Synthesis of N-7-Substituted Purines from Imidazole Precursors

Synthesis of purines and purine mono-1-oxides from the respective 5-substituted 4-nitroimidazole derivatives is described. The commercially available 4-nitroimidazole, in a few simple steps (including Vicarious Nucleophilic Substitution of Hydrogen), was transformed into: 4-aminoimidazoyl-5-carboximes, N-[1-(4-amino-1H-imidazol-5-yl)methylidene]-N-(aryl)amine derivatives or into 4-amino-5-aminomethyl-imidazole. These intermediates were cyclized to the final products.


Introduction
In the past years we have reported new and efficient approaches to fused pyrimidines [1][2][3], using aromatic nitrocompounds which were commercially or very readily available as starting materials.In these publications, the synthesis of many fused pyrimidines based on carbocyclic (benzene, naphthalene) and heterocyclic rings (quinoline, thiophene, imidazole) was described.Among them, occasionally a few examples of purine synthesis were also reported.In this methodology, the appropriate nitroaromatic compounds were transformed into the desired ortho-substituted derivatives, and the crucial step of these syntheses was realized by Vicarious Nucleophilic Substitution of Hydrogen (VNS) [4].Hence, the use of VNS reaction has been found to be an excellent tool for introduction of the dihalomethyl group [5] and isocyanomethyl group [3,6] in the ortho position of aromatic or heteroaromatic nitrocompounds, which thus giving an opportunity for further transformations leading to cyclizable intermediates.
Here, we would like to present the application of these approaches to pyrimidines for synthesis of purines.As outlined on the general scheme in the Abstract, the starting material in this synthesis is the commercially available 4-nitroimidazole, in which the presence of the nitro group allows introduction of the desired substituents Y at the carbon atom bearing a hydrogen in position 5-.
Synthesis of purines from imidazole precursors has been described in the literature [7,8].Among these works, the latest reviews [7b,7c] should be mentioned, as they cover most of the synthetic aspects of purines with the use of this method.However, thus far, a serious limitation for these syntheses was the rather difficult access to suitable key-intermediates or to their precursors.Now, based on the VNS methodology, the new approach to purines presented here seems to be easy and efficient.We tried to fill the gap concerning the preparation of intermediates, and to illustrate this development by several new syntheses of purines.

Synthesis of Purines via Oximes (3)
7-Substituted purines can be efficiently obtained from the 4-aminoimidazole-5-carbaldehyde oximes (4).Appropriate oximes were easily prepared from 4-nitroimidazole via the VNS in the reactions of 1a,b with chloroform followed by subsequent standard transformations [1,9] (Scheme 1).Because of their instability, the aminooximes 4 should be directly used in the next steps after preparation from the corresponding 4-nitroimidazole-5-carboximes (3a,b).Condensation of the amino group in oximes 4a,b with orthoesters resulted in formation of iminoethers 5.These imidates underwent the previously unknown cyclocondensation of the ortho-neighbouring iminoether and oxime groups with ammonia (carried out in a sealed tube), and led to the purine ring systems 6 in good overall yield.This approach (Scheme 2) was exemplified by preparation of a few purines and it was described in the previous communication [1].Full data for the all synthesized compounds is included in this experimental.
Preparation of purines of this type is rather a difficult task.In the earlier work [7,8], in the synthesis from imidazole derivatives, substituted purines (6-amino-, hydroxy-, etc.) or dihydro-compounds were usually obtained.The yields for the last step of these syntheses varied from 40 to 70%.On the other hand, compounds 6a and 6b, thus far described in the literature [23], were obtained via catalytic dehalogenation of already prepared purine derivatives (i.e.chloropurines) in yields of 60-80% (given for the last step as well).
In the method presented here, synthesis of purines which do not contain any substituent in position 6-is possible directly from imidazole precursors, in overall yields (calculated for the last four steps) 20-66%; depending on R in position 2-.

Synthesis of Purine mono-N-Oxides via Oximes (3)
Aromatic or heteroaromatic ortho-aminocarboximes are very valuable intermediates for preparation of fused pyrimidine mono-N-oxides by the cyclization reaction with the use of orthoesters [10].This is a method for the selective introduction of the N->O function into the desired position in heteroaromatic compounds having more than one nitrogen atom.However, only a few examples of application of this method (or its modifications) have been described in the literature [11][12][13], probably because of the limited availability of the proper ortho-aminocarboximes.Now, easy access to these intermediates by the VNS methodology allows us to develop this very useful cyclization, as it was demonstrated for synthesis of many heterocyclic systems [2].
Aminooximes such as 4 are proper key-intermediates for synthesis of [7H]-purine mono-N-oxides.These substrates can easily cyclize to purine 1-oxides, preparation of which would be a difficult task and practically impossible to realize by direct N-oxidation of the corresponding purine.
Condensation of 4 with various orthoesters resulted in formation of ortho oxime-imidates 5. Then nitrogen attack of the oxime moiety on the imidate electrophilic "C=N" double bond, followed by the loss of a proton, leads selectively to the desired purine mono-N-oxides.However, for purine derivatives this cyclization was very erratic and the yields were changeable, because of a few competitive reactions observed, i.e.: back-hydrolysis of imidate 5 to the starting amine 4, and the formation of nitrile 12.It can be explained by the fact that ethanol (solvent for this reaction) always contains traces of water -hence the hydrolysis to starting amine 4 was possible.Moreover, this imidate (5) even when heated in dry ethanol can be hydrolyzed by alcohol moiety.When a dry acetonitrile has been used instead of ethanol, the decomposition was also observed, because the water which is necessary for this hydrolysis, at higher temperature can be produced in the competitive dehydration of oxime 5 to nitrile 12. So, for each case of the purine N-oxide preparation, optimization of the conditions was necessary.All these possible transformations are outlined in Scheme 3 and the optimized conditions were listed in Table 1.
In addition, the method presented here allows not only for selective introduction of the N->O function into the 1-position of the target N-oxides, but it can also be applied for functionalization at C-2 (introduction of alkyl substituents) by using diverse orthoesters.For some reactions the attempts to convert 5 into 11 failed.For example, the optimization in the case of orthopropionate triethyl ester, while reacting with 4b, was unsuccessful.Finally in ethanol, at temperature 200°C small amounts (<5%) of the proper N-oxide 11g were detected by 1 H NMR.  The structures of N-oxides 11a-e,g were confirmed by 1 H NMR and 13 C NMR spectra.According to our knowledge 13 C NMR spectra have not so far been published for any of this type of purine mono-Noxides and this is the first attempt of the assignment of chemical shifts to carbon atoms of the structure.
First, C-8 signals were easily identified due to the almost identical chemical shifts in all compounds.Also the assignment of C-2.1, C-2.2, C-2.3, C-2.4 and C-7.1 was a trivial problem.However, it was difficult to assign correctly the remaining values on the basis of 13 C decoupled spectra because of a considerable upfield shifts of C-6 (in compounds 11b,d,e) and the closeness of these signals to aromatic peaks of benzyl group.Also a pair of very weak signals of quaternary carbons C-4 and C-5, despite significant difference in their chemical shifts (over 25 ppm), was difficult to assign correctly (in the case of 11a,c some even were not detected).In addition, a theoretical calculations of the chemical shifts for all of the investigated purine N-oxides using the ACD/Labs program gave very poor agreement between the calculated data and those determined experimentally.
Therefore, for the selected compound 11d two-dimensional HETCOR spectrum (GHMBC technique) was recorded.The proton chemical shifts of H-6 and H-8 were differentiated easily as the signal of H-8 always appears at higher field [7].However, the direct proof came from its strong correlation with already identified C-8 signal.Therefore, it was assigned definitively: δ(H-8) = 8.22 ppm and δ(H-6) = 8.43 ppm.We found that the proton signal at δ = 8.43 ppm (H-6) corresponds to carbon atom δ = 129.6 ppm (C-6!) and its strong correlation with carbon atom ca δ = 151 ppm was also observed.In this spectral technique it is typical to find correlation via three-bond distance, while the two-bond correlation with carbon atom at δ = 123.4ppm is very weak.Such a situation usually occurs due to a smaller two-bond coupling constant as compared to 3 J H-C .On the other hand, proton H-8 (δ = 8.22 ppm) also shows the correlation with the carbon atom at δ = 151.3ppm and with another one at δ = 123.4ppm (characteristic correlations via three bonds).Hence, it was assigned definitively: δ(C-4) = 151.-7.4').In this case, the correlations between CH 2 group and the respective aromatic carbon atoms were the most diagnostic.Therefore, on the basis of this Noxide GHMBC spectrum we assigned the values for the remaining structures in this series of compounds.All the 13 C NMR parameters are listed in Table 2.
Analysis of the spectra, reported in this paper, revealed a high regularity in the chemical shifts at C-4, C-5, C-6, C-8 of the corresponding purine bicyclic system of all investigated structures.While, at C-2 small signal deviations in C-2 substituted products (11c,d,e) occurred.However, a considerably upfield shift was observed in the C-2 substituted products when compared to with the corresponding unsubstituted compounds at C-2 (ca 10 ppm for 11d,e as compared to 11b and ca 2 ppm for 11c as compared to 11a).
In these spectra a considerable influence of the substituent at N-7 on the chemical shifts of C-6 was also observed.Replacement of methyl group at this position by a benzyl group (compounds 11b,d,e) is reflected in the upfield shift (ca 10 ppm) of the corresponding δ(C-6) values as compared to C-6 in Nmethyl derivatives 11a and 11c.It can be an effect of the steric bulk of the CH 2 Ph group or of a local aromatic ring current effect of the phenyl group.

Synthesis of Purines via Schiff bases (13)
In the previous case (Chapter 2.1) the efficient synthesis of 7-substituted purines starting from 4nitroimidazole-5-carboxime derivatives was presented.In that method along with the proper products, even if obtained with good yields, the formation of small amounts of the respective N-oxides was occasionally observed.It was due to a condensation (without ammonia) between the ortho-situated iminoether and the oxime groups.This type of cyclization was utilized independently in the synthesis of N-oxides and it was discussed in Chapter 2.2.1 (Scheme 3).
Herein, an alternative approach to the purine skeleton is reported, because the aldehydes 2 can be also easily transformed into Schiff bases of N-protected (R = CH 2 Ph, CH 3 ) 4-nitroimidazolecarbaldehydes (13) and they were used (instead of oximes) in the condensation with ammonia.So N-oxide formation is avoided with this method.
Selective reduction of the nitro group in imines 13 on 10% Pd/C in ethanol (25-40 psi, ca 4 h) fol-lowed by condensation with orthoesters results in the formation of imidates (15).Their heating in a sealed tube with an excess of ammonia in ethanol (80-140°C, 3-6 h) gives purines 6.Perhaps the first step of this cyclocondensation is the addition of the ammonia moiety to the carbon atom on the >C=N double bond of the Schiff base, and then replacement of the OEt group by the amino species by addition-elimination.If this postulated reaction sequence operates as drawn on Scheme 4, then the last step of this transformation must be the elimination of aryloamino-moiety from the dihydro-intermediate 16.This could indeed be the case, as a small amounts of 16 (e.g.16b) were identified in the reaction mixture along with the main products 6.It is worth to mention that the ArNH-moiety, in basic conditions, is rather a poor leaving group.Hence, one can expect difficulties on the elimination step leading to the corresponding purine ring.
The reaction of the Schiff base prepared from the 1-methyl-4-nitro-1H-imidazole-5-carbaldehyde and para-toluidine (13a; X = para-Me) afforded the purine 6a with moderate total yield (20%).One can expect, that electron-withdrawing groups in the aryl substituents should make this elimination easier.Looking for a good arylamine to prepare the desired Schiff base, a para-cyanoaniline (X = para-CN) was selected to check this postulated substituent effect and to improve the yields of purines.Initially, it was found that the CN group was resistant to reduction in these catalytic conditions.However, in preliminary reactions when the methyl substituent in para-position (Hammett constant: σpara(CH 3 ) = -0.17[14]) was replaced by the strong electron-withdrawing CN group (σpara(CN) = 0.66 [14c]), the yield of purines 6a or 6b decreased dramatically (< 5%), while in the post-reaction mixture, the starting para-cyanoaniline and its N-formylated derivative 17 were found (40-45%).Similar results were obtained, or no product was detected for para-chloro-substituent (this was examined on other model compounds [15]), for a 2,4,6-tribromo-derivative (13e) and for the bulky trityl moiety (13f).It was found that these strong electron-withdrawing groups increase a reactivity of the imine >C=N double bond; hence, its hydrogenation on the reduction step (13 -> 14) probably occurred, decreasing the overall yield.Additionally, the degradation of this labile >C=N double bond during the cyclocondensation (15 -> 6) is also possible.This problem was investigated carefully on model compounds (benzene and naphthalene derivatives) and the results will be published soon [15].
On the other hand, the best results were obtained in the case of the moderate electron-withdrawing group X = para-Br, σ = 0.23 [14] (imine 13h, 36%).Optimization of this reaction amounted the yield of 6b up to 47% and the use of para-bromoarylimine 13g gave satisfactorily results as well.However, the introduction of alkyl substituents into position 2-(entries i,j; Table 3), probably due to an additional steric effects, gave poor yields.One can expect that this is a limitation for the presented synthesis.

Synthesis of Purines via Isonitriles
We have described lately [6] a general method for the synthesis of ortho-(isocyanomethyl)nitro-aromatic or nitroheteroaromatic compounds.As well, a practical application of this method for the synthesis of fused pyrimidines has been published already [3].In this chapter an approach to purines via the 4-nitro-5-isocyanomethylimidazole derivative ( 18) is presented.The above isonitrile (18) was efficiently obtained from the 1-methyl-4-nitro-1H-imidazole 1a by the Vicarious Nucleophilic Substitution of Hydrogen (VNS) in 86% yield.
The attempts to cyclize isonitrile 18 to purine 21a (R' = H) in various catalytic conditions was unsuccessful.It could be due to inhibition of the catalyst by the isocyano moiety [16].However, using platinum as catalyst (10% Pt/C) traces of the desired purine were detected.
The exhaustive hydrolysis of 18 (concentrated HCl, EtOH-H 2 O, reflux, 3 h) afforded the hydrochloride of 19, which after recrystallization from methanol is a very valuable and stable intermediate for the next steps of this synthesis.Its catalytic reduction (10% Pd/C, 25 psi, 4 h, in EtOH) and treatment of the crude diamine formed (20) with appropriate orthoesters (reflux, 6 h) gave purines 6a,c in ca 20% overall yields (path A).The last step, an aromatization of the dihydro-derivative 21 occurred spontaneously while exposed to air.However, this may be a more complicated red-ox process, because in the post-reaction mixture traces of a compound with molecular mass equal 138 was detected by MS (m/z = 138, 100%).This could be the tetrahydro-compound 22, but the attempts to isolate this product failed, therefore we do not have any additional spectroscopic evidence for this structure.To improve the yields of the corresponding purines the condensation with orthoesters was carried out in boiling ethanol to assure lower temperature (path B).At this temperature the reaction was stopped at the stage of the dihydro-compounds 21a,b giving also considerably higher overall yields (61% and 39.5% respectively).However, the condensation of 20 with orthoesters of carboxylic acids bearing a longer carbon chain (orthopropionate, orthovalerate) or an aromatic substituent (orthobenzoate) failed or only traces of the expected products were detected.In these cases the condensation is probably a slower process and the degradation of the relatively unstable amino-derivative occurred [17].

VNS
The hydrolysis of the isocyano group must be carried-out exhaustively, because after incomplete hydrolysis of isonitrile 18, followed by catalytic reduction and cyclocondensation of the crude mixture with triethyl orthoacetate, two products 6c and 6a (!) can be obtained unexpectedly.It was a case in a one of our preliminary experiments.The explanation of this fact is given on Scheme 6. Partial hydrolysis of N=C group gave the formamido-derivative 23, which after reduction of the nitro group followed by cyclization and spontaneous oxidation [reflux in CH 3 C(OC 2 H 5 ) 3 ] leads to purine 6a.
To assure full conversion to amine 19, the hydrolysis had to be carried out during 3 h in boiling ethanol.Otherwise, formation of small amounts of compound 23 was always observed!

Conclusions
New and efficient approaches to fused pyrimidines, previously described in several papers from our laboratory [1][2][3], can also be applied for the synthesis of purines.The scope and limitations of the above methods for this synthesis is presented.With these methods, the preparation of purines and their N-oxides which do not contain any substituent in position 2-and/or 6-is possible directly from imidazole precursors.We used the new reaction (Vicarious Nucleophilic Substitution of Hydrogen) and the new substituent pattern to achieve this goal.Until now, this was rather a difficult task, because in earlier work [7,8] substituted purines (6-amino-, hydroxy, etc.) or dihydro-compounds were obtained usually in the synthesis from imidazole derivatives.
Some of the approaches presented herein, i.e. synthesis via cyclocondensation of iminoether and oxime groups with ammonia, or selective preparation of purine mono-N-oxides, are of substantial value.Taking into account commercially available starting material (4-nitroimidazole) these approaches probably can receive more attention.
By chance we also determined 13 C NMR spectra for a series of newly prepared purine N-oxides.

Experimental
NMR spectra were recorded with a Varian GEMINI-200 spectrometer operating at 200 MHz for 1 H and 50 MHz for 13 C. Coupling constants J are expressed in hertz [Hz].Two-dimensional HETCOR GHMBC spectra were recorded with a Varian INOVA-500 spectrometer (500 MHz / 125 MHz).The assignments of the chemical shifts accompanied by asterisks* ) may be exchanged. 13C NMR spectra were assigned on the basis of literature data and theoretical calculations using the ACD/Labs program. 13C NMR of purine N-oxides 11a-e are discussed in Chapter 2.2.2.IR spectra were measured with a Perkin Elmer 1600 FTIR spectrometer and mass spectra with an AMD 604 (AMD Intectra GmbH, Germany) spectrometer (electron impact and LSIMS methods); m/z values are given as a % of relative intensity.Melting points are uncorrected.TLC analysis was performed on aluminum foil plates precoated with silica gel 60F 254 Merck.Silica gel 200-300 mesh (Merck AG) was used for column chromatography.
Preparation of Purines 6a-f.General Procedure The 4-nitro-5-imidazole carbaldehyde oxime derivative (3a, 3b; 0.5 mmol) [9] was dissolved in ethanol (5 mL) and it was hydrogenated in Parr Apparatus using a catalytic amount of 10% Pd/C (ca 20 mg) at 40 psi during 4 h.The catalyst was filtered off, washed with alcohol and the filtrate after evaporation to dryness gave crude aminooximes 4a, 4b quantitatively.
Examples of 1 H NMR data for crude imidates 5b-d are given below.Another imidate (5g) is described in the next paragraph.
Preparation of Purine mono-N-Oxides (11a-e, g).General Procedure The intermediates 5a-e,g were prepared according to the procedure used previously for preparation of purines 6a-f.
The appropriate imidate 5a-e,g was dissolved in a dry ethanol or dry acetonitrile (4 mL for ca 50 mg) and the reaction mixture was heated in a sealed tube for 2-4 h at 100-200°C (see Table 1).After cooling and evaporation of the solvent the residue was purified by chromatography to give the desired purine N-oxides (eluent: from CHCl 3 to CHCl 3 /MeOH -20:1; except for 11a: from CHCl 3 to CHCl 3 /MeOH -5:1).Yields are given in Table 1, 13 C NMR spectra -in Table 2.
As the transformation of 5g to N-oxide 11g proved somewhat troublesome, an analytical sample of 5g was purified by column chromatography for characterization (the yield was not determined).

Transformations to Schiff bases
Procedure A. -To a solution of aldehyde (2a, 2b correspondingly; 0.50 mmol) and aniline derivative (0.70 mmol) in toluene (5 mL) a catalytic amount of p-TsOH acid (5 mg, 0.03 mmol) was added and the mixture was heated to reflux for 2.5-3 h.After cooling and evaporation to dryness, crude products were purified by chromatography (CHCl 3 as eluent) to give a mixture of E/Z isomers (products 13a, 13b and 13h were obtained).
Data for products 13a and 13b were described earlier [9].
For crude intermediate 20 (as hydrochloride) MS spectrum was recorded.

Table 1 .
Reaction conditions and the yields.