Model studies on the modification of proteins by lipoxidation-derived 2-hydroxyaldehydes

Article abstract of DOI:10.1021/tx020095d

2-Hydroxyaldehydes have been previously identified as products of lipid peroxidation, and although they represent the simplest reducing sugars, their potential for modification of proteins under physiological conditions has not been investigated. Here, 2-hydroxyaldehydes were found to condense with amines in two ways, implicating potential pathways for modification of lysine residues. A fluorescent 4,5-dialkyl-3-hydroxypyridinium with ex/em 327/390 nm and a nonfluorescent 4-alkylimidazolium cross-linking product were isolated and characterized by ¹H NMR, ¹³C NMR, high-resolution mass spectrometry, and, in the former case, through independent synthesis. Both reactions appear to proceed through Amadori rearrangement of the initial Schiff base. On the basis of the UV absorbance of the 3-hydroxypyridinium, the latter was estimated to represent modification of 1.5% of the lysines of RNase incubated with 0.5 mM 2-hydroxyheptanal for 10 days at 25 °C. The 4-alkylimidazolium is proposed to contribute to the protein cross-linking observed by gel electrophoresis in the incubation of RNase with higher concentrations of 2-hydroxyheptanal.

Full text of DOI:10.1021/tx020095d

232
Chem. Res. Toxicol. 2003, 16, 232-241
Mod el Stu d ies on th e Mod ifica tion of P r otein s by
Lip oxid a tion -Der ived 2-Hyd r oxya ld eh yd es†
Zhongfa Liu and Lawrence M. Sayre*
Department of Chemistry, Case Western Reserve University, Cleveland, Ohio 44106
Received October 17, 2002
2-Hydroxyaldehydes have been previously identified as products of lipid peroxidation, and
although they represent the simplest reducing sugars, their potential for modification of proteins
under physiological conditions has not been investigated. Here, 2-hydroxyaldehydes were found
to condense with amines in two ways, implicating potential pathways for modification of lysine
residues. A fluorescent 4,5-dialkyl-3-hydroxypyridinium with ex/em 327/390 nm and a
nonfluorescent 4-alkylimidazolium cross-linking product were isolated and characterized by
1H NMR, 13C NMR, high-resolution mass spectrometry, and, in the former case, through
independent synthesis. Both reactions appear to proceed through Amadori rearrangement of
the initial Schiff base. On the basis of the UV absorbance of the 3-hydroxypyridinium, the
latter was estimated to represent modification of 1.5% of the lysines of RNase incubated with
0.5 mM 2-hydroxyheptanal for 10 days at 25 °C. The 4-alkylimidazolium is proposed to
contribute to the protein cross-linking observed by gel electrophoresis in the incubation of RNase
with higher concentrations of 2-hydroxyheptanal.
limited to Schiff base formation under basic conditions,
as supported by characterization of the secondary amine
product generated upon quenching with NaBH4 (18).
However, 2-hydroxyaldehydes are the simplest forms of
reducing sugars, which are known to condense with
lysine groups to give, following Amadori rearrangement,
R-ketoamines (eq 1) (19, 20). We suspected that the latter
chemistry would ensue under neutral conditions at
relatively long incubation times that could be relevant
in vivo.
In tr od u ction
A large variety of aldehydic products of lipid peroxi-
dation have been shown in recent years to be capable of
modifying nucleophilic protein side chains, most promi-
nently the ꢀ-amino group of lysines. Such modifications
are believed to play key roles in aging and pathophysi-
ological processes, such as atherogenesis (1). The early
emphasis on the easily measurable malondialdehyde was
soon augmented by an increasing interest in R,â-
unsaturated aldehydes (2, 3), most notably trans-4-
hydroxy-2-nonenal (HNE) (4-11). In addition, short
chain and long chain aliphatic 2-hydroxyaldehydes were
more recently identified by trapping methods in the
oxidation of unsaturated fatty acids, low-density lipopro-
tein (LDL), and some tissues in vitro by Spiteller (12-
15). Short and long chain 2-hydroxyaldehydes were
proposed to arise from two different sources: the former
from decomposition of bis-dihydroperoxides of unsatur-
ated fatty acids (14) and the latter from hydrolysis of the
epoxides of the enol ether plasmalogens formed upon
enzyme-mediated epoxidation at the expense of mono-
hydroperoxides of unsaturated fatty acids such as linoleic
acid (15).
The 2-hydroxyaldehyde function also occurs naturally
in the maturation of the connective tissue protein col-
lagen. Lysyl oxidase-mediated deamination of D-hydroxy-
lysine side chains formed by the action of lysine hydrox-
ylase results in “hydroxylysine aldehyde” moieties 1 that
condense with other δ-hydroxylysine or lysine residues,
ultimately forming the trifunctional pyridinium cross-
link structures pyridinoline (5a ) or deoxypyridinoline
(5b), respectively (Scheme 1) (21, 22). These stable and
fluorescent moieties (λmax 325 nm at neutral pH) are
widely used urinary markers to assess turnover of
connective tissue in diseases such as osteoporosis and
bone cancer (23, 24). The biosynthesis of these pyridinium
species is described to be initiated by condensation/
Amadori rearrangement of hydroxylysine aldehyde (1)
with the ꢀ-amino groups of δ-hydroxylysine or lysine.
Studies on the total syntheses of pyridinoline or deoxy-
pyridinoline (25, 26), which can be generated in high yield
from bis-R-ketoamines 2 independently prepared from
the corresponding R-bromoketones, support the biosyn-
thetic route shown in Scheme 1, involving an initial
The most abundant 2-hydroxyaldehyde found in tissue
is 2-hydroxyheptanal, which has been shown to arise
from the ω-6 polyunsaturated arachidonic and linoleic
acids (16). 2-Hydroxybutanal has also been found to arise
from ω-3 polyunsaturated fatty acids (17). The reaction
of 2-hydroxyheptanal with amines was reported to be
† Presented in preliminary form: Liu, Z., Xu, G., and Sayre, L. M.
(2001) Abstracts of Papers, 222nd National Meeting of the American
Chemical Society, Chicago, IL, August 26-30, 2001, American Chemi-
cal Society, Washington, DC (ORGN 25). Sayre, L. M. (2001) Abstracts
of Papers, 222nd National Meeting of the American Chemical Society,
Chicago, IL, August 26-30, 2001, American Chemical Society, Wash-
ington, DC (TOXI 17).
* To whom correspondence should be addressed. Tel: (216)368-3704.
Fax: (216)368-3006. E-mail: lms3@po.cwru.edu.
10.1021/tx020095d CCC: $25.00 © 2003 American Chemical Society
Published on Web 01/18/2003

Lipoxidation-Derived 2-Hydroxyaldehydes
Chem. Res. Toxicol., Vol. 16, No. 2, 2003 233
Sch em e 1
570 power supply (Fisher Scientific, Pittsburgh, PA). The water
that was used for all studies was purified with a Millipore
system. UV-vis spectra were obtained using a Perkin-Elmer
Lambda 20 spectrophotometer. The fluorescent spectra were
obtained using a Varian Cary Eclipse fluorescence spectrometer.
Sodium phosphate buffers in D2O were prepared from the
corresponding pH-adjusted sodium phosphate buffer in H2O by
complete removal of H2O and reconstitution of the residue to
the same volume with D2O, repeated two more times.
Diethyl acrolein acetal, 2-methoxyethylamine, 70% ethyl-
amine solution, 6-aminocaproic acid, and NR-Cbz-Lys were
purchased from Aldrich Chemical Co. Ribonuclease A (RNase
A) was purchased from Sigma Chemical Company (St. Louis,
MO). All other reagents and chemicals were AR or ACS grade.
Unless stated otherwise, all operations were conducted at room
temperature and all concentrations were conducted at reduced
pressure.
double condensation/Amadori process. In the synthetic
studies, base-catalyzed cyclization of 2 typically results
in pyridiniums 5 without detection of dihydropyridone
intermediate 3, apparently because of the ease of autoxi-
dation of the latter under the basic conditions. However,
in a model reaction wherein 1-phenethylamine was
alkylated by 2 equiv of 1-bromopentan-2-one, Allevi and
co-workers showed that base treatment of the resulting
bis-R-ketoamine in the absence of oxygen permitted the
corresponding 3-like dihydropyridone to be isolated and
completely characterized (25). The latter result indicated
that conversion of 3 to 5 represented strictly an O2-
dependent autoxidation (presumably of the enol-enamine
tautomer 4) rather than a redox disproportionation
process.
The present paper deals with an exploration of the
reaction of simple 2-hydroxyaldehydes with amines as
lysine surrogates. The finding in preliminary studies that
the incubation of amines with 2-hydroxyaldehydes gave
rise to a fluorescent product with ex/em 327/390 nm led
to its characterization as a 4,5-dialkyl-3-hydroxypyri-
dinium, apparently duplicating the chemistry of pyridi-
noline biosynthesis. In addition, as it was observed that
treatment of proteins with 2-hydroxyheptanal led to
protein cross-linking in addition to fluorophore genera-
tion, model studies were directed at determining a cross-
link structure, leading to identification of a 4-alkylimi-
dazolium.
2-Hyd r oxybu ta n a l Dieth yl Aceta l (26). Acrolein diethyl
acetal (11.44 g, 88 mmol), benzonitrile (9.2 mL), and 30%
aqueous H2O2 (10.56 g) were added sequentially to a suspension
of KHCO3 (1.5 g, 15 mmol) in methanol (50 mL). The mixture
was stirred for 2 h and then heated to 40-45 °C for 4 h.
Methanol was removed, and the residue was dissolved in water
(50 mL). The mixture was filtered, and the filtrate was extracted
with chloroform (3 × 50 mL). The organic layers were combined
and evaporated, and hexane (100 mL) was added to the residue
to precipitate benzamide. Benzamide was filtered off, and
hexane was evaporated. The residue was subjected to silica gel
chromatography eluted with ethyl acetate-hexane (1:4, v/v) to
yield 2-(diethoxymethyl)oxirane (27) in 60% yield. A solution of
the latter (1.46 g, 10 mmol) and cuprous bromide (40 mg) in
THF (20 mL) was cooled with ice and charged with argon, to
which a solution of methylmagnesium bromide in THF (4 mL,
2.8 M) was added in 30 min. The reaction mixture was then
heated to 60-70 °C for 17 h and then cooled, and water (100
mL) was added. The water layer was separated and extracted
with ether (3 × 50 mL). The organic layers were combined and
evaporated, and the crude product was subjected to silica gel
chromatography (eluant hexanes-ethyl acetate 1:1, v/v) to
afford 2-hydroxybutanal diethyl acetal in a yield of 92.6%.
2-Hyd r oxyh ep ta n a l Dieth yl Aceta l. A solution of 2-(di-
ethoxymethyl)oxirane (1.46 g, 10 mmol) and cuprous bromide
(40 mg) in THF (20 mL) was cooled with ice and charged with
argon, to which a solution of n-butylmagnesium chloride in THF
(5.6 mL, 2.0 M) was added dropwise in 30 min. The reaction
mixture was then heated to 60-70 °C for 17 h and worked up
as above to yield the crude product, which was subjected to
chromatography (eluant hexanes-ethyl acetate 3:1, v/v) to
Exp er im en ta l P r oced u r es
Gen er a l Meth od s. 1H NMR (300 or 200 MHz) and 13C NMR
(75.1 or 50.1 MHz) spectra were recorded on Varian Gemini 300
or 200 instruments. In all cases, the solvent peak served as the
internal standard for reporting chemical shifts, which are
expressed as parts per million downfield from TMS (δ scale).
In the 13C NMR line listings, attached proton test (APT)
designations are given as (+) or (-) following the chemical shift.
High-resolution mass spectra (HRMS) were obtained at 20 eV
on a Kratos MS-25A instrument. TLC was performed on glass
plates precoated with silica gel 60F254. Compounds on the
developed plate were visualized by short wavelength UV light
(λ ) 254 nm) by placing the plate in a chamber filled with iodine
vapor or by spraying with ammonium phosphomolybdic acid
solution or ninhydrin solution. Preparative column chromatog-
raphy was performed using 32-63 µm silica gel. Electrophoresis
studies utilized an SE 200 series minigel caster (Hoefer Scien-
tific Instruments, San Francisco, CA) and a Fisher Biotech FB
1
afford 2-hydroxyheptanal diethyl acetal in a yield of 85.3%. H
NMR (CDCl3): δ 0.89 (t, 3H, J ) 5.86 Hz), 1.22 (t, 3H, J ) 7.0
Hz), 1.25 (t, 3H, J ) 8.1 Hz), 1.30-1.42 (6H), 1.52-1.56 (2H),

234 Chem. Res. Toxicol., Vol. 16, No. 2, 2003
Liu and Sayre
2.16 (br, 1H), 3.54-3.62 (3H), 3.69-3.79 (2H), 4.26 (dd, 1H, J
) 6.00, 0.80 Hz). 13C NMR (CDCl3): δ 14.12 (-), 15.48 (-), 15.54
(-), 22.69 (+), 25.29 (+), 31.80 (+), 31.99 (+), 63.46 (+), 63.55
(+), 71.84 (-), 105.25 (-). EI HRMS m/z calcd for C9H19O2 (M-
OEt)+, 159.1385; found, 159.1380.
(318 mg, 2.5 mmol) in methylene chloride (4 mL) at -78 °C
under nitrogen (31). After the mixture was stirred for 30 min,
a solution of 11 (262 mg, 1 mmol) in methylene chloride (3 mL)
was added via a syringe at -78 °C. The mixture was stirred for
3.5 h and allowed to warm to -60 °C slowly. Triethylamine (1.2
mL, 10 mmol) was added at -60 °C, and the mixture was stirred
for 15 min. The reaction was quenched with water (2 mL), and
the aqueous layer was separated and extracted with methylene
chloride (2 × 3 mL). The combined organic layer was washed
with brine (2 mL), dried (anhydrous Na2SO4), and evaporated,
and the residue was subjected to silica gel chromatography
(eluant hexanes-ethyl acetate 7:3, v/v) to afford 12 (233 mg) in
83.3% yield. 1H NMR (CDCl3): δ 1.04 (t, 3H, J ) 7.32 Hz), 1.06
(t, 3H, J ) 7.38 Hz), 1.39 (s, 9H), 2.34-2.48 (4H), 3.98 (s, 2H),
4.10 (s, 2H). 13C NMR (CDCl3): δ 7.40 (+), 7.57 (+), 28.19 (+),
32.85 (-), 33.05 (-), 56.63 (-), 57.10 (-), 80.94 (-), 155.29 (-),
206.92 (-), 207.22 (-). EI HRMS m/z calcd for C13H23NO4 M+,
257.16270; found, 257.16341.
Gen er a l P r oced u r e for Isola tion of 3-Hyd r oxyp yr i-
d in iu m fr om Lon g-Ter m In cu ba tion of 2-Hyd r oxybu ta n a l
w ith Sim p le Am in es. Dowex 50W-X8 H+ (400 mg) was added
to a solution of 2-hydroxybutanal diethyl acetal (1.20 g, 7.4
mmol) in phosphate buffer (200 mL, 1.5 M, pH 7.4). The mixture
was stirred at room temperature for 6 h, and then, 30 mmol of
either 2-methoxyethylamine or ethylamine was added, and the
mixture was stirred at room temperature for 2 weeks. The
reaction mixture was extracted with ethyl acetate, and the
aqueous layer was evaporated. The residue was subjected to
silica gel chromatography eluted with methanol to give the
corresponding 3-oxidopyridinium zwitterionic “inner salts”.
1,5-Dieth yl-4-m eth yl-3-oxid op yr id in iu m (6). 1H NMR
(methanol-d4): δ 1.23 (t, 3H, J ) 7.36 Hz), 1.56 (t, 3H, J ) 7.32
Hz), 2.27 (s, 3H), 2.70 (q, 2H, J ) 7.52 Hz), 4.30 (q, 2H, J )
7.34 Hz), 7.53 (s, 1H), 7.62 (s, 1H). 13C NMR (methanol-d4): δ
12.48 (-), 14.52 (-), 17.09 (-), 25.24 (+), 56.72 (+), 127.23 (-),
129.76 (-), 143.00 (+), 145.23 (+), 167.54 (+). FAB HRMS m/z
calcd for C10H16NO (M + H)+, 166.1232; found, 166.1232.
5-E t h yl-1-(2-m et h oxyet h yl)-4-m et h yl-3-oxid op yr id in i-
u m (20). 1H NMR (D2O): δ 1.13 (t, 3H, J ) 7.44 Hz), 2.19 (s,
3H), 2.62 (q, 2H, J ) 7.44 Hz), 3.31 (s, 3H), 3.87 (t, 2H, J )
4.26 Hz), 4.39 (t, 2H, J ) 4.26 Hz), 7.49 (s, 1H), 7.55 (s, 1H).
13C NMR (D2O): δ 11.90 (-), 13.14 (-), 23.81 (+), 58.40 (-),
59.47 (+), 70.69 (+), 128.87 (-), 129.43 (-), 142.40 (+), 145.18
(+), 162.78 (+). EI HRMS m/z calcd for C11H17NO2 M+, 195.1259;
found, 195.1259.
N,N-Bis(2-h yd r oxyb u t yl)b en zyla m in e (9) (29). Benzyl-
amine (2.72 mL, 25.0 mmol) was added to 1,2-epoxybutane (3.6
g, 50 mmol) under nitrogen, and the reaction mixture was
heated to 100 °C in a sealed tube for 24 h. The mixture was
cooled and purified by silica gel column chromatography (eluant
hexanes-ethyl acetate 3:2, v/v) to afford 9 in 85% yield.
N,N-Bis(2-h yd r oxybu tyl)a m in e (10). To a solution of 9
(259 mg, 1 mmol) in 4.4% methanolic formic acid solution (26
mL) was added 10% Pd/C (26 mg) (30). The mixture was stirred
at room temperature for 5 h and filtered through a pad of Celite
powder, and the pad was washed with methanol. The solvent
was evaporated, and the residue was used for the next step
directly. 1H NMR (CDCl3): δ 0.97 (t, 6H, J ) 7.49 Hz), 1.47-
1.57 (4H), 2.89-3.16 (4H), 3.92(m, 2H), 8.48 (br, 3H).
5-Eth yl-4-m eth yl-3-h yd r oxyp yr id in e (13). According to a
published method (26), a suspension of potassium t-butoxide (96
mg, 0.8 mmol) in THF (1.5 mL) was cooled in an ice bath under
nitrogen, and a solution of 12 (51.4 mg, 0.2 mmol) in THF (0.6
mL) was added. After 30 min, the bath was removed, and the
reaction mixture was stirred at room temperature for 24 h and
then recooled (ice bath) and diluted with methanol (3 mL). The
solvent was evaporated, and the residue was dissolved in ethyl
acetate (5 mL) and water (2 mL). The aqueous layer was
separated and extracted with ethyl acetate (2 × 3 mL). The
combined organic layer was washed with brine (2 mL), dried
(anhydrous Na2SO4), and evaporated, and the residue was
subjected to silica gel column chromatography (eluant methanol-
ethyl acetate 5:95, v/v) to afford 13 in 73.0% yield. 1H NMR
(CDCl3): δ 1.22 (t, 3H, J ) 7.7 Hz), 2.27 (s, 3H), 2.66 (q, 2H, J
) 7.7 Hz), 7.90 (s, 1H), 8.13 (s, 1H). 13C NMR (CDCl3): δ 11.08
(-), 14.51 (-), 23.93 (-), 132.89 (-), 134.07 (+), 138.18 (+),
138.57 (-), 152.84 (+). EI HRMS m/z calcd for C8H11NO M+,
137.0841; found, 137.0847.
P r ep a r a tion of 1,5-Dieth yl-4-m eth yl-3-oxid op yr id in iu m
(6) fr om 13. Compound 13 (68.5 mg, 0.5 mmol) was added to a
solution of ethyl iodide (41.5 mg, 0.25 mmol) dissolved in
anhydrous 1,4-dioxane (2.5 mL) under nitrogen. The mixture
was heated to reflux for 10 h, the solvent was evaporated, and
the crude product was subjected to silica gel column chroma-
tography eluted with methanol to afford 6, in 85.6% yield, as
the inner salt as verified by the lack of reaction with AgNO3
expected if the product was the 3-hydroxypyridinium iodide.
ter t-Bu tyl N,N-Bis(2-h yd r oxybu tyl)ca r ba m a te (11). The
crude amine 10 (159 mg, 1 mmol) was dissolved in THF (2 mL)
and di-tert-butyl dicarbonate (545 mg, 2.5 mmol) was added at
room temperature under nitrogen. The mixture was stirred at
room temperature for 4 h, the solvent was evaporated, and the
crude product was dissolved in ethyl acetate (5 mL) and water
(2 mL). The aqueous layer was separated and extracted with
ethyl acetate (2 × 3 mL). The combined organic layer was
washed with brine (2 mL), dried (anhydrous Na2SO4), and
concentrated. The crude product was subjected to silica gel
column chromatography (eluant hexanes-ethyl acetate 3:2, v/v)
to afford 240 mg of 11 in 90% yield as a mixture of RR/SS and
RS/SR diastereomers. Fast moving fraction (11a ): 1H NMR
(CDCl3): δ 0.97 (t, 6H, J ) 7.56 Hz), 1.37-1.51 (4H), 1.46 (s,
9H), 2.83 (br, 2H), 3.38 (br, 2H, exchangeable), 3.62 (app dd,
2H, J ) 2.1, 14.4 Hz), 3.86 (br, 2H). 13C NMR (CDCl3): δ 9.82
(-), 28.03 (+), 28.49 (-), 57.14 (+), 72.9 (br, -), 80.27 (+), 164.5
(-). Slow moving fraction (11b): 1H NMR (CDCl3): δ 0.97 (t,
6H, J ) 7.46 Hz), 1.37-1.51 (m, 4H), 1.46 (s, 9H), 2.92 (s, 2H,
exchangeable), 3.28 (br, 2H), 3.79 (br m, 2H). 13C NMR
(CDCl3): δ 9.99 (-), 27.97 (+), 28.47 (-), 55.36 (br, +), 72.13
(-), 80.57 (+), 167.91 (-). EI HRMS m/z calcd for C13H28NO4
M+, 262.2018; found, 262.2032.
Gen er a l P r oced u r e for Isola tion of 3-Hyd r oxyp yr id in i-
u m s fr om Lon g-Ter m In cu ba tion of 2-Hyd r oxybu ta n a l
w ith Nr-Cbz-Lys a n d 6-Am in oca p r oic Acid . Dowex 50W-
X8 H+ (400 mg) was added to a solution of 2-hydroxybutanal
diethyl acetal (1200 mg) in phosphate buffer (200 mL, 0.1 M,
pH 7.4). The mixture was stirred at room temperature for 6 h
and then decanted into a solution of either 6-aminocaproic acid
or N-Cbz-Lys (4 equiv) in 1.5 M pH 7.4 sodium phosphate buffer
(600 mL), and the mixture was stirred at room temperature for
2 weeks. The reaction mixture was extracted with ethyl acetate,
and the aqueous layer was evaporated. The residue was
subjected to silica gel chromatography eluted with methanol to
give the corresponding 3-hydroxypyridinium carboxylates in
5-10% yield.
1-(5-Ben zyloxycar bon ylam in o-5-car boxypen tyl)-5-eth yl-
4-m eth yl-3-h yd r oxyp yr id in iu m (14a ). 1H NMR (methanol-
d4): δ 1.20 (t, 3H, J ) 7.65 Hz), 1.39 (m, 2H), 1.71 (m, 1H),
1.83-1.92 (3H), 2.24 (s, 3H), 2.66 (q, 2H, J ) 7.62 Hz), 4.01 (t,
1H, J ) 3.78 Hz), 4.18 (t, 2H, J ) 6.45 Hz), 5.05 (s, 2H), 7.28-
7.33 (5H), 7.45 (s, 1H), 7.53 (s, 1H). 13C NMR (methanol-d4): δ
13.35 (-), 15.41 (-), 24.32 (+), 26.07 (+), 32.90 (+), 34.36 (+),
58.12 (-), 62.13 (+), 68.26 (+), 128.223 (-), 129.70 (-), 129.83
(-), 130.34 (-), 130.93 (+), 139.27 (+), 143.58 (+), 146.07 (+),
159.01 (+), 168.61 (+), 170.06 (+). FAB HRMS m/z calcd for
ter t-Bu tyl N,N-Bis(2-oxobu tyl)ca r ba m a te (12). Dimethyl
sulfoxide (0.35 mL, 5 mmol) in methylene chloride (2.5 mL) was
added slowly via syringe to a cooled solution of oxalyl chloride
C
22H29N2O5 (M + H)+, 401.2076; found, 401.2059.

Lipoxidation-Derived 2-Hydroxyaldehydes
Chem. Res. Toxicol., Vol. 16, No. 2, 2003 235
1-(5-Ca r b oxyp en t yl)-5-et h yl-4-m et h yl-3-h yd r oxyp yr i-
d in iu m (14b). 1H NMR (D2O): δ 1.10 (t, 3H, J ) 7.65 Hz), 1.24
(m, 2H), 1.50 (m, 2H), 1.85 (m, 2H), 2.11 (t, 2H, J ) 7.29 Hz),
2.17 (s, 3H), 2.61 (q, 2H, J ) 7.65 Hz), 4.23 (t, 2H, J ) 7.18
Hz), 7.67 (s, 1H), 7.73 (s, 1H). 13C NMR (D2O): δ 11.50 (-), 13.27
(-), 23.77 (+) 25.18 (+), 25.30 (+), 30.35 (+), 37.12 (+), 60.43
(+), 128.47 (-), 130.22 (-), 142.66 (+), 144.22 (+), 161.01 (+),
183.07 (+). FAB HRMS m/z calcd for C14H22NO3 (M + H)+,
252.1600; found, 252.1602.
0.87-0.98 (6H), 1.28-1.67 (14H), 1.91 (m, 2H), 2.16 (t, 2H, J )
7.47 Hz), 2.63-2.73 (4H), 4.21 (t, 2H, J ) 7.35 Hz), 7.46 (s, 1H),
7.56 (s, 1H). 13C NMR (methanol-d4, two aliphatic carbon signals
overlapping): δ 14.38 (-), 23.53 (+), 24.35 (+), 26.93 (+), 27.27
(+), 27.40 (+), 31.46 (+), 31.71 (+), 32.22 (+), 32.84 (+), 38.66
(+), 61.34 (+), 128.20 (-), 130.53 (-), 141.12 (+), 149.67 (+),
167.42 (+), 182.30 (+). FAB HRMS m/z calcd for C20H34NO3 (M
+ H)+, 336.2539; found, 336.2529.
1,3-Bis-(6-ca r b oxyp en t yl)-4-p en t ylim id a zoliu m
(23).
1H NMR Sp ectr oscop ic Mon itor in g of th e Rea ction of
2-Hyd r oxybu t a n a l w it h E xcess 2-Met h oxyet h yla m in e.
Dowex 50W-X8 H+ (10 mg) was added to a solution of 2-hy-
droxybutanal diethyl acetal (32.4 mg, 0.2 mmol) in phosphate
buffer in D2O (0.5 mL, 0.1 M, pH 7.4). 1H NMR spectroscopy
monitoring indicated that the reaction was complete in 6 h.
Yield: 90 mg, 10.6%. 1H NMR (methanol-d4): δ 0.94 (t, 3H, J
) 7.08 Hz), 1.32-1.42 (8H), 1.44-1.67 (6H), 1.70-1.91 (4H),
2.17 (t, 4H, J ) 7.20 Hz), 2.67 (t, 2H, J ) 7.35 Hz), 4.13 and
4.15 (2t, 2H each, J ) 7.5 Hz), 7.42 (s, 1H), 8.93 (s, 1H). 13C
NMR (methanol-d4): δ 14.33 (-), 23.42 (+), 24.37 (+), 26.67 (+),
26.75 (+), 27.10 (+), 27.20 (+), 28.18 (+), 30.53 (+), 30.77 (+),
32.41 (+), 120.13 (-), 137.47 (+), 175.23 (+). 13C NMR (D2O,
one aliphatic carbon signal overlapping): δ 13.23 (-), 21.63 (+),
22.71 (+), 24.84 (+), 24.95 (+), 25.16 (+), 25.30 (+), 26.31 (+),
28.76 (+), 28.89 (+), 30.35 (+), 36.65 (+), 46.42 (+), 49.16 (+),
118.60 (+), 136.16 (-), 182.58 (2C, +). FAB HRMS m/z calcd
for C20H35N2O4 M+, 367.2597; found, 367.2589.
1
Then, 2-methoxyethylamine (0.8 mmol) was added, and the H
NMR spectrum of the mixture was taken after 2 h.
1H NMR Sp ectr oscop ic Mon itor in g of th e Rea ction of
2-H yd r oxyb u t a n a l w it h 2-Met h oxyet h yla m in e u n d er
Bu ffer Con d ition s. Dowex 50W-X8 H+ (2.7 mg) was added to
a solution of 2-hydroxybutanal diethyl acetal (0.05 mmol, 8.1
mg) in phosphate buffer in D2O (100 µL), and the mixture was
kept at room temperature for 6 h. The resulting solution (100
µL) was decanted into a solution of 2-methoxyethylamine (0.2
mmol) in 1.5 M pH 7.4 sodium phosphate in D2O (400 µL) in an
NMR tube. The progress of the reaction was monitored by 1H
NMR spectroscopy at 0, 2, 5, 10, 24, and 48 h.
In cu ba tion of RNa se w ith 2-Hyd r oxyh ep ta n a l. To 2.50
mL aliquots of RNase A (7.5 mM based on Lys) in 0.1 M pH 7.4
phosphate buffer were added 2.50 mL aliquots of freshly
prepared 2-hydroxybutanal (15, 7.5, 3.75, 1.88, 0.94, 0.47, 0.24,
and 0 mM) in a 3:1 mixture of 0.1 M pH 7.4 phosphate buffer
and acetonitrile, and the UV-vis spectrum of each incubation
was taken from 200 to 600 nm at 1, 2, 3, 4, 8, 10, and 14 days.
Aliquots of the reaction mixtures (20 µL) at 2 and 4 days were
mixed with 20 µL of denaturing buffer for electrophoresis (0.3
M pH 6.8 Tris, 5% SDS, 5% 2-mercaptoethanol, 50% glycerol,
and 0.02% bromophenol blue, pH 6.8) and heated in boiling
water for 5 min. Then, 3 µL samples were analyzed by SDS-
PAGE using 5% stacking and 14% resolving gels. Protein bands
were stained by Gelcode Blue stain reagent (Pierce, Rockford,
IL).
Tr in itr oben zen esu lfon ic Acid (TNBS) Assa y for P r otein
Lys Mod ifica tion (32). Solutions of TNBS (30 mM) and 0.1 M
Na2B2O7‚10H2O in 0.1 M NaOH were prepared in doubly
distilled water. The protein solution (100 µL) was added to 2.90
mL of the borate buffer, and then, 75 µL of the TNBS solution
was added. The mixture was allowed to stand at room temper-
ature for 30 min. The absorbance was measured at 420 nm. The
values were used to estimate the percent of reactive lysyl groups
modified relative to the native protein.
Red u ctive Qu en ch in g (Na BH4) of th e Rea ction of 2-Hy-
d r oxybu ta n a l w ith 2-Meth oxyeth yla m in e. Dowex 50W-X8
H+ (40 mg) was added to a solution of 2-hydroxybutanal diethyl
acetal (120 mg) in phosphate buffer (20 mL, 0.1 M, pH 7.4). The
mixture was stirred at room temperature for 6 h and then
decanted into a solution of 2-methoxyethylamine (4 equiv) in
1.5 M pH 7.4 phosphate buffer (60 mL), and the mixture was
stirred at room temperature for 10 h. Then, NaBH4 (40 mg) was
added to the mixture, and the reaction mixture was stirred at
room temperature for 24 h and adjusted to pH 12 with 1 N
NaOH. The solution was extracted with ethyl acetate (6 × 50
mL), and the combined organic layer was evaporated to give a
residue that was subjected to silica gel chromatography (eluant
hexanes-ethyl acetate 1:3, v/v) to give 21 and 22.
N-(2-Meth oxyeth yl)-2-h yd r oxybu tyla m in e (21). 1H NMR
(CDCl3): δ 0.93 (t, 3H, J ) 7.53 Hz), 1.42 (m, 2H), 2.43 (dd, 1H,
J ) 12.03, 9.33 Hz), 2.69 (dd, 1H, J ) 12.03, 3.03 Hz), 2.77 (m,
2H), 3.33 (s, 3H), 3.44-3.52 (3H). 13C NMR (CDCl3): 10.01 (-),
27.94 (+), 49.00 (+), 54.93 (+), 58.81 (-), 70.85 (-), 71.92 (+).
EI HRMS m/z calcd for C7H17NO2 M+, 147.1259; found, 147.1255.
N-(2-Met h oxyet h yl)-N-(2-h yd r oxyb u t yl)-2-h yd r oxyb u -
Resu lts a n d Discu ssion
1
Syn th esis of 2-Hyd r oxya ld eh yd es. 2-Hydroxybuta-
nal and 2-hydroxyheptanal were prepared from acrolein
diethyl acetal according to a modification of literature
methods (26, 27). In brief, acrolein diethyl acetal was
converted to its epoxide with hydrogen peroxide under
basic conditions in the presence of benzonitrile. Opening
of the epoxide with methylmagnesium bromide or n-
butylmagnesium chloride in the presence of a catalytic
amount of cuprous bromide yielded the corresponding
2-hydroxyalkanal diethyl acetals. The free 2-hydroxy-
alkanals were generated in situ by hydrolysis of the
diethylacetals with Dowex 50W-X8 H+ in pH 7.4 phos-
phate buffer (for 2-hydroxybutanal) or in 95% aqueous
acetonitrile (for 2-hydroxyheptanal).
F lu or escen t Ad va n ced En d P r od u ct, 3-Oxid op y-
r id in iu m , fr om th e In cu ba tion of 2-Hyd r oxya lk a -
n a ls w it h P r im a r y Am in es. Exposure of 2-hydroxy-
butanal to primary amines in aqueous buffer over the
course of several days led to production of a fluorescent
product with a λmax near 325 nm. Using the UV absor-
bance as an indicator of fluorophore generation, it was
found that the use of higher concentrations and an excess
tyla m in e (22). H NMR (CDCl3): δ 0.93 (t, 6H, J ) 7.30 Hz),
1.38 and 1.39 (2q, 2H each), 2.40 (dd, 2H, J ) 13.32, 9.84 Hz),
2.62 (dd, 2H, J ) 13.32, 3.30 Hz), 2.79 (t, 2H, J ) 5.12 Hz),
3.35 (s, 3H), 3.41 (t, 2H, J ) 5.30 Hz), 3.32-3.52 (4H). 13C NMR
(CDCl3): δ 10.03 (-), 27.47 (+), 56.40 (+), 58.87 (-), 63.48 (+),
71.30 (-), 71.42 (+). EI HRMS m/z calcd for C11H25NO3 M+,
219.1834; found, 219.1815.
Rea ction of 2-Hyd r oxyh ep ta n a l w ith 6-Am in oca p r oic
Acid . Dowex 50W-X8 H+ (330 mg) was added to a solution of
2-hydroxyheptanal diethyl acetal (5 mmol, 1.002 g) in 95%
aqueous acetonitrile (30 mL), and the mixture was sonicated
for 3.5 h. The solution was diluted with acetonitrile (50 mL)
and decanted to a solution of 6-aminocaproic acid (10 mmol, 1.31
g) in 0.1 M pH 7.4 phosphate buffer (300 mL). The mixture was
stirred at room temperature for 2 weeks. Acetonitrile was
evaporated, and the aqueous solution was extracted with ethyl
acetate (3 × 50 mL). The aqueous layer was evaporated under
vacuum, and the residue was dissolved in methanol (100 mL).
The mixture was filtered, the filtrate was evaporated, and the
residue was subjected to silica gel chromatography (eluant
methanol-ethyl acetate 2:1, v/v) to give 15 and 23 as their
carboxylate inner salts.
4-Bu t yl-1-(6-ca r b oxylp en t yl)-5-p en t yl-3-h yd r oxyp yr i-
d in iu m (15). Yield: 39 mg, 4.7%. 1H NMR (methanol-d4): δ

236 Chem. Res. Toxicol., Vol. 16, No. 2, 2003
Liu and Sayre
F igu r e 2. Observed NOEs in compound 6.
to the methylene signal at 4.3 ppm. The betaine structure
was also supported by comparing the aryl hydrogen and
aryl carbon NMR chemical shifts in DMSO-d6 and D2O
with those reported for 1-methyl-3-oxidopyridinium (33,
34).
The core 3-oxidopyridinium structure was unambigu-
ously confirmed by independent synthesis of 6, conducted
according to a strategy used for the total synthesis of
pyridinoline and model compounds (26), as shown in
Scheme 2. Thus, 1,2-epoxybutane (8) was reacted with
F igu r e 1. Fluorescence spectrum of compound A in H2O (647
mM).
of amine over aldehyde resulted in optimal apparent
yields of the fluorophore. To simplify the structural
characterization of what was presumed to be a single
major chromophoric product, the incubation of 2-hydroxy-
butanal (40 mM) with 2-methoxyethylamine (160 mM)
was carried out in pH 7.4 phosphate buffer for 2 weeks.
The choice of amine was based on the advantage of the
Sch em e 2
1
methoxy singlet in the H NMR spectrum in assessing
the stoichiometry of the reaction by integration. Workup
of the reaction yielded a fluorescent species A with ex/
em 327/390 nm (Figure 1) as the main isolable product
(silica gel column chromatography and preparative TLC)
that was formed in a 50% yield based on starting
aldehyde, as revealed by integration of the crude 1H NMR
1
spectrum. The H NMR spectrum (D2O) of the purified
product showed that it had one methoxy group from
2-methoxyethylamine, an ethyl, and a methyl group
implicating two molecules of 2-hydroxybutanal and two
aryl singlets at 7.6 and 7.7 ppm. The 13C NMR APT
spectrum showed that five of the total of 11 carbons are
in an aromatic ring; two bonded to hydrogen resonate at
relatively high field and the other three resonate at
relatively low field. Electron ionization (EI) HRMS
indicated a M+ peak at m/z 195.1259, consistent with one
molecule of 2-methoxyethylamine and two molecules of
2-hydroxybutanal with the molecular formula C11H17NO2.
Because the sum formula for two 2-hydroxybutanals and
one 2-methoxyethylamine would predict C11H25NO5, the
stoichiometry requires three dehydrations and one two
electron oxidation. A similar product was isolated from
the incubation of 2-hydroxybutanal with ethylamine.
benzylamine to afford N,N-bis(2-hydroxybutyl)benzyl-
amine (9) (29). Hydrogenolytic N-debenzylation (30) to
give 10 and treatment with di-tert-butyl dicarbonate
afforded two diastereomers (meso and d.l) of the Boc
derivative 11, which were individually characterized but
recombined and converted to azadiketone 12 by Swern
oxidation (31). Treatment of azadiketone 12 with potas-
sium tert-butoxide gave 5-ethyl-4-methyl-3-hydroxypy-
ridine (13), which was alkylated with ethyl iodide to
afford target molecule 6, found to be identical to that
isolated from the incubation of 2-hydroxybutanal and
ethylamine.
To confirm that the 3-hydroxypyridinium would also
form from the reaction of the ꢀ-amino group of Lys with
2-hydroxybutanal, the incubation of NR-Cbz-Lys (50 mM)
with 2-hydroxybutanal (12.5 mM) was carried out in 0.1
M pH 7.4 phosphate buffer for 2 weeks. The 1H NMR
spectrum of the reaction product in D2O after several
column chromatographic and preparative TLC separa-
tions exhibited two characteristic peaks at about 7.4 and
7.5 ppm. Also, the 13C NMR APT spectrum supports the
3-hydroxypyridinium core structure. FAB HRMS showed
the exact mass is 401.2059, which is consistent with the
quasi-ion mass of 1-(5-benzyloxycarbonylamino-5-car-
boxypentyl)-5-ethyl-4-methyl-3-hydroxypyridinium (14a ).
This model study suggests that the 3-hydroxypyridinium
motif could represent a stable oxidative stress protein
modification generated by the action of 2-hydroxyalde-
hydes on lysyl residues. Thus, to prepare 3-hydroxypy-
ridinium haptens, from the most predominant and next
predominant lipoxidation-derived 2-hydroxyaldehydes,
that could be conveniently conjugated to a carrier protein
for production of antibodies, the reaction was repeated
using 2-hydroxyheptanal or 2-hydroxybutanal with 6-ami-
nocaproic acid as a tether, leading to isolation of 14b and
Taking the ethylamine-derived product as an example,
two possible hydroxypyridine-like structures 6 and 7
were considered presuming starting material-related
connectivity. Structure 6 is shown as the 3-oxidopyri-
dinium inner salt (betaine) (33, 34) that would be the
expected neutral form eluting from a chromatography
column. The structure was assigned as 6 rather than 7
based on the two-dimensional nuclear Overhauser effect
(NOE) spectrum. All NOEs observed for the aryl singlets
were as indicated in Figure 2: the distinguishing findings
were that (i) no aryl singlet showed an NOE to the C4
methyl singlet and (ii) both aryl singlets exhibited NOEs

Lipoxidation-Derived 2-Hydroxyaldehydes
Chem. Res. Toxicol., Vol. 16, No. 2, 2003 237
15. In the cases of 14 and 15, the presence of the carboxyl
moiety led to the neutral compounds being isolated as
3-hydroxypyrinidinium carboxylates rather than as the
3-oxido species.
starting aldehyde diastereotopic C3 multiplets at 1.39
and 1.62 ppm (CH3CH2CH(OH)CH(OH)2) (Figure 3A)
and growth of a quartet at 2.4 ppm (Figure 3B), a
transformation that was nearly complete after 10 h
(Figure 3C). The quartet at 2.4 ppm is consistent with
R-aminoketone 17 (CH3CH2C(dO)CH2N) formed via the
Amadori reaction (eq 2). These results suggested that
whereas the simple Schiff base is stable under basic
conditions, tautomerization (Amadori rearrangement)
proceeds under pH 7.4 buffer conditions. This explains
the difference between our results and those obtained by
Spiteller (18).
Mech a n ism of th e Rea ction of 2-Hyd r oxya lk a n a ls
w ith Am in es. To investigate the mechanism of pyri-
dinium formation, which contrasts the simple Schiff base
formation observed by Spiteller (18), the reaction of
2-hydroxybutanal with 2-methoxyethylamine was moni-
tored by 1H NMR spectroscopy. First, generation of
2-hydroxybutanal from the diethyl acetal precursor was
monitored in buffered D2O by observing the release of
1
ethanol. The aldehydic group did not appear in the H
NMR spectrum, indicating that 2-hydroxybutanal exists
as its hydrate form in this medium. With addition of
2-methoxyethylamine (4 equiv) to the solution, which
Further spectral monitoring over the period of 10-24
h revealed the appearance of a singlet at 1.7 ppm and a
quartet at 2.2 ppm, the intensities of which were seen to
grow with a corresponding decrease in intensity of the
presumed R-aminoketone peak at 2.4 ppm (Figure 3D).
The two newly formed peaks, which are distinct from the
singlet at 2.0 ppm and quartet at 2.6 ppm seen in the
final pyridinium product 20 (previously isolated com-
pound A), are consistent with structure 19 proposed as
1
exceeded the buffer and made the reaction basic, the H
NMR spectrum after 2 h showed a doublet at 7.65 ppm
and no resonance from 1.8 to 2.6 ppm, suggesting the
formation of the Schiff base 16 (CH3CH2CH(OH)CHd
N-) (eq 2). In contrast, when the pH 7.4 buffer concen-
tration was increased to exceed that of the added amine,
1
the H NMR spectrum showed slow disappearance of the
1
F igu r e 3. Progress of the reaction of 2-hydroxybutanal (SM) (0.1 mmol) with 2-methoxyethylamine (0.4 mmol) as monitored by H
NMR in D2O (pH 7.4 sodium phosphate buffer): (A) 5 min progress, (B) 2 h progress, (C) 10 h progress, and (D) 24 h progress.

238 Chem. Res. Toxicol., Vol. 16, No. 2, 2003
Liu and Sayre
the 3-oxidopyridinium precursor (25). Although the pos-
sible intermediacy of the azadiketone 18 could not be
discerned distinctly from 17 by 1H NMR, an identical
reaction mixture to that monitored by NMR was quenched
with NaBH4 after 10 h to permit isolation of the more
stable alcohols potentially reflective of the coexistence of
17 and 18. After work up, two major products were
obtained and characterized as the N-(2-hydroxybutyl) 21
and N,N-bis(2-hydroxybutyl) 22 derivatives of 2-meth-
oxyethylamine (eq 3). Once isolated, compounds 21 and
22 could be determined to form in a ratio of 4:3 from the
crude 1H NMR spectrum. The finding of 21 and 22 is
consistent with the sequence of reaction 17 f 18 f 19
f 20 (eq 2).
F igu r e 4. Fluorescence spectrum of RNase A (10 mg/mL)
treated with 7.5 mM 2-hydroxyheptanal in 0.1 M pH 7.4
phosphate buffer for 10 days.
3-hydroxypyridinium in the sample of RNase (10 mg/mL,
7.5 mM of Lys) treated with 7.5 mM 2-hydroxyheptanal
for 10 days could be estimated to involve 3.8% of the
lysines in RNase A. Because the TNBS assay (32) on the
same sample indicated that 75% of the lysines of the
protein was modified, it is clear that only a small fraction
of the lysines modified by 2-hydroxyheptanal were con-
verted to the 3-hydroxypyridinium. This is not surprising
in view of the fact that formation of the pyridinium
requires the initial 1:1 R-aminoketone Amadori product
to react with a second molecule of 2-hydroxyheptanal
before it is converted (e.g., oxidized) to another species.
Although it constitutes a minor fraction of Lys modifica-
tion, the pyridinium adduct represents a convenient and
stable marker for protein modification by 2-hydroxyhep-
tanal. As far as the nature of the adducts responsible for
the main loss of TNBS reactivity of lysines, it is probable
that the initial R-aminoketone mono-Amadori product (eq
4, P-NH2 ) protein Lys ꢀ-amine) is the major contribu-
tor, although this would have to be confirmed by further
studies, e.g., by analysis of protein hydrolysate following
reductive stabilization.
On the basis of all of the above information, we propose
that generation of 3-oxidopyridiniums in the reaction of
primary amines with 2-hydroxyaldehydes follows the
same general pathway described in Scheme 1 for the
biosynthesis of pyridinoline. The initial stages of reaction
follow well-precedented Amadori chemistry of reducing
sugars, and following the known cyclization to dihydro-
pyridone (25), its enol-enamine tautomer is readily
autoxidized to the pyridinium. On the basis of incorpora-
tion of two molecules of aldehyde per molecule of amine,
it is not readily apparent why the yield of pyridinium
was found to be optimal using excess amine over alde-
hyde. One possibility is that use of excess amine com-
pletely and efficiently ties up the aldehyde in the form
of the Schiff base-Amadori equilibrium mixture 16/17
and that the protonated Schiff base 16H+ is a more
reactive aldehyde donor in reacting with 17 (to give 18)
than is the free aldehyde.
Detection of a F lu or escen t Ad d u ct in th e In cu ba -
tion of RNa se w ith 2-Hyd r oxya ld eh yd es. Whereas
a fluorescent product (ex/em 327/390 nm) was observed
in the model reaction of 2-hydroxyaldehydes with pri-
mary amines, fluorescence (ex/em 325/409 nm) was also
seen when the 13.7 kDa protein RNase A was incubated
with 7.5 mM 2-hydroxyheptanal as shown in Figure 4.
The similarity of the absorbance-emission spectrum is
consistent with the 3-hydroxypyridinium adduct identi-
fied above in the model studies. We found no evidence
for an alternative amine-derived fluorescent product in
the above model reactions, and other studies (results not
shown) failed to provide evidence for a fluorophore (or
any other stable adduct) when other potentially reactive
protein side chain groups (4-alkylimidazole for His and
N-alkylguanidine for Arg) were incubated with 2-hy-
droxyheptanal.
Figure 5 shows that the 3-hydroxypyridinium began
to form after only 2 days of incubation and reached a
plateau on the eighth day. Incubations of various con-
centrations of 2-hydroxyheptanal with RNase A for 10
days demonstrated that the formation of the fluorophore
was dependent on the concentration of 2-hydroxyhepta-
nal (Figure 6). Even a concentration of 2-hydroxyheptanal
as low as 0.5 mM resulted in pyridinylation of 1.5% of
RNase lysines. At even lower concentrations of aldehyde
likely to be present under physiological conditions, the
3-hydroxypyridinium adduct would be expected to form
only after a longer period of time.
Isola tion a n d Ch a r a cter iza tion of a P oten tia l
2-Hyd r oxya ld eh yd e-In d u ced P r otein Cr oss-Lin k -
in g Moiety. In the incubation of RNase with 2-hydroxy-
heptanal, in addition to fluorophore generation, gel
electrophoretic analysis revealed that 2-hydroxyheptanal
also induces protein cross-linking (Figure 7), although
the pyridinium adduct would not explain this. Notwith-
Assuming a protein-bound 3-hydroxypyridinium that
has a similar extinction coefficient at 325 nm as does the
model pyridinium compound 15 at 327 nm, the level of

Lipoxidation-Derived 2-Hydroxyaldehydes
Chem. Res. Toxicol., Vol. 16, No. 2, 2003 239
gested that the compound had an imidazolium structure,
where it is known that C2 undergoes equilibrium depro-
tonation to give a ylide (35, 36). Furthermore, because
of the H(D) exchange at C2, the carbon resonance was
not observed in the 13C NMR spectrum. Nonetheless, the
FAB HRMS indicated the exact mass was 367.2589,
consistent with the molecular formula C20H35N2O4
(367.2597) for the 4-pentylimidazolium compound 23.
F igu r e 5. Time course of formation of 3-hydroxypyridinium
in the incubation of RNase (10 mg/mL) with 2-hydroxyheptanal
(7.5 mM) in a 3:1 (v/v) mixture of 0.1 M pH 7.4 phosphate buffer
and acetonitrile.
The generation of 23 from 2-hydroxyheptanal requires
one intact aldehyde and a second aldehyde molecule that
serves as the donor of the lone C2 carbon. A possible
mechanism starting from the previously described enol-
enamine intermediate is shown in Scheme 3. The enol-
Sch em e 3
F igu r e 6. Concentration dependence of 3-hydroxypyridinium
formation in 2-hydroxyheptanal-treated RNase for 10 days.
enamine species 24 can suffer autoxidation in the reac-
tion mixture to produce the 2-oxoaldehyde Schiff base 25,
which can react with another molecule of amine to form
the same R-diimine intermediate 26 that appears to be
involved in the formation of imidazolium structures in
the incubation of 2-oxoaldehydes with amines (36). The
species 26 may then react with another molecule of
2-hydroxyaldehyde to give an adduct 27 that decomposes
with loss of chain-shortened aldehyde to the species 28
that readily dehydrates to 23. According to this mecha-
nism, the second molecule of 2-hydroxyheptanal that
donates its C1 would end up as hexanal, a well-known
aldehyde generated in lipid peroxidation.
Although generation of the imidazolium in our reac-
tions here requires 2 mol of Lys side chain and 2 mol of
2-hydroxyaldehyde, the same 2:2 stoichiometry is re-
quired to generate the corresponding imidazoliums from
glyoxal and methylglyoxal, termed GOLD and MOLD,
respectively, which have been observed in physiological
samples (37). Thus, imidazolium formation from 2-hy-
droxyaldehydes under physiological situations should not
be deterred on the basis of a seemingly unlikely stoichi-
ometry.
F igu r e 7. Modification of RNase with 2-hydroxyheptanal.
Solutions of RNase A (10 mg/mL, control RNase A in lane 1)
containing 0.117, 0.235, 0.469, 0.938, 1.875, 3.75, 7.5, 15, and
30 mM 2-hydroxyheptanal (lanes 2-10), respectively, were
incubated at room temperature for 4 days.
standing, in addition to the pyridinium products isolated
1
in the incubations with 2-hydroxybutanal, the crude H
NMR spectra indicated the presence of a minor nonfluo-
rescent product. Under the conditions of the incubation
of 6-aminocaproic acid with 2-hydroxyheptanal (mixed
phosphate buffer and CH3CN), this alternate product was
actually the major product accompanying the generation
1
of pyridinium 15. The H NMR spectrum of the purified
Con clu sion
compound in CD3OD exhibited two singlets at 7.42 and
8.93 ppm in a ratio of 1:1. The hydrogen resonating at
8.93 underwent exchange over time. This result sug-
2-Hydroxyaldehydes are important products of lipid
oxidation, but their potential for modification of nucleo-

240 Chem. Res. Toxicol., Vol. 16, No. 2, 2003
Liu and Sayre
(8) Itakura, K., Osawa, T., and Uchida, K. (1998) Structure of a
fluorescent compound formed from 4-hydroxy-2-nonenal and NR-
hippuryllysine: a model for fluorophores derived from protein
modifications by lipid peroxidation. J . Org. Chem. 63, 185-187.
(9) Xu, G., and Sayre, L. M. (1998) Structural characterization of a
4-hydroxy-2-alkenal-derived fluorophore that contributes to li-
poperoxidation-dependent protein cross-linking in aging and
degenerative disease. Chem. Res. Toxicol. 11, 247-251.
(10) Tsai, L., Szweda, P. A., Vinogradova, O., and Szweda, L. I. (1998)
Structural characterization and immunochemical detection of a
fluorophore derived from 4-hydroxy-2-nonenol and lysine. Proc.
Natl. Acad. Sci. U.S.A. 95, 7975-7980.
(11) Xu, G., Liu, Y., and Sayre, L. M. (1999) Independent synthesis,
solution behavior, and studies on the mechanism of formation of
a primary amine-derived fluorophore representing cross-linking
of proteins by (E)-4-hydroxy-2-nonenal (HNE). J . Org. Chem. 64,
5732-5745.
(12) Loidl-Stahlhofen, A., Hannemann, K., and Spiteller, G. (1994)
Generation of a-hydroxyaldehydic compounds in the course of lipid
peroxidation. Biochim. Biophys. Acta 1213, 140-148.
(13) J ira, W., and Spiteller, G. (1996) Plasmalogens and their oxidative
degradation products in low and high density lipoprotein. Chem.
Phys. Lipids 79, 95-100.
(14) Loidl-Stahlhofen, A., Hannemann, K., and Spiteller, G. (1995)
Detection of short-chain a-hydroxyaldehydic compounds as pen-
tafluorbenzyloxime derivatives in bovine liver. Chem. Phys. Lipids
77, 113-119.
philic side chains in proteins under aerobic physiomi-
metic conditions has not been explored. 2-Hydroxyalde-
hydes were found to be capable of condensing with
amines in two ways, implicating potential pathways for
modification of Lys residues. A fluorescent 4,5-dialkyl-
3-hydroxypyridinium with ex/em 327/390 nm and a
nonfluorescent 4-alkylimidazolium cross-linking product
were isolated and characterized by NMR and HRMS and,
in the former case, through independent synthesis.
Mechanisms have been proposed for formation of these
adducts, both of which depend on a two electron autoxi-
dation step. The 4-alkylimidazolium is proposed to
contribute to the protein cross-linking observed by gel
electrophoresis in the incubation of RNase with 2-hy-
droxyheptanal. In this same incubation, the extent of
generation of the fluorescent 3-hydroxypyridinium was
estimated based on the UV absorbance. The results of
this study suggest that the 3-hydroxypyridinium is a
convenient marker for the modification of proteins by
2-hydroxyaldehydes. It is of interest to consider the
possible biological significance of this pyridinium relative
to the several pyridines that form enzymatically and
nonenzymatically in the connective tissue proteins col-
lagen and elastin (21-26, 38, 39).
(15) Dudda, A., Spiteller, G., and Kobelt, F. (1996) Lipid oxidation
products in ischemic porcine heart tissue. Chem. Phys. Lipids 82,
39-51.
(16) Mlakar, A., and Spiteller, G. (1996) Previously unknown aldehydic
lipid peroxidation compounds of arachidonic acid. Chem. Phys.
Lipids 79, 47-53.
(17) Mlakar, A., and Spiteller, G. (1994) Reinvestigation of lipid
peroxidation of linolenic acid. Biochim. Biophys. Acta 1214, 140-
148.
(18) Kern, W., and Spiteller, G. (1996) Synthesis and properties of
natural occurring a-hydroxyaldehydes. Tetrahedron 52, 4347-
4362.
(19) Baynes, J . W., Watkins, N. G., Fisher, C. I., Hull, C. J ., Patrick,
J . S., Ahmed, M. U., Dunn, J . A., and Thorpe, S. R. (1989) The
Amadori product on protein: structure and reactions. Prog. Clin.
Biol. Res. 304, 43-67.
(20) Acharya, A. S., Sussman, L. G., and Manning, J . M. (1983) Schiff
base adducts of glyceraldehyde with hemoglobin. Differences in
the Amadori rearrangement at the R-amino groups. J . Biol. Chem.
258, 2296-2302.
(21) Fujimoto, D., Moriguchi, T., Ishida, T., and Hayashi, H. (1978)
The structure of pyridinoline, a collagen cross link. Biochem.
Biophys. Res. Commun. 84, 52-57.
(22) Ogawa, T., Ono, T., Tsuda, M., and Kawanishi, Y. (1982) A novel
fluor in insoluble collagen: a cross linking moiety in collagen
muscle. Biochem. Biophys. Res. Commun. 107, 1252-1257.
(23) Coleman, R. E. (2002) The clinical use of bone resorption markers
in patients with malignant bone disease. Cancer 94, 2521-2533.
(24) Christenson, R. H. (1997) Biochemical markers of bone metabo-
lism: an overview. Clin. Biochem. 30, 573-593.
The preparative conditions chosen here to conveniently
access end products involved nonphysiological, high
concentrations of aldehyde; nonetheless, the identified
structures would be expected to form physiologically in
small amounts over time. Because 2-hydroxyheptanal is
the predominant 2-hydroxyalkanal found in lipid oxida-
tion, a 6-aminocaproic acid-based hapten for the 4-butyl-
5-pentyl-3-hydroxypyridinium was prepared for the pur-
poses of conjugating to keyhole limpet hemocyanin, to
permit the production of antibodies recognizing this
epitope. Studies on the use of these antibodies to confirm
the generation of the 3-hydroxypyridinium in proteins
exposed to oxidized lipid preparations and in in vitro
oxidized LDL will be reported separately.
Ack n ow led gm en t. We thank the National Institutes
of Health for support of this work through Grant HL
53315.
Su p p or tin g In for m a tion Ava ila ble: 1H and 13C NMR
spectra of new compounds. This material is available free of
charge via the Internet at http://pubs.acs.org.
(25) Allevi, P., Longo, A., and Anastasia, M. (1999) Total synthesis of
deoxypyridinoline, a biochemical marker of collagen turnover.
Chem. Commun., 559-560.
Refer en ces
(26) Adamczyk, M., J ohnson, D. D., and Reddy, R. E. (1999) Collagen
cross-links: synthesis of pyridinoline, deoxypyridinoline and their
analogues. Tetrahedron 55, 63-68.
(1) Baynes, J . W., and Thorpe, S. R. (2000) Glycoxidation and
lipoxidation in atherogenesis. Free Radical Biol. Med. 28, 1708-
1716.
(27) Kobori, Y., Myles, D. C., and Whitesides, G. M. (1992) Substrate
specificity and carbohydrate synthesis using transketolase. J . Org.
Chem. 57, 5899-5907.
(28) Durrwachter, J . R., Drueckhammer, D. G., Nozaki, K., Sweers,
H. M., and Wong, C. H. (1986) Enzymic aldol condensation/
isomerization as a route to unusual sugar derivatives. J . Am.
Chem. Soc. 108, 7812-7818.
(29) Auge, J ., and Leroy, F. (1996) Lithium trifluoromethanesulfonate-
catalyzed aminolysis of oxiranes. Tetrahedron Lett. 37, 7715-
7716.
(30) ElAmin, B., Anantharamaiah, G. M., Royer, G. P., and Means,
G. E. (1979) Removal of benzyl-type protecting groups from
peptides by catalytic transfer hydrogenation with formic acid. J .
Org. Chem. 44, 3442-3444.
(2) Esterbauer, H., Schaur, R. J ., and Zollner, H. (1991) Chemistry
and biochemistry of 4-hydroxynonenal, malondialdehyde and
related aldehydes. Free Radical Biol. Med. 11, 81-128.
(3) Witz, G. (1989) Biological interactions of R,â-unsaturated alde-
hydes. Free Radical Biol. Med. 7, 333-349.
(4) Uchida, K., and Stadtman, E. R. (1993) Covalent attachment of
4-hydroxynonenal to glyceraldehyde-3-phosphate dehydrogenase.
J . Biol. Chem. 268, 6388-6393.
(5) Friguet, B., Stadtman, E. R., and Szweda, L. I. (1994) Modification
of glucose-6-phosphate-dehydrogenase by 4-hydroxy-2-nonenal.
Formation of cross-linked protein that inhibits the multicatalytic
protease. J . Biol. Chem. 269, 21639-21643.
(6) Nadkarni, D. V., and Sayre, L. M. (1995) Structural definition of
early lysine and histidine adduction chemistry of 4-hydroxynon-
enal. Chem. Res. Toxicol. 8, 284-291.
(31) Mancuso, A. J ., Huang, S.-L., and Swern, D. (1978) Oxidation of
long-chain and related alcohols to carbonyls by dimethyl sulfoxide
“activated” by oxalyl chloride. J . Org. Chem. 43, 2480-2482.
(32) Bubnis, W. A., and Ofner, C. M., III (1992) The determination of
ꢀ-amino groups in soluble and poorly soluble proteinaceous
(7) Sayre, L. M., Sha, W., Xu, G., Kaur, K., Nadkarni, D., Subbana-
gounder, G., and Salomon, R. G. (1996) Immunochemical evidence
supporting 2-pentylpyrrole formation on proteins exposed to
4-hydroxy-2-nonenal. Chem. Res. Toxicol. 9, 1194-1201.

Lipoxidation-Derived 2-Hydroxyaldehydes
Chem. Res. Toxicol., Vol. 16, No. 2, 2003 241
materials by a spectrophotometric method using trinitrobenzene-
sulfonic acid. Anal. Biochem. 207, 129-133.
(33) Takeuchi, Y., and Dennis, N. (1976) Carbon-13 NMR spectra of
pyridinium betaines and related compounds. Org. Magn. Reson.
8, 21-27.
(37) Frye, E. B., Degenhardt, T. P., Thorpe, S. R., and Baynes, J . W.
(1998) Role of the Maillard reaction in aging of tissue proteins.
Advanced glycation end product-dependent increase in imidazo-
lium cross-links in human lens proteins. J . Biol. Chem. 273,
18714-18719.
(38) Chen, J . R., Takahashi, M., Kushida, K., Suzuki, M., Suzuki, K.,
Horiuchi, K., and Nagano, A. (2000) Direct detection of cross links
of collagen and elastin in the hydrolysates of human yellow
ligament using single-column high performance liquid chroma-
tography. Anal. Biochem. 278, 99-105.
(39) Umeda, H., Takeuchi, M., and Suyama, K. (2001) Two new elastin
cross-links having pyridine skeleton. Implication of ammonia in
elastin cross-linking in vivo. J . Biol. Chem. 276, 12579-12587.
(34) Katritzky, A. R., and Takeuchi, Y. (1971) 1,3-Dipolar character
of six-membered aromatic rings. Part I. 1-Methyl-3-oxidopyri-
dinium. J . Chem. Soc. C, 874-877.
(35) Grimmett, M. R. (1980) Advances in imidazole chemistry. Adv.
Heterocycl. Chem. 27, 241-326.
(36) Wells-Knecht, K. J ., Brinkmann, E., and Baynes, J . W. (1995)
Characterization of an imidazolium salt formed from glyoxal and
NR-hippuryllysine: a model for Maillard reaction cross links in
proteins J . Org. Chem. 60, 6246-6247.
TX020095D