Evolution of Bordetella pertussis in the acellular vaccine era in Norway, 1996 to 2019

We described the population structure of Bordetella pertussis (B. pertussis) in Norway from 1996 to 2019 and determined if there were evolutionary shifts and whether these correlated with changes in the childhood immunization program. We selected 180 B. pertussis isolates, 22 from the whole cell vaccine (WCV) era (1996–1997) and 158 from the acellular vaccine (ACV) era (1998–2019). We conducted whole genome sequencing and determined the distribution and frequency of allelic variants and temporal changes of ACV genes. Norwegian B. pertussis isolates were evenly distributed across a phylogenetic tree that included global strains. We identified seven different allelic profiles of ACV genes (A–F), in which profiles A1, A2, and B dominated (89%), all having pertussis toxin (ptxA) allele 1, pertussis toxin promoter (ptxP) allele 3, and pertactin (prn) allele 2 present. Isolates with ptxP1 and prn1 were not detected after 2007, whereas the prn2 allele likely emerged prior to 1972, and ptxP3 before the early 1980s. Allele conversions of ACV genes all occurred prior to the introduction of ACV. Sixteen percent of our isolates showed mutations within the prn gene. ACV and its booster doses (implemented for children in 2007 and adolescents in 2013) might have contributed to evolvement of a more uniform B. pertussis population, with recent circulating strains having ptxA1, ptxP3, and prn2 present, and an increasing number of prn mutations. These strains clearly deviate from ACV strains (ptxA1, ptxP1, prn1), and this could have implications for vaccine efficiency and, therefore, prevention and control of pertussis. Supplementary Information The online version contains supplementary material available at 10.1007/s10096-022-04453-0.


Background
Pertussis, caused by Bordetella pertussis (B. pertussis), is a highly contagious acute respiratory infection. The disease affects people of all ages, but is especially severe in infants. Although pertussis is a vaccine-preventable disease with high vaccine coverage worldwide, a global resurgence of this infection has been observed [1]. Pertussis is endemic in many countries, with epidemic cycles occurring every 2-5 years [2]. In 2014, 24.1 million pertussis cases were estimated globally, with 160,700 deaths from pertussis in children less than 5 years of age [3]. In Europe, a notification rate of 8.2 cases per 100,000 population was reported in 2018, with the highest notification rate in infants followed by 10-14-year olds [4]. In Norway, a notification rate of 47.4 cases per 100,000 population was reported in 2019 (www. msis. no). Norway has, since 2011, consistently reported the highest notification rate of pertussis in all of Europe, and in contrast to the majority of European countries, adults (≥ 18 years of age) accounted for the majority of cases (55%) [5].
During the mid-1990s and in the early twenty-first century, whole cell vaccines (WCVs) were replaced by acellular pertussis vaccines (ACVs) in high-income countries, including Norway, where ACVs were introduced from 1998 [2]. To combat the increasing incidence of pertussis, Norway and other European countries implemented childhood and adolescent ACV boosters [6,7]. In Norway, ACV boosters were introduced for 7-year olds in 2006/2007 and for 15-year olds in 2013/2014. Recently, maternal immunization was introduced in some countries to prevent severe pertussis in infants, but has yet to be introduced in Norway [8].
Several factors are contributing to the re-emergence of pertussis, like improved diagnostics [9], increased awareness of the disease [10,11], decreased vaccine efficacy [12], waning immunity [8,13,14], and pathogen adaption [2,[15][16][17][18][19]. Growing evidence suggests that genetic divergence of circulating B. pertussis population away from vaccine antigens and the emergence of strains with increased pertussis toxin production might be contributing factors [17,[19][20][21]. Deficiency in pertactin (Prn), one of the components in ACVs, is widely distributed globally, and the rapid spread of Prn deficiency is likely vaccine driven [22]. Mutations in other ACV antigens such as pertussis toxin (ptxA), its promoter (ptxP), and fimbria (fim2 and fim3) are also reported and shown to have quickly spread throughout the B. pertussis population [17,23]. Currently, more than 90% of B. pertussis circulating in Europe have ptxA1, ptxP3, and prn2 genotypes and the recent replacement of ptxP1 with ptxP3 is associated with increased pertussis toxin production [17,19,24]. Comparative genomic analysis has demonstrated that worldwide transmission of new strains is rapid, and that the global population of B. pertussis is evolving in response to vaccine introduction, potentially enabling vaccine escape [17,25]. All these changes raise concern, as ACV-induced selection pressure could potentially threaten vaccine efficacy [22,26].
Pertussis has resurged in Norway with high notification rates since the late 1990s. Although lethal cases are very rare, pertussis is poorly controlled [7,27]. No overview of the molecular epidemiological situation of B. pertussis in Norway exists. Only a few Norwegian isolates from the last decade have previously been included in European studies, showing that the ptxA1, ptxP3, and prn2 genotypes are frequent and that Prn-deficient isolates are present among isolates examined [16,18,20,28]. In Norway, the three-component ACVs used in the childhood immunization program and as booster vaccine for 15-year olds include vaccine antigens from Tohama I: ptxA2, ptxP1, prn1, and fhaB1 (https:// bigsdb. paste ur. fr/ borde tella/), whereas the two-component vaccine used as a booster vaccine for 7-year olds harbors vaccine antigens from the "Pillemer (P134)" strain, with ptxA2 present, but no information on the allelic variant of fhaB [29]. This suggests that the Norwegian B. pertussis population might have ACV antigen profiles evolving away from the antigens used in ACVs. The aim of this retrospective study was to characterize the population structure of B. pertussis in Norway over the period from 1996 to 2019 and determine whether there have been antigenic shifts in the wake of the introduction of ACVs. As such change may have implications for vaccine efficiency, the results can inform vaccine policy in Norway.

B. pertussis isolates
Clinical microbiological laboratories from each of five regions of Norway (Eastern Norway, Southern Norway, Western Norway, Northern Norway, and Trøndelag County) referred B. pertussis isolates or PCR-positive samples from clinical specimens to the National Reference Laboratory (NRL) for pertussis at the Norwegian Institute of Public Health (NIPH) on a voluntary basis. Before 2002, pertussis cases were diagnosed with culture and serological methods; however, from 2002, the use of PCR gradually increased, first among the youngest age groups, and from 2012, the majority of cases with pertussis in Norway were diagnosed by PCR. Stratified convenience sampling of 180 B. pertussis isolates from the microbiological biobank at the NRL was performed, taking into account the time of sampling (year), place of residence (five regions of Norway), and age of the case at the time of sampling (< 20 years or ≥ 20 years). We included eight and 14 isolates from 1996 and 1997, the WCV era, respectively, and a median of eight (range 5-10) isolates annually from 1998 to 2019, the ACV era (Table S1). No isolates or PCR-positive samples were received at the NRL for the years 2010 and 2011. Twenty-five of the isolates have previously been included in European Pertussis Surveillance studies [16,18,20,28] (Table S1).

Culture, DNA extraction, and whole genome sequencing
Approximately 60 PCR-positive clinical samples were cultured on charcoal agar with 40 µg/ml cephalexin and 250 µg/ ml amphotericin B (made in-house at the NIPH) at 35℃ without CO 2 under humid conditions for up to 10 days. Pure cultures were verified by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) (Daltonics) and stored in the microbiological biobank at − 70℃. Strains selected from the biobank were cultured on charcoal agar as described above for 24-72 h prior to genomic DNA extraction using QIAamp DNA Mini QIAcube kit (Qiagen) on QIAcube (Qiagen) following the manufacturer's procedure. DNA concentration was measured using a Qubit fluorometer (Invitrogen, Thermo Fisher Scientific). DNA libraries were constructed with KAPA HyperPlus (Roche Life Science) with KAPA Unique Dual-Indexed Adapter kit (Roche Life Science), following the manufacturer's instructions. Clean-up and size selection were performed using magnetic AMPure XP beads (Beckman Coulter). DNA was then quantified using a Qubit fluorometer, and the size of fragments was measured using an Agilent 4200 TapeStation system (Agilent Technologies). Paired-end reads (300 bp × 2) were produced on an Illumina MiSeq platform according to standard protocols using the MiSeq Reagent kits (v2 600 cycles; Illumina). Illumina data was deposited in the European Nucleotide Archive under the accession numbers ERS7736033 to ERS7736212. FastQC (Babraham Bioinformatics) was used for quality control of the raw reads. Sequence reads were trimmed and adapters removed using Trimmomatic v0.36. FASTA contigs were obtained from the trimmed reads using SPAdes v3.13.2. The final assembly files consisted of a median number of 277 contigs/sample (range: 261 to 1568 contigs) and had a median of 75.9 coverage/sample (range: 14.9 to 520.4 coverage). The mean length/sample was 3,885,772 bp. The B. pertussis genome is ~ 4.1 Mb, and the estimated mean coverage across all sequencing runs was 94.2%. Kraken (version 1.1.1) was used to assign taxonomic label species identification. All, except one sample, showed pure cultures of B. pertussis. The last sample was contaminated with Streptococcus (Table S1).

Defining molecular mechanisms associated with mutations within the prn gene
A region including the prn gene was extracted from all 180 whole genome sequences (approximately 2748 bp). The sequences were imported and aligned in MEGA X [31] and manually examined for mutations using Tohama I (NC_002929.2), carrying prn1, as the reference. All sequences were screened for insertion sequence (IS) elements in prn using ISMapper (https:// github. com/ jhawk ey/ IS_ mapper).

Time-measured evolutionary analysis
Phylogenetic tree scaled to time was created using the program treetime v0.8.1 with the following non-default options: least-squares rerooting, accounting for covariation, and coalescent rate set to constant. The R 2 value was 0.63. Figures were annotated using ITOL [33].

Molecular mechanisms involved in mutations within the prn gene
Sixteen percent (29/180) of the Norwegian B. pertussis isolates harbored mutations within the prn gene, including partial deletions and IS481 insertions (Table 1). Among these, 97% (28/29) carried prn2 and one had prn1. Of the isolates with prn mutations, allelic profile B was dominating (79%, 23/29), whereas profile A1 occurred in four isolates (14%, 4/29) and profiles A2 and F in one isolate each, respectively. Twelve (41%) isolates had insertion of IS481 in positions between 1613 and 1614 or between 241 and 242, including isolates with allelic profiles B (n = 7), A1 (n = 4), and F (n = 1). Eight (28%) isolates, all allelic profile B, had C > T mutation leading to stop codon at positions 1258 (n = 6) and 223 (n = 2), respectively. The last nine isolates (31%) had deletion in the start of prn, eight with allelic profile B and one with profile A2. Seven of the nine isolates had deletion of IS1663 downstream of the prn gene including deletion of the proximal part of prn (Table 1). Mutations within the prn gene were more common from 2007, and in 2019, 60% (6/10) of the selected Norwegian B. pertussis isolates had prn mutations (Fig. S2).

Evolution of B. pertussis in Norway from 1996 to 2019 and its correlation to alterations in pertussis vaccination
Temporal changes of genes encoding ACV-related antigens showed that alterations of these genes occurred prior to introduction of ACV in 1998, but mainly after implementation of WCV in 1952 (Fig. 2). The prn2 allele likely emerged prior to 1972 and the ptxP3 allele before the early 1980s. B. pertussis isolates with allelic profile B (ptxA1, ptxP3, prn2,  fim2-1, fim3-1, and fhaB1) were introduced around 1980, and allelic profile B is still the dominating profile of Norwegian B. pertussis isolates. B. pertussis isolates with allelic profiles C-F, all carrying ptxP1 and prn1/prn2, were present before and during the WCV era, but disappeared around 2007. In the mid-1990s, the fim3-2 allele appeared and allelic profile A1 (ptxA1, ptxP3, prn2, fim2-1, fim3-2, fhaB1) was introduced. After the mid-1990s, but before the introduction of ACV, allelic profile A2 occurred, carrying the fhaB7 allele. However, profile A2 disappeared around 2005 prior to introduction of the first ACV booster dose. After 2005, only two allelic profiles (A1 and B) were present in the Norwegian B. pertussis population, both carrying ptxA1, ptxP3, prn2, fim2-1, and fhaB1 and either fim3-2 (A1) or fim3-1 (B) (Fig. 2). An increase in the number of mutations within the prn gene was seen after the introduction of ACV and its booster doses, and these mutations were mainly seen in B. pertussis isolates with allelic profiles A1 and B (Table 1 and Fig. 2).

Comparison of the Norwegian B. pertussis population with the global population
The available global B. pertussis isolates covered the time period 1936 to 2016 and 18 countries in six different continents, although 81% (300/370) of the isolates were from North America (Table S2). The majority of the isolates showed ST2.
The maximum likelihood tree showed that the Norwegian B. pertussis population was distributed across the global population of ST2 and ST83 isolates (Fig. 3). However, a few branches consisted entirely of Norwegian B. pertussis isolates. Branch 1 included isolates from 1997 to 2005, and they showed allelic profile A2, with the fhaB7 allele present. Branch 2 consisted of a subgroup of Norwegian B. pertussis isolates with allelic profile B. These Table 2 Allelic profiles of acellular vaccine gene variants of B. pertussis in Norway, 1996-2019 a Including one isolate with ptxP34 instead of ptxP3 and another isolate with fim2-15 instead of fim2-1 (both indicated as B* in Fig. 2 and Table S1) b Including two isolates with prn3 instead of prn1 (both indicated as C* in Fig. 2 (Fig. 3).

Discussion
A global resurgence of pertussis has been seen after the introduction of ACVs in the 1990s [34,35]. Changes in the B. pertussis population and pathogen adaptation have been suggested as one of the causes of this increase [2,15,[17][18][19]. For Norway, little data was available on the B. pertussis population before this study and none of the isolates had previously been whole genome sequenced. In the Norwegian B. pertussis population, a heterogeneous mix of strains with a distinct time-dependent shift in allelic profiles was observed. The Norwegian strains were distributed across the global phylogenetic tree, supporting previous observations of rapid strain flow of B. pertussis between countries, even though a few country-specific branches were observed [17,21,25,36,37].
Temporal changes of genes encoding the ACV antigens in the Norwegian B. pertussis population were comparable with findings from other countries. The changes supported the evidence that genetic alterations are driven by the selective pressure imposed by vaccine-induced immunity, mainly from the WCV era, and accentuated in the ACV era [15,17,23,38]. In concordance with the global population, B. pertussis with ptxP3, a gene variant leading to increased pertussis toxin production and associated with more severe Fig. 3 SNP-based phylogenetic tree of Norwegian B. pertussis isolates, 1996-2019, put in a global context. Norwegian strains (n = 180, purple circles) were distributed across the global phylogenetic tree; however, a few country-specific branches were detected (1)(2)(3)(4)(5). Worldwide strains (n = 370) were isolated from 1939 to 2016 and covered six different continents (Table S2) disease in young infants, emerged prior to the early 1980s and dominated by the end of the 1990s [15,17,24,36,39]. The prn2 allele emerged from the 1970s, as indicated by others, and is now the dominating prn allele in both WCVs and ACVs using countries, except for China and India [17,19,[40][41][42]. Furthermore, the fim3-2 allele appeared in the start of the 1980s and has since been present alongside fim3-1 in the B. pertussis population worldwide [15,18,43].
Like other countries, we observed an increasing trend of mutations within the prn gene, including partial deletions and IS481 insertions, associated with time since implementing ACVs [16,44,45]. Even though Prn expression was not examined in our study, five of our isolates with prn mutations have previously been shown to be Prn-deficient [28]. Interestingly, few prn-negative isolates have been reported from countries still using WCVs as well as in countries without a Prn component included in their ACVs, indicating that this alteration is driven by ACVs [16,19,41,42,[46][47][48]. This has clearly been demonstrated by Japan who implemented ACVs as early as 1981 followed by increased proportion of Prn-deficient strains from the mid-1990s. However, in 2012, Japan introduced ACVs without Prn and a decline in Prn-deficient strains was observed [19,37,47,48].
During the WCV era, a number of antigenic allele conversions emerged, and the introduction of ACVs led to clonal expansions of strains with mutations that did not match the ACV profile. It has been shown that B. pertussis isolates with ptxP3 and Prn expressed had increased fitness during the WCV era, whereas after implementing ACV, enhanced fitness was associated with Prn-deficiency, ptxP3 and fim3-1 [37]. Consequently, the diversity in allelic profiles decreased during the ACV era and approximately 90% of recent circulating B. pertussis isolates harbor ptxA1, ptxP3, and prn2 with an increasing number of mutations within the prn gene [2,21]. This demonstrates that the current B. pertussis population clearly deviates from the vaccine strains (ptxA2, ptxP1, prn1). This might affect the preventive potential of the ACVs and thus contribute to resurgence of pertussis [12,38]. However, resurgence of pertussis is a complex phenomenon, and several other factors might favor the rise in pertussis disease. For instance, ACVs have been shown to be inferior to WCVs by not inducing mucosal immune response leading to increased colonization and transmission of B. pertussis and by inducing earlier waning immunity [12,49]. Thus, novel ACVs with antigen targets matching the current circulating B. pertussis population or new targets, as well as appropriate adjuvants to stimulate both antibody and cellmediated immune response, are highly warranted [49,50].
Macrolide-resistant isolates are only sporadically seen in Europe, the Middle East, and North and South America [2]. This is in concordance with our findings, showing molecular determinants for macrolide resistance in only one B. pertussis isolate. This isolate carried the msrD_2 and mefA_10 genes, to our knowledge, never detected in B. pertussis previously; however, this sequence was contaminated with Streptococcus. Interestingly, in China, antibiotic abuse has led to a high level of erythromycin resistance in the B. pertussis population and contributed to the spread of B. pertussis lineages harboring ptxA1, ptxP1, and prn1, clearly deviating from the dominating clone seen in Europe, even though similar vaccine regimes were implemented both places years ago [32].
Limitations of the study include that the representativeness of our material can be questioned, because the primary laboratories in Norway submit B. pertussis samples to the NRL at the NIPH only on a voluntary basis and few isolates were included annually. Nonetheless, the similarities with previous studies indicate that our selection of isolates probably is representative for the Norwegian population. Second, we used short-sequence technology which is incapable of detecting rearrangements and large duplications responsible for the genome diversity seen in B. pertussis [51][52][53]. However, studies using long-sequence technology in combination with short sequencing technology show similar results to ours when it comes to temporal changes of ACV antigens [17,21]. Finally, we have examined alterations at the genomic level and not at the transcriptomic or proteomic levels. The two latter might also be important for vaccine efficacy. Several countries have reported B. pertussis strains that do not express ACV antigens such as Prn, FhaB, and Ptx, although Prn deficiency is the only change frequently reported in the B. pertussis population [15,19,38,54,55].

Conclusion
We observed molecular epidemiological shifts in the Norwegian B. pertussis population during the WCV and ACV eras. ACV use and its booster doses have led to lineagespecific proliferation of more successful, ACV-deviating strains, leading to a more uniform B. pertussis population. These clones have the ptxP3 and prn2 alleles fixed and are characterized by increasing incidence of mutations within the prn gene. The emergence of B. pertussis isolates with ACV antigens deviating from the vaccine antigens might have implications for vaccine efficiency and therefore prevention and control of pertussis. We recommend continued collection of B. pertussis for surveillance purposes, and to further explore the role of genetic adaptation. Developing new vaccine candidates, including allelic variants of ACV antigens present in the current circulating population, as well as exploring novel candidates, requires attention.