Catanionic Vesicles as a Facile Scaffold to Display Natural N-Glycan Ligands for Probing Multivalent Carbohydrate–Lectin Interactions

Abstract

Multivalent interactions are a key characteristic of protein–carbohydrate recognition. Phospholipid-based liposomes have been explored as a popular platform for multivalent presentation of glycans, but this platform has been plagued by the instability of typical liposomal formulations in biological media. We report here the exploitation of catanionic vesicles as a stable lipid-based nanoparticle scaffold for displaying large natural N-glycans as multivalent ligands. Hydrophobic insertion of lipidated N-glycans into the catanionic vesicle bilayer was optimized to allow for high-density display of structurally diverse N-glycans on the outer membrane leaflet. In an enzyme-linked competitive lectin-binding assay, the N-glycan-coated vesicles demonstrated a clear clustering glycoside effect, with significantly enhanced affinity for the corresponding lectins including Sambucus nigra agglutinin (SNA), concanavalin A (ConA), and human galectin-3, in comparison with their respective natural N-glycan ligands. Our results showed that relatively low density of high-mannose and sialylated complex type N-glycans gave the maximal clustering effect for binding to ConA and SNA, respectively, while relatively high-density display of the asialylated complex type N-glycan provided maximal clustering effects for binding to human galectin 3. Moreover, we also observed a macromolecular crowding effect on the binding of ConA to high-mannose N-glycans when catanionic vesicles bearing mixed high-mannose and complex-type N-glycans were used. The N-glycan-coated catanionic vesicles are stable and easy to formulate with varied density of ligands, which could serve as a feasible vehicle for drug delivery and as potent inhibitors for intervening protein–carbohydrate interactions implicated in disease.

Graphical Abstract

INTRODUCTION

Glycans are ubiquitous molecules of the cell that play active roles in biological recognition processes, including cell signaling, cell adhesion, and host immunity.^1,2 The biological functions of glycans are usually mediated by carbohydrate-binding proteins (CBPs or lectin) that recognize specific glycan structures attached to glycoproteins, glycolipids, or other glycoconjugates. While the affinity of a CBP for a single monosaccharide is usually weak (K_D = ~10⁻³ M), biological systems frequently adopt multivalent display of carbohydrate ligands to enhance the binding affinity and specificity of CBPs.^3,4 Tremendous efforts have been devoted to the design, synthesis, and evaluation of various multivalent carbohydrate ligands for studying protein–carbohydrate interactions and for developing efficient inhibitors to block specific protein–carbohydrate interactions associated with viral and bacterial infections, inflammation, and cancer progression.^5,6 Attempts have been made to use various scaffolds such as liposomes, dendrimers, cyclodextrin, and metal nanoparticles for the multivalent display of small glycans.^7–16 However, the multivalent display of the more biologically relevant natural O- and N-linked glycans remains to be an interesting area deserving further investigation.^17–19 For example, phospholipid-based liposomes have been used to display modified N-glycan ligands for targeted delivery of anticancer drugs and for immunomodulation.^18,19 Nevertheless, major drawbacks of liposome-based vesicles include the tedious procedures of formulation, polydispersity, and the inherent instability under biological conditions. On the other hand, catanionic vesicles, prepared from cationic and anionic surfactants in water, offer an alternative multivalent display scaffold that is easy to use and retains a unilamellar equilibrium phase for extended periods in buffer.^20,21 The advantage of these catanionic vesicles over conventional liposomes is that they are derived from cheap ionic surfactants, do not require solubilization by organic solvents or tedious extrusion procedures, provide vesicles with a narrow size distribution, and exhibit long-term stability (months typically) in aqueous media.²²

Catanionic vesicles were first employed for the multivalent display of sugars by Raghavan and co-workers in 2005, where C12-functionalized chitosan was shown to form a gel in the presence of catanionic vesicles.²³ In 2008, the versatility of catanionic vesicles to display structurally diverse carbohydrates was demonstrated by DeShong and co-workers, where monoand tri-saccharides bearing a C8 lipid tail were incorporated into the vesicle bilayer and the resulting sugar-coated vesicles were evaluated for lectin-mediated hemagglutination.²⁴ Particularly, DeShong and co-workers have demonstrated that the sugar density can be controlled on the vesicle leaflet. Since then, catanionic vesicles have been employed in carbohydrate microarrays and in the display of proteins, and bacterial lipopolysaccharides for prophylactic vaccines against bacteria.^21,25–27 The exceptional stability of catanionic vesicles and their ability to display a diverse array of biomolecules on their surface make them attractive scaffolds to characterize protein–carbohydrate interactions; however, this scaffold has not been employed for the multivalent display of the more biologically relevant natural O- and N-linked glycans. We report in this paper the synthesis and evaluation of catanionic vesicle scaffolded natural N-glycans as multivalent glycan ligands for lectins. Our results indicate that catanionic vesicles are powerful scaffolds that allow high-density multivalent display of high-mannose and complex-type N-glycans on the vesicle surface, which significantly enhances the affinity for the respective lectins with a clear clustering glycoside effect. These novel glycan-coated vesicles offer a new and valuable platform for the investigation of carbohydrate–protein interactions and potentially for targeted drug delivery.

RESULTS AND DISCUSSION

Synthesis of N-Glycan Lipids.

Previous studies have shown that glycolipids containing simple sugars and a C8–C12 lipid tail can be efficiently incorporated into the lipid bilayer of catanionic vesicles.^24,25 Thus, we sought to prepare N-glycan-based glycolipids and evaluate their efficiency of insertion into the catanionic vesicle bilayer. For this purpose, we chose to use the asparagine-containing natural N-glycans as the starting materials and take advantage of the free amino group in the asparagine residue to introduce a lipid chain. We selected four typical N-glycans as model sugars, including the sialylated bi-antennary complex-type glycan (S2G2-Asn), the asialylated bi-antennary Asn-glycan (S0G2), and the two high-mannose type N-glycans (Man9-Asn and Man5-Asn), which we have previously prepared by digestion of the chicken egg yolk sialoglycopeptide (SGP) and soybean agglutinin with pronase, respectively, followed by size-exclusion chromatographic purification.^28,29 Glycolipids bearing lipids with different lengths were synthesized to investigate if a longer lipid chain is required to incorporate the large natural N-glycans in the vesicles. Thus, reaction of the asialo-complex-type N-glycan (1) with the N-hydroxysuccinimide (NHS)-activated ester of dodecanoic acid gave the corresponding N-glycolipid (5) containing a C12 lipid chain in excellent yield. Treatment of 1 with the NHS ester of palmitic acid gave the N-glycolipid (6) containing a C16 lipid chain. Similarly, reaction of the sialylated N-glycan (S2G2-GlcNAc₂-Asn, 2), high-mannose type N-glycan (Man9-GlcNAc₂-Asn, 3), and the smaller high-mannose glycan (Man5GlcNAc2-Asn, 4) with the NHS ester of palmitic acid afforded the corresponding glycolipids (7−9) carrying a C16 lipid chain, respectively (Scheme 1). The glycolipids were purified by reverse-phase high-performance liquid chromatography (HPLC), and their identity was confirmed by mass spectrometry (MS) and nuclear magnetic resonance (NMR) analysis (see the Supporting Information).

Scheme 1.

Synthesis of N-Glycan Lipids Using Free Asparagine-Linked N-Glycans as the Starting Materials

Preparation and Characterization of N-Glycan-Coated Vesicles.

With the synthetic glycolipids (5−9) in hand, glycan-coated vesicles were prepared by mixing catanionic vesicles with varying concentrations of N-glycolipids in water (Scheme 2). Three different concentrations were used for each of the glycolipids as an attempt to prepare vesicles with low (L), medium (M), and high (H) glycolipid incorporation. Importantly, preformed vesicles were utilized to restrict N-glycan presentation to the outer membrane leaflet (Scheme 2).

Scheme 2.

Assembly of N-Glycan-Coated Catanionic Vesicles^a

^aCatanionic vesicles were formulated from sodium dodecylbenzene sulfonate (SDBS) and cetyltrimethylammonium tosylate (CTAT) detergents in water and allowed to equilibrate. Preformed vesicles were mixed with N-glycolipids 5−9 at room temperature for 16 h to obtain N-glycan-coated vesicles

N-Glycan-coated vesicles (N-gCVs) were then purified by size-exclusion chromatography (SEC) using a Sepharose G50 column, with characteristic vesicle elution between 13 and 19 mL as determined by dynamic light scattering (DLS) intensity of the SEC fractions (Figure 1A). Vesicle N-glycolipid incorporation was determined for pure vesicle-containing SEC fractions by the optimized phenol-sulfuric acid assay reported by Lee and co-workers.³⁰ This assay is robust and requires minimal work-up. Incorporation values were calculated relative to the recorded weight of dried vesicle and reported in mol/wt %. In initial experiments, the uncharged S0G2-GlcNAc₂-Asn N-glycan was used as a model, functionalized with C12 (5) and C16 (6) lipid tails, and tested for incorporation into the vesicle bilayer. Figure 1B showed the glycolipid incorporation values for the C12- and C16-functionalized N-glycans as a function of N-glycolipid concentration. It was found that the C16 functionalized N-glycan achieved up to 10-fold higher incorporation than the shorter C12 glycolipid at the same glycolipid concentration. This result suggests that for large glycans, a relatively long lipid chain is essential to have an efficient coating of the preformed catanionic vesicles.

Figure 1.

(A) Vesicle elution profile from the Sepharose G-50 column, peak observed between 13 and 19 mL elution volume by DLS at 633 nm. (B) Glycolipid incorporation for C12- and C16-functionalized S0G2-GlcNAc₂-Asn at varying concentrations of N-glycolipid (0.2, 0.45, and 0.85 mM). (C) Glycolipid incorporation of S2G2-GlcNAc₂-Asn-C16 at 0.2, 0.8, and 1.6 mM N-glycolipid. (D) Glycolipid incorporation of Man9-GlcNAc₂-Asn-C16 at 0.2, 0.6, and 1.6 mM N-glycolipid. (D) Glycolipid incorporation of Man5-GlcNAc₂-Asn-C16 at 0.2, 0.8, and 1.6 mM N-glycolipid. All measurements of glycolipid incorporation were done in triplicate.

Having demonstrated that the C16-functionalized S0G2-GlcNAc₂-Asn has superior glycolipid incorporation, presumably due to stronger interactions with lipids of the vesicle bilayer than the C12 glycolipid, C16-functionalized N-glycans were used in all subsequent experiments. N-Glycolipids containing S2G2, Man9, and Man5 headgroups were synthesized and tested for glycolipid incorporation at different concentrations (Figure 1C–E, Scheme 2). Glycolipid incorporation was found to have a roughly linear relationship with final N-glycolipid concentration for all four glycolipids, indicating that the amount of N-glycan on the vesicle surface could be predictably tuned by adjusting the concentration of N-glycolipid in the vesicle formulation (Figure 1B–E). Up to 9.9 × 10⁻⁶, 1 × 10⁻⁵, 1.7 × 10⁻⁵, and 3.1 × 10⁻⁵ mol/wt % of S0G2-GlcNAc₂-Asn-C16, S2G2-GlcNAc₂-Asn-C16, Man9-GlcNAc₂-Asn-C16, and Man₅-GlcNAc₂-Asn-C16 were incorporated into catanionic vesicles, respectively.

N-Glycan-coated vesicles and bare vesicles (containing no sugar) were characterized for the hydrodynamic size by DLS. Bare vesicles were found to have a hydrodynamic diameter of 228 ± 51 nm, which matched well with the previously reported values.^24,25 N-Glycan-coated vesicles were found to be 107–299 nm in diameter, with vesicles containing higher N-glycolipid concentrations typically being larger (Table S1). Nevertheless, there was no clear correlation between the vesicle size and the types of incorporated N-glycans. All vesicles were fairly monodisperse with most vesicle samples having polydispersity indices below 0.1, suggesting homogeneity in the vesicle size distribution (Table S1, Figure S1). Glycan-coated vesicles were also found to be extraordinarily stable, retaining their particle size for up to 6 months as judged by DLS. To confirm that the N-glycolipids had been incorporated into the catanionic vesicles, SEC-purified N-gCVs were treated with endoglycosidase-CC (Endo-CC) from Coprinopsis cinerea, an endoglycosidase enzyme that cleaves between the two N-acetylglucosamines (GlcNAc) of the chitobiose core and with activity toward complex-type and high-mannose N-glycans.³¹ After Endo-CC digestion of each sample for 36 h at 37 °C, the reaction mixture was analyzed by matrix-assisted laser-desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF MS), and the observed m/z values were found to match the expected mass values of the free N-glycans cleaved from the vesicle surface (Figure 2A–D), confirming that each of the synthesized N-glycolipids was incorporated into a catanionic vesicle, respectively.

Figure 2.

Endo-CC glycan release experiments on N-glycan-coated vesicles to confirm N-glycolipid incorporation (A) MALDI-TOF MS spectra for S0G2-GlcNAc₁, calculated, M = 1437.5 Da; found, m/z 1460.7 [M + Na]⁺, (B) S2G2-GlcNAc₁, calculated, M = 2019.7 Da; found, m/z 2040.8 [M + Na − 2H]⁻, (C) Man9-GlcNAc₁, calculated, M = 1679.6 Da; found, m/z 1702.6 [M + Na]⁺, and (D) Man₅-GlcNAc₁, calculated, M = 1031.3 Da; found, m/z 1054.9 [M + Na]⁺.

Based on the particle sizes measured by DLS, and the glycolipid incorporation values quantified by phenol-sulfuric acid assay, the average number of N-glycans per vesicle was estimated using the average number of vesicle particles in solution calculated with the equations as described previously.³² The estimated number of N-glycans per particle for a typical vesicle of medium loading was several thousands of glycan molecules per particle and, except for vesicles containing the Man9-Asn-C16 glycolipid, there was a conserved trend of increasing N-glycan molecules per particle as the concentration of glycolipids in the vesicle formulation was increased (see Table S1, Supporting Information). However, when these values were normalized to the vesicle surface area (assuming that the vesicle is a perfect sphere and the N-glycolipids are homogenously distributed about the surface), the glycan density predictably increased with higher glycolipid concentrations for all vesicle samples (Table S1). These values provide additional insight into the extent of glycan modification of individual vesicle particles to draw important conclusions on the molecular basis of carbohydrate recognition by lectins.

Binding Affinity of the Glycan-Coated Vesicles for Respective Lectins.

The binding affinity of the N-glycan-coated vesicles for several lectins was measured by a competitive inhibition assay. Specifically, N-glycan-coated vesicles were tested for their ability to inhibit binding of Concavalin A (ConA), Sambucus nigra lectin (SNA), and human Galectin-3 (Gal-3) to synthetic N-glycan-BSA conjugates in a modified Enzyme-Linked Lectin Assay (ELLA; Figure 3A). Lectins ConA, SNA, and Gal-3 have specificity for terminal mannose, sialic acid, and galactose residues, respectively. N-Glycan-coated vesicles of varying N-glycan densities were used as inhibitors to reveal the influence of N-glycan presentation on lectin recognition.

Figure 3.

(A) Design of microplate-based competitive inhibition assay, N-glycan-coated vesicles inhibit lectin binding to synthetic N-glycan-BSA conjugate coating antigens. Detection is done by Streptavidin-HRP (Strep-HRP), or a mAb-HRP conjugate in the case of Galectin-3. Colored shapes represent different lectins. (B) Inhibition curve of ConA binding to Man9-BSA conjugates with Man9-gCVs as inhibitors. (C) Inhibition of SNA binding to S2G2-BSA conjugates with S2G2-gCVs. (D) Inhibition of Gal-3 binding to S0G2-BSA conjugates with S0G2-gCVs as the inhibitors. (E) Inhibition curve of ConA binding to Man5-BSA conjugates with Man5-gCVs as inhibitors. Insets in each graph provide the IC₅₀ and multivalent enhancement (β) for the corresponding inhibitors shown on the right side. The molar concentrations are defined as the concentration of the equivalent sugar ligands on the particle, instead of the molar concentration of the particles themselves.

The competitive inhibition curves (Figure 3B−D) demonstrate that ConA, SNA, and Gal-3 bind strongly to the clustered N-glycans on the vesicle surface. Figure 3B depicts the inhibition curve for ConA, where the monovalent Man9-Asn is the weakest inhibitor (IC50 of 1.06 μM) and a clear enhancement in lectin inhibition is observed for the Man9 glycan-coated vesicles (Man9-gCVs). The multivalent enhancement, or β value as coined by Whitesides and co-workers,⁵ was calculated by comparing the IC₅₀ of each glycan-coated vesicle to its corresponding monovalent N-glycan ligand. In the case of ConA, the highest β value was observed for the Man9-gCV (M) sample with intermediate N-glycolipid incorporation (β = 7.6, IC₅₀ = 140 nM). Notably, inhibition of ConA binding was less effective with the densely coated Man9-gCV (H) vesicle, with a β value of 1.6 and an IC₅₀ value of 650 nM. These data suggest that ConA prefers multivalent N-glycan presentations of intermediate density and that higher-density N-glycan presentations sterically occlude lectin binding, possibly by limiting access to the core trimannoside which comprises the main epitope for ConA recognition.³³ Similar observations have been reported by Kiessling and co-workers on mannose-containing glycopolymers as ConA inhibitors.³⁴ Importantly, the IC₅₀ found for Man9-Asn in this assay matches the K_D previously reported for this lectin toward the Man9 oligosaccharide,³⁵ suggesting that the IC₅₀ data obtained by this method are a good approximation of the affinities of the investigated lectins for their N-glycan ligands.

Similarly, the sialic acid-specific lectin SNA displayed a weak affinity for the monovalent S2G2-Asn N-glycan (IC₅₀ = 20 μM) but a significantly enhanced affinity for the S2G2-coated vesicles of intermediate glycan loading [β value = 38, IC₅₀ =520 nM for S2G2-gCV (M)]. The multivalent binding enhancement was more pronounced than that observed for ConA; however, the enhancement was decreased for vesicle samples containing high levels of the S2G2-Asn-C16 glycolipid (i.e., S2G2-gCV (H)). The results suggest that dense N-glycan clustering may hinder accessibility to important glycan epitopes such as the internal LacNAc moiety of the α(1,3)-mannose arm needed for SNA binding.³⁶

Following these results, competitive inhibition experiments were performed with human galectin-3, a lectin overexpressed in several hepatic, pancreatic, and colorectal cancers.² Galectin-3 was found to have a weak affinity for the monomeric galactosylated bi-antennary N-glycan (S0G2-Asn, IC₅₀ = 7.8 μM), consistent with previous reports on this lectin.³⁷ Very modest affinities were observed for S0G2-coated vesicles of low and intermediate glycan density [S0G2-gCV (L) and gCV (M); Figure 3]. However, a pronounced enhancement in lectin binding was observed with the high-density vesicle, S0G2-gCV (H) (β value = 11, IC 50 = 710 nM; Figure 3D). Similar IC₅₀ values have been reported for proteins bearing high-affinity synthetic ligands for Gal-3 and for LacdiNAc-bearing multivalent particles, making S0G2-coated vesicles potent multivalent inhibitors of human Gal-3.^38–40 In contrast to ConA and SNA, Gal-3 prefers high-density presentations of N-glycans bearing the cognate LacNAc ligand. The glycan density dependence in the multivalent binding with Gal-3 may be explained by the mode of recognition as demonstrated by the crystal structure of human Gal-3 in complex with N-acetyllactosamine. The Gal-3 carbohydrate recognition domain (CRD) interacts mainly with the terminal galactose residue of the LacNAc moiety which is more deeply buried in the binding pocket and only makes contacts with the N-Acetyl and C3 hydroxyl groups of the internal GlcNAc residue.⁴¹ Therefore, highly clustered S0G2 N-glycans could still be recognized by Gal-3 at the galactose terminus. Furthermore, Gal-3 is known to self-oligomerize into pentamers through the N-terminal domain, making a lattice of intermolecular glycan-lectin networks that would enhance binding affinity to highly clustered glyconanoparticles.⁴²

Competitive inhibition experiments on ConA with Man5-containing vesicles revealed a remarkable enhancement of affinity for the catanionic vesicle-based Man5 ligands. While the monovalent Man5-Asn glycan had a moderate affinity for ConA in the competitive assay (IC₅₀ = 38.3 μM), the Man5-containing vesicle with a relatively low ligand loading, Man5-gCV (L), showed over 200-fold enhanced affinity for ConA (IC₅₀ = 0.16 μM, β value = 239). Interestingly, vesicles displaying higher densities of the Man5 glycolipid showed less enhancement of the affinity for ConA (Figure 3E). This result may be explained by the higher incorporation efficiencies of the Man5 glycolipid as compared to the Man9 glycolipid (Figure 1), meaning that the Man5 headgroups are in closer proximity to one another and are not as accessible to protein recognition.

Synthesis and Characterization of Mixed N-Glycan-Coated Vesicles.

Glycan-coated vesicles displaying structurally diverse sugars are desirable as better mimics of heterogenous mammalian glycosylation of the cell surface. As a proof of concept, we sought to prepare vesicles displaying different types of sugars such as the S2G2 and Man5 N-glycans, which also feature in variable regions 1 and 2 (V1/V2) of the HIV-1 gp120 trimer.⁴³ To enable quantification of multiple N-glycans on the vesicle surface, a new N-glycolipid incorporation assay based on endo-glycosidase-catalyzed glycan release, 2-aminobenzoic acid (2AA) labeling, and HPLC analysis was developed.

Initial attempts to perform endo-glycosidase release directly from intact N-glycan-coated vesicles were inefficient and required prolonged incubation periods (approximately 3 days). Furthermore, vesicle-associated detergents hampered reaction monitoring by MALDI-TOF MS. To maximize N-glycolipid purity and minimize sample loss, a small-scale purification method was adapted from the biphasic water: phenol/chloroform extraction protocol commonly used in DNA/RNA purification. Due to partial solubility of the N-glycolipids in chloroform, however, dichloromethane (DCM) was used as the organic solvent for the extraction. Importantly, both vesicle-associated detergents (CTAT and SDBS) are soluble in dichloromethane. Pooled N-glycan-coated vesicle samples were mixed with an equal volume of DCM and vortexed vigorously to extract vesicle-associated detergents, the suspension was centrifuged at high speed, and this process was repeated three times to afford pure N-glycolipids in the aqueous phase. The glycolipids were cleaved by wild-type Endo-CC, and the enzymatic hydrolysis was monitored by MALDI-TOF MS analysis. After reaction completion (24−30 h at 37 °C), reductive amination of the reducing oligosaccharides with 2-aminobenzoic acid in the presence of sodium cyanoborohydride afforded the 2AA-labeled N-glycans, which could be separated by analytical reverse-phase HPLC and quantified relative to standard curves of pure 2AA-labeled N-glycan standards by peak integration. The results are summarized in Figure 4B.

Figure 4.

(A) Scheme for the preparation of heteromultivalent vesicles from glycolipids 7 and 9. (B) N-Glycolipid incorporation of heteromultivalent S2G2/Man5-gCV vesicles as quantified by HPLC-based assay.

To validate the N-glycolipid incorporation assay, S2G2-Asn-C16 vesicles of different glycan densities were subjected to the above procedure, and N-glycolipid incorporation was compared to values obtained by the phenol-sulfuric acid assay. As seen in Supplementary Figure S9, no significant difference in N-glycolipid incorporation was found between the two methods, indicating that the HPLC-based assay is a reliable quantitation method giving comparable incorporation values to the well-established phenol-sulfuric acid assay.

Vesicles bearing both Man5 and S2G2 N-glycans were initially prepared by mixing bare vesicles with the respective N-glycolipids at equimolar concentrations of 0.2, 0.8, and 1.6 mM [S2G2/Man5-gCV (LL), S2G2/Man5-gCV (MM), and S2G2/Man5-gCV (HH)]. The glycan content was then quantified by HPLC as described above. It was found that Man5-Asn-C16 incorporation values were very similar for vesicles Man5-gCV (M) and S2G2/Man5-gCV (MM), with values of 1.5 × 10⁻⁶ mol/wt % and 1.4 × 10⁻⁶ mol/w%, respectively. In addition, it was observed that an equimolar ratio of glycolipids S2G2-Asn-C16 and Man5-Asn-C16 was incorporated into samples S2G2/Man5-gCV (LL) and S2G2/Man5-gCV (MM), which were prepared by simply adding the two glycolipids at the same final concentration (Figure 4B). In the case of saturating glycolipid concentrations [i.e., S2G2/Man5-gCV (HH)], approximately two times as much Man5-Asn-C16 was incorporated relative to S2G2-Asn-C16 when added at the same molar concentration, reflecting saturation of the vesicle surface with the bulky S2G2 headgroup and a corresponding increase in the incorporation of the smaller Man5 N-glycolipid. These results indicate that the smaller Man5 glycolipid is incorporated into the vesicle bilayer at the expense of the bulky S2G2 glycolipid in vesicles bearing both glycans.

Heteromultivalent vesicles were also prepared which contain one glycolipid at a fixed concentration and the other glycolipid at variable concentrations. These samples serve to further clarify the influence of an unrelated glycan structure on recognition of the cognate glycan ligand. Thus, the Man5-GlcNAc₂-Asn-C16 glycolipid was fixed at 0.2 mM concentration, while the S2G2-GlcNAc2-Asn-C16 concentration ranged from 0.2 to 1.6 mM, to prepare heteromultivalent vesicles of low Man5 and S2G2 incorporation [i.e., S2G2/Man5-gCV (LL)], low Man5 and medium S2G2 incorporation [i.e., S2G2/Man5-gCV (ML)], and low Man5 and high S2G2 incorporation [i.e., S2G2/Man5 gCV (HL)]. Gratifyingly, these heteromultivalent vesicles contained comparable amounts of the Man5 glycolipid, and with a proportional increase in the S2G2 glycolipid as its concentration was increased (Figure 4B). We also measured the sizes of the nanoparticles by DLS analysis. It was found that heteromultivalent vesicles bearing both S2G2 and Man5 N-glycans were similar in size to the homomultivalent vesicles bearing the S2G2 or Man5 glycans only (Table S1).

Lectin Recognition of Mixed Glycan-Containing Cationic Vesicles.

With the heteromultivalent N-gCVs in hand, we examined their interactions with mannose-specific lectin ConA and sialic acid-specific lectin SNA. Previous work has demonstrated that an irrelevant glycan could affect the affinity and specificity of carbohydrate-binding proteins, such as some lectins and antibodies, for their cognate ligands and carbohydrate epitopes, which was coined as macromolecular crowding effect.^44–50 This effect might be a result of various combined factors such as the interactions of the “irrelevant” glycans with a secondary binding site on the lectin, the steric shielding, and/or the favorable entropic contributions imparted by rapid protein binding and dissociation in a sliding mechanism across the glycoclusters.^51–53 Using a competitive binding assay, we first tested the lectin binding of the heteromultivalent vesicles containing equimolar amounts of S2G2 and Man5 glycans. It was found that at low loading of the Man5 ligand, the presence of the irrelevant sialylated complex N-glycan (S2G2) could significantly enhance the affinity (11 fold) of the Man5 ligand to lectin ConA in comparison with the catanionic vesicles containing comparable loading of the Man5 glycolig and (Figure 5A) [S2G2/Man5-gCV (LL), IC50 = 0.014 μM, β = 11]. Nevertheless, it was observed that increasing the surface density of the sialylated S2G2 glycolipid while keeping a fixed concentration of the Man5 glycolipid [S2G2/Man5 gCV (ML) and S2G2/Man5 gCV (HL)], did not lead to further enhancement in ConA binding. The data suggest that the initial enhancement in ConA affinity is not due to specific recognition of the S2G2 glycan by ConA, but a result of the macromolecular crowding effect.

Figure 5.

(A) Competitive inhibition of ConA binding to Man5-BSA conjugates by S2G2/Man5-C16 heteromultivalent N-gCVs. (B) Competitive inhibition of SNA binding to S2G2-BSA conjugates by S2G2/Man5-C16 heteromultivalent N-gCV and homomultivalent S2G2-C16 N-gCV.

We next tested the competitive inhibition of SNA with homo- and heteromultivalent vesicles containing similar amounts of the S2G2 N-glycolipid, namely, vesicles S2G2-gCV (L) and S2G2/Man5-gCV (LL). Interestingly, the heteromultivalent vesicles were found to have slightly reduced affinity for the SNA lectin in comparison with the S2G2 glycolipid containing vesicles with similar ligand loading (Figure 5B). This result suggests that the presence of the irrelevant Man5 ligand is actually detrimental to the recognition of the S2G2 glycan by SNA, showing negative macromolecular crowding effect. Similar results were observed when comparing heteromultivalent vesicles containing higher concentrations of the S2G2 glycolipid as SNA ligands (Supplementary Figure S13). By using synthetic sugar-bearing polymers to install a mucin-like shield on red blood cell surface, Godula and co-workers have shown that glycocalyx crowding slows the rate of lectin SNA association with the sialylated glycan ligands on the cell surface but enhances the binding complex stability, resulting in an overall enhanced binding of soluble and virus-associated SNA to the host glycan receptors.⁵⁴ On the other hand, Kikkeri and co-workers have shown that several CBPs show decreased affinities for heteroglycoclusters when compared to homoglycoclusters bearing their cognate sugar ligands.⁴⁷ Taken together, these results suggest that macromolecular crowding can modulate the affinity of CBPs to their glycan ligands, and the effects (positive or negative) are dependent on the nature of the proteins/glycan ligands and the mode of the interactions. The glycan-bearing catanionic vesicles described here provide a facile platform for further investigating specific lectin–glycan interactions.

CONCLUSIONS

A facile synthesis and characterization of catanionic vesicle-based multivalent N-glycan ligands for lectins are described. The N-glycan-coated vesicles show significantly enhanced affinity for lectins Sambucus nigra agglutinin (SNA), concanavalin A (ConA), and human galectin-3 over their specific monovalent N-glycan ligands (sialylated, high-mannose, and asialylated N-glycans), respectively, demonstrating a clear clustering glycoside effect. In addition, a positive macromolecular crowding effect was observed on the binding of ConA to high-mannose N-glycan ligands when catanionic vesicles bearing mixed high-mannose and complex-type N-glycans were used. These N-glycan-coated catanionic vesicles are stable and easy to formulate with variable density of ligands. The glycan-bearing nanoparticles should be useful as vehicles for drug delivery and as potent inhibitors for intervening specific carbohydrate–protein interactions in disease processes.

MATERIALS AND METHODS

Materials.

Lauric acid was purchased from Alfa Aesar. Sodium dodecylbenzenesulfonate (SDBS) and cetyltrimethylammonium tosylate (CTAT) were purchased from TCI America and Merck, respectively. N,N′-dicyclohexylcarbodiimide (DCC) coupling reagent was purchased from TCI America. High-binding polystyrene 96-well ELISA plates were purchased from Santa Cruz Biotechnology, Inc. Biotin-ConA and biotin-SNA were purchased from Vector Laboratories. Galectin-3 with a His tag at the N-terminus was purchased from Abcam. All other chemicals were purchased from Sigma-Aldrich unless sated otherwise and used as received.

Methods.

The bi-antennary complex-type containing sialoglycoprotein (SGP) was prepared from dried hen’s egg yolk powder following the reported procedures.^55,56 The crude soybean agluttinin was prepared from soybean flour following a previously reported method.²⁸ Analytical reverse-phase HPLC was performed on a Waters Alliance e2695 HPLC system equipped with a dual-absorbance 2489 UV/Vis detector. Separations were performed using a C18 column (YMCTriart C18, 4.6 × 250 mm, 5 μm) at a flow rate of 1 mL/min using a linear gradient of 30–70% MeCN containing 0.1% TFA (trifluoroacetic acid) over 30 min and at 50 °C. Preparative HPLC was performed on a Waters 600 HPLC instrument with a preparative reverse-phase C18 column (Waters Symmetry 300, 19 × 300 mm). NMR spectra were measured with a Bruker AV III 600 MHz NMR spectrometer, and the chemical shifts were assigned in parts per million. MALDI-TOF MS analysis was performed using a Bruker UltrafleXtreme (UTX) mass spectrometer with TOF/TOF detection and a dihydroxybenzoic acid/dimethylaniline (DHB/DMA) matrix, and samples were analyzed under reflectron ion mode. LC ESI-TOF MS was performed using an X-Bridge Shield RP18 3.5 μm (2.1 × 50 mm) short column coupled to a Micromass ZQ-4000 single quadruple mass spectrometer. High-resolution mass was taken on an Exactive Plus Orbitrap (Thermo Scientific) with an Agilent Poroshell 300SB-C8 column (5 μm, 75 × 1 mm).

Synthesis of Lauric Acid and Palmitic Acid NHS Esters.

NHS-activated lipids NHS-C12 and NHS-C16 were prepared using N,N′-dicyclohexylcarbodiimide (DCC) as the coupling reagent. Briefly, 1 g of lauric acid (C12) or palmitic acid (C16) was dissolved in 5 mL of DCM and 5 mL of tetrahydrofuran in a 50 mL round-bottom flask. The solution was then placed in an ice bath, capped with a rubber septum, and degassed under positive argon pressure using a Schlenk-line. With argon still passing through the solution, 1.1 mole equivalents of NHS were added followed by 1.1 mole equivalents of DCC. Finally, 0.05 mole equivalents of dimethyl-aminopyridine (DMAP) were added and the solution was stirred for 2 h when TLC indicated the completion of the reaction. The dicyclohexylurea (DCU) byproduct was precipitated by adding 5 mL of hexanes and 5 mL of diethyl ether, followed by vacuum filtration through a Celite pad. The filtrate was then concentrated under vacuum, and the residue was purified by flash chromatography using a Biotage SNAP Ultra 25 g cartridge: Mobile phase A: hexanes, mobile phase B: ethyl acetate; a 5–30% gradient of ethyl acetate over 25 column volumes. Fractions containing the product were pooled and concentrated in vacuo to obtain the target NHS-activated lipids.

N-Hydroxysuccinimidyl Lauric Acid.

1.29 g of NHS-C12 (87.2% yield) as a white solid.⁵⁷ NHS-C12 ¹H-NMR (400 MHz, CDCl₃): δ(ppm) 2.84 (s, 4H, CH2 from NHS), 2.60 (t, 2H, J = 5.0 Hz, carbon C2), 1.74 (p, 2H, J = 5.0 Hz, carbon C3), 1.42 (m, 2H, CH2, carbon C4), 1.28 (m, 14H, CH2, carbons C5−11), 0.89 (t, 3H, J = 4.6 Hz, CH3).

N-Hydroxysuccinimidyl Palmitic Acid.

1.31 g NHS-C16 (94.8% yield) as a white solid.⁵⁷¹H-NMR (400 MHz, CDCl₃): δ(ppm) 2.83 (s, 4H, CH2 from NHS), 2.60 (t, 2H, J = 7.5 Hz, carbon C2), 1.73 (p, 2H, J = 7.5 Hz, carbon C3), 1.41 (m, 2H, carbon C4), 1.26 (m, 22H, CH2 carbons C5−15), 0.89 (t, 3H, J = 7.0 Hz, CH3).

General Method for Synthesis of the N-Glycolipids.

Each Glycan-Asn (10 mg) was first dissolved in 200 μL of water, to which 800 μL of N,N′-dimethylformamide (DMF) was added. Separately, 1.1 mole equivalents of the NHS-activated lauric or palmitic acid was dissolved in 1 mL of DMF and the solution was added to the glycan-Asn solution in a 5 mL amber vial. To this reaction mixture was added 20 μL of triethylamine (TEA). The resulting mixture was stirred at r.t. for 3 h when LC–MS indicated the completion of the reaction. The mixture was diluted in 50% aqueous acetonitrile and lyophilized. The residue was then redissolved in 2 mL of water and purified by preparative RP-HPLC using a 30–70% MeCN (0.1% TFA) gradient over 30 min (mobile phase A: H₂O containing 0.1% TFA, mobile phase B: acetonitrile containing 0.1% TFA). The fractions containing the glycolipid were pooled and lyophilized to obtain the respective glycolipids as a white solid. Proton chemical shifts for the N-glycan portions were assigned according to the previous NMR assignments of N-glycans reported by Kajihara, Pancera, and Shahzad-ul-Hussan.^58–60

S0G2-GlcNAc₂-Asn-C12 (5).

7.4 mg (67% yield after HPLC purification). ¹H-NMR (400 MHz, D₂O): δ 5.11 (s, 1H, Man-4 H-1), 5.04–5.03 (d, 1H, J = 9.2 Hz, GlcNAc-1 H-1), 4.94–4.90 (s, 1H, Man-4′ H-1), 4.72–4.68 (m, 1H, Man-3 H-1) 4.64–4.54 (m, 3H, GlcNAc-2, GlcNAc-5 and GlcNAc-5′ H-1), 4.49–4.43 (m, 2H, Gal-6 and Gal-6′ H-1), 4.24 (s, 1H, Man-3 H-2), 4.18 (s, 1H, Man-4 H-2), 4.10 (s, 1H, Man-4′ H-2), 3.98–3.47, 2.85–2.74 (m, 2H, Asn-CH₂), 2.29–2.22 (m, 2H, C2 of lipid), 2.07–1.99 (m, 12H, NHAc), 1.60–1.53 (m, 2H, C3 of lipid), 1.31–1.20 (m, 16H, C4–C11 of lipid), 0.84 (t, 3H, J = 3.5 Hz, CH₃ of lipid). ¹³C-NMR (125 MHz, D₂O): δ 175.7, 174.2, 174.1, 174.0, 173.6, 172.1, 102.5, 100.9, 99.9, 98.9, 96.6, 79.8, 79.1, 78.2, 78.0, 77.8, 75.9, 75.9, 75.7, 74.9, 74.2, 73.9, 73.1, 72.4, 72.2, 72.0, 71.6, 71.5, 70.5, 69.7, 68.9, 68.1, 66.8, 65.2, 61.2, 60.5, 59.5, 54.4, 53.4, 48.6, 38.2, 36.6, 35.1, 31.0, 28.7, 28.5, 28.4, 28.3, 28.1, 24.8, 21.9, 21.8, 21.7, 13.2. Analytical HPLC: t_R = 10.55 min. HRMS (ESI-MS): calcd for C₇₈H₁₃₂N₆O₄₉, [M + 2H]²⁺ = 1938.8167 Da; found, m/z 1938.8165 [M + 2H]²⁺.

S0G2-GlcNAc₂-Asn-C16 (6).

8 mg (71% yield after HPLC purification). ¹H-NMR (400 MHz, D₂O): δ 5.10 (s, 1H, Man-4 H-1), 5.06–4.99 (m, 1H, GlcNAc-1 H-1), 4.9 (s, 1H, Man-4′ H-1), 4.58–4.51 (m, 2H, GlcNAc-5 and GlcNAc-5′ H-1), 4.47–4.40 (m, 2H, Gal-6 and Gal-6′ H-1), 4.21 (s, 1H, Man-3 H-2), 4.16 (s, 1H, Man-4 H-2), 4.08 (s, 1H, Man-4′ H-2), 3.95–3.47, 2.85–2.65 (m, 2H, Asn-CH₂), 2.25–2.15 (m, 2H, C2 of lipid), 2.09–1.95 (m, 12H, NHAc), 1.59–1.45 (m, 2H, C3 of lipid), 1.27–1.16 (m, 24H, C4–C11 of lipid), 0.84–0.78 (m, 3H, CH₃ of lipid). ¹³C-NMR (125 MHz, D₂O): δ 174.7, 174.2, 173.9, 173.3, 102.5, 100.9, 100.1, 99.1, 96.6, 79.7, 79.2, 78.1, 75.9, 74.9, 74.3, 74.0, 73.1, 72.1, 71.6, 71.5, 70.5, 69.8, 69.0, 68.1, 66.9, 65.3, 61.3, 60.6, 59.6, 54.4, 53.5, 48.5, 38.3, 35.3, 31.4, 29.3, 29.3, 29.2, 28.9, 28.7, 25.0, 22.1, 22.0, 21.8, 13.4. Analytical HPLC: t_R = 18.85 min. HRMS (ESI-MS): calcd for C₈₂H₁₄₀N₆O₄₉, [M + 2H]²⁺ = 1994.8793 Da; found, m/z 1984.8799 [M + 2H]²⁺.

S2G2-GlcNAc₂-Asn-C16 (7).

10.1 mg (92% yield after HPLC purification). ¹H-NMR (400 MHz, D₂O): δ 5.16 (s, 1H, Man-4 H-1), 5.09–5.02 (m, 1H, GlcNAc-1 H-1), 4.95 (s, 1H, Man-4′ H-1), 4.66–4.56 (m, 3H, 2, GlcNAc-5 and GlcNAc-5′ H-1), 4.45 (s, 2H, Gal-6 and Gal-6′ H-1), 4.25 (s, 1H, Man-3 H-2), 4.20 (s, 1H, Man-4 H-2), 4.11 (s, 1H, Man-4′ H-2), 4.06–3.48, 2.87–2.74 (m, 2H, Asn-CH₂), 2.70–2.61 (m, 2H, NeuAc-7 and NeuAc-7′ H-3_eq), 2.33–2.19 (m, 2H, C2 of lipid), 2.12–1.98 (m, 18H, NHAc), 1.86–1.76 (m, 2H, C3 of lipid), 1.62–1.51 (m, 2H, NeuAc-7 and NeuAc-7′ H-3_ax), 1.32–1.20 (m, 24H, C4–C15 of lipid), 0.86 (t, 3H, J =6.9 Hz, CH₃ of lipid). ¹³C-NMR (125 MHz, D₂O): δ 174.4, 174.2, 103.1, 100.0, 98.8, 95.1, 80.1, 79.4, 77.9, 75.9, 73.9, 73.1, 72.3, 72.0, 71.6, 70.7, 70.2, 69.8, 69.7, 68.9, 67.8, 67.2, 66.9, 66.4, 65.5, 62.7, 62.4, 61.2, 60.6, 59.7, 54.2, 53.5, 51.6, 51.3, 38.8, 38.5, 31.4, 29.2, 28.9, 24.9, 22.1, 22.00, 21.7, 21.6, 13.4. Analytical HPLC: t_R = 17.82 min. HRMS (ESI-MS): calcd for C₁₀₄H₁₇₄N₈O₆₅, [M + 2H]²⁺ = 1289.0368 Da; found, m/z 1289.0327 [M + 2H]²⁺.

Man9-GlcNAc₂-Asn-C16 (8).

5.3 mg (47% yield after HPLC purification). ¹H-NMR (400 MHz, D₂O): δ 5.39 (s, 1H, ManA H-1), 5.35–5.25 (m, 2H, Man4, ManC H-1), 5.12 (s, 1H, ManB H-1), 5.08–4.99 (m, 3H, Man D1-D3), 4.85 (s, 1H, Man4′ H-1), 4.20 (s, 1H, Man3 H-2), 4.14–3.50, 2.87–2.65 (m, 2H, Asn-CH₂), 2.28–2.14 (m, 2H, C2 of lipid), 2.05 (s, 3H, NHAc), 1.97 (s, 3H, NHAc), 1.60–1.47 (m, 2H, C3 of lipid), 1.30–1.16 (m, 24H, C4–C15 of lipid), 0.86–0.79 (m, 3H, CH₃ of lipid). ¹³C-NMR (125 MHz, D₂O): δ 174.7, 174.1, 173.3, 101.7, 100.2, 99.0, 97.5, 78.2, 77.9, 72.7, 72.3, 70.7, 69.9, 69.5, 69.0, 66.4, 65.2, 60.7, 60.6, 38.3, 36.7, 35.2, 31.4, 29.3, 29.2, 28.9, 28.7, 25.00, 22.1, 21.8, 13.4. Analytical HPLC: t_R = 18.55 min. HRMS (ESI-MS): calcd for C₉₀H₁₅₄N₄O₅₉, [M + 2H]²⁺ = 2236.9319 Da; found, m/z 2236.9345 [M + 2H]²⁺.

Man5-GlcNAc₂-Asn-C16 (9).

6.5 mg (46% yield after HPLC purification). ¹H-NMR (400 MHz, D₂O): δ 5.09–5.00 (m, 2H, ManE, ManF H-1, and ManG H-1), 4.87–4.81 (m, 2H, ManD H-1 and ManC H-1), 4.61–4.52 (m, 1H, GlcNAc-B H-1), 4.25–4.17 (m, 1H, ManC H-2), 4.12–4.07 (m, 1H, ManD H-2), 4.05–4.00 (m, 2H, ManE and ManF H-2), 3.89–3.44, 2.83–2.67 (m, 2H, Asn-CH₂), 2.26–2.11 (m, 2H, C2 of lipid), 2.02 (s, 3H, NHAc), 1.94 (s, 3H, NHAc), 1.57–1.44 (m, 2H, C3 of lipid), 1.26–1.12 (m, 24H, C4–C15 of lipid), 0.81–0.75 (m, 3H, CH₃ of lipid). ¹³C-NMR (125 MHz, D₂O): δ 174.6, 173.9, 173.4, 172.0, 105.3, 102.1, 101.8, 101.4, 101.2, 100.1, 99.3, 98.9, 94.9, 93.3, 89.9, 81.0, 80.4, 79.2, 78.6, 78.1, 75.8, 73.9, 72.9, 72.3, 71.6, 70.5, 70.2, 70.0, 69.8, 69.6, 69.0, 66.5, 66.4, 65.3, 64.8, 60.7, 60.6, 59.4, 54.6, 53.6, 50.3, 48.7, 46.4, 44.1, 35.3, 31.5, 29.5, 29.3, 29.0, 28.8, 25.1, 22.2, 22.0, 21.9, 13.4. Analytical HPLC: t_R = 19.92 min. HRMS (ESI-MS): calcd for C₆₆H₁₁₄N₄O₃₉, [M + H]⁺ = 1587.7133 Da; found, m/z 1587.7001 [M + H]⁺.

Formulation of Bare and Glycan-Coated Vesicles.

Catanionic vesicles were prepared according to a previously published procedure.²⁴ Briefly, 35.3 mg of SDBS (0.1 mmol) was dissolved in 5 mL of water, and 15.4 mg of CTAT (3.4 × 10⁻² mmol) was directly added to achieve a 1% w/v final surfactant concentration. The suspension was stirred for 1 hour to ensure complete dissolution and then allowed to equilibrate in the dark for 48 h. Vesicles were purified from free detergent by Sepharose G-50 and elution was confirmed by DLS intensity.

Glycan-coated vesicles were prepared from 1 mL aliquots of crude vesicle solution taken prior to purification by Sepharose G-50 (preformed vesicle). Then, varying amounts of each N-linked glycolipid were added from a 10 mg/mL stock solution to reach final concentrations ranging from 0.2 to 1.6 mM in water. The mixture was then stirred at room temperature for 24 h, and the vesicles were purified by Sepharose G-50.

DLS Measurements of the Hydrodynamic Size.

Intact vesicles were characterized for the mean hydrodynamic radius using a Photocor-FC DLS instrument. Measurements were done using a 5 mW laser polarized at 633 nm, and the scattering angle was 90° for all measurements. All measurements were conducted at room temperature in 1 mL glass vials.

General Method for Endo-CC-Catalyzed Glycan Release.

Aliquots of each glycan-coated vesicle (0.5 mL) were incubated with ethanol at a 20% final concentration (v/v) for 1 h and then lyophilized. The lyophilized film was dissolved in 1 mL of water containing 0.1% TFA and loaded onto a preconditioned 100MG Hypersep C18 column (per manufacturer instructions). The column was then washed with 1 mL of water a gradient 5%−90%) of aqueous acetonitrile containing 0.1% TFA. The fractions were analyzed by MALDI-TOF MS with DHB as the matrix under reflectron-positive or negative mode. Efficient separation of the glycolipids from the surfactants was achieved, with the 30 and 40% acetonitrile fractions containing the glycolipids and the 90% acetonitrile fraction containing the vesicle-associated detergents. Fractions containing the glycolipid were combined and lyophilized and then resuspended in 200 μL of 100 mM PB buffer (pH 7.4) prior to hydrolysis by 1 μL of Endo-CC (1 mg/mL) at 37 °C for 36 h.

Endo-CC Cleavage of S0G2-GlcNAc2-Asn-C16.

MALDI-TOF-MS of the released glycan, calcd for S0G2-GlcNAc₁ (C₅₄H₉₁N₃O₄₁), M = 1437.5 Da; found, m/z 1460.7 [M + Na]⁺.

Endo-CC Cleavage of S2G2-GlcNAc₂-Asn-C16.

MALDI-TOF-MS of the released glycan, calcd for S2G2-GlcNAc₁ (C₇₆H₁₂₅N₅O₅₇), M = 2019.7 Da; found, m/z 2040.8 [M + Na − 2H]⁻.

Endo-CC Cleavage of Man9-GlcNAc₂-Asn-C16.

MALDI-TOF-MS of the released glycan, calcd for Man9-GlcNAc₁ (C₆₂H₁₀₅NO₅₁), M = 1679.6 Da; found, m/z 1702.6 [M + Na]⁺.

Endo-CC Cleavage of Man5-GlcNAc₂-Asn-C16.

MALDI-TOF-MS of the released glycan, calcd for Man5-GlcNAc₁ (C₃₈H₆₅NO₃₁), M = 1031.3 Da; found, m/z 1054.9 [M + Na]⁺.

Quantification of Glycan Incorporation by Phenol-Sulfuric Acid (PSA) Assay.

Glycan incorporation was measured using a variation of the PSA assay method employed by Lee and co-workers.³⁰ First, stock solutions of the relevant glycan-Asn (S0G2-GlcNAc₂-Asn, Man9-GlcNAc₂-Asn, and S2G2-GlcNAc₂-Asn) were prepared at 2 mg/mL in water for use as standards. These stock solutions were serially diluted in 1.5 mL microcentrifuge tubes to obtain 0.8, 0.6, 0.4, 0.2, and 0 μg/μL final concentrations in 50 μL volumes. G-50 fractions of glycan-coated vesicles were combined into a 1 mL aliquot in a preweighed microcentrifuge tube before being lyophilized. The final weight of the vesicle was recorded, and the solid was resuspended in 250 μL of water. Fifty microliters of each vesicle solution were taken for analysis by PSA assay. For samples with high final concentrations of the glycolipid (≥0.4 mM), the resuspended solution was diluted 1:2 in water.

To each 50 μL of the solution in a microcentrifuge tube was added 150 μL of concentrated sulfuric acid quickly followed by 30 μL of 5% aqueous phenol. The mixtures were vortexed and then incubated in a 90 °C heat block for 5 min, followed by a room temperature water bath for 5 min. When the tubes had reached room temperature, 60 μL of 95% ethanol were added to each tube to a final concentration of 20% v/v. The solution in the microcentrifuge tubes was then transferred to a 96-well microplate and allowed to incubate for 1 h before reading the absorbance at 490 nm. A bare vesicle sample was used as a negative control, and all samples were plated in triplicate. Glycan incorporation was quantified using the line of best fit for the plot of N-glycan-Asn concentration vs absorbance at 490 nm as a standard curve. Incorporation percent (mol/wt %) was then calculated by dividing the amount of vesicle-incorporated glycolipid by the recorded weight of dry vesicle.

Competitive ELLA.

BSA conjugates (100 μL of a 5 μg/mL solution in PBS, pH 7.4) were added to a 96-well plate (Santa Cruz Biotechnology, Inc. high-binding polystyrene) and incubated at 4 °C overnight. Liquid was removed, and the wells were washed three times with 200 μL of PBST (PBS +0.05% Tween-20). The wells were then blocked with 300 μL of the blocking buffer (3% BSA w/v in PBS) at 37 °C for 1 h. After washing, 100 μL serial dilutions of each lectin [10–2 to 10⁻⁸ mg/mL lectin, in PBS for biotin-SNA and His-Galectin-3, in 10 mM HEPES (+0.15 M NaCl, 0.01 mM Mn²⁺ and 0.1 mM Ca²⁺) for biotin-ConA] were added to the wells and incubated at room temperature for 1 h. The wells were washed with PBST four times. For the biotinylated lectins (i.e., ConA and SNA), 100 μL of a 1: 15,000 dilution Streptavidin-HRP (abcam) in PBS were added and incubated for 1 h at room temperature. For Galectin-3, 100 μL of a 1: 1000 dilution of anti-His tag mouse mAb (R&D Systems) in PBS were added to each well and incubated for 1 h at room temperature. Wells were washed five times with PBST before mixing equal volumes of KPL TMB peroxidase substrate and KPL TMB peroxidase substrate Solution B and adding 100 μL to each well. The plate was incubated for 30 min at room temperature in the dark, and the reaction was stopped by addition of 100 μL of 1 M phosphoric acid. Absorbance was measured at 450 nm with a Spectramax M5e microplate reader with background correction at 550 nm. The optimal concentration of each lectin was taken as that resulting in an absorbance reading of 0.8–1 AU and was as follows: 5 × 10⁻⁶ mg/mL biotin-ConA, 1 × 10⁻⁵ mg/mL biotin-SNA, and 2 × 10⁻³ mg/mL His-Galectin-3.

For inhibition experiments, 2-fold serial dilutions of each monovalent glycan-Asn (1 mM stock solution) and 4-fold serial dilutions of each N-glycan-coated vesicle were prepared. One hundred microliters of each inhibitor solution was premixed with the corresponding biotinylated lectin (at the optimal concentration noted above) and incubated for 30 min at r.t. in a separate U-bottom 96 well plate. The lectin/inhibitor mixture was then added to the blocked, N-glycan-BSA-coated plate and incubated for 1 h at r.t. The rest of the procedure was followed as outlined above. Percent inhibition was determined using the equation below, where “A” represents absorbance:

$$\%\text{inhibition}=\left(A_{\left(\text{no inhibitor}\right)}-A_{\left(\text{with inhibitor}\right)}/A_{\left(\text{no inhibitor}\right)}\right)\times 100$$

Each sample was plated in triplicate for competitive ELLA and in direct binding assays. IC₅₀ values were determined by nonlinear regression using Graphpad Prism software.

ABBREVIATIONS

GlcNAc

N-acetylglucosamine

TEA

triethylamine

DMSO

dimethyl sulfoxide

MALDI

matrix-associated laser-desorption/ionization

CTAT

cetyltrimethylammonium tosylate

SDBS

sodium dodecylbenzenesulfonate

HPLC

high-performance liquid chromatography

SGP

sialoglycopeptide

SBA

soybean agglutinin