Enantioselective Synthesis of Dideoxy-tetra ﬂ uorinated Hexoses

: Carbohydrates typically have low a ﬃ nities to protein binding sites, and the development of carbohydrate mimetics with improved binding is therefore of interest. Tetra ﬂ uorination of monosaccharides is one of the strategies currently under investigation for that purpose. The synthesis of the required tetra ﬂ uorinated monosaccharides is achieved by a ﬂ uorinated building block approach. The enantioselective synthesis of tetra ﬂ uorinated hexose derivatives is described here, in both pyranose and furanose forms. In particular, the optimization of the enantioselective synthesis of the previously reported 2,3-dideoxy-2,2,3,3-tetra ﬂ uoro- D - threo -hexopyranose 3 , 2,3-dideoxy-2,2,3,3-tetra ﬂ uoro- D - threo -hexofuranose 4 , and 2,3-dideoxy-2,2,3,3-tetra ﬂ uoro- D - erythro -hexopyranose 5 is described as is the synthesis of two novel sugar derivatives, 3,4-dideoxy-3,3,4,4-tetra ﬂ uoro- D - threo -hexopyranose 6 and 3,4-dideoxy-3,3,4,4-tetra ﬂ uoro- D - erythro -hexopyranose 7 . The key step of all syntheses is a per ﬂ uoroalkyl lithium-mediated C − C bond formation, either intramolecular or intermolecular, which proceeds in good to excellent yields. NMR and X-ray crystallographic analyses of the tetra ﬂ uorinated methyl pyranoside derivatives con ﬁ rm their 4 C 1 conformation.


■ INTRODUCTION
The pronounced hydrophilicity of carbohydrates is an inherent significant contributor to the typically low affinity found for protein−carbohydrate interactions. 1 This unfavorable factor has to be taken into account when developing inhibitors of carbohydrate-processing enzymes or carbohydrate-binding proteins starting from a carbohydrate structure, at least for non-mechanism-based inhibitors. Given that carbohydrates play a central role in many fundamental processes 2 and that glycosylation of proteins and natural products can significantly alter their stability and/or biological activity, 3,4 the design of carbohydrate-based analogues with greater affinity to carbohydrate-processing proteins is of interest for use as probes or therapeutics. 2,5−9 We are interested in investigating an approach in which the carbohydrate ring is modified by extensive fluorination. The rationale for this approach is that the combination of aqueous desolvation of perfluoroalkylidene groups 10 as well as attractive multipolar interactions mediated by the individual polar C−F bonds would positively contribute to the sugar-binding affinity and selectivity. 11,12 Such C−F-mediated polar interactions have been recognized and described in detail by Diederich et al. 13−15 Recently, our group and collaborators have described the synthesis and biological evaluation of dideoxy-tetrafluorinated uridine diphosphate (UDP)−Gal analogues 1 and 2 (Chart 1) as inhibitors of the enzyme UDP−galactopyranose mutase (UGM). 16 Inhibition assays and competition STD NMR experiments clearly showed that 2 possessed much higher affinities compared to its nonfluorinated parent, and that the tetrafluorinated structures occupied the same binding site. This was confirmed by structural studies using X-ray crystallography of the Mycobacterium tuberculosis UGM, which further revealed that binding of both 1 and 2 occurred with extensive substrate− enzyme interactions, including  Hence, with the obvious caveat that sugar C−OH for C−F replacement will result in a loss of hydrogen-bond-donating capacity at these positions, these results demonstrate the potential of tetrafluorinated derivatives for investigation as potential carbohydrate mimetics as inhibitors of carbohydrateprocessing enzymes.
■ RESULTS AND DISCUSSION Retrosynthetic Analysis. The synthesis of all compounds followed a fluorinated building block approach, with the sugars fluorinated at positions 2 and 3 (sugar numbering) originating from 1-bromo-2-iodotetrafluoroethane 16 (Scheme 1) and those at positions 3 and 4 from 4-bromo-3,3,4,4-tetrafluorobut-1-ene 19 (Scheme 1). Both fluorinated building blocks are commercially available. In all cases, the key step consisted of an intramolecular 21 or intermolecular 22−24 C−C bond formation, through perfluoroalkyl lithium intermediates 9−13, all formed by MeLi-mediated halogen−lithium exchange. Two approaches are described for 6 and 7. In all cases, chirality was introduced via asymmetric dihydroxylation, with enantiopure vicinal diol intermediates 14 and 18 obtained from alkenes 15 and 19. The C4-diastereomers 3 and 5 both originated from a syn-diol, with 5 requiring inversion of configuration at one of the chiral centers of the diol. An alternative synthesis of 3 and 5 was recently described by Konno et al., using an intermolecular addition of the perfluoroalkyl anion derived from 19 with glyceraldehyde acetonide. 22 Synthesis of the Enantiopure syn-Diols 14 (Scheme 2). Following from our earlier communication, 18 the synthesis of diols 14 was optimized on a large scale. Intermolecular atom transfer addition of the 2-bromotetrafluoroethyl radical, formed by sodium-dithionite-mediated single electron transfer, to benzyl allyl ether yielded iodide 21 in excellent yield. Regioselective elimination through reaction of the most acidic proton led to alkene 15 in quantitative yield, with virtually complete stereoselectivity. This reaction required extensive optimization to achieve this result, as product and starting material were not separable, and initial attempts at large-scale optimization invariably resulted in incomplete reaction. It was found that laboratory-grade DMF (<0.2% H 2 O) gave much better conversions than did extra dry DMF (<0.005% H 2 O). These conditions gave 99% yield of 15 on a 23 g (50 mmol) scale, with 99:1 E/Z selectivity. The low temperature was required to achieve high diastereoselectivity.
The previously established conditions for the Sharpless AD reaction with the deactivated 25 alkene 15 led to 14 in 97% ee using (DHQD) 2 AQN as the chiral ligand. 18 Further optimization consisted of the development of a recrystallization protocol that resulted in virtually enantiopure material (>99% ee). On a 25 g (75 mmol) scale, a 93% reaction yield and 80% recrystallization yield was obtained, which denotes an overall yield of 75%. In addition, recovery of the mother liquor provides more diol of inferior ee, which can then be recrystallized again to increase the enantiopurity. Recovery of the expensive ligand was easily achieved by a modified extraction method. 26 Formation of the enantiomeric diol ent-14 was achieved by using a catalyst having the AQN spacer and the pseudoenantiomeric DHQ ligand, which resulted in a 97% ee, which was increased to 99% by recrystallization. On a 14 g (40 mmol) scale, this transformation was achieved in an 84% yield, with an 80% yield for the recrystallization, accounting for an overall yield of 67%.
Pyranose Ring Formation: Synthesis of the Dideoxytetrafluoro "Galactopyranose" 3 (Scheme 3). The required selective benzylation is possible due to the increased acidity of the hydroxyl group proximal to the perfluoroalkyl moiety. 27 Thus, deprotonation was carried out using 1 equiv of a strong base (NaH) followed by treatment with 1 equiv of BnBr. The reaction has now been performed on a large scale (12 g, 30 mmol), giving a 79% yield. The dibenzylation product (23) was also isolated in 5% yield, and 16% yield of starting diol was recovered. None of the undesired "reverse" monobenzylation product has ever been observed, suggesting that the first benzylation proceeded with complete selectivity. The next step is the formylation of the remaining hydroxyl group in 22. Although this was previously achieved using diisopropyl carbodiimide (DIC) and formic acid in the presence of dimethylaminopyridine (DMAP), we serendipitously found that the use of a Vilsmeier−Haack-type reagent led to superior results. Activation of DMF with tosyl chloride in pyridine solvent is presumed to generate iminium salts 24 or 25 that then react with alcohol 22 to give the intermediate 26 Aqueous workup then results in hydrolysis and loss of dimethylamine to give the formate 27, which was confirmed to proceed with retention of configuration. This reaction has been achieved in a 92% yield on a 12 g (25 mmol) scale and has been employed in favor of the DIC/HCO 2 H method as it is more reliable, efficient, and uses cheaper reagents. More importantly, a simple workup consisting of extraction into hexane provided clean product suitable for the next step, with no chromatography necessary. Finally, cyclization to give pyranose 28, achieved in a yield of 81% on an 11 g (20 mmol) scale, and hydrogenolysis of the benzyl ethers gave the deprotected dideoxy-tetrafluorinated sugar 3 in 99% yield on a 1 g (3 mmol) scale.
Pyranose Ring Formation: Synthesis of the Dideoxytetrafluoro "Glucopyranose" 5. Using the strategy used for the synthesis of the dideoxy-tetrafluoro "galactose" 3, the synthesis of the diastereomeric "glucose" 5 would require asymmetric dihydroxylation of (Z)-15. However, a highly diastereoselective synthesis of (Z)-15 was expected to be cumbersome (separation with the E-isomer is not possible), and its asymmetric dihydroxylation would unlikely proceed with high enantioselectivity. Hence, a synthesis starting from (E)-15 was envisaged that involved an inversion of configuration at one carbon center, for example, by opening of the corresponding epoxide. Unfortunately, all attempts to effect a Sharpless asymmetric epoxidation on the corresponding (E)allylic alcohol failed (not shown), which led us to investigate inversion of configuration of the alcohol group at C4 in 14 or at C5 in ent-14 (sugar numbering). Because S N 2 reactions next to perfluorinated carbon atoms are difficult, inversion at C5, through activation of the alcohol group in ent-22 was investigated first (Table 1). Mitsunobu inversion 29,30 with formic acid gave no reaction (entry 1). With chloroacetic acid, 31 a tiny amount of inversion product 32 was isolated, in addition to the elimination product 33 (entry 2). The Zconfiguration of 33 showed that elimination occurred at the activated alcohol stage. Isourea-mediated alcohol inversion also proceeds with inversion of stereochemistry, 32,33 but all attempts with ent-22 led to recovered starting material (entry 3) or direct ester formation with retention of configuration (entry 4, leading to 34). The relative configuration of 34 was proven by ester cleavage to give ent-22 (not shown). Then, the alcohol group was activated as mesylate 29. Reaction with cesium formate, 34,35 with 36 or without DMAP, only led to elimination (entries 5 and 6). Following a procedure by Otera using formic acid and CsF, 37 we observed only elimination product (entry 7). Finally, reacting the triflate 30 with sodium nitrite (entry 8) did give the desired inverted alcohol 31, but it was again accompanied by elimination product as well as a range of decomposition products.
Clearly, the increased C−H acidity due to the fluorination, combined with the deactivation toward S N 2 reaction by the electronegative substitution pattern, promoted E2 elimination reaction, and this line of research was terminated. When the MOM-ether 35 became available (see below), inversion of the alcohol group next to the fluorination was investigated ( Table  2). Activation as the corresponding triflate 36 was highyielding, provided that short reaction times were employed given the lability of the MOM protecting group under the reaction conditions. Triflate 36 was noticeably more stable compared to triflate 30, due to the electronegativity of the tetrafluoroalkylidene group.
Pleasingly, displacement with NaNO 2 at room temperature (entry 1) gave the inversion product, even if in low yield. However, no elimination product was observed, and starting material was recovered. Raising the temperature (entry 2) increased the yield of 37 but unfortunately also gave rise to the elimination side reaction. Again, the alkene configuration indicated that the elimination process took place from the starting material. The addition of a crown ether 38 (entry 3) or attempting to generate Bu 4 NNO 2 (entry 4), which was reported as an effective nucleophile for triflate displacement, 39 did not lead to improved results. Lowering the temperature to an intermediate 40°C did lead to a reasonable 52% yield (entry 5) but still with a substantial amount of elimination byproduct. Carboxylate-based nucleophiles were not successful. With benzoate, 40 a low yield of 39 was obtained, but the major product was the elimination byproduct (entry 6). With trichloro-and trifluoroacetates (entries 7−9), mainly elimination was observed for incomplete conversions, even with cesium trifluoroacetate. 41 Hence, it was decided to return to the previously developed cyclic sulfate route (Scheme 4), in which the regioselectivity is due to the fluorination hampering S N 2 reactions in adjacent positions. 18 The cyclic sulfate formation was further optimized to give excellent overall yield on a large scale (10 mmol), mainly achieved by modifying the workup procedure. The direct cyclic sulfate formation using SO 2 Cl 2 was also investigated but only gave a 68% yield. As reported, the direct formation of the formate ester was shown to be low-yielding, due to its lability under the conditions of the required subsequent hydrolysis of the sulfate group. Hence, the reaction was optimized toward complete formate hydrolysis to obtain the anti-diol. After extensive efforts, it was found that nonaqueous conditions involving HCl generated in situ (AcCl, MeOH) were superior to the use of aqueous sulfuric acid. The use of 3 equiv of in situ generated HCl gave the desired antidiol 41 in an acceptable yield on an 8 mmol scale. Interestingly, on this scale, ketone 42 was isolated as a byproduct, which presumably is formed via competitive elimination of the formate group followed by equilibration to the keto-tautomer. Finally, with the anti-diol 41 in hand, completion of the synthesis of 5 was achieved in a manner similar to that shown for 3 (Scheme 5), via selective diol benzylation, which, given its anti-stereochemistry, was lower-yielding than for the syn-diol as reported in Scheme 3, formate introduction, and anionic cyclization.
Furanose Ring Formation: Synthesis of (Protected) Dideoxy-tetrafluoro "Galactofuranose". The synthesis of tetrafluorinated furanose 47 has been communicated as part of the synthesis of the UGM inhibitor 2. 16 Briefly (Scheme 6), starting from 14, selective silylation of the more nucleophilic alcohol 42,43 yielded the desired silyl ether 45 in good yield and selectivity (6% of the regioisomer and 10% of bis-silylated derivative, not shown). Formylation of 45 with TsCl/DMF never reached completion, no doubt due to the reduced nucleophilicity of the hydroxyl group in close proximity to the perfluoroalkyl group, but could be accomplished with the conventional formic acid/DIC method. Finally, MeLi-induced anionic cyclization led to the protected dideoxy-tetrafluoro furanose derivative 47. However, a significant amount of the isomerized pyranose 48 was also isolated.
As shown in Scheme 7a, Kitazume had described a similar isomerization process in which deprotonation of the furanose 49 led to ring opening and silyl transfer, followed by cyclization to give the pyranose 54 in quantitative yield. 42,43 The direction of the equilibrium was explained by the thermodynamic driving force toward formation of the more stable pyranose form, with the equilibrium easily established due to an energetically favorable silyl migration (pK a difference of the alcohols). Hence, formation of 48 (Scheme 7b) is similarly explained starting from anion 55, which is obtained from the cyclization step. While the isomerization of 49 to give 54 was reported to be complete in 3 h at −78°C, it is interesting to observe that the anionic cyclization step starting from 46 as described above (Scheme 6), which took 4.5 h at temperatures up to −60°C, only led to the formation of 48 in 11% yield. This suggests a much slower isomerization process, due to the higher stability of anion 56 compared to that of anion 57.
Indeed, in a separate experiment in which 47 was subjected to the Kitazume isomerization conditions (Scheme 8a), only minimal reaction to 48 was observed (TLC) at −78°C, and quantitative conversion only occurred after warming the reaction mixture to room temperature overnight. This clearly confirms that the tetrafluorinated pyranose form is the more stable ring and, as expected, that the silyl migration from 56 to  57 is an energetically unfavorable process with a low reaction rate at low temperature. Unfortunately, attempts to prevent the formation of 48 in the anionic cyclization reaction starting from 46 by decreasing the reaction temperature, or by shorter reaction times, only resulted in incomplete reaction.
The much larger stability of the tetrafluorinated pyranose was further confirmed by rapid isomerization upon silyl cleavage (Scheme 8b). Hence, to avoid the isomerization, a different protecting group was used (Scheme 9). Reaction of diol 14 with MOMCl led to the desired monoprotected 35 in 70% yield (contaminated with an additional 5% of 60, which was separated after the next step). With 35 in hand, formylation of the remaining alcohol group and anionic cyclization gave the desired furanose 8 without any isomerization.
Pyranose Ring Formation: Synthesis of 6 and 7. Given the success of the anionic cyclization reaction, this strategy was initially investigated for the synthesis of the 3,3,4,4-tetrafluorinated sugar derivatives 6 and 7 (Scheme 10), which here required the formation of the C2−C3 bond. Hence, starting from the (racemic) alcohol derivative 62, obtained in two steps as described previously, 26 functionalization with α-bromoacetate esters to obtain suitable cyclization precursors was required. Initially, alkylation with methyl bromoacetate 63 was carried out to investigate the cyclization on a simplified precursor 65. Pleasingly, using the standard conditions, the 1deoxysugar derivative 67 was obtained in very good yield and was isolated mainly as the hydrate, as shown. The structure of the hydrate was proven by X-ray crystallographic analysis ( Figure 1). Unfortunately, subsequent functionalization of the anomeric position of 67 using radical bromination was unsuccessful (not shown).
Hence, the synthesis of precursor 66, already containing an anomeric substituent, was envisioned. This was achieved in high yield by reaction of 62 with known bromoether 64, 44 which was obtained by radical bromination of methyl methoxyacetate (not shown). Also, on this substrate, anionic cyclization proceeded in excellent yield, leading to an inseparable mixture of anomers. Subsequent reduction of the C2 keto group led to a mixture of four diastereomers, with the existing anomeric configuration directing the hydride attack. 45,46 Hence, the cis-1,2-disubstituted diastereomers β-70 and α-71 were obtained as major products, each as a 10:1 mixture with the other C2 epimer. Unfortunately, despite extensive efforts, a high-yielding and complete separation of these compounds was never achieved. In addition, anomeric deprotection proved to be difficult. Hence, a different approach involving intermolecular addition with 13 (cf. Scheme 1) leading directly to hemiacetal structures was investigated (Scheme 11).
In contrast to Konno's strategy, 22 which involved lithiation of 19 and reaction with a chiral aldehyde, the synthesis of 6/7 called for the use of a chiral lithiated fluorinated building block and an achiral aldehyde. Hence, 19 was converted to enantiopure monoprotected diol 72 on a large scale as described by us previously. 26 Analysis via the corresponding Mosher ester derivative confirmed its enantiopurity (>99% ee; see Supporting Information). Protection of the remaining alcohol was initially achieved by benzylation or p-methoxybenzylation, and while the generation of the corresponding lithiated species and intermolecular addition to cinnamaldehyde proved high-yielding (80%, not shown), it was decided, in the interest of atom economy, to protect the secondary alcohol as naphthylmethylidene acetal 73 (Scheme 11). Hence, DDQmediated oxidation of 72 under anhydrous conditions 47 led to 73 as a crystalline 1:1 mixture of diastereomers, which could be easily separated by column chromatography. Assignment of relative stereochemistry was achieved by X-ray crystallographic analysis of both isomers (see Supporting Information).
Starting from each acetal diastereomer, bromine−lithium exchange followed by reaction with cinnamaldehyde proceeds in excellent yields (80%), but in each case, an inseparable 1:1 mixture of alcohol diastereomers was obtained (not shown). Hence, on a large scale (7 g), the acetal diastereomers were not separated before lithiation and cinnamaldehyde addition, leading to 74 as a mixture of four diastereomers. These were not separated and directly subjected to acetal deprotection, leading to an inseparable mixture of syn and anti isomers 75 in 71% overall yield. An added advantage of this two-step procedure is that the small amount of MeLi/cinnamaldehyde addition product formed in the first step, which is inseparable from the addition products 74, is easily removed after acetal cleavage. In addition, the ozonolysis step is not compatible with the naphthyl acetal protecting group. Finally, ozonolysis provided a mixture of sugar derivatives 6 and 7 in 97% yield (89% on multigram scale) after removal of the benzaldehyde and DMSO byproducts by column chromatography. Unfortunately, separation of these sugar derivatives was not possible. Derivatization as the peracetates 76 and 77 proceeded in excellent yield (Scheme 12) but did not allow for a practical diastereomeric separation.
Interestingly, β-77 could be obtained as pure crystals and analyzed by X-ray diffraction (see below). Analogous perbenzoylation was also achieved but proved equally ineffective in separating the C2-diastereomers (not shown).
During the extensive derivatization/separation efforts, it was noticed that very often the two diastereomers having the same anomeric configuration but different C2-stereochemistry were separable. Hence, a protection strategy aiming at selective glycoside formation was pursued. It was found that after 6-OH protection as silyl ether (Scheme 13), anomeric naphthylmethylation proceeded with excellent yield and β-selectivity. 18,48,49 Only low amounts (<3%) of α-anomers could be observed by 19 F NMR. Interestingly, 6.5% of presumably 2-naphthylmethylated "tetrafluoromannose" byproduct was formed during the reaction, whereas no regioisomer could be observed for the "gluco" analogue. TBDMS cleavage then gave the free naphthylmethyl glycosides 82 and 83, which were separable by column chromatography. Finally, the individual sugar analogues 6 and 7 could be obtained after hydrogenolysis using Pearlman's catalyst.
Conformational Analysis. The conformation of carbohydrates is an important element in considering protein− carbohydrate interactions, and hence, conformational analysis of modified carbohydrates is of interest. 50,51 Solution-phase NMR studies (2D 1 H− 19 F HOESY) of the spectra of both αand β-methyl glycosides of 3 in CDCl 3 and D 2 O have been reported previously and were consistent with a 4 C 1 chair conformation. 19 These studies have been extended with the novel sugar derivatives described herein, and given the extensive spectral overlap of the free hemiacetals prevented clear analysis by 1D and 2D 1 H− 19 F HOESY NMR, they were conducted with the corresponding methyl glycosides. The syntheses of the methyl glycosides are shown in Scheme 14. Anomeric alkylation of 44 with methyl iodide followed by benzyl hydrogenolysis led to the methyl glycoside derivatives α-85 and β-85. For the 2-OH sugar derivatives, the mixture of 70 and 71, obtained as described above, was subjected to hydrogenolysis, leading to β-86, which was obtained in pure form, and α-87, which could not be separated from small amounts of its β-anomer.
Clearly, the 1 H− 1 H NOESY and 1 H− 19 F HOESY analysis shown in Figure 2 shows that all methyl pyranoside structures adopt a 4 C 1 conformation, both in deuterated chloroform and in water. This confirms the minimal influence of dideoxytetrafluorination on the monosaccharide chair conformation.
Crystal structure analysis of heavily fluorinated carbohydrate derivatives has also shown that the 4 C 1 chair conformation is retained, with generally minimal distortion. Examples include the 1,6-dibenzoate ester of a hexafluorinated pyranose described by DiMagno, 11,12 as well as structures from our group such as 28, 20 the αand β-methyl glycosides of 3, 19 and the UDP derivative 1. 17 In addition to the structure of 67, shown in Figure 1, the structures of the other crystalline sugar derivatives described above (α-58, and β-77), as well as that of β-88, which was isolated after an incomplete hydrogenolysis of 28 (Scheme 15), were obtained and are shown in Figure 3. These crystal structures also show a relatively undistorted 4 C 1 chair conformation. The crystal packing of all structures (see Supporting Information) shows that hydrogen bonding of the alcohol groups with oxygen-containing groups is maximized, an effect which presumably determines the anomeric configuration of the hemiacetals: 20 the benzyl ether 58 crystallizes as the αanomer, while 88 and 77 are obtained as the β-anomer.
Pleasingly, both unprotected 3,3,4,4-tetrafluorinated sugar derivatives 6 and 7 proved to be crystalline (Figure 4), and crystallization was achieved from hexane/acetone. These are the first crystal structures of fully deprotected tetrafluorinated pyranoses. Both compounds crystallized as the β-anomer, with the hydroxymethyl group as the gt-rotamer. For 6, the axial C2 hydroxyl group and C4 fluorine atom are somewhat splayed (13.7°). Their corresponding methyl glycosides were also crystallized and showed very similar conformations.
Where  The Journal of Organic Chemistry Featured Article molecule allow for 3D network formation; however, the change of the 2-OH configuration is enough to disrupt this network such that 7 now forms 2D sheets in the bc plane. The arrangement in the third direction is now directed by weaker interactions with hydrophobic F surfaces stacking along the a axis, forming alternating hydrophilic and hydrophobic layers. Similar molecules 58 and 67 also form sheets (bc plane) with a layer structure characterized by interdigitating benzyl groups at the sheet interfaces. Compounds 82, 86, and 87 have supramolecular structures dominated by hydrogen-bonded ladder chains aligned in a manner that colocates the benzyl groups of adjacent chains, forming hydrophobic columns. This colocation of certain groups gives rise to a related colocation of the CF 2 groups. In 82, hydrogen-bonded ladder chains are still dominant, but the larger nature of the naphthalene rings now gives rise to a more pronounced layer structure with pseudoparallel naphthalenes. Compounds cis-73 and trans-73 have structures dominated by interdigitation of the naphthalene rings forming hydrophobic layers. Interestingly, the change between cis/trans is enough to significantly change the level of interdigitation and switch the arrangement from herringbone to pseudoparallel. Compound β-77 has neither hydrogen bonding opportunities nor peripheral aromatic rings, and its packing is thus directed by van der Waals interactions.

■ CONCLUSIONS
It has been shown that the intramolecular tetrafluoroalkylidene lithium addition to ester groups is a suitable method for the synthesis of tetrafluorinated monosaccharides, even on a large scale. The formation of both the C1−C2 (tetrafluorination at C2/C3) as well as the C2−C3 (tetrafluorination at C3/C4) bonds (sugar numbering) is possible. Controlling the protection group pattern allows selective formation of the furanose or the pyranose form. In addition, following the work of Konno for the synthesis of 2,2,3,3-tetrafluorinated pyranose derivatives, an intermolecular strategy is also demonstrated here for the synthesis of pyranose derivatives with tetrafluorination at C3/C4. For all sugar derivatives investigated, NMR and Xray crystallographic analyses show that they exist in the 4 C 1 conformation, with only minimal distortion from the ideal chair conformation. The monosaccharides synthesized will be of interest for the synthesis of carbohydrate mimetics for possible use as inhibitors or probes for carbohydrate-processing enzymes.

■ EXPERIMENTAL SECTION
All air/moisture-sensitive reactions were carried out under an inert atmosphere (Ar) in oven-dried glassware. Solvents distilled prior to use: CH 2 Cl 2 (from CaH 2 ), THF (from Na and benzophenone), and MeCN (from CaH 2 ). Where appropriate, other reagents and solvents were purified by standard techniques. TLC was performed on aluminum-precoated plates coated with silica gel 60 with an F254 indicator; they were visualized under UV light (254 nm) and/or by staining with KMnO 4 (10% aq). Flash column chromatography was performed with silica gel (40−63 nm). Chemical shifts are reported in δ units using CHCl 3 as an internal standard. Fourier transform infrared spectra were measured using an ATR accessory using neat samples (solids and liquids). Electrospray mass spectra were run in HPLC methanol or MeCN. HRMS samples were run on an ESI-TOF MS or an ESI FT-ICR MS spectrometer. Optical rotations were measured at 589 nm, and all reducing carbohydrate derivatives were equilibrated in the used solvent for 3 days prior to measurement.