HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by Packard, E. R. Articles by Fry, N. K. Search for Related Content PubMed PubMed Citation Articles by Packard, E. R. Articles by Fry, N. K. Agricola Articles by Packard, E. R. Articles by Fry, N. K.

Sequence variation and conservation in virulence-related genes of Bordetella pertussis isolates from the UK

¹Division of Infection and Immunity, Institute of Biomedical and Life Sciences, University of Glasgow, Glasgow G12 8QQ, Scotland, UK ²Health Protection Agency, Respiratory and Systemic Infection Laboratory, Specialist and Reference Microbiology Division, 61 Colindale Avenue, London NW9 5HT, UK

Correspondence Norman K. Fry Norman.Fry{at}HPA.org.uk

Received October 22, 2003
Accepted January 12, 2004

To determine the value of gene markers for surveillance and to assess the genetic stability of potential acellular pertussis vaccine components, the sequence variation in ten virulence-related genes of Bordetella pertussis was investigated in strains isolated in the UK between 1920 and 2002. These genes encode: pertactin (prnA); pertussis toxin subunits S1 (ptxA) and S3 (ptxC); tracheal colonization factor (tcfA); bordetella autotransporter protein C (bapC); bordetella resistance to killing protein (brkA); fimbrial antigen 2 (fim2); outer-membrane protein Q (ompQ); virulence-activated gene 8 (vag8) and adenylate cyclase toxin (cyaA). The encoded proteins are either components of current acellular vaccines (ACVs), or potential virulence markers for B. pertussis. Three strains used in the pertussis UK whole-cell vaccine (WCV), strain Tohama-I used for production of ACV components and the type strain of B. pertussis (18323^T) were also analysed. Several novel alleles were found. The UK isolates were assigned multi-locus sequence types (MLSTs) according to a previously described scheme for B. pertussis based on three of these genes (ptxA, ptxC and tcfA). Compared with isolates from other countries, the UK clinical strains showed a distinct distribution of MLSTs. Apart from one strain that was MLST-3, all other recent isolates (2000–2002) were identified as MLST-5. These isolates differed from the three WCV strains, which were MLST-2 or MLST-3, the Tohama-I strain (MLST-2) and the type strain of B. pertussis (MLST-9). MLST-3 and MLST-5 differ only by a single synonymous mutation, but this method does indicate that currently circulating strains of B. pertussis are not identical to the vaccine types, and they may differ in other important characteristics. Two new MLSTs were identified amongst historical UK isolates. Sequence-based typing offers a convenient method of analysing and comparing populations of B. pertussis from different time periods and from different countries. The variation exhibited by prnA and fim2 suggests that they could be useful, additional epidemiological markers in such a typing scheme.

	INTRODUCTION

TOP INTRODUCTION METHODS RESULTS AND DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
The identification of gene polymorphisms in bacteria by nucleotide sequence analysis can be useful in the epidemiological surveillance of bacterial pathogens (Maslow et al., 1993). The advent of identification libraries derived from sequence-based typing systems that utilize non-selective housekeeping genes, such as the multilocus sequence typing (MLST) schemes (Maiden et al., 1998; Spratt, 1999) (http://www.mlst.net), has facilitated the ability to identify and track particular strains over time. Other markers that may be under selective evolutionary pressure, including those encoding antibiotic resistance (Oliveira et al., 2001) and surface proteins (Maiden et al., 1991; van Loo et al., 2002; Gaia et al., 2003), can also be used, particularly in clonal organisms such as Bordetella pertussis, where the selection of useful markers can be problematic.

Nucleotide polymorphisms in genes encoding two virulence factors of B. pertussis, pertactin (prnA) and pertussis toxin S1 subunit (ptxA), were first demonstrated in Dutch isolates by Mooi et al. (1998) following a large outbreak of pertussis in The Netherlands in 1996. One possible explanation for this resurgence was that the circulating B. pertussis population consisted of strains that were antigenically distinct from the vaccine strains (Mooi et al., 1998). Subsequent studies have reported allelic variation in these two markers in Finland, the UK, Poland, the USA and Canada (Mooi et al., 1999; Cassiday et al., 2000; Fry et al., 2001; Gzyl et al., 2002; Peppler et al., 2003). Additional targets under presumed selective pressure were examined for nucleotide sequence variation by van Loo et al. (2002), including other genes encoding proteins used as acellular vaccine (ACV) components and other surface-associated proteins. Three of these, encoding tracheal colonization factor (tcfA) and pertussis toxin subunits S1 and S3 (ptxA and ptxC), showed sufficient polymorphisms to differentiate B. pertussis isolates from The Netherlands, Finland, Italy, the USA and Japan into eight designated MLSTs.

The genes chosen for the present study included eight of the genes investigated by van Loo et al. (2002), namely, prnA, ptxA, ptxC, tcfA, fim2 (fimbrial antigen 2), ompQ (outer-membrane protein OmpQ), brkA (bordetella resistance to killing protein) and vag8 (virulence-activated gene 8), and two additional genes, cyaA (adenylate cyclase toxin) and bapC (bordetella autotransporter protein C). The latter gene sequence was originally submitted to GenBank as a putative autotransporter protein gene (accession no. AF081494) and named bap5 (Blackburn, 2000), but later submitted independently as bapC (AJ277634). The protein BapC, like pertactin, Vag8, BrkA and TcfA, is a member of the bordetella autotransporter family (Henderson & Nataro, 2001), and is therefore a potential virulence factor and protective antigen. In addition to its possible use as a component of future ACVs, adenylate cyclase toxin (CyaA), is being investigated for use in multipurpose vaccines, by exploiting its ability to deliver foreign epitopes to antigen-presenting cells (Ladant & Ullmann, 1999). Before using CyaA or BapC as vaccine components, their properties should be fully characterized, including their potential for antigenic variation.

The aims of the present study were (i) to determine the level of sequence variation or conservation in ten genes in historical and recent clinical isolates of B. pertussis and to compare these with UK vaccine strains, and (ii) to identify further potential genetic markers for the epidemiological surveillance of B. pertussis.

	METHODS

TOP INTRODUCTION METHODS RESULTS AND DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Bacterial strains and culture conditions.

All isolates were obtained from the Health Protection Agency, Respiratory and Systemic Infection Laboratory, London, and were cultured on charcoal agar with 10 % horse blood (Media Services, CPHL) at 37 °C for up to 5 days with 5 % CO₂. The origin of 285 UK isolates has been described previously (Fry et al., 2001). Numbers of isolates and years of isolation for additional UK isolates included in this study were as follows: 1977, 2; 1983, 1; 2000, 5; 2002, 42. The total number of isolates and years of isolation are shown in Table 1. All isolates in this study were assumed to be epidemiologically unrelated. The three strains (CN2992, CN5476 and CN3099) used in the preparation of the current UK pertussis whole-cell vaccine (Medeva Pharma), the Tohama-I strain (components of which are used in acellular pertussis vaccines in the UK and other countries) and the type strain of B. pertussis (18323^T = NCTC 10739^T), were also characterized.

Study design.

Sequence polymorphism was investigated in regions of ten genes using various subsets of the 335 B. pertussis isolates. Not all isolates were examined for all targets, and data from a previous study (Fry et al., 2001) were used to aid the selection of subsets of isolates containing representatives of all known UK pertactin gene alleles, prnA(1–3), and pertussis toxin S1 gene alleles, ptxA(1, 2). The number of isolates from each year analysed for each gene is shown in Table 1. Sequences of all ten gene targets were determined for the four vaccine strains and the type strain (18323^T). Pertactin and pertussis toxin S1 gene type and serotype were determined for the 50 isolates from 1997–2002. Only the major polymorphic region designated region 1 of the prnA gene was sequenced for these 50 isolates. Serotype and sequence data of the ptxA and prnA genes from the other UK clinical isolates and the three UK WCV strains were taken from a previous study (Fry et al., 2001).

Extraction of DNA.

Genomic DNA was extracted from B. pertussis cells grown on charcoal blood agar using the Nucleon BACC2 genomic DNA extraction kit (Amersham Biosciences) including an RNase A step. DNA concentration was determined using a GeneQuant II spectrophotometer (Amersham Biosciences) at A₂₆₀.

Gene nomenclature and numbering.

The description of the pertactin and pertussis toxin S1 gene variants, first described by Mooi et al. (1998, 1999, 2000), is as defined by Fry et al. (2001). Other gene variants described by van Loo et al. (2002) or novel variants found in this study follow the format of Fry et al. (2001), i.e. gene designation followed by allele number in parentheses. Throughout the text, the location of numbered positions on the genes is with respect to the GenBank or EMBL reference sequence indicated.

Serotyping.

Serotyping was performed with polyclonal antisera to agglutinogens 1, 2 and 3, namely B. pertussis anti-agglutinogen 1 (no. 89/596), B. pertussis anti-agglutinogen 2 (no. 89/598) and B. pertussis anti-agglutinogen 3 (no. 89/600) (National Institute for Biological Standards and Control, UK), in a slide-agglutination assay.

PCR amplification and sequencing.

PCR amplification and sequencing of prnA and ptxA was performed essentially as described by Mooi et al. (1998, 2000), with minor modifications (Fry et al., 2001). Amplification and sequencing of brkA, ptxC, tcfA and vag8 was as described by van Loo et al. (2002), with minor modifications using Montage PCR₉₆ filter plates (Millipore), the MultiScreen vacuum manifold (Millipore) and the CEQ 8000 genetic analysis system (Beckman Coulter). Nucleotide sequence was determined for the complete open reading frame of brkA (3033 bp) and the polymorphic sites of ptxC, vag8, tcfA, fim2 and ompQ reported by van Loo et al. (2002). Amplification and sequencing primers designed for this study are described in Table 2. Additional primers used were taken from previous studies (Mooi et al., 1998, 2000; Boursaux-Eude et al., 1999; van Loo et al., 2002). Primer sequences and PCR parameters for amplification and dideoxynucleotide sequencing of brkA, tcfA, vag8 and ptxC were kindly provided by F. Mooi, National Institute of Public Health and the Environment, Bilthoven, The Netherlands. The cyaA and bapC genes were also examined for variation by dideoxy sequencing. Oligonucleotide primers were designed from available sequences for cyaA (Y00545) (Glaser et al., 1988) and bapC (AF081494). For cyaA, five primer pairs were used to amplify approx. 330–620 bp products targeting bases 3586–6098 (i.e. 2512 bp in total), with respect to Y00545 (Table 2; Fig. 1). For bapC, five primer sets were used to amplify approx. 418–677 bp products, targeting bases 219–2548 (i.e. 2280 bp in total) (AF081494). For three genes, ptxC, fim2 and ompQ, the single nucleotide polymorphisms (SNPs) were taken to define these alleles (van Loo et al., 2002), and were determined by a real-time pyrophosphate DNA sequencing method named pyrosequencing (Ronaghi et al., 1998). For five isolates, the ptxC sequence type was determined initially by conventional dideoxy sequencing prior to pyrosequencing. The allele type designations by both methods were in complete agreement. Assignment of SNP alleles was made by comparison to those sequences previously described (van Loo et al., 2002).

View larger version (17K): [in this window] [in a new window]   Fig. 1. Schematic of the B. pertussis adenylate cyclase toxin (CyaA), showing the enzymic (AC) domain (amino acids 1–400), the hydrophobic membrane-spanning domain, the site of post-translational modification and the 42 calcium-binding nonapeptide repeats. Numbering represents amino acid residues. Arrows indicate the positions on the cyaA gene amplified by the five primer pairs (I–V) used to target the immunodominant region. Adapted from Ladant & Ullmann (1999) with permission.

All PCR mixtures contained 1.5 mM MgCl₂ and 200 µM of each deoxynucleotide in a final volume of 50 µl. In addition, mixtures designed to amplify cyaA and tcfA contained 10 mM Tris/HCl (pH 8.3), 50 mM KCl, 1.0 µM each primer (except for cyaA-2F and cyaA-2R, 0.1 µM), 5 % (v/v) DMSO (Sigma-Aldrich) and 1 U Taq DNA polymerase (Invitrogen). Other target reaction mixtures contained HotStarTaq DNA polymerase with Q-solution (Qiagen) and 1.0 µM each primer (Invitrogen), except for bapC-5F and bapC-5R (0.1 µM). Template DNA (approx. 100 ng) was added and reaction mixtures containing no added DNA served as negative controls. Amplification was performed using a DNA Engine (MJ Research) with the following conditions: initial denaturation for 10 min at 94 °C, then 30 cycles of denaturation for 2 min at 94 °C, annealing for 2 min at 50–60 °C, extension for 1 min at 72 °C and a final extension step of 10 min at 72 °C. The annealing temperatures were as follows: ptxC, 50 °C; ompQ, 55 °C; fim2, 58 °C; and brkA, vag8, tcfA, cyaA and bapC, 60 °C. Amplified products were purified using Montage PCR₉₆ filter plates (Millipore) and sequenced with the primers used for amplification. Nucleotide sequences determined by the dideoxynucleotide method used the dye terminator cycle sequencing quick start kit (Beckman Coulter) and the products were analysed on a CEQ 8000 genetic analysis system (Beckman Coulter).

Pyrosequencing was performed following the manufacturer's instructions (Pyrosequencing). Typically, 20 µl biotinylated PCR product from a 50 µl PCR was immobilized onto streptavidin-coated beads and converted to single-stranded DNA template by denaturation. After washing, removal of the unmodified strand and neutralization, the sequencing primer was annealed to the template. Samples were incubated at 80 °C for 2 min, allowed to cool to room temperature and analysed using the PSQ 96MA system (Pyrosequencing).

Sequence analysis.

Sequence analyses were performed using software packages BioNumerics and Kodon (Applied Maths). For dideoxynucleotide analyses, sequence data from forward and reverse primers were combined and aligned manually. New sequence data were aligned with the available sequences for B. pertussis from GenBank, cyaA, Y00545; tcfA(1), U16754; tcfA(2), AJ009785; tcfA(3), AJ420991; tcfA(4), AJ507643; tcfA(5), AJ420992; brkA, U12776; vag8(1), U90124; bapC(1), AF081494, and The Sanger Centre, Cambridge, UK (http://www. sanger.ac.uk/Projects/B_pertussis/). Novel alleles found in this study were designated by the authors.

Nucleotide sequence accession numbers.

The sequence comprising the complete open reading frame of the tracheal colonization factor gene (tcfA) from B. pertussis strain DCH154 isolated in 1983 was deposited in GenBank, under the accession number AY375533.

	RESULTS AND DISCUSSION

TOP INTRODUCTION METHODS RESULTS AND DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
Variation in the genes encoding two or more B. pertussis components, including the ACV components (prnA, ptxA-E, fhaB, fim2 and fim3), and other surface-associated proteins (tcfA, brkA, vag8 and ompQ), has been reported in isolates from The Netherlands, Finland, Italy, Japan, Poland and the USA (Mooi et al. 1998, 1999; Cassiday et al., 2000; van Loo et al., 2002; Gzyl et al., 2002). Genotypic variation in historical and recent UK isolates in two of these genes, ptxA and prnA, has been reported previously (Fry et al., 2001). In the present study, sequence variation was determined in regions of ten virulence-associated genes (prnA, ptxA, ptxC, tcfA, fim2, brkA, vag8, ompQ, bapC and cyaA) in a selection of historical and recent UK isolates of B. pertussis, together with the three strains used to produce the UK whole-cell pertussis vaccine, the strain used to produce components for the acellular pertussis vaccine (Tohama-I) and the type strain (18323^T) of B. pertussis (Tables 3 and 4). The vaccine and type strains were included as reference strains to help identify vaccine and non-vaccine allele types and to increase the probability of identifying polymorphic sites, as the type strain has been shown to be genotypically distinct from other B. pertussis strains (Musser et al., 1986; Gerlach et al., 2001). In B. pertussis isolates from the UK for the period 1920–2002, polymorphism was found in five genes, prnA, ptxA, ptxC, tcfA and fim2. In a selection of these isolates, no polymorphism was found in the brkA, bapC, cyaA, vag8 and ompQ genes, but differences were observed between the type strain and all other strains examined in all genes except brkA (Table 3).

Pertussis toxin S1 and S3 genes

Pertussis toxin S1 subunit gene (ptxA) types are defined by the number and type (synonymous and non-synonymous) of point mutations in defined regions of the gene. To date, six alleles have been described, ptxA(1–4) (Mooi et al., 2000) and ptxA(5–6) (AJ506994, AJ506995). Polymorphism within ptxA has been described in five major regions within an approximately 600 bp region (position 478–1062, AJ006155). The data for UK isolates for ptxA (Table 3) were mainly taken from a previous study (Fry et al., 2001) using primers described by Mooi et al. (1998). Fifty additional isolates were included in the present study, but no new alleles were found. Thus, all 324 UK isolates were ptxA(1) or ptxA(2). Temporal analysis of isolates revealed that ptxA(2) variants have not been seen in the UK since 1985 and all subsequent isolates have been ptxA(1). Two of the three whole-cell pertussis vaccine strains were ptxA(1), the other was ptxA(2); the ACV strain (Tohama-I) was ptxA(2), and the type strain was ptxA(4).

The pertussis toxin S3 subunit gene (ptxC) alleles are defined by an SNP at position 3625 (M13223, van Loo et al., 2002): ptxC(1) (TGC, Cys) and ptxC(2) (TGT, Cys). Both allele types were seen in the 138 UK isolates analysed (Table 3). However, there has been an apparent shift in the predominant allele of ptxC. From 1920 until 1998, all isolates (n = 53) were ptxC(1). In 1998, eight of 21 isolates were ptxC(2) and, in 1999, 14 of 30 isolates were ptxC(2). Of 34 isolates from 2000 to 2002, 33 were ptxC(2) and one was ptxC(1). The vaccine strains and the type strain were ptxC(1).

Tracheal colonization factor gene

Tracheal colonization factor is unique to B. pertussis, and has been reported to be important for colonization of the trachea in a mouse model of infection (Finn & Stevens, 1995). The nucleic acid sequence of the tcfA gene (between bases 688–1339, with respect to U16754) was determined for 138 UK clinical isolates, the four vaccine strains and the type strain (18323^T). The tcfA gene exhibited the greatest allelic variation of the ten genes investigated in this study, with five different alleles, and all of the changes were non-synonymous. Four previously described types, tcfA(2), tcfA(3), tcfA(4) and tcfA(5), were found amongst the UK B. pertussis population (Table 3; Fig. 2). A novel tcfA type, here designated tcfA(6), was found in two isolates from 1983, and contained a 15 bp deletion not seen in the other alleles (Fig. 2). To date, tcfA(1), which contains a 75 bp segment, has been described only in the type strain (18323^T) (Table 4; Fig. 2). In the alleles tcfA(2–4 and 6), this segment is missing and these alleles differ from each other in both nucleotide and amino acid sequence.

View larger version (15K): [in this window] [in a new window]   Fig. 2. Primary structure of the B. pertussis tracheal colonization factor gene (tcfA) showing polymorphism in the six types described to date, designated tcfA(1) to tcfA(5) (van Loo et al., 2002) and tcfA(6) (this study). Numbering is with respect to tcfA(1), U16754. Nucleotide polymorphisms are shown and associated amino acid changes are indicated beneath the relevant codon. Dots indicate identity, and codons are separated by dashes. In tcfA(1), [N]75 indicates a region of 75 bp that is absent from the remaining alleles. Deletions are indicated by X. The arrow below tcfA(5) indicates an inserted G, which changes the reading frame and results in premature translational termination two codons downstream. The novel allele tcfA(6) reported in this study contains a 15 bp deletion not found in tcfA(1–5).

Of 138 UK clinical isolates from 1920–2002 investigated for tcfA polymorphism, 124 (89.9 %) were tcfA(2), two (1.4 %) tcfA(3), nine (6.5 %) tcfA(4), one (0.7 %) tcfA(5) and two (1.4 %) tcfA(6). The isolates of type tcfA(3) were from a pertussis case in paediatric intensive care isolated in 1999 (Crowcroft et al., 2003) and from an unvaccinated individual in 1999. Allele tcfA(4) was found in isolates from 1970–1999 and tcfA(5) in an isolate from 1942. The two strains with the novel tcfA type, tcfA(6), were isolated in 1983. All of the isolates (n = 34) from the most recent period (2000–2002) were tcfA(2). All four vaccine strains were tcfA(2) and the type strain was confirmed as tcfA(1).

The predominant tcfA allele type in this study, tcfA(2), has been reported in isolates from The Netherlands, Finland, Italy, Japan and the USA, and this allele was predominant in all countries except Italy (van Loo et al., 2002). Of ten Italian B. pertussis isolates studied, six were tcfA(3) and four were tcfA(2). Type tcfA(3) was not present in the American or Japanese B. pertussis isolates. In The Netherlands, Finland and Italy, the allele tcfA(3) has been reported in isolates from the period 1990–1999 only. This allele is apparently also more dominant in The Netherlands than in the UK, in that 27 of 85 (31.8 %) Dutch B. pertussis isolates from 1990–1999 were tcfA(3), compared with two of 51 (4 %) UK B. pertussis isolates from the same period. Eight of nine UK tcfA(4) isolates were from 1970–1989, and the other was from 1999. The allele tcfA(4) has not been reported in B. pertussis isolates from The Netherlands, Finland, Italy, Japan or the USA, but has been seen in at least one isolate from Australia (AJ507643). The UK isolate that was tcfA(5) was unusual as this allele has previously been reported only in isolates from the USA from 1990–1999 (van Loo et al., 2002). It may be significant that the unusual alleles, tcfA(4) and tcfA(6), except for one isolate, were present in UK isolates from the late 1970s to the mid 1980s, following a drop in vaccine uptake that resulted in two epidemic peaks of pertussis in the UK around 1979 and 1983.

Temporal analysis of MLSTs of B. pertussis in the UK

Nine MLSTs (MLST-1 to -9), based on point mutations and the presence of insertions and deletions in the three genes ptxA, ptxC, and tcfA have been described by van Loo et al. (2002) (Table 5). These combined allelic profiles were used to analyse temporal trends in MLST frequencies in the UK (Fig. 3). As strains were not available from all years, time periods corresponding to six, approximately 10-year periods from 1920–1999, together with recent isolates, from 2000–2002, were used.

View larger version (25K): [in this window] [in a new window]   Fig. 3. Temporal trends in the frequency of MLST types in the UK B. pertussis population.

Seven distinct MLSTs were identified in 132 isolates selected from each of the different time periods. The earliest available isolate from 1920 was MLST-2, i.e. ptxA(2), ptxC(1), tcfA(2). Two MLSTs, MLST-2 and MLST-3 [ptxA(1), ptxC(1), tcfA(2)], appear to have been predominant prior to the widespread introduction of pertussis vaccination in the UK in the 1960s. Of the 23 isolates from the pre-vaccination era, 10 were MLST-2 and 12 were MLST-3. However, MLST-2 strains were not found after 1989 (Fig. 3). MLST-3 was found in all time periods studied. Nine isolates were MLST-6, found only from 1970–1999. Two isolates with novel MLSTs were identified, MLST-10, defined as ptxA(1), ptxC(1), tcfA(6), from 1983, and MLST-11, defined as ptxA(2), ptxC(1), tcfA(5), from 1942. MLST-4 [ptxA(1), ptxC(1), tcfA(3)], was seen only in two isolates from 1999. In 24 of the most recent isolates, from 2000–2002, only two MLSTs were found, MLST-5 (n = 23), detected only from 1990 onwards, and MLST-3 (n = 1). It is important to note that two of the three UK WCV strains (CN5476 and CN3099) were MLST-3 and the other (CN2992) was MLST-2. The Tohama-I strain (components of which are used in ACVs) was MLST-2. The type strain of B. pertussis was MLST-9.

Using a combination of defined alleles to describe isolates allows the ready comparison of datasets of different geographical origins and from different time periods. Analysis of 132 UK isolates revealed that there has been an apparent shift of sequence types from the pre-vaccination era to recent isolates (Fig. 3). For example, MLST-2 was prevalent in the earliest time period but has not been detected since 1989. MLST-3 was the predominant sequence type from 1920 until 1999, but this appears to have been replaced by MLST-5, although the difference between these MLSTs is conferred by a single silent nucleotide difference in the ptxC gene (van Loo et al., 2002). In the American and Dutch B. pertussis populations, the trends appear to have been different (van Loo et al., 2002). MLST-1 and MLST-2 isolates were present in the pre-vaccination era, with MLST-2 predominating. In addition, in the USA, MLST-9 was present in the pre-vaccination era, but neither MLST-9 nor MLST-1 has been identified amongst recent (1990–1999) isolates, although MLST-2 was present (10 % of isolates). In recent (1990–1999) Dutch isolates, MLST-1 and MLST-2 were present at low frequencies (1 and 4 %, respectively). In 1990–1999, MLST-3 was present in all three countries, and comprised 52 % of UK isolates, 10 % of American isolates and 25 % of Dutch isolates. MLST-4 was present in the UK and Dutch B. pertussis populations from 1990 until 1999 but was not found amongst American isolates from the same time period. The frequency of MLST-4 differed between UK and Dutch populations, being 4 and 32 %, respectively. MLST-5 was the predominant type in recent (1990–1999) B. pertussis populations from the USA, The Netherlands and the UK. MLST-7 and MLST-8 were found only in recent isolates from the USA, and at very low frequencies. Van Loo et al. (2002) noted that the resurgence of pertussis in The Netherlands since 1996 coincided with the appearance of the new MLSTs, MLST-4 and MLST-5. In the UK, these two types were detected from 1999 onwards, although MLST-5 was the predominant type in the most recent time period. MLST-5 was also the predominant type in The Netherlands in the most recent time period studied, 1996–1999, comprising 40 % of the isolates. In contrast to The Netherlands, however, there has been no resurgence of pertussis in the UK and therefore no evidence that vaccine efficacy has been compromised by change in the antigenic make-up of circulating strains.

MLST and pertactin gene types

A number of prnA alleles have been described, prnA(1–8) (Mooi et al., 2000), prnA(9) (AF218785) and prnA(11) (AJ507642). They differ by the number (n = 4 to 7) and the composition of the repeat units (GGxxP) found in the major polymorphic region (823–897, AJ011091), designated region 1, in another repeat region (PQP)_4–5 (1762–1827, AJ011091), designated region 2, or, in one type, prnA(7), a point mutation approx. 150 bases upstream of region 2. The prnA data presented in Table 3 were mainly taken from a previous study (Fry et al., 2001). Fifty additional isolates were included in the present study, but no new alleles were found. All 330 UK isolates of B. pertussis analysed were either prnA(1), prnA(2) or prnA(3). Of the 35 most recent (2000–2002) isolates, 31 were prnA(2), three were prnA(1) and one was prnA(3). The four pertussis vaccine strains were prnA(1) and the type strain was prnA(6).

A comparison of the MLSTs of UK isolates and their prnA types is shown in Fig. 4. MLST-2, -6, -10 and -11 isolates were all exclusively prnA(1). The MLST-3 isolates were prnA(1) prior to 1980 but, after 1980, were associated with three prnA types, prnA(1–3). Of the MLST-4 isolates, one was prnA(2) and the other was prnA(3). MLST-5 isolates, which predominate in recent years, were mainly prnA(2) (22 of 23 isolates), with one prnA(3). Two of the three UK pertussis WCV strains were MLST-3, prnA(1), and the other was MLST-2, prnA(1). The ACV strain was MLST-2, prnA(1), and the type strain was MLST-9, prnA(6).

View larger version (15K): [in this window] [in a new window]   Fig. 4. B. pertussis MLST frequencies in the UK from 1920 to 2002 showing distribution of pertactin gene type; prnA(1) (filled bars), prnA(2) (open bars), prnA(3) (hatched bars). The number of isolates of each MLST is indicated.

Polymorphism in the fim2 gene and relationship to serotype

Eighty UK clinical isolates were investigated for the SNP defining the fim2 alleles. Sixty (75 %) were fim2(1) and 20 (25 %) were fim2(2) (Table 3). All four vaccine strains and 18323^T were fim2(1) (Table 4). The allele fim2(2) was not found in isolates prior to 1998. This allele showed association with serotype 1,2 such that all fim2(2) isolates identified (n = 20) were serotype 1,2, although not all serotype 1,2 strains were fim2(2). Using a larger sample, it would be possible to investigate whether fim2(2) is a potential marker for virulence due to its association with serotype 1,2. Isolates of B. pertussis of serotype 1,2 have been linked to more severe cases of pertussis (van Buynder et al., 1999).

Adenylate cyclase toxin gene

Forty-two clinical UK isolates of B. pertussis, the four vaccine strains and 18323^T were investigated for polymorphism in the region of the cyaA gene encoding the immunodominant moiety (approx. 2.5 kb). Only two allelic types were detected. The allelic type of cyaA in strain 18323^T sequenced in this study was designated cyaA(1) (Fig. 5) and that of all the other isolates was designated cyaA(2). The two allele types differ by a single nucleotide at position 4403 (Y00545) corresponding to a synonymous change. The cyaA(2) sequence, which is also present in the genome sequence of Tohama-I strain (Parkhill et al., 2003), should perhaps be regarded as the prototype sequence. Comparison of the cyaA sequence from the type strain (18323^T = NCTC 10739^T) determined in this study with that of the same strain obtained from GenBank (Y00545) (Glaser et al., 1988) showed one nucleotide difference: C (this study) compared with G at position 3981 with respect to Y00545. In our study, sequencing from forward and reverse primers covering this region was performed in triplicate with concordant results and, therefore, the most likely explanation is that there is a sequencing error at position 3981 in the previously deposited sequence, Y00545.

View larger version (8K): [in this window] [in a new window]   Fig. 5. Alleles of genes encoding bordetella autotransporter protein C (bapC) and adenylate cyclase toxin gene (cyaA). Numbering is relative to the start of the GenBank accession numbers AF081494 (bapC) and Y00545 (cyaA). Dots indicate identity.

The lack of polymorphism in the region encoding the immunodominant moiety of B. pertussis CyaA is in contrast to strains of Bordetella bronchiseptica, where variation was reported in this region (Betsou et al., 1995b). There are also nucleotide differences between the cyaA sequences of Bordetella parapertussis and of B. pertussis, but the two available B. parapertussis cyaA sequences (strain 63.2, AJ249835; NCTC 13253, BX470249) are identical. The immunodominant region of the B. pertussis cyaA gene is extremely stable which suggests that antigenic variation within cyaA would be predicted to be extremely rare. This lack of variation in cyaA could perhaps be attributed to the toxin not being subjected to the same immune pressure as the other virulence factors, such as pertussis toxin S1 subunit (ptxA), pertactin (prnA) and tracheal colonization factor (tcfA). Adenylate cyclase toxin (CyaA) has been shown to be protective in mouse studies (Betsou et al., 1995a; Hormozi et al., 1999) and, in view of the stability of its immunodominant region, it may be suitable as a component in new generation ACVs.

Outer-membrane protein Q, virulence-activated gene 8, bordetella resistance to killing protein and bordetella autotransporter protein C

Polymorphism in ompQ was investigated in 42 UK isolates chosen from different time periods, in the four vaccine strains and in the type strain. The alleles of ompQ are defined by an SNP at position 1465 (U16266), ompQ(1) (CTG, Leu), found only in the type strain, and ompQ(2) (CCG, Pro), found to be present in all 42 UK isolates and the four vaccine strains (Tables 3 and 4). Nine isolates selected from each of the different time periods, the four vaccine strains and the type strain were investigated for variation in all, or part of, vag8, brkA and bapC. There was no sequence variation in vag8 between the UK isolates and the UK vaccine strains, but there were three single nucleotide differences in this gene between the type strain and the other strains. These differences occurred at positions 510, 782 and 2697 (U90124), vag8(1) (CGT, Arg; GTC, Val; CGC, Arg), in the type strain, and vag8(2) (CGC, Arg; GCC, Ala; CGT, Arg) in all other strains. In bapC, there was a single nucleotide difference at position 2159 (AF081494), bapC(1) (CCT, Pro) between all isolates, including the vaccine strains, and bapC(2) (CCC, Pro) in the type strain (Tables 3 and 4; Fig. 5). Similarly, no sequence variation was found in brkA between the UK isolates, the vaccine strains or the type strain and no variation was found in brkA in any of the strains examined by van Loo et al. (2002). As strain 18323^T is, in many respects, an atypical strain of B. pertussis, the ompQ(2) and vag8(2) sequences, which are also present in the whole-genome sequence of the strain Tohama-I (Parkhill et al., 2003), should perhaps be regarded as prototype sequences. With bapC, the sequence found in all isolates except the type strain and the sequence deposited in GenBank has already been designated allele bapC(1).

Vaccination and epidemiology

WCVs for pertussis have been in use since widespread vaccination began in the UK in the 1960s (Therre & Baron, 2000). However, the reactogenicity of these vaccines stimulated research into the various virulence factors expressed by B. pertussis and the protection that these proteins could provide if included in ACVs. Some countries have replaced WCVs with ACVs or included them as part of their vaccination programme as booster vaccines for adolescents and adults (Campins-Martí et al., 2001). The impact of the use of ACVs with their limited antigenic repertoire on the antigenic make-up of the circulating B. pertussis population is unknown at the present time. However, as use of ACVs becomes more widespread, it will be important to monitor the genotypic and phenotypic characteristics of B. pertussis isolates for significant changes.

Characterization of strains by nucleotide sequence analysis has allowed epidemiological tracking of shifts in the circulating populations of B. pertussis of several countries. The choice of targets for tracking clonal organisms such as B. pertussis remains a difficult one, but prnA, ptxC, tcfA and fim2 appear to offer useful data. In the UK, the absence of historical isolates of B. pertussis for which patient information (such as age, vaccination status, etc.) is available has hindered this choice. However, prospective studies to identify useful markers including those described here will focus on a combination of matched isolate and patient data.

Regarding the current B. pertussis MLST scheme, there is no guarantee that, for example, an MLST-5 isolate in the UK is genotypically or phenotypically the same as an MLST-5 isolate from another country. There may be differences in other gene sequences and/or gene expression. The MLST scheme used here has thus been limited by the number of genes chosen. Of the allele types described to date, there are a possible 72 MLSTs. To help define the genotype and presumptive phenotype of isolates, it would be more informative to include all other targets that have demonstrated sequence variation, such as prnA and fim2. The addition of the prnA and fim2 alleles described would increase the number of possible B. pertussis sequence types to 1440. The other genes that were investigated for sequence variation revealed no polymorphism in UK isolates, which precludes them from use as epidemiological markers.

	ACKNOWLEDGEMENTS

TOP INTRODUCTION METHODS RESULTS AND DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 
We gratefully acknowledge Professor F. Mooi and colleagues for provision of laboratory protocols and C. Arnold for help with Pyrosequencing. We also thank E. Miller and colleagues in the Immunization Division, Communicable Diseases Surveillance Centre for epidemiological information, S. Neal and J. Duncan for additional serotyping and gene typing data on UK B. pertussis isolates, J. Parkhill and colleagues at The Sanger Institute, Cambridge, UK for access to unpublished sequence data of B. pertussis and B. parapertussis and T. Harrison and R. George for constructive comments on this manuscript. E. P. was the recipient of a Medical Research Council collaborative studentship.

	REFERENCES

TOP INTRODUCTION METHODS RESULTS AND DISCUSSION ACKNOWLEDGEMENTS REFERENCES

 

Gzyl, A., Augustynowicz, E., van Loo, I. & lusarczyk, J. (2002). Temporal nucleotide changes in pertactin and pertussis toxin genes in Bordetella pertussis strains isolated from clinical cases in Poland. Vaccine 20, 299–303.