Portable non‐enzymatic electrochemical biosensor based on caffeine for Alzheimer's disease diagnosis

Caffeine and its derivatives can effectively bind amyloid beta 16–22 (Aβ16–22) fragment of amyloid beta 1–42 (Aβ1–42), a biomarker for the early diagnosis of Alzheimer's disease (AD), by means of conformation selection, π–π stacking, van der Waals forces, and hydrogen bonding, so as to achieve high specificity and quantitative detection of Aβ1–42. In this study, 3‐mercaptopropionic acid (MPA), conductive polymer poly(thiophene‐3‐acetic acid) (PTAA), and pine‐like/Au pine PTAA (pine PTAA) were applied to modify the electrodes, and the non‐enzymatic caffeine was used as specific biorecognition element to study the analytical performance of the electrochemical sensor platform for Aβ1–42 oligomer (AβO). It was found that caffeine/pine PTAA‐based sensor with large surface area, high active sites, and excellent electrical conductivity demonstrated the widest linear range (10−8 to 100 nM) and highest sensitivity (743.77 Ω/log nM) in comparison. The prepared caffeine‐based sensor was afterward applied to cerebrospinal fluid and blood tests for real sample analysis, demonstrating its potential for practical use in detecting AβO at the attomolar level. Furthermore, the non‐enzymatic caffeine was constructed on the pine‐like PTAA‐modified screen‐printed electrodes for the rapid detection of AβO using portable meter.


INTRODUCTION
Alzheimer's disease (AD) is a progressive and neurodegenerative disease. 1 Currently, the presence of dementia is confirmed by analyzing the cerebrospinal fluid (CSF)/blood (BLD) with established biomarkers such as amyloid beta peptide (Aβ), 2 tau protein, 3 and phospho-tau expression levels. 4Among all the biomarkers, Aβ, particularly that of 42 amino acids in length (Aβ 1-42 ) may bind cellular prion protein (PrP C ) with high affinity on the neuronal surface and then trigger the activation of intracellular signaling cascade to combine with the protein tyrosine kinase Fyn, the stimulation of which, afterwards leading to the phosphorylation of N-methyl-D-aspartate receptor subtype 2B, dysregulation of the N-methyl-Daspartic acid receptor, and finally causes the excitotoxicity and restriction of dendritic spines. 5Hence, quantitative detection of Aβ 1-42 in CSF/BLD samples is one of the effective methods for early diagnosis of AD.
Recently, several methods have been introduced to detect Aβ 1-42 oligomer (AβO), such as an enzyme-linked immunosorbent assay, 6 electrochemistry, 7 electrochemiluminescence immunoassay, 7b surface-enhanced Raman spectroscopy, 8 colorimetric assay, 9 localized surface plasmon resonance, 10 etc.Among them, the electrochemical immunoassay, as a powerful analysis tool, could be effectively applied in AβO detection owing to its high sensitivity, speed, and good portability. 11Wei et al. used Mg 2+ -dependent DNAzyme to drive DNA bipedal walking, and a strategy with faster walking kinetics, larger walking area, and higher amplification efficiency was developed for the specific detection of AβO in serum.The aptamer electrochemical sensor allows the wide detection range of 0.1 pM-1 nM and the detection limit of 46 fM. 12 Liu et al. reported a biosensor with Aβ 1-42 antibody immobilized on a self-assembled single-layer functionalized interfinger chain-shaped electrode to detect Aβ 1-42 in serum with high sensitivity.The sensing performance of interfinger electrodes can be enhanced by modulating sensing area to be comparable in size to the biomarker.The sensor demonstrates a linear range of 10 −3 to 10 3 ng/mL, a low detection limit of 100 pg/mL, and high sensitivity. 13lthough these electrochemical methods can effectively improve sensing performance, biological materials such as enzymes, antibodies, and peptides are costly and susceptible to denaturation and inactivation by surrounding humidity, temperature, and chemical factors.Additionally, the attached biological materials and complicated fabrication may restrict the electron transfer on the electrode surface and cause limitations in the detection range.
Unlike enzymatic materials, the advantages of nonbiological molecular recognition components or nanocatalysts are that they are highly stable and can be easily fabricated and preserved, which has been widely recognized in recent years. 14Specifically, plastic antibodies were obtained by molecularly imprinted polymer (MIP) technology to mimic the recognition characteristics of biological materials. 15Most metal and metal oxide nanoparticles can be used as non-enzymatic sensors to analyze glucose, H 2 O 2 , dopamine, etc. 14d, 16 Metal-organic frameworks (MOFs) with high specific surface areas and ease of synthesis can be directly applied to the electrocatalysis and electrode materials. 17Previously, we prepared curcumin 18 and MOF 19 as biometrics to develop Aβ 1-42specific enzyme-free sensors, which reduced the cost of the sensor and improved the sensitivity and specificity of the sensor, thus expanding the application of enzyme-free sensors to a large extent and obtaining a number of patents, providing a new idea for the early diagnosis of AD.However, such non-enzymatic sensors have some barriers, such as cumbersome characterization of the electrodes, low selectivity to analysts and uncertainty in the mechanism of the sensing process. 20Specifically, MIPs are susceptible to conformational changes in large biomolecules, which can produce mismatched blotting sites.Furthermore, most metal oxide materials and MOF applied in electrochemical characterization are mainly based on the redox reactions of analysts, which limits the development of the sensor. 21affeine is a xanthine alkaloid compound that can be used as a central nervous stimulant, 22 which can not only effectively reduce the risk of coronary heart disease, stroke, and other cardiovascular diseases, 23 but also inhibit brain aseptic inflammation and reduce the generation and deposition of abnormal proteins, thus helping to resist brain degenerative diseases such as AD. 24Furthermore, due to the difference in conformation between Aβ 1-40 and Aβ 1-42 , caffeine can specifically bind Aβ 16-22 fragments in Aβ 1-42 through conformation selection, π-π stacking (interacting with phenylalanine Phe19 and 20 in Aβ 1-42 ), van der Waals forces, hydrogen bonds, etc. 25 Additionally, as a nonbiological receptor, caffeine is more stable, safer, easier to be modified, and relatively more specific than nanostructures and nanomaterials and can be an ideal recognition molecule for the detection of AβO.
In this study, label-free impedimetric biosensors, which employed non-enzymatic caffeine as bioreceptor, were constructed based on pine-poly(thiophene-3-acetic acid) (PTAA)/3-mercaptopropionic acid (MPA)-modified gold electrodes (Figure S1).Beside bare gold, the gold pine structure with microsized and nanosized assemblies are of considerable interest owing to extremely large surface areas for charge and mass transport in electrochemistry. 26urthermore, compared with the self-assembled MPA, the carboxylic acid-modified PTAA 27  of bioreceptors, which effectively improves the stability, adhesion, electrochemically activated surface area, and conductivity 28 (Figure 1).PTAA is a macromolecule composed of π-conjugated organic monomers with delocated electrons, whose conductivity can be improved when doped with an electron donor (n-doping) or acceptor (pdoping). 29After that, the sensing performance of the caffeine-based sensors was subsequently examined, and the caffeine-pine PTAA-based sensor had the highest performance in terms of detection range, sensitivity, and limit of detection.Artificial blood and brain samples from AD mice were afterward studied to demonstrate that the potential utility for early diagnosis of AD in vitro and ex vivo.In addition, the caffeine-pine PTAA was also electromodified on the screen-printed gold electrodes for the rapid detection of AβO with portable meter.

Preparation of non-enzymatic electrochemical sensors
Before modification, the gold electrode was polished by physical, chemical, and electrochemical methods to obtain the mirror morphology and remove the surface impurities.MPA was self-assembled on the gold electrode surface via immersing the polished electrode in MPA solution (deionized water:EtOH = 3:1, v:v) for 24 h.The pinelike gold was prepared on the polished gold electrode by chronoamperometry at a potential of −0.55 V for 600 s in an aqueous solution of 0.1 M Na 2 SO 4 and 30 mM AuCl 3 .26a,30 PTAA was electrochemically polymerized on the gold electrode by cyclic voltammetry (CV) between 0.55 and 1.8 V at a scan rate of 50 mV/s in the electrolyte of 10 mM 3-thiophene acetic acid (3-TAA) and 50 mM tetrabutylaminium perchlorate in acetonitrile (ACN), followed by the investigation of electrochemical properties.N-Boc-ethylenediamine hydrochloride was applied to carboxyl caffeine (caffeine-COOH) activated with 1-(3dimethyl aminopropyl)-3-ethylcarbodiimide hydrochloride (EDC)/N-hydroxy-succinimide (NHS) and connected to caffeine by amide reaction.Trifluoroacetic acid was then added to remove Boc and obtain the amino caffeine for the synthesis of caffeine-NH 2 .In order to prepare the caffeine-pine PTAA/MPA/PTAA biosensors in Figure S1, the aforementioned electrodes modified with MPA, PTAA, or the pine-like PTAA were activated by dipping into morpholino ethanesulfonic acid (MES) buffer containing EDC/NHS followed by the incubation of caffeine-NH 2 .Finally, the sensors were blocked with bovine serum albumin to reduce the effect of non-specific binding.AβO was prepared, analyzed (Figure S2), and stored at −20 • C before use.

Rapid detection on portable meter
The pine PTAA was also electro-modified on the screenprinted gold electrode.After amide reaction with caffeine-NH 2 , different concentrations of AβO in artificial blood were dropped on the electrodes for 15 min.After removal and rinsing with phosphate buffer saline (PBS), the 20 mM Fe(CN) 6 3−/4− /PBS (pH 7.4) was dropped on the electrode and recorded with portable meter.

Preparation of non-enzymatic caffeine-based sensors modified with different linking agents
The preparation of MPA self-assembled single layer (SAM), conductive polymer PTAA, and Au pine PTAA (pine PTAA) are discussed in Supporting Information and   31 Moreover, the i p of the pine PTAA electrode in Figure 2B was larger and grew faster with the electrochemical polymerization of 3-TAA compared to the PTAA formed on the ordinary gold electrode, indicating a larger surface area of pine PTAA.In addition, the morphology of the pine-like gold was observed via scanning electron microscopy imaging in Figure S3 with a branch size of about 100-500 nm.Compared with pure PTAA, pine PTAA exhibited better electrochemical performance due to its high conductivity and large active specific surface area.
The aforementioned electrodes were activated by EDC/NHS in MES buffer, followed by the connection to amine-modified caffeine through amidation reaction.Successful step immobilization and the interface properties of the caffeine-based sensors were verified by electrochemical measurements using the ternary electrode method, as shown in Figure S1.In order to study the interaction mode between caffeine and human Alzheimer's β-amyloid protein, molecular docking was performed using AutoDock tool (version 1.5.6), and the protein pocket was determined by analysis with PyMOL plugin-parKVFinder.In the docking simulation, caffeine was used as the ligand, and the Aβ 1-42 was used as the receptor.Each docking run output 20 complexes, and the results were ranked based on their binding energy values.Then, the compound with the lowest energy for each was selected as a reliable binding mode.As shown in Figure S4, caffeine can specifically bind Aβ 16-22 fragments in Aβ 1-42 through conformation selection, π-π stacking (interacting with phenylalanine Phe19 and 20 in Aβ 1-42 ), van der Waals forces, hydrogen bonds, etc. 25 CV analysis indicated a gradually increased peak potential difference of the electrode and a decreased peak current value with the gradual modification of the sensor, characterizing the successful modification of MPA, PTAA, pine PTAA, and caffeine on the electrode surface and their binding ability of Aβ 1-42 (Figure S5A-C).In addition, electrochemical impedance spectroscopy (EIS) can effectively evaluate the electron transfer efficiency at different stages of modification with monitoring changes in electron transfer resistance at each step (R et , whose magnitude is related to the radius of the semicircle displayed on the Z re axis), which also characterizes the successful modification of caffeine-based electrodes at each step.As shown in Figure S5D-F, a significant increase in the resistance value can be observed after each step.Moreover, significantly increased R et can also be detected when AβO was connected to the caffeine electrode, possibly because AβO hindered the transfer of electrons through the surface layer.These results further demonstrated the successful preparation of the sensors and their ability to bind Aβ 1-42 .Compared with other modified electrodes, the pine PTAA-modified electrode showed a much higher R et value after incubation in AβO, demonstrating a larger specific surface area and thus binding more caffeine and AβO (Table S1).

AβO sensing performance
Furthermore, the stepwise preparations of the nonenzymatic caffeine-based sensors with different substrates were characterized by EIS in incremental concentrations of AβO.It can be seen in Figure 3A-C that R et increased gradually with the increase in AβO concentration, and the variation in resistance values was linearly correlated with the logarithmic concentration of AβO (Figure 3D).As shown in Table S2, compared with the MPA-modified (linear range of 10 −5 to 1 nM and sensitivity of 658.97 Ω/log nM) and the PTAA-based sensors (linear range of 10 −6 to 100 nM and sensitivity of 84.7 Ω/log nM), the pine PTAA-modified electrode showed the widest linear range (10 −8 to 100 nM) and the largest sensitivity (743.77Ω/log nM) because the large specific surface area of pine PTAA can fix more biorecognition element of caffeine and capture more electroactive molecules (e.g., ferrocyanide) to improve the detection sensitivity and its good conducive to electron transfer, which improves the sensing performance and achieves aM level detection of AβO.The overall performance of the non-enzymatic caffeine/pine PTAA-based sensors was then compared with other reported AβO sen-sors in terms of sensitivity and detection range in Table 1, suggesting that the non-enzymatic caffeine/pine PTAA sensor with superior sensing performance could be used for the early diagnosis of AD.

Real sample analysis in vitro and ex vivo
To simulate clinical blood analysis, different concentrations of AβO were dispersed in 10% fetal bovine serum.The non-enzymatic caffeine-pine PTAA-based sensor effectively determined small amounts of AβO in the presence of several interferents with appropriate recoveries and relative standard deviations (RSDs), demonstrating satisfactory reliability and good application potential in clinical use, as shown in Figure 4 and Table S3.
Moreover, brain sample analysis was also used to test the feasibility of the prepared non-enzymatic caffeinebased biosensor for AβO detection.Mouse brain samples were extracted and dissolved in a mixture of T-PER and protease inhibitors to extract homogeneous proteins.After adequate mixing and centrifugation, extracted   protein samples were diluted in artificial CSF for real sample detection.It can be seen in Figure 5 that compared with the amount of AβO in hemizygous 5xFAD AD mice (0.05 mg/mL, 1.37 × 10 -5 nM), the concentration of AβO in wild-type (0.05 mg/mL, 1.27 × 10 -8 nM) mice is much lower.As shown in Table S4, spiked recoveries of different AβO from AD mice brain samples were found to be 83.80%-101.63%,and the RSDs were less than 4.25% for these measurements, indicating that the non-enzymatic caffeine-based sensor is promising for Aβ detection with satisfactory reliability in actual implementation.

Stability and portable meter testing
The prepared sensor based on pine PTAA was stored in the refrigerator at 4 • C and immersed in 1 nM Aβ 1-42 CSF simulation solution every 5 days.The change of resistance before and after the reaction was detected by EIS, and the stability of the sensor was measured, as shown in Figure S6.The results showed that non-enzymatic caffeine can effectively identify and combine with AD biomarker Aβ 1-42 to avoid the degeneration and deactivation of bioactive materials such as antibodies, enzymes, and peptides, which are easily affected by the surrounding environment, thus greatly improving the stability of the sensor.
The non-enzymatic caffeine was also immobilized on the pine PTAA polymerized screen-printed sensor for the rapid detection on portable meter in Figure 6.The perfect linear relationship between the rapid counts and the loga-rithmic concentration of AβO indicates high stability and good potential application in rapid detection.

CONCLUSION
In this paper, non-enzymatic caffeine was utilized for the capture and identification of Aβ 1-42 with pine PTAA for the improvement of conductivity and activated surface area.The electrochemical properties and sensing performance of prepared non-enzymatic caffeine-pine PTAA-based sensor were compared with widest detection range (10 −8 to 100 nM), highest sensitivity (743.77Ω/log nM), and lowest detection limit of AβO.Real sample analysis tests using AβO in blood and AD mice samples in artificial CSF were also successfully evaluated, indicating the potential use of the sensor for early diagnosis and real-time detection of AD patients.Moreover, the sensor was also designed in a screen-printed electrode for rapid detection using a portable meter, highlighting its versatility and potential for clinical applications.However, the current work is an exploration, based on which we will further improve the specificity of caffeine in recognizing AβO within different Aβ1-42forms, so as to achieve the purpose of selective detection while ensuring the stability of the biosensor.Overall, the non-enzymatic caffeine-pine PTAA-based sensor offers a promising avenue for the development of sensitive and reliable biosensors for AD diagnosis and monitoring.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflict of interest.

R E F E R E N C E S
was selected to deposit on the electrode surface by electropolymerization and the residual COOH can support an immobilization F I G U R E 1 Schematic diagram of a novel caffeine electrochemical biosensor for detecting amyloid beta peptide oligomer (AβO) in blood or cerebrospinal fluid based on non-enzymatic caffeine/pine poly(thiophene-3-acetic acid) (PTAA)/Au electrode.

F
I G U R E 2 Electrochemical polymerization process of conductive polymer poly(thiophene-3-acetic acid) (PTAA) (A) and pine PTAA (B) using cyclic voltammetry (CV).CV operation conditions: the electrolyte was 10 mM 3-TAA in ACN containing 50 mM tetrabutylaminium perchlorate with a scanning rate of 50 mV/s and an applied potential of 0.55-1.8V.

Figure S1 .
Figure S1.To characterize their successful preparation, CV was used to compare the changes in the redox signal on the electrode with the deposition of PTAA and pine PTAA.It can be seen from Figure 2A,B that the characteristic oxidation peak (E = 1.4 V) was increased gradually with the increase in the number of cycles of electrochemical polymerization.According to Randle-Sevcik equation,  p = (2.68648× 10 5 ) 3∕2  1∕2   *   1∕2, where i p represents the peak current, A represents the electrode surface area, n indicates the scanning rate, D represents the diffusion coefficient, C represents the analyte concentration, v represents the reaction rate with other parameters fixed, and an increase in i p indicates an increase of the active surface of the electrode, demonstrating the successful preparation of PTAA.31Moreover, the i p of the pine PTAA electrode in Figure2Bwas larger and grew faster with the electrochemical polymerization of 3-TAA compared to the PTAA formed on the ordinary gold electrode, indicating a larger surface area of pine PTAA.In addition, the morphology of the pine-like gold was observed via scanning electron microscopy imaging in FigureS3with a branch size of about 100-500 nm.Compared with pure PTAA, pine PTAA exhibited better electrochemical performance due to its high conductivity and large active specific surface area.The aforementioned electrodes were activated by EDC/NHS in MES buffer, followed by the connection to amine-modified caffeine through amidation reaction.Successful step immobilization and the interface properties of the caffeine-based sensors were verified by electrochemical measurements using the ternary electrode method, as shown in FigureS1.In order to study the interaction mode between caffeine and human Alzheimer's β-amyloid protein, molecular docking was performed using AutoDock tool (version 1.5.6), and the

F
I G U R E 3 (A) Electrochemical impedance spectroscopy (EIS) of the mercaptopropionic acid (MPA)-based biosensor after incubation in a series of concentrations of Aβ oligomer (AβO) solution (10 −5 to 10 nM).(B) EIS of the poly(thiophene-3-acetic acid) (PTAA)-based biosensor after incubation in various concentrations of AβO solution (10 −6 to 1000 nM).(C) EIS of the pine PTAA-based biosensor after incubation in a constant concentration of AβO solution (10 −8 to 1000 nM).(D) Linear relationship between resistance changes and logarithm of AβO concentration obtained from (A)-(C).Error bars represent mean ± SD, where n = 3 replicates.

F I G U R E 4
(A) Nyquist plot and (B) histogram of Aβ oligomer (AβO) detection in blood (BLD) simulation fluid.Error bars represent mean ± SD, where n = 3 replicates.
This work was financially supported by the National Natural Science Foundation of China (82003150 and 32101153), the China Postdoctoral Science Foundation (2023M730260), Shanghai Medical Innovation Project (21Y11905800), the "Chenguang Program" supported by Shanghai Education Development Foundation & Shanghai Municipal Education Commission (20CG25), and Beijing Institute of Technology Research Fund Program for Young Scholars.