Heterogeneous Reconstitution of the PQQ-Dependent Glucose

Jan 29, 2014 - A highly sensitive electroanalytical method for determination of PQQ in solution down to subpicomolar concentrations is proposed. It is...
1 downloads 11 Views 872KB Size
Article pubs.acs.org/ac

Heterogeneous Reconstitution of the PQQ-Dependent Glucose Dehydrogenase Immobilized on an Electrode: A Sensitive Strategy for PQQ Detection Down to Picomolar Levels Ling Zhang,∥,†,‡ Rebeca Miranda-Castro,∥,† Claire Stines-Chaumeil,§ Nicolas Mano,§ Guobao Xu,‡ François Mavré,*,† and Benoît Limoges*,† †

Laboratoire d’Electrochimie Moléculaire, UMR 7591 CNRS, Université Paris Diderot, Sorbonne Paris Cité, 15 rue Jean-Antoine de Baïf, F-75205 Paris Cedex 13, France ‡ State Key Laboratory of Electroanalytical Chemistry, Changchun Institute of Applied Chemistry, University of Chinese Academy of Sciences, Chinese Academy of Sciences, 5625 Renmin Street, Jilin 130022, China § Centre de Recherche Paul Pascal, Université de Bordeaux, UPR8641, Avenue Albert Schweitzer, 33600 Pessac, France S Supporting Information *

ABSTRACT: A highly sensitive electroanalytical method for determination of PQQ in solution down to subpicomolar concentrations is proposed. It is based on the heterogeneous reconstitution of the PQQ-dependent glucose dehydrogenase (PQQ-GDH) through the specific binding of its pyrroloquinoline quinone (PQQ) cofactor to the apoenzyme anchored on an electrode surface. It is shown from kinetics analysis of both the enzyme catalytic responses and enzyme surface-reconstitution process (achieved by cyclic voltammetry under redox-mediated catalysis) that the selected immobilization strategy (i.e., through an avidin/biotin linkage) is well-suited to immobilize a nearly saturated apoenzyme monolayer on the electrode surface with an almost fully preserved PQQ binding properties and catalytic activity. From measurement of the overall rate constants controlling the steady-state catalytic current responses of the surface-reconstituted PQQ-GDH and determination of the PQQ equilibrium binding (Kb = 2.4 × 1010 M−1) and association rate (kon = 2 × 106 M−1 s−1) constants with the immobilized apoenzyme, the analytical performances of the method could be rationally evaluated, and the signal amplification for PQQ detection down to the picomolar levels is well-predicted. These performances outperform by several orders of magnitude the direct electrochemical detection of PQQ in solution and by 1 to 2 orders the detection limits previously achieved by UV−vis spectroscopic detection of the homogeneous PQQ-GDH reconstitution. bioaffinity binding assays.11−13,15−17 The first category is based on the use of modified cofactors that can be either chemically or allosterically activated (or deactivated) for enzyme reconstitution, therefore opening the door to the development of homogeneous competitive binding assays.11−13,15 The other is related to strategies involving an overamplification of signal for increasing the sensitivity of heterogeneous bioaffinity binding assays.16,17 The quinoprotein glucose dehydrogenase (or PQQ-dependent glucose dehydrogenase, PQQ-GDH) is a prominent example of redox enzyme whose catalytic activity can be switched on via its reconstitution.19 PQQ-GDH is a watersoluble homodimeric enzyme where each subunit (50 kDa) can tightly bind one pyrroloquinoline quinone (PQQ), a redoxactive cofactor, and three calcium ions.20,21 One of the calcium ions is required to bind and activate the PQQ in the active site,

T

hough many enzymes exhibit full activity as soon as they are in vivo expressed and spontaneously folded into their proper three-dimensional structure, many others require the additional presence of nondiffusional cofactors (e.g., flavins, hemes, metal ions, iron−sulfur clusters) to be entirely catalytically active. These nondiffusional cofactors (often termed prosthetic groups) may be covalently or noncovalently bound to the protein. In the latter case, the possibility to reversibly remove the cofactor to yield an inactive enzyme form (apoenzyme) and then to reinsert back the cofactor (or an analogue) to reactivate the enzyme (holoenzyme) was proven to be a valuable tool not only in fundamental studies of structural and functional properties of enzymes1−4 but also for the design of artificial biocatalysts displaying new catalytic properties.5−10 It was also shown useful in the development of sensitive bioanalytical methods taking advantage of the switching-on activation of a reconstituted enzyme for indirectly detecting a target analyte in a sample.11−18 To achieve this, two different categories of switching-on activation strategies have been proposed, mostly in the context of enzyme-amplified © XXXX American Chemical Society

Received: January 13, 2014 Accepted: January 29, 2014

A

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

enzyme catalysis, electroanalytical methods look as an attractive alternative (to optical methods) for indirectly monitoring the specific binding and activation of apo-GDH at low levels of PQQ concentrations.22 Besides, electrochemical readouts offer a range of advantages including low-cost instrumentation, simplicity, ease of miniaturization and integration, field-portable capability, high sensitivity in small volumes39 (not limited as with optical methods by the Beer−Lambert law), and compatibility with direct analysis in turbid and/or colored samples. Another attractive prospect that has not been exploited yet, is to take advantage of the electrode surface to immobilize the apoenzyme such that, after heterogeneous reconstitution with its cofactor in solution, the catalytic activity of the surface-reconstituted enzyme is sensitively electrochemically detected. The gain of sensitivity under such conditions is expected to be directly linked to the close proximity of the enzyme to the electrode sensing area and, as a result, a function of the enzyme coverage and reactivity.39−43 Another key advantage of such a configuration is that it lets an immediate determination of the activated enzyme by simple measurement of the enzyme-generated catalytic current, thereby paving the way for an in situ and real-time monitoring of the heterogeneous enzyme reconstitution. By combining the attractive features of an electrochemical detection with those of a heterogeneous PQQ-GDH reconstitution, one can anticipate to develop a highly specific and sensitive method for quantitative analysis of PQQ in aqueous media. This is typically what we have explored in the present work, including the additional prospect to establish and rationalize the analytical performances (sensitivity, detection limit, and linear dynamic range of PQQ concentration) reached with such an approach, and to evaluate the potentialities offered by this new PQQ detection strategy for improving the sensitivity of bioaffinity binding assays such as those developed by Meyerhoff’s group.16,17,25 The PQQ binding reaction and in situ electrochemical detection scheme used throughout this work is schematically represented in Scheme 1. For the immobilization of apo-GDH,

whereas the other two are needed for functional dimerization of the protein. This enzyme exhibits a particularly high catalytic efficiency and turnover number not only toward the oxidation of glucose but also toward miscellaneous mono- and disaccharides. It can be regenerated in its oxidized form by a wide range of natural or artificial electron acceptors. We recently reported a detailed mechanistic and kinetic study of the redox-mediated electrochemistry of PQQ-GDH in solution, from which a ping-pong type mechanism including substrate inhibition and cooperativity (enhanced enzyme activity at high substrate and cosubstrate concentrations) could be fully characterized.22 The reconstitution of PQQ-GDH by specific binding of PQQ to the apo-GDH is of particular interest for the development of a switching-on biocatalyst for the following reasons: (i) The inactive apoenzyme can be easily overproduced in an Escherichia coli recombinant strain and isolated with a high yield and purity (i.e., entirely free of PQQ because the latter is not produced in E. coli),23,24 therefore circumventing the traditional cumbersome protocols for cofactor removal as well as the difficulty for getting an apoenzyme completely free of residual activity.5 (ii) The holoenzyme has a particularly high turnover number (>1500 s−1).22 (iii) The reaction of reconstitution is spontaneous and fast in the presence of Ca2+ (vide infra) and driven by a high binding affinity of PQQ to the apo-GDH (a dissociation constant in the nanomolar range has been reported).19 These favorable properties have only been recently exploited by Meyerhoff’s group in highly sensitive sandwich-type immunoassays associated with a spectrophotometric microtiter plate reader.16,17,25 In these approaches, the signal amplification strategy takes advantage of PQQ-doped nanocontainer tracers which, after their specific heterogeneous immunoaffinity binding on a solid phase, are dissolved. The resulting free PQQ molecules released in solution are then available to homogeneously activate the apo-GDH which, once reconstituted, can then be quantified spectrophotometrically from the catalytic conversion of a chromogenic cosubstrate. In these approaches, PQQ was determined down to subnanomolar concentrations, thereby allowing the determination of picomolar concentrations of Creactive protein in a sandwich-type immunoassay.16 Independent of these applications, the development of analytical methods for sensitive and selective determination of PQQ in biological, food, and environmental samples is in itself relevant. PQQ is in fact a molecule of great biological importance. Since its identification in the late 1970s,26,27 a number of physiological properties and functions have been attributed to this ortho-quinone. PQQ is a ubiquitous compound present in bacteria, plant and animal cells, as well as in many foods and biological fluids, such as milk, at pM to nM levels.28,29 It was demonstrated to take part in microbial metabolism and as a cell and plant growth promoter for a variety of organisms.30,31 Although its role in eukaryotic metabolism remains to be fully elucidated and its nutritional importance as a potentially new vitamin needs to be clarified,32 there is evidence that PQQ plays a role in pathways important to cell signaling.33−35 PQQ can also serve as an antioxidant as well as a cardio- and neuroprotector.36−38 As a consequence, precise quantification of PQQ in diverse body fluids is of nutritional, pharmacological, and clinical importance.29,35 Because of their ability to easily and efficiently monitor the activity of PQQ-GDH either by direct detection of an electroactive enzyme-generated product or by redox-mediated

Scheme 1. Principle of the Heterogeneous Reconstitution of PQQ-GDH on an Electrode Surface and Its Consecutive Electrochemical Detection

we have resorted to a gentle immobilization strategy taking advantage of the strong and specific binding of a biotinylated apo-GDH (b-apo-GDH) to an avidin monolayer passively adsorbed on a surface. This versatile immobilization method was previously shown to lead to stable depositions of highly active enzyme monolayers on an electrode surface with minimal B

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Instrumentation. Cyclic voltammetric measurements were performed with an AUTOLAB PGSTAT 12 potentiostat/ galvanostat, and the data were acquired by using the GPES 4.9007 software (EcoChemie B.V. Utrecht, The Netherlands). A conventional thermostated (20 ± 1 °C) water-jacketed threeelectrode cell was employed in all electrochemical studies. Unless otherwise stated, standard glassy carbon (GC) disk electrodes (3-mm diameter) were used as the working electrode, whereas a saturated calomel electrode (SCE) and a platinum wire were used as reference and counter electrodes, respectively. Before use, the GC electrodes were polished on a cloth using alumina slurries of 1, 0.3, and 0.05 μm particle sizes (ESCIL, France), sonicated in Milli-Q water for 5 min each, and finally subjected to several potential cycles between 0 and 0.8 V in EB at 0.1 V/s until stable capacitive background current was obtained. For homogeneous PQQ-GDH kinetics studies, disposable screen-printed carbon-based electrodes (4 mm diameter, printed with a DEK model-65 screen-printer and using the commercial carbon-based ink Electrodag PF-470A39) were used. These electrodes are more practical for performing a large number of experiments, and they possess a particularly good signal/noise ratio, which is useful for working with low substrate and cosubstrate concentrations. Before use, the screen-printed carbon electrodes (SPCEs) were saturated with a monolayer of BSA by immersion for 15 min in EB containing 1 mg/mL BSA. The BSA coating allows minimizing nonspecific adsorption of PQQ-GDH on the electrode surface and improves CV reproducibility. For the PQQ binding kinetics studies, a rotating GC disk electrode (EDI 01 Tacussel) was used. UV−vis absorption spectra were measured with a Specord S600 spectrophotometer (Analytic Jena). The electrochemical cell and glassware used for the on-electrode enzyme reconstitution were beforehand heated at 600 °C for 1 h in order to remove any PQQ contamination by carryover. Immobilization of Biotinylated Apo- or Holo-GDH on GC Electrodes. All steps during the immobilization procedure were carried out at room temperature under a water-saturated atmosphere to minimize solvent evaporation. A 5 μL aqueous droplet of 1 mg/mL neutravidin (in PBS) was locally deposited onto the sensing area of a GC electrode turned head-to-tail perpendicular to the bench. After 2 h of passive adsorption, the GC electrode surface was thoroughly rinsed with PBS. Subsequently, the electrode surface was blocked with BSA by dipping the extremity of the GC electrode in a 1 mg/mL solution of BSA in PBS for 15 min, followed by rinsing with PBS and then pre-equilibrating in EB for at least 5 min. Thereafter, the GC surface was covered with a 5 μL droplet of 0.1 μM biotinylated apo- or holo-GDH (in EB-BSA). After 2 h of incubation in the refrigerator at 4 °C, the droplet was removed from the electrode surface by simply immersing/ rinsing the GC electrode tip in a 0.5 mL EB-BSA. The resulting enzyme activity of this diluted solution was then characterized spectrophotometrically as described in Supporting Information (SI). In a few control experiments, the neutravidin adsorption step was replaced by an incubation step with a BSA solution (1 mg/mL) for 2 h. The modified electrodes were finally thoroughly rinsed with EB and then stored in EB at 4 °C until used.

loss of catalytic activity (e.g., by steric hindrance and/or enzyme surface deactivation) and without hindering the accessibility of small molecules in solution (i.e., substrates and cosubstrates) to the underlying conductive surface.39,40,44 Submicromolar concentrations of biotinylated enzyme are also enough to saturate an avidin-coated surface by a close-packed enzyme monolayer.39,40,44 To evaluate the key parameters governing the magnitude of the electroanalytical responses and thus to rationalize and predict the analytical performances of the system, a quantitative analysis of the amount of active immobilized biotinylated enzyme has been addressed first. This is accompanied by a detailed kinetic characterization of the immobilized enzyme by cyclic voltammetry (CV) under redox mediated catalysis, completed by a comparative study with kinetics in homogeneous solution. The feasibility to trigger the heterogeneous reconstitution of an immobilized apo-GDH in the presence of PQQ in solution and so to monitor the PQQ binding kinetics/ enzyme activation by in situ CV detection (as illustrated in Scheme 1) has been investigated next. Several important questions have also been addressed such as how much apoenzyme in the apo-GDH monolayer is activated, and are the heterogeneous PQQ binding constant and surface reconstitution rate the same as in homogeneous solution? The obtained results also raise questions about strategies based on the heterogeneous reconstitution of apoproteins at cofactormodified electrodes. This point is discussed critically.



EXPERIMENTAL SECTION Reagents. PQQ, D-glucose, bovine serum albumin (BSA), Triton X-100, phenazine methosulfate (PMS), and 2,6dichlorophenolindophenol (DCPIP) were purchased from Sigma and used without further purification. Ferrocene methanol (FcMeOH) was also obtained from Sigma and recrystallized twice from toluene and cyclohexane. The apoGDH was produced by controlled expression in an E. coli strain as previously described.45 The holo-GDH (or native PQQGDH) was obtained from the reconstitution of apo-GDH with PQQ according to previously published procedures.22,23 Neutravidin, sulfosuccinimidyl-6-[biotin-amido]hexanoate (EZ-Link Sulfo-NHS-LC-Biotin, spacer arm: 22.4 Å), and NHS-PEG12-biotin (EZ-Link NHS-PEG12-Biotin, spacer arm: 56 Å) were purchased from Thermo Scientific. Salts for buffer solutions (Tris-HCl, CaCl2, and phosphate buffered saline tablets) were obtained from Sigma. Double-deionized water (18.2 MΩ cm, TKA Micro-Pure UV) was used to prepare all aqueous solutions. Three different buffers were used. The enzyme biotinylation buffer was a phosphate-buffered saline (PBS) solution prepared from 0.01 M phosphate buffer, 0.0027 M KCl, and 0.137 M NaCl (pH 7.4). The Tris buffer (TB) was a 0.1 M Tris-HCl (pH 7.5). The enzyme reconstitution buffer (EB) consisted of TB with 3 mM CaCl2. In order to minimize adsorption of the enzyme on the electrochemical cell walls and electrodes and thus to improve the reproducibility of the results during the homogeneous electrochemical kinetic studies, BSA was added to the buffer solutions at a concentration of 1 mg/mL. These buffer solutions were denoted as TB-BSA or EB-BSA. Glucose solutions were allowed to mutarotate to the anomeric equilibrium for 1 day before use. All kinetic data were expressed in terms of analytical glucose concentration, even though GDH is known to be specific for β-D-glucose.46 C

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry



Article

in the absence of substrate), the “purely” catalytic plateau current (represented here in current density, jpl) is obtained (Figure 1B).The magnitude of this catalytic current is directly proportional to the enzyme coverage. From precise knowledge of the pertaining catalytic rate constants controlling the enzyme catalysis, one should thus assess quantitatively to the enzyme surface concentration.40,44 However, to achieve this, it is necessary to determine to what extent the immobilized enzyme remains as active as in homogeneous solution, an issue that can be solved by quantitative analysis of the b-holo-GDH kinetics both in homogeneous solution and after immobilization on an electrode surface. Quantitative Analysis of the b-Holo-GDH Kinetics. The homogeneous rate constants of the biotinylated PQQ-GDH prepared from biotinylation of apo-GDH followed by reconstitution with PQQ were characterized in detail from the analysis of the redox-mediated catalytic CV responses at various concentrations of glucose and FcMeOH. This was achieved using the same methodology as previously described22 and considering the following catalytic reaction mechanism. At the electrode, the redox mediator is reversibly oxidized according to

RESULTS AND DISCUSSION Enzyme Biotinylation. The degree of biotinylation of PQQ-GDH was found an important parameter not only to ensure an efficient immobilization of PQQ-GDH on a neutravidin-coated electrode (Scheme 1) but also for preserving high enzyme activity after biotinylation. This was particularly true for the biotinylation of apo-GDH (b-apoGDH) where chemical coupling of biotin residues at the entrance of the PQQ binding site might hinder the enzyme reconstitution/reactivation. The chemical coupling of a sulfosuccinimidyl-6-[biotin-amido]hexanoate, a water-soluble biotinylation reagent able to react with exposed lysine residues on the enzyme outer shell (∼20−30 lysines exposed on the surface of the PQQ-GDH), was at first considered.21 The extent of biotinylation was experimentally controlled by adjusting the coupling reaction time and the molar excess of the labeling reagent to the dimeric protein (see SI for details). After chemical derivatization and reconstitution (see SI), activities of biotinylated holo-GDH (b-holo-GDH) were controlled in homogeneous solution by measuring the magnitude of the redox-mediated catalytic plateau current in CV (vide infra). It was found by comparison with the native PQQ-GDH that the activity of b-holo-GDH was significantly decreased when a too large excess of biotinylated reagent was used (for molar excess of 12 and 40, enzymatic activity decreases of 30% and 50% were determined, respectively), whereas for the lowest molar excess of 5, a nearly identical activity was found. This molar excess ratio was thus selected for the further studies. Effectiveness of enzyme biotinylation was next established by recording the redox-mediated catalytic current response at a neutravidin electrode preincubated in a diluted solution of bholo-GDH (0.1 μM) for 2 h. As illustrated by the CV curves in Figure 1, in contrast to the absence of catalytic response at a

R ⇄ O + e‐

(1)

(where R is for the reduced FcMeOH and O for the oxidized Fc+MeOH) In the diffusion-reaction layer, the ping-pong mechanism can be described by the following successive steps: (i) a reductive half-reaction with glucose (G) k1(″)

kc(″)

G + PQQ‐GDHox XoooY [G/PQQ‐GDHox ] ⎯⎯→ Glu + PQQ‐GDH red k−(″1)

(2)

(where PQQ-GDHox and PQQ-GDHred are the fully oxidized and reduced forms of the enzyme) (ii) an oxidative half-reaction by the electrogenerated oxidized mediator (″) kox

2O + PQQ‐GDH red ⎯→ ⎯ 2R + PQQ‐GDHox

(3)

and (iii) an enzyme inhibition by the substrate K i(″)

PQQ‐GDH red + G XooooY PQQ‐GDH i

(4)

When including the cooperative effects, the catalytic reaction scheme of PQQ-GDH can be appropriately described by two sets of four parameters22 denominated kox, kc, KM, Ki and k″ox, kc″, KM ″ , Ki″, where the first one accounts for the noncooperative mode (occurring at low mediator concentrations) and the second one for the cooperative mode (occurring at high mediator and substrate concentrations). The parameters K(M″) (in M) here stand for the Michaelis−Menten constants given by

Figure 1. (A) CVs (v = 0.1 V/s) at (red) a neutravidin-modified GC electrode and (blue) BSA-modified GC electrode both preincubated with a 0.1 μM b-holo-GDH solution for 2 h and then immersed in a TB solution (pH 7.5) containing 3 μM FcMeOH and 3 mM glucose (the currents were here normalized to the geometric electrode area). (B) Pure catalytic response resulting from subtraction of the CV recorded in the absence of glucose (but presence of mediator) to the red one in A.

(″) KM =

k −(″1) + kc(″) k1(″)

whereas the parameters K(i ″) (in M) are the equilibrium inhibition constants. The resulting set of kinetic and equilibrium constants determined for both the b-holo-GDH and native PQQ-GDH are summarized in Table 1. The high similarity between the two set of parameters clearly demonstrates that the reactivity of bholo-GDH is not discernibly perturbed by the biotinylation,

BSA-coated electrode (showing only the reversible wave of FcMeOH), a well-defined steady-state catalytic S-shaped wave was obtained at the neutravidin electrode, demonstrating that a noteworthy amount of biotinylated enzyme was specifically bounded to the electrode surface. After correction from the diffusion-controlled reversible wave of mediator (i.e., recorded D

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

Table 1. Kinetic Parameters of the Different Forms of PQQ-GDH (0.1 M Tris buffer, 3 mM CaCl2, pH 7.5, T = 20°C) biotinylated PQQ-GDHa

native PQQ-GDH kox (M−1 s−1) kc (s−1) KM (M) Ki (M−1) k″ox (M−1 s−1) kc″ (s−1) KM ″ (M) K″i (M−1) a

homogeneous rates

homogeneous rates

heterogeneous rates

(1.9 ± 0.3) × 1500 ± 300 (2.7 ± 0.5) × 52 ± 6 (3.3 ± 0.6) × 6000 ± 1000 (4.7 ± 0.3) × 127 ± 20

(1.8 ± 0.5) × 1500 ± 300 (4.8 ± 1.2) × 24 ± 2 (2.7 ± 0.3) × 5000 ± 1000 (9.6 ± 3.2) × 53 ± 13

(9.5 ± 2.0) × 107 1500 ± 400 (7.3 ± 0.9) × 10−4 30 ± 4 n.d.b n.d. n.d. n.d.

108 10−4 108 10−3

108 10−4 108 10−3

Obtained for a biotinylation reagent/protein molar excess ratio of 5. bn.d.: not determined.

and so the selected conditions for biotin coupling (molar excess ratio of 5) are mild enough to fully preserve the enzyme activity. It is to be noted that the previously reported values of native enzyme22 slightly differ from those here, but this is because of the different conditions of buffer and pH (in the previous work, a 0.1 M phosphate buffer of pH 7 was used). Quantitative Characterization of the Immobilized PQQ-GDH. The kinetics of the immobilized b-holo-GDH was next investigated by CV in the presence of different substrate and cosubstrate concentrations. The concentrations of FcMeOH were restricted to the low micromolar range in order to have catalytic responses that satisfy steady-state conditions (i.e., conditions leading to a plateau current). At too high mediator concentration, peak-shaped catalytic waves were obtained, pointing out to a partial control of the catalytic response by the mass transport of glucose to the electrode. Although these peaks contain kinetic information relevant to the catalysis,40 it is less straightforward to extract (no analytical expression for relating the catalytic current to the pertinent rate constants) and also not relevant to the scope of the present work. In consequence, only the noncooperative mode has been fully investigated with the immobilized enzyme. The analytical equation that relates the steady-state catalytic current in CV to the active enzyme surface concentration (Γ0holo in mol/cm2) and enzyme kinetic parameters has been derived considering the ping-pong mechanism with substrate inhibition.40 Under these conditions and considering only the noncooperative mode, the catalytic steady-state plateau current density jpl (in A/cm2, obtained after subtraction of the diffusive CV response of mediator from the total CV current) may then be expressed by eq 5 (see SI for its establishment) jpl = 2F

Figure 2. (A) Linear voltammetric scans recorded (v = 0.1 V/s) at a biotinylated PQQ-GDH/neutravidin-modified GC electrode in EB containing 1 μM FcMeOH and 0.3 (gray), 0.5 (orange), 1 (violet), 3 (olive), 5 (red), 10 (black), 30 (blue), 50 (brown), and 100 mM glucose (pink). (B) Steady-state plateau currents as a function of glucose concentration for 0.5 (blue), 1 (red), and 3 μM (olive) FcMeOH. Plain lines are theoretical fits of eq 5 to the experimental data using Γ0holo= 1.1 pmol/cm2 and the kinetic constants reported in Table 1. Errors bars represent standard deviation from the average of six experimental data.

C0G and C0R (Figure 2B) and an independent estimation of Γ0holo, one can determine the values of kox, kc, KM, and Ki by nonlinear regression fit of eq 5. The value of Γ0holo was independently obtained using the droplet depletion method that was formerly used by us for other immobilized enzymes.39,44 It consists of depositing a droplet of biotinylated enzyme solution on the surface of a neutravidin-modified electrode and, after an incubation step, in determining spectrophotometrically the enzyme activity remaining in the droplet (see SI for detailed explanations). An average value of Γ0holo = 1.1 ± 0.1 pmol cm−2 was finally determined from a series of modified electrodes incubated in a saturating concentration of 0.1 μM b-holo-GDH. Such an enzyme surface coverage corresponds approximately to ∼75% of the theoretical value that can be calculated for a saturated closed-packed monolayer of PQQ-GDH deposited on a perfectly flat surface (a theoretical maximal coverage of 1.5 pmol cm−2 can be determined using an enzyme radius of 4.5 nm and a protein packing factor of 0.6). The value also agrees with those found for other enzymes of similar size and using the same immobilization strategy.39,40 This independently determined value of Γ0holo was then inserted into eq 5, and the resulting equation was used to fit the overall experimental data in Figure 2B with a single set of parameters (i.e., kox, kc, KM, and Ki). The best fit was achieved with the data listed in Table 1 (heterogeneous rates column).

Γ 0holo 1 kc

+

KM kcCG0

+

1 koxC R0

+

K iCG0 koxC R0

(5)

where F is the Faraday constant, C0G the bulk concentration of glucose (in M), and C0R the bulk concentration of FcMeOH (in M). A typical set of experimental steady-state catalytic waves recorded at a saturated b-holo-GDH-electrode and for different concentrations of glucose are shown in Figure 2A (all plots were corrected from the current response of FcMeOH recorded in the absence of substrate). As expected, the steady-state catalytic current first increases with the increase of glucose concentration until reaching a maximal value and then steadily decreases upon further increase of the glucose concentration due to the substrate inhibition. From the resulting experimental bell-shaped plots of jpl as a function of E

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

an exponential increase over several hours until it levels off at a constant value characteristic of an equilibrated reaction. After the equilibrium was reached, the addition of a large excess of PQQ (10 nM) to the solution leads to a fast and marked increase of the current up to a new constant value (jpl,max) representative of a saturation of the apo-GDH monolayer by PQQ, that is, 0.76 pmol/cm2 (further additions of PQQ do not increase the catalytic current). By repeating similar PQQ saturating experiments on several b-apo-GDH-modified electrodes, a maximal activatable apo-GDH coverage of Γ0apo,m = 0.78 ± 0.12 pmol/cm2 was obtained, which corresponds to 71% of the maximal coverage independently determined for the direct immobilization of b-holo-GDH (1.1 ± 0.1 pmol/cm2). This result suggests that not all of the immobilized b-apo-GDH can be reconstituted, probably because of some problems of steric constrains, limiting the access of PQQ to a few apo-GDH binding sites, or possibly because of some apo-GDH deactivation/denaturation on the surface or dissociation of the dimeric enzyme into less active monomeric subunits during the immobilization step. To better compare and analyze the time course of heterogeneous enzyme reconstitution at different apo-GDH electrodes and PQQ concentrations, the maximal coverage of apo-GDH that can reconstitute was systematically determined by the addition of an excess of PQQ at the end of each experiments, and the resulting value then used to normalize the data. In this way, the kinetic plots can be represented as fractional enzyme coverage (θ = jpl/jpl,max = Γ0holo/Γ0apo,m) as a function of time. The role of PQQ mass transport on the reconstitution kinetics is clearly visible in Figure 3B where the time-course fractional coverage depends on the electrode rotation rate at low value. This dependence ceases when the rotation rate is made larger than 500 rpm, which is a clue that the kinetics is no longer controlled by the mass transport of PQQ but solely by the binding rate of PQQ to the immobilized apo-GDH (or more rigorously by the surface reconstitution/activation of the enzyme). These conditions offer the possibility to determine the kinetic and thermodynamic parameters of the heterogeneous enzyme reconstitution. For such purpose, we have assumed that the binding of PQQ to the immobilized apoGDH obeys a simple Langmuirian kinetics according to the following reversible reaction

These values are in very good agreement with those obtained for the b-holo-GDH in homogeneous solution, clearly demonstrating that the reactivity of the immobilized biotinylated PQQ-GDH is the same as in solution and also that the immobilization strategy is well-suited to anchor a close-packed monolayer of PQQ-GDH on an electrode surface with an almost fully preserved activity. Heterogeneous Reconstitution of PQQ-GDH. We have next characterized the heterogeneous reconstitution of PQQGDH from b-apo-GDH immobilized on a neutravidin-modified electrode (as depicted in Scheme 1). The same immobilization procedure as for the b-holo-GDH was used. By extrapolation with the results obtained for b-holo-GDH, it can be assumed that a comparable saturating amount of b-apo-GDH could be immobilized on a neutravidin-modified electrode (i.e., Γ0apo,m ∼ 1.1 pmol cm−2). In order to carefully control the rate of PQQ mass transport at the electrode/solution interface, the b-apo-GDH-electrode was mounted on a rotating disk electrode (RDE) system. Moreover, the progress of the enzyme reconstitution was electrochemically monitored in situ by immersing the apoGDH RDE in an enzyme reconstitution buffer containing not only the PQQ analyte but also glucose (0.3 mM) and FcMeOH (1 μM) and then by recording at any time the catalytic CV response. The volume of solution has been voluntarily set at high value (25 mL) in such a way that the absolute amount of PQQ extracted by the apo-GDH electrode from the solution is negligible, and so the bulk concentration of PQQ (C0PQQ) can be considered as constant during the whole course of the experiment. The quantity of heterogeneously reconstituted bholo-GDH as a function of incubation time can thus be inferred from the magnitude of the catalytic plateau current and eq 5 (under our selected conditions of 1 μM FcMeOH and 0.3 mM glucose, the proportionality coefficient between Γ0holo and jpl was equal to (0.07 ± 0.01) pmol/μA). An illustrating example of time-course heterogeneous enzyme reconstitution carried out at an electrode rotation rate of 1500 rpm is shown in Figure 3A for a PQQ concentration of 12 pM. The catalytic current response follows

kon

apo‐GDH + PQQ XooY holo‐GDH koff

(6)

where one immobilized apo-GDH binding site reversibly reacts with one PQQ molecule in solution to give one active holoGDH binding site (the two PQQ binding sites of the dimeric enzyme were considered equivalent). Here, kon (in M−1 s−1) and koff (in s−1) are the binding and dissociation rate constants, respectively. The equilibrium binding constant is then given by Kb = kon/koff. The rate of the surface binding reaction under controlled steady-state convective regime may thus be expressed as

Figure 3. Binding kinetics of PQQ to a b-apo-GDH-modified RDE. (A) CV catalytic plateau current density as a function of time for an apo-GDH electrode (rotation rate: 1500 rpm) immersed in 25 mL EB containing 12 pM PQQ, 0.3 mM glucose, and 1 μM FcMeOH. The arrow shows the injection of an excess of PQQ to the solution up to a final concentration of 10 nM. (B) Fractional enzyme coverage as a function of time at apo-GDH electrodes incubated in 12 pM PQQ with the electrode rotating at 1500 rpm (red circles) or in 100 pM PQQ with the electrode rotating at 500 rpm (purple stars), 1500 rpm (green triangles), and 2500 rpm (open blue diamonds). Other conditions are the same as in A.

d[Γ 0holo] 0 =kon Γ apo [PQQ]x = 0 − koff Γ 0holo dt DPQQ 0 = (C PQQ − [PQQ]x = 0 ) δ F

(7)

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

where Γ0apo and Γ0holo are the surface concentrations of the free and occupied enzyme binding sites (in mol/cm2), DPQQ is the diffusion coefficient of PQQ (in cm2/s), δ is the diffusionconvection layer thickness (in cm), and C0PQQ and [PQQ]x = 0 the concentrations of PQQ in solution and at the electrode/ solution interface (in mol/cm3), respectively. In eq 7, the last term accounts for the flux of PQQ arriving at the electrode surface by diffusion−convection. By normalizing to the maximal enzyme coverage and taking into account the mass balance (Γ0apo + Γ0holo = Γ0apo,m), we can rewrite eq 7 into eq 8 0 kon(1 − θ )C PQQ − koff θ dθ = 0 kon Γ apo,m dt 1 + DPQQ (1 − θ ) δ

PQQ concentrations (from 2 pM to 150 pM). The overall kinetic plots were satisfactorily fitted with the following single set of parameters: kon = (2.0 ± 0.4) × 106 M−1 s−1 and Kb = (2.4 ± 0.1) × 1010 M−1. An identical value of Kb could also be inferred from the Langmuir isotherm plot in Figure 4B (i.e., obtained by plotting the equilibrium values of θ as a function of C0PQQ). The obtained Kb value is slightly higher than previously determined by Matsushita et al. in homogeneous solution (Kb ∼ 109 M−1 in a Tris buffer with 3 mM Ca2+).19 To determine if it could be due to some difference between our experimental conditions and the previous one, we have independently determined the homogeneous value of Kb (see SI and Figure S2). A binding constant of Kb = (3.0 ± 1.5) × 1010 M−1 was found, which is very close to the one obtained from the heterogeneous experiments. This result clearly shows that the thermodynamics of the binding reaction of PQQ to the immobilized apo-GDH is not affected by the immobilization. These overall results finally point out that, under our conditions, the heterogeneous apo-GDH reconstitution is a viable strategy which can be achieved with a high efficiency, yield, and reproducibility. A confirmation of the absence of kinetic interference from diffusion can be realized, a posteriori, by calculating the dimensionless competitive kinetic parameter konΓ0apo,m/(DPQQ/ δ). Considering δ = 10.2 × 10−4 cm at 1500 rpm, DPQQ = 5 × 10−6 cm2 s−1, Γ0apo,m = 0.78 pmol cm−2, and the above determined kon kinetic rate constant expressed in mol−1 cm3 s−1, a value of 0.26 is calculated. This value is smaller than 1 in eq 8, which thus turns out unequivocally to eq 10. It was next interesting to determine how much the absolute PQQ binding rate constant kon compares with its homogeneous counterpart. The absence of data in the literature has led us to determine kon in homogeneous solution. For such a purpose, stopped-flow kinetics experiments were carried out by measuring the transient UV−vis absorbance change at 338 nm (corresponding to the conversion of the oxidized reconstituted holo-GDH into its reduced form) immediately after rapid mixing of apo-GDH with glucose and various concentrations of PQQ (see SI). The resulting exponential kinetic plots were in agreement with a bimolecular reaction from which a second-order rate constant of (1.2 ± 0.1) × 106 M−1 s−1 was inferred. This kon value is clearly in line with the one found from heterogeneous binding kinetics (kon = 2 × 106 M−1 s−1), indicating that the surface reconstitution behaves very similarly as in solution. This value was also confirmed by determining spectrophotometrically the homogeneous biomolecular enzyme reactivation rate under steady-state catalytic conditions (i.e., at low enzyme concentration and in presence of a large excess of substrate and cosubstrates, see SI and Figure S3 for details). Finally, the fact that the monolayer of apo-GDH can be nearly entirely activated (at least by 70%) and that the kinetics and thermodynamics of the heterogeneous reconstitution process are very similar to those in solution definitely demonstrates that the immobilized active apo-GDH as well as its reconstituted form behave like in solution and are not significantly affected by the immobilization process or by the neutravidin layer and/or electrode surface proximity. In an attempt to recover the ∼30% of apparently inactive apo-GDH (by comparison with the direct immobilization of bholo-GDH), the apoenzyme was chemically coupled to a twicelonger spacer arm biotin derivative (NHS-PEG12-Biotin). By this way, we could expect to slightly improve the accessibility of

(8)

Explicit integration of this differential equation leads to a general analytical solution that allows for prediction of all possible fractional coverage-time curves.47 As previously discussed from the results in Figure 3B, under appropriate electrode rotation rates, the PQQ mass transport is no longer rate determining (meaning that in eq 8 the competitive kinetic parameter konΓ0apo,m/(DPQQ/δ) is low). Consequently, eq 8 simplifies into dθ 0 = kon(1 − θ )C PQQ − koff θ dt

(9)

which after integration leads to 0 θ = θeq {1 − exp[−kon(1/Kb + C PQQ )t ]}

(10)

where θeq is the fractional surface coverage at equilibrium defined by the Langmuir isotherm equation θeq =

0 KbC PQQ 0 1 + KbC PQQ

(11)

Equation 10 was then used to globally fit the experimental kinetics traces (Figure 4A) obtained for a range of different

Figure 4. (A) Binding kinetics of PQQ to b-apo-GDH/neutravidin electrodes (represented as fractional reconstituted enzyme coverage with time) rotated at 1500 rpm and for different PQQ concentrations in solution (concentrations are mentioned on the graph). Solid lines are the theoretical kinetic plots calculated with eq 10 and using kon = 2 × 106 M−1 s−1 and Kb= 2.4 × 1010 M−1. (B) Plots of the fractional enzyme coverage as a function of PQQ concentration (red circle) for 40 min binding reaction time or (blue square) at the equilibrium. Blue solid line: Langmuir isotherm fit (eq 11) using Kb = 2.4 × 1010 M−1. Red dashed line: fit with eq 10 using kon = 2 × 106 M−1 s−1, Kb = 2.4 × 1010 M−1, and t = 40 min. Conditions: 0.1 M EB (pH 7.5) containing 0.3 mM glucose and 1 μM FcMeOH. G

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

works where the homogeneous reconstitution of PQQ-GDH was attempted with PQQ analogues, the catalytic activity of the reconstituted enzyme was systematically strongly decreased, if not totally suppressed53−55 (a simple esterification of the three carboxylic groups of PQQ could lead to a complete loss of the enzyme activity55). It is thus rather unlikely that the surface reconstitution of apo-GDH with a PQQ-modified electrode could lead to an enzyme reactivity higher than that for the native PQQ-GDH in homogeneous solution (a turnover rate of k = 11800 s−1 was reported in ref 49, which is much higher than the turnover rates determined for the native PQQ-GDH in solution, see Table 1). It is also unexpected that the current response attributed to the biocatalytic oxidation of glucose by the reconstituted wired PQQ-GDH does not present the characteristic natural inhibition of PQQ-GDH at high glucose concentrations (a property of PQQ-GDH toward glucose that has no fundamental grounds to be suppressed after heterogeneous enzyme reconstitution on a surface, exactly like shown in the present work). All of these remarks finally raise questions on the real efficiency of the heterogeneous reconstitution of apo-GDH to the PQQ-modified electrode and, as a result, on the effectiveness of the direct wiring of PQQ-GDH to an electrode. Similar questions also arise for the reconstitution of apoglucose oxidase on a FAD-modified gold electrode and for which a direct enzyme wiring with the electrode has been also asserted.50 These uncertainties and questions are reinforced by a work published by Gooding and his collaborators,51 in which they were unable to observe an electrocatalytic enzyme activity from the binding of apo-glucose oxidase to a FAD-terminated self-assembled monolayer on gold. To interpret this lack of enzyme turnover through a supposed direct electron transfer, the authors suggested a decrease in the electronic coupling between the redox active FAD and the electrode following reconstitution of the glucose oxidase. However, a more plausible explanation would be a dramatic loss of the binding recognition properties of apo-glucose oxidase to the FAD derivative attached to the electrode, and/or a loss of the catalytic activity of the reconstituted enzyme due to an inappropriate arrangement of the FAD analogue in the enzyme binding site. A detailed and careful quantitative study as is done in the present work would certainly help to elucidate these questions and to obtain a better understanding of such heterogeneous reconstitution strategies. On account of the very high affinity binding of PQQ to the apo-GDH, very low levels of nonspecifically immobilized (adsorbed) cofactor could lead to a non-negligible apo-GDH reconstitution. The consequence is that it would result in a false interpretation of the origin of the enzyme surface-activation process (which typically could occur with the surface reconstitution of apoenzymes on cofactor-modified electrodes), and in our specific case of quantitative detection of PQQ, it would lead to a risk of bias in the results. This is also the reason that, after each use, we have taken care to cautiously remove from the electrochemical cell and electrodes any residual traces of PQQ, thus avoiding cross-contamination Evaluation and Rationalization of the Analytical Performances. From the kinetic experiments in Figure 4A, PQQ standard calibration plots can be drawn. The analytical performances of the method are obviously a function of the time the apo-GDH-electrode is incubated in the PQQ solution. If we report the fractional enzyme coverage after 40 min of incubation time (i.e., out of equilibrium), we obtained a

the immobilized apo-GDH binding sites by an increase of the flexibility and distance between the immobilized enzymes as well as between the enzymes and the underlying neutravidin layer. Though the homogeneous and heterogeneous kinetic parameters of the new b-holo-GDH were quite similar to those of native and previously biotinylated enzymes (Table S1 and Figure S4), the yield of surface reconstitution was still only ∼70% to that obtained with the direct immobilization of bholo-GDH, meaning that no improvement of the reconstitution yield could be achieved with this longer biotin linker. Besides, no enhancement of the PQQ binding rate could be observed from the heterogeneous reconstitution kinetic plots of this new biotinylated PQQ-GDH (data not shown). At this point, it is interesting to discuss our results relative to the works in which an on-electrode surface enzyme reconstitution has been proposed.48−51 Niemeyer and collaborators have studied the reconstitution kinetics of two apohemoproteins with DNA-linked heme on gold electrode surfaces by means of surface plasmon resonance.48 Their approach substantially differs from ours by the fact that the cofactor is chemically coupled to the electrode surface and the apoenzyme is the reconstitution partner in solution. For apomyoglobin and apo-horseradish peroxidase, the measured association kinetic rate constants were 2.0 × 104 and 0.2 × 104 M−1 s−1, whereas the association thermodynamic constants were 3.3 and 0.6 × 107 M−1, respectively. Differences in the kinetics were interpreted in terms of access to a more or less buried site, which makes sense because the accessibility of the cofactor under such a heterogeneous reconstitution configuration is much more prone to steric effects on the surface compared to our approach, where the native cofactor in solution is free to diffuse and bind to the immobilized apoprotein. Although the authors do not report and compare their heterogeneous binding constants with those that could be obtained in homogeneous solution, a several orders of magnitude larger homogeneous binding constant between apomyoglobin and its natural heme is reported in the literature (∼1014 M−1).52 It therefore suggests that the chemical coupling of the heme cofactor to the electrode gives rise to a strong alteration of the biomolecular recognition, resulting thus in a drastic decrease of the binding strength between the apomyoglobin in solution and the immobilized heme. It is also important to point out that the chemical modification of the cofactor by a tethered linker may alter not only the apoprotein binding recognition but also the catalytic activity of the surface-reconstituted enzyme. According to a similar strategy, the heterogeneous reconstitution of PQQ-GDH on the surface of a PQQmodified electrode has been proposed by Willner and collaborators in the frame of direct electrical wiring of this enzyme to a gold electrode.49 In this work, the PQQ was covalently attached to the conductive surface via an amide bond linkage between a primary amino group anchored on the gold surface and one of the three carboxylic acid moieties of PQQ, and the resulting modified electrode then reacted with the apoGDH in solution. Although not considered by the authors, the same issues as just mentioned above should apply here too. It is indeed highly likely that the chemical modification and coupling of PQQ to the electrode (which in ref 49 is not well-defined because PQQ was linked randomly to the surface) significantly affect not only the enzyme reconstitution process (both kinetically and thermodynamically) but also the enzyme reactivity. This statement is supported by the fact that in all H

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

calibration plot which overall fits to eq 10 and, within the shortrange of 0−50 pM, varies approximately linearly with C0PQQ (Figure 4B). Under these conditions, a detection limit of ∼2 × 10−12 M PQQ can be estimated. An even better detection limit reaching the subpicomolar level can be retrieved from the plot of the equilibrium binding (i.e., θeq) as a function of C0PQQ (Figure 4B). These remarkable performances outperform by several orders of magnitude the direct electrochemical detection of PQQ in solution (a detection limit of 0.1 μM can be at best achieved by direct CV reduction of PQQ at a GC electrode), demonstrating the huge amplification brought by the enzyme surface confinement and catalysis. This is also between 1 and 2 orders of magnitude better than the detection limits of PQQ attained by the UV−vis spectroscopic determination of the homogeneous reconstitution of PQQGDH.16,17 However, in spite of the relatively fast enzyme reconstitution process (kon = 2 × 106 M−1 s−1), it takes a relatively long time to reach the equilibrium binding at low PQQ concentrations. A usual compromise between analysis time and analytical sensitivity would thus have to be found. In order to better appreciate the key parameters governing the analytical performances of the method and to predict how much it could be further improved by adjusting these parameters, it is interesting to formally express the theoretical equation that relates the sensitivity, detection limit, and dynamic range of PQQ detection to the analytical current response. This relationship can be obtained from the combination of eqs 5 and 10, leading to 0 jpl = 2Fk Γ apo,m

⎫ ⎧ ⎡ ⎞ ⎤⎪ ⎛ 1 ⎪ 0 ⎬ ⎨1 − exp⎢− kon⎜ + C PQQ ⎟t ⎥⎪ ⎪ ⎠ ⎥⎦⎭ ⎝ Kb ⎣⎢ ⎩

0 KbC PQQ 0 KbC PQQ

1+

with the multilayer assembling of glucose oxidase56 or horseradish peroxidase43). Improving the accessibility of binding sites by a better control of the orientation of apoGDH on the surface (for instance, by immobilizing a his-tagged apo-GDH on a nitrilotriacetic acid-functionalized electrode57,58) is a route to explore. Optimizing the value of k (here arbitrarily fixed at 77 s−1) by increasing the mediator concentration could also significantly enhance the analytical current response. This is well-illustrated in Figure 2, where a 3fold increase of the FcMeOH concentration leads to a 3-fold enhancement of the catalytic plateau current. Finally, to assess to what extent it can be advantageous to use the apoenzyme immobilized on the electrode surface rather than simply dissolved in solution, it is useful to refer to the analytical equations expressing the redox-mediated catalytic current as a function of the enzyme concentration. In the immobilized case, it is proportional to the PQQ-GDH surface concentration (eq 5), whereas in the homogeneous configuration it is a function of the square root of the bulk PQQGDH concentration (eq 1 in ref 22). This distinction would mean that for increasing the sensitivity of the homogeneous redox-mediated catalysis by a factor 10, it is necessary to increase the bulk enzyme concentration by 100-fold. Moreover, on the basis of these two theoretical equations and the use of the rate constants determined for both the immobilized and solubilized form of PQQ-GDH (see Table 1), it can be demonstrated that the magnitude of the catalytic current density of a 10 μM PQQ-GDH in solution is approximately the same as that at a PQQ-GDH electrode covered with merely 0.78 pmol/cm2 holoenzyme. This comparison clearly illustrates the interest to immobilize the apo-GDH on an electrode surface, not only for reaching a high sensitivity but also for avoiding a too high cost per analysis.

(12)

where k is the global apparent enzyme reaction rate given by k=



1 1 kc

+

KM kcCG0

+

1 koxCR0

+

K iCG0

CONCLUSIONS The present work demonstrates that the concept of an onelectrode reconstitution/activation of the apo-GDH by specific binding reaction with its natural PQQ cofactor in solution is feasible and that it can be achieved with a high efficiency, yield, and reproducibility. It also shows that the PQQ binding properties of the immobilized apo-GDH as well as its reactivity once reconstituted are almost identical to those in homogeneous solution. In addition to the combined high affinity constant (Kb = 2.4 × 1010 M−1) and fast binding rate (kon = 2 × 106 M−1 s−1) leading to an efficient capture of PQQ in solution, even at low concentration, the high catalytic activity (tuned by k) of the reconstituted enzyme and the high enzyme concentration at the vicinity of the detector ensure an efficient and sensitive transduction of the binding event. It results in a remarkably high sensitive switchable off/on bioaffinity electrode for PQQ detection down to subpicomolar concentrations. These noteworthy performances outperform by several orders of magnitude the direct electrochemical detection of PQQ in solution, and by 1 to 2 orders of magnitude the detection limits previously achieved by UV−vis spectroscopic monitoring of the homogeneous reconstitution of PQQGDH.16,17 These results are very promising for applying this electroanalytical detection strategy not only for the selective and sensitive detection of PQQ in biological fluids and foods but also for the development of highly sensitive bioaffinity assays (e.g., through the use of a primary enzyme label able to convert an inactivated form of PQQ into an active one or by the release of the content of nanocontainer labels filled with

koxCR0

Considering that for analytical purposes it is more convenient to have an analytical response linearly correlated to the analyte concentration, it is interesting to extract from eq 12 the limiting cases that allow us to linearize eq 12 with C0PQQ. A first limiting case can be obtained when t → 0, since from series expansion approximation, eq 12 leads to 0 0 jpl = 2Fk Γ apo,m kontC PQQ

(13)

Another possibility is to consider the equilibrium limiting case (i.e., when t → ∞) and the linear approximation of the Langmuir isotherm at low PQQ concentrations (for instance, for conditions satisfying the inequality KbC0PQQ < 0.1), which gives 0 0 jpl = 2Fk Γ apo,m KbC PQQ

(14)

From these two relatively simple expressions, we can highlight that the magnitude of steady-state catalytic current response is dependent on the following key parameters: t, k, Γ0apo,m, Kb, kon, and C0PQQ. Waiting for a sufficiently long incubation time until reaching the equilibrium binding is an evident way to increase the sensitivity up to the maximal equilibrium value (eq 14), but this is at the expense of the analysis time. Another possibility is to raise Γ0apo,m by increasing the amount of available apo-GDH binding sites. This can be achieved either by an improvement of the accessibility of the immobilized binding sites or by the assembling of a multilayer film of apo-GDH (likewise shown I

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

PQQ16,17). For such analytical developments, the use of disposable electrochemical microtiter plates such as those we have recently developed (involving screen-printed electrodes at the bottom of flat microwells)59,60 would assuredly be better adapted than the present RDE in a standard electrochemical cell. The work is in progress to achieve these goals.



(15) Simon, P.; Dueymes, C.; Fontecave, M.; Décout, J.-L. Angew. Chem., Int. Ed. 2005, 117, 2824−2827. (16) Shen, D.; Meyerhoff, M. E. Anal. Chem. 2009, 81, 1564−1569. (17) Zimmerman, L. B.; Lee, K.-D.; Meyerhoff, M. E. Anal. Biochem. 2010, 401, 182−187. (18) Virel, A.; Saa, L.; Koster, S. D.; Pavlov, V. Analyst 2010, 135, 2291−2295. (19) Matsushita, K.; Toyama, H.; Ameyama, M.; Adachi, O.; Dewanti, A.; Duine, A. J. Biosci., Biotechnol., Biochem. 1995, 59, 1548−1555. (20) Olsthoorn, A. J. J.; Otsuki, T.; Duine, J. A. Eur. J. Biochem. 1998, 255, 255−261. (21) Oubrie, A.; Rozeboom, H. J.; Kalk, K. H.; Duine, J. A.; Dijkstra, B. W. J. Mol. Biol. 1999, 289, 319−333. (22) Durand, F.; Limoges, B.; Mano, N.; Mavré, F.; Miranda-Castro, R.; Savéant, J.-M. J. Am. Chem. Soc. 2011, 133, 12801−12809. (23) Olsthoorn, A. J. J.; Duine, J. A. Arch. Biochem. Biophys. 1996, 336, 42−48. (24) Yu, Y.; Wei, P.; Zhu, X.; Huang, L.; Cai, J.; Xu, Z. Eng. Life Sci. 2012, 12, 574−582. (25) Zimmerman, L. B.; Worley, B. V.; Palermo, E. F.; Brender, J. R.; Lee, K.-D.; Kuroda, K.; Ramamoorthy, A.; Meyerhoff, M. E. Anal. Biochem. 2011, 411, 194−9. (26) Salisbury, S. A.; Forrest, H. S.; Cruse, W. B. T.; Kennard, O. Nature 1979, 280, 843−844. (27) Westerling, J.; Frank, J.; Duine, J. A. Biochem. Biophys. Res. Commun. 1979, 87, 719−724. (28) Stites, T. E.; Mitchell, A. E.; Rucker, R. B. J. Nutr. 2000, 130, 719−727. (29) Kumazawa, T.; Sato, K.; Seno, H.; Ishii, A.; Suzuki, O. Biochem. J. 1995, 307, 331−333. (30) Misra, H. S.; Rajpurohit, Y. S.; Khairnar, N. P. J. Biosci. 2012, 37, 313−325. (31) Sashidhar, B.; Podile, A. R. J. Appl. Microbiol. 2010, 109, 1−12. (32) Rucker, R.; Storms, D.; Sheets, A.; Tchaparian, E.; Fascetti, A. Nature 2005, 433, E10−E11. (33) Hirakawa, A.; Shimizu, K.; Fukumitsu, H.; Furukawa, S. Biochem. Biophys. Res. Commun. 2009, 378, 308−312. (34) Sato, K.; Toriyama, M. J. Dermatol. Sci. 2009, 53, 140−145. (35) Bauerly, K.; Harris, C.; Chowanadisai, W.; Graham, J.; Havel, P. J.; Tchaparian, E.; Satre, M.; Karliner, J. S.; Rucker, R. B. PLoS One 2011, 6, e21779. (36) Kim, J.; Kobayashi, M.; Fukuda, M.; Ogasawara, D.; Kobayashi, N.; Han, S.; Nakamura, C.; Inada, M.; Miyaura, C.; Ikebukuro, K.; Sode, K. Prion 2010, 4, 26−31. (37) Misra, H. S.; Khairnar, N. P.; Barik, A.; Indira Priyadarsini, K.; Mohan, H.; Apte, S. K. FEBS Lett. 2004, 578, 26−30. (38) Hara, H.; Hiramatsu, H.; Adachi, T. Neurochem. Res. 2007, 32, 489−495. (39) Limoges, B.; Marchal, D.; Mavré, F.; Savéant, J.-M.; Schöllhorn, B. J. Am. Chem. Soc. 2008, 130, 7259−7275. (40) Limoges, B.; Marchal, D.; Mavré, F.; Saveant, J.-M. J. Am. Chem. Soc. 2006, 128, 2084−2092. (41) Limoges, B.; Marchal, D.; Mavré, F.; Savéant, J.-M. J. Am. Chem. Soc. 2006, 128, 6014−6015. (42) Mavré, F.; Bontemps, M.; Ammar-Merah, S.; Marchal, D.; Limoges, B. Anal. Chem. 2007, 79, 187−194. (43) Andrieux, C. P.; Limoges, B.; Savéant, J.-M.; Yazidi, D. Langmuir 2006, 22, 10807−10815. (44) Limoges, B.; Savéant, J.-M.; Yazidi, D. J. Am. Chem. Soc. 2003, 125, 9192−9203. (45) Durand, F.; Stines-Chaumeil, C.; Flexer, V.; André, I.; Mano, N. Biochem. Biophys. Res. Commun. 2010, 402, 750−754. (46) Olsthoorn, A. J. J.; Duine, J. A. Biochemistry 1998, 37, 13854− 13861. (47) Bourdillon, C.; Demaille, C.; Moiroux, J.; Savéant, J.-M. J. Am. Chem. Soc. 1993, 115, 2−10. (48) Fruk, L.; Kuhlmann, J.; Niemeyer, C. M. Chem. Commun. 2009, 230−232.

ASSOCIATED CONTENT

* Supporting Information S

Experimental details for the biotinylation of apo-GDH, the homogeneous reconstitution of PQQ-GDH from b-apo-GDH, the determination of the maximal surface concentration of bholo-GDH, and the measurement of the homogeneous affinity binding constant of PQQ for the apo-GDH binding sites. Other additional information is as indicated in the text. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions ∥

These authors contributed equally.

Notes

The authors declare no competing financial interests.



ACKNOWLEDGMENTS This research has been partially supported by Sino-Europe Joint Doctoral Promotion of the Chinese Academy of Sciences, 2012. R.M.C. thanks Fundación Ramón Areces for a postdoctoral grant. N.M. and C.S.C. thank the continuous support of the Région Aquitaine.



REFERENCES

(1) Roth, J. P.; Wincek, R.; Nodet, G.; Edmondson, D. E.; McIntire, W. S.; Klinman, J. P. J. Am. Chem. Soc. 2004, 126, 15120−15131. (2) Yorita, K.; Misaki, H.; Palfey, B. A.; Massey, V. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 2480−2485. (3) Efimov, I.; Cronin, C. N.; Bergmann, D. J.; Kuusk, V.; McIntire, W. S. Biochemistry 2004, 43, 6138−6148. (4) Grunden, A. M.; Jenney, F. E., Jr.; Ma, K.; Ji, M.; Weinberg, M. V.; Adams, M. W. Appl. Environ. Microbiol. 2005, 71, 1522−1530. (5) Fruk, L.; Kuo, C.-H.; Torres, E.; Niemeyer, C. M. Angew. Chem., Int. Ed. 2009, 48, 1550−1574. (6) Hefti, M. H.; Vervoort, J.; van Berkel, W. J. H. Eur. J. Biochem. 2003, 270, 4227−4242. (7) Hayashi, T.; Hitomi, Y.; Ando, T.; Mizutani, T.; Hisaeda, Y.; Kitagawa, S.; Ogoshi, H. J. Am. Chem. Soc. 1999, 121, 7747−7750. (8) Sato, H.; Watanabe, M.; Hisaeda, Y.; Hayashi, T. J. Am. Chem. Soc. 2005, 127, 56−57. (9) Kawakami, N.; Shoji, O.; Watanabe, Y. ChemBioChem 2012, 13, 2045−2047. (10) Berggren, G.; Adamska, A.; Lambertz, C.; Simmons, T. R.; Esselborn, J.; Atta, M.; Gambarelli, S.; Mouesca, J. M.; Reijerse, E.; Lubitz, W.; Happe, T.; Artero, V.; Fontecave, M. Nature 2013, 499, 66−69. (11) Morris, D. L.; Ellis, P. B.; Carrico, R. J.; Yeager, F. M.; Schroeder, H. R.; Albarella, J. P.; Boguslaski, R. C.; Hornby, W. E.; Rawson, D. Anal. Chem. 1981, 53, 658−665. (12) Tyhach, R. J.; Rupchock, P. A.; Pendergrass, J. H.; Skjold, A. C.; Smith, P. J.; Johnson, R. D.; Albarella, J. P.; Profitt, J. A. Clin. Chem. 1981, 27, 1499−1504. (13) Dosch, M.; Weller, M. G.; Bückmann, A. F.; Niessner, R. Fresen. J. Anal. Chem. 1998, 361, 174−178. (14) Wilson, R. Analyst 1992, 117, 1547−1551. J

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

(49) Zayats, M.; Katz, E.; Baron, R.; Willner, I. J. Am. Chem. Soc. 2005, 127, 12400−12406. (50) Xiao, Y.; Patolsky, F.; Katz, E.; Hainfeld, J. F.; Willner, I. Science 2003, 299, 1877−1881. (51) Liu, J.; Paddon-Row, M. N.; Gooding, J. J. Chem. Phys. 2006, 324, 226−235. (52) Hargrove, M. S.; Barrick, D.; Olson, J. S. Biochemistry 1996, 35, 11293−11299. (53) Dewanti, A. R.; Duine, J. A. Biochemistry 2000, 39, 9384−9392. (54) Shinagawa, E.; Matsushita, K.; Nonobe, M.; Adachi, O.; Ameyama, M.; Ohshiro, Y.; Itoh, S.; Kitamura, Y. Biochem. Biophys. Res. Commun. 1986, 139, 1279−1284. (55) Shen, D. The study of apoenzyme/prosthetic groups and their application in chemical analysis. Ph.D. Dissertation, University of Michigan, 2009. (56) Bourdillon, C.; Demaille, C.; Moiroux, J.; Savéant, J.-M. J. Am. Chem. Soc. 1994, 116, 10328−10329. (57) Blankespoor, R.; Limoges, B.; Schöllhorn, B.; Syssa-Magalé, J. L.; Yazidi, D. Langmuir 2005, 21, 3362−3375. (58) Balland, V.; Hureau-Sabater, C.; Cusano, A. M.; Liu, Y.; Tron, T.; Limoges, B. Chem.Eur. J. 2008, 14, 7186−7192. (59) Miranda-Castro, R.; Marchal, D.; Limoges, B.; Mavré, F. Chem. Commun. 2012, 48, 8772−8774. (60) Kivlehan, F.; Mavré, F.; Talini, L.; Limoges, B.; Marchal, D. Analyst 2011, 136, 3635−3642.

K

dx.doi.org/10.1021/ac500142e | Anal. Chem. XXXX, XXX, XXX−XXX