Metabolic activation of deferiprone mediated by CYP2A6

Article abstract of DOI:10.1080/00498254.2021.1931729

Deferiprone (DFP) is a metal chelating agent generally used to treat patients with thalassaemia, due to iron overload in clinical settings. Studies have revealed that long-term use of DFP can induce hepatotoxicity, however, mechanisms of its toxic action remain unclear. The present studies are aimed to characterize the reactive metabolite of DFP, to define the metabolic pathway, and to determine the P450 enzymes participating in the bioactivation. A demethylation metabolite (M1) was observed in rat liver microsomal incubations. Additionally, a glutathione (GSH) conjugate (M2) and an N-acetylcysteine (NAC) conjugate (M3) were detected in microsomal incubations fortified with DFP and GSH/NAC. Biliary M2 and urinary M3 were respectively found in animals administered DFP. CYP2A6 enzyme dominated the catalysis to bioactivate DFP.

Full text of DOI:10.1080/00498254.2021.1931729

Xenobiotica
the fate of foreign compounds in biological systems
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ixen20
Metabolic Activation of Deferiprone Mediated by
CYP2A6
Xiaojiao Zheng, Xu Wang, Zifang Ding, Wei Li, Ying Peng & Jiang Zheng
To cite this article: Xiaojiao Zheng, Xu Wang, Zifang Ding, Wei Li, Ying Peng & Jiang
Zheng (2021): Metabolic Activation of Deferiprone Mediated by CYP2A6, Xenobiotica, DOI:
10.1080/00498254.2021.1931729
To link to this article: https://doi.org/10.1080/00498254.2021.1931729
Accepted author version posted online: 18
May 2021.
Submit your article to this journal
Article views: 33
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=ixen20

Title page
Metabolic Activation of Deferiprone Mediated by CYP2A6
Xiaojiao Zheng,1 Xu Wang,1 Zifang Ding,1 Wei Li,1 Ying Peng,1* and Jiang Zheng1,2,3*
1Wuya College of Innovation, Shenyang Pharmaceutical University, Shenyang, Liaoning
110016, P. R. China
2State Key Laboratory of Functions and Applications of Medicinal Plants,
Guizhou Provincial Key Laboratory of Pharmaceutics, Guizhou Medical University,
Guiyang, Guizhou 550025, P. R. China
3Key Laboratory of Environmental Pollution, Monitoring and Disease Control, Ministry of
Education, Guizhou Medical University, Guiyang, 550025, P. R. China
Corresponding Authors
Jiang Zheng, PhD
Wuya College of Innovation, Shenyang Pharmaceutical University, Shenyang, Liaoning
110016, P. R. China; State Key Laboratory of Functions and Applications of Medicinal
Plants, Key Laboratory of Pharmaceutics of Guizhou Province, Guizhou Medical
University, Guiyang, Guizhou 550025, P. R. China
E-mail: zhengneu@yahoo.com
Tel: +86-24-23984215
Fax: +86-24-23986510
Ying Peng, PhD
Wuya College of Innovation, Shenyang Pharmaceutical University, Shenyang, Liaoning
110016, P. R. China
E-mail: yingpeng1999@163.com
Tel: +86-24-23986361
Fax: +86-24-23986510

Abstract
1. Deferiprone (DFP) is a metal chelating agent generally used to treat patients with
thalassemia, due to iron overload in clinical settings.
2. Studies have revealed that long-term use of DFP can induce hepatotoxicity, however,
mechanisms of its toxic action remain unclear. The present studies are aimed to
characterize the reactive metabolite of DFP, to define the metabolic pathway, and to
determine the P450 enzymes participating in the bioactivation.
3. A demethylation metabolite (M1) was observed in rat liver microsomal incubations.
Additionally, a glutathione (GSH) conjugate (M2) and an N-acetylcysteine (NAC)
conjugate (M3) were detected in microsomal incubations fortified with DFP and
GSH/NAC.
4. Biliary M2 and urinary M3 were respectively found in animals administered DFP.
5. CYP2A6 enzyme dominated the catalysis to bioactivate DFP.
Keywords: Deferiprone, Metabolic Activation, CYP2A6

Introduction
Deferiprone (DFP, 3-hydroxy-1,2-dimethyl-4(1H)-pyridone) is a metal chelating
agent (Huang et al. 2006). In 2011, the Food and Drug Agent of the USA approved DFP
for the treatment of thalassemia with poor response to existing chelation therapy and
excessive iron load, due to blood transfusion (Belmont et al. 2017; Chuansumrit et al.
2016). Nowadays, it is the only metal chelating agent that can be taken orally with a
strong chelating effect on iron (Kontoghiorghe et al. 2016). Previous studies showed that
DFP can be rapidly absorbed from the upper gastrointestinal tract with Tmax at 45 min,
rapidly metabolized and excreted by the liver and kidney (Fredenburg et al. 1996).
However, long-term use of DFP has been shown to cause liver injury with an abnormal
increase of serum alanine (Cohen et al. 2003; Maggio et al. 2002). Unfortunately, the
mechanisms of DFP induced hepatotoxicity are not understood.
Metabolic activation is one of the mechanisms of drug-induced liver injury (Chi et al.
2016). Those drugs execute toxicities after metabolic transformation. A number of drug
metabolizing enzymes have been found to participate in the bioactivation and toxicities of
the drugs (Yao et al. 2016). Cytochrome P450 enzymes are involved in such undesired
toxic effects (Lee et al. 2003; Venkatakrishnan et al. 2001). Specifically, the enzymes can
catalyze the oxidation via addition of oxygen and dehydrogenation. Oxidative N-, O-, and
S-dealkylations are the metabolic pathways catalyzed mainly by P450 enzymes.
Structurally, DFP has a frame of pyridinone, along with a methyl group attached on the
nitrogen. We speculated that N-demethylation catalyzed by P450 enzymes would take
place with DFP, which may produce a quinone-like reactive metabolite. Quinones are

highly reactive electrophiles that can react with nucleophilic group occurring in
biomolecules, i.e. DNA and protein (Bolton et al. 2000; Graham et al. 1978). The present
study aimed to characterize metabolite(s) of DFP in vitro and in vivo, determine metabolic
activation pathway(s) of DFP, and identify cytochromes P450 enzymes participating in the
production of reactive metabolite(s) of DFP.
Materials and methods
Chemicals and materials. Deferiprone and maltol (purity > 98%) were obtained from
Aladdin Reagent Co., Ltd. (Shanghai, China). Propranolol, glutathione (GSH),
nicotinamide adenine dinucleotide phosphate (NADPH), N-acetyl-L-cysteine (NAC) and
methoxsalen were acquired from Sigma-Aldrich Co. (St. Louis, MO). Solid phase
extraction columns (CleanertTM PAX) were obtained from Agela Technologies Co., Ltd.
(Tianjin, China). Rat (Sprague-Dawley, male) liver microsomes were prepared in our
laboratory according to our published procedure (Ma et al. 2014). Recombinant human
P450 enzymes and human liver microsomes were supplied by BD Gentest (Woburn, MA).
All organic solvents were purchased from Fisher Scientific (Springfield, NJ). All of the
reagents and solvents were of analytical or HPLC grade.
Microsomal incubations. DFP was dissolved in PBS buffer to prepare the stock
solution. A microsomal incubations mixture was prepared by mixing DFP (100 μM) or M1
(100 μM) with rat or human hepatic microsomes (1.0 mg protein/mL) supplemented with
GSH or NAC (20 mM) in PBS buffer (pH 7.4, final volume of 250 μL). The concentration
was selected according to a study where the fasting serum concentration of DFP can reach

126 μmol/L at a dose of 25 mg/kg (Bellanti et al. 2014). Reactions were launched by
adding NADPH (1.0 mM), and the incubations were carried out at 37 °C for 45 min. The
reactions were terminated by the same volume of ice-cold acetonitrile, vortexed for 3 min,
and centrifuged at 19,000 g for 10 min. The resultant supernatants (5 μL) were analyzed
by an LC-MS/MS system. Control incubations lacked NADPH, NAC or GSH. All
incubation experiments were carried out in duplicate.
Chemical synthesis of M1. Synthetic M1 was prepared (Scheme 1) as described by
Dong and co-workers (Dong et al. 2018). Briefly, maltol (0.756 g, 6 mmol) was mixed
with aqueous ammonia (5 mL, 20%) in a round bottom flask. The resultant mixture was
slowly stirred at 45 °C for 2 h and then mixed dropwise with carbonic acid (1.73 g, 18
mmol) dissolved in 5 mL water within 5 h with stirring and refluxed for 5 h. Thin layer
chromatography was used to monitor the reaction. The resulting mixture was cooled to
0 °C, filtered, rinsed with water, and dried to offer crude products. The crude product was
submitted to a semipreparative HPLC system for purification, followed by
characterization by NMR and mass spectrometry.
Synthesis of M2 and M3. A mixture of M1 (0.625 g, 5 mmol) and Ag2O (3 g, 6
mmol) was stirred in methanol (40 mL) at 45 °C for 2 h (Li et al. 1999). The resulting
solid was removed by filtration and the filtrates were condensed by rotary evaporation.
The crude product was recrystallized from ether to give bright yellow crystals. GSH
(1.842 g, 6 mmol) or NAC (0.978 g, 6 mmol) was dissolved in CH2Cl2 (75 mL) containing
synthetic M1 (0.492 g, 4 mmol). The reaction mixture was stirred for 8 h at room

temperature and condensed by rotary evaporation and submitted to the semipreparative
HPLC system for purification. The purified product was characterized by NMR and mass
spectrometry.
Animal experiments. Sprague-Dawley rats (male, 200 ± 10 g) were obtained from
the Animal Center of Shenyang Pharmaceutical University. All animal raising and
experiments were conducted by following the Animal Experimental Regulations of Ethics
Review Committee of Shenyang Pharmaceutical University (license number:
SYPU-IACUC-C2019-9-2-204). Animals were kept in a controlled environment with a 12
h cycle of light and darkness and fed with water and food freely. Rats were randomly
divided into two groups (three for each). Animals were fasted for 12 h before the
experiment and anesthetized by administering (i.p.) urethane (20%) at a dose of 1.35 g/kg.
We used at least three different batches of rats. Bile ducts were cannulated, and blank bile
was harvested. Subsequently the rats were administered (i.p.) DFP (140 mg/kg) dissolved
in saline, according to early literature (Gong et al. 2010), and 0-4 h bile samples were
harvested. The other three animals were given the same dose of DFP and placed in
metabolism cages, and 0-24 h urine samples were harvested individually. Blank urine
samples were collected from these animals before DFP administration.
Sample preparation. The resulting urine or bile samples (300 µL) were mixed with
acetonitrile (900 µL), vortexed, and centrifuged. The supernatants were condensed by
blowing with nitrogen flow and cleaned by solid phase extraction column. Briefly, SPE
columns were washed with methanol (10 mL) and then glacial acetic acid-water solution

(10 mL, pH=3.5). The samples were acidified with acetic acid to pH 3 and loaded onto the
column. The resulting column was eluted with a gradient of methanol-water (pH=3.5)
solution at concentrations of 0, 20, 40, 60, and 80% (10 mL for each). The target eluates
were pooled and concentrated by blowing with N2 flow. The resulting concentrates were
dissolved in 100 μL of 5% acetonitrile in water, followed by 10 min centrifugation at
19,000 g. The resultant supernatants (5 µL) were subjected to LC-MS/MS for analysis.
Recombinant human P450 enzyme incubations. To identify the specific P450
enzymes participating in the production of reactive metabolites of DFP, nine human
recombinant P450s, i.e. CYPs3A5, 3A4, 2E1, 2D6, 2C19, 2C9, 2B6, 2A6, and 1A2, were
tested. Individual human recombinant P450 enzymes (100 nmol enzyme with a total
volume of 100 µL in each incubation) were incubated with DFP and NAC under similar
condition for microsomal incubations. The resulting reactions were quenched by ice-cold
acetonitrile (100 µL) spiked with propranolol (5.0 ng/mL) as the internal standard.
Produced metabolites of DFP were assessed using an LC-MS/MS system. Each incubation
was carried out in triplicate.
Microsomal inhibition study. The incubation system contained DFP (100 μM),
MgCl2 (3.2 mM), methoxsalen (10 μM), NAC (20 mM), PBS (100 mM, pH 7.4), rat or
human hepatic microsomes (1.0 mg protein/mL), and NADPH (1.0 mM). The control
reaction mixture lacked methoxsalen. The resulting reactions were quenched by the same
volume of ice-cold acetonitrile spiked with propranolol (5.0 ng/mL) as the internal
standard.

Cytotoxicity Evaluation. MTT assay was performed in cultured rat primary
hepatocytes to evaluate the cytotoxicity of DFP. Cells were isolated according to a
modified two-step collagenase perfusion method (Shen et al. 2012). After dissociating
from anesthetized adult rats, hepatocytes were seeded in 96-wells plates at a density of 1.8
× 104 cells/well and incubated for nearly 8 h for recovering and adhering. The resulting
cells were treated with DFP at concentrations of 0, 5, 10, 25, 50, 100 and 200 μM. In
control groups, cells received PBS buffer. The resultant cells were incubated at 37 °C in a
humidified incubator with 5% CO2 for 12 h, followed by addition of 10 μL of MTT
solution (5.0 mg/mL). The media were discarded 4 h later, and DMSO (100 μL) was
added to dissolve the formazan crystals. Cell viability was determined by measuring the
absorbance at 562 nm. The cytotoxicity tests were carried out in triplicate.
LC-MS/MS methods. Metabolite analysis and assessment, along with internal
standard quantification, were conducted on an AB SCIEX Instruments 4000 Q-Trap
(Applied Biosystems, Foster City, CA) equipped with an Agilent 1260 infinity HPLC
system (Agilent Technologies, Biblingen, Germany). Analytes were separated on an
Accuore C18 column (4.6×250 mm, 5 μm, Dikma Technological). The column was eluted
with a mobile system composed of acetonitrile with 0.1 % formic acid (A) and 0.1 %
formic acid in water (B), using a gradient elution program as follows: 5 % A at 0-2 min;
5-30 % A at 2-4 min; 30-90 % A at 4-10 min; 90 % A at 10-11 min; 90-10 % A at 11-12
min; 5 % A at 12-14 min. The flow rate was set at 0.5 mL/min, and the column
temperature was kept at 25 °C. Mass spectrometric parameters were set as follows: ion
spray voltage: 5500 V; curtain gas 1: 20 psi; curtain gas 2: 20 psi; temperature of the turbo

ion spray: 600 °C; entrance potential: 10 V; and cell exit potential: 3 V. Optimal response
for the analysis in positive ion mode was achieved by multiple-reaction monitoring (MRM)
scanning. Characteristic ion pairs (declustering potential (DP), collision energy (CE) )
were m/z 140→96 (95, 35) for DFP, m/z 126→108 (85, 35) for M1, m/z 431→158 (100,
40) for DFP-GSH conjugate, m/z 287→158 (100, 40) for DFP-NAC conjugate, and m/z
260→116 (70, 30) for propranolol separately. Information-dependent acquisition (IDA)
was used to initiate enhanced product ion (EPI) scanning by analyzing MRM signals. EPI
scanning was run in positive ion mode for product ion ranging from m/z 100 to 900. IDA
was set to select ions more than 2000 cps with the exclusion of former target ions after
three occurrences for 10 s and collision energy was set at 35 eV with a spread of 15 eV.
Additionally, an AB SCIEX Instruments 5500 triple quadrupole mass spectrometry
(Applied Biosystems, Foster City, CA) equipped with an Agilent 1260 infinity LC system
(Agilent Technologies, Biblingen, Germany) was also employed to analyze DFP and its
metabolites with the same parameters as those mentioned above. The AB SCIEX
Instruments 4000 triple quadrupole mass spectrometry equipped with Q-trap was used to
acquire tandem mass spectra of analytes. The AB SCIEX Instruments 5500 triple
quadrupole mass spectrometry was employed for quantitative analysis, due to higher
sensitivity. All data obtained were processed by AB SCIEX Analyst 1.6.3 software
(Applied Biosystems).
In addition, a hybrid quadrupole-time-of-flight mass spectrometry (Bruker micro
Q-TOF, Germany) with an ESI source coupled to an Agilent 1200 Series LC system was
used for the characterization of synthetic products in positive ion mode. Mass

spectrometry parameters were optimized as below: ionization temperature was maintained
at 180 °C; nebulizer gas pressure, 1.2 bar; dry gas flow rate, 8.0 L/min; capillary voltage,
-4500 V; end plate offset, 2500 V. Mass spectra were acquired at 2 s per spectrum from
m/z 50 to 1500. Analytes were separated using the same LC protocol described above. The
resulting data were processed by use of Bruker Daltonics Data Analysis 3.4 software.
Statistical analysis. All obtaining data were processed using GraphPad Prism 8
(GraphPad Softrare, San Diego, CA, USA) and expressed as means ± SD. Unpaired
Student’s t-test was applied for statistical analysis. The values of p < 0.05 were deemed to
have significant differences.

Results
Oxidative demethylation of DFP in microsomal incubations. In microsomal
incubations supplemented with DFP, a demethylation metabolite (M1) with retention time
of 6.38 min was detected by monitoring ion pair m/z 126/108. Product ion m/z 108 of M1
was acquired by MRM-IDA-EPI scanning and generated by the breaking of the C-O bond
(Figure 1C). No such metabolite was detected in microsomal incubations lacking NADPH
(Figure 1A). M1 was synthesized by amination of maltol in aqueous NH3 (Scheme 1) and
1
then characterized by NMR and mass spectrometry. H-NMR (400 MHz, DMSO-d6): δ
11.39 (s, 1H), 7.40 (d, J = 6.9 Hz, 1H), 6.09 (d, J = 6.9 Hz, 1H), 2.17 (s, 3H). As expected,
synthetic M1 displayed similar retention time (6.35 min) and mass spectrum as that
detected in microsomal incubations (Figure 1D and E).
Formation of quinone metabolite derived from DFP in microsomal incubations.
Microsomal incubations were conducted in which GSH was included as the trapping agent.
A GSH conjugate (M2) at retention time of 10.37 min was detected in GSH-supplemented
rat microsomal incubations. Formation of M2 was monitored by acquiring
precursor/product ion pair at m/z 431/158 in positive ion mode (Figure 2B). The tandem
mass spectrum of M2 demonstrated product ions at m/z 158, 302, 355 and 199. The
fragment ion at m/z 158, the major fragment ion, was generated by the breaking of the C-S
bond of the conjugate. Fragment ion m/z 302 was formed by the neutral loss (NL) of the
-glutamyl moiety (-129 Da) from m/z 431. Formation of the product ion at m/z 355
resulted from cleavage of the glycine group (-75 Da). The characteristic ion at m/z 199
derived from m/z 302 was formed by the loss of the N-formylglycinyl group (-103 Da).

Control incubation lacking NADPH failed to produce such conjugate (Figure 2A). Similar
in vitro incubations were carried out in human liver microsomes. As expected, we detected
a GSH conjugate with similar retention time and mass spectral property (Figure S1).
Biomemetic synthesis of M2 was carried out by oxidation of synthetic M1 using Ag2O
and then conjugating with GSH. The obtained product demonstrated similar
chromatographic property and mass spectrum (Figure 2D and E) as that detected in
GSH-fortified rat microsomal incubations of DFP.
Similar microsomal incubations were conducted which contained NAC in place of
GSH as the tapping agent. By monitoring ion pair m/z 287/158, an NAC conjugate (M3)
with the retention time of 10.41 min was detected (Figure 3B) in the incubations. The
tandem mass spectrum obtained from MRM-IDA-EPI scanning showed product ions at
m/z 158, 130 and 84. The characteristic product ion at m/z 158 was generated by the
breaking of the C-S bond in NAC moiety (-129 Da) and fragment ion m/z 130 due to S-C
cleavage at NAC side. In addition, product ion m/z 84 was the most characteristic
fragment ion for the parent moiety (Figure 3C). As expected, no such NAC conjugate was
detected in microsomal incubations lacking NADPH (Figure 3A). Meanwhile, we found
the same NAC conjugate in human liver microsomal incubations (Figure S2). M3 was
synthesized through oxidation of M1 by Ag2O and then reaction with NAC. The resultant
synthetic product was characterized by NMR and mass spectrometry. The protonated
molecular ion of the product was m/z 287.0691 (HPLC-Q-TOF). According to the
predicted formula, the error between the accurate mass observed in the high-resolution MS
1
system and the corresponding theoretical mass was less than 5 ppm (Table 1). H-NMR

(600 MHz, Deuterium Oxide): δ 7.00 (s, 1H), 4.49 (dd, J = 7.4, 4.1 Hz, 1H), 3.60 (dd, J =
14.6, 4.2 Hz, 1H), 3.36 (dd, J = 14.6, 7.5 Hz, 1H), 2.44 (s, 3H), 1.90 (s, 3H) (Figure S3B
and 4B). The synthetic M3 displayed similar chromatographic behavior and mass spectral
property as M3 observed in NAC-fortified microsomal incubations of DFP.
Biliary and urinary phase II metabolites of DFP. M2 was observed in bile of rats
given DFP by aquiring ion pair m/z 431/158, while we failed to observe biliary M2 in rats
given vehicle (Figure 5A and 5B). M2 was found to show similar retention time and mass
spectral behavior as synthetic M2. Urinary M3 was found in rats administered DFP. The
detected metabolite displayed the same chromatographic identity and similar mass spectral
property as that of synthetic M3 (Figure 5E and 5F). As expected, M3 was not detected in
the urine of animals given vehicle.
Requirement of N-demethylation for metabolic activation of DFP. Microsomal
incubations of synthetic M1 supplemented with GSH or NAC were performed to define
the metabolic activation pathway. Both GSH and NAC conjugates were detected in the
GSH- or NAC-supplemented microsomal incubations (Figure S4). The two conjugates
demonstrated similar chromatographic and mass spectral identities as biliary M2 and
urinary M3 observed in animals treated with DFP. As expected, M2 and M3 were not
detected in NADPH-free microsomal incubations.
P450 enzymes responsible for bioactivation of DFP. Individual CYPs3A5, 3A4, 2E1,
2D6, 2C19, 2C9, 2B6, 2A6, and 1A2 were incubated with DFP supplemented with
NADPH and NAC. The resultant mixtures were submitted to LC-MS/MS analysis, and

NAC conjugate was determined to monitor the formation of DFP-derived the reactive
metabolite. CYP2A6 was found to be the primary enzyme participating in the generation
of M3 (Figure 6).
Effect of methoxsalen on formation of M3. Methoxsalen as a general inhibitor of
CYP2A6 (Sellers et al. 2000) was used to verify the role of CYP2A6 in the formation of
M3 in vitro. The contents of M3 were assessed by LC-MS/MS. M3 detected in a
solvent-treated group was normalized as 100%. As depicted in Figure 7, the presence of
methoxsalen resulted in 78% inhibition of the formation of M3 in human microsomal
incubations, and 64% inhibition in rat microsomal incubations.
Effect of DFP on cell viability. The cytotoxicity was evaluated in rat primary
hepatocytes. No significant cell death was observed after 12 h cell exposure at up to
concentration of 200 M (Figure S5).

Discussion
DFP is commonly used as an iron depleter in clinic practice, due to its low price.
Recently, many cases of liver damage caused by DFP have been reported (Xie et al. 2016;
Ceci et al. 2015), but the mechanisms of the toxicity have not been elucidated. A study
demonstrated that DFP is mainly excreted from urine as parent and glucuronide forms
(Fradette et al. 2016), and limited DFP is metabolized through glucuronidation (Gong et al.
2010). Here, we proposed a metabolic activation pathway of DFP mediated by P450 for
the better understanding of the hepatotoxicity of DFP.
Firstly, we examined mass spectrometric identity of DFP, which would assist the
identification of metabolites. DFP displayed the molecular ion at m/z 140 and fragments of
m/z 122, 108, and 96 (Figure S6) in positive ion mode.
Metabolism study was carried out in rat liver microsomal incubations supplemented
with DFP. M1 was not detected in control group which lacked NADPH (Figure 1A). This
suggests that NADPH participated in the generation of the oxidative metabolite. M1
displayed its molecular ion at m/z 126 with 14 Da less than that of DFP, suggesting that
M1 was a demethylation metabolite after metabolic loss of -CH3 of DFP. We also
performed similar incubations using human liver microsomes, but failed to detect M1,
possibly due to the detection limitation, while M1 was detected in CYP2A6 incubations
(Figure S7). This suggests that the metabolic demethylation was taking place in human
microsomes. Then we conducted similar incubations fortified with M1 in place of DFP
and then detected M3 (Figure S8), indicating DFP is activated via the same pathway in
humans as that in rats.

GSH and NAC were used to capture reactive intermediates resulting from metabolic
activation of DFP (Chen and Cai 1999; Du et al. 2017; Zhang et al. 2017). GSH conjugate
M2 and NAC conjugate M3 were detected in rat and human liver microsomal incubations
containing DFP and GSH/NAC (Figures 2B and 3B; S1 and S2). This clearly indicates
that an electrophilic intermediate was formed in DFP-supplemented rat liver microsomes.
Then we incubated synthetic M1 with rat liver microsomes fortified with GSH or NAC
and detected M2 and M3 in the corresponding incubation systems. This suggests that M1
was a metabolite responsible for the formation of the two conjugates.
Chemical synthesis was conducted for metabolite identification. The synthetic M1
characterized by NMR (Figures S3A and 4B) revealed the same retention time and
fragments as M1 detected in microsomal incubations (Figures 1D and E). The resultant
conjugates showed similar retention time and fragments (Figures 2D, 2E, 3D and 3E) as
M2 and M3 detected in microsomal incubations as well as in bile and urine samples
obtained from rats given DFP. Unfortunately, we did not obtain enough amount of M2 for
NMR analysis.
Metabolic oxidation of M1 could also produce a p-quinone methide intermediate
which can react with NAC and form a NAC conjugate with the same molecular weight as
that of M3. If the quinone methide were generated, we would have obtained the NAC
conjugate resulting from 1,4-addition at benzylic position. The NMR spectrum of M3
(Figure 4) excludes the possibility for the formation of such quinone methide intermediate,
since NAC was found to attach on the aromatic ring of M3.

Recombinant human P450 enzyme incubation study showed that CYP2A6 played the
major role in the metabolic activation of DFP (Figure 6). However, we cannot exclude the
contribution of other P450 enzyme to the bioactivation, particularly multiple P450
enzymes are involved in the metabolic activation of DFP. Although some drugs are
reportedly metabolized by human CYP2A6 but not necessarily by rat CYP2A1/2 (Miksys
et al. 2000; Messina et al. 1997), rat CYP2A1/2 are known to share about 60% homology
in amino acid sequence with human CYP2A6 (Martignoni et al. 2006). Our microsomal
inhibition study showed that the presence of methoxsalen inhibited the formation of M3 in
both human and rat microsomal incubations (Figure 7). Despite this, more work is in need
to ensure rat CYP2A1/2 and human CYP2A6 share the similarity in catalysis of the
metabolic activation of DFP.
GSH conjugates are often considered to be biomarkers for exposure to electrophiles
(Chen and Cai 1999). Two-step metabolic degradation of GSH conjugates mediated by
γ-glutamyltranspeptidase and dipeptidases produces the corresponding cysteine conjugates.
Sequential N-acetylation of cysteine conjugates offers NAC conjugates (mercapturic acids)
(Clapp and Young 1970). As expected, both biliary DFP-derived GSH conjugate and
urinary DFP-derived NAC conjugate were detected in DFP-administered rats (Figures 5B
and E). Besides, we analyzed plasma in rats, but neither M1 nor DFP-derived GSH/NAC
conjugates were detected in rats given DFP.
The observed GSH/NAC conjugates suggest the formation of the o-quinone
metabolite of DFP, and this electrophilic species may react with nucleophilic group of
liver proteins. These GSH/NAC conjugates were detected in both rat and human

microsomal incubations fortified GSH or NAC. It is our speculation that the metabolic
action of DFP would take place in patients administered DFP. Besides, the elimination of
DFP in liver is slow relative to that of other tissues. The accumulated DFP in liver
(Gong et al. 2010) could be responsible for the reported DFP-associated liver injury.
Given together, DFP was metabolized to M1 by oxidative N-demethylation (Scheme
2). M1, a hydroquinone derivative, was sequentially oxidized to o-quinone intermediate
M4 which reacted with GSH to produce M2 (a GSH conjugate). M2 was biotransformed
to M3 (a NAC conjugate).
In summary, the present work appeared to report, for the first time, the metabolic
activation of DFP. The bioactivation was achieved by N-demethylation and then oxidation
to produce o-quinone intermediate. CYP2A6 was involved in the metabolic activation of
DFP. This finding lays the groundwork for better understanding the mechanisms of the
reported DFP-induced liver injury.

Funding Information
This work was supported in part by the National Natural Science Foundation of China
[Grants
81773813 and U1812403].
Conflict of interest
The authors declare no competing financial interest.
Abbreviation
NADPH, β-nicotinamide adenine dinucleotide 2′-phosphate reduced tetra sodium salt;
RLMs, rat liver microsomes; HLMs, human liver microsomes; PBS, phosphate-buffered
saline; DMSO, dimethyl sulfoxide; NMR, nuclear magnetic resonance; ESI, electrospray
ionization;
MRM,
multiple-reaction
monitoring;
LC-MS/MS,
liquid
chromatography-tandem mass spectrometry; Q-TOF, quadrupole-time-of-flight; SPE,
solid phase extraction.

Reference
Bellanti F, Danhof M, Della Pasqua O. (2014). Population pharmacokinetics of
deferiprone in healthy subjects. British journal of clinical pharmacology, 78(6),
1397–1406.
Belmont A, Kwiatkowski JL. (2017). Deferiprone for the treatment of transfusional iron
overload in thalassemia. Expert review of hematology, 10, 493-503.
Bolton JL, Trush MA, Penning TM, Dryhurst G, Monks TJ. (2000). Role of quinones in
toxicology. Chemical research in toxicology, 13, 135-60.
Ceci A, Baiardi P, Felisi M, MD Cappellini, Tricta F. (2015). The safety and effectiveness
of deferiprone in a large-scale, 3-year study in italian patients. British Journal of
Haematology, 118(1), 330-336.
Chen YF, Cai H. (1999). Investigation of liver damage associated with traditional Chinese
medicines. Adverse Drug Reactions Journal, 1, 27-32.
Chi M, Peng Y, Zheng J. (2016). Characterization of glutathione conjugates derived from
reactive metabolites of bakuchiol. Chemico-biological interactions, 244, 178-86.
Chuansumrit A, Songdej D, Sirachainan N, Wongwerawattanakoon P, Kadegasem P,
Sasanakul W. (2016). Safety profile of a liquid formulation of deferiprone in young
children with transfusion-induced iron overload: a 1-year experience. Paediatrics and
international child health, 36, 209-13.
Clapp JJ, Young L. (1970). Formation of mercapturic acids in rats after the administration
of aralkyl esters. The Biochemical journal, 118, 765-71.
Cohen AR, Galanello R, Piga A, De Sanctis V, Tricta F. (2003). Safety and effectiveness of
long-term therapy with the oral iron chelator deferiprone. Blood, 102, 1583-7.
Dong LJ, Sun JG, Liu J. (2018). A preparation method of pantoprazole intermediate 2-
chloromethyl 3,4-dimethoxypyridine hydrochloride. CN 105111187 A.
Du Q, Wang Z, Yun N, Huang Y, Xu Q, Wang BH. (2017). Literature Analysis of 185
Cases of ADR Induced by Xianling Gubao Capsule. China Pharmacy, 28,
3785-3787.
Fradette C, Pichette V, Sicard É, Stilman A, Jayashankar S, Tsang Y C, Spino M, Tricta F.
(2016). Effects of renal impairment on the pharmacokinetics of orally administered

deferiprone. British journal of clinical pharmacology, 82(4), 994-1001.
Fredenburg AM, Sethi RK, Allen DD, Yokel RA. (1996). The pharmacokinetics and
blood-brain barrier permeation of the chelators 1,2 dimethly-, 1,2 diethyl-, and
1-[ethan-1'ol]-2-methyl-3-hydroxypyridin-4-one in the rat. Toxicology, 108, 191-9.
Gong QQ, Liu P, Zhang YN, Xu HL, Jia XJ. (2010). Pharmacokinetics and tissue
distribution of deferiprone in rats. Chinese Journal of Pharmacology and Toxicology,
24, 59-63.
Graham DG, Tiffany SM, Bell WR Jr, Gutknecht WF. (1978). Autoxidation versus
covalent binding of quinones as the mechanism of toxicity of dopamine,
6-hydroxydopamine, and related compounds toward C1300 neuroblastoma cells in
vitro. Molecular pharmacology, 14, 644-53.
Huang XP, Spino M, Thiessen JJ. (2006). Transport kinetics of iron chelators and their
chelates in Caco-2 cells. Pharmaceutical research, 23, 280-90.
Knutsen HK, Alexander J, Barregård L, Bignami M, Brüschweiler B, Ceccatelli S, Cottrill
B, Dinovi M, Edler L, Grasl-Kraupp B and others. (2017). Risks for public health
related to the presence of furan and methylfurans in food. European Food Safety
Authority, 15.
Kontoghiorghe CN, Kontoghiorghes GJ. (2016). New developments and controversies in
iron metabolism and iron chelation therapy. World journal of methodology, 6, 1-19.
Lee S, Kim JM, Chung CS, Cho KJ, Kim JH. (2003). Polymorphism in CYP2C9 as a
non-critical factor of warfarin dosage adjustment in Korean patients. Archives of
pharmacal research, 26, 967-72.
Li MS, Ma L, Zhang XC. (1999). Synthesis of substituted 1,2-benzoquinones and
substituted 3-hydroxy-4 (1H) -pyridinones: Application of oxidation-Michael
addition in organic synthesis. Chinese Journal of Chemistry, 17, 377-383.
Maggio A, D'Amico G, Morabito A, Capra M, Ciaccio C, Cianciulli P, Di Gregorio F,
Garozzo G, Malizia R, Magnano C and others. (2002). Deferiprone versus
deferoxamine in patients with thalassemia major: a randomized clinical trial. Blood
cells, molecules & diseases, 28, 196-208.
Ma JY, Zhou X, Fu J, Hu T, Or PM, Feng R, He CY, Chen WJ, Zhang X, Chen Y and

others. (2014). Metabolite profiling analysis of FR429, an ellagitannin purified from
Polygonum capitatum, in rat and human liver microsomes, cytosol and rat primary
hepatocytes in vitro. Chemico-biological interactions, 220, 33-40.
Martignoni M, Groothuis G M, de Kanter R. (2006). Species differences between mouse,
rat, dog, monkey and human CYP-mediated drug metabolism, inhibition and
induction. Expert opinion on drug metabolism & toxicology, 2(6), 875-894.
Messina E S, Tyndale R F, Sellers E M. (1997). A major role for cyp2a6 in nicotine
c-oxidation by human liver microsomes. Journal of Pharmacology and Experimental
Therapeutics, 282(3), 1608-1614.
Miksys S, Hoffmann E, Tyndale R F. (2000). Regional and cellular induction of
nicotine-metabolizing cyp2b1 in rat brain by chronic nicotine treatment. Biochemical
Pharmacology, 59(12), 1501-1511.
Rodrigues AD. (1999). Integrated cytochrome P450 reaction phenotyping: attempting to
bridge the gap between cDNA-expressed cytochromes P450 and native human liver
microsomes. Biochemical pharmacology, 57, 465-80.
Sellers E M, Kaplan H L, Tyndale R F. (2000). Inhibition of cytochrome P450 2A6
increases nicotine's oral bioavailability and decreases smoking. Clinical
pharmacology and therapeutics, 68(1), 35-43.
Shen L, Hillebrand A, Wang D Q, Liu M. (2012). Isolation and primary culture of rat
hepatic cells. Journal of visualized experiments : JoVE, (64), 3917.
Venkatakrishnan K, Von Moltke LL, Greenblatt DJ. (2001). Human drug metabolism and
the cytochromes P450: application and relevance of in vitro models. Journal of
clinical pharmacology, 41, 1149-79.
Xie YY, Lu Z, Kong XL, Zhou T, Bansal S, Hider R. (2016). Systematic comparison of the
mono-, dimethyl- and trimethyl 3-hydroxy-4(1H)-pyridones - Attempted optimization
of the orally active iron chelator, deferiprone. European journal of medicinal
chemistry, 115, 132-40.
Yao H, Peng Y, Zheng J. (2016). Identification of glutathione and related cysteine
conjugates derived from reactive metabolites of methyleugenol in
rats. Chemico-Biological Interactions, 253, 143-152.

Table legend
Table 1. Data of DFP-NAC Conjugate (M3) Obtained from HPLC-Q-TOF MS Analysis.
Figure legends
Figure 1. Characterization of oxidative M1. Extracted ion (m/z 126/108) chromatograms
collected from LC-MS/MS analysis of rat liver microsomal incubations containing DFP in
the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M1 formed in
rat microsomal incubations; D: Extracted ion (m/z 126/108) chromatogram collected from
LC/MS-MS analysis of synthetic M1; E: Tandem mass spectrum of synthetic M1.
Figure 2. Characterization of M2. Extracted ion (m/z 431/158) chromatograms collected
from LC-MS/MS analysis of rat liver microsomal incubations containing DFP and GSH in
the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M2 generated
in rat microsomal incubations; D: Extracted ion (m/z 431/158) chromatogram collected
from LC/MS-MS analysis of synthetic M2; E: Tandem mass spectrum of synthetic M2.
Figure 3. Characterization of M3. Extracted ion (m/z 287/158) chromatograms acquired
from LC-MS/MS analysis of rat liver microsomal incubations containing DFP and NAC
in the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M3
generated in rat microsomal incubations; D: Extracted ion (m/z 287/158) chromatogram
collected from LC/MS-MS analysis of synthetic M3; E: Tandem mass spectrum of
synthetic M3.
Figure 4. Structures (A) and 1H-NMR profiles (B) of M1 and M3.

Figure 5. Characterization of biliary M2 and urinary M3. Chromatograms of extracted ion
(m/z 431/158 for M2) acquired from LC-MS/MS analysis of bile harvested from rats
before (A) and after (B) administration of DFP; C: Tandem mass spectrum of biliary M2
in rats treated with DFP; Chromatograms of extracted ion (m/z 287/158 for M3) obtained
from LC-MS/MS analysis of urine of rats before (D) and after (E) treatment with DFP; F:
Tandem mass spectrum of urinary M3 in rats administered DFP.
Figure 6. Role of individual human recombinant P450 enzymes in the metabolic
activation of DFP. The catalytic capabilities of the enzymes were evaluated by monitoring
the formation of M3 after normalization according to the relative contents of the
corresponding P450 enzymes in human liver microsomes. The most abundant metabolite
M3 was deemed as 100%. Data represent the mean ± SD (n = 3).
Figure 7. Inhibitory effect of methoxsalen (10 μM) on the formation of metabolite M3 in
human (A) and rat (B) liver microsomal incubations containing NAC and NADPH. The
generation of M3 in control groups without methoxsalen was deemed as 100%. Data
represent the mean ± SD (n = 3). *p < 0.05; **p < 0.01 versus their respective incubations
without methoxsalen.
Scheme legends
Scheme 1. Synthetic route of M1.
Scheme 2. Proposed metabolic pathways of DFP mediated by CYP2A6.

Table 1. Data of DFP-NAC Conjugate (M3) Obtained from HPLC-Q-TOF MS Analysis.
[M+H]+
Error
Compound
M3
Formula
C11H15N2O5S
Calculated
287.0696
Detected
287.0691
ppm
1.7
mDa
0.50
mSigma
7.70
Figure 1. Characterization of oxidative M1. Extracted ion (m/z 126/108) chromatograms
collected from LC-MS/MS analysis of rat liver microsomal incubations containing DFP in
the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M1 formed in
rat microsomal incubations; D: Extracted ion (m/z 126/108) chromatogram collected from
LC/MS-MS analysis of synthetic M1; E: Tandem mass spectrum of synthetic M1.

Figure 2. Characterization of M2. Extracted ion (m/z 431/158) chromatograms collected
from LC-MS/MS analysis of rat liver microsomal incubations containing DFP and GSH in
the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M2 generated
in rat microsomal incubations; D: Extracted ion (m/z 431/158) chromatogram collected
from LC/MS-MS analysis of synthetic M2; E: Tandem mass spectrum of synthetic M2.

Figure 3. Characterization of M3. Extracted ion (m/z 287/158) chromatograms acquired
from LC-MS/MS analysis of rat liver microsomal incubations containing DFP and NAC
in the absence (A) or presence (B) of NADPH; C: Tandem mass spectrum of M3
generated in rat microsomal incubations; D: Extracted ion (m/z 287/158) chromatogram
collected from LC/MS-MS analysis of synthetic M3; E: Tandem mass spectrum of
synthetic M3.

Proton
H-1
M1
M3
2.17 (3H,s)
2.44 (3H,s)
7.00 (1H,s)
H-5
6.09 (1H, d, J=6.9 Hz)
H-6
7.40 (1H, d, J=6.9 Hz)
H-7
-
3.60 (1H, dd, J=14.6, 4.2 Hz)
3.36 (1H, dd, J=14.6, 7.5 Hz)
4.49 (1H, dd, J=7.4, 4.1 Hz)
1.90 (3H,s)
-
H-8
-
H-11
N-H
-
11.39 (1H,s)
-
Figure 4. Structures (A) and 1H-NMR profiles (B) of M1 and M3.

Figure 5. Characterization of biliary M2 and urinary M3. Chromatograms of extracted ion
(m/z 431/158 for M2) acquired from LC-MS/MS analysis of bile harvested from rats
before (A) and after (B) administration of DFP; C: Tandem mass spectrum of biliary M2
in rats treated with DFP; Chromatograms of extracted ion (m/z 287/158 for M3) obtained
from LC-MS/MS analysis of urine of rats before (D) and after (E) treatment with DFP; F:
Tandem mass spectrum of urinary M3 in rats administered DFP.