Enantioselective oxazaborolidinium-catalyzed Diels-Alder reactions without CH·...O hydrogen bonding

Article abstract of DOI:10.1002/anie.200802002

(Chemical Equation Presented) All cis-tems go! The first detailed computational investigations into the title reaction validate Corey's two pre-transition-state models and reveal a third Lewis acid coordination mode (see picture), which operates for esters having s-cis C=C-C=O groups. The new pre-transition-state model explains the unexpected enantioselectivity witnessed for several Diels-Alder reactions and does not involve a C-H...O hydrogen bond.

Full text of DOI:10.1002/anie.200802002

Angewandte
Chemie
DOI: 10.1002/anie.200802002
Cycloaddition
Enantioselective Oxazaborolidinium-Catalyzed Diels–Alder Reactions
without CH····O Hydrogen Bonding**
Michael N. Paddon-Row,* Laurence C. H. Kwan, Anthony C. Willis, and Michael S. Sherburn*
The oxazaborolidine Diels–Alder (DA) precatalysts first
introduced by Corey et al. in 2002[1] are remarkable for
several reasons. The derived oxazaborolidinium cations, for
example 1a·HNTf2 or 1a·MXn (formed by exposure to either
Brønsted[1–3] or Lewis[4] acids, Scheme 1), are active catalysts
for enantioselective cycloadditions involving a broad range of
dienes and dienophiles.[1–12] Perhaps the most significant
quality of this family of Lewis acid (LA) catalysts, however,
lies in their singular ability to promote cycloadditions of
dienophiles carrying simple, common functional groups.
Interestingly, aldehyde-activated dienophiles give the oppo-
site p-enantiofacial selectivity relative to that of the corre-
sponding esters, carboxylic acids, and ketones in reactions
promoted by 1a. This observation was explained by Corey
and co-workers[2] in terms of the pre-transition-state assem-
blies shown in Scheme 1. A pivotal feature of these pre-
transition-state models is the presence of an O···HC hydrogen
bond, which is believed to facilitate p-enantiofacial selectivity
by constraining the orientation of the dienophile–LA com-
plex. Herein we present computational and experimental
evidence that certain enantioselective DA reactions may take
place in the absence of O···HC hydrogen bonding and that
they might even prefer this pathway over the normal Corey
pathway.
Certain intermolecular, intramolecular, and transannular
DA reactions of ester-activated dienophiles catalyzed by
oxazaborolidinium species furnish products which do not
correlate with the Corey pre-transition-state model shown in
Scheme 1. Thus, under oxazaborolidinium catalysis,[13]
a-methylene-g-butyrolactone (2) undergoes a highly exo-
selective intermolecular DA reaction with cyclopentadiene to
give adduct 3 (Scheme 2a) as the major stereoisomer, and
triene 4 undergoes a highly cis-selective intramolecular DA
reaction to give 5 (Scheme 2b). In a series of elegant
Scheme 1. Corey’s oxazaborolidinium catalyst and pre-transition-state
models (i.e. diene absent). a) Proposed mode of complexation for
aldehydes. b) Proposed mode of complexation for ketones, esters, and
acids.
[*] Prof. M. N. Paddon-Row[$]
School of Chemistry
The University of New South Wales, Sydney, NSW 2052 (Australia)
E-mail: m.paddonrow@unsw.edu.au
Dr. L. C. H. Kwan, Dr. A. C. Willis,[+] Assoc.Prof. M. S. Sherburn[#]
Research School of Chemistry
Australian National University, Canberra, ACT 0200 (Australia)
Fax: (+61)2-6125-8114
E-mail: sherburn@rsc.anu.edu.au
Homepage: http://rsc.anu.edu.au/research/sherburn.php
[$] Computational studies
[+] Correspondence author for crystallographic data
(willis@rsc.anu.edu.au).
[#] Synthetic studies
[**] Funding from the Australian Research Council (ARC) is gratefully
acknowledged. M.N.P.-R. acknowledges computing time from the
Australian Partnership for Advanced Computing (APAC) awarded
under the Merit Allocation Scheme.
Scheme 2. DA reactions proceeding with enantioselectivities that vary
relative to those predicted by the Corey O···HC hydrogen-bonding
model. a) and b) are results from this work, and c) is a result from
the work of Balskus and Jacobsen.[14]
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.200802002.
Angew. Chem. Int. Ed. 2008, 47, 7013 –7017
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7013

Communications
Table 2: Relative enthalpies (Hrel°) and relative Gibbs free energies
(Grel°) between the two most stable 1b·H+-catalyzed DA TSs leading to
opposite enantiomeric cycloadducts and predicted ee values based on
experiments, Balskus and Jacobsen recently demonstrated[14]
that macrocyclic lactones such as E,E,E-triene 6 undergo
stereoselective transannular DA reactions to furnish tricycles
such as 7 (Scheme 2c). The reason for the unexpected
enantioselectivities obtained for these reactions has now
been identified by DFT calculations.
the DGrel values.[a]
°
Dichloromethane solvent[b]
ee [%][d]
Exp. ee [%][d]
°[c]
°[c]
Entry
Substrate
Hrel
Grel
The Corey catalyst used in the calculations is 1b·H+,
protonated 1b, which was used in an early experimental
study.[1] In general, we limit our investigation to enantiose-
lectivities in DA reactions leading to the observed major
diastereomer (i.e. cis- or trans-fused bicycle, exo- or endo-
intermolecular DA adduct). Comprehensive sets of reactant–
1b complexes and plausible transition states (TSs) leading to
the two enantiomeric cycloadducts were located for each
reaction. Geometry optimizations and thermal corrections
were calculated by using B3LYP/6-31G(d) theory[15] and more
reliable single-point energies were calculated at the MPW1
K/6-31 + G(d,p) level of theory.[16,17] The polarizable contin-
uum model (PCM)[18] approximation was used to model the
dichloromethane solvent (e = 8.93).[19]
1
2
3
4
5
6
7
8
2
4
8
9
10
11
12
13
5.3
7.5
6.6
5.0
2.8
6.2
7.1
6.3
7.1
10.6
6.9
1.9
6.9
85
82
97
94
36
87
96
77
70[e]
53[e]
–
90[f]
80[f]
93[f,g]
86[f]
88[h]
9.4
5.09
8.13
[a] MPW1K/6-31+G(d,p)//B3LYP/6-31G(d)
+ B3LYP/6-31G(d) ther-
mal corrections. The reported values are in kJmolÀ1, calculated at
298.15 K. [b] Nonspecific solvent effect using the polarizable continuum
model. [c] Hrel° =H°(minor enantiomer)ÀH°(major enantiomer);
Grel° =G°(minor enantiomer)ÀG°(major enantiomer); [d] In all cases,
computational and experimental results give the same major enantiomer
(see Figure 1 for depictions of TSs leading to the major enantiomer.)
[e] This work. [f] Reference [5]. [g] The -CO2CH2CF3 group was used
experimentally. [h] Reference [2].
The Gibbs free energies (DGcomp) for formation of the
most stable reactant–LA complex for several dienophiles are
given in Table 1 together with the free energies of activation
Table 1: Gibbs free energies of formation (DGcomp) of the most stable
1b·H+ complexes of reactants and DA Gibbs free energies of activation
(DGa°) relative to these complexes.[a]
Gas phase
DGcomp
Dichloromethane solvent[b]
DGcomp
Substrate
DGa°
DGa°
2
4
9
12
13
À34.7
À23.2
À41.7
À16.1
À38.9
90.0
91.2
66.8
85.3
99.4
À28.0
À13.7
À33.9
À4.4
93.1
86.3
69.7
85.2
101.4
Figure 1. Schematic representation of the favored TS for each 1b·H+-
catalyzed DA reaction studied. The depicted diastereoselectivities
(cis or trans for triene precursors and endo or exo for dienophiles) and
p-enantiofacial selectivities are those favored experimentally
(LA=1b·H+).
À30.2
[a] MPW1K/6-31+G(d,p)//B3LYP/6-31G(d)
+
B3LYP/6-31G(d) ther-
mal corrections. Values reported in kJmolÀ1, calculated at 298.15 K.
[b] Nonspecific solvent effect by using the polarizable continuum model.
See Figure 1 for schematic transition structures.
ment with those observed experimentally (Table 2, entries 1
and 2, and 4–8). With one exception (Table 2, entry 5), the
predicted ee values are in good agreement with the exper-
imental values.[21]
(DGa°), which are relative to the precursor complex (or
include the diene for 2 and 13), for the most favorable DA
transition state.
These data clearly show that, for each system, the
numerical value of the free energy of complexation is
considerably smaller than the corresponding free energy
activation for the DA reaction. Assuming that the complex
formation is essentially barrierless, we conclude that the
complex formation is rapidly reversible on the DA reaction
timescale.[20] Consequently, DA enantioselectivities in these
LA-catalyzed DA reactions is determined solely by the
relative free energies of competing TSs, upon which we now
focus attention.
The essential features of Coreyꢀs pre-transition-state
assembly models are clearly evident in the TS geometries
(Figure 2). Thus, the lowest energy TSs[22] for the a,b-enals 8,
9, and 10 have the s-trans C C C O conformation[23] and the
= À =
boron center complexed to the oxygen atom through the lone
=
pair anti to the C C bond. This coordination mode, typified
by 9-TS (Figure 2a), is consistent with in the presence of a
formyl CH···OB hydrogen bond having an H···O distance of
2.47 Š, which lies within the range from 2.41 to 2.59 Š as
determined from X-ray crystal structures of aldehyde–boron
complexes.[24] The 3.38 Š separation between the formyl
carbon atom and C1’ of the exo-phenyl group at C5 might
signal a stabilizing p–p interaction postulated in the Corey
model.
Table 2 presents the relative enthalpies (Hrel°) and free
energies (Grel°) between the most favorable TSs leading to
each enantiomer for each system, and the ee values as
calculated from the Grel° values. Figure 1 presents schematic
representations of the predicted favored TSs. The predicted
p-enantiofacial selectivities are in complete qualitative agree-
The final feature of the Corey model—having the exo-
phenyl group at C5 hinder approach of the diene from the C5
7014
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7013 –7017

Angewandte
Chemie
Figure 2. PCM B3LYP/6-31G(d) calculated transition-state structures for intramolecular Diels–Alder reactions catalyzed by oxazaborolidinium
1b·H+. a) and b) TSs for 9; c) and d) TSs for 11; e) TSs for 4; f)TSs for 12.
direction—is apparent in 9-TS; the diene component, there-
fore approaches from the more open side which has the
phenyl group attached to B. The lowest energy TS leading to
the minor enantiomer (ent-9-TS, Figure 2b) does indeed
involve diene approach from the hindered side (C5–Ph), but
the developing steric congestion is eased by a substantial 568
able steric interaction between the syn-coordinated LA and
[25]
=
the C C bond of the dienophile in s-trans-11-TS; thus the
=
B···O C angle widens from 1268 in the s-trans TS to 1338 in
the s-cis TS. The b-CH···O distance in s-cis-11-TS is 2.55 Š,
which is indicative of a b-CH···O hydrogen bond, but this
energetic advantage is overwhelmed by the aforementioned
steric interaction.
À
twisting of the dienophile about the formyl O B bond away
from the exo phenyl group at C5, resulting in a closest diene–
exo-phenyl group distance of 3.5 Š. However, this twisting
comes at the cost of destroying the formyl CH···O hydrogen
bond, making the H···O distance 3.46 Š.
In TSs with ester-linked tethers, for example, 4 and 12, the
= À
carbonyl group of the tether is forced to adopt the s-cis C C
=
À
C O conformation with the C O bond approximately anti to
= À À
the carbonyl group (the dihedral angle O C O C is about
The favored TS for 9E-ester 11, s-trans-11-TS (Figure 2c),
resembles the Corey model for a,b-unsaturated esters, that is,
1608). Coordination of the LA to the carbonyl group in 4-TS
(Figure 2e) takes place at the less hindered anti lone pair,
rather than at the more congested syn lone pair. Syn-
complexed TSs for 4 are predicted to be 15–22 kJmolÀ1 less
stable than the anti-complexed TSs, notwithstanding the
presence of an CO···OB distance of only 2.92 Š in the anti-
complexed TS. The recently reported examples of enantiose-
lective oxazaborolidinium-catalyzed transannular DA reac-
tions of macrocyclic unsaturated esters (Scheme 2c) may be
explained in terms of anti-complexed TSs akin to that for 4.[14]
= À =
the C C C O conformation is s-trans and the LA is
=
coordinated syn with respect to the C C; the diene
approaches from the more open side having the B–Ph
fragment. This TS displays an a-CH···O hydrogen bond of
2.62 Š.
= À =
The lowest free energy TS having an s-cis C C C O
conformation, s-cis-11-TS (Figure 2d), is 24 kJmolÀ1 less
stable than s-trans-11-TS, which reflects the highly unfavor-
Angew. Chem. Int. Ed. 2008, 47, 7013 –7017
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7015

Communications
The most favored TSs for the intermolecular DA reactions
combination of destabilizing electrostatic dipole–dipole and
overlap repulsion interactions with the in-plane lone pair of
the alkoxy oxygen center.[28] This destabilizing interaction is
essentially removed upon formation of the anti-complexed
species at the tether, but remains if cooordination occurs
syn to the -CO2Me group.
The first possibility is discounted because the most stable
TS for the minor trans-IMDA pathway, trans-12-TS-I
(Figure 3 ), has an LA coordinated to the carbonyl group of
the tether, even though this carbonyl group is exo to the diene.
The large energy differences between 12-TS-III and 12-TS-
II—each having the terminal -CO2Me group in an s-trans
of 2 and 13 are readily explained in terms of the above
discussion, namely anti-complexed 2 avoids steric interactions
with the exocyclic methylene group, and syn-complexed 13
produces a stabilizing a-CH···O hydrogen bond in the TS.
For fumarate 12, coordination to the LA in the TS may
occur either at the syn position of the terminal -CO2Me group
= À =
(s-trans C C C O conformation) or at the anti position of
the ester tether (12-TS, Scheme 2), with the Corey model
predicting the former, owing to the formation of an
a-CH···O hydrogen bond in this coordination mode. Because
both coordination modes give the same enantiomeric prod-
uct, product analysis cannot distinguish between these
mechanistic alternatives. Our calculations resolved this mech-
anistic ambiguity. Of the 24 DA TSs located for the
experimentally observed cis addition (i.e. the terminal
-CO2Me group exo to the diene), the most stable cis-12-TS-I
(Figures 2 f and 3) involves an anti-complexed species at the
= À =
C C C O conformation—for both the cis- and trans-IMDA
reactions are consistent with the second explanation.
In all mechanistic discussions on the oxazaborolidinium-
catalyzed DA reactions published to date, it has been
implicitly assumed that reactant complexes and the TSs
occurs at the less congested exo (convex) face of the catalyst,
and our calculations confirm this preferred coordination
mode in all reactant complexes and in the dominant TSs
giving rise to the major enantiomers for all systems studied
(e.g. Figure 2). However, this assumption fails when consid-
ering the formation of the minor enantiomer. Although the
dominant TSs leading to the minor enantiomers in the DA
reactions of 2, 8–10, and 13 involve exo (convex) face
complexation, those for 4, 11, and 12 involve complexation
to the endo (concave) face of the catalyst, as depicted for the
case of 4 in Figure 4.[29] It appears that the steric congestion
associated with endo (concave) face complexation in the TSs
Figure 3. Relative free energies (kJmolÀ1) for selected a) cis-TSs and
b) trans-TSs for fumarate 12 (LA=1b·H+).
carbonyl group of the tether and an uncomplexed -CO2Me
= À =
group which adopts the s-cis C C C O conformation (cis-12-
TS-II, Figure 3). The -CO2Me-complexed TS, cis-12-TS-III, is
33 kJmolÀ1 higher in free energy than cis-12-TS-I and there-
fore plays no role in product formation. Why does the TS
prefer the anti-complexation at the carbonyl group of the
tether, which lacks a CH···O interaction, rather than the
Corey-type complexation seen in cis-12-TS-III, having a
stabilizing a-CH···O hydrogen bond?
There are two plausible explanations: 1) The uncatalyzed
intramolecular Diels–Alder (IMDA) reaction of 12 is strongly
cis-selective[26] and LA coordination at the endo carbonyl
group of the tether will additionally stabilize the cis TS
through strengthened secondary orbital interactions;[27] 2) the
anti lone pair of the carbonyl group of the tether is more basic
than the syn lone pair of the terminal -CO2Me group (s-trans
Figure 4. The lowest free energy TS for the DA reaction of 4 leading to
formation of the minor enantiomer.
for these systems is more than compensated by the diene
being able to approach from the more open side having the
B–Ph fragment.[30] This finding suggests that increasing steric
congestion on the endo (concave) face, by judicial placement
of ring substituents, might improve enantioselectivity.
In summary, we report herein the first computational
investigations into oxazaborolidinium-catalyzed Diels–Alder
reactions, which validate Coreyꢀs pre-transition-state models
with aldehyde- and ester-activated dienophiles and, further-
more, indicate an alternative coordination mode for esters
= À =
C C C O conformation), because the former suffers from a
7016
www.angewandte.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2008, 47, 7013 –7017

Angewandte
Chemie
[11]S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006, 128, 1346 – 1352.
[12]Y.-Y. Yeung, S. Hong, E. J. Corey, J. Am. Chem. Soc. 2006, 128,
which does not involve a CH···O hydrogen bond. The new
pre-transition-state model (Figure 5) operates with substrates
6310 – 6311.
= À =
undergoing DA reactions through s-cis C C C O conforma-
[13]We find that SnCl
is an excellent activator for Coreyꢀs
4
oxazaborolidine 1a. Corey recently reported the use of AlBr3
as an activator for 1a.[4a] Control experiments demonstrate that
triflimide-activated catalyst 1a·HNTf2 gives the same major
enantiomer as does 1a·SnCl4 (see the Supporting Information
for details).
[14]E. P. Balskus, E. N. Jacobsen, Science 2007, 317, 1736 – 1740.
[15]a) C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785 – 789;
b) A. D. Becke, J. Chem. Phys. 1993, 98, 5648 – 5652.
[16]B. J. Lynch, P. L. Fast, M. Harris, D. G. Truhlar, J. Phys. Chem. A
2000, 104, 4811 – 4815.
[17]Justification for the choice of the level of theory is given in the
Supporting Information. MP2/6-31G(d) single-point energies
were also calculated but they gave inferior results compared to
the MPW1K results (see the Supporting Information for details).
[18]a) S. Miertus, E. Scrocco, J. Tomasi, Chem. Phys. 1981, 55, 117 –
129; b) S. Miertus, J. Tomasi, Chem. Phys. 1982, 65, 239 – 245.
[19]Unless stated otherwise, the discussion refers to the PCM results.
[20]Indeed, it has been shown that the association of an a,b-enal and
a cognate oxazaborolidine catalyst is rapidly reversible on the
NMR timescale: E. J. Corey, T.-P. Loh, T. D. Roper, M. D.
Azimioara, M. C. Noe, J. Am. Chem. Soc. 1992, 114, 8290 – 8292.
[21]The agreement might even improve if the experimentally used
precatalyst 1a were employed in the calculations because it gives
higher ee values than 1b.[1]
Figure 5. The new pre-transition-state model explaining the unexpected
= À =
enantioselectivity with s-cis C C C O, ester-activated dienophiles.
tions and involves LA coordination at the anti lone pair of the
ester carbonyl group. Our studies show that this new pathway
may actually be preferred over the Corey pathway when the
two are in competition. Importantly, this model explains the
unexpected enantioselectivities seen in certain intermolecu-
lar, intramolecular, and transannular DA reactions.
Received: April 29, 2008
Revised: June 25, 2008
Published online: August 1, 2008
[22]It turns out that for all systems investigated the most stable TS
has both the smallest enthalpy and Gibbs free energy.
[23]In contrast, for the corresponding uncatalyzed DA reactions, the
= À =
formyl group in the TS prefers the s-cis C C C O conformation
Keywords: asymmetriccatalysis·densityfunctionalcalculations·
Lewis acids · Lewis bases · transition states
= À =
over the s-trans C C C O conformation by approximately
.
2.5 kJmolÀ1
.
[24]E. J. Corey, T. W. Lee, Chem. Commun. 2001, 1321 – 1329.
[25]Although the boron center may coordinate to the carbonyl
=
oxygen atom lone pair, which is anti to the dienophile C C;
[1]E. J. Corey, T. Shibata, T. W. Lee, J. Am. Chem. Soc. 2002, 124,
3808 – 3809.
[2]D. H. Ryu, T. W. Lee, E. J. Corey, J. Am. Chem. Soc. 2002, 124,
9992 – 9993.
[3]D. H. Ryu, E. J. Corey, J. Am. Chem. Soc. 2003, 125, 6388 – 6390.
[4]a) D. Liu, E. Canales, E. J. Corey, J. Am. Chem. Soc. 2007, 129,
1498 – 1499; b) K. Futatsugi, H. Yamamoto, Angew. Chem. 2005,
117, 1508 – 1511; Angew. Chem. Int. Ed. 2005, 44, 1484 – 1487.
[5]G. Zhou, Q.-Y. Hu, E. J. Corey, Org. Lett. 2003, 5, 3979 – 3982.
[6]D. H. Ryu, G. Zhou, E. J. Corey, J. Am. Chem. Soc. 2004, 126,
4800 – 4802.
[7]Q.-Y. Hu, P. D. Rege, E. J. Corey, J. Am. Chem. Soc. 2004, 126,
5984 – 5986.
[8]Q.-Y. Hu, G. Zhou, E. J. Corey, J. Am. Chem. Soc. 2004, 126,
13708 – 13713.
[9]D. H. Ryu, G. Zhou, E. J. Corey, Org. Lett. 2005, 7, 1633 – 1636.
[10]S. A. Snyder, E. J. Corey, J. Am. Chem. Soc. 2006, 128, 740 – 742.
anti complexation is 5 kJmolÀ1 less stable than syn complexa-
tion.
[26]E. L. Pearson, L. C. H. Kwan, C. I. Turner, G. A. Jones, A. C.
Willis, M. N. Paddon-Row, M S. Sherburn, J. Org. Chem. 2006,
71, 6099 – 6109.
[27]a) J. I. Garcia, V. Martꢁnez-Merino, J. A. Mayoral, L. Salvatella,
J. Am. Chem. Soc. 1998, 120, 2415 – 2420; b) D. A. Singleton, J.
Am. Chem. Soc. 1992, 114, 6563 – 6564.
[28]a) K. B. Wiberg, K. E. Laidig, J. Am. Chem. Soc. 1988, 110,
1872 – 1874; b) X. Wang, K. N. Houk, J. Am. Chem. Soc. 1988,
110, 1870 – 1872.
[29]The free energy preferences for endo-face over exo-face com-
plexation are (kJmolÀ1): 4 (6.4); 11 (0.3); 12 (8.7).
[30]The corresponding endo-face-complexed TS for 12 still involves
preferential complexation at the carbonyl group of the tether
rather than at the CO2Me group.
Angew. Chem. Int. Ed. 2008, 47, 7013 –7017
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7017