Sodium chloride (NaCl) was obtained from CDH (New Delhi, India). Sodium dihydrogen phosphate-1-hydrate (Na₂HPO₄.H₂O) and monosodium phosphite (NaH₂PO₃) were acquired from Merck (India). Sodium bicarbonate (NaHCO₃), sodium carbonate anhydrous (Na₂CO₃), potassium dihydrogen orthophosphate (KH₂PO₄), magnesium chloride (MgCl₂), sodium citrate tribasic dehydrate (C₆H₅Na₃O₇.2H₂O), potassium ferri- and ferro-cyanide (K₃Fe(CN)₆ and K₄Fe(CN)₆·3H₂O) were purchased from SRL (India). 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC), N-hydroxysuccinimide (NHS), and graphene oxide (GO) were procured from Sigma (India). β-Site amyloid precursor protein-cleaving enzyme 1 (BACE1) antigen (cat: 931-AS-050) and antibody (cat: MAB9311) were acquired from R&D Systems (Minnesota, United States). Human neurofilament (NFL) antigen (cat: ab224840) was purchased from Abcam (Massachusetts, United States). Chemicals, reagents, and solvents used in this research were of analytically graded high quality.

Raman spectroscopy and Fourier-transform infrared (FT-IR) spectroscopy were performed on Thermo Scientific-Nicolet 6700 Raman Spectroscope and iS50 FT-IR (Bangalore, India), respectively. Surface morphology was visualized via scanning electron microscopy (SEM), and elemental composition was analyzed via energy dispersive X-ray (EDX) on ZEISS EVO SEM coupled with SmartSEM software (Germany). Cyclic and differential pulse voltametric experiments were carried out on PalmSens4 potentiostat from PalmSens (The Netherlands).

rGO was synthesized by removing oxygen from GO (Habte et al., 2019; Roberts et al., 2022). Here, the reducing agent used was ascorbic acid (AA). First, 4 g AA was added to 100 μg/ml GO under constant stirring at 60°C and later centrifuged at 12,000 rpm. The remaining AA was oxidized by adding excess 30 wt% H₂O₂ to the black paste, which was constantly stirred at 60°C, and then centrifuged at 12,000 rpm. Further washing with 70% ethanol followed by distilled water was carried out three times each, and dried for 24 h at 120°C. Using carbodiimide chemistry, the resultant rGO powder after sonication in 0.05 M phosphate buffer (PB, pH 7.5) was labeled with BACE1 Ab. A mixture of equimolar ratio (75 µM) of EDC and NHS was added to 100 μg/ml rGO and incubated for 2 h at room temperature to activate the carboxyl groups on rGO via carbodiimide chemistry, which can then react with the amine groups present on the Ab for immobilization forming a covalent bond. This was followed by adding 90 µg of BACE1 Ab dropwise to the activated rGO, incubating at 4°C overnight, and then adding BACE1 Ag. Each synthesis, conjugation, and capture step were thoroughly characterized using Raman spectroscopy, FTIR spectroscopy, EDX, and SEM.

In this work, 100 μl of activated rGO was allowed to air dry onto the conductive face of an FTO slide at room temperature, followed by carbodiimide immobilization of BACE1 Ab (100 μl). A platinum counter and silver/silver chloride reference electrode were used to carry out the electrochemical experiments for cyclic voltammetry and differential pulse voltammetry. The different stages of fabrication and testing of the modified electrode were confirmed by CV. Different parameters including nanomaterial/Ab concentration, response time, pH, temperature, and rate of scan were standardized for maximum current output via CV/DPV. The limit of detection (LOD) was determined form the calibration curve by testing BACE1 Ag samples ranging from 1 fM to 1 µM in 0.05 M PB and spiked serum. 0.1 ml PB/serum samples spiked with various BACE1 Ag concentrations were added to the electrochemical 3-electrode cell containing redox potassium ferri-/ferro-cyanide buffer containing the immersed electrodes. Furthermore, 4-week storage stability as well as 6-cycle repeatability was also evaluated. The specificity was investigated via cross-reactivity of the FTO/rGO/BACE1Ab electrode against the NFL protein at 1 μM concentration in the buffer as well as spiked serum and 1X artificial CSF samples. 10X artificial CSF was made by adding 1.25 M NaCl, 260 mM NaHCO₃, 12.5 mM NaH₂PO₃, 25 mM KCl, and 10 mM MgCl₂ to 1 L double-distilled water. After use, the electrodes were regenerated by rinsing thoroughly with water and storing in a dust-free environment until coating for the next experiment.

The schematic and working principle of the fabricated electrode proposed in this research work is based on electrochemical detection of current variation upon interaction of antibody and antigen, as shown in Figure 1. GO (Figure 1A) was reduced and activated rGO (Figure 1B) was drop cast onto FTO electrodes (Figure 1C). The BACE 1 Ab bioreceptor was immobilized onto the rGO (Figure 1D) using EDC-NHS carbodiimide chemistry, where EDC activated the carboxyl group on rGO by converting the less-reactive carboxyl group into unstable O-acyl-urea derivative, which then reacted with the amine group of NHS to form NHS ester of rGO, which is more stable than the O-acyl-derivative. The NHS ester was then allowed to react with the primary amine group present on BACE1 Ab, resulting in an amide bond between rGO and BACE1 Ab. The fabricated immunosensor was then used to detect BACE 1 Ag (Figure 1E) via specific Ag–Ab interaction (Figure 1F) via electrochemical analysis (Figure 1G). Hence, the addition of BACE 1 Ag on the FTO/rGO/BACE1Ab-fabricated sensor produced a change in conductivity by inducing variation in the redox potential which was detected by a potentiostat.

In our work, 1 mg/ml rGO powder, after being synthesized from GO, was suspended to form a uniform black solution upon sonication in 50 mMPB. The FT-IR spectra of GO and rGO are shown in Figure 2A, where defined peaks at 1,630 cm⁻¹ (C = C stretching) and 1,070 cm⁻¹ (C-O stretching) were observed for GO, whereas only C=C stretching at 1,640 cm⁻¹ was seen for rGO due to removal of oxygen groups upon reduction of GO to rGO. rGO synthesis was further observed via Raman spectra (Figure 2B), where D-band and G-band appeared at a very similar peak height in case of rGO, but G-band appeared higher than D-band by 10 a.u. intensity in case of GO. The reduction in oxygen upon synthesis of rGO from GO using ascorbic acid was further corroborated by comparing EDX graphs, where reduction in the oxygen content was from 50.7 wt% in the case of GO (Figure 2Ci) to 20.3 wt% in the case of rGO (Figure 2Cii), since rGO contains less oxygen than GO. The morphological studies of the fabrication and working of the electrode were analyzed using SEM. In the corresponding scanning electron micrographs, bare rGO appeared as flakes (Figure 2Di), immobilized BACE1 Ab was observed as white globular structures on the rGO surface (Figure 2Dii), and BACE1 Ag layer captured by the specific Ab was observed (Figure 2Diii).

The electrochemical parameters for the developed electrode were optimized for efficient performance giving maximum current output. CV depends on the redox reaction, while DPV depends on oxidation reaction. As depicted in Figure 3A, DPV of the bare FTO shows a peak current at 3,567 μA, which increased due to oxidation. When the experiment was repeated on a rGO-immobilized FTO electrode, the corresponding peak current increased by 7,129 μA, which corresponds to the conductivity of rGO-enabled increase in surface electron and increased surface area. Following the conjugation of Ab to rGO using EDC-NHS chemistry, a 2,150 μA reduction in peak current was observed, relative to rGO-immobilized electrode. A further 2,662 μA decrease was observed after the attachment of Ag to the Ab. All these are most likely due to the blocking effect of the electron transfer by the proteins^. As shown in Figure 3B, among different response times (5-30 s) tested for stable output, 5 s was determined as the least time-point required for a stable response. Since sample preparation/incubation is often required in other BACE1 protein detection techniques, the proposed sensor in comparison gave a stable rapid result consuming less time. The electrode performance was checked within a range of rate of scan from 0.01 to 0.1 V/s (Figure 3C) and reduced current output was seen with reduction in the rate of scan, which was also depicted by a linear regression of peak current vs. scan rate (Figure 3D). As the peak current decreased with scan rate, 0.1 V/s was optimized as the most efficient rate of scan for electrode testing.

By studying the effect of rGO concentration (1, 0.5, 0.24, 0.1 mg/ml) used to prepare the rGO-immobilized electrode, the highest peak current for the oxidation of ferrocyanide was measured at 0.5 mg/ml rGO, as depicted in Figure 4A, since each FTO binding point became saturated at this concentration. Also, the ideal Ab amount to be immobilized on rGO was chosen from varying concentrations (1.5, 1, 0.5, 0.25 µg), and 1 µg resulted in the highest peak current (Figure 4B), since beyond this concentration, all rGO binding sites became saturated with no current increase. While testing environmental parameters, the maximum current output was recorded at pH 7.5 (Figure 4C) and room temperature (RT) (Figure 4D), when the electrode was checked at varying pH (6.0, 6.5, 7.0, 7.5, 8.0) and temperature (4°C, room temperature, 37°C, 45°C), since the immobilized Ab showed maximum activity and did not degrade under these optimum conditions.

DPV was plotted for the detection of BACE1 Ag, and in Figure 5, it was observed that with the increase in concentration of Ag, the current output reduced as added layers of protein result in masking the effect that reduces surface electron transfer. Figure 5Ai shows differential pulse voltammograms of the electrodes tested with increasing concentration of BACE1 Ag in buffer samples from 1 fM to 1 µM. The peak current of the electrodes was observed to decrease as a function of BACE1 Ag concentration. Figure 5Aii shows a plot of the peak current/blank current in Figure 5Ai vs. log of BACE1 Ag concentration in buffer samples, which can be represented by the linear equation shown in Figure 5Aii, where the slope, ordinate intercept, and r ² are −0.05434 ± 0.002333, 0.6210 ± 0.01245, and 0.9509, respectively. The p value of the calibration graph was significant (p value < 0.0001), and hence the regression line was linear. Similarly, Figure 5Bi shows the differential pulse voltammograms of the electrodes tested with increasing concentration of BACE1 Ag spiked in serum samples from 1 fM to 1 µM to verify changes in electrode functioning due to matrix effect. The peak current of the electrodes was again observed to decrease as a function of BACE1 Ag concentration. Figure 5Bii shows a plot of the peak current/blank current in Figure 5Bi vs. log of BACE1 Ag concentration spiked in serum samples, which can be represented by the linear equation shown in Figure 5Bii, where the slope, ordinate intercept, and r ² are −0.01029 ± 0.0002079, 0.9031 ± 0.001110, and 0.9887, respectively. The p value of the calibration graph was significant (p value < 0.0001), and hence the regression line was linear. The LOD in the case of buffer samples was calculated to be 0.64 fM (Figure 5Aii) and 1 fM (Figure 5Bii) in spiked serum. This slight decrease in sensitivity in serum samples is due to the matrix effect caused by other components such as different proteins present in serum, in addition to the target antigen. The formula used to calculate the LOD was 3(S_y/S), where S_y is the standard deviation of response, S is slope, and signal to noise ratio is 3. Since the testing was repeated on multiple electrodes and the average was plotted, the fabricated electrode also showed high reproducibility. Table 1 shows other developed electrochemical immunosensors for the detection of AD, and it can be observed that the LOD of the immunosensor developed in this research work is the lowest, making it the most sensitive electrochemical detection method reported till date.

Specificity of the fabricated electrode was determined through cross-reactivity experiments to study any non-specific binding of another neuronal Ag. While a reduction in conductivity was observed with BACE1 Ag present in buffer, serum, and artificial CSF samples, no such change was seen on testing with NFL Ag present in buffer, serum, and artificial CSF samples, which showed similar output as the blank samples (non-spiked buffer/serum/artificial CSF), when both Ag was spiked at a maximum concentration of 1 μM (Figure 6).

Furthermore, storage stability and repeatability of the developed sensor were determined. In Figure 7A, the proposed sensor showed a stable output upon storage for 1 month when tested at 7-day intervals, with a negligible decrease of 401 μA and 225 μA observed in the 3rd and the 4th week, respectively. This showed that the sensor could be kept in a fridge at 4°C for up to a month without affecting the functioning of the electrode as there is no degradation or reduction in the activity of the immobilized Ab. Multiple readings up to 6 times were taken on each individual electrode (Figure 7B), and a decrease of 1839 μA was seen at the fifth trial; hence, a single electrode may be reused 4 times without any compromise in results since the washing of the electrode after every trial may reduce the concentration and activity of the coated rGO and Ab.