Formation of Aminocyclopentadienes from Silyldihydropyridines: Ring Contractions Driven by Anion Stabilization

Abstract

Highly functionalized aminocyclopentadienes are formed by rearrangement of anions generated from readily prepared 6-silyl-1,2-dihydropyridines by desilylation with fluoride. The scope and generality of the reaction are defined, and diverse transformations are performed on the highly functionalized products. A mechanism and driving force for rearrangement is identified from experiment and DFT calculations.

A new contract(ion)

Densely functionalized aminocyclopentadienes are efficiently accessed by anionic rearrangement of 6-silyl-1,2-dihydropyridines. DFT calculations have been performed to elucidate the mechanism of this transformation.

We previously reported the Rh(I)-catalyzed C–H bond addition/electrocyclization cascade reaction of readily available α,β-unsaturated imines 1 with trimethylsilyl alkynes, providing efficient access to diverse 6-trimethylsilyl-substituted 1,2-dihydropyridines (DHPs) 3 (Scheme 1).^[1] Subsequent protonation/desilylation of the silyl-substituted DHPs 3 generated unstabilized ylides 4 that were converted to pharmaceutically relevant tetrahydropyridines 5 and tropanes 6.

Scheme 1.

Reactions of silyl-substituted DHPs 3.

Upon investigation of the desilylation of DHPs 3 under basic conditions using tetrabutylammonium fluoride hydrate, we were surprised to obtain aminocyclopentadienes 8. This transformation provides rapid entry to densely substituted derivatives that would be difficult to access by other methods.^[2] Notably, aminocyclopentanes are present in many natural products^[3] and pharmaceuticals such as the recently approved drugs Ticagrelor^[4] and Peramivir.^[5] We now report on the scope and limitations of this new transformation and demonstrate that products 8 are versatile intermediates for further elaboration. We have determined a mechanism for this transformation from experiment and computation.

A variety of conditions were explored for mediating desilylative rearrangement of silyl DHP 3 (see Table S1). Bu₄NF hydrate proved to be the optimal reagent for effecting this transformation, but the presence of water was also necessary. The optimized conditions were next applied to a number of silyl-substituted DHPs 3 (Table 1). We first evaluated DHPs with an ester group at the R⁵ position to provide aminocyclopentadienes 8a–8e. DHPs 3a with R¹, R² and R³ = Me and 3b with R¹ and R² = Me and R³ = H provided products 8a and 8b, respectively, in high yields. The bicyclic product 8c was similarly obtained in a high yield. Moreover, the aminocyclopentadiene product 8d demonstrates efficient access to densely substituted products with differential substitution at all five of the R¹ to R⁵ sites. Replacement of the methyl ester (8a) with the more sterically encumbered tert-butyl ester (8e) resulted in minimal reduction in yield.

Table 1.

Substrate scope.^[a]

^[a]Conditions: 3 (1 equiv), Bu₄NF hydrate (1.2 equiv) in THF (0.1 M). Reactions generally proceeded much more rapidly than 18 h, but for evaluating scope were allowed to continue for longer times. Isolated yields after silica gel chromatography.

^[b]X-ray structure shown with anisotropic displacement ellipsoids at the 50% probability level. The pircrylsulfonate counterion and hydrogen atoms omitted for clarity.

^[c]Conditions: Bu₄NF (2.0 equiv), 72 h.

Different R1-nitrogen substituents could be incorporated successfully into the aminocyclopentadiene product, including benzyl (8f), alkyl (8g and 8h), and even aryl (8i) N-substituents. The structure of the rearrangement product was rigorously confirmed to be the aminocyclopentadiene motif by X-ray structural analysis of the amine salt of 8i.^[6] As demonstrated with aminocyclopentadienes 8f to 8l, an aryl group could be incorporated at the R⁵ position instead of the ester group, including trifluoromethylphenyl (8f–8i), phenyl (8j), 2-fluorophenyl (8k), and 3-chlorophenyl (8l) groups.

To enhance synthetic utility, the reaction sequence from imine 1 was carried out without isolation of DHPs 3 (Table 2). Previously prepared aminocyclopentadienes 8a and 8f were obtained in 52% and 67% overall yields from 1a, respectively. The additional aminocyclopentadienes, 8m displaying different alkyl substituents from R² to R⁴ and bicyclic 8n, further establish that these densely functionalized products can be rapidly prepared from readily available TMS alkynes and imines.

Table 2.

Direct synthesis of aminocyclopentadienes 8 from imine 1.^[a]

^[a]Conditions: 1^st step: alkyne (2 equiv), from 1.25 to 15 mol % of [RhCl(coe)₂]₂ with equimolar amounts of 4-Et₂PPhNMe₂ and from 2 to 48 h depending on DHP 3 synthesized (see Supporting Information). 2^nd step: Bu₄NF hydrate in THF (0.1 M). Isolated overall yield for sequence based on 1 after silica gel chromatography.

While not all substitution patterns for DHPs 3 underwent clean rearrangement to aminocyclopentadienes 8, the side products obtained for these derivatives provided very useful information for ascertaining a mechanism for this new transformation (vide infra). For DHP 3o with an R⁵ methyl group, no reaction occurred under the standard conditions, and upon heating to 50 °C, while the rearrangement product 8o was not observed, 1,4-dihydropyridine 9o was obtained in moderate yield (Scheme 2).^[7] In contrast, DHP 3p with the electron-rich 4-methoxyphenyl group at the R⁵ position, gave a mixture of rearrangement product 8p and 1,4-dihydropyridine 9p (entry 2).

Scheme 2.

DHPs 3 that provide the 1,4-dihydropyridines. Conditions: 3 (1 equiv), Bu₄NF hydrate in THF (0.1 M). [a] 18 h at 50 °C. [b] 72 h at rt. Yields determined by ¹H NMR spectroscopy with 2,6-dimethoxytoluene as the external standard.

The N-benzyl DHP 3q with an R⁵ phenyl group also resulted in an unexpected outcome. The aminocyclopentadiene 8q was produced in very low yield with the pyridine byproduct 10 obtained in 52% isolated yield instead (eq 1). Extrusion of the tolyl anion from the N-benzyl substituent likely occurred due to the considerable resonance stabilization achieved by aromatization after generating the anion upon silyl cleavage. In support of this hypothesis, pyridine 10 was not detected for the reaction of DHP 3j, which possesses an N-cyclohexyl instead of an N-benzyl group (see 8j, Table 1). Extrusion of the more basic cyclohexyl anion would be less favorable in this case.

(1)

Synthetic transformations were performed upon the aminocyclopentadiene products 8 to demonstrate their utility as intermediates for further elaboration. Diels-Alder reaction of aminocyclopentadiene 8f with N-phenyl maleimide provided the norbornene adduct 11 as a single isomer with the facial and endo selectivity rigorously determined by X-ray crystallography (eq 2).^[8,2b]

(2)

More extensive elaboration of the fused bicyclic aminocyclopentadiene 8c was also carried out (Scheme 3). Diels-Alder cycloaddition with N-phenyl maleimide and methyl acrylate gave the polycyclic adducts 12 and 13, respectively, in good yields and with very high regio- and diastereoselectivity. Dihydroxylation with catalytic osmium under acidic conditions occurred exclusively at only one of the two double bonds and proceeded with moderate diastereoselectivity to provide 14.^[9] Selective reduction of the ester to the corresponding alcohol 15 was accomplished in high yield with DIBAL. Finally, straightforward acetylation of the secondary amine gave amide 16.

Scheme 3.

Elaboration of aminocyclopentadiene 8c. Reaction conditions: a) N-phenylmaleimide, CH₂Cl₂; b) methyl acrylate; c) OsO₄, NMO, citric acid, tBuOH/H₂O; d) DIBAL, CH₂Cl₂; e) (CH₃CO)₂O, NEt₃, CH₂Cl₂.

The most closely related transformation is that reported by Fusi and Adamo for ring contraction of highly electron-deficient N-trifluoroacetyl-3,5-dicyano-1,4-dihydropyridines (Scheme 4).^[10,11]

Scheme 4.

Previously reported transformation by Fusi and Adamo.

We investigated the mechanism of the ring-contraction of 3 to aminocyclopentadiene 8 with DFT calculations. The mechanism in Pathway A (Scheme 5), analogous to that proposed by Fusi and Adamo, is consistent with the experimental observation that an electron-withdrawing R⁵ substituent is necessary to obtain 8 in a good yield. After desilylation of dihydropyridine 3 to yield anion 17, 6π disrotatory electrocylization occurs via TS1 to afford aziridine intermediate 18, with its formal negative charge stabilized by the electron-withdrawing R⁵ substituent. Aziridine intermediate 18 opens regioselectively in the presence of a proton source to furnish 8. An alternative mechanism, Pathway B, where the C6–N1 bond cleaves heterolytically to yield a carbene intermediate 19 that undergoes 6π-electrocyclization to provide 8 was also considered. We calculated the energy profiles for each pathway with the results shown in Table 3. DFT computations were performed with B3LYP geometry optimizations and M06-2X energy evaluations^[12] in implicit CPCM solvent.^[13] Computational details and discussion on the possible TMS elimination pathways to give 18 are given in the Supporting Information.^[14]

Scheme 5.

Possible mechanistic pathways from DHP 3 to aminocyclopentadiene 8.

Table 3.

Computed free energies (kcal/mol) at the M06-2X/6-311G++(d,p)-CPCM (THF)// B3LYP/6-31G(d) level of theory of transition states and intermediates along Pathways A and B relative to desilylated DHP 17 (R¹ to R⁴ = Me).

R5	Structure	ΔG‡	Structure	ΔG
4-CF3-Ph	TS1a	5.9	18a	−8.2
Ph	TS1b	7.9	18b	−2.8
CO2Me	TS1c	1.2	18c	−14.7
Me	TS1d	15.3	18d	11.3
4-CF3-Ph	TS2a	48.2	19a	18.6

The relative energies of the two pathways A and B were determined with a trifluoromethylphenyl group at the R⁵ position and with the N-R¹ substituent modeled with a methyl group to simplify the calculations (Table 3). The ΔG^‡ of the aziridine-forming transition state TS1a was calculated to be 5.9 kcal/mol, indicating the feasibility of Pathway A. In contrast, the ΔG^‡ of the carbene-forming transition state TS2a was calculated to be 48.2 kcal/mol, clearly excluding Pathway B from consideration.

For electron-withdrawing substituents at the R⁵ position the ring-contracting step along Pathway A was calculated to have free energies of activation TS1a–c ranging from 1.2 kcal/mol (R⁵ = CO₂Me) to 7.9 kcal/mol (R⁵ = Ph), with the length of the forming C2–C6 bond ranging from 1.92 Å (R⁵ = Ph) to 2.01 Å (R⁵ = CO₂Me). With a methyl group as the R⁵ substituent, the free energy of activation TS1d is much higher at 15.3 kcal/mol,^[15] and the forming C2–C6 bond length of 1.79 Å indicates a considerably later transition state. The formation of 18 was found to be exergonic for all electron-withdrawing R⁵ substituents, with ΔG for 17a–c to 18a–c ranging from −14.7 kcal/mol for R⁵ = CO₂Me to −2.8 kcal/mol for R⁵ = Ph. In contrast, for 17d with R⁵ = Me, the formation of 18d is endergonic with a ΔG of +11.3 kcal/mol. These results indicate that both TS1 and 18 are strongly stabilized by an electron-withdrawing R⁵ substituent. For 17a with R⁵ = 4-CF₃-Ph, the free energy of activation is 2.0 kcal/mol lower and the reaction is >5 kcal/mol more exergonic than for R⁵ = Ph, showing that for R⁵ groups with similar steric profiles, stronger electron-withdrawing ability provides better stabilization.

The calculated energies are consistent with the experimental results. In particular, when 3o with R⁵ = Me was submitted to the reaction conditions, the aminocyclopentadiene product 8o was not observed and DHP 9o was instead obtained (Scheme 2). This result is in agreement with high endergonicity of 17d to 18d, with direct protonation of 17d occurring to give 9o rather than rearrangement.

The energy profiles of the two regiodivergent pathways, C and D, of opening up 18a are shown in Figure 1, with Pathway C leading to the experimentally observed aminocyclopentadiene isomer 8a. Spontaneous opening of the aziridine ring in 18a would result in a highly basic amide anion, which is calculated to be unstable. This result indicates that a proton source is required for the ring-opening, consistent with the observation that anhydrous fluoride sources result in greatly diminished yields (Table S1). Computed geometries of TS4 involve mainly C–N bond breaking with very little proton transfer. At the level of theory used for geometry optimizations, hydrogen-bonded amide intermediates 20 were located, but these local minima disappear at the higher level of theory used for the energetics (Figure 1), suggesting that the C–N breaking and proton transfer are likely energetically concerted but highly asynchronous. While the energetics are not expected to be accurate, these results do provide information on how hydrogen bonding and proton transfer influence the reaction barrier. At both levels of theory, the calculations show the C–N breaking to be rate-determining, with TS4a being lower than TS4b by 2.2 kcal/mol. This result is in excellent agreement with the experimentally observed product, isomer 8a.

Figure 1.

Energy profiles (in kcal/mol) of Pathways C and D for the regiodivergent aziridine ring-opening of 18a to yield aminocyclopentadiene isomers at the M06-2X/6-311G++(d,p)-CPCM (THF)// B3LYP/6-31G(d)-CPCM (THF) level of theory. Distances denoted in Angstroms.

We sought to provide direct experimental support for the aziridine intermediate 18 based upon the observation that silyl DHP 3p provided a mixture of 8p and 9p albeit in low yield (Scheme 2). We hypothesized that deprotonation of the secondary amine in 8p might result in the aziridine intermediate, which upon protonation could provide a mixture of 8p and 9p. One caveat is that rearrangement of 3 requires that Bu₄NF be employed as the hydrate while deprotonation of 8p requires a strong base under anhydrous conditions. Perhaps for this reason, we did not detect the formation of 1,4-DHP 9p upon deprotonation under a variety of conditions. However, deprotonation with BuLi and 12-crown-4^[16] at −40 °C followed by protonation with acetic acid gave the aminocyclopentadiene isomer 21p along with 8p, providing significant support for aziridine intermediate 18p (Scheme 6).

Scheme 6.

Support for aziridine intermediate. [a] Yields determined by ¹H NMR spectroscopy with 2,6-dimethoxytoluene as the external standard.

In conclusion, we have discovered a new type of ring-contraction of 6-azacyclohexadienyl anions to obtain densely functionalized aminocyclopentadienes 8 from readily accessible silyl DHPs 3. The scope and limitations of the reaction were defined with many different types of products 8 obtained in good to excellent yields. The mechanism of this transformation was elucidated by DFT computations. A variety of regio- and stereoselective transformations of aminocyclopentadienes demonstrate the versatility of these products as synthetic intermediates.