A New Perspective on Using Glycols in Glutamate Biosensor Design: From Stabilizing Agents to a New Containment Net

Glutamate is a major excitatory neurotransmitter in the brain. It is involved in many normal physiological brain activities, but also neurological disorders and excitotoxicity. Hence, glutamate measurement is important both in clinical and pre-clinical studies. Pre-clinical studies often use amperometric biosensors due to their low invasiveness and the relatively small size of the devices. These devices also provide fast, real-time measurements because of their high sensitivity. In the present study, diethylene glycol (DEG), neopentyl glycol (NPG), triethylene glycol (TEG), and glycerol (GLY) were used to increase the long-term stability of glutamate biosensors. The evaluation was made by measuring variations of the main enzymatic (VMAX and KM) and analytical (Linear Region Slope (LRS)) parameters. Of the glycols tested, TEG was the most promising stabilizer, showing about twice as high VMAX maintained over a greater duration than with other stabilizers tested. It is also yielded the most stable linear region slope (LRS) values over the study duration. Moreover, we highlighted the ability of glycols to interact with enzyme molecules to form a containment network, able to maintain all the layered components of the biosensor adhering to the transducer.


Introduction
Glutamate is a major excitatory neurotransmitter in the brain [1,2] implicated in normal physiological brain activities [3,4]. These activities include neurotransmission, learning and memory processes, synaptic plasticity, neuronal development, and aging [5,6]. However, glutamate is also implicated in dysfunction, with high levels of glutamate associated with neuronal death [7] and neurological disorders [8][9][10]. Glutamate is involved in excitotoxicity as well, due to an over-activation of glutamate receptors [11,12]. Given the importance of glutamate in both normal and abnormal brain function, the measurement of glutamate levels is relevant in both clinical and pre-clinical studies.
In clinical investigations, various techniques have been used to measure glutamate, including spectroscopic and neuroimaging techniques [13,14] used to correlate glutamate levels to brain L-Glutamate + H 2 O + GluOx/FAD − → α-ketoglutarate + NH 3 + GluOx/FADH 2 (1) Hydrogen peroxide (H 2 O 2 ), which is generated during the enzymatic reaction, is measured on a platinum (Pt) surface when an anodic potential of +0.7 V vs. Ag/AgCl is applied [29][30][31][32]. The current generated from H 2 O 2 oxidation is proportional to the glutamate concentration in the matrix.
Sensitivity, efficiency, and performance over time are crucial and continually improving with new biosensor designs [24,35,36]. The most common approach to improve the efficiency of biosensors is the use of polyethyleneimine (PEI) [24,25,[28][29][30]37]. Very recently, it has been demonstrated that glycols (and polyols in general) can improve biosensor efficiency in the presence of PEI, as well as increase stability over time [24,38]. Indeed, it has been shown that propylene glycol (PG) improves glutamate biosensor activity over time [24]. Some polyols can act as enzyme stabilizers by interacting with proteins to control the enzyme microenvironment and protect against denaturation [24,39].
Different parameters were considered to evaluate the stability of the glutamate biosensors over time. Among the most important evaluated was V MAX, which represents the maximum enzymatic rate at which substrate is converted into end products, when all catalytic sites are saturated, and enzymes can form an enzyme-substrate complex [24,35,38,40,41]. V MAX indicates the number of active enzyme molecules loaded on the biosensor surface [24,34,35,37,38,42].
In addition, the Michaelis-Menten constant, K M , was determined as this parameter is correlated with the affinity between the enzyme and substrate, giving the substrate concentration that produces half of the V MAX [25,34,35]. K M variations have been linked to the linear operating range (equal to about a half of K M ). Indeed, some authors have demonstrated that a high value of K M produces a broad linear range for biosensors [30,35,43]. Moreover, K M is essential for determining one of the most critical analytical parameters: the linear region slope (LRS). LRS indicates the sensitivity of the biosensor for the substrate and can be considered a K M function [35,43,44].
In this study, different glycols were examined for their ability to improve the sensitivity and stability over time of glutamate biosensors. The compounds were chosen based on their different structures and possible interactions with GluOx.

Instrumentation and Software
All electrochemical experiments were performed using a classical three-electrode cell. It consisted of a beaker holding 20 mL of PBS, four glutamate biosensors as working electrodes, an Ag/AgCl (NaCl 3M) electrode (Bioanalytical Systems, Inc., West Lafayette, IN, USA), and a stainless needle as the auxiliary electrode. A four-channel potentiostat (eDAQ Quadstat, e-Corder 410, eDAQ Europe, Poland) and the software Chart (v 5.5, eDAQ Europe, Poland) were employed for all electrochemical procedures.

Biosensor Construction and Characterization
All biosensors were constructed using the same cylindrical geometry (1 mm in length and 125 µm in diameter) ( Figure 1). Briefly, a 3 cm portion of Pt/Ir wire was cut, and from one edge, 3 mm of Teflon ® insulation was removed to allow the bare metal to be welded to a support. From the other edge of the wire, 1 mm of bare metal was exposed for modifications. On Day 0, ortho-phenylenediamine polymer (PPD) electropolymerization was carried out by immersing the Pt/Ir wires in an OPD 0.30 M solution prepared in deoxygenated PBS (N 2 100%). It was prepared by bubbling ultrapure N 2 gas in 12 mL of PBS for 30 min, and by applying a positive potential of +0.7 V vs. Ag/AgCl electrode (RE) for 30 min. After polymerization, Pt/p-OPD cylinders were rinsed in pure water and dipped twice in a PEI 1% solution, waiting 5 min between each dipping. At that point, electrodes were dipped in the GlutOx solution (400 U/mL) for five times, leaving to dry for 5 min at room temperature between each dip. Finally, one layer of 0.1% glycol was deposited on the top of each biosensor using a single dip, after which the biosensors were left to dry at room temperature for 30 min. Depending on the glycol used, different biosensors were manufactured with the following general design: where Glycol is represented by DEG, NPG, TEG, and GLY.
Chemosensors 2020, 8, 23 4 of 16 for 5 minutes at room temperature between each dip. Finally, one layer of 0.1% glycol was deposited on the top of each biosensor using a single dip, after which the biosensors were left to dry at room temperature for 30 minutes. Depending on the glycol used, different biosensors were manufactured with the following general design: where Glycol is represented by DEG, NPG, TEG, and GLY. Moreover, a control biosensor design was constructed, where the final coating was omitted: Ptc/PPD/PEI(1%)2/GlutOx5 Biosensors were put in 20 ml of fresh PBS and then polarized at a constant potential of +0.7 V vs. Ag/AgCl overnight for stabilization. On Day 1, a full glutamate calibration ranging from 0 to 50 mM was conducted in fresh PBS by the addition of fixed volumes of Glut stock solutions (0.01 M and 1 M) ( Figure 2). In Figure S1 of the Supplementary Material, calibration plots are shown, from 21 days, of the two representative glycol designs (DEG and TEG). All designs were subjected to the same calibration protocol, which was repeated on Day 7, 14, and 21 to assess biosensor aging. Biosensor performances were evaluated from calibration data in terms of enzymatic kinetic (VMAX and KM) and analytical performances (linear region slope-LRS and LOD). After each calibration, all biosensors were rinsed in double-distilled water and stored at +4 °C.
All the electrochemical procedures as electropolymerization or calibrations were carried out using constant potential amperometry (CPA). Moreover, a control biosensor design was constructed, where the final coating was omitted:

Statistical Analysis
Biosensors were put in 20 mL of fresh PBS and then polarized at a constant potential of +0.7 V vs. Ag/AgCl overnight for stabilization. On Day 1, a full glutamate calibration ranging from 0 to 50 mM was conducted in fresh PBS by the addition of fixed volumes of Glut stock solutions (0.01 M and 1 M) ( Figure 2). In Figure S1 of the Supplementary Material, calibration plots are shown, from 21 days, of the two representative glycol designs (DEG and TEG). All designs were subjected to the same calibration protocol, which was repeated on Day 7, 14, and 21 to assess biosensor aging. Biosensor performances were evaluated from calibration data in terms of enzymatic kinetic (V MAX and K M ) and analytical performances (linear region slope-LRS and LOD). After each calibration, all biosensors were rinsed in double-distilled water and stored at +4 • C.

Results
As shown in Figure 2, kinetic parameters (VMAX and KM) were calculated using nonlinear fitting of calibrations data in the range of 0-50 μM of glutamate (Panel A). Meanwhile, the analytical parameter LRS was obtained by means of linear fitting of calibration data in the range between 0 and 200 μM of glutamate (Panel B). In the inset of Panel B, the graph inherent in the concentration range 0-10 μM of glutamate is highlighted. All the electrochemical procedures as electropolymerization or calibrations were carried out using constant potential amperometry (CPA).

Statistical Analysis
After in vitro calibrations, biosensor currents were plotted versus Glut concentrations. Linear regression was calculated at low Glut concentrations (0-0.2 mM), to determine the value of LRS, while a nonlinear fitting (Michaelis-Menten equation) was performed on the entire concentration range (0-50 mM) to evaluate V MAX and K M . While V MAX was expressed as nA, the LRS was expressed as current density (nA µM −1 cm −2 ). Statistical significance (p values) was evaluated by ANOVA using GraphPad Prism v 5.02 software.
As previously published [29], the limit of detection (LOD) was determined using a statistical method based on the standard deviation (σ) of the response and the LRS of the calibration curve.

Scanning Electron Microscopy
Biosensors were also examined by scanning electron microscopy (SEM). After construction, biosensors with different glycols were immobilized on a holder for gold coating. As previously published [38], after plating, microphotographs were collected using a DSM 962 Zeiss conventional SEM (Oberkochen, Germany) by applying an accelerating voltage of 25 kV. Microphotographs were taken at different magnifications of 2000, 5000, and 1000×.

Results
As shown in Figure 2, kinetic parameters (V MAX and K M ) were calculated using nonlinear fitting of calibrations data in the range of 0-50 µM of glutamate (Panel A). Meanwhile, the analytical parameter LRS was obtained by means of linear fitting of calibration data in the range between 0 and 200 µM of glutamate (Panel B). In the inset of Panel B, the graph inherent in the concentration range 0-10 µM of glutamate is highlighted.      TEG biosensors produced the highest values of V MAX between Day 1 and 7 when the parameter was equal to 317.10 ± 5.46 nA (not statistically different from Day 1). Over time, V MAX decreased moderately but significantly (p < 0.001 vs. Day 1) on both Days 14 and 21, to a minimum of 221.50 ± 4.98 nA.

Impact of Glycols on VMAX Over Time
GLY-loaded biosensors generated a V MAX of 222.20 ± 6.21 nA at Day 1, and a constant significant decrease (p < 0.001 vs. Day 1) was observed over time to 142.50 ± 5.22 nA at Day 21.
All glycol-containing designs showed a significant difference compared to the glycol-free design in terms of V MAX (p < 0.001), over 21 days.  All glycol-containing designs showed a significant difference compared to the glycol-free design in terms of VMAX (p < 0.001), over 21 days.

Impact of Glycols on Analytical Parameters Over Time
In Figure 5, the variation of LRS over time of different biosensor designs is shown and expressed as a function of current density. The control design showed the lowest values of LRS, from Day 1 (36.346 ± 0.487 nA µM −1 cm −2 ). Moreover, the absence of glycol resulted in a significant decrease, starting from Day 7 (6.121 ± 0.268 nA µM −1 cm −2 ) to Day 21 (2.366 ± 0.244 nA µM −1 cm −2 ).   Comparisons between the different designs demonstrated a significant difference compared to the glycol-free design in terms of V MAX (p < 0.001).
As shown in Figure 6, and as expected, the control biosensor without any glycol had the best LOD on Day 1 (0.164 ± 0.052 µM), but significant decreases over time from Day 7 (p < 0.001 vs. Day 1), with the LOD increasing to 3.256 ± 1.053 µM by Day 21. As shown in Figure 6, and as expected, the control biosensor without any glycol had the best LOD on Day 1 (0.164 ± 0.052 μM), but significant decreases over time from Day 7 (p < 0.001 vs. Day 1), with the LOD increasing to 3.256 ± 1.053 μM by Day 21.
Designs with glycols as the final layer had higher LODs on Day 1 (average value = 0.450 μM). However, all but the DEG design had better LODs over time than the control design.
For the NPG design, the LOD was 0.780 ± 0.042 μM and 0.915 ± 0.092 μM on Days 14 and 21, respectively. A slight decrease relative to Day 1, although not significant, was recorded on Day 7.
With regard to TEG and GLY, there were no significant differences in the LOD values over the observation period, with LODs of about 0.400 μM and 0.497 μM, respectively. Designs with glycols as the final layer had higher LODs on Day 1 (average value = 0.450 µM). However, all but the DEG design had better LODs over time than the control design.

Scanning Electron Microscopy (SEM) Analysis
For the NPG design, the LOD was 0.780 ± 0.042 µM and 0.915 ± 0.092 µM on Days 14 and 21, respectively. A slight decrease relative to Day 1, although not significant, was recorded on Day 7.
With regard to TEG and GLY, there were no significant differences in the LOD values over the observation period, with LODs of about 0.400 µM and 0.497 µM, respectively.

Scanning Electron Microscopy (SEM) Analysis
Following previously published studies [38,45,46], the designs were further characterized by SEM to examine surface characteristics. Microphotographs were taken with a conventional SEM, at Day 1 after construction, with a magnification of 2000×, as shown in Figure 7. Microphotographs with the magnifications of 5000 and 1000× are reported in Figure S2   Following previously published studies [38,45,46], the designs were further characterized by SEM to examine surface characteristics. Microphotographs were taken with a conventional SEM, at Day 1 after construction, with a magnification of 2000x, as shown in Figure 7. Microphotographs with the magnifications of 5000 and 1000x are reported in Figure S2 of the Supplementary Material.
As shown in Figure 6, all the biosensor's surfaces differed from bare Pt (Panel A) because of the presence of different layered components. In panel B, surface modification due to the presence of PPD was evident, which was consistent with previous findings [38]. In panels C, D, E, and F, the deposition of PEI, GlutOx, and glycols resulted in the formation of craters on the surface, with a peculiar size and distribution for each design. DEG and NPG (Panel C and D of Figure 7, respectively) showed craters on average that were smaller than TEG and GLY (Panel E and F of Figure 7, respectively). A closer look at these cavities ( Figure S2) suggested a degree of depth greater for DEG  Figure 6, all the biosensor's surfaces differed from bare Pt (Panel A) because of the presence of different layered components. In panel B, surface modification due to the presence of PPD was evident, which was consistent with previous findings [38]. In panels C, D, E, and F, the deposition of PEI, GlutOx, and glycols resulted in the formation of craters on the surface, with a peculiar size and distribution for each design. DEG and NPG (Panel C and D of Figure 7, respectively) showed craters on average that were smaller than TEG and GLY (Panel E and F of Figure 7, respectively). A closer look at these cavities ( Figure S2) suggested a degree of depth greater for DEG (Panels 1A and 1B from Figure S2) than all the others (Panels 2A-B, Panels 3A-B, Panels 4A-B of Figure S2). Moreover, DEG (Panel C), together with TEG (Panel E) had better-sharpened edges compared to NPG (Panel D) and GLY (Panel F).

Discussion
In the present study, the impact of different glycols on glutamate biosensor activity and stability over time was studied to develop better biosensors for in vivo applications. As previously described [24,38], glycols, and polyols in general, stabilize enzyme activity by forming a stable hydration layer around the enzyme molecules, which maintains the native enzyme conformation (folded state) [38,47,48] and diminishes unfavorable interactions between the enzyme and surrounding solution [38,49]. This is important for biosensor stability as enzyme activity is the first parameter to suffer from prolonged biosensor use and enzyme denaturation during immobilization for biosensor manufacture [34,50]. The loss of enzyme activity could also be due to a rearrangement over time within the biosensor, which could result in less favorable interactions with the substrate over time, as the enzymes are not covalently linked to the biosensor platinum surface [38]. V MAX reflects the number of active molecules present on the transducer surface [24,34,35,37,38,42], so an increase (or a decrease) in V MAX reflects a variation in enzyme activity.
In our experiments, a control design was investigated, which lacked glycol as the final layer. As expected, in this design, the V MAX decreased from Day 7, demonstrating that the lack of a containment network did not prevent the loss of enzymatic activity, causing a sudden loss of the biosensor performances.
In this study, all the glycol designs showed better V MAX than the control. While DEG and NPG showed a consistent loss of enzymatic activity (with a sustained decrease in V MAX ), TEG and GLY showed a lesser decrease. In particular, TEG showed a slight (although not significant) increase between Day 1 and 7 and the smallest pronounced decrease among all designs. The obtained V MAX values are in line with measurements of a comparable factor J max in the literature [33], where a conversion could easily be obtained considering the surface area of the 1 mm Pt/Ir cylinder (4.05 × 10 −3 cm 2 ). These data confirmed that glycols could preserve enzymatic activity, in particular, TEG. Notably, all glycol compounds tested were monomeric and easier to handle than PG [38]. Their deposition increased the retention of enzyme molecules on the biosensor surface.
Usually, containment net or crosslinking agents are used to entrap layered components on glutamate biosensor surfaces [24,25,37,51]. Aggressive crosslinking agents are sometimes employed, such as glutaraldehyde, that can affect biosensor enzyme performance over time. Very recently [24], it has been found that PG, a small monomeric molecule never used before in biosensing, exhibits the same characteristics as the polyethylene glycol diglycidyl ether (PEDGE). This has been widely used as a cross-linking agent for biosensors fabrication [24,25]. In the present study, some of the glycols tested showed a behavior similar to PG. Their deposition plausibly created a containment net, likely due to the intramolecular glycol-protein interactions, able to maintain the layered compound on the transducer surface, improving enzyme performance.
In the past, glycerol has been shown to stabilize proteins by influencing the hydration environment, as well as establishing hydrogen bonds due to interaction of the outer oxygen atoms of glycerol with nitrogen and oxygen atoms in the amino acids composing proteins [52]. The study identified glycerol-protein interactions even at the hydrophobic surface of proteins. These interactions are more pronounced with less hydrophilic glycerol. The occurrence of hydrogen bonds, which are characterized by weak to moderate stabilization energies [53] aside from hydrophobic interactions, is likely responsible for the formation of a glycol-protein network. This network can maintain biosensor components on the transducer surface.
Among the glycols used here, GLY and TEG had the most favorable impact on V MAX (highest values) while also exhibiting the smallest decrease over time. In particular, for TEG, after an initial (although not significant increase) on Day 7, V MAX values remained relatively high to Day 21, indicating good enzymatic activity. It has recently been shown that TEG can establish hydrogen bonds not only with terminal OH groups but also with ether oxygens atoms [54], resulting in higher enzyme stability over time. Furthermore, the structural flexibility of TEG, and likely, the molecular length allows this compound to be positioned between the enzyme amino acids to maximize interactions, including simultaneous hydrophilic (such as hydrogen bonds) and hydrophobic interactions. Such interactions could result in high and persistent values of V MAX over time.
Concerning K M , the deposition of glycols influenced the affinity of the enzyme for the substrate. In general, an increase over time was observed, as expected. Surprisingly, NPG and TEG showed relative stability in K M values over 14 days, more likely due to stronger intramolecular forces established by these glycols with the enzyme compared to the other glycols. Once again, the absence of a containment net led to higher K M values than in the control design, and therefore, to worse performance compared to designs with glycols.
As previously noted [35,43,44], LRS is one of the essential analytical parameters. It reflects biosensor efficiency in converting the substrate into byproducts but also represents sensitivity to the analyte. LRS is a function of V MAX and K M [43], so these two parameters affect the LRS.
As expected, and highlighted above, the absence of a containment network resulted in a loss of active enzymes on the transducer surface. This phenomenon had an important impact on LRS values. In fact, in the control design in which glycol was absent, from Day 7 and up to Day 21, there was a significant decrease in LRS. Almost all of the glycols used here yielded interesting LRS values, except for DEG, which revealed quite low values and was particularly unstable over time. Other glycols yielded quite high LRS values, which decreased with time, except for TEG, which surprisingly produced stable LRS values over the study period. The LRS values were found to be in line with the values already present in the literature [25].
Glycols had an interesting effect on the LOD values. All glycols increased LOD values on Day 1 when compared to the glycol-free design. Biosensors without glycol exhibited increasing LODs starting from Day 7 until Day 21. Among the glycols, only DEG and NPG showed a comparable behavior to the glycol-free design. At the same time, TEG and GLY gave LOD values that did not significantly change over the study period, demonstrating good analytical performance.
The behavior of biosensors with DEG over time is probably due to the few, and not very effective, interactions between glycol and the enzyme. This indicates that hydrophobicity and an appropriate molecular length glycol can positively influence biosensor performance. DEG-protein interactions did not allow protection of enzyme molecules, nor their effective orientation in space, reducing the enzyme-substrate interactions, as denoted by the low values of V MAX and LRS, and the high values of K M . Moreover, DEG did not have adequate stabilizing effects.
Biosensors with NPG and GLY showed comparable behavior over time. The relatively higher values of V MAX and LRS indicated enzyme stabilization, suggesting efficacious intermolecular interactions, maintaining a higher number of active molecules on the transducer surface. These phenomena may be attributed to the ability of the glycols to establish effective hydrogen bonds through the OH moieties, but also to create hydrophobic connections, the latter being more pronounced in NPG.
Surprisingly, the deposition of TEG revealed interesting results in terms of enzymatic kinetics (V MAX and K M ) and analytical efficacy (LRS and LOD) over time. The high number of active molecules and high sensitivity, as well as the constant enzyme affinity, reflect the elevated capability of TEG to stabilize enzymatic activity, over the time frame studied. Moreover, the persistence of many active enzyme molecules that were well-oriented for interaction with the substrate could be due to hydrophilic and hydrophobic interactions of TEG through OH moieties and ether oxygens [55,56]. Moreover, the structural flexibility of TEG probably allows better interactions with the amino acids of the enzyme.
High vacuum conditions used for SEM imaging strongly influenced the presence of water in the sample, thereby altering hydrophilic interactions. Since glycols form the external layer, enzyme and PEI are mediated by water molecules, and H 2 O removal can lead to a collapse of the network. In this way, craters observed on biosensor surfaces would represent the tears of the texture. On the other hand, hydrophobic interactions can be maintained even under the conditions used for SEM. Among all investigated glycols, NPG is the most hydrophobic molecule, and this characteristic can explain the blunt or unsharpened surface (Figure 7, Panel D). NPG and GLY share a more marked propensity to three-dimensionality compared to linear DEG and TEG, noting that NPG is also the most hindered molecule. The larger molecular volume of NPG and relative three-dimensionality of GLY could hamper a uniform, well-sharpened coating. Nevertheless, the high hydrophobicity of NPG provided enzymatic stability even seven days after biosensor construction. The deposition of TEG (Panel E of Figure 7 and Panels 3A,3B of Figure S2 of the Supplementary material) resembles a tight overlapping of distinct pages, composed of an underlying and uniform PPD and the collapsed cratered network. In SEM images of the TEG biosensors, the glycol appeared tightly embedded in the matrix, making it potentially better able to stabilize the enzymes.

Conclusions
The stability of biosensor analytical parameters over time are essential features when developing biosensors for in vivo use. Based on the results obtained in the present study, glycols and TEG, in particular, can be used to improve biosensor performance. The results obtained here for glutamate biosensors could also be confirmed for other enzymatic biosensors, such as glucose and lactate biosensors. Studies of these biosensors are in progress already. Work is also ongoing to evaluate the ability of the containment net created by interacting glycols and proteins to protect the biosensors from extracellular enzymes, such as proteases, during in vivo applications.