Synthesis of Highly Oxygenated Bicyclic Carbasugars. Remarkable Difference in the Reactivity of the d-gluco and d-xylo- Derived Trienes

2,3,4-Tri-O-benzyl-D-xylopyranose was used as a starting material in the preparation of the corresponding triene, which underwent smooth cyclization to a polyhydroxylated hydrindane, as a single diastereoisomer. The analogous triene prepared from D-glucose did not undergo any cyclization even under high pressure.


Introduction
Carbasugars are glycomimetics obtained through a substitution of the endocyclic oxygen atom with a methylene group [1,2]. Due to their similarity to carbohydrates, these compounds often possess interesting biological properties, e.g., they may act as glycosidase inhibitors [3][4][5]. Some of these compounds found an application in medicine; for example, acarbose is commercialized to treat obesity and type 2 diabetes mellitus [6] and validamycin is an antibiotic [7]. Elaboration of the convenient synthetic methods for the preparation of carbobicyclic sugar mimetics is one of the leading trends in our laboratory [8]. We have proposed a useful route to polyhydroxylated decalins (such as 5) from sugar allyltins as shown in Scheme 1 [9]. The model synthesis was initiated from D-gluco-derivative 1, which-upon treatment with mild Lewis acid (preferably ZnCl 2 )-afforded dienoaldehyde 2 with the E-configuration across the internal double bond. This compound was converted into phosphonate 3 and-by the Horner-Wadsworth-Emmons (HWE) reaction with the corresponding aldehyde-into triene 4, subsequently cyclized to 5. The methodology proposed in Scheme 1 allowed the preparation of a number of diastereoisomeric polyhydroxylated bicyclic derivatives in optically pure form starting from different sugars: d-glucose, d-mannose, and d-galactose [9][10][11][12].
It is known that polyhydroxylated decalins might have significant biological activity [13][14][15], thus there is an interest in the preparation of such targets. Our methodology, however, starting from organostannanes-being assumed to be highly toxic-cannot be used for the preparation of potential therapeutics. Other, environmentally friendly method(s), for the synthesis of such important derivatives is needed.
We have proposed an efficient method for the stereoselective preparation of sugar dienes with the E or Z geometry across the internal double bond (8-E or 8-Z; Scheme 2) from D-glucosep-derivative 6 (via intermediate 7) [16]. These compounds could be eventually used as convenient precursors for the Alcohol 21 was oxidized to the corresponding aldehyde 22 which reacted with phosphonate 23 [24], to afford triene 24 with the E-configuration across the newly created double bond (Scheme 6). This triene was subjected to the cyclization induced by Me 2 AlCl; no reaction was, however, noticed. Application of other Lewis acids, as well as, high pressure (10 kbar) also did not give any positive result. This lack of cyclization was very surprising, especially in the context of the results obtained by Evans. We prepared also two other trienes: 25 and 26 by reaction of aldehyde 22 with appropriate phosphoranes: Ph 3 P=CHCO 2 Me and Ph 3 P=CHCO 2 Me. None of the resulting trienes, however, underwent cyclization (Scheme 6).
Why could the cyclization of compounds-similar to those successfully used in Evans' synthesis-not be realized? Is it a result of only the different configuration across the double bond in a diene system (Z in 24 in our case vs. E in 14 in Evans case)? Or maybe other reasons are responsible for that strange result. This phenomenon is not clear.
We turned, therefore, our attention to shorter analogs being precursors of perhydroindane derivatives.

Synthesis of Derivatives of D-Xylose with the Diene System at the C1
The synthesis was initiated from known lactone 27 [25], obtained by an oxidation of 2,3,4-tri-O-benzyl-D-xylose. It was reacted with the in situ generated complex of Me 3 Al and N,O-dimethylhydroxylamine which provided the Weinreb amide 28. Since this compound was relatively unstable (easily underwent decomposition to 27), it was converted immediately into silylated derivative 29 and then into aldehyde 30. Reactivity of 30 was similar to the previously obtained derivatives. Its conversion into 31 followed the same route as for D-glucose derivatives (depicted in Scheme 1); this silanol (31) was also unreactive under the acidic conditions and underwent conversion into Z-diene 32 with the simultaneous removal of the silyl block upon treatment with fluoride anion (Scheme 7). The configuration of the internal double bond in the diene part was proven by 1 H NMR data in which the coupling constant J = 11.2 Hz was observed (see Scheme 7).
The synthesis was initiated from known lactone 27 [25], obtained by an oxidation of 2,3,4-tri-Obenzyl-D-xylose. It was reacted with the in situ generated complex of Me3Al and N,Odimethylhydroxylamine which provided the Weinreb amide 28. Since this compound was relatively unstable (easily underwent decomposition to 27), it was converted immediately into silylated derivative 29 and then into aldehyde 30. Reactivity of 30 was similar to the previously obtained derivatives. Its conversion into 31 followed the same route as for D-glucose derivatives (depicted in Scheme 1); this silanol (31) was also unreactive under the acidic conditions and underwent conversion into Z-diene 32 with the simultaneous removal of the silyl block upon treatment with fluoride anion (Scheme 7). The configuration of the internal double bond in the diene part was proven by 1 H NMR data in which the coupling constant J = 11.2 Hz was observed (see Scheme 7). Scheme 7. Stereoselective synthesis of bicyclic derivative 35 from D-xylose.
Compound 32 was converted into aldehyde 33 and further into triene 34, with the E-configuration across the newly created double bond (J = 15.9 Hz). Cyclization of this triene induced with Me2AlCl provided bicyclic compound 35 as a single stereoisomer (Scheme 3); only two olefinic signals at δ = 128.0 and 124.3 ppm were seen in the 13 C NMR spectrum.
The configuration of the product was assigned by the Nuclear Overhauser Effect (NOE) measurement ( Figure 1).  Small NOE values between the H1-H7a and H3-H3a confirmed structure 35 and-at the same time-excluded the alternative one: 35a in which the strong H1-H7a and H3-H3a interaction should have been observed. Compound 32 was converted into aldehyde 33 and further into triene 34, with the E-configuration across the newly created double bond (J = 15.9 Hz). Cyclization of this triene induced with Me 2 AlCl provided bicyclic compound 35 as a single stereoisomer (Scheme 3); only two olefinic signals at δ = 128.0 and 124.3 ppm were seen in the 13 C NMR spectrum.
The configuration of the product was assigned by the Nuclear Overhauser Effect (NOE) measurement ( Figure 1). Small NOE values between the H1-H7a and H3-H3a confirmed structure 35 and-at the same time-excluded the alternative one: 35a in which the strong H1-H7a and H3-H3a interaction should have been observed.

General
NMR spectra were recorded in CDCl3 (internal Me4Si) with a Varian AM-600 (600 MHz 1 H, 150 MHz 13 C) spectrometer (Sugar Land, TX, USA) at rt. Chemical shifts (δ) are reported in ppm relative to Me4Si (δ 0.00) for 1 H and residual chloroform (δ 77.00) for 13 C. All significant resonances (carbon skeleton) were assigned by COSY ( 1 H-1 H), HSQC ( 1 H-13 C), and HMBC ( 1 H-13 C) correlations. The relative configuration of the stereogenic centers was assigned on the basis of 1D-NOESY spectra. Mass spectra (ESI) were recorded with an Applied Biosystems 4000 Q-TRAP (low resolution) (Toronto, ON, Canada) and Waters AutoSpec Premier (Waters, Milford, MA, USA) or Waters MALDISynapt G2-S HDMS (high resolution) spectrometers (Manchester, UK). HPLC analyses were conducted on Merck-Hitachi apparatus (Darmstadt, Germany) equipped with Merck LiChrospher Small NOE values between the H1-H7a and H3-H3a confirmed structure 35 and-at the same time-excluded the alternative one: 35a in which the strong H1-H7a and H3-H3a interaction should have been observed.
The scans of the NMR data for all compounds are provided as Supplementary Materials.
The scans of the NMR data for all compounds are provided as Supplementary Materials.

Olefin 19
Generation of the titanium reagent 18: To a cooled to −78 • C solution of allyltrimethylsilane (3.7 g, 32.4 mmol, 10.0 eq) in dry THF (32 mL), a solution of 2.5 M BuLi in hexane (12 mL, 29.1 mmol, 9.0 eq.) was added within 1h with a syringe pump; after another 30 min. the mixture became yellowish. Then, a solution of 1.0 M ClTi(O i Pr) 3 in methylene chloride (29 mL) was added within 30 min (syringe pump) and the red mixture was stirred for another 30 min. To such prepared reagent 18, a solution of 17 (2.10 g, 3.19 mmol) in dry THF (6.4 mL) was added within 1h, the mixture was stirred overnight at −78 • C (using immersion cooler with temperature control Huber TC100E, temperature range −100 to +40 • C) and partitioned between water (70 mL) and ether (70 mL). The white precipitate was filtered off and discarded; the organic layer was separated, washed with brine, and concentrated to afford a crude mixture of (anticipated) anti-isomers, pure enough to be used in the next steps ( Figure 3). Generation of the titanium reagent 18: To a cooled to −78 °C solution of allyltrimethylsilane (3.7 g, 32.4 mmol, 10.0 eq) in dry THF (32 mL), a solution of 2.5 M BuLi in hexane (12 mL, 29.1 mmol, 9.0 eq.) was added within 1h with a syringe pump; after another 30 min. the mixture became yellowish. Then, a solution of 1.0 M ClTi(O i Pr)3 in methylene chloride (29 mL) was added within 30 min (syringe pump) and the red mixture was stirred for another 30 min.
To such prepared reagent 18, a solution of 17 (2.10 g, 3.19 mmol) in dry THF (6.4 mL) was added within 1h, the mixture was stirred overnight at −78 °C (using immersion cooler with temperature control Huber TC100E, temperature range −100 to +40 °C) and partitioned between water (70 mL) and ether (70 mL). The white precipitate was filtered off and discarded; the organic layer was separated, washed with brine, and concentrated to afford a crude mixture of (anticipated) antiisomers, pure enough to be used in the next steps ( Figure 3).

Dienoalcohol 21
To a solution of crude 19 obtained above in THF (16 mL) a solution of 1.0 M TBAF in THF (16 mL, 16.0 mmol, 5.0 eq.) was added and the mixture was stirred at room temperature overnight. Then it was concentrated and the residue was subjected to column chromatography (hexane-ethyl acetate, 92:8→36:64) to afford the title product 21 (1.11 g, 73% over two steps) as an oil ( Figure 4).

Dienoalcohol 21
To a solution of crude 19 obtained above in THF (16 mL) a solution of 1.0 M TBAF in THF (16 mL, 16.0 mmol, 5.0 eq.) was added and the mixture was stirred at room temperature overnight. Then it was concentrated and the residue was subjected to column chromatography (hexane-ethyl acetate, 92:8→36:64) to afford the title product 21 (1.11 g, 73% over two steps) as an oil (Figure 4). Generation of the titanium reagent 18: To a cooled to −78 °C solution of allyltrimethylsilane (3.7 g, 32.4 mmol, 10.0 eq) in dry THF (32 mL), a solution of 2.5 M BuLi in hexane (12 mL, 29.1 mmol, 9.0 eq.) was added within 1h with a syringe pump; after another 30 min. the mixture became yellowish. Then, a solution of 1.0 M ClTi(O i Pr)3 in methylene chloride (29 mL) was added within 30 min (syringe pump) and the red mixture was stirred for another 30 min.
To such prepared reagent 18, a solution of 17 (2.10 g, 3.19 mmol) in dry THF (6.4 mL) was added within 1h, the mixture was stirred overnight at −78 °C (using immersion cooler with temperature control Huber TC100E, temperature range −100 to +40 °C) and partitioned between water (70 mL) and ether (70 mL). The white precipitate was filtered off and discarded; the organic layer was separated, washed with brine, and concentrated to afford a crude mixture of (anticipated) antiisomers, pure enough to be used in the next steps ( Figure 3).

Dienoalcohol 21
To a solution of crude 19 obtained above in THF (16 mL) a solution of 1.0 M TBAF in THF (16 mL, 16.0 mmol, 5.0 eq.) was added and the mixture was stirred at room temperature overnight. Then it was concentrated and the residue was subjected to column chromatography (hexane-ethyl acetate, 92:8→36:64) to afford the title product 21 (1.11 g, 73% over two steps) as an oil (Figure 4).  To a cooled to 0 • C solution of alcohol 21 (350 mg, 0.62 mmol) and TEMPO (1.0 mg; 6.0 µmol) in dry CH 2 Cl 2 (8 mL), trichloroisocyanuric acid (155 mg; 0.66 mmol; 1.1 eq) was added and the mixture was stirred for 15 min. Then it was filtered through short pad of Celite, the filtrate was diluted with Et 2 O (6.0 mL) and washed with 5% Na 2 S 2 O 3 (2.0 mL), 1M NaOH (5 mL), 1M H 2 SO 4 (5 mL), and water (3.0 mL). The organic phase was dried and concentrated, and the crude aldehyde was used immediately in the next step ( Figure 5). To a cooled to 0 °C solution of alcohol 21 (350 mg, 0.62 mmol) and TEMPO (1.0 mg; 6.0 μmol) in dry CH2Cl2 (8 mL), trichloroisocyanuric acid (155 mg; 0.66 mmol; 1.1 eq) was added and the mixture was stirred for 15 min. Then it was filtered through short pad of Celite, the filtrate was diluted with Et2O (6.0 mL) and washed with 5% Na2S2O3 (2.0 mL), 1M NaOH (5 mL), 1M H2SO4 (5 mL), and water (3.0 mL). The organic phase was dried and concentrated, and the crude aldehyde was used immediately in the next step ( Figure 5).

Weinreb Amide 29
To a cooled to 0 °C suspension of MeNHOMe x HCl (5.85 g, 60.0 mmol, 3.0 eq) in dry CH2Cl2 (175 mL), a 2M solution of Me3Al in toluene (30 mL, 60 mmol, 2 eq.) was added dropwise during 30 min. by a syringe pump. The mixture was stirred for an additional 30 min., then lactone 27 (8.37 g, 20 mmol) in dry CH2Cl2 (25 mL) was added within 30 min. by a syringe pump, and the mixture was

Weinreb Amide 29
To a cooled to 0 °C suspension of MeNHOMe x HCl (5.85 g, 60.0 mmol, 3.0 eq) in dry CH2Cl2 (175 mL), a 2M solution of Me3Al in toluene (30 mL, 60 mmol, 2 eq.) was added dropwise during 30 min. by a syringe pump. The mixture was stirred for an additional 30 min., then lactone 27 (8.37 g, 20 mmol) in dry CH2Cl2 (25 mL) was added within 30 min. by a syringe pump, and the mixture was

Weinreb Amide 29
To a cooled to 0 • C suspension of MeNHOMe x HCl (5.85 g, 60.0 mmol, 3.0 eq) in dry CH 2 Cl 2 (175 mL), a 2M solution of Me 3 Al in toluene (30 mL, 60 mmol, 2 eq.) was added dropwise during 30 min. by a syringe pump. The mixture was stirred for an additional 30 min., then lactone 27 (8.37 g, 20 mmol) in dry CH 2 Cl 2 (25 mL) was added within 30 min. by a syringe pump, and the mixture was stirred at room temperature for 3 h. Aqueous H 2 SO 4 (1M solution, 100 mL) was carefully added and the organic phase was separated, washed with water (100 mL), brine (100 mL), and dried. Imidazole (4.08 g, 60 mmol, 3.0 eq) was added to this containing crude 28 and the resulting mixture was cooled to 0 • C. A solution of tert-butyldiphenylchlorosilane (7.8 mL, 30.0 mmol, 1.5 eq.) in CH 2 Cl 2 (20 mL) was added dropwise within 1 h by a syringe pump and the mixture was stirred for additional 16 h. Aqueous H 2 SO 4 (1M solution, 50 mL) was carefully added, the organic phase was separated, washed with water (100 mL), brine (100 mL), dried, and the crude product was purified by column chromatography (hexanes-ethyl acetate: 13:1→7:1) to give the title product 29 (12.35 g, 86% over two steps) as a colorless oil (Figure 9).

Silanol 31
This compound was prepared analogously as 19, from 30 (2.79 g, 4.23 mmol) as a colorless oil as a mixture of (anticipated) two anti-isomers ( Figure 11).

Conclusions
We have proposed a convenient and simple route to optically pure bicyclic carbasugar derivatives from D-xylose derivative. The methodology is based on the introduction of the Z-diene system at the anomeric position of a sugar and an olefinic unit at the terminal position. The so obtained triene underwent smooth and highly stereoselective cyclization providing the bicyclic  Figure 1 in the text).