Characterization of scavengers of γ-ketoaldehydes that do not inhibit prostaglandin biosynthesis

Article abstract of DOI:10.1021/tx900407a

Expression of cyclooxygenase-2 (COX-2) is associated with the development of many pathologic conditions. The product of COX-2, prostaglandin H₂ (PGH₂), can spontaneously rearrange to form reactive γ-ketoaldehydes called levuglandins (LGs). This γ-ketoaldehyde structure confers a high degree of reactivity on the LGs, which rapidly form covalent adducts with primary amines of protein residues. Formation of LG adducts of proteins has been demonstrated in pathologic conditions (e.g., increased levels in the hippocampus in Alzheimer's disease) and during physiologic function (platelet activation). On the basis of knowledge that lipid modification of proteins is known to cause their translocation and to alter their function, we hypothesize that modification of proteins by LG could have functional consequences. Testing this hypothesis requires an experimental approach that discriminates between the effects of protein modification by LG and the effects of cyclooxygenase-derived prostanoids acting through their G-protein coupled receptors. To achieve this goal, we have synthesized and evaluated a series of scavengers that react with LG with a potency more than 2 orders of magnitude greater than that with the ε-amine of lysine. A subset of these scavengers are shown to block the formation of LG adducts of proteins in cells without inhibiting the catalytic activity of the cyclooxygenases. Ten of these selective scavengers did not produce cytotoxicity. These results demonstrate that small molecules can scavenge LGs in cells without interfering with the formation of prostaglandins. They also provide a working hypothesis for the development of pharmacologic agents that could be used in experimental animals in vivo to assess the pathophysiological contribution of levuglandins in diseases associated with cyclooxygenase up-regulation.

Full text of DOI:10.1021/tx900407a

240
Chem. Res. Toxicol. 2010, 23, 240–250
Characterization of Scavengers of γ-Ketoaldehydes That Do Not
Inhibit Prostaglandin Biosynthesis
Irene Zagol-Ikapitte,† Venkataraman Amarnath,‡ Manju Bala,† L. Jackson Roberts II,†,§
John A. Oates,†,§ and Olivier Boutaud*,†
Departments of Pharmacology, Pathology, and Medicine, Vanderbilt UniVersity,
NashVille, Tennessee 37232-6602
ReceiVed NoVember 6, 2009
Expression of cyclooxygenase-2 (COX-2) is associated with the development of many pathologic
conditions. The product of COX-2, prostaglandin H2 (PGH2), can spontaneously rearrange to form reactive
γ-ketoaldehydes called levuglandins (LGs). This γ-ketoaldehyde structure confers a high degree of
reactivity on the LGs, which rapidly form covalent adducts with primary amines of protein residues.
Formation of LG adducts of proteins has been demonstrated in pathologic conditions (e.g., increased
levels in the hippocampus in Alzheimer’s disease) and during physiologic function (platelet activation).
On the basis of knowledge that lipid modification of proteins is known to cause their translocation and
to alter their function, we hypothesize that modification of proteins by LG could have functional
consequences. Testing this hypothesis requires an experimental approach that discriminates between the
effects of protein modification by LG and the effects of cyclooxygenase-derived prostanoids acting through
their G-protein coupled receptors. To achieve this goal, we have synthesized and evaluated a series of
scavengers that react with LG with a potency more than 2 orders of magnitude greater than that with the
ε-amine of lysine. A subset of these scavengers are shown to block the formation of LG adducts of
proteins in cells without inhibiting the catalytic activity of the cyclooxygenases. Ten of these selective
scavengers did not produce cytotoxicity. These results demonstrate that small molecules can scavenge
LGs in cells without interfering with the formation of prostaglandins. They also provide a working
hypothesis for the development of pharmacologic agents that could be used in experimental animals in
ViVo to assess the pathophysiological contribution of levuglandins in diseases associated with cyclooxy-
genase up-regulation.
Both COX isoforms catalyze the oxygenation of arachidonic
acid to the endoperoxide prostaglandin H2 (PGH2). PGH2 is
further metabolized to the prostanoids PGD2, PGE2, PGF2a,
thromboxane A2, and prostacyclin by specific enzymes. In
addition, prostaglandin H2 can rapidly undergo nonenzymatic
rearrangement to PGE2 and PGD2, and for 20% to the γ-ke-
toaldehydes levuglandin (LG) E2 and D2 (13).
Introduction
Chronic overexpression of cyclooxygenases (COX1) is as-
sociated with the development of a number of diseases, including
cancer and Alzheimer’s disease (for review see refs 1-4).
Chronic inflammation in the gastrointestinal tract precedes
carcinogenesis. Consistent evidence indicates that the inducible
COX isoform (COX-2) can promote the multistep sequence of
events that lead to colon cancer, and it has been demonstrated
that COX-2 is also overexpressed in gastric cancer as well as
in esophageal carcinoma, pancreatic carcinoma, prostate carci-
noma, lung cancer, and mammary cancer (5-10). In addition,
substantial epidemiologic evidence suggests that long-term use
of nonsteroid anti-inflammatory drugs (NSAIDs) prevents the
development of Alzheimer’s disease (11, 12).
Levuglandins are highly reactive molecules. They react with
free amines to form covalent adducts on lysine residues in
proteins (14-17), and their reactivity with proteins is at least 1
order of magnitude greater than that of 4-hydroxynonenal (15).
Free levuglandins have eluded detection in biological fluid
because of their high reactivity with amine groups, including
those on proteins. Several isomers of levuglandins are also
formed by nonenzymatic free radical-mediated peroxidation of
lipids (also known as isolevuglandins or isoketals), and these
γ-ketoaldehydes share the same properties as levuglandins (18).
Using liquid chromatography-tandem mass spectrometry, we
have characterized the LG adducts of proteins that are formed
by the reaction of lysine with LGE2 or PGH2 (15, 16), and
knowledge of their structures has provided a method to detect
the adducts in protein digests. We have shown that the LG-
lysine adduct is formed in activated cells (19) and in the brains
of transgenic mice (20) in a cyclooxygenase-dependent manner.
We have found that the average amount of LG-lysine lactam
in the hippocampus of patients with Alzheimer’s disease is 12-
fold higher than in that age-matched control brains; the levels
of LG-lysine lactam demonstrate a significant positive relation-
* To whom correspondence should be addressed. Division of Clinical
Pharmacology, Department of Pharmacology, Vanderbilt University, Nash-
ville, TN 37232-6602. Tel: (615) 343-7398. Fax: (615) 322-4707. E-mail:
olivier.boutaud@vanderbilt.edu.
† Department of Pharmacology.
‡ Department of Pathology.
§ Department of Medicine.
1 Abbreviations: COX, cyclooxygenase; COX, prostaglandin H2 synthase;
PG, prostaglandin; DMEM, Dulbecco’s modified Eagle’s medium; AA,
arachidonic acid; LC/ESI/MS/MS, liquid chromatography/electrospray
ionization/tandem mass spectrometry; LG, levuglandin; Iso-LG, iso-
levuglandins; HoBA, hydroxybenzylamine; EtPM, ethylpyridoxamine; PPM,
pentylpyridoxamine; BzPM, benzylpyridoxamine; SA, salicylamine; ClSA,
5-chlorosalicylamine; MeSA, methylsalicylamine; 5/3 MoSA, 5/3-methoxy-
salicylamine; EtSA, 5-ethylsalicylamine; PSA, 5-pentylsalicylamine; HxSA,
5-hexylsalicylamine; BzSA, 5-benzylsalicylamine; BoSA, 5-benzyloxysali-
cylamine.
10.1021/tx900407a  2010 American Chemical Society
Published on Web 12/31/2009

ScaVengers of γ-Ketoaldehydes
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 241
Torrance CA) with a flow rate of 0.2 mL/min. The mobile phase
consisted of solvent A (10% ACN/90% water/0.1% acetic acid)
and solvent B (ACN/0.1% acetic acid). The gradient was as follows:
0-3 min, 100% A; 3-7 min, 100 to 10% A; 7-10 min, 10% A;
10-12 min, 10 to 100% A. The ion transfer tube was operated at
35 V and 300 °C. The tube lens voltage varied depending on the
compound being analyzed. The ions were subjected to collision-
induced dissociation at -28 eV, and daughter ions were scanned
between m/z 50 and 500. Xcalibur software (version 1.3; Ther-
moFinnigan) was used to operate the instrument and to process
the data.
ship with Braak stages and CERAD neuritic plaque scores (21).
We showed that formation of LG adducts on amyloid ꢀ (Aꢀ)
accelerates its aggregation (22) into oligomers that are toxic to
cortical neurons in primary culture (23). This concerted evidence
suggests that LGs could be involved in the development of
Alzheimer’s disease.
The fact that other types of lipid modification of proteins can
alter their cellular location and function also suggests that the
formation of LG adducts of proteins could have functional
consequences.
In order to determine whether the formation of LG adducts
can alter protein function, it would be optimal to employ an
experimental intervention that will block the formation of LG-
protein adducts without inhibiting the biosynthesis of cyclooxy-
genases-derived prostanoids and thromboxane A2 that act
through G-protein coupled receptors to exert numerous patho-
logic and physiologic functions. Such an approach would consist
in using small molecules that react with LGs much faster than
lysine and would not inhibit COX activity. Pyridoxamine (PM,
13), an inhibitor of advanced glycation end-products (AGEs),
has been shown to react with LGs at a rate 2300-fold higher
than that with lysine (24). We have recently shown that
salicylamine (SA, 1) and the PM (13) analogue pentyl pyri-
doxamine (PPM, 15) prevent the formation of LG adducts on
proteins in cells without interfering with the activity of COXs
(25). In this article, we describe the synthesis and characteriza-
tion of a series of SA (1) and PM (13) analogues with the goal
of selecting compounds that have enhanced effectiveness in
inhibiting the formation of covalent LG adducts on proteins in
cells, are not inhibitors of either COX isoform, and are not
cytotoxic.
Nuclear magnetic resonance spectra were acquired by a 9.4
Bruker Spectrospin magnet equipped with a Bruker AV spectrom-
eter operating at the proton Lamour frequency of 400.13 MHz.
Proton chemical shifts are reported in parts per million (ppm) on
the δ scale relative to [2H6] DMSO (δ 2.5 ppm) or H2O deuterium
2
(δ 4.6 ppm), which also served as the H lock solvent, and the
coupling constants (J) are reported in Hz.1H spectra were acquired
in 5 mm NMR tubes using a Bruker 5 mm broadband NMR probe.
Typical experimental conditions included 32K data points, a 13-
ppm sweep width, a recycle delay of 1.5 s, and 32-256 scans,
depending on sample concentration.
Chemical Syntheses. Experimental details for the syntheses,
1
analyses by H NMR, and mass spectroscopic characterization of
pyridoxamine (13), salicylamine (1), and their analogues are
provided in Schemes 1 and 2, and Figure 1.
Synthesis of Analogues of Pyridoxamine (13). Pyridoxine was
converted to 3,4′-O-isopropylidenepyridoxine (28) and was treated
with NaH in THF followed by 1-iodopentane. The pentyl derivative
was hydrolyzed with formic acid to 5′-O-pentylpyridoxine (29).
The side chain alcohol at the 4-position was oxidized with MnO2
to aldehyde in chloroform. The aldehyde was converted to oxime
with a slight excess of hydroxylamine hydrochloride and sodium
acetate. The oxime was reduced to pentylpyridoxamine (15) with
zinc dust in acetic acid at 10-15 °C. After filtration of zinc and
removal of acetic acid, pH was raised to 8.5 with 1 M NH4OH.
The crude product was purified by flash chromatography (10-30%
methanol in ethyl acetate); mp 118-120 °C (23% overall yield).
1H NMR (DMSO-d6), δ 8.61 (brd s, 2H, NH2), 8.21 (s, 1H, 6-H),
4.78 (s, 2H, 5-CH2), 4.15 (s, 2H, 4-CH2), 3.47 (t, 2H, J ) 6.60 Hz,
5′-O-CH2), 2.69 (s, 3H, 2-CH3), 1.55 and 1.27 (dd, 6H,
CH2CH2CH2), 0.85 (t, 3H, J ) 6.76 Hz, CH3). MS m/z 239 (M +
1), 222 (M - NH2), 151 (222 - C5H11), 136 (151 - CH3).
Experimental Details
Materials. Dazoxiben was a generous gift from Pfizer Ltd.
(Sandwich, UK). Dulbecco’s modified Eagle’s medium (DMEM),
fetal bovine serum (FBS), lipopolysaccharide (LPS), INF-γ, IL-
1ꢀ, arachidonic acid, pyridoxamine dihydrochloride (13), pyridox-
ine, sodium citrate, and citric acid were purchased from Sigma-
Aldrich (St Louis, MO). [14C]-Arachidonic acid (AA) was obtained
from Perkin-Elmer Life Sciences (Boston, MA). Sepharose 2B was
purchased from Amersham Biosciences (Uppsala, Sweden). Sep-
Pak tC18 cartridges were obtained from Waters Corp. (Milford,
MA). Pronase and aminopeptidase M from porcine kidney were
purchased from Calbiochem (Gibbstown, NJ). Silica gel 60A thin
layer chromatography plates were purchased from Whatman
(Clifton, NJ). Cyclooxygenase 1 (COX-1) and cyclooxygenase 2
(COX-2) were a generous gift from Dr. Larry Marnett. COX-1 was
purified from ram seminal vesicle as described (26), and wild-type
murine COX-2 was expressed in SF-9 cells (Novagen, Madison
WI) and purified as described (27). A549 (human alveolar basal
epithelial) cells were a generous gift from Dr. Robert Coffey
(Vanderbilt University). HCA-7 (human colonic adenocarcinoma),
HepG2 (human hepatocellular carcinoma), and RAW 264.7 (murine
leukemia virus-induced monocyte/macrophage) cells were obtained
from American Type Tissue Collection (Rockville, MD). Cellular
ATP content was measured using Cytolux L001-1001 assay kits
with modifications (EG&G Wallac, Finland).
In the second step of the sequence, instead of 1-iodopentane,
benzyl bromide or iodoethane was used to obtain benzylpyridox-
amine (BzPM, 16) or ethylpyridoxamine (EtPM, 14), respectively,
in similar yields.
Benzylpyridoxamine (BzPM, 16). 1H NMR δ 8.30 (brd s, 2H,
NH2), 8.26 (s, 1H, 6-H), 7.36 and 7.31 (m, 5H, C6H5), 4.78 (s,
2H, 5-CH2), 4.61 (s, 2H, 4-CH2), 4.18 (s, 2H, C6H5CH2), 2.63
(s, 3H, 2-CH3). MS m/z 259 (M + 1), 242 (M - NH2), 151
(222 - C5H11), 91 (C6H5CH2).
Ethylpyridoxamine (EtPM, 14). 1H NMR δ 8.61 (brd s, 2H,
NH2), 8.22 (s, 1H, 6-H), 4.78 (s, 2H, 5-CH2), 4.15 (s, 2H,
4-CH2), 3.55 (q, 2H, J ) 6.96 Hz, CH2CH3), 2.69 (s, 3H,
2-CH3), 1.16 (t, 3H, J ) 6.96 Hz, CH2CH3). MS m/z 197 (M +
1), 180 (M - NH2), 151 (180 - C2H5), 136 (151 - CH3).
Synthesis of Salicylamine (1) and Its Analogues. The reduction
of appropriately substituted 2-hydroxybenzaldehyde oxime to the
amine, using zinc in acetic acid, was the final step for preparing
salicylamine (SA, 1) and its analogues. In the case of salicylal-
doxime (Acros, 2.9 g, 20 mmol), it was dissolved in acetic acid
(20 mL), cooled in an ice-water bath (10 °C), followed by the
addition of zinc dust (5 g) with stirring. The stirring was continued
at 10-15 °C for 1 h and at 20 °C for 2 h. The reaction mixture
was filtered, and the filtrate was evaporated. The solid was
crystallized from ethanol to give salicylamine acetic acid salt
(SA.AcOH, 2); 2.7 g (74%), mp 187-188 °C. It was completely
converted to hydrochloride salt (SA.HCl, 1) by refluxing in 1 M
HCl for 1 h. Excess acid was removed, and the residue was
Characterization of Structures. The identity of each compound
was confirmed by mass spectrometry (MS) and nuclear magnetic
resonance (NMR).
For mass spectrometry, the ThermoFinnigan (San Jose, CA) TSQ
Quantum triple quadrupole mass spectrometer equipped with a
standard electrospray ionization source was used. Nitrogen was used
for both the sheath and auxiliary gas. The mass spectrometer was
operated in the positive ion mode, and the electrospray needle was
maintained at 4000 V. The compounds were chromatographed on
a Prodigy 5 µm ODS (3), 150 × 1.00 mm column (Phenomenex,

242 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Scheme 1. Synthesis of Pyridoxamine (13) and Its Analogues
Zagol-Ikapitte et al.
coevaporated with ethanol, and ether was added to get a white solid;
mp 150-152 °C. 5-Chlorosalicyladehyde, 2-hydroxy-3-methoxy-
benzaldehyde, and 2-hydroxy-5-methoxybenzaldehyde (Alfa Aesar)
were converted to the oxime with a slight excess of hydroxylamine
hydrochloride and sodium acetate. Reduction of these oximes led
to 5-chlorosalicylamine (ClSA, 7), 3-methoxysalicylamine (3-
MoSA, 6), and 5-methoxysalicylamine (5-MoSA, 5), respectively.
4-Cresol, 3-ethylphenol, 4-pentylphenol, 4-hexylphenol, 4-ben-
zylphenol, and 4-benzyloxyphenol were formylated (30) to obtain
substituted salicylaldehydes that were converted to oximes and then
to 5-methylsalicylamine (MeSA, 3), 4-ethylsalicylamine (EtSA, 8),
5-pentylsalicylamine (PSA, 9), 5-hexylsalicylamine (HxSA, 10),
5-benzylsalicylamine (BzSA, 11), and 5-benzyloxysalicylamine
(BoSA, 12).
undergo rapid intermediate or slow exchange. This explains why
we do not see the peaks from NH2 and OH in the spectrum.
Salicylamine (3-SA). H NMR δ 9.63 (s, 1H, OH), 8.28 (s,
1
2H, NH2), 7.18 (t, 1H, 7.8 Hz, 3-H), 6.85 (dt, 2H, J ) 8.72 and
11.01 Hz, 2,4-H), 6.78 (dt, 1H, J ) 1.64 and 8.17 Hz, 8-H),
3.38 (s, 2H, CH2). MS m/z 124 (M + 1), 107 (124 - NH2), 77
(C6H5).
1
Salicylamine (4-SA). H NMR δ 10.22 (s, 1H, OH), 8.21 (s,
2H, NH2), 7.28 (dd, 2H, J ) 8.53 and 14.73 Hz, 2,5-H), 6.77
(d, 2H, J ) 11.25, 3,4-H), 3.86 (s, 2H, CH2). MS m/z 124 (M
+ 1), 107 (124 - NH2), 77 (C6H5).
5-Chlorosalicylamine (ClSA, 7). 1H NMR δ 10.4 (s, 1H, OH),
8.10 (s, 2H, NH2), 7.38 (d, 1H, J ) 2.6 Hz, 6-H), 7.23 (dd, 1H,
J ) 2.6 and 8.6 Hz, 3-H), 6.92 (d, 1H, J ) 8.6 Hz, 4-H), 3.34
(s, 2H, CH2). MS m/z 158 (M + 1), 141 (158 - NH2), 105
(141 - Cl), 107 (124 - NH2) and 77 (C6H5).
1
Salicylamine (2-SA, 1). H NMR δ 10.22 (s, 1H, OH), 8.28
(s, 2H, NH2), 7.30 (dd, 1H, J ) 1.5 and 7.4 Hz, 6-H), 7.17 (dt,
1H, J ) 1.5 and 8.0 Hz, 4-H), 6.94 (d, 1H, J ) 8.0 Hz, 3-H),
6.79 (dt, 1H, J ) 6.80 and 7.4 Hz, 5-H), 3.37 (s, 2H, CH2). MS
m/z 124 (M + 1), 107 (124 - NH2), 77 (C6H5).
1
5-Methylsalicylamine (MeSA, 3). H NMR δ 9.96 (s, 1H,
1
OH), 8.25 (s, 2H, NH2), 7.11 (d, J ) 1.5 Hz, 1H, 6-H), 6.97
(dd, 1H, J ) 1.5 and 8.20 Hz, 3-H), 6.85 (d, 1H, J ) 8.2 Hz,
4-H), 3.85 (s, 2H, CH2), 2.17 (s, 3H, CH3). MS m/z 138 (M +
1), 121 (138 - NH2), and the acetic acid salt gave m/z 180 (M
- H2O).
Salicylamine (SA·AcOH, 2). H NMR δ 7.04 (dt, 2H, J )
1.44 and 5.68 Hz, 2,4-H), 6.69 (dt, 2H, J ) 1.92 and 6.85 Hz,
4-H), 3.87 (s, 1H, CH2), 1.62 (s, 3H, acetyl). MS of SA.AcOH
(2) m/z 165 (M - H2O), 124 (165 - acetyl), 107 (124 - NH2),
77 (C6H5). SA·AcOH (2) has been dissolved in deuterium oxide.
Hydrogen bonding between the solvent and the molecule
decreases the electron density around the proton and may
1
5-Methylsalicylamine.AcOH (MeSA·AcOH, 4). H NMR δ
6.89 (t, J ) 2.56 Hz, 2H, CH), 6.63 (d, J ) 8.08 Hz, 1H, CH),

ScaVengers of γ-Ketoaldehydes
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 243
Scheme 2. Synthesis of Salicylamine (1) and Its Analogues
1H, 6-H), 6.85 (s, 2H, 3-H and 4-H), 4.99 (s, 2H, CH2), 3.36
(s, 2H, CH2N). MS m/z 230 (M + 1), 213 (M - NH2).
Reaction Rate of Compounds in Vitro. Compounds (1 mM)
were reacted with 4-oxopentanal (OPA, 1 mM) in 1:1 acetonitrile/
100 mM phosphate buffer, pH 7.4, at 25 °C. Aliquots were taken
to measure the concentration of pyrrole, and second-order rate
constants were calculated as previously described (24).
Analysis of COX Activity Using [14C]-Arachidonic Acid.
Because the analogues may inhibit the COXs, they were incubated
with pure COX-1 and -2 enzymes. Ovine COX-1 (3.3 nM final
concentration; specific activity, oxygenation of 377 nmol of AA/
min/mg enzyme) or wild-type murine COX-2 (7.7 nM final
concentration; specific activity, oxygenation of 154 nmol of AA/
min/mg enzyme) was preincubated on ice for 30 min with 2 mol
equiv of hematin in Tris-HCl buffer, pH 8.0, with 500 µM phenol.
The enzyme solution was incubated in the absence (control) or
presence of compounds (final concentrations, 0.1 or 1 mM) for 30
min at 37 °C. Then, 20 µL [14C]-AA (4.8 nCi, 0.5 µM final
concentration) in Tris-HCl buffer, pH 8.0, was added, and the
activity of COXs was assayed as previously described (31). Activity
of COXs is expressed as percent of control, in which no compound
was added.
Preparation of Washed Platelets. Human blood was obtained
following a protocol approved by the Institutional Review Board
of Vanderbilt University. Washed human platelets were isolated
following a previously described protocol (31). Briefly, 45 mL of
blood was drawn into a syringe containing 5 mL of 3.8% sodium
citrate and centrifuged in plastic tubes at 300g for 10 min at room
temperature. The supernatant (platelet-rich plasma) was acidified
to pH 6.4 with 0.15 M citric acid and then centrifuged at 1,000g
for 10 min at room temperature. The pellet was resuspended in
wash buffer (24.4 mM sodium phosphate, pH 6.5, 0.113 M NaCl,
and 5.5 mM glucose). After 15 min at room temperature, the
platelets were purified on a Sepharose 2B column equilibrated with
wash buffer. The eluted platelets were counted with a Coulter
counter and diluted with resuspension buffer (8.3 mM sodium
phosphate, pH 7.5, 0.109 M NaCl, and 5.5 mM glucose) to a final
concentration of 600,000 platelets/µL.
3.87 (s, 2H, 1-H), 2.01 (s, 3H, 3-H), 1.67 (s, 3H, Acetyl). MS
of acetic acid salt gave m/z 180 (M - H2O), 138 (M + 1), 121
(138 - NH2). MeSA·AcOH (4) has been dissolved in deuterium
oxide. The peak from NH2 and OH cannot be seen in the
spectrum.
3-Methoxysalicylamine (3-MoSA, 6). 1H NMR δ 9.21 (s, 1H,
OH), 8.38 (s, 2H, NH2), 6.96 (d, 1H, J ) 7.9 Hz, 5-H), 6.93 (s,
1H, 6-H), 6.78 (d, 1H, J ) 7.9 Hz, 4-H), 3.79 (s, 2H, CH2),
3.35 (s, 3H, CH3). MS m/z 154 (M + 1), 137 (154 - NH2),
107 (137 - OCH2).
Cell Culture. A549, HCA-7, and RAW 264.7 cells were cultured
in 6-well plates at 37 °C in a humidified atmosphere containing
5% CO2. HepG2 cells were cultured in T75 Falcon flasks. Cells
were grown in DMEM supplemented with 10% heat-inactivated
FBS, 100 units/mL penicillin, 100 µg/mL streptomycin, and 2.5
µg/mL amphotericin B.
5-Methoxysalicylamine (5-MoSA, 5). 1H NMR δ 9.68 (s, 1H,
OH), 8.30 (s, 2H, NH2), 6.9 (d, J ) 3.0 Hz, 1H, 6-H), 6.86 (d,
1H, J ) 8.80 Hz, 4-H), 6.76 (dd, J ) 1H, 3.0 and 8.80 Hz,
3-H), 3.88 (d, 2H, J ) 3.58 Hz, CH2), 3.66 (s, 3H, CH3). MS
same as that of 3-MOSA (6).
Cytotoxicity of Compounds in HepG2 Cells. Toxicity of the
compounds on HepG2 cells was assessed by measuring total cellular
levels of ATP according to the previously described protocol (25).
To summarize, confluent HepG2 cells were seeded in multiple
CulturPlate TC-96 plates (Perkin-Elmer, Boston, MA) at 10,000
cells per well, and the cells were allowed to adhere for 4 h. The
experimental compounds were dissolved in water to create 70 mM
stock solutions. The compounds were serially diluted in growth
medium and were added to the wells in triplicate. The cells were
then incubated under normal growth conditions for 24 h. Six wells
were left untreated as controls. After incubation, 100 µL of culture
medium was removed and assayed according to the manufacturer’s
instructions. The viability of cells was quantified by measuring ATP
levels with the ATPlite luminescence ATP detection assay system
(Perkin-Elmer) using a Packard Lumicount luminescent microplate
reader (Global Medical Instrumentation, Ramsey, MN). Cytotoxicity
was quantified on the basis of the level of ATP in compound-treated
wells compared to controls (defined as 100% viability).
Effect of the Compounds on the Level of LG-Lysine Ad-
ducts in Human Platelets. Washed platelets were preincubated
with the thromboxane synthase inhibitor dazoxiben (10 µM, final
concentration) and then with either vehicle or varying concentrations
of the different compounds (0.1 or 1 mM, final concentration) for
30 min at room temperature. Next, arachidonic acid (20 µM, final
concentration) was added, and the platelets were incubated at room
temperature for 2 h. After incubation, the platelets were pelleted at
2,000g for 10 min at room temperature. The medium was removed
4-Ethylsalicylamine (EtSA, 8). 1H NMR δ 10.00 (s, 1H, OH),
8.11 (s, 2H, NH2), 7.12 (d, 1H, J ) 7.6 Hz, 6-H), 6.72 (s, 1H,
3-H), 6.58 (d, 1H, J ) 7.6 Hz, 5-H), 3.78 (d, 2H, J ) 4.88 Hz,
CH2N), 2.42 (q, 2H, J ) 7.65 Hz, CH2), 1.04 (t, J ) 7.65 Hz,
3H, CH3). MS m/z 152 (M + 1) 135 (M - NH2).
1
5-Pentylsalicylamine (PSA, 9). H NMR δ 9.23 (brd s, 2H,
NH2), 8.11 (s, 1H, 6-H), 7.9 (d, 1H, J ), 8.16 Hz, 3-H), 7.84
(d, 1H, J ) 8.16 Hz, 4-H), 4.85 (s, 2H, CH2N), 3.43 and 2.28
(m, 8H, (CH2)4), 1.82 (t, 3H, J ) 6.68 Hz, CH3). MS m/z 194
(M + 1), 177 (M - NH2), 107.
1
5-Hexylsalicylamine (HxSA, 10). H NMR δ 10.94 (s, 1H,
OH), 9.19 (brd s, 2H, NH2), 8.11 (d, 1H, J ) 1.64 Hz, 6-H),
7.97 (dd, 1H, J ) 1.64 and 8.15 Hz, 3-H), 7.83 (d, 1H, J )
8.15 Hz, 4-H), 4.85 (s, 2H, CH2N), 4.33 (s, 2H, 5-CH2), 3.3
and 2.23 (m, 8H (CH2)4), 1.82 (t, 3H, J ) 6.3 Hz, CH3). MS
m/z 208 (M + 1), 191 (M - NH2).
1
5-Benzylsalicylamine (BzSA, 11). H NMR δ 10.05 (s, 1H,
OH), 8.17 (s, 2H, NH2), 7.25 and 7.18 (5H, C6H5), 7.04 (dd, J
) 1H, 1.7 Hz, 8.20 Hz, 6-H), 6.86 (d, 2H, J ) 8.20 Hz, 3-H
and 4-H), 3.34 (s, 2H, CH2N), 3.80 (m, 2H, CH2O). MS m/z
214 (M + 1), 197 (M - NH2).
5-Benzyloxysalicylamine (BoSA, 12). 1H NMR δ 9.73 (s, 1H,
OH), 8.29 (s, 2H, NH2), 7.40 and 7.34 (m, 5H, C6H5), 7.04 (s,

244 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Zagol-Ikapitte et al.
Figure 1. Chemical structures of aldehyde-scavenging amine compounds discussed in the text.
for analysis of prostaglandins by mass spectrometry. The pelleted
cells were collected, and the LG-lysine lactam adduct was isolated
from proteins by proteolytic digest and was analyzed by LC/MS/
MS as previously described (15).
derivatized for GC/NICI/MS analysis, monitoring selected ions as
described previously (16). The signals for the different molecules
are m/z 524 for PGD2/E2 and m/z 528 for the [2H4]-PGD2/E2 internal
standard.
Statistical Methods. Statistical analysis was performed using
GraphPad Prism 4.0 (GraphPad Software, Inc., San Diego, CA).
The significance of difference between the control and compound-
treated experiments, in different cells lines, using COX-1/2 enzymes
or for different isomers of salicylamine (1) was analyzed by one-
way analysis of variance (Tukey’s Mutliple Comparaison Test).
Quantification of LG-Lysine Adducts in A-549, HCA-7,
and RAW 267.7 Cell Lines. To assess the efficiency of compounds
to scavenge LGs in different cell lines expressing COX-2, we
selected four compounds that greatly inhibited the formation of
adducts without inhibiting COX activity. At 90% confluence, A549
cells were activated with 1 ng/mL IL-1ꢀ, and RAW 264.7 cells
were activated with 10 µg/mL LPS and 10 units/mL INF-γ
overnight, in serum-free DMEM containing penicillin and strep-
tomycin. Overconfluent HCA-7 cells in serum-free DMEM were
used without need for activation. After the addition of fresh medium
to the cells, compounds were added in desired concentrations (0.1
or 1 mM, final concentration) and incubated at 37 °C for 30 min.
Then, arachidonic acid (20 µM, final concentration) was added,
and the cells were incubated at 37 °C for 1 h. The medium was
removed for analysis of prostaglandins by mass spectrometry.
Adherent cells were scraped; protein concentration was estimated
and was used to quantify the LG-lysine lactam adduct. Experiments
were repeated twice in triplicate.
Results
Reaction of the Compounds with 4-Oxopentanal. SA (1)
and PM (13) react extremely fast with LG when compared with
N-acetyl lysine (∼1000 and ∼2300 times greater, respectively)
(24). We synthesized a series of analogues containing the
salicylamine structure and investigated their reactivity with a
γ-ketoaldehyde. The scavenging ability of the different ana-
logues was tested using 4-oxopentanal (OPA) because the
reaction forms a pyrrole, which is stable to oxidation, unlike
the reaction with LGs. The absorption spectra of the reaction
between the 2-methylpyrroles and 4-dimethylaminobenzalde-
hyde (Ehrlich’s reagent) exhibited two maxima at 528 and 568
nm, and the sum of the absorbance at the two wavelengths was
used as a measure of the pyrrole concentration (24). The second-
Quantification of Prostaglandins by GC/MS. The effect of
compounds on COX activity was determined by measuring PGE2
and PGD2 production. [2H4]-PGE2 (2 ng) was added as an internal
standard to 30 µL of the platelet supernatant or 200 µL of
conditioned cell culture medium. Prostaglandins were isolated and

ScaVengers of γ-Ketoaldehydes
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 245
Table 1. Second-Order Rate Constants for Pyrrole
Formation with Oxopentanala
Table 2. List of Compounds Devoid of Cytotoxicity up to
1 mMa
amine
kb × 103 M-1 s-1
compounds without cytotoxicity
ehtylsalicylamine (8)
1190 ( 53
1209 ( 34
1248 ( 58
1631 ( 139
1863 ( 27
2225 ( 220
2236 ( 94
2238 ( 1.6
2387 ( 193
2402 ( 186
2801 ( 245
3767 ( 220
1730 ( 76
1835 ( 169
1868 ( 78
2419 ( 172
ethylsalicylamine (8)
salicylammonium acetate (2)
salicylamine (1)
methylsalicylammonium acetate (4)
methylsalicylamine (3)
3-methoxysalicylamine (6)
5-methoxysalicylamine (5)
pyridoxamine (13)
pentylpyridoxamine (15)
ethylpyridoxamine (14)
5-chlorosalicylamine (7)
salicylammoniun acetate (2)
salicylamine (1)
methylsalicylammoniun acetate (4)
benzylsalicylamine (11)
pentylsalicylamine (9)
hexylsalicylamine (10)
methylsalicylamine (3)
3-methoxysalicylamine (6)
5-methoxysalicylamine (5)
benzoylsalicylamine (12)
benzylpyridoxamine (16)
pyridoxamine (13)
a HepG2 cells were exposed to various concentrations of compounds
for 24 h at 37 °C, after which total cellular ATP concentration was
measured. Non-toxicity was determined as more than 75% of the initial
ATP concentration remaining in the presence of a concentration of
analogue of 1,000 µM.
pentylpyridoxamine (15)
ethylpyridoxamine (14)
a A reaction mixture (1 mL) containing 2 mM each of compounds
and OPA in 100 mM phosphate buffer (pH 7.4) was taken in a 1.5 mL
Eppendorf tube that was mixed at 25 °C in an Eppendorf ThermoMix.
An aliquot of 40 µL was diluted with 460 µL of water and heated with
0.5 mL of Ehrlich’s reagent (80 mM 4-dimethylaminobenzaldehyde in
1:1 methanol (0.6 M HCl)) for 2 min at 67-69 °C. After the mixture
was cooled, the absorbances at 528 and 568 nm were measured
(Shimadzu UV-1601 spectrophotometer) and used for calculating the
concentration of 2-methylpyrrole. The sum of extinction coefficients for
Ehrlich adducts of pyrroles in this study was 68 000 ( 500. b k: second
order rate constant for the amine.
adjacent to the hydroxyl group on the aromatic ring was critical
for the increased reactivity of the scavengers for LG (24, 25).
We further investigated the effects of the different positions of
the hydroxyl group relative to the aminomethyl group using
human platelets. We found that 1 mM (final concentration)
2-HoBA (SA, 1), 3-HoBA, and 4-HoBA did not inhibit COX-1
activity, but the amount of LG-lysine lactam measured by LC/
MS varied with the position of the hydroxyl. As shown in Figure
2, formation of the LG-lysine lactam adduct was reduced by
85% and 75%, respectively, for 2-HoBA (1) and 3-HoBA.
However, 4-HoBA reduced the level of the LG adduct by only
30%, confirming the importance of the position of the hydroxyl
group for the reactivity of the scavengers.
order rate constants for amines are compared in Table 1. Our
results show that the addition of an aliphatic substituent increases
the reactivity of salicylamine by up to 1.5-fold. Increasing
hydrophobicity with methoxy and benzyloxy groups increases
reactivity of pyridoxamine.
Effect of the Hydroxyl Position on the Inhibition of LG
Lactam Adduct Formation. We have previously published data
suggesting that the relative position of the aminomethyl group
Cytotoxicity of Compounds in Hepatoma Cell Line
HepG2. The possible toxicity of the analogues was tested in a
cellular system using the hepatocarcinoma cell line HepG2.
Cytotoxicity was assessed by measuring intracellular ATP levels
using the luciferin-luciferase reaction (32). This assay is based
Figure 2. Inhibition of LG-lysine adducts and of PGE2 production by salicylamine (1) isomers in human platelets. (A) Human platelets were
preincubated with 10 µM dazoxiben and 1 mM of the specified compound or vehicle (control) for 30 min. After incubation, 20 µM arachidonic acid
was added for 2 h, and then the proteins were precipitated and digested to single amino acids by step-digestion with proteases. The daughter ions
at m/z 84.1 for the lactam and m/z 89.1 for the internal standard were monitored. Values are means ( SD (n ) 4). *P < 0.01 vs control. +P < 0.05
vs 4-HoBA. (B) COX activity was measured by the formation of PGE2 after exposure to exogenous 20 µM arachidonic acid for 2 h. Values are
means ( SD (n ) 9). (C) Structure of salicylamine analogues.

246 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Zagol-Ikapitte et al.
Table 3. Induction of Cytotoxicity by Candidate
Compoundsa
ridoxamine (BzPM, 16), which contain 6-membered rings,
caused some cytotoxicity at higher concentrations (g 500 µM),
the strongest effects were seen with 5-benzyloxysalicylamine
(BoSA, 12), 5-hexylsalicylamine (HxSA, 10), and 5-pentylsali-
cylamine (PSA, 9) at lower concentration (100 µM).
ATP levels (% of control)
concentration
(µM)
10
100
500
1000
BoSA (12)
BzSA (11)
HxSA (10) 107.36 ( 10.13
PSA (9)
90.89 ( 20.40 24.85 ( 8.29
0.25 ( 0.06 0.50 ( 0.18
Effect of the Compounds on Activity of the Two Iso-
forms of COX. To assess whether the compounds directly
inhibit cyclooxygenase activity, they were incubated with
purified ovine COX-1 or murine recombinant COX-2. The
amounts of the two COX isoforms were selected to yield equal
catalytic activity in control reactions that did not include
experimental compounds. The analogues were tested at the
highest concentration used for the cytotoxicity experiment (1
mM final concentration). The nontoxic compounds did not
inhibit COX-1 or COX-2 activity, suggesting that these ana-
logues could be used to inhibit the formation of levuglandin
adducts in cells without blocking COX activity (Figure 3). The
most toxic two compounds inhibited both of the COX iso-
forms.
98.25 ( 19.12 95.37 ( 18.27 0.475 ( 0.09 0.37 ( 0.03
0.52 ( 0.06
109.44 ( 12.87 20.14 ( 7.84
97.86 ( 3.04 101.44 ( 5.73 98.36 ( 1.64 42.22 ( 2.43
0.33 ( 0.08 0.26 ( 0.04
0.34 ( 0.02 0.31 ( 0.02
ClSA (7)
BzPM (16) 106.01 ( 14.19 100.02 ( 13.67 59.65 ( 7.67 0.56 ( 0.15
a HepG2 cells were exposed to various concentrations of compounds
for 24 h at 37 °C, after which total cellular ATP concentration was
measured. Each bar represents the mean ( SD of six values, expressed
as the percent of control (vehicle-treated cells).
on the knowledge that reduction in cellular ATP is indicative
of a cytotoxic effect. Pilot experiments demonstrated that
none of the tested compounds interferes with the enzymatic
reaction of the assay (data not shown). The compounds’ effects
after 24 h of exposure were determined. Seven analogues of
the salicylamine series and three analogues of the pyridoxamine
series presented no cytotoxicity up to a concentration of 1 mM
(Table 2). Cytotoxic effects were observed for several com-
pounds containing hydrophobic substitutions or chloride (Table
3). Although 5-benzylsalicylamine (BzSA, 11) and benzylpy-
Effects of Scavengers on LG-Lysine Lactam Adduct
Formation and on COX-1 Activity in Platelets. We have
shown previously that PM (13), PPM (15), and SA (1) inhibit
the formation of levuglandin adducts on proteins in cells (25).
We used the same cellular system to assess the effects of the
Figure 3. Inhibition of COX-1 and COX-2 by selected compounds. Ovine COX-1 (1.5 µM) and murine COX-2 (2.7 µM) were reconstituted in 100
mM Tris-HCl, pH 8.0, with 500 µM phenol, two molar equivalents of hematin, and 1 mM final concentration of the selected compounds at 4 °C
for 30 min, as described in the text. Values are means ( SD (n ) 4). *P < 0.001 vs control.

ScaVengers of γ-Ketoaldehydes
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 247
Figure 4. Effect of the analogues on the formation of LG-lysine adducts and COX-1 activity in human platelets. (A) Inhibition of the formation
of LG adducts in platelets. Human platelets were preincubated with 10 µM dazoxiben, a thromboxane synthase inhibitor, and 1 mM of the specified
compound or vehicle (control) for 30 min. After incubation, 20 µM arachidonic acid was added for 2 h, and then the proteins were precipitated and
digested to single amino acids by step-digestion with proteases. The daughter ions at m/z 332.1 and 84.1 for the lactam were monitored. Values are
means ( SD (n ) 9) *P < 0.001 vs control; +P < 0.001 vs 100 µM concentration. (B) Production of PGE2 in human platelets. COX activity was
measured by the formation of PGE2 after exposure to exogenous 20 µM arachidonic acid for 2 h. Prostanoids were analyzed by GC/NICI/MS as
previously described (16). The amount of PGE2 present in the medium is represented as a percentage of the vehicle-treated control. Values are
means ( SD (n ) 9).
ring substitutions on both the ability to inhibit the formation of
LG-adducts on proteins and the catalytic activity of COX-1 in
human platelets. We have previously demonstrated that inhibi-
tion of the thromboxane synthase with dazoxiben for 30 min
leads to a 2.1-fold increase in the levels of the LG-lysine lactam
adduct (19). We used the same method to increase the sensitivity
of our assay in these studies. Accordingly, we measured the
levels of LG-lysine lactam as the marker of protein modification
and PGE2 as the marker of COX-1 activity. As shown in Figure
4A, the formation of LG-lysine lactam was reduced in a dose-
dependent manner by all of the compounds tested with maximal
inhibition up to 80% by 1 mM SA (1), SA·AcOH (2), 3-MoSA
(6), EtSA (8), and PPM (15); 70% by 5-MoSA (5), MeSA (3),
MeSA·AcOH (4), and PM (13), and 60% by EtPM (14)
compared to the control. Surprisingly, the least potent compound
in platelets (EtPM, 14) is one of the most reactive analogues in
Vitro, suggesting that cellular efficacy is not merely predicted
by reactivity but that other parameters are involved. It is
interesting to note that the acetate salts of SA (2) and MeSA
(4) (SA·AcOH and MeSA·AcOH) had activities similar to their
free amines in platelets, whereas they had slightly decreased
reactivity in Vitro (Table 1). The acetate forms a bond between
the amine group and hydroxyl group and is easily hydrolyzable
in physiological or aqueous solution. Their physical properties
change, but their reactivity in aqueous solution is the same as
that of the free amine. This indicates that the acetate readily
dissociates in the presence of cells, releasing the free compound.
In order to exclude the possibility that the analogues reduce
LG adducts by inhibiting COX-1, we measured PGE2, the major
product of PGH2 metabolism, in dazoxiben-treated platelets.
None of the compounds decreased the level of PGE2 (Figure
4B), demonstrating that the reduced formation of protein adducts
by these compounds is due to direct scavenging of levuglandins
and not by the inhibition of cyclooxygenase activity.
Inhibition of LG-Lysine Lactam Adduct Formation in
Cells That Express COX-2. On the basis of the results with
purified COXs and human platelets, we selected the two most
potent analogues from each series, SA (1), 3-MoSA (6), PM
(13), and PPM (15), and used them to investigate their ability
to inhibit formation of LG lactam adducts in three different cell
types expressing COX-2. We used the human epithelial lung
carcinoma cell line A549 activated with IL-1ꢀ and the murine
monocyte-macrophage cell line RAW 264.7 activated with LPS-
and INF-γ as COX-2-inducible cells, and the human colon

248 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Zagol-Ikapitte et al.
Figure 5. Inhibition of LG-lysine adducts in cell lines that express COX-2. Cells were preincubated with 100 or 1000 µM of compounds or vehicle
(control) for 30 min. After incubation with 20 µM arachidonic acid for 1 h, the proteins were scraped and digested to single amino acids by
step-digestion with proteases. The daughter ion at m/z 84.1 for the lactam was monitored. (A) RAW 264.7 cells. Values are means ( SD (n e 6).
*P < 0.001 vs control. +P < 0.001 vs 100 µM concentration. (B) HCA-7 cells. Values are means ( SD (n e 6). *P < 0.001 vs control. +P < 0.01
vs 100 µM concentration. (C) A549 cells. Values are means ( SD (n e 6). *P < 0.001 vs. control. +P < 0.001 vs 100 µM concentration.
adenocarcinoma cell line HCA-7 as a cell that expresses COX-2
constitutively. We found that SA (1), 3-MoSA (6), PM (13),
and PPM (15) inhibited LG-lysine adduct formation in a dose-
dependent manner (Figure 5). Treating with compounds at a
concentration of 1 mM reduced the formation of the LG-lysine
lactam by 75 to 85% in RAW 264.7 cells, 60 to 80% in A549
cells, and 50 to 60% in HCA-7 cells, when compared to the
control. PGE2 production was analyzed as a measure of COX-2
activity in all three cells lines, and no inhibition of COX-2 by
the compounds was detected (data not shown).
hydroxyl group can interact with the oxygen atom of the
ketone moiety in LG, holding it in position, thereby facilitat-
ing the attack of the hemiaminal amine intermediate (24).
This would predict that the scavengers would become less
reactive as the hydroxyl is moved farther from the aminom-
ethyl group. We investigated the effects of the position of
the phenolic hydroxyl group on the reactivity of salicylamine
(1) by using analogues with the hydroxyl in ortho (2-HoBA,
SA 1) meta (3-HoBA), and para (4-HoBA) positions. We
found that the most reactive analogue is 2-HoBA (1) followed
by 3-HoBA and that 4-HoBA only decreased the formation
of the LG-lysine lactam adduct by 30%. Because the effect
of the hydroxyl on the reactivity of the amine should be the
same for 2- (1) and 4-HoBA, the observed difference of
reactivity suggests that the major determinant for the reactiv-
ity of the scavengers is not the electron density of the ring
but is the distance between the hydroxyl and the amine group
that favors formation of the hydrogen bond with LG.
Discussion
Evaluation of the pathophysiologic consequences of LG
adducts of cellular proteins and other molecules require an
experimental intervention that will block LG adduct formation
without altering the formation of other cyclooxygenase products.
We here investigate an approach that employs molecules that
react with LGs much faster than the amino groups found on
proteins, e.g., the ε-amine of lysine. Previously, we had found
that pyridoxamine (13) and its analogues, salicylamine (1) and
pentyl pyridoxamine (15), react rapidly with γ-ketoaldehydes
and are potent scavengers of LG in Vitro and in cells (25). On
the basis of the observations with these compounds, we now
report on the synthesis and evaluation of a series of analogues
of salicylamine and pyridoxamine. From this group, a subset
was found to inhibit LG adduct formation in cells without
blocking the biosynthesis of prostanoids and to be free of
toxicity to cells in culture.
The acetate salts of SA (2) and MeSA (4) inhibit the
formation of LG-lysine adducts to an extent similar to that of
their respective free amines in cellular systems, whereas their
reactivity in Vitro is decreased by 24% and 22%, respectively.
This observation is important with regard to cellular and in ViVo
experiments as it suggests that the water-soluble acetate salts
can be used without loss of efficacy instead of the free amine
compounds that require the use of organic solvents.
In addition to the levuglandins, the primary amine group of
these scavengers would react with other aldehydes, such as
malondialdehyde (MDA), methylglyoxal (MGO), HNE, and
4-oxo-2-nonenal (ONE). MDA and MGO target amino groups
and react rather slowly with the SA group of scavengers. In
these reactions, the advantage of SA (1) analogues seems to be
their lypophilicity (33). The primary reaction of HNE and ONE
The reactivity of the amino group of the scavengers is
influenced by two factors: (1) the presence of the phenolic
group increases the electron density in the aromatic ring. This
effect is maximal when the hydroxyl group is ortho or para
to the aminomethyl group. (2) The hydrogen atom of the

ScaVengers of γ-Ketoaldehydes
Chem. Res. Toxicol., Vol. 23, No. 1, 2010 249
is Michael addition, and Cys and His residues are much more
reactive than lysyl groups (33). Again, pyridoxamine (13) did
not exhibit particularly high reactivity toward HNE (33). ONE,
that is ∼100 times more reactive than HNE, may react with
lysyl residues in two separate pathways (34). Michael addition
products of ONE are capable of condensing with amino groups
to form pyrroles, and ONE may react with lysyl residues in
two sequential steps to form pyrrolinone. In both of these
reactions, it is possible to visualize SA (1) analogues with a
phenolic group at the 2-position being much more reactive than
lysyl residues.
Utilization of LG scavengers to evaluate the biological
consequences of LG adducts requires that the observed actions
of the compounds on cells are not due to cytotoxicity. Thus, as
a screen to eliminate compounds with obvious cytotoxicity from
further evaluation, we measured ATP depletion in HepG2 cells
as a surrogate marker with which to screen for cellular toxicity
(32, 35). Several analogues of SA (1) and of PM (13) were
found to be cytotoxic in this assay. Among the SA (1) series of
analogues, those with the most lipophilic substituents were toxic
to these cells. Interestingly, the toxic analogues were also
inhibitors of both COX isoforms in Vitro. The pharmacological
effects of the compounds that did not cause ATP depletion were
then further evaluated.
The objective was to synthesize scavengers that prevent the
formation of LG-lysine adducts without inhibiting COX activity.
Determination of whether the new compounds had effects on
COX-1 activity was carried out in platelets. Induction and
chronic up-regulation of COX-2 is associated with inflammatory
conditions and pathogenesis of many diseases. We investigated
whether the compounds could scavenge LGs produced by cells
in response to COX-2 activity. Three cancer cell lines with
different levels of COX-2 expression and activity were treated
with the analogues, and exogenous arachidonic acid was added
to achieve maximum prostanoid biosynthesis (36). Our results
indicate that several analogues inhibit the formation of COX-
1-derived LG-lysine lactam on platelet proteins without affecting
enzyme activity. As with platelets, the same analogues inhibited
the formation of LG-adducts on proteins without inhibiting the
release of PGE2 in cells expressing COX-2, although with slight
variations in efficacy: HCA-7 cells are less sensitive to
scavengers than are A549 or RAW264.7 cells.
John Oates is the Thomas F. Frist, Sr. Professor of Medicine.
This work was presented, in part, at the 9th international
Eicosanoids and Other Bioactive Lipids in Cancer, Inflammation
and Related Disease conference, Montreal, Canada, September
15-17, 2007.
References
(1) Wang, M. T., Honn, K. V., and Nie, D. (2007) Cyclooxygenases,
prostanoids, and tumor progression. Cancer Metastasis ReV. 26, 525–
534.
(2) Cha, Y. I., and DuBois, R. N. (2007) NSAIDs and cancer prevention:
targets downstream of COX-2. Annu. ReV. Med. 58, 239–252.
(3) Minghetti, L. (2004) Cyclooxygenase-2 (COX-2) in inflammatory and
degenerative brain diseases. J. Neuropathol. Exp. Neurol. 63, 901–
910.
(4) Pasinetti, G. M. (1998) Cyclooxygenase and inflammation in Alzhe-
imer’s disease: experimental approaches and clinical interventions.
J. Neurosci. Res. 54, 1–6.
(5) Harris, R. E. (2009) Cyclooxygenase-2 (cox-2) blockade in the
chemoprevention of cancers of the colon, breast, prostate, and lung.
Inflammopharmacology 17, 55–67.
(6) Harris, R. E., Beebe-Donk, J., and Alshafie, G. A. (2007) Cancer
chemoprevention by cyclooxygenase 2 (COX-2) blockade: results of
case control studies. Subcell. Biochem. 42, 193–202.
(7) Harris, R. E., Beebe-Donk, J., and Alshafie, G. A. (2007) Reduced
risk of human lung cancer by selective cyclooxygenase 2 blockade:
results of a case control study. Int. J. Biol. Sci. 3, 328–334.
(8) Agrawal, A., and Fentiman, I. S. (2008) NSAIDs and breast cancer:
a possible prevention and treatment strategy. Int. J. Clin. Pract. 62,
444–449.
(9) Roberts, R. O., Jacobson, D. J., Girman, C. J., Rhodes, T., Lieber,
M. M., and Jacobsen, S. J. (2002) A population-based study of daily
nonsteroidal anti-inflammatory drug use and prostate cancer. Mayo
Clin. Proc. 77, 219–225.
(10) Rostom, A., Dube, C., Lewin, G., Tsertsvadze, A., Barrowman, N.,
Code, C., Sampson, M., and Moher, D. (2007) Nonsteroidal anti-
inflammatory drugs and cyclooxygenase-2 inhibitors for primary
prevention of colorectal cancer: a systematic review prepared for the
U.S. Preventive Services Task Force. Ann. Intern. Med. 146, 376–
389.
(11) Stewart, W. F., Kawas, C., Corrada, M., and Metter, E. J. (1997) Risk
of Alzheimer’s disease and duration of NSAID use. Neurology 48,
626–632.
(12) in t’ Veld, B. A., Ruitenberg, A., Hofman, A., Launer, L. J., van Duijn,
C. M., Stijnen, T., Breteler, M. M. and, and Stricker, B. H. 2001)
Nonsteroidal antiinflammatory drugs and the risk of Alzheimer’s
diseaseN. Engl. J. Med. 345, 1515–1521.
(13) Salomon, R. G. (1985) Prostaglandin endoperoxide reaction mecha-
nisms and the discovery of levuglandins. Acc. Chem. Res. 18, 294–
301.
(14) Iyer, R. S., Kobierski, M. E., and Salomon, R. G. (1994) Generation
of pyrroles in the reaction of levuglandin E2 with proteins. J. Org.
Chem. 59, 6038–6043.
Conclusions
(15) Brame, C. J., Salomon, R. G., Morrow, J. D., and Roberts, L. J. (1999)
Identification of extremely reactive gamma-ketoaldehydes (isole-
vuglandins) as products of the isoprostane pathway and characterization
of their lysyl protein adducts. J. Biol. Chem. 274, 13139–13146.
(16) Boutaud, O., Brame, C. J., Salomon, R. G., Roberts, L. J., II, and and
Oates, J. A. (1999) Characterization of the lysyl adducts formed from
prostaglandin H2 via the levuglandin pathway. Biochemistry 38, 9389–
9396.
(17) Zagol-Ikapitte, I., Bernoud-Hubac, N., Amarnath, V., Roberts, L. J.,
II, Boutaud, O., and Oates, J. A. (2004) Characterization of bis(le-
vuglandinyl) urea derivatives as products of the reaction between
prostaglandin H(2) and arginine. Biochemistry 43, 5503–5510.
(18) Davies, S. S., Amarnath, V. and, and Roberts, L. J., 2nd. (2004)
Isoketals: highly reactive gamma-ketoaldehydes formed from the H2-
isoprostane pathway. Chem. Phys. Lipids 128, 85–99.
(19) Boutaud, O., Li, J., Zagol, I., Shipp, E. A., Davies, S. S., Roberts,
L. J., 2nd, and Oates, J. A. (2003) Levuglandinyl adducts of proteins
are formed via a prostaglandin H2 synthase-dependent pathway after
platelet activation. J. Biol. Chem. 278, 16926–16928.
In summary, we have identified a series of compounds that
inhibit the formation of LG-lysine adducts in cells without
interfering with the activity of either COX isoform. These
compounds do not exhibit cytotoxicity when assessed in
HepG2 cells up to millimolar concentrations. Also, we
provide more evidence in support of the role of the
o-hydroxyl group of benzylamine in the mechanism respon-
sible for the high reactivity of these molecules with γ-ke-
toaldehydes. These results provide the rationale and lead
structures for future studies examining the biological con-
sequence of the formation of levuglandin adducts on proteins
that can result from both lipid peroxidation and catalytic
activity of COXs in cells as well as in ViVo.
Acknowledgment. We thank Dr. Jim Smith for his valuable
advice. We thank Taneem Amin for expert assistance in cell
culture and Bharati Kakkad for assistance in prostanoid mea-
surement by GC/MS. This work was supported in part by the
American Health Assistance Foundation, Award Numbers
AG026119 and GM42056 from the National Institutes of Health.
(20) Boutaud, O., Andreasson, K. I., Zagol-Ikapitte, I., and Oates, J. A.
(2005) Cyclooxygenase-dependent lipid-modification of brain proteins.
Brain Pathol. 15, 139–142.
(21) Zagol-Ikapitte, I., Masterson, T. S., Amarnath, V., Montine, T. J.,
Andreasson, K. I., Boutaud, O., and Oates, J. A. (2005) Prostaglandin
H-derived adducts of proteins correlate with Alzheimer’s disease
severity. J. Neurochem. 94, 1140–1145.

250 Chem. Res. Toxicol., Vol. 23, No. 1, 2010
Zagol-Ikapitte et al.
(22) Boutaud, O., Ou, J. J., Chaurand, P., Caprioli, R. M., Montine, T. J.,
and Oates, J. A. (2002) Prostaglandin H2 (PGH2) accelerates formation
of amyloid ꢀ1-42 oligomers. J. Neurochem. 82, 1003–1006.
(23) Boutaud, O., Montine, T. J., Chang, L., Klein, W. L., and Oates, J. A.
(2006) PGH-derived levuglandin adducts increase the neurotoxicity
of amyloid beta1-42. J. Neurochem. 96, 917–923.
(24) Amarnath, V., Amarnath, K., Davies, S., and Roberts, L. J., II. (2004)
Pyridoxamine: an extremely potent scavenger of 1,4-dicarbonyls.
Chem. Res. Toxicol. 17, 410–415.
(29) Yang, D., Shih, Y., and Liu, H. (1991) Chemical synthesis of
stereospecifically labeled pyridoxamine 5′-phosphate. J. Org. Chem.
56, 2940–2946.
(30) Hofslokken, N. U., and Skattebol, L. (1999) Convenient method for
the ortho-formylation of phenols. Acta Chem. Scand. 53, 258–262.
(31) Boutaud, O., Aronoff, D. M., Richardson, J. H., Marnett, L. J., and
Oates, J. A. (2002) Determinants of the cellular specificity of
acetaminophen as an inhibitor of prostaglandin H2 synthases. Proc.
Natl. Acad. Sci. U.S.A. 99, 7130–7135.
(25) Davies, S. S., Brantley, E. J., Voziyan, P. A., Amarnath, V., Zagol-
Ikapitte, I., Boutaud, O., Hudson, B. G., Oates, J. A., and Ii, L. J.
(2006) Pyridoxamine analogues scavenge lipid-derived gamma-
ketoaldehydes and protect against H(2)O(2)-mediated cytotoxicity.
Biochemistry 45, 15756–15767.
(26) Marnett, L. J., Siedlik, P. H., Ochs, R. C., Pagels, W. R., Das, M.,
Honn, K. V., Warnock, R. H., Tainer, B. E., and Eling, T. E. (1984)
Mechanism of the stimulation of prostaglandin H synthase and
prostacyclin synthase by the antithrombotic and antimetastatic agent,
nafazatrom. Mol. Pharmacol. 26, 328–335.
(27) Marnett, L. J., Rowlinson, S. W., Goodwin, D. C., Kalgutkar, A. S.,
and Lanzo, C. A. (1999) Arachidonic acid oxygenation by COX-1
and COX-2. Mechanisms of catalysis and inhibition. J. Biol. Chem.
274, 22903–22906.
(28) Wu, Y. K., and Ahlberg, P. (1989) On the formation of isopropyl-
idenepyridoxines. A convenient method for the preparation of 2,2,8-
trimethyl-4H-1,3-dioxino[4,5-c]pyridin-5-ylmethanol from pyridoxine
hydrochloride. Acta Chem. Scand. 43, 1009–1011.
(32) Crouch, S. P., Kozlowski, R., Slater, K. J., and Fletcher, J. (1993)
The use of ATP bioluminescence as a measure of cell proliferation
and cytotoxicity. J. Immunol. Methods 160, 81–88.
(33) Doorn, J. A., and Petersen, D. R. (2002) Covalent modification of
amino acid nucleophiles by the lipid peroxidation products 4-hy-
droxy-2-nonenal and 4-oxo-2-nonenal. Chem. Res. Toxicol. 15,
1445–1450.
(34) Liu, Z., Minkler, P. E., and Sayre, L. M. (2003) Mass spectroscopic
characterization of protein modification by 4-hydroxy-2-(E)-nonenal
and 4-oxo-2-(E)-nonenal. Chem. Res. Toxicol. 16, 901–911.
(35) Cree, I. A., and Andreotti, P. E. (1997) Measurement of cytotoxicity
by ATP - based luminescence assay in primary cell cultures and cell
lines. Toxicol. In Vitro 11, 553–556.
(36) Mitchell, J. A., Larkin, S., and Williams, T. J. (1995) Cyclooxygenase-
2: regulation and relevance in inflammation. Biochem. Pharmacol. 50,
1535–1542.
TX900407A