APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Mar. 1979, p. 521-526 0099-2240/79/03-0521/06$02.00/0

Vol. 37, No. 3

Automated Electrical Impedance Technique for Rapid Enumeration of Fecal Coliforms in Effluents from Sewage Treatment Plants MELVIN P. SILVERMAN* AND ELAINE F. MUNOZ Extraterrestrial Research Division, National Aeronautics and Space Administration, Ames Research Center, Moffett Field, California 94035 Received for publication 18 December 1978

Fecal coliforms growing in a selective lactose-based broth medium at 44.5°C generate a change in the electrical impendance of the culture relative to a sterile control when populations reach 106 to 107 per ml. The ratio of these changes was measured automatically, and the data were processed by computer. A linear relation was found between the logio of the number of fecal coliforms in an inoculum and the time required for an electrical impedance ratio signal to be detected. Pure culture inocula consisting of 100 fecal coliforms in log phase or stationary phase were detected in 6.5 and 7.7 h, respectively. Standard curves of logio fecal coliforms in wastewater inocula versus detection time, based on samples collected at a sewage treatment plant over a 4-month period, were found to vary from one another with time. Nevertheless, detection times were rapid and ranged from 5.8 to 7.9 h for 200 fecal coliforms to 8.7 to 11.4 h for 1 fecal colifonn. Variations in detection times for a given number of fecal coliforms were also found among sewage treatment plants. A strategy is proposed which takes these variations into account and allows for rapid, automated enumeration of fecal coliforms in wastewater by the electrical impedance ratio technique.

The most widely used methods for enumerating fecal coliforms are the standard most-probable-number (MPN) and membrane filter techniques (1). They are time consuming, however, requiring from 24 to 72 h to complete. In recent years, emphasis has been placed on reducing the time required for fecal coliform detection and enumeration to improve the efficiency of sewage treatment plant operations and to better assess the quality of water in routine as well as emergency situations. As a result, number of techniques that yield results in less than 24 h have been developed, some of which can be automated. Bachrach and Bachrach (2) proposed a radiometric method for detecting coliforms based on the release of 14C02 from labeled lactose. Trinel and Leclerc (16), using an autoanalyzer, developed an automated glutamate decarboxylase assay for detecting Escherichia coli. This method was extended to the enumeration of coliforms in milk (8) and water and sewage plant effluents (9). Coliforms have been detected rapidly by sensing gas pressure changes over cultures (18) or by electrochemical measurements of hydrogen production (19, 20); the latter procedure has been automated (21). Warren et al. (17) developed a rapid ,B-galactosidase assay for fecal coliforms in water. A rapid presumptive test for

coliforms has been proposed based on gas chromatographic detection of ethanol (11). Munoz and Silverman (10) developed a rapid, singlestep MPN procedure for enumerating fecal coliforms in wastewater based on electrical impedance measurements. In this paper we present the results of our studies on an automated electrical impedance ratio technique for the rapid enumeration of fecal coliforms in sewage treatment plant effluents. Changes in the electrical impedance of a culture medium inoculated with fecal coliforms are compared with the electrical impedance changes of uninoculated controls. The time at which the ratio of these changes can be detected was found to be related inversely to the logio of the number of fecal coliforms in the inoculum. A single fecal coliform in an inoculum of sewage treatment plant wastewater effluent can be detected in less than 12 h, and 200 fecal coliforms can be detected in less than 8 h by this auto-

j

mated electrical impedance technique.

MATERIALS AND METHODS Instrumentation and data processing. Imped-

ance due to growth was detected with a Bactometer model 32 (Bactomatic, Inc., Palo Alto, Calif.). The instrument measures the increase with time in the

521

522

APPL. ENVIRON. MICROBIOL.

SILVERMAN AND MUNOZ

electrical impedance ratio, Rz, between an inoculated Experimental procedures. Chlorinated and unsample vial and an uninoculated sterile reference vial chlorinated final effluent samples were collected at according to the relation Rz = Zref/(Zref + Zsample), two local sewage treatment plants. Chlorinated efwhere Zref and Zsampie are the impedance of the refer- fluent samples were dechlorinated at the sewage treatence and sample vials, respectively. In theory, the ment plant by using 2.0 ml of 0.025 N sodium thiosulimpedance ratio at zero time should be 0.5000 with fate per liter of effluent. All effluent samples were identical solutions in the sample and reference vials. received at our laboratory within 30 min of collection, In practice, initial impedance ratios ranged between and experimental procedures were begun immediately 0.4500 and 0.5500. More detailed descriptions of the thereafter. Bactometer and the theory and practice of impedance For the experiment illustrated in Fig. 1, the followratio measurements of microbial growth and metabo- ing procedures were used. E. coli was grown in modilism have been published (4, 5, 7). fied medium at 35°C for 24 h, diluted in the same The Bactometer 32 accepts up to 32 sample/refer- medium, and 2.0 ml of the 10' dilution were inocuence pairs, reads the impedance ratio of a single pair lated into 18 ml of modified medium in each of 26 in 3 s, then indexes to the next pair. A complete cycle sample vials. The reference vials contained 20 ml of of 32 sample/reference pairs is read every 96 s. The modified medium. Sample and reference vials were impedance ratio output for each pair is recorded on connected to the Bactometer and incubated at 44.50C. individual channels of a 32-channel strip chart re- Individual sample vials were removed at appropriate corder. To facilitate data handling, an interface was intervals for surface plate counts, using modified medesigned which also permits the output to be stored dium as diluent, Trypticase soy agar (BBL) plates, on a Kennedy model 1610 magnetic tape recorder for and an incubation temperature of 350C. subsequent processing by an IBM 360 computer. For the experiments illustrated in Fig. 2, the fecal The computer was programmed to recognize and coliform pure culture was grown in modified medium print out the detection time for fecal coliforms for at 44.5°C and log-phase and stationary-phase cultures each sample/reference pair according to the following criteria. First, all impedance ratio increases during the first 3 h of incubation were ignored to allow for temperature and other equilibria to occur and to avoid 9.5 false positives. Thereafter, the time at which the 10 impedance ratio began to increase at a rate of 0.0039 9.0 h-1 or greater and continued for at least the next 1.5 h was recognized as the detection time. 8.5 -9 Media and organisms. The medium used for the automated electrical impedance technique was a mod8.0ification of the fecal coliform medium of Reasoner et 8 al. (12). It contained (grams per liter) proteose peptone 0 7.5no. 3 (Difco Laboratories), 5.0; yeast extract (Baltimore Biological Laboratory [BBL]), 3.0; lactose (Difco), 10.0; NaCl, 7.5; sodium lauryl sulfate (Mathe7.0 son Coleman & Bell [MCB]), 0.2; and sodium deoxycholate (MCB), 0.1. The pH was adjusted to 6.5 with 6.5-6 NaOH. The medium was dispensed (20 ml single strength or 10 ml double strength) into 40-ml screw O 6.0capped vials fitted with stainless-steel wire electrodes (Bactomatic, Inc., Palo Alto, ?alif.) and sterilized by autoclaving at 15 lb/in2 for 'no longer than 5 min. Unless otherwise noted, 1.0-ml or 10-ml inocula were added to 20 ml of medium or 10 ml of double-strength medium, respectively. A total final volume of ca. 20 ml per 40-ml vial was selected, because preliminary experiments showed that smaller volumes did not permit adequate electrode immersion, and larger volumes lengthened the time required to detect an electrical impedance ratio response. A pure culture of E. coli was isolated from San Francisco Bay water adjacent to the outlet of a local sewage treatment plant. It was identified as E. coli by its typical colonial morphology on Endo agar (Difco) 2 4 6 12 and by its biochemical responses using the API 20E 8 10 HOURS system (Analytab Products, Plainview, N.Y.). A pure culture of a fecal coliform was isolated from a local FIG. 1. Relation between the growth of E. coli in sewage treatment plant effluent and identified as such modified medium and changes in the electrical by its ability to grow in the modified medium at 44.50C impedance ratio. Symbols: ®D, viable count; El, elecand by its ability to grow and produce gas at 44.50C in trical impedance ratio. One A impedance ratio unit EC medium (1). equals a change in impedance ratio of 0.0039.

ss

A 1~ ~ ~ ~ ~ ~ 1

VOL. 37, 1979

RAPID ENUMERATION OF FECAL COLIFORMS

4-

0

4

7 6 5 MEAN DETECTION TIME, hours

8

9

FIG. 2. Detection times for fecal coliforms by automated electrical impedance ratio technique. Symbols: (), log phase cells; A, stationary-phase cells. Each point is the mean of four replicates ± one standard deviation. were diluted serially in modified medium. Individual dilutions were inoculated into each of four replicate sample vials containing sufficient modified medium so that the final volume was 20 ml. The reference vials, containing 20 ml of modified medium, and sample vials were connected to the Bactometer and incubated at 44.5°C. The number of fecal coliforms in the inoculum was determined by the standard MPN multiple-tube technique for fecal coliforms, using phosphatebuffered water as diluent, lauryl tryptose broth, and EC medium as described in Standard Methods (1). In the experiments illustrated in Fig. 3 and 4, unchlorinated effluent samples were split into two equal portions. One portion was first filtered through Whatman no. 42 filter paper, then filtered through a 0.20,um pore size sterile Nalge filter unit and reserved for use as a diluent. Serial dilutions of the unfiltered portion were made in filter-sterilized effluent. Each serial dilution was inoculated into each of four sample vials of modified medium (final volume, 20 ml). The corresponding reference vials received an equivalent volume of filter-sterilized effluent. Sample and reference vials were connected to the Bactometer and incubated at 44.50C. The number of fecal coliforms in the effluent was determined by the standard MPN

technique (1). To achieve the desired range of fecal coliforms in a chlorinated-dechlorinated matrix for the experiments related to Fig. 5, it was necessary to use unchlorinated effluents diluted 1,000- to 10,000-fold with chlorinateddechlorinated effluent because of the low number of fecal coliforms in chlorinated effluent alone. Thereafter, all procedures for these experiments were as given for the experiments related to Fig. 3 and 4. The Bactometer operates at a fixed frequency of either 2 kHz or 400 Hz. All experiments were run at 2 kHz, except for the experiments illustrated in Fig. 5,

523

impedance ratio can be detected. Thereafter, the impedance ratio increased rapidly and was linear, extending well into the stationary phase of growth before it began to decrease. The total impedance ratio increase consistently reached 0.0390 or greater over the complete growth cycle. These findings were reproducible and aided in defining a detection time as outlined in Materials and Methods. This definition was included in our computer program and was applied to all subsequent experiments. The relation between the detection time and the number of fecal coliforms in an inoculum was examined. Both log-phase and stationaryphase pure culture inocula were tested. Figure 2 shows that a linear relation exists in which the detection time is inversely proportional to the 5 r-

Y= z -i

-

0.81x+8.08

3

J

0 (0 4

2

RI

a

0

O

I

I

5

6

7 8 9 DETECTION TIME, hours

10

11

FIG. 3. Detection times for fecal coliforms in unchlorinated final effluent from sewage treatment plant A. Each point is the mean of four replicates ± one standard deviation. 10

r

CURVE

DATE

1

DEC. 8 DEC. 21 JAN. 4

2

z

3 4 5 6 7

JAN. 7 JAN. 12 MAR. 30 APR. 20

cn 6 r x

0

0

u4 -. u0 I4.5 )

O '

which were run at 400 Hz.

RESULTS The results of a typical experiment on the ability of the impedance ratio technique to detect the growth of E. coli are illustrated in Fig. 1. It is apparent that populations must grow to between 106 and 107 per ml before an increase in

12

DETECTION TIME, hours

FIG. 4. Standard curves of detection times for fecal coliforms in unchlorinated final effluent from sewage treatment plant A. Linear correlation coefficients ranged from 0.957 to 0.993.

524

APPL. ENVIRON. MICROBIOL.

SILVERMAN AND MUNOZ

9 8

z

7

L.

z

(a: 2

0

a LJ

SL

0

2

13 O,

Y= -1.OOX +7.56

1 0

1

2

3

7 6 5 4 DETECTION TIME, hours

8

9

10

FIG. 5. Composite standard curve of detection times for fecal coliforms in chlorinated-dechlorinated final effluent from sewage treatment plant B. Each point is the mean of four replicates ± one standard deviation.

log1o of the number of fecal coliforms in the inoculum. As expected, inocula taken from log phase cultures were detected earlier than were inocula taken from stationary-phase cultures. For example, Fig. 2 shows that an inoculum containing 100 log-phase fecal coliforms was detected in approximately 6.5 h as compared to approximately 7.7 h for a stationary-phase inoculum of the same size. This 1.2-h difference in detection time was consistent throughout the range of inoculum sizes tested. Studies were extended to unchlorinated sewage treatment plant final effluents to determine whether there was a similar relation between the number of fecal coliforms and detection time to that found with a fecal coliform in pure culture. Figure 3 shows that this was indeed the case; there was an inverse relation between the detection time and the logio of the number of fecal coliforms in unchlorinated sewage treatment plant final effluent. The detection time for 100 fecal coliforms was 7.5 h (Fig. 3), which agreed closely with a detection time of 7.7 h for the same number of stationary-phase pure-cultured fecal coliforms (Fig. 2). These results suggested the feasibility of constructing a standard curve relating detection time to the number of fecal coliforms in a wastewater inoculum. Figure 4 shows a series of seven standard curves constructed over a 4-month period, using dilutions of unchlorinated final effluent as inocula. Standard curves 3, 4, and 5 were constructed within an 8-day period and agreed very closely with one another. Curves 1,

2, 6, and 7, constructed several weeks earlier or 2 to 3 months later, diverged considerably from the three January curves, suggesting that any given standard curve, at least in this sewage treatment plant, would have a finite useful life. Nevertheless, the seven curves in Fig. 4 show that 200 fecal coliforms can be detected by the automated impedance ratio technique in 5.8 to 7.9 h, and one fecal coliform in 8.7 to 11.4 h. A different approach was taken at a second sewage treatment plant in an attempt to assess the utility and longevity of standard curves. Instead of constructing a series of standard curves over time, a single composite standard curve was constructed over a 1-month period, using unchlorinated final effluent diluted in a chlorinated-dechlorinated final effluent matrix (Fig. 5). Each point represents the results obtained on a single day with a single dilution. The inoculum (point) with the highest number of fecal coliforms was tested on November 10 and the lowest, on December 12. The largest discrepancy in the number of fecal coliforms (i.e., any single point) from the least squares line was approximately 0.5 log units. This suggests that a standard curve might have a useful life of approximately one month at this sewage treatment plant. The composite curve in Fig. 5, run at a Bactometer frequency of 400 Hz, shows that 100 fecal coliforms per ml were detected in 5.5 h and one fecal coliforn per ml in 7.4 h. A parallel experiment run at 2 kHz resulted in a similar composite standard curve except that detection times were approximately 30 min later (not illustrated).

DISCUSSION It appears feasible to automate the rapid detection and enumeration of fecal coliforms by the electrical impedance ratio technique as shown by the results of our experiments with pure cultures (Fig. 2) and wastewater from sewage treatment plants (Fig. 3, 4, and 5). Although detection times for a given number of fecal coliforms may vary by 2 to 3 h in effluents from a particular sewage treatment plant sampled at different times of the year, the longest detection time for one fecal coliform remained less than 12 h (Fig. 4). Many reasons probably exist for the divergence of the standard curves in Fig. 4. Factors over which one has no control, such as weather, temperature, rainfall, fluctuations in the volume and quality of influent, variations in residence time of sewage in the various sewage treatment processes, strain differences among fecal coliform populations and variations in the stresses to which they are subjected, etc., will affect the detection time to varying degrees. By its very

VOL. 37, 1979

RAPID ENUMERATION OF FECAL COLIFORMS

nature, the present procedure of relating fecal coliform numbers determined by the MPN method to detection time by impedance ratio measurements is subject to some uncertainty because a correlation is being made between a procedure that is relatively independent of time (MPN) and one that is strictly time dependent (impedance). Those factors that increase or decrease the lag phase or growth rate of fecal coliform populations are bound to have a more direct effect on detection time than on MPN. One such factor may be differences in the inherent growth rates of different fecal coliform strains that enter the sewage treatment plant at different times. Added to this are the different stresses to which fecal coliforms are subjected during sewage treatment and their relative ability to recover in the different media used in the MPN procedure and the impedance ratio measurements. It has been demonstrated that stressed coliforms show greater recovery if incubated in a favorable medium at lower temperatures before being subjected to more rigorous selective media and temperatures (3, 6, 14, 15). In our procedure for constructing standard curves, the same population of stressed fecal coliforms serves as the inoculum for the standard MPN and impedance ratio measurements. In the MPN procedure, stressed organisms are afforded an opportunity to recover in lauryl tryptose broth at 35°C before being subjected to the more rigorous demands of growth at 44.5°C in EC medium. By contrast, the determination of detection time by impedance ratio response requires these same stressed organisms to recover and grow in the selective modified fecal coliform medium of Reasoner et al. (12) at 44.50C without a preliminary opportunity to recover under more favorable conditions. It is not surprising, therefore, that detection times for a given number of fecal coliforms in sewage treatment plant effluents vary among samples taken at different times, and it seems unreasonable to assume that a single standard curve could be constructed for use over long periods of time in any sewage treatment plant. It will likely be necessary to construct new standard curves periodically; perhaps at weekly intervals in some sewage treatment plants, (Fig. 4), or at longer intervals in other sewage plants (Fig. 5). Every sewage treatment plant is probably unique in this respect, and the useful life of a standard curve will have to be determined empirically for each facility. Roesler et al. (13), in reviewing the current status of automation of wastewater treatment in the United States, identified a number of problem areas in need of research, including the lack of suitable sensors for biological indicators. Such

525

sensors could improve the efficiency and reduce the cost of disinfection by providing positive feedback for disinfection control. They would be especially useful where disinfectants other than chlorine are employed (ultraviolet irradiation, ozone, etc.) and residual concentrations of disinfectants are not measurable (13). The electrical impedance ratio technique meets the criteria for a biological indicator sensor in that it enumerates fecal coliforms throughout a broad range in minimal time, and is amenable to automation with computer control and data reduction. It should merit serious consideration as a sensor system for use in disinfection control of wastewater and in monitoring the quality of the receiving waters. ACKNOWLEDGMENTS We thank D. DeLeuran and N. Arango for excellent technical assistance, W. Gore for the design and fabrication of the interface between the Bactometer and the tape recorder, C. H. Schulbach for computer programming, and personnel at the local sewage treatment plants (E. Becker, V. Alford, H. Sanders, E. Jacobi, R. Nice, H. Hollstein, L. Pasqual) for their cooperation and assistance in obtaining effluent samples. LITERATURE CITED American Public Health Association. 1971. Standard 1. methods for examination of water and wastewater, 13th ed. American Public Health Association, Inc., New York. 2. Bachrach, U., and Z. Bachrach. 1974. Radiometric method for the detection of coliform organisms in water. Appl. Microbiol. 28:169-171. 3. Bissonnette, G. K., J. J. Jezeski, G. A. McFeters, and D. G. Stuart. 1977. Evaluation of recovery methods to detect coliforms in water. Appl. Environ. Microbiol. 33: 590-595. 4. Cady, P. 1975. Rapid automated bacterial identification by impedance measurements, p. 73-99. In C. G. Heden and T. Illeni (ed.), New approaches to the identification of microorganisms. John Wiley & Sons, Inc., New York. 5. Cady, P. 1978. Progress in impedance measurements in microbiology, p. 199-239. In A. N. Sharp and P. S. Clarke (ed.), Mechanizing microbiology, Charles C Thomas, Publisher, Springfield. Ill. 6. Green, B. L., E. M. Clausen, and W. LUtsky. 1977. Twotemperature membrane filter method for enumerating fecal coliform bacteria from chlorinated effluents. Appl. Environ. Microbiol. 33:1259-1264. 7. Hadley, W. K., and G. Senyk. 1975. Early detection of microbial metabolism and growth by measurement of electrical impedance, p. 12-21. In D. Schlessinger (ed.), Microbiology-1975. American Society for Microbiology, Washington, D.C. 8. Moran, J. W., and L. D. Witter. 1976. An automated rapid test for Escherichia coli in milk. J. Food Sci. 41: 165-167. 9. Moran, J. W., and L. D. Witter. 1976. An automated rapid method for measuring fecal pollution. Water Sewage Works 123:66-67. 10. Munoz, E. F., and M. P. Silverman. 1979. Rapid, singlestep most-probable-number method for enumerating fecal coliforms in effluents from sewage treatment plants. Appl. Environ. Microbiol. 37:527-530. 11. Newman, J. S., and R. T. O'Brien. 1975. Gas chromatographic presumptive test for coliform bacteria in water. Appl. Microbiol. 30:584-588. 12. Reasoner, D. J., J. C. Blannon, and E. E. Geldreich.

526

13.

14.

15.

16.

SILVERMAN AND MUNOZ

1976. The seven-hour fecal coliform test. Technol. Conf. Proc. Am. Water Works Assoc. 3B-5:1-10. Roesler, J. F., D. F. Bishop, and I. J. Kugelman. 1977. Current status of research in automation of wastewater treatment in the United States. Prog. Water Res. 9: 659-671. Rose, R. E., E. E. Geldreich, and W. Litsky. 1975. Improved membrane filter method for fecal coliform analysis. Appl. Microbiol. 29:532-536. Stuart, D. G., G. A. McFeters, and J. E. Schillinger. 1977. Membrane filter technique for the quantification of stressed fecal coliforms in the aquatic environment. Appl. Environ. Microbiol. 34:42-46. Trinel, P. A., and H. Leclerc. 1972. Automatisation de l'analyse bacteriologique de l'eau. I. Etude d'un nouveau test specifique de contamination fecale et des conditions optimales de sa mise en evidence. Water Res. 6:1445-

APPL. ENVIRON. MICROBIOL. 1458. 17. Warren, L. S., R. E. Benoit, and J. A. Jessee. 1978. Rapid enumeration of fecal coliforms in water by a colorimetric,B-galactosidase assay. Appl. Environ. Microbiol. 35:136-141. 18. Wilkins, J. R. 1974. Pressure transducer method for measuring gas production by microorganisms. Appl. Microbiol. 27:135-140. 19. Wilkins, J. R., and E. H. Boykin. 1976. Electrochemical method for early detection and monitoring of coliforms. J. Am. Water Works Assoc. 68:257-263. 20. Wilkins, J. R., G. E. Stoner, and E. H. Boykin. 1974. Microbial detection method based on sensing molecular hydrogen. Appl. Microbiol. 27:949-952. 21. Wilkins, J. R., R. N. Young, and E. H. Boykin. 1978. Multichannel electrochemical microbial detection unit. Appl. Environ. Microbiol. 35:214-215.

Automated electrical impedance technique for rapid enumeration of fecal coliforms in effluents from sewage treatment plants.

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Mar. 1979, p. 521-526 0099-2240/79/03-0521/06$02.00/0 Vol. 37, No. 3 Automated Electrical Impedance Techniqu...
878KB Sizes 0 Downloads 0 Views