Evaluating bioreceptor immobilization on Gold Nanospike (AuNS)–modified Screen-Printed Carbon Electrode (SPCE) as enzymatic glucose biosensor

Abstract

Integration of gold nanoparticles onto electrochemical biosensor electrodes has been widely conducted to improve the performance of biosensors. Gold nanospikes (AuNS), as one of the gold nanoparticle morphologies, can be integrated into biosensors through electrodeposition and has the potential to immobilize bioreceptor on biosensors using the self-assembled monolayer (SAM) method. This paper examines the potential of AuNS-deposited Screen-Printed Carbon Electrodes (SPCEs) on immobilizing enzymes as label-based electrochemical biosensor by evaluating the optimum parameter for glucose oxidase (GOx) enzyme immobilization on the SPCE that consists of incubation time and concentration of SAM molecule—L-cysteine—and GOx enzyme, then reviews its performances. The developed biosensor exhibits excellent performance in detecting glucose (linear range of 0.2–15 mM and limit of detection (LOD) of 116 µM), with good selectivity against uric acid, urea, ascorbic acid, dopamine, and lactic acid, and superiority towards gold nanosphere modified biosensor.

GRAPHICAL ABSTRACT

Keywords:

• Gold nanospike
• electrodeposition
• screen-printed carbon electrode
• self-assembled monolayer
• L-cysteine
• glucose oxidase
• glucose
• amperometric electrochemical biosensor

Introduction

Gold nanospikes (AuNS), as one of the gold nanoparticle morphologies, show enormous potential for biosensor applications. This potential arises from their high electrocatalytic ability, primarily due to the presence of Au (111) as the dominant facet, coupled with increased active sites on numerous Au (311) facets [1]. AuNS can easily be formed and integrated on the electrode surface using electrodeposition method [1,2]. Until this day, AuNS has been evaluated to be used as dopamine [1] and arsenic [3] sensors. As gold nanoparticles, AuNS is known to have a high surface area-to-volume ratio, allowing them to accommodate a significant volume of bioreceptors. It has been studied to immobilize deoxyribonucleic acid (DNA) using layer-by-layer methods [4], and antigens using self-assembled monolayer (SAM) methods [5,6]. Surface modification of biosensors using bioreceptors is recognized for its ability to facilitate specific and selective interactions with substrates [7]. However, studies assessing bioreceptor immobilization on AuNS and their applications as label-based biosensors remain limited. Therefore, further evaluation of bioreceptor immobilization on AuNS is needed.

In this study, the bioreceptors immobilization on AuNS is conducted using the SAM method, which has been previously proven to immobilize bioreceptor on gold surfaces [5,6, 8–10]. The SAM molecule connects and immobilizes the bioreceptor on gold surfaces using strong covalent bonds, forming a self-assembled, tidy, and dense connector monolayer [11]. The strong covalent bond makes biosensor with SAM has good stability [11,12]. The chosen SAM molecule, L-cysteine, is easily acquired and has been studied for immobilizing various bioreceptors, including antibodies [13,14] and enzyme [15]. As for the bioreceptor, we used glucose oxidase (GOx), one of the most common bioreceptor for biosensors that is often utilized for blood glucose detection which need to fulfill detection range of 4.9–6.6 mM for the normal person and > 11.1 mM for diabetes patients [8,16,17]. Reaction between GOx and glucose leads to electron transfer in the biosensor (Eq. (1)(1) $$\begin{array}{l}\beta -D‐\mathrm{\text{glucose}}+\mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}\right)\to \mathrm{\text{gluconolactone}}+\mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}{\mathrm{H}}_2\right)\\ \mathrm{G}\mathrm{\text{Ox}}\left(\mathrm{\text{FAD}}{\mathrm{H}}_2\right)\leftrightarrow \mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}\right)+2H^{+}+2e^{-}\end{array}$$(1) ) [16]. In the presence of water and oxygen, GOx oxidizes glucose into gluconic acid and hydrogen peroxide (H₂O₂) (Eq. (2)(2) $$\begin{array}{l}\beta -D‐\mathrm{\text{glucose}}+{\mathrm{O}}_2\mathrm{G}\mathrm{\text{Ox}}\to {\mathrm{H}}_2{\mathrm{O}}_2+g\mathrm{\text{luconolactone }}{\text{H}}_2\mathrm{O}\to \text{gluconic acid}\\ {{\mathrm{H}}_2{\mathrm{O}}_2\to \mathrm{O}}_2+2H^{+}+2e^{-})\end{array}$$(2) ) [18,19]. The electron transfers of the GOx–glucose reaction is then evaluated through amperometry methods. (1) $$\begin{array}{l}\beta -D‐\mathrm{\text{glucose}}+\mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}\right)\to \mathrm{\text{gluconolactone}}+\mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}{\mathrm{H}}_2\right)\\ \mathrm{G}\mathrm{\text{Ox}}\left(\mathrm{\text{FAD}}{\mathrm{H}}_2\right)\leftrightarrow \mathrm{\text{GOx}}\left(\mathrm{\text{FAD}}\right)+2H^{+}+2e^{-}\end{array}$$(1) (2) $$\begin{array}{l}\beta -D‐\mathrm{\text{glucose}}+{\mathrm{O}}_2\mathrm{G}\mathrm{\text{Ox}}\to {\mathrm{H}}_2{\mathrm{O}}_2+g\mathrm{\text{luconolactone }}{\text{H}}_2\mathrm{O}\to \text{gluconic acid}\\ {{\mathrm{H}}_2{\mathrm{O}}_2\to \mathrm{O}}_2+2H^{+}+2e^{-})\end{array}$$(2)

Currently, there is no research of enzyme immobilization on AuNS-deposited SPCE, specifically the one that utilizes L-cysteine as SAM to immobilize GOx enzyme. However, Lović et al. successfully used L-cysteine to immobilize GOx on Au electrodes [15]. Therefore, this work aims to evaluate the potential of AuNS-deposited SPCE as a label-based electrochemical biosensor, by assessing the optimal GOx enzyme immobilization parameters that consist of SAM molecule (L-cysteine) and GOx enzyme, then review its performances.

Methods

Tools

The electrochemical activity was evaluated using ZP1000080 Anapot EIS Rev.01 potentiostat from Zimmer & Peacock, Ltd. (Coventry, UK). Material characterization was evaluated using JIB-4610F field emission scanning electron microscope with energy dispersive X-ray spectroscope (FESEM-EDX) from JEOL, Ltd. (Tokyo, Japan), SU3500 Scanning Electron Microscope (SEM) from Hitachi, Ltd. (Tokyo, Japan), Nicolet iS5 Fourier Transform Infrared Spectrometer (FT-IR Spectrometer) from ThermoScientific/ASB (Madison, WI, USA), D8 Advance X-ray Diffraction (XRD) from Bruker (Madison, WI, USA), iHR320 Raman Spectroscopy from HORIBA Scientific (Kyoto, Japan).

Materials

The electrode used was hyper value screen-printed electrode (HVSPE) purchased from Zimmer & Peacock, Ltd. (Coventry, UK), with polyethylene terephthalate (PET) as substrate material, carbon as the working electrode (WE) (WE area = 4.15 mm²), and Ag/AgCl as the reference (RE) and the counter electrode (CE). The geometric area is delimited by PET layer. Chemicals such as Gold(III) chloride trihydrate (HAuCl₄.3H₂O) ≥ 99.9%, glucose oxidase (GOx) type X-S from Aspergillus niger (147019 units/g, protein ≥ 65%), and L–cysteine (cys) 97%, and synthetic urine solution (Sigma-trix Urine Diluent) were purchased from Sigma Aldrich (St. Louis, Missouri, United States). The synthetic urine solution contains calcium chloride (CaCl₂), magnesium chloride (MgCl₂), potassium chloride (KCl), sodium chloride (NaCl), sodium phosphate, sodium sulfate (Na₂SO₄), urea, and creatinine with sodium azide (NaN₃) as a preservative. Potassium chloride (KCl) ≥ 99.5%, glutaraldehyde (glut) 25%, D(+)–glucose anhydrous (C₆H₁₂O₆) 97.5%–102.0%, and urea (CH₄N₂O) 99.0%–100.5% were purchased from Merck (Darmstadt, Germany). Uric acid (C₅H₄N₄O₃) ≥ 99% and dopamine hydrochloride (C₈H₁₂CINO₂) 98% powder were purchased from Sigma-Aldrich (Steinheim, Germany). Lactic Acid (CH₃CHCOOH) 88%–92% were purchased from Rofa Laboratorium (Bandung, Indonesia). Sulfuric acid (H₂SO₄) 95%–97% were purchased from Merck (Darmstadt, Germany) and Mallinckrodt (Staines-upon-Thames, United Kingdom). The phosphate buffer (PB) solution used consists of disodium hydrogen phosphate (Na₂HPO₄.12H₂O) 98% purchased from Pudak Scientific (Bandung, Indonesia) and sodium dihydrogen phosphate (NaH₂PO₄.2H₂O) 99% purchased from Loba Chemie (Mumbai, India). All aqueous solutions (marked with _(aq)) were prepared using sterilized distilled water as the solvent. The chemicals used in the study exhibit a range of toxicological properties and require careful handling for safety precautions. HAuCl₄.3H₂O may causes skin and eye burns, glutaraldehyde may leads to respiratory and dermatological complications, cysteine is associated with respiratory depression and ataxia, GOx poses risks of skin and eye damage, uric acid induces irritation, while ascorbic acid, glucose, lactic acid, and sulfuric acid are linked to various forms of eyes, skin, and respiratory irritations.

Biosensor fabrication

Gold Nanospike electrodeposition

First, for the activation procedure, the screen-printed carbon electrode (SPCE) was pre-treated and cleaned using cyclic voltammetry (CV) in 0.5 M H₂SO_4(aq) to activate higher electrode electrochemical active area. The gold nanospike (AuNS) was then electrodeposited by immersing the electrode in 1.5 mL of 10 mM HAuCl_4(aq) + 0.1 M KCl_(aq), running constant potential amperometry (CPA) at −0.2 V vs Ag/AgCl for 1200s, and then finally rinsing it with deionized water. The electrochemical surface area (ECSA) of the SPCE was then evaluated by analyzing surface oxide reduction while running CV (0.4–1.5 V vs Ag/AgCl; V_step = 0.01 V; scan rate = 0.05 V/s) in 0.5 M H₂SO_4(aq) solution. All the activation, electrodeposition, and electrochemical characterization procedures are conducted based on our previous studies on AuNS synthesis [1]. AuNS that has been electrochemically deposited on the SPCE’s working electrode was then materially characterized by FESEM-EDX, cross-sectional SEM, Raman spectroscopy, and XRD.

Enzyme immobilization

Three steps were involved in GOx enzyme immobilization (Figure 1). (i) Firstly, 3 µL of 30 mM cysteine (0.1 M PB_(aq) pH 7) in three aliquots of 1 µL each was drop-casted onto the AuNS-deposited SPCE at room temperature, incubated for 1 day, then rinsed with 0.01 M PB_(aq) pH 7. (ii) Next, 1.5 µL of 2.5% glutaraldehyde (0.1 M PB_(aq) pH 8) in three aliquots of 0.5 µL each was drop-casted onto the modified SPCE/AuNS/cys at room temperature, incubated for 1 h, then rinsed with 0.01 M PB_(aq) pH 8. (iii) Lastly, 1.5 µL of 3 mg/mL GOx (0.1 M PB_(aq) pH 7) in three aliquots of 0.5 µL each was drop-casted onto the modified SPCE/AuNS/cys/glut, incubated at 4 °C for 3 h. After incubation, the SPCE/AuNS/cys/glut was washed with 0.01 M PB_(aq) pH 7 and stored at 4 °C. All the materials, their respective concentration level, incubation time, and temperature used in these procedures are adopted from the study by Lović et al. with adjustment on the electrode modification method (drop casting instead of immersion) [15]. Electrochemical performance before and after enzyme immobilization were evaluated using CV (−0.6 to 1.2 V vs Ag/AgCl; scan rate = 100 mV/s; V_step = 0.01 V) in 0.1 M PB_(aq) pH 7 with and without the presence of 5 mM glucose.

Figure 1. Biosensor Fabrication Scheme.

Optimization of enzyme immobilization variables

In this study, the enzyme immobilization variables optimized were (i) incubation time of cysteine solution in 0.1 M PB_(aq) pH 7 (1, 2, 3, 4, and 24 h), (ii) concentration of cysteine solution in 0.1 M PB_(aq) pH 7 (1.5, 2.5, 5, 10, 20, and 30 mM), (iii) incubation time of GOx solution in 0.1 M PB_(aq) pH 7 (3, 20, 22, and 24 h), and (iv) concentration of GOx solution in 0.1 M PB_(aq) pH 7 (1, 3, 5, and 10 mg/mL). Each optimization variable was evaluated using chronoamperometry (+0.77 V vs Ag/AgCl for 120 s) in three concentrations of glucose solution: 0 mM (blank), 5 mM (normal blood glucose), and 10 mM (high range/diabetic blood glucose) in 0.1 M PB_(aq) pH 7. The optimal value obtained for each variable was then used to optimize the next variable until the optimum parameter for all variables was reached. In this work, no optimization for the concentration and incubation time of glutaraldehyde was conducted due to the complexity of its parameters, such as pH and temperature, which could easily affect the molecule conformation in the solution [20]. Additional optimization of analyte condition such as variation of pH (5, 6, 7, 8) of 5 mM glucose in 0.1 M PB_(aq) and variation of temperature (15, 20, 25, 30, 35, 40, and 45 °C) of 5 mM glucose in 0.1 M PB_(aq) pH 7 were conducted using the same chronoamperometry parameter as previous optimization step, to choose desired analyte condition for evaluation of biosensor performances in real world application.

Biosensor performance

The biosensor was fabricated using optimized variables. The electrochemical performances of the biosensor were evaluated using CV (−0.6 to 1.2 V vs Ag/AgCl; scan rate = 100 mV/s; V_step = 0.01 V) and chronoamperometry (+0.77 V vs Ag/AgCl for 120 s). All glucose solutions used were diluted in 0.1 M PB_(aq) pH 7 and previously stored at 4 °C for 24 h to allow mutarotation [19]. The biosensors were then materially characterized using FESEM-EDX and FTIR before and after enzyme immobilization.

To evaluate the excellence of AuNS-modified-SPCE biosensor, the performance of this sensor is also compared with gold nanosphere (AuNP)-modified-SPCE biosensor and bare-modified -SPCE biosensor (n = 2). Both modifications were first started by pre-treating SPCE with the same method as AuNS-modified-SPCE as mentioned in previous section. Then, AuNP modification was acquired by electrodepositing SPCE in 1.5 mL 10 mM HAuCl_4(aq) + 0.1 M H₂SO_4(aq) using CV (−0.2 to 1.2 V vs Ag/AgCl; 25 cycles; scan rate of 100 mV/s). This procedure is based on the study by Chiang et al. [21], but with several modifications such as the type of electrode used (SPCE was used instead of GCE), the CV parameters used, and the concentration of the electrodeposition precursor solution used. Each biosensor was then immobilized with GOx using the same optimal variable immobilization. The performance was evaluated using CV in 5 mM glucose (0.1 M PB_(aq) pH 7).

Other performance characteristics of the biosensors were also evaluated, which include the effect of scan rate variation, glucose detection performance (linear range, sensitivity, limit of detection, Michaelis-Menten constant), reproducibility, repeatability, storage stability, selectivity with interference reagents (urea, uric acid, ascorbic acid, dopamine, lactic acid, and combined analyte of glucose and all interference reagents) and selectivity for glucose in synthetic urine (pure and diluted 10x in 0.1 M PB_(aq) pH 7) evaluation.

Results and discussion

Biosensor fabrication

Gold Nanospike electrodeposition

FESEM image of electrodeposited AuNS on SPCE (Figure 2) exhibited a successfully formed gold nanospike on the electrode surface. One of the sharpest spikes measures 109.49 nm in diameter and thickness of 35.8 µm (Fig. S1). XRD data of AuNS-deposited SPCE (Fig. S2) shows Au structure formed on the surface: Au(111) as dominant facets and numerous Au(311) facets. the diffraction peaks are seen at 2θ values of 38.18°, 44.38°, 64.57°, 77.56, and 81.72° which correspond to the crystalline planes (111), (200), (220), (311), and (222) of Au metal, respectively. Raman spectroscopy was used to further characterize the electrode materials of SPCE after deposition with AuNS on the surface of SPCE. The result of the Raman spectrum of SPCE with AuNS deposited on the surface is seen in Fig. S3. The D and G bands represent the sp³ and sp² carbon hybridizations, respectively. The disorder band and graphitic band of bare-SPCE is at 1364 cm⁻¹ and 1590 cm⁻¹, correspondingly, with the I_D/I_G ratio of the D and G bands is 0.90 [22]. After deposition the AuNs on the surface of the SPCE, the I_D/I_G ratio of SPCE/AuNS is 0.96. The increasing of the I_D/I_G ratio corresponds to the surface plasmon resonance of AuNS which confirm the existence of the AuNS deposited on the electrode of SPCE.

Figure 2. FESEM image of AuNS-deposited SPCE.

The charges for gold oxide reduction obtained from the integration of reduction area of 13.9 µAV for a scan rate of 50 mV/s (Fig. S4(a)) was 278 µC. Assuming an Au density charge of 390 µC/cm² [23], the Electrochemical Surface Area (ECSA) was approximately calculated to be 0.713 cm², with a roughness factor of 17.18 times, close enough to the reference for the same electrode type [1]. The calculated ECSA and roughness factor imply that the deposition of AuNS increases the active electrochemical surface area compared to the geometric surface area. In further analyses over the next 10 CV cycles, the ECSA of the AuNS-deposited SPCE exhibited a slight reduction before stabilizing, demonstrating consistent values across the following cycles (Fig. S4(b)). This pattern indicates stability of deposited AuNS in maintaining constant ECSA throughout the study.

Enzyme immobilization

It is shown in Figure 3 that non-modified SPCE exhibits no electron transfer in 0.1 M PB_(aq) pH 7, as evidenced by the absence of prominent oxidation and reduction peaks. The redox behavior is evident in AuNS-deposited SPCE (SPCE/AuNS) data, with gold oxide formation peaking at 0.7 V. After addition of glucose, there was an increased peak in the 0.2–0.6 V range for enzyme-immobilized-SPCE (SPCE/AuNS/cys/glut/GOx), which comes from H₂O₂ oxidation [24]. This may indicate that GOx enzyme is successfully immobilized on SPCE surface since H₂O₂ is the product of glucose oxidation by GOx enzyme.

Figure 3. CV results of unmodified SPCE, AuNS-deposited SPCE, and GOx-immobilized biosensors with and without the presence of 5 mM glucose in 0.1 M PB_(aq) pH 7 (scan rate = 100 mV/s).

The oxidation peak of SPCE increased after enzyme immobilization, indicating an increase in the electron transfer rate. The immobilization conducted uses cysteine solution in 0.1 M PB_(aq) pH 7 and glutaraldehyde solution in 0.1 M PB_(aq) pH 8 to immobilize enzymes on amino ends, enabling glucose molecules to interact with the enzyme’s active site and adopt the optimal catalytic activity conformation, resulting in good enzyme activity and high electron transfer rate [15]. The glucose incubation allows the glucose molecules undergo mutarotation into D-glucose conformation, resulting in a specific reaction between D-glucose and glucose oxidase [25]. The oxidation peak of enzyme-immobilized-SPCE at +0.77 V vs Ag/AgCl was then used as a chronoamperometric parameter.

Evaluation of enzyme immobilization variables

Evaluation of cysteine variables

Incubation times of 1 h and 2 h generated a great linear response (R² > 0.9) for the applied glucose range with coefficients of determination (R²) of 0.976 and 0.978, respectively (Figure 4(a)). The drop-casted cysteine electrode was fully dried in 1 h since SAM formation on gold surface occurred within minutes and continued to stabilize until the molecule was perfectly absorbed on the electrode [11]. Therefore, a longer incubation time was not necessary. Hence, 1 h was chosen as the optimum incubation time for cysteine solution.

Figure 4. Effect of cysteine (a) incubation time and (b) concentration on the biosensor chronoamperometry (E = +0.77 V vs Ag/AgCl) steady-state response for glucose in 0.1 M PB_(aq) pH 7.

For the variation of cysteine solution (0.1 M PB_(aq) pH 7) concentration, Figure 4(b) shows that 30 mM cysteine in (0.1 M PB_(aq) pH 7) had the best linearity (R² = 1). However, it does not exhibit any pattern of linearity in accordance with the variation of cysteine concentration. Pooi See and colleagues found that the formation of cysteine SAM on gold surface saturates at 30 mM cysteine concentration [26]. Hence, based on the reference, 30 mM was chosen as the optimum cysteine concentration.

Evaluation of GOx variable

Figure 5(a) shows that 3 h of incubation of GOx solution in 0.1 M PB_(aq) pH 7 resulted in poor sensitivity for the applied glucose range, implying that the initial incubation time was not effective enough to immobilize the enzyme onto the electrode. This might have occurred because the drop-casted GOx solution on the electrode was not completely dried out due to the higher humidity in the incubation environment (4 °C inside the refrigerator) compared to room condition. We extended the incubation time, and the results showed that for 20–, 22–, and 24-h incubation time, GOx had fully dried on the electrode and generated higher sensitivity for the applied glucose range with coefficients of determination (R²) of 0.996, 0.997, and 0.998, respectively. Hence, 20 h was chosen as the optimum GOx incubation time.

Figure 5. Effect of GOx (a) incubation time and (b) concentration on the biosensor chronoamperometry (E = +0.77 V vs Ag/AgCl) steady-state response for glucose in 0.1 M PB_(aq) pH 7.

Concentrations of 1 and 3 mg/dL GOx in 0.1 M PB_(aq) pH 7 exhibited a poor linear response for the applied glucose range with coefficients of determination (R²) of 0.880 and 0.899, respectively (Figure 5(b)). While 5 mg/dL GOx in 0.1 M PB_(aq) pH 7 showed a good linear response (R² = 0.948), the best linear response (R² = 1) was achieved using 10 mg/dL GOx concentration. Hence, the optimum parameter for GOx was a concentration of 10 mg/dL with an incubation time of 20 h.

Evaluation of pH and temperature variation of analyte

The catalytic activity of GOx enzyme is closely influenced by the pH and analyte temperature [16]. Fig. S5(a) demonstrates that glucose in 0.1 M PB at pH 6 yields the highest current response. Despite the initial optimal pH for free GOx enzyme activity being 5.5, the process of immobilization may induce a shift in the optimum pH due to alterations in the ionizable groups within the enzyme’s active site. To align with the intended application for the developed biosensor, which targets human blood as the analyte, a pH closest to that of human blood (pH 7) was selected for the analyte conditions, despite the 6% deviation from the response observed at the optimum pH. Fig. S5(b) additionally illustrates that 30 °C represents the temperature at which the biosensors exhibit the highest current response. However, considering the intended practical application of the developed biosensor under real-world conditions, characterized by room temperature, the working condition was carefully set at 25 °C, resulting in a 5% deviation from the response observed at the optimum temperature. This selection aims to ensure the relevance and accuracy of the biosensor’s performance in practical human blood glucose monitoring scenarios.

Biosensor performance

Material characterization

It is shown in Fig. S6(a) that the SPCE-AuNS only contained Au from AuNS and C from the carbon on WE. However, after immobilization, there was an increased value of C with the presence of N and S (Fig. S6(b)). Since there was no N element involved before enzyme immobilization, it is possible that the N value came from cysteine and GOx, and the S value may have come from cysteine, which created a thiol bond with Au. The increased C value might come from the presence of cysteine, glutaraldehyde, and GOx.

Each step of enzyme immobilization was confirmed using FTIR measurements (Fig. S7). The spectrum for SPCE/AuNS/cys revealed characteristic peaks of amine (–NH₂) groups (N–H bending at 1606 cm⁻¹, C–N bending at 1240 cm⁻¹) and carboxylic acid (–COOH) groups (C=O stretching at 1770 cm⁻¹, O–H stretching at 3020 cm⁻¹, and O–H bending at 1402 cm⁻¹). These peaks were attributed to cysteine interaction with Au, forming a SAM [27]. Furthermore, the immobilization of the GOx enzyme onto the electrode was confirmed by the SPCE/AuNS/cys/glut/GOx spectrum, which displayed N–H stretching at 3282 cm⁻¹, amide II bands at 1531 cm⁻¹ associated with N–H bending and C–N stretching, and both overlapping amide I bands and an imine peak at 1633 cm⁻¹, reflecting C=O stretching vibrations from cysteine–GOx interaction and the bonding of GOx and cysteine amine groups with glutaraldehyde’s aldehyde group, respectively [15, 28]. These findings collectively indicated successful enzyme immobilization on the electrode.

Variation of gold-deposited modification

The biosensor performance of the AuNS-modified biosensors is compared with AuNP-modified biosensors. The CV results (Figure 6) show higher anodic and cathodic peaks for AuNS-modified biosensors compared to the other two modifications. The absence of anodic and cathodic peaks in the bare-modified biosensors was due to cysteine’s inability to bind on the carbon surface [20], resulting in no enzyme immobilization. The graph also indicates that the bare carbon electrode could not initiate a glucose redox reaction, implying that the immobilized enzyme bonded with the gold surface not the carbon electrode. The inability of bare-modified biosensors to immobilize enzymes is also supported by lack of C, N, and S values in its EDX spectra (Fig. S8(b)).

Figure 6. CV results of 5 mM glucose in 0.1 M PB_(aq) pH 7 for variation of SPCE modification (scan rate = 100 mV/s).

Comparing AuNP-modified biosensors and AuNS-modified biosensors, redox reactions occurred for AuNP-modified biosensors, but the peaks were not as high as the peaks of AuNS-modified biosensors. This difference might be due to the lower ECSA value—the ECSA of the AuNP-modified biosensor was 0.175 cm² with roughness factor 4.22—which results in poorer electrocatalytic performance. Furthermore, AuNS has more Au (111) and Au (311) facets (Fig. S2) [1], on which the thiol bond of the SAM relatively takes place. It is shown in Fig. S8(d) that the N, C, S values of AuNP-modified biosensors were lower than AuNS-modified biosensors, which indicates lower load of immobilized enzyme. Hence, it made AuNS superior in immobilizing a higher load of enzyme and increased electron transfer rate for glucose detection.

Effects of scan rate variation

Figure 7(a) presents CV results for 5 mM glucose in 0.1 M PB_(aq) pH 7 at scan rate values ranging from 20 to 200 mV/s. The linear increase of the anode peak and linear decrease of the cathode peak with the square root of scan rate (Figure 7(b)) satisfy the Randles-Sevcik equation [15, 29], implying that the reaction occurred is a surface electrochemical reaction and diffusion-controlled process.

Figure 7. (a) CV results of 5 mM glucose in 0.1 M PB_(aq) pH 7 in various scan rates. (b) Linearity curve between the anodic-cathodic peak current with the square root of the potential scan rate.

Linear range, sensitivity, limit of detection, and Michaelis-Menten constant

The linear range was evaluated through a chronoamperometry measurement calibration plot of two biosensors in the range of 0–30 mM glucose (0.1 M PB_(aq) pH 7) (Figure 8 inset). The linear range was determined to be 0.2–15 mM with a high correlation coefficient of linear regression (R²) of 0.994 (Figure 8). This range satisfies the human blood glucose detection range: 4.9–6.6 mM for the normal range and > 11.1 mM for random diabetic blood glucose range diabetes [17], as well as the human urine glucose detection range: 2.78–5.5 mM for normal range and > 5.5 mM for diabetic urine glucose range [30]. The detection limit of 116 µM was experimentally obtained from calculation of blank measurement using Equation 3(3) $$\text{Detection limit}=3.3\times \frac{\text{Standard Deviation}}{\text{Slope}}$$(3) [31], with a sensitivity of 1.01 µA/mM or equivalent to 24.41 µA/(mM.cm²) for WE area 4.15 mm². (3) $$\text{Detection limit}=3.3\times \frac{\text{Standard Deviation}}{\text{Slope}}$$(3)

Figure 8. Biosensor linear range towards glucose in 0.1 M PB_(aq) pH 7 (E = +0.77 V vs Ag/AgCl) (inset) calibration plot of biosensor towards 0–30 mM glucose in 0.1 M PB_(aq) pH 7 (E = +0.77 V vs Ag/AgCl).

The Michaelis-Menten constant (K_m) was determined using the Lineweaver-Burk curve that satisfies Equation 4(4) $$\frac{1}{I}=\frac{K_m}{I_{\mathrm{\text{max}}}}\frac{1}{C}+\frac{1}{I_{\mathrm{\text{max}}}}$$(4) . (4) $$\frac{1}{I}=\frac{K_m}{I_{\mathrm{\text{max}}}}\frac{1}{C}+\frac{1}{I_{\mathrm{\text{max}}}}$$(4) with I = steady state current (µA), I_max = maximum current measured (µA), and C = concentration of analyte (mM) [32]. In this case, the analyte is glucose. Based on the plotted Lineweaver-Burk curve (Fig. S9), the Michaelis-Menten constant (K_m) obtained was 16.44 mM. The K_m value is often used to assess enzymatic activity and affinity with the substrate. It implies the glucose concentration needed to reach half the maximum current response. Biosensors are considered to have a good linear response and sensitivity when the linear range does not exceed the K_m value [33]. In this work, the calculated K_m is higher than the linear range, supporting the chosen linear range’s suitability for generating an excellent biosensor response. By satisfying the Lineweaver-Burk curve, it also suggests that the reaction is diffusion-controlled [34].

Reproducibility, repeatability, and storage stability

Biosensor reproducibility was assessed using chronoamperometry of five biosensors in 5 mM glucose (0.1 M PB_(aq) pH 7) (Figure 9), resulting in mean current response in 7.42 µA with a relative standard deviation (RSD) of ±2.36% relative to the mean response. The acquired RSD is relatively lower compared to other works by Ang, Anusha, Bi, and Lović (Table 1), suggesting good biosensor reproducibility [15, 18,19, 35].

Figure 9. Biosensors current response towards 5 mM glucose in 0.1 M PB_(aq) pH 7 (E = +0.77 V vs Ag/AgCl) for 5 biosensors.

Table 1. Electrochemical performance comparison with published works regarding glucose oxidase-based glucose sensing.

Author	Transducer	Linear Range (mM)	Limit of Detection (μM)	Sensitivity	Km (mM)	Reproducibility (RSD)	Repeatability (RSD)
[28]	GR/AuND/SAM-GOx-glut	0.1–10	19	–	40.42 ± 5.72	–	–
[28]	GR/AuND/GOx-glut	0.1–10	22	–	25.22 ± 2.64	–	–
[14]	Pt/CS-glut-GOx	0.01–15	50	0.0597 μA/mM	14.235	3.70% (n = 6)	2.3% (n = 20)
[12]	Au-cys-glut-GOx	1.5–7	940	2.65 μA/mM cm^2	1.15	5% (n = 6)	2.5% (n = 6)
[29]	Au/CS/GOx	0.001–1	1.5	146.2 μA/mM cm^2	0.73	3.28% (n = 8)	1.4% (n = 5)
[30]	CC/AuND-SAM-GOx (EPC)	0.02–31.7	6.7	72.45 μA/mM cm^2	–	3.73% (n = 10)	–
[31]	MWCNT/GOx/Nafion	0.001–0.5	1.3 ± 0.1	12.1 uA/mM	9.8	2.1% (n = 7)	–
[28]	PLA/PEG/SPCE/PB/GOx	1.0–6.0	197	0.024 A/(M cm^2)	–	3.2% (n = 3)	3.1% (n = 1)
This work	SPCE/AuNS/cys/glut/GOx	0.2–15	116	24.41 μA/mM cm^2	16.44	2.36% (n = 5)	2.49%(n = 5)

GR: Graphite rod; AuND: gold nanodendrite; SAM: self-assembled monolayer; GOx: glucose oxidase; glut: glutaraldehyde; Pt: platinum electrode; CS: chitosan; Au: gold electrodes; cys: L-cysteine; CC: Carbon Cloth; EPC: Enzyme Precipitation Coating; MWCNT: multiwall carbon nanotube; PLA: polylactic acid; PEG: polyethylene glycol; PB: Prussian Blue.

Biosensor repeatability was evaluated by plotting repeated chronoamperometry measurements of each biosensor (n = 2) in 5 mM glucose (0.1 M PB_(aq) pH 7) (Figure 10(a)). For the first 5 times measurements, the calculated mean current response was 8.16 µA with an RSD of ±2.49%, slightly higher than that observed in Anusha’s works, but comparable with Lović’s works for almost the nearly equivalent number of measurements (Table 1) [15, 18]. To provide further evidence, measurements were continued until a total of 30 repetitions were reached. As the measurements were repeated, the mean current response gradually decreased while the RSD increased. At the 30^th measurement, the mean current response was 7.60 µA with an RSD of ±4.41%, resulting in an acceptable RSD value for a good biosensor repeatability.

Figure 10. Biosensors current response towards 5 mM glucose in 0.1 M PB_(aq) pH 7 (E = +0.77 V vs Ag/AgCl) (a) for 30 times repetitive measurements (n = 2) and (b) through 27 days storage (n = 2).

The storage stability of the biosensors was assessed through chronoamperometry measurements conducted with two biosensors in 5 mM glucose (0.1 M PB_(aq) pH 7), from day 1 to day 27. The mean measured current, relative to the initial measurement taken on the first day, was plotted as a function of time (Figure 10(b)). All biosensors were stored in a closed container at 4 °C when not in use. On the initial day, the current response averaged 7.43 ± 0.22 µA. Then, there was a 7.69% decrease in the current response by the end of the first week (day 6), followed by 19.27% decrease after 2 weeks (day 16), and a 26.72% decrease by day 27, compared to the initial response. The observed reduction in current responses over time can be attributed to two primary factors: the temperature difference between the storage condition (4 °C) and the experimental condition (25 °C) [19], and the impact of conducting repeated measurements with the same biosensor over time. This stability response closely resembles to Lović’s work, which uses L–cysteine SAM to immobilize GOx [15]. L-cysteine SAM has been proven to improve biosensor storage stability, with greater stability relatively compared to works by Ang, Anusha, and Sakalauskiene, who immobilized GOx covalently without the use of SAM molecule [18,19,36].

Selectivity

Biosensor selectivity was evaluated by comparing chronoamperometry measurements of two biosensors in the presence of 5 mM (90 mg/dL) glucose (in 0.1 M PB_(aq) pH 7) and other possible interference agents present in blood, including 0.45 mM (7.5 mg/dL) uric acid, 3.33 mM (20 mg/dL) urea, 85.17 µM (1.5 mg/dL) ascorbic acid, 6.53 µM (0.1 mg/dL) dopamine, 1.2 mM (10.8 mg/dL) lactic acid, and combined analytes of all mentioned interference agents [37–39] with all of them were also diluted in 0.1 M PB_(aq) pH 7. The result demonstrates excellent selectivity, with current response relative to all analytes remaining below 30% of the response to glucose (Figure 11). This great selectivity performance is also supported by the specific reaction of GOx with D-glucose [25].

Figure 11. Comparison of biosensor current response (n = 2) of 0.45 mM (7.5 mg/dL) uric acid (UA), 3.33 mM (20 mg/dL) urea, 85.17 µM (1.5 mg/dL) ascorbic acid (AA), 6.53 µM (0.1 mg/dL) dopamine, 1.2 mM (10.8 mg/dL) lactic acid (LA), and combined analytes of all mentioned interference agents with 5 mM (90 mg/dL) glucose in 0.1 M PB_(aq) pH 7 (E = +0.77 V vs Ag/AgCl).

To mimic real sample condition, we added 5 mM glucose in synthetic urine, both in its undiluted form and 10x dilution with 0.1 M PB_(aq) pH 7, and performed chronoamperometry measurements (Figure 12). With ion Cl^- as possible interference that decreases the Au active site, the concentration of synthetic urine affects the measurements result significantly [40], resulting in 40% decrease of response in 0.1x synthetic urine and 50% decrease of response in pure synthetic urine). Sodium azide (NaN₃) as preservative contained in the synthetic urine also may inhibited GOx enzyme activity resulting in decrease response of biosensor [41]. To enhance the biosensor’s adaptability to synthetic sample conditions, further adjustment to the calibration plot may be needed and further evaluation using human samples should be considered.

Figure 12. Comparison of biosensor current response (n = 2) of 5 mM glucose in 0.1 M PB_(aq) pH 7, in 10x diluted synthetic urine with 0.1 M PB_(aq) pH 7, and in pure synthetic urine (E = +0.77 V vs Ag/AgCl).

Conclusion

Our work has demonstrated the great potential of AuNS-deposited SPCE as a label-based biosensor through the immobilization of GOx enzyme using SAM cysteine. Optimal biosensor performance was achieved when the drop-casted reagents on the SPCE surface were completely dried. The developed biosensor has an electrochemical surface area (ECSA) for AuNS-deposited SPCE of 0.713 cm² with a roughness factor of 17.18. Using amperometry, biosensor performance characteristics towards glucose detection in 0.1 M PB_(aq) pH 7 were obtained with a linear range of 0.2–15 mM, limit of detection (LOD) of 116 µM, sensitivity of 24.41 µA/(mM.cm²), and Michaelis-Menten constant (K_m) of 16.44 mM. Our biosensor exhibits good performance in terms of reproducibility (RSD 2.36% for n = 5), repeatability (RSD 4.42% for 30 times measurement), storage stability (retained 73.28% of initial response after 27 days), good selectivity against uric acid, urea, ascorbic acid, dopamine, and lactic acid (all were diluted in 0.1 M PB_(aq) pH 7), and superiority towards AuNP-modified biosensor.

Our developed biosensor showcased a great foundation for future improvement of AuNS-deposited SPCE as a label-based biosensor. Despite showcasing excellent repeatability, the gradual reduction observed in the long-term stability study indicated that repeated measurement for the same biosensor over time might influence the biosensor’s stability performance, suggesting the need for further stability studies. Moreover, this study identified opportunities to improve performance in synthetic samples, laying the groundwork for future exploration using human samples. This raises the possibility of further studies that cover unrepeated biosensor stability performance, optimization using synthetic samples and human samples, and glutaraldehyde optimization as an immobilization parameter. In the future, we hope that these immobilization parameters and methods could be applied in the development of label-based biosensor for other bioreceptors as well.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported through financial grants from the Indonesia Endowment Fund for Education (LPDP), Ministry of Finance of the Republic of Indonesia under the RISPRO Invitasi program with grant number PRJ-81/LPDP/2019.

Notes on contributors

Salma Majidah

Salma Majidah received B.Sc. in Biomedical Engineering degree from Bandung Institute of Technology, Indonesia, in 2023. Her research interests include electrochemical biosensors and bioelectronics.

Lavita Nuraviana Rizalputri

Lavita Nuraviana Rizalputri received her MSc in Nanotechnology from Bandung Institute of Technology, Indonesia, in 2022, and before, she received her B.Sc. in Biomedical Engineering from the same university in 2019. Her current research focuses on the electrochemical biosensing and the lab-on-chip systems in healthcare field.

Eduardus Ariasena

Eduardus Ariasena received B.Sc. in Biomedical Engineering degree from Bandung Institute of Technology, Indonesia, in 2022. His research interests include electrochemical sensors, bioelectronics, and nanomedicine-related topics.

Aldyla Nisa Raditya

Aldyla Nisa Raditya received her PhD in Bioengineering from Imperial College London, UK, in 2022 as a member of Professor Danny O’Hare Group at Biosensors Lab. She is a microbiologist-turned-biomedical engineer who received her B.Sc. in Microbiology from Bandung Institute of Technology, Indonesia, in 2011. She also received two masters in 2013 and 2015: MSc in Sustainable Energy Technologies and Management from Brunel University London, UK, and MSc in Environmental Engineering and Business Management from Imperial College London, UK. Her current research interests broadly lies within the implementation of bioengineering, biomedical science, and precision medicine in healthcare, particularly in the domain of cardiology and infectious diseases. This includes the practical applications of electrochemical biosensors, lab-on-a-chip platforms, and wearables.

Bejo Ropii

Bejo Ropii received his B.Sc. in Biology and M.Sc. in Biomedical Engineering from Bandung Institute of Technology, Indonesia. His research interests include molecular biology, immunology, and cancer-related topics.

Nadia Salsabila

Nadia Salsabila received the bachelor’s degree in chemical engineering from the University of Indonesia, Indonesia, in 2020. She is currently pursuing master’s degree in Nanotechnology at Bandung Institute of Technology, Indonesia. Her research interests include electrochemical biosensors and nanomedicines.

Uperianti

Uperianti received B.Sc. in Biomedical Engineering degree from Bandung Institute of Technology, Indonesia, in 2023. Her research interests include electrochemical sensors, nanoscience, and nanotechnology-related topics.

Murni Handayani

Murni Handayani received the B.Sc. degree from the Chemistry Department, Sebelas Maret University, Indonesia, in 2003, and the M.Sc. and Ph.D. degrees from the Chemistry Department, Osaka University, Japan, in 2013 and 2016, respectively, under supervision of Prof. Takuji Ogawa. After her doctoral graduation, in 2017, she was a Specially Assigned Assistant Professor at the Chemistry Department, Osaka University. From 2005 to 2021, she was a Researcher at the Research Center for Metallurgy and Materials, Indonesian Institute of Sciences (LIPI). In 2021, the Indonesian Institute of Sciences merged into the National Research and Innovation Agency (BRIN). She is currently a Researcher at the Research Center for Nanotechnology Systems, National Research and Innovation Agency (BRIN). Her current research interests include nanoscience, advanced materials synthesis, molecular architectonics, nanocomposites, and nanomaterials, especially nanocarbons (graphene and carbon nanotubes).

Yeni Wahyuni Hartati

Yeni Wahyuni Hartati is an Associate Professor in Analytical Chemistry, Department of Chemistry, Padjadjaran University, Indonesia. She received her MSc degree in Biochemistry from Bandung Institute of Technology, Indonesia, and finished her doctoral degree in Analytical Chemistry from the Faculty of Mathematics and Natural Sciences, Padjadjaran University, Indonesia, in 2009. Her research topics include the study and development of electrochemical sensors and biosensors, as well as focusing on the use of nanomaterial modification such as gold nanoparticles, nanoceria, and graphene, application of experimental design, and various biosensing elements.

Isa Anshori

Isa Anshori received the B.Eng. degree from the Engineering Physics Department, Bandung Institute of Technology, Indonesia, in 2009, and the M.Eng. degree in materials science and the Ph.D. degree in nanoscience and nanotechnology from the University of Tsukuba, Japan, in 2015 and 2018, respectively. He has been working as an Assistant Professor with the Biomedical Engineering Department, Bandung Institute of Technology, since 2018. His current research interests include bio/chemical sensors, microfluidics, IoT devices, and lab-on-chip.