Sigmatropic Rearrangements of 2,4-Dinitrophenyl Oximes

Reaction of 6-amino-1,3-dimethyluracil (14) with the 2,4-dinitrophenyl oxime (21) of iodoacetone led to initial condensation followed by an unexpected [3,3] sigmatropic rearrangement, followed in turn by ring closure to give 5-(2-hydroxy-3,5-dinitrophenyl)-1,3,6-trimethylpyrrolo[2,3-d ]pyrimidine-2,4(1 H ,3 H )-dione (22) . 6-Methylamino-1,3-dimethyluracil under the same conditions underwent a similar sigmatropic rearrangement, but the product cyclised instead to give 1,3-dimethyl-6-methylamino-5-(2-methyl-5,7-dinitro-3-benzo[ b ]furanyl)pyrimidine-2,4(1 H ,3 H )-dione (32) . A third type of product, 4-amino-6-(2-hydroxy-3,5-dinitrophenyl)-5-methyl-2-phenylfuro [2,3-d ]pyrimidine (35) , was obtained when 6-amino-2-phenyl-4(3 H )-pyrimidinone (33) reacted with the 2,4-dinitrophenyl oxime (19) of chloroacetone in presence of potassium iodide. Mechanisms are suggested for these unusual reactions.


Introduction
It was shown recently by Narasaka [1][2][3][4] and co-workers that 2,4-dinitrophenyl oximes can serve as a source of electrophilic nitrogen, and that in suitable cases intramolecular electrophilic cyclisation of nitrogen on to a benzene ring may lead ultimately to the formation of quinolines.It was shown, 1 for example, that 2-methyl-8-hydroxyquinoline (2) could be obtained from the 2,4dinitrophenyl oxime (1).It was therefore of interest to see if 2,4-dinitrophenyl oximes such as (3) or (4) could undergo intramolecular cyclisation on to the pyrimidine 5-position to give 7,8dihydropteridines (5) or (6) respectively, which could be readily oxidised to the corresponding fully aromatic pteridines.No pteridine has yet been synthesised by a route of this type.

Results and Discussion
Various approaches to the synthesis of oximes (3) and (4) were explored, but they all encountered problems.For example, the diethyl acetal (8) could be prepared readily by reaction of 6-chloro-1,3-dimethyluracil 5 (7) with the diethyl acetal of aminoacetaldehyde. 6Hydrolysis of (8) with HCl, however, afforded none of the required aldehyde (9), and led only to the pyrrolo[2,3-d]pyrimidine 7 (10), formed in situ by cyclisation of the aldehyde (9).Milder methods of acetal hydrolysis based on silica failed to hydrolyse the acetal group in (8). 8t was also found impractical to use the methyl ketone (12) as a precursor for oxime (4).Although (12) was prepared as described below it was found to cyclise extremely readily to the pyrrolo [2,3-d]pyrimidine (13), and could be obtained only in small amounts.It was prepared from alcohol (11), which could be obtained in good yield by refluxing 6-chloro-1,3dimethyluracil (7) in aqueous solution with 1-amino-2-propanol.Oxidation of (11) using a variety of methods, including the Swern oxidation, always led to the cyclised material (13) as the main product, however, formed by in situ cyclisation of the initially formed ketone (12).A small amount of (12) could be isolated by chromatography, but a solution of it in ethanol was converted very rapidly into (13) when stirred with silica.Reaction of (7) with 2-amino-1phenylethanol gave the phenyl analogue of alcohol (11), namely 6-(2-hydroxy-2phenylethylamino)-1,3-dimethyluracil.Attempts to oxidise the latter to the corresponding ketone were also unsuccessful, however. oxidn.
Another approach towards the preparation of methyl ketone (12) involved the treatment of 6amino-1,3-dimethyluracil (14) with iodoacetone.No reaction occurred on refluxing these reagents together in water solution.Reaction in methanol in a pressure tube at 120 °C did afford two products, however.The first of these proved to be 6-methoxy-1,3-dimethyluracil (15), formed by reaction of (14) with the solvent.It was shown to be identical with an authentic sample of the same material prepared by treatment of 6-chloro-1,3-dimethyluracil (7) with methanolic sodium methoxide.The second product was shown to be 1,3,5-trimethylpyrrolo [2,3d]pyrimidine-2,4(1H,3H)-dione (13).For example, it gave a single HPLC peak, homogeneous by UV, when it was mixed with authentic (13), prepared as described earlier by the oxidation of (11).Its 1 H and 13 C NMR spectra were also identical with those of authentic (13), and with those reported by Kawahara et al. for the same compound. 9These observations confirm that the product obtained was (13) and not its regioisomer 10 (16), which might have been formed in this reaction.
The next experiment involved a reaction between 6-amino-1,3-dimethyluracil (14) and the 2,4-dinitrophenyl oxime (21) of iodoacetone.This was carried out with a view to obtaining oxime (4), and to investigating the possible cyclisation of it to the dihydropteridine (6).Quite unexpectedly, however, the product actually isolated was shown to have structure (22), and in order for this to be formed an unusual sigmatropic rearrangement must have taken place.The 2,4-dinitrophenyl oxime (21) of iodoacetone required for this reaction could be prepared by reaction of O-(2,4-dinitrophenyl)hydroxylamine (18) with iodoacetone, the latter being obtained by a Finkelstein reaction on chloroacetone. 11Since iodoacetone is a dangerously lachrymatory compound its preparation by this route is not to be recommended except under very carefully controlled conditions.Iodooxime (21) is also highly lachrymatory, and was used as a crude solid without being completely characterised.Iodooxime (21) could also be prepared from the corresponding chloro compound (19) by reaction of the latter with sodium iodide in dimethylformamide.The chloroacetone  12 and also by treatment of tert-butyl N-(2,4-dinitrophenoxy)carbamate (20) with trifluoroacetic acid. 13In our hands the preparation of (18) from (17) was the preferred route, using a modification of the reaction conditions described by Tamura. 12Compound (18) is heat labile, and decomposes to 2,4-dinitrophenol on standing at room temperature.It was therefore stored at 4 °C and used as quickly as possible after preparation.When the 2,4dinitrophenyloxime (19) of chloroacetone was treated with 6-amino-1,3-dimethyluracil (14) under a variety of conditions, including reaction at 120 o C in a pressure tube, the only product isolated was 2,4-dinitrophenol, formed by thermal decomposition of (19).However, if a one molar amount of sodium or potassium iodide was added to the reaction mixture at 120 o C, reaction did occur to give the new product (22).As already mentioned, the same product was obtained by reaction of the iodooxime (21) with 6-amino-1,3-dimethyluracil (14) in methanolic solution at 120 °C in a pressure tube.Presumably the chloro compound (19) reacts via its more reactive iodo analogue (21).The proposed structure of the new product (22) rests on its spectral properties and microanalysis.Thus its high resolution mass spectrum showed a single peak at m/z 376.0895 (M+H + ), indicating an empirical structure of C 15 H 13 N 5 O 7 for (22).This was confirmed by microanalysis.The 1 H NMR spectrum unexpectedly showed only two aromatic protons at δ 7.79 and 8.65 and the appearance of these as two doublets with coupling constants of J = 3 Hz, suggested two meta aromatic protons.The presence of a low field exchangeable signal (D 2 O) at δ 11.76 was indicative of a phenolic OH, and this was supported by an OH stretching band at 3401 cm -1 in the infrared spectrum.In agreement with the above, the 13 C NMR spectrum for (22) showed only two aromatic CH signals at δ 123.41 and 128.83.Together, these results suggested the presence of a 2-hydroxy-3,5-dinitrophenyl moiety in the new molecule.The 1 H NMR spectrum also showed three methyl groups at δ 2.05, 2.51 and 3.14.Significantly, there was no signal attributable to a pyrimidine 5-H.Other important features in the NMR spectra were the exchangeable NH signal and the ten quaternary carbon signals.All of these data are in agreement with the proposed structure (22).These data are also consistent with the regioisomeric structure (23).This latter was ruled out, however, on the basis of NOE results, which showed a positive increase in intensity of the pyrrolo CH 3 when the NH signal at δ 7.46 was irradiated.Similarly, when the pyrrolo CH 3 was irradiated, the NH signal increased in intensity.This shows that both these groups are in close proximity to each other in space, which result is compatible with structure (22), but not for the isomeric structure (23).Bennett and Mason 14 prepared by a different route a compound similar to (22), which exists in the indolenine tautomeric form shown in (24).This was ascribed to steric interactions of the tertiary butyl group, however, and there is no reason to suppose that compound (22) should exist in its indolenine tautomeric form.Various other mechanisms for the reaction of ( 14) with (21) can be postulated.However, the products resulting from these can be ruled out on the basis of the spectroscopic data already described and the known reactivity of 6-amino-1,3-dimethyluracil (14).No analogous reaction was observed if 1,3-dimethyluracil or 6-methoxy-1,3-dimethyluracil were used instead of 6-amino-1,3dimethyluracil (14).
A suggested mechanism for the formation of (22) is shown below.The first step involves electrophilic substitution at the 5-position of (14), to give intermediate (25).A double bond shift leads to (26) and sets the system up for a [3,3] sigmatropic rearrangement in which (26) is converted into (27).This step is similar to both the Claisen allyl ether rearrangement and the key step in the Fischer indole synthesis.An intramolecular nucleophilic attack by nitrogen on the imine carbon centre in (27) leads to ring closure, giving the fused ring compound (28), which with loss of ammonia then gives the fully aromatic final product (22).Surprisingly, when the reaction was repeated using 1,3-dimethyl-6-methylaminouracil in place of the 6-amino compound (14), a dinitrobenzo[b]furan product (32) was obtained instead, and not the expected (29).The 1,3-dimethyl-6-methylaminouracil was prepared by treating 6-chloro-1,3dimethyluracil (7) with methylamine, using a modification of a method described by Pfleiderer. 5 Again the structure of (32) follows from its properties.Significantly, the lack of a phenolic OH group signal in both its NMR and IR spectra indicated that the final cyclisation had followed a different route from that described already for the formation of (22).It is assumed that formation of (32) follows initially a pathway similar to that described for the formation of (22).The intermediate (30) formed by the sigmatropic rearrangement, however, now carries a substituted 6-amino group at the pyrimidine 6-position, and this appears to promote preferential intramolecular cyclisation on to the phenolic oxygen, leading to (31), which with loss of ammonia gives the fully aromatic benzo[b]furan product (32).Yet another variation of this type of reaction was observed when 6-amino-2-phenyl-4(3H)pyrimidinone (33) was reacted with the chloro oxime (19), using the same experimental conditions as already described for the preparation of (22) and (32).This time a furo [2,3d]pyrimidine was obtained and was shown to have structure (35).Its high resolution mass spectrum showed a single peak at m/z 408.0976 (M+H + ), indicating an empirical formula for (35) of C 19 H 13 N 5 O 6 .The 1 H NMR spectrum showed peaks at δ 8.14 and 8.65 corresponding to a pair of meta aromatic protons.Peaks at δ 6.84 and 12.06 showed the presence of an NH 2 group and a phenolic OH group, respectively, while there was no peak that could have been attributed to a pyrimidine 5-H.The first step in the formation of (35) involves pyrimidine (33) reacting nucleophilically at its oxygen with the chloro oxime (19) to give (34).This latter is converted into (35) by a tautomeric double bond shift, followed by a [3,3] sigmatropic rearrangement, intramolecular cyclisation onto the nucleophilic pyrimidine 5-position, and final loss of ammonia.Other structures isomeric with (35) may be drawn which fit the spectroscopic data, for example structure (36).Formation of the latter would involve (33) reacting nucleophilically at its 5-position with the chloro oxime (19).It is known, however, that (33) reacts with chloroacetone at its exocyclic oxygen atom, 15 and to our knowledge there are no examples of (33) reacting with halo compounds at the 6-amino group, or at the 5-position.It always seems to prefer reaction at the lactam oxygen.It is also possible to envisage several other routes by which (33) and (19) could react together, but in all cases they lead to products that would be incompatible with the observed spectroscopic data. (36)
sigmatropic rearrangement of 2,4dinitrophenyl oximes, in which two of the six reacting atoms were nitrogen and oxygen.The initially formed product was found to undergo intramolecular cyclisation in a variety of ways All reagents not prepared in the laboratory were purchased either from Aldrich Chemical Company or from Lancaster Synthesis Ltd. (UK).Thin layer chromatography (TLC) was carried out using Merck (silica gel 60) TLC aluminium sheets.Visualisation was achieved by the use of either UV light or by development over iodine adsorbed onto silica.Column chromatography was carried out using manually prepared columns containing 'flash' silica gel, which was packed under pressure.Elemental analysis was carried out in the microanalytical laboratory, Department of Chemistry, University College Dublin.Melting points are uncorrected and were measured on an Electrothermal 9100 heating ramp.NMR spectra were recorded on a Bruker DPX 400 instrument, operating at 400.14 MHz for 1 H analysis and 100.14 MHz for