Cytotoxic Bacillus spp. belonging to the B. cereus and B. subtilis groups in Norwegian surface waters


Øyvin Østensvik, Department of Food Safety and Infection Biology, The Norwegian School of Veterinary Science, PO Box 8146 Dep., N-0033 Oslo, Norway (e-mail:


Aims:  To investigate the presence and numbers of Bacillus spp. spores in surface waters and examine isolates belonging to the B. cereus and B. subtilis groups for cytotoxicity, and to discuss the presence of cytotoxic Bacillus spp. in surface water as hazard identification in a risk assessment approach in the food industry.

Methods and Results:  Samples from eight different rivers with variable degree of faecal pollution, and two drinking water sources, were heat shocked and examined for the presence of Bacillus spp. spores using membrane filtration followed by cultivation on bovine blood agar plates. Bacillus spp. was present in all samples. The numbers varied from 15 to 1400 CFU 100 ml−1. Pure cultures of 86 Bacillus spp. isolates representing all sampling sites were characterized using colony morphology, atmospheric requirements, spore and sporangium morphology, and API 50 CHB and API 20E. Bacillus spp. representing the B. cereus and B. subtilis groups were isolated from all samples. Twenty-one isolates belonging to the B. cereus and B. subtilis groups, representing eight samples, were screened for cytotoxicity. Nine strains of B. cereus and five strains belonging to the B. subtilis group were cytotoxic.

Conclusions:  The presence of cytotoxic Bacillus spp. in surface water represents a possible source for food contamination. Filtration and chlorination of surface water, the most common drinking water treatment in Norway, do not remove Bacillus spores efficiently. This was confirmed by isolation of spores from tap water samples.

Significance and Impact of the Study:  Contamination of food with water containing low numbers of Bacillus spores implies a risk for bacterial growth in foods. Consequently, high numbers of Bacillus spp. may occur after growth in some products. High numbers of cytotoxic Bacillus spp. in foods may represent a risk for food poisoning.


Aerobic, endospore-forming bacteria are a heterogenic group of micro-organisms, and the genus Bacillus is one of 16 validly published genera in this group (Fritze 2002). Most Bacillus spp. are widely distributed in nature, and because of passive distribution of endospores their occurrence is not necessarily restricted to their natural habitat. In relation to food poisoning and food spoilage, two groups of Bacillus spp. are of special interest; the B. cereus group and the B. subtilis group. Members of the B. cereus group are B. cereus, B. anthracis, B. mycoides, B. thuringiensis, B. pseudomycoides and B. weihenstephanensis (Fritze 2002). The role of B. cereus as a food-borne pathogen was recently reviewed by Granum (2002). Other Bacillus spp., primarily members of the B. subtilis group have also been related to food-borne illness (Kramer and Gilbert 1986; Granum and Baird-Parker 2000). The most important organisms of this group are B. subtilis, B. licheniformis and B. pumilus (Fritze 2002). Additionally, Bacillus spp. has been associated with spoilage of diary products and other foods (Heyndrickx and Scheldeman 2002).

Spores of Bacillus spp. are easily spread from soil and other environments to foods. Water is an environment where spores and vegetative cells of Bacillus spp. are expected to be present. However, the role of water as a risk factor for contamination of food industry with spores of toxigenic Bacillus spp. is not clearly outlined. Bacillus cereus has been isolated from tap water in a hospital where patients with severe burns were infected by B. cereus (Torregrossa et al. 2000). In a Japanese study, B. thuringiensis was isolated from 53 of 107 water samples (Ichimatsu et al. 2000). Spores present in water may pass through simple drinking water treatment, such as filtration and disinfection (chlorination), and then unexpectedly introduce Bacillus spp. into food products. Contamination of food with low numbers of Bacillus spp. does not primarily represent a risk for food poisoning or spoilage, but contamination may result in bacterial growth. Under conditions favourable for bacterial growth, the numbers of bacteria in food may increase significantly within 24 h, and some lightly heat-treated foods are expected to have a shelf life of many weeks. High numbers of Bacillus spp. in foods may cause spoilage and/or food poisoning.

No standard method is described for the isolation and enumeration of Bacillus spp. from water samples. Young et al. (2000) compared a membrane filtration method with centrifugation. The recovery was higher using the membrane filtration technique than centrifugation. A membrane filtration method is also described by Francis et al. (2001). Nonselective agars have been recommended as isolation media (Young et al. 2000; Francis et al. 2001). The identification of Bacillus spp. is traditionally based on phenotypic characteristics, including morphological and biochemical properties. Based on spore and sporangium shape, Bacillus spp. is arranged into five groups (Fritze 2002). In addition, biochemical tests, such as API 50 CHB/API 20 E, are commonly used for species identification (Logan 2002). Genetic methods have been developed during the last decades based on 16S rRNA (Stackebrandt and Swiderski 2002). The B. cereus and the B. subtilis groups form well-separated clusters or subgroups within the Bacillus RNA group 1.

The aims of the present study were to investigate the presence of spores from Bacillus spp. in surface waters, and to examine strains identified as members of the B. cereus and B. subtilis groups for cytotoxicity. To describe the influence of surface runoff and sewage as sources of Bacillus spp. in surface water, samples from a limited selection of river waters with variable degree of faecal pollution were examined. In addition, two drinking water samples were examined for the presence and numbers of Bacillus spp. Presence of cytotoxic Bacillus spp. in surface water will be discussed as hazard identification in a risk assessment approach in the food industry.

Materials and methods


Ten water samples, eight of them from different rivers and two drinking water samples (tap water), were collected in sterile 500 ml glass bottles. The sample sites are described in Table 1. Four river samples (3–6) were chosen from the catchment area and inlets to the lake Maridalsvannet in Oslo, which is the water source for the largest drinking water supply system in Norway. These rivers are located in a recreation area with minor point-source pollution, and were chosen as representatives for rivers mainly receiving surface runoff. Sample numbers 1, 2, 7 and 8 were from rivers with variable degree of faecal pollution. These samples would give information about the numbers of Bacillus spp. in faecally contaminated water, and they were chosen to compare the influence of surface runoff and faecal pollution as sources of Bacillus spp. Sample no. 9 (Table 1) was tap water from The Norwegian School of Veterinary Science, Section of Food Safety, connected to the drinking water distribution in Oslo (Norway). Water from hypolimneon in Maridalsvannet is the raw water, and the water is treated by filtration and chlorination. This water is used without any further treatment by both food and diary industries. The other drinking water sample, no. 10 (Table 1) was tap water from a well in rural Norway, that is used as drinking water for one household without any form of treatment. This water supply was also used on a diary farm until 1995. Private water supplies from wells with a mixture of ground water and surface water are still commonly used in Norwegian diary farms, and this sample was chosen as a representative for such a water supply. All samples were kept refrigerated until analysis. Nine of them were analysed the same day as sampling, and sample number 10 was analysed within 24 h.

Table 1.  Sample sites, date of sampling and characterization of the sample sites
1Akerselva (Nydalsdammen)2002-04-29Downstream the lake Maridalsvannet, the source water of Oslo. Receiving sewage leakages and faecal material from birds and dogs
2Akerselva (Frysjadammen)2002-04-29 
3Skjersjøelva (Hammeren)2002-04-29Rivers in the catchment area of the lake Maridalsvannet. Recreation areas, some wildlife and limited settlement and agricultural activities
4Skarselva (Sørbråtan gård)2002-04-29 
6Dausjøelva (Nesbukta)2002-04-29 
7Sandvikselva2002-04-29Sewage polluted river 20 km west of Oslo
8Sognsvannbekken (Frognerparken)2002-04-29Sewage polluted river (Oslo)
9Drinking water (Oslo, Norway)2002-06-03Tap water, treated drinking water (filtration and chlorination), The Norwegian School of Veterinary Science
10Drinking water (Tynset, Norway)2002-06-03Untreated well water supplying one household in rural Norway, 300 km north from Oslo This water was used in diary farming until 1995

Isolation of Bacillus spp.

Bacillus spp. was isolated and quantified using a membrane filter technique. In order to achieve plates without overgrowth, 1, 10 and 100 ml volumes of heat-shocked samples (80°C for 30 min) were filtered through 0·45 μm membrane filters (Millipore HA, Millipore Corporation, Bedford, MA, USA). After membrane filtration, the membrane filters were placed on bovine blood agar plates and incubated aerobically at 30 ± 1°C for 21 ± 3 h. Membrane filters with more than 70–80 colonies showed various degree of confluent growth, and were not used for quantification. All colonies that grew on the membrane filters, both β-haemolytic and anhaemolytic, were defined as Bacillus spp. The results are given as Bacillus spp. 100 ml−1. Colonies producing β-haemolysis were easily identified by examination of the blood agar plates from the bottom side, and the results were expressed as β-haemolytic Bacillus spp. 100 ml−1.

Identification of Bacillus spp.

Well-isolated colonies, representing the different morphological groups observed, were subcultured on blood agar plates. Subculturing was made from filters with maximum 20–30 colonies. Eighty-six colonies, 16 β-haemolytic and 70 anhaemolytic, representing the 10 sample sites were identified. The number of isolates from each sample is shown in Table 2. To compare the species composition in faecally polluted river water and drinking water an extended picking strategy was used from two of the samples. From sample 8 (a highly faecally polluted river) all 14 colonies from the 1 ml sample were identified. From sample 9 (drinking water) 14 of 38 colonies from the 100 ml sample were identified. These colonies represented all visible different morphologies observed.

Table 2.  Anhaemolytic and β-haemolytic Bacillus spp. colonies counted on membrane filters from 10 surface water samples, numbers of colonies picked for identification and numbers of isolates used in the test for cytotoxicity. Sample site numbers refer to Table 1
Sample siteColonies countedColonies picked
Bacillus spp.β-haemolytic Bacillus spp.Anhaemolytic Bacillus spp.β-haemolytic Bacillus spp.For cytotoxicity

After subculturing, typical B. cereus-colonies were defined as circular, opaque, β-haemolytic colonies with a diameter of 2–4 mm. Further tests to differentiate between B. cereus, B. weihenstephanensis and B. thuringiensis were not carried out. Bacillus mycoides was defined as β-haemolytic, rhizoid colonies on blood agar plates. Anhaemolytic, and colonies with weak β-haemolysis, were not identified directly by morphological characteristics. These colonies were tested for for their ability to grow anaerobically after 24 and 48 h at 30 ± 1°C on blood agar plates that were preincubated anaerobically before use. Further identification included spore and sporangium morphology according to the classification scheme described by Fritze (2002). Gram staining was made from colonies on blood agar plates that had been kept at room temperature for 1 week. Additionally, 63 of the 68 anhaemolytic isolates were typed by API 50 CHB and API 20 E (bioMérieux, Marcy l‘Etiole, France).

Production of crude toxin extracts

Cell extracts for the cytotoxicity test were produced as follows: 5 ml of BHIG (brain–heart infusion; Difco, with 1% glucose added) were inoculated with pure Bacillus spp. cultures and incubated overnight at 32°C with agitation (ca 100 rpm). From the overnight cultures 0·5 ml were transferred to 50 ml BHIG and incubated at 32°C with agitation for 6 h. Extra cellular components were harvested by centrifugation of the cultures at 12 000 g at 4°C for 20 min. Aliquotes of the supernatants containing the extra cellular components were immediately tested, or frozen at −20°C.

Testing for cytotoxicity

From the 86 isolates Bacillus spp. identified a random selection of 21 strains, representing the B. cereus and B. subtilis groups were tested for cytotoxicity. The strains were selected from eight of the sample sites, and the number of isolates from each sample site is presented in Table 2. The selection included 12 B. cereus, four B. pumilus, three B. licheniformis and two B. subtilis isolates. The assay for cytotoxicity is described by Sandvig and Olsnes (1982), and measures the inhibition of protein synthesis in Vero cells, caused by the toxin(s). Cytotoxins reduce the incorporation of 14C-leucine into proteins, so that cytotoxic strains obtain very low radioactive counts. For each of the strains, two parallels of 100 μl crude toxin extracts were applied on Vero cell monolayers. Supernatant of the B. cereus reference strains 1230–88 (Lund and Granum 1997) was used as positive control (100 μl). Growth medium for Vero cells without addition of bacterial extracts was used as negative control. The percentage inhibition of protein synthesis was calculated from the mean of the duplicates: 100% −(100 × mean count of test/mean count of negative control –mean count of positive control).

Faecal indicator bacteria

Coliform bacteria and Escherichia coli were quantified using Colilert-18/QuantitrayTM (IDEXX Laboratories, Westbrook, ME, USA), according to the manufacturer's instruction, except for the incubation temperature (37 ± 1°C).


Presence of Bacillus spp.

Bacillus spp. was isolated from all samples (Table 3), and the numbers varied from 15 to 1400 CFU 100 ml−1. Beta-haemolytic Bacillus spp. was present in all samples in numbers ranging from 1 to 160 CFU 100 ml−1.

Table 3.  Coliform bacteria (CB), Escherichia coli (EC), Bacillus spp. (including β-haemolytic Bacillus spp.) and β-haemolytic Bacillus spp. in eight river and two drinking water samples. Sample site numbers refer to Table 1
Sample siteCB (100 ml−1)EC (100 ml−1)Bacillus spp. (100 ml−1)β-haemolytic Bacillus spp. (100 ml−1)
811 53012201400100

Identification of Bacillus spp.

A selection of anhaemolytic and β-haemolytic colonies, representing different morphologies, were picked for species identification (Table 2). Species identification of 86 Bacillus spp. representing the 10 samples is shown in Table 4. The identification was based on the combination of colony morphology on blood agar, atmospheric requirements, cell and spore morphology and API 50CHB/API 20E identification. The majority of the Bacillus spp. isolates belonged to the B. cereus and B. subtilis groups. Bacillus cereus, B. subtilis, B. pumilus and B. licheniformis were the most frequently identified species. Bacillus brevis, B. sphaericus and B. circulans were identified in low numbers. Thirteen isolates (15·1%) could not be identified to species with the methods used in this study.

Table 4.  Species identification of 86 Bacillus spp. from water. The diagnosis are based on the combination of colony morphology on blood agar, atmospheric requirements, cell and spore morphology and API 50CHB/20E-diagnosis
SpeciesTotal numberPercentage
B. cereus16 18·6
B. mycoides2  2·3
B. megaterium4  4·7
B. subtilis16 18·6
B. pumilus14 16·3
B. licheniformis12 14·0
B. brevis6  7·0
B. circulans1  1·2
B. sphaericus2  2·3
Unclassified13 15·1


From eight different sample sites 12 B. cereus isolates and nine isolates belonging to the B. subtilis group were screened for cytotoxicity in a cell test (Table 2). Fifteen isolates were classified as moderatly cytotoxic (Table 5). Cytotoxic Bacillus spp.were isolated from all eight sample sites. Nine B. cereus isolates from six different sample sites and six isolates representing the B. subtilis group from three different sample sites were cytotoxic. Six isolates did not show cytotoxicity. These six isolates represented B. cereus, B. pumilus and B. licheniformis. The percentage cytotoxic β-haemolytic and anhaemolytic isolates was 75 and 67, respectively. Interestingly, the highest percentage inhibition of 14C-leucine observed, 71%, was caused by a B. pumilus isolate. No clear relationship between β-haemolytic and anhaemolytic strains and cytotoxicity was observed (Table 6).

Table 5.  Cytotoxicity of 21 Bacillus spp. isolates from water in a leucine-starved Vero-cell test*
Bacillus speciesNumber testedHighModerateNegative
  1. *Cytotoxicity tested on 100 μl supernatant (crude toxin). High toxicity: 80–100% inhibition of protein synthesis; moderate toxicity: 20–80% inhibition of protein synthesis; negative toxicity: <20% inhibition of protein synthesis.

B. cereus12093
B. pumilus4031
B. licheniformis3012
B. subtilis2020
Table 6.  Relation between β-haemolysis on bovine blood agar plates and Vero cell cytotoxicity from 21 isolates of Bacillus spp. isolated from surface water
Sum 12921

Relation to faecal pollution

The numbers of Bacillus spp. and E. coli were compared. The highest counts for Bacillus spp., 570 and 1400 CFU 100 ml−1, were observed in the samples with the highest E. coli counts, 496 and 1220 MPN 100 ml−1, respectively (Table 3). From four samples, E. coli was not detected, and the numbers of Bacillus spp. in these samples varied from 15 to 65 CFU 100 ml−1. Escherichia coli and coliform bacteria were not detected in the two drinking water samples, and the numbers of Bacillus spp. in these samples were 15 and 38 CFU 100 ml−1. Bacillus cereus was present in both drinking water samples. The Bacillus spp. composition from one faecally contaminated river sample and one drinking water sample did not show obvious differences. From both of these samples B. cereus, B. subtilis, B. licheniformis and B. pumilus were isolated.


Cytotoxic Bacillus spp. was isolated from eight different surface water samples. The toxic strains included B. cereus, B. pumilus, B. licheniformis and B. subtilis. To our knowledge, the presence of toxic Bacillus spp. in surface water has not previously been reported. However, toxigenic strains of B. cereus are frequently isolated from foods and faecal samples derived from patients with diarrhoea. Toxigenic strains of B. licheniformis, B. pumilus and B. subtilis have previously been reported from food samples, and have also been linked to cases of foodborne disease (Kramer and Gilbert 1986; Salkinoja-Salonen et al. 1999; Rowan et al. 2001; Phelps and McKillip 2002). From all 10 water samples both haemolytic and anhaemolytic Bacillus spp. were detected. However, the numbers of spores in the two drinking water samples and the samples from the rivers with minor faecal pollution were low compared with river water containing high levels of faecal indicator bacteria. This observation indicates that both surface runoff from the environment and wastewater are important sources of Bacillus spp. in water. In a sample of untreated wastewater from Bekkelaget Wastewater Treatment Plant in Oslo (Norway), the numbers of Bacillus spp. and β-haemolytic Bacillus spp. were 490 and 100 CFU ml−1, respectively (Ø. Østensvik, unpublished data). In the Vero cell cytotoxicity assay, both β-haemolytic and anhaemolytic Bacillus spp. showed cytotoxicity. No clear correlation between haemolysis and cytotoxicity was observed. This is not surprising, as most Bacillus spp. that belong to the B. subtilis group are anhaemolytic. However, some strains of B. subtilis may show a weak β-haemolysis on blood agar plates. Interestingly, the strain with the highest cytotoxicity was an anhaemolytic B. pumilus, indicating that cytotoxicity is not necessarily linked to β-haemolysis.

Bacillus spp. was isolated from water using a membrane filter technique, which allows filtration of a wide range of sample volumes. The use of bovine blood agar plates as isolation medium made it possible to differentiate between β-haemolytic and anhaemolytic colonies. After 1 day of incubation, the presence of B. cereus was observed with a high degree of certainty. The members of the B. subtilis group could not be differentiated using colony morphology on blood agar as the only criterion. Further identification of anhaemolytic isolates was necessary.

The presence of toxigenic Bacillus spp. in surface water indicates that water may play a role as a risk factor for introducing food poisoning Bacillus spp. to the food industry. Spores are present in the source water, and may pass through drinking water treatment (filtration and chlorination). Five of the sample sites represented the largest drinking water supply system in Norway (Oslo). Four of these samples were taken from rivers in the catchment area and absence or low numbers of E. coli in these samples indicated that surface runoff was the main source of Bacillus spp. (Table 3). From the treated drinking water sample Bacillus spp. was detected in numbers in the same order of magnitude as the samples from the catchment area. These results agree with the hypothesis that filtration and chlorination have minor or no effects on removal of Bacillus spp. spores in drinking water. In a study carried out in 2002 and 2003, the numbers of Bacillus spp. at this drinking water treatment plant in Oslo were reduced with 11% as an average from six samples (Ø. Østensvik, unpublished data). The results from the present study show that spores of Bacillus spp. may be present in drinking water, which satisfy the microbiological regulations (no coliform bacteria and E. coli detected in 100 ml volumes). In the diary industry, several investigations have pointed to milk as the most important source for B. cereus contamination (Sutherland and Murdoch 1994; Lin et al. 1998; Heyndrickx and Scheldeman 2002). The present study indicates that water may be another possible source of cytotoxic Bacillus spp. in diary plants and other food industries using insufficiently treated surface water. In Norway, the majority of the drinking water supply systems use surface water as source water. The most common treatment is filtration and chlorination, and spores of Bacillus spp. were isolated from treated drinking water. Consequently, Bacillus spp. spores present in drinking water can be transferred to food products. The present study includes few samples. However, the results suggest that surface water treated by filtration and chlorination can be regarded as a hazard for contamination in the food industry. To achieve more knowledge of water as a risk factor for contamination of food with cytotoxic Bacillus spp., further investigations to characterize this hazard should be carried out. The cytotoxic strains we have isolated are currently investigated for the presence of possible food poisoning virulence factors.