APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Oct. 1982, p.

814-819

Vol.

44, No. 4

0099-2240/82/100814-06$02.OO/O Copyright 1982, American Society for Microbiology

Effect of Antecedent Growth Conditions on Sensitivity of Escherichia coli to Chlorine Dioxide

JAMES D. BERG,'* ABDUL MATIN,2 AND PAUL V. ROBERTS1 Civil Engineering Department, Terman Engineering Center,1 and Department of Medical Microbiology, Stanford University School of Medicine,2 Stanford University, Stanford, California 94305 Received 28 December 1981/Accepted 17 May 1982

Bacterial resistance to inactivation by antibacterial agents that is induced by the growth environment was studied. Escherichia coli was grown in batch culture and in a chemostat, and the following parameters were varied: type of substrate, growth rate, temperature, and cell density during growth. Low doses (0.75 mg/ liter) of chlorine dioxide were used to inactivate the cultures. The results demonstrated that populations grown under conditions that more closely approximated natural aquatic environments were more resistant than those grown under commonly employed batch culture conditions. In particular, bacteria grown at submaximal rates were more resistant than their counterparts grown at pLmax. The most resistant populations encountered in this study were those grown at D values of 0.02 h-1 and 0.06 h-1 at 25C. Growth at 15C led to greater resistance than did growth at 37C. The conditions that produced relatively resistant phenotypes were much closer to those found in most natural environments than are the typical conditions of batch culture methods. The importance of major physiological changes that can be induced by the antecedent growth environment is discussed in light of the possible modes of action of several disinfectants.
Bacteria that are resistant to inactivation by chemical disinfecting agents are commonly encountered in a diverse set of aquatic environments, including drinking water reservoirs and distribution systems (20), evaporative air conditioners (21, 22), and humidifiers used in respiratory therapy (9). The apparent resistance of the target organism has most often been attributed indirectly to protection by physical means, e.g., association with particulate matter (13) or occlusion within a biological film on a surface (6). Equally important is the genotypic provision of a protective capsule or spore, as well as abiotic factors such as extraneous chemical demand for the disinfectant. Often overlooked, however, is the resistance proffered by environmentally induced changes in phenotype that can occur during growth before the disinfection process. These phenotypic changes can be quite significant, particularly during a laboratory evaluation of a disinfectant. We have shown, for example, that a fecal isolate of Escherichia coli grown in batch culture with nutrient broth is orders of magnitude more sensitive to chlorine and chlorine dioxide than are in situ wastewater-grown coliforms (1). Other studies (4, 5, 7), including a study by Milbauer and Grossowicz in 1959 (18), also corroborate the view that antecedent growth conditions greatly influence the sensitivity of an organism to a variety of disinfectants.
Pseudomonas aeruginosa and atypical mycobacteria were shown by Carson et al. (4, 5) to be capable of growth in deionized water. The organisms grown in that environment possessed markedly greater resistance to chlorine dioxide than did those grown on Trypticase soy agar (BBL Microbiology Systems). Similarly, in a study by Favero and Drake (7), swimming pool water isolates grown in "pool water" demonstrated much greater resistance to iodine than did their counterparts grown on Trypticase soy agar. The objective of the present study was to investigate differences in sensitivity attributable to the antecedent growth environment. Particularly, we contrasted those environmental parameters which are routinely employed in laboratory studies with general ambient aquatic conditions. For example, the optimum growth temperature of 37C is often used to culture enteric bacteria in the laboratory, whereas in nature enteric bacteria are capable of growth at a temperature as low as 15C in water reservoirs (20). Also, the growth rates (,u) encountered in natural aquatic environments are often as low as 0.01 h-1, as shown by Jannasch (14) and Brock (3). We hypothesized that such differences in growth temperature and rate may affect importantly the sensitivity of organisms to disinfectants. Therefore, we compared the sensitivity to disinfection

814

VOL. 44, 1982

SENSITIVITY OF E. COLI TO CHLORINE DIOXIDE

815

of cultures grown at submaximal rates and suboptimal temperatures with that of conventional batch cultures in which lLma, was attained. In this study the chemostat was employed to attain nutrient-limited growth while another aspect of the growth environment, e.g., temperature, was manipulated separately. Manipulations such as changing the specific growth rate have been shown by Matin (16) and others (17, 19) to have important effects on the physiology of bacterial cells. These include alterations in the cell surface-to-volume ratio and in the levels of specific enzyme activity. Therefore, these and other phenotypic characteristics that may affect important target sites for disinfectants may be altered by changes in growth conditions, conferring a change in sensitivity as a function of the antecedent growth environment.
MATERIALS AND METHODS Organism. A fecal isolate of E. coli was obtained from primary effluent of the Palo Alto Wastewater Treatment Facility by being grown on EMB agar (Difco Laboratories). The isolate is a gram-negative, motile rod that ferments lactose at 44.5C and is indole and methyl red positive and Voges-Proskauer and citrate negative (IMViC [+ + - -]). Colonies produce a brilliant green metallic sheen on EMB agar. Cultures were maintained on nutrient agar (Difco) slants at 4C and subcultured bimonthly. Growth media. Two media were used throughout the study. The first was a glucose mineral salts medium with either 0.02% or 0.15% glucose, containing the following concentrations of inorganic salts (in mg/ liter): NH4Cl, 200; MgSO4 * 7H20, 100; NaCl, 6.6; and CaC12, 1.0; and (in g/liter): K2HPO4 * 3H20, 1.13 and KH2PO4, 0.88. To the above solution in deionized water was added 0.2 ml of trace salts solution per liter as described by Vishniac and Santer (23). The second medium consisted of 0.16% nutrient broth (BBL) adjusted to pH 7; deionized water was used in the preparation of this medium. Growth conditions. Populations were grown either in a chemostat (Bio Flo; New Brunswick Scientific Co.) or in batch culture. Chemostat cultures were grown under forced aeration at 275 rpm in a 1-liter vessel; the working volume was 350 ml. The growth rates were fixed by adjusting the dilution rate (D) between 0.02 and 0.40 h-1. The concentration of both nutrient broth and glucose was determined to be limiting in control batch experiments conducted before any of the continuous culture experiments. Temperatures ranged from 15 to 37C in different experiments. A constant pH of 7.0 + 0.2 was maintained by the automatic addition of 5% Na2CO3. D, temperature, aeration rate, agitation, and pH were checked at least once daily and often more frequently. Populations were harvested after a minimum of five volume changes, but only when several consecutive optical density (OD) measurements indicated that a steady state had been attained. Batch cultures were grown in magnetically stirred Erlenmeyer flasks. Cells were harvested when the culture density, as measured by the OD at 660 nm, had

attained a value equivalent to the steady-state cell density of the corresponding chemostat culture. This usually occurred at the late logarithmic phase of growth. The relationship between the OD and the concentration of bacteria, as determined by colony formation on m-Endo agar (Difco) with batch or chemostat cultures, was identical within the precision attainable by serial dilutions (for colony counts) and by absorption measurements (for OD values). Disinfectant. Chlorine dioxide (CI02) stock solutions of approximately 2,000 mg of C102 per liter were generated by acid activation of a sodium chlorite solution as previously described (1). Working stock solutions of 100 mg of C102 per liter, prepared by dilution of the original stock, were standardized before each experiment. These solutions were stored at 4C in the dark and discarded after 10 days. Dose-response experiments indicated that a dose of 0.75 mg of C102 per liter resulted in significant killing while providing measurable concentrations of residual disinfectant and surviving organisms. Residuals were measured by iodometric back titration (2). Inactivation conditions. The disinfection reactor consisted of a baffled, well-mixed, pressurized 1-liter vessel held at 23 1C. Bacteria were suspended in 600 ml of C102-demand-free buffered basal salts solution at a concentration of approximately 107 cells per ml. Bacterial concentration was adjusted by using OD as the criterion. Due to the C102 demand exerted by nutrient broth, cells grown in this medium were centrifuged and suspended in demand-free buffered salts solution. Control experiments showed that the centrifugation did not affect viability, and the rinsing step had no effect on cell sensitivity to C102. Cells grown in glucose medium were placed directly into the inactivation solution, since this medium showed no significant C102 demand. The cell suspension in the reactor was rapidly injected with C102 to yield a dose of 0.75 mg/liter. Dye studies showed that complete mixing occurred within 5 s. Samples were withdrawn at times ranging from 15 s to 15 min in sterile bottles containing sufficient Na2S203 to quench any residual C102. Recovery conditions. Samples were immediately diluted to yield 20 to 200 colonies and recovered on membrane filters (Gelman GN-6) in triplicate. Filters were placed on m-Endo medium and incubated at 35.5C for 22 + 2 h. Presentation of data. We have quantified resistance in terms of a survival ratio (N,/NO) after t minutes of contact with the disinfectant C102. It proved unfeasible to express the differences in terms of pseudokinetic rate constants because of the biphasic rate behavior, characterized by an extremely rapid initial die-off followed by an asymptotic approach to an approximately constant surviving fraction which was reached within 15 min in all experiments (see Fig. 1). Therefore, the ratio N15/No is used in this work as the criterion for comparing the effects of the various antecedent growth conditions on the sensitivity of the populations. In previous experiments we have consistently observed non-ideal, biphasic survival kinetics in a variety of situations, ranging from field studies of secondary-treated municipal wastewater treated with 10 mg of C102 per liter to laboratory studies which used C102 doses of 0.5 mg/liter in demand-free distilled water.

816

BERG, MATIN, AND ROBERTS

APPL. ENVIRON. MICROBIOL.

-3
0

> -5
n

-6

15

CONTACT TIME (min) FIG. 1. Reproducibility of two inactivation experiments with steady-state chemostat-grown cells sampled on different days; experiment 1 (0), experiment 2 (0). An E. coli culture was grown at 25C at a D of 0.06 h-1. Glucose was the limiting nutrient (SR, 0.02% glucose). Cells were inactivated at 23C with a dose of 0.75 mg of C102 per liter. Error bars represent one standard deviation.

Since detectable and stable residuals also have been observed under these conditions, it is assumed that the constant surviving fraction (e.g., N15/NO) is a resistant subpopulation. General experimental design. Experiments were designed to facilitate the use of analysis of variance to permit comparisons between various treatments. A treatment consisted of one antecedent growth environment (e.g., 15C on 0.16% nutrient broth at D = 0.40 h-1) from which a population was harvested for an experiment. An experiment consisted of triplicate trials (i.e., bacterial contact with the disinfectant) performed on the sample on the same day. The variance among trials and the similarity or difference between experimental treatments could then be ascertained. Duplicate experiments run on different days could also be checked for reproducibility. Statistical analyses were carried out on an IBM 3033 computer with either the BMDP or Minitab packages.

RESULTS

Validity of the experimental design. Analysis of variance indicated that the variation among triplicate trials on the same day (i.e., one experiment) was not significant at the 0.05 level. A measure of the similarity is the F ratio: F(2,45) = 1.43 < 3.23. Duplicate experiments performed on different days were also shown not to be different at the 0.05 level: F(1,54) = 0.52 < 4.00.

A graphic example of the reproducibility of these experiments is shown in Fig. 1. To ascertain whether the tail of the biphasic survivor curve was due to a resistant fraction of bacteria or to the absence of a disinfectant residual, we measured the residual C102. At a dose of 0.75 mg/liter, residuals ranged from 0.50 to 0.35 mg of C102 per liter for all experiments, with a standard deviation of 0.10 determined during individual experiments. Comparison with control experiments conducted without the addition of a bacterial suspension showed that approximately 13% of the C102 consumed in each experiment was actually due to the bacterial suspension itself, whereas 26% of the C102 was lost due to redox reactions irrelevant to disinfection. Control experiments conducted in a headspace-free reactor showed that 14% of the C102 was lost due to volatilization. Effect of growth temperature. The effect of growth temperature on sensitivity to C102 was examined under two individual limitations, glucose and some component of the nutrient broth. At a submaximal growth rate of 0.06 h-1 under glucose limitation in the chemostat, an increase in growth temperature from 25 to 37C enhanced sensitivity (log N151No standard deviation, -3.21 0.15 and -5.48 0.48, respectively). In contrast, at maximal growth rates in batch cultures, a change in temperature caused essentially no change in sensitivity (log N151No, -4.69 0.21 and -4.78 0.34 at 25 and 37C, respectively). At 25C, batch culture-grown cells were more sensitive than their chemostat-grown counterparts, but at 37C the opposite was true. Cells grown in nutrient broth, whether grown at a submaximal rate (D, 0.20 h-1) in the chemostat or in conventional batch culture (log phase), showed an increase in sensitivity in response to an increase in growth temperature. The log N151No for the chemostat-grown cells was -2.82 0.20, -4.21 0.22, and -5.51 0.36 at 15, 25, and 37C, respectively; for the batch culture-grown cells it was 0.45 0.07 and 0.85 0.32 at 25 and 37C, respectively. At both 25 and 37C, the batch culture-grown cells were more sensitive. Control experiments in which viability was determined at growth temperatures of 15 or 37C and after cultures had been equilibrated in the test reactor at 23C demonstrated that the temperature shift itself-from that of the growth environment to the test reactor temperature of 23C-caused no measurable change in viability. Effect of D. The different sensitivity of batchand chemostat-grown cells at a fixed growth temperature observed above could be due to their different growth rates. We therefore directly measured the effect of D (the specific growth rate) on sensitivity to C102. Two different media

VOL. 44, 1982

SENSITIVITY OF E. COLI TO CHLORINE DIOXIDE

817

z 0

__

\ 250
l Z z 5

~0 IV) -i
c#
Jr
I I

-6
I
I
I

-A

.02 .06

.20

.40

.60

1.00

1.20

1.40

DILUTION

RATE

( hr' )

FIG. 2. Effect of D on sensitivity to chlorine dioxide under glucose limitation at two growth temperatures. SR (glucose) was 0.02%. Growth at IUmax values (0, *) was obtained with batch cultures grown in 0.02% glucose at the two temperatures. Symbols: *, 25C; O, 37C. Error bars represent one standard deviation.

(glucose and nutrient broths) and two different temperatures (25 and 37C) were included in these measurements. A composite of the experiments based on the fraction of cells surviving after 15 min is shown in Fig. 2 and 3 for glucose and nutrient broth growth media, respectively. The resistant fraction (N15/NO) is plotted as a function of D and growth temperature for cells grown in 0.02% glucose (Fig. 2) and in 0.16% nutrient broth (Fig. 3). It is clear that in either medium and at either temperature, the D at

which the antecedent growth occurred had a pronounced effect on the sensitivity of the culture to C102 (Fig. 2 and 3). In glucose medium, the sensitivity was lowest at an intermediate growth rate (0.20 h-1) and increased with either decreasing or increasing D. In nutrient broth the pattern was less complex, and there was in general a decrease in resistance with increasing D. In both media the resistance tended to be greater after growth at 25 than at 37C, especially at low D values.

-3

)00 -41-

H\ A
_

z
4

.51

\ 37
-6
I

>

Lir

t ,

.40 .60 .20 RATE ( hr ) DILUTION FIG. 3. Effect of D on sensitivity to C102 under the limitation of a component of nutrient broth at two growth temperatures. SR (nutrient broth) was 0.16%. Growth at Pmax values (0, 0) was obtained with batch cultures grown in 0.16% nutrient broth at the two temperatures. Symbols: *, 25C; O, 37C. Error bars represent one
.02 .06
standard deviation.

IN-110-0

818

BERG, MATIN, AND ROBERTS

on

APPL. ENVIRON. MICROBIOL.

population growing more rapidly was hypothesized to be more sensitive. For example, active substrate transport systems or genetic material SR (% glucose) D (h-1) Cell density' Log N15/NOC could be more easily damaged in a growing than 0.06 1.10 -3.12 0.16 0.15 0.02 0.06 0.20 -4.24 0.20 in a senescent population. Grunau (10, 11) reported similar results, correlating an increased amount of replicating DNA during rapid growth 0.40 1.10 -3.09 0.13 0.15 with increased sensitivity to UV light. 0.20 0.40 -5.06 0.13 0.02 The effect of growth temperature was more a Growth temperature was 25C. variable than that of growth rate on s;nsitivity. b Average OD at 660 nm. c Survival ratio after 15 min of contact with a dose of Some distinct differences existed, yielding more sensitive populations at higher growth tempera0.75 mg of C102 per liter at 23C. tures. If it is assumed that the permeability of the outer membrane to small molecules is due in part to the fluidity of the lipid bilayer, then those Effect of SR in medium inflow. In addition to results are reasonable. This effect of permeabilitemperature and D, the concentration of the ty with respect to temperature-modified lipids limiting nutrient (SR) in the inflow medium was has been demonstrated previously (8). Also, also found to influence sensitivity to C102. Table Haas, in his work with atypical mycobacteria, 1 shows this effect for extreme values of D at a suggested that the dense lipid layers of the fixed temperature. At both D values, popula- mycobacteria offer protection against penetrations grown at the lower SR concentration were tion of chlorine (12). A less permeable memmore sensitive to inactivation by C102 than were brane could retard the leakage of other small those grown at the higher concentration. constituents (e.g., K+) critical for viability (2) as Since there is a direct relationship between SR well, especially for a sublethally injured bacteriand steady-state cell biomass (Table 1), the um. On the other hand, our data showed cases to results indicate that a lower cell density during which this hypothesis did not apply, i.e., when growth is conducive to enhanced resistance to temperature had no significant effect on sensitivC102. We emphasize that the density during ity. This may have'been due to an overriding growth is the determining factor here and not the factor controlling the sensitivity in that particudensity of populations subjected to the inactiva- lar growth regime. tion procedure: the latter density was kept conThe SR increased the steady-state biomass. stant in all experiments, as described above. However, a decrease in sensitivity with increased SR was unexpected. We can offer no DISCUSSION explanation for this finding. The results presented here show that populaA major objective of this study was to critically test the premise that antecedent growth condi- tions grown under conditions that more closely tions can influence bacterial sensitivity to disin- approximate natural environments are more refectants. The results clearly establish the sistant to disinfectants than are those grown validity of this premise, and it is evident that the under commonly used laboratory conditions. sensitivity of E. coli is markedly influenced by Since the conventional growth conditions (e.g., all of the four environmental parameters of batch culture at 37C) are representative of growth tested here, i.e., the qualitative nature of techniques commonly employed in laboratory the growth environment (as evidenced by the evaluations of disinfectants and in the elucidadifference between glucose and nutrient broth- tion of mechanisms of inactivation, it appears grown cells cultivated under otherwise similar that the alternative methods presented in this conditions), degree of nutrient limitation, tem- work merit further study. perature, and density of the culture. These results strengthen the notion that the often-obACKNOWLEDGMENTS served differences in sensitivity between natural J.D.B. was supported by funds from the Marx-Moreno populations and their laboratory-grown counter- Fellowship of Stanford University School of Engineering, parts is due to their different antecedent growth Stanford, Calif. conditions. Review of the manuscript by C. N. Hass, Illinois Institute The relationship between growth rate and of Technology, is gratefully acknowledged. sensitivity became evident during other experiments in our laboratory with batch cultures (8). LITERATURE CITED Similarly, for E. coli cells grown in the chemo1. E. D. Berg, P. V. Roberts, and R. C. Cooper. Aieta, A., J. stat in nutrient broth at either 25 or 37C, our 1980. Comparison of chlorine dioxide and chlorine in the results showed an increase in sensitivity as the disinfection of wastewater. J. Water Pollut. Control Fed. growth rate increased. Based on those results, a 52:810-822.
TABLE 1. Effect of SR in inflowing medium

sensitivity to ClO2a

VOL. 44, 1982

SENSITIVITY OF E. COLI TO CHLORINE DIOXIDE

819

2. American Public Health Association. 1976. Standard methods for examination of water and waste water, 14th ed. American Public Health Association, Inc., New York. 3. Brock, T. D. 1971. Microbial growth rates in nature. Bacteriol. Rev. 35:39-58. 4. Carson, L. A., M. S. Favero, W. W. Bond, and N. J. Petersen. 1972. Factors affecting the comparative resistance of naturally occurring and subcultured Pseudomonas aeruginosa to disinfectants. Appl. Microbiol. 23:863-869. 5. Carson, L. A., N. J. Petersen, M. S. Favero, and S. M. Aguero. 1978. Growth characteristics of atypical mycobacteria in water and their comparative resistance to disinfectants. Appl. Environ. Microbiol. 36:839-846. 6. Characklas, E. G. 1973. Attached microbial growths. II. Frictional resistance due to microbial slimes. Water Res. 7:1249-1258. 7. Favero, M. S., and C. H. Drake. 1966. Factors influencing the occurrence of high members of iodine-resistant bacteria in iodinated swimming pools. Appl. Microbiol. 14:627635. 8. Gill, C. O., and J. R. Suisted. 1978. The effects of temperature and growth rate on the proportion of unsaturated fatty acids in bacterial lipids. J. Gen. Microbiol. 104:3136. 9. Grieble, H. C., F. R. Colton, and T. J. Bird. 1970. Fineparticle humidifiers: source of Pseudomonas aeruginosa infections in a respiratory disease unit. N. Engl. J. Med. 282:531-535. 10. Grunau, J. A. 1978. The ultraviolet sensitivity of Klebsiella pneumoniae as a function of growth rate. FEMS Lett. 4:47-50. 11. Grunau, J. A. 1978. Ultraviolet lethality and the sensitivity of chromosome replication forks. FEMS Lett. 4:51-54. 12. Haas, C. N., and R. S. Englebrecht. 1980. Physiological alterations of vegetative microorganisms resulting from chlorination. J. Water Pollut. Control Fed. 52:1976-1989.

13. Hoff, J. C. 1978. The relationship of turbidity to disinfection of potable water, p. 103-117. In C. W. Henricks (ed.), Evaluation of the microbiology standards for drinking water. U.S. Environmental Protection Agency, Cincinnati. 14. Jannasch, H. W. 1969. Estimations of bacterial growth rates in natural waters. J. Bacteriol. 99:156-160. 15. Lambert, P. A., and S. M. Hammond. 1973. Potassium fluxes, first indications of membrane damage in microorganisms. Biochim. Biophys. Res. Commun. 54:796-799. 16. Matin, A., A. Grootjins, and H. Hogenhuis. 1976. Influence of dilution rate on enzymes of intermediary metabolism in two fresh water bacteria grown in continuous culture. J. Gen. Microbiol. 94:323-332. 17. Matin, A., and H. Veldkamp. 1978. Physiological basis of the selective advantage of a Spirillum sp. in a carbonlimited environment. J. Gen. Microbiol. 105:187-197. 18. Milbauer, R., and N. Grossowicz. 1959. Effect of growth conditions on chlorine sensitivity of Escherichia coli. Appl. Microbiol. 7:71-74. 19. Niekus, H. G. D., W. deVries, and A. H. Stouthamer. 1977. The effect of different oxygen tensions on growth and enzyme activities of Campylobacter sputorum subspecies bubulis. J. Gen. Microbiol. 103:215-222. 20. Seidler, R. J., J. E. Morrow, and S. T. Bagley. 1977. Klebsiellae in drinking water emanating from redwood tanks. Appl. Environ. Microbiol. 33:893-900. 21. Skaliy, P., and H. V. McEachern. 1979. Survival of legionnaires disease bacterium in water. Ann. Int. Med. 90:662663. 22. SkalHy, P., T. A. Thompson, G. W. Gorman, G. K. Morris, H. V. McEachern, and D. C. Mackel. 1980. Laboratory studies of disinfectants against Legionella pneumophila. Appl. Environ. Microbiol. 40:697-700. 23. Vishniac, W., and M. Santer. 1957. The thiobacilli. Bacteriol. Rev. 21:195-213.