Screen-Printed Carbon Electrodes (Ref. 110) were purchased from Metrohm DropSens. Hydrophilic Dynabeads™ with carboxylic acid groups (MBs, Ø = 2.8 µm, Ref. 14305D) were purchased from Invitrogen-Thermo Fisher. N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide (EDC), N-hydroxysulfosuccinimide (NHSS), ethanolamine (ETA), hydroquinone (HQ), and hydrogen peroxide (H₂O₂) (30%, v/v) were purchased from Sigma-Aldrich. Blocking Buffer (BB) Pierce™ Protein-Free (Ref. 37572) was purchased from Thermo Scientific. Human β-Amyloid Peptide (1–42) (Ref. 932502), Purified anti-β-Amyloid 1–42 Antibody (Ref. 805501, used as the capture antibody, CAb) and HRP anti-β-Amyloid 1–16 Antibody (Ref. 803012, used as the detection antibody, DAb-HRP) were purchased from BioLegend Inc. All solutions were prepared with ultrapure water from a Millipore Milli-Q water purification system (18.2 MΩ cm). Phosphate-buffered saline (PBS) (NaCl 137 mmol L⁻¹ and KCl 2.7 mmol L⁻¹) pH 7.4 and pH 8.0; phosphate buffer 0.05 mol L⁻¹, pH 6.0; phosphate buffer 0.1 mol L⁻¹, pH 8.0; MES buffer 25 mmol L⁻¹, pH 5.0 and Tris-HCl buffer 0.1 mol L⁻¹, pH 7.4. All reagents were of analytical grade and were used without further purification.

In the process of constructing sensing interfaces on magnetic microparticles, an Agimaxx Thermoshaker, model AG-100R, was utilized for the incubation steps, while a Gehaka vortex, model AV-2, was employed to solubilize the dispersions. Magnetic separation during the incubation and washing processes of the magnetic microparticles was carried out using a DynaMag™-2 magnet (Invitrogen-Thermo Fisher). The capture of modified magnetic microparticles onto the surface of the working electrode (SPCE) was achieved using a homemade polymethylmethacrylate (PMMA)-coated holding block with an encapsulated neodymium magnet (AIMAN GZ), which holds the SPCE electrode.

The chronoamperometry measurements were conducted using a PGSTAT 302N potentiostat/galvanostat from Autolab^® (Eco Chemie, Netherlands) connected to a microcomputer running NOVA software (version 2.1.3 - Metrohm Autolab B.V., Netherlands). All electrochemical measurements were carried out at room temperature in a 15 mL electrochemical cell with a SPCE working electrode, Ø = 5 mm.

The modification of magnetic beads (MBs) was characterized before and after each step of immunosensor construction. Dynamic Light Scattering (DLS) was utilized to assess the hydrodynamic diameter and distribution of the samples in suspension. The measurements were conducted in triplicate at room temperature using a Malvern spectrometer Nano-ZS (Malvern Instruments), model ZEN3601. Additionally, the zeta potential of the MBs after each functionalization step was evaluated using the same DLS equipment. The results are reported as mean ± standard deviation (SD) based on three independent measurements.

Transmission Electron Microscopy (TEM) images were obtained using JEOL JEM-100CXII equipment.

Infrared spectra were recorded using a Nicolet IS20 FTIR spectrometer (Thermofisher, United States) with a nitrogen gas purge. The spectral resolution was set at 4.0 cm⁻¹, and they were recorded with 64 scans in transmission mode, covering the wavenumber range of 500–4,000 cm⁻¹. Samples (5 µL) were placed on a silicon wafer (Si/SiO₂/B) and dried under vacuum conditions. The measurements were conducted at room temperature immediately after drying, with constant purging and a corrected background. The overall spectra were normalized.

Raman spectra were recorded using the Renishaw micro-Raman spectroscopy system in the 500–1800 cm⁻¹ range (static mode) with a 532 nm laser excitation wavelength. The exposure time was set to 1 s, and 50 scans were accumulated (total exposure time of 50 s). The incident laser light was adjusted to 1% power to avoid excessive heating and fluorescence of the sample. Each sample (5 µL) was placed on a silicon wafer (Si/SiO₂/B) and dried under vacuum conditions. Raman spectra were collected at room temperature after drying the sample and calibrated using the silicon peak. Data processing included background subtraction and eleven-order polynomial fitting for fluorescence background removal, followed by normalization.

The amperometric immunosensor was constructed using magnetic beads (MBs) as a platform to immobilize monoclonal antibodies selective for Aβ1-42. The detection antibody is conjugated to the enzyme horseradish peroxidase (HRP), which reacts with hydrogen peroxide (H₂O₂). Thus, amperometric transduction was performed on screen-printed carbon electrodes (SPCE) using the hydroquinone (HQ)/H₂O₂ system.

All incubation steps were performed in a thermoshaker operating at 950 rpm and 25°C. After each incubation, the microtubes were placed in a magnetic separator for 4 min, and then the supernatant was discarded. Initially, a 5 μL aliquot of hydrophilic beads with carboxylic acid groups was placed in a 1.5 mL microtube. Then, 50 μL of MES buffer was added and incubated for 5 min; the supernatant was removed, and the washing step was repeated. In the second step, the carboxyl groups on the surface of the magnetic beads (−COOH-MBs) were activated with EDC/NHSS by incubating in 25 μL of a 100 mmol L⁻¹ EDC/NHSS solution prepared with MES buffer for 35 min. The activated −COOH-MBs were washed twice with 50 μL of MES buffer. For covalent immobilization of capture antibodies (CAb), the activated MBs were resuspended in 25 μL of a 10 μg mL⁻¹ CAb solution prepared in MES buffer for 60 min. The CAb-modified MBs were washed twice with 50 μL of MES buffer. Any activated −COOH-MBs that did not react with CAb were blocked by incubation in 25 μL of a 50 mmol L⁻¹ ethanolamine solution prepared in PBS (pH 8.0) for 60 min. The CAb-modified MBs were washed once with 50 μL of Tris-HCl buffer and twice with 50 μL of commercial blocking buffer solution (BB).

The detection protocol consists of incubating the MBs modified with capture antibodies (CAbs) in 25 μL of solutions of varying concentrations of Aβ_1-42 and 0.5 μg mL⁻¹ of the detection antibody labeled with the enzyme horseradish peroxidase (HRP) (DAb-HRP), prepared in blocking buffer (BB), for 60 min. For amperometric detection, the MBs modified with the immunocomplexes were washed twice with 50 μL of BB and then resuspended in 50 μL of the electrolyte, which is a 50 mmol L⁻¹ sodium phosphate buffer (pH 6.0).

The SPCEs were embedded in the PMMA block, and 50 µL of the suspension of MBs modified with the immunocomplex was magnetically captured by the magnet embedded in the block in a simple, reproducible, and stable manner (Gamella et al., 2020). The block-SPCE assembly was then immersed in an electrochemical cell containing 10 mL of 50 mmol L⁻¹ phosphate buffer (pH 6.0), which contained 1.0 mmol L⁻¹ of hydroquinone (HQ) (freshly prepared) (Conzuelo et al., 2012a). Amperometric measurements were conducted in solutions under magnetic stirring by applying a detection potential of −200 mV (vs. the Ag pseudo-reference electrode) (Gamella et al., 2012). The current was recorded upon reaching a steady state before (B) and after (S) the addition of 50 μL of a 0.1 mol L⁻¹ hydrogen peroxide (H₂O₂) solution. The amperometric response of the immunosensor corresponds to the difference between the steady-state currents (S−B). The S−B values reported in this study are the mean of three measurements, and the error bars represent three times the standard deviation of the triplicates. Confidence intervals were calculated at a significance level of α = 0.05.

The size and surface charge of the magnetic beads (MBs) modified with the capture antibodies (MBs-COOH-CAb) and the immunocomplex (MBs-COOH-CAb@peptide-DAb-HRP) were investigated using dynamic light scattering, zeta potential measurements, and transmission electron microscopy (TEM) images, as shown in Figure 1 and Table 1.

As shown in Figures 1C, D and Table 1, the magnetic microparticles exhibited a mean size of 2.8 ± 0.1 µm (according to the manufacturer’s specifications) and a charge of −37 ± 1 mV without any pretreatment, as measured by dynamic light scattering (DLS) and zeta potential, respectively. The introduction of EDC/NHSS and capture antibody (CAb) molecules to the magnetic beads (MBs) affected the size, reducing it to 1.8 ± 0.1 µm, and the charge to −42 ± 2 mV, corresponding to a decrease of approximately 1.0 µm in size and 5 mV in charge. The surface charge and hydration layer are the main factors to contribute with this fact since MBs are often surrounded by a hydration layer or an electric double layer in a liquid medium. When antibodies are attached to the surface, the zeta potential of the NP changes, altering the structure or thickness of this layer, which might result in a smaller effective hydrodynamic diameter being measured, as observed in our DLS results that decrease from 2,793 ± 105 to 1832 ± 127 nm. Moreover, some NPs come with a stabilizing layer like polymers or surfactants that maintain their stability and prevent aggregation. During antibody conjugation, this layer may be partially or completely replaced by the antibody molecules, resulting in a net decrease in overall particle size. These changes were also reflected in the conductivity profile, which significantly increased from 0.011 to approximately 0.800 m/cm after the MBs’ surface modification and by an increase in polydispersity index (PdI) values from 0.202 to 0.503.

The biorecognition step was evidenced by a slight variation in the zeta potential values, changing from −42 ± 2 to −35 ± 1 mV due to the incorporation of @peptide-DAb-HRP onto the MBs-COOH-CAb, as well as an increase in size of approximately 0.3 µm. These changes were expected due to the presence of biomolecules on the protein/peptide surface contributing to the surface charge. Transmission electron microscopy (TEM) images further confirmed the modification steps (Figures 1A, B). The soft pattern observed around the magnetic microparticles is attributed to the presence of biological molecules on the surface.

The interaction of magnetic beads (MBs) with modifiers at each stage of the immunosensor construction was analyzed using Fourier-transform infrared spectroscopy (FTIR), as shown in Figure 2.

Initially, FTIR spectra were obtained for as-received MBs-COOH and compared with MBs-COOH washed with MES buffer. Figures 2–A displays the typical functional groups associated with the MBs-COOH surface structure for both samples. In the single bond region (4,000–2,500 cm⁻¹), O−H bending modes at 3,307 cm⁻¹ and C−H stretching with a broad band at 2,921 cm⁻¹ were observed. Figures 2–B shows the double bond region (2,000–1,500 cm⁻¹) associated with C−C and C=C stretching modes, as well as N−H bending from lipids, esters, amide I, and amide II bands (Miller et al., 2013; Burt et al., 2004). Notably, three main bands were observed at 1716, 1,605, and 1,523 cm⁻¹, corresponding to C−C stretching, N−H bending, and C=C stretching, respectively, from the MBs-COOH (Byun, 2016). Figures 2–C, D highlight the main differences in the spectra in the region from 1,500 to 500 cm⁻¹. Two primary bands were observed at 1,456 cm⁻¹ and 1,347 cm⁻¹, corresponding to the C−H₂ bending and C−N stretching modes, respectively. Additionally, bands at 1,263 and 1,225 cm⁻¹ may represent the SO₃ vibration and residues from cysteine at 1,077 cm⁻¹ (Miller et al., 2013; Burt et al., 2004; Byun, 2016).

The second step in the construction of the immunosensor consists of activating the −COOH groups on the MBs with EDC/NHSS to facilitate the covalent immobilization of the capture antibody. This activation was characterized by a decrease in the intensity of the band at 3,353 cm⁻¹, which is related to the N−H stretching mode. The decrease occurred due to the reaction of N−H groups with EDC/NHSS, forming secondary amine bonds in the same region as the O−H stretching band (Byun, 2016). The immobilization of the capture antibody (CAb) was evidenced by a decrease in the band intensity associated with primary amine stretching groups in the range of 1,500 to 1,050 cm⁻¹. This reduction is likely caused by the rearrangement of the amino acids in the monoclonal antibody CAb with the carboxylic acid C=O, which results in a decrease in the total dipole moment, as described in the literature (Moran and Zanni, 2014).

The blocking step with ethanolamine (ETA) was confirmed by an increase in the band intensities in the regions corresponding to O−H, N−H, and C−H stretching modes, attributed to the presence of amine and hydroxyl groups in the ETA molecule. Finally, the cyan spectrum in Figure 2 revealed the biorecognition step of the complex composed of the peptide bound to the detection antibody labeled with the HRP enzyme and captured by the capture antibody immobilized on the MBs. Studies from the literature have demonstrated the presence of native β-amyloid by the existence of primary amide bands and anti-parallel β-sheet structures between 1,630–1,643 cm⁻¹ and 1,670–1,695 cm⁻¹, respectively (Moran and Zanni, 2014; Zandomeneghi et al., 2004; Sarroukh et al., 2013; Boelens et al., 2022). The stretching band between 1,650 and 1,658 cm⁻¹ is indicative of an α-helical structure, while the band at 1,644 cm⁻¹ is related to random coil structures in the secondary protein structure. Consequently, the broad band around 1,636 cm⁻¹ may be associated with β-sheet proteins and primarily arises from the C=O stretching vibration with a minor contribution from C−N stretch (Miller et al., 2013; Sarroukh et al., 2013; Boelens et al., 2022; Breydo et al., 2016). In a β-sheet, the normal modes are delocalized both along and across the structure, and its frequency provides information on its size and rigidity (Moran and Zanni, 2014; Zandomeneghi et al., 2004; Sarroukh et al., 2013; Guivernau et al., 2016).

In addition to the FTIR studies, Raman spectroscopy was used to gather further information on the interaction of MBs with the modifiers at each stage of immunosensor construction, as illustrated in Figure 3.

The characteristic peaks from MBs-COOH are observed around 1,060 and 1,600 cm⁻¹, corresponding to aliphatic and aromatic C−C stretching vibrations from the carboxylic groups on the MBs’ surface, corroborating the FTIR results (van Apeldoorn et al., 2005). Additionally, the immunocomplex MBs-COOH-CAb@peptide-DAb-HRP exhibited two peaks in the same region, corresponding to the benzene ring breathing and C=C vibration associated with phenylalanine, an amino acid present in the protein complex (Imanbekova et al., 2021; Ishigaki et al., 2020). Confirming these findings, a peak at 1,235 cm⁻¹ was observed, corresponding to the stretching mode of C−C₆H₅ of phenylalanine (Ishigaki et al., 2020). The presence of CAb on the MB particles was confirmed by the peak at 1,688 cm⁻¹ (amide I), which can be assigned to C=O and a minor contribution of C−N stretching. These peaks increased in intensity after peptide biorecognition on the surface of the MBs-COOH-CAb (Boelens et al., 2022; Imanbekova et al., 2021; Ishigaki et al., 2020; Chen et al., 2015; Michael et al., 2014). The peptide backbone is also observed at 1,452 cm⁻¹, attributed to CH₃ deformation and a ternary amide belonging to the N−H bend (Ishigaki et al., 2020; Michael et al., 2014). This peak emerges with the presence of biomolecules CAb and Aβ_1-42 peptide. Furthermore, other peaks in the spectra were attributed to residues of specific amino acids, including a tyrosine doublet at 723 and 637 cm⁻¹, alanine at 847 cm⁻¹, proline at 950 cm⁻¹, and cysteine at 534 cm⁻¹ (Burt et al., 2004; Boelens et al., 2022; Imanbekova et al., 2021; Ishigaki et al., 2020). The cysteine residues in CAb must be free to interact and couple with the −COOH groups of MBs, consistent with the analysis of FTIR spectra (Burt et al., 2004; Boelens et al., 2022; Michael et al., 2014).

The experimental variables at each step of the sandwich amperometric immunosensor construction were optimized by considering the amperometric responses of both unmodified and CAb-modified MBs, in the absence (B) of the Aβ_1-42 standard and in the presence (S) of 5.0 ng mL⁻¹ of the Aβ_1-42 standard, along with the corresponding S/B ratio values. The same conditions and reagents were used for both S and B measurements to ensure that any variation in the S/B ratio was solely attributable to the presence or absence of Aβ_1-42. Figures 4–A demonstrates that there is no nonspecific adsorption of Aβ_1-42 and DAb-HRP in the absence of CAb, thus confirming the functionality and feasibility of the proposed sandwich immunoassay.

The crucial experimental variables for the preparation of the sandwich amperometric immunosensor were optimized. The comparison of the signal-to-blank ratio values (S/B) corresponding to the amperometric current measured in the absence (B) and presence (S) of 5.0 ng mL⁻¹ of Aβ_1-42 was used to determine the optimal variable values for each step. The amount of magnetic microbeads (5 µL) used aligns with that reported in the literature (Conzuelo et al., 2012b; Eguílaz et al., 2010). The use of 50 mM ethanolamine as a blocking solution follows the manufacturer’s protocol for Dynabeads^® M-270 Carboxylic Acid (MBs). According to the study by Eguiláz et al. (Eguílaz et al., 2010), the reduction potential for the decomposition of H₂O₂ in the HRP/H₂O₂/HQ system using a 0.05 mol L⁻¹ sodium phosphate buffer electrolyte at pH 6.0 should be −200 mV to achieve good sensitivity and accuracy in the amperometric response (Gamella et al., 2020; Conzuelo et al., 2012b; Eguílaz et al., 2010). The pH value of 6.0 (Conzuelo et al., 2012b) is consistent with the optimal pH range (6.0–6.5) for HRP enzyme activity (Ding et al., 2023). The tested ranges and selected values for each variable are summarized in Table 2, while Figure 4 illustrates the results.

In the development of an immunosensor, several important variables must be considered, including the consumption of immunoreagents, sensor preparation time, and sample analysis time. The goal is to minimize reagent consumption, shorten sensor preparation time, and achieve rapid analytical responses. In this study, a concentration of 25 μg mL⁻¹ for the capture antibody (CAb) was selected, along with an incubation time of 60 min, based on the optimal signal-to-blank (S/B) ratios that provide the best efficiency for the analysis.

The number of steps required for assembling the immunosensor was determined by analyzing the results of two work protocols, both starting from the modification stage of the magnetic beads (MBs) with the capture antibody (CAb). Protocol one involved a single incubation step in a mixed solution containing Aβ_1-42 and DAb-HRP for 60 min (1 step). Protocol 2, on the other hand, included two successive incubation steps: first, with the standard Aβ_1-42 solution for 60 min, followed by incubation with the detection antibody conjugated with horseradish peroxidase (DAb-HRP) for another 60 min (2 steps). Figures 4–D shows that the best signal-to-background (S/B) ratio was achieved with Protocol 1 (a single step), likely due to the higher efficiency of biorecognition events occurring in a homogeneous mixture of antigens and antibodies.

Figures 4–E shows the effect of the concentration of the HRP-labeled detection antibody (DAb-HRP) on the response of the immunosensor. The signal-to-background ratio (S/B) increased with rising DAb-HRP concentrations up to 0.5 μg mL⁻¹ but decreased at a concentration of 1.0 μg mL⁻¹. This decrease is attributed to the reduced specific response to Aβ_1-42 (S), indicating impaired molecular recognition, which may be caused by excessive amounts of DAb-HRP. As for the incubation time of the Aβ_1-42 + DAb-HRP mixture, an incubation period of 60 min was selected, as shown in Figures 4–F.

The calibration curve for Aβ_1-42 was constructed using the immunosensor prepared with 25 μg mL⁻¹ of the capture antibody (CAb) immobilized on the magnetic beads (MBs) (Figures 5–A).

Figure 5A shows that the response current of the immunosensor exhibits an increasing linear relationship with the logarithm of the Aβ_1-42 concentration, in a concentration range of 100 to 10,000 pg mL⁻¹ according to the equation: −Δi (nA) = (357.5 ± 5.2) + (342.8 ± 7.8) log[Aβ_1-42] (ng mL⁻¹). The limit of detection (LOD) was calculated according to formula 3 × Sb/m, where Sb represents the standard deviation of the arithmetic mean of the measurements of 10 blanks, and m is the slope value of the analytical curve. The immunosensor has an LOD of 7.4 pg mL⁻¹, a value consistent with the values reported in the literature (Table 3) for electrochemical immunosensors. When compared to the lower reference limit for Aβ_1-42 in cerebrospinal fluid (CSF) (192 pg mL⁻¹), the response of the immunosensor is significantly lower. Consequently, the immunosensor is suitable for the detection of Alzheimer’s disease, given that the onset of the pathology occurs when CSF-Aβ_1-42 drops below this threshold of 192 pg mL⁻¹ (Ding et al., 2023).

Selectivity is a crucial parameter for the application of the proposed immunosensor in clinical plasma or cerebrospinal fluid (CSF) samples. The selectivity of the immunosensor was evaluated against several proteins potentially found in plasma and CSF samples, including human IgGs, hemoglobin (Hb), human serum albumin (HSA), and β-amyloid peptide (1–40). The tests were conducted by measuring Aβ_1-42 concentrations of 0.0 and 2.5 ng mL⁻¹ both in the absence and in the presence of these interferents, at concentrations equivalent to those found in the serum of healthy individuals. The presence of these interferents did not affect the determination of Aβ_1-42, as shown in Figures 5–B.

The developed immunosensor was prepared in 4 h, without involving exhaustive synthesis procedures, demonstrating good reproducibility (RSD of 4.7%) and stability. Reproducibility measurements were performed for the concentration of 500 pg mL⁻¹ of Aβ_1-42 using 8 identically prepared immunosensors tested on the same day. For interday stability measurements, the immunosensors were prepared from CAb-MBs stored in filtered PBS at 4°C for at least 15 days, and no significant difference was observed in the signal-to-noise (S/B) ratio for measurements of 0.0 and 500 pg mL⁻¹ of Aβ_1-42.