Glucose Biosensor Based on a Glassy Carbon Electrode Modified with Polythionine and Multiwalled Carbon Nanotubes

A novel glucose biosensor was fabricated. The first layer of the biosensor was polythionine, which was formed by the electrochemical polymerisation of the thionine monomer on a glassy carbon electrode. The remaining layers were coated with chitosan-MWCNTs, GOx, and the chitosan-PTFE film in sequence. The MWCNTs embedded in FAD were like “conductive wires” connecting FAD with electrode, reduced the distance between them and were propitious to fast direct electron transfer. Combining with good electrical conductivity of PTH and MWCNTs, the current response was enlarged. The sensor was a parallel multi-component reaction system (PMRS) and excellent electrocatalytic performance for glucose could be obtained without a mediator. The glucose sensor had a working voltage of −0.42 V, an optimum working temperature of 25°C, an optimum working pH of 7.0, and the best percentage of polytetrafluoroethylene emulsion (PTFE) in the outer composite film was 2%. Under the optimised conditions, the biosensor displayed a high sensitivity of 2.80 µA mM−1 cm−2 and a low detection limit of 5 µM (S/N = 3), with a response time of less than 15 s and a linear range of 0.04 mM to 2.5 mM. Furthermore, the fabricated biosensor had a good selectivity, reproducibility, and long-term stability, indicating that the novel CTS+PTFE/GOx/MWCNTs/PTH composite is a promising material for immobilization of biomolecules and fabrication of third generation biosensors.


Introduction
Glucose has always been an analyte that has received much attention because of its vital effect in clinical and environmental analysis. Since the first concept of an enzyme-based biosensor was proposed by Clark and Lyons [1], tremendous interest has been focused on the fabrication of glucose enzyme biosensors. The enzyme biosensor has experienced three stages in previous years. Due to its better selectivity and stability, the third-generation biosensor based on the direct electrochemistry of glucose oxidase has been widely developed [2][3][4].
However, for the flavin adenine dinucleotide (FAD) that is the active centre of glucose oxidase (GOx) and is deeply embedded in the molecular cavity [5], it is generally difficult to establish direct electron transfer (DET) between the active centre of GOx and conventional electrodes. To enhance this communication, nanomaterials have been extensively used in biosensors to achieve DET. Various nanomaterials composed of carbon [6,7] metal oxides [8,9] and conducting polymers [10,11] have demonstrated the capabilities of DET. However, significant challenges remain in the fabrication of practicable biosensors that show superior bioelectrocatalysis, precise specificity, and high stability.
Multiwalled carbon nanotubes (MWCNTs) have unique electronic, mechanical, and thermal properties, which leads to their potential application in the fabrication of biosensors. Many recent reports have demonstrated that the integration of MWCNTs and conducting polymers [12,13] or metal nanoparticles [14,15] could greatly improve the direct electrochemistry of GOx.
Recently, conducting polymers have been extensively employed for the construction of electrochemical biosensors due to their conductive nature and superior stability [16]. Thionine (TH) is a phenothiazine dye (CH 3 COOC 12 H 10 N 3 S) with two amino groups symmetrically distributed on each side of its aromatic rings [17]. Under a suitable voltage, the aromatic rings of TH can be linked together via -NH-bridges to form polythionine (PTH) [18]. The PTH has good conductivity and exhibits a fast charge transfer capacity, showing great potential for various future applications [19,20].
In this study, we investigated the step-wise immobilizations of PTH, MWCNTs, GOx, and chitosan-PTFE composite films onto a glassy carbon electrode to build a biosensor for measuring glucose. A mediator-free GOx-based glucose biosensor was constructed though a layer-by-layer self-assembly approach. Electrochemical impedance spectroscopy was used to monitor the biosensor modification process. DET between the active site of the GOx and the electrode was achieved, and its behaviour as a biosensor and its enzyme stability were investigated by electrochemical techniques.

Apparatus and measurements
Amperometric experiments and cyclic voltammetric experiments were performed on an Autolab AUT72230 electrochemical work station (Metrohm China Ltd.). A conventional threeelectrode system was applied: the modified or unmodified GCE electrode (3 mm diameter) acted as the working electrode, and a saturated calomel electrode (SCE) and a platinum wire served as the reference and counter electrodes, respectively. All of the electrochemical experiments were carried out at room temperature.

Electrode modification
The glassy carbon electrode (GCE, W = 3 mm) was polished with 1.0 and 0.05 mm a-Al 2 O 3 slurries successively and was then rinsed with water, acetone, absolute ethanol, and doubly-distilled water. According to the method described by Wang [21], the PTH/GCE was built and covered with a golden PTH film after 50 cycles of the electrodeposition of thionine (the potential range was between +0.1 and 20.55 V at 50 mV s 21 ).
Two mg MWCNTs was dispersed by ultrasonication in 1 mL 0.5% chitosan (CTS) solution. One mg GOx and 7.5 mg bovine serum albumin (BSA) were added to 100 mL 0.1 M PBS to prepare a 10 mg mL 21 GOx aqueous solution.
A CTS+PTFE/GOx/MWCNTs/PTH/GCE coating was prepared facilely by successively casting 6 mL 2 mg mL 21 MWCNTs, 6 mL 10 mg mL 21 GOx aqueous solution (using glutaraldehyde crosslinking as the fixation method), and 5 mL 2% (w/w) chitosan/ PTFE suspension onto the pre-treated PTH/GCE. Each casting  was performed after the previous cast had been air-dried. Finally, the modified GCE was immersed in PBS with a pH = 7.0 to remove the loosely absorbed GOx and was stored at 4uC under dry conditions. For comparison, MWCNTs/GCE,PTH/GCE,MWCNTs/PTH/ GCE, and GOx/MWCNTs/GCE coatings were also prepared according to the same procedure. The assembly process of CTS+ PTFE/GOx/MWCNTs/PTH is showed in Figure 1.

SEM characterisations of MWCNTs and electropolymerisation of thionine
SEM characterisation of the MWCNTs. Figure 2 shows the SEM image of the MWCNTs deposited onto the glassy carbon surface. The disordered filamentous carbon nanotubes, with tube diameters of tens of nanometres, built a porous and threedimensional structure, which can provide a conductive, porous, and biocompatible microenvironment for enzyme immobilization.
Electropolymerisation of thionine. During the process of thionine electropolymerisation (Figure 3), the currents of a pair of reversible redox peaks increased gradually with increasing scan numbers, which can be attributed to the redox reaction of the PTH at the electrode. The anodic peak was located at 20.22 V and the cathodic peak was at 20.26 V after 50 scans. The results are very similar to Wang's study [21], and the electrode is also covered by a firm golden film after electropolymerisation. In addition, a derived redox peak was located at 20.07 V, which may be due to the differences in the raw materials.

Electrochemical behaviour of CTS+PTFE/GOx/MWCNTs/ PTH/GCE
EIS can give more detailed information on the impedance changes during the modification process. When presented in a Nyquist plot, the impedance spectrum includes a semi-circular portion corresponding to the electron-transfer-limited process and a linear part resulting from the diffusion process [22]. The diameter of the semicircle corresponds to the electron transfer resistance (RET) of the redox probe at the electrode interface. A smaller RET value implies that the probe has a higher interfacial electron transfer rate.
Compared with bare GCE (Figure 4), the diameter of the semicircle greatly and successively decreased after the modification with MWCNTs/GCE, PTH/GCE, and MWCNTs/PTH/GCE, which indicated that both the MWCNTs and the PTH are good conductive materials, with the latter showing better conductive performance. The combination of MWCNTs and PTH on the GCE (MWCNTs/PTH/GCE) can greatly reduce the electrode resistance and promote a higher electron transfer rate between the redox probe and the electrode surface. However, an increase in  the electron transfer resistance was observed after CTS+PTFE/ GOx was immobilised on the probe, suggesting a decrease in the conductivity of CTS+PTFE/GOx/MWCNTs/PTH/GCE due to the immobilization of non-conductive PTFE and biomacromolecules. The results of the impedance change during the electrode modification can provide evidence for the successful immobilization of MWCNTs, PTH, GOx, and other materials on the electrodes. Figure 5 (A) shows that for the GCE, there was no obvious redox peaks; two pairs of distinct redox peaks (20.2 V and 20.07 V, respectively) were observed for PTH/GCE, which were confirmed as the characteristic peaks of PTH according to its electropolymerisation. For GOx/MWCNTs/GCE, there was a pair of standard redox peaks, whose reduction and oxidation peak potentials were 20.443 V and 20.471 V, respectively. These peak potentials were very close to the standard electrode potential of GOx, which is derived from the FAD/FADH 2 GOx redox cofactor [23,24]. The results also proved that the MWCNTs can directly promote electron transfer between GOx and the electrode. At 20.45 V, 20.27 V, and 20.13 V, three pairs of well-defined and nearly symmetric redox peaks were observed in GOx/ MWCNTs/PTH. This result indicated that both GOx and PTH maintained good redox activities and that the modification layers provided a proper ''micro room'' for GOx. Both of the redox peak potentials were negatively shifted, which may be due to the parallel multi-component reaction system (PMRS). On the one hand, the redox species on the biosensor kept their separate redox characteristics; on the other hand, they were interdependent in the modification layers as a whole. Due to the high-impedance of GOx and PTFE, the peak current value of PMRS declined  From Figure 5 (B), it can be observed that the ratio of the GOx oxidation and reduction peaks is close to 1 (i pa1 /i pc1 <1) and DEp 1 <30 mV; however, that ratio for PTH is slightly greater than 1 (i pa2 /i pc2 .1) and DEp 2 <20 mV. These results further demonstrated the good reversibility of PMRS.
Both the anodic and cathodic peak currents of GOx and PTH increased linearly with an increase in the scan rate from 10 to 300 mV s 21 (Figure 6 (A), (B), and (C)), which is characteristic of a surface-controlled quasi-reversible electrochemical process [25]. These results suggested that GOx and PTH were immobilised as adsorbed reactants on the electrode surface.

Biocatalytic activity of GOx on the biosensor
The electrocatalytic activity of GOx for oxygen was convincing proof of the biocatalytic activity of the GOx immobilised on the electrode surface [26]. From Figure 7, three pairs of redox peaks were observed in both N 2 -saturated and air-saturated PBS solutions. Compared to the peak currents of both GOx and PTH in the corresponding N 2 -saturated solution, an increase in the cathodic peak current and decrease in anodic peak current were observed in the air-saturated solution. This response can be explained as follows: In N 2 -saturated PBS, DET occurred between the active site of the enzyme and the stable bioelectrode. When in an air-saturated solution, the electrochemically formed GOx (FADH 2 ) catalysed the reduction of oxygen dissolved in the solution. The relative consumption of GOx (FADH 2 ) and generation of GOx (FAD) directly led to the changes in the peak currents of GOx. Because GOx and PTH were both in the PMRS of the electrode, the peak currents of PTH had the same tendency to change, which can be used as an indicator of the change in GOx in turn.

Glucose Sensing Performance of CTS+PTFE/GOx/ MWCNTs/PTH/GCE
In Figure 8 (A), three pairs of redox peaks (GOx located at 20.45 V and PTH located at 20.3 V) were observed in the PBS without glucose. With increases in the glucose concentration, both the anodic and cathodic peak currents decreased, which can be explained by the following reactions: Electrode : GOx(FAD)z2e { z2H z ?GOx(FADH 2 ) ð4Þ The equation indicates that GOx (FAD) is directly involved in the glucose oxidation reaction. As the glucose concentration  increases, GOx (FAD) is competitively employed by glucose and increasingly converted into GOx (FADH 2 ), leading to a continuous decrease in the cathodic and corresponding anodic peak currents of both GOx and PTH. Figure 8 (B) depicts the plot of the cathodic peak currents of PTH vs. the glucose concentrations. The cathodic current decreased obviously with an increasing glucose concentration up to 10 mM. When the glucose concentration continued to increase, the current decreased slowly and finally flattened out. This result arose from GOx consumption and is similar to the research results of Si Peng [14], which suggested that the GOx on the surface of such an electrode is able to simultaneously demonstrate DET with the electrode and to retain its catalytic activity towards glucose. The sensing mechanism in our system is different from the first and second generation glucose sensors. Figure 9 (A) shows that each injection of 0.25 mM glucose resulted in a decrease of 50 nA in the cathodic current for the first 10 additions, which gave rise to a sensitivity of 2.80 mA mM 21 cm 22 . In addition, the current decreased steeply and then reached 95% of the steady-state current in less than 15 s. The current verses time plot showed a well-defined behaviour typical of an enzymatic reaction with a linear range up to 2.5 mM, and a current plateau was observed when the glucose concentration was higher than 2.5 mM, which suggests typical enzymatic reaction kinetics [28,29].
The calibration curve (current vs. glucose concentration) and the Lineweaver Burk plot (current 21 vs. concentration 21 ) were obtained from the amperometric response (Figure 9 (B)). The linear regression equation of I (mA) = 0.196 (mA mM 21 ) C (mM)-0.01 (mA) (R = 0.997) was derived from the calibration curve and revealed that the glucose sensor has a detection limit as low as 5.0 mM (S/N = 3).
The apparent Michaelis-Menten constant K M was calculated using the electrochemical version of the Lineweaver Burk where I ss is the steady-state current, I max is the saturation current, and C is the glucose concentration. K M was determined to be 3.1 mM from the intercept and slope of the Lineweaver Burk plot (Figure 9 (B)), which is much lower than the values of 8.5 mM for GOx immobilised on literature angle nanotubes [30], 13.9 mM for GOx immobilised on silica/multi-walled carbon nanotube/ polyacrylonitrile nanocomposite layer redox polymers [31], and 14.6 mM for GOx immobilised on gold nanoparticles [32]. The smaller K M indicates that this biosensor has a superior enzymatic activity and a higher affinity for the glucose substrate [33].
Optimization conditions for the glucose detection system As shown in Figure 10, the sequence of the absolute differences between the starting and ending current values was Ic start 2Ic end , Ia start 2Ia end ,Ib start 2Ib end , which was attributed to the combination of differently modified materials. For the glucose detection range, the size order was Step C.
Step B.
Step A, which showed that the biosensor using the CTS+PTFE/GOx/MWCNTs/ PTH/GCE composite had a better amperometric response. The reason was that the CTS+PTFE film had good permeability and could make glucose molecules diffuse into the electrode surface smoothly and orderly.
We also did a series of research experiments to optimise the working conditions of the glucose sensor (the related figures and data are not shown). The biosensor has an optimum working voltage of 20.42 V, an optimum working temperature of 25uC, and an optimum working pH of 7.0. In addition, the best percentage of PTFE in the outer composite film was 2%.

Selectivity, reproducibility, and stability of the glucose biosensor
To evaluate the selectivity of the biosensor, many interfering biomolecules, including uric acid (UA), ascorbic acid (AA), and L-cysteine (L-cys), that normally co-exist with glucose in real samples (human blood) were investigated ( Figure 11). The current generated from the interfering species, such as 1.0 mM of UA, AA, and L-cysteine, was negligible, indicating that the biosensor had a strong anti-interference ability. This strong anti-interference ability can mainly be attributed to the low working voltage and reduction method we adopted in this study, which can effectively avoid the oxidation of the interfering substances at high voltages.
We put three sensors in PBS solution, which had a glucose concentration of 1.0 mM, to perform current tests. Then, the sensors were saved in the refrigerator at 4uC to perform stability tests. As Figure 12 shows, the current response had no obvious changes at the beginning after 2 days. After 3 weeks, the current responses decreased to 89.6%, 86.7%, and 91.7% of the initial response. In Figure 12, the relative standard deviation (RSD) was small and the reproducibility was good. The biosensor displayed an acceptable stability, which can be attributed to the following two properties: the immobilization of the MWCNTs and the thionine electropolymerisation on the electrode surface were very stable, which can provide a friendly environment for GOx and maintain its bioactivity; and the CTS+PTFE composite can effectively prevent enzyme leakage and prolong the life of the biosensor.
The mechanism of DET of the sensor In Figure 13, the active center of GOx (FAD) is buried deeply, only a few GOx can occur DET (DET-GOx), and the other part is just loaded up without DET-GOx (WDET-GOx). Both of them have biological catalysis activity, and can catch glucose and oxide them.
The current response of CTS+PTFE/GOx/MWCNTs/PTH is from the competitive reaction of glucose. Glucose competitively spreads to DET-GOx and reacts with it, thereby the DET-GOx is reduced, and thus the response current is decreased.
DET-GOx influences detection performance. The MWCNTs embedded in FAD are like ''conductive wires'' connecting FAD with electrode, reduce the distance between them and are propitious to DET. Combining with good electrical conductivity of PTH and MWCNTs, the current response is enlarged. According to response current, glucose concentration can be detected.

Conclusions
In this work, we successfully introduced a promising glucose biosensor based on the immobilization of GOx onto a MWCNTs/ PTH nanocomposite film. The evaluation of the CTS+PTFE/ GOx/MWCNTs/PTH/GCE demonstrated that DET of GOx  was achieved on this nanocomposite film. Cyclic voltammetry showed three pairs of well-defined redox peaks corresponded to DET of GOx (FAD/FADH 2 ) and PTH. The parallel multicomponent reaction system (PMRS) of the electrode played an important role in facilitating the electron transfer between GOx and the electrode surface. To the best of our knowledge, this is the first report about the direct electrochemistry of GOx based on PMRS. Furthermore, the potential use of the resulting electrode as a third generation glucose biosensor was confirmed by the amperometric response for successive additions of glucose. The biosensor showed a high sensitivity, fast amperometric response and low detection limit. Possible interfering species in blood, such as uric acid, ascorbic acid, and L-cysteine, did not influence the glucose determination. In addition, the CTS+PTFE film had good permeability and could greatly improve glucose detection range and stability. The MWCNTs embedded in FAD were like ''conductive wires'' connecting FAD with electrode, reduced the distance between them and were propitious to DET. Combining with good electrical conductivity of PTH and MWCNTs, the current response was enlarged. These results demonstrate that the CTS+PTFE/GOx/MWCNTs/PTH is an attractive material for the fabrication of efficient amperometric biosensors. The method presented can be used for the immobilization and evaluation of DET of other enzymes or proteins.