Generation and cycloaddition of semi-stabilized 2-azaallyllithiums via electrocyclic ring-opening or cycloreversion reactions

Semi-stabilized 2-azaallyl anions, i


Introduction
The [π4s+π2s] cycloadditions of 2-azaallyl anions, particularly 2-azaallyllithiums 1 with alkenes (1 → 6 → 7, Scheme 1) is a method of growing importance for the synthesis of pyrrolidines.Kauffmann discovered that semi-stabilized 1 anions bearing two or more aryl rings could be generated by the deprotonation of imines, 2, with bases such as LiN i Pr 2 . 2 We have reported that the deprotonation method could be extended to generate semi-stabilized 2-azaallyllithiums bearing only one aryl group when their cycloadditions were intramolecular. 3While nonstabilized 1 2-azaallyllithiums (i.e., those bearing only alkyl or hydrogen) are not available by the deprotonation technique, we found that tin-lithium exchange on (2-azaallyl)stannanes, 3, was useful for the generation of such anions for both inter-and intramolecular cycloadditions. 4,5 spite these successes, we continue to explore other routes to 2-azaallyl anions, since deprotonation is limited to semi-stabilized-(or stabilized-) anion formation and tin-lithium exchange is subject to the usual undesirable attributes of tetraalkyltin chemistry.Herein we report studies on two alternative approaches to 2-azaallyllithium formation, namely the electrocyclic ring-opening of N-lithioaziridines, 4, and the 1,3-anionic cycloreversion of Nlithioimidazolidines, 5. 6 ISSN 1424-6376 Page 92 El-X The aziridine 13 was synthesized as shown in Scheme 2. Sulfide 9, 9 which was prepared from 8, was oxidized to the sulfilimine 10. 10 The sodium salt of 10 was condensed with benzaldehyde to give the oxirane 11 as a 9:1 mixture of trans-and cis-isomers. 11Nucleophilic ring-opening of 11 with sodium azide gave 12.At room temperature, only the trans-epoxide reacted, producing 65% of one diastereomer of 12 and leaving 20% of 11 enriched in the cis isomer.If the reaction was carried out at 70 °C, both isomers of 11 reacted, producing an undesired mixture of diastereomers of 12.A Blum reaction 12 on 12 gave the desired aziridine 13 as the transdiastereomer only.Deprotonation of 13 with n-BuLi failed to afford useful quantities of ring-opened products at ambient temperature in a variety of solvents.However, heating a solution of the lithium salt 15 at 110 °C in benzene produced the pyrroline 14 in 81% yield (Scheme 3).A solid, presumed to be lithium hydride, was deposited on the walls of the reaction vessel and gave gas evolution upon quenching with water.A likely pathway is a thermally allowed conrotatory ring-opening of 15 to the 2-azaallyllithium 16, which undergoes a [π4s+π2s] cycloaddition yielding the Nlithiopyrrolidine 17.At this temperature, lithium hydride elimination is apparently facile, a phenomenon that has been observed with lithiated benzylic amines in other systems. 13h N Conducting the reaction at lower temperatures (ca.90 °C) led to low conversions, but traces of the pyrrolidine 18 were observed.Further evidence for the intermediacy of 17 was obtained by subjection of authentic 18 3 to n-BuLi at 110 °C in benzene (eq.1), where clean conversion into 14 was observed, with formation of lithium hydride again in evidence.Deprotonation of 13 with KH in THF in the presence of 18-crown-6 at reflux led to rapid ring-opening but no evidence of cycloaddition.In an attempt to extend this chemistry to the generation of non-stabilized 2-azaallyl anions, the aziridine 19 14 was prepared and deprotonated with a variety of bases (eq.2).No cycloadduct 21 (or the pyrroline) was observed.The material identified, other than starting material, was the isomerized alkene 20, formed in low yield when KH/18-crown-6 was used.In summary, while the aziridine route to 2-azaallyl anions can be extended to include semistabilized anions bearing one phenyl group, it cannot be used to make non-stabilized anions.Further, the conditions required to promote ring-opening of the mono-phenyl N-lithioaziridines is significantly higher than that required to open diphenyl N-lithioaziridines (110 °C in benzene vs. 66 °C in THF).At this higher temperature, lithium hydride elimination from the initially formed N-lithiopyrrolidine is observed.

1,3-Anionic cycloreversion of N-lithioimidazolidines
As described in the Introduction, our early work on 2-azaallyl anion chemistry showed that semistabilized 2-azaallyllithiums bearing a single phenyl group could be made by deprotonation, as illustrated by the first intramolecular cycloaddition of such an anion (Scheme 4). 3 During the course of the present work, we uncovered a new route to 2-azaallyl anions, namely the 1,3anionic cycloreversion of N-lithioimidazolidines (eq.3).We now report the discovery of this method, and studies of its scope.
When the reaction in Scheme 4 was followed by GC, the expected first-order kinetics were not observed; in fact, evidence for an intermediate was obtained.Performing the deprotonation in THF-d 8 allowed observation of the rapid buildup of two new compounds by 1 H-NMR, which were assigned as the lithio-imidazolidines 23 and 24 (Scheme 5).These were slowly converted to the bicyclic N-lithiopyrrolidine 17 after several hours at room temperature, producing the pyrrolidines 18 after workup.Apparently, the initial deprotonation of 22 is slow enough to allow intermolecular cycloaddition with the imine portion of a second molecule of 22. Cycloadditions of 2-azaallyl anions with imines to give imidazolidines have been observed by others. 2,15,16The reverse process, namely the anionic cycloreversion of the lithioimidazolidines to the 2-azaallyl anion, is now implicated by the results shown in Scheme 5.This process has not been previously observed, although similar anionic cycloreversions are known.In a separate experiment (Scheme 6), workup of the deprotonation reaction at partial conversion allowed the isolation of the imidazolidines 25 and 26 of undefined stereochemistry.Re-subjection of these imidazolidines to the same reaction conditions (LDA, THF, room temperature) also gave 18.Hence, for the first time, it has been demonstrated that N- lithioimidazolidines are subject to anionic cycloreversion to 2-azaallyl anions.This may allow a new route to 2-azaallyl anions that does not rely on imine deprotonation, thereby perhaps obviating the need for a non-enolizable imine.For example, imidazolidines might be synthesized from carbonyl compounds and vicinal diamines, then subjected to base to produce N- lithioimidazolidines and thus 2-azaallyl anions (vide infra).Deprotonation of the imine 27, which cannot lead to cycloaddition owing to the absence of an alkene, resulted in the formation of the imidazolidines 28 and 29 (Scheme 7).Interestingly, the regioselectivity of the dimerization was found to depend on the solvent, providing complementary results.The stereochemistry of these imidazolidines is unknown, although the phenyl groups on 29 were later found to be trans by synthesizing this compound via an unambiguous route.The imidazolidines 28 and 29 were found to be good precursors of 2-azaallyl anions (Scheme 8).Addition of LDA to a mixture of 28, 29, and trans-stilbene afforded the pyrrolidine 30 as a mixture of three stereoisomers.The imidazolidines used were those formed in the THF reaction in Scheme 7. Thus, definitive evidence for the generation of a 2-azaallyl anion (namely, 31) from imidazolidines was gathered.Since imidazolidines may be prepared from diamines and aldehydes, we sought to explore this method as a route to 2-azaallyl anions.Initially, we prepared the diamines 32 and 33 from the imidazolidines 28 and 29 (Scheme 9).The imidazolidines used in this case were derived from the benzene experiment in Scheme 7. The diamine 33 was formed as the major product upon acidic hydrolysis.Condensation of 33 with 5-hexenal in a Dean-Stark apparatus afforded the imidazolidine 34, which was treated with LDA.Cycloreversion to the 2-azaallyl anion and Nbenzylidinehexylamine occurred, leading to the bicyclic pyrrolidines 18.Unfortunately, the Nbenzylidinehexylamine by-product dimerized to the imidazolidines 28 and 29, making the workup and isolation of 18 difficult.Nonetheless, this experiment showed that diamines may be used as starting materials for 2-azaallyl anion chemistry.To avoid the dimerization of the imine by-product of the imidazolidine cycloreversion, we synthesized the N-isopropyl diamine 37 from (d,l-)-stilbenediamine 35 20 by a reductive amination approach (Scheme 10).We hoped that imidazolidines derived from 37 would form N-benzylidineisopropylamine as a by-product, which might not dimerize, owing to steric considerations.Scheme 10 shows the successful conversion of 37 to the imidazolidines 38 and 39.The relative configuration at C-2 of these compounds is assumed to be as shown, based on molecular modeling (MM2), with the assumption that the most stable diastereomer would be formed under the condensation conditions.Scheme 12 summarizes the stereochemical outcome of the imidazolidine method and the deprotonation method for 2-azaallyl anion synthesis.Note that in each case, the two routes give opposite stereochemical outcomes.The pyrrolidines 18b and 40a are the likely result of cycloaddition through the "W"-geometry of the 2-azaallyl anion, while pyrrolidines 18a and 40b must arise through a "sickle"-geometry.It is certainly believable that the imidazolidine method and the deprotonation method may give different anion geometries.However, there are several curious aspects to this hypothesis.First, as determined above, the deprotonation of 22 goes through imidazolidines!Why, then, would these imidazolidines and 38 give different results?Possible explanations include, (1), the idea that imidazolidines 25 and 26 (Scheme 6) each produce different 2-azaallyl anion geometries, and, (2), that the stereochemistry of 38 (particularly at C-2) is different from the stereochemistry of imidazolidine 25 (Scheme 6), perhaps leading to a different 2-azaallyl anion geometry.This explanation is, however, contradicted by the results shown at the bottom of Scheme 12.Note that 39, an imidazolidine analogous to 38, gives products resulting from the sickle-rather than W-anion geometry.How can the simple change of placing a phenyl group on the anionophile cause such a reversal?Also surprising is the fact that the deprotonation of 41 gives results opposite to those observed in the deprotonation of 22.However, no evidence for imidazolidine intermediates was obtained in the deprotonation of 41, so these two experiments are not entirely comparable.The relative configurations of the bicyclic pyrrolidines shown in Scheme 12 were vouchsafed by difference-NOE 1  Studies meant to explore the scope of the N-lithioimidazolidene cycloreversion route were implemented.Scheme 13 shows some imidazolidines that failed to undergo cycloreversion / cycloaddition.The first compound shows that an N-aryl substituent inhibits the reaction.The remaining compounds show that, regardless of the N-substituent, no cycloreversion is observed where the 2-azaallyl anion would be non-stabilized (i.e., no C-aryl substituent).Our tin-lithium exchange methodology thus remains as the only method available for the generation of such nonstabilized anions.

Conclusions
The electrocyclic ring-opening of N-lithioaziridines and the 1,3-anionic cycloreversion of Nlithioimidazolidines to 2-azaallyl anions have been shown to produce semi-stabilized 2azaallyllithiums bearing a single phenyl group.However, extension of either of these methods to the generation of non-stabilized 2-azaallyl anions fails, leaving tin-lithium exchange as the only viable method for the generation of such anions.

Scheme 9 .
Scheme 9. Hydrolysis of imidazolidines to diamines and their use in the preparation of new imidazolidines.

17 16 22 Ph 18a : 18b : Ph Scheme 4. Deprotonation route to the 2-azaallyllithium 16 and its cycloaddition. ISSN 1424-6376 Page 96 © ARKAT USA, Inc
Scheme 11).Smooth cycloreversion of the N-lithioimidazolidines occurred, leading to the cycloadducts 18 and 40.No by-products were observed from the dimerization of N-benzylidineisopropylamine, which was observed in nearly quantitative amounts by GC.
Scheme 12. Stereochemical comparison of the deprotonation and imidazolidine routes.