Intramolecular Polar [4^⊕ + 2]-Cycloadditions of Aryl-1-aza-2-azoniaallene Salts: Unprecedented Reactivity Leading to Polycyclic Protonated Azomethine Imines

Abstract

The first example of aryl-1-aza-2-azoniaallene salts undergoing a [4^⊕ + 2]-cycloaddition reaction in which the azo bond and one aromatic π-bond make up the 4π component is described. This intramolecular reaction appears to be concerted and provides high yields of protonated azomethine imine products that contain a 1,2,3,4-tetrahydrocinnoline core. Substituted alkenes provided products that contain all carbon or nitrogen bearing quaternary centers in high yield.

Polar cycloadditions, in which one of the reacting partners is ionic, are less common than cycloadditions involving uncharged or dipolar components but often occur more readily.^[1] For example, whereas [4+2]-cycloaddition reactions that involve styrene subunits as the 4π component generally require reactive dienophiles or harsh reaction conditions to proceed,^[2] the Povarov reaction, which is the stepwise^[3] [4^⊕+ 2]-cycloaddition of N-aryliminium ions with electron rich olefins, occurs readily at or below room temperature.^[4],[5] The charge that is present in the ionic partner of polar cycloadditions is often due to the presence of a heteroatom and these systems can provide useful routes to heterocyclic products,^[1a] which are prevalent scaffolds in biologically active molecules.^[6] Although uncharged heteroallenes have been used extensively in the preparation of heterocyclic compounds,^[7] cationic heteroallenes have received less attention. To this end, we have been exploring the use of 1-aza-2-azoniaallene cations (e.g. 2, Scheme 1) in intramolecular reactions as a means of preparing a variety of aza-heterocycles. In this communication we report our discovery that aryl-1-aza-2-azoniaallene cations can react by an unprecedented intramolecular [4^⊕ + 2]-cycloaddition with pendant alkenes wherein the azo bond and one aromatic π-bond make up the 4π component.

Scheme 1.

Varied Reactivity of Aryl-1-aza-2-azoniaallene Salts

1-Aza-2-azoniaallene cations are known to react by several different pathways leading to a variety of products. For example, these species can add nucleophiles at carbon to provide azo products,^[8] can undergo 3,3-sigmatropic rearrangements,^[9] and can react in stereospecific intramolecular C-H amination reactions to provide pyrazolines by what appears to be a concerted nitrenoid-type insertion (e.g. 2 to 3, Scheme 1).^[10] In addition, these cationic heteroallenes can act as 1,3-dipoles in [3+2]-cycloaddition reactions with a variety of π-systems to provide 5-membered ring heterocycles. ^[11] We have taken advantage of this latter reactivity to prepare bicyclic diazenium salts (e.g. 6a and 6b, Scheme 1). ^[12]

While continuing our studies on intramolecular reactions of 1-aza-2-azoniaallene cations we recently prepared heteroallene 8a (Scheme 1), which could in principle form a 5,5-bridged bicyclic diazenium salt similar to 6b. However, orbital alignment in the transition state leading to that product would not be ideal; this asynchronous ring closure would be Baldwin-disfavored in the same way that 5-(enolendo)-exo-trig aldol condensations are disfavored.^[13] We were interested to observe that in fact heteroallene 8a did not undergo intramolecular [3+2]-cycloaddition, but instead reacted by an unprecedented intramolecular [4^⊕ + 2]-cycloaddition to provide a tricyclic protonated azomethine imine containing a 1,2,3,4-tetrahydrocinnoline scaffold (9a, Scheme 1). Cinnoline derivatives, including 1,2,3,4-tetrahydrocinnolines, show diverse biological activity^[14] and although tetrahydrocinnolines can be prepared by several classical methods,^[15] the development of new methods to prepare this useful scaffold continues to be an active area of research.^[16] The unprecedented nature of this reaction, the uniqueness of the protonated azomethine imine products^[17] and the potential that these products will have diverse reactivity and thus be useful synthetic intermediates encouraged us to examine this polar cycloaddition in more detail. Our preliminary results are presented here.

Our first task was to optimize the reaction conditions for the conversion of 7a to 9a. Our initial results were obtained by adding 1.2 equiv of SbCl₅ to a −78 °C solution of 7a in CH₂Cl₂ and then allowing the reaction to warm to room temperature. After some experimentation, we discovered that using a slight deficiency of SbCl₅ (0.95 equiv) resulted in cleaner crude reaction mixtures. Alternative Lewis acids were screened (Table 1) and while we were pleased to see that AlCl₃, AgOTf and TMSOTf could each mediate the reaction, they did not improve the yield or product purity compared to the use of SbCl₅. However, the triflate counter ion did give a product that was more crystalline, which allowed us to confirm the structure of 9a by X-ray crystallography.^[18] Factoring in both cost and the simplicity of using a liquid Lewis acid caused us to select SbCl₅ as the Lewis acid of choice for further studies.

Table 1.

Assessment of Lewis Acids

entry	Lewis acid	yield (%)a
1	SbCl5	86
2	AlCl3	84
3	AgOTfbc	70
4	TMSOTfcd	84

^aYield determined by ¹H NMR vs an internal standard and are based on the limiting reagent.

^bReaction conducted at room temperature for 2 h.

^c1 equiv of Lewis acid was used.

^dReaction conducted at room temperature for 24 h.

With optimized reaction conditions in hand, we next explored the scope of this intramolecular cycloaddition (Table 2).^[19] We were pleased to note that increased substitution adjacent to the heteroallene carbon was well tolerated; the more sterically hindered isopropyl derivative 7b (entry 2) provided the desired product in 88% yield, whereas the cyclohexanone derived α-chloroazo 7c (entry 3) provided tetracycle 9c in 83% yield as a 2:1 mixture of diastereomers.^[20] Incorporation of a silyloxy group adjacent to the heteroallene carbon was also well tolerated and silyl ether 7d provided the more heteroatom rich product 9d in 71% yield (entry 4). In this case, one could envision the cationic heteroallene intermediate undergoing a competitive 1,2-hydride migration to the electrophilic carbon facilitated by the adjacent oxygen, but this product was not observed.

Table 2.

Substrate Scope

entry	substrate	product	yield(%)a
1	7a	9a	86
2	7b	9b	88
3	7c	9c	83 (2:1 dr)
4	7d	9d	71
5	7e	9e	97
6	7f	9f	98
7	7g	9g	81
8	7h	9h	82
9	7i	9i	89
10	7j	9j	85
11	7k	9k	33
12	7l	9l	93

^aYield determined by ¹H NMR with 1,3,5-trimethoxybenzene as internal standard.

We next examined the scope of this reaction with respect to alkene substitution. We were pleased to observe that α-chloroazo compounds 7e and 7f (entries 5 and 6), which contained a di- and tri-substituted olefin respectively, provided excellent yields (97% and 98% respectively) of the desired product. Importantly, these examples show that this transformation can efficiently form nitrogen-bearing and all-carbon quaternary centers.

In an effort to gain some understanding of the concertedness of the bond forming events, we prepared trans and cis alkenes 7g and 7h (entries 7 and 8) and subjected each to the cyclization conditions. The cycloaddition reaction proved to be stereospecific; each substrate led to a unique diastereomer of product, which suggests that the cycloaddition process is concerted.^[21]

Incorporation of the alkene component into a ring provided tetracyclic products 9i and 9j (entries 9 and 10) as single diastereomers in high yield. These results highlight the ability of this transformation to provide structurally complex products from structurally simple starting materials.

The electronic nature of the dienophile can have a dramatic effect on the efficiency of polar cycloadditions. For example, the Povarov reaction fails when the dienophile is electron deficient.^[4d] To test the scope of this reaction with respect to the electronics of the pendant dienophile we prepared cyclization precursors 7k and 7l (entries 11 and 12) which contain electron rich and electron deficient olefins respectively. The electron rich olefin (7k) was surprisingly difficult to prepare as both it and the hydrazone precursor decomposed readily. Treating 7k to the reaction conditions provided cycloaddition product 9k in a modest 33% yield. We suspect that this low yield is not due to the cycloaddition step itself, but rather to the instability of 7k which became noticeably dark in color while setting up the reaction. It is interesting to note that this highly electron rich olefin reacted to provide a single diastereomer of product. In view of the cationic nature of the heteroallene intermediate, we expected an electron deficient alkene to be a poor reaction partner. We were surprised to observe that enoate 7l (entry 12) provided cycloadduct 9l in 93% yield. This high yield further demonstrates the broad scope of this reaction.

In view of the similarity between heteroallenes 5 and 8a (Scheme 1), which are identical except for the length of the tether that separates the heteroallene from the pendent alkene, it is interesting that 5 provides only diazenium salt product and none of the corresponding protonated azomethine imine. In addition, intermolecular cycloadditions of 1-aza-2-azoniaallene cations proceed by the [3^⊕+ 2] manifold^[11a–g] and taken together these facts indicate that the [3^⊕+ 2] pathway is intrinsically more favorable than the alternative [4^⊕+ 2]-cycloaddition described here. It seems likely that this latter reaction occurs in the cases described here because of the orbital alignment constraints discussed above which stem from the intramolecular nature of these reactions.

In conclusion, we have discovered an unprecedented reactivity of aryl-1-aza-2-azoniaallene salts. The [4^⊕+ 2]-cycloaddition reaction described herein is likely concerted and provides high yields of protonated azomethine imine products that contain a 1,2,3,4-tetrahydrocinnoline core. This reaction occurs at low temperature, is quite general with respect to alkene substitution and delivers products that contain all carbon or nitrogen bearing quaternary centers in high yield. Further studies on the scope and mechanism of this transformation including computational studies and application of this cycloaddition in natural product synthesis are planned.