Catalyst-controlled torquoselectivity switch in the 4π ring-opening reaction of 2-amino-2-azetines giving β-substituted α,β- unsaturated amidines

Article abstract of DOI:10.1021/ja202576e

The torquoselectivity of the 4π electrocyclic ring-opening reaction of 2-azetines can be controlled by the Bronsted acidity of the catalyst and the polarity of the solvent. DFT calculations provided insight into the mechanism of this remarkable switch. Anti and syn stereoisomers of α,β-unsaturated amidines were selectively synthesized from ynamides and aldimines in the presence of Tf₂NH and CSA, respectively.

Full text of DOI:10.1021/ja202576e

COMMUNICATION
pubs.acs.org/JACS
Catalyst-Controlled Torquoselectivity Switch in the 4π Ring-Opening
Reaction of 2-Amino-2-azetines Giving β-Substituted r,β-Unsaturated
Amidines
Naoya Shindoh, Kazuo Kitaura, Yoshiji Takemoto,* and Kiyosei Takasu*
Graduate School of Pharmaceutical Sciences, Kyoto University, Yoshida, Sakyo-ku, Kyoto 606-8571, Japan
S Supporting Information
b
ABSTRACT: The torquoselectivity of the 4π electrocyclic
ring-opening reaction of 2-azetines can be controlled by the
Brønsted acidity of the catalyst and the polarity of the
solvent. DFT calculations provided insight into the mechan-
ism of this remarkable switch. Anti and syn stereoisomers of
R,β-unsaturated amidines were selectively synthesized from
ynamides and aldimines in the presence of Tf2NH and CSA,
respectively.
Figure 1. Modes of torquoselectivity in the 4π ring-opening reaction.
Substituents, D, A, S, and L, signify electronically donating, electronically
accepting, sterically small, and sterically large groups, respectively.
4π Electrocyclic ring-opening reactions of cyclobutene analo-
gues to give 1,3-dienyl compounds have attracted the attention of
many synthetic and theoretical chemists.1 The stereochemical
process by which geometrical isomers are formed in these
reactions is referred to as “torquoselectivity”.1a The thermal
conrotatory 4π ring-opening allows two senses of rotation of
the terminal substituents of the double bonds of the product
(inward versus outward rotation). Both steric and electronic
factors affect torquoselectivity (Figure 1). Using ab initio calcula-
tions, Houk and co-workers showed that electronic effects dom-
inantly control the torquoselectively and an electron-donating
substituent at the C-4 position tends to rotate outward, whereas
an electron-accepting substituent rotates inward (mode a).1a
This theoretical prediction successfully explains many experi-
mental results.2 Steric effects can influence on the torquoselec-
tivity of the 4π ring-opening when the electronic properties of
the substituents are not significantly different. When the C-3
substituent is sterically smaller than the forming C(=X)R moiety,
the bulky C-4 substituent tends to undergo an outward rotation
(mode b). When the C-3 substituent is sterically bulky, an inward
rotation of the larger C-4 substituent is preferred (mode c).3 We
have recently reported a selective synthesis of R,β-unsaturated
amidines bearing a TIPS group at the R-position from ynesul-
tams and aldimines by a domino [2 þ 2] cycloaddition—4π
ring-opening in which the torquoselectivity was determined by
the steric bulk of the C-3 substituent (mode c).3
Scheme 1. Catalytic Reaction of Ynecarbamate 1 with Aldi-
mine 2 Giving R,β-Unsaturated Amidine 3
During the course of our efforts to synthesize a variety of R,
β-unsaturated amidines, we observed reverse stereoselectivity in
the reaction of ynecarbamate 14 with aldimines 2 (Scheme 1).
When the reaction of 1a (R1 = Me, R2 = TMS) with 2a (R3 = R4 =
p-CF3C6H4) in toluene was carried out with triflic imide
(Tf2NH) as a catalyst, a mixture of two geometrical isomers,
anti- and syn-3aa, was obtained in 95% yield (anti:syn = 92:8)
(Table 1, entry 1). Using triflic acid (TfOH) and anhydrous
tetrafluoroboric acid as catalysts also gave 3aa in similar yields
and selectivities (entries 2 and 3). Surprisingly, the stereoselec-
tivity of 3aa was inverted (anti:syn = ∼15:∼85) when methane-
sulfonic acid (MsOH) and 10-camphorsulfonic acid (CSA) were
In most reported cases of 4π ring-opening, including our earlier
effort, the torquoselectivity of the electrocyclic reaction is critically
dependent on the properties of the substrates; a stereoselective
synthesis of both geometrical isomers of the 1,3-dienyl compound
from the same substrate is difficult.4 We now disclose a catalyst-
controlled synthesis of R,β-unsaturated amidines in which the
rotation modes of the torquoselective electrocyclic ring-opening
step can be switched by the properties of the acid catalysts.
Received: March 22, 2011
Published: May 10, 2011
r
2011 American Chemical Society
8470
dx.doi.org/10.1021/ja202576e J. Am. Chem. Soc. 2011, 133, 8470–8473
|

Journal of the American Chemical Society
COMMUNICATION
employed as catalysts (entries 5 and 6). Using p-toluenesulfonic
acid (TsOH) resulted in a low yield of 3aa with low selectivity
(entry 4). These results suggested that the acidity of the catalysts
affected selectivity; strong Brønsted acids (the pKa of TfOH in
CH3CN is 2.6)5 primarily produced the anti-product, whereas
weaker acids (the pKa of MsOH in CH3CN is 10)6 selectively
furnished the syn-adduct. We have confirmed that isomerization
between anti- and syn-3aa does not occur under the reaction
conditions.
It has been shown that the stereochemistry of 3 is kinetically
determined during the conrotatory ring-opening of 2-azetine
4.7,8 The possibility of acid-catalyzed anti/syn isomerization of 3
was ruled out; no isomerization of anti-3aa and syn-3aa was
observed after 24 h in the presence of Tf2NH and CSA. As
previously reported by Houk et al., 4π electrocyclic ring-opening
of 2-azetines may proceed exothermically without the assistance
of catalysts.7a However, our data clearly demonstrate that the
torquoselectivity of 4 can be switched by the properties of the
acid catalysts. To the best of our knowledge, these are the first
observations of a significant influence on the torquoselectivity of
the 4π ring-opening reaction by a catalyst.9
salts cis- and trans-[4ab H]þ is small, and the diastereomers
3
rapidly equilibrate via deprotonation, inversion of a nitrogen
atom, and reprotonation under the reaction conditions. Follow-
ing the CurtinꢀHammett principle,11 we considered four pos-
sible transition states: inward (TSs A and B) and outward (TSs C
and D) rotations of the R3 substituent and cis (TSs A and C) and
trans (TSs B and D) relationships between the R4 and R3
substituents (Figure 2). The E/Z stereochemistry of the CdN
moiety of 3 (oxazolidinone vs R4) readily epimerizes,12 and the
E geometry is much more stable than the Z geometry. Therefore,
only the geometry of the CdC bond reports the torquoselec-
tivity of the ring-opening reaction.
Table 2 summarizes the relative Gibbs free energy (ΔΔGq of
the TSs AꢀD under three sets of conditions: (1) in the presence
of proton (a naked ammonium cation), (2) in the presence of
TfOH (an intimate ion pair), and (3) in the presence of MsOH
(an intimate ion pair). Under each set of conditions for the
reaction of 2-azetinium salt [4 H]þ, TSs A and D were more
3
stable than TSs B and C and were the structures that produced
syn- and anti-3, respectively. These results indicated that the
To clarify the origins of catalyst-induced reverse torquoselec-
tivity, the transition states for the 4π electrocyclic ring-opening
of the simplified model compound 4ab (R1 = Me, R2 = TMS,
R3 = Ph, R4 = p-CF3C6H4) were calculated using density
functional theory (DFT) at the B3LYP/6-31(d) level.10 The
energy difference (ΔGo) between the diastereomeric ammonium
Table 1. Effect of a Catalyst on anti/syn Selectivity of 3aaa
entry
catalyst
% yield of 3aa
anti/sync
1b
2
3b
Tf2NH
95
93
97
5
92:8
TfOH
95:5
HBF4 OMe2
94:6
3
4
TsOH H2O
55:45
17:83
15:85
3
5
MsOH
40
79
6
CSA
a Conditions: 1a (R1 = Me, R2 = TMS), 2a (R3 = R4 = pCF3C6H4) (1.2
equiv), catalyst (20mol%), solvent(0.3M), rt, 24 h. b Catalyst(10mol%),
30 min. c Ratios were determined by 1H NMR.
Figure 2. Four possible transition states TSs AꢀD for conrotatory ring-
opening of 4ab. oxz = dimethyloxazolydinoyl, Ar = p-CF3C6H4.
Table 2. Calculated Ring-Opening Free Energy for Protonated 4ab with/without a Counteranion
transition
entry
azetinea
structure
ΔGq (kcal/mol)b
ΔΔGq (kcal/mol)
final product
syn-3ab
1.1
1.2
1.3
1.4
2.1
2.2
2.3
2.4
3.1
3.2
3.3
3.4
cis-[4ab H]þ
TS A1
TS B1
TS C1
TS D1
TS A2
TS B2
TS C2
TS D2
TS A3
TS B3
TS C3
TS D3
6.5
9.0
þ1.4
þ3.9
þ3.5
0
3
trans-[4ab H]þ
3
cis-[4ab H]þ
8.6
anti-3ab
syn-3ab
anti-3ab
syn-3ab
anti-3ab
3
trans-[4ab H]þ
5.1
3
cis-4ab TfOH
8.8
0
3
trans-4ab TfOH
22.0
15.4
9.9
þ13.2
þ6.6
þ1.1
0
3
cis-4ab TfOH
3
trans-4ab TfOH
3
cis-4ab MsOH
8.9
3
trans-4ab MsOH
23.2
15.2
10.2
þ14.3
þ6.3
þ1.3
3
cis-4ab MsOH
3
trans-4ab MsOH
3
a ΔGo(cisꢀtrans) = þ3.1 kcal/mol for [4ab H]þ, ΔGo(cisꢀtrans) = þ1.8 kcal/mol for 4ab TfOH, ΔGo(cisꢀtrans) = þ1.8 kcal/mol for 4ab MsOH. b ΔGq
3
3
3
values were calculated based on the more stable trans-substrate.
8471
dx.doi.org/10.1021/ja202576e |J. Am. Chem. Soc. 2011, 133, 8470–8473

Journal of the American Chemical Society
COMMUNICATION
Table 3. Solvent Effect on anti/syn Selectivity of 3aaa
entry
catalyst
solvent
% yield of 3aa
anti/sync
1b
2
Tf2NH
Tf2NH
CSA
CH3CN
86
0
95: 5
—
DMF
3
CH3CN
50
45
77
33: 67
25: 75
13: 87
4
CSA
ClCH2CH2Cl
CF3C6H5
Figure 3. Orbital interactions in the transition states for the ring-
5
CSA
opening of [4 H]þ. TMS and oxazolidinone moieties are omitted for
a Conditions: 1a, 2a (1.2 equiv), catalyst (20 mol %), solvent (0.3 M),
3
rt, 24 h. b Catalyst (10 mol %), 30 min. Ratios were determined by
c
clarity.
1H NMR.
Table 4. Stereoselective Synthesis of r,β-Unsaturated Ami-
dines Catalyzed by Tf2NH or CSAa
a Conditions: 1, 2 (1.2 equiv), rt. b Method A: Tf2NH (20 mol %),
MeCN (0.3 M), 1ꢀ6 h. Method B: CSA (20 mol %), TFT (0.3 M),
24ꢀ48 h. c The ratios were determined by 1H NMR. d Reactions were
carried out at 60 °C.
Figure 4. The most stable transition states leading to anti- and syn-3ab
in the presence of TfOH and MsOH, respectively. Selected distances in
Å. ΔΔGq (kcal/mol) between TS A and TS D are shown in parentheses.
intimately interacts with a counteranion. For the solvated-ion
model, preference for TS D giving anti-3 over TS A giving syn-3
(þ1.4 kcal/mol)14 might be explained by the steric effect. In the
cis-orientation, the steric repulsion of the two aromatic rings
would somewhat destabilize TS A. For the ionic pair model, the
calculated structure of TS D indicated steric and/or electronic
repulsion between the aromatic ring on C-4 and the sulfonate
anion (details for the important atomic distances of TSs A and D
in the Supporting Information [SI]). Consequently, the ring-
opening reaction would proceed preferably via TS A in the
presence of MsOH or CSA (Figure 4).
To further clarify the proposed mechanisms and improve
selectivity, we next examined the effect of solvents on anti/syn
selectivity in the reaction of 1a with 2a (Table 3). When the
reaction was carried out in a polar solvent (CH3CN) in the
presence of Tf2NH, the anti-selectivity of 3aa was slightly in-
creased relative to toluene (Table 3, entry 1 vs Table 1, entry 1).
Unfortunately, the reaction did not proceed in more polar DMF
(Table 3, entry 2). A considerable solvent effect was observed in
the reaction with CSA. The proportion of anti-3aa versus syn-3aa
increased in polar solvents such as CH3CN (entry 3), whereas
nonpolar solvents, such as toluene and trifluorotoluene (TFT),
led to preferable formation of syn-3aa (entries 4 and 5). The
experimentally observed solvent effect supports the hypothesis
that the Tf2NH- and CSA-catalyzed reactions proceed preferably
via the naked cation model and the intimate ionic-pair model,
respectively. Thus, the reaction using Tf2NH in a polar solvent
would be optimal for selective synthesis of the anti-adduct,
whereas the syn-adduct would be selectively produced by using
torquoselectivity is predominantly determined by the substitu-
ents on the ammonium nitrogen. We propose that the key
interaction is a hyperconjugation between the ammonium pro-
ton (NþꢀH) and the π system of the 4-membered ring.
Stabilization resulting from alignment of the filled σNꢀH orbital
with the unfilled π*CdC orbital of the alkene might be expected in
TSs A and D (Figure 3). In contrast, the stabilizing contribution
of the interaction between the σNꢀAr orbital and the π*CdC
orbital in TSs B and C would be smaller, and an interaction
between σNꢀH and π*CdC is unavailable.13 A related interaction
was proposed to explain torquoselectivity in the ring-opening of
2-azetine (a neutral amine). Houk et al. suggested that one of the
lone pairs on the nitrogen atom controls the torquoselectivity by
aligning with the π* orbital of the alkene.7a
In the absence of a counteranion (Table 2, entries 1.1ꢀ1.4),
TS D giving anti-3ab was calculated to be the most stable
transition state, which is consistent with the experimental results
using TfOH. When a sulfonate anion was present, TS A giving
syn-3ab was calculated to be the most stable transition state
(entries 2.1ꢀ2.4 and 3.1ꢀ3.4). These computational results
agree with the MsOH-catalyzed reaction but are inconsistent
with TfOH-catalyzed one. Thus, the results indicate that the ring-
opening reaction of azetine 3 with strong Brønsted acid catalysts
such as TfOH and Tf2NH might be preferably promoted by a
solvated model. The experiment using an HBF4 catalyst, which
would produce a naked ammonium cation with a coordinate
saturated anion (BF4ꢀ), strongly supports this mechanism.
However, weaker sulfonic acids such as MsOH might be favored
by an ionic pair model, in which the cationic NꢀH moiety
8472
dx.doi.org/10.1021/ja202576e |J. Am. Chem. Soc. 2011, 133, 8470–8473

Journal of the American Chemical Society
COMMUNICATION
CSA in a nonpolar solvent (details for the computational study in
the SI).
Hietbrink, B. N.; Thomas, B. E., IV; Nakamura, K.; Kallel, E. A.; Houk,
K. N. J. Org. Chem. 2001, 66, 6669–6672. (b) Bachrach, S. M.; Liu, M.
J. Org. Chem. 1992, 57, 209–215. (c) Mangelinckx, S.; Van Speybroeck,
V.; Vansteenkiste, P.; Waroquier, M. W.; De Kimpe, N. J. Org. Chem.
2008, 73, 5481.
(8) Studies without discussion of the torquoselectivity: (a) Kai, H.;
Iwamoto, K.; Chatani, N.; Murai, S. J. Am. Chem. Soc. 1996,
118, 7634–7635. (b) Shindo, M.; Oya, S.; Sato, Y.; Shishido, K.
Heterocycles 2000, 52, 545–548.
(9) Niwayama and Houk have reported that the torquoselectivity of
4π ring-opening of acetylcyclobuten was changed in the absence or
presence of Lewis acid. Niwayama, S.; Houk, K. N. Tetrahedron Lett.
1993, 34, 1251–1254.
(10) Hehre, W. J.; Radom, L.; Schleyer, P. V. R.; Pople, J. A. Ab Initio
Molecular Orbital Theory; John Wiley: New York, 1986 and references
cited therein.
Having determined the optimal conditions for selective synth-
esis of both isomers, we then examined the scope of the reaction
using several ynecarbamates and aldimines as substrates. As
shown in Table 4, anti- and syn-3 were obtained in good to
excellent yield and selectivity in reactions with Tf2NH and CSA,
respectively. The stereoselectivity change was also observed in
the reaction of ynesultam 1d (entries 9 and 10), although no anti
selectivity was observed in the reaction with Tf2NH.
In conclusion, we have shown that external factors, such as the
Brønsted acidity of the catalyst and the polarity of the solvent,
affect the torquoselectivity of 4π ring-opening of 2-azetine
compounds. The inward/outward torquoselectivity may be
inverted by the use of Tf2NH or CSA. This is the first successful
example of a catalyst-controlled selectivity switch in electrocyclic
ring-opening reactions.
(11) For a review: see Seeman, J. I. Chem. Rev. 1983, 83, 83–134.
(12) Cunninham, I. D.; Hagarty, A. F. J. Chem. Soc., Perkin Trans. 2
1986, 537–541.
(13) The electronic effect of the N-Ar of 4 on the torquoselectivity
was also evaluated. See SI.
’ ASSOCIATED CONTENT
(14) The anti/syn selectivities based on the calculated ΔΔGqare
estimated to be 91: 9 and 11: 89 in the solvated and ionic pair models,
respectively, at 25 °C.
S
Supporting Information. Details of the computational
b
study, experimental methods, spectral data for all new com-
pounds, and X-ray crystallographic data of anti-3ac. This material
is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
takemoto@pharm.kyoto-u.ac.jp (Y.T.); kay-t@pharm.kyoto-u.
ac.jp (K.T.)
’ ACKNOWLEDGMENT
This research was partially supported by a Grant-in Aid for
Scientific Research from MEXT. N.S. also acknowledges the
Japan Science Society.
’ REFERENCES
(1) For reviews, see: (a) Dolbier, W. R., Jr.; Koroniak, H.; Houk,
K. N.; Sheu, C. Acc. Chem. Res. 1996, 29, 471–477. (b) Dolbier, W. R., Jr.;
Koroniak, H.; Burton, D. J.; Bailey, A. R.; Shaw, G. S.; Hansen, S. W.
J. Am. Chem. Soc. 1984, 106, 1871–1872. (c) Kirmse, W.; Rondan, N. G.;
Houk, K. N. J. Am. Chem. Soc. 1984, 106, 7989–7991. (d) Shindo, M.;
Mori, S. Synlett 2008, 2231–2243.
(2) (a) Murakami, M.; Miyamoto, M.; Ito, Y. J. Am. Chem. Soc. 2001,
123, 6441–6442. (b) Lee, P. S.; Zhang, X.; Houk, K. N. J. Am. Chem. Soc.
2003, 125, 5072–5079. (c) Shindo, M.; Mori, Kita, T.; Kumagai, T.;
Matsumoto, K.; Shishido, K. J. Am. Chem. Soc. 2006, 128, 1062–1063.
(d) Liang, Y.; Jiao, L.; Zhang, S.; Yu, Z.-X.; Xu, J. J. Am. Chem. Soc. 2009,
131, 1542–1549. (e) Yoshikawa, T.; Mori, S.; Shindo, M. J. Am. Chem.
Soc. 2009, 131, 2092–2093. (f) Um, J. M.; Xu, H.; Houk, K. N.; Tang, W.
J. Am. Chem. Soc. 2009, 131, 6664–6665. (g) Qi, H.; Li, X.; Xu, J. Org.
Biomol. Chem. 2011, 9, 2702–2714.
(3) Shindoh, N.; Takemoto, Y.; Takasu, K. Chem.—Eur. J. 2009,
15, 7026–7030.
(4) (a) De Korver, K. A.; Li, H.; Lohse, A. G.; Hayashi, R.; Lu, Z.;
Zhang, Y.; Hsung, R. P. Chem. Rev. 2010, 110, 5064–5106. (b) Evano,
G.; Coste, A.; Jouvin, K. Angew. Chem., Int. Ed. 2010, 49, 2840–2859.
(5) Henry, R. M.; Shoemaker, R. K.; DuBois, D. L; DuBois, M. R.
J. Am. Chem. Soc. 2006, 128, 3002–3010.
(6) Eckert, F.; Leito, I.; Kaljurand, I.; Kutt, A.; Klamt, A.; Diedenhofen,
M. J. Comput. Chem. 2009, 30, 799–808.
(7) 4π Electrocyclic ring-opening of neutral 2-azetines has been
studied with discussion of the torquoselectivity: (a) Walker, M. J.;
8473
dx.doi.org/10.1021/ja202576e |J. Am. Chem. Soc. 2011, 133, 8470–8473