and not by its metabolites.20 Because of this, toxicity experiments and BSEP inhibition might not be directly comparable. However, as the correlation was clear, it is possible that cellular toxicity observed in the experiments is not due to metabolites but the compound itself. Another hepatic canalicular transporter, MRP2, is a carrier for organic anions such as glucuronides and GSH conjugates to bile.41 Interestingly, multiple drugs such as NSAIDs may also inhibit MRP2 (in addition to BSEP), which presumably causes a higher risk for DILI than BSEP inhibition alone.42 It is possible that the compounds studied here and/or their metabolites inhibit human MRP2 or BSEP and contribute to their toxicity and DILI risk. In fact, such data exist that telmisartan and montelukast strongly inhibit MRP2.43,44 Interestingly, evidence exits that acyl glucuronides and acylCoA conjugates are protein-reactive and thus could react with canalicular membrane proteins, affecting their function.45 This could result in disturbances in transporting other substrates from the hepatocyte, with the risk of IDTs.
■
CONCLUSIONS The half-lives of degradation were determined for the acyl glucuronides of 13 drugs, and longer half-lives were detected for drugs in the safe category compared to those of drugs in the warning or withdrawn categories. This difference was yet clearer when the relative half-lives, based on acyl migration rate only, after exclusion of the hydrolysis reaction, were compared. This correlation was further improved when the daily dose of the drug was taken into account. The relative amount of acyl migrated isomers at the end of the in vitro incubation also showed correlation to the drug safety classification, i.e., in most cases the relative abundance of acyl migration isomers compared to the 1-O-β-acyl glucuronide was lower for safe drugs than for nonsafe drugs. General formation rate of acyl glucuronides in the HepaRG incubation did not correlate with safety classification. When comparing formation of CoA-route metabolites, the most common conjugates detected were those with taurine and glutathione thioester, and CoA conjugates were detected only for ibuprofen and valproic acid. With the exception of telmisartan, all of these CoA-route conjugates or GSH-thioesters were detected for drugs in the nonsafe categories, suggesting these may possibly contribute to the toxicity. However, these conjugates were not observed for isoxepac and gemfibrozil, which both have a nonsafe classification. The reason for detecting some CoA-derived J
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology
human serum albumin: Identification of binding sites and mechanisms of reaction by tandem mass spectrometry. Drug Metab. Dispos. 3, 369− 376. (6) Bolze, S., Bromet, N., Gay-Fuetry, C., Massiere, F., Boylieu, R., and Hulot, T. (2002) Development of an in vitro screening model for the biosynthesis of acyl glucuronide metabolites and the assessment of their reactivity toward human serum albumin. Drug Metab. Dispos. 30, 404−413. (7) Wang, J., Davis, M., Li, F., Azam, F., Scatina, J., and Talaat, R. (2004) A novel approach for predicting acyl glucuronide reactivity via Schiff base formation: Development of rapidly formed peptide adducts for LC/MS/MS measurements. Chem. Res. Toxicol. 17, 1206−1216. (8) Sawamura, R., Okudaira, N., Watanabe, K., Murai, T., Kobayashi, Y., Tachibana, M., Ohnuki, T., Masuda, K., Honma, H., Kurihara, A., and Okazaki, O. (2010) Predictability of idiosyncratic drug toxicity risk for carboxylic acid-containing drugs based on the chemical stability of acyl glucuronides. Drug Metab. Dispos. 38, 1857−1864. (9) Benet, L. Z., Spahn-Langguth, H., Iwakawa, S., Volland, C., Mizuma, T., Mayer, S., Mutschler, E., and Lin, E. T. (1993) Predictability of the covalent binding of acidic drugs in man. Life Sci. 53, 141−146. (10) Grillo, M. P., Lohr, M. T., and Wait, J. C. M. (2012) Metabolic activation of mefenamic acid leading to mefenamyl-S-acyl-glutathione adduct formation in vitro an in vivo in rat. Drug Metab. Dispos. 40, 1515−1526. (11) Li, C., Benet, L. Z., and Grillo, M. P. (2002) Studies on the chemical reactivity of 2-phenyulpropionic acid 1-O-acyl glucuronide and S-acyl-CoA thioester metabolites. Chem. Res. Toxicol. 15, 1309− 1317. (12) Olsen, J., Bjornsdottir, I., Tjornelund, J., and Hansen, S. H. (2002) Chemical reactivity of the naproxen coenzyme A thioester towards bionucleophiles. J. Pharm. Biomed. Anal. 29, 7−15. (13) Shore, L. J., Fenselau, C., King, A. R., and Dickinson, R. G. (1995) Characterization and formation of the glutathione conjugate of clofibric acid. Drug Metab. Dispos. 1, 119−123. (14) Grillo, M. P., Knotson, C. G., Sanders, P. E., Waldon, D. J., Hua, F., and Ware, J. A. (2003) Studies on the chemical reactivity of diclofenac acyl glucuronide with glutathione: Identification of diclofenac-S-acyl-glutathione in rat bile. Drug Metab. Dispos. 31, 1327−1336. (15) Horng, H., and Benet, L. Z. (2013) Characterization of the acyladenylate linked metabolite of mefenamic acid. Chem. Res. Toxicol. 26, 465−476. (16) Horng, H., and Benet, L. Z. (2013) The nonenzymatic reactivity of the acyl-linked metabolites of mefenamic acid towards amino and thiol functional group bionucleophiles. Drug Metab. Dispos. 41, 1923− 1933. (17) Darnell, M., and Weidolf, L. (2013) Metabolism of xenobiotic carboxylic acids: Focus on coenzyme A conjugation, reactivity, and interference with lipid metabolism. Chem. Res. Toxicol. 26, 1139−1155. (18) Chen, Z., Holt, T. G., Pivnichny, J. V., and Leung, K. (2007) A simple in vitro model to study the stability of acylglucuronides. J. Pharmacol. Toxicol. Methods 55, 91−95. (19) Chen, M., Vijay, V., Shi, Q., Liu, Z., Fang, H., and Tong, W. (2011) FDA-appproved drug labeling for the study of drug-induced liver injury. Drug Discovery Today 16, 697−703. (20) Pedersen, J. M., Matsson, P., Bergström, C. A. S., Hoogstraate, J., Norén, A., LeCluyse, E. L., and Artursson, P. (2013) Early identification of clinically relevant drug interactions with the human bile salt export pump (BSEP/ABCB11). Toxicol. Sci. 136, 328−343. (21) Aninat, C., Piton, A., Glaise, D., Le Charpentier, T., Langouët, S., Morel, F., Guguen-Guillouzo, C., and Guillouzo, A. (2006) Expression of cytochromes P450, conjugating enzymes and nuclear receptors in human hepatona HepaRG cells. Drug Metab. Dispos. 34, 75−83. (22) Wallace, K. B., and Starkov, A. A. (2000) Mitochondrial targets of drug toxicity. Annu. Rev. Pharmacol. Toxicol. 40, 353−388.
conjugates, but not CoA-conjugate itself, was probably the low stability of the CoA conjugate that seemed to correlate with acyl-glucuronide stability from the same compounds. In vitro cytotoxicity experiments did not show correlation to the drug safety classification, although it did correlate well with the human BSEP inhibition. Short relative half-life of the acyl glucuronides, combined with the high daily dose and the detection of CoA conjugates or its further conjugate metabolites, collectively points to some link to possible DILI and can be useful in the early in vitro screening of drug candidates. However, because it was not possible to distinguish between drugs in the warning and withdrawn categories, the decision on terminating drug candidate development cannot be based solely on the observation of these features. The method used here to produce acyl glucuronides and determine the acyl migration in one incubation by stopping the production of glucuronides with UDP proved to be fast, easy to perform, and suitable to drug screening, without the need for chemical synthesis of the conjugates. More optimized sample preparation methods might improve the detection of CoA conjugates if they are unstable.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemrestox.5b00315. MS/MS data for the detected conjugates (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: toni.lassila@oulu.fi. Notes
The authors declare no competing financial interest.
■
ABBREVIATIONS AG, acyl glucuronide; AGC, automated gain control; BSA, bovine serum albumin; BSEP, bile salt export pump; Car, carnitine; CCCP, carbonyl cyanide m-chlorophenyl hydrazine; DILI, drug-induced liver injury; ESI, electrospray ionization; FDA, US Food and Drug Administration; fwhm, full width at half-maximum; KPB, potassium phosphate buffer; IDT, idiosyncratic toxicity; MRP2, multidrug resistance-associated protein 2; MS/MS, tandem mass spectrometry; NSAID, nonsteroidal anti-inflammatory drug; PDA, photodiode-array; Q-TOF-MS, quadrupole-time-of-flight mass spectrometer; SG, S-acyl-glutathione thioester; Tau, taurine; UDP, uridine diphosphate; UDPGA, uridine diphosphate glucuronic acid; UPLC, ultrahigh performance liquid chromatography
■
REFERENCES
(1) Uetrecht, J. (2007) Idiosyncratic drug reactions: current understanding. Annu. Rev. Pharmacol. Toxicol. 47, 513−539. (2) Lasser, K. E., Allen, P. D., Woolhandler, S. J., Himmelstein, D. U., Wolfe, S. M., and Bor, D. H. (2002) Timing of new black box warnings and withdrawals for prescription medications. JAMA 287, 2215−2220. (3) Kaplowitz, N. (2005) Idiosyncratic drug hepatotoxicity. Nat. Rev. Drug Discovery 4, 489−499. (4) Regan, S. L., Maggs, J. L., Hammond, T. G., Lambert, C., Williams, D. P., and Park, B. K. (2010) Acyl glucuronides: The good, the bad and the ugly. Biopharm. Drug Dispos. 31, 367−395. (5) Ding, A., Zia-Amirhosseini, P., Mcdonagh, A. F., Burlingame, A. L., and Benet, L. Z. (1995) Reactivity of tolmetin glucuronide with K
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX
Article
Chemical Research in Toxicology (23) Lai, Y. (2014) Transporters in Drug Discovery and Development: Detailed Concepts and Best Practice, Woodhead Publishing, Cambridge, UK. (24) Gulmez, S. E., Larrey, D., Pageaux, G. P., Lignot, S., Lassalle, R., Jové, J., Gatta, A., McCormick, P. A., Metselaar, H. J., Monteiro, E., Thorburn, D., Bernal, W., Zouboulis-Vafiadis, I., de Vries, C., PerezGutthann, S., Sturkenboom, M., Bénichou, J., Montastruc, J. L., Horsmans, Y., Salvo, F., Hamoud, F., Micon, S., Droz-Perroteau, C., Blin, P., and Moore, N. (2013) Transplantation for acute liver failure in patients exposed to NSAIDs or paracetamol (acetaminophen): the multinational case-population SALT study. Drug Saf. 36, 135−144. (25) Chang, T. K. H., and Abbott, F. S. (2006) Oxidative stress as a mechanism of valproic acid-associated hepatoxicity. Drug Metab. Rev. 38, 627−639. (26) Grillo, M. P., and Hua, F. (2008) Enantioselective formation of ibuprofen-S-acyl-glutathione in vitro in incubations of ibuprofen with rat hepatocytes. Chem. Res. Toxicol. 21, 1749−1759. (27) Boelsterli, U. A. (2003) Diclofenac-induced liver injury: a paradigm of idiosyncratic drug toxicity. Toxicol. Appl. Pharmacol. 192, 307−322. (28) Chen, Q., Doss, G. A., Elaine, C. T., Wensheng, L., Tang, Y. S., Braun, M. P., Didolkar, V. D., Strauss, J. R., Wang, R. W., Stearns, R. A., Evans, D. C., Baillie, T. A., and Tang, W. (2006) Evidence for the bioactivation of zomepirac and tolmetin by an oxidative pathway: Identification of glutathione adducts in vitro in human liver microsomes and in vivo in rats. Drug Metab. Dispos. 34, 145−151. (29) Ju, C., and Uetrecht, P. (1998) Oxidation of a metabolite of indomethacin (desmethyldeschlorobenzoylindomethacin) to reactive intermediates by activated neutrophils, hypochlorous acid and the myeloperoxidase system. Drug Metab. Dispos. 26, 676−680. (30) Sidenius, U., Skonberg, C., Olsen, J., and Hansen, S. H. (2004) In vitro reactivity of carboxy acid-CoA thioesters with glutathione. Chem. Res. Toxicol. 17, 75−81. (31) Shirley, M. A., Guan, X., Kaiser, D. G., Halstead, G. W., and Baillie, T. A. (1994) Taurine conjugation ibuprofen in human and in rat liver in vitro. Relationship to metabolic chiral inversion. J. Pharm. Exp. Ther. 269, 1166−1175. (32) Hyneck, M. L., Munafo, A., and Benet, L. Z. (1988) Effect of pH on acyl migration and hydrolysis of tolmetin glucuronide. Drug Metab. Dispos. 16, 322−324. (33) Olsen, J., Li, C., Bjorsdottir, I., Sidenius, U., Hansen, S. H., and Benet, L. Z. (2005) In vitro an in vivo studies on acyl-coenzyme Adependent bioactivation of zomepirac in rats. Chem. Res. Toxicol. 18, 1729−1736. (34) Darnell, M., Breitholtz, K., Isin, E. M., Jurva, U., and Weidolf, L. (2015) Significantly different covalent binding of oxidative metabolites, acyl glucuronide, and S-acyl CoA conjugates formed form xenobiotic carboxylic acids in human liver microsomes. Chem. Res. Toxicol. 28, 886−896. (35) Grillo, M. P., and Hua, F. (2003) Identification of zomepirac-Sacyl-glutathione in vitro in incubations with rat hepatocytes and in vivo in rat bile. Drug Metab. Dispos. 31, 1429−1436. (36) Grillo, M. P., Hua, F., Knutson, C. G., Ware, J. A., and Li, C. (2003) Mechanistic studies on the bioactivation of diclofenac: Identification of diclofenac-S-acyl-glutathione in vitro in incubations with rat and human hepatocytes. Chem. Res. Toxicol. 16, 1410−1417. (37) Olsen, J., Li, C., Skonberg, C., Bjornsdottir, I., Sidenius, U., Benet, L. Z., and Hansen, S. H. (2007) Studies on the metabolism of tolmetin to chemically reactive acyl-coenzyme A thioester intermediates in rats. Drug Metab. Dispos. 35, 758−764. (38) Dawson, S., Stahl, S., Paul, N., Barber, J., and Kenna, J. G. (2012) In vitro inhibition of the bile salt export pump correlates with risk of cholestatic drug-induced liver injury in humans. Drug Metab. Dispos. 40, 130−138. (39) 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, 1015−1022.
(40) Le Vee, M., Noel, G., Jouan, E., Stieger, B., and Fardel, O. (2013) Polarized expression of drug transporters in differentiated human hepatoma HepaRG cells. Toxicol. In Vitro 27, 1979−1986. (41) Nies, A. T., and Keppler, D. (2007) The apical conjugate efflux pump ABCC2 (MRP2). Pfluegers Arch. 453, 643−659. (42) Kubitz, R., Dröge, C., Stindt, J., Weissenberger, K., and Häussinger, D. (2012) The bile salt export pump (BSEP) in health and disease. Clin. Res. Hepatol. Gastroenterol. 36, 536−553. (43) Weiss, J., Sauer, A., Divac, N., Herzog, M., Schwedhelm, E., Böger, R. H., Haefeli, W. E., and Benndorf, R. A. (2010) Interaction of angiotensin receptor type 1 blockers with ATP-binding cassette transporters. Biopharm. Drug Dispos. 31, 150−161. (44) Roy, U., Chakravarty, G., Honer Zu Bentrup, K., and Mondal, D. (2009) Montelukast is a potent and durable inhibitor of multidrug resistance protein 2-mediated efflux of taxol and saquinavir. Biol. Pharm. Bull. 32, 2002−2009. (45) Boelsterli, U. A. (2002) Xenobiotic acyl glucuronides and acyl CoA thioester as protein-reactive metabolites with the potential to cause idiosyncratic drug reactions. Curr. Drug Metab. 3, 439−450. (46) FDA Online Label Repository. http://labels.fda.gov/.
L
DOI: 10.1021/acs.chemrestox.5b00315 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX