The Development and Evaluation of Reagentless Glucose Biosensors Using Dendritic Gold Nanostructures as a Promising Sensing Platform

Reagentless electrochemical glucose biosensors were developed and investigated. A graphite rod (GR) electrode modified with electrochemically synthesized dendritic gold nanostructures (DGNs) and redox mediators (Med) such as ferrocenecarboxylic acid (FCA), 1,10-phenathroline-5,6-dione (PD), N,N,N′,N′-tetramethylbenzidine (TMB) or tetrathiafulvalene (TTF) in combination with glucose oxidase (GOx) (GR/DGNs/FCA/GOx, GR/DGNs/PD/GOx, GR/DGNs/TMB/GOx, or GR/DGNs/TTF/GOx) were developed and electrochemically investigated. A biosensor based on threefold-layer-by-layer-deposited PD and GOx (GR/DGNs/(PD/GOx)3) was found to be the most suitable for the determination of glucose. To improve the performance of the developed biosensor, the surface of the GR/DGNs/(PD/GOx)3 electrode was modified with polypyrrole (Ppy) for 5 h. A glucose biosensor based on a GR/DGNs/(PD/GOx)3/Ppy(5 h) electrode was characterized using a wide linear dynamic range of up to 39.0 mmol L−1 of glucose, sensitivity of 3.03 µA mM−1 cm−2, limit of detection of 0.683 mmol L−1, and repeatability of 9.03% for a 29.4 mmol L−1 glucose concentration. The Ppy-based glucose biosensor was characterized by a good storage stability (τ1/2 = 9.0 days). Additionally, the performance of the developed biosensor in blood serum was investigated.


Introduction
Interest in accurate, inexpensive, sensitive, selective, miniature, and rapid electrochemical biosensors [1][2][3][4][5] for clinical and pharmaceutical chemistry, drug discovery, and food and environmental quality monitoring has increased in the last few decades [1][2][3][4][5][6][7]. Glucose, as a monosaccharide, has significant importance as a vital energy source for organisms [2,7]. It is estimated that the number of deaths from diabetes in 2030 will account for 3.5% of deaths caused by non-infectious diseases [8]. The concentration of glucose in the bloodstream serves as the primary indicator for diagnosing and monitoring patients with diabetes mellitus [1,3]. Because of the high concentration of glucose in the blood of diabetic patients, the retina, kidneys, and the nervous and circulatory systems are damaged [3]. More than USD 13 billion and approximately 85% of the global biosensor market are used for the medical diagnostics of glucose to treat and prevent this disease [3,8].
The performance of glucose biosensors, as well as their longer operational stability and sensitivity, depends on many factors, including proper glucose oxidase (GOx) immobilization. For this purpose, the adsorption of GOx on the working electrode surface is ineffective in most cases; thus, additional crosslinking of GOx, covalent binding to a which is formed during the enzymatic GOx reaction, by generating radical cations of pyrrole [41,42]. The radical cations combine to first form oligomers and then polymers [42]. A conductive Ppy film can be deposited on the surface of the electrode, providing a stable and porous matrix for the incorporation of nanomaterials and immobilization of enzymes (e.g., horseradish peroxidase (HRP), GOx, or laccase) [27,36,39]. The formed linear or branched polymers with broad molecular weight distributions are useful for biomedical and biosensing applications [36].
The aim of the present study was to select a suitable redox mediator for the development of reagentless glucose biosensors based on DGNs, to investigate the impact of the polypyrrole layer on the performance of the developed biosensors, and to test an analytical system for the determination of glucose in the serum.

The Preparation of the Graphite Rod Electrode
The GR electrode sealed in a silicone tube (0.071 cm 2 of the area) was covered with long, thin, and branched DGNs, according to a previously presented methodology [18,19]. Electrochemical synthesis of the DGNs was performed with a solution of 6 mmol L −1 HAuCl 4 with 0.1 mol L −1 KNO 3 using a computerized potentiostat/galvanostat Autolab/PGSTAT 302N (EcoChemie, Utrecht, The Netherlands) with GPES 4.9 software at a constant potential of −0.4 V vs. Ag/AgCl (3 mol L

−1
KCl) for 400 s [18]. Then, 4.5 µL of redox mediator (38 mmol L −1 FCA, PD, and TMB or 72 mmol L −1 TTF) was added onto the GR/DGN electrode, and the solvent was evaporated at +20 ± 2 • C. After that, the GOx from 3 µL of 25 mg mL −1 solution was immobilized on the GR/DGNs/Med electrode, and the solvent was evaporated at +20 ± 2 • C. Then, the modified GR electrode was stored in a closed vessel in a 25% solution of GA for 15 min at +20 ± 2 • C ( Figure S1). A schematic representation of the preparation of the GR/DGNs/FCA/GOx, GR/DGNs/PD/GOx, GR/DGNs/TMB/GOx, and GR/DGNs/TTF/GOx is presented in Figure 1A. The GR/DGNs/(PD/GOx) 3 electrode was prepared according to the described procedure using the layer-by-layer technique: PD and GOx were deposited three times onto the GR surface modified with dendritic gold nanostructures ( Figure 1B). GR/DGNs/TMB/GOx, and GR/DGNs/TTF/GOx is presented in Figure 1A. The GR/DGNs/(PD/GOx)3 electrode was prepared according to the described procedure using the layer-by-layer technique: PD and GOx were deposited three times onto the GR surface modified with dendritic gold nanostructures ( Figure 1B).
To determine the electroactive surface area (EASA) of the electrodeposited DGNs, cyclic voltammograms were recorded in an aqueous 0.5 mol L −1 H2SO4 solution in the range of 0.0 to +1.4 V. The scan rate was 0.025 V s −1 . The EASA was estimated using the following equation [43]: where A-the integrated peak of gold oxide reduction, v-the potential scan rate (V s −1 ), and 386 µC cm −1 -the charge density per unit area associated with the electrochemical reduction of a monolayer of chemisorbed oxygen on polycrystalline gold.
To determine the EASA of the modified graphite rod electrodes, the potential was swept in the range of −0.70 to 0.0 V with varying scan rates (0.025, 0.050, 0.075, 0.100, 0.125, and 0.150 V s −1 ). The cyclic voltammograms were registered in a solution consisting of 0.1
To determine the EASA of the modified graphite rod electrodes, the potential was swept in the range of −0.70 to 0.0 V with varying scan rates (0.025, 0.050, 0.075, 0.100, 0.125, and 0.150 V s −1 ). The cyclic voltammograms were registered in a solution consisting of 0.1 mol L −1 KCl and 1 mmol L −1 Ru(NH 3 ) 6 Cl 3 . The EASA was calculated for the GR/DGNs/TTF/GOx, GR/DGNs/(PD/GOx) 3 , GR/DGNs/TTF/GOx/Ppy (3.5 h) , and GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrodes using the Randles-Sevcik equation: where i p -maximum peak current (A), n-the number of electrons appearing in the halfreaction for the redox pair, D-diffusion coefficient (cm 2 s −1 ), C-concentration of electroactive species (mol·cm −3 ), and υ-scan rate (V s −1 ). All electrochemical measurements were repeated at least three times and evaluated as the mean value. The statistic software SigmaPlot (version 12.5) was used to estimate the intercept, slope, and correlation coefficient of the calibration curve, the limit of detection (LOD), the difference in maximal current response registered during the enzymatic reaction (∆I max ), and the apparent Michaelis constant (K M(app) ). ∆I max and K M(app) were calculated as the a and b parameters of the hyperbolic function y = ax/(b + x) and were used for the approximation of the received results. The value of the LOD was estimated statistically with the software SigmaPlot 12.5 and defined as the lowest amount of glucose which provided a current response greater than the background current response value plus 3 σ.

The Enzymatic Synthesis of Polypyrrole and the Stability of the Developed Glucose Biosensors
To improve the analytical characteristics of the developed glucose biosensors, the enzymatic polymerization of pyrrole was performed by immersing the GR/DGNs/TTF/GOx or GR/DGNs/(PD/GOx) 3 electrodes in a polymerization solution in the dark for 1.8, 3.5, 8, and 24 or 2, 5, 17, 21, and 46 h at +20 ± 2 • C. The polymerization solution contained 0.05 mol L −1 of SA buffer at pH 6.0, 0.2 mol L −1 of pyrrole, and 0.05 mol L −1 of glucose. Glucose oxidase, glucose, and oxygen are required for enzymatic polymerization to proceed [36].

The Application of a Biosensor Based on a GR/DGNs/(PD/GOx) 3 /Ppy (5 h) Electrode for Glucose Detection in Serum
The serum was diluted 10 times and centrifuged (8 min, 14,600× g) using an IEC CL31R Multispeed centrifuge (Aze Bellitourne, Château-Gontier, France) according to the previously described method [19]. CPA investigations were performed for a glucose biosensor based on the GR/DGN/(PD/GOx) 3 /Ppy (5 h) electrode. The developed glucose biosensor was tested in serum diluted 10 times with 10.0 mmol L −1 of glucose before and after the addition of 1.0 mmol L −1 of carbohydrate (saccharose, xylose, galactose, mannose, or fructose). This study involved an evaluation of the influence of electroactive species on the current response. Experiments were performed in serum diluted 10 times with either 10.0 mmol L −1 of glucose alone or 10.0 mmol L −1 of glucose with 0.01 or 0.1 mmol L −1 of AA or with 0.01 or 0.05 mmol L −1 of UA.

Electrochemical Characterization of the Developed Biosensors
The nanostructurization of the electrode surface is one of the methods for improving the analytical parameters of the developed biosensors. In this work, electrodeposited gold nanostructures were used for this purpose. DGNs were synthesized according to a previously published study [18,19]. DGNs have a high surface area, which increases the analytical value of their signals. The EASA of the DGNs was evaluated using Equation (1) ( Figure S2). The EASA of the synthesized gold nanostructures was found to be 1.23 ± 0.01cm 2 , which is significantly larger than the geometric area (0.071 cm 2 ) of the graphite rod electrode.
The data regarding the mechanism of the redox process and the electron transfer can be obtained in the form of a cyclic voltammogram and peak shift [37]. The electrochemical behavior of the glucose biosensors based on GR/DGNs/FCA/GOx, GR/DGNs/PD/GOx, GR/DGNs/(PD/GOx) 3 , GR/DGNs/TMB/GOx, and GR/DGNs/TTF/GOx was studied in 0.05 mol L −1 SA buffer of pH 6.0 with 0.1 mol L −1 KCl via CV according to the methodology described in Section 2.3. The registered cyclic voltammograms are presented in Figure 2A.
( Figure S2). The EASA of the synthesized gold nanostructures was found to be 1.23 ± 0.01cm 2 , which is significantly larger than the geometric area (0.071 cm 2 ) of the graphite rod electrode.
The data regarding the mechanism of the redox process and the electron transfer can be obtained in the form of a cyclic voltammogram and peak shift [37]. The electrochemical behavior of the glucose biosensors based on GR/DGNs/FCA/GOx, GR/DGNs/PD/GOx, GR/DGNs/(PD/GOx)3, GR/DGNs/TMB/GOx, and GR/DGNs/TTF/GOx was studied in 0.05 mol L −1 SA buffer of pH 6.0 with 0.1 mol L −1 KCl via CV according to the methodology described in Section 2.3. The registered cyclic voltammograms are presented in Figure 2A. It is clear that electron transfer from the GOx redox center to electrodes without mediators does not take place when using the GR/GOx and GR/DGNs/GOx electrodes ( Figure 2B). As can be seen from Figure 2B, electron transfer begins only in the presence of PD on the GR/DGNs electrode, and a similar behavior was observed with another redox mediator. The cyclic voltammogram registered using the GR/DGNs/FCA/GOx electrode ( Figure 2A) was characterized by a small anodic peak at the potential of +0.50 V vs. Ag/AgCl(3 mol L −1 KCl). However, the cyclic voltammograms obtained using the GR/DGNs/PD/GOx, GR/DGNs/TMB/GOx, and GR/DGNs/TTF/GOx electrodes were characterized by clearly visible anodic and cathodic peaks ( Figure 2A). Anodic peaks were noted at +0.010 V vs. Ag/AgCl(3 mol L −1 KCl) for the GR/DGNs/PD/GOx and GR/DGNs/(PD/GOx)3 electrodes and at + 0.54 V vs. Ag/AgCl(3 mol L −1 KCl) for the GR/DGNs/TMB/GOx electrode. The shape of the cyclic voltammograms and the position of the anodic peak for the GR/DGNs/FCA/GOx and GR/DGNs/PD/GOx electrodes are very similar to those obtained using the GC/NPG/Cys/GA/GOx electrode in the presence of FCA (+0.316 V) [17] and the GR/PD/GOx (+0.010 V) [27] electrode, respectively. The two anodic peaks at + 0.30 and +0.67 V vs. Ag/AgCl(3 mol L −1 KCl) were obtained in the case of the cyclic voltammograms recorded using the GR/DGNs/TTF/GOx electrode. The presence of two peaks can be explained by the reversible oxidation/reduction process between the neutral TTF, radical cation (TTF + ), and dication (TTF 2+ ) states [44]. Similar cyclic voltammograms were registered previously for the GR/TTF/GOx-electrode-based glucose biosensor, with the first oxidation peak at +0.307 V vs. Ag/AgCl(3 mol L −1 KCl) [27]. The obtained TTF + is considered an effective redox mediator for electron transfer [31].
To ensure efficient electron transfer on the GR/DGNs/PD/GOx or GR/DGNs/TTF/GOx electrodes, the potential value of +0. It is clear that electron transfer from the GOx redox center to electrodes without mediators does not take place when using the GR/GOx and GR/DGNs/GOx electrodes ( Figure 2B). As can be seen from Figure 2B, electron transfer begins only in the presence of PD on the GR/DGNs electrode, and a similar behavior was observed with another redox mediator. The cyclic voltammogram registered using the GR/DGNs/FCA/GOx electrode (Figure 2A) was characterized by a small anodic peak at the potential of +0.50 V vs. Ag/AgCl ( KCl) were obtained in the case of the cyclic voltammograms recorded using the GR/DGNs/TTF/GOx electrode. The presence of two peaks can be explained by the reversible oxidation/reduction process between the neutral TTF, radical cation (TTF + ), and dication (TTF 2+ ) states [44]. Similar cyclic voltammograms were registered previously for the GR/TTF/GOx-electrode-based glucose biosensor, with the first oxidation peak at +0.307 V vs. Ag/AgCl (3 mol L
To ensure efficient electron transfer on the GR/DGNs/PD/GOx or GR/DGNs/TTF/GOx electrodes, the potential value of +0.40 V vs. Ag/AgCl (3 mol L

The Selection of the Optimal Redox Mediator for Glucose Biosensor Construction
The nature of the redox mediator has a significant influence on the sensitivity of electrochemical biosensors. To evaluate the impacts of the FCA, PD, TMB, and TTF on the current responses, five types of working electrodes, including GR/DGNs/FCA/GOx, GR/DGNs/PD/GOx, GR/DGNs/(PD/GOx) 3 , GR/DGNs/TMB/GOx, and GR/DGNs/ TTF/GOx, were prepared. The calibration plots using 38 mmol L −1 of FCA, PD, or TMB or 72 mmol L −1 of TTF are shown in Figure 3A. The redox mediators used were compared with each other in terms of effectiveness. It was observed that all the calibration plots were in agreement with Michaelis-Menten kinetics.
or TMB or 72 mmol L −1 of TTF are shown in Figure 3A. The redox mediators used were compared with each other in terms of effectiveness. It was observed that all the calibration plots were in agreement with Michaelis-Menten kinetics.
The difference in maximal current response depends on the Med used. In the case of the FCA, TMB, PD, and threefold-deposited PD redox mediators, the ΔImax was calculated as 2.25 ± 0.92, 25.9 ± 2.8, 34.1 ± 9.6, and 69.1 ± 7.7 µA, respectively. The highest value of ΔImax was achieved for the glucose biosensor based on the GR/DGNs/TTF/GOx electrode −107 ± 7 µA. Meanwhile, in the case of the GR/DGNs/GOx-electrode-based glucose biosensor without a Med, no current response was recorded, which was expected, since direct electron transfer is difficult in the case of GOx, because FAD, as a cofactor, is located deep in the molecule globule, which is electrically well-isolated [45]. The value of ΔImax for the biosensor based on the GR/DGNs/(PD/GOx)3 electrode was 2.03 times higher than that obtained using the GR/DGNs/PD/GOx electrode. DGN fabrication leads to an increase in the electroactive surface area, and thus, an improved signal is recorded. This can be explained by the one-dimensional hollow tubular form of PD, which promotes electron transfer [46]. For further studies, biosensors based on the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx)3 electrodes were selected, as the highest current responses were registered for these electrodes.

Influence of the Polypyrrole Layer on Biosensor Performance
The kind of electrode The difference in maximal current response depends on the Med used. In the case of the FCA, TMB, PD, and threefold-deposited PD redox mediators, the ∆I max was calculated as 2.25 ± 0.92, 25.9 ± 2.8, 34.1 ± 9.6, and 69.1 ± 7.7 µA, respectively. The highest value of ∆I max was achieved for the glucose biosensor based on the GR/DGNs/TTF/GOx electrode −107 ± 7 µA. Meanwhile, in the case of the GR/DGNs/GOx-electrode-based glucose biosensor without a Med, no current response was recorded, which was expected, since direct electron transfer is difficult in the case of GOx, because FAD, as a cofactor, is located deep in the molecule globule, which is electrically well-isolated [45].
The value of ∆I max for the biosensor based on the GR/DGNs/(PD/GOx) 3 electrode was 2.03 times higher than that obtained using the GR/DGNs/PD/GOx electrode. DGN fabrication leads to an increase in the electroactive surface area, and thus, an improved signal is recorded. This can be explained by the one-dimensional hollow tubular form of PD, which promotes electron transfer [46]. For further studies, biosensors based on the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx) 3 electrodes were selected, as the highest current responses were registered for these electrodes.

Influence of the Polypyrrole Layer on Biosensor Performance
To improve the performance of the biosensors based on the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx) 3 electrodes, the enzymatic formation of the Ppy layer was performed. Firstly, the Ppy layer was enzymatically synthesized on the surfaces of the GR/DGNs/TTF/GOx or GR/DGNs/(PD/GOx) 3 electrodes according to the protocol described in Section 2.4. Then, the dependences of the current responses on the glucose concentration after various polymerization times were registered ( Figures 4B and 5B, respectively), and the ∆I max and K M(app) were evaluated.
A high KM(app) value is considered to be a significant indicator of the linear dynamic range (LDR) width of the glucose biosensor. The value of KM(app), which was determined for the GR/DGNs/TTF/GOx electrode without Ppy and after 1. According to the hyperbolic dependences shown in Figures 4B and 5B, the LDR for the fabricated glucose biosensors was evaluated. As seen in Figures 4C and 5C, the LDRs were without an intercept on the x-or y-axis and could be used for glucose determination. The LDRs of the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx)3 electrodes were extended by increasing the duration of Ppy synthesis and depended on the kind of redox mediator Based on hyperbolic dependences, the calculated ∆I max values for the glucose biosensors based on the GR/DGNs/TTF/GOx or GR/DGNs/(PD/GOx) 3 electrodes after the Ppy layer's formation depended on the Ppy layer formation time. As can be seen from Figures 4A and 5A, as the polymerization time increased, a decrease in ∆I max was observed using both of the electrodes tested. In the case of the electrode prepared using TTF, when the Ppy was synthesized for 1.8, 3.5, 8, or 24 h of enzymatic polymerization ( Figure 4A), the ∆I max decreased from 108 ± 7.5 µA (calculated for the GR/DGNs/TTF/GOx electrode) by 4.48 (∆I max = 24.1 ± 5.8 µA), 5.68 (∆I max = 19.0 ± 3.4 µA), 6.32 (∆I max = 17.1 ± 3.7 µA), and 31.1 (∆I max = 3.47 ± 1.32µA) times, respectively. In the case of the GR/DGNs/(PD/GOx) 3electrode-based biosensor modified with the Ppy layer during 2, 5, 17, 21, or 46 h of polymerization, the ∆I max decreased from 68.0 ± 4.3 µA (calculated for the GR/DGNs/(PD/GOx) 3 electrode) by 2.17 (∆I max = 31.3 ± 3.3), 2.63 (∆I max = 25.9 ± 2.4), 3.51 (∆I max = 19.4 ± 4.5), 5.31 (∆I max = 12.8 ± 3.4), and 6.67 (∆I max = 10.2 ± 4.0 µA) times, respectively ( Figure 5A). Comparing the obtained results for the glucose biosensors based on the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx) 3 electrodes, we noticed that the enzymatic formation of Ppy significantly affected the performance of the tested electrodes. However, a stronger decrease in ∆I max was observed using TTF. This could be attributed to the poor stability of the biosensor modified with TTF [29,35].
A high K M(app) value is considered to be a significant indicator of the linear dynamic range (LDR) width of the glucose biosensor. The value of K M(app) , which was determined for the GR/DGNs/TTF/GOx electrode without Ppy and after 1. According to the hyperbolic dependences shown in Figures 4B and 5B, the LDR for the fabricated glucose biosensors was evaluated. As seen in Figures 4C and 5C, the LDRs were without an intercept on the xor y-axis and could be used for glucose determination. The LDRs of the GR/DGNs/TTF/GOx and GR/DGNs/(PD/GOx) 3 electrodes were extended by increasing the duration of Ppy synthesis and depended on the kind of redox mediator used. In the case of the GR/DGNs/TTF/GOx electrodes ( Figure 4C) modified with a Ppy layer for 1.8, 3.5, 8, and 24 h of polymerization, the LDRs were extended from 1 mmol L −1 to 2.00, 2.99, 4.48, and 9.92 mmol L −1 of glucose. As seen in Figure 5C, the LDRs after 2 and 5 h of Ppy layer formation on the GR/DGNs/(PD/GOx) 3 electrode were extended from 16.5 mmol L −1 to 26.2 mmol L −1 and 39.0 mmol L −1 of glucose; thus, the LDR was extended by 1.59 and 2.36 times, respectively. When the polymerization time was increased to 17 h, there was no extension of the LDR. A further increase in the polymerization duration to 24 h prolonged the LDR but was accompanied by a significant decrease in the recorded current response. Taking into account the discussed analytical parameters, the GR/DGNs/TTF/GOx/Ppy (3.5 h) or GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrodes were chosen for further comparison and can be recommended as more suitable for glucose biosensor construction and the determination of glucose. The LDRs for the unmodified and selected electrodes were characterized by a correlation coefficient above 0.9900.
The LDR for the developed reagentless biosensor based on the GR/DGNs/TTF/GOx electrode was the same as that obtained by other authors using a reagentless biosensor based on a Au electrode modified with a Cys self-assembled monolayer, TTF, GOx, and alcohol oxidase (AOx) (Au-Cys/TTF/GOx-AOx) (up to 1.0 mmol L −1 of glucose) [31]. Modification with the Ppy layer increased the LDR, but this increase was not enough, as compared to almost all the other biosensors in Table 1. As mentioned earlier, the LDR was much wider (up to 39.0 mmol L −1 ) when using a reagentless biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode. Thus, the LDR of this electrode was at least twice as wide as that of the other electrodes ( Table 1).
The electroactive surface area of the electrodes was calculated according to the methodology described in Section 2.3. Ru(NH 3 ) 6 Cl 3 was used as a redox probe, because its oxidation occurs in a region where the studied electrodes do not yield an additional signal (Figures 2 and S5). For the calculation, a value of the Ru(NH 3 ) 6 Cl 3 diffusion coefficient equal to 9.1 × 10 −6 cm 2 s −1 was used [49]. The EASA of the GR/DGNs/TTF/GOx, GR/DGNs/TTF/GOx/Ppy (3.5 h) , GR/DGNs/(PD/GOx) 3 , and GR/DGNs/(PD/GOx) 3 / Ppy (5 h) electrodes was calculated as 0.504, 0.648, 0.736, and 0.901 cm 2 , respectively, according to the results presents in Figure S6. Modifying the electrodes with the Ppy layer allowed for an EASA increase of more than 20%. The EASA of the electrodes obtained using PD was larger than that of the electrodes fabricated using TTF.

The Stability of Glucose Biosensors
The stability of biosensors is an essential parameter that defines the duration of their use [7]. Thus, the GR/DGNs/TTF/GOx, GR/DGNs/TTF/GOx/Ppy (3.5 h) , GR/DGNs/ (PD/GOx) 3 , and GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrodes were stored at +4 • C in SA buffer solution at pH 6.0 for up to 22 and 25 days, respectively. The variation in the current responses over time using the glucose biosensors based on electrodes modified with TTF or PD is shown in Figure 6A,C, respectively. As presented in Figure 6B,D, the calibration plots were in agreement with Michaelis-Menten kinetics.
larger than that of the electrodes fabricated using TTF.

The Stability of Glucose Biosensors
The stability of biosensors is an essential parameter that defines the duration of their use [7]. Thus, the GR/DGNs/TTF/GOx, GR/DGNs/TTF/GOx/Ppy(3.5 h), GR/DGNs/(PD/GOx)3, and GR/DGNs/(PD/GOx)3/Ppy(5 h) electrodes were stored at +4 °C in SA buffer solution at pH 6.0 for up to 22 and 25 days, respectively. The variation in the current responses over time using the glucose biosensors based on electrodes modified with TTF or PD is shown in Figure 6A,C, respectively. As presented in Figure 6B,D, the calibration plots were in agreement with Michaelis-Menten kinetics. As seen in Figure 6A,C, the current responses of the glucose biosensors based on the GR/DGNs/TTF/GOx (line 1) and GR/DGNs/(PD/GOx)3 (line 3) electrodes significantly decreased during the first two days, and only 12.9 or 49.0% of their initial value was retained. However, the modification of the developed electrodes with the Ppy layer could increase the stability of the biosensors. The current responses obtained using the GR/DGNs/TTF/GOx/Ppy(3.5 h) ( Figure 6A, line 2) and GR/DGNs/(PD/GOx)3/Ppy(5 h) ( Figure  6C, line 4) electrodes were found to decrease over the studied period and retained 11. As seen in Figure 6A,C, the current responses of the glucose biosensors based on the GR/DGNs/TTF/GOx (line 1) and GR/DGNs/(PD/GOx) 3 (line 3) electrodes significantly decreased during the first two days, and only 12.9 or 49.0% of their initial value was retained. However, the modification of the developed electrodes with the Ppy layer could increase the stability of the biosensors. The current responses obtained using the GR/DGNs/TTF/GOx/Ppy (3.5 h) (Figure 6A, line 2) and GR/DGNs/(PD/GOx) 3 /Ppy (5 h) ( Figure 6C, line 4) electrodes were found to decrease over the studied period and retained 11.3 and 24.9% of the initial current response after 22 and 25 h, respectively. The decrease in current response could be related to GOx inactivation and Ppy degradation during the maintenance of the glucose biosensor over time. It can be seen that the biosensor based on GR/DGNs/TTF/GOx/Ppy (3.5 h) retained 44.7% of the initial current response after the first two days, whereas the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode was much more stable, as it retained 92.5% of the initial current response. These results suggest that the Ppy-modified electrodes are preferable to the unmodified ones.
The noticeably, the poor TTF stability noted in a previous paper [29] and established in our studies and narrow LDR influenced the choice of the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode for the reagentless glucose biosensor's construction and further glucose determination in blood serum.

Glucose Determination in Blood Serum Using the Developed Glucose Biosensor Based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) Electrode
Saccharose, xylose, galactose, mannose, and fructose are considered as species interfering in glucose biosensing. The ascorbic acid and uric acid belong to electroactive species that can affect the current response of glucose biosensors in the blood serum [4,[50][51][52].
The selectivity of the reagentless glucose biosensor based on the GR/DGNs/(PD/GOx) 3 / Ppy (5 h) electrode to interfering and electroactive species was studied in 10-times-diluted serum according to the protocol described in Section 2.5. Firstly, the selectivity of the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode was evaluated, and it is summarized in Figure 7A. As can be seen from the obtained results, the current response registered by the biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode was almost the same after the addition of similar carbohydrates, which confirms its selectivity to glucose. days) electrode [47].
The noticeably, the poor TTF stability noted in a previous paper [29] and established in our studies and narrow LDR influenced the choice of the GR/DGNs/(PD/GOx)3/Ppy(5 h) electrode for the reagentless glucose biosensor's construction and further glucose determination in blood serum.

Glucose Determination in Blood Serum Using the Developed Glucose Biosensor Based on the GR/DGNs/(PD/GOx)3/Ppy(5 h) Electrode
Saccharose, xylose, galactose, mannose, and fructose are considered as species interfering in glucose biosensing. The ascorbic acid and uric acid belong to electroactive species that can affect the current response of glucose biosensors in the blood serum [4,[50][51][52].
The selectivity of the reagentless glucose biosensor based on the GR/DGNs/(PD/GOx)3/Ppy(5 h) electrode to interfering and electroactive species was studied in 10-times-diluted serum according to the protocol described in Section 2.5. Firstly, the selectivity of the GR/DGNs/(PD/GOx)3/Ppy(5 h) electrode was evaluated, and it is summarized in Figure 7A. As can be seen from the obtained results, the current response registered by the biosensor based on the GR/DGNs/(PD/GOx)3/Ppy(5 h) electrode was almost the same after the addition of similar carbohydrates, which confirms its selectivity to glucose. The next step was to study the effect of ascorbic acid and uric acid on the current response registered in the presence of glucose. It is established that the normal physiological concentration of these electroactive species in human blood serum does not normally exceed 0.141 mmol L −1 of AA [53] and 0.1 mmol L −1 of UA [50,54]. These acids, when present in biological samples, can be oxidized at a higher positive potential than that required for glucose oxidation [14]. The addition of 10.0 mmol L −1 glucose with 0.01 or 0.1 mmol L −1 of AA increased the current response to glucose obtained the using GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode by 1.80 and 5.39%, compared with the results obtained without AA ( Figure 7B). The GR/DGNs/(PD/GOx) 3 /Ppy (5 h) -electrode-based reagentless glucose biosensor developed in this study was 1.79 and 1.14 times more resistant to 0.01 and 0.1 mmol L −1 of AA than the GR/Ppy/GNPs (AuCl4 − ) -GOx/GOx electrode (the interferences of 3.23 and 6.16% for 0.01 or 0.1 mmol L −1 AA were registered) [47]. The addition of 10.0 mmol L −1 glucose containing 0.01 and 0.05 mmol L −1 of UA increased the current response of the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode by 4.69 and 10.2%, respectively, compared with the results registered after adding 10.0 mmol L −1 of glucose without UA ( Figure 7B). The current responses of the previously developed glucose biosensor based on the GR/Ppy/GNPs (AuCl4 − ) -GOx/GOx electrode changed by 2.19 and 13.4%, respectively, after the addition of 10.0 mmol L −1 of glucose containing 0.01 or 0.05 mmol L −1 of UA [47]. In general, an interference of 10% is considered acceptable for electroactive species [3,52]. The reagentless biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode is characterized by good glucose selectivity and can be used for blood serum glucose analysis. The stated normal glucose concentration in human blood is usually below 5.6 mmol L −1 [3,14,39,54], while in diabetic patients, it can rise to 30 mmol L −1 [3,39]. The suitability of the developed reagentless biosensor based on the GR/DGNs/(PD/GOx) 3 / Ppy (5 h) electrode for glucose detection in serum was tested using the addition method. The results obtained from three measurements were expressed as average values and are presented in Table 2. The recovery ratios using the developed reagentless glucose biosensor were in the range of 97.5 ± 6.5 to 98.0 ± 6.3%. The developed glucose biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode, according to its recovery ratio (97.5-98.0%) and relative error (4.76-6.50%), is no worse than the commercial methods and sensors. As reported by Conzales's group, the recovery ratio and relative error depend on the country, agency, method of detection, and amount of glucose [55]. For example, in the USA, for ≥75 mg dL −1 (4.17 mmol L −1 ) of glucose, the detection criterion is 98 ± 15%; in Europe, for ≥100 mg dL −1 (5.55 mmol L −1 ) of glucose, it is −95 ± 15%; and for people with type 1 diabetes, it is −99% [55]. It should be noted that the advantages of the developed reagentless glucose biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode are as follows: (i) a low LOD (0.683 mmol L −1 ), wide LDR (up to 39.0 mmol L −1 of glucose), and good repeatability (the RSD was 9.03%); (ii) a good storage stability (50% of the current response was retained after 9.0 days); (iii) the fast determination of glucose (approximately 8 s); and (iv) high resistance to electroactive species and suitability for glucose detection in blood serum (97.5-98.0%). The electrochemical biosensor, after some improvements and mandatory clinical validation, could be used in clinical practice for the quantitative determination of glucose.

Conclusions
In this study, we developed an enzymatic reagentless glucose biosensor based on a GR electrode modified with electrochemically synthesized DGNs in combination with redox mediators such as TTF and PD, which had the simplicity to operate at a low cost with a quick response, high current response, and good sensitivity. After the formation of the Ppy layer, the reagentless biosensor based on the GR/DGNs/(PD/GOx) 3 /Ppy (5 h) electrode was characterized by a wider LDR and higher stability than that based on the GR/DGNs/TTF/GOx/Ppy (3.5 h) electrode. The limitation of the GR/DGNs/(PD/GOx) 3 / Ppy (5 h) electrode was its 3.03 µA mM −1 cm −2 sensitivity in comparison to GR/DGNs/ (PD/GOx) 3 electrode (13.4 µA mM −1 cm −2 ). However, characterized by a wide LDR (up to 39.0 mmol L −1 ) and good stability, the biosensor based on the GR/DGNs/(PD/GOx) 3 / Ppy (5 h) electrode was adapted for practical application and the determination of glucose in blood serum in the presence of interfering species. This principle of biosensor development can be adapted to other types of reagentless biosensors suitable for the analysis of real samples. The developed and improved reagentless glucose biosensor, due to its wide LDR, could be applied in various research areas, e.g., for biomedical purposes, for the control of beverages, for biofuel cell fabrication, and in bioelectronics devices.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/bios13070727/s1, Figure S1: Image of modified GR electrode storage in a closed vessel in a 25% solution of GA for 15 min at +20 ±2 • C; Figure S2: Cyclic voltammogram registered using a GR/DGNs electrode in 0.5 mol L −1 H 2 SO 4 solution; Figure S3: Amperograms registered using enzymatic glucose biosensors based on GR electrodes modified by DGNs, without or with redox mediators. Experiments were performed in 0.05 mol L −1 SA buffer, pH 6.0, with 0.1 mol L −1 KCl; Figure S4: Amperograms registered using biosensors based on GR/DGNs/TTF/GOx/Ppy (A) and GR/DGNs/(PD/GOx) 3 Figure S6: The relationships between the square root of the scan rate and registered peak anodic current; Figure S7: Amperograms registered at different times after the preparation of biosensors based on GR/DGNs/TTF/GOx (A) and GR/DGNs/TTF/GOx/Ppy (3.