Fused bicyclic piperidines and dihydropyridines by dearomatising cyclisation of the enolates of nicotinyl-substituted esters and ketones

Summary The silyl enol ether derivatives of ketones or esters tethered by a hydrocarbon or ether linkage to the 3-position of a pyridine ring undergo dearomatising nucleophilic attack on the ring once it is activated (as an acylpyridinium species) by the addition of methyl chloroformate. The bicyclic dihydropyridine products are in some cases unstable, but may be isolated after hydrogenation as fused bicyclic piperidines.

In this paper we report the results of cyclising the enolates of ester and ketones tethered to a nicotinyl nucleus via chains which do not incorporate an amide linkage. The starting materials for these cyclisations do not benefit from the favourable conformational disposition of amides 1 and 3, making the reactions more challenging. Likewise, the products are evidently less stable than those produced by the reactions in Scheme 1a, but nonetheless they allow new, partially saturated "drug-like" heterocyclic systems to be formed.

Results and Discussion
Formation of a carbocyclic ring by dearomatising cyclisation The study was initiated with the synthesis of the δ-nicotinyl ketone 7 as illustrated in Scheme 2. Ethyl benzoylacetate was alkylated with 3-(3-iodopropyl)pyridine 5 and the product 6 hydrolysed and decarboxylated to yield the pyridine 7 in moderate yield. On treatment of 7 with LDA in THF at −78 °C, followed by trapping with methyl chloroformate, a yield of 40% of the bicyclic hexahydroisoquinoline 8 was obtained (Scheme 3), which even after extensive experimentation could not be improved. Lack of crystallinity and overlapping 1 H NMR signals prevented us from confirming the relative stereochemistry, and the assignment shown in Scheme 3 is on the basis that the benzoyl group of 8 is likely to lie on the exo face of the azadecalin system. We assume, in line with previous results [39], that cyclisation occurs only after the addition of the electrophilic trap (which, precedent suggests, attacks the pyridine lone pair and activates the ring as an acylpyridinium species even in the presence of the lithium enolate). Attempts to use bases with a sodium or potassium counter ion led instead to a high yield of the Claisen product 6 (R = Me), presumably because the sodium and potassium enolates are more reactive than the lithium enolate and compete too well with N-acylation.
We surmised that effective cyclisation onto the acylpyridinium species, avoiding the N-vs. C-acylation problem, might be made possible by decreasing the reactivity of the enolate still further, transforming it into a silyl enol ether 9. Silyl enol ethers and ketene acetals are known to add effectively to pyridinium species in an intermolecular manner [42][43][44][45][46]. Thus 7 was converted to silyl enol ether 9 in excellent yield under standard conditions. A single geometrical isomer was obtained, presumably Z as shown. On treatment with methyl chloroformate, enol ether 9 cyclised to yield 8 again as a single diastereoisomer but in a greatly improved yield of 93%. The strategy of using a less nucleophilic specific enolate equivalent is clearly an effective way of improving selectivity, allowing the chloroformate to activate the pyridine without competing attack by the enolate.
Next we extended the reaction to the cyclisation of a δ-nicotinyl butyrate ester 12 encouraged by the observations of Onaka [47], who demonstrated that silyl ketene acetals can be added (in an intermolecular fashion) to electron deficient pyridines in the presence of trimethylsilyl triflate, tetrabutylammonium fluoride or a montmorillonite clay.
The cyclisation precursor was synthesised by using the procedure of Hayashi [48] employing a Horner-Wadsworth-Emmons olefination between nicotinaldehyde and phosphonate 10. The resulting mixture of dienes 11 gave ester 12 after hydrogenation (Scheme 4). It proved challenging to isolate cleanly the silyl ketene acetal derived from 12, so instead we decided to form and cyclise the silyl derivative in a single pot. Thus, ester 12 was added to LDA at −78 °C, and the enolate quenched with trimethylsilyl chloride. After 15 min methyl chloroformate was added and the solution warmed to room temperature. Complete consumption of starting material (by TLC) was accompanied by the appearance of a single less polar spot (R f 0.77; EtOAc-petroleum ether 1:1). 1 H NMR analysis of the crude product after rapid work-up showed two significant sets of new signals at 6.55-6.80 ppm (2H) and 4.65-5.10 ppm (1H) consistent with the dihydropyridine protons of the expected dearomatised product 14 (Scheme 5). However, in contrast with the clean spectra and dearomatised product 8 derived from ketone 7, duplication of many of the signals in the crude 1 H NMR spectrum of 14 suggested the existence of either a mixture of diastereoisomers or rotamers caused by restricted rotation of the carbamate group. No dihydropyridine was isolable from this mixture by flash chromatography, probably due to rapid re-aromatisation. However, immediate hydrogenation at ambient pressure using the conditions developed by Arnott for related 3,4-fused dihydropyridines [39] gave 15 in 45% yield after chromatography as an inseparable mixture of two diastereoisomers in a Unfortunately, again the lack of crystallinity and the large number of overlapping signals in the 1 H NMR spectrum frustrated an unequivocal assignment of the stereochemistry. However, hydrogenation of related fused dihydropyridines has always led to cis stereochemistry at the ring junction [32,39]. The consequent expected axial-equatorial relationship between the protons at the ring junction is supported by a coupling constant of 4.2 Hz between these protons in 15 (Figure 1) in the major product diastereoisomer. The corresponding 12.9 Hz coupling to the proton α to the ester group is consistent with adoption of an exo-equatorial orientation by this substituent.

Formation of a tetrahydrofuran by dearomatising cyclisation
Encouraged by the successful formation of carbocyclic rings in dearomatising cyclisations of nicotinyl ketones and esters, we moved to extend the reaction to the analogous formation of tetrahydofuranyl esters by cyclisation of starting materials incorporating an enolate nucleophile and a nicotinyl electrophile tethered through an ether linkage. Alkylation of 3-hydroxymethylpyridine by t-butyl bromoacetate 17a or bromopropionate 17b suffered from competing N-alkylation but returned acceptable yields of the esters 18a and 18b (Scheme 6). As with 13, we anticipated that the silyl ketene acetal derivatives 19 would be challenging to isolate, so both starting esters 18a and 18b were treated with LDA and Me 3 SiCl followed by methyl chloroformate (Scheme 7). As with 14, re-aromatisation was fast and the crude products 19 were therefore hydrogenated at atmospheric pressure to give 20a in up to 32% yield from 18a and 20b in up to 35% yield from 18b. The instability of the two non-isolable intermediates meant however that these yields were not consistently reproducible and yields around 25% were more commonly observed. However, scrupulous avoidance of contact with oxygen before the hydrogenation step improved the yield of 20a to 41%. Attempted cyclisation without formation of the silyl enol ether (i.e. omitting Me 3 SiCl) led to a complex mixture of products. In both cases the cyclic products were obtained as single diastereoisomers, indicating a diastereoselective cyclisation and a face-selective hydrogenation. An nOe experiment on cyclic ether 20b, irradiating the 7a ring junction proton, showed nOe enhancements of protons 3a, 6 (1H) and 7 (1H) (Figure 2). This result is consistent with a cis-fused ring junction. A lack of conclusive nOes prevented determination of the stereochemistry at the ester-bearing centres of 20a or 20b. However, a similar cyclisation with an amide tether [39] had resulted in an endo-orientated ester substituent, and the stereochemistries of 20 are accordingly shown with the ester orientated endo.

Conclusion
Tethered ketone or ester enolate nucleophiles undergo dearomatising attack on a pyridine ring to yield bicyclic products. Yields are greatest if the enolate is first stabilised as a silyl enol ether, presumably because acylation of the pyridine ring to give the electrophilic acylpyridinium species is cleaner. The bicyclic dihydropyridine products are unstable towards re-aromatisation, but can be isolated in moderate to excellent yield if they are hydrogenated in situ, especially when oxygen is excluded prior to and during the hydrogenation.

5-Ethyl 2-methyl octahydroisoquinoline-2,5-(1H) dicarboxylate (15)
n-Butyllithium (0.36 mL of a 1.8 M solution in hexane) was added to a solution of diisopropylamine (0.11 mL, 0.75 mmol) in THF (15 mL) at 0 °C and the mixture stirred for 15 min before cooling to −78 °C. A solution of ester 12 (0.104 g, 0.5 mmol) in THF (5 mL) and then trimethylsilyl chloride (0.10 mL, 0.75 mmol) were added using a cannula. The solution was stirred at −78 °C for 15 min, methyl chloroformate (0.19 mL, 2.5 mmol) was added and the solution warmed to room temperature. The solution was rapidly added to a saturated sodium hydrogen carbonate solution (30 mL), extracted with EtOAc (2 × 30 mL), dried (MgSO 4 ), and concentrated to yield an oil. The crude oil was dissolved in isopropanol (6 mL), and 10% palladium/carbon (0.053 g, 0.05 mmol) was added and the suspension immediately placed under a hydrogen atmosphere. The suspension was warmed to 60 °C for 50 h, filtered through celite and evaporated under reduced pressure. The residue was purified by flash chromatography (SiO 2 ; EtOAc-petroleum ether 1:19 to 1:1) to yield the title compound (0.061 g, 45%) as a yellow oil which was approximately a 6:1 mixture of diastereoisomers; R f (EtOAc-petroleum ether 1:1) 0.45; IR (film) ν max (cm