Molecular Genetics and Metabolism 84 (2005) 90–99

www.elsevier.com/locate/ymgme

Rapid, species-speciWc detection of uropathogen 16S rDNAand rRNA at ambient temperature by dot-blot hybridization

and an electrochemical sensor array

Chien-Pin Suna,1, Joseph C. Liaob,1, Yao-Hua Zhangc, Vincent Gaud,Mitra Mastalie,f, Jane T. Babbitte,f, Warren S. Grundfesta, Bernard M. Churchillb,

Edward R.B. McCabec,g, David A. Haakee,f,¤

a Biomedical Engineering Department, Henry Samueli School of Engineering and Applied Science, UCLA, Los Angeles, CA, USAb Department of Urology, David GeVen School of Medicine at UCLA, Los Angeles, CA, USA

c Department of Pediatrics, David GeVen School of Medicine at UCLA, Los Angeles, CA, USAd GeneFluidics Inc., Monterey Park, CA, USA

e Division of Infectious Diseases, Veterans AVairs Greater Los Angeles Healthcare System, Los Angeles, CA, USAf Department of Medicine, David GeVen School of Medicine at UCLA, Los Angeles, CA, USA

g Department of Human Genetics, David GeVen School of Medicine at UCLA, Los Angeles, CA, USA

Received 10 November 2004; accepted 10 November 2004Available online 13 December 2004

Abstract

Development of rapid molecular approaches for pathogen detection is key to improving treatment of infectious diseases. For thisstudy, the kinetics and temperature-dependence of DNA probe hybridization to uropathogen species-speciWc sequences wereexamined. A set of oligonucleotide probes were designed based on variable regions of the 16S gene of the Escherichia coli, Proteusmirabilis, Klebsiella oxytoca, and Pseudomonas aeruginosa. A universal bacterial probe and probes-speciWc for gram-positive andgram-negative organisms were also included. The oligonucleotide probes discriminated among 16S genes derived from 11 diVerentspecies of uropathogenic bacteria applied to nylon membranes in a dot-blot format. SigniWcant binding of oligonucleotide probes totarget DNA and removal of nonspeciWc binding by membrane washing could both be achieved rapidly, requiring as little as 10 min.An oligonucleotide probe from the same species-speciWc region of the E. coli 16S gene was used as a capture probe in a novel electro-chemical 16-sensor array based on microfabrication technology. Sequence-speciWc hybridization of target uropathogen 16S rDNAwas detected through horseradish peroxidase acting as an electrochemical transducer via a second, detector probe. The sensor arraydemonstrated rapid, species-speciWc hybridization in a time course consistent with the rapid kinetics of the dot-blot hybridizationstudies. As in the dot-blot hybridization studies, species-speciWc detection of bacterial nucleic acids using the sensor array approachwas demonstrated both at 65 °C and at room temperature. These results demonstrate that molecular hybridization approaches canbe adapted to rapid, room temperature conditions ideal for an electrochemical sensor array platform.Published by Elsevier Inc.

Keywords: Bacteria; Diagnosis; Molecular; DNA probes; Microbiology; Molecular microbiology; Nucleic acid hybridization/methods; Oligonucleo-tide probes; RNA, Ribosomal; 16S; Urinary tract infections; Molecular diagnosis

1096-7192/$ - see front matter. Published by Elsevier Inc.doi:10.1016/j.ymgme.2004.11.006

* Corresponding author. Fax: +1 310 268 4928.E-mail address: dhaake@ucla.edu (D.A. Haake).

1 These authors contributed equally to the work described in this article.

C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99 91

Introduction

Prompt and accurate detection of microbial patho-gens is essential for improving the management of infec-tious diseases. Current clinical microbiology laboratorypractices continue to rely on cultivation of samples inartiWcial media for isolation and phenotypic identiWca-tion of etiologic agents [1,2]. In the case of urinary tractinfections, the delay between specimen submission anddiagnosis results in empiric and frequently inappropriateantimicrobial therapy [3].

The application of molecular approaches to clinicaldiagnostics has a number of advantages over standardmicrobiological techniques including sensitivity, speed,and ease of specimen processing. The ability of geneticprobes to detect infectious agents that would otherwisebe unculturable is well known [4–7]. In terms of speed,DNA-based sensors potentially reduce the time to diag-nosis by coupling probe–target hybridization to a mea-surable electronic signal [8–10]. Direct detection ofpathogen–speciWc target sequences by complementaryoligonucleotide probes requires less technically demand-ing sample preparation methods than enzymatic ampliW-cation methods that are sensitive to inhibition byspecimen components [11]. Despite these advantages,there has been a lag in the application of DNA hybrid-ization techniques to clinical diagnostic problems.

DNA hybridization techniques utilized in theresearch laboratory are not easily translated to the clini-cal setting. Southern hybridization and dot-blot hybrid-ization techniques involve detection of target DNAmolecules immobilized on nitrocellulose or nylon mem-branes. Nucleic acid probes are typically incubated withthe membrane for 6–24 h, followed by washing toremove nonspeciWcally bound probes [12–16]. Hybrid-ization and washing requires precise temperature controlto prevent nonspeciWc hybridization at low temperaturesand loss of correctly hybridized probes at high tempera-tures. X-ray Wlm is then exposed to blots to detecthybridization of labeled probes to target DNA mole-cules.

The extended hybridization time and rigid tempera-ture requirements have limited the use of direct hybrid-ization methods for clinical diagnostic purposes. For thisreason, there is a need to examine the speed andeYciency of DNA/DNA hybridization for species-spe-ciWc identiWcation of microbial pathogens. In this study,we examined the kinetics and temperature-dependenceof oligonucleotide probes for the development of a rapiddetection and identiWcation system for urinary patho-gens. We Wrst studied the hybridization of species-spe-ciWc DNA probes to PCR-ampliWed uropathogen 16SrDNA using a dot-blot format. We then tested whetherrapid, room temperature oligonucleotide probe hybrid-ization approaches could be translated to the detectionof uropathogen 16S rRNA using a novel electrochemical

DNA biosensor array based on microfabrication tech-nology that we described previously [17].

Materials and methods

Bacterial strains and media

Clinical isolates of the following uropathogens wereobtained from the UCLA Clinical Microbiology Labo-ratory with approval of the UCLA Medical InstitutionalReview Board: Escherichia coli, Klebsiella pneumoniae,Klebsiella oxytoca, Proteus mirabilis, Pseudomonas aeru-ginosa, Enterobacter cloacae, Enterobacter aerogenes,Citrobacter freundii, Providencia stuartii, Enterococcusfaecalis, and Staphylococcus saprophyticus. Specimenswere received in vials containing tryptic soy broth (TSB)with glycerol and hypertonic additives, and were storedat ¡70 °C.

PCR ampliWcation

Template DNA was obtained from gram-negativebacteria by incubating organisms in 100 �l water at100 °C for 10 min. Template DNA was obtained fromgram-positive bacteria by incubating organisms in 100�lof 25 mM NaOH, 0.25% sodium dodecyl sulfate (SDS)at 100 °C for 10 min. PCR ampliWcation was performedusing the highly conserved 16S rDNA primer pair,p8FPL and p806R [18]. 16S rRNA genes were ampliWedusing Taq polymerase (8 U, Perkin–Elmer, Norwalk,CT) as follows: initial denaturation at 94 °C for 30 s, fol-lowed by 35 cycles of denaturation at 94 °C for 30 s,annealing at 55 °C for 30 s, and extension at 72 °C for1 min, and completed with a Wnal extension at 72 °C for2 min.

Design of oligonucleotide probes

The species-speciWc probes were designed by examin-ing 16S rDNA sequences obtained from the NCBI data-base (Bethesda, MD). Sequence data for the type strainsof the following species were aligned by the ClustalWmethod (GenBank accession numbers in parentheses):E. coli (X80725), C. freundii (AJ233408), E. aerogenes(AB004750), E. cloacae (AJ251469), E. faecalis(AB012212), K. oxytoca (Y17655), K. pneumoniae(AB004753), P. mirabilis (AF008582), P. stuarti(AF008581), P. aeruginosa (Z76651), and S. saprophyti-cus (L37596). Seven-speciWc oligonucleotide probes weredesigned based on the 16S DNA sequences of the uro-pathogenic bacteria (Table 1 and Fig. 1). A universalprobe was designed to hybridize with all bacteria. Gram-negative and gram-positive probes were designed to dis-tinguish sequences from organisms diVering accordingto their gram stain characteristics. Four species-speciWc

92 C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99

probes were designed to identify the respective uropath-ogens E. coli, P. mirabilis, P. aeruginosa, and K. oxytoca.Regions 20–26 bp in length with at least four sequencemismatches between the sequence of the species of inter-est and the sequences of the other bacterial species wereselected as candidate species-speciWc probes.

Dot-blot hybridization technique

Probes were labeled at the 5� end with [�-32P]ATPusing T4 kinase at a temperature of 37 °C for 45 min.The dot-blot mixtures were prepared with 40 �l PCRampliWcation products, 51 �l deionized water (MilliporeMilli-Q, Bedford, MA), 4 �l of 10 M NaOH, and 5 �l of0.5 M EDTA. Each mixture was divided into two 50 �lportions and Wltered through 0.45 �m nylon Wlters (Pall,Ann Arbor, MI) in a Bio-dot dot-blot manifold (96wells, Bio-Rad, Hercules, CA). The target DNA wasWxed to the membranes using a UV crosslinker (Strata-gene, La Jolla, CA) and dried in an 80 °C oven for 1 h.Dot-blot hybridization was performed as previouslydescribed [19]. BrieXy, the dot-blots were Wrst prehybrid-ized with 3 ml Church buVer (0.5 M Na2HPO4, 7% SDS,and 0.001 M EDTA) at 55 °C for 1 h. The membraneswere then hybridized with 2 ml Church buVer, 50 �lSalmon sperm DNA (Gibco-BRL, Grand Island, NY),and 30–50 �l of radioactive oligonucleotide probe at55 °C overnight. The amount of probe in the hybridiza-tion mixture was varied to adjust for diVerence in radio-activity of the probes determined using a LiquidScintillation Analyzer (1500 Tri-CARB, Packard Instru-ment, Downers Grove, IL). The blots were then washedat 55 °C in Church wash buVer (0.04 M Na2HPO4, 1%SDS) until background radiation was no longer detect-able with the Geiger–Muller counter. For the rapid,

ambient or “room temperature” approach, the dot-blotswere prehybridized with Church buVer at 25 °C for10 min, hybridized with the hybridization mixture at25 °C for 10 min, and washed with Church wash solutionat 25 °C for 10 min, with 25 °C chosen as a temperaturethat could be maintained constantly in the laboratoryand that approximated room temperature. The dot-blotswere then autoradiographed at ¡80 °C for 12–24 h usingKodak X-OMAT AR Wlm (Rochester, NY).

Electrochemical sensor array

Disposable electrochemical sensor arrays pretreatedwith an alkanethiolate self-assembled monolayer (SAM)were obtained (GeneFluidics, Monterey Park, CA, http://www.geneXuidics.com/). The basic design of the 16-sen-sor array was as previously described [17] with theexception that the gold sensors were batch fabricated onplastic instead of silicon. Each sensor in the array con-tained three electrodes: a central working electrode, areference electrode, and an auxiliary electrode.

Preparation of functional layers on the electrochemical sensor

Using a 16-channel multi-potentiostat (GeneFlui-dics), cyclic voltammetry (CV) was Wrst applied to char-acterize the integrity of the alkanethiolate SAM on thesensor surface with 50 �l of 0.1 mM K3Fe(CN)6 (Sigma,St. Louis, MO) as previously described [17]. After satis-factory CV characterization, the sensor was washed anddried. All washing steps were carried out with a streamof deionized (DI) H2O to the front side of the sensor forapproximately 2–3 s followed by 5 s drying under astream of nitrogen. Streptavidin (Calbiochem) was next

Table 1Oligonucleotide probes used in dot-blot hybridization and electrochemical sensor array studies

a Probe designations are based on the Wrst letters of the genus and species, the length of the probe, and the E. coli 16S rRNA gene sequence posi-tion of the Wrst base in the probe sequence. Uni, Universal; GmNeg, gram-negative; GmPos, gram-positive; Ec, E. coli; Pm, P. mirabilis; Ps, P. aeru-ginosa; Ko, K. oxytoca.

b The Wrst seven probes use positive strand sequences for use in the dot-blot hybridization experiments. The last two probes use negative strandsequences for hybridization to 16S rRNA.

c Positions based on the E. coli 16S rRNA gene sequence.d Melting temperature (Tm) was calculated using the Wallace rule: Tm D 2 (A + T) + 4 (G + C).

Probesa Probe sequence (5�–3�)b Positionsc Probe length (bp)

Chemical modiWcation

Melting tempd (°C)

Uni21-684 TGTAGCGGTG AAATGCGTAG A 684–704 21 None 62GmNeg24-358 TGGGGAATAT TGCACAATGG GCGC 358–381 24 None 74GmPos20-358 TAGGGAATCT TCGGCAATGG 358–377 20 None 60Ec20-465 AATACCTTTG CTCATTGACG 465–484 20 None 56Pm26-184 GTCTACGGAC CAAAGCAGGG GCTCTT 184–209 26 None 82Pa20-179 ATACGTCCTG AGGGAGAAAG 179–198 20 None 60Ko-455 GAGTGAGGTT AATAACCTCA TTCAT 455–479 25 None 68Ec35C-449 capture probe GTCAATGAGC AAAGGTATTA

ACTTTACTCC CTTCC483–449 35 5�-Biotin —

Ec35D-408 detector probe CTGAAAGTAC TTTACAACCC GAAGGCCTTC TTCAT

442–408 35 5�-Fluorescein —

C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99 93

added (2.5�l, 0.5 U/ml in phosphate-buVered saline),incubated for 10 min and washed oV. Biotinylated cap-ture probes (2.5 �l, 1 �M in GeneFluidics’ Probe Diluent)were then added to the individual sensors. After 30 minincubation, the sensor array was washed and dried, com-pleting the surface preparation.

Detection of bacterial rRNA

Uropathogenic bacterial strains were grown to logphase and the concentration determined by plating onLB agar. Bacteria were pelleted and resuspended in 90�l

of water at a concentration of approximately2 £ 109 bacteria/ml. In some experiments, 90 �l samplesof urine were tested containing uropathogenic E. coligrown to a concentration of approximately 1 £108 bacteria/ml. Ten microliters of 0.4 M NaOH wereadded to the bacterial suspension followed by incuba-tion at room temperature for 5 min. Fifty microliters ofneutralization solution containing the detector probe(0.5�M) in GeneFluidics’ Probe Diluent, containing2.5% bovine serum albumin (Sigma) was added to thebacterial lysate. After 10 min of hybridization at thedesired temperature, 4 �l of the bacterial lysate/detector

Fig. 1. Sequence alignment of bacterial 16S rDNA sequences. (A) The region of the P. aeruginosa (yellow) and P. mirabilis (blue) probes. (B) Theregion of the gram-negative (yellow) and gram-positive (blue) probes. The Wrst line of (C) shows the region of the E. coli detector (left) and capture(right) probes in yellow. (C) The region of the E. coli (yellow) and K. oxytoca (blue) dot-blot probes. (D) The region of the universal bacterial probe(yellow). Nucleotides diVering from the consensus sequence are shaded. Helix locations of probes are given according to the secondary structuremodel for 16S rRNA [23]. Nucleotide positions are based on the E. coli 16S rDNA gene sequence. Two letter codes for bacteria are as follows: Cf, C.freundii; Ea, E. aerogenes; Ec, E. coli; El, E. cloacae; Ef, E. faecalis; Ko, K. oxytoca; Kp, K. pneumoniae; Pa, P. aeruginosa; Pm, P. mirabilis; Ps, P. stu-artii; and Ss, S. saprophyticus.

94 C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99

probe mixture were deposited on each of the workingelectrodes in the sensor array. The sensor array was incu-bated for 10 min at the desired temperature. After wash-ing and drying, 2.5 �l of 0.5 U/ml anti-Xuoresceinperoxidase Fab fragments (Roche, diluted in 0.5% caseinin 100 mM sodium phosphate buVer, pH 7.4) weredeposited on each of the working electrodes for 10 min.After washing and drying, a prefabricated plastic wellmanifold (GeneFluidics) was bonded to the sensor arrayto prevent cross-contamination of the substrate solution.Fifty microliters of horseradish peroxidase substrate(GeneFluidics) were placed on each of the sensors in thearray so as to cover all three of the electrodes of eachsensor. Electrochemical measurements were immediatelyand simultaneously taken for all 16 sensors. The entireassay protocol was completed within 40 min from thetime when bacterial lysis was commenced. Current vs.time was measured using a multichannel potentiostat(GeneFluidics). The voltage was Wxed at ¡100 mV (vs.reference), and the electroreduction current was mea-sured at 60 s after reaching steady state. All measure-ments were carried out in duplicate.

Results

Dot-blot hybridization

Alignment of uropathogen 16S sequences revealedconserved regions where the DNA sequences were iden-tical among all species, and divergent regions where thesequences diVer between each individual species orgroups of species (Figs. 1A–D). A conserved region wasused to design the universal probe, while divergentregions were used to design the gram-negative, gram-positive, and species-speciWc probes. The universal PCRprimer pair, p8FPL and p806R, successfully ampliWedthe 16S ribosomal DNA of all 11 bacterial specimens(data not shown). PCR ampliWcation products approxi-mately 800 base pairs in length were ampliWed andapplied to membranes as target rDNA. In the initialphase of the experiment, the E. coli and P. aeruginosaoligonucleotide probes demonstrated species-speciWchybridization. However, the K. oxytoca and P. mirabilisprobes hybridized both with their PCR products fromtheir target species and those of other species (data notshown). The Klebsiella probe showed positive resultswith C. freundii and E. aerugenes, while the Proteusprobe hybridized with C. freundii, K. oxytoca, and P. stu-artii. The lack of speciWcity was attributed to the factthat the percentage of mismatches between each probeand the sequences of the false positives was less than24% of the probe’s length. The Proteus and Klebsiellaprobes were redesigned using alternative divergentregions of the 16S rDNA sequence that provided higherpercentage of mismatches between the probes and their

nontargets for better species speciWcity (Figs. 1A and C,respectively).

Standard and rapid dot-blot hybridization assays

Dot-blot hybridization was performed using standard1 h prehybridization and overnight hybridization condi-tions at 55 °C to determine results obtainable under opti-mal conditions. The oligonucleotide probes showedexcellent speciWcity and had positive hybridizations withtheir corresponding DNA targets (data not shown). Theuniversal probe chosen as a positive control, correctlylabeled all 11 PCR targets. The gram-positive probe suc-cessfully hybridized with the targets from organisms E.faecalis and S. saprophyticus, and the gram-negativeprobe with the targets from all gram-negative organisms.The redesigned set of species-speciWc probes correctlyidentiWed their target species namely, E. coli, P. mirabilis,P. aeruginosa, and K. oxytoca. The dot-blots werewashed at 55 °C for 30–60 min until background signalswere no longer detected (Fig. 2). To study the eVects oftime and temperature on the dot-blot results, the prehy-bridization, hybridization, and washing times were eachreduced to 10 min, and all steps were performed at 25 °C.All seven probes retained sensitivity and speciWcity fortheir intended targets (Fig. 3).

Species-speciWc detection of bacterial rRNA with an electrochemical sensor array

A pair of E. coli-speciWc 35 bp capture and detectorprobes, Ec35C-449 and Ec35D-408 (Table 1) was used todetect bacterial rRNA in the electrochemical sensorarray. The capture probe was derived from the sameregion of the 16S rRNA sequence as the E. coli-speciWcprobe used in the dot-blot experiments. Fig. 4 shows theresults using a hybridization temperature of 65 °C. Asexpected, the highest current output was generated usingE. coli nucleic acids. Current output for nucleic acidsfrom other uropathogenic bacteria, including P. mirabi-lis, K. pneumoniae, K. oxytoca, E. aerogenes, and E. cloa-cae was signiWcantly lower. Most of these uropathogensyielded a current output in the same range as the nega-tive control that contained no bacterial nucleic acids.Similar results were obtained in several independentexperiments (data not shown).

Kinetics and temperature-dependence of E. coli nucleic acid detection using the electrochemical sensor array

Consistent with the kinetics found for dot-blothybridization, E. coli rRNA was detected using the elec-trochemical sensor with a hybridization time as short as10 min. This was consistent with the kinetics found fordot-blot hybridization. We were also interested inwhether species-speciWc detection of E. coli nucleic acids

C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99 95

could be achieved at room temperature as in the dot-blothybridization experiments. Fig. 5 shows the results usingthe E. coli 35 bp 16S capture and detector probes todetect rRNA of E. coli, K. oxytoca, and K. pneumoniae at65 °C and at room temperature. The signal for E. coliwas 5-fold higher at 65 °C hybridization than at roomtemperature, possibly indicating an eVect of higherhybridization stringency with elevated temperature. Fig.5 also demonstrated that at room temperature hybrid-ization, the E. coli signal retained speciWcity and was stillsigniWcantly higher than K. oxytoca and K. pneumoniae.In separate experiments using the E. coli sensor probepair, rRNAs of other uropathogens including P. mira-blis, E. aerogenes, E. cloacae, P. aeruginosa, and C. freun-dii yielded signals comparable to background noise of690 nA (data not shown). The diVerences in signalsbetween E. coli and other uropathogens were statisticallysigniWcant based on Student’s t test for unpairedsamples.

Discussion

Our goal is to develop rapid molecular diagnosticapproaches to the identiWcation of uropathogensinvolving techniques that can be applied in a variety ofsettings and platforms, including a point-of-care micro-fabricated device. To achieve this goal, the eVects of timeand temperature on the speciWcity of oligonucleotideprobe binding to uropathogen 16S rDNA sequenceswere examined using both a rapid dot-blot hybridizationmethod and a novel electrochemical sensor array plat-form. The rapid dot-blot hybridization methoddescribed here involves 10 min each for the prehybridiza-tion, hybridization, and washing steps, or a total of only30 min for all three steps. In the standard dot-blotmethod, hybridization typically involves an overnightincubation. Our rapid approach demonstrates that ifrRNA is used rather than PCR-generated rDNA, then amuch shorter analytical time would be suYcient to

Fig. 2. Hybridization results using standard conditions. Dot-blots using overnight hybridization and washing at 55 °C resulted in speciWc hybridiza-tion of the gram-negative, gram-positive, and species-speciWc probes to PCR ampliWed 16S rDNA genes.

96 C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99

generate a robust signal, allowing an approximately 24-to 36-fold reduction in procedural time (from approxi-mately 12–18 h to 40 min).

The success of the rapid dot-blot method is criticallydependent on probe characteristics. Oligonucleotideprobes were designed by comparing the 16S genesequences of eleven uropathogenic species to locatevariable regions containing signature sequences for foururopathogens: E. coli, P. mirabilis, P. aeruginosa, and K.oxytoca. Probes were selected that contained at least Wvemismatches per 20 bp region in each of the nontargetsequences. Probes with fewer than Wve mismatches per20 bp region were not considered robust, since theyyielded false-positive results and therefore they were re-designed (data not shown). Probes with mismatcheslocated in the center of the sequence had greater speciWc-ity than probes with mismatches concentrated at theends. This degree of sequence mismatching was suYcientto obtain target speciWcity not only at 55 °C but also in

the much less stringent condition of room temperaturehybridization.

The dot-blot experiments were informative regardingprobe design, kinetics, and temperature-dependence ofhybridization using a well-established technique. Weexamined whether the results of the rapid dot-blothybridization studies could be applied to a novel electro-chemical sensor array platform. There is enormous inter-est in development of DNA biosensors that are able todetect nucleic acid hybridization events without the needfor target ampliWcation [8,20]. An electrochemical sensorprovides instantaneous and quantitative readout ofnucleic acid hybridization, which oVers tremendousadvantages for clinical microbiology and genetic diag-nostics. The sensor strategy combines approaches fromdiverse disciplines including microfabrication technol-ogy, self-assembled monolayer (SAM) generation,DNA/RNA hybridization, and enzyme signal ampliWca-tion. SAMs allow immobilization of biological mole-

Fig. 3. Rapid, room-temperature hybridization. These dot-blots show that speciWcity was retained using rapid hybridization and washing at roomtemperature. The universal bacterial, gram-negative, gram-positive, and species-speciWc probes were allowed to hybridize to PCR ampliWed 16SrDNA genes for 10 min followed by washing for 10 min. Both the hybridization and washing steps were performed at room temperature.

C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99 97

cules such as streptavidin onto microfabricated devicesto provide surface functionalization.

In this study, the E. coli-speciWc capture probe wasbiotinylated to allow it to be anchored onto the working

electrode surface that contained a monolayer of strepta-vidin. Detection of E. coli DNA/RNA occurred when amixture containing the bacterial lysate and the E. coli-speciWc Xuorescein-conjugated detector probe is

Fig. 4. SpeciWcity of the E. coli 35 bp probe. Current output in nanoamperes obtained using the GeneFluidics sensor with the E. coli-speciWc captureand detector probe pair. Bacteria tested include E. coli (Ec), P. mirabilis (Pm), K. oxytoca (Ko), P. aeruginosa (Pa), K. pneumoniae (Kp), and E. cloa-cae (El). Negative control (NC) readings were from a sample without bacteria. Standard deviations are shown by the error bars. Readings for E. coliwere signiWcantly greater than for the Wve other species of uropathogenic bacteria tested (P < 0.0001).

Fig. 5. EVect of hybridization temperature on electrochemical detection of uropathogens. Current output in nanoamperes obtained using the Gene-Fluidics sensor with the E. coli-speciWc capture and detector probe pair at 65 °C and room temperature (RT). Standard deviations are shown by theerror bars. Readings for E. coli were signiWcantly greater than for K. oxytoca and K. pneumoniae at both 65 °C (P < 0.001) and RT (P < 0.01). Otheruropathogens including P. mirablis, E. aerogenes, P. aeruginosa, and C. freundii were tested at room temperature and yielded signals comparable tobackground noise of 690 nA (data not shown).

98 C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99

deposited onto the sensor surface. Hybridization wasdetected using a peroxidase-conjugated anti-Xuoresceinantibody, which ampliWed the hybridization signalthrough catalysis of peroxidase substrate. The chemicalsignal was converted to an electronic signal measured bya multichannel potentiostat. Amperometric measure-ment was taken at a Wxed potential between the workingand reference electrodes after the reaction had reachedsteady state. The strength of the current output reXectedthe number of stable hybrids on the sensor surface. Ourdetection strategy is a molecular analogue of theenzyme-linked immunosorbent assay (ELISA) withimportant diVerences: electrochemical measurement,rather than optical measurement, is utilized, and the ana-lyte is a DNA/RNA species, rather than protein.

Using a capture probe derived from the E. coli probedesigned for the dot-blot hybridization experiments, theelectrochemical sensor array demonstrated species-spe-ciWc detection of E. coli (Fig. 4). Twenty base pair probeswere used in the dot-blot platform and 35 bp captureand detector probe pairs in the electrochemical sensorarray platform. The sequence used for the E. coli dot-blot probe corresponds to the 5� end of the E. coli cap-ture probe used in the sensor array. Attempts to use thedot-blot E. coli 20 bp probe in the sensor array were notsuccessful (data not shown) whereas the 35 bp probe wassuccessful. The need for a longer probe in the sensorarray platform is likely due to inherent diVerences in oli-gonucleotide probe hybridization to 16S rRNA [21]. ThespeciWcity of the capture and detector probe pair for E.coli 16S was retained at room temperature, as indicatedby high signal-to-noise ratio (Fig. 5). Background noisewas consistently <50 nA in samples where no targetswere present, and <100 nA in the majority of samplescontaining nonspeciWc bacteria. The low backgroundmeasurements indicate that nonspeciWc binding of theperoxidase to the gold electrodes did not occur to a sig-niWcant extent. The results highlight the advantages ofthe electrochemical sensor array approach, including thesimplicity of the bacterial lysis procedure, the minimalbackground noise of the sensor, the small volume ofsample needed, and the short time period (approxi-mately 40 min, see Materials and methods) from samplelysis to readout of results.

Rapid, species-speciWc detection of uropathogenDNA/RNA is a major improvement over current meth-ods. The current standard for identiWcation of uropatho-gens is urine culture followed by a battery of biochemicaland molecular assays. Even with current automated sys-tems, this method remains expensive, time-consuming,and labor intensive. In contrast, the oligonucleotideprobes we describe in this report allow rapid hybridiza-tion and species-speciWc detection of uropathogen nucleicacids in a dot-blot format over a range of temperatures,including ambient conditions. Results obtained by dot-blot hybridization studies were translated onto a novel

microfabricated electrochemical sensor platform. Thiselectrochemical rRNA microsensor can eliminate theneed for PCR ampliWcation and signiWcantly reducedetection time. The abundance of the 16S rRNA target inbacterial cells should allow detection of small numbers ofpathogens directly by rRNA analysis without requiringPCR of rDNA. We anticipate that detection sensitivity ofthe electrochemical sensor will be equal to or greater thancurrent culture-based methods, and will identify noveluropathogens that cannot be grown in culture [4–7]. TheGeneFluidics electrochemical sensor array is compatiblewith future integration into a micro total analysis system(�TAS) or so-called ‘lab-on-chip’ involving a microXui-dics-based sample preparation module [22]. Such a sys-tem would provide rapid and eYcient detection andidentiWcation of uropathogens at the point-of-care thatwould result in signiWcant reductions in medical cost,improvement in the quality of care for patients with sus-pected urinary tract infection, and more judicious use ofantimicrobial agents.

Acknowledgments

We thank James Matsunaga (UCLA/VAGLAHS),Shu-ching Ma (GeneFluidics), Paul Swanton (GeneFlui-dics), and Melissa Kelley (Haake Laboratory) for experttechnical assistance. E.R.B. McCabe serves as an advisorto GeneFluidics. This paper represents work carried outby C.-P. Sun in partial fulWllment of the requirements fora Master of Science degree. This study was funded byBioengineering Research Partnership Grant EB00127(to B.M.C.) from the National Institute of BiomedicalImaging and Bioengineering and a Fellowship (to J.C.L.)from the American Foundation for Urologic Disease(AFUD). Finally, we gratefully acknowledge the gener-ous support of the Ken & Wendy Ruby Fund forResearch Excellence in Pediatric Urology.

References

[1] C. Kunin, in: Urinary Tract Infections: Detection, Prevention, andManagement, Williams & Wilkins, Baltimore, 1997, pp. 78–101.

[2] J. Eisenstadt, J. Washington, in: H. Mobley, J. Warren (Eds.), Uri-nary Tract Infections: Molecular Pathogenesis and Clinical Man-agement, ASM Press, Washington, DC, 1996, pp. 29–66.

[3] W.E. Stamm, Am. J. Med. 113 (Suppl. 1A) (2002) 1S–4S.[4] D. Relman, T. Schmidt, R. MacDermott, S. Falkow, N. Engl. J.

Med. 327 (1992) 293–301.[5] D. Relman, Science 284 (1999) 1308–1310.[6] I. Kroes, P.W. Lepp, D.A. Relman, Proc. Natl. Acad. Sci. USA 96

(1999) 14547–14552.[7] M.A. Tanner, D. Shoskes, A. Shahed, N.R. Pace, J. Clin. Micro-

biol. 37 (1999) 1863–1870.[8] T.G. Drummond, M.G. Hill, J.K. Barton, Nat. Biotechnol. 21

(2003) 1192–1199.[9] J. Wang, et al., Anal. Chim. Acta 347 (1997) 1–8.

C.-P. Sun et al. / Molecular Genetics and Metabolism 84 (2005) 90–99 99

[10] J. Albers, T. Grunwald, E. Nebling, G. Piechotta, R. Hintsche,Anal. Bioanal. Chem. 377 (2003) 521–527.

[11] F. Martineau, F. Picard, C. Menard, P. Roy, M. Ouellette, M. Ber-geron, J. Clin. Microbiol. 38 (2000) 3280–3284.

[12] K. Rose, J.O. Mason, R. Lathe, Biotechniques 33 (2002) 54–56 58.[13] J. Kellokoski, S. Syrjanen, M. Yliskoski, K. Syrjanen, Oral Micro-

biol. Immunol. 7 (1992) 19–23.[14] N.J. Fuller, D. Marie, F. Partensky, D. Vaulot, A.F. Post, D.J.

Scanlan, Appl. Environ. Microbiol. 69 (2003) 2430–2443.[15] A. Manero, A.R. Blanch, J. Microbiol. Methods 50 (2002) 115–121.[16] K. Greisen, M. LoeVelholz, A. Purohit, D. Leong, J. Clin. Micro-

biol. 32 (1994) 335–351.

[17] V.J. Gau, E. Lan, B. Dunn, C.M. Ho, J.C.S. Woo, Biosen. Bioelec-tron. 16 (2001) 745–755.

[18] K. McCabe, G. Khan, Y. Zhang, E. Mason, E. McCabe, Pediatrics95 (1995) 165–169.

[19] K.M. McCabe, Y.H. Zhang, B.L. Huang, E.A. Wagar, E.R. McC-abe, Mol. Genet. Metab. 66 (1999) 205–211.

[20] J. Wang, Nucleic Acids Res. 28 (2000) 3011–3016.[21] B.M. Fuchs, G. Wallner, W. Beisker, I. Schwippl, W. Ludwig, R.

Amann, Appl. Environ. Microbiol. 64 (1998) 4973–4982.[22] S.J. Lee, S.Y. Lee, Appl. Microbiol. Biotechnol. 64 (2004) 289–299.[23] Y. Van de Peer, S. Nicolai, P. De Rijk, R. De Wachter, Nucleic

Acids Res. 24 (1996) 86–91.