Nucleophilic Trapping of Alkoxy-Stabilized Oxyallyl Systems Generated from Inosose 2-O-Mesylates

Abstract

Protected inosose 2-O-mesylates generate oxyallyl systems by elimination of the mesylate group after enolization. The reaction is promoted by weak bases such as triethylamine and azide, and the oxyallyl systems are then trapped by nucleophilic alcohols or azide. Computations using DFT/B3LYP confirm that the singlet planar oxyallyl is stabilized by the exocyclic alkoxy group.

Key words

nucleophilic addition - carbocycle - carbohydrates - rearrangement - carbocation

The oxyallyl system is a transient reactive group often used in [4+3]-cycloaddition reactions to form seven-membered and bicyclic structures.[1] [2] [3] Macmillan et al. has recently expanded their use as electrophiles that react with heteroatoms and soft π-systems when generated in the solvent 2,2,2-trifluoroethanol.[4] Stabilization of the oxyallyl system can be promoted by heteroatoms such as nitrogen[5–7] and oxygen,[1b] [2] and a methoxy-stabilized oxyallyl was recently proposed as an intermediate following a cyclopropane ring-opening reaction.[8]

Oxyallyls have most notably been implicated as intermediates in the Favorskii rearrangement which transforms cyclic α-halo ketones into ring-contracted cycloalkanecarboxylic acids and acyclic alkanones into chain-extended carboxylic acids or derivatives (Scheme [1]).[9] [10] Evidence suggests that the reaction proceeds by the intramolecular cyclization of an enolate generated from an α-halo ketone 1 to give a cyclopropanone 2, in equilibrium with the oxyallyl intermediate 3 (Scheme [1]).[10] [11] [12] The cyclopropanone intermediate is attacked by alcohol or water giving a hemiacetal which then undergoes cyclopropane ring opening affording the final product 4. The rearrangement is usually conducted using alkoxide or hydroxide as both base and nucleophile, affording esters or acids, respectively. When conformationally constrained α-halo ketones are used, both ring-contracted products 4 and alkoxide α-addition products 5 can be observed.[10] The retention of stereochemistry observed in some addition products,[13] trapping reactions,[14] as well as reaction rates for alcoholysis reactions[15] have been used as evidence for the oxyallyl intermediate 3, while stereospecific examples of the rearrangement provide evidence for direct synchronous cyclopropanone formation.[9] [16]

Scheme 1 General form of the Favorskii reaction

We have recently observed nucleophilic addition to 2-O-alkyl-6-O-mesylinosose derivatives through an oxyallyl intermediate without ring contraction and now detail our findings, mechanistic investigations and the scope of the reaction.

In a recent report, we described the synthesis of a series of tetra-O-alkylinosose derivatives from tetra-O-alkyldialdoses using organocatalysis and converted the products into allo- and epi-inositol.[17] To examine the utility of the primary products of cyclization such as 6a described in this report, we attempted the synthesis of some aminodeoxyinositols. Thus, allo-inosose derivative 6a was converted into mesylate 6b using mesyl chloride and an equivalent of triethylamine in good yield (Scheme [2]). Treating the mesylate with sodium azide in DMF did not give the expected 6-azido-6-deoxyinosose, rather two diastereomers of azidoketal 7a and 7b were isolated in a 1:1 ratio in excellent yield. An evaluation of the reaction outcome suggested that the products arose from a Favorskii-type process with opening of the cyclopropanone or direct formation of a dipolar alkoxy-stabilized oxyallyl intermediate. If this was the case, then other nucleophiles might also be used to trap the intermediate formed in the reaction. Thus, mesylate 6b was stirred with triethylamine in benzyl alcohol which fulfilled the role of solvent and nucleophile and to our delight afforded a good yield of ketal 8.

Scheme 2 Rearrangement of mesylate 6b

The observed reaction outcome is similar to the Mattox rearrangement which converts a steroidal dihydroxyacetone motif into an enol aldehyde under acidic conditions.[18] A base-promoted variant of this rearrangement on betamethasone dipropionate has been reported by Li et al. who proposed hydrolysis and rearrangement of an intermediate enol.[19]

The characterization of ketal 8 was performed using 1D and 2D ¹H and ¹³C NMR spectroscopy. A methylene group, not present in the starting material, was assigned on the basis of two upfield resonances at δ = 2.87 and 2.73 ppm in the ¹H NMR spectrum. Assignment of protons around the ring was straightforward using the COSY spectrum, and the relative stereochemistry of the product was found to match the starting material although the conformation was ring-flipped, on the basis of the trans-diaxial coupling constant (J = 8.8 Hz) observed between H-3 and H-4 and a gauche H-4 to H-5 coupling. Ketal 8 exhibited both ketone (δ = 202 ppm) and ketal (δ = 102 ppm) resonances in the ¹³C NMR spectrum consistent with the proposed structure.

Many reactions on carbohydrates and inositols show dependence on the stereochemistry of the substrate. In addition, the Favorskii reaction on cyclohexyl substrates is influenced by the relative geometry of both the acidic proton and the leaving group.[10] To examine the scope of the reaction and the effect of stereochemistry, a series of inososes were subjected to the reaction conditions (Scheme [3]). Mesylation of the alcohols under standard conditions afforded 9b, 12b, and 14b which were then treated with triethylamine in benzyl alcohol or in the case of 14b with triethylamine in methanol. The reaction was generally tolerant to stereochemistry and good yields of ketal 8, 13, and 15 were obtained from 9b, 12b, and 14b, respectively. The product isolated from the reaction of 9b was spectroscopically identical to the product obtained from the reaction of 6b as both of the epimeric α-centers are eliminated in the rearrangement. Attempted isolation of the mesylate from 10 was unsuccessful and so the crude mesylate obtained after precipitating ammonium salts using ethyl acetate was treated with benzyl alcohol and triethylamine which afforded minor amounts of ketal 11. Inosose 10 was a difficult substrate, prone to decomposition, especially on silica which may have contributed to the low yield of 11. Conformations of the products were assigned on the basis of coupling constants which indicated that the products 11, 13, and 15 were ring-flipped relative to the starting materials.

Scheme 3 Synthesis and rearrangements of 9b, 10, 12b, and 14b showing conformations of the products

All mesylates underwent varying degrees of decomposition on silica so we applied our chromatography-free procedure for 10 to the other hydroxy ketones (Table [1]).

The yields obtained from 6a and 12a were superior using this chromatography-free methodology, however, both 9a and 14a afforded inferior isolated yields of ketals. Application of the conditions to the methyl ether 16 afforded a poor yield of ketal 17.

Table 1 Chromatography-Free Rearrangement Starting with Hydroxy Ketone^a

Entry	Starting material	Product	Yield (%)
1	6a	8	58
2	9a	8	30
4	12a	13	64
5	14a	15	45
6	16	17	27

^a Reactions were performed on 40–60 mg of hydroxy ketone using 2 equiv of MsCl and Et₃N and once complete, amine salts were precipitated using EtOAc, the mixture filtered, concentrated, and BnOH or MeOH (2 mL) and Et₃N (4 equiv) were added.

A mechanism for the reaction of 6b which explains the observed product 8 is shown in Scheme [4]. Axial α-hydrogens are more acidic than equatorial hydrogens as the orbitals are aligned such that the developing carbanion can be stabilized by the carbonyl group. Equatorial leaving groups are required for the formation of the cyclopropanone as this makes the C–X σ* orbital accessible to the enolate carbanion.[10] The conformation in solution for 6b is shown in Scheme [3] based upon the 10 Hz trans-diaxial coupling seen for H-3 and H-4. Thus, starting with 6b, the triethylamine in the reaction mixture promotes the formation of enolate 16a from which the cyclopropanone 17 can form by direct ring closure due to the equatorial mesylate. Ring opening of the cyclopropanone 17 promoted by the benzyloxy group gives the alkoxy oxyallyl intermediate 18. Alternatively, a ring flip of enolate 16a to give 16b makes the mesylate axial and elimination can afford 18 without the involvement of the cyclopropanone 17. Formation of the final product could proceed by protonation of 18 to give 21 and then nucleophilic addition of the alcohol, or by addition of alcohol, to give 19 and then protonation. Formation of tautomer 20 from mesylate 6b could avoid oxyallyl 18, proceeding through the oxocarbenium ion 21, and such a mechanism would be similar to that proposed for the Mattox rearrangement.

Scheme 4Mechanism for the rearrangement of 6b

Attempted small-scale trapping of the proposed benzyloxyoxyallyl species from 9b with two equivalents of 2,6-dimethylfuran in CDCl₃ gave only reaction with adventitious water, then tautomerism affording unstable 22 as well as 8 formed from eliminated benzyl alcohol (Scheme [5]). Attempted trapping with furan and cyclopentadiene gave complex mixtures with no clear evidence for any [4+3] reactions.

Scheme 5

To examine the reaction mechanism further we have used computational chemistry to study a model methoxyoxyallyl system using DFT B3LYP/6311++G^**.[20] Previous computational studies have examined the semibenzylic and cyclopropanone pathways for the Favorskii rearrangement but have not examined the effect of alkoxy substitution adjacent to the ketone.[21] The oxyallyl species has previously been studied computationally which has shown the ground state to be a singlet diradical.[23]

There were three energy-minimized structures found for 2-methoxyoxyallyl 23, corresponding to the planar singlet structure 23P, the chair cyclopropanone 24C, and the boat cyclopropanone 24B (Figure [1,]Table [2]). Both 24C and 24B are drawn with a formal cyclopropyl ring as these energy-minimized structures are isoenergetic to those found starting with 23. In vacuum, the planar conformation 23P was 0.9 kcal·mol^–1 more stable than the lowest-energy boat cyclopropanone 24B, while the alternative chair 24C was destabilized by 4.9 kcal·mol^–1 as the methoxy group eclipsed the neighbouring C–H bond. In vacuum, the triplet diradical was calculated to be 10.8 kcal·mol^–1 higher in energy than 23P.

Figure 1 Energy-minimized structures for the model system 23

The p-type lone pair found on the alkoxy oxygen stabilizes the planar conformation 23P as rotation around the C–O bond taking these electrons out of conjugation causes the structure to collapse to the cyclopropanone 24B. The empirical solvation model[22] implemented in Spartan ’14 shows significant stabilization of the planar 23P relative to the cyclopropanones. The polar solvents used would stabilize 23P as an intermediate and this may explain why ring-contraction products are not observed.

Table 2Selected Properties of Energy-Minimized Structures Using DFT B3LYP/6311++G^**

	Interatomic distance (Å)					Relative energy (kcal·mol–1)	Solvation energy (kcal·mol–1)	Natural charges
	C1–C6	C1–C2	C2–C6	C1–O1	C2–O2			C2	C6
24C	1.48	1.47	1.61	1.20	1.39	+4.9	+6.30	+0.21	–0.33
23P	1.41	1.46	2.33	1.26	1.31	0	0	+0.52	–0.16
24B	1.47	1.47	1.65	1.21	1.38	+0.9	+6.16	+0.23	–0.30

The long C1–O1 bond favors resonance hybrid 23 and the short C1–C6 bond relative to the C1–C2 bond as well as the short C2–O2 bond relative to the same bond in 24C or 24B favor canonical form 23(ii). Calculation of the natural charges[24] showed cationic character at C2 (+0.52) and anionic character at C6 (–0.16). The charges and bond lengths explain why addition is observed adjacent to the alkoxy group at C2 and not at the allylic C6 position. In addition, the enolate formed by reaction of methanol at C2 of 23P is computed to be 11.4 kcal·mol^–1 more stable than the enolate formed from addition at C6.

In conclusion, we have demonstrated that 2-O-alkyl-6-O-mesylinososes undergo an oxyallyl or oxyallyl cation mediated rearrangement in the presence of alcohols and amine base yielding solely addition products. We were able to trap the intermediates (18 or 21) using both azide and alcohols. Importantly, the conditions are extremely mild and the examples demonstrate the reaction is general and tolerant to stereochemistry around the ring. The reaction could find use transferring oxidation within carbohydrate structures and in generating masked 1,2-diones.

• References

• 1a Harmata M. Chem. Commun. 2010; 46: 8886
• 1b Harmata M. Chem. Commun. 2010; 46: 8904
• 2 Lee JC, Jin S.-j, Cha JK. J. Org. Chem. 1998; 63: 2804
• 3 Meilert KT, Schwenter M.-E, Shatz Y, Dubbaka SR, Vogel P. J. Org. Chem. 2003; 68: 2964
• 4 Vander Wal MN, Dilger AK, MacMillan DW. C. Chem. Sci. 2013; 4: 3075
• 5 Lohse AG, Hsung RP. Chem. Eur. J. 2011; 17: 3812
• 6 Du Y, Krenske EH, Antoline JE, Lohse AG, Houk KN, Hsung RP. J. Org. Chem. 2013; 78: 1753
• 7 Krenske EH, He S, Huang J, Du Y, Houk KN, Hsung RP. J. Am. Chem. Soc. 2013; 135: 5242
• 8 Goodell JR, Poole JL, Beeler AB, Aubé J, Porco JA. J. Org. Chem. 2011; 76: 9792
• 9 Kende AS. Org. React. 1960; 11: 261
• 10 Baretta A, Waegell B In Reactive Intermediates. Springer; Berlin: 1982: 527
• 11 Loftfield RB. J. Am. Chem. Soc. 1951; 73: 4707
• 12 Hamblin GD, Jimenez RP, Sorensen TS. J. Org. Chem. 2007; 72: 8033
• 13 House HO, Thompson HW. J. Org. Chem. 1963; 28: 164
• 14 Fort AW. J. Am. Chem. Soc. 1962; 84: 4979
• 15 Fort AW. J. Am. Chem. Soc. 1962; 84: 2620
• 16 Stork G, Borowitz IJ. J. Am. Chem. Soc. 1960; 82: 4307
• 17 Stockton KP, Greatrex BW, Taylor DK. J. Org. Chem. 2014; 79: 5088
• 18 Mattox VR. J. Am. Chem. Soc. 1952; 74: 4340
• 19 Li M, Chen B, Lin M, Chan T.-M, Fu X, Rustum A. Tetrahedron Lett. 2007; 48: 3901
• 20 Spartan ’14. Wavefunction Inc; Irvine, CA: 2014. 1.1.4 ed.
• 21 Moliner V, Castillo R, Safont VS, Oliva M, Bohn S, Tuñón I, Andrés J. J. Am. Chem. Soc. 1997; 119: 1941
• 22 Chambers CC, Hawkins GD, Cramer CJ, Truhlar DG. J. Phys. Chem. 1996; 100: 16385
• 23 Ichino T, Villano SM, Gianola AJ, Goebbert DJ, Velarde L, Sanov A, Blanksby SJ, Zhou X, Hrovat DA, Borden WT. Angew. Chem. Int. Ed. 2009; 48: 8509
• 24 Reed AE, Weinstock RB, Weinhold F. J. Chem. Phys. 1985; 83: 735