Abstract

The effect of solvents on the 1,3-dipolar cyclization reaction between ethyl propiolate and 2-furfuryl nitrile oxide was studied in various organic solvents. As expected, the major product was ethyl-3-(2-furanyl)-5-carboxylate. The relative ratio of the 3,5- to 3,4- disubstituted isoxazoles in dichloromethane, toluene, ethanol and dimethyl sulfoxide were 3.4, 2.0, 1.9 and 1.5 respectively. Experimental regioselectivity was found to be dissimilar to density functional theory predictions.

1. Introduction

Isoxazoles possess a wide range of biological properties including antitumor, antifungal, antibiotics, antimalarial, antituberculosis and herbicidal activities ^{1, 2, 3, 4}. Classic examples include the mefloquine-isoxazole carboxylic ester 1 ¹, the penicillin-resistant antibiotics oxacillin (2) ⁴, and the histone deacetylase (HDC) inhibitor 3 ³.

While 3,5-disubstituted isoxazoles are well-documented for their medicinal properties and as synthetic precursors ^{3, 4, 5, 7}, their 3,4-disubstituted counterparts are less prevalent. Notwithstanding, the latter also exhibits significant biological activities and are key starting materials for some natural products. Examples within this class of compounds include the tetrahydroindazole-isoxazole (4), a mild inhibitor of Myco-bacterium Tuberculosis ⁹, compound 5 (a voltage-gated sodium (NaV1.8) channel modulator) and the carboxamide carboxylic isoxazole 6, a growth inhibitor of some phytopathogenic fungi ¹⁰. Other examples include key precursors for the biologically active natural product trachyspic acid ¹¹. Compounds 7 and 8 were side products prepared in an attempt to construct the perhydrophenanthrene system of the anti-inflammatory triterpenoid, celastrol ¹², (Figure 1).

Several methods are used to prepare isoxazoles, including the reaction of hydroxylamine with α,β - unsaturated carbonyl compounds ^{13, 14} and the intermolecular [3+2] cycloaddition of alkynes and nitrile oxides ^{15, 16}.

The latter is efficient, convergent and among the more popular methods. Previous studies reveal that this type of regioselectivity favors the 3,5- to the 3,4-disubstituted adduct ^{13, 15, 16}. Molecular orbital calculations highlighted that a more favorable overlap of the HOMO of the dipolarophile and the LUMO of the dipole aligned with the selectivity observed experimentally ^{18, 19}.

Sharpless and coworkers synthesized these isoxazoles using copper as catalyst. Under these reaction conditions the 3,5-isomer were prepared selectively and with high yields ²⁰. Other groups have capitalized on their results and applied this method of synthesis to their research ²¹.

The synthesis of the 3,4-disubstituted isoxazole adduct is hardly reported in the literature ^{9, 16, 22}. The most recent citation by Chalyk and coworkers. exploited the 1,3-dipolar cycloaddition reaction between in situ generated nitrile oxides and enamines to generate 3,4-disubstituted isoxazole products ¹⁶. Using hybrid density functional theory (DFT) calculations, the Houk’s group predicted and explained that the regioselectivity for the unfavored 3,4-isomer in the cycloaddition reaction between mesitonitrile oxide and the methyl propiolate should increase in non-polar solvents ¹⁹. This regioselectivity was due to the lesser polar character of the 3,4-disubstutued isoxazole transition state in comparison to that of the 3,5 –disubstituted one.

Herein, we report the effect of solvent on the regioselectivity of the uncatalyzed [3+2] cycloaddition reaction between the in situ generated 2-furfuryl nitrile oxide 10 and ethyl propiolate 11. DFT calculations revealed an unexpected disparity between theory and experimental regioselectivity.

2. Materials and Methods

Ethyl propiolate 11 was reacted with nitrile oxide 10, which was generated in situ from furfural oxime 9 ²³ and sodium hypochlorite (NaOCl), (Scheme 1).

All isoxazoles were synthesized by following the general procedure given for selected solvents. The compounds were characterized by IR, ¹H NMR and ¹³C NMR spectral data. The ratio of the regioisomers was determined by comparing the integrals of the methine protons for the isoxazole mixture in the crude products. A Perkin Elmer Spectrum BX FT IR system. A JEOL ECS 400 MHz FT-NMR spectrometer with TMS was used as internal standard for recording the ¹H and ¹³C NMR spectra of the compounds in deuterated chloroform-d (CDCl₃). Thin Layer Chromatograpy (TLC) was done to analyze reaction using pre-coated silica gel plates. These were viewed under UV light at 254 nm and developed using potassium permanganate stain solution.

To a mixture of ethyl propiolate, 11 (0.32 g, 3.27 mmoles), furfural oxime, 9 (0.20 g, 1.80 mmoles) and solvent (5 mL) in a 100 mL round bottom flask was added bleach (0.354 M NaOCl, 13.8 mL, 4.88 mmoles) and the reaction mixture was stirred overnight at room temperature. The reaction was extracted with dichloromethane (3 x 10 mL) and the combined organic layer was washed with water (2 x 10 mL) and brine (10 mL), then dried with anhydrous Na₂SO₄. The mixture was filtered and concentrated by rotary evaportion. The crude product was then examined using proton NMR to identify the ratio of the 12/13 formed in the reaction. In some cases, flash column chromatography of the cude product produced the isoxazole products as a yellow oil.

Ethyl 3-(2-furanyl)-5-isoxazolecarboxylate (12) and Ethyl 3-(2-furanyl)-4-isoxazole carboxylate (13) Reaction in dichloromethane:

Sample data for reaction ran in dichloromethane: yield (85.7 mg, 23%), ratio of 12/13 is 1:3, ¹H NMR (400 MHz, CDCl₃): δ = 8.98 (s, 0.23 H), 7.63 (d, J = 3.6 Hz, 0.23 H). 7.61 - 7.58 (m, 1 H), 7.19 (s, 0.77 H), 6.99 (d, J = 3.6 Hz, 0.77 H), 6.58 – 6.55 (m, 1 H), 4.46 (q, J = 6.8 Hz, 1.54 H), 4.37 (q, J = 6.8 Hz, 0.46 H), 1.44 (t, J = 7.3 Hz, 2.31 H), 1.40 (t, J = 7.0 Hz, 0.69 H).

¹³ C NMR (100 MHz, CDCl₃) δ = 14.3, 14.4, 61.5, 62.6, 107.0, 111.3, 111.9, 112.1. 116.4, 143.3, 144.6, 144.8, 155.5. 156.7, 160.8, 164.3. IR (cm^-1) v = 3131, 2983, 1739, 1617, 1582, 1505, 1517.

3. Results and Discussion

Different solvents provided varying ratios of the 3,4- and the 3,5-disubstituted isoxazoles (12 and 13) in very low yields after column purification (Table 1). Careful analysis of the ¹H NMR spectrum indicated that the methine proton on the isoxazole rings of each product after workup is distinct. These singlets resonated at 7.19 ppm and 8.98 ppm for compounds 12 and 13 respectively. The product ratios were determined from the integrations of these resonances. Although the 3,5-disubstituted product was always favored, the ratio of 12/13 decreased from 3.4 to 1.5 as the solvent polarity increased. (Table 1).

To rationalize the decrease in selectivity for 12, the transition states TS-3,5 and TS-3,4 were optimized in the gas phase, DMSO and dichloromethane using the Gaussian 09 ²⁵ at the B3LYP/6-31+G(d) ^{26, 27, 28, 29, 30} level of theory.

These transition states were shown to be single-order saddle points each with one imaginary frequency. The imaginary vibrational frequencies correlated with the direction of the formation of the two sigma bonds. (C3-C4) and (C5-O1).

Our calculations revealed that in the gas phase the 1,3-dipolar cyclization favors the formation of 3,4 disubstituted product 13 both kinetically and thermodynamically by 0.22 kcal/mol and 1.09 kcal/mol, respectively (Table 2). This result was expected as the interaction between the LUMO of the electron-deficient 11 and the HOMO of 10 is more favorable ¹⁸.

Our results predicted that in dichloromethane TS-3,4 was also favored kinetically by 0.82 kcal/mol but in DMSO, this stability was reversed, and TS-3,5 was more stable by 0.97 kcal/mol. This neglibable effect of solvent on the difference in activation energy is no surprise, as these values are generally small and in our case less than 1 kcal/mol. A similar trend was reported in the literature by Mekelleche and coworkers ³¹.

Our results were quite similar to Houk’s. predictions for the most part ¹⁹. In his hands, the 1,3-dipolar cycloaddition reaction between mesitonitrile oxide and methyl propiolate favored the more polar 3,5-transition state to that of the less polar 3,4-transition state by 0.60 kcal/mol in CCl₄ and 5.6 kcal/mol in water). This stability was reversed in the gas phase by 1.3 kcal/mol. Thus an increase in solvent polarity was should favor the synthesis of the 3,5-disubstituted product and decreased the formation of the less polar 3,4- disubstituted isoxazole.

It should be noted that no experimental data was gathered at the time and these predictions were contingent on the relative dipole moments between the reactants and the transition states.

In Mekelleche’s report, the transition states of 1,3-dipolar reactions when less polar than those of the reactants, may be poorly stabilized in polar solvents ³¹.

In the reaction between ethyl propiolate 11 and furfural oxime 9, the difference in dipole moments values between the transition states and the starting material were mostly negative except for TS-3,5 in DMSO and dichloromethane which were less than one (0.79 and 0.43 respectively) (Table 2). Thus it is possible that more polar solvent could disfavor the formation of the more polar TS-3,5 to that of the less polar TS-3,4 ³¹. This could account for the unexpected decrease in regioselectivity for the 3,5-disubstituted isoxazole 12 with increase in solvent polarity. Such computational result correlated well with the experimental result of roughly a 50-50 mixture in DMSO.

This conflict between experiment and theory may also imply that additional interactions may exist between the solvent (especially in DMSO) and the transition states that were not accounted for in the calculations provided. Hence, we are considering more robust calculations to further elucidate the cause of this unexpected experimental regioselectivity.

4. Conclusion

The effect of solvent polarity on the uncatalyzed 1,3-dipolar cycloaddition of ethyl propiolate and furfural oxime was studied both experimentally and using DFT calculations. The difference in the dipole moments of the reactants and the transition states were mostly negative. Thus pointing to the fact that the more polar transition state TS-3,5 which led to the major regioisomers 13 might have been destabilized with increase in solvent polarity thus disfavoring the formation of the 3,5-disubstituted isoxazole (12) and unexpectedly improving the 3,4-disubstituted isoxazole 13.

This project is pertinent to the synthesis of antibiotics such as oxacillin and its derivatives.

Acknowledgements

We acknowledge support from a Hofstra Faculty Research and Development Grant. The Doris Lister Endowed Fellowship in Chemistry Research to Jaffarguriqbal Singh and the Hofstra University department of chemistry for assistance with supplies and chemicals. The author would also like to acknowledge support from the Department of Chemistry, Barnard College-Columbia University for additional support in purchasing chemicals and use of the IR instrument.

References