In-a-Day Electrochemical Detection of Coliforms in Drinking Water

The method relies on the detection of phenol released after the hydrolysis of phenyl β-d-galactopyranoside (PG) by β-galactosidase. Under the optimi...
0 downloads 0 Views 183KB Size
Anal. Chem. 2005, 77, 8115-8121

In-a-Day Electrochemical Detection of Coliforms in Drinking Water Using a Tyrosinase Composite Biosensor Beatriz Serra,†,‡ M. Dolores Morales,† Jinbiao Zhang,† A. Julio Reviejo,† Elizabeth H. Hall,‡ and Jose M. Pingarron*,†

Department of Analytical Chemistry, Faculty of Chemistry, Complutense University of Madrid, Avenida Complutense s/n 28040 Madrid, Spain

A rapid method for the detection of fecal contamination in water based on the use of a tyrosinase composite biosensor for improved amperometric detection of β-galactosidase activity is reported. The method relies on the detection of phenol released after the hydrolysis of phenyl β-D-galactopyranoside (PG) by β-galactosidase. Under the optimized PG concentration and pH (4.0) values, a detection limit of 1.2 × 10-3 unit of β-galactosidase/mL-1 was obtained. The capability of the sensor for the detection of Escherichia coli was evaluated using polymyxin B sulfate to allow permeabilization of the bacteria membrane. A detection limit of 1 × 106 cfu of E. coli mL-1 was obtained with no preconcentration or pre-enrichment steps. To improve the analytical characteristics for bacteria detection, the processes involving galactosidase induction during incubation and membrane permeabilization were optimized. Using 0.25 mM isopropyl β-Dthiogalactopyranoside for the enzyme activity induction, and 10 µg mL-1 polymyxin B sulfate as permeabilizer agent, it was possible to detect bacteria concentrations as low as 10 cfu mL-1 after 5 h of enrichment. The possibility of detecting E. coli at the required levels for drinking water quality assessment (1 cfu/100 mL) is demonstrated, the time of analysis being shorter than 6.5 h and involving a simple methodology. Public and environmental health protection requires drinking water free of pathogenic bacteria. Nevertheless, this type of microorganism contaminates water areas sporadically and does not survive for long times. These facts, together with the low concentrations in which they are usually present, make them difficult to detect, requiring complex and time-consuming (days) methods. The elapsed time is too great to provide useful information that may require immediate remedial measures. Coliforms, which are found in large numbers among the intestinal flora of humans and other warm-blooded animals, can be detected in higher concentration than pathogenic bacteria and, therefore, are used as an index of the potential presence of enteropathogens * To whom correspondence should be addressed: Phone: +34913944315. Fax: +34913944329. E-mail: [email protected]. † Complutense University of Madrid. ‡ Present address: Institute of Biotechnology, University of Cambridge, Tennis Court Rd., Cambridge CB2 1QT UK. 10.1021/ac051327r CCC: $30.25 Published on Web 11/12/2005

© 2005 American Chemical Society

from fecal contamination in water environments. In particular, Escherichia coli is employed as an indicator of the possible presence of pathogens in aquatic systems. Conventional microbiological analyses of drinking water for coliform bacteria include the multiple-tube fermentation technique and the membrane filter technique using different selective media and different incubation conditions. These methods have the limitations of long incubation times (24-48 h), antagonistic organism interference, lack of specificity, and poor detection of slow-growing or viable but nonculturable microorganisms.1 To improve sensitivity and decrease time of analysis, several methods have been developed based on the detection of β-D-galactosidase activity,2-14 as this enzyme is involved in the lactose fermentation process, which is indicative of coliforms.1 The enzyme activity is observed via the measurement of a specific property (electroactivity, color, fluorescence, or luminescence) of the hydrolyzed product of a defined β-D-galactosidase substrate. These enzymatic methods have been shown to be a suitable alternative to the classical techniques, but are more expensive, and the incubation time, even though reduced, remains too long for in-a-day results. More sophisticated molecular-based methods are proposed that allow very specific and rapid detection without the need of a cultivation step and rely on immunological, polymerase chain (1) Rompre´, A.; Servais, P.; Baudart, J.; de-Roubin, M.-R.; Laurent, P. J. Microbiol. Methods 2002, 49, 31-54. (2) Ley, A.; Barr, S.; Fredenburgh, D.; Taylor, M.; Walter, N. Can. J. Microbiol. 1993, 39, 821-825. (3) Park, S. J.; Lee, E. J.; Lee, D. H.; Lee, S. H.; Kim, S. J. Appl. Environ. Microbiol. 1995, 61, 2027-2029. (4) Clark, J. A.; El-Shaarawi, A. H. Appl. Environ. Microbiol. 1993, 59, 380388. (5) MacCarty, S. C.; Standridge, J. H.; Stasiak, M. C. J. AWWA 1992, 84, 9197. (6) Fricker, E. J.; Fricker, C. R. Water Res. 1996, 30, 2226-2228. (7) Griffith, K. L.; Wolf, R. E., Jr. Biochem. Biophys. Res. Commun. 2002, 290, 397-402. (8) Pe´rez, F.; Tryland, I.; Mascini, M.; Fiksdal, L. Anal. Chim. Acta 2001, 427, 149-154. (9) Nistor, C.; Osvick, A.; Davidsson, R.; Rose, A.; Wollenberger, U.; Pfeiffer, D.; Emne´us, J.; Fiksdal, L. Water Sci. Technol. 2002, 45, 191-199. (10) Mittelmann, A. S.; Ron, E. Z.; Rishpon, J. Anal. Chem. 2002, 74, 903-907. (11) Masuda-Nishimura, I.; Fukuda, S.; Sano, A.; Kasai, K.; Tatsumi, H. Lett. Appl. Microbiol. 2000, 30, 130-135. (12) Berg, J. D.; Fiksdal, L. Appl. Environ. Microbiol. 1988, 54 (8), 2118-2122. (13) Van Poucke, S. O.; Nelis, H. J. Appl. Environm, Microbiol. 1995, 61 (12), 4505-4509. (14) Van Poucke, S. O.; Nelis, H. J. J. Microbiol. Methods 2000, 42, 233-244.

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005 8115

reaction (PCR), and in situ hybridization (ISH) techniques.15-20 Despite the advantages they present, each of these methods has its own unique set of drawbacks: immunological techniques show low antibody specificity, often require enrichments, and cannot differentiate viable and nonviable cells; PCR can be used to detect coliform bacteria by means of signal amplification of DNA sequence coding for the lacZ gene (β-galactosidase gene) and is more rapid than plating techniques, but it can be inhibited by components of the sample matrix and cannot be used to determine viability. The ISH methods provide quantitative data within one day, but the enumeration of cells is linked to the physiological status of the bacteria. Moreover, these molecular-based methods are in general more complex and expensive. The objective of the present study is to evaluate the use of a tyrosinase composite biosensor for an improved amperometric detection of β-galactosidase activity as a rapid method for the detection of fecal contamination of water. The method relies on the detection of a phenolic compound that is released after the hydrolysis of the corresponding phenyl galactopyranoside compound by the galactosidase. Our research group had developed tyrosinase composite electrodes previously, which exhibited a high sensitivity for phenolic compounds, as well as great stability and reproducibility.21 This research was focused on decreasing analysis time, increasing sensitivity, and accuracy using inexpensive instruments and reagents to obtain a procedure that can ultimately be automated for the continuous on-line monitoring of water. EXPERIMENTAL SECTION Apparatus, Electrodes, and Electrochemical Cell. Amperometric measurements were performed on a BAS model LC4C amperometric detector connected to a Linseis model L250 recorder. The electrochemical cell was a BAS VC-2 cell with a BAS RE-5 Ag/AgCl/KCl (3 M) reference electrode and a Pt wire auxiliary electrode. Graphite-Teflon-tyrosinase composite electrodes were used as the working electrode and were fabricated in the form of cylindrical pellets as described previously.21 A Raypa AES-75 steam sterilizer, a P-Selecta Agimatic-N magnetic shaker, a Griffin flask shaker, a P-Selecta Ultrasons ultrasonic bath, and a P-Selecta Tectron 200 thermostatic bath were also used. Reagents and Solutions. Wild type E. coli were obtained from the Coleccio´n Espan ˜ola de Cultivos Tipo. Graphite (Carbon of America), Teflon (Fluka), and tyrosinase (from mushroom, EC 1.14.18.1, activity 6750 units/mg of solid; Sigma) were used for the fabrication of the biosensor. Phenol (Sigma) and p-aminophenol (Sigma) were used as tyrosinase substrates, phenyl β-Dgalactopyranoside (PG) (Fluka) and p-aminophenyl D-thiogalactopyranoside (PAPG) (Sigma) as β-galactosidase substrates, lactose (Sigma) and isopropyl β-D-galactopyranoside (IPTG) (Sigma) as galactosidase inductors, and polymyxin B sulfate (15) Chapman, P. A.; Ellin, M.; Ashton, R.; Shafique, W. Int. J. Food Microbiol. 2001, 68, 11-20. (16) Uyttendaele, M.; Van Bosxtael, S.; Debevere, J. Int. J. Microbiol. 1999, 52, 85-95. (17) Pe´rez, F. G.; Mascini, M. Anal. Chem. 1998, 70, 2380-2386. (18) Tu, S.-I.; Patterson, D.; Uknalis, J.; Irwin, P. Food Res. Int. 2000, 33, 375380. (19) Su, X.-L.; Li, Y. Biosens. Bioelectron. 2004, 19, 563-574. (20) Garcia-Armisen, T.; Servais, P. J. Microbiol. Methods 2004, 58, 269-279. (21) Serra, B.; Jime´nez, S.; Mena, M. L.; Reviejo, A. J.; Pingarro´n, J. M. Biosens. Bioelectron. 2002, 17, 217-226.

8116

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

(Sigma) and sodium dodecyl sulfate (SDS) (Fluka) as permeabilizers. All chemicals were of analytical-reagent grade, and the water used was obtained from a Milli-Q purification system (Millipore). Stock solutions of the phenolic compounds and the phenyl β-Dgalactopyranoside compounds were prepared in a 0.05 mol L-1 phosphate buffer of pH 6.5 or 4.0. Culture media, consisting of Luria broth (LB) (Scharlau), and LB agar (Scharlau), were used for plate colony counting. Culture media and flasks were sterilized by autoclaving at 121 °C for 15 min. Galactosidase Determination. Amperograms in stirred solutions were recorded after immersing the biosensor, at room temperature, in the electrochemical cell containing 5.0 mL of phosphate buffer solution, which was mechanically stirred at a constant rate. The selected potential was applied, and the background current was allowed to stabilize. Next, the appropriate volume of stock solution of the corresponding galactosidase substrate (PAPG or PG) was added with a micropipet. When a steady-state current was reached, the corresponding amount of β-galactosidase was also added, and the current was recorded for 10 min. Bacteria Cultivation. Wild-type E. coli belongs to the microorganism biosafety level 2 group, and consequently, all safety considerations concerning this group22 were accomplished in the manipulation of these bacteria. E. coli cultures were grown overnight in LB medium at 37 °C with aeration by shaking, which allowed the growing stationary phase to be reached. Then, bacterial cultures were serially diluted (10-fold steps), and 10µL aliquots of samples were applied to LB agar plates and incubated for 24 h at 37 °C, for enumeration of colonies. At the same time, the stationary-phase cultures were diluted to 10-108 cfu/mL in 50 mL of LB medium, containing IPTG to a final concentration of 0.25 mM, to induce galactosidase activity. The bacterial suspensions were incubated during defined periods of time (1-5 h). Bacterial samples of different concentrations, not subjected to growing/enrichment procedures, were obtained from the stationary-phase suspensions by dilution to 10-108 cfu/mL with 0.05 mol L-1 phosphate buffer (pH 4.0), containing 10µg/mL polymyxin B sulfate. Detection of E. coli. After the incubation step, bacteria were recovered by membrane filtration (0.45-µm pore size, 47-mmdiameter nylon filter; Osmonics) and transferred to a 5 mL of 0.05 mol L-1 phosphate buffer solution, containing 10 µg/mL polymyxin B sulfate. This solution acted as the working solution in the electrochemical cell. When the background current was stabilized, the appropriate volume of phenyl β-D-galactopyranoside solution was added with a micropipet and the current was recorded for 10 min. The current change was compared with the obtained for bacteria stock solutions for concentration estimation. RESULTS AND DISCUSSION A composite graphite-Teflon-tyrosinase biosensor, previously developed by our group,21 showed excellent analytical performance for the detection of phenolic compounds, such as low limits of detection (9.9 × 10-8 mol L-1 for phenol), high reproducibility of (22) Protection of workers from risks related to exposure to biological agents at work Directive 2000/54/EC.

Figure 1. Schematic diagram displaying the reactions involved in the detection of the phenolic compounds formed from the galactosidase substrate.

Figure 2. Current-time recordings for the addition of 4.0 × 10-4 mol L-1 PAPG or PG followed by the addition of 0.03 unit of β-galactosidase/mL.

the amperometric measurements, renewability of the electrode surface by simply polishing, and long useful lifetime, defined as the period of time in which one single biosensor provided everyday mean values of the amperometric signals inside the control limits, set at (3 × standard deviation of the central value (1 month for a single biosensor).21 Moreover, responses from different biosensors fabricated in the same way yielded very reproducible responses. These advantageous features have been exploited in this work for the detection of galactosidase activity, following the reaction pathway depicted in the schematic diagram of Figure 1. Optimization of Galactosidase Detection. Two different enzyme substrates, PAPG and PG, were evaluated for β-galactosidase detection. The products obtained after substrate hydrolysis by galactosidase are p-aminophenol and phenol, respectively. The amperometric response obtained at a graphite-Teflontyrosinase electrode at an applied potential of -0.15 V showed a remarkably more sensitive response to phenol than to p-aminophenol, and thus, the limit of detection for the enzyme activity would be lower using PG as the substrate. Figure 2 shows current-time recordings obtained for galactosidase at the composite tyrosinase electrode using both enzyme substrates. As can be observed, the addition of the substrates to the electrochemical cell yielded an amperometric response, even when no galactosidase was present, proving that self-hydrolysis occurred with both compounds. It is also observed that the galactosidase signal was remarkably larger when PG is used as

the substrate. Furthermore, another important observation is that when PAPG was used as the substrate, the steady-state amperometric response for β-galactosidase was reached in a longer time: the lower the amount of β-galactosidase the longer the time required to reach the steady state. As a comparison, the steady state was reached in 5 min for a galactosidase concentration of 0.02 unit mL-1 using PG and for a concentration of 8.25 units mL-1 when PAPG was employed. Hence, to attain a compromise between rapidity and sensitivity, the current was monitored 10 min after the addition of the substrate. To obtain the best signal-to-background ratio for β-galactosidase, the substrate concentration and the working solution pH were optimized for both PG and PAPG (Figure 3). Different β-galactosidase concentrations were used for PG (0.025 unit mL-1) and PAPG (0.825 unit mL-1) in order to allow a visual comparison of the trend for both substrates, as the response for PAPG is much smaller. As can be seen, a remarkable increase in the blank response was observed when the PG concentration was increased. This avoided obtaining a plateau for the response to galactosidase. Therefore, the PG concentration selected for further work was that exhibiting the highest signalto-blank ratio (1.2 × 10-3 mol L-1 PG). The same criterion was applied to choose the PAPG concentration (4.0 × 10-4 mol L-1). Moreover, it is also evident that, instead of the higher blank signals for PG, a better signal-to-background ratio was obtained with this compound. Moreover, Figure 3b shows that the optimum pH value is not the same for each of the substrates, with the reaction more favored at more acidic pHs for PG. This could be due to the difficulty with which PAPG is hydrolyzed at lower pH, when the amino group is protonated. According to the same criterion used above for the choice of the substrate concentration, pH values of 4.0 and 6.5 were selected for further work with PG and PAPG, respectively. Figure 4 compares calibration plots for galactosidase with both substrates. The slope value of the linear portion of the calibration plot using PG as the substrate was more than 30 times higher than that of the calibration plot with PAPG. However, because of the background substrate response, the limit of detection (calculated as 3σb/m) achieved with PG, 1.2 × 10-3 unit mL-1, was only ∼6 times better than that obtained using PAPG, 6.7 × 10-3 unit mL-1. Nevertheless, this detection limit is comparable to that obtained by chemiluminometric methods, which have been claimed to be the most sensitive for β-galactosidase detection,13,14 Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

8117

Figure 3. (a) Influence of substrate concentration on the tyrosinase biosensor amperometric response to galactosidase. Addition of PG (0); response to galactosidase (0.025 unit mL-) after PG addition (9); addition of PAPG (4); response to galactosidase (0.825 unit mL-1) after PAPG addition (2). 5.0 mL of 0.05 M phosphate buffer, pH 6.5. Applied potential, -0.15 V. (b) Influence of pH on the tyrosinase biosensor amperometric response to galactosidase. Addition of 1.2 × 10-3 mol L-1 PG (0); response to galactosidase (0.025 unit mL-1) after PG addition (9); Addition of 4.0 × 10-4 mol L-1 PAPG (4); response to galactosidase (0.825 unit mL-1) after PAPG addition (2). Other conditions as in (a).

Figure 4. Calibration plots for β-galactosidase at a graphiteTeflon-tyrosinase biosensor; 5.0 mL of 0.05 mol L-1 phosphate buffer of pH 4.0 or 6.5 when 1.2 × 10-3 mol L-1 PG ([) or 4.0 × 10-4 mol L-1 PAPG (0) used as the substrates, respectively. Eapp ) -0.15 V. Inset: linear portion of the calibration plot using PG.

and much better than the limits of detection provided by other amperometric, colorimetric, and fluorometric methods.7-12 This high sensitivity must be attributed to the amplification of the amperometric response by substrate recycling at the tyrosinase electrode. The repeatability of the β-galactosidase measurements was evaluated by performing sets of measurements from 10 8118 Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

different β-galactosidase solutions using both PG and PAPG as the substrates. Two different enzyme concentrations were checked in each case: one in the upper part of the linear calibration plots (0.025 unit mL-1 for PG and 0.825 unit mL-1 for PAPG) and another close to the calculated limit of detection (1.5 × 10-3 and 8.5 × 10-3 unit mL-1, respectively). The RSD values respectively obtained for the amperometric currents were 4.8 and 4.1% for PG and PAPG at the higher concentration levels and 8.1 and 7.3% close to the detection limit. It must be noticed that no cleaning or regeneration of the tyrosinase electrodes was needed between these measurements. Detection of E. coli. Under the optimized conditions for β-galactosidase detection (PG 1.2 × 10-3 mol L-1; pH 4.0) the capability of the sensor for the detection of E. coli was evaluated. Different bacterial solutions were prepared, from a culture that reached the stationary phase, in 5.0 mL of phosphate buffer containing polymyxin B sulfate to allow permeabilization of the bacteria membrane, as it is described in the Experimental Section. After stabilization of the background current, the optimized amount of PG was added to the electrochemical cell and the current at the bioelectrode recorded. The analytical response measured after 10 min of current recording was significantly different from the blank solution when the concentration of bacteria was equal to or larger than 1 × 106 cfu mL-1, and the amperometric response showed an increase when the bacteria concentration was increased (see below). These results were very promising, taking into account that neither preconcentration nor pre-enrichment steps were employed. In fact, and as a comparison, the sensitive chemiluminometric method for galactosidase reported by Van Poucke and Nelis13 provided a detection limit of 105 cfu E. coli mL-1, for bacteria solution whose galactosidase activity had been induced, as they were prepared from overnight Luria broth cultures containing IPTG. Consequently, once the galactosidase detection was optimized, the processes involving galactosidase induction during incubation and membrane permeabilization were evaluated in order to improve the analytical characteristics for bacteria detection. Optimization of Galactosidase Induction during Bacteria Enrichment. When bacteria were cultivated in a medium containing a galactosidase substrate, the enzyme activity was markedly increased.10 Then, two different galactosidase substrates for the enzyme activity induction were evaluated: IPTG and lactose. The hydrolysis of these substrates does not release any phenolic compound, and therefore, the possibility of their interference in the measurement of bacteria concentration is avoided. Three different 1 × 106 E. coli cultures were prepared in 100 mL of LB medium containing no inductor, 2.5 mM lactose, and 2.5 mM IPTG, respectively. After 3 h of incubation at 37 °C, the cultures were filtrated through a 0.45-µm pore size filter, and bacteria were recovered in 5 mL of phosphate buffer containing 10 µg mL-1 polymyxin B sulfate, which were placed in the electrochemical cell for the amperometric measurements. The obtained results are displayed in Figure 5, showing the importance of the inducer presence during cultivation and that IPTG was the substrate with higher induction ability. The effect of IPTG concentration in the culture medium on the amperometric signal was then checked by culturing 1 × 106 cfu mL-1 in 100 mL of LB medium containing different amounts

Figure 5. Amperometric responses to the addition of PG 1.2 × 10-3 mol L-1 to the electrochemical cell, consisting of 5.0 mL of 0.05 mol L-1 phosphate buffer, pH 4.0. containing 10 µg mL-1 polymyxin B sulfate and the cultures filtrates. Bacteria cultivated for 3 h in 100 mL of LB medium.

of IPTG during 3 h. Similar responses were obtained in the 0.2525 mM concentration range, the lowest concentration tested being employed in further work. Optimization of the Permeabilization Step. Permeabilization of the cellular membrane is a critical step for bacteria detection because they show relative impermeability for the substrate and partial intracellular retention of the enzymatic cleavage product.13 Although it is possible to find several different methods for membrane permeabilization in the literature,7 we decided to use polymyxin B sulfate as the permeabilizer. Polymyxin B is an antibiotic that disrupts the outer and cytoplasmic membranes of Gram-negative bacteria.23 Permeabilization of E. coli was carried out using the combined action of polymyxin B and lysozime.13 With this combination, 80% of the enzyme was released from the inner part of bacteria, but no further benefit was obtained for galactosidase detection in comparison to polymyxin B alone, suggesting a free access of the substrate. We also evaluated the effect of the introduction of SDS in the culture media, as this surfactant, besides providing selectivity for coliform bacteria, appeared to enhance activity by improving the transfer of the substrate, enzyme, or both across the outer membrane.12 Additionally, the results obtained with polymyxin B or SDS were also compared with those obtained when permeabilization was accomplished by sonic disruption7 (Figure 6). As can be deduced, an improvement due to sonication was only significant when no permeabilizer was used. Furthermore, this improvement was not as noticeable as those obtained by using SDS or polymyxin B. In addition, it is worth mentioning that sonication delayed the whole analytical procedure by 10 min. As can be observed, the combination of polymyxin B and SDS yielded the most sensitive detection. The effect of SDS can improve the response due to both effects: increasing permeabilization and allowing a better growth of coliforms during cultivation. Nevertheless, as the error bars displayed in Figure 6 point out, the responses obtained in the presence of SDS were less reproducible. This irreproducibility might be attributable to a nonreproducible partial loss of galactosidase during the filtration step, since some of the enzyme molecules might have migrated out from the bacteria by the SDS action. On the other hand, due to foam (23) Teuber, M. Arch. Mikrobiol. 1970, 73, 61-64. (24) Drinking water Directives 80/778/EEC and 98/83/EEC.

Figure 6. Amperometric responses (n ) 5) obtained with a graphite-Teflon-tyrosinase electrode after the addition of 1.2 × 10-3 mol L-1PG when bacteria are filtrated and recovered with 5.0 mL of 0.05 mol L-1 phosphate buffer solution containing no polymyxin B sulfate (a, b) and containing 10 µg mL-1 polymyxin B sulfate (c, d). In all cases, bacteria were grown in LB media containing 0.25 mM IPTG. In (b, d) the LB media also contained 1 mM SDS. Eapp ) -0.15 V.

Figure 7. Amperometric responses obtained at a graphite-Teflontyrosinase electrode for 1.2 × 10-3 mol L-1 PG in 5.0 mL of 0.05 mol L-1 phosphate buffer containing 10 µg mL-1 polymyxin B and the bacteria recovered after filtration. Eapp ) -0.15 V.

formation when SDS was in the culture medium, filtration was slower and more difficult to implement. All these reasons led us to use only polymyxin B as the permeabilizer agent. The concentration of polymyxin B used (10 µg mL-1) was high enough to permit the disruption of the outer and cytoplasmic membranes of E. coli, as reported previously.13 Optimized Detection of E. coli. Once all the steps involving the analytical methodology (pre-enrichment, permeabilization, detection) were optimized, different bacterial concentrations were assayed after different periods of cultivation (Figure 7). As can be observed, it is possible to detect bacteria concentrations as low as 10 cfu mL-1 after 5 h of enrichment. Furthermore, high E. coli concentrations and long incubation times produced a leveling off of the corresponding plots, indicating tyrosinase or galactosidase saturation by the substrates. To compare the performance of the method, we have summarized in Table 1 the analytical characteristics of relevant previously reported methods for the detection of E. coli. As can be seen, the detection limit of 10 cfu mL-1 is similar to those reported in previous methods with amperometric detection8-10 but the assay time is considerably shorter (at least 2 h less) using our approach. When compared with data obtained with the most sensitive methods reported using Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

8119

Table 1. Characteristics of Other Galactosidase Detection Methods galactosidase detection

limit of detection

assay time (h)

method steps

ref

colorimetry substrate: 2-nitrophenyl galactopyranoside

not specified

∼8

culture; permeabilization (2 min); incubation with substrate (10 min); colorimetric detection

7

amperometry substrate: 4-aminophenyl galactopyranosid

4.5 cfu/mL

∼8

filtration; culture (7.3 h); flow injection amperometric detection

8

amperometry: using a biosensor with cellobiose dehydrogenase

2.9 cfu/mL

∼11

filtration; culture (10.5 h); flow injection amperometric detection

9

amperometry substrate: 4-aminophenyl galactopyranoside

10 cfu/mL

∼7

culture (6 h); filtration; permeabilization (45 min); amperometric detection (∼10 min)

10

bioluminiscence substrate: 4-methylumbelliferyl galactopyranoside (LuGal) coupled reactions: LuGal f galactosidase f luciferin + galactose luciferin + ATP + O2 f luciferase(Mg2+) f oxyluciferin + AMP + PPi + CO2 + hν

no quantification: positive/absent test

∼7

culture (6 h); incubation with LuGal (1 h); luminometric detection (10 min)

11

fluorescence substrate: 4-methylumbelliferone galactopyranoside

1 cfu/100 mL

6

filtration; culture on agar plates (6 h); manual counting under a UV lamp

12

chemiluminiscence substrate: 3-(4-methoxyspiro[1,2dioxetane-3,2’-tricyclo(3.3.1.1)decan]-yl)phenyl galactopyranoside

1 cfu/100 mL

7-10

filtration; culture (6-9 h); incubation with substrate and permeabilizer (45 min); chemiluminometric detection (10 s)

13

luminiscence: solid-phase cytometer substrate: several fluorogenic and chemiluminogenic

1 cfu/100 mL

4

filtration; culture on cellulose pad (3 h); transference to a second cellulose pad and incubation for 30 min at 0 °C; transference to a black nitrocellulose filter and solid-phase cytometric detection

14

fluorescence and chemiluminescence detection,12-14 it can be seen that the assay times are in some cases longer. In addition, the methodology using the tyrosinase biosensor is of a high simplicity when it is compared to those described in refs 12 and 14 and allows the possibility of a real automation implementation. Obviously, the setup used is simple and inexpensive as corresponds to amperometric detection, and the measurements can be performed in the bacteria medium even when it is cloudy. On the other hand, the repeatability of the E. coli measurements was checked for an initial concentration of 105 cfu mL-1, which was subjected to the enrichment procedure for 4 h. The RSD value for the amperometric currents (n ) 5) was 1.1%, indicating an excellent repeatability under the mentioned conditions. Concerning the reusability of the biosensor, one of the most claimed advantages of the use of composite bioelectrodes is the possibility of obtaining a fresh electrode surface simply by polishing.21 So, the tyrosinase biosensor was polished for ∼5 s on a 150-grit SiC paper at the beginning of each working day, and no cleaning or pretreatment procedure was then necessary between measurements during the whole working day. When not in use, the biosensors were stored at 4 °C in a refrigerator. Furthermore, the detection of bacteria did not affect the useful lifetime of one single biosensor, which was the same as that the 8120 Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

Figure 8. Amperometric response obtained with the graphiteTeflon-tyrosinase biosensor after the addition of 1.2 × 10-3 mol L-1PG to the electrochemical cell containing 5.0 mL of 0.05 mol L-1 phosphate buffer, pH 4.0, 10 µg mL-1 polymyxin B, and bacteria recovered from a water sample containing 1 cfu/100 mL. The sample was cultured in 50 mL of LB medium containing 0.25 mM IPTG.

reported before when the bioelectrode was used for the detection of phenolic compounds (1 month). The possibility of interference in the detection of E. coli was evaluated by following different approaches. First, the same procedure used for the detection of E. coli was applied to a culture

medium without bacteria. No significant difference with respect to the blank signal was observed after the addition of the substrate, indicating no interference from the culture medium components. Second, the same procedure was applied to an E. coli culture medium. This medium was added to the electrochemical cell containing 5.0 mL of phosphate buffer of pH 4.0 containing no PG, and no amperometric response was obtained in this case. Concerning the specificity of the enzyme used for coliform assay, interference from non-coliform bacteria is low and insignificant, since the β-galactosidase activity of these bacteria is at least 2 log units below that of induced coliforms, affecting the results only when bacteria are present in high concentrations.10 Detection of E. coli in Drinking Water. Despite the excellent analytical performance of the method for the quantification of E. coli using tyrosinase composite biosensors, the detection limit of E. coli at the required levels for drinking water quality assessment,23 which regulates that the water must be free of bacteria (0 cfu/100 mL on the 95% of the analyses), makes necessary the detection of 1 single bacteria in 100 mL. Therefore, to get this achievement in the shortest time of analysis, we modified the method by introducing a preconcentration step prior to the sample enrichment. A liter of water containing 10 cfu, i.e., a 1 cfu/100 mL concentration, was filtered using a 45-µm pore size filter. Then, the filter was placed in a culture flask containing 50 mL of LB and 0.25 mM IPTG. Bacteria were cultivated for 6 h, and after this step, the culture was filtered again and recovered in the phosphate buffer working medium containing polymyxin B sulfate. Figure 8 shows the amperometric response obtained when compared with that of a blank water sample with no bacteria treated following the same procedure.

As can be observed, the water sample containing E. coli yielded a response perfectly distinguishable from that of the blank sample, thus allowing detection of these bacteria at the required level for the quality assessment of drinking water. It is also important to remark that this detection can be made in a total time of analysis shorter than 6.5 h, which is similar to the best methodologies reported previously. CONCLUSIONS A sensitive, simple, and in-a-day electrochemical method for the detection of coliforms in water has been developed making use of the capabilities of a composite tyrosinase biosensor as the amperometric detection system. The substrate recycling at the tyrosinase biosensor, together with an appropriate optimization of the β-galactosidase detection and induction and membrane permeabilization processes, allows the required sensitivity for coliform detection in drinking water. The overall time of analysis of less than 6.5 h and the procedure involved compares advantageously in terms of simplicity, capability of automation, and cost with other methods reported in the literature. ACKNOWLEDGMENT We are grateful for the support of the Ministerio de Ciencia y Tecnologı´a (Project BQU2003-00365). J.Z. thanks the Ministerio de Educacio´n, Cultura y Deporte for a fellowship.

Received for review July 26, 2005. Accepted October 12, 2005. AC051327R

Analytical Chemistry, Vol. 77, No. 24, December 15, 2005

8121