Clostridium botulinum and Clostridium perfringens
Jim McLauchlin and Kathie A. Grant
1. INTRODUCTION
Clostridium is a diverse genus of Gram-positive, endospore-bearing obligate anaerobes that are widespread in the environment. This genus includes more than 100 species, and the overall range in the G+C content (22–55 mol%) reflects the enormous phylogenetic variation encompassed within this group. The principal foodborne pathogens are Clostridium botulinum and Clostridium perfringens that cause toxin-mediated disease either by preformed toxin (foodborne botulism) or by the formation of toxin in the enteric tract (infant botulism and C. perfringens diarrhea). These two bacteria and their foodborne diseases will be discussed here.
2. CLOSTRIDIUM BOTULINUM 2.1. Introduction
Botulism is a rare but potentially fatal disease caused by neurotoxins (BoNTs) usually produced by C. botulinum. Disease symptoms result from paralysis owing to the inhibition of neurotransmitter release by BoNTs. Human foodborne disease results from either ingestion of preformed BoNT in foods contaminated by C.
botulinum, or production of BoNT following intestinal colonization by C. botu- linum (usually in infants). The disease can also be transmitted by nonfoodborne routes including the production of BoNT from C. botulinum-infected wounds as well as accidental and deliberate release of BoNT (1). Although botulism has been reported as causing a tooth abscess, and hence was probably foodborne (2), wound botulism is most often associated with the trauma at other sites (especially at injec- tion sites to illegal drug users) and are outside the scope of this work and will not be discussed here.
Foodborne botulism represents a potential national (or international) public health emergency. Prompt diagnosis and early treatment are essential to reduce the considerable morbidity and mortality of this disease. The laboratory investigation of botulism is highly specialized, but is essential for the rapid identification of reservoirs of infection and provides vital information allowing the most appropriate interventions to be implemented. Key reviews on C. botulinum can be found in the literature(1,3–17).
From: Infectious Disease: Foodborne Diseases Edited by: S. Simjee © Humana Press Inc., Totowa, NJ
41
2.2. Classification, Identification, Isolation and Diagnosis 2.2.1. Taxonomy of C. botulinum
The species C. botulinum is defined on the basis of a single phenotypic characteristic, i.e., neurotoxin production, and is classified into four distinct taxonomic lineages (des- ignated groups I–IV; [18]). Characteristics of these four groups are shown in Table 1.
This grouping combines highly diverse phylogenetic organisms that consequently show considerable differences in genotype and phenotype. Indeed the genetic (phylogenetic) differences within C. botulinum are greater than that between Bacillus subtilis and Staphylococcus aureus (18). This situation is further, and somewhat inconsistently, complicated because there are genetically very closely related Clostridium species that do not possess neurotoxins and are, therefore, not named C. botulinum (i.e., Clostridium sporogenes, Clostridium novyi, and un-named organisms) and neurotoxin containing Clostridium baratii and Clostridium butyricum (18). Further complications exist becaue there is evidence for instability and horizontal gene transfer of BoNT genes as well as nontoxigenic C. botulinum variants containing cryptic or fragmentary copies of BoNTs.
It has been proposed (18) that the four taxonomic lineages should be reclassified into separate species, and a precedence has been set by a proposal to rename C. botulinum Group IV (as well as some Clostridium subterminale and Clostridium hastiforme) as Clostridium argentinense (148). However, this proposed revision for the classification of C. botulinum has not found common use and the taxonomy remains unresolved.
The genome size of both C. botulinum groups I and II has been estimated at between 3.6 and 4.1 Mbp by pulsed-field gel electrophoresis analysis (19,149). A complete genome sequence for a single C. botulinum Type A (Hall strain, ATCC 3502) is available, and at the time of writing the annotation is not yet complete: the genome size is 3,886,916 bp in size, with a G+C content of 28.2 mol %, and one 16,344 bp plasmid was detected (20).
2.2.2. Neurotoxin Types
BoNTs are proteins that can be divided into seven antigenically distinct types, designated A to G, and are among the most potent toxins known (1). The distribution of BoNTs among the four taxonomic groups is shown in Table 1. Group I includes the proteolytic strains producing BoNTs A, B, and F, either singly or as dual toxin types AB, AF, and BF. Group II includes the nonproteolytic strains producing toxins B, F, and E. Group III organisms produce either type C or D toxins and are generally nonproteolytic. Group IV strains produce type G toxin and differ from other groups in not producing lipase. Important C. botulinum for human foodborne diseases are in groups I (BoNTs A, B, and F) and II (BoNTs B, E, and F), with type F being the least common (1). However human infection resulting from type C has been reported (21).
Organisms from groups I–III cause diseases among animals especially aquatic birds, horses, and cattle. Group IV organisms producing BoNTG have been isolated from autopsy specimens, although evidence that botulism was the cause of death in these cases was not demonstrated (22). BoNTG has not been associated with any other naturally occurring cases of botulism in humans or any other animals (9). BoNT production, pre- sumably as a result of horizontal gene transfer, has also been detected in C. baratii (type F) and C. butyricum (type E), both of which have been associated with human diseases (23–26).
2.2.3. Detection of BoNTs By Mouse Bioassay
Detection of BoNT production is the cardinal feature for both the classification and identification of C. botulinum as well as providing confirmation of a clinical diagnosis of botulism. Traditionally this has relied on the use of an in vivo mouse bioassay to detect BoNT activity (8,11).
Toxin extracts for bioassays are prepared as described here and injected intra- peritoneally into mice. The biological action of the toxin is generally observed within 24 h and initially often presents with hyperactivity followed by reduced mobility, ruffled fur, labored breathing, contraction of the abdominal muscles (wasp waist), and total paralysis. Respiratory failure and death follow unless animals are humanely euthanized.
Animals are observed initially after injection and then at 2, 4, 8, 12, and 24 h, and then daily for up to 4 d. Those showing signs within 2–6 h usually die within 24 h. The rapidity of symptoms (and time of death) is proportional to the amount of toxin present in the sample. Confirmation of the toxin (together with the determination of the toxin type) is achieved by an absence of the above symptoms in the mice following injection of the inoculum after preincubation with specific antitoxin (8,11).
Extracts from food or feces are generally prepared by homogenization in gelatin- phosphate buffer (pH 6.2) which is centrifuged and filtered (to prevent infection) prior to injection. When low levels of toxin or nonproteolytic strains are suspected, the preparation can be pretreated with trypsin to activate the neurotoxin.
Table 1
Characteristics of C. botulinum Groups
Group
I II III IV
Neurotoxin types A, B, F B, E, F C, D G
Human disease Yes Yes Noa No
Growth temperatures
Minimum 10 3.3 15 ND
Optimum 35–40 18–25 40 37
Minimum pH for growth 4.6 5.0 5.0 ND
Minimum awfor growth 0.94 0.97 ND ND
Inhibition by NaCl 10% 5% 2% ND
D100°Cof spores (min) 25 <0.1 0.1–0.9 0.8–1.12
D121°Cof spores (min) 0.1–0.2 <0.001 ND ND
Proteolytic activity Yes No No Yes (weak)
Lipolytic activity Yes Yes Yes No
Saccharolytic activity No Yes Yes (weak) No
Related non-neurotoxigenic C. sporogenes C. beijerinkii C. novyi C. subterminale Clostridium species C. putrificum C. haemolyticum C. histolyticum C. linosium Location of neurotoxin Chromosomal Chromosomal Phage Plasmid
genes
Overall G+C content 26–29 27–29 26–28 28–30
(mol%)
aOne case of infant botulism due to type C has been reported (21).
ND, not determined.
Assays for toxin are applied to direct analysis of clinical material (serum), extracts from clinical material (feces, gastric contents, and vomitus) and food, as well as to the supernatants from enrichment and pure cultures growing in broths. In vitro growth of C. botulinum in enrichment cultures is achieved by inoculation of broths with feces, gastric contents, and vomitus as well as food samples. For nonfoodborne infections (i.e., material from infected wounds), additional toxin tests are performed to the extracts from pus or tissue, together with application of cultural methods in these samples.
It should be noted that the detection of toxin by bioassays is highly specialized and is likely to require specific authority (including ethical considerations) for their performance.
2.2.4. Identification and Isolation of C. botulinum
As stated earlier, C. botulinum is defined on the basis of a single phenotypic characteristic, i.e., neurotoxin production, and hence more conventional phenotypic reactions such as the fermentation of carbohydrates, cellular fatty-acid analysis, and 16S rDNA gene sequence in not sufficient to provide an unequivocal identification (18,27,28). The identification process, therefore, relies on the detection of neurotoxins (or their genes).
The initial process for culture of C. botulinum is the enrichment in prereduced cooked meat medium with or without glucose, chopped-meat glucose-starch medium, or tryptone–peptone glucose yeast extract broth. Trypsin may be added to inactivate bacteriocins and to activate neurotoxin. Replicate inoculated broths are treated either with or without a heat shock (60–80°C for 10–20 min). Lysozyme is also added to assist in the reversal of heat-injured spores, especially for nonproteolytic C. botulinum strains.
Because of the different physiologies of the C. botulinum groups (Table 1), the opti- mal incubation temperature for isolation is problematic. Some authors recommend 35°C, and 28°C for which the Group II organisms are suspected (i.e., when examining fish or shellfish; [11]). Others recommend 30°C as a compromise to cover all groups, with the proviso that Group I organisms will not be growing optimally (8). Broths should be tested for toxin production using bioassay after 5 d, and an additional 10 d may be necessary to detect the growth from the delayed germination of injured spores.
Broths should also be examined microscopically: typical C. botulinum in Gram-stained smears have a ‘drum stick’ or ‘tennis racket’ appearance with a subterminal spore markedly swelling the vegetative cell.
Broths suspected to contain C. botulinum together with foods and clinical samples (especially feces) should be subcultured onto prereduced solid media and incubated in an anaerobic environment. Specific media for C. botulinum isolation have been described, which incorporates egg yolk to detect lipase-positive C. botulinum colonies (29) as well as the selective antimicrobial agents: cycloserine, sulfamethoxazole, and trimethoprim (29–31). Because of the considerable diversity within this species, especially between C. botulinum groups I and II, plates with and without added antimicrobial agents should both be inoculated and incubated in parallel. Care should be taken to analyze all suspicious colonies since lipase negative C. botulinum occur, although these are rare (1). Typical colonial growth showing a C. botulinum, together with an atypical lipase negative culture are shown in Fig 1.
Classically, C. botulinum colonies are usually gray-white on blood containing agars with circular irregular edges, and a clear zone of G-hemolysis. However a large amount of phenotypic variation in colonial appearance should be expected. Purified cultures suspected to be C. botulinum should be reinoculated into the broths, and following incu- bation retested in the bioassay to ensure production of BoNT. Additional methods based on 16S rDNA sequencing and detection of BoNT gene fragments by PCR are increasingly used for identification (see Section 2.2.5.). However, because 16S rDNA does not equivocally differentiate C. botulinum from its closest relative (e.g., C. sporogenes and C. novyi), tests for neurotoxin or the presence of neurotoxin genes are still necessary.
2.2.5. Diagnosis of Botulism in Humans
The diagnosis of botulism is based primarily on clinical symptoms and for both food- borne and infant botulism illnesses presents with variable severity, presumably because of the amount of toxin presented to the patient. Common symptoms are diplopia, blurred vision, and bulbar weakness and rapidly progressing symmetric paralysis: a more detailed description of the presentation of this disease is given in Section 2.6.
Criteria for a confirmed laboratory diagnosis for foodborne, infant, and wound botulism have been produced (32) and these include the detection of botulinum toxin in serum, stool, or patient’s food and/or the isolation of C. botulinum from stool. Cases are further classified as: confirmed, those with a clinically compatible presentation and occurring among persons who ate the same food as persons who have laboratory-confirmed cases of botulism; and probable, those clinically compatible and with an epidemiologic link (e.g., ingestion of a home-canned food within the previous 48 h) but without laboratory confirmation (32). Laboratory confirmation of a clinical diagnosis may not always be possible by the detection of BoNT in patients’ body fluids because, an appropriate sample is not always available, the toxin concentration is below the detection limit of the bioassay, or toxin degradation has occurred during transit prior to testing in the laboratory. As stated previously, in 30–40% of patients, the concentration of BoNT Fig. 1. Typical colonial growth of C. botulinum, together with an atypical lipase negative culture.
(Please see color insert.)
is below the detection level for the bioassay (9). Arnon (4) summarized results from 21 studies and showed that C. botulinum (but not C. botulinum neurotoxin) could be recov- ered from the feces of 1.9% of >1400 individuals feces which were not suspected to have botulism. Thus, the isolation of C. botulinum from feces in the absence of this disease symptoms is very rare, and the above case definition with respect to the isolation of the organism is appropriate in almost all circumstances. Difficulties in the laboratory diagno- sis of foodborne botulism are illustrated by large outbreaks where despite appropriate testing regimes, none or a very low percentage of laboratory tests provide evidence for the disease (33,34).
All clinical specimens for BoNT detection must be collected prior to the administration of botulinum antitoxin. Suspect foods should also be examined by direct detection of BoNT together with isolation and identification of C. botulinum vegetative cells and spores (8,11). In infant botulism, implicated sources are identified by the detection of C. botulinum spores in food or in the environment (8,11).
2.2.6. Alternatives to In Vivo Methods
PCR-based assays have been reported for BoNT gene fragments using conventional block-based PCR assays with gel electrophoresis for detection of amplified products (35–47). These assays either detect a single toxin type or suffer from a lack of specificity owing to low melting temperatures of the primers, some requiring subsequent use of hybridization probes to confirm the specific toxin type, hence having shortcomings for the routine identification of cultures or for the detection of the bacterium in enrichment broths. A conventional multiplex PCR assay has been described for BoNT A, B, E and F, and although primers with higher annealing temperatures are used, specific toxin gene detection is still achieved by differential migration of amplified products during agarose gel electrophoresis (45). Advances in nucleic-acid detection technologies have seen a plethora of real-time PCR-based methods for the detection of a wide range of microbial pathogens. Fluorescence-based PCR using hybridization probes allows online monitoring of amplified gene fragments at each cycle of PCR, thus, permitting simulta- neous amplification and detection at high sensitivities of pathogen specific nucleic acids within 1–2 h. A real-time PCR assay has been described for a BoNTE (48) and for BoNT A, B and E (49). The later method showed that rapid detection of neurotoxin gene fragments could be performed directly on foods, clinical samples, enrichment broths, and bacterial cultures (49). It was also demonstrated that the use of PCR has consider- able advantages for the investigation of both foodborne and infant botulism in terms of speed of analysis, nonsubjectivity of testing regimes, reduction in cost and reduction (although not total elimination) in the numbers of animal tests (49).
Considerable efforts have been made in the development and evaluation of immunoassays for the detection of BoNTs, because of the recent increase in interest in bioterrorism agents (50–54). However, despite these efforts, immunoassays have not been found to be suitable alternatives to the mouse for public health applications. This is probably because of both the sensitivity and specificity of the mouse bioassay, and the need to detect relatively small amounts of toxin complex in various matrices. To increase both the sensitivity and specificity of tests, Wictome et al. (55,56) described assays based on both the immunoassays as well as the detection of the specific zinc endopeptidase activities of the BoNTs (see Section 2.5.). Recent data has shown that
the BoNT toxins A–E and F can be unequivocally identified by matrix assisted laser desorption ionization and electrospray mass spectrometry (57,58). The approaches of improved immunoassays, functional in vitro assays, and mass spectrometric approaches have yet to be proven for the detection of these toxins in complex matrices such as serum, feces, and food. However, applications of these technologies to detect other targets suggest that this will be possible in the near future and is likely to be of a level of a similar level of sensitivity to that of the mouse bioassay.
2.2.7. Epidemiological Typing
Despite the realization that C. botulinum is widespread in the environment, the use of epidemiological typing systems to track strains of the bacterium through the food chain is relatively underdeveloped, and an association between food sources is assumed on the results of toxin type alone. This is probably adequate where there is a strong epidemiological evidence linking a specific product in a case of food botulism (together with large numbers of the bacterium in incriminated foods); however, typing systems may be invaluable where low levels of the organism are present (as may be the case in infant botulism). From our experience, with a single infant botulism case in the United Kingdom, heterogeneous populations of C. botulinum can be present in both the infant’s feces and in single packets of implicated food (dried infant-formula feed; [59]).
In addition to toxin typing of C. botulinum, DNA-based typing methods have been applied to this bacterium and these include pulsed field gel electrophoresis (19,60,61), randomly amplified polymorphic DNA (62), ribotyping (63,64), and amplified fragment length polymorphism analysis (59). These studies indicate a high level of heterogeneity of the bacterium in the environment.
2.3. Reservoirs
C. botulinum is widely distributed in soils and in the sediments of oceans and lakes.
However, their types, and the numbers present vary widely. In the Western USA, parts of South America, and China, type A predominates. However, in the Eastern USA, proteolytic type B is most common; and in Europe, nonproteolytic type B. Type E is particularly associated with temperate aquatic environments worldwide (hence, botulism resulting from this type is associated with fish and other seafood), and types C and D in warmer environments. The numbers of the organism present in natural environments also varies widely, some soils being frequently contaminated with >1000 organisms per kg. Examples of the incidence of C. botulinum in different soils and sediments are shown in Table 2.
Because of the widespread distribution of this bacterium, foods can contain viable C. botulinum spores, but will not cause foodborne botulism unless the organism is able to grow and produce toxin. However, for infant botulism, since viable organisms are required to colonize the gut, C. botulinum spores in food (especially, honey and syrup) may act as the reservoir for infection (65,66). In the majority of infant botulism cases, a reservoir has not been identified and other environmental exposures (e.g., spores in dust) have been hypothesized as the reservoir of infection (65,66). A recent case in England identified contaminated dried infant-formula milk as a possible source of infection (59).
Historically, poorly processed canned and bottled foods have been associated with foodborne botulism outbreaks. However, because canning and bottling processes are
better controlled by the food industry, home produced products now represent the greatest risk for intoxication. However, swollen or flat cans from any source (flat cans imply loose seams sufficient to allow loss of gas) suggest poor processing and should be considered as likely reservoirs when investigating suspect botulism cases.
2.4. Incidents of Botulism and Risk Factors 2.4.1. Foodborne Incidents
Foodborne outbreaks of botulism occur worldwide and have the potential to cause large morbidity and mortality that require considerable public heath and acute care resources. In addition, the costs for treatment, investigation, food recalls, and any legal actions are likely to be in millions of dollars/pounds even for a relatively small outbreak (67,68). The potential for large outbreaks is illustrated by an outbreak of 91 botulism cases with 18 deaths related to type E and which were associated with traditionally salted fish dish (fesaikh) in Egypt in 1991 (69). Fesaikh is prepared from ungutted fish that, in this outbreak, were suspected to have been salted, warmed (placed in the sun) and then sealed into barrels. This environment was clearly sufficient to allow germination and growth of C. botulinum type E, probably in the fish viscera.
Table 2
Occurrence of C. botulinum in Soils and Sediments (7)
Sample BoNT type (% Identified)
Location size (g) % Positive MPN/kga A B C & D E F
USA, eastern, soil 10 19 21 12 64 12 12 0
USA, western, soil 10 29 33 62 16 14 8 0
USA, Green Bay, 1 77 1280 0 0 0 100 0
sediment
USA, Alaska, soil 1 41 660 0 0 0 100 0
Britain, soil 50 6 2 0 100 0 0 0
Britain, coastal 2 4 18 0 100 0 0 0
sediment
Scandinavian, 6 100 >780 0 0 0 100 0
coastal sediment
The Netherlands, soil 0.5 94 2500 0 22 46 32 0
Switzerland, soil 12 44 48 28 83 6 0 27
Rome, Italy, soil 7.5 1 2 86 14 0 0 0
Caspian Sea, Iran, 2 17 93 0 8 0 92 0
sediment
Sinkiang, China, soil 10 70 25000 47 32 19 2 0
Japan, Hokkaido, soil 5–10 4 4 0 0 0 100 0
Japan, Ishikawa, soil 40–50 56 16 0 0 100 0 0
Brazil, soil 5 35 86 57 7 29 0 7
Paraguay, soil 5 24 10 14 0 14 0 71
South Africa, soil 30 3 1 0 100 0 0 0
Thailand, sediment 10 3 3 0 0 83 17 0
New Zealand, soil 20 55 40 0 0 100 0 0
aMPN/kg, highest most probable number/kg of sample.
As previously stated, foodborne outbreaks are now most often associated with home preservation (e.g., canning, bottling, and preservation in oil) and hence are most common in parts of the United States, central and southern Europe where these preservation practices are more common (8,70,71). In the United States, between 9 and 10 outbreaks of botulism have been reported each year since 1899 (8,70). Home-canned vegetables have traditionally been recognized since the 1950s as the most common cause and were responsible for over 40% of all outbreaks (8,70). Foodborne botulism in an infant has been reported resulting from home-canned infant food (72). Most of the remainder of the cases have been related to readily identifiable poor processing and handling practices as the likely causes allowing the growth of C. botulinum (Table 3). The majority of incidents of foodborne botulism in the United States are related to C. botulinum type A and to a lesser extent type B, except in Alaska (8,70). Recent changes in ‘traditional’
fish and marine mammal preservation practices in Alaska have resulted in the highest inci- dence in this state where 90% of outbreaks are related to type E (70). A list of foods and processing problems for foodborne outbreaks in the United States is shown in Table 3.
All identified foodborne botulism incidents in the United Kingdom are shown in Table 4. The two most recent outbreaks were associated with home-produced products (bottled mushrooms in 1998 and sausage in 2002), both of which were owing to type B and were produced in areas of Europe where botulism is much more common (150,151). The three most recent outbreaks associated with commercially prepared products all had evidence of poor handling and processing. The outbreak associated with tinned salmon in 1978 was caused by a defect in a single can that was contaminated from the factory environment: most probably from factory overalls which were allowed to dry over cans during cooling after being autoclaved (73). The second outbreak was related to a ‘shelf-stable’ airline meal which was given an unfeasibly long shelf-life which supported the growth of C. botulinum type A (74). The third recent outbreak (the largest recorded in the United Kingdom) associated with a commercial product was caused by the consumption of hazelnut yogurt in 1989 (75). This is an unusual vehicle for transmission because the pH of yogurt is normally too low to allow the growth of C. botulinum. However, the toxin production occurred in cans of hazelnut conserve which were not sufficiently heated to kill this bacterium and did not contain any other physico-chemical hurdles to prevent the growth of C. botulinum. The conserve (together with neurotoxin) was added as a flavoring at the end of the yogurt manufacture, and clearly sufficient toxin retained its biological activity in the acid environment to cause illness.
2.4.2. Infant Botulism
Infant botulism results from the colonization of the immature enteric tract by C.
botulinum that produces neurotoxin in vivo. As stated previously, risk factors for infant botulism include consumption of honey and corn syrup, age (95% of case occur in the first 6 mo after delivery, the remainder at less than 12 mo), and location (California has the highest rate wordwide). However, the consumption of honey and corn syrup is an identifiable risk factor in <20% of cases (65). Although the disease is orally acquired, it is not clear what proportion is foodborne, or contracted via oral exposure to the environment. The attack rate for infection is likely to be very low and all cases appear sporadic. Almost all cases are owing to C. botulinum types A and B (6), with additional
very rare cases related to C. botulinum type C (21), C. baratii (type F) and C. butyricum (type E; [23–26]).
In the United Kingdom, there have been six cases of infant botulism identified between 1978 and 2001 (Table 5). Infant-formula milk powder was identified as a source of infection in only one case (59). All these six cases have been at the severe end of the spectrum of illness and were ventilated. This raises the possibility of under-recognition in the United Kingdom from the fact that not all cases of infant botulism require ventilation (3).
Table 3
Foodborne botulism in the United States, 1990–2000. Adapted from ref. (7)
Food type Incidents Cases Comments
Continental USA and Hawaiia
(Total 160 cases: 130 type A, 16 type B, 6 type E, 3 type F, and 5 not known)
Noncommercial home-canned 47 70 Likely to be weak acid foods (pH >4.6).
vegetables Asparagus (9 incidents, 14 cases)
and olives (4 cases) were the most common vehicles. Also includes two episodes with garlic in oil which was insufficiently heated to kill
C. botulinum spores
Noncommercial home-prepared 7 9 Sausages (3 cases) were the most meat products (sausage, pate, common vehicle. Failure to refrigerate beef chili, meatballs, roast beef chili after cooking associated
beef, hamburger) with two cases
Noncommercial home prepared 14 18 Soup (4 cases) of unspecified type was other products (salsa, potato the most common vehicle. Salsa salad, bread pudding, soup, was responsible for 2 cases and was apple pie, potatoes, and prepared from raw vegetables and
pickled herring) stored at room temperature in an
airtight plastic container
Commercial (preserved fish, 5 10 All outbreaks involved poor handling
burrito, and bean dip) practices
Restaurant made (cheese sauce 2 25 One outbreak (17 cases) was associated
and skordalia) with skordalia which involved
potatoes baked in aluminium foil and left at ambient temperature for several days. The second outbreak (8 cases) involved a cheese sauce which was left unrefrigerated
Other and unknown 26 27 One case associated with peyote tea Alsakab
(103 cases: 11 type B, 91 type E, and 1 not known)
Noncommercial preserved fish 49 92 All associated with native foods
or marine mammals consisting of whale, beaver, seal, or fish
Unknown 3 11
Total 160 263
aIncidence rate <0.1–0.6 cases per million.
bIncidence rate 19 cases per million.
2.5. Pathogenicity and Virulence Factors 2.5.1. Characteristics and Action of Toxins
BoNT is directly responsible for the characteristic clinical symptoms of botulism which arise as a result of blocking of acetylcholine release from peripheral cholinergic sites, particularly neuromuscular junctions leading to typical flaccid paralysis (14,17).
All seven BoNTs (A–G) are synthesized as single polypeptide chains and have an approximate molecular weight of 150 kDa. To become biologically active, the single polypeptide is proteolytically cleaved, usually by a botulinum enzyme, to a heavy (~100 kDa) and a light chain (~50 kDa) linked by a single disulfide bond and other noncovalent bonds to give the biologically active toxin (Fig 2.; [15]). The BoNT heavy chain contains two functional domains: the C-terminal end, responsible for both receptor binding and internalization; and the N-terminal end, involved in translocation of the light chain across the endocytic membrane into the neuronal cell cytoplasm (15,16). The central region of the light chain shows a high degree of homology conserved in all toxin types and contains the active site of the enzyme, now identified as a zinc endopeptidase, specific for SNARE proteins. SNARE proteins are part of the hosts neurotransmitter release apparatus and when inactivated by the light-chain prevents acetylcholine release.
X-ray crystallography analysis of BoNTA demonstrates the three structurally distinct functional domains, namely, the enzymatic domain corresponding to the light chain, the translocation domain corresponding to the N-terminal part of the heavy chain, and the binding domain composed of two subdomains corresponding to the C-terminal part of the H chain. The enzymatic domain is composed of a mixture of G-sheets and F-helices.
The translocation domain, at the N-terminal of the heavy chain, is characterized by two longF-helices that are twisted around each other in coiled-coil-like fashion and a long loop that wraps around the catalytic domain and occludes the active site. The two subdomains comprising the C-terminal of the heavy chain are approximately equivalent Table 4
Foodborne Botulism in the United Kingdom 1922–2002
Implicated food (Country
Year Cases (Deaths) Home-prepared of origin, if outside the UK) BoNT type
1922 8 (8) No Duck pâté A
1932 2 (1) Yes Rabbit and pigeon broth NK
1934 1 (0) Yes Jugged hare NK
1935 5 (4) Yes Vegetarian nut brawn A
1935 1 (1) Yes Minced meat pie B
1949 5 (1) Yes Macaroni cheese NK
1955 2 (0) NK Pickled fish (Mauritius) A
1978 4 (2) No Canned salmon (USA) E
1987 1 (0) No Rice and vegetable shelf A
stable airline meal
1989 27 (1) No Hazelnut yoghurt B
1998 2 (1) Yes Bottled mushrooms (Italy) B
2002 1 (1) Yes Sausage (Poland) B
NK, not known.
Based on ref. (140,151).
in size and contain predominantly G-sheet structures. The N-terminal subdomain has a so-called “jelly roll fold,” whereas the C-terminal subdomain forms a G-trefoil fold and has a G-hairpin at the base of the domain (15).
BoNTs are released from bacterial cells noncovalently complexed with hemagglutinin (HA) and nontoxic, nonhemagglutinin (NTNH) proteins, which, although do not play a role in blocking neurotransmitter release, appear to have a crucial role in protecting BoNT from low pH and enzymic degradation in the human gut (13). BoNT protein complexes dissociate at physiological pH and ionic strength, and thus, intact complex does not reach the peripheral nerve endings and most likely dissociates in the circulation and lymph (17). BoNT complexes vary in size from 300 to 900 kDa depending on toxin type, strain and growth conditions, particularly the availability of iron (12).
Once orally ingested, the BoNT complex is transported across the lumen of the small intestine via intestinal epithelial cells and passes into interstitial fluid and into the circu- lation. Although transportation across the gut lumen is thought to involve receptor binding, the exact location of the binding domain on the BoNT complex remains controversial with either HA protein or BoNT itself having been proposed as mediating binding (17). On gaining access to the bloodstream and lymph, BoNT is transported to its target site, the neuronal cell. In order to exert its effect, BoNT must leave the vasculature although, at present, it remains to be elucidated if this transportation is by diffusion between the endothelial cells or if an active process of endocytosis and transcytosis is involved (17).
The BoNT mechanism of inhibition of acetylcholine release occurs in four stages:
binding, internalization, translocation and intracellular enzymic activity (Fig. 2). Firstly, BoNT binds to the plasma membrane of the neuronal cell followed by internalization by receptor-mediated endocytosis. Then, BoNT is translocated across the endosome membrane, and finally, there is a intracellular endopeptidase action on the SNARE proteins leading to the abolition of endocytosis of acetycholine (13,17). BoNT binds to specific receptors on unmyelinated areas of the presynaptic membrane. Although the exact identity of the BoNT receptor has not been fully elucidated, there is considerable experimental and structural evidence to suggest that binding involves a dual receptor composed of a polysialoganglioside of the G1b series and a specific glycoprotein on the surface of the neuromuscular junction, and it has been proposed for BoNT/B that synaptotagmin II associated with GT1bcould be the specific receptor (13,14,17). Different neurotoxins bind to specific receptors as evidenced by the considerable sequence diversity in the binding domain of the heavy chains of the different BoNT serotypes despite their Table 5
Infant Botulism in the United Kingdom
Laboratory confirmation by analysis of
Year Age (in mo)/Sex Toxin Type Feces Serum Ref.
1978 5.5/F A Toxin & organism ND (141)
1987 4/M B+F Toxin & organism ND (142)
1989 2/F B Toxin ND (143)
1993 4/F B Toxin & organism Toxin (144)
1994 4.5/M B Toxin & organism Toxin (145)
2001 5/F B Toxin & organism ND (146)
M, male; F, female; ND, not detected.
similar structural homology. Receptor specificity together with specificity for the intra- cellular target accounts for the different sensitivity of animal species to the different serotypes (13). Once bound to its specific receptor, BoNT is internalized by receptor- mediated endocytosis and remains at the nerve terminal where the catalytically active light chain is translocated into the cytosol. The precise mechanism of internalization is not known although it involves clathrin-coated vesicles and is temperature- and energy-dependent (13,16).
During translocation, the light chain of BoNT crosses the endocytic membrane to reach the neuronal cytosol. The process is triggered by acidification of the endocytic vesicle which induces a conformational change in the N-terminal portion (translocation domain) of the heavy chain such that it inserts into the lipid bilayer forming possibly either a channel or cleft which facilitates translocation of the partially unfolded light chain through the membrane (15,16). Although the precise details of the translocation Fig. 2. Schematic representation of the structure and mode of action of C. botulinum neurotoxin.
(A) Cleavage of single neurotoxin peptide to heavy and light chain. The black area (heavy chain) is the translocation domain, and grey the binding domain. (B) Schematic mechanism of action of botulinum neurotoxin.
process have yet to be established, at some point both the disulfide bond and noncovalent bonds linking the heavy and light chains are cleaved. Once in the cytoplasm, the light-chain refolds owing to the neutral pH (13) and is able to proteolytically cleave highly specific synaptic proteins involved in neurotransmitter release.
The light chain of the BoNT containing metallo-protease activity acts on specific synaptic proteins known as soluble NSF-attachment protein receptors (SNAREs) that are involved in the exocytosis of acetylcholine. SNARE proteins form part of a complex that is involved in fusion of the acetylcholine containing synaptic vesicle with the pre- synaptic membrane. BoNT A and E cleave SNAP-25 (synaptosomal-associated proteins of 25 kDa); BoNTs B, D, F and G cleave synaptobrevin (also known as vesicle-associated membrane or VAMP) and C can act on both the SNAP-25 and syntaxin. Each clostridial neurotoxin recognizes its substrate at specific binding sites (SNARE motifs) and each cleaves a different peptide bond even when the substrate is the same (17). SNARE proteins must be in the uncomplexed form for BoNT action to occur and although cleavage does not prevent formation of the SNARE complex, it results in a nonfunctional complex in which the uncoupling of Ca2+ influx and fusion prevents neurotransmitter being released (13,15). The duration of action of the different BoNTs varies significantly with BoNTA having the most persistent action. Duration of BoNT action may be effected by differences in either the persistence of BoNT LC in the cytosol and or persistence of the specific cleaved protein in the SNARE complex. An important factor effecting recovery from BoNT is how quickly and successfully nerve sprouting occurs.
2.5.2. Molecular Genetics of Clostridium botulinum Neurotoxins
C. botulinum types A, B, E and F neurotoxin genes are encoded chromosomally, whereas in types C and D the genes are found on a bacteriophage (Table 1). The BoNT genes are clustered in close proximity to the nontoxic nonhemagglutinin (NTNH) and in most strains with hemagglutanin (HA) genes, forming what is known as the botulinum neurotoxin locus. A positive regulatory gene botR coding a 21–22 kDa protein is asso- ciated with the neurotoxin gene locus. C. botulinum strains usually produce only one type of neurotoxin and the neurotoxin locus is present in only one copy on the genome.
However, in a few strains, two toxin types are synthesized, one usually being produced in excess of the other. These strains contain two BoNT genes. Some C. botulinum type A strains have silent bont B genes (76), which, although present on the genome with a fully functional BoNTA gene, contain several mutations within the coding region leading to transcription disruption.
The ability of clostridial species other than botulinum to produce neurotoxin, the presence of more than one bont gene in some botulinum strains, as well as the similarity between BoNT and tetanus toxin, indicates that bont genes are derived from a common ancestor and are transferable between different clostridial strains; it is likely that this is mediated by mobile genetic elements (18).
2.6. Clinical Characteristics
Early symptoms of foodborne or infant botulism affect the gastrointestinal tract and often involve nausea, abdominal pain, vomiting and diarrhea followed by constipation.
Neurological symptoms then follow and are the same, irrespective of the route of entry of botulinum neurotoxin. The classical picture is of descending symmetrical flaccid
paralysis, with no fever. Early symptoms include cranial nerve palsies—blurred and double vision (diplopia), difficulty in focusing, drooping eyelids (ptosis), facial weakness, sluggishly reacting or enlarged pupils, difficulty in swallowing (dysphagia), difficulty in speaking (dysphonia) and slurred speech (dysarthria). Weakness in the neck and arms, loss of the gag reflex, weakness in lower limbs, and respiratory paralysis may follow.
There may be autonomic signs with dry mouth, fixed or dilated pupils, and gastro- intestinal, urinary, and cardiovascular dysfunction. Altered sensory awareness and fever are not associated with botulism. Deep tendon reflexes may decrease over time in some cases. Initial differential diagnoses from foodborne botulism includes: Guillain-Barré syndrome, myasthenia gravis, spinal/paralytic poliomyelitis, viral or bacterial encephalitis, rabies, cerebrovascular accident, tick paralysis, paralytic shellfish poisoning, diphtheria, chemical intoxications (carbon monoxide, barium, methyl chloride, methyl alcohol, organic phosphorus, and atropine), mushroom poisoning, and reaction to pharmaceutical com- pounds such as antibiotics (151,8).
For foodborne botulism, the incubation period ranges from 6 h to 10 d, but symptoms are generally present 18–36 h after consumption of contaminated food. In both foodborne and infant botulism there is a spectrum of disease: severity being related to the amount of neurotoxin. If illness is severe, involvement of respiratory muscles occurs, and venti- latory failure and death can occur unless supportive care is provided. Recovery follows regeneration of new neuromuscular connections and 2–8 wk of ventilatory support is common: some patients may require >7 mo before the return of muscular function.
The presentation of infant botulism is very similar to foodborne disease; however, since the infants are unable to complain, symptoms often develop suddenly. In mild cases, weakness, lethargy, and reduced feeding occur. In severe infant botulism cases weak- ened cry, suck, and swallowing are observed together with muscle weakness, diminished gag reflex, and loss of head control. Infants are described as “floppy”. In infants, sepsis (especially meningitis), electrolyte-mineral imbalance, metabolic encephalopathy, Reye syndrome, Werdnig–Hoffman disease, congenital myopathy, and Leigh disease should also be considered (8).
Although intestinal colonization by C. botulinum is classically associated with infants of <1 yr old, this has also been described, albeit very rarely, in an older infant and among adults (77–82). The diagnosis is supported by prolonged excretion of toxin, C. botulinum in the feces and an absence of a specific contaminated food vehicle. A single case has been reported in a 3-yr-old neuroblastoma patient following intense immuno- suppression and antibiotic treatment (83). Specific risk-factors among adults include Chron’s disease, gastrointestinal surgery, and prior antibiotic treatment. A cases of C. baratii intestinal colonization in an adult has been reported (79). The reservoirs of infec- tion in these cases of adult colonization are equally obscure as for the majority of infant botulism cases, but clearly food and other environmental sources such as dust are the likely candidates.
Similarities exist between infant botulism and sudden infant death syndrome (SIDS) and causal link has been suggested (6). Botulinum neurotoxin (especially type E) was detected in the feces of 12% of SIDS infants (84) although this was not a universal finding (85) and this connection remains controversial.
As noted previously, most cases of botulism are related to BoNT A and BoNT B, BoNTE in which there is an association with marine products, and occasionally BoNT F
(1,6,9). In addition, a single case of infant botulism owing to C. botulinum type C (21) together with food and infant botulism related to C. baratii (type F), and C.
butyricum (type E; [23–26]) have been described. However, there are no recognized differences between the clinical characteristics of botulism and the toxin type or the Clostridium species.
2.7. Treatment and Control
The mainstay of treatment of foodborne botulism is the inactivation of toxin in the patient by intravenous administration of equine antitoxin. Early administration is essential because the action is directed towards free toxin which has not yet bound to nerve endings.
A single dose is given because the antibodies have a half life of 5–8 d in the patients serum. Treatment is not without risk because up to 9% of patients are hypersensitive to horse serum; an initial skin test prior to administration is advised to establish this. In contrast with foodborne botulism, treatment with equine antitoxin is not recommended in infants because of the possibility of serum sickness. A human derived antitoxin product has been used in the United States (86). Treatment of infant botulism is primarily through meticulous supportive care. Antibiotic treatment has no role in the treatment of foodborne or infant botulism because this may increase toxin release.
Because the administration of antitoxin is the only specific therapy for foodborne botulism, and that this is only effective early in the course of neurological dysfunction (87), the decision for treatment is almost always based on clinical observation, case history, and physical findings; and should not be delayed for the results of laboratory tests.
The control of C. botulinum in the food chain relies on the killing of organisms in foods that will support the growth of this bacterium, or the formulation of food ingredi- ents and processes to prevent growth. Although many foods satisfy the nutritional requirements for the growth of C. botulinum, not all of these have the necessary anaerobic environments to support growth. Anaerobic requirements are, however, supplied in many canned and bottled vegetables, meat, and fish products, and these should either be subjected to a botulinum 12D ‘kill’ or have one or more of the following to prevent growth: low pH, low water activity, high salt, high sodium nitrite, and/or other preserv- atives. Refrigeration will not prevent growth and toxin production unless the temperature is kept below 3°C. Foods processed to prevent spoilage but not refrigerated are the most common vehicles of botulism. Because of the association of infant botulism with honey, and since this food does not undergo a botulinum ‘cook,’ some countries and manufactur- ers recommend not feeding honey to infants less than 1 yr old.
Because of the extremely protracted course of both foodborne and infant botulism, the cost of treatment can be considerable. For example, it was estimated that an average cost for hospitalization and treatment of a single infant botulism case was $80,000, with those infants with most protracted illness (>10 mo hospitalization) to be >$635,000 (6).
2.8. Summary and Conclusions
Foodborne botulism results from the ingestion of preformed toxin in contaminated foods which may still be available to cause illness in others and a single case can precede a severe public health emergency. It is, therefore, critical for acute care clinicians to be able to rapidly recognize botulinum cases and to alert public health microbiologists to investigate incidences, identify implicated food sources, and advise those responsible
for food regulation on rapid withdrawal of implicated product and appropriate addi- tional interventions. Identical considerations apply to a deliberate release (bioterrorism) incident if botulinum toxin or C. botulinum (or indeed another species of bacteria engineered to produce botulinum neurotoxins) were to be introduced into the food chain. The use of a mouse bioassay to detect active toxin has been the mainstay for laboratory diagnosis and identification of this very diverse group of bacteria; however, molecular biological advances (either for the detection of neurotoxin genes by PCR, detection of toxin by immunoassay, or the detection of biologically active toxins using specific in vitro assays) are likely to become more widespread in the future.
Infant botulism is also a foodborne disease, but occurs via the production of neuro- toxin in the immature intestinal tract. The epidemiology is less well-understood than foodborne botulism, but the molecular advances outlined above may also contribute to improved diagnosis and a greater understanding of the disease.
3. CLOSTRIDIUM PERFRINGENS 3.1. Introduction
Clostridium perfringens is probably one of the most widely occurring bacterial pathogens and is ubiquitous in the environment and as part of the normal intestinal flora of humans and other animals (1). This bacterium causes a wide range of diseases in humans, indeed there is hardly a site that has not been reported as infected by C. per- fringens especially following penetrating wounds (88). The most commonly occurring C. perfringens infections are summarized in Table 6.
The pioneering work of Hobbs et al. (89) extended earlier observation and firmly established this bacterium as a frequent cause of foodborne diarrheal disease (see Section 3.4.1.). Among those diseases which are transmitted to humans via the oral route it is now known that C. perfringens (type A) is not only responsible for diarrhea associated with the consumption of contaminated food as originally described by Hobbs et al. (89), but also antibiotic associated and infectious diarrhea. C. perfringens (type C) is also a cause of necrotizing enterocolitis (Pig-bel or Darmbrand) in humans which Table 6
Diseases Caused by C. perfringens
C. perfringens Diseases in
type Humans Other animals
A Gas gangrene; foodborne, Enterotoxemia (lambs, cattle, goats, antibiotic-associated, horses, dogs, alpacas, kangaroos, pigs, and infectious diarrhea rabbits, reindeer, and others)
Necrotic enteritis (domestic and wild birds) Acute gasrtic dilation (nonhuman primates
and others)
B, D, and E None identified Enteritis and enterotoxemia (sheep, goats, guinea pigs, rabbits, and others) C Necrotizing enteritis (jejunitis) Enterotoxemia (lambs)
Necrotic enteritis (piglets, lambs, calves, foals, and birds)
is also foodborne. This bacterium also causes enteric infections in a very wide range of other animals (Table 6). A more detailed description of the intestinal diseases in humans is given in later sections.
Studies of infectious intestinal disease in the 1990s identified C. perfringens as responsible for diarrhea in 4 and 2.3% of patients in the community in England and Wales (152) and in Holland (90), respectively. The study in England and Wales also esti- mated that for each case identified in the laboratory, an additional 186 cases present to a medical practitioner, and 343 cases occur in the community (91). Thus demonstrating that this disease is considerably underdiagnosed. Further analysis in England and Wales (92) and in the United States (93) estimated C. perfringens as the second and the ninth most common foodborne diseases, respectively (Table 7). Adak et al. (92) further estimated that C. perfringens was responsible for 2% of the total hospitalizations and 18% of all the deaths from preventable foodborne diseases in England and Wales (Table 7).
More detailed reviews about C. perfringens and gastrointestinal disease can be found in the following literature (94–100).
Table 7
Estimated Illness Resulting From 10 Most Common Foodborne Pathogens in the United States, and England and Wales. Adapted for refs. (92,93)
Numbers of cases
Total Hospitalizations Deaths USA (After 1993)
Total 76,000,000 323,000 5200
Norovirus 23,000,000 50,000 310
Rotavirus 3,900,000 50,000 30
Astrovirus 3,900,000 12,500 10
Campylobacter 2,453,926 13,174 124
Giardia 2,000,000 5,000 10
Salmonella 1,412,498 16,430 582
Shigella 448,240 6,231 70
Cryptosporidium 300,000 1,989 66
C. perfringens 248,520 41 7
Toxoplasma gondii 225,000 5,000 750
England and Wales (After 1992)
Total 1,338,772 20,759 480
Campylobacter 359,466 16,946 86
C. perfringens 84,081 354 89
Norovirus 57,781 37 9
Yersinia 45,144 216 1
Salmonella 41,616 1,516 119
Astrovirus 17,291 12 4
Bacillus 11,144 27 0
Rotavirus 8,979 46 4
Staphylococcus aureus 2,276 57 0
Cryptosporidium 2,063 39 3
Giardia 1,673 5 0
