Background Neisseria meningitidis is a diverse commensal bacterium that occasionally causes severe invasive disease. The relationship between meningococcal genotype and capsular polysaccharide, the principal virulence factor and vaccine component, was investigated in carried meningococci isolated from 8000 children and young adults in Bavaria, Germany
MethodsOf the 830 meningococci isolated (carriage rate, 10.4%) by microbiological techniques, 822 were characterized by serogrouping, multilocus sequence typing, and genetic analysis of the capsule region. Statistical and population genetic analyses were applied to these data
ResultsThe rapid increase in carriage rates with age of carrier, the low prevalence of hyperinvasive meningococci, and the relative prevalence of the 4 disease-associated serogroups were consistent with earlier observations. There was no genetic structuring of the meningococcal population by age of carrier or sampling location; however, there was significant geographic structuring of the meningococci isolated in civil, but not military, institutions. The rate of capsule gene expression did not vary with age of carrier or meningococcal genotype, except for serogroup C, for which increased expression was associated with ST-11 (formerly ET-37) complex meningococci
ConclusionsSerogroup C capsule expression during carriage may contribute to the invasive character of ST-11 complex meningococci and to the high efficacy of meningococcal serogroup C conjugate polysaccharide vaccine
Neisseria meningitidis although responsible for bacterial meningitis and septicemia worldwide, causes disease infrequently relative to the prevalence of its asymptomatic carriage [1, 2]. Phenotypic and genotypic investigations have shown that carried meningococci are more diverse [3, 4] than disease-associated meningococci [5, 6]. For example, carried meningococci may be either encapsulated—such that they express 1 of 13 capsular polysaccharides, each of which corresponds to a particular serogroup [7]—or acapsulate, because of genetic down-regulation of capsule expression [8, 9], inactivation of genes in the capsule gene cluster (cps) or the absence of cps in capsule null (cnl) meningococci [10, 11]. In contrast, the majority of disease-associated isolates express polysaccharide of 1 of 5 serogroups: A, B, C, W-135, and Y [12]. The polysaccharide acid capsules corresponding to serogroups B, C, W-135, and Y predominate in Europe, Australia, and the Americas, whereas the polysialic acid capsules corresponding to serogroup A are predominant in Africa and Asia [12, 13]. The cps regions of serogroups B, C, W-135, and Y differ at the siaD locus, where 4 distinct allele classes occur (siaD B , siaD C , siaD W and siaD Y), each of which encodes serogroup-specific polysialyltransferases [14, 15]
Analysis of meningococcal genotypes by multilocus enzyme electrophoresis [5] and multilocus sequence typing (MLST) [6] has demonstrated that relatively few genotypes, termed “hyperinvasive lineages,” are responsible for most disease. Disease-causing meningococci belong to particular groups of related genotypes, termed “clonal complexes” [6, 13]—for example, the ST-11 (formerly ET-37) complex [6, 16, 17]—that are overrepresented in disease isolates relative to their carriage prevalences [3]. The nature of meningococcal transmission suggests that disease outbreaks follow the spread of carried hyperinvasive meningococci [1, 18]. In many countries, disease mainly occurs in infants and young children, whose carriage rates are low, although increased disease incidence also occurs in young adults, whose carriage rates are much higher [19, 20]
The most widely used meningococcal vaccines are composed of purified polysaccharide [21] and have the ability to contain disease outbreaks; however, they are ineffective in infants and do not confer memory responses in adults [22]. Polysaccharide-protein conjugates address these problems for serogroup C and probably for serogroups A, W-135, and Y, but poor immunogenicity and similarity to host antigens have hampered the development of serogroup B vaccines [23, 24]. A number of protein-based outer membrane vesicle vaccines have been developed [25 –27], but they are poorly cross-protective [28]. Because disease is not necessary for meningococcal transmission [29, 30], understanding the carrier state is important for resolving the epidemiology of meningococcal disease, which will enable the population effects of immunization—such as herd immunity and vaccine escape—to be exploited in immunization campaigns [1, 31]
The present study investigated meningococcal carriage in individuals 3–26 years old in Bavaria, Germany, at a time when meningococcal disease rates reported in Germany were <1/100,000 persons and when (in 2001) ∼20% of disease isolates expressed serogroup C capsule. The few disease isolates available from Bavaria at the time of this study were consistent with the national data. Disease clusters caused by serogroup C meningococci of the ET-15 variant of the ST-11 complex had occurred in Bavaria before the study was initiated [32]. Combining genotypic and phenotypic typing data for the carriage isolates elucidated relationships among capsular operons, capsule expression, and clonal complexes and determined the relative prevalence of genotypes among regions, institutions, and age groups
Isolation of carried meningococci Sampling was conducted from November 1999 to March 2000. The study protocol was approved by the ethics committee of the Medical Faculty of the University of Würzburg (study 137/99). At each sampling location, several educational institutions covering different age groups were chosen by the local health authorities, in consultation with school directors. In many cases, an informational event was organized for parents and guardians of students at the school, and the parents and guardians were provided with written information, including a consent form. At military camps, the commander chose the company to be sampled; military recruits were sampled within 2 weeks of arrival. Participation was voluntary, which was documented via a signed consent form; consent from parents and guardians was obtained for participants <18 years old. Most of the individuals who were approached agreed to participate. Retropharyngeal swab sampling was performed at each institution, and a single meningococcus was isolated from each culture-positive sample by direct plating onto Martin-Lewis agar plates (gift from Becton Dickinson). Gram-negative, oxidase-positive colonies were tested for β-galactosidase and α-glutamyltransferase activity, and the identity of the bacterium was confirmed by use of the API NH system (gift from BioMerieux)
MLST Isolates were characterized by MLST, as described elsewhere [4, 6]; an ABI Prism 3700 automated sequencer was used to separate the labeled extension products. The data were assembled by use of the Staden suite of computer programs [33]. Alleles, sequence types, and clonal complexes were assigned on the basis of the Neisseria MLST database (available at: http://pubmlst.org/neisseria) [34]
Serogroup analysis Previously described methods [35] were employed to measure the expression of serogroup A, B, C, W-135, and Y capsules (by ELISA with monoclonal antibodies); the presence of the mynB gene specific to serogroup A meningococci; and the presence of the siaD allele classes specific to serogroup B, C, W-135, and Y meningococci (the latter 2 by dot-blot hybridization with specific probes) [36]. The positive control isolates were as follows: serogroup A isolate Z2491 (ST-4) (gift from M. Achtman, Max-Planck Institut für Infektionsbiologie, Berlin, Germany), serogroup B isolate MC58 (ST-74), serogroup C isolate 2120 (ST-11), serogroup W-135 isolate 171 (ST-11), and serogroup Y isolate 172 (ST-166). The negative controls were gonococcal strain FA1090 (gift from M. Achtman) and Neisseria lactamica strain 4691 (German Collection of Microorganisms and Cell Cultures)
Statistical analyses The association between siaD allele class and age was evaluated for serogroups B, C, W-135, and Y. Age and 2-term fractional polynomial forms of age were used to test for monotonic or polytonic continuous associations [37]. Age was grouped as follows: 3–9, 10–14, 15–19, and ⩾20 years. The χ2 test was used to test for the association between age group and siaD allele class. The association between age and capsule expression was evaluated by use of a logistic regression model, with 2-term fractional polynomial forms of age as explanatory variables. Analyses were conducted for all 4 siaD allele classes and for each siaD allele class on its own
The association between siaD allele class and capsule expression was estimated by use of logistic regression, to provide odds ratios with 95% confidence intervals, and by Fisher’s exact test, to provide P values. A logistic regression model was used to assess the degree to which carrier age group might have confounded the results. The association between clonal complex and capsule expression was estimated by likelihood ratio tests across all 4 serogroups. Within each siaD allele class, this association was assessed by Fisher’s exact test with the Bonferroni correction for multiple comparisons
Analysis of molecular variance (AMOVA) Significant genetic differentiation among groups of isolates was assessed by AMOVA [38], as implemented in Arlequin software (version 2.000) [39]. This program computed an F statistic [40, 41] by applying a permutation test to assess statistical significance. AMOVAs were performed on the data as grouped by institution, institution excluding military camps, military camps versus all other institutions, age of carrier, carrier age group, and geographic location of the institutions. Two further analyses were performed on geographic subsets of the data, one excluding military camps and one including military camps only. For the analysis by carrier age group, the age groups described above were used. The locations of the institutions sampled were used for all geographic analyses in this study. Although the locations of origin (mainly Bavaria) were known for most of the military recruits, this was not amenable to analysis, because they were distributed throughout Bavaria
The following genetic characteristics were analyzed: nucleotide sequence of an individual locus, concatenated nucleotide sequences of all 7 loci, allele designations, concatenated allele designations of all 7 loci (known as the “allelic profile”), and sequence type. For the small genetic differences detected, analysis of the nucleotide sequences of individual loci lacked statistical power, but power was recovered when the concatenated nucleotide sequences, concatenated allele designations, and sequence types were used. Where appropriate, the Bonferroni correction was applied, to account for multiple comparisons
Mantel test Correlation between genetic distance and geographic distance was assessed by use of the Mantel test [42]. Square n-by-n matrices were generated, where n was the number of isolates and, with the exception of the diagonal elements, each element of the matrix corresponded to an isolate pair. The correlation coefficient was calculated for the distance matrices, and its significance was assessed by permutation [35]. Mantel tests were performed on the complete MLST data set (n=822), the data set excluding military camps (n=441), and the data set including military camps only (n=381). The geographic distance for a pair of isolates was taken to be the geographic distance between the towns at which the isolates were collected. For a pair of isolates, the genetic distance was defined variously, as follows: the proportion of nucleotide sites that differed at an individual locus, the proportion of nucleotide sites that differed across all 7 loci, a number that indicated whether the allele designation for a particular locus was the same (0) or different (1), the proportion of alleles that differed across all 7 loci, and a number that indicated whether the sequence types were the same (0) or different (1). All definitions of genetic distance were used to investigate the effect on the analysis; where appropriate, the Bonferroni correction was applied, to account for multiple comparisons
Meningococcal carriage rates From the 8000 children and young adults sampled, 830 (10.4%) meningococci were isolated; carriage rates were lowest in young children and highest in participants 25 years old (figure 1 and table 1). The 6821 participants who provided samples at schools (age range, 3–21 years) yielded 446 isolates (München, 75 isolates; Ingolstadt, 61 isolates; Erlangen, 57 isolates; Coburg, 38 isolates; Passau, 47 isolates; Würzburg, 47 isolates; Rottal-Inn, 47 isolates; Augsburg, 23 isolates; Sonthofen, 20 isolates; Dinkelsbühl, 17 isolates; and Weiden, 9 isolates). The remaining 384 meningococci were isolated from 1179 military recruits 18–26 years old at 6 camps (Roth, 109 isolates; Volkach, 96 isolates; Bayreuth, 95 isolates; Kempten, 50 isolates; Sonthofen, 21 isolates; and Ebern, 10 isolates)
Age-specific meningococcal carriage rates. The size of the populations for each year of age were as follows: 3 years, 109; 4 years, 247; 5 years, 339; 6 years, 420; 7 years, 525; 8 years, 532; 9 years, 569; 10 years, 538; 11 years, 465; 12 years, 469; 13 years, 460; 14 years, 423; 15 years, 397; 16 years, 320; 17 years, 361; 18 years, 411; 19 years, 420; 20 years, 405; 21 years, 236; 22 years, 121; 23 years, 67; 24 years, 45; 25 years, 12; and 26 years, 6. A Age-specific carriage rates by serogroup. Black serogroup B; cross-hatched to the right serogroup C; cross-hatched vertically serogroup W-135; cross-hatched to the left serogroup Y; white all other serogroups and not groupable. B Age-specific carriage rates by the gene present at the capsular region. Black, siaD B ; cross-hatched to the right, siaD C ; cross-hatched vertically, siaD W ; cross hatched to the left, siaD Y ; grey, cnl; white not groupable
Meningococcal capsule genes and serogroups The majority of the 822 MLST-typed meningococci (677/822 [82%]) were also characterized at cps. Most isolates (541/822 [66%]) possessed a siaD gene, with 136 (17%) containing cnl in place of cps [10]. The most prevalent serogroup was B (261/822 [32%]), and the most common siaD allele class was siaD B (333/822 [41%], corresponding to 4% of participants sampled). Serogroup C was of much lower prevalence (20/822 [2%]), as was the siaD C allele class (56/822 [7%] of all meningococci, corresponding to 0.7% of participants sampled) (table 1). Among the 4 siaD allele classes, siaD B serogroup B meningococci comprised the majority of isolates in all age groups (range, 55%–68%). There was some variation among age groups in excess of that expected by chance (P=.02, χ2 test). The main difference was a higher proportion of siaD C isolates found in the participants 15–19 years old and ⩾20 years old (overall, 12%; for each group, 9% and 16%, respectively) than in the participants 3–9 years old and 10–14 years old (overall, 4%; for each group, 1% and 9%, respectively), with associated reductions in the proportion of siaD B isolates found in the participants 15–19 years old and siaD Y isolates in the participants ⩾20 years old. The proportion of meningococci expressing their capsule genes was calculated from the prevalence of the serogroup-specific siaD allele classes and serogrouping data (table 2). Expression of capsule genes was much less common among siaD C and siaD Y isolates than among siaD B and siaD W isolates (table 2). Adjustment for age group had no substantial effect on this pattern (data not shown). There was no evidence for an effect of age of carrier on capsule expression, either overall or within any serogroup
Genetic diversity Among the 822 MLST-typed isolates, there were 323 unique sequence types, 221 of which occurred only once. The number of alleles per locus varied from 30 (adk) to 57 (pdhC). Most isolates (543/822 [66%]) were assigned to 1 of 20 previously identified clonal complexes, with 153 sequence types among the 279 unassigned isolates. With the exception of the serogroup A–associated clonal complexes (ST-1, ST-4, and ST-5), most of the previously described hyperinvasive lineages were present: ST-41/44 complex (138 isolates [17%]); ST-23 complex (71 isolates [9%]); ST-32 complex (41 isolates [5%]); ST-11 complex (8 isolates [1%]); and ST-8 complex (2 isolates) (table 3). The full data set, including isolate details, is available as an MLST dbNet database [34] at http://pubmlst.org/neisseria. A small number (21) of Bavarian disease-associated isolates contemporary to the present study were available; of these, 3 were ST-8 complex, serogroup C; 4 were ST-11 complex, serogroup C; 3 were ST-32 complex, serogroup B; and 5 were ST-42/44 complex, serogroup B. The remaining 6 isolates, all serogroup B, belonged to various unrelated sequence types
Association between clonal complexes with the siaD allele class and proportion of isolates expressing disease-associated capsule
Relationship between clonal complex and capsule genes The distribution of cps variants was not random among clonal complexes; certain clonal complexes were associated with particular cps genotypes. Furthermore, siaD C ST-11 complex isolates were more likely to express their capsule than were siaD C isolates that belonged to the other clonal complexes (table 3). All ST-53 complex (n=59) and ST-198 complex (n=44) isolates were cnl meningococci, as reported elsewhere [10]. The 401 siaD gene–containing isolates that were assigned to a known clonal complex were distributed among 16 clonal complexes (table 3), and there was no strong evidence showing an association between clonal complex and capsule expression (P = .30, by a likelihood ratio test from a logistic regression model that excluded 15 isolates because their clonal complexes were represented by ⩽3 isolates). In a logistic regression model (n=386) that included the siaD allele class, there was weak statistical support for an overall association between clonal complex and capsule expression (P=.09). No formal interaction test of reasonable power was feasible for this data set, given the strong correlation between clonal complex and serogroup and the limited number of isolates for some clonal complexes. Within individual serogroups, there was an association between clonal complex and capsule expression in siaD C isolates that remained significant after application of the Bonferroni correction for 4 comparisons (P=.03; n=34) but not in any other siaD allele class, with P>.1 for siaD B (n=252), siaD W (n=31), and siaD Y (n=84). This association was due to the high level of expression among ST-11 complex and ST-8 complex isolates and the low level of expression among other siaD C isolates (table 3)
Genetic structuring AMOVA provided no evidence for structuring of the concatenated nucleotide sequence data for any of the groupings tested except geographic location, where low but statistically significant levels of structuring were observed (table 4). Further AMOVAs showed that there was significant geographic structuring for the nucleotide sequences of 3 alleles, for all but 1 allele designation (aroE) for the allelic profile, and for the sequence type (table 5). When the data were categorized by institution type (military or school), the evidence for geographic structuring remained for the schools but not for the military camps. The correlation coefficients derived from the Mantel test (ρ) provided similar evidence of geographic structuring for the concatenated nucleotide sequence, the allele designations, the allelic profile, and the sequence type for the complete data set and for the schools. These analyses demonstrate that the degree of genetic differentiation correlated with geographic distance excepting only military camps, for which there was no support for geographic structuring (table 5)
Analysis of molecular variance for various groupings of data, performed on concatenated nucleotide sequences
Meningococcal vaccines protect individuals from disease by eliciting serum bacterial antibodies [43] but can also induce mucosal immunity, affecting carriage and transmission [44]. In terms of public health, the resultant population effects can be either advantageous, if transmission of the disease-associated meningococcus is interrupted, or disadvantageous, if they promote the emergence of vaccine escape variants [31]. Quantitative description of meningococcal carriage and transmission in terms of capsule expression and genotype is, therefore, necessary to understand these effects
Here, a carried meningococcal population is analyzed in terms of host characteristics, genotype, and capsule expression. The age-specific carriage prevalence and frequency of serogroups observed were consistent with the findings of previous studies, which have consistently shown low rates of carriage in young children that rise dramatically with age [11, 19, 45 –48], presumably as a result of behavioral changes. Our results also confirmed that some cps genotypes are associated with particular clonal complexes [10, 13]. Except for a marginal increase in the carriage of serogroup C meningococci with age, there was no evidence for associations between meningococcal serogroups and geographic regions, institutions, or age groups. Meningococci with the siaD C or siaD Y allele classes were less likely to express capsule than were those with the siaD B or siaD W allele classes. There were significant differences among the meningococci with the siaD C allele class, with evidence for a possibly very strong association between the ST-11 complex and siaD C expression. This novel observation is potentially important, because, notwithstanding their low carriage prevalence, members of the ST-11 complex have caused elevated levels of disease on numerous occasions, recently spreading across several continents [49]. Because capsule expression is a major virulence determinant [12], this property may contribute to the invasive character of ST-11 siaD C meningococci. If these high levels of capsule expression are also necessary for fitness in terms of transmission, these meningococci may be particularly vulnerable to the mucosal immunity generated by conjugate polysaccharide vaccines [48]. This may have contributed to the high efficacy of the meningococcal serogroup C conjugate polysaccharide vaccines recently introduced in the United Kingdom [50] and elsewhere
The only genetic structuring observed was by geographic location, with the observed F ST indicating that 0.7% of the total genetic diversity could be attributed to differences between locations. This statistic has not been widely used in bacterial populations, but the level was lower than the differentiation (1.5%–9.4%) observed between environmental and farm-animal isolates of the zoonotic pathogen Campylobacter jejuni [51]. Although the AMOVA showed that there was a strong signal of association between particular genotypes and given geographic locations (P<.001), the magnitude was not great. This low but statistically significant result for meningococci was consistent with the idea that meningococcal populations contain distinct genotypes—the clonal complexes—whose members have global distribution but are unevenly distributed among different geographic regions at any given time. The data also suggested that this genetic differentiation is very recent, perhaps arising as a consequence of the fact that clonal complexes spread faster than novel clonal complexes emerge. For an obligate inhabitant of humans such as the meningococcus, the spread of genotypes is likely to be related to host demographic behavior. In this context, the present data suggest that the rate of spread is sufficiently slow to result in detectable geographic structuring among major population centers in a region the size of Bavaria, providing some indication of the relative intensity of transmission between and within population centers
Further analysis of geographic structuring by AMOVA and the Mantel test confirmed that significant but low levels of geographic structuring were present and that structuring was the result of isolation by distance. Indeed, this structuring was often undetectable when individual loci were analyzed, caused by a lack of statistical power. MLST is a typing procedure, optimized for epidemiological applications rather than population-genetics analysis [52], and the present results suggest that, for meningococcal populations, contiguous nucleotide sequences in excess of the ∼500 bp used in MLST are required to detect geographic structuring at the level of individual loci. Geographic structuring was, however, detectable at the level of concatenated nucleotide sequences, allele designations, allelic profile, and sequence types. Each of these levels of analysis emphasized differences among the regions, allowing them to be detected against the background of high genetic variation that is present in meningococcal populations [4]. These analyses further established that the geographic structuring was present among the schools but absent among the different military centers, which was consistent with the wide geographic catchment areas of the latter institutions, with the majority of recruits originating in diverse locations in Bavaria
The present study provides support for the idea that variations in disease incidence are due to the gradual spread of hyperinvasive meningococci, with intensity of transmission declining with geographic distance. Geographic structuring by distance is a likely explanation for the differences in the predominant clonal complexes that cause invasive disease in different European countries at any given time [13]. Furthermore, notwithstanding the differences in carriage rates among different cohorts, the meningococcal transmission system is the community as a whole, rather than a subpopulation of a particular age or peer group
Although a large number of individuals were included in the present study, the high diversity of the meningococcal population, the low levels of carriage of the ST-11 complex, and the lack of a large set of contemporary disease isolates limited some of the conclusions that could be drawn. In particular, although the finding that ST-11 meningococci with the siaD C allele class may be unusual in terms of capsule expression has implications for disease control, because the number of isolates on which this observation was based is small, the finding has to be confirmed in a study that includes more isolates. Nonetheless, the present study has indicated the types of analyses that can be performed on these data; it has also generated hypotheses and data that will be essential to the design of the future studies needed to further explore the associations described here
The Bavarian Government and the German Armed Forces are gratefully acknowledged for their assistance. We also thank Man-Suen Chan, for help with data analysis; Dirk Alber, Marc Oberkötter, Silke Getzlaff, and Carmen Kantelberg, for assistance with sampling; Gabi Heinze, for providing expert technical assistance; Rainer Maag, for providing help with serogrouping; and Man-Suen Chan, for developing, with K.A.J., the Neisseria multilocus sequence typing database (available at: http://pubmlst.org/neisseria)
↵Presented in part: 13th International Pathogenic Neisseria Conference, Oslo, Norway, 1–6 September 2002 (abstract 49)
Financial support: Deutsche Forschungsgemeinschaft (grant VO718/3, SPP 1047); German Federal Ministry of Education and Research (project III/2 of the Pathogenomik network); Deutsche Gesellschaft für Hygiene und Mikrobiologie; Senator Kurt und Inge Schuster-Stiftung. M.C.J.M. is a Wellcome Trust Senior Research Fellow, N.D.M. is a Wellcome Trust Clinical Training Fellow, and D.J.W. is a Biotechnology and Biological Sciences Research Center Research Student
IDSA Members: For your free access to this journal, log in via the IDSA members area.
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy