In vitro liver metabolism of aclidinium bromide in preclinical animal species and humans: Identification of the human enzymes involved in its oxidative metabolism

Biochemical Pharmacology 81 (2011) 761–776

Contents lists available at ScienceDirect

Biochemical Pharmacology

journal homepage: www.elsevier.com/locate/biochempharm

In vitro liver metabolism of aclidinium bromide in preclinical animal

species and humans: Identiﬁcation of the human enzymes involved

in its oxidative metabolism

Joan J. Albert ı´ *, S o` nia Sentellas, Miquel Salv a`

Department of Pharmacokinetics and Drug Metabolism, Almirall, S.A., Laure a` Mir o´ 408-410, 08980 Sant Feliu de Llobregat, Barcelona, Spain

A R T I C L E I N F O

A B S T R A C T

Article history:

The metabolism of aclidinium bromide, a novel long-acting antimuscarinic drug for the maintenance

treatment of chronic obstructive pulmonary disorder, has been investigated in liver microsomes and

hepatocytes of mice, rats, rabbits, dogs, and humans. Due to the rapid hydrolysis of this ester compound,

two distinct radiolabeled forms of aclidinium were studied. The main biotransformation route of

aclidinium was the hydrolytic cleavage of the ester moiety, resulting in the formation of the alcohol

metabolite (M2, LAS34823) and carboxylic acid metabolite (m3, LAS34850), which mainly occurred non-

enzymatically. By comparison, the oxidative metabolism was substantially lower and the metabolite

proﬁles were similar across all ﬁve species examined. Aclidinium was metabolized oxidatively to four

minor primary metabolites that were identiﬁed as monohydroxylated derivatives of aclidinium at the

phenyl (M4) and glycolyl (m6 and m7) moieties of the molecule. The NADPH-dependent metabolite m4

involved the loss of one of the thiophene rings of aclidinium. Incubations with human recombinant P450

isoforms and inhibition studies with selective chemical inhibitors and antibodies of human P450

enzymes demonstrated that the oxidative metabolism of aclidinium is primarily mediated by CYP3A4

and CYP2D6. Additionally, up to eight secondary metabolites were also characterized, involving further

hydrolysis, oxidation, or glucuronidation of the primary metabolites. Also, the liver oxidative metabolism

of the alcohol metabolite (LAS34823) resulted in the production of one hydroxylated metabolite (M1)

mediated by human CYP2D6, whereas the acid metabolite (LAS34850) was not metabolized

enzymatically, although a minor non-enzymatic and NADPH-dependent reduction was observed.

Received 28 October 2010

Accepted 10 December 2010

Available online 22 December 2010

Keywords:

Aclidinium bromide

In vitro metabolism

Interspecies differences

Enzyme identiﬁcation

metabolites [2,3]. The main human esterase involved in the

enzymatic hydrolysis of aclidinium was identiﬁed as butyrylcho-

linesterase (BChE), which is found mainly in human plasma.

Cytochrome P450-catalyzed ester cleavage was not observed in

human liver microsomes [4].

1

. Introduction

Aclidinium bromide (AB) (also known as 3R-(2-hydroxy-2,2-

dithiophen-2-yl-acetoxy)-1-(3-phenoxy-propyl)1-azonia bicyclo

2.2.2] octane bromide) is novel, long-acting muscarinic

[

a

In vitro incubations with liver microsomes and/or hepatocytes

can be used to predict potential biotransformations in humans and

in those animal species used for preclinical safety studies.

Hepatocyte incubations retain Phase I and Phase II enzyme

activities and are therefore useful in determining overall metabo-

lism. They also mimic in vivo metabolism more accurately than

incubations with subcellular fractions [5]. Previous data using

diagnostic substrates have shown that P450 activities in rat, dog,

monkey, and human hepatocyte suspensions are not signiﬁcantly

decreased by cryopreservation [6].

antagonist [1] undergoing Phase III clinical trials for the mainte-

nance treatment of chronic obstructive pulmonary disorder. This

ester compound displayed non-enzymatic hydrolysis of its ester

bond at neutral and basic pH. Furthermore, AB was rapidly

hydrolyzed in plasma of different animal species and humans to

yield its alcohol (LAS34823) and carboxylic acid (LAS34850)

Abbreviations: AB, aclidinium bromide; BChE, butyrylcholinesterase; CID, collision-

induced dissociation; P450, cytochrome P450; ESI, electrospray ionization; FMO,

ﬂavin-containing monooxigenases; glyc, glycolyl; K

h

, hydrolysis rate constant; LC,

The objectives of this study were (a) to compare the in vitro

metabolism of aclidinium in liver microsomes and hepatocytes

of different preclinical animal species and humans; (b) to identify

the oxidative metabolites; and (c) to identify and kinetically

characterize the human enzymes responsible for the oxidative

liquid chromatography; MRM, multiple reaction monitoring; MS, mass spectrome-

try; m/z, mass-to-charge ratio; phe, phenyl; SPE, solid-phase extraction; TEA,

triethylamine.

*

Corresponding author. Tel.: +34 93 291 29 48; fax: +34 93 291 29 97.

E-mail address: joan.alberti@almirall.com (J.J. Albert ı´ ).

doi:10.1016/j.bcp.2010.12.013

7

62

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

Hydrolysis

m2

-

[

M-H] m/z 155

Ox

Hydrolysis

M2

(LAS34823)

[M ] m/z 262

^M²gluc

Hydrolysis

M3/m4

M ] m/z 402

+

[

[M ] m/z 438

Hydrolysis

major)

(

CYP2D6

m1

LAS101563)

M-H] m/z 157

(

CYP3A4

CYP2D6

-

[

Aclidinium

M5/m6, M6/m7

M1

Bromide

+

[

M ] m/z 500

(LAS188638)

M ] m/z 278

+

[

M ] m/z 484

+

[

Hydrolysis

CYP2D6

CYP1A1)

m_a

(major)

(

LAS101565)

-

[

M-H] m/z 223

non-enzymatic

NADPH

Hydrolysis

M1gluc

non-enzymatic

+

M6/m5

M ] m/z 500

[M ] m/z 454

+

[

artifact

m3

(

LAS101563)

(LAS34850)

[M-H] m/z 239

-

+

[

M+H] m/z 195

Fig. 1. Proposed in vitro metabolic pathways for aclidinium bromide in animal species and humans. Dashed arrows indicate possible paths of metabolite formation. Symbols *

1

4

14

and # denote the positions of C-labeled carbon atoms in C-phe-AB and C-glyc-AB, respectively. Metabolites generated from C-phe-AB incubations were coded as ‘‘M’’,

1

4

whereas metabolites generated from C-glyc-AB incubations were coded as ‘‘m’’.

metabolism. Due to the hydrolysis mechanism, two distinct

radiolabeled forms of aclidinium were prepared with the radioac-

tive carbon-14 label incorporated into the phenyl or the glycolyl

moieties of the molecule (Fig. 1). The phenotyping reaction of

aclidinium and its hydrolysis metabolites was performed using

human-expressed recombinant P450 isoforms and P450-speciﬁc

chemical inhibitors and inhibitory antibodies.

[dithienyl-glycolic acid, sodium salt]), LAS188638 and LAS101563

standards were synthesized at Ranke Qu ı´ mica S.L. (Barcelona,

1

4

14

Spain). Working solutions of C-phe-AB and C-glyc-AB were

prepared daily before use in 0.01N HCl:acetonitrile (80:20, v/v) to

prevent aclidinium hydrolysis. LC-grade methanol, acetonitrile,

dimethylsulphoxide (DMSO), and triethylamine (TEA) were

obtained from Scharlab S.L. (Barcelona, Spain). Glucose-6-phos-

+

phate, NADP , glucose-6-phosphate dehydrogenase, 2-thiophene-

2

. Materials and methods

glyoxylic acid, dithienyl-2-ketone, and Krebs–Henseleit medium

with D-glucose were purchased from Sigma–Aldrich (Steinheim,

2

.1. Chemicals

Germany). Williams’ E medium and other chemicals used in

hepatocyte incubations were purchased from Gibco Invitrogen

(Paisley, UK). All other chemicals used were purchased from

Sigma–Aldrich (Steinheim, Germany).

14

C-phenyl-AB ( C-phe-AB, 30.2 mCi/mmol) and C-glycolyl-

14

AB ( C-glyc-AB, 26.0 mCi/mmol) were synthesized at Quotient

BioResearch Ltd. (Northamptonshire, UK). The radiolabeled

hydrolysis products

14

C-LAS34823 (23.9 mCi/mmol) and

C-

2.2. Biological materials

LAS34850 (24.4 mCi/mmol) were prepared by basic hydrolysis

1

4

14

from

C-phe-AB and

C-glyc-AB, respectively, and further

CD-1 mouse (male and female), Sprague–Dawley rat (male and

female), New Zealand white rabbit (female), Beagle dog (male), and

human (mixed gender, n = 50 donors) liver microsomes were

purchased from XenoTech (Lenexa, KS, USA). CD-1 mouse (male),

Wistar rat (male), New Zealand white rabbit (female) and human

(mixed gender, n = 10 donors) cryopreserved hepatocytes were

purchased from Celsis International (Chicago, USA). Cryopreserved

Beagle dog (male) hepatocytes were purchased from XenoTech.

Recombinant human P450 (CYP1A1, 1A2, 2C8, 2C9*1, 2C19*1, 2A6,

2B6, 2D6*1, 2E1, 3A4, 3A5, 4A11, 2F2, 4F3A, 4F3B) and ﬂavin-

containing monooxigenases (FMO1, FMO3, and FMO5) expressed

puriﬁcation (Pharmacokinetics & Drug Metabolism Department,

Almirall S.A., Barcelona, Spain). All radiolabeled compounds

exhibited purity over 98% (determined by liquid chromatography

LC] with UV and radiometric detection). Stock solutions of 5 mM

C-phe-AB and 5 mM

HCl:acetonitrile (10:90, v/v). Stock solutions of 5 mM

LAS34823 and 10 mM C-LAS34850 were prepared in aqueous

basic solution containing 20% acetonitrile. Non-radiolabeled

aclidinium (>99% purity), its alcohol metabolite (LAS34823;

3(R)-hydroxy-1-(3-phenoxy-propyl)-1-azonia-bicyclo[2.2.2]oc-

tane, bromide]) and its carboxylic acid metabolite (LAS34850;

[

1

4

14

C-glyc-AB were prepared in 0.1N

1

4

C-

1

4

[

TM

in microsomes of baculovirus-infected cells (Supersomes ) were

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

763

obtained from BD Biosciences (San Jose, CA, USA). All the P450

2.5. Incubations with human recombinant P450 and FMO isoforms

isoforms contained NADPH-P450 reductase. Selective human P450

antibodies (CYP1A2, 2A6, 2B6, 2C8, 2D6, 2E1, and 3A4/5) were

obtained from Gentest (Woburn, MA, USA).

1

4

14

The compounds

m

C-Phe-AB (5

m

M),

C-Glyc-AB (5 and

1

4

25

M), and C-LAS34823 (5 mM) were incubated in duplicate

at 37 8C for 15, 30 and 60 min, respectively, with a panel of

2.3. Liver microsome incubations

recombinant human P450 enzymes (CYP1A1, 1A2, 2B6, 2C8, 2C19,

2D6, 2E1, 3A4, 3A5, 4F2, 4F3A, 4F3B, CYP2A6, 2C9, and 4A11) at

All microsomal incubations were carried out in open-to-air

polyethylene tubes at 37 8C in a shaking water bath. Microsomal

protein (0.5 mg/ml) in 50 mM phosphate buffer (pH 7.4) contain-

25 pmol of P450/ml and with recombinant FMO1, FMO2, and

FMO3 at 0.2 mg of protein/ml. Protein concentration in recombi-

nant human P450 incubations was normalized (0.25 mg/ml) using

insect control protein, without esterase and P450 activity. The

incubation conditions and sample workup were similar to those

described above (Section 2.3).

2

ing 3 mM MgCl , 1 mM EDTA, and co-factor generating system

+

(

1 mM NADP , 5 mM glucose-6-phosphate and 2.5 units/ml

glucose-6-phosphate dehydrogenase) were placed in a shaking

water bath for 3 min at 37 8C. Reactions were initiated by the

1

4

14

addition of the test compounds ( C-phe-AB and C-glyc-AB

2.6. Chemical and antibody inhibition studies

14

incubations) or NADPH-generating system ( C-LAS34823 and C-

LAS34850 incubations) to give a ﬁnal incubation volume of 1 ml.

Incubations were terminated by the addition of 0.5 ml of ice-cold

.2N HCl. Aliquots of the incubation samples (0.3 ml) in time-

dependent studies were taken at different pre-deﬁned times (15,

0, and 60 min) and terminated by the addition of 0.15 ml of ice-

Incubations were conducted in human liver microsomes as

described above at the following test substance concentrations:

1

4

14

0

5 mM ( C-LAS34823), 5 and 25 mM ( C-phe-AB and C-glyc-AB,

respectively) in a ﬁnal volume of 1 ml. The selective chemical

human P450 inhibitors assayed and ﬁnal concentrations were

3

cold 0.2N HCl. Enzyme kinetics studies in human liver microsomes

were performed under initial linear conditions with respect to

1

m

M

a

-naphthoﬂavone (CYP1A2), 2.5

m

M 8-methoxypsolaren

M quercetin (CYP2C8), 10 M sulfaphenazole

omeprazole (CYP2C19), 10 quinidine

M ketoconazole (CYP3A4), and 2.5 M 4-methyl-

(CYP2A6), 10

(CYP2C9), 10

(CYP2D6), 2

m

14

incubation time (15 min for C-phe-AB and C-glyc-AB and

m

M

mM

m

1

4

14

3

0 min for C-LAS34823 and C-LAS34850) and microsomal

m

14

protein concentration (0.25 mg/ml for C-phe-AB and C-glyc-AB

pyrazole (CYP2E1). After 3 min of pre-incubation at 37 8C,

reactions were started by addition of the NADPH generating

system ( C-LAS34823) or the test substance ( C-phe-AB and C-

14

and 0.5 mg/ml for C-LAS34823 and C-LAS34850). All incuba-

tions were carried out in duplicate, and the concentration of

organic solvent (acetonitrile) was kept below 2% (v/v). Parallel

incubations were always performed in the incubation buffer as a

non-enzymatic hydrolysis evaluation. Sample analysis was con-

ducted using solid phase extraction (SPE)–LC system with

radiometric detection.

1

4

14

1

4

14

glyc-AB). After 15 min ( C-phe-AB and C-glyc-AB) or 30 min

1

4

(

C-LAS34823), reactions were stopped by the addition of 0.5 ml

ice-cold 0.2N HCl. Incubations in the presence of the irreversible

CYP2A6 inhibitor 8-methoxypsolaren were preceded by 15 min of

pre-incubation time with NADPH before the addition of the test

compounds. The ﬁnal concentration of acetonitrile in the incuba-

tion mixtures was 0.7%. To determine the effect of anti-P450

monoclonal antibodies on the metabolism of C-LAS34823, C-

phe-AB, and

presence of 5

2.4. Hepatocyte incubations

1

4

14

1

4

Immediately before use, hepatocytes were thawed according

C-glyc-AB, incubations were conducted in the

ml antibody/100 mg microsomal protein in all cases

to the recommended protocol. In brief, frozen cells (approxi-

6

mately 5 ꢀ 10 cells/vial) were thawed quickly by gentle shaking

except for CYP3A4 antibody (10 ml/100 mg). The different mono-

in a 37 8C water bath. Immediately after thawing, the hepatocyte

clonal human P450 antibodies were added to human liver

microsomes (0.5 ml) and incubated for 20 min on ice. Incubations

were conducted as described above in a ﬁnal volume of 1 ml. The

ﬁnal concentration of acetonitrile in the incubation mixtures was

0.2%. All incubations were carried out in duplicate. Sample analysis

was conducted using SPE–LC system with radiometric detection.

suspension was diluted with Williams’ E medium (pH 7.4)

supplemented with 2.5

m

M dexamethasone, 4

mg/ml insulin,

1

00 units/ml penicillin, 100

m

g/ml streptomycin, 2 mM gluta-

mine, and 10% fetal bovine serum pre-warmed to 37 8C. After

centrifugation (25 8C, 50 ꢀ g, 5 min) and medium aspiration, the

hepatocyte pellet was resuspended by gentle inversion in 2.0 ml

of pre-warmed Krebs–Henseleit buffer (pH 7.4) supplemented

2.7. LC analysis with radiometric detection

2 3

with 1 mM CaCl , 25% NaHCO , 20 mM HEPES, diluted to 30 ml

with the same buffer and washed once. Hepatocyte viability was

measured using the trypan blue exclusion method. The initial cell

2.7.1. Method A

Microsomal incubation samples were centrifuged for 10 min

(4 8C) at 2000 ꢀ g (Heraeus Omnifuge 2.0RS, Heraeus Sepatech

GmbH, Osterode, Germany), and the supernatant (1.5 ml) was

analyzed by on-line SPE (Prospekt, Spark-Holland, Emmen,

Netherlands) coupled to LC (Waters Alliance 2695 system, Milford,

MA, USA) with radiometric ﬂow scintillation detection (Packard

150TR, PerkinElmer, Downers Grove, IL, USA), and UV detection at

220 and 240 nm (Waters 2996 Photo Diode Array). Brieﬂy,

following activation of an SPE Waters Oasis HLB cartridge

6

suspension was diluted at 2.1 ꢀ 10 viable cells/ml using Krebs–

Henseleit buffer (pH 7.4). Incubations were conducted in

6

suspension (2 ꢀ 10 viable cells/ml in a total volume of 0.1 ml)

for 1 and 2 h in round-bottom glass tubes at 37 8C in an

atmosphere of 5% CO

121, CO water jacketed incubator, Thermo Scientiﬁc, Marietta,

OH, USA). Radiolabeled test compounds were incubated at ﬁnal

concentrations of 20 M. To minimize P450 inhibition, the

concentration of organic solvent (acetonitrile) was kept below 1%

v/v). Reactions were terminated by addition of 0.1 ml of ice-cold

2

and 95% relative humidity (Forma Series II

3

2

m

(10 ꢀ 2 mm, 30

mm, 15 mg) with 1 ml of methanol and condition-

(

ing with 1 ml of pure water, incubation samples (1.5 ml) were

passed through the cartridges and cleaned up with 0.5 ml of

40 mM formic acid. The compounds retained were eluted directly

into the LC column using the mobile phase for 20 min.

Chromatographic separation was achieved using a Symmetry

acetonitrile/1N HCl (90/10, v/v) and incubation samples were

immediately frozen at ꢁ80 8C until analysis using LC with

radiometric detection. All incubations were performed in

triplicate and parallel incubations with 75

mM 7-ethoxycou-

marin were also performed as a positive control for Phases I and II

metabolism.

C

18 column (250 ꢀ 4.6 mm; 5

mm; Waters), eluted at a constant

ﬂow rate of 1 ml/min. The mobile phase consisted of (A) 40 mM

7

64

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

ammonium formate containing 0.1% triethylamine (pH 4.5) and (B)

acetonitrile as solvents. Samples were separated using a linear

gradient from 10% to 55% B within 30 min. The mobile phase was

returned to the starting solvent mixture in 1 min and the system

equilibrated for 8 min between runs. Incubation samples during LC

analysis were always kept at 4 8C in order to prevent aclidinium

ester cleavage. Chromatographic data were processed using

nebulizing gas. LC/MS and LC/MS/MS experiments were performed

at a voltage between 2.0 and 4.5 kV (positive mode) and 1.5 kV

(negative mode). The collision energy was between 30–45 eV in

positive mode and 5–30 eV in negative mode.

Metabolite identiﬁcation was established by comparison with

synthetic standards, when available, using retention time and

mass spectral data.

32

Millenium software, version 3.2 (Waters). The analytical method

was validated for selectivity, linearity, quantitation limit, intra-

and inter-batch precision and accuracy, recovery, and stability,

demonstrating the suitability of this method for the quantitative

measurements of test compounds and their respective metabo-

lites.

2.9. NMR analysis

M1 (approximately 50

radiolabeled LAS34823 with rat and rabbit liver microsomes in the

presence of NADPH was characterized using H nuclear magnetic

resonance (NMR). The H NMR spectrum was recorded in a Varian

mg) isolated from an incubation of non-

1

2.7.2. Method B

Mercury-plus NMR spectrometer (Palo Alto, CA, USA) at 400 MHz.

Hepatocyte incubation samples were centrifuged for 10 min

The sample was dissolved in deuterated DMSO-d

were reported in the scale (ppm) by using DMSO-d

2.49 ppm as a reference.

6

. Chemical shifts

(

4 8C) at 2000 ꢀ g and the supernatant was analyzed by LC with

radiometric detection. Chromatographic separation was achieved

using m;

d

6

signal at

a

Luna Phenyl-Hexyl column (250 ꢀ 4.6 mm;

5 m

Phenomenex, Torrance, CA, USA) and eluted at a constant ﬂow

rate of 0.8 ml/min. The mobile phase consisted of mixtures of (A)

2.10. Statistics and data analysis

1

M ammonium acetate:water:acetonitrile (10:890:100, v/v/v)

Data from kinetic studies were analyzed by non-linear

regression analysis with Graﬁt 5.0 (Erithacus Software, Staines,

UK) using models for Michaelis–Menten kinetics, biphasic kinetics,

or sigmoidal kinetics. In those cases where non-enzymatic

hydrolysis occurred, the determination of the apparent Michae-

and (B) 1 M ammonium acetate:water:acetonitrile (10:290:700, v/

v/v). An initial proportion of 10% B was maintained for 1 min after

sample injection (75

0% B within 7 min. B was maintained at 90% for 12 min and the

mobile phase was returned to the starting solvent mixture in

min. The system was equilibrated for 7 min between runs. All LC

ml), followed by a linear gradient from 10% to

9

lis–Menten parameters (K

m

and Vmax) and non-enzymatic hydro-

h

lysis rate constant (K ) were carried out using the following

1

analysis samples were also kept at +4 8C in order to prevent

aclidinium ester cleavage. Chromatographic data were processed

as described above.

equation [4]:

V

max½Sꢂ

y

^¼K

þ K ½Sꢂ

h

The designation of the individual metabolites was coded to

specify the structure and the origin of the biotransformation

product. Thus, the initial letter refers to the parent compound

m

þ ½Sꢂ

where is the formation rate, [S] is the substrate concentration,

y

1

4

14

origin: ‘‘M’’ metabolites formed from

C-phe-AB and

C-

K

m

is the apparent Michaelis–Menten constant, Vmax is the

is the non-enzymatic

1

4

LAS34823 incubations, and ‘‘m’’ metabolites formed from C-

apparent maximum velocity, and K

h

14

glyc-AB and C-LAS34850 incubations. The number given to a

metabolite corresponds to a speciﬁc chemical entity identiﬁed

according to the observed chromatographic retention time in the

radiochromatograms. For example, metabolites M4 and m5

correspond to the same biotransformation product obtained in

hydrolysis rate constant. The goodness-of-ﬁt criteria used to

select the model comprised visual inspection, consideration of the

randomness of residuals, and the standard error of the param-

eters.

14

C-phe-AB or in C-glyc-AB incubations, respectively. Other

minor LAS34823, LAS34850, and aclidinium metabolites were

coded as M/m, followed by a lower-case subscript letter (e.g.: M ).

3. Results

f

3.1. Validation of the LC method with radiometric detection

The identiﬁcation of metabolites was based on comparison with

standards and LC–MS/MS analysis.

14

The oxidative and NADPH-dependent metabolites of C-phe-

1

4

AB and C-glyc-AB were formed at a very low extent. Therefore,

on-line SPE was necessary to improve the sensitivity of the

radiometric detection method. To ensure the applicability of this

method for the analysis of incubation samples, the method was

fully validated for accuracy, precision, linearity, limit of quantita-

tion, and compound stability. Overall, the analytical method

demonstrated its suitability for the quantitative measurement of

AB metabolites originated in vitro. The extraction recovery for test

compounds and oxidative and hydrolysis metabolites ranged from

80% to 108%. For each day of incubation, calibration standards

(n = 5, r > 0.995) and quality-control samples (n = 2 at three

different concentrations) of the corresponding radiolabeled test

compounds were prepared in incubation buffer and treated as

incubation samples. Intra- and inter-batch precision and accuracy

of measurements of quality-control samples, expressed as coefﬁ-

cient of variation, were always lower than 10% at all concentration

2

.8. LC–MS and LC–MS/MS analysis

The LC–MS system used for analysis of incubation supernatants

consisted of an Agilent 1100 LC separation module (Waldbronn,

Germany) coupled to an API Sciex 3000 triple quadrupole mass

spectrometer (PerkinElmer, Boston, MA, USA). Chromatographic

separation was achieved using

a Symmetry C18 column

(

100 ꢀ 2.6 mm; 3.5 m; Waters) with elution at a constant ﬂow

m

rate of 0.5 ml/min. The mobile phase consisted of (A) 10 mM

ammonium formate at pH 4.5 and (B) acetonitrile as solvents. An

initial proportion of 100% A was maintained for 2 min after sample

injection, followed by a linear gradient from 0% to 45% B within

2

8 min. The mobile phase was returned to the starting solvent

mixture in 1 min, and the system was equilibrated for 7 min

between runs. Mass spectrometry was conducted with electro-

spray ion source operating in positive ion mode (AB, LAS34823 and

their metabolites) or negative ion mode (AB, LAS34850 and their

metabolites). The ion spray interface temperature was set at 450 8C

1

4

14

levels. Following incubations of C-phe-AB, C-glyc-AB, C-

1

4

LAS34823, and C-LAS34850 in human liver microsomes in the

presence of NADPH, the supernatant recoveries based on total

radioactivity peak area (sum of peak areas of all metabolites and

(positive mode) and at 250 8C (negative mode) using nitrogen as a

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

765

100

The radioactive metabolite proﬁles of liver microsomal

1

4

14

incubations of C-phe-AB and C-glyc-AB (50

mM) with NADPH

8

6

4

2

0

were qualitatively similar across species and representative

chromatograms are shown in Fig. 3. After incubation of C-phe-

AB, the major metabolite observed in all species was M2

14

Buffer

-

NADPH

(

LAS34823, alcohol metabolite), whilst other minor and NADPH-

+

dependent metabolites M1, M3, M4, M5, and M6 were also

observed in different species. In the same way, the acid metabolite

m3 (LAS34850) and metabolite m2 were the main metabolites in

1

4

C-glyc-AB incubations. In addition, trace amounts of other

a

metabolites (m1, m , m4, and m5) were also observed in some

MOUSE MOUSE

m) (f)

RAT

(m)

RAT

(f)

RABBIT

(f)

DOG

(m)

HUMAN

(mix)

species.

(

1

4

Incubation of C-LAS34823 with pooled human liver micro-

somes revealed the NADPH-dependent metabolism of this

compound to metabolite M1 (Fig. 3). Traces of other hydroxylated

LAS34823 metabolites were also observed using MS/MS detection

1

4

Fig. 2. Percentage of metabolism of C-phe-AB (50

mM) after 60 min in incubation

buffer and in liver microsomes of different species (m): male; (f): female; (mix):

mixed gender. Each bar represents the mean of duplicate determinations (<15%

variance).

(

Section 3.4.14).

After incubation of

1

4

C-LAS34850 in incubation buffer, a

remaining parent compound) were, in all cases, greater than 85%

with respect the radioactivity peak area of the test compound in

control incubations. All test compounds were shown to be stable

during the sample analysis.

radioactive peak was observed with a retention time of

31.5 min. The formation of this compound was independent of

the incubation time. According to its chromatographic retention

time and comparison of the MS spectra with a commercial

standard, this artifact was identiﬁed as the compound di-2-thienyl

ketone (see below). In the presence of NADPH-generating system,

the artifact disappeared and a new radioactive peak (tret 19 min)

3.2. Metabolism of aclidinium bromide in liver microsomes

14

Incubations of C-phe-AB and C-glyc-AB in liver microsomes

0.5 mg protein/ml) of animal species and humans were conducted

at a concentration of 50 M in order to detect minor metabolites

a a

named m was detected (Fig. 3). The formation of m was very

(

rapid and did not increase with incubation time. Further

investigations at different microsomal protein concentrations

m

formed during incubations. Both radiolabeled forms of AB were

incubated for 60 min in order to evaluate differences in their

metabolism. Aclidinium was shown to be an unstable ester in the

incubation buffer (pH 7.4). A species comparison of the overall

metabolism of C-phe-AB after 60 min of incubation is repre-

sented in Fig. 2. The hydrolysis half-life in the assay buffer (pH 7.4,

a

(0.25, 0.5, and 1 mg/ml) demonstrated that m formation was

non-enzymatic and dependent only on NADPH. This point was

conﬁrmed by performing additional incubations where the

+

presence of NADP (oxidized form) in buffer solution failed to

14

a a

generate m (results not shown). The chemical structure of m was

tentatively elucidated by LC–MS/MS as the compound 2-dithie-

nylacetic acid (see below). The presence of an acidic moiety in the

3

7 8C) was around 70 min. The in vitro disappearance half-life of

C-phe-AB in the absence of NADPH followed the rank order rat

1

4

a

chemical structure of m was also conﬁrmed by the chro-

(

4

male, female), 100 min ꢃ human, 93 min > mouse (male),

matographic behaviour of this compound, increasing its retention

time with decreasing pH of the mobile phase in the same way

observed for LAS34850. No other radioactive peaks were observed

in the chromatograms, which suggests that enzymatic metabolism

of LAS34850 does not occur.

2 min ꢃ mouse (female), 37 min > dog 19 min ꢃ rabbit, 16 min.

14

The results obtained following incubation of C-glyc-AB were

14

similar to those obtained with C-phe-AB. Possible metabolites

other than the hydrolysis metabolites (LAS34823 and LAS34850)

were not observed in incubations without NADPH, suggesting that

the oxidative metabolism of AB in liver microsomes is primarily

mediated by P450.

14

The metabolites found in liver microsomal incubations of C-

1

4

phe-AB and C-glyc-AB are summarized in Tables 1 and 2,

respectively.

According to these results, enzymatic hydrolysis of aclidi-

nium in human and rat liver microsomes was not observed,

suggesting that the potential presence of esterases responsible

for AB hydrolysis in the microsomal fractions of these two

species has little relevance. To conﬁrm this, additional incuba-

3.3. Metabolism of aclidinium bromide in hepatocytes

The stability of AB in Krebs–Henseleit incubation medium (pH

7.4) showed an apparent elimination half-life (t1/2) of 56 min,

indicative of considerable non-enzymatic hydrolysis in the

1

4

tions of C-phe-AB (5

mM) in human liver microsomes (1 and

1

4

2

mg/ml) were investigated in the absence of NADPH and

incubation conditions. The percentages of metabolism for C-

1

4

incubation buffer. There were slight differences in the formation

of the hydrolysis metabolites between human liver microsomes

and the incubation buffer. In fact, the net enzymatic hydrolysis

of AB at 2 mg/ml was less than 20% after 60 min of incubation.

These results are in agreement with previous studies where

BChE content was demonstrated to be low in human liver

microsomes and consequently, the enzymatic hydrolysis of AB

had little scientiﬁc relevance compared to non-enzymatic

hydrolysis [4].

phe-AB and C-glyc-AB were high in all species evaluated after 2 h

of incubation, ranging from 60% (human) to almost 100% (rabbit).

1

4

The major radioactive metabolites of C-phe-AB (20

mM) in

all species were M2 (LAS34823) and M1 (LAS188638). Represen-

tative chromatograms obtained in rabbit and human hepatocytes

are shown in Fig. 4. Rabbit hepatocytes showed the highest rate of

M1 formation (around 50%) and intermediate values (between 9%

and 16%) were observed for mouse, rat and human hepatocytes.

In contrast, dog hepatocytes showed the lowest formation of M1

(Table 1). Also, two polar radioactive metabolites were observed

in mouse, rat, and rabbit hepatocytes. Their structures were

identiﬁed by LC–MS/MS as the glucuronide conjugates of M1 and

M2, i.e., M1gluc and M2gluc, respectively (see below). Rabbit

hepatocytes produced the highest amount of glucuronide

conjugates. Although M1 and M2 were generated to a similar

Additional oxidative metabolism of aclidinium was observed in

the presence of NADPH. In this case, the in vitro disappearance

14

half-life value for the overall metabolism of C-phe-AB followed

the rank order rat (female), 73 min > rat (male), 56 min ꢃ human,

5

8

2 min > mouse (male, female), 23 min > dog 13 min ꢃ rabbit,

14

min. Similar values were also obtained for C-glyc-AB.

7

66

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

1

4

14

C-phe-AB and C-glyc-AB in phosphate buffer (A, B), 50

14

Fig. 3. Representative chromatograms (Method A) showing metabolite proﬁles of 50

mM

m

M

C-phe-AB and C-

1

4

14

glyc-AB in human liver microsomes (HLM) with NADPH (C, D), 10

buffer and human liver microsomes with/without NADPH (F).

mM

C-LAS34823 in human liver microsomes with NADPH (E), and 10 mM

C-LAS34850 in phosphate

1

4

extent in incubations, M2gluc was formed at much lower levels

than M1gluc Thus, it is reasonable to propose that M1

glucuronidation occurs at the hydroxyl moiety located at the

para position of the phenyl ring.

The major radioactive metabolite of C-glyc-AB (20 mM) in all

.

species was m3 (LAS34850), which accounted for 85%, 51%, 100%,

72%, and 64% in mouse, rat, rabbit, dog, and human hepatocytes,

respectively. Two minor polar metabolites were also observed in

Table 1

14

Metabolites of C-phe-aclidinium bromide detected in liver microsomes and hepatocytes. C-phe-AB was incubated with liver microsomes (0.5 mg/ml and NADPH-

generating system) and hepatocytes (2 M cells/ml) for 60 and 120 min, respectively. Results expressed as percentage of total peak area of the radiochromatograms.

Metabolite

RT

Mouse (m)

Mouse (f)

Rat (m)

Rat (f)

Rabbit (f)

Dog (m)

Human (mix)

Liver microsomes

M1

M2

M3

M4

M5

M6

6.2 (8.8)

3.7

75.7

d

3.5

75.4

ND

1.3

1.0

7.1

36.5

ND

2.0

1.4

2.0

4.3

36.5

ND

1.1

ND

8.2

91.0

d

9.3

85.7

ND

0.9

ND

0.9

47.5

2.1

2.5

0.7

1.1

12.6 (13.2)

19.9 (20.2)

21.8 (22.4)

22.6 (25.3)

23.1 (25.4)

1.1

d

ND

d

0.9

Hepatocytes

M1gluc

M2gluc

M1

3.9B (3.9)

4.4B (4.4)

8.7B (8.8)

12.2B (13.2)

2.7

–

2.5

–

14.8

1.4

ND

1.7

ND

9.2

0.5

0.6

15.9

73.1

13.9

59.5

51.2

32.6

M2

87.3

76.2

Mean values from duplicate (liver microsomes) and triplicate (hepatocytes) experiments.

m, male; f, female; mix, mixed gender.

RT, retention time (minutes) using method A, if not otherwise indicated. In brackets, retention times using LC–MS/MS.

: assay not performed.

–

ND, not detected, below 0.5% of total radioactive peak area.

d, metabolite detected at shorter incubation times.

14

Metabolites generated from C-phe-AB incubations were coded as ‘‘M’’. Metabolites M1gluc and M2gluc were not observed in microsomal incubation, whereas M3, M4, M5,

and M6 were not observed in hepatocyte incubations.

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

767

Table 2

14

Metabolites of C-glyc-aclidinium bromide detected in liver microsomes and hepatocytes. C-glyc-AB was incubated with liver microsomes (0.5 mg/ml and NADPH-

generating system) and hepatocytes (2 M cells/ml) for 60 and 120 min, respectively. Results expressed as percentage of total peak area of the radiochromatograms.

Metabolite

RT

Mouse (m)

Mouse (f)

Rat (m)

Rat (f)

Rabbit (f)

Dog (m)

Human (mix)

Liver microsomes

m1

m2

m3

4.4 (1.5)

3.1

23.5

57.2

3.4

d

2.3

ND

15.5

33.1

2.8

d

ND

9.6

36.0

2.0

ND

0.9

ND

1.5

1.8

19.5

74.1

4.6

d

ND

16.3

76.1

4.9

2.5

5.1 (3.0)

29.9

52.6

3.0

13.7

33.7

2.8

14.2 (16.6)

19.3 (18.6)

19.9 (20.2)

21.8 (22.4)

22.6 (25.3)

23.1 (25.4)

m

a

m4 (M3)

m5 (M4)

m6 (M5)

m7 (M6)

ND

1.3

ND

1.4

3.1

1.2

ND

1.7

1.0

ND

d

3.3

1.0

ND

0.9

Hepatocytes

m

x

3.5B (ND)

ND

0.5

–

0.9

–

ND

100

ND

0.8

ND

63.7

ND

m1

m2

m3

ND

1.4

85.4

ND

6.7

50.6

ND

0.8

3.9B (3.0)

8.1B (16.6)

ND

71.8

ND

m

a

Mean values from duplicate (liver microsomes) and triplicate (hepatocytes) experiments.

m, male; f, female; mix, mixed gender.

RT, retention time (minutes) using method A, if not otherwise indicated. In brackets, retention times using LC–MS/MS.

: assay not performed.

–

ND, not detected, below 0.5% of total radioactive peak area.

d, metabolite detected at shorter incubation times.

1

4

Metabolites generated from C-glyc-AB incubations were coded as ‘‘m’’. Metabolite m

observed in hepatocyte incubations.

x

, was not observed in microsomal incubation, whereas m4, m5, m6, and m7 were not

1

4

14

Fig. 4. Representative chromatograms (Method B) showing metabolite proﬁles of C-phe-AB and C-glyc-AB in human hepatocytes (A, B), C-phe-AB and C-glyc-AB in

14

rabbit hepatocytes (C, D), C-LAS34823 and C-LAS34850 in human hepatocytes (E, F). Test substances were incubated at a ﬁnal concentration of 20

mM.

7

68

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

1

4

mouse, rat, and dog cryopreserved hepatocytes. One of these minor

metabolites was identiﬁed as 2-thiopheneglyoxylic acid (m2),

Following incubation of C-LAS34823 (20 mM) with human

hepatocytes, only M1 (LAS188638) was observed, accounting for

28% of the radioactivity after an incubation time of 2 h. The acid

metabolite C-LAS34850 (20 mM) was not metabolized by human

x

accounting for 1–7% of total radioactivity. Metabolite m was only

1

4

observed in mouse, rat, and dog hepatocytes at a very low

concentration (Table 2).

hepatocytes during the 2 h of incubation (Fig. 4).

14

Metabolite m2 was only observed in C-glyc-AB incubations

14

and not in C-phe-AB incubations (Table 2, Fig. 3), which supports

the hypothesis that the metabolic attack took place at the

thiopheneglyoxylic moiety of the molecule. Nevertheless, this

3.4. Metabolite identiﬁcation

The identiﬁcation of metabolites was based on comparison with

standards (when available) using LC-ESI-MS/MS in either positive

or negative mode depending on the compound or metabolite of

1

4

metabolite was not observed following incubation of

C-

LAS34850, suggesting that m2 might originate from the parent

compound only.

Fig. 5. MS/MS spectra of (A) aclidinium bromide, (B) metabolite M2 (LAS34823), and (C) metabolite M1 (LAS188638).

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

769

interest. The exact position of hydrolylation on the terminal phenyl

ring of M1 was conducted using H NMR.

retention time of 4.4 min showed a peak at the MRM transition

of m/z 454 > 278, which corresponds to a glucuronide of M1.

Similarly, the radioactive peak observed at 5.1 min presented a

peak at the transition m/z 438 > 262, which strongly suggests the

formation of a glucuronide derivative of metabolite M2. Additional

1

3.4.1. Aclidinium bromide

The MS/MS spectrum of aclidinium in positive mode gave a

major fragment ion at m/z 262, corresponding to the molecular ion

of the alcohol derivative LAS34823. This fragment ion m/z 262 led

incubations of rabbit and mouse hepatocyte samples with b-

glucuronidase from Helix Pomatia demonstrated a decrease in the

radioactive peak areas, although both glucuronide metabolites

were highly resistant to enzymatic hydrolysis (results not shown).

2 2 2

to the sequential loss of H O (m/z 244) and H C 55 CH (m/z 216) as

shown in Fig. 5A. This fragmentation pattern is in accordance with

the MS/MS fragmentation of different N-alkyl derivatives of 3-

quinuclidinol molecules described by Bedn a´ r et al. [7].

3.4.5. Metabolite m3 (LAS34850)

This compound presented two major peaks at m/z 239 and m/z

195 in negative ESI mode (Table 3). The fragment of m/z 239

3

.4.2. Metabolite M2 (LAS34823)

ꢁ

LC–MS analysis of LAS34823 using ESI in positive mode showed

correspondstothemolecularion[M ꢁ H] , whereasthefragmentm/

+

a molecular ion [M ] of m/z 262. The product ion spectrum showed

a fragment ion of m/z 140 as the most abundant. The fragment ions

observed at m/z 168 and 107 suggest the loss of the phenox-

ymethyl moiety. The fragment of m/z 135 points to the loss of the

phenoxypropyl moiety (Fig. 5B).

z 195 may be assigned to decarboxylation (in source collision-

induced dissociation) of the molecular ion [8]. Additionally, a minor

ion at m/z 285 was consistent with the formation of an adduct with

ꢁ

the formate present in the mobile phase [M + HCOO] , a common

process described in the literature [9,10]. The product ion mass

spectrum of m/z 239 gave only the fragment of m/z 195, suggesting

3

.4.3. Metabolite M1 (LAS188638)

2

the loss of CO from the carboxylic acid group. The mass

+

This metabolite presented a molecular ion [M ] at m/z 278. The

fragmentation of the product ion of m/z 195 gave a main fragment

ion at m/z83, whichisinaccordancewiththe lossofathiophenering.

fragments obtained at m/z 168, 140, and 124 were also observed in

the spectrum of M2 (LAS34823), which suggests that the 3-

quinuclidinol moiety remains unchanged. The observed fragments

at m/z 123 (107 + 16 amu) and 151 (135 + 16 amu) clearly suggest

hydroxylation at the phenyl ring (Fig. 5C). In order to determine the

exact position of hydroxylation, this metabolite was isolated and

puriﬁed from rat and rabbit liver microsomal incubations for further

3.4.6. Metabolite m1 (LAS101563)

MS/MS (negative mode) characterization showed the molecular

ion at m/z 157 and a peak at m/z 113 (Table 3). Moreover, a formate

adduct was also observed in the MS spectrum (157 + 46),

conﬁrming the presence of a carboxylic acid moiety in its structure,

as observed for LAS34850 and m2. The chromatographic retention

time and the mass fragmentation were identical to those of the

authentic standard LAS101563 (hydroxyl-2-thienylacetic acid).

1

analysis by H NMR. The integrals of the aromatic region revealed

that the point of hydroxylation was at the 4-position, since the

spectrum showed two doublets (ata proportion 1:1) with a coupling

constant of J = 8 Hz. This was later conﬁrmed with the synthesis of

the authentic standard. The LC–MS/MS analysis showed the same

chromatographic retention time and mass fragmentation for both

M1 and the authentic standard (LAS188638).

3.4.7. Metabolite m2 (2-thiopheneglyoxylic acid)

The MS (negative mode) spectrum showed a molecular ion at

m/z 155. The peak observed at m/z 201 would be also compatible

with the formation of an adduct with formic acid (Table 3). Further

MS/MS characterization gave a single fragment at m/z 83,

corresponding to the thiophene ring. The LC-MS analysis showed

the same chromatographic retention time and mass fragmentation

of the authentic standard 2-thiopheneglyoxylic acid.

3.4.4. Metabolites M1gluc and M2gluc

The formation of these conjugated metabolites was observed in

14

hepatocyte incubations of C-Phe-AB, and their identiﬁcation was

conducted using the LC–MS/MS in multiple reaction monitoring

(MRM) mode. The theoretical transitions corresponding to the loss

of the glucuronic acid moiety (176 Da) of the potential glucur-

onides of M1 and M2 (m/z 454 > 278 and m/z 438 > 262,

respectively) were monitored. The radioactive peak with a

a

3.4.8. Metabolite m (LAS101565)

The full scan spectrum (negative mode) presented peaks at m/z

269, m/z 223, and m/z 179 (Table 3). The peak of m/z 223

Table 3

1

4

MS and MS/MS spectral data of C-glyc-AB metabolites.

Metabolite

MS spectra

MS/MS spectra

m/z

Tentative assignation

Relevant product ions (m/z)

ꢁ

m3

285 (15)

239 (91)

[M + HCOO]

239 (100); 195 (35)

(LAS34850)

[M ꢁ H]^ꢁ

239 (20); 195 (100)

ꢁ

1

95 (100)

37 (5)

[M ꢁ H ꢁ CO

2

]

195 (100); 83 (10)

na

–

ꢁ

m1

LAS101563)

203 (20)

[M + HCOO]

[M ꢁ H]^ꢁ

–

(

157 (100)

157 (15); 113 (100)

ꢁ

113 (20)

[M ꢁ H ꢁ CO

2

ꢁ

]

–

m2

2-thiopheneglyoxylic acid)

201 (5)

[M + HCOO]

201 (50); 155 (100)

(

155 (100)

[M ꢁ H]^ꢁ

155 (15); 83 (100)

8

9 (7)

na

–

ꢁ

3 (55)

[C

4

H

3

S]

–

ꢁ

m

a

269 (77)

223 (61)

[M + HCOO]

[M ꢁ H]^ꢁ

269 (75); 223 (63); 179 (100)

(LAS101565)

223 (100); 179 (100)

ꢁ

179 (100)

[M ꢁ H ꢁ CO

2

]

178/179 (75/50); 146/147 (75/50); 133 (92); 107 (80); 83 (100)

Data in parenthesis: relative abundance (%) in the spectrum.

na: not assigned.

–

: MS/MS spectrum not performed.

7

70

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

Fig. 6. MS/MS spectra of (A) M3/m4 and (B) M4/m5 metabolites.

ꢁ

corresponds to the molecular ion [M ꢁ H] . The peak observed at

m/z 269 is compatible with the formation of a formate adduct,

which was conﬁrmed by further MS/MS analysis. The fragment of

m/z 195 may be assigned to the decarboxylation of the molecular

ion, in analogy to LAS34850 and other acidic metabolites. Product

ion spectrum of m/z 179 showed abundant ions at m/z 178, 146,

fragments indicate that the biotransformation occurs at the

thiopheneglyoxylic acid moiety of the molecule. The product ion

spectrum showed fragments at m/z 308 and m/z 280, which

suggests that a thiophene ring has been removed from the

structure, corresponding to a net loss of 82 Da (Fig. 6A).

1

33, 107, and 83. The LC–MS analysis showed the same

3.4.11. Metabolite M4/m5

chromatographic retention time of the authentic standard

LAS101565.

This metabolite presented an ion mass spectrum of m/z 500,

which is 16 Da higher than that of the parent compound. The most

prominent fragment (m/z 278) pointed to

a hydroxylation

3

.4.9. LAS34850 artifact

occurring at the quinuclidinol portion of the molecule. The

presence of the minor fragments of m/z 140, 362, and 390

suggested that the hydroxylation took place at the phenyl ring in a

manner similar to that observed in the formation of M1 (Fig. 6B). To

conﬁrm this hypothesis, an M4-puriﬁed sample obtained from

incubations in liver microsomes was subjected to basic hydrolysis

in order to generate the alcohol derivative of this metabolite, which

matched the retention time and MS/MS spectrum of M1.

14

In incubations of

C-LAS34850 in phosphate buffer, a

radioactive chromatographic peak appeared at a retention time

of 31.4 min (Fig. 3), which was associated with a degradation

product or artifact of LAS34850. The commercially available

compound dithienyl-2-ketone presented the same retention time

and identical UV spectrum with a

using the LC–MS method conﬁrmed identical retention times

26.6 min) and identical mass fragmentation pattern in positive ESI

lmax at 315 nm. Further analysis

(

mode (molecular ion at m/z 195). Product ion scans of m/z 195

showed a fragment at m/z 111, corresponding to the loss of one

thiophene ring. This compound was not ionizable in negative ESI

mode.

3.4.12. Metabolites M5/m6 and M6/m7

Metabolites M5/m6 and M6/m7 were obtained in very low

quantities and the chromatographic resolution was poor. These

two metabolites presented a molecular ion of m/z 500 which is

consistent with oxidation of the parent compound. In both cases,

the MS/MS spectra showed a fragment of m/z 262, which is

characteristic of an intact quinuclidinol moiety. In addition, a

fragment of m/z 401 clearly points out to an oxidation at one of the

thiophene rings. Nonetheless, the MS/MS spectra did not provide

3.4.10. Metabolite M3/m4

The MS/MS (positive mode) of the molecular ion of m/z 402

presented major fragments at m/z 262 and 216, corresponding to

the intact quinuclidinol portion of the molecule. These intact

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

771

sufﬁcient information to distinguish between C-oxidation (hy-

droxylation) and S-oxidation.

3

.4.13. Metabolite m

x

This polar compound was present as a minor metabolite in

14

incubations of C-glyc-AB in mouse, rat, and dog hepatocytes,

although the low amounts prevented its characterization by LC–

MS/MS.

3.4.14. Other minor metabolites

Additionally, other minor metabolites were detected by LC–MS

but were below the level of radiometric detection. Four minor

metabolites (M , M , M , and M ) with a molecular ion at m/z 278

were observed in AB and LAS34823 incubations. The MS spectra of

two of them (M and M ) conﬁrmed the hydroxylation on the

phenyl ring of LAS34823, hence constituting a positional isomer of

M1. M presented a mass fragmentation pattern different from M1,

with a clear loss of water and the appearance of an ion at m/z 156

140 + 16) as the major fragment ion, suggesting that hydroxyl-

a

b

c

d

a

b

c

(

ation would take place at the 3-quinuclidinol group. Supporting

this hypothesis, the fragments at m/z 107 and 135 remained

unchanged. M

concentration. Metabolite M

d

characterization was not possible due to its low

presented a molecular ion m/z 408

e

and a major fragment of 186 Da, which is 76 Da less than the major

fragment (m/z 262) of LAS34823, consistent with the loss of the

phenyl ring. Another metabolite presented a molecular ion of m/z

4

f

00 (M , tret 23.8 min). This molecular ion and the fragments at m/z

1

52, 135, 124, and 107 suggested that the quinuclidinol portion of

the molecule remained intact. This fact implies that the

biotransformation should take place at one of the thiophene rings,

in a similar way already observed for M3/m4, with the ﬁnal loss of a

thiophene ring. The presence of a fragment of m/z 111 strongly

suggests oxidation of the alcohol moiety of M3/m4 to a ketone

derivative. In addition, other minor glucuronide metabolites in rat

and mouse hepatocytes were also observed in LC–MS/MS (MRM

mode) analysis using the transition m/z 454 > 278, consistent with

1

4

Fig. 7. Kinetics of (A) M1 formation from C-LAS34823, and (B) m

a

formation from

14

C-LAS34850 in human liver microsomes. Symbols represent observed data (mean

of duplicate determinations). Solid lines represent calculated rates with (A) bi-

structural isomers of M1gluc

.

enzymatic Michaelis–Menten equation in the range of concentration of 2–100

and (B) linear regression.

mM

3.5. Enzyme kinetics

1

4

14

The enzyme kinetics of C-LAS34823, C-LAS34850, C-phe-

192.7 pmol/min/mg protein, respectively. The catalytic efﬁciency

(Vmax/K ) of the high-afﬁnity component was approximately

14

AB, and C-glyc-AB metabolism in human liver microsomes was

determined. The metabolites formed in incubations were quanti-

ﬁed by LC with radiometric detection (method A) as described in

Section 2.7.

m

three-fold higher than that observed for the low-afﬁnity compo-

nent (Table 4).

1

4

On the other hand, following incubations of C-LAS34850

with liver microsomes, the formation of the radioactive peak m

a

1

4

14

3

.5.1. Kinetics of enzymatic metabolism of C-LAS34823 and C-

was observed in phosphate buffer and liver microsomes

only in the presence of NADPH-generating system or reduced

LAS34850

14

+

The kinetics of C-LAS34823 metabolism was investigated in

pooled human liver microsomes over the concentration range 0.5–

NADPH, but not with NADP (oxidized form). The fact that

liver microsomal protein was not essential for m

a

formation

5

00 mM. A single radioactive metabolite M1 was observed in all the

strongly suggests a non-enzymatic reaction. Furthermore, the

kinetics of C-LAS34850 metabolism was investigated in pooled

human liver microsomes over the concentration range 0.5–

1

4

radiochromatograms (Fig. 3). The kinetic proﬁle of M1 formation

exhibited substrate inhibition at LAS34823 concentrations greater

than 100

m

M, indicating that LAS34823 may inhibit its own

500

LAS34850 metabolites apart from m

concentration and the time course of m

with increasing LAS34850 concentrations, conﬁrming that its

formation was non-enzymatic and only NADPH-dependent

(Fig. 7B).

m

M in the presence of NADPH-generating system. No other

were observed at any

formation was linear

metabolism (Fig. 7A). Unfortunately, experimental data could not

be ﬁtted to any substrate inhibition model due to the severity of

the inhibition. Within LAS34823 concentration range between 2

a

and 100

mM, the Eadie–Hofstee plot of the kinetic data obtained

exhibited a biphasic pattern, suggesting that two enzymes of

different afﬁnity could be involved in the formation of M1 (Fig. 7A).

Although limited experimental data were available, data ﬁtting

was carried out assuming a bi-enzymatic Michaelis–Menten

model within this concentration range. The high-afﬁnity compo-

1

4

3.5.2. Kinetics of enzymatic metabolism of C-phe-AB

1

4

The esterase-mediated hydrolysis of C-phe-AB (5

mM) in

human liver microsomes (1 and 2 mg/ml) was investigated in the

absence of NADPH. Parallel incubations in buffer (pH 7.4) were also

carried out as a control of non-enzymatic hydrolysis. There were

slight differences in the formation of LAS34823 between human

nent exhibited apparent Km1 and Vmax1 values of 2.1

m

M and

2

0.1 pmol/min/mg protein, respectively. The low-afﬁnity compo-

nent exhibited apparent Km2 and Vmax2 values of 65.5

mM and

7

72

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

Table 4

1

4

14

Apparent kinetic parameters for the oxidative metabolism of C-LAS34823, C-phe-AB and C-glyc-AB. Test compounds (0.5–100

LAS34823) were incubated in phosphate buffer (pH 7.4) and with human liver microsomes in the presence and absence of NADPH.

mM

C-phe-AB and 2–100

m

M

C-

ꢁ1

Metabolite

M1

Test compound

System

K

m

(

m

M)

V

max (pmol/min/mg protein)

Clint

(

ml/min/mg protein)

K

h

(min

)

1

4

C-LAS34823

HLM + NADPH

K

m1 2.1

Vmax1 20.1

9.6

2.9

NA

m2 65.5

Vmax2 192.7

M2

C-phe-AB

Buffer

–

0.014

0.012

0.010

HLM–NADPH

HLM + NADPH

0.43

1.78

37.7

87.7

142.8

253.9

1

4

M3/m4

M4/m5

C-phe-AB

C-glyc-AB

HLM + NADPH

22.6

18.3

283.4

229.2

12.5

12.6

NA

1

4

C-phe-AB

C-glyc-AB

HLM + NADPH

4.6

5.8

106.0

100.4

23.0

17.4

NA

14

m2

C-glyc-AB

HLM + NADPH

0.4

328.3

812.5

0.022

NA, not applicable.

liver microsomes and the incubation buffer. In fact, the net

enzymatic hydrolysis of aclidinium at a microsomal protein

concentration of 2 mg/ml was less than 20% after 60 min of

incubation. These results suggest that the esterase(s) responsible

for the aclidinium hydrolysis have low catalytic activity in human

liver microsomes.

Moreover, at the highest protein concentration assayed (1 mg/ml)

and incubation times of 15 and 30 min, aclidinium disappearance

was also considerable due to the enzymatic formation of the

oxidative metabolites M3 and M4. As a compromise, an incubation

time of 15 min and 0.25 mg of microsomal protein/ml was

selected.

In terms of substrate depletion, the rapid non-enzymatic

hydrolysis of aclidinium compromised the selection of the

incubation time for further kinetic studies. Incubation times of

above 15 min led to a substrate depletion higher than 20%.

The inhibitory potential of the main hydrolysis metabolites

towards human esterases was of low relevance [4]. In addition,

human CYP1A2, 2A6, 2C8, 2C9, 2C19, 2D6, 2E1, 3A4/5, and A9/11

activities were not inhibited by LAS34850 in vitro, whereas

CYP2D6 was inhibited competitively by LAS34823 with an

inhibition constant of 15.5

mM [11]. Thus, taking into account

all these data, any potential interference of the hydrolysis

14

metabolites on the characterization of the kinetics of C-phe-

1

4

AB and C-glyc-AB is very unlikely.

1

4

The kinetics of C-phe-AB metabolism was investigated in

pooled human liver microsomes over the concentration range 0.5–

1

4

1

00

buffer and with human liver microsomes without NADPH-

generating system. The K in the incubation buffer was calculated

from the slope of the regression line (r = 0.9991) and presented a

mM. In parallel, C-phe-AB was also incubated in phosphate

h

2

ꢁ

1

value of 0.014 min

(Table 4). The major metabolite was M2

(

LAS34823) and following inspection of the Eadie–Hofstee plot

obtained in human liver microsomes, two different phases could

clearly be differentiated (Fig. 8A). At low aclidinium concentra-

tions, enzymatic hydrolysis may occur via esterases, and it was

assumed that this process could be well described using the typical

Michaelis–Menten equation. However, non-enzymatic hydrolysis

becomes predominant at high substrate concentrations, leading to

a constant value of the

enzymatic hydrolysis rate constant (K

were calculated using the mixed model described above (Section

.10). The formation of M2 (LAS34823) without NADPH exhibited

apparent K and Vmax values of 0.43 M and 37.7 pmol/min/

mg protein, respectively. In contrast, higher apparent K and Vmax

values (1.78 M and 253.9 pmol/min/mg protein, respectively)

were obtained in the presence of NADPH. The apparent K values of

C-phe-AB in human liver microsomes with and without NADPH

v/[S] ratio, which is represented by the non-

h

). The kinetic parameters

2

m

h

1

4

ꢁ1

were 0.012 and 0.010 min , respectively. The calculated intrinsic

clearance (Vmax/K ) with and without NADPH was 87.7 and

42.8 l/min/mg protein, respectively, which is consistent with an

m

1

m

additional NADPH-dependent metabolism of AB apart from

enzymatic hydrolysis.

The formation of the minor metabolites M3 and M4 was only

observed in the presence of human liver microsomes and NADPH-

generating system, conﬁrming that their formation was enzymatic

and NADPH-dependent. The kinetics of the formation of metabo-

lites M3 and M4 followed the Michaelis–Menten model (Fig. 8B).

1

4

Fig. 8. Eadie–Hofstee plots of (A) M2 formation from C-phe-AB with (ﬁlled

diamonds) and without (open circles) NADPH and (B) M3 and M4 formation from

1

4

C-phe-AB in human liver microsomes. Symbols represent observed data (mean of

duplicate determinations); solid lines represent rates calculated by non-linear

regression.

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

773

The formation of M3 exhibited apparent K

2.6 M and 283.4 pmol/min/mg protein, respectively. The for-

mation of M4 exhibited apparent K and Vmax values of 4.6 M and

06 pmol/min/mg protein, respectively. The enzymatic intrinsic

m

and Vmax values of

calculated parameters, the proﬁle was best ﬁtted to a mixed model

assuming non-enzymatic degradation as well as to an enzymatic

process (Michaelis–Menten), as described above. Assuming this

2

m

1

model, the apparent hydrolysis rate constant K

h

was very rapid

half-life of 32 min. The

and Vmax were 0.40 mM and 328.3 pmol/mg protein/

ꢁ1

clearance values (Clint) for M3 and M4 were 12.5 and 23.0

mg protein/min, respectively (Table 4).

m

l/

(0.022 ꢄ 0.002 min ), representing

apparent K

a

m

min, respectively. The intrinsic clearance was calculated to be

812.5 ml/min/mg protein. The formation of m4 was enzymatic and

NADPH-dependent. Its formation followed Michaelis–Menten kinet-

1

4

3

.5.3. Kinetics of enzymatic metabolism of C-glyc-AB

14

The formation of

C-glyc-AB metabolites in human liver

microsomes was conducted in the incubation conditions estab-

ics with K

m

and Vmax values of 18.3 mM and 229.2 pmol/min/

14

lished for C-phe-AB. Parallel incubations were also conducted in

the incubation buffer as a control of non-enzymatic hydrolysis. The

principal metabolite was the hydrolysis metabolite m3 (LAS34850)

and was generated non-enzymatically as described above. Thus,

m3 formation was linear in incubation buffer and a similar proﬁle

was also obtained in liver microsomes without NADPH. Interest-

ingly, m3 (LAS34850) formation was lower after incubation in

human liver microsomes with NADPH. This was a general

observation for all concentrations tested (Fig. 9A). As an example,

mg protein, respectively, which are practically identical to those

1

4

found for M3 after incubation of C-phe-AB, conﬁrming that M3 and

m4 are actually the same metabolite. The generation of metabolite

m5 was also enzymatic and NADPH-dependent. The apparent

Michaelis–Menten parameters

m

K and Vmax were 5.8 mM and

100.4 pmol/min/mg protein, respectively, also similar to those found

for M4 (Table 4).

14

a

The formation of m was only observed at high C-glyc-AB

concentrations and required the presence of a NADPH-generating

system, as already described for LAS34850. The formation of

metabolites m1 and m5 was negligible, thus, precluding any

attempt of identiﬁcation.

14

the rates of formation at a C-glyc-AB concentration of 5

53 and 171 pmol/min/mg protein, without and with NADPH,

respectively.

mM were

2

The formation of m2 in incubations with human liver

microsomes and NADPH followed a clear biphasic kinetic proﬁle,

which is in agreement with the participation of at least two

different processes in its formation (Fig. 9B). Following visual

inspection of the Eadie–Hofstee plot and the standard errors of

3.6. Identiﬁcation of P450 isoforms responsible for the metabolism of

1

4

14

C-phe-AB and C-glyc-AB

The identiﬁcation of the P450 enzymes responsible for the

1

4

14

formation of the oxidative metabolites of C-LAS34823, C-

1

4

14

LAS34850, C-phe-AB, and C-glyc-AB was conducted using

three different approaches: (a) incubations of test compounds

with commercially available human cDNA-expressed P450 and

FMO isoenzymes, (b) effect of selective chemical inhibitors of

P450 isoenzymes in human liver microsomes, and (c) effect of

selective antibodies against P450 isoenzymes in human liver

microsomes.

1

4

3

.6.1. Identiﬁcation of P450 isoforms responsible for the C-

LAS34823 metabolism

1

4

The incubations of C-LAS34823 with human recombinant

P450 (25 pmol/ml) and FMO (0.2 mg/ml) isoforms were carried

out at a LAS34823 concentration of 5

mM. The formation of M1

was catalyzed only by human recombinant CYP2D6 with

practically complete conversion of LAS34823 into M1. Additional

metabolites were not observed in the presence of other P450

isoforms or ﬂavin-containing monooxygenases (FMO1, FMO3,

and FMO5). The kinetic parameters of M1 formation in the

presence of recombinant human CYP2D6 were determined,

corrected by P450 content (2 pmol/ml) and incubation time

(

20 min), in order to ensure linear conditions and low substrate

depletion. An important inhibitory effect on M1 formation was

observed at substrate concentrations above 5 M and complete

inhibition was obtained at the highest LAS34823 concentrations

assayed (100–200 M). Within the LAS34823 concentration

range 0.5–5 M, the formation of M1 by CYP2D6 followed

Michaelis–Menten kinetics with apparent K and Vmax values

of 4.9 M and 22.1 pmol/min/pmol P450, respectively. These

m

results are in agreement with the high-afﬁnity enzyme observed

in human liver microsomes (Section 3.5.1). Incubations at higher

LAS34823 concentrations to identify the possible low-afﬁnity

component observed in human liver microsomes were not

conducted, as it was not considered relevant due to the low

systemic LAS34823 concentrations (<0.5 nM) observed in clinical

trials [12]. Following incubation in the presence of different

chemical and antibody inhibitors, the formation of M1 was

strongly inhibited by quinidine (CYP2D6 inhibitor) and by the

antibody against CYP2D6 (Table 5).

1

4

Fig. 9. (A) Formation of metabolite m3 (LAS34850) from C-glyc-AB in incubation

buffer (ﬁlled triangles), human liver microsomes without NADPH, and human liver

1

4

microsomes with NADPH. (B) Eadie–Hofstee plot of m2 from C-glyc-AB in human

liver microsomes. Symbols represent observed data (mean of duplicate

determinations); solid lines represent rates calculated by non-linear regression.

7

74

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

Table 5

Summary of in vitro human P450 reaction phenotyping of aclidinium bromide metabolism obtained in recombinant human P450 isoforms and human liver microsomes.

Metabolite

Substrate

Recombinant human P450 isoforms

n-fold basal activity)

Inhibition (%)

Chemical

(

Antibody

1

4

m1

m2

C-glyc-AB

–

No effect

" 2D6 (ꢃ2.6-fold at 5 and 25

m

M)

Ketoconazole: 58%

Quercetin: 25%

Quinidine: >80%

–

CYP3A4: 52%

"

3A4 (ꢃ1.4-fold at 5 and 25

mM)

1

4

M1

C-LAS34823

C-phe-AB

" 2D6 (>40-fold)

" 2D6 (>4.3-fold)

CYP2D6: >80%

–

1

4

M2

m3

C-phe-AB

C-glyc-AB

C-phe-AB

" 2D6 (2.2-fold at 5

# 2D6 (0.7-fold at 5

No effect

m

M)

Ketoconazole: 28%

No effect

CYP3A4: 23%

No effect

m

a

–

M3/m4

–

Ketoconazole: >77%

Quercetin: 44%

Quinidine: 20%

Ketoconazole: >35%

CYP3A4: >70%

CYP2D6: 31%

14

C-glyc-AB

–

CYP2D6: 34%

CYP3A4: 32%

1

4

M4/m5

C-phe-AB

C-glyc-AB

" 2D6 (>6.5-fold)

Quinidine: >77%

Quinidine: >48%

CYP2D6: 62%

CYP2D6: 40%

" 2D6 (>1.6-fold at 5

m

M, >3.2-fold at 25

mM)

"

1A1 (>1.8-fold at 25 mM)

See Sections 2.5 and 2.6 for incubation details.

: Metabolite not detected at any of the concentrations assayed.

M5/m6 and M6/m7 were not detected in the experiments.

–

"

/#: increased/decreased activity with human recombinant CYP compared to control incubation (insect control). The remaining human rCYPs did not have any inﬂuence on

metabolite formation.

No effect: enzyme activity was not signiﬁcantly inhibited (<20% inhibition) against control incubations.

1

4

3

.6.2. Identiﬁcation of P450 isoforms responsible for the C-phe-AB

metabolism

The incubation of

25 pmol/ml) and FMO (0.2 mg/ml) isoforms revealed that the

The formation of the hydrolysis metabolite m3 (LAS34850) was

similar for all rCYPs assayed and control incubations (approxi-

mately 12 pmol/min/pmol CYP), except in the case of CYP2D6,

where a slight reduction was observed (9.5 pmol/min/pmol CYP).

Metabolite m3 formation was not signiﬁcantly affected by any of

the human P450 inhibitors and inhibitory antibodies, indicating

that its formation was not P450-dependent.

1

4

C-phe-AB (5

mM) with human P450

(

formation of M1 (hydroxylated LAS34823) and M4 was only

catalyzed by CYP2D6. Remarkably, the rate of formation of

M2 (LAS34823) was also higher with human rCYP2D6

(

2.92 pmol/min/pmol CYP) compared to the other rCYP isoforms

1.3–1.5 pmol/min/pmol CYP). Unfortunately, metabolite M3

Metabolite m1 was not observed in the presence of any of the

rCYP isoforms assayed, most probably because the formation of

this metabolite is dependent on m2 and/or m4 concentrations,

which were always low. Metabolite m2 was formed to a similar

extent in insect control incubations and with most human rCYPs

was not observed in the incubation samples at the assay

concentrations. Therefore, further inhibition experiments with

chemical P450 inhibitors and antibodies in liver microsomes

1

4

were studied at a C-phe-AB concentration of 25

m

M. At this

(approx. 0.38 and 1.00 pmol/min/pmol CYP at 5 and 25

respectively), with the exception of human CYP2D6 (1.02 pmol/

min/pmol CYP at 5 M and 2.5 pmol/min/pmol CYP at 25 M) and

CYP3A4 (0.49 pmol/min/pmol CYP at 5 M and 1.40 pmol/min/

pmol CYP at 25 M), suggesting that these enzymes may be

mM,

concentration, the formation of M2 was slightly inhibited (28%)

by the presence of ketoconazole (CYP3A4 inhibitor) and the

formation of M3 was decreased by ketoconazole and slightly by

quinidine. The formation of M4 was catalyzed by human

rCYP2D6 and totally blocked by quinidine in human liver

microsomes. Interestingly, human CYP1A1, which appears to be

expressed in human liver at low levels [13,14], could be also

involved in the formation of M4. This means that the oxidative

metabolism in the human lungs from smokers could be

increased, since smoking-induced elevated levels of CYP1A1

have been observed [15].

m

involved in its formation. The formation of m2 was inhibited

substantially by ketoconazole (CYP3A4 inhibitor) and slightly by

quercetin (CYP2C8 inhibitor). Moreover, the formation of m2 in

human liver microsomes was slightly affected by the presence of

a

CYP3A4 antibodies. Metabolite m was observed to a similar extent

in incubations with insect control and human rP450 isoforms

(from 0.8 to 1.3 pmol/min/pmol CYP), which is in accordance with

the formation of LAS34850 and further chemical reduction by

NADPH, because this co-factor was always present in the

incubations. The formation of metabolite m4 was not catalyzed

by any human CYP isoform at the two different substrate

concentrations, although these results should be taken with

caution due to the similar retention times of metabolites m4 and

The remaining chemical inhibitors had no particular inﬂuence

14

on C-phe-AB metabolism. These results were also conﬁrmed in

the studies conducted with selective P450 antibodies. M2

formation was slightly affected by the presence of anti-CYP3A4,

whilst M3 and M4 formation were inhibited by CYP3A4 and

CYP2D6 inhibitory antibodies, respectively (Table 5).

a

m . Curiously, metabolite m4 was not observed in incubations with

1

4

14

3

.6.3. Identiﬁcation of P450 isoforms responsible for the C-glyc-AB

recombinant CYP isoforms at any of the concentrations of C-glyc-

AB assayed. In contrast, m4 was detected in inhibition studies with

human liver microsomes. In these studies, some degree of

inhibition was observed in the presence of human CYP3A4 and

2D6 antibodies, although conclusive results could not be obtained

due to the low activity observed.

metabolism

14

Incubations of C-glyc-AB with recombinant human P450

isoforms (rP450) were conducted at two different substrate

concentrations (5 and 25

mM), in order to monitor the metabolite

formation adequately. Further studies on the effect of different

P450 chemical inhibitors and antibodies were studied at a single

The formation of the hydroxylated metabolite m5 (M4) was

14

C-glyc-AB concentration of 25

are summarized in Table 5.

m

M. The results of these studies

mainlycatalyzedbyCYP2D6(1.08 pmol/min/pmolCYPat25

observed in C-phe-AB incubations. This metabolite was also

m

M), as

1

4

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

775

formed to a lesser extent in the presence of enzyme CYP1A1

0.60 pmol/min/pmol CYP at 25 M). CYP1A1 involvement in the

formationofM4 wasnot observedafter C-phe-AB incubation, most

probably due to the low concentration assayed (5 M). Metabolite

m5 formation in human liver microsomes was inhibited only by

quinidine (CYP2D6 inhibitor) and human CYP2D6 antibodies, which

is in agreement with the results obtained above (Table 5).

The rapid generation of m2 in liver microsomes and hepato-

cytes is difﬁcult to interpret. According to its chemical structure,

the most plausible explanation for its formation is related to the

primary oxidation of m4/M3 and further hydrolysis, as represented

in Fig. 1. Indeed, the formation of m2 followed a biphasic kinetic

proﬁle in incubations with human liver microsomes and NADPH,

consistent with the participation of two different processes

(

m

14

m

(

Fig. 9B).

The apparent K

for the hydrolysis metabolite M2, suggesting a faster hydrolysis of

its tentative primary metabolite M into m2 and M2 (LAS34823).

h

obtained for m2 was higher than that observed

4

. Discussion

f

The non-enzymatic hydrolysis of aclidinium bromide (AB)

Furthermore, the enzymatic origin of m2 was demonstrated by its

formation in the presence of rCYP2D6 and in lower amounts by

rCYP3A4. The involvement of CYP2D6 and CYP3A4 was further

conﬁrmed by inhibition studies. The fact that m2 formation was

more affected by CYP3A4 inhibitors could be explained by the

relative content of these enzymes in the human liver, with average

values of 30% (CYP3A4) and 2.5% (CYP2D6) of the total P450

content [16]. The slight inhibition caused by quercetin could be

explained by its lack of selectivity as a CYP2C8 inhibitor, as

observed in the incubation buffer (pH 7.4) was high and accounted

for approximately 45% of overall metabolism in incubations

conducted with liver microsomes. The rate of enzymatic hydrolysis

of both radiolabeled forms, measured in the absence of NADPH,

was higher in of rabbits and dogs, followed by mice. Remarkably,

the formation rate of the hydrolysis metabolites in rat and human

liver microsomes was similar to that obtained in buffer incubation,

suggesting a low esterase activity in these liver subcellular

fractions. This observation has already been done in a previous

study using human liver microsomes [4].

described by Obach [17]. The low apparent K

m

and the high

intrinsic clearance suggest that this metabolic pathway would be

relevant in vivo.

The net oxidative metabolism obtained by subtraction of non-

NADPH and NADPH-dependent metabolism was similar across all

species examined, but its extent was far lower compared to the

hydrolysis process. In hepatocytes, the overall rate of aclidinium

metabolism followed the rank order: rabbit > mouse > dog >

rat ꢃ human. Although the enzymatic hydrolysis of aclidinium

in hepatocytes cannot be differentiated from its overall metabo-

lism, the metabolism pattern was similar to the obtained in liver

microsomes.

The minor aclidinium metabolites M3/m4, M4/m5, M5/m6, M6/

m7, and m1 accounted for less than 5% of total metabolism in liver

microsomes and were not detected in hepatocytes, suggesting that

these metabolites are actually intermediate metabolites that are

likely to be further metabolized. As it was expected, the formation

of these intermediate metabolites was similar following incuba-

1

4

14

tions of C-phe-AB and C-glyc-AB with liver microsomes. The

origin of metabolite M3/m4 is an intriguing issue that cannot be

completely addressed for the moment. However, the loss of a

thiophene ring from aclidinium could be explained by thiophene

oxidation (hydroxylation or S-oxidation) and further ring opening

in a similar manner that has been described for compounds such as

suprofen and tienilic acid [18]. This complex process would mean

that the minor metabolites M5/m6 and M6/m7 could actually be

precursors for the formation of M3/m4. An interesting observation

was that the primary metabolite M3/m4 was generated to a greater

14

The major radioactive metabolites of C-phe-AB and C-glyc-

AB detected in liver microsomes and hepatocytes of all species

were the alcohol and acid metabolites, M2 (LAS34823) and m3

(

LAS34850), respectively. These two hydrolysis metabolites are

devoid of in vitro antimuscarinic activity and show no relevant

bronchodilatory activity in preclinical models [3].

The formation of M2 (LAS34823) was slightly higher in human

14

liver microsomes with NADPH at low C-phe-AB concentrations.

Inspection of Eadie–Hofstee plots conﬁrmed the differentiation

between enzymatic and non-enzymatic hydrolysis, as previously

observed in human plasma [4]. The slight inhibition of LAS34823

formation produced by ketoconazole and CYP3A4 antibody may be

explained by the inhibition of the precursor metabolite M3/m4, the

formation of which was also mediated by CYP3A4. In contrast, m3

formation (LAS34850) was lower after incubation in human liver

microsomes with NADPH, which could be related to the increased

oxidative pathway of primary metabolites such as M3/m4, whose

hydrolysis leads to the formation of metabolites other than m3. On

the other hand, m3 formation was similar with all recombinant

CYPs assayed with the exception of CYP2D6, where small amounts

were formed. This fact conﬁrms that other metabolic routes

involving the thiopheneglyoxylic moiety of aclidinium are in

competition with the hydrolysis process.

extent than its oxidized metabolite M

f

, which was only detected by

MS. In contrast, the metabolite m2 (hydrolytic metabolite of M

f

)

was always formed to a greater extent than m1 (hydrolytic

metabolite of M3/m4), strongly supporting faster hydrolysis of the

f

primary metabolite M compared to M3/m4. The formation of the

p-hydroxylated metabolite of aclidinium (M4/m5) was catalyzed

by human CYP2D6, in analogy to the formation of M1. Metabolite

m1 (hydroxyl-2-thienylacetic acid) was observed at very low

levels in liver microsomes of mouse, rabbit, and human, which

could be associated with the hydrolysis of its primary metabolite

M3/m4. In this study, one of the minor metabolic routes identiﬁed

involved the oxidation of the thiophene rings of aclidinium (M5/

m6 and M6/m7). The formation of these metabolites could be

related to the formation of potential reactive metabolites.

However, the recovery of total radioactivity in the supernatans

14

Other major metabolites obtained in incubations of C-phe-AB

of incubations of C-phe-AB, C-glyc-AB, C-LAS34823, and C-

LAS34850 in human liver microsomes was greater than 85%, which

suggests that this metabolic route would be a minor process.

Additional investigations were also performed although the results

of these studies are outside the scope of the current study.

14

and C-glyc-AB were identiﬁed as hydroxylated LAS34823 (M1)

and 2-thiopheneglyoxylic acid (m2), respectively.

In humans, the formation of M1 from LAS34823 was catalyzed

by CYP2D6 and a substrate inhibition proﬁle was apparent at

higher concentrations, which is in agreement with a previous in

vitro study where this compound was shown to be a moderate

inhibitor of the catalytic activity of human CYP2D6 with an

1

4

The incubation of C-LAS34850 in human liver microsomes

revealed some interesting features. An artifact was generated in

the incubation buffer that was identiﬁed as the compound di-2-

thienyl ketone. The biomimetic oxidative formation of a similar

compound has been also described for denaverine hydrochloride

i

apparent K of 15 mM [11]. Following incubation of AB, the

formation of M1 could be explained by two different routes: (a)

direct hydroxylation of the hydrolysis product LAS34823 (M2) and

a

[19]. The formation of metabolite m was only observed in

(b) hydrolysis of metabolite M4/m5.

incubations where the NADPH (reduced form) was present with or

7

76

J.J. Albert ı´ et al. / Biochemical Pharmacology 81 (2011) 761–776

clo[2.2.2]octane bromide (aclidinium bromide). J Med Chem 2009;52:5076–

2.

3] Sentellas S, Ramos I, Albert ı´ J, Salv a` M, Ant o´ n F, Miralpeix M, et al. Aclidinium

bromide, a new, long-acting, inhaled muscarinic antagonist: in vitro plasma

inactivation and pharmacological activity of its main metabolites. Eur J Pharm

Sci 2010;39:283–90.

4] Albert ı´ J, Martinet A, Sentellas S, Salv a` M. Identiﬁcation of the human enzymes

responsible for the enzymatic hydrolysis of aclidinium bromide. Drug Metab

Disp 2010;38:1202–10.

5] Placidi L, Scott EC, de Sousa G, Rahmani R, Placidi M, Sommadossi JP. Inter-

species variability of TNP-470 metabolism, using primary monkey, rat, and dog

cultured hepatocytes. Drug Metab Dispos 1997;25:94–9.

[6] Hewitt NJ, Fischer T, Zuehlke U, Oesch F, Utesch D. Metabolic activity of fresh

and cryopreserved cynomolgus monkey (Macaca fascicularis) hepatocytes.

Xenobiotica 2000;30:665–81.

7] Bedn a´ r P, Lemr K, Bart a´ k P, Sevc ı´ k J, Hlav a´ c J, St y´ skala J, et al. Capillary

electrophoresis/mass spectrometry: a promising tool for the control of some

physiologically hazardous compounds. I-derivatives of 3-quinuclidinol. J Mass

Spectrom 2002;37:1213–8.

8] Pati S, Losito I, Palmisano F, Zambonin PG. Characterization of caffeic acid

enzymatic oxidation by-products by liquid chromatography coupled to elec-

trospray ionization tandem mass spectrometry. J Chromatogr A 2006;1102:

without liver microsomes, which would be compatible with a non-

enzymatic reduction process of the hydroxyl moiety by NADPH.

9

[

a

Furthermore, the formation of m was linear with increasing

LAS34850 concentrations, conﬁrming that its formation was non-

enzymatic. On the other hand, it is important to note that the polar

metabolites m1 and m2 were not generated during the incubation

[

14

of C-LAS34850 with human liver microsomes and hepatocytes.

These observations conﬁrm that these metabolites can be formed

only from the parent compound but not from m3 (LAS34850),

which is the main circulating metabolite in humans [12].

In summary, the present study indicates that aclidinium

hydrolysis is the main biotransformation route in liver microsomes

and hepatocytes of mouse, rat, rabbit, dog, and human, with rabbit

and dog exhibiting the highest rates of enzymatic hydrolysis.

Oxidative metabolism in liver microsomes and hepatocytes was

substantially lower when compared to ester hydrolysis. The

metabolite proﬁles were similar across all ﬁve species examined

and up to ﬁve primary metabolites and eight secondary

metabolites were characterized. The alcohol metabolite

[

184–92.

9] Ma YC, Kim HY. Determination of steroids by liquid chromatography/mass

spectrometry. J Am Soc Mass Spectrom 1997;8:1010–20.

[

10] Volmer DA, Hui JPM. Rapid determination of corticosteroids in urine by

combined solid phase microextraction/liquid chromatography/mass spec-

trometry. Rapid Commun Mass Spectrom 1997;11:1926–33.

(

LAS34823) is transformed to a single hydroxylated metabolite

that is subsequently glucuronidated. In contrast, the acid

metabolite (LAS34850) was not metabolized enzymatically in

vitro. The oxidative conversion of AB and its alcohol metabolite

was primarily catalyzed by human CYP3A4 and CYP2D6, respec-

tively.

[11] Almirall SA, data on ﬁle.

[12] Jansat JM, Lamarca R, Garcı

´

a-Gil E, Ferrer O. Safety and pharmacokinetics of

single doses of aclidinium bromide, a novel long-acting, inhaled muscarinic, in

healthy subjects. J Clin Pharmacol Ther 2009;47:460–8.

[

13] Stiborova M, Martınek V, Rydlova H, Hodek P, Frei E, Sudan I. Is a potential

carcinogen for humans: evidence for its metabolic activation and etoxication

by human recombinant cytochrome P450 1A1 and liver microsomes. Cancer

Res 2002;62:5678–84.

´

Acknowledgments

14] Drahushuk AT, McGarrigle BP, Larsen KE, Stegeman JJ, Olson JR. Detection

of CYP1A1 protein in human liver and induction by TCDD in precision-

cut liver slices incubated in dynamic organ culture. Carcinogenesis 1998;

We would like to acknowledge the technical support provided

by Francisco Jimenez Berbell. We also thank Josep M. Huerta,

Department of Computational and Structural Drug Discovery

1

9:1361–8.

15] Smith GBJ, Harper PA, Wong JMY, Lam MSM, Reid KR, Petsikas D, et al. Human

lung microsomal cytochrome P4501A1 (CYP1A1) activities. Impact of smoking

status and CYP1A1, Aryl hydrocarbon receptor, and glutathione S-transferase

M1 genetic polymorphisms. Cancer Epidemiol Biomarkers Prev 2001;10:839–

1

(

Almirall, S.A.), for analyzing the H NMR spectrum of metabolite

M1.

53.

References

[

16] Rendic S, DiCarlo FJ. Human cythocrome P450 enzymes: a status report

summarizing their reactions, substrates, inducers and inhibitors. Drug Metab

Rev 1997;29:413–580.

[17] Obach R. Inhibition of human cytochrome P450 enzymes by constituents of St.

John’s Wort, an herbal preparation used in the treatment of depression. J

Pharmacol Exp Ther 2000;294:88–95.

[

1] Gavald a` A, Miralpeix M, Ramos I, Otal R, Carre n˜ o C, Vi n˜ als M, et al. Characteri-

zation of aclidinium bromide, a novel inhaled muscarinic antagonist, with long

duration of action and a favorable pharmacological proﬁle. J Pharmacol Exp

Ther 2009;331:740–51.

[

2] Prat M, Fern a´ ndez D, Buil MA, Crespo MI, Casals G, Ferrer M, et al. Discovery of

novel quaternary ammonium derivatives of (3R)-quinuclidinol esters as po-

tent and long-acting muscarinic antagonists with potential for minimal sys-

temic exposure after inhaled administration: identiﬁcation of (3R)-3-

[18] O’Donnell JP, Dalvie DK, Kalgutkar AS, Obach RS. Mechanism-based inactiva-

tion of human recombinant P450 2C9 by the nonsteroidal anti-inﬂammatory

drug suprofen. Drug Metab Dispos 2003;31:1369–77.

[19] Smolinka K, G o¨ ber B. Biomimetic oxidation of denaverine hydrochloride. Eur J

Org Chem 1999;3:679–83.

{[hydroxy(di-2-thienyl)acetyl]oxy}-1-(3-phenoxypropyl)-1-azoniabicy-

Article Doi

In vitro liver metabolism of aclidinium bromide in preclinical animal species and humans: Identification of the human enzymes involved in its oxidative metabolism

DOI: 10.1016/j.bcp.2010.12.013

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.bcp.2010.12.013

Full text of DOI:10.1016/j.bcp.2010.12.013

Products guided by the article

R&D Labs maybe for 4075-59-6

Relevant to this article

Hot Product