Bacillus licheniformis - Applied and Environmental Microbiology

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Oct. 1999, p. 4637–4645 0099-2240/99/$04.00⫹0 Copyright © 1999, American Society for Microbiology. All Rights Reserved.

Vol. 65, No. 10

Toxigenic Strains of Bacillus licheniformis Related to Food Poisoning ¨ MPFER,2 M. C. ANDERSSON,3 M. S. SALKINOJA-SALONEN,1* R. VUORIO,1 M. A. ANDERSSON,1 P. KA 4 T. HONKANEN-BUZALSKI, AND A. C. SCOGING5 Department of Applied Chemistry and Microbiology1 and Animal Reproduction, Department of Clinical Sciences, Saarentaus,3 FIN-00014 University of Helsinki, and Department of Food Microbiology, National Veterinary and Food Research Institute (EELA), 00231 Helsinki,4 Finland; Institut fu ¨r Angewandte Mikrobiologie, Justus-Liebig Universita ¨t, D-35390 Giessen, Germany2; and Food Hygiene Laboratory, Central Public Health Laboratory, Public Health Laboratory Service, London NW9 5HT, United Kingdom5 Received 18 November 1998/Accepted 5 May 1999

Toxin-producing isolates of Bacillus licheniformis were obtained from foods involved in food poisoning incidents, from raw milk, and from industrially produced baby food. The toxin detection method, based on the inhibition of boar spermatozoan motility, has been shown previously to be a sensitive assay for the emetic toxin of Bacillus cereus, cereulide. Cell extracts of the toxigenic B. licheniformis isolates inhibited sperm motility, damaged cell membrane integrity, depleted cellular ATP, and swelled the acrosome, but no mitochondrial damage was observed. The responsible agent from the B. licheniformis isolates was partially purified. It showed physicochemical properties similar to those of cereulide, despite having very different biological activity. The toxic agent was nonproteinaceous; soluble in 50 and 100% methanol; and insensitive to heat, protease, and acid or alkali and of a molecular mass smaller than 10,000 g molⴚ1. The toxic B. licheniformis isolates inhibited growth of Corynebacterium renale DSM 20688T, but not all inhibitory isolates were sperm toxic. The food poisoningrelated isolates were beta-hemolytic, grew anaerobically and at 55°C but not at 10°C, and were nondistinguishable from the type strain of B. licheniformis, DSM 13T, by a broad spectrum of biochemical tests. Ribotyping revealed more diversity; the toxin producers were divided among four ribotypes when cut with PvuII and among six when cut with EcoRI, but many of the ribotypes also contained nontoxigenic isolates. When ribotyped with PvuII, most toxin-producing isolates shared bands at 2.8 ⴞ 0.2, 4.9 ⴞ 0.3, and 11.7 ⴞ 0.5 or 13.1 ⴞ 0.8 kb. Bacillus licheniformis, Bacillus subtilis, and Bacillus pumilus comprise the subtilis group, which has been associated with a range of clinical conditions, food spoilage such as ropy bread, and incidents of food-borne gastroenteritis (27). B. licheniformis has also been associated with septicemia, peritonitis, ophthalmitis, and food poisoning in humans, as well as with bovine toxemia and abortions (14, 28). B. licheniformis is a common contaminant of dairy products (7). Most food poisoning incidents attributed to Bacillus species are associated with Bacillus cereus, but the relevance of the subtilis group as food poisoning organisms is being increasingly recognized. B. cereus toxins have been well documented (12), but involvement of toxins produced by B. licheniformis has not yet been demonstrated. Food-borne B. licheniformis outbreaks are predominantly associated with cooked meats and vegetables (20, 24, 26). We report here on toxin-producing isolates of B. licheniformis obtained from foods involved in food poisoning incidents, from raw milk, and from industrially produced baby food. (A preliminary report on this work has been presented previously [29].)

elsewhere (3). The bacteria were cultured for toxicity assays on blood or brain heart infusion (Difco Laboratories, Detroit, Mich.) or in Trypticase soy agar (LAB M, Bury, England) plates or Trypticase soy broth. Temperature tolerance was tested at 10°C (with B. cereus DSM 31T as a positive control) and at 55°C (⫾0.5°C) in liquid medium on a shaking incubator (150 rpm) and at 28, 37, and 55°C (⫾0.5°C) on agar plates. Biological analyses. The following phenotypic traits were assayed as described by Smibert and Krieg (25): hydrolysis of lecithin, of starch, and of casein and lysozyme resistance (7 days, 25°C); hemolysis (bovine blood agar plates, read after 24 and 72 h at 37°C); and anaerobic growth (read after 1 and 7 days, 37°C). The brain heart infusion agar plates were preincubated in anaerobic chambers before use. Lipase activity was assayed with modified Sierra lipolysis agar containing peptone (25) (10 g), NaCl (5 g), CaCl2 (0.1 g), beef extract (3 g), ferrous citrate (0.2 g), and agar (15 g) per liter. After autoclaving, 0.5 ml of Victoria Blue B (stock, 0.1 g/150 ml) and 0.1 ml of Tween 80 were added per 10 ml of the medium. Microconcentrator membranes were obtained from Amicon Ltd., Stonehouse, United Kingdom. Physiological tests. The food poisoning isolates (coded F) of B. licheniformis were identified by 25 phenotypic and biochemical tests as described in reference 10. Good anaerobic growth and utilization of propionate were used to distinguish the strains from B. subtilis. All B. licheniformis isolates were characterized by using API 50 CH cassettes (bioMe´rieux, Marcy l’Etoile, France), read after 24 and 48 h at 37°C with Bacillus identification profile database API Lab⫹ (version 2.1) and with a battery of 87 physiological tests, as described previously (17). The reaction profiles of these tests were compared with a database (16). Toxicity tests. Bacteria were grown on tryptic soy agar plates for 10 days at 28°C to obtain mainly sporulated and lysed cells, verified by phase-contrast microscopy. Colonies were scraped from the agar and suspended in aqua destillata to 100 mg ml⫺1. The suspension was treated by repeated freeze-thaw cycles and filtered (0.2-␮m pore size). The permeate was diluted in dimethyl sulfoxide and tested for toxicity by using the same concentration of dimethyl sulfoxide as the blank. The motility inhibition of boar spermatozoa by the cell extracts was tested as described for the emetic toxin of B. cereus (3). Of each bacterial strain, two to five independently prepared extracts were tested. The sperm motility inhibition by the extracts was given as the concentration required to block motility of 50% of the cells (see Table 1) or by indicating the percentage of motile cells (see Table 3). Three microscopic fields of 102 spermatozoa (magnification, ⫻200) were analyzed for motility with a Hamilton-Thorne sperm analyzer (HTM-S, version 7.2; Hamilton-Thorne Research, Danvers, Mass.) as described in reference 3. The results were confirmed by subjective estimation of motility by phase-contrast microscopy (5 to 10 fields; magnification, ⫻200).

MATERIALS AND METHODS Bacterial cultures. Corynebacterium renale DSM 20688T, B. licheniformis DSM 13T, Bacillus amyloliquefaciens DSM 7T, B. pumilus DSM 27T, and B. cereus DSM 31T were obtained from the German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany, and B. subtilis ATCC 6051T was obtained from the American Type Culture Collection, Manassas, Va. The emetic-toxinproducing strains of B. cereus, 4810/72, NC7401, and F-5881, are described

* Corresponding author. Mailing address: Department of Applied Chemistry and Microbiology, P.O. Box 56 (Biocenter), 00014 University of Helsinki, Finland. Phone: 358-9-70859300. Fax: 358-9-70859301. E-mail: [email protected]. 4637

4638

SALKINOJA-SALONEN ET AL.

The cell membrane-damaging capacity of the bacterial extracts was measured by selective staining with ethidium homodimer and calcein AM, carried out as described in reference 15. Damage to acrosomes was recorded by light microscopy as described elsewhere (30). Mitochondrial damage was documented by transmission electron microscopy of spermatozoan thin sections as described elsewhere (3). ATP loss from the spermatozoan was assayed as described elsewhere (3). C. renale DSM 20688T growth inhibition (4) was read after 2 days at 28°C from two to three replicate Trypticase soy agar plates with wells each holding 150 to 200 ␮l of the cell extract of the isolate to be tested. Ribotyping. Ribotyping was performed with a robotized instrument as described in reference 31. The B. licheniformis strains were prepared and analyzed similarly to the B. cereus strains (23). In short, the DNA was restriction endonuclease cut with EcoRI or PvuII and hybridized to phosphorescently labeled Escherichia coli whole ribosomal operon. Fragment sizes were determined with the GelCompar program (version 4.0; Applied Maths BVBA, Kortrijk, Belgium) from the ribotypes produced by the RiboPrinter (Qualicon, Wilmington, Del.) with DNA molecular size markers (1.1, 2.2, 3.2, 6.5, 9.6, and 48 kb) in every third lane.

RESULTS Detection of toxin-producing B. licheniformis isolates. In total, 210 B. licheniformis isolates involved in food poisoning or suspected food poisoning and 29 strains and isolates originating from veterinary samples, food packaging material, air, and industrial contaminants were studied for toxicity by two tests, the spermatozoan motility test and the C. renale DSM 20688T growth inhibition test. Thirteen strains were found to be positive in one or both toxicity tests. Isolates toxic by both tests (n ⫽ 10) originated from incident-associated food or clinical specimens from Finland and the United Kingdom over a period of ⬎10 years, mainly from cases where B. licheniformis was isolated from food in high numbers (104 to 105 CFU g⫺1). In addition, toxic isolates were obtained from the udder of a cow that had clinically recovered from severe mastitis. Table 1 is a compilation of the properties of the 13 positive and 9 nontoxic strains. Table 1 shows that crude cell extracts (filtered through 0.2␮m-pore-size filters) prepared from 10 B. licheniformis isolates inhibited boar spermatozoan motility when spermatozoa were exposed to the extracts. This protocol has been shown to be a sensitive indicator for the presence (in nanograms per milliliter) of the emetic toxin of B. cereus (3). The 10 toxic isolates included two isolates (553/1 and 553/2) cultured from infant feed formula following an infant fatality (Table 1). Cell extracts prepared from these 10 isolates also inhibited growth of C. renale DSM 20688T. The type strain of B. licheniformis, DSM 13T, was not toxic by either test. The type strains of B. subtilis (ATCC 6051T) and B. pumilus (DSM 27T) were also tested and found to be nontoxic to sperm cells. Two B. licheniformis isolates (575U/5 and 575E/P) of eight tested from unused infant feed packages of the same brand as that connected to the fatal case blocked sperm motility and inhibited C. renale (Table 1; the six nontoxic isolates are not shown). Toxic strains were also isolated from a fecal specimen of a hospital patient with acute-phase food poisoning symptoms (F287/91) and from food poisoning cases connected with curry rice and fast foods (F2943/92, F5520/96, and F231/97). Two of three isolates from milk (each from a separate quarter of the udder) of a postmastitic cow were inhibitory to C. renale DSM 20688T (Table 1; the nontoxic isolate is not shown) and blocked boar spermatozoan motility. Even though all spermtoxic isolates inhibited growth of C. renale, the reverse was not true. However, the sperm cells were exposed to extracts corresponding to 2 to 4 mg (wet weight) of B. licheniformis cells ml⫺1, whereas the amount used in the C. renale test corresponded to 20 to 40 mg (wet weight) of B. licheniformis cells per well in the agar plates. The C. renale test, using a higher amount of the agent, may thus have detected weak toxin pro-

APPL. ENVIRON. MICROBIOL.

ducers which were undetectable in the sperm test. High doses could not be tested with sperm cells because of nonspecific interference by the crude bacterial extracts. The results thus do not exclude the possibility that both effects may have been caused by the same toxic agent. Extracts prepared from the type strain of B. cereus (DSM 31T) strongly inhibited growth of C. renale at amounts equivalent to 20 to 40 mg of B. cereus cells per well. Description of the toxigenic B. licheniformis isolates. All toxic B. licheniformis isolates grew anaerobically; were lecithinase negative, lipolytic (i.e., they hydrolyzed Tween 80; isolates coded F and 123/3 were not tested), and lysozyme sensitive; and hydrolyzed starch and casein (123/3 was not tested). The 10 sperm-toxic strains grew at 55 but not at 10°C in Trypticase soy broth. Twelve of 17 strains inhibitory to C. renale (Table 1) were beta-hemolytic, and 9 of the 10 sperm-toxic isolates were (the exception was F231/97) also beta-hemolytic. Four of the 15 beta-hemolytic isolates were not inhibitory to C. renale and not sperm toxic. The type strain of B. licheniformis (DSM 13T) was nonhemolytic and nontoxic. The growth inhibition of C. renale and/or toxicity to boar spermatozoa was thus not due to the production of beta-hemolysins by the B. licheniformis isolates. All isolates listed in Table 1 could use propionate as the sole carbon source, the characteristic which distinguishes B. licheniformis from B. subtilis and B. pumilus. The sperm-toxic and nontoxic strains of B. licheniformis were compared by using 87 biochemical traits. The 21 B. licheniformis isolates presented in Table 1 were characterized as follows. The results presented for the 21 isolates are identical to those obtained for the type strain DSM 13T. Type strain DSM 13T and all isolates assimilated N-acetyl-D-glucosamine, L-arabinose, p-arbutin, D-cellobiose, D-fructose, D-galactose, gluconate, D-glucose, D-mannose, D-maltose, ␣-D-melibiose, L-rhamnose, D-ribose, sucrose, salicin, D-trehalose, D-xylose, i-inositol, maltitol, D-mannitol, D-sorbitol, acetate, propionate, cis-aconitate, trans-aconitate, 4-aminobutyrate, citrate, fumarate, DL-3-hydroxybutyrate, DLlactate, oxoglutarate, pyruvate, L-alanine, L-aspartate, L-ornithine, and L-proline. Type strain DSM 13T and all isolates did not assimilate adonitol, putrescine, adipate, azelate, glutarate, itaconate, L-malate, mesaconate, suberate, ␤-alanine, L-histidine, L-leucine, L-phenylalanine, L-serine, L-tryptophan, 3-hydroxybenzoate, 4-hydroxybenzoate, and phenylacetate. All isolates (including type strain DSM 13T) did not hydrolyze esculin, para-nitrophenyl-␤-D-galactopyranoside, para-nitrophenyl-␣-D-glucopyranoside, para-nitrophenyl-␤-D-glucopyranoside, bis-para-nitrophenyl-phosphate, and L-glutamyl-␥-3carboxy-para-nitroanilide. Type strain DSM 13T but none of the isolated hydrolyzed para-nitrophenyl-␤-D-glucuronide, para-nitrophenyl-phenyl-phosphonate, para-nitrophenyl-phosphorylcholine, 2-deoxythymidine-5⬘-para-nitrophenyl-phosphate, L-alanine-para-nitroanilide, and L-proline-para-nitroanilide. The results confirmed their identity as B. licheniformis, with a Willcox probability of P ⬎ 0.99 (17). No biochemical difference was detected between the isolates that were toxic and those that were nontoxic to sperm cells or C. renale DSM 20688T: all biochemical reactions were the same as those of the type strain B. licheniformis DSM 13T. Ribotype patterns of toxic and nontoxic B. licheniformis strains. The 21 isolates and the type strain of B. licheniformis in Table 1 were ribotyped by using EcoRI and PvuII, and the multiband patterns obtained are shown in Fig. 1. Ten distinct ribotype patterns were obtained with EcoRI, and 11 were obtained with PvuII. In the ribotype patterns obtained with PvuII (Fig. 1A), the sperm-toxic isolates clustered in four groups, in three (Fig. 1, lanes A, B, and E) of which the fragment patterns were closely similar. Two of these ribotypes (A and B) also

4639 TOXIGENIC BACILLUS LICHENIFORMIS FROM FOOD VOL. 65, 1999

Isolate code

Ice cream Feces of food poisoning patient Curried chicken and mayonnaise sandwich Minced beef pie Tandoori king prawnb Blue cheese, salad dressingb Pancake Curry riceb Profiterolesb Vanilla sauce Vanilla pudding (reconstituted)b Infant feed (formula)b Infant feed (formula)b Infant feed formula, unused package Infant feed formula, unused package Milk from postmastitic cow, 1st quarter Milk from postmastitic cow, 2nd quarter Milk from postmastitic cow, 3rd quarter Unused food packaging paperboard Unused food packaging paperboard Air (garbage dump) Type strain, culture collection

Source of isolate

UK UK UK UK UK UK UK UK UK Norway Finland Finland Finland Finland Finland Finland Finland Finland Finland Finland Finland

Country

12

Acute Acute 5 12

Onset (h) or phase

SC, V, D N, V, D, AP N, SC, D AP, D

Symptom(s)

7 6–8

N, V, B SC, V, D

N, AP, D N, V V

No. ill/no. risk

6/9 1/? 1/1 3/? ⬎2 111/124 (6 hospitalized, 1 fatal) 1/1 (fatal)c 1/1 (fatal)c

2/2 1/1 1/1? 1/1

Description of illness

TABLE 1. Origins and toxicities of B. licheniformis isolatesa

F3648/90 F287/91 F2943/92 F2667/94 F2896/95 F9229/95 F4647/96 F5520/96 F231/97 F5734/93 123/3 553/1 553/2 575U/5 575E/P Hulta 52/97 Hulta 53/97 Hulta 54/97 TSP19 TSP29a 48/87 DSM 13T

CFU g of food⫺1

3 ⫻ 106 1 ⫻ 108 1.1 ⫻ 108 1.1 ⫻ 106 3.1 ⫻ 105

⫺ ⫹⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹⫹ ⫺ ⫺ ⫹⫹ ⫹⫹ ⫹⫹ ⫹ ⫹⫹ ⫺ ⫹⫹ ⫺ ⫺ ⫺ ⫺

Toxicity to boar spermd

⬍2 ⬎20 ⬎20 ⬍2 ⬍2 ⬍2 ⬎10 ⬎10 ⬎10 ⬍2 ⬎20 ⬎10 ⬎10 ⬎10 ⬎10 ⬎10 ⬎10 ⬎10 ⬍2 ⬎10 ⬍2 ⬍2

Inhibition zone (mm) of C. renale

Toxic properties of the B. licheniformis isolates

Hemolysis

Beta Beta Beta Beta Beta Beta Beta

ND Alpha, beta Alpha, beta Beta Beta Beta Beta Beta Beta

a Abbreviations: UK, United Kingdom; SC, stomach cramps; V, vomiting; D, diarrhea; N, nausea; AP, abdominal pain; B, oral burning; ND, not determined. b More than one Bacillus sp. present in the associated food. Isolates 553/1 and 553/2 are from the same incident. Sperm cell motility inhibition is indicated as follows: ⫹⫹, inhibition at ⬍2 mg of B. licheniformis cells; ⫹, inhibition at 4 mg (wet weight) of cells per ml of extended boar semen; ⫺, no inhibition at 4 mg ml⫺1. c d

4640



FIG. 1. Ribotyping of 25 isolates and strains of B. licheniformis of different origins, with PvuII (A) or EcoRI (B) and hybridization with labeled whole ribosomal operon of E. coli. The patterns obtained from B. amyloliquefaciens DSM 7T, B. cereus DSM 31T, and B. subtilis ATCC 6051T are also shown. Strains indicated as being of the same ribotype exhibited patterns with a similarity value of ⬎0.95. (A) Lane A, DSM 13T, 553/2 (toxic), 575E/P (toxic), and TSP29a; lane B, 553/1 (toxic), F287/91 (toxic), F231/97 (toxic), Hulta 53/97, and Hulta 54/97 (toxic); lane E, F2943/92 (toxic); lane H, 575U/5 (toxic), F2667/94, F5520/96 (toxic), Hulta 52/97 (toxic), and TSP19; lanes C, D, F, G, I, J, and K, no toxic isolates; lanes L, M, and N, reference strains B. amyloliquefaciens DSM 7T, B. subtilis ATCC 6051T, and B. cereus DSM 31T, respectively. (B) Lane A, DSM 13T, 553/1 (toxic), 575U/5 (toxic), F281/91 (toxic), F5734/93, TSP29a, and Hulta 52/97 (toxic); lane C, F9229/95, Hulta 53/97, and Hulta 54/97 (toxic); lane D, F2943/92 (toxic), F2896/95, and F231/97 (toxic); lane E, 575E/P (toxic); lane F, 553/2 (toxic); lane I, F5520/96 (toxic); lanes B, G, H, and J, no toxic isolates; lanes K, L, and M, reference strains B. amyloliquefaciens DSM 7T, B. subtilis ATCC 6051T, and B. cereus DSM 31T, respectively.

contained nontoxic isolates. The three related ribotype patterns shared bands at 2.7 ⫾ 0.2, 5.0 ⫾ 0.2, and 11.6 ⫾ 0.1 kb (Table 2). The type strain B. licheniformis DSM 13T (nontoxic) has ribotype A together with another nontoxic strain and two toxic strains (lane A, Fig. 1). The fourth toxic strain ribotype (lane H in Fig. 1) shared the bands at 2.7 and 5.0 kb but had, in addition, four larger fragments (13 to 35 kb). When cut with EcoRI, one ribotype pattern contained the type strain DSM 13T and seven other isolates, of which five were toxic (Fig. 1B, lane A). The other sperm-toxic isolates were scattered among six EcoRI ribotype patterns. The results thus show that among the toxic strains there was considerable diversity in ribotypes, indicating that the toxic isolates were not a clone. However, the PvuII ribotypes containing toxic isolates shared bands of similar sizes (2.5, 5.0, 7 to 8, 11.6, or 13.6 kb [Table 2]). The fragment sizes obtained by the unweighted pair group method with averages algorithm and a commercial computer program for 22 isolates analyzed with an automated ribotyper over a period of 1 year showed standard deviations of approximately 5% for fragments of ⬍10 kb and 5 to 10% for fragments of 10 to 36 kb. Considering this reproducibility, an automated search for fragments of specific sizes may be useful as a preliminary screening method for toxic strains. Effects of toxic B. licheniformis extracts on boar spermatozoa. The responses, shown as four different viability parameters, of boar spermatozoa to cell extracts prepared from nine sperm-toxic isolates of B. licheniformis (Table 1) are compiled in Table 3. None of the extracts was cytolytic toward spermatozoa. All nine isolates inhibited motility of the exposed spermatozoa and damaged cell membrane integrity (Fig. 2), depleted the cellular ATP content (Table 3), and swelled the acrosome (Fig. 3). None of these extracts swelled the mitochondria as observed by transmission electron microscopy (Fig. 4). Extracts prepared from the type strains of B. licheniformis, B. subtilis, B. pumilus, Bacillus mycoides, and B. cereus caused none of the effects exhibited by the strains of B. licheni-

formis (Table 3). Extracts prepared from three emetic-toxinproducing strains of B. cereus (4810/72, NC7401, and F-5881) were tested in the same assay; all inhibited spermatozoan motility at extremely low concentrations (corresponding to ⬍0.002 mg [wet weight] of cells ml⫺1) and swelled the mitochondria in the sperm tail but had no effect on the membrane integrity, cell ATP content, or the acrosome (Table 3). The extracts of each of the three emetic B. cereus strains had no effect on the growth of C. renale at doses corresponding to 20 to 40 mg (wet weight) of cells per well in the agar plate.

TABLE 2. Fragment sizes observed in the B. licheniformis PvuII ribotypes containing toxic isolates Fragment size range (kb)

35.1–36.0 19.1–35.0 16.1–19.0 13.1–16.0 10.1–12.0 9.1–10.0 8.1–9.0 7.1–8.0 6.6–7.0 6.1–6.5 4.1–6.0 2.0–4.0

Size of fragment for PvuII ribotypea: A

11.7 (⫾0.5) 9.9 (⫾0.5) 8.4 (⫾0.3) 7.1 (⫾0.2) 6.8 (⫾0.2) 6.2 (⫾0.2) 5.2 (⫾0.1) 2.6 (⫾0.1)

B

11.8 (⫾0.1) 10.0 (⫾0.2) 8.5 (⫾0.1) 7.2 (⫾0.2) 7.0 (⫾0.1) 6.3 (⫾0.1) 4.8 (⫾0.1) 2.6 (⫾0.1)

E

13.8 11.3 7.7 5.7 2.8

H

35.5 (⫾3.7) 19.3 (⫾0.9) 16.8 (⫾1.0) 13.1 (⫾0.8) 8.3 (⫾0.4) 7.0 (⫾0.3) 4.9 (⫾0.2) 2.8 (⫾0.1)

a Sizes are shown in kilobases ⫾ standard deviations calculated for the isolates with identical ribotypes. For the patterns of the isolates in ribotypes A, B, E, and H (similarity value of 0.95), see the ribotype images in lanes A, B, E, and H, respectively, in Fig. 1. The sizes are means of all strains sharing the ribotype. Ribotype A included two toxic and two nontoxic isolates, ribotype B included four toxic isolates and one nontoxic isolate, ribotype E included one toxic isolate and no nontoxic isolates, and ribotype H included three toxic and two nontoxic isolates.

TOXIGENIC BACILLUS LICHENIFORMIS FROM FOOD

VOL. 65, 1999

4641

TABLE 3. Effects of cell extracts from food- and food poisoning-related isolates of B. licheniformis and from reference strains on boar spermatozoaa Observed response Extract prepared from each of the strains or isolates

Damage to energy metabolism of sperm cells

% of sperm cells in which morphological damage observed

% of sperm cells motilec

ATP (␮g ml⫺1) of extended sperm

Membrane integrityd

Acrosome swellingb

Mitochondrial swellinge

⬍1

⬍0.1

⬎70

⬎50

⬍1

⬍1

ND f

⬎70

ND

⬍1

⬍1

⬎3

⬍20

⬍10

⬎50

Type strains of related species B. licheniformis DSM 13T B. subtilis ATCC 6051T B. pumilus DSM 27T B. mycoides ATCC 6462T B. cereus DSM 31T

⬎60 ⬎60 ⬎60 ⬎60 ⬎60

⬎3 ⬎3 ⬎3 ⬎3 ⬎3

⬍20 ⬍20 ⬍20 ⬍20 ⬍20

⬍10 ND ND ⬍10 ⬍10

⬍1 ⬍1 ⬍1 ⬍1 ⬍1

Control exposures Reagents only Freeze-thaw

⬎60 ⬍1

⬎3 ⬍0.1

⬍20 ⬎70

⬍10 ND

⬍1 ND

B. licheniformis isolates from food F287/91, F2943/92, F5220/96, F231/97, 553/1, 553/2, and 575U/5 Raw milk isolates (postmastitic cow) Hulta 52/97 and Hulta 54/97 Emetic-toxin-producing strains of B. cereus 4810/72, NC7401, and F-5881

a Extended boar semen (30 ⫻ 106 to 60 ⫻ 106 sperm cells ml⫺1) was exposed for 1 to 3 days to 10 ␮l of cell extracts (0.2-␮m-pore-size filter filtrate) ml⫺1 prepared from plate-grown cultures. Two to three extracts of each strain or isolate were tested with sperm from three individual boars. The average difference between the results was ⬍20% in all cases. b Degenerated acrosomes were detected after Giemsa staining by light microscopy (Fig. 3). c Percentage of sperm cells expressing high motility as shown by the Hamilton-Thorne sperm motility analyzer (3). d As observed by vitality staining and fluorescence microscopy (15) (Fig. 2). e Swollen and disrupted mitochondria as observed by transmission electron microscopy (Fig. 4D). f ND, not determined.

These results show that the sperm-toxic agent(s) produced by the toxic strains of B. licheniformis was uniform in action. The exerted toxic effect differed in biological activity from that of the cereulide produced by the emetic strains of B. cereus. It is interesting that the extracts prepared from the two spermtoxic B. licheniformis isolates (Hulta 52/97 and Hulta 54/97) originating from raw milk from a postmastitic cow exhibited toxic responses in spermatozoan cells quantitatively and qualitatively similar to those seen after exposure to the toxic food poisoning isolates, including those from an unused package of branded infant feed (Tables 1 and 3). The effects on the spermatozoan plasma membrane and on the acrosomal response to B. licheniformis extracts were dose dependent. The toxic threshold of these effects was equivalent to 2 to 4 mg (wet weight) of bacterial cells per ml of extended boar semen for all toxic B. licheniformis strains (Fig. 2 and Table 1). Toxic activities of extracts prepared from the B. licheniformis isolates 553/1, 553/2, 575U/5, Hulta 52/97, and Hulta 54/97 were insensitive to heat (100°C for 20 min), inactivation by pronase (200 ␮g ml⫺1, 3 h), acid (pH 2 with HCl for 30 min), and alkali (pH 12 with NaOH for 30 min). The observed heat stability suggests that the toxin(s) was not an enzyme. The toxic agent was more soluble in 50 or 100% methanol than in water. As a solution in 50% (vol/vol) methanol, the toxic agent was filterable through microconcentrator membranes with a nominal cutoff of 10,000 g mol⫺1 but not as a water extract, indicating a tendency to hydrophobic interactions. The 10,000-g mol⫺1 filtrates of the toxic extracts exhibited unaltered toxic effects on the spermatozoa. These results show that the B. licheniformis sperm-toxic agent was nonproteinaceous, heat stable, and nonpolar and of an apparent mass smaller than 10,000 g mol⫺1. The agent inhibiting the growth of C. renale also

survived the treatments listed above, indicating that it was the same or a similar type of compound as the agent blocking spermatozoan motility. DISCUSSION This paper is to our knowledge the first demonstration of toxins produced by B. licheniformis isolates associated with human disease. Toxins produced by B. licheniformis were detected by the boar spermatozoan motility inhibition assay, which has been reported as a sensitive and specific test for detecting emetic-toxin-producing B. cereus strains (3). The toxins of B. licheniformis inhibited sperm motility (Table 1) by interfering with the cellular energy metabolism in a manner different from that shown with the emetic toxin of B. cereus, a toxin that causes swelling of mitochondria (3, 21). The toxic threshold of the B. licheniformis extracts was ⱖ100 times higher than those observed for emetic-toxin-producing B. cereus strains (3). The occurrence of B. cereus in sensitive foods is regulated in many countries (in Finland, the limit is ⱕ103 CFU g⫺1). There is no restriction on B. licheniformis, which often occurs in high numbers in foods (we observed 104 to 108 CFU g⫺1 [Table 1]). The toxigenicity observed in the present work may thus be of food poisoning significance. The toxic B. licheniformis extracts induced the acrosome reaction (Table 3 and Fig. 3), a very likely novel trait among bacterial toxins, and possibly indicating an impact on the cellular signalling system. Despite differences in biological activity, the sperm-toxic agents from the isolates of B. licheniformis studied (Table 3) were similar in many physicochemical properties to cereulide, the emetic toxin of B. cereus (1, 3). Cereulide is a dodecadep-

4642



FIG. 2. Fluorescence micrographs of boar spermatozoa stained for determination of viability after being exposed to cell extracts of different B. licheniformis strains. (A) Sperm cells (5 ⫻ 106) in 1 ml of extended (with commercial extender) boar semen were exposed to cell extract from 4 mg (wet weight) of B. licheniformis DSM T 13 cells. Over 80% of the spermatozoa showed intact cell membranes. Similar results were obtained with spermatozoa exposed to the negative control (staining green). (B) Effect of extract prepared from 2 mg (wet weight) of cells of B. licheniformis 553/2. Fifty percent of the sperm cells lost integrity in the cell membrane (staining orange). (C) Sperm cells were exposed to extract prepared from 4 mg (wet weight) of the same isolate as in panel B. Seventy percent of sperm cells lost cell membrane integrity (red). Magnification in all panels, ⫻2,000 (i.e., sperm head dimensions are 2 to 3 by 3 to 5 ␮m).

VOL. 65, 1999


4643

FIG. 3. Light micrographs of Giemsa stained boar spermatozoa exposed to cell extracts of B. cereus and toxic and nontoxic strains of B. licheniformis. (A and B) Extended boar semen (1 ml) exposed to extract from 4 mg (wet weight) of cells of the emetic B. cereus strain 4810/72 (A) or the (nontoxic) B. licheniformis type strain DSM 13T (B). Over 90% of the exposed sperm cells showed heavily staining dark intact acrosomes, similar to spermatozoa exposed to negative control extract (data not shown). (C) Extended boar semen (2 ml) exposed to B. licheniformis 553/2 (extract from 4 mg [wet weight] of cells). Over 50% of the cells showed lightly staining fused acrosomes or swollen and disrupted acrosomes. Magnification in all panels, ⫻2,000 (i.e., sperm head dimensions are 2 to 3 by 3 to 5 ␮m).

4644



FIG. 4. Thin cross sections of the middle segments of boar spermatozoa exposed for 4 days to cell extracts of B. licheniformis 553/2 and F287/91, the B. licheniformis type strain DSM 13T, and an emetic toxin producer strain, B. cereus 4810/72. (A and B) Sperm cells exposed to cell extracts prepared from 4 mg (wet weight) of isolates 553/2 and F287/91. These sperm cells had lost motility and ATP and showed damaged cell plasma membrane, but the mitochondria were intact. (C) Spermatozoon exposed to extract of the type strain B. licheniformis DSM 13T (nontoxic). These cells displayed normal motility and cellular ATP content after exposure, and the figure shows an intact plasma membrane. (D) Sperm cell exposed to cell extract from 2 mg (wet weight) of the emetic strain B. cereus 4810/72 (12) ml⫺1. These cells have a morphologically intact plasma membrane and normal ATP content, but over 90% of the cells exposed showed no motility, and their mitochondria were swollen with a disrupted outer membrane. Bars, 200 nm.

sipeptide structurally resembling valinomycin (1). Peptide toxins are nonribosomally produced by a wide range of microorganisms (18). Also, some strains of B. licheniformis are known to produce peptide antibiotics (13), such as bacitracin (11, 18) and amoebicins (8, 9, 19), some of which have been used as antimicrobial agents, but none have so far been shown to be associated with food poisoning. The B. licheniformis toxins

reported in this paper also possessed antimicrobial activity, demonstrated by the induction of large inhibition zones by cell extracts of many of the B. licheniformis isolates when introduced onto plate cultures of the actinomycete C. renale DSM 20688T (Table 1). C. renale has been shown elsewhere to be susceptible toward human pathogenic Staphylococcus aureus strains producing staphylococcin BacR1 (4) and to be associ-


VOL. 65, 1999

ated with the production of the exfoliative toxin B (22). The anticorynebacterial activity of B. licheniformis isolates was caused by an agent(s) physicochemically similar to that toxic to spermatozoa (both are resistant to heat, acid, and alkali and soluble in methanol). Sperm toxin-producing B. licheniformis strains were not readily differentiated from nontoxigenic strains by only biochemical and physiological criteria (Table 1; see also above). Ribotyping revealed great genetic diversity; the toxigenic strains thus formed no clone. However, the toxic ribotypes were related (shared bands [Table 2]); only 4 of the 11 PvuII ribotypes contained toxigenic strains. Automated ribotyping may be useful as a preliminary screening test for putative toxin producers. It is interesting that the ribotype of a toxic isolate obtained from the baby food associated with a fatal food poisoning (553/1) was identical to that of the isolate obtained from an unopened package of the same brand (575U/5) and to that of a toxic isolate cultured from the raw milk of a cow that had apparently recovered from mastitis (Hulta 52/97). As B. licheniformis is a sporeformer and likely to survive all industrial processing of milk, such as the manufacture of milk powder and whey concentrate, such a finding may indicate a possible route of infection. These results also indicate genetic diversity among the toxic B. licheniformis isolates: two different toxigenic ribotypes were isolated from two different quarters of the same udder of a cow, two different toxigenic ribotypes were isolated from the same batch of baby food, and two toxigenic strains with different ribotypes were recovered from an unused package of commercial baby food. It is possible that genotypically distinguishable different toxigenic isolates might have been detected also in the other food poisoning cases listed in Table 1, had more than one isolate been available for study. Recombinant strains of B. licheniformis are used to produce industrial enzymes on a large scale, e.g., carbohydrase and protease used in food processing (5, 6). The species is considered safe and has generally recognized-as-safe status with the U.S. Food and Drug Administration (6). In the light of our findings, the generally recognized-as-safe status of the species B. licheniformis may require reassessment. ACKNOWLEDGMENTS This work was financially supported by the Academy of Finland (M.S.S.) and the Centre of Excellence Fund of the University of Helsinki (M.S.S.). We thank the Institute of Biotechnology of the University of Helsinki for the use of the electron microscope. We are grateful to Jyrki Juhanoja for expert technical support in electron microscopy, to Paula Hyvo ¨nen (EELA, Kuopio) for donating strain 123/3, to Irina Tsitko for help with the GelCompar program, and to Irmgard Suominen and Camelia Apetroaie for their contributions in sperm toxicity testing. REFERENCES 1. Agata, N., M. Ohta, M. Masashi, and M. Isobe. 1995. A novel dodecadepsipeptide, cereulide, is an emetic toxin of Bacillus cereus. FEMS Microbiol. Lett. 129:17–20. 2. Andersson, M., M. Laukkanen, E.-L. Nurmiaho-Lassila, F. A. Rainey, S. I. Niemela ¨, and M. S. Salkinoja-Salonen. 1995. Bacillus thermosphaericus, sp. nov., a new thermophilic ureolytic Bacillus isolated from air. Syst. Appl. Microbiol. 18:203–220. 3. Andersson, M. A., R. Mikkola, J. Helin, M. C. Andersson, and M. S. Salkinoja-Salonen. 1998. A novel sensitive bioassay for detection of Bacillus cereus emetic toxin and related depsipeptide ionophores. Appl. Environ. Microbiol. 64:1338–1343. 4. Crupper, S. S., A. J. Gies, and J. J. Iandolo. 1997. Purification and characterization of staphylococcin BacR1, a broad-spectrum bacteriocin. Appl. Environ. Microbiol. 63:4185–4190. 5. de Boer, S. A., F. G. Priest, and B. Diedrichsen. 1994. On the industrial use

4645

of Bacillus licheniformis: a review. Appl. Microbiol. Biotechnol. 40:595–598. 6. Food and Drug Administration. 1983. 21 CFR 184:1027. Fed. Regist. 48: 329–340. 7. Frank, J. F. 1997. Milk and dairy products, p. 100–116. In M. P. Doyle, L. R. Beuchat, and T. J. Montville (ed.), Food microbiology: fundamentals and frontiers. American Society for Microbiology, Washington, D.C. 8. Galvez, A., E. Valdivia, A. Gonzalez-Segura, M. Lebbadi, M. MartinezBueno, and M. Maqueda. 1993. Purification, characterization and lytic activity against Naegleria fowleri of two amoebocins produced by Bacillus licheniformis A12. Appl. Environ. Microbiol. 59:1480–1486. 9. Galvez, A., M. Maqueda, P. Cordovilla, M. Martinez-Bueno, M. Lebbadi, and E. Valdivia. 1994. Characterization and biological activity against Naegleria fowleri of amoebocins produced by Bacillus licheniformis D-13. Antimicrob. Agents Chemother. 38:1314–1319. 10. Gordon, R. E., W. C. Haynes, and C. Hor-Nay Pang. 1973. The genus Bacillus, p. 3–14. In Agriculture handbook no. 427. United States Department of Agriculture, Washington, D.C. 11. Gra ¨fe, U. 1992. Biochemie der Antibiotika: Struktur-Biosynthese-Wirkmechanismus, p. 164–170. Spektrum Akademischer Verlag GmbH, Heidelberg, Germany. 12. Granum, P. E. 1997. Bacillus cereus, p. 327–336. In M. P. Doyle, L. R. Beuchat, and T. J. Montville (ed.), Food microbiology: fundamentals and frontiers. American Society for Microbiology, Washington, D.C. 13. Horowitz, S., and W. M. Griffin. 1991. Structural analysis of Bacillus licheniformis 86 surfactant. J. Ind. Microbiol. 7:45–52. 14. Johnson, C. T., G. R. Lupson, and K. E. Lawrence. 1994. The bovine placentome in bacterial and mycotic abortions. Vet. Rec. 134:263–266. 15. Juonala, T., S. Lintukangas, T. Nurttila, and M. Andersson. 1998. Relationship between semen quality and fertility in 106 A1-boars. Reprod. Domest. Anim. 33:155–158. 16. Ka ¨mpfer, P. 1991. Application of miniaturized physiological and biochemical tests in numerical classification and identification of some bacilli. J. Gen. Appl. Microbiol. 37:225–247. 17. Ka ¨mpfer, P., E. B. M. Denner, S. Meyer, E. R. B. Moore, and H.-J. Busse. 1997. Classification of “Pseudomonas azotocolligans” Anderson 1955, 132, in the genus Sphingomonas as Sphingomonas trueperi sp. nov. Int. J. Syst. Bacteriol. 47:577–583. 18. Kleinkauf, H., and H. von Do ¨hren. 1996. A nonribosomal system of peptide synthesis. Eur. J. Biochem. 236:335–351. 19. Lebbadi, M., A. Galvez, E. Valdivia, M. Martinez-Bueno, and M. Maqueda. 1994. Purification of amoebicin m4-A from Bacillus licheniformis M-4. Arch. Microbiol. 162:98–102. 20. Lund, B. M. 1990. Foodborne diseases due to Bacillus and Clostridium species. Lancet 336:982–986. 21. Mahler, H., A. Pasi, J. M. Kramer, P. Schulte, A. C. Scoging, W. Baer, and S. Kraehenbuehl. 1997. Fulminant liver failure in association with the emetic toxin of Bacillus cereus. N. Engl. J. Med. 336:1143–1148. 22. Melish, M. E., and L. A. Glasgow. 1971. Staphylococcal scalded skin syndrome: the expanded clinical syndrome. J. Pediatr. 78:958–967. 23. Pirttija ¨rvi, T. S., M. A. Andersson, A. Scoging, and M. S. Salkinoja-Salonen. 1999. Evaluation of methods for recognizing strains of Bacillus cereus with food poisoning potential among industrial contaminants. Syst. Appl. Microbiol. 22:133–144. 24. Rosenkvist, H., and Å. Hansen. 1995. Contamination profiles and characterization of Bacillus species in wheat bread and raw materials for bread production. Int. J. Food Microbiol. 26:353–363. 25. Smibert, R. M., and N. R. Krieg. 1994. Phenotypic characterization, p. 607–654. In P. Gerhardt, R. G. E. Murray, W. A. Wood, and N. R. Krieg (ed.), Methods for general and molecular bacteriology. American Society for Microbiology, Washington, D.C. 26. Tatzel, R., W. Ludwig, K.-H. Schleifer, and P. R. Wallno ¨fer. 1994. Identification of Bacillus strains isolated from milk and cream with classical and nucleic acid hybridization methods. J. Dairy Res. 61:529–535. 27. Turnbull, P. 1997. The role of Bacillus genus in infection. Culture 18:5–8. 28. Turnbull, P. C. B., and J. M. Kramer. 1995. Bacillus, p. 349–356. In P. R. Murray, E. J. Baron, M. A. Pfaller, F. C. Tenover, and R. H. Yolken (ed.), Manual of clinical microbiology, 6th ed. American Society for Microbiology, Washington, D.C. 29. Vuorio, R., M. A. Andersson, T. Johansson, T. Honkanen-Buzalski, and M. S. Salkinoja-Salonen. 1998. Toxin producing Bacillus licheniformis strains in infant food. Poster presented at International Conference on Foodborne Pathogens, Detection and Typing, The Hague, The Netherlands, 20 to 21 April 1998. 30. Watson, P. F. 1975. The use of Giemsa stain to detect changes in acrosomes of frozen ram spermatozoa. Vet. Rec. 97:12–13. 31. Zaitsev, G. M., I. V. Tsitko, F. A. Rainey, N. I. Govorukhina, Y. A. Trotsenko, J. S. Uotila, E. Stackebrandt, and M. S. Salkinoja-Salonen. 1998. New aerobic ammonium-dependent and obligately oxalotrophic bacteria: description of Ammoniphilus oxalaticus, gen. nov., sp. nov., and Ammoniphilus oxalivorans gen. nov., sp. nov. Int. J. Syst. Bacteriol. 48:151–161.