Evolution of French Bordetella pertussis and Bordetella parapertussis isolates: increase of Bordetellae not expressing pertactin


  • N. Hegerle,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • A.-S. Paris,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • D. Brun,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • G. Dore,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • E. Njamkepo,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • S. Guillot,

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author
  • N. Guiso

    1.  Molecular Prevention and Therapy of Human Diseases, Institut Pasteur, National Centre of Reference of whooping cough and other bordetelloses, Paris
    2.  URA-CNRS3012, Paris, France
    Search for more papers by this author

Corresponding author: N. Guiso, Institut Pasteur, Molecular Prevention and Therapy of Human Diseases Unit, 25 rue du Dr Roux, 75724 Paris Cedex 15, France
E-mail: nicole.guiso@pasteur.fr


Clin Microbiol Infect 2012; 18: E340–E346


Bordetella pertussis and Bordetella parapertussis are closely related bacterial agents of whooping cough. Whole-cell pertussis (wP) vaccine was introduced in France in 1959. Acellular pertussis (aP) vaccine was introduced in 1998 as an adolescent booster and was rapidly generalized to the whole population, changing herd immunity by specifically targeting the virulence of the bacteria. We performed a temporal analysis of all French B. pertussis and B. parapertussis isolates collected since 2000 under aP vaccine pressure, using pulsed-field gel electrophoresis (PFGE), genotyping and detection of expression of virulence factors. Particular isolates were selected according to their different phenotype and PFGE type and their characteristics were analysed using the murine model of respiratory infection and in vitro cell cytotoxic assay. Since the introduction of the aP vaccines there has been a steady increase in the number of B. pertussis and B. parapertussis isolates collected that are lacking expression of pertactin. These isolates seem to be as virulent as those expressing all virulence factors according to animal and cellular models of infection. Whereas wP vaccine-induced immunity led to a monomorphic population of B. pertussis, aP vaccine-induced immunity enabled the number of circulating B. pertussis and B. parapertussis isolates not expressing virulence factors to increase, sustaining our previous hypothesis.


The isolation of the aetiological agent of whooping cough, Bordetella pertussis, allowed the development of whole-cell pertussis (wP) vaccines [1]. As early as 1959, wP vaccine was intensively used in France for primary vaccination of infants at 3–5 months of age and for the first booster at 24 months. This vaccination programme resulted in a dramatic decrease in the incidence of pertussis among young children but led, during the next 20 years, to adolescents and adults becoming a new threat for newborns not yet vaccinated and who required hospitalization after being infected by their elders [2–4]. As a consequence, in 1995, the French authorities changed the age of primary vaccination with a wP vaccine to 2 months and the age for the first booster to 16–18 months. Despite the efficacy of the French wP vaccine, its side effects encouraged the development of subunit vaccines that were equally efficacious but devoid of undesirable effects. Acellular pertussis (aP) vaccines contain only a few detoxified bacterial proteins and were introduced for adolescent booster vaccination in France in 1998 [5]. Use of aP vaccines was rapidly generalized and wP vaccine was retired from the French market in 2002.

We previously observed that the herd immunity resulting from wP vaccination had been highly effective in controlling isolates similar to those used to develop the wP vaccines [6,7]. First genomic analysis showed that the B. pertussis isolates currently circulating in France have lost genetic material and possess a higher number of insertion sequence elements [8]. Similar genomic changes have been observed in other European regions [9–12] but not in Bordetella parapertussis isolates [13]. Hence, the generalized use of wP vaccine revealed polymorphisms within the B. pertussis population because it was not able to control all the circulating isolates [13]. The wP vaccine-induced herd immunity enabled isolates, as virulent as but different from the vaccine strains, to emerge [3].

Unlike wP, the aP vaccine includes only a small number of detoxified bacterial proteins involved in pathogenicity. All aP vaccines contain detoxified pertussis toxin (PT) in addition to either filamentous haemagglutinin (FHA); or FHA and pertactin (PRN); or FHA, PRN and fimbrial proteins 2 and 3 (FIM2 and FIM3) [14]. We have proposed that aP vaccine-induced immunity would cause the number of circulating B. pertussis isolates lacking the expression of one or more vaccine antigens to increase [14]. It was thought that it should remain ineffective against B. parapertussis [15–17] although some efficacy of aP vaccine was observed in the field [18]. Since 2005, 7 years after France began using aP vaccines, surveillance has revealed the appearance of B. pertussis and B. parapertussis isolates lacking expression of vaccine antigens [19,20].

Here, we first analysed all B. pertussis and B. parapertussis isolates collected since 2000 using pulsed-field gel electrophoresis (PFGE) and expression of virulence factors. We show a steady increase in the number of B. pertussis as well as B. parapertussis isolates lacking expression of PRN. Then, several isolates belonging to the pre-wP, post-wP or post-aP vaccine eras were compared using models of in vivo and in vitro infection.

Materials and Methods

Isolates and growth conditions

Isolates belonging to the pre-vaccine era are part of the collection of the Institut Pasteur. Since 1993 the National Centre of Reference of whooping cough and other bordetelloses has been integrated into our laboratory. The National Centre of Reference isolated bacteria during epidemiological studies [2,3,21] and has collected, identified and archived the bacteria isolated by microbiologists from 43 paediatric hospitals since 1996 [3]. Hence, 1030 B. pertussis and 49 B. parapertussis isolates were identified and analysed since 2000. For murine respiratory infections and murine macrophage cytotoxicity assays, isolates from the French pre-wP (5/23), post-wP (4/878) or post-aP (7/1030) vaccine eras were selected on the basis of the phenotypic and genetic differences listed in Table 1. Bacteria were grown on Bordet–Gengou agar (Difco, Franklin Lakes, NJ, USA) supplemented with 15% defibrinated sheep blood at 36°C for 72 h and plated again for 24 h before each experiment.

Table 1.   Characteristics of selected Bordetella pertussis and Bordetella parapertussis isolates
IsolatesYearEraWestern analysisaFIMbPFGE ptxP prn ptxARD1RD2RD4LD50Reference
  1. AC-Hly, adenylate cyclase haemolysin; aP, acellular pertussis; CIP, Institut Pasteur; FHA, filamentous haemagglutinin; FIM, fimbrial protein; LD50, median lethal dose; NA, non applicable; ND, undetermined; PFGE, pulsed-field gel electrophoresis; PRN, pertactin; PT, pertussis toxin; RD, region of difference; wP, whole-cell pertussis.

  2. aWestern analyses were conducted using specific anti-FHA, anti-PRN, anti-PT and anti-AC-Hly antisera. If not specified, the considered isolate expresses all tested virulence factors and is designated by +. Question marks (?) are used when reasons of non-expression are unknown. Δ stands for deletion of the genes and IS for insertion of an Insertion Sequence inside the gene of interest.

  3. bNumbers in the ‘FIM’ column stand for the FIM phenotype: 2 = FIM2, 3 = FIM3 and 2/3 = FIM2 and FIM3.

  4. cNL, non-lethal in the mouse model of respiratory infection. LD50 is given in bacteria by mouse.

CIP16721950Pre-vaccine eraFHA (?)2II111+8 × 107This study
CIP30771950+3II114+2 × 107This study
CIP54561953+3II214+++1 × 107This study
Tohama1954+2II112+++8 × 107[19]
Bp14141959+2/3II114+++8 × 106This study
Bp14161959+3III112+3 × 107This study
BpHav1993wP vaccine era+3IVα121+3 × 107This study
FR1451995+3III312+5 × 107This study
Bp2871996+3V131+8 × 107This study
Bp7431999+3IVβ3212 × 108This study
FR34692005aP vaccine eraPT− (?)3IVγ321NLcThis study
FR36932007PRN− (IS)3IVα3216 × 107[19]
FR37132007+3IVγ3212 × 107This study
FR37492007PT− (Δ)3IVγ2NL[19]
FR37932007PRN− (Δ)3IVα3212 × 108[19]
FR46152009+2IVγ3214 × 107This study
FR46242009FHA− (?)/PRN− (IS)3IVβ317 × 107This study
Bpp 63-21963wP vaccine era+/BppIND1.4a-2.9bNDNANANA8 × 107This study
Bpp 11990+/BppIND1.4a-2.9bNDNANANA1 × 108[20]
Bpp 128221993+/BppIND1.4a-2.9bNDNANANA1 × 108[20]
FR 37282007aP vaccine eraPRN− (ΔG)/BppIIND1.4a-2.9bNDNANANA1 × 108[20]
FR 37432007PRN− (ΔG)/BppIIND1.4a-2.9bNDNANANA3 × 107This study
FR47042010PRN− (ΔG)/BppIIND1.4a-2.9bNDNANANA5 × 107This study

Pulsed-field gel electrophoresis

DNA fingerprint by PFGE was performed as previously described on all isolates [22].

DNA extraction and genotyping

For genotyping, bacterial DNA extractions were performed using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s recommendations. Genotyping of the repeated regions I and II of the gene encoding PRN, the S1 subunit of PT and the ptx promoter (ptxP) were performed as described elsewhere [23]. A study of genetic evolution of B. pertussis isolates was also conducted by detection of regions of differences (RD): RD1 (BP0911–BP0937), RD2 (BP1135–BP1141) and RD4 (BP1948–BP1966) as previously described [8], which were shown to disappear over time after wP vaccine introduction.

Western blot analysis

Western blot analyses were performed as described [7] using specific antibodies directed against FHA, PRN, PT and adenylate cyclase haemolysin (AC-Hly) on all isolates since 2000.


FIM2 and FIM3 detection was performed using monoclonal antibodies as described elsewhere [6].

J774A.1 cell culture and cytotoxicity assays

Culture of the murine monocyte/macrophage-like cell line J774A.1 (ATCC ref TIB-67) and cytotoxicity assays were performed as described previously [19].

Murine intranasal model of infection

All procedures involving animals were conducted in accordance with the Institut Pasteur animal care and use committee guidelines. These experiments were conducted as described previously [19].


Description of the isolates

The annual number of B. pertussis and B. parapertussis isolates received since 2000 is shown in Fig. 1. The number of B. parapertussis isolates collected remains low compared with the number of B. pertussis isolates. The interval time-period between two peaks of pertussis cycles seems to increase over time. All isolates collected since 2000 possess biochemical characteristics similar to the charactersitics of isolates collected during the pre-wP and post-wP vaccine eras.

Figure 1.

 Numbers of Bordetella pertussis and Bordetella parapertussis isolates analysed in France since 2000. Annual numbers of analysed isolates are represented for B. pertussis (grey bars) and B. parapertussis (black bars).

Pulsed-field gel electrophoresis

Our previous results showed that B. pertussis isolates collected before the wP vaccination era (before 1959) belonged to French PFGE groups II and III, which include the French vaccine strains [7] and were rapidly controlled by wP vaccine-induced immunity. In all, 93.8% of all B. pertussis isolates collected since 2000 belong to PFGE group IV. However, PFGE group IV subgroups α [7], β [7] and γ (this study) were observed and their proportions evolved following aP introduction (Fig. 2). The B. parapertussis isolates only segregated into two closely related PFGE groups [20]. Of the B. parapertussis isolates collected since 2007, 71.4% belong to PFGE group II, first identified in 2007 [20].

Figure 2.

 Evolution of the proportion of French Bordetella pertussis pulsed-field gel electrophoresis (PFGE) group IV subgroups. Acellular pertussis (aP) vaccination was first introduced in France in 1998 as an adolescent booster before being generalized to the whole population in 2000. Since 2000, 94% of all the B. pertussis isolates analysed are part of PFGE group IV. This PFGE group can be subdivided in three subgroups, for which the annual proportions are shown on this figure.


Most B. pertussis isolates currently circulating express a type 1 ptxS1 subunit (ptxA1), a type 2 prn and harbour a ptx operon promoter P3 (ptxP3, Table 1). This promoter type is found in over 90% of all isolates collected since 2000. The RD genotyping shows that RD1, RD2 and RD4 have been progressively lost over time and that all isolates collected since 2000 lack these three regions. Bordetella parapertussis genotyping indicates that all isolates possess a type 1.4a-2.9b prn gene.

Virulence factor expression

All B. pertussis isolates collected since 2000 express AC-Hly; 97.2% (1001/1030) express FIM3, 1.3% (13/1030) express FIM2 and 0.6% (6/1030) express FIM2 and FIM3. Forty isolates collected since 2000 were found not to express one or more virulence factors. PT, PRN and PRN FHA isolates account for 5.0% (2/40), 92.5% (37/40) and 2.5% (1/40) of these collected isolates, respectively. Collection of B. pertussis isolates deficient in PRN expression was sporadic between 2000 and 2008 but increased since 2009 and their prevalence reached 13.3% (16/120) of all B. pertussis isolates collected in 2011 (Fig. 3). The phenotype of the PRN isolates is mainly the result of the insertion of the IS481 insertion sequence element in region II of the prn gene (51.4%, 19/37). A prn gene deletion (2.7%, 1/37), a 25-nucleotide deletion in the first repeated region (2.7%, 1/37), an 89-nucleotide deletion at the 5′ end of the prn gene (2.7%, 1/37) or a mutation leading to a stop codon at position 1479 of the prn gene (2.7%, 1/37) was also found in isolates with such a phenotype. The PT phenotype is due to the deletion of the entire ptx operon in one isolate but further analyses are needed to understand this phenotype in the other isolate and to understand the FHA phenotype. The PRNB. parapertussis was first isolated in 2004, but since 2007, 94.3% (32/35) of all collected B. parapertussis isolates have been PRN. This phenotype is caused by the deletion of an A in region I of the prn gene (position 988, 12.1%, 4/33) or a G in region II (position 1895, 75.8%, 25/33) both of which lead to a stop codon.

Figure 3.

 Evolution of the proportion of Bordetella pertussis isolates not expressing pertactin (PRN) since 2000. Detection of PRN expression was performed using Western blot analysis and specific anti-PRN antiserum. No PRN B. pertussis isolates were collected before 2003.

Macrophage cytotoxicity

All selected B. pertussis isolates were cytotoxic for murine macrophages whereas none of the B. parapertussis isolates tested were cytotoxic (data not shown).

Murine model of respiratory infection

All selected isolates, except the two PT ([19] and this study), were lethal to mice, including the PRN, FHA and PRN FHA isolates, and exhibited similar median lethal dose (LD50) values (Table 1). Isolates harbouring a ptxP3 promoter type supposed to express more PT are not more virulent. The B. parapertussis isolates exhibit similar LD50 (Table 1) whether they express PRN or not ([20] and this study).


In the present study we show that since the introduction of aP vaccines in France, 93.8% of the isolates collected are part of the same PFGE group IV, which can be divided in three subgroups, α [7], β [7] and γ (which appeared in 2004, this study). Distributions within these subgroups seem to have reached equilibrium because no major changes have been observed since 2005. Although wP vaccine-induced immunity, which targets the whole bacterium, effectively controlled circulating isolates resembling vaccine strains in France, i.e. isolates belonging to PFGE groups II and III [6,7,12], it did not control PFGE group IV isolates [6,7]. The aP vaccine-induced immunity therefore maintained the monomorphic population of B. pertussis observed after the use of wP vaccine despite the emergence of the PFGE group IV γ subgroup. However, emergence of a new PFGE group was observed within the B. parapertussis population in 2007 [20] and 71.4% of all B. parapertussis isolates collected since then are part of this new group.

Along with evolutionary changes in PFGE group distribution after wP vaccine introduction, particular genomic regions have shown evidence of evolution as well. In addition to changes previously detected in the ptxS1 subunit (ptxA1 replaced ptxA2 and ptxA4 alleles) and the PRN structural gene (prn2 replaced prn1 allele), it was reported that the promoter of the ptx operon has evolved towards a ptxP3 type promoter since the late 1980s [23]. Similar evolutions were observed in France [7,8] and in other European countries [10,12,13] after wP vaccine introduction. It has to be noted that although all isolates harbouring a ptx P3 type promoter are suggested to express more PT [23], they are not more virulent in the murine respiratory model.

It was hypothesized that the B. pertussis population could cope with aP-induced immunity by losing the expression of vaccine antigens [14], by gaining expression of new antigens or by overexpressing antigens not included in aP vaccines such as AC-Hly, because clinical isolates presenting cyaA gene duplication were already described [24].

By controlling expression of virulence factors by collected isolates we recorded a steady increase in the number of isolates not expressing PRN since 2007, 7 years after generalization of aP vaccination in France. They represent 7.8% (32/408) of all isolates collected since 2007 and their proportion reached 13.3% (16/120) in 2011. It is important to note that these data rely on annual surveillance and not on isolates collected during an outbreak of pertussis [3]. Although PT, FHA or PRN isolates were collected during the pre-vaccine and post-vaccine eras in France and in Italy ([25] and this study), they did not expand like PRN isolates have since 2007. This expansion does not appear to be clonal because the PRN phenotype can result from different genomic events. Although PRN was thought to be a major B. pertussis adhesin, its loss does not seem to hinder lethality or transmission. PRN isolates exhibit LD50 values similar to those of PRN+ isolates in the respiratory murine model and retain their ability to be transmitted in the human population (H. Bodilis and N. Guiso, unpublished data). A possible explanation for this phenotype might be found in the great diversity of auto-transporter proteins found in B. pertussis, which may compensate for the loss of PRN expression [26]. Currently, the other well-known virulence factors, targeted or not by aP vaccine-induced immunity, do not seem to be affected. All isolates still express AC-Hly, which is not part of the aP vaccines. As a consequence of the universal expression of this toxin, all selected isolates for this study were cytotoxic for murine J774A.1 macrophages as expected [27]. Even though variable, no increase in this toxin expression, as measured by adenylate cyclase activity, was observed over time (data not shown).

One could speculate that the loss of PRN expression could be the result of host adaptation because B. pertussis recently emerged from B. bronchiseptica [28] and might still undergo genetic rearrangements. Even though loss of genetic material through RD elimination or gain in genome plasticity through increase in insertion sequence element numbers [8] could fit host adaptation, strong evidence for aP vaccine-driven loss of PRN expression can be found. Indeed, it was recently published that in Japan, a country using aP vaccines since the 1980s, circulation of PRN isolates has been observed since 2000 [29]. Even though it is too soon to assume that our hypothesis is verified, other countries using aP vaccines should survey virulence factor expression of circulating isolates to verify if they encounter such a phenomenon.

Another point of interest is the evolution of the B. parapertussis population. Since 2007, 71.4% of the isolates are part of a new PFGE group ([20] and this study). Furthermore, 94.2% (33/35) of the B. parapertussis isolates collected since 2007 are PRN and this phenomenon does not seem to be related to PFGE group distribution. The lack of expression of PRN is the consequence of different genetic events, indicating that these isolates did not arise from the same ancestor. As observed with B. pertussis PRN isolates, B. parapertussis PRN isolates exhibit LD50 values similar to their PRN+ counterparts.

As shown previously, wP vaccine-induced immunity does not seem to protect against B. parapertussis in the field [13,18]. However, whereas aP vaccine-induced immunity was hypothesized to be ineffective against B. parapertussis [16] in the field, an observation confirmed in an animal model of infection [15,17], another study performed during a clinical trial comparing wP and aP vaccines suggested that aP vaccination might be effective in controlling B. parapertussis [18]. It is therefore tempting to assign the emergence of PRNB. parapertussis isolates to aP herd immunity.

In conclusion, even though genetic rearrangements took place in the absence of human interference [28], human vaccination strategies introduced a new challenge that bacterial populations had to overcome, promoting evolution and selection among these populations. It is clearly known now that in France, wP vaccination favoured genotypic changes (PFGE IV, ptxP3, prn2, ptxA1) in the B. pertussis population [6,7], changes which could not be observed in regions of low vaccine coverage (Senegal) [21]. Gene loss (RDs) and increase in insertion sequence elements are also part of the adaptation of the B. pertussis population to its host. Whether they are influenced by vaccination should be appreciated in regions with no or very low vaccine coverage. Continuous surveillance for antigenic change in B. pertussis and B. parapertussis strains is therefore required as a tool for monitoring pertussis vaccine effectiveness.


We thank the members of the hospital-based surveillance, RENACOQ: Isabelle Bonmarin, Emmanuel Belchior and Daniel Levy-Bruhl from the Institut de Veille Sanitaire and all the clinicians and microbiologists of the 43 paediatric hospitals involved in the surveillance: Dr Theveniau, Dr Chardon (Aix-En-Provence); Dr Garnier, Dr La Scola (Marseille); Dr Guillois, Dr Leclercq, Dr Vergnaud (Caen); Dr Guillot, Dr Paris (Lisieux); Dr Romanet, Dr Biessy (La Rochelle); Dr Huet, Dr Duez (Dijon); Dr Dagorne, Dr Vaucel (Saint Brieuc); Dr Hoen, Dr Couetdic (Besançon); Dr de Parscaud, Dr Picard (Brest); Dr Sarlangue, Dr Lehours (Bordeaux); Dr Reygrobellet, Dr Jean Pierre (Montpellier); Dr Bonnemaison, Dr Lanotte (Tours); Dr Bost-Bru, Dr Croize, Dr Pelloux (Grenoble); Dr Mouzard, Dr Gibaud (Nantes); Dr Bentata-Durupt, Dr Barthez-Carpentier (Orléans); Dr Leneveu, Dr Le Coustumier, Dr Wilhems, (Cahors); Dr Savagner, Dr Cottin (Angers); Dr Chomienne, Dr Laurens (Cholet), Dr Morville, Dr Brasme (Reims); Dr Monin, Dr Alauzet (Nancy); Dr Martinot, Dr Courcol (Lille); Dr Blanckaert, Dr Verhaeghe (Dunkerque); Dr Parlier, Dr Darchis (Compiègne); Dr Labbe, Dr Bonnet (Clermont-Ferrand); Dr Sheftel, Dr Kiesler (Strasbourg); Dr De Briel, Dr Kretz (Colmar); Dr Floret, Dr Etienne (Lyon); Dr Bonardi, Dr Boyer (Le Mans); Dr Grimprel, Dr Garbarg-chenon, Dr Moissenet (Trousseau Hospital, Paris); Dr Bourrillon, Dr Bingen, Dr Bonacorsi (R. Debré Hospital, Paris); Dr Cheron, Dr Descamps, Dr Ferroni (Necker Hospital Paris); Dr Gendrel, Dr Raymond, Dr Poyard (Saint-Vincent-de-Paul Hospital, Paris); Dr Meunier, Dr Le Luan (Fécamp); Dr Mallet, Dr Lemeland, Dr Nouvellon, Dr Lemée (Rouen); Prof Eb, Dr Hamdad-Daoudi (Amiens); Dr Fortier, Dr Lefrand (Avignon); Dr Menetrey, Dr Ploy (Limoges); Dr Gaudelus, Dr Poilane (Bondy); Dr Delacour, Dr Estrangin, Dr Aberrane (Créteil); Dr Carrière, Dr Prère, Dr Delmas (Toulouse); Dr Parez, Dr Valdes (Colombes); Dr Tara-Maher, Dr Reveil (Charleville-Mezières); Dr Grattard (St Etienne).

We also thank the Collection of the Institut Pasteur for the gift of the pre-vaccine era isolates.

Conflicts of Interest

Nothing to declare.

Author Contributions

Nicolas Hegerle, Sophie Guillot, Nicole Guiso conceived and designed the experiments, analysed the data and wrote the paper. Nicolas Hegerle, Anne-Sophie Paris, Delphine Brun, Gregory Dore, Elisabeth Njamkepo performed the experiments. Nicole Guiso obtained the financial support.

Transparency Declaration

This work was supported by the Institut Pasteur Fondation, URA CNRS3012, and GlaxoSmithKline Biologicals, Rixensart, Belgium.