A highly enantioselective and diastereoselective synthesis of cyclobutanes via boronic esters

Article abstract of DOI:10.1021/ol990579+

(equation presented) Deprotonation of enantiopure (R,R)-1,2-dicyclohexyl-1,2-ethanediol 1-chloro-4-cyanobutylboronates 5 with LDA followed by treatment with anhydrous magnesium bromide yields (R)-(trans-2-cyanocyclobutyl)boronic esters 7 in high diastereomeric and enantiomeric purity. No cyclobutane formation has been observed in the absence of at least a catalytic amount of magnesium halide.

Full text of DOI:10.1021/ol990579+

ORGANIC
LETTERS
1999
Vol. 1, No. 3
379-381
A Highly Enantioselective and
Diastereoselective Synthesis of
Cyclobutanes via Boronic Esters
Hon-Wah Man, William C. Hiscox, and Donald S. Matteson*
Department of Chemistry, Box 644630, Washington State UniVersity,
Pullman, Washington 99164-4630
dmatteson@wsu.edu
Received April 11, 1999
ABSTRACT
Deprotonation of enantiopure (R,R)-1,2-dicyclohexyl-1,2-ethanediol 1-chloro-4-cyanobutylboronates 5 with LDA followed by treatment with
anhydrous magnesium bromide yields (R)-(trans-2-cyanocyclobutyl)boronic esters 7 in high diastereomeric and enantiomeric purity. No
cyclobutane formation has been observed in the absence of at least a catalytic amount of magnesium halide.
We have found a highly enantioselective and diastereo-
selective general synthesis of cyclobutanes having two or
three stereocenters in the ring, which utilizes asymmetric (4-
cyano-1-chlorobutyl)boronic esters as the key intermediates.
R-Chloro boronic esters have previously been shown (1) to
be capable of providing exceptionally high stereocontrol in
the construction of asymmetric centers1 and (2) to allow the
assembly of molecules containing several adjacent stereo-
centers bearing a variety of functional substituents.2 However,
this chemistry has not been used for the preparation of
carbocycles.
pyl)boronate (2). Treatment of 2 with sodium cyanide in
DMSO then yielded 3a (R ) H), which was transesterified
with (R,R)-1,2-dicyclohexyl-1,2-ethanediol4 to make boronic
ester 4a (Scheme 1).
Chain extension of 4a with (dichloromethyl)lithium ac-
cording to the established method1,2 yielded chloro boronic
ester 5a (R ) H). Cyclization of 5a was initially ac-
complished by treatment with commercial lithium diisopro-
pylamide (LDA).5 Presumably the cyclic borate 6a is formed
immediately, but ring contraction with carbon-carbon bond
formation is very slow. Yields of 7a up to ∼65% (containing
∼6% cis isomer) were obtained irreproducibly. Dark batches
of LDA yielded only ∼20%, and LDA freshly prepared from
butyllithium and diisopropylamine failed entirely. We then
noted that our purchased LDA solutions contained a small
amount of magnesium diisopropylamide as a stabilizer.
Addition of 0.1-1.0 equiv of anhydrous magnesium bromide
in THF (from 1,2-dibromoethane and magnesium) after the
The reaction of allyl bromide with boron trichloride and
triethylsilane was used to prepare (3-bromopropyl)dichlo-
roborane (1),3 which was converted to pinacol (3-bromopro-
(1) Tripathy, P. B.; Matteson, D. S. Synthesis 1990, 200-206.
(2) (a) Matteson, D. S.; Sadhu, K. M.; Peterson, M. L. J. Am. Chem.
Soc. 1986, 108, 812-819. (b) Matteson, D. S. Chem. ReV. 1989, 89, 1535-
1551. (c) Matteson, D. S. Stereodirected Synthesis with Organoboranes;
Springer-Verlag: Berlin, 1995; pp 162-220. (d) Ho, O. C.; Soundararajan,
R.; Lu, J.; Matteson, D. S.; Wang, Z.; Chen, X.; Wei, M.; Willett, R. D.
Organometallics 1995, 14, 2855-2860. (e) Matteson, D. S.; Man, H.-W.;
Ho, O. C. J. Am. Chem. Soc. 1996, 118, 4560-4566. (f) Matteson, D. S.;
Man, H.-W. J. Org. Chem. 1996, 61, 6047-6051. (g) Matteson, D. S.;
Yang, J.-J. Tetrahedron: Asymmetry 1997, 8, 3855-3861.
(4) (a) Hiscox, W. C.; Matteson, D. S. J. Org. Chem. 1996, 61, 8315-
8316. (b) Wang, Z.-M.; Sharpless, K. B. J. Org. Chem. 1994, 59, 8302-
8303. (c) Hoffmann, R. W.; Ditrich, K.; Ko¨ster, G.; Stu¨rmer, R. Chem.
Ber. 1989, 122, 1783-1789.
(3) Matteson, D. S.; Soundararajan, R. Organometallics 1995, 14, 4157-
4166.
(5) Aldrich Chemical Co., Milwaukee, WI.
10.1021/ol990579+ CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/15/1999

of 7c was replaced by an isopropenyl group by Zweifel’s
method8 to provide 8c, which has a monoterpenoid carbon
skeleton.
Scheme 1a
Function of Magnesium Ion. The effect of magnesium
ion was especially surprising after zinc chloride had failed
to catalyze ring closure in an earlier experiment. Perhaps
zinc chloride is a strong enough Lewis acid to form a
complex with the cyano group of 6 and break the carbon-
boron bond, resulting in reversion to 5 as its zinc keten-
iminate derivative.
Before the role of magnesium cation was recognized, it
was observed that a very strict experimental protocol was
required. It was essential that the LDA solution be added to
the boronic ester 5, not vice versa, and that the amount of
LDA added must not exceed 1 equiv. A reaction followed
1
by H NMR showed only ∼10% completion after 1 day,
but ∼80% after 4 days. These observations are consistent
with the hypothesis that the magnesium diisopropylamide
present in the fresh samples of LDA liberates magnesium
halide as the reaction mixture becomes less basic. As the
LDA ages, the magnesium cation is apparently sequestered
in the dark precipitate.
The catalytic role of the magnesium halide in the ring
contraction of 6 to 7 presumably involves complex formation
between the departing chloride ion and the magnesium cation.
The transition state may be analogous in principle, though
not in all details, to that proposed for the rearrangement of
(B-alkyl)(B-dichloromethyl)borates.9
a (a) Pinacol, ether, -78 °C; (b) NaCN in DMSO, 50 °C, 12 h;
(c) if H replaced by R (series b, c), LDA in THF, -78 °C; R-I;
(d) (R,R)-1,2-dicyclohexyl-1,2-ethanediol; (e) LiCHCl2 in THF,
-100 °C; ZnCl2; to 25 °C 16-24 h; (f) LDA, -78 °C; (g) MgBr2,
THF, 3 days; (h) isopropenyl-MgBr, -78 °C; I2, Mg(OMe)2.
1,2,3-Tri- and 1,1,2,3-Tetrasubstituted Cyclobutanes.
Our mode of construction of the cyclobutane ring allows
other substitution patterns with full stereocontrol. 1,2,3-
Trisubstituted cyclobutanes were prepared by starting from
(bromomethyl)boronic ester 9, which is easily prepared by
the published method.10 Treatment of 9 with lithioacetonitrile
yielded 10a (R1 ) H), or 9 with lithiopropanenitrile yielded
racemic 10b (R1 ) CH3). Transesterification of 10 with
(R,R)-1,2-dicyclohexyl-1,2-ethanediol to form 11 was fol-
lowed by reaction with (dichloromethyl)lithium to form
R-chloro boronic esters 12. Treatment of 12a with methyl-
magnesium chloride led to 13a. Reactions of 12b with
butyllithium or with lithium benzyl oxide led to 13b and
13c, respectively. (Dichloromethyl)lithium converted 13 to
R-chloro boronic esters 14. This work was done prior to our
recognition of the role of magnesium halide in the ring
closure, and treatment of 14 with LDA containing a small
amount of magnesium diisopropylamide resulted in erratic
yields of cyclobutanes 15 together with 5-10% of their
diastereomers 16 (Scheme 2). The yield of 15a was 62%,
that of 15b was 50%, and that of 15c was 15-20%. Aside
from possible future synthetic utility, the immediate signifi-
cance of this work lies in the 15/16 pairs suitable for NOE
studies.
LDA resulted in very high yields of 7a routinely, and the
cis isomer was not detected by NMR, implying that <2%
could be present.
To enter series b or c, cyanopropylboronic ester 3a (or
4a) was deprotonated with LDA and treated with the alkyl
iodide. Prior to the discovery of magnesium bromide
catalysis, cyclization of 5b yielded up to 60% 7b, trans/cis
ratio 8:1, and cyclization of 5c to 7c was <20%, trans/cis
∼3:1. With 1 equiv of magnesium bromide, yields of 7c were
80-90%, and some of the 13C NMR peaks attributed to the
diastereomer appeared to be absent in the crude sample,
though the possibility of a ∼10:1 isomer ratio is not
positively ruled out by our data. The strong diastereomeric
preference for B trans to CN matches the stereocontrol found
in acyclic reactions of R-bromo boronic esters with carbox-
ylic esters6 or oxazolidinones7 and is supported by the NOE
observations described below.
Cyclobutylboronic esters 7 are potentially useful for a wide
variety of synthetic applications.2 As a demonstration of the
synthetic potential of this chemistry, the boronic ester group
(8) (a) Zweifel, G.; Arzoumanian, H.; Whitney, C. C. J. Am. Chem. Soc.
1967, 89, 3652-3653. (b) Brown, H. C.; Basavaiah, D.; Kulkarni, S. U.;
Bhat, N. G.; Vara Prasad, J. V. N. J. Org. Chem. 1988, 53, 239-246.
(9) (a) Reference 2c, pp 187-189. (b) Midland, M. M. J. Org. Chem.
1998, 63, 914-915. (c) Corey, E. J.; Barnes-Seeman, D.; Lee, T. W.
Tetrahedron: Asymmetry 1997, 8, 3711-3713.
(6) Matteson, D. S.; Michnick, T. J. Organometallics 1990, 9, 3171-
3177.
(7) Matteson, D. S.; Man, H.-W. J. Org. Chem. 1994, 59, 5734-5741.
(10) Michnick, T. J.; Matteson, D. S. Synlett 1991, 631-632.
380
Org. Lett., Vol. 1, No. 3, 1999

spectrum and coupling constants calculated from an MMX
program (PCModel) and formed a self-consistent NOE set.
Scheme 2a
NOE data for 15c and 16c (R1 ) CH3, R3 ) OCH2C6H5)
likewise support the structure assignments. The distinctive
CHOR peak at δ 4.15 in 15c [H3R] showed a 1.9-3.1% NOE
with the methyl singlet at δ 1.47 [CH31R], and the corre-
sponding peaks of 16c [H3R and CH31â] showed no detectable
NOE. The other ring hydrogens formed a self-consistent set,
and the NOESY spectrum was consistent with the assign-
ments.
There is nothing inherent in the foregoing synthetic method
that would restrict the size to four-membered rings, which
were chosen as our initial goal because they presented an
interesting challenge for asymmetric synthesis.11,12
Acknowledgment. We thank the National Science Foun-
dation for support (CHE9303074 and CHE9613857) and NIH
(RR 0631401) and NSF (CHE-911528) grants to the WSU
NMR Center.
Supporting Information Available: Detailed preparative
information for all new compounds; NOE data for 15a, 16a,
15c, and 16c, NOE spectrum of 16a. This material is
available free of charge via the Internet at http://pubs.acs.org.
OL990579+
(11) Typical Ring Closure (7c). Chloro boronic ester 5c was prepared
according to the previously described method.1,2 Crude 5c (14.2 g, 0.0278
mol) in THF (200 mL) was stirred at -78 °C during the dropwise addition
of LDA and for an additional 30 min. Magnesium bromide in THF (140
mL, ∼0.2 M, from magnesium metal and 1,2-dibromoethane) was added
slowly to the cold solution. The mixture was stirred for 3 days at 20-25
°C (until monitoring by NMR indicated complete reaction). Aqueous workup
followed by chromatography on a short silica column with 10% ethyl acetate
in pentane yielded 7c, Rf ∼0.9, as an oil (12.36 g, ∼95% 7c with ∼5%
unchanged 5c by 1H NMR [δ 2.8, 3.95], 89% yield of 7c contained): 300
MHz 1H NMR (CDCl3) δ 0.03 (s, 6), 0.85 (s, 9), 0.85-1.74 (m, 23), 1.97
(t, 2), 2.28 (m, 2), 2.4 (m, 2), 3.73 (m, 2), 3.83 (m, 2); 75 MHz 13C NMR
(DEPT) (CDCl3) δ -5.5 (CH3), 18.2 (w, CSi?), 19.4 (CH2), 25.75 (CH3),
25.82 (CH2), 25.90 (CH2), 26.3 (CH2), 27.3 (CH2), 27.5 (CH2), 32.8 (CH2),
35.4 (w, CCN?), 38.4 (CH2), 42.9 (CH), 60.1 (CH2), 84.0 (CH), 125.0 (CN);
CB (br) not observed. HRMS calcd for C27H48BNO3SiN (MH+) 474.3574,
found 474.3565. The analytical sample was further purified by chroma-
tography. Anal. Calcd for C27H48BNO3SiN: C, 68.48; H, 10.22; B, 2.28;
N, 2.96; Si, 5.93, Found: C, 68.36; H, 10.17; B, 1.84; N, 2.82; Si, 5.84.11
(12) All new compounds were characterized by 300 MHz 1H and 75
MHz 13C NMR. Satisfactory elemental analyses for all elements present
except O were obtained for 2, 3a, 4a, 4b, 7a, 7b, 7c, 11a, 13a, 13b, 15a,
15b, and 15c. Satisfactory HRMS data were obtained for all of the foregoing,
as well as 3b, 3c, 8c, 10a, 12a, 13c, 14a, 14b, and 14c.
a (d) (R,R)-1,2-Dicyclohexyl-1,2-ethanediol; (e) LiCHCl2 in THF,
-100 °C; ZnCl2; to 25 °C 16-24 h; (f) LDA, -78 °C; 25 °C, 3
days; (i) RMgBr or RLi.
NOE Studies. The NOE spectra of major product 15a (R1
) H, R3 ) CH3) and its isomer 16a support the structure
assignments. Pure 15a was obtained by crystallization, and
a small amount of a 1:2 mixture of 15a and 16a was
separated by chromatography. The 3-methyl and 1-H protons,
assigned by chemical shifts and coupling constants, showed
no NOE interaction in 15a and a substantial NOE signal in
16a, easily seen in the spectrum of the 1:2 mixture (see
supplementary data) but not easily quantified. The other
proton assignments for 15a were consistent with a NOESY
Org. Lett., Vol. 1, No. 3, 1999
381