Fabrication of Carbohydrate Chips Based on Polydopamine for Real-Time Determination of Carbohydrate–Lectin Interactions by QCM Biosensor

A novel approach for preparing carbohydrate chips based on polydopamine (PDA) surface to study carbohydrate–lectin interactions by quartz crystal microbalance (QCM) biosensor instrument has been developed. The amino-carbohydrates were immobilized on PDA-coated quartz crystals via Schiff base reaction and/or Michael addition reaction. The resulting carbohydrate-chips were applied to QCM biosensor instrument with flow-through system for real-time detection of lectin–carbohydrate interactions. A series of plant lectins, including wheat germ agglutinin (WGA), concanavalin A (Con A), Ulex europaeus agglutinin I (UEA-I), soybean agglutinin (SBA), and peanut agglutinin (PNA), were evaluated for the binding to different kinds of carbohydrate chips. Clearly, the results show that the predicted lectin selectively binds to the carbohydrates, which demonstrates the applicability of the approach. Furthermore, the kinetics of the interactions between Con A and mannose, WGA and N-Acetylglucosamine were studied, respectively. This study provides an efficient approach to preparing carbohydrate chips based on PDA for the lectin–carbohydrate interactions study.


Introduction
Carbohydrates located on the surfaces of cells play a crucial role in cell interactions, cell communication, cell proliferation, as well as cell death, where the interactions between carbohydrates and proteins are identified to be essential important in biological processes [1][2][3]. To date, several methods have been used to analyze the interactions between carbohydrates and proteins, such as enzyme-linked lectin assays (ELLAs), mass-spectrometric techniques, nuclear magnetic resonance, X-ray crystallography, microarray technologies, and biosensors [4][5][6][7][8][9][10][11][12]. Among them, biosensors are well known to its label-free and real-time detection, which have been wildly used to study biomolecular interactions. The quartz crystal microbalance (QCM) biosensor, which is based on the effect of piezoelectric, has turned out to be an efficient analytical tool for the assessment of carbohydrate-protein interactions with protein chips, carbohydrate chips, and cell chips [13][14][15][16][17][18][19]. Recently, as a useful analytical tool, QCM carbohydrate chip technology has attracted wide attentions for the carbohydrate-protein interaction study [20][21][22].
Bruker 500 MHz Spectrometer (BrukerDaltonics Inc., Billerica, MA, USA) recorded 1 H NMR spectra, using residual signals from D2O ( 1 H: δ 4.79 ppm) or CDCl3 ( 1 H: δ 7.26 ppm) as internal standards. Chemical composition of the sensor surface was determined by an X-ray photoelectron spectroscopy (XPS) instrument (Thermo Fisher Scientific Inc., Waltham, MA, USA). The Attana Cell A200 QCM (Attana AB, Stockholm, Sweden) under a continuous flow mode was used to carry out the interaction measurements ( Figure 1) [15]. Attana gold sensor chips (Attana, Stockholm, Sweden), with 8 mm in diameter and a 4.5-mm diameter gold electrode on each side, were used for further modification to produce the carbohydrate sensor chips.

Gold Surface Functionalization by PDA Coating
The gold chip surface was cleaned for 30 min in a piranha solution of 30% H2O2 and H2SO4 (1:3 v/v) at 80 °C, and thoroughly rinsed with ultrapure water and dried under nitrogen stream. Then, 50 μL dopamine solution (2 mg/mL in 10 mM Tris buffer, pH 8.5) was added to the gold surface at room temperature for 30 min. After incubation in dopamine solution, the chip surface was operated by rinsing with ultrapure water and drying under nitrogen [35,36].

Immobilization of Amino-Carbohydrate on the PDA-Coated Surface
50 μL carbohydrate solution (100 mg/mL in PBS, pH 7.4) was applied to the chip surface coated by PDA, then incubated for 4 h at room temperature. After incubation with the carbohydrate solution, the sensor chip surface was further rinsed with PBS in order to remove the carbohydrate which was un-immobilized on the surface.

XPS Measurement
In order to evaluate the chemical composition of the sensor chip surface, XPS spectra of these samples were obtained after drying the samples overnight.

Gold Surface Functionalization by PDA Coating
The gold chip surface was cleaned for 30 min in a piranha solution of 30% H 2 O 2 and H 2 SO 4 (1:3 v/v) at 80 • C, and thoroughly rinsed with ultrapure water and dried under nitrogen stream. Then, 50 µL dopamine solution (2 mg/mL in 10 mM Tris buffer, pH 8.5) was added to the gold surface at room temperature for 30 min. After incubation in dopamine solution, the chip surface was operated by rinsing with ultrapure water and drying under nitrogen [35,36].

Immobilization of Amino-Carbohydrate on the PDA-Coated Surface
50 µL carbohydrate solution (100 mg/mL in PBS, pH 7.4) was applied to the chip surface coated by PDA, then incubated for 4 h at room temperature. After incubation with the carbohydrate solution, the sensor chip surface was further rinsed with PBS in order to remove the carbohydrate which was un-immobilized on the surface.

XPS Measurement
In order to evaluate the chemical composition of the sensor chip surface, XPS spectra of these samples were obtained after drying the samples overnight. was achieved (frequency shift < 0.2 Hz/min), the flowrate was adjusted to 25 µL/min. To block the PDA areas which were not react with carbohydrate, BSA (50 µg/mL) was subsequently injected over the PDA-coated carbohydrate chip surface until the surface was saturated. After the injection of the BSA, the measurements of interactions can be recorded. The lectin diluted in running buffer at 50 µg/mL was injected to the chip surface. The association time is 84 s, and the dissociation time is 300 s. After each association and dissociation cycle, the chip surface regeneration was performed with injection of 10 mM Glycine (pH 1.5), and the bound analyte was removed. The recorded data was analyzed using the evaluation software of the Attana Cell A200. The kinetic evaluations were calculated by fitting the curves with a theoretical 1:1 binding model. The curve fitting data were used to interpret the kinetic constants, including the association rate constant (k on ), the dissociation rate constant (k off ), and the equilibrium dissociation constant (K D ). ClampXP, developed by Tom Morton and David Myszka, is a global analysis of the association and dissociation phases of the interaction by fitting a theoretical 1:1 interaction model compensated for mass-transport limitation [17] and used for the biosensor kinetic data analysis.

Reuse of the Sensor Chip
The PDA coating was removed after the measurement with QCM according to Doriane Del Frari's method [40]. Briefly, the sensor chip was immersed in NaClO solution (1 g/L) and then washed with ultrapure water. The chip can be reused for fabrication of PDA coated chips by following the protocol in Section 2.2.

Results and Discussion
The fabrication of the carbohydrate chips and their interaction with lectins were illustrated in Figure 2. The carbohydrate chips were readily prepared using a two-step process. Firstly, a QCM gold sensor surface was coated with PDA in alkaline environment through the dopamine self-polymerization. Following PDA coating, the aminated carbohydrates were immobilized on the PDA coated chip surface via Schiff base and/or Michael addition reaction to yield the carbohydrate chips, respectively. The interactions between the lectins and the carbohydrates were evaluated by a QCM biosensor. In order to study the interaction of the immobilized carbohydrate on the sensor surface with its interacting lectin, the flow rate was set to 25 µL/min. The interacting lectin was diluted in running buffer and injected over the surface. The resonant frequency of the quartz crystal and the frequency shift (∆f) coupled to the association or dissociation were recorded with the Attester software in real time. The data were analyzed using Attester Evaluation software, where the biosensor kinetic data analysis was performed by using software ClampXP v3.50 [17].

XPS Analysis of the Sensor Chip Surface
XPS was used to analyze the chemical composition of these chip surfaces and the results were shown in Figure 3. From the analysis of XPS, compared to the Attana gold sensor chips, PDA-coated gold sensor chip was observed the disappearance of the photoelectron peaks of Au (56.7 eV for Au5p3, 84.2 eV for Au4f7, 87.9 eV for Au4f5, 111.8 eV for Au5s, 334.3 eV for Au4d5,

XPS Analysis of the Sensor Chip Surface
XPS was used to analyze the chemical composition of these chip surfaces and the results were shown in Figure 3. From the analysis of XPS, compared to the Attana gold sensor chips, PDA-coated gold sensor chip was observed the disappearance of the photoelectron peaks of Au (56.7 eV for Au5p3, 84.2 eV for Au4f7, 87.9 eV for Au4f5, 111.8 eV for Au5s, 334.3 eV for Au4d5, 353.1 eV for Au4d3, 545.7 eV for Au4p3, 641.1 eV for Au4p1, and 762.1 eV for Au4s, Figure 3 A), and at the same time, the emergence of the peaks of carbon (C1s, 283.1 eV), oxygen (O1s, 531.5 eV), and nitrogen (N1s, 398.2 eV) (Figure 3 B). The XPS spectrum indicated evident signals for the atomic composition of PDA. The nitrogen-to-carbon molar ratio (N/C ratio) of the PDA-coated sensor chip surface is 0.120. The result is similar to the theoretical value of dopamine (N/C ratio = 0.125) and the same as the value reported [41]. These implied that the gold chip surface was coated completely with PDA.
XPS was used to analyze the chemical composition of these chip surfaces and the results were shown in Figure 3. From the analysis of XPS, compared to the Attana gold sensor chips, PDA-coated gold sensor chip was observed the disappearance of the photoelectron peaks of Au (56.7 eV for Au5p3, 84.2 eV for Au4f7, 87.9 eV for Au4f5, 111.8 eV for Au5s, 334.3 eV for Au4d5, 353.1 eV for Au4d3, 545.7 eV for Au4p3, 641.1 eV for Au4p1, and 762.1 eV for Au4s, Figure 3 A), and at the same time, the emergence of the peaks of carbon (C1s, 283.1 eV), oxygen (O1s, 531.5 eV), and nitrogen (N1s, 398.2 eV) (Figure 3 B). The XPS spectrum indicated evident signals for the atomic composition of PDA. The nitrogen-to-carbon molar ratio (N/C ratio) of the PDA-coated sensor chip surface is 0.120. The result is similar to the theoretical value of dopamine (N/C ratio = 0.125) and the same as the value reported [41]. These implied that the gold chip surface was coated completely with PDA.
Compared to the PDA-coated sensor chip, it was found that XPS spectrum signals of the sensor chip surface chemical composition (N1s, C1s, and O1s) changed after Compound 6 immobilized (Figure 3 C). The N/C ratio of the PDA-coated chip surface was 0.120. With the immobilization of Compound 6, the ratio changed to 0.097, which is close to the theoretical value of Compound 6 (N/C = 0.083). These showed clearly that Compound 6 were immobilized on the PDA-coated chip surface.  Compared to the PDA-coated sensor chip, it was found that XPS spectrum signals of the sensor chip surface chemical composition (N1s, C1s, and O1s) changed after Compound 6 immobilized (Figure 3 C). The N/C ratio of the PDA-coated chip surface was 0.120. With the immobilization of Compound 6, the ratio changed to 0.097, which is close to the theoretical value of Compound 6 (N/C = 0.083). These showed clearly that Compound 6 were immobilized on the PDA-coated chip surface.

Binding of FITC-Con A on the PDA Coated Carbohydrate Chip Surface
When a stable baseline was achieved (frequency shift < 0.2 Hz/min), in order to block the PDA areas which was not modified with carbohydrate, bovine serum albumin (BSA) was repeatedly injected over the carbohydrate chip surface until the surface was saturated and no more binding could be detected. In order to prove the interaction between lectin and carbohydrate by fluorescence imaging, FITC-Con A with green fluorescence was used for the interaction study. After the injection of the BSA, FITC-Con A (50 µg/mL) was subsequently injected onto the sensor chip surface. As shown in Figure 4, the binding response of the injection of FITC-Con A was recorded, resulting in a frequency shift about 50 Hz. Before and after the injection of the FITC-Con A, a fluorescence microscopic evaluation was performed. Before injection, only a black field was observed, while green fluorescence could be observed after injection of FITC-Con A onto the carbohydrate chip surface. These different results confirmed that the frequency shift monitored by QCM specifically reflected the surface interactions between FITC-Con A and the carbohydrate which was immobilized on the PDA-coated surface. recorded, resulting in a frequency shift about 50 Hz. Before and after the injection of the FITC-Con A, a fluorescence microscopic evaluation was performed. Before injection, only a black field was observed, while green fluorescence could be observed after injection of FITC-Con A onto the carbohydrate chip surface. These different results confirmed that the frequency shift monitored by QCM specifically reflected the surface interactions between FITC-Con A and the carbohydrate which was immobilized on the PDA-coated surface.

QCM Measurements of Lectin-Carbohyhdrate Interactions
The method of fabrication of the carbohyhdrate sensor chip surface was evaluated using a QCM instrument. Three carbohydrates synthesized in the study were immobilized onto the sensor chip surface of PDA coating, and five different lectins were interacted with the three carbohydrates, respectively.
Firstly, the carbohydrate sensor chip was inserted to QCM, and then equilibrated under PBS running buffer at a flowrate of 25 μL/min. In order to minimize nonspecific binding, BSA solution was used to block the surfaces with several injections before the injection of lectins. When a stable baseline was achieved (frequency drift < 0.2 Hz/min), BSA (50 μg/mL) was injected to the carbohydrate chip surface. Then, an injection consisting of 50 μg/mL lectin (WGA, SBA, PNA, UEA-I, Con A) was performed in order to assess the interactions between the immobilized carbohydrate and the different lectins. Following each lectin injection, 10 mM glycine at pH 1.5 was injected to the surface to remove the interacting lectin. The processes of interaction and regeneration between lectin and carbohydrate were monitored using the Attester software. Figure 5 shows five different lectins (WGA, SBA, PNA, UEA-I, Con A) binding to GlcNAc-functionalized sensor chip surface. From the different lectin binding results, it can be noted that WGA proved to be the highest binding with more than 30 Hz based on binding specificity, small amounts of potential nonspecific lectin bindings were detected, hypothetically owing to nonspecific binding between unblocked PDA-coated carbohydrate sensor chip surface and lectins.

QCM Measurements of Lectin-Carbohyhdrate Interactions
The method of fabrication of the carbohyhdrate sensor chip surface was evaluated using a QCM instrument. Three carbohydrates synthesized in the study were immobilized onto the sensor chip surface of PDA coating, and five different lectins were interacted with the three carbohydrates, respectively.
Firstly, the carbohydrate sensor chip was inserted to QCM, and then equilibrated under PBS running buffer at a flowrate of 25 µL/min. In order to minimize nonspecific binding, BSA solution was used to block the surfaces with several injections before the injection of lectins. When a stable baseline was achieved (frequency drift < 0.2 Hz/min), BSA (50 µg/mL) was injected to the carbohydrate chip surface. Then, an injection consisting of 50 µg/mL lectin (WGA, SBA, PNA, UEA-I, Con A) was performed in order to assess the interactions between the immobilized carbohydrate and the different lectins. Following each lectin injection, 10 mM glycine at pH 1.5 was injected to the surface to remove the interacting lectin. The processes of interaction and regeneration between lectin and carbohydrate were monitored using the Attester software. Figure 5 shows five different lectins (WGA, SBA, PNA, UEA-I, Con A) binding to GlcNAc-functionalized sensor chip surface. From the different lectin binding results, it can be noted that WGA proved to be the highest binding with more than 30 Hz based on binding specificity, small amounts of potential nonspecific lectin bindings were detected, hypothetically owing to nonspecific binding between unblocked PDA-coated carbohydrate sensor chip surface and lectins.
The five different lectins (WGA, SBA, PNA, UEA-I, Con A) binding to Man-and Gal-functionalized chip surfaces were also studied, respectively. The binding results between the different lectins and the three carbohydrate sensor chip surfaces were represented in Figure 6. From the binding analysis, Man-functionalized sensor chip surface displayed binding specificity to Con A, Gal-functionalized sensor chip surface displayed binding specificity to PNA. Comparison of the results in sensitivity among the three kinds of carbohydrate sensor chip surfaces shows that the surfaces had a specificity binding to different lectins, which is consistent with traditional ELLA test results [42]. Furthermore, the previous study of the interaction between the above-mentioned carbohydrates and lectins based on carbohydrate chips by other biosensors demonstrated the same specificity binding results [20,43,44].
Gal-functionalized sensor chip surface displayed binding specificity to PNA. Comparison of the results in sensitivity among the three kinds of carbohydrate sensor chip surfaces shows that the surfaces had a specificity binding to different lectins, which is consistent with traditional ELLA test results [42]. Furthermore, the previous study of the interaction between the above-mentioned carbohydrates and lectins based on carbohydrate chips by other biosensors demonstrated the same specificity binding results [20,43,44].

Kinetic Studies
PDA-coated carbohydrate sensor surface was utilized to study the interaction kinetics between lectins and the immobilized carbohydrate. To study the interaction kinetic properties between Con A and mannose, a dilute solution of Con A in the running buffer was injected to the surface. The concentrations were 20, 10, and 5 μg/mL. The interaction between lectin and carbohydrate was monitored for 85 s association while for 300 s dissociation. After each cycle, 0.1 M glycine (pH 1.5) results in sensitivity among the three kinds of carbohydrate sensor chip surfaces shows that the surfaces had a specificity binding to different lectins, which is consistent with traditional ELLA test results [42]. Furthermore, the previous study of the interaction between the above-mentioned carbohydrates and lectins based on carbohydrate chips by other biosensors demonstrated the same specificity binding results [20,43,44].

Kinetic Studies
PDA-coated carbohydrate sensor surface was utilized to study the interaction kinetics between lectins and the immobilized carbohydrate. To study the interaction kinetic properties between Con A and mannose, a dilute solution of Con A in the running buffer was injected to the surface. The concentrations were 20, 10, and 5 μg/mL. The interaction between lectin and carbohydrate was monitored for 85 s association while for 300 s dissociation. After each cycle, 0.1 M glycine (pH 1.5)

Kinetic Studies
PDA-coated carbohydrate sensor surface was utilized to study the interaction kinetics between lectins and the immobilized carbohydrate. To study the interaction kinetic properties between Con A and mannose, a dilute solution of Con A in the running buffer was injected to the surface. The concentrations were 20, 10, and 5 µg/mL. The interaction between lectin and carbohydrate was monitored for 85 s association while for 300 s dissociation. After each cycle, 0.1 M glycine (pH 1.5) was injected for regeneration of the surface which enable remove the interacting lectin. Figure 7 shows the resulted binding responses. Using the ClampXP software, the global fit with the theoretical 1:1 interaction model was performed, giving the resultant association rate constant k on = 3.51 × 10 4 M −1 ·s −1 , dissociation rate constant k off = 1.60 × 10 −3 s −1 , and equilibrium dissociation constant K D = 45.6 nM.
In order to study the interaction affinity properties between WGA andGlcNAc, a dilute solution of WGA in the running buffer was injected to the surface. The concentrations were 10, 5, and 2 µg/mL. Figure 8 shows the measured responses, giving the resultant including the association rate constant k on = 3.59 × 10 5 M −1 ·s −1 , dissociation rate constant k off = 1.70 × 10 −2 s −1 , and equilibrium dissociation constant K D = 47.2 nM. KD = 45.6 nM.
In order to study the interaction affinity properties between WGA andGlcNAc, a dilute solution of WGA in the running buffer was injected to the surface. The concentrations were 10, 5, and 2 μg/mL. Figure 8 shows the measured responses, giving the resultant including the association rate constant kon = 3.59 × 10 5 M −1 ·s −1 , dissociation rate constant koff = 1.70 × 10 −2 s −1 , and equilibrium dissociation constant KD = 47.2 nM. Figure 7. Kinetic evaluation of Con A and mannose interaction. The Con A at 20, 10, 5 μg/mL was injected over the sensor surface, respectively, and the measured responses were recorded as black lines, the theoretical 1:1 fits using the ClampXP software (Attana) were shown in red lines.  Kinetic evaluation of Con A and mannose interaction. The Con A at 20, 10, 5 µg/mL was injected over the sensor surface, respectively, and the measured responses were recorded as black lines, the theoretical 1:1 fits using the ClampXP software (Attana) were shown in red lines.
In order to study the interaction affinity properties between WGA andGlcNAc, a dilute solution of WGA in the running buffer was injected to the surface. The concentrations were 10, 5, and 2 μg/mL. Figure 8 shows the measured responses, giving the resultant including the association rate constant kon = 3.59 × 10 5 M −1 ·s −1 , dissociation rate constant koff = 1.70 × 10 −2 s −1 , and equilibrium dissociation constant KD = 47.2 nM. Figure 7. Kinetic evaluation of Con A and mannose interaction. The Con A at 20, 10, 5 μg/mL was injected over the sensor surface, respectively, and the measured responses were recorded as black lines, the theoretical 1:1 fits using the ClampXP software (Attana) were shown in red lines. Figure 8. Kinetic evaluation of WGA and GlcNAc interaction. The WGA at 10, 5, 2 μg/mL was injected over the sensor surface, respectively, and the measured responses were recorded as black lines, the theoretical 1:1 fits using the ClampXP software (Attana) were shown in red lines.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1. Author Contributions: K.S. and S.S. designed the experiments and performed the majority of the experimental work; Y.C. participated in synthesizing amino-carbohydrates; L.G. performed the part of the experimental work; Y.P. supervised the work and wrote the manuscript; X.L. performed the part of the synthesis work, T.A. participated in developing the analytical method and the interpretation of QCM data; Z.P. co-supervised the work and revised the manuscript. Figure 8. Kinetic evaluation of WGA and GlcNAc interaction. The WGA at 10, 5, 2 µg/mL was injected over the sensor surface, respectively, and the measured responses were recorded as black lines, the theoretical 1:1 fits using the ClampXP software (Attana) were shown in red lines.

Conclusions
In this study, a novel approach for fabricating carbohydrate chips based on PDA surface to determinate carbohydrate-lectin interactions by QCM was developed. Three kinds of carbohydrate chips were prepared by immobilizing the amino-monosaccharide derivatives on PDA-coated sensor surface. The surface via PDA coating was prepared through simple incubation by immersing the gold chip surface in the alkaline dopamine solution. Five kinds of plant lectins, including Con A, SBA, WGA, PNA, and UEA I, were evaluated for their binding to different kinds of carbohydrate chips, where the resulting frequency shifts show that the predicted lectin selectively bind to the respective carbohydrates. The further kinetic studies of the interactions between ConA and mannose, WGA and GlcNAc were performed, respectively. This work provides a good example for fabricating carbohydrate chips to evaluate carbohydrate-lectin interactions in real-time by QCM biosensor, which shows a potential application for studying biological processes.