Rearrangement of Protected Epoxy-Alcohols into Tetrahydrofuran Derivatives: The Protecting Group Matters!

Abstract

We have explored an interesting rearrangement of protected epoxy-alcohols into tetrahydrofuran derivatives. Our protocol is operationally simple and involves treatment of a substrate with catalytic quantities of boron trifluoride diethyl etherate in methylene chloride without any special precautions to exclude air or ambient moisture. The nature of the protecting group dictates the stereochemical outcome of the cyclization. For example, with trans-di-substituted epoxides bearing pendant esters or carbamates, the rearrangement gives tetrahydrofurans with contiguous stereocenters in a syn configuration. With these substrates, we hypothesize that the transformation initiates upon attack of the epoxide by the carbonyl oxygen of the ester or carbamate. Conversely, with trans-di-substituted epoxides bearing free alcohols or ethers, cyclization gives tetrahydrofurans with contiguous stereocenters in an anti configuration. Here, we believe that a simple S_N2 attack on the epoxide is taking place. We also examined the cyclization with aziridine alcohols and their derivatives and with oxetane esters and found that some of these substrates were compatible with the reaction conditions.

Graphical Abstract

Our laboratory has a programmatic focus on the ring-opening of epoxides and aziridines with interesting and sometimes unusual tethers,¹ such as sulfamates and di-tert-butyl silanols.^2–8 As a natural continuation of this agenda, we wondered about the ring-opening of an epoxide with a malonate tether, a transformation which, based on our search of the literature, does not yet exist. In a model substrate such as compound A, there are two conceivable entries to attack (Scheme 1). Deprotonation of the acidic protons of the 1,3-dicarbonyl moiety would form a carbanion, which could then ring-open the epoxide to form a lactone with three contiguous stereocenters. Under the right conditions, the oxygens of the malonate could also cleave the epoxide and form a functionalized tetrahydrofuran.

Scheme 1.

The ring-opening of an epoxide with a pendant malonate could proceed with the formation of a new C−C bond or a new C−O bond.

While we were unsuccessful in forging a new C–C bond by treatment of A with a variety of bases both in the presence and absence of additives, simply stirring A with a Lewis acid such as Sc(OTf)₃ led to tetrahydrofuran formation with concomitant rearrangement of the malonate (Scheme 1). We were quite excited by this result because of our laboratory’s history with rearrangement reactions^{3, 7, 9} and because we could find few analogous literature reports, each containing a very limited substrate scope.^10–19 Pioneering contributions in this area came from the groups of Coxon^{11, 12, 20} (rearrangements of trans- and cis-1-acetoxy-3,4-epoxypentane) and Wasserman¹³ (thermal rearrangement of the epoxide of 1-allylcyclopropyl acetate). Since then, there have been sporadic reports examining the use of this rearrangement in natural products synthesis.^{14–17, 21–23} Despite these efforts, there was no systematic study of the scope of this reaction with respect to either the alcohol protecting group or the alkene substrate. Even the ring-opening of epoxides by pendant alcohols to form analogous tetrahydrofuran products is underexplored. The majority of what exists concerns the ring-opening of activated epoxides, i.e. those adjacent to arenes, alkenes, or alkynes.^24–29 For unactivated epoxides, literature protocols for tetrahydrofuran formation suffer from harsh conditions, require extended reaction times, or give mixtures of products.^30–33 Thus, we felt that this area³⁴ was quite open for further investigation.

Our first priority was to determine the stereochemical outcome of the tetrahydrofuran formation. In their seminal investigation of the rearrangements of trans- and cis-1-acetoxy-3,4-epoxypentanes, Coxon and co-workers¹² proposed a dioxolenium dance-orthoester mechanism, which was supported by clever ¹⁸O-labeling experiments (Scheme 2A, Mechanism 1). Later investigations by Giner and co-workers supported these preliminary studies.¹⁶ In their initial investigation, the Coxon laboratory used super-stoichiometric quantities of Lewis acids during the rearrangement. We wondered whether a dioxolenium dance mechanism was operative under our milder, catalytic conditions or whether a simple cyclization-protecting group transfer was taking place (Scheme 2A, Mechanism 2). We also realized that a mechanistic switch could occur based on the hydroxyl protecting group. With esters, a dioxolenium dance is plausible, but only Mechanism 2 is likely with ethers.

Scheme 2.

(A) Mechanistic proposals and (B) Salient Investigations. (Note: All compounds shown are racemic, and relative stereochemistry is depicted).

Treatment of epoxide 1 with a catalytic amount of boron trifluoride diethyl etherate (BF₃•OEt₂) gave tetrahydrofuran 2 in an excellent yield of 85% (Scheme 2B). Removal of the benzoate using potassium carbonate in methanol gave known alcohol 3.³⁵ Since 3 is a syn tetrahydrofuran, Mechanism 1 was likely operative, and this result agrees with Coxon’s hypothesis. The same sequence of reactions was done with epoxide 4, which gave known alcohol 6, again suggesting Mechanism 1.³⁶ Cyclization with TBS ether substrate 7 gave alcohol 6. With this substrate, the cyclization likely proceeded through a single inversion of the epoxide, as depicted in Mechanism 2. Collectively, these experiments show protecting-group dependent modes of cyclization, and both diastereomers of a given tetrahydrofuran can be synthesized from a common epoxy-alcohol precursor!

A brief examination of alternate conditions for this rearrangement reaction revealed that none were superior to using catalytic quantities of BF₃•OEt₂ or of Ph₃CBF₄ (Table 1, Entries 8–9). Stirring 8 with catalytic amounts of Sc(OTf)₃ in CH₂Cl₂ at room temperature gave desired product 9 in a respectable yield of 57% (Table 1, Entry 1). There were some side products which we hypothesized were arising from attack of the epoxide by adventitious water. We reasoned that adding NaHCO₃ would serve to quench trace triflic acid and would function as a desiccant. Unfortunately, adding 1 equivalent of NaHCO₃ to the reaction with Sc(OTf)₃ led to unproductive decomposition of starting material (Table 1, Entry 2). Stirring 8 with catalytic quantities of Bi(OTf)₃ or of Al(OTf)₃ gave 9 in a similar yield and accompanied by the same profile of side products as with Sc(OTf)₃ (Table 1, Entries 3 and 4). The reaction was markedly worse with Cu(OTf)₂, Mg(ClO₄)₂, or Zn(OTf)₂ (Table 1, Entries 5–7). With Cu(OTf)₂ and Zn(OTf)₂, there was unproductive decomposition of starting material while no reaction was observed with Mg(ClO₄)₂.³⁷

Table 1.

Optimization Experiments.

	Lewis Acida	Temp.	Yieldb
1	Sc(OTf)3 (0.3)	23 °C	57%
2	Sc(OTf)3 (0.3)c	23 °C	0%
3	Bi(OTf)3 (0.15)	23 °C	59%
4	Al(OTf)3 (0.15)	23 °C	70%
5	Cu(OTf)2 (0.15)	23 °C	0%
6	Mg(ClO4)2 (0.15)	23 °C	0%
7	Zn(OTf)2 (0.15)	23 °C	0%
8	Ph3CBF4 (0.15)	23 °C	87%
9	BF3•OEt2 (0.3)	0 °C to 23 °C	>90%

^aEquivalents are given in parentheses.

^bEstimated by ¹H NMR integration against an internal standard. Relative configuration of the product is shown.

^c1 equivalent of NaHCO₃ was added to the reaction.

We next wished to investigate this rearrangement reaction with a variety of groups attached to 2-((2S*,3S*)-3-ethyloxiran-2-yl)ethan-1-ol (Scheme 3). Groups that did not significantly attenuate the reactivity of the nucleophilic oxygen included esters (Scheme 3, Entries 1 – 3), carbamates (Scheme 3, Entry 4), and benzyl ethers (Scheme 3, Entry 7). Clean reactions were not observed with more electron-deficient moieties such as benzyloxycarbonyl (Cbz), mesyl (Ms), and phosphate groups (Scheme 3, Entries 5, 6, and 8).

Scheme 3.

Protecting Group-Reactivity Relationship.

We next examined our optimized conditions with a variety of substrates (Table 2). A range of functional groups were compatible, including aromatic ethers (Table 2, Entry 2), a Boc-protected piperidine (Table 2, Entry 2), and aliphatic ethers (Table 2, Entry 4). The substrates in Table 2, Entry 4 also illustrate that nucleophilic attack by the ester oxygens is more favorable than that from the ether oxygens. Depending on the position of the epoxide in the substrate, products containing esters inside (Table 2, majority of examples) and outside (Table 2, Entry 8) the tetrahydrofuran ring could be prepared. Epoxides synthesized from a variety of substituted olefins (cis/di-substituted, trans/di-substituted, terminal/mono-substituted, terminal/di-substituted, and tri-substituted) were amenable to cyclization. In addition to epoxy-esters, epoxy-alcohols (Table 2, Entry 6) and aziridine alcohols (Table 2, Entry 10) also cyclized in excellent yields. Thus, our protocol nicely complements existing ones that use more forcing conditions for similar transformations or that give mixtures of isomeric products.^{30, 31, 33, 38–40} If a reaction works with an epoxide, it is always worth trying with an oxetane, a related, similarly strained heterocycle.⁴¹ Relative to rearrangements of epoxy-esters,²³ there are even fewer literature examples with oxetane esters.^{42, 43} We were pleased to see formation of tetrahydrofuran 52 from oxetane 51 (Table 2, Entry 9), but we note that the efficiency of the transformation was poorer than with analogous epoxy-esters.

Table 2.

Substrate scope exploration.

^aBF₃•OEt₂ (0.3 equiv.), CH₂Cl₂ (0.1 M), 0 °C to RT

^b(substrate number, product number)

^cEstimated by ¹H NMR integration against an internal standard due to co-elution of an impurity.

Not all substrates were compatible with our optimized conditions (Figure 1). Epoxides derived from protected allylic alcohols rapidly decomposed when treated with BF₃•OEt₂. Cis-disubstituted aziridine alcohol 55 and cis-disubstituted aziridine acetate 56 were similarly unsuccessful, decomposing into a variety of products which were difficult to characterize (see the Supporting Information for spectra of the unpurified reaction mixtures). While oxetane 51 (Table 2, Entry 9) did give some tetrahydrofuran product, isomeric compound 57 (Figure 1) failed to yield anything discernible. Finally, with substrate 58 (Figure 1), the benzoate of compound 53 (Table 2, Entry 10), a complex mixture containing mainly uncyclized products formed.

Figure 1.

Poor Performers.

The reaction scale could be increased from 0.3 mmol to 1.6 mmol without loss of yield or selectivity (Scheme 4A). Tetrahydrofuran 42 can be conceptualized as containing differentially protected alcohols. Hydrogenolysis of the benzyl group using 1 atmosphere of hydrogen gas and Pd/C gave alcohol 59 in an excellent yield of 96% (Scheme 4B). Removal of the acetate from 40 was facile using potassium carbonate in methanol (Scheme 4B).

Scheme 4.

(A) Scale up and (B) Applications.

In summary, we have explored an interesting rearrangement of protected epoxy-alcohols into tetrahydrofuran derivatives. Our protocol is operationally simple and involves treatment of the substrate with catalytic quantities of boron trifluoride diethyl etherate in methylene chloride without any special precautions to exclude air or ambient moisture. The nature of the protecting group dictates the stereochemical outcome of the cyclization. For example, with trans-di-substituted epoxides bearing pendant esters or carbamates, the rearrangement gives tetrahydrofurans with contiguous stereocenters in a syn configuration. With these substrates, we hypothesize that the transformation initiates upon attack of the epoxide by the carbonyl oxygen of the ester or carbamate. Conversely, with trans-di-substituted epoxides bearing free alcohols or ethers, cyclization gives tetrahydrofurans with contiguous stereocenters in an anti configuration. Here, we believe that a simple S_N2 attack on the epoxide is taking place. We also examined the cyclization with aziridine alcohols and their derivatives and with oxetane esters and found that some of these substrates were compatible with the reaction conditions. We expect these reactions to be notable to both physical organic chemists interested in rearrangement mechanisms and to synthetic chemists tasked with preparing oxygenated heterocycles.

Supplementary Material

Additional experimental details including reaction procedures and NMR spectra.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.