Synthesis of Disaccharides Containing 6-Deoxy-α-L-talose as Potential Heparan Sulfate Mimetics

A 6-deoxy-α-L-talopyranoside acceptor was readily prepared from methyl α-L-rhamnopyranoside and glycosylated with thiogalactoside donors using NIS/TfOH as the promoter to give good yields of the desired α-linked disaccharide (69–90%). Glycosylation with a 2-azido-2-deoxy-D-glucosyl trichloroacetimidate donor was not completely stereoselective (α:β = 6:1), but the desired α-linked disaccharide could be isolated in good overall yield (60%) following conversion into its corresponding tribenzoate derivative. The disaccharides were designed to mimic the heparan sulfate (HS) disaccharide GlcN(2S,6S)-IdoA(2S). However, the intermediates readily derived from these disaccharides were not stable to the sulfonation/deacylation conditions required for their conversion into the target HS mimetics.


Introduction
The fibroblast growth factors FGF-1 and FGF-2 are heparan sulfate (HS)-binding proteins that play key roles in tumor angiogenesis, a critically important process in tumor growth and development [1,2].

OPEN ACCESS
They promote angiogenesis by binding with HS and their receptors (FGFRs) to form a ternary HS:FGF:FGFR complex which leads to receptor dimerization/activation and subsequent initiation of cell signaling [3]. Inhibiting angiogenesis by blocking ternary complex formation with HS mimetics is thus a promising strategy for the development of anticancer drugs [4][5][6] without the side effects sometimes associated with other antiangiogenic therapies [7].
A number of studies have described the synthesis of specific HS or HS-like oligosaccharides to interact with FGF-1 or FGF-2 [8][9][10][11] in order to obtain information about structural requirements for HS-FGF binding and activation. Despite much recent progress [12][13][14], the synthesis of native HS oligosaccharides remains a difficult and labour-intensive exercise and has thus lead to interest in less synthetically challenging oligosaccharide mimetics as FGF antagonists [10,[15][16][17]. As part of a program aimed at the development of angiogenesis inhibitors, we recently described [18,19] the synthesis of simple disaccharides such as 2-6 which mimic the HS disaccharide GlcN(2S,6S)-IdoA(2S) (1, Figure 1), postulated from X-ray crystallographic analyses as a minimal HS consensus sequence for FGF binding [20]. The compounds were designed to maintain the -(14) linkage between the two monosaccharide units and the spatial orientation of the two key sulfo groups [GlcN(2S) and IdoA(2S)]. The conformationally flexible disaccharides 2-4 were designed to mimic this known property [21,22] of IdoA residues. Disaccharides 5 and 6, on the other hand, were designed to investigate the other extreme: a locked 1 C 4 conformation. Molecular docking calculations indicated that the predicted locations of disaccharide sulfo groups in the binding site of FGF-1 and FGF-2 were consistent with the positions observed for co-crystallized heparin-derived oligosaccharides. These studies suggest that it may be possible to mimic HS oligosaccharides with simpler structures.  [20], conformationally flexible mimetics 2-4 [18], conformationally locked mimetics 5 and 6 [19], and proposed disaccharides of intermediate flexibility 7-10. 2 In order to extend the above investigations we herein describe the design and synthesis of simple disaccharides with intermediate conformational flexibility compared with 2-6, which once sulfated, could mimic disaccharide 1.
In order to extend the above studies, it was decided to investigate IdoA mimics with intermediate degrees of conformational flexibility. The 6-deoxy-L-taloside 14 was thus selected as a potential glycosyl acceptor because, like the majority of the L-sugars, it was expected to adopt the 1 C 4 conformation in solution but not be strictly held in this conformation like 5 and 6. It was anticipated that the use of 14 would, after deprotection and sulfonation, lead to target disaccharides such as 7-10. In addition to the desired 2-O-sulfate, an additional sulfate at O-3 could provide additional electrostatic interactions with the target proteins. Acceptor 14 was thus prepared in a straightforward manner from methyl -L-rhamnopyranoside 15 [27], as outlined in Scheme 1. Triol 15 was treated with 2,2-dimethoxypropane and toluenesulfonic acid as catalyst to give the isopropylidene 16 which was subsequently oxidised with Dess-Martin periodinane to the ketone 17 in good yield (70%, 2 steps). Stereoselective reduction with sodium borohydride in methanol gave exclusively the 6-deoxy--Ltaloside 18 which was converted into the diol 20 in moderate yield (54%, 3 steps) via allylation at C4 followed by toluenesulfonic acid catalysed methanolysis of the isopropylidene group. Diol 20 was then benzylated (NaH/benzyl bromide, 87%) and de-O-allylated with PdCl 2 in methanol at reflux to afford the alcohol 14 in excellent yield (95%), ready for use in the glycosylation studies.
Glycosylation of acceptor 14 with methyl thiogalactoside donor 11 in dichloromethane at −20 °C with NIS/TfOH as the promoter was very rapid and gave the disaccharide 22 in high yield (90%) following purification by flash chromatography (Scheme 2). Analysis of the 1 H-NMR spectrum of 22 confirmed the presence of the newly formed -glycosidic linkage (doublet at 5.77 ppm, J 1,2 = 3.6 Hz), and that the L-taloside ring remained in the desired 1  Attention was then turned towards conversion of 22 into the sulfated target disaccharides 7 and 8. Unfortunately, the compounds in this series proved to be unusually unstable to the standard transformations [18,19] used to successfully prepare disaccharides 2-6 (Scheme 2). Hydrogenolytic debenzylation of disaccharide 22 was hampered by apparent poisoning of the palladium catalyst by trace sulfur-containing impurities from the glycosylation step. However, by replacing the catalyst four times during reaction, and in the presence of glacial acetic acid, triol 23 was obtained in low yield (33%), but good purity after flash chromatography. Subsequent attempted sulfonation with concomitant deacetylation of 23 (SO 3 .Me 3 N followed by aqueous 3 M NaOH) gave rise to complex mixtures from which no pure product could be isolated by the chromatographic procedures previously used [18,19] (size exclusion chromatography on Bio-Gel P-2). It is known that sulfonation of carbohydrate polyols with sulfur trioxide-amine complexes can induce cleavage of acid labile groups and glycosidic linkages [28]. Evidently disaccharide 23 is not stable to these harsh conditions. In attempts to prepare the trimethyl derivative, both the Zemplén deacetylation/methylation and hydrogenolysis steps were low yielding (29-46% and 55%, respectively). The former caused significant degradation and in one case resulted in the isolation of the monosaccharide 27 as the dominant product (71%). Cleavage of glycosidic bonds under basic conditions in the presence of atmospheric oxygen is known [29,30], and could account for the degradation seen here, but it is unclear why these disaccharides are so sensitive compared with the earlier series. Attempted sulfonation of the small amounts of available 26 produced also gave rise to complex mixtures from which the desired products could not be isolated.
Attention was then turned to the alternative disaccharide series using imidate 13 as the glycosyl donor (Scheme 3). Following literature precedent [31], TBDMSOTf was selected as the promoter for the glycosylation of 14 with donor 13. The reaction proceeded rapidly in 1,2-dichloroethane at −20 °C (10 min, thenrt over 20 min), however, it was not completely stereoselective and resulted in an inseparable mixture of  and  anomers 28 (: = 6:1). The crude mixture was therefore deacetylated under Zemplén conditions (NaOMe in MeOH) and the crude triol 29 then benzoylated with benzoyl chloride in pyridine. The resultant mixture of benzoates was separable by careful column chromatography from which the desired -linked disaccharide 30 was isolated in 60% overall yield along with the -anomer 31 (10%).  We were not able to transform disaccharide 30 into the desired sulfated products 9 or 10 (Scheme 4). Disaccharide 30 was subjected to catalytic transfer hydrogenation (Pearlman's catalyst/ammonium formate) to presumably give the crude amine. However, attempted sulfonation only gave a complex mixture of products and 1 H-NMR analysis indicated some loss of benzoates. The mixture was subjected to standard benzoylation conditions (excess benzoyl chloride/pyridine) but this did not result in simplification of the mixture and no pure products could be recovered. Compound 30 was subjected to the Zemplén and methylation procedures to give the trimethyl derivative 32 in moderate yield (69% over 2 steps). However, when this compound was subjected to the same azide reduction/sulfonation procedure as above, once again a complex mixture resulted from which no identifiable products were isolated.

Experimental
General 1 H-NMR spectra were recorded at 400 MHz for 1 H, 100 MHz for 13 C in deuteriochloroform (CDCl 3 ) with residual CHCl 3 ( 1 H,  7.26) employed as internal standard, at ambient temperatures (298 K) unless specified otherwise. Where appropriate, analysis of 1 H-NMR spectra was aided by gCOSY experiments. Flash chromatography was performed on Merck silica gel (40-63 m) under a positive pressure with the specified eluants. All solvents used were of analytical grade. The progress of the reactions was monitored by TLC using commercially prepared Merck silica gel 60 F 254 aluminium-backed plates. Compounds were visualized by charring with 5% sulfuric acid in MeOH and/or by visualization under ultraviolet light. The term 'workup' refers to dilution with water, extraction into an organic solvent, sequential washing of the organic extract with aq. 1 M HCl (where appropriate), saturated aq. NaHCO 3 and brine, followed by drying over anhydrous MgSO 4 , filtration and evaporation of the solvent by means of a rotary evaporator at reduced pressure and where appropriate, extensive drying of the residue at <1 mmHg.
Attempted sulfonation. The polyol was dissolved in anhydrous DMF (0.04 M) and sulfur trioxide pyridine complex (2 eq. per hydroxyl) or sulfur trioxide trimethylamine complex (3 eq. per hydroxyl) was added. The mixture was stirred at 60 C under a nitrogen (6-16 h), cooled (0 C), treated with MeOH (2 mL) and then made basic to pH  9 by addition of 3 M NaOH solution. The mixture was filtered and evaporated to dryness and the residue was purified by size exclusion chromatography (Bio-Gel P-2, 5 × 100 cm, 2.8 mL/min, 0.1 M NH 4 HCO 3 , 2.8 min per vial). The fractions were analyzed for carbohydrate content by TLC (charring) or the 1,9-dimethylmethylene blue test [32] and for purity by CE [33].
Attempted sulfonation/deacylation. The polyol was sulfonated according to the general procedure for sulfonation, however, the residue obtained from evaporation of basified (pH = 9) crude mixture was redissolved in 3 M NaOH (0.16 M) and stirred at rt (o/n) before purification.

Methyl 2,3-di-O-benzyl-6-deoxy--L-talopyranoside 14. (A)
The diol 20 (110 mg, 0.50 mmol) in DMF (1 mL) was added dropwise to a stirred suspension of pre-washed (hexane) NaH (500 mg of 50% oil suspension, 10 mmol) in DMF (5 mL) and the combined mixture stirred (0 °C→rt, 30 min). The mixture was then cooled (0 °C) and benzyl bromide (250 L), 2.0 mmol) was introduced and stirring continued (0 °C→rt, o/n). The mixture was cooled (0 °C) and MeOH (3 mL) was added with continued stirring (5 min) prior to evaporation of the solvent. The residual oil was subjected to rapid silica filtration (10-40% EtOAc/hexanes) to yield, presumably, the dibenzyl ether 21 as a pale yellow oil (174 mg, 87%). This was used for the next reaction without further characterisation or purification.  In a separate experiment, following the above deacetylation and benzylation procedures, the triacetate 22 (67 mg, 90.9 mol) was converted to trimethyl ether 25 as a minor product (17 mg, 29%, R f = 0.08, hexanes/EtOAc = 65:35). The major fraction was the decomposed by-product methyl   [19] 13 (291 mg, 0.66 mmol) and alcohol 14 (157 mg, 0.44 mmol) in 1,2-DCE (5 mL) was stirred in the presence of activated mol. sieves (300 mg of 3 Å powder) under an atmosphere of argon (rt, 30 min) and then cooled (−20 °C) with continued stirring (10 min). TBDMSOTf (30 L, 0.132 mmol) was introduced dropwise and the mixture was warmed (−20 °C→0 °C, 20 min). Et 3 N (100 L) was introduced and the mixture was filtered and evaporated. The residue was subjected to workup (EtOAc) and flash chromatography (EtOAc/hexanes 1:92:3) to yield a fraction presumed to contain the disaccharide product 28 as a pale yellow oil (301 mg). 1 H-NMR analysis indicated that a monosaccharide component was also present in the mixture. This residue was co-evaporated (2 × 10 mL CH 3 CN) and used in the next reaction without further which the desired product was isolated in excellent yield. Glycosylation of 14 with the 2-azido-2deoxy-glucosyl imidate 13 was not completely stereoselective (: = 6:1) and resulted in an inseparable mixture. However, deacetylation and conversion into the corresponding tribenzoates allowed for the isolation of the desired -linked disaccharide 30 in good overall yield (60%). Unfortunately, the intermediates readily derived from 22 and 30 were not stable to the sulfonation/deacylation conditions required for their conversion into the target HS mimetics, resulting in complex mixtures from which no products could be isolated.