N. Moons et al. / Tetrahedron Letters 53 (2012) 6806–6809
6807
method.16 To facilitate the purification of all intermediates, we first
O
converted both steviol and isosteviol into their corresponding
methyl esters using Cs2CO3 and MeI, instead of the known proce-
dure with ethereal diazomethane.17 Alternatively, the procedure
with methyl tosylate can also be used.18 The carbonyl ene reaction
on the methyl ester 6 was first evaluated. There are numerous
examples of this reaction in the literature,19 and recent examples
on terpenes as substrates or as target molecules.20 We first tried
the carbonyl ene reaction with ethyl glyoxalate. Without any cata-
lysts, even during heating at higher temperatures (>140 °C) no
reaction was observed. Although this reaction is usually successful
when using strong Lewis acid catalysts,19 the rearrangement of
steviol ester 6 to its isosteviol analogue was predominant for most
conventional acids used in carbonyl ene reactions. To avoid this
side reaction, we tried some milder acids and reaction conditions,
and ZnBr2 was found to be the best acid catalyst for the carbonyl
ene reaction of steviol. The reaction was performed with equimolar
amounts of ZnBr2 at room temperature for 2 days. Higher temper-
atures increased the amount of undesired isosteviol in the reaction
mixture. The reaction conditions were optimized for the reaction
O
OH
OH
(S)
H
H
O
PTSA
H
H
OEt
HO
Δ
7
9
CO2Me
CO2Me
Scheme 2. Formation of a novel lactone 9. Major constituent shown.
Ph
O (R)
OH
O
H
H
O
PTSA
H
H
Ph
HO
Δ
7
10
CO2Me
CO2Me
Scheme 3. Cyclization of the a-hydroxy ketone 7 (main diastereomer shown).
with ethyl glyoxalate, and the
a-hydroxy ester 7 was isolated as
a mixture of isomers (85:15) in a total yield of 85%. These reaction
conditions were slightly modified for the reaction of steviol ester 6
with phenylglyoxal, resulting in a comparable yield (86%). Surpris-
ingly, for phenylglyoxal only one isomer 8 was formed after 3 days
(Scheme 1). As our initial idea was to dehydrate both isomers to
2-phenyldihydro-2H-pyran-3(4H)-one (Scheme 3). The formation
of this particular ring can be explained by an addition of the C13
hydroxyl group on the ketone, followed by dehydration of the
formed hemiacetal and keto–enol tautomerization. We were able
to assign the exact configuration of the stereogenic center in the
newly attached ring as the R-configuration. In this configuration,
the phenyl ring is in the equatorial position, which can also be
determined via cross couplings observed in the NOESY spectrum.
For validation of our proposed structure, crystallographic X-ray
measurements were obtained. The corresponding 3D structure is
shown in Fig. 2, and confirms our hypothesis.
their corresponding dienes, no structural elucidation of either
a-
hydroxy ester 7 or the
a-hydroxy ketone 8 was carried out
(Scheme 1).
However, upon examining various procedures for the dehydra-
tion of products 7 or 8, we were surprised to find that a cyclization
had occurred for both products. Compound 7 was cyclized to the
corresponding lactone 9 by refluxing in toluene with 0.1 equiv of
p-toluenesulfonic acid for 40 minutes. The lactone was isolated
in a good yield of 59%, however, the sample contained 2 diastere-
oisomers in a ratio of 88:12. Their relative amounts were easily cal-
culated via 1H NMR integration. The absolute configuration of both
Pericyclic reactions on isosteviol
As for steviol 1, the ent-beyerane isosteviol 3 was first converted
into its methyl ester 1121 by the same procedure. This ester was
then submitted to the Grignard reaction with vinyl magnesium bro-
mide. As the ester of isosteviol is rather sterically hindered, the
Grignard reaction was performed with multiple equivalents of
Grignard reagent. Although this reaction seems quite straightfor-
ward, the yield of vinyl adduct 12 was rather low (34%). This was
probably due to the instability of the adduct, which seemed to
dehydrate and rearrange to multiple unidentifiable side products.
Therefore, other pathways to obtain this compound were also
investigated. As the addition of ethynyl magnesium bromide to es-
ter 11 was significantly more successful, this ethynyl adduct 13 was
reduced with Lindlar’s catalyst, resulting in the same product 12, al-
beit with a higher total yield of 68%. (Scheme 4). We continued by
investigating the dehydration of this vinylated compound 12. Var-
ious methods were tried, but the outcome of these reactions varied
significantly. In the presence of a nucleophile, even a weak one, the
dehydration often resulted in an addition at the formed carbocat-
ion, rather than the formation of the desired diene. For example,
we were able to characterize the addition of toluene to the vinyl ad-
duct 12 by refluxing in the presence of PTSA. As an alternative route,
the dehydration of acetylene derivative 13 was investigated. The
outcome for these reactions when using acid (e.g. 98% sulfuric acid
or 85% phosphoric acid) was roughly the same. Surprisingly, when
using thionyl chloride as a dehydrating agent, a chloroallene22 14
was formed instead of the desired dehydrated compound (Scheme
4). We were able to optimize this reaction to a yield of 52%, yet no
further reactions on the allene 14 were carried out.
a
-hydroxy esters was determined via the 1H coupling constants of
the lactone ring. For the main constituent of the mixture, the
hydrogen atom in -position of the ketone group displayed both
a
an axial (12.6 Hz) and an equatorial (6.6 Hz) coupling constant,
therefore the hydroxyl group must be in equatorial position. Based
on these spectroscopic data, for the major product, the stereogenic
center in the lactone ring corresponds with an S-configuration
(Scheme 2). Despite several attempts using different separation
methods including HPLC, we were unable to separate both
diastereomers.
When submitting the phenyl ketone 8 to the same conditions,
the reaction outcome was a crystalline product 10. CIMS measure-
ments revealed a loss of water, while 1H NMR showed the presence
of two diastereoisomers in a 93:7 ratio. We were able to isolate the
major constituent from the mixture via recrystallization in diethyl
ether to afford a yield of 81%, and its structure was determined via
1D and 2D NMR techniques. Compound 10 was found to be a
OH
OH
O
H
H
O
R
O
R
H
H
HO
ZnBr2
CO2R
1: R=H
CO2Me
Cs2CO3
MeI
85%
86%
7: R=OEt
8: R=Ph
6: R=Me
As the instability of the intermediate carbocation after removal
of the hydroxyl group was obvious, we envisaged a one pot dehy-
Scheme 1. Carbonyl ene reactions of steviol methyl ester 6.