MICROBIAL ECOLOGY IN HEALTH AND DISEASE

VOL.9: 67-77 (1996)

Analysis of the Virulence Characteristics of Bacteria Isolated from Bottled, Water Cooler, and Tap Water S. C. EDBERG*?$, P. GALLOf and C. KONTNICKf ?Department of Laboratory Medicine and Internal Medicine, Yale University Scliool of Medicine and f Clinical Microbiology Laboratory, Yale New Haven Hospital, New Haven, CT. USA Received 4 January 1995; revised 12 December 1995

A study was conducted to determine the numbers and types of bacteria found in three water sources-bottled water, water cooler water, and tap water-and to determine their virulence characteristics. A wide variety of water types were collected and each was analysed for Heterotrophic Plate Count (HPC) bacteria, Pseudomonas aeruginosa, and total coliforms. For each isolate, virulence characteristics were determined by enzyme analysis (10 associated with virulence), antibiotic susceptibility testing (natural and semi-synthetic antibiotics), acid lability (survival a t pH 3.5). and cytotoxicity testing (HEp-2 cells). Results showed that all water sources had a normal bacterial content. Only 2 per cent of bottled water sources had P. aeruginosa. Total coliforms were isolated only from bottled water that used mixed (water alternating with milk) filling lines. Environmental bacteria did not produce significant enzymes associated with virulence, were not acid resistant, were susceptible to semi-synthetic antibiotics, and did not produce appreciable cytotoxicity. These natural aqueous bacteria were adopted to a water environment, did not grow well at conditions analogous to the human host, and did not have the characteristics associated with virulence. Future drinking water revisions and changes to the treatment processes should be directed towards the elimination of specific pathogens and to the prevention of exogenous sources of contamination rather than the elimination of natural water microbial populations. KEY WORDS: virulence factors; drinking water; health risk; heterotrophic bacteria; water regulations.

INTRODUCTION There are three primary factors that will determine if a host will develop an infection.12First, there is the number of the particular microbe the host contacts. Second, there is the virulence component the microbe possesses. Third, there is the resistance of the challenged target organ system of the host. The assessment of health risk for each microbe concerns the nature of its pathogenesis, the vehicle of delivery, the ability of the host to respond, and the particular microenvironment to activate the virulence factor^.^^'^^'^^^'*^^ There are a variety of direct ways to assay the number of microbes challenging the host.46 Likewise, it is possible to assess the immunological status of the gastrointestinal system." The means to determine pathogenic *Author to whom correspondence shouid be addressed at: Department of Laboratory Medicine, Yale University School of Medicine, Box 208035, New Haven, CT 06520-8035, USA.

CCC 0891-060W96/020067-11

0 1996 by John Wiley & Sons, Ltd.

potential has been studied in the last years and a number of specific virulence factors associated with gastrointestinal pathogenesis has been elucidated.l o There have been a number of recent major reviews of drinking water regulations in the European Community and the USA in the last several years. ",18 These regulations have ascertained that the principal threat to human health is from the primary pathogens that may enter the water distribution system and escape the treatment processes. However, there has been increased interest in attempting to assess risk from the natural microbiota found in drinking water. These microbes have been referred to as the heterotrophic plate count bacteria (HPC).4M8 If these natural HPC could be shown to have significant health risk characteristics, then the nature of water treatment and distribution would be significantly affected. Epidemiological studies that examined the possible health effects of HPC in drinking water have

68

yielded mixed results. Payment et al.,42*43 studying reverse osmosis (filtered and non-filtered tap water) published some data that showed an association of HPC concentrations with gastroenteritis and other data that did not. Calderon and M ~ o d , in ~ . two ~ US Environmental Protection Agency sponsored studies, found no association of HPC concentrations with point-of-entry devices and gastroenteritis. In order to rationally assess health risk from naturally occurring HPC, it is critical to determine if these microbes posses virulence com onents known to be associated with p a t h o g e n e ~ i s . ~ Epi’~ demiological studies involving human population groups are complex, with many variables, and expensive. Animal experimentation studies suffer from a number of inherent difficulties. Animal gastrointestinal systems can be significantly different from the human, animals have to be unrealistically immunosuppressed in order to bring about an infection, and animal studies are very expensive. Therefore, it is not likely that animal models will be generated that will be able to determine the risk from the broad variety of HPC bacteria we daily ingest. Accordingly, because virulence factors are essential for the ability of a microbe to cause disease, determination of the presence or absence of these factors in naturally occurring HPC will be of great help to epidemiologists, medical microbiologists, and regulators in determining the future course of drinking water regulations. Moreover, an assessment of these virulence factors can assist hospital and community based health care professionals in determining if individuals in particular high risk groups should have modified drinking water intake. There have been several extracellular enzymes and microbial products described which have been associated with bacterial pathogenesis.16*19.27*37 These include coagulase, which can sequester a microbe; elastase, which dissolves cellular glue; proteinase, which disrupts the body’s proteins; gelatinase, which hydrolyses the body’s proteins; haemolysin, which destroys red blood cells; DNAse, which destroys the nucleic acid of cells; lipase, which dissolves lipids; and lecithinase, which dissolves a cell’s membrane^.^^'^*^^,^*,^' While the possession of any one of these virulence factors does not mean that a microbe can cause disease, it is generally believed that their known association with pathogenesis is one requirement for the development of infection. Moreover, recent work has allowed a determination of overall

S. C. EDBERG ETAL.

virulence potential, cytotoxicity, to be performed on large number of microbial isolate^.^'.^^ Therefore, a study was conducted to determine the numbers and types of bacteria found in the three types of drinking water: tap water, bottled water, and water obtained from water coolers. From each of the HPC bacteria isolated from each of the drinking water types, the presence or absence of virulence factors associated with bacterial pathogenesis was determined. This study.was the first to examine the virulence characteristics of HPC isolated simultaneously from the three major sources of drinking water-tap water, bottled water, and cooler water. MATERIALS AND METHODS Water sources examined Tap water. Tap water from distribution systems in three areas of the United States were examined. All tap water samples were sent by overnight delivery using cold packs to maintain a temperature between 4” and 10°C.These areas included the northeast, midwest, and south. Fifty tap water samples were examined from each site. All tap waters came from distribution systems in full compliance with the US Environmental Protection Agency regulations.” Bottled water. Bottles of less than or equal to 1.5 litres were bought from stores throughout the United States and Canada. These included waters bottled in the US, Canada, and Europe. Different brands of water were chosen to provide the greatest variety possible. All bottles were purchased from supermarket shelves and represented those which the consumer could obtain. Because some brands were national and others local, there was not an even distribution of each water sold. The number of times a particular brand was tested reflected its distribution across the United States. Gallon bottles (1 gallon) were purchased from supermarkets off the shelves at the same time the smaller bottles were obtained. Gallon bottles were of two basic types: ‘label brands’ and ‘supermarket brands’. Label brands were waters manufactured by a company not affiliated with a supermarket chain and were sold in more than one particular store chain. Supermarket brands were bottled waters manufactured either by a supermarket chain or for a particular supermarket chain and sold only in that chain.

WATER BACTERIAL VIRULENCE CHARACTERISTICS

Cooler water. Cooler water was obtained in sealed 5 gallon bottles. All cooler water was natural (and ozonated) spring water. Water coolers contained two spigots: room temperature water and chilled water. The chilled water spigot was utilised to collect water for analysis because this was the tap more commonly used. Coolers were installed, sanitised, and maintained according to directions from the manufacturer and in accordance with guidelines established by the International Bottled Water Association (Alexandria, VA, USA). Nine coolers were utilised during the study. Coolers were installed and decontaminated with a hypochlorite solution and then thoroughly rinsed to remove residual chlorine. Each bottle was sampled for bacterial analytes before it was placed on the cooler. On a weekly basis for 9 wks each of the coolers was sampled. The nine locations were varied and included lunch rooms and workshops. Microbiological methods Total microbiological counts (Heterotrophic Plate Count [HPC] bacteria) were made using R2A agar4' (Difco, Detroit, MI, USA). Incubation was at 37°C for 3d. These conditions were chosen because they were recommended by the U.S. Environmental Protection Agency.34 For each water sample, 1.0 and 0.01 ml was spread on the surface of an R2A agar plate. After incubation, each colony type was counted with the aid of a dissecting microscope. Each colony type was transferred to R2A agar to ensure purity. From each colony a Gram stain was made. If the Gram stain was not definitive to determine if the isolate was gram-positive or gram-negative, the isolated colony was grown for 4 to 6 h in trypticase soy broth (BBL, Cockeysville, MD, USA). A Gram stain was then made from log phase growth. In addition, each bacterial isolate was subcultured to trypticase soy agar with 5 per cent sheep blood (BD Biological Systems, Cockeysville, MD, USA) to determine if it was able to grow on media with constituents analogous to the human state. The database of Palleroni4' was used for identification of the isolates. If a bacterial isolate was not definitively identified by this approach, it was further analysed based on the analysis of their metabolic products by Gas-liquid chromatography (MIDI system)." Because different species of HPC can appear similar on the surface of R2A agar, at least three identical colonies from each isolation plate were analysed.

69 In addition, each water sample was processed for the presence of P. aeruginosa and the total coliform group. For P. aeruginosa, a 250ml sample was filtered through a 0.45 micron filter and placed on the surface of m-PA agar and incubated at 42°C.' Each colony was identified according to an established d a t a b a ~ e . ~ ' For total coliforms the Colilert method (Idexx, Portland, ME, USA) was ~ t i l i s e d . ' ~This . ' ~ method has been approved by the US Environmental Protection Agency for the direct detection of both total coliforms and Escherichia coli from. water samples. A standard 100 ml sample was tested for each water. The Colilert method has the particular advantage for the detection of coliforms because it is refractory to the presence of high concentrations of HPC bacteria.8 Virulence methods The virulence tests fell into three basic categories: enzyme analysis and acid lability; antibiotic susceptibility analysis; and cytotoxicity potential. Enzyme analysis. Chondroitinase: Chondroitin sulphate type A was obtained from whale shark chondroitin (Sigma, St. Louis, MO, USA), incorporated in agar, spot inoculated with a pure culture from a colony, incubated for 48 h at 37"C, after which the plates were flooded with 2 N acetic acid, with the acid allowed to remain in contact with the plate for 10min. The appearance of a clear zone around the colonies was a positive test.54 Coagulase: Rabbit citrated plasma containing EDTA (ethylenediaminetetraacetate) was dispensed in 1 ml aliquots, each bacterial colony was inoculated to produce a turbid suspension, and incubated for 48 h at 37°C. The development of a clot represented a positive test.22 DNAse: DNase agar (Difco) supplemented with 0-01 per cent toluidine blue was employed as the substrate. Bacterial colonies were inoculated on the surface of the agar and allowed to incubate for 72h at 37°C. The development of a red colour or the development of a zone of clearing after flooding of the plate with 0.1 per cent 1 N HCl represented a positive test.39 Elastase: Elastin powder from bovine neck ligament (Sigma) was made as a 1 per cent suspension in agar plates. Colonies were inoculated onto the surface of the agar and initially incubated for 48 h at 37°C. Subsequently, the plates were held at

70 room temperature for up to 5d. Clearing of the opaque medium around the inoculum spot represented a positive test.5‘ Fibrinolysin: Fibrinogen (Sigma) was mixed with nutrient agar to yield a final concentration of 280 mg per 100 ml. Colonies were inoculated on the surface of the agar and allowed to incubate for 48 h at 37°C. Clear zones around the inoculum spot with diameters greater than 2 mm indicated the lysis of human fibrinogen and represented a positive test.27 Gelatinase: X-ray film served as the gelatin substrate support. A turbid bacterial suspension was made in 1.0ml of distilled water. A gelatin strip of x-ray film of approximately 3 mm by 20 mm was placed in the suspension. The substrate film was incubated for 48 h at 37’C with the removal of the gelatin coating from the surface of the x-ray film evidence of a positive test.I5 Haemolysin: Five per cent defibrinated sheep blood agar in a trypticase soy agar base (Difco) was the substrate. Colonies were inoculated onto the surface of the agar with incubation at 72 h at 37°C. Lysing of the red blood cells indicated by a clear halo around the inoculum spot was evidence of a positive test.” Hyaluronidase: Hyaluronic acid from human umbilical cord (Sigma) served as a substrate. It was incorporated into nutrient agar with colonies inoculated onto the surface. Incubation occurred for 48 h at 37°C after which the surface of the plate was flooded with 2 N acetic acid. After IOmin, clearing around the inoculum spot was evidence of a positive test.54 Lecithinase: Lecithinase was measured by utilising an egg yolk (50 per cent) agar base (Difco). Colonies were inoculated onto the surface of the agar and allowed to incubate for 72 h at 37°C. A white precipitate around or beneath the inoculum spot was taken as a positive test.39 Lipase: Trypticase soy agar plates supplemented with 1 per cent tween 80 (polyoxyethylene sorbitan monooleate) served as a substrate. Colonies were inoculated onto the surface of the agar plate and allowed to incubate for 72 h at 37°C. The appearance of a turbid halo around the inoculum spot was taken as evidence of a positive test.” Proteinase: Skim milk agar incorporated into dialysed brain heart infusion broth base (Difco) was the substrate. Colonies were inoculated onto the surface of the agar and allowed to incubate for 72 h at 37’C. A zone of clearing around the inoculum spot was evidence of a positive test.55

S. C. EDBERG ET

AL

Pyocyanin and Fluorescein: The presence of pyocyanin and fluorescein from Pseudornonas aeruginosa was determined utilising P agar (pyocyanin) and F agar (fluorescein) as the substrates (Difco). Production of pigment or fluorescence were taken as positive tests.39 Acid lability. From the surface of R2A agar, each bacterial isolate was suspended in sterile, dechlorinated tap water to produce a 1.0 McFarland turbidity equivalent. This suspension was kept at room temperature for 24 h in order for bacteria to adapt to the aqueous environment. A previously titrated amount of acetic acid was mixed with 0.1 ml of this suspension to produce a final pH of 3.5. After rapid mixing, incubation was at room temperature for 10 min. 0.1 ml were subcultured onto a nutrient agar plate which was buffered with HEPES to produce a final pH of 7.5. This buffered agar plate neutralised the acetic acid while allowing the bacteria to grow. Percent survival was calculated based on the initial inoculum concentration obtained from quantitative plating of the bacterial suspension after 24 h at room temperature and the final number of CFU/ml that survived on the buffered agar plate. Antibiotic susceptibility analysis. Two basic classes of antibiotics were tested. First, there was a representative from the major groups of original natural and first generation antibiotics. Second, there were representatives of major groups of the synthetic and later generation antibiotics. All bacterial isolates were tested to both groups by a commonly quality controlled disk diffusion method with the use of R2A agar.40 The natural and first generation antibiotics included: penicillin, ampicillin, erythromycin, kanamycin, streptomycin, tetracycline, chloramphenicol, and sulphonamide. The synthetic and later generation antibiotics included: gentamicin, cefoxitin, cefoperazone, oxacillin, piperacillin, imipenem, ciprofloxacin, and sulphonamideftrimethoprim. Cytotoxicity analysis. The cytotoxicity test was performed as described by the US Environmental Protection Agency.34 Briefly, HEp-2 cells were seeded at approximately 2 x lo5 cells per ml into 60 mm tissue culture plates and incubated to form a monolayer. Five ml of agar medium was placed on the surface of the tissue culture monolayers. Because membrane filters from drinking water often have bacterial colonies that overlap and can

71

WATER BACTERIAL VIRULENCE CHARACTERISTICS

produce cytotoxicity results that are not related to a specific individual colony type, in this study three colonies representing each distinct HPC morphotype were selected and passaged in 5 per cent sheep blood agar (Becton-Dickinson, Baltimore, MD, USA) with Columbia agar base prior to examination. Therefore, whereas other studies have only utilised total membranes in contact with the surface of cell c ~ l t u r e s , ~this ~ ' ~study * additionally examined individual HPC colonies. The surface of the agar overlay was contacted from each of three identical colony types from the surface of the blood agar plate. After exposure for 2 h, the contact was cleansed by washing three times in tissue culture medium (Hams F10 growth medium supplemented with horse serum and fetal calf serum). Cells were incubated and examined for changes in the monolayers. After 48 h incubation, the cells were fixed and stained. A cytopathic effect was one in which there was a distinct well demarcated circular area of complete monolayer destruction. Salmonella typhi served as the positive control and bacterial culture medium served as the negative control. RESULTS Bacteriological analysis Table 1 presents the bacteriological findings from the three water sources. A large variety of bacteria were isolated from the various water samples. Approximately 95 per cent of bacterial colonies isolated from R2A agar did not grow when transferred to blood agar. In order to be able to present the results of the microbial analyses in a cogent tabular fashion, the results for each water source was averaged. For each of the water sources the species name of the individual bacterial isolate is presented and the average number of samples containing the species is presented with the concentration range of that particular bacterial species in the sample. Each of the water sources had a naturally occurring HPC concentration. There were somewhat different species present in each of the different water sources but there was also considerable variation within each grouping (bottled water, tap water, and cooler water). A small number of both tap water and bottled water sources contained Pseudomonas aeruginosa, but only in very small numbers. Total coliforms were only isolated from bottled water made from a supermarket that used

filling lines which alternated between milk and water. These mixed lines use a single set of pipes for both milk filling and water filling. It should be noted that the two bottles containing total coliforms were produced by a supermarket in one location. Virulence test results Table 2 presents the percentage of isolates possessing a positive virulence characteristic for each water source. Also included are the results of the acid lability test. Except for the enzyme proteinase, there was a very low prevalence of enzymes associated with bacterial pathogenesis from any of the water sources. In all cases, except for proteinase (produced by 20 per cent of isolates), less than 6 per cent of bacterial isolates produced a single extracellular enzyme associated with virulence. It is generally considered necessary to contain more than one extracellular enzyme in order for a microbe to have sufficient armaments to be virulent. 19*27*28*34 Antibiotic susceptibility test results Table 3 presents the results of the antibiotic susceptibility tests for the two broad categories of antibiotics examined. There was a clear differentiation in the susceptibilities in the two groups of antibiotics tested. There was generally a small percentage of HPC susceptible to the early generation antibiotics and a greater than 95 per cent susceptibility to the later generation of synthetic antibiotics. Cytotoxicity test results Each bacterial isolate was examined for its ability to damage HEp-2 cells. Table 4 presents the results of these cytotoxicity tests. There was generally a small percentage of HPC that were able to demonstrate cytotoxicity. There was no difference in cytotoxicity characteristics among the three water sources examined.

DISCUSSION All natural a ueous systems have a bacterial c~ntent.*'.? ~6*48,52*61 ~ , ~ ~ , These bacteria are adopted to a water environment and do not find the conditions of the human host congenial. Because naturally occurrin HPC go through life cycles in a container,25,33*35.3'onewould expect at any single point in time to only obtain a snapshot of the

72

S. C. EDBERG

ET AL.

Table 1. Microbial analysis of drinking water

Species name

Acinetobacter sp. Achrornobacter sp. Agrobacterium sp. Bacillus bre vis Bacillus cereus Bacillus licheniformis Bacillus circulans Bacillus firmus Bacillus megaterium Bacillus polymyxa Bacillus pumilus Bacillus sphericus Bacillus macerans CDC Group IV (GNR)’ Corynebacteriwn sp. Coryneform sp. Methanococcus Moraxella sp. Comamonas acidovorans Pseudomonas aeruginosa Burkholderia cepacia Pseudomonas juorescens Comamonas testosteroni Staphylococcus sp. Streptomyces sp. Xanthomonas maltophilia Total coliforms

Bottled water (n= 150)”

Cooler water (n=81)b

Tap water (n= I50)c

% Samples with species (concentration range CFU/ml)

% Samples with species (concentration range

% Samples with species (concentration range CFU/ml)

5 (2-30) 5 (8-31) 2 (7-12) 0 0 0 50 (1900-2 1 000) 35 (600-39 000) 15 (21-4500) I5 (18-1700) 0 0 25 (660-68 000) 1 (45)

10 (8-40) 0 85 (30-76 000) 10 (3-38) 6 (1-28) 3 (2-26) 3 (1-15) 12 (1-65) 2 (3-10) 2 (25-600) 0 2 (2-22) 2 samples ‘Supermarket”: (2, m c 3

CFU/ml) 10 (100-350)

5 (6-21)

< 5 (>20) 20 (2-1 10)

4 (5-25)

0 0 0 80 (3-17 000) 20 (900-35 000) 20 (55-600) 20 (80-2100) 0 0 30 (200-45 000) < 5 (