Solution structure of a chiral dialkylboron enolate by NMR spectroscopy and simulated annealing: Unusual stabilization of this intermediate by a baron-nitrogen interaction

Full text of DOI:10.1021/jo0001885

J . Org. Chem. 2000, 65, 5031-5033
5031
During our investigations of the enantioselective syn-
thesis of propanolamine based compounds, we observed
that the aldol reaction of a chiral oxazolidinone acylated
with phenyl acetic acid yielded the desired aldol product
with high diastereo- and enantioselectivity. However, the
corresponding reaction using the 2-pyridyl acetic acid
derivative for acylation led exclusively to the dialkylboryl
enolate intermediate 1 (Scheme 1). This situation was
further complicated because chiral 3-pyridyl acetic acid
derivatives behaved as expected during an aldol reaction.
To understand this remarkable stability of 2-pyridyl
dialkylboron enolates, we studied the conformation of the
dibutylboryl enolate in CDCl3 solution using 2D-NMR
spectroscopy and restrained simulated annealing, as
suitable crystals for X-ray crystallography were not
obtained.
Solu tion Str u ctu r e of a Ch ir a l
Dia lk ylbor on En ola te by NMR
Sp ectr oscop y a n d Sim u la ted An n ea lin g:
Un u su a l Sta biliza tion of Th is In ter m ed ia te
by a Bor on -Nitr ogen In ter a ction
Karl-Heinz Baringhaus,* Hans Matter, and
Michael Kurz
Aventis Pharma Deutschland GmbH, DI&A Chemistry,
D-65926 Frankfurt/ Main, Germany
Karl-Heinz.Baringhaus@aventis.com
Received February 10, 2000
The understanding of stereo- and enantioselective
reactions is indispensable for todays challenges in me-
dicinal chemistry, as most enzymes and receptors strictly
discriminate between stereoisomers. Spatial complemen-
tarity between ligands and target enzymes can be
achieved by designing all relative and absolute stereo-
genic centers of new ligands. The combination of multi-
dimensional NMR spectroscopy with computational ap-
proaches such as simulated annealing and distance
geometry yields information about the conformation and
relative configuration of key intermediates in those
stereospecific syntheses of relevant compounds.
The aldol reaction is still one of the most popular
carbon-carbon bond forming transformation, while us-
ing preformed enol derivatives provides an effective way
to control stereoselectivity.1 In particular, the introduc-
tion of architecturally refined enolate metal centers
improves its stereochemical attributes remarkably. Es-
pecially boron enolates2 and tetrachlorotitanium enolates
leads to well-defined transition states and high stereo-
selective yields.3
All NMR spectra were recorded on a Bruker DRX 600
spectrometer at 300 K using 20 mg of the dibutylboryl
1
enolate 1 in 0.5 mL of CDCl3.5 The H and 13C assign-
ments are given in the Supporting Information (Table
S1), diastereotopic protons at C9 and C11 could be
assigned using NOE derived distances, 3J (C,H) and
3J (H,H) coupling constants. Furthermore, the known
absolute configuration at C10 and the constrained five
membered ring system permitted the unambiguous as-
signment of protons at C9. Here the experimental
coupling constant of 7.8 Hz between H10 and H9proR is
in agreement with a dihedral angle of 9.6° in the five
membered ring system after simulated annealing. This
translates to a theoretical coupling constant of 8.4 Hz.
On the other hand the experimental coupling constant
proS
J H10-H9
is 2.7 Hz, which corresponds to the calculated
dihedral angle of -117.4° (theoretical coupling constant
2.8 Hz). The population of three rotameric states around
the oxazolidine side chain torsion N7-C10-C11-C12
proS
was calculated based on coupling constants J H10-H11
and J H10-H11proR.6 This side chain torsion predominantly
adopts the 180° orientation (65%), while +60° and -60°
are less populated (5%, 30%), respectively, due to steric
hindrance. Only one distinct set of signals was found in
Chiral oxazolidinones have been utilized as chiral
auxiliaries in different types of reactions4 with high
enantioselectivity. The asymmetric induction of the boron
mediated aldol reaction is accomplished in a noninter-
active way; i.e., the reaction is simply blocked at one side
of the substrate. For dialkylboryl enolates, the stereo-
chemistry of the kinetic aldol product was shown to be
related to enolate geometry.
the H and 13C NMR spectra, no conformational change
1
slow on the NMR time scale could be determined by line-
broadening effects. All other criteria for conformational
homogeneity were fulfilled,7 all experimental data are
* To whom correspondence should be addressed.Phone: ++49-69-
305-84048. Fax: ++49-69-331399.
(1) (a) Evans, D. A.; Nelson, J . V.; Taber, T. Top. Stereochem. 1982,
13, 1-115. (b) Evans, D. A. Aldrichim. Acta 1982, 15, 23-32. (c)
Heathcock, C. H. In Asymmetric Synthesis; Morrison, J . D., Ed.;
Academic Press: New York, 1983; Vol. 3, pp 111-212.
(2) (a) Masamune, S.; Mori, S.; Van Horn, D.; Brooks, D. W.
Tetrahedron Lett. 1979, 1665-1668. (b) Hirama, M.; Garvey, D. S.;
Lu, L.; Masamune, S. Tetrahedron Lett. 1979, 3937-3940. (c) Evans,
D. A.; Nelson, J . V.; Vogel, E.; Taber, T. R. J . Am. Chem. Soc. 1981,
103, 3099-3111. (d) Evans, D. A.; Bartroli, J .; Shih, T. L. J . Am. Chem.
Soc. 1981, 103, 2127-2129. (e) Masamune, S.; Choy, W.; Kerdesky, F.
A. J .; Imperiali, B. J . Am. Chem. Soc. 1981, 103, 1566-1568. (f)
Gennari, C.; Cardani, S.; Colombo, L.; Scolastico, C. Tetrahedron Lett.
1984, 25, 2283-2286. (g) Prashad, M.; Mahavir, K.; Kim, H.-Y.; Lu,
Y.; Liu, Y.; Har, D.; Repic, O.; Blacklock, T. J .; Giannousis, P. J . Org.
Chem. 1999, 64, 1750-1753.
(5) Homonuclear 2D-NMR experiments (DQF-COSY (Derome, A.;
Williamson, M. J . Magn. Reson. 1990, 88, 177-185.), TOCSY (Bax A.;
Davis, D. G. J . Magn. Reson. 1985, 65, 355-360.), and NOESY (J eener,
J .; Meier, B. H.; Bachmann, P.; Ernst, R. R. J . Chem. Phys. 1979, 71,
4546-4553.)) were performed with a spectral width of 8 ppm. These
spectra were recorded with 1024 increments in t1 and 4096 complex
data points in t2. For the NOESY 16 transients were averaged for each
t1 value, for COSY and TOCSY 8 transients. Mixing times of 70 or
150 ms were used for TOCSY and ROESY spectra, respectively. Data
processing was done using the XWIN NMR software from Bruker. For
HMQC (Bax, A.; Subramanian, S. J . Magn. Reson. 1986, 67, 565-
569.) spectra 1024 increments (16 scans) with 4096 complex data points
in t2 were collected using a sweep width of 8 ppm in the proton and
165 ppm in the carbon dimension. The HMBC (Bax, A.; Summers, M.
F. J . Am. Chem. Soc. 1986, 108, 2093-2094.) spectra were acquired
with a sweep width of 8 ppm in the proton and 200 ppm in the carbon
dimension. A total of 48 transients were averaged for each of 1024
increments in t1, and 4096 complex points in t2 were recorded. A delay
of 70 ms was taken for the development of long-range correlations.
(6) (a) Pachler, K. G. R. Spectrochim. Acta 1963, 19, 2085-2092.
(b) Pachler, K. G. R. Spectrochim. Acta 1964, 20, 581-587.
(3) Evans, D. A.; Rieger, D. L.; Bilodeau, M. T.; Urpi, F. J . Am.
Chem. Soc. 1991, 113, 1047-1049.
(4) (a) Ager, D. J .; Prakash, I.; Schaad, D. R. Chem. Rev. 1996, 96,
835-875. (b) Ager, D. J .; Prakash, I.; Schaad, D. R. Aldrichim. Acta
1997, 30, 3-12.
(7) Kessler, H. Angew. Chem., Int. Ed. Engl. 1982, 21, 512-523.
10.1021/jo0001885 CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/19/2000

5032 J . Org. Chem., Vol. 65, No. 16, 2000
Notes
rmsd method,13 for each structure the pairwise rmsd to
all structures with lower target function values was
computed and the maximum rmsd as a function of the
constraint violation value was plotted. Structures corre-
sponding to the first plateau were selected, leading to
an ensemble of 15 acceptable conformers for the first run
with an average restraint violation of 0.041 Å and 19
conformers for the second run (average restraint violation
0.037 Å) with the B-N bond. Only the constraint H9proS
to H11proS is violated more than 0.1 Å, demonstrating a
good agreement between experiment and simulation,
when taking only one conformation into account. There
is no evidence like conflicting NOE data to account for a
conformational equilibrium.
Sch em e 1. Syn th esis a n d Atom Nu m ber in g of th e
Dibu tylbor yl En ola te 1
Both simulations result in a short distance between
N1 and B1 with a bonding interaction forming a six-
membered heterocyclus and a similar shape of all con-
formers. The short distance between protons H4 and H6
(exptl 2.5 Å, calcd 2.43 Å) is only possible when the
2-pyridyl side chain allows a close proximity between
boron and its aromatic nitrogen. Additional support of a
single conformer is given by remarkable chemical shift
differences for both diastereotopic boron-complexing butyl
groups, indicating a rigid, conformationally different
environment, which is confirmed by NOEs involving both
butyl groups to protons H1, H10, H11, and H13.
While the planarity of pyridine and phenyl rings is
without question, the six-membered heterocyclus B1-
O1-C7-C6-C5-N1 significantly deviates from planar-
ity. The characteristic dihedral angle B1-O1-C7-C6
adopts an average value of -16.2 ° (std: 10.6°) for 15
conformers from the first run and -13.1° (std: 10.2°) for
the 19 conformers from the second run. To check, whether
this is an artifact from force field optimizations, those
15 conformers were fully geometry optimized using
MOPAC 6.0 with the AM1 Hamiltonian.14 Twelve out of
15 conformers from simulated annealing led to a com-
pletely planar six-membered ring (dihedral angle B1-
O1-C7-C6: -2.2° SD: 5.8). Their average heat of
formation (∆Ηf) is -111.62 kcal/mol (SD 0.49 kcal/mol),
while for the alternative conformers with a six-membered
ring deviating from planarity, higher heats of formation
between -93.18 and -98.89 kcal/mol are computed
(dihedral angle B1-O1-C7-C6: between -72.6° and
-86.1°). The acceptable 12 superimposed conformers
after MOPAC geometry optimization are superimposed
in Figure 1.
consistent with the assumption of a predominant single
conformation, except for the butyl side chains.
Quantitative information on interproton distances for
the structure determination was obtained from analyzing
a 600 MHz 2D NOESY spectrum in CDCl3 with 500 ms
mixing time. Volume integrals of the individually as-
signed cross-peaks were converted into distance con-
straints using the isolated spin pair approximation. The
NOE restrained simulated annealing calculations8 were
performed utilizing 23 nontrivial interproton distances
(Table S2).9 Partial charges were calculated using MO-
PAC 6.010 with the AM1 Hamiltonian.11
As an interaction between boron and nitrogen will not
be modeled correctly using force field methods (e.g., the
used TRIPOS 6.0 force field,12 two different 600 ps
simulations (100 cycles at 6 ps) were carried out. The first
run did not utilize any additional B-N bond in the
molecular topology, while the postulated B - N interac-
tion was incorporated as additional bond for the starting
geometry during a second simulation. Acceptable struc-
tures were selected based on the maximum pairwise
(8) (a) Nilges, M.; Clore, G. M.; Gronenborn, A. M. FEBS Lett. 1988,
229, 317-324. (b) Bru¨nger, A. T.; Adams, P. A.; Rice, L. M. Structure
1997, 5, 325-336.
(9) All modelling work was performed using SYBYL (Versions 6.4
and 6.5, Tripos, St. Louis, MO, 1998.) on Silicon Graphics workstations.
A time step of 0.5 fs was used for the integration of Newton’s equation
of motion. Kinetic energy was included by coupling the system to a
thermal bath (Berendsen, H. J . C.; Postma, J . P. M.; van Gunsteren,
W. F.; Di Nola, A.; Haak, J . R. J . Chem. Phys. 1984, 81, 3684-3690).
The dielectricity function was set to 1. The NOE derived distance
constraints were applied as a biharmonic constraining function with
a force constant of 10 kcal/mol*Å2. Upper and lower distance limits
were set to plus and minus 10% of the experimental distances, for
nondiastereotopically assigned methylene and methyl protons, 0.9 and
1.0 Å were added to the upper bound. The atomic velocities were
applied following a Boltzmann distribution about the center of mass
to obtain a starting temperature of 1000 K. After simulating for 1.0
ps at this high temperature, the system temperature was stepwise
reduced using an exponential function over a 5.0 ps period to reach a
final temperature of 100 K. This results in 100 cycles of a 6.0 ps
simulated annealing protocol for each simulation.
To address the questions, whether a close boron -
nitrogen distance lead to a very stable structure, relaxed
conformational maps were computed for the stable dibu-
tylboryl enolate 1 with an ortho-substituted pyridine
fragment and the corresponding reactive dibutylboryl
enolate 2 containing a meta-substituted pyridine nucleus.
Those relaxed conformational maps were computed by
incrementing the dihedral angles C7-C6-C5-N1 in 1
or C7-C6-C5-C1 in 2 in steps of 10°. The resulting
conformers were minimized using the TRIPOS 6.0 force
field using a quasi Newton-Raphson procedure and then
subjected to a full geometry optimization using MOPAC
AM1. In Figure 2 the computed heats of formation (in
kcal/mol) are plotted on the y-axis versus the dihedral
(10) Stewart, J . J . P.; Seiler, F. J . MOPAC (Version 6.0), QCPE
program No. 455, Quantum Chemistry Program Exchange, University
of Indiana, Bloomington, IN, 1989.
(11) (a) Dewar, M. J . S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J . J .
P. J . Am. Chem. Soc. 1985, 107, 3902-3907. (b) Dewar, M. J . S.; J ie,
C.; Zoebisch, E. G. Organometallics 1988, 7, 513-521.
(12) Clark, M.; Cramer, R. D.; Van Opdenbosch, N. J . Comput.
Chem. 1989, 10, 982-1912.
(13) Widmer, H.; Widmer, A.; Braun, W. J . Biomol. NMR 1993, 3,
307-324.
(14) Full geometry optimization level using all bonds and angles,
the convergence option precise and the additional MOPAC keywords
NOMM, and XYZ were used.

Notes
J . Org. Chem., Vol. 65, No. 16, 2000 5033
F igu r e 1. Stereoview of the superimposed ensemble of 12 acceptable conformers for the dibutylboryl enolate 1 after restrained
simulated annealing and MOPAC AM1 full geometry optimization. Protons are omitted for clarity.
angle C7-C6-C5-N1 (deg) for the stable dibutylboryl
enolate 1 and the corresponding angle C7-C6-C5-C1
in the reactive dibutylboryl enolate 2. Only for 1 a
minimum value is observed, when this torsion angle
adopts values between -30 and +20°, corresponding to
a close B-N distance (1.58 Å). Such a close distance is
not possible in the meta-substituted pyridine derivative
2, there is no orientation leading to a stabilization of this
regioisomer. Due to internal structural constraints in 2,
formation of a seven-membered ring for stabilization by
adding another intramolecular coordination pattern to
the central boron is impossible. Although two X-ray
structures of boron enolates15 with boron-nitrogen bonds
(distance: 1.60, 1.64 Å) are present in the Cambridge
crystallographic database,16 those nitrogens have no
aromatic character, causing the formed six-membered
heterocycles to deviate significantly from planarity. In
contrast, the herein reported conformation suggests a
nearly planar, stable 6-membered cyclus.
F igu r e 2. Heats of formation (in kcal/mol) versus dihedral
angle C7-C6-C5-N1 (deg) for the stable dibutylboryl enolate
The remarkable stability of the dialkylboron enolate
1 from this study is caused by the coordination of the
2-pyridyl nitrogen with boron, as could be demonstrated
using the combined application of NMR spectroscopy and
quantum chemical calculations. Only 1 is internally
stabilized by formation of a six-membered heterocyclus
with a boron-nitrogen bond (1.58 Å), while this is not
possible for 2. As this coordination causes a reduced
Lewis acidity of the boron, an activation of the benzal-
dehyde carbonyl-oxygen atom is no longer possible.
Hence, the cyclic six-membered Zimmermann/Traxler
transition state could not be formed and no subsequent
1 and the corresponding angle C7-C6-C5-C1 in the reactive
dibutylboryl enolate 2 obtained using full MOPAC AM1
geometry optimization using grid searching.
reaction to the desired aldol product occurs. The present
work demonstrates that the combined application of
modern techniques for conformational analysis led to an
improved understanding of very different reactivities in
chemically slightly different derivatives for a well-
documented reaction in many syntheses of biologically
relevant molecules.
Su p p or t in g In for m a t ion Ava ila b le: Additional tables
with 1H, 13C chemical shifts, experimentally derived distance
constraints, computed averaged distances for the diarylboryl
enolate 1 and experimental procedures for the preparation of
1. This material is available free of charge via the Internet at
http://pubs.acs.org.
(15) (a) Ko¨ster, R.; Kucznierz, R.; Schu¨ssler, W.; Bla¨ser, D.; Bo¨se,
R. Liebigs Ann. Chem. 1993, 189-200. (b) Ansorge, A.; Brauer, D. J .;
Bu¨rger, H.; Do¨rrenbach, F.; Hagen; Pawelke, G.; Weuter, W. J .
Organomet. Chem. 1990, 396, 253-267. (c) CSD Refcode SUHNUE,
VIBZEL.
(16) Cambridge Structural Database (CSD), UNITY release 5.15,
Cambridge Crystallographic Data Center (CCDC), Cambridge, UK,
1998.
J O0001885