Allyl Cyanate/Isocyanate Rearrangement in Glycals: Stereoselective Synthesis of 1-Amino and Diamino Sugar Derivatives.

The [3,3]-sigmatropic allyl cyanate/isocyanate rearrangement of glycals in the presence of O-, N-, and C-nucleophiles afforded β-N-glucosyl and galactosyl carbamates, ureas, and amides in good yields. The unsaturated products were elaborated to N-glycosides by dihydroxylation, to 1,3-diaminosugars by tethered aminohydroxylation, or to 1,2-diaminosugars by iteration of the sigmatropic rearrangement. This metal-free methodology represents an excellent and general method for the stereoselective synthesis of N-glycosides and diamino sugars with complete transmission of stereochemical information.

Aminosugars are widespread in Nature, possessing a variety of biological roles.[1]n class="Chemical">N-Glycosides (1-aminosugars and derivatives)[2] are a subclass which may serve as synthetic intermediates, besides playing a key role in N-linked glycoproteins involved in many recognition events or in nucleosides and antibiotics.[2b] Representative examples (Figure ) include the N-acetyl-d-glucosamine asparagine moiety 1, a key agent in cell-recognition and signal transduction,[3] glycocinnamoylspermidines 2, broad spectrum antibacterial agents,[4] staurosporine 3, displaying anticancer activity,[5] and trehazoline 4, a potent and specific inhibitor of trehalase.[6]

Figure 1

Examples of natural N-glycosides.

Several synthetic efforts have been devoted to accessing diverse N-glycosides (n class="Chemical">amides, ureas, heterocycles), in view of their potential for medicinal chemistry.[7,8] However, compared to O- and C-glycosides, efficient and stereoselective syntheses of N-glycosides are less investigated and remain challenging.[1d−1f,9] Major issues are formation of anomeric mixtures and configurational stability. Selectivity for a single anomer is an important target and represents a substantial synthetic challenge. Synthesis of 2- and 3-aminosugars and challenging 1,2(or 3)-diaminosugar backbones, such as in compounds 1–3, is also of paramount importance,[1] and new versatile methods for achieving this task are desired.

We tackled this topic following our interest in iminon class="Chemical">sugar glycomimetics,[10] recently extended to aminosugars.[11] A straightforward way for introducing a nitrogen atom onto a carbohydrate backbone involves the use of unsaturated sugar derivatives. Glycals (1,2-deoxyanhydro carbohydrates, 5) are readily accessible from inexpensive sources and serve as useful precursors for the synthesis of 1-aminosugars.[12] Appropriate strategies rely on epoxidation of glycals followed by regioselective epoxide opening with an N-nucleophile (Scheme , route a)[13] and related approaches.[14,12] Alternatively, introduction of the nitrogen moiety at C-1 with a concomitant double bond shift has been pursued through Ferrier rearrangement (Scheme , route b). However, this approach met with limited success and required peculiar N-nucleophiles and use of catalysts, affording mostly moderate yields of α/β anomeric mixtures.[15] The O-allyl to N-allyl rearrangements (cyanate/isocyanate[16] or Overman[17]), reliable methods for introducing nitrogen moieties on allyl alcohol skeletons with high degree of stereochemical control, might serve the purpose. These methods are largely neglected in synthesis; particularly, the allyl cyanate/isocyanate rearrangement has been used in limited examples of elegant syntheses mainly by Ichikawa following his reliable procedure for generating the required cyanate functionality[16a,18] and more recently by Stecko’s and Carreaux’s groups.[19] Although glycals and other unsaturated sugars are readily affordable, applications of these methodologies to carbohydrates are rare. Only Ichikawa and co-workers have studied the allyl cyanate/isocyanate rearrangement with unsaturated sugars, however focusing on 2- or 3-hexenopyranosides for the synthesis of 2- and 4-aminosugars, respectively.[20] Donohoe and co-workers also synthesized 2-aminosugars applying the related Overman rearrangement to 2-hexenopyranosides derived from Ferrier reaction of glycals.[21]

Scheme 1

Known Procedures for the Synthesis of N-Glycosides from Glycals and the Strategy Reported Herein Based on the Allyl Cyanate/Isocyanate Rearrangement

Pd(II)-catalyzed rearrangements of glycal imidates to yield α- and β-N-glycosyl trichloroacetimidates (Scheme , route c)[22] or N-heterocyclic glycosides[23] have been reported. However, no example of a thermal Overman or cyanate/isocyanate rearrangement on glycals has been described, despite the value of N-glycosides. The latter rearrangement is particularly attracting, in view of several advantages, including use of manageable carbamate precursors, much milder rearrangement under metal-free conditions, high degree of stereocontrol, and versatility of the resulting isocyanates. Absence of applications to glycals for accessing glycosyl isocyanates is surprising and suggested that the envisioned rearrangement might be challenging. Ichikawa reported the synthesis of N-glycosyl isocyanates, generated however by oxidation of the corresponding isonitriles, obtained in turn from glycosyl azides.[24] An additional pro of the envisaged rearrangement rests in the opportunity to manipulate the shifted C=C double bond for installing other oxygen and/or nitrogen functionalities.

In this paper, we report our results on the application of the allyl cyanate/isocyanate rearrangement to glycals to give N-glycosides (Scheme , route d) and diaminosugar derivatives.

In order to synthesize allyl n class="Chemical">carbamates of simple d-glycals, 4,6-O-protected d-glucal 5a and d-galactal 5b(25) were reacted with trichloroacetyl isocyanate followed by potassium carbonate to give 6a and 6b in quantitative yield (Scheme ). Their transformation to the desired products involves three steps,[16] which are usually performed in one-pot. Initial dehydration of the carbamate to the corresponding elusive cyanate 7(26) is followed by spontaneous [3,3]-sigmatropic rearrangement to the isocyanate 8, which is conveniently trapped with a nucleophile to afford the final products 9.

Scheme 2

Synthesis of the d-Glucal Carbamate 6a and the d-Galactal Carbamate 6b and the Expected Rearrangement to Yield the N-Glycosides 9

The d-glucal carbamate 6a was selected for studying the process and optimizing the reaction conditions. As anticipated, however, the rearrangement proved troublesome. All attempts employing the sn class="Chemical">tandard Appel’s conditions for dehydration (CBr4, PPh3, NEt3 in CH2Cl2, entries 1 and 2, Table ) were unsuccessful. Addition of BnOH as nucleophile either at rt or 0 °C after consumption of the starting material resulted in a complex mixture of products from which isolation of 10a was impossible, providing a hint for the dearth of reports of this rearrangement on glycals. Early formation (TLC) of the complex mixture after addition of the dehydrating reagents suggested that carbamates 6 are sensitive to these conditions, possibly due to competitive Ferrier type processes. Then, use of trifluoroacetic anhydride (TFAA) as the dehydrating agent, according to recent work by Stecko,[19] was investigated. No reaction was obtained by treatment of 6a with TFAA (2 equiv) and NEt3 (3 equiv) in THF at −20 °C (entry 3, Table ), but raising the temperature to 0 °C led to its complete and clean conversion (entries 4–6). Optimal conditions resulted by using 3 equiv of BnOH at 0 °C for 3 h and afforded the desired benzyl carbamate 10a in 92% yield (entry 6).

Table 1

Reaction Conditions of the Three-Step Reaction of Scheme with 6a and Benzyl Alcohol As Nucleophile

	step 1		step 2
entry	reagents	temp °C, time	BnOH (equiv)	temp (°C), time (h)	yield of 10a (%)
1	CBr4, PPh3, NEt3,CH2Cl2	0, 6 h	1	rt, 16	complex mixture
2	CBr4, PPh3, NEt3,CH2Cl2	0, 10 h	1	0, 3	complex mixture
3	TFAA, NEt3, THF	–20, 35 min			no reaction
4	TFAA, NEt3, THF	0, 35 min	6	–20, 3	80
5	TFAA, NEt3, THF	0, 35 min	6	0, 3	92
6	TFAA, NEt3, THF	0, 35 min	3	0, 3	92

Gratified by this result, we addressed the scope of the process with both d-glucal and d-galactal carbamates 6a and 6b. Several O-, N-, and C-nucleophiles were used to yield n class="Chemical">N-glucosyl and N-galactosyl carbamates, ureas, and amides a and b, respectively (Table ). Alcohols gave carbamates 10–13, several of which display common amino protecting groups at the anomeric amine, such as benzyloxy, methoxy, and fluorenylmethoxy carbonyl (Cbz, Moc, and Fmoc, entries 1–3) suitable to be removed under different conditions.[9a] Ammonia and primary and secondary amines gave the corresponding ureas 14–17 (entries 5–8). Most of the nucleophiles afforded the desired N-glycosides in good to excellent yields, considering the challenging three-step transformation. The yield dropped with bulky nucleophiles, as in case of the fluorenylmethyl alcohol. Addition of MeMgBr as a model C-nucleophile also gave the expected amides in more moderate yields (entry 9). In the reaction from 6b an unexpected product, characterized by NMR and MS as an aminal derived from a double addition and trapped as its trifluoroacetic ester (see the SI), was also isolated besides the desired acetamide 18b.

Table 2

Nucleophiles and Reaction Products of the Cyanate/Isocyanate Rearrangement from d-Glucal and d-Galactal Carbamates 6a and 6b

Yield of the double addition product (see text and SI).

All products 10–18a/b were obtained as single isomers, assumed to be the β-anomers in consideration of the concerted mechanism of the [3,3]-sigmatropic rearrangement through the cyclic TS 19 occurring with complete 1,3-chirality transfer (Scheme ).[16,3,3] Sigmatropic Cyanate–Isocyanate Rearrangement. Tetrahedron Lett.. 2001 ">26,27] However, formation of the α-anomer which may experience anomeric effect[2b,9a,28] consequent to a different mechanism, e.g., a Ferrier-type, could not be ruled out. Then, single crysn class="Chemical">tals of products 18a from d-gluco and 10b from d-galacto series were subjected to X-ray structural determination (SI), which confirmed the N-glycosides β-configuration.

Scheme 3

TS Involved in the Rearrangement

We have then proved the feasibility and generality of the desired process, which works satisfactorily in both gluco and galacto series under mild conditions without the use of metal mediators, affording β-n class="Chemical">N-glycosides in good yields and with complete stereocontrol.

In order to illustrate the potential of this methodology for accessing a variety of rare carbohydrate derivatives, such as those of n class="Chemical">allose or gulose, which constitute the backbone of important biologically active substances and are not economically accessible from natural sources,[29] we aimed to elaborate the obtained unsaturated products into amino and diamino carbohydrates. Dihydroxylation to N-glycosides was demonstrated with the reaction of the d-galacto carbamate 11b under Upjohn conditions,[30] which afforded the single gulose derivative 20 derived from selective attack of OsO4 to the α-face (Scheme ). Final desilylation of 20 with TBAHF in THF gave the β-N-guloside 21 in 87% yield.

Scheme 4

Synthesis of N-Guloside 21 by cis-Dihydroxylation of Hexenopyranoside 11b and of 1,3-Diaminoallose 25 by Tethered Aminohydroxylation of Hexenopyranoside 24

Access to 1,3-diaminosugars was secured via an n class="Chemical">osmium-catalyzed tethered syn-aminohydroxylation (TA)[31] of 2-hexenopyranosides, extending the methodology on glycals reported recently by us for the synthesis of 2-aminosugars.[11] Toward this goal, a sequential deprotection of the hydroxy groups of 10a followed by selective protection of the primary alcohol gave 22 in 86% yield (Scheme ). Conversion of the free OH to the suitable p-chlorobenzoyloxy carbamate 24 for the key Os-catalyzed TA was achieved as reported.[11] Reaction of 24 with 1 mol % K2OsO2(OH)4 in tBuOH/water 3:1 for 2 d gave the isoxazolidinone 25, a protected 1,3-diamino allose, in a satisfactory 60% yield. The reaction occurred with complete cis-stereoselectivity and afforded exclusively the diastereoisomer with the installed functionalities at C-2 and C-3 trans to the anomeric substituent, as attested by a large diaxial H-1/H-2 coupling constant in the 1H NMR spectrum, in agreement with the proposed mechanism through a key N-imido osmium TS. The accomplishment of the TA reaction from 24 is noteworthy, considering our previously experienced failure of the same reaction from a related d-gluco derivative bearing an α-OMe at C-1.[11]

Having proved the access to 1,3-diaminosugars by n class="Chemical">TA, we envisaged that iteration of the cyanate/isocyanate rearrangement would furnish complementary alternative for introduction of a second amino group at C-2 of the carbohydrate. This opportunity arises since products 10–18 of the one-pot rearrangement/nucleophilic addition process contain a protected allyl alcohol. Iteration of the process would allow to access 1,2-diaminosugars leaving another double bond suited for further elaboration. The allylic alcohol 22 was converted as before to the carbamate 26 in 82% yield. The one-pot dehydration/rearrangement was performed as usual and the resulting isocyanate trapped in situ using MeOH as the nucleophile, affording the orthogonally protected 1,2-diamino sugar 27 in 51% yield with complete transmission of the α-configuration (Scheme ). The trans relationship of the two aminocarbonyl groups in 27 was assigned on the basis of the large H-1/H-2 coupling constant (8.5 Hz) in the 1H NMR spectrum.[32] It is worthy of note that compound 27 possesses the same stereochemical 1,2-diamino pattern of N-glycosides 1 in Figure .

Scheme 5

Iterative Allyl Cyanate/Isocyanate Rearrangement to Yield the 1,2-Diamino Sugar 27

In conclusion, we proved the viability of the [3,3]-sigmatropic allyl n class="Chemical">cyanate/isocyanate rearrangement with glycals to give a straightforward access to various functionalized β-N-glycosides with complete chirality transfer by reaction with several O-, N- and C-nucleophiles. The resulting hexenopyranosides were elaborated to 1-amino and 1,3-diaminosugars by cis-dihydroxylation and tethered aminohydroxylation, respectively. 1,2-Diamino unsaturated sugars were also obtained by double allyl cyanate/isocyanate rearrangement. The variety of glycals available and all possible elaboration of the rearranged products make this simple and mild method widely applicable in principle to the synthesis of constitutionally and configurationally diversified aminosugars.

Further studies are ongoing in our laboratory to extend this methodology to other carbohydrate derivatives and to address the scope of the multiple rearrangement for the conversion of chiral polyols to polyamines.