Article pubs.acs.org/accounts
Computational Approach to Diarylprolinol-Silyl Ethers in Aminocatalysis Published as part of the Accounts of Chemical Research special issue “Computational Catalysis for Organic Synthesis”. Kim Søholm Halskov, Bjarke S. Donslund, Bruno Matos Paz, and Karl Anker Jørgensen* Department of Chemistry, Aarhus University, DK-8000 Aarhus C, Denmark CONSPECTUS: Asymmetric organocatalysis has witnessed a remarkable development since its “re-birth” in the beginning of the millenium. In this rapidly growing field, computational investigations have proven to be an important contribution for the elucidation of mechanisms and rationalizations of the stereochemical outcomes of many of the reaction concepts developed. The improved understanding of mechanistic details has facilitated the further advancement of the field. The diarylprolinol-silyl ethers have since their introduction been one of the most applied catalysts in asymmetric aminocatalysis due to their robustness and generality. Although aminocatalytic methods at first glance appear to follow relatively simple mechanistic principles, more comprehensive computational studies have shown that this notion in some cases is deceiving and that more complex pathways might be operating. In this Account, the application of density functional theory (DFT) and other computational methods on systems catalyzed by the diarylprolinol-silyl ethers is described. It will be illustrated how computational investigations have shed light on the structure and reactivity of important intermediates in aminocatalysis, such as enamines and iminium ions formed from aldehydes and α,β-unsaturated aldehydes, respectively. Enamine and iminium ion catalysis can be classified as HOMO-raising and LUMO-lowering activation modes. In these systems, the exclusive reactivity through one of the possible intermediates is often a requisite for achieving high stereoselectivity; therefore, the appreciation of subtle energy differences has been vital for the efficient development of new stereoselective reactions. The diarylprolinol-silyl ethers have also allowed for novel activation modes for unsaturated aldehydes, which have opened up avenues for the development of new remote functionalization reactions of poly-unsaturated carbonyl compounds via di-, tri-, and tetraenamine intermediates and vinylogous iminium ions. Computational studies have played a pivotal role in the elucidation of the regioselectivities observed for such systems because these pose a challenge due to the presence of multiple reactive sites in these intermediates. Charge distribution and π-orbital coefficient calculations have been applied to explain the observed regioselectivity of the given reactions. The calculation of more elaborate energetic pathways has allowed for in silico identification of high-energy intermediates, such as zwitterions, and transition-state structures, which have also provided information on the driving force controlling the reaction course and outcome.
1. INTRODUCTION
aldehydes, respectively. Furthermore, these catalysts have allowed for novel activation modes of unsaturated aldehydes and cascade reactions and have been combined with transitionmetal catalysis and biocatalysis.4 An important part of the development of novel catalytic asymmetric reactions is elucidation of mechanistic details, based on experimental and computational studies. In this Account, we will demonstrate how computational investigations of diarylprolinol-silyl ethers have provided an understanding of intermediates and reaction pathways, which has been fundamental for the advancement of organocatalysis. We will outline the various activation modes and reactions, where computational investigations have helped rationalizing experimental results.
Organocatalysis has manifested itself as a cornerstone for the construction of chiral molecules. Since the start of the millennium, organocatalysis has expanded from being applied in enantioselective reactions, which were also possible using enzymes and metal complexes, to novel enantioselective reactions, concepts, and activation modes.1 The advent of organocatalysis has led to new opportunities in the synthesis of chiral molecules, and its application has found its way from academia to industry. The activation of organic molecules by organocatalysts can be mediated through covalent or noncovalent interactions.2 Among the organocatalysts used for covalent activation, chiral secondary amines play a central role.3 Within this category, the diarylprolinol-silyl ethers are among the most extensively applied and have demonstrated excellent stereocontrolling properties for αand β-functionalizations of aldehydes and α,β-unsaturated © XXXX American Chemical Society
Received: January 12, 2016
A
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
which leads to a HOMO-energy increase compared with the corresponding enol. The application of the same catalytic system for the activation of α,β-unsaturated aldehydes can form either a dienamine or an iminium ion. In the former intermediate, the HOMO energy is raised, while for the latter, the LUMO energy is lowered relative to the α,β-unsaturated aldehyde. 2,4-Dienals can also be activated by diarylprolinol-silyl ethers in two ways: by HOMO energy raising, found for trienamines and crosstrienamines, and LUMO energy lowering via the vinylogous iminium ion. Finally, diarylprolinol-silyl ethers have been used for activation of poly-unsaturated aldehydes involving tetraenamines.
The computational methods used in many of the investigations of intermediates and reaction paths are based on density functional theory (DFT). We will not go into detail with functionals and basis sets used in these DFT calculations, and the reader should consult another review for these.5 Our mission is to illustrate the application of computational studies for diarylprolinol-silyl ethers catalyzed reactions. The general approach for studying these reactions starts by conformational searches using force field calculations, which provides starting points for optimization using DFT calculations. The advent of the current computing power has made it possible to perform calculations to study transition-state structures of key organocatalytic steps using hybrid functionals, of which B3LYP with basis set 631G(d) stands out due to its efficient compromise between computational cost and agreement with experimental results. However, one should be aware that dispersion corrections are essential to capture the key aspects of chemical structures and catalyst design.6 The discussion originates in the different activation modes of aldehydes and their stereoselective reactions. Furthermore, we will discuss reactions where it has been necessary to go beyond the diarylprolinol-silyl ether catalysts to develop novel reaction concepts. Unless otherwise noted, all energies are free energies reported in kcal/mol.
3. ENAMINES The reaction between an aldehyde and a secondary amine catalyst gives an enamine intermediate and the stereochemical outcome of the α-functionalization depends on the approach of the electrophile to the enamine. The diarylprolinol-silyl ethers were first introduced for the enantioselective α-sulfenylation of aldehydes.8 Shortly after, the catalytic system was applied for the enantioselective α-fluorination of aldehydes.9 To account for the high enantioselectivity for the α-fluoroaldehydes (>91% ee), DFT calculations were performed and showed that the lowest energy structure of the enamine has one of the 3,5trifluoromethylphenyl groups in the diarylprolinol-silyl ether 1a shielding the Re face (Figure 2). Consequently, the electrophilic fluorine approaches the Si face providing enantioselectivities in agreement with experimental observation.
2. ACTIVATION MODES The diarylprolinol-silyl ether catalysts can activate aldehydes and unsaturated aldehydes in different ways (Figure 1).7 For aldehydes, condensation provides the enamine intermediate,
Figure 2. DFT-calculated (B3LYP/6-31G*) model of the enamine formed from isovaleraldehyde and 1a.
Further computational studies of the enamine for four possible intermediates, formed from two different aldehydes (R = Me, tBu) and 1a, focused on the relative energies of configurations and conformations (Figure 3).10 The free energies of the four enamines from propanal and 1a show that E-s-trans (0 kcal/mol) and E-s-cis (−0.1 kcal/mol) having E-configuration are more stable than Z-s-trans (1.9 kcal/ mol) and Z-s-cis (5.6 kcal/mol) having Z-configuration. The
Figure 1. HOMO energy raising and LUMO energy lowering activations of aldehydes and unsaturated aldehydes applying diarylprolinol-silyl ethers.
Figure 3. Relative energies (kcal/mol) of enamine geometries (B3LYP/ 6-31G(d)). B
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research energy difference originates from steric repulsion between the methyl group and the hydrogens adjacent to the nitrogen in the pyrrolidine. This is confirmed by changing the methyl with the sterically demanding tert-butyl group, which gives an increase in energy difference for Z-s-cis and Z-s-trans compared with E-strans (12.1 and 7.5 kcal/mol, respectively). The calculations suggest that the two major conformers present in the reaction mixture are E-s-trans and E-s-cis. Comparison of the calculated intermediate E-s-trans with the X-ray structure of the enamine having R = Ph in the aldehyde and Ar = Ph (catalyst 1b) showed that the CC bond had E-configuration and the conformation of the bond between the sp2-C and the N atom was s-trans.11 Thus, the X-ray structure supports the intermediate proposed in Figure 2 and E-s-trans in Figure 3. Furthermore, during NMR investigations only E-s-trans was observed in solution.12 These experimental findings show that the calculations failed to describe the relative stability of E-s-trans vs E-s-cis, which were predicted to be equally stable. The similarity of the calculated Es-trans structure and the crystal structure is notable; the major difference observed is the orientation of the substituent at the stereocenter in the pyrrolidine. In the crystal structure, the OTMS-group is in sc-exo conformation, while it is sc-endo in the calculated structure (Figure 4). The difference between the two
Figure 5. Two lowest energy transition-state structures (kcal/mol) for the α-fluorination of aldehydes (B3LYP/6-31G(d)).
an enantioselectivity of 96% ee in favor of the (S)-product, in agreement with the experimental result (97% ee). In the reaction, the fluorine is transferred to the enamine in an SN2-like reaction. The electrophilic fluorine is partially transferred in the transition-state structure and is positioned closer to the nitrogen in NFSI (1.711 Å) than to the carbon in the enamine (2.190 Å), indicating an early transition state. The energies of the transition states for the Z-configurated enamines are significantly higher than those for the E-configurated enamines (≥10.5 kcal/mol). Calculations also showed that attack on the “upper side” (Re face) of the E-s-trans is disfavored (≥9.5 kcal/mol) due to shielding of the enamine by the bulky substituent in the catalyst. The α-amination of aldehydes using 1a and diethyl azodicarboxylate has been studied computationally (B3LYP/631G(d)) to account for the observed (S)-α-aminated aldehyde.9a,10 Calculations of different pathways showed that amination of E-s-trans had the lowest transition-state energy, compared with, for example, E-s-cis. The lower transition-state energy for the fluorination and amination of the E-s-trans enamine may be attributed to a change in hybridization from sp2 to sp3 of the nucleophilic α-carbon as it attacks the incoming electrophile. This change in hybridization induces steric clash with the substituent at the stereogenic center in the pyrrolidine, which shields the opposite face of the enamine. This mode of stereoinduction, based on steric shielding of the incoming electrophile, in combination with control of the geometry of the reacting enamine has proven to be a general recipe for enantioselective α-functionalization of aldehydes. However, it should be noted that elaborate studies have shown that for some enamine-catalyzed reactions a more complex picture emerges in which “downstream species” play a role in the stereoinduction.13 Computational studies have been applied to provide insight into mechanistic details of the enantioselective intramolecular [6 + 2]-cycloaddition of fulvene derivatives catalyzed by 1b (Figure 6).14 The authors modeled the [6 + 2]-cycloaddition using dimethyl enamine as model for the enamine formed from condensation of the aldehyde with 1b. Both the inter- and intramolecular versions of the reaction were studied. For the intermolecular version, geometry optimization was performed employing DFT (B3LYP) and MP2 (6-311G(d,p)), while CCSD(T) at the basis set limit
Figure 4. Substituent orientations at the stereocenter in the diarylprolinol-silyl ethers.
structures might be due to differences between the studied systems. In the calculations, an alkyl substituted aldehyde and 1a were applied, while the crystal structure was based on an aromatic aldehyde and 1b. The latter shows that the silyl-ether protecting group is responsible for shielding the nucleophilic carbon of the enamine. According to the calculated enamines in Figure 3, E-s-trans and E-s-cis are close in energy for R = Me, while for R = t-Bu, E-strans is favored. However, for the fluorination reaction applying catalyst 1a, high enantioselectivity is achieved, which was intriguing at the time of the study considering that the two conformers of the enamine were believed to be close in energy and thus the reaction was expected to give low selectivity. Computational studies on the α-fluorination reaction tried to shed light on this peculiarity;10 however, as described above, experimental work has since shown E-s-trans to be significantly more stable. This is an example of a discrepancy between computational and experimental work, which, in this case, could be due to the relatively low level of theory applied for the computational study. Optimization of the transition-state structures for the approach of N-fluorobenzenesulfonimide (NFSI) to the enamine obtained from 1a and 3,3-dimethylbutanal (R = t-Bu, Figure 3) for the two most likely intermediates E-s-trans and E-scis gave TS-A as the lowest energy transition state (Figure 5, left). The calculations showed that NFSI approaches the Si face of E-strans leading to (S)-α-fluoroaldehyde. The second most stable transition state found (TS-B) is 2.4 kcal/mol higher in energy than TS-A and corresponds to fluorination of the Re face of the E-s-cis, forming the (R)-configured product (Figure 5, right). The energy difference between TS-A and TS-B corresponds to C
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
kinetically favored compared with 10, while the transition-state energy for cyclization to 9 is relatively high (26.0 kcal/mol). Formation of 8 from 7 is reversible and eventually the formation of 10 as opposed to 9 should prevail. Recently, it was shown that cyclopropanes can be activated by catalytic formation of an enamine. By combination of cyclopropylacetaldehyde with 1b, the cyclopropane moiety was activated, which facilitated an unexpected asymmetric [2 + 2]cycloaddition reaction with 3-olefinic oxindoles (Figure 9).15 Figure 6. Enantioselective [6 + 2]-cycloaddition of fulvene derivatives catalyzed by 1b.
was used to obtain single point enthalpies. The reaction proceeds via a stepwise mechanism initiated by addition of enamine 3 to fulvene 2 resulting in the zwitterionic intermediate 4 (Figure 7). Figure 9. Enamine-activation of cyclopropylacetaldehydes leading to [2 + 2]-cycloaddition with 3-olefinic oxindoles.
The nature of the activation of the cyclopropane in the cyclopropylacetaldehyde was investigated on a simplified system. These calculations revealed that enamine formation weakens the C1−C2-bond in the cyclopropane as observed by an elongation (1.559 Å) compared with the corresponding iminium ion (1.522 Å) and cyclopropylacetaldehyde (1.517 Å) (Figure 10). Figure 7. Intermolecular [6 + 2]-cycloaddition of fulvene and an enamine. Zero-point energy corrected relative enthalpies (kcal/mol) at 298 K (CCSD(T)limit//MP2/6-311G(d,p)) taking solvation into account (CPCM).
Formation of 4 was slightly unfavored and proceeded with a transition-state energy of 9.1 kcal/mol. A low transition-state energy (5.2 kcal/mol) for the cyclization to the energetically favored product 5 was calculated. The product 6 and dimethylamine, arising from elimination of the dimethylamine from 5, were higher in energy than 5. Interestingly, the authors found that the intramolecular version proceeded in a different fashion (Figure 8). Calculations indicated that fulvene 7 would form the experimentally observed cis-fused tricyclic product 10 directly by an asynchronous concerted pathway. In contrast to this, formation of the transfused cycloadduct 9 was stepwise and takes place via formation of the zwitterionic intermediate 8. This intermediate (8) is
Figure 10. Calculated bond lengths (wB97XD/pcseg-1) in activated and unactivated cyclopropyl moieties.
Optimizing the enamine formed from condensation of the cyclopropylacetaldehyde with pyrrolidine shows conformational changes. In the cyclopropylacetaldehyde, the C1−C2 bond is in the same plane as the aldehyde functionality; however, as enamine formation occurs, the cyclopropyl moiety rotates to be perpendicular to the enamine functionality to allow for favorable interactions between the π-orbital of the enamine and the σ*C−C orbital of the cyclopropyl moiety (Figure 11). It was postulated that donation of electron density from this π-orbital into the σ*C−C orbital facilitates cleavage of the C1−C2-bond in the cyclopropyl
Figure 8. Relative energies (kcal/mol) for the intramolecular [6 + 2]cycloaddition of fulvenes (B3LYP/6-311G(d,p).
Figure 11. Optimized structure (wB97XD/pcseg-1) of enamine formed from cyclopropylacetaldehyde. D
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
lprolinol-silyl ethers have included computational studies. One of these is the enantioselective phosphonylation of α,β-unsaturated aldehydes (Figure 14).18
ring leading to the formation of a dienamine that reacts with 3olefinic oxindoles providing cyclobutane products (Figure 9).
4. IMINIUM IONS The condensation of a secondary amine with an α,β-unsaturated aldehyde generates an iminium ion. This intermediate plays a key role in reactions of α,β-unsaturated aldehydes, as the LUMO energy of the iminium ion is lower than that in the α,βunsaturated aldehyde. Several studies of the induction of stereoselectivity in iminium ion catalysis have been performed. The first thing to consider is the influence of the E/Z ratio of the CN bond in the iminium ion on the stereochemical outcome. If both isomers showed the same reactivity, the ratio between the enantiomers of the product would depend on the E/Z ratio of the iminium ion.16 For iminium ions formed by condensation of cinnamaldehyde with diarylprolinol-silyl ethers, the E/Z ratio in solution ranges from 88:12 to 98:2. Although there is a preference for the Eisomer, these ratios imply enantioselectivity ranging from 75% to 95% ee, which is inferior to the experimentally observed values.16 MP2 calculations showed the Z-isomer to be 1.73 kcal/mol higher in energy than the E-isomer, consistent with experimental E/Z-ratios (Figure 12).17
Figure 14. Enantioselective phosphonylation of α,β-unsaturated aldehydes.
The reaction between trimethyl phosphite and the iminium ion from 2-hexenal was used as a model. The transition-state structures for the approach of trimethyl phosphite were optimized, and three approaches were investigated, two for the E-iminium ion and one for the Z-iminium ion (Figure 15). The
Figure 12. Relative energies (kcal/mol) of E- and Z-iminium ions based on MP2-calculations (6-311G(d,p)). Figure 15. Relative transition-state energies (kcal/mol) for approaches of trimethyl phosphite to E- and Z-iminium ions (B3LYP/6-31G(d)).
The high enantioselectivities observed suggest that the Eiminium ion is more reactive than the Z-iminium ion. An explanation for this comes from the changes in hybridization of the β-carbon and nitrogen upon the formation of an enamine by addition of a nucleophile.16 While the newly formed tetrahedral center is bulkier, the pyramidalization of the nitrogen forces the enamine side chain to move toward the bulky group of the catalyst, and these effects should be more energetically demanding for the Z-iminium ion (Figure 13). The predominance of the E-isomer, in combination with these effects, might explain the observed enantioselectivities. A series of enantioselective additions of heteroatomic nucleophiles to α,β-unsaturated aldehydes catalyzed by diary-
transition states in which the trimethyl phosphite attacks the iminium ion anti to the sterically shielding group are lower in energy (0.0 for the E-iminium ion, 2.5 kcal/mol for Z-iminium ion) compared with the syn-approach (9.2 kcal/mol). The addition of N-heterocycles to α,β-unsaturated aldehydes (Figure 16) was investigated by DFT calculations for three transition-state structures, two for the E-iminium ion and one for the Z-iminium ion.19 The calculations showed that the transition state for the addition to the E-iminium ion is 5.7 kcal/mol lower in energy than the Z-iminium ion. The addition anti to the sterically shielding group of the catalyst is 1.8 kcal/mol lower in energy than the syn-approach, in agreement with experimental results (Figure 17).
Figure 16. Enantioselective addition of N-heterocycles to α,βunsaturated aldehydes.
Figure 13. Explanation of favored E-isomer reaction path for nucleophilic attack to E- and Z-iminium ions. E
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
Figure 17. Relative transition-state energies (kcal/mol) for the 1,4addition of triazole to an iminium ion (B3LYP/6-311G(d,p)/CPCM).
The calculations also included an investigation of proton transfers between the protonated triazole and the enamine, with and without mediation by water (Figure 18). The former pathway is 2.3 kcal/mol lower in energy than the intramolecular proton transfer.
Figure 19. Relative electronic energies (kcal/mol) of four s-trans (C−N bond) dienamines (sc-endo) (B3LYP/6-31G(d)). Energies are not temperature or zero-point corrected. Later studies revealed sc-exo to be more stable (right).
Figure 20. Dienamine-catalyzed γ-amination of α,β-unsaturated aldehydes.
Figure 18. Relative energies (kcal/mol) for proton transfers following the addition of triazole to the iminium ion (B3LYP/6-311G(d,p)/ CPCM).
simple addition of the electrophilic amination reagent to the most stable s-trans,E,s-trans,E-dienamine. Instead, a Diels− Alder reaction with the s-trans,E,s-cis,E-dienamine was found to be kinetically and thermodynamically favored (Figure 21). This could be formed by rotation around the C−C-bond from the strans,E,s-trans,E-dienamine with a barrier of 7.5 kcal/mol. A reaction pathway involving a Diels−Alder reaction followed by hydrolysis of the aminal intermediate would account for the observed stereochemical configuration of the product. The
5. DIENAMINES Dienamines can be generated by deprotonation of iminium ions, formed from condensation of diarylprolinol-silyl ethers with α,βunsaturated aldehydes. Compared with an enamine, the presence of an additional conjugated double bond increases the number of possible geometries of the dienamine, which may induce different stereoselectivities upon reaction. Furthermore, additional reactive centers are present, indicating that regioselectivity might be a challenge. Nevertheless, a number of highly stereoselective transformations based on dienamines have been reported. In the initial work introducing dienamines, the electronic energies of the different intermediates formed from condensation of 1a with 2-pentenal were calculated.20 These calculations indicated that the s-trans,E,s-trans,E-dienamine was the lowest energy intermediate, with the s-trans,E,s-trans,Z-dienamine being less stable (Figure 19). This was supported by NMR studies showing these two intermediates in a 2:1 ratio in solution. Later studies of these intermediates revealed that an sc-exo conformation was the most stable orientation at the exocyclic C− C-bond.17 The calculated energies of the intermediates having scis conformation at the C−N-bond were higher in energy compared with their s-trans counterparts. Dienamine catalysis was first applied for the asymmetric γamination of α,β-unsaturated aldehydes (Figure 20).20 The product was formed in excellent regio- and enantioselectivity. It was noticed, however, that the opposite stereochemical configuration was found compared with what would be expected from reaction through the most stable s-trans,E,s-trans,Edienamine. The mechanism of the reaction was subjected to computational investigations to shed light on the reaction pathway. These calculations showed that the reaction did not proceed through a
Figure 21. Intermediates, reaction paths, and relative energies (kcal/ mol) for dienamine-catalyzed γ-amination of α,β-unsaturated aldehydes (B3LYP/6-31G(d)/CPCM//B3LYP/6-31G(d)). F
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research reaction pathways concerning addition through the α-position of the dienamine were found to have significantly higher transitionstate energies and were thermodynamically disfavored compared with the γ-aminated intermediates due to loss of conjugation. The diarylprolinol-silyl ethers have shown limitations in remote functionalizations of dienamines, and in order to provide solutions for such challenges, bifunctional organocatalysts have been developed.21 Such a catalyst (1c), based on the pyrrolidine scaffold carrying a hydrogen-bond donating squaramide, was applied in an asymmetric formal [2 + 2]-cycloaddition between α,β-unsaturated aldehydes and nitroolefins (Figure 22). The
Figure 24. Trienamine-catalyzed Diels−Alder reaction yielding chiral spirocyclic oxindoles.
The reaction displayed complete regioselectivity for the β,εpositions of the 2,4-dienal. In an attempt to understand the preference for β,ε-reactivity over ipso,γ-reactivity, which is known from dienamines, computational methods were employed. The relevant conformers of the trienamine were optimized using DFT calculations, and the linear all-trans conformer A was lowest in energy (Figure 25) in accordance with
Figure 22. Asymmetric formal [2 + 2]-cycloaddition promoted by a bifunctional organocatalyst.
catalyst design includes a pyrrolidine moiety, which activates the aldehyde by dienamine formation, and a hydrogen-bonding squaramide coordinating the nitro group thereby activating and directing the olefin toward reaction at the remote dienamine πbond. Given that stereoselective [2 + 2]-cycloadditions are challenging to perform and that catalyst 1c was applied, the mechanism of the reaction was subjected to DFT calculations. These pointed toward a stepwise mechanism initiated by a conjugate addition of the dienamine to the nitroolefin (Figure 23). This step occurs with a transition-state energy of 9.8 kcal/ mol, whereas the second barrier was determined to be 17.6 kcal/ mol.
Figure 25. Conformations and energies (kcal/mol) of the trienamine (B3LYP/6-31G(d)).
NMR studies. The two conformers B and C, which might take part in Diels−Alder reactions, were close in energy and only 3.8 and 3.6 kcal/mol higher in energy than A, respectively. The barrier for single-bond rotation around C2−C3 (11.8 kcal/mol) was higher than the rotation barrier around C4−C5 (10.0 kcal/ mol). Coefficients for the relevant HOMO and LUMO in B and C were calculated (HF/STO-3G) and the HOMO-coefficients on the carbons of the reacting dienes are −0.32 (C1, B), 0.43 (C4, B), −0.23 (C3, C), and 0.34 (C6, C). While these coefficients pointed to B as the reactive conformer, the energy levels of the orbitals pointed to C; the HOMO of C, set up for interaction with the LUMO of a dienophile, was 2.17 eV higher in energy than the corresponding HOMO in B. Also, the LUMO of C, capable of interacting with the HOMO of a dienophile, was 4.41 eV lower than the corresponding LUMO in B. Hence, it was concluded that the energies of the FMOs and the lower rotational barrier around C4−C5 favor reactivity at the remote β,ε-positions. A more comprehensive computational study was performed on the trienamine-catalyzed thio-Diels−Alder reaction (Figure 26). The reaction between 2,4-dienals and dithioesters catalyzed by ent-1b gave chiral dihydrothiopyrans.23 An interesting regioselectivity was observed as a bond is formed between the nucleophilic ε-position in the trienamine and the sulfur of the thiocarbonyl rather than the carbon. The regioselectivity was probed applying DFT calculations. Initially, a representative dithioester was optimized and the relevant LUMO
6. TRIENAMINES Trienamine catalysis was introduced as a new activation concept in 2011. It was demonstrated that stereoselective Diels−Alder reactions between 2,4-dienals and 3-olefinic oxindoles could be facilitated by 1d (Figure 24).22
Figure 23. Calculated reaction pathway (kcal/mol) for the formal [2 + 2]-cycloaddition by 1c (M062x/6-31G(d)).
Figure 26. Trienamine-catalyzed thio-Diels−Alder reaction. G
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research was slightly larger at the sulfur of the thiocarbonyl (Figure 27). Localized charges were calculated and a charge separation of 0.2 was found over the CS-bond with the electronegative part localized at the sulfur.
Figure 27. LUMO and charge distribution for simplified thiodienophile.
Detailed insights into regio- and diastereoselectivity were obtained by investigation of four reaction pathways modeled with a simplified trienamine. First the endo- and exo-approaches leading to the observed regiochemistry were examined, and nonconcerted pathways involving high-energy zwitterionic intermediates were discovered (Figure 28). The initial attack leading to the observed major diastereoisomer by an endopathway had a transition-state energy of 13.9 kcal/mol, while the exo-pathway had a transition-state energy of 15.0 kcal/mol. Due to relatively high barriers for the reverse reaction (>20 kcal/mol), the diastereoselectivity in the reaction was argued to be governed by kinetics and the difference of 1.1 kcal/mol between the two pathways corresponds with the experimental diastereomeric ratio. The routes leading to products not observed experimentally involving attack on the carbon of the dithioesters were investigated as well, but in these cases concerted, although highly asynchronous, cycloaddition pathways were observed because no zwitterionic intermediates were located (Figure 29). Significantly higher barriers were calculated, which explains the observed regioselectivity. Finally, the attention was turned to the enantioselectivity of the reaction. Hence, the bulky group of the catalyst was taken into account, and these calculations were performed with a smaller basis set for geometry optimization to accommodate the larger system. Dienophile approach from the face opposite to the bulky group of the catalyst (Figure 30) was favored by 1.4 kcal/
Figure 29. Energetic paths (kcal/mol) involving initial attack on carbon of the dithioester (wB97xd/tzvp/IEFPCM).
mol, corresponding to 83% ee, which is in the range of experimental observations. Another interesting utilization of trienamine catalysis is the activation of anthracenes. Attachment of an acetaldehyde unit to the central ring allows for catalytic formation of an enamine, which activates the anthracene unit for Diels−Alder reactions with nitroolefins, hereby dearomatizing the central ring (Figure 31).24 Catalyst 1c was applied to facilitate both trienamine activation of the aldehyde and activation of the nitroolefin via hydrogen-bonding. To evaluate the effects of catalyst condensation on the aromaticity, nucleus-independent chemical shifts (NICS) for the anthracene unit were calculated. Negative NICS(1)zz values indicate aromaticity, and addition of an acetaldehyde substituent reduces the aromaticity of the central ring (Figure 32). Changing the substituent to a conjugated enamine further lowers the aromaticity. The energies for the HOMOs of the substituted anthracenes were also calculated, and the HOMO of the enamine species was 0.8 and 0.7 eV higher in energy than the aldehyde and the unsubstituted anthracene, respectively. These results show
Figure 28. Energetic pathways (kcal/mol) leading to the major and minor diastereomers of the observed products (wB97xd/tzvp/IEFPCM). H
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
by the application of cyclic 2,4-dienals reactions proceeded exclusively through the cross-trienamine intermediate as opposed to the intermediate with a linear conjugated system (Figure 33).25 This reaction concept was applied in asymmetric Diels−Alder reactions with, for example, 3-olefinic oxindoles and conjugate additions to vinyl bis-sulfones (Figure 34). The observation that the reaction occurred exclusively through the cross-trienamine intermediate was puzzling since DFT calculations (MPW1K/631+G(d,p)) on a model system of the two nucleophilic intermediates showed that the linear trienamine is favored by 4.2 kcal/mol relative to the cross-trienamine. This was in accordance with NMR-studies, in which only the linear trienamine was detected in solution. The authors postulated that cross-trienamine reactivity may prevail due to a thermodynamic preference for the formation of a dienamine with a fully conjugated π-system following the Diels− Alder reaction. This would, however, not explain the observed preference for conjugate additions to vinyl bis-sulfones via the cross-trienamine over the linear trienamine. In light of these peculiarities, the Diels−Alder reaction with 3olefinic oxindoles was studied applying DFT calculations utilizing model substrates.26 These results revealed that the reactions occurred through a stepwise mechanism, and zwitterionic intermediates could be identified in silico by taking solvent effects into account. With regards to the linear trienamine A, two intermediates F1 and F2, formed from an initial addition reaction with transition-state energies of 20.5 and 17.5 kcal/mol, respectively, were identified (Figure 35). The barriers for subsequent cyclization are small, and thus the formation of the two diastereomeric products D1 and D2 from their respective intermediates F is predicted to occur rapidly. With regards to the cross-conjugated system, intermediate B is thermodynamically less favored than the linear system, in agreement with the previous results. From reaction of B with 3olefinic oxindole C via transition states TS-G1 and TS-G2, the diastereomeric zwitterionic intermediates G1 and G2 are formed (Figure 36). Intermediate G1, from which the experimentally observed major diastereomer of product is formed, has a lower energy barrier for formation than G2. By initially focusing on G1, it is realized that this intermediate may cyclize to form a variety of cycloadducts. Whereas products H1, I1, and J1 are relatively high in energy, the experimentally observed product E1 is more stable and is therefore the thermodynamically favored product from G1. Similar observations were made for G2 leading to the minor
Figure 30. Transition-state structures and relative energies (kcal/mol) leading to the two enantiomers of product (wB97xd/cc-pVDZ/ IEFPCM//wB97xd/6-31G(d)/IEFPCM).
Figure 31. Trienamine activation of anthracenes for Diels−Alder reaction with nitroolefins.
Figure 32. NICS(1)zz values for the three rings (A, B, C) and HOMO energy levels in anthracenes (GIAO-B3LYP/6-31G(d)//M062X/631+G(d,p)).
that catalyst condensation and enamine formation activates the system toward reaction with electron-poor dienophiles.
7. CROSS-TRIENAMINES In 2012, it was shown that asymmetric reactions could be promoted by cross-trienamine intermediates. It was found that
Figure 33. Linear trienamine (top) versus cross-trienamine (bottom) reaction pathways from cyclic 2,4-dienals. I
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
Figure 34. Cross-trienamine-catalyzed conjugate additions (left) and Diels−Alder reactions (right).
Figure 35. Calculated reaction pathways (kcal/mol) for linear trienamine A with 3-olefinic oxindole C (M06-2X/def2-TZVPP/IEFPCM//B97D/631+G(d,p)/IEFPCM).
Figure 36. Calculated pathways for cross-conjugated trienamine B with 3-olefinic oxindole C. Energies are relative to A + C (kcal/mol) (M06-2X/def2TZVPP/IEFPCM//B97D/6-31+G(d,p)/IEFPCM).
J
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
studies were employed to investigate the regioselectivity. The study involved a 1,6/1,4-addition of hydroxyarenes to 2,4-dienals (Figure 38).28 As displayed in Figure 39, four regioselective outcomes were plausible. The sequence could be initiated by a Friedel−Crafts or an oxa-addition of the hydroxyarene to the β-position or to the δposition followed by ring closure. A model system of the vinylogous iminium ion of 2,4-hexadienal and pyrrolidine with various naphthol systems were constructed to investigate the relative transition-state energies and stabilities of the intermediates. Three different nucleophilic species were tested: free 1naphtholate, 1-naphthol/1-naphtholate complex, and 1-naphthol/DABCO complex. After the conformational degrees of freedom of the systems were sampled applying the MMFF force field, full optimization was performed by DFT calculations. The barrier for the reaction leading to the observed product was found to be higher (Figure 39, lower left), while the barriers for addition through the oxygen were lower (right). Nevertheless, all calculated transition-state energies implied that the reactions could take place at room temperature (14 kcal/mol higher than for TS-G1 ruling out a concerted pathway.
8. VINYLOGOUS IMINIUM IONS LUMO-activating aminocatalytic strategies have also been applied to 2,4-dienal systems for functionalization in the δ-
Figure 37. Geometry optimization and π-orbital coefficients calculated with HF/STO-3G. Charges (CHELPG) calculated with HF/6-31G(d). All values are solvent corrected (CPCM).
Figure 38. Vinylogous iminium ion activation of 2,4-dienals and 1,6/ 1,4-addition of hydroxyarenes.
9. CONCLUSION AND OUTLOOK Computational and experimental investigations have worked hand-in-hand in the development and understanding of asymmetric organocatalysis. In this Account, we have documented how important computational studies have been for the advancement of diarylprolinol-silyl ethers in aminocatalysis. For both “the classical intermediates”, enamines and iminium ions, and those made possible by the introduction of the diarylprolinol-silyl ethers, dienamines, trienamines, and vinylogous iminium ions, computational investigations have been important
position. Hayashi et al. studied the addition of nitromethane to 2,4-dienals and exclusively observed β-functionalization.27 They calculated the π-orbital coefficients and localized charges of simplified vinylogous iminium ions (Figure 37). This showed higher charges and orbital coefficients at the β-position (C4) compared with the δ-position (C6) explaining the observed βselectivity. Despite these findings, some stereoselective δ-functionalizations of 2,4-dienals have been disclosed and computational
Figure 39. Pathways for initial addition of 1-naphthol to a vinylogous iminium ion. The three relative energies (kcal/mol) refer to the nucleophile being modeled as 1-naphtholate/1-naphthol−1-naphtholate complex/1-naphthol−DABCO, respectively (wB97xd/pcseg-1/IEFPCM). K
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research
(5) Cheong, P. H.-T.; Legault, C. Y.; Um, J. M.; Cekebi-Ö lcum, N.; Houk, K. N. Quantum Mechanical Investigations of Organocatalysis: Mechanism, Reactivities and Selectivities. Chem. Rev. 2011, 111, 5042− 5137. (6) Wagner, J. P.; Schreiner, P. R. London Dispersion in Molecular Chemistry - Reconsidering Steric Effects. Angew. Chem., Int. Ed. 2015, 54, 12274−12296. (7) Donslund, B. S.; Johansen, T. K.; Poulsen, P. H.; Halskov, K. S.; Jørgensen, K. A. The Diarylprolinol-Silyl Ethers: 10 Years After. Angew. Chem., Int. Ed. 2015, 54, 13860−13874. (8) Marigo, M.; Wabnitz, T. C.; Fielenbach, D.; Jørgensen, K. A. Enantioselective Organocatalyzed α−Sulfenylation of Aldehydes. Angew. Chem., Int. Ed. 2005, 44, 794−797. (9) (a) Franzén, J.; Marigo, M.; Fielenbach, D.; Wabnitz, T. C.; Kjærsgaard, A.; Jørgensen, K. A. A General Organocatalyst for Direct α− Functionalization of Aldehydes: Stereoselective C−C, C−N, C−F, C− Br, and C−S Bond-Forming Reactions. Scope and Mechanistic Insight. J. Am. Chem. Soc. 2005, 127, 18296−18304. See also (b) Hayashi, Y.; Gotoh, H.; Hayashi, T.; Shoji, M. Diphenylprolinol Silyl Ethers as Efficient Organocatalysts for the Asymmetric Michael Reaction of Aldehydes and Nitroalkenes. Angew. Chem., Int. Ed. 2005, 44, 4212− 4215. (10) Dinér, P.; Kjærsgaard, A.; Lie, M. A.; Jørgensen, K. A. On the Origin of the Stereoselectivity in Organocatalyzed Reactions with Trimethylsilyl-Protected Diarylprolinol. Chem. - Eur. J. 2008, 14, 122− 127. (11) Seebach, D.; Groselj, U.; Badine, D. M.; Schweizer, W. D.; Beck, A. K. Isolation and X-Ray Structures of Reactive Intermediates of Organocatalysis with Diphenylprolinol Ethers and with Imidazolidinones. Helv. Chim. Acta 2008, 91, 1999−2034. (12) (a) Schmid, M. B.; Zeitler, K.; Gschwind, R. M. Formation and Stability of Prolinol and Prolinol Ether Enamines by NMR: Delicate Selectivity and Reactivity Balances and Parasitic Equilibria. J. Am. Chem. Soc. 2011, 133, 7065−7074. (b) Schmid, M. B.; Zeitler, K.; Gschwind, R. M. Distinct conformational preferences of prolinol and prolinol ether enamines in solution revealed by NMR. Chem. Sci. 2011, 2, 1793−1803. (13) Burés, J.; Armstrong, A.; Blackmond, D. G. Explaining Anomalies in Enamine Catalysis: “Downstream Species” as a New Paradigm for Stereocontrol. Acc. Chem. Res. 2016, 49, 214−222. (14) Hayashi, Y.; Gotoh, H.; Honma, M.; Sankar, K.; Kumar, I.; Ishikawa, H.; Konno, K.; Yui, H.; Tsuzuki, S.; Uchimaru, T. Organocatalytic, Enantioselective Intramolecular [6 + 2]-cycloaddition Reaction for the Formation of Tricyclopentanoids and Insights on Its Mechanism from a Computational Study. J. Am. Chem. Soc. 2011, 133, 20175−20185. (15) Halskov, K. S.; Kniep, F.; Lauridsen, V. H.; Iversen, E. H.; Donslund, B. S.; Jørgensen, K. A. Organocatalytic Enamine-Activation of Cyclopropanes for Highly Stereoselective Formation of Cyclobutanes. J. Am. Chem. Soc. 2015, 137, 1685−1691. (16) Seebach, D.; Gilmour, R.; Groselj, U.; Deniau, G.; Sparr, C.; Ebert, M.-O.; Beck, A. K.; McCusker, L. B.; Sisak, D.; Uchimaru, T. Stereochemical Models for Discussing Additions to α,β-Unsaturated Aldehydes Organocatalyzed by Diarylprolinol or Imidazolidinone Derivatives − Is There an ‘(E)/(Z)-Dilemma’? Helv. Chim. Acta 2010, 93, 603−634. (17) Groselj, U.; Seebach, D.; Badine, D. M.; Schweizer, W. D.; Beck, A. K.; et al. Structures of the Reactive Intermediates in Organocatalysis with Diarylprolinol Ethers. Helv. Chim. Acta 2009, 92, 1225−1259. (18) Maerten, E.; Cabrera, S.; Kjærsgaard, A.; Jørgensen, K. A. Organocatalytic Asymmetric Direct Phosponylation of α,β-Unsaturated Aldehydes: Mechanism, Scope, and Application in Synthesis. J. Org. Chem. 2007, 72, 8893−8903. (19) Dinér, P.; Nielsen, M.; Marigo, M.; Jørgensen, K. A. Enantioselective Organocatalytic Conjugate Addition of N-Heterocycles to α,β-Unsaturated Aldehydes. Angew. Chem., Int. Ed. 2007, 46, 1983−1987. (20) Bertelsen, S.; Marigo, M.; Brandes, S.; Dinér, P.; Jørgensen, K. A. Dienamine Catalysis: Organocatalytic Asymmetric γ-Amination of α,βUnsaturated Aldehydes. J. Am. Chem. Soc. 2006, 128, 12973−12980.
for obtaining insight into the reactivities and reaction courses. Furthermore, computational investigations have been valuable for describing intermediates and to account for observed regioand stereoselectivities in the obtained products. Whereas computational chemistry will without doubt continue to play an important role in rationalizing observed results thereby contributing to the understanding of the organocatalytic systems (as well as other fields), the increasingly advanced computational methods may allow for it to be applied more as a predictive tool in the future, by which it may further advance the development of new chemistry.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest. Biographies Kim Søholm Halskov completed his Ph.D. studies in 2015 from Aarhus University under the supervision of Karl Anker Jørgensen and currently holds a postdoctoral research position in the same group. His research focuses on the development of new methodologies in asymmetric organocatalysis. Bjarke S. Donslund obtained his bachelor degree in Medicinal Chemistry in 2011 and is now pursuing his Ph.D. studies under the guidance of Karl Anker Jørgensen. His research is mainly focused on aminocatalytic cycloaddition reactions. Bruno Matos Paz received his bachelor degree in 2013 from State University of Campinas, Brazil, and is now pursuing his Ph.D. studies under the supervision of Karl Anker Jørgensen. His research interests are focused on aminocatalytic asymmetric annulations and cascade reactions. Karl Anker Jørgensen received his Ph.D. from Aarhus University in 1984. He was a postdoctoral fellow with Roald Hoffmann, Cornell University, 1985. In 1985, he became Assistant Professor at Aarhus University, and in 1992 he was promoted to Professor. His research interests are the development, understanding, and application of asymmetric catalysis.
■
ACKNOWLEDGMENTS This work was supported by Aarhus University and Carlsberg Foundation. B. M. Paz thanks CAPES Foundation, Ministry of Education of Brazil, for fellowship (No. 9525-13-0). Thanks are expressed to a reviewer for comments about limitations in DFT calculations.
■
REFERENCES
(1) See, for example, (a) Dalko, P. I. Comprehensive Enantioselective Organocatalysis: Catalysts, Reactions, and Applications; Wiley-VCH: Weinheim, Germany, 2013. (b) List, B. Asymmetric Organocatalysis; Springer: Heidelberg, Germany, 2009. (2) Dondoni, A.; Massi, A. Asymmetric Organocatalysis: From Infancy to Adolescence. Angew. Chem., Int. Ed. 2008, 47, 4638−4660. (3) Bertelsen, S.; Jørgensen, K. A. Organocatalysis − After the Gold Rush. Chem. Soc. Rev. 2009, 38, 2178−2189. (4) For a review on aminocatalysis in cascade reactions, see (a) Grondal, C.; Jeanty, M.; Enders, D. Organocatalytic Cascade Reactions as a New Tool in Total Synthesis. Nat. Chem. 2010, 2, 167− 178. For a review on combined catalysis, see (b) Shao, Z.; Zhang, H. Combining Transition Metal Catalysis and Organocatalysis: a Broad New Concept for Catalysis. Chem. Soc. Rev. 2009, 38, 2745−2755. L
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX
Article
Accounts of Chemical Research (21) Albrecht, Ł.; Dickmeiss, G.; Acosta, F. C.; Rodríguez-Escrich, C.; Davis, R. L.; Jørgensen, K. A. Asymmetric Organocatalytic Formal [2 + 2]-cycloadditions via Bifunctional H-Bond Directing Dienamine Catalysis. J. Am. Chem. Soc. 2012, 134, 2543−2546. (22) Jia, Z.-J.; Jiang, H.; Li, J.-L.; Gschwend, B.; Li, Q.-Z.; Yin, X.; Grouleff, J.; Chen, Y.-C.; Jørgensen, K. A. Trienamines in Asymmetric Organocatalysis: Diels-Alder and Tandem Reactions. J. Am. Chem. Soc. 2011, 133, 5053−5061. (23) Jiang, H.; Cruz, D. C.; Li, Y.; Lauridsen, V. H.; Jørgensen, K. A. Asymmetric Organocatalytic Thio-Diels-Alder Reactions via Trienamine Catalysis. J. Am. Chem. Soc. 2013, 135, 5200−5207. (24) Jiang, H.; Rodríguez-Escrich, C.; Johansen, T. K.; Davis, R. L.; Jørgensen, K. A. Organocatalytic Activation of Polycyclic Aromatic Compounds for Asymmetric Diels-Alder Reactions. Angew. Chem., Int. Ed. 2012, 51, 10271−10274. (25) Halskov, K. S.; Johansen, T. K.; Davis, R. L.; Steurer, M.; Jensen, F.; Jørgensen, K. A. Cross-trienamines in Asymmetric Organocatalysis. J. Am. Chem. Soc. 2012, 134, 12943−12946. (26) Dieckmann, A.; Breugst, M.; Houk, K. N. Zwitterions and Unobserved Intermediates in Organocatalytic Diels-Alder Reactions of Linear and Cross-Conjugated Trienamines. J. Am. Chem. Soc. 2013, 135, 3237−3242. (27) Hayashi, Y.; Okamura, D.; Umemiya, S.; Uchimaru, T. Organocatalytic 1,4-Addition Reaction of α,β-γ,δ-Diunsaturated Aldehydes versus 1,6-Addition Reaction. ChemCatChem 2012, 4, 959−962. (28) Poulsen, P. H.; Feu, K. S.; Paz, B. M.; Jensen, F.; Jørgensen, K. A. Organocatalytic Asymmetric 1,6-Addition/1,4-Addition Sequence to 2,4-Dienals for the Synthesis of Chiral Chromans. Angew. Chem., Int. Ed. 2015, 54, 8203−8207.
M
DOI: 10.1021/acs.accounts.6b00008 Acc. Chem. Res. XXXX, XXX, XXX−XXX