Chemical Communications

glucoBox—a new carbohydrate-based bis(oxazoline) ligand. Synthesis and first application

Mustafa Irmak, Annika Groschner and Mike M. K. Boysen*
Institute of Organic Chemistry, Leibniz University of Hannover, Schneiderberg 1B, D-30167 Hannover, Germany. E-mail: mike.boysen@oci.uni-hannover.de; Fax: +49 511 762 3011; Tel: +49 511 762 4643

The synthesis of a new bis(oxazoline) ligand from D-glucosamine and its application in enantioselective copper(I) catalysed cyclopropanations of olefins is described.

Over the last two decades bis(oxazolines) have emerged as one of the most versatile ligand classes for metal catalysed asymmetric transformations.¹ They are prepared from chiral -amino alcohols and dicarboxylic acid derivatives by condensation^1c and are applied in several important and synthetically valuable enantioselective reactions. Among them the cyclopropanation of olefins with diazoesters first reported independently by Pfaltz et al.,^2a Masamune and co-workers^2b,2c and Evans et al.^2d is a prominent example and also a kind of sharpening stone for new ligand structures.

Even though D-glucosamine, an inexpensive amino sugar from the chiral pool, contains a -amino alcohol substructure and its N-acylated derivatives form oxazolines quite easily, this carbohydrate has rarely been employed for the synthesis of chiral oxazoline ligands. To date only a handful of examples of such structures have appeared. The groups of Kunz and Gläser³, Uemura and co-workers⁴ as well as Diéguez and colleagues⁵ reported on bidentate phosphorus-containing mono(oxazolines) based on glucosamine and their application in palladium catalysed reactions. The only contribution so far concerning carbohydrate bis(oxazolines) was published by Hartinger et al.⁶ in 2005 describing two ferrocene-bridged structures also derived from glucosamine. The publication included MS studies of palladium complexes of these ligands but no data on application have been reported so far.

In this communication we disclose a facile synthesis of a new carbohydrate bis(oxazoline) ligand with a dimethylmethylene bridge (glucoBox) from glucosamine hydrochloride. Further, the first results obtained for its application in the copper(I) catalysed cyclopropanation of olefins with ethyl diazoacetates are presented.

For ligand synthesis we selected a route via the bis(amide) of glucosamine with dimethylmalonyl dichloride and subsequent cyclisation to yield the bis(oxazoline). All attempts to form the bis(amide) directly from unprotected glucosamine and malonyl dichloride under mildly basic conditions (sodium methoxide in methanol, triethylamine in DMF) only led to decomposition of the acid chloride component. Therefore glucosamine hydrochloride (1) was first treated with trimethylsilylchloride (TMSCl) and hexamethyl disilazane (HMDS) in pyridine^7a to exclusively yield per-O-silyl protected derivative 2 as a crystalline solid. Under these conditions the amino function remains unprotected.^7b Reaction of 2 with dimethylmalonyl dichloride in the presence of triethyl amine as base, conditions originally employed for preparation of bis(amides) from cyclohexyl amine derivatives,⁸ yielded bis(amide) 3. After O-silyl deprotection with trifluoroacetic acid in methanol, the unprotected bis(amide) 4 was obtained and directly used for further transformations (Scheme 1).

	Scheme 1 Preparation of glucoBox ligand 6. Conditions: (a) TMSCl, HMDS, pyridine, rt; (b) Et3N, DCM, 0 °C to rt; (c) MeOH–TFA 9 : 1, rt; (d) AcCl, rt; (e) NaHCO3, NEt4Cl, MeCN, rt.

Most syntheses of carbohydrate oxazolines from N-acyl glucosamines use 1-O-acylated compounds as starting materials that are cyclised under Brønsted acid or Lewis acid catalysis. We therefore attempted to activate the anomeric position of peracetylated or perbenzoylated derivatives of 4 with different Lewis acids such trimethylsilyl triflate,^9a,9b stannic chloride^9c or ferric chloride.^9d We also treated the acylated derivatives with hydrogen bromide^9e,9f to subsequently activate the resulting anomeric bromide with tetraalkyl ammonium bromide via in situ anomerisation.^9g However, all these methods were not successful with our acylated precursors, leading either to incomplete transformation or extensive decomposition of the starting material. Therefore we were extremely pleased to find that the desired bis(oxazoline) 6 could be obtained directly from unprotected bis(amide) 4 by simple stirring in acetyl chloride and subsequently treating the resulting bis(glucosyl chloride) 5 with sodium hydrogen carbonate and triethylammonium chloride in a one-pot reaction. These conditions were originally described for oxazoline synthesis from N-acetyl glucosamine.¹⁰ Thus glucoBox ligand 6 was prepared from 1 in four steps in an excellent overall yield of 81% (Scheme 1).

With glucoBox ligand 6 in hand we began our evaluation studies regarding its efficiency in metal catalysed asymmetric transformations. As a first model reaction we chose the enantioselective copper(I) catalysed cyclopropanation of alkenes with diazoacetates. Styrene (8a), para-methoxy styrene (8b) and 1,1-diphenyl ethene (8c) as alkene components were reacted with ethyl diazoacetate in the presence of the preformed complex of ligand 6 with copper(I) triflate (Scheme 2).§ The results are summarised in Table 1.

	Scheme 2 Asymmetric cyclopropanations performed with copper(I)-glucoBox. Absolute configurations of the products were assigned by optical rotation and comparison with literature values (cf. electronic supplementary information).

The cyclopropanation of all alkenes proceeded in good yields with trans : cis ratios for 9a and 9b comparable to those reported in the literature.² We were pleased to find that enantiomeric excesses for all trans as well as for all cis products are well above 70%. These results are very promising, as the carbohydrate scaffold of 6 offers many opportunities for structural modifications. By optimising the ligand structure through modification, it should be possible to further increase the optical yields and thus to obtain tailor-made ligands for the cyclopropanation reaction.

Fig. 1 shows a comparison of the results achieved in the cyclopropanation of styrene (8a) with ethyl diazoacetate (7) with new ligand 6 to those reported for known bis(oxazolines)^2d derived from biogenic amino acid (S)-valine (10) and non-biogenic (R)-tert-leucine (11). The absolute stereochemistry of the major trans products is displayed as well. While all three ligands lead to essentially the same trans : cis ratios, the enantomeric excesses differ substantially. While glucoBox 6 is inferior to tert-butyl substituted ligand 11, it gives substantially better results than valine derived ligand 10. This can be explained by the steric demand of the substituent at the stereocenter adjacent to the coordinating nitrogen atom: exchanging the isopropyl residue of 10 for a bulkier tert-butyl group in 11 results in increased enantioselectivity. The steric demand of the substituent in ligand 6 (the sugar moiety) appears to be somewhere between that of the isopropyl and the tert-butyl group, as is reflected by the results obtained for the cyclopropanations.

	Fig. 1 Comparison of glucoBox ligand 6 to known bis(oxazolines).2d All values given refer to the cyclopropanation of styrene (8a) with ethyl diazoacetate (7), configurations for the major trans products are given.

It is important to note that ligand 6 is prepared in a few simple steps from inexpensive glucosamine whereas ligand 11 requires the rather costly tert-leucine as a starting material. Thus, by improving the efficiency of ligand 6 by modification, it might be possible to obtain a low-price alternative to 11.

We thank Edison Díaz Gómez for assistance with the determination of optical yields via the dirhodium method. Financial support of this work by the VW Stiftung and the Fonds der Chemischen Industrie is gratefully acknowledged.

Footnotes

Electronic supplementary information (ESI) available: Experimental details and spectra for the new glucoBox ligand, spectroscopic and other analytical details of cyclopropanation products. See DOI: 10.1039/b612986b

Spectroscopic data for glucoBox ligand 6: []D25 + 53 (c 1.9 in chloroform); H (400 MHz; CDCl3) 1.58 (6H, s, (CH3)2C), 1.98, 2.05, 2.16, (each 6H, each s, CH3CO), 3.86 (2H, ddd, J4,5 9.2, J5,6 4.8, H-5), 4.13 (2H, dd, J5,6 2.7, J6,6 12.3, H-6), 4.18 (2H, ddd, J1,2 7.5, J2,3 2.4, J2,4 1.3, H-2), 4.22 (2H, dd, J5,6 4.8, J6,6 12.3, H-6), 4.94 (2H, ddd, J3,4 2.4, J2,4 1.3, J4,5 9.2, H-4), 5.31 (2H, dd t, J2,3J3,4 2.4, H-3), 6.05 (2H, d, J1,2 7.5, H-1); C (100 MHz, CDCl3) 20.7, 20.8, 20.9 (each 2 × t, CH3CO), 23.8 (2 × t, (CH3)2C), 39.2 (q, (CH3)2C), 62.8 (2 × s, C-6), 64.8 (2 × t, C-2), 67.8 (2 × t, C-5), 68.3, (2 × t, C-4), 70.1, (2 × t, C-3), 100.0 (2 × t, C-1), 170.6, 170.2, 169.5, 169.2 (each 2 × q, CH3CO, CN).; HRMS (ESI) C29H39O16N2 [M + H]+ calcd. 671.2300, found 671.2298, C29H38O16N2Na [M + Na]+ calcd. 693.2119, found 693.2036.

§ General conditions for cyclopropanation: in a dry box CuOTf·0.5C6H6 (1.8 mg, 7.2 µmol, 1 mol%) and glucoBox ligand 6 (5.2 mg, 8 µmol, 1.1 mol%) were placed into a flame-dried flask. Under nitrogen dichloromethane (2 cm3) was added and the resulting mixture was stirred for 2 h at room temperature. To this preformed catalyst solution the alkene component (5 mmol) was added. The mixture was cooled to 0 °C and ethyl diazoacetate (0.7 mmol) dissolved in dichloromethane (1 cm3) was slowly added over a period of 2.5 h with a syringe pump at the same temperature. The mixture was allowed to warm to room temperature and was stirred for another 16 h. The solvent was removed under reduced pressure and the residue purified by flash chromatography on silica gel.