Silylative Dieckmann-like cyclizations of ester-lmides (and diesters)

Article abstract of DOI:10.1021/ol061988q

Trialkylsilyl triflates effect cyclization of ester-imides such as 2 to produce adducts such as 4a. Trapping of the in situ generated, nucleophilic ketene acetal (cf. 5a) is a key aspect of the transformation. A range of substrates amenable to this operationally simple reaction is reported. In many instances the levels of diastereoselectivity are very high. Mechanistic points are inferred from spectroscopic observations.

Full text of DOI:10.1021/ol061988q

ORGANIC
LETTERS
2006
Vol. 8, No. 23
5191-5194
Silylative Dieckmann-Like Cyclizations
of Ester-Imides (and Diesters)
Thomas R. Hoye,* Vadims Dvornikovs, and Elena Sizova
Department of Chemistry, UniVersity of Minnesota, Minneapolis, Minnesota 55455
hoye@chem.umn.edu
Received August 10, 2006
ABSTRACT
Trialkylsilyl triflates effect cyclization of ester-imides such as 2 to produce adducts such as 4a. Trapping of the in situ generated, nucleophilic
ketene acetal (cf. 5a) is a key aspect of the transformation. A range of substrates amenable to this operationally simple reaction is reported.
In many instances the levels of diastereoselectivity are very high. Mechanistic points are inferred from spectroscopic observations.
Upon contemplating modes of construction of the hetero-
cyclic core of the telomerase inhibitor UCS1025A (1), we
considered the cyclization of an ester enolate equivalent with
an imide carbonyl group as in 2. The base sensitivity of 2
quickly revealed itself; no trace of an adduct such as 3
(Scheme 1) was ever observed under basic reaction condi-
serving the role of acceptor. This approach proved to be very
effective for generating species such as 4,2 and this tactic
was successfully implemented in two syntheses of 1.3
Silyl ketene acetals are typically generated by silylation
of an ester enolate with a silyl chloride. Attempts to prepare
them by reaction with silyl triflates in the presence of weak
bases [e.g., Et3N (TEA)], by analogy with silyl enol ether
formation, are usually thwarted by overconversion to the
R-silylated esters. However, in appropriate settings one can
imagine simultaneous silylative generation of both a SKA
and an activated electrophile, which would co-react faster
than conversion of the SKA to its isomeric R-silylated ester.
In particular, when ester-imide 2 was treated with TMSOTf
and TEA in chlorocarbon solvents (CH2Cl2 or CDCl3), the
bicyclic lactam 4a was produced in high yield and diaster-
eomeric ratio (dr) (Scheme 2). The endo orientation of the
carbomethoxy group in the predominant product, a fortunate
outcome vis-a`-vis synthesis of 1, was established on the basis
of a single-crystal X-ray analysis (Figure 1a) and eventual
elaboration into UCS1025A.3 The sense of diastereoselec-
tivity of this reaction is consistent with C-C bond formation
through an open transition state geometry for the addition
Scheme 1
tions. This turned our attention to a cationic process involving
a silyl ketene acetal (SKA) derived from the enolizable ester
in 2. SKAs serve as nucleophiles in various addition reactions
to CdX containing electrophiles, including aldehydes/
ketones, imines, enones, and acid halides to provide â-hy-
droxyesters, â-aminoesters, δ-ketoesters, and â-ketoesters,
respectively.1 We are not aware of an imide functional group
(1) Mahrwald, R. Chem. ReV. 1999, 99, 1095-1120.
(2) Dvornikovs, V.; Hoye, T. R. Abstr. Pap. Am. Chem. Soc. 2005, 229,
665-ORGN.
(3) (a) Lambert, T. H.; Danishefsky, S. J. J. Am. Chem. Soc. 2006, 128,
426-427. (b) Hoye, T. R.; Dvornikovs, V. J. Am. Chem. Soc. 2006, 128,
2550-2551.
10.1021/ol061988q CCC: $33.50
© 2006 American Chemical Society
Published on Web 10/11/2006

Scheme 2
Scheme 3
of an imide carbonyl group in 5, which allows the low
steady-state population of SKA to be dragged to the desired
products.
To probe if in situ activation of an ester electrophile might
also serve to induce a net, silylative Dieckmann cyclization,
we investigated simple dimethyl R,ω-dicarboxylates (Scheme
4). The reaction proved to be successul for five- and six-
of the SKA to an O-silylated imide carbonyl as suggested
in 7-endo (as opposed to the synclinal geometry of 7-exo).
This reaction could be extended to higher trialkylsilyl
triflates. TBSOTf and TIPSOTf gave products 4b and 4c,
respectively, with comparable outcomes; the only difference
was a progressively slower overall reaction rate (t1/2 ≈ 6
min, 4a; 15 min, 4b; and 7 h, 4c at [2] ) 0.3 M in CDCl3
and 1.5 equiv of R3SiOTf). In addition, we observed a low,
steady-state level of each of the intermediate SKAs 5b and
Scheme 4
membered ring formation, giving excellent yields of the
ketals 12 (dr 2:1) from dimethyl adipate and dimethyl
pimelate (11, n ) 1 and 2). No cyclization was observed
(1H NMR analysis) for substrates that would have given rise
to four-, seven-, and eight-membered rings (11, n ) 0, 3,
and 4).
1
5c by H NMR spectroscopy.4
When the reaction mixture was undercharged with TM-
SOTf, major product 4a was accompanied by a small amount
of R-silylated ester 6 (Scheme 2). This suggested that the
SKA 5 was competitively and, perhaps, reversibly silylated
at the C* vs O* atoms in 5. Indeed, resubjection of isolated
6 to TMSOTf/Et3N in CDCl3 resulted in its clean conversion
to lactam 4a. Simple esters are known to be silylated with
R3SiOTf in diethyl ether.5 As a model we examined the
silylation of methyl propionate (8), but under conditions more
similar to those used for the cyclization of 2 (Scheme 3).
Namely, in CDCl3 solution we could easily monitor (1H
NMR) the extent of conversion to a mixture of SKA 9 and
the R-silylated ester 10. Starting with 2 equiv each of
TMSOTf and TEA, the system equilibrated to a steady-state
mixture of 8:9:10 in a ratio of ∼5:trace:1.
We have also defined some of the scope of the imide
cyclization reaction, particularly with respect to ring size.
Substrate 13 (the one-carbon-longer homolog of 2) smoothly
cyclized to the 6/5-bicyclic lactams 14 and 15 with slightly
lower levels of dr (Scheme 5) compared with 2. The
glutarimide substrate 16 also cyclized in high yield, but the
stereochemical outcome was quite different. TMSOTf gave
rise, predominantly, to the exo adduct 18a, suggesting that
the saturated (and puckered) glutarimide ring deterred an anti-
addition analogous to that indicated in 7-endo (the assign-
ment of relative configuration for 17 vs 18 is based on
convincing differences in NOE effects and coupling constant
data). On the other hand, the bulkier TIPSOTf reagent gave
a nearly 1:1 mixture of adducts 17b and 18b, reflective of a
competitive interaction with the silyl moiety in a synclinal
mode of addition similar to that in 7-exo. The less hindered
(and more reactive) TMSOTf induced competitive elimina-
tion of TMS2O from 17a and/or 18a under the reacion
conditions to give the pyrroline byproduct, 19. Fortunately
(again), in the UCS1025A-relevant cyclization of 2 (and 20,
see below), this type of elimination event is thwarted by
virtue of destabilization of the carbocationic intermediate.
The behavior of chiral imide substrates was also of interest,
so we studied the cyclization of various ether-protected
This observation implies that the success of the cycliza-
tions to form 4 derives from the in situ O-silylative activation
(4) Z-O-Silylketene acetals are more stable than their E-isomers: Wilcox,
C. S.; Babston, R. E. J. Org. Chem. 1984, 49, 1451-1453.
(5) Emde, H.; Simchen, G. Liebigs Ann. Chem. 1983, 816-834.
Figure 1. ORTEP representation of 4a (panel a) and 24 (panel b).
5192
Org. Lett., Vol. 8, No. 23, 2006

Further, selective removal of the two TMS groups in 21b
gave the crystalline, mono-TBS ether 24, which was
subjected to single-crystal X-ray structure determination (cf.
Figure 1b).
Scheme 5
Another chiral substrate, hydroxysuccinimide 25, derived
from L-malic acid, was subjected to TMSOTf and Et3N
(Scheme 7). However, an entirely different reaction course
Scheme 7
tartarimides 20 (Scheme 6). When chiral imide 20a,2,3a
derived from l-tartaric acid, was subjected to TMSOTf/Et3N,
lactam 21a was obtained with excellent diastereoselectivity
(dr 20:1:0:0). In the endo transition states (23a-si vs 23a-
re), approach from the re face of either of the homotopic
imide carbonyl groups is disfavored by repulsion between
OTMS and the SKA R-H. Each of the products 21a-d was
obtained in good to excellent yield and diastereoselectivity.
The stereochemical assignment of the major product was
made on the basis of several lines of reasoning. The
conversion of 21b into the natural enantiomer of 1 proved
both the relative and absolute configurations of 21b.3a In our
studies, the 1H NMR characteristics of 21a and 22a indicated
that each of these diastereomers was an endo adduct (cf. 4a).
ensued. Following an initial, essentially instantaneous O-
silylation step, a rapid cyclization reaction occurred to give
a labile product, to which we have assigned the structure of
â-ketolactam 28. The skeleton of this indolizidinedione is
different from that of all of the previous products. We
surmise that rapid silylketeneaminal formation now diverts
the imide functionality to an electron-rich pyrrole; silylative
activation of the ester would then produce 26. This species
is apparently sufficiently reactive to cyclize to form the new
six-membered ring present, initially, as 27. GC-MS analysis
of an aliquot taken directly from the reaction mixture gave
evidence for an intermediate containing three trimethylsilyl
groups (M+ ) 431). Workup following addition of aqueous
sodium bicarbonate solution gave 28. This alternative reac-
tion course arises from a reversal of the roles of the imide
and ester carbonyl carbons (here the nucleophile and elec-
trophile, respectively) and represents a silylative umpolung.
In retrospect, the tartarimides (Scheme 6) are apparently
resistant to this alternative mode of cyclization because of a
steric barrier toward silylative enolization of the imide. More
specifically, loss of an R-proton, which is required for initial
silylketenaminal formation, is disfavored. Finally and in
contrast, the unsubstituted, parent succinimide substrate 29
quickly formed the 2,5-bis(trimethylsilyloxy)pyrrole 30 (in
Scheme 6
a Each of the minor isomers was identified by GC-MS; 22a
was isolated and fully characterized; 22b and 22d were observed
by 1H NMR analysis of the crude product mixture and had
characteristics similar to those of 22a. b Hu¨nig’s base (DIPEA) was
used instead of TEA, because use of the latter resulted in
competitive elimination of the â-OBn group.
1
situ H NMR analysis),6 which was apparently resistant to
(6) Barrett, A. G. M.; Broughton, H. B.; Attwood, S. V.; Gunatilaka, A.
A. L. J. Org. Chem. 1986, 51, 495-503.
Org. Lett., Vol. 8, No. 23, 2006
5193

Figure 2. 1H NMR (500 MHz, CDCl3) snapshot (time ) 35 min) of the reaction of starting imide 20a to give product lactam 21a.
Resonances assigned to arise from intermediates 32 and 33 are labeled with lower case letters.
cyclization. In the presence of excess TMSOTf and over
time, 30 gave rise instead to a mixture of mono- and bis-
C-silylated succinimides 31.
temperature and at relatively high concentrations (0.3-0.5
M in substrate) and are easy to work up. Potential applica-
tions to other carbonyl-containing compounds as well as
bimolecular versions are currently being investigated.
Finally, most of the reactions described above were
monitored directly by proton NMR analysis (in CDCl3) at
one time or another. As an example of the insights gained
from such studies, consider the spectrum shown in Figure
2. It is of the reaction mixture for the tartarimide 20a, shown
in Scheme 6. Starting material (20a), intermediate ketene
acetal (32), isomeric C-silylated intermediates (33), and (a
trace of) product 21a are all identifiable in this snapshot taken
after 35 min at room temperature. These assignments are
based upon and consistent with observations of the 1H NMR
profile over the entire course of the reaction.
Acknowledgment. This project was supported by Grant
GM-65597 from the National Institute of General Medical
Sciences, National Institutes of Health. We thank William
W. Brennessel, Victor G. Young, Jr., and the X-ray Crystal-
lographic Laboratory, Department of Chemistry, University
of Minnesota.
Supporting Information Available: Spectroscopic char-
acterization data and procedures for preparation of all new
compounds (4, 6, 12-15, 17-21, 24, 25, 28, and 31) and
crystallographic data in CIF format. This material is available
free of charge via the Internet at http://pubs.acs.org.
In conclusion, we have discovered a highly efficient and
operationally simple method for silylative Dieckmann-like
cylizations that is applicable to ester/imide- and diester-
containing substrates. The reactions can be run at room
OL061988Q
5194
Org. Lett., Vol. 8, No. 23, 2006