Explorations of an intramolecular route to pyrrolo[3,4-b ]isoxazoles: an unexpected retro-Claisen reaction

Potential precursors have been prepared for intramolecular 1,3-dipolar cycloaddition to form a pyrrolo[3,4-b ]isoxazole. The cycloaddition has not to date been accomplished, however an unexpected retro-Claisen reaction is reported


Introduction
We have reported over a number of years on synthetic approaches to the 3-acyltetramic acid group of metabolites, which display an N-heterocyclic enolised tricarbonyl core, as illustrated in general structure 1 (Figure 1). 1 These metabolites show a range of biological activities, ranging through antibiotic, antitumour, antifungal and antiviral; 2 they also have a relationship to the inducers of bacterial quorum sensing. 3[6]  Most recently we have reported on our 2 nd generation approach that employs the 5,6dihydropyrrolo [3,4-b]isoxazol-4-one building block 2 through the route shown in Scheme 1, which involves the 1,3-dipolar cycloaddition of an amino acid-derived nitrile oxide dipole with a -ketoester enamine as dipolarophile. 5One of the less efficient steps in this sequence is the ring closure (lactam formation) of 3-(1-aminoalkyl)-5-methylisoxazole-4-carboxylates to complete the pyrroloisoxazole framework 2. We were interested to investigate whether this sequence could be improved by reversing the order of N1-C2 bond formation and cycloaddition, making the amide bond first, which would also have the possible benefit of making the cycloaddition intramolecular.We report here our efforts to date, which generated an intermediate having a nitro group as dipole precursor and a -keto amide as dipolarophile precursor, but revealed an unexpected retro-Claisen reaction on attempting to generate the dipolarophile.

Results and Discussion
As the desired dipolarophile in the cycloaddition is an enamine prepared from a -keto acid derivative, an obvious initial approach based on intermediates we had in hand was Nacetoacetylation of an amino acid-derived oxime.The nitrile oxide dipole would be prepared from the aldoxime via C-chlorination and 1,3-dehydrochlorination.Using valine as our test amino acid scaffold, the aldoxime 4a was prepared from N-tert-butyloxycarbonyl-S-valine methyl ester (DIBAL-H, toluene, -78 °C, 2 h; then NH2OH.HCl, NaOAc, EtOH aq, 70 °C, 10 min) in 89% yield as a 2:1 mixture of geometric isomers (Scheme 2).Acetoacetylation was attempted using 2,2,6-trimethyl-1,3-dioxin-4-one, the diketene-acetone adduct (PPTS, toluene reflux, 3 h). 7The results were inconsistent, with product 1 H NMR spectra sometimes showing loss of the tert-butyloxycarbonyl group and sometimes not.Samples retaining the tertbutyloxycarbonyl group had 1 H NMR spectra that implied acylation of the oxime N-atom rather than the amine function, with a signal for the proton NHBoc being observed at  4.62 (1H, d, J 9.6 Hz).More surprisingly, the NMR spectra suggested the reduced structure 5a, having an These results led us to avoid the oxime as a precursor functional group for the nitrile oxide dipole.Instead we targeted the dehydration of a primary nitro group as the route to the nitrile oxide.Our target thus became an N-protected S-2-methyl-1-nitromethyl-1-propanamine 6a or 6b (Scheme 3), which could be N-acetoacetylated.Thus S-valinol 7 (prepared by reduction of Svaline: TMSCl, NaBH4, THF, 0 °C, 24 h; 9 78%) was selected as starting material.A good leaving group was now needed at the alcohol function for substitution to generate a primary nitro group, but attempted toluene-4-sulfonylation (TsCl, DCM, 020 °C, 17 h) unsurprisingly afforded the N-sulfonylated compound 8 (100%) rather than the desired O-sulfonylated compound.Thus S-valinol 7 was N-protected as the tert-butyloxycarbonyl derivative 9a (Boc2O, Et3N, DCM, 020 °C, 17 h; 100%) and benzyloxycarbonyl derivative 9b (PhCH2OCOCl, NaHCO3, EtOAc-H2O, 020 °C, 17 h; 90%).The corresponding mesylate derivatives 10a and 10b were formed by a standard method (methanesulfonyl chloride, Et3N, DCM, 020 °C, 4 h) in crude yields of over 90%.Attempts to use these crude materials in a nucleophilic substitution with nitrite were unsuccessful (NaNO2, toluene or DMF, 060 °C, 17 h) with the alcohols 9a and 9b being recovered after aqueous workup.The mesylates were instead treated under Finkelstein conditions (NaI, acetone reflux, 17 h) in an attempt to generate the corresponding iodo compounds 11a and 11b, respectively.The N-tert-butyloxycarbonyl derivative 10a however, underwent protecting group cleavage to afford the cyclic carbamate 12; the Nbenzyloxycarbonyl compound 10b did give the iodo compound 11b (42%) along with some cyclic carbamate 12 (7%).It is assumed that anchimeric assistance by the N-protecting group carbonyl oxygen atom is followed by loss of a carbenium ion (to afford the cyclic carbamate) competing with iodide nucleophilic attack.In an alternative direct iodination from alcohol 9a, iodotriphenylphosphonium iodide was prepared (triphenylphosphine, iodine, DCM) to which imidazole and alcohol 9a was added and the mixture heated under reflux for 17 h to afford the iodide 11a (31% after chromatography).The yield and product purification was improved by replacing the triphenylphosphine with a polymer-supported triphenylphosphine, 10 to afford the iodides 11a (43%) and 11b (80%) from the corresponding alcohols 9a and 9b, respectively.Traces of the cyclised by-product 12 were sometimes observed in these reactions.The nitro substitution was undertaken with the iodo compounds, now using sodium nitrite in DMF (20 °C, 20 h) with the addition of phloroglucinol and urea to minimize the formation of nitrite ester. 11he corresponding N-tert-butyloxycarbonyl and N-benzyloxycarbonyl S-2-methyl-1nitromethylpropanamines 6a (64%) and 6b (70%) were isolated along with small recoveries of the corresponding alcohols 9a or 9b.Scheme 3. Preparation of protected S-2-methyl-1-nitromethylpropanamines 6. Reagents: i, TsCl, DCM, 020 °C; ii, For 9a: Boc2O, Et3N, DCM, 020 °C.For 9b: PhCH2OCOCl, NaHCO3, EtOAc-H2O, 020 °C; iii, MsCl, Et3N, DCM, 020 °C; iv, NaI, acetone reflux; v, PS-PPh3, I2; then imidazole; vi, NaNO2, DMF, phloroglucinol, urea, 20 °C.This route to optically active N-protected S-2-methyl-1-nitromethylpropanamines was complemented by a reported approach to the N-tert-butyloxycarbonyl R-compound 6c via the aza-Henry reaction (Scheme 4). 12Thus, sodium benzenesulfinate in water was added to tertbutyl carbamate (THF, 20 °C), followed by 2-methylpropanal and methanoic acid.After 17 h, the white precipitate was recrystallized to afford the sulfone 13 (75%).Reaction with nitromethane in the presence of potassium hydroxide (toluene, -7820 °C, 48 h) and catalytic benzylquininium chloride led to the nitro compound 6a (72%).Replacing potassium by cesium hydroxide led to incomplete reaction after 48 h, in contrast to one of the earlier reports.The nitro compound prepared by this route was formed highly stereoselectively, whereas the route from Svaline was stereospecific.Scheme 4. Alternative preparation of R-nitromethylpropanamine 6a.Reagents: i, BocNH2, PhSO2Na, THF aq., 20 °C; ii, MeNO2, KOH, benzylquininium chloride (12 mol %), toluene, -7820 °C, 48 h.The next step was N-acetoacetylation of the amino-nitro compound 6a, which was initially attempted as described earlier, using thermolysis of the diketene-acetone adduct (pyridinium toluene-4-sulfonate, toluene reflux, 3 h), but starting material was recovered.As an alternative acylating agent we employed S-tert-butyl 3-oxobutanthioate 14, itself generated by base treatment of 2-methylpropane-2-thiol (NaH, THF, -150 °C) followed by addition of diketene to the thiolate at -5 °C. 13Thus amino-nitro compound 6a and the thioester 14 were reacted in the presence of silver(I) trifluoroacetate (THF in the dark) but starting carbamate 6a was recovered. 14We reasoned this was due to reduced nucleophilicity of the carbamate N-atom, so we removed the tert-butyloxycarbonyl group (TFA, followed by 2M hydrochloric acid) to afford the stable amine hydrochloride salt 15 (Scheme 5).Acylation was completed using both protocols described above, to afford the acetoacetamide 16 (68% from diketene-acetone adduct; just 19% from thioester 14).The precursor functionality for generation of the required dipole (nitrile oxide from dehydration of the primary nitro group) and dipolarophile (enamine derived from the -ketoamide) is present in the acetoacetamide 16.The next step was planned to be the enamine formation, so the amide 16 was treated with pyrrolidine in toluene at reflux under Dean-Stark water removal conditions.After workup, a product was recovered (55%) but the 1 H NMR spectrum was missing the CH3 and CH signals expected for the enamine 17, although it did show signals for the pyrrolidine ring.The structure was revealed by an X-ray crystal structure determination to be the unexpected mixed urea 18 (Figure 2).The solid-state structure displays two independent molecules in the asymmetric unit, differing in the detailed conformation of the pyrrolidine sub-unit and linked via a strong N-H•••O hydrogen bond. 15cheme 5. Attempted enamine formation from acetoacetamide 16.Reagents: i, TFA, 20 °C, 4.5 h, then 2M HCl aq., 20 °C, 0.5 h; ii, Et3N, DCM, then 2,2,6-trimethyl-1,3-dioxin-4-one, pyrH +- OTs, toluene reflux; iii, Et3N, DCM, then MeCOCH2COSBu t 14, CF3CO2Ag, THF, -15 °C, 30 min; iv, pyrrolidine, toluene, Dean-Stark reflux.We propose the mechanism shown in Scheme 6 wherein the iminium ion formed en route to the desired enamine 17, or in equilibrium with it, is attacked by a second pyrrolidine molecule to promote a retro-Claisen reaction, affording the observed urea and presumably the pyrrolidine enamine of propanone as an unusual leaving group, although this was not isolated.Scheme 6. Possible mechanism for formation of mixed urea 18.
An alternative potential sequence towards an intramolecular dipolar cycloaddition would be nitrile oxide formation and then cycloaddition to the enol form of the -keto amide.With this in mind we treated the nitro compound 16 under several reported dehydration conditions: di-tertbutyl dicarbonate 16 or ethyl chloroformate 17 (Et3N, 0.1 mol equiv.4-DMAP, in DCM, acetonitrile or toluene at reflux), but in all cases starting material was recovered unchanged.Likewise, an attempt to form a silyl nitronate as dipole (TMSCl, Et3N, in THF or acetonitrile at reflux) returned starting nitro-amide 16.

Conclusions
In conclusion, based on these explorations we have not been able to form the isoxazole building block 2 via an intramolecular approach , and continued to focus instead on developing the pyrroloisoxazole strategy towards masked acyltetramic acids via the previously reported intermolecular sequence.

Experimental Section
General.Commercial dry solvents were used in all reactions except for light petroleum and EtOAc, distilled from CaCl2, and CH2Cl2 distilled over P2O5.THF was distilled from sodium and benzophenone.Light petroleum refers to the b.p. 40-60 °C fraction.Sodium hydride was obtained as 60% dispersion in oil and washed with light petroleum.Melting points were determined on a Leica Galen III hot stage apparatus. 1H (250 MHz), 1 H (400 MHz) and 13 C (100 MHz) NMR spectra were recorded on Bruker AC-250 or AC-400 spectrometers in CDCl3 solutions with Me4Si or (CD3)2SO as internal standard unless otherwise specified.Chemical shifts δ are given in parts per million (ppm) and 1 H coupling constants J in Hz, with multiplicities: s (singlet), d (doublet), t (triplet) and m (multiplet).Mass spectra were recorded on a JEOL SX102 spectrometer, or carried out by the EPSRC National Mass Spectrometry Service Centre (Swansea) or on a ZQ2000 spectrometer with Waters 600 series liquid handling system, dual wavelength UV and ELS detectors at Novartis, Horsham UK.MicroMass LCT was detected using a TOF spectrometer with Agilent 1100 series HPLC and Gilson 215 liquid handling; diode array and CAD detectors; Platform LC spectrometers with Agilent 1100 series HPLC diode array and ELS detectors; ThermoElectron LTQ Linear Quadrupole Ion Trap MS with ESI probe and APCI source were used for MS/MS; ThermoElectron DSQ MS with TRACE GC, at Novartis.GCMS was carried out on a Fisons 8000 series instrument using a 15 m x 0.25 mm DB-5 column and an EI low resolution MS at Novartis.IR spectra were recorded on a Perkin-Elmer Paragon 1000 FT-IR spectrophotometer on NaCl plates, in the range 4000-600 cm -1 .Elemental analyses were determined on a Perkin Elmer 2400 CHN Elemental Analyser in conjunction with a Perkin Elmer AD-4 Autobalance, or on a LECO CHNS-932 Analyser, at Novartis.TLC using silica gel as absorbent was carried out on aluminium backed plates coated with silica gel (Merck Kieselgel 60 F254), and TLC using alumina as absorbent was carried out on aluminium backed plates coated with neutral aluminium oxide (Merck 150 F254, Type T).Silica gel (Merck Kieselgel 60 H silica) was used for column chromatography unless otherwise specified.Column chromatography using alumina was carried out with Aldrich aluminium oxide, activated neutral, Brockmann 1, STD Grade, 150 mesh sizes.Preparative TLC was carried out using aluminium oxide (Merck 60 F254, Type E).