Reactions of adenosine with bromo- and chloromalonaldehydes in aqueous solution: Kinetics and mechanism

Article abstract of DOI:10.1002/1099-0690(200006)2000:12<2315::AID-EJOC2315>3.0.CO;2-G

Reactions of adenosine nucleosides with halogen substituted acetaldehydes and malonaldehydes have been studied and pseudo first-order rate constants have been determined. All the reactions yield 1,N⁶-etheno adducts, and with malonaldehydes; in addition to this, 11-formyl-1,N⁶-etheno adducts are also formed. Particular attention has been paid to the formation of the formyletheno products. The results obtained suggest that the reactions of adenine base with halogenated acetaldehydes and malonaldehydes are basically similar. It also seems that in reactions of halomalonaldehydes with adenosine, the etheno and formyletheno products are formed through the same initial reaction pathway i.e. the attack of the 6-amino group of the adenine base at the carbonyl carbon atom of the aldehyde.

Full text of DOI:10.1002/1099-0690(200006)2000:12<2315::AID-EJOC2315>3.0.CO;2-G

FULL PAPER
Reactions of Adenosine with Bromo- and Chloromalonaldehydes in Aqueous
Solution: Kinetics and Mechanism
Satu Mikkola,*[a] Niangoran Koissi,[a] Kaisa Ketomäki,[a] Susanna Rauvala,[a]
Kari Neuvonen,[a] and Harri Lönnberg[a]
Keywords: Nucleobases / DNA alkylations / Kinetics / Ethenoadenosine / Aldehydes
Reactions of adenosine nucleosides with halogen substituted
acetaldehydes and malonaldehydes have been studied and
suggest that the reactions of adenine base with halogenated
acetaldehydes and malonaldehydes are basically similar. It
pseudo first-order rate constants have been determined. All also seems that in reactions of halomalonaldehydes with ade-
the reactions yield 1,N6-etheno adducts, and with malonal- nosine, the etheno and formyletheno products are formed
dehydes, in addition to this, 11-formyl-1,N6-etheno adducts
are also formed. Particular attention has been paid to the for-
mation of the formyletheno products. The results obtained
through the same initial reaction pathway i. e. the attack of
the 6-amino group of the adenine base at the carbonyl car-
bon atom of the aldehyde.
Introduction
Formation of etheno adducts of nucleic acid bases has
a dual role in the chemistry of DNA. Firstly, miscoding in
DNA synthesis owing to such adducts has been established
both in vitro and in vivo,[1] and their formation has been
identified as the reason for mutagenicity of several bifunc-
tional halo compounds including vinyl chloride,[1a,1d,2]
a
widely used industrial chemical, and chlorinated hydroxyfu-
ranones,[3] side products of chlorination of tap water. Many
of the mutagenic halo compounds are converted intracellul-
arly to halogenated acetaldehydes or malonaldehydes,
which are assumed to react with DNA bases.[4]
In addition to the undesired mutagenic effects, the etheno
adducts of nucleobases also have properties that make them
useful tools for nucleic acid chemistry. Since the etheno ad-
ducts are fluorescent, they are extensively used as surrogates
of natural nucleotides in biochemical studies. Ever since the
early works of Kochetkov,[5] the rapid and clean reactions
of haloacetaldehydes 1a,b with nucleic acid bases, yielding
ethenonucleosides, have been utilized extensively in the
probing of nucleic acid structures.[6] The corresponding re-
actions of halomalonaldehydes 2a,b appear even more at-
tractive, since the reactive formyl function of the resulting
formyletheno adduct allows for the further derivatization of
the polynucleotide chain.
guanine[8] adducts, respectively. It has been suggested that
the carbonyl carbon atom of the aldehyde is initially at-
tacked by the exocyclic amino function of a nucleobase and
the halogen substituent is then displaced by intramolecular
nucleophilic attack of the neighboring ring nitrogen
(Scheme 1). The cyclic carbinolamine intermediate is finally
dehydrated to the etheno product.
The kinetics and mechanisms of the reactions of halo-
acetaldehydes 1a,b with nucleosides and DNA have been
studied rather extensively.[7] Chloro- and bromoacetal-
dehydes have been shown to react with adenine, cytosine
and guanine nucleosides giving 1, N6-ethenoadenine,[3c,6,7]
3,N4-ethenocytosine[3c,6,7aϪ7c] and 1,N2-or 2,N3-etheno-
[a]
Department of Chemistry, University of Turku,
The reactions of halomalonaldehydes 2a,b with nucleo-
bases have been less thoroughly studied. The reaction with
adenosine (3a) has been reported to give two etheno prod-
FIN-20014 Turku, Finland
Fax: (internat.) ϩ358-2-3336700
E-mail: satu.mikkola@utu.fi
Eur. J. Org. Chem. 2000, 2315Ϫ2323
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
1434Ϫ193X/00/0612Ϫ2315 $ 17.50ϩ.50/0
2315

S. Mikkola, N. Koissi, K. Ketomäki, S. Rauvala, K. Neuvonen, H. Lönnberg
FULL PAPER
petition between the formation of 4a and 5a under various
conditions and detection of unstable intermediates. Com-
parative measurements were also carried out with 9-methyl-
adenine (3b) to simplify the LC-MS analysis of the reaction
intermediates observed, and with N6-methyladenosine (3c)
to facilitate the detection of possible carbinolamine inter-
mediates. Replacement of one of the N6-hydrogens of aden-
osine with a methyl group may be expected to retard the
dehydration of carbinolamine intermediates, and hence
such intermediates are in all likelihood accumulated to a
larger extent.
Results
The reactions of the adenine substrates 3aϪc with halo-
malonaldehydes 2a,b and haloacetaldehydes 1a,b were
studied at 60 °C by determining the composition of aliquots
withdrawn at appropriate intervals by RP-HPLC. The
Scheme 1
ucts 1,N6-ethenoadenosine (4a) and 11-formyl-1,N6-
ethenoadenosine (5a),[2d,3b] but the mechanism is not
known in detail. According to Nair et al.,[2d] the carbonyl
carbon atom of the aldehyde is initially attacked by the 6-
amino group of adenosine, and the carbinolamine interme-
diate is either cyclized and dehydrated to 5a or decomposed
to haloacetaldehyde and N6-formyladenosine (Scheme 2).
Compound 4a is then formed as a secondary product by
the reaction of adenosine with the haloacetaldehyde as de-
picted in Scheme 1. The suggestion is, however, based only
on product analysis and the fact that secondary amines are
known to accelerate the cleavage of bromomalonaldehyde
to bromoacetaldehyde by attacking the carbonyl carbon
atom.[9] Kronberg et al.[3a,10] have more recently proposed
a similar mechanism, but regarded spontaneous cleavage of
halomalonaldehyde as a reasonable alternative for the
formation of the haloacetaldehyde. The present paper de-
scribes our kinetic studies on the reactions of adenosine
(3a) with chloromalonaldehyde (CMA; 2a) and bromoma-
lonaldehyde (BMA; 2b). For comparative purposes, and to
facilitate the mechanistic interpretation of the kinetic re-
sults, the reactions with chloroacetaldehyde (CAA; 1a) and
bromoacetaldehyde (BAA; 1b) were also studied under
identical conditions. Special attention was paid to the com-
measurements were carried out under pseudo first-order
conditions, the concentration of 3aϪc being 0.1 m, and
that of the aldehyde at least 10 times higher. Under these
conditions, the decrease in the concentration of the alde-
hyde can be considered to become insignificant. The de-
crease in the concentration of the adenine derivative was
followed as a function of time, and a first-order rate con-
stant was calculated by applying the integrated first-order
rate law. The products formed in the reactions of adenosine
(3a) were identified by spiking with authentic samples.
Those from the reactions of 9-methyladenine (3b) and N6-
methyladenosine (3c) were isolated by semipreparative RP-
HPLC and identified by 1H- and/or 13C NMR spectroscopy
and LC-ES/MS analysis. Consistent with previous re-
ports,[6,7] the reactions of 3a,b with halomalonaldehydes
were observed to give both 1,N6-etheno (4a,b) and 11-for-
myl-1,N6-etheno (5a,b) adducts, and those with haloacetal-
dehydes only the etheno adducts. It is worth noting that the
formation of 4b by the reaction of 3b with BMA (2b) is in
contradiction with a previous report.[2d] The products of
the subsequent decomposition of the etheno adducts were
not fully characterized in the present work, but it has been
previously shown that 4a is cleaved under acidic[7d] and ba-
sic[11] conditions to form bisimidazole derivative 6.
Figure 1 shows the pseudo first-order rate constants for
the disappearance of adenosine (3a) as a function of the
concentration of the halomalonaldehyde or haloacetal-
dehyde. As seen, the reactions of adenosine with BMA are
about one order of magnitude slower than the reactions
with BAA. With chlorosubstituted aldehydes such reactivity
difference was observed only at high aldehyde concentra-
tion. With both types of aldehydes, the bromosubstituted
compounds seem to be more reactive than the chloro-sub-
stituted ones. The higher reactivity of BAA relative to CAA
Scheme 2
2316
Eur. J. Org. Chem. 2000, 2315Ϫ2323

Reactions of Adenosine with Bromo- and Chloromalonaldehydes
FULL PAPER
Figure 2. Logarithmic rate constants for the reactions of adenosine
(3a) and 9-methyladenine (3b) with halomalonaldehydes (2a, b) as a
function of pH at 60 °C; [aldehyde] ϭ 10 mm, I ϭ 0.1  (NaNO3);
notation: ᮀ 9-methyladenine with BMA; ᭿ 9-methyladenine with
CMA; ᭹ adenosine with BMA and ᭺ adenosine with CMA
Figure 1. Logarithmic rate constants for the reactions of adenosine
(3a) with haloacetaldehydes (1a, b) and halomalonaldehydes (2a,
b) as a function of the aldehyde concentration at pH 4.7 and 60
°C. I ϭ 0.1  (NaNO3); notation: ᮀ bromoacetaldehyde; ᭿ chlo-
roacetaldehyde; ᭹ bromomalonaldehyde and ᭺ chloromalonal-
dehyde
has also previously been observed on reacting these alde-
hydes with ATP.[12] The reactivity difference between BMA
and CMA is, however, smaller than that between BAA and
CAA. The dependence of reaction rate on the aldehyde con-
centration is also different with halomalonaldehydes and
haloacetaldehydes. While the reaction with haloacetal-
dehydes is approximately first-order in the aldehyde concen-
tration over the entire concentration range studied, a clear
deviation from the first-order dependence is observed with
halomalonaldehydes. With CMA, the rate constants ob-
tained seem to level off to a constant value at high aldehyde
concentrations, whereas with BMA the phenomenon is
much less pronounced.
Figure 2 shows the pH rate profiles for the reactions of
the adenine substrates 3a, 3b with CMA and BMA. The
rate profiles of all the reactions share a common feature:
the reaction rate passes through a broad maximum under
mildly acidic conditions, and hence only a rather moderate
effect of pH on the rate is observed. The greatest depend-
ence on pH was observed in reactions with CMA, where an
Figure 3. Logarithmic rate constants for the reactions of adenosine
(3a) with BAA (᭺) and with CAA (᭹) as a function of pH at 60
°C; [aldehyde] ϭ 10 mm, I ϭ 0.1  (NaNO3)
increase of three units in pH resulted in only a ten-fold rate practically no effect on the rate of the disappearance the
enhancement. Similar behaviour has previously been ob- adenine substrate. The pseudo first-order rate constant for
served with haloacetaldehydes.[5,6b,7a,12] This was also seen the reaction of adenosine with BMA, for example, remained
in the present work (Figure 3). The pH at which the max- unchanged (1.2 ϫ 10Ϫ5 sϪ1) on increasing the concentration
imum reactivity is observed seems, to some extent, to de- of a 1:1 formic acid/sodium formate buffer from 0.1  to
pend on the identity of the reactants in question, in particu- 0.5  (data not shown).
lar on the halogen substituents of the aldehyde. The max-
The rate constants for the breakdown of 1,N6-ethenoad-
imum value is observed between pH 4 and 5 with BMA enosine (4a) as a function of pH are shown in Figure 4. The
and between pH 4.5 and 5.5 with CMA. The reaction of dependence of the rate of the reaction on [Hϩ] seems to
N6-methyl adenine with BMA is almost entirely pH-inde- become more significant as the pH decreases. The reaction
pendent (data not shown). The buffer concentration had is nearly pH-independent at pH Ͼ 3; on going from pH 6
Eur. J. Org. Chem. 2000, 2315Ϫ2323
2317

S. Mikkola, N. Koissi, K. Ketomäki, S. Rauvala, K. Neuvonen, H. Lönnberg
FULL PAPER
Figure 4. Logarithmic rate constants for the disappearance of
ethenoadenosine (4a) and formylethenoadenosine (5a) in buffer so-
lutions (c ϭ 0.1 ) as a function of pH at 60 °C; I ϭ 0.1 
(NaNO3); ᭺ ethenoadenosine; ᭹ formylethenoadenosine
Figure 6. Logarithmic rate constants for the formation of the
etheno product (4b; ᭺) and the formyletheno derivative (5b; ᭹) in
the reaction of 9-methyladenine with 10 m BMA as a function of
pH at 60 °C; I ϭ 0.1  (NaNO3)
to pH 3, the rate increases only by a factor of less than
ten. Below pH 3 the dependence approaches a first-order
dependence on [Hϩ], an indication of an acid-catalyzed pro-
cess. Under slightly acidic conditions, the 1,N6-formyl-
etheno adduct (5a) decomposed 2 to 4 times less readily
than 4a.
As mentioned above, the reactions of adenosine and 9-
methyladenine with BMA yield two stable products, viz. the
1,N6-etheno (4a,b) and 11-formyl-1,N6-etheno (5a,b) ad-
ducts. The products were formed in parallel, obeying first-
order kinetics. No intermediates could be detected. Com-
pounds 5a,b were not converted into 4a,b under the reac-
tion conditions. Figure 5 shows as an illustrative example
the time-dependent product distribution for the reaction of
3b with BMA at pH 2.7. At longer reaction times, break-
down of the etheno adduct 4b is observed. The pH-depend-
ence of the pseudo first-order rate constants observed for
the formation of 4b and 5b is depicted in Figure 6. It can
be seen that in the reaction of 3b with BMA, the etheno
(4b) and formyletheno (5b) adducts were formed in an ap-
proximately equimolar ratio at pH Ͻ 3.5, while at higher
pH, 4b predominated. The optimum pH for the formation
of 4b (pH 4.7) is more than one unit greater than that for
the formation of 5b (pH 3.3). The pH-dependent product
distribution of the reaction of adenosine with BMA was
observed to be very similar (data not shown). The concen-
tration of BMA does not seem to affect the product ratio,
as is shown by the rate contants in Table 1. The formyle-
theno adduct (5a,b) was accumulated to a clearly lesser ex-
tent when using CMA as starting material. With CMA, the
formation of 5a,b only accounts for up to 10% of the total
disappearance of 3a,b over the entire pH range (data not
shown).
Table 1. Rate constants of formation of the etheno (4a) and formyl-
etheno (5a) derivatives of adenosine in the reaction of adenosine
(3a) with BMA at pH 4.7 and 60 °C
Ϫ6
[BMA]/mol dmϪ3
k (4a)/10
sϪ1
k (5a)/10Ϫ6 sϪ1
0.005
0.010
0.020
0.030
0.040
2.3±0.1
6.3±0.1
10.4±0.1
14.0±0.9
16.0±0.9
2.9±0.6
3.5±0.4
6.1±0.6
9.4±0.6
13.1±0.6
Figure 5. Time-dependent product distribution in the reaction of
9-methyladenine (3b) with BMA at pH 2.7 at 60 °C. I ϭ 0.1 
(NaNO3); notation: ᭿ 9-methyladenine; ᭺ etheno product (4b);
᭹ formyletheno product (5b)
2318
Eur. J. Org. Chem. 2000, 2315Ϫ2323

Reactions of Adenosine with Bromo- and Chloromalonaldehydes
FULL PAPER
The reaction of N6-methyladenosine (3c) with BMA and 1). On using BAA (10 m) as starting material, accumula-
CMA yielded only one major stable product in the pH tion of the cyclic carbinolamine at a level up to 10% of the
range from 3 to 6. This compound was characterized by 1H initial concentration of 3a (10 m) was observed by HPLC-
and 13C NMR spectroscopy and FAB mass spectroscopy MS. With CAA, the cyclic carbinolamine is not clearly ac-
and assigned as the hydrated 1,N6-etheno product of 3c, cumulated. This is expected since the intramolecular dis-
i. e. 7,8-dihydro-8-hydroxy-9-methyl-3-(β--ribofuranosyl)- placement of chlorine, i. e. the cyclization step, is un-
imidazo[2,1-i]purine (7). Additionally, the formation of a doubtedly slower than that of bromide, while the sub-
small amount of another stable product was observed at sequent breakdown of the intermediate is naturally as rapid
pH 6. The HPLC-MS data suggests that this compound in both cases. The proposed mechanism is also consistent
could be the 11-formyl-1,N6-etheno adduct of N6-methylad- with the observed reactivity order of CAA and BAA; repla-
enosine (8). While N6-methyladenosine reacted with halom- cing of chlorine with bromine may well accelerate the nucle-
alonaldehydes at pH 4 to 5 as rapidly as adenosine, its reac- ophilic attack on the neighbouring carbonyl carbon atom.
tion rate with CAA was only one third of that of adenosine. Consistent with this, the base-catalysed hydrolysis of α-
bromo acetic acid ethyl ester, for example, is faster than that
of the corresponding chloro compound.[14] In comparison
to the situation with α-halo esters, the reactivity difference
observed in this work is, however, unexpectedly large (more
than one order of magnitude). Accordingly, the possibility
that the cyclization step is partially rate-limiting should not
be strictly excluded.
Analogously, the reactions of halomalonaldehydes 2a,b
with the adenine substrates 3a,b may also be assumed to be
initiated by an attack of the 6-amino group on one of the
carbonyl carbon atoms. The formyletheno product (5a,b)
Discussion
could then be obtained by subsequent cyclization by the
displacement of the halogen substituent with the N1 atom
Reaction Order in the Aldehyde Concentration
As indicated in the introduction, the reaction of adenos- and dehydration of the resulting cyclic carbinolamine
ine with halogenated acetaldehydes has been suggested to (Scheme 2). To obtain the etheno product 4a,b, the haloma-
be a three-step process.[7aϪ7c] A cyclic carbinolamine inter- lonaldehyde must undergo deformylation. As discussed be-
mediate is obtained by two consecutive reactions: initial nu- low in more detail, we believe that this branching of the
cleophilic addition of the 6-amino group to the carbonyl reaction pathway takes place at the initially formed acyclic
group and subsequent SN2 substitution of the halogen sub- carbinolamine (Scheme 3). Accordingly, while formation of
stituent by the N1 atom of the adenine ring (Scheme 1). the acyclic carbinolamine intermediate may, in principle, be
While the cyclic carbinolamine intermediate is detectably reversible, its subsequent cyclization or deformylation/cyc-
accumulated, no direct observation of the existence of the lization must be virtually irreversible processes in dilute
acylic carbinolamine has ever been reported. The suggested aqueous solution. This forms the basis for the following
regioselectivity, i. e. the attack of N6 on the carbonyl carbon discussion concerning the dependence of the overall reac-
atom and not on C2 of the aldehyde, receives support from tion rate on the concentration of the halomalonaldehyde.
the observation that the reaction of α-bromopropionaldehy-
On the basis of the kinetic results obtained, it can be
de[7c] or α-chlorobutyraldehyde[13] with adenosine yields suggested that in reactions of adenosine (or its analogs)
only one product, viz. the one having the original carbonyl with halogenated malonaldehydes, the first step is no longer
carbon bonded to N6. The first nucleophilic attack has been rate limiting. The formal kinetics of the reactions of halo-
suggested to be the rate-limiting step of the formation of malonaldehydes differ from those of haloacetaldehydes in
the cyclic carbinolamine intermediate: once the linear carbi- one important aspect: The reaction is not first-order in the
nolamine has been formed, it is assumed to undergo virtu- aldehyde concentration but the rate increase observed at
ally irreversible cyclization which is faster than the collapse low aldehyde concentration begins to level off at high alde-
of the linear carbinolamine back to starting materials. The hyde concentrations (Figure 1). This kind of kinetic be-
etheno product is then formed by dehydration of the cyclic havior is a consequence of saturation of the adenine sub-
carbinolamine. This final step is sufficiently fast to prevent strate with the aldehyde. It seems that the acyclic carbinol-
extensive accumulation of the cyclic carbinolamine.
amine is accumulated in a pre-equilibrium, and its cycliza-
The kinetic results of the present study are consistent tion or deformylation/cyclization constitutes the rate-
with the mechanism presented above. The reaction of aden- limiting step for the formation of the cyclic carbinolamine
ine substrates 3a,b with haloacetaldehydes 1a,b was ob- intermediate. In other words, the subsequent reactions of
served to be first-order in the aldehyde concentration, con- initially formed intermediate are slower than the collapse of
sistent with the assumed rate-limiting attack on the car- the intermediate back to the starting materials. The lower
bonyl carbon atom. Only a very modest downward curva- reactivity of halomalonaldehydes 2a,b relative to haloacet-
ture could be observed on plotting the pseudo first-order aldehydes 1a,b would hence result from the less rapid sub-
rate constants against the aldehyde concentration (Figure sequent reactions of the acyclic carbinolamine intermediate.
Eur. J. Org. Chem. 2000, 2315Ϫ2323
2319

S. Mikkola, N. Koissi, K. Ketomäki, S. Rauvala, K. Neuvonen, H. Lönnberg
FULL PAPER
Scheme 3
The kinetic saturation is more marked with CMA than with ity of 3a,b. However, the reactions are only moderately re-
BMA. Since the intramolecular displacement of chlorine is tarded on the acidic side of the pKa values of 3a,b; the as-
more difficult than that of bromine, and since chlorine may cending parts of the pH-rate profiles (pH Ͻ 3.5) exhibit a
also be expected to facilitate the hydrolytic deformylation reaction-order of less than 0.5 in the hydroxide ion concen-
less efficiently than bromine, the cyclization step alone may tration. At pH Ͼ 5, the reactions are, in turn, decelerated
well be rate-limiting with CMA. The reaction of BMA may with increasing pH, although no change in the ionic form
still represent a transition from rate-limiting formation to of the adenine substrate takes place. Although the inter-
rate-limiting breakdown of the acyclic carbinolamine inter- pretation of the pH-rate profiles of multistep reactions, such
mediate.
as the present ones, is a complicated task, it appears clear
The fact that methylation of the 6-amino group of aden- that the protolytic equilibrium of the adenine substrate
osine (conversion of 3a to 3c) retards the reaction of aden- alone does not explain the observed pH-rate profiles. The
osine with CAA but not with CMA may also be accounted protolytic equilibria of the malonaldehyde also affect the
for by the proposed change in the rate-limiting step. For pH-rate profiles. The pKa values of BMA and CMA are not
steric reasons, the 6-methylamino group of 3c may be ex- known, but the pKa values of their 2-methyl and 2-methoxy
pected to be a less efficient nucleophile than the 6-amino counterparts have been reported to be 4.7 and 4.1, respect-
group of 3a. Since with CAA the initial nucleophilic attack ively.[16] Accordingly, the pKa values of BMA and CMA
of the amino group constitutes the rate-limiting step, a rate- bearing a more electronegative 2-substituent than 2-me-
retardation is observed. With CMA, the amino and car- thoxymalonaldehyde may be expected to lie around 3.5 and
bonyl groups react in pre-equilibrium step, and hence the BMA is quite probably more acidic than CMA. In other
steric hindrance caused by the methyl group plays a less words, the adenine substrate and the aldehyde both undergo
important role.
deprotonation on passing pH 3.5: adenine substrate from a
cationic form to a more nucleophilic neutral species, and
halomalonaldehyde from a neutral molecule to a less elec-
trophilic enolate anion. The influences of these two depro-
tonations on the overall reaction rate evidently cancel each
other at least partially.
pH-Rate Profiles
As an indication of the overall likeness of the reactions
of halomalonaldehydes and haloacetaldehydes with adenine
substrates, the pH-dependence of all the reactions are rather
similar: the pH-rate profiles are bell shaped exhibiting a
broad maximum in the pH range 3Ϫ5. The maximum with
BMA occurs at a lower pH than with CMA. The maximum
rate is hence achieved under conditions where the adenine
Mechanisms of the Formation of the Etheno and
Formyletheno Products
As indicated above, the reactions of adenine substrates
substrates are converted from neutral species to monoca- 3a,b with BMA or CMA give both etheno (4a,b) and for-
tions. The pKa values of 3a and 3b are 3.7 and 4.0, respect- myletheno (5a,b) products, although with CMA the etheno
ively, the site of protonation being N1.[15] This protonation products clearly predominate. It has previously been sug-
might be expected to reduce dramatically the nucleophilic- gested that the etheno product is actually formed by initial
2320
Eur. J. Org. Chem. 2000, 2315Ϫ2323

Reactions of Adenosine with Bromo- and Chloromalonaldehydes
FULL PAPER
conversion of halomalonaldehyde to haloacetaldehyde: from a common intermediate which still contain the hal-
either the malonaldehyde is cleaved spontaneously to the ogen substituent. We tentatively suggest that the branching
corresponding acetaldehyde,[9] or the acetaldehyde is takes place at the acyclic carbinolamine intermediate, which
formed by initial nucleophilic attack by the 6-amino func- may either undergo cyclization by displacement of the hal-
tion of the adenine substrate onto the carbonyl carbon of ogen substituent by the adenine N1 atom (Route A in
the malonaldehyde (Scheme 2).[2d] Neither of these mechan- Scheme 3), or hydrolytic deformylation prior to cyclization
isms is, however, consistent with the kinetic data of the pre- to a hydroxyethano adduct (Route B in Scheme 3). Since
sent study. The formation of 1,N6-ethenoadenosine (4a) bromine undergoes nucleophilic displacement more readily
from CMA and adenosine is clearly too fast to be explained than chlorine, formation of the formyletheno adduct is fa-
by a mechanism involving intermediate formation of CAA voured with BMA. In all likelihood, the proposed hydro-
by a reaction of CMA with adenosine. If this kind of mech- lytic deformylation is base-catalyzed, proceeding by initial
anism is followed, the maximum concentration of CAA pre- deprotonation of one of the hydroxy functions of the cova-
sent is of the order of 0.1 m, i. e. equal to the initial con- lent hydrate, and hence the formation of the formyletheno
centration of adenosine. Otherwise adenosine should pro- adduct is most marked under acidic conditions.
duce CAA catalytically, which means that adenosine should
It is also worth noting that if the reaction were initiated
be rapidly regenerated from N6-formyladenosine. This can- by displacement of the halogen substituent of CMA or
not be the case, since the N-formyl derivatives of pyrimid- BMA with the N1 atom of the adenine ring, and not by the
ines are known to be hydrolytically rather stable.[17] It is attack of the 6-amino function on the carbonyl carbon
also worth noting that no sign of accumulation of N6-for- atom, N6,N6-dimethyladenosine could be expected to be al-
myladenosine during a kinetic run could be detected. Ac- kylated by CMA and BMA. No such reaction was ob-
cordingly, the pseudo first-order rate constant for the served.
formation of 4a from CMA and adenosine may be estim-
The structural information obtained on the intermediates
ated to be of the order of 10Ϫ7 sϪ1 at pH 4.7, if one assumes formed upon reacting N6-methyladenosine (3c) with BMA
that 4a is actually formed by the reaction of CAA with is consistent with the mechanistic suggestion depicted in
adenosine (refers to [CAA] ϭ 0.1 m). The rate constant Scheme 3. The major intermediate formed was identified as
1
observed for the reaction between CMA and adenosine un- the hydrated etheno product (7) by HPLC-MS, and by H
der these conditions is, however, 3 ϫ 10Ϫ6 Ϫ1. In other and 13C NMR spectroscopy. Another minor intermediate
s
words, the mechanism depicted in Scheme 2 for the forma- that could be isolated and characterized was the formyl-
tion of the etheno product cannot be the major pathway. etheno product 8. Furthermore, an early intermediate
The mechanism where the halomalonaldehyde spontan- having m/z ϭ 352 was detected by HPLC-MS. This mass
eously releases haloacetaldehyde seems equally unlikely. number can naturally be attributed to a number of different
The stability of BMA was studied by NMR spectroscopy structures, but it is clear that this intermediate has already
in several different buffer solutions from pH 4.5 to 7.2 at lost the bromine at C2, since the isotope peaks resulting
60 °C. Under these conditions, no decomposition was ob- from the presence of bromine were not observed. The dif-
served. The results of the kinetic experiments are also in- ferences between the mass number of this intermediate and
consistent with this mechanism. The formation of 1,N6- those of 7 and 8 are 28 and 18, respectively. Such differences
ethenoadenosine (4a) follows first-order kinetics. This could may be accounted for by replacement of one hydrogen atom
not be the case if the haloacetaldehyde was formed simul- in 7 with a formyl group and addition of a molecule of
taneously with 4a; the continuous increase of the concentra- water to 8. Accordingly, the early intermediate may have
tion of one of the starting materials (BAA or CAA) could structure 9.
be expected to lead to a sigmoidal dependence of the con-
In conclusion, it seems that the reactions of adenine base
centration of 4a on time.
with halogenated acetaldehydes and malonaldehydes pro-
Our results suggest that rather than being formed ceed through similar pathways. The exocyclic amino func-
through two separate pathways, the etheno and formyl- tion of adenine first attacks the carbonyl carbon atom of
etheno products are formed through a common initial inter- the aldehyde. While this step is rate-limiting in reactions
mediate. The pH-rate profiles obtained for the overall reac- with acetaldehyde derivatives, in reactions with malonal-
tion of the adenine substrates with BMA and CMA are dehyde derivatives the first step can be regarded as a pre-
rather similar (Figure 2). The profiles differ only slightly equilibrium. In the case of haloacetaldehydes, the cycliza-
with respect to the position of the rate maximum, which tion and the subsequent dehydration yield the etheno prod-
can probably be attributed to higher pKa value of CMA. uct. With halogenated malonaldehydes, the acyclic interme-
Yet at a low pH, a large proportion of the reaction of aden- diate may react in two different ways: A deformylation, fol-
osine with BMA yields the formyletheno product, while lowed by cyclization and dehydration yield the etheno prod-
only a barely detectable amount of this product is formed ucts, whereas cyclization and dehydration of the initial
with CMA. The fact that the shape of the pH-rate profiles intermediate result in the formation of a formyletheno de-
of the overall disappearance of the adenine substrate re- rivative. It hence seems that not only α-haloacetaldehydes,
mains similar with BMA and CMA, even though the pH- but also other halogenated carbonyl compounds may have
dependent product distribution is different, suggests that mutagenic properties under physiological conditions. Such
the reaction pathways leading to the two products branch compounds are, for example, α-halomalonaldehydes,
Eur. J. Org. Chem. 2000, 2315Ϫ2323
2321

S. Mikkola, N. Koissi, K. Ketomäki, S. Rauvala, K. Neuvonen, H. Lönnberg
FULL PAPER
(1.51 mmol; 230 mg) and 9-methyladenine (1.50 mmol; 230 mg) in
acetic acid buffer (pH 4.7, 100 mL), was heated for three days at
60 °C, and for one day at 90 °C. The progress of the reaction was
monitored by RP-HPLC. The reaction solution was then concen-
trated, and purified on a semipreparative RP-HPLC, using a Li-
Chrospher RP-18 (250 ϫ 10 mm, 5 µm) column. The eluent employed
was a mixture of 0.1  NH4OAc and acetonitrile (88:12). The col-
lected fractions were concentrated, and desalted on the same semi-
preparative column. The eluent was aqueous acetonitrile con-
taining 14% and 16% acetonitrile for purification of 4b and 5b,
known to be formed as decomposition products of α-hy-
droxy furanones.[3] In in vitro experiments, however, the
malonaldehyde derivatives appear to be less reactive than
the corresponding acetaldehyde compounds. As for the
chemical modification of DNA bases, the reaction with bro-
moacetaldehyde under slightly acidic conditions seems to
be the most efficient way to produce ethenobases, whereas
bromomalonaldehyde below pH 3 gives the highest yield of
formyletheno products.
1
respectively. For 4b: H NMR (D2O): δ ϭ 8.65 (1 H, s, H-5); 8.07
(1 H, s, H-2); 7.65 (1 H, s, H-8); 7.30 (1 H, s, H-7); 3.62 (3 H, s,
1
CH3). HPLC-MS: m/z 174.2 (M ϩ 1). Ϫ For 5b: H NMR (D2O,
Experimental Section
ppm from TMS): δ ϭ 9.70 (1 H, s, CHO); 9.67 (1 H, s, H-5); 8.26
(1 H, s, H-2); 8.13 (1 H, s, H-8); 3.79 (3 H, s, CH3). Ϫ HPLC-MS:
m/z ϭ 202.0 (M ϩ 1). Ϫ HRMS of 5b: HMRS(EI) Mϩ found
201.064970, C9H7N5O requires 201.065060. The UV spectra of 4b
and 5b were identical to those of the corresponding ribosides 4a
and 5a, respectively.
Equipment: The kinetic experiments were carried out by using
PerkinϪElmer Integral 4000 HPLC equipped with a diode array
detector. HPLC-MS analysis was performed on a PerkinϪElmer
Sciex API 365 LC/MS/MS triple quadrupole mass spectrometer.
A JEOL JNM-A 500 Fourier Transform NMR spectrometer was
employed for the NMR characterization of the isolated com-
pounds. A VG Analytical organic Mass Spectrometer and a Fisons
Intruments VG ZapSpec were used for HRMS and FAB-MS ana-
lysis, respectively.
7,8-Dihydro-8-hydroxy-9-methyl-3-(β-d-ribofuranosyl)imidazo[2,1-
i]purine (7): N6-Methyladenosine (3c; 0.071 mmol; 20 mg) was re-
acted with BMA (2b; 0.115 mmol; 17.4 mg) in aqueous solution
(3.4 mL) at pH 4.5. After 48 h incubation under nitrogen at 60 °C,
the mixture was cooled and the major product (7) was separated
by RP-HPLC. 1H NMR ([D6]DMSO): δ ϭ 8.88 (s, 1 H, H-5); 8.83
Materials: The solutions of chloroacetaldehyde were prepared from
a commercially available (Merck) 45% aqueous stock solution, the
concentration of which was determined as reported in the literat-
ure.[18] Bromoacetaldehyde was synthesized from diethyl acetal as
described before.[12] The preparation of chloromalonaldehyde[19]
and bromomalonaldehyde[20] has been reported previously. The
buffer constituents employed were of reagent grade.
(s, 1 H, H-2); 6.00 (d, 1 H, 1J ϭ 5.2 Hz, H-1Ј); 5.89 (dd, 1 H, J ϭ
1
1
4.4 and 7.0 Hz, H-8); 4.77 (dd, 1 H, J ϭ 12.5 and 7.0, H-7); 4.48
(m, 2 H, H-7 and H-2Ј); 4.16 (dd, 1 H, 1J ϭ 4.2 and 4.2 Hz, H-3Ј);
1
3.99 (m, 1 H, H-4Ј); 3.68 (dd, 1 H, J ϭ 12.1 and 3.8 Hz; H-5Ј);
3.57 (dd, 1 H, J ϭ 12.1 and 3.2 Hz, H-5ЈЈ); 3.57 (s, 3 H, CH3). Ϫ
1
13C NMR ([D6]DMSO, ppm from TMS): δ ϭ 149.9 (C-2); 148.2
(C-9a); 144.9 (C-3a); 143.5 (C-9b); 116.1 (C-5); 88.1 (C-1Ј); 85.9
(C-4Ј); 84.6 (C-8); 74.5 (C-2Ј); 69.9 (C-3Ј); 60.8 (C-5Ј); 54.8 (C-7),
30.3 (CH3). FAB-MS: m/z ϭ 324 (Mϩ, 100%); 192 (71%).
Adenosine (3a) and 1,N6-Ethenoadenosine (4a) were products of
Sigma, and were used as received. 11-Formyl-1,N6-ethenoadenos-
ine (5a) was a generous gift of Dr Kronberg, Abo Akademi Univer-
˚
sity, Turku. 9-Methyladenine was synthesised as described previ-
ously, and purified by repeated washes with a mixture of hot acet-
one and 40% aqueous tetrabutylammonium hydroxide (98:2, v/
v).[21] The compound exhibited a molecular ion signal [M ϩ 1]ϩ at
m/z ϭ 149.2 in HPLC-MS. The 1H NMR and UV spectra were
identical to those reported previously.[22]
Kinetic Measurements: The reaction solutions were prepared in
sterilized water. The pH was adjusted with formic acid (pH 3Ϫ4),
acetic acid (pH 4Ϫ5.2), MES (pH 5.2Ϫ6.5) or HEPES (pH
6.5Ϫ7.5) buffers. Below pH 3, aqueous solutions of hydrochloric
acid were employed. The pH of the buffer based reaction solutions
was measured on a pH meter at 60 °C prior to the kinetic experi-
ments. The ionic strength was adjusted with NaNO3.
N6-Methyladenosine (3c): N6-Methyladenosine was synthesised by
a procedure modified from that of Johnson et al.[23] Accordingly, a
solution of 6-chloropurine riboside (0.78 mmol; 250 mg) in aque-
ous methylamine (40%, 12.5 mL) was heated in a Parr reactor at
70 °C and 1.5 bar for 16 hours. The solution was concentrated
down to a small volume, and the precipitate was filtered. The fil-
trate was evaporated to dryness at 40 °C, and the residue was
recrystallized from methanol. The yield of pure crystalline product
was 141 mg (56%). The methanolic solution was evaporated, and
the recrystallization procedure was repeated. After three days at
The reactions of adenosine (3a), 9-methyladenine (3b), N6-methyl-
adenosine (3c) and 1,N6-ethenoadenosine (4a) were followed at 60
°C. Reactions were carried out in stoppered tubes, which were im-
mersed in a thermostated water bath. The initial concentration of
the starting material was 0.1 m. Aliquots of 100 µL were with-
drawn at appropriate intervals to cover approximately two half-lives
of the reaction. The aliquots were cooled down in an ice bath to
quench the reaction. When the aliquots could not be analyzed im-
mediately, they were frozen until analyzed.
1
Ϫ25 °C, more pure product (80 mg, 30%) was obtained. H NMR
([D6]DMSO): δ ϭ 8.32 (s, 1 H, H-2); 8.21 (s, 1 H, H-8); 7.77 (br s,
The HPLC analysis was performed on Hypersil ODS RP-18 col-
umn (250 ϫ 5 mm, 5µm particle size). The eluents were mixtures
of acetonitrile and 0.05  acetic acid buffer (pH 4.3, I ϭ 0.1  with
NH4Cl). The acetonitrile content varied depending on the sub-
strate: with adenosine (3a) it was 5%, with ethenoadenosine (4a)
3.5%, and with N6-methyladenosine (3c) 10%. Aliquots from reac-
tions of 9-methyladenine were analyzed by using a mixture of 0.1
 NH4OAc and acetonitrile (93:7, v/v). UV-detection at wave-
lengths of 260 and 325 nm was employed.
1 H, NH); 5.87 (d, 1 H, 1J ϭ 5.6 Hz, H-1Ј); 5.40 (m, 2 H, 2 ϫ OH);
1
5.14 (d, 1 H, OH); 4.61 (dd, 1 H, J ϭ 4.7 Hz, H-2Ј); 4.15 (dd, 1
H, H-3Ј); 3.96 (ddd, 1 H, H-4Ј); 3.67 (m, 1 H, H-5Ј); 3.56 (m, 1 H,
H-5ЈЈ); 2.95 (d, 3 H, CH3). 13C NMR ([D6]DMSO, ppm from
TMS): δ ϭ 155.2 (C-6); 152.5 (C-2); 148.1 (C-4); 139.7 (C-8); 119.9
(C-5); 88.0 (C-1Ј); 86.0 (C-4Ј); 76.3 (C-3Ј); 70.7 (C-2Ј); 61.5 (C-5Ј);
34.2 (CH3).
3-Methylimidazo[2,1-i]purine (1,N6-Etheno-9-methyladenine; 4b)
and 7-Formyl-3-methylimidazo[2,1-i]purine (1,N6-Etheno-11-formyl-
The concentrations of the adenine substrate (3a,b), the etheno
9-methyladenine; 5b):
A
solution of bromomalonaldehyde product (4a,b) and the formyletheno product (5a,b) were deter-
2322
Eur. J. Org. Chem. 2000, 2315Ϫ2323

Reactions of Adenosine with Bromo- and Chloromalonaldehydes
FULL PAPER
Swenberg, N. Fedtke, F. Ciroussel, A. Barbin, H. Bartsch, Car-
mined by using calibration curves. The sum of the concentrations
generally remained constant during the reaction. As the breakdown
of the etheno products is an acid-catalyzed process, a slight de-
crease in the total concentration (substrate ϩ etheno products) was
observed under acidic conditions.
[2d]
cinogenesis 1992, 13, 727Ϫ729. Ϫ
V. Nair, R. J. Offerman,
G. A. Turner, J. Org. Chem. 1984, 49, 4021Ϫ4025.
[3a] L. Kronberg, R. Sjöholm, S. Karlsson, Chem. Res. Toxicol.
[3]
[4]
[5]
1992, 5, 852Ϫ855. Ϫ [3b] L. Kronberg, S. Karlsson, R. Sjöholm,
[3c]
Chem. Res. Toxicol. 1993, 6, 495Ϫ499. Ϫ
D. Asplund, L.
Kronberg, R. Sjöholm, T. Munter, Chem. Res. Toxicol. 1995,
8, 841Ϫ846.
Calculation of the Rate Constants: Pseudo first order rate constants
for the reaction of adenosine (3a), 9-methyladenine (3b), N6-
methyladenosine (3c) and 1,N6-ethenoadenosine (4a) were calcu-
lated by using the integrated first-order rate law. The calculation
was based on the decrease of the concentration of the starting mat-
erial as a function of time.
[4a] A. Barbin, H. Bresil, A. Croisy, P. Jacquinon, C. Malaveille,
R. Montesano, H. Bartsch, Biochem., Biophys. Res. Commun.
1975, 67, 596Ϫ603. Ϫ [4b] C. Malaveille, H. Bartsch, A. Barbin,
A. M. Camus, R. Montesano, A. Croisy, P. Jacquinon, Bio-
chem. Biophys. Res. Commun. 1975, 63, 363Ϫ370.
N. K. Kochetkov, V. N. Shibaev, A. A. Kost, Tetrahedron Lett.
1971, 22, 1993Ϫ1996.
[6] [6a]
J. A. Secrist III, J. R. Barrio, N. J. Leonard, G. Weber,
The rate constants obtained for the disappearance of adenosine
and 9-methyladenine in the presence of BMA were divided into the
rate constants for the formation of the etheno product (4a,b) and
formyletheno product (5a,b) by Equation (1) and (2).[24] Here k1
refers to the formation of an etheno product, 4a or 4b, k2 to the
formation of a formyletheno product, 5a or 5b, k3 to the overall
disappearance of the starting material, 3a or 3b, and kd2 to the
disappearance of the etheno product.
[6b]
Biochemistry 1972, 11, 3499Ϫ3506. Ϫ
J. R. Barrio, J. A.
Secrist III, N. J. Leonard, Biochem. Biophys. Res. Commun.
1972, 46, 597Ϫ604.
[7] [7a]
J. Biernat, J. Ciesiolka, P. Gornicki, R. W. Adamiak, W.
Krzyzosiak, M. Wiewiorowski, M., Nucleic Acids Res. 1978, 5,
[7b]
789Ϫ804. Ϫ
W. J. Krzyzosiak, J. Biernat, J. Ciesiolka, P.
Gornicki, M. Wiewiorowski, Polish J. Chem. 1979, 53,
[7c]
243Ϫ252. Ϫ
W. J. Krzyzosiak, J. Biernat, J. Ciesiolka, P.
Gornicki, M. Wiewiorowski, Polish J. Chem. 1983, 57,
[7d]
779Ϫ787. Ϫ
P. D. Sattsangi, J. R. Barrio, N. J. Leonard, J.
x(4a,b) ϭ [k1/(k3Ϫkd2)] [exp(Ϫkd2t)Ϫexp(Ϫk3t)]
k3 ϭ k1 ϩ k2
(1)
(2)
Am. Chem. Soc. 1980, 102, 770Ϫ774.
[8] [8a]
F. Oesch, G. Doerjer, Carcinogenesis 1982, 3, 663Ϫ638. Ϫ
[8b]
J. T. Kusmierek, B. Singer, Chem. Res. Toxicol. 1992, 5,
[8c]
634Ϫ638. Ϫ
F. P. Guengrich, M. Persmark, Chem. Res.
With adenosine, the calculation was based on the mole fraction of
ethenoadenosine (4a) as a function of time. The rate constants for
the overall disappearance of adenosine and the disappearance of
4a were known, and the rate constant for the formation of 4a was
obtained by nonlinear fitting. In the case of 9-methyladenine (3b),
only the rate constant for the overall disappearance of the starting
material was known, and the rate constants for the formation and
disappearance of the products were obtained by fitting.
Toxicol. 1994, 7, 205Ϫ208.
[9]
C. Reichardt, K. Halbritter, Angew. Chem. 1975, 87, 124Ϫ132;
Angew. Chem. Int. Ed. Engl. 1975, 14, 86Ϫ94.
L. Kronberg, D. Asplund, J. Mäki, R. Sjöholm, Chem. Res.
Toxicol. 1996, 9, 1257Ϫ1263.
K. F. Yip, K. C. Tsou, Tetrahedron Lett. 1973, 33, 3087Ϫ3090.
K. Kayasuga-Mikado, T. Hashimoto, T. Negishi, K Negishi,
H. Hajatsu, Chem. Pharm. Bull. 1980, 28, 932Ϫ938.
A. H.-J. Wang, L. G. Dammann, J. R. Barrio, I. C. Paul, J.
Am. Chem. Soc. 1974, 96, 1205Ϫ1213.
E. K. Euranto, in The Chemistry of Carboxylic Acids and Es-
ters, S. Patai, Ed., Interscience Publishers, John Wiley & Sons
Ltd. 1969, pp. 518.
H. Lönnberg, in Biocoordination Chemistry: Coordination Equi-
libria in Biologically Active Systems, K. Burger, Ed., Ellis Hor-
wood, Chichester 1990, pp. 284Ϫ346.
[10]
[11]
[12]
[13]
[14]
Acknowledgments
˚
The authors wish to thank Dr. Kronberg (Abo Akademi Univer-
[15]
[16]
sity, Turku) for the generous gift of 1,N6-formylethenoadenosine,
and Mr. Kristo Hakala, M. Sc., for the MS analyses.
R. C. Moschel, N. J. Leonard, J. Org. Chem. 1976, 41,
294Ϫ300.
[1] [1a]
B. Singer, S. J. Spengler, F. Chavez, J. T. Kusmierek, Carci-
[17]
[18]
P. Lehikoinen, H. Lönnberg, Chem. Scripta 1986, 26, 103Ϫ108.
W. P. McCann, L. M. Hall, W. K. Nododez, Anal. Chem. 1983,
55, 1454Ϫ1455.
W. Dieckmann, L. Platz, Chem. Ber. 1904, 37, 4638Ϫ4646.
S. Trofimenko, J. Org. Chem. 1963, 28, 3243Ϫ3245.
M. Hedayatullah, J. Heterocycl. Chem. 1982, 19, 249Ϫ251.
H. Lönnberg, J. Ylikoski, J. Arpalahti, E. Ottoila, A.Vesala,
Acta Chem. Scand. 1985, A39, 171Ϫ180.
[1b]
nogenesis 1987, 8, 745Ϫ747. Ϫ
M. M. Mroczkowska, J. T.
[1c]
Kusmierek, Mutagenesis 1991, 6, 385Ϫ390. Ϫ
B. Singer, J.
T. Kusmierek, W. Folkman, F. Chavez, M. K. Dosanjh, Carci-
[19]
[20]
[21]
[22]
nogenesis 1991, 12, 745Ϫ747. Ϫ [1d] K. C. Cheng, B. D. Preston,
D. S. Cahill, M. K. Dosanjh, B. Singer, L. A. Loeb, Proc. Natl.
[1e]
Acad. Sci. U.S.A 1991, 88, 9974Ϫ9978. Ϫ
H. Bartch, A.
Barbin, M. J. Marion, J. Nair, Y. Guillard, Drug. Metab. Rev.
1994, 26, 349Ϫ371; M. Moriya, W. Zhang, F. Johnson, A. P.
[23]
[24]
Grollman, A.P., Proc. Natl. Acad. Sci. U.S.A. 1994, 91,
J. A. Johnson Jr., H. J. Thomas, H. J. Schaeffer, J. Am. Chem.
Soc. 1958, 699.
[1f]
11899Ϫ11903. Ϫ
G. A. Pandya, M. Moriya, Biochemistry
1996, 35, 11487Ϫ11492.
N. M. Rodiguin, E. N. Rodiguina, Consecutive Chemical Reac-
tions, Van Nostrand Company Inc., New York, 1964, p. 49.
Received August 11, 1999
[2] [2a]
N. Fedtke, J. A. Boucheron, M. J. Turner Jr., J. A. Swen-
[2b]
berg, Carcinogenesis 1990, 11, 1279Ϫ1285. Ϫ
rich, F.P., Chem. Res. Toxicol. 1992, 5, 852Ϫ855. Ϫ
F. P. Gueng-
[2c]
J. A.
[O99493]
Eur. J. Org. Chem. 2000, 2315Ϫ2323
2323