Potentiometric Biosensor Based on Artiﬁcial Antibodies for an Alzheimer Biomarker Detection

: This paper presents a potentiometric biosensor for the detection of amyloid β -42 (A β -42) in point-of-care analysis. This approach is based on the molecular imprint polymer (MIP) technique, which uses covalently immobilised A β -42 to create speciﬁc detection cavities on the surface of single-walled carbon nanotubes (SWCNTs). The biosensor was prepared by binding A β -42 to the SWCNT surface and then imprinting it by adding acrylamide (monomer), N,N (cid:48) -methylene-bis-acrylamide (crosslinker) and ammonium persulphate (initiator). The target peptide was removed from the polymer matrix by the proteolytic action of an enzyme (proteinase K). The presence of imprinting sites was conﬁrmed by comparing a MIP-modiﬁed surface with a negative control (NIP) consisting of a similar material where the target molecule had been removed from the process. The ability of the sensing material to rebind A β -42 was demonstrated by incorporating the MIP material as an electroactive compound in a PVC/plasticiser mixture applied to a solid conductive support of graphite. All steps of the synthesis of the imprinted materials were followed by Raman spectroscopy and Fourier transform infrared spectroscopy (FTIR). The analytical performance was evaluated by potentiometric transduction, and the MIP material showed cationic slopes of 75 mV-decade − 1 in buffer pH 8.0 and a detection limit of 0.72 µ g/mL. Overall, potentiometric transduction conﬁrmed that the sensor can discriminate A β -42 in the presence of other biomolecules in the same solution.


Introduction
Several peptides/proteins are being considered as potential biomarkers for predicting the early stages of Alzheimer's disease (AD). Numerous clinical studies show that the concentrations of these biomarkers in biological fluids correlate with the progression of early-stage disease [1]. The mechanism driving the progression of AD is related to the formation of senile plaques and neurofibrillary tangles, which are responsible for the death of neurons and lead to memory loss and symptomatic behavioural changes [2][3][4][5]. Current hypotheses also include circulating Aβ-42 oligomers as potentially neurotoxic (not just the plaques) [6][7][8][9][10][11][12]. The detection of Aβ-42 is based on immunochemical reactions, e.g., conventional ELISAs [13], which are highly sensitive and selective, but time-consuming, unstable, and expensive. Several biosensors have been described in the literature for the detection of Aβ-42 by electrochemical [14,15], surface plasmon resonance (SPR) [16][17][18][19][20][21] and field-effect transistor Potentials were measured using a Crison pH meter GLP 21 (±0.1 mV sensitivity) at room temperature with constant stirring. The output signal was passed to a commutation unit and connected to one of the six outputs, which allowed for synchronized measurements. The setup of the potentiometric cell was as follows: epoxy-graphite|Aβ-42 selective membrane | buffered sample solution of 3-(N-morpholino)propanesulfonic acid (MOPS), 1 × 10 −3 mol/L, pH 8.0 | AgCl (s)|Ag. An Ag/AgCl electrode served as the reference electrode.
Infrared spectra were recorded using a Nicolet 6700 FTIR spectrometer coupled to an attenuated total reflectance (ATR) and equipped with a Nicolet diamond contact crystal. Raman spectroscopy data were obtained with a Thermo Scientific DXR Raman microscope equipped with a 532 nm laser. A laser power of 8 mW was left on the sample at a slit aperture of 50 µm.
The SEM analysis was performed using a high-resolution (Schottky) scanning electron microscope with X-ray microanalysis and electron backscattered diffraction analysis: Quanta 400 FEG ESEM/EDAX Genesis X4M.
Phosphate buffer solution (PBS) buffer was prepared with the following composition: NaCl, 8 g/L; KCl, 0.2 g/L, Na 2 HPO 4 , 1.42 g/L; KH 2 PO 4 , 0.24 g/L. The pH was adjusted to a value of 8.0. The standard solutions of Aβ-42 were prepared from a successive dilution of the stock solution (0.05 mg/mL) previously prepared in ultrapure water. The MOPS 1.0 × 10 −3 mol/L were prepared in deionized water.

SWCNT Oxidation
Briefly, 0.5 g of SWCNT was added to a ten-fold diluted solution of concentrated H 2 SO 4 :H 2 O 2 (4:1). Then, the solution was stirred overnight, centrifuged and washed three times with water. The material was then stirred in PBS buffer, pH 7.2, for 2 h to increase the pH. This procedure was repeated three times. Finally, the nanotubes were oven-dried overnight at 40 • C.

Imprinting Stage
Approximately 10 mg of oxidized SWCNT was suspended in 1.0 mL of deionized water by ultrasonically stirring the mixture for approximately 1 min. Then, 1.0 mL of an aqueous solution containing 5 mmol/L NHS and 2 mmol/L EDAC was added, and the mixture was stirred at room temperature for 45 min. The suspension was then filtered, and the solid was thoroughly rinsed with deionized water to remove unreacted reagents. After activation of the carboxylic acids by EDAC/NHS treatment, 1 mL of 0.01 µg/mL Aβ-42 prepared in MOPs buffer pH 8.0 was added to the modified material for 2.0 h at 4 • C to bind the template. The imprinting phase began with overnight incubation of the modified SWCNTs with 1 mL 5.0 × 10 −4 mol/L AEMA. The materials were then washed several times with deionized water. The polymerization phase starts with the addition of 0.5 mL of a monomer (AAM, 5.0 × 10 −4 mol/L) and a crosslinker (NNMBA, 1.0 × 10 −4 mol/L) to the SWCNTs previously modified with the peptide. After incubation at room temperature for 60 min, 0.5 mL of APS, 1 × 10 −4 mol/L and 0.5 mL of TEMED, 10%, solution in MOPS buffer pH 8.0 were added to start polymerization. The polymerization was carried out at room temperature for 2 h. After that, the sensor was washed thoroughly with deionized water. Finally, the template was removed by exposed to proteinase K (500 µg/mL) for 3 h at 37 • C. The imprinted sensor material was washed with phosphate buffer, pH 7.2, to remove the peptide fractions formed by the enzyme treatment, and dried in an oven at 40 • C.

Preparation of Sensory Membranes
Different compositions of PVC membranes were prepared to evaluate the potentiometric response of the MIP-based sensors. Approximately 0.05 g PVC, 0.06 g plasticizer (oNPOE) and 2.0 mg MIP and NIP were mixed (see Table 1). The mixture was stirred until complete dispersion of the PVC in 3.5 mL THF. These membranes were dropped onto the graphite electrode surface and dried for 48 h. Then, the membranes were conditioned in an Aβ-42, 0.01 µg/mL solution. The electrodes were stored in this solution when not in use.

Potentiometric Procedures
The analytical performance of the electrodes was evaluated using calibration curves according to IUPAC recommendations. Potentiometric measurements were performed at two pH values (6 and 8), at room temperature, with constant agitation and recording of potentials after stabilization of ±0.2 mV. The e.m.f. value was measured at constant values of ionic strength, and pH. MOPS solution was used for this purpose; aliquots of 0.4 to 15 mL of 1.0 × 10 −4 mol/L MOPS were transferred to 0.25 mL of Aβ-42 solution 0.01 µg/mL and 1.75 mL of buffer in a 100 mL beaker. Calibration data for each calibration included limit of detection (LOD), slope, linear ranges and squared correlation coefficient (R 2 ). Serum samples were analyzed by adding 100-fold diluted commercial human Cormay ® serum and an appropriate amount of Aβ-42 standard. The Aβ-42 concentration was calculated from the analytical data of a control calibration curve in serum.

Selectivity Study
The potentiometric selectivity coefficients were determined by mixed solutions method (MSM). The selectivity of the biosensor was assessed by quantifying LOD of each calibration curve through the presence of an interfering compound [62]. The selectivity coefficients were calculated according to this equation: In Equation (1), a J is the concentration of interfering species, Z the ionic charges of the interfering ions and a Aβ42 the intersection of the extrapolated linear section of the plot Emf vs. the logarithm of Aβ-42 concentration. Alb, Crea and urea were evaluated as interfering species. For this propose, a fixed concentration of interfering species within the physiological level in serum was added as background, while the concentration of primary ion (Aβ-42) was varied. (Crea, 1.2 × 10 −5 g/mL), (Alb, 5.2 × 10 −3 g/mL) and (urea, 4.0 × 10 −4 g/mL) were evaluated as interfering species.

Design of the Plastic Antibodies
Herein, a plastic antibody synthesized on the oxidized SWCNT surface was successfully used for Aβ-42 detection with potentiometric transduction. The SWCNTs were selected for this propose due to their large surface area, which promotes an increase in the number of binding sites of the MIP material. The overall process for preparing the MIP involved the attachment of Aβ-42 to the surface of the SWCNTs (Figure 1), filling the vacant places around the Aβ-42 with an appropriate rigid structure and removing the peptide from the polymer. The polymerization steps were carried out under mild conditions in order to guarantee that 3D-structure and electrostatic environment of Aβ-42 remained stable.

Design of the Plastic Antibodies
Herein, a plastic antibody synthesized on the oxidized SWCNT surface was successfully used for Aβ-42 detection with potentiometric transduction. The SWCNTs were selected for this propose due to their large surface area, which promotes an increase in the number of binding sites of the MIP material. The overall process for preparing the MIP involved the attachment of Aβ-42 to the surface of the SWCNTs (Figure 1), filling the vacant places around the Aβ-42 with an appropriate rigid structure and removing the peptide from the polymer. The polymerization steps were carried out under mild conditions in order to guarantee that 3D-structure and electrostatic environment of Aβ-42 remained stable. First, the peptide was covalently bound to the previously oxidized SWCNT. The carboxyl groups were activated by an EDAC and NHS reaction. In the first step, a highly reactive O-acylurea intermediate is formed [59], which reacts rapidly with NHS to form a more stable ester (succinimidyl intermediate). This ester performs a nucleophilic substitution with any accessible amine group on Aβ-42, leading to the formation of an amide bond between the SWCNTs and Aβ-42.
The formation of stable complexes between the monomers and the target molecule and their preservation in solution are crucial for achieving a suitable selectivity of the imprinting polymer. In this work, electrostatic interactions were established between the target and the monomers. For this purpose, 2-AEMA was used as the functional monomer, forming hydrogen bonds with Aβ-42 [63]. Polymerization then starts with the addition of the monomer AAM, the crosslinker NNMBA and the initiators APS and TEMED after peroxide (OO) bond cleavage by the initiator. Crosslinking between the NNMBA chains was ensured by two terminal vinyl groups. The modified SWCNTs were washed three times with the MOPS buffer [64]. The Aβ-42 was then removed from the polymer by proteinase K treatment. Several washing steps were then performed to remove the free peptide fractions from the surface of the biomimetic material. First, the peptide was covalently bound to the previously oxidized SWCNT. The carboxyl groups were activated by an EDAC and NHS reaction. In the first step, a highly reactive O-acylurea intermediate is formed [59], which reacts rapidly with NHS to form a more stable ester (succinimidyl intermediate). This ester performs a nucleophilic substitution with any accessible amine group on Aβ-42, leading to the formation of an amide bond between the SWCNTs and Aβ-42.
The formation of stable complexes between the monomers and the target molecule and their preservation in solution are crucial for achieving a suitable selectivity of the imprinting polymer. In this work, electrostatic interactions were established between the target and the monomers. For this purpose, 2-AEMA was used as the functional monomer, forming hydrogen bonds with Aβ-42 [63]. Polymerization then starts with the addition of the monomer AAM, the crosslinker NNMBA and the initiators APS and TEMED after peroxide (OO) bond cleavage by the initiator. Crosslinking between the NNMBA chains was ensured by two terminal vinyl groups. The modified SWCNTs were washed three times with the MOPS buffer [64]. The Aβ-42 was then removed from the polymer by proteinase K treatment. Several washing steps were then performed to remove the free peptide fractions from the surface of the biomimetic material.

Raman Spectroscopy Studies
The chemical profile of the different materials was evaluated by analysing the Raman spectra. The analysis was performed for the following materials: (i) SWCNT, (ii) SWCNT-COOH and (iii) MIP and NIP. RAMAN spectra give information about the efficiency of the nanotube carboxylation, polymerization and template removal [65].
In general, the structure of the spectrum highlights four distinctive peaks ( Figure 2): the first peak in the region around 1300 cm −1 (D), the second in the region around 1600 cm −1 (G), the third in the high frequencies around 2600 cm −1 (G ) and the fourth in the region around 1930 cm −1 . Figure 2 shows the Raman spectra for each sample separately, and the ratio of the intensity between G and D band, which is useful for determining the purity of and the degree of the chemical modification of the material. As a result, we observe a significant difference in terms of the peak intensity and shift between different materials. The shifts are essentially perceptible for the G and D bands.

Raman Spectroscopy Studies
The chemical profile of the different materials was evaluated by analysing the Raman spectra. The analysis was performed for the following materials: (i) SWCNT, (ii) SWCNT-COOH and (iii) MIP and NIP. RAMAN spectra give information about the efficiency of the nanotube carboxylation, polymerization and template removal [65].
In general, the structure of the spectrum highlights four distinctive peaks ( Figure 2): the first peak in the region around 1300 cm −1 (D), the second in the region around 1600 cm −1 (G), the third in the high frequencies around 2600 cm −1 (G') and the fourth in the region around 1930 cm −1 . Figure 2 shows the Raman spectra for each sample separately, and the ratio of the intensity between G and D band, which is useful for determining the purity of and the degree of the chemical modification of the material. As a result, we observe a significant difference in terms of the peak intensity and shift between different materials. The shifts are essentially perceptible for the G and D bands. Carbon nanotubes are rolled-up graphene sheets that have been wound into tubes. These nanomaterials are cylindrical tubes with a single outer wall and a diameter typically around 1-2 nm. Due to their exceptional mechanical, electrical and thermal assets, carbon nanotubes are widely used in the field of carbon nanotechnology nowadays.
The Raman spectrum of an SWCNT is similar to graphene, which is to be expected as it is a rolled-up sheet of graphene. In Figure 2, in the Raman spectrum of SWCNT, prominent G and G' peaks are visible. The band relative to the frequency of tangential displacement (G) is close to 1590 cm −1 , and the band associated with the second order peak (G') is related to the charge transfer between the nanotubes and the reaction mixture [66]. It is also possible to see a well-defined band near 1350 cm −1 . This band is denominated as the D band. The D band is ascribed to hybridized vibrational mode, which is related to the graphene edges, and it designates the presence of some disorder to the carbon structure [67].
Comparing the RAMAN spectra of SWCNT and SWCNT−COOH, the latter shows two additional bands near 1540.7 and 267.7 cm −1 , and a peak shift of the G' and G bands approximately 60 and 9 cm −1 to the right. These changes are associated with defects in the walls of the nanotubes caused by chemical oxidation by introducing groups (OH, CHO and COOH) on their surface. Additionally, a decrease in the IG/ID of the SWCNT after chemical modification from 7.9 to 2.1 was observed, confirming the presence of additional defects in the nanomaterial. Carbon nanotubes are rolled-up graphene sheets that have been wound into tubes. These nanomaterials are cylindrical tubes with a single outer wall and a diameter typically around 1-2 nm. Due to their exceptional mechanical, electrical and thermal assets, carbon nanotubes are widely used in the field of carbon nanotechnology nowadays.
The Raman spectrum of an SWCNT is similar to graphene, which is to be expected as it is a rolled-up sheet of graphene. In Figure 2, in the Raman spectrum of SWCNT, prominent G and G peaks are visible. The band relative to the frequency of tangential displacement (G) is close to 1590 cm −1 , and the band associated with the second order peak (G ) is related to the charge transfer between the nanotubes and the reaction mixture [66]. It is also possible to see a well-defined band near 1350 cm −1 . This band is denominated as the D band. The D band is ascribed to hybridized vibrational mode, which is related to the graphene edges, and it designates the presence of some disorder to the carbon structure [67].
Comparing the RAMAN spectra of SWCNT and SWCNT-COOH, the latter shows two additional bands near 1540.7 and 267.7 cm −1 , and a peak shift of the G' and G bands approximately 60 and 9 cm −1 to the right. These changes are associated with defects in the walls of the nanotubes caused by chemical oxidation by introducing groups (OH, CHO and COOH) on their surface. Additionally, a decrease in the I G /I D of the SWCNT after chemical modification from 7.9 to 2.1 was observed, confirming the presence of additional defects in the nanomaterial.
Regarding the RAMAN spectra of MIP and NIP materials, the most relevant information involves changes at the level of the G and D bands and the respective peak ratio (Figure 2). The I G /I D for MIP and NIP were 4.4 and 4.7, respectively. These values are very similar, which would be expected, since the polymeric matrix is identical. Besides, this ratio increased in comparison to SWCNT-COOH, indicating the presence of a crystalline structure, since the G band is the Raman signature for sp2 carbons. Additionally, we can observe a shift of the peaks G', G and D to the right compared to the previous step of the chemical modification of SWCNT-COOH, and an enhancement of an additional band at approximately 1900 cm −1 . The shift of these bands was approximately 13.9 and 12 cm −1 , respectively. Overall, this result confirms the SWCNT oxidation and MIP and NIP polymerization.

FTIR Analysis
FTIR analysis is useful to evaluate the chemical structure of the nanomaterial and to identify existing compounds by surface analysis. The FTIR / ATR spectra were plotted for SWCNT, SWCNT-COOH, MIP and NIP (Figure 3). Regarding the RAMAN spectra of MIP and NIP materials, the most relevant information involves changes at the level of the G and D bands and the respective peak ratio (Figure 2). The IG/ID for MIP and NIP were 4.4 and 4.7, respectively. These values are very similar, which would be expected, since the polymeric matrix is identical. Besides, this ratio increased in comparison to SWCNT-COOH, indicating the presence of a crystalline structure, since the G band is the Raman signature for sp2 carbons. Additionally, we can observe a shift of the peaks G', G and D to the right compared to the previous step of the chemical modification of SWCNT-COOH, and an enhancement of an additional band at approximately 1900 cm −1 . The shift of these bands was approximately 13.9 and 12 cm −1 , respectively. Overall, this result confirms the SWCNT oxidation and MIP and NIP polymerization.

FTIR analysis
FTIR analysis is useful to evaluate the chemical structure of the nanomaterial and to identify existing compounds by surface analysis. The FTIR / ATR spectra were plotted for SWCNT, SWCNT-COOH, MIP and NIP (Figure 3). As observed in Figure 3, the SWCNT does not present relevant visible peaks; however, after acidic treatment, many new peaks appear. The peak observed for SWCNT-COOH at 1701 cm −1 can be attributed to the C = O stretch of the carboxylic group, whereas the peak at 1545 cm −1 can be related to the C = C stretch of SWCNTs and the peak at 1150 cm −1 to the C-O stretch [68,69]. These bands may be related to the chemical oxidation of SWCNT.
Both MIP and NIP materials showed intense peaks at 1723, 1539, 1444, 1087.67 and 856 cm −1 , respectively. These peaks have been attributed to the carboxylic/acetone groups, C = C stretch and C-O/C=O stretch bonds. Although these peaks are also present in the SWCNT; however, they showed a peak shift to the left or right. This characteristic allows us to confirm the modification of the surface after polymerization. As observed in Figure 3, the SWCNT does not present relevant visible peaks; however, after acidic treatment, many new peaks appear. The peak observed for SWCNT-COOH at 1701 cm −1 can be attributed to the C=O stretch of the carboxylic group, whereas the peak at 1545 cm −1 can be related to the C=C stretch of SWCNTs and the peak at 1150 cm −1 to the C-O stretch [68,69]. These bands may be related to the chemical oxidation of SWCNT.
Both MIP and NIP materials showed intense peaks at 1723, 1539, 1444, 1087.67 and 856 cm −1 , respectively. These peaks have been attributed to the carboxylic/acetone groups, C=C stretch and C-O/C=O stretch bonds. Although these peaks are also present in the SWCNT; however, they showed a peak shift to the left or right. This characteristic allows us to confirm the modification of the surface after polymerization.

SEM Analysis
The SEM images of the carboxylated SWCNTs and MIP materials are shown in Figure 4. The SWCNTs have a thickness of a few nanometers; however, the SEM images show that the oxidized SWCNTs have some impurities or are in an agglomerated state. The chemical modification of the surface was clearly visible. This was confirmed by magnifying the original material 100,000 times. However, the occurrence of the imprinting could not be verified by SEM, since we cannot observe cavities. In addition, the surface modification with the imprinting polymer significantly increased the thickness of the SWCNT, and some agglomerates were observed, which can be attributed to the presence of the polymer. (see Figure 4).

SEM Analysis
The SEM images of the carboxylated SWCNTs and MIP materials are shown in ure 4. The SWCNTs have a thickness of a few nanometers; however, the SEM images that the oxidized SWCNTs have some impurities or are in an agglomerated state chemical modification of the surface was clearly visible. This was confirmed by mag ing the original material 100,000 times. However, the occurrence of the imprinting c not be verified by SEM, since we cannot observe cavities. In addition, the surface m cation with the imprinting polymer significantly increased the thickness of the SW and some agglomerates were observed, which can be attributed to the presence o polymer. (see Figure 4).

Sensory Surfaces
The materials MIP and NIP were incorporated into the PVC membrane and test sensing materials with potentiometric transduction. The effect of membrane compos on the potential response was investigated and the results are shown in Table 1.
Overall, the membranes were prepared by dispersing the active compounds (M NIP) in a high dielectric constant plasticizer, oNPOE and PVC. The type of plasticize ployed is important to the analytical performance of PVC-based sensors. Some plastic cause leakage or migration from the membrane, which limits the lifetime of sensor mat based on plasticized membranes. Almeida and co-authors have previously describe effects of the plasticizer dielectric constant on the analytical performance of sensors. W oNFOE was used as plasticizing solvent, the detection limit of the sensor was lower d the higher dielectric constant compared to other conventional solvents [70].

PH Effect
ISEs were calibrated in the MOPS buffer, 1.0 × 10 −1 mol/L, pH 6.0 and 8.0. Under conditions, the peptide is slightly negatively charged because its isoelectric point [64]. However, all electrodes exhibited cationic slopes, and an opposite behavior wa pected. A possible explanation for this behaviour is related to the presence of severa izable functional groups on the amyloid surface that can interact with the PVC memb causing the protein to approach the MIP binding sites within the PVC membrane b positively charged groups. At pH 6, the MIP showed linear behavior after 1.23 μg/m cationic slope of 62.1 mV/dec and a detection limit of 1.20 μg/mL. The NIP showed a linear behavior with an R 2 of 0.92, indicating that the (re)binding mechanism o

Sensory Surfaces
The materials MIP and NIP were incorporated into the PVC membrane and tested as sensing materials with potentiometric transduction. The effect of membrane composition on the potential response was investigated and the results are shown in Table 1.
Overall, the membranes were prepared by dispersing the active compounds (MIP or NIP) in a high dielectric constant plasticizer, oNPOE and PVC. The type of plasticizer employed is important to the analytical performance of PVC-based sensors. Some plasticizers cause leakage or migration from the membrane, which limits the lifetime of sensor materials based on plasticized membranes. Almeida and co-authors have previously described the effects of the plasticizer dielectric constant on the analytical performance of sensors. When oNFOE was used as plasticizing solvent, the detection limit of the sensor was lower due to the higher dielectric constant compared to other conventional solvents [70].

PH Effect
ISEs were calibrated in the MOPS buffer, 1.0 × 10 −1 mol/L, pH 6.0 and 8.0. Under these conditions, the peptide is slightly negatively charged because its isoelectric point is 5.5 [64]. However, all electrodes exhibited cationic slopes, and an opposite behavior was expected. A possible explanation for this behaviour is related to the presence of several ionizable functional groups on the amyloid surface that can interact with the PVC membrane, causing the protein to approach the MIP binding sites within the PVC membrane by its positively charged groups. At pH 6, the MIP showed linear behavior after 1.23 µg/mL, a cationic slope of 62.1 mV/dec and a detection limit of 1.20 µg/mL. The NIP showed a nonlinear behavior with an R 2 of 0.92, indicating that the (re)binding mechanism of the peptide in the imprinted material dominates the potentiometric response (Table 1). Overall, the ISEs with the MIP material showed the best operating characteristics in terms of the slope and R 2 .
For the calibration curve at pH 8.0, the MIP sensor showed overnernstian slopes of 75.0 mV/decade and an NIP material of 44.0 mV/decade, respectively. The MIP showed a linear response after 1.20 µg/mL and LOD after 0.97 µg/mL ( Figure 5). The particles of NIP showed acceptable properties in terms of LOD, but a lower Nernstian slope compared to the MIP material, indicating that the binding sites of MIP led to the analytical response.
peptide in the imprinted material dominates the potentiometric response (Table 1). Overall, the ISEs with the MIP material showed the best operating characteristics in terms of the slope and R 2 .
For the calibration curve at pH 8.0, the MIP sensor showed overnernstian slopes of 75.0 mV/decade and an NIP material of 44.0 mV/decade, respectively. The MIP showed a linear response after 1.20 μg/mL and LOD after 0.97 μg/mL ( Figure 5). The particles of NIP showed acceptable properties in terms of LOD, but a lower Nernstian slope compared to the MIP material, indicating that the binding sites of MIP led to the analytical response. Overall, the best results in terms of the analytical performance were obtained for the membranes with MIP as an ionophore at pH 8.0. These conditions were used for further studies.
The analytical performance of the biosensor was compared (Table S2) with previously reported electrochemical biosensors with different recognition elements for Aβ-42 detection. Overall, these methods show a better analytical performance compared to this approach. However, to our knowledge, this research demonstrates for the first time a simple approach based on potentiometric transduction. We are aware that improvements should be made in terms of the analytical performance. However, we present a low-cost and simple system based on two electrodes made of graphite carbon electrode (reference and working electrode).

Response Time
The time required to reach a stable potential (±3 mV) in the Aβ-42, 5 μg/mL solution with a rapid 10-fold increase in concentration was 300 s. A low potential drift and longterm stability were observed with successive calibrations. Wash periods with the buffer were inserted between calibrations to remove bound proteins and adjust the potential to a "blank" value. In general, LODs, response times, the linear range and calibration slopes were reproducible within ±5% of their original values over ≈2 weeks.

Selectivity Study
The selectivity study of ISEs is a critical key factor for the success of the overall performance of the biosensing device in applications in real or close-to-real conditions. There are several methodologies adopted for estimating the potentiometric selectivity coefficients, including the mixed solutions, the separated solutions or the matched potential. Overall, the best results in terms of the analytical performance were obtained for the membranes with MIP as an ionophore at pH 8.0. These conditions were used for further studies.
The analytical performance of the biosensor was compared (Table S2) with previously reported electrochemical biosensors with different recognition elements for Aβ-42 detection. Overall, these methods show a better analytical performance compared to this approach. However, to our knowledge, this research demonstrates for the first time a simple approach based on potentiometric transduction. We are aware that improvements should be made in terms of the analytical performance. However, we present a low-cost and simple system based on two electrodes made of graphite carbon electrode (reference and working electrode).

Response Time
The time required to reach a stable potential (±3 mV) in the Aβ-42, 5 µg/mL solution with a rapid 10-fold increase in concentration was 300 s. A low potential drift and longterm stability were observed with successive calibrations. Wash periods with the buffer were inserted between calibrations to remove bound proteins and adjust the potential to a "blank" value. In general, LODs, response times, the linear range and calibration slopes were reproducible within ±5% of their original values over ≈2 weeks.

Selectivity Study
The selectivity study of ISEs is a critical key factor for the success of the overall performance of the biosensing device in applications in real or close-to-real conditions. There are several methodologies adopted for estimating the potentiometric selectivity coefficients, including the mixed solutions, the separated solutions or the matched potential. Herein, we selected the mixed solution method and tested against the MIP−based ISE sensor. The selectivity behaviour of Aβ-42 ISEs was estimated by calculating the logarithmic potentiometric selectivity coefficients, log K POT . As the log K POT decreased, the sensor increased its preference for Aβ-42, as these coefficients established the ability of an ISE to differentiate a particular (primary) ion from others (interfering species). The interfering compounds designated for this study were, Crea, Alb and urea prepared in spiked serum. (see Table S1). Overall, a decreased log K POT was expected to be observed for Crea and urea when compared with Alb. As Alb is a protein with many amino acids with functional groups, such as amino and carboxyl groups, and a high molecular weight, an increased log K POT was expected to be observed, and, in contrast, all interfering species were expected to demonstrate quite similar behaviour. This could be ascribed to their net charge at pH 8.0. Thus, the addition of a cationic additive to the membrane is expected to improve the selectivity. However, this study could be conducted in future studies.

Spiked Serum Samples Analysis
An analytical application of the sensor was performed on the membrane with MIP in a spiked serum sample. Each analysis was performed immediately after the calibration of the ISEs in 100-fold diluted serum.
The concentration of Aβ-42 in serum was 1.31, 1.08 and 0.71 µg/mL, respectively ( Table 2). The results obtained for the MIP sensor showed an average recovery of 104.2%, ranging from 102.3% to 107.05%. The relative error ranged from 2.34% to 7.49%. Thus, the results show that the MIP-ISE membranes are suitable for practical use in serum samples. However, the sensor still needs to be improved in terms of the detection limit required to sample analysis.

Conclusions
This research describes the development and characterization of potentiometric sensors based on MIP for the determination of the target molecule Aβ-42.
In essence, MIP technology allows for the production of synthetic receptors with binding constants comparable to those of natural receptors but capable of withstanding extreme conditions, such as temperature, pressure, pH and organic solvents. In theory, synthetic antibodies are very stable and robust, much cheaper than natural receptors and can be stored for long periods of time. The proper selection of the imprinting method is critical, as is the type of polymer material. Conventional bulk imprinting has proven to be a complicated process. Major difficulties include a reduced mass transfer and entrapment of the analyte in the polymer matrix, compromised integrity of the polymer structure, limited solvent choices and formation of heterogeneous binding sites. These limitations can be overcome by surface polymerizations. In surface imprinting, nanometer-sized molecules are immobilized on the surface of nanostructured materials.
In this work, the synthesis of a nanostructured molecular imprint polymer immobilized on a SWCNT surface is described. Due to its small size, a high surface-to-volume ratio is achieved, which favors miniaturization. This material showed good affinity and sensitivity for the Aβ-42 peptide.
The imprinted material was immobilized on a PVC membrane and evaluated by potentiometry. The potentiometric response was controlled by the properties of the medium in which the membranes were located so that the pH had a major influence. In alkaline, more precisely at pH 8.0, the membranes showed good analytical properties. Overall, these membranes have several advantages: a simple design, low response time, and good selectivity. Moreover, this method proved to be simple, cheap, accurate and cost effective in terms of the reagent and equipment consumption. Further developments are needed to reduce the LOD of the imprinted materials.