Multi component coupling reactions of N -acetyl-2-azetine

N -Acetyl-2-azetine undergoes Lewis acid catalysed formal [4+2]-cycloaddition with imines derived from aromatic amines to initially give an approximately 1:1 mixture of exo-endo - diastereoisomeric 1-(2a,3,4,8b-tetrahydro-2 H -1,4-diaza-cyclobuta[ a ]naphthalen-1-yl)-ethanone cycloadducts which were detected by proton NMR spectroscopy. These products, which were too unstable to isolate, and characterise, reacted further with aromatic amines to give 2,3,4-trisubstituted tetrahydroquinolines in good to excellent yield, predominantly as a single diastereoisomer, with the minor diastereoisomer converting to the major diastereoisomer on silica. The cycloaddition was irreversible and a mechanism is presented for the formation of the major diastereoisomer from the mixture of diastereoisomeric intermediates. A range of conditions is described for converting the 2,3,4-trisubsitituted tetrahydroquinolines into 2,3-disubstituted quinolines.


Introduction
The imino-Diels-Alder reaction, originally described by Povarov, 1 involves coupling of imines, derived from aromatic amines, or their surrogates, with electron rich alkenes and has emerged as a powerful tool for the synthesis of tetrahydroquinolines, and the subject has been reviewed several times. 2,3The original reaction was catalysed by boron trifluoride etherate 4,5 but more recently it was demonstrated that a wide range of Lewis acids, [6][7][8][9] or protic acids [10][11][12] also catalyse this process.No catalyst is required if the reaction is carried out in an ionic liquid 13 or 2,2,2trifluoroethanol. 14There does not appear to be one generic mechanism to adequately explain all reactions in this class, with some substrates appearing to react by a formal concerted [4+2] cycloaddition 15 whilst others appear to react via a stepwise process with discrete ionic intermediates. 16,17Scheme 1. Elimination-oxidation approach to 2,3-disubstituted quinolines.
8][29] It was envisaged that if this compound participated in imino-Diels-Alder reactions, with an aromatic imine, then the products would be much more prone to participate in elimination-oxidation reactions regardless of the substituent at C-2.The initial results of imino-Diels-Alder reactions of N-acetyl-2-azetine, with imines derived from aromatic amines, have been previously reported in preliminary form 30,31 and we now present full details of this chemistry.

Results and Discussion
N-Acetyl-2-azetine 6 was prepared by the literature procedure 28 as outlined in Scheme 2. Although this compound was previously known, no detailed experimental procedures for the final two steps were reported.Alcohol 4 was converted into N-acetyl-2-azetine 6 uneventfully in 32% overall yield by formation of the mesylate followed by elimination using potassium tertiary butoxide as base.The moderate overall yield can be attributed to the instability of the azetine and water solubility of azetine and mesylate intermediate 5.This compound proved to be remarkably stable for a cyclic enamide, and could be stored indefinitely at -5 o C in a refrigerator.

Scheme 2. Preparation of N-acetyl-2-azetine.
Initially the imino Diels-Alder reaction of N-acetyl-2-azetine 6 with imine 7a, using yttrium triflate as Lewis acid, proved extremely problematic.Although both starting materials were rapidly consumed, proton NMR analysis of the crude reaction mixture indicated that the reaction was not clean and that many products had formed from which quinoline 11a could be sometimes isolated.The reproducibility with respect to yield, was poor and ranged from 0-39%.In order to identify the troublesome step, in this multi step coupling sequence, the reaction was carried out in d-3-acetonitrile with yttrium triflate as catalyst and directly monitored by proton NMR spectroscopy.Gratifyingly, the imino Diels-Alder reaction proceeded rapidly, cleanly, and completely regioselectively at room temperature.After one hour all the N-acetyl-2-azetidine 6 was consumed and a 55:45 mixture of exo-endo-diastereoisomeric 1-(2a,3,4,8b-tetrahydro-2H-1,4-diaza-cyclobuta[a]naphthalen-1-yl)-ethanones 8a were present, Scheme 3.These compounds were readily identified by the similarity of spectra to the more stable exo/endohexahydropyrolloquinolines 2b, previously isolated and characterised, 21 with ring junction proton H8b coming into resonance at the distinctive positions of δ 5.38 and 5.20 as doublets for both isomers.On standing overnight, and recording the NMR spectrum, all peaks for tricyclic compound 8 had gone and new extremely broad peaks were evident indicating that the material had possibly polymerised.On standing for several days, or heating for four hours no quinoline product 11a was detected by proton NMR spectroscopy, strongly suggesting that the quinolines could be a product of workup.This raises the interesting question as to what the structure of the suspected polymeric material was.One interesting idea was that 1-(3-phenyl-2a,3,4,8btetrahydro-2H-1,4-diaza-cyclobuta[a]naphthalen-1-yl)-ethanone 8a was undergoing a ring opening reaction, facilitated by the tetrahydroquinoline nitrogen, to iminium salt 9 which then reacted with additional 8a to give a dimer 10, which then oligomerised and ultimately polymerised by a repetition of the aforementioned events.To test this theory, an additional NMR experiment was run which contained aromatic amine.Again the exo/endo-cycloadducts 8a, were rapidly formed inside one hour at room temperature, with the primary aromatic amine having no detrimental effect on the catalyst, and these cycloadducts slowly reacted with the aromatic amine, typically over twelve hours, and gave the 2,3,4-tetrahydroquinolines diastereoisomers 12a and 13a as the sole reaction products.With conditions worked out on how to perform the multicomponent coupling reaction, a range of imines 7a-f were reacted with N-acetyl-2-azetine and aromatic amine, from which the imine was derived, and the results are summarized in Table 1.In most cases the reaction proceeded smoothly and gave a crude product as a mixture of diastereoisomers 12 and 13.However, unless the silica gel used for chromatography was prewashed with triethylamine, only the major diastereoisomer 12 could be isolated.Clearly these diastereoisomers are interconverting.The low yield for entry f probably reflects the instability of the imine and the reaction failed completely for imines derived from aliphatic aldehydes with only intractable material produced.Assigning the relative stereochemistry of the major diastereoisomer 12 proved difficult, by proton NMR spectroscopy, as the preferred conformation of the saturated ring was also unknown.This problem was solved by single crystal X-ray crystallography on 12d 32 and the other members of the series by comparison of their NMR spectra with that of 12d.Interestingly, this molecule adopts a conformation which puts the two substituents at C-3 and C-4 transdiaxial, presumably to minimise 1,3-allylic strain of H-5 with the aniline substituent at C-4, 33 with the phenyl substituent at C-2 equatorial.The molecule adopts the same conformation in solution with a coupling constant J H3-H4 of less than the line width, 1.4 Hz, which is consistent with both H-3 and H-4 being trans-diequatorial.Although the minor diastereoisomers 13a-f could be detected by proton NMR spectroscopy, in the crude reaction mixtures, their isolation and purification proved to be extremely challenging.Using standard flash chromatography these minor isomers were never isolated, probably due to an equilibration of 13 to 12 on the column.However, in one case (entry a), the column was pre-washed with triethylamine, prior to chromatography, and 13a was isolated in 3% yield.It also proved difficult to determine the relative stereochemistry of this diastereoisomer by proton NMR spectroscopy.The coupling constants J H2-H3 and J H3-H4 , 2.5 Hz and less than 1 Hz respectively, were very similar to that of the major diastereoisomer, 2.8 Hz and less than 1.4 Hz.However, saturation at H-2 gave a 4.4% and 3.7% nOe to H-3 and H-4 respectively, supporting the proposed structure of 13a, and indicating that both H-2 and H-4 had a cis-1,3-diaxial relationship.In the case of imine 7c the isolated yield of product was low, 28%.The reason for this was that the intermediate cycloadduct 8c was not very soluble in acetonitrile and precipitated during the course of the reaction.It proved possible to filter this solid and it was soluble enough in d-3acetonitrile for a proton NMR spectrum to be recorded.When p-methoxyaniline was added to the NMR sample and the reaction monitored by proton NMR spectroscopy, nothing happened, with cycloadduct 8c being stable under these reaction conditions.When one crystal of yttrium triflate was added to the NMR sample the reaction proceeded and gave a 99:1 mixture of diastereoisomers 12c:13c.These results prove that the ring opening reaction of 8c with aromatic amine was Lewis acid catalysed.When imine 7c was reacted with N-acetyl-2-azetidine and aniline, a complex mixture of products resulted due to imine equilibration, giving rise to two different imines and amines leading to a permutation of four different products plus their minor diastereoisomers.However, when precipitated cycloadduct 8c was reacted with aniline under standard reaction conditions this gave 12g as the sole reaction product.These findings prove that the imino Diels-Alder reaction giving rise to 8c is irreversible.This then raises the interesting question as to how a single diastereoisomeric tetrahydroquinoline product, with respect to centres C-2 and C-3, is formed from a 55:45 mixture of exo-endo diastereoisomers 8. Chiral centres C-2 or C-3, or both, must be epimerising at some stage in the subsequent sequence and the most likely place for this to happen is at the stage of intermediate 9, prior to reaction with aromatic amine.
With tetrahydroquinolines 12 at hand, attempts were made to achieve the initial goal and convert these to 2,3-disubstituted quinolines, in a reproducible manner.This transformation was readily achieved by stirring with acetic acid, heating with yttrium triflate, or oxidation with DDQ and the results are summarised in Scheme 4. Presumably, the reason for the success in this sequence is that monomeric tetrahydroquinolines 12b, c, e are more reactive than their polymeric counterparts that would have resulted if the aromatic amine was omitted from the initial cycloaddition.The second reaction, heating with yttrium triflate, was particularly attractive, because once the initial multicomponent coupling is complete, heating the mixture initiates the eliminationoxidation sequence turning this into a one pot procedure.The results of these one pot couplings to give 2,3-disubstituted quinolines are summarised in Scheme 5.Although it is known that imino Diels-Alder reactions of meta-substituted aromatic imines can be completely regioselective, 34,35 in the the case of imine 7h the regioselectivity for the cycloaddition was poor.
In conclusion, we have demonstrated that the multicomponent reaction comprising of an imino Diels-Alder reaction, fragmentation-oxidation to give 2,3-disusbtsituted quinolines, is facilitated by the addition of aromatic amines, which functions by preventing unstable intermediates from polymerising.

Experimental Section
General Procedures.Melting points were recorded using a Kofler hot stage apparatus and are uncorrected.I.R spectra were recorded on a Perkin-Elmer Model 983G instrument coupled to a Perkin-Elmer 3700 Data Station as potassium bromide (KBr) disks, or films (liquids). 1H nuclear magnetic resonance (NMR) spectra were recorded at 300MHz using Bruker DPX 300 and at 500MHz using Bruker DRX 500 NMR spectrometers.Chemical shifts are given in parts per million (δ) down field from tetramethylsilane as internal standard and coupling constants are given in Hertz.Unless otherwise stated, deuteriochloroform was used as solvent.Spectra splitting patterns are designated as s: singlet, d: doublet, t: triplet, q: quartet, m: multiplet and br: broad.Mass spectra were recorded using Double Focusing Triple Sector VG Auto Spec and accurate molecular masses were determined by the peak matching method using perfluorokerosene as standard reference and were accurate to within +/-0.006 a.m.u.Analytical tlc was carried out on Merck Kielselgel 60 254 plates and the spots visualised using a Hanovia Chromatolite uv lamp.Flash chromatography was effected using Merck Kielselgel 60 (230-400 mesh).Imines 7a-f [36][37][38][39] were prepared by condensation of aldehyde (2.6 mmol) with amine (2.6 mmol) in benzene (30 mL) with azeotropic removal of water.