Immunochemical detection of hepatic cocaine-protein adducts in mice

Article abstract of DOI:10.1021/tx970147c

Cocaine is capable of producing hepatic necrosis in laboratory animals and humans. Studies in mice indicate that N-oxidative metabolism of cocaine is required for hepatotoxicity and have suggested that toxicity may result from the adduction of proteins by cocaine-reactive metabolites. To aid in identifying protein targets for cocaine-reactive metabolites, an antibody was raised in rabbits immunized with cocaine linked via the tropane nitrogen to a carrier protein (bovine serum albumin). Hepatic proteins from cocaine- treated mice (ICR males, 50 mg of cocaine/kg of body weight, ip) and saline- treated controls were prepared from whole liver homogenate or following subcellular fractionation, and Western blot analyses of hepatic Proteins using this antibody were conducted following one- and two-dimensional SDS- PAGE. Analysis of liver homogenate from cocaine-treated mice revealed major protein targets with approximate molecular masses of 20 kDa (pI = 6.0), 44 kDa (two proteins with pI's of 5.0 and 7.0), 52-54 kDa (pI = 4.5), and 64 kDa (pI = 5.5). These specific protein targets were shown to be localized in the mitochondria and microsomes. Several minor bands of immunoreactivity were also seen in mice treated with cocaine, but not in saline-treated controls. Pretreatment of mice with the P450 inhibitor SKF 525A diminished or eliminated the formation of these cocaine-protein adducts. Liver sections from cocaine-treated mice immunostained using the antibody indicated the presence of cocaine-adducted proteins in the centrilobular and midzonal regions of the lobule, corresponding to areas of hepatocyte swelling and necrosis. This study indicates that reactive metabolites from cocaine bind to discrete proteins in specific regions of the liver, consistent with a role for protein adduction in cocaine hepatotoxicity.

Full text of DOI:10.1021/tx970147c

Chem. Res. Toxicol. 1998, 11, 185-192
185
Im m u n och em ica l Detection of Hep a tic Coca in e-P r otein
Ad d u cts in Mice
Florence M. Ndikum-Moffor, J ohn W. Munson, Nagraj K. Bokinkere,†
J ennifer L. Brown,† Nigel Richards,† and Stephen M. Roberts*
Center for Environmental & Human Toxicology and Department of Chemistry, University of
Florida, Gainesville, Florida
Received August 18, 1997
Cocaine is capable of producing hepatic necrosis in laboratory animals and humans. Studies
in mice indicate that N-oxidative metabolism of cocaine is required for hepatotoxicity and have
suggested that toxicity may result from the adduction of proteins by cocaine-reactive
metabolites. To aid in identifying protein targets for cocaine-reactive metabolites, an antibody
was raised in rabbits immunized with cocaine linked via the tropane nitrogen to a carrier
protein (bovine serum albumin). Hepatic proteins from cocaine-treated mice (ICR males, 50
mg of cocaine/kg of body weight, ip) and saline-treated controls were prepared from whole
liver homogenate or following subcellular fractionation, and Western blot analyses of hepatic
proteins using this antibody were conducted following one- and two-dimensional SDS-PAGE.
Analysis of liver homogenate from cocaine-treated mice revealed major protein targets with
approximate molecular masses of 20 kDa (pI ) 6.0), 44 kDa (two proteins with pI’s of 5.0 and
7.0), 52-54 kDa (pI ) 4.5), and 64 kDa (pI ) 5.5). These specific protein targets were shown
to be localized in the mitochondria and microsomes. Several minor bands of immunoreactivity
were also seen in mice treated with cocaine, but not in saline-treated controls. Pretreatment
of mice with the P450 inhibitor SKF 525A diminished or eliminated the formation of these
cocaine-protein adducts. Liver sections from cocaine-treated mice immunostained using the
antibody indicated the presence of cocaine-adducted proteins in the centrilobular and midzonal
regions of the lobule, corresponding to areas of hepatocyte swelling and necrosis. This study
indicates that reactive metabolites from cocaine bind to discrete proteins in specific regions of
the liver, consistent with a role for protein adduction in cocaine hepatotoxicity.
binding) diminish or prevent toxicity (8, 10). The cor-
relation between covalent binding and hepatic necrosis
in these experiments has led to speculation that protein
adduction may be responsible for the hepatotoxicity of
cocaine (8). Definitive testing of this hypothesis, how-
ever, requires the identification of the protein targets.
Efforts to identify proteins adducted by cocaine-reac-
tive metabolites have been limited, due in large part to
the absence of effective tools for detecting cocaine-
protein adducts. Radiolabeled cocaine (3H and 14C) has
been available for several years, and has been used to
measure the overall extent of irreversible protein binding
that accompanies a cocaine dose. These isotopes are of
relatively low energy, however, making radiochemical
detection of the adduction of individual proteins difficult.
Numerous anti-cocaine antibody preparations have been
developed, principally for use in the immunoassay of
cocaine and metabolites for drug screening purposes. In
nearly all instances, these antibody preparations are
created using an antigen in which the cocaine hapten is
linked to the carrier protein through one of the cocaine
ester groups. In contrast, adduction of hepatic proteins
by cocaine in situ is thought to occur via the nitrogen,
resulting in a very different presentation of the cocaine
molecule for antibody recognition. This may explain why
several attempts by this laboratory to detect cocaine-
protein adducts by Western blots using commercially
available anti-cocaine antibodies were unsuccessful.1
In tr od u ction
Clinical reports have shown that cocaine is capable of
producing severe hepatic necrosis, sometimes resulting
in death (1-5). Studies of cocaine hepatotoxicity in mice
indicate that the necrogenic effects are a consequence of
cocaine bioactivation. While the predominant pathways
for cocaine biotransformation involve hydrolysis of its
phenyl and methyl esters, cocaine also undergoes se-
quential N-oxidation of the tropane nitrogen resulting in
the formation of norcocaine, N-hydroxynorcocaine, and
norcocaine nitroxide (in order of increasing oxidation) (6,
7). This N-oxidative pathway leads additionally to the
formation of a reactive metabolite that binds covalently
to proteins and is postulated to be a nitrosonium ion
created from a one-electron oxidation of norcocaine
nitroxide (8, 9). Experimental manipulations that result
in increased cocaine N-oxidative metabolism and covalent
binding (e.g., pretreatment with certain cytochrome
P450-inducing agents or inhibitors of competing esteratic
metabolism) increase cocaine hepatotoxicity, while in-
hibitors of oxidative metabolism (and thus covalent
* Address all correspondence to: Dr. Stephen M. Roberts, Center
for Environmental
& Human Toxicology, University of Florida,
Box 110885, Bldg. 471, Mowry Road, Gainesville, FL 32611. Phone:
(352) 392-4700. Fax: (352) 392-4707. E-mail: sroberts.vetmed1@
mail.health.ufl.edu.
† Department of Chemistry.
S0893-228x(97)00147-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 02/12/1998

186 Chem. Res. Toxicol., Vol. 11, No. 3, 1998
Ndikum-Moffor et al.
1. Cou p lin g of CBZ-â-a la n in e to Nor coca in e th r ou gh
Ca r bod iim id e-Med ia ted Rea ction : A reaction mixture
Sch em e 1. Syn th esis of NOR-a cid fr om
Nor coca in ea
a
containing 10 mL of dichloromethane, 0.865 mmol of norcocaine,
1.17 mmol of CCD, and 1.12 mmol of CBZ-â-alanine was stirred
for 30 min at 4 °C and then for 24 h at room temperature. The
mixture was washed twice with water and twice with 1% (w/v)
sodium carbonate, dried over anhydrous magnesium sulfate, and
evaporated to about 1 mL. The concentrated material was
streaked on three preparative TLC plates (1000 µm, 20 × 20
cm) and developed with ethyl acetate. The major band was
collected, dissolved in ethyl acetate, filtered, separated by
analytical TLC, and developed with ethyl acetate. A single spot
(Rf ) 0.48) was obtained, dissolved in ethyl acetate, and
evaporated to a clear oil. The reaction product (NOR CBZ-â-
alanine) was confirmed by NMR, and the yield was 75%.
2. F or m a tion of Nor coca in e Am in e (NOR-a m in e) fr om
NOR CBZ-â-a la n in e: The carbobenzoxycarbonyl protection
group was removed by catalytic dehydrogenation to yield a free
amine. NOR CBZ-â-alanine (0.404 mmol, 200 mg) was dissolved
in 10 mL of freshly distilled tetrahydrofuran containing 20 equiv
of cyclohexene and 60 mg (30%, w/w) of palladium on carbon.
The mixture was refluxed for 3 h, filtered through Celite, and
evaporated to dryness as a clear oil. The product was confirmed
as NOR-amine by NMR, and the yield was 86%.
a
Final product was confirmed by NMR and HPLC analyses.
The objective of this research was to develop an
antibody to cocaine capable of detecting cocaine-protein
adducts, both in Western blots and in liver sections from
cocaine-treated mice. This was accomplished by raising
polyclonal antibodies to cocaine linked to a carrier protein
(bovine serum albumin, BSA) through the tropane ni-
trogen. We report here the method of synthesis of the
antigen, the results of studies characterizing the affinity
and specificity of the antibody preparation, and the use
of the antibody preparation to detect by Western blot the
presence of several hepatic proteins adducted by cocaine
following its administration in vivo. The use of the
antibody preparation for immunohistochemical localiza-
tion of cocaine-protein adducts in liver sections is also
evaluated.
3. F or m a tion of NOR-a cid fr om NOR-a m in e: NOR-acid
was prepared by reacting the free amine with succinic anhy-
dride. NOR-amine (0.332 mmol, 120 mg) was added to 10 mL
of dichloromethane containing succinic anhydride (0.332 mmol),
and the reaction was carried out for 24 h at room temperature.
Similarly, radiolabeled NOR-acid was prepared by reacting
NOR-amine (0.0138 mmol, 5 mg) with succinic anhydride-1,4-
14C (0.0138 mmol, 2 µCi). NMR and HPLC analyses using
radiochemical and ultraviolet detection revealed that conversion
of NOR-amine to NOR-acid was complete. The NOR-acid
(hapten) was coupled to BSA to produce the immunogen (NOR-
BSA conjugate) that was used to raise a polyclonal antibody in
rabbits. NOR-acid was also attached to wells of Covalink plates
and served as the antigen for enzyme-linked immunosorbent
assays (ELISAs) performed to characterize the antibody.
Syn th esis of th e Im m u n ogen . Reactions were set up to
bind NOR-acid or [14C]NOR-acid to BSA. Briefly, NOR-acid was
combined with 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide
HCl (1 equiv) and N-hydroxysuccinimide (2 equiv) in 50/50
dimethylformamide/water and incubated at room temperature
for 2 h. BSA (1/140 equiv) in 1 N sodium bicarbonate was added
to the mixture, which was chilled to 4 °C and agitated for 24 h.
The coupled protein was dialyzed with four changes of 50 mM
NaHPO4 (pH 8.0) over a 24-h period. A 50% coupling efficiency
of NOR-acid to available lysine binding sites occurred, as
determined by counting radioactivity present in the product of
the reaction of [14C]NOR-acid and BSA. Binding of BSA to
NOR-acid was also confirmed by one-dimensional SDS-PAGE.
Electrophoretic migration of the NOR-BSA conjugate was
compared to that of BSA.
Exp er im en ta l P r oced u r es
Ma ter ia ls. (-)-N-Norcocaine was obtained from the Na-
tional Institute on Drug Abuse through the Research Triangle
Institute (Research Triangle Park, NC). Succinic anhydride,
succinic anhydride-1,4-14C (1.7 mCi/mmol), N-carbobenzoxy-â-
alanine (CBZ-â-alanine), 1-cyclohexyl-3-(2-morpholinoethyl)-
carbodiimide p-toluenesulfonate (CCD), 1-ethyl-3-[3-(dimethyl-
amino)propyl]carbodiimide HCl, N-hydroxysuccinimide, p-
nitrophenyl phosphate, goat anti-rabbit IgG coupled to alkaline
phosphatase, normal sheep serum, bovine serum albumin (BSA),
3,3′-diaminobenzidine (DAB), and prestained molecular weight
markers were purchased from Sigma Chemical Co. (St. Louis,
MO). Dimethyl sulfoxide (DMSO), sodium chloride, magnesium
sulfate, magnesium chloride, Tween 20, cyclohexane, tetrahy-
drofuran, dichloromethane, and ethyl acetate were purchased
from Fisher Scientific Co. (Pittsburgh, PA). Palladium on
activated carbon was purchased from Aldrich Chemical Co.
(Milwaukee, WI). Covalink ELISA plates were purchased from
Nunc Inc. (Naperville, IL). TLC analytical sheets, TLC pre-
parative plates, and reagents for SDS-PAGE were obtained
from Fisher Scientific (Pittsburgh, PA). PVDF membrane was
purchased from Micron Separations Inc. (Westborough, MA).
All chemicals and reagents were of reagent grade or higher.
Im m u n iza tion P r otocol. A polyclonal antibody against the
NOR-BSA antigen was raised in three female New Zealand
white rabbits by Rockland Immunochemicals (Boyerstown, PA).
Rabbits were bled initially to obtain preimmune serum and then
immunized with an intradermal injection (2 mL) of 0.5 mg of
NOR-BSA in Freund’s complete adjuvant. After 1 week, rabbits
received a booster injection intradermally of 0.5 mg of NOR-
BSA in Freund’s incomplete adjuvant (FIA). On day 14 and
for every 2 weeks for the next 4 months, 0.5-mg booster
injections in FIA were administered. After 4 months, boosters
were administered once a month with periodic production
bleeds. A termination bleed was performed 7 months after
initial immunization.
Syn th esis of th e Ha p ten . Norcocaine acid (NOR-acid) was
synthesized by attaching a free acid moiety to the nitrogen of
norcocaine using a three-reaction procedure (Scheme 1) (11-
15).
ELISAs. ELISAs were performed using polystyrene plates
grafted with secondary amino groups (Nunc Covalink NH
plates). NOR-acid (antigen) was bound to plate wells via
N-hydroxysuccinimide linkage through a carbodiimide-mediated
reaction. Fifty microliters (50 µL) of a solution containing NOR-
1
S. Roberts, unpublished observations.

Immunodetection of Cocaine-Protein Adducts
Chem. Res. Toxicol., Vol. 11, No. 3, 1998 187
acid in DMSO and N-hydroxysuccinimide (1.3 equiv) in co-
vabuffer (phosphate-buffered saline, pH 7.2, containing 116.9
g of sodium chloride, 10 g of magnesium sulfate heptahydrate,
and 0.5 mL of Tween 20) was placed in each well. Then, 50 µL
of 1-ethyl-3-[3-(dimethylamino)propyl]carbodiimide HCl (1.5
mM) was added to each well to initiate the reaction, and plates
were incubated for 90 min at room temperature. After incuba-
tion, wells were washed three times with covabuffer, blocked
with 3% normal sheep serum for 2 h at room temperature,
blotted dry, covered, and stored at -20 °C until needed. The
amount of antigen bound to wells varied according to experi-
mental protocols.
gel, 10% T, 2.7% Cbis; stacking gel, 4% T, 2.7% Cbis) (16).
Following one-dimensional SDS-PAGE, proteins in whole liver
homogenate were further characterized by two-dimensional (2D)
PAGE. Samples were diluted with isoelectric focusing (IEF)
sample buffer (9 M urea, 2% NP-40, 2% ampholine), and
proteins were separated according to their isoelectric points,
followed by electrophoretic separation on a slab gel according
to their molecular weights (separating gel, 10% T, 2.7% Cbis
;
stacking gel, 4% T, 2.7% Cbis) (17). Each sample was run in
duplicate simultaneously. The gels either were stained with
Coomassie blue and destained to visualize proteins or were used
unstained for protein transfer to membrane which was either
stained with Coomassie blue or probed with anti-NOR antibody.
To determine the optimum antibody titer for assays, ELISA
was performed using different antibody dilutions in wells
containing varying concentrations of antigen bound to Covalink
NH. Wells were blocked with 3% normal sheep serum; then
100 µL of rabbit anti-NOR antibody (1:125-1:8000 dilution in
3% normal sheep serum in covabuffer) was added. Plates were
incubated for 90 min at room temperature and washed three
times with covabuffer. Alkaline phosphatase-conjugated goat
anti-rabbit IgG (100 µL of 1:20000 dilution) was added to each
well, followed by incubation for 90 min at room temperature.
Plate wells were washed three times with covabuffer, and 50
µL of p-nitrophenyl phosphate (1 mg/mL substrate in Tris buffer,
pH 9) was added to develop color. Absorbance was read after 1
h at 450 nm using an ELISA plate reader (Molecular Devices,
Menlo Park, CA).
P r otein Tr a n sfer a n d Wester n Blot An a lysis for Co-
ca in e Bin d in g. Proteins in unstained gels were electrotrans-
ferred (Trans-Blot Cell, Biorad Laboratories, Hercules, CA) to
PVDF membrane in morpholinoethanesulfonic acid buffer (10
mM morpholinoethanesulfonic acid, pH 6, 20% methanol) for
16 h at 20 V, based on the method of Towbin and others (18).
Membrane was equilibrated in TTBS (20 mM Tris, pH 7.5, 0.5
M sodium chloride, 0.02% sodium azide, 0.05% Tween 20) for
30 min, incubated in blocking buffer (3% normal sheep serum
in TTBS) for 19 h to minimize nonspecific binding, and washed
three times (5 min each) with TTBS. Membrane was then
incubated with anti-NOR antibody (1:100 in blocking buffer) for
3 h, washed four times (5 min each) in TTBS, and incubated
with goat anti-rabbit IgG-alkaline phosphatase conjugate (1:
15000 in blocking buffer) for 1 h. Following three washes (5
min each) with TTBS, the membrane was placed in chromogen
[0.6% (v/v) nitro blue tetrazolium, 0.3% (v/v) 5-bromo-4-chloro-
3-indolyl phosphate, in alkaline phosphatase buffer (100 mM
sodium chloride, 5 mM magnesium chloride, 100 mM Tris, pH
9.5)] for development of binding signals. The color reaction was
stopped by washing the membrane with distilled water.
For competitive ELISAs, wells of microtiter plates coated with
10 µM NOR-acid (antigen) were incubated with antibody (1:1000
dilution) which had been preadsorbed with varying concentra-
tions (1-105 µM) of cocaine and its major metabolites (norco-
caine, N-hydroxynorcocaine, benzoylecgonine, and ecgonine
methyl ester). ELISA was carried out as before to determine
the relative ability of each metabolite to inhibit antibody binding
to the antigen.
For competitive Western blots, proteins in liver homogenate
(200 µg of protein/gel) from cocaine-treated and saline-treated
animals were separated by 2D SDS-PAGE. Proteins were
transferred to PVDF membrane and probed with anti-NOR
antibody as described above. To test specificity of the antibody
for proteins adducted by cocaine metabolites, Western blot
analysis was repeated on replicate samples using antibody
preadsorbed with 10 mM norcocaine for 16 h.
An im a ls a n d Tr ea tm en ts. Experiments evaluating hepatic
protein adduction by cocaine were conducted using ICR male
mice (Harlan Sprague-Dawley, Indianapolis, IN) weighing 28-
36 g. Mice were housed in temperature-controlled animal
quarters with a 12-h light/dark cycle and given free access to
food and water. Mice were normally kept 5/cage in polycarbon-
ate cages with corn cob bedding. Mice were administered a
single dose of cocaine (50 mg/kg, ip), with vehicle (saline)-treated
animals serving as controls. To test the requirement of meta-
bolic activation for cocaine-product adduct formation, some
mice were pretreated with a P450 inhibitor, SKF 525A (Smith-
Kline Beecham Pharmaceuticals, Philadelphia, PA) (50 mg/kg,
ip), 30 min before the cocaine dose. Mice were euthanized by
carbon dioxide asphyxiation 6 h after the cocaine dose, and livers
were collected. All procedures were approved by the Institu-
tional Animal Care and Use Committee at the University of
Florida.
Im m u n oh ist och em ica l Det ect ion of Coca in e Bin d in g
Usin g An ti-NOR An tibod y. Liver tissue was collected from
mice 6 h following a single injection of either cocaine HCl (50
mg/kg, ip) or saline. The tissue was fixed in neutral buffered
10% formalin for 3 h, placed in saline at 4 °C, trimmed,
embedded in formalin, and sectioned at 4-6 µm. Sections were
deparaffinized and hydrated by treatment with xylene, 100%
ethanol, 95% ethanol, and distilled water. Sections were covered
with quenching solution (3% H2O2, 0.1% sodium azide) for 10
min, rinsed in water (three times, 2 min each) and TBS (20 mM
Tris, pH 7.6, 0.5 M sodium chloride), and then incubated with
blocking solution [25% (v/v) normal sheep serum in TBS] for 2
h at 37 °C in a humid box. Slides were washed twice in TBS
and then incubated with anti-NOR antibody [1:100 dilution in
3% (v/v) normal sheep serum in TBS] for 2 h at 37 °C. Sections
were washed in TBS (four times, 2 min each), incubated with
biotinylated rabbit IgG (1:750 dilution; Biostain Rabbit IgG
System, Biomeda Corp., Foster City, CA) for 1 h at 37 °C,
washed in TBS (three times, 2 min each), and treated with
peroxidase-conjugated streptavidin [1:200 in 3% (v/v) normal
sheep serum in TBS], streptavidin-HRP (Southern Biotechnol-
ogy Associates Inc., Birmingham, AL) for 30 min at 37 °C.
Sections were rinsed in TBS, treated with chromogen solution
(3.5 mg of DAB/urea H2O2 in PBS, 0.06% + 550,uL of 3% NiCl2)
for 8 min, rinsed in running tap water for 5 min, counterstained
with Mayer’s modified hematoxylin for 2 min, rinsed in tap
water (5 min), blued in Scott’s tap water (30 mM NH4OH, 12
slow dips), and rinsed again in tap water. Sections were
dehydrated through ethanol (95% and 100%) and xylene and
examined with a light microscope. To serve as negative controls,
Liver Cell F r a ction a tion a n d SDS-P AGE. Liver was
collected at different time points after cocaine injection and
placed in ice-cold preparation buffer (PB; 10 mM Tris-HCl, 0.25
M sucrose, 1 mM magnesium chloride, 0.05%, w/v, BSA, pH 7.4).
Liver tissue was chopped into small pieces and hand homog-
enized in preparation buffer using a Kontes Dounce-type tissue
grinder (Fisher Scientific). The homogenate was centrifuged
at 700g for 15 min, and the supernatant was centrifuged at
7000g for 15 min to pellet the mitochondria. The pellet was
rinsed and resuspended in PB, and the supernatant was
centrifuged at 106000g for 90 min to pellet the microsomes. The
microsomal pellet was washed by resuspension in buffer and
centrifugation at 106000g for 60 min. All manipulations were
carried out on ice, and homogenate and cell fractions were stored
at -80 °C until needed.
For one-dimensional SDS-PAGE, whole liver homogenate
and cell fractions were diluted with solubilization buffer (2%
SDS, 5% 2-mercaptoethanol, 20% glycerol, 0.025% bromophenol
blue, 0.1 M Tris, pH 7.0), and proteins were separated by
electrophoresis according to their molecular weights (separating

188 Chem. Res. Toxicol., Vol. 11, No. 3, 1998
Ndikum-Moffor et al.
Figu r e 2. ELISA to determine optimum conditions for antigen-
antibody interaction. Wells containing varying amounts of
antigen were incubated with different dilutions of antibody as
described in Experimental Procedures.
F igu r e 1. One-dimensional SDS-PAGE to verify coupling of
BSA to NOR-acid. Bovine serum albumin and NOR-BSA
conjugate were separated on a 10% acrylamide gel (20 µg of
protein/well).
replicate sections were incubated with either preimmune serum
or antibody preadsorbed with 10 mM norcocaine or blocking
solution in place of the primary antibody. Other negative
controls were sections incubated with the primary antibody but
using blocking buffer in place of the secondary antibody. To
test specificity of the antibody, sections from necrotic liver of
mice treated with acetaminophen, a known hepatotoxicant, were
also used for immunohistochemistry.
Resu lts
NOR-a cid BSA Cou p lin g. BSA was successfully
linked to the hapten (norcocaine acid, NOR-acid) to form
the immunogen, with approximately a 50% coupling
efficiency as determined radiometrically. Additional
confirmation of coupling was provided by one-dimen-
sional SDS-PAGE, which showed a higher molecular
weight (about 75 kDa) for the NOR-BSA conjugate than
for BSA (68 kDa) alone (Figure 1). Based on the change
in molecular weight, it appears that approximately 15-
16 molecules of NOR-acid (mass 460) were bound to each
BSA molecule. It has been reported that 30-35 lysines
are available for binding (19).
F igu r e 3. Competitive ELISA to determine the relative ability
of cocaine and metabolites to inhibit antibody binding to NOR-
acid (antigen) bound to wells of microtiter plates. Cocaine,
norcocaine, N-hydroxynorcocaine, ecgonine methyl ester, and
benzoylecgonine were tested at varying concentrations. The
N-oxidative metabolites (norcocaine and N-hydroxynorcocaine)
were more potent inhibitors than the esteratic metabolites
(benzoylecgonine and ecgonine methyl ester).
antibody to the NOR-acid antigen, with IC50 values of
0.062, 0.061, and 0.022, respectively (Figure 3). For the
esteratic cocaine metabolite benzoylecgonine, about 100-
fold higher concentrations were required to produce 50%
inhibition, and affinity of the antibody for ecgonine
methyl ester was sufficiently weak that a 50% inhibition
could not be obtained using ecgonine methyl ester
concentrations as high as 100 mM.
Wester n Blots. Mice were treated with either saline
or a hepatotoxic dose of cocaine (50 mg/kg, ip) and
euthanized at varying time intervals (30-360 min) to test
for the presence of cocaine-protein adducts in liver.
Hepatic proteins from whole liver homogenate were
separated by SDS-PAGE, transferred to PVDF mem-
brane, and probed with the anti-NOR antibody. As
shown in Figure 4, antibody binding to several proteins
was observed in cocaine-treated mice, with major bands
An tibod y Titer . Figure 2 shows the effects of anti-
body dilution at varying levels of antigen concentration.
For each concentration of antigen, the binding signal
declined with increasing dilution of antibody, while for
each antibody dilution binding got stronger as antigen
concentration increased. Antigen-antibody interactions
were highest for antigen concentrations between 5 and
10 µM and antibody dilutions between 1:100 and 1:1000.
On the basis of these results, an antigen concentration
of 10 µM and an antibody dilution of 1:100 or 1:1000 were
used in subsequent assays.
In h ibition of Bin d in g. Competitive ELISA revealed
a distinct contrast between cocaine and its N-oxidative
metabolites versus esteratic cocaine metabolites in terms
of immunoreactivity with the anti-NOR antibody. Co-
caine, norcocaine, and N-hydroxynorcocaine were ap-
proximately equally effective in inhibiting binding of the

Immunodetection of Cocaine-Protein Adducts
Chem. Res. Toxicol., Vol. 11, No. 3, 1998 189
F igu r e 4. Detection of target proteins by Western blot analysis
following a hepatotoxic dose of cocaine. Liver homogenate was
prepared from tissue collected at varying times after a dose of
cocaine (50 mg/kg, ip). Proteins (80, µg/lane) were separated by
one-dimensional SDS-PAGE, transferred to PVDF membrane,
and probed with anti-NOR antibody. Representative blot indi-
cates protein adduction across time after cocaine treatment;
each lane contains liver homogenate from a different mouse.
evident at approximately 20, 44, 52-54, and 64 kDa
using one-dimensional SDS-PAGE. Antibody binding
to each of these major bands was evident at each of the
time points examined, including as early as 30 min after
the cocaine dose, and appeared to increase somewhat
with time. Several other minor binding signals were
observed in cocaine-treated animals. Little or no anti-
body binding to these protein bands occurred in liver from
saline control mice, except for some nonspecific binding
at 29, 37, 66, and 102 kDa (Figure 4). The 66-kDa band
is a result of immunoreactivity of the antiserum to
albumin in the preparation buffer, since the immunogen
was a NOR-BSA conjugate. The identity of the other
proteins is not known. As an additional control, Western
blots from cocaine-treated mice were probed with anti-
NOR antibody preadsorbed with norcocaine (10 mM). No
antibody binding was observed for these blots (not
shown).
Further characterization of proteins in whole liver
homogenate by Western blot analysis of second-dimen-
sional gels showed that the target proteins had pI’s
ranging from 4.5 to 7.0 (arrows, Figure 5, top). The
protein at approximately 20 kDa had a pI of about 6.0.
Two binding signals were observed around 44 kDa with
pI’s of 5.0 and 7.0. The 52-54- and 64-kDa proteins had
pI’s of 4.5 and 5.5, respectively (Figure 5, top). Corre-
sponding binding signals were not observed in samples
from saline-treated controls (Figure 5, bottom). Some
nonspecific binding was present between 12 and 20 kDa
in both samples.
F igu r e 5. Western blot analysis of proteins in whole liver
homogenate from mice injected with either (top) cocaine (50 mg/
kg, ip) or (bottom) saline. Proteins (200 µg/gel) were separated
by 2D SDS-PAGE, transferred to membrane, and probed with
anti-NOR antibody. Blots indicate intense binding to proteins
with approximate molecular masses of 20 kDa (pI ) 6.0), 44
kDa (pI’s of 5.0 and 7.0), 52-54 kDa (pI ) 4.5), and 64 kDa (pI
) 5.5) only in samples from cocaine-treated mice.
kDa observed in whole liver homogenate from cocaine-
treated mice (described above) were found in both mito-
chondria and microsomal fractions. The 64-kDa adduct-
ed protein was present only in mitochondria (Figure 6).
None of the major cocaine-protein adducts observed in
whole liver homogenate were found in the cytosolic
fraction. As with whole liver homogenate, several minor
binding signals were observed in all fractions (Figure 6),
and the intensity of binding to target proteins appeared
to increase over time after cocaine treatment (data not
shown).
Pretreatment with SKF 525A inhibited the formation
of the major mitochondrial and microsomal cocaine-
protein adducts (Figure 6). The binding pattern for all
the fractions was similar between the group pretreated
with SKF 525A and the saline-treated controls.
Im m u n oh istoch em istr y. Liver sections were taken
from mice treated with cocaine (50 mg/kg, ip) at 30, 60,
120, 240, and 360 min after the dose and, for comparison,
from saline-treated mice. Immunostaining was evident
as early as 30 min after the cocaine dose and was
restricted to hepatocytes in the midzonal and centrilobu-
lar regions (Figure 7, top), matching regions of necrosis
observed with hematoxylin-eosin (H&E) staining (Figure
7, middle). Preadsorption of the antibody with norcocaine
prior to use completely prevented immunostaining (Fig-
ure 7, bottom), and no immunostaining with the anti-
NOR antibody was present in liver sections from mice
injected with saline vehicle only. When preimmune
serum or blocking buffer was used in place of the anti-
Su bcellu la r Distr ibu tion of Coca in e-P r otein Ad -
d u ct s a n d E ffect s of P 450 In h ib it ion on Ad d u ct
F or m a tion . Subcellular fractionations corresponding to
cytosol, mitochondria, and microsomes were isolated from
cocaine-treated (50 mg/kg, ip) and saline control mice and
evaluated by Western blot analysis following one-
dimensional SDS-PAGE. Bands resulting from specific
antibody binding at approximately 20, 44, and 52-54

190 Chem. Res. Toxicol., Vol. 11, No. 3, 1998
Ndikum-Moffor et al.
F igu r e 6. Western blot analysis of liver cell fractions. Mice
received a single dose of cocaine (50 mg/kg, ip) with or without
pretreatment with SKF 525A (50 mg/kg, ip), a P450 inhibitor.
Subcellular fractions were prepared as described in Experimen-
tal Procedures; proteins in fractions were separated by one-
dimensional SDS-PAGE (50 µg/lane), transferred to mem-
branes, and probed with anti-NOR antibody.
NOR antibody, or when the secondary antibody was
replaced with blocking buffer, no immunoreactivity was
observed. To further insure that antibody binding was
not in response to some hepatic antigen unmasked by
hepatic necrosis, the anti-NOR antibody was tested with
liver sections from mice administered a hepatotoxic dose
of acetaminophen (400 mg/kg, ip). No immunoreactivity
was observed (not shown).
Discu ssion
The mechanism by which cocaine produces hepatic
necrosis is not well-understood. While it is certain that
N-oxidative metabolism of cocaine is a key aspect,
precisely how this leads to cytotoxicity has not been
established. Hypotheses have been proposed that im-
plicate either the generation of reactive oxygen speciess
through a futile redox cycle between N-hydroxynorco-
caine and norcocaine nitroxide or from poorly coupled
microsomal oxidationsor the adduction of critical cellular
proteins by cocaine-reactive metabolites (see ref 20 for a
review). Evidence for protein adduction as the mecha-
nism comes primarily from observations regarding the
strong correlation between the extent of protein adduc-
tion and toxicity. Though some interventions, such as
cotreatment of hepatocytes in culture with dithiothreitol,
have been observed to reduce toxicity without affecting
the extent of protein binding (21), these observations do
not rule out protein adduction as the initiating event in
toxicity. It can be argued that such interventions do not
necessarily mean that protein adduction is not essential,
only that modulation of cellular response to the adduction
can significantly alter the outcome in terms of cytotox-
icity. It is significant to note that there has been no
demonstration of cocaine hepatotoxicity without protein
adduction and that manipulations that decrease protein
binding always diminish toxicity.
F igu r e 7. Immunohistochemical detection of cocaine binding
in mouse liver. Sequential liver sections were cut from the liver
of a mouse treated with cocaine (50 mg/kg, ip) and euthanized
6 h later: (top) immunostain using anti-NOR antibody, (middle)
hematoxylin-eosin stain, and (bottom) immunostain using anti-
NOR antibody preadsorbed with norcocaine (10 mM). For
reference purposes, the arrow in each panel points to the same
central vein. Original magnification, 100× (reproduced at 50%
of original).
Despite evidence for the importance of protein adduc-
tion in cocaine hepatotoxicity, very little is known about
this binding. By administering cocaine radiolabeled in
various positions to mice, Evans (8) demonstrated that
the reactive metabolite is formed after N-demethylation
and that the reactive metabolite bound to proteins has
both ester groups intact. These observations have been
interpreted as indicating that protein adduction occurs
via oxidation of the nitrogen to a reactive species. The
rate of adduction of proteins after a cocaine dose is
rapidspeak levels of radiolabel irreversibly bound to
protein have been observed 4 h after administration of a

Immunodetection of Cocaine-Protein Adducts
Chem. Res. Toxicol., Vol. 11, No. 3, 1998 191
radiolabeled dose of cocaine in phenobarbital-induced
Swiss-Webster mice (8) and 1 h after the dose in naive
ICR mice.2 Very little information is available regarding
potential protein targets. One study in which [benzoyl-
3H-]cocaine was incubated with isolated hepatocytes from
phenobarbital-pretreated rats found the radiolabel to be
associated with a 53-kDa protein which was immunore-
active with anti-rat CYP2B1 antibody (22). No other
attempts to identify cocaine-adducted proteins have
appeared in the literature.
cocaine is similar to that observed previously in a study
from this laboratory using a commercially obtained anti-
cocaine antibody3 (25).
While it is possible that some of the observed immu-
nohistochemical staining reflects binding to residual,
unbound cocaine (or N-oxidative metabolites) remaining
in the liver section rather than cocaine-protein adducts,
the contribution of this to total immunostaining is
probably minor. Most of the soluble cocaine near the
surface of the liver section is no doubt lost during routine
processing, including the incubation and washing steps
associated with immunostaining. Also, the time course
for the presence of positive immunostaining is consistent
with protein-adduct formation rather than free cocaine
or metabolites. Hepatic concentrations of cocaine and
norcocaine peak quickly after a cocaine dose in mice,
reaching a maximum 15 min after the dose and falling
to nearly undetectable levels at 90 min postadministra-
tion (26). Immunostaining, in contrast, appears soon
after the cocaine dose but remains stable for at least 6
h, consistent with the several hours required to clear
cocaine-adducted protein (8).
Using an antigen created by linking the cocaine
molecule to a carrier protein through the tropane nitro-
gen, an antibody was generated that may be very useful
in identifying cocaine-adducted proteins. Given the
orientation of the cocaine molecule with respect to the
carrier protein and its presentation for immune recogni-
tion, it is not surprising that the antibody was fully cross-
reactive with cocaine metabolites modified at the nitrogen
(norcocaine and N-hydroxynorcocaine) but had greatly
diminished binding to the cocaine molecule when either
of the two ester groups was lost (benzoylecgonine or
ecgonine methyl ester). This selectivity makes the
antibody well-suited to the detection of cocaine bound to
proteins through oxidative metabolism.
While protein adduction by cocaine-reactive metabo-
lites has been a promising mechanism to explain cocaine
hepatotoxic effects for nearly 20 years, the ability to
explore this mechanism has been impeded by the lack of
effective tools for detecting cocaine-protein adducts. The
antibody described here provides this capability. With
SDS-PAGE, it can be used to detect adducted proteins
that could be subsequently sequenced and identified.
With this information, hypotheses regarding the effects
of adduction of specific proteins and their role in toxicity
can be developed and tested. Immunohistochemistry can
also provide useful information regarding the distribution
of protein adduction, not only within the lobule but also
within subcellular organelles using electron microscopy.
Together, these approaches offer the potential to signifi-
cantly expand our understanding of the effects of cocaine
on the liver.
Probing blots of hepatic proteins from cocaine-treated
mice with the anti-NOR antibody following one-dimen-
sional SDS-PAGE resulted in relatively intense staining
of proteins at approximately 20, 44, 52-54, and 64 kDa.
These proteins were not immunoreactive in saline-treated
mice, and several control experiments indicated the
immunoreactivity was specific for cocaine. Two-dimen-
sional SDS-PAGE revealed that the 44-kDa band in fact
contained two cocaine-adducted proteins. This suggests
that there are at least five proteins which are preferential
targets for adduction, and the Western blot results
(Figures 4 and 5) indicate that there are perhaps several
more that are adducted to a lesser extent. A previous
study (21) using isolated hepatocytes from phenobarbital-
induced rats found CYP2B1/B2 to be the principal target
of cocaine adduction. It is possible that one of the bands
of cocaine-adducted protein between 52 and 54 kDa in
the present study may also be a P450, though perhaps
not CYP2B1/B2. While CYP2B1/B2 is important for
cocaine bioactivation and toxicity in the phenobarbital-
induced rat (23), cocaine bioactivation in the mouse
appears to require instead CYP3A activity (24). This
would suggest that, if any of the 52-54-kDa proteins is
a P450, CYP3A may be the more likely target for
adduction.
Ack n ow led gm en t. This research was supported by
Grant DA-06601 from the National Institute on Drug
Abuse.
Refer en ces
(1) Kanell, G. C., Cassidy, W., Shuster, L., and Reynolds, T. B. (1990)
Cocaine-induced liver cell injury: Comparison of morphological
features in man and in experimental models. Hepatology 11, 646-
651.
Immunochemical staining using the anti-NOR anti-
body showed binding only in liver sections from cocaine-
treated mice. Immunostaining was lost when (1) preim-
mune serum was substituted for the primary antibody
preparation, (2) blocking buffer was substituted for the
secondary antibody, (3) antibody was preadsorbed with
norcocaine, or (4) hepatotoxicity was produced by a
different toxicant (acetaminophen), indicating that the
immunoreactivity was specific for cocaine. Immuno-
staining for cocaine was confined to midzonal and cen-
trilobular regions of the lobule, which is where hepatic
necrosis from cocaine characteristically occurs in this
mouse strain. The close spatial relationship between
positive immunoreactivity and the hepatic lesion from
(2) Perino, L. E., Warren, G. H., and Levine, J . S. (1987) Cocaine-
induced hepatotoxicity in humans. Gastroenterology 85, 122-129.
(3) Radin, D. R. (1992) Cocaine-induced hepatic necrosis: CT dem-
onstration. J . Comput. Asst. Tomog. 16, 155-156.
(4) Wanless, I. R., Dore, S., Gopinath, N., Tan, J ., Caeron, R.,
Heathcote, E. J ., Blendis, L. M., and Levy, G. (1990) Histopath-
ology of cocaine hepatotoxicity. Gastroenterology 98, 497-501.
(5) Silva, M. O., Roth, D., Reddy, K. R., Fernandez, J . A., Albores-
Saavedra, J ., and Schiff, E. R. (1991) Hepatic dysfunction ac-
companying acute cocaine intoxication. J . Hepatol. 12, 312-315.
(6) Kloss, M. W., Rosen, G. M., and Rauckman, E. J . (1984) Cocaine-
mediated hepatotoxicity. A critical review. Biochem. Pharmacol.
33, 169-173.
3
The commercial antibody used previously was effective for immu-
nohistochemical staining but could not be used to probe Western blots
due to considerable nonspecific binding.
2
S. Roberts, unpublished observations.

192 Chem. Res. Toxicol., Vol. 11, No. 3, 1998
Ndikum-Moffor et al.
(7) Roberts, S. M., J ames, R. C., and Harbison, R. D. (1992) Cocaine-
induced hepatotoxicity. In Cocaine: Pharmacology, Physiology,
and Clinical Strategies (Lakoski, J . M., Galloway, M. P., and
White, F. J ., Eds.) pp 15-33, CRC Press, Boca Raton, FL.
(8) Evans, M. A. (1983) Role of protein binding in cocaine-induced
hepatic necrosis. J . Pharmacol. Exp. Ther. 224, 73-79.
(9) Charkoudian, J . C., and Shuster, L. (1985) Electrochemistry of
norcocaine nitroxide and related compounds: Implications for
cocaine hepatotoxicity. Biochem. Biophys. Res. Commun. 130,
1044-1045.
(10) Thompson, M. L., Shuster, L., and Shaw, K. (1979) Cocaine-
induced hepatic necrosis in mice - The role of cocaine metabolism.
Biochem. Pharmacol. 28, 2389-2395.
(11) Hoare, D. G., and Koshland, D. E. (1967) A method for the
quantitative modification and estimation of carboxylic acid groups
in proteins. J . Biol. Chem. 242, 2447-2453.
(12) Bauminger, S., and Wilchek, M. (1980) The use of carbodiimides
in the preparation of immunizing conjugates. Methods Enzymol.
70, 151-159.
(13) Erlanger, B. F. (1980) The preparation of antigenic hapten-carrier
conjugates: A survey. Methods Enzymol. 70, 85-104.
(14) Goodfriend, T. L., Levine, L., and Fasman, G. D. (1964) Antibodies
to bradykinin and angiotensin: A use of carbodiimides in im-
munology. Science 144,1344-1346.
sheets: Procedure and some applications. Proc. Natl. Acad. Sci.
U.S.A. 76, 4350-4354.
(19) Harlow, E., and Lane, D. (1988) Antibodies: A Laboratory
Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor,
New York.
(20) Boelsterli, U. A., and Goldlin, C. (1991) Biomechanisms of cocaine-
induced hepatocyte injury mediated by the formation of reactive
metabolites. Arch. Toxicol. 65, 351-360.
(21) Goldlin, C., and Boelsterli, U. A. (1994) Dissociation of covalent
protein adduct formation from oxidative injury in cultured
hepatocytes exposed to cocaine. Xenobiotica 24, 251-264.
(22) Boelsterli, U. A., Oertle, M., and Goldlin, C. (1992) Bioactivated
cocaine irreversibly binds to cytochrome P450IIB1 in rat hepa-
tocytes. Toxicologist 12, 67.
(23) Poet, T. S., McQueen, C. A., and Halpert, J . R. (1996) Participation
of cytochromes P4502B and P4503A in cocaine toxicity in rat
hepatocytes. Drug Metab. Dispos. 24, 74-80.
(24) Pellinen, P., Honkakoski, P., Stenback, F., Niemitz, M., Alhava,
E., Pelkonen, O., Lang, M. A., and Pasanen, M. (1994) Cocaine
N-demethylation and the metabolism-related hepatotoxicity can
be prevented by cytochrome P450 3A inhibitors. Eur. J . Phar-
macol.-Environ. Toxicol. Pharmacol. 270, 35-43.
(25) Roth, L., Harbison, R. D., J ames, R. C., Tobin, T., and Roberts,
S. M. (1992) Cocaine hepatotoxicity: Influence of hepatic enzyme
inducing and inhibiting agents on the site of necrosis. Hepatology
15, 934-940.
(15) Laundry, D. W., Zhao, K., Yang, G. X. Q., Glickman, M., and
Georgiadis, T. M. (1993) Antibody-catalyzed degradation of
cocaine. Science 259, 1899-1901.
(16) Laemmli, U. K. (1970) Cleavage of structural proteins during the
assembly of bacteriophage T4. Nature 227, 680-685.
(17) O’Farrell, P. H. (1975) High resolution two-dimensional electro-
phoresis of proteins. J . Biol. Chem. 250, 4007-4021.
(18) Towbin, H., Staelin, T., and Grodon, J . (1979). Electrophoretic
transfer of proteins from polyacrylamide gels to nitrocellulose
(26) Roberts, S. M., Munson, J . W., J ames, R. C., and Harbison, R. D.
(1992) An assay for cocaethylene and other cocaine metabolites
in liver using high-performance liquid chromatography. Anal.
Biochem. 202, 256-261.
TX970147C