Reactions of genistein with alkylperoxyl radicals

Article abstract of DOI:10.1021/tx000015a

Antioxidant actions of the soy isoflavone genistein are believed to contribute to its overall chemopreventive activity. However, the mechanisms of its antioxidant reactions remain unknown. The objective of this study was to characterize the reaction products of genistein (5,7,4'- trihydroxyisoflavone) with peroxyl radicals generated by thermolysis of 2,2'- azobis(2,4-dimethylvaleronitrile) (AMVN). Genistein oxidations with AMVN- derived peroxyl radicals yielded orobol (5,7,3',4'-tetrahydroxyisoflavone), a hydroxylated derivative of genistein, and several stable adducts of 4'- oxogenistein with AMVN-derived radicals. Some of these adducts include novel structures resulting from secondary oxidations of the AMVN-derived moiety. For all the observed oxidation products, the modifications occurred on the B- ring of the molecule. Genistein oxidation product structures provide potentially useful markers of genistein antioxidant chemistry.

Full text of DOI:10.1021/tx000015a

638
Chem. Res. Toxicol. 2000, 13, 638-645
Rea ction s of Gen istein w ith Alk ylp er oxyl Ra d ica ls
Arti Arora,† Susanne Valcic,† Santiago Cornejo,† Muralee G. Nair,‡
Barbara N. Timmermann,† and Daniel C. Liebler*,†
Department of Pharmacology and Toxicology and Southwest Environmental Health Sciences Center,
College of Pharmacy, The University of Arizona, Tucson, Arizona 85721, and Center of Food Safety
and Toxicology, Michigan State University, East Lansing, Michigan 48824
Received J anuary 31, 2000
Antioxidant actions of the soy isoflavone genistein are believed to contribute to its overall
chemopreventive activity. However, the mechanisms of its antioxidant reactions remain
unknown. The objective of this study was to characterize the reaction products of genistein
(5,7,4′-trihydroxyisoflavone) with peroxyl radicals generated by thermolysis of 2,2′-azobis(2,4-
dimethylvaleronitrile) (AMVN). Genistein oxidations with AMVN-derived peroxyl radicals
yielded orobol (5,7,3′,4′-tetrahydroxyisoflavone), a hydroxylated derivative of genistein, and
several stable adducts of 4′-oxogenistein with AMVN-derived radicals. Some of these adducts
include novel structures resulting from secondary oxidations of the AMVN-derived moiety.
For all the observed oxidation products, the modifications occurred on the B-ring of the molecule.
Genistein oxidation product structures provide potentially useful markers of genistein
antioxidant chemistry.
Sch em e 1. F or m a tion of th e Gen istein P h en oxyl
Ra d ica l by On e-Electr on Oxid a tion
The potential role of soy products in cancer prevention
has received much attention in recent years. Epidemio-
logical data indicate that consumption of soybean-
containing diets is associated with a lower incidence of
certain human cancers in Asian than in Caucasian
populations (1, 2). Genistein is one of the two principal
isoflavones found in soy (3) and, as such, has been the
primary focus in studying the role of soy isoflavones in
cancer prevention. Numerous biological activities have
been reported for genistein, of which the protective effects
against cancer are the most notable (4-7). Genistein is
thought to act as an anticancer agent in large part
through its ability to scavenge oxidants involved in
carcinogenesis. It has been reported to inhibit tumor
promoter-induced formation of hydrogen peroxide in vivo
and in vitro in mouse skin (8). Other studies have
demonstrated the antioxidant activity of genistein in a
coupled linoleic acid/â-carotene system (9), in a liposomal
system against UV or peroxyl radical-induced peroxida-
tion (10), and against microsomal lipid peroxidation
induced by the Fe(II)/ADP/NADPH system (11).
Although published data describe structure-activity
relationships for genistein antioxidant activity, the oxi-
dative fate of genistein in mechanistic terms remains
poorly understood. Like other phenolic antioxidants,
genistein would primarily act as a chain-breaking anti-
oxidant by scavenging peroxyl radicals, thereby sup-
pressing radical chain autoxidations. Genistein has been
proposed to react with peroxyl radicals by a single
electron transfer followed by deprotonation (Scheme 1);
however, the specific products formed during these
antioxidant reactions have not been characterized.
The objective of this investigation was to identify
reaction products formed by oxidation of genistein by
peroxyl radicals. In the study presented here, we have
succeeded in isolating and characterizing reaction prod-
ucts of genistein with alkylperoxyl radicals from AMVN1
in a homogeneous acetonitrile system. The use of a
selective generator of peroxyl radicals permitted selective
study of the major radical-scavenging reaction of phenolic
antioxidants. The oxidation products formed are orobol,
a hydroxylated derivative of genistein, and a group of
adducts of 4′-oxogenistein with AMVN-derived radicals.
Genistein (Scheme 1) fulfills many of the structural
requirements considered essential for effective radical
scavenging by flavonoids and isoflavonoids. It has a C-2,3
double bond in conjugation with a 4-oxo function in the
C-ring, which together can participate in electron delo-
calization from the B-ring (12). Additionally, the positions
of its phenolic hydroxyl groups favor a high antioxidant
activity. Previous structure-activity studies suggest that
hydroxyl groups at positions C-5 and C-7 on the A-ring
(13) and C-4′ of the B-ring (14, 15) contribute to inhibition
of lipid peroxidation.
1 Abbreviations: AMVN, 2,2′-azobis(2,4-dimethylvaleronitrile); APCI-
MS, atmospheric-pressure chemical ionization mass spectrometry;
BSTFA, N,O-bis(trimethylsilyl)trifluoroacetamide; ESI-MS, electro-
spray ionization mass spectrometry; FAB-MS, fast atom bombardment
mass spectrometry; GC/MS, gas chromatography/mass spectrometry;
HMBC, heteronuclear multiple-bond correlation; HSQC, heteronuclear
single-quantum coherence; LC/MS, liquid chromatography/mass spec-
trometry; MPLC, medium-pressure liquid chromatography; MS-MS,
tandem mass spectrometry; NOESY, nuclear Overhauser effect spec-
troscopy; TMCS, trimethylchlorosilane; TMS, trimethylsilyl.
* To whom correspondence should be addressed: Department of
Pharmacology and Toxicology, College of Pharmacy, The University
of Arizona, Tucson, AZ 85721. E-mail: Liebler@pharmacy.arizona.edu.
Phone: (520) 626-4488. Fax: (520) 626-2466.
† The University of Arizona.
‡ Michigan State University.
10.1021/tx000015a CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/20/2000

Antioxidant Chemistry of Genistein
Chem. Res. Toxicol., Vol. 13, No. 7, 2000 639
the end of 3 h by chilling the reaction mixture on ice. Excess
AMVN was removed from the reaction mixture by extractions
with hexane. The residue was evaporated in vacuo and stored
at -80 °C until further analyses could be carried out.
Isola tion of Rea ction P r od u cts. The reaction mixture was
first subjected to reverse-phase MPLC using a 0.1% aqueous
CF3COOH/CH3CN mixture (60:40, v/v) as the mobile phase at
a rate of 15 mL/min. This was changed to a 0.1% aqueous CF3-
COOH/CH3CN mixture (30:70, v/v) at 18 min and the run
terminated at 30 min. MPLC runs yielded three crude fractions
that were further separated using a HPLC system. Analytical-
scale HPLC was conducted using gradient elution at a rate of 1
mL/min and detection at 260 nm. Initially, the mobile phase
consisted of 100% solvent A (CH3CN/H2O/CF3COOH, 28:72:0.1,
v/v/v). At 15 min, solvent B (0.1% CF3COOH in CH3CN) was
introduced in a linear gradient to 40% by 25 min. From 40 to
45 min, the system was returned to 100% solvent A in a linear
gradient.
Gen istein Oxidation s in Aceton itr ile an d H218O. Genistein
(1 µmol) and AMVN (10 µmol) were dissolved in 5 mL of an
acetonitrile/18O-labeled water mixture (95:5, v/v) and heated for
3 h in a screw-cap test tube at 50 °C. The reaction mixture was
extracted with hexane, and the acetonitrile fractions were
evaporated in vacuo and then redissolved in the HPLC mobile
phase. Products were purified by reversed-phase HPLC as
described above. Orobol samples were collected and analyzed
by positive APCI-MS. Control oxidations were conducted in
acetonitrile and unlabeled water (95:5, v/v).
These product structures indicate reactivity of the
genistein B-ring in reactions with peroxyl radicals.
Exp er im en ta l P r oced u r es
Ch em ica ls. N,O-(Bistrimethylsilyl)trifluoroacetamide (BST-
FA) and trimethylchlorosilane (TMCS) were purchased from
Pierce Chemical Co. (Rockford, IL). H218O was from Isotec, Inc.
(Miamisburg, OH). Acetone-d6 and CD3CN were obtained from
Cambridge Isotopes (Woburn, MA). All other reagents were of
the highest purity available.
Genistein (1) was synthesized as yellowish needle-like crys-
tals as described previously (16): UV (CH3CN/H2O) λmax 200,
260 nm; GC/MS [M - 15]+ m/z 471; positive APCI-MS [M +
H]+ m/z 271; negative ESI-MS [M - H]- m/z 269; positive high-
resolution FAB-MS for C15H11O5 [M + H]+ calcd 271.0607, found
271.0608.
AMVN was obtained from Polysciences, Inc. (Warrington,
PA): 1H NMR (CD3CN, 500 MHz) δ 1.95 (m, CH2), 1.83 (m,
CH2), 1.65 (m, CH), 1.50 (s, CH3), 0.87 (d, J ) 6.5 Hz, CH3),
0.77 (d, J ) 6.5 Hz, CH3); 13C NMR (CD3CN, 125 MHz) δ 120.12
(CtN), 72.84 (C), 47.04 (CH2), 25.79 (CH), 25.19 (CH3), 23.80
(CH3), 23.62 (CH3).
Liqu id Ch r om a togr a p h y. HPLC analyses were conducted
using a Hewlett-Packard model 1040M diode-array detector
coupled to a Hewlett-Packard 1050 four-channel gradient pump
(Hewlett-Packard, Palo Alto, CA) and an Allsphere ODS-2
HPLC column (5 µm, 4.6 mm × 250 mm, Alltech Associates,
Deerfield, IL). System control and data analyses were performed
using Hewlett-Packard Chemstation Software. The medium-
pressure liquid chromatography (MPLC) system that was used
was equipped with a Bu¨chi 688 pump, a D-Star Instruments
DFV-20 fixed-wavelength detector, a Linseis L200E recorder,
and a Bu¨chi borosilicate 3.3 column (15 mm × 460 mm) filled
with Polygoprep 100-30 C18 (25-40 µm, Machery & Nagel,
Du¨ren, Germany).
Ma ss Sp ectr om etr y. LC/MS was performed on a Finnigan
TSQ-7000 triple-quadrupole mass spectrometer using either an
atmospheric-pressure chemical ionization (APCI) or electrospray
ionization (ESI) source (Finnigan MAT, San J ose, CA). Spectra
obtained with high-resolution fast atom bombardment mass
spectrometry (FAB-MS) were recorded with a J EOL HX 110
instrument equipped with a FAB (Xe) ionization gun. GC/MS
spectra were recorded on a Fisons-VG MD-800 analyzer coupled
to a Carlo Erba 8000 gas chromatograph fitted with a Fisons
A200S autosampler and an on-column injector (Fisons Instru-
ments, Beverly, MA). Samples for GC/MS were derivatized with
200 µL of a BSTFA/TMCS mixture (10:1, v/v) at 60 °C for 60
min. TMS-derivatized samples were introduced into the GC/
MS apparatus by on-column injection at 140 °C and separated
on a 30 m × 0.25 mm DB-5MS column (J & W Scientific, Folsom,
CA). One minute after injection, the column temperature was
increased at a rate of 20 °C/min to 320 °C and held at that
temperature for 20 min. Compound detection was conducted in
the electron ionization mode at 70 eV. Fisons-VG Lab-Base
system software was used for system control and data process-
ing.
NMR Sp ectr oscop y. 1H, 13C, and 1H-13C heteronuclear
single-quantum coherence (HSQC), heteronuclear multiple-bond
correlation (HMBC), and nuclear Overhauser effect (NOESY)
spectra were acquired using either a Bruker AMX-500 or a
Bruker DRX-500 (500 MHz for 1H and 125 MHz for 13C)
instrument at 25 °C. Samples were dissolved in 0.6 mL of
deuterated acetone or acetonitrile. All spectra were referenced
to residual undeuterated acetone or acetonitrile.
Oxid a tion of Gen istein w ith AMVN. Genistein was oxi-
dized with a 10-fold molar excess of AMVN in acetonitrile.
Typical oxidation mixtures contained 1 µmol of genistein and
10 µmol of AMVN in 7 mL of acetonitrile. Large-scale oxidations
were carried out using 150 µmol of genistein and 1.5 mmol of
AMVN in 250 mL of acetonitrile. The reaction mixture was
incubated at 50 °C for 3 h. The oxidation was terminated at
Resu lts
AMVN-In itia ted P er oxyl Ra d ica l Oxid a tion s. Pre-
vious work from our laboratory and others has employed
the azo initiator AMVN as a source of peroxyl radicals
for studies of antioxidant chemistry. AMVN decomposes
thermally to yield alkyl radicals (R•) which, in turn, react
with molecular oxygen to generate peroxyl radicals
(ROO•) (eqs 1 and 2) (17):
RNdNR f 2R• + N2
R• + O2 f ROO•
(1)
(2)
Advantages of using AMVN for these studies include the
selectivity with which AMVN yields peroxyl radicals, the
ease of controlling oxidation rates (via the AMVN con-
centration and temperature), and the lack of transition
metals used in other oxidant generating systems, which
often degrade primary oxidation products.
HP LC An a lysis of Oxid a tion P r od u cts of Gen istein .
Genistein was oxidized with peroxyl radicals generated
by thermolysis of AMVN in oxygenated acetonitrile.
These oxidations were conducted using a 10:1 molar ratio
of AMVN to genistein. This ratio was selected because it
resulted in a low rate of genistein oxidation. A low rate
of oxidation ensured that genistein was the major com-
ponent present in the reaction mixture at all times,
thereby suppressing the generation of secondary oxida-
tion products. Although this molar ratio appears high,
thermolysis at 50 °C consumes AMVN with nearly zero-
order kinetics and allows the oxidation of genistein to
be reproducibly controlled. Analysis of the reaction
mixture at 3 h by reverse-phase HPLC with diode-array
UV detection (Figure 1) indicated the presence of residual
genistein 1 and formation of five oxidation products 2-6.
Compound 2 was more polar than genistein, whereas the
other oxidation products were less polar as suggested by
RP-HPLC retention times.

640 Chem. Res. Toxicol., Vol. 13, No. 7, 2000
Arora et al.
HPLC and characterized on the basis of their spectro-
scopic data. GC/MS analysis of TMS-derivatized genistein
yielded a small molecular ion at m/z 486 and a strong
[M - 15]+ base peak at m/z 471. In contrast, the TMS
derivative of 2 exhibited a molecular ion at m/z 574 and
a strong [M - 15]+ ion at m/z 559. This mass corres-
ponded to the molecular weight of fully derivatized gen-
istein plus an additional OTMS group, suggesting that
2 was a hydroxylated derivative of genistein. Positive
APCI-MS yielded a [M + H]+ ion at m/z 287 which cor-
responds to the molecular weight of hydroxylated geni-
stein. The presence of an additional oxygen in 2 was con-
firmed by high-resolution positive FAB-MS data which
gave a [M + H]+ at m/z 287.0549 for 2 (calcd 287.0556),
corresponding to a molecular formula of C15H11O6.
Further information about the position of the hydroxyl
group on 2 was obtained by positive APCI-MS-MS
analysis of the [M + H]+ ion at m/z 287. Fragmentation
of the m/z 287 ion yielded product ions at m/z 271, 259,
241, 231, 161, and 153. The ion at m/z 271 resulted from
the loss of a hydroxyl group. The ions [MH - CO]+, [MH
- H2O - CO]+, and [MH - COCO]+ at m/z 259, 241, and
231, respectively, represented typical losses for hydroxy-
isoflavones (18). The base peak observed at m/z 153 was
attributed to a dihydroxylated A-ring fragment following
retro-Diels-Alder rearrangement characteristic of all
flavonoids. The presence of this fragment demonstrated
that ring A in 2 had remained unchanged when compared
to the parent genistein. The ion at m/z 161 had only been
observed earlier with isoflavones as a rearrangement
product for a dihydroxylated B-ring fragment (18). Taken
together, the MS-MS data suggested that hydroxylation
had occurred on ring B for product 2.
F igu r e 1. Reverse-phase HPLC analysis of oxidation products
formed during the reaction of genistein and AMVN-derived
peroxyl radicals in oxygenated acetonitrile for 3 h at 50 °C.
1
In the H NMR (CD3CN) spectrum of 2 (Table 1), the
signals of H-2 (δH 7.99) in ring C and H-6 (δH 6.25) and
H-8 (δH 6.38) in ring A were similar to those observed
earlier for genistein (16). The signals observed for ring
B in 2 were a doublet at δH 6.86 (J ) 8 Hz), a double
doublet at δH 6.90 (J ) 8 and 2 Hz), and a doublet at δH
7.04 (J ) 2 Hz), indicating that the additional OH group
is located at C-3′ in 2. On the basis of the spectroscopic
data and comparison to literature 1H NMR data in
acetone-d6 (19, 20), compound 2 was assigned as orobol
(5,7,3′,4′-tetrahydroxyisoflavone) (Figure 3).
F igu r e 2. Time course of the appearance of oxidation products
in oxygenated acetonitrile containing 1 µmol of genistein and
10 µmol of AMVN incubated at 50 °C.
Tim e Cou r se Stu d ies of P r od u ct F or m a tion . The
time course of formation of the oxidation products was
investigated using HPLC analysis. The reaction was
followed over 9 h, and the results are shown in Figure 2.
Relative product yields were estimated from integrated
peak areas. This is based on the similar UV/vis spectra
of genistein and its oxidation products (vide infra) and
the presumption of similar molar absorptivities. Time
course studies indicated that oxidation products 2 (λmax
) 202 and 260 nm) and 5 (λmax ) 210 and 254 nm) were
The other major oxidation product that formed, com-
pound 5, was very labile and decomposed readily. Com-
pound 5 could not be analyzed by GC/MS as it decom-
posed in part to genistein during either sample workup
or analysis. Positive APCI-MS analysis indicated a strong
signal at m/z 271, which was equivalent to genistein. A
small ion at m/z 444 observed in the positive APCI-MS
spectrum and the base peak at m/z 442 in the negative
ESI-MS spectrum suggested a product with a molecular
weight of 443. This molecular weight was confirmed by
positive high-resolution FAB-MS which gave a [M + H]+
ion at 444.1288 (calcd 444.1295), corresponding to a
molecular formula of C22H22O9N for 5. The presence of a
nitrogen atom in 5, as deduced from the high-resolution
MS data, together with the 1H and 13C NMR data
suggested the incorporation of an AMVN-derived moiety
on the genistein molecule.
formed in much greater yields than products 3 (λmax
)
208 and 254 nm), 4 (λmax ) 208 and 254 nm), and 6 (λmax
) 208 and 254 nm). The distribution of products changed
over time. Compound 5 was the principal product formed
until 3 h and then rapidly disappeared from the reaction
mixture. Compound 2, the other major oxidation product,
was formed more gradually and was still present in the
reaction mixture at the end of the 9 h time course study.
Oxidation products 3 and 4 were formed in very low
yields and had essentially disappeared by the end of the
9 h oxidation. The other minor oxidation product, com-
pound 6, began to form in very low yields after 3 h of
oxidation and was still present in the reaction mixture
at the end of the 9 h study.
1
The H NMR spectrum of 5 (Table 1) showed the H-2
of ring C at δH 8.15 and the H-6 and H-8 of ring A at δH
6.24 and 6.34, respectively. The 13C NMR data (Table 2)
and the HSQC and HMBC correlations of the protons
mentioned above indicated that rings A and C were
Isolation an d Ch ar acter ization of Oxidation P r od-
u cts. The oxidation products were isolated by MPLC and

Antioxidant Chemistry of Genistein
Ta ble 1. 1H NMR Ch em ica l Sh ifts of Com p ou n d s 2-5 in CD3CN
δH
Chem. Res. Toxicol., Vol. 13, No. 7, 2000 641
position
2
3
4
5
2
7.99 (s)
8.14 (s)
8.13 (s)
8.15 (s)
3
4
4a
5
5-OH
6
7
12.96 (s)
6.25 (d) (J ) 2 Hz)
12.56 (s)
6.26 (d) (J ) 2 Hz)
12.50 (s)
6.25 (d) (J ) 2.5 Hz)
12.52 (s)
6.24 (d) (J ) 2 Hz)
8
8a
1′
6.38 (d) (J ) 2 Hz)
6.38 (d) (J ) 2 Hz)
6.37 (d) (J ) 2 Hz)
6.34 (d) (J ) 3 Hz)
2′
3′
4′
7.04 (d) (J ) 2 Hz)
7.20-7.16 (m)
6.32 (br d) (J ) 10.5 Hz)
7.19-7.15 (m)
6.31 (br d) (J ) 10.5 Hz)
7.17-7.13 (m)
6.30 (br d) (J ) 9.5 Hz)
5′
6′
6.86 (d) (J ) 8 Hz)
6.90 (dd) (J ) 2 and 8 Hz)
6.32 (br d) (J ) 10.5 Hz)
7.20-7.16 (m)
6.31 (br d) (J ) 10.5 Hz)
7.19-7.15 (m)
6.30 (br d) (J ) 9.5 Hz)
7.17-7.13 (m)
1′′
2′′a
2′′b
3′′
4′′
5′′
6′′
7′′
2.14 (d) (J ) 15 Hz)
2.07 (d) (J ) 15 Hz)
2.13 (d) (J ) 15.5 Hz)
2.06 (d) (J ) 15 Hz)
2.26 (d) (J ) 15 Hz)
2.16 (d) (J ) 15 Hz)
1.40a (s)
1.39a (s)
1.39 (s)
1.38 (s)
1.36 (s)
1.33 (s)
1.57 (s)
1.56 (s)
1.62 (s)
a
Assignments interchangeable.
Ta ble 2. 13C NMR a n d HMBC Da ta of Com p ou n d s 3-5 in
CD3CN
3
4
5
a
position
δC
HMBC
3, 8a
δC
δC
HMBC
3, 4, 8a
2
3
4
4a
5
157.4
119.9
180.9
106.7
163.9
157.1
119.7
180.6
106.5
163.7
118.8
105.9
163.0
5-OH
6
7
4a, 5, 6
4a, 6
4a, 5, 6
4a, 5, 7, 8
99.7
100.8
165.5
95.3
100.4
165.0
95.0
8
4a, 6, 7, 8a
8a
1′
158.0
76.7
159.1
78.0
158.8
77.6
2′
3′
4′
5′
147.1
131.6
186.4
131.6
147.1
67.2
147.1
131.2
186.2
131.2
147.0
78.9
4′, 6′
1′, 5′
1′, 5′
1′, 3′
1′, 3′
2′, 4′
F igu r e 3. Structures of 1 (genistein), and oxidation products
2 (orobol) and 3-6.
6′
146.4
66.1
48.1
1′′
2′′a
2′′b
3′′
4′′
5′′
6′′
7′′
1′′, 3′′, 6′′
1′′, 3′′, 6′′
49.4
45.2
1′′, 3′′-7′′
1′′, 3′′-7′′
unchanged in 5 when compared to those of genistein. This
left ring B as the only possible location where the
modification could have occurred.
83.0
25.2
25.2
123.1
29.9
84.0
26.3
26.3
124.3
30.9
82.7
26.5
25.4
121.1
25.2
2′′, 3′′, 5′′
2′′-4′′
2′′, 3′′, 5′′
2′′-4′′
The NOESY spectrum revealed a small cross-peak
between the H-2 of ring C and the H-2′ and H-6′ of ring
B. Both B-ring protons (H2′ and H6′) showed HSQC
correlations to two overlapping carbons at δC 147.1. The
broad doublet at δH 6.30 (2H, J ) 9.5 Hz), which
correlated to two overlapping carbons at δC 131.2, was
assigned to the H-3′ and H-5′ of the B-ring. The HMBC
experiment revealed correlations between H-2′ and/or
H-6′ and a carbonyl at δC 186.2, indicating that the 4′-
hydroxyl in genistein was oxidized to a keto group in 5.
HMBC correlations between H-3′ and/or H-5′ and a
quaternary carbon at δC 77.6 (C-1′) were also observed,
indicating that the C-1′ location was the only possible
connection between the B-ring of the isoflavone moiety
and the AMVN-derived moiety in 5.
1′′, 2′′, 6′′
1′′, 2′′, 6′′
a
Based on HMBC data.
2.26 (J ) 15 Hz) and δH 2.16 (J ) 15 Hz). The isopropyl
proton which appears at δH 1.65 in AMVN was not seen
in the spectrum of 5, indicating a substituent at that
position. Additionally, the isopropyl methyl groups in 5,
when compared to those in AMVN, exhibited HMBC
correlations to a downfield-shifted quaternary carbon at
δC 82.7 (C-3′′) and to the methylene carbon at δC 45.2
(C-2′′) (Figure 4). The methyl signal at δH 1.62 (C-7′′)
exhibited HMBC cross-peaks to the nitrile carbon at δC
121.1 (C-6′′), to the methylene carbon C-2′′, and to a
quaternary carbon at δC 78.9 that was assigned as C-1′′.
When compared to those in AMVN, C-1′′ and C-3′′ in 5
experienced downfield shifts of 6.1 and 56.9 ppm, respec-
1
For the AMVN-derived moiety, the H NMR spectrum
of 5 revealed three methyl groups at δH 1.33, 1.36, and
1.62 and two doublet signals of a methylene group at δH

642 Chem. Res. Toxicol., Vol. 13, No. 7, 2000
Arora et al.
NMR data, 3 was determined to be 1′-(alkyloxyhydro-
peroxide)-4′-oxogenistein. This is equivalent to 4′-oxo-
genistein with an O-X-OOH substituent at the 1′-position,
where X corresponds to C(CN)(CH3)CH2C(CH3)2 derived
from AMVN.
We were able to generate sufficient quantities of
compound 4 to acquire 13C NMR data. The 13C NMR data
for 4 were very similar to the carbon data of 3 derived
from HMBC experiments. Both compounds exhibited the
same chemical shifts when compared to 5. In the case of
4, C-1′′ was at δC 67.2 and C-3′′ was at δC 84.0, again
suggesting that the AMVN radical substitution in 4 was
O-X-OOH instead of the OO-X-OOH observed for 5
[where X is C(CN)(CH3)CH2C(CH3)2 derived from AMVN].
The negative ESI-MS-MS data were also similar for
compounds 3 and 4. On the basis of their similar MS and
NMR spectra, we infer that 3 and 4 may be stereoisomers
of each other, perhaps differing in configuration at C-1′′.
F igu r e 4. HMBC correlations (C f H) of the AMVN moiety of
5.
tively, indicating oxygen substituents on both ends of the
AMVN-derived moiety in 5. On the basis of all spectro-
scopic data, the structure of 5 was assigned as 1′-
(alkyldioxyhydroperoxide)-4′-oxogenistein. This is equiva-
lent to 4′-oxogenistein with an OO-X-OOH substitution
at the 1′-position of the molecule, where X corresponds
to C(CN)(CH3)CH2C(CH3)2 derived from AMVN.
Compounds 3 and 4 could not be analyzed by GC/MS
either as they apparently decomposed in part to genistein
during sample workup or analysis. For both compounds,
positive APCI-MS yielded a small peak at m/z 428 and
negative ESI-MS showed a base peak at m/z 426. High-
resolution positive FAB-MS gave a [M + H]+ peak at
428.1357 (calcd 428.1346) for 3 and 428.1341 (calcd
428.1346) for 4, corresponding to a molecular formula of
C22H22O8N for both compounds. The MS data indicated
that compounds 3 and 4 differed from 5 only by possess-
ing one fewer oxygen. The 1H NMR data for 3 and 4 were
very similar to those of 5, suggesting that all three
compounds were structurally related. Since 3 and 4 were
formed in very low yields and decomposed over periods
of extended storage, it was very difficult to generate
sufficient quantities of these compounds to conduct 13C
NMR and HMBC experiments.
The extremely low yield of compound 6 coupled with
very fast degradation precluded us from acquiring high-
resolution MS or NMR data for the compound. Positive
APCI-MS indicated a [M + H]+ ion at m/z 412 for 6. This
corresponded to a molecular mass of 411 for 6, which
would be equivalent to the mass of compound 5 minus
two oxygen atoms. Under our RP-HPLC conditions, 5
eluted at 33 min whereas 6 eluted at 42 min. This
indicated that 6 was less polar than 5. On the basis of
the MS information for 6, we hypothesized that it was
1′-(alkyldioxy)-4′-oxogenistein. This is equivalent to 4′-
oxogenistein with an OOX [where X is C(CN)(CH3)CH2-
CH(CH3)2] substitution at the 1′ position instead of the
OO-X-OOH [where X is C(CN)(CH3)CH2C(CH3)2] substi-
tution observed for 5.
Gen istein Oxid a tion s in Aceton itr ile a n d H218O.
To ascertain whether the hydroxyl group at C-3′ in orobol
originated from water or atmospheric oxygen, we con-
ducted experiments with 18O-labeled water. These ex-
periments were conducted by carrying out the genistein
and AMVN oxidations in an acetonitrile and 18O-labeled
water mixture (95:5, v/v) rather than in acetonitrile alone.
Corresponding oxidations also were carried out using
unlabeled water. Positive APCI-MS analysis of orobol
purified from the 18O-labeled and unlabeled water oxida-
tions indicated that the molecular ion for orobol remained
unchanged at [M + H]+ m/z 287 for both samples (data
not shown). If water was the source of oxygen necessary
for genistein hydroxylation to orobol, we would have
expected the 18O label to be incorporated into orobol and
for its mass to be shifted by 2 mass units. The 18O-labeled
water experiment demonstrated that the hydroxylation
of genistein during peroxyl radical oxidation does not
occur through hydrolysis.
A possible concern was that the 18O label had been
incorporated into orobol but was lost during the HPLC
purification step due to an exchange with unlabeled
water from the mobile phase. To check for exchange of
the 18O label, we incubated our orobol standard with 1
mL of acetonitrile and 18O-labeled water (28:72, v/v) for
3 h at room temperature. The orobol standard was
immediately evaporated in vacuo and analyzed by posi-
tive APCI-MS. The observed [M + H]+ for orobol re-
mained unchanged at m/z 287, confirming that no
exchange of the label had occurred during the HPLC step
(data not shown).
We were able to acquire HMBC data for 3 and used
the HMBC data to assign most of the 13C NMR reso-
nances for that compound. The 1H, 13C, and two-
dimensional NMR data for the isoflavone moiety of 3
were very similar to those in 5, suggesting that this part
of the molecule remained unchanged in 3 when compared
to 5. The only changes observed in the NMR spectra of 3
belonged to the AMVN-derived moiety, indicating the loss
of one oxygen in 3 when compared to 5. As in 5, the
methyl group (δH 1.57, H-7′′) geminal to the nitrile group
showed HMBC correlations to the nitrile carbon C-6′′, to
the methylene carbon C-2′′, and to the quaternary C-1′′
that had experienced an upfield shift of 12.8 ppm in 3,
when compared to that in 5. As in 5, the methyl protons
H-4′′ (δH 1.40) and H-5′′ (δH 1.39) showed HMBC cor-
relations to the quaternary C-3′′ (δC 83.0) that had
experienced an upfield shift of only 0.3 ppm in 3. On the
basis of the NMR data for 3, we concluded that the only
location possible for the loss of one oxygen atom, as
indicated by MS data, could be the connection between
the isoflavone- and the AMVN-derived moieties between
C-1′ and C-1′′ in 3, when compared to 5. Negative ESI-
MS-MS experiments confirmed the site of the oxygen loss
in 3. Fragmentation of the [M - H]- ion at m/z 442
yielded strong product ions at m/z 268 and 285. These
fragments corresponded to masses of the 4′-oxogenistein
moiety and 4′-oxogenistein moiety with one oxygen,
respectively. In contrast to the negative ESI-MS-MS data
of 5, we did not observe a fragment at m/z 301, which
would have corresponded to the 4′-oxogenistein fragment
with two oxygens attached. On the basis of our MS and

Antioxidant Chemistry of Genistein
Chem. Res. Toxicol., Vol. 13, No. 7, 2000 643
Discu ssion
The purpose of this investigation was to isolate and
characterize the reaction products of genistein with
alkylperoxyl radicals derived from AMVN in a homoge-
neous solution. The choice of AMVN as the model oxidant
for these studies was dictated by the specificity of peroxyl
radical generation from AMVN (17). In previous studies
with R-tocopherol (vitamin E), antioxidant reactions with
AMVN-derived peroxyl radicals in acetonitrile and ac-
etonitrile/water mixtures (21, 22) were similar to those
subsequently detected in lipid bilayers (23, 24), in sub-
cellular fractions (25), and in whole organs (26) in vitro
and in vivo (27). Thus, the model system used for these
studies is reasonably predictive of antioxidant chemistry
in more complex biological systems.
Here we report that genistein reacts with AMVN-
derived peroxyl radicals in oxygenated acetonitrile to
form orobol (2) and 1′-(alkyldioxyhydroperoxide)-4′-oxo-
genistein (5) as principal products, and 1′-(alkyloxyhy-
droperoxide)-4′-oxogenistein (3 and 4) and 1′-(alkyldioxy)-
4′-oxogenistein (6) as minor oxidation products. Of these
five oxidation products, orobol is the only compound that
has been characterized previously. Ours, however, is the
first study to report the formation of orobol as an
oxidation product during the antioxidant reactions of
genistein. A recent paper demonstrated that rat and
human cytochrome P450 metabolism of genistein re-
sulted in the formation of orobol (18). Orobol has also
been isolated previously from the seeds of Bolusanthus
speciosus (Bolus) Harms (19) and from the Podalyrieae
and Lipariae papilionoid tribes of the Fabaceae (20),
legume plants indigenous to South Africa. Orobol has
biological activities that are similar to that of genistein.
It has been shown to induce topoisomerase II-dependent
DNA cleavage at a rate comparable to that of genistein
(28), inhibit dihydroxyphenylalanine decarboxylase with
an activity stronger than that of genistein (29), inhibit
15-lipoxygenase (30) and phosphatidylinositol turnover
in cell culture with an IC50 similar to that of genistein
(31), and increase cytoplasmic free calcium levels in
isolated rat hepatocytes (32).
F igu r e 5. Proposed mechanism for the formation of 2 (orobol)
during the peroxyl radical oxidation of 1 (genistein).
We were interested in understanding the mechanism
for formation of orobol as a product of genistein antioxi-
dant chemistry. Our experiments with 18O-labeled water
suggested that instead of water, atmospheric oxygen was
the source of the oxygen needed for hydroxylation of
genistein to orobol. On the basis of our findings, we
proposed a mechanism for the formation of orobol during
peroxyl radical oxidation of genistein (Figure 5).
The other family of oxidation products that we identi-
fied, the stable AMVN-derived radical adducts observed
for genistein, appears to be unique in two ways. First,
although previously observed with other phenolic anti-
oxidants, peroxyl radical addition products have not been
reported in flavonoid oxidations. Second, further oxida-
tion of the AMVN-derived moiety in these adducts is un-
precedented. Of these products, compound 5 was the ma-
jor product that was formed. Compounds 3, 4, and 6 were
generated in very low yields. For all four products, it ap-
peared that AMVN-derived alkoxyl (HXO•) or peroxyl ra-
dicals (HXOO•) attached at the C-1′ position on the B-ring
of the genistein phenoxyl radical. Further oxidation of
the AMVN moiety led to formation of the corresponding
O-X-OOH or OO-X-OOH hydroperoxide. A mechanism for
F igu r e 6. Proposed mechanism for the formation of compounds
3-6 during the peroxyl radical oxidation of 1 (genistein).
the formation of these compounds is proposed in Figure
6.
An initial one-electron oxidation of genistein by a
peroxyl radical yields a genistein phenoxyl radical.
Recombination with a peroxyl or alkoxyl radical would
form either 1′-(alkyldioxy)-4′-oxogenistein or 1′-(alkyloxy)-
4′-oxogenistein, respectively. Further abstraction of a
proton from the alkyl side chain of the AMVN moiety,
followed by addition of oxygen in the presence of a proton
donor, would result in the formation of product 5 or

644 Chem. Res. Toxicol., Vol. 13, No. 7, 2000
Arora et al.
products 3 and 4. These appeared to be stable end
products of genistein reactions with AMVN.
similar to those observed in this study in a homogeneous
solution system.
Interestingly, 1′-(alkyldioxyhydroperoxide)-4′-oxogeni-
stein (5) was the major product formed during these
oxidations of genistein with alkylperoxyl radicals. Its
precursor, 1′-(alkyldioxy)-4′-oxogenistein (6), was formed
in almost negligible yield, suggesting that 6 was less
stable than genistein in oxidations with AMVN. Products
3 and 4, which were formed through a parallel oxidation
pathway involving an AMVN alkoxyl radical, were also
generated in low yields, indicating that this was a minor
pathway for the oxidation of genistein by AMVN.
Ack n ow led gm en t. We thank Dr. Neil E. J acobsen
for help with the acquisition of NMR data and Dr. Arpad
Somogyi for determination of FAB spectra. This inves-
tigation was supported by NIH Grants CA75599 and
ES06694.
Refer en ces
(1) King, H., and Locke, F. B. (1980) Cancer mortality among Chinese
in the United States. J . Natl. Cancer Inst. 65, 1141-1148.
(2) Locke, F. B., and King, H. (1980) Cancer mortality risk among
J apanese in the United States. J . Natl. Cancer Inst. 65, 1149-
1156.
(3) Murphy, P. A. (1982) Phytoestrogen content of processed soybean
products. Food Technol. 36, 60-64.
(4) Akiyama, T., Ishida, J ., Nakagawa, S., Ogawara, H., Watanabe,
S.-I., Itoh, N., Shibuya, M., and Fukami, Y. (1987) Genistein, a
specific inhibitor of tyrosine-specific kinases. J . Biol. Chem. 262,
5592-5595.
(5) Verma, S. P., and Goldin, B. R. (1998) Effect of soy-derived
isoflavonoids on the induced growth of MCF-7 cells by estrogenic
environmental chemicals. Nutr. Cancer 30, 232-239.
(6) Booth, C., Hargreaves, D. F., Hadfield, J . A., McGown, A. T., and
Potten, C. S. (1999) Isoflavones inhibit intestinal epithelial cell
proliferation and induce apoptosis in vitro. Br. J . Cancer 80,
1550-1557.
(7) Wei, H., Bowen, R., Zhang, X., and Lebwohl, M. (1998) Isoflavone
genistein inhibits the initiation and promotion of two-stage
carcinogenesis in mice. Carcinogenesis 19, 1509-1514.
(8) Wei, H., Wei, L., Frenkel, K., Bowen, R., and Barnes, S. (1993)
Inhibition of tumor promoter-induced hydrogen perxoide forma-
tion in vitro and in vivo by genistein. Nutr. Cancer 20, 1-12.
(9) Pratt, D. E., and Birac, P. M. (1979) Source of anitoxidative
activity of soybeans and soy products. J . Food Sci. 44, 1720-1722.
(10) Record, I. R., Dreosti, I. E., and McInerney, J . K. (1995) The
antioxidant activity of genistein in vitro. J . Nutr. Biochem. 6,
481-485.
(11) J ha, H. C., von Recklinghausen, G., and Zilliken, F. (1985)
Inhibition of in vitro microsomal lipid peroxidation by isofla-
vonoids. Biochem. Pharmacol. 34, 1367-1369.
(12) Bors, W., Heller, W., Michel, C., and Saran, M. (1990) Flavonoids
as antioxidants: Determination of radical-scavenging activities.
Methods Enzymol. 186, 343-355.
(13) De Whalley, C. V., Rankin, S. M., Henlt, J . R. S., J essup, W., and
Leake, D. S. (1990) Flavonoids inhibit the oxidative modification
of low-density lipoproteins by macrophages. Biochem. Pharmacol.
39, 1743-1750.
(14) Chen, Z. Y., Chan, P. T., Ho, K. Y., Fung, K. P., and Wang, J .
(1996) Antioxidant activity of natural flavonoids is governed by
number and location of their aromatic hydroxyl groups. Chem.
Phys. Lipids 79, 157-163.
(15) Arora, A., Nair, M. G., and Strasburg, G. M. (1998) Structure-
activity relationships for antioxidant activities of a series of
flavonoids in a liposomal system. Free Radical Biol. Med. 24,
1355-1363.
(16) Chang, Y.-C., Nair, M. G., Santell, R. C., and Helferich, W. G.
(1994) Microwave-mediated synthesis of anticarcinogenic isofla-
vones from soybeans. J . Agric. Food Chem. 42, 1869-1871.
(17) Niki, E. (1990) Free radical initiators as source of water- or lipid-
soluble peroxyl radicals. Methods Enzymol. 186, 100-108.
(18) Roberts-Kirchhoff, E. S., Crowley, J . R., Hollenberg, P. F., and
Kim, H. (1999) Metabolism of genistein by rat and human
cytochrome P450s. Chem. Res. Toxicol. 12, 610-616.
(19) Asres, K., Mascagni, P., O’Neill, M. J ., and Phillipson, J . D. (1985)
Isoflavonoids from Bolusanthus speciosus (Bolus) Harms Legu-
minosae. Z. Naturforsch. 40c, 617-620.
There have been many studies conducted to character-
ize the oxidation products of reactions of other phenolic
antioxidants with AMVN-derived peroxyl radicals. With
R-tocopherol and other phenolic antioxidants, adducts
that are structurally analogous to compound 6 have been
previously observed (21, 22, 33). However, further oxida-
tion of the AMVN moiety to give products analogous to
3, 4, or 5 has not been observed. We cannot presently
explain this surprising secondary oxidation of AMVN-
derived alkyl chains in genistein-AMVN oxy radical
adducts. It seems highly unlikely that these isopropyl
moieties are intrinsically more reactive toward radicals
than genistein itself. However, noncovalent association
of genistein molecules and/or initial reaction products
(such as 6) may alter relative reactivities enough to favor
formation of products 3-5. Further work will be required
to sort out the chemistry that is involved.
The site of antioxidant chemistry for formation of all
these reaction products was the B-ring of genistein. This
finding agrees with earlier studies that suggested that
ring B was the principal site of antioxidant reactions for
flavonoids (12, 34). Our study provides the first conclusive
proof of this hypothesis for genistein oxidations by perox-
yl radicals. A previous paper from our laboratory had
reported that the site of antioxidant reactions for the
green tea catechin (-)-epigallocatechin-3-gallate when
oxidized with peroxyl radicals also resides on the B-ring
of the molecule (35). What is particularly interesting
about the results of our product studies with flavonoids
is the variety of antioxidant chemistries displayed by the
B-ring phenols. Genistein produced products that are
relatively typical of well-documented phenol antioxidant
reactions: B-ring hydroxylation and formation of radical
addition products. In contrast, (-)-epigallocatechin and
(-)-epigallocatechin-3-gallate yield neither B-ring hy-
droxylation nor stable peroxyl radical addition products.
Instead, one-electron oxidations appear to initiate radical
polymerization reactions, and transient radical addition
products decompose with B-ring expansion.2 Thus, anti-
oxidant chemistry of simpler phenols does not necessarily
predict the course of flavonoid phenol antioxidant reac-
tions.
The characterization of these product structures pro-
vides potential markers of genistein antioxidant reac-
tions. Sensitive assays for specific genistein oxidation
products may thus be useful markers for antioxidant
reactions of the isoflavone in biological systems. Cur-
rently, we are investigating whether reactions of genistein
and peroxyl radicals in lipid bilayers in vitro and in
biological membranes in vitro and in vivo yield products
(20) de Nysschen, A.-M., van Wyk, B.-E., and van Heerden, F. R. (1998)
Seed flavonoids of the Podalyrieae and Liparieae (Fabaceae). Plant
Syst. Evol. 212, 1-11.
(21) Yamauchi, R., Matsui, T., Satake, Y., Kato, K., and Ueno, Y.
(1989) Reaction products of R-tocopherol with
a free radical
initiator, 2,2′-azobis(2,4-dimethylvaleronitrile). Lipids 24, 204-
209.
(22) Liebler, D. C., Baker, P. F., and Kaysen, K. L. (1990) Oxidation
of vitamin E: Evidence for competing autoxidation and peroxyl
radical trapping reactions of the tocopheroxyl radical. J . Am.
Chem. Soc. 112, 6995-7000.
2
S. Valcic, A. Muders, J . A. Burr, B. N. Timmermann, and D. C.
Liebler, manuscript submitted for publication.

Antioxidant Chemistry of Genistein
Chem. Res. Toxicol., Vol. 13, No. 7, 2000 645
(23) Liebler, D. C., Kaysen, K. L., and Burr, J . A. (1991) Peroxyl radical
trapping and autoxidation reactions of R-tocopherol in lipid
bilayers. Chem. Res. Toxicol. 4, 89-93.
(24) Yamauchi, R., Yagi, Y., and Kato, K. (1994) Isolation and
characterization of addition products of R-tocopherol with peroxyl
radicals of dilinoleoylphosphatidylcholine in liposomes. Biochim.
Biophys. Acta 1212, 43-49.
(30) Kohyama, T., Hasumi, K., Murakawa, S., and Endo, A. (1994)
Inhibition of 15-lipoxygenase by orobol. J . Antibiot. 47, 1069-
1071.
(31) Imoto, M., Yamashita, T., Sawa, T., Kurasawa, S., Naganawa,
H., Takeuchi, T., Bao-quan, Z., and Umezawa, K. (1988) Inhibition
of cellular phosphatidylinositol turnover by psi-tectorigenin.
FEBS Lett. 230, 43-46.
(25) Ham, A. J . L., and Liebler, D. C. (1995) Vitamin E oxidation in
rat liver mitochondria. Biochemistry 34, 5754-5761.
(26) Ham, A. J . L., and Liebler, D. C. (1997) Antioxidant reactions of
vitamin E in the perfused rat liver: product distribution and effect
of dietary vitamin E supplementation. Arch. Biochem. Biophys.
339, 157-164.
(27) Kramer-Stickland, K. A., Krol, E. S., and Liebler, D. C. (1999)
UV-B-induced photooxidation of vitamin E in mouse skin. Chem.
Res. Toxicol. 12, 187-191.
(28) Yamashita, Y., Kawada, S.-Z., and Nakano, H. (1990) Induction
of mammalian topoisomerase II dependent DNA cleavage by
nonintercalative flavonoids, genistein and orobol. Biochem. Phar-
macol. 39, 737-744.
(29) Umezawa, H., Tobe, H., Shibamoto, N., Nakamura, F., and
Nakamura, K. (1975) Isolation of isoflavones inhibiting DOPA
decarboxylase from fungi and streptomyces. J . Antibiot. 28, 947-
952.
(32) Tomonaga, T., Mine, T., Kojima, I., Taira, M., Hayashi, H., and
Isono, K. (1992) Isoflavonoids, genistein, psi-tectorigenin, and
orobol, increase cytoplasmic free calcium in isolated rat hepato-
cytes. Biochem. Biophys. Res. Commun. 182, 894-899.
(33) Horswill, E. C., and Ingold, K. U. (1966) The oxidation of phenols.
II. The oxidation of 2,4-di-tert-butylphenol with peroxy radicals.
Can. J . Chem. 44, 269-277.
(34) Rice-Evans, C., Miller, N. J ., and Paganga, G. (1996) Structure-
activity relationships of flavonoids and phenolic acids. Free
Radical Biol. Med. 20, 933-956.
(35) Valcic, S., Muders, A., J acobsen, N. E., Liebler, D. C., and
Timmermann, B. N. (1999) Antioxidant chemistry of green tea
catechins. Identification of products of the reaction of (-)-
epigallocatechin gallate with peroxyl radicals. Chem. Res. Toxicol.
12, 382-386.
TX000015A