- Email: [email protected]
Rare occurrence of heterotrophic bacteria with pathogenic potential in potable water
International Journal of Food Microbiology 92 (2004) 249 – 254 www.elsevier.com/locate/ijfoodmicro
Rare occurrence of heterotrophic bacteria with pathogenic potential in potable water Gerard N. Stelma Jr. a,*, Dennis J. Lye a, Bennett G. Smith a, James W. Messer a, Pierre Payment b b
a National Exposure Research Laboratory, U.S. Environmental Protection Agency, Cincinnati, OH 45268, USA Centre de Recherche en Virologie, Institut Armand-Frappier, 531 Boulevard des Prairies, Laval, Quebec, Canada H7N 4Z3
Presented at the NSF International/World Health Organization Symposium on HPC Bacteria in Drinking Water, April 22 – 24, 2002, Geneva, Switzerland.
Abstract Since the discovery of Legionella pneumophila, an opportunistic pathogen that is indigenous to water, microbiologists have speculated that there may be other opportunistic pathogens among the numerous heterotrophic bacteria found in potable water. The US Environmental Protection Agency (USEPA) developed a series of rapid in vitro assays to assess the virulence potential of large numbers of bacteria from potable water to possibly identify currently unknown pathogens. Results of surveys of potable water from several distribution systems using these tests showed that only 50 of the approximately 10,000 bacterial colonies expressed one or more virulence characteristics. In another study, 45 potable water isolates that expressed multiple virulence factors were tested for pathogenicity in immunocompromised mice. None of the isolates infected mice that were compromised either by treatment with carrageenan (CG), to induce susceptibility to facultative intracellular pathogens, or by cyclophosphamide (CY), to induce susceptibility to extracellular pathogens. These results indicate that there are very few potential pathogens in potable water and that the currently developed in vitro virulence screening tests give an overestimation of the numbers of heterotrophic bacteria that may be pathogens. Current efforts are focused on using the animal models to screen concentrated samples of waters known to contain large numbers of heterotrophic bacteria and newly discovered Legionella-like organisms that parasitize amoebae. D 2003 Elsevier B.V. All rights reserved. Keywords: Virulence; Heterotrophic bacteria; Potable water
1. Introduction Treated potable water contains a variety of heterotrophic bacteria that are not well characterized. Many of these organisms grow slowly and require nutrient-
* Corresponding author. E-mail address: [email protected] (G.N. Stelma). 0168-1605/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2003.08.011
poor media for culturing (Reasoner and Geldreich, 1985). Although there is evidence that these bacteria are not hazardous to the general healthy population (Calderon and Mood, 1988; Calderon, 1991; Regunathan and Beauman, 1994), there is a possibility that some of them may be opportunistic pathogens and may be capable of causing adverse health effects in individuals with impaired body defenses. At least two types of slow-growing heterotrophic bacteria, Legion-
250
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
ella spp. (Broome and Fraser, 1979) and Mycobacterium spp. (Goslee and Wolinsky, 1976) commonly found in potable water are known opportunistic pathogens. In addition, some strains of at least two types of rapidly growing heterotrophic bacteria found in potable water, Aeromonas (Clark et al., 1982; Le Chevallier et al., 1982) and Pseudomonas (Morrison and Wenzel, 1984), are also capable of causing infections, if present in infectious doses. There is no reason to assume that the currently known opportunistic pathogens are the only opportunistic pathogens indigenous to potable water. Many cases of respiratory infections and digestive system infections still are of unknown etiology and it is possible that some of them could be due to pathogens that are currently unknown. The US Environmental Protection Agency (USEPA) is interested to learn whether there are additional opportunistic pathogens among the heterotrophic bacteria indigenous to treated potable water. The initial approach taken by USEPA scientists was to develop several in vitro tests to screen colonies of heterotrophic bacteria on membrane filters for expression of traits frequently associated with virulence and to use these tests to characterize heterotrophic bacteria found in potable water samples from a community (Lye and Dufour, 1993). The results of these preliminary experiments suggested that fewer than 1% of isolates from potable water possess these characteristics and that these organisms occur in small numbers (Lye and Dufour, 1993). Similar results were observed in a study performed at Yale University (Edberg et al., 1996). The results of these studies suggest that heterotrophic plate count bacteria are very unlikely to constitute a health hazard. In contrast, isolates of heterotrophic bacteria obtained during an epidemiological study in Canada (Payment et al., 1997), expressed virulence factors at a much higher frequency suggesting that, under some circumstances, significant numbers of bacteria with pathogenic potential may be found in potable water. In this study, we investigated the potential pathogenicity of some of these isolates in immunocompromised mice to ascertain whether any of them were opportunistic pathogens and to determine how well the in vitro tests for virulence factors correlated with actual pathogenicity as measured in animal models. The data suggest that none of these isolates tested are
opportunistic pathogens and that the results of the currently available in vitro tests overestimate the virulence potential of a bacterial isolate.
2. Materials and methods 2.1. Bacterial isolates The test strains of heterotrophic bacteria from potable water were isolated at the Institut ArmandFrappier, Laval, Quebec, Canada. These strains were isolated from water collected from domestic taps in large buildings in the Montreal urban area where water can more easily stagnate in the extensive piping systems. Water samples were collected in 1 l, sterile containers containing sodium thiosulfate (20 mg/l) to neutralize residual chlorine. As the objective was to obtain heterotrophic bacteria, water was not flushed before sampling. The Listeria monocytogenes and Listeria innocua control strains were provided by Dr. A.D. Hitchins, U.S. Food and Drug Administration, Washington, DC, and the Vibrio vulnificus control strains were provided by Dr. James Oliver, University of North Carolina, at Charlotte. Aeromonas hydrophila strain SSU was provided by Dr. Ashok Chopra, University of Texas Medical Branch, Galveston TX, and the Legionella pneumophila serotype 1 was a cooling tower isolate from our own culture collection. 2.2. Characterization of bacteria isolates Species identifications were achieved using the Crystal ID system (BBL). The in vitro virulence tests were performed using in situ membrane filter transfers as described in earlier studies from our laboratory (Lye and Dufour, 1991, 1993). 2.3. Bioassays White, outbred Swiss Webster mice (18 – 20 g) were treated either with carrageenan (CG) to suppress their defenses against facultative intracellular pathogens or with cyclophosphamide (CY) to suppress their defenses against extracellular pathogens. CG (Sigma type II) was dissolved in sterile distilled water and injected IP into the mice (200 mg/kg) 24 h prior to
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
challenge (Tamura and Tanaka, 1984; Stelma et al., 1987). The mice were challenged IP the following day with 1 103 to 5 104 cfu from overnight cultures of heterotrophic bacterial isolates. Positive control mice were challenged with 1 103 to 5 104 virulent L. monocytogenes and sterile PBS was injected into negative controls. Mice that showed signs of infection (ruffled fur, labored breathing, or crusting of conjunctival fluid) were sacrificed and their livers and spleens were aseptically removed. The livers and spleens were also removed from animals that died within 5 days after challenge and from all of the surviving animals, which were sacrificed at the end of the experiments. The organs were macerated and cultured for the challenge organisms. Isolation of the infecting organisms from livers or spleens was considered a positive for infection. An organism was considered pathogenic if it infected three of five compromised animals (Stelma et al., 1987). CY (Sigma) was dissolved in sterile distilled water and injected IP into mice (150 mg/kg) 72 h prior to challenge (Tamura and Tanaka, 1984; Stelma et al., 1992). Positive controls were challenged with 1 to 103 to 5 104 virulent Vibrio vulnificus and negative controls were injected with sterile PBS. Criteria for infection were as described above.
3. Results 3.1. Characteristics of the isolates The isolates sent to us were not a typical array of heterotrophic bacteria but they were specifically chosen because they expressed one or more key virulence factors. All virulence tests were repeated in our laboratory to determine whether any of the virulence factors had been lost during storage or transit to our laboratory. A summary of the virulence characteristics and other properties of the 45 isolates used in this study is shown in Table 1. One surprising result was the high proportion of Gram-positive isolates (73%). Thirty-six of these isolates were identified to the genus or species level, seven of the unidentified isolates were Gram-positive rods, one was a Grampositive coccus and one was a Gram-negative rod. Several isolates were from species sometimes associated with opportunistic infections, including Stenotro-
251
Table 1 Characteristics of 45 heterotrophic isolates 33 (73%) 12 (27%) 6 (13%) 2 (4%) 34 (76%) 35 (78%) 25 (56%)
Gram-positive Gram-negative Cytotoxic Beta-hemolytic Collaginase-positive Gelatinase-positive Elastase-positive
phomonas maltophila, Chromobacterium violaceum, Sphingomonas paucimobilis and three isolates of Bacillus cereus. One isolate was identified as Yersinia pseudotuberculosis, a species that includes frank pathogens. 3.2. Bioassays The models adopted for this study utilized infection as an endpoint rather than death to minimize suffering by the animals. Earlier studies using CG and CY with the bacterial species selected as controls for this study (Stelma et al., 1987, 1992) used 50% lethal doses as endpoints. To verify that the avirulent L. innocua and the avirulent strain of V. vulnificus do not infect the livers and spleens of compromised mice at the planned doses, compromised mice were challenged with these avirulent strains. Mice that had been treated with CG were challenged with graded doses of L. innocua up to 106 cfu and mice that had been treated with CY were challenged with up to 106 cfu avirulent V. vulnificus. None of the mice showed signs of illness during the 5 days of observation in either experiment. The animals were sacrificed; the livers and spleens were aseptically removed and cultured for the challenge organisms. Both avirulent strains failed to invade the livers or spleens of the compromised mice at any time during the 5 days after challenge. The positive control strains, however, were consistently isolated from the livers and spleens of compromised mice after challenge doses of only 103 to 104 cfu. Two additional isolates were tested in an attempt to better define the parameters of the pathogenicity tests. The L. pneumophila serotype 1 isolate was tested to verify that the CG model will work for pathogens that cause pneumonia when the mice are challenged IP, and A. hydrophila strain SSU was used to determine whether these models would identify
252
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
bacteria that cause diarrhea but not disseminating illness. The results showed that the L. pneumophila isolate did invade the livers and spleens of the CGtreated mice and, therefore, tested positive for pathogenicity in spite of the fact that the organisms were not delivered to the lungs. The A. hydrophila strain did not infect the livers or spleens of immunocompromised mice and did not cause diarrhea when given to the mice via IP injection. The 45 heterotrophic bacterial strains were used to challenge both CG-treated mice and CY-treated mice. The mice were observed for signs of illness for 5 days. Those mice that showed no signs of illness, including the negative controls, were sacrificed at the end of the experiments, their livers and spleens were removed and macerated, and attempts were made to culture the challenge organisms from the macerated organs. All 45 heterotrophic bacterial isolates were negative in both animal models. In fact, no signs of infection were observed in any mouse; and no challenge bacteria were isolated from any of the organs. The positive control mice were always positive for infection, with most of the animals dying within the first day after challenge. No organisms were cultured from the organs of the negative control mice in any of the experiments, indicating that none of the mice were attacked by their own indigenous flora or other organisms from their environment during the course of the experiments. Even those isolates that expressed multiple virulence factors including cytotoxin production were negative, as were isolates identified as fitting into a genus or species sometimes associated with opportunistic infections. Representative data obtained from Gram-negative isolates that appeared from the results of the in vitro virulence tests and/or from their genus identification to have potential pathTable 2 Characteristics of representative Gram-negative isolates Isolate
CYT HEM GEL ELA COL Bioassay Dose (cfu)
Stenotrophomonas + Chromobacterium Sphingomonas + CYT = cytotoxin. HEM = hemolysin. GEL = gelatinase. ELA = elastase. COL = collaginase.
+ +
+
+ +
0/5 0/5 0/5
1 103 3 103 3 104
Table 3 Characteristics of representative gram positive isolates Isolate
CYT HEM GEL ELA COL Bioassay Dose (cfu)
Bacillus spp. + Staphylococcus Corynebacterium
+ + +
+ +
+ + +
0/5 0/5 0/5
3 103 1 104 5 104
CYT = cytotoxin. HEM = hemolysin. GEL = gelatinase. ELA = elastase. COL = collaginase.
ogenicity are shown in Table 2. All three of these isolates are from genera that have been associated with opportunistic infections. The Stenotrophomonas, in particular, was considered a likely candidate for pathogenicity because it produced four virulence factors including cytotoxin. Representative data obtained from Gram-positive isolates are shown in Table 3. Again, all three isolates are from genera that have been associated with infections. Even the Bacillus isolate, which expressed four virulence factors, including cytotoxin, was negative for pathogenicity.
4. Discussion The animal models used in this study should have identified any opportunistic pathogens that could cause disseminating infections, such as those caused by the species used as positive controls in this study. CG effectively eliminates macrophage function, which is essential for the body to eradicate facultative intracellular pathogens; and CY drastically reduces the number of polymorphonuclear cells, which are essential for the body to eliminate extracellular pathogens (Tatsukawa et al., 1979). The facultative intracellular pathogen, L. monocytogenes, was consistently isolated from the livers and spleens of positive control mice that had been treated with CG and challenged with low doses of L. monocytogenes; and the extracellular pathogen, V. vulnificus, was consistently isolated from the livers and spleens of positive control mice treated with CY and challenged with low doses of V. vulnificus. Because a test for infectivity rather than lethality was used, there was some concern that avirulent organisms that would have been negative in LD50 experiments would be able to establish sublethal
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
infections and appear to be pathogenic. Therefore, control experiments were performed with known avirulent strains of L. innocua and V. vulnificus. These organisms were negative for pathogenicity at doses several orders of magnitude higher than those used for the test strains of heterotrophic bacteria, indicating that there was little potential for false positive results. The CG model was effective for detecting virulent L. pneumophila in earlier experiments when intratracheal challenges were used (Davis-Hoover et al., 1990). However, it was not certain that L. pneumophila or other respiratory system pathogens would infect if given by IP injections The fact that our virulent isolate of L. pneumophila invaded the livers and spleens of CG-treated mice after IP challenge suggests that the model also has potential to identify isolates that cause pneumonia. The failure to find any A. hydrophila in the organs or to observe any signs of diarrhea in the mice challenged with A. hydrophila strain SSU indicates that this model is not useful for organisms that cause diarrhea and there is a need for an effective and rapid in vivo screening method for potential to produce diarrhea. The results of this study with the 45 isolates reinforce the conclusions of previous studies that slow-growing heterotrophic bacteria have little potential for virulence (Lye and Dufour, 1993; Edberg et al., 1996). In fact, the results of this study suggest there is even less risk than one would assume from the results of the previous studies. This is because the isolates we tested, even those isolates that produced cytotoxins plus three or more additional putative virulence factors, showed no signs of pathogenicity in compromised mice, which suggests that the array of in vitro virulence tests currently available for rapid screening of bacterial colonies is insufficient. Bacterial virulence is multifactorial and not well understood; and until we have better understanding of virulence and can develop an additional array of virulence tests, the current tests will give us an overestimation of the number of potentially pathogenic heterotrophic bacteria in potable water. The possibility of unknown opportunistic pathogens in potable water still exists but it is unlikely that any will be found by screening bacterial colonies individually in vivo or by screening large numbers of colonies using the currently available in vitro virulence tests. There are other approaches that are more likely to be successful. One approach would be to test concen-
253
trates of microorganisms from potable waters that contain exceptionally high numbers of heterotrophic bacteria in the animal models. Any organisms that can be isolated from the normally sterile livers and spleens would be regarded as potential pathogens. Before any conclusions could be made, it would be necessary to verify that the same organisms occur in significant numbers in the water. It would also be necessary to perform epidemiological studies to determine whether these organisms also infect humans and whether they occur in large enough numbers to constitute an infectious dose for humans. Human pathogens typically do not grow on diluted media such as R2A, nor do organisms that grow on these media express virulence factors (Edberg et al., 1996). However, there are two exceptions to these rules; Legionella species and nontuberculous mycobacteria are found in potable water and require unusual growth media (Dufour and Jakubowski, 1982; Du Molin and Stottmeier, 1986). Pathogenic legionellae survive and grow in amoebae, and this may relate to their ability to parasitize human cells (Rowbotham, 1980); this may also be true for nontuberculous mycobacteria. Therefore, another approach would be to specifically look for new types of pathogens of amoebae and test these organisms in the CG model for facultative intracellular pathogens. Again, it would be necessary to use an epidemiological approach to verify whether or not these organisms occur in numbers sufficient to infect humans and that they are human as well as mouse pathogens. It would also be possible to start from an epidemiological approach looking for unusual opportunistic infections in a community with a high occurrence of heterotrophic bacteria in its potable water and a community with very low occurrence, then verifying that the causative agent is present in the potable water of the first community but not the second.
5. Summary and conclusions A total of 45 isolates of heterotrophic bacteria were tested for pathogenicity in two animal models, one designed to identify facultative intracellular pathogens and the other to identify extracellular pathogens. Although these isolates were chosen because they expressed multiple putative virulence factors, all were negative for pathogenicity in both models. In addition,
254
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
several of these isolates were identified as members of genera or species known to contain opportunistic pathogens. The results provide no evidence for additional opportunistic pathogens on potable water and suggest that the battery of in vitro virulence tests currently available is insufficient to predict pathogenicity for animals. The results also suggest that if there are any currently unknown opportunistic pathogens in potable water, new approaches are needed to identify them.
Acknowledgements We thank Dr. Pierre Payment, Dr. A.D. Hitchins, Dr. James Oliver and Dr. Ashok Chopra for providing the bacterial strains used in this study and Mrs. Doris Morris for assistance in preparing the manuscript.
References Broome, C.V., Fraser, D.W., 1979. Epidemiologic aspects of legionellosis. Epidemiol. Rev. 1, 1 – 16. Calderon, R.G., 1991. Bacteria colonizing point-of-entry, granular activated carbon filters and their relationship to human health. Final Report CR 813978-01-0, USEPA. Calderon, R.G., Mood, E.W., 1988. Bacteria colonizing point-ofuse, granular activated carbon filters and their relationship to human health. Final Report CR 811904-01-0, USEPA. Clark, J.A., Burger, C.A., Sabatinos, L.E., 1982. Characterization of indicator bacteria in municipal raw water, drinking water and main water samples. Can. J. Microbiol. 28, 1002 – 1013. Davis-Hoover, W.J., Dufour, A.P., Stelma Jr., G.N., Scarpino, P.V., 1990. Use of an in vivo model for identifying opportunistic pathogens isolated from drinking water. Proc. AWWA Annu. Conf. Part 2, 1693 – 1706. Dufour, A.P., Jakubowski, W., 1982. Drinking water and Legionnaires disease. J. AWWA, 631 – 637. Du Molin, G.C., Stottmeier, K.D., 1986. Waterborne mycobacteria: an increasing threat to health. ASM News 52, 525 – 529.
Edberg, S.C., Gallo, P., Kontnick, C., 1996. Analysis of the virulence characteristics of bacteria isolated from bottled, water cooler and tap water. Microb. Ecol. Health Dis. 9, 67 – 77. Goslee, S., Wolinsky, E., 1976. Water as a source of potentially pathogenic mycobacteria. Am. Rev. Respir. Dis. 113, 287 – 292. Le Chevallier, M.W., Evans, T.M., Seidler, R.J., Daily, O.P., Merrell, B.R., Rollins, D.M., Joseph, S.W., 1982. Aeromonas sobria in chlorinated drinking water supplies. Microb. Ecol. 8, 325 – 333. Lye, D.J., Dufour, A.P., 1991. A membrane filter procedure for assaying cytotoxic activity in heterotrophic bacteria isolated from drinking water. J. Appl. Bacteriol. 70, 89 – 94. Lye, D.J., Dufour, A.P., 1993. Virulence characteristics of heterotrophic bacteria commonly isolated from potable water. Environ. Toxicol. Water Qual. 8, 13 – 23. Morrison, A.J., Wenzel, R.P., 1984. Epidemiology of infections due to Pseudomonas aeruginosa. Rev. Infect. Dis., S627 – S642. Payment, P., Siemiatycki, J., Richardson, L., Renaud, G., Franco, E., Pr vost, M., 1997. A prospective epidemiological study of gastrointestinal health effects due to the consumption of drinking water. Int. J. Environ. Health Res. 7, 5 – 31. Reasoner, D.J., Geldreich, E.E., 1985. A new medium for the enumeration and subculture of bacteria from potable water. Appl. Environ. Microbiol. 49, 1 – 7. Regunathan, P., Beauman, W.H., 1994. Significance of heterotrophic bacteria in drinking water. Annu. Conf. Water Qual. Assoc. Rowbotham, T.J., 1980. Preliminary report on the pathogenicity of Legionella pneumophila for freshwater amoebae. J. Clin. Pathol. 33, 1179 – 1183. Stelma Jr., G.N., Reyes, A.L., Peeler, J.T., Francis, D.W., Hunt, J.M., Spaulding, P.L., Johnson, C.H., Lovett, J., 1987. Pathogenicity test for Listeria monocytogenes using immunocompromised mice. J. Clin. Microbiol. 25, 2085 – 2089. Stelma Jr., G.N., Reyes, A.L., Peeler, J.T., Johnson, C.H., Spaulding, P.L. 1992. Virulence characteristics of clinical and environmental isolates of Vibrio vulnificus. Appl. Environ. Microbiol. 58, 2776 – 2782. Tamura, Y., Tanaka, S., 1984. Effect of antiflagellar serum in the protection of mice against Clostridium chauvoei. Infect. Immun. 43, 612 – 616. Tatsukawa, K., Mitsuyama, M., Takeya, K., Nomoto, K., 1979. Differing contribution of polymorphonuclear cells and macrophages to protection of mice against Listeria monocytogenes and Pseudomonas aeruginosa. J. Gen. Microbiol. 115, 161 – 166.
Rare occurrence of heterotrophic bacteria with pathogenic potential in potable water Gerard N. Stelma Jr. a,*, Dennis J. Lye a, Bennett G. Smith a, James W. Messer a, Pierre Payment b b
a National Exposure Research Laboratory, U.S. Environmental Protection Agency, Cincinnati, OH 45268, USA Centre de Recherche en Virologie, Institut Armand-Frappier, 531 Boulevard des Prairies, Laval, Quebec, Canada H7N 4Z3
Presented at the NSF International/World Health Organization Symposium on HPC Bacteria in Drinking Water, April 22 – 24, 2002, Geneva, Switzerland.
Abstract Since the discovery of Legionella pneumophila, an opportunistic pathogen that is indigenous to water, microbiologists have speculated that there may be other opportunistic pathogens among the numerous heterotrophic bacteria found in potable water. The US Environmental Protection Agency (USEPA) developed a series of rapid in vitro assays to assess the virulence potential of large numbers of bacteria from potable water to possibly identify currently unknown pathogens. Results of surveys of potable water from several distribution systems using these tests showed that only 50 of the approximately 10,000 bacterial colonies expressed one or more virulence characteristics. In another study, 45 potable water isolates that expressed multiple virulence factors were tested for pathogenicity in immunocompromised mice. None of the isolates infected mice that were compromised either by treatment with carrageenan (CG), to induce susceptibility to facultative intracellular pathogens, or by cyclophosphamide (CY), to induce susceptibility to extracellular pathogens. These results indicate that there are very few potential pathogens in potable water and that the currently developed in vitro virulence screening tests give an overestimation of the numbers of heterotrophic bacteria that may be pathogens. Current efforts are focused on using the animal models to screen concentrated samples of waters known to contain large numbers of heterotrophic bacteria and newly discovered Legionella-like organisms that parasitize amoebae. D 2003 Elsevier B.V. All rights reserved. Keywords: Virulence; Heterotrophic bacteria; Potable water
1. Introduction Treated potable water contains a variety of heterotrophic bacteria that are not well characterized. Many of these organisms grow slowly and require nutrient-
* Corresponding author. E-mail address: [email protected] (G.N. Stelma). 0168-1605/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2003.08.011
poor media for culturing (Reasoner and Geldreich, 1985). Although there is evidence that these bacteria are not hazardous to the general healthy population (Calderon and Mood, 1988; Calderon, 1991; Regunathan and Beauman, 1994), there is a possibility that some of them may be opportunistic pathogens and may be capable of causing adverse health effects in individuals with impaired body defenses. At least two types of slow-growing heterotrophic bacteria, Legion-
250
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
ella spp. (Broome and Fraser, 1979) and Mycobacterium spp. (Goslee and Wolinsky, 1976) commonly found in potable water are known opportunistic pathogens. In addition, some strains of at least two types of rapidly growing heterotrophic bacteria found in potable water, Aeromonas (Clark et al., 1982; Le Chevallier et al., 1982) and Pseudomonas (Morrison and Wenzel, 1984), are also capable of causing infections, if present in infectious doses. There is no reason to assume that the currently known opportunistic pathogens are the only opportunistic pathogens indigenous to potable water. Many cases of respiratory infections and digestive system infections still are of unknown etiology and it is possible that some of them could be due to pathogens that are currently unknown. The US Environmental Protection Agency (USEPA) is interested to learn whether there are additional opportunistic pathogens among the heterotrophic bacteria indigenous to treated potable water. The initial approach taken by USEPA scientists was to develop several in vitro tests to screen colonies of heterotrophic bacteria on membrane filters for expression of traits frequently associated with virulence and to use these tests to characterize heterotrophic bacteria found in potable water samples from a community (Lye and Dufour, 1993). The results of these preliminary experiments suggested that fewer than 1% of isolates from potable water possess these characteristics and that these organisms occur in small numbers (Lye and Dufour, 1993). Similar results were observed in a study performed at Yale University (Edberg et al., 1996). The results of these studies suggest that heterotrophic plate count bacteria are very unlikely to constitute a health hazard. In contrast, isolates of heterotrophic bacteria obtained during an epidemiological study in Canada (Payment et al., 1997), expressed virulence factors at a much higher frequency suggesting that, under some circumstances, significant numbers of bacteria with pathogenic potential may be found in potable water. In this study, we investigated the potential pathogenicity of some of these isolates in immunocompromised mice to ascertain whether any of them were opportunistic pathogens and to determine how well the in vitro tests for virulence factors correlated with actual pathogenicity as measured in animal models. The data suggest that none of these isolates tested are
opportunistic pathogens and that the results of the currently available in vitro tests overestimate the virulence potential of a bacterial isolate.
2. Materials and methods 2.1. Bacterial isolates The test strains of heterotrophic bacteria from potable water were isolated at the Institut ArmandFrappier, Laval, Quebec, Canada. These strains were isolated from water collected from domestic taps in large buildings in the Montreal urban area where water can more easily stagnate in the extensive piping systems. Water samples were collected in 1 l, sterile containers containing sodium thiosulfate (20 mg/l) to neutralize residual chlorine. As the objective was to obtain heterotrophic bacteria, water was not flushed before sampling. The Listeria monocytogenes and Listeria innocua control strains were provided by Dr. A.D. Hitchins, U.S. Food and Drug Administration, Washington, DC, and the Vibrio vulnificus control strains were provided by Dr. James Oliver, University of North Carolina, at Charlotte. Aeromonas hydrophila strain SSU was provided by Dr. Ashok Chopra, University of Texas Medical Branch, Galveston TX, and the Legionella pneumophila serotype 1 was a cooling tower isolate from our own culture collection. 2.2. Characterization of bacteria isolates Species identifications were achieved using the Crystal ID system (BBL). The in vitro virulence tests were performed using in situ membrane filter transfers as described in earlier studies from our laboratory (Lye and Dufour, 1991, 1993). 2.3. Bioassays White, outbred Swiss Webster mice (18 – 20 g) were treated either with carrageenan (CG) to suppress their defenses against facultative intracellular pathogens or with cyclophosphamide (CY) to suppress their defenses against extracellular pathogens. CG (Sigma type II) was dissolved in sterile distilled water and injected IP into the mice (200 mg/kg) 24 h prior to
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
challenge (Tamura and Tanaka, 1984; Stelma et al., 1987). The mice were challenged IP the following day with 1 103 to 5 104 cfu from overnight cultures of heterotrophic bacterial isolates. Positive control mice were challenged with 1 103 to 5 104 virulent L. monocytogenes and sterile PBS was injected into negative controls. Mice that showed signs of infection (ruffled fur, labored breathing, or crusting of conjunctival fluid) were sacrificed and their livers and spleens were aseptically removed. The livers and spleens were also removed from animals that died within 5 days after challenge and from all of the surviving animals, which were sacrificed at the end of the experiments. The organs were macerated and cultured for the challenge organisms. Isolation of the infecting organisms from livers or spleens was considered a positive for infection. An organism was considered pathogenic if it infected three of five compromised animals (Stelma et al., 1987). CY (Sigma) was dissolved in sterile distilled water and injected IP into mice (150 mg/kg) 72 h prior to challenge (Tamura and Tanaka, 1984; Stelma et al., 1992). Positive controls were challenged with 1 to 103 to 5 104 virulent Vibrio vulnificus and negative controls were injected with sterile PBS. Criteria for infection were as described above.
3. Results 3.1. Characteristics of the isolates The isolates sent to us were not a typical array of heterotrophic bacteria but they were specifically chosen because they expressed one or more key virulence factors. All virulence tests were repeated in our laboratory to determine whether any of the virulence factors had been lost during storage or transit to our laboratory. A summary of the virulence characteristics and other properties of the 45 isolates used in this study is shown in Table 1. One surprising result was the high proportion of Gram-positive isolates (73%). Thirty-six of these isolates were identified to the genus or species level, seven of the unidentified isolates were Gram-positive rods, one was a Grampositive coccus and one was a Gram-negative rod. Several isolates were from species sometimes associated with opportunistic infections, including Stenotro-
251
Table 1 Characteristics of 45 heterotrophic isolates 33 (73%) 12 (27%) 6 (13%) 2 (4%) 34 (76%) 35 (78%) 25 (56%)
Gram-positive Gram-negative Cytotoxic Beta-hemolytic Collaginase-positive Gelatinase-positive Elastase-positive
phomonas maltophila, Chromobacterium violaceum, Sphingomonas paucimobilis and three isolates of Bacillus cereus. One isolate was identified as Yersinia pseudotuberculosis, a species that includes frank pathogens. 3.2. Bioassays The models adopted for this study utilized infection as an endpoint rather than death to minimize suffering by the animals. Earlier studies using CG and CY with the bacterial species selected as controls for this study (Stelma et al., 1987, 1992) used 50% lethal doses as endpoints. To verify that the avirulent L. innocua and the avirulent strain of V. vulnificus do not infect the livers and spleens of compromised mice at the planned doses, compromised mice were challenged with these avirulent strains. Mice that had been treated with CG were challenged with graded doses of L. innocua up to 106 cfu and mice that had been treated with CY were challenged with up to 106 cfu avirulent V. vulnificus. None of the mice showed signs of illness during the 5 days of observation in either experiment. The animals were sacrificed; the livers and spleens were aseptically removed and cultured for the challenge organisms. Both avirulent strains failed to invade the livers or spleens of the compromised mice at any time during the 5 days after challenge. The positive control strains, however, were consistently isolated from the livers and spleens of compromised mice after challenge doses of only 103 to 104 cfu. Two additional isolates were tested in an attempt to better define the parameters of the pathogenicity tests. The L. pneumophila serotype 1 isolate was tested to verify that the CG model will work for pathogens that cause pneumonia when the mice are challenged IP, and A. hydrophila strain SSU was used to determine whether these models would identify
252
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
bacteria that cause diarrhea but not disseminating illness. The results showed that the L. pneumophila isolate did invade the livers and spleens of the CGtreated mice and, therefore, tested positive for pathogenicity in spite of the fact that the organisms were not delivered to the lungs. The A. hydrophila strain did not infect the livers or spleens of immunocompromised mice and did not cause diarrhea when given to the mice via IP injection. The 45 heterotrophic bacterial strains were used to challenge both CG-treated mice and CY-treated mice. The mice were observed for signs of illness for 5 days. Those mice that showed no signs of illness, including the negative controls, were sacrificed at the end of the experiments, their livers and spleens were removed and macerated, and attempts were made to culture the challenge organisms from the macerated organs. All 45 heterotrophic bacterial isolates were negative in both animal models. In fact, no signs of infection were observed in any mouse; and no challenge bacteria were isolated from any of the organs. The positive control mice were always positive for infection, with most of the animals dying within the first day after challenge. No organisms were cultured from the organs of the negative control mice in any of the experiments, indicating that none of the mice were attacked by their own indigenous flora or other organisms from their environment during the course of the experiments. Even those isolates that expressed multiple virulence factors including cytotoxin production were negative, as were isolates identified as fitting into a genus or species sometimes associated with opportunistic infections. Representative data obtained from Gram-negative isolates that appeared from the results of the in vitro virulence tests and/or from their genus identification to have potential pathTable 2 Characteristics of representative Gram-negative isolates Isolate
CYT HEM GEL ELA COL Bioassay Dose (cfu)
Stenotrophomonas + Chromobacterium Sphingomonas + CYT = cytotoxin. HEM = hemolysin. GEL = gelatinase. ELA = elastase. COL = collaginase.
+ +
+
+ +
0/5 0/5 0/5
1 103 3 103 3 104
Table 3 Characteristics of representative gram positive isolates Isolate
CYT HEM GEL ELA COL Bioassay Dose (cfu)
Bacillus spp. + Staphylococcus Corynebacterium
+ + +
+ +
+ + +
0/5 0/5 0/5
3 103 1 104 5 104
CYT = cytotoxin. HEM = hemolysin. GEL = gelatinase. ELA = elastase. COL = collaginase.
ogenicity are shown in Table 2. All three of these isolates are from genera that have been associated with opportunistic infections. The Stenotrophomonas, in particular, was considered a likely candidate for pathogenicity because it produced four virulence factors including cytotoxin. Representative data obtained from Gram-positive isolates are shown in Table 3. Again, all three isolates are from genera that have been associated with infections. Even the Bacillus isolate, which expressed four virulence factors, including cytotoxin, was negative for pathogenicity.
4. Discussion The animal models used in this study should have identified any opportunistic pathogens that could cause disseminating infections, such as those caused by the species used as positive controls in this study. CG effectively eliminates macrophage function, which is essential for the body to eradicate facultative intracellular pathogens; and CY drastically reduces the number of polymorphonuclear cells, which are essential for the body to eliminate extracellular pathogens (Tatsukawa et al., 1979). The facultative intracellular pathogen, L. monocytogenes, was consistently isolated from the livers and spleens of positive control mice that had been treated with CG and challenged with low doses of L. monocytogenes; and the extracellular pathogen, V. vulnificus, was consistently isolated from the livers and spleens of positive control mice treated with CY and challenged with low doses of V. vulnificus. Because a test for infectivity rather than lethality was used, there was some concern that avirulent organisms that would have been negative in LD50 experiments would be able to establish sublethal
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
infections and appear to be pathogenic. Therefore, control experiments were performed with known avirulent strains of L. innocua and V. vulnificus. These organisms were negative for pathogenicity at doses several orders of magnitude higher than those used for the test strains of heterotrophic bacteria, indicating that there was little potential for false positive results. The CG model was effective for detecting virulent L. pneumophila in earlier experiments when intratracheal challenges were used (Davis-Hoover et al., 1990). However, it was not certain that L. pneumophila or other respiratory system pathogens would infect if given by IP injections The fact that our virulent isolate of L. pneumophila invaded the livers and spleens of CG-treated mice after IP challenge suggests that the model also has potential to identify isolates that cause pneumonia. The failure to find any A. hydrophila in the organs or to observe any signs of diarrhea in the mice challenged with A. hydrophila strain SSU indicates that this model is not useful for organisms that cause diarrhea and there is a need for an effective and rapid in vivo screening method for potential to produce diarrhea. The results of this study with the 45 isolates reinforce the conclusions of previous studies that slow-growing heterotrophic bacteria have little potential for virulence (Lye and Dufour, 1993; Edberg et al., 1996). In fact, the results of this study suggest there is even less risk than one would assume from the results of the previous studies. This is because the isolates we tested, even those isolates that produced cytotoxins plus three or more additional putative virulence factors, showed no signs of pathogenicity in compromised mice, which suggests that the array of in vitro virulence tests currently available for rapid screening of bacterial colonies is insufficient. Bacterial virulence is multifactorial and not well understood; and until we have better understanding of virulence and can develop an additional array of virulence tests, the current tests will give us an overestimation of the number of potentially pathogenic heterotrophic bacteria in potable water. The possibility of unknown opportunistic pathogens in potable water still exists but it is unlikely that any will be found by screening bacterial colonies individually in vivo or by screening large numbers of colonies using the currently available in vitro virulence tests. There are other approaches that are more likely to be successful. One approach would be to test concen-
253
trates of microorganisms from potable waters that contain exceptionally high numbers of heterotrophic bacteria in the animal models. Any organisms that can be isolated from the normally sterile livers and spleens would be regarded as potential pathogens. Before any conclusions could be made, it would be necessary to verify that the same organisms occur in significant numbers in the water. It would also be necessary to perform epidemiological studies to determine whether these organisms also infect humans and whether they occur in large enough numbers to constitute an infectious dose for humans. Human pathogens typically do not grow on diluted media such as R2A, nor do organisms that grow on these media express virulence factors (Edberg et al., 1996). However, there are two exceptions to these rules; Legionella species and nontuberculous mycobacteria are found in potable water and require unusual growth media (Dufour and Jakubowski, 1982; Du Molin and Stottmeier, 1986). Pathogenic legionellae survive and grow in amoebae, and this may relate to their ability to parasitize human cells (Rowbotham, 1980); this may also be true for nontuberculous mycobacteria. Therefore, another approach would be to specifically look for new types of pathogens of amoebae and test these organisms in the CG model for facultative intracellular pathogens. Again, it would be necessary to use an epidemiological approach to verify whether or not these organisms occur in numbers sufficient to infect humans and that they are human as well as mouse pathogens. It would also be possible to start from an epidemiological approach looking for unusual opportunistic infections in a community with a high occurrence of heterotrophic bacteria in its potable water and a community with very low occurrence, then verifying that the causative agent is present in the potable water of the first community but not the second.
5. Summary and conclusions A total of 45 isolates of heterotrophic bacteria were tested for pathogenicity in two animal models, one designed to identify facultative intracellular pathogens and the other to identify extracellular pathogens. Although these isolates were chosen because they expressed multiple putative virulence factors, all were negative for pathogenicity in both models. In addition,
254
G.N. Stelma Jr. et al. / International Journal of Food Microbiology 92 (2004) 249–254
several of these isolates were identified as members of genera or species known to contain opportunistic pathogens. The results provide no evidence for additional opportunistic pathogens on potable water and suggest that the battery of in vitro virulence tests currently available is insufficient to predict pathogenicity for animals. The results also suggest that if there are any currently unknown opportunistic pathogens in potable water, new approaches are needed to identify them.
Acknowledgements We thank Dr. Pierre Payment, Dr. A.D. Hitchins, Dr. James Oliver and Dr. Ashok Chopra for providing the bacterial strains used in this study and Mrs. Doris Morris for assistance in preparing the manuscript.
References Broome, C.V., Fraser, D.W., 1979. Epidemiologic aspects of legionellosis. Epidemiol. Rev. 1, 1 – 16. Calderon, R.G., 1991. Bacteria colonizing point-of-entry, granular activated carbon filters and their relationship to human health. Final Report CR 813978-01-0, USEPA. Calderon, R.G., Mood, E.W., 1988. Bacteria colonizing point-ofuse, granular activated carbon filters and their relationship to human health. Final Report CR 811904-01-0, USEPA. Clark, J.A., Burger, C.A., Sabatinos, L.E., 1982. Characterization of indicator bacteria in municipal raw water, drinking water and main water samples. Can. J. Microbiol. 28, 1002 – 1013. Davis-Hoover, W.J., Dufour, A.P., Stelma Jr., G.N., Scarpino, P.V., 1990. Use of an in vivo model for identifying opportunistic pathogens isolated from drinking water. Proc. AWWA Annu. Conf. Part 2, 1693 – 1706. Dufour, A.P., Jakubowski, W., 1982. Drinking water and Legionnaires disease. J. AWWA, 631 – 637. Du Molin, G.C., Stottmeier, K.D., 1986. Waterborne mycobacteria: an increasing threat to health. ASM News 52, 525 – 529.
Edberg, S.C., Gallo, P., Kontnick, C., 1996. Analysis of the virulence characteristics of bacteria isolated from bottled, water cooler and tap water. Microb. Ecol. Health Dis. 9, 67 – 77. Goslee, S., Wolinsky, E., 1976. Water as a source of potentially pathogenic mycobacteria. Am. Rev. Respir. Dis. 113, 287 – 292. Le Chevallier, M.W., Evans, T.M., Seidler, R.J., Daily, O.P., Merrell, B.R., Rollins, D.M., Joseph, S.W., 1982. Aeromonas sobria in chlorinated drinking water supplies. Microb. Ecol. 8, 325 – 333. Lye, D.J., Dufour, A.P., 1991. A membrane filter procedure for assaying cytotoxic activity in heterotrophic bacteria isolated from drinking water. J. Appl. Bacteriol. 70, 89 – 94. Lye, D.J., Dufour, A.P., 1993. Virulence characteristics of heterotrophic bacteria commonly isolated from potable water. Environ. Toxicol. Water Qual. 8, 13 – 23. Morrison, A.J., Wenzel, R.P., 1984. Epidemiology of infections due to Pseudomonas aeruginosa. Rev. Infect. Dis., S627 – S642. Payment, P., Siemiatycki, J., Richardson, L., Renaud, G., Franco, E., Pr vost, M., 1997. A prospective epidemiological study of gastrointestinal health effects due to the consumption of drinking water. Int. J. Environ. Health Res. 7, 5 – 31. Reasoner, D.J., Geldreich, E.E., 1985. A new medium for the enumeration and subculture of bacteria from potable water. Appl. Environ. Microbiol. 49, 1 – 7. Regunathan, P., Beauman, W.H., 1994. Significance of heterotrophic bacteria in drinking water. Annu. Conf. Water Qual. Assoc. Rowbotham, T.J., 1980. Preliminary report on the pathogenicity of Legionella pneumophila for freshwater amoebae. J. Clin. Pathol. 33, 1179 – 1183. Stelma Jr., G.N., Reyes, A.L., Peeler, J.T., Francis, D.W., Hunt, J.M., Spaulding, P.L., Johnson, C.H., Lovett, J., 1987. Pathogenicity test for Listeria monocytogenes using immunocompromised mice. J. Clin. Microbiol. 25, 2085 – 2089. Stelma Jr., G.N., Reyes, A.L., Peeler, J.T., Johnson, C.H., Spaulding, P.L. 1992. Virulence characteristics of clinical and environmental isolates of Vibrio vulnificus. Appl. Environ. Microbiol. 58, 2776 – 2782. Tamura, Y., Tanaka, S., 1984. Effect of antiflagellar serum in the protection of mice against Clostridium chauvoei. Infect. Immun. 43, 612 – 616. Tatsukawa, K., Mitsuyama, M., Takeya, K., Nomoto, K., 1979. Differing contribution of polymorphonuclear cells and macrophages to protection of mice against Listeria monocytogenes and Pseudomonas aeruginosa. J. Gen. Microbiol. 115, 161 – 166.