An investigation of the reactions between azido alcohols and phosphoramidites

Abstract

The reactions of several β-, γ-, and δ-azido alcohols with dibenzyl and dimethyl N,N-diisopropylphosphoramidites were examined. Detailed analysis of the intermediates and products formed from the reactions under different conditions provided useful information to gain insights into their mechanisms involving intramolecular Staudinger reaction, as well as the structure-reactivity relationships of both substrates. The reactions of γ- and δ-azido alcohols with dibenzyl N,N-diisopropylphosphoramidite could produce 6- and 7-membered cyclic phosphoramidates, thereby providing a new synthetic method for these biologically important molecules.

Graphical Abstract

The Staudinger reaction is valuable for creating phosphorous-nitrogen bonds via the reaction between phosphites and organic azides (Scheme 1).¹ This reaction forms first a phosphazide ylide intermediate, which can convert into a phosphazine intermediate by losing a molecule of nitrogen. Next, the resultant phosphazine is transformed into phosphoramidate as the end-product either spontaneously or under mild acidic conditions.

Scheme 1.

Previously, we observed the undesirable formation of a cyclic phosphoramidate (1) in the synthesis of a CD52 GPI anchor when an azido alcohol (2) was reacting with a phosphoramidite.² We hypothesized that the reaction first formed an azido phosphite intermediate 3, which underwent an intramolecular Staudinger reaction to result in 1 (Scheme 2). According to this mechanism, it was expected that other azido alcohols might behave similarly, thereby forming a novel synthesis of cyclic phosphoramidates.

Scheme 2.

To the best of our knowledge, the Staudinger reaction has not been used to synthesize cyclic phosphoramidates, which are useful compounds in the design and development of pesticides, cancer chemotherapeutics and enzyme inhibitors.^3–7 Current synthetic routes to these molecules include the reactions of phosphoryl trichloride with amino alcohols or other methods using relatively harsh reaction conditions.^3–7 One exception to this is the recent development by the Johnston group of a new method for the synthesis of C- and P-chiral cyclic phosphoramidates employing Brønsted acid-catalyzed intramolecular phosphoramidic acid addition to alkene.⁸

To investigate the above hypothesis, we prepared several azido alcohols and probed their reactions with phosphoramidites. The reaction of dibenzyl N,N-diisopropylphosphoramidite 4 with a δ-azido alcohol 5 in the presence of tetrazole was very fast (finished in about 10 min at room temperature) to produce a less polar product, which remained unchanged in the mixture for hours. However, our attempt to isolate this product by silica gel column chromatography failed. We suspected that it might be the azido phosphite intermediate that was oxidized in air on the column. To examine this assumption, we treated the reaction mixture with m-chloroperbenzoic acid (m-CPBA) before workup, in an effort to transform the azido phosphite into the corresponding phosphate. Indeed, under revised conditions, azido phosphate 6 was isolated as the major product in a 75% yield (Table 1, entry 1) after silica gel column chromatography. Similarly, the reaction of 4 with γ-azido alcohol 7 or β-azido alcohols 9 and 11 afforded azido phosphates 8, 10 and 12, respectively, in good to excellent isolated yields (Table 1, entries 2–4). These results have proved unambiguously that while the reactions between 4 and alcohols are easy and fast to afford azido phosphites that are stable in the reaction mixture at room temperature, under these conditions inter- and intramolecular Staudinger reactions were very slow. Buoyed by our previous discovery of cyclic phosphoramidate formation,² we began to explore different conditions to promote the intramolecular Staudinger reaction of the azido phosphite intermediates formed from the above reactions. We found that in the presence of tetrazole, the reactions of 4 with 5 and 7 at 80 °C gave cyclic phosphoramidates 13 and 14, respectively, in good to moderate yields (Table 2, entries 1 and 2). The reaction products were conveniently isolated by column chromatography and fully characterized with HR-MS and NMR. To our surprise, however, the reaction between 4 and cis-β-azido alcohol 9 under the same conditions resulted in a complex mixture but did not produce a significant amount of the corresponding cyclic phosphoramidate, even though the azido and the hydroxyl groups in 9 were at cis positions, which are favorable for intramolecular reactions. Not surprisingly, the reaction between 4 and unfavorable trans-azido alcohol 11 did not afford the cyclic phosphoramidate either. The results suggested that the ring size of the reaction intermediate or product might play a critical role in intramolecular Staudinger reactions of azido phosphites.

Table 1.

Reaction of azido alcohols with (BnO)₂PNiPr₂ 4 at rt

Table 2.

Reaction of azido alcohols with (BnO)₂PNiPr₂ 4 at 80 °C

In view of the potential influence of electronic and steric effects of the phosphite group on its Staudinger reactions,⁹ we studied subsequently the reactions of the same set of azido alcohols with phosphoramidite 15 bearing methyl instead of benzyl esters. We anticipated that this would facilitate the intramolecular Staudinger reaction of the resultant azido phosphite intermediates. Indeed, even at room temperature the reaction of 15 with 5 produced a modest yield (40%) of the desired cyclic phosphoramidate 16 (Table 3, entry 1). More interestingly, a substantial quantity (10%) of cyclic phosphorazide 17 was also obtained. Clearly, 17 was stable to column chromatography. The reactions of 15 with azido alcohols 7 and 9 were quite complex and the isolated major products were phosphorazides 18 and 19 (Table 3, entries 2, 3), respectively. For azido alcohol 11, the reaction with 15 was clean but the product was isolable only after oxidation with m-CPBA to afford azido phosphate 20 in an excellent yield (91%), suggesting the formation of a stable azido phosphite intermediate. These results revealed that the azido phosphite intermediates formed from the reactions of dimethyl phosphoramidite 15 with azido alcohols could undergo intramolecular Staudinger reactions even at room temperature, except where the cyclization was hindered by unfavorable trans configurations in substrate 11. Evidently, these reaction conditions were much milder than those required for the intramolecular Staudinger reaction to occur with dibenzyl phosphoramidite 4.

Table 3.

Reaction of azido alcohols with (MeO)₂PNiPr₂ 15 at rt

^*Reaction conditions: 1.2 equiv tetrazole, CH₂Cl₂, rt, 10 h; then m-CPBA (1.2 equiv), rt, 0.5 h.

The results shown in Table 2 and Table 3 clearly demonstrated that the reaction yield and stability of the cyclic phosphoramidate products followed an order of: 7-membered ring > 6-membered ring > 5-membered ring. In agreement with this trend, the results in Table 3 also suggested that the reaction yield of phosphorazide intermediates was in a reversed order, namely, 19 > 18 > 17. For the reaction of 15 with 7 or 9, no cyclic phosphoramidate product was observed, whilst the latter reaction gave the highest isolated yield of phosphorazide 19. Both the cyclic phosphoramidate and the cyclic phosphorazide products should have been formed via intramolecular Staudinger reaction (see Scheme 2). However, the observation of mainly cyclic phosphorazide for the reactions of 15 with 7 and 9, as well as a mixture of cyclic phosphoramidate and phosphorazide for the reaction of 15 with 5, also indicated that intramolecular Staudinger reactions of the azido phosphite intermediates were not the limiting factor for the transformation into phosphoramidate products. Yet, another interesting finding was that at room temperature 17 could not be converted into 16 on treatment with tetrazole and diisopropanylamine, indicating that 16 obtained from the reaction of 15 with 5 might be formed through a more reactive intermediate rather than through cyclic phosphorazide 17. Furthermore, heating β-azido phosphites at 80 oC gave complex results that were different from the reactions of β-azido alcohols with phosphites or phosphines.^10,11

To explain the above results and observations, we propose that the phosphazide ylide intermediate generated by intramolecular Staudinger reaction of azido phosphite endured two competitive pathways as shown in Scheme 3. One of these pathways involved ring contraction via automatic elimination of a nitrogen molecule to produce phosphazine 21 that could undergo protonation and nucleophilic de-O-alkylation to afford cyclic phosphoramidate 1 (Scheme 3A). Another pathway involved direct protonation and nucleophilic dealkylation of the phosphazide ylide intermediate to give cyclic phosphorazide 22. It seemed that once phosphazide 22 was formed it could not be converted into phosphoramidate under the probed conditions. However, as some of the reactions were very complex and gave low yields of identified products, we could not completely eliminate the possibility of intermolecular Staudinger reaction of azido phosphites to form oligo-/polymers, which are difficult to purify and characterize.

Scheme 3.

Moreover, the results indicated that the ring contraction reaction was highly dependent on the reaction condition and the substrate structure. It was favorable at higher temperature and with larger cyclic phosphazide ylides (≥ 8-membered ring, producing ≥ 6-membered ring product). At 80 °C, cyclic phosphoramidates were obtained as the major products and no phosphazide was isolated in a significant yield from any of the reactions, probably because, even if phosphazides were generated, they would decompose at a high temperature. In addition to the 7- and 6-membered ring products 13 and 14 that were readily formed from the reactions of 4 with 5 and 7 (Table 2), the by-product observed in our synthesis of a CD52 GPI anchor was an 8-membered ring cyclic phosphoramidate, which could be formed at room or lower temperature.2 Therefore, it seemed that the 8-membered cyclic phosphoramidate product was even more readily generated from intramolecular Staudinger reaction than the 7-membered cyclic phosphoramidate, such as 13. Continuing this trend, 5-membered cyclic phosphoramidate products were not observed from the reaction of 9 with either 4 or 15.

To further verify the conclusion that it would be difficult to form 5-membered cyclic phosphoramidates through intramolecular Staudinger reaction, we prepared two more β-azido alcohols, 23 and 24, and probed their reactions with 4 and 15 under above-described conditions (Scheme 4). The results were in agreement with our prediction. As shown, the reactions of 23 and 24 with dibenzyl phosphoramidite 4 at room temperature produced the corresponding azido phosphites as major products, which were converted into azido phosphates 25 and 28, respectively, after m-CPBA oxidation; however, at 80 °C, these reactions were complex and no major products could be isolated and identified. For the reactions of 23 and 24 with dimethyl phosphoramidite 15 at room temperature, a significant amount of phosphazides 26 and 29 were isolated. The results were similar to that of the reactions between related 1,2-azido alcohol 9 and phosphoramidites 4 and 15. In addition, aziridine phosphoramidate 27 was also observed in the reaction of 23 with 15, and similar products were reported previously.^12–14

Scheme 4.

In summary, the reactions between phosphoramidites and azido alcohols were examined in detail. It was disclosed that the results of these reactions were dependent on the substrate structure and reaction conditions. The capture of azido phosphite intermediates or isolation of cyclic phosphorazide products from the reactions carried out at room temperature enabled us to gain more insights into the mechanisms of these reactions and the structure-reactivity relationships of both substrates. Significantly, it was observed that the reactions of γ- and δ-azido alcohols and their larger homologs with dibenzyl N,N-diisopropylphosphoramidite could produce corresponding cyclic phosphoramidates in good yields. These yields may be improved further via optimization of reaction conditions, paving the foundation for a new and facile method to access these important molecules under neutral and mild conditions. This method could be especially useful for synthesizing large cyclic phosphoramidates, such as macrocyclic phosphoramidates.¹⁵