Solvent-dependent stereoselectivity in a Still-Wittig rearrangement: An experimental and ab initio study

Article abstract of DOI:10.1021/ol015764d

(formula presented) The Still-Wittig rearrangement gave opposite selectivities for (Z:E)-alkenes in THF (3:1) vs toluene (1:3) in the synthesis of serine-proline dipeptide amide isosteres. Four transition states leading to (Z)-and (E)-alkenes with THF and

Full text of DOI:10.1021/ol015764d

ORGANIC
LETTERS
2001
Vol. 3, No. 12
1789-1791
Solvent-Dependent Stereoselectivity in a
Still−Wittig Rearrangement: An
Experimental and ab Initio Study
,‡
Scott A. Hart,† Carl O. Trindle,† and Felicia A. Etzkorn*
Department of Chemistry, UniVersity of Virginia, CharlottesVille, Virginia 22904, and
Chemistry Department (0212), Virginia Tech, Blacksburg, Virginia 24061-0212
fetzkorn@Vt.edu
Received February 26, 2001
ABSTRACT
The Still−Wittig rearrangement gave opposite selectivities for (Z:E)-alkenes in THF (3:1) vs toluene (1:3) in the synthesis of serine−proline
dipeptide amide isosteres. Four transition states leading to (Z)-and (E)-alkenes with THF and without (representing toluene) were identified by
ab initio calculations at the 3-21G* level. The calculated (Z:E)-ratios with THF (4.7:1) and without THF (1:3.2) suggested that the transition state
geometries and energies were well-represented by the calculations.
In the synthesis of cis-proline dipeptide mimics,1 we have
used the Still-Wittig [2,3]-sigmatropic rearrangement2 to
form trisubstituted (Z)-alkenes selectively. Others have
similarly synthesized disubstituted (E)-alkene dipeptide mim-
ics.3 The general [2,3]-sigmatropic rearrangement has been
studied extensively,2,4 providing detailed accounts of how
substitution of the allylic ether precursor determines the
stereochemical outcome of the rearrangement in a predictable
fashion; the tabular survey compiled by Nakai gives an
excellent overview of these effects.4b While (E)- or (Z)-alkene
selectivity of [2,3]-sigmatropic rearrangements can be pre-
dicted by conformational analysis of the expected transition
state, little has been demonstrated regarding solvent effects
on selectivity, since these rearrangements are usually per-
formed in THF or mixtures containing THF.4b We report
here an example of solvent-dependent stereoselectivity in this
rearrangement and provide an ab initio theoretical analysis
of the possible transition states to account for the stereo-
selectivity.
† University of Virginia.
‡ Virginia Tech.
(1) (a) Hart, S. A.; Sabat, M.; Etzkorn, F. A. J. Org. Chem. 1998, 63,
7580-7581. (b) Hart, S. A.; Etzkorn, F. A. J. Org. Chem. 1999, 64, 2298-
2299. (c) Hart, S. A.; Etzkorn, F. A. Peptides for the New Millenium: Proc.
16th Am. Peptide Symp. 2000, 478-480.
(2) Still, W. C.; Mitra, A. J. Am. Chem. Soc. 1978, 100, 1927-1928.
(3) (a) Bol, K. M.; Liskamp, R. M. J. Tetrahedron Lett. 1991, 32, 5401-
5404. (b) Bol, K. M.; Liskamp, R. M. J. Tetrahedron 1992, 48, 6425-
6438. (c) Yong, Y. F.; Lipton, M. A. Bioorg. Med. Chem. Lett. 1993, 3,
2879-2882. (d) Bohnstedt, A. C.; Prasad, J. V. N. V.; Rich, D. H.
Tetrahedron Lett. 1993, 34, 5217-5220.
(4) (a) Hoffmann, R. W. Angew. Chem., Int. Ed. Engl. 1979, 18, 563-
572. (b) Nakai, T.; Mikami, K. Chem. ReV. 1986, 86, 885-902. (c) Midland,
M. M.; Kwon, Y. C. Tetrahedron Lett. 1985, 26, 5013-5016. (d) Marshall,
J. A. In ComprehensiVe Organic Synthesis; Pattenden, G., Ed.; Pergamon
Press: Oxford, 1991; Vol. 3, pp 975-1014. (e) Nakai, T.; Mikami, K. Org.
React. 1994, 46, 105-209.
Solvent variations in the Still-Wittig rearrangement
include examples where the use of THF resulted in either
reduced yields3c or reduced yields coupled with significant
[1,2]-rearrangement.3a,b In these cases, hexane was found to
be a superior solvent, leading to [2,3]-rearrangement in higher
yields. However, these examples were in systems where (E)-
alkene formation was significantly favored by sterics and
thus provided little information concerning the role of solvent
in (Z)- or (E)-selectivity. In solvents other than THF, the
alkene geometry was not variable as a result of formation
of a cyclic product.4e
10.1021/ol015764d CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/11/2001

Most conformational analyses of the [2,3]-sigmatropic
rearrangement have excluded the role of the counterion for
simplicity, with some exceptions. One specific proposal was
that the presence of a methoxymethyl ether protecting group
allowed the substrate to act as a tridentate ligand, chelating
the Li+ counterion and forcing a locked conformation in the
transition state. While this proposal accounts for the observed
results, theoretical analysis was not performed.5 The effect
of counterion chelation in the transition state on the outcome
of [1,2]-sigmatropic rearrangements was analyzed recently,
though solvent was not at issue.6 Computational studies on
the minimal, nonsubstituted [2,3]-rearrangement substrate
found that a five-membered ring saddle point was located
without the cation present, but the Li+ counterion had to be
included in energy calculations in order to locate a true
transition structure.7 In these cases solvent effects were not
examined. A review by Collum that discusses the chelation
of Li+ by TMEDA vs THF is particularly relevant for our
results.8
3:1 (E:Z)-2 (Scheme 1). Rearrangement at +50 °C yielded
a 1:1 mixture of (Z)- and (E)-alkenes in each solvent.9
This selectivity difference led us to examine possible
transition states in an effort to explain these results. The
transition state proposed by Still2 leading to the (Z)-
trisubstituted homoallylic alcohol (Scheme 1) was expected
to be favored over the analogous transition state leading to
the (E)-trisubstituted alkene as a result of unfavorable steric
interactions in the equatorial plane. This simple analysis is
consistent with our results in THF but not in toluene. Because
of these differences, we used ab initio calculations to clarify
the geometries and energies of the transition states leading
to the (Z)- and (E)-alkenes in the presence and absence of
THF.
Calculations. The tributylstannyl group was not included
in the calculations after initial results indicated that this group
quickly distanced itself from the reacting carbanion once the
rearrangement began. The tributyltin may influence the
conformation of the ground state, but it was not a factor in
comparison of the transition states. The starting structures
for calculations were truncated versions of 1b; methyl groups
were used in place of the benzyl protecting groups. This
truncation greatly reduced the calculation time while retaining
the most proximal steric interactions involved at the benzylic
positions. Restricted Hartree-Fock (RHF) calculations at the
3-21G* level located transition state structures on the paths
leading to major and minor products in the presence and
absence of one coordinating THF molecule. The size of the
systems in question limited the sophistication of the calcula-
tions we could complete. RHF calculations in the small
3-21G* basis produced reasonable bond distances and angles
for the reacting fragments and reproduced to a good
approximation the results of Wu, Houk, and Marshall on the
simplified system they studied.7 Although differences in
transition state energies were very small, the transition states
being compared were stereoisomers and we expected major
basis set errors to cancel out. The energy ranking of the four
transition states was consistent from MNDO through RHF/
3-21G* levels.
Still-Wittig Rearrangement. Beginning with tributyl-
stannane 1a, our system favored the (Z)-alkene upon treat-
ment with n-BuLi in THF,1 as expected when a pseudoaxial
transition state is preferred (Scheme 1).2 However, in 1b,
Scheme 1. Solvent-Dependent Stereoselectivity.
The effect of counterion chelation in the Wittig rearrange-
ment was found to be crucial in our system. Because THF
is capable of coordinating Li+, the resulting transition state
structure that included a THF molecule was quite different
than the transition state structure without THF. The transition
state structure was governed by the manner in which Li+
coordinates to the substrate and solvent. Figure 1 shows the
resulting structures and relative energies found in each case.
For the system with THF present, the solvent molecule
chelated the Li+ (Figure 1A). The cation also interacted with
the amine and the ether oxygen adjacent to the reacting
carbanion, thus forming a five-membered chelated ring that
locked the steric interactions of the transition state in a
defined orientation. The transition structure leading to the
the case where R ) OBn, (Z)-selectivity was only 3:1 in
THF at -78 °C (Scheme 1).1c The rearrangement was thus
performed in toluene to examine the effect of solvent on
(E)- or (Z)-alkene selectivity. A reversal of selectivity was
observed when toluene was the solvent at -78 °C, giving
(5) Wittman, M. D.; Kallmerten, J. J. Org. Chem. 1988, 53, 4631-4633.
(6) Maleczka Jr., R. E.; Geng, F. J. Am. Chem. Soc. 1998, 120, 8551-
8552.
(7) Wu, Y.-D.; Houk, K. N.; Marshall, J. A. J. Org. Chem. 1990, 55,
1421-1423.
(8) Collum, D. Acc. Chem. Res. 1992, 25, 448-454.
(9) The product ratios for the reactions in toluene and the reaction at
+50 °C in THF were determined by integration of the appropriate alkene
signal in the 1H NMR of the reaction mixture (Supporting Information).
Chemical shifts of the alkene protons had been previously determined by
isolation and characterization of both alkenes from the reaction in THF at
-78 °C.
1790
Org. Lett., Vol. 3, No. 12, 2001

The only difference in these two sets of diastereomeric
transition states was the presence or absence of one THF
molecule, which implied that the steric demands of the THF
molecule on the system altered the stabilities of the resulting
transition states. The striking difference was that, with THF
present, the chelation was of the form O-Li(THF)-N, while
without THF the cation instead preferred to chelate the two
ether oxygens, O-Li-O. There is some evidence that Li+
prefers to chelate oxygen over nitrogen (THF over
TMEDA).8 Despite the presence of a tertiary amine in the
molecule, our calculations showed chelation to only the two
internal oxygen atoms when THF was absent. When THF
was present, the internal tertiary amine replaced the methoxy
as ligand. In this model, the THF molecule was interacting
in a steric manner to determine the conformation of the
transition state. The transition structures lacking THF adopted
distinct structures, binding only the two internal oxygen
atoms. These steric interactions were not obvious in qualita-
tive analysis of this system, but ab initio transition state
analysis enabled us to visualize such differences.
In summary, ab initio calculations were consistent with
the stereochemical outcome of these Still-Wittig rearrange-
ments. Transition state structures located by Hartree-Fock
methods in the presence or absence of a chelating THF
molecule demonstrated differences in the manner in which
the Li+ counterion was chelated by the electronegative atoms
of the substrate. Examination of the transition state structures
found at the 3-21G* level did not reveal any discrete
interactions that might account for the observed selectivity.
Beyond the obvious differences in the manner of Li+
chelation, the diastereomeric transition structure differences
were qualitatively subtle. Despite the approximate nature of
the computational level applied, the ab initio calculations
were a useful tool for predicting plausible transition states
because they yielded subtle structural details that were not
obvious by qualitative analysis but they were instrumental
in favoring one transition state over another. The solvent-
dependent differences in these transition states can be used
as a guide in future applications of the Still-Wittig rear-
rangement where specificity in alkene geometry is desired.
Figure 1. Transition states found leading to (Z)- and (E)-alkenes
2: (A) with one THF and (B) without solvent, to represent the
reaction in toluene. Li+ is shown in yellow. The new bond is formed
to carbon marked with pink dot. Only hydrogens at the reacting
centers are shown. Calculations used Gaussian 98. Graphics were
prepared with MacroModel v. 6.0.
(Z)-alkene was more stable by 0.6 kcal/mol than the structure
leading to the (E)-alkene. At -78 °C, this value would result
in a 4.7:1 (Z:E) ratio, consistent with the predominance of
(Z)-alkene experimentally (3:1 Z:E).
Results for the system in the absence of THF revealed a
crucial difference in the way the substrate chelated Li+.
Figure 1B shows the transition structures located in the
absence of THF (representing the reaction in toluene). At
the 3-21G* level, the transition structure leading to the (E)-
alkene was more stable by 0.45 kcal/mol than the structure
leading to the (Z)-alkene. In both (E)- and (Z)-transition states
without the THF present, the Li+ cation lay between the two
ether oxygens, constraining the system to a six-membered
ring that significantly altered the stereochemistry of the
reaction.10 In this case, the calculated energy difference of
the transition states would result in a 3.2:1 (E:Z) ratio,
consistent with the observed 3:1 (E:Z) ratio of products.
Acknowledgment. This work was supported by NIH
Grant R01 GM63271-01.
Supporting Information Available: Experimental pro-
cedures for the synthesis and analytical data of the (E)- and
(Z)-alkenes, calculations and NOE spectra confirming alkene
geometries. This material is available free of charge via the
Internet at http://pubs.acs.org.
(10) Calculations at the 3-21G* level demonstrated that structures
containing O-Li-O chelation instead of N-Li-O in the presence of THF
were higher in energy by 7.58 kcal/mol (frontside approach to give Z) and
6.01 kcal/mol (backside approach to give E). In the absence of THF,
structures containing O-Li-N chelation were higher in energy than those
containing O-Li-O chelation by 7.71 kcal/mol (frontside approach to give
Z) and 6.17 kcal/mol (backside approach to give E).
OL015764D
Org. Lett., Vol. 3, No. 12, 2001
1791