How the binding of substrates to a chiral polyborate counterion governs diastereoselection in an aziridination reaction: H-bonds in equipoise

Article abstract of DOI:10.1021/ja103863j

The stereochemistry-determining step of the self-assembled chiral Bronsted acid-catalyzed aziridination reactions of MEDAM imines and three representative diazo nucleophiles has been studied using ONIOM(B3LYP/6-31G:AM1) calculations. The origin of cis selectivity in the reactions of ethyldiazoacetate and trans selectivity in reactions of N-phenyldiazoacetamide can be understood on the basis of the difference in specific noncovalent interactions in the stereochemistry-determining transition state. A H-bonding interaction between the amidic hydrogen and an oxygen atom of the chiral counterion has been identified as the key interaction responsible for this reversal in diastereoselectivity. This hypothesis was validated when a 3° diazoamide lacking this interaction showed pronounced cis selectivity both experimentally and calculationally. Similar trends in diastereoselection were observed in analogous reactions catalyzed by triflic acid. The broad implications of these findings and their relevance to chiral Bronsted acid catalysis are discussed.

Full text of DOI:10.1021/ja103863j

Published on Web 08/31/2010
How the Binding of Substrates to a Chiral Polyborate Counterion Governs
Diastereoselection in an Aziridination Reaction: H-Bonds in Equipoise
Mathew J. Vetticatt,*,† Aman A. Desai,‡ and William D. Wulff*,‡
Department of Chemistry, Michigan State UniVersity, East Lansing, Michigan 48824, and Department of
Biochemistry, Albert Einstein College of Medicine, Bronx, New York 10461
Received May 13, 2010; E-mail: wulff@chemistry.msu.edu
Scheme 2. General Mechanism of Aziridination Reactions of Imines
Abstract: The stereochemistry-determining step of the self-
assembled chiral Brønsted acid-catalyzed aziridination reactions
of MEDAM imines and three representative diazo nucleophiles
has been studied using ONIOM(B3LYP/6-31G*:AM1) calcula-
tions. The origin of cis selectivity in the reactions of ethyldiaz-
oacetate and trans selectivity in reactions of N-phenyldiazoace-
tamide can be understood on the basis of the difference in specific
noncovalent interactions in the stereochemistry-determining tran-
sition state. A H-bonding interaction between the amidic hydrogen
and an oxygen atom of the chiral counterion has been identified
as the key interaction responsible for this reversal in diastereo-
selectivity. This hypothesis was validated when a 3° diazoamide
lacking this interaction showed pronounced cis selectivity both
experimentally and calculationally. Similar trends in diastereose-
lection were observed in analogous reactions catalyzed by triflic
acid. The broad implications of these findings and their relevance
to chiral Brønsted acid catalysis are discussed.
and Diazo Compounds Catalyzed by Brønsted Acids (the
Diastereomer of the Aziridine Formed from Each TS Is Indicated in
Blue)
In the preceding communication (DOI 10.1021/ja1038648), we
described the extension of our catalytic asymmetric cis-aziridination
methodology1 to the selective formation of trans-aziridines. This
reversal of diastereoselectivity, which is accomplished by a simple
change in one of the substrates used, was accompanied by the
observation of the opposite facial selectivity to the imine (Scheme 1).
Herein we present a combined experimental and computational study
that examines the mode of catalysis of this unique catalyst and the
origins of the enantio- and diastereoselection in this unprecedented
universal catalytic asymmetric aziridination methodology.
iminium carbon to form an R-diazonium-ꢀ-amino ester/amide
intermediate. There are four limiting orientations for this attack:
two cisoid approaches (quasi-[3 + 2]) and two antiperiplanar
transoid approaches, as shown in Scheme 2.3 This carbon-carbon
bond-forming step is the enantioselectivity- and diastereoselectivity-
determining step of the reaction.4 Diastereoselection can be achieved
if one of these approaches is preferred. This step is rendered
enantioselective if the nucleophile can effectively discriminate
between the Si and Re faces of the activated imine. After the
carbon-carbon bond has been formed, N2 is eliminated in an SN2-
like fashion from the diazonium intermediate (directly from the
anti intermediates and after bond rotation from the gauche
intermediates) to form the three-membered aziridine ring.
Scheme 1. Universal Catalytic Asymmetric Aziridination
The active catalyst in our universal aziridination protocol is a
self-assembled chiral polyborate Brønsted acid having a unique
structure. Analogous to earlier studies with VAPOL, we found by
11B and 1H NMR spectroscopy that VANOL induces self-assembly
of an identical VANOL-B3 catalyst-imine complex.5 The iminium
ion can potentially be bound to one of the four oxygen atoms O1,
O2, O3, or O4 of the chiral counterion (Figure 1). NBO analysis
(RHF/3-21G) performed on the polyborate anion showed that the
oxygen atom with the highest electron density is O2. Consequently,
the O2-bound iminium ion was found to be the preferred geometry
for the catalyst-iminium complex (Figure 1).6 In addition to being
bound to the most electron-rich oxygen, the O2-bound species also
benefits from stabilizing noncovalent interactions between the
MEDAM protecting group and the VANOL ligand. The biaryl
system appears to effectively shield the Re face of the iminium
ion, keeping the Si face accessible for nucleophilic attack. This is
consistent with the absolute configuration of the cis-aziridine 3a
The mechanism of Lewis/Brønsted acid-catalyzed aziridination
reaction of imines and diazo compounds has received little
experimental and no calculational scrutiny.2 The widely accepted
mechanism invokes initial attack of the diazo nucleophile at the
† Albert Einstein College of Medicine.
‡ Michigan State University.
9
13104 J. AM. CHEM. SOC. 2010, 132, 13104–13107
10.1021/ja103863j  2010 American Chemical Society

C O M M U N I C A T I O N S
geometries from the ONIOM calculations computed at the B3LYP/
6-31+G* level of theory.8 This approach has been reasonably
successful in qualitative predictions of stereoselectivity in similar
reactions.9
The lowest-energy transition structures TS1 and TS2 leading to
the observed enantiomers of the cis- and trans-aziridines, respec-
tively, in the reaction of 1 and 2a are shown in Figure 2.10 The
key observation in both of these structures (and in all of the
structures discussed below) is that unlike in the catalyst-imine
complex depicted in Figure 1, the iminium proton is no longer
closest to O2. As a general trend, protonated 1 is H-bonded to O1
in all of the cis transition structures and O3 in all of the trans
transition structures. There also exists a stabilizing non-covalent
interaction between the acidic R-CH of 2a and O2/O1 of the catalyst
core in TS1 and TS2.11 In order to accommodate this interaction,
attack of 2a occurs via a syn-clinical approach relative to the
iminium double bond in TS1 and via a trans-antiperiplanar approach
in TS2 (similar to the cisoid and transoid transition structures in
Scheme 2). TS1 is 3.1 kcal/mol lower in energy than TS2, and this
difference is qualitatively consistent with the experimentally
observed >50:1 cis/trans ratio for this reaction at room temperature.
The transition structure for the formation of the minor enantiomer
of the cis-aziridine was found to be 1.9 kcal/mol higher in energy
than TS1, in good agreement with the experimental 97% ee.12
Figure 1. Optimized geometry of the (S)-VANOL-B3 catalyst-imine complex.
Also shown is the division of layers for the ONIOM calculations.
in the reaction of 1 and 2a.1 However, on the basis of Figure 1,
one cannot rationalize the Re-facial attack that must occur in order
to form the observed enantiomer of the trans-aziridine 3b in the
reaction of 1 and 2b. There clearly must be some interaction
between 2b and the catalyst (which is absent in the reaction of 2a)
in the stereochemistry-determining transition state that reverses both
the diastereoselectivity and the facial selectivity to the imine in
this reaction.
Transition structures TS3 and TS4 for the reaction of 1 and 2b
(Figure 2) were then located. Both TS3 and TS4 are characterized
by non-covalent H-bonding interactions between (a) the iminium
proton and O1/O3, (b) the R-CH of 2b and O2, and (c) the amidic
hydrogen of 2b and O3/O1. While H-bonding interactions (a) and
(b) are also present in TS1 and TS2, (c) is exclusive to TS3 and
TS4. Therefore, the reversal of diastereoselectivity likely has its
origins in the relative strength of H-bonding interaction (c) in TS3
Transition structures for the key carbon-carbon bond-forming
step of the reactions of 1 with 2a, 2b, and 2c catalyzed by the
(S)-VANOL-B3 catalyst were located using ONIOM(B3LYP/6-
31G*:AM1) calculations.7 The color scheme in Figure 1 illustrates
the division of layers for the ONIOM calculations. In addition to
the portions shown in red, the diazo compound was also calculated
using the DFT method. All distances are reported in angstroms,
and all reported energies are single-point energies of fully optimized
Figure 2. Geometries, key interactions, and relative energies of the diastereoselectivity-determining transition states in the reactions of 1 and 2a-c.
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13105

C O M M U N I C A T I O N S
versus TS4. H-bond strengths are characterized both by the bond
distance and the donor-H-acceptor angle, with short, linear
H-bonds having the strongest interactions.13 The amidic hydrogen-
oxygen H-bond is shorter (1.94 vs 2.04 Å) and closer to linearity
(176 vs 159°) in TS4 than in TS3 (Figure 2). TS4 is 2.6 kcal/mol
lower in energy than TS3, and this is consistent with the
experimental trans selectivity.14 Finally, the TS for the Si-facial
attack of 2b to give the minor enantiomer of the trans-aziridine
was found to be 11.4 kcal/mol higher in energy than TS4.12
As a mechanistic probe of the importance of this third H-bonding
interaction in impacting the diastereoselectivity, we decided to
explore the aziridination reaction of 1 and N-methyl-N-benzyldia-
zoacetamide (2c). Being a 3° diazoamide, 2c lacks the key amidic
hydrogen, and we expected that the reaction of 1 and 2c would
revert to a cis-selective reaction analogous to the reaction of 1 and
2a. Not surprisingly, this was indeed the case: the reaction carried
out at room temperature gave almost exclusively the cis-aziridine
(Scheme 1). Transition structures TS5 and TS6 were then located
for the reaction of 1 and 2c (Figure 2). Comparison of the pairs of
transition structures TS3/TS5 and TS4/TS6 shows that all of the
key distances are almost identical, and clearly, the only difference
of note is the absence of the H-bond between the amidic hydrogen
and O3/O1 in TS5 and TS6. As a result, TS5 is favored over TS6
by 4.3 kcal/mol, once again in complete accord with the experi-
mental (>50:1) cis/trans ratio.
This is a unique template in asymmetric catalysis. We have
shown that the polyborate catalyst self-assembles only in the
presence of the imine substrate.5 During a catalytic cycle, the
boroxinate core executes key functions that are quintessential for
asymmetric catalysis. It activates the imine electrophile by proton-
ation and imparts enantioselection in nucleophilic additions to the
imine by serving as a chiral counterion. Diastereoselection is
achieved when the polyborate core directs the orientation and
approach of the diazo nucleophile to the “active site”. It also lowers
the energy of the transition state via multiple H-bonding interactions
with both substrates.
importance of considering multiple non-covalent interactions as a
control element in other Brønsted acid-catalyzed aziridinations.17
Figure 3. Transition structures TS7 and TS8 leading to the minor (cis) and
major (trans) products in the triflic acid-catalyzed reaction of 4 and 2b.
The mode of catalysis described in this work adds a new
dimension to counterion catalysis by integrating into it some of
the key features of H-bonding catalysis. We are currently investi-
gating the application of this highly tunable “template catalyst” to
a wide variety of nucleophilic additions to imines and carbonyl
compounds.
Acknowledgment. This work was supported by NSF Grant
CHE-0750319.
Supporting Information Available: Experimental details, coordi-
nates of all calculated structures, alternative transition structures,
mechanistic discussions, and pdb files. This material is available free
of charge via the Internet at http://pubs.acs.org.
References
(1) (a) Zhang, Y.; Lu, Z.; Wulff, W. D. Synlett 2009, 2715–2739. (b) Zhang,
Y.; Lu, Z.; Desai, A.; Wulff, W. D. Org. Lett. 2008, 10, 5429–5432.
(2) Key reports: (a) Casarrubios, L.; Pere´z, J. A.; Brookhart, M.; Templeton,
J. L. J. Org. Chem. 1996, 61, 8358–8359. (b) Williams, A. L.; Johnston,
J. N. J. Am. Chem. Soc. 2004, 126, 1612–1613. (c) Rasmussen, K. G.;
Jørgensen, K. A. J. Chem. Soc., Perkin Trans. 1 1997, 1287-1292. Review:
(d) Johnston, J. N.; Muchalski, H.; Troyer, T. L. Angew. Chem., Int. Ed.
2010, 49, 2290–2298, and references therein.
(3) Intermediate orientations are also possible on the basis of the steric bulk
of the substrates. If both faces of the imine are considered, there are eight
limiting orientations. Cisoid attack leads to gauche intermediates, whereas
transoid attack leads to anti intermediates. The terminology used here is
similar to that in the following reference: Aggarwal, V. K.; Harvey, J. N.;
Richardson, J. J. Am. Chem. Soc. 2002, 124, 5747–5756.
Scheme 3. Triflic Acid-Catalyzed Aziridination Reactions
(4) On the basis of the relative rates for p-substituted imines, it has been
proposed that the C-C bond-forming step might also be the rate-
determining step of a Lewis acid-catalyzed aziridination reaction (see ref
2c).
(5) See the Supporting Information for spectra showing the existence of the
VANOL-B3 catalyst. For VAPOL-B3 studies, see: Hu, G.; Huang, L.;
Huang, R. H.; Wulff, W. D. J. Am. Chem. Soc. 2009, 131, 15615–15617.
(6) See the Supporting Information for higher-energy optimized complexes.
(7) (a) Svensson, M.; Humbel, S.; Morokuma, K. J. Chem. Phys. 1996, 105,
3654–3661. (b) Vreven, T.; Morokuma, K. J. Comput. Chem. 2000, 21,
1419–1432. (c) Dapprich, S.; Koma’romi, I.; Byun, K. S.; Morokuma, K.;
Frisch, M. J. J. Mol. Struct. 1999, 461, 1–21.
(8) The reported energies do not include zero-point energy (zpe) or entropy
corrections, since frequency calculations at the B3LYP/6-31+G* level of
theory would have been prohibitive. The zpe- and entropy-corrected energies
from the ONIOM calculations, PCM single-point energy calculations using
other DFT methods, and relevant discussions are included in the Supporting
Information.
(9) (a) Dudding, T.; Houk, K. N. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,
5770–5775. (b) Simo´n, L.; Goodman, J. M. J. Am. Chem. Soc. 2009, 131,
4070–4077. (c) Anderson, C. D.; Dudding, T.; Gordillo, R.; Houk, K. N.
Org. Lett. 2008, 10, 2749–2752. (d) Li, N.; Chen, X.-H.; Song, J.; Luo,
S.-W.; Fan, W.; Gong, L.-Z. J. Am. Chem. Soc. 2009, 131, 15301–15310.
(10) See the preceding communication for a discussion of facial bias in the
formation of cis- and trans-aziridines in the reactions of 1 and 2a/2b.
(11) The acidity of the R-H of diazo compounds is well-established. Even a
simple base such as DBU has been shown to cause deprotonation. See: (a)
Jiang, N.; Wang, J. Tetrahedron Lett. 2002, 43, 1285–1287. (b) Xiao, F.;
Liu, Y.; Wang, J. Tetrahedron Lett. 2007, 48, 1147–1149. (c) Wang, W.;
Shen, K.; Hu, X.; Wang, J.; Lui, X.; Feng, X. Synlett 2009, 1655–1658.
For a review, see: (d) Maas, G. Angew. Chem., Int. Ed. 2009, 48, 8186–
8195. Numerous other transition structures similar to TS1 and TS2 that
did not have this non-covalent interaction were also located. While they
On the basis of our understanding of the non-covalent interactions
that stabilize the transition states in the aziridination reactions catalyzed
by the B3 catalyst and the idea that the triflate anion could function as
a H-bonding counterion,15 we decided to explore the aziridination
reactions of 4 with 2a,2b 2b, and 2c catalyzed by triflic acid. Our
hypothesis was that in the event that the three oxygen atoms of the
triflate anion could stabilize the aziridination transition state in a manner
similar to our B3 catalyst, we could have tunable diastereoselectivity
even in this simple reaction, depending on the diazo nucleophile used.
To our delight, identical to the trends observed in our system, we
observed trans-selective aziridination in the reaction of 4 and 2b and
cis-selective aziridination in the reactions of 4 and 2a/2c (Scheme 3).16
The trans-antiperiplanar orientation of the double bonds of 4 and 2b
in TS8 (Figure 3) sets up the three strong H-bonding interactions with
the triflate anion, virtually identical to the situation in TS4. While this
result reinforces the validity of our model, it also emphasizes the
9
13106 J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010

C O M M U N I C A T I O N S
reproduced the diastereoselectivity, they were on an average 2-4 kcal/
mol higher in energy than either TS1 or TS2 (see the Supporting
Information).
Zuend, S. J.; Jacobsen, E. N. J. Am. Chem. Soc. 2009, 131, 15358–15374.
(c) Klausen, R. S.; Jacobsen, E. N. Org. Lett. 2009, 11, 887–890. (d) Xu,
H.; Zuend, S. J.; Woll, M. G.; Tao, Y.; Jacobsen, E. N. Science 2010, 327,
986–990.
(12) See the Supporting Information for structures and detailed discussion.
(13) Taylor, M. S.; Jacobsen, E. N. Angew. Chem., Int. Ed. 2006, 45, 1520–
1543.
(14) Isomerization of 1 to the (Z)-imine and approach of 2b identical to TS3
would also result in the observed enantiomer of trans-3b. This possibility
was considered and found to be higher in energy (see the Supporting
Information for coordinates of this structure).
(15) Jacobsen and co-workers recently illustrated the concept of “cooperative
catalysis” in Brønsted acid-catalyzed reactions using chiral ureas/thioureas
in a series of elegant publications. In one example, hydrogen bonding to a
triflate anion is invoked (reference 15d). See: (a) Zuend, S. J.; Coughlin,
M. P.; Lalonde, M. P.; Jacobsen, E. N. Nature 2009, 461, 968–970. (b)
(16) These reactions are slow (imine consumption is not always complete in
24 h) and are typically accompanied by some decomposition. See the
Supporting Information for full experimental details.
(17) (a) Hashimoto, T.; Uchiyama, N.; Maruoka, K. J. Am. Chem. Soc. 2008,
130, 14380–14381. (b) Zeng, X.; Zeng, X.; Xu, Z.; Lu, M.; Zhong, G.
Org. Lett. 2009, 11, 3036–3039.
JA103863J
9
J. AM. CHEM. SOC. VOL. 132, NO. 38, 2010 13107