A

wbEtbEiots©

K

1

bmtPla2tpem

0d

Biosensors and Bioelectronics 22 (2007) 2408–2414

Interdigitated array microelectrode based impedance biosensorcoupled with magnetic nanoparticle–antibody conjugates for

detection of Escherichia coli O157:H7 in food samples

Madhukar Varshney a, Yanbin Li a,b,∗a Department of Biological and Agricultural Engineering, University of Arkansas, Fayetteville, AR 72701, United States

b O-411 Poultry Science Building, Center of Excellence for Poultry Science, University of Arkansas, Fayetteville, AR 72701, United States

Received 7 April 2006; received in revised form 25 July 2006; accepted 23 August 2006Available online 12 October 2006

bstract

An impedance biosensor based on interdigitated array microelectrode (IDAM) coupled with magnetic nanoparticle–antibody conjugates (MNAC)as developed and evaluated for rapid and specific detection of E. coli O157:H7 in ground beef samples. MNAC were prepared by immobilizingiotin-labeled polyclonal goat anti-E. coli antibodies onto streptavidin-coated magnetic nanoparticles, which were used to separate and concentrate. coli O157:H7 from ground beef samples. Magnitude of impedance and phase angle were measured in a frequency range of 10 Hz to 1 MHz in

he presence of 0.1 M mannitol solution. The lowest detection limits of this biosensor for detection of E. coli O157:H7 in pure culture and groundeef samples were 7.4 × 104 and 8.0 × 105 CFU ml−1, respectively. The regression equation for the normalized impedance change (NIC) versus. coli O157:H7 concentration (N) in ground beef samples was NIC = 15.55N − 71.04 with R2 = 0.95. Sensitivity of the impedance biosensor was

mproved by 35% by concentrating bacterial cells attached to MNAC in the active layer of IDAM above the surface of electrodes with the help

f a magnetic field. Based on equivalent circuit analysis, it was observed that bulk resistance and double layer capacitance were responsible forhe impedance change caused by the presence of E. coli O157:H7 on the surface of IDAM. Surface immobilization techniques, redox probes, orample incubation were not used in this impedance biosensor. The total detection time from sampling to measurement was 35 min.

2006 Elsevier B.V. All rights reserved.

c nano

if

bscfaont

eywords: Interdigitated array microelectrode; Impedance biosensor; Magneti

. Introduction

Escherichia coli O157:H7 is one of the most harmful food-orne pathogenic bacteria and is alone responsible for an esti-ated 73,000 cases of infection and 61 deaths that occur in

he United States each year (Centers of Disease Control andrevention, 2005). The infection of E. coli O157:H7 may cause

ife-threatening complications – hemolytic uremic syndromend hemorrhagic colitis in humans (Deisingh and Thompson,004; Yu et al., 2001). Consumption of beef, sprouts, let-uce, salami, unpasteurized milk, and juice contaminated with

athogens are some sources of infection (Lekowska-Kochaniakt al., 2002). As the loss caused by E. coli O157:H7 is enor-ous in terms of medical cost and product recall, it is extremely

∗ Corresponding author. Tel.: +1 479 575 2424; fax: +1 479 575 7139.E-mail address: yanbinli@uark.edu (Y. Li).

danrmi2

956-5663/$ – see front matter © 2006 Elsevier B.V. All rights reserved.oi:10.1016/j.bios.2006.08.030

particles; Bacterial detection; E. coli O157:H7

mportant to rapidly and specifically detect E. coli O157:H7 inood products.

Till now most of biosensors used for detection of pathogenicacteria are label-dependent immunosensors based on labeledecondary antibodies. Label-free biosensors including quartzrystal microbalance (QCM) (Fung and Wong, 2001) and sur-ace plasmon resonance (SPR) (Koubova et al., 2001) havettractive advantages with respect to speed and simplicity ofperation. Impedance is yet another rapid and inexpensive alter-ative technique to develop label-free biosensors for the detec-ion of bacteria. Generally impedance measurement can beivided into two categories: faradic and non-faradic (Yang etl., 2004). Faradic measurement requires a redox probe, whileon-faradic measurement can be performed in the absence of a

edox probe. Both faradic and non-faradic impedance measure-ent can be performed by capturing bacterial cells to antibodies

mmobilized on the surface of electrodes (Radke and Alocilja,005). The major problem associated with antibody immobi-

d Bio

lstftoCg1im

silaMctfstscl(

bcMtcattmtrum

2

2

tKaaaI

2

(

Gbnuw1

2

FMcm4ag(la

2

(piu(d

2b

caca3tBsoPf(4iwmou

M. Varshney, Y. Li / Biosensors an

ization is the low capture efficiency (CE) of the immobilizedurface. As a result, the functional surface area (where target bac-erial cells are detected) of an electrode is not optimally utilizedor detection. Anti-E. coli O157:H7 antibodies immobilized onhe surface of indium tin oxide coated glass electrode showednly 16% CE for E. coli O157:H7 (Ruan et al., 2002) and theE of anti-Salmonella antibodies functionalized on roughenedlass surface was less than 1% for Salmonella (Brewster et al.,996). Therefore, this issue needs to be addressed by develop-ng impedance methods without immobilization of biosensing

aterial on the electrodes.In recent years, use of nanoparticles has opened new dimen-

ions in immunomagnetic separation (IMS). From a functional-ty point of view nanoparticles are used as magnetic carriers orabels (Fritzsche and Taton, 2003; Liberti et al., 1996; Mao etl., 2006; Stoimenov et al., 2002) in a variety of applications.agnetic particles are very useful for the separation of target

ells from the mixture of bacteria and food matrix and also helpo concentrate separated cells into a very small volume suitableor impedance measurements with IDAM. The microelectrodescan a region called “active layer”, which is few microns abovehe surface with maximum strength of an electric field. The sen-itivity of IMS based impedance methods can be improved byoncentrating bacterial cells attached to MNAC in the activeayer of microelectrodes under the influence of a magnetic fieldGerwen et al., 1998).

In this study, we developed and evaluated a label-free MNACased impedance biosensor using IDAM for the rapid and spe-ific detection of E. coli O157:H7 in ground beef samples.NAC were used for: (1) separation of target cells from the mix-

ure of bacteria and food matrix, (2) concentration of separatedells into a small volume, and (3) concentration of cells in thective layer of IDAM. No redox probe, chemical immobilizationechniques, and sample incubation were used for the separa-ion, concentration or detection of bacteria. A low conductivity

annitol solution was used to keep cells alive and to minimizehe effect of medium impedance on direct detection of bacte-ial cells. An electrical equivalent circuit was proposed to betternderstand parameters, which were significant for impedanceeasurement of target bacteria using IDAM.

. Materials and methods

.1. Culture and plating of bacteria

Frozen stock of E. coli O157:H7 (ATCC 43888) was main-ained in brain heart infusion broth (BHI, Remel Inc., Lenexa,S) at −70 ◦C. The culture was harvested in BHI maintained

t 37 ◦C for 18–22 h. For enumeration, pure cultures were seri-lly diluted in 0.01 M, pH 7.4 phosphate buffered saline (PBS)nd surface plated on sorbitol MacConkey (SMAC) agar (Remelnc., Lenexa, KS), which was incubated at 37 ◦C for 20 to 22 h.

.2. Chemicals and reagents

PBS (0.01 M, pH 7.4) was obtained from Sigma–AldrichSt. Louis, MI). Bovine serum albumin (BSA; EM Science,

2

i

electronics 22 (2007) 2408–2414 2409

ibbstown, NJ), 1.0% (wt vol−1) was prepared in PBS as alocking buffer (PBS BSA). One-tenth molar solution of man-itol (Sigma–Aldrich, St. Louis, MI)) in deionized water wassed for washing and resuspension of bacteria. All solutionsere prepared with deionized water from Millipore (Milli-Q,8.2 M� cm, Bedford, MA).

.3. Nanoparticles and antibodies

Magnetic nanoparticles (average diameter 145 nm, 0.5 mge ml−1) conjugated with streptavidin were obtained fromolecular Probes Inc. (Eugene, OR). Magnetic nanoparti-

les have more than 85% of oxide as Fe3O4, approxi-ately 80% wt wt−1 of magnetite, and an approximately× 1011 particles mg−1 Fe. Affinity purified polyclonal goatntibodies against E. coli (specific for O and K antigens) conju-ated with biotin were obtained from Biodesign InternationalSaco, ME). The concentration of stock solution of biotin-abeled antibodies was 4–5 mg ml−1. A 1:10 dilution of thentibodies was prepared in PBS (0.01 M, pH 7.4) before use.

.4. Interdigitated array microelectrode

Gold IDAM chip was obtained from ABtech Scientific Inc.Richmond, VA), of total size 2 cm × 1 cm × 0.05 cm with 50air of fingers each measuring 15 �m in width and 4985 �mn length. The space between the fingers was 15 �m. Beforese, the IDAM chip was cleaned with 0.1 M sodium hydroxide15 min), 0.1 M hydrochloric acid (15 min), acetone (5 min), andeionized water, and then was dried in a stream of nitrogen.

.5. Immunomagnetic separation and concentration ofacteria

MNAC were prepared in 1.7 ml of sterile polypropyleneentrifuge tubes. Biotin-labeled polyclonal goat anti-E. colintibodies (7.5 �l) were continuously mixed with streptavidin-oated magnetic nanoparticles (15 �l) in 250 �l PBS BSAt 7 rpm on a variable speed rotator (ATR, Laurel, MD) for5 min at room temperature. Following antibody immobiliza-ion, MNAC were mixed with 150 �l of biotin solution (in PBSSA) for 15 min to block unbound streptavidin present on the

urface of magnetic nanoparticles. Excess biotin was washedut with PBS BSA, and MNAC were resuspended in 450 �l ofBS BSA. Serial dilutions of pure culture of E. coli O157:H7rom 7.4 × 102 to 7.4 × 108 CFU ml−1 were prepared in PBS0.01 M, pH 7.4) buffer. A 50 �l of pure culture was mixed with50 �l of MNAC for an immunoreaction time of 15 min. Follow-ng the immunoreaction, nanoparticles-bacteria complexes wereashed three times with 0.1 M mannitol solution with an inter-ittent magnetic separation, and were concentrated in 100 �l

f mannitol solution. Only 2 �l of the concentrated sample wassed for impedance measurement.

.6. Impedance measurement

Impedance measurement was performed using an IM-6mpedance analyzer (BAS, West Lafayette, IN) with IM-

2 d Bioelectronics 22 (2007) 2408–2414

6mastt1tanEtpatI7e

wacramOig

N

wpctc

2

A2tUtsSd8

3

3t

r

Fig. 1. Bode diagrams of impedance spectra of IDAM based impedance biosen-sor for mannitol solution, MNAC (in the absence of a magnetic field), purebacterial cells (7.4 × 107 CFU ml−1 of E. coli O157:H7), and bacterial cells(7.4 × 107 CFU ml−1 of E. coli O157:H7) attached to MNAC in the pres-ence and absence of a magnetic field. Impedance measurement was performedi1

wcrsfid2b((it3tfi(tWittand2i

410 M. Varshney, Y. Li / Biosensors an

/THALES software. For all impedance measurements, a sine-odulated ac potential of 100 mV was applied across IDAM

nd the magnitude of impedance and phase angle were mea-ured for frequency ranging from 10 Hz to 1 MHz. Assuminghat the MNAC were prepared before the test, the total detec-ion time from sampling to measurement was 35 min (15, 10, and0 min for immunoreaction, washing, and measurement, respec-ively). A 2 �l of sample was uniformly spread over the exposedrea of electrodes for impedance measurement. Nickel-platedeodymium rare earth magnet (0.4 in. × 0.2 in. × 0.105 in.) (Alllectronics Corp., Van Nuys, CA) was used to concentrate bac-

erial cells attached to MNAC in the active layer of IDAM. Oneole of the IDAM chip was connected to test and sense probes,nd the other pole was connected to counter and reference elec-rodes of the IM-6 impedance analyzer. At the end of each test,DAM chips were rinsed with 0.1% quat and immersed into0% ethanol for 30 min. Finally, they were washed as discussedarlier in Section 2.4.

Mannitol solution with MNAC and without E. coli O157:H7as used as a control for all tests. To observe the effect of MNAC

nd magnetic field on the impedance measurement of bacterialells, the magnitude of impedance was compared for pure bacte-ial cells, and bacterial cells attached to MNAC in the presencend absence of a magnetic field. A calibration curve for nor-alized impedance change (NIC) and concentrations of E. coli157:H7 was drawn based on the difference of magnitude of

mpedance with respect to the control. The value of NIC wasiven by following formula:

IC = Zsample − Zcontrol

Zcontrol× 100 (1)

here Zcontrol is the magnitude of impedance for control sam-le, and Zsample is the magnitude of impedance for a sampleontaining E. coli O157:H7. An average of three readings andheir standard deviation were calculated and analyzed for eachoncentration of bacteria.

.7. Preparation of ground beef samples

Ground beef was purchased from a local supermarket.sample of 25 g of ground beef was homogenized with

25 ml of 0.1% buffered peptone water in a Whirl-pak plas-ic bag using a laboratory Stomacher 400 (Seward, Norfolk,K) for 2 min. After stomaching, the sample was centrifuged

wo times at 250 × g for 15 min in order to separate largeize particles present in the ground beef stomaching water.upernatant of food samples was inoculated with decimallyiluted cultures of E. coli O157:H7 ranging from 8.0 × 101 to.0 × 107 CFU ml−1.

. Results and discussion

.1. Effect of magnetic nanoparticles and magnetic field on

he impedance measurement of E. coli O157:H7

In classical electrochemical impedance spectroscopy, whereadox probes are used, impedance is measured and analyzed

sdEM

n the presence of 0.1 M mannitol solution for frequency range of 10 Hz toMHz.

ith nyquist plot (Ruan et al., 2002; Yang et al., 2004), specifi-ally parameters unrelated to frequency such as charge transferesistance and solution resistance. However, impedance mea-urement based on direct detection of bacterial cells, or per-ormed in the presence of a medium with low conductivity,s best analyzed with bode plot, as it is suitable for studyingirect relationship of impedance with frequency (Gawad et al.,004). Fig. 1 shows the bode plot of IDAM based impedanceiosensor for mannitol solution, MNAC, pure bacterial cells7.4 × 107 CFU ml−1 of E. coli O157:H7), and bacterial cells7.4 × 107 CFU ml−1 of E. coli O157:H7) attached to MNACn the presence and absence of a magnetic field. The magni-ude of impedance at 40 kHz was found to decrease by 15%,6%, 53%, and 71% (with respect to mannitol solution) dueo the presence of MNACs (in the absence of a magneticeld), pure bacterial cells, bacterial cells attached to MNACin the absence of a magnetic field), and bacterial cells attachedo MNAC (in the presence of a magnetic field), respectively.

hen compared with mannitol solution, MNAC showed lowermpedance value due to the presence of highly conductive pro-eins (bovine serum albumin, streptavidin, and antibodies) onhe surface of nanoparticles. Like MNAC, pure bacterial cellslso showed decrease in impedance value. However, MNAC didot interfere with impedance detection of bacterial cells as theecrease in the value of impedance for pure bacterial cells was1% lower than MNAC. Magnetic nanoparticles were used formparting specificity and improving sensitivity of IDAM biosen-or. E. coli O157:H7 cells attached to MNAC were shown to

ecrease magnitude of impedance by 17% as compared to pure. coli O157:H7 cells, because the formation of clusters betweenNAC and bacterial cells resulted in fast settling and concen-

M. Varshney, Y. Li / Biosensors and Bioelectronics 22 (2007) 2408–2414 2411

F oparti(

tcoapAtIpelmfipofiaIO

dotadowacntcatvvmeioc

3bO

apEieslthe use of a simple capacitor (Gawad et al., 2004). Rs accounts forchange in conductivity and charge transport across the bulk solu-tion. Cs represents the stray capacitance of the system associatedwith electrodes, shielding, wiring, connecting cables, and adhe-

Fig. 3. (a) Equivalent circuit for impedance measurement system based onIDAM coupled with MNAC for the detection of E. coli O157:H7. (b) Bode dia-

ig. 2. SEM micrographs of (a) E. coli O157:H7 cells attached to magnetic nan1250×), and (b) a part of the electrode shown in (a) with 5000 magnification.

ration of cells in the active layer of IDAM. Fig. 2 clearly showsluster formation between E. coli O157:H7 cells and MNACn the surface of IDAM. Interdigitated microelectrodes usu-lly result in a sensitive impedance change if bacterial cells areresent in the active layer (Gerwen et al., 1998). Radke andlocilja (2004) calculated the range of active layer with an elec-

rode width and spacing measuring 3 and 4 �m, respectively.DAM best detects impedance change when bacterial cells areresent in the active layer range of 10 �m above the surface oflectrodes. When cells are present out of the range of the activeayer, impedance change is minimized. Impedance measure-

ent of E. coli O157:H7 cells attached to MNAC in a magneticeld decreased the magnitude of impedance by 18% when com-ared to the magnitude of impedance measured in the absencef a magnetic field. This clearly demonstrated that the magneticeld was advantageous in concentrating E. coli O157:H7 in thective layer of IDAM and resulted in enhanced sensitivity ofDAM based impedance biosensor for the detection of E. coli157:H7.It has been shown that E. coli O157:H7 cell acts like a con-

uctor in the presence of mannitol solution. Some componentsf a cell (cell wall and cell cytoplasm) are more conductivehan mannitol solution (Suehiro et al., 2003). The resistance ofn E. coli cell was calculated based on linear parameters, con-uctivities, and the equivalent circuit of different componentsf an E. coli cell (as referred by Suehiro et al., 2003). Thisas then compared with the mannitol solution based on par-

llel and serial connections of cell membrane, cell wall, andytoplasm. The resistance of a cell, calculated with parallel con-ection was found to be three log values lower when comparedo mannitol solution of same volume as a bacterial cell. Thesealculations were in agreement with the results of this studynd others (Suehiro et al., 1999, 2005). Calculations based onhe serial connection showed resistance of a bacterium as 5 logalues higher as compared to the mannitol solution of sameolume as a bacterial cell, which was contrary to the experi-ental results obtained in this research and presented by Suehiro

t al. (2003). Hence it can be concluded that when measuringmpedance, the cell behavior is similar to parallel connectionf cell wall, cell membrane and cytoplasm in an electrical cir-uit.

grEmd

cles–antibody conjugates on the surface of interdigitated array microelectrodes

.2. Equivalent circuit analysis of IDAM based impedanceiosensor coupled with MNAC for the detection of E. coli157:H7

Data of the impedance spectrum can be simulated withn equivalent circuit of the system, which consists of variousarameters leading to impedance change due to the presence of. coli O157:H7 on the surface of IDAM. This equivalent circuit

ncluded two constant phase elements (CPE) one for each set oflectrodes, resistance of medium or bulk resistance (Rs), and atray capacitance (Cs) (Fig. 3a). CPE was used to model doubleayer capacitance of electrodes. This is commonly favored over

rams of impedance spectra of experimental and simulated data in a frequencyange of 10 Hz to 1 MHz. The spectrum was obtained for 7.4 × 107 CFU ml−1 of. coli O157:H7 attached to MNAC in the presence of 0.1 M mannitol solutioneasured in the presence of a magnetic field. Solid lines indicate experimental

ata and crosses indicate simulated data.

2412 M. Varshney, Y. Li / Biosensors and Bioelectronics 22 (2007) 2408–2414

Table 1Simulated valuesa of Rs, CPE, and Cs in the equivalent circuit for control sample and samples with 7.4 × 107 CFU ml−1 of E. coli O157:H7 attached to MNACmeasured in the presence and absence of a magnetic field and their respective percentage change with respect to control sample

Samples Rs (k�) CPE (nF) Cs (pF)

Control 3.11 ± 0.26 a 117.70 ± 1.84 a 222.10 ± 0.30 aSample without magnetic field 2.46 ± 0.19 b 116.50 ± 3.04 a 223.25 ± 0.91 aChange (%) −20.9 −1.0 +0.5Sample with magnetic field 1.52 ± 0.17 c 126.30 ± 1.69 b 222.65 ± 0.35 aC

( nifica

sbEiImeor

apMacr1imtooto(o

ncmd(mtbIcTr

abto(a

3i

Fia

hange (%) −51.1

a–c) Means within a column followed by the different lower case letters are siga Fifty data points were selected for simulation.

ive layer between gold and glass. Fig. 3b shows a representativeode diagram of experimental and fitted impedance spectra of. coli O157:H7 cells (7.4 × 107 CFU ml−1) attached to MNAC

n the presence of mannitol solution. The fitting was done byM-6/THALES software using complex non-linear least-squareethod. The mean modulus of impedance and phase angle

rrors were 0.2% and 0.1◦, respectively, and maximum errorsf modulus of impedance and phase angle were 3.2% and 1.7◦,espectively.

Using the same equivalent circuit, the values of CPE, Rs,nd Cs for the impedance spectra of control sample and sam-les with 7.4 × 107 CFU ml−1 of E. coli O157:H7 attached toNAC measured in the presence and absence of a magnetic field

re summarized in Table 1. The results showed that the bacterialells attached to MNAC in the medium (without magnetic field)esulted in a decrease in the values of Rs and CPE by 20.9% and%, respectively, as compared to control sample. The decreasen values of Rs was due to the increase in conductivity of the

edium caused by the presence of E. coli O157:H7 attachedo MNAC in a mannitol solution. The value of Rs comprisedf two components—resistance of the medium and resistancef bacterial cells attached to MNAC. The value of Rs for resis-

ance of the medium for the sample with 7.4 × 107 CFU ml−1

f E. coli O157:H7 cells was found to decrease by less than 7%out of a total decrease of 21%) as compared to the resistancef the medium of the control sample with MNAC only (data

(t

ig. 4. Normalized impedance change (NIC) at 40 kHz for the samples with a range on pure culture, and (b) 8.0 × 101 to 8.0 × 107 CFU ml−1 in ground beef with respect tre the regression lines. A positive value of NIC indicates an increase, whereas a neg

+7.1 +0.2

ntly different (P < 0.05).

ot shown). This decrease was caused by the leakage of moreonductive ions from inside the bacterial cells into the outeredium. Thus the major change in the value of Rs (>14%) was

ue to the presence of E. coli O157:H7 cells only. Suehiro et al.2005) showed an improved sensitivity of impedance measure-ent of E. coli by decreasing detection limit from 104 CFU ml−1

o 102 CFU ml−1 by releasing conductive ions from inside theacterial cells into the outer medium using electro-permeation.t has already been shown in Section 3.1 that E. coli O157:H7ell behaves as a conductor in the presence of mannitol solution.hus decrease in the value of Rs was due to the combination of

esistance of medium and the resistance of bacterial cells.The concentration of cells in the active layer in the presence of

magnetic field, resulted in a further decrease in the value of Rsy 30% and an increase in the value of CPE by 6% as compared tohe samples of E. coli O157:H7 attached to MNAC in the absencef a magnetic field. The values of Cs did not change significantlyP > 0.05) from test to test because connecting wires, IDAM chipnd all other parameters in the system remained constant.

.3. Effect of E. coli O157:H7 present in ground beef onmpedance measurement

Fig. 4a and b shows the NIC values (calculation based on Eq.1) at 40 kHz for E. coli O157:H7 of concentrations 7.4 × 101

o 7.4 × 107 CFU ml−1 and 8.0 × 101 to 8.0 × 107 CFU ml−1

f E. coli O157:H7 concentrations (N) from (a) 7.4 × 101 to 7.4 × 107 CFU ml−1

o control sample. Solid lines indicate the experimental data, while dashed linesative value indicates a decrease in the magnitude of medium impedance.

d Bio

ptlp8tOtno(cE8bvpNT8wocCaomi

mpcttflipbf3Seaat1s

gaptaElm

eia

4

pfbssitb8pOdpolitmmMboit

A

s

R

B

C

DF

FFG

G

K

M. Varshney, Y. Li / Biosensors an

resent in pure culture and in ground beef samples, respec-ively. Results showed that for impedance measurement, theower detection limits for detection of E. coli O157:H7 inure culture and in ground beef samples were 7.4 × 104 and.0 × 105 CFU ml−1, respectively. The difference in detec-ion limits was caused by low CE of MNCs for E. coli157:H7 in ground beef samples due to the presence of pro-

ein, fat, and other components of food matrix. In IMS tech-iques, the food matrix is found to interfere with the capturef bacteria by MNAC and thus results in lower CE valuesVarshney et al., 2005). IDAM based impedance biosensoroupled with MNAC was successfully used for detection of. coli O157:H7 from 7.4 × 104 to 7.4 × 107 CFU ml−1 and.0 × 105 to 8.0 × 107 CFU ml−1 in pure culture and in groundeef samples, respectively. The regression equation for NICersus cell numbers from 7.4 × 104 to 7.4 × 107 CFU ml−1 inure culture was NIC = 12.62N − 44.71 with R2 = 0.99, where

is the concentration of E. coli O157:H7 in log CFU ml−1.he similar regression equation for the presence of 8.0 × 105 to.0 × 107 CFU ml−1 of E. coli O157:H7 in ground beef samplesas NIC = 15.55N − 71.04 with R2 = 0.95. Total viable countf bacteria in ground beef samples was 102 CFU g−1 and no E.oli was detected based on surface plate growth on SMAC agar.ombination of IMS, impedance measurement and influence ofmagnetic field for the concentration of cells in the active layerf IDAM showed promising results for the detection of a mini-um of 7.4 × 104 and 8.0 × 105 CFU ml−1 of E. coli O157:H7

n pure culture and ground beef samples, respectively.The detection limit of the present label-free impedance

ethod based on IDAM and MNAC for detection of food-borneathogens is comparable with other conventional electrochemi-al methods. However, in comparison to conventional methods,his method has the following distinctive features—capabilityo handle small volume, ease of surface regeneration, no sur-ace immobilization and enrichment growth, and potential forab-on-a-chip design. Radke and Alocilja (2005) developed anmpedance biosensor for the direct detection (without any radoxrobe) of a range of E. coli O157:H7 from 104 to 107 CFU ml−1

ased on anti-E. coli O157 antibodies immobilized on the sur-ace of IDAM. The volume of the sample used for detection was0 ml as compared to 2 �l in the present impedance biosensor.usmel et al. (2003) demonstrated a classical label-free three-lectrode electrochemical system employing radox probe andntibody immobilization, for the detection of bacterial cells suchs Listeria monocytogenes and Bacillus cereus. Immunoreactionime and detection volume were 60 min and 3 ml as compared to5 min and 2 �l, respectively, in the present impedance biosen-or.

When compared to classical methods involving enrichmentrowth, selective concentration using immunomagnetic beadsnd detection of E. coli O157:H7 by surface plating method, theresent impedance biosensing methods is rapid and detects bac-erial cells in 35 min and has a high detection limit. Fratamico et

l. (1992) selectively recovered an initial level of 10 CFU ml−1

. coli O157:H7 in meat products after 24 h enrichment time fol-owed by detection using surface plating method. The impedance

ethod presented in this paper can also be used for differ-

L

L

electronics 22 (2007) 2408–2414 2413

ntiating between live and dead bacterial cells by measuringmpedance change in the conductivity of the medium beforend after an enrichment growth time (1–4 h).

. Conclusions

In this study an impedance biosensor based on IDAM cou-led with MNAC was successfully developed and evaluatedor rapid and specific detection of E. coli O157:H7 in groundeef samples. The use of MNAC provided an efficient andpecific way to separate E. coli O157:H7 from ground beefamples and to concentrate bacterial cells attached to MNACn the active layer of IDAM with the help of a magnetic field,hereby enhancing the sensitivity of IDAM based impedanceiosensor. Impedance measurement was used to detect as low as.0 × 105 CFU ml−1 of E. coli O157:H7 in ground beef sam-les. The lowest detection limit of the biosensor for E. coli157:H7 in pure culture was 7.4 × 104 CFU ml−1, with a totaletection time of 35 min from sampling to measurement. Theresent biosensing method may be easily adopted for detectionf other bacterial pathogens by substituting primary biotiny-ated antibodies. This research focused on the design of anmpedance method without any surface modification of the elec-rodes and pre-enrichment of the sample for cell growth. A

icrochannel with embedded IDAM may be used for auto-ated bio-detection techniques. Additionally, the collection ofNAC attached cells in a region between the electrode fingers

y dielectrophoresis technique may improve the performancef an impedance biosensing method. However, in either casempedance measurement may be different from that observed inhis study.

cknowledgments

This project was supported in part by the Food Safety Con-ortium and Arkansas Experimental Station.

eferences

rewster, J.D., Gehring, A.G., Mazenko, R.S., Van Houten, L.J., Crawford, C.J.,1996. Anal. Chem. 68 (23), 4153–4159.

enters for Disease Control and Prevention (CDC), 2005. Disease Information-E. coli O157:H7. Available at: www.cdc.gov. Accessed December 2005.

eisingh, A.K., Thompson, M., 2004. J. Appl. Microbiol. 96, 419–429.ratamico, P.M., Schultz, F.J., Buchanan, R.L., 1992. Food Microbiol. 9,

105–113.ritzsche, W., Taton, T.A., 2003. Nanotechnology 14, R63–R73.ung, Y.S., Wong, Y.Y., 2001. Anal. Chem. 73, 5302–5309.awad, S., Xheung, K., Seger, U., Bertsch, A., Renaud, P., 2004. Lab Chip 4,

241–251.erwen, P.V., Laureyn, W., Laureys, W., Huyberechts, G., Beeks, M., Baert,

K., Suls, J., Sansen, W., Jacobs, P., Hermans, L., Mertens, R., 1998. Sens.Actuators B 49, 73–80.

oubova, V., Brynda, E., Karasova, L., Skvor, J., Homola, J., Dostalek, J.,Tobiska, P., Rosciky, J., 2001. Sens. Actuators B 74, 100–105.

ekowska-Kochaniak, A., Czajkowska, D., Popowski, J., 2002. Acta Microbiol.Pol. 51, 327–337.

iberti, P.A., Chiarappa, J.N., Hovsepian, A.C., Rao, C.G., 1996. Fine Parti-cles Science and Technology. Kluwer Academic Publishers, Dordrecht, TheNetherlands, 777–790.

2 d Bio

M

RRRS

S

S

S

414 M. Varshney, Y. Li / Biosensors an

ao, X., Yang, L., Su, X., Li, Y., 2006. Biosens. Bioelectron. 21 (7), 1178–1185.

adke, S.M., Alocilja, E.C., 2004. IEEE Sens. J. 4 (4), 434–440.adke, S.M., Alocilja, E.C., 2005. Biosens. Bioelectron. 20 (8), 1662–1667.

uan, C., Yang, L., Li, Y., 2002. Anal. Chem. 74, 4814–4820.toimenov, P.K., Klinger, R.L., Marchin, G.L., Klabunde, K.J., 2002. Langmuir

18, 6679–6686.uehiro, J., Yatsunami, R., Hamada, R., Hara, M., 1999. J. Appl. Phys. 32,

2814–2820.

S

VYY

electronics 22 (2007) 2408–2414

uehiro, J., Hamda, R., Noutomi, D., Shutou, M., Hara, M., 2003. J. Electrostat.57, 157–168.

uehiro, J., Hatano, T., Shutou, M., Hara, M., 2005. Sens. Actuators B 109,209–215.

usmel, S., Guibault, G.G., O’Sullivan, C.K., 2003. Biosens. Bioelectron. 18,881–889.

arshney, M., Yang, L., Su, X., Li, Y., 2005. J. Food Prot. 68 (9), 1804–1811.ang, L., Li, Y., Erf, G.F., 2004. Anal. Chem. 76, 1107–1113.u, L.S.L., Uknalis, J., Tu, S., 2001. J. Immunol. Methods 256, 11–18.