ABSTRACT
The choice of adequate methods for epidemiological purposes remains a challenging problem in Neisseria gonorrhoeae molecular monitoring. In this study, the collection of geographically unrelated gonococci (n = 103) isolated in Russian clinics was comparably tested by (i) a traditional serotyping scheme, (ii) por typing, (iii) Neisseria gonorrhoeae multiantigen sequence typing (NG-MAST), and (iv) multilocus sequence typing (MLST). It is shown that, according to sequencing data, a third of the strains carried new porB1 alleles, as well as tbpB ones, and more than half of the samples had new sequence types (STs) as determined by NG-MAST or MLST. The discriminatory power for each typing method was calculated by using the Hunter-Gaston discriminatory index, D. Commonly, modern nucleic acid-based typing methods (por typing, NG-MAST, and MLST) appeared to be more efficient than the classical serotyping scheme. While the traditional serotyping gave a D value of 0.82, the por typing, NG-MAST, and MLST approaches yielded D values of 0.97, 0.98, and 0.91, respectively. Each typing technique revealed the distribution of gonococci slightly correlated with their geographical sources. However, only the MLST method STs were highly associated with certain phenotypes. Although ST1594, ST1892, and ST6720 were typical for susceptible gonococci, ST1901 and ST6716 were undoubtedly associated with a multidrug-resistant phenotype. We conclude that every tested nucleic acid-based typing method is suitable for N. gonorrhoeae molecular surveillance. However, the MLST method seems to serve large-scale epidemiological purposes, whereas the NG-MAST and por typing approaches are more appropriate for the investigation of local outbreaks.
The choice of an optimal epidemiological approach for Neisseria gonorrhoeae typing remains a problem for public health control strategies. The worldwide spread of drug-resistant strains requires typing methods to be introduced into a national surveillance programs that are already realized within the Australian Gonococcal Surveillance Programme (33). In Russia, strains that are resistant to penicillin, tetracycline, and fluoroquinolones are common, reaching up to 60% prevalence in some regions (14). As is generally known, most resistance mechanisms in N. gonorrhoeae are linked to mutations in genomic DNA, and a wide dissemination of such mutations has been demonstrated in subjects with gonorrhea in the Russian population (11). In fact, chromosomally mediated drug resistance expands clonally in a bacterial population. Thus, tracing the spread of gonorrhea by using strain differentiation methods is of great importance. In addition, population genetics are important in understanding the evolutionary history, epidemiology, and population dynamics of pathogens.
Widespread adoption of the molecular diagnosis of gonorrhea could compromise traditional bacteriological cultivation in routine practice and highlights the need for the development of molecular tools for N. gonorrhoeae typing. Several promising approaches exist in this field. One of them is por typing (7, 8, 27), based on comparative analysis of porB gene nucleotide sequences. It possesses rather high discriminatory power (39) and presents data in a format comparable with serotyping data. It should be mentioned that PorB1 protein plays a key role in the adhesion of gonococci to epitheliocytes and reflects the pathogenicity of isolates. Moreover, in earlier studies the multidrug-resistant phenotype of N. gonorrhoeae was bound to certain serotypes determined in accordance with the antigenic properties of PorB1 proteins (7, 21, 29).
Today, the N. gonorrhoeae multiantigen sequence typing (NG-MAST) (19) appears to be a leading method for N. gonorrhoeae typing. It is based on the analysis of internal fragments of two hypervariable genes, porB and tbpB, encoding superficial gonococcus antigens and therefore under positive selection (19). An online database (www.ng-mast.net ) collects current information on the sequence types of gonococcus isolates in different regions of the world, and it can be an efficient instrument for objective estimation of the genetic variability of the microbial population and tracing the spread of infection (2). Moreover, an association of the NG-MAST sequence type with the antibiotic resistance profile has been demonstrated by some investigators (20, 26).
In addition, NG-MAST, which is usually carried out on DNA extracted from a pure bacterial culture, can be performed directly from noncultured samples such as a piece of clothing (18) or clinical specimens (40). Although some researchers have reported a successful application of NG-MAST to urogenital specimens (urine specimens and swabs from the cervix, urethra, and vagina) and to rectal swabs, this typing method was found to be less suitable for throat swabs due to cross-reaction with commensal Neisseria species. It seems that direct typing schemes utilizing probe hybridization methods (16, 17) for a broad spectrum of clinical samples are more convenient than sequencing.
The opposite approach—analysis of conservative, presumably selective neutral housekeeping genes—is taken by multilocus sequence typing (MLST) (1, 38). Possessing a sufficient number of allelic variants and characterized by the slow accumulation of mutations, these genes reflect the natural evolution of the microbial population and common trends in the spread of gonorrhea.
Often, the effectiveness of N. gonorrhoeae surveillance depends on the methods used for species identification and the typing of clinical isolates. However, considering the variety of typing systems but the lack of world experience in their application to monitoring gonococci, it has not yet been determined whether there is a unique objective method for revealing the relationships between clinical isolates. The goal of the present study was to evaluate a number of approaches, including por typing, NG-MAST, MLST, and serotyping for typing geographically unrelated gonococcal isolates in Russia.
(This study was partially presented during the 16th International Pathogenic Neisseria Conference in 2008 in the Netherlands.)
MATERIALS AND METHODS
Bacterial isolates. N. gonorrhoeae clinical isolates collected in different regions of Russia from 2004 to 2005 were obtained from the Central Research Institute of Dermatology and Venereology. All isolates were identified as N. gonorrhoeae and tested for serotype and for susceptibility to penicillin G (PEN), tetracycline (TET), ciprofloxacin (CIP), spectinomycin, and ceftriaxone as a part of a previous investigation (11).
Genetic analysis.Genomic DNA from N. gonorrhoeae was extracted according to the method of Boom et al. (4). When necessary, the prepared DNA samples were stored at −20°C.
Amplification reactions were carried out in 10 μl of 66 mM Tris-HCl (pH 9.0), 16.6 mM (NH4)2SO4, 2.5 mM MgCl2, 0.2 mM concentrations of each deoxynucleoside triphosphate (dNTP), 5 pmol of each primers (see Table S1 in the supplemental material), and 1 U of Taq polymerase (Fermentas, Lithuania) under the following conditions: 35 cycles of 94°C for 20 s, 60°C for 20 s, and 72°C for 15 s. A programmed thermocycler Tetrad DNA Engine (MJ Research, Inc.) was used.
Dephosphorylation of the 5′-end phosphate groups of dNTPs and cleavage of primers in the postamplification reaction mixture was done by incubation with 0.5 U of shrimp alkaline phosphatase and 0.1 U of exonuclease I (both enzymes from Fermentas, Lithuania) for 20 min at 37°C, followed by the inactivation by heating for 10 min at 85°C.
Sequencing procedure was performed by the modified Sanger method using a ABI Prism BigDye terminator cycle sequencing ready reaction kit and an ABI Prism 3100 genetic analyzer (Applied Biosystems, Hitachi, Japan) according to the manufacturer's instructions.
Analysis of the nucleotide sequences was carried out by using Vector NTI Advance v.9.0 software (Infomarks, Inc.). The DNA sequences of porB and tbpB genes collected during the present study were submitted to GenBank (accession numbers EU530732 to EU530817 and EU532618 to EU532759, respectively). The DNA sequences of housekeeping genes were uploaded on the http://pubmlst.org website.
NG-MAST or MLST.Characterization of all isolates by NG-MAST and MLST was performed as originally described (1, 19). The corresponding sequences were submitted to the NG-MAST (http://www.ng-mast.net ) or MLST (http://pubmlst.org/neisseria ) websites for comparison with the existing alleles for determination of the allele types and sequence types (STs) of the isolates.
Data analysis.Sequences were aligned by using CLUSTAL X 1.8 software (http://www.clustal.org ). At the next step, aligned nucleotide sequences were converted to MEGA version 4.0.2 software (http://www.megasoftware.net ), and distance matrices were estimated (using the nucleotide Tajima-Nei model). Further classification was constructed by using the neighbor-joining method based on a bootstrap tree stability test (500 iterations).
Visual mapping of multidrug resistance and place of strain isolation were carried out by using the Dendroscope tree editor (http://www-ab.informatik.uni-tuebingen.de/software/dendroscope/welcome.html ) (10).
The discriminatory power for each typing method was defined by calculation of a single numerical index of discrimination (i.e., the Hunter-Gaston discriminatory index [D]) (9), based on the probability that two unrelated isolates would be placed into different typing groups: $$mathtex$$\[D{=}1{-}\frac{1}{N(N{-}1)}\ {{\sum}_{j{=}1}^{S}}n_{j}(n_{j}{-}1)\]$$mathtex$$ where N is the number of unrelated strains tested, S is the number of different types, and nj is the number of strains belonging to the jth type.
The calculated data was interpreted with commonly accepted decision that a D of >0.90 is desirable for a confident typing method (9).
A recombination test to determine the index of association and the standardized index of association (IA and IS A, respectively) was performed using the START2 program (12). Allele profile data were analyzed in eBURST (version 3; http://eburst.mlst.net/v3/enter_data/single/ ) to define clonal complexes or groups (6).
Associations between sequence type and susceptibility level were examined by using chi-square (χ2) and Cramer (V) tests. All statistical calculations were two-tailed and were conducted with a significance level of at least P < 0.05. Statistical analysis was performed with Statistica 7.0 software.
RESULTS
In all, 103 clinical N. gonorrhoeae isolates from five remote regions of Russia—Irkutsk (n = 22), Samara (n = 10), Arkhangelsk (n = 28), Murmansk (n = 14), and St. Petersburg (n = 29)—were investigated (see Table S2 in the supplemental material).
Protein-based typing method.The traditional serotyping procedure divided all typing isolates into 11 different groups, with a prevalence of the PIB serovar (92/103 [89.3%]). Within this serovar the PIB2, PIB3, PIB5, and PIB22 serotypes were detected in 30 (29.1%), 27 (26.2%), 10 (9.7%), and 10 (9.7%) cases, respectively. The remaining PIB serovar isolates (n = 15) belonged to five different serotypes: PIB3/6 (n = 4), PIB4 (n = 4), PIB8 (n = 2), PIB9 (n = 3), and PIB26 (n = 2). Among the PIA serovar isolates, the PIA6 serotype was found in 10 (9.7%) isolates, and one sample belonged to the PIA10 serotype. The calculated D value was found to be 0.82, which shows that the discriminatory power of serotyping is not sufficient to make it an effective typing method.
Nucleic acid-based typing methods.Nucleic acid-based techniques, such as por typing, NG-MAST, and MLST, divided all of the isolates into 58, 61, and 30 groups, respectively.
por typing.Alignment of the entire porB gene nucleotide sequences revealed the sufficient heterogeneity of N. gonorrhoeae clinical isolates. The dendrogram constructed by the neighbor-joining algorithm is shown in Fig. 1. When a cutoff value of 0.1 (genetic distance) was used, 58 clusters could be delineated. Expectedly, all gonococcal isolates were separated into two large groups corresponding to the porB1a (n = 11) and porB1b (n = 92) alleles of the porB gene. Comparative analysis of the nucleotide polymorphism showed a much higher variability for the porB1b allele than for porB1a. Variability for the porB1b allele became apparent in a number of exposed allelic variants and polymorphic sites within each group. Moreover, 3-, 6-, or 12-nucleotide insertions, as well as 6-nucleotide deletions, were found for several isolates possessing the porB1b allele. Analysis of the nonsynonymous/synonymous substitution rate ratio revealed different trends in the natural selection of porB alleles (Table 1). While the influence of positive diversifying selection was shown for the porB1b allele, the porB1a allele seemed to be under the influence of purifying selection. Commonly, the discriminatory index for this typing method was 0.97, which is sufficient for epidemiological purposes.
Phylogenetic tree based on analysis of the entire porB1 gene nucleotide sequences. Serovars P1A and P1B are indicated by vertical lines. Clinical N. gonorrhoeae strains (n = 103) are named according to Table S2 in the supplemental material. Isolate origins: Irkutsk (n = 22), black square; Samara (n = 10), gray square; Murmansk (n = 14), large black circle; St. Petersburg (n = 29), small black circle; Arkhangelsk (n = 28), large gray circle. Multiresistant isolates (i.e., isolates resistant to at least two antibiotics) are indicated by thick lines.
Variability of entire porB gene nucleotide sequences for 103 N. gonorrhoeae clinical isolates
Comparative analysis of internal fragments of porB genes (490 bp) via the public database on the NG-MAST website (http://www.ng-mast.net ) showed that a third of tested group carried new por alleles unrepresented in the nucleotide library. The 49 isolates formed six clusters carrying por-37 (n = 10), por-90 (n = 10), por-91 (n = 8), por-164 (n = 6), por-232 (n = 7), and por-685 (n = 8) alleles. Among these, the por-90 cluster was clearly associated with the PIA6 serotype, and the por-91, por-164, and por-232 clusters contained gonococci isolated in Irkutsk, Samara, and Murmansk, respectively.
NG-MAST.Based on analysis of porB and tbpB fragments, all isolates were characterized by NG-MAST. We found 51 and 27 variants of porB and tbpB alleles, respectively, which resulted in the assignment of 61 different sequence types (STs). A phylogenetic tree based on the concatenated sequences of two loci is shown in Fig. 2. The 57 (55%) isolates belonged to 15 different STs containing from two to eight members. The largest of them were ST1043 (n = 8), ST285 (n = 7), ST206 (n = 6), and ST972 (n = 5). The other 46 STs were represented by individual isolates, including four penicillinase-producing N. gonorrhoeae (PPNG) isolates collected in St. Petersburg (n = 3) and Murmansk (n = 1). Meanwhile, approximately half of the isolates (n = 50) belonged to the 39 new STs described here for the first time. Generally, there was no clear geographic clustering of specific NG-MAST types found within the country. In some cases, certain STs were found for gonococci isolated in the same place, for example, ST205 in Samara, ST285 in Murmansk, and ST972 in Irkutsk. However, this probably does not reflect local outbreaks but rather the global spread of infection. The calculated index of discrimination for this typing method was 0.98, which reflected the strong discriminatory power of NG-MAST.
Phylogenetic tree based on analysis of concatenated sequences of porB1 and tbpB gene fragments (NG-MAST scheme). Newly described STs are indicated in red; serovars P1A and P1B are indicated too. Clinical N. gonorrhoeae strains (n = 103) are named according to Table S2 in the supplemental material. Isolate origins: Irkutsk (n = 22), black square; Samara (n = 10), gray square; Murmansk (n = 14), large black circle; St. Petersburg (n = 29), small black circle; Arkhangelsk (n = 28), large gray circle. Multiresistant isolates (i.e., isolates resistant to at least two antibiotics) are indicated by thick lines.
MLST.According to MLST data, the 30 different groups (STs) were discovered. Among them, 13 ST clusters of between 2 and 25 isolates were identified. ST6716 (n = 25) were found to be the most prevalent, the other largest groups were ST1901 (n = 11), ST1905 (n = 11), ST1892 (n = 8), ST1584 (n = 8), and ST6720 (n = 5). The 17 STs were represented by single isolates. Concatenated sequences of seven housekeeping gene fragments were analyzed by using START 2.0 software. A phylogenetic tree constructed by using the neighbor-joining method is shown in Fig. 3. The calculated D value was determined to be 0.91 by MLST, which demonstrates this method's applicability as an effective typing method.
Phylogenetic tree based on analysis of concatenated sequences of seven housekeeping gene fragments (MLST scheme). Newly described STs are indicated in red. Clinical N. gonorrhoeae strains (n = 103) are named according to Table S2 in the supplemental material. Isolate origins: Irkutsk (n = 22), black square; Samara (n = 10), gray square; Murmansk (n = 14), large black circle; St. Petersburg (n = 29), small black circle; Arkhangelsk (n = 28), large gray circle. Multiresistant isolates (i.e., isolates resistant to at least two antibiotics) are indicated by thick lines.
Analysis by eBURST showed that four nonoverlapping groups (clonal complexes) contained STs, which matched at least one other ST at five or more loci (Fig. 4). Four STs—ST1901, ST1927, ST6716, and ST6715—were recognized as hyperinvasive genotypes. Among them, ST1901 and ST6716 were evolutionary related, sharing six of seven alleles.
Population snapshot of clinical N. gonorrhoeae isolates (n = 103) typed in the present study. The snapshot was created by the eBURST algorithm applied for the analysis of MLST data. The circles represent STs differing in only one housekeeping gene sequence from the founder genotype (in the center). Lines connect the other evolutionarily related STs, i.e., those sharing six of seven alleles. Each centered group was considered a clonal complex. ST1594 was identified as a singleton.
Based on MLST data, the examined N. gonorrhoeae isolates were compared to gonococcal isolates collected in the United Kingdom (1) in terms of allelic diversity (Table 2) and recombination parameters (Table 3). The most changeable loci were found to be similar to these collections, and significant linkage disequilibrium was detected for both populations.
Allele frequencies within N. gonorrhoeae isolates collected in the Russian Federation and the United Kingdoma
Multilocus linkage disequilibrium analysis of N. gonorrhoeae isolates collected in the Russian Federation and the United Kingdoma
Genotyping data and antimicrobial resistance. N. gonorrhoeae clinical isolate distribution according to different nucleic acid-based typing schemes (por typing, NG-MAST, and MLST) was compared to their susceptibility profiles. Since many STs were represented by occasional isolates, only clusters that contained five or more members were considered. Among them, NG-MAST ST205 (n = 6), ST285 (n = 7), and ST282 (n = 5) were unambiguously associated with multidrug resistance, i.e., intermediate susceptibility or resistance to penicillin, tetracycline and ciprofloxacin. NG MAST ST1043 (n = 8) included gonococci not susceptible to PEN, resistant to TET, and susceptible to CIP. Appropriately, the same isolates also formed corresponding clusters in accordance with the por typing results.
A relationship between the N. gonorrhoeae phenotype and certain STs was observed by MLST (Table 4). Although ST1594, ST1892, and ST6720 were typical for susceptible gonococci, ST1901 and ST6716 were undoubtedly associated with multidrug resistance. ST1905 comprised isolates displaying intermediate susceptibility to PEN, resistance to TET, and (especially) susceptibility to CIP.
Distribution of 103 N. gonorrhoeae isolates by susceptibility level against different sequence types according to MLST
DISCUSSION
In recent years, many molecular approaches for N. gonorrhoeae typing have been suggested: opa typing (23, 25), por typing (7, 15, 34, 39), amplified fragment length polymorphism (24, 32), restriction fragment length polymorphism (3, 15), and some others, but no single typing scheme has been generally adopted, and the lack of such a standard typing method has impeded the sharing of epidemiological data between laboratories. The recently developed NG-MAST (19) and MLST (1, 38) approaches look the most promising in this field.
In our research, four different schemes of molecular typing have been adopted and used to study gonococcal strains circulating in the Russian Federation. The traditional serotyping procedure is based on an analysis of protein variability. The other typing schemes—por typing, NG-MAST, and MLST—are nucleic acid-based techniques.
Each tested technique typed all of the isolates, but their discriminatory powers were different. According to our data, the discriminatory power of the traditional serotyping approach was 0.82 and did not satisfy the value of ≥0.90 desirable for effective typing systems (9). Although this disadvantage of serotyping was discovered 10 years ago (37), this method can still be useful when a comparison of modern and old data is required.
The discriminatory power of each of the nucleic acid-based techniques was quite high. Both por typing and NG-MAST satisfied the value of ≥0.95 recommended for new typing methods by the European Society of Clinical Microbiology and Infectious Diseases (ESCMID) (36), in contrast to the MLST scheme.
Generally, the N. gonorrhoeae typing data obtained by por typing and NG-MAST appeared to be very similar. Both methods divided gonococcal strains into two large nonoverlapping groups in accordance with porB alleles (or serovars in terms of proteins). Both methods had comparable discriminatory power (0.97 and 0.98 for por typing and NG-MAST, respectively) sufficient for identifying transmission clusters of N. gonorrhoeae. In addition, there was no clear geographic clustering of specific por types or STs according to NG-MAST found within the examined group.
As far as por typing is concerned, different trends in the evolution of porB alleles were found. Earlier, such a phenomenon was described by Smith et al. in 1995 (28) and Posada et al. in 2000 (31). This fact can be explained by different epidemiologies in the P1A and P1B homology groups. It was shown that blood isolates obtained during disseminated gonococcal infection generally belong to the P1A serovar (5, 29), whereas isolates from mucosal surfaces more often belong to the P1B serovar (21). It may be that the invasive process requires certain antigens exposed on the microbial cell surface that elucidate the influence of purifying selection on PIA gonococci.
In contrast to por typing, NG-MAST allows international comparisons of genotyping data. Unfortunately, most of the STs identified in our research were newly described STs. Of 23 known STs, ST1527, ST1534, ST1544, and ST1548 had previously been detected in N. gonorrhoeae samples obtained in Arkhangelsk (35). It should be noted that the most common ST in a European population, ST225 (26), was found in one isolate from Irkutsk.
Although the discriminatory power of MLST was slightly lower than that of por typing and NG-MAST, the obtained genotyping data looked very promising. First, the tested gonococci were divided into 30 different clusters aside from their serotypes. In spite of the fact that N. gonorrhoeae has earlier been proposed to have a nonclonal population structure (22, 30), four clonal complexes were identified in the analyzed group.
Moreover, using data from public MLST site (http://pubmlst.org/neisseria ) allowed us to compare the N. gonorrhoeae populations in the Russian Federation and the United Kingdom. The similar variabilities of alleles in both populations reflect the universality of this typing scheme and its applicability for worldwide studies.
Despite the significant linkage disequilibrium detected for both groups, the N. gonorrhoeae population in Russia seems more homogeneous, more clonal than in the United Kingdom. For Russian isolates, the observed variance was much greater than the maximum variance obtained in 1,000 trials (P = 0.000). In combination with the IA and IS A values, this suggests that the population structure of the examined N. gonorrhoeae isolates is highly clonal. It was confirmed by the eBURST analysis data shown that the majority of gonococci can be combined into four clonal complexes with only a singleton formed by ST1594.
Only 6 STs discovered here by MLST were described elsewhere, and 24 STs were newly elaborated. It should be mentioned that ST1905 was previously identified only in gonococci isolated in Uzbekistan and Russia (1). In accordance with MLST findings, there was no evidence for geographical structuring among the isolates examined, and a strong relationship between the susceptibility profile of N. gonorrhoeae and certain STs was discovered. Moreover, two STs clearly associated with multidrug resistance were determined by eBURST to be evolutionarily related (Fig. 4). If we take into account the geographic heterogeneity of each cluster (Fig. 4; see also Table S2 in the supplemental material), it seems that the multiresistant strains in Russia spread as two large clonal groups corresponding to MLST ST6716 and ST1901.
In conclusion, we assert that the methods based on nucleic acid variability are more suitable for N. gonorrhoeae molecular typing than the ones based on changes in proteins. Among these, the NG-MAST displayed the highest resolution that is useful in monitoring short-term transmission patterns of the organism and is more appropriate for local outbreak investigations. Meanwhile, MLST, in spite of its lower discriminating potential, seems to be the best choice for global epidemiological studies.
ACKNOWLEDGMENTS
This study was supported by the Russian Ministry of Health (development contract 06/311 [30 May 2008]).
We thank V. A. Karpov for oligonucleotide primer synthesis and M. M. Chukin and T. A. Akopian for assistance with the sequence instrumentation.
We used the NG-MAST database (http://www.ng-mast.net ), Department of Infectious Disease Epidemiology, Imperial College London, and the Neisseria MLST website (http://pubmlst.org/neisseria/ ) developed by Keith Jolley and Man-Suen Chan (13) and located at the University of Oxford.
FOOTNOTES
- Received 16 March 2010.
- Returned for modification 28 April 2010.
- Accepted 12 July 2010.
- Copyright © 2010 American Society for Microbiology
