Covalent protein modification by reactive drug metabolites using online electrochemistry/liquid chromatography/mass spectrometry

Article abstract of DOI:10.1021/ac801699g

We present a rapid and convenient method to perform and evaluate the covalent protein binding of reactive phase I metabolites. The oxidative metabolism of the drugs paracetamol, amodiaquine, and clozapine is simulated in an electrochemical (EC) flow-through cell, which is coupled online to an LC/MS system. Adduct formation of the reactive metabolites with the proteins β-lactoglobulin A and human serum albumin proceeds in a reaction coil between EC cell and injection system of the HPLC system. The formed drug-protein adducts are characterized with online time-of-flight mass spectrometry, and the modification site is localized using FTICR-mass spectrometry. Due to its simple setup, easy handling, and short analysis times, the method provides an interesting tool for the rapid risk assessment of covalent protein binding as well as for the synthesis of covalent drug-protein adducts in high purity and high yield.

Full text of DOI:10.1021/ac801699g

Anal. Chem. 2008, 80, 9714–9719
Covalent Protein Modification by Reactive Drug
Metabolites Using Online Electrochemistry/Liquid
Chromatography/Mass Spectrometry
Wiebke Lohmann,† Heiko Hayen,‡ and Uwe Karst*,†
Institut fu¨r Anorganische und Analytische Chemie, Westfa¨lische WilhelmssUniversita¨t Mu¨nster, Corrensstrasse 30,
48149 Mu¨nster, Germany, and ISASsInstitute for Analytical Sciences, Bunsen-Kirchhoff-Strasse 11,
44139 Dortmund, Germany
We present a rapid and convenient method to perform
and evaluate the covalent protein binding of reactive phase
I metabolites. The oxidative metabolism of the drugs
paracetamol, amodiaquine, and clozapine is simulated in
an electrochemical (EC) flow-through cell, which is coupled
online to an LC/MS system. Adduct formation of the
reactive metabolites with the proteins ꢀ-lactoglobulin A
and human serum albumin proceeds in a reaction coil
between EC cell and injection system of the HPLC system.
The formed drug-protein adducts are characterized with
online time-of-flight mass spectrometry, and the modifica-
tion site is localized using FTICR-mass spectrometry. Due
to its simple setup, easy handling, and short analysis
times, the method provides an interesting tool for the
rapid risk assessment of covalent protein binding as well
as for the synthesis of covalent drug-protein adducts in
high purity and high yield.
the modification directly affects the metabolizing CYP enzyme,
irreversible inactivation of that enzyme can emerge.5,6
First reports about covalent drug-protein adduct formation
date back to the 1940s.7 Since then, the methodology for the
detection and identification of drug-protein adducts improved
significantly, particularly after introduction of electrospray mass
spectrometry.8 However, the detection and identification of
proteins that are covalently modified by a drug or a drug
metabolite is still a challenging task since the isolation of a single
protein species from a complex mixture such as a microsomal
incubation or even an in vivo assay is quite intricate and
time-consuming.9,10 Often, radiolabeled drugs are used for the
quantification of covalent protein binding,11 but thus far, only little
information on the nature of the modified proteins can be obtained.
Using two-dimensional liquid chromatography/mass spectrometry
(LC/MS), covalent adduct formation of metabolites of two model
compounds with a CYP enzyme was detected.12 For more detailed
information on the nature of the modified protein, however,
isolation and proteolytic digest in combination with mass spec-
trometric measurements cannot be avoided.2,9,13-15
The mechanistic details of drug toxicity are still poorly
understood. One hypothesis explaining toxic side effects of a drug
is the covalent binding of drug metabolites to proteins. Xenobiotics
such as drugs are often metabolized by cytochrome P450 enzymes
(CYP) to oxidative metabolites. While in most cases drug
metabolism is regarded as a detoxification process, sometimes
bioactivation to highly reactive species occurs.1 The majority of
reactive metabolites are electrophilic compounds seeking nucleo-
philic centers as reaction partners,2 which are omnipresent in
proteins in the cellular environment where metabolism proceeds.
The covalent modification of proteins can alter their functionality
or disrupt regulatory pathways, which can cause cell death,3 or
covalently modified proteins can initiate an immune response.4 If
Furthermore, the modification grade of a protein species is
generally low, which makes the detection of a modified protein
even more difficult. Therefore, the development of a simple
method is desirable that allows a rapid assessment of the potential
of a drug candidate to covalently modify proteins as well as the
characterization of the modified protein including the localization
(5) Kalgutkar, A. S.; Obach, R. S.; Maurer, T. S. Curr. Drug Metab. 2007, 8,
407–447
(6) Hollenberg, P. F.; Kent, U. M.; Bumpus, N. N. Chem. Res. Toxicol. 2008,
21, 189–205
(7) Miller, E. C.; Miller, J. A. Cancer 1947, 47, 468–480
(8) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science
1989, 246, 64–71
(9) To¨rnqvist, M.; Fred, C.; Haglund, J.; Helleberg, H.; Paulsson, B.; Rydberg,
P. J. Chromatogr., B 2002, 778, 279–308
(10) Zhou, S. J. Chromatogr., B 2003, 797, 63–90
(11) Evans, D. C.; Watt, A. P.; Nicoll-Griffith, D. A.; Baillie, T. A. Chem. Res.
Toxicol. 2004, 17, 3–16
.
.
.
.
.
.
* To whom correspondence should be addressed. E-mail: uk@
uni-muenster.de. Fax: +49-251-83 36013.
.
(12) Bateman, K. P.; Baker, J.; Wilke, M.; Lee, J.; LeRiche, T.; Seto, C.; Day, S.;
† Westfa¨lische WilhelmssUniversita¨t Mu¨nster.
Chauret, N.; Ouellet, M.; Nicoll-Griffith, D. A. Chem. Res. Toxicol. 2004,
‡ ISASsInstitute for Analytical Sciences.
17, 1356–1361
(13) Yukinaga, H.; Takami, T.; Shioyama, S.; Tozuka, Z.; Masumoto, H.; Okazaki,
O.; Sudo, K. Chem. Res. Toxicol. 2007, 20, 1373–1378
(14) Sleno, L.; Varesio, E.; Hopfgartner, G. Rapid Commun. Mass Spectrom.
2007, 21, 4149–4157
.
(1) Kalgutkar, A. S.; Gardner, I.; Obach, R. S.; Shaffer, C. L.; Calligari, E.; Henne,
K. R.; Mutlib, A. E.; Dalvie, D. K.; Lee, J. S.; Nakai, Y.; O’Donnell, J. P.;
.
Harriman, S. P. Curr. Drug Metabol. 2005, 6, 161–225
(2) Ma, S.; Subramanian, R. J. Mass Spectrom. 2006, 41, 1121–1139
(3) Boelsterli, U. A. Drug Metab. Rev. 1993, 25, 395–451
(4) Naisbitt, D. J.; Gordon, S. F.; Pirmohamed, M.; Park, B. K. Drug Saf. 2000,
23, 483–507
.
.
.
.
(15) Damsten, M. C.; Commandeur, J. N. M.; Fidder, A.; Hulst, A. G.; Touw,
D.; Noort, D.; Vermeulen, N. P. E. Drug Metab. Dispos. 2007, 35, 1408–
1417.
.
9714 Analytical Chemistry, Vol. 80, No. 24, December 15, 2008
10.1021/ac801699g CCC: $40.75 2008 American Chemical Society
Published on Web 11/12/2008

of the modification site. As was shown in recent publications, the
electrochemical oxidation of drug compoundsssimulating the
CYP catalyzed oxidative metabolismscan provide valuable infor-
mation about metabolically labile sites in a molecule, particularly
in an online coupling with electrochemistry/liquid chromatogra-
phy/mass spectrometry (EC/LC/MS).16-21 The use of a mi-
crospray device connected to a mass spectrometer for tagging
cysteine residues in synthetic or tryptic peptides22,23 or in proteins
in a nanospray24 or a microspray electrode25 with electrogenerated
benzoquinone is published as well. However, the reaction products
generated in the electrospray interface are obtained only in low
yield and cannot be synthesized on a preparative scale for further
characterization. Until now, nothing was reported about the use
of an EC/(LC)/MS system for the investigation of a drug
compound in terms of its tendency to covalently modify proteins.
Therefore, we decided to set up an online EC/LC/MS system
for assessing the risk of covalent protein binding of drug
metabolites. Three model compounds that are known for bioac-
tivation to reactive metabolites were selected for validation of the
EC/LC/MS setup. Paracetamol is an analgesic and antipyretic
drug that is oxidized by CYP to the reactive N-acetyl-p-benzo-
quinoneimine (NAPQI), which is known for covalent protein
binding thus causing hepatotoxicity.26,27 For this reason, it was
recently decided by the German Federal Institute for Pharma-
ceuticals and Medical Products to classify paracetamol, which was
so far a classical over-the-counter drug, in package sizes of more
than 10 g as a prescription drug.28 The antimalarial agent
amodiaquine undergoes bioactivation to a respective quinoneimine
(AQQI) and has been withdrawn from malaria prophylaxis due
to the occurrence of agranulocytosis and hepatotoxicity,29 and
clozapine which is an antipsychotic agent that is used for the
treatment of schizophrenia is metabolized by CYP to a reactive
electrophilic nitrenium ion30 (Figure 1a). The oxidative metabo-
lism of all model compounds as well as the detoxification of the
reactive species by addition of glutathione has been successfully
Figure 1. (a) Structures of the reactive metabolites of paracetamol
(N-acetyl-p-quinoneimine, NAPQI), amodiaquine (amodiaquine quino-
neimine, AQQI), and clozapine (clozapine nitrenium ion, CLZox). (b)
Principle of covalent protein binding of drugs after activation by
cytochrome P450 (CYP) enzymes to reactive metabolites.
mimicked in EC/(LC)/MS systems.31-34 In the present work, a
setup for the online synthesis of covalent drug-protein adducts
that are subsequently detected and characterized by mass
spectrometric techniques is developed.
EXPERIMENTAL SECTION
Reagents and Materials. ꢀ-Lactoglobulin A from bovine milk,
human serum albumin, and trypsin from bovine pancreas
(g10000 BAEE units/mg of protein) were purchased from Sigma-
Aldrich Chemie GmbH (Steinheim, Germany) and were used
without further purification. All chemicals and solvents were
ordered from Sigma-Aldrich Chemie GmbH, Fluka Chemie GmbH
(Buchs, Switzerland), ABCR (Karlsruhe, Germany), or Molekula
(Shaftesbury, U.K.) in the highest quality available. Disposable
PD-10 Sephadex G-25 M desalting columns were obtained from
GE Healthcare (Buckinghamshire, U.K.). Water used for HPLC
was purified using a Milli-Q Gradient A 10 system and filtered
through a 0.22-µm Millipak 40 (Millipore, Billerica, MA).
Online Electrochemical Generation of Drug-Protein Ad-
ducts. A solution containing a 1 × 10-4 M concentration of the
desired drug compound in 20 mM aqueous ammonium formate
(adjusted to pH 7.4 with ammonia) and acetonitrile (50/50, v/v)
was pumped with a flow rate of 10 µL min-1 through a com-
mercially available electrochemical flow-through cell (model 5021,
conditioning cell, ESA Biosciences Inc., Chelmsford, MA). The
potential at the working electrode was controlled by a Coulochem
II potentiostat (ESA Biosciences). Paracetamol was oxidized at
600 mV, amodiaquine at 300 mV, and clozapine at 400 mV versus
Pd/H2 in the EC cell. Via a T-piece, a solution containing the
protein of interest (2 × 10-5 M) in 8 M urea was added after the
EC cell at the same flow rate, and the combined solutions were
allowed to react in a reaction coil (Teflon tubing, 0.762-mm inner
diameter, 438-mm length) for 10 min. The effluent from the
(16) Jurva, U.; Wikstro¨m, H. V.; Bruins, A. P. Rapid Commun. Mass Spectrom.
2000, 14, 529–533
(17) Jurva, U.; Wikstro¨m, H. V.; Weidolf, L.; Bruins, A. P. Rapid Commun. Mass
Spectrom. 2003, 17, 800–810
(18) Gamache, P.; Smith, R.; McCarthy, R.; Waraska, J.; Acworth, I. Spectroscopy
2003, 18, 14–21
(19) Johansson, T.; Weidolf, L.; Jurva, U. Rapid Commun. Mass Spectrom. 2007,
21, 2323–2331
(20) Permentier, H. P.; Bruins, A. P.; Bischoff, R. Mini Rev. Med. Chem. 2008,
8, 46–56
(21) Lohmann, W.; Karst, U. Anal. Bioanal. Chem. 2008, 391, 79–96
(22) Dayon, L.; Roussel, C.; Prudent, M.; Lion, N.; Girault, H. H. Electrophoresis
2005, 26, 238–247
(23) Abonnenc, M.; Dayon, L.; Perruche, B.; Lion, N.; Girault, H. H. Anal. Chem.
2008, 80, 3372–3378
(24) Rohner, T.; Rossier, J. S.; Girault, H. H. Electrochem. Commun. 2002, 4,
695–700
(25) Dayon, L.; Roussel, C.; Girault, H. H. J. Proteome Res. 2006, 5, 793–800
(26) Jollow, D. J.; Mitchell, J. R.; Potter, W. Z.; Davis, D. C.; Gillette, J. R.; Brodie,
B. B. J. Pharmacol. Exp. Ther. 1973, 187, 195–202
(27) Potter, W. Z.; Davis, D. C.; Mitchell, J. R.; Jollow, D. J.; Gillette, J. R.; Brodie,
B. B. J. Pharmacol. Exp. Ther. 1973, 187, 203–210
.
.
.
.
.
.
.
.
.
.
.
.
(28) Bundesinstitut fu¨r Arzneimittel und Medizinische Produkte (German
Federal Institute for Pharmaceuticals and Medical Products), Ergebnispro-
tokoll der 60. Sitzung des Sachversta¨ndigen-Ausschusses fu¨r Verschrei-
bungspflicht vom 15. Januar 2008.
(31) Getek, T. A.; Korfmacher, W. A.; McRae, T. A.; Hinson, J. A. J. Chromatogr.,
A 1989, 474, 245–256
(32) Lohmann, W.; Karst, U. Anal. Bioanal. Chem. 2006, 386, 1701–1708
(33) Lohmann, W.; Karst, U. Anal. Chem. 2007, 79, 6831–6839
(34) Van Leeuwen, S. M.; Blankert, B.; Kauffmann, J.-M.; Karst, U. Anal. Bioanal.
Chem. 2005, 382, 742–750
.
(29) Neftel, K. A.; Woodtly, W.; Schmid, M.; Frick, P. G.; Feh, J. Br. Med. J.
.
1986, 292, 721–723
(30) Maggs, J. L.; Williams, D.; Pirmohamed, M.; Park, B. K. J. Pharmacol. Exp.
Ther. 1995, 275, 1463–1475
.
.
.
.
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008 9715

reaction coil was collected in a 10-µL injection loop, which was
mounted on a 10-port switching valve. By switching this valve,
the content of the injection loop was flushed by the HPLC pumps
into the LC/MS system.
FTICR-MS Conditions. Nano-ESI-FTICRMS experiments
were carried out using the hybrid LTQ FT Fourier transform ion
cyclotron resonance hybrid mass spectrometer (Thermo Fisher
Scientific, Bremen, Germany), equipped with a 7.0-T actively
shielded superconducting magnet and nano-ESI source. Gold-
plated nano-ESI pipets (tip i.d. ∼5 µm) were from MasCom
(Bremen, Germany). The instrument was operated in the positive
ionization mode and the mass resolution was set to 50 000 fwhm
at m/z 400. Ion transmission into the linear trap and signal
intensity were automatically optimized for maximum ion signal
of the unmodified peptide T13. The parameters were as follows:
source voltage 1.0-1.2 kV, capillary voltage 25 V, capillary
temperature 200 °C, and tube-lens voltage 105 V. Accurate mass
measurements of peptides were carried out by a zoom scan in a
narrow mass window (±10 Da) and displayed mass spectra were
averaged from five single spectra. Fragmentation experiments
were performed by collisionally induced dissociation (normalized
collision energy of 35% with activation q ) 0.25 and activation time
of 30 ms) in the linear ion trap part of the hybrid instrument.
Mass analysis of the resulting fragments was done in the FTICR
cell, and displayed spectra were averaged from 100 single spectra.
Off-Line Electrochemical Synthesis of Drug-Protein Ad-
ducts and Tryptic Digest. The offline electrochemical oxidation
was performed as described for the online setup except with a
higher flow rate of 50 µL min-1 and in a 5-fold higher concentra-
tion of the drug compound (5 × 10-4 M). The effluent from the
EC cell was collected in a vial containing 1 mL of a 1 × 10-4
M
solution of the protein of interest in 8 M urea. After 20 min of
collection, corresponding to a total volume of 1 mL of reactive
metabolite solution (5-fold molar excess based on the amount of
protein), the sample was incubated at 37 °C for 30 min. After the
incubation time, the excess reagent was removed on a PD-10
desalting column with pure water as mobile phase according to
the supplier’s protocol. Reduction of the disulfide bridges was
performed at 37 °C for 30 min using a 100 mM stock solution of
tris(2-carboxylethyl)phosphine hydrochloride (TCEP) in 0.3 M
aqueous ammonia. A 960-mg sample of urea was added to 2 mL
of the purified drug-protein adduct and a 9-fold molar excess of
TCEP per oxidized S-atom in the protein was added from the stock
solution. The reduction was terminated by passing the solution
again through a PD-10 desalting column. Digestion with trypsin
was performed by adding 5 µL of a stock solution of 2.3 mg of
trypsin in 1 mL of water to 1 mL of the reduced drug-protein
adduct solution containing ∼1.6 × 10-5 M modified protein and
50 mM ammonium bicarbonate. After 90 min at 37 °C, the digest
was terminated by adding 200 µL of ice-cold acetonitrile containing
1% formic acid.
LC/MS Conditions. An HPLC separation was performed in
the online EC/LC/MS setup to remove excess reactive metabolite
and urea before MS analysis. The LC/MS setup comprised a
Shimadzu (Duisburg, Germany) LC system and a micrOTOF mass
spectrometer (Bruker Daltonics, Bremen, Germany), equipped
with an electrospray ionization (ESI) source. For controlling the
micrOTOF and data handling, micrOTOFControl 1.1 and DataAnal-
ysis 3.3 (Bruker Daltonics) software were used. The LC system
consisted of two LC-10ADVP pumps, a DGC-14A degasser, a SIL-
RESULTS AND DISCUSSION
Generating Drug-Protein Adducts in the Online EC/LC/
MS System. First experiments were performed using ꢀ-lactoglo-
bulin A (ꢀ-LGA) as model target protein. Despite its occurrence
as the main whey protein in milk, this protein was selected due
to its structural homogeneity and relatively small molecular mass
(162 amino acids, 18.4 kDa), which allows an easier elucidation
of its fate after reaction with reactive metabolites. Paracetamol
(APAP), amodiaquine (AQ), and clozapine (CLZ) were selected
as model drug compounds that undergo bioactivation by dehy-
drogenation to reactive electrophilic species (Figure 1a). This
bioactivation process is simulated in an EC flow-through cell
(Figure 2). The cell is commercially available and comprises a
three-electrode arrangement with a porous glassy carbon working
electrode, a Pd counter electrode, and a Pd/H2 reference elec-
trode. The porosity of the working electrode results in a large
surface area, which allows for an almost quantitative turnover of
the analytes and low maintenance intervals of the EC cell. A
solution containing the drug of interest at pH 7.4 was pumped
through the EC cell at a constant flow rate of 10 µL/min. The EC
potential at the working electrode was selected for each model
compound on the basis of previously reported EC oxidation results
so that the yield of the respective reactive metabolite was optimal.
APAP was thus oxidized to NAPQI at 600 mV versus Pd/H2, AQ
was converted to AQQI at 300 mV versus Pd/H2, and CLZ was
dehydrogenated to CLZox at 400 mV versus Pd/H2. The outlet of
the EC cell was connected via a T-piece with the outlet of a second
syringe pump, which delivered the denaturized model protein
ꢀ-LGA in 8 M urea in a 5-fold lower concentration as the reactive
metabolite. The combined flows containing the reactive metabolite
and the protein were directed into a reaction coil. For ꢀ-LGA, a
reaction time of 10 min has proven to be sufficient for a good
yield of modified protein. The effluent from the reaction coil was
then collected in an injection loop, which was mounted on a 10-
port switching valve for injection into the LC/MS system.33 From
the time-of-flight mass spectra of unmodified ꢀ-LGA (a) and ꢀ-LGA
after 10 min of reaction with NAPQI (b), AQQI (c), and CLZox
HTA autosampler, a CTO-10AVP column oven, and a SPD-10AVVP
UV detector. The separation was performed on a Discovery BIO
Wide Pore C5 column (2.1 × 150 mm, 5 µm) from Supelco at
40 °C. The injection volume was 10 µL. A binary gradient with
aqueous formic acid (0.1%) and acetonitrile containing 0.1% formic
acid at a total flow rate of 300 µL min-1 was used. After 2 min of
an isocratic step with 5% organic mobile phase, the acetonitrile
content was increased to 80% in 10 min and held at 80% for 2 min,
and finally, the system was equilibrated for 6 min. On the
micrOTOF instrument, full scan spectra (m/z 50-4000) were
recorded after HPLC separation using ESI(+)-MS under the
following conditions: end plate offset, -500 V; capillary, 4000 V;
nebulizer gas (N2), 0.8 bar; drying gas (N2), 8.0 L min-1; drying
temperature, 200 °C; capillary exit, 150.0 V; skimmer 1, 50.0 V;
skimmer 2, 23.0 V; hexapole 1, 23.0 V; hexapole 2, 17.0 V; hexapole
rf, 500 V; transfer time, 70.0 µs; prepulse storage, 30.0 µs; detector,
-1000 V. Internal calibration was performed using sodium formate
clusters at the beginning of each HPLC run.
9716 Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

Figure 2. EC/LC/MS setup for the generation and detection of covalent drug-protein adducts. The drug of interest is oxidized in the EC
flow-through cell to its oxidative metabolites, which subsequently react online with a protein. HPLC separation is carried out to remove excess
reactive metabolites, and the drug-protein adducts are detected with mass spectrometry.
Figure 4. FTICR-MS spectra of the tryptic peptide T13 containing
the amino acid residues 102-124 of ꢀ-LGA with (a) the [M + 2H]2+
of the unmodified protein, (b) the [M + 3H]3+ of ꢀ-LGA modified with
NAPQI, (c) the [M + 3H]3+ of ꢀ-LGA modified with AQQI, and (d) the
[M + 3H]3+ of ꢀ-LGA modified with CLZox.
Figure 3. Time-of-flight mass spectra and deconvolution results of
(a) unmodified ꢀ-LGA, (b) ꢀ-LGA after reaction with NAPQI, (c) ꢀ-LGA
after reaction with AQQI, and (d) ꢀ-LGA after reaction with CLZox.
The mass spectra of the modified protein were obtained after online
reaction of the unmodified protein with electrochemically generated
reactive metabolites.
mass of the protein to 18 714.0 Da, which is in good accordance
to the attachment of one molecule of AQQI with a monoisotopic
mass of 353.1 Da. As could already be seen from the nondecon-
voluted mass spectrum, the addition of CLZox with a monoisotopic
mass of 325.1 Da even yielded multiple adducts with more than
one CLZ molecule, with the monoadduct with a molecular mass
of 18 685.5 Da being the most abundant one, next to a bis-adduct
with 19009.7 Da and a tris-adduct with 19333.9 Da. It is remarkable
that, especially in case of NAPQI and AQQI, and to a certain extent
also for CLZox, only a monodrug-protein adduct is formed despite
a relatively large excess of reactive metabolite. It thus seems likely
that this is due to the protein sequence, probably providing only
a single readily available modification site. The electrophilic
addition of reactive intermediates often occurs at nucleophilic
centers in the protein, being for example the free thiol group of
a cysteine. Since ꢀ-LGA contains exactly one free cysteine residue
that is not bound in disulfide bridges, it is very likely that the
modification occurs at this free cysteine residue. In case of CLZox,
(d) (Figure 3), it can clearly be seen that the MS signals are
shifted toward higher m/z values after reaction with one of the
reactive metabolites, thus proving that a covalent drug-protein
adduct formation has occurred. While for the ꢀ-LGA-NAPQI
adduct, a certain amount of unmodified ꢀ-LGA could be observed,
the modification grade of ꢀ-LGA with AQQI is 100% (no unmodi-
fied ꢀ-LGA was detectable) and CLZox even seems to generate
multiple ꢀ-LGA-CLZox adducts. Deconvoluted mass spectra of
the modified ꢀ-LGA are shown next to the respective raw spectra.
Deconvolution of unmodified ꢀ-LGA resulted in a molecular mass
of the protein of 18 362.5 Da, which corresponds well with the
theoretical value of 18 363.1 Da calculated from the amino acid
sequence. After reaction with NAPQI, the molecular mass of
ꢀ-LGA increased to 18 513.6 Da, corresponding to adduct forma-
tion with one unit of NAPQI with a monoisotopic mass of 149.1
Da. Modification of ꢀ-LGA with AQQI increased the molecular
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008 9717

Figure 5. (a) FTICR-MS/MS spectrum of [T13 + AQQI + 3H]3+ resulting in loss of the N-diethyl group of the AQQI modification. (b) Further
fragmentation (MS3) of the fragment ion with m/z 985.7593 and assignment of b- and y-ion series to the observed signals.
the electrophilic nitrenium ion seems to be more reactive than
NAPQI or AQQI, thus attacking also other nucleophilic centers.
Off-Line Analysis of the Drug-Protein Adducts. To obtain
more information about the modification site in the protein, a
tryptic digest of the modified proteins after reduction of disulfide
bonds was performed. Therefore, the modified proteins were
synthesized in an off-line approach according to the online assay,
except that the effluent from the EC cell was collected in a vial
containing a ꢀ-LGA solution, which was incubated at 37 °C after
termination of the collection. Urea and excess reactive metabolite
were removed on a PD-10 desalting column, and disulfide bridges
were reduced using TCEP before tryptic digest. Trypsin is a
proteolytic enzyme that cleaves proteins specifically at the C-
terminal site of lysine and arginine into smaller peptides. In case
of a covalently modified protein, one or more of these peptides
carry the modification and these peptides have to be analyzed in
more detail. Particular attention was paid to the peptide bearing
the only free cysteine thiol group in the protein, which is the
peptide containing the amino acid residues 102-124 (YLLFC-
MENSAEPEQSLVCQCLVR, peptide T13). Cysteine 121 is the
amino acid residue of interest. The nonmodified peptide T13 has
a monoisotopic neutral mass of 2674.2263 Da and was detected
using positive nanoelectrospray Fourier transform ion cyclotron
resonance-mass spectrometry (FTICR-MS) as [M + 2H]2+ in the
tryptic digest of unmodified ꢀ-LGA (Figure 4a). If this peptide is
modified by one unit of either NAPQI, AQQI, or CLZox, the
monoisotopic neutral mass increases by the molecular weight of
the modification. All these modified peptides were detected as
[M + 3H]3+ in the respective tryptic digests of the modified
ꢀ-LGA, and the measured mass spectra are in good agreement
with the calculated spectra (Figure 4b-d).
The site of the modification could thus be restricted to one of
the amino acids in peptide T13. To further narrow the location,
the respective peptides were fragmented by collisionally induced
dissociation in the linear quadrupole ion trap of the hybrid MS.
The resulting fragment ions were then analyzed in the FTICR cell
for accurate mass information at high resolution. The fragmenta-
tion pathway of peptide T13, which is modified with AQQI (T13
+ AQQI) should be discussed representatively for all modified
peptides. In the MS/MS spectrum of this peptide (Figure 5a),
one main fragmentation product of T13 + AQQI was observed at
m/z 985.7593, which corresponds to a loss of C4H11N. Obviously,
9718 Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

residue (nonmercaptalbumin, HNA) were detected (Figure 6b).
The reactive CLZox thus seems to exhibit a reactivity toward other
nucleophilic functional groups in the amino acid side chains, as
for example the additional amino group in lysine or the hydroxyl
group of serine, as was already observed for ꢀ-LGA, where adducts
of more than one activated drug molecule were detected. Even
though the adduct yield of HSA with CLZox was not as high as
for ꢀ-LGA, it is evident that the online EC/LC/MS method also
has high potential in the risk assessment of covalent protein
binding, for example, to the biologically relevant plasma protein
HSA.
CONCLUSIONS
We present a simple and rapid online method for the genera-
tion and identification of drug-protein adducts. In an EC flow-
through cell, the oxidative phase I metabolism of drugs is
mimicked by an electrochemical oxidation. The LC/MS equipment
used is currently standard instrumentation in drug metabolism
laboratories. Upgrading an LC/MS system by an EC cell is thus
easily accomplished, which makes the method interesting for all
researchers being active in the field of drug metabolism.
The experiments demonstrate that our method offers interest-
ing features in the risk assessment of covalent protein binding of
drug metabolites. Rapid information on the possibility of covalent
protein binding may be obtained, and in combination with the
off-line approach for the characterization, the modification site can
be localized. If reactive metabolites mainly react with free thiol
groups, endogenous detoxification pathways as, for example, the
detoxification by addition of glutathione may lower the risk of
undesired side effects; but if protein modification also occurs at
other functional groups, the risk of covalent binding to other
macromolecules such as DNA, which has more serious conse-
quences as covalent protein binding regarding the toxicity, should
also be considered. For such considerations, our method com-
poses an excellent basis since it provides rapid and reliable
information with minimum technical effort and analysis time.
In terms of complexity, our method is superior to convention-
ally used techniques. All standard procedures, such as microsomal
incubations or in vivo experiments, yield only a small percentage
of modified protein, which has to be isolated from a complex
biological matrix for identification and characterization, while our
method yields very clean modified proteins with a high degree
of modification. This saves time, labor, and money, and it facilitates
the subsequent determination of the position of the modification
site.
Figure 6. (a) Section of a time-of-flight mass spectrum of com-
mercially available human serum albumin. Mercaptalbumin (HMA, with
a free thiol group), and nonmercaptalbumin (HNA, with the thiol group
in a mixed disulfide bond with cysteine) detected with a deconvoluted
molecular mass of 66 437.6 (calculated 66 437.5 Da) and 66 557.8
Da (calculated 66 556.6 Da), respectively. (b) Section of a time-of-
flight mass spectrum of commercially available human serum albumin
after reaction with CLZox. Both HMA and HNA were modified by
CLZox to HMA* with a deconvoluted molecular mass of 66 768.0 Da
(calculated 66 762.3 Da) and HNA* with a deconvoluted molecular
mass of 66 879.4 Da (calculated 66 881.4 Da), respectively.
instead of fragmentation of the amino acid sequence, the AQQI
modification is the main site where fragmentation occurs, and the
AQQI residue is cleaved under loss of the N-diethyl functionality
in the MS/MS experiment, which provides further evidence for
the presence of AQQI in this peptide. This resulting ion with m/z
985.7593 was fragmented further in an MS3 experiment, and this
time, fragmentation of the amino acid chain was observed (Figure
5b). The b-ion and particularly the y-ion series are covered well,
and since both y3 without AQQI modification and y5 with AQQI
modification could be detected, the modification site could be
assigned to Cys121, which is the free cysteine in the native protein.
Applicability of the Online EC/LC/MS Method to Serum
Albumins. The newly developed method should be applied not
only to the whey protein ꢀ-LGA but also to human plasma proteins
of relevance in drug metabolism studies such as human serum
albumin (HSA). HSA is the most abundant plasma protein and is
synthesized in the liver, where extensive drug metabolism occurs.
Therefore, it is a potential target protein for reactive drug
metabolites. Adduct formation of CLZox with HSA was examined
with the EC/LC/MS setup according to the experiments with
ꢀ-LGA. Figure 6 shows a section of the mass spectrum of HSA
before and after reaction with CLZox. The unmodified protein is
microheterogeneous (Figure 6a), which is mostly due to partial
blockage of the only free cysteine thiol functionality via a mixed
disulfide bond with small bioavailable thiols such as cysteine or
glutathione.35 After reaction with CLZox, adducts of both HSA
with the free thiol group (mercaptalbumin, HMA) and the HSA
variant in which the free cysteine is blocked with another cysteine
To sum up, our simple and elegant method offers two new
possibilities: the risk assessment of covalent protein binding of a
new chemical entity and the rapid and clean synthesis of covalently
modified proteins.
ACKNOWLEDGMENT
Financial support by the Fonds der Chemischen Industrie
(Frankfurt/Main, Germany) and the Deutsche Forschungsge-
meinschaft (Bonn, Germany) is gratefully acknowledged.
Received for review August 13, 2008. Accepted October
10, 2008.
(35) King, T. P. J. Biol. Chem. 1961, 236, PC5
.
AC801699G
Analytical Chemistry, Vol. 80, No. 24, December 15, 2008 9719