Identification of an isoform catalyzing the CoA conjugation of nonsteroidal anti-inflammatory drugs and the evaluation of the expression levels of acyl-CoA synthetases in the human liver

Article abstract of DOI:10.1016/j.bcp.2020.114303

Nonsteroidal anti-inflammatory drugs (NSAIDs) containing carboxylic acid are conjugated with coenzyme A (CoA) or glucuronic acid in the body. It has been suggested that these conjugates are associated with toxicities, such as liver injury and anaphylaxis,

Full text of DOI:10.1016/j.bcp.2020.114303

Journal Pre-proofs
Identification of an isoform catalyzing the CoA conjugation of nonsteroidal
anti-inflammatory drugs and the evaluation of the expression levels of acyl-
CoA synthetases in the human liver
Hiroki Hashizume, Tatsuki Fukami, Kanji Mishima, Hiroshi Arakawa, Kenji
Mishiro, Yongjie Zhang, Masataka Nakano, Miki Nakajima
PII:
DOI:
Reference:
S0006-2952(20)30539-6
https://doi.org/10.1016/j.bcp.2020.114303
BCP 114303
To appear in:
Biochemical Pharmacology
Received Date:
Revised Date:
Accepted Date:
9 September 2020
21 October 2020
22 October 2020
Please cite this article as: H. Hashizume, T. Fukami, K. Mishima, H. Arakawa, K. Mishiro, Y. Zhang, M.
Nakano, M. Nakajima, Identification of an isoform catalyzing the CoA conjugation of nonsteroidal anti-
inflammatory drugs and the evaluation of the expression levels of acyl-CoA synthetases in the human liver,
Biochemical Pharmacology (2020), doi: https://doi.org/10.1016/j.bcp.2020.114303
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2020 Elsevier Inc. All rights reserved.

Identification of an isoform catalyzing the CoA conjugation of nonsteroidal anti-
inflammatory drugs and the evaluation of the expression levels of acyl-CoA synthetases
in the human liver
1
1,2
1
1
3
Hiroki Hashizume , Tatsuki Fukami , Kanji Mishima , Hiroshi Arakawa , Kenji Mishiro ,
2
,4
1,2
1,2
Yongjie Zhang , Masataka Nakano , and Miki Nakajima
1Drug Metabolism and Toxicology, Faculty of Pharmaceutical Sciences, Kanazawa
University, Kanazawa, Japan
2WPI Nano Life Science Institute (WPI-NanoLSI), Kanazawa University, Kanazawa, Japan
3Institute for Frontier Science Initiative, Kanazawa University, Kanazawa, Japan
4Clinical Pharmacokinetics Laboratory, School of Basic Medicine and Clinical Pharmacy,
China Pharmaceutical University, Nanjing, China
To whom all correspondence should be sent:
Tatsuki Fukami
Drug Metabolism and Toxicology
Faculty of Pharmaceutical Sciences
Kanazawa University
Kakuma-machi, Kanazawa 920-1192, Japan
Tel: +81-76-234-4438/Fax: +81-76-264-6282
E-mail: tatsuki@p.kanazawa-u.ac.jp
1

Abstract
Nonsteroidal anti-inflammatory drugs (NSAIDs) containing carboxylic acid are
conjugated with coenzyme A (CoA) or glucuronic acid in the body. It has been suggested that
these conjugates are associated with toxicities, such as liver injury and anaphylaxis, through
their binding via trans-acylation to cellular proteins. Although studies on glucuronidation
have progressed, studies on CoA conjugation of drugs catalyzed by acyl-CoA synthetase
(ACS) enzymes are still in the early stages. This study aimed to clarify the human ACS
isoforms responsible for CoA-conjugation of NSAIDs through consideration of the hepatic
expression levels of ACS isoforms. We found that among 10 types of NSAIDs, propionic
acid-class NSAIDs, namely, alminoprofen, flurbiprofen, ibuprofen, ketoprofen, and
loxoprofen, were conjugated with CoA in the human liver, whereas NSAIDs in the other
classes, including diclofenac and mefenamic acid, were not. qRT-PCR revealed that among
the 26 ACS isoforms, ACSL1 was the most highly expressed in the human liver, followed by
ACSM2B. The propionic acid-class NSAIDs were conjugated with CoA by recombinant
human ACSL1. The protein binding abilities of the CoA conjugates and the glucuronide
forms of propionic acid-class NSAIDs were compared as an index of toxicity. The CoA
conjugates had stronger adduct formation with liver microsomal proteins than glucuronides
for all 5 propionic acid-class NSAIDs. In conclusion, we found that propionic acid-class
NSAIDs could be conjugated to CoA by ACSL1 in the human liver to form CoA conjugates,
which likely cause toxicity by protein adduct formation.
Keywords: CoA conjugation, Acyl-glucuronide, Acyl-CoA synthetase, NSAID, Protein
binding
2

1
. Introduction
Drugs containing carboxylic acids have been suggested to cause various toxicities, such as
anaphylaxis, liver injury, and Stevens-Johnson syndrome (SJS) [1]. It is well known that they
are metabolized to acyl-glucuronides, which are likely related to their toxicities because of
their instability and tendency to bind to biomolecules [2]. The relevance of acyl-glucuronides
to their toxicities has not been fully and experimentally proven, but it has been demonstrated
that their instability is highly correlated with the degree of toxicity of the parent compounds
[3].
Drugs containing carboxylic acids are also metabolized to acyl-CoA conjugates in
addition to acyl-glucuronides. Acyl-CoA conjugation is a thioesterification two-step reaction
as follows [4]: (1) the carboxylate ion is replaced with ATP to form acyl adenylate, and then
(2) nucleophilic attack of CoA on the carbonyl carbon releases AMP to form a thioester bond
with the substrate. These reactions are catalyzed by acyl-CoA synthetase (ACS), which
consists of 26 isoforms in humans that are categorized into six groups: short-chain family
(
ACSS), medium-chain family (ACSM), long-chain family (ACSL), very long-chain family
ACSVL), bubblegum family (ACSBG), and other family (ACSF) [5]. The ACS isoforms are
(
localized in various organelles in cells. For example, ACSS and ACSM are mainly localized
in mitochondria, ACSL and ACSVL are mainly localized in the endoplasmic reticulum
system, and ACSBG and ACSF are mainly localized in the cytoplasm [5,6,7]. The ACSS,
ACSM, ACSL, and ACSVL families are classified by their substrate selectivity, which is
based on the length of the carbon chain of the substrate fatty acids C2-4, C4-10, C12-20, and
C18-26, respectively. ACSBG is the human ortholog of the bubblegum gene expressed in the
Drosophila optic lobe, and the ACSF family deviates from the above five families by
phylogenetic tree analysis based on amino acid sequences [4,5]. CoA conjugation in lipid
metabolism has been extensively studied, but CoA conjugation of drugs is still incompletely
understood.
Nonsteroidal anti-inflammatory drugs (NSAIDs) are used for the treatment of pain in
various parts of the body, such as the bones, joints, and muscles [8]. NSAIDs often cause
3

gastrointestinal injury and rarely cause anaphylaxis, SJS, or liver injury as side effects.
Suggested possible mechanisms to cause these side effects are the accumulation of
arachidonic acid in cells, inhibition of mitochondrial function, and the production of reactive
metabolites [9,10]. Many NSAIDs possess carboxylic acids in their structures and are mainly
subjected to glucuronidation in humans [11]. It has been reported that acyl-glucuronides are
relevant to adverse reactions owing to their instability, although their involvement in toxicity
has not been experimentally proven [12]. Recently, Darnell et al. [13] reported that the
covalent binding ability of acyl-CoA conjugates of ibuprofen, one of the NSAIDs containing
carboxylic acid, via trans-acylation to proteins is higher than that of its acyl-glucuronide,
implying the relevance of acyl-CoA conjugation to toxicity. ACS isoforms involved in the
acyl-CoA conjugation of drugs and their preferences for various NSAIDs remain to be
clarified. In this study, we sought to clarify the ACS isoforms catalyzing acyl-CoA
conjugation of NSAIDs by evaluating the expression levels of 26 ACS isoforms in the human
liver.
4

2
2
. Materials and Methods
.1. Chemicals and Reagents. Alminoprofen, indomethacin, ketoprofen, salicylic acid,
sulindac, and valeric acid were purchased from Wako Pure Chemical Industries (Osaka,
Japan). Ibuprofen acyl-glucuronide (IBP Glu) was purchased from Toronto Research
Chemical (Toronto, Canada). Adenosine 5′-triphosphate magnesium salt, alamethicin,
coenzyme A sodium salt hydrate cofactor for acyl transfer (CoA), diclofenac, flurbiprofen,
ibuprofen sodium salt, lauric acid, loxoprofen sodium salt, mefenamic acid, and UDP-
glucuronic acid (UDPGA) were purchased from Sigma-Aldrich (St. Louis, MO). RNAiso and
random hexamers were obtained from Takara Bio (Shiga, Japan). ReverTra Ace and Luna
Universal qPCR Master Mix were purchased from Toyobo (Osaka, Japan) and New England
Biolabs (Ipswich, MA), respectively. All primers were commercially synthesized at
Integrated DNA Technologies (Coralville, IA). Rabbit anti-human ACSL1 and rabbit anti-
human ACSM2A antibodies were obtained from Cell Signaling Technology Japan (Tokyo,
Japan) and Abcam (Cambridge, MA), respectively. IRDye 680-labeled goat anti-rabbit
secondary antibody and Odyssey Blocking buffer were from LI-COR Biosciences
(Cambridge, UK). All other chemicals and solvents were of the highest grade commercially
available.
2
.2. Human Livers. Human liver samples obtained from 28 donors (Table 1) were supplied
by the National Disease Research Interchange (Philadelphia, PA) through the Human and
Animal Bridging Research Organization (Chiba, Japan). The use of the human livers was
approved by the Ethics Committees of Kanazawa University (Kanazawa, Japan). Human liver
homogenate (HLH) was prepared by homogenizing the tissue in 100 mM Tris-HCl buffer (pH
7
.4) containing 20% glycerol, 100 mM KCl, and 1 mM EDTA (pH 7.4), and pooling samples
if necessary. Pooled human liver microsomes (HLM) were prepared from the above HLHs as
reported previously [14].
5

2
.3. CoA conjugation of NSAIDs in HLH and recombinant ACSs. The ten NSAIDs shown
in Fig. 1, alminoprofen, diclofenac, flurbiprofen, ibuprofen, indomethacin, ketoprofen,
loxoprofen, mefenamic acid, salicylic acid, and sulindac, were used as substrates. The
reaction mixtures (final volume of 0.2 ml) contained 100 mM Tris-HCl (pH 7.4), 15 mM
MgCl , 150 mM KCl, 0.5 mg/mL pooled HLH (10 donors, No. 1–10 in Table 1), and 3 mM
2
ATP. Ibuprofen sodium salt was dissolved in water, and the other substrates were dissolved in
DMSO. The final concentration of the substrates was 500 µM. After preincubation at 37°C
for 2 min, reactions were initiated by the addition of 20 µL of CoA (final concentration: 0.3
mM). After a 6-hr incubation at 37°C, the reaction was terminated by the addition of 10 µL of
6
0% perchloric acid. The mixture was centrifuged at 20,380 g for 5 min, and a 10 µL aliquot
of the supernatant was subjected to liquid chromatography-tandem mass spectrometry (LC-
MS/MS). An LCMS-8045 (Shimadzu, Kyoto, Japan) equipped with an LC-20AD HPLC
system was used. The column was a Develosil ODS-UG-3 (2.0 × 150 mm, 3 µM; Nomura
Chemical, Seto, Japan). The flow rate was 0.2 mL/min, and the column temperature was
4
0°C. Nitrogen was used as the nebulizer gas and drying gas. The operating parameters were
optimized as follows: nebulizer gas flow, 3 L/min; drying gas flow, 15 L/min; desolvation
line temperature, 300°C; and heat block temperature, 40°C. LC-MS/MS was performed with
positive or negative electrospray ionization. For measurement of CoA conjugation from 10
NSAIDs in HLH, CoA conjugates of NSAIDs were monitored in single ion monitoring
(SIM). The mobile phase was (A) 10 mM ammonium acetate (pH 5.6) and (B) acetonitrile.
The conditions of the mobile phase and m/z values are summarized in Table 2. The analytical
data were processed using LabSolutions (version 5.82.1; Shimadzu).
For propionic acid-class NSAIDs, further evaluation of CoA conjugation was performed
by incubating 50 µM propionic acid-class NSAIDs with recombinant enzymes or HLH after
confirmation of linearity in terms of protein concentration and incubation time. The reaction
mixtures were the same as described above except for the concentration of HLH. The
concentrations of the enzyme sources and the incubation times are shown in Table 3. After
incubation, the mixture was centrifuged at 20,380 g for 5 min, and a 10 µL aliquot of the
6

supernatant was subjected to LC-MS/MS as described above. CoA conjugates were
monitored by multiple reaction monitoring (MRM), which was established by performing a
product ion scan and voltage optimization (Table 4). The other conditions were the same as
described above.
2
.4. Real-Time RT-PCR Analysis. Total RNA was extracted from 3 individual human liver
samples (Nos. 1–3 in Table 1) using RNAiso, and cDNA was prepared using ReverTra Ace
according to the manufacturer's protocols. The ACS mRNA levels were quantified by real-
time RT-PCR using Mx3000P (Stratagene, San Diego, CA). A 1 μl portion of the reverse-
transcription mixture was added to a PCR mixture containing Luna Universal qPCR Master
Mix solution containing 5 pmol primers in a final volume of 20 µL. For 20 out of the 26
isoforms, multiple splicing variants have been registered in the National Center for
Biotechnology Information (https://www.ncbi.nlm.nih.gov/). Therefore, primers were
designed to amplify all splicing variants in each isoform. The sequences of the primers are
shown in Table 4. The PCR conditions were as follows: After an initial denaturation at 95°C
for 60 s, amplification was performed by denaturation at 94°C for 20 s, then annealing and
extension at the temperature and time shown in Table 5 for 40 cycles. The copy number was
calculated by standard curves prepared from the purified PCR product as described previously
[15].
2
.5. Construction of recombinant human ACSL1 or ACSM2B expressed in baculovirus-
infected Sf21 cells. Expression systems of human ACSL1 and ACSM2B were prepared using
the Bac-to-Bac Baculovirus Expression System (Invitrogen) according to the manufacturer’s
protocol. Human ACSL1 and ACSM2B cDNA were obtained by PCR using cDNA prepared
from the human liver and the following primer sets: sense primer (5-
ATGCAAGCCCATGAGCTGTT-3) and antisense primer (5-CTAAACCTTGATAGTGG-
AATAGAG-3) for ACSL1 and sense primer (5-TTCCGACAGTGCAGGGATTC-3) and
antisense primer (5-TGTCTCCTAGACGCCTCACT-3) for ACSM2B. The PCR products
7

were subcloned into the pTargeT mammalian expression vector (Promega, Maison, WI). The
nucleotide sequences were confirmed to be consistent with the reference sequences (accession
no. NM_001995.4 and NM_182617.3 for human ACSL1 and ACSM2B, respectively) by
DNA sequence analysis (FASMAC, Kanagawa, Japan). The ACSL1 or ACSM2B cDNA in
the pTargeT vector was resubcloned into the pFastBac1 vector using BamH I and Sal I. The
pFastBac1 vector containing ACSL1 or ACSM2B cDNA was transformed into DH10Bac
competent cells, and the other steps were performed according to the method reported
previously [16]. A homogenate of Sf21 cells infected with the constructed baculoviruses was
prepared, and the human ACSL1 and ACSM2B proteins were obtained. The protein
concentration was determined according to the method of Bradford [17] using γ-globulin as a
standard.
2
.6. Immunoblot analyses for ACSL1 and ACSM2B. SDS-PAGE and immunoblot
analyses were performed according to a method reported previously [18] with slight
modifications. HLH (10 or 50 µg) and Sf21 cell homogenates expressing human ACSL1 (2.5
µg), ACSM2B (15 µg), or mock (15 µg) were separated on 7.5% polyacrylamide gels and
electrotransferred onto Immobilon-P (Millipore, Billerica, MA) membranes. The membranes
were probed with polyclonal rabbit anti-human ACSL1 or rabbit anti-human ACSM2A
antibody and the corresponding fluorescent dye-conjugated secondary antibody. An Odyssey
infrared imaging system (LI-COR Biosciences, Lincoln, NE) was used for detection.
2
.7. CoA conjugation of lauric acid or valeric acid by HLH and recombinant ACS. CoA
conjugation of lauric acid or valeric acid was measured according to a method reported
previously [19]. The reaction mixture and assay conditions were the same as described above,
except that the concentrations of the enzyme sources (0.1 mg/ml recombinant ACSL1,
ACSM2B, or HLH) and incubation times for lauric acid and valeric acid were 10 min and 20
min, respectively. After incubation, the mixture was centrifuged at 20,380 g for 5 min, and a
1
0 µL aliquot of the supernatant was subjected to LC-MS/MS analysis using the same
8

systems described above. The column was a Zorbax SB-C18 (2.0 × 50 mm, 3.5 µm; Aglient
Technologies, Tokyo, Japan), the flow rate was 0.3 mL/min, and the column temperature was
4
0°C. Nitrogen was used as the nebulizer gas and the drying gas. The operating parameters
were optimized as follows: nebulizer gas flow, 3 L/min; drying gas flow, 15 L/min;
desolvation line temperature, 300°C; and heat block temperature, 40°C. The LC-MS/MS was
operated in positive electrospray ionization. The CoA conjugates were monitored by MRM.
The conditions of the mobile phase and m/z values for the detection and collision energies
(CE) are summarized in Table 2.
2
.8. Synthesis of the CoA conjugate of ibuprofen. To chemically synthesize the CoA
conjugate of ibuprofen (IBP CoA), ibuprofen N-hydroxysuccinimide (NHS) ester was first
synthesized. To a stirred mixture of ibuprofen (500 mg, 2.42 mmol) and NHS (335 mg, 2.91
mmol) in CH Cl (2.4 mL), EDCI·HCl (558 mg, 2.91 mmol) was added at room temperature.
2
2
After stirring for 1 hr, complete consumption of the ibuprofen was confirmed by thin-layer
chromatography. The formed ibuprofen NHS ester was purified with a SiO column (eluent:
2
CH Cl ) as a colorless solid (615 mg, 84%), and its structure was confirmed to be identical to
2
2
1
1
that reported previously by H NMR [20] [ H NMR (400 MHz, CDCl ) δ 7.29–7.22 (m, 2H),
3
7
1
.17–7.10 (m, 2H), 4.03 (q, J = 6.4 Hz, 1H), 2.81 (br s, 4H), 2.46 (d, J = 6.8 Hz, 2H), 1.92–
.77 (m, 1H), 1.63 (d, J = 6.8 Hz, 3H), 0.90 (d, J = 6.4 Hz, 6H)].
Next, to a stirred solution of CoA (10 mg, 0.026 mmol) in MeCN/0.5 M aqueous
NaHCO (1.3 mL/1.3 mL), ibuprofen NHS ester (40 mg, 0.26 mmol) was added at room
3
temperature under an Ar atmosphere. After stirring for 10 hr, the mixture was cooled to 0°C,
and an acidic ion exchange resin (DOWEX 50W8, hydrogen form, 500 mg) was added to the
mixture. After stirring for 30 min at 0°C, the resin and reaction solution were separated by
filtration. The compounds bound to the resin were extracted with water (1 mL, 3 times). The
extracts and the former filtrate were combined, and low polar impurities in the mixture were
removed by liquid-liquid extraction with AcOEt (10 mL, 3 times). The aqueous phase was
freeze-dried to give a colorless solid. The solid was suspended in AcOEt/MeCN (4 mL/4
9

mL). The suspension was centrifuged at 2,700 g for 5 min, and the supernatant was removed.
The washing process with AcOEt/MeCN was repeated 3 times. The remaining solid was dried
1
in vacuo to give IBP CoA as a colorless solid (10.0 mg). The product was characterized by H
1
NMR [ H NMR (400 MHz, D O) δ 8.70 (s, 1H), 8.43 (s, 1H), 7.27 (d, J = 8.0 Hz, 2H), 7.20
2
(d, J = 8.0 Hz, 2H), 6.23 (d, J = 5.6 Hz, 1H), 4.91–4.85 (m, 2H), 4.62 (s, 1H), 4.27 (s, 2H),
4
1
1
3
1
.04 (s, 1H), 4.03 (q, J = 6.8 Hz, 1H), 3.88 (dd, J = 8.4, 4.8 Hz, 1H), 3.60 (dd, J = 9.6, 4.8 Hz,
H), 3.38–3.26 (m, 4H), 3.10–2.90 (m, 2H), 2.45 (d, J = 7.2 Hz, 2H), 2.31–2.22 (m, 2H),
.88–1.72 (m, 1H), 1.48 (d, J = 7.6 Hz, 3H), 0.94 (s, 3H), 0.85 (d, J = 6.8 Hz, 6H), 0.81 (s,
H)], and the product purity was 83% (confirmed by qNMR using sodium 3-(trimethylsilyl)-
-propanesulfonate as an internal standard). Taking the purity into account, the product yield
was 33%.
2
.9. Protein adduct formation abilities of IBP CoA and IBP Glu. The protein adduct
formation ability of IBP CoA via trans-acylation and that of IBP Glu via glycation were
evaluated as follows: incubation mixtures (final volume of 0.2 ml) contained 10 pmol of IBP
CoA or IBP Glu, 100 mM Tris-HCl (pH 7.4), 1.0 mg/mL HLH as well as 3 mM ATP and 0.3
mM CoA (for IBP CoA) or 25 µg/mL alamethicin and 2.5 mM UDPGA (for IBP Glu). These
mixtures were incubated for 60 min at 37°C, and then the proteins were removed by the
addition of 10 µL of 60% perchloric acid and centrifugation at 20,380 g for 5 min. A 10 µL
aliquot of the supernatant was subjected to LC-MS/MS analysis. The analytical conditions for
the detection of IBP CoA and IBP Glu are shown in Table 2 and Table 6, respectively. The
mobile phase was (A) 0.1% formic acid and (B) acetonitrile, as shown in Table 6. Protein
adduct formation was calculated with the following equation:
Protein adduct formation (%) = 100 × (RA(HLH-) – RA(HLH+))/RA(HLH-)
where RA(HLH+) and RA(HLH-) are the remaining amounts of conjugate incubated with and
without HLH, respectively.
1
0

2
.10. Protein adduct formation abilities of propionic acid-class NSAIDs CoA conjugate
and acyl-glucuronide. The protein adduct formation ability of the CoA conjugate of
propionic acid-class NSAIDs via trans-acylation and that of acyl-glucuronide via glycation
were evaluated by not using their authentic standards as follows: For CoA conjugation,
incubation mixtures (final volume of 180 µL) contained 50 µM propionic acid-class NSAIDs,
1
00 mM Tris-HCl (pH 7.4), and 1.0 mg/mL HLM. After preincubation at 37°C for 2 min,
reactions were initiated by the addition of 20 µL of 3 mM CoA. After a 60 min incubation at
7°C, the reaction was terminated on ice, and then 100 µL of supernatant containing CoA
3
conjugate was collected after centrifugation at 105,000 g for 60 min. The collected
supernatant was mixed with 100 µL of 200 mM Tris-HCl (pH 7.4) and 10 µM triacsin C
(ACS inhibitor) with or without 2.0 mg/mL HLM to a final volume of 200 µL. After a 60 min
incubation at 37°C, the proteins were denatured by the addition of 10 µL perchloric acid. The
mixtures were centrifuged at 20,380 g for 5 min, and a 50 µL aliquot of the supernatant was
subjected to LC-MS/MS.
For glucuronidation, the incubation mixtures (final volume of 190 µL) contained 50 µM
propionic acid-class NSAIDs, 100 mM Tris-HCl (pH 7.4), 1.0 mg/mL HLM, and 25 µg/mL
alamethicin. After preincubation at 37°C for 2 min, the reactions were initiated by the
addition of 10 µL of 50 mM UDPGA. After a 60 min incubation at 37°C, the reaction was
terminated on ice, and then 100 µL of supernatant containing acyl-glucuronide was collected
after centrifugation at 105,000 g for 60 min. The collected supernatant was mixed with 100
µL of 200 mM Tris-HCl (pH 7.4) and 1 mM mefenamic acid (an inhibitor of UGT2B7 and
UGT1A9) with or without 2.0 mg/mL HLM to a final volume of 200 µL. After a 60 min
incubation at 37°C, the proteins were denatured by the addition of 10 µL perchloric acid. The
mixtures were centrifuged at 20,380 g for 5 min, and a 50 µL aliquot of the supernatant was
subjected to LC-MS/MS. Protein adduct formation was calculated with the following
equation:
Protein adduct formation (%) = 100 − (100 × PA(HLM+)/PA(HLM-)
)
1
1

where PA(HLM+) and PA(HLM-) are the peak areas of the conjugate in incubation with and
without HLM, respectively. It had been preliminarily confirmed that CoA conjugations or
glucuronidations of propionic acid-class NSAIDs were inhibited by over 90% by 5 µM
triacsin C or 500 µM mefenamic acid, respectively (data not shown).
2
.11. Statistical analysis. Data are expressed as the mean ± SD. Comparisons of two groups
were made with two-tailed Student’s t-tests. Correlation analyses were performed using
Pearson’s test. A value of Pꢀ < ꢀ0.05 was considered statistically significant.
1
2

3
3
. Results
.1. CoA conjugation of NSAIDs containing carboxylic acid. To investigate which
NSAIDs are conjugated with CoA in the human liver, the formation of CoA conjugates with
0 kinds of NSAIDs was evaluated using HLH as an enzyme source. As a result, the CoA
1
conjugates of alminoprofen, flurbiprofen, ibuprofen, ketoprofen, and loxoprofen, which are
propionic acid-class NSAIDs, were detected in positive mode (Fig. 2), whereas those of
diclofenac, indomethacin, mefenamic acid, salicylic acid, and sulindac were not detected by
SIM or Q1 scan in the range of the predicted m/z ± 100 in either positive or negative
electrospray mode (data not shown). Thus, it was demonstrated that CoA conjugates are
produced from propionic acid-class NSAIDs.
3
.2. Expression levels of 26 ACS mRNAs in human livers. To determine the expression
levels of 26 kinds of ACS mRNA in human livers, real-time RT-PCR was performed using
cDNA samples from 3 individual livers (Table 7). Among the 26 ACS isoforms, ACSM4,
ACSVL2, ACSF1, ACSF2, ACSBG1, and ACSBG2 could not be detected; the absolute
hepatic expression levels of the other 20 kinds of ACS isoforms are shown in Table 7. Among
them, ACSL1 revealed the highest expression (% in total hepatic ACS levels: 46.6%, 31.2–
5
6.1%), followed by ACSM2B (14.7%, 9.6–20.6%), ACSVL6 (8.8%, 3.1–17.0%), ACSL5
6.7%, 3.4–12.3%), and ACSVL1 (5.8%, 2.7–10.5%). Thus, ACSL1 and ACSM2B are the
major isoforms in the liver, occupying approximately 60% of the total.
(
3
.3. Confirmation of the protein levels and activities of recombinant ACSL1 and
ACSM2B. To investigate the possibility that propionic acid-class NSAIDs may be CoA
conjugated by ACSL1 and ACSM2B, the most abundant ACS isoforms in the human liver,
recombinant human ACSL1 and ACSM2B, were constructed. In the immunoblot analysis,
HLH was used as a positive control because the ACS enzymes are located in various
organelles, including the plasma membrane, endoplasmic reticulum, mitochondrial
1
3

membrane, and mitochondrial matrix [5]. By immunoblot analysis, a band at approximately
0 kDa was observed in recombinant ACSL1, and a band with the same mobility was
7
detected in HLH (Fig. 3A). Lauric acid CoA conjugation by recombinant ACSL1 was
approximately 2.5-fold higher than that by HLH (Fig. 3B), which can be explained by the
higher expression level of ACSL1 in expression systems than HLH (3.5-fold). Thus,
recombinant ACSL1 with a specific activity almost equivalent to ACSL1 expressed in the
human liver was established. Two bands were detected for the recombinant ACSM2B.
Similarly, two bands were detected in HLH, and the molecular weights of the common bands
between recombinant ACSM2B and HLH was 64 kDa, which corresponds to the estimated
molecular weight of ACSM2B (Fig. 3C). The intensity of the 64 kDa band in the recombinant
ACSM2B was 4.3-fold higher than that in the HLH. Valeric acid CoA conjugation by
recombinant ACSM2B was 1.3-fold higher than that in HLH (Fig. 3D). Taken together, we
judged that recombinant ACSL1 and ACSM2B were successfully constructed.
3
.4. Propionic acid-class NSAID CoA conjugation by recombinant ACSL1 or ACSM2B.
We investigated, by using the constructed recombinant enzymes, whether ACSL1 or
ACSM2B catalyzes CoA conjugation of propionic acid-class NSAIDs. As shown in Fig. 4,
recombinant ACSL1 catalyzed CoA conjugation of alminoprofen, flurbiprofen, ibuprofen,
ketoprofen, and loxoprofen at 3.5-, 2.0-, 1.7-, 3.8-, and 3.0-fold higher activities, respectively,
than HLH. The CoA conjugates of NSAIDs other than propionic acid-class NSAIDs after
exposure to the recombinant enzymes were not detected by SIM (data not shown), consistent
with the lack of production of CoA conjugates in HLH. These results suggested that the CoA
conjugations of propionic acid-class NSAIDs in the human liver are catalyzed by ACSL1.
3
.5. Correlation between ACSL1 protein levels and CoA conjugations for propionic
acid-class NSAIDs in 28 individual human liver samples. To confirm the responsibility of
ACSL1 for CoA conjugation of propionic acid-class NSAIDs in human liver, the correlations
between the ACSL1 protein levels and the CoA conjugations for propionic acid-class
1
4

NSAIDs were evaluated using a panel of 28 individual human liver samples (Fig. 5). There
were 11.9-fold interindividual variations in ACSL1 protein levels and 59.5-, 12.2-, 26.1-,
2
3.0-, and 8.8-fold variations in CoA conjugations for alminoprofen, flurbiprofen, ibuprofen,
ketoprofen, and loxoprofen, respectively. Significant positive correlations were observed
between ACSL1 protein levels and CoA conjugations for alminoprofen (R = 0.598, P <
0
0
.001), flurbiprofen (R = 0.623, P < 0.001), ibuprofen (R = 0.564, P < 0.01), ketoprofen (R =
.618, P < 0.001), and loxoprofen (R = 0.495, P < 0.01). These results supported the
proposition that ACSL1 is responsible for the CoA conjugations of propionic acid-class
NSAIDs in the human liver.
3
.6. Protein adduct formation abilities of the IBP CoA conjugate and glucuronide. The
protein adduct formation abilities of IBP CoA and IBP Glu were evaluated by quantification
of their unbound amounts in the mixture containing HLH. After the addition of 10 pmol IBP
CoA, the remaining amount of IBP CoA was 3.8 pmol in the presence of HLH, whereas it
was 4.9 pmol in the absence of HLH (Fig. 6A), which may be due to the instability of IBP
CoA in the buffer. Accordingly, the protein adduct formation of IBP CoA was calculated to
be 22% by equation (1). The remaining amount of IBP Glu was 4.0 pmol in the presence of
HLH, whereas it was 4.2 pmol in the absence of HLH (Fig. 6B). The protein adduct formation
of IBP Glu was calculated to be 5%. Thus, it was demonstrated that IBP CoA has a higher
protein adduct formation ability than IBP Glu.
3
.7. Protein adduct formation abilities of other propionic acid-class NSAID CoA
conjugates and glucuronides. CoA conjugates of propionic acid-class NSAIDs other than
IBP CoA are not commercially available. As an alternative method, we tried to evaluate the
protein adduct formation abilities of the CoA conjugate versus the acyl-glucuronide, which
were produced by an in vitro metabolism reaction. First, we evaluated the validity of this
strategy by examining the binding of IBP CoA and IBP Glu, which were produced by a
metabolic reaction with HLM. Protein adduct formations of IBP CoA and IBP Glu were
1
5

calculated according to equation (2) and were 28% or 5%, respectively (Fig. 7A), which were
close to those obtained using authentic standards of IBP CoA and IBP Glu (Fig. 6). Thus, we
judged that this strategy is reliable. Next, the protein adduct formation of CoA conjugates and
acyl-glucuronides of the other propionic-acid NSAIDs was evaluated. As shown in Figs. 7B-
E, the protein adduct formation of the CoA conjugates of propionic-acid NSAIDs were
significantly higher than that of acyl-glucuronides (21% vs −4% for alminoprofen, 45% vs
2
% for flurbiprofen, 25% vs −10% for ketoprofen, and 16% vs −15% for loxoprofen,
respectively). The negative values for the protein adduct formation observed for the acyl-
glucuronides of some NSAIDs is due to the error of the measurement by LC-MS/MS,
meaning no adduct formation. Collectively, it was demonstrated that the protein adduct
formation ability of the CoA conjugate of propionic acid-class NSAIDs was significantly
higher than that of their acyl-glucuronide form.
1
6

4
. Discussion
NSAIDs containing a carboxylic acid in their structure rarely cause adverse reactions such
as anaphylaxis, liver injury, Stevens-Johnson syndrome, gastrointestinal disorders, renal
dysfunction, or acute bronchospasm [21]. Some of these have been considered to be due to
the covalent binding of the parent drug or its metabolites to cellular proteins [12]. Among
their reactive metabolites, acyl-glucuronides have received attention for NSAID-induced
toxicity. Recently, it has been reported that the amount of ibuprofen bound to HLM proteins
was greater in the presence of CoA than in the presence of UDPGA, and the protein adduct
formation of IBP CoA to proteins was 116.3-fold higher than that of IBP Glu [13]. Thus, it
has been suggested that acyl-CoA conjugates, rather than acyl-glucuronides, may be
responsible for the toxicity.
To examine what kinds of NSAIDs are subjected to CoA conjugation in humans, ten
NSAIDs belonging to the anthranic acid, indole acetic acid, phenyl acetic acid, or propionic
acid classes were used (Fig. 1). We found that alminoprofen, flurbiprofen, ibuprofen,
loxoprofen, and ketoprofen, which are all propionic acid-class NSAIDs, were conjugated with
CoA in HLH (Fig. 2). Our results supported earlier study showing that propionic acid-class
NSAIDs including ibuprofen and ketoprofen are conjugated with CoA by HLH [22]. It has
been reported that a taurine conjugate of ibuprofen was detected in urine after oral intake of
ibuprofen in humans [23]. Because CoA conjugation is followed by amino acid conjugation,
CoA conjugation of ibuprofen would indeed occur in the human body. In addition, S-acyl
glutathione conjugate of ibuprofen was detected in rat or human hepatocytes [24]. It has been
reported that S-acyl glutathione conjugate of ibuprofen is formed from CoA-conjugate of
ibuprofen [24]. S-Acyl-glutathione conjugate was also formed from diclofenac [25], but in
this study, we found that the CoA conjugate was not formed from diclofenac. Thus, S-acyl
glutathione conjugates would not be necessarily formed from CoA conjugates.
To obtain a clue about the ACS isoform catalyzing CoA conjugation for propionic acid-
class NSAIDs, we evaluated the expression levels of the various ACS isoforms in the human
1
7

liver by real-time RT-PCR and found that ACSL1 had the highest expression in the human
liver (46.6%), followed by ACSM2B (14.7%) (Table 7). Although most isoforms could be
quantified, the expression of ACSM4, ACSVL2, ACSBG1, ACSBG2, ACSF1, and ACSF2
were not observed in the human liver. ACSVL2 is expressed in the heart and placenta;
ACSBG1 is expressed in the brain, adrenal gland, and testis; ACSBG2 is expressed in the
adult testis; ACSF1 is expressed in the kidney, heart, or brain; and ACSF2 is expressed in
adipose tissue [5]. Although information about the tissue distribution of human ACSM4 is
unavailable, it has been reported that ACSM4 is specifically expressed in the olfactory organ
in rats [4]. Thus, they appear to be extrahepatic isoforms.
To investigate whether ACSL1 and ACSM2B, abundant isoforms in the human liver, can
catalyze CoA-conjugation of drugs, their recombinant enzymes were constructed. By
immunoblot analysis and the measurement of CoA conjugation for lauric acid, we confirmed
that a functional recombinant ACSL1 was successfully constructed (Figs. 3A and B). The
expression of ACSM2B was observed using an antibody against ACSM2A, whose amino acid
sequence identity in the epitope is 97% (195 out of 201 amino acids) identical with that of
ACSM2B. Unexpectedly, two bands were detected in the expression system of ACSM2B
(Fig. 3C). To explain this, posttranslational modification was considered. A search for
glycosylation sites using the NetOGlyc 4.0 Server (Technical University of Denmark)
revealed three potential O-linked glycosylation sites in ACSM2B. However, treatment with
O-glycosidase did not change the mobility of the two bands in immunoblot analyses (data not
shown). Another possible reason was considered to be the usage of another start codon
because an ATG codon is located 51 bp downstream from the start codon according to the
mRNA sequence of human ACSM2B (Accession No. NM_182617.4). If this ATG is utilized
as a start codon, a protein lacking 17 amino acids at the N-terminus could be produced. Thus,
the band with high mobility in recombinant ACSM2B might be derived from this shortened
protein.
Two bands were also observed in HLH, and the molecular weight of the band with high
mobility was different from that with high mobility in the recombinant human ACSM2B (Fig.
1
8

3
C). According to the NCBI website (http://www.ncbi.nih.gov/gene/123876), a shorter
variant composed of 498 amino acids is also registered for human ACSM2A. This variant
lacks 79 amino acids at the N-terminus and can be recognized by the anti-ACSM2A antibody
used in this study. The band with high mobility in HLH may be derived from this shorter
variant of ACSM2A. The intensity of the band with low mobility in the recombinant
ACSM2B was 4-fold higher than that of the corresponding band in HLH, whereas the activity
for the formation of the CoA conjugate for valeric acid was almost the same between
recombinant ACSM2B and HLH (Fig. 3D). Because human ACSM2A shows the same
molecular weight as ACSM2B, the band with low mobility in HLH would be expected to
contain both ACSM2A and ACSM2B, which may cause a contradiction between band
intensity and activity. Since the constructed recombinant ACSM2B showed efficient CoA
conjugation for valeric acid, it was judged that a functional ACSM2B expression system had
been successfully developed.
Among the two recombinant ACS isoforms prepared in this study, human ACSL1 showed
efficient CoA conjugation for propionic acid-class NSAIDs (Fig. 4), and its responsibility was
demonstrated by a positive correlation between ACSL1 protein levels and CoA conjugations
for propionic acid-class NSAIDs in a panel of 28 human liver samples (Fig. 5). By
comparison of ACSL1 protein levels in HLH and recombinant system by immunoblot
analysis (Fig. 3), the contributions of ACSL1 to the CoA conjugations of alminoprofen,
flurbiprofen, ibuprofen, ketoprofen, and loxoprofen were calculated to be 99%, 57%, 50%,
1
09%, and 87%, respectively. Thus, ACSL1 would be a major contributor to the CoA
conjugation of three NSAIDs. It has been reported that CoA conjugate of ibuprofen is rapidly
epimerized and hydrolyzed to ibuprofen in HLH [22]. The moderate contribution of ACSL1
to the CoA conjugation of ibuprofen may be due to the difference in the efficiencies of
epimerization and hydrolysis between HLH and Sf21 cells. This theory may be applied to the
moderate contribution of ACSL1 to the CoA conjugation of flurbiprofen.
Possible mechanisms to cause the adverse reactions by NSAIDs include the accumulation
of arachidonic acid in cells, inhibition of mitochondrial function, and the production of
1
9

reactive metabolites [9,10]. As the former two mechanisms, the direct inhibition of ACS
enzymes or fatty acid oxidation enzymes in mitochondrion by NSAIDs, leading to apoptosis
or necrosis via ATP depletion, has been proposed [26,27]. As the latter mechanism, the
formation of CoA conjugates would be involved. As another mechanism, the excessive
consumption of cellular CoA by CoA conjugation reaction, which may cause microvesicular
steatosis of the liver [28,29], has been considered, because CoA is required for the formation
of acetyl-CoA in glycolysis [30] and in β-oxidation of fatty acids [31]. In addition, a decrease
in NADH, which is produced in the tricarboxylic acid cycle driven by acetyl-CoA, may cause
a decrease in the production of ATP, which is essential for cellular function [32]. The
interindividual variations of CoA conjugations for propionic acid-class NSAIDs were 59.5-
fold at the maximum (Fig. 5). Individuals with high ACSL1 activity may be susceptible to
adverse reactions, although a possibility could not be denied that other players having other
mechanisms for toxicity may also be involved in the susceptibility.
It has been suggested that liver injury is caused by reactive metabolites binding to
macromolecules such as proteins and DNA [33,34]; therefore, the protein adduct formation
ability of compounds has often been evaluated for toxicity relevance [13,35]. In this study,
prior to the evaluation of the adduct formation of the conjugates, the stabilities of IBP CoA
and IBP Glu in Tris-HCl buffer were assessed. The amount of IBP Glu was decreased by a 60
min incubation from 10 pmol to 4.2 pmol, even in the absence of any proteins (Fig. 6B). This
result is consistent with a previous report showing the instability of IBP Glu in phosphate
buffer [3]. The amount of IBP CoA was also decreased from 10 pmol to 4.9 pmol simply by
incubation (Fig. 6A); therefore, IBP CoA would also be unstable even in a buffer without any
proteins. Next, the adduct formation of IBP conjugates was evaluated by the quantification of
unbound conjugates. The protein adduct formation of IBP CoA (22%) was higher than that of
IBP Glu (2%) (Fig. 6). In addition, we evaluated the protein adduct formation of the
conjugates by an alternative evaluation method with no use of authentic standards, resulting
in the protein adduct formation of IBP CoA (28%) and IBP Glu (5%), in which the tendency
was similar to that obtained in analyses using authentic standards (Figs. 6 and 7A). Thus, we
2
0

judged that this evaluation method was reliable. Darnell et al. [13] reported, by using
radioactive ibuprofen, that the binding rates of IBP CoA or IBP Glu produced by metabolic
reactions with microsomal proteins were 15% or 0.03%, respectively, which were lower than
those obtained in this study. However, the trend of higher adduct formation of IBP CoA than
IBP Glu was consistent. For the other propionic acid-class NSAIDs, the evaluation with
conjugates produced by the metabolic reaction demonstrated that the covalent binding of the
CoA conjugates of alminoprofen, flurbiprofen, ketoprofen, and loxoprofen was significantly
higher than that of acyl-glucuronides (Figs. 7B, C, D and E). From these results, it is
conceivable that CoA conjugates, rather than acyl-glucuronides, might be relevant to the
toxicity of propionic acid-class NSAIDs.
According to LiverTox database (https//www.ncbi.nim.nih.gov/books/ NBK547852/), the
likelihood score of liver injury of ibuprofen (A: well known but rare cause of clinically
apparent liver injury) is higher rather than that of flurbiprofen and ketoprofen (C: probable
rare cause of clinically apparent liver injury). This is unlikely corresponding to the protein
adduct formation in Fig. 7. Recommended dosage of ibuprofen, 400–800 mg orally 3 to 4
times daily, is much higher than those of flurbiprofen (50–100 mg 2 to 4 times daily) and
ketoprofen (50–75 mg 3 or 4 times daily). This fact supports a theory that daily dose would
also be a key factor for the risk of adverse reactions [35,36].
In conclusion, we revealed that ACSL1, the most highly expressed isoform among the ACS
family in the human liver, is responsible for the CoA conjugation of propionic acid-class
NSAIDs. It was demonstrated that the protein adduct formation of CoA conjugates via trans-
acylation was higher than that of acyl-glucuronides. It would be interesting to directly
investigate the involvement of ACSL1 in the incidence of adverse reactions induced by
propionic acid-class NSAIDs containing carboxylic acid.
2
1

Conflict of interest
The authors declare no conflict of interest.
Abbreviations
ACS, Acyl-CoA synthetase; ACSBG, ACS bubblegum family; ACSL, ACS long-chain
family; ACSM, ACS medium-chain family; ACSS, ACS short-chain family; ACSVL, ACS
very long-chain family; CE, Collision energy; HLH, Human liver homogenate; HLM, Human
liver microsomes; IBP CoA, CoA conjugate of ibuprofen; IBP Glu, ibuprofen acyl-
glucuronide; MRM, Multiple reaction monitoring; NHS, N-Hydroxysuccinimide; NSAID,
Nonsteroidal anti-inflammatory drug; SIM, Single ion monitoring; SJS, Stevens-Johnson
syndrome; UDPGA, UDP-glucuronic acid; UGT, UDP-glucuronosyltransferase
2
2

References
[
[
[
1] T. Lee, Y.S. Lee, S.Y. Yoon, S. Kim, Y.J. Bae, H.S. Kwon, Y.S. Cho, H.B. Moon, T.B.
Kim, Characteristics of liver injury in drug-induced systemic hypersensitivity reactions,
J. Am. Acad. Dermatol. 69 (2013) 407-415.
2] J. Olsen, I. Bjørnsdottir, J. Tjornelund, H.S. Honoré, Chemical reactivity of the naproxen
acyl glucuronide and the naproxen coenzyme A thioester towards bionucleophiles, J.
Pharm. Biomed. Anal. 29 (2002) 7-15.
3] R. Sawamura, N. Okudaira, K. Watanabe, T. Murai, Y. Kobayashi, M. Tachibana, T.
Ohnuki, K. Masuda, H. Honma, A. Kurihara, O. Okazaki, Predictability of idiosyncratic
drug toxicity risk for carboxylic acid-containing drugs based on the chemical stability of
acyl glucuronide, Drug Metab. Dispos. 38 (2010) 1857-1864.
[
4] P.A. Watkins, D. Maiguel, Z. Jia, J. Pevsner, Evidence for 26 distinct acyl-coenzyme A
synthetase genes in the human genome, J. Lipid Res. 48 (2007) 2736-2750.
5] G. Pérez-Chacón, A.M. Astudillo, D. Balgoma, M.A. Balboa, J. Balsinde, Control of
free arachidonic acid levels by phospholipases A2 and lysophospholipid
acyltransferases, Biochim. Biophys. Acta 1791 (2009) 1103-1113.
[
[
6] S. Yan, X.F. Yang, H.L. Liu, N. Fu, Y. Ouyang, K. Qing, Long-chain acyl-CoA
synthetase in fatty acid metabolism involved in liver and other diseases: an update,
World J. Gastroenterol. 21 (2015) 3492-3498.
[
7] E. Soupene, F.A. Kuypers, Mammalian long-chain acyl-CoA synthetases, Exp. Biol.
Med. 233 (2008) 507-521.
[
8] A.T. Harder, Y.H. An, The mechanisms of the inhibitory effects of nonsteroidal anti-
inflammatory drugs on bone healing: A concise review, J. Clin. Pharmacol. 43 (2003)
8
07-815.
[
9] T. Mahmud, S. Rafi, D.L. Scott, J.M. Wrigglesworth, I. Bjarnason, Nonsteroidal
antiinflammatory drugs and uncoupling of mitochondrial oxidative phosphorylation,
Arthritis Rheum. 39 (1996) 1998-2003.
2
3

[10] D. Wu, A.I. Cederbaum, Sodium salicylate increases CYP2E1 levels and enhances
arachidonic acid toxicity in HepG2 cells and cultured rat hepatocytes, Mol. Pharmacol.
5
9 (2001) 795-805.
[
[
[
11] U.A. Boelsterli, V. Ramirez-Alcantara, NSAID acyl glucuronides and enteropathy, Curr.
Drug Metab. 12 (2011) 245-52.
12] M.J. Bailey, R.G. Dickinson, Acyl glucuronide reactivity in perspective: Biological
consequences, Chem. Biol. Interact. 145 (2003) 117-137.
13] M. Darnell, K. Breitholtz, E.M. Isin, U. Jurva, L. Weidolf, Significantly different
covalent binding of oxidative metabolites, acyl glucuronides, and S-acyl CoA conjugates
formed from xenobiotic carboxylic acids in human liver microsomes, Chem. Res.
Toxicol. 28 (2015) 886-896.
[
14] Y. Kobayashi, T. Fukami, A. Nakajima, A. Watanabe, M. Nakajima, T. Yokoi, Species
Differences in Tissue Distribution and Enzyme Activities of Arylacetamide Deacetylase
in Human, Rat, and Mouse, Drug Metab. Dispos. 40 (2012) 671-679.
15] T. Izukawa, M. Nakajima, R. Fujiwara, H. Yamanaka, T. Fukami, M. Takamiya, Y.
Aoki, S. Ikushiro, T. Sakaki, T. Yokoi, Quantitative analysis of UDP-
glucuronosyltransferase (UGT) 1A and UGT2B expression levels in human livers, Drug
Metab. Dispos. 37 (2009) 1759-1768.
[
[
[
[
[
16] T. Yoshida, T. Fukami, T. Kurokawa, S. Gotoh, A. Oda, M. Nakajima, Difference in
substrate specificity of carboxylesterase and arylacetamide deacetylase between dogs
and humans, Eur. J. Pharm. Sci. 111 (2018) 167-176.
17] M.M. Bradford, A rapid and sensitive method for the quantifiation of microgram
quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72
(1976) 248-254.
18] A. Watanabe, T. Fukami, M. Nakajima, M. Takamiya, Y. Aoki, T. Yokoi, Human
arylacetamide deacetylase is a principal enzyme in flutamide hydrolysis, Drug Metab.
Dispos. 37 (2009) 1513-1520.
19] Y. Hisanaga, H. Ago, N. Nakagawa, K. Hamada, K. Ida, M. Yamamoto, T. Hori, Y. Arii,
2
4

M. Sugahara, S. Kuramitsu, S. Yokoyama, M. Miyano, Structural basis of the substrate-
specific teo-step catalysis of long chain fatty acyl-CoA synthetase dimer, J. Biol. Chem.
2
79 (2004) 31717-31726.
[20] D. Wang, N. Zhu, P. Chen, Z. Lin, G. Liu, Enantioselective decarboxylative cyanation
employing cooperative photoredox catalysis and copper catalysis, J. Am. Chem. Soc.
1
39 (2017) 15632-15635.
[
[
[
21] D.A. Henry, Side-effects of non-steroidal anti-inflammatory drugs, Baillieres Clin.
Rheumatol. 2 (1988) 425-454.
22] S.D. Hall, X. Quen, The role of coenzyme A in the biotransformation of 2-arylpropionic
acids, Chem. Biol. Interact. 90 (1994) 235-251.
23] M.A. Shirley, X. Guan, D.G. Kaiser, G.W. Halstead, T.A. Baillie, Taurine conjugation
of ibuprofen in humans and in rat liver in vitro. Relationship to metabolic chiral
inversion, J. Pharmacol. Exp. Ther. 269 (1994) 1166-1175.
[
[
[
24] M.P. Grillo, F. Hua, Enantioselective formation of ibuprofen-S-acyl-glutathione in vitro
in incubations of ibuprofen with rat hepatocytes, Chem. Res. Toxicol. 21 (2008) 1749-
1
759.
25] M.P. Grillo, F. Hua, C.G. Knutson, J.A. Ware, C. Li, 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 (2003) 1410-1417.
26] F. Kasuya, M. Kazuhiro, H. Tatsuya, K. Nakamoto, S. Tokuyama, T. Masuyama, Effect
of the non-steroidal anti-inflammatory drugs on the acyl-CoA synthetase activity toward
medium-chain, long-chain and polyunsaturated fatty acids in mitochondria of mouse
liver and brain, J. Enzyme Inhib. Med. Chem. 28 (2013) 223-227.
[
27] M. Darnell, L. Weidolf, Metabolism of xenobiotic carboxylic acids: focus on coenzyme
A conjugation, reactivity, and interference with lipid metabolism, Chem. Res. Toxicol.
2
6 (2013) 1139-1155.
[
28] J.F. Bravo, M.P. Jacobson, B.F. Mertens, Fatty liver and pleural effusion with ibuprofen
therapy, Ann. Intern. Med. 87 (1977) 200-201.
2
5

[29] J.P. Dutertre, F. Bastides, A.P. Jonville, A. De Muret, A. Sonneville, D. Larrey, E.
Autret, Microvesicular steatosis after ketoprofen administration, Eur. J. Gastrol. Hepatol.
3
(1991) 953-954.
[
[
[
30] L.J. Ruff, E.P. Brass, Metabolic effects of pivalate in isolated rat hepatocytes, Toxicol.
Appl. Pharmacol. 110 (1991) 295-302.
31] F. Bernard, P. Dominique, Inhibition of mitochondrial beta-oxidation as a mechanism of
hepatotoxicity, Pharmacol. Ther. 67 (1995) 101-154.
32] G.A. Mitchell, N. Gauthier, A. Lesimple, S.P. Wang, O. Mamer, I. Qureshi, Hereditary
and acquired diseases of acyl-coenzyme A metabolism, Mol. Genet. Metab. 94 (2008) 4-
1
5.
[
33] J.D. Miner, T.P. Kissinger, Evidence for the involvement of N-acetyl-p-quinoneimine in
acetaminophen metabolism, Biochem. Pharmacol. 28 (1979) 3285-3290.
[
34] G.K. Poon, Q. Chen, Y. Teffera, J.S. Ngui, P.R. Griffin, M.P. Braun, G.A. Doss, C.
Freeden, R.A. Stearns, D.C. Evans, T.A. Baillie, W. Tang, Bioactivation of diclofenac
via benzoquinone imine intermediates - Identification of urinary mercapturic acid
derivatives in rats and humans, Drug Metab. Dispos. 29 (2001) 1608-1613.
35] T. Usui, M. Mise, T. Hashizume, M. Yabuki, S Komuro, Evaluation of the potential for
drug-induced liver injury based on in vitro covalent binding to human liver proteins,
Drug Metab. Dispos. 37 (2009) 2383-2392.
[
[
36] J.N. Bauman, J.M. Kelly, S. Tripathy, S.X. Zhao, W.W. Lam, A.S. Kalgutkar, R.S.
Obach, Can in vitro metabolism-dependent covalent binding data distinguish hepatotoxic
from nonhepatotoxic drugs? An analysis using human hepatocytes and liver S-9 fraction,
2
2 (2009) 332-340.
2
6

Figure Legends
Fig. 1. Chemical structures of 10 NSAIDs containing carboxylic acid investigated in this
study. NSAIDs are classified based on their chemical structure.
Fig. 2. LC-MS chromatograms for CoA conjugations for 5 kinds of NSAIDs containing
carboxylic acid in human liver homogenate. HLH was incubated with NSAIDs in the
absence (A, C, E, G, I) or presence (B, D, F, H, J) of CoA, and each conjugate was monitored
by SIM.
Fig. 3. Immunoblot analyses of recombinant human ACSL1 and ACSM2B and
evaluation of their activities for fatty acid conjugations. ACSL1 (A) or ACSM2B (C)
expression was evaluated by Immunoblot analyses. HLH [10 µg (A) or 50 µg (C)],
homogenates of Sf21 cells expressing human ACSL1 (2.5 µg) or ACSM2B (15 µg), and
mock homogenate (15 µg) were separated by electrophoresis using 7.5% SDS-polyacrylamide
gel. CoA conjugations for lauric acid (B) and valeric acid (D) in HLH and recombinant
ACSL1 or ACSM2B were evaluated by LC-MS/MS. Enzyme sources (0.1 mg/mL) were
incubated with 30 µM lauric acid (B) or 100 µM valeric acid (D). Each column represents the
mean ± SD (n = 3).
Fig. 4. CoA conjugations for propionic acid-class NSAIDs by HLH or recombinant
ACSL1 and ACSM2B. CoA conjugations for alminoprofen (A), flurbiprofen (B), ibuprofen
(C), ketoprofen (D), and loxoprofen (E) by HLH or recombinant ACSL1 and ACSM2B were
evaluated at 50 µM substrate concentrations by LC-MS/MS. Each column represents the
mean ± SD (n = 3). ND: Not detected.
Fig. 5. Correlation between ACSL1 protein levels and CoA conjugations for propionic
acid-class NSAIDs in a panel of 28 human livers. Correlation analyses between ACSL1
2
7

protein levels and CoA conjugations of alminoprofen (A), flurbiprofen (B), ibuprofen (C),
ketoprofen (D), or loxoprofen (E) were performed. Pearson’s rank correlation method was
used. ACSL1 protein levels are relative to the sample with the lowest ACSL1 expression.
Each point represents the mean of triplicate determinations.
Fig. 6. Protein adduct formation of CoA conjugate and acyl-glucuronide of ibuprofen.
Ten pmol of IBP CoA (A) or IBP Glu (B) was incubated for 60 min at 37˚C in 100 mM Tris-
HCl (pH 7.4), 1.0 mg/mL HLH as well as 3 mM ATP and 0.3 mM CoA (A) or 25 µg/mL
alamethicin and 2.5 mM UDPGA (B). Proteins were removed by the addition of 60%
perchloric acid, and subsequent centrifugation at 20,380g for 5 min. The supernatants were
subjected to LC-MS/MS. Each column represents the mean ± SD of triplicate determinations.
Fig. 7. Protein adduct formation of CoA conjugate and acyl-glucuronide of propionic
acid-class NSAIDs. Protein adduct formations of CoA conjugate and acyl-glucuronide of
ibuprofen (A), alminoprofen (B), flurbiprofen (C), ketoprofen (D), or loxoprofen (E) were
evaluated via their productions by in vitro metabolic reaction. Each column represents the
mean ± SD of triplicate determinations. ***P < 0.001.
2
8

Table 1. Characterization of 28 donors of livers.
Age
ꢀ
No
Sex
Ethnicity
Cause of death
(yr)
1
2
3
4
5
6
7
8
9
F
F
33
47
34
63
32
56
33
52
53
60
47
36
54
62
36
47
46
35
68
16
51
52
59
33
49
41
38
35
Hispanic
Caucasian
Black
Cerebrovascular accident
Intracerebral hematoma
Intracerebral hematoma
Intracerebral hematoma
Subarachnoid hemorrhage
Intracerebral hematoma
Primary brain tumor
Cerebrovascular accident
Intracerebral hematoma
Cerebral apoplexy
F
M
F
Caucasian
Hispanic
Caucasian
Caucasian
Caucasian
Caucasian
Hispanic
Caucasian
Caucasian
Caucasian
Asian
M
F
M
M
M
F
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
S/P code
M
M
M
M
M
M
F
Intracerebral hemorrhage
Cerebrovascular accident
Intracerebral hematoma
Head gun wound
Caucasian
Caucasian
Caucasian
Hispanic
Caucasian
Caucasian
Caucasian
Asian
Aneurysm
Hypoxia
Intracerebral hemorrhage
Cerebrovascular accident
Traffic accident
M
M
M
F
Cerebrovascular accident
Cerebrovascular accident
Traffic accident
F
Caucasian
Caucasian
Hispanic
Caucasian
Caucasian
Hispanic
M
M
F
Traffic accident
Head injury
Intracerebral hemorrhage
Head injury
M
F
Cerebral apoplexy
2
9