INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Meningococcal disease remains an important public health problem in the United States and worldwide. Multilocus enzyme electrophoresis (MEE) is the "gold standard" method for Neisseria meningitidis subtyping, allowing the identification of transcontinental clonal complexes that have an increased propensity to cause epidemic disease (1, 3, 4, 12, 14, 15, 20). Although it has been successfully applied in many epidemiologic investigations, MEE is time-consuming, expensive, and subject to difficulties in data analysis and interlaboratory comparison. As a result, a system that uses multilocus sequence typing (MLST) of housekeeping genes has recently been developed for the characterization of N. meningitidis (13). One of the biggest advantages of MLST over MEE is the electronic portability of the nucleotide sequence data, enabling rapid global exchange of molecular typing information for epidemiologic comparisons. While the technology for nucleotide sequencing has substantially improved in the last few years, sequencing of multiple housekeeping genes remains time-consuming, even when carried out with an automated sequencer. Cleavase fragment length polymorphism (CFLP) is a subtyping system based on the single-stranded DNA patterns resulting from digestion with the enzyme cleavase, a structure-specific, thermostable nuclease (2). This enzyme recognizes and cleaves secondary structures that consist of double-stranded hairpin regions interspersed with single-stranded regions of DNA and that are formed after denaturation and cooling to an intermediate temperature, in a pattern unique to the nucleotide sequence. The objectives of this study were to evaluate CFLP as a means for rapidly identifying N. meningitidis isolates of two major epidemic-associated electrophoretic type (ET) complexes and to assess its possible usefulness as an initial screening survey for genes that might be used for DNA sequence-based subtyping.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

N. meningitidis strains. Thirty N. meningitidis serogroup B and 50 serogroup C isolates were assayed by CFLP and MEE. The serogroup B isolates were 19 serotyping and serosubtyping reference strains (9, 10) and 11 sporadic-case isolates obtained through a population-based surveillance system for N. meningitidis that is part of a multistate population-based surveillance project coordinated by the Centers for Disease Control and Prevention (CDC) as part of the Active Bacterial Core Surveillance/Emerging Infections Program Network; 4 of these 11 isolates belonged to the ET-5 complex, and the remaining 7 were only distantly related to this complex. Seventeen of the serogroup C isolates were chosen to represent four epidemiologically defined outbreaks that occurred in California in 1993 (11, 18), New Mexico in 1994 Arizona in 1994 (11), and Georgia in 1998. Strains isolated from a series of cases that occurred in Massachusetts in 1998 were also analyzed. Four or five sporadic-case isolates were selected as controls for each outbreak by use of the available isolate with the closest temporal and geographic proximity. A total of six serogroup C isolates of six ETs closely related to the ETs of the outbreak-associated isolates and therefore within the ET-37 complex (4, 19) and five isolates of ETs only distantly related to this complex were also included. All strains were isolated from normally sterile body sites, such as blood or cerebrospinal fluid, except for the reference strain for serosubtyping, for which information was not available. Serogroup B and serogroup C strains and their selection criteria are listed in Tables 1 and 2, respectively.

CFLP analysis. The bacteria were incubated on blood agar plates with 5% sheep blood (BBL Microbiology Systems, Cockeysville, Md.) overnight at 35°C in a 5% CO₂ atmosphere. Whole-cell suspensions in 0.01 M Tris buffer (pH 8.0) were boiled for 10 min and used as a template for PCR amplification. PCR products were derived from five target genes: adenylate kinase (adk), aspartate transaminase (aspC), carbamoylphosphate synthetase (carA), dihydropteroate synthase (dhp), and glutamine synthetase (glnA). The criteria for gene selection included enzymes that are used in MEE and that have a high degree of diversity and availability of sequences in GenBank. The strands were labeled with 6-carboxyfluorescein (FAM) or 4,7,2',7'-tetrachloro-6-carboxyfluorescein (TET) fluorescence dye (Glen Research, Sterling, Va.) and purified with a Qiaquick 8 PCR Purification Kit (Qiagen Inc., Valencia, Calif.). Approximately 100 fmol of the labeled DNA substrate was denatured at 95°C for 2 min, cooled to a suitable collapse temperature, and digested with 25 U of the structure-specific cleavase I enzyme provided in the CFLP Evaluation Kit (Third Wave Inc., Madison, Wis.). The manufacturer's instructions were followed, except for the use of 95% formamide instead of the stop solution to terminate the reaction because of the background caused by the dye present in that solution. The cleavase reaction optimization of time and temperature was performed at temperatures ranging from 50 to 65°C and times ranging from 2 to 6 min as previously described (2). The labeled primer sequences, annealing temperatures for PCR amplification, DNA product sizes, and CFLP digestion conditions are described in Table 3.

MEE analysis. All strains were subtyped by MEE; 24 enzymes were used with methodology described previously (16). In order to compare CFLP to MEE, the genetic relatedness of the serogroup B strains was analyzed by use of the respective dendrograms generated with PHYLIP software (8).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

CFLP analysis of five metabolic genes in serogroup B strains. Among the 30 serogroup B isolates, we found 13 CFLP patterns or types for glnA, 10 for dhp, 9 for carA, 6 for aspC, and 3 for adk. By MEE, 25 distinct ETs were identified. More variability for the adk gene was identified by CFLP than by MEE, since three types were observed by CFLP and no diversity was observed for this locus by MEE. Similarly, nine CFLP types and five MEE types were observed for carA. For aspC, 6 CFLP types and 10 MEE types were observed with all combinations of the two MEE bands. The genetic relatedness of the strains analyzed by CFLP and MEE is illustrated by the dendrograms shown in Fig. 1. The majority of strains were grouped into similar clusters by both CFLP and MEE analyses. The nine ET-5 complex strains showed almost identical clustering patterns in both methods. Relationships of some strains were not the same in CFLP and MEE analyses. For example, the serological reference strains B16B6 and 2996 were found closely related by MEE but were found different by CFLP.

View larger version (37K): [in this window] [in a new window]   FIG. 1.   Dendrograms generated from CFLP and MEE subtyping of the 30 N. meningitidis serogroup B isolates. The group of nine closely related strains of the ET-5 complex is clustered together by both methods (box). The bar indicates genetic distance for both CFLP and MEE.

CFLP analysis of single metabolic genes in serogroup C strains. With a single exception (strain M1532), all of the outbreak-related strains from California, New Mexico, Arizona, and Georgia and three strains from a cluster in Massachusetts showed the same glnA and carA CFLP patterns. Strain M1532, which was isolated in New Mexico and which displayed a unique pattern for both genes, was characterized as ET-147 (which differs from ET-24 in 17 of 24 enzymes), while the ETs of all other outbreak-related strains were members of the ET-37 complex (Table 2). Only 1 of 19 sporadic-case isolates (M239) was identified to be of an ET not closely related to the ET-37 complex (Table 2), and this isolate could be differentiated from the outbreak-associated strains by both glnA and carA. Two additional sporadic-case isolates (M393 and M4594) could be differentiated from the outbreak-associated strains by glnA. Isolate M393 also had a distinct carA pattern. Figure 2 shows glnA and carA CFLP patterns of an outbreak-related strain (M140) and a sporadic-case isolate (M239) from Los Angeles County, Calif. Analysis of N. meningitidis serogroup C isolates revealed that the glnA gene was more diverse than the carA gene, as was observed for the serogroup B isolates. No differences were observed for the carA gene among the six isolates of ETs closely related to the ET-37 complex (Table 2). Only two of those isolates (M2964 and M3733) showed minor differences in glnA patterns. In contrast, the five isolates of ETs distantly related to the ET-37 complex (Table 2) exhibited unique patterns for both genes, confirming the screening capacity of CFLP with a single metabolic gene in differentiating ET-37 complex strains from strains of other ETs.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Electropherograms generated by the 672 GeneScan software showing the CFLP patterns of serogroup C isolates M140 (an outbreak-associated ET-24 isolate) and M239 (a sporadic-case isolate not related to the ET-37 complex) from California. The two upper electropherograms are cleaved single-stranded PCR products derived from the glnA gene, and the two lower electropherograms were derived from the carA gene. The x axis shows the scan number, and the y axis shows relative fluorescence.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, CFLP and MEE were compared to investigate the capacity of CFLP to discriminate meningococci belonging to the ET-5 complex, which have been responsible for many outbreaks in Europe, South Africa, Central America, and South America (3, 4, 15) and, more recently the United States (5, 14, 20), from other serogroup B strains. All strains of the ET-5 complex were clearly distinguishable from other serogroup B strains by both CFLP and MEE. CFLP detected more variation than MEE for some genes, resulting in more alleles per locus, similar to the results obtained by MLST analysis (13). The number of CFLP patterns obtained generally correlated with the relative amount of DNA sequence diversity observed for the genes used. For instance, meningococcal adk gene sequences may differ at ~1% of nucleotide sites (7); conversely, the glnA gene has been found to be unusually variable, with diversity observed at 4.4% of nucleotide sites (17, 21). The congruence between CFLP and MEE was excellent for most of the strains tested, but some differences were found, as also reported by Maiden et al. (13) when comparing MLST and MEE. The reasons for these differences are not known, nor are the actual relationships of the strains. Because the two methods examine different properties of strains, it is possible that both typing schemes are correct. More importantly, the sensitivity of the CFLP method for differentiating the ET-5 complex strains is equivalent to that of the standard MEE method.

The usefulness of CFLP in rapidly screening serogroup C isolates of the ET-37 complex (1, 4, 11, 12, 18, 19) was also evaluated. CFLP analysis of a single metabolic gene could consistently differentiate strains belonging to the ET-37 complex from serogroup C strains unrelated to this complex. However, it was of particular epidemiologic significance to determine if ET-17 and ET-24 could be differentiated from other ETs of this complex and from each other, as these ETs are the most frequently identified ETs of the ET-37 complex and are also most frequently associated with serogroup C meningococcal disease outbreaks in the United States. Although ET-17 and ET-24 differ from each other for only 3 of 24 enzymes in MEE, the outbreaks that they cause have always been epidemiologically distinct. No outbreaks in which both of these ETs were simultaneously detected have been reported. The CFLP patterns of strains of ET-17 and ET-24 were indistinguishable by use of both glnA and carA genes, suggesting that perhaps the selection and evaluation of another gene(s) may be needed in this particular situation; possible genes are those encoding the peptidase, phosphoenolpyruvate carboxykinase, or fumarase enzymes that differentiate ET-17 and ET-24 in MEE. A weakness of MEE analysis results from the fact that ET-17 and ET-24 are also frequently identified in strains isolated from sporadic cases through population-based epidemiologic surveillance; therefore, it is impossible to distinguish outbreak-associated and sporadic-case isolates of these ETs. This is also generally true for strains of other ETs of this complex, as specific ETs are identified in both outbreak-associated and sporadic-case isolates. It is not completely unexpected for CFLP based on a single gene not to be able to differentiate outbreak-associated from sporadic-case isolates of the ET-37 complex, especially because of the large number of loci examined by MEE (24 in our study). Further CFLP evaluation of genes used in the MEE panel may result in the identification of a gene(s) capable of providing this differentiation.

CFLP has the potential to be widely used for initial screening because, unlike MEE, CFLP can easily and rapidly screen a large number of strains. With the model 377 automated sequencer, 96 strains can be assayed for a single gene within 48 h. Additionally, CFLP analysis is a DNA-based subtyping method with the capacity for direct assignment of alleles based on the nucleotide sequences of genes. CFLP can be applied to the rapid screening of a large number of strains during investigations of outbreaks and/or surveillance systems. CFLP is as efficient as and more rapid than MEE in identifying strains of the ET-5 and ET-37 complexes. Because of the sensitivity in detecting nucleotide changes (2), CFLP based on a single metabolic gene with a high degree of diversity but not under selective pressure may be used to generate an appropriate level of discrimination among isolates. As clearly demonstrated in this study, CFLP not only can rapidly screen for the major epidemic clones of serogroups B and C, e.g., clonal complexes ET-5 and ET-37 but also, more importantly, can provide guidance regarding the choice of genes for the DNA sequence-based approach to the molecular subtyping of N. meningitidis. The development of such a subtyping system with a level of discrimination comparable to or surpassing that achieved by the current gold standard, MEE, will enable rapid global exchange of data and comparison of molecular subtyping information for both research and epidemiologic purposes. The use of CFLP as a screening tool to develop a DNA sequence-based subtyping method will greatly simplify the creation of a reliable and reproducible database for N. meningitidis or other microorganisms.