The enzyme glucose oxidase (GOx) from Aspergillus niger (180,200 U·g⁻¹; EC 1.1.3.4, type VII-S) was obtained from Sigma-Aldrich, as were ortho-phenylenediamine (oPD), α-d-( + )-glucose, ascorbic acid (AA), hydrogen peroxide (HP, 30% w/w aqueous solution) and potassium chloride. All reagents were used as supplied. All experimental calibrations were carried out in PBS (pH 7.4) prepared by adding NaCl (Sigma, 150 mM), NaH₂PO₄ (Fluka, 40 mM) and NaOH (Fluka, 40 mM) to distilled water, bubbled with N₂ for 15 min, and stored at 4 °C. Solutions of monomer, oPD, were prepared in 25 mL of PBS, unless stated otherwise, and sonicated at room temperature until dissolved. A stock solution of 1 M glucose was prepared in distilled water and left for 24 h at room temperature to allow equilibration of the anomers, and then stored at 4 °C. Stock solutions of 10 mM HP and 100 mM AA were prepared in distilled water and 100 mM HCl, respectively.

Calibrations for HP, AA and glucose were performed in a standard three-electrode cell containing 20 mL PBS at room temperature, a saturated calomel reference electrode (SCE), a stainless steel auxiliary electrode and either bare or modified platinum-iridium (90:10) working electrodes. Constant potential amperometry was performed at an applied potential of +0.7 V versus SCE, using Chart (v 5.2) software (AD Instruments Ltd., Oxford, UK) and a low-noise potentiostat (Biostat IV, ACM Instruments, Cumbria, UK). The working electrodes were allowed to settle in quiescent PBS to give a steady background current before the addition of small known aliquots of the analyte of interest.

Cylinder electrode preparation has been described in detail recently [44]. Briefly, 125 μm diameter Teflon-coated Pt-Ir wire (90:10, Advent Research Materials Ltd., Eynsham, England) was stripped of 1 mm Teflon to expose the bare metal, which displays many of the electrochemical properties of pure Pt [44]. Electropolymerization was carried out in oPD solutions (of varied monomer concentration, background electrolyte and enzyme concentration) at +0.7 V versus SCE for 15 minutes for these Pt_C electrodes [39,41]. Three main enzyme immobilization protocols were used in this work. In the first, the enzyme was immobilized by adsorption and dip-evaporation before PoPD deposition [23]. Each electrode was dipped in a 200 U·mL⁻¹ solution of GOx for 5 minutes, allowed to dry for 5 min, and then dipped quickly into the GOx solution four more times with 5 minutes drying between each dip, followed by electropolymerization. This protocol was previously found to optimize enzyme loading for biosensors of the type Pt_C/EOx/PoPD [24]. The second design immobilized the enzyme by adsorption and dip-evaporation after PoPD deposition followed by exposure to glutaraldehyde (GA) vapour for 15 min to crosslink the enzyme [23], and are termed Pt_C/PoPD/GOx-GA. The third method used co-immobilization, whereby either 1 mg·mL⁻¹ (∼650 U·mL⁻¹; ∼5 μM) or 5 mg·mL⁻¹ GOx was dissolved in oPD, and electropolymerized at +0.7 V vs. SCE for 15 min [30,31] to give Pt_C/PoPD-GOx; see Figure 1.

First generation biosensors of the general design Pt_C/PoPD∼EOx (i.e., various configurations of enzyme deposited before, over, or simultaneously with, the PoPD layer) display Michaelis-Menten kinetics, as discussed recently [11]. A previous study has shown that substrate diffusion is not limiting for non-conducting PoPD layers incorporating enzyme [32], due to their ultrathin nature (10–30 nm [31,45,46]). Therefore, the basic Michaelis-Menten enzyme parameters used here provide more readily accessible insights into factors affecting the responsiveness of biosensors fabricated from this polymer, and avoids the use of more complex analyses such as those involving the Thiele modulus [47].

Although a ping-pong mechanism describes the enzyme interaction with the substrate and co-substrate [Equations (1) and (2)], the oxygen effect was not included in the analysis here for simplicity. This is justified on two grounds: the concentration of oxygen was fixed in the present experiments (air saturation); and oxygen effects on biosensors of these designs are minimal for the range of substrate and oxygen levels encountered during neurochemical monitoring in vivo for both glucose [23,48] and glutamate [27,28]. Thus, the one-substrate form of the Michaelis-Menten equation contains the parameters used here to compare biosensor performance [Equation (4)], where J_S is the current-density normalized, background subtracted, biosensor response to a specified concentration of enzyme substrate ([S]): (4) J s = J max 1 + K M / [ S ]

J_max is the maximum, or plateau, current density response, obtained when all enzyme sites are saturated with substrate (see Figure 1). Different values of J_max, determined under the same conditions, reflect differences in the amount of active (not total) enzyme on the surface, provided k_cat and the sensitivity of the electrode to H₂O₂ [Equation (3)] does not vary much [11]. K_M is the apparent Michaelis constant, and phenomenologically defines the concentration of substrate that gives half the J_max response. Changes in K_M are sensitive to variations in enzyme-substrate access/binding, and have been interpreted in terms of barriers to enzyme-substrate access [26,49], as it is in the present study. K_M is also useful for defining the range of the linear response to S (up to ∼½K_M), as well as determining the linear region slope (LRS), i.e., LRS ≈ J_max/K_M for a true hyperbolic response [11,27]. A plot of J_S versus [S] up to enzyme saturation therefore provides the basic kinetic parameters J_max and K_M, as well as the nonlinear coefficient of determination, R²; a similar plot up to ∼½K_M provides the analytically key substrate sensitivity parameter, LRS, and the linear coefficient of determination, R² (see Figure 1). To account for any variations in LRS caused by differences in polymer-enzyme composite (PEC) biosensor sensitivity to HP, the parameter BE% was defined as the HP-normalized LRS [Equation (5)], which can be considered to reflect the efficiency of the biosensor in converting substrate to HP [11]. This parameter also allows the efficiency of the enzyme layer to be compared across diverse biosensor designs: (5) BE % = LRS at Pt / PEC slope ( HP )   at Pt / PEC × 100 %

From the calibration plots for HP, AA and glucose, a number of parameters were calculated to quantify the performance of the different designs. The apparent analyte permeabilities to HP and AA were calculated from Equations (6) and (7) [11], which is similar to other studies [50], and discussed in detail recently [41]: (6) P ( HP ) % = slope ( HP )   at Pt / PEC slope ( HP )   at bare Pt × 100 %

The slopes (μA cm⁻² mM⁻¹) of the linear responses for HP on bare metal and on the PoPD-modified electrodes were obtained from linear regression analysis of the respective calibration plots of the steady-state HP responses versus HP concentration up to 0.1 mM, as were the AA slopes up to 1 mM for the bare electrodes. In contrast, the AA response was nonlinear and self limiting (see Figure 2), as observed previously for similar PoPD-based designs [30,41], and linked with “self-blocking” by AA-related species trapped in the polymer matrix [45]. Hence, the current density at 1 mM AA [J_AA(1 mM)] was used as a measure of the AA response (see Figure 2). This AA concentration is relevant to neurochemical applications because baseline brain AA levels are ∼0.5 mM [51], reaching millimolar levels during periods of behavioral stimulation [52–54]. The ideal values of P(HP)% and P(AA)% for biosensor applications are therefore 100% and 0%, respectively. In addition, the concentration of other electroactive interference compounds in the brain, such as dopamine and its metabolites [55], are orders of magnitudes smaller than that of AA, ensuring that their contributions to PPD-based biosensor responses are insignificant [24,25,56–58]: (7) P ( AA ) % = J AA ( 1   mM )   at Pt / PEC J AA ( 1   mM )   at bare Pt × 100 %

A polymer selectivity parameter, S% (Equation 8), has been defined as the percentage interference by AA in HP detection for equimolar concentrations [41,44], with an optimum value of 0%: (8) S % = J AA ( 1   mM )   at Pt / PEC J HP ( 1   mM )   at Pt / PEC × 100 %

Although S% has been a useful parameter for gaining insights into the performance of the enzyme-free electrosynthesized polymer [11], it is not a sufficient index of the selectivity of PEC-based biosensors. Hence S_S%, the equimolar enzyme substrate selectivity parameter [Equation (9), e.g., S_G% for glucose], has also been described [11], which is similar to, but more straightforward than, non-equimolar equivalents [50,59]. The ideal value of S_S% is zero, and compares the capacity of the PEC layer to generate current from the enzyme reactions, J_S [see Equations (1–3)], to the interference response produced by AA, J_AA, for 1 mM of each analyte, a concentration which is close to brain extracellular fluid values in vivo for both compounds [51,60]: (9) S s % = J AA ( 1   mM )   at Pt / PEC J s ( 1   mM )   at Pt / PEC × 100 %

The permeability and selectivity parameters, as well as BE% [Equations (5–9)], reflect intrinsic properties of PoPD which are normalized with respect to actual electrochemical surface area, rather than geometrically calculated area. All parameters were determined for individual electrodes and then averaged over populations of sensors for each design. Results are reported are mean ± standard error (SEM), with n = number of electrodes. Currents are presented as current density, calculated using the geometric area of these smooth wire electrodes. Linear and nonlinear regression analyses were performed using GraphPad Prism (version 5.02, San Diego, CA, USA). The statistical significance of variations between parameters for the different designs was calculated using Student’s two-tailed unpaired t-tests (Prism 5.02), with values of p < 0.05 considered to indicate statistical significance of the difference.

Three main methods of GOx immobilization were examined here: dip-evaporation before polymer electrosynthesis (Pt_C/GOx/PoPD), dip-evaporation over the polymer using glutaraldehyde (GA) as a crosslinker (Pt_C/PoPD/GOx-GA), and co-polymerization from the monomer solution (Pt_C/PoPD-GOx). Glucose calibrations for biosensors of all designs followed Michaelis-Menten hyperbolic behavior [Equation (4), see Figure 1]; the corresponding J_max, K_M and LRS values for biosensors fabricated using our standard background electrolyte and monomer concentration (PBS containing 300 mM oPD [24,30]) are given in Table 1, as well as parameters for the PoPD-free design, Pt_C/GOx-GA, for comparison.

The PoPD-free design (Pt_C/GOx-GA) displayed a moderate J_max value indicative of good active enzyme loading. The corresponding K_M value was the lowest of these six designs, indicating that the enzyme was readily accessible to the substrate, although values lower than 5 mM have been reported previously [66,67]. Addition of the PoPD layer after enzyme deposition (GOx/PoPD) decreased the J_max and increased the K_M, the latter indicating that PoPD hindered access of the substrate to the surface-bound enzyme, as observed previously for a PoPD layer containing the protein, bovine serum albumin, BSA [23]. The decrease in J_max following deposition on the PoPD could be due to either covering of the GOx by the polymer, or displacement of the enzyme off the surface during the PoPD deposition. The finding that the J_max for Pt_C/GOx-GA/PoPD was three-fold greater than that for Pt_C/GOx/PoPD (see Table 1) suggests that, in the absence of crosslinking with GA, much of the enzyme is removed from the surface by electro-deposition of the polymer. This interpretation is consistent with the ultrathin nature of electrosynthesized PoPD [31,45,46] which is considered not to overwhelm immobilized enzyme [31,68]; see Figure 1.

Immobilization of GOx over the PoPD layer (PoPD/GOx) showed a 7-fold increase in active enzyme loading compared with the GOx/PoPD configuration, and was indistinguishable from the PoPD-free configuration (p > 0.34). Surprisingly, the K_M value was similar for PoPD/GOx and GOx/PoPD, suggesting that enzyme-substrate binding was compromised by the presence of PoPD (see Section 3.2). In an attempt to increase GOx loading further, GOx was dip-evaporated both before and after polymer fabrication (GOx/PoPD/GOx), with no benefit obtained (Table 1). Neither was any significant enhancement in J_max observed when a 2,000 U mL⁻¹ GOx solution was used in the dip/evaporation procedure (data not shown).

In the final basic design, GOx (5 mg mL⁻¹) was dissolved in the monomer solution for co-immobilization during electropolymerization, as described previously for other conditions and electrode geometries [30–32,50]. This PoPD-GOx configuration showed the highest active enzyme loading, nearly twice the value of the PoPD-free design (p < 0.001). The mean K_M value for Pt_C/PoPD-GOx was not statistically different from the PoPD-free design (p > 0.33), indicating similar substrate-enzyme access. In addition, K_M for the PoPD-GOx configuration was significantly lower than for PoPD/GOx (p < 0.02). One speculation is that the different structure of PoPD deposited in the presence of solution GOx, which has been observed in scanning electron microscopy studies [30], is less obstructive to substrate binding, a notion supported by AA permeability data below (see Section 3.2). Irrespective of the mechanism, however, clearly the co-immobilization of GOx from the monomer solution is superior in terms of active enzyme loading and affinity (see Table 1) compared with these, and other BSA-containing, Pt_C/PoPD∼GOx biosensor designs [23]. This detailed comparison supports the protocols advanced previously for the co-immobilization of enzyme with PoPD for biosensor fabrication [30–32,48,50,68].

The linear region slope (LRS) of the glucose calibration, a parameter determined by both the J_max and K_M values [11], is a better index of the functional sensitivity of the different designs. As expected, the deposition of GOx before the polymer led to the lowest LRS sensitivity, ∼8-fold lower than incorporation of enzyme over the polymer. This trend is in line with that reported recently, where BSA was incorporated in the PoPD matrix [23]. Co-polymerization of GOx displayed the highest LRS sensitivity, twice as good as the next ranking PoPD-based biosensor, Pt_C/PoPD/GOx-GA (p < 0.003). LRS values are influenced by two main factors: the ability of the enzyme layer to convert substrate to HP (Equations (1)–(2)), and the sensitivity of the electrode to HP [Equation (3)]. This latter can be determined as the biosensor HP calibration slope, and normalization of LRS with respect to this HP slope provides an index of the efficiency of the biosensor to convert substrate to HP [BE%, Equation (5)]. As well as being of intrinsic interest, this parameter becomes of practical importance when biosensors are used in environments where HP is produced by other components in the medium, such as mitochondria in brain tissue [69]. The maximum value of BE% for the designs shown in Table 1 was ∼2% for the Pt_C/GOx-GA and Pt_C/PoPD-GOx configurations. This low value contrasts with estimates of ∼50% for PoPD-based glutamate biosensors, mainly due to the much higher affinity of glutamate oxidase for its substrate compared with the GOx-glucose system [70].

Response times were recorded in constantly stirred solution, using the PowerLab module operating at a data acquisition rate of > 100 Hz. A t_90% parameter was defined as the time taken for the analyte response to reach 90% of its maximum value from the start of the current upswing, and is similar to definitions used previously [71–73]. The co-immobilized PoPD-GOx design was used in this study because of its high enzyme loading and LRS sensitivity (Table 1). The response time for glucose was fast (t_90% = 1.7 ± 0.1 s, n = 6), with the corresponding response time for injections of HP aliquots (t_90% = 1.3 ± 0.1 s) indicating that only ∼0.4 s of the glucose response was attributable to the enzyme reactions [Equations (1–2)]. This compares favorably with a glutamate Pt_C-based biosensor where glutamate oxidase was in the PoPD layer, and the glutamate component of the response time was ∼0.6 s [27]. These results are also consistent with previous time-response studies of the PoPD-GOx system [31], and with the ultrathin nature of PoPD, electrosynthesized under non highly-acidic conditions, allowing fast interaction of the enzyme with its substrate (see Figure 1). It appears, therefore, that K_M is a much more sensitive index of hindrance in enzyme-substrate interactions for these glucose biosensors (Table 1), as in the case of PoPD-based glutamate biosensors, where little difference in response time was observed across diverse PEC configurations with largely different K_M values [27].

Good permeability of the PEC membrane to HP is important for practical first-generation biosensor designs. The apparent HP permeability, P(HP)% defined by Equation (6), was similar for all designs (Table 2), with an average value of 106 ± 8% (n = 41) which was not significantly different from the ideal value of 100% (p > 0.42). Therefore differences in the permselectivity parameter, S% [Equation (8)], across the designs should be influenced mainly by polymer interference-rejection properties. Similarly, differences in the biosensor selectivity parameter, S_S% [Equation (9)], should be due to a combination of polymer interference rejection and the ability of the PEC layer to generate HP. The finding that some P(HP)% values were greater than 100% is unexpected, but has been observed before for PoPD layers containing a variety of macromolecular modifiers [44]. These supra-optimal values may be due minor disproportionation of HP on metals, and its possible inhibition by polymer coatings [74].

The apparent AA permeability, P(AA)%, was calculated using Equation (7); all PoPD-modified designs blocked the 1 mM AA flux by ≥ 99% compared with the bare metal, with P(AA)% ≤ 1% (Table 2). The best blocking characteristics were displayed by Pt_C/PoPD. i.e., by the pure PoPD, with P(AA)% = 0.11 ± 0.02%, a value similar to that reported for PoPD deposited under the same conditions on pure, low edge-density, Pt microfiber electrodes [41]. The finding that P(AA)% for Pt_C/PoPD/GOx (0.34 ± 0.05%, p < 0.001) was significantly greater than for the native polymer (Pt_C/PoPD) suggests that the enzyme does not simply sit on top of the PoPD, but imbeds in the polymer, opening its structure and undermining its interference blocking to a small degree. Surprisingly, the GOx deposited before the PoPD had a similar effect on P(AA)% compared with the PoPD/GOx configuration. This view is consistent with the K_M data in Table 1, which shows that the barrier to GOx-glucose interactions was significantly greater for Pt_C/PoPD/GOx compared with Pt_C/GOx (p < 0.002), but not with Pt_C/GOx/PoPD (p > 0.13). It is interesting to note that P(AA)% for the GOx/PoPD and PoPD/GOx configurations was additive when compared with Pt_C/GOx/PoPD/GOx, the worst AA-rejecting configuration (1.0 ± 0.1%). This indicates that successive fabrication steps (in this case, GOx deposited both before and after the PoPD) perturb the polymer, compounding the slight undermining of its interference blocking properties. Finally, P(AA)% for the co-immobilized enzyme (GOx-PoPD, 0.24 ± 0.04%), which had the best glucose LRS value (Table 1), was second only to pure PoPD in terms of AA rejection (99.76%) for these basic designs.

To investigate whether the presence of GOx on the metal surface (GOx/PoPD design) or in the monomer solution (PoPD-GOx design) affected the rate of deposition of the PoPD, the collapse of electropolymerization current associated with the self-sealing nature of this polymer as it deposits on the electrode surface was analysed. The anodic electropolymerization current fell off very rapidly following the initial surge induced by the application of 0.7 V versus SCE, leading to a ∼99% loss of initial current by ∼10 s for electropolymerizations carried out in PBS containing 300 mM oPD. A 2-phase exponential decay model gave a significantly better nonlinear regression fit compared with a 1-phase analysis, as observed [22] and discussed [75] recently. The half-life values for the associated two time domains, t_½(fast) and t_½(slow), are a measure of the rate at which the blocking layer of PoPD builds up on the metal, and so might be expected to be influenced by the presence of protein macromolecules near the electrode surface. The reference values determined in the absence of GOx (i.e., for Pt_C/PoPD) were 60 ± 10 ms and 0.40 ± 0.02 s (n = 8), respectively. The presence of GOx on the surface prior to electro-deposition (Pt_C/GOx/PoPD) did not significantly affect the rate of electropolymerization current decay: 50 ± 10 ms and 0.48 ± 0.06 s (n = 11, p > 0.28). In contrast, the present of GOx (5 mg mL⁻¹) in the monomer solution did significantly slow the current collapse in both time domains: 140 ± 10 ms and 0.93 ± 0.06 s (n = 8, p < 0.001). Work is currently underway to understand more fully the significance of these fast and slower components of PoPD electrosynthesis [75], but here, as in recent findings [42], there does not appear to be any correlation between the rate of the electropolymerization current collapse and the apparent permeability of AA in the PoPD deposited (see Table 2 and Section 3.3).

As expected from the corresponding definitions and the relatively constant value of P(HP)% across the biosensor designs studied here (Table 2), the trend in the permselectivity parameter (S%), calculated using Equation (8), paralleled that of P(AA)% (see Table 2). It is the equimolar substrate selectivity (S_G%), defined by Equation (9), which most clearly reveals the superiority of the co-immobilized configuration among these basic designs (Table 2). The mean S_G% value for Pt_C/PoPD-GOx (7 ± 1%, n = 8) was between 4 and 50 times smaller (better) than for the other designs. Thus, in media containing equal concentrations of glucose and AA at the ∼1-mM level, as is the case in brain extracellular fluid [51,60], the baseline response of the Pt_C/PoPD-GOx biosensor would have a ∼7% interference contribution from AA. However, because changes in the biosensor signal are far more important than their absolute output, interference in monitoring glucose changes would be considerably less than this value due to the self-limiting shape of the AA response (see Figure 2).

Biosensors fabricated using the co-immobilization conditions described above show more than adequate substrate selectivity for glucose monitoring in most biological fluids. That said, it is always desirable to reduce biosensor interference as far as possible, especially in the design of biosensors of substrates which exist at much lower levels, such as when monitoring the key neurotransmitters, glutamate [76,77] and acetylcholine [78,79]. Electropolymerization conditions were therefore modified in attempts to lower further S_G% for Pt_C/PoPD-GOx devices compared with the standard conditions of 300 mM oPD in PBS containing 5 mg mL⁻¹ GOx.