An unusual decarboxylative maillard reaction between L-DOPA and D-glucose under biomimetic conditions: Factors governing competition with Pictet-Spengler condensation

Article abstract of DOI:10.1021/jo010078d

In 0.1 M phosphate buffer at pH 7.4 and 37°C, the tyrosine metabolite L-3,4-dihydroxyphenylalanine (L-DOPA) reacts smoothly with D-glucose to afford, besides diastereoisomeric tetrahydroisoquinolines 1 and 2 by Pictet-Spengler condensation, a main product shown to be the unexpected decarboxylated Amadori compound N-(1-deoxy-D-fructos-1-yl)-dopamine (3). Under similar conditions, dopamine gave only tetrahydroisoquinoline products 4 and 5, whereas L-tyrosine gave exclusively the typical Amadori compound 6. Fe³⁺ and Cu²⁺ ions, which accumulate in relatively high levels in parkinsonian substantia nigra, both inhibited the formation of 3. Cu²⁺ ions also inhibited the formation of 1 and 2 to a similar degree, whereas Fe³⁺ ions increased the yields of 1 and 2. Apparently, the formation of 3 would not be compatible with a simple decarboxylation of the initial Schiff base adduct, but would rather involve the decarboxylative decomposition of a putative oxazolidine-5-one intermediate assisted by the catechol ring. These results report the first decarboxylative Maillard reaction between an amino acid and a carbohydrate under biomimetic conditions and highlight the critical role of transition metal ions in the competition with Pictet-Spengler condensation.

Full text of DOI:10.1021/jo010078d

5048
J . Org. Chem. 2001, 66, 5048-5053
An Un u su a l Deca r boxyla tive Ma illa r d Rea ction betw een L-DOP A
a n d D-Glu cose u n d er Biom im etic Con d ition s: F a ctor s Gover n in g
Com p etition w ith P ictet-Sp en gler Con d en sa tion
Paola Manini, Marco d’Ischia, and Giuseppe Prota*
Department of Organic Chemistry and Biochemistry, University of Naples Federico II,
Complesso Universitario di Monte S. Angelo, Via Cinthia 4, I-80126 Naples, Italy
prota@cds.unina.it
Received J anuary 22, 2001
In 0.1 M phosphate buffer at pH 7.4 and 37 °C, the tyrosine metabolite L-3,4-dihydroxyphenylalanine
(L-DOPA) reacts smoothly with D-glucose to afford, besides diastereoisomeric tetrahydroisoquinolines
1 and 2 by Pictet-Spengler condensation, a main product shown to be the unexpected decarboxylated
Amadori compound N-(1-deoxy-D-fructos-1-yl)-dopamine (3). Under similar conditions, dopamine
gave only tetrahydroisoquinoline products 4 and 5, whereas L-tyrosine gave exclusively the typical
Amadori compound 6. Fe3+ and Cu2+ ions, which accumulate in relatively high levels in parkinsonian
substantia nigra, both inhibited the formation of 3. Cu2+ ions also inhibited the formation of 1 and
2 to a similar degree, whereas Fe3+ ions increased the yields of 1 and 2. Apparently, the formation
of 3 would not be compatible with a simple decarboxylation of the initial Schiff base adduct, but
would rather involve the decarboxylative decomposition of a putative oxazolidine-5-one intermediate
assisted by the catechol ring. These results report the first decarboxylative Maillard reaction between
an amino acid and a carbohydrate under biomimetic conditions and highlight the critical role of
transition metal ions in the competition with Pictet-Spengler condensation.
In tr od u ction
including alcoholism,3 phenylketonuria,4 and neurode-
generative disorders such as Parkinson’s disease.5 In this
framework, a plausible class of target compounds for
L-DOPA is provided by carbohydrate-related metabolites,
which play a very prominent role in neuron energetics
and functioning. Glucose, the brain’s source of energy,
and other hexoses are transported across the blood brain
barrier and may react with L-DOPA, usually adminis-
tered at relatively high doses, and other catecholamines
to give tetrahydroisoquinolines and other potentially
neurotoxic species that could exacerbate neuronal dam-
age in the long term. Along this line of thought, we
recently demonstrated that L-DOPA6 as well as the
catecholamine neurotransmitter dopamine7 reacts under
physiologically relevant conditions with D-glyceraldehyde,
the simplest member of the carbohydrate family, to give
stereoisomeric tetrahydroisoquinolines. Transition metal
cations of wide occurrence in biological systems, including
the parkinsonian substantia nigra,8 were found to cause
dychotomous effects on the reaction course: whereas Fe3+
accelerated tetrahydroisoquinoline formation, Cu2+ caused
an opposite and rate-decreasing effect. Overall, the
results were consistent with a mechanistic picture gov-
erned by Schiff base formation as the key event that is
L-3,4-Dihydroxyphenylalanine (L-DOPA), the primary
product of tyrosine metabolism in neural crest-derived
cells and the main percursor of catecholamine neu-
rotransmitters, has gained widespread credit over the
past decades as being the mainstay for the pharmacologi-
cal treatment of Parkinson’s disease,1 a disabling neu-
rodegenerative disorder characterized by selective loss
of pigmented dopaminergic neurons of the substantia
nigra. However, despite having initially beneficial effects,
in the long term this drug causes severe, adverse
(untoward) effects, including involuntary movements,
on-off fluctuation of effectiveness, and dyskinesias at
peak doses, possibly as a result of the hastened degen-
eration of nigrostriatal neurons.2 Although several hy-
potheses have been put forward, the chemical mecha-
nisms underlying these side effects have remained elusive;
consequently, progress in the development of an effective
therapeutic regimen based on L-DOPA has been consid-
erably slowed.
A possible contributory mechanism in L-DOPA-induced
side effects involves Pictet-Spengler condensation with
aldehyde metabolites leading to tetrahydroisoquinoline
derivatives and oxidation products thereof. Tetrahy-
droisoquinolines are generally regarded as neurotoxic
compounds and are putatively involved in a variety of
pathological conditions of the central nervous system,
(3) Yamanaka, Y.; Walsh, M. J .; Davis, V. E. Nature 1970, 227, 1143.
(4) Lasala, J . M.; Coscia, C. J . Science 1979, 203, 283.
(5) Sandler, M.; Bonham-Carter, S.; Hunter, K. R.; Stern, G. M.
Nature 1973, 241, 439.
(6) Manini, P.; d’Ischia, M.; Prota, G. Bioorg. Med. Chem. 2001, 9,
923.
(7) Manini, P.; d’Ischia, M.; Lanzetta, R.; Parrilli, M.; Prota, G.
Bioorg. Med. Chem. 1999, 7, 2525.
(8) (a) Dexter, D. T.; Carayon, A.; J avoy-Agid, F.; Agid, Y.; Wells,
F. R.; Daniel, S. E.; Lees, A. J .; J enner, P.; Marsden, C. D. Brain 1991,
114, 1953. (b) Riederer, P.; Sofic, E.; Rausch, W. D.; Schmidt, B.;
Reynolds, G. P.; J ellinger, K.; Youdim, M. B. H. J . Neurochem. 1989,
52, 515.
* To whom correspondence should be addressed. Phone: +39-081-
674128. Fax: +39-081-674393. E-mail: prota@cds.unina.it.
(1) (a) Napolitano, A.; Pezzella, A.; Misuraca, G.; Prota, G. Exp.
Opin. Ther. Patents 1998, 8, 1249. (b) Dunnett, S. B.; Bjo¨rklund, A.
Nature 1999, 399, A32. (c) Yahr, M. D. Neurology 1993, 60, 11.
(2) Fahn, S. In The Scientific Basis for the Treatment of Parkinson’s
Disease; Olanow, C. W., Lieberman, A. N., Eds.; Parthenon Publishing
Group: Carnforth, U.K., 1992; Vol. 89, p 111.
10.1021/jo010078d CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/23/2001

L-DOPA-D-Glucose Decarboxylative Maillard Reaction
J . Org. Chem., Vol. 66, No. 15, 2001 5049
Sch em e 1. Ma in P r od u cts Obta in ed by Rea ction
of L-DOP A a n d L-Tyr osin e w ith D-Glu cose
F igu r e 1. ROESY correlations for compound 3.
Besides 1 and 2, HPLC analysis of the reaction mixture
displayed another major condensation product (about
30% yield) that was subjected to extensive spectral
analysis and was shown to be the Amadori compound
N-(1-deoxy-D-fructos-1-yl)-dopamine (3) (Scheme 1). For-
mation of 3 is remarkable since it represents, to the best
of our knowledge, the first example of a decarboxylative
Amadori rearrangement involving D-glucose and an
amino acid. Indeed, the Amadori reaction, the initial step
in Maillard type condensations of amino acids14 and
proteins15 with carbohydrates, involves two sequential
tautomerization steps in converting an aldimine (the
Schiff base) into a ketoamine, which proceeds normally
without concomitant decarboxylation.
Diagnostic features in support of the proposed decar-
boxylated structure were the lack of the carboxyl carbon
resonance and the presence of four CH2 carbons versus
only three aliphatic CH signals. The noticeable finding
was that the product appeared to be in a single form, as
â-fructopyranose, at variance with the mixtures of forms
in equilibrium that are commonly reported for Amadori
compounds derived from D-glucose and amino acids.16
This fact confirms the suggested participation of the free
carboxy group in ring-opening and ring-closing reactions
of Amadori compounds.
followed by a finely modulated ring-closure reaction in
accordance with the Felkin-Anh model of asymmetric
induction.9
As an extension of that study, we have now investi-
gated the reaction of L-DOPA and dopamine with D-
glucose under physiologically relevant conditions to
specifically elucidate the nature of competing pathways,
the possible role of transition metal cations, and the
chemical viability of nonenzymatic glycation processes
in relation to the motor problems attending long-term
L-DOPA therapy.
Resu lts a n d Discu ssion
Rea ction of L-DOP A w ith D-Glu cose. In 0.1 M
phosphate buffer at pH 7.4 and 37 °C under an argon
atmosphere, L-DOPA (50 mM) reacted smoothly with
D-glucose (1.5 M) to give eventually three main products,
two of which were readily identified as (1R,1′S, 3S)-1-
(D-gluco-pentitol-1′-yl)-3-carboxy-6,7-dihydroxy-1,2,3,4-
tetrahydroisoquinoline (1) and (1S,1′S, 3S)-1-(D-gluco-
pentitol-1′-yl)-3-carboxy-6,7-dihydroxy-1,2,3,4-tetrahy-
droisoquinoline (2) (25% and 15% yields, respectively),
differing exclusively in the configuration of the stereo-
genic center at C-1. Detailed inspection of the NMR
spectra permitted the complete assignment of all proton
and carbon resonances for the products. The configura-
tion at C-1 was assigned on the basis of the following
arguments: (a) In the 13C NMR spectra of 1,3-disubsti-
tuted tetrahydroisoquinolines10 and â-carbolines,11 both
C-1 and C-3 carbons are shifted upfield in the trans
isomer due to a compression effect for 1,4-gauche interac-
tions12 that are precluded in the cis isomer. (b) The H-1
proton is usually upfield in the cis isomer.6,7,13 The
preferential formation of the 1R isomer (the cis form) is
predicted by the Felkin-Anh model of asymmetric induc-
tion9 and is consistent with previous observations in the
case of the reaction of L-DOPA with D-glyceraldehyde.6
The fructopyranose structure 3 was inferred from the
chemical shift of the anomeric carbon, resonating at a δ
of 97.0 (fructofuranose anomeric carbons usually appear
at a δ of ca. 103).17 The â configuration of the anomeric
carbon was deduced from a ROESY experiment indicat-
ing a close spatial proximity between the CH2NH protons
of fructose and the axial protons on C-3′ and C-4′ (Figure
1). Finally, the values of the coupling constants between
H-3′ and H-4′ protons and between H-4′ and H-5′ protons
2
indicated a C5 conformation of the pyranose ring.14a
Kinetic analysis indicated that the product ratios do
not change during the course of the reaction and remain
unaffected at higher temperatures, e.g., 80 °C. The
reaction showed pH-dependent kinetics, increasing in
alkaline medium (Figure 2), without significant changes
in product distribution.
Rea ction s of Dop a m in e a n d L-Tyr osin e w ith D-
Glu cose. From the complex of these results, it appears
that L-DOPA can react with D-glucose via two main
competing pathways, viz., a Pictet-Spengler condensa-
tion and a Maillard process leading in the initial steps
to a decarboxylated Amadori compound. Because of the
(9) Eliel, E. L.; Wilen, S. H.; Mander, L. N. Stereochemistry of
Organic Compounds; Wiley: New York, 1995; p 115.
(10) Iban˜ez, A. F.; Yaculiano, G. B.; Moltrasio-Iglesias, G. Y. J .
Heterocycl. Chem. 1989, 26, 907.
(11) Ungemach, F.; Soerens, D.; Weber, R.; DiPierro, M.; Campos,
O.; Mokry, P.; Cook, J . M.; Silverton, J . V. J . Am. Chem. Soc. 1980,
102, 6976.
(12) Whitesell, J . K.; Minton, M. A. In Stereochemical Analysis of
Alicyclic Compounds by 13C NMR Spectroscopy; Chapman and Hall:
New York, 1987.
(14) (a) Mossine, V. V.; Glinsky, G. V.; Feather, M. S. Carbohydr.
Res. 1994, 262, 257. (b) Ge, S. J .; Lee, T. C. J . Agric. Food Chem. 1997,
45, 1619.
(15) Higgins, P. J .; Bunn, H. F. J . Biol. Chem. 1981, 256, 5204.
(16) J akas, A.; Horvat, Sˇ. J . Chem. Soc., Perkin Trans. 2 1996, 789.
(17) Bock, K.; Pedersen, C. In Advances in Carbohydrate Chemistry
and Biochemistry; Academic Press: New York, 1983; Vol. 41, p 48.
(13) Brossi, A.; Focella, A.; Teitel, S. Helv. Chim. Acta 1972, 55, 15.

5050 J . Org. Chem., Vol. 66, No. 15, 2001
Manini et al.
role of the polyol side chain of glucose in the metal-
assisted Pictet-Spengler reaction.
Mech a n istic Issu es. L-DOPA, dopamine, and L-ty-
rosine displayed quite different reactivities with D-glucose
that could not be rationalized in terms of the currently
accepted mechanism. In particular, the following points
seem worthy of attention:
L-DOPA behaves differently from all of the other amino
acids investigated so far in that it suffers an exclusively
decarboxylative Maillard condensation with D-glucose.
L-Tyrosine does not give rise to Pictet-Spengler prod-
ucts, whereas dopamine does not produce Maillard ad-
ducts. These findings imply that the OH group on the
3-position of the aromatic ring is essential for cyclization
to tetrahydroisoquinoline products, whereas the carboxyl
group is critical for Maillard reaction.
The most plausible explanation would invoke a critical
effect of the OH group on the 3-position of L-DOPA, which
would not only activate the 6-position toward Schiff base
cyclization to give tetrahydroisoquinolines but somehow
also induce decarboxylation and assist the critical Ama-
dori rearrangement that drives Maillard condensation.
Formation of tetrahydroisoquinoline products is clearly
indicative of a Pictet-Spengler reaction involving a
transient Schiff base, which might well be an intermedi-
ate in the Maillard process, as is commonly reported.14b,15
The Schiff base per se is not prone to decarboxylation
and, hence, is unlikely to be the ultimate precursor of
the Maillard product. An attractive hypothesis is that
decarboxylation occurs at the expense of an oxazolidine-
5-one intermediate formed by intramolecular cyclization
of the Schiff base. This intermediate would thus afford
an azomethine ylide that would rapidly establish tau-
tomerization and protonation equilibria to yield the
decarboxylated Amadori product (Scheme 2).
F igu r e 2. pH dependence of product yields of the reaction of
L-DOPA with D-glucose. Data are averages of at least three
determinations. Standard deviation did not exceed (5% of
mean values. L-DOPA ) 10 mM; D-glucose ) 100 mM. Product
yields determined at 3 days: ([) 1, (9) 2, (2) 3.
cogent biological implications, a more detailed under-
standing of the structural and experimental factors
governing this competition was desirable. This was
pursued through a detailed analysis of the reactivity of
two biologically relevant compounds related to L-DOPA,
namely, the catecholamine dopamine, lacking the car-
boxyl group, and the amino acid L-tyrosine, lacking the
hydroxyl group on the 3-position of the aromatic ring,18
which is critical to activate the 6-position toward Pictet-
Spengler type cyclizations.
Under conditions comparable to those of the L-DOPA-
D-glucose reaction, dopamine reacted at a rate similar
to that of L-DOPA to give the diastereoisomeric tetrahy-
droisoquinolines 4 and 5 in an approximate ratio of 4:1
(Scheme 1). These products were isolated by preparative
1
HPLC and identified by comparing the H NMR spectra
with that reported in the literature.19
Notably, no detectable formation of products from
Maillard reactions was observed.
A quite different behavior was displayed by L-tyrosine
under the same experimental conditions. This amino acid
reacted with D-glucose more slowly than did L-DOPA
(initial rate ratio of vDOPA/vTyr ) 8.5) to give a single major
product that proved to be the typical Amadori compound
6, in which the carboxyl group of the amino acid was
retained (Scheme 1).20 The structure was assigned on the
basis of 1H NMR and 13C NMR. No trace of Pictet-
Spengler products could be detected in the L-tyrosine-
D-glucose reaction.
Evolution of the putative azomethine ylide must be
accompanied by a concomitant proton shift to yield the
enol intermediate in the Amadori rearrangement. This
azomethine ylide route is clearly precluded from dopam-
ine in which the Schiff base would be more prone to
cyclization to Pictet-Spengler products than to the
Maillard process.
Whereas reversible formation of the putative oxazoli-
dine-5-one can be envisaged as a common route to all
Schiff bases,21 its facile decomposition, even at room
temperature, is unique to the L-DOPA reaction, and
seems to be the critical step affected by the catechol ring.
The mechanism by which the catechol ring would facili-
tate decarboxylation from the putative oxazolidine-5-one
system is intriguing. Theoretical analysis by semiem-
pirical methods (AM1, PM3) identifies the 6- and 2-posi-
tions as the most reactive in terms of HOMO coefficients
and total charge density, respectively.22 It is therefore
tempting to speculate that the electron rich catechol ring
can be engaged in intramolecular interactions with the
oxazolidine-5-one system and lower, in some way, the
barrier for decarboxylation by stabilizing the developing
positive charges, e.g., on the acyl carbon during fission
of the C-C bond or on the nitrogen center via interaction
with the NH proton. Alternately, perhaps the role of the
OH group on the 3-position of L-DOPA is to enable the
Effects of Tr a n sition Meta l Ca tion s on th e Rea c-
tion of L-DOP A w ith D-Glu cose. In further experi-
ments, the possible effect of transition metal cations of
biological interest (e.g., Fe3+ and Cu2+) on the reactions
of L-DOPA with D-glucose was investigated. Fe3+ ions
markedly accelerated tetrahydroisoquinoline formation
at the expenses of the Maillard process, which was
significantly inhibited. Conversely, Cu2+ ions decreased
the kinetics of both processes without affecting their
relative extents. Other metal cations, e.g., Zn2+ and Mn2+
,
did not affect the kinetic and chemical course of the
reaction. Both Fe3+ and Cu2+ ions also accelerated
tetrahydroisoquinoline formation from dopamine and
D-glucose; this observation supported the general char-
acter of the metal-dependent effects and ruled out any
(18) For the sake of simplicity, we adopted the common system for
nomenclature of catecholamines, which assigns positions 3 and 4 to
the aromatic carbons bearing the hydroxyl groups.
(19) Piper, I. M.; MacLean, D. B.; Kvarnstro¨m, I.; Szarek, W. Can.
J . Chem. 1983, 61, 2721.
(21) (a) Grigg, R.; Surendrakumar, S.; Thianpatanagul, S.; Vipond,
D. J . Chem. Soc., Perkin Trans. 1 1988, 2693. (b) Grigg, R.; Malone, J .
F.; Mongkolaussavaratana, T.; Thianpatanagul, S. Tetrahedron 1989,
45, 3849.
(20) Hashiba, H. J . Agric. Food Chem. 1976, 24, 70.
(22) Manini, P.; d’Ischia, M.; Prota, G. J . Org. Chem. 2000, 65, 4269.

L-DOPA-D-Glucose Decarboxylative Maillard Reaction
J . Org. Chem., Vol. 66, No. 15, 2001 5051
Sch em e 2. P r op osed Mech a n ism for th e F or m a tion of 3 by Rea ction of L-DOP A w ith D-Glu cose
and 22 × 250 mm, respectively) with UV detection at 280 nm
formation of a ketotautomer that can assist as a general
acid in protonating the azomethine ylide as it is formed.23
for L-DOPA and dopamine reactions and at 260 nm for
L-tyrosine reactions. The flow rate was maintained at 1 mL/
To gain some insight into the mechanism of the
min for analytical runs and at 15 mL/min for preparative
Amadori reaction, L-DOPA was incubated with D-glucose
chromatography.
under the same conditions as described above using,
L-DOPA, dopamine, and D-glucose were purchased from
however, phosphate-buffered D2O as the medium. The
Aldrich, while L-tyrosine was purchased from Sigma. Metal
reaction proceeded at a significantly slower rate (initial
ion solutions were prepared from Fe(NH4)(SO4)2 and CuSO4‚
7H2O.
rate ratio of vH/vD ) 3.1) to give a similar product mixture
from which a compound that was chromatographically
indistinguishable from 3 was isolated and subjected to
spectral analysis. The 1H NMR spectrum (Figure 3A)
revealed complete deuteration of the methylene group of
the sugar moiety linking the secondary amino functional-
ity and replacement of only one of the CH2 protons
flanking the nitrogen on the opposite side. This result,
which is supported by the 13C NMR spectrum (Figure 3B),
in which the carbon signals for the relevant methylene
groups are evidently affected by deuteration, is not in
contrast with the postulated pathway. In particular,
monodeuteration on the former carboxyl-bearing carbon
is compatible with conventional decarboxylation mech-
anisms, including that from the oxazolidine-5-one. Com-
plete deuteration of the alternate NHCH2 group is likely
to be indicative of rapid tautomeric equilibria involving
the ketoamine functionality in the Amadori compound.
In this frame, the kinetic isotope effect observed in the
presence of deuterium is likely to be due to a decrease in
the rate of Schiff base formation since both of the
competing Pictet-Spengler and Amadori processes are
affected in accordance with the proposed scheme.
Rea ction s of L-DOP A a n d Dop a m in e w ith D-Glu cose.
Appropriate amounts of L-DOPA or dopamine and D-glucose
were incubated in 0.1 M phosphate buffer, pH 7.4, kept at 37
°C in a rubber-capped tube under argon. When necessary,
aliquots of stock aqueous solutions of metal ions were added.
For product analysis, aliquots of the reaction mixtures were
withdrawn with a syringe, acidified to pH 3, and injected into
the HPLC system. All kinetic experiments were run at least
in triplicate. The eluant used for L-DOPA-glucose reactions
was 5 mM octane-1-sulfonic acid in 0.1 M H3PO4 (pH 3)/
acetonitrile 97:3 v/v (eluant a), while the eluant used for the
dopamine-glucose reactions was 5 mM octane-1-sulfonic acid
in 0.1 M H3PO4 (pH 3)/acetonitrile 95:5 v/v (eluant b).
(1R,1′S,3S)-1-(D-gluco-P en titol-1′-yl)-3-ca r boxy-6,7-d ih y-
d r oxy-1,2,3,4-tetr a h yd r oisoqu in olin e (1), (1S,1′S,3S)-1-(D-
gluco-P en titol-1′-yl)-3-ca r boxy-6,7-d ih yd r oxy-1,2,3,4-tet-
r a h yd r oisoqu in olin e (2), a n d N-(1-Deoxy-D-fr u ctos-1-yl)-
d op a m in e (3). A solution of L-DOPA (100 mg, 0.5 mmol) in
0.1 M phosphate buffer, pH 7.4 (10 mL), previously purged
with argon, was treated with D-glucose (1.4 g, 15 mmol) at 80
°C. After 4 h, HPLC analysis (eluant a) revealed three main
products along with small amounts of the starting material.
The mixture was acidified to pH 2 with 3 M HCl and subjected
to ion-exchange chromatography (Dowex 50W-X4 (H+), 2 × 60
cm) using water (250 mL), 0.1 HCl (250 mL), 0.5 M HCl (500
mL), and 1 M HCl (500 mL) as the eluants. Fractions eluted
with 0.5 and 1 M HCl were collected, evaporated to dryness
under reduced pressure, and subjected to preparative HPLC
(eluant, 0.1 M HCOOH/acetonitrile 97:3 v/v) to afford 1 (25%),
2 (15%), and 3 (30%).
Exp er im en ta l Section
1
Gen er a l Meth od s. H NMR spectra were obtained at 400
and 200 MHz with tert-butyl alcohol (δ 1.23) as the internal
standard. 13C NMR spectra were run at 50 MHz. 1H-1H COSY
and ROESY experiments were performed at 400 MHz. Mass
spectra were recorded using the fast atom bombardment (FAB)
technique with glycerol as the matrix. Analytical and prepara-
tive HPLC were carried out on C18 columns (4.6 × 250 mm
1: UV λmax (log ꢀ) (H2O) 286 (2.79) nm; [R]21 +6.76 (c 0.5,
D
1
0.5 M HCl); H NMR (D2O) δ 3.03-3.21 (2H, m), 3.67-3.82
(1H × 4, m), 4.10-4.14 (1H × 2, m), 4.40 (1H, s), 4.75 (1H, s),
6.83 (1H, s), 6.91 (1H, s); 13C NMR (D2O) δ 28.7 (t), 55.9 (d),
60.5 (d), 63.4 (t), 71.0 (d), 72.0 (d), 72.5 (d), 73.4 (d), 114.6 (d),
117.1 (d), 121.7 (s), 126.0 (s), 144.3 (s), 145.0 (s), 173.4 (s); MS
(23) We acknowledge a suggestion by one of the referees.

5052 J . Org. Chem., Vol. 66, No. 15, 2001
Manini et al.
F igu r e 3. (A) 1H NMR spectrum (400 MHz, high-field region) of 3 obtained in phosphate-buffered H2O (upper trace) and in
phosphate-buffered D2O (lower trace). (B) 13C NMR spectrum (50 MHz, high-field region) of 3 obtained in phosphate-buffered
H2O (upper trace) and in phosphate-buffered D2O (lower trace).
(FAB) m/z 360; HRMS (M+ + 1) calcd for C15H22NO9 360.1295,
found 360.1297. Anal. Calcd for C15H21NO9: C, 50.12; H, 5.89;
N, 3.90. Found: C, 50.10; H, 5.85; N, 3.88.
3: UV λmax (log ꢀ) (H2O) 281 (2.64) nm; [R]21 -2.77 (c 0.9,
D
1
0.5 M HCl); H NMR (D2O) δ 3.15 (2H, m), 3.31 (2H, s), 3.72
(1H, d, J ) 9.6 Hz), 3.73 (1H, d, J ) 12 Hz), 3.88 (1H, dd, J )
9.6, 3.4 Hz), 3.97 (1H, d, J ) 12 Hz), 4.01 (1H, d, J ) 3.4 Hz),
4.07 (2H, m), 6.73 (1H, dd, J ) 8, 2.5 Hz), 6.85 (1H, d, J ) 2.5
Hz), 6.92 (1H, d, J ) 8 Hz); 13C NMR (D2O) δ 36.5 (t), 45.5 (t),
54.5 (t), 65.8 (t), 70.7 (d), 71.0 (d), 71.9 (d), 97.0 (s), 118.5 (d),
119.0 (d), 123.7 (d), 128.6 (s), 145.2 (s), 145.9 (s); MS (FAB)
m/z 316; HRMS (M+ + 1) calcd for C14H22NO7 316.1396, found
316.1394. Anal. Calcd for C14H21NO7: C, 53.31; H, 6.72; N,
4.44. Found: C, 53.29; H, 6.71; N, 4.42.
2: UV λmax (log ꢀ) (H2O) 286 (2.76) nm; [R]21 -4.33 (c 0.7,
D
1
0.5 M HCl); H NMR (D2O) δ 3.17 (1H, dd, J ) 17.0, 4.5 Hz),
3.31 (1H, dd, J ) 17.0, 5.2 Hz), 3.57-3.80 (1H × 5, m), 4.39
(1H, m), 4.67 (1H, t, J ) 5.2 Hz), 4.99 (1H, d, J ) 4.6 Hz),
6.77 (1H, s), 6.85 (1H, s); 13C NMR (D2O) δ 28.9 (t), 54.6 (d),
58.7 (d), 63.6 (t), 70.3 (d), 72.0 (d), 72.9 (d), 73.2 (d), 115.0 (d),
117.7 (d), 120.8 (s), 124.6 (s), 144.8 (s), 145.5 (s), 173.1 (s); MS
(FAB) m/z 360; HRMS (M+ + 1) calcd for C15H22NO9 360.1295,
found 360.1296. Anal. Calcd for C15H21NO9: C, 50.12; H, 5.89;
N, 3.90. Found: C, 50.11; H, 5.87; N, 3.87.
(1R,1′S)-1-(D-gluco-P en titol-1′-yl)-6,7-d ih yd r oxy-1,2,3,4-
tetr a h yd r oisoqu in olin e (4) a n d (1S,1′S)-1-(D-gluco-P en -

L-DOPA-D-Glucose Decarboxylative Maillard Reaction
J . Org. Chem., Vol. 66, No. 15, 2001 5053
t it ol-1′-yl)-6,7-d ih yd r oxy-1,2,3,4-t e t r a h yd r oisoq u in o-
lin e (5). A solution of dopamine (100 mg, 0.5 mmol) in 0.1 M
phosphate buffer, pH 7.4 (10 mL), previously purged with
argon, was treated with D-glucose (1.4 g, 15 mmol) at 80 °C.
After 4 h, HPLC analysis (eluant b) revealed two main
products along with small amounts of the starting material.
The mixture was acidified to pH 2 with 3 M HCl and subjected
to ion-exchange chromatography (Dowex 50W-X4 (H+), 2 × 60
cm) using water (250 mL), 0.1 M HCl (250 mL), 0.5 M HCl
(250 mL), 1 M HCl (500 mL), and 2 M HCl (500 mL) as the
eluants. Fractions eluted with 1 and 2 M HCl were collected,
evaporated to dryness under reduced pressure, and subjected
to preparative HPLC (eluant, 0.1 M HCOOH/acetonitrile 95:5
v/v) to afford 4 (60%) and 5 (15%).
pressure, and subjected to preparative HPLC (eluant, 0.1 M
HCOOH/acetonitrile 97:3 v/v) to afford 6 (15%).
1
6: H NMR (D2O) δ 3.31 (2H, m), 3.43 (2H, s), 3.75 (1H, d,
J ) 10 Hz), 3.76 (1H, d, J ) 12 Hz), 3.92 (1H, dd, J ) 10, 3.4
Hz), 4.03 (1H, d, J ) 3.4 Hz), 4.04 (1H, d, J ) 12 Hz), 4.34
(1H, m), 6.92 (1H × 2, dd, J ) 8.4, 2.7 Hz), 7.23 (1H × 2, dd,
J ) 8.4, 3 Hz); 13C NMR (DEPT) (D2O) δ 35.2 (t), 46.4 (d),
53.5 (t), 65.3 (t), 70.1 (d), 70.2 (d), 71.0 (d), 117.4 (d), 132.3 (d);
MS (FAB) m/z 344; HRMS (M+ + 1) calcd for C15H22NO8
344.1345, found 344.1341. Anal. Calcd for C15H21NO8: C,
52.31; H, 6.44; N, 4.07. Found: C, 52.29; H, 6.41; N, 4.02.
Ack n ow led gm en t. This work was supported by
grants from MURST (PRIN 1997) and CNR (Rome). We
thank Miss Silvana Corsani for technical assistance, the
Centro Interdipartimentale di Metodologie Chimico-
Fisiche (CIMCF, University of Naples) for NMR spectra,
and the Servizio di Spettrometria di Massa del CNR e
dell’Universita` (Naples) for mass spectra.
NMR and other spectral data were in accordance with
literature.19
N-(1-Deoxy-D-fr u ctos-1-yl)-L-tyr osin e (6). A solution of
L-tyrosine (100 mg, 0.6 mmol) in 0.1 M phosphate buffer, pH
7.4 (6 mL), was treated with D-glucose (1.4 g, 15 mmol) at 100
°C. After 24 h, HPLC analysis (eluant, 0.1 M HCOOH/
acetonitrile 98:2 v/v) revealed one main product along with
large amounts of the starting material. The mixture was
acidified to pH 2 with 3 M HCl and subjected to ion-exchange
chromatography (Dowex 50W-X4 (H+), 2 × 60 cm) using water
(250 mL), 0.1 M HCl (250 mL), 0.5 M HCl (500 mL), and 1 M
HCl (500 mL) as the eluants. Fractions eluted with 0.5 and 1
M HCl were collected, evaporated to dryness under reduced
Su p p or tin g In for m a tion Ava ila ble: 1H NMR, 13C NMR,
1
DEPT, H-1H COSY, and ROESY spectra of compounds 1, 2,
3, and 6. This material is available free of charge via the
Internet at http://pubs.acs.org.
J O010078D