A

migiwmrTpf©

K

1

ifgi(acG

tui

0d

Biosensors and Bioelectronics 22 (2007) 3021–3027

Amperometric glucose biosensor based on self-assembly hydrophobinwith high efficiency of enzyme utilization

Zi-Xia Zhao, Ming-Qiang Qiao, Feng Yin, Bin Shao, Bao-Yan Wu, Yan-Yan Wang,Xin-Sheng Wang, Xia Qin, Sha Li, Lei Yu, Qiang Chen ∗

The Key Laboratory of Bioactive Materials Ministry of Education, College of Life Sciences, Nankai University, Tianjin 300071, China

Received 5 September 2006; received in revised form 10 November 2006; accepted 8 January 2007Available online 20 January 2007

bstract

Hydrophobins are a family of natural self-assembling proteins with high biocompability, which are apt to form strong and ordered assembly ontoany kinds of surfaces. These physical-chemical and biological properties make hydrophobins suitable for surface modification and biomolecule

mmobilization purposes. A class II hydrophobin HFBI was used as enzyme immobilization matrix on platinum electrode to construct amperometriclucose biosensor. Permeability of HFBI self-assembling film was optimized by selecting the proper HFBI concentration for electrode modification,n order to allow H2O2 permeating while prevent interfering compounds accessing. HFBI self-assembly and glucose oxidase (GOx) immobilizationas monitored by quartz crystal microbalance (QCM), and characterization of the modified electrode surface was obtained by scanning electronicroscope (SEM). The resulting glucose biosensors showed rapid response time within 6 s, limits of detection of 0.09 mM glucose (signal-to-noise

atio = 3), wide linear range from 0.5 to 20 mM, high sensitivity of 4.214 × 10−3 A M−1 cm−2, also well selectivity, reproducibility and lifetime.

he all-protein modified biosensor exhibited especially high efficiency of enzyme utilization, producing at most 712 �A responsive current forer unit activity of GOx. This work provided a promising new immobilization matrix with high biocompatibility and adequate electroactivity forurther research in biosensing and other surface functionalizing.

2007 Elsevier B.V. All rights reserved.

etbceup

r(fir

eywords: Biosensor; Self-assembly; Hydrophobin; Glucose oxidase

. Introduction

Enzyme immobilization has become a widely concernedssue in biosensing and surface functionalizing for years. Asor amperometric biosensors, the generally applied strategy toet stronger enzymatic responses is to increase the quantity ofmmobilized enzyme, such as deposition of enzyme multilayersChen et al., 1998; Hoshi et al., 2001; Ferreira et al., 2004; Davisnd Higson, 2005), or bringing in nanomaterials with large spe-ific surface area to immobilize more enzyme (Zhou et al., 2005;uan et al., 2005; Yang and Zhu, 2006; Xian et al., 2006).While we focus on another promising strategy: to improve

he efficiency of enzyme utilization. That means making bestse of the catalytic activity for per unit of enzyme and produc-ng electrochemical signal as much as possible under limited

∗ Corresponding author. Tel.: +86 22 23506173; fax: +86 22 23506122.E-mail address: qiangchen@nankai.edu.cn (Q. Chen).

oebsfct

956-5663/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2007.01.007

nzyme quantity. To realize high efficiency of enzyme utiliza-ion, the crucial first step should be searching for a highlyiocompatible immobilization matrix which well conserves theatalytic activity of enzymes and provides a mild and stablenvironment for enzymatic reactions. Hydrophobins, small nat-ral self-assembling proteins, provide new choices with greatotential.

Hydrophobins consisting approximately 100 amino acidesidues are the most powerful surface-active proteins knownHakanpaa et al., 2004). They are ubiquitously produced bylamentous fungi, playing various roles in fungal physiologyelated to surface phenomena, such as adhesion, formationf surface layers, and lowering of surface tension (Lindert al., 2002). As a remarkable property, hydrophobins haveoth hydrophobic and hydrophilic parts and self-assemble into

trong, highly ordered, amphiphilic films at almost any inter-aces, helping fungi adapt to a wide variety of environmentalonditions and to attach to various surfaces. Depending on theype of surface, various molecular interactions, such as van

3 Bioe

das

(fscs2ebdmba

iwHabfad(2

bctwhet

iaoaHaac(dhewta

tb

HteolSHestt

tTtcpawpbspu

2

2

1CwtNc(tACRwbpSHgi2

022 Z.-X. Zhao et al. / Biosensors and

er Waals interactions, hydrogen bonding, electrostatic inter-ctions, or hydrophobic interactions, may be involved in theelf-assembling process (Linder et al., 2002).

Since the first hydrophobin was identified in early 1990sWessels et al., 1991), a big family has been isolated from manyungi but not yet found in any other organism. The sequenceimilarity of hydrophobins is generally weak, except for theharacteristic and unique pattern of eight Cys residues with con-erved spacing, forming four disulfide bridges (Hakanpaa et al.,004; Hektor and Scholtmeijer, 2005). On the basis of differ-nces in hydropathy patterns of the amino acid sequences andiophysical properties, two different types of hydrophobins areistinguished, namely class I and class II (Wessels, 1994). For-ulas of the amino acid sequences for each class were given

elow, with X representing any other amino acid than cysteinend the subindex the number of amino acids:

Class I: X25–158-C-X5–9-C-C-X4–44-C-X7–23-C-X5–7-C-C-X6–18-C-X2–13.Class II: X17–165-C-X7–10-C-C-X11-C-X15/16-C-X6–9-C-C-X10/11-C-X3–8.

The members of both classes share properties such as highnterfacial activity and high tendency to adhere to surfaces,hich are essential for the surface modifying applications.owever, the class I hydrophobins tend to form aggregates

nd surface layers that are highly insoluble, which can onlye dissolved in strong acids such as trifluoroacetic acid orormic acid. Whereas the dissolution of class II hydrophobinssemblages seem to be relatively easier, for example, theyissolve in 2% sodium dodecyl sulfate (100 ◦C) or 60% ethanolLinder et al., 2002; Lumsdon et al., 2005; Kisko et al.,005).

Class I hydrophobins have been more intensively researchedecause they were discovered earlier. It’s reported that non-ovalent immobilization of lipases to the hydrophobic side ofhe hydrophobins resulted in an increase in lipase activity thatas comparable to the activation of lipases on conventionalydrophobic supports (Hektor and Scholtmeijer, 2005). Thelectrode modification with class I hydrophobins was also inves-igated (Bilewicz et al., 2001; Corvis et al., 2005).

But recently, class II hydrophobins, which used to be lessnvestigated, aroused much attention because their purificationnd structural studies have made great progress. HFBII is thenly hydrophobin for which the atomic resolution structure isvailable (Hakanpaa et al., 2004). Another class II hydrophobin,FBI, is a 7.5 kDa small protein having the same origin of fil-

mentous fungus Trichoderma reesei as HFBII, with an aminocid similarity of 69%. The purification and growth of HFBIrystals suitable for X-ray crystallography is also reportedAskolin et al., 2004). Atomic force microscopy and otheretailed analysis showed that both the two hydrophobins formedighly ordered monolayer films and crystalline fibrils (Torkkeli

t al., 2002; Linder et al., 2005). Specific structural informationould help understanding the self-assembling process on elec-

rodes, interactions between hydrophobin films and enzymes,nd their electrochemical behaviors during amperometric detec-

2

w

lectronics 22 (2007) 3021–3027

ions. That’s why we focus on class II hydrophobins in ouriosensor research.

Study on self-assembly of recombinant class I hydrophobinYDPt-1 from Pisolithus tinctorius on electrodes has indicated

hat hydrophobin layers blocked the access of small hydrophiliclectroactive probe K3Fe(CN)6 to the electrode surface and itsxidation (Bilewicz et al., 2001). While the same team reportedater that electrodes covered with another class I hydrophobinC3 from Schizophyllum commune was not only accessible to2O2, but also showed higher sensitivity to H2O2 than bare

lectrodes (Corvis et al., 2005). Considering H2O2 is a muchmaller molecular than K3Fe(CN)6, these results suggested ushat hydrophobin layers on electrodes may have size exclusionowards electroactive molecules in aqueous solutions.

To testify our assumption, we used cyclic voltammetry (CV)o check the permeability of HFBI films on platinum electrodes.hen HFBI was used as enzyme immobilization matrix, with

he well-studied glucose oxidase (GOx) as model enzyme toonstruct an amperometric glucose biosensor. Self-assembly ofrotein multilayers was monitored by quartz crystal microbal-nce (QCM), characterization of the modified electrode surfaceas obtained by scanning electron microscope (SEM), and theroperties of the resulting glucose biosensors were measuredy amperometric experiment and cyclic voltammetry (CV). Thetudy would provide a promising platform for fabricating all-rotein modified electrodes with high efficiency of enzymetilization.

. Experimental

.1. Reagents

�-d-Glucose oxidase from Aspergillus niger (GOx, EC.1.3.4, pI 4.5, 202 units/mg protein), was purchased from Sigmao. Hydrophobin (HFBI, pI 5.7, 22.3% in weight) from T. reeseias donated by Prof. M.Q. Qiao (The Key Laboratory of Bioac-

ive Materials Ministry of Education, College of Life Sciences,ankai University). 0.1 M acetate buffer solution (ABS, pH 5.1)

onsisting of NaAc and HAc, 0.1 M phosphate buffer solutionPBS, pH 6.8) consisting of Na2HPO4 and KH2PO4 were usedo dissolve proteins and employed as supporting electrolyte.luminum oxide nanopowder was purchased from Aldricho. Glucose was purchased from Tianjin damao Chemicaleagent Co. (Tianjin, China) and the stock solution of glucoseas allowed to mutorotate at room temperature overnightefore use. Uric acid purchased from Sigma Co., ascorbic acidurchased from Fluka Co., and acetaminophen purchased fromhanghai Chemical Reagents Co. (Shanghai, China) were ofPLC grade. All other chemical reagents were of analyticalrade and dissolved with deionized distilled water. All exper-ments were performed at room temperature, approximately5 ◦C.

.2. Self-assembly thin films analysis

A quartz crystal microbalance (QCM, Model QCA917ith a software Winchem, SEIKO EG&G Co.) was used to

Bioelectronics 22 (2007) 3021–3027 3023

mopdtbrsrtnbs

o2T1o

2

mecHowPt3wfT3staatw

2

PEiw(ewtwms

Fig. 1. Cyclic voltammetry showing the redox process of 0.1 mM H2O2 onHFBI film modified Pt electrodes. Electrodes were prepared in HFB solutionsof different concentrations as (2) 200 �g/mL, (3) 20 �g/mL, (4) 2 �g/mL, (5)0b5

3

3

vfiebPstaar

aga1ah1aHbpir

w

Z.-X. Zhao et al. / Biosensors and

onitor the assembly of protein films. The sensor composedf an AT-cut 9-MHz piezoelectric quartz resonator withlatinum thin layers (effective surface area: 0.19 ± 0.01 cm2)eposited on its two faces is excited to oscillation in thehickness shear mode at its fundamental resonant frequency,y applying a RF voltage across the electrodes near theesonant frequency. Protein concentrations for QCM mea-urement were 20 �g/mL for HFBI and 2 mg/mL for GOx,espectively. The volume of every injection was 20 �L andhe existing solution in the cell would be removed beforeext injection. Different pH environments were controlledy different buffer solutions as protein solvents and rinsingolution.

Surface images of the modified QCM substrates werebtained by scanning electron spectroscopy (SEM, QUANTA00, FEI Co.) operated at an accelerating voltage of 20.0 kV.he substrates were rinsed with deionized distilled water for0 min before SEM imaging in order to avoid the interferencef salt crystallization from buffer solutions.

.3. Electrodes modification with proteins

Platinum electrodes (3 mm diameter) were polished with alu-inum oxide nanopowder, and then washed ultrasonically in

thanol and deionized distilled water, each for an hour. Theleaned electrodes were dried with high-purified N2 stream.FBI was dissolved in different buffer solutions (pH 6.8 PBSr pH 5.1 ABS) at desired concentrations (200, 20 or 2 �g/mL),hile GOx was dissolved in different buffer solutions (pH 6.8BS or pH 5.1 ABS) at a concentration of 2 mg/mL. The pre-

reated electrodes were immersed in certain HFBI solutions for0 min and rinsed in corresponding buffer solutions for 10 minhich was applied at the end of each assembly deposition

or dissociating the weak adsorption without drying-procedure.hen electrodes were transferred into proper GOx solutions for0 min, followed by a 10-min-rinsing in corresponding bufferolutions. Finally, to get more uniform films, the modified elec-rodes were put on a shelf with the electrode surface upward tollow it dry overnight at 4 ◦C. Each electrode was dust-protectednd kept at constant humidity conditions. The modified elec-rodes were stored in corresponding buffer solutions at 4 ◦Chen not in use.

.4. Electrochemical measurements

Electrochemical measurements were carried out by aotentiostat–Galvanostat (Model 283 with a software M270,G&G Co.). A conventional three-electrode system compris-

ng the modified Pt electrode as working electrode, a Ptire (1 mm diameter) as counter electrode and an Ag/AgCl

saturated KCl) as reference electrode was employed for alllectrochemical experiments in an electrochemical cell filledith 20 mL of 0.1 M ABS or 0.1 M PBS at room tempera-

ure. In cyclic voltammetry (CV) measurements, the scan rateas set at 50 mV/s. In steady-state amperometric measure-ents, the potential was set at 400 mV under gently magnetic

tirring.

thhw

�g/mL while no enzyme was immobilized, compared with blank reference (1)are electrode with no substrate. Measurements were taken at pH 5.1, scan rate0 mV/s.

. Results and discussion

.1. Permeability of the HFBI films

The permeability of HFBI films was checked by cyclicoltammetry (CV). Fig. 1 showed the responses of HFBIlm modified Pt electrodes to 0.1 mM H2O2, H2O2 generatednzymatically is the signal molecular of most amperometriciosensors whose oxidation current can be accurately detected.t electrodes prepared in 20 and 2 �g/mL HFBI solutions wereensitive to H2O2 although the current responses are a little lowerhan that of bare electrode. But H2O2 was apparently blocked toccess the surface of Pt electrode prepared in 200 �g/mL HFBI,s its current response was less than half of the bare electrodeesponse.

The responses of HFBI film modified Pt electrodes to 0.1 mMscorbic acid, uric acid and acetaminophen were also investi-ated, which are common interfering compounds in blood formperometric measurements. With larger molecular weights of68.11, 176.13 and 151.17, respectively, ascorbic acid, uriccid and acetaminophen were more easily to be blocked byydrophobin films than H2O2 (MW: 34.01). Supplemental Fig.showed the current responses to ascorbic acid, uric acid and

cetaminophen of Pt electrodes prepared in 200 and 20 �g/mLFBI solutions ranged from 1.4% to 19% of the correspondingare electrode responses, while the responses of Pt electrodesrepared in 2 �g/mL HFBI solution were much higher, reach-ng 62%, 70% and 79% of the corresponding bare electrodeesponses.

These results indicated that the permeability of HFBI filmsas controllable though varying the hydrophobin concentra-

ions during electrode modification, which thus affected the

ydrophobin density on electrode surfaces. As a result, optimumydrophobin concentration for biosensors could be selected,hich was 20 �g/mL in this work. Under this concentration,

3024 Z.-X. Zhao et al. / Biosensors and Bioe

Fig. 2. Frequency changes of quartz crystal microbalance during sequentialHFBI self-assembly and GOx immobilization on Pt coated substrate at pH 5.1.The start baseline was obtained when the substrate was exposed to correspondingbiG

Hee

attf

3

sctw(61ipnhia

�

wcμ

qa5

s

3cs

atipfitip

cGamddpdis(latagw

3

s5ietegfiswtgcbgGtk

uffer solution. The arrows indicated the removing of existing solution andnjection of 20 �L new solution: 20 �g/mL HFBI, buffer solution, 2 mg/mL forOx, and buffer solution in turn.

FBI film formed on Pt electrode was permeable to H2O2 butfficiently prevents interfering compounds from accessing thelectrode surface.

CV analysis of the H2O2 redox process (Fig. 1) also exhibitedshift of oxidation peak potential from +550 mV on bare elec-

rode to +400 mV on HFBI film modified electrodes, suggestinghat HFBI films might provide catalysis surfaces for H2O2 andacilitate its oxidation.

.2. Enzyme immobilizing analysis

The self-assembly of HFBI films on Pt coated substrates andubsequent immobilization of GOx was monitored by quartzrystal microbalance (QCM). After rinsing with buffer solutiono wash down the weak adsorption, the average frequency shiftsere −56.1 ± 0.8 for HFBI, −157.6 ± 2.0 for GOx at pH 5.1

Fig. 2), and −50.1 ± 0.5 for HFBI, −79.5 ± 3.0 for GOx at pH.8 (Supplemental Fig. 2), with RSD of 0.45%, 0.88%, 0.57%,.5%, respectively. The decrease in resonant frequency (�f) isnduced by small mass added to the substrate (�m), which isroportional to �f, provided the mass is evenly distributed, doesot slip on the electrode, and is sufficiently rigid and thin toave negligible internal friction. Relationship between the twos given by the following equation (Sauerbrey, 1959; Hook etl., 1998):

f = −2F20 (ρqμq)−1/2 �mA−1

here F0 is the fundamental oscillation frequency of the dryrystal (9.0 × 106 Hz), ρq the density of quartz (2.648 g cm−3),q the shear modulus (2.947 × 1011 g cm−1 s−2), and A is theuartz area (0.19 cm2). For the quartz crystal used in this work,

frequency change of 1 Hz corresponds to a mass increase of

.45 ng cm−2.According to the Sauerbrey equation, the masses of HFBI

elf-assembly on Pt coated surfaces were calculated to be

tcas

lectronics 22 (2007) 3021–3027

06 ng cm−2 at pH 5.1 and 273 ng cm−2 at pH 6.8. They werelose to the reported binding level of 290 ng cm−2 on hydrophilicilanized surface (Linder et al., 2002).

With respective isoelectric points (pI) of 5.7 and 4.5, HFBInd GOx were oppositely charged at pH 5.1, which was exactlyhe optimum pH for catalytic activity of GOx. With electrostaticnteractions enhancing the affinity between HFBI and GOx, thisH value was more suitable for GOx immobilization on HFBIlm than the commonly used moderate pH 6.8, at which the

wo proteins were both negatively charged. QCM measurementsndicated the mass of immobilized GOx was 859 ng cm−2 atH5.1, compared with 433 ng cm−2 at pH 6.8.

Scanning electron spectroscopy (SEM) images (Fig. 3)onfirmed that a slightly acidic environment promoted theOx immobilization. Immobilized GOx, clearly differentiated

s bright particles, appeared to be aggregations rather thanonomolecules. For their sizes varied from less than one hun-

red nanometers to several hundred nanometers, while theimensions of native GOx dimer were 70 A × 55 A × 80 A, com-osed of two compact spheroidic subunits with approximateimensions of 60 A × 52 A × 37 A (Hecht et al., 1993). SEMmages of surface modified at pH 5.1 (Fig. 3a and b) showedignificant quantities of aggregated GOx, whereas at pH 6.8Fig. 3c and d) GOx could only be occasionally seen, often witharger particle sizes suggesting stronger aggregation at moder-te pH. The highly ordered structure of HFBI films, of whichhe unit cell ranged from a = b = 38 A (Torkkeli et al., 2002) to= b = 54 ± 1 A (Kisko et al., 2005), were not able to be distin-uished from the SEM images. And the roughness of the surfaceas due to the Pt coated QCM substrate itself.

.3. Amperometric response of the glucose biosensor

Pt electrode modified by HFBI and GOx exhibited rapid andensitive current response to successively added glucose at pH.1 with +400 mV working potential versus Ag/AgCl, as shownn Fig. 4. The current response represented the velocity of thenzymatic reaction. It well corresponded to the kinetics ofypical enzyme catalytic reaction, defined by Michaelis–Mentenquation: V = Vmax [S]/([S] + Km). At very low concentration oflucose, the enzyme catalytic reaction displayed approximatelyrst-order kinetics, with reaction velocity (V) proportional toubstrate concentration ([S]), because the quantity of GOxas constant and far from saturated by glucose. Increasing

ogether with the glucose concentration, the reaction velocityradually stopped increasing linearly but displayed compli-ated mixed-order kinetics, for the substrate concentrationecame comparable to the quantity of enzyme. When thelucose concentration increased overmuch for the immobilizedOX, the reaction velocity became more and more close to

he maximum velocity, displaying a approximate zero-orderinetics.

Michaelis constant (Km) was the concentration of substrate

hat leads to half-maximum velocity. The apparent Michaelisonstant (Kapp

m ) was determined by analysis of the slopend intercept for the plot of the reciprocals of the steady-tate current versus glucose concentration, for they fitted the

Z.-X. Zhao et al. / Biosensors and Bioelectronics 22 (2007) 3021–3027 3025

Fig. 3. Scanning electron spectroscopy (SEM) images of QCM substrates modified b(c) pH 6.8, scale bar 20.0 �m; (d) pH 6.8, scale bar 2.0 �m.

Fig. 4. Amperometric glucose response of Pt electrode modified by HFBI andGOx at pH 5.1 with +400 mV working potential. The inset was the front part ofthe whole curve with enlarged scale of Y-axis, showing specific details.

L

wic

Sltuwi

atndli

y HFBI and GOx. (a) pH 5.1, scale bar 20.0 �m; (b) pH 5.1, scale bar 2.0 �m;

ineweaver–Burk equation (Kamin and Wilson, 1980):

1

iss=(

Kappm

imax

)(1

C

)+(

1

imax

)

here iss is the steady-state current after the addition of substrate,max the maximum current measured under saturated substrateondition, and C is the bulk concentration of the substrate.

After five independent repeating experiments (data shown inupplemental Fig. 3), the K

appm of immobilized GOx was calcu-

ated to be 16.8 mM. While the Km value of dissolved GOx athe same condition (pH 5.1, 25 ◦C) was 25 mM, given by prod-ct introductions of Sigma Co., The K

appm of immobilized GOx

as a little lower, indicating higher affinity for glucose aftermmobilization.

Parameters of the amperometric glucose biosensor werelso obtained from the five measurements. The maximumime from glucose addition to achieving steady state current,

amely response time, was no more than 6 s. The limits ofetection (LOD) of glucose was 0.09 mM (S/N = 3), and theinear range was 0.5–20 mM, wide enough to cover the clin-cally significant glucose concentrations ranged from 0.5 to

3 Bioelectronics 22 (2007) 3021–3027

5aeyto

fb(b(abcaSbfao2rfot

cwArtcttde

3

o

Fig. 5. The amperometric response to 20 mM glucose of Pt electrode modifiedbs

dcrcaipi

ialiodtwa

TE

I

CPLLZCS

H

026 Z.-X. Zhao et al. / Biosensors and

mM in tear to 2–12 mM in blood for diabetics (Badugu etl., 2005). The steady state calibration curve within the lin-ar range (inset of Supplemental Fig. 3) was characterized as= 0.00176 + 0.2979x, (R = 0.99994, P < 0.0001). Considering

he surface area of the electrode was 0.0707 cm2, the sensitivityf the glucose biosensor was 4.214 × 10−3 A M−1 cm−2.

Glucose biosensors are susceptible to the possible inter-erences such as ascorbic acid, uric acid and acetaminophenecause electrooxidation of H2O2 occurs at high potentialHoshi et al., 2001), while these influences may be restrictedy the application of appropriate inner or outer membranesThevenot et al., 2001). Thevenot et al. (2001) recommendedn easily quantified procedure to determine the selectivity ofiosensor: interfering substances are added, at their expectedoncentration, into the measuring cell, already containing usualnalyte concentration, at the mid-range of its expected value.electivity is then expressed as the percentage of variation of theiosensor response. The physiological concentrations of inter-ering substances were 0.1 mM for ascorbic acid, 0.5 mM for uriccid and 0.1 mM for acetaminophen, while the average physi-logical concentration of blood glucose was 4 mM (Wu et al.,007). Compared with the glucose current, the interfering cur-ents were 48.2% for ascorbic acid, 164% for uric acid and 29.6%or acetaminophen, respectively. As discussed in Section 3.1, theptimum permeability of hydrophobin film contributes a lot tohe selectivity of the glucose biosensor, due to size exclusion.

The lifetime of the enzyme electrode was determined by suc-essive and repeating detection of 20 mM glucose in 100 days,hile the electrode was stored in 4 ◦C buffer solution after use.s shown in Fig. 5, more than 85% of its maximum current

esponse remained after 100 days, indicating that the Pt elec-rode modified by HFBI and GOx films well conserved theatalytic activity of GOx. It’s also demonstrated in the figurehat the maximum current response did not appear right afterhe enzyme modification of electrode, but presented itself oneay after the modification due to protein rearrangement on thelectrode surface and bioactivity recovery of enzyme.

.4. Efficiency of enzyme utilization

To characterize the responsive current produced by per unitf immobilized enzyme for amperometric biosensors, we intro-

vpre

able 1fficiency of enzyme utilization for different glucose biosensors

mmobilization matrix Maximum current(�A)

Enzymequantity (ng)

hitosan and gold nanocomposite 13.2 641oPD/GOD-GA/PB composite 7.96 24,000atex 58.8 500,000aponite 80.4 66,000n3Al(OH)8Cl·2H2O 81.7 400,00/Fe nanocomposite 11.6 500iO2 contained nanocomposite ∼15a 22,500

ydrophobin HFBI 8.76 60.7

a Not accurately provided, estimated from amperometric response figure.b Not provided.

y HFBI and GOx after storage in 4 ◦C pH 5.1 ABS. Error bars showed thetandard deviation (S.D.) of five repeating measurements.

uced the concept “efficiency of enzyme utilization”, which wasalculated when all immobilized enzyme molecules were satu-ated by substrates and the biosensor achieved its maximumurrent. High efficiency of enzyme utilization depended on twospects: well conservation of the original enzyme catalytic activ-ty during immobilization and storage, potent electron transferathway to realize transfer from enzymatic reaction informationnto electronic signal.

QCM measurements in Section 3.2 indicated the mass ofmmobilized GOx was 859 ng cm−2. Considering the surfacerea of the electrode was 0.0707 cm2, the quantity of immobi-ized GOx on the electrode surface would be 60.7 ng. GOx usedn this work contained 202 catalytic units per milligram protein,f which one unit would oxidize 1.0 �mol �-d-glucose to-gluconic acid and H2O2 per minute at pH 5.1 at 35 ◦C. So the

otal original activity of immobilized enzyme on the electrodeould be 1.23 × 10−2 units. It must be mentioned that the

ctual activity might be lower than the theoretically calculated

alue, due to loss of protein activity during the immobilizingrocedures. But the original activity of immobilized enzyme,ather than the residual activity, was used in calculation forfficiency of enzyme utilization, because it reflected the

Total activity(unit)

Efficiency of enzymeutilization (�A unit−1)

Reference

0.129 102 Wu et al. (2007)2.4 3.32 Deng et al. (2006)

57.5 1.02 Cosnier et al. (2006)6.8 11.8 Cosnier et al. (2006)8.7 0.939 Cosnier et al. (2006)

–b – Wu et al. (2005)4.016 ∼3.74 Yang and Zhu (2005)

0.0123 712 This work

Bioe

pwcecb

stOirisfopGu

4

bbtww4ihfichilm

mermt

A

etdfaw

A

i

R

A

B

B

C

C

C

DD

F

G

H

H

H

H

KHK

L

L

L

ST

T

W

WW

W

X

Z.-X. Zhao et al. / Biosensors and

reserval circumstance of enzyme catalytic activity in thehole immobilizing strategy. While the average maximum

urrent in five measurements was 8.76 �A, the efficiency ofnzyme utilization was calculated to be 712 �A unit−1. Weompared this value with some previously reported glucoseiosensors.

In Table 1 recently reported glucose biosensors withtrong amperometric responses were chosen for contrast withhe Pt electrode modified by HFBI and GOx in our work.bviously, HFBI self-assembling film exhibited superiority

n efficiency of enzyme utilization, with hundred times higheresponsive current for per enzyme unit than commonly usedmmobilization matrix. That might be due to its unique proteintructure, providing a particularly suitable microenvironmentor GOx, another protein, to keep the active conformation andrientation. On the other hand, the simple thin film ensuredrompt and sufficient signal transfer from the active sites ofOx to electrode surface. As a result, HFBI helped making fullse of GOx catalytic activity.

. Conclusions

In this work an amperometric glucose biosensor haseen fabricated based on Pt electrode modified successivelyy HFBI and GOx. Under optimized conditions for elec-rode preparation, the biosensor showed rapid response timeithin 6 s, limits of detection of 0.09 mM glucose (S/N = 3),ide linear range from 0.5 to 20 mM, high sensitivity of.214 × 10−3 A M−1 cm−2, also well selectivity, reproducibil-ty and lifetime. But its most prominent characteristic is theigh efficiency of enzyme utilization. The all-protein modi-ed electrode is able to produce at most 712 �A responsiveurrent for per unit activity of GOx, which is hundred timesigher than normal polymer matrix modified electrodes. Sot saves a large amount of enzyme, suggesting the immobi-ization strategy especially meaningful for expensive bioactive

olecules.This all-protein-based surface functionalization will show

ore significance in future biosensing research, such as microlectrodes, microfluidic devices, lab-on-a-chip, which call forestricted sizes of equipments, limited quantity of bioactiveolecules and therefore high efficiency of biomaterial utiliza-

ion.

cknowledgements

The supports by the Grant-in-Aid for the High-Tech Coop-ration Project of Nankai University & Tianjin University fromhe Chinese Ministry of Education, the National Science Foun-

ation of China (Grant no. 90409021), and the Innovation Fundor Technology Based Firms from Tianjin Municipal Sciencend Technology Commission (Grant no. 05ZHCXGX11600) areell acknowledged.

YYZ

lectronics 22 (2007) 3021–3027 3027

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at doi:10.1016/j.bios.2007.01.007.

eferences

skolin, S., Turkenburg, J.P., Tenkanen, M., Uotila, S., Wilson, K.S., Penttila,M., Visuri, K., 2004. Acta Cryst. D: Biol. Crystallogr. 60 (10), 1903–1905.

adugu, R., Lakowicz, J.R., Geddes, C.D., 2005. Curr. Opin. Biotechnol. 16 (1),100–107.

ilewicz, R., Witomski, J., Van der Heyden, A., Tagu, D., Palin, B., Rogalska,E., 2001. J. Phys. Chem. B 105 (40), 9772–9777.

hen, Q., Kobayashi, Y., Takeshita, H., Hoshi, T., Anzai, J., 1998. Electroanal-ysis 10 (2), 94–97.

orvis, Y., Walcarius, A., Rink, R., Mrabet, N.T., Rogalska, E., 2005. Anal.Chem. 77 (6), 1622–1630.

osnier, S., Mousty, C., Gondran, C., Lepellec, A., 2006. Mater. Sci. Eng. C:Biomim. Supramol. Syst. 26 (2–3), 442–447.

avis, F., Higson, S.P., 2005. Biosens. Bioelectron. 21 (1), 1–20.eng, C., Li, M., Xie, Q., Liu, M., Tan, Y., Xu, X., Yao, S., 2006. Anal. Chim.

Acta 557 (1), 85–94.erreira, M., Fiorito, P.A., Oliveira, O.N., de Torresi, S.I.C., 2004. Biosens.

Bioelectron. 19 (12), 1611–1615.uan, W.J., Li, Y., Chen, Y.Q., Zhang, X.B., Hu, G.Q., 2005. Biosens. Bioelec-

tron. 21 (3), 508–512.akanpaa, J., Paananen, A., Askolin, S., Nakari-Setala, T., Parkkinen, T., Pent-

tila, M., Linder, M.B., Rouvinen, J., 2004. J. Biol. Chem. 279 (1), 534–539.echt, H., Kalisz, H., Hendle, J., Schmid, R., Schomburg, D., 1993. J. Mol.

Biol. 229 (1), 153–172.oshi, T., Saiki, H., Kuwazawa, S., Tsuchiya, C., Chen, Q., Anzai, J., 2001.

Anal. Chem. 73 (21), 5310–5315.ook, F., Rodahl, M., Kasemo, B., Brzezinski, P., 1998. Proc. Natl. Acad. Sci.

95 (21), 12271–12276.amin, R.A., Wilson, G.S., 1980. Anal. Chem. 52 (8), 1198–1205.ektor, H.J., Scholtmeijer, K., 2005. Curr. Opin. Biotechnol. 16 (4), 434–439.isko, K., Torkkeli, M., Vuorimaa, E., Lemmetyinen, H., Seeck, O.H., Linder,

M., Serimaa, R., 2005. Surf. Sci. 584 (1), 35–40.inder, M., Szilvay, G.R., Nakari-Setala, T., SoDerlund, H., Penttila, M., 2002.

Protein Sci. 11 (9), 2257–2266.inder, M.B., Szilvay, G.R., Nakari-Setala, T., Penttila, M.E., 2005. FEMS

Microbiol. Rev. 29 (5), 877–896.umsdon, S.O., Green, J., Stieglitz, B., 2005. Colloid Surf. B: Biointerf. 44 (4),

172–178.auerbrey, G., 1959. Z. Phys. 155 (2), 206–222.hevenot, D.R., Toth, K., Durst, R.A., Wilson, G.S., 2001. Biosens. Bioelectron.

16 (1–2), 121–131.orkkeli, M., Serimaa, R., Ikkala, O., Linder, M., 2002. Biophys. J. 83 (4),

2240–2247.essels, J.G., De Vries, O.M., Asgeirsdottir, S.A., Springer, J., 1991. J. Gen.

Microbiol. 137 (10), 2439–2445.essels, J.G., 1994. Annu. Rev. Phytopathol. 32, 413–437.u, B.Y., Hou, S.H., Yin, F., Li, J., Zhao, Z.X., Huang, J.D., Chen, Q., 2007.

Biosens. Bioelectron. 22 (6), 838–844.u, J., Zou, Y., Gao, N., Jiang, J., Shen, G., Yu, R., 2005. Talanta 68 (1–2),

12–18.ian, Y., Hu, Y., Liu, F., Xian, Y., Wang, H., Jin, L., 2006. Biosens. Bioelectron.

21 (10), 1996–2000.ang, H., Zhu, Y., 2005. Anal. Chim. Acta 554 (1–2), 92–97.ang, H., Zhu, Y., 2006. Talanta 68 (3), 569–574.hou, H., Chen, H., Luo, S., Chen, J., Wei, W., Kuang, Y., 2005. Biosens.

Bioelectron. 20 (7), 1305–1311.