α -Azido ketones, Part 6. † Reduction of acyclic and cyclic α -azido ketones into α -amino ketones: old problems and new solutions

Comparative experiments on the selective reduction of α -azido ketones to α -amino ketones revealed that tin(II) chloride reduction followed by immediate protection with Boc group is the method of choice. This methodology proved to be useful for more complex substrates, too. Chromium(II) acetate also resulted in the desired products but in lower yields due to a competitive deazidation procedure. A mechanism to explain this deazidation was suggested


Introduction
α-Azido ketones 1 represent useful precursors of the synthetically important 1,2-amino alcohols.The survey of the literature revealed that this transformation is usually executed in two steps.Either the carbonyl or the azide groups can be reduced chemoselectively and numerous protocols have also been published for the synthesis of enantiomerically pure or enriched 2-azido-1alcohols.The major problem during the reduction of α-azido ketones 1 to the corresponding αamino ketones 2 lies in the well-known propensity of the products to the intermolecular condensation followed by dehydrogenation affording pyrazines 4 (Scheme 1). oxidation

Scheme 1
Anselme and his co-workers 1 studied the catalytic reduction of various phenacyl azides and aliphatic α-azido ketones 2 (R 1 = i-Pr, R 2 = H; R 1 = Me, R 2 = Et) over Pd-C in ethanol in the presence of a few drops of acetic acid at 3.4 atm.pressure and isolated the corresponding pyrazines 4. In one case, the intermediate dihydropyrazine 3 has also been obtained which oxidized to the pyrazine 4 spontaneously by standing in air.However, the outcome of the reaction was somewhat structure dependent.Suzuki and his co-workers 2 have also reported the formation of symmetrical 2,5-substituted-or 2,3,5,6-tetrasubstituted-pyrazines 4 by treating αazido ketones 2 with sodium hydrogen telluride in ethanol.The reaction could be performed in a "crossed" manner to synthesize complex pyrazines such as the naturally occurring cephalostatin 7, cephalostatin 12, and ritterazin K. 3 In some cases the catalytic reduction over Pd-charcoal, 4 Pd-calcium carbonate 5 or platinum oxide 6 was reported to give stable α-amino ketones but usually the products should be protected by their immediate transformation into a salt or an acylated / alkoxycarbonylated derivative to avoid the pyrazine formation.Hydrogen chloride or perchloric acid was added to the solution of the substrate prior to the hydrogenation [7][8][9][10][11] , or as an alternative, concentrated hydrochloric acid or dry hydrogen chloride was added to the reaction mixture just after filtering the catalyst off. 8,9n situ derivatization of amino ketones was accomplished by adding acetic anhydride 12 or ditert-butyl dicarbonate 13 (Boc anhydride) to the solution of the substrate prior to the hydrogenation.Acetylation 12 or aroylation by an active ester 10 immediately after the reduction has also been reported.
Only sporadic reports are available on the use of other reducing agents.Pulici et al. 14 applied tin(II) chloride dihydrate in ethanol for the preparation of 2-acylamino ketones but they presented 2-amino-1-phenylpropane-1-one (2, R 1 = Ph, R 2 = Me) as an only example in their paper.In a systematic study on the reduction of azides to amines by the combination of zinc and bismuth (III) chloride in water or aqueous ethanol, phenacyl azide was shown as a single example. 15Trivalent phosphorus compounds such as phosphines and phosphites are generally useful reagents to convert azides to amines but this methodology is not applicable for the reduction of α-azido ketones because of the concurrent secondary reactions such as pyrazine [16][17][18] or aziridin 19,20 formation.In the only reported exception, triphenylphosphine was applied in the presence of p-toluenesulfonic acid and the intermediate iminophosphoranes were immediately cleaved and the formed α-amino ketone were trapped as their tosylates. 21nother reduction method leading directly to N-acylated α-amino ketones using thioacids as reducing and acylating agent has also been reported.The method which was originally developed for the reduction of simple azides by Rosen et al. 22 was successfully applied first to α-azido ketones having a protected amino group in their α' position, the reducing and acylating thioacids were N-protected L-aminothiocarboxylic S-acids. 23This methodology was also applied by other research groups for the reduction of complex α-azido ketones using thioacetic or thiobenzoic acid. 24,25A mechanism involving an interesting triathiazoline intermediate has also been proposed. 25onsequently, an efficient method for the transformation of α-azido ketones into α-amino ketones is still a need.In this contribution we wish to present our comparative studies using various reducing systems and to demonstrate the usefulness of tin(II) chloride in this transformation.

Results and Discussion
The usefulness of the transfer hydrogenation using ammonium formate as hydrogen source in the presence of palladium on charcoal in hot methanol was investigated first; this methodology has not been tested so far.Unfortunately, the reaction 2-azidoacetophenone (5a) or 2-azidopropiophenone (5e) did not result in the desired aminoketones 8a,e, only the corresponding 1,2-amino-alcohols 6a,e were obtained in low or moderate yields.2-Amino-1phenylethanol (6a) was isolated as its p-nitrobenzoate, 7a (Scheme 2).Interestingly, the reduction of azido ketone 5e afforded anti-2-amino-1-phenyl-1-propanol (anti-6e) in nearly diastereo-pure form, only traces (≤5 %) of syn-6e was detected in the worked-up reaction mixture.The relative configurations of amino-alcohol anti-6e 26 and the minor product syn-6e 27 were verified by comparison of the chemical shifts with the literature data.We can conclude that the afore-mentioned chemoselectivity of the reduction was completely lost under these conditions.
Next, we tested the synthetic value of the catalytic hydrogenation by using Lindlar's catalyst instead of the previously reported Pd-charcoal, 4 Pd-calcium carbonate 5 or platinum oxide. 6This catalyst was found effective in the reduction of azido group 28 but has never been tried in the case of α-azido ketones.The 2-Azidoacetophenones 5a,b, 2-azidopropiophenone (5e), and the heterocyclic α-azido ketones 12a,d were hydrogenated at atmospheric pressure in the presence of Lindlar's catalyst.The product was immediately derivatized with (Boc) 2 O in the presence of sodium hydrogencarbonate to avoid the formation of pyrazines from the primary product αamino ketones 8a,b,e and 13a,d.The corresponding Boc-protected derivatives 11a,b,e and 14a,d were isolated but the yields were low or moderate (7.6-32%) in all cases.No other products could be isolated from the reaction mixture by column chromatography.In conclusion, although this reduction method works for the α-azido ketones, the observed low efficiency, particularly keeping the high price of the catalyst in mind, diminishes its synthetic value.

Scheme 2
Low-valent transition (LVT) metal ions offer another possibility to reduce the azido group.Recently, we have applied successfully chromium(II) acetate to reduce prochiral ketones into alcohols and this reduction could be performed with moderate-to-good enantioselectivity in the presence of α-amino acids. 29The same reducing system was also used for the enantioselective reduction of C=N double bonds. 30,31LVT metal ions such as tin(II), iron(II) and chromium(II) were used in the reduction of simple azides into the corresponding amines, 32 but this approach has never been tested in the case of α-azido ketones.First, we investigated the reduction of α-azido ketones with chromium(II) ions.When 2-azidopropiophenone (5e) was treated with chromium(II) acetate in water-dioxane medium and the worked-up reaction mixture was purified by column chromatography, 2,5-dimethyl-3,6-diphenylpyrazine (15) was the only isolable product.This observation provided a further proof for the necessity of the immediate protection of the α-amino ketone products.We studied various protecting groups such as 4-nitrobenzoyl, benzyloxycarbonyl and tert-butoxycarbonyl but no marked difference was found in the yields (Table 1).Moreover, the same moderate yields were observed when intermediate 8e was treated with (Boc) 2 O under different conditions.These results support that reason of the low yields is in the reduction and in not the protection step.The reduction and derivatization of the phenacyl azides 5a,b,d also gave similarly low yields.More surprisingly, the reduction of 3-azidochromanone (12a) and 3-azido-1-thiochromanone (12d) with chromium(II) acetate did not result in any expected products 14a,d, only compounds 16,17, the products of a deazidation reaction, were obtained.The same deazidation was observed in the case of another open-chain substrate.The treatment of 2-azido-1,2-diphenylethanone (18)  with chromium(II) acetate gave deoxybenzoin (20) exclusively.Therefore, it is very likely that this side-reaction is responsible for the lower yields in the case of other substrates.The deazidation may be explained in terms of the SET mechanism of the reduction.The second SET step of the reduction leads to anion 23 which, instead of a protonation, loses an azide ion giving enol 24.The tautomerization of the enol 24 yields the final product ketone 25 (Scheme 3).
Scheme 3 Finally, we studied the reduction of α-azido ketones 5a,c,d,e and 12a-d with tin(II) chloride in methanolic solution under nitrogen atmosphere, the α-amino ketone intermediates 8a,c,d,e,  13a-d were derivatized with (Boc) 2 O in the presence of TEA.Although the yields varied in relatively wide range (24-84%), the values were generally better than by using any previous method (Scheme 2, Table 1).We can conclude that tin(II) chloride is the reagent of choice for the reduction of α-azido ketones.
4][35] In our first attempt the sequential reduction and protection of 2-azido-3-hydroxy-1-phenyl-1-butanone (26) according to the procedure described above resulted in 2-(N-tert-butoxycarbonylamino)-1phenyl-1-ethanone (11a) as the only product instead of the expected compound 27.Obviously, a competitive and faster retro-aldol cleavage leading to the phenacyl azide (5a) took place prior to the reduction.To avoid this side reaction the 3-hydroxy group should be blocked with an appropriate protecting group.Previously, we reported 33 on the efficient tertbutyldimethylsilylation of this compound by treating the azido-alcohol 26 with tertbutyldimethylsilyl chloride in DMF and in the presence of imidazole to give silyl ether 28 and demonstrated the lack of any epimerization during the protection step.Fortunately, we managed to find proper chromatographic conditions for the separation of the syn and anti diastereomers of azide 28.The reduction of the pure diastereomers of azides 28 followed by reaction with (Boc) 2 O in the presence of TEA afforded the expected derivatives 29 in good (66-72%) yields and in diastereomerically pure form.This methodology was also found to work in the case of heterocyclic systems.The silylation of chromanone 30 followed by chromatographic separation resulted in the pure diastereomers of protected compound 31.These azide derivatives were reduced and derivatized with Boc protecting groups successfully and the desired compounds 32 were obtained in good (54-79%) yields.

Conclusions
In conclusion, tin(II) chloride gave the best results in the selective reduction of α-azido ketones and proved to be useful in the wide range of substrates.Immediate protection of the amino group without any attempted purification seems necessary to avoid the secondary dimerization by condensation followed by dehydrogenation.Another low-valent metal ion, chromium(II), has also considerable reducing potential but this procedure suffers from a competitive de-azidation reaction.

Experimental Section
General Procedures.Chromatographic separations were performed using silica gel (Merck, 70-230 mesh).Thin-layer chromatography was carried out on Kieselgel 60 F 254 (0.25 mm layer thickness, Merck).Melting points were determined on a Boetius hot-stage apparatus and are uncorrected. 1H-and 13 C-NMR spectra were recorded with a Bruker AM 360 (360 MHz for 1 H-; 90 MHz for 13 C-nuclei) or a Bruker WP 200 SY (200 MHz for 1 H-nuclei) spectrometer in CDCl 3 solution unless otherwise specified (internal standard TMS, δ = 0 ppm).IR spectra were recorded with a Perkin-Elmer 16 PC-FT-IR instrument in KBr disks.Elemental analyses were performed in house with a Carlo Erba 1106 EA instrument.
To the ethereal extract of the alkaline aqueous phase di-tert-butyl dicarbonate (2.36 g, 10.80 mmol), triethylamine (0.91 mL, 6.50 mmol) and 4-(N,N-dimethylamino)pyridine was added and the mixture was stirred at room temperature for 23 hrs.The reaction mixture was concentrated in vacuo and purified by column chromatography (hexane-ethyl acetate: 4:1, v/v) to give the carbamate 11a (183 mg, 30%) as white crystals.