Laboratory for Infectious Diseases Research (LIO), National Institute of Public Health and Environment, 3720 BA Bilthoven, The Netherlands1
Eijkman Winkler Institute for Medical Microbiology, University Medical Centre Utrecht, Utrecht, The Netherlands2
Author for correspondence: Frits R. Mooi. Tel: +31 30 2743091. Fax: +31 30 2744449. e-mail: fr.mooi{at}rivm.nl
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Keywords: population structure, serotyping, gene polymorphism, fingerprint typing
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The Bordetella pertussis population changed significantly after the introduction of vaccination in a number of countries (Cassiday et al., 2000 ; Fry et al., 2001
; Gzyl et al., 2001
; Mastrantonio et al., 1999
; Mooi et al., 1999
, 1998
; Weber et al., 2001
). Antigenic divergence between clinical isolates and vaccine strains was observed, suggesting that adaptation of the B. pertussis population to vaccine-induced immunity may be an important factor in the resilience of B. pertussis against vaccination. DNA fingerprinting revealed two decreases in the genotypic diversity of the Dutch B. pertussis population in the periods 19501972 and 19821996, which were associated with the emergence of novel pertussis toxin and pertactin types, respectively (van Loo et al., 1999
). These types were antigenically distinct from the type(s) present in the pertussis vaccine, suggesting that the novel strains were escape variants.
Changes in the B. pertussis population in the 1950s, the period when pertussis vaccines were introduced, are particularly interesting as they may illustrate how pathogens adapt to mass vaccination within a short time span. In the period 19491972, changes in frequencies of alleles for the S1 subunit of pertussis toxin were observed (Mooi et al., 1998 ). Thus expansion of escape variants may have contributed to the observed decrease in genotypic diversity. However, only a limited number of isolates and genes encoding immunologically relevant proteins were analysed. Here we investigated changes in the B. pertussis population structure in this period in greater detail. Further, we looked for evidence that strain adaptation affected mortality.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
DNA sequencing.
PCR conditions for the amplification of ptxS1 and prn were described previously (Mooi et al., 1998 ). Amplification of the tcfA and fim2 genes and the 500 bp fragment of fhaB, which contained the polymorphic locus(van Loo et al., 2002
) was performed in 20 µl, containing 1 µl DNA and 19 µl buffer comprising 50% HotstarTaq Master mix (Qiagen), 1 µM each primer and 5% DMSO. The tcfA gene from five isolates was sequenced completely. Two polymorphic regions were identified between bases 1 and 945, which were amplified with the primers tcfAF (5'-TTCTTGCGCGTCGTGTCTTC-3') and tcfAR3 (5'-GCGGTTGCGGACCTTCAT-3'). A region composed of bases 395945 was sequenced for all isolates with primers tcfAR3 and tcfAF2 (5'-TCGTCTGGCGGACATAACCC-3'). As polymorphism in the second region, composed of bases 1395, appeared to be linked to polymorphism in region 395945 (i.e. it was non-informative), it was sequenced for one out of five Dutch isolates with primer tcfAF. Primers used for amplification and sequencing of fim2 were fim2for (3'-GCGCCGGGCCCTGCATGCAC-5') and fim2rev (3'-GGGGGGTTGGCGATTTCCAGTTTCTC-5'). For fhaB primers fha9F (3'-ATCTCCGGCGAGGGG-5') and fha4450R (3'-CGGTGTAATCGCCCTGTATC-5') were used for amplification, and primers fha39R (3'-GATTGGCGTGCAGCGAGTTC-5') and fha41R (3'-GGCTAGCGCTGGCTCTGTCC-5') were used for sequencing. PCR amplification was performed by preheating the samples for 15 min at 95 °C. Subsequent amplification of tcfA was performed by heating the samples for 15 s at 95 °C, initial annealing at 72 °C for 15 s and elongation for 1 min at 72 °C. The annealing temperature was decreased every two cycles by 2 °C until 62 °C. Amplification of fim2 was performed by heating the samples for 15 s at 95 °C, initial annealing at 67 °C for 15 s and elongation for 1 min at 72 °C. The annealing temperature was decreased every cycle by 1 °C until 62 °C. Amplification of fhaB was performed by heating the samples 15 s at 95 °C, initial annealing at 65 °C for 15 s and elongation for 4 min at 72 °C. The annealing temperature was decreased every two cycles by 2 °C until 55 °C. The cycle with the final annealing temperature was repeated 25 times and a final elongation was performed for 10 min at 72 °C in all cases. PCR fragments were purified with Qiaquick (Qiagen). Sequencing reactions were performed with an ABI Prism Big Dye terminator reaction kit and the reactions were analysed with a model 377 or 3700 ABI DNA Sequencer (Perkin-Elmer Applied Biosystems).
Serotyping.
On a grease-free slide, bacterial colonies were emulsified in a drop of serum that contained Fim2 or Fim3 polyclonal antibodies, after which agglutination was determined macroscopically. To examine autoagglutination the drop of serum was replaced by physiological saline.
IS1002 DNA fingerprinting.
IS1002-based DNA fingerprinting was performed as described previously by digesting chromosomal DNA with SmaI (van Loo et al., 1999 ). The samples were analysed on agarose gels (Pulse Field Certified; Bio-Rad), and transferred to Hybond N+ membrane (Amersham Pharmacia Biotech) using standard DNA blotting techniques. A 293 bp IS1002 probe was used for hybridization. Labelling of the probe with peroxidase and detection of hybridizing bands was performed according to the instructions of the Enhanced Chemiluminescence Gene Detection System (Amersham Pharmacia Biotech). The exposed films were scanned at 190 d.p.i. (HP Scanjet IIcx/T; Hewlett Packard). The films were analysed using the Bionumerics software (Applied Maths). Assignment of IS1002 fingerprint patters to similarity groups was performed by the calculation of pairwise similarities using the Dice coefficient and cluster analysis with the UPGMA algorithm.
Genotypic diversity.
This was calculated with the following equation: genotypic diversity=(n/n-1)(1-xi2), where n is the number of isolates and xi is the frequency of ith fingerprint type (Nei & Tajima, 1981
).
Statistical analyses.
The statistical significance of the ptxS1 frequencies and differences in fimbrial serotypes and fingerprint types between periods were calculated by applying the 2 test. To comply with the conditions for the
2 test the analysis was performed on the summed fingerprint types of the similarity groups. The significance of the genotypic diversity was estimated as follows. By computer simulation we randomly generated 1000 cross-tabulations similar to Table 2, under the null hypothesis of no differences in frequency of fingerprint types between periods or ptxS1 allele groups. Under that hypothesis, fingerprint types essentially occur at random in the different periods or in ptxS1 groups. We then counted the number of times the squared differences in genotypic diversity in the simulated tables exceeded the observed squared difference and used these counts (divided by 1000) as an estimate of the P value of the difference in genotypic diversity.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
IS1002-based DNA fingerprinting
Isolates for which the ptxS1 allele was identified were also analysed by IS1002-based DNA fingerprinting. Twenty-one fingerprint types were observed, which were grouped into six (AF) similarity groups (Fig. 2). To observe temporal trends, isolates were stratified into three periods: 19491952, 19531958, 19651972 (Table 2
). The first two periods did not significantly differ in fingerprint frequencies (P=0·507). In these two periods most isolates were found in similarity group B (37% and 35%, respectively) and similarity group E (33% and 42%, respectively). The genotypic diversity of 0·87 and 0·90 in these two periods did not differ significantly (P=0·531). Although the first two periods were very similar, there was a significant difference in fingerprint types and frequencies between the periods 19531958 and 19651972 (P<0·0001). In the period 19651972, most (65%) fingerprint types were found in similarity group A, while in the two previous periods 9% and 0% of the fingerprint types were found in this similarity group. Further, in this period 50% of the isolates were associated with one fingerprint type, ft29, a type that was not detected previously. The expansion of ft29 strains was reflected in the genotypic diversity, which decreased to 0·73 in the period 19651972 (P=0·004, 19531958 vs 19651972).
|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Of the 15 investigated genes, four were found to be polymorphic in the investigated period (fim2, fhaB, prn and ptxS1). Three allelic combinations were observed in the pre-vaccination period and the period 19531958: [fim2-2 fhaB2 prn7 ptxS1D], [fim2-1 fhaB1 prn1 ptxS1B] and [fim2-1 fhaB1 prn10 ptxS1E]. The first two combinations were present in the two vaccine strains, used since the early 1960s, while the third combination of alleles was observed in one isolate only. In the 1960s, only one allelic combination from the pre-vaccination period was observed [prn1 fim2-1 fhaB1 ptxS1B], and a novel, fourth, allelic combination was found [prn1 fim2-1 fhaB1 ptxS1A]. Thus, the only mismatch we detected in the 1960s between the vaccine strains and circulating strains was with respect to the PtxS1 subunit, suggesting that it played an important role in driving the observed changes in the B. pertussis population.
In the first period after the introduction of vaccination (19531958), no significant changes in the frequencies of fingerprint types and the alleles investigated were observed (Tables 1 and 2
). However, changes in frequencies of fimbrial serotypes occurred. The pre-vaccination B. pertussis population was characterized by equal frequencies of Fim2 and Fim3 isolates (29%), while Fim2,3 isolates were found at a higher frequency (43%) (Table 1
). Similar frequencies were found in other countries in the pre-vaccination period (Bronne-Shanbury et al., 1976
; Eldering et al., 1969
). In the period 19531958 a decrease in Fim3 isolates was observed to 8%, possibly implicating that the vaccine contained a Fim3 strain in this period.
In the second period after the introduction of vaccination (19651972), significant changes in frequencies of ptxS1 alleles, fingerprint types and fimbrial serotypes were observed (Tables 1 and 2
). The ptxS1A allele, not previously detected, was found in 89% of the isolates, while ptxS1B was observed in 29% of the isolates. The ptxS1D allele was not detected in this period. Further, fimbrial serotype 3 increased from 8% to 92%. The most notable change in fingerprint types was the emergence of ft29, associated with 50% of the isolates. The observed changes in the B. pertussis population in the second period after the introduction of vaccination were much more dramatic compared to the first period. Mortality was 543-fold and fivefold lower in the periods 19651972 and 19531958, respectively, compared to the pre-vaccination period (Cohen, 1963
). This suggested that vaccination was more effective in the second period and consequently had a greater effect on the B. pertussis population. Further, the decrease in mortality indicates that, although vaccination affected the competitive balance between strains resulting in the expansion of escape variants, the total circulation of B. pertussis was decreased.
Isolates with ptxS1D disappeared before ptxS1B isolates (Table 1). This may have been due to the fact that early vaccine strains contained the ptxS1D allele (as indeed strain 509 does), while the second strain, 134 (with ptxS1B), was added to the vaccine in the 1960s (Cohen, 1963
). It is also possible that the ptxS1B allele confers a higher degree of fitness on strains compared to ptxS1D. Consistent with this, ptxS1D was not detected in isolates from the pre-vaccination era in the United Kingdom and detected in low frequencies (14%) in the United States (Fry et al., 2001
; Cassiday et al., 2000
). In the same period, ptxS1B was found at frequencies of 50% and 81%, respectively, in these countries. Although ptxS1A was not detected in the Dutch pre-vaccination population, we presume that it was present at low frequencies, as it was found in the pre-vaccination period in both the United Kingdom and the United States (frequencies 50% and 5%, respectively) (Fry et al., 2001
; Cassiday et al., 2000
). Strains with ptxS1A predominate in many countries with a high vaccination coverage (Cassiday et al., 2000
; Fry et al., 2001
; Gzyl et al., 2001
; Mastrantonio et al., 1999
; Mooi et al., 1999
, 1998
; Weber et al., 2001
). Most likely, ptxS1A strains are less affected by immunity induced by the vaccine strains, which harbour ptxS1B and ptxS1D. However, as yet we cannot exclude the possibility that the emergence of ptxS1A strains was (also) caused by other (unknown) loci, which increased strain fitness and which were linked to ptxS1A.
The rise to predominance of Fim3 subsequent to the introduction of vaccines with both fimbrial serotypes (in the 1960s) was observed not only in The Netherlands, but also in other countries (Blaskett et al., 1971 ; Bronne-Shanbury et al., 1976
; Eldering et al., 1969
; Preston, 1976
). It has been suggested that this phenomenon is due to the fact that Fim3 is less immunogenic than Fim2 (Preston, 1976
; Preston & Carter, 1992
). Since the rise of Fim3 frequency coincides with the rise in ptxS1A frequency, it is also conceivable that the Fim3ptxS1A combination has a higher fitness than other Fim combinations with ptxS1A. It is, however, unlikely that fimbriae played a role in the expansion of particular clones. B. pertussis contains both fimbrial genes, which are switched on or off randomly by insertions or deletions in a homopolymeric C-tract (Willems et al., 1990
). Thus strains can switch between fimbrial serotypes with relatively high frequency (Robinson et al., 1989
).
In general, there was congruence between clustering based on fingerprint type and ptxS1 type (Table 2). Two exceptions were observed, isolates with ptxS1D and ptxS1A were found in similarity groups A and B, in which ptxS1A and ptxS1B predominated, respectively. Especially the presence of ptxS1D in similarity group A was striking, as this allele was normally found in similarity group E, which showed a deep branching point with similarity group A. This observation could be explained by horizontal transfer of ptxS1D to a similarity group A strain, or chromosomal rearrangements resulting in a change in fingerprint type. The ptxS1D allele is generally found associated with fim2-2, fhaB2 and prn7, while ptxS1A was linked to fim2-1, fhaB1 and prn1. As the ptxS1D isolate found in similarity group A harboured fim2-2, fhaB2 and prn7, it seems likely that its close relationship with ptxS1A isolates is due to chromosomal rearrangements. Weber et al. (2001)
and Cassiday et al. (2000)
studied the relationship between B. pertussis isolates with PFGE. In contrast to our observation, they did not find congruence between clustering based on PFGE and ptxS1 type. However, they did observe an association between PFGE type and prn alleles. As changes in prn alleles occurred more recently compared to ptxS1 (Mooi et al., 1998
), this may suggest that fingerprinting based on IS1002-based and PFGE reveal slower and faster evolutionary clocks, respectively.
The genotypic diversity as determined by DNA fingerprinting decreased significantly from 0·90 to 0·73 in the periods 19531958 and 19651972, respectively. This may have been caused by an evolutionary bottleneck and/or by clonal expansion of particular strains, possibilities that are both consistent with an effect resulting from the introduction of a vaccine. In the period 19651972, 50% of the isolates belonged to one fingerprint type (ft29), and all of these isolates contained the ptxS1A allele, indicating that clonal expansion had occurred. The 114-fold decrease in mortality observed after the period 19531958 suggested a very significant decrease in circulation of B. pertussis, and thus an evolutionary bottleneck. Thus the balance of evidence suggests that the introduction of vaccination resulted in an evolutionary bottleneck and clonal expansion of strains harbouring PtxS1A. Strains with PtxS1A may have been introduced by import or mutation, or may have existed at a low frequency before the introduction of vaccination. Although we favour the latter possibility, in all cases the competitive balance between the strains was changed, most likely due to the introduction of vaccination, resulting in a significantly higher fitness of PtxS1A strains compared to PtxS1B and PtxS1D strains.
Although the differences between the ptxS1 alleles are small, it seems likely that they affect strain fitness (Fig. 1). All of the polymorphisms observed in ptxS1 were non-synonymous, which was consistent with the emergence of these alleles as a consequence of positive immune selection. This is also suggested by the observation that one of the polymorphic sites of the PtxS1 subunit has been implicated in binding to the T-cell receptor (De Magistris et al., 1989
; Scarselli et al., 1998
). Hausman & Burns (2000)
did not find a difference in the ability of antibodies raised with an acellular vaccine to neutralize pertussis toxin variants derived from B. pertussis and Bordetella bronchiseptica. However, the effect of variation in PtxS1 on immunological memory and cellular immunity has not been studied. The effect of variation of PtxS1 on fitness is probably small and may be difficult to study in vitro or in animal models.
Although in The Netherlands mortality due to pertussis decreased in the 1950s and 1960s, an increase was observed in the 1990s (data not shown). We have observed potentially adaptive mutations in three B. pertussis genes since the introduction of vaccination: ptxS1, tcfA and prn (van Loo et al., 2002 ). Mutations in prn have been shown to affect efficacy of a whole-cell vaccine in the mouse model (King et al., 2001
). Together, these changes may act synergistically and reduce efficacy of pertussis whole-cell vaccines.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Bass, J. W. & Wittler, R. R. (1994). Return of epidemic pertussis in the United States. Pediatr Infect Dis J 13, 343-345.[Medline]
Blaskett, A. C., Gulasekharam, J. & Fulton, L. C. (1971). The occurrence of Bordetella pertussis serotypes in Australia, 19501970. Med J Aust 1, 781-784.[Medline]
Brennan, M., Strebel, P., George, H., Yih, W. K., Tachdjian, R., Lett, S. M., Cassiday, P., Sanden, G. & Wharton, M. (2000). Evidence for transmission of pertussis in schools, Massachusetts, 1996: epidemiologic data supported by pulsed-field gel electrophoresis studies. J Infect Dis 181, 210-215.[Medline]
Bronne-Shanbury, C. J., Miller, D. & Stanfast, A. F. B. (1976). The serotypes of Bordetella pertussis isolated in Great Britain between 1941 and 1968 and a comparison with the serotypes observed in other countries over this period. J Hyg 76, 265-275.
Cassiday, P., Sanden, G., Heuvelman, C., Mooi, F., Bisgard, K. M. & Popovic, T. (2000). Polymorphism in Bordetella pertussis pertactin and pertussis toxin virulence factors in the United States, 19351999. J Infect Dis 182, 1402-1408.[Medline]
Cohen, H. H. (1958). Establishment of a dried standard pertussis vaccine. Antonie Leeuwenhoek 24, 33-48.[Medline]
Cohen, H. H. (1963). Development of pertussis vaccine production and control in the National Institute of Public Health in The Netherlands during the years 19501962. Antonie Leeuwenhoek 29, 183-201.[Medline]
Cohen, H. H. & Leppink, G. J. (1956). Selection of Haemophilus pertussis strains for vaccine production in the mouse protection test in a balanced design. J Immunol 77, 299-304.[Medline]
De Magistris, M. T., Romano, M., Bartoloni, A., Rappuoli, R. & Tagliabue, A. (1989). Human T cell clones define S1 subunit as the most immunogenic moiety of pertussis toxin and determine its epitope map. J Exp Med 169, 1519-1532.[Abstract]
de Melker, H. E., Conyn van Spaendonck, M. A., Rumke, H. C., van Wijngaarden, J. K., Mooi, F. R. & Schellekens, J. F. (1997). Pertussis in The Netherlands: an outbreak despite high levels of immunization with whole-cell vaccine. Emerg Infect Dis 3, 175-178.[Medline]
de Melker, H. E., Schellekens, J. F., Neppelenbroek, S. E., Mooi, F. R., Rumke, H. C. & Conyn-van Spaendonck, M. A. (2000). Re-emergence of pertussis in the highly vaccinated population of The Netherlands: observations on surveillance data. Emerg Infect Dis 6, 348-357.[Medline]
De Serres, G., Boulianne, N., Douville Fradet, M. & Duval, B. (1995). Pertussis in Quebec: ongoing epidemic since the late 1980s. Can Commun Dis Rep 21, 45-48.[Medline]
Eldering, G., Holwerda, J., Davis, A. & Baker, J. (1969). Bordetella pertussis serotypes in the United States. Appl Microbiol 18, 618-621.[Medline]
Fry, N. K., Neal, S., Harrison, T. G., Miller, E., Matthews, R. & George, R. C. (2001). Genotypic variation in the Bordetella pertussis virulence factors pertactin and pertussis toxin in historical and recent clinical isolates in the United Kingdom. Infect Immun 69, 5520-5528.
Guris, D., Strebel, P. M., Bardenheier, B., Brennan, M., Tachdjian, R., Finch, E., Wharton, M. & Livengood, J. R. (1999). Changing epidemiology of pertussis in the United States: increasing reported incidence among adolescents and adults, 19901996. Clin Infect Dis 28, 1230-1237.[Medline]
Gzyl, A., Augustynowicz, E., van Loo, I. & Slusarczyk, J. (2001). Temporal nucleotide changes in pertactin and pertussis toxin genes in Bordetella pertussis strains isolated from clinical cases in Poland. Vaccine 20, 299-303.[Medline]
Hausman, S. Z. & Burns, D. L. (2000). Use of pertussis toxin encoded by ptx genes from Bordetella bronchiseptica to model the effects of antigenic drift of pertussis toxin on antibody neutralization. Infect Immun 68, 3763-3767.
King, A. J., Berbers, G., van Oirschot, H. F. L. M., Hoogerhout, P., Knipping, K. & Mooi, F. R. (2001). Role of the polymorphic region 1 of the Bordetella pertussis protein pertactin in immunity. Microbiology 147, 2885-2895.
Mastrantonio, P., Spigaglia, P., van Oirschot, H., van der Heide, H. G. J., Heuvelman, C., Stefanelli, P. & Mooi, F. R. (1999). Antigenic variants in Bordetella pertussis strains isolated from vaccinated and unvaccinated children. Microbiology 145, 2069-2075.[Abstract]
Mooi, F. R., van Oirschot, H., Heuvelman, C., van der Heide, H. G. J., Gaastra, W. & Willems, R. J. L. (1998). Polymorphism in the Bordetella pertussis virulence factors P.69/pertactin and pertussis toxin in The Netherlands: temporal trends and evidence for vaccine-driven evolution. Infect Immun 66, 670-675.
Mooi, F. R., He, Q., van Oirschot, H. & Mertsola, J. (1999). Variation in the Bordetella pertussis virulence factors pertussis toxin and pertactin in vaccine strains and clinical isolates in Finland. Infect Immun 67, 3133-3134.
Nei, M. & Tajima, F. (1981). DNA polymorphism detectable by restriction endonucleases. Genetics 97, 145-163.
Preston, N. W. (1976). Prevalent serotypes of Bordetella pertussis in non-vaccinated communities. J Hyg 77, 85-91.
Preston, N. W. & Carter, E. J. (1992). Serotype specificity of vaccine-induced immunity to pertussis. Commun Dis Rep CDR Rev 2, R155-R156.[Medline]
Robinson, A., Gorringe, A. R., Funnell, S. G. P. & Fernandez, M. (1989). Stereospecific protection in mice against intranasal infection with Bordetella pertussis. Vaccine 7, 321-324.[Medline]
Scarselli, M., Esposito, G., De Magistris, M. T., Domenighini, M., Rappuoli, R., Burroni, G., Bernini, A. & Niccolai, N. (1998). NMR studies on the structure/function correlations of T-cell-epitope analogs from pertussis toxin. Eur J Biochem 254, 313-317.[Abstract]
Senzilet, L. D., Halperin, S. A., Spika, J. S., Alagaratnam, M., Morris, A. & Smith, B. (2001). Pertussis is a frequent cause of prolonged cough illness in adults and adolescents. Clin Infect Dis 32, 1691-1697.[Medline]
Strebel, P., Nordin, J., Edwards, K., Hunt, J., Besser, J., Burns, S., Amundson, G., Baughman, A. & Wattigney, W. (2001). Population-based incidence of pertussis among adolescents and adults, Minnesota, 19951996. J Infect Dis 183, 1353-1359.[Medline]
van Loo, I. H., van der Heide, H. G., Nagelkerke, N. J., Verhoef, J. & Mooi, F. R. (1999). Temporal trends in the population structure of Bordetella pertussis during 19491996 in a highly vaccinated population. J Infect Dis 179, 915-923.[Medline]
van Loo, I. H. M, Heuvelman, C. J., King, A. J. & Mooi, F. R. (2002). Multilocus sequence typing of Bordetella pertussis based on surface protein genes. J. Clin Microbiol 40, 1994-2001.
Weber, C., Boursaux-Eude, C., Coralie, G., Caro, V. & Guiso, N. (2001). Polymorphism of Bordetella pertussis isolates circulating for the last 10 years in France, where a single effective whole-cell vaccine has been used for more than 30 years. J Clin Microbiol 39, 4396-4403.
Willems, R., Paul, A., van der Heide, H. G., ter Avest, A. R. & Mooi, F. R. (1990). Fimbrial phase variation in Bordetella pertussis: a novel mechanism for transcriptional regulation. EMBO J 9, 2803-2809.[Abstract]
Received 6 December 2001;
revised 26 March 2002;
accepted 2 April 2002.