N-heteroatom substitution effect in 3-aza-cope rearrangements

Background The nature of the heteroatom substitution in the nitrogen of a 3-aza-Cope system is explored. Results While N-propargyl isoxazolin-5-ones suffer 3-aza-Cope rearrangements at 60°C, the corresponding N-propargyl pyrazol-5-ones need a higher temperature of 180°C for the equivalent reaction. When the propargyl group is substituted by an allyl group, the temperature of the rearrangement for both type of compounds is less affected by the nature of the heteroatom present. Treatment with a base, such as ethoxide, facilitates the rearrangement, and in the case of isoxazol-5- ones other ring opening reactions take precedence, involving N–O ring cleavage of the 5-membered ring. However when base-catalysed decomposition is prevented by substituents, products arising from a room temperature aza-Cope rearrangement are isolated. A possible mechanistic pathway based on free energies derived from density functional calculations involving cyclic intermediates is proposed. Conclusions The nature of the heteroatom substitution in the nitrogen of a 3-aza-Cope system leads to a remarkable difference in the energy of activation of the reaction.


Results and discussion
The required compounds 1 for the rearrangement studies were obtained by alkylation of 3 using a Mitsunobu reaction [9] as in Scheme 1. The isolated compounds 1 and 2 were easily separated by chromatography (cf. Experimental). Heating either 1 or 2 gave rise to a 3,3sigmatropic rearrangement [5,10] and the same final compound 4 as in Scheme 1 ( Table 1, entries [1][2][3][4]. In the presence of EtOK and 18-crown-6 ether at r.t., it was found that the propargyl group of 1a isomerized to the corresponding allene 5a as in Scheme 2, which could in turn suffer upon heating a 3,3-sigmatropic rearrangement leading to 6a. This last compound could also be reached from 2a by treatment with base to give 7a, followed by thermolysis. Thus coupling of these two reactions opened the way to 3 isomeric compounds of 1a. It is worth noticing that, while the N-allenyl 5a ( Table 1, entry 5) reacts in a similar way to the N-propargyl 1a (entry 1) in the 3-aza-Cope rearrangement, the corresponding O-allenyl 7a (entry 6) and the O-propargyl 2a (entry 3) rearrange faster and give better yields. Similar results are found for the pair 1a/2a and 1b/2b. The reason is probably related to the more facile nature of the 3-oxa-Cope (Claisen) versus the 3-aza-Cope rearrangements [3].
Next, the N-propargyl isoxazolinones 8a,b and the N-allyl 8c,d [11] were synthesised (cf. Experimental) and heated to give the corresponding rearranged compounds, the allenes 9a,b, (Scheme 3, Table 2, entries 1,2) and the allyl 9c,d (Scheme 4, Table 2, entries 3,4). It was found that the first ones required a much lower temperature for the rearrangement.
The synthesis of 8e is depicted in Scheme 5 as well as its product after heating (Table 2, entry 5). Desilylation [12] of compound 10 yielded the corresponding enehydroxylamine 11, which was found to be unstable and to lactonise spontaneously to the 2-allyl-isoxazolone 8e (Scheme 5). The latter on thermolysis (180°C) furnished 2-cyano-pent-4-enoic acid ethyl ester 12, as shown. In this case the thermodynamic gain resulting from the loss of carbon dioxide from the rearranged product 9e is also accompanied by the formation of the cyano derivative 12 [13].When treated with potassium ethoxide, 8e led to the formation of N-allyl-malonamic acid ethyl ester 13 instead of the rearranged product (Scheme 6).
Again loss of carbon dioxide, as found earlier by Woodman [14], occurred with a rearrangement possibly involving cyclic intermediates to provide the amide group of 13 [15].The energetics of possible mechanistic pathways for this process were explored using a density functional approach (ωB97XD functional, 6-311G(d,p) basis set and SCRF(CPCM) continuum solvation method for ethanol as a model polar solvent). The reaction is sufficiently complex that a significant number of possible mechanistic pathways can be envisaged using the classical "arrow pushing" approach. Here we adopt the approach of incrementally locating pathways to the product and optimizing them for the lowest overall activation free energy. Such an approach of course does not guarantee finding the global energetic minimum in mechanistic pathways. Instead the objective is to find a thermally reasonable pathway that might approximately correspond to the observed rate of the reaction, and along the way eliminating pathways that have unreasonably high activation free energies.
The mechanistic exploration is set out in Scheme 6. The first route explored involved 8e and ethoxide anion acting as a nucleophile. Addition of the ethoxide gives 8e-1 as the first intermediate, followed by ring closure to 8e-2 [16]. The most direct route (Occams's razor) to 13 is by a dyotropic rearrangement of 8e-2 to 8e-3, followed by protonation and decarboxylation. The energetic high point of this pathway is the transition state for the dyotropic rearrangement of the intermediate oxaziridine (8e-TS1) which manifests as unreasonably high (Additional file 1: Table S1, entry 4) a .
The next route explored was with ethoxide acting as a base, abstracting the allylic proton to give the intermediate 8e-4. A ring closure to 8e-5, followed by 1,3 intramolecular proton transfer to give 8e-6, a second proton transfer to reform alkoxide anion and final ring opening to 8e-7 gives 8e-8, which is merely a tautomer of 8e-3 and can decarboxylate as before to give 13. The high point of this pathway is 8e-TS2, which shows as an entirely reasonable barrier for the reaction (Additional file 1: Table S1, entry 5). Whilst this does not constitute a formal mechanistic proof of the reaction, it does imply a reasonable one, and further that any alternative mechanism must have an overall lower energy barrier than this one. Whereas it is traditional, indeed conventional, for the mechanism of a synthetic pathway to be speculated upon using mechanistic reaction arrows, we suggest here that increasingly such speculations must be supported by a computational exploration of the potential free energy surface. In this case, this has been done with a procedure which, including as it does dispersion corrections for all species, a triple-zeta quality basis with polarization functions, and a correction for continuum solvation, is suggested to be the minimum in quality appropriate for such exploration.
Substrate 8b, where formation of the carbanion adjacent to the ring nitrogen of the oxazolidinone is blocked by substitution, was next treated with potassium ethoxide (0.1 eq), in the presence of 18-crown-6 ether, at room temperature (5 days), and gave rise to 9b in 55% yield (cf. Scheme 8). Here the opening of the five membered ring leading to 8b-1 might lower the energy of the reaction either by allowing an easier access to the conformation leading to the transition state or by anionic charge acceleration of the rearrangement. The importance of EtOK was noticed further when, after 5 days at r. t., in its absence, the rearrangement of 8b was observed to occur, but in a much lower yield (<10%). The same compound treated with LDA or tert-BuLi, in the   presence of 12-crown-6 ether and in the same conditions of concentration and temperature, showed no reaction. Thus the anion derived from the possible base removal of the terminal triple bond proton does not appear to accelerate the rearrangement, but the oxyanion formed upon the isoxazolidinone ring opening did have a moderately positive effect enabling the rearrangement to occur at room temperature.

Conclusions
In conclusion: 1) substitution of the nitrogen-1 of pyrazolin-5-ones by oxygen lowers the temperature of the 3-aza-Cope rearrangement of isoxazolin-5-ones vis-à-vis the corresponding pyrazolin-5-ones when the rearranging element includes a propargyl group attached to the heterocyclic nitrogen-2; 2) when the propargyl group is replaced by an allyl group the rearranging temperature is higher and similar in both compounds; 3) treatment with potassium ethoxide as base in the presence of 12-crown-6 ether at room temperature, while leaving the pyrazolin-5-one heterocycle untouched, leads to ring opening of isoxazolin-5-ones, followed by further reactions. The putative oxaziridine or other cyclic intermediate then suffer base catalysed ring opening reaction whenever there are protons available in the carbon α to the ring nitrogen, leading to a 6-membered aza,oxa-ring system. When such position is blocked by substituents the 3-aza-Cope rearrangement occurs at room temperature.

Experimental General
Melting points were determined on a Reichert Thermovar apparatus and are uncorrected. Ordinary mass spectra were recorded on a Fisons TRIO 2000 or AEI MS-9 spectrometer. High-resolution MS spectra (HRMS) were obtained on a FT-ICR/ MS Finnigan FT/MS 2001-DT spectrometer at 70 eV by electron impact or on a Finnigan MAT 900 ST spectrometer by ESI. Infrared (IR) spectra were recorded on a Perkin-Elmer 1000X FT-IR spectrometer. Proton and 13 C NMR spectra were recorded in CDCl 3 on a Bruker ARX 400 spectrometer (400 MHz for 1 H, 100.63 MHz for 13 C). Chemical shifts are reported relative to tetramethylsilane as the internal reference (δ H 0.00) for 1 H NMR spectra and to CDCl 3 (δ C 77.00) for 13 C NMR spectra. IR spectra were run on an FT Perkin-Elmer 1000 instrument, with absorption frequencies expressed in reciprocal centimeters. Thin-layer chromatography was performed on Merck silica gel 60 F 254 0.2 mm thick plates, visualized under UV light or by exposing to iodine vapour. For preparative separations the plates were 0.5-1 mm thick. For flash chromatography silica Merck Kieselgel 60, 70-230 mesh was used. Usual work-up implies drying the water-or brine-washed organic extracts over anhydrous sodium sulfate or magnesium sulfate, followed by filtration and solvent removal under reduced pressure. Anhydrous solvents were dried and freshly distilled by standard methods [18].

Synthesis of starting materials Ethyl 3-oxo-2-phenyl-butanoate
To a solution of ethyl acetoacetate (1.5 g, 11.5 mmol) in CH 2 Cl 2 (45 mL), was added triphenylbismuth carbonate [19] (6.33 g, 12.6 mmol). After being stirred under argon for 24 h at 40°C, the mixture was filtered and concentrated at reduced pressure. The crude was purified by CC (silica; Et 2 O / n-hexane 1:3) to give the title compound as a yellow oil [20]

Reactions of pyrazolinones with base
In a round bottom flask, the propargylated pyrazolinone 1a or its isomer 2a is dissolved in dry THF (ca. 0.07 M) (1 eq.) under argon and 18-crown-6 ether (1 eq.) is added. After total dissolution of the reactants, EtOK (0.1 eq.) is added under anhydrous conditions. The reaction is left to react at room temperature. After consumption of the starting material (TLC control) the reaction is stopped by addition of a saturated solution of NH 4 Cl and extraction with Et 2 O (2x). The product is purified by flash chromatography [AcOEt / n-hexane (1:6)].