The stereoselective intramolecular reductive etherification of δ-trialkylsilyloxy substituted ketones with catalytic bismuth tribromide and triethylsilane provides a convenient method for the construction of cis-2,6-disubstituted tetrahydropyrans. This method was highlighted in the key step of an expeditious total synthesis of the antibiotic, (—)-centrolobine.

The stereoselective construction of cyclic ethers remains a topical area of synthetic interest, particularly with respect to C-glycosides that are ubiquitous in biologically important natural products.¹ This may be attributed to their diverse and significant biological activity and the challenges associated with the design of stereochemically flexible approaches that provide access to either diastereoisomer. We have recently developed a two-component etherification reaction for the diastereoselective construction of cis- and trans-2,6-di- and trisubstituted tetrahydropyrans in excellent yield.^2,3 These studies provided compelling evidence for hydrogen bromide and bismuth oxybromide, derived from the in situ hydrolysis of bismuth tribromide, to be responsible for the catalysis.^4,5

Herein, we now describe the extension of the intramolecular reductive etherification process using a series of tert-butylsilyloxy ketones 1 to highlight this strategy for the stereoselective construction of cis-2,6-disubstituted tetrahydropyrans 2 (Scheme 1). This work also provides additional support for Brønsted acid rather than Lewis acid catalysis, and highlights this transformation in an expeditious total synthesis of the antibiotic (—)-centrolobine (4).⁶

Scheme 1 General Strategy for the Stereoselective Construction of cis-2,6-Disubstituted Tetrahydropyrans with an Intramolecular Reductive Etherification

Recent studies by Bajwa and co-workers implicated triethylsilyl bromide as the Lewis acid catalyst in a series of intermolecular reductive etherification reactions with catalytic bismuth tribromide and triethylsilane.^7,8 We have demonstrated that the two-component etherification is catalyzed by hydrogen bromide and bismuth oxybromide, derived from the hydrolysis of bismuth tribromide.² This subtle difference in catalyst and similarity in the reagents prompted the reinvestigation of the reductive etherification to determine which of these species was responsible for the catalysis.

The combination of bismuth tribromide and triethylsilane is known to generate elemental bismuth, hydrogen bromide, and triethylsilyl bromide.⁸ Preliminary studies focused on the hypothesis that bismuth tribromide and triethylsilyl bromide provide an in situ source of hydrogen bromide, which then functions as the active Brønsted acid catalyst (Table 1, entries 1, 4, and 7). The addition of activated molecular sieves to each of these reagents, which should sequester only the HBr and water, completely suppresses the reaction (entries 2, 5, and 8).¹⁰ Moreover, the addition of 2,6-di-tert-butyl-4-methylpyridine (DTBMP), which should neutralize any hydrogen bromide, leads to analogous results (entries 3, 6, and 9).² Hence, the reductive etherification reactions are consistent with Brønsted acid rather than Lewis acid catalysis.^11,12

Table 1 Elucidation of the Role of Bismuth Tribromide in the Intramolecular Reductive Etherification Reaction (Scheme 1; 1a, R1 = PhCH2, R2) Ph)a

These experimental findings prompted the mechanistic revision of the original catalytic cycle proposed by Bajwa and co-workers, as outlined in Figure 1. The Brønsted acid formed through the reduction of bismuth trihalide is clearly not effectively buffered by the acetonitrile.⁸ Protodesilylation of the δ-trialkylsilyloxy substituted ketone i catalyzed by the Brønsted acid should afford the hemiketal iii. Acid-catalyzed dehydration of iii should afford the intermediary oxocarbenium ion iv, which will be reduced to afford the cis-2,6-disubstituted tetrahydropyran v. The Brønsted acid can then be recycled through the hydrolysis of the trialkylsilyl halide, with the water derived from the dehydration of the hemiketal completing the catalytic cycle.

Figure 1 Proposed catalytic cycle for the reductive etherification with bismuth trihalides and trialkylsilanes.

Table 2 summarizes the scope of the intramolecular reductive etherification. The δ-tert-butyldimethylsilyloxy ketones 1a—i furnished the cis-2,6-disubstituted tetrahydropyrans 2a—i in 82-97% yield with excellent diastereoselectivity (Scheme 1).¹³ Interestingly, the reaction is remarkably tolerant to a wide array of substituents. For example aryl-, α-branched alkyl-, alkyl halide-, and alkene-containing substituents (entries 1-5), β-keto esters (entry 6), and hydroxymethyl derivatives (entries 7-9) are suitable substrates.

Table 2 Scope of the Diastereoselective Intramolecular Reductive Etherification Reactions (Scheme 1; R1 = PhCH2)a

(—)-Centrolobine A (4) is an antibiotic that was isolated from the heartwood of Centrolobium robustum and from the stem of Brosimum potabile in the amazon rain forest.⁶ Solladie and Rychnovsky have recently completed independent enantioselective total syntheses of this agent, and thereby determined its absolute configuration.¹⁴ We envisioned that the reductive etherification using bismuth tribromide and triethylsilane would provide an expeditious route to this agent, as outlined in Scheme 2.

Scheme 2 Retrosynthetic Analysis for (—)-Centrolobine; Potential Reductive Etherification Reactions

The initial approach to (—)-centrolobine (4) involved the examination of the reductive etherification of an aryl ketone i (Scheme 2; Route A). Enantioselective allylation of aldehyde 5,¹⁵ and protection of the resulting secondary alcohol (95% ee), furnished the triethysilyl ether 6 in 77% overall yield, as detailed in Scheme 3.¹⁶ The alkene 6 was then subjected to cross-metathesis by using the second-generation Grubbs' catalyst, to afford the corresponding α,β-unsaturated ketone.¹⁷ Selective hydrogenation of the alkene with Wilkinson's catalyst furnished the aryl ketone 7 in 74% yield from 6. Treatment of the δ-triethylsilyloxy aryl ketone 7 with bismuth tribromide and triethylsilane at room temperature, followed by in situ removal of the tert-butyldimethylsilyl group afforded (—)-centrolobine (4) in 93% yield, with excellent diastereoselectivity. Synthetic (—)-centrolobine (4) was identical in all respects with the reported spectral data for the natural substance (¹H/¹³C NMR and IR), including optical rotation [α]²² _D −90.3 (c = 0.88, CHCl₃) {lit.^6b [α]_D −92.2 (c = 1, CHCl₃)}. Overall, this total synthesis was accomplished in 5 steps from aldehyde 5, in 53% overall yield, making it the most efficient route developed to date.

Scheme 3 Stereoselective Synthesis of (—)-Centrolobine with an Intramolecular Reductive Etherification

The alternative approach to (—)-centrolobine (4) involved the reductive coupling of a benzylic triethylsilyl ether ii (Scheme 2; Route B). Treatment of the ketone 8, which was prepared in 4 steps by using an analogous reaction sequence, with bismuth tribromide and triethylsilane furnished none of the desired product. This observation is presumably the result of the solvolysis of the activated benzylic triethylsilyl ether and/or alcohol formed through protodesilylation.

In conclusion, we have demonstrated that the intramolecular reductive etherification using bismuth tribromide and triethylsilane provides a versatile route for 2,6-disubstituted tetrahydropyrans. These studies also provide additional support for the notion that the hydrolysis of bismuth tribromide leads to the generation of hydrogen bromide, which functions as a Brønsted acid catalyst. Finally, this methodology was highlighted in an expeditious total synthesis of the antibiotic, (—)-centrolobine.

We sincerely thank the National Institutes of Health (GM58877) for generous financial support. We also thank Johnson and Johnson for a Focused Giving Award, and Pfizer Pharmaceuticals for the Creativity in Organic Chemistry Award. The Camille and Henry Dreyfus Foundation is thanked for a Camille Dreyfus Teacher-Scholar Award (P.A.E.).