Efficient Assembly of Quantum Dots with Homogenous Glycans Derived from Natural N-linked Glycoproteins

Abstract

Coating inorganic nanoparticles with polyethylene glycol (PEG)-appended ligands, as means to preserve their physical characteristics and promote steric interactions with biological systems, including enhanced aqueous solubility and reduced immunogenicity, has been explored by several groups. Conversely, macromolecules present in the human serum and on the surface of cells are densely coated with hydrophilic glycans that act to reduce non-specific interactions, while facilitating specific binding and interactions. In particular, N-linked glycans are abundant on the surface of most serum proteins and are composed of a branched architecture that is typically characterized by a significant level of molecular heterogeneity. Here we provide two distinct methodologies, covalent bioconjugation and self-assembly, to functionalize two types of Quantum Dots with a homogenous, complex-type N-linked glycan terminated with a sialic acid moiety. A detailed physical and functional characterization of these glycan-coated nanoparticles has been performed. Our findings support the potential use of such fluorescent platforms to sense glycan-involved biological processes, such as lectin recognition and sialidase-mediated hydrolysis.

Graphical Abstract

INTRODUCTION

Luminescent semiconductor Quantum Dots (QDs) are powerful tools that have great utility for use as imaging agents and sensing platforms.¹ They offer the possibility to study complex processes in biology,² and research aimed at improving their luminescence and surface properties is ongoing.³ QD-based sensors have impacted bioimaging research more than any other nanomaterial, due to their unique photophysical properties and versatility in applications ranging from in vitro and in vivo process monitoring, cell tracking and substance detection.⁴ The development of new synthetic methods,⁵ coatings⁶ and surface modifications⁷ have been essential to allow access to stable, biocompatible, high quantum yield and reactive nanocrystals.⁸ QDs present a large surface area that allows for the conjugation of multiple biomolecules of interest with the additional benefit of controlling the density and orientation of those biomolecules. Following an initial coating step with polyethylene glycol (PEG) or an analogous biocompatible linker, the water-soluble QDs can be readily coupled to any biological macromolecule of interest.⁹

PEG has been established as a molecular coating that allows for QD solubilization with low non-specific binding.¹⁰ PEG-modified QD surfaces can permit easy attachments, via simple chemical ligation strategies, to various peptides or proteins.⁷ However, PEG is a molecule of non biological origin and while toxicity is negligible, cellular vacuolation has been observed in patients treated with PEGylated Biopharmaceuticals.¹¹ In the effort to find biologically compatible alternatives to PEG, the use of homogeneous complex bi-antennary N-glycans that can be isolated in large quantities from chicken egg yolks has gained attention.¹² Since nanoparticle glycoconjugates with high-value glycans have been synthesized and successfully utilized in the context of viral nanoparticle immunogens¹³ or host cell infection among others,¹⁴ we hypothesized that a straightforward access to glycan-functionalized QDs could provide highly biologically compatible nanoparticles while maintaining the desirable properties of PEG. In this study, we focus on displaying a complex homogenous N-linked glycan at the surface of QDs to afford nanoparticles of greater intrinsic compatibility in biological systems in tandem with capabilities as sensors. A glycan-functionalized QD also features intrinsic tools for biophysical assays not present in regular proteins. In developing a QD with a surface featuring an N-linked glycan, it would be advantageous to present the biomolecule with a surface density similar to the quantity encountered on a glycoprotein, thereby enabling the study of the interactions between biological surfaces; these interactions are highly relevant to a number of processes.^15,16 Furthermore, a QD functionalized with glycans in such a manner more closely resembles human serum proteins than typical PEG coatings.

Sialic acid (SA) terminated N-linked glycans are an important class of cell surface-displayed glycans with roles in IgG-mediated immunity,¹⁷ viral particle binding,¹⁸ cancer cell metathesis,¹⁹ and cellular fluid uptake.²⁰ These functions stem from specific protein binding events as well as the general physical properties of hydrophilicity combined with bulk negative charge on the cell surface derived from clusters of sialyl moieties. As specific binding partner for lectins and antibodies, SA-terminated glycans are involved in a number of both physiological and pathological responses.²¹ A majority of SA-terminated glycan studies are currently pursued via linkage-specific sialidases, esterases, lyases and/or lectins in tandem with routine chromatography and mass spectrometry methods. ^{12, 22} The fusion of SA-terminated glycans with state-of-the-art QD biosensor technology would afford a powerful platform for the characterization of sialoglycoconjugates across a number of biological settings. In addition, sialic acid coatings on proteins have been proven to prevent accelerated blood clearance and extend circulation times, thereby avoiding phagocytic pathways.²³

The SA moiety and related oligosaccharides²⁴ have been conjugated to QDs, enabling quantitative analysis of the diffusion dynamics and endocytosis of cell membrane-bound SAs²⁵ and other analytical applications.^{26, 27} However, to study their role in glycobiology conjugation of full-length glycans presenting SA at their termini on the QD surfaces is highly desirable.²⁸ SA-terminated glycans have been conjugated to gold nanoparticles, which allowed for the detection of the influenza virus at nanomolar concentrations.²⁹ Here we use a complex bi-antennary N-glycan found in humans that is readily isolated from chicken egg yolks.¹² In contrast to the tedious and challenging total chemical synthesis of complex glycans, the SA-glycan presented here can be obtained in high yield and quantity, while allowing for functionalization on its reducing end as well as conjugation to peptides and proteins. ³⁰ QDs have previously been coated with some complex carbohydrates as well such as dextran ³¹ and the complex bi-antennary glycan presented on this work,³² where an efficient tool to detect hemagglutinin interaction through fluorescence polarization was described. Here we show straightforward conjugation of glycans onto QDs and their bioavailability towards lectins and glycan modifying enzymes by using a readily available, hydrophilic coating that is identical to that present on the glycoproteins coating mammalian cells.³³

Controlling the architecture of nanoparticle-conjugates is an essential step towards the wide-ranging use of such nanomaterials. With the potential changes produced by molecule type and load, specific methodologies are needed to enable characterization of the assembled nanoparticle constructs. Understanding the physicochemical properties of the generated nanocrystals will help to understand their cellular behavior. For example, cellular targeting could be mediated by surface or coating charge, as it has been shown how a particular coating can selectively drive specific cellular internalization. ^34,35 Our experience in developing QD-peptide bioconjugates suggests that QD functionalization can be mediated through non-covalent self-assembly as well as through direct covalent attachment using bioconjugation reactions, and that both approaches have distinct utility in nanoparticle construction. In this work, two complimentary approaches have been developed to introduce SA-glycan to the surface of QDs: via His-tag-driven self-assembly or covalent oxime ligation. The resulting SA-QDs were characterized through electrophoretic mobility, lectin binding and sialidase digestion assays, thereby confirming that QD functionalization with complex glycans preserves the properties of the carbohydrate on the surface of the QDs (Figure 1).

Figure 1.

Approaches for SA-glycan detection on QDs. (A) Chemical structures of unprotected and protected SA termini. (B) Tagged lectin recognition by FRET. (C) Sialidase cleavage of SA.

RESULTS AND DISCUSSION

Coupling of the glycan to the QD surfaces to form bio-functional fluorescent conjugates with potential utility for glycosylation, was achieved via two distinct conjugation strategies (as schematically represented in Figure 2). The first relies on direct metal-coordination between the Zn-rich surfaces of the QDs and polyhistidine-appended glycan, while the second uses covalent conjugation promoted by oxime ligation between aldehyde presenting QDs and the glycan molecules. The self-assembly strategy builds on previous works, which showed that biological macromolecules such as peptides, proteins and nucleic acids modified with specifically designed polyhistidine (e.g., His₆, His₇ or His₈) linkers can self-assemble onto ZnS-overcoated QDs surface coated with small lipoic acid (LA)-based ligands.^{36, 37, 38, 39} Those studies have shown that this self-assembly route provides QD-biomolecule conjugates that are highly stable and biologically active. Additionally, the valence of these conjugates can be controlled by varying the biomolecule-to-QD ratio. The oxime ligation exploits a highly efficient coupling reaction, and control over the number of bound molecules is achieved by adjusting the fraction of LA-PEG-CHO ligands per QD; this is accomplished during the phase transfer step, where ligand exchange of the native hydrophobic QDs is carried using a mixture of LA-PEG-OMe (inert) and LA-PEG-CHO. Here a small percentage (5–10% of LA-PEG-CHO mixed with 95–90% LA-PEG-OMe) were targeted for the QDs used.⁴⁰ Both strategies are easy to implement and can be straightforwardly extended to other nanomaterials.⁴¹

Figure 2.

(top) Strategies for glycan immobilization on QDs. Shown are: (A) self-assembly driven by metal-His coordination between QDs photo-ligated with either LA-PEG or LA-zwitterion (LA-ZW) and polyhistidine-appended glycan. (B) Covalent modification between QDs photoligated with LA-PEG-CHO and SA-glycans. Loading control is achieved at self-assembly when using His-tagged peptides and at coating exchange when covalently attaching peptides. (bottom) (A) Synthesis of glycan derivatives for conjugation onto QDs. (B) Glycan 1a structure. The 11-unit polysaccharide is composed of N-acetylglucosamine (blue square), mannose (green circle), galactose (yellow circle) and α(2,6) linked SA (red diamond) units.

Self-assembly of Bi-antennary N-Glycans on LA-PEG- or LA-ZW-capped QDs.

Functionalized glycans were first chemically synthesized starting from the α-bromoacetylated glycan derivatives containing either an unprotected SA glycan 1a or a benzyl ester-modified sialic acid 1b (serving as an internal control), as shown in Figure 2 (bottom). The thioether bridged SA constructs were synthesized via nucleophilic substitution by a cysteine residue, in 100 mM ammonium bicarbonate buffer at pH 8. The reaction was carried at room temperature for 3h (See Figure 2 bottom). The compounds 1a/1b were conjugated with the peptide Cys-Trp-Ahx-Pro₉-Gly₂-His₆ designed to incorporate a proline helix spacer between the glycan and the His tag to yield 2a/2b. Self–assembly of QD-conjugates was carried out by incubation of the His-tagged derivatives 2a/2b with zwitterion-QDs (ZW-QDs) or PEGylated-QDs in 0.1x PBS buffer (pH 7) for 30 min at room temperature. The glycan-to-QD ratios used were 0, 5, 10, 20, 30, 40 and 50.⁴²

Covalent Conjugation of Bi-antennary N-Glycans onto Aldehyde-QDs.

This strategy exploits the presence of an aminooxyacetyl (Aoa) group in SA glycans 3a/3b to enable oxime ligation with CHO-QDs. Peptidoglycans with Aoa (3a/3b) were prepared in solution by reaction of SA-glycan with Aoa-Ahx-Cys peptide in a similar fashion detailed above (Figure 2, bottom). Briefly, dispersions of QDs photoligated with 5% or 10% LA-PEG-CHO (3 μM) were mixed with 3a/3b in 50 mM ammonium acetate buffer at pH 4.5, at a 25-fold excess with respect to the surface aldehydes. The reaction mixture was stirred at room temperature for 4h, and the QDs were purified from excess glycan using Amicon Ultra Centrifugal device (50 MW). In order to support rapid oxime product formation under these reaction conditions, a control reaction was performed in the absence of QDs coupling 3b and LA-PEG-CHO under the same above conditions (see Scheme S1). All LA-PEG-CHO reagent was converted to LA-PEG-3b product after 2 hours, as monitored by reversed-phase (RP) HPLC (Figures S1–S2).

Characterization of the QD-Glycan Conjugates using Gel Electrophoresis.

The electrophoretic mobility was assessed for both sets of QD-glycan conjugates, prepared via either His-driven self-assembly or covalent oxime ligation. Measurements were run using 1.5% Agarose gels in Tris-acetate-EDTA (TAE) pH 8 buffer at 110 V for 10 min. QD samples were mixed with an equal volume of a mixture of 30% Glycerol in H₂O (loading buffer) before loading. Figure 3 shows that there are marked differences in mobility shifts measured for each QD-assembly. Dispersions of LA-PEG-QDs alone essentially exhibited no migration (no mobility shift), consistent with their neutral surfaces.⁴³ In comparison, a slight shift in the mobility band towards the anode was measured when increased loading of the glycan 2a was introduced, indicating that the added negative charges from the sialyl moieties contribute to the electrophoretic mobility. The rather modest shift measured for all glycan ratios reflects the competing effects of charge and overall size when more molecules are self-assembled on the QDs (Figure 3A, left image). When the neutral, esterified SA glycan 2b is self-assembled on the QDs a reduction on mobility shift was measured compared to QDs alone, which reflects the fact that glycan coupling increases the conjugate diameter but without introducing new charges (Figure 3A, right image).

Figure 3.

Gel mobility of His-tagged glycans 2a/2b self-assembled onto LA-PEG QDs (A) and LA-ZW QDs (B). Glycan:peptide loading ratios were 0, 5, 10, 20, 30, 40 and 50. The QDs or QD-conjugates were mixed with loading buffer (30% glycerol in H₂O), loaded onto a 1.5% Agarose gel in Tris-acetate-EDTA (TAE) pH 8 buffer and run for 10 min at 110 V.

The gel mobility pattern for conjugates self-assembled with zwitterionic-QDs are drastically different. A net negative charge due to the sulfate moieties in the coating drives the largest mobility shift (towards the anode) for unconjugated QDs.³⁷ When conjugated with either SA-glycans, 2a or 2b, a decreased electrophoretic mobility commensurate with the number of glycan moieties self-assembled per QD is observed (see Figure 3B). The decrease in mobility shift is consistent with the increased conjugate size and shielding/compensation of the zwitterion surface charges brought by the His-tag-driven conjugate self-assembly. The gel mobility patterns exhibit a few small differences, nonetheless. The decrease in mobility shift measured though systematic for both QD-2a and QD-2b conjugates, reaches saturation at much lower conjugate valences for conjugates self-assembled with glycan 2b, with zero mobility essentially measured at a ratio 20:1 (Figure 3B, right image). A non-zero mobility was still measured at 50:1 loading ration for QD-2a conjugates (Figure 3B, left image). This reflects the presence of additional negative charge in the 2a glycan but not in 2b (see Figure 2). Apart from the presence of two net negative charges per molecule in 2a glycans, each self-assembled glycopeptide also introduces a His₆-tag that contributes additional charge shielding/compensation and affects the electrophoretic mobility. To test this hypothesis, gel mobility data were collected for samples made of self-assembled QD-His₆-tag-unit without any glycan at different loading concentration. A slower electrophoretic mobility was measured for increasing His₆-tag loading ratio, similar to what was observed in Figure 3B (right image) (see Figure S3).

Electrophoretic mobility measurements collected from covalently formed conjugates using CHO-QDs reacted with glycans 3a and 3b show that only weak charge effects (from the carboxylic acid group in SA) are accounted for in the mobility shifts shown in Figure 4. Unconjugated CHO-QDs show zero mobility, and when coupled to unprotected glycan 3a they migrate through the gel as expected due to an increase in the net negative charges. When covalently conjugated with benzyl-protected glycan 3b, no migration was measured, since there is no added negative charge in the conjugate. Marginal difference in mobility shift was recorded for conjugates prepared using 5% of 10% CHO-QDs, a result attributed to rather small differences in the overall net charge and the fact that conjugates prepared with 10% CHO-QDs are larger which cancel the effects of extra charge COOH groups displayed on the conjugate surfaces at higher loading ratios.

Figure 4.

Differences in the gel mobility exhibited by covalently coupled QD-SA-glycan conjugates. Conjugates with 3a and 3b, prepared using 5% and 10% CHO-QDs, were loaded onto a 1.5% Agarose gel in Tris-acetate-EDTA (TAE) buffer at pH 8 and run for 10 min at 110V.

Lectin Recognition by QD-sialic acid Glycan Conjugates.

QD-glycan conjugates (QD-2a or QD-2b) were first self-assembled using LA-ZW-QDs (~0.05 μM) at a ratio of 10 SA-glycan per QD. Aliquots of the dispersions were incubated for 15 min with Tetramethylrhodamine-conjugated Sambucus nigra lectin (TRITC-SNA) at 0, 8 and 15 molar equivalents with respect to QDs. This lectin specifically recognizes the terminal sialic acid – galactose (α-2,6 linkage) disaccharide unit at the end of complex-type N-linked glycans. Similarly, dispersions of QDs alone (no glycan) were incubated for 15 min with TRITC-SNA also at 0, 8 and 15 molar equivalents and used for control experiments. The photoluminescence spectra were then recorded using an excitation signal at 350 nm.

The data shown in Figure 5 indicate that a sizable quenching of the QD emission was specifically measured for dispersions QD-2a conjugates mixed with TRITC-SNA, and the quenching was commensurate with the added TRITC-conjugate concentration. In comparison, weak quenching was observed when the TRITC-SNA was incubated with QD-2b conjugates, ascribed to collisional interactions between freely diffusing conjugates and dyes in solution (also referred as collisional quenching); this indicates that the benzyl-protected SA is not recognized by lectin. When unconjugated LA-ZW-QDs were incubated with TRITC-labeled SNA lectin negligible quenching was also was measured. These findings clearly indicate that the TRITC-labeled SNA specifically interacts with the glycan 2a in the conjugates, resulting in efficient, proximity-driven, fluorescence resonance energy transfer (FRET) interactions and quenching of the QD emission, as shown in Figure 5. In contrast, no recognition was exhibited by QD-2b conjugates or QDs alone. Similar results were collected for covalently linked QD-glycan conjugates, when they were incubated with TRITC-labeled lectin. Dispersions of 10% CHO-QDs (~0.02 μM) coupled to 3a or 3b along with control samples of unconjugated 10% CHO-QDs were incubated with different concentrations of TRITC-SNA for 15 min, then the fluorescence spectra were recorded using an excitation at 350 nm, as done above.

Figure 5.

Quenching of QD photoluminescence through SA-glycan recognition by TRITC labelled lectin in self-assembled SA-QDs. Comparison of lectin recognition of QDs conjugated with 2a (A), 2b (B) and a control with no glycan self-assembled on LA-ZW-QDs (C). Samples were prepared at a ratio of 10 SA-glycan per QD and incubated with TRITC conjugated SNA at 0, 8 and 15 equivalents to the QD concentration. Photoluminescence scans were recorded after 15 min from 500 to 600 nm with an excitation signal at 350 nm. The PL spectra were normalized with respect to the peak values of the QD-glycan conjugate dispersions (no SNA, i.e., reference).

The fluorescence data in Figure 6 show that a large quenching of the QD signal (exceeding ~ 50%) was measured only for the dispersions of QD-3a conjugates mixed with TRITC-labelled SNA. In comparison, quenching was rather modest for dispersions of QD-3b conjugates or unconjugated CHO-QDs (the PL was smaller than 30%). These losses are slightly larger than those measured for the control samples (QD-2b conjugates or ZW-QDs control) and may be due to the presence of non-specific interactions between the dye and CHO-QDs, since QD-3b conjugates contain a benzyl-protected SA and should not be specifically recognized by the lectin. We hypothesize that a low amount of residual benzaldehyde on the surface of the QD could be producing nonspecific binding to the lectin since the conjugates were employed without a capping step. The FRET quenching results shown in Figure 5 and Figure 6 provide a strong validation that only the QD-conjugates prepared with the SA-glycan, either covalently linked or His-tag self-assembled, can recognize lectin molecules in solutions.

Figure 6.

QD photoluminescence is quenched through SA-glycan recognition by TRITC labelled lectin in covalently formed QD-SA-glycan conjugates. Comparison of lectin recognition of 3a (A), 3b (B) and unconjugated 10% LA-PEG-CHO-QDs (C). Samples were incubated with TRITC conjugated SNA at 0, 8 and 15 equivalents to the QD concentration. Photoluminescence scans were recorded after 15 min using an excitation at 350 nm. The PL spectra were normalized with respect to the peak values of the QD-glycan conjugate dispersions (no SNA, i.e., reference).

Sialidase recognition and binding to SA-QDs.

Having proven that multiple copies of glycan molecules can be attached onto luminescent QDs and that only those glycans presenting exposed sialic acid groups can specifically sense lectin, we decided to test the ability of sialidase enzyme to specifically cleave the bound SA-glycans. Prior to experiments involving QDs, sialidase cleavage of glycan 2a was performed in solution. A solution of 2a (172 μM) in 50 mM sodium acetate buffer at pH 4.5 was incubated for 10 min at 37 °C with sialidase (11 nM). The mixture was analyzed by HPLC-MS and the hydrolyzed product of 2a was found (M+H 3911), which confirmed that SA cleavage of the glycan has taken place. The enzyme assay was applied to the self-assembled QD-glycan bioconjugates prepared using PEG-QDs and His-tagged SA-glycans (1.5–2 μM) at a SA-glycan:QD molar ratio of 50. The dispersions were incubated with sialidase (10–30 nM) for 10 min at 37 °C in 50 mM acetate buffer pH 4.5. Loading buffer (30% glycerol in H₂O) was added to the samples and the electrophoretic mobility was examined using 1.5% agarose gels (Figure 7).

Figure 7.

Gel image showing the effects of sialidase cleavage of unprotected SA from glycans self-assembled on QDs. 1.5 μM of QD-conjugates either mixed with (+) or without (−) sialidase at a glycan:QD ratio of 50. The experiments were performed in 50 mM sodium acetate buffer pH 4.5 and incubated at 37 °C for 10 min. Loading buffer (30% glycerol in H₂O) was added to the samples and the gel was run in Tris-acetate-EDTA (TAE) buffer at pH 8 for 10 min at 110V.

The gel image shown in Figure 7 clearly proves that treatment with sialidase affects the electrophoretic mobility on QD-2a conjugate only, where the mobility shift decreased to zero after incubation of the conjugates with sialidase. Here, unprotected SA-glycans on the QDs are recognized and cleaved by sialidase. In the absence of the enzyme the band shifts towards the anode, due to the effects of the sialic acid charge. The reduced mobility shift (to zero) measured in the presence of sialidase can be attributed to a loss of negative charge on the QD-conjugate surfaces. Conversely, the electrophoretic mobility of the conjugates formed with benzyl SA-glycan (2b) was baseline and, furthermore, not affected by sialidase treatment, suggesting a lack of recognition by the enzyme. Similar findings were collected for the electrophoretic mobility after sialidase treatment for covalently formed QD-SA-glycan conjugates using 5% and 10% CHO-QDs. The QD-glycan conjugates presenting unprotected SA (QD-3a) showed a reduced electrophoretic mobility in the presence of sialidase compared with samples of conjugates prepared with the benzyl-protected-glycan. No difference in the gel electrophoretic mobility was measured with and without added sialidase, with essentially no gel migration observed for any of the wells loaded with QD-3b conjugates (Figure 8). These observations provide additional proof that the SA-glycan presented on the QD surface is being recognized and cleaved by sialidase.

Figure 8.

Sialidase cleaves unprotected SA from covalently assembled QD-SA-glycan conjugates. Dispersions of QD-conjugates (1 μM), either with (+) or without (−) sialidase enzyme incubation, were loaded onto agarose gel and run for 10 min at 110 V. Image shows that incubation with the sialidase cleaves the SA off of the conjugates, therefore limiting their mobility shift in the gel.

CONCLUSIONS

In order to better integrate nanotechnology with biological systems, it is critical to engineer the surface chemistry of nanomaterials to mimic macromolecules found in those environments. Significant progress has been achieved in developing highly stable coatings of various nanocrystal that facilitate aqueous solubility and reduce undesired binding to biological molecules. Here we extend these studies by establishing a suite of robust methods for coating of QDs with homogenous complex-type N-linked glycans that have the potential to mimic macromolecules found in serum. Importantly, access to large quantities of the N-linked glycan through purification from chicken eggs facilitates the broad application of this approach as a biologically compatible nanoparticle coating. We have successfully shown that N-linked glycan can be conjugated onto QDs via either His-tag self-assembly or covalent linkage through oxime ligation. Our studies demonstrated that the glycan-conjugated QDs were specifically recognized by their corresponding lectin, where interaction was accounted for through FRET between the central QD and conjugated fluorophores. In addition, the glycans displayed on the QD-conjugate surfaces were available to undergo hydrolysis through sialidase recognition, showing they can mimic natural processes and serve as reporters. The glycans conjugated on the QDs show the same characteristics as in their native cellular environment, such as bulk negative charge, lectin recognition and sialidase recognition, showing the utility of these bioconjugates to be used as nanoparticles with a coating that mimics naturally occurring macromolecules. In addition to confirming the SA glycans can successfully be conjugated to QDs, these methods showed the particles produced by both methods to be equivalent in their properties, demonstrating the applicability of either functionalization approach. Overall, our study proves that SA-glycan conjugated onto fluorescent QDs can mimic free glycan behavior. Taking advantage of the intrinsic photophysical properties of QDs, these conjugates may serve as sensors for glycan involved biological processes such as lectin recognition, enzyme cleavage and surface charge modifications.

EXPERIMENTAL PROCEDURES

Reagents.

All solvents and chemicals were used without further purification, unless otherwise specified. The amino acids used in this study were purchased from Bachem. DMF (Dimethylformamide, HPLC grade), CH₃CN and CH₂Cl₂ from Fisher, TFA (trifluoroacetic acid), DIEA (N,N-diisopropylethylamine), N,N′-Diisopropylcarbodiimide (DIC) and Boc-aminooxyacetic acid (Boc-Aoa-OH) and Sialidase (from Clostridium perfringens) were purchased from Sigma-Aldrich. DI-water (18 MΩ) was prepared using a Millipore Milli-Q water purification system. HCTU (o-(1-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate) and MBHA resin (4-methylbenzhydrylamine, 0.65 mmol/g) were purchased from Peptides International. 1-Hydroxybenzotriazole (HOBT) was purchased from GL Biochem Shanghai. Rink amide (0.68 mmol/g) and Aminohexanoic acid (Ahx) were purchased from Novabiochem. TIS (Triisopropylsilane) was purchased from Oakwood Chemical. AmiconUltra Centrifugal filtration devices (0.5 mL, 50 MW cutoff) and TRITC-dye-conjugated Sambucus nigra (Elderberry Bark, SNA I) were respectively purchased from Millipore Inc. and EY Laboratories. Glycans 1a and 1b were obtained from chicken eggs as described by Kajihara, ^12,30,44 and provided as a generous gift by GlyTech Inc. (Kyoto, Japan).

Instrumentation.

Analytical reverse phase HPLC was carried out in a Varian ProStar Model 210 equipped with a Dynamax Absorbance Rainin Detector. Analytical injections were monitored by tracking the absorbance at 214 nm using a flow rate of 1 mL/min. Preparative HPLC was performed on a Waters Delta Prep 4000 equipped with a Waters UV detector model 486 and a Phenomenex Proteo C18 column (10 μm, 90 Å, 250 × 21.20 mm) at a flow rate of 15 mL/min, employing a gradient specified for each product. Preparative injections were monitored by tracking the absorbance at 220 nm. Products were characterized using electro spray ionization MS on a LC/MS API 2000 Plus triple quadrupole mass spectrometer (Sciex). The peptide masses were calculated from the experimental mass-to-charge (m/z) ratios from all observed protonation states using the onboard Analyst software package (Sciex). UV measurements were carried out using a Genesys 6 UV/vis spectrometer (Thermo Electron Corporation). The fluorescence spectra were recorded using an AVIV automated titrating differential/ratio spectrofluorometer (Model ATF 105). Agarose gels were imaged using a Bio-Rad ChemiDoc MP Imaging system, which runs Image Lab software.

Peptide Synthesis.

The cysteine-modified and His₆-terminated polyproline peptide, Cys-Trp-Ahx-Pro₉-Gly₂-His₆, was chain assembled manually using in situ neutralization cycles on 0.2 mmol MBHA resin (4-Methylbenzhydrylamine, 0.65 mmol/g), via Boc-solid-phase-peptide synthesis (Boc-SPPS), following procedures described in the literature.⁴⁵ For this, Boc-amino acid (1.5 mmol) was dissolved in 2.66 mL of 0.5 M HCTU in DMF (1.33 mmol) and then DIEA (2.66 mmol, 464 μL) was added. After 30 s, the solution was added to the resin and reacted for 20 min. Then, 2 sequential, 1 min TFA Boc-deprotection steps were applied and the process was repeated to grow the full sequence. To incorporate the proline sequence, the resin was first neutralized with 10% DIEA in DMF (2 washes of 1.5 min each) after each TFA deprotection prior to the introduction of the amonoacid. Briefly, Boc-proline (2.0 mmol, 430 mg) was dissolved in 4 mL of 0.5 M HCTU in DMF (2 mmol) and DIEA (2.0 mmol, 348 μL) were added, incubated for 30 s, and then added to the resin. Double couplings were performed for 1 h. Following chain assembly, the peptide was cleaved from the resin with HF and 10% of anisole for 1 h at 0 °C. After that, HF was evaporated and cold ether was added to precipitate the crude peptides. The resin was filtered, and the crude product was dissolved in 30% buffer B (0.05% TFA, 90% CH₃CN, 10% H₂O) in Buffer A (0.05% TFA in H₂O) and lyophilized. The crude peptide was purified by preparative HPLC using a Phenomenex Proteo column using a gradient of 5–25% Buffer B in A for 45 mins. Pure peptide fractions were collected at 18–19% Buffer B.

The Aoa-Ahx-Cys sequence was chain assembled by manual Fmoc-SPPS using 0.2 mmol Rink amide resin. A 2 sequential, 2 min 4-methyl-piperidine Fmoc-deprotection steps were applied before each aminoacid addition. Fmoc-amino acid (0.5 mmol) was dissolved in 1.25 mL of 0.4 M HCTU in DMF (0.5 mmol), mixed with DIEA (0.75 mmol, 130 μL), then added to the resin after 30 s; coupling times were 25 min. Boc-Aoa-OH (0.25 mmol, 47.8 mg) was dissolved in 0.5 mL of 0.5 M HOBT in DMF (0.25 mmol), DIC (0.25 mmol, 38.6 μL) was added and the solution was added to the resin after 30s. Coupling was complete within 30 min and checked with a Kaiser test. Following chain assembly, the peptide was cleaved from the resin using 90 min treatment with a cleavage cocktail that contained TFA (95%), TIS (2.5%) and H₂O (2.5%). The resin was then filtered and TFA evaporated using a gentle stream of N₂. The crude peptide was precipitated with cold ether, dissolved in 30% buffer B (0.05% TFA, 90% CH₃CN, 10% H₂O) in buffer A (0.05% TFA in H₂O) and lyophilized. The crude peptide was used in the next synthetic step without further purification.

Synthesis of Histag- and Aminooxy-modified Glycans 2a/2b and 3a/3b.

1a/b (1 Eq) was dissolved in 0.1M NH₄HCO₃ buffer at pH 8.0 (~2.5 mg/mL) and Cys-Trp-Ahx-Pro₉-Gly₂-His₆ or Aoa-Ahx-Cys (1.2 Eq) were added. The reaction was stirred at room temperature for 4h, and the product was purified on a C18 semi preparative column using a gradient 0–25% Buffer B in 45 min. The pure fractions were lyophilized, affording the products as white solids with yields ranging from 33 to 62%, 2a (1.55 mg, 51%), 2b (1.41 mg, 47%), 3a (2.86 mg, 33%), 3b (13.5 mg, 62%).

Growth of Core–Shell QDs and Photoligation with LA-PEG and LA-zwitterion ligands.

Growth of the Quantum Dots.

The CdSe-ZnS core-shell QDs used in this study were grown (in two steps) via reduction of organometallic precursors at high temperature. The growth medium consisted of coordinating solution containing a small fraction of n-hexylphosphonic acid in a mixture of n-trioctyl phosphine (TOP), n-trioctyl phosphine oxide (TOPO), and alkylamines. Core growth was then followed by overcoating with a few monolayers of ZnS. ^{46, 47, 48, 49, 50}

Phase Transfer of the QDs.

Phase transfer of the QDs was promoted by photoinduced ligand exchange of QDs with aldehyde- or zwitterion-modified LA ligands, as described in references. ^{37, 40, 51} We briefly describe the photoligation of CdSe-ZnS QDs with a mixture of LA-PEG₇₅₀-OCH₃ and LA-PEG₁₀₀₀-CHO (i.e., mixed surface ligand exchange). The molar fraction of LA-PEG₁₀₀₀-CHO used was varied between 0, 5% and 10% to allow the preparation of QDs with controllable numbers of surface reactive groups. First, hydrophobic QDs were precipitated using excess ethanol, followed by centrifugation, yielding a precipitate of the QD materials. The solvent was removed and the wet solid paste was re-dispersed in 500 μL of hexane. Separately, a mixture of LA-PEG₇₅₀-OCH₃ and LA-PEG₁₀₀₀-CHO with the desired molar ratio was dissolved in methanol (500 μL) mixed with a catalytic amount of tetramethylammonium hydroxide (TMAH, ~10 mM). The two contents were combined in one scintillation vial equipped with a stirring bar. The vial was sealed with a septum and the atmosphere was switched to nitrogen. This vial was placed inside a UV reactor (Luzchem Research Inc., Ottawa, Canada), and then irradiated using UV light (irradiation peak centered at 350 nm and a power = 4.5 mW/cm²) for 30 minutes with constant stirring. This promoted a complete phase transfer of the QDs from hexane to methanol, confirming ligand exchange. Following removal of the organic solvents under a mild vacuum, the QDs were purified from excess TOP, TOPO, alkyl amine and alkyl phosphonic acid ligands via precipitation and centrifugation. The QD paste was gently dried then dispersed in DI water. Two or three rounds of concentration/dilution using a centrifugal membrane filtration device (MW cutoff = 50 kDa) were applied to remove excess free hydrophilic and solubilized hydrophobic ligands.³⁷

Photoligation of the QDs with LA-zwitterion (LA-ZW) was carried out following the above protocol but with a few slight changes.^{37, 40} Hydrophobic QDs were first precipitated in a mixture of toluene/hexane/ethanol, then re-dispersed in 500 μL hexane. Separately, 500 μL of a fresh solution of LA-ZW in CH₃OH (containing ~40 mg of LA-ZW) was prepared in a scintillation; a catalytic amount of TMAH was added and a slight heating (at 50 – 60 °C) was applied to speed up the dissolution of the ligands. These two solutions were combined in a vial, the vial was sealed, and the atmosphere was switched to nitrogen. The reaction mixture was placed inside the UV reactor (as above) with vigorous stirring, producing a macroscopic precipitate of the QD materials.⁴⁰ The clear solution containing excess and solubilized ligands was removed and the precipitate was washed with methanol (2 times). After gentle drying under vacuum, DI-water with small amount of TMAH/NaOH was added, yielding a homogeneous dispersion of QDs. ³⁷ Finally 3–4 rounds of concentration/dilution using a membrane filtration device (Amicon Ultra, 50kD) were applied to provide a dispersion of LA-ZW–QDs in water.

Self-assembly of QD-Glycan Conjugates.

Conjugates with increasing valences were prepared by mixing (in 0.1xPBS buffer) compound 2a or 2b with the QD dispersion at glycan-to-QD molar ratios of 5, 10, 20, 30, 40, 50 and incubating for 30 min at room temperature. Gel electrophoresis experiments were carried out using 5 μL aliquots of QD or QD-conjugate dispersions (~1 μM).

Oxime ligation of Amino-Glycan onto CHO-QDs.

50 μL aliquots of CHO-QDs (~5–7 μM stock solution) were mixed with 25 equiv of 3a or 3b in 200 μL ammonium acetate buffer (50 mM, pH = 4.5) and incubated for 5 h. The QDs were purified using Amicon ultra centrifugal units (cut-off MW = 50 KDa) by performing 3 washes with water. The equivalent glycan amounts were calculated based on the percentage of reactive aldehydes used during phase transfer and assuming that each QD has an average of ~200 ligands.

Gel electrophoresis.

Agarose gel electrophoresis experiments were performed using a 1.5% gel with pH=8 TAE running buffer. QD dispersions (5 μL, ~1 μM) were mixed with an equal volume of the loading buffer (30% glycerol in H₂O). The samples were placed in the loaded into the gel and run for 10 min at 110 V. The gels were then imaged using a Bio-Rad ChemiDoc MP Imaging system.

Lectin conjugation.

Self-assembled QD-glycan (glycan:QD = 10:1) and covalently linked QD-glycan conjuagtes (starting with 10%-LA-PEG-CHO-QDs) were prepared in 250 μL dispersions at concentrations of 0.02–0.05 μM. A TRITC-conjugated SNA (lectin) stock solution was prepared at 1 mg/mL in a 0.01 M phosphate buffer (pH 7.2–7.4), with 0.15 M NaCl and 0.05% sodium azide as a preservative. Lectin was added to the QD dispersions to afford a final concentration of 0, 8 and 15 Equiv. with respect to QDs, and the samples were mixed at room temperature for 15 min. Fluorescence scans were recorded over the range 500 – 600 nm using excitation at 350 nm. The photoluminescence spectra were normalized with respect to the emission collected from a control containing 0 Equiv. of lectin.

Sialidase Enzyme Assay.

The enzyme assays were carried out in the absence and the presence of the QDs. 1) Assay without QDs. Glycan 2a was dissolved in 50 mM sodium acetate buffer at pH 4.5 at a concentration of 172 μM and incubated with sialidase (11 nM) for 10 min at 37 °C. The mixture was then analysed by HPLC-MS to identify the hydrolyzed product (M+H 3911). 2) Assay with QDs. For the sialidase cleavage, the assay was performed using self-dispersions of self-assembled QD-glycan conjugates, PEG-QD-2a or PEG-QD-2b, at a glycan:QD ratio of 50:1. The conjugates were dispersed in 5 μL sodium acetate buffer (50 mM, pH 4.5) at 1 μM. 1 μL of sialidase solution (from a 83.3 nM stock) was added to the samples; the final sialidase concentration was 13.8 nM. For negative controls 1 μL of buffer was added to the QD dispersion. The mixtures were incubated for 10 min at 37 °C and then 3 μL of loading buffer (30% glycerol in H₂O) was added. Samples were directly loaded on an agarose gel and ran in pH 8 Tris-acetate-EDTA (TAE) buffer for 10 min at 110V. For the sialidase cleavage assays performed using covalent QD-glycan conjugates, formed using 5% and 10% CHO-QDs, the following procedure was performed: To a 2 μL solution of 1 μM QD-glycan conjugates in 50 mM sodium acetate buffer pH 4.5, 1 μL of sialidase (from a stock solution of 83.3 nM) was added, yielding an enzyme concentration of 27 nM. The samples were incubated for 10 min at 37 °C, and then 3 μL of loading buffer (30% glycerol in water) was added. The samples were loaded onto the gels and ran in the same Tris-acetate-EDTA (TAE) buffer for 10 min at 110V, as above.