A new standardized electrochemical array for drug metabolic profiling with human cytochromes P450

Article abstract of DOI:10.1021/ac200309q

Over the past two decades, a wealth of information on the human cytochrome P450 enzymes and their role in drug metabolism both in vitro and in vivo has been gathered. Our understanding of this area has progressed greatly, but our confidence in the develop

Full text of DOI:10.1021/ac200309q

ARTICLE
pubs.acs.org/ac
A New Standardized Electrochemical Array for Drug Metabolic
Profiling with Human Cytochromes P450
Andrea Fantuzzi,‡ Lok Hang Mak,‡ Ennio Capria,‡ Vikash Dodhia,‡ Paola Panicco,‡ Stephen Collins,§ and
Gianfranco Gilardi*,^
‡Division of Molecular Biosciences, Imperial College London, Biochemistry Building, South Kensington, London, SW7 2AY,
United Kingdom
§NanoBioDesign Ltd., Woodstock House, Winch Road, Kent Science Park, Sittingbourne, Kent, ME9 8EF, United Kingdom
^Department of Human and Animal Biology, University of Torino, Via Accademia Albertina 13, Torino, Italy
ABSTRACT: Over the past two decades, a wealth of informa-
tion on the human cytochrome P450 enzymes and their role in
drug metabolism both in vitro and in vivo has been gathered.
Our understanding of this area has progressed greatly, but our
confidence in the development of quantitative projections of
drug interactions, made from in vitro data, is somehow still
shaky. There are therefore no doubts in the necessity for reliable
and fast methodologies for P450 drug metabolism analysis,
capable of providing accurate and precise in vitro data. This
paper reports on the first integration of a P450-electrode into a microtiter plate format for the rapid determination of the affinity
parameters (KM) for a set of known drugs. The most relevant human drug metabolizing cytochromes P450, isoforms 3A4, 2D6, and
2C9, have been covalently bound to a gold electrode via a 10-carboxydecanethiol and 8-hydroxyoctanethiol (1:1) self-assembled
monolayer at the bottom of an eight-well microtiter plate. The electrochemical response of the P450-electrode and the performance
of the platform have been validated using a set of 30 known drugs with KM values spanning from less than 1 to more than 100 μM.
The KM values obtained using this platform show an excellent error, and their ranking is within the range of those present in the
literature determined from conventional incubation experiments with cytochrome P450s 3A4, 2D6, and 2C9.
uman cytochromes P450 are responsible for the metabo-
lism of the large majority of known drugs.1 Characterization
limitations and that, thanks to the full metabolic characterization,
is more information rich, but also more costly and slow, is the use
of mass spectrometry.4 One of the major challenges encountered
when screening drugs for metabolic studies is that a great variety
of chemicals with distinct molecular structures must be analyzed
using a generic method. For a method to achieve high-through-
put rates, the measurements of the output signal should be
preferably compound-independent and the data acquisition must
be rapid. This has proved to be challenging for the mass
spectrometry-based methodologies, though progress is con-
stantly being made, and sparked the need for investigating
alternative methodologies more likely to be high throughput.4
Recombinant enzymes, microsomes, hepatocytes, and liver
slices have all been used as in vitro model systems for testing the
metabolic stability of new drug candidates leading to a certain
controversy when attempting to compare data obtained using
enzymes from different sources.4 Human liver microsomes have
in the past provided the most convenient way to study drug
metabolism. They consist of a subcellular fraction of tissue
obtained by differential high-speed centrifugation and can be
obtained commercially, with or without prior phenotyping, for most
H
of the specific metabolic pathways and of the P450 isoenzymes
involved constitutes an essential part of the drug development
process. There are 11 major drug-metabolizing P450s that are
expressed in the human liver, and among these the isoforms, 3A4,
2D6, and 2C9 are responsible for the metabolism of nearly 90%
of all drugs.2 Metabolic stability of a drug is a major factor that
will ultimately determine the concentration of the drug found in
the systemic circulation. Since the practice of simultaneous pre-
scription of more than one drug to treat one or more conditions
in a single patient has become a common practice, drugꢀdrug
and drugꢀP450 interactions have been cited as one of the major
reasons for hospitalization and even death.3 Many drugꢀdrug
interactions are metabolism based and result from two or more
drugs competing for the same enzyme that in the majority of the
cases involves P450s.4 Therefore, it has become increasingly
important to develop methods that enable screening for meta-
bolic stability and P450 inhibition in a simple, rapid, high-
throughput format. At present, high-throughput screening of
drugꢀP450 interaction is achieved using a competitive fluores-
cence assay with fluorogenic substrates.5,6 This approach, based
on a competitive reaction, can only yield indirect values for the
kinetic parameters and is unable to distinguish between sub-
strates and inhibitors. An approach that does not possess these
Received: February 4, 2011
Accepted: April 6, 2011
Published: April 06, 2011
r
2011 American Chemical Society
3831
dx.doi.org/10.1021/ac200309q Anal. Chem. 2011, 83, 3831–3839
|

Analytical Chemistry
ARTICLE
important drug metabolizing enzymes. All of the P450 enzymes are
collected together in the microsomal fraction. Microsomes from
different donors are used, either individually or pooled together, to
avoid reliance on microsomes that are deficient in one or more
metabolic pathways. With the use of selective chemical inhibitors for
each major pathway, the metabolic pathways for a new drug can be
demonstrated or ruled out. With the cloning of most of the common
P450s,1 the recombinant human enzymes have been expressed in a
variety of cells allowing a more controlled system and becoming the
new standard for drug metabolism and toxicological studies.
Incubations with these enzymes still requires a lengthy, multistep
procedure followed by either fluorescence or mass spectrometry
analysis.
The reaction catalyzed by cytochromes P450 involves the use of
two reducing equivalents donated by NADPH via a reductase.
Electrodes can be used as an alternative source of electrons in place
of NADPH: here an electric current is measured with an intensity
that is proportional to the rate of substrate turnover by the P450
enzymes. Reports can be found in the most recent literature on the
electrochemical characterization and potential exploitation of P450
for the characterization of specific substrates.7ꢀ20 Our laboratory has
previously reported the successful covalent immobilization of
cytochromes P450 on gold electrodes by using self-assembled
monolayers that exploited different methods to tether the protein
to the surface, and their activity was demonstrated in terms of
product formation.7,18ꢀ21 The bioelectrochemistry of cytochromes
P450 has been adopted in some cases for the determination of
unknown quantities of specific analytes in a classical biosensor setup.
In these approaches the sensors were designed to work with a linear
response to increasing concentration of substrate with the specific
aim of determining the unknown concentration of such analyte in
biological samples.22ꢀ27 The novelty of the approach presented in
this work stands on the design of a system that responds hyperbo-
lically to increasing concentrations of substrate, following the
enzymatic MichaleisꢀMenten model that allows the determination
of the substrate-specific parameters such as KM that is important in
guiding the determination of the dosage for each drug.
In this paper we present the construction and characterization of
an array that draws together the requirements for a high-throughput
methodology with output that does not require substrate-dependent
analytical methods. It uses a controlled and reproducible source of
recombinant human cytochrome P450 isoforms, enabling the
investigation of the specific contribution of different enzymes in a
microtiter plate format where no reagents, reductases, or NADPH
are needed, but simply requires increasing concentrations of pure
drug for titration in the well.
The platform presented in this paper is not a biosensor
intended to detect an analyte in a complex mixture, but rather
a device for the identification of specific metabolic parameters
(KM) of pure samples of new potential drugs in controlled
experimental conditions. For this reason, a set of drugs, used as
pure samples, with KM values for the three P450s ranging over 1
order of magnitude, represents the ideal set for the validation of
the array.
NBS Biological Ltd. (UK). All other chemicals were obtained from
Sigma and were of highest purity available and used as received. All
solutions were prepared using Milli-Q water purifying system
(Millipore, UK)
Array Preparation. The surface of the gold working electrode
was cleaned before modification with the self-assembled mono-
layer (SAM) using oxygen plasma. The array was placed in a
K1050X Plasma Treatment Unit (Emitech, UK), and oxygen
plasma with 90 W RF was applied for 10 min. Subsequent to the
plasma treatment, ethanol was added to each well and incubated
for 20 min. A 200 μL amount of ethanolic solution containing
1 mM CDT and 1 mM HOT was added to each well and
incubated for 48 h. The excess solution was then removed, and
the electrodes washed three times with ethanol to remove the
unbound alkanethiols. The ethanol was removed, and 250 μL of
water was added to each well to wash out the ethanol. This
washing step was repeated twice. A 200 μL amount of 2-(N-
morpholino)ethanesulfonic acid (MES) buffer containing 1 M
NaCl (pH 6) was added to each well. A 200 μL amount of 25 mM
1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride
(EDC) and 25 mM N-hydroxysulfosuccinimide (S-NHS) dis-
solved in 50 mM MES buffer was added to each well to activate
the SAM’s carboxylic groups. After 20 min, the EDC solution was
removed and washed twice with water. Protein solution of either
P450 3A4, 2D6, or 2C9 was diluted with 50 mM phosphate
buffer containing 500 mM potassium chloride (pH 7.4) to the
concentration of 5 μM. Ten microliters of protein solution was
added to each well, thereby making sure the whole working
(gold) electrode was covered by the protein solution. The array
was incubated with the protein for 1 h at 4 ꢀC in a high-humidity
chamber to limit evaporation of the droplet. After the protein
immobilization, the electrodes were washed three times with
phosphate buffer and stored in the same buffer until use.
Electrochemical Measurements. No commercially available
electrochemical arrays with the desired characteristics of gold
working, carbon counter, and silver/silver chloride reference
electrodes in each well are available. Two electrode systems are
commercially available, but none have reference electrodes in
each well. Therefore, it was chosen to design and construct an
array with the desired characteristics as described in Results and
Discussion. The choice of materials and design was guided by
integration with the existing technology, electrodes design,
disposability, and materials compatibility with the final goal of
constructing a disposable array with a cost predicted to be no
more than 10 US$ per unit. The array was manufactured
according to the specifications by Conductive Technologies
Inc. (York, PA).
Furthermore, only few multichannel potentiostats are equip-
ped with more than 8 independent channels and in general no
more than 16. The commercially available potentiostats with a
higher number of working electrodes normally have a common
reference for all the working electrodes. This configuration is not
ideal for our application; therefore, all the electrochemical
measurements were performed using an eight-channel multi-
potentiostat PG580-RM commercially available from Uniscan
(Uniscan, UK) interfaced to a PC. This instrument is capable of
controlling eight independent working, counter, and reference
electrodes, making it ideal for our application. The array’s
connector was custom designed and constructed in collaboration
with Uniscan, allowing easy connection of the array to the
potentiostat (Figure 1). All potentials are reported versus the
0.5 M Ag/AgCl reference electrode. Data fittings were carried
’ MATERIALS AND METHODS
Materials. Cytochrome P450 3A4, 2D6, and 2C9 were
expressed and purified as previously described.28 The proteins
were stored at ꢀ20 ꢀC in 50 mM phosphate buffer, 500 mM
potassium chloride, and 50% glycerol until use. 10-carboxydecanethiol
(CDT) and 8-hydroxyoctanethiol (HOT) were purchased from
3832
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
out using a custom-made macro that integrates with the acquisi-
tion software of the Uniscan multichannel potentiostat. The
macro was written in collaboration with Uniscan and is based on
a MichaelisꢀMenten model.
Cyclic Voltammetry. Insulation properties of SAM-modified
electrodes were tested by cyclic voltammetry from ꢀ0.2 V to
þ0.65 V using a solution containing 10 mM potassium
hexacyanoferrate(III) (K3[Fe(CN)6]) with 1 M KCl as support-
ing electrolyte.
Cyclic voltammograms of P450 modified arrays were per-
formed under anaerobic conditions in a Belle technology glove-
box. Buffer solutions containing 50 mM phosphate and 500 mM
potassium chloride (pH 7.4) were used. The scan range was from
0 V to ꢀ0.65 V versus Ag/AgCl.
Titration of P450 with Specific Drug Substrates. Electroche-
mical measurements were carried out at 25 ꢀC in 50 mM potassium
phosphate pH 7.4 containing 500 mM potassium chloride (working
buffer). Drugs were purchased at the purest grade from Sigma
Aldrich and were dissolved in the appropriate solvent: either
methanol, ethanol, or DMSO. The choice of the appropriate solvent
was made according to the solubility of the specific drug and the
potential interference of the solvent with the P450 metabolism.
Stock solutions of 1, 10, and 50 mM of the drugs were used to
generate the desired dilutions into the aforementioned buffer.
The prepared arrays with the immobilized P450 containing
200 μL of 50 mM phosphate and 500 mM potassium chloride
(pH 7.4) buffer were inserted in the specifically designed
connector fitted on the platform of a microtiter plate shaker.
The wells were left uncovered for full oxygenation of the buffer
during the titration.
Figure 1. (A) Scheme of the design of the electrode array on the PET
layer. The materials used for the construction of the array are color
coded: yellow for the gold of the working electrode and the track; blue
for the carbon ink of the counter and of the contacts pads; cyan for the
Ag/AgCl ink of the reference electrode; gray for the silver ink used for
the connecting tracks. (B) Photograph of the array from the top. It is
possible to observe the polypropylene housing and the green layer of
insulation ink sandwiched between the housing and the polyethylene
terephthalate (PET)/contacts layer. The comb structure of the contacts
is also noticeable. (C) Side view of the array housed in the holder for
measurement.
An initial cyclic voltammetry scan was recorded before any
addition to determine the catalytic current in resting state. The
substrate was then added with a multichannel pipet in 1 μL
aliquots to minimize the addition of large quantities of organic
solvents. After each addition, the array was stirred with the plate
shaker and rested for 1 min to allow for substrate binding to the
enzyme. Cyclic voltammetry was then performed after each
addition. All CVs were recorded between 0 and ꢀ0.65 V vs
Ag/AgCl. Control experiments were carried out by titrating
equivalent volumes of buffer containing the solvents used to
dissolve the drugs. In a second set of control experiments, arrays
modified with only the SAM but no protein were used and
titrated with the drugs.
because precise control is needed over the applied potential in order
to limit the adverse effect of nonspecific reactions, mainly associated
with oxygen, at the electrode surface. Therefore, it was decided to
keep the three-electrode configuration in each well with every
electrode independently wired to a multichannel potentiostat
(Figure 1).
The electrode materials are crucial for the performance of the
amperometric sensor and were therefore chosen with care. The
working electrode was constructed by laser ablation of the
desired features, working area, and contacts, from a sheet of gold
(50 nm thick) sputtered on polyethylene terephthalate (PET).
The choice of gold stems from the successful immobilization of
several P450s onto gold surfaces modified with a self-assembled
monolayer (SAM),7,18,20,29 while the low cost of the sputtered
material makes it ideal for a disposable consumable. The
reference and counter electrodes were instead built by screen
printing methodologies on the PET support. Inks of carbon and
silver/silver chloride were used for the counter and reference
electrodes, respectively, while silver was used to draw the
contacts (Figure 1). An initial design making use of a gold
counter electrode fabricated by laser ablation together with the
working electrode was adopted to minimize the number of
fabrication steps but was abandoned because of the instability
of such electrodes in buffers containing a high concentration of
chlorides. A layer of insulating ink was printed in the area outside
the wells in order to improve the adhesion of the PET support to
the polypropylene housing.
’ RESULTS AND DISCUSSION
Design and Construction of the Array. The design of the
array was driven by the following considerations: (1) need of
integration with the existing automated fluidic technologies existing
in the pharma industry; (2) the constraints in the three-electrode
design; (3) the disposability of the array; (4) the compatibility
between the materials used. Most of the P450 assays used in the
pharma industry are run in microtiter plates. It was therefore decided
to construct an eight-well array with the standard dimensions of a
row of a 96-well microtiter plate. The housing forming the eight
wells was molded in polypropylene (Figure 1). Other materials,
such as polycarbonate and polystyrene, were also tested, but
polypropylene was chosen because of its excellent properties of
chemical resistance and biocompatibility.
The geometric layout of the electrodes on the bottom of the
well was chosen in order to maintain the working electrode disk
at the center of the well, the reference electrode on the left side,
and the counter electrode extending for about two-thirds of the
In all previously published electrochemical applications of
cytochromes P450, a three-electrode configuration, working, coun-
ter, and reference, has been adopted as the set up of choice7,18,29
3833
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
circumference. Positioning the working electrode at the center of
the well facilitates the deposition of the protein during the
preparation of the array.
function not only as the anchoring point for the protein but also
as a barrier limiting the diffusion of molecules to the gold surface.
For this reason, a long chain alkenthiol was used to improve the
insulation properties of the SAM, as the strong van der Waals
interactions between the long aliphatic chains will make a
compact and uniform layer (Figure 2).
Figure 3A shows the cyclic voltammograms (CVs) obtained
at 50 mV/s, in a solution of 10 mM K3[Fe(CN)6] in 1 M KCl,
on a bare gold electrode and on an electrode on which the
SAM was immobilized. The dramatic disappearance of the
peaks in the second sample modified with the 10-carboxyde-
canethiol and 8-hydroxyoctanethiol (SAM, C8/C10) is a con-
firmation of the presence of an insulating layer on the gold
surface. The self-assembled monolayer is creating a barrier that
limits the diffusion of the redox molecules from the solution to
the electrode but allows electron transfer to and from the
protein by covalently binding the P450 enzyme to the elec-
trode surface at a distance that is compatible with electron
tunneling to the heme. Diffusion-controlled electron transfer
is therefore limited in favor of the tunneling from the cova-
lently bound protein.
Following the immobilization of the P450 onto the (C8/C10)
SAM-modified gold surface, the electrochemical activity of the
bound protein was investigated by CV in anaerobic conditions.
Figure 3B shows the CV recorded in the absence of oxygen for
P450 3A4 immobilized onto the C8/C10 modified gold electrode
in one of the wells of the array. The background signal of the
solely SAM modified electrode is represented for comparison
(dotted line). The CV of the bound enzyme shows two peaks
at ꢀ480 mV and ꢀ290 mV with a resulting midpoint potential
of ꢀ395 mV versus 0.5 M Ag/AgCl. These values are in the range
of those reported in the literature for P450 on different electrode
and different immobilization approaches.10
The stability of the screen printed reference electrode was
tested. Its potential stability was measured in the working buffer
containing 0.5 M KCl using a saturated calomel electrode as
standard. A medium value of the potential of 7 ( 0.2 mV with a
drift of 0.25 mV hꢀ1 was recorded for 24 h. An elapse time of
20 min was measured due to electrode hydration. The long-term
stability was checked by measuring the potential value in a 0.5 M
KCl solution, and it was found to be stable for more than two
weeks. Furthermore, the mechanical stability of the printed
electrode was investigated before and after plasma treatment
and incubation in the ethanolic solution. The DuPont ink 5870
was found to possess the best combination of mechanical and
electrochemical stability.
Electrochemical Characterization of the Array. As the P450
catalytic reaction is driven by the application of a negative
potential bias, some of the drugs that have electroactive proper-
ties could be directly reduced at the gold surface. Therefore, we
identify this as the only potential source of interference because
the array platform is intended to be titrated with increasing
concentrations of the pure drug. Nevertheless, in order to
minimize this possible interference, the SAM was deemed to
Testing of the Array. The pharmaceutical industry has shown
a high interest in the ability to determine the KM in vitro and in a
high throughput way because the drug’s specific KM is an
important parameter that contributes to the determination of
the dosage in vivo. At present, KM values are obtained from
lengthy and costly LC-MS measurements carried out after
incubation of the drug with a P450 reconstituted system, i.e.,
P450 plus a reductase and NADPH.
Figure 2. Scheme of the electrode modification. Deposition of 10-
carboxydecanethiol and 8-hydroxyoctanethiol on the gold electrode to
form a compact self-assembled monolayer, followed by the activation of the
carboxylic groups with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide
hydrochloride (EDC) and N-hydroxysulfosuccinimide (S-NHS). The
P450 is then covalently linked via its surface amino groups.
In our previous work we have shown that all the P450 chosen
for this work, 3A4, 2D6, and 2C9, respond in the presence of
Figure 3. (A) Cyclic voltammetry of a solution of 10 mM K3[Fe(CN)6] in 0.1 M KCl on bare gold (dotted) and on 10-carboxydecanethiol and
8-hydroxyoctanethiol modified gold (solid line). Scan rate was 50 mV/s. (B) Cyclic voltammetry recorded from one well in the absence of oxygen using
the 10-carboxydecanethiol and 8-hydroxyoctanethiol modified electrode with (solid line) and without (dotted line) bound P450 3A4. The CVs were
recorded at 50 mV/s in 50 mM phosphate and 500 mM potassium chloride (pH 7.4) buffer.
3834
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
Figure 4. (A) Cyclic voltammetry recorded from one well of the electrode modified with 10-carboxydecanethiol and 8-hydroxyoctanethiol with bound
P450 3A4 in the absence (black thin line) and in the presence (red line) of oxygen. The blue line represents the catalytic current recorded in the presence
of oxygen and saturating concentrations (50 μM) of quindine. The CVs were recorded at 50 mV/s in 50 mM phosphate and 500 mM potassium chloride
(pH 7.4) buffer. (B) Δ current as a function of the applied potential obtained by subtracting the CV recorded prior to any addition of substrate from the
CVs recorded upon addition of substrate. The arrow indicates the direction of the catalytic current during the titration.
Figure 5. Screenshot example of the data analysis carried out by the custom-made macro in the Uniscan software. The eight panels represent the
catalytic currents recorded in each individual well and plotted as a function of the added substrate. The data points are fitted with the MichaelisꢀMenten
model, and the best fit is represented by the blue line. The fitting parameters are listed in the bottom right panel. All the wells contain the immobilized
P450 2C9. Well number 5 and 6 are the control wells where equivalent volumes of buffer and solvent but no substrate are added. In all the other wells,
fluoxetine was added and the concentrations (in μM) are indicated on the x-axis. The fitting parameters are given in the table on the bottom right. The
V
max and KM values are expressed in nA and μM respectively.
specific substrates with catalytic currents that increase as a
function of the concentration of substrate, with a hyperbolic
behavior in a MichaelisꢀMenten fashion.18,21 Figure 4A shows
the CVs recorded in the presence of oxygen and with saturating
concentrations of substrate for P450 3A4 immobilized onto the
C8/C10 modified gold electrode in one of the wells of the array. The
CV recorded in the absence of oxygen is also shown for comparison.
The measured cathodic current increases significantly as expected
3835
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
Figure 6. MichaelisꢀMenten plot based on the current measured at ꢀ390 mV upon addition of aliquots with increasing concentrations of substrate.
(A) P450 3A4 titrated with quinidine, (B) P450 2D6 titrated with propranolol, and (C) P450 2C9 titrated with diclofenac. Structures of the substrates
and products formed in the respective P450-catalyzed reactions are also shown.
from the consumption of electrons in the P450 catalytic cycle as
represented by the overall reaction mechanism in eq 1.
are often indicated as a range because very often incubations with
cytochromes P450 carried out on the same substrate but by
different groups yield different results. One of the sources of
variability has been associated with the source of the P450
enzymes used to determine the kinetic data, for example purified
P450, microsomes, or hepatocytes. For this reason, the source of
the enzyme used in the cited reference has been added to Table 1.
Another source of variability are the conditions of the assay, for
example the ratio between the P450 and its reductase. The range
of values found in the literature, often quite large, is the
motivation in the use of the standardized electrochemical array
proposed here but at the same time makes it difficult to directly
compare the results. For these reasons here it is proposed to look
at these data in terms of ranking of the various substrates and not
as absolute values.
From this point of view, the KM values measured with the array
rank the substrates in accordance to the ranking obtained from the
literature values. This indicates the ability of the array to correctly
determine the affinity of drugs for the specific P450, opening the
possibility for its use in the early stage drug discovery process.
Array validation in terms of intra- and interarray variability was
carried out for all the substrates. Precision is more difficult to
evaluate, as there is no accepted gold standard in the literature
where often imprecise values of KM are accepted. If a more
reliable method, such as the one proposed in this work, will be
adopted, it will be possible to attain a set of data on which
precision evaluation can be carried out. We therefore ranked and
clustered the compounds according to defined ranges, and we
used the standard deviation as indication of the performance of
the assay. For each substrate, the standard deviation calculated
across the wells of an array compare well with the standard
deviation between wells of different arrays. The standard devia-
tions indicated in Table 1 for each substrate were determined
from the titrations in 24 individual wells (4 arrays) and are an
indicator of both intra- and interarray variability. Titrations were
performed as described in Materials and Methods , and in each
array two wells were dedicated to the control experiments. In the
control experiments, aliquots of buffer containing the same solvent
used to dissolve the drugs were added to the well containing the
RH þ O2 þ 2Hþ þ 2eꢀ f ROH þ H2O
ð1Þ
Figure 4B shows the current/voltage profiles as a function of
increasing concentrations of substrate recorded in one well and
represented as the current difference by subtracting at each CV
the one recorded before any addition of substrate. The maximum
difference in catalytic current occurs at values below ꢀ390 mV.
Plotting the values of catalytic current at ꢀ390 mV as a
function of the concentration of added substrate shows a
hyperbolic behavior that can be fitted with a MichaelisꢀMenten
model. Figure 5 shows an example, as a screenshot, of the output
recorded and analyzed from the eight wells. Using a custom-
made macro written in collaboration with the manufacturer of
the multichannel potentiostat, Uniscan Ltd., the series of CVs
recorded from each channel in each well are analyzed in parallel.
The catalytic currents are plotted as a function of the concentra-
tion of the added substrate, and the values are fitted with the
MichaelisꢀMenten model. The fitting parameters are displayed
together with the R2 of the fitting.
The data obtained from the titrations with the same substrate
in different wells were collated and refitted to improve the
statistical significance of the determined KM. Three examples
of collated data obtained from the titration of three different
substrates to their respective P450 are shown in Figure 6. The
values of current are expressed as the current difference, and the
structure of the products produced during the catalytic reaction
are represented.
The data points were fitted to a MichaelisꢀMeneten model,
and the calculated KM values are presented in Table 1. The
measured KM values for each substrate (Table 1) are compared
to the values reported in the literature.
These drugs are all known substrates of either P450 3A4, 2C9,
or 2D6 for which KM have been determined in vitro. They have
been chosen to span a range of affinity from high (from less than
1 to 10 μM) to low (from 10 to more than 100 μM) for each
respective P450 enzyme. The values reported from the literature
3836
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
Table 1. MichaelisꢀMenten Constants (KM) for P450 3A4, 2D6, and 2C9 with Their Respective Specific Substrates Determined
Using the Microtiter Plate Electrochemical Platforma
literature
substrate
system used
range of KM (μM) [reference]
electrochemistry KM (μM)
P450 3A4
high affinity
low affinity
cyclosporine A
terfenadine
quinidine
insect microsomes
1.2ꢀ2.230
7.32ꢀ14.131
1.532
1.8 ( 0.8
3.8 ( 2.2
3A4/CPR fusion
Iinsect microsomes, HLM
insect microsomes HLM
4.1 ( 0.6
domperidone
omeprazole
chlorzoxazone
verapamil
3ꢀ15.733
7.1 ( 1.9
insect and human liver microsomes
HLM
6034
14.6 ( 4.8
15.3 ( 6.5
39.0 ( 6.7
41.1 ( 1.3
43.7 ( 14.2
206.4 ( 29.4
16035
microsomes
5.4ꢀ12736
100ꢀ23037
81ꢀ12838
1400ꢀ250039
clozapine
Insect microsomes
yeast microsomes, HLM
HLM microsomes
propafenone
theophylline
P450 2D6
high affinity
low affinity
perhexiline
fluoxetine
HLM
3.340
2.4 ( 0.6
4.4 ( 1.1
HLM
1.4ꢀ3.841
5.9ꢀ31.942
0.543
propranolol
propafenone
clozapine
HLM
4.6 ( 0.6
HLM
7.1 ( 2.1
baculosomes reconstituted system
human insect microsomes
yeast microsomes
mice baculosomes
HLM baculosomes
HLM
2537
9.3 ( 1.2
sparteine
54.8ꢀ6444
19ꢀ4645
2.8ꢀ2246
12.147 13046
150048 15046
9.7 ( 1.0
16.5 ( 2.8
33.0 ( 5.4
62.7 ( 8.2
204.2 ( 27.8
metoprolol
dextromethorphan
debrisoquine
codeine
P450 2C9
high affinity
low affinity
warfarin
HLM
2.549 1850
1.850 1551
9.741
19.452
27.953
30.552
120ꢀ19254 60ꢀ58046
222ꢀ34355
116ꢀ30756 92ꢀ16046
114ꢀ38557
1.3 ( 0.6
4.3 ( 2.4
diclofenac
fluoxetine
HLM
HLM
4.3 ( 1.5
flurbiprofen
ibuprofen
baculosomes
HLM
13.6 ( 5.1
17.5 ( 6.8
36.1 ( 7.1
84.7 ( 6.7
109.1 ( 10.6
117.1 ( 31.3
203.7 ( 23.3
piroxicam
baculosomes
yeast and E.coli microsomes
HLM
tolbutamide
dextromethorphan
naproxen
insect microsomes
HLM
melatonin
a The KM values are compared to the literature data obtained from conventional incubation methodologies. The source of P450 used in the literature is
also indicated.
immoblized P450. The position of the two control wells in the array
was changed in different experiments to avoid the bias toward specific
positions in the array. No catalytic current was detected in these
control experiments, confirming that the measured catalytic current in
the presence of the drug is generated by the P450 turnover.
A second type of control was also performed using unmodified
arrays where only the SAM was deposited and no protein was added.
Using these arrays, the substrates were titrated in order to assess any
potential interference of the drugs on the electrochemical signal. For
all tested drugs, no signal was detected in these experiments, as
expected thanks to the insulating properties of the SAM.
proportional to the turnover rate. Nevertheless, since in the P450
catalytic cycle the reducing equivalents can also be used for
uncoupling reactions where oxygen is converted into superoxide
radicals, hydrogen peroxide, or water, the proportionality factor
that would allow the determination of the kcat from the catalytic
current remains unknown.
The time necessary to obtain a KM value for the compound
under investigation is 15 min. This is a very short time when
compared with typical fluorescent measurements that are nor-
mally limited to end point measurements. Furthermore, assays
making use of fluorescent substrates are based on competitive
inhibition experiments with the drug yielding only indirect
measurements of the affinity for a specific P450. Conversely,
the platform presented in this work allows for a direct, reagent-
less, and fast determination of the KM.
Similarly to what was previously observed in an electrochemi-
cal microfluidic platform, the uncoupled nature of the human
cytochrome P450s prevented the calculation of accurate kcat
values.29 The intensity of current measured during catalysis is
3837
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
’ CONCLUSIONS
(19) Ferrero, V. E. V.; Andolfi, L.; Di Nardo, G.; Sadeghi, S. J.;
Fantuzzi, A.; Cannistraro, S.; Gilardi, G. Anal. Chem. 2008,
80, 8438–8446.
(20) Mak, L. H.; Sadeghi, S. J.; Fantuzzi, A.; Gilardi, G. Anal. Chem.
2010, 82, 5357–5362.
The lack of a gold standard for the high throughput in vitro
determination of the kinetic parameters for P450 drug metabo-
lism results in the inability to predict with confidence the
toxicological and dosage parameters for new potential drugs.
The platform presented in this paper addresses some of these
issues by developing a validated assay based on recombinant
purified P450s, on highly pure reagents, and by having a
disposable and standard format. The results of the titrations
with the specific substrates clearly show how the electrochemical
determination of the KM with the array accurately ranks the
affinity for the different drugs with high precision. Furthermore,
the microtiter plate format allows for easy integration in the
systems currently employed in the early drug discovery process
with ease of handling and incorporation.
(21) Panicco, P.; Dodhia, V. R.; Fantuzzi, A.; Gilardi, G. Anal. Chem.
2011, 83, 2179–2186.
(22) Alonso-Lomillo, M.; Gonzalo-Ruiz, J.; Domínguez-Renedo, O.;
Mu~noz, F. J.; Arcos-Martínez, M. J. Biosens. Bioelectron. 2006,
23, 1733–1737.
(23) Hendricks, N. R.; Waryo, T. T.; Arotiba, O.; Jahed, N.; Baker,
G. L.; Iwuoha, E. I. Electrochim. Acta 2009, 54, 1925–1931.
(24) Ignaszak, A.; Hendricks, N.; Waryo, T. T.; Songa, E.; Jahed, N.;
Ngece, R.; Al-Ahmed, A.; Kgarebe, B.; Baker, P.; Iwuoha, E. I. J. Pharm.
Biomed. Anal. 2009, 49, 498–501.
(25) Shumyantseva, V. V.; Bulko, T. V.; Kuznetsova, G. P.;
Samenkova, N. F.; Archakov, A. I. Biochemistry (Moscow, Russ. Fed.)
2009, 74, 438–444.
(26) Shumyantseva, V. V.; Supurun, E. V.; Bulko, T. V.; Dobrinina,
O. V.; Archakov, A. I. Biochemistry (Moscow, Russ. Fed.) Suppl. 2010,
4, 25–36.
(27) Shumyantseva, V. V.; Bulko, T. V.; Suprun, E. V.; Chalenko,
Y. M.; Vagin, M. Y.; Rudakov, Y. O.; Shatskaya, M. A.; Archakov, A. I.
Biochim. Biophys. Acta 2011, 1814, 94–101.
’ AUTHOR INFORMATION
Corresponding Author
*Phone: þ39-011-6704593. Fax: þ39-011-6704643. E-mail:
gianfranco.gilardi@unito.it.
(28) Dodhia, V. R.; Fantuzzi, A.; Gilardi, G. J. Biol. Inorg. Chem.
2006, 11, 903–916.
(29) Fantuzzi, A.; Capria, E.; Mak, L. H.; Dodhia, V. R.; Sadeghi, S. J.;
Collins, S.; Somers, G.; Huq, E.; Gilardi, G. Anal. Chem. 2010,
82, 10222–10227.
’ ACKNOWLEDGMENT
This research was supported by NanoBioDesign Ltd.
(30) Dai, Y.; Iwanaga, K.; Lin, Y. S.; Hebert, M. F.; Davis, C. L.;
Huang, W.; Kharasch, E. D.; Thummel, K. E. Biochem. Pharmacol. 2004,
68, 1889–1902.
’ REFERENCES
(31) Rodrigues, A. D.; Mulford, D. J.; Lee, R. D.; Surber, B. W.;
Kukulka, M. J.; Ferrero, J. L.; Thomas, S. B.; Shet, M. S.; Estabrook, R. W.
Drug Metab. Dispos. 1995, 23, 765–775.
(1) Ortiz de Montellano, P. Cytochrome P450: Structure, Mechanism,
and Biochemistry, 3rd ed.; Kluwer Academics/Plenum Publishers: New
York, 2005.
(32) Ngui, J. S.; Tang, W.; Stearnis, R. A.; Shou, M.; Miller, R. R.;
Zhang, Y.; Lin, J. H.; Baillie, T. A. Drug Metab. Dispos. 2000,
28, 1043–1050.
(2) Rendic, S.; Di Carlo, F. J. Drug Metab. Rev. 1997, 29, 413–580.
(3) Walsky, R. L.; Obach, R. S. Drug Metab. Dispos. 2004,
32, 647–660.
(33) Ward, B. A.; Morocho, A.; Kandil, A.; Galinsky, R. E.; Flockhart,
D. A.; Desta, Z. Br. J. Clin. Pharmacol. 2004, 58, 277–287.
(34) Yamazaki, H.; Inoue, K.; Shawn, P. M.; Checovich, W. J.;
Guengerich, P. F.; Shimada, T. J. Pharmacol. Exp. Ther. 1997,
283, 434–442.
(35) Shimada, T.; Tsumura, F.; Yamazaki, H. Drug Metab. Dispos.
1999, 27, 1274–1280.
(36) Tracy, T. S.; Korzekwa, K. R.; Gonzalez, F. J.; Wainer, I. W. Br. J.
Clin. Pharmacol. 1999, 47, 545–552.
(4) Ansede, J. H.; Thakker, D. R. J. Pharm. Sci. 2004, 93, 239–255.
(5) Crespi, C. L.; Miller, V. P.; Stresser, D. M. Methods Enzymol.
2002, 357, 276–284.
(6) Cohen, L. H.; Remley, M. J.; Raunig, D.; Vaz, A. D. N. Drug
Metab. Dispos. 2003, 31, 1005–1015.
(7) Fantuzzi, A.; Fairhead, M.; Gilardi, G. J. Am. Chem. Soc. 2004,
126, 5040–5041.
(8) Bistolas, N.; Wollenberger, U.; Jung, C.; Scheller, F. W. Biosens.
Bioelectron. 2005, 20, 24082423.
(37) Linnet, K.; Olsen, O. V. Drug Metab. Dispos. 1997, 25, 1379–1382.
(38) Botsch, S.; Gautier, J.-C.; Beaune, P.; Eichelbaum, M.; Kroe-
mer, H. K. Mol. Pharmacol. 1992, 43, 120–126.
(39) Zhang, Z-Y; Kaminsky, L. S. Biochem. Pharmacol. 1995,
50, 205–211.
(40) Sorensen, L. B.; Sorensen, R. N.; Miners, J. O.; Somogyi, A. A.;
Grgurinovich, N.; Birkett, D. J. Br. J. Clin. Pharmacol. 2003, 55, 635–638.
(41) Ring, B. J.; Eckstein, J. A.; Gillespie, J. S.; Binkley, S. N.;
Vandenbranden, M.; Wrighton, S. A. J. Pharmacol. Exp. Ther. 2001,
297, 1044–1050.
(42) Johnson, J. A.; Herring, V. L.; Wolfe, M. S.; Relling, M. V.
J. Pharmacol. Exp. Therap. 2000, 294, 1099–1105.
(43) Kroemer, H. K.; Botsch, S.; Heinkele, G.; Schick, M. Methods
Enzymol. 1996, 272, 99–105.
(9) Fleming, B. D.; Johnson, D. L.; Bond, A. M.; Martin, L. L. Expert
Opin. Drug Metab. Toxicol. 2006, 2, 581–589.
(10) Sadeghi, S. J.; Fantuzzi, A.; Gilardi, G. Biochim. Biophys. Acta
2011, 1814, 237–248.
(11) Joseph, S.; Rusling, J. F.; Lvov, Y. M.; Friedberg, T.; Fuhr, U.
Biochem. Pharmacol. 2003, 65, 1817–1826.
(12) Shumyantseva, V. V.; Ivanov, Y. D.; Bistolas, N.; Scheller, F. W.;
Archakov, A. I.; Wollenberger, U. Anal. Chem. 2004, 76, 6046–6052.
(13) Johnson, D. L.; Lewis, B. C.; Elliot, D. J.; Miners, J. O.; Martin,
L. L. Biochem. Pharmacol. 2005, 69, 1533–1541.
(14) Peng, L.; Yang, X.; Zhang, Q.; Liu, S. Electroanal. 2008,
20, 803–807.
(15) Liu, S.; Peng, L.; Yang, X.; Wu, Y.; He, L. Anal. Biochem. 2008,
375, 209–216.
(44) Evert, B.; Eichelbaum, M.; Haubruck, H.; Zanger, U. M.
Naunyn-Schmiedeberg’s Arch. Pharmacol. 1997, 355, 309–318.
(45) Ellis, S. W.; Rowland, K.; Ackland, M. J.; Rekka, E.; Simula,
A. P.; Lennard, M. S.; Wolf, C. R.; Tucker, G. T. Biochem. J. 1996,
316, 647–654.
(16) Yang, M. L.; Kabulski, J. L.; Wollenberg, L.; Chen, X. Q.;
Subramanian, M.; Tracy, T. S.; Lederman, D.; Gannett, P. M.; Wu, N. Q.
Drug Metab. Dispos. 2009, 37, 892–899.
(17) Mie, Y.; Suzuki, M.; Komatsu, Y. J. Am. Chem. Soc. 2009,
131, 6646–6647.
(18) Dodhia, V. R.; Sassone, C.; Fantuzzi, A.; Di Nardo, G.; Sadeghi,
S. J.; Gilardi, G. Electrochem. Commun. 2008, 10, 1744–1747.
(46) Pelkonen, O.; Turpeinen, M.; Hakkola, J.; Honkakoski, P.;
Hukkanen, J.; Raunio, H. Arch. Toxicol. 2008, 82, 667–715.
3838
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839

Analytical Chemistry
ARTICLE
(47) Granvil, C. P.; Krausz, K. W.; Gelboin, H. V.; Idle, J. R.;
Gonzalez, F. J. J. Pharmacol. Exp. Ther. 2002, 301, 1025–1032.
(48) Yue, Q. Y.; Sawe, J. Eur. J. Clin. Pharmacol. 1997, 52, 41–47.
(49) Tatsumi, A.; Ikegami, Y.; Morii, R.; Sugiyama, M.; Kadobayashi,
M.; Iwakawa, S. Biol. Pharm. Bull. 2009, 32, 517–519.
(50) Yamazaki, H.; Inoue, K.; Chiba, K.; Ozawa, N.; Kaway, T.;
Suzuki, Y.; Goldstein, J. A.; Guengerich, F. P.; Shimada, T. Biochem.
Pharmacol. 1998, 56, 243–251.
(51) Mancy, A.; Antignac, M.; Minoletti, C.; Dijols, S.; Mouries, V.;
Duong, N-T. H.; Battioni, P.; Dansette, P. M.; Mansuy, D. Biochemistry
1999, 38, 14264–14270.
(52) Tracy, T. S.; Hutzler, J. M.; Haining, R. L.; Rettie, A. E.;
Hummel, M. A.; Dickmann, L. J. Drug Metab. Dispos. 2002, 30, 385–390.
(53) Andersson, T. B.; Bredberg, E.; Ericsson, H.; Sjoberg, H. Drug
Metab. Dispos. 2004, 32, 715–721.
(54) Iwata, H.; Fujita, K.; Kushida, H.; Suzuki, A.; Konno, Y.;
Nakamura, K.; Fujino, A.; Kamataki, T. Biochem. Pharmacol. 1998,
55, 1315–1325.
(55) Yu, A.; Haining, R. L. Drug Metab. Dispos. 2001, 29, 1514–1520.
(56) McGinnity, D. F.; Griffin, S. J.; Moody, G. C.; Voice, M.;
Hanlon, S.; Friedberg, T.; Riley, R. J. Drug Metab. Dispos. 1999, 27, 1023.
(57) Facciola’, G.; Hidestrand, M.; von Bahr, C.; Tybring, G. Eur. J.
Clin. Pharmacol. 2001, 56, 881–888.
3839
dx.doi.org/10.1021/ac200309q |Anal. Chem. 2011, 83, 3831–3839