Department of Drug Safety and Disposition, Eisai Research
Institute, Wilmington, Massachusetts
 |
Introduction |
N-Methyl-[1-[1-(2-fluorophenethyl)piperidine-4-yl]-1H-indol-6-yl]
acetamide (E21012) is currently under development
as a central acting muscle relaxant. This 5-hydroxytryptamine (5-HT)
receptor antagonist potently binds to the 5-HT1A
and 5-HT2 receptors with
IC50 in low nanomolar concentration ranges. It
reduced muscle tension in a series of physiological and pharmacological
experiments. Despite the complexity of 5-HT system in stress, anxiety,
depression, and muscle contraction (Saxena 1995
; Pauwels 2000), the
results from experimental animal models are largely consistent with
those from healthy volunteers (Graeff et al., 1996).
Many of the drugs targeting the central nervous system (CNS) are either
substrates or inhibitors of cytochromes P450 (P450s), a superfamily of
enzymes responsible for the metabolism of the majority of drugs and
xenobiotics (Harvey and Preskorn, 1996; Richelson, 1997). This is
particularly true for 5-HT- or serotonin-selective reuptake inhibitors
(SSRIs) including sertraline and fluvoxamine (Harvey and Preskorn,
1996; van Harten, 1996). 5-HT1A antagonists may
synergistically act with SSRIs, thus markedly increase the SSRI-treatment efficacy (Hjorth et al., 2000). The drug-drug
interactions after clinical administration of SSRIs have been, in fact,
extensively studied and mostly attributed to the inhibition of P450
enzymes (Vella and Sayegh, 1998; Larsen et al., 1999; DeSilva et al., 2001). The P450 forms responsible for the metabolism of CNS drugs may
exhibit their own characteristics. For example, the well known polymorphic P450 members, especially CYP2C19 and CYP2D6, are among the
most common metabolizers of CNS drugs, besides CYP3A4 (Roddam et al.,
2000; Sata et al., 2000; Hsieh et al., 2001). Moreover, the profile of
P450 expression in the brain is different compared with that in the
liver. Although the total P450s in brain might be low, individual
members of P450s, such as CYP2D6, might be relatively abundant (Siegle
et al., 2001). Therefore, in addition to common liver metabolism, CNS
drugs may potentially undergo the metabolism at the target tissues.
Because of the involvement of the polymorphic P450 forms and possible
target tissue disposition, the risk of drug-drug interaction for CNS
drugs, potentially varying among different individuals, is complicated
to predict and tends to be higher (Larsen et al., 1999; DeSilva et al.,
2001; Yu et al., 2001).
Metabolic studies of new chemical entities in vitro are currently an
important approach for the prediction of the potential drug-drug
interactions in clinic. In contrast to SSRIs, the interactions between
5-HT antagonists and P450s have received little study either in vitro
or in vivo. Therefore, to gain the knowledge of the potential drug-drug
interaction for 5-HT antagonists, and also as part of the preclinical
safety evaluation, P450-mediated E2101 metabolism was studied in vitro
with emphases on the potential enzyme inhibition and induction. A
variety of enzyme preparations, including recombinant enzymes, liver
subcellular fractions, and primary hepatocytes were used in the study.
The results of this present study are expected to be useful for the
prediction of potential drug-drug interactions for future clinical
applications of E2101.
| |
Experimental Procedures |
Materials.
Chemicals
ER2101, or
N-methyl-[1-[1-(2-fluorophenethyl)piperidine-4-yl]-1H-indol-6-yl]
acetamide, was obtained from Tsukuba Research Laboratories of Eisai
Co., Ltd. (Ibaraki, Japan). (±)-Bufuralol, (±)-1'-hydroxybufuralol,
6-hydroxychlorozoxazone, S-mephenytoin, 4'-hydroxy-S-mephenytoin, and monohydroxylated warfarin
metabolites (6-, 7- and 10-hydroxywarfarin) were purchased from Gentest
Corp. (Woburn, MA). Chlorzoxazone, coumarin, albendazole,
R-propranolol, 4'-chlorowarfarin, rifampicin, NADPH, TRIZMA,
magnesium chloride, potassium phosphates, ketoconazole, quinidine, and
rac-warfarin were obtained from Sigma-Aldrich (St.
Louis, MO). 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) was
purchased from National Cancer Institute Chemical Carcinogen Reference Standard Repositories (Midwest Research Institute, Kansas City, MO). Optically pure R- and S-warfarin were
prepared from the racemic mixture by the differential crystallization
method (West et al., 1961). The purity of warfarin enantiomers was at least 98% as determined by chiral HPLC, mass spectrometer, and NMR
analyses. All solvents used for the HPLC analyses were HPLC grade.
Enzymes and hepatocytes.
Pooled human liver microsomal preparations and S9 fractions were
purchased from Gentest Corp. The insect microsomal preparations containing cDNA-expressed recombinant human P450s, and the control insect microsomal preparations were also purchased from Gentest Corp.
Primary cultures of human hepatocytes from three female Caucasian
donors (ages 55, 56, and 79) and the culture media were purchased from
In Vitro Technologies Inc. (Baltimore, MD). These donors did not have
recorded liver diseases and damages. Two of the donors were smokers
(ages 55 and 79).
Others.
Polyclonal goat anti-human CYP1A1/2 and anti-human CYP3A4 antibodies,
and monoclonal mouse anti-human CYP2D6 antibodies were obtained from
Gentest Corp. Polyclonal rabbit anti-human CYP2C19 antibodies were from
Research Diagnostics Inc. (Flanders, NJ). All secondary antibodies were
from Sigma-Aldrich. The electrophoresis apparatus and accessories were
obtained from Bio-Rad (Hercules, CA) or Pierce Chemical (Rockford, IL).
The antibiotics (streptomycin/penicillin) and buffers for the
electrophoresis were purchased from Invitrogen (Carlsbad, CA) or
Bio-Rad.
Metabolite and Metabolic Enzyme Identification.
The metabolism of E2101 was determined by the disappearance of E2101 or
the appearance of E2101 metabolites in the reaction mixtures compared
with the respective controls. E2101 was incubated in the reconstituted
in vitro reaction systems containing pooled human liver S9 fractions,
microsomal preparations, or recombinant hepatic P450 forms including
CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4. The reaction
mixture (total volume 250 µl) contained 0.5 mg of liver microsomal
protein, 50 or 25 pmol of recombinant P450, and E2101 (50-80 µM) in
50 mM Tris buffer containing 15 mM MgCl2 (pH
7.4). After being incubated at 37°C with gentle shaking for 1 min,
the reaction was initiated by adding 25 µl of NADPH solution (20 mg/ml) and carried out for 1 h. The reaction was terminated by
quickly cooling the test tube on ice, followed by the addition of equal
volume of 100% methanol. The samples were centrifuged in a desktop
centrifuge at 14,000 rpm for 5 min, and the supernatants were filtered
through syringe filters (13 mm, 0.45 µ). The filtrates were analyzed
using LC/MS (ion trap) or LC/MS/MS.
P450 Inhibition Assays.
The incubations were carried out in test tubes (12.0 × 75 mm). The incubation mixtures (250 µl) contained 0.5 mg of HLM
protein, 50 (CYP1A2 and CYP2E1) or 25 pmol (CYP2C9, CYP2C19, CYP2D6,
and CYP3A4) of recombinant human P450s, 10 or 50 µM E2101, probe
substrate (R-, S-warfarin, chlorzoxazone,
S-mephenytoin, or bufuralol) at different concentrations,
and 0.5 mg of NADPH in 50 mM Tris buffer containing 15 mM
MgCl2 (pH 7.4). E2101 was added to the reaction mixture 5 min prior to the addition of the probe substrate. After being
incubated in a 37°C waterbath with gentle shaking for 1 min, the
reaction was initiated by adding 25 µl of NADPH solution (20 mg/ml)
and carried out for 15 to 60 min. The reaction mixture containing only
the probe substrate was used as the control. After the incubation, the
reaction mixture was extracted by mixing with 250 µl of methanol
containing the appropriate internal standard (IS). After vortex mixing
and centrifuging in a desktop centrifuge at 14,000 rpm for 5 min, the
supernatant was filtered through a syringe filter (13 mm, 0.45 µ)
into a HPLC vial. The filtrates were analyzed using LC/UV/fluorimetric
detection and LC/MS/MS.
If inhibition was detected, the inhibitory potency was determined using
recombinant P450s. E2101 appeared to inhibit the activities of CYP2D6
and CYP2C19. Therefore, three concentrations of
S-mephenytoin (30, 80, and 200 µM) and bufuralol (5, 10, and 40 µM), and a range of concentrations (0-200 µM) of E2101 were
used for the construction of Dixon plots and simultaneously nonlinear
regression analyses (SNLR). The incubation conditions and sample
preparations were the same as previously described.
The quantification was based on the calibration curves, and the quality
control samples were applied to ensure the quality of the experiments.
Samples for the standard curves and quality control were prepared in
the similar manner as those of the reaction samples.
P450 Induction Assays.
Hepatocyte treatment and sample
preparation
Upon the reception of the primary culture of human hepatocytes in
6-well plates, the culture media containing streptomycin/penicillin were refreshed (2 ml/well). After being acclimatized in 5%
CO2 and 37°C overnight, the cells were treated
with the vehicle solution (negative control), the prototypic P450
inducers including TCDD (0.4 µM) for CYP1A, rifampicin (50 µM) for
CYP3A and possibly CYP2C (Feng et al., 1998; Xu et al., 2000;
Gerbal-Chaloin et al., 2001; Liu et al., 2001), and E2101 (10 and 50 µM) for 72 h. E2101 stock solution was added to the culture
media at a 1:400 ratio (v/v). The culture media and testing compounds
were replenished every 24 h. At the end of the treatment, the
cells were harvested into 1 ml of phosphate-buffered saline (PBS) after
being washed twice. The cells were precipitated by centrifugation and
resuspended in PBS (approximately 50 µl). The samples for
electrophoresis were prepared after mixing with 200 µl of Laemmli
buffer (62.5 mM Tris-HCl containing 2% SDS, 25% glycerol, 0.01%
bromophenol, pH 6.8) and rocked for 2 to 4 h. The samples were
later mixed with 10 µl of 2-mercaptoethanol and heated at 90°C for
10 min before electrophoresis.
Western immunoblotting analysis.
Proteins were resolved in a 12% SDS-polyacrylamide gel electrophoresis
gel using a mini gel apparatus at a constant voltage (60 mV/gel) for 70 to 80 min and transferred onto a polyvinylidene difluoride membrane
using a membrane-transferring unit at a constant voltage (60 mV/membrane) for 60 min. The membrane was blocked by 5% nonfat dried
milk blotting buffer (PBS containing 0.05% Tween 20) at 4°C
overnight. The membrane was rinsed with the blotting buffer and probed
by 1:1000 diluted anti-human P450 antibodies in 2.5% nonfat dried milk
blotting buffer for 1 h at room temperature. The membrane was
rinsed six times (10 min each time) with the blotting buffer and
exposed to 1:10,000 diluted secondary antibodies labeled with
horseradish peroxidase for 1 h at room temperature. After being
extensively rinsed with the blotting buffer, the membrane was exposed
to the substrate of peroxidase (enhanced chemiluminescence reagent).
P450 proteins were detected by the fluorescence using an X-ray developer.
Instrumentation.
Identification of metabolites and enzymes
LC/MS/MS systems were applied for the metabolite identification. The
operation conditions were described as follows.
LC/MS was performed with LCQ ion trap mass spectrometer (Thermo
Finnigan/GC & GC/MS Div., Austin TX). Hewlett Packard 1100 HPLC
system (Hewlett Packard GmbH, Waldbronn, Germany) consisted of a binary
pump, an autosampler, a column compartment unit, and an online variable
wavelength UV detector. The detector was monitored at 270 nm, which is
the UVmax of E2101 predetermined by scanning between 225 to 400 nm using a photodiode array detector (PDA). The
metabolites were separated on a Hewlett Packard Eclipse
C18 column (150 × 2.1 mm). The mobile
phases were 10 mM ammonium acetate at pH 4.8 (A), and acetonitrile (B).
The gradient (B) was 10% (0-6 min), 22% (12-22 min), 90% (25-30
min), and 10% (31 min and after). The flow rate was 0.25 ml/min.
Software Navigator (version 1.2, Thermo Finnigan/GC & GC/MS Div.) was
used to control the HPLC and MS and to acquire the data. The MS was
operated at positive electrospray ionization (ESI) with 5.2 kV
ionization potential and 220°C heated capillary temperature. The
product ion spectra were generated under 24 V collision energy, which
was optimized for the fragmentation of E2101.
LC/MS/MS was performed with Sciex API 2000 triple quadrupole mass
spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA):
The same HPLC system described above was coupled with a tandem mass
spectrometer. The metabolites were separated on a Supelco Discovery
C18 column (150 × 2.1 mm; Sigma-Aldrich),
and the mobile phase described previously was run at 1.0 ml/min with a
1:10 split. The gradient (B) was 20% (0-6 min), 60% (14-18 min),
95% (20-24 min), and 20% (25 min and after). The operation of the
HPLC and the MS/MS was controlled by MacChrom (version 1.6;
Applied Biosystems, Foster City, CA). The MS/MS was operated at
positive ESI with 5.0 kV ionization potential and 400°C ion source
temperature. The product ion spectra were generated applying
collision-induced dissociation (CID) with optimized ion optic parameter settings.
P450 Substrate Assays.
CYP1A2/2C9/3A4-mediated warfarin 6-, 7-, 8-, and 10-hydroxylation were
as described by Brian et al., 1990; Rettie et al., 1992; and Zhang et
al., 1995. The system consisted of a Hewlett Packard 1100 HPLC
system with a binary pump, an autosampler, a column compartment unit, a
PDA, and a fluorescent detector. The system was controlled by
ChemStation (version 6.03, Hewlett-Packard). Warfarin and the
metabolites were resolved on a Hewlett Packard Zorbax ODS
C18 column (250 × 4.6 mm). The flow rate
was 1 ml/min. The mobile phases were 250 mM ammonium acetate at pH 4.9 (A) and 100% acetonitrile (B). The gradient (B) was 10% (0 min), 40%
(5-10 min), 60% (16-19 min), and 90% (22-26 min), 10% (27 min and
after). 6-, 8-, and 10-Hydroxywarfarin were monitored by UV absorbance at 313 nm and 7-hydroxywarfarin by the fluorescence at Ex320
nm/Em380 nm.
CYP2E1-mediated chlorzoxazone 6-hydroxylation was as described by Court
et al. (1997). The equipment was similar to the system described for
warfarin hydroxylations, except the substitution of a PDA by a variable
wavelength detector. The HPLC column and mobile phases were also the
same as for the warfarin assay. The gradient (B) was 10% (0 min), 40%
(6-14 min), and 90% (16-20 min), 10% (21 min and after).
6-Hydroxychlorzoxazone was monitored by UV absorbance at 280 nm.
CYP2D6-mediated bufuralol 1'-hydroxylation was as described by Boobis
et al. (1985). The HPLC system described was interfaced with a Sciex
API 2000 triple quadrupole mass spectrometer. A Supelco Discovery
C18 column (150 × 2.1 mm), and isocratic
mobile phase was applied. The mobile phase, 10 mM ammonium
acetate-acetonitrile/60:40, was run at 0.2 ml/min. The operation of the
LC/MS/MS was controlled by software MacChrom (version 1.6). The MS/MS
was operated at positive ESI with 5.5 kV ionization potential and
500°C ion source temperature. Multiple reaction monitoring (MRM) was
applied for the quantification. The MRM transition ions were
m/z 278
186 for 1'-hydroxybufuralol, and
m/z 266234 for the IS albendazole.
CYP2C19-mediated S-mephenytoin 4'-hydroxylation was as
described by Goldstein et al. (1994). The LC/MS/MS system, including the separation column, described for the bufuralol hydroxylation was
used. The isocratic mobile phase, 100 mM formic
acid-acetonitrile/75:25, was run at 0.25 ml/min. The operation of the
LC/MS/MS was controlled by MacChrom (version 1.6). The MS/MS was
operated at positive ESI with a 5.5 kV ionization potential and 100°C
ion source temperature. MRM was applied for the quantification. The MRM
transition ions were m/z 235150 for
4'-hydroxymephenytoin, and m/z 260183 for the
IS R-propranolol.
Data Analysis.
Data were acquired and analyzed by ChemStation (version 6.03, Hewlett
Packard) for 6-, 7-, and 10-hydroxywarfarin and 6-hydroxychlorzoxazone and MacChrom (version 1.6) for 1'-hydroxybufuralol and
4'-hydroxymephenytoin. Quantification was based on peak area ratios of
metabolites over the respective IS against the respective concentration
of the metabolites. The standard curves were generated by linear
regression with (6-, 7-, and 10-hydroxywarfarin, 1'-hydroxybufuralol
and 4'-hydroxymephenytoin) or without (6-hydroxychlorzoxazone) a
weighting factor (1/x2). The metabolic rates were
determined using Excel (Microsoft Office 97; Microsoft Corporation,
Redmond, WA) or SigmaPlot (version 6.00; SPSS Inc., Chicago, IL).
Apparent inhibition constants (Ki) were estimated by Dixon plots generated by the linear regression analyses and by SNLR analyses applying the reversible inhibition models
of Michaelis-Menten kinetics (Engel, 1996). The equations of velocity
or turnover rate derived from these models are as follows:
|
(1)
|
|
(2)
|
|
(3)
|
|
(4)
|
Equation 1 is for the competitive inhibition model, eq. 2 is for
the uncompetitive model, eq. 3 is for the noncompetitive model, and eq.
4 is for the mixed inhibition model. S is the substrate concentration, and I is the inhibitor concentration.
Vmax is the maximum turnover rate, and
Km is the substrate concentration at which the turnover rate is half of the maximum.
Ki is the competitive inhibition
constant, whereas Ki' is the uncompetitive
inhibition constant. Ks is the
dissociation constant of the enzyme-substrate complex. Statistical
analyses were performed using SigmaPlot and SigmaStat (version 2.03;
SPSS Inc.). PowerPoint (Microsoft Corporation) was applied for the
reconstruction of the images of Western immunoblots after being scanned.
| |
Results |
Identification of Metabolites and Metabolic P450 Forms.
Metabolite identification
Several metabolites were detected in the reconstituted systems
containing pooled HLM or HLS9 using mass spectrometers. Formation of
these metabolites was NADPH-dependent indicating the involvement of
P450s. The possible biotransformations were mono-oxidations at several
positions, N-dealkylation at either the piperidinyl or
methyl amido moiety, and multiple oxidations at different sites such as
sequential mono-oxidations to form diol metabolites.
The identification of the major metabolites was undertaken using
MS/MS spectral analyses. These metabolites (M1-M4), listed in
Table 1, were detected online by both UV
absorbance at 270 nm (Fig. 1) and MS
total ion current scanned between 60 to 1800 atomic mass units. The
MS/MS fragmentation patterns of the phase 1 metabolites and parent
compound are often similar. The MS and the MS/MS product ion spectrum
of E2101, serving as references for the spectral interpretation for the
metabolites, were first determined (Fig.
2). The predominant MS ion at
m/z 394 was the MH+ ion of
E2101, and its intensive MS/MS product ions at
m/z 178, 206, 123, and 229 were likely formed
after the CID fragmentations at the positions proposed in Fig. 2. Two
of the major metabolites (M1 and M2) formed in reconstituted systems
containing HLM or HLS9 exhibited the MH+ ion at
m/z 410. Apparently, these were the
monohydroxylated metabolites because of 16 mass unit increment compared
with E2101. M1, one of the most abundant microsomal metabolites, was
dissociated under CID in the triple quadrupole or the ion trap MS to
the product ions at m/z 194, 222, 229, 241, and
392 (Fig. 3). These characteristic product ions suggested that the metabolite was the hydroxylated E2101
at the moiety of fluorophenethyl piperidine. The product ions of the
metabolite at m/z 194 and 222 were likely the
counterparts of E2101 at m/z 178 and 206, respectively. The product ions at m/z 176 and
164, however, were possibly secondary. The abundant ion at
m/z 392, apparently formed by the
H2O elimination from the
MH+ ion, would further suggest the position of
hydroxyl group was aliphatic rather than aromatic. Therefore, it would
be reasonable to assign this metabolite as monohydroxylated E2101 with
the hydroxyl group at one of the two carbons between the piperidinyl
and the fluorophenyl group, preferably at the 
-carbon to the
piperidinyl nitrogen. In contrast, the other major hydroxylated
metabolite (M2) produced identical MS/MS product ions at
m/z 206, 178, and 123, as E2101 (Fig.
4B). Therefore, the hydroxyl group of the metabolite was not likely at the fluorophenethyl piperidinyl moiety, which was further supported by the MS2 spectrum
produced by the ion trap mass spectrometer (Fig. 4A). The formation of
product ions at m/z 351 and 379 in the ion trap MS indicated that the site of the hydroxyl group should be at the
indolyl acetamide, likely either on the indolyl ring or the carbon
between the indolyl and amido group. However, the actual position of
the M2 hydroxyl group could not be clarified although the lack of the
MS2 product ion at m/z 392 did imply that the hydroxyl group was aromatic rather than aliphatic.
E2101 could be also hydroxylated at more than one site, evidenced by
the detection of M3. M3 was one of the major, if not the only, diol
metabolite detected because of the increment of 32 atomic mass units of
the MH+ ion (m/z 426) as
compared with that of E2101 (Fig. 5A).
M3, similar to M2 and E2101, produced the MS2
product ions at m/z 206, 178, and 123, thus
possessed the intact E2101 fluorophenethyl piperidinyl moiety (Fig.
5B). Interestingly, the contrast between the abundant product ion at
m/z 408 and the lack of the product ion at
m/z 390 indicated the possible elimination of
one, but not likely two, H2O molecule from the
MH+ ion of M3 during the CID fragmentations, thus
suggesting the possible coexistence of both aliphatic and aromatic
hydroxyl group. Therefore, the sites where the hydroxyl groups attached
would likely be in the indolyl ring and between the indolyl and amido group. N-Dealkylated metabolite of E2101 (M4), in
addition, was also detected. The structural elucidation for the
dealkylated metabolite was straightforward, particularly when the
nitrogen rule was applied (McLafferty and Turecek, 1993). The
MH+ ion at m/z 272 (Fig.
6A) indicated that M4 was a cleavage
product of E2101, likely a dealkylated metabolite. The even
m/z number of the MH+ ion
of M4 would suggest that the metabolite possessed an odd number of
nitrogen atoms, as the case of that of E2101 (Fig. 2). Therefore, the
metabolite should have either one or three nitrogen atoms, and such a
requirement would be fulfilled only if M4 was formed by the
dealkylation at the piperidinyl nitrogen (Figs. 6B and
7).
View this table:
[in this window]
[in a new window]
|
TABLE 1
Relative amount of major metabolites formed in reconstituted reaction
systems containing HLM and recombinant human P450s
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 1.
E2101 metabolic profiles in reconstituted in
vitro human enzyme systems.
The metabolic profiles were monitored by the UV absorbance at
270 nm. The final protein concentration was 2 mg/ml in the reaction
mixtures containing the subcellular fractions and 50 pmol/ml in the
mixtures containing the recombinant proteins. The other reaction
conditions were described under Experimental
Procedures.
|
|
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 2.
E2101 MS (A) and MS/MS spectrum (B).
The spectra were generated by an online MS/MS (API 2000) after
the compound was eluted from the HPLC column. The MS/MS product ion
spectrum was generated after the fragmentation of the quasi-molecular
ions of E2101 (MH+) applying the collision energy. The
detail LC/MS and LC/MS/MS conditions were described under
Experimental Procedures.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 3.
Product ion spectra of metabolite M1 formed
in the reaction mixture containing HLM: the MS2 spectrum by
the ion trap MS (A) and the MS/MS spectrum by the MS/MS (B).
The product ion spectra were generated after the CID
fragmentation of the quasi-molecular ions (MH+) of M1
eluted from HPLC column. The in vitro metabolic reaction and LC/MS (ion
trap) and LC/MS/MS condition were described under Experimental
Procedures.
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4.
Product ion spectra of metabolite M2 formed
in the reaction mixture containing HLM: the M2 spectrum by the ion trap
MS (A) and the MS/MS spectrum by the MS/MS (B).
The product ion spectra were generated after the CID fragmentation of
the quasi-molecular ions (MH+) of MS2 eluted
from HPLC column. The in vitro metabolic reaction and LC/MS (ion trap)
and LC/MS/MS condition were described under Experimental
Procedures.
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 5.
MS and MS2 spectrum of
metabolite M3 formed in reaction mixture containing HLM: the MS
spectrum (A) and the MS2 spectrum (B).
The spectra were generated by an online ion trap MS after M3 was eluted
from the HPLC column. The MS2 product ion spectrum was
generated after the CID fragmentation of the quasi-molecular ions
(MH+) of M3. The detail LC/MS (ion trap) condition was
described under Experimental Procedures.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 6.
MS and MS2 spectrum of
metabolite M4 formed in reaction mixture containing HLM: the MS
spectrum (A) and the MS2 spectrum (B).
The spectra were generated by an online ion trap MS after M4 was eluted
from the HPLC column. The MS2 product ion spectrum was
generated after the CID fragmentation of the quasi-molecular ion
(MH+) of M4. The detail LC/MS (ion trap) condition was
described under Experimental Procedures.
|
|
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 7.
Proposed MS2 fragmentation
pathways of M4: an example of nitrogen rule in LC/ESI-MS/MS
application.
The MS and MS2 spectrum were shown in Fig. 6, and the
condition for the generation of the spectra was described under
Experimental Procedures.
|
|
Several minor E2101 metabolites including the
N-demethylated metabolite, detected only by LC/MS, were not
presented and discussed. The major metabolic pathways of E2101 are
proposed in Fig. 8.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 8.
Proposed major P450-mediated E2101 metabolic
pathways in humans.
The relatively minor metabolites, including the
N-demethylated metabolite and the monohydroxylated
metabolites other than M1 and M2, were not included.
|
|
Metabolic enzyme identification.
The metabolic enzymes were identified by detecting the formation of
metabolites in reconstituted enzyme systems. The rate of
NADPH-dependent metabolism was found to be faster in the reaction mixture containing pooled HLM than that containing pooled HLS9 (Fig.
1). Therefore, the major metabolic enzymes of E2101 would be microsomal
oxidases, or hepatic P450s.
The responsible P450 forms for E2101 metabolism were further determined
using a panel of recombinant human P450s including CYP1A2, CYP2C9,
CYP2C19, CYP2D6, CYP2E1, and CYP3A4. In the presence of approximately
50 µM E2101, CYP2C19, CYP2D6, and CYP3A4, among the major P450 forms
tested, metabolized E2101 (Fig. 1). The formation of the major E2101
metabolites was P450 form-dependent (Table 1). CYP3A4 produced the
broadest spectrum of metabolites including M1-M4, similar to that
generated by HLM or HLS9 preparations. However, CYP2C19 and CYP2D6
produced rather distinctive metabolite profiles. CYP2D6 preferably
converted E2101 to M3, and M2 to a much less extent, whereas CYP2C19
converted E2101 to form all the major metabolites but M3. Apparently,
the formation of the hydroxylated metabolite M2 was not metabolizing
P450 form-specific.
P450 Inhibition.
A panel of P450 substrate assays was applied to determine the P450
form-specific inhibition as described under Experimental Procedures. The quantifications were based on the standard
calibration curves. The correlation coefficient
(r2) for each calibration curve was at
least 0.990. The experimental quality was also ensured by quality
control samples. As shown in Table 2, no
inhibitory effect of E2101 at 10 or 50 µM on CYP1A2, CYP2C9, CYP2E1,
and CYP3A4 activity was detected, as assessed by R-warfarin
6- (CYP1A2) and 10-hydroxylation (CYP3A4), S-warfarin 7-hydroxylation (CYP2C9), and chlorzoxazone 6-hydroxylation (CYP2E1). However, the activities of CYP2C19-mediated S-mephenytoin
4'-hydroxylation and CYP2D6-mediated bufuralol 1'-hydroxylation were
significantly reduced at the presence of E2101 in a
concentration-dependent manner.
The apparent inhibition constants (Ki)
of CYP2C19 and CYP2D6 were determined using the microsomal preparations
containing cDNA-expressed recombinant P450 proteins. The apparent
Ki values were first estimated by the
Dixon plots. The estimated Ki varied between 25 to 45 µM for CYP2C19 (Fig.
9A), and approximately 20 µM for CYP2D6
(Fig. 9B). The apparent Ki values were
also determined by SNLR analysis using the common reversible inhibition
models of Michaelis-Menten kinetics including the competitive,
uncompetitive, noncompetitive, and mixed inhibition model. The
competitive inhibition model was selected to determine the
Ki because of the best curve fitting
(r2 > 0.97 for CYP2C19 inhibition;
r2 > 0.99 for CYP2D6 inhibition). In
consistence with the Dixon plots, the apparent
Ki values determined by SNLR were 48 µM for CYP2C19 inhibition and 15 µM for CYP2D6 inhibition. Both the
Dixon plots and SNLR analyses suggested that the inhibitions of CYP2C19 and 2D6 by E2101 were primarily, if not fully, competitive.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 9.
Dixon plots for E2101 inhibition of
CYP2C19-mediated S-mephenytoin 4'-hydroxylation (A), and
CYP2D6-mediated bufuralol 1'-hydroxylation (B).
The recombinant enzymes were applied to obtain the single enzyme
kinetics. The detail experimental condition was described under
Experimental Procedures.
|
|
P450 Induction.
P450 induction was evaluated applying primary culture of human
hepatocytes from three donors (two smokers and one nonsmoker). At the
time of the reception of the hepatocyte cultures, the confluence of the
cells from the nonsmoker was slightly denser than those from the
smokers. Moreover, the cell density tended to reduce slightly although
marked morphologic change was not observed among the control cells and
those exposed to E2101 during the treatment. Therefore, it may not be
suitable to compare the protein expression among the cells from
different donors. With the exception of the monoclonal anti-CYP2D6
antibodies, the polyclonal primary antibodies used for immunochemical
detection would cross-react with the members in the same P450
subfamilies. Therefore, the enzymes determined were CYP1A1/2,
CYP2C8/9/19, CYP2D6, and CYP3A4/5. As shown in Fig.
10, the cells responded to the
inductions of CYP1A1/2, CYP3A4/5, and CYP2C8/9/19 as demonstrated by
the elevated expressions of these proteins in the hepatocytes after the
exposure to TCDD and rifampicin. However, the expression of CYP1A1/2,
CYP2C8/2C9/2C19, CYP2D6, or CYP3A4/5 did not increase in the cells
exposed to E2101 at 10 or 50 µM for 72 h. Therefore, E2101 did
not appear to induce the expression of these P450 forms.
View larger version (54K):
[in this window]
[in a new window]
|
Fig. 10.
Effect of E2101 on P450 expression in
primary human hepatocytes after 72-h exposure.
The protein expression was determined immunochemically, and anti-P450
antibodies, and chemiluminescent reagents were used for the detection.
Cells treated with TCDD (0.4 µM) were used to serve as the positive
control for CYP1A induction, and cells treated with rifampicin (50 µM) were used to serve as the positive control for CYP3A, and
probably CYP2C19, induction. The detail experimental condition was
described under Experimental Procedures.
|
|
| |
Discussion |
The advance in molecular biology during the last decade has led us
to determine the metabolic roles of human P450s and to uncover the
possible mechanism of the clinical drug-drug interactions (Fang and
Gorrod, 1999; Malaty and Kuper, 1999; Venkatakrishnan et al., 2000).
Today it has been acknowledged that the majority of drug-drug
interactions were due to the interactions between the therapeutic
agents and their metabolic enzymes, especially P450s. Therefore,
metabolic interaction studies of new drug entities in vitro for
predicting potential risk of drug-drug interaction before the clinical
trials are becoming useful.
E2101, similar to some of the 5-HT1/2 antagonistic amine derivatives,
was metabolized by CYP2C19, CYP2D6, and CYP3A4 to a series of oxidative
metabolites, including the mono- and dihydroxylated, and
N-dealkylated metabolites. Based on the MS/MS spectral
analyses, the structures of the major microsomal metabolites were
proposed. Without synthetic standards, the MS/MS spectra may not be
sufficient for the identification of the exact sites of the hydroxyl
groups for M1-M3. However, the potential mechanism of P450 catalytic reactions may assist, at least to some extent, the structural interpretation of these metabolites. For example, the hydroxylation to
form M1 could possibly take place at the positions other than the
proposed, especially the 
-carbon next to the piperidinyl nitrogen
since the CID fragmentations would possibly produce the same MS/MS
spectra as shown in Fig. 3. However, the formation of such metabolite
would not be feasible based on well known P450-mediated oxidative
dealkylations. If the hydroxylation did take place at this putative
site, the metabolite would exist rather as the more stable
N-dealkylated metabolite M4.
Although exemplified only for the structural elucidation of the
N-dealkylated metabolite M4, the nitrogen rule was indeed applicable for the interpretation of all of the MS spectra shown (Figs.
2-6). Noticeably, the nitrogen rule in ESI-MS/MS would be the reverse
of what originally applied in GC/MS due to the preferred formation of
quasi-molecular ion (or quasi-product ion). For example, as shown in
Fig. 2, the MH+ ion of the odd nitrogen atoms
containing E2101 exhibited an even m/z number
(394). In addition, two of the proposed MS/MS product ions of E2101
containing the even nitrogen atoms (0 and 2) exhibited odd
m/z numbers (123 and 229), and the other product
ions containing the odd nitrogen atom (1) exhibited even
m/z numbers (178 and 206). Therefore, such a
useful tool might be applied more often for the interpretation of
ESI-MS/MS spectra because the formation of odd-electron ions or radical
ions is usually unstable, thus rare in ESI-MS/MS.
The potential contribution of individual E2101-metabolizing P450
forms would not be accurately estimated without knowing the kinetic
parameters. However, the metabolic profile of E2101 generated by
recombinant CYP3A4, the most abundant human hepatic P450 form (Shimada
et al., 1994), was similar to that observed in the reconstituted system
containing HLM (Fig. 1). This metabolic profile in HLM was virtually
concentration-independent as determined using 10, 50, and 200 µM
E2101 (data not shown). Moreover, at 50 µM E2101 concentration, 16 µM CYP3A4 inhibitor ketoconazole inhibited at least 75% of the
formation of all major metabolites (M1-M4), whereas 1.6 µM CYP2D6
inhibitor quinidine inhibited approximately 40% of the formation of M3
but M2 to a much less extent in the reaction mixture containing HLM.
The concentrations of the inhibitors used were at least several-fold
higher than their respective Ki values (Bourrie et al., 1996; Fogelman et al., 1999), thus they should be able
to inhibit the majority, if not all, of these P450 form-specific activities in HLM. Therefore, the metabolic profile in HLM
appeared to be largely controlled by the hepatic CYP3A4 activity, and
CYP3A4 is likely the one responsible for the hepatic metabolism of
E2101, with the contribution of CYP2D6, especially for the formation of M3.
The structural requirements to be the substrates (or inhibitors) of
CYP2D6 are probably the best defined among the major metabolic P450
forms. E2101 containing basic nitrogens, and two or more hydrophobic
regions, is structurally similar to some of the well known CYP2D6
substrates (Ekins et al., 1999). However, such structural requirements
for CYP2C19 or CYP3A4 could not be depicted due to the limited
structure-activity relationship studies reported for CYP2C19 (Lock et
al., 1998; Ekins et al., 2001) or the diverse substrate selectivity of
CYP3A4 (Smith et al., 1997). The lack of CYP3A4 inhibition by E2101 as
determined by CYP3A4-mediated R-warfarin 10-hydroxylation
may also indicate atypical enzyme kinetics for CYP3A4-mediated
reactions including possibly two or more different substrate (or
inhibitor) binding sites (Wang et al., 2000; Oda and Kharasch, 2001;
Shou et al., 2001). Therefore, our data would not exclude the
possibility of the E2101 alteration of CYP3A4-mediated metabolism other
than R-warfarin, which may require further studies to
clarify. On the other hand, based on Michaelis-Menten enzyme kinetics,
the apparent Ki values for CYP2C19 and
CYP2D6 inhibition were estimated by both Dixon plots and SNLR analyses
in this study. The values generated by these two methods were quite
consistent. Dixon plots, providing the visual indication of the enzyme
inhibition mechanism, were used for the initial estimation of the
Ki values, which were confirmed using
SNLR method (Kakkar et al., 1999). The data obtained from a single
experiment were simultaneously analyzed by both methods, and the
consistent results would be thus expected reliable for the inhibition
of CYP2C19 and CYP2D6 detected.
The induction of the members in the subfamilies of CYP1A, CYP2C, and
CYP3A, and as well as CYP2D6 at protein level was not detected in the
primary human hepatocytes after E2101 exposure. CYP1A enzymes are
responsible for the bioactivation of a variety of polycyclic aromatic
hydrocarbons (Liu et al., 2001), whereas CYP3A4 is the most common
drug-metabolizing P450 form (Shimada et al., 1994). In addition, the
effect of E2101 on the expression of the minor but inducible CYP2C19,
and constitutive CYP2D6, was also determined for the elimination of
possible feedback from enzyme inhibition. Instead of the possible
induction, the expression of some P450s appeared to be slightly
suppressed in cells exposed to 50 µM E2101. However, such a weak
suppression by E2101 at higher concentration may need to be confirmed
since Western blots were at most semiquantitative, which may not be
applied appropriately to interpret subtle differences. Nevertheless,
the lack of induction of the major E2101-metabolizing P450 forms in
primary culture of human hepatocytes suggests that the potential
drug-drug interaction in clinic due to P450 induction would be unlikely.
In summary, the drug-drug interaction between E2101, a potential
muscle relaxant, and human P450s was characterized in vitro. E2101 was
metabolized by CYP2C19, CYP2D6, and CYP3A4 to form a series of phase 1 oxidative metabolites including mono- and dihydroxylated, and
N-dealkylated metabolites. The structures of these
metabolites were elucidated based on the MS/MS spectral analyses. E2101
moderately inhibited the activities of CYP2C19 and CYP2D6 in a
competitive manner but did not induce the expression of any
E2101-metabolizing P450 or CYP1A enzymes in human primary hepatocytes.
Therefore, although the understanding of in vitro interaction between
E2101 and P450s would be beneficial for the clinical trials, in
particular for patients taking coadministered medications that are
metabolized by CYP2C19 and/or CYP2D6, a low risk of drug-drug
interaction in clinic would be anticipated.
We thank L.-Q. Nguyen and P. Saxton for technical assistance.
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
Abstract
Introduction
