Biosensors& Bioekctmu’cs 6 (1991) 575-580
Electrochemical detection of viable bacteria in urine and antibiotic selection Noriyuki
6 Tadashi Matsunaga*
Department of Biotechnology, Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan (Received 30 November 1988; revised version received 24 July 1990, accepted 18 December 1990)
Abatraet: An electrode system consisting of a basal-plane pyrolytic graphite (BPG) electrode and a porous nittocellulose membrane filter to trap bacteria was used for the detection of bacteria in urine. The peak current of a cyclic voltammogram increased with increasing initial cell concentmtion of Escherichia coli in urine. Urine containing from 5 X 102to 5 X 10r cells ml-’ was measured with this system. The susceptibility of bacteria to various antibiotics was also determined from the peak current. The minimum inhibitory concentration values obtained by the electrochemical method were in good agreement with those obtained by the conventional method. Keywords: electrochemical
detection, urine, cyclic voltammetry, antibiotics.
INTRODUCTION Rapid detection of viable bacteria and antibiotic selection are important in clinical bacteriology (e.g. for diagnosis and treatment of urinary tract infections). Methods based on colony formation have been used for the detection of viable cells. Sensitivity to antibiotics has been determined by diffusion methods using solid media or dilutions in liquid media. However, these methods are time-consuming. Various electrochemical methods have been developed for the rapid determination of viable cell numbers (Wilkins et aZ., 1974; Hadley & Senyk, 1975;Matsunaga &al., 1981).However, the numbers of viable cells were measured indirectly from bacterial metabolites or from oxygen *To whom correspondence should be addressed
consumption. Recently, viable cells have been directly detected by cyclic voltammetry with a basal-plane pyrolytic graphite electrode (Matsunaga & Namba, 1984). Classification of Gram-negative and Gram-positive bacteria was also possible by this system (Matsunaga & Nakajima, 1985). The minimum detectable cell numbers were lo7 ml-’ for yeast and 10sml-’ for bacteria. Patients with urinary tract infections typically have [email protected]
bacteria cells in 1 ml of urine. Therefore, it is necessary to increase the cell numbers to allow direct electrochemical detection of bacteria. In this paper, bacteria were added to filter urine, concentrated on a membrane filter, and cyclic voltammetry was applied to detect viable bacteria. The susceptibility of bacteria to various antibiotics was determined from the effect on the peak current of bacteria.
0956-5663/91/$03.50 (0 1991 Elsevier Science Publishers Ltd.
Biosensors& Bioelectmnics 6 (1991) 515-580
MATERIALS AND METHODS Materials
Trypticase and tryptose were purchased from Oxoid (London, UK). Yeast extract was obtained from Kyokuto Pharmaceutical Co. (Tokyo, Japan). Other reagents were commercially available analytical reagents or laboratory-grade materials. Distilled water was used in all procedures. Microorganisms
Two species of Escherichiu coli K-12 (a wild strain and an ampicillin- and tetracycline-resistant strain) were used for the experiments. Bacteria were cultured aerobically at 37°C for 12 h in 100 ml of the medium (pH 7.0) containing 1 g of trypticase, 0.3 g of tryptose, 0.5 g of yeast extract, 0.3 g of KI-12P04, 03 g of KZHPO.+ 02g of diammonium citrate, 2 g of glucose, 0.1 g of meen-80,0*02 g of cysteine hydrochloride, and 0.5 ml of a salt solution (115% MgS04 - 7Hz0, 068% FeS04 - 7H20, 2.4% MnS04 - 2Hz0, w/v). Cultured cell suspensions (10 ml) were centrifuged at 4°C for 10min at 3000g and washed with 10 ml of phosphate buffer (PH 7.0).
8’ Fig. 1. Schematic diagmm of the electrodesystemfor the electrochemicaldetection of bacteria in urine: 1, function generator; 2, potentiostat;3, X-Y reconler; 4, reference elect&e (SCE); 5, counter-electrode(platinum wire); 6, working electrode(BPG); 7, microbial cells; 8, membmne filter: 9, holder.
by immersion in a glass tube terminated by a sintered glass frit. Procedures
The electrode system is depicted in Fig. 1. The working electrode consisted of a basal-plane pyrolytic graphite (BPG) electrode (surface area, 0.19 cm2, Union Carbide Corp., New York, NY) and a nitrocellulose membrane filter (Advantec, Tokyo; O-45-pm pore size, 25 mm diameter) to retain microbial cells. Cyclic voltammograms were obtained by using a potentiostat (model HA301; Hokuto Denko), a tinction generator (model HB 104; Hokuto Denko), and an X-Y recorder (F35; Riken Denshi). After each run, the graphite electrode was polished with emery paper (No. 2000; N&ken Kogyo, Tokyo). The cell for cyclic voltammetry was of all-glass construction, approximately 25 ml in volume, and incorporated a conventional three-electrode system. The counter-electrode was a platinum wire and the reference electrode was a saturated calomel electrode (SCE). The reference electrode was separated from the main cell compartment 576
Urine samples were filtered using a nitrocellulose membrane filter (O-45Frn pore size, 25 mm diameter) immediately after they were obtained from a normal adult (a 25-year-old male). A model infective urine was prepared as follows: the cultured cells of E. coli were added to 50 ml of the filtered urine and the mixture was allowed to stand anaerobically at 37°C for 3 h. The model infective urine (50 ml) containing E. coli was added to 50 ml of the medium described above, and allowed to stand at 37°C for 3 h. Then, 50 ml of the urine-medium mixture containing E. coli cells was centrifuged for 10 min at 4°C and 3OOOg, suspended in 25 ml of 0 1 M phosphate buffer (pH 7-O),and 4 ml of the suspension was dropped onto the membrane filter. Immediately, the cells were fixed on the membrane filter by filtration using an aspirator. The number of bacteria is reproducible with a relative error of 7% in the
Biosensors & Bioelectmnics 6 (1991) 575-580
filtration step; more than 95% of microbial cells on the surface of the membrane filter were viable (Matsunaga, T. et al., unpublished). The membrane filter retaining the microbial cells was attached to the graphite electrode, which was then inserted into the reaction cell filled with 10 ml of 0 1 M phosphate buffer (pH TO), and a sweeping potential (10 mVs-‘) was applied to the electrode. The susceptibility to antibiotics was tested by addition of antibiotic solution (0.1 ml) to 100 ml of the urine-medium mixture described above, which was then allowed to stand at 37°C for 3 h. The cells were centrifuged for 10 min at 4°C and 3OOOg, and resuspended in 50 ml of 0.1 M phosphate buffer (pH 7.0). Determination of cell numbers
The viable cell number of E. coli in model infective urine was determined by plating suitably diluted samples of a culture and counting colonies on LB medium (10 g of tryptone, 5 g of yeast extract and 10 g of NaCl per litre of water; Gherna et aZ.,1989)containing 1.5% agar, after a 24 h incubation at 37°C. The number of colonies was reproducible with an average
detection of bacteria in urine
relative error of 6% when lo5 E. coli cells were employed for the experiments (Matsunaga,.T. et al., unpublished).
RESULTS Cyclic voltammogram of microbial cells in urine
Cyclic voltammograms were directly obtained from urine containing ld cells ml-’ of E. coli. However, no peak current was found. Therefore, an equal amount of the medium described in the experimental section was added to the urine for incubation of bacteria. Cyclic voltammograms were obtained after incubation of the urine mixture for l-3 h. The peak current appeared after 3 h of incubation. These results show that the cells in the urine mixture were cultured to detectable cell numbers (more than 10’ cells on the membrane filter) when they were incubated for 3 h. On the other hand, little current was obtained for the sample incubated for less than 2 h. The samples incubated in urine show little peak current. Figure 2 shows the cyclic voltammograms of E. coli in urine when initial cell concentrations of lo3 and 10’ cells ml-’ were -
Potential ( V vs SCE 1 Fig. 2. Cyclic voltammograms of E. coli in urine when Iti cells ml-’ (A) and I# cells ml-l (B) of cellswereincubated with the medium for 3 h. The scan rate was 10 m Vs-I. 577
h! Nakamum et al.
Biosensors & Bioelectmnics 6 (1991) 575-580
incubated with nutrient medium for 3 h. Anodic waves appeared at around O-7V vs SCE on the first scan in the positive direction. Upon scan reversal, no corresponding reduction peak was obtained. The peak currents were 0.2~A for lo3 cells ml-’ and 05pA for lo5 cells ml-’ respectively. Determination of cell concentration
Figure 3 shows the relationship between the peak current and the cell concentration of E. coli in the model infective urine. Urine samples containing E. coli in the range of 5 X ld-5 X 16 cells ml-’ at the initial cell concentration were employed in the experiments. The peak current rose with initial cell concentration. In increasing particular, the peak current increased sharply for the initial cell concentration of more than lo5 cells ml- ‘. The peak currents were reproducible with an average relative error of 10%. Patients with a urinary tract infection typically have 105106cells in 1 ml of the urine. Therefore, the urine of these patients could be discriminated from the peak current around 07V vs SCE after the incubation. Susceptibility of bacteria in urine to various antibiotics
Table 1 shows the peak currents of wild and resistant strains of E. coli (initial cell concentration 16 cells ml-‘) in urine, when they
TABLE 1 Peak currents of wild and resistant strains when these strains were incubated in the mixture of urine and the medium for 3 h Antibiotic
Peak current (PA) Wild strain
Ampicillin (12pg ml-‘) Tetracycline (15 c(g ml-‘)
Initial cell concentration 10s cells ml-‘.
of E. coli was adjusted to
were incubated for 3 h in the mixture of urine and the medium containing various antibiotics (ampicillin, tetracycline, gentamycin and colistin). The peak current of the wild strain was below O-2PA because its growth was inhibited by antibiotics. On the other hand, the resistant strain can be increased in the presence of antibiotics, and peak currents are over O-3PA. Therefore, the susceptibility of urinary bacteria to antibiotics can be tested from the peak current. Figure4 shows the relationship between the peak current and ampicillin concentration. The peak current obtained from bacteria was 0.2pA at an ampicillin concentration of 12.0,ug ml-‘. Peak current became constant above this concentration. By varying the antibiotic
Fig. 3. Relationship between peak current and the cell concentration of E. coli in urine when they were incubatad for 3 h with the medium. 578
I 40 ( p/ml )
Fig. 4. Relationship between peak current and ampicillin concentmtion. The peak currents were obtainedfrom cyclic voltammogmms of E. coli wild strain (Iti cells ml-!, in urine when they were incubated with the medium for 3 h.
Biosensors& Bioekctmnics 6 (1991) 575-580
Electrochemicaldetection of bacteria in urine
concentration of the medium in which the wild strain was cultured, we determined the minimum concentration which gives the lowest value and is defined as electrochemical minimum inhibitory concentration (MIC). Using this method, the sensitivity of E. coli to other antibiotics was examined and its MIC was determined. The MIC of the wild strain of E. coli was 12pg ml-‘, 38 ,ug ml-‘, 1.3 pg ml-’ and 2.5 lug ml-’ respectively for ampicillin, tetracycline, colistin and gentamycin (Table 2; Tanaka & Nakamura, 1988).The MIC values were reproducible with an average relative error of 5%.These results were in fairly good agreement with those obtained by the conventional method. The MIC value of colistin measured by the electrochemical method was lower than that obtained by the conventional colony method.
DISCUSSION Recently, viable cell numbers have been determined by cyclic voltammetry with a BPG electrode, and 10’ cells ml-’ of bacteria could be detected (Matsunaga & Namba, 1984b). When bacteria were attached to the surface of the electrode (0.19 cm2), using a membrane filter, 10’ cells were detectable. Patients with urinary tract infections usually have more than 10’ cells ml-’ bacteria in their urine, whereas the urine of a healthy person contains fewer than ld cells ml-’ bacteria. Therefore, diagnosis of urinary tract infection was impossible by cyclic voltammetry using a graphite electrode. In this paper, bacteria present in the urine were incubated with the medium to increase cell numbers to a measurable level. We have employed several media, including the nutrient medium for culture of E. coli and to TABLE 2
Sensitivityof E. coli to antibiotics Antibiotic
Minimum inhibitory concentration (Pg ml-t) Electmchemical method
Ampicillin Tetracycline Gentamycin Colistin
12.0 3.8 2.5 1.3
Colony method 6.3-13 2-s-20 0.8-25 24-5.0
enrich urine. Use of the nutrient medium has the merit of shortening the lag phase of E. coli. Therefore, the peak current obtained from the urine of a patient containing over 16 cells ml-’ was detectable, whereas a low peak current was obtained from that of a healthy person which contains fewer than ld cells ml-‘. As a result, the urine of infected and healthy people can be differentiated by the electrochemical method. Moreover, this method can be used for selection of a suitable antibiotic to cure the patient of the urinary tract infection during diagnosis. When the wild and resistant strains of E. coli were incubated for 3 h in the mixture of urine and the medium containing various antibiotics, the peak current values of the wild strain was below O-2PA, whereas those of the resistant strain were more than 0*3pA The turbidimetric method provided a relatively quick and convenient way of estimating cell concentrations, and the minimum detectable cell concentration was lo’-108 cells ml- ‘. However, viable cells and precipitates of culture could not be distinguished by the turbidimetric method (Harris & Kell, 1985). Consequently, the turbidimetric method is not applicable to diagnosis of urine infections. Recently, we reported that electron transfer from microbial cells to a graphite electrode is correlated with coenzyme A (CoA) in the cell for Saccharomyces cerwisiae,Lactobacillusacidophilus and E. coli (Matsunaga & Namba, 1984a,b). In
particular, CoA in the periplasmic space is responsible for the electrochemical reaction (Matsunaga ef al., in preparation). In this study, peak currents obtained around 0.7 V vs SCE from E. coli cells seems to be related to CoA in the cell. Automated, rapid and reliable methods of detecting bacteriuria and testing susceptibility patterns have been developed by several groups (Lamb etaZ.,1976;Colvin & Sherris, 1977;Cady et al., 1978; Strassburger & Tiller, 1984, Smith et al., 1985; Baynes et al., 1986). The impedance measurement of culture media has been proposed for this purpose (Hadley 8zSenyk, 1975; Colvin 8z Sherris, 1977; Cady et aZ., 1978). Nutrients are converted to various charged metabolites, such as organic acids and other compounds, by bacteria. As a result, the impedance of the medium decreases with increasing cultivation time. The conductance method is based on a similar principle (Smith et aZ., 1985; Baynes er aZ.,1986). It takes 6 h for the 579
N Nakamura et al.
determination of antimicrobial susceptibility. An electrochemical method based on the potentiometric determination of hydrogen molecules produced by bacteria has also been developed, for the early detection of urinary tract infections (Lamb et aZ., 1976). Bacteria in urine were detected in 3.5-9 h. However, these methods detect cell numbers indirectly from bacterial metabolites. The results obtained are sometimes not correlated with cell numbers. Our method requires a 3 h incubation and a filtration-based concentration before the assemblage of the immobilized-cell electrode. Further developmental studies in our laboratory are being directed towards automation of these processes and their application to natural urine samples.
ACKNOWLEDGEMENT This work was supported by Grant-in Aid for Scientific Research No. 59850134 from the Ministry of Science and Culture.
REFERENCES Baynes, N. C., Comrie, J. & Harper, I. A (1986). Antimicrobial susceptibility determination by an automated conductance technique. Med. Lab. ScL 43,232-40. Cady, P., Dufour, S. W., Lawless, P., Nunke, B. & Krasger, S. J. (1978). Impedimetric screening for bacteriuria. .I Clin. Mirrobiol., 7, 273-8. Colvin, H. J. & Sherris, J. C. (1977). Electrical impedance measurements in the reading and monitoring of broth dilution susceptibility tests. Antimicrob. Agenrs Chemother., 12,61-6.
Biosensorst Bioelectronics6 (1991) 575-580 Ghema, R, Pienta, P. & Cote, R (1989). In: American i’jpe Culture Collection Catalogue of Bacteria and Phages, 17th edn. ATCC, Rockville, MD, p. 367. Hadley, W. K. & Senyk, G. (1985). In: Microbiology,ed Society for D. Schlessinger. American Microbiology, Washington, DC, pp. 12-21. Harris, C. M. & Kell, D. B. (1985). The estimation of microbial biomass. Biosensors,1, 17-84. Lamb, V. A, Dalton, H. P. 8c Wilkins, J. R (1976). Electrochemical method for the early detection of urinary-tract infections. Am.L Clin. Path..66,91-5. Matsunaga, T. & Nakajima, T. (1985).Electrochemical classification of Gram-negative and Grampositive bacteria. Appl. Envinwz. Microbial., 50, 238-42. Matsunaga, T. & Namba, Y. (1984a). Detection of microbial cells by cyclic voltammetry. Anal Chem, 56,798-801. Matsunaga, T. & Namba, Y. (19843). Selective determination of microbial cells by graphite electrode modified with absorbed 4,4’-bipyridine. Anal. Chim. Acta 159,87-94. Matsunaga, T., Karube, I., Nakahara, T. & Suzuki, S. (1981).Amperometric determination of viable cell numbers based on sensing microbial respiration. Eur. J. Appl. Microbial.Biotechnol., 12, 97-101. Smith, T. K., Eggington, R, Pease, A A, Harris, D. M. & Spencer, R C. (1985).Evaluation of Malthus 128H microbiological growth analyser for detecting signiticant bacteriuria. J Clin. Path., 38, 926-8. Strassburger, J. & Tiller, F. W. (1984). Bacteriuria screening and antimicrobial susceptibility testing of aerobic bacteria by an electrochemical method. Zbl. Bakt. Hm., A256,466-74. Tanaka, N. & Nakamura, S. (1988). In: Outline of Antibiotics, 3rd edn. University of Tokyo Press, Tokyo, p. 21. Wilkins, J. R, Stoner, G. E. & Boykin, E. H. (1974). Microbial detection method based on sensing molecular hydrogen. Appl. Environ. Microbial.,27, 949-52.