We noted at the outset that access to ‘retention’ spiroketals in the erythro-glycal series (3a–g) is particularly challenging. The corresponding ‘inversion’ spiroketals (4) are thermodynamically favored in most cases,² owing to double anomeric stabilization.³ Further, the erythro-glycal epoxides 2a–g should be kinetically predisposed to spirocyclization with inversion of configuration via favorable trans-diaxial epoxide opening. However, we recognized that the problem at hand bears a notable similarity to a key challenge in carbohydrate synthesis, namely the synthesis of β-mannosides.⁴ One effective solution has been to direct the desired β-glycosylation reaction syn to the axial C2-hydroxyl group of mannose using a covalent tether to the nucleophile.⁵ By analogy, we reasoned that, in our spiroketal synthesis, an appropriate multidentate Lewis acid might serve as a non-covalent tether between the epoxide oxygen and the sidechain hydroxyl of 2 (Figure 2). The Lewis acid could then activate the epoxide electrophile (5) to form an oxonium intermediate (6), then deliver the sidechain nucleophile to the desired β-face of the anomeric carbon (7). In this manner, the required epoxide opening with retention of configuration might be achieved in a kinetically controlled reaction, overriding the inherent thermodynamic and kinetic preferences of the system.

To explore this hypothesis, we carried out initial experiments with erythro-glycal 1a.^2,6 Epoxidation with DMDO provided the reactive glycal epoxide 2a, which began to cyclize spontaneously even at reduced temperatures (NMR, −65 °C). Since isolation of 2a was, thus, precluded, we added various multidentate Lewis acids directly to the nascent epoxide at −78 °C and analyzed the resulting product ratios after warming to room temperature (Table 1).⁷ Despite our initial concerns that the acetone cosolvent used in the epoxidation reaction might interfere with substrate coordination by these Lewis acids, we were encouraged to find that all of the reagents tested provided an improved ratio of 3a:4a compared to the spontaneous cyclization (entry 1). In particular, Ti(Oi-Pr)₄ provided the retention spiroketal 3a as a single stereoisomer (entry 6), albeit in low purity (≈55%). Further investigations revealed that warming the reaction to 0 °C immediately after addition of Ti(Oi-Pr)₄ dramatically improved the yield of 3a by avoiding the formation of various glycoside and overoxidation products (2 equiv Ti(Oi-Pr)₄, −78 °C; then 0 °C, ≤ h; >98:2 dr, 81% isolated yield). Importantly, exposure of the inversion spiroketal 4a to the reaction conditions did not result in equilibration to 3a, establishing that this Ti(Oi-Pr)₄-mediated spkocyclization is, indeed, kinetically controlled.⁸ We observed reduced stereoselectivity using substoichiometric amounts of Ti(Oi-Pr)₄, suggesting that the metal may remain coordinated to the product (7), although this complex is not responsible for the stereochemical outcome of this kinetically controlled reaction.

We next explored the effectiveness of this reaction in spirocyclizations of stereochemically diverse substrates with various sidechain lengths (Figure 3 and Supporting Information). We were gratified to find that, in the erythro series, the reaction provided contrathermodynamic five- and six-membered ring retention spiroketals with complete stereocontrol and good yields (3a–f), including 3b, which has no anomeric stabilizations. The seven-membered ring spiroketal 3g was also formed stereoselectively, but in low yield. The reaction was similarly effective in the threo series (3h–n)⁹ and, in particular, provided the retention spiroketal 3j, which we have previously found to be contrathermodynamic in spite of double anomeric stabilization.²

To evaluate our proposed tethered spirocyclization mechanism (Figure 2), we carried out conformational analysis of transition state models developed by Deslongchamps for oxonium-based kinetic spiroketalizations.^9,10 In particular, our analysis revealed that alternate non-chelated mechanisms are inconsistent with the observed stereochemical preference for 3. Conversely, a metal-chelated early transition state model (cf. 6) appears energetically favorable and is consistent with formation of the retention spiroketals.

We recognized that this strategy might also provide a means to achieve the related intermolecular glycosylations of glycal epoxides to generate β-mannosides.¹¹ Indeed, early investigations of this idea have produced promising results, with β-selectivity as high as 10:1 achieved in a model system.⁹