OUP user menu

Multilocus sequence typing for global surveillance of meningococcal disease

Carina Brehony, Keith A. Jolley, Martin C.J. Maiden
DOI: http://dx.doi.org/10.1111/j.1574-6976.2006.00056.x 15-26 First published online: 1 January 2007


The global surveillance of bacterial pathogens is particularly important for bacteria with diverse and dynamic populations that cause periodic epidemics or pandemics. The isolate characterization methods employed for surveillance should: (1) generate unambiguous data; (2) be readily implemented in a variety of scenarios and be reproducible among laboratories; (3) be scalable and preferably available in a high throughput format; and (4) be cost effective. Multilocus sequence typing (MLST) was designed to meet these criteria and has been implemented effectively for a wide range of microorganisms. The ‘Impact of meningococcal epidemiology and population biology on public health in Europe (EU-MenNet)’ project had amongst its objectives: (1) to disseminate meningococcal MLST and sequence-based typing throughout Europe by establishing a centre for training and data generation, and (2) to produce a comprehensive Europe-wide picture of meningococcal disease epidemiology for the first time. Data produced from the project have shown the distribution of a relatively small number of STs, clonal complexes and PorA types that account for a large proportion of the disease-associated isolates in Europe. The project demonstrates how molecular typing can be combined with epidemiological data via the Internet for global disease surveillance.

  • Neisseria meningitidis
  • epidemiology
  • population biology
  • isolate characterization
  • internet


The surveillance of pathogenic bacteria is an essential element of national and international public health strategies for the control of infectious disease. Trans-national epidemiological and population studies of Neisseria meningitidis, the meningococcus, have been particularly important in understanding meningitis and septicaemia caused by this globally distributed bacterium. Despite having a reputation as a fearsome pathogen, the meningococcus is frequently carried harmlessly in the human nasopharynx in c. 10% of the population and can be considered to be part of the normal commensal flora (Broome, 1986; Cartwright, 1995; Maiden, 2004).

Comparisons of meningococci isolated from asymptomatic carriage and disease have demonstrated that meningococcal populations are genetically and antigenically highly diverse (Caugant et al., 1987a, 1998), but that a minority of genotypes and antigenic types, the ‘hyperinvasive lineages’, account for the majority of disease (Caugant et al., 1988; Urwin et al., 2004; Yazdankhah et al., 2004). Detailed and accurate isolate characterization is therefore a particularly important element in studies of meningococcal epidemiology, population biology and evolution.

The first isolate characterization methods to gain broad acceptance for meningococci were immunological, employing polyclonal sera or monoclonal antibodies (Frasch et al., 1985). Much valuable information has been obtained from the use of these methods, and serogrouping on the basis of the polysaccharide capsule remains the most important characteristic in routine laboratory use (Vedros, 1987). For many studies however, serological methods exhibit a number of limitations, including an incomplete coverage of the antibody panels employed (Feavers et al., 1996; Sacchi et al., 1998); difficulties in accommodating antigenic variants (Feavers et al., 1996); failure to recognize epitopes due to the variable expression caused by phase variation (van der Ende, 1995); and inconsistent correspondence with genetic relationships (Maiden & Feavers, 1994).

In an attempt to provide an alternative typing paradigm, many methods based on DNA technology have been proposed, including: ribotyping (Woods et al., 1992), random amplified polymorphic DNA (RAPD) (Woods et al., 1994; Bart et al., 1998; Schmink et al., 2001); fluorescent amplified-fragment length polymorphism (AFLP) (Goulding et al., 2000), restriction fragment length polymorphism (RFLP) (Campos et al., 1992), multiple-locus variable-number tandem repeat analysis (MLVA) (Schouls et al., 2006), and pulsed-field gel electrophoresis (PFGE). PFGE, in particular, has been widely employed in the surveillance of diseases caused by enteric bacteria in the PulseNet network (Swaminathan et al., 2001, 2006; Ribot et al., 2006), however it is not reliable in its identification of hyperinvasive meningococci and comparison among laboratories is difficult. As a result, the PulseNet System has not been extended to meningococci. Although these methods can be discriminatory and effective in recognizing the meningococci responsible for localized outbreaks, they do not reliably indicate the long-term relationships of isolates and are generally poorly portable among laboratories. Consequently they are rarely employed beyond the laboratory that developed them (Achtman, 1996).

A further complication in the characterization of meningococci is their high rates of recombination (Feavers et al., 1992; Spratt et al., 1992), which have led to an essentially nonclonal population structure (Holmes et al., 1999) without a simple bifurcating tree-like phylogeny (Maiden, 1993; Feil et al., 1995, 1999; Holmes et al., 1999; Jolley et al., 2000). For such bacteria, accurate isolate characterization depends on an analysis of multiple loci distributed around the genome. Relatively high recombination rates exacerbate the problems of typing methods that index the variation in genes under positive or diversifying selection, such as those encoding antigens or antibiotic resistance determinants. Whilst such genes provide high levels of discrimination, for long-term and global surveillance it is preferable to examine variation at loci that are evolving more slowly, for example housekeeping genes that encode proteins forming part of central metabolism and which are under stabilizing selection for conservation of their metabolic function (Urwin & Maiden, 2003).

The first technique to exploit the multilocus approach was multilocus enzyme electrophoresis (MLEE) (Selander et al., 1986). MLEE analyses the variation in metabolic enzymes by measuring the electrophoretic mobility of the proteins on a starch gel, followed by detection using specific staining methods. In N. meningitidis, early MLEE studies were central to the discovery of groupings (clonal complexes) within meningococcal populations, and the identification that some of these – the hyperinvasive lineages – were particularly associated with disease. These studies also enabled a large-scale analysis of the global clonal spread of hyperinvasive meningococci (Caugant et al., 1986; Olyhoek et al., 1987; Moore et al., 1989; Achtman, 1990; Caugant, 1998). MLEE is, however, technically demanding, and it is difficult to compare the results obtained from different laboratories. Furthermore, the technique indirectly measures genetic variation and is relatively low-resolution, requiring large numbers of loci (up to 20) to be examined.

Multilocus sequence typing of meningococci

Multilocus sequence typing (MLST) was conceived as a generic typing method that fulfilled the criteria of reproducibility, reliability, cost and throughput. The approach was first developed and implemented for the meningococcus (Maiden et al., 1998). MLST was based upon the principles of MLEE, but also exploited high-throughput nucleotide sequencing and data dissemination via the Internet (Urwin & Maiden, 2003).

MLST indexes the variation present in nucleotide sequences of 400–500 bp internal fragments from housekeeping genes. This fragment size can be efficiently sequenced on both strands using a single primer extension reaction in each direction whilst providing sufficient sequence diversity for discrimination (Urwin & Maiden, 2003). Nucleotide sequence data are unambiguous and can be easily transferred and compared among laboratories, and the techniques required are both generic and highly reproducible. The high resolution of nucleotide sequencing means that fewer loci are needed compared to MLEE, and synonymous and nonsynonymous base changes are differentiated. In developing the Neisseria MLST scheme, 11 loci were examined, but it was found that six, later expanded to seven loci (Holmes et al., 1999), gave sufficient resolution to identify meningococcal clonal complexes.

MLST data are accessible via the Internet (http://pubmlst.org/neisseria/) (Maiden et al., 1998; Chan et al., 2001; Jolley et al., 2004). Following interrogation of the MLST profile database, each sequence is assigned an allele number based upon its sequence. For a full MLST scheme with seven loci this results in a sequence type (ST) comprising seven integers, one for each MLST locus. Each unique ST is assigned an arbitrary number which is equivalent to the electrophoretic type (ET) employed by MLEE. There is sufficient nucleotide sequence variation at each locus to provide many alleles and a very large number of allelic profiles. Despite this variation, MLST identifies clonal complexes that correspond to the hyperinvasive lineages (Caugant, 1998; Maiden et al., 1998; Yazdankhah et al., 2004).

A clonal complex is defined in the Neisseria MLST profile database as a group of STs that share at least four of the seven loci in common with a central ST. The central ST is the putative ‘ancestral genotype’ which the other genotypes in the complex are descended from and that the complex is named after, e.g. the ‘ST-11 clonal complex’ (Urwin & Maiden, 2003). In one case, that of the ST-41/44 complex, it has been necessary to define two central STs as a consequence of its size and diversity. Clonal complexes and their central STs have been identified by a combination of clustering methods such as split decomposition (Bandelt & Dress, 1992; Huson, 1998) or eBURST (Feil et al., 2004), as well as epidemiological data on the distribution of the STs in space and time; central STs are generally both common and persistent. The status of a particular clonal complex is confirmed by a review of the data by an international management group.

Greater resolution can be obtained, if necessary, by combining the MLST data with sequence data from loci that are under diversifying selection, e.g. the outer-membrane proteins PorA, PorB, and FetA (Bygraves et al., 1999; Feavers et al., 1999; Harrison et al., 2006). This approach can add discrimination to the typing scheme cost-effectively, and can be helpful in defining outbreaks. It is also informative to assess the distribution of antigenic markers that are potential vaccine candidates (Urwin et al., 2004).

The European Meningococcal MLST centre

One of the major advantages of MLST is its inherent portability, not only of data, which being electronic is easily shared via the Internet, but also in terms of the samples required. This portability can be exploited to overcome the problem that not all laboratories may have access to cost-effective high-throughput nucleotide sequencing facilities.

The European Meningococcal MLST Centre (EMMC) was established as part of the ‘Impact of meningococcal epidemiology and population biology on public health in Europe (EU-MenNet)’ project. The goal of the EU-MenNet consortium was to co-ordinate and integrate epidemiological and population genetic studies across the participating countries in Europe in order to provide a better understanding of the spread of hyperinvasive lineages in the region. The objectives of the EMMC were: (i) to disseminate meningococcal MLST and sequence-based typing throughout Europe by establishing a centre for training and data generation and dissemination, and (ii) to produce, for the first time, a comprehensive survey of meningococcal disease epidemiology in Europe. The EMMC is a paradigm for how integrated transnational surveillance of bacterial pathogens can be achieved in practice.

The EMMC was established with the specific aim of producing complete MLST profiles for at least 1000 meningococci isolated from invasive disease for each of the 3 years of the EU-MenNet project. These isolates were to be representative of those that had been submitted to national reference laboratories across Europe during this period. In addition, the EMMC was to disseminate MLST technology by providing resources, including cost-effective access to the latest automated DNA analysers, and the provision of protocols and training in the form of workshops. Nucleotide sequence determination of the gene fragment encoding the subtype antigen PorA variable regions VR1 and VR2 was also undertaken to add further discrimination and to assess the levels of diversity of this putative vaccine component both geographically and temporally.

At the simplest level a laboratory could submit a specimen (DNA or a sacrificed cell suspension) to the EMMC (Table 1), which would perform the isolate characterization and make the data available to the submitting laboratory via the Internet. For those laboratories with MLST and other sequence-based typing in place in high-throughput format, compiled MLST data could be submitted to the EMMC. All intermediate levels of participation were possible; the most common being the sending of completed sequence extension reactions to the EMMC for separation on the DNA analysers and the return of the sequence chromatograms for assembly.

View this table:
Table 1

Number of isolates for which MLST data was compiled by the EMMC by country and year

Czech Rep.50 (4)(67)36 (35)192Boilate (Data)
E & W(77)(150)22 (76)325Boilate (Data)
France(93)19 (103)(125)340Boilate (Data)
Netherlands176167 (72)204619DNA (Data)
Spain14593 (27)143 (4)412Boilate (Data)

The EMMC approach allowed a rapid dissemination of the MLST technology and comprehensive participation. The collated data were then linked to the epidemiological records held by the European Meningococcal Epidemiology Centre (EMEC) (Trotter et al., 2006) and made available on the EU-MenNet website.

Sample choice, preparation and submission

For each of three years 2000, 2001 and 2002 a representative collection of meningococcal specimens was assembled on the basis of an algorithm devised by the EU-MenNet management committee. For national reference laboratories receiving 80 meningococcal disease isolates per year or fewer, all isolates were submitted to the EMMC. For laboratories receiving more than 80 isolates, every third isolate received was submitted, with the exception of the England & Wales Meningococcal Reference Unit, which received more than a thousand annual cases and submitted every tenth isolate.

To ensure maximum participation it was necessary to employ a simple, robust and inexpensive method of sample preparation. Since MLST utilizes the PCR to amplify the loci to be sequenced, it is not necessary for the sample DNA to be free from other cellular components. PCR amplification can work well with DNA of varying quality ranging from that present in crude lysed cell suspensions or clinical samples up to high-quality DNA samples purified by expensive and/or time-consuming protocols. Satisfactory results can be obtained from samples prepared by simply boiling cells suspended in phosphate buffered saline. These samples are stable at room temperature and, as they comprise sacrificed cell suspensions, can be submitted by conventional post without the expense of refrigeration or bio-security measures.

Data generation

Nucleotide sequencing is a generic technique that is available in many biological laboratories world-wide. Consequently, the equipment and training necessary to perform MLST is readily accessible. The latest generation of automated capillary DNA analysers offer robust performance and have eliminated the problems associated with older slab gel-based DNA analysers. Perhaps the greatest advantage of the use of capillaries is in the reduction of manpower and consumables costs. It is not necessary for every laboratory to have its own DNA analyser, as sequencing products can easily be analysed remotely and the raw or compiled data returned via the Internet.

A further advantage of PCR-based MLST is that it can be used to characterize the meningococci present in specimens from which no microbiological culture can be obtained. In these cases, however, greater care is required to avoid cross-contamination since the template copy number will be lower than from samples obtained from culture. Contaminant DNA in the sample preparation area may be preferentially amplified, so physical separation of the pre and post amplification areas is important. If the lack of laboratory space does not permit such separation, the use of small bench-top PCR cabinets that fully enclose the preparation area and from which contaminating DNA can be removed by UV light prior to use, can be employed. Protocols have been developed to include a two-step PCR amplification with enhanced sensitivity to detect the low numbers of genome copies present in some clinical samples (Kriz et al., 2002; Diggle et al., 2003; Birtles et al., 2005).

Where samples are prepared by external laboratories and submitted for nucleotide sequencing, consideration needs to be given as to the template concentration used in the sequencing reactions, as it may be difficult to standardize the samples. The latest generation of DNA analysers are very sensitive compared to earlier instruments and consequently require considerably less labelled terminator to produce a signal, markedly reducing the costs of sequencing. Reducing the concentration of labelled terminator, however, has the side effect of also making the window of acceptable DNA concentrations smaller. The experience of the EMMC found that using 1: 16 scale reactions (i.e., using 0.5 μL of BigDye dye terminator mix per reaction, compared to the 8 μL initially recommended by the manufacturer) worked robustly with the variety of samples submitted without the need to quantitate each reaction when run on a 3730 or 3100 DNA analyser. With careful on-site control of sample preparation, however, it is possible to routinely reduce dye terminator use to a 1:32 scale.

Bioinformatics and data handling

Sequence assembly

With current improvements in nucleotide sequencing technology, the bottlenecks in high-throughput sequence determination are most often located in the downstream processing of the unassembled sequence data. Automated assembly and allele assignment of sequences from multiple isolates is possible using tools such as stars (http://www.cbrg.ox.ac.uk/~mchan/stars/) or Ridom TraceEdit Pro (http://www.ridom.de/traceeditpro/). Using these tools with the good quality data produced by the latest DNA analysers, hundreds of forward and reverse sequencing reads can be assembled into double-stranded allele sequences and automatically assigned MLST allele designations in a few minutes. Apart from the obvious benefits compared to manual assembly, of increased throughput, the use of automated tools ensures consistency, with an ability to set thresholds on sequence quality that may be important for quality assurance purposes.

Linux distributions tailored for bioinformatics use, such as Bio-Linux (Field et al., 2005) (http://envgen.nox.ac.uk/biolinux.html), enable laboratories to easily and cheaply set up software to facilitate high-throughput sequence assembly. The stars software comes preinstalled on Bio-Linux, allowing rapid sequence clipping and allele assignment. The EMMC also provided access to software, training courses, and a server available from the Internet, ensuring that all participants had access to the latest software.

MLST databases

Large-scale MLST projects require streamlining of the data flow, with particular attention to the avoidance of manual transcription of allele numbers, as this can introduce human-generated assignment errors. With the introduction of automated assembly, it is possible to store the assigned allele numbers in a database with minimal user interaction. This can either be achieved by interfacing the assembly software directly with an isolate database, or more simply by outputting the assembly results in a standard text format that can be readily parsed by a simple script for use with local or other systems.

Databases designed for storing MLST data fall into two types – those for profiles and allele sequences and those for isolate data. Separation of the two offers many advantages. The use of a definitive profile/sequence database to which all new assignments are made ensures data integrity and avoids duplication. The Neisseria MLST profile database is hosted at http://pubmlst.org/neisseria/(Jolley et al., 2004), where all submissions are entered via a curator who checks new allele sequence traces for accuracy. Isolate databases can then behave as clients of the profile database to look up sequence types or profiles, without the need to submit isolate data to a central authority.

A single profile database potentially presents a single point of failure if multiple isolate databases rely on it for their assignments. As profile databases are very straightforward to implement and as the data stored within it can be represented as simple text, it is very easy to mirror or establish a local copy. Provided updates are made to only one profile database that remains definitive, any number of synchronized mirror sites can be set up so that high availability is achieved. The Neisseria MLST profile database is mirrored at different geographical servers in the UK, New Zealand and the USA, providing a robust service for the research community.

The largest public isolate database is Neisseria PubMLST (http://pubmlst.org/neisseria/), which contains at least one isolate representing every known ST and to which data can be submitted by anyone. At the time of writing, the database had over 7500 submitted isolates from a total of 77 countries. However, as anyone can submit to the PubMLST isolate database, and many submit only to obtain new ST or allele designations, this database is a nonpopulation based collection of isolates that describes the extent of meningococcal diversity. It is not necessarily a reliable indication of distribution or of the frequency of different genotypes.

For surveillance purposes, more focused and specialized isolate databases are required. Isolate databases can have their own access restrictions and be customized to fulfil the needs of an individual project or laboratory, but still benefit from the single definitive ‘dictionary’ of allele, ST, and clonal complex definitions. Software for setting up distributed MLST databases on Linux systems is freely available (Chan et al., 2001; Jolley et al., 2004) although other systems could be used since the ST definitions are published in a simple text format on the PubMLST site that can be readily parsed by automated software tools.

The EMMC database

At the outset of the EU-MenNet project it was decided that the MLST data should be held in a web-accessible database that permitted participating reference laboratories to access the status of all the isolates they had submitted in real time. All the reference laboratories agreed that access restrictions would be created such that individual laboratories could only view detailed data from their own country until such a time that the full data set was published. The EMMC isolate database was established as a client of the PubMLST profile database using the mlstdbnet software (Jolley et al., 2004) on a Linux system. This arrangement ensured that the software assigned the ST and clonal complex designations automatically as the allele data were collated (Fig. 1). As the mlstdbnet software is published freely as open-source software, updates that allowed restricted and variable access, graphical breakdown statistics and rapid data upload interfaces which were developed for the needs of the EU-MenNet project were then made freely available for use by others.

Figure 1

Interactions and structure of the MLST and epidemiology databases hosted at the EMMC and EMEC. The MLST isolate database retrieves ST and clonal complex definitions from the MLST profiles database hosted at http://pubmlst.org/neisseria/. The EMEC database server synchronizes a local copy of the isolate MLST database nightly so that the epidemiological information can be linked.

Epidemiological data for the isolates processed by the EMMC were sent to the EMEC by the participating countries. This information was stored in a database system that was compatible with the mlstdbnet software used to store the MLST profiles. Since the EMMC and EMEC are separate entities, it was not appropriate that the EMMC stored the epidemiological information as it was tasked only to generate the MLST profiles for the isolates. It was, however, necessary for the MLST data to be merged with the epidemiological data at the EMEC. To achieve this, a mirroring system was put in place so that the two database servers could connect over an encrypted link to upload a copy of the MLST database to the EMEC server every night. This was then linked locally to the epidemiological database. The only requirement within the data for the linking to be achieved was that a common identifier would be used when submitting isolates or data to the EMMC and the EMEC. The common identifier consisted of a two letter country code followed by the isolate number, in whichever format was used locally, provided by the submitting laboratory.

Data analysis

Once a MLST profile has been completed for an isolate, the next stage is clonal complex assignment. Where an isolate database is a client of the PubMLST central profiles database, this assignment will be automatic. An advantage of using MLST over many other typing methods is that the nucleotide sequence data produced can also be used for more complex phylogenetic analysis. Using powerful coalescent methods, phylogenetic parameters such as the relative rates of recombination and mutation, and the mean recombination fragment size can be estimated (McVean et al., 2002; Jolley et al., 2005). The data can also be used to determine the rates of gene flow between populations and genetic structuring (Excoffier et al., 1992; Schneider et al., 2000; Jolley et al., 2005). Elucidation of these phylogenetic parameters is informative in generating evolutionary models that can address clonal complex emergence and stability and the effects of vaccination.

Meningococcal clonal complex distribution in Europe 2000–2002

The genetic characterization of isolates reveals the spread of the major disease-associated clonal complexes. MLEE first identified clonal complexes using cluster analysis on ET profiles (Caugant et al., 1987a, b). Analysis of over 4000 European disease isolates from the 18 countries involved in the EU-MenNet project for the 3 years 2000–2002 (Table 1) revealed the predominance of hyperinvasive complexes (Fig. 2). While there was much diversity in the STs (∼1000 types), only ten accounted for half of the isolates. The STs resolved into 31 distinct clonal complexes, the most prevalent being the ST-41/44 complex (1014 isolates, 25%), ST-11 complex (901 isolates, 22%), ST-32 complex (706 isolates, 17%), ST-8 complex (273 isolates, 7%) and ST-269 complex (256 isolates, 6%).

Figure 2

Distribution of clonal complexes in Europe. Eighteen countries participated in the EU-MenNet project, and this figure shows the distribution of clonal complexes found overall in each country for the years 2000–2002 inclusive. Note: Isolates were not submitted for: Iceland 2000 and 2002; Portugal 2000 and 2002; or Sweden 2000.

These major disease-associated complexes have been found world-wide and over a number of years (Achtman, 1995; Muros-Le Rouzic et al., 2004). The ST-11 complex is the predominant complex accounting for most serogroup C disease as well as being associated with outbreaks of disease in Europe, Australia, the USA and Canada (Whalen et al., 1995; Jelfs & Munro, 2001). An increase in ST-11 serogroup C disease in the UK in the 1990s prompted the introduction of a Meningococcal C Conjugate (MCC) vaccine campaign (Ramsay et al., 1997; Miller et al., 2001). The complex has more recently also been associated with serogroup W-135 outbreaks in Africa and also with those returning from the Hajj pilgrimage (Aguilera et al., 2002; Mayer et al., 2002; Nicolas et al., 2005). Prior to the 1970s the ST-32 complex (which is mainly linked with serogroup B) was rarely associated with epidemic disease. Since then, however, it has spread world-wide to cause raised levels of disease in Europe, South Africa, South America and the USA (Caugant et al., 1986; Wedege et al., 1995; Diermayer et al., 1999). Apart from in European countries, the ST-41/44 complex has been found to cause disease in various countries such as the USA and in New Zealand where it has been responsible for an epidemic since 1991 (Martin et al., 1998; Whitney et al., 2006). The ST-8 complex is associated with serogroups B and C and has been known to have caused cases of disease world-wide since the 1970s (Caugant et al., 1987a). The ST-269 complex associated with serogroup B disease has recently emerged in Quebec, Canada (Law et al., 2006).

The geographical distribution of the clonal complexes found in Europe was broadly similar, although the prevalence of each was slightly different among countries. For example, in Greece the ST-162 complex (18%) showed a higher prevalence in comparison with other countries (Yazdankhah et al., 2004, 2005). In Scotland the ST-213 complex was present at a higher prevalence (10%) than in most other countries. The range of prevalence of the ST-8 complex varied from 0% (Czech Republic, Iceland, Sweden, Finland) to 39% (Portugal). In addition, the ST-23 complex, which has been associated with serogroup Y (Yazdankhah et al., 2004), was found at a much higher prevalence in Sweden (11%) and Finland (8%) than in most other countries. Overall there were no major changes in the distribution of types over the 3 years; however, there was a decrease in ST-8 complex from 9% to 5% over this period. At the individual country level, the ST-11 complex increased in some countries (Germany, Netherlands, France), while it declined in prevalence in others, particularly those that had implemented MCC vaccination programmes during this time (England and Wales, Scotland, Republic of Ireland, Belgium). In Spain there was a sizeable decrease in ST-8 complex from 27% to 7% over the 3 years.

Some complexes were more diverse than others in terms of the number of different STs observed (Yazdankhah et al., 2004). In the EU-MenNet study, the ST-11 complex was the least diverse with the central genotype accounting for 90% of isolates. The other major hyperinvasive complexes, i.e., ST-41/44, ST-32, ST-8 and ST-269, were much more diverse in comparison. As is discussed in detail in the accompanying paper by Trotter et al. (2006) there is a link between clonal complex and serogroup in the European isolates. For example, the ST-32, ST-41/44 and ST-269 complexes are associated with serogroup B, and the ST-11 and ST-8 complexes mainly associated with serogroup C. There also appears to be some association between genotype and the antigenic gene porA variable region (VR) types (Urwin et al., 2004; Devoy et al., 2005) which will be of particular interest in vaccine design. In the European isolates there were strong relationships between the ST-41/44 complex and PorA type 7-2,4 (52%), ST-32 complex with 7,16 and 19,15 (55%), ST-269 complex with 19-1,15-11 and 22,9 (58%), ST-11 complex with 5,2 and 5-1,10-8 (79%), and ST-8 complex with 5,2 (80%). There were 409 different PorA/clonal complex types with 10 accounting for 55% of the isolates. The main types being 7-2,4/ST-41/44 complex (14.8%), 5,2/ST-11 complex (10.2%), 5-1,10-8/ST-11 complex (6.6%), 7,16/ST-32 complex (6.2%) and 5,2/ST-8 complex (5.7%).

Analysis of MLST allelic profile and sequence data can help in assessing the degree of genetic structuring across the population and whether it can be divided into subpopulations corresponding to country. A means of measuring population subdivision is by use of the F-statistic (FST) (Wright, 1951). Wright's F-statistic measures the extent and/or presence of genetic variation and subdivision by a comparison of the alleles within subpopulations (e.g., countries) and the alleles within the total population. If there is a high amount of gene flow in the total population and no subdivision is apparent then the FST is zero. However, if there appears to be subdivision and structure in the population, then the FST is greater than zero. In the European disease isolates, calculations of FSTs for the allelic profiles, porA and concatenated locus sequences were carried out using the arlequin 2.0 program (Schneider et al., 2000). The results (Table 2) have shown a significant presence of structure across all countries except for between Norway and Sweden. This shows that while overall, countries have similar STs and clonal complex distributions, at the finer level of allele sequence comparison, we can find a degree of structuring and, as may be expected, there may be gene flow between similar and neighbouring countries (e.g., Norway and Sweden).

View this table:
Table 2

FST values* of pair-wise comparisons of concatenated MLST sequences of the 18 participating EUMenNet countries

AustriaBelgiumCzech Rep.DenmarkE & WFinlandFranceGermanyGreeceIcelandIrelandItalyNethlndsNorwayPortugalScotlandSpain
Czech Rep.0.033110.02229
E & W0.021980.010140.023340.07912
  • * FST measures the extent of structuring in a population by comparing the alleles within sub-populations (e.g. countries) and alleles within the total population. An FST value of 0 indicates no subdivision and therefore high gene flow in the population. A value greater than 0 indicates the presence of structure. Pair-wise comparisons of the concatenated MLST sequences from the 18 EU-MenNet countries were carried out using the arlequin 2.0 program. FST values in bold have P values>0.05 and therefore there is no significant subdivision between the two countries compared.

Concluding remarks

The EU-MenNet project has demonstrated how a highly cost-effective international infrastructure can be established that integrates microbiology and molecular typing with surveillance data, for the investigation of the epidemiology and population genetics of a bacterial pathogen. Coordination among European laboratories and the availability of the EMMC resources resulted in efficient data collection and technology transfer. The MLST results provided an overview of circulating types, both temporally and geographically and identified the major hyperinvasive lineages and their prevalence. This information is of great importance for public health strategies such as vaccination. As the current MCC vaccine is not comprehensive, the combination of MLST and antigen gene data will aid the design, development and implementation of novel vaccines.

MLST is particularly suitable for the long-term surveillance of meningococcal disease. It takes advantage of the latest advances in sequencing and bioinformatics and produces data that are highly reproducible between laboratories. The data generated can be deposited, accessed, and readily compared among laboratories, rapidly and inexpensively. The linkage of the MLST data with other databases, such as that established by the EMEC – which contains epidemiological information (serogroup, disease outcome, etc.) for invasive meningococcal disease cases in Europe – provides a wealth of additional information. Data collected by both the EMEC and EMMC will be integrated and further analysed, to be published in future papers. The EMEC is outlined in more detail in the accompanying paper by Trotter et al. (2006). Future possible developments include the addition of other data, such as sequences from the antigen-encoding FetA locus, as supplementary information. With the infrastructure and coordination established under the auspices of the EU-MenNet project it is possible for the surveillance of meningococcal disease across Europe to continue inexpensively, providing appropriate resources are made available. In principle, such surveillance could eventually be performed in real time, providing appreciable added value in terms of reactive public health intervention.


M.C.J.M. is a Wellcome Trust Senior Research Fellow in Basic Biomedical Sciences. The authors were funded by the European Union as part of the EU-MenNet project (QLK2-CT-2001-01436) and the Wellcome Trust.

We thank all the EU-MenNet participants/EMMC submitting centres: Statens Serum Institut, Copenhagen, Denmark; National Reference Centre for Meningococci, Graz, Austria; Dept. of Clinical Microbiology, Landspitali University Hospital, Reykjavik, Iceland; Istituto Superiore di Sanità, Rome, Italy; Department of Infectious Disease Epidemiology, National Public Health Institute, Helsinki, Finland; Antibiotic Resistance Unit, National Institute of Health Dr Ricardo Jorge, Lisbon, Portugal; Institut für Hygiene und Mikrobiologie, Würzburg, Germany; National Meningococcal Reference Laboratory, Scientific Institute of Public Health, Brussels, Belgium; Scottish Meningococcus and Pneumococcus Reference Laboratory, Glasgow, UK; National Reference Centre for Meningococci, Pasteur Institute, Paris, France; National Meningococcal Reference Laboratory, National School of Public Health, Athens, Greece; National Reference Laboratory for Meningogoccal Infections, National Institute of Public Health, Prague, Czech Republic; Meningococcal Reference Laboratory, Madrid, Spain; Reference Laboratory for Bacterial Meningitis, Academic Medical Centre, Department of Medical Microbiology, Amsterdam, Netherlands; Irish Meningococcal and Meningitis Reference Laboratory, Dublin, Ireland; Norwegian Institute of Public Health, Oslo, Norway; National Reference Laboratory for Pathogenic Neisseria, Dept. of Clinical Microbiology, Örebro University Hospital, Örebro, Sweden; Meningococcal Reference Unit, Manchester, UK.


  • Editor: Matthias Frosch


View Abstract