The pseudo-Michael reaction of 1-aryl-4,5-dihydro-1H-imidazol-2-amines with ethyl ethoxymethylenecyanoacetate

Abstract The pseudo-Michael reaction of 1-aryl-4,5-dihydro-1H-imidazol-2-amines with ethyl 2-cyano-3-methoxyprop-2-enoate (ethyl ethoxymethylenecyanoacetate) is investigated. At −10 °C reaction takes place on the exocyclic nitrogen atom, giving exclusively ethyl esters of 2-cyano-3-[(1-phenyl-4,5-dihydro-1H-imidazol-2-yl)amino]prop-2-enoic acid. The formation of isomeric enamines which may be a theoretical product of the reaction on N3 ring nitrogen atom is not observed. The N6 enamines, heated in boiling acetic acid, yield cyclic 1-aryl-5-oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carbonitriles. Heating of the enamines to the temperature of 120–140 °C without a solvent makes it possible to obtain a mixture of 1-aryl-5-oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carbonitriles and ethyl 1-aryl-5-imino-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carboxylates. The reaction of the respective hydrobromides of 1-aryl-4,5-dihydro-1H-imidazol-2-amines with ethyl ethoxymethylenecyanoacetate in the presence of triethylamine gives selectively 1-aryl-5-oxo-1,2,3,5-dihydroimidazo[1,2-a]pyrimidine-6-carbonitriles. Graphical Abstract

In this paper we present the pseudo-Michael reaction of 4,5-dihydro-1H-imidazol-2-amines with EMCA and structural studies of the respective products.

Results and discussion
The hydrobromides 3a-3f of 4,5-dihydro-1H-imidazol-2amines 4a-4f were obtained from the respective N-aryl-1,2-diaminoethanes and cyanogen bromide, as previously reported [29,39,40]. The hydrobromides 3a-3f were then transformed into free bases 4a-4f by action of sodium hydroxide and extraction with methylene chloride. The 4,5dihydro-1H-imidazol-2-amines 4a-4f were subjected to the pseudo-Michael reaction with EMCA in propan-2-ol solution at -10°C (Scheme 2). In these conditions the only isolated products were chain enamines (5a, 5c-5h), formed as a result of the reaction on the N6 exocyclic nitrogen atom. The attempt to obtain derivative 5b with 2-chloro substituent failed, probably due to the steric hindrance. The reaction on the N6 nitrogen atom was in contrast to our earlier results on the pseudo-Michael reaction of 4,5-dihydro-1H-imidazol-2-amines with DEEM [29]. In the case of DEEM, at -10°C reaction took place on the N3 ring nitrogen atom, but the isolation of the respective chain enamines was not possible due to the fast cyclization process, even at low temperature. Instead, ethyl 1-aryl-7-oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carboxylates were obtained exclusively. As was found previously, the pseudo-Michael reaction of 4,5-dihydro-1H-imidazol-2-amines with DEEM was temperaturedependent; therefore, we tried first to perform the corresponding reaction with EMCA at room temperature and then at higher temperatures (up to the boiling point of propan-2-ol). Theoretically, formation of the mixture of isomeric N3 and N6 enamines is possible in the case of both Michael reagents, but in the case of EMCA we observed the formation of N3 enamines neither at -10°C nor at higher temperatures. The N6 enamines 5a, 5c-5h formed exclusively underwent cyclization to 1-aryl-5-oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carbonitriles 6a, 6c-6h when the pseudo-Michael reaction was performed at ambient or higher temperature (the derivative 6b could not be obtained by this method as the preparation of enamine 5b was not successful). On the contrary, in the case of DEEM, at room temperature the reaction yielded mixtures with varying ratio of isomeric 1-aryl-5-oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carboxylates and 1-aryl-7oxo-2,3-dihydroimidazo[1,2-a]pyrimidine-6-carboxylates. Alternatively, to obtain compounds 6a and 6c-6h, the enamines 5a and 5c-5h should be heated in boiling acetic acid (Scheme 2). In such conditions the cyclization takes place via the ester group of 5a and 5c-5h, giving 6a and 6c-6h exclusively. Myiamoto [41], when performing the similar cyclization reaction under acidic conditions (MeOH

Scheme 2
saturated with HCl), observed the cyclization via cyano group only, leading to appropriate imines. Analogs of 6a-6h were obtained by Myiamoto [41] in basic conditions (Et 3 N/MeOH).
In the next stage we heated enamines 5a and 5c-5h to 120-140°C without a solvent. In such conditions the mixtures of the cyclization products both via ester group (6a and 6c-6h) and via cyano group, 1-aryl-5-imino-2,3dihydroimidazo[1,2-a]pyrimidine-6-carboxylates (7a and 7c-7h) were obtained (Scheme 2). The mixtures were separated with preparative thin-layer chromatography. After preparative TLC separation, compounds 7a and 7c-7h were extracted from silica gel with methanol. During extraction process some transesterification occurred which resulted in mixtures of ethyl and methyl esters. The percentage of methyl ester in the mixture depended on the substituent in the phenyl group and varied from 10 to 45 % by NMR. The obtained mixtures of ethyl and methyl esters were practically impossible to separate, and thus, they were analysed as received.
Finally, the reactions of respective 4,5-dihydro-1Himidazol-2-amine hydrobromides 3a-3h with EMCA in the presence of base (Et 3 N) led to reaction on N6 nitrogen atom (which is in agreement with our previous results obtained for DEEM [29]) and cyclization via the ester group, resulting in compounds 6a-6h (Scheme 3). This method made it possible to obtain the derivative 6b with 2-chloro substituent.
The course of all reactions was confirmed by elementary analysis and spectral data ( 1 H and 13 C NMR, MS). All the compounds were characterized with the aid of NMR spectroscopy. Assignments of 1 H and 13 C chemical shifts and 1 H-1 H coupling constants were achieved by a combination of several 2D NMR techniques. Since there are not that many protons in heterocyclic rings, the most important experiment for assignments was HMBC spectrum. By HMBC, one can see a long range correlation between H-2 and aromatic substituent at N-1. This proves the assignment of H-2 and H-3 protons. The protons and the carbons at the aromatic substituent are easily assigned by DQF-COSY and HSQC spectra. The most difficult part is the assignment of quaternary carbons at positions 5, 6, and 8a. H-7 showed HMBC correlations to all of those. However, C-8a can be identified by a long range correlation from both H-2 and H-3. A correlation between H-3 and C-5 was not observed, so the assignments of C-5 and C-6 were done based on the very different chemical shift values. Therefore, as an example, the structures of 5f and 7f were modeled by DFT method B3LYP/6-31G(d,p) and the NMR chemical shifts were calculated by the same method [42]. The good agreement (the results are not shown here) with experimental values proved the correct assignments. The multiplicities of proton signals H-2 and H-3 are noteworthy. Those methylene protons at positions 2 (H-2a and H-2b) and 3 (H-3a and H-3b) have nearly the same chemical shifts, respectively. However, the vicinal coupling constants between H-2 and H-3 differ. Actually there are four coupling constants: J 2a,3a , J 2a,3b , J 2b,3a , and J 2b,3b and the signals can be analysed, for example by simulation and iteration software PERCH [43]. 1 It was found that they are all between 7 and 12 Hz. This makes the signals look roughly like triplets, but the unequal coupling constants as well as the second order effects make the signals rather complicated. Therefore, these are stated as multiplets (m) at the experimental section.
The primary fragmentations of compounds 5a and 5c-5h as well as 7c-7h are mainly initiated from the ester function (see ' other it appears that compounds 5 may partially rearrange into compounds 7 under EI conditions (see ''Experimental''). However, some differences still exist. M ?Á is the base peak only for 5g but in case of compounds 7 for 7e, 7g, and 7h. The base peak for 5a, 5e, 5f, and 5h is [M-C 2 H 4 CO 2 ] ?Á but only for 7d and 7f. The clearest difference is seen in the abundance of [M-ROH] ?Á ion (R = Et for 5 and Me or Et for 7) which is by far more abundant for compounds 5 being even the base peak for 5d and 5e. Compounds 5e and 5g show some ions which are indicative for the 2-Me and 2-OMe substitutions and are missing from the spectra of 5f and 5g. We also succeeded in the preparation of crystals suitable for X-ray structure determination for 6c and 6d [44]. The X-ray analyses of compounds 6c and 6d were performed in order to confirm the synthetic pathway and the position of the oxo group (5-oxo/7-oxo; Scheme 2). The crystal structure of 6c is shown in Fig. 1. It was found that the investigated compounds contain an oxo group at position 5. The geometric parameters (bond lengths, angles, torsion angles, and planarity of the rings) are very similar to those observed in a previously reported structure of 1-(4-chlorophenyl)-5-oxo-1,2,3,5-tetrahydroimidazo[1,2-a]pyrimidine-6-carbonitrile (6d) [44]. The molecule as a whole adopts a nearly planar conformation with the torsion angle C2-N1-C21-C26 of 1.2(4)°. This conformation is stabilized by an intramolecular C26-H26ÁÁÁN6 interaction leading to the formation of a six-member ring described by the S(6) graph-set symbol [2,45] (Fig. 2). Moreover, the p-electron systems of the pairs of pyrimidine ring at (x, y, z) and phenyl ring at (x, 1 ? y, z) and phenyl ring at (x, y, z) and pyrimidine ring at (x, -1 ? y, z) overlap each other, with centroid-tocentroid separation of 3.5695(18) Ǻ . The pÁÁÁp distances between overlapping planes are alternately 3.4648(11) and 3.4564(13) Å and the angle between them is 3.21(14)°.
Theoretical calculations at DFT/B3LYP/6-311 ?? G(d,p) ab initio level [42] show that that 5-oxo isomeric form of 6c and 6d (the initial geometries were built from their crystallographic data) obtained after energy minimization and geometry optimization in the gaseous phase is more energetically stable than 7-oxo form, with a difference in the energy between the 7-oxo and 5-oxo forms of 45.6 and 46.0 kJ mol -1 for 6c and 6d, respectively. In solution (water (e = 78.35) and chloroform (e = 4.71), CPCM model [46]) the energy difference between form 7-oxo and 5-oxo is 18.8 kJ mol -1 (aqueous solution) and 27.2 kJ mol -1 (chloroform solution) for 6c and 19.3 kJ mol -1 (aqueous solution) and 27.6 kJ mol -1 (chloroform solution) for 6d. Thus, the population of the Fig. 1 A view of the X-ray molecular structure of 6c with the atomic labeling scheme 7-oxo form in vacuum and polar (water) and non-polar (chloroform) solutions estimated using a non-degenerate Boltzmann distribution is below the threshold of the detectability of conventional analytical methods.

Experimental
All reagents and solvents were purchased and used without additional purification. In particular, EMCA was purchased from Merck. Reactions were routinely monitored by thinlayer chromatography (TLC) in silica gel (60 F 254 Merck plates, DS horizontal chamber, Chromdes, Lublin, Poland) in toluene-ethyl acetate-methanol (1:3:0.5) eluent system and the products were visualized with ultraviolet light of 254 nm wavelength.
NMR spectra were acquired using Bruker Avance 500 spectrometer (equipped with BBO 5 mm Z-grad probe) operating at 500.13 MHz for 1 H and 125.77 MHz for 13 C. Spectra were recorded at 25°C using DMSO-d 6 as solvent with a non-spinning sample in 5 mm NMR tubes. Spectra were processed by a PC with Windows XP operating system and TopSpin software. Proton and carbon spectra were referenced to tetramethylsilane (TMS: 0.00 ppm). In addition to normal 1 H and 13 C NMR spectra, also a variety of gradient selected 2D measurements were used to receive an unequivocal assignment of all compounds. DQF-COSY spectra were acquired with cosygpmfqf pulse program (pulse programs refer to original ones installed by Bruker) and NOESY spectra were acquired with noesygpph pulse program with mixing time of 300 ms. 1 H-13 C HSQC spectra were acquired with hsqcetgpsisp.2 pulse program (using shaped pulses) with 145 Hz one-bond coupling constant. 1 H-13 C HMBC spectra were acquired with hmbcgplpndqf pulse program with 10 Hz long-range coupling constant. Computational methods were used to confirm the assignment of some quaternary carbons. The geometry optimizations and NMR chemical shift calculations were done by density functional B3LYP equipped with basis set 6-31G(d,p). Calculations were done by Gaussian 03 W software [42]. Numbering of atoms for 5a-5g in NMR assignments corresponds to the numbering of the final products 6a-6g and 7a-7g and is shown in Scheme 2.
The electron ionization (EI) mass spectra were recorded on a VG Analytical (Manchester, UK) ZABSpec instrument, equipped with Opus data system. Samples were introduced using a direct insertion probe at ambient temperature. Accurate mass measurements were performed at a resolving power of 8,000-10,000 (10 % valley definition) using peak matching technique and perfluorokerosene (PFK) as a reference compound. The elementary analyses were performed on a Perkin-Elmer analyzer. Melting points were determined with a Boetius apparatus.
X-ray data of 6c were collected on a Kuma KM4 fourcircle diffractometer at room temperature; crystal sizes 0.60 9 0.30 9 0.20 mm, CuKa (k = 1.54178 Å ) radiation, x-2h scans. The XABS2 absorption correction was applied [47]; T min = 0.2071, T max = 0.7898. The structure was solved by direct methods using SHELXS97 [48] and refined by full-matrix least-squares with SHELXL97 [48]. The H atoms were positioned geometrically and treated as riding on their parent C atoms with C-H distances of 0.93 Å (aromatic) and 0.97 Å (CH 2 ). All H atoms were refined with isotropic displacement parameters taken as 1.5 times those of the respective parent atoms. The flack parameter of -0.02 (2) confirmed that the correct absolute structure was refined [49]. All calculations were performed using WINGX version 1.64.05 package [50]. General procedure to obtain compounds 5a-5h       General procedure to obtain compounds 6a-6h Method A Enamine 5a-5h (0.01 mol) was dissolved in 10 cm 3 of glacial acetic acid and refluxed under mild boiling for 4 h. The solvent was distilled off and the solid residue was crystallized from DMF.
Method B 1-Aryl-4,5-dihydro-1H-imidazol-2-amine hydrobromides (3a-3h, 0.01 mol) and 1.69 g of EMCA (1, 0.01 mol) were dissolved in 20 cm 3 of ethanol. The solution was stirred under reflux for 4 h and then 1.02 g of triethylamine (0.01 mol) was added dropwise over a period of 15 min and the mixture was refluxed for additional 6 h. The precipitate was filtered off and washed with methanol.