An experimentum crucis for testing purine vs imidazo-
[1,5-c]imidazole formulas is found in tracing the conversions
of discretely labeled 3,9-dimethyl[5-13C]uric acid ([5-13C]-
1) by NMR, for the species in the two projected pathways
clearly differ in connectivity and hybridization of their
labeled carbons (Figure 1 and Scheme 1). 3,9-Dimethyl[5-
13C]uric acid ([5-13C]-1) was synthesized using a modification
of literature procedures (see Supporting Information).3d,9
Generation of the initial intermediates was carried out by
introducing chlorine into a suspension of [5-13C]-1 (158 mg,
0.8 mmol) in dry acetic acid-d4 (0.6 mL) under strictly
anhydrous conditions. After all the solid had gone into
solution (10 min), the mixture was purged with argon and
transferred to a 5 mm NMR tube. The initial spectrum
showed only a broad 13C peak at about 85.0 ppm, exhibiting
a barely resolved substructure, and probably consisted of
several peaks. This peak gradually decreases, and new peaks
grow, first at 82.1 ppm, and then a smaller one at 79.3 ppm
(Figure 2). The initial species reveal NMR signals charac-
isocyanate 4 and rotation about the C5-C6 bond (ca. 100°)
to place C-2 in juxtaposition to N-7 for closure to 5. Support
for the rearrangement from 2 (or a diadduct 3) to 5, or the
competing ring contraction to 6, comes from the fact that it
has proved possible to trap a species in an early stage of the
conversion that still contained an intact purine ring. Forma-
tion of the cis-4,5-dimethoxy-3,9-dimethyltetrahydropurine-
2,6,8-trione (5c) upon treatment with dry methanol clearly
differentiates the initial chloro-adduct(s) from the rearranged
product 5 that does not react. Since this behavior eliminates
structure III as the end product,11 the problem reduced itself
to tracing the label in the reductive dechlorination product
of [7a-13C]-5. NMR results conclusively ruled out the formula
IV and supported structure [7a-13C]-7. In an aprotonic solvent
such as DMSO-d6, it exists in the enediamine tautomeric
form that exhibits a signal at 88.3 ppm. However, in an acidic
solution (TFA-d), the amidine tautomeric form, being
stabilized by amidinium resonance not available to the
enediamine, exhibits a peak at 61.4 ppm; removal of solvent
and redissolving in DMSO-d6 gave again the signal at 88.2
ppm. The reverse conversion of 7 to 5 by chlorination and
the hydrolysis to [7a-13C]-8, showing a peak at 62.7 ppm
(see Supporting Information), conforms to the pathway which
is, indeed, a case of oxidative skeleton rearrangement.
The strained bicyclic system 8, with imide nitrogen at the
bridgehead, undergoes an interesting decay upon heating in
water. This reaction, involving initial ring opening at C-7
or C-1 and decarboxylation to a mixture of isomeric
carbamoylhydantoins 9 and 10, provides clean degradative
evidence for the structure 8 (see Supporting Information).
Retrograde thermal skeleton rearrangement of 7, which had
been so misleading, implies an alternative ring opening and
reclosure to the more stable 1, probably also through an
intermediary isocyanate (Scheme 1).
Accordingly, even with favorable N-blocking in 1, no
species resembling II and IV could be detected along the
reaction path. However, several features of this essentially
ad hoc assignment caused us to question such proposals. This
species, if formed, would likely be extremely short-lived,
due to the inherent instability of the iminium functionality
flanked by two carbonyls. Probably the most serious in-
consistency is that in reactions with uric acids or 8-oxo-
guanines, the electrophile heads for C-5, the site of highest
electron density, to give the most stable amidinium inter-
mediate, which can then be intercepted by a nucleophile.
Thus, no matter how reactive the iminium species might be,
this is an unlikely incipient intermediate. Not only was the
probability of an iminium species arising from the diadduct
3 viewed as low, but also if such were to occur, there would
be no obvious route for the formation of the observed
products. It should be pointed out that chlorination in
methanol affords cis-4,5-dimethoxytetrahydropurine-2,6,8-
triones,12 with no traces of any trans isomer. Stereospecific
syn addition is due to the geometric constraints imposed by
Figure 2. Disappearance of transient intermediate(s) at 85.0 ppm
and increase of 5 at 82.1 ppm and 6 at 79.3 ppm with time (20, 50,
90, and 120 min after start of chlorination of [5-13C]-1).
teristic of sp3 carbon that eliminate 4-chloro-3,9-dimethyl-
4,9-dihydro-3H-purine-2,6,8-trione (II) as a possible inter-
mediate.10 The peaks in the final spectrum were assigned to
the main crystalline product [7a-13C]-5, which separated after
NMR acquisition, and to the byproduct [5-13C]-6 isolated
from the mother liquor. A reasonable mechanism would
involve a transient amidinium intermediate 2, which is
capable of acting as an ambident electrophile. One possible
route would then result from the C2-N3 bond cleavage to
(11) Species such as [5-13C]-III (6a, Scheme 1) could, of course, account
for the observed broad signal at ca. 85 ppm due to transient intermediate(s),
but it is clear that such a structure, characterized by its inherent reactivity
in nucleophic substitutions (ref 5), cannot explain the observed inertness
of the actual chlorination end product toward methanol.
(10) Positions of appropriate chemical shift (60-90 ppm) values clearly
demonstrate the presence of tetrasubstituted carbons; imine carbon would
be expected to resonate above 150 ppm.
Org. Lett., Vol. 5, No. 23, 2003
4267