Naming the guilty – identifying the causes of food poisoning outbreaks
When an outbreak of foodborne infection is reported, one of the most important tasks facing investigators is to identify the microorganism responsible. But determining the species is often not enough. Outbreak strains of bacteria or viruses need to be characterised (typed and subtyped) in much greater detail, both to confirm that suspected cases are linked, and to help trace the outbreak back to its source. This can be done to some extent by phenotypic analysis, but today it is often genetic analysis that proves conclusive. Now the latest sequencing technology is promising to revolutionise the investigation of food poisoning outbreaks.
There are currently around 8,000 recognised species of bacteria, but only a relative handful, along with a few viruses and protozoa, are known to cause foodborne infections. When a probable food poisoning outbreak is reported, clues to the identity of the cause often lie in the patients’ symptoms. Analysis of clinical samples based on those clues can then isolate the agent responsible and identify the species. Once that information is available it becomes possible to look for the agent in food samples and identify the source of the outbreak, so that the offending food can be taken out of circulation and further cases prevented.
Unfortunately, things are rarely that simple. Take Salmonella, one of the commonest causes of foodborne infection, for example. Few non-microbiologists realise that there are only two recognised species and six subspecies of Salmonella, which is of little use when identifying isolates. In fact Salmonella has been classified by serological testing for more than 80 years, and familiar names like Salmonella Enteritidis are not those of species but names assigned to individual serovars (serotypes). There are now about 2,500 serovars in the serotyping system for Salmonella (the Kauffmann-White scheme), differentiated by the reaction between specific antibodies and antigens on the cell wall and the flagellae. If the serovar involved in a food poisoning outbreak is very rare, serotyping may be sufficient identification, but most outbreaks are caused by serovars such as S. Enteritidis and S. Typhimurium, which are common in animals, the environment and in human infection. In the vast majority of outbreaks more discriminating typing methods are needed.
A long established method of typing isolates beyond the serovar level is phage typing. This relies on the fact that some viruses that infect bacteria (bacteriophages) are very specific and only attack individual strains. By testing against a number of different bacteriophages specific to the species or serovar, a bacterial isolate can be characterised further. This technique has been in use for several decades and is still a valuable tool to type some food poisoning bacteria, including Salmonella. But there are problems with phage typing. The interpretation of the results can be subjective and may vary between laboratories, so that the same strain may not always be classified as the same phage type. Furthermore, some phage types, such as S. Enteritidis phage type 14b, are common and regularly cause human infection.
One technique routinely used in outbreak investigation is to take advantage of the fact that many bacterial isolates are resistant to one or more antibiotics. By testing the susceptibility of an isolate to a range of different antimicrobial drugs, it is possible to build up an antibiotic resistance pattern. This provides an additional level of characterisation, which can be very useful in epidemiological investigations for comparing isolates from different patients and from food samples.
All the techniques mentioned so far are based on phenotypic characteristics and although they are all well established and relatively inexpensive to perform, there are drawbacks. Some tests are difficult to interpret and may not give consistent results, but more importantly, using phenotypic characteristics to type bacteria may not provide enough information to fully differentiate closely related strains of some pathogens. But the adoption of molecular biology-based techniques has lead to a dramatic improvement in the ability of laboratories to distinguish individual bacterial strains.
One of the first techniques to be applied to typing foodborne pathogens, and still regarded as the gold standard by many epidemiologists, is pulsed field gel electrophoresis (PFGE), sometimes referred to as genetic fingerprinting. This technique involves breaking open bacterial cells and then adding an endonuclease enzyme, which splits up the cell’s DNA into a number of large fragments. These can be separated by applying a pulsed electrical field to the cell extract on a gel plate and then visualised by fluorescent staining. The result is a pattern of distinct bands formed by the DNA fragments, which can be used to identify and differentiate many bacterial strains by comparing the pattern with those held in reference databases. PFGE is routinely used to type foodborne pathogens, including Salmonella, Listeria and E. coli O157, involved in outbreaks. The major disadvantages of PFGE are the length of time – about three days – it takes to complete, and the relatively high cost.
The development of polymerase chain reaction- (PCR) based testing has added another useful weapon to the epidemiologist’s arsenal. PCR can be used to amplify and detect individual genes on the bacterial genome, such a genes that confer the ability to produce specific toxins and other virulence factors. PCR can confirm the findings of phenotypic typing and add additional information very quickly.
Some more recent and potentially very powerful genetic techniques are also now finding their place in outbreak investigation. Multi locus sequence typing (MLST) is one example. MLST involves sequencing 400-500 base pair fragments of DNA of seven different ‘housekeeping genes’ and allows small variations within a species to be detected. Results can be compared against databases held at Imperial College in London. A related technique now being used by reference laboratories to type Salmonella isolates is multi locus variable number of tandem repeats analysis (MLVA), which is based on PCR amplification and sequencing of rapidly mutating repetitive DNA sequences called tandem repeats. The result is an MLVA pattern characteristic of the bacterial strain under investigation. MLVA has been used successfully to differentiate Salmonella Typhimurium strains of the same phage type and is also being developed for typing other Salmonella serovars.
Other nucleic acid-based techniques employed to type bacteria include ribotyping, which relies on the relative stability of the 16S and 23S rRNA genes coding for ribosomal RNA to produce a characteristic electrophoresis fingerprint of DNA fragments, and repetitive sequence-based PCR (rep-PCR). Both of these have been developed into commercial typing systems that can be used to analyse the genotypes of a range of bacterial species.
Clearly the ideal way to identify and differentiate bacterial isolates would be to sequence the entire genome. Until recently, this would have been completely impractical, very costly and taking several years to complete. But new, next-generation sequencing systems, such as Life Technologies semiconductor-based Ion Torrent PGM™ system, make it possible to sequence the microbial genome in days rather than months or years. Sequencing has already been used to investigate the genetics of human pathogens involved in outbreaks and promises to revolutionise the process in the near future. Sequencing hepatitis A viruses involved in outbreaks associated with green onions in the USA revealed that they were closely related to strains commonly found in people living in the area where the onions were grown, valuable information for the outbreak investigators. Sequencing is likely to replace PFGE as the gold standard for pathogen typing in the near future.
A recent example of how all these techniques play their part in the investigation of foodborne outbreaks is the 2011 German E. coli O104:H4 outbreak. The technical report from the Robert Koch Institute shows how the isolates from the outbreak were characterised and the range of tests that were used. These included, serotyping, antibiotic susceptibility testing, PFGE, PCR screening for specific virulence factors and MLST. Finally it is worth noting that the entire genome of the outbreak strain was sequenced, within three days, confirming the other findings and adding valuable information about the likely origin of the strain and its relationship with other isolates.