Development of an Extended Gate Field Effect Transistor Enzymatic Sensor to Monitor Glucose in Human Plasma

Abstract

Integration of direct electron transfer-type (DET-type) Burkholderia cepacia glucose dehydrogenase (BcGDH) with an extended gate field effect transistor (EGFET) transducer to measure glucose in human plasma is a promising approach to overcome technology limitations in commercial continuous glucose monitors (CGM). Sensors were fabricated using microwire electrodes and were characterized for selectivity against interferents, reversibility, stability, and validated ex vivo. DET-type EGFET sensors showed low signal bias against a variety of interfering compounds and demonstrated acute reversibility and stability, while also successfully measuring glucose ex vivo in human plasma with a limit of detection of 0.94 mM. The DET-type EGFET glucose sensor was operated ex vivo over a physiological concentration range, demonstrating the feasibility of using EGFET-based transduction of DET-BcGDH for future use in CGM applications.

Introduction

The state of continuous glucose monitoring (CGM) has been greatly advanced using novel electrochemical architectures, enzyme engineering, and predictive algorithms.^1-10 Several such architectures involve the use of field effect transistors (FETs); early electrochemical FETs allowed for analyte quantification based on changes in a transistor’s drain current (I_D) resulting from reactions at the electrode surface occurring on or near the semiconducting channel spanning the source and drain of the FET.^11-17 Recently, extended gate FETs (EGFETs) have been used to measure a variety of small molecules, such as cortisol and serotonin. Unlike other electrochemical sensors using FETs, EGFETs use a conventional solid-state, semiconductor-based FET that is physically separate from the electrochemical cell where the interaction between the analyte and the electrode occurs. The FET gate terminal is electrically connected to a working/gate electrode immersed in the electrochemical cell, which is the site of analyte recognition. A potential (relative to the transistor source terminal) is applied to both the transistor drain terminal and the reference electrode (commonly Ag/AgCl), to bias the transistor and facilitate measurement. Analyte interactions occurring on the working electrode surface generate changes to the working electrode junction potential, which shifts the FET gate-source potential, thereby altering the channel conductivity and the measured drain current.

One engineered enzyme of great interest to glucose biosensing, direct electron transfer-type Burkholderia cepacia-derived glucose dehydrogenase (DET-type BcGDH), has been used extensively for glucose monitoring; prior use of this enzyme is comprehensively documented in earlier works.^9,10,18,19 Here, we characterize the ability of a DET-BcGDH-integrated EGFET sensor to measure physiological concentrations of glucose in various real and simulated biological media including artificial serum, potassium phosphate buffer (PPB), and ex vivo human plasma. In addition, we investigate the impact of key interferent species on sensor operation and interrogate the stability and reversibility of the sensor. Integrating BcGDH into an EGFET sensor has several key advantages compared to previously described glucose sensors using BcGDH. The primary advantage is that the EGFET is tailored to both amplify changes in the junction potential while also transducing the signal into a simple change in current. This enables the advantages of potentiometric sensors (ie, size independent signal), ²⁰ while amplifying the signal to increase the signal-to-noise ratio as compared to conventional glucose sensors that employ BcGDH. ²¹ To the best of our knowledge, this is the first demonstration of either using EGFET or potentiometry with BcGDH for the detection of glucose in human plasma. This work builds on previous efforts in developing glucose sensors using BcGHD with various electrochemical modalities including chronoamperometry, impedimetric, and open circuit potential.^5,21-31

Figure 1 shows the scheme of the proposed EGFET-based sensor: a bias potential is applied to the reference electrode (V_Ref), which, along with the junction potential of the working electrode, establishes a potential between the FET gate and source terminals (V_GS). Upon the addition of glucose, DET-BcGDH is reduced, changing the junction potential of the working electrode. This shifts V_GS, thereby changing the measured drain current (I_D) given a specific applied drain-source potential (V_DS). ²¹

Figure 1.

Scheme of extended gate field effect transistor glucose sensor leveraging direct electron transfer-type glucose dehydrogenase.

Materials and Methods

For detailed materials and methods, please refer to the Supplemental Information. Briefly, BcGDH was produced and purified according to the method described in previous studies.^33,34 Two electrode form factors were prepared: gold disk electrodes (GDE) and gold microwire (GMW) electrodes. Dithiobis(succinimidyl hexanoate) (DSH) was immobilized at 100 µM onto both systems overnight, followed by overnight immobilization of 0.014 mg/mL BcGDH. Prior to OCP measurement, 0.2 V vs Ag/AgCl was applied over the working electrode for 30 seconds to initialize the electrode. ²⁶ Following this potential bias, the OCP between the working electrode and Ag/AgCl reference electrode was measured over a three-minute period, with OCP values being collected and analyzed at the end of the three minutes. EGFET sensors were characterized and tested using a commercial N-channel, enhancement-type FET (ZVNL120A, Diodes Incorporated, TX, USA) as previously described by Probst et al. ²¹ Briefly, a potential of 0.5 V was held between the source and drain terminal (V_DS), while the potential between the Ag/AgCl reference electrode and FET source terminal (V_Ref) was either varied from 0 V to 1 V at a sweep rate of 25 mV/sec or held constant at 1 V. Electrochemical evaluations were performed using cells containing 10 mL of solution, which were actively mixed using a magnetic stir rod at 250 RPM. Evaluation in buffer was performed using a 3-electrode configuration, using a platinum counter electrode and an Ag/AgCl reference electrode. All experiments were performed at room temperature.

Results and Discussion

Characterization of DET-Type EGFET Sensor in Artificial Serum

The DET-type EGFET sensor was evaluated using artificial serum and GDE working electrodes, either holding V_Ref at a constant potential or utilizing a potential sweep. Figure 2 shows the sensor responding in a glucose-dependent manner, including changes to I_D over time (Figure 2a) and a calibration curve generated from these data (Figure 2b).

Figure 2.

(a) EGFET I_D response to sequential glucose additions in artificial serum, under a constant applied V_Ref. (b) Derived calibration curve between glucose and I_D tested under constant V_Ref. Standard error was calculated from three replicates.

Further characterization of the sensor investigated its reversibility, stability in artificial serum, the impact of various interferent species, and the impact of the ionic conductivity of the electrolyte solution (Figure 3a-d). The sensor retains nearly 100% reversibility over several cycles alternating between a glucose-free blank and 20 mM glucose (both in 100 mM PPB; Figure 3a). This reversibility is attributed to the gate-source leakage current of the ZVNL120A transistor (approximately 20 nA), which equates to about 61 picomoles of enzyme being re-oxidized over the course of a five-minute period. Beyond this, there was no significant change in the overall signal response between each cycle (such as a visible lag or variations in noise; Supplemental Figure S3a-d). The sensor showed adequate stability over 24 hours of continuous use in artificial serum spiked with 20 mM glucose, with V_Ref and V_DS being held constant at 1 V and 0.5 V, respectively. The total drift in artificial serum over 24 h was 10.7%, or 4.6 µA/h (equivalent to a 0.03 mM change in effective glucose readout). Several interferent species were added to artificial serum to examine the non-specific response of the sensor to species other than glucose, including 10 mM KCl, 10 mM lactate, 400 µM ascorbic acid, and 200 µM acetaminophen, in order to mimic concentrations likely to be encountered in a physiological context. ³⁵ These interferents were selected to produce specific types of signal artifact, including direct interference from electroactive molecules which may spontaneously oxidize at the working electrode surface (ascorbic acid and acetaminophen), as well as ions which may adversely influence the double layer (lactate) or the reference electrode (KCl). Each interferent addition produced a change to the baseline signal of less than 10%, suggesting little impact on the specific measurement. While other interferent species (including uric acid, dopamine, and structurally similar sugars) may impact the measured signal, there are several pathways to mitigate this interference; this includes the addition of a Nafion outer membrane, a strategy which has previously been employed with EGFET sensors.^26,36-38 The impact of electrolyte concentration was evaluated due to the potential effect of changes in the solution’s ionic composition on the length of the Debye layer of the working electrode (and thereby its junction potential).^39-41 Figure 3d shows the effects of low versus high ionic strength buffer (10 vs 100 mM PPB) on the sensor response. As the ionic strength increases, the sensitivity of the sensor will decrease due to the shrinkage of the Debye layer. ⁴² While this was observed to impact the signal, there still remained a strong glucose-dependent response across the physiological window; further, physiological ion concentrations are generally stable in vivo, with moderate fluctuations. ⁴³

Figure 3.

(a) Reversibility of the electrode response over three cycles alternating between 20 mM and 0 mM glucose. (b) Twenty-four-hour stability of the sensor in artificial serum. (c) Impact of various electroactive interferents on measured I_D in 3 mM glucose. (d) Impact of solution ionic strength on the EGFET response using high ionic strength (100 mM PPB) and low ionic strength (10 mM PPB) electrolyte. Standard deviation was calculated from three replicates.

EGFET- and OCP-Based Sensor Testing in Human Plasma

A two-electrode configuration was used for the evaluation of the OCP and EGFET sensors using GMW working electrodes. The GMW form factor was selected due to being a practical and translational approach for future in vivo studies. This has been demonstrated in both academic and commercial settings for glucose sensing and for other biosensors such as electrochemical aptamer sensors.^44,45

The OCP and EGFET response was compared in 100 mM PPB, showing a small shift in signal but maintaining a similar glucose-dependent response across the physiological glucose range (Supplemental Figure S4a and b). The sensor was then evaluated in human plasma spiked with glucose across a relevant physiological range (3-23 mM). Figure 4 depicts the EGFET response obtained using three distinct electrodes, showing a slope of −0.226 mA/decade glucose, and a limit of detection of 0.94 mM within the linear region. OCP measurement of each plasma sample was used as an additional control, parallel to EGFET measurements (Supplemental Figure S5). OCP configuration gave a sensitivity of −11.67 mV/decade glucose, with a calculated limit of detection of 1.3 mM.

Figure 4.

EGFET responses using microwire electrodes in human plasma, including an I_D calibration curve over the physiological glucose range. Standard deviation was determined using three distinct working and reference electrodes.

This work suggests that EGFET sensor architectures are suitable for glucose monitoring, sharing acceptable ex vivo sensitivity, specificity, and stability compared to previously demonstrated OCP glucose sensors. Human plasma was selected as a surrogate complex matrix to interstitial fluid (ISF) since sampling ISF is very challenging in practice; we anticipate that ISF measurement will be comparably achievable, based on our successful characterization in plasma. Our group previously demonstrated the feasibility of using a DET-type enzyme in an EGFET configuration, as well as the relationship between EGFET and OCP sensing. Here we have expanded upon this earlier work, demonstrating for the first time an ex vivo EGFET sensor which uses DET-type BcGDH.

Conclusion

This work represents the first attempt at quantifying glucose concentrations in ex vivo human samples using an EGFET sensor functionalized with DET-type BcGDH on a miniaturized electrode, suitable for future CGM applications. We investigated the sensor response in buffer and artificial serum over a physiological range of glucose concentrations. Additionally, we investigated the impact of several common interferent species on the measured sensor output and determined the impact of these interferents to be negligible compared to changes in the measured drain current resulting from variations in glucose concentration. This work demonstrates that this technology may be suitable for both point-of-care and in vivo sensing. There remain several broad challenges that must be addressed prior to the deployment of EGFET-based glucose sensors, including leakage into the FET gate, which may lead to drift. Reference electrode instability over prolonged sensing may also lead to shifts in V_Ref and V_DS, which could be misinterpreted as changes in glucose concentration.^46,47

Supporting Information

Supporting information includes methods for enzyme preparation, electrode fabrication, electrochemical experimental design and parameters, characterization of clean gold microwires for ex vivo applications, calibration of GLUCOCARD Shine glucometer (Arkray USA Inc., Minneapolis, MN, USA) for ex vivo glucose validation, impact of reference electrode miniaturization on signal response, and open circuit potential measurements of ex vivo samples.

Supplemental Material

sj-docx-1-dst-10.1177_19322968251384979 – Supplemental material for Development of an Extended Gate Field Effect Transistor Enzymatic Sensor to Monitor Glucose in Human Plasma

Supplemental material, sj-docx-1-dst-10.1177_19322968251384979 for Development of an Extended Gate Field Effect Transistor Enzymatic Sensor to Monitor Glucose in Human Plasma by David Probst, Jack Twiddy, Mika Hatada, Michael Daniele and Koji Sode in Journal of Diabetes Science and Technology