Synthesis of chondroitin/dermatan sulfate-like oligosaccharides and evaluation of their protein affinity by fluorescence polarization

1 Synthesis of chondroitin/dermatan sulfate-like oligosaccharides and evaluation of their protein affinity by fluorescence polarization Susana Maza, M. Mar Kayser, Giuseppe Macchione, Javier López-Prados, Jesús Angulo, José L. de Paz,* Pedro M. Nieto* Glycosystems Laboratory, Instituto de Investigaciones Químicas (IIQ), Centro de Investigaciones Científicas Isla de La Cartuja, CSIC and Universidad de Sevilla, Americo Vespucio, 49, 41092 Sevilla, Spain. * Corresponding author. Fax: +34 954 460565; e-mail: [email protected] Abstract Here, we present a novel approach for the chemical synthesis of chondroitin and dermatan sulfate oligosaccharides. A key point of this strategy is the preparation and use of an N-trifluoroacetyl galactosamine building block containing a 4,6-O-di-tert- butylsilylene group. Glycosylation reactions proceeded in good yields (74-91%) with our protecting group distribution. Using this approach, we have synthesized, for the first time, a chondroitin/dermatan sulfate-like tetrasaccharide that contains both types of uronic acids, D-glucuronic and L-iduronic acid. Moreover, we have employed a fluorescence polarization competition assay to evaluate the interactions between the synthesized oligosaccharides and FGF-2 (basic Fibroblast Growth Factor). Our results show that this method, using standard instrumentation and minimal sample consumption, is a powerful tool for the rapid analysis of the glycosaminoglycan affinity for proteins in solution. Introduction 2 Chondroitin sulfate (CS) and dermatan sulfate (DS) are highly heterogeneous and sulfated, linear polysaccharides that belong to the glycosaminoglycan (GAG) family.1, 2 CS is formed by the repetition of disaccharide units of D-glucuronic acid (GlcA) and N- acetyl-D-galactosamine (GalNAc), following the sequence GlcA-β(1→3)-GalNAc- β(1→4). The disaccharide repeating unit of the structurally related DS mainly contains L-iduronic acid (IdoA) in place of GlcA. Both polysaccharides may contain sulfate groups at different positions of the chain (Figure 1). These sulfate groups are introduced during the biosynthesis of these polymers, through the action of specific sulfotransferases, giving rise to GAG chains with a high level of structural diversity. The microheterogeneity of CS and DS can be considered as a capacity to encode information and control a wide variety of biological processes by specific interactions with certain proteins.3 As in the case of others members of the GAG family, such as heparin and heparan sulfate,4-8 it is proposed that defined CS and DS oligosaccharide sequences are responsible for specific protein recognition and subsequent activity.3 However, little is known about the exact structural requirements for these interactions. In this context, synthetic CS and DS oligosaccharides9-13 are useful tools for the establishment of structure-activity relationships and the preparation of mimetics that potentially modulate the biological functions of the natural products.14-16 Interestingly, CS and DS are often found as co-polymeric structures.17 These hybrid CS/DS chains, containing both types of uronic acids, GlcA and IdoA, are involved in growth factor signalling and neuronal growth and development.17 The interaction between CS/DS and several chemokines and growth factors, including FGF-2 (basic Fibroblast Growth Factor), is mediated by oversulfated oligosaccharide sequences containing GlcA/IdoA-GalNAc (4,6-OSO3).18-21 In order to study these interactions at the molecular level, it would be very useful the access to well-defined synthetic CS/DS 3 oligosaccharides. However, to the best of our knowledge, the synthesis of a CS/DS oligomer, containing both GlcA and IdoA, has not yet been reported. Despite significant advances in the field,22-33 the synthesis of GAG oligosaccharides, including DS and CS, is still challenging, mainly due to the low reactivity of the building blocks required.34-37 There is still a great demand for efficient synthetic strategies, involving a robust and reliable set of carbohydrate building blocks. This is an important point for the successful automation of oligosaccharide synthesis.33, 38-40 Here, we present a novel approach for the synthesis of CS and DS oligomers that is based on the use of an N-trifluroacetyl-protected galactosamine building block. Glycosylation reactions proceeded in high yields using our design of protecting groups. The efficiency of this strategy is illustrated with the total synthesis of the oversulfated tetrasaccharide 1 (Scheme 1) that contains both IdoA and GlcA and bears sulfate groups at positions 4 and 6 of the GalNAc units, position 2 of the uronic acid moieties, and position 4 of the non-reducing terminus. All these positions may be sulfated in the natural products (Figure 1), except position 4 of the non-reducing end. Interestingly, the introduction of a “non-natural” sulfate group at the reducing end did not significantly affect the FGF-2 affinity of a synthetic heparin hexasaccharide41 and an “artificial” IdoA monosaccharide, sulfated at positions 2 and 4, showed considerable binding to several proteins.42, 43 Moreover, we have developed a fluorescence polarization assay to analyse the binding of tetrasaccharide 1 and its di-O-benzylated precursor to a model heparin-binding protein, FGF-2. Fluorescence polarization measurements allowed the study of GAG oligosaccharide-protein interactions in solution, with minimal time and sample consumption, and using a standard fluorescence reader. Therefore, our results indicate that this method can be considered as a powerful tool to evaluate the binding 4 affinities of synthetic oligosaccharides for receptors of biological relevance, helping to establish structure-activity relationships. Results and discussion Synthesis of chondroitin/dermatan sulfate-like oligosaccharides For the synthesis of the CS/DS-related tetrasaccharide 1, we first prepared the required monosaccharide building blocks 2-4 (Scheme 1). Glucuronic acid trichloroacetimidate 2 was prepared from known diol 5,44 as shown in Scheme 2. Selective oxidation at position 6 was performed by treatment of the diol 5 with calcium hypochlorite and catalytic TEMPO, under phase-transfer conditions.44 Strict control of the reaction time and temperature, and quenching with Na2SO3 were required to avoid the chlorination of the electron-rich 4-methoxyphenyl ring.45 The carboxylate intermediate was then esterified with BnBr and Bu4NI in DMF at 60ºC to give the benzyl uronate 6. Levulinoylation at position 4, followed by oxidative removal of the 4-methoxyphenyl group with CAN, and trichloroacetimidate formation, afforded glycosyl donor 2. Starting from diol 9,46, 47 iduronic acid trichloroacetimidate 3 was obtained by levulinoylation, selective desilylation and anomeric activation with Cl3CCN and K2CO3. Regarding the galactosamine unit, two types of building blocks, possessing different protections on the amino group, have been employed, up to date, for the synthesis of CS and DS oligosaccharides.10 2-Azido-2-deoxy-galactose derivatives11, 12, 48-51 present some limitations for their general use due to the non-participating character of the azido moiety. On the contrary, 2-deoxy-2-trichloroacetamido-galactose building blocks lead to the stereoselective formation of the required 1,2-trans glycosidic bond. Impressive synthesis of CS oligomers have been reported using N-trichloroacetyl(TCA)-protected units.3, 32, 52-55 However, this amino protecting group is associated with some problems. For example, we56 and others57 have detected the formation of stable trichlorooxazoline 5 side products during glycosidation of 2-deoxy-2-trichloroacetamido donors. Moreover, several difficulties are occasionally encountered in the final transformation to the desired 2-acetamido group. Thus, the deprotection of multiple N-TCA groups by basic hydrolysis, followed by selective N-acetylation, requires very long reaction times.57 Alternatively, radical reduction using tributylstannane afford, in some cases, significant amounts of mono- and dichloroacetamide intermediates.58, 59 For these reasons, we considered the use of an alternative amine-protecting group for the synthesis of CS/DS oligosaccharides. We chose an N-trifluoroacetyl (TFA) group because it can be easily removed under mild conditions while ensures high β selectivities in glycosylation reactions.60 Thus, we first planned the preparation of a N-TFA-protected galactosamine unit that should act as an efficient glycosyl acceptor in coupling reactions with uronic acid donors. The synthesis of such a compound was challenging (Scheme 3). Known tetraacetate 1261, 62 was synthesized from galactosamine hydrochloride in 66% yield by treatment with NaOMe and then with trifluoroacetic anhydride and Et3N in MeOH, followed by extensive acetylation (Ac2O, Py, DMAP). Compound 12 was transformed into 4-methoxyphenyl glycoside 15 by glycosylation with 4-methoxyphenol, followed by de-O-acetylation and benzylidenation. This compound was an ideal candidate for our synthetic approach because the 4,6-O-benzylidene acetal would allow the selective sulfation of these positions at the end of the synthesis to generate, among others, biologically relevant type E sulfation sequences.3 However, 15 suffered from poor solubility in organic solvents such as dichloromethane, toluene and acetonitrile, and glycosylation attempts with uronic acid donor 263 in THF or THF/CH2Cl2 mixtures did not give any desired disaccharide. We then decided to prepare compound 17 with a silyl ether group at the anomeric position. Selective cleavage of the anomeric acetyl group of 6 12 was followed by treatment with TDSCl to afford derivative 16. Acetate hydrolysis and benzylidenation gave compound 17. Surprisingly, this derivative proved to be unstable during silica gel chromatography. Therefore, 17 was not considered anymore as building block for our synthetic scheme. We then directed our attention to the 6-O- chloroacetylated building block 18, efficiently synthetized from 16 in two steps. It was anticipated that diol 18 would be selectively glycosylated at position 3. Unfortunately, glycosylation reactions between diol 18 and uronic acid donor 263 proceeded in low yield (<15%) and regioselectivity. Finally, we decided to prepare monosaccharide 4 containing a di-tert-butylsilylene group. It has been reported that this protecting group, compared to benzylidene acetals, confers desirable properties to glycosyl acceptors, such as better solubility in most organic solvents and higher stability under acidic glycosylation conditions.58, 64 Thus, we treated monosaccharide 14 with di-tert- butylsilyl bistriflate in pyridine to obtain 4 in high yield. Gratifyingly, coupling of 4 with glucuronic acid trichloroacetimidate 2 gave the desired β (1→3) disaccharide 19 in excellent yield (Scheme 4). These results highlight the profound effect that protecting group distribution of the building blocks has on glycosylation reactions.65 We also perfomed the glycosylation reaction between 4 and iduronic acid donor 3. The target α (1→3) disaccharide 20 was also obtained in high yield. The small-to-zero coupling constants for IdoA protons indicated that this residue mainly exists in 1C4 conformation, and the value of the 1JC, H (172 Hz) confirmed the α configuration of the new glycosidic linkage.66 Next, we studied the 2+2 assembly of the disaccharide units 19 and 20 to generate a tetrasaccharide sequence containing both GlcA and IdoA. The 4,6-di-tert-butylsilylene group67 of galactose and galactosamine donors leads to the selective formation of α glycosides despite the presence of participating groups at position 2.68 In some cases, 7 even 4,6-O-benzylidene derivatives of galactosamine donors may lead to loss of stereocontrol in coupling reactions with glucuronic acid-derived acceptors.69 For these reasons, we decided to transform disaccharide 20 in a suitable protected donor containing less sterically hindered acetyl groups at positions 4 and 6 to obtain the desired 1,2-trans glycoside with excellent stereoselectivity. Cleavage of the silylene group gave diol 21. The isolation of this compound was tricky because 21 formed gels in several solvents, such as toluene, CH2Cl2 and EtOAc. Therefore, diol 21 was directly acetylated, without further purification, to yield compound 22. Removal of the 4- methoxyphenyl group followed by treatment with Cl3CCN and catalytic DBU gave donor 24. On the other hand, 19 was transformed into acceptor 25 by treatment with hydrazine monohydrate in a pyridine/acetic acid solution. Coupling of disaccharides 24 and 25 gave the target β tetrasaccharide 26 in high yield. This result demonstrates the utility of our synthetic route for the assembly of CS and DS oligosaccharides, paving the way for other sequences with different sulfation patterns. Tetrasaccharide 26 was submitted to the deprotection/sulfation steps to obtain final compound 1. Cleavage of the silylene group was performed by treatment with (HF)n·Py complex. Hydrolysis of the acyl groups, benzyl and methyl esters, and trifluoroacetamides was carried out by treatment with lithium hydroperoxide and then NaOH to give compound 28. The amine groups were selectively acetylated with Ac2O in MeOH to provide intermediate 29, which was purified by gel permeation chromatography. Then, extensive O-sulfation using SO3·Me3N in DMF at 100 °C under microwave heating41, 70 gave cleanly the corresponding hepta-O-sulfated tetrasaccharide 30, which proved to be soluble in water. This compound was converted into the corresponding calcium salt for NMR characterization because the 1H-NMR spectrum of the sodium salt showed considerable signal overlap. NMR spectra showed the 8 characteristic downfield shifts of the proton and carbon signals at positions bearing a sulfate group (see Table 1 and 2). COSY, HSQC and TOCSY NMR experiments were employed for the structural assignment. Additionally, mass spectroscopic analysis confirmed the structure of 30. Finally, hydrogenolysis of 30 gave the fully deprotected tetrasaccharide 1 in good yield. The structure of 1 was confirmed by NMR and mass spectroscopic analysis. The values of 1H and 13C NMR chemical shifts for sulfated positions are in good agreement with those reported in the literature for similar GAG sulfated sequences.41,53 Fluorescence polarization measurements Fluorescent polarization is a powerful tool for the study of biomolecular interactions in solution.71 It is based on the observation that when a fluorescent molecule is excited with plane-polarized light, the remaining polarization of the emitted light depends on the rotational rate of the fluorescent molecule in solution that is inversely related to its molecular weight. Thus, the light emitted by a small fluorescent compound, which rotates quickly in solution, is highly depolarized and, therefore, the polarization value is low. If the fluorescent probe binds to a high molecular weight molecule, for example, a protein, the large complex rotates slower in solution, the emitted light remains polarized, and the polarization value is higher. Importantly, fluorescence polarization measurements do not require immobilization of protein or ligand to a surface for the analysis of the interaction. Thus, ligand’s bound/free ratio can be directly measured in solution, avoiding potential misleading effects derived from the attachment of the biomolecules to solid supports. Moreover, this technique is adequate for high throughput screening, requires very little amount of samples, and is well suited for the binding analysis of small ligands, such as oligosaccharides, to a protein receptor. Despite these advantages, fluorescence polarization has had a limited use in the study of 9 carbohydrate-protein interactions,72, 73 in part due to the lack of sensitive and appropiate instrumentation until the late 90’s. Interestingly, competition assays can be easily designed to study the binding affinities of nonfluorescent ligands.72 In fact, we employed a competition experiment to evaluate the affinity of synthetic tetrasaccharides 1 and 30 to FGF-2, as described below. First, we prepared five different fluorescein-conjugated glycosaminoglycan oligosaccharides (36-40) to select an optimal probe for binding studies with FGF-2 (Scheme 7 and Supplementary information). Commercially available oligosaccharides 31-35, derived from natural heparin by enzymatic depolimerization, were functionalized by reaction of the aldehyde group of the reducing end of the chain with a hydrazidecontaining fluorescein molecule.74 The corresponding glycosyl hydrazides were obtained in good yield, after purification by reverse phase C-18 chromatography. Then, fluorescent labelled sugars 36-40 were mixed with a fixed concentration of FGF-2, and polarization was measured in 384-well microplates using a standard fluorescence microplate reader (Figure 2). Control wells containing only the fluorescent probe, without any protein, were included in the experiment. As expected,75, 76 heparin hexasaccharide 38 and tetrasaccharide 37 bound to FGF-2 since a significant increased polarization value was observed in FGF-2 containing wells. No interaction was detected for heparin disaccharide 36 and hyaluronic acid oligosaccharides 39 and 40, ruling out any binding of FGF-2 to the fluorescein tag. Hexa 38, which gave best binding, was chosen as optimal probe for inhibition experiments (see below). Importantly, the use of 384-well plates allowed the minimization of the sample quantities required for these experiments: the standard assay was performed with 10 nM fluorescent probe and approximately 100 nM protein in a final volume of 40 µL per well. 16 3.95 (dd, 1H, J 2,3 = J 3,4 = 8.3 Hz, H-3), 2.58-2.21 (m, 4H, CH2(Lev)), 2.09 (s, 3H, CH3(Lev)); 13C-NMR (75 MHz, CDCl3) (selected data for β anomer from HMQC experiment): δ 96.3 (C-1), 77.6 (C-3), 72.7 (C-5), 70.8 (C-4); HR MS: m/z: calcd for C32H32O10Na: 599.1893; found: 599.1911 [M+Na]+. O-(Benzyl 2-O-benzoyl-3-O-benzyl-4-O-levulinoyl-α,β-D-glucopyranosyluronate) trichloroacetimidate (2): Trichloroacetonitrile (0.83 mL, 8.3 mmol) and K2CO3 (126 mg, 0.91 mmol) were added to 8 (480 mg, 0.83 mmol) in dry CH2Cl2 (4.8 mL). After stirring at room temperature for 4 h, the mixture was filtered off and concentrated in vacuo to obtain 2 (574 mg, 96%). TLC (2:1 toluene/EtOAc) Rf 0.55; 1H-NMR (300 MHz, CDCl3): (for α anomer): δ 8.59 (s, 1H, NH), 7.95 (m, 2H, Ar), 7.58 (m, 1H, Ar), 7.43 (m, 2H, Ar), 7.37 (m, 5H, Ar), 7.20 (m, 5H, Ar), 6.74 (d, 1H, J 1,2 = 3.3 Hz, H-1), 5.44 (dd, 1H, H-2), 5.38 (dd, 1H, H-4), 5.15 (2d, 2H, CH2(Bn)), 4.72 (2d, 2H, CH2(Bn)), 4.52 (d, 1H, J 4,5 = 10.1 Hz, H-5), 4.29 (dd, 1H, J 2,3 = J 3,4 = 9.7 Hz, H-3), 2.66-2.20 (m, 4H, CH2(Lev)), 2.12 (s, 3H, CH3(Lev)); 13C-NMR (75 MHz, CDCl3) (for α anomer): δ 206.4 (CO(Lev)), 171.4, 168.5, 165.7 (CO(COOBn, Bz, Lev)), 160.0 (C=NH), 137.4-127.7 (Ar), 93.1 (C-1), 90.8 (CCl3), 75.7 (C-3), 74.9 (CH2(Bn)), 71.3, 71.2, 70.9 (C-2, C-5, C-4), 67.9 (CH2(Bn)), 37.6 (CH2(Lev)), 29.7 (CH3(Lev)), 27.7 (CH2(Lev)); HR MS: m/z: calcd for C34H32Cl3NO10Na: 742.0989; found: 742.0984 [M+Na]+. Methyl (dimethylthexylsilyl 3-O-benzyl-2,4-di-O-levulinoyl-β-L-idopyranoside) uronate (10): Lev2O preparation: LevOH (4.7 mL, 45.4 mmol) was added at 0ºC to a solution of 1,3-dicyclohexylcarbodiimide (4.68 g, 22.7 mmol) in CH2Cl2 (38 mL). After stirring for 5 min at room temperature, the mixture was cooled and filtered to give a solution of Lev2O in CH2Cl2. 17 Lev2O (15.0 mL of a 0.76 M solution in CH2Cl2) was added at room temperature to a mixture of 9 (1.0 g, 2.3 mmol) and DMAP (41 mg, 0.34 mmol) in dry Py (40 mL). The mixture was stirred for 22 h, diluted with CH2Cl2, and washed with 1 M HCl, saturated aqueous NaHCO3, and H2O. The organic phase was dried (MgSO4), filtered and concentrated to dryness. The residue was purified by column chromatography (hexane/EtOAc 2:1) to afford 10 (1.55 g, 98%). TLC (hexane/EtOAc 1:1) Rf 0.26; [α] 20 D +19º (c 1.0, CHCl3); 1H-NMR (300 MHz, CDCl3): δ 7.34 (m, 5H, Ar), 5.16 (m, 1H, H- 4), 5.08 (d, 1H, J1,2 = 1.6 Hz, H-1), 4.93 (m, 1H, H-2), 4.76, 4.69 (2d, 2H, CH2(Bn)), 4.59 (d, 1H, J4,5 = 2.1 Hz, H-5), 3.86 (t, 1H, J2,3 = J3,4 = 2.7 Hz, H-3), 3.78 (s, 3H, COOMe), 2.92-2.50 (m, 8H, CH2(Lev)), 2.18 (s, 6H, CH3(Lev)), 1.61 (hp, 1H, CH(CH3)2), 0.87-0.83 (12H, CH(CH3)2, C(CH3)2), 0.23, 0.14 (2s, 6H, Si(CH3)2); 13C-NMR (75 MHz, CDCl3): δ 206.5, 206.3 (CO(Lev)), 172.3, 172.1 (CO(Lev)), 167.9 (COOMe), 137.3 (Ar-C), 128.6, 128.2, 127.8 (Ar-CH), 93.1 (C-1), 74.4 (C-3), 73.1 (CH2(Bn)), 72.7 (C-5), 68.0 (C-2), 67.2 (C-4), 52.4 (COOMe), 38.0, 37.8 (CH2(Lev)), 34.1 (CH(CH3)2), 30.0, 29.9 (CH3(Lev)), 28.1 (CH2(Lev)), 25.0 (C(CH3)2), 20.3, 20.0, 18.7, 18.5 (CH(CH3)2, C(CH3)2), -1.8, -3.5 (Si(CH3)2); HR MS: m/z: calcd for C32H48O11SiNa: 659.2864; found: 659.2847 [M+Na]+. Methyl 3-O-benzyl-2,4-di-O-levulinoyl-α,β-L-idopyranosuronate (11): An excess of (HF)n·Py (7.2 mL) was added at −10°C under an argon atmosphere to a solution of 10 (1.63 g, 2.56 mmol) in dry THF (37 mL). After 19 h at 0ºC the mixture was diluted with CH2Cl2 and washed with H2O and saturated NaHCO3 solution until neutral pH. The organic layers were dried (MgSO4), filtered and concentrated in vacuo. The residue was purified by column chromatography (toluene/EtOAc 1:3) to afford 11 as a mixture of α/β anomers (1:1) (970 mg, 77%). TLC (hexane/EtOAc 1:2) Rf 0.13; 1H-NMR (300 MHz, CDCl3): δ 7.34 (m, 10H, Ar, Ar’), 5.30 (m, 1H, H-1), 5.26 (m, 1H, H-4), 5.16 (m, 18 1H, H-4’), 5.14 (bs, 1H, H-1’), 4.98 (d, 1H, J4,5 = 2.3 Hz, H-5), 4.95 (m, 1H, H-2’), 4.85 (m, 1H, H-2), 4.78, 4.74 (2m, 4H, CH2(Bn), CH2(Bn)’), 4.69 (d, 1H, J4,5 = 2.0 Hz, H- 5’), 4.18 (bd, 1H, OH), 3.96 (bt, 2H, H-3, H-3’), 3.81, 3.80 (2s, 6H, COOMe, COOMe’), 2.91-2.45 (m, 16H, CH2(Lev), CH2(Lev)’), 2.19-2-17 (3s, 12H, CH3(Lev), CH3(Lev)’); 13C-NMR (75 MHz, CDCl3): δ 207.3, 206.5, 206.3 (CO(Lev), CO(Lev)’), 172.4, 171.9, 171.8 (CO(Lev), CO(Lev)’), 168.8, 168.1 (COOMe, COOMe’), 137.0, 136.6 (Ar-C), 129.1, 128.7, 128.6, 128.4, 128.3, 128.2, 128.0, 127.8 (Ar-C, Ar-CH), 93.0 (C-1), 92.1 (C-1’), 73.5, 73.4 (CH2(Bn), CH2(Bn)’), 73.1, 72.7, 72.3 (C-3, C-3’, C-5’), 68.0 (C-2’), 67.1 (C-4, C-4’), 66.7 (C-2), 65.6 (C-5), 52.6 (COOMe, COOMe’), 38.1, 37.8, 37.7 (CH2(Lev), CH2(Lev)’), 29.8 (CH3(Lev), CH3(Lev)’), 28.0, 27.9, 27.8 (CH2(Lev), CH2(Lev)’); HR MS: m/z: calcd for C24H30O11Na: 517.1686; found: 517.1666 [M+Na]+. O-(Methyl 3-O-benzyl-2,4-di-O-levulinoyl-α,β-L-idopyranosyluronate) trichloroacetimidate (3): K2CO3 (207 mg, 1.67 mmol) and trichloroacetonitrile (3.0 mL, 30 mmol) were added at room temperature to a solution of 11 (750 mg, 1.52 mmol) in dry CH2Cl2 (15 mL). After stirring for 11 h at room temperature, the reaction mixture was filtered through a pad of Celite and concentrated to dryness to obtain 3 as a mixture of α/β anomers (920 mg, 95%). TLC (toluene/EtOAc 1:4) Rf 0.44, 0.58 (α and β anomers); 1H-NMR (300 MHz, CDCl3) (data for major anomer): δ 8.68 (s, 1H, NH), 7.34 (m, 5H, Ar), 6.23 (d, 1H, J1,2 = 1.8 Hz, H-1), 5.27 (m, 1H, H-2), 5.22 (m, 1H, H-4), 4.80 (d, 1H, J4,5 = 2.0 Hz, H-5), 4.80-4.68 (m, 2H, CH2(Bn)), 3.99 (t, 1H, J2,3 = J3,4 = 3.0 Hz, H-3), 3.78 (s, 3H, COOMe), 2.90-2.49 (m, 8H, CH2(Lev)), 2.16 (s, 6H, CH3(Lev)); 13C-NMR (75 MHz, CDCl3) (data for major anomer): δ 206.4, 206.3 (CO(Lev)), 172.1, 172.0 (CO(Lev)), 167.2 (COOMe), 160.4 (C=NH), 136.9 (Ar-C), 128.6, 128.3, 127.9 (Ar-CH), 94.4 (C-1), 90.8 (CCl3), 73.6 (C-3), 73.4 (CH2(Bn)), 73.2 (C-5), 67.3 (C-4), 19 65.8 (C-2), 52.7 (COOMe), 37.8, 37.7 (CH2(Lev)), 29.9, 29.8 (CH3(Lev)), 28.0, 27.9 (CH2(Lev)); 1H-NMR (300 MHz, CDCl3) (data for minor anomer): δ 8.69 (s, 1H, NH), 7.34 (m, 5H, Ar), 6.40 (bs, 1H, H-1), 5.27 (m, 1H, H-4), 5.11 (m, 1H, H-2), 5.04 (d, 1H, J4,5 = 1.8 Hz, H-5), 4.80-4.68 (m, 2H, CH2(Bn)), 3.87 (m, 1H, H-3), 3.78 (s, 3H, COOMe), 2.90-2.49 (m, 8H, CH2(Lev)), 2.17 (s, 6H, CH3(Lev)); 13C-NMR (75 MHz, CDCl3) (data for minor anomer): δ 206.4, 206.3 (CO(Lev)), 171.8, 171.7 (CO(Lev)), 168.1 (COOMe), 160.1 (C=NH), 137.2 (Ar-C), 128.4, 127.9, 127.7 (Ar-CH), 95.0 (C-1), 90.5 (CCl3), 72.6 (CH2(Bn)), 71.5 (C-3), 67.8 (C-5), 67.3 (C-4), 65.2 (C-2), 52.7 (COOMe), 37.9, 37.8 (CH2(Lev)), 29.9, 29.8 (CH3(Lev)), 28.0, 27.9 (CH2(Lev)); HR MS: m/z: calcd for C26H30Cl3NO11Na: 660.0782; found: 660.0782 [M+Na]+. 1,3,4,6-Tetra-O-acetyl-2-deoxy-2-trifluoroacetamido-α,β-D-galactopyranose (12)61, 62: Galactosamine hydrochloride (21 g, 93.4 mmol) was suspended in MeOH (250 mL). NaOMe (105 mL, 1.3 M solution in MeOH) was added at room temperature. After stirring for 30 min, TFA anhydride (14.2 mL, 98.1 mmol) was added at 0ºC. After stirring for 10 min, Et3N (13.6 mL, 94.0 mmol) was added. After stirring for 28 h at room temperature, the reaction mixture was concentrated in vacuo. The residue was dissolved in Py (420 ml) and DMAP (6.4 g, 46.7 mmol) and acetic anhydride (105.8 mL, 1.1 mol) were added at 0 ºC. The reaction was stirred for 72 h at room temperature, diluted with EtOAc, washed with H2O, 1 M HCl and NaHCO3, dried (MgSO4), filtered, and concentrated. Flash chromatography (hexane/EtOAc, 3:1) gave 12 (28.4 g, 66%) as a mixture of α/β anomers (1:0.8). TLC (2:1 hexane/EtOAc) Rf 0.27; 1H-NMR (300 MHz, CDCl3) (data for α anomer): δ 6.55 (bs, 1H, NH), 6.28 (d, 1H, J 1,2 = 3.6 Hz, H- 1), 5.46 (dd, 1H, H-4), 5.29 (dd, 1H, J 2,3 = 11.4 Hz, J 3,4 = 3.5 Hz, H-3), 4.68 (m, 1H, H-2), 4.28-4.03 (m, 3H, H-5, H-6a, H-6b), 2.18-2.00 (m, 12H, CH3); 1H-NMR (300 MHz, CDCl3) (data for β anomer): δ 7.02 (bs, 1H, NH), 5.77 (d, 1H, J 1,2 = 8.7 Hz, H- 20 1), 5.40 (dd, 1H, H-4), 5.18 (dd, 1H, J 2,3 = 11.3 Hz, J 3,4 = 3.3 Hz, H-3), 4.49 (m, 1H, H-2), 4.28-4.03 (m, 3H, H-5, H-6a, H-6b), 2.18-2.00 (m, 12H, CH3); ESI MS: m/z: calcd for C16H20F3NO10Na: 466.1; found: 466.1 [M+Na]+. 4-Methoxyphenyl 3,4,6-tri-O-acetyl-2-deoxy-2-trifluoroacetamido-β-D- galactopyranoside (13): TMSOTf (320 µL, 1.7 mmol) was added to a cooled (0ºC) solution of 12 (5.1 g, 11.5 mmol) and 4-methoxyphenol (2.6 g, 20.7 mmol) in dry CH2Cl2 (51 mL). The mixture was stirred for 1 h at 0ºC and TEA (1 mL) was then added. The mixture was diluted with EtOAc and washed with H2O, saturated aqueous NaHCO3 and H2O. The organic phase was dried (MgSO4), filtered, and concentrated. The residue was purified by silica gel chromatography (toluene/EtOAc 4:1 → 1:1) to give 13 (2.4 g, 42%) and starting material (α anomer, 2.4 g, 40%). TLC (4:1 toluene/EtOAc) Rf 0.17; [α] 20 D – 3.5 º (c 1.0, CHCl3); 1H-NMR (300 MHz, CDCl3): δ 6.98 (d, 1H, J2,NH = 9.1 Hz, NH), 6.92 (m, 2H, Ar), 6.78 (m, 2H, Ar), 5.41 (d, 1H, J3,4 = 2.9 Hz, H-4), 5.31 (dd, 1H, J2,3 = 11.3 Hz, H-3), 5.04 (d, 1H, J1,2 = 8.3 Hz, H-1), 4.40 (m, 1H, H-2), 4.26-4.12 (2dd, 2H, J6a,6b = 11.5 Hz, H-6a, H-6b), 4.03 (br dd, 1H, J5,6a = J5,6b = 6.5 Hz, H-5), 3.76 (s, 3H, Me(OMP)), 2.18-2.00 (3s, 9H, CH3(Ac)); 13C-NMR (75 MHz, CDCl3): δ 170.7, 170.6, 170.3 (COCH3), 158.0 (q, 2JC,F = 34.1 Hz, COCF3), 155.9, 151.0 (Ar-C), 118.8, 114.7 (Ar-CH), 115.6 (q, 1JC,F = 287.0 Hz, COCF3), 100.4 (C-1), 71.1 (C-5), 69.5 (C-3), 66.4 (C-4), 61.6 (C-6), 55.7 (Me(OMP)), 51.8 (C-2), 20.5 (CH3(Ac)) ; HR MS: m/z: calcd for C21H24NO10F3Na: 530.1250 ; found: 530.1268 [M+Na]+. 4-Methoxyphenyl 3,4,6-trihydroxy-2-deoxy-2-(trifluoroacetamido)-β-D- galactopyranoside (14): Compound 13 (2.8 g, 5.5 mmol) was dissolved in MeOH (39 mL) and NaOMe (365 µL, 2.17 M solution in MeOH) was added. After 50 min, Amberlite acidic resin was added until pH 7. The Amberlite resin was filtered off, and 21 the solvent was removed in vacuo to give 14 (2.1 g, quantitative). TLC (16:1 CH2Cl2 /MeOH) Rf 0.21; 1H NMR (300 MHz, MeOD): δ 6.97 (m, 2H, Ar), 6.82 (m, 2H, Ar), 4.91 (d, 1H, J1,2 = 8.5 Hz, H-1), 4.25 (dd, 1H, J2,3 = 10.7 Hz, H-2), 3.91 (d, 1H, J3,4 = 3.1 Hz, H-4), 3.87-3.71 (m, 6H, H-6a, H-6b, H-3, CH3(OMP)), 3.64 (dd, 1H, H-5); HR MS: m/z: calcd for C15H18NO7F3Na: 404.0933; found: 404.0923 [M+Na]+. 4-Methoxyphenyl 4,6-O-benzylidene-2-deoxy-2-trifluoroacetamido-β-D- galactopyranoside (15): Benzaldehyde dimethyl acetal (1.0 mL, 6.7 mmol) and ptoluenesulfonic acid (0.08 g, 0.45 mmol) were added to a solution of 14 (1.7 g, 4.5 mmol) in MeCN (31 mL). After stirring at room temperature for 4 h, EtOAc was added and the mixture was washed with saturated aqueous NaHCO3. The organic phase was dried with MgSO4, filtered and concentrated in vacuo. Flash chromatography ( toluene/acetone 5:1 → 1:1) gave 15 (1.8 g, 87%). TLC (2:1 toluene/acetone) Rf 0.43; [α] 20 D – 8.4º (c 1.0, acetone); 1H-NMR (400 MHz, (CD3)2CO): δ 8.56 (d, 1H, J2,NH = 9.4 Hz, NH), 7.59 (m, 2H, Ar), 7.40 (m, 3H, Ar), 7.02 (m, 2H, Ar), 6.86 (m, 2H, Ar), 5.72 ( s, 1H, PhCHO), 5.19 (d, 1H, J1,2 = 8.3 Hz, H-1), 4.48-4.35 (m, 3H, H-2, H-4, OH), 4.32- 4.18 (m, 2H, H-6a, H-6b), 4.16-4.06 (m, 1H, H-3), 3.85 (s, 1H, H-5), 3.76 (s, 3H, Me (OMP)), 13C-NMR (100 MHz, (CD3)2CO): δ 157.3 (q, 2JC,F = 36.0 Hz, COCF3), 155.5- 115.5 (Ar), 116.3 (q, 1JC,F = 289.0 Hz, COCF3), 100.7 (PhCHO), 100.3 (C-1), 75.4 (C- 4), 69.8 (C-3), 68.8 (C-6), 66.9 (C-5), 55.0 (Me(OMP)), 53.8 (C-2); HR MS: m/z: calcd for C22H22NO7F3Na: 492.1246; found: 492.1234 [M+Na]+. Dimethylthexylsilyl 3,4,6-tri-O-acetyl-2-deoxy-2-trifluoroacetamido-β-D- galactopyranoside (16): Benzylamine (3.0 mL, 28 mmol) was added to a solution of 12 (4.0 g, 9.0 mmol) in THF (40 mL). After stirring for 3 h at room temperature, the mixture was diluted with CH2Cl2, and washed with 1M HCl and H2O. The organic phase was dried over Mg2SO4, filtered and the solvent was removed under reduced 22 pressure. The residue [TLC (3:2 hexane/EtOAc) Rf 0.36] (3.9 g) was dissolved in CH2Cl2 (19.5 mL). Imidazole (1.78 g, 26.2 mmol) and thexyldimethylsilyl chloride (2.30 mL, 11.7 mmol) were added. After 24 h, the mixture was diluted with CH2Cl2 and washed with H2O. The organic layer was dried over MgSO4, filtered, and the solvent was removed in vacuo. Flash chromatography on silica gel (3:1 hexane/EtOAc) afforded 16 (2.43 g, 51%). TLC (3:1 hexane/EtOAc) Rf 0.33; [] 20 D -8.7º (c 1.0, CHCl3); 1H-NMR (400 MHz, CDCl3):  6.71 (d, 1H, J 2,NH = 9.3 Hz, NH), 5.38 (dd, 1H, J 3,4 = 3.8 Hz, H-4), 5.22 (dd, 1H, J 3,4 = 3.5 Hz, J 2,3 = 11.3 Hz, H-3), 4.84 (d, 1H, J 1,2 = 8.0 Hz, H-1), 4.29- 4.08 (m, 3H, H-2, H-6a, H-6b), 3.95 (m, 1H, H-5), 2.28-1.94 (3s, 9H, COCH3), 1.62 (m, 1H, CH(CH3)2), 0.97-0.76 (m, 12H, C(CH3)2 and CH(CH3)2), 0.29- 0.07 (2s, 6H, Si(CH3)2); 13C-NMR (100 MHz, CDCl3):  170.9, 170.6, 170.4 (COCH3), 157.3 (q, 2JC,F = 36.9 Hz, COCF3), 115.8 (q, 1JC,F = 290.0 Hz, COCF3), 96.0 (C-1), 71.0 (C-5), 69.8 (C-3), 66.8 (C-4), 61.9 (C-6), 53.6 (C-2), 33.9, 24.8, 20.7, 20.5, 19.8, 18.4 (TDS, OAc), -2.0, -3.0 (TDS); HR MS: m/z: calcd for C22H36NO9F3NaSi: 566.2009; found: 566.1991 [M+Na]+. Dimethylthexylsilyl 6-O-chloroacetyl-2-deoxy-2-trifluoroacetamido-β-D- galactopyranoside (18): Compound 16 (2.2 g, 4.1 mmol) was dissolved in MeOH (31 mL) and NaOMe (273 µL, 2.17 M solution in MeOH) was added. After 50 min, Amberlite acidic resin was added until pH 7. The Amberlite resin was filtered off, and the solvent was removed in vacuo to give the desired triol (1.7 g, quantitative). An aliquot of this triol (0.5 g, 1.2 mmol) was dissolved in anhydrous CH2Cl2 (45 mL) and collidine (4.5 mL), and chloroacetic anhydride (112 µL, 1.4 mmol) was then added dropwise at -60 ºC. After stirring for 1 h at -60 ºC, the mixture was diluted with EtOAc and washed with 1M HCl, NaHCO3 and H2O. The organic phase was dried (MgSO4), filtered, and concentrated. Flash silica chromatography (3:2 hexane/EtOAc) afforded 18 23 (454 mg, 77%). TLC (3:2 hexane/EtOAc) Rf 0.14; [] 20 D + 17 º (c 1.0, MeOH); 1H- NMR (300 MHz, CDCl3):  7.31 (d, 1H, J 2,NH = 7.9 Hz, NH), 4.70 (d, 1H, J 1,2 = 7.5 Hz, H-1), 4.56-4.30 (m, 2H, H-6a, H-6b), 4.10 (s, 2H, COCH2Cl), 4.02-3.80 (m, 3H, H- 3, H-4, H-2), 3.75 (m, 1H, H-5), 1.58 (m, 1H, CH(CH3)2), 0.98-0.70 (m, 12H, C(CH3)2 and CH(CH3)2), 0.23-0.03 (2s, 6H, Si(CH3)2); 13C-NMR (75 MHz, CDCl3):  167.7 (COCH2Cl), 159.0 (q, 2JC,F = 36.8 Hz, COCF3), 115.8 (q, 1JC,F = 285.0 Hz, COCF3), 95.8 (C-1), 72.5 (C-5), 70.4, 68.3 (C-3, C-4), 65.1 (C-6), 55.9 (C-2), 40.5 (CH2Cl), 33.9, 24.8, 19.8, 18.5 (Ac, TDS), -2.0, -4.0 (TDS); ESI MS: m/z: calcd for C18H31ClF3NO7SiNa: 516.2; found: 516.2 [M+Na]+. 4-Methoxyphenyl 4,6-O-di-tert-butylsilylene-2-deoxy-2-trifluoroacetamido-β-D- galactopyranoside (4): Compound 14 (0.66 g, 1.73 mmol) was dissolved in dry Py (30 mL) and cooled (0ºC). Di-tert-butylsilyl bis(trifluoromethanesulfonate) (0.63 mL, 1.9 mmol) was added and the mixture was stirred at room temperature for 12 min. The reaction was quenched with MeOH (3 mL), diluted with EtOAc (120 mL), and washed with 1 M HCl, saturated aqueous NaHCO3, and H2O. The organic phase was dried (MgSO4), filtered and concentrated to dryness. The residue was purified by column chromatography (toluene/EtOAc 5:1) to afford 4 (0.8 g, 89%). TLC (toluene/EtOAc 3:1) Rf 0.30; [α] 20 D –17º (c 1.0, CHCl3); 1H-NMR (300 MHz, CDCl3): δ 6.92 (m, 2H, Ar), 6.87 (d, 1H, J2,NH = 9.1 Hz, NH), 6.78 (m, 2H, Ar), 4.97 (d, 1H, J1,2 = 8.4 Hz, H-1), 4.39 (d, 1H, J3,4 = 2.7 Hz, H-4), 4.24 (m, 2H, H-6a, H-6b), 4.09 (bq, 1H, H-2), 3.89 (dd, 1H, J2,3 = 10.5 Hz, H-3), 3.75 (s, 3H, Me(OMP)), 3.46 (bs, 1H, H-5), 2.68 (bs, 1H, OH), 1.09, 1.06 (2s, 18H, C(CH3)3); 13C-NMR (75 MHz, CDCl3): δ 158.2 (q, 2JC,F = 37.2 Hz, COCF3), 155.9, 151.2 (Ar-C), 119.7 (Ar-CH), 115.9 (q, 1JC,F = 288.0 Hz, COCF3), 114.6 (Ar-CH), 100.2 (C-1), 72.0 (C-4), 71.5 (C-5), 70.7 (C-3), 66.9 (C-6), 55.7 24 (Me(OMP)), 55.1 (C-2), 27.5 (C(CH3)3), 23.4, 20.9 (C(CH3)3); HR MS: m/z: calcd for C23H34F3NO7SiNa: 544.1954; found: 544.1968 [M+Na]+. 4-Methoxyphenyl 3-O-(benzyl 2-O-benzoyl-3-O-benzyl-4-O-levulinoyl-β-D- glucopyranosyluronate)-4,6-O-di-tert-butylsilylene-2-deoxy-2-trifluoroacetamido- β-D-galactopyranoside (19): Acceptor 4 (0.18 g, 0.3 mmol) and glucuronic acid trichloroacetimidate 2 (0.38 g, 0.5 mmol) were combined in a flask, coevaporated with toluene and dried under vacuum. The starting materials were disolved in dry CH2Cl2 (5 mL) and further dried by stirring over freshly activated 4Å molecular sieves for 15 min. TMSOTf (10 µL, 0.05 mmol) was added at 0ºC. After 10 min, the reaction was quenched with Et3N (2.5 mL) and filtered, and the solvent was removed under reduced pressure. Purification by flash cromathography (8:1 toluene/EtOAc) yielded 19 (345 mg, 91%). TLC (5:1 toluene/EtOAc) Rf 0.57; [α] 20 D +16.5º (c 1.0, CH2Cl2); 1H-NMR (500 MHz, CDCl3) δ 8.01 (m, 2H, Ar), 7.62 (m, 1H, Ar), 7.47 (m, 2H, Ar), 7.42-7.31 (m, 5H, Ar), 7.18-7.07 (m, 5H, Ar), 6.94 (m, 2H, Ar), 6.81(m, 2H, Ar), 5.41-5.30 (m, 4H, H-2B, H-1B, H-4B, H-1A), 5.18 (2d, 2H, CH2(Bn)),4.68-4.53 (m, 3H, H-4A, CH2(Bn)), 4.43 (dd, 1H, J2,3 = 11.3 Hz, J3,4 = 2.4 Hz, H-3A), 4.21-4.03 (m, 4H, H-6aA, H-6bA, H-5B, H-2A), 3.86 (dd, 1H, J2,3 = J3,4 = 8.7 Hz, H-3B), 3.78 (s, 3H, Me(OMP)), 3.42 (s, 1H, H-5A), 2.62-2.22 (m, 4H, CH2(Lev)), 2.13 (s, 3H, CH3(Lev)), 1.09, 0.98 (2s, 18H, C(CH3)3); 13C-NMR (100 MHz, CDCl3): δ 206.0 (CO(Lev)), 171.3, 167.0, 164.9 (CO(COOBn, Bz, Lev)), 157.6 (q, 2JC,F = 38.6 Hz, COCF3), 155.8, 151.1, 137.2, 134.6 (Ar-C), 133.5 (Ar-CH), 129.8-127.8 (Ar), 120.0 (Ar-CH), 115.5 (q, 1JC,F = 288.5 Hz, COCF3), 114.5 (Ar-CH), 99.5, 99.4 (C-1B, C-1A), 79.6 (C-3B), 74.5 (C-3A), 74.2 (CH2(Bn)), 73.2 (C-4A), 72.5 (C-4B or C-2B), 72.2 (C-5B), 71.3 (C-5A), 70.7 (C-2B or C-4B), 68.1 (CH2(Bn)), 67.0 (C-6A), 55.6 (Me(OMP)), 53.7 (C-2A), 37.6 (CH2(Lev)), 29.8 (CH3(Lev)), 27.7, 27.6, 27.4 (CH2(Lev), C(CH3)3), 23.2, 20.8 25 (C(CH3)3); HR MS: m/z: calcd for C55H64NO16F3NaSi: 1102.3844; found: 1102.3820 [M+Na]+. 4-Methoxyphenyl 3-O-(methyl 3-O-benzyl-2,4-di-O-levulinoyl-α-L- idopyranosyluronate)-4,6-O-di-tert-butylsilylene-2-deoxy-2-trifluoroacetamido-β- D-galactopyranoside (20): Donor 3 (0.92 g, 1.44 mmol) and acceptor 4 (0.5 g, 0.96 mmol) were dissolved in dry CH2Cl2 (12 mL) in the presence of freshly activated 4Å molecular sieves. After stirring for 30 min at 0ºC, TMSOTf (348 µL of a 0.41 M solution in dry CH2Cl2) was added under an argon atmosphere. After stirring for 15 min at 0ºC, the reaction mixture was neutralized with Et3N and concentrated to dryness. The residue was purified by column chromatography (toluene/EtOAc 1:1) to afford 20 (756 mg, 79%). TLC (toluene/EtOAc 1:1) Rf 0.19; [α] 20 D –18º (c 1.0, CHCl3); 1H-NMR (300 MHz, CDCl3): δ 7.34-7.27 (m, 5H, Ar), 7.00 (d, 1H, J2,NH = 7.2 Hz, NH), 6.94 (m, 2H, Ar), 6.79 (m, 2H, Ar), 5.34 (d, 1H, J1,2 = 8.3 Hz, H-1A), 5.28 (bt, 1H, H-4B), 5.15 (bs, 1H, H-1B), 5.07 (d, 1H, J4,5 = 2.6 Hz, H-5B), 4.88 (bt, 1H, H-2B), 4.72 (m, 2H, CH2(Bn)), 4.54 (bd, 1H, H-4A), 4.35 (dd, 1H, J2,3 = 11.0 Hz, J3,4 = 2.3 Hz, H-3A), 4.20 (m, 2H, H-6aA, H-6bA), 3.99 (m, 1H, H-2A), 3.80-3.75 (m, 7H, H-3B, Me(OMP), COOMe), 3.48 (bs, 1H, H-5A), 2.83-2.44 (m, 8H, CH2(Lev)), 2.16 (s, 6H, CH3(Lev)), 1.06, 0.97 (2s, 18H, C(CH3)3); 13C-NMR (75 MHz, CDCl3): δ 206.8, 206.3 (CO(Lev)), 171.7, 171.6 (CO(Lev)), 168.9 (COOMe), 157.8 (q, 2JC,F = 36.8 Hz, COCF3), 155.9, 151.2, 138.0 (Ar-C), 128.4, 127.7, 127.5, 120.0 (Ar-CH), 115.6 (q, 1JC,F = 287.8 Hz, COCF3), 114.6 (Ar-CH), 100.3 (C-1B), 99.1 (C-1A), 78.1 (C-3A), 73.3 (C-3B), 72.7, 72.6 (C-4A, CH2(Bn)), 71.3 (C-5A), 68.6 (C-4B), 67.8 (C-2B), 67.4 (C-5B), 67.0 (C-6A), 55.7 (COOMe or Me(OMP)), 54.0 (C-2A), 52.6 (COOMe or Me(OMP)), 37.9, 37.8 (CH2(Lev)), 29.9 (CH3(Lev)), 28.0, 27.7, 27.4 (CH2(Lev), C(CH3)3), 23.3, 20.9 32 4,6-di-O-sulfo-β-D-galactopyranoside (30): An excess of (HF)n·Py (56 µL, 2.1 mmol) was added at 0°C under an argon atmosphere to a solution of 26 (20 mg, 0.011 mmol) in dry THF (1.0 mL). After 24 h at 0ºC, the mixture was diluted with CH2Cl2 and washed with H2O and saturated NaHCO3 solution until neutral pH. The organic layer was dried (MgSO4), filtered and concentrated in vacuo to give 27. TLC (toluene/EtOAc 1:2) Rf 0.14; [α] 20 D –4º (c 0.9, CHCl3); 1H-NMR (400 MHz, CDCl3): δ 7.85 (d, 2H, Ar), 7.56 (t, 1H, Ar), 7.48-7.23 (m, 12H, Ar), 7.13 (d, 1H, J2,NH = 8.2 Hz, NH), 7.06 (m, 5H, Ar), 7.00 (d, 1H, J2,NH = 6.4 Hz, NH), 6.86 (m, 2H, Ar), 6.74 (m, 2H, Ar), 5.36 (d, 1H, J1,2 = 8.4 Hz, H-1A), 5.25 (q, 2H, CH2(Bn)), 5.23 (m, 2H, H-2B, H-4D), 5.16 (bs, 1H, H-4C), 4.98 (bs, 1H, H-1D), 4.93 (bs, 1H, H-5D), 4.80 (m, 3H, H-1B, H-2D, CH2(Bn)), 4.63 (q, 2H, CH2(Bn)), 4.54 (m, 2H, H-1C, CH2(Bn)), 4.46 (bd, 1H, J2,3 = 10.6 Hz, H-3A), 4.20 (t, 1H, J3,4 = J4,5 = 8.2 Hz, H-4B), 4.13 (bs, 1H, H-4A), 4.08 (bd, 1H, J4,5 = 8.8 Hz, H- 5B), 4.02 (m, 1H, H-2C), 3.89 (m, 1H, H-6aC), 3.83-3.71 (m, 12H, H-2A, H-6aA, H-6bA, H-3B, H-3C, H-3D, Me(OMP), COOMe), 3.63 (m, 2H, H-6bC, H-5C), 3.54 (bt, 1H, H-5A), 2.82-2.49 (m, 8H, CH2(Lev)), 2.18 (s, 6H, CH3(Lev)), 1.90, 1.64 (2s, 6H, CH3(Ac)); 13C-NMR (100 MHz, CDCl3): δ 207.1, 206.4 (CO(Lev)), 171.7, 171.5, 170.5, 170.3, 168.9, 168.8, 165.2 (CO(Lev), CO(Ac), CO(Bz), COOBn, COOMe), 158.0 (bq, 2JC,F = 37.4 Hz, 2xCOCF3), 155.8, 151.1, 137.8, 137.6, 134.8 (Ar-C), 133.6, 129.9, 129.5, 129.3, 129.2, 128.9, 128.5, 128.2, 128.1, 127.9, 127.8, 127.7, 119.0 (Ar-C, Ar-CH), 116.0 (q, 1JC,F = 287.6 Hz, COCF3), 115.2 (q, 1JC,F = 287.6 Hz, COCF3), 114.7 (Ar-CH), 101.6 (C-1B), 100.5 (C-1D), 100.1 (C-1C), 98.7 (C-1A), 80.0 (C-3B), 78.5 (C-3A), 77.8 (C-4B), 76.5 (C-3C), 75.1 (CH2(Bn)), 74.7 (C-5C), 74.3 (C-5B), 73.3 (C- 3D), 72.9 (CH2(Bn)), 72.6 (C-2B), 71.3 (C-5A), 69.4 (C-4D), 68.7 (CH2(Bn)), 68.6 (C-4A), 68.1 (C-4C), 68.0 (C-2D), 67.3 (C-5D), 62.8 (C-6C), 61.4 (C-6A), 55.8 (COOMe or Me(OMP)), 54.7 (C-2A), 53.9 (C-2C), 52.8 (COOMe or Me(OMP)), 37.9, 33 37.8 (CH2(Lev)), 29.9, 29.8 (CH3(Lev)), 28.0, 27.7 (CH2(Lev)), 20.6, 20.2 (CH3(Ac)); HR MS: m/z: calcd for C78H84F6N2O31Na: 1681.4860; found: 1681.4868 [M+Na]+. H2O2 (30%, 0.44 mL) and an aqueous solution of LiOH (0.7 M, 0.27 mL) were added at -5ºC to a solution of 27 (11.1 µmol) in THF (1.5 mL). After stirring for 20 h at room temperature, MeOH (1.5 mL) and an aqueous solution of NaOH (4 M, 0.28 mL) were added. After stirring for 6 d at room temperature, the reaction mixture was neutralized with Amberlite IR-120 (H+) resin, filtered, and concentrated to give 28. ESI MS: m/z: calcd for C45H57N2O22: 977.3; found: 977.2 [M+H]- Triethylamine (0.4 mL of a 0.36 M solution in dry MeOH) and acetic anhydride (21 µL, 0.22 mmol) were added to a cooled (0ºC) solution of 28 (11.1 µmol) in dry MeOH (2.5 mL). After stirring for 2 h at room temperature, triethylamine (0.3 mL) was added and the mixture was concentrated to dryness. The residue was purified by Sephadex LH-20 chromatography (CH2Cl2-MeOH 1:1) to give 29 as triethylammonium salt. The sodium salt of 29 was obtained by treatment with Amberlite IR-120 H+ resin in MeOH (pH ~ 3), followed by filtration, treatment with 0.1 M NaOH (pH ~ 7) and concentration. 1H-NMR (500 MHz, MeOD, data for sodium salt): δ 7.58 (d, 2H, Ar), 7.45 (d, 2H, Ar), 7.34-7.21 (m, 6H, Ar), 6.98 (m, 2H, Ar), 6.82 (m, 2H, Ar), 5.06 (d, 1H, CH2(Bn)), 4.93 (d, 1H, J1,2 = 8.4 Hz, H-1A), 4.89 (d, 1H, J1,2 = 4.4 Hz, H-1D), 4.77 (q, 2H, CH2(Bn)), 4.73 (d, 1H, CH2(Bn)), 4.62 (d, 1H, J1,2 = 8.5 Hz, H-1C), 4.43 (m, 1H, H-1B), 4.41 (d, 1H, J4,5 = 3.7 Hz, H-5D), 4.27 (dd, 1H, J2,3 = 10.6 Hz, H-2A), 4.23 (dd, 1H, J2,3 = 10.4 Hz, H-2C), 4.13 (d, 1H, J3,4 = 2.7 Hz, H-4A), 4.05 (d, 1H, J3,4 = 2.7 Hz, H-4C), 3.99- 3.95 (m, 2H, H-4B, H-4D), 3.84-3.69 (m, 9H, H-6aA, H-6aC, H-6bA or H-6bC, H-3A (3.80), Me(OMP) (3.74), H-5B (3.72), H-3C (3.71)), 3.65-3.58 (m, 3H, H-6bA or H-6bC, H-2D, H-5A or H-5C), 3.55 (t, 1H, J2,3 = J3,4 = 5.9 Hz, H-3D), 3.52 (m, 1H, H- 5A or H-5C), 3.47-3.44 (m, 2H, H-2B, H-3B), 2.05, 1.99 (2s, 6H, NHAc); 13C-NMR 34 (125 MHz, MeOD) (selected data from HSQC experiment): δ 129.9-128.4, 118.9, 115.1 (Ar-CH), 105.6 (C-1B), 103.9 (C-1D), 102.1 (C-1A), 101.3 (C-1C), 83.7 (C-3B), 81.8 (C-3C), 81.7 (C-3A), 80.3 (C-3D), 78.9 (C-4B), 77.7 (C-5B), 77.2, 76.5 (C-5A, C-5C), 76.4 (CH2(Bn)), 74.0 (C-2B), 73.6 (CH2(Bn)), 71.7 (C-5D), 71.2 (C-4D), 70.6 (C-2D), 69.4 (C-4C), 68.8 (C-4A), 62.7, 62.3 (C-6A, C-6C), 55.7 (Me(OMP)), 52.7 (C-2C), 52.6 (C-2A), 23.1, 22.8 (NAc); ESI MS: m/z: calcd for C49H61N2O24: 1061.4; found: 1061.2 [M+H]-. Compound 29 (11.1 µmol) and sulfur trioxide–trimethylamine complex (54 mg, 0.39 mmol) were dissolved in dry DMF (1.5 mL) and heated at 100ºC for 2 h using microwave radiation (20 W average power). The reaction vessel was cooled and Et3N (150 µL) and MeOH (1 mL) were added. The solution was layered on the top of a Sephadex LH-20 chromatography column which was eluted with MeOH to obtain 30 as triethylammonium salt. The residue was converted into the sodium salt by elution from a column of Dowex 50WX4-Na+ with MeOH-H2O 9:1 (6.9 mg, 34% from 26; 4 steps, 76% average yield per step). Due to extensive overlap of 1H-NMR signals of the sodium salt at 25ºC, 30 was characterized as calcium salt at 40ºC. The calcium salt of 30 was obtained by adding a 0.9 M solution of CaCl2 in D2O. 1H-NMR (500 MHz, D2O, 40ºC, data for calcium salt): δ 7.73 (d, 2H, Ar), 7.63-7.46 (m, 8H, Ar), 7.23 (m, 2H, Ar), 7.11 (m, 2H, Ar), 5.44 (bs, 1H, H-1D), 5.37 (d, 1H, J1,2 = 8.6 Hz, H-1A), 5.30 (bs, 1H, H-5D), 5.11 (bd, 1H, H-4A), 5.02 (bs, 1H, H-4C), 4.98 (q, 2H, CH2(Bn)), 4.91-4.81 (m, 5H, H-1B (4.90), H-4D (4.89), CH2(Bn), H-1C (4.84)), 4.52 (t, 1H, J1,2 = J2,3 = 7.5 Hz, H-2B), 4.47 (dd, 1H, J5,6a = 2.6 Hz, J6a,6b = 11.3 Hz, H-6aA or H-6aC), 4.43-4.19 (m, 10H, H-2D (4.43), H-6aA or H-6aC, H-3A (4.39), H-3D (4.36), H-6bA, H-6bC, H-5A or H-5C (4.33), H-4B (4.29), H-2C (4.24), H-2A (4.22)), 4.14 (m, 1H, H-3C), 4.10 (bt, 1H, H-5A or H-5C), 3.99-3.94 (m, 5H, H-3B (3.98), H-5B (3.97), Me(OMP) (3.94)), 35 2.21 (s, 6H, NHAc); 13C-NMR (125 MHz, D2O, 40ºC) (selected data from HSQC experiment): δ 129.6-128.5, 118.6, 115.2 (Ar-CH), 102.2 (C-1B), 101.5 (C-1D), 100.6 (C-1A), 100.4 (C-1C), 80.2 (C-3B), 79.6 (C-3C), 79.1 (C-2B), 77.5 (C-4B), 77.4 (C- 5B), 76.7 (C-3A), 75.7 (C-4A), 75.3 (C-4C), 74.2 (CH2(Bn)), 72.8 (C-5A or C-5C), 72.6 (C-4D), 72.4 (C-5A or C-5C), 71.9 (C-3D), 71.4 (CH2(Bn)), 71.2 (C-2D), 68.2 (C- 5D), 67.8, 67.4 (C-6A, C-6C), 56.0 (Me(OMP)), 52.4 (C-2A), 52.2 (C-2C), 22.8 (NAc); ESI MS: m/z: calcd for C49H55N2O45S7Na5: 865.0; found: 865.0 [M+5Na+2H]2-; HR MS: m/z: calcd for C49H60N2O45S7: 810.0262; found: 810.0253 [M+7H]2-. 4-Methoxyphenyl O-(2,4-di-O-sulfo-α-L-idopyranosyluronic acid)-(1→3)-O-(2- acetamido-2-deoxy-4,6-di-O-sulfo-β-D-galactopyranosyl)-(1→4)-O-(2-O-sulfo-β-D- glucopyranosyluronic acid)-(1→3)-2-acetamido-2-deoxy-4,6-di-O-sulfo-β-D- galactopyranoside (1): A solution of 30 (4.9 mg, 2.7 µmol, sodium salt) in H2O/MeOH (4.5 mL/0.5 mL) was hydrogenated (1.5 atm) in the presence of Pd(OH)2. After 22 h, the suspension was filtered over Celite and concentrated. The residue was purified by Sephadex G-25 chromatography (H2O/MeOH 9:1) to give 1 as sodium salt after lyophilisation (3.6 mg, 82%; 28% from 26, 5 steps, 78% average yield per step). 1H-NMR (500 MHz, D2O): δ 7.10 (d, 2H, Ar), 6.98 (d, 2H, Ar), 5.29 (bs, 1H, H-1D), 5.22 (d, 1H, J1,2 = 8.4 Hz, H-1A), 4.98 (bs, 1H, H-5D), 4.95 (bs, 1H, H-4A), 4.76 (m, 1H, H-4C), 4.72 (m, 1H, H-1C), 4.70 (d, 1H, J1,2 = 7.7 Hz, H-1B), 4.66 (bs, 1H, H-4D), 4.43 (m, 1H, H-3D), 4.35 (bd, 1H, H-6aA or H-6aC), 4.31-4.20 (m, 7H, H-6aA or H- 6aC, H-3A (4.27), H-6bA, H-6bC, H-5A or H-5C (4.23), H-2B (4.22), H-2D (4.20)), 4.14-4.07 (m, 4H, H-2A, H-2C, H-5A or H-5C, H-3C), 3.93 (t, 1H, J3,4 = J4,5 = 9.1 Hz, H-4B), 3.82-3.81 (m, 4H, Me(OMP), H-3B), 3.73 (d, 1H, J4,5 = 9.4 Hz, H-5B), 2.11, 2.05 (2s, 6H, NHAc); 13C-NMR (125 MHz, D2O) (selected data from HSQC experiment): δ 118.8, 115.5 (Ar-CH), 103.1 (C-1B), 101.3 (C-1C), 101.0 (C-1A), 100.7 36 (C-1D), 81.0 (C-4B), 80.1 (C-2B), 77.0 (C-5B, C-3A), 76.6 (C-3C), 76.4 (C-4A), 76.1 (C-4C), 74.8 (C-4D), 73.9 (C-3B), 73.2, 72.7 (C-5A, C-5C), 72.6 (C-2D), 68.5, 67.9 (C- 6A, C-6C), 67.0 (C-5D), 66.5 (C-3D), 56.2 (Me(OMP)), 52.8, 52.4 (C-2A, C-2C), 23.2, 22.8 (NAc); ESI MS: m/z: calcd for C35H43N2O45S7Na5: 774.9; found: 774.8 [M+5Na+2H]2-; HR MS: m/z: calcd for C35H48N2O45S7: 719.9792; found: 719.9786 [M+7H]2-. Fluorescence polarization assays Fluorescence polarization measurements were performed in 384-well microplates (black polystyrene, non-treated, Corning) using a TRIAD multimode reader (Dynex). Fluorescent probes 36-40, recombinant human FGF-2 (Peprotech), and inhibitors were dissolved in PBS buffer (10 mM, pH 7.4). For direct binding, 20 µL of a fluorescent probe solution (20 nM) were transferred to each well. Then 20 µL of FGF-2 solution (concentration ranging from 1.45 µM to 23 nM) were added and the microplate was shaked in the dark for 5 min, before reading. The total sample volume in each well was 40 µL. Control wells contained 20 µL of the fluorescent probe solution and 20 µL of PBS buffer. Blank wells contained 20 µL of FGF-2 solution and 20 µL of PBS buffer and their measurements were substracted from all values. All samples were performed in replicates of three. For inhibition assay, 10 µL of probe and 20 µL of protein at fixed concentration (40 nM and 205 nM, respectively) were mixed with 10 µL of inhibitor solution (100 µM). The total sample volume in each well was again 40 µL. After stirring for 5 min in the dark, fluorescence polarization was recorded. Two control wells containing no inhibitor and probe only (no FGF-2) were included in the study. For the determination of IC50 value, wells containing probe and FGF-2 at fixed concentration, as described above, were incubated with 6 different concentrations of inhibitor, ranging from 0.025 µM to 100 µM. Average polarization values of six replicate wells were 37 plotted against the logarithm of inhibitor concentration, and the curve was fitted to the simplified formula corresponding to a one-site competitive interaction. All the experiments were repeated at least twice. Occasionally, we found loss of FGF-2 activity when working with low concentrated aliquots in PBS buffer. For this reason, all measurements were alternatively done in PBS + 0.5% BSA (data not shown), getting similar results than those obtained with PBS. Figure 1. Disaccharide repeating units of chondroitin (left) and dermatan (right) sulfate with potential sites of sulfation indicated. Scheme 1. Building blocks required for the synthesis of tetrasaccharide 1. 38 Scheme 2. a) TEMPO, Bu4NBr, KBr, Ca(ClO)2, NaHCO3, CH2Cl2/H2O, 0ºC; BnBr, DMF, Bu4NI, 60ºC, 56%; b) LevOH, DCC, DMAP, CH2Cl2, 81%; c) CAN, toluene/CH3CN/H2O, 60%; d) Cl3CCN, K2CO3, CH2Cl2, 96%; e) Lev2O, Py/CH2Cl2, DMAP, 98%; f) (HF)n·Py, THF, 0ºC, 77%; g) Cl3CCN, K2CO3, CH2Cl2, 95%. Scheme 3. a) 4-methoxyphenol, TMSOTf, CH2Cl2, 0ºC, 42% + 40% starting material; b) NaOMe, MeOH, quantitative; c) PhCH(OMe)2, p-TsOH, CH3CN, 87%; d) tBu2Si(OTf)2, Py, 89%; e) BnNH2, THF; TDSCl, imidazole, CH2Cl2, 51%; f) NaOMe, MeOH; PhCH(OMe)2, p-TsOH, CH3CN/DMF; g) NaOMe, MeOH; (ClAc)2O, collidine, CH2Cl2, -60ºC, 77%. 39 Scheme 4. a) TMSOTf, CH2Cl2, 0ºC, 91%; b) NH2NH2·H2O, Py/AcOH, CH2Cl2, 84%; c) TMSOTf, CH2Cl2, 0ºC, 79%. Scheme 5. a) (HF)n·Py, THF, 0ºC, 97%; b) Ac2O, Py, DMAP, 89%; c) CAN, toluene/CH3CN/H2O, 0ºC, 76%; d) Cl3CCN, DBU, CH2Cl2, 84%; e) 25, TMSOTf, CH2Cl2, 0ºC, 74%. 40 Scheme 6. a) (HF)n·Py, THF, 0ºC; b) LiOH, H2O2, THF; NaOH, MeOH; c) Ac2O, MeOH, Et3N; d) SO3·Me3N, DMF, 100ºC, MW, 34% from 26, 4 steps, 76% average yield per step; e) H2, Pd(OH)2, H2O/MeOH, 82%. Table 1. 1H-NMR chemical shifts for sulfated positions of compounds 30 and 1 and the corresponding non-sulfated positions of 29. Compound H-4A H-6A H-2B H-4C H-6C H-2D H-4D 29a 4.13 3.84-3.58 3.47-3.44 4.05 3.84-3.58 3.65-3.58 3.99-3.95 30b 5.11 4.47-4.19 4.52 5.02 4.47-4.19 4.43 4.89 1c 4.95 4.35-4.20 4.22 4.76 4.35-4.20 4.20 4.66 a sodium salt, in MeOD; bcalcium salt, in D2O; csodium salt, in D2O Table 2. 13C-NMR chemical shifts for sulfated positions of compounds 30 and 1 and the corresponding non-sulfated positions of 29. Compound C-4A C-6A C-2B C-4C C-6C C-2D C-4D 29a 68.8 62.7 or 62.3 74.0 69.4 62.7 or 62.3 70.6 71.2 30b 75.7 67.8 or 67.4 79.1 75.3 67.8 or 67.4 71.2 72.6 1c 76.4 68.5 or 67.9 80.1 76.1 68.5 or 67.9 72.6 74.8 a sodium salt, in MeOD; bcalcium salt, in D2O; csodium salt, in D2O 41 Scheme 7. a) Fluorescein hydrazide, DMSO/phosphate buffer pH 5.5 (1:1), 30ºC, 91% (36), 91% (37), 94% (38), 92% (39), 95% (40). 0 10 20 30 40 50 60 70 80 90 100 36 37 38 39 40 Polarization (mP) Figure 2. Fluorescence polarization values (right, in grey) from wells containing fluorescent GAG oligosaccharides 36-40 (10 nM) and FGF-2 (97 nM) are compared with the values obtained in the absence of the protein (left, in white). For each oligosaccharide, polarization values are the average of three replicate wells and the error bars show the standard deviations for these measurements.