Interestingly, both the γ-butyrolactones and MMFs are
biosynthetically derived from the same starting materials,
dihydroxyacetone phosphate and a coenzyme A ꢀ-ke-
tothioester (Scheme 1). These substrates are condensed by
distinct enzymes belonging to the AfsA superfamily.3 Based
on the inherent reactivity of 1,3-dicarbonyl compounds and
ketones, we suspected that catalysts used in synthetic organic
chemistry could be to used to effect condensations reminis-
cent of the enzymatic reactions in the biosyntheses of the
signaling molecules. In model studies, we tested this
hypothesis using a ꢀ-ketoester (methyl 3-oxononanoate) in
place of the coenzyme A ꢀ-ketothioester and ketal-protected
dihydroxyacetone in place of dihydroxyacetone phosphate.
Our expectation was that these starting materials would
undergo a Knoevenagel-type reaction to give an R,ꢀ-
unsaturated dicarbonyl compound that would subsequently
lactonize upon ketal deprotection, yielding the butenolide
intermediate (1) in γ-butyrolactone biosynthesis (Scheme 1).
In initial condensations, we examined the catalytic activity
of piperidine, ammonium acetate, and diethylamine, which
are known to catalyze carbon-carbon bond-forming reac-
unsaturated dicarbonyl compound product was not observed;
however, a methyl 2-alkyl-4-hydroxymethylfuran-3-carboxylate
product was isolated in good yield. This product has the identical
skeleton observed in the S. coelicolor MMF signaling mol-
ecules. In subsequent experiments, we found that this reaction
is also catalyzed by ytterbium triflate but not significantly by
bismuth triflate or zinc chloride (Table 1). In the absence of
scandium or ytterbium triflate, no reaction of the substrates was
observed. Apparently, Sc(OTf)3 and Yb(OTf)3 catalyze the
condensation and dehydrative aromatization of dihydroxyac-
etone and ꢀ-ketoesters.
Given the mechanistic curiosity of the metal-catalyzed
condensation and its potential use in the chemical synthesis
of the MMFs, we studied the reaction in more detail. The
structure of the product indicated that in situ deprotection
of the dihydroxyacetone ketal took place prior to furan
formation. Indeed, we observed rapid (<1 min) deprotection
of the dihydroxyacetone by scandium triflate in methanol
by thin-layer chromatography.8 The rate of deprotection is
clearly important because condensations in THF, where
deprotection is markedly slower, yield much less product
than those performed in methanol (42% vs 72% yield). Next,
we examined the importance of the order of substrate and
catalyst addition because both dihydroxyacetone and ꢀ-ke-
toesters are prone to self-condensation. We found that
premixing scandium triflate with either of the substrates
followed by addition of the other substrate led to an increase
in side reactions and a ∼30% decrease in furan yield. The
optimal reaction conditions involved premixing of the
substrates followed by addition of the catalyst. Then, we
investigated the importance of reactant stoichiometry given
the propensity of the ꢀ-ketoester and dihydroxyacetone to
self-condense. It was found that a 1:1 stoichiometry was
optimal. Finally, catalyst loadings were screened and the
reaction was found to be highest yielding when 10 mol %
of scandium triflate was used. Any more than 10 mol %
catalyst resulted in side reactions that lowered the yield, while
5 mol % led to a slower reaction rate. With these optimized
reaction conditions, we obtained methyl 2-hexyl-4-hy-
droxymethylfuran-3-carboxylate from ketal-protected dihy-
droxyacetone and methyl 3-oxononanoate in 72% yield.
Furan formation likely involves an aldol reaction of the
ꢀ-ketoester and dihydroxyacetone. To examine the substrate
scope of this reaction, we reacted a ꢀ-ketoester with two
different substrates containing R-hydroxy groups, hydroxy-
acetone, and glycolaldehyde (Table 2). These substrates
reacted to form the expected furans in 51 and 56% yields,
respectively. In contrast, only starting materials were recov-
ered from reactions with ketones lacking an R-hydroxy group
(cyclohexanone, cyclopentanone, or 3-pentanone). The ab-
sence of aldol products in those reactions suggested that the
R-hydroxy group is crucial for irreversible carbon-carbon
bond formation. Thus, we propose that under these conditions
ꢀ-ketoesters and R-hydroxycarbonyl compounds undergo a
domino reaction.5b,9 They first react in a reversible aldol
Table 1. Condensation Catalyst Screeninga
catalyst
loading (mol %)
% yield
Sc(OTf)3
Yb(OTf)3
Bi(OTf)3
ZnCl2
10
10
10
10
72
56
trace
trace
a Reactions were performed at room temperature for 15 h with equimolar
substrates.
tions of 1,3-dicarbonyls.5 No reaction was observed with any
of the amine catalysts. These results can be ascribed to the
lower reactivity of ketones relative to that of aldehydes in
Knoevenagel condensations.6 Next, we tested Lewis acidic
metals as condensation catalysts because they have been
shown to effect additions of 1,3-dicarbonyl compounds to
aldehydes or ketones.7 For practical and mechanistic consid-
erations, we focused our efforts on air- and moisture-insensitive
Lewis acid catalysts (Table 1). First, we reacted methyl
3-oxononanoate and ketal-protected dihydroxyacetone in metha-
nol using scandium triflate as a catalyst. The expected R,ꢀ-
(5) (a) Tietze, L. F.; Beifuss, U. The Knoevenagel Reaction. In
ComprehensiVe Organic Syntheses; Trost, B. M., Fleming, I., Eds.;
Pergamon: Oxford, 1991; Vol. 2, pp 341-394. (b) Tietze, L. F.; Rackelmann
N. The Domino Knoevenagel-Hetero-Diels-Alder Reaction and Related
Transformations. In Multicomponent Reactions; Zhu J., Bie, H., Eds.; Wiley-
VCH: Weinheim, 2005; Vol. 12, pp 1-168.
(6) Smith, M., March, J., Eds. March’s AdVanced Organic Chemistry:
Reactions, Mechanisms, and Structure, 5th ed.; John Wiley & Sons, Inc.:
New York, 2001.
(7) (a) Rao, P. S.; Venkataratnam, R. V. Tetrahedron Lett. 1991, 32,
5821–5822. (b) Narsaiah, A. V.; Nagaiah, K. Synth. Commun. 2003, 33,
3825–3832.
(8) The deprotection of acetals by scandium triflate is known. Oriyama,
T.; Watahiki, T.; Kobayashi, Y.; Hirano, H.; Suzuki, T. Synth. Commun.
2001, 31, 2305–2311.
Org. Lett., Vol. 11, No. 14, 2009
2985