Stereoselective Dynamic Cyclization of Allylic Azides: Synthesis of Tetralins, Chromanes, and Tetrahydroquinolines

Abstract

This report describes the stereoselective synthesis of 3-azido-tetralins, -chromanes, and -tetrahy-droquinolines via a tandem allylic azide rearrangement/Friedel–Crafts alkylation. Exposure of allylic azides with a pendant trichloroacetimidate to catalytic quantities of AgSbF₆ proved optimal for this transformation. This cascade successfully differentiates the equilibrating azide isomers, providing products in excellent yield and selectivity (>25 examples, up to 94% yield and >25:1 dr). In many cases, the reactive isomer is only a trace fraction of the equilibrium mixture, keenly illustrating the dynamic nature of these systems. We demonstrate the utility of this process via a synthesis of hasubanan.

Tetralins, chromanes, and tetrahydroquinolines represent privileged structural motifs present in pharmaceuticals, agrochemicals, and natural products. Many feature an amino substituent at the C-3 position (Figure 1). These molecules display a wide range of activity including treatments or potential treatments for cancer, pain, depression, thrombosis, Parkinson’s disease, and malaria.^1–9 Due to this rich history, numerous synthetic methods can generate these systems including cyclization,^10–15 annulation,^16,17 cycloaddition,^18,19 partial reduction,^20,21 and others.^6,22–24 Presented herein is a distinctive carbon–carbon bond forming reaction that generates these privileged motifs with the required 3-amino-functionality by manipulating the allylic azide rearrangement.

Figure 1.

Representative biologically active compounds.

Allylic azides exist as an equilibrating mixture of isomers. This was first noted by Winstein, who documented the rate of isomerization for prenyl and crotyl azides.²⁵ Since then, chemists have struggled to exploit this rearrangement synthetically because of difficulty differentiating the azide isomers.^26–33 Only a few reports accomplish selective elaboration. The Craig lab reported a tandem Claisen rearrangement (Figure 2a).³⁴ Selectivity was achieved by orchestrating a second irreversible sigmatropic process. The Aubé lab reported a tandem Schmidt reaction that attained selectivity through cyclization and achieved diastereocontrol via chairlike intermediates (Figure 2b).³⁵ Recently, we initiated a program to explore the allylic azide rearrangement’s potential.³⁶ We reported the first example of enantioselective resolution of symmetric systems.³⁷ Herein, we report a synergistic approach that combines the use of proximal functionality with a complexity generating cyclization. By merging these tactics (Figure 2c), we obtain selectivity on unsymmetric systems and accomplished a tandem Friedel–Crafts alkylation (Figure 2d). This reaction results in the highest combined yield and diastereoselectivity reported to date for a dynamic allylic azide functionalization. We demonstrate the utility of this chemistry through a succinct synthesis of hasubanan.

Figure 2.

Selectivity for Allylic Azide Functionalization.

Our strategy incorporated a second proximal allylic functional group. When the azide is proximal (Figure 2c, OIM group, isomers i E and iv Z), the second group would be primary. However, when the azide is distal (Figure 2c, ii trans isomer), the second group would be allylic and therefore substantially more reactive. We chose a trichloroacteimidate because it could be readily activated under electrophilic conditions, such as those for glycosylation,^38,39 rearrangement,^40–45 or substitution.^46–50 A pendent arene might intercept the putative electrophile in a Friedel–Crafts alkylation. Ring closure could potentially proceed from either allylic terminus and the tether would enforce regioselectivity. Imposing selectivity by these combined approaches could afford a highly chemo-, site-, regio-, and diastereoselective protocol for the synthesis of 3-azido-tetralins, -chromanes, and -tetrahydroquinolines (X = CH₂, O, or N-PG, respectively; Figure 2d).

We began with allylic azide 1a (Table 1). This allylic azide exists as an equilibrium mixture (1.3:1.0:0.7:trace E:trans:Z:cis, representations shown in Figure 2c). The reactive trans isomer is only ∼30% of the mixture. We exposed azide 1a to a number of activators, including Lewis acids (entries 1–3), Brønsted acids (entries 4–7), and transition metal complexes (entries 8–10). Most conditions led to decomposition (entry 1), poor selectivity (entries 2,3), or poor reactivity (entries 3–5, 8, and 9). Control experiments, conducted after an initial hit with cationic gold(I) (entry 10), provided encouraging results with silver salts. Silver salts with noncoordinating counterions were efficacious (entries 12–14) and those with more lipophilic counterions provided superior results. Conditions with catalytic AgSbF₆ are mild, high yielding, and highly stereoselective (entry 14). In a control experiment, 2,6-di-tert-buty-4-methyl-pyridine inhibited the reaction (entry 15). The reaction is slow in ethanol stabilized chloroform (entry 16) or in the presence of deliberately added water (entry 17). These observations implicate general acid catalysis.⁵¹ The reaction was tolerant to ambient conditions (entry 18).

Table 1.

Optimization of Tandem Rearrangement Friedel-Crafts Alkylation

entrya	catalyst	yield %b	drc
1	BF3·OEt2	35	28:1
2	Cu(OTf)2	75	9:1
3	Zn(OTf)2	58	9:1
4	TFA	4	4:1
5	TsOH	4	7:1
6	Tf2NH	80	13:1
7	TfOH	49	7:1
8	PdCl2	0	nd
9	[(COP)PdCl]2	0	nd
10	JohnPhosAuSbF6	75	21:1
11	AgOTs	0	nd
12	AgClO4	81	7:1
13	AgOTf	80	7:1
14	AgSbF6	92	21:1
15d	AgSbF6	3	3:1
16e	AgSbF6	55	16:1
17f	AgSbF6	71	21:1
18g	AgSbF6	77	20:1

^a0.10 mmol substrate at 0.1 M in CHCl₃ for 24 h.

^bDetermined by GC-FID analysis using naphthalene as an internal standard. Values are the average of duplicate trials.

^cDetermined by GC-FID analysis.

^d20 mol % 2,6-di-tert-butyl-4-methylpyridine was added. nd = not determined.

^eThe reaction was conducted in CHCl₃ stabilized with EtOH.

^f5 equiv of water were deliberately added.

^gThe reaction was conducted under ambient conditions.

We explored the scope of this cyclization with a series of allylic azides (Scheme 1). Common substituents on the aryl ring were tolerated (H, OMe, Cl, or Br, 2a–2d). Conveniently, derivative 2d provided diffraction quality crystals. Analysis unambiguously demonstrated the relative configuration of 2d. Other compounds were assigned by analogy to compound 2d. Different tethers were tolerated. The 3-methyl group is not required for diastereoselectivity (2e). Azide 2e maps onto biologically active 3-amino-tetralins (Figure 1). Interestingly, imidate 1f exists almost exclusively as the unreactive isomers by ¹H NMR analysis (>99% E and Z). The reactive trans isomer is destabilized by syn-pentane interactions with the geminal methyl groups. However, this does not inhibit reactivity and compound 2f was isolated in acceptable yield. This observation supports Curtin–Hammett kinetics with rate limiting aromatic substitution. Moving the methyl group to the center of the allylic system eroded the dr (1.7:1 for compound 2g). We propose a putative stereochemical model based on chair like transition states (Figure 3). The pathways differ in the orientation of the vinyl group. The major diastereomer could arise from the pseudoequatorial orientation. When the hydrogen atom is replaced with a methyl (1g), then the relative energy approaches unity and diastereoselectivity is reduced (2g). Lastly, a group larger than methyl was permitted in the backbone (2h).

Scheme 1.

Formation of Tetralins 2a–h by Tandem Process^a

^aYields reported for isolated material as the average of duplicate trials. The dr was determined by ¹H NMR.

Figure 3.

Putative cyclic stereochemical model.

Incorporating a heteroatom into the system would generate valuable heterocycles. A number of ethereal allylic azides were converted into chromanes in high yield and selectivity (Scheme 2, 4a). Activating groups such as methyl (4b) or methoxy (4c) were competent as were compounds with a halogen substituent (entries 4d–4f). Poly substituted aromatics were tolerated (entries 4g–4i). These results are gratifying because the heteroatom is a basic site that could slow catalysis. The proximal oxygen would inductively reduce the stability of a cationic intermediate and could disrupt the azide equilibrium.⁵² These substrates also indicate limitations in this method. The formation of 4j was slow and imidate decomposition occurred. Chromane 4k was isolated as a mixture of regioisomers (1.4:1).

Scheme 2.

Formation of Chromanes 4a–l by Tandem Process^a

^aYields reported for isolated material as the average of duplicate trials. The dr was determined by ¹H NMR.

We explored aniline derived allylic azides that lead to azido-tetrahydroquinolines (Scheme 3). Several azides were prepared with varying arene substitution and N-protecting group. All imidate precursors in this class contained less than 10% of the reactive trans isomer at equilibrium. Gratifyingly, these azides readily afforded tetrahydroquinolines in good yield with high dr. A number of aryl-substituents were tolerated (entries 6a–6f).

Scheme 3.

Formation of Tetrahydroquinolines 6a–f^a

^aYields reported for isolated material as the average of duplicate trials. The dr was determined by ¹H NMR.

We conducted the reaction on a gram scale (Scheme 4). Using >1 g of imidate 1a provided tetralin 2a in 82% yield. The product could be oxidized under the Upjohn protocol⁵³ to afford diol 7. Reduction using palladium on carbon provided amine 8. Selective reduction of the azide with LiAlH₄ afforded amine 9. The protocol of Evans⁵⁴ was used to form pyrrolidine 10. Cycloaddition provided triazole 11.

Scheme 4.

Gram Scale Reaction and Diversification of Product

We were emboldened to attempt a synthesis of hasubanan (Scheme 5). Hasubanan is the parent structure of the hasubanan alkaloids, a large family of botanical natural products. Members of this family have been studied synthetically and display diverse biological activities.^55–60 The vicinal tetrasubstituted centers in the core provide a significant synthetic challenge. To the best of our knowledge, there are no prior reported syntheses of the parent molecule hasubanan. We began with ester 12, which is directly available from commercial material.^61–63 Reduction and partial reoxidation afforded aldehyde 13, which was elaborated by Corey–Chaykovsky epoxidation. Epoxide 14 was opened regioselectively with NaN₃ in acetone/water yielding allylic azide 15. Imidate 16 was isolated after activation with trichloroaceto-nitrile. The key cyclization proceeded as expected to establish the necessary stereochemical relationship and the quaternary carbon center. The minor diastereomer was not observed. The final functional group manipulation was accomplished through reduction and ring closure with dicyclohexyl borane.⁵⁴ This last step clearly illustrates the advantage of using an allylic azide in this synthesis because the final amine functionality was established with concurrent C–N bond formation.

Scheme 5.

Synthesis of Hasubanan by Tandem Process

In conclusion, we report a tandem allylic azide rearrangement-Friedel–Crafts alkylation. This dynamic cyclization process resolves a mixture of equilibrating allylic azide isomers by exploiting the enhanced reactivity of the allylic electrophile. We demonstrated that this process is general and leads to valuable 3-azido-tetralin, - chromane, and -tetrahydroquinoline hetero-cycles. Furthermore, the products of this reaction are readily diversifiable.