Lectin Interactions on Surface-Grafted Glycostructures: Influence of the

We performed quantitative analysis of the binding kinetics and affinity of carbohydrate–lectin binding and correlated them directly with the molecul...
1 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Lectin Interactions on Surface-Grafted Glycostructures: Influence of the Spatial Distribution of Carbohydrates on the Binding Kinetics and Rupture Forces Kai Yu,† A. Louise Creagh,§ Charles A. Haynes,*,§ and Jayachandran N. Kizhakkedathu*,†,‡ †

Centre for Blood Research and Department of Pathology & Laboratory Medicine and ‡Department of Chemistry, University of British Columbia, Vancouver, British Columbia V6T 1Z3, Canada § Michael Smith Laboratories and Department of Chemical and Biological Engineering, University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada S Supporting Information *

ABSTRACT: We performed quantitative analysis of the binding kinetics and affinity of carbohydrate−lectin binding and correlated them directly with the molecular and structural features of ligands presented at the nanoscale within the glycocalyx mimicking layers on surfaces. The surface plasmon resonance analysis identified that the mode of binding changed from multivalent to monovalent, which resulted in a near 1000-fold change in the equilibrium association constant, by varying the spatial distribution of carbohydrate ligands within the surface-grafted polymer layer. We identified, for the first time, that the manner in which the ligands presented on the surface has great influence on the binding at the first stage of bivalent chelating, not on the binding at the second stage. The rupture forces measured by atomic force microscope force spectroscopy also indicated that the mode of binding between lectin and ligands changed from multiple to single with variation in the ligand presentation. The dependence of lectin binding on the glycopolymer composition and grafting density is directly correlated with the nanoscale presentation of ligands on a surface, which is a determining factor in controlling the clustering and statistical effects contributing to the enhanced binding

P

of the surface-displayed saccharide residues, can thereby be explored to help elucidate the molecular mechanisms of protein−carbohydrate recognition. While improvements in the affinity of multivalent lectins for carbohydrate-carrying surfaces have been achieved through changes in the surface characteristics,13−19 the fundamental understanding of the reaction mechanism connecting binding affinity/avidity and surface structure remains lacking. This limitation could be addressed through the development of carbohydrate arrays where the binding interactions can be tuned from monovalent to multivalent using chemistry that allows for precise control of the presentation of ligands in the nanoscale. The implementation of such a technology would permit direct correlation of binding affinity/avidity with engineered changes in surface structure and thereby aid in the design and development of potent lectin inhibitors and biological effectors. In this study, we developed a novel system based on the grafted glycopolymer layers (Scheme 1), where polymer chains carrying carbohydrate residues were grafted on a planar surface where the grafting density, degree of polymerization, and carbohydrate density can be independently and tightly controlled at the nanoscale.

rotein−carbohydrate interactions mediate a variety of biological processes involving highly specific events such as cell signaling, cell adhesion, fertilization, and inflammatory response.1−3 The specific interactions typically involve the association of glycoproteins, glycolipids, or polysaccharides displayed on the glycocalyx of the cell with lectins or other carbohydratebinding proteins. The affinity between a single carbohydrate residue and a protein is usually modest.4,5 This limitation is overcome in nature through multivalent interactions, i.e., the engineering of multivalency in the lectin to permit multipoint attachment to carbohydrates displayed on the cell surface and the clustering of glycosidic groups on the cell surface in a manner that serves to strengthen each of the multivalent complexes formed. The overall strength of binding can then be enhanced by the chelating effect6−8 or proximity/statistical effect,9 which is directly correlated with the presentation of ligands on a surface.10−14 An improved understanding of the contribution of the molecular and structural features of ligands presented on a surface at the nanoscale to the mechanisms of carbohydrate recognition would therefore be of great use, in part by helping to resolve their roles in various biological processes. The presentation of carbohydrates on an array surface can provide a means to model (mimic) oligosaccharides found on a cell surface. Tuning the valency and density of binding epitopes, as well as the composition, spatial arrangement, and orientation © 2013 American Chemical Society

Received: April 18, 2013 Accepted: July 23, 2013 Published: July 23, 2013 7786

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Scheme 1. Representation of Binding of Con A to End-Grafted Glycopolymer Layers Containing Mannose (Green Circles) and Galactose (Red Titled Squares) Residuesa

a Due to its bivalency, Con A can bind either a single mannose unit in the layer, with the other binding site of Con A unoccupied (first binding event), or two mannose units to achieve binding-site saturation (second binding event). The association rate constant and dissociation rate constant for the monovalent binding reaction are ka1 and kd1. Those for the bivalent binding reaction are ka2 and kd2.

dissociation rate constant, a conventional regeneration protocol was used. The flow cell was set at 30 or 80 μL/min (the curves obtained are able to be overlaid, confirming the kinetic control of the experiments, Figure S1, Supporting Information). After each injection of Con A solution, the chip was regenerated by 300 μL of 5 M mannose buffer solution twice and 10 μL of 0.1 M phosphoric acid. The data were then evaluated using the BIAevaluation software, version 4.1. The bulk refractive index effect brought on by variation of the analyte concentration was fixed during the fitting. The dissociation rate constant for the interaction between Con A and the layer presenting 0.8% mannose was obtained by fitting the data to one-phase exponential decay using Origin 7.0 software. AFM Force Spectroscopy Measurements. AFM tips were first functioned with NHS-PEG6000-NHS22 and then immersed in phosphate buffer (25 mM, pH 7.4) containing Con A (2 mg/mL). The tips were removed after 2 h and washed briefly in pH 7.4 phosphate buffer and then with pH 4.8 phosphate buffer to remove the tetrameric form of Con A as well as any larger aggregates. AFM force measurements were performed on a commercially available multimode system with an atomic head of 130 × 130 μm2 scan range which used a NanoScope IIIa controller (Digital Instruments, Santa Barbara, CA). A commercially manufactured V-shaped silicon nitride (Si3N4) cantilever with gold on the back for laser beam reflection (Veeco, NP-S10) was used. The spring constant of the AFM cantilever was measured using the thermal equipartition theorem.23 Force measurements were performed in PBS buffer (pH 4.8, containing 1 mM CaCl2 and 1 mM MnCl2). The collection of force curves was performed after incubation for 10 min. On tip approach, the onset of the region of constant compliance was used to determine the zero distance, and on retraction, the region in which the force was unchanged was used to determine the zero force. The rate of tip− sample approach or retraction was set as 0.5 μm/s. We followed our published protocol for the calculations of the adhesive

A systematic investigation of the interaction of concanavalin A (Con A), a bivalent lectin, with glycopolymer layers is conducted and used to explore the influence of molecular and structural features of the ligands on the binding characteristics. Binding equilibrium and kinetic data collected by surface plasmon resonance (SPR) are reported and used to determine equilibrium and forward and reverse rate constants for each possible association/dissociation reaction that can occur between bivalent Con A and the grafted layer displaying the binding partner, mannose. The intermolecular unbinding force between Con A and carbohydrate units on the surface is also investigated by atomic force microscope (AFM) force spectroscopy. Results obtained show that, by tuning the presentation of ligands on the surface, the binding mode can be changed from monovalent to multivalent. In the bivalent Con A−mannose binding interaction, the distance between the ligands within the layer greatly influences the energetics of binding of the first sugar-binding epitope on bivalent Con A, but not the second epitope.

■

EXPERIMENTAL SECTION Glycopolymer layers containing different mannose contents were synthesized by adapting methods we reported earlier.20,21 Detailed synthetic procedures and characterization methods are presented in the Supporting Information. SPR Analysis. SPR measurements were performed on a BIAcore 3000 (BIAcore, Uppsala, Sweden) operated using the BIAcore control software. The flow rate of the analyte solution and phosphate buffered (PBS) (pH 4.8, containing 1 mM Ca2+ and 1 mM Mn2+) through the flow cells was set at 30 μL/min. A series of 750 μL Con A solutions were injected and allowed to flow through the channels continually without regeneration between consecutive injections. The equilibrium response was determined from duplicate measurements, plotted against the analyte concentration, and fitted with the model with a Hill slope by using GraphPad Prism 5 to determine the association constant values. For calculating the association rate constant and 7787

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Table 1. Characteristics of pAAEM (pM; M = Mannose) Layers and p(AAEM-co-AAEGal) (pMXGY; M = Mannose, G = Galactose) Layers sample pM-high pM-med pM-low pM1G5 pM1G10 pM1G20 pM1G50 pM1G120 pG

feed ratio of AAEM (mannose) to mannose AAEGal (galactose) composition (%) 100:0 100:0 100:0 1:5 1:10 1:20 1:50 1:120 0:100

100 100 100 16.7 9 4.8 2 0.8 0

dry thickness (nm)

grafting density, molecular weight of free distance between equilibrium σa (chain/nm2) polymer, Mn (Mw/Mn) chains, db (nm) thickness, Lec (nm)

19.6 4.6 1.3 20 16.3 22 23 17 24.4

0.10 0.02 0.006 0.09 0.08 0.09 0.09 0.09 0.09

123000 (1.3) 130000 (1.3) 130000 (1.3) 131000 (1.3) 122000 (1.4) 141000 (1.4) 148000 (1.3) 118000 (1.4) 158000 (1.3)

3.2 7.1 12.9 3.3 3.5 3.3 3.3 3.3 3.3

59 ± 3.6 15.5 ± 0.7 N/Dd 74 ± 5.7 55 ± 3.2 67 ± 4.7 71 ± 6.3 58 ± 5.1 70 ± 2.7

a The grafting density (σ) for glycopolymer layers was estimated by using the equation σ = (hρNA)/Mn, where Mn is the molecular weight of free polymer in the solution, NA is the Avogado number, h is the polymer layer thickness measured by an elliposometer, and ρ is the density of the glycopolymer (we assumed the density of the glycopolymer is equal to 1 g/cm3). bd = 1/σ1/2 nm. cThe equilibrium thickness (Le) was determined by atomic force microscopy as the critical distance from the substrate surface beyond which no repulsive force was detectable.24 dN/D = not determined.

Figure 1. (A) SPR sensorgram for the interaction of Con A with glycopolymer layers with different mannose contents: (a) 100%, (b) 16.7%, (c) 9%, (d) 2%, and (e) 0.8%. For the copolymer layers composed of 100% and 16.7% mannose, eight different concentrations (in the sequence of injection, 0.1, 0.2, 0.4, 1, 4, 10, 20, and 50 μM) of Con A were injected and flowed through the channel. For the copolymer layers composed of 9% and 2% mannose, eight different concentrations (1, 2, 4, 10, 20, 32, 64, and 108 μM) of Con A were injected and flowed through the channel. For the copolymer layer composed of 0.8% mannose, eight different concentrations (1, 4, 10, 20, 32, 64, 108, and 150 μM) of Con A were injected and flowed through the channel. (B) SPR sensorgram for the interaction of Con A with a galactose layer. The Con A concentrations used, in descending order of response units (RU), were 64, 32, 10, 1, and 0.1 μM. (C) Isotherms for Con A binding to glycopolymer layers with different mannose contents: (a) 100%, (b) 16.7%, (c) 9%, (d) 2%, and (e) 0.8%. (D) Req vs C obtained from the SPR data for Con A binding to a surface grafted with a mannose layer and the derivation of the fitted curve (red line) with the Hill model. The R2 value for the fitting is 0.97.

force.24 Average values ± SD from different force curves from five different spots on the substrate are reported.

utilizing aqueous surface-initiated atom transfer radical polymerization (SI-ATRP)20,21 to investigate the influence of various characteristics of the glyco structures on the carbohydrate binding of Con A. Galactose is a neutral sugar which has no specific binding interaction with Con A17 and was therefore used to dilute the ligand density within the grafted layer without affecting the properties of the layer structure, such as hydration, steric

■

RESULTS AND DISCUSSION Synthesis and Characterization of Glycopolymer Layers. End-grafted galactose-based copolymers with increasing amount of mannose were prepared on gold chips (Scheme 1) 7788

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Table 2. Equilibrium Association Constant (KA) and Hill Coefficient (h) Data for Binding of Con A to Various End-Grafted Glycopolymers sample

mannose composition (%)

grafting density, σ (chain/nm2)

association constant (KA) from eq 1 (M−1)

association constant (KA) from kinetic study (M−1)

h

pM-low pM-med pM-high pM1G5 pM1G10 pM1G20 pM1G50 pM1G120

100 100 100 16.7 9 4.8 2 0.8

0.006 0.02 0.10 0.09 0.08 0.09 0.09 0.09

(7.4 ± 1.3) × 105 (1.5 ± 0.6) × 106 (3.1 ± 0.5) × 106 (1.2 ± 0.5) × 105 (3.1 ± 0.2) × 104 (2.6 ± 0.3) × 104 (8.4 ± 0.3) × 103 (3.8 ± 0.6) × 103

N/D N/D 5.5 × 106 6.2 × 105 N/D 5.2 × 104 9.2 × 103 4.2 × 103

0.55 ± 0.14 0.71 ± 0.09 0.69 ± 0.12 0.65 ± 0.06 0.80 ± 0.05 0.83 ± 0.04 0.96 ± 0.06 0.99 ± 0.10

the classic sigmoidal shape, while those for layers with lower mannose content are second-order polynomials in form. The equilibrium isotherms for layers with different mannose contents (pMXGY) were also fitted with the Hill equation (Table 2)31,32

factors, etc. Eight glycopolymer layers were synthesized (Table 1) that cover a range of relevant structures through differences in the layer thickness (determined by ellipsometry), grafting density, and mannose content. Since the molecular weight and composition of the grafted chains on the surface are reported to be similar to those formed in solution along with surface-grafted chains,25−27 we used their characteristics for the estimation of the graft density of our glycopolymer system. Since the galactose and mannose monomers have similar polymerization kinetics,20 the molar ratio of mannose to galactose units within the copolymer is very close to the monomer feed ratio, which was supported by the 1H NMR analysis of the glycopolymer formed in solution along with surface-grafted polymer (Figure S2, Supporting Information). Utilizing this synthesis method, the composition of mannose residues (0.8 to 100 mol %) within the grafted layer was precisely controlled while holding the molecular weight of the grafted chains, grafting density, and hydrated layer thickness (equilibrium thickness) (Table 1) nearly constant. In addition, mannose homopolymer layers with different grafting densities from 0.10 to 0.006 chain/nm2 were grown from a diluted initiator layer28 to investigate whether steric factors influence the binding characteristics (Table 1). Our AFM surface morphology analysis showed that the grafted glycopolymer layers are quite uniform (Figure S3, Supporting Information) with very low surface roughness (0.3 nm), suggesting even distribution of polymer chains and sugar ligands on the surface. Effect of the Composition of the Glycopolymer Layer on Its Interaction with Lectin. Con A is a dimeric protein with two carbohydrate-binding sites at a distance 6.5 nm apart at pH less than 5.29 Given the defined separation of the binding sites on Con A, we initially utilize SPR to investigate whether the bivalent lectin interacts differently with the grafted layer with a change in the mannose content. SPR data were initially collected in a surface nonregeneration mode30 (Figure 1A). Avoiding regeneration of the unloaded sorbent surface between consecutive Con A injections, while having it done at the final stage, better preserved the properties and structure of the grafted glycopolymer layer. In addition, the results were comparable to those obtained using the more conventional surface-regeneration SPR protocol.17 A galactose layer was used as a negative control to exclude the bulk effect brought on by variation of the analyte concentration as there was no affinity observed between Con A and the galactose layer (Figure 1B). Adsorption isotherms for Con A binding to σ = 0.1 chain/nm2 pMXGY (M = mannose, G = galactose) layers (Table 1) displaying different mannose contents are reported in Figure 1C. The resulting data set reveals a strong dependence of the isotherm form on the mannose content within the layer. The log−log isotherm for the 100% mannose layer adopts (pM-high)

R eq = RUmax

Ch KD + C h

(1)

where KD (=1/KA) is the equilibrium dissociation constant, h is the Hill coefficient, Req is the SPR equilibrium response to a solution of Con A of concentration C, and RUmax is the maximal SPR response brought on by binding extrapolated to the saturation concentration of lectin. An R2 value close to 1 was obtained for fitting of eq 1 to SPR data for each layer (Figure S4, Supporting Information). h > 1 or h < 1 is usually interpreted as indicating positive or negative cooperativity, respectively, between the two mannose residues in the glycopolymer layer binding to a Con A molecule. h < 1 would therefore provide evidence that the first binding event (Scheme 1, one mannose binding to Con A) is not beneficial to the second binding event (binding of the other mannose site nearby to the same Con A molecule). That is, binding of Con A to a layer-displayed mannoside residue through both of its sites is less than the product of the per-site affinity constant for binding free mannose ((4 × 103)2 = 1.6 × 107 M−1). Global fitting of eq 1 to the SPR data for Con A binding to the 100% poly(mannose) layer yields a KA of (3.1 ± 0.5) × 106 M−1 (Figure 1D; fitting curves for layers with different mannose contents are given in the Supporting Information, Figure S4). This is comparable to the value reported previously for the multivalent interaction of Con A with a surface-displayed mannose polymer as measured by SPR, quartz crystal microbalance, or other methods.17,33,34 Importantly, both the KA, which is less than 1.6 × 107 M−1, and the regressed Hill coefficient, which is less than unity, indicate that there is a small energetic penalty associated with multivalent binding of mannose residues to Con A through both of its binding sites. As a result, only partial bivalent binding avidity from negative cooperativity is observed. A decrease in the mannose content within the layer (σ ≈ 0.1 chain/nm2) from 100% to 0.8% (pM1G120) reduced KA from (3.1 ± 0.5) × 106 to (3.8 ± 0.6) × 103 M−1 and h from 0.69 to 1 (Table 2). In accordance with the larger range of grafting and ligand densities explored here, this change in association constant is much more significant than that which has been observed previously using carbohydrate arrays; those studies reported 2−3-fold changes in affinity as a function of the surface density of ligand residues.12,13,16−18 At 0.8% mannose content (pM1G120), the interaction of the polymer layer with Con A is monovalent with an association constant equal to KA for Con A binding to methyl α-D-mannopyranoside in free solution.4,5 Also, at this 7789

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Figure 2. Kinetics analyses of Con A binding to glycopolymer layers. Shown are experimental data (black curves) and the derivation of the fitted sensorgram (red lines) by bivalent model fitting of Con A binding to glycopolymer layers with different mannose contents: 100% (A), 16.7% (B), 4.7% (C), and 2% (D). (E) SPR sensorgram of binding Con A to a glycopolymer layer with a mannose content of 0.8%. (F) Fitting of the dissociation curves (scattered points) of a 0.8% mannose polymer layer and Con A with the 1:1 Langmuir binding model (red lines).

Table 3. Equilibrium and Reaction Rate Constants for Binding of Con A to Glycopolymer Layers (Grafting Density 0.10 ± 0.02 chain/nm2) Having a Mannose Content Ranging from 2% to 100% sample pM-high pM1G5 pM1G20 pM1G50

mannose composition (%)

ka1 (M−1 s−1)

kd1 (s−1)

100 16.7 4.8 2

(7.6 ± 1.9) × 10 (1.4 ± 0.8) × 103 (3.3 ± 1) × 102 (2.1 ± 1) × 102 3

kd2 (s−1) −3

(8.4 ± 1.3) × 10 (7.5 ± 1) × 10−2 (1.1 ± 0.3) × 10−1 (1.6 ± 0.7) × 10−1

−4

(5.3 ± 0.7) × 10 (6.1 ± 1) × 10−4 (8.5 ± 1.2) × 10−4 (2.8 ± 0.9) × 10−3

KA1 (M−1)

KA2 (M−1)

χ2/Rmax

9 × 10 1.9 × 104 3 × 103 1.3 × 103

6.1 32.6 17.3 7.1

0.01−0.1 0.03−0.1 0.004−0.1 0.005−0.1

5

9 × 105 M−1. These are the hallmarks of the avidity effect,35 present in this system through the ability of the two binding sites on Con A to independently interact with immobilized mannose residues. It is also noticed that the bivalent model of binding contributes less to the overall interaction (greater KA1/KA2 ratio), which is different from the study by Tassa et al.36 and Jayaraman et al.37,38 evaluating the multivalent interactions between α-D-mannopyranoside glycolipid micelles and lectin. It seems that the very high mannose content, which was due to the dense packing of polymer chains on the surface, generated an unfavorable bivalent binding interaction between the sugar ligands and the lectins. The saturation of the ligands by lectins presented on the surface in the first stage of binding significantly lowered the surface density of available ligands to the second stage of bivalent binding. Also, the lectin−mannose complex formed in the first stage (monovalent binding) may create steric hindrance for the second stage of bivalent binding. Individually, each binding interaction may strengthen due to two well-known effects. The first is through the creation of grafts that serve to locally cluster mannose units in a manner that improves binding to each site, an effect that several groups have shown can be significant in lectin−sugar interactions.4,39 The observed increase in KA1 with mannose content supports this finding, showing that ligand-clustering effects are indeed significant. The second effect is related to changes in the local concentration. In particular, when the surface density of available

composition, the binding event occurred in a monovalent way. An h value of 1 was obtained accordingly. To gather further insight into the nature of Con A and glycopolymer layer interactions, we performed detailed SPRbased kinetic analyses. Figure 2 reports kinetic curves for the interaction of Con A with layers having different mannose contents (Figure 2A−D,F). Fitting the data to a bivalent analyte model gave reasonably good confidence intervals, with χ2/Rmax less than 0.1 for all layers with a mannose content greater than 2% (pM1G50). In contrast, dissociation curves (Figure 2F) for the 0.8% layer (pM1G120) are poorly fit by this model and are instead well described by a classic Langmuir monovalent binding model, for which R2 values are between 0.94 and 0.97. Together, these results further support our conclusion of a transition from monovalent to bivalent binding of Con A with increasing mannose content in the layers. Evidence of contributions from lectin clusters is provided in Table 3, which reports association and dissociation rate constants (ka1 and kd1) for the monovalent Con A binding reaction, as well as dissociation rate constants (kd2) and equilibrium association constants (KA2) for the bivalent binding reaction. Increasing the mannose content from 2% to 100% results in an increase in ka1 from 2.1 × 102 to 7.6 × 103 M−1 s−1 and a significant decrease in kd1 from 0.16 to 0.0084 s−1. The equilibrium association constant KA1 for the monovalent complex (KA1 = ka1/kd1) therefore increases by nearly 3 orders of magnitude from 1.3 × 103 to 7790

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Figure 3. (A) Representative approach (black line) and retraction (red line) force curves for the interaction of a mannose polymer layer (σ = 0.1 chain/ nm2 and height 59 nm) with a Con A-modified tip. (B) Representative approach (black line) and retraction (red line) force curves for the interaction of a galactose polymer layer with a Con A-modified tip (inset, histogram of the unbinding force during retraction of the Con A-modified tip from the galactose polymer layer). (C) Probability distribution histograms of the rupture distance during retraction of the Con A-modified tip from the mannose polymer layer. (D) Probability distribution histograms of the maximum unbinding force during retraction of the Con A-modified tip from the mannose polymer layer.

and divalent interactions of Con A with mannose at a similar loading rate are 46 and 68 pN, respectively.41 The multiple peaks observed in the unbinding force histograms suggest that the retraction of Con A from a pure mannose layer (σ = 0.1 chain/nm2 and height 59 nm) involves the dissociation of roughly 2-3 individual divalent linkages formed at different rupture distances. This might be due to the fact that Con A or mannose may not be present on the tip or surface as single molecules in the present case. Finally, force measurements were performed in the presence of 100 mM mannose in solution to verify that the measured unbinding force originated from the specific interaction between Con A and mannose within the grafted layer. As expected, a much lower unbinding force (16.7 ± 12.3 pN) and shorter rupture distance (50.7 ± 9.2 nm) (Figure S5, Supporting Information) were recorded. The mean unbinding force was found to decrease from 171 ± 46 to 51 ± 12 pN when the mannose content changed from 100% to 9% (Figure S6, Supporting Information). The mean unbinding force (Figure S6C) is seen to fall toward that limit with decreasing mannose content in the layer, confirming that avidity effects are lost and monovalent binding dominates at sufficiently low mannose densities. Multiple rupture events were recorded in ca. 25% of retractions from the 9% mannose layer, indicating weak avidity in this system. Effect of the Graft Density of Mannose Polymer Layers. We argued above that the increase in ka1 with increasing mannose content on the surface suggests that the presentation geometry of the mannose residues, including clustering effects, influences the

(possibly clustered) ligand is high, the dissociation of a single binding site on Con A is short-lived, such that the occupancy of that site is re-established before the time required for the protein to diffuse away (t = Dx2, where x is the distance away from the surface and D is the protein diffusivity). Avidity is therefore reflected in the values of kd1 and kd2, the latter of which also decreases with increasing mannose content in the σ = 0.10 chain/nm2 layer. AFM Force Spectroscopy Studies of Lectin Binding to Glycopolymer Layers: Variation of the Ligand Composition. To further quantify the interaction between lectins and glycopolymer layers, force spectroscopy measurements on the influence of the layer composition on the binding characteristics were made. Glycopolymer layers having compositions and structures similar to those employed in the SPR analyses were utilized (Table 1), and AFM tips were functionalized with Con A using a PEG spacer.22 The retraction force curves differ, with that for the mannose layer (Figure 3A) showing features40 indicative of a specific interaction between Con A and mannose residues on the surface. As shown in the histogram reported in Figure 3C,D, the binding ruptured at 98.8 ± 22.3 nm with forces from 70 to 250 pN recorded at a loading rate of ∼30 ± 7 nN s−1 (0.06 nN/nm × 500 nm/s), with the mean value equal to 171 ± 46 pN. In addition, the identification of multiple adhesion peaks (63% of the total event) in our system indicates that multiple interactions between Con A and ligand units distributed within the layer are sequentially broken during tip retraction.41 Previous reports showed that the mean rupture forces for individual monovalent 7791

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

Figure 4. (A) SPR sensorgram for the interaction of Con A with a mannose layer having grafting densities of 0.1 (a), 0.02 (b), and 0.006 (c) chain/nm2. Eight different concentrations of Con A were injected and flowed through the channel. Concentration of Con A in the sequence of injection: 0.1, 0.2, 0.4, 1, 4, 10, 20, and 50 μM. (B) Isotherms for Con A binding to glycopolymer layers with different grafting densities: 0.1 (a), 0.02 (b), and 0.006 (c) chain/nm2. (C) Probability distribution histograms of the maximum unbinding force during retraction of the Con A-modified tip from the mannose polymer layer with a grafting density of 0.02 chain/nm2. (D) Dependence of the association constant (■) and unbinding force (▲) on the grafting density of the mannose polymer layer.

the distance between two binding sites (6.5 nm) of Con A. This shows that strong avidity can be realized through ligand clustering on an individual chain and bivalent chelation of Con A by multiple ligands on an individual chain. The dependence of the spacing between proximal ligands on the association constant also suggests that there may be cross-linking of lectin within the glycopolymer layer system, which was notably demonstrated by Bertozzi and co-workers using a density-variant glycan microarray.12 AFM studies (Figure 4C,D; Figure S8, Supporting Information) revealed that the rupture forces are independent of the grafting density, which also supports the arguments above.

binding affinity of the lectin. The improvement in the stability of the complex formed at each glycoside recognition site on Con A and increase in the potential to form a stable bivalent complex either with two ligands (clusters) on a single grafted chain or with ligands on different chains (cross-linking by lectins) are recognized as factors. To better define the latter multichain binding contribution, the grafting density of the 100% mannose layer was tuned to vary the interchain distance (pM-low, pM-med, and pM-high samples in Table 1). The SPR response (Figure 4A) increases with increasing grafting density. Calculation of the corresponding adsorption isotherms (Figure 4B) shows that the overall affinity of Con A binding also increases with increasing grafting density (Table 2). From this, we infer that bivalent complexes formed with Con A through multichain bridging, as well as an increased probability for dissociated-site reoccupancy by a free ligand available on a different chain, contribute to overall binding avidity. The overall equilibrium association constants for Con A binding to this series of mannose layers, obtained by fitting the SPR data to eq 1 (Figure S7, Supporting Information), show only a 4-fold change from (3.1 ± 0.5) × 106 to (7.4 ± 1.3) × 105 M−1, similar in extent to the others.18 Thus, the overall affinity remains high (near 106 M−1), indeed about 3 orders of magnitude higher than the binding constant for an individual mannose−Con A complex, even at grafting conditions where intermolecular interactions between chains are absent, or at least very limited, as the distance between the chains is larger (12.9 nm for the layer having a density of 0.006 chain/nm2) than

■

CONCLUSIONS Glycopolymer layers with different spatial distributions of mannose residues were prepared on gold-coated glass chips by the careful control of the feed ratio of glycomonomers and grafting density of the polymer chains. It was found that the composition of ligands within the layer is a greater contributor toward binding characteristics than the grafting density. While glycopolymer layers presenting 100% mannose residues exhibited multivalent interaction with Con A, equivalent grafting density layers containing 0.8% mannose residues showed only monovalent interaction. The kinetic study revealed that the mannose content has a greater influence on monovalent binding than bivalent binding. This is explained in part by clustering and statistical effects, which lead to differences in the binding behavior. The weaker dependence of the association constants 7792

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793

Analytical Chemistry

Article

(16) Liang, P. H.; Wang, S. K.; Wong, C. H. J. Am. Chem. Soc. 2007, 129, 11177−11184. (17) Smith, E. A.; Thomas, W. D.; Kiessling, L. L.; Corn, R. M. J. Am. Chem. Soc. 2003, 125, 6140−6148. (18) Meng, X. L.; Fang, Y.; Wan, L. S.; Huang, X. J.; Xu, Z. K. Langmuir 2012, 28, 13616−13623. (19) Zhang, Y. L.; Li, Q. A.; Rodriguez, L. G.; Gildersleeve, J. C. J. Am. Chem. Soc. 2010, 132, 9653−9662. (20) Yu, K.; Kizhakkedathu, J. N. Biomacromolecules 2010, 11, 3073− 3085. (21) Yu, K.; Lai, B.; Kizhakkedathu, J. N. Adv. Healthcare Mater. 2012, 1, 199−213. (22) Ebner, A.; Wildling, L.; Kamruzzahan, A. S. M.; Rankl, C.; Wruss, J.; Hahn, C. D.; Holzl, M.; Zhu, R.; Kienberger, F.; Blaas, D.; Hinterdorfer, P.; Gruber, H. J. Bioconjugate Chem. 2007, 18, 1176−1184. (23) Wang, M.; Cao, Y.; Li, H. B. Polymer 2006, 47, 2548−2554. (24) Zou, Y. Q.; Rossi, N. A. A.; Kizhakkedathu, J. N.; Brooks, D. E. Macromolecules 2009, 42, 4817−4828. (25) Ejaz, M.; Ohno, K.; Tsujii, Y.; Fukuda, T. Macromolecules 2000, 33, 2870−2874. (26) Brittain, W. J.; Boyes, S. G.; Granville, A. M.; Baum, M.; Mirous, B. K.; Akgun, B.; Zhao, B.; Blickle, C.; Foster, M. D. Adv. Polym. Sci. 2006, 198, 125−147. (27) Wu, T.; Efimenko, K.; Vlcek, P.; Subr, V.; Genzer, J. Macromolecules 2003, 36, 2448−2453. (28) Jones, D. M.; Brown, A. A.; Huck, W. T. S. Langmuir 2002, 18, 1265−1269. (29) Derewenda, Z.; Yariv, J.; Helliwell, J. R.; Kalb, A. J.; Dodson, E. J.; Papiz, M. Z.; Wan, T.; Campbell, J. EMBO J. 1989, 8, 2189−2193. (30) Tang, Y. J.; Mernaugh, R.; Zeng, X. Q. Anal. Chem. 2006, 78, 1841−1848. (31) Dam, T. K.; Gerken, T. A.; Brewer, C. F. Biochemistry 2009, 48, 3822−3827. (32) Katrlik, J.; Skrabana, R.; Mislovicova, D.; Gemeiner, P. Colloids Surf., A 2011, 382, 198−202. (33) Kitano, H.; Takahashi, Y.; Mizukami, K.; Matsuura, K. Colloids Surf., B 2009, 70, 91−97. (34) Yu, L.; Huang, M.; Wang, P. G.; Zeng, X. Anal. Chem. 2007, 79, 8979−8986. (35) Monsigny, M.; Mayer, R.; Roche, A. C. Carbohydr Lett. 2000, 4, 35−52. (36) Tassa, C.; Duffner, J. L.; Lewis, T. A.; Weissleder, R.; Schreiber, S. L.; Koehler, A. N.; Shaw, S. Y. Bioconjugate Chem. 2010, 21, 14−19. (37) Murthy, B. N.; Voelcker, N. H.; Jayaraman, N. Glycobiology 2006, 16, 822−832. (38) Jayaraman, N. Chem. Soc. Rev. 2009, 38, 3463−3483. (39) Corbell, J. B.; Lundquist, J. J.; Toone, E. J. Tetrahedron 2000, 11, 95−111. (40) Ratto, T. V.; Langry, K. C.; Rudd, R. E.; Balhorn, R. L.; Allen, M. J.; McElfresh, M. W. Biophys. J. 2004, 86, 2430−2437. (41) Ratto, T. V.; Rudd, R. E.; Langry, K. C.; Balhorn, R. L.; McElfresh, M. W. Langmuir 2006, 22, 1749−1757.

and unbinding force on the grafting density showed that strong avidity can be realized by intrachain glycoside clustering and binding. By covering a large range of grafted carbohydrate structures, our results therefore provide new insights into the presentation of ligands at the nanoscale on surfaces and their contribution to protein binding characteristics.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (J.N.K.); [email protected] (C.A.H.). Phone: 604-822-7085 (J.N.K.); 604-822-5136 (C.A.H.). Author Contributions

K.Y. performed all the experiments. All authors contributed to the analysis and design of the experiments. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We acknowledge the funding provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) and Canadian Institutes of Health Research (CIHR). The Laboratory of Molecular Biophysics (LMB) Macromolecular and Biothermodynamics Hubs at the University of British Columbia (UBC) were funded by the Canada Foundation for Innovation and Michael Smith Foundation of Health Research. K.Y. is a recipient of a CIHR/Canadian Blood Services (CBS) postdoctoral fellowship in transfusion science. J.N.K. is a recipient of a CIHR/CBS new investigator award in transfusion science and Michael Smith Foundation of Health Research Career Investigator Award.

■

REFERENCES

(1) Sharon, N.; Lis, H. Glycobiology 2004, 14, 53R−62R. (2) Bertozzi, C. R.; Kiessling, L. L. Science 2001, 291, 2357−2364. (3) Dube, D. H.; Bertozzi, C. R. Nat. Rev. Drug Discovery 2005, 4, 477− 488. (4) Mandal, D. K.; Kishore, N.; Brewer, C. F. Biochemistry 1994, 33, 1149−1156. (5) Schwarz, F. P.; Puri, K. D.; Bhat, R. G.; Surolia, A. J. Biol. Chem. 1993, 268, 7668−7677. (6) Mammen, M.; Choi, S. K.; Whitesides, G. M. Angew. Chem., Int. Ed. 1998, 37, 2755−2794. (7) Lundquist, J. J.; Toone, E. J. Chem. Rev. 2002, 102, 555−578. (8) Dam, T. K.; Brewer, C. F. Biochemistry 2008, 47, 8470−8476. (9) Mangold, S. L.; Cloninger, M. J. Org. Biomol. Chem. 2006, 4, 2458− 2465. (10) Wolfenden, M. L.; Cloninger, M. J. Bioconjugate Chem. 2006, 17, 958−966. (11) Ting, S. R. S.; Chen, G.; Stenzel, M. H. Polym. Chem. 2010, 1, 1392−1412. (12) Godula, K.; Bertozzi, C. R. J. Am. Chem. Soc. 2012, 134, 15732− 15742. (13) Wilczewski, M.; Van der Heyden, A.; Renaudet, O.; Dumy, P.; Coche-Guerente, L.; Labbe, P. Org. Biomol. Chem. 2008, 6, 1114−1122. (14) Wang, X.; Ramstrom, O.; Yan, M. D. Anal. Chem. 2010, 82, 9082− 9089. (15) Wang, X.; Ramstrom, O.; Yan, M. D. Adv. Mater. 2010, 22, 1946− 1953. 7793

dx.doi.org/10.1021/ac401306b | Anal. Chem. 2013, 85, 7786−7793