DNA biosensor for assay of E. coli based on detecting DNA products from PCR amplification

Biosensors and Bioelectronics 22 (2006) 506–512

Flow analysis coupled with PQC/DNA biosensor for assay of E. coli based on detecting DNA products from PCR amplification Hui Sun, Youyu Zhang, Yingsing Fung ∗ Department of Chemistry, The University of Hong Kong, Pokfulam Road, Hong Kong SAR, China Received 28 February 2006; received in revised form 30 July 2006; accepted 9 August 2006 Available online 12 September 2006

Abstract A flow-through PQC/DNA biosensor system is developed by combining sequential flow polymerase chain reaction (PCR) products denaturing prior to piezoelectric quartz crystal (PQC) detection via hybridization of ssDNA. The PQC/DNA biosensor is fabricated based on complex formation of neutravidin/biotinylated probe in 0.2 M NaCl in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5). Results show that the coating fabricated provides a desirable quality with satisfactory performance. Its application for Escherichia coli detection under controlled flow at 0.02 mL/min for denaturing PCR products and 10 mL/min for transferring solution between reactors and delivering samples to detector to reduce rehybridization leads to significant improvement in repeatability (R.S.D. < 6%, n = 5) and sensitivity (F = 34 Hz/1000 E. coli cells) as compared to existing manual method (R.S.D. = 19%, n = 5 and F = 26 Hz/1000 E. coli cells, respectively). Down to 23 E. coli cells are detected, satisfying the HKEPD requirements for E. coli count in beach water. © 2006 Elsevier B.V. All rights reserved. Keywords: Biosensor; Piezoelectric quartz crystal; E. coli; Flow analysis; PCR DNA sensor

1. Introduction To ensure the public that beach water is safe for bathing and swimming, environmental protection agencies worldwide require monitoring of bacteria in water during the swimming season. The enteric pathogens, such as Escherichia coli, are the most frequently encountered bacteria found in environmental water. Some E. coli act as a typical inhabitant of human intestinal tract and a causative agent of intestinal and extra-intestinal infections (Buchanan and Doyle, 1997). Thus, E. coli count of water samples collected within the bathing area has been selected as an important indicator in determining the microbiological quality of beach water for use in rating the suitability of bathing in a given beach. Although existing microbiological technique based on morphological evaluation allows the detection of low level of bacterium, it requires the growth of cells into colonies, which is time-consuming, requiring 24–48 h for presumptive results, and necessitates a further 48 h for confirmation. In addition, some pathogens may not be able to grow in labora-

∗

Corresponding author. Tel.: +852 28592162; fax: +852 25482132. E-mail address: [email protected] (Y. Fung).

0956-5663/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2006.08.006

tory culture media or such media may simply not exist for their propagation. Thus, it is important to develop a rapid field monitoring method which does not depend on culture techniques, allowing the detection of E. coli with low concentrations, and capable to deliver quick results in case of an outbreak in E. coli infection. Three common types of biosensors have been developed for bacteria determination based on electrochemical (Serra et al., 2005; Tang et al., 2006), optical (Acharya et al., 2006; Biran et al., 2003) and piezoelectric (Chang et al., 2006; Mo et al., 2002; Su and Li, 2004; Cui et al., 2003; Tombelli et al., 2000) detection. For the determination of E. coli, the electrochemical biosensors are less sensitive and indirect in detection, whereas the optical biosensors are more suited for laboratory rather than field determination at the present stage of development. The piezoelectric quartz crystal (PQC) biosensor provides an attractive system as it offers a real-time output to enable field monitoring, simplicity of use and cost effectiveness. As the most sensitive PQC bacterial sensors based on antibody-antigen interaction give a detection limit of 170 cells/mL (Fung and Wong, 2001), its sensitivity is insufficient for monitoring environmental beach water requiring detection of 24 cells in 100 mL (Hong Kong Environmental Protection Department, 2006).

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

The polymerase chain reaction (PCR) technique has been shown to provide a selective and sensitive means for detecting pathogens in waters without the need of a complex cultivation and additional confirmation steps. Consequently, results can be obtained within hours, instead of days as by traditional methods. Coupling PQC detection with PCR-based DNA technique, it will produce a biosensor with outstanding performance as it combines the rapid and extremely high amplification of the PCR procedure with the versatility of the PQC technology and specificity of the DNA probe. In addition, the use of bacterial DNA has clear advantage over the use of viable cells for detection (Fung and Wong, 2001; Wong et al., 2002), as DNA is not infectious, more stable during storage and would not cause dampening of the quartz crystal vibration due to its much smaller molecular size. However, most previous works developed using PCR/PQC biosensors aim mainly at the identification of a specific isolates (such as the E. coli O157:H7) of E. coli (Deisingh and Thompson, 2001), the target gene region selected is unique for a given strain. Thus, they are not useful to provide a general indicator for total E. coli count. Very little work has been done to quantify E. coli in environmental water using the PQC method (Mao et al., 2006). Since the current regulation on beach water is based on the number of cells per 100 mL of water, more detailed work is needed to establish the link between PQC response and the number of cells present and to enhance the detection sensitivity of the DNA/PQC electrode to meet the demand for low E. coli count in environmental water. A commonly used diagnostic parameter for E. coli determination is to detect the presence and activity of the enzyme ␤-d-glucuronidase (GUD) encoded by the uid gene (Bej et al., 1991). The presence of GUD is usually detected using a fluorogenic substrate methylumbelliferyl glucuronide (MUG). As some E. coli strains such as serotype O157:H7 show MUGnegative (Rompr´e et al., 2002; Monday et al., 2001), it is needed to detect uid gene of E. coli directly in water to circumvent the above problem. Thus, a structural region (166 bp) designated uid A is selected to be amplified by PCR with a pair of primers, UAL-1939 and UAR-2105, located at the carboxyl coding region of the uid A gene of E. coli, as the carboxyl end of the uid A gene has been reported to be unique and conserved in general E. coli (Iqbal et al., 1997; Rompr´e et al., 2002). Using this pair of primers, even the MUG negative isolates of E. coli show amplification at the target DNA region, while no amplification has been observed when DNAs from other bacterial strains are used as template for PCR (Bej et al., 1991; Farnleitner et al., 2000). In the present work, the immobilization reaction between neutravidin and biotinylated-ssDNA probe is adopted to fabricate the DNA/PQC electrode. To enhance its detection sensitivity, a simple and cost-effective approach based on the improvement of the currently used off-line denaturing of dsDNA from PCR products is adopted for PQC detection. The offline approach often leads to considerable recombination of denatured-ssDNA during the transfer of the denatured DNA sample to the detection cell. Sample loss occurs due to splashing at the heating/cooling denaturing process. Although the freez-

507

ing centrifuge method can be used for sample collection, too many transfer steps could increase the probability of recombination between denatured PCR products. As DNA/PQC electrode can only detect ssDNA and not recombined dsDNA, this can lower the detection sensitivity and increase the variability of the results. A flow analysis PQC/DNA system is thus built, incorporating the dsDNA denaturing process online prior to PQC/DNA hybridization to reduce rehybridization of complementary ssDNA after denaturation and to enhance the probability of ssDNA reaching the PQC electrode surface. The use of flow analysis can also shorten the analysis time to allow a large number of samples to be handled daily and enhance the repeatability of the method. The system developed is optimized to obtain the best performance with online dsDNA denaturing process and PQC/DNA hybridization detection. Its applicability for monitoring of E. coli in environmental water will be given and discussed. 2. Materials and methods 2.1. Materials Neutravidin was purchased from Pierce Biotechnology Inc. (Rockford, USA). Agarose gel powders, Tris and other reagents for making up the buffer were obtained from Sigma–Aldrich Inc. (St. Louis, USA). Biotinylated-ssDNA probe, primers UAL1939 and UAR-2150 and other oligonucleotides were purchased and customer-made from Tech Dragon Limited (Hong Kong). The dNTP (deoxynucleoside triphosphate) mix, 10× PCR buffer II (including 100 mM Tris–HCl, 500 mM KCl, pH 8.3), Gold Taq DNA polymerase and 25 mM Mg2+ solution were obtained from Applied Biosystems (Foster, USA). X174 DNA/BsuRI (HaeIII) marker was purchased from the Fermentas Life Sciences (Hanover, USA). 2.2. Apparatus PCR amplification was performed in a GeneAmp PCR System 9700 (Applied Biosystems, Foster, USA). An electrophoresis system (model DYY-5, Liuyi Instrument Factory, Beijing, China) was used to conduct gel electrophoresis with electropherograms recorded by an UV trans-illuminator (Bio-Rad Mini Trans-illuminator) and a Polaroid Direct Screen Instant Camera (Polaroid, UK). The 9 MHz AT-cut quartz crystals (12.5 mm diameter) with gold evaporated (28.3 mm2 area) on both sides were obtained from CH Instruments, Inc. (Shanghai, China). The quartz crystal was placed inside a methacrylate housing with only one side of the crystal in contact with the solution. A schematic diagram showing the flow analysis system is given in Fig. 1. Prior to coating, the gold electrode surface of the crystal was washed with K2 Cr2 O7 –conc. H2 SO4 (1:30, g/g) solution for 2 min and distilled water for another 2 min. A self-constructed oscillator circuit powered by a 5 V dc voltage regulator was used to resonate the piezoelectric crystal with frequency output monitored by a frequency counter (Heathkit model IM-4120). The electrochemical impedance spectrum (EIS) and cyclic voltammograms (CV) were measured in an electrochemical cell with

508

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

Fig. 1. Schematic diagram showing the flow-through PQC/PCR DNA biosensor system.

a three-electrode configuration (9 MHz Au-PQC working electrode; Ag/AgCl reference electrode and Pt counter electrode). The electrolyte solution contained 0.2 mol/L NaCl, 1 mmol/L K4 Fe(CN)6 , 1 mmol/L K3 Fe(CN)6 and 0.05 mol/L Na2 HPO4 with pH adjusted to 7.5. The electrochemical impedance measurements were carried out using an EG&G 263 potentiostat and a model 5201 lock-in amplifier controlled by a microcomputer under the M398 software. Five points per decade were measured. The cyclic voltammetry experiment was conducted by the Princeton Applied Research model 263 potentiostat under software-control (EG&G model 270). 2.3. PCR amplification of target DNA Genomic DNA of E. coli (ACTT 25922) was extracted as described previously (Leung et al., 2001). For the detection of E. coli, a 166 base pair (bp) region of E. coli ␤-dglucuronidase gene (uid A) was amplified using 20-mer primers UAL-1939 and UAR-2105 (Iqbal et al., 1997). The 25-␮L PCR mixture was consisted of 3 ␮L of DNA extract, 1 ␮M of each primer, 0.2 mM of each dNTP, 1.5 mM MgCl2 and 1.25 units Gold Taq polymerase. The target DNA was amplified using an initial denaturation at 95 ◦ C for 5 min, and then 50 PCR cycles of three-step PCR amplification (denaturation at 95 ◦ C for 15 s, primer re-annealing at 50 ◦ C for 30 s and primer extension at 72 ◦ C for 30 s). A final elongation was carried out at 72 ◦ C for 7 min, following by a final cooling stage at 4 ◦ C until the post-PCR analysis. PCR products were examined by gel electrophoresis to ascertain successful amplification. The PCR products were indexed against a X174 DNA/BsuRI

(HaeIII) marker to estimate the length of the sequence to be amplified. 2.4. Flow-through analysis of target E. coli DNA Initially, the Tris buffer (10 mM Tris, 1 mM EDTA, 0.2 M NaCl, pH 7.5) was passed through the flow system. The inlet/outlet pumps were controlled to deliver 100 ␮L of solution to the detection cell. After the baseline was stabilized, 20 ␮L of neutravidin (1 mg/mL) was added into the cell through the reagent reservoir and the pump was stopped for 20 min. After the immobilization of neutravidin was completed, the pump was restarted again. The detection cell was then washed using the buffer prior to injection of 15 ␮L biotinylated probe (approximately 15 ␮g/mL). After another interval of 20 min (pump stopping), the cell was washed by buffer for 5 min. At the same time, 20 ␮L PCR products were driven by pump to firstly pass through the hot coiled tubing (100 ◦ C) for 5 min and then rapidly passing through the cold coiled tubing (0 ◦ C) for 2 min prior to a rapid transfer to the detection cell. Two flow rates were maintained with a slow flow at 0.02 mL/min for denaturing PCR products and a fast flow at 10 mL/min for transferring solution between reactors and delivering samples to the detection cell. After the denatured PCR products were injected into the PQC sensor for 15 min, the cell was washed by the buffer again prior to the recording of frequency shift after a stable frequency was reached. For the regeneration of the neutravidine/biotinylated probe modified electrode, 1 mM HCl was added to the electrode surface for 30 s, following by a thorough washing using the buffer solution.

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

509

Table 1 The sequence of different oligonucleotides Oligonucleotides

Sequence

Biotinylated-ssDNA probe Complementary oligo Non-complementary oligo Primer UAL-1939 Primer UAR-2105 PCR producta

5 -CAAAAACGCTGGACTGGC-Biotin-3 5 -GCCAGTCCAGCGTTTTTG-3 5 -ATTCCAATGTTACCGCCA-3 5 -TATGGAATTTCGCCGATTTT-3 5 -TGTTTGCCTCCCTGCTGCGG-3 5 -TGTTTGCCTCCCTGCTGCGGTTTTTCACCGAAGTTCATGCCAGTCCAGCGTTTTTGCAGCAGAAAAGCCGCCGACTTCGGTTTGCGGTCGCGGGTGAAGATCCCCTTCTTGTTACCGCCAACGCGCAATATGCCTTGCGAGGTCGCAAAATCGGCGAAATTCCATA-3

a

The sequence of PCR products in bold represents the sequence of the primers. The sequence underlined is complementary to the probe.

3.1. Optimization of buffer composition for DNA/PQC biosensor

neutravidin (Fig. 2b), the maximum frequency shift due to the interaction of biotinylated-ssDNA with neutravidin is found at pH 7.5 and no frequency shift has been observed when the buffer pH is over 11 or below 4. Thus, the buffer pH is adjusted to 7.5 during the entire procedure for sensor fabrication and analyte determination. The effects of buffer ionic strength on the modification of PQC electrode and detection of target ssDNA are investigated with results shown in Fig. 3. The decrease of the NaCl concentration in the buffer solution (10 mM Tris, 1 mM EDTA and x M NaCl, pH 7.5) is found to give rise to a rapid increase in the maximum frequency response upon the addition of neutravidin. A plateau is observed at NaCl concentration below 0.4 mol/L. For the interaction between biotinylated-ssDNA and neutravidin, the buffer ionic strength does not show significant effect as the frequency shift is almost independent of NaCl concentration from 0 to 1.0 mol/L. For hybridization between target ssDNA and the biotinylated-ssDNA probe, only 7 Hz has been observed in the TE-0 M NaCl. However, a gradual increase in the frequency shift is found with the increase of the NaCl concentration in TE buffer. A maximum binding is obtained when NaCl concentration is more than 0.2 mol/L. It seems that NaCl plays an important role not only on the modification of the electrode with neutravidin but also on the detection of the target ssDNA. To obtain the best

The DNA/PQC biosensor is fabricated based on the formation of a neutravidin–biotin complex (Ghafouri and Thompson, 1999) on the Au electrode. Two steps have been included: immobilization of neutravidin onto PQC gold surface, following by biotinylated-ssDNA probes. Maximum coverage on PQC electrode is reached at 0.1 mg/mL of neutravidin in the detection cell and a maximum immobilization of the biotinylated probes onto the neutravidin-modified electrode is obtained with at least 1.0 ␮g/mL of biotinylated probe present in the detection cell. As the buffer pH strongly affects the adsorption of protein onto the gold electrode surface, its effect on the fabrication of PQC sensor surface is shown in Fig. 2. Results (Fig. 2a) show that the frequency shift caused by the adsorption of neutravidin is decreased rapidly when buffer pH (10 mM Tris, 1 mM EDTA and 0.2 M NaCl) is increased from 4 to 5.5, or from 8.5 to 11. However, it is almost independent of pH in the range between 5.5 and 8.5. Thus, buffer pH at 7.5 is used in subsequent experiments to study the adsorption of neutravidin on PQC electrode. To investigate the binding of biotinylated-ssDNA probe with

Fig. 2. The effect of buffer pH on the frequency responses of the PQC sensor: (a) frequency responses of PQC to the immobilization of neutravidin and (b) frequency responses of neutravidin-modified PQC to the immobilization of biotin-ssDNA probe. Buffer background: 10 mM Tris–1 mM EDTA–0.2 M NaCl.

3. Results and discussion In the present work, the amplified 166-bp gene fragments are determined by PQC biosensor previously modified by neutravidin and biotinylated-ssDNA probes. The sequence of the biotinylated-ssDNA probe is targeted at location close to the end of the amplified 166-bp gene fragment. This design facilitates the placement of the probe to its complementary sequence in the long gene fragment and overcomes to some extent the stereochemical and orientational hindrance for the improvement of the hybridization efficiency. To calibrate and optimize the performance of the PQC biosensor, the 18-mer synthetic ssDNA oligonucleotides complementary to the biotinylatedssDNA probe is used as standard oligonucleotides for positive control. Two negative controls, non-complementary DNA sample and PCR blank (without adding E. coli. genomic DNA template during the PCR process), have been used under matching working conditions. The sequences of different DNA fragments are listed in Table 1. Results on the fabrication of the DNA/PQC biosensor and its characterization are given in the following sections.

510

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

Fig. 3. The effect of NaCl in buffer solution on the frequency responses of the PQC sensor: (a) frequency responses of the PQC to the immobilization of neutravidin, (b) frequency responses of neutravidin-modified PQC to the immobilization of biotinylated-ssDNA probe and (c) frequency responses of neutravidin/biotinylated-ssDNA modified PQC to complementary ss-DNA standard. Buffer background: 10 mM Tris–1 mM EDTA–x M NaCl at pH 7.5.

modification of electrode and to achieve the highest response of the sensor to the target ssDNA, 0.2 M NaCl in TE buffer (10 mM Tris, 1 mM EDTA, pH 7.5) is adopted for use in subsequent experiments. 3.2. Study and characterization of the modified PQC coating surface Using the modified PQC under the optimum conditions, the maximum frequency shifts (Fsol ) due to the absorption of neutravidin and biotinylated probe are 310 and 53 Hz, respectively. The maximum frequency shift due to the hybridization of complementary ssDNA standard with biotinylated-ssDNA probe is found to be 44 Hz. Although the Fsol values cannot be directly transposed to mass changes due to the presence of water within the adsorbed layer and/or viscoelastic effects arising from the solvent and the coating layer, as well as slip between the film and substrate (Ghafouri and Thompson, 1999), the Fair values can be converted into reagent loadings per unit area using the Sauerbrey equation. For an AT-cut 9 MHz PQC electrode, a change of 1 Hz in air is corresponding to a mass change of 5.5 ng/cm2 on the surface of the crystal. Since the average frequency shift in air due to the absorption of neutravidin is found to be about 148 Hz, the surface coverage calculated from the Fair values is thus 1.36 × 10−11 mol/cm2 . Since neutravidin has been reported (Vermette et al., 2003) to possess basically the molecular dimension of 4.5 nm × 4.5 nm × 5.8 nm, it indicates the formation of essentially, a close-packed layer of neutravidin on the Au sur-

face. The Fair values for biotin-DNA cannot be measured, as neutravidin could be denatured during drying (Caruso et al., 1998). Assuming that the ratio of Fsol /Fair for the binding of biotin-DNA (18-mer) probe or target ssDNA (18-mer) is the same as neutravidin (Fsol /Fair = 2:1) (Caruso et al., 1997), it can be estimated that the loadings of biotinylated-ssDNA and target ssDNA onto the surface of PQC electrode are 2.43 × 10−11 and 2.21 × 10−11 mol/cm2 , respectively, and the binding ratio between neutravidin, biotinylated probe and the complementary ssDNA on the PQC electrode are about 1:1.8:1.6 (Table 2). From the atomic force microscopy (AFM) image of the PQC surface, the image roughness mean square (RMS) value of the electrode is found to decrease from 5.344 nm (for gold electrode) to 1.788 nm after the electrode has been modified with neutravidin and biotinylated probe. The results indicate that a successive grafting of neutravidin and biotinylated probe onto the electrode leads to a reduction in the roughness of the gold surface by forming a dense coverage onto the electrode surface. With successive immobilization of neutravidin and biotinylated probe onto the PQC surface, the peak current obtained from cyclic voltammetry (CV) study is found to decrease with a corresponding increase in the peak potential difference (Ep). The results indicate that the charge transfer resistance is increased after successive immobilization of the neutravidin and biotinylated probes onto the bare gold electrodes. Results from the Niquist plot of the electrochemical impedance spectra measurements show a straight line at all frequency range investigated for a bare PQC electrode. However, an arc in the high frequency range and a line in the low frequency range are obtained for the impedance spectra of PQC electrodes modified with neutravidin and biotinylated-ssDNA probe. The magnitude of the arc is found to increase with successive immobilization of neutravidin and biotinylated probe onto the PQC surface. The results are attributed to the increase in both the contact resistance and charge transfer resistance (Pyun and Bae, 1996) at the neutravidin and biotinylated-ssDNA probe coating/electrode interface. The CV and EIS spectra of the modified electrode are found to be more or less the same after storage at 4 ◦ C in humid environment for up to 5 days. The modified electrode is shown to be stable and can be prepared prior to onsite determination of E. coli. 3.3. Application of PQC biosensor for E. coli detection In the present work, the following procedures have been adopted to enhance the sensitivity and improve the repeatability of the biosensor: firstly, an enclosed sequential flow-through

Table 2 The maximum loadings of different reagents onto the PQC electrode Reactants

Frequency shift (Fsol , Hz)

Repeatability (%R.S.D., n = 5)

Loadinga (mol/cm2 )

Neutravidin Biotinylated DNA probe Complementary hybridized DNA Non-complementary DNA

310 53 44 Not detected

3.2 3.9 4.6

1.36 × 10−11 2.43 × 10−11 2.21 × 10−11 –

Buffer background: 10 mM Tris–1 mM EDTA–0.2 M NaCl, pH 7.5. a The reagent loadings calculated is using the Sauerbrey equation based on F (F /F = 2:1). air sol air

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

system with PCR product denaturation at 100 ◦ C and freezing at 0 ◦ C is built to eliminate the loss of sample due to splashing and to minimize the recombination of denatured ssDNA during the transfer from the freezing coiled tubing to the detection cell; secondly, the sample delivering tube is placed just above the immersed PQC sensor in the detection cell to allow immediate dilution when a small sample volume (20 ␮L) of the denatured PCR products is added rapidly into the detection cell containing about 100 ␮L buffer. At the same time, it also reduces the chance of recombination of denatured PQC products and hence maximizes the detection of ssDNA; thirdly, the flow rates of the samples and hence their corresponding resident times for denaturation and freezing are controlled carefully to optimize the sensitivity and repeatability of the procedure. To allow sufficient time for PCR product denaturation and to reduce transfer time between reactors, two different flow rates are used. For denaturation, a slow flow rate at 0.02 mL/min is used whereas for transfer between reactors and from freezing tubing to detector cell, a fast flow rate at 10 mL/min has been applied. To adjust the resident time for the denaturation of the PCR products, the lengths of the two coiled tubings in the oven at 100 ◦ C and in the freezing bath at 0 ◦ C are kept at 50 and 25 cm, respectively, so as to attain resident times of 5 and 2 min, respectively. Using the flow-through system as described above, a decrease of frequency shift at about 34 Hz is observed with the injection of PCR products from about 1000 E. coli cells with R.S.D. (n = 5) below 6%. Whereas, using an off-line manually denaturing method, the R.S.D. (n = 5) is found to be 19% with mean frequency shift reduced to 26 Hz. Thus, both the repeatability and sensitivity of the method are improved using the flowthrough system in comparison to the traditional off-line manually denaturing method. The improvement in performance is attributed to the reduction in the rehybridization between the denatured PCR products during the transfer process and the elimination of sample loss due to splashing during denaturing DNA. The response of the amplified PCR products on the PQC biosensor is given in Fig. 4. The initial sharp spike is due to the temperature disturbance when sample at 0 ◦ C is injected into the detection cell. As the injected volume (20 ␮L) is much less than the buffer volume (100 ␮L) in contact with the PQC biosensor at room temperature, an equilibrium is established quickly after a couple of minutes. For the injection of non-complementary DNA sample or PCR blank (without addition of E. coli genomic DNA during PCR process) as negative control, no apparent frequency shift is found, whereas for the injection of PCR products from E. coli cells, increasing responses are obtained for solutions with higher E. coli counts. The results indicate that the PQC sensor provides a high specificity against non-complementary ssDNA or the background PCR solution. For the enhancement in sensitivity, PCR-amplified DNA from as little as 100 fg E. coli genomic DNA or 23 bacterial cells can be consistently detected with mean frequency shift of 10 Hz. This level of detection is equivalent to a PCR product concentration at 0.15 × 10−4 mmol/L. The E. coli cells can be collected from seawater on small filters for

511

Fig. 4. The frequency responses caused by the interaction of different DNA samples (20 ␮L) with the PQC electrode modified with neutravidin and biotinssDNA probe: (a) non-complementary DNA sample, (b) PCR products from about 23 E. coli cells and (c) PCR products from 1000 E. coli cells. The arrow marks the time for addition of sample into the detection cell.

direct insertion into PCR vial for amplification (Bej et al., 1991; Rompr´e et al., 2002). According to the Annual Ranking System of the HKEPD (Hong Kong Environmental Protection Department, 2006), if the E. coli count is less than 24 cells per 100 mL beach water, the rank of the beach water is considered to be suitable for bathing. The PQC method is able to detect change more than 1 Hz in frequency shift. Thus, the method developed using PCR/DNA/PQC possesses sufficient sensitivity to meet the demand for a quick classification of beach water under acceptable quality. 4. Conclusions A flow-through PQC/PCR DNA biosensor system has been developed by combining an enclosed sequential flow dsDNA denaturing and PQC detection via hybridization of ssDNA for E. coli detection. The results indicate improvement in repeatability (R.S.D. < 6%, n = 5) and enhancement in sensitivity (F = 34 Hz/1000 E. coli cells) as compared to the traditional manual method (R.S.D. = 19%, n = 5 and F = 26 Hz/1000 E. coli cells). Down to 23 E. coli cells can be detected by the biosensor developed which satisfy the demand from the Annual Ranking System of the HKEPD for E. coli count less than 24 cells per 100 mL for bathing water. Thus, the method developed is shown to have sufficient sensitivity to meet the demand for a quick classification of beach water into different qualities. Acknowledgements We would to thank Dr. S.P. Ng and Dr. W.C. Yam, Microbiology Department, Hong Kong University, to provide the E. coli samples and helpful discussion. This research is supported by the Hong Kong Research Grants Council of the Hong Kong Special Administrative Region, China (HKU 7043/03P), and the Hong Kong University Research and Conference Grants Committee.

512

H. Sun et al. / Biosensors and Bioelectronics 22 (2006) 506–512

References Acharya, G., Chang, C.L., Savran, C., 2006. J. Am. Chem. Soc. 128, 3862–3863. Bej, A.K., Dicesare, J.L., Haff, L., Atlas, R.M., 1991. Appl. Environ. Microbiol. 57, 1013–1017. Biran, I., Rissin, D.M., Ron, E.Z., Walt, D.R., 2003. Anal. Biochem. 315, 106–113. Buchanan, R.L., Doyle, M.P., 1997. Food Technol. 51, 69–76. Caruso, F., Furlong, D.N., Niikura, K., Okahata, Y., 1998. Colloids Surf. B—Biointerfaces 10, 199–204. Caruso, F., Rodda, E., Furlong, D.F., Niikura, K., Okahata, Y., 1997. Anal. Chem. 69, 2043–2049. Chang, K.S., Jang, H.D., Lee, C.F., Lee, Y.G., Yuan, C.J., Lee, S.H., 2006. Biosens. Bioelectron. 21, 1581–1590. Cui, S., Schroeder, C.M., Zhang, D.Y., Meng, J., 2003. J. Appl. Microbiol. 95, 129–134. Deisingh, A.K., Thompson, M., 2001. Analyst 126, 2153–2158. Farnleitner, A.H., Kreuzinger, N., Kavka, G.G., Grillenberger, S., Rath, J., Mach, R.L., 2000. Appl. Environ. Microbiol. 66, 1340–1346. Fung, Y.S., Wong, Y.Y., 2001. Anal. Chem. 73, 5302–5309. Ghafouri, S., Thompson, M., 1999. Langmuir 15, 564–572. Hong Kong Environmental Protection Department, 2006. Beach Water Quality. http://www.info.gov.hk/epd/.

Iqbal, S., Robinson, J., Deere, D., Saunders, J.R., Edwards, C., Porter, J., 1997. Lett. Appl. Microbiol. 24, 498–502. Leung, P.H., Yam, W.C., Ng, W.W., Peiris, J.S., 2001. Epidemiol. Infect. 126, 173–179. Mao, X.L., Yang, L.J., Su, X.L., Li, Y.B., 2006. Biosens. Bioelectron. 21, 1178–1185. Mo, X.T., Zhou, Y.P., Lei, H., Deng, L., 2002. Enzyme. Microb. Technol. 30, 583–589. Monday, S.R., Whittam, T.S., Feng, P.C., 2001. J. Infect. Dis. 184, 918–921. Pyun, S.I., Bae, J.S., 1996. Electrochim. Acta 41, 919–925. Rompr´e, A., Servais, P., Baudart, J., de-Roubin, M.R., Laurent, P., 2002. J. Microbiol. Methods 49, 31–54. Serra, B., Morales, M.D., Zhang, J.B., Reviejo, A.J., Hall, E.H., Pingarron, J.M., 2005. Anal. Chem. 77, 8115–8121. Su, X.L., Li, Y.B., 2004. Biosens. Bioelectron. 19, 563–574. Tang, H., Zhang, W., Geng, P., Wang, Q.J., Jin, L.T., Wu, Z.R., Lou, M., 2006. Anal. Chim. Acta 562, 190–196. Tombelli, S., Mascini, M., Sacco, C., Turner, A.P.F., 2000. Anal. Chim. Acta 418, 1–9. Vermette, P., Gengenbach, T., Divisekera, U., Kambouris, P.A., Griesser, H.J., Meagher, L., 2003. J. Colloid Interface Sci. 259, 13–26. Wong, Y.Y., Ng, S.P., Ng, M.H., Si, S.H., Yao, S.Z., Fung, Y.S., 2002. Biosens. Bioelectron. 17, 676–684.