Introduction

In the majority of countries in Western Europe, Salmonella is a zoonotic pathogen with its primary reservoirs being poultry, cattle and pigs. Salmonella organisms are transmitted through the food chain to humans with contaminated foodstuffs such as beef, chicken, turkey and pork being important sources of infection. In recent years, salad products have also been implicated as vehicles of infection (1-3). International trade both in food animals and food products ensures that salmonella organisms are widely distributed throughout the European Union (EU), and that international outbreaks occur regularly.
Much salmonellosis prevention and control depends on early outbreak recognition through a suitable surveillance system based on isolate subtyping. The principal internationally accepted method for the subtyping of salmonellas is serotyping, followed by phage typing for discrimination within the most common serotypes. The value of phenotypic typing methods as surveillance tools is well established but because of the predominance of certain serotypes and phage types in many countries, DNA fingerprinting is often used as an adjunct in outbreak investigations in which enhanced strain discrimination is needed. A number of DNA based methods, including ribotyping, insertion sequence 200 fingerprinting, and pulsed field gel electrophoresis (PFGE) are available. Of these, PFGE has become the gold standard for strain subdivision both within serotypes and phage types (4).

Analysis by PFGE is highly discriminatory and can subdivide bacterial isolates relating to possible outbreak situations. As a method it can therefore be used to make decisions of epidemiological importance. In the EU, PFGE has already been applied to international outbreaks of S. enterica, including Enteritidis, Typhimurium, Anatum, Virchow and Hadar, for which contaminated food products have often been the vehicles. It has also been used elsewhere in the world to establish foodborne outbreaks of Typhimurium, Paratyphi, Agona, Stanley, Saphra, and Javiana. Of particular note is the recent outbreak of Stanley associated with peanuts produced on one continent with cases being identified on three others (5).
In the United States (US), a subtyping network, PulseNet US, based at the Centres for Disease Control and Prevention (CDC), Atlanta, uses PFGE technology to track E. coli O157 throughout the country, and this has recently been extended to S. enterica (6). Having demonstrated its effectiveness as a tool for foodborne disease surveillance, this model has been implemented in Canada and is at various stages of development in the Asia Pacific region and South America. However, phage typing of Enteritidis and Typhimurium strains is not undertaken routinely in the US.

The Salm-gene EU research project was therefore designed to assess the added value for outbreak recognition of routine molecular subtyping, using PFGE in particular, of the two predominant salmonella serotypes in an environment where phage-typing is also employed routinely. To establish PFGE as an effective method for the subtyping of Salmonella within the EU, it is essential that all parameters are internationally agreed and harmonised.
The ten laboratories participating in the Salm-gene project are already members of Enter-net (7), the international surveillance network for human gastrointestinal infections. The main aims of Salm-gene are: i) to develop standard laboratory operating procedures for PFGE and for computer recognition of the results, ii) to create a searchable database of PFGE profiles for the major Salmonella serovars currently in circulation within Europe, iii) to DNA fingerprint in real time a large sample of salmonella strains in each of several countries, using selection criteria that maximise outbreak detection power, and analyse the data continuously in the online database, iv) to establish an external quality assurance (EQA) scheme for PFGE.
Satisfactory harmonising of PFGE testing across laboratories is essential if comparable data are to be collected and is fundamental to the Salm-gene project. We report the results of testing a panel of carefully selected strains in each of the nine participating Salm-gene laboratories, using a harmonised protocol.

Methods

Participants in the Salm-gene project include the Laboratory of Enteric Pathogens (acting as the Project Co-ordinator in conjunction with the Enter-net surveillance hub, Communicable Disease Surveillance Centre, United Kingdom) together with eight other national reference laboratories within Europe. Each participating centre received a set of 16 S. enterica strains to be used for EQA of the method. These strains included the serovars Typhimurium, Enteritidis, Hadar, Virchow, Agona, Heidelberg, Indiana, Montevideo, Mbandaka, Livingstone, Anatum, London, Senftenberg, and Poona; two phage types of Typhimurium and Enteritidis were included. These strains were selected to provide a wide variety of PFGE profiles such that well defined chromosomal fragments were present in all areas of the gel.

An agreed protocol for PFGE was performed using the Bio-Rad CHEF® system. This involved proteinase K lysis of cells, a series of washes at 50oC followed by digestion with 30U XbaI (minimum 4h, 37oC). Electrophoresis conditions were as follows: RAMP - Initial 2s; Final 64s; 6V/cm; 14oC, 22h (CHEF DRII®), 20h (CHEF DRIII®), 18h (CHEF Mapper®). DNA macrorestriction fragments were resolved on 1% agarose gels (Bio-Rad Pulsed Field Certified® or Seakem Gold®) with a S. Braenderup strain (kindly supplied by PulseNet US, CDC) as a molecular reference marker.

Gel images were exchanged in tag image file format (TIFF files). PFGE patterns were analysed with BioNumerics software using Dice coefficient and Unweighted Pair Group Method of Averages (UPGMA) with a 1.5% tolerance limit and 1.5% optimisation. Pulsed field profiles were assigned on a temporal basis and types were designated on the basis of one or more band differences between strains.

Results and discussion

PFGE profiles for the external quality assessment of the set of S. enterica demonstrated between 12 and 20 resolvable chromosome fragments, ranging from approximately 20 kb to 1 140 kb. By using a harmonised method for PFGE with defined parameters for electrophoresis, the gel images produced were comparable between each centre despite slight variations in DNA preparation (table 1). In most cases, there was at least 90% similarity between isolates tested in the different European laboratories and there was usually >95% similarity. Where there were differences in banding patterns, this was due to the absence/faintness of the smallest bands on the gel, with molecular masses of <30 kb (figure 1).
Reference laboratories participating in this project are currently in the process of selecting and testing a further 500-1 000 S. enterica isolates, representing currently defined serotypes of epidemiological importance within their country. Electronic recording and transmission of data between laboratories means that these isolates do not need to be exchanged. The PFGE patterns sent as TIFF files are being analysed using BioNumerics software at the coordinating laboratory where the profiles are stored in a central database and compared to a library of such patterns. Each new pattern is given a unique designation and added to the library of PFGE profiles. This designation is in the form of a six letter code together with a four digit numerical identifier. For example, the first pattern for S. Enteritidis digested with the enzyme XbaI is SENTXB.0001.

Table 1
Similarity rate of the PFGE profile for the 16 S. Enteritidis strains analysed by nine European laboratoires

PFGE produces reproducible, stable fingerprints with well resolved fragments that represent the entire genome. It is this stability that is crucial for PFGE as a strain typing method. We use a harmonised PFGE protocol that takes into account some of the technical variation between different European centres. While standardisation of DNA preparation and digestion were not considered to be essential, standardisation of the parameters for electrophoresis was considered to be an absolute requirement. During the development of the Salm-gene project, there has also been consultation with PulseNet in the US and Canada to ensure comparability of data between Europe and North America.
The initial results from the EQA data show that it is possible to reproduce results at different centres and transfer this information electronically to a central database. The EQA scheme takes place every six months to ensure integrity of the results obtained with the standardised procedures.

Nomenclature of profiles in this scheme is for ease of communication between laboratories. While not intended as a 'bacterial classification' system, it is important to establish universally recognised profile numbers for each unique pattern with single band differences being considered potentially significant. It is intended that this will serve as the basis for exchange of information between laboratories. However, the reporting of single band differences and the identity of PFGE profiles alone does not prove unequivocally whether isolates are related or not. Such information should be considered together with epidemiological evidence. The practical application will be that the organisms responsible for food related outbreaks of salmonellosis in different countries in the EU can be compared on the basis of their DNA fingerprints, together with other subtyping data and epidemiological information, thereby providing a sound basis for the introduction of appropriate intervention strategies.

We are currently creating an online, web-based searchable database of information for the most prevalent salmonella serotypes and phage types within Europe. Participants submit all DNA fingerprints and associated epidemiological information electronically to the coordinating hub where it is incorporated into the central web-based database. This international database will be developed, managed and maintained at the coordinating hub and will be accessible to all participants via the internet. The epidemiological database will be routinely analysed and results reported back to all participants. Public domain outputs and reports will be developed and published by the epidemiological coordinating centre for the project.

Recommendations will be developed for incorporating DNA fingerprinting into national laboratory based surveillance of human salmonella isolates, based on the cost effectiveness of different laboratory methods, sampling criteria, and the incidence of particular phage types.

Acknowledgements

The Salm-gene project is funded by Directorate General RESEARCH of the European Commission under the Framework Programme 5 (Contract QLK2-CT-2001-1940).

* Participants in Salm-gene are:
The Laboratory of Enteric Pathogens, England and Wales, the Bakteriologisch-serologische Untersuchungsanstalt, Austria; Statens Seruminstitut, Denmark; National Public Health Institute, Finland; Robert Koch-Institut, Germany; Instituto Superiore di Sanita, Italy; National Institute of Public Health & the Environment, the Netherlands; Scottish Salmonella Reference Laboratory, Scotland and Instituto de Salud Carlos III, Spain with the reference laboratory at the Pasteur Institute, France acting as the software compatibility advisor, and the Communicable Disease Surveillance Centre, England & Wales being the epidemiological co-ordinating centre.