Synthesis, binding affinity, and inhibitory capacity of cyclodextrin-based multivalent glycan ligands for human galectin-3

Abstract

Human galectin 3 (Gal-3) has been implicated to play important roles in different biological recognition processes such as tumor growth and cancer metastasis. High-affinity Gal-3 ligands are desirable for functional studies and as inhibitors for potential therapeutic development. We report here a facile synthesis of β-cyclodextrin (CD)-based Tn and TF antigen-containing multivalent ligands via a click reaction. Binding studies indicated that the synthetic multivalent glycan ligands demonstrated a clear clustering effect in binding to human Gal-3, with up to 153-fold enhanced relative affinity in comparison with the monomeric glycan ligand. The GalNAc (Tn antigen) containing heptavalent ligand showed the highest affinity for human Gal-3 among the synthetic ligands tested, with an EC₅₀ of 1.4 μM in binding to human Gal-3. A cell-based assay revealed that the synthetic CD-based multivalent ligands could efficiently inhibit Gal-3 binding to human airway epithelial cells, with an inhibitory capacity consistent with their binding affinity measured by SPR. The synthetic cyclodextrin-based ligands described in this study should be valuable for functional studies of human Gal-3 and potentially for therapeutic applications.

1. Introduction

Lectin-carbohydrate interactions are key mediators of numerous natural biological processes such as early development, tissue repair, and immune recognition, as well as pathological conditions, including cancer growth and metastasis, and host attachment and entry of potential pathogens^1-4. Lectins are usually composed of polypeptide subunits each carrying a carbohydrate recognition domain (CRD), resulting in oligomeric multivalent carbohydrate-binding proteins ^4-5. The specificity and affinity of a lectin’s CRD is determined by a unique set of amino acids that interact with the carbohydrate ligand usually via hydrogen bonds and stacking interactions. The multivalent binding of the lectin subunits’ CRDs to multivalent ligands increases the avidity of the lectin-ligand interaction ^6-8. Based on the unique sequence motifs in the CRD, cation requirements, and structural folds, lectins are currently classified into distinct types, including C-type, I-type, F-type, galectins and others. Among these, galectins are β-galactoside-binding lectins, which have generated great interest due to their multiple functions in health and disease ^{1-3, 9-10}. Based on their domain organization, they are classified as “proto”, “chimera”, and “tandem-repeat” types, each type comprising one or multiple galectin members.

Galectin-3 (Gal-3), the only member of the “chimera”-type, is involved in diverse biological processes, some of which are beneficial to the host, such as modulation of signaling pathways, cell adhesion, and pathogen recognition, whereas others are clearly detrimental ¹¹. Among the latter, Gal-3 promotes cancer metastasis by facilitating adhesion of circulating tumor cells to the vascular endothelium ^12-16. Similarly, in influenza A patients, Gal3 can increase adhesion of Streptococcus pneumoniae to the airway epithelial cells, potentially promoting a secondary bacterial infection and pneumonia ¹⁷. Therefore, there is great interest in preventing or disrupting the above-mentioned Gal3-ligand interactions by the use of natural and synthetic glycan-displaying multivalent inhibitors.

The Thomsen–Friedenreich disaccharide (TFD), Gal-β-1,3-GalNAc, which has been found in about 90 % of all human cancers, is a natural ligand for Gal-3¹⁸. Previous reports have shown that both natural and synthetic TFD can inhibit Gal3-mediated cell adhesion of TFD-decorated tumor cells ^12,19 Similarly, Tn antigen (GalNAc linked to serine or threonine)-based ligands can also function as potent inhibitors for Gal-3 ^20-22. Since in their natural state Gal-3 subunits oligomerize as trimers and pentamers, considerable effort has been devoted to investigating multivalent interactions between oligomeric Gal-3 and multivalent glycan clusters, that would yield high affinity/avidity inhibitors of the aforementioned detrimental Gal-3 binding ^20,23-34. Our first attempt to develop high affinity/avidity synthetic multivalent inhibitors was based on using human serum albumin as a scaffold ³⁵. To enable higher multivalency that would increase avidity of the Gal3-ligand interaction, we sought to construct Tn or TFD-containing glycoclusters, using β-cyclodextrin (CD) as the cyclic oligosaccharide scaffold that comprises seven glucose units linked by α-1,4-glycosidic bonds. β-Cyclodextrin has also been widely used as the scaffold for multivalent display and as a vehicle for selective drug delivery, due to its biocompatibility and non-immunogenicity ^14,36-40. β-CD has previously been used as a scaffold for constructing multivalent carbohydrate ligands for lectins ^29,4150. In 2004, Gabius and coworkers reported the first cyclodextrin-based glycan ligands for Gal-3, where they have shown that the CD-based heptavalent N-acetyllactosamine (LacNAc) ligand has an EC₅₀ of 106 μM, which is about 28-fold more potent than free LacNAc ²⁹. Other synthetic Gal-3 multivalent inhibitors on different scaffolds have also been reported, with EC₅₀ values ranging from μM to mM. ^{20-21,25,27-30,32,35,42,43} We report in this paper a facile synthesis of β-CD-based multivalent ligands carrying GalNAc (Tn antigen) and Gal-β-1,3-GalNAc (T antigen) moieties via a copper catalyzed azide-alkyne cycloaddition reaction. Surface plasmon resonance binding analysis indicated that the CD-based GalNAc and Gal-β-1,3-GalNAc ligands showed up to 153-fold enhanced affinity to human Gal-3 in comparison with free disaccharide ligand. In addition, a cell-based assay revealed their significantly increased inhibitory capacity for Gal-3 binding to human airway epithelial cells, as compared with a CD-based lactose inhibitor.

2. Results and discussion

2.1. Synthesis of the azide-functionalized mono- and di-saccharide ligands

We sought to apply the copper-catalyzed azide-alkyne cycloaddition reaction to conjugate mono- and disaccharide ligands to a β-cyclodextrin scaffold to make the heptavalent glycoclusters as ligands for galectin-3. For that purpose, we decided to introduce an azide group in the monomeric ligands of galectin-3 with polyethylene glycol (PEG) linker attached to the reducing end, including the GalNAc, Gal-β-1,3-GalNAc, and lactose derivatives (1–3), respectively (Fig. 1).

Figure 1.

Structures of azide-functionalized GalNAc (Tn antigen), Gal-β-1,3-GalNAc (T antigen) and lactose to be conjugated to β-cyclodextrin scaffold.

The synthesis of 1 and 2 commenced with the use of the 1-bromo-2-azido-galactose derivative (5) as the starting material, which was prepared from galactose following the previously reported procedure ⁴⁴. Glycosylation of 5 with the alcohol under the promotion of AgClO₄/AgCO₃ gave the α-glycoside (6) in a moderate yield (Scheme 1). The α-galactosidic linkage was confirmed by ¹H NMR, which showed a relatively small coupling constant (J_1,2 = 3.2 Hz) between the H-1 and H-2, indicating an α-glycosidic bond. The azide group at C-2 was then reduced to give the GalNAc derivative (7). Introduction of a terminal azide was achieved by reaction of 7 with NaN₃ to give 8. De-O-acetylation of 8 with catalytic amount of NaOMe in MeOH afforded 1 in excellent yield. To synthesize the azide-tagged T antigen disaccharide (2), 1 was first reacted with benzaldehyde dimethyl acetal to give the 4,6-O-benzylidene protected GalNAc derivative (9). Then 9 was glycosylated with acetobromogalactose under the promotion of AgOTf to give the disaccharide derivative (10). The β-glycosidic configuration was confirmed by ¹H NMR which showed a relatively large coupling constant (J_1,2 = 7.6 Hz) between the H-1′ and H-2′, suggesting the presence of a β-glycosidic bond. Finally, global deprotection afforded the azide-tagged T antigen disaccharide (2) (Scheme 1). The synthesis of the azide-functionalized lactose derivative (3) was performed following the previously procedure ⁴⁵. On the other hand, the azide-lactose derivative (3) was synthesized following the previously reported procedure ²⁸.

Scheme 1.

Chemical synthesis of the disaccharide ligands for making GalNAc azide and T antigen azide.

2.2. Synthesis of the cyclodextrin-based multivalent glycan ligands.

To synthesize the cyclodextrin-based multivalent glycan ligands, the 7 primary hydroxyl groups in β-cyclodextrin was selectively converted to amino groups to give 13, following the previously reported procedure ⁴⁶ (Scheme 2). Thus, iodination of β-cyclodextrin followed by nucleophilic substitution with NaN₃ gave the azide derivative (12) in 70 % yield in two steps. Then reduction of the azide groups with triphenyl phosphine gave the amine derivative (13). Finally, coupling between the amine groups in 13 and the N-hydroxylsuccinimide ester of 4-pentynoic acid (14) afforded the alkyne-functionalized cyclodextrin (15) after size exclusion chromatography.

Scheme 2.

Synthesis of cyclodextrin-based multivalent inhibitors for galectins.

With the alkyne-functionalized cyclodextrin (15) in hands, we sought to synthesize the multivalent glycan ligands using the Cu(I)-catalyzed alkyne-azide cycloaddition reaction between the alkyne (15) and the azide derivatives (1–3). Thus, the Cu(I)-catalyzed click reaction between 15 and azide-tagged GalNAc derivative (1) (2 equiv. of azide per alkyne group) was performed in an aqueous solution at room temperature and the reaction was monitored by mass spec. After 24 h, an essential quantitative conversion was achieved. The desired conjugate (16) was purified by size-exclusion chromatography on a Sephadex G-25 column in a 79 % isolated yield. In a similar fashion, the conjugations of the azide-tagged T antigen (2) and the lactose derivative (3) with 15 provided the desired conjugates (17 and 18) in 70 % and 74 % isolated yields, respectively (Scheme 2). The structures of all final glycoconjugates were confirmed by ¹H and ¹³C NMR and ESI-HRMS analysis.

2.3. SPR analysis of the binding of the synthetic CD-based glycan ligands with human Gal-3.

The affinity of the synthetic multivalent ligands for human Gal-3 was measured by surface plasmon resonance (SPR) technology. Gal-3 was immobilized on a CM5 chip through an amine coupling method, in which the free amino groups of Gal-3 were reacted with the NHS ester activated carboxymethyl groups of the CM5 surface. The data were collected with Biacore^™ T200 system and plotted with GraphPad Prism software to obtain EC₅₀ values (Fig. 2). It was found that all the monovalent mono- and disaccharide derivatives (1–3) showed very weak binding affinities for Gal-3, and there was no apparent binding at a concentration of 500 μM of the monomeric ligands (1–3). To obtain an estimate EC₅₀ value, we selected lactose as a reference and measured its binding with Gal-3 starting with a high concentration (5 mM) of the ligand, which gave an EC₅₀ of 1.47 mM (Fig. S1). The data was consistent with previous published results ^24,47-48. in contrast, all the multivalent glycan ligands (16–18) showed significantly enhanced affinities for Gal-3 with EC₅₀ values in the micromolar range. An analysis of the binding data gave an EC₅₀ of 1.37 μM, 3.16 μM, and 25.3 μM for the GalNAc (Tn antigen), Gal-β-1,3-GalNAc (T antigen), and lactose-containing heptavalent glycan ligands (16–18), respectively (Table 1). In comparison with the affinity of free lactose, the relative affinity (per glycan ligand) of the heptavalent Tn-, TFD- and lactose-containing multivalent ligands (16–18) showed 153-fold, 66-fold, and 8-fold enhanced affinity, respectively. These results indicate a clear multivalent clustering effect of the CD-based ligands for interactions with Gal-3. It is worthy to note that the GalNAc (Tn-antigen)-containing multivalent ligand (16) showed higher affinity for Gal-3 than the disaccharide (T antigen and lactose)-containing multivalent ligands (17 and 18). The results suggest that the GalNAc moiety is an excellent subunit for designing multivalent ligands for Gal-3, as demonstrated for some previously reported Gal-3 ligands^20-22. The structurally well-defined, Gal-NAc-containing heptavalent ligand (16), with an EC₅₀ of 1.4 μM, is among the best synthetic oligomeric ligands for human Gal-3 with natural glycans as ligands ^20-22.

Figure 2.

SPR analysis of the binding of the synthetic inhibitors with immobilized galectin-3. (A) monovalent inhibitor Lac-linker (3) starting concentration at 1 mM with serial dilutions; (B) multivalent inhibitor CD-Lac (18) starting concentration at 100 μM with 1:2 serial dilutions; (C) monovalent inhibitor TFD-linker (2) starting concentration at 1 mM with serial dilutions; (D) multivalent inhibitor CD-TFD (17) starting concentration at 25 μM with 1:2 serial dilutions; (E) monovalent inhibitor GalNAc-linker (1) starting concentration at 1 mM with serial dilutions; (F) multivalent inhibitor CD-GalNAc (16) starting concentration at 10 μM with 1:2 serial dilutions.

Table 1.

Affinity of the glycan ligands for human galectin 3^a..

Ligand	EC50 (μM)	Relative affinityb	Relative affinity per unitc
Lac	1470	1	–
mono Tn (1)	> 1,000	< 1.47	–
mono TFD (2)	> 1000	< 1.47	–
mono Lac (3)	> 1,000	< 1.47	–
CD-Tn (16)	1.37 ± 0.39	1070	153
CD-TFD (17)	3.16 ± 0.25	464	66.4
CD-Lac (18)	25.3 ± 3.8	58.1	8.29

^aSPR binding experiments were carried out following our previous studies, and a general method was descripted in the Experimental section.

^bRelative potency (rel pot.) was calculated relative to the free lactose.

^cRelative affinity per unit (affinity per unit) was calculated for each sugar unit on cyclodextrin.

2.4. Inhibition of Gal-3 binding to human airway epithelial cells by the synthetic CD-based glycan ligands

A cell-based galectin-3 binding assay was also conducted to measure the inhibition of Gal-3 binding to human airway epithelial cells by the cyclodextrin based glyco-clusters. Thus, A549 cells were mixed with human Gal-3 in the presence of the CD-based glyco-clusters (16–18) at varied concentrations, and the amount of Gal-3 that bind to the cell surface was measured (Fig. 3). In this cell assay, the EC₅₀ values for CD-Tn (16), CD-TFD (17), and CD-Lac (18) were measured as 1.5 μM, 10.5 μM, and 23.5 μM, respectively. The CD-Tn was found to be the most efficient for inhibiting Gal-3 binding to the human airway epithelial cells. It should be noted that in a previous report, the similar synthetic CD-based heptavalent disaccharide ligands carrying LacNAc and Lac moieties gave the minimal inhibitory concentration (MIC) of 106 μM and 452 μM, respectively, in a cell-based Gal-3 binding assay ²⁹. In our present study, the synthetic CD-Lac showed much better inhibitory activity, which could be attributed partially to the fact that a longer linker (the PEG linker) was used to present the same lactose ligands, which may make the lactose units more accessible to the binding sites on Gal-3.

Figure 3.

The inhibition of Gal-3 binding to A549 cells by the synthetic multivalent glycan ligands.

3. Conclusion

A facile synthesis of β-cyclodextrin-based Tn and T antigen containing multivalent ligands is described. The synthetic multivalent glycan ligands show a clear clustering effect in binding to human Gal-3, with up to 150-fold enhanced relative affinity in comparison with the monomeric glycan ligand. The GalNAc (Tn antigen) containing heptavalent ligand is shown to have the highest affinity for Gal-3 among the synthetic ligands tested, with an EC₅₀ of 1.4 μM in binding to human Gal-3. This study suggests that the GalNAc moiety is an excellent subunit for designing multivalent ligands for human galectin-3 with high potency. The synthetic cyclodextrin-based ligands described in this study should be valuable for functional studies of human Gal-3.

4. Experimental

4.1. Materials and methods

All chemicals, reagents, and solvents were purchased from Sigma – Aldrich and TCI, and applied in the reaction without further purification, unless specially noted. All moisture sensitive reactions were carried out under argon atmosphere, using standard Schlenk techniques. All dry solvents were prepared according to standard procedures. 4 Å molecular sieves was flame dried before use. Thin-layer chromatography was performed on silica gel 60-F₂₅₄ on glass plates (Merck, Germany) and revealed with p-anisaldehyde stain. Silica gel (200–425 mesh) used in flash chromatography for large-scale reactions was purchased from Sigma-Aldrich. Columns for flash chromatography for small-scale reactions were performed on Isolera One system with ZIP KP-Sil columns (Biotage, Sweden) with elution condition specified for each target compound. Solvent gradients were given refer to stepped gradients and concentrations are reported as % v/v. LC-ESI-MS analysis was performed on HPLC-SQ2 detector (Waters, US) with a XBridge BEH Shield RP18 column (2.1 × 50 mm, 3.5 μm) (Waters, US) using water containing 0.1 % formic acid as phase A, MeCN containing 0.1 % formic acid as phase B. LC-ESI-HRMS was performed on an Exactive Plus Orbitrap Mass Spectrometer (Thermo Scientific, US) equipped with a XBridge BEH Shield RP18 column (2.1 × 50 mm, 3.5 μm) (Waters, US) at a flow rate of 0.4 ml/min using water containing 0.1 % formic acid as phase A, MeCN containing 0.1 % formic acid as phase B, at a gradient of 5–95 % B in 6 min. ¹H, ¹³C, and ¹H─¹H COSY NMR spectra were recorded on either 400 or 600 MHz spectrometer (Bruker, Japan) with CDCl₃, D₂O or DMSO-d₆ as the solvent. All ¹³C NMR spectra were performed with proton decoupling, and all chemical shifts are reported in part per million (ppm) and referenced to residual solvent. All spectra are in the supporting information. Binding affinities between Gal-3 and analytes were performed on a Biacore T200 (GE Healthcare, US) instrument. Recombinant Gal-3 was expressed following the reported procedure ⁴⁹.

4.2. Synthesis of the monovalent Gal-3 ligands

2-(2-(2-Chloroethoxy)ethoxy)ethyl-2-azido-3,4,6-tri-O-acetyl-2-desoxy-α-d-galactopyranoside (6).

A mixture of 2-(2-(2-chloroethoxy) ethoxy)ethanol (5.0 g, 30 mmol) and molecular sieves (20 g) in dichloromethane (60 ml) and toluene (50 ml) was stirred at room temperature for 30 min. Then the solution of AgClO₄ (0.8 g, 4 mmol) and Ag₂CO₃ (8.3 g, 30 mmol) in dichloromethane (50 ml) and toluene (50 ml) was added and stirred at 0 °C for 30 min. The solution of glycosyl donor 5 (8.1 g, 20.6 mmol) in dichloromethane (50 ml) and toluene (50 ml) was added dropwise and stirred at 0 °C for 30 min, then the solution was warmed up to room temperature and stirred at for 16 h under argon. The reaction mixture was diluted with dichloromethane (150 ml) then filtered through a Celite pad. The filtrate was washed with sat. NaHCO₃ and brine, then dried over Na₂SO₄, and filtered. The filtrate was concentrated in vacuo and purified by flash silica gel chromatography (hexane/EtOAc = 100/0 to 60/40) to give the pure alpha derivative (6) (2.35 g, 4.87 mmol, 32 %) as a colorless oil. ¹H NMR (CDCl₃, 400 MHz): δ 5.45 (dd, J_4,5 = 1.0 Hz, J_3,4 = 3.4 Hz, 1H, H-4), 5.37 (dd, J_3,4 = 3.3 Hz, J_2,3 = 11.1 Hz, 1H, H-3), 5.09 (d, J_1-2 = 3.5 Hz, 1H, H-1), 4.31 (dt, J_4,5 = 3.3 Hz, J_5,6 = 11.1 Hz, 1H, H-5), 4.05–4.20 (m, 2H, H-2, H-6_a), 3.83–3.93 (m, 1H, H-6_b), 3.60–3.80 (m, 12H, ethylene glycol chain), 2.17 (s, 3H, Ac), 2.08 (s, 3H, Ac), 2.07 (s, 3H, Ac). ¹³C NMR (CDCl₃, 100 MHz): δ 170.78, 170.47, 170.24, 98.63, 71.77, 71.08, 70.55, 68.56, 68.14, 68.02, 66.98, 61.98, 57.82, 43.19, 21.10, 21.07, 21.03. ESI-MS: [M + Na]⁺ calcd for C₁₈H₂₈ClN₃O₁₀Na⁺, 504.16; found (m/z), 504.33.

2-(2-(2-Chloroethoxy)ethoxy)ethyl-2-actamido-3,4,6-tri-O-acetyl-2-desoxy-α-d-galactopyranoside (7).

6 (2.35 g, 4.87 mmol) was dissolved in a mixture of THF (100 ml), acetic anhydride (66 ml) and acetic acid (33 ml). Zinc (2.0 g) treated with CuSO₄ solution (2 % 80 ml) was added to the mixture, the solution was stirred at room temperature for 16 h, then filtered through Celite pad. The filtrate was diluted with dichloromethane (300 ml) and washed with saturated NaHCO₃ till no bubble was forming. The organic layer was separated and washed with brine and dried over MgSO₄ and filtered. The filtrated was concentrated in vacuo and then subjected to flash chromatography (EtOAc/MeOH = 100/0 to 90/10) to give acetamide (7) (1.88 g, 3.78 mmol, 78 %) as a colorless oil. ¹H NMR (CDCl₃, 400 MHz): δ 5.97 (d, J_NH,H-2 = 9.6 Hz, 1H, NH-GalNAc), 5.28 (d, J_4,5 = 2.5 Hz, 1H, H-4), 5.06 (dd, J_3-4 = 3.2 Hz, J_2,3 = 11.3 Hz, 1H, H-3), 4.83 (d, J_1-2 = 3.6 Hz, 1H, H-1), 4.45–4.52 (m, 1H, H-2), 4.15 (t, J_5-6 = 6.6 Hz, 1H, H-5), 3.94–4.06 (m, 2H, H-6_a, H-6_b), 3.73–3.78 (m, 1H, α-OCH_a), 3.65–3.71 (m, 2H, −CH₂Cl); 3.54–3.62 (m, 9H, α-OCH_b, ethylene glycol chain), 2.07 (s, 3H, Ac), 1.96 (s, 3H, Ac), 1.90 (s, 3H, Ac), 1.89 (s, 3H, Ac). ¹³C NMR (CDCl₃, 100 MHz): δ 171.09, 170.69, 170.66, 98.32, 71.66, 70.90, 70.69, 70.23, 68.78, 67.85, 67.71, 67.05, 62.21, 48.06, 43.20, 23.48, 21.02. ESI-MS: [M + H]⁺calcd for ${\mathrm{C}}_{20}{\mathrm{H}}_{33}{\text{CINO}}_{11}^{+}$, 498.17; found (m/z), 498.29.

2-(2-(2-Azido)ethoxy)ethyl-2-actamido-3,4,6-tri-O-acetyl-2-desoxy-α-d-galactopyranoside (8).

A mixture of 7 (0.75 g, 1.5 mmol), NaN₃ (0.49 g, 7.5 mmol) and tetrabutylammonium iodide (1.11 g, 3 mmol) in THF (50 ml) was warmed up to 90 °C and stirred for 6 h under argon, the resulting mixture was diluted with EtOAc (100 ml) and washed twice with distilled water (50 ml). The mixture was concentrated in vacuo and subjected to flash chromatography (EtOAc/MeOH = 100/0 to 90/10) to give compound 8 (0.665 g, 1.32 mmol, 88 %) as a white solid. ¹H NMR (CDCl₃, 400 MHz): δ 5.95 (d, J_NH,H-2 = 9.7 Hz, 1H, NH-GalNAc), 5.33 (d, J_4,5 = 2.5 Hz, 1H, H-4), 5.12 (dd, J_3,4 = 3.3 Hz, J_2,3 = 11.3 Hz, 1H, H-3), 4.87 (d, J_1,2 = 3.6 Hz, 1H, H-1), 4.51–4.57 (m, 1H, H-2), 4.20 (t, J_5,6 = 6.5 Hz, 1H, H-5), 4.00–4.11 (m, 2H, H-6_a, H-6_b), 3.77–3.84 (m, 1H, α-OCH_a), 3.57–3.68 (m, 9H, α-OCH_b, ethylene glycol chain), 3.34 (t, J = 4.8 Hz, 2H, −CH₂N₃), 2.12 (s, 3H, Ac), 2.01 (s, 3H, Ac), 1.95 (s, 3H, Ac), 1.94 (s, 3H, Ac). ¹³C NMR (CDCl₃, 100 MHz): δ 171.19, 170.76, 170.73, 170.69, 98.35, 71.03, 70.87, 70.46, 70.37, 68.83, 67.87, 67.75, 67.11, 62.27, 51.02, 48.12, 23.51, 21.11, 21.06. ESI-MS: [M + H]⁺ calcd for ${\mathrm{C}}_{20}{\mathrm{H}}_{33}{\mathrm{N}}_4{\mathrm{O}}_{11}^{+}$, 505.21; found (m/z), 505.44.

2-(2-(2-Azido)ethoxy)ethyl-2-actamido-2-desoxy-α-d-gal-actopyranoside (1).

To a solution of 5 (0.668 g, 1.33 mmol) in dry MeOH (10 ml) NaOMe (0.2 mM, 400 μL, 0.5 mmol) was added at room temperature. The reaction mixture was stirred under argon at room temperature for 30 min, then neutralized with Dowex (50 W X8 hydrogen form) and filtered. The filtrate was concentrated in vacuo to dryness to yield the GalNAc derivative (1) (0.458 g, 91 %) as white solid. ¹H NMR (D₂O, 400 MHz): δ 4.90 (d, J_1,2 = 3.8 Hz, 1H, H-1), 4.15 (dd, J_1,2 = 3.7 Hz, J_2,3 = 11.0, 1H, H-2), 3.94–3.99 (m, 2H, H-4, H-5), 3.89–3.94 (dd, J_3,4 = 3.2 Hz, J_2,3 = 11.0 Hz, 1H, H-3), 3.81–3.88 (m, 1H, H-6_a), 3.69–3.77 (m, 10H, ethylene glycol chain), 3.48 (t, J = 4.7 Hz, 2H, −CH₂N₃), 2.03 (s, 3H, Ac). ¹³C NMR (D₂O, 150 MHz): δ 174.92, 97.65, 71.38, 70.11, 69.95, 69.90, 69.64, 68.90, 68.20, 66.95, 61.63, 50.50, 50.18, 22.34. ESI-MS: [M + H]⁺ calcd for ${\mathrm{C}}_{14}{\mathrm{H}}_{27}{\mathrm{N}}_4{\mathrm{O}}_8^{+}$, 379.18; found (m/z), 379.38.

2-(2-(2-Azido)ethoxy)ethyl-2-actamido-4,6-O-benzyliden-2-desoxy-α-d-galactopyranoside (9).

A mixture of 1 (255 mg, 0.647 mmol), α,α-Dimethoxytoluene (153 mg, 1.01 mmol) and camphorsulfonic acid (15.1 mg, 0.065 mmol) in dry acetonitrile (10 ml) was stirred at room temperature under argon for 1.5 h. The mixture was concentrated in vacuo and subjected to flash chromatography (EtOAc/iPrOH = 100/0 to 85/15) to give compound 9 (264 mg, 0.566 mmol, 88 %) as white solid. ¹H NMR (CDCl₃, 400 MHz): δ 7.51–7.57 (m, 2H, H_ar-Bzn), 7.35–7.43 (m, 3H, H_ar-Bzn), 6.34 (d, J_NH,H2 = 7.3 Hz, 1H, NH-GalNAc), 5.59 (s, 1H, −CH-Bzn), 5.00 (d, J_1,2 = 3.0 Hz, 1H, H-1), 4.46–4.53 (m, 1H, H-2), 3.83–4.31 (m, 5H, H-3, H-4, H-5, H-6_a, H-6_b), 3.80 (s, 1H, H-4), 3.66–3.74 (m, 10H, ethylene glycol chain), 3.43 (t, J = 5.1 Hz, 2H, −CH₂N₃), 2.09 (s, 3H, Ac). ¹³C NMR (CDCl₃, 100 MHz): δ 192.87, 174.41, 137.98, 134.91, 130.17, 129.56, 129.43, 128.65, 126.81, 101.68, 98.93, 75.95, 71.16, 70.91, 70.82, 70.55, 70.48, 69.81, 69.31, 67.90, 63.49, 51.19, 51.07, 23.59. ESI-MS: [M + H]⁺ calcd for ${\mathrm{C}}_{21}{\mathrm{H}}_{31}{\mathrm{N}}_4{\mathrm{O}}_8^{+}$, 467.21; found (m/z), 467.38.

2-(2-(2-Azido)ethoxy)ethyl-2-actamido-4,6-O-benzyliden-2-desoxy-3-O-[2,3,4,6-tetra-O-acetyl-β-d-galactopyranosyl]-α-d-galactopyranoside (10).

The mixture of 2,3,4,6-Tetra-O-acetyl-α-d-galactopyranosylbromide (300 mg, 0.763 mmol), the acceptor 9 (180 mg, 0.386 mmol), and spherical molecular sieves in dichloromethane (5 ml) were stirred under argon for 2 h, then cooled to −40 °C. The solution of AgOTf (200 mg, 0.778 mmol) and tetramethylurea (180 mg, 1.55 mmol) in dichloromethane (2 ml) was added dropwise under argon. The mixture was stirred at room temperature for 16 h, then filtered through a Celite pad. The filtrated was diluted with EtOAc and washed with NaHCO₃, and then dried over MgSO₄ and filtered. The filtrated was concentrated in vacuo and subjected to flash chromatography (Hexane/EtOAc = 50/50) to give 10 (164 mg, 0.206 mmol, 53 %) as a white solid. ¹H NMR (CDCl₃, 400 MHz): δ 7.53–7.57 (m, 2H, H_ar-Bzn), 7.33–7.41 (m, 3H, H_ar-Bzn), 5.98 (s, 1H, NH-GalNAc), 5.56 (s, 1H, −CH-Bzn), 4.95–5.40 (m, 4H, H-1, H-2′, H-4, H-4′), 4.75 (d, J_1′,2′ = 7.8 Hz, 1H, H-1′), 4.46–4.53 (m, 1H, H-2), 3.73–4.44 (m, 9H, H-3, H-5, H-6_a, H-6_b, H-3′, H-5′, H-6_a’, H-6_b’), 3.66–3.74 (m, 10H, ethylene glycol chain), 3.43 (m, 2H, −CH₂N₃), 2.16 (s, 3H, Ac), 2.05 (2 s, 6H, Ac), 2.00 (s, 3H, Ac), 1.98 (s, 3H, Ac). ¹³C NMR (CDCl₃, 100 MHz): δ 192.93, 170.76, 170.62, 170.59, 17.039, 138.19, 136.86, 134.86, 130.14, 129.40, 129.19, 128.59, 128.53, 126.63, 126.49, 101.09, 99.04, 71.30, 71.13, 71.07, 70.91, 70.58, 70.44, 69.72, 69.28, 67.38, 67.34, 61.71, 51.04, 21.11, 21.05, 20.96, 20.90. ESI-MS: [M + H]⁺ calcd for ${\mathrm{C}}_{35}{\mathrm{H}}_{49}{\mathrm{N}}_4{\mathrm{O}}_{17}^{+}$, 797.31; found (m/z), 797.35.

2-(2-(2-Azido)ethoxy)ethyl-2-actamido-2-desoxy-3-O-[-β-d-galactopyranosyl]-α-d-galactopyranoside (2).

To the solution of 10 (131 mg, 0.164 mmol) in dichloromethane (5 ml) 80 % TFA solution was added at 0 °C. The mixture was stirred at 0 °C for 2 h and dried in vacuo. The residue was dissolved in dry MeOH (5 ml), and then treated with MeONa (0.5 M, 1 ml). The mixture was stirred under at room temperature for 3 h, then neutralized with Dowex (50 W X8 hydrogen form) and filtered. The filtrate was concentrated and subjected to G-10 size exclusion column to obtain the pure T antigen derivative 2 (63.7 mg, 118 mmol, 72 %) as a white solid after lyophilization. ¹H NMR (D₂O, 400 MHz): δ 4.94 (d, J_1,2 = 3.6 Hz, 1H, H-1), 4.48 (d, J_1,2 = 7.6 Hz, 1H, H-1′), 4.36 (dd, J_1,2 = 3.6 Hz, J_2,3 = 11.2 Hz, 1H, H-2), 4.26 (d, J_4-5 = 2.4 Hz, 1H, H-4), 4.06 (dd, J_3,4 = 2.8 Hz, J_2,3 = 10.8 Hz, 1H, H-3), 3.62–4.05 (m, 18H, H-5, H-6_a, H-6_b, H-3′, H-4′, H-5′, H-6_a’, H-6_b’, ethylene glycol chain), 3.51–3.57 (m, 3H, H-2′, −CH₂N₃), 2.05 (s, 3H, Ac). ¹³C NMR (D₂O, 150 MHz): δ 174.08, 104.30, 96.93, 76.85, 74.46, 72.05, 70.21, 70.12, 69.28, 69.08, 68.77, 68.31, 68.09, 66.03, 60.71, 60.45, 49.63, 48.02, 21.54. ESI-HRMS: [M + H]⁺ calcd for ${\mathrm{C}}_{20}{\mathrm{H}}_{37}{\mathrm{N}}_4{\mathrm{O}}_{13}^{+}$, 541.24; found (m/z), 541.23.

4.3. Synthesis of the cyclodextrin-based Gal-3 ligands

β-Cyclodextrin-alkyne (15).

per-6-amino β-cyclodextrin (13) (50 mg, 44.3 μmol) and NHS activated 4-pentynoic acid (14, 121 mg, 620 μmol) was dissolved in DMF (5 ml) with 1 % TEA, the mixture was stirred under Ar overnight and dried in vacuo. The residue was extracted with dichloromethane and water. The aqueous solution was concentrated under vacuo. The filtrate was concentrated and subjected to P2 size exclusion column to obtain the compound 15 (43.8 mg, 26.0 μmol, 59 %). ¹H NMR (DMSO-d₆, 400 MHz) δ 5.85 (d, J = 6.8 Hz,7H, OH), 5.81 (s, 7H, OH), 4.83 (d, J_1,2 = 3.2 Hz, 7H, H-1), 3.77 (d, J = 12.8 Hz, 7H, H-6_a), 3.68 (m, 7H, H-5), 3.590 (m, 7H, H-3), 3.22–3.42 (m, 21H, H-2, H-4, H6_b), 2.65 (s, 7H, CC-H), 2.3–2.45 (m, 28H, CH₂-CH₂). ¹³C NMR (DMSO-d₆, 100 MHz) 171.04, 102.19, 83.81, 83.52, 72.58, 72.22, 71.00, 70.08, 33.85, 14.27. ESI-MS: [M + H]⁺ calcd for ${\mathrm{C}}_{77}{\mathrm{H}}_{106}{\mathrm{N}}_7{\mathrm{O}}_{35}^{+}$, 1688.67, fond (m/z), 1688.83.

β-Cyclodextrin-GalNAc (16).

β-Cyclodextrin-alkyne (15) (1.5 mg, 0.89 μmol), azido GalNAc derivative (1) (4.7 mg, 12 μmol), CuSO₄ (0.4 mg, 2.5 μmol), and sodium ascorbate (2.4 mg, 13.3 μmol) were dissolved in distilled water. The reaction mixture was stirred at room temperature for 24 h. The crude product was purified by G-25 size exclusion column to yield 16 (3.1 mg, 0.70 μmol, 79 %) as a white solid. ¹H NMR (D₂O, 400 MHz): δ 7.83 (br s, 7H, triazole), 4.92–5.10 (br, 7H, H-1 of Glc in CD), 4.89 (br, 7H, H-1 of GalNAc), 4.16 (dd, J_1,2 = 3.7 Hz, J_2,3 = 11.0, 7H, H-2 of GalNAc), 3.51–4.02 (m, 161H, the rest protons on the sugar rings, ethylene glycol chain), 2.84–3.08 (br, 14H, COCH2CH2-triazole), 2.47–2.86 (br, 14H, COCH₂CH₂-triazole), 2.01 (s, 21H, Ac). ¹³C NMR (D₂O, 100 MHz): δ 174.83, 97.64, 71.40, 70.06, 69.83, 68.91, 68.21, 66.93, 61.63, 50.20, 22.35. ESI-HRMS: [M + 3H]³⁺ calcd for ${\mathrm{C}}_{175}{\mathrm{H}}_{290}{\mathrm{N}}_{35}{\mathrm{O}}_{91}^{3+}$, 1446.64; found (m/z), 1446.66.

β-Cyclodextrin-TFD (17).

β-Cyclodextrin-alkyne (15) (1.5 mg, 0.89 μmol), azido TFD-Linker (2) (6.5 mg, 12 μmol), CuSO₄ (0.4 mg, 2.5 μmol), and sodium ascorbate (2.4 mg, 13.3 μmol) were dissolved in distilled water. The reaction mixture was stirred at room temperature for 24 h. The crude product was purified by G-25 size exclusion column to yield 17 (3.4 mg, 0.62 μmol, 70 %) as a white solid. ¹H NMR (D₂O, 400 MHz): δ 7.83 (br s, 7H, triazole), 4.92–5.10 (br, 7H, H-1 of Glc in CD), 4.89 (br, 7H, H-1 of GalNAc), 4.44 (d, J_1,2 = 7.6 Hz, 7H, H-1 of Gal), 4.32 (dd, J_1,2 = 3.5 Hz, J_2,3 = 11.2 Hz, 7H, H-2 of GalNAc), 4.32 (br, J_4,5 = 2.4 Hz, 7H, H-4 of GalNAc), 3.96 (br d, J_2,3 = 10.8 Hz, 7H, H-3 of GalNAc), 3.48–4.00 (m, 189H, the rest protons on the sugar rings, ethylene glycol chain), 2.84–3.08 (br, 14H, COCH₂CH₂-triazole), 2.47–2.86 (br, 14H, COCH₂CH₂-triazole), 2.01 (s, 21H, Ac). ¹³C NMR (D₂O, 150 MHz): δ 174.08, 104.30, 96.93, 76.85, 74.46, 72.03, 70.21, 70.12, 69.28, 68.77, 68.31, 68.09, 66.03, 60.71, 60.45, 49.63, 48.02, 21.54. ESI-HRMS: [M + 3H]³⁺ calcd for ${\mathrm{C}}_{217}{\mathrm{H}}_{360}{\mathrm{N}}_{35}{\mathrm{O}}_{126}^{3+}$, 1824.76; found (m/z), 1824.71.

β-Cyclodextrin-Lactose (18).

β-Cyclodextrin-alkyne (15) (1.5 mg, 0.89 μmol), azido Lactose derivative (3) (6.3 mg, 12 μmol), CuSO₄ (0.4 mg, 2.5 μmol), and sodium ascorbate (2.4 mg, 13.3 μmol) were dissolved in distilled water. The reaction mixture was stirred at room temperature for 24 h. The crude product was purified by G-25 size exclusion column to yield 18 (3.4 mg, 0.66 μmol, 74 %) as a white solid. ¹H NMR (D₂O, 400 MHz): δ 7.84 (br s, 7H, triazole), 4.85–5.01 (br, 7H, H-1 of Glc in CD), 4.48 (d, J_1,2 = 7.9 Hz, 7H, H-1 of Lactose), 4.45 (d, J_1′,2′= 7.8 Hz, 7H, H-1′ of Lactose), 3.55–4.05 (m, 210H, the rest protons on the sugar rings, ethylene glycol chain), 2.84–3.08 (br, 14H, COCH₂CH₂-triazole), 2.47–2.86 (br, 14H, COCH₂CH₂-triazole). ¹³C NMR (D₂O, 100 MHz): δ 103.37, 102.50, 78.82, 75.76, 75.16, 74.82, 73.20, 72.93, 71.35, 70.03, 69.86, 69.09, 68.94, 61.41, 60.48, 50.39. ESI-HRMS: [M + 3H]³⁺ calcd for ${\mathrm{C}}_{203}{\mathrm{H}}_{339}{\mathrm{N}}_{28}{\mathrm{O}}_{126}^{3+}$,1729.03; found (m/z), 1729.05.

4.4. Surface plasmon resonance (SPR) binding analysis

The SPR experiments and data analysis were performed following our previous method ⁴⁰. The experiments were carried out by immobilizing Gal-3 onto a CM5 chip and flowing serial dilutions of synthetic inhibitor as the analyte. Gal-3 (25 μg/mL) in acetate solution (pH 5.0, 10 mM) was immobilized onto a CM5 chip with a density of 1000 response units (RU) using an amine coupling kit following the instructions of the manufacturer. A separate channel immobilized with ethanolamine was used as a reference channel. In all measurements, different analytes were diluted in HEPES buffer (0.01 M, pH 7.4) containing NaCl (0.15 M) and 0.005 % surfactant P20. The analysis was carried out at 25 °C with a flow rate of 30 μL/min. The analyte association time was 120 s and the dissociation time was 300 s. After each measurement, the chip was regenerated with MgCl₂ solution (3 M). Data were collected with Biacore^™ T200 system and analyzed with GraphPad Prism 7 for EC₅₀ calculation.

4.5. Cell-based assays on the inhibition of Gal-3 binding to A549 cells by the synthetic multivalent glycan ligands

A549 cells were cultured in 96-well plate and fixed. The fixed cells were washed and blocked with 3 % BSA in PBS overnight at 4 °C, followed by incubation with biotinylated Gal-3 (5 μg/ml) pre-mixed with indicated concentration (3 ~ 300 μM) of multivalent compounds (16, 17, 18) for 1 h at 4 °C. The binding was revealed by incubation with streptavidin-conjugated horseradish peroxidase (HRP) followed by TMB. The reaction was stopped with 1 M HCl and the optical density (OD) at 450 nm from each well was measured by SpectraMax 340 plate reader (Molecular Device). The OD of each sample with inhibitors was compared with a standard curve with different concentration of Gal-3 vs OD to interpolated the “relative concentration”, which was used to calculated the percentage inhibition (%) = (original concentration-relative concentration)/original concentration × 100. The percentage inhibition over inhibitor concentration was shown. EC₅₀ of each inhibitor was calculated with GraphPad Prism’s nonlinear fit function [dose–response-special, asymmetric (five parameter), X is concentration] to interpolate the concentration of inhibitor to achieve 50 % inhibition.

Data availability

All experimental details and data are described in the manuscript and in the supporting information