Organocatalytic Asymmetric Addition of Aldehyde to Nitroolefin by H-d-Pro-Pro-Glu-NH2: A Mechanistic Study.

The mechanism of the asymmetric addition of aldehyde (butanal) to nitroolefin (β-nitrostyrene) catalyzed by H-d-Pro-Pro-Glu-NH2 (dPPE-NH2; 1) was explored using density functional theory methods in chloroform. By conformational search, it was confirmed that catalyst 1 and its enamine intermediate adopted a dominant conformation with a βI structure stabilized by a C10 H-bond between the C=O of d-Pro1 and C-terminal NH2 proton and by an additional H-bond between the side chain and the backbone of Glu3. This βI turn structure was conserved all along the catalytic cycle. Consistently with the kinetic studies, the C-C bond formation between the enamine and electrophile was also confirmed as the rate-determining step. The stereoselectivity results from a re → re prochiral approach of enamine and β-nitrostyrene with a gauche- orientation of the double bonds. Although it was suggested as the possible formation of dihydrooxazine oxide species, this process was confirmed to be kinetically less accessible than the formation of acyclic nitronate. In particular, our calculated results supported that the carboxylic acid group of Glu3 in 1 played a central role by acting as general acid/base all along the catalytic cycle and orienting the asymmetric C-C bond formation.

Introduction

The advent of organocatalysis has brought the prospect of a mode of catalysis, complementary to organometallic systems, with the potential for savings in cost, time, and energy; easier experimental procedures; and reductions in chemical waste.[1] Among environmentally friendly organocatalysts, peptides and peptide-based molecules have emerged as promising candidates (for reviews, see ref (2)). Both combinatorial and de novo methods of the catalyst design have been demonstrated successfully to perform a variety of asymmetric conjugate addition reactions including aldol, Mannich, Michael, and Morita–Baylis–Hillman reactions.[2a,2e,3] From a synthetic perspective, after the recognition of proline’s aptitude in formation of carbon–carbon bond in a stereoselective fashion,[4] remarkable efforts have been endorsed in the development of chiral-amine and peptide-based organocatalysts bearing a pyrrolidine moiety.[5] It is widely accepted that in such transformations an enamine intermediate is generated by condensing a carbonyl bearing reagent with the catalytic pyrrolidine. Reaction of the enamine then proceeds via addition on an electrophile partner, and the resulting iminium ion is finally hydrolyzed to afford the products. Another less explored possible mechanism involves enol as activated species for addition to the electrophile.[6]

In such a field, the organocatalytic Michael addition reaction of ketones or aldehydes to nitroolefins has recently attracted considerable attention because of the importance of resultant chiral nitroalkanes as synthetically valuable building blocks.[7] Considering activity and selectivity, Wennemers tripeptide H-d-Pro-Pro-Glu-NH2 (dPPE-NH2; peptide 1 in Scheme ) deserves special mention as a highly active and stereoselective catalyst for Michael conjugate addition of aldehydes to nitroethylene[8] and β-monosubstituted nitro-olefins in the protic CHCl3/iPrOH (9:1 v/v) environment[9] (Scheme ). It has noteworthy achieved the highest levels of stereocontrol under the lowest catalyst loading reported to date without significant catalyst deactivation. ESI–MS back-reaction screening experiments support an enamine mechanism because enamine and iminium have been detected as intermediates.[10] Considering such a pathway, the catalytic cycle should be divided into four main steps consisting of (i) enamine formation, (ii) reaction with the nitroolefin to yield a new C–C bond, (iii) protonation of the nitronate intermediate, and finally (iv) hydrolysis of the iminium. Kinetic studies provided insights that the reaction of the enamine with the electrophile is rate limiting and highlight a double role of the acidic group. It first orients the reactivity and stereoselectivity and secondly, it improves the reaction rate by promoting protonation of the iminium nitronate.[11,12] Although the acid group probably coordinates the nitronate function, its exact role as well as the protonation states along the catalytic pathway are not fully understood. Especially, its presence is not indispensable because a powerful catalytic tripeptide without the acid group has been reported for the addition of nitro-Michael between aldehydes and β,β-disubstituted nitro-olefins.[13]

Scheme 1

(a) H-d-Pro-Pro-Glu-NH2-Catalyzed Conjugate Addition Reaction of Butanal to β-Nitrostyrene; (b) Proposed Catalytic Cycle: (i) Enamine Formation, (ii) C–C Bond Formation, (iii) Protonation of the Nitronate and (iv) Hydrolysis[10]

Alternative mechanisms have been proposed for the catalysts which do not have acidic hydrogen. As an example, with chiral prolinol ether derivatives, cyclic intermediates such as dihydrooxazine and cyclobutane (1-D′ and 1-C′ in Chart a, respectively) were identified as resting states of catalysts,[14−16] and the rate-determining step (rds) of the reaction was suggested to be the protonation of the dihydrooxazine oxide species[16] instead of the iminium (refs (14a,15a)) or enamine nitronates (ref (15b)). An analogous cyclobutane intermediate has also been found in the reaction of butanal and nitrostyrene with dPPE-NH2 methyl ester.[11] Hence, even they are not populated to a significant extent, dihydrooxazine and cyclobutane intermediates could not be excluded in dPPE-NH2 as a catalyst.

Chart 1

(a) Cyclic Intermediates Dihydrooxazine (1-D′) and Cyclobutane (1-C′) for the Michael Addition of Aldehydes and Nitroalkenes Catalyzed by Diaryl Prolinol Ethers; (b) the Enamine Intermediate from Catalyst 1 by Reaction with Phenylacetaldehyde (ref (17))

Recently, Rigling et al. investigated the conformational preferences of dPPE-NH2 (1) and its enamine intermediate (1-En′ in Chart b) obtained by reaction with phenylacetaldehyde in a solution of CDCl3/CD3OH (9:1) using NMR spectroscopy and density functional theory (DFT) methods.[17] In dPPE-NH2, a strong nuclear Overhauser effect (NOE) was found between the amide NH of Glu3 and one of the C-terminal amide protons, which supports the formation of a β-turn conformation with a H-bond between the carbonyl group of d-Pro1 and the amide proton of the C-terminal group. By the lower vicinal coupling constants (close to 4 Hz) from the Hβ protons to Hα and Hγ1, it was suggested as distinct conformational preferences of the side chain torsion angles χ1 and χ2 of Glu3 at ∼60° (g+) and −60° (g–), respectively, which indicates a conformation with the side chain of Glu3 pointing toward d-Pro1. However, it was found that 1-En′ with the s-trans configuration still adopts a β-turn conformation being less populated compared to dPPE-NH2 by observing a stronger NOE between Hα of d-Pro1 and NH of Glu3 and by a significantly higher temperature dependence of the C-terminal CONH1 amide proton chemical shift in 1-En′ than in peptide 1. The higher flexibility of the Glu3 side chain and backbone in 1-En′ was suggested to be essential to stereoselectively accommodate the incoming nitroolefin for the C–C bond formation. However, the detailed pathways for the formation of iminium-nitronate, protonation, and hydrolysis were not reported.

The Michael addition of propanal to β-nitrostyrene catalyzed by diarylprolinol silyl ether has been the subject of computational studies using DFT methods to investigate the stereoselctivity of the C–C bond formation and mechanistic pathways.[16,18] In earlier studies, the lowest-energy transition state was obtained for the (R,S) pathway (∼23 kcal mol–1) at the B3LYP/6-31G(d) level of theory.[18] In a recent study, the free energy of the transition state for the (R,S) pathway was computed as 16.2 kcal mol–1 in chloroform by the sum of the free energy at the ωB97X-D/6-311++G(3df,3pd)//ωB97X-D/6-311G(d,p) level of theory and polarizable continuum model (PCM) solvation free energy at the ωB97X-D/6-311G(d,p) level of theory.[16b] In particular, the stereoselectivity was suggested to be primarily controlled by the C–C bond formation even though the reaction rate was dictated by the subsequent protonation step.[16b]

Here, we explored the plausible pathways of addition of aldehyde (butanal) to nitroolefin (β-nitrostyrene) catalyzed by dPPE-NH2 in chloroform by DFT calculations. Our results highlight the central role of the carboxylic acid group by acting as a general acid/base all along the catalytic cycle and orienting the asymmetric C–C bond formation.

Results and Discussion

Preferred Conformations of dPPE-NH2

For dPPE-NH2 (1), we identified 11 local minima with ΔE < 10 kcal mol–1 at the M06-2X/6-31G(d) level of theory in the gas phase. However, there were only four local minima with the relative Gibbs free energy (ΔG) < 10 kcal mol–1 at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform. The torsion angles and thermodynamic properties of these local minima in chloroform are listed in Table .

Table 1

Torsion Angles (°) and Thermodynamic Properties (kcal mol–1) of Local Minima (ΔG < 10 kcal mol–1) for the Catalyst dPPE-NH2a

conformer	ψ1	ψ2	ϕ3	ψ3	χ31	χ32	χ33	ΔEb	ΔHb	ΔGb	wc
1-dPPE-NH2	–157.9	–16.1	–63.0	–21.2	75.8	–64.6	152.4	0.0	0.0	0.0	100
2-dPPE-NH2	–147.4	61.5	–111.8	156.4	–52.8	78.8	45.5	5.5	5.6	4.1	0
3-dPPE-NH2	–141.2	58.8	–85.4	69.7	–54.0	89.0	–152.4	7.2	7.2	5.6	0
4-dPPE-NH2	–129.6	67.3	–144.2	–60.4	–8.7	70.9	–174.5	7.7	7.8	6.3	0

Torsion angles are for the backbone of dPPE-NH2. Calculated at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform.

ΔE, ΔH, and ΔG stand for relative electronic energy, enthalpy, and Gibbs free energy at 25 °C and 1 atm.

The population calculated by the ΔG values at 25 °C.

The most preferred conformation 1-dPPE-NH2 was dominantly populated in chloroform (ΔG = 0.0 kcal mol–1 with a population of 100%) and the next preferred conformers were 2-dPPE-NH2, 3-dPPE-NH2, and 4-dPPE-NH2 with ΔG = 4.1, 5.6, and 6.3 kcal mol–1, respectively (Table and Figure ). The most preferred conformer 1-dPPE-NH2 adopted a type I β-turn (βI) structure, which appeared to be stabilized by a C10 H-bond between the C=O of d-Pro1 and the proton of the C-terminal amide NH2 with a distance of d(C=O···H–NC-terminal) = 2.01 Å (Figure a). In addition, there were two additional H-bonds between the side chain Cδ=Oε1 and backbone NH proton of Glu3 with a distance of d(Cδ=OGlu3ε1···H–NGlu3) = 1.92 Å and between the side chain carboxylic proton of Glu3 and the amide nitrogen of d-Pro1 with a distance of d(Oε2–HGlu3···N) = 1.59 Å (Figure a). Although the second preferred conformer 2-dPPE-NH2 had two H-bonds between the C=O of d-Pro1 and the NH proton of Glu3 with a distance of d(C=O···H–NGlu3) = 2.03 Å and between the side chain carboxylic proton of Glu3 and the amide nitrogen of d-Pro1 with a distance of d(Oε2–HGlu3···N) = 1.66 Å (Figure b), 2-dPPE-NH2 was energetically and enthalpically less favored than 1-dPPE-NH2 (Table ). Although 3-dPPE-NH2 and 4-dPPE-NH2 had two H-bonds as found in 2-dPPE-NH2 and an additional H-bond of the C-terminal NH2 proton with the C=O of Pro2 and the side chain carboxylic Cδ=Oε1 of Glu3, respectively (Figure c,d), they are 1.6 and 2.1 kcal mol–1 higher in ΔE than 2-dPPE-NH2 (Table ).

Figure 1

Conformations of local minima for the catalyst dPPE-NH2 (1) with ΔG < 10 kcal mol–1 optimized at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform: (a) 1-dPPE-NH2 (ΔG = 0.00), (b) 2-dPPE-NH2 (ΔG = 4.10), (c) 3-dPPE-NH2 (ΔG = 5.57), and (d) 4-dPPE-NH2 (ΔG = 6.27). All ΔG are in kcal mol–1. H-Bonds are represented by dotted lines and in Å.

The backbone torsion angles of the most preferred conformation 1-dPPE-NH2 in chloroform were calculated as (ψ1, ψ2, ϕ3, ψ3) = (−158°, −16°, −63°, −21°) (Table ), which are consistent with the mean values of (−152 ± 1°, −36 ± 10°, −66 ± 11°, 2 ± 15°) obtained by simulated annealing calculations with the restraints of NOEs, residual dipolar couplings, and vicinal coupling constants from NMR experiments.[17] In addition, our calculated torsion angles (χ31, χ32) = (76°, −65°) of the side chain of Glu3 (Table ) are in accord with the conformational preference of the side chain torsion angles χ1 and χ2 of Glu3 at g+ and g–, respectively, deduced from vicinal coupling constants between protons at Cα, Cβ, and Cγ in Glu3.[17]

Preferred Conformations of the Enamine 1-En

For the enamine intermediate 1-En, we identified 40 local minima with ΔE < 10 kcal mol–1 at the M06-2X/6-31G(d) level of theory in the gas phase and 30 local minima with ΔG < 5 kcal mol–1 at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform. The torsion angles and thermodynamic properties of 12 local minima of 1-En with ΔG < 3 kcal mol–1 in chloroform are listed in Table . All 30 local minima with ΔG < 5 kcal mol–1 in chloroform are listed in Table S1 of the Supporting Information.

Table 2

Torsion Angles (°) and Thermodynamic Properties (kcal mol–1) of 12 Local Minima (ΔG < 3 kcal mol–1) for the Enamine Intermediatea

conformer	ψ1	ψ2	ϕ3	ψ3	χ31	χ32	χ33	ΔEb	ΔHb	ΔGb	wc
En-01	–167.4	–18.5	–78.2	–13.7	84.7	–66.0	174.8	0.0	0.0	0.0	45
En-02	40.1	–16.6	–82.6	70.2	–58.7	86.7	–178.9	3.3	2.7	0.6	17
En-03	–101.4	–11.8	–80.6	–9.1	–55.0	–169.3	–168.9	1.4	1.6	0.6	16
En-04	–150.4	150.2	–89.6	67.4	–74.9	55.6	–106.0	2.9	2.8	1.0	9
En-05	–153.9	66.3	–145.5	17.6	70.1	–65.9	125.7	1.9	1.7	1.2	6
En-06	34.7	146.1	–147.6	159.4	–57.8	–65.4	169.9	4.3	4.1	1.9	2
En-07	–162.4	70.5	–90.0	64.9	–48.5	–46.2	–56.2	3.3	3.1	2.0	2
En-08	–94.2	–10.2	79.4	–51.2	–54.5	–71.0	–172.7	3.5	3.6	2.2	1
En-09	–168.6	67.0	–86.0	71.3	–59.4	71.8	49.1	2.8	2.7	2.3	1
En-10	–123.1	153.1	–88.8	64.3	–76.4	53.4	–101.6	3.1	3.1	2.6	1
En-11	–106.3	142.5	–158.5	166.7	–99.7	–67.7	179.3	5.8	5.5	2.6	1
En-12	42.9	59.1	–93.3	63.1	–51.6	–48.0	–58.2	4.5	4.3	2.8	0

Torsion angles are for the backbone of dPPE-NH2. Calculated at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform.

ΔE, ΔH, and ΔG stand for relative electronic energy, enthalpy, and Gibbs free energy at 25 °C and 1 atm.

The population is calculated by the ΔG values at 25 °C.

The conformer En-01 of 1-En (Figure a) was the most preferred, as indicated by the ΔG values with a population of 45%; the next most preferred conformers were En-02, En-03, and En-04 (Figure b–d) [ΔG = 0.6, 0.6, and 1.0 kcal mol–1, respectively (Table ); populated at 17, 16, and 9%, respectively], confirming the higher flexibility of 1-En compared with dPPE-NH2. The most preferred conformer En-01 adopted a βI structure stabilized by a C10 H-bond between the C=O of d-Pro1 and the proton of the C-terminal amide NH2 with a distance of d(C=O···H–NC-terminal) = 2.02 Å (Figure a). In addition, there was an additional H-bond between the side chain Cδ=Oε1 and the backbone NH proton of Glu3 with a distance of d(Cδ=OGlu3ε1···H–NGlu3) = 1.92 Å (Figure a). Although a H-bond between the side chain carboxylic proton of Glu3 and the amide nitrogen of d-Pro1 was lost in En-01 because of the enamine formation, the overall conformation of En-01 with the torsion angles (ψ1, ψ2, ϕ3, ψ3, χ31, χ32, χ33) = (−167°, −19°, −78°, −14°, 85°, −66°, 175°) (Table ) was quite similar to the most preferred conformer 1-dPPE-NH2 of peptide 1 (Table ).

Figure 2

Conformations of local minima for the enamine intermediate (1-En) with ΔG < 1 kcal mol–1 optimized at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform: (a) En-01 (ΔG = 0.00), (b) En-02 (ΔG = 0.59), (c) En-03 (ΔG = 0.61), and (d) En-04 (ΔG = 0.97). All ΔG are in kcal mol–1. H-Bonds are represented by dotted lines and in Å.

The second most preferred conformer En-02 had two H-bonds between the C=O of Pro2 and the proton of the C-terminal amide NH2 with a distance of d(C=OPro2···H–NC-terminal) = 2.07 Å and between the side chain Cδ=Oε1 and the amide NH of Glu3 with a distance of d(Oε1···H–NGlu3) = 2.24 Å (Figure b). Conformer En-02 was 2.7 kcal mol–1 less favored in ΔH relative to the most preferred En-01, but the former had more conformational flexibility of −TΔS = −2.1 kcal mol–1 than the latter. The third most preferred conformer En-03 had only one H-bond between the C=O of d-Pro1 and the proton of the C-terminal amide NH2 with a distance of d(C=O···H–NC-terminal) = 1.99 Å (Figure c). Conformer En-03 was 1.6 kcal mol–1 less favored in ΔH relative to the most preferred En-01, but the former had more conformational flexibility of −TΔS = −1.0 kcal mol–1 than the latter. The fourth most preferred conformer En-04 had two H-bonds between the C=O of Pro2 and the proton of the C-terminal amide NH2 with a distance of d(C=OPro2···H–NC-terminal) = 2.17 Å and between the C=O of d-Pro1 and the side chain carboxylic proton of Glu3 and the C=O of d-Pro1 with a distance of d(Oε2–HGlu3···C=O) = 1.73 Å (Figure d). Conformer En-04 was 2.8 kcal mol–1 less favored in ΔH relative to the most preferred En-01, but the former had more conformational flexibility of −TΔS = −1.8 kcal mol–1 than the latter. The other conformers En-05 to En-12 exhibited values of ΔG = 1.2–2.8 kcal mol–1 and populations less than 6% in chloroform (Table ).

Our results are in agreement with the NMR data reported by Rigling et al. for the enamine intermediate named 1-En′ obtained by condensation of dPPE-NH2 (1) with phenylacetaldehyde.[17] Despite a predominant β-turn conformation, the enamine intermediates appeared more flexible than the parent catalyst dPPE-NH2. We also predicted that the puckerings of d-Pro1 and Pro2 in En-01 are both Cγ-endo, which are consistent with NMR data for 1-En′.[17] The main difference between En-01 and 1-En′ relates to the conformation of the Glu3 side chain. While (χ31, χ32) = (−71°, 78°) for 1-En′,[17] it was (χ31, χ32) = (85°, −66°) for En-01, which is similar to the values calculated for 1-dPPE-NH2. The assignment of the g+g– side chain conformation of Glu3 to 1-En′ probably resulted in a clash between the carboxylic group of Glu3 and the C=C bond that did not occur with the butanal reactant.

Stereoselective C–C Bond Formation

We then studied the stereoselective C–C bond formation resulting from the addition of En-01 of 1-En to β-nitrostyrene. Depending on the relative orientations of the prochiral centers,[18a,19] four different enamine → nitrostyrene facial approaches were considered. The re → re, re → si, si → re, and si → si approaches lead, respectively, to the four possible diastereomers (2S,3R), (2S,3S), (2R,3R), and (2R,3S) for the nitronate 1-Nit (Scheme ). For each enantiomer, three different orientations [i.e., gauche– (g–), gauche+ (g+), and trans (t)] of the double bonds of enamine and nitrostyrene were considered.

Scheme 2

Four Different Approaches of the Prochiral Centers of Enamine 1-En and β-Nitrostyrene to Form the C–C Bond

The re → re, re → si, si → re, and si → si approaches resulted the (2S,3R), (2S,3S), (2R,3R), and (2R,3S) enantiomeric products of nitronate 1-Nit (Scheme b), respectively. For each enantiomer, three different orientations [i.e., gauche– (g–), gauche+ (g+), and trans (t)] were considered for the double bonds of enamine and nitrostyrene around the prochiral centers.

The procedure to construct the 12 initial 1-Nit structures (only four enantiomers) and the corresponding transition states are described in the Supporting Information. The torsion angles and relative thermodynamic properties of enantiomeric transition states and products for the stereoselective C–C bond formation in chloroform are listed in Table S2 of the Supporting Information. The free-energy profiles for 12 stereoselective C–C bond formations between En-01 and β-nitrostyrene in chloroform at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory are depicted in Figure .

Figure 3

Free-energy profiles for the 12 feasible stereoselective C–C bond formations between En-01 and β-nitrostyrene depending on the relative orientations of the prochiral centers. All free energies are relative to the reaction complex (RC) + β-styrene system.

The torsion angles of the backbone of dPPE-NH2 moiety in all transition states and products were quite similar to each other. The five lowest energy barriers were obtained for the re → re/t, re → si/g–, re → si/t, si → re/t, and re → re/g– prochiral approaches. Although the barriers of the four first prochiral approaches were 1.6, 2.3, 1.0, and 2.2 kcal mol–1 lower in free energy than that of the re → re/g– one, the corresponding products were less preferred than the (2S,3R)-1-Nit (ΔΔG = 12.4, 4.9, 3.8, and 1.8 kcal mol–1, respectively). The nitronate intermediates resulting from re → re/g+ and si → si/g+ approaches were lower in free energy (ΔΔG = −1.4 and −1.8 kcal mol–1); however, the energy barriers were found to be 1.4 kcal mol–1 higher than that of the re → re/g– prochiral approach. Hence, the re → re/g– prochiral approach between enamine (1-En) and β-nitrostyrene was kinetically and/or thermodynamically favored and appeared to be the major route for the C–C bond formation, which produced the major (2S,3R)-1-Nit intermediate.

The three re → si/g–, re → si/t, and si → re/t prochiral approaches have the lower ΔG⧧ barriers for the C–C bond formation than the re → re/g– approach, and the si → re/g+ approach has comparable ΔG⧧ barrier to the re → re/g– approach. In addition, these four prochiral approaches resulted in the somewhat higher ΔG values for nitronate intermediate (1-Nit) than the re → re/g– approach. Hence, the backward process (i.e., the C–C bond dissociation) from 1-Nit to intermediate I5 is kinetically feasible for these four prochiral approaches as well. The ΔG⧧ barriers for backward process were calculated as 12.1, 14.5, 17.7, and 15.3 kcal mol–1 for the re → si/g–, re → si/t, si → re/g+, and si → re/t prochiral approaches, respectively, which are 7.2, 4.8, 1.6, and 4.0 kcal mol–1 lower than that of the re → re/g– approach, respectively.

In addition, the possibility of further protonation of nitronate intermediates produced by these four prochiral approaches was investigated. The torsion angles and thermodynamic properties for the protonation in chloroform at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory are listed in Table S7 of the Supporting Information. The free-energy profiles for the protonation are depicted in Figure . The ΔG⧧ barriers for protonation from nitronate intermediate were calculated as 17.5, 17.0, 19.1, and 21.3 kcal mol–1 for the re → si/g–, re → si/t, si → re/g+, and si → re/t prochiral approaches, respectively, which are 5.4, 2.5, 1.4, and 6.0 kcal mol–1 higher than those of the corresponding backward processes, respectively. Hence, the C–C bond formations via the four re → si/g–, re → si/t, si → re/g+, and si → re/t prochiral approaches are kinetically less feasible because of the lower barriers for the backward process and the higher barriers for protonation.

Figure 4

Free-energy profiles for the C–C bond dissociation (via ts4) and protonation (via ts5) of nitronate intermediates (1-Nit) produced by the four re → si/g–, re → si/t, si → re/g+, and si → re/t prochiral approaches. All free energies are relative to the RC + β-styrene system.

The optimized structures of transition states and 1-Nit for re → re/g– are depicted in Figure . Interestingly, the re → re/g+, re → si/g+, and si → si/g– prochiral approaches produced dihydrooxazines as products (i.e., 1-D in Scheme b), but the barriers ΔG⧧ were 1.4, 3.1, and 5.9 kcal mol–1 higher than that of the re → re/g– prochiral approach and those appeared to be kinetically less favored. In the Michael addition of propanal to β-nitrostyrene catalyzed by diarylprolinol silyl ether, the ΔG⧧ barrier of the most preferred transition state for the C–C bond formation via the (R,S) pathway in chloroform was recently calculated as 16.2 kcal mol–1 at the ωB97X-D/6-311++G(3df,3pd)//ωB97X-D/6-311G(d,p) level of theory and the PCM solvation free energy at the ωB97X-D/6-311G(d,p) level of theory,[16b] which is 5.7 kcal mol–1 lower than that of the re → re/g+ prochiral approaches produced dihydrooxazines as products (i.e., 1-D in Scheme b) in the present work.

Figure 5

Conformations of transition states and intermediates for the enamine (re) → β-nitrostyrene (re) prochiral approaches to form the C–C bond between En-01 and β-nitrostyrene with (a) gauche– and (b) gauche+ orientations of the double bonds in chloroform. The distances d(C···C) in transition states are represented by dotted lines and in Å. H-Bonds were also represented by dotted lines.

In addition, we studied the stereoselective C–C bond formation of the next most preferred conformers [i.e., En-02, En-03, and En-04 shown in Figure ; populated at 17, 16, and 9%, respectively]. Only the re → re prochiral approach in the g– orientation of the double bonds was considered for these addition reactions, which resulted the major (2S,3R)-1-Nit species. The procedure to construct these initial 1-Nit structures and the corresponding transition states are described in the Supporting Information. The torsion angles and relative thermodynamic properties of transition states and products for these stereoselective C–C bond formation in chloroform at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory are listed in Table S3 of the Supporting Information.

The free-energy profiles for the C–C bond formation of En-02, En-03, and En-04 with β-nitrostyrene via the re → re/g– prochiral approach in chloroform at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory are depicted in Figure . The torsion angles of the backbone of dPPE-NH2 moiety in the transition state and 1-Nit of each enamine species were similar to the corresponding values of En-02, En-03, and En-04, although there were somewhat large changes of +46°, +36°, and +106° in ϕ3, ψ3, and χ32 of En-03, respectively (Table S3 of the Supporting Information). The ΔG⧧ barriers of the re → re/g– prochiral approaches for En-02, En-03, and En-04 with β-nitrostyrene were 9.9, 9.1, and 6.1 kcal mol–1 higher than that of En-01, and the corresponding products were much less preferred than En-01 (ΔΔG = 19.4, 18.4, and 15.6 kcal mol–1, respectively). Hence, the re → re/g– prochiral approach between En-01 and β-nitrostyrene was kinetically and thermodynamically favored and appeared to be the major route for the C–C bond formation, which produced the major (2S,3R)-1-Nit intermediate.

Figure 6

Free-energy profiles for the C–C bond formation of En-02, En-03, and En-04 with β-nitrostyrene via the re → re/g– prochiral approach in chloroform. All free energies are relative to the RC + β-styrene system.

Exploration of the Nitronate or Dihydrooxazine Pathways

Whatever the pathway considered, i.e., 1-Nit (re → re/g–) or 1-D (re → re/g+) pathway, the next step of the catalytic cycle requires a water molecule that participates in the protonation of the species. In the 1-D hypothesis, conversion of dihydrooxazine into a cyclobutane adduct should also be considered because this species has been experimentally observed with diphenylprolinol trimethylsilyl ether as catalyst.[16]

Considering the 1-Nit pathway, the initial ts structure was constructed by introducing a water molecule into the (2S,3R)-1-Nit model obtained from the most preferred re → re/g– prochiral approach and was optimized at the SMD M06-2X/6-31+G(d) level of theory in chloroform. The same procedure was used to construct the ts structure for the protonation step of the (2S,3R)-1-D dihydrooxazine intermediate. The free energy of each intermediate and transition state in chloroform was calculated at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory.

The six-membered dihydrooxazine ring in the hydrated (2S,3R)-1-D model adopted a boat conformation. However, the structure of (2S,3R)-1-D with a half-chair conformation was also considered because such a conformation was proposed in the nitro-Michael addition catalyzed by diphenylprolinol trimethylsilyl ether.[16a] Nevertheless, the later was 3.1 kcal mol–1 less stable in chloroform than the corresponding structure in boat conformation. The ΔG⧧ barriers for the protonation step of (2S,3R)-1-Nit and (2S,3R)-1-D in boat conformation were calculated to be 12.4 and 14.3 kcal mol–1 (see Figure a). Again, this confirms the re → re/g– prochiral approach as the most preferred pathway.

Figure 7

Free-energy profiles for (a) the protonation of dihydrooxazine oxide species (2S,3R)-1-D and (b) its transition to the cyclobutane adduct (2S,3R)-1-C in chloroform (see Scheme b). H-Bonds are represented by dotted lines. All free energies are relative to the RC + β-styrene system.

Next, we explored the transition of the hydrated dihydrooxazine species (1-D) to the cyclobutane adduct (1-C). The structure of the hydrated (2S,3R)-1-C was constructed from (2S,3R)-1-D and then optimized at the SMD M06-2X/6-31+G(d) level of theory in chloroform. The initial transition state was obtained by the relaxed scan of (2S,3R)-1-D at the SMD M06-2X/6-31+G(d) level of theory in chloroform. The distance between the C3 and C6 atoms of the oxazine decreased from 2.851 to 1.574 Å, whereas the length of intracyclic C6–O1 bond increased from 1.512 to 3.196 Å over 10 steps. The maximum energy was found at the eighth step, and then, the ts structure was reoptimized at the same level of theory in chloroform.

The free-energy profile for the protonation of dihydrooxazine oxide species (2S,3R)-1-D and its transition to the cyclobutane adduct (2S,3R)-1-C in chloroform is depicted in Figure . The ΔG⧧ barrier associated with the transition of (2S,3R)-1-D to (2S,3R)-1-C in chloroform was calculated as 11.3 kcal mol–1, which is 3.0 kcal mol–1 lower than that found for the protonation of 1-D (Figure b). In addition, the structure of (2S,3R)-1-C was 5.5 kcal mol–1 more stable than (2S,3R)-1-D, indicating that, when formed, the dihydrooxazine oxide species (1-D) should readily convert to cyclobutane adduct. Because 1-C was never observed experimentally with dPPE-NH2 as catalyst, the 1-D pathway involving the re → re/g+ prochiral approach should be excluded.

In the Michael addition of propanal to β-nitrostyrene catalyzed by diarylprolinol silyl ether, the ΔG⧧ barriers for the transition of dihydrooxazine oxide species D into the cyclobutane adduct C and for the C–C bond formation and protonation of dihydrooxazine oxide species D in the presence of p-nitrophenol were calculated as 11.9, 15.9, and 18.1 kcal mol–1, respectively, at the ωB97X-D/6-311++G(3df,3pd)//ωB97X-D/6-311G(d,p) level of theory and the PCM solvation free energy at the ωB97X-D/6-311G(d,p) level of theory.[16b] The ΔG⧧ barrier for the D → C transition is quite similar to 11.3 kcal mol–1 for the transition of (2S,3R)-1-D to (2S,3R)-1-C in chloroform in the present work. However, the ΔG⧧ barriers for the C–C bond formation and protonation of dihydrooxazine oxide species D were 6.0 kcal mol–1 lower and 3.8 kcal mol–1 higher, respectively, than those of the re → re/g+ prochiral approaches produced dihydrooxazines as products (i.e., 1-D in Scheme b) in the present work.

Plausible Organocatalytic Cycle by dPPE-NH2

The procedure to construct the initial structures for the plausible organocatalytic cycle catalyzed by dPPE-NH2 was described in the Computational Methods. The plausible pathway for the addition of butanal to β-nitrostyrene catalyzed by dPPE-NH2 (1) obtained at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory is depicted in Figure (reactant complex, transition state, intermediate, and product complex are abbreviated as RC, ts, I, and PC, respectively). Their corresponding thermodynamic properties are shown in Table . All the ΔG values were calculated with respect to the total Gibbs free energy of the reactant complex system formed by dPPE-NH2 and β-nitrostyrene. The torsion angles of dPPE-NH2 for reactant, transition states, intermediates, and product for the organocatalytic addition of butanol to β-nitrostyrene by dPPE-NH2 are listed in Table S7 of the Supporting Information. The free-energy profile for the organocatalytic cycle is depicted in Figure . The essential reaction species such as ts3 for the enamine formation, I4 (the enamine intermediate En), ts5 for the protonation of nitronate, ts7 for hydrolysis, and the final PC are shown in Figure . All other structures of RC, transition state, intermediate, and PC optimized in chloroform are depicted in Figure S1 of the Supporting Information.

Figure 8

Plausible pathways for the H-d-Pro-Pro-Glu-NH2-catalyzed conjugate addition reaction of butanal to β-nitrostyrene into (2S,3R)-2-ethyl-4-nitro-3-phenylbutanal in chloroform. The C–C bond formation was confirmed as the rds. The corresponding thermodynamic properties are listed in Table . H-Bonds are represented by dotted lines. All optimized structures are shown in Figure S1 of the Supporting Information.

Table 3

Relative Thermodynamic Properties (kcal mol–1) of Reactant, Transition States, Intermediates, and Products for the Addition of Butanal to β-Nitrostyrene Catalyzed by dPPE-NH2 (1) in Chloroforma

chemical species	ΔEb	ΔHb	ΔGb
RC + nitrostyrene	0.0	0.0	0.0
ts1 + nitrostyrene	2.8	2.7	6.2
I1 + nitrostyrene	–9.4	–7.7	–3.2
I2 + nitrostyrene	–5.1	–4.2	–0.9
ts2 + nitrostyrene	6.0	4.9	8.5
I3 + nitrostyrene	–2.2	–2.4	–0.9
ts3 + nitrostyrene	19.3	14.6	17.5
I4 (En) + nitrostyrene	1.4	1.3	2.4
I5 + water	2.0	0.7	5.2
ts4 (C–C bond formation) + water	13.5	11.9	20.5
I6 + water	–8.3	–8.6	1.2
I7	–12.9	–11.1	7.2
ts5 (protonation)	2.4	0.0	19.6
I8	–24.6	–21.5	–3.5
I9	–21.8	–18.6	–0.9
ts6 (carbinol amine)	–16.4	–14.3	5.8
I10	–23.0	–19.2	0.6
I11	–19.2	–14.0	7.3
ts7 (hydrolysis)	–14.2	–11.3	9.5
PC	–21.3	–18.2	–1.2

Calculated at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory.

ΔE, ΔH, and ΔG stand for relative electronic energy, enthalpy, and Gibbs free energy at 25 °C and 1 atm.

Figure 9

Free-energy profiles for the conjugate addition reaction of butanal and β-nitrostyrene catalyzed by H-d-Pro-Pro-Glu-NH2 in chloroform: (a) the enamine formation and (b) C–C bond formation, protonation, and hydrolysis. The relative Gibbs free energies to the RC + β-nitrostyrene system were calculated at the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory in chloroform.

Figure 10

Essential reaction species for addition of butanal to β-nitrostyrene catalyzed by H-d-Pro-Pro-Glu-NH2 in chloroform: (a) ts3 for the enamine formation, (b) I4 (the enamine intermediate En), (c) ts5 for the protonation of nitronate, (d) ts7 for hydrolysis, and (e) the final PC. H-Bonds are represented by dotted lines and in Å.

All intermediates and transition states of the catalytic cycle adopted a βI structure similar to that found for peptide 1 (see 1-dPPE-NH2 in Figure a). The structures are stabilized by a C10 H-bond between the C=O of d-Pro1 and the C-terminal amide NH2 [distance d(C=O···H–NC-terminal) = 1.97–2.09 Å]. A second H-bond involving the Cδ=Oε1 and NH proton of Glu3 was also conserved all along the catalytic pathway [d(Cδ=OGlu3ε1···H–NGlu3) = 1.75–1.95 Å].

In the first step of the catalytic cycle, peptide 1 and butanal [d(Cδ=OGlu3ε1···H–Obutanal) = 1.64 Å] proceed to a first carbinol amine intermediate I1 via the transition state ts1 [d(N···C=Obutanal) = 2.29 Å and d(C=Obutanal···H–OGlu3ε2) = 1.60 Å] with a barrier ΔG⧧ = 6.2 kcal mol–1. The carbinol amine I2 results from the prototropy of the d-Pro1 NH to the carboxylate of Glu3. Dehydration of I2 under the control of Glu COOH leads to the iminium species I3 via the transition state ts2 (ΔG⧧ = 8.5 kcal mol–1). The water molecule abstracts a proton from the CH2β of iminium I3 in the transition state ts3 (Figure a) with ΔG⧧ = 17.5 kcal mol–1, which yields the enamine intermediate I4 (Figure b). In particular, I4 adopts a structure identical to the most preferred conformer En-01 (Figure a), except that a water molecule was bound to the side chain carboxylic H–Oε2 of Glu3 in I4 [d(Owater···H–OGlu3ε2) = 1.71 Å].

In the second step of the catalytic cycle, the re → re/g– prochiral approach of β-nitrostyrene is assisted by the Glu COOH that binds the nitro group. The bimolecular intermediate I5 proceeds to the nitronic acid I6, in which the carboxylic hydrogen is transferred to the nitronate group [d(NO-HNO2···OGlu3ε2–) = 1.55 Å]. The transition state ts4 described in Figure a has the highest barrier of the catalytic cycle (ΔG⧧ = 20.5 kcal mol–1) and the stereoselective (2S)-C–(3R)-C bond formation is thermodynamically and/or thermodynamically controlled.

The third step of the catalytic cycle consists in the protonation of the nitronic acid. A second bimolecular species I7 is formed by introduction of a water molecule that is doubly coordinated to the β-nitronate group and the carboxylic group of Glu3. In the transition state ts5 (Figure c), the nitronate picks up a proton from the water molecule with the assistance of the Glu COOH group to form I8. The protonation step is recognized as the second highest barrier of the reaction pathway (ΔG⧧ = 19.6 kcal mol–1).

The final step consists of the iminium hydrolysis. The water molecule shifts in close proximity of the iminium moiety. Intermediate I9 then proceeds to the carbinol amine I10 via the transition state ts6 with a relatively low energy barrier (ΔG⧧ = 5.8 kcal mol–1). In ts6, the water molecule bridges both the iminium Cα and the carboxylate group. Prototropy of acid proton, that is initially H-bonded to the nitrate group in I10, to the nitrogen atom of d-Pro1 leads to I11. I11 is stabilized by a new H-bond between the OH group of carbinol amine and the carboxylate of Glu3 in I11. Finally, the abstraction of the proton of the hydroxyl group by the Glu3 carboxylate as highlighted in Figure d [ts7: d(N···C=Oproduct) = 2.09 Å and d(C=Oproduct···H–OGlu3ε2) = 1.58 Å] facilitates the breaking of the C–N bond and the recycling of the catalyst. This elementary step requires a relatively low ΔG⧧ = 9.5 kcal mol–1. The final product complex, in which the product and catalyst are H-bonded [d(C=Oproduct···H–OGlu3ε2) = 1.67 Å], is 1.2 kcal mol–1 more stable in ΔG than the RC + β-nitrostyrene system.

According to the calculated ΔG⧧ values with respect to the total Gibbs free energy of the RC + β-nitrostyrene system, ts4 for the C–C bond formation and ts5 for the protonation of the nitronic acid intermediate exhibited higher barriers of ΔG⧧ = 20.5 and 19.6 kcal mol–1, respectively. Hence, it confirms that the C–C bond formation between the enamine and the electrophile is the rds. However, the protonation of nitronic acid requires energy comparable with the C–C bond formation. These calculated results are consistent with the kinetic studies, which provided insights that not the enamine formation but the reaction of the enamine with the electrophile is rate limiting in case of catalysts bearing an acid group in a position that allows for coordination of the nitrate.[9b,9c,11]

Conclusions

By conformational search, it was confirmed that the catalyst H-d-Pro-Pro-Glu-NH2 (dPPE-NH2; 1) and its enamine intermediate (1-En) adopted a dominant conformation with a βI structure stabilized by a C10 H-bond between the C=O of d-Pro1 and the C-terminal NH2 proton and with an additional H-bonds between the side chain and the backbone of Glu3. The βI structure is conserved along the catalytic cycle.

The stereoselective (2S)-C–(3R)-C bond formation between enamine (1-En) and β-nitrostyrene proceeds via the re → re prochiral approach with the gauche– orientation of the double bonds of reactants. Although it was suggested the possible formation of dihydrooxazine oxide species (1-D), this process was confirmed to be kinetically less accessible than the acyclic nitronate pathway.

By exploring the pathways for addition of butanal to β-nitrostyrene catalyzed by catalyst dPPE-NH2 using DFT methods in chloroform, the C–C bond formation between the enamine and the electrophile (β-nitrostyrene) was confirmed as the rds. However, the protonation by a water molecule to nitronic acid requires energy comparable to the C–C bond formation. These calculated results are consistent with the kinetic studies. In particular, our calculated results supported that the central role of the carboxylic acid group of Glu3 in dPPE-NH2 by acting as general acid/base all along the catalytic cycle and orienting the asymmetric C–C bond formation.

Computational Methods

DFT Calculations

All Hartree–Fock (HF) and density functional calculations were performed using the Gaussian 09 programs.[20] All density functional calculations of optimizations and vibrational frequencies were carried out using the M06-2X functional[21] and the solvation model based on density (SMD) method.[22] GaussView[23] was used to generate and edit the structures of all intermediates and transition states. The M06-2X is a hybrid-meta-GGA functional with the improved medium-range correlation energy and showed good performance in predicting noncovalent interactions of small molecules and structures and relative stabilities of biological compounds such as peptides.[24]

For all reactants, intermediates, transition states, and products optimized at the SMD M06-2X/6-31+G(d) level of theory in chloroform, the relative energy (ΔE) of each local minimum in chloroform was calculated as the sum of the relative single-point energy (ΔE0) using the ωB97X-D functional[25] with the 6-311++G(d,p) basis set and the relative solvation free energy (ΔΔGs) at the SMD M06-2X/6-31+G(d) level of theory in chloroform. For all local minima and transition states, vibrational frequencies were calculated at the SMD M06-2X/6-31+G(d) level of theory in chloroform at 25 °C and 1 atm to confirm the nature of the stationary points and to obtain relative enthalpies and Gibbs free energies of intermediates and transition states at the same level of theory. The scale factor used was 0.9440 at the SMD M06-2X/6-31+G(d) level of theory; this value was chosen to reproduce the experimental frequency for the amide I band of N-methylacetamide in Ar and N2 matrices.[26] The intrinsic reaction coordinate method[27] was used to confirm each transition state whether it connects the corresponding reactants and products. In addition, each transition state was confirmed by checking the motions of normal modes using GaussView,[23] which were involved in the formation and dissociation of the bonds in the corresponding reactant and product. Notably, consistent with CD and NMR experiments, the ωB97X-D/6-311++G(d,p)//SMD M06-2X/6-31+G(d) level of theory correctly predicted the conformational populations of the backbone and prolyl peptide bond for the Pro dipeptide in solution.[28] In addition, the ωB97X-D functional exhibited better performance than double-hybrid functionals DSD-BLYP, B2GPPLYPD, and B2PLYP with dispersion corrections with respect to the benchmark CCSD(T)/CBS-limit energies of Diels–Alder reactions (the DARC set[29]).[30]

Conformational Search

We investigated the conformational preferences of dPPE-NH2 and its enamine intermediate (1 and 1-En in Scheme , respectively) obtained by reaction with butanal. The initial structure of peptide 1 was constructed using the X-ray structure of H-d-Pro-Pro-Asp-NH2[9b] and optimized at the M06-2X/6-31+G(d) level of theory. Then, the initial structure of the enamine 1-En in the s-trans configuration was constructed from the optimized structure of peptide 1 and optimized at the same level of theory. Using these optimized structures, 157 and 189 initial structures for 1 and 1-En, respectively, were generated by the systematic search of the Discovery Studio package[31] using the CHARMm force field with the maximum systematic conformations = 1000 and the energy threshold = 10 kcal mol–1. In the conformational search of 1 and 1-En, a systematic variation of each of the torsion angles ψ1, ψ2, ϕ3, and ψ3 of the backbone and χ31, χ32, and χ33 of the side chain of Glu3 was done using steps of 60°. These initial structures were then optimized at the HF/3-21G(d) level of theory, and we obtained 21 and 57 local minima with the relative energy ΔE < 10 kcal mol–1, respectively, which were reoptimized at the M06-2X/6-31G(d) level of theory and further optimized at the SMD M06-2X/6-31+G(d) level of theory in chloroform.

Construction of Organocatalytic Cycle

The most preferred conformations 1-dPPE-NH2 of peptide 1 (Table and Figure a) and En-01 of 1-En (Table and Figure a) obtained by the conformational search were used to construct the initial structures in the organocatalytic cycle. In particular, the C–C bond formation between enamine (1-En) and β-nitrostyrene was considered by the re → re/g– prochiral approach, which was kinetically and/or thermodynamically favored, as described earlier. In addition, the initial structures of intermediates and transition states for addition of butanal to β-nitrostyrene catalyzed by 1 were constructed using the structures of intermediates and transition states suggested by Clemente and Houk[32] for the enamine mechanism of intramolecular aldol reactions catalyzed by proline as a reference, which involves an enamine intermediate with concerted C–C bond formation and proton transfer from the carboxylic acid group to the carbonyl acceptor. The overall reaction mechanism of intramolecular aldol reactions catalyzed by proline appears to be quite similar to the addition reaction of butanal to β-nitrostyrene catalyzed by dPPE-NH2, except for the protonation step being not necessary for the intramolecular aldol reactions.