Molecular architecture of the immunosensor
Here, we report the development of a real-time immunosensor which can perform highly sensitive Aβ measurements. The general molecular architecture and working principle of the immunosensor is shown in Figure 1. To construct this immunosensor, three distinct antibodies were placed on each of the barrels of the carbon microelectrode. Antibodies mHJ2 placed on E1(40) and mHJ7.4 placed on E2(42) are highly specific to the detection of Aβ1–40 and Aβ1–42 respectively. Antibody mHJ5.1 which binds to the central region of the Aβ peptide can capture all Aβ isoforms and thus it act as a positive control for the immunosensor. The function of antibodies in the developed immunosensor is three-fold.
Figure 1. Schematic illustration of the microbioelectronic detection system for Alzheimer’s disease related Aβ1–40 and Aβ1–42 peptides.
Download figure to PowerPoint
Antibodies capture/bind Aβ molecules to the sensor surface which decreases the electron tunneling distance between the sensor surface and the Aβ molecules. The decrease in the tunneling distance allows for electrons generated due to the oxidation of Aβ to travel to the sensor surface during SWV and generate a current signal. Therefore, lower quantities of Aβ can also generate a measurable signal and improve sensor performance.
The capture/binding of Aβ to the surface of the carbon fiber eliminates cross-talk between the three microelectrodes during simultaneous measurements of Aβ1–40 and Aβ1–42.
Antibodies also enable highly specific detection of Aβ1–40 and Aβ1–42.
In this work, square wave voltammetry for the oxidation of Aβ was performed so that the magnitude of the resultant oxidation peak could be related to the amount of Aβ1–40/42 captured by the specific antibodies onto the electrode surface. Parameters involved in the preparation of an immunosensor play a vital role on its voltammetric response. Therefore, the immunosensor was optimized for a maximum voltammetric response by optimizing key parameters such as amount of antibody immobilized on the electrode surface and incubation time. A cyclic square wave E-field was applied to accelerate the movement of antibodies towards the electrode surface which resulted in faster immobilization time.
Optimization of antibody immobilization parameters
In the development of electrochemical immunosensors the immobilization of the bio-recognition element on the electrode is critical towards sensor performance. The surface of carbon fibers is well known to exhibits many reactive sites including hydroxylic, phenolic and carboxylic functionalities (Pantano and Kuhr 1991). To expose these groups, the carbon fiber microelectrodes are electrochemically pre-treated (Prabhulkar and Li 2010). The microelectrodes were electroactivated using a 150-mM NaCl solution (pH = 10) at 1.2 V (vs. Ag/AgCl) for 8 min at 25°C. The next step towards the construction of the immunosensor was to employ a carbodimide reaction to selectively attach the three antibodies on the surface of each of the barrels of the microelectrode. To test binding of the antibody to the electrode surface, we first conjugated horseradish peroxidase (HRP) to the antibody. HRP provides a large oxidative signal that is easily detected by the electrode; HRP-labeled antibodies were only used to test for antibody absorption to the electrode. This involved the incubation of the pre-treated microelectrodes in EDC and NHSS solutions. Initially, the EDC and NHSS solutions were prepared in PBS buffer; however, a negligible amount of antibody was bound to the electrode surface as measured by the electroactivity of the HRP label. We therefore switched the buffer to MES which showed a much higher peak from the HRP-labeled (mHJ5.1) antibodies as seen in Figure 3a.
Figure 3. (a) DPV response obtained from HRP-labeled antibodies immobilized using PBS and MES buffer solutions. (b) Bare triple barrel carbon fiber microelectrode incubated with a 1.5 μg/mL solution of HRP-labeled anti-Aβ antibody (mHJ5.1) solution. DPV was every 1 min performed using 10 mM PBS buffer and 1 mM KCl as supporting electrolyte. Oxidation peak current is plotted versus responding incubation time. (c) Effects of driving potential on the amount of antibody immobilized on electrode surface using 1000 μg/mL antibody solution and incubation time of 7 min. (d) HRP signal generated by electrode following the application of an increasing load of HRP-labeled antibody.
Download figure to PowerPoint
HRP-labeled (mHJ.2) antibodies were used to optimize the parameters of the CSW E-field which was applied to accelerate the incubation process by promoting the transport of the antibodies to the electrode surface (Wei et al. 2009). The CSW E-field creates a mass transportation effect in the incubation solution which creates good mixing in shorter time as compared with diffusion kinetics. The voltage applied to three electrodes was varied from ±0.1 V to ±0.5 V for equal positive and negative cycles. The electrodes were monitored by conducting DPV at every 1-min interval for a period of 12 min. Based on the data shown in Figure 3b and Figure 3c, the optimum applied voltage and incubation time were chosen as ±0.3 V and 7 min respectively. During the immobilization of antibodies on the immunosensor, only a fraction of the total antibody applied gets bound to the electrode surface. In order to conserve a precious resource such as anti-Aβ antibodies, the optimal concentration was established empirically. In these experiments, an electrochemically activated and pre-treated microelectrode was incubated in solutions ranging from 0.1 to 1.5 μg/mL of mHJ5.1 antibody. The binding efficiency peaked and plateaued when more than 0.8 μg/mL of mHJ5.1 antibody was applied as shown in Figure 3d. The low binding efficiency can be attributed to the presence of hydroxylic and phenolic groups in conjunction to the carboxylic groups which are not activated by EDC/NHSS and do not bind antibodies.
After completing the optimization of key steps in the molecular architecture of the sensor, we assessed the ability of the immunosensor to simultaneously quantify Aβ1–40 and Aβ1–42 in CSF solutions using SWV with 10 mM PBS and 1 mM KCl as the supporting electrolyte. Figure 4 shows the voltammetric responses obtained after the incubation of four triple barell microelectrodes with mouse CSF solutions containing 0 nM, 100 nM, 500 nM and 1 μM Aβ1–40 and Aβ1–42 in equal concentrations. In each case when the potential was scanned between 0.0 and 0.8 V, a well-defined voltammogram consisting of an oxidation peak between 0.6 and 0.65 V was obtained. The oxidation current was obtained almost instantaneously in our experiments with no delay needed for equilibrium. Furthermore, no stringent temperature control was required during electrochemical evaluation. The signal obtained for E3(TOT) which acts as the control electrode was always higher as compared with the signals obtained for E1(40) and E2(42). The response obtained for E3(TOT) correlates with the activity of the mHJ5.1 antibody which recognizes most isoforms of Aβ as it is targeted towards the central region of the peptide.
Figure 4. Square wave voltammetric responses of three barrel of the immunosensor obtained in rat (a) CSF solution, spiked with (b) 100 nM Aβ1–40 and 100 nM Aβ1–42, (c) 500 nM Aβ1–40 and 500 nM Aβ1–42, (d) 1 μM Aβ1–40 and 1 μM Aβ1–42 solutions. Square wave voltammetry parameters: Init E (V) = 1; Final E (V) = 0; Incr E (V) = 0.004; Amplitude (V) = 0.04; Frequency (Hz) = 500.
Download figure to PowerPoint
Linear calibration plots shown in Figure 5a and Figure 5b were then constructed based on peak oxidation currents obtained versus Aβ1–40 and Aβ1–42 concentrations respectively, to normalize the variations between electrodes. Based on the evaluation of the calibration plots the linearity of the immunosensor for the detection of Aβ1–40 and Aβ1–42 lies within the detection range of 20–50 nM and 20–140 nM respectively. The immunosensor demonstrates a sensitivity of 20 nM towards the detection of Aβ1–40 and Aβ1–42. The imprecision observed with the detection capabilities of the immunosensor towards Aβ1–40 and Aβ1–42 detection might be for several reasons: (i) related to the inherent immunoreaction kinetics of each antibody, (ii) differential aggregation of Aβ which could limit signal, (iii) more non-specific absorption (‘sticking’) of Aβ1–42 versus Aβ1–40 to the apparatus.
Figure 5. (a) Calibration curve: the linear relationship between peak oxidation current and Aβ1–40 concentration. (b) Calibration curve: the linear relationship between peak oxidation current and Aβ1–42 concentration. (c) Oxidation peak current obtained from three anti-Aβ modified carbon fiber microelectrodes after being incubated with 1 μM of Aβ1–40, 1 μM of Aβ1–42 and 500 nM of Aβ1–40 + 500 nM of Aβ1–42 spiked into rat CSF solutions respectively. Measurements were performed in 10 mM PBS buffer and 1 mM KCl versus Ag/AgCl.
Download figure to PowerPoint
Specificity is a crucial parameter which influences the performance of an immunosensor in real matrices, especially when an analyte at low concentrations needs to be detected in the presence of a much larger concentration of non-specific molecules. We need to prove that the presented sensor responds only to the Aβ and anti-Aβ immunoreaction and not to the non-specific adsorption of other proteins. CSF contains free Tyr and other proteins containing Tyr amino acid residues which could cause false signaling. Ethanolamine which has been proven to be a successful blocking agent in previous studies (Frederix et al. 2004) was used to minimize these non-specific binding reactions.
To measure the specificity of each of the barrels towards a specific Aβ isotope, three immunosensors were challenged using each of the following solutions: (i) 1000 nM Aβ1–40, (ii) 1000 nM Aβ1–42, and (iii) 500 nM Aβ1–40 and 500 nM Aβ1–42. SWV was performed before and after incubation of each microelectrode with analyte solution. Figure 5c shows the average values of peak oxidation currents obtained after incubation from each of the barrel of three separate electrodes incubated in each of the above mentioned solutions. From the evaluation of Figure 5c we can deduce that each of the barrels showed highly specific recognition properties. The signal obtained due the non-specific adsorption of non-Aβ proteins and non-corresponding Aβ isoforms from each of the barrels was found to be minimum and statistically insignificant.
As the electrochemical detection of Abeta is based on the electrochemical oxidation of Tyr group in Abeta, other electroactive groups, especially the Methionine 35 (Met35), located at residue 35 of Aβ(1–42), might have interfere to the oxidation current of the Tyr group. Met is an easily oxidizable amino acid and can undergo 2-electron oxidation to form methionine sulfoxide (Schöneich et al. 2003). We compared the electrochemical oxidation potential of the two electrochemical active groups. We found that the oxidation of Tyr10 residue which occurs at a potential of 0.65 V (vs. Ag/AgCl reference electrode), whereas the oxidation potential of zwitterionic Met35 at pH = 7 is −0.057 V (vs. Ag/AgCl reference electrode) (Brunelle and Rauk 2004), which is much more negative than the oxidation potential of Tyr10. Therefore, in the present study the sweeping potential window for Abeta isoforms analysis was set up in the range from 0 to 0.8 V, in which the oxidization interference of Met35 on the electrochemical current of Tyr10 should be excluded.
Another consideration is regarded to the glycosylation of Tyr10, which could affect the electrochemical properties of Tyr10 leading to the miss interpretation of the electrochemical signal of Tyr10. The glycosylation of Tyr10 was commonly observed for shorter Abeta isoforms with sizes varying from Abeta 1–15, 1–16, 1–17, 1–18, 1–19, 1–20, 3–15, 4–15, 4–17, 5–17 (Halim 2011). However, the most abundant isoforms of Abeta found in human CSF are with sizes from 1 to 38, 1 to 40, 1 to 42, which did not show glycosylation of Tyr10. The antibodies utilized in the immunosensor are specific towards the detection of Abeta-40 and Abeta-42 and do not show any significant cross-reactivity as observed from Figure 5c. Hence, as Tyr10 present in Abeta 1–40 and Abeta 1–42 does not undergo glycosylation, it should not have any effect on the immunosensor response.
The stability of the microelectrodes with anti-Aβ antibodies immobilized on their surface was evaluated over a period of 1 month. The micro-immunosensors were stored in 0.1 M PBS solution at 4°C after fabrication. The stability of the stored electrodes was studied by incubating them in freshly prepare solution containing 0.1% bovine serum albumin, 500 nM Aβ1–40, 500 nM Aβ1–42 in 0.1 M PBS at ph 7.2. The stability of the electrodes was evaluated by measuring their electrochemical response using square wave voltammetry. The electrochemical response of the microelectrodes progressively degraded after 4 days in storage. To test the inter-sensor reproducibility of the proposed system, three samples of different concentrations (100, 500 and 1000 nM) of Aβ1–40 and Aβ1–42 were tested first with different electrodes. The maximum value of the relative standard deviations was 14.2% (n = 3) for inter-assay. This indicates that our detection strategy offers an acceptable reproducibility towards the detection of Aβ1–40 and Aβ1–42.