Characterization of mixtures of recombinant human cytochrome P450s as a screening model for metabolic stability in drug discovery

Article abstract of DOI:10.1080/00498250210147124

1. Recombinant human cytochrome P450 (rhCYP) has become an important screening model in drug metabolism studies due to the high cost of human and animal hepatic tissue. Until now, rhCYPs have been evaluated and used as separate forms, but a mixture of CYP forms comparable with the human liver could be of value in early drug discovery. 2. In the present study, rhCYP2C9, rhCYP2D6 and rhCYP3A4 co-expressed with reductase in Escerichia coli were mixed and evaluated with regards to kinetic properties (K_m and V_max). Furthermore, antioxidant was added to investigate whether a free radical scavenger would affect the kinetic parameters. Results were compared with data obtained in human liver microsomes (HLM). 3. Results showed a good correlation between mixed rhCYP data and HLM data for K_m and V_max. K_m varied < 3-fold between matrices for CYP2C9 and CYP3A4, whereas the K_m for CYP2D6 varied up to 4.5-fold. V_max differed up to 3-fold between matrices for the CYP forms investigated. However, the discrepancy in V_max may depend on the anticipated level of each form in HLM. The addition of antioxidant increased V_max for CYP2C9 and CYP2D6 by 75 and 50%, respectively, whereas V_max for CYP3A4 was unchanged. 4. In conclusion, the rhCYP mixture shows promising results as a predictor of CYP kinetic parameters. Furthermore, addition of antioxidant can in certain cases increase catalytic activity.

Full text of DOI:10.1080/00498250210147124

xenobiotica, 2002, vol. 32, no. 9, 749±759
Characterization of mixtures of recombinant
human cytochrome P450s as a screening model
for metabolic stability in drug discovery
N. HAGEN*, A. K. OLSEN{, J. V. ANDERSEN{, J. TJéRNE-
LUND{} and S. H. HANSEN{
{
Department of Analytical and Pharmaceutical Chemistry, The Royal Danish School of
Pharmacy, DK-2100 Copenhagen, Denmark
Department of Drug Metabolism, Novo Nordisk A/S, DK-2760 Ma
{
Ê
lùv, Denmark
Received 22 January 2002
1
. Recombinant human cytochrome P450 (rhCYP) has become an important
screening model in drug metabolism studies due to the high cost of human and animal
hepatic tissue. Until now, rhCYPs have been evaluated and used as separate forms, but a
mixture of CYP forms comparable with the human liver could be of value in early drug
discovery.
2
. In the present study, rhCYP2C9, rhCYP2D6 and rhCYP3A4 co-expressed with
reductase in Escerichia coli were mixed and evaluated with regards to kinetic properties
Km and Vmax). Furthermore, antioxidant was added to investigate whether a free radical
(
scavenger would a€ect the kinetic parameters. Results were compared with data obtained
in human liver microsomes (HLM).
3
. Results showed a good correlation between mixed rhCYP data and HLM data for
Km and Vmax. Km varied < 3-fold between matrices for CYP2C9 and CYP3A4, whereas
the Km for CYP2D6 varied up to 4.5-fold. Vmax di€ered up to 3-fold between matrices for
the CYP forms investigated. However, the discrepancy in Vmax may depend on the
anticipated level of each form in HLM. The addition of antioxidant increased Vmax for
CYP2C9 and CYP2D6 by 75 and 50%, respectively, whereas Vmax for CYP3A4 was
unchanged.
4
. In conclusion, the rhCYP mixture shows promising results as a predictor of CYP
kinetic parameters. Furthermore, addition of antioxidant can in certain cases increase
catalytic activity.
Introduction
In vitro drug metabolism studies, investigating the extent and route of
candidate drug metabolism, have been recognized as important methods in the
early discovery screening of new drug candidates (White 1998, Eddershaw and
Dickins 1999). The number of new drug candidates and thereby screening
expenses has increased signi®cantly due to combinatorial chemistry and other
developments in medicinal research (Roberts et al. 2001). Therefore, ecient low-
cost metabolism study protocols are needed.
Cytochrome P450 (CYP) is the major family of Phase I drug-metabolizing
enzymes in the human liver. CYP-catalysed reactions are often the main reason for
drugs failing to make it to the market and therefore of great interest for the drug
metabolism scientist (Bajpai and Adkison 2000). Hepatic microsomes have proven
*
}
Author for correspondence; e-mail: nhag@novonordisk.com
Present address: Analytical Support, H. Lundbeck A/S, DK-2500 Valby, Denmark.
Xenobiotica ISSN 0049±8254 print/ISSN 1366±5928 online # 2002 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/00498250210147124

750
N. Hagen et al.
to be the simplest and best suited model for high-throughput screening among the
traditional metabolic stability in vitro models (Roberts et al. 2001). As the supply
of human tissue is expensive and of varying quality, animal liver tissue is often
used for early metabolic stability screening purposes presently. However, due to
interspecies di€erences, a more human-like alternative would be preferable.
Therefore, it is an advantage to have other in vitro possibilities for human
metabolism screening than hepatic microsomes.
cDNA-expressed human cytochrome P450 enzymes provide a reproducible,
consistent source of single CYP forms for many types of studies. Owing to its rapid
and inexpensive production in, for example, bacterial systems, recombinant
human CYPs (rhCYP) have become a common tool in drug metabolism studies
(Crespi and Miller 1999). rhCYPs expressed in Escherichia coli have been reported
generally to exhibit the same kinetic pro®les and inhibition properties as shown in
HLM experiments when incubated separately (Jung et al. 1997, McGinnity et al.
1999, Moody et al. 1999). However, others have reported that enzyme activity
varies among the in vitro systems (Crespi and Penman 1997). The level of
NADPH-cytochrome P450 oxidoreductase (reductase) and/or other cofactors
may be suboptimal, as the catalytic activity is not solely determined by the CYP
itself but, also, by the electron transport partners present. For example, the
reductase:CYP ratio is higher in expression systems compared with in vivo thereby
increasing the risk for accumulation of reactive oxygen species (Crespi and Miller
1999). Single rhCYPs are typically used to determine potential inhibition of the
speci®c enzymes or they are used in the identi®cation of speci®c enzymes involved
in metabolism (McGinnity et al. 2000).
The use of single enzyme systems, relative to multi-enzyme systems, has
distinct advantages and disadvantages depending on the speci®c application.
Mixing the rhCYPs in levels comparable with HLM would be another approach
to study metabolic stability in the early discovery phase with reduced cost and
would be a way to produce `arti®cial livers’ representing subpopulations (e.g.
alcoholics, polymorphisms). Crespi and Miller (1999) and Eddershaw and Dickins
(1999) have reported a reasonable correlation to HLM metabolism pro®les by this
approach, but a formal evaluation has not been published. Hence, it is essential to
know whether the rhCYPs a€ect the catalytic activity of each other in vitro and to
study the conditions that are likely to a€ect this.
The aim of the present study was, therefore, to compare the enzyme kinetics of
a mixture of rhCYPs to HLM, and furthermore to evaluate the e€ect of
antioxidant addition. A simpli®ed mixture containing rhCYP2C9, rhCYP2D6,
and rhCYP3A4 (expressed in E. coli) in ratios equivalent to those reported to be
present in HLM has been studied. The three forms were chosen based on their
high contribution to the total CYP-catalysed metabolism of marketed drug
substances. Furthermore, they represent CYPs with both high and low natural
abundance in human liver.
Kinetic pro®les for reactions containing both separate and mixed rhCYPs were
compared with data obtained in HLM incubations. Also, the addition of a free
radical scavenger, the antioxidant ascorbate, was investigated. Ascorbate was
added to decrease possible accumulation of reactive oxygen species caused by an
excess of reductase or uncoupling of the CYP catalytic cycle in order to test its
e€ect on the enzyme kinetics.

Characterization of recombinant human CYP
Materials and methods
751
Chemicals
S-warfarin, 7-hydroxywarfarin and dextrorphan were purchased from Ultra®ne Chemicals (Man-
chester, UK). Dextromethorphan hydrobromide, testosterone, 6 ­ -hydroxytestosterone and corticos-
terone were purchased from Sigma (St Louis, MO, USA). All other chemicals were of analytical grade.
Test systems
HLM (mixed pool, 15 donors) were obtained from In Vitro Technologies (Baltimore, MD, USA).
Transformed E. coli co-expressing either CYP2C9, CYP2D6 or CYP3A4 with reductase were provided
by the LINK consortium (a collaboration between University of Dundee, UK, and a number of
pharmaceutical companies). Expression of the recombinant proteins and the subsequent preparation of
E. coli membranes were carried out according to earlier published protocols (Pritchard et al. 1998, Voice
et al. 1999).
Cytochrome P450 content was determined by CO-di€erence UV spectrum as described by Omura
and Sato (1964). Protein content was determined by the method described by Lowry et al. (1951) using
BSA as standard protein for quanti®cation. Reductase activity was estimated by its ability to reduce
cytochrome c in the presence of NADPH in a 300mm potassium phosphate bu€er at pH 7.7 (Vermilion
and Coon 1978).
Testosterone 6 ­ -hydroxylase assay. Incubations (500ml) contained (1) 24 pmol rhCYP3A4, or (2)
2
0
4 pmol rhCYP3A4, 3.5 pmol rhCYP2D6 and 15.5pmol rhCYP2C9, or (3) HLM (0.2 mg protein) in
.1 m sodium phosphate bu€er (pH 7.4). Microsomal incubations furthermore contained magnesium
chloride (5 mm). Testosterone dissolved in methanol was added to the incubation mixture in eight
concentrations covering a range around the expected Km (9±230mm). The ®nal methanol concentration
was 1% v/v. For samples testing the e€ect of antioxidant, ascorbate was added to a ®nal concentration of
5
mm. The incubation mixture was pre-incubated at 37 8C for 5 min before 1 mm NADPH was added to
start the reaction. Incubation was carried out for 5min at 37 8C and stopped by the addition of 3 ml
dichloromethane. Corticosterone (1mm) was subsequently added as an internal standard and the samples
were vigorously shaken and centrifuged. The aqueous phase was then removed followed by evaporation
of 2 ml dichloromethane phase at 35 8C in the presence of N2. The residue was redissolved in 100ml
methanol before analysis (injection volume 35 ml). The metabolites were separated by HPLC on a
LiChrospher 100 RP-18 endcapped (5mm) 250 4 mm column (Merck, Darmstadt, Germany)
operating at 40 8C using a gradient based on (A) methanol/water (30/70v/v), (B) methanol/water (70/
£
¡
1
3
0 v/v) and (C) tetrahydrofuran/water (50/50 v/v) at a ¯ow rate of 1 ml min . Gradient: Isocratic from 0
to 8 min at 87% A/13% C, linear gradient from 8 to 28 min going from 87% A/13% C to 17% A/70% B/
3% C and from 28 to 29 min going to 100% B. Isocratic from 29 to 34min at 100% B, and linear
1
gradient from 34 to 35 min going to 87% A/13% C. UV detection was performed at 240 nm. The assay
was based on protocols published by Van der Hoeven (1984) and Reinerink et al. (1991).
S-warfarin 7-hydroxylase assay. Incubations (200 ml) contained (1) 6.2 pmol rhCYP2C9, or (2)
6
0
.2 pmol rhCYP2C9, 1.4pmol rhCYP2D6 and 9.6 pmol rhCYP3A4 or (3) HLM (0.08 mg protein) in
.1 m sodium phosphate bu€er (pH 7.4). Microsomal incubations furthermore contained magnesium
chloride (5 mm). S-warfarin dissolved in acetonitrile was added to the incubation mixture in eight
concentrations covering a range around the expected Km (0.4±53mm). The S-warfarin concentration was
9
mm in the experiment investigating time dependency. The ®nal acetonitrile concentration was 0.1%.
For samples testing the e€ect of antioxidant, ascorbate was added to a ®nal concentration of 5 mm. The
incubation mixture was pre-incubated for 5 min at 37 8C before 1mm NADPH was added to start the
reaction. Incubations were carried out at 378C for 15 min (or 10, 20, 30, 45, 60, 90 and 120min for the
time dependency experiment) then stopped by the addition of 100ml 12% perchloric acid. Following
centrifugation, the metabolites in the supernatant were separated by HPLC on a LiChrospher 100 RP-
£
1
8 endcapped (5 mm) 250 4 mm column (Merck) operating at 408C using an eluent containing 0.5%
phosphoric acid/acetonitrile (67/33v/v) at a ¯ow rate of 1 ml min . The injection volume was 100 ml.
Fluorescence detection at ¶ex315nm and ¶em395nm was performed. The HPLC assay was based on the
¡
1
È
method described by Lang and Bocker (1995).
Dextromethorphan O-demethylase assay. Incubations (200ml) contained (1) 1.4 pmol rhCYP2D6, or (2)
1
0
.4 pmol rhCYP2D6, 9.6 pmol rhCYP3A4 and 6.2 pmol rhCYP2C9, or (3) HLM (0.08 mg protein) in
.1 m sodium phosphate bu€er (pH 7.4). Microsomal incubations furthermore contained magnesium
chloride (5 mm). Dextromethorphan dissolved in sodium phosphate bu€er was added to the incubation
mixture in eight concentrations covering a range around the expected Km (0.25±22 mm). For samples
testing the e€ect of antioxidant, ascorbate was added to a ®nal concentration of 5 mm. The incubation
mixture was then pre-incubated for 2 min at 378C followed by the addition of 1 mm NADPH to start the

752
N. Hagen et al.
reaction. Incubation was carried out for 10 min at 37 8C and stopped by the addition of 200 ml
acetonitrile. Following centrifugation, the metabolites in the supernatant were separated by HPLC
on a LiChrospher 100 CN (5mm) 125 4 mm column (Merck) operating at 40 C using an eluent
£
8
containing 100 mm ammonium acetate (pH 5.0)/triethylamine/acetonitrile (80/0.06/20v/v/v). The
¡
1
samples were diluted 1:1 with eluent. The injection volume was 50 ml. The ¯ow rate was 1 ml min
.
Fluorescence detection was performed at ¶ex220nm and ¶em305nm. The HPLC method was based on a
protocol by Jones et al. (1996).
Kinetic analysis
Km and Vmax were determined by non-linear regression ®t to Michaelis±Menten enzyme kinetics
(
GraphPad Prism version 3.0 for Windows, GraphPad Software, San Diego, CA, USA).
Results
CYP content and reductase activity of the rhCYPs used are shown in table 1.
The CYP content in the recombinant systems was essentially the same for the
three forms. The reductase activity in the rhCYP3A4 preparation was approxi-
mately 3-fold higher than in rhCYP2C9 and rhCYP2D6 preparations.
Mixture of rhCYPs versus separate rhCYPs
Comparisons of kinetic parameters between rhCYPs incubated separately and
as a mixture with other rhCYPs as well as those determined in HLM incubations
are shown in table 2. The data for maximal catalytic activity (Vmax) in HLM are in
this study corrected for the average content in a Caucasian human liver of the CYP
in question, i.e. 18% CYP2C9, 4% CYP2D6 and 28% CYP3A4 (Guengerich 1995,
Lewis 1998). In this way the results obtained in HLM incubations can be directly
compared with the results obtained in the recombinant system. Vmax for the
rhCYP mixture was comparable with rhCYPs incubated separately for CYP2C9
and CYP2D6 (table 2 and ®gure 1). In contrast, CYP3A4 Vmax was slightly
decreased when the other rhCYPs were included in the incubation (table 2 and
®
gure 1). However, Vmax determined in separate rhCYP or the rhCYP mixture
were 2±3-fold lower than obtained in HLM for both CYP2D6 and CYP3A4,
whereas for CYP2C9 the results were opposite with approximately 2-fold higher
Vmax for the rhCYPs compared with HLM.
Apparent anity constants (Km) determined in single rhCYP forms, the
rhCYP mixture, and HLM varied <3-fold for CYP2C9 and CYP3A4, whereas
Km for CYP2D6 varied up to 4.5-fold between matrices (table 2).
Table 1. CYP content and reductase activity of the recombinant systems used determined by spectral
analysis (CO-di€erence spectrum), as described by Omura and Sato (1964), and the activity
towards cytochrome c (Vermillion and Coon 1978). All values are means of duplicate
determinations.
CYP-speci®c content
(
Reductase activity
(nmol cyt c min mg
protein)
Reductase activity/CYP
(nmol cyt c min pmol
CYP)
¡
1
¡1
¡1
¡1
¡1
pmol mg
protein)
CYP form
rhCYP2C9
rhCYP2D6
rhCYP3A4
349
306
387
72
61
205
0.206
0.199
0.530

Characterization of recombinant human CYP
753
Table 2. Km and Vmax for the studied CYPs as determined in incubations including a single rhCYP,
mixed rhCYPs and HLM for CYP2C9 catalysed S-warfarin 7-hydroxylation, CYP2D6-
catalysed dextromethorphan O-demethylation, and CYP3A4-catalysed testosterone 6­ -
ˆ
§
hydroxylation (n 3; SD). HLM data are corrected for the expected contant of CYP2C9
(18%), CYP2D6 (4%) and CYP3A4 (28%) (Guengerich 1995, Lewis 1998).
Vmax
¡1
¡
1
(pmol min nmol relevant CYP)
CYP form
CYP2C9
Matrix
Km (mm)
§
§
single rhCYP
rhCYP mixture
HLM
7:5 0:7
74
80
37
2
2
3
§
§
§
6:4 0:5
§
3:4 1:0
§
§
CYP2D6
CYP3A4
single rhCYP
rhCYP mixture
HLM
1:1 0:1
2700 100
§
§
2700 300
0:6 0:2
§
§
2:7 0:7
8000 600
§
§
21000 2000
single rhCYP
rhCYP mixture
HLM
37
50 18
8
§
§
14000 2000
§
53 12
§
31000 2000
Figure 1. E€ect of addition of ascorbate on (a) CYP2C9 catalysed S-warfarin 7-hydroxylation, (b)
CYP2D6 catalysed dextromethorphan O-demethylation and (c) CYP3A4-catalysed testosterone
6
­ -hydroxylation using rhCYPs separately (*, solid line) and in a rhCYP mixture (&, dashed
line). Data from experiments where ascorbate was added are presented with closed legends.
ˆ 3 § S:D.).
Control data (no ascorbate) is presented with open legends (n
Antioxidant addition
Results from rhCYP2C9, rhCYP2D6 and rhCYP3A4 experiments where
ascorbate was added are shown ®tted to the Michaelis±Menten equation in ®gure
1
a±c. Vmax and Km are listed in table 3. Addition of ascorbate gave rise to an
increased Vmax for rhCYP2C9 (about 75%) and rhCYP2D6 (about 50%) incubated
separately. Likewise, an increase was observed for Km when ascorbate was added
to rhCYP2C9 (2-fold) and rhCYP2D6 (50%). Results obtained in experiments
using the rhCYP mixture were comparable with those obtained using the separate
rhCYPs. For rhCYP3A4, on the other hand, ascorbate addition appeared to result

754
N. Hagen et al.
Table 3. E€ect of ascorbate addition on Km and Vmax as determined in rhCYPs both separately and
mixed for CYP2C9-catalysed S-warfarin 7-hydroxylation, CYP2D6-catalysed dextromethorphan O-
demethylation and CYP3A4 catalysed testosterone 6 ­ -hydroxylation (n
ˆ 3 § SD).
¡
1
¡1
CYP form
CYP2C9
Matrix
Km (mm)
Vmax (pmol min nmol relevant CYP)
§
§
single rhCYP
rhCYP mixture
14
11
2
2
130
115
6
6
§
§
§
§
CYP2D6
CYP3A4
single rhCYP
rhCYP mixture
1:6 0:2
3800 100
4100 200
§
§
0:9 0:2
§
§
20000 2000
single rhCYP
rhCYP mixture
27
30
8
7
§
§
12000 1000
Figure 2. Time dependence of the e€ect of ascorbate on the 7-hydroxylation of S-warfarin in rhCYP
ˆ 3
§
mixture (n
SD). Controls (no ascorbate added) are presented as open bars.
in a greater anity towards testosterone (Km decreased) whereas no change in
Vmax was observed. Experiments to determine the potential time-dependent e€ect
of ascorbate addition in prolonged incubations versus short incubations showed no
time dependency (®gure 2).
Discussion
The yields of rhCYP in the E. coli expression system are similar to those
reported by Pritchard et al. (1997, 1998) and McGinnity et al. (1999) for similar
cDNA-constructs from the LINK consortium. The reductase activities were
¡
1
¡1
typically between 100 and 600 nmol cytochrome c reduced min mg protein in
the above-mentioned papers, thus the reductase activities reported here are slightly
lower for the rhCYP2C9 and rhCYP2D6 membrane preparations (72 and 61 nmol
¡
1
¡1
cytochrome c reduced min mg protein, respectively) as shown in table 1.
Mixture of rhCYPs versus separate rhCYPs
The mixture of rhCYPs investigated was based on the relative abundance of the
CYP2C9, CYP2D6 and CYP3A4 in a Caucasian human liver. Generally, CYP2C9
is reported to account for 15±20% of the total cytochrome P450 content, whereas
CYP2D6 and CYP3A4 have been reported to account for 1±5% and approximately
3
1
0%, respectively (Shimada et al. 1994, Guengerich 1995, Inoue et al. 1997, Lewis
998, Pelkonen et al. 1998). However, human hepatic tissue is subject to large

Characterization of recombinant human CYP
755
interindividual di€erences in CYP expression due to both genetic polymorphisms
and induction/inhibition caused by environmental exposures or drug treatments.
Polymorphisms have been reported for CYP2C9 (Crespi and Miller 1997, Kimura
et al. 1998, Yamazaki et al. 1998), CYP2D6 (Gut et al. 1986, Ko
and recently for CYP3A4 (Eiselt et al. 2001, Dai et al. 2001). CYP2D6 poly-
morphisms are due to both inactive alleles (Kohler et al. 1997) as well of gene
multiplication (Ingelman-Sundberg 1997) leading to an expression variability of
000-fold or more (Guengerich 1995). CYP2C9 allelic variants have been shown to
È
hler et al. 1997),
È
1
change the catalytic activity as well as the substrate anity rather than a€ect the
expression level (Crespi and Miller 1997, Inoue et al. 1997). Among the CYP3A4
allelic variants, one is reported to be expressed at signi®cantly lower levels than
wild type CYP3A4, and for another variant an altered testosterone metabolite
pro®le is observed (Eiselt et al. 2001). With regard to induction, several inducers of
CYP2C9 and CYP3A4 are known (e.g. barbiturates and rifampicin), but no
induction of CYP2D6 has been detected so far. In an attempt to minimize the
large interindividual di€erences in our study, a pool of 15 di€erent donors was
used. However, a comparison of the catalytic activities reported by the supplier of
1
4 mixed pools of HLM revealed a 10-fold variation with regards to CYP2C9
catalytic activity, CYP2D6 catalytic activity varied 4-fold, and CYP3A4 catalytic
activity varied 5-fold (data not shown). Furthermore, it has to be kept in mind that
even larger variability is observed between individuals. The overall levels of total
CYP vary only a few fold among most humans (Guengerich 1995), which was also
re¯ected by the fact that the total CYP content varied <3-fold and reductase
activity varied <4-fold in the above-mentioned mixed pools of HLM. In short,
the CYP ratios chosen, i.e. 18% CYP2C9, 4% CYP2D6 and 28% CYP3A4
(Guengerich 1995, Lewis 1998), re¯ect an `average’ Caucasian liver, without
being an absolute standard.
The kinetic properties (Km and Vmax) of rhCYP2C9 were not signi®cantly
a€ected by the addition of CYP2D6 and CYP3A4 (table 2 and ®gure 1a).
Furthermore, these kinetic parameters are within 2-fold range of the HLM data.
This suggests that rhCYP2C9 co-expressed with reductase can be mixed with
other rhCYPs for kinetic investigations without a€ecting the kinetic parameters.
CYP2D6 substrate anity and maximal catalytic activity show a good correlation
between rhCYP2D6 and the rhCYP mixture. Km determined in rhCYP2D6 and
the rhCYP mixture were, however, low compared with data obtained in incuba-
tions with HLM. Vmax determinations in rhCYP2D6 and the rhCYP mixture were
3
-fold lower than in HLM (table 2). Both Km and Vmax determined in rhCYP3A4
were within 2-fold of the values obtained in HLM (table 2). For rhCYP3A4, Vmax
was somewhat decreased by the addition of rhCYP2C9 and rhCYP2D6 (table 2
and ®gure 1c). On the contrary, the Km for CYP3A4 was not a€ected. It could be
speculated that competition for reductase takes place as shown by Li et al. (1999).
This latter group co-expressed rhCYP2D6 and rhCYP3A4 in the same system. By
adding substrate for both CYP forms it was demonstrated that competition did
take place. The experiments described in the current paper were, however,
di€erent as CYP2D6 and CYP3A4 were expressed in separate systems. Further-
more, one substrate only was added in each experiment. The decrease in CYP3A4
activity upon mixing forms is rather surprising considering that more rhCYP3A4
is present in the rhCYP mixture than rhCYP2C9 and rhCYP2D6 in total. On the
other hand, it may not be surprising that CYP3A4 seems to be the one enzyme

756
N. Hagen et al.
showing peculiar kinetics when considering various reports on CYP3A4 complex-
ity such as non Michaelis±Menten kinetics and variability in kinetic behaviour due
to cofactors such as cytochrome b5 or the use of di€erent bu€ers (Korzekwa et al.
1998, Houston and Kenworthy 2000).
The above-mentioned comparisons of rhCYP catalytic activity and substrate
anity to values obtained in HLM agree with earlier published reports where
separate rhCYPs were compared with HLM. Generally, the variation between the
two in vitro models are reported to be up to 3-fold with regard to Km and up to 6-
fold with regard to Vmax (Rettie et al. 1992, Yamazaki and Shimada 1997, Von
Moltke et al. 1998, McGinnity et al. 1999). The higher variability between Vmax is
most likely due to the above-mentioned possible discrepancy between anticipated
and real expression level of the speci®c forms in the HLM used. These variations
do not seem to be increased when the rhCYPs are mixed.
Crespi and Miller (1999) mixed rhCYPs according to another approach where
catalytic activity was used as a marker and the rhCYP mixture was `formulated to
achieve ratios of individual enzyme activities similar to those observed in pooled
HLMs’. The latter authors have reported a reasonable correlation of the catalytic
activity of pooled HLM preparations with catalytic activities of the rhCYP
mixtures generally being 2-fold higher than for HLM. Furthermore, Crespi and
Miller (1999) stated that individually expressed rhCYPs can be `predictably
combined to prepare mixtures whose pro®le of metabolic activities is similar to
pooled HLMs’.
Antioxidant addition
The level of NADPH-cytochrome P450 oxidoreductase (reductase) and/or
other cofactors could be suboptimal, as the catalytic activity is not solely deter-
mined by the CYP itself but, also, by the electron transport partners present. For
example, the reductase/CYP ratio is higher in expression systems compared with
liver microsomes (Crespi and Miller 1999) thereby increasing the possibility of
accumulation of reactive oxygen species. Calculated from data of Venkatakrishnan
et al. (2000), the reductase to CYP ratio is about 1:5 or more in HLM depending
on the CYP level/induction state compared with the typical 1:2 or more ratio in the
E. coli expression system. This renders the reductase to be the limiting component
in HLM, as the di€erent CYP forms may have to compete for the reductase.
However, reductase is known to transfer electrons much faster than CYP can use
them (Parkinson 1996) and could thereby make up for the di€erence between CYP
and reductase expression. Furthermore, the low reductase level may be a protec-
tive measure by the cells to avoid one-electron reduction reactions catalysed by
reductase (Parkinson 1996). The possible accumulation of reactive oxygen species
could also be caused by uncoupling of the catalytic cycle, e.g. oxygen would be
released as a superoxide anion if the catalytic cycle was interrupted following
introduction of the ®rst electron (Parkinson 1996). Such uncoupling could be
speculated to be more likely to occur in the `foreign’ environment of an E. coli
membrane than in HLM, the native environment of human CYP.
Addition of an antioxidant to the rhCYPs and the rhCYP mixture was therefore
expected to result in an increase in enzyme stability, as the negative in¯uence of
accumulating free radicals would be suppressed. Thereby, the Vmax for the
reactions would be anticipated to increase due to more active enzyme available.

Characterization of recombinant human CYP
757
rhCYP2C9 and rhCYP2D6 both exhibited an increased Vmax when the anti-
oxidant ascorbate was added to the incubation mixture (®gure 1). The increase was
about 75% for S-warfarin 7-hydroxylation and about 50% for dextromethorphan
O-demethylation. Thereby, rhCYP2C9 Vmax was further increased to a higher
level than observed in HLM; rhCYP2D6 Vmax was still 2-fold below the HLM
Vmax level. These changes all ®t with the hypothesis of the free radical suppression
described above. Km was also a€ected demonstrating a 2-fold increase for both
rhCYP2C9 and rhCYP2D6. In contrast, rhCYP3A4 Vmax in testosterone incuba-
tions was not a€ected by the addition of ascorbate, although the reductase activity
of the rhCYP3A4 preparation was slightly higher than for the other two rhCYP
preparations. The anity of rhCYP3A4 was, on the other hand, slightly increased
(table 3).
The e€ect of addition of antioxidant could possibly be related to incubation
time, i.e. there was no e€ect towards rhCYP3A4 catalytic activity (5-min incuba-
tion) whereas rhCYP2C9 catalytic activity (15-min incubation) was substantially
a€ected. However, there was no increase of the e€ect of ascorbate when the
incubation time was prolonged (®gure 2). Therefore, the e€ect of ascorbate
appears to be related to the speci®c CYP form or speci®c substrate metabolism
rather than the experimental procedure.
In conclusion, Km and Vmax determinations in rhCYP mixtures produce results
comparable with those obtained in HLM. The rhCYP mixture would, therefore,
even with more CYP forms incorporated, be an in vitro model to use in early drug
discovery screenings of metabolic stability with the advantages of consisting of
human enzymes, being lower-cost, well-de®ned, reproducible and biohazard-free.
Finally, addition of antioxidant can, in certain cases, be used as a means to obtain
an increase in Vmax.
Acknowledgements
Portions of the paper were presented at the 9th and 10th North American
ISSX Meetings in October 1999 and October 2000.
References
Bajpai, M. and Adkison, K. K., 2000, High-throughput screening for lead optimization: a rational
approach. Current Opinions in Drug Discovery and Development, 3, 63±71.
Crespi, C. L. and Miller, V. P., 1997, The R144C change in the CYP2C9*2 allele alters interaction of
the cytochrome P450 with NADPH:cytochrome P450 oxidoreductase. Pharmacogenetics, 7, 203±
2
10.
Crespi, C. L. and Miller, V. P., 1999, The use of heterologously expressed drug metabolizing enzymes
state of the art and prospects for the future. Pharmacology and Therapeutics, 84, 121±131.
±
Crespi, C. L. and Penman, B. W., 1997, Use of cDNA-expressed human cytochrome P450 enzymes to
study potential drug±drug interactions. Advances in Pharmacology, 43, 171±188.
Dai, D., Tang, J., Rose, R., Hodgson, E., Bienstock, R. J., Mohrenweiser, H. W. and Goldstein,
J. A., 2001, Identi®cation of variants of CYP3A4 and characterization of their abilities to
metabolise testosterone and chlorpyrifos. Journal of Pharmacology and Experimental
Therapeutics, 299, 825±831.
Eddershaw, P. J. and Dickins, M., 1999, Advances in in-vitro drug metabolism screening.
Pharmaceutical Science and Technology Today, 2, 13±19.
Eiselt, R., Domanski, T. L., Zibat, A., Mueller, R., Presecan-Siedel, E., Hustert, E., Zanger,
U. M., Brockmoller, J., Klenk, H.P., Meyer, U. A., Khan, K. K., He, Y.A., Halpert, J. R.
and Wojnowski, L., 2001, Identi®cation and functional characterization of eight CYP3A4
protein variants. Pharmacogenetics, 11, 447±458.

758
N. Hagen et al.
Guengerich, F. P., 1995, Human cytochrome P450 enzymes. In P. R. Ortiz de Montellano (ed.),
Cytochrome P450: Structure, Mechanism, and Biochemistry, 2nd edn (New York: Plenum), pp.
4
73±535.
Gut, J., Catin, T., Dayer, P., Kronbach, T., Zanger, U. and Meyer, U. A., 1986, Debrisoquine/
sparteine-type polymorphism of drug oxidation. Journal of Biological Chemistry, 261, 11734±
1
1743.
Houston, J. B. and Kenworthy, K. E., 2000, In vitro±in vivo scaling of CYP kinetic data not consistent
with the classical Michaelis±Menten model. Drug Metabolism and Disposition, 28, 246±254.
Inoue, K., Yamazaki, H., Imiya, K., Akasaka, S., Guengerich, F. P. and Shimada, T., 1997,
Relationship between CYP2C9 and CYP2C19 genotypes and tolbutamide methyl hydroxylation
and S-mephenytoin 4 -hydroxylation activities in livers of Japanese and Caucasian populations.
0
Pharmacogenetics, 7, 103±113.
Jones, D. R., Gorski, J. C., Hamman, M. A. and Hall, S. D., 1996, Quanti®cation of
dextromethorphan and metabolites: a dual phenotypic marker for cytochrome P450 3A4/5 and
2
D6 activity. Journal of Chromatography B, 678, 105±111.
Jung, F., Richardson, T. H., Raucy, J. L. and Johnson, E. F., 1997, Diazepam metabolism by cDNA-
expressed human 2C P450s. Drug Metabolism and Disposition, 25, 133±139.
Kimura, M., Ieiri, I., Mamiya, K., Urae, A. and Higuchi, S., 1998, Genetic polymorphism of
cytochrome P450s, CYP2C19, and CYP2C9 in a Japanese population. Therapeutic Drug
Monitoring, 20, 243±247.
Kohler, D., H
A
È
rtter, S., Fuchs, K., Sieghart, W. and Hiemke, C., 1997, CYP2D6 genotype and
phenotyping by determination of dextromethorphan and metabolites in serum of healthy
controls and of patients under psychotropic medication. Pharmacogenetics, 7, 453±461.
Korzekwa, K. R., Krishnamachary, N., Shou, M., Ogai, A., Parise, R. A., Rettie, A. E., Gonzalez,
F. J. and Tracy, T. S., 1998, Evaluation of atypical cytochrome P450 kinetics with two-substrate
models: evidence that multiple substrates can simultaneously bind to cytochrome P450 active
sites. Biochemistry, 37, 4137±4147.
È
Lang, D. and B
O
cker, R., 1995, Highly sensitive and speci®c high-performance liquid chromatographic
analysis of 7-hydroxywarfarin, a marker for human cytochrome P4502C9 activity. Journal of
Chromatography B, 672, 305±309.
Lewis, D. F. V., 1998, The CYP2 family: models, mutants and interactions. Xenobiotica, 28, 617±661.
Li, D. N., Pritchard, M. P., Hanlon, S. P., Burchell, B., Wolf, C. R. and Friedberg, T., 1999,
Competition between cytochrome P450 isozymes for NADPH-cytochrome P-450 oxidor-
eductase a€ects drug metabolism. Journal of Pharmacology and Experimental Therapeutics, 289,
6
61±667.
Lowry, O. H., Rosebrough, N. J., Farr, A. L. and Randall, R. J., 1951, Protein measurement with the
Folin phenol reagent. Journal of Biological Chemistry, 193, 265±275.
McGinnity, D. F., Griffin, S. J., Moody, G. C., Voice, M., Hanlon, S., Friedberg, T. and Riley,
R. J., 1999, Rapid characterization of the major drug-metabolizing human hepatic cytochrome
P-450 enzymes expressed in Escherichia coli. Drug Metabolism and Disposition, 27, 1017±1023.
McGinnity, D. F., Parker, A. J., Soars, M. and Riley, R. J., 2000, Automated de®nition of the
enzymology of drug oxidation by the major human drug metabolizing cytochrome P450s. Drug
Metabolism and Disposition, 28, 1327±1334.
Moody, G. C., Griffin, S. J., Mather, A. N., McGinnity, D. F. and Riley, R. J., 1999, Fully
automated analysis of activities catalysed by the major human liver cytochrome P450 (CYP)
enzymes: assessment of human CYP inhibition potential. Xenobiotica, 29, 53±75.
Omura, T. and Sato, R., 1964, The carbon monoxide-binding pigment of liver microsomes. Journal of
Biological Chemistry, 239, 2370±2378.
Parkinson, A., 1996, Biotransformation of xenobiotics. In C. D. Klaassen (ed.), Casarett & Doull’s
Toxicology: The Basic Science of Poisons, 5th edn (New York: McGraw-Hill), pp. 113±186.
Pelkonen, O., M enp, J., Taavitsainen, P., Rautio, A. and Raunio, H., 1998, Inhibition and induction
A
È
of human cytochrome P450 (CYP) enzymes. Xenobiotica, 28, 1203±1253.
Pritchard, M. P., Glancey, M. J., Blake, J. A. R., Gilham, D. E., Burchell, B., Wolf, C. R. and
Friedberg, T., 1998, Functional co-expression of CYP2D6 and human NADPH-cytochrome
P450 reductase in Escherichia coli. Pharmacogenetics, 8, 33±42.
Pritchard, M. P., Ossetian, R., Li, D. N., Henderson, C. J., Burchell, B., Wolf, C. R. and
Friedberg, T., 1997, A general strategy for the expression of recombinant human cytochrome
P450s in Escherichia coli using bacterial signal peptides: expression of CYP3A4, CYP2A6, and
CYP2E1. Archives of Biochemistry and Biophysics, 345, 342±354.
Reinerink, E. J., Doorn, L., Jansen, E. H. and Van Iersel, A. A., 1991, Measurement of enzyme
activities of cytochrome P-450 isoenzymes by high-performance liquid chromatographic analysis
of products. Journal of Chromatography, 553, 233±241.

Characterization of recombinant human CYP
759
Rettie, A. E., Korzekwa, K. R., Kunze, K. L., Lawrence, R. F., Eddy, A. C., Aoyama, T., Gelboin,
H. V., Gonzalez, F. J. and Trager, W. F., 1992, Hydroxylation of warfarin by human cDNA-
expressed cytochrome P450: a role for P-4502C9 in the etiology of (S)-warfarin±drug
interactions. Chemical Research in Toxicology, 5, 54±59.
Roberts, S. A., 2001, High-throughput screening approaches for investigating drug metabolism and
pharmacokinetics. Xenobiotica, 31, 557±589.
Shimada, T., Yamazaki, H., Mimura, M., Inui, Y. and Guengerich, F. P., 1994, Interindividual
variation in human liver cytochrome P-450 enzymes involved in the oxidation of drugs,
carcinogens and toxic chemicals: studies with liver microsomes of 30 Japanese and 30
Caucasians. Journal of Pharmacology and Experimental Therapeutics, 270, 414±423.
Van der Hoeven, T., 1984, Assay of hepatic microsomal testosterone hydroxylases by high performance
liquid chromatography. Analytical Biochemistry, 138, 57±65.
Venkatakrishnan, K., Von Moltke, L. L., Court, M. H., Harmatz, J. S., Crespi, C. L. and
Greenblatt, D. J., 2000, Comparison between cytochrome P450 (CYP) content and relative
activity approaches to scaling from cDNA-expressed CYPs to human liver microsomes: ratios of
accessory proteins as sources of discrepancies between the approaches. Drug Metabolism and
Disposition, 28, 1493±1504.
Vermilion, J. L. and Coon, M. J., 1978, Puri®ed liver microsomal NADPH-cytochrome P450
reductase. Journal of Biological Chemistry, 253, 2694±2704.
Voice, M. W., Zhang, Y., Wolf, C. R., Burchell, B. and Friedberg, T., 1999, E€ects of human
cytochrome b5 on CYP3A4 activity and stability in vivo. Archives of Biochemistry and Biophysics,
366, 116±124.
Von Moltke, L. L., Greenblatt, D. J., Grassi, J. M., Granda, B. W., Venkatakrishnan, K.,
Schmider, J., Harmatz, J. S. and Shader, R. I., 1998, Multiple human cytochromes contribute
to biotransformation of dextromethorphan in-vitro: role of CYP2C9, CYP2C19, CYP2D6, and
CYP3A. Journal of Pharmacy and Pharmacology, 50, 997±1004.
White, R. E., 1998, Short- and long-term projections about the use of drug metabolism in drug
discovery and development. Drug Metabolism and Disposition, 26, 1213±1216.
Yamazaki, H. and Shimada, T., 1997, Progesterone and testosterone hydroxylation by cytochromes
P450 2C19, 2C9, and 3A4 in human liver microsomes. Archives of Biochemistry and Biophysics,
346, 161±169.
Yamazaki, H., Inoue, K. and Shimada, T., 1998, Roles of two allelic variants (Arg144Cys and
Ile359Leu) of cytochrome P4502C9 in the oxidation of tolbutamide and warfarin by human liver
microsomes. Xenobiotica, 28, 103±115.