Regio- and stereoselective synthesis of α -hydroxy- β -azido tetrazoles †

Unreported α -hydroxy- β -azido tetrazoles were prepared in one step from readily available α , β -epoxy nitriles. This reaction involves a dibutyltin oxide-catalyzed cycloaddition of the nitrile reacting with TMSN 3 leading to the tetrazole moiety, and opening of the epoxide by the azide anion. High levels of regio- and stereoselectivity are obtained in this reaction and are discussed, also by means of quantum mechanical DFT calculations. The azido group in these compounds could be uneventfully reduced to the corresponding amine thus leading to an α -hydroxy- β -amino tetrazole, surrogate of the corresponding carboxylic acid, while reaction with triphenylphosphine led to propargylic amines.


Introduction
Tetrazoles have found applications in various domains including energetic materials, owing to their high nitrogen content, and medicinal chemistry, due to the fact that 5-substituted tetrazoles (5-ST) are bioisosteres of carboxylic acids. 1 This last property has been notably popularized with the release of the antihypertensive drug Losartan, 2 which soon became a blockbuster. The synthesis of tetrazoles has been extensively studied, cycloaddition of azides (anion or derivatives) with nitriles being the most popular way to efficiently produce this heterocycle. Many key improvements in this reaction, which can be promoted either by Brønsted or Lewis acids have appeared, including the use of sodium azide or TMSN 3 with NH 4 Cl, 3 ZnBr 2 , 4 Me 3 Al, 5 I 2 , 6 or AgNO 3 . 7 Microwave irradiation 8 has also been used with much success, but this reaction still requires elevated temperatures (typically above 100°C) to proceed, thus narrowing its scope to quite robust nitriles. Another possibility lies in the use of dibutyltin oxide as a catalyst, in conjunction with TMSN 3 . 9 In contrast to the above methods, this reaction involves neutral reaction medium and a weak Lewis acid, thus allowing cycloaddition of nitriles fitted with a Lewis base, such as amino nitriles. 10 We therefore decided to study this cycloaddition with α,β-epoxy nitriles, aiming at the preparation of functionalized tetrazoles suitable for further synthetic transformations. Our findings are exposed in the next section.

Results
We first studied reaction depicted in Scheme 1 with epoxide 1, readily prepared by a Darzens reaction. 11 Thus, reacting this epoxide (18 h) in toluene at 60°C with TMSN 3 (5 equiv.) and a substoichiometric amount of Bu 2 SnO (0.5 equiv.) led, after acidic hydrolysis, to the α-hydroxy-β-azido tetrazole 2 in good yield. Much to our delight, this reaction, involving both cycloaddition and epoxide opening, occurred with high regioselectivity and complete inversion at the β-carbon. 12 The structure of 2 was verified by X-ray crystallography (see ESI †). 13 We were surprised by the low temperature required for completion of this reaction and we first screened the amount of catalyst and TMSN 3 needed in order to maintain a high yield. Reactions were run in toluene at 60°C for 18 h. (Table 1).
Increasing the amount of Bu 2 SnO (entry 2) gave an excellent yield of 2, and lowering the amount of TMSN 3 to three equiv. (entry 3) maintained a high yield, but decreasing of the amount of catalyst to 20 mol% lowered the yield significantly (entry 5). In order to determine which event first occurred (cycloaddition or epoxide opening), we also used one equivalent of TMSN 3 and Bu 2 SnO ( preheated until dissolution) and isolated after reaction and acidic workup β-chloroα-hydroxy tetrazole, resulting from the opening of the epoxide by the chloride anion (entry 6), suggesting that cycloaddition first occurs. Thus, we chose to examine the scope of this reaction with other epoxides (shown in Fig. 1) with 3 equiv. of TMSN 3 , using 50 mol% of Bu 2 SnO (conditions of entry 3). Structures of the isolated compounds are shown in Fig. 2.
As depicted in these figures, the scope of this reaction was very good, tolerating aryl, alkenyl and alkyl groups at the β-position of the epoxide. Yields are modest to good (38-76%) and these compounds are isolated as crystalline materials. ‡ β-Disubstituted epoxides 12-18 also reacted very well and tertiary azides were obtained in good yields. Only α-disubstituted epoxide 10 failed to react in these conditions, leaving unreacted starting material. Regioselectivity was constantly excellent, leading to a unique β-azido regioisomer whatever the degree and the nature of substitution at the β-carbon. The stereoselectivity of this reaction could be evaluated with the cis or trans epoxides 1, 3-9 and 11. For 1, 3-7 and 11 the reaction appears to be stereospecific and involves an inversion at the β-carbon as shown from the X-ray structure of 2 and slightly different NMR data for diastereoisomers 2 : 19, 20 : 21 and 22 : 23 (see ESI †). Starting with 8 (1 : 2 cis-trans mixture), a 1 : 1.3 mixture of isomers 24 was obtained, albeit in low yield, which also might reflect the stereospecificity of this reaction. Only the starting trans epoxide 9 reacted with epimerization at the β-carbon, leading to 25 as a 1 : 2 mixture of diastereoisomers.
Next we briefly examined the reactivity of these α-hydroxyβ-azido tetrazoles. We first tried to reduce the azido moiety into an amino group, in order to access α-hydroxy-β-amino tetrazoles, 14 which can be viewed as surrogates of the corresponding α-hydroxy-β-amino acid. 15 The latter compounds occupy a very important place within the family of β-amino acids, being constituents of several drugs and natural molecules such as inter alia Taxol, 16 Taxotere, 17 Bestatin, 18 Microginin 19 and the HIV protease inhibitor R-87366. 20 Moreover, α-hydroxy-β-amino tetrazoles have found uses in the design of bioactive peptides with cis-conformationally restricted peptide bonds. 21   Thus, aiming to reduce the azido through Staudinger conditions, 2 was reacted with triphenylphosphine in refluxing THF for 2 h, but the crude reaction mixture unexpectedly showed formation of propargylic amine 34, together with phosphine oxide. This compound was acetylated for easier purification and 35 was isolated with an overall yield of 52% (Scheme 2).
The scope of this reaction, conducted in one pot without isolation of the intermediate amine, was briefly screened ( Table 2 and Fig. 3) and it was found to be general, with yields varying from 23 to 71% in the case of secondary azides. However, no trace of acetylenic compound could be detected in the crude reaction mixture starting from tertiary azides 28 or 32.

Discussion
The unexpected formation of propargylic amines during the reaction of α-hydroxy-β-amino tetrazoles with triphenylphosphine deserves comment. The Blum-Ittah aziridine synthesis is an established procedure to convert 1,2-azido alcohols into aziridines by treatment with a tertiary phosphine. 22 It is accepted that it goes through an oxazaphospholidine 40 that collapses to the aziridine 41 with release of phosphine oxide. A possible mechanism that would explain our results is the following. In our starting compounds, proton transfer from the acidic tetrazole to the nitrogen of the intermediate oxazaphospholidine 42 would produce 5-methylene-5H-tetrazole 43 with release of triphenylphosphine oxide, that could further decompose to vinylic carbene 44. 23 This intermediate would ultimately lead to the propargylic amine 45 through a Fritsch-Buttenberg-Wiechell rearrangement (Scheme 4). Recent precedents in the literature have demonstrated the possibility of smoothly generating vinylic carbenes from α-hydroxy tetrazoles upon activation with DCC. 24 The second point which is worth discussing is the high regio-and stereoselectivity observed during ring-opening of the epoxide. All substrates, except allylic azide 25, were obtained as single compounds, and the erosion of stereoselectivity in that particular case might be due to a dynamic [3,3] equilibration process of the allylic azide which reflects thermodynamic control. 25 Though uncatalyzed ring opening of glycidates with TMSN 3 has been reported to proceed with varying regio-and stereoselectivity (S N 2 or S N i), depending on the stereochemistry of the epoxide, 12 in our case, no reaction occurred in the absence of Bu 2 SnO suggesting the crucial role of the tin catalyst for both cycloaddition and epoxide opening. The mechanism of Bu 2 SnO-catalyzed cycloaddition of TMSN 3 with alkynes has been studied in details. 26 It was demonstrated that the active catalytic species is Bu 2 Sn(OTMS)N 3 , and that regeneration of this catalyst occurs through a S N 2 displa-Scheme 2 Reaction of α-hydroxy-β-azido tetrazole 2 with triphenylphosphine leads to the formation of propargylamine 34. Table 2 Reaction of α-hydroxy-β-azido tetrazoles with P(Ph) 3 (refluxing THF, 2 h, followed by acetylation (Ac 2 O and Et 3 N, 3 equiv.)

Entry
Starting azide Product Yield (%)  Scheme 4 A plausible mechanism accounting for the formation of propargylic amine upon the reaction of α-hydroxy-β-azido tetrazoles with triphenylphosphine.
cement at the silicon atom, which was calculated to require only 28 kcal mol −1 , followed by fast ligand exchange at the tin atom (Scheme 5). In our case, the tin atom in the produced tetrazole 46 is ideally located to assist in the opening of the epoxide by the azide anion at the β-position. This concerted reaction would account for the regioselectivity of the opening and the S N 2 process. In this case, regeneration of the catalyst would then imply reaction of the produced tin alkoxide 47 with TMSN 3 , to produce OTMS derivative 48, an exchange that can be promoted from a thermodynamic viewpoint considering the much stronger O-Si bond (190 kcal mol −1 ) compared to the O-Sn bond (130 kcal mol −1 ). In order to evaluate the feasibility of this tin to silicon exchange (47→48), simplified reaction depicted in Scheme 6 was considered. Quantum mechanical calculations at the B3LYP level of theory [with LANL2DZ ECP for tin atom 27 and 6-31G** basis set 28 for other atoms] were performed with the Firefly 8.0.1 package of programs. 29 The structure of the optimized transition state (TS) of this concerted reaction, located at only 17.9 kcal mol −1 , together with the structures RC1 and RC2 of pre-and post-reaction complexes are shown in Fig. 4, while Fig. 5 outlines the energetic profile of this reaction.
Indeed, calculations demonstrate that this exchange is favored both from kinetic and thermodynamic viewpoints, thus reinforcing our hypothesis.

Conclusions
In conclusion, we have described a straightforward entry to so far unreported α-hydroxy-β-azido tetrazoles, together with a    brief examination of their reactivity. Considering the stereospecificity of the opening of the epoxide, this process should allow the preparation of non-racemic molecules starting from readily available enantiopure epoxy nitriles. 30 Further work is in progress to extend the scope and applications of this reaction.