Chieh-Hung
Tien
a,
Alan J.
Lough
b and
Andrei K.
Yudin
*a
aDavenport Research Laboratories, Department of Chemistry, University of Toronto, Toronto, ON M5S 3H6, Canada. E-mail: andrei.yudin@utoronto.ca
bX-Ray Crystallography Laboratory, Department of Chemistry, University of Toronto, Toronto, ON M5S 3H6, Canada
First published on 10th October 2022
The discovery of new reactions enables chemists to attain a better understanding of fundamental chemical reactivity and push the boundaries of organic synthesis. Our understanding and manipulation of high-energy states such as reactive conformations, intermediates, and transition structures contribute to this field. Herein we interrogate epoxide ring-closure by inserting the CN functionality into a well-known precursor to nucleophilic epoxide ring-closure. The synthesis of tetrasubstituted, nitrile-tethered epoxides takes place via activation of iminologous diols followed by fragmentation. Mechanistic study reveals the transformation to be stereospecific, which is consistent with the concerted nature of the epoxide ring-closure.
Unlike the double bond in vinylogy, the utility of the imino-fragment (CN) as a two-atom relay unit has not garnered attention in the synthetic community. Structurally speaking, oximes can be considered as the iminologues of alcohols and related ethers and esters (Scheme 1b).11,12 Driven by the ongoing interest in the energy-rich N–O bond of oximes,12 we considered the iminologue of 1,2-diols (1). Doing so appeared attractive because of the known migration, ring-closing and fragmentation transformations of 1,2-diols (Scheme 1c).8,13 Insertion of an imino fragment into the C–O bond of one of the hydroxy groups would afford structures such as 1, and subsequently the iminologous pseudohalohydrin upon oxime activation (Scheme 1d).11e,14
Entry | Substrate | Activator | 2 (%) | 3 (%) |
---|---|---|---|---|
a The crude reaction was subjected to analysis by NMR spectroscopy, but no purification was attempted. b Analysis by NMR spectroscopy indicated formation of nitrile-containing products, but no purification was attempted. c DIPEA (1.1 eq.) was added to 1a, followed by TsCl (1.1 eq.). After 1 h, KOtBu (2.1 eq.) was added. d KOtBu (2.4 eq.) and TsCl (1.4 eq.) were used. KOtBu = potassium tert-butoxide, TFAA = trifluoroacetic anhydride, MsCl = methanesulfonyl chloride, TsCl = p-toluenesulfonyl chloride, NsCl = 2-nitrosulfonyl chloride, DIPEA = diisopropylethylamine. | ||||
1a | 1a (R = H) | Ph2POCl | n/a | n/a |
2a | 1a | TFAA | n/a | n/a |
3a | 1a | MsCl | n/a | n/a |
4b | 1a | TsCl | n/a | n/a |
5a | 1a | NsCl | n/a | n/a |
6c | 1a | TsCl | 23 | 8 |
7 | 1b (R = Me) | TsCl | 12 | 14 |
8 | 1c (R = CONBn2) | TsCl | n/a | 78 |
9d | 1c | TsCl | n/a | 88 |
With the optimized conditions in hand, a series of iminologous diols 1d–q were synthesized, and the scope of the reaction was investigated (Scheme 2). Switching the dibenzylamide moiety to dimethylamine, derived from DMF, slightly lowered the yield to 68% (3d). The presence of an aniline on the amide substituent did not hinder the reaction and 3e was obtained in high yield. Other amides derived from cyclic amines such as pyrrolidine (3f) and morpholine (3g) were also well tolerated. Single crystals of 3f were grown and analyzed by X-ray crystallography to confirm the identity of the molecule. By switching to a less electron-rich amide derived from N-methylaniline, the reaction became less efficient, but otherwise gave 3h in synthetically useful yields. A substrate containing the antidepressant desipramine fragment was isolated in 81% yield (3i). Changing the amide fragment to an equally electron-withdrawing CF3 group afforded epoxide 3j in 24% yield. Novel [2.2]- and [2.4]-spirocyclic epoxides 3k–m were synthesized in low to good yields. Notably, the olefin group in 3m was tolerated, which could be sensitive towards other oxidant-based epoxidation conditions. Reducing ring size of the precursor to five atoms gave propionitrile-tethered epoxide 3n in 57% yield. Increasing the ring size to seven atoms gave 3o in 32% yield. Interestingly, an isobaric side product (3oa) that does not contain a nitrile functional group was isolated. Based on the 13C NMR chemical shifts of the carbonyl species, and the coupling patterns of the methylene proton (see ESI†), the side product was proposed to be in a rigid, cyclic system. Thus, the structure of 3oa was assigned to be a bicyclo[4.2.1]-dihydroisoxazole bearing an exocyclic amide and a strained bridgehead carbonyl. This product was presumed to arise from the direct intramolecular substitution on the electrophilic nitrogen atom of the respective iminologous pseudohalohydrin by the negatively charged alkoxide. Lastly, substrates with substituents on the ring can also smoothly undergo the reaction to afford 3p and 3q in moderate to good yields.
To further expand the scope of this process, 4a and 4b, where the alcohol moiety is external to the ring system, were synthesized (Scheme 3). Treatment of 4a under standard conditions did not afford the desired epoxynitrile (Scheme 3a). To our surprise, similar to 3oa, a direct displacement of the pseudohalide by the alkoxide onto the nitrogen atom occurred to exclusively give dihydroisoxazole 5a. The identity of this species was further confirmed by single crystal X-ray crystallography. Analogous reactivity was observed when 4b was employed, and the methylated dihydrooxazines 5b was isolated in high yields (Scheme 3b). The presence of the diester substituent could lower the barrier of cyclization to favour dihydroisoxazole formation instead of the desired pathway to afford epoxides.16
Scheme 3 Synthesis of dihydroisoxazoles 5a (shown as X-ray crystal structure, displacement ellipsoid set at 30% probability level) and 5bvia the activation of iminologous diols 4a and 4b. |
The epoxynitriles can undergo further transformations to access other highly decorated molecules (Scheme 4). Treatment of 3c with SmI2 in THF effected the regioselective ring-opening of the epoxide to afford allylic alcohol 6. Lewis acid-mediated fluorination of 3g using BF3 gave fluorohydrin 7a in 37% yield. To our surprise, a significant amount of borylated heterocycle 7b did not undergo hydrolysis under reversed-phase chromatography conditions, and was isolated in 48% yield. The structure was identified by X-ray crystallography. Lastly, in the presence of a strong nucleophile such as nBuLi, the nitrile was reduced to the corresponding ketone, while maintaining the epoxide moiety, albeit in low yields.
Scheme 4 Downstream functionalization of epoxynitriles 3c and 3g. X-ray crystal structure of 7b (displacement ellipsoid set at 30% probability level). |
We considered the mechanism of this transformation (Scheme 5). Without TsCl, the reaction does not proceed, and 1c was recovered, indicating the intermediacy of the corresponding iminologous pseudohalohydrin (Scheme 5a). Using only one equivalent of KOtBu and TsCl gave 3c in 44% isolated yield (Scheme 5b). It was previously established that when 1a and 1b were subjected to the reaction conditions, both the ring-closing products, 3a and 3b, and the H-migration products, 2a and 2b, were isolated (Scheme 5c). The ketones are presumed to arise from the [1,2]-H migration (semipinacol rearrangement) of the intermediates that contain adjacent carbocations. To further probe the presence of carbocations in amide-tethered substrates (since no migration was observed), oximes (±)-cis-1r and (±)-trans-1r were synthesized and isolated. The major isomer was crystallized, and a crystal structure was obtained to confirm the relative stereochemistry. Both isomers, (±)-cis-1r and (±)-trans-1r, were subsequently subjected to the standard conditions, and to our surprise, the epoxide products were afforded as diastereomerically pure species (±)-trans-3r and (±)-cis-3r, respectively (Scheme 5d). The relative configuration of the two diastereomers were assigned based on NOESY NMR assignment (see ESI†). This suggests the lack of planarization of the α-carbon atom (carbon in the α-position relative to the oxime), which is inconsistent with the formation of carbocations in this process. Therefore, C–O bond formation and C–C bond cleavage must occur in a concerted or otherwise stereospecific fashion.
Scheme 5 Mechanistic studies. X-ray crystal structure of (±)-cis-1r (displacement ellipsoid set at 30% probability level). |
Two other substrates (±)-1s and (±)-1t were synthesized and the diastereomers were separated (Scheme 6). Subjection of (±)-trans-1s and (±)-cis-1s to the optimized conditions provided epoxides (±)-cis-3s and (±)-trans-3s, respectively, with good yields in a stereospecific fashion. Ethylene ester-substituted (±)-cis-1t did not react under the reaction conditions, and the starting material was recovered upon hydrolysis during reversed-phase purification. Interestingly, diastereomeric (±)-trans-1t gave the desired epoxide (±)-cis-3t, albeit in low yields.
Considering the aforementioned results, we put forth a potential mechanism of the epoxidation process (Scheme 7a). First, the oxime is activated to form the corresponding iminologous pseudohalohydrin A. If the ring size is sufficiently large, or if the alkoxide is spatially close to the electrophilic nitrogen, a transannular or intramolecular substitution could occur to give dihydroisoxazoles B (pathway a). Otherwise, A undergoes a stereospecific C–C fragmentation, where the group antiperiplanar to the tosylate migrates onto the nitrogen atom to generate nitrilium C as in the abnormal Beckmann rearrangement.11d Facile fragmentation of C is expected to give the stabilized carbocation D, which could further undergo a semipinacol rearrangement to afford ketones 2 (pathway b). However, in amide-tethered substrates, the anionic oxygen atom in A likely undergoes direct intramolecular substitution to expel the oxime functional group by C–C cleavage in a concerted iminologous ring-closure to directly afford 3 (pathway c).17 This pathway is only possible if the oxygen atom is equatorial (Scheme 7b, green), which makes the C–O bond antiperiplanar to the breaking C–C bond as seen in the X-ray crystal structure of (±)-cis-1r. Since the equilibrium distribution of conformers is different between diastereomers, a change in reactivity would be expected, thus explaining the higher reactivity of (±)-trans-1t with respect to (±)-cis-1t.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details; mass spectrometry data; NMR spectroscopy data; X-ray crystallography data. CCDC 2194278–2194281. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d2sc04496j |
This journal is © The Royal Society of Chemistry 2022 |