Origin of High Diastereoselectivity in Reactions of Seven-Membered-Ring Enolates

Abstract

Unlike many reactions of their six-membered-ring counterparts, the reactions of chiral seven-membered-ring enolates are highly diastereoselective. Diastereoselectivity was observed for a range of substrates, including lactam, lactone, and cyclic ketone derivatives. The stereoselectivity arises from torsional and steric interactions that develop when electrophiles approach the diastereotopic π-faces of the enolates, which are distinguished by subtle differences in the orientation of nearby atoms of the ring.

Graphical Abstract

The diastereoselective reactions of chiral seven-membered-ring enolates have been systematically investigated. Diastereoselectivity was found to be general for ε-lactams, ε-lactones, and cycloheptanones with various substituent patterns. Computational investigations provided insight into the origin of diastereoselectivity. A model for predicting the stereochemistry of seven-membered-ring enolate alkylations is proposed.

The stereoselective synthesis of cyclic carbonyl compounds is often achieved using reactions of enolates to form carbon–carbon and carbon–heteroatom bonds.^[1] Reactions of chiral endocyclic enolates can be useful in the cases of four-^[2] and five-membered^[3]-rings and for macrocyclic systems (eight-membered and larger rings^[4]), but predicting the stereochemical outcomes of reactions of six-membered and seven-membered-rings is more challenging. Reactions of six-membered-ring enolates are often unselective,^[5] which is believed to be the result of particularly early transition states that do not allow for differentiation of the diastereofaces.^[6] The reactions of seven-membered-ring enolates, which are complicated by the many conformations available to seven-membered-rings,^[7] have received less attention,^[7b] although the few examples reported were diastereoselective.^[8] Considering that seven-membered-ring enolates could be useful for the syntheses of biologically active natural products and synthetic drugs,^[9] it would be valuable to know whether reactions of seven-membered-ring enolates are generally stereoselective.

In this Communication, we demonstrate that the enolates of many substituted seven-membered-ring compounds undergo reactions with high diastereoselectivity. Experimental results and computational investigations show that stereoselectivity can be explained by considering the different steric and torsional interactions that arise when electrophiles approach the diastereotopic π-faces of these enolates. These data suggest that an early, enolate-like transition state is not responsible for the lack of stereoselectivity in six-membered-ring enolates.

Initial observations focused on the alkylations of seven-membered ε-lactam substrates. These reactions represent model systems for the synthesis of azepane-containing alkaloids.^[9e,10] Treatment of C4-substituted lactam 1 with lithium diisopropylamide followed by alkylation with allyl bromide furnished product cis-2a as a 96:4 mixture of diastereomers (eq 1). The use of THF as the solvent proved to be optimal with respect to the yield of the reaction. The use of other solvents (toluene, diethyl ether, n-hexane, and methyl tert-butyl ether) gave similar stereoselectivities but poorer conversion to product. Analysis by X-ray crystallography^[11] revealed that the electrophile was installed on the same face as the tert-butyl substituent.^[12] This observation indicates that the remote substituent does not interfere directly with the approach of the electrophile to the intermediate. Instead, inherent torsional changes must occur in the transition state to control stereoselectivity. By contrast, alkylation of the enolate of the six-membered-ring lactam 3 was not stereoselective (eq 2), just as these reactions are not generally stereoselective for other types of six-membered-ring enolates.^[5] These different outcomes indicate that, in contrast to the accepted explanation for lack of selectivity in six-membered-ring alkylations,^[6] it is not likely that the position along the reaction coordinate determines the diastereoselectivity of reactions of cyclic enolates.

(1)

(2)

Systematic exploration on the scope of diastereoselectivity in seven-membered-ring alkylations began with exploring the reactions of C3-substituted ε-lactams. Alkylation of C3-substituted lactam 5 with allyl bromide furnished product trans-6 as a single diastereomer (eq 3). Lower diastereoselectivity was observed in the formation of trans-8a (eq 3). The observed preference for anti-alkylation is consistent with what was observed with other endocyclic enolates that bear β-alkyl substituents, regardless of ring size.^{[2a,3c,3d,8b,13]} The favored conformer of the enolate 9 would position the substituent pseudoequatorially, leaving the peripheral face exposed (eq 4). The dependence of stereoselectivity on the size of the substituent suggests that attack from the opposite face is disfavored by a developing steric interaction between the substituent and the incoming electrophile.

(3)

(4)

The importance of the C3 substituent is illustrated by the reactions of seven-membered-ring lactam 10 (eq 5) and lactones 12 and 14 (eq 6). The propensity for alkylation to be controlled by the substituent at C3 appears to supersede any influence on stereoselectivity provided by a remote substituent. This hypothesis is consistent with what has been reported in the alkylations of more highly substituted seven-membered-ring enolates.^{[8b,13g,13h,14]}

(5)

(6)

Alkylations of C4-substituted ε-lactams were generally diastereoselective for various substrates (eq 7). The magnitude of stereoselectivity depended on a number of factors, however. Stereoselectivity decreased as the size of the substituent at C4 decreased across the series t-Bu > Ph > Me (Table 1, cis-2a > cis-17a > cis-19a).^[12] The size of the substituent likely determines the magnitude of the conformational preference of the enolate and the first-formed product of alkylation, thus exerting an influence on stereoselectivity. The same trend was observed for the alkylation with MeI, although the diastereoselectivities were generally lower with this electrophile. The decreased selectivities with MeI suggest that steric interactions in the diastereomeric transition states play some role in stereoselectivity, although, unlike for the C3-substituted enolate, these interactions are not likely to be solely responsible for stereoselectivity.^[15]

(7)

Table 1.

Alkylations of C4-substituted ε-lactams.

R1	Electrophile	Product	dr[a]	Yield[b] (%)
t-Bu	H2C=CHCH2Br	cis- 2a	96:4	77
Ph	H2C=CHCH2Br	cis- 17a	92:8	61
Me	H2C=CHCH2Br	cis- 19a	83:17	68
t-Bu	MeI	cis- 2b	89:11	53
Ph	MeI	cis- 17b	72:28	75
Me	MeI	cis- 19b	52:48	33[c]

^[a]Determined by ¹³C{¹H} analysis of the crude reaction mixture.^[16]

^[b]Isolated yield of purified product.

^[c]Reaction was ceased at 63% conversion to suppress competing di-alkylation.

The diastereoselective alkylations of C4-substituted enolates from the same face as the substituent^[12] were general for other electrophiles and substrates (Scheme 1). Benzyl, allyl, and ethyl electrophiles reacted with high stereoselectivity. Addition of the enolate to an aldehyde was also stereoselective for attack from the same face as the substituent, although the facial selectivity on the aldehyde was low, as anticipated for similar endocyclic lithium enolates.^[17] Acylation with methyl benzoate^[18] afforded cis-2g stereoselectively. The nature of the alkyl substituent on the nitrogen atom did not influence selectivity, as seen with formation of the substituted product cis-21. Even in the case of a substrate with a relatively small alkoxy group at C4, the reaction was diastereoselective (lactam cis-23). Stereoselectivity was not limited to lactam-derived enolates. Similar diastereoselectivity was seen upon alkylation of the enolate of C4-substituted ε-lactone cis-24 (eq 8). The conditions employed in this case were modified to improve the yield of alkylation of this lactone-derived enolate, which was sensitive to decomposition.^[19] This example reinforces the distinction between seven-membered-ring enolates and six-membered-ring enolates because alkylations of C4-substituted six-membered-ring lactones proceed with low stereoselectivity.^[5a–c]

(8)

Scheme 1.

Scope of stereoselectivity^[a] of alkylations of C4-substituted ε-lactam enolates with various electrophiles.

^[a] Determined by ¹³C{¹H} analysis of the crude reaction mixture.^{[16] [b]} Isolated yield of purified product. ^[c] Product of the reaction of 1 with benzaldehyde. The ratio of syn- to anti-aldol adducts was 67:33.

The observed diastereoselectivity of reactions of C4-substituted seven-membered-ring enolates is the result of avoiding both unfavorable steric and torsional interactions upon approach of the electrophile. Calculations^[20] using a simplified computational model of the enolate^[20] showed that seven-membered-ring enolate 26 adopts a chair conformation^[21] in which the tert-butyl substituent occupies a pseudoequatorial position and the nitrogen atom is tetrahedral^[22] (Figure 1a). Approach of an electrophile from the peripheral β-face, which would lead to the major product, would be less hindered compared to approach from the non-peripheral α-face (as shown in the Newman projection of 26, Figure 1b, 1c).^[4a,23]

Figure 1:

C4-substituted ε-lactam enolates. a) Transition state geometries of TS-26α and TS-26β. b) Enlarged Newman projections of enolate 26, TS-26α and TS-26β, viewing along the C2–C3 bond. c) Schematic depictions of Figure 1b.

Transition-state calculations illustrate more subtle interactions that cause attack from the different faces to have such different energies. Attack from the favored peripheral β-face would enable the transition state to avoid having the vicinal protons at C2 and C3 pass through an eclipsed transition state, as illustrated in the conversion of 26 to TS-26β (Figure 1b, 1c). ^[24] Conversely, attack from the disfavored α-face through transition state TS-26α would require the H–C2–C3–H torsional angle to pass through an eclipsed conformer on the way to the product (Figure 1b, 1c).^[25] This transition state TS-26α also suffers from a developing gauche interaction between the incipient bond to the electrophile and the C3–C4 bond of the seven-membered ring (C_Me–C2–C3–C4 dihedral = 50°, Figure 1b, 1c). This interaction, which is accompanied by steric interactions with the C4 and C6 methylene groups (Figure 1a), would become more destabilizing with the larger electrophile.^[20] This observation is consistent with calculations using a larger electrophile, allyl chloride, as the electrophile (Figure 1c) and with experimental results (Table 1).

The calculations also provide an explanation for why six-membered-ring lactam enolates do not react stereoselectively (eq 2). In neither transition state TS-27α and TS-27β must the enolate pass through an eclipsed conformer. Both faces of the enolate are also sterically accessible; neither mode of attack develops destabilizing steric interactions. Consequently, the energy difference between the two modes of attack are similar, leading to low stereoselectivity (Figure 2b, 2c).

Figure 2:

C4-substituted δ-lactam enolates. a) Transition state geometries of TS-27α and TS-27β. b) Enlarged Newman projections of enolate 27, TS-27α and TS-27β, viewing along the C2–C3 bond. c) Schematic depictions of Figure 2b.

The reactions of enolates of C5-substituted seven-membered lactams illustrate how small changes in the structure of the ring can exert large influences on stereoselectivity. These enolates did not react diastereoselectively with electrophiles, regardless of the size of the substituent at C5 (eq 9).

(9)

These low diastereoselectivities for the C5-substituted enolate are consistent with the analysis used to explain the high stereoselectivity of the C4-substituted enolates (Figure 1). In contrast to the enolate with a substituent at C4, substitution at the C5 position causes the seven-membered ring to flatten in the vicinity of the enolate to accommodate the large substituent.^[20] This flattening of the ring changes the orientation of the carbon–carbon double bond with respect to the ring, as illustrated in the Newman projection of 34 viewing along the C2–C3 bond (Figure 3b, 3c). This change results in the torsional interactions being minimized in one transition state, TS-34α, but steric interactions being minimized in the other (TS-34β, Figure 3b, 3c). As a result, the two modes of attack are more similar in energy, resulting in low diastereoselectivity.^[20]

Figure 3:

C5-substituted ε-lactam enolates. a) Transition state geometries of TS-34α and TS-34β. b) Enlarged Newman projections of enolate 34, TS-34α and TS-34β, viewing along the C2–C3 bond. c) Schematic depictions of Figure 3b.

Alkylations of seven-membered-ring lactam enolates with substituents at C6, like the C3-substituted ones, were highly anti-selective (eq 10). The alkylation of C6-substituted lactam 35 with allyl bromide provided product trans-36a as a 97:3 mixture of isomers.^[12] This stereoselectivity is consistent with what has been reported for alkylations of similarly substituted six-membered lactam enolates.^[26] Just as observed for the C3-substitued lactams, an enolate with a smaller substituent (37) reacted with lower stereoselectivity. Computational analysis revealed that C6-substituted seven-membered lactam enolates prefer geometries with pseudoaxial substituents to minimize allylic strain^[27] between the C6 substituent and the alkyl group on the nitrogen atom (39b, Scheme 2). This computational prediction is supported by the fact that a crystal structure of substituted lactam trans-36b (eq 11) shows that the tert-butyl group adopts a pseudoaxial orientation.^[12]

(10)

(11)

Scheme 2.

Enolates of C6-substituted seven-membered-rings.

The importance of allylic strain in determining the stereoselectivity shown in Scheme 2 is reinforced by the allylations of a C6-substituted cycloheptanones. Regioselective deprotonation of ketone 41 at the less sterically hindered position followed by alkylation occurred with high syn diastereoselectivity for several electrophiles (Scheme 3). Additionally, trapping of the silicon enolate of 41 with MCPBA^[28] occurred with complete facial selectivity, suggesting that the stereochemical outcomes of reactions of seven-membered-ring enolates may also apply more generally to cycloheptenes. The apparent reversal in stereoselectivity compared to the selectivities observed with lactams 35 and 37 can be understood by examining the conformational preferences of the enolate of ketone 35 (Scheme 2). In contrast to the lactam substrates, the lowest energy conformer of the enolate would position the tert-butyl substituent in a pseudoequatorial position (40a, Scheme 2), leading to preferential formation of cis-disubstituted products. Similar to what was observed for the C4-substituted enolate 26 (Figure 1) the C2 carbon atom of the enolate of 35 is slightly twisted in the same direction as the peripheral face of the enolate (H–C–C–H dihedral = +–5°),^[20] which likely accounts for the observed high syn diastereoselectivity.

Scheme 3.

Alkylations of C6-substituted cycloheptanones.

^[a] Determined by ¹³C{¹H} analysis of the crude reaction mixture.^{[16] [b]} Isolated yield of purified product. ^[c] Product was synthesized by oxidation of the trimethylsilyl enol ether of 41.

The diastereofacial preference for alkylation of ε-lactam substrates allows for the selective construction of α-carbon quaternary stereocenters (Scheme 4).^[10a,29] C3-substituted alkylation product trans-8a was alkylated a second time with benzyl bromide to afford the product trans-8b with high diastereoselectivity at the α-quaternary stereocenter.^[30] The alkylation of C4-substituted cis-2a with benzyl bromide afforded product cis-2h as a single diastereomer. The synthesis of other diastereomer, trans-2h, was achieved with high diastereoselectivity by alkylating the α-benzyl compound cis-2c with allyl bromide. The acylation of cis-2a with methyl benzoate occurred with similarly high selectivity to afford cis-2i. The alkylation of C6-substituted product trans-25a was only moderately selective, indicating that diastereoselectivity may be eroded in cases where the α-carbon atom is highly congested. The stereoselective synthesis of the fully substituted products in Scheme 4 provide evidence that the alkylations are under kinetic control because equilibration of these products by deprotonation and reprotonation is not possible.

Scheme 4:

Construction of α-carbon quaternary stereocenters.

^[a] The reaction was performed at −20 °C.

In conclusion, reactions of seven-membered-ring enolates with electrophiles generally occur with high diastereoselectivity, in contrast to observations with six-membered-ring enolates. The conformation of a seven-membered-ring enolate, unlike for a six-membered-ring enolate, causes the two diastereotopic faces of the enolate to be differentiated, resulting in diastereoselective reactions. Diastereoselectivity was observed for reactions of seven-membered-ring enolates derived from lactams, lactones, and ketones, and these reactions can be used to control stereoselectivity at quaternary carbon stereocenters. The patterns of selectivity suggest that conformational preferences and torsional effects determine selectivity. The major products of alkylations of endocyclic seven-membered-ring enolates can be readily predicted by establishing the preferred conformation of the enolate and considering torsional and steric interactions that develop upon approach of the electrophile.