Development of a highly sensitive CNT-metal graphene hybrid nano-IDA electrochemical biosensor for the diagnosis of Alzheimer's disease

Abstract

The blood-based detection of Alzheimer's disease (AD) is becoming challenging since the blood–brain barrier (BBB) restricts the direct circulation of AD molecules in the blood, thereby precluding the detection of AD at an early-stage. Herein, we report the development of a novel CNT-metal-porous graphene hybrid (CNT-MGH) nano-interdigitated array (n-IDA) electrochemical 8-well biosensor for the successful early-stage diagnosis of AD from blood. Laser-induced graphene (LIG) technology has been used to fabricate the proposed CNT-MGH n-IDA 8-well sensor. Firstly, the electrochemical characterization (i.e., electrode gap, material composition, etc.) of the proposed sensor was demonstrated by measuring p-aminophenol (PAP) with a limit of detection (LOD) of 0.1 picomole. Subsequently, the CNT-MGH n-IDA 8-well sensor was then used to diagnose AD via novel blood biomarkers p-Tau 217 and p-Tau 181 using an electrochemical enzyme-linked immunosorbent assay (e-ELISA) enzyme by-product PAP. During e-ELISA, the alkaline phosphatase enzyme (IgG-AP) tagged to the detection antibody produced an electroactive ELISA by-product PAP by reacting with the enzyme–substrate 4-aminophenyl phosphate (PAPP). Finally, the CNT-MGH n-IDA 8-well sensor was then used to measure the current generated by the redox reaction via the e-ELISA by-product PAP. While quantified, the proposed CNT-MGH n-IDA 8-well sensor successfully detected p-Tau 217 and p-Tau 181 proteins in blood with LODs of 0.16 pg ml⁻¹ and 0.08 pg ml⁻¹, respectively.

---

1. Introduction

Over recent years, there has been increased interest in accurate and less invasive AD diagnosis from blood or plasma because these are less invasive body fluids and can be used to accurately diagnose Alzheimer's patients.^1–3 Moreover, the concentrations of AD biomarkers in blood are 10 to 100 times lower than that in the CSF,⁴ since the blood–brain barrier restricts AD biomarkers from directly circulating into the blood from the brain.^5,6 Amyloid beta and tau proteins are the most promising blood biomarkers for the early and accurate diagnosis of AD.⁷ Briefly, amyloid beta is an amyloid precursor protein, whereas tau is a brain-specific multifunctional microtubule-associated protein that is a more effective and favorable biomarker than amyloid beta.⁸ Recently developed novel blood biomarkers, p-Tau 217 and p-Tau 181, are superior for diagnosing Alzheimer's patients with higher accuracy.^9–11 Additionally, the standard levels of tau proteins in the blood of healthy individuals are usually less than 0.2 pg ml⁻¹ for p-Tau 217 and less than 2 pg ml⁻¹ for p-Tau 181.^12,13 Moreover, p-Tau 217 and p-Tau 181 levels increase in healthy individuals earlier in the disease process, often before significant cognitive symptoms emerge.^14,15 Therefore, p-Tau 217 and p-Tau 181 are hallmarks of AD biomarkers for early-stage diagnostics.

Several methods have been developed to diagnose AD utilizing blood samples, such as enzyme-linked immunosorbent assay (ELISA), electrochemical biosensors, field-effect transistor-based biosensors, surface-enhanced Raman spectroscopy, and surface plasmon resonance (SPR).^16,17 A novel electrochemical ELISA (e-ELISA) technique has been developed by combining conventional ELISA with electrochemical biosensors that may accurately diagnose AD in its early stages.¹⁸ The e-ELISA technique offers several advantages compared to conventional ELISA¹⁹ and traditional electrochemical biosensors,²⁰ making it a powerful tool for the sensitive detection of AD biomarkers. Unlike conventional ELISA, which relies on optical detection and requires bulky and expensive instrumentation, e-ELISA leverages electrochemical detection, enabling miniaturization and integration into portable devices. This transition reduces costs and also enhances sensitivity and detection limits due to the low background noise and high signal-to-noise ratio inherent in electrochemical measurements. However, the traditional electrochemical biosensors and specific antigens and antibodies are immobilized on the electrode surface.^21,22 Therefore, the surface needs covalent and noncovalent functionalization before immobilization using different functional groups (–NH₂, –COOH, –OH) depending on the chemical groups on the electrode in the presence of or absence of supporting materials for specifically binding the antibodies with corresponding antigens.²³ Improper functionalization may cause loss of activity, thus providing less specificity and low biocompatibility. The e-ELISA technique does not require the functionalization of the electrode surface, thus facilitating robust specificity with the high sensitivity of electrochemical transduction, and allowing for the precise detection of AD biomarkers. Because of higher measured electrochemical signals, the interdigitated array (IDA) sensor has extensively been used to diagnose AD, among other electrochemical biosensors,^24–26 and, therefore, the IDA sensor has a significantly lower detection limit (LOD).

In recent years, nanoscale functional nanomaterials, such as zero-dimensional (0D), one-dimensional (1D), and two-dimensional (2D) materials, have been frequently utilized as electrochemical biosensors in the detection of AD.^27–29 Metal nanomaterials (i.e., gold, silver, platinum, and copper nanoparticles) and metal–oxide nanomaterials (i.e., ZnO NPs, NiMoO₄ NPs, Co₃O₄ NPs, and CoP NPs) have been used as 0D nanomaterials for electrochemical detection of AD because 0D nanomaterials have exhibited better stability and electrical conductivity.³⁰ Moreover, ink-based AgNP electrochemical biosensors have been developed for AD detection using an inkjet printing technique.³¹ AgNPs are also more highly conductive than copper nanoparticles and are cost-effective when compared with gold and platinum nanoparticles; as a result, AgNPs have received special attention in electrochemical biosensors. On the other hand, metal–oxide nanomaterials,³² NiMoO₄ NPs,^31,32 and Co₃O₄ NPs³³ have been reported for the sensitive and selective detection of AD. However, 0D nanomaterials have a lower surface area, low reproducibility, and a lower detection limit up to concentrations of μg ml⁻¹.^34–36 In the past few years, considerable attention has been directed towards the hybridization of metal nanomaterials with carbon nanomaterials, such as 1D nanomaterials CNT and 2D nanomaterials graphene to improve the electrical conductivity and surface area. The hybridization of metal nanomaterials with CNTs has shown high electric catalytic activity, good stability, and excellent electrochemical performance during AD detection.^37,38 Similarly, graphene has been hybridized with metal nanomaterials (AgNPs) that exhibit good chemical stability during redox activities.^39,40 The metal–graphene and metal-CNT hybrid electrochemical biosensors still have a low surface area and insufficient detection limits up to concentrations of ng ml⁻¹ that cannot diagnose AD in the early stages.^39,41 To overcome the current limitations, we have proposed hybridizing metal nanomaterials (AgNPs), graphene, and CNT, thus providing ample surface area due to their hybrid structures, higher electrical conductivity, and faster electron transfer rates.

Herein, a novel CNT-metal-porous graphene hybrid (CNT-MGH) n-IDA 8-well sensor has been developed using a laser-induced graphene (LIG) direct writing technique on a commercial polyimide (PI) substrate. Metal nanoparticles (i.e., AgNPs) and MWCNTs were bridged using LIG; the MWCNTs were mixed with ethanol and AgNPs and then appropriately intermingled at room temperature. The mixture solution was coated on the PI substrate by using a spin coater and dried on a hot plate. Laser direct writing technology was used to fabricate the CNT-MGH n-IDA 8-well sensor under the optimal condition of the laser parameters, and the stripper solution removed the uncovered graphene area. The proposed CNT-MGH n-IDA 8-well sensor was designed with a large sensing area of 5 mm × 3 mm, cantoning many pairs of working electrodes. Each finger electrode of the MHG IDA was 400 μm wide and the gap between the electrodes was 100 μm. The counter and reference electrodes were also fabricated along with the working electrode. The proposed sensor was used to measure the different concentrations of PAP, achieving a detection limit (LOD) of 0.1 pM, and was applied in the detection of AD novel blood biomarkers p-Tau217 and p-Tau 181 using an electroactive by-product of PAP via e-ELISA. For e-ELISA, 4-aminophenyl phosphate (PAPP) and goat anti-rabbit IgG (H+L) secondary antibody AP (IgG-AP) are usually used as an enzyme label and an electroactive enzyme–substrate, respectively. To the best of our knowledge, this is the first report on developing electrochemical measurements of e-ELISA by-products in the 96-well reaction chamber using CNT-MGH n-IDA 8-well sensors. In particular, the target antigen was conjugated by the capture antibody-coated on a 96-well microplate to make a capture antibody-target antigen complex, then the detection antibody and IgG-AP enzyme were sequentially applied. Upon applying the enzyme–substrate PAPP, the AP enzyme tagged at the detection antibody produced an electroactive enzyme product of PAP by reacting with the enzyme–substrate PAPP. Finally, the CNT-MGH n-IDA 8-well electrodes measured the current generated by the redox reaction of the e-ELISA by-product PAP. Thus, the proposed CNT-MGH n-IDA 8-well sensors were applied to detect various concentrations of p-Tau 217 and p-Tau 181 protein in blood with the limit of detection (LOD) of 0.16 pg ml⁻¹ and 0.08 pg ml⁻¹.

2. Biosensing principle behind the proposed sensor

Graphene is a zero-bandgap semiconductor with a hexagonal carbon structure that includes porous flakes and exhibits excellent electrocatalytic activities. The commercial PI substrate, which transforms into graphene under laser irradiation, has been utilized to develop the CNT-MGH n-IDA 8-well sensor. When the PI substrate, coated with metal particles (i.e., AgNPs) and CNT, is irradiated with a laser, the AgNPs melt and are firmly anchored along with the transformed porous graphene flakes and CNT, resulting in a highly conductive structure with a large surface area. This CNT-MGH structure increases the electron transfer rate and is an ideal electrode material for electrochemical sensing electrodes. The electrochemical measurements of CNT-MGH n-IDA rely on the redox reaction between two electrode fingers, as shown in Fig. 1(g) and Fig. (S3).† ^42,43 The redox species is recycled between the anode and cathode during the redox reaction. The anode (one electrode array) is oxidized to generate reduced molecules, and the cathode (another electrode array) is reduced to oxidized molecules. Due to the recycling of redox molecules, the electrochemical signal generated by n-IDA is simultaneously amplified.^44,45Fig. 1 shows the conceptual view of the proposed CNT-MGH n-IDA 8-well sensor fabrication and measurement of Human Phosphorylated Tau (p-Tau 181 and p-Tau 217), biomarkers of AD, using the e-ELISA by-product. The IDA of the two electrodes (anode and cathode) was made as a pair of combs with a 125 μm gap between the pairs. Since the fingers of the two electrode combs are interdigitally arranged, the entire CNT-MGH n-IDA 8-well sensor increases the effective surface area with a minimal electrode gap and number of pairs, thus maximizing the signal for early-stage AD detection.

Fig. 1 The stepwise fabrication scheme of the proposed 8-well sensors and e-ELISA measurements using the CNT-MGH n-IDA 8-well sensor. (a) Commercial polyamide tape was attached to the 1 mm acrylic wafer, and the homogeneously mixed CNT-AgNPs solution was coated on the polyamide-acrylic substrate using a spin coater at 1000 rpm; (b) the coated wafer was dried on the hot plate; (c) laser engraving for making a CNT-MGH n-IDA sensor; (d) the fabricated CNT-MGH n-IDA sensor, including four electrodes (i.e., reference electrode, counter electrode, and two working electrodes), and the CNT-MGH structure. (e) AD biomarkers p-Tau 217 and p-Tau 181 in the human neuron, (f) the proposed n-IDA 8-well sensor dipped in the 96-well microplate for e-ELISA, (g) electrochemical redox reaction on n-IDA 8-well sensor, (h) electrochemical signal.

3. Methods and materials

3.1 Materials

Spherically shaped AgNPs (purchased from Join M. Co. Ltd. Korea), flexible polyimide film (PI) with 125 μm thickness (purchased from Join M Co. Ltd Korea), and MWCNTs (purchased from Sigma-Aldrich, Korea) were used to fabricate the CNT-MGH n-IDA 8-well sensor. The Tris Buffer (TBS) powder and P-aminophenol (PAP) were purchased from Sigma-Aldrich, Korea, and 0.1 M Tris Buffer (TBS) solution was prepared using deionized water (DI water). The PAP stock solution 10 mM was prepared by adding 54.565 milligrams of PAP to 50 mL of 0.1 M Tris Buffer solution. The different concentrations of PAP ranging from 1 mM to 0.01 pM were prepared by serial dilution from a 10 mM stock solution. The goat anti-Rabbit IgG (H+L) Secondary Antibody AP (IgG-AP) and 4-aminophenyl phosphate (PAPP) were purchased from ThermoFisher Scientific U.S. and 0.1 M Tris Buffer solution was used as a diluent buffer for PAPP. The Ag/AgCl ink was purchased from Sigma-Aldrich, Korea, and Ag/AgCl ink was used to paint the reference electrode on the 8-well n-IDA sensor. The conductive silver, provided by M.G. Chemicals Ltd, Canada, was used as the contact pad of the developed sensor. The 50% stripper solution (purchased from Dongjin Semichem, Korea) was prepared with DI water and the diluted stripper solution was cooled at room temperature for 40 min. For e-ELISA, p-Tau 181 ELISA kits including standard protein, standard diluent buffer, and wash buffer concentrate (25×) were purchased from ThermoFisher Scientific U.S., and Human Phosphorylated Tau 217 (p-Tau217) ELISA kits including standard protein, standard diluent buffer, and wash buffer concentrate (25×) were purchased from SUNLONG, China. The p-Tau 181 and p-Tau 212 standards supplied with the ELISA kit were reconstituted (stock solution) using a sample diluent and different concentrations were prepared from the stock solution using a sample diluent at the ratio of 1 : 1. Human plasma was collected from the Korea University Ansan Hospital and stored in the refrigerator below 4 °C. To conduct an e-ELISA using a real human plasma sample, p-Tau 181 antigens (ThermoFisher Scientific, U.S.) at concentrations of 0.04, 0.08, 0.16, 0.31, 0.63, 1.25, 2.5, and 5 pg ml⁻¹ were prepared using the plasma as a diluent buffer.

3.2 Fabrication of the CNT-metal graphene hybrid (MGH) n-IDA 8-well sensor

CNTs were dispersed using a surfactant solution containing 1% Tween 20 (Sigma-Aldrich Company) in DI water.^46,47 The CNTs in the surfactant solution were sonicated in an ultrasonic bath for 1 hour. Then, a magnet was placed at the bottom of the container to separate the metal impurities from the solution. The CNT-MGH n-IDA 8-well sensor was fabricated in four steps. First, dispersed MWCNTs (30 wt%) were mixed with ethanol (70 wt%) and left for 180 seconds to mix homogeneously at room temperature. Then, AgNPs (10 wt%) were combined with the prepared solution and left for 10 minutes to mix homogeneously at room temperature. The mixed solution was then coated on a polyamide (PI) acrylic substrate by a spin coater at 1000 rpm and baked at 60 °C on a hot plate. Secondly, the desired pattern was made using a computer-controlled CO₂ laser machine under optimized conditions (i.e., 80 mm s⁻¹ scan speed, 5.8 watts, and 16 mm vertical distance between the laser nozzle and the surface substrate) on the coated substrate.^48,49 The CO₂ laser power and scan speed were optimized according to the sheet resistance of the CNT-MGH structure, as shown in Fig. S2.† Thirdly, the laser-treated wafer was dipped into a 50% stripper solution for 300 seconds to remove the untreated area. Finally, conductive silver epoxy and Ag/AgCl ink were used to paint the contact pad and the reference electrode using a screen-printing mask, followed by 1 hour of baking at 90 °C in a convection oven.

3.3 Electrochemical experimental procedures

The electrochemical analyzer system connected 8-array pins to hook up the CNT-MGH n-IDA 8-well sensor. Fig. 2(a and b) show the digital images of the 8-well sensor and a single-well sensor, which shows an IDA electrode including four electrodes (i.e., a reference electrode, a counter electrode, and two working electrodes). The 8-array pin connects four electrode contact pads to record the redox cycle's electrochemical activity. Fig. 2(d and e) respectively show the 8-well sensor connector with and without closing the connector array hook. Fig. 2(f) shows the printed circuit board (PCB) for recording the electrochemical activity of an 8-well sensor with eight different concentrations from low to high concentrations for AD biomarkers, and the PCB circuit diagram for electrochemical measurements for an 8-well sensor is shown in Fig. S1.† The 8-well sensor was connected to the 8-array pin and the array hook was closed, as shown in Fig. 2(g). In the capture antibody-coated 96-well microplates, as shown in Fig. 2(h), 100 μl of target antigen was loaded on 96-well microplates followed by 2 h of incubation at room temperature. The microplates were then washed to remove the non-specially bound antigen using 260 μl of 1x wash buffer. After that, the 100 μl of biotin-conjugated detection antibody was loaded onto each 96-well microplate, incubated for 1 hour at room temperature, and then thoroughly washed with 260 μl of the same wash buffer. The IgG-AP (0.1 mM), diluted using phosphate-buffered saline (PBS, pH 7.4), was then applied to the 96-well. Subsequently, the microwell, post-incubated for 30 min at room temperature, was thoroughly washed with 260 μl of wash buffer, and 260 μl of enzyme–substrate PAPP (1 mM) solution was applied to the 96-well microplate and left for 30 min to ensure a significant reaction between IgG-AP and PAPP. The IgG-AP and enzyme–substrate PAPP were rapidly hydrolyzed into the electroactive PAP enzyme by-product. The developed CNT-MGH n-IDA 8-well electrochemical sensors were dipped into the 96-well microplate to record the electrical signal of the redox cycle of PAP, as shown in Fig. 2(i), whereas +0.35 V and −0.35 V were applied to the anode and the cathode of the working electrode, respectively. Fig. 2(j) shows the layout of the developed software for electrochemical measurements of an 8-well sensor.

Fig. 2 Images of (a and b) the developed CNT-MGH n-IDA 8-well sensor along with four electrodes (i.e., WE-1, WE-2, RE and CE), (c) SEM image of hybrid graphene, (d) the 8-well sensor array connector opening the hook, (e) closing the 8-well sensor array connector hook, (f) the electrochemical circuit for the 8-cell sensor, (g) the sensor fixed to the array connector, (h) the 96-well microplate, (i) the 8-well sensor dipped into the 96-well microplate wells, and (j) the layout of developed electrochemical software and the measured electrochemical signal.

4. Results and discussion

4.1 Physical characterization of AgNPs, porous graphene flakes, and the CNT-MGH structure

The surface morphologies of metal NPs, porous graphene flakes, and the CNT-MGH structure were studied using SEM images. The AgNPs were mixed with ethanol on the glass substrate and the regular spherical shape of the particles was observed as shown in Fig. 3(a). However, when the AgNPs were laser irradiated using a CO₂ laser source, the AgNPs melted due to photothermal and photochemical reactions as shown in Fig. 3(b). Fig. 3(c) and (d) show the properties of the MCNT before and after laser treatment. To study the anchoring of AgNPs with graphene and CNTs, the PI film was treated with a laser source to transform the PI film into graphene flakes. Fig. 3(f) shows the laser-induced bare graphene structure under the ambient conditions of the laser parameters. When the PI film was successively coated with AgNPs, CNTs, and ethanol, followed by laser irradiation, the AgNPs were found to be melted down and strongly anchored along with porous graphene flakes, and the CNT were strongly connected among the AgNPs, thus forming the CNT-MGH structure, and providing a large effective surface area as shown in Fig. 3(h). The conceptual view of the hybrid structure is shown in Fig. 1(d) and the magnified view of the CNT-MGH structure is shown in Fig. 3(i).

Fig. 3 SEM images of AgNPs mixed with ethanol on the glass substrate (a) before and (b) after CO₂ laser treatment; (c) MWCNT before CO₂ laser treatment on the glass substrate, (d) MWCNT after CO₂ laser treatment, (e) the control PI substrate, (f) laser-treated porous graphene on the PI substrate without MWCNT and AgNPs, (g) laser-treated porous graphene on the PI substrate with MWCNT, (h and i) the resulting MGH structure after coating the PI film with the mixture of CNT, AgNPs, and ethanol followed by laser treatment, and the magnified view of the CNT-MGH structure. (j–k) Surface element distributions of the CNT-MGH structure. (l) EDS of the CNT-MGH structure showing the number of surface elements. (m) Raman spectra of LIG and the CNT-MGH structure.

The interaction of the CNT-MGH structure with LIG was investigated by Raman spectroscopy, as shown in Fig. 3(m). In the Raman spectra, the peaks were observed for graphene materials in two harmonics regions, below and above 2000 cm⁻¹, characterized by D, G, and 2D bands.^50,51 The D peak was generated at the ∼1350 cm⁻¹ position after the hydrogenation reaction representing the sp³ vibration of the carbon atom's surface. Herein, the G peaks are common for graphite-related materials observed at ∼1581 cm⁻¹, which refers to the sp² in-plane vibration of the carbon atom's surface. The second harmonics region is the double resonance of the D peak at approximately ∼2700 cm⁻¹, which presents the bilayer structure of graphene. As shown in Fig. 3(m), the Raman spectra of LIG and CNT-MGH exhibited a D peak at ∼1345 cm⁻¹, a G peak at ∼1578 cm⁻¹, and a 2D peak at ∼2686 cm⁻¹. The ratio of the D-Raman peak (I_D) and G-Raman peak (I_G) was 0.57 for both LIG and CNT-MGH. The same ratio of I_D/I_G confirmed that there was no change in graphene formation after the AgNPs were melted and strongly anchored along with porous graphene flakes.^52,53

4.2 Optimization of the different ratios of AgNPs and CNTs, the different gaps of CNT-MGH-IDA electrodes, and a demonstration of the developed sensor measuring PAP

We studied the signals with different compositions of CNTs and AgNPs to optimize the conductivity. Fig. 4(a) shows the signal changes (redox potential) and Fig. 4(b) shows the current changes in the CNT-MGH n-IDA 8-well sensor according to different ratios of CNTs. The electrical conductivity of the CNT-MGH n-IDA 8-well sensor was found to increase due to an increase in the ratio of CNTs, thereby increasing the number of redox cycles of PAP molecules. As shown in Fig. 4(b), the current increased linearly for up to 30% of CNTs, and the signal was saturated beyond this, while 0.01 μM PAP concentration was used for each measurement. Since the current is stabilized beyond 30% of CNTs, this ratio (30%) was selected for the CNT-MGH n-IDA 8-well sensor for our experiments.

Fig. 4 (a) Electrochemical signal and (b) stabilized current changes according to the different concentrations of CNTs with a 10% ratio of AgNPs and a 125 μm electrode gap for the 0.01 μM PAP concentration. (c) Real-time electrochemical signals for different electrode gaps and (d) measured current changes for the different gaps of the CNT-MGH n-IDA 8-well sensor with 10% AgNPs and 30% CNTs for a PAP concentration of 0.01 μM. (e) The current changes measured for different ratios of AgNPs with a PAP concentration of 0.01 μM. (f) Current changes in the developed CNT-MGH n-IDA 8-well sensor for the different concentrations of PAP under the optimized ratio of CNTs and AgNPs, and the electrode gap between two electrodes.

In the design and fabrication of the CNT-MGH n-IDA 8-well sensor, the gap between the n-IDA electrodes was a major concern. Therefore, we structured the CNT-MGH n-IDA sensor for the electrode gap ranging from below 100 μm to 350 μm. Fig. 4(c) and (d) show signal changes and the current changes of the CNT-MGH n-IDA 8-well sensor, respectively, according to the different gaps between the two electrodes. During the redox cycle, a PAP molecule could be oxidized at one electrode, diffuse to another electrode, and then PAP molecules could be reduced by diffusing back to the first electrode. Since the diffusion time is dependent on the distance of the electrodes, the number of redox cycles increases with the decreasing electrode gap. As shown in Fig. 4(d), the current decreased with the increasing electrode gap. Since the highest current was found at the electrode gap of 125 μm and below this the device was short-circuited, 125 μm was selected as the optimized electrode gap for the developed CNT-MGH n-IDA 8-well sensor.

Fig. 4(e) shows the current changes in the CNT-MGH n-IDA 8-well sensor according to different ratios of AgNPs. The electrical conductivity of the CNT-MGH n-IDA 8-well sensor was found to increase due to an increase in the ratio of AgNPs, thereby increasing the number of redox cycles of PAP molecules. The greater the ratio of AgNPs, the more AgNPs melted and were strongly anchored along with porous graphene flakes as well as CNTs by laser irradiation, which resulted in more electron transfer in the CNT-MGH n-IDA 8-well sensor during the redox cycle. As shown in Fig. 4(e), the current increased linearly up to a ratio of 10% of AgNPs, and beyond this, the signal was saturated while a PAP concentration of 0.01 μM was used for each measurement. Since the current was stabilized beyond the 10% ratio of AgNPs, this ratio (10%) was used for the CNT-MGH n-IDA 8-well sensor for optimized conditions.

The CNT-MGH n-IDA 8-well sensor was developed and demonstrated using different concentrations of PAP. The optimized composition for CNTs (30% CNTs) and AgNPs (10% AgNPs), and the electrode gap (125 μm) were applied. During the electrochemical measurement for the CNT-MGH n-IDA 8-well sensor, firstly, the current signal of 1 mM tris-buffered saline (1 mM TBS) was measured several times for the calibration of the CNT-MGH n-IDA 8-well sensor, which ensured error-free measurement. The post-calibrated TBS signal was treated as a control (baseline) signal. The measured PAP electrochemical current signal was subtracted from the control signal to get the actual PAP signal. When the PAP molecules interact with the IDA electrode surface, oxidation occurs at one electrode and reduction occurs at another electrode due to the redox reaction among the IDA comb structures, changing the electrochemical signal beyond 10–20 s. After that, the authentic PAP signal for individual concentration was stabilized. Fig. 4(f) shows the measured stabilized current for different concentrations of PAP measured by the CNT-MGH n-IDA 8-well sensor. As seen in Fig. 4(f) the current increased linearly according to the different PAP concentrations with a detection limit of 0.1 pM.

5. ELISA with the developed CNT-MGH n-IDA 8-well sensor for the detection of Alzheimer's biomarkers

Based on the proof-of-concept from the different compositions of CNTs, AgNPs, and different gaps of IDA electrodes and the demonstration in measuring PAP concentrations, the developed CNT-MGH n-IDA 8-well sensor was applied in the detection of p-Tau 217 and p-Tau 181, novel biomarkers for Alzheimer's disease. In the ELISA procedure for the characterization of p-Tau 217 and p-Tau 181 via the electrochemical method using the developed sensor, the ELISA protocol had to be modified, replacing the streptavidin horseradish peroxidase (HRP) enzyme and 3,3′,5,5′-tetramethylbenzidine (TMB) substrate with IgG-AP and PAPP, respectively. In the modified protocol, the reaction time of IgG-AP with PAPP was optimized prior to carrying out the e-ELISA. Fig. 5(a) shows the individual reaction time of IgG-AP (0.1 μM) with PAPP (1 mM) and the corresponding current. As seen in Fig. 5(a), the current increased linearly until the reaction time of 25 min, and then the current was found to be saturated for further reaction phenomena. Thus, the reaction time of 30 min was decided as the optimized value for the e-ELISA steps of applying the IgG-AP enzyme and PAPP as the enzyme–substrate. Fig. 5(b) shows the reaction time of enzyme IgG-AP (0.1 μM) with the detection antibody; the current changed until the reaction time of 1 hour and then became saturated. Therefore, the IgG-AP reaction time of 1 hour was selected as the optimized time for e-ELISA.

Fig. 5 The quantitative detection of human p-Tau 217 and p-Tau 181. (a) The reaction time for enzyme IgG-AP (0.1 μM) and enzyme–substrate PAPP (1 mg ml⁻¹) for a fixed concentration of p-Tau 181 (1.25 pg ml⁻¹). (b) The measured current for the different conjugation times of IgG-AP (0.1 μM) with the detection antibody for a 1.25 pg ml⁻¹ concentration of p-Tau 181. (c) The current signal and (d) measured current for the different concentrations of p-Tau181 using the CNT-MGH n-IDA electrochemical sensor. (e) The current signal and (f) measured current for the different concentrations of p-Tau 217 using the CNT-MGH n-IDA electrochemical sensor, the measured detection limits of 0.08 pg ml⁻¹ for p-Tau181 and 0.16 pg ml⁻¹ for p-Tau217. (g) The measured current signal and (h) current changes for different concentrations of p-Tau 181 in plasma using the developed sensor, and (i) a comparative analysis using the proposed sensor for different concentrations of p-Tau 181 mixed with buffer and diluted plasma. All the measurements were performed using 0.1 μM IgG-AP and 1 mM PAPP, and the error bar represents ±SD from three time-independent measurements.

Finally, the proposed CNT-MGH n-IDA 8-well sensor was applied in e-ELISA. In ELISA, the assay steps were executed following the suggested protocol for the conjugation of the targets of p-Tau 217 and p-Tau 181 antigens with the capture antibody in microwells and the biotin-conjugated detection antibody with primary antibody–antigen conjugation. In the following steps of enzyme IgG-AP and enzyme–substrate PAPP, the previously optimized conjugation time was maintained. The CNT-MGH n-IDA 8-well sensor was dipped into the 96-microwell to measure the electrochemical signal of AD novel biomarkers from the ELISA by-product PAP. When the CNT-MGH n-IDA 8-well sensor was dipped into the 96-microwell, the ELISA by-product came in contact with the surface of the n-IDA 8-well sensor and the electrochemical signal rapidly increased and then stabilized for the redox reaction from the ELISA by-product. Fig. 5(c) and (e) show the measured current signal for the different concentrations of p-Tau 181 and p-Tau217. As shown in Fig. 5(d) and (f), the measured current increased linearly with the increasing concentration of target antigen p-Tau 181 and p-Tau217 AD biomarkers. This is because as the concentration of target antigen p-Tau 181 and p-Tau217 increases, the monoclonal capture antibody (mAb) specific to p-Tau 181 and p-Tau217 binds more strongly, further enhancing the conjugation of enzyme IgG-AP. Consequently, more enzyme–substrate PAPP strongly interacts with enzyme product IgG-AP, thereby producing a greater amount of the electroactive enzymic product PAP. Therefore, the measured electrochemical signal increased linearly with increasing concentrations of p-Tau 181 and p-Tau217. However, the current was measured from the stabilized signal for different concentrations of p-Tau 181 and p-Tau217 ranging from 0.04 pg ml⁻¹ to 5 pg ml⁻¹ with the limits of detection of 0.08 pg ml⁻¹ for p-Tau 181 and 0.16 pg ml⁻¹ for p-Tau 217. The limit of detection (LOD) was calculated by LOD = Y_dl/trend line Slope, where, Y_dl = Y_blank + 3 × Stdv.⁵⁴

To evaluate the proposed CNT-MGH n-IDA 8-well sensor use in real plasma sample analysis for clinical applications, a study measuring different concentrations of Alzheimer's biomarker (p-Tau 181) was carried out following the e-ELISA with various concentrations that were prepared with buffer diluent as shown in Fig. 5(d). In the comparative analysis, the different concentrations of p-Tau 181 (0.04, 0.08, 0.16, 0.31, 0.63, 1.25, 2.5, and 5 pg ml⁻¹) were mixed with real plasma samples. Fig. 5(g) shows the current signal and Fig. 5(h) shows changes in current according to different concentrations of p-Tau 181 in plasma, whereas Fig. 5(i) shows the comparative measurements mixed with plasma and buffer. As seen in Fig. 5(i), the current linearly increased with various concentrations of p-Tau 181 in plasma, and in the buffer, it seems to be comparable with a similar LOD. Therefore, the proposed sensor is useful for clinical application in Alzheimer's patients in the early-stage with higher accuracy.

6. Conclusions

In this work, a novel ultrasensitive CNT-MGH n-IDA 8-well sensor has been developed using the direct writing LIG technique. The CNT-MGH n-IDA 8-well sensor exhibits a large surface area, higher electrical conductivity, and faster electron transfer between the two electrode fingers. The developed sensor was successfully demonstrated by measuring different concentrations of PAP and was then applied to detect novel AD biomarkers p-Tau 181 and p-Tau 217. The enzyme by-products of e-ELISA were electrochemically measured in plasma with Alzheimer's biomarkers using the CNT-MGH n-IDA 8-well sensor. Since the CNT-MGH n-IDA sensor detects AD markers at a lower detection limit, it can be applied in the early diagnosis of AD patients.

Author contributions

M. Mahabubur Rahman and Joon S. Shim conceived the concept of this approach. M. Mahabubur Rahman designed and fabricated the 8-well sensor, carried out the experiments, collected the data, prepared the figures, and wrote and reviewed the manuscript. Bappa Sarkar, M. Mahabubur Rahman, and Nabil H. Bhuiyan finalized Fig. 1. Md. Tareq Rahman, Gyeong J. Jin, and M. Jalal Uddin reviewed the manuscript. Joon S. Shim critically revised the manuscript and supervised the entire project. All authors have read and approved the final manuscript.

Data availability

The data that support the findings of this study are available upon request from the corresponding author Joon Sub Shim.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors gratefully acknowledge the partial support for this work provided by a 2023 KwangWoon Research and DIPS 1000+ (Deep Tech Incubator for Startup). This research was also supported by the Korea Dementia Research Project through the Korea Dementia Research Center (KDRC), funded by the Ministry of Health & Welfare and Ministry of Science and ICT, Republic of Korea (grant number: HU20C0414) and the Korea Medical Device Development Fund grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Trade, Industry, and Energy, the Ministry of Health & Welfare, the Ministry of Food and Drug Safety) (Project Number: 1711196703, RS-2023-00255063).

References

1. M. L. Shaw, M. Korecka, M. C. Clark, Y. M. V. Lee and Q. J. Trojanowski, Nat. Rev. Drug Discovery, 2007, 6(4), 295–303.
2. H. Gao, J. Chen, Y. Huang and R. Zhao, Talanta, 2024, 268, 125311.
3. K. Kim, J. M. Kim, W. D. Kim, Y. S. Kim, S. Park and B. C. Park, Nat. Commun., 2020, 11(1), 119.
4. E. Zenaro, G. Piacentino and G. Constantin, Neurobiol. Dis., 2017, 107, 41–56.
5. F. L. Lue, A. Guerra and G. D. Walker, Neurol. Ther., 2017, 6, 25–36.
6. K. Henriksen, E. S. O'Bryant, H. Hampel, Q. J. Trojanowski, J. T. Montine, A. Jeromin, K. Blennow, A. Lonneborg, T. Wyss-Coray, H. Soares and C. Bazenet, Alzheimer's Dementia, 2014, 10(1), 115–131.
7. M. Ye, M. Jiang, J. Cheng, X. Li, Z. Liu, W. Zhang, M. S. Mugo, N. Jaffrezic-Renault and Z. Guo, Sens. Actuators, B, 2020, 308, 127692.
8. M. Mila-Aloma, J. N. Ashton, M. Shekari, G. Salvado, P. Ortiz-Romero, L. Montoliu-Gaya, L. A. Benedet, K. T. Karikari, J. Lantero-Rodriguez, E. Vanmechelen and A. T. Day, Nat. Med., 2022, 28(9), 1797–1801.
9. R. P. Kac, F. Gonzalez-Ortiz, J. Simrén, N. Dewit, E. Vanmechelen, H. Zetterberg, K. Blennow, J. N. Ashton and K. T. Karikari, Alzheimer's Res. Ther., 2022, 14(1), 65.
10. M. A. Brickman, J. J. Manly, S. L. Honig, D. Sanchez, D. Reyes-Dumeyer, A. R. Lantigua, J. P. Lao, Y. Stern, P. J. Vonsattel, F. A. Teich and C. D. Airey, Alzheimer's Dementia, 2021, 17(8), 1353–1364.
11. R. T. de Oliveira, R. C. Erbereli, R. P. Manzine, N. T. Magalhaes, L. M. Balthazar, R. M. Cominetti and C. R. Faria, ACS Sens., 2020, 5(4), 1010–1019.
12. S. Janelidze, E. Stomrud, R. Smith, S. Palmqvist, N. Mattsson, D. C. Airey and O. Hansson, Nat. Commun., 2020, 11(1), 1683.
13. E. H. Thijssen, R. La Joie, A. Strom, C. Fonseca, L. Iaccarino, A. Wolf and A. L. Boxer, Lancet Neurol., 2021, 20(9), 739.
14. F. Gonzalez-Ortiz, P. R. Kac, W. S. Brum, H. Zetterberg, K. Blennow and T. K. Karikari, Mol. Neurodegener., 2023, 18(1), 18.
15. N. J. Ashton, A. L. Benedet, T. A. Pascoal, T. K. Karikari, J. Lantero-Rodriguez, W. S. Brum, S. Mathotaarachchi, J. Therriault, M. Savard, M. Chamoun, E. Stoops, C. Francois, E. Vanmechelen, S. Gauthier, E. R. Zimmer, H. Zetterberg, K. Blennow and P. Rosa-Neto, EBioMedicine, 2022, 76, 103836.
16. M. Ding, S. Ding, D. Du, X. Wang, X. Hu, P. Guan, Z. Lyu and Y. Lin, TrAC, Trends Anal. Chem., 2023, 117087.
17. L. K. Rahn, U. Peramune, T. Zhang and K. R. Anand, Annu. Rev. Anal. Chem., 2023, 16, 49–69.
18. M. M. Rahman, N. H. Bhuiyan, M. Park, J. M. Uddin, G. J. Jin and J. S. Shim, Biosens. Bioelectron., 2024, 244, 115803.
19. J. McCarthy, Detect. Pathog. Food, 2003, 84, 241–258.
20. L. Zhang, W. Guo, C. Lv, X. Liu, M. Yang, M. Guo and Q. Fu, Adv. Sens. Energy Mater., 2023, 100081.
21. G. Ozcelikay, M. Gamella, M. A. Unal, K. Gucuyener, A. Montero-Calle, R. Barderas and S. A. Ozkan, Talanta, 2022, 246, 123526.
22. M. Ding, Q. Shu, N. Zhang, C. Yan, H. Niu, X. Li and X. Hu, Electroanalysis, 2022, 34(2), 263–274.
23. S. C. Barman, M. A. Zahed, M. Sharifuzzaman, S. G. Ko, H. Yoon, S. J. Nah and J. Y. Park, Adv. Funct. Mater., 2020, 30(14), 1907297.
24. S. Boobphahom and N. Rodthongkum, Int. J. Biol. Macromol., 2023, 253, 126316.
25. Y. M. Emran, A. M. Shenashen, A. Elmarakbi, M. M. Selim and A. S. El-Safty, Anal. Chim. Acta, 2022, 1192, 339380.
26. S. J. Shim, J. M. Rust and H. C. Ahn, J. Micromech. Microeng., 2013, 23(3), 035002.
27. R. Kour, S. Arya, J. S. Young, V. Gupta, P. Bandhoria and A. Khosla, J. Electrochem. Soc., 2020, 167(3), 037555.
28. B. Conklin, M. B. Conley, Y. Hou, M. Chen and B. K. Lee, Adv. Drug Delivery Rev., 2023, 192, 114636.
29. V. Caratelli, E. Di Meo, N. Colozza, L. Fabiani, L. Fiore, D. Moscone and F. Arduini, J. Mater. Chem. B, 2022, 10(44), 9021–9039.
30. J. Ahlawat, M. Sharma and S. C. Pundir, Microchem. J., 2023, 108620.
31. M. Sun, X. Wang, X. Guo, L. Xu, H. Kuang and C. Xu, Chem. Sci., 2022, 13(11), 3069–3081.
32. B. Chauhan, S. Patel, G. B. Prajapati and S. Singh, in Alzheimer's Disease and Advanced Drug Delivery Strategies, 2024, pp. 361–371.
33. T. Lin, Y. Xu, A. Zhao, W. He and F. Xiao, Anal. Chim. Acta, 2022, 1207, 339461.
34. M. Negahdary and L. Angnes, Biomater. Adv., 2022, 135, 112689.
35. S. Nasraoui, A. Al-Hamry, R. P. Teixeira, S. Ameur, G. L. Paterno, B. M. Ali and O. Kanoun, J. Electroanal. Chem., 2021, 880, 114893.
36. M. Mahmoudpour, S. Ding, Z. Lyu, G. Ebrahimi, D. Du, N. E. J. Dolatabadi, M. Torbati and Y. Lin, Nano Today, 2021, 39, 101177.
37. R. Das, R. Pal, S. Bej, M. Mondal, K. Kundu and P. Banerjee, Mater. Adv., 2022, 3(11), 4421–4459.
38. L. Xu, J. Xu, W. Liu, D. Lin, J. Lei, B. Zhou, Y. Shen and X. Deng, Sens. Actuators, B, 2022, 367, 132016.
39. P. J. Tessonnier, O. Ersen, G. Weinberg, C. Pham-Huu, S. D. Su and R. Schlogl, ACS Nano, 2009, 3(8), 2081–2089.
40. S. J. Shim, H. Y. Yun, J. M. Rust, J. Do, V. Shanov, J. M. Schulz and H. C. Ahn, Nanotechnology, 2009, 20(32), 325607.
41. S. Nasraoui, A. Al-Hamry, R. P. Teixeira, S. Ameur, G. L. Paterno, B. M. Ali and O. Kanoun, J. Electroanal. Chem., 2021, 880, 114893.
42. L. J. Hudson and T. T. Tsotsis, Chem. Eng. Sci., 1994, 49(10), 1493–1572.
43. T. A. V. Dam, W. Olthuis and A. van den Berg, Analyst, 2007, 132(4), 365–370.
44. S. J. Shim and H. C. Ahn, Biosens. Bioelectron., 2012, 34(1), 208–214.
45. Y. Li, C. Wang, Y. Zhu, X. Zhou, Y. Xiang, M. He and S. Zeng, Biosens. Bioelectron., 2017, 89, 758–763.
46. J. S. Shim, Y. H. Yun, M. J. Rust, J. Do, V. Shanov, M. J. Schulz and H. C. Ahn, Nanotechnology, 2009, 20(32), 325607.
47. J. S. Shim, H. Y. Yun, W. Cho, V. Shanov, J. M. Schulz and H. C. Ahn, Langmuir, 2010, 26(14), 1642–11647.
48. M. M. Rahman, J. M. Uddin, H. J. Hong, N. H. Bhuiyan and S. J. Shim, Sens. Actuators, B, 2022, 355, 131336.
49. M. T. Rahman, S. Hossen, K. J. Jeong, N. H. Bhuiyan, M. M. Rahman, B. Sarkar and J. S. Shim, Small Methods, 2024, 2400189.
50. C. A. Ferrari, C. J. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, S. K. Novoselov, S. Roth and K. A. Geim, Phys. Rev. Lett., 2006, 97(18), 187401.
51. A. M. Zahed, C. S. Barman, M. Sharifuzzaman, S. Zhang, H. Yoon, C. Park, H. S. Yoon and Y. J. Park, Adv. Funct. Mater., 2021, 31(25), 2009018.
52. A. D. Sokolov, R. K. Shepperd and M. T. Orlando, J. Phys. Chem. Lett., 2010, 1(18), 2633–2636.
53. A. M. Zahed, C. S. Barman, S. P. Das, M. Sharifuzzaman, S. H. Yoon, H. S. Yoon and Y. J. Park, Biosens. Bioelectron., 2020, 160, 112220.
54. U. Şengul, J. Food Drug Anal., 2016, 24(1), 56–62.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4bm00654b

---