Dehydrogenation–halogenation of a 4,5,6,7-tetrahydroisoxazolo[4,3-c ]pyridin-4-one to provide a scaffold for acylpyridones

The dehydrogenation and halogenation of 3-methyl-4,5,6,7-tetrahydroisoxazolo[4,3-c ]pyridin-4-one has been investigated to provide a suitable 7-halo-4,5-dihydroisoxazolo[4,3-c ]pyridine-4-one to act as a masked scaffold for the acylpyridone natural products and analogues.


Introduction
The 3-acyl-4-hydroxypyridin-2-one moiety 1 (Figure 1) is the common structural unit of a family of natural products with a range of interesting biological activities. 1 Examples are the pigments tenellin 2a and bassianin 2b from insect pathogenic fungus Beauveria bassiana, 2 pyridovericin 2c 3 (a tyrosine kinase inhibitor) and the elfamycin antibiotics. 4Farinosone A 2d, isolated from Paecilomyces farinosus, induces and enhances neurite outgrowth in the PC-12 cell line, although it is not clear whether the pyridines in general display neuritogenic properties. 5The 5-substituent is commonly an aryl ring or derivative (biogenesis from tyrosine), and the 3-acyl substituent is commonly polyenoyl The biosynthesis of tenellin and bassianin in Beauveria bassiana has recently been studied in detail using genetic techniques, and been shown to involve conversion from an acyltetramic acid via oxidative ring expansion. 6s part of an ongoing programme of synthesis towards metabolites and analogues containing the enolised heterocyclic tricarbonyl motif 3, [7][8][9][10] we have reported a nitrile oxide dipolar cycloaddition strategy which affords 3-methyl-4,5-dihydroisoxazolo [4,3-c]pyridin-4-one 4 as a 2 nd generation masked non-polar scaffold for the 3-acyl-4-hydroxypyridin-2-one nucleus. 8,9We have reported two syntheses of building block 4, from 3-aminopropanoic acid (β-alanine) or from 2,3-diaminopropanoic acid; 10 the more economical starting material for the synthesis is βalanine and a key intermediate is the tetrahydroisoxazolopyridone 5. Furthermore, elaboration of dihydroisoxazolopyridone 4 at C-7 to append the aromatic rings found at C-5 of many of the pyridone natural products is an important step in our sequence. 8,10Two key conversions for our overall strategy are thus the 6,7-desaturation of lactam 5 to give scaffold 4, and generation of a suitable alkenyl halide at C-7 to facilitate Pd-mediated cross-couplings.We report here in detail our investigations to establish reliable methodology for these two conversions.

Results and Discussion
Having in hand the tetrahydroisoxazolopyridone 5, 10 we needed to introduce C-6,7 unsaturation to generate the dehydro compound 4. Our initial attempts were using dehydrogenation reagents, thus we investigated the following methods: DDQ in 1,4-dioxane at reflux, and also in the presence of N,O-bis(trimethylsilyl)trifluoroacetamide as this reagent had been shown to improve the dehydrogenation of steroidal lactams by facilitating the breakdown of the intermediate quinone-lactam complex; 12 10% Pd-C in 1,4-dioxane at reflux, and in diphenyl ether at 200 °C; 13 Pb(OAc)4 in acetic acid. 14In all except the latter case, none of the desired pyridone was detected, and only in the latter case was a trace of the dehydrogenation product observed spectroscopically (NMR, MS), although not in synthetically useful quantities.The next approach to be investigated was N-chlorination to give 6 followed by elimination of HCl, on the presumption that an imine so-produced would tautomerize to the required enamine 4 (Scheme 1).Our initial studies used a reported procedure: potassium hexamethyldisilazide (KHMDS) as base, and a positive chlorine source, N-chlorosuccinimide (NCS; THF, 20 °C). 15fter purification, some of the desired N-chlorolactam 6 was observed, contaminated by NCS, but to our surprise the major product was the 3-trichloromethyl compound 7 (9%) (Scheme 2). 16he formation of the C-chloro compound 7 can be rationalized by competitive deprotonation at C-3(Me) (which we have observed separately 10 ) and chlorination either directly by NCS, or with the N-chlorolactam 6 as an alternative chlorinating agent.Once mono-chlorination has taken place, successive deprotonation-chlorination is favoured by the existing chloro-substituent, in a sequence mirroring the haloform reaction. 17NCS used in the absence of base did afford the Nchloro compound 6 but contaminated with NCS and succinimide, such that a pure sample could not be isolated.In an alternative N-chlorination protocol, treatment of the pyrroloisoxazole 5 with freshly prepared t-butyl hypochlorite (MeOH, 0 °C) 18 in the absence of direct light did provide the N-chlorolactam 6 in good yield (94%).Scheme 1. Plan for C-6,7 desaturation of lactam 5 via N-chlorination and dehydrochlorination.Scheme 2. N-Chlorination & photolysis, and base mediated reactions of lactam 5. Reagents: i, KHMDS, NCS, THF, 20 °C; ii, t-BuOCl, MeOH, 0 °C; iii, DBU, benzene, 20 °C; iv, BuLi, THF, -78 °C, MeI; v, BuLi, THF, -78 °C, H2C=CHCH2Br; vi, h, Hanovia Hg-lamp, MeOH.
We next attempted elimination of HCl from 6 using DBU as base (benzene, 20 °C), but the products observed were the N-H lactam 5 and the trichloromethylisoxazolopyridone 7 (Scheme 2).It is probable that the N-chlorolactam 6 acted as a chlorinating agent for anions formed at the C-3 substituent, as proposed above.The competitive deprotonation at C-3(Me) of N-H lactam 5 was further illustrated by alkylation experiments.In the first, methylation was investigated (1 mole equiv BuLi, THF, -78 °C; 1 mol equiv MeI) to afford a mixture containing unchanged lactam 5, the C-methyl compound 8a and the C,N-dimethyl-lactam 9, that was not separated (Scheme 2).An allylation experiment (excess BuLi, THF, -78 °C; 1.5 mol equiv prop-2-enyl bromide) afforded recovered 5 and C-allyl compound 8b (20%), identified by the presence of an NH signal (δ 5.82, 1H, br s) in the 1 H NMR spectrum along with the appropriate alkene proton signals, and the absence of the C-3(Me) signal.It was therefore deemed necessary to avoid basic conditions for dehydrohalogenation of chlorolactam 6, to avoid reaction at C-3(Me).
Our attention was therefore drawn to a report of the photolysis of N-chlorolactams to afford N-(α-methoxyalkyl)lactams. 19 After some experimentation, a solution of N-chloro compound 6 in MeOH was degassed (N2 flow, 10 min) and irradiated with a medium-pressure mercury lamp for 1 h to afford the desired dihydropyrroloisoxazole 4 in an optimum yield of 80%, along with some recovered N-H compound 5 (Scheme 2).In some experiments the yield of 4 was lower but more dechlorinated material was returned so that the yield based on recovered N-H lactam was maintained around 80%.Other light sources were less efficient; for example, use of two home Solaria UV lamps gave 4 in just 40% (60% based on recovered 5).The mechanism for the dehydrochlorination of 5 is proposed to be via a radical chain process (Scheme 3): 19 homolysis of the N-Cl bond followed by H-atom abstraction from C-6 of 6, would leave a carbon radical at C-6 which, if it fragments to expel a chlorine atom, would generate an imine that could simply tautomerize to the stable acyl enamine form 4. The accompanying production of the N-H lactam 5 is proposed to arise from H-abstraction from the MeOH solvent; the HCl so-formed may also cause acid-promoted decomposition of 6 with generation of chlorine.The structure of target dihydroisoxazolopyridone 4 was confirmed by an X-ray crystal structure determination (Figure 2); 20 the solid-state structure shows H-bonded pyridone dimers with some - stacking.Having secured the desired overall dehydrogenation of tetrahydroisoxazolopyridone 5, we required a substituent at C-7 to enable cross-coupling to the aromatic residues commonly found in the acylpyridone natural products.This proved straightforward (Scheme 4); treatment of 4 with iodine monochloride (CH2Cl2-MeOH, 20 °C, 16 h) provided 7-iodo compound 10 (70%). 21heme 4. C-7 Iodination of unsaturated lactam 4. Reagents: i, ICl, CH2Cl2-MeOH, 20 °C, 16 h.Meanwhile, we had identified additional trace products in the photolysis reaction of Nchlorolactam 6, as the 7-chloroisoxazolopyridone 11 and 7-chloro-6-methoxy-adduct 12a (which co-eluted on column chromatography), and pseudo-dimer 13.The 7-chloro derivatives are believed to arise by interaction of major product 4 with chlorine formed during the radical process, and either proton loss or MeOH addition to an imine intermediate in the chlorination.The dimer is attributed to chlorination of MeOH solvent to provide methanal or an equivalent.The identification of 7-chloroisoxazolopyridone 11 prompted us to investigate further, as the direct one-step formation of a 7-haloisoxazolopyridone from chlorolactam 6 would shorten the synthetic sequence.During investigation of alternative light sources, a tungsten UV lamp was employed (Scheme 5).After complete disappearance of starting material 6 (TLC), only low yields of the dihydroisoxazolopyridone 4 were obtained and the N-H lactam 5 was the major compound returned (up to 60%).However, the main product of interest was the 7chloroisoxazolopyridone 11 (optimum yield 40%), sometimes isolated with the 6,7-dichloro adduct 12b; in these instances the mixture could be converted on base treatment (K2CO3 aq, 50 °C) to 7-chlorodihydroisoxazolopyridone 11 by an HCl elimination.As outlined above, we propose the 7-chloro compound 11 to arise from chlorination of the dihydroisoxazolopyridone 4. In an attempt to increase the yield of 11 we thus added a potential chlorine source (NCS, 1 mol equiv) to the photolysis reaction mixture but no change to the outcome was observed.For the planned cross-coupling of 7-haloisoxazolopyridones, it was anticipated that a 7bromo derivative would be more effective than the above 7-chloro compound.We thus investigated access to 7-bromodihydroisoxazolopyridone 14.As the photolysis route was believed to proceed via a radical mechanism, we attempted bromination of lactam 5 using Nbromosuccinimide (NBS) under radical initiation.After less successful attempts using 1,1azobis(cyclohexanecarbonitrile) (ACCN) initiation in cyclohexane at reflux, and azobis(isobutyronitrile) (AIBN) in cyclohexane or toluene at reflux, we found that AIBN in carbon tetrachloride at reflux afforded 7-bromo compound 14 in 30% yield; changing to tbutanol solvent at reflux, and adding AIBN in portions every hour for 5 h, afforded an improved 35% yield (Scheme 6).As we did not find any evidence for dihydroisoxazolopyridone 4 in these experiments, we propose that this conversion proceeds as a 'benzylic' bromination via H-atom abstraction at C-7, 22 and that the 7-bromo compound so-formed undergoes a second Habstraction/bromination followed by HBr elimination, rather than elimination to form enamine 4 and a polar bromination.Attempts to extend this protocol to radical iodination of 5 using Niodosuccinimide (NIS) were unsuccessful, producing only traces of iodo compound 10.Scheme 6. Radical bromination of lactam 5. Reagents: i, NBS, AIBN, t-butanol reflux, 5 h.
In one further approach, we very briefly experimented with direct arylation of the nucleophilic acyl enamine 4 via a presumed electrophilic substitution [Pd(PPh3)4, Ag2CO3, water, 60 °C, 24 h] to afford the 7-phenylisoxazolopyridone 15a but in low yield (12%). 25 This process would potentially avoid the iodination step, but we have not optimized further.

Conclusions
We can therefore conclude that 7-iododihydroisoxazolopyridone 10 was both the most accessible 7-halo derivative from intermediate 5, and (as expected) the best substrate for cross-coupling, and have proceeded using this sequence for the key conversions in our programme of synthesis towards the acylpyridones.

Experimental Section
General. 1 H NMR spectra were recorded using Bruker DPX 400 MHz or Varian 400 MHz spectrometers, and 13 C NMR spectra the Bruker instrument operating at 100.62 MHz.Chemical shift, , values are given in ppm, coupling constants in Hertz and multiplets denoted by: s, singlet; d, doublet; t, triplet; q, quartet and m, multiplet.All spectra were recorded using tetramethylsilane (TMS) as the internal reference in CDCl3 or d6-DMSO solvent.IR spectra were determined using a Perkin Elmer FT-IR Paragon 1000 spectrometer or a Perkin Elmer Spectrum One spectrometer and recorded in the range 4000-600 cm -1 .Mass spectra were recorded on a Joel SX-102 spectrometer (FAB and EI) and the Thermo Exactive (Orbi) accurate mass spectrometer (ESI), fitted with a Triversa Advion Nanomate sample delivery system using nano-ESI of MeOH or MeOH-AcOH (99:1 w/w).GC-MS used was Fisons GC 800 with autosampler, Fisons mass lab MD 800 EI + and DB5-MS 30 m column.Melting points (m.p.; °C) were determined using an Electrothermal-IA 9100 and are uncorrected.TLC using silica gel as the absorbent was carried out with aluminium backed plates; column chromatography using silica gel was carried out with Zeoprep 60 HYD 40-63 Micron silica.HPLC analysis was conducted in a Waters Fraction Lynx system comprising a 2767 injector/collector with a 2525 gradient pump, CFO, 2996 photodiode array, 2420 ELSD and Micromass ZQ2000 equipped with a Waters XBridge dC18 column (column length 20 mm, internal diameter of column 3 mm, particle size 3.5 micron).The analysis was conducted using a three minute run time using H2O with 10mM NH4OAc and CH3CN.Distillation of reagents was carried out at atmospheric pressure unless otherwise stated.Solvents were distilled before use.Light petroleum (b.p. 40-60 °C) and EtOAc were distilled from CaCl 2 , CH 2 Cl 2 from CaH 2 ; THF was freshly distilled from sodium and benzophenone under an atmosphere of nitrogen, MeOH from Mg(OMe)2.