Biomimetic 2-Imino-Nazarov Cyclizations via Eneallene Aziridination

Abstract

Amidoallyl cations are appealing three-carbon synthons for the preparation of complex amine-containing carbocycles; however, methods to generate and utilize these reactive species are limited and underexplored compared to those for oxallyl cations. Here we disclose a bioinspired strain-driven ring opening of bicyclic methyleneaziridines to 2-amidopentadienyl cation intermediates that readily engage in Nazarov cyclizations. Advantages of this strategy include ease of generation and improved reactivity compared to 3-pentadienyl cations, control over the ultimate position of the alkene, the potential for high dr between vicinal stereocenters, and the ability to further elaborate the products to fully substituted aminocyclopentanes. Experimental and computational studies support a dual role for the Rh₂L_n complex as both a nitrene transfer catalyst and a Lewis acid promoter, insight that provides a framework for the future development of asymmetric 2-imino-Nazarov cyclizations.

Flexible methods to construct functionalized, densely substituted carbocycles have long been of intense interest to the synthetic community, as these motifs occur frequently in useful bioactive molecules and natural products. For example, amine-bearing cyclopentanes are found in diverse natural products and pharmaceuticals, including the antiprotozoal compound jogyamycin, the anti-influenza drug peramivir, and nitriloprostaglandin I2 (PGI₂), which is used in the treatment of arteriosclerosis, cardiac failure, and thrombosis (Scheme 1A).

Scheme 1.

Background and Proposed 2-Imino-Nazarov Reaction

The aza version of the classic Nazarov cyclization¹ of divinyl ketones to prepare 2-cyclopenten-1-imines is challenging. This is due to better stabilization of the key pentadienyl cation in a “3-imino-Nazarov” reaction compared with the 3-oxyallyl cation intermediates implicated in a typical Nazarov reaction.² Tius, Hsung, West, Huang, and others have reported creative solutions to overcome this issue;³ nonetheless, reaction development has been hampered by the dearth of methods for convenient generation of 3-amidoallyl cation intermediates.⁴ The aza-Piancatelli reaction, which moves the N to C1, has also been used to address this challenge;⁵ however, it benefits from a “push–pull” enol–iminium intermediate that ultimately furnishes the amine-substituted cyclopentenone product.

Our work was inspired by the biosynthesis of epi-jasmonic acid (Scheme 1B), which proceeds via a Nazarov-type electrocyclization of allene oxide 2a to furnish α,β-unsaturated cyclopentenone 2b. This enzyme-catalyzed process readily controls the site of unsaturation in the product and yields excellent dr and ee of the new vicinal stereocenters.⁶ We envisaged that a nitrogen version of this process could be readily mimicked by a tandem allene aziridination/electro- cyclization (Scheme 1C) from easily obtained eneallenes of the form 3a. Rh₂-catalyzed nitrene transfer yields conjugated bicyclic methyleneaziridine 3b, analogous to 2a. The ring strain inherent in 3b (~35 kcal/mol) provides kinetically competent access to versatile 2-amidoallyl cation intermediate 3c without the need for stoichiometric Lewis or Brønsted acids⁷ or competing formation of iminocyclopropanes.^7d,f Additionally, the modified 2-N positioning in 3c reduces stabilization of the pentadienyl cation relative to traditional 3-amidopentadienyl cations to facilitate productive cyclization. A 4π conrotatory electrocyclization of 3c to 3d and subsequent loss of the Rh complex furnishes 3e, ideally with high dr. The intermediacy of 3d enables the formation of 3e with predictable positioning of the alkene, as substrate-controlled elimination present in the typical Nazarov reaction is not operative here. In fact, no elimination or tautomerization is required from 3d in this oxidative manifold, which avoids loss of stereochemical information between the vicinal stereocenters installed in the conrotatory cyclization of 3c.⁸ This Communication describes our “2-imino-Nazarov” reaction, including factors influencing the mechanism and stereo-selectivity of the cyclization, the elaboration of the products to increase molecular complexity from simple precursors, and potential strategies to secure enantioenriched products.

Hydrozirconation/Zr=O elimination of allyl propargyl alcohol with Schwartz’s reagent enables rapid access to eneallene 4 (Table 1).⁹ Preliminary studies found that sulfamates were the best nitrene precursors and that a two-carbon tether between the allene and the sulfamate was optimal. Nitrene transfer conditions previously reported by our group served as a starting point for investigating the feasibility of the 2-imino-Nazarov reaction of 4.^7a Catalytic Rh₂(TPA)₄ (TPA = triphenylacetate) transformed 4 to 4a in 24% yield with >19:1 dr (entry 1). Varying the concentration (entries 2 and 3) had little impact; however, increasing the temperature to 50 °C (entry 1 vs 4) improved the yield of 4a to 41%. Adding the catalyst last, as opposed to the oxidant, was not beneficial (entry 5).

Table 1.

Selected Optimization Studies

entry	solvent	catalyst (mol %)	temp (°C)	yield (%)a	notes
1	CH2CI2	Rh2(TPA)4 (1)	rt	24	15% rsm
2	CH2Cl2 (0.2 M)	Rh2(TPA)4 (1)	rt	28	–
3	CH2Cl2 (0.05 M)	Rh2(TPA)4 (1)	rt	33	–
4	CH2Q2	Rh2(TPA)4 (1)	50	41	–
5	CH2Q2	Rh2(TPA)4 (1)	50	33	catalyst added last
6	CH2Q2	Rh2(TPA)4 (1)	50	21	PhI(OAc)2
7	MeNO2	Rh2(TPA)4 (1)	45	23	–
8	CH2CI2	Rh2(TPA)4 (5)	50	55	LUMO = −2.90 eV
9	CH2CI2	Rh2(TPA)4 (5)	50	58	0.05 equiv of PhIO × 3, added every 20 min
10	CICH2CH2CI	Rh2(TPA)4 (5)	80	49	–
11	CH2CI2	Rh2(OAc)4 (5)	50	42	LUMO = −2.68 eV
12	CH2CI2	Rh2(esp)2 (5)	50	41	LUMO = −2.71 eV

^aNMR yields using mesitylene as an internal standard.

Switching the oxidant to PhI(OAc)₂ (entry 6) or the solvent to MeNO₂ (entry 7) gave low yields of 21% and 23%, respectively. However, increasing the loading of Rh₂(TPA)₄ to 5 mol % (entries 8 and 9) in CH₂Cl₂ improved the yield of 4a to 58%; further increases in temperature in DCE (entry 10) were not beneficial. Interestingly, Rh₂(OAc)₄ (entry 11) and Rh₂(esp)₂ (entry 12) were inferior to Rh₂(TPA)₄. Computations of the LUMO energies for Rh₂(TPA)₄ (–2.90 eV), Rh₂(OAc)₄ (−2.68 eV), and Rh₂(esp)₂ (−2.71 eV) suggested that Lewis acidity of the catalyst was important to reaction success (see the Supporting Information (SI) for details).

With the optimized conditions in hand, the scope of the tandem allene aziridination/2-imino-Nazarov reaction was explored (Table 2). All of the substrates were either racemic or a 1:1 racemic mixture of diastereomers. Freshly prepared 4 provided 4a in 63% yield with >19:1 dr, bearing an anti relationship between the hydrogens at the two newly formed stereocenters through a thermally allowed conrotatory 4π electrocyclization (Table 2, entry 1). Engaging cis-eneallene 5 gave 5a in moderate yield (entry 2), favoring the syn isomer (NOESY-NMR studies detailed in the SI); we considered that the lower dr compared to 4a could result from isomerization of the alkene geometry in 5, partial epimerization of 5a under the reaction conditions, or a competing “non-Nazarov” pathway. Control experiments confirmed that the alkene of 5 does not isomerize in the presence of Rh₂TPA₄ or PhIO, while epimerization of 5a is unlikely, as we would expect a similar dr for 4a if this pathway were operative. Isomerization of intermediate species or a competing non-Nazarov pathway cannot be ruled out.

Table 2.

Scope of the Tandem Allene Aziridination/2-Imino-Nazarov Reaction

^aIsolated yields

^bUsing 1 mol % Rh₂(TPA)₄.

^cThe yield decreased as allene aged.

Isopropyl substitution was tolerated in 6 to give 6a in 74% yield with excellent dr (entry 3). To our delight, the preparation of 6a could be run on a 1 g scale to give a 79% yield with >19:1 dr. Protected alcohols were suitable precursors, with 7 giving 7a in 65% yield with >19:1 dr (entry 4). The ability of the external stereocenter in 8 to control the dr in the all-carbon stereotriad in 8a was investigated (entry 5); while the dr between a and b was >19:1, the dr between a/b and c was only moderately improved to 1.2:1.

Substitution at C3 (R⁴) and C4 (R²) in 9 and 10 (entries 6 and 7) gave 9a and 10a in good yields; NOE correlations for 10a showed anti stereochemistry in the major diastereomer. Here, moderate dr may result from isomerization of either the precursor or intermediates, although a competing non-Nazarov pathway is also possible. Extending the conjugation of the alkene in styrenyl allene 11 (entry 8) resulted in an unoptimized 36% yield of 11a with >19:1 dr, though this substrate was prone to decomposition at high temperatures. Increased steric congestion in eneallenes 12 (entry 9), where both R¹ and R³ are alkyl substituents, gave lower yields but excellent dr in forming a challenging all-carbon quaternary center in 12a. Eneallenes 13–15, where both R¹ and R² are substituted, gave good yields of the 2-cyclopenten-1-imines 13a–15a. As the alkene in these cases cannot undergo isomerization, the less-than-perfect dr is puzzling. Prolonged exposure of 14a as a mixture to the reaction conditions revealed no change in dr over 24 h, while resubjecting diastereomerically pure anti-14a (>19:1 dr) to the reaction conditions gave no change in the dr, providing support for lack of epimerization of the imine itself (see the SI for details). The isopropenyl substituent on cyclohexene 15a promotes high anti selectivity between the adjacent stereocenters a and b.

Density functional theory calculations (see the SI for computational details) were carried out to gain more insight into the mechanism of this unusual 2-imino-Nazarov reaction (Figure 1). The fate of methyleneaziridine 16, which is readily formed upon reaction of eneallene 4a with Rh₂ catalyst in the presence of PhIO, was explored first. Exothermic coordination of the aziridine nitrogen to the Rh₂ catalyst leads to INT1, which evolves to amidoallyl cation INT2 through TS1, a saddle point associated with aziridine ring opening (ΔE^⧧ = 13.6 kcal/mol). The intermediacy of this achiral intermediate was supported by subjecting enantioenriched 4 (92% ee; see the SI for details) to the standard conditions and noting degradation of the axial chirality to only 12% ee in the product 4a. Ring closure converts INT2 into bicyclic INT3 via TS2, which is associated with the formation of a new C–C bond. The lower barrier for this step (ΔE^⧧ = 5.0 kcal/mol) is consistent with its computed high exothermicity (ΔE_R = −39.8 kcal/mol); however, the barrier is similar to that for an aza-Piancatelli reaction with the N at C1, highlighting the favorable electronic benefits of positioning the N at C2 versus C3.⁵ Decoordination of the Rh₂ fragment gives the observed product 4a and releases the catalyst. The reaction profile for the corresponding (Z)-eneallene isomer 16-Z was computed to be higher in energy than that computed for 16 along the entire reaction coordinate. In addition, the E/Z isomerization barriers computed for either INT1 or INT2 (>20 kcal/mol; see Figure S-2 in the computational portion of the SI) are much higher than the barriers associated with the ring opening and subsequent cyclization. This finding could help to explain the higher yields noted with 4 versus 5.

Figure 1.

Computed reaction profile for the 2-imino-Nazarov reaction of aziridine 16. Relative energies and bond distances are given in kcal/mol and angstroms, respectively. All of the data were computed at the SMD(CH₂Cl₂)-B3LYP-D3/def2-TZVPP//SMD(CH₂Cl₂)-B3LYP-D3/def2-SVP level.

The dual role of the Rh₂L_n nitrene transfer catalyst as a Lewis acid promoter was intriguing; while these modes of reactivity are precedented individually, there are few examples where both features of a Lewis acidic dinuclear Rh catalyst are utilized in a synergistic fashion.¹⁰ During optimization studies, higher catalyst loadings of the most Lewis acidic Rh₂L_n complex gave more efficient cyclization. Dissociation of Rh₂(OAc)₄ prior to cyclization had a prohibitively large computed energy barrier, leading to the hypothesis that the use of a racemic allene with a chiral Rh₂L_n catalyst might generate enantioenriched products. Indeed, subjecting (rac)-4 (Figure 1 inset) to the reaction conditions using Rh₂(R-PTAD)₄ produced (+)-4a and (−)-4b in a 60:40 er, a promising result considering the large distance between the catalyst and the site of the stereodetermining C–C bond formation event. Efforts to identify better chiral Rh₂L_n catalysts and counteranions for accessing enantioenriched, densely functionalized amino-cyclopentenes are currently underway. This result also provides compelling evidence that Rh₂ is involved in promoting cyclization as a Lewis acid, in addition to facilitating the nitrene transfer.

Finally, the 2-cyclopenten-1-imine products of tandem allene aziridination/2-imino-Nazarov reaction proved to be flexible intermediates for the preparation of complex aminated cyclopentanes. As shown in Scheme 2, a variety of transformations were successfully carried out on 6a (see the SI for conditions). Global reduction of 6a with NaBH₃CN produced 17 in 82% yield with 4.8:1 dr. Boc protection of the amine, separation of the diastereomers, and double displacement with NaI/NaH generated fused pyrrolidine 18 in 64% yield over two steps (>19:1 dr), showcasing one of many potential uses of the tether.^6a Higher-order cuprates, such as Bu₂Cu(CN)Li₂, gave diastereoselective 1,4-addition to 6a; subsequent protonation of the metalloenamine and imine reduction afforded 19 with >19:1 dr. Lastly, to demonstrate the application of 6a to the synthesis of fully substituted aminated cyclopentanes, nucleophilic epoxidation with H₂O₂ and catalytic NaOH provided 20 in 58% yield with >19:1 dr.

Scheme 2.

Flexible Transformations of 6a

In conclusion, we have developed an efficient 2-imino-Nazarov cyclization reaction from simple precursors. Stereo- controlled, site-selective Rh₂-catalyzed eneallene aziridination initiates an electrocyclization that furnishes structurally diverse α,β-iminocyclopentene scaffolds in good yields and dr. Investigation of the reaction mechanism suggested the formation of discrete achiral intermediates, implying that strain-promoted methyleneaziridine ring opening to give a 2- amidopentadienyl cation is operative. Computations indicated that the nitrene transfer catalyst remains associated during cyclization as a mild Lewis acid, providing a new framework for developing asymmetric 2-imino-Nazarov electrocyclizations.