Stable isotope labeling pattern of resveratrol and related natural stilbenes.

Article abstract of DOI:10.1021/jf011103j

The stable isotope characterization of resveratrol 1 from Polygonum cuspidatum and of related natural stilbenes 11 and 12 obtained by hydrolysis of the corresponding glucosides 2 and 3 from Rheum is reported. The C(6)-C(2)-C(6) framework of suitably protected derivatives of 1, 2, and 3 has been degraded with ozone to the C(6)-C(1) aldehydes 4, 5, 9, and 10, retaining all hydrogen atoms of the precursors. The natural and synthetic derivatives are characterized and distinguished by natural abundance deuterium nuclear magnetic resonance studies. In the case of anisaldehyde 4 the two series show, as expected, the characteristic difference of the aromatic labeling. The formyl deuterium contents of 4 and 5 from resveratrol are remarkably different, seemingly reflecting the different enrichments existing between positions 3 and 2, respectively, of the phenylpropanoid precursor. The positional delta(18)O values of the extractive materials 1-3 were also determined. In this instance a selective deoxygenation procedure was adopted, leading from 1 to the products 6, 7, and 8. The delta(18)O values of the latter compounds reveal, respectively, those at position 4' and positions 3 and 5 of 1. Similarly, the phenolic products 11 and 12 were converted into 13 and 14. From the delta(18)O values of the single components it is possible to design a detailed map of the oxygen fractionations which characterizes the stilbenes 1-3. In particular, the oxygen present at position 4' of the phenylpropanoid moiety of 1-3 shows delta(18)O values of +11.5, +1.8, and +6.7 per thousand, respectively. Moreover, the phenolic oxygen atom at position 3' of rhapontin 3 shows a value of +11.7 per thousand. The data are compared with those previously obtained on structurally related compounds. These results show the utility of simple chemical degradations in the stable isotope characterization of structurally complex food components.

Full text of DOI:10.1021/jf011103j

2748 J. Agric. Food Chem. 2002, 50, 2748−2754
Stable Isotope Labeling Pattern of Resveratrol and Related
Natural Stilbenes
GIOVANNI FRONZA,† CLAUDIO FUGANTI,*,† STEFANO SERRA,†
§
MARCO CISERO,‡ AND JOSEPH KOZIET
Dipartimento di Chimica del Politecnico and CNR, Centro di Studio sulle Sostanze Organiche
Naturali, Via Mancinelli 7, 20131 Milano, Italy; San Giorgio Flavors SpA, Via Fossata 114,
10147 Torino, Italy; and Centre de Recherche Pernod Ricard, 120 Avenue du Mare´chal Foch,
94015 Creteil Cedex, France
The stable isotope characterization of resveratrol 1 from Polygonum cuspidatum and of related natural
stilbenes 11 and 12 obtained by hydrolysis of the corresponding glucosides 2 and 3 from Rheum is
reported. The C6-C2-C6 framework of suitably protected derivatives of 1, 2, and 3 has been degraded
with ozone to the C6-C1 aldehydes 4, 5, 9, and 10, retaining all hydrogen atoms of the precursors.
The natural and synthetic derivatives are characterized and distinguished by natural abundance
deuterium nuclear magnetic resonance studies. In the case of anisaldehyde 4 the two series show,
as expected, the characteristic difference of the aromatic labeling. The formyl deuterium contents of
4 and 5 from resveratrol are remarkably different, seemingly reflecting the different enrichments existing
between positions 3 and 2, respectively, of the phenylpropanoid precursor. The positional δ18O values
of the extractive materials 1-3 were also determined. In this instance a selective deoxygenation
procedure was adopted, leading from 1 to the products 6, 7, and 8. The δ18O values of the latter
compounds reveal, respectively, those at position 4′ and positions 3 and 5 of 1. Similarly, the phenolic
products 11 and 12 were converted into 13 and 14. From the δ18O values of the single components
it is possible to design a detailed map of the oxygen fractionations which characterizes the stilbenes
1-3. In particular, the oxygen present at position 4′ of the phenylpropanoid moiety of 1-3 shows
δ18O values of +11.5, +1.8, and +6.7‰, respectively. Moreover, the phenolic oxygen atom at position
3′ of rhapontin 3 shows a value of +11.7‰. The data are compared with those previously obtained
on structurally related compounds. These results show the utility of simple chemical degradations in
the stable isotope characterization of structurally complex food components.
KEYWORDS: Resveratrol; rhapontin; stilbenes; natural; deuterium NMR; oxygen-18; isotopic charac-
terization
INTRODUCTION
The distribution of the stable isotopes within the atomic
species constituting an organic molecule is not random but
depends instead upon the way in which it was constructed (1).
Accordingly, through the determination of the site-specific
isotopic enrichment of a compound it is possible to elucidate
the nature of the operations that presided over the synthesis,
thus revealing its origin. In this context we now report on the
stable isotope characterization of resveratrol 1 from Polygonum
cuspidatum (2) and of the related stilbenic glycosides 2 and 3
from Rheum (3, 4). Resveratrol 1, a long known plant secondary
metabolite that received recent attention due to its biological
activity (5), is commercially available as a food additive. There
are two sources for this material: the costly plant extraction
(6) and the chemical synthesis from two readily accessible
C-6-C-1 precursors, that is, anisaldehyde 4 and 3,5-dimethoxy-
benzaldehyde 5 (7-9). Due to the preference of consumers for
* Corresponding author (telephone +39 2 2399 3040; fax +39 2 2399
3080; e-mail claudio.fuganti@polimi.it).
† Dipartimento di Chimica del Politecnico and CNR, Centro di Studio
sulle Sostanze Organiche Naturali.
‡ San Giorgio Flavors SpA.
§ Centre de Recherche Pernod Ricard.
10.1021/jf011103j CCC: $22.00 © 2002 American Chemical Society
Published on Web 04/06/2002

Natural Abundance 2H and 18O Studies on Resveratrol
J. Agric. Food Chem., Vol. 50, No. 10, 2002 2749
positional oxygen depletion values are significant for identifica-
tion purposes (12). To specifically identify a single oxygen atom
of the many present in a multiply oxygenated molecule, it is
necessary to rely on the techniques of the chemical synthesis,
which are sometimes difficult to apply. However, the diagnostic
potential of the positional δ18O measurements in the definition
of the extractive rather than synthetic origin of molecules of
importance in food chemistry has been recently demonstrated
in the case of raspberry ketone and vanillin, phenolic compounds
structurally related to 1-3 (13-15). Accordingly, in the present
study 1-3 were specifically deoxygenated, thus allowing also
the δ18O values relative to the oxygen atoms present at different
positions of the C-6-C-2-C-6 framework to be acquired. The
results of the multiple isotope characterization of stilbenes 1-3
are presented below.
Scheme 1. Degradation of Resveratrol 1 into the Aldehydes 4 and 5
for the Deuterium Isotopic Measurements and to the Compounds 6−8
for the Positional 18O Isotopic Determinationa
EXPERIMENTAL PROCEDURES
a (i) diazomethane; (ii) O /CH Cl2−MeOH 1:1, −78 °C; then Ph3P; (iii)
3
2
Ph3PdCHCO Et/CH Cl2; (iv) H /Pd−C/AcOEt; (v) LiAlH /THF; (vi) Ph3P/CCl4
2
2
2
4
Origin of the Samples. Resveratrol from P. cuspidatum (from
China) was a gift of Dr. P. Maggioni (Chemprogress, Sesto Ulteriano,
Italy). Rhapontin, containing ∼20% of the â-glycoside of 4′-O-methyl-
resveratrol, was a commercial sample. Natural anisaldehyde was from
Robertet (Grasse, France), whereas the synthetic modifications were
from Aldrich (Milwaukee, WI) and BBE (Milano, Italy), respectively.
Synthetic 3,5-dimethoxybenzaldehyde was from Aldrich.
Degradation of Natural Resveratrol. (a) Anisaldehyde 4 and
3,5-Dimethoxybenzaldehyde 5. Tri-O-methylresveratrol was obtained
quantitatively by treating a dilute methanolic solution of resveratrol at
10 °C with a large excess of ethereal diazomethane over ∼24 h. The
above material, 8.1 g (0.03 mol), in 80 mL of CH2Cl2/MeOH (1:1)
was treated at -78 °C with ozonized oxygen until the appearance of a
blue color. The ozone stream was replaced with nitrogen, and Ph3P
(7.8 g; 0.03 mol) in CH2Cl2 (30 mL) was added dropwise. After 30 min,
the temperature was raised and the crude mixture was evaporated under
vacuum at 30-40 °C. The residue was taken up with Et2O/hexane,
and the precipitated phosphine oxide was removed by filtration. The
oily residue was chromatographed on SiO2, eluting with increasing
quantities of ethyl acetate in hexane to give anisaldehyde 4, 3 g (73%),
purified by bulb-to-bulb vacuum distillation, then mixed fractions, and,
finally, 3,5-dimethoxybenzaldehyde 5, 3 g (60%), mp 46-48 °C (from
hexane). There was no formation of the acids or of the alcohols
corresponding to 4 and 5.
(b) 3-(4-Methoxyphenyl)-1-chloropropane 6 and 3-(3,5-Dimethoxy-
phenyl)-1-chloropropane 7. Anisaldehyde 4 (2.5 g; 0.018 mol) in
CH2Cl2 (30 mL) was treated at reflux for 5 h with carboxyethyl-
methylenetriphenylphosphorane (6.28 g; 0.018 mol). The reaction
mixture was evaporated, and the residue was treated with Et2O/hexane.
The precipitated triphenylphosphine oxide was filtered, and the oily
residue obtained upon evaporation of the solvent was chromatographed
on a short SiO2 column, yielding with hexane/AcOEt (6:4), ethyl
p-methoxycinnamate, 2.8 g (76%): 1H NMR (500 MHz, CDCl3) δ 7.63
(1H, d, J ) 16.2 Hz), 7.46 (2H, d, J ) 8.2 Hz), 6.89 (2H, d, J ) 8.2
Hz), 6.29 (1H, d, J ) 16.2 Hz), 4.24 (2H, q, J ) 6.7 Hz), 3.82 (3H, s,
OCH3), 1.32 (3H, t, J ) 6.7 Hz). The latter material (2.4 g; 0.011 mol)
in AcOEt (30 mL) was submitted to hydrogenation at room temperature
in the presence of 10% Pd/C (0.2 g). At the end of the adsorption, the
reaction mixture was filtered and evaporated, and the residue, 2.4 g,
was treated at reflux in dry THF (50 mL) with LiAlH4 (∼0.5 g; 0.012
mol) for 5 h. After usual workup, 3-(4-methoxyphenyl)propan-1-ol,
1.4 g (76%), was isolated after purification by SiO2 column chroma-
tography: 1H NMR (500 MHz, CDCl3) δ 7.09 (2H, d, J ) 8.8 Hz),
6.81 (2H, d, J ) 8.8 Hz), 3.76 (3H, s, OCH3), 3.63 (2H, t, J ) 6.2
Hz), 2.63 (2H, t, J ) 7.5 Hz), 2.19 (1H, s, OH), 1.84 (2H, m).
The latter material (1.4 g; 0.0084 mol), in CCl4 (40 mL), was refluxed
for 8 h with Ph3P (2.2 g; 0.0084 mol). The reaction mixture was
evaporated and the residue taken up with Et2O/hexane. The precipitated
triphenylphosphine oxide was removed by filtration, and the oil that
remained upon evaporation of the solvent was chromatographed on
SiO2, eluting first with 5% AcOEt in hexane some unreacted triphenyl-
phosphine and, by increasing the amount of AcOEt, 3-(4-methoxy-
reflux; (vii) H /Pd/C.
2
what is natural in food, it seemed desirable to verify if isotopic
methods could be suitable also in the case of resveratrol 1 for
the definition of its extractive rather than synthetic origin.
Because resveratrol 1 is accessible in quantity only from P.
cuspidatum, the structurally related stilbenes 2 and 3 were
analyzed as comparison substances.
In food chemistry, perhaps the most useful analytical method
for the definition of the origin of an organic molecule is
2
deuterium natural abundance H NMR spectroscopy (SNIF-
NMR) (10). However, a prerequisite for the applicability of this
technique is that the deuterium signals in the NMR spectrum
of the molecule under examination are dispersed enough to allow
a proper integration of the individual signals. Unfortunately,
this is not the case for resveratrol 1 and the related products 2
and 3, which are characterized by the presence of only C-H
signals on sp2 carbon atoms, the 1H NMR signals of which occur
in a restricted region (for 1 between 6.20 and 7.40 ppm). It
was therefore considered to be necessary to specifically degrade
1 and the structurally related natural compounds 2 and 3 to
smaller molecules meeting the above-mentioned prerequisites,
that is, (baseline) separation of the deuterium signals in the NMR
spectrum. The procedure applied in the present instances simply
consisted of the ozonolytic splitting of the double bonds of 1,
2, and 3, respectively, performed on suitably protected deriva-
tives. By these means, the molecules produced from the
permethylated form of 1 are anisaldehyde 4 and 3,5-dimethoxy-
benzaldehyde 5, respectively (Scheme 1). These two C-6-C-1
molecules are the same as those used in the chemical synthesis
of 1 (7-9). Thus, comparison of the deuterium content of 4
and 5 from extractives 1-3 with chemically identical samples
of synthetic origin would represent a comparison between
extractive and synthetic specimens.
In addition to natural abundance deuterium NMR studies,
δ18O measurements were also carried out on 4 and 5 of natural
and synthetic derivation for characterization purposes. The
studies were done on specific degradation products of extractives
1-3, allowing the assignment of the positional δ18O values of
the oxygen atoms attached to the sp2 carbon atoms of these
phenolic products. Despite the fact that isotopic oxygen deple-
tion measurements are considered to be of paramount impor-
tance in defining the mechanism of natural phenomena in earth
sciences (11), a limited number of applications of this diagnostic
tool in the definition of the origin of oxygenated organic
molecules have been reported to date. This might be due to the
fact that in the case of polyoxygenated compounds only

2750 J. Agric. Food Chem., Vol. 50, No. 10, 2002
Fronza et al.
phenyl)-1-chloropropane 6, 1.2 g (78%), purified by bulb-to-bulb
vacuum distillation, 99% by GC-MS: 1H NMR (500 MHz, CDCl3) δ
7.7 (2H, d, J ) 8.8 Hz), 6.81 (2H, d, J ) 8.8 Hz), 3.74 (3H, s, OCH3),
3.47 (2H, t, J ) 6.7 Hz), 2.68 (2H, t, J ) 7.3 Hz), 2.03 (2H, m).
By repeating the same sequence, starting from 3,5-dimethoxybenz-
aldehyde 5 was obtained 3-(3,5-dimethoxyphenyl)-1-chloropropane 7,
99% by GC-MS: 1H NMR (500 MHz, CDCl3) δ 6.36 (2H, d, J ) 2.1
Hz), 6.32 (1H, t, J ) 2.1 Hz), 3.78 (3H, s, OCH3), 3.53 (2H, t, J ) 6.4
Hz), 2.72 (2H, t, J ) 7.3 Hz, 2.08 (2H, m).
(c) 3,5-Dimethoxytoluene 9. 3,5-Dimethoxybenzaldehyde 5 (1.7 g;
0.01 mol), in 25 mL of absolute ethanol, was hydrogenated at room
temperature in the presence of 10% Pd/C (0.5 g). At the end of the
absorption, the filtered solution was evaporated and the residue distilled
bulb-to-bulb under vacuum to provide 3,5-dimethoxytoluene 9, 0.8 g
(87%), 99% by GC-MS.
Degradation of Commercial Rhapontin. The commercial product
(50 g) in 200 mL of pyridine was treated at 0 °C with 300 mL of
acetic anhydride. After 24 h at room temperature, the reaction mixture
was poured into ice water. The semisolid precipitate was decanted and
taken up into AcOEt (500 mL). The organic phase was washed
repeatedly with dilute HCl, NaHCO3, and brine. The residue obtained
upon evaporation of the dried solution was chromatographed on SiO2
eluting with ∼25% AcOEt in hexane the pentaacetate of the â-gluco-
pyranoside of 4′-O-methylresveratrol (pentaacetate of 2), 12.2 g
(20%): 1H NMR (500 MHz, CDCl3) δ 7.43 (2H, d, J ) 8.6 Hz), 7.02
(1H, d, J ) 16.3 Hz), 6.96 (2H, m), 6.90 (2H, d, J ) 8.6 Hz), 6.86
(1H, d, J ) 16.3 Hz), 6.62 (1H, t, J ) 2.1 Hz), 5.32-5.26 (2H, m),
5.18-5.11 (2H, m), 4.28 (1H, dd, J ) 5.8 and 12.4 Hz), 4.18 (1H, dd,
J ) 2.4 and 12.4 Hz), 3.90 (1H, ddd, J ) 2.4, 5.4, and 9.9 Hz), 3.83
(3H, s, OCH3), 2.30 (3H, s, COCH3), 2.07 (3H, s, COCH3), 2.056 (3H,
s,COCH3), 2.055 (3H, s, COCH3), 2.04 (3H, s, COCH3). The amount
of AcOEt in the eluting mixture was then increased, yielding the
hexaacetate of rhapontin 3, 30 g (62%): 1H NMR (500 MHz, CDCl3)
δ 7.29 (1H, dd, J ) 2.2 and 8.5 Hz), 7.21 (1H, d, J ) 2.2 Hz), 6.97
(1H, d, J ) 16.3 Hz), 6.96-6.93 (3H, m), 6.84 (1H, d, J ) 16.3 Hz),
6.63 (1H, t, J ) 2.1 Hz), 5.32-5.26 (2H, m), 5.16 (1H, t, J ) 9.4 Hz),
5.12 (1H, d, J ) 7.4 Hz), 4.28 (1H, dd, J ) 5.7 and 12.3 Hz), 4.18
(1H, dd, J ) 2.4 and 12.3 Hz), 3.90 (1H, ddd, J ) 2.4, 5.6, and 10.0
Hz), 3.85 (3H, s, OCH3), 2.33 (3H, s, COCH3), 2.30 (3H, s, COCH3),
2.07 (3H, s, COCH3), 2.06 (3, s, COCH3), 2.05 (3H, s, COCH3), 2.04
(3H, s, COCH3).
6.25 (1H, t, J ) 2.1 Hz), 3.84 (3H, s, OCH3)]. Products 11 and 12
were converted, in separate experiments, into monooxygenated 1-(4-
methoxyphenyl)-2-phenylethane 13 by first submitting the materials
to catalytic hydrogenation, which saturated the double bond, followed
by deoxygenation of the phenolic moieties by conversion into the
corresponding phosphate ester, followed by cleavage with Li metal in
liquid ammonia, operating exactly as described in the deoxygenation
of vanillin (13), in ∼35% overall yield: 1H NMR (500 MHz, CDCl3)
δ 7.33 (2H, m), 7.25 (3H, m), 7.15 (2H, d, J ) 8.9 Hz), 6.88 (2H, d,
J ) 8.9 Hz), 3.85 (3H, s, OCH3), 2.94 (4H, m).
Conversion of Acetylisovanillin 10 into 3-(3-Hydroxy-4-methoxy-
phenyl)-1-chloropropane 14. Acetylisovanillin 10 was converted into
3-(3-hydroxy-4-methoxyphenyl)-1-chloropropane 14 by first reacting
the aldehyde with carbethoxymethylenetriphenylphosphorane and con-
verting the ethyl 3-acetoxy-4-methoxycinnamate obtained into 14 as
described above for anisaldehyde and 3,5-dimethoxybenzaldehyde,
respectively. The overall yield of 14 was ∼30%, 99% by GC-MS: 1H
NMR (500 MHz, CDCl3) δ 6.78 (1H, dd, J ) 1.8 and 8.1 Hz), 6.78
(1H, d, J ) 8.1 Hz), 6.67 (1H, dd, J ) 1.8 and 8.1 Hz), 5.60 (1H, s,
OH), 3.87 (3H, s, OCH3), 3.51 (2H, t, J ) 6.4 Hz), 2.68 (2H, t, J )
7.3 Hz), 2.04 (2H, m).
Natural Abundance Deuterium Nuclear Magnetic Resonance
Measurements. Deuterium NMR data (46.076 MHz) were recorded
at 308 K on a Bruker AC300 spectrometer equipped with a process
controller, a 10 mm selective deuterium probehead, and a 19F lock
channel, under broad-band proton decoupling conditions.
Samples (0.3-0.9 g), hexafluorobenzene for 19F lock (150 mg,
Merck, Darmstadt, Germany) and tetramethylurea as internal (D/H)
standard (50-150 mg), were carefully weighed directly into the NMR
tube. Then 3 mL of acetonitrile was added as a solvent, and the tube
was warmed and shaken until complete dissolution. For isovanillin
samples, a 50:50 mixture of acetone/acetonitrile was used. Tetramethyl-
urea (TMU) was a BCR reference product certified for a (D/H) value
of 136.67 ppm.
Eight to 10 spectra were run for each sample, collecting 1600-
2000 scans and using the following parameters: 6.8 s acquisition time,
0.05 s relaxation delay, 1200 Hz spectral width, 16K memory size,
and 15 µs (90°) pulse length. Each FID was Fourier transformed with
no zero filling (0.15 Hz/point digital resolution) and line broadening
of 2 Hz.
Bruker WIN NMR software was used for FID processing and peak
deconvolution.
(a) Ozonolysis of the Pentaacetate of 2 and of the Hexaacetate of
3: Anisaldehyde 4 and AcetylisoVanillin 10. The pentaacetate of 2 (10
g; 0.015 mol) was submitted to the action of ozonized oxygen, as
indicated above, eventually obtaining, after chromatographic separation,
anisaldehyde 4, 1.4 g (67%), and the 3,5-disubstituted benzaldehyde
9, 4.5 g (60%): 1H NMR (500 MHz, CDCl3) δ 9.92 (1H, s, CHO),
7.37 (1H, dd, J ) 1.2 and 2.2 Hz), 7.34 (1H, dd, J ) 1.2 and 2.1 Hz),
7.01 (1H, t, J ) 2.2 Hz), 5.32-5.26 (2H, m), 5.18-5.12 (2H, m), 4.23
(1H, dd, J ) 5.5 and 12.2 Hz), 4.19 (1H, dd, J ) 2.6 and 12.2 Hz),
3.93 (1H, ddd, J ) 2.4, 5.6, and 9.9 Hz), 2.32 (3H, s, COCH3), 2.09
(3H, s, COCH3), 2.06 (3, s, COCH3), 2.05 (3H, s, COCH3), 2.03 (3H,
s, COCH3). When this sequence was applied to the hexaacetate of
rhapontin 3, acetylisovanillin 10 and the aldehyde 9 were obtained in
62 and 59% yields, respectively.
Internal isotopic ratios are
Rij ) njSi/Sj
(1)
where Si is the area of the ith site, Sj is the area of a reference peak,
and nj is the number of isochronous hydrogens at the j site. A statistical
distribution of deuterium among the n molecular sites would originate
Rij factors equal to the number of hydrogens of the corresponding ith
sites.
The absolute values of the site-specific (D/H) ratios were calculated
according to the formula (8)
(D/H)i ) nWSgWS(MW)LSi(D/H)WS/nigL(MWWS)SWSPL (2)
(b) Enzymic Hydrolysis of Commercial Rhapontin: 1-(4-Methoxy-
phenyl)-2-phenylethane 13. A mixture was made up composed of 18 g
of commercial rhapontin, 1 g of emulsin (Fluka, Milano, Italy), distilled
water (1 L), and acetone (0.1 L) at pH 6. After 24 h of stirring at room
temperature, an additional quantity (0.5 g) of emulsin was added, and
the incubation was continued for a further 48 h. The reaction mixture
was extracted with AcOEt (3 × 400 mL). The residue obtained upon
evaporation of the organic phase was chromatographed on SiO2, giving
4′-O-methylresveratrol 11, 1.3 g (60%) [1H NMR (500 MHz, acetone-
d6) δ 8.11 (2H, s, 2 OH), 7.50 (2H, d, J ) 8.6 Hz), 7.03 (H, d, J )
16.3 Hz), 6.91 (1H, d, J ) 16.3 Hz), 6.91 (2H, d, J ) 8.6 Hz), 6.54
(2H, d, J ) 2.1 Hz), 6.27 (1H, t, J ) 2.1 Hz), 3.80 (3, s, OCH3)] and
rhapontinin 12, 5 g (56%) [1H NMR (500 MHz, acetone-d6) δ 8.38
(2H, s br, 2 OH), 7.66 (1H, s br, OH), 7.08 (1H, d, J ) 2.0 Hz), 6.97
(1H, d, J ) 16.3 Hz), 6.97 (1H, dd, J ) 2.1 and 8.4 Hz), 6.91 (1H, d,
J ) 8.4 Hz), 6.88 (1H, d, J ) 16.3 Hz), 6.53 (2H, d, J ) 2.1 Hz),
where WS stands for the working standard (TMU) with a known isotope
ratio (D/H)WS and L for the product under examination; nWS and ni
are the number of equivalent deuterium atoms of TMU and of the ith
peak, respectively; gWS and gL are the weights of the standard and the
sample, respectively; MWL and MWWS are the corresponding molecular
weights; Si and SWS are the areas of the ith peak and of the standard,
respectively; PL is the purity of the sample, measured by capillary gas
chromatography. (D/H)WS is the working standard isotope ratio as
determined by isotope ratio mass spectrometry on the Vienna Standard
Ocean Water (V-SMOW) scale (14).
Measurement of the 18O/16O Isotopic Ratios. The determination
of the relative stable isotopic ratio was carried out using a Finnigan
MAT Delta S mass spectrometer coupled with a pyrolysis unit including
a commercially available pyrolysis system (Leco VTF900, St. Joseph,
MI), fitted with a Carlo Erba S-200LS autosampler (ThermoQuest, San

Natural Abundance 2H and 18O Studies on Resveratrol
J. Agric. Food Chem., Vol. 50, No. 10, 2002 2751
Scheme 2. Degradation of the â-Glucopyranoside of 4′-O-Methylresveratrol 2 and of Rhapontin 3 to the Aldehydes 4 and 10 for the (D/H)
Measurements and to Compounds 13 and 14 for the Positional 18O Isotopic Determinationa
a (i) Ac2O/Pyr; (ii) O /CH Cl2−MeOH 1:1, −78 °C; (iii) Ph3P; (iv) emulsin, H O, pH 5.6; (v) H /Pd/C; (vi) (EtO)2POCl/NaOH; (vii) Li/NH (l).
3
2
2
2
3
Jose, CA). The analytical procedures for the determination of the 18O/
16O ratio have been described in a previous study (15).
The values are expressed in the δ per mil (‰) scale versus the
international standard V-SMOW (14), using the formula
δ18O (‰)V-SMOW ) [(Rsample/RV-SMOW) - 1] × 103
(3)
where R )18O/16O. The reproducibility of the measurements was
estimated within (0.7‰.
RESULTS AND DISCUSSION
Resveratrol 1 was first isolated over 60 years ago from
Veratrum grandiflorum by Takaoka (2), who confirmed the
structure through chemical synthesis from 4-hydroxybenz-
aldehyde and 3,5-dihydroxyphenylacetic acid. More recently,
several additional more practical syntheses of 1 were reported
(7-9) in which the C6-C2-C6 skeleton is constructed by
coupling C6-C1 synthons derived from anisaldehyde 4 and
3,5-dimethoxybenzaldehyde 5, respectively. The present study
concerns the isotopic characterization of extracted resveratrol
from P. cuspidatum and of a commercial sample of rhapontin
from Rheum (3). The latter material contains, in addition to
rhapontin 3, ∼20% of product 2, that is, the â-glucopyranoside
of 4′-O-methylresveratrol. In addition to these natural stilbenes,
a sample of extracted C6-C1 anisaldehyde 4 was also examined,
together with synthetic samples of 4, 3,5-dimethoxybenzalde-
hyde 5, and acetyl isovanillin 10, obtained in an unspecified
manner.
Resveratrol 1 and the related products 2 and 3, because they
are highly unsaturated compounds, as indicated above, have
NMR spectra not suitable for the determination of the deuterium
content at natural abundance. Their degradation to smaller
molecules using procedures not affecting the isotopic content
was therefore necessary. The pathway followed involves cleav-
age with ozone of the double bond of suitably protected
derivatives of 1-3, which provides the C6-C1 aldehydes 4, 5,
9, and 10, retaining all of the hydrogen atoms originally present
in the parent compounds. The cleavage with ozone of the
ethylenic substrates was quantitative, and in the Ph3P-mediated
decomposition of the intermediate ozonides there is no formation
of either the acids or the alcohols corresponding to the aldehydes
4, 5, 9, and 10. To avoid the danger of isotope fractionation
during the purification processes, the materials were purified
by chromatography, pooling all of the identical fractions.
Figure 1. Natural abundance deuterium NMR spectra of anisaldehyde 4:
(A) synthetic from BBE; (B) natural from resveratrol 1. The signal at 2.8
ppm is due to TMU (calibrated tetramethylurea used as internal reference).
Extracted resveratrol 1 was thus converted into the corre-
sponding tri-O-methyl ether by treatment with ethereal diazo-
methane. Ozonolysis of the latter at low temperature, followed
by triphenylphosphine treatment, yielded anisaldehyde 4
and 3,5-dimethoxybenzaldehyde 5, respectively, separated by
column chromatography (Scheme 1). Anisaldehyde 4 was
similarly obtained by ozonolysis of the acetylated form of
the â-glucoside 2. In this instance, the crystalline aldehyde 9
was also isolated. Similarly, rhapontin 3, once acetylated,
on ozonolysis afforded acetylisovanillin 10 and, as for 2, the
aldehyde 9 (Scheme 2).
The natural abundance deuterium NMR spectra of anisalde-
hyde 4, obtained from tri-O-methylresveratrol (Figure 1B) and
from the peracetylated form of the glucoside 2, respectively,
were thus acquired and compared with those of a sample of
natural anisaldehyde of extractive origin and of two synthetic
samples (Figure 1A), respectively. Inspection of the results
(Table 1) indicated in the synthetic series (D/H)2 > (D/H)3,
whereas the reverse is true in the natural series (entries 3-5).
However, for entries 1-4 the (D/H)2 values are included
between 140 and 147 ppm, whereas the value relative to 4 (entry
5) obtained from 2 is to some extent lower (129 ppm). The
main numerical difference between the two series is in the (D/
H)3 values, which range from 135 to 130 ppm for entries 1 and
2 and up to 199, 195, and 167 ppm, respectively, for entries
3-5. Within the two sets, the difference in the (D/H) values of

2752 J. Agric. Food Chem., Vol. 50, No. 10, 2002
Fronza et al.
Table 3. (D/H) Isotopic Ratiosa of Acetylisovanillin 10 of Different
Table 1. (D/H) Isotopic Ratiosa of Anisaldehyde 4 of Different Origin
i
i
Origin
entry
origin
synthetic (Aldrich)
(D/H)
(D/H)2
(D/H)3
formyl
entry
origin
synthetic (Fluka)
natural (ex rhapontin 3 130 (27)
from Rheum)
(D/H)
(D/H)6
(D/H)2
(D/H)5
formyl
1
2
3
288 (17)
268 (12)
184 (9)
147 (8) 135 (7)
140 (6) 130 (6)
145 (9) 199 (10)
synthetic (BBE)
1
2
319 (15)
142 (14) 160 (8)
155 (15) 145 (17) 173 (32)
131 (5)
natural (ex resveratrol 1
from P. cuspidatum)
natural (Robertet)
natural (ex resveratrol glucoside 2
from Rheum)
4
5
133 (5)
110 (7)
142 (6) 195 (5)
129 (5) 167 (5)
a (D/H) values are expressed in ppm; the standard deviations are reported in
parentheses.
i
a (D/H) values are expressed in ppm; the standard deviations are reported in
parentheses.
i
Table 4. (D/H) and R Isotopic Ratiosa of Isovanillin of Different Origin
i
entry
origin
(D/H)
(D/H)2
(D/H)5
R
R
2,5
formyl
formyl,5
Table 2. (D/H) Isotopic Ratiosa of 3,5-Dimethoxybenzaldehyde 5 of
Different Origin
i
1
2
synthetic (Fluka)
272 (16) 119 (40) 138 (16) 1.96 1.73
natural (ex neohesperidin 105 (8) 143 (6) 157 (16) 0.67 1.83
from Citrus)
entry
origin
(D/H)
(D/H)2
(D/H)4
formyl
a (D/H) values are expressed in ppm; the standard deviations are reported in
parentheses.
i
1
2
3
synthetic (Aldrich)
synthetic (Fluka)
natural (ex resveratrol 1
153 (6)
130 (7)
96 (10)
144 (6)
129 (4)
140 (10)
129 (18)
134 (13)
134 (18)
from P. cuspidatum)
analysis was made in comparison with a synthetic sample. The
S/N ratio in the NMR spectra of 10 from rhapontin was rather
low, owing to the low sample quantity, so standard deviations
were quite high. However, Table 3 shows clear differences
between the two origins. Comparison of the data again indicates
as the major element of distinction the formyl site enrichment,
being 319 ppm for the synthetic material and only 130 ppm for
the product isolated from 3.
a (D/H) values are expressed in ppm; the standard deviations are reported in
i
parentheses.
the aldehydic moiety is significant, going down from 288 and
268 ppm (synthetic products, entries 1 and 2) to 184, 133, and
110 ppm (natural derivation, entries 3-5), respectively.
The deuterium labeling pattern of the aromatic part of 1 and
2, as measured on anisaldehyde 4, is similar to that of raspberry
ketone [4-(p-hydroxyphenyl)butan-2-one] from Taxus baccata
and Aeonium lindleyi (13, 14) and is in perfect agreement with
measurements made on 4-hydroxybenzaldehyde accompanying
vanillin in Vanilla pods (18). More problematic is the interpreta-
tion of the formyl deuterium enrichment values of anisaldehyde
4 from 1 and 2 in comparison with those of vanillin and
4-hydroxybenzaldehyde from Vanilla (18) and of benzaldehyde
from different natural sources (19, 20). The deuterium content
of 4 obtained from 1 (entry 3) is very different from that
measured on 4 obtained from 2 (entry 5). The same difference
was measured between vanillin and 4-hydroxybenzaldehyde
from Vanilla (18). Aldehyde 4 retains, in the formyl moiety,
the hydrogen atom of the stilbene framework of 1 and 2. This
hydrogen originates from position 3 of the biosynthetic precursor
p-coumaric acid, and thus the formyl deuterium values of entries
3 and 5 seemingly should reflect the differences in the deuterium
enrichment at the corresponding position of p-coumaric acid.
It follows from these observations that both the aromatic and
the methine moieties of resveratrol 1 from P. cuspidatum and
of its glucosidated 4′-O-methyl ether 2 from Rheum possess a
different mean deuterium enrichment, the major element of
distinction being the position R to the p-substituted ring.
The (D/H) values of 3,5-dimethoxybenzaldehyde 5, obtained
from permethylated 1, and of two synthetic samples (Table 2),
are nearly indentical as far as the aromatic positions are
concerned, whereas the (D/H)formyl of the material from res-
veratrol 1 (entry 3) is 96 ppm, significantly lower with respect
to those (153 and 130 ppm, respectively) of the synthetic
counterparts (entries 1 and 2). Following the above reasoning,
the formyl hydrogen of 5 obtained from resveratrol 1 reflects
the deuterium enrichment at position 2 of the side chain of
p-coumaric acid. Unfortunately, aldehyde 9 from acetylated 2
gave an NMR spectrum unsuitable for a quantification of the
deuterium at the position of interest, that is, the formyl moiety.
Rhapontin 3 was then submitted to the above degradative
sequence. Its hexaacetate, with ozone, provided almost quan-
titatively aldehyde 9 and acetylisovanillin 10 (Scheme 2). NMR
Finally, for comparison purposes, isovanillin from the basic
cleavage of neohesperidin from unripe bitter orange was also
submitted to deuterium NMR studies, in comparison with a
synthetic sample. Unfortunately, owing to the low solubility,
part of the sample crystallized at the bottom of the NMR tube
during measurements, so in this case we prefer to rely on internal
ratios R and not on (D/H). The (D/H) and the R values of
synthetic and “extractive” isovanillin are reported in Table 4.
Rformyl,5 is the ratio between the formyl group and meta hydrogen
enrichment. R2,5 is the ratio between ortho and meta hydrogen
enrichment. Rformyl,5 shows a strong depletion of deuterium in
the formyl site of the natural sample. The same trend is also
observed on (D/H)formyl of the “natural” material, which is much
lower with respect to that of the correponding synthetic
counterparts, as occurs in the case of anisaldehyde 4.
Isovanillin is obtained from neohesperidin through basic
treatment, the latest stage of the sequence being a retro-aldol
C-C cleavage similar to that occurring in the generation of
benzaldehyde from cinnamaldehyde (19). From a chemical point
of view, it is expected that the basic treatment may cause
equilibration with the water hydrogen atoms of the aromatic
protons in positions ortho and para to the phenolic oxygen,
whereas the hydrogen of the formyl group of isovanillin should
be the one present in position 3 of the chain of p-coumaric acid
as in the case of 1-3.
The positional δ18O values of 1 and that of the aglyconic
part of 2 and 3 were finally determined. This goal was achieved
by measuring the global values of a set of products derived
from the above materials by selective deoxygenation. To avoid
the handling of compounds unsuitable for the measurements
due to high volatility, anisaldehyde 4 and 3,5-dimethoxy-
benzaldehyde 5, obtained from resveratrol 1, were converted
to the compounds 6 and 7, respectively, which retain only
the original oxygen atom in the aromatic ring and possess
suitable molecular characteristics. Thus, extractive resveratrol
1, possessing δ18O ) +11.8‰, was converted, via 4 and 5
(Scheme 1), into 3-(4-methoxyphenyl)-1-chloropropane 6,
δ18O ) +11.5‰, and 4-(3,5-dimethoxyphenyl)-1-chloropropane

Natural Abundance 2H and 18O Studies on Resveratrol
J. Agric. Food Chem., Vol. 50, No. 10, 2002 2753
Chart 1. Positional δ18O Values of Resveratrol 1 Isolated from P. cuspidatum and of 4′-O-Methylresveratrol 11 and Rhapontinin 12 Obtained by
Enzyme Hydrolysis of the Corresponding Glucosides from Rheum
7, δ18O ) +15.2‰. Moreover, 3,5-dimethoxybenzaldehyde 5
derived from 1 was converted by catalytic hydrogenation into
3,5-dimethoxytoluene 8 with δ18O ) +13.2‰ (Scheme 1). The
latter material and the chloro derivative 7 retain the two oxygen
atoms at positions 3 and 5 of resveratrol 1 (mean δ18O )
+14.2‰), whereas 6 conserves the phenolic oxygen at position
4′ of the stilbene moiety (Scheme 1).
In conclusion, the approach to the stable isotope characteriza-
tion of the practically important food components 1-3 based
on selective degradation, followed by isotopic measurements
of the derived fragments, seems to be satisfactory. Indeed, by
ozonolytic splitting of the unsaturated products a set of simple
aldehydes are obtained, on which the deuterium analysis can
be performed quite easily. Additionally, the acquisition of the
positional δ18O values of selected degradation products might
be of substantial utility for identification purposes. Additionally
this kind of analysis, at variance with the deuterium NMR,
requires only a minute amount (few milligrams) of material.
The progressive deoxygenation procedure here applied allows
the positional rather than global δ18O values to be obtained,
thus dramatically enlarging the diagnostic capacity of the
investigation.
The determination of the positional δ18O values of 2 and 3
required a different degradation sequence. To start, commercial
rhapontin was submitted to hydrolysis in the presence of
emulsin. The reaction is very slow, but eventually the aglycons
11 and 12, later separated by column chromatography, were
obtained. 4-O-Methylresveratrol 11 showed δ18O ) +8.5‰,
whereas the value for rhapontinin 12 was +9.4‰. Subsequently,
diphenol 11 and triphenol 12, in separate experiments, were
submitted to selective deoxygenation by means of cleavage with
Li metal in liquid ammonia of the corresponding phosphate
esters (Scheme 2), both yielding monooxygenated 1-(4-methoxy-
phenyl)-2-phenylethane 13. The product from 2 showed
δ18O ) +1.8‰, whereas the identical material from 3 possessed
a value of +6.7‰. Moreover, acetylisovanillin 10, obtained
from rhapontin 3, was converted by chain elongation, reduction,
and transformation of the primary alcohol into the chloride 14
(Scheme 2), showing a δ18O value of +9.2‰.
From consideration of the values relative to 13 and 14 it
follows, by calculation, that the oxygen at position 3′ of 3
possesses a δ18O value of +11.7‰. Moreover, when the global
values of 12-14 are considered, the mean δ18O value for the
oxygen atoms at positions 3 and 5 of 12 results in a value ca.
+9.6‰. In summary, it follows that the δ18O values of the
aromatic oxygen atom at position 4′ of 1, 2, and 3 are +11.5,
+1.8, and +6.7‰, respectively. The oxygen atoms in the
malonate-derived part of these molecules, that is, those at
positions 3 and 5, show, for the same compounds, mean values
of +14.2, +11.8, and +9.6‰, respectively. Finally, the oxygen
atom at position 3′ of rhapontin 3 shows a value of +11.7‰
(Chart 1).
LITERATURE CITED
(1) Schmidt, H.-L.; Kexel, H.; Butzenlechner, M.; Schwarz, S.;
Gleixner, G.; Thimet, S.; Werner, R. A.; Gensler, M. In Stable
Isotopes in the Biosphere; Wada, E., Yoneyama, T., Minagawa,
M., Ando, T., Fry, B. D., Eds.; Kyoto University Press: Kyoto,
Japan, 1995; pp 17-35.
(2) Takaoka, M. Die Synthese von Resveratrol und seinen Derivaten.
U¨ ber die phenolischen Substanzen der weissen Nieswurz
(Veratrum grandifolium Loes. Fil). Proc. Imp. Acad. (Tokyo)
1940, 16, 405-407; Chem. Zentralbl. 1941, 2117.
(3) Ruszkowski, M. Beitrag zum Studium in Polen kultivierten
Rhabarbers. Chem. Zentralbl. 1930, 1339.
(4) Gorham, J. The Stilbenoids. Prog. Phytochem. 1980, 6, 203-
252.
(5) Soleas, G. J.; Diamandis, E. P.; Goldberg, D. M. Resveratrol:
A molecule whose time has come? And gone? Clin. Biochem.
1997, 30, 91-113.
(6) Kubo, M.; Kimura, Y.; Shin, H.; Haneda, T.; Tani, T.; Namba,
K. Studies on the antifungal substances of crude drug (II). On
the roots of Polygonum cuspidatum Sieb. Et Zuc. (Polygonaceae).
Shoyakugaky Zasshi 1981, 35, 58-61.
(7) Moreno-Manas, M.; Pleixatas, R. Dehydroacetic acid chemistry.
A new synthesis of resveratrol, a photoalexine of Vitis Vinifera.
Anal. Quim. 1985, 81, 157-161.
(8) Alonso, E.; Ra`mon, D. J.; Yus, M. Simple synthesis of
5-substituted resorcinols: a revisited family of interesting
bioactive molecules. J. Org. Chem. 1997, 62, 417-421.
(9) Meier, H.; Dullweber, U. Extension of the squaraine chromo-
phore in symmetrical bis(stilbenyl)squaraines. J. Org. Chem.
1997, 62, 4821-4826.
(10) Martin, G. J.; Martin, M. L. Stable isotope analysis of food and
beverages by NMR spectroscopy. Annu. Rep. NMR Spectrosc.
1995, 31, 91-104.
The selected oxygen depletion figures found for the oxygen
atoms located at different positions of the examined stilbenes
have been, in part, calculated on the basis of the biochemical
mechanisms for natural formation of oxygenated organic
molecules (12). The oxygen depletion values for position 4′ of
1-3 can be compared with those obtained for structurally related
plant phenolics of relevance in food chemistry. The δ18O value
(6.7‰) of rhapontin 3 is in line with those found for the
corresponding position of vanillin from Vanilla pods and from
lignin (∼6.5‰) (15). The one from resveratrol 1 (+11.5‰) fits
with those observed in the Citrus glycosides naringin and
(11) Thiemens, M. H.; Savarino, J.; Farquhar, J.; Bao, H. Mass-
independent isotopic compositions in terrestrial and extra-
terrestrial solids and their applications. Acc. Chem. Res. 2001,
34, 645-652.
neohesperidin (14 and 11‰, respectively) (14). Finally, the 18
O
content of the phenol of 2 (+1.8‰) is in line with those
observed in the two specimens of raspberry ketone from T.
baccata and A. lindleyi examined so far (-0.8 and +0.6‰)
(13, 14).
(12) Schmidt, H.-L.; Werner, R. A.; Rossmann, A. 18O Pattern and
biosynthesis of natural plant products. Phytochemistry 2001, 58,
9-32.

2754 J. Agric. Food Chem., Vol. 50, No. 10, 2002
Fronza et al.
(13) Fronza, G.; Fuganti, C.; Pedrocchi-Fantoni, G.; Serra, S.; Zucchi,
G.; Fauhl, C.; Guillou, C.; Reniero, F. Stable isotope character-
ization of raspberry ketone extracted from Taxus baccata and
obtained by oxidation of the accompanying alcohol (betuligenol).
J. Agric. Food Chem. 1999, 47, 1150-1155.
and extracts: Application of the SNIF-NMR method to vanillin
and p-hydroxybenzaldehyde. J. Agric. Food Chem. 1997, 45,
859-866.
(19) Hagedorn, M. L. Differentiation of natural and synthetic benz-
aldehydes by 2H nuclear magnetic resonance. J. Agric. Food
Chem. 1992, 40, 634-637.
(14) Unpublished results.
(15) Fronza, G.; Fuganti, C.; Serra, S.; Burke, A.; Guillou, C.; Reniero,
F. The positional δ18O values of extracted and synthetic vanillin.
HelV. Chim. Acta 2001, 84, 351-359.
(16) Gonfiantini, R. Standards for stable isotope measurements in
natural compounds. Nature 1978, 271, 534-536.
(20) Remaud, G.; Debon, A. A.; Martin, Y. L.; Martin, G. G.; Martin
G. J. Authentication of bitter almond oil and cinnamon oil:
Application of the SNIF-NMR method to benzaldehyde. J. Agric.
Food Chem. 1997, 45, 1042-4048.
(17) Koziet, J. Isotope ratio mass spectrometric method for the on-
line determination of oxygen-18 in organic matter. J. Mass
Spectrom. 1997, 32, 103-108.
(18) Remaud, G. S.; Martin, Y. L.; Martin, G. G.; Martin, G. J.
Detection of sophisticated adulterations of natural vanilla flavors
Received for review August 14, 2001. Revised manuscript received
February 13, 2002. Accepted February 14, 2002. We thank CNR
Progetto Finalizzato Biotecnologie for financial support.
JF011103J