Screening and Genomic Characterization of Filamentous Hemagglutinin-Deficient Bordetella pertussis

ABSTRACT

Despite high vaccine coverage, pertussis cases in the United States have increased over the last decade. Growing evidence suggests that disease resurgence results, in part, from genetic divergence of circulating strain populations away from vaccine references. The United States employs acellular vaccines exclusively, and current Bordetella pertussis isolates are predominantly deficient in at least one immunogen, pertactin (Prn). First detected in the United States retrospectively in a 1994 isolate, the rapid spread of Prn deficiency is likely vaccine driven, raising concerns about whether other acellular vaccine immunogens experience similar pressures, as further antigenic changes could potentially threaten vaccine efficacy. We developed an electrochemiluminescent antibody capture assay to monitor the production of the acellular vaccine immunogen filamentous hemagglutinin (Fha). Screening 722 U.S. surveillance isolates collected from 2010 to 2016 identified two that were both Prn and Fha deficient. Three additional Fha-deficient laboratory strains were also identified from a historic collection of 65 isolates dating back to 1935. Whole-genome sequencing of deficient isolates revealed putative, underlying genetic changes. Only four isolates harbored mutations to known genes involved in Fha production, highlighting the complexity of its regulation. The chromosomes of two Fha-deficient isolates included unexpected structural variation that did not appear to influence Fha production. Furthermore, insertion sequence disruption of fhaB was also detected in a previously identified pertussis toxin-deficient isolate that still produced normal levels of Fha. These results demonstrate the genetic potential for additional vaccine immunogen deficiency and underscore the importance of continued surveillance of circulating B. pertussis evolution in response to vaccine pressure.

INTRODUCTION

Bordetella pertussis is the causative agent of whooping cough (pertussis), a highly contagious respiratory disease most severe in unvaccinated infants. The introduction of vaccines against pertussis during the 1940s dramatically reduced disease incidence in the United States. Acellular pertussis (aP) vaccines replaced whole-cell preparations in the United States during the 1990s, and despite high vaccine coverage, pertussis cases have since risen, with notable recent epidemics (1–3). Multiple factors likely contribute to increased disease reporting, including heightened awareness, expanded surveillance, and improved laboratory diagnostic testing (3, 4). There is also growing evidence to suggest that waning protection conferred by aP vaccine formulations has led to increased disease rates among vaccinated individuals (3, 5–7).

The United States exclusively uses aP vaccines for childhood series and adolescent/adult boosters, composed of inactivated pertussis toxin (Pt), pertactin (Prn), and filamentous hemagglutinin (Fha), either with or without fimbria (Fim) types 2 and 3. However, genetic divergence of the circulating population has resulted in allelic mismatches with vaccine reference strains, which may contribute to disease resurgence within aP-vaccinated populations (8, 9). Mutations to vaccine immunogens, such as Pt (subunit ptxA and the promoter ptxP) and Fim (fimH), have spread quickly throughout the circulating population with little evidence of geographic restriction (10, 11), prompting the description of B. pertussis as a monomorphic pathogen. Yet pertussis epidemics are not clonal (1, 2, 12). Circulating isolates recovered in the United States have become predominantly Prn deficient by one of at least 16 independent mutations, including missense substitutions, insertions, deletions, and promoter disruptions, but most frequently through IS481 insertion at one of three positions (13). The global emergence of Prn deficiency is well documented (13, 14) but, although likely vaccine driven (15), does not appear to impact the effectiveness of aP vaccines (16). Together, such genetic shifts illustrate how circulating B. pertussis has significantly diverged from vaccine reference strains since widespread vaccine introduction, which many researchers have attributed to vaccine-driven immune selection (4, 10, 15, 17–20).

The regulation of virulence genes in B. pertussis, including aP vaccine immunogens, is controlled by the BvgAS two-component system (21, 22). Although immunogens are included in aP vaccine formulations for their antigenic properties, knowledge of their activity during infection varies. Fha is understood to function as both a surface-associated and secreted protein (23). Numerous studies have demonstrated that Fha contributes to both colonization and persistent infection, acting not only as an adhesin to mediate respiratory epithelium attachment (23, 24) and promote biofilm formation (25, 26) but also as a modulator of immune responses (24, 27–31). Mutants lacking Fha exhibit reduced colonization (25, 26) and elicit a more robust inflammatory response, leading to faster clearance compared to the wild type (27). It should be noted that much of this knowledge is derived from experiments in animal models, and often with the closely related Bordetella bronchiseptica, which do not exhibit the full spectrum of symptoms observed in human pertussis (32, 33). While Prn appears to be unessential for disease (13, 15), Fha on the other hand may critically mediate persistent infection such that the emergence of deficient isolates would represent a remarkable shift in B. pertussis host-pathogen interaction.

The emergence and spread of immunogen-deficient isolates raises concern about the stable production of antigenic proteins, and we have previously developed electrochemiluminescent antibody capture (ECL) assays for detecting Prn and Pt (34). To further improve surveillance of circulating B. pertussis, here we modified the same assay platform to screen 787 isolates for Fha production and identify rare Fha-deficient mutants. Subsequent genomic characterization of mutant isolates revealed putative, underlying genetic changes analogous to those contributing to the rise of Prn deficiency. These results demonstrate the potential for additional, multiple-immunogen deficiency, presenting an emerging public health challenge that emphasizes the importance of continued monitoring of circulating B. pertussis for production of all vaccine immunogens.

RESULTS

Isolates deficient in Fha production.

Seven hundred twenty-two isolates collected in the United States through disease surveillance from 2010 to 2016, as well as 65 historic clinical isolates and laboratory strains from the CDC collection, were screened for Fha production using an ECL assay. Among recent circulating strains, two isolates were identified with severely reduced Fha production (J014, J199) compared to the wild type (Fig. 1A). Three historic laboratory strains (B199, J042, and J043) were also deficient for Fha (Fig. 1A). Additionally, J014 and J199 were Prn deficient (Table 1) and J043 was deficient for both Prn and Pt (Fig. 1B; see also Table S1 in the supplemental material) when screened using the same platform (34). Altered Fha production was confirmed in all five isolates by Western blotting following growth in liquid culture (Fig. 2) and on agar media (data not shown). Under both growth conditions, no Fha or cross-reacting material was detected in B199, J014, J042, or J043 while a very low concentration of protein was observed in J199, consistent with initial screening results. Deficient clinical isolates were collected from cases of infected unvaccinated children, none of which were fatal. Clinical and epidemiologically derived data for each are listed in Table 1.

FIG 1.

Vaccine immunogen detection by electrochemiluminescent assay. Fha (A) and Pt (B) protein production of Fha-deficient mutants J014, J199, J042, J043, and B199 compared to select Fha-producing isolates H627, I762, I959, J010, and J174 measured by secondary antibody capture. Purified protein and phosphate-buffered saline (PBS) were included for positive and negative control, respectively. Electrochemiluminescent (ECL) values are averages, and error bars represent the standard deviations for 50 replicates for Fha (A) and 18 replicates for Pt (B).

TABLE 1.

Epidemiologic data of vaccine immunogen-deficient clinical isolates

Isolate	Phenotypec	Yr	State	Age of child	Vaccination status	Symptoms	Hospitalized
J014a	Prn-def, Fha-def	2013	Minnesota	<2 mo	Too young for vaccination	Paroxysmal cough, whooping cough, apnea, cyanosis	Yes
J199a	Prn-def, Fha-def	2014	Colorado	4 yrs	Parental refusal	Paroxysmal cough, posttussive vomiting, apnea	No
J365	Prn-def, Pt-defb	2014	California	<2 mo	Allegedly vaccinated	Paroxysmal cough	No

^aIsolates collected through Enhanced Pertussis Surveillance (61).

^bIdentified in reference 34.

^cPrn-def, Prn deficient; Fha-def, Fha deficient; Pt-def, Pt deficient.

FIG 2.

Fha detection by Western blotting. Fha production determined by ECL assay was found to be deficient in J014, J199, J042, J043, and B199 compared to Fha-producing isolates J010 and J365. Each well was loaded with 10 μg of total protein. The anti-Fha antibody recognized multiple polypeptides in J010, consistent with the frequent observation of various FhaB degradation products in B. pertussis cell extracts (28).

Mutation identification by whole-genome sequencing.

Shotgun sequencing reads from Fha-deficient isolates were mapped to the genome sequences of Fha-producing vaccine references Tohama I (E476) and 10536 to identify putative, underlying mutations. Genes encoding known proteins within the Fha biosynthesis pathway were further checked manually for nonsynonymous sequence variation or IS481 disruption. Identified mutations were confirmed to be absent in 148 Fha-producing isolates (Table S1) and are detailed in Table 2.

TABLE 2.

Characteristics, identified mutations, and genome sequence accession numbers of isolates analyzed in this study^g

Isolate	Source (alias)	Yr isolated	PFGE profile	Detected mutations	GenBank accession no.	Reference
B199	FDA (105)	1935	CDC059	fhaB: IS481, pos 3124; 28 nonsynonymousb; 127-kb duplicationc; 34.5-kb duplicationc	CP022361	This study
J014a	Minnesota, USA	2014	CDC237	RD16_14675: del AGGCCC, pos 479	CP012135	This study
J042	NIH (325)	1947	CDC232	bipA: IS481, −88d; RD16_14485: C10T; RD16_09925: C352T; 57.5-kb duplicationc	CP019869	This study
J043	NIH (326)	1947	CDC372	bvgS: ins C, pos 3312; fhaB: IS481, pos 9865; 58 nonsynonymousb; 6 IS481b	CP016887	This study
J199a	Colorado, USA	2014	CDC237	fhaB: del G, pos 1087; bipA: 5′-UTR C/Te	CP011245	This study
J365	California, USA	2014	CDC002	28-kb ptx-ptl delf; fhaB: IS481, pos 9865	CP013867	This study
10536	Sanofi vac	1939	CDC054	NA	CP012128	37
B1917	Netherlands	2000	NT	NA	CP009751	78
B1920	Netherlands	2000	NT	NA	CP009752	78
B3582	Sweden	2009	NT	fhaB: ins G, pos 1087	CP011443	35
B3585	Sweden	2009	NT	Unknown	CP011444	35
C393	Chinese vac	1951	CDC052	NA	CP010963	12
E476	Tohama I	1954	CDC232	NA	CP010964	12
FR3749	France	2007	CDC046	NA	CP010966	45
H378	California, USA	2010	CDC253	NA	CP010839	12
H559	California, USA	2010	CDC253	NA	CP010844	12
H622	California, USA	2010	CDC217	NA	CP010847	12
H627	California, USA	2010	CDC217	NA	CP010962	12
I468	Vermont, USA	2012	CDC002	NA	CP010251	12
I472	Vermont, USA	2012	CDC046	NA	CP010253	12
I483	Vermont, USA	2012	CDC237	NA	CP010256	12
I518	Vermont, USA	2012	CDC002	NA	CP010259	12
I521	Vermont, USA	2012	CDC237	NA	CP010260	12
I762a	New York, USA	2013	CDC002	NA	CP011745	36
I959a	Oregon, USA	2012	CDC253	NA	CP011746	36
I979	New York, USA	2013	CDC306	NA	CP010966	45
J010a	Connecticut, USA	2013	CDC237	NA	CP012085	36
J174a	Colorado, USA	2014	CDC237	NA	CP013900	36

^aIsolates collected through Enhanced Pertussis Surveillance (61).

^bDetails listed in Data Set S1 in the supplemental material.

^cDetails listed in Data Set S3 in the supplemental material.

^dDetails listed in Data Set S2 in the supplemental material.

^e164 bp upstream of the translational start site at position 1147474 (position 1169765 in E476).

^fDescribed by Williams et al. (45).

^gAbbreviations: del, deletion; ins, insertion; pos, position; vac, vaccine; UTR, untranslated region; NT, not tested; NA, not applicable.

Isolate J199 produced very little Fha, and genome sequencing revealed a single deleted G at position 1087 in fhaB, which encodes the precursor to mature Fha (Fig. 3). This deletion occurred within a homopolymeric tract, typically 10 bp long in Fha-producing isolates, causing a frameshift mutation and presumably a truncated protein product. The Swedish Fha-deficient isolate B3582 sequenced by Bart et al. (35) contained an inserted G within the same homopolymeric tract, also causing a frameshift mutation (Fig. 3). Because the PacBio RSII platform is prone to single-base insertion-deletion errors in homopolymeric tracts, Illumina reads were manually checked to confirm the deletion in 36 (64.3%) of 56 reads mapped to this locus. The genome sequence of J199 did not contain any additional disruptive IS481 insertions but did include a strain-specific single nucleotide polymorphism (SNP) in the 5′ untranslated region (5′ UTR) of the BvgAS-regulated bipA (RD16_05540 in E476 [Table 2]). The presence of opposing variation in the same homopolymeric tract within fhaB in genomes from two Fha-deficient isolates suggests that this locus is subject to strand slippage during replication.

FIG 3.

Observed mutations in fhaB. Mutations within domains of the FhaB preprotein: signal peptide (SP), two-partner secretion (TPS), mature C-terminal (MCD), N-terminal prodomain (PNT), proline-rich region (PRR), and extreme C-terminal (ECT) domains. Genomes of isolates J199, B199, and B3585 exhibited mutations that altered mature Fha, while J043 and J365 included IS481 insertion within the C-terminal prodomain. Predicted IS481 insertion target sites at positions 2785, 3124, and 9865 (RD16_09375 in E476) are indicated with red arrows.

The fhaB coding region includes three predicted IS481 target sites (36), and an insertion was observed at position 3124 in B199, disrupting the central β-helical shaft domain of mature Fha (Fig. 3). B199 is most closely related to vaccine strain 10536 (37) and differed in genome sequence by an additional 83 variants, including 28 nonsynonymous substitutions, none of which were located in genes known to be associated with Fha biosynthesis (see Data Set S1 in the supplemental material). In J043, IS481 insertion was observed in fhaB at position 9865, near the C-terminal end of the FhaB prodomain (Fig. 3). However, this insertion was also observed in J365, which produced Fha (Fig. 2), indicating that IS481 at this site does not lead to Fha deficiency.

No mutations were detected in any genes of the known Fha biosynthesis pathway in the deficient clinical isolate J014. Instead, a 6-bp in-frame deletion was identified within a gene encoding a possible alternative sigma factor (RD16_14675 in E476). This mutation was not observed in any of the 148 Fha-producing isolates and removed two amino acids adjacent to a predicted Sigma70 region 4 (SR4) DNA-binding motif (see Fig. S1 in the supplemental material). Additional annotation suggested that this sigma factor belonged to the extracytoplasmic function (ECF) class (38).

Historic laboratory strains J042 and J043 are phylogenetically related to Tohama I, and their genomes differed from that of E476 by 2 and 106 sequence variants, respectively. In J042, one nonsynonymous variant was detected in a predicted amidotransferase (RD16_14485 in E476) and the other within a hypothetical membrane protein pseudogene (RD16_09925 in E476 [Table 2]). Three additional IS481 insertions were present in J042: two adjacent to existing insertions in E476 and one within the 5′ UTR of bipA (Table 2). The insertion upstream of bipA not only disrupted known BvgA binding sites (39, 40) but also facilitated a unique ∼22-kb inversion, separating the bipA promoter and transcriptional start site from its coding region (see Fig. S1 and Data Set S2 in the supplemental material).

Historic laboratory strain J043 was deficient for Prn, Fha, and Pt. Of the observed sequence variants, 58 were predicted to cause amino acid changes or frameshift mutations in encoded proteins (Data Set S1). Among these was a single C insertion at position 3307 within a homopolymeric tract in bvgS, the central histidine kinase of the B. pertussis virulence regulon (Table 2; Fig. S1). Illumina sequencing reads mapped to this locus confirmed the insertion in 89 (87.3%) of 102 reads. The mutation to bvgS occurred between the receiver (REC) and histidine phosphotransfer (Hpt) domains, which are necessary for phosphorylation of BvgA to activate expression of Prn, Pt, and Fha. The genome of J043 also contained seven additional IS481 insertions compared to E476, including within fhaB (Data Set S1; Fig. 3). IS481 insertion was also detected in the methyl-directed mismatch repair gene mutS, the disruption of which increases mutation rates in various bacterial species (41). Insertion into mutS was not observed in any other Fha-deficient or Fha-producing isolates.

Large genome duplications and deletions.

The chromosomal structure of the B. pertussis genome is subject to IS481-mediated rearrangement (36), and the genomes of historic Fha-deficient laboratory strains B199 and J042 included large chromosomal duplications (Table 2). These duplications were not resolvable by high-throughput sequencing alone, and assembly required enzyme mapping with high-molecular-weight genomic DNA. The genome of J042 included the direct duplication of a 57.5-kb repeat, while B199 harbored a more complex repeat structure; three copies of a 34.5-kb repeat nested within one copy of a larger 127-kb duplication. Accurate assembly of the complex repeat in B199 could be determined only with the application of Nabsys HD-Mapping. The gene content within the duplications differed between the two strains, including a locus of flagellar biosynthesis in B199 and adenylate-cyclase hemolysin (ACT) in J042 (see Data Set S3 in the supplemental material). Duplications of genes in both regions have been reported previously (42–44) and were observed in a small number of Fha-producing isolates here based on high Illumina sequencing read coverage (Table S1), suggesting that neither confers Fha deficiency. Conversely, the genome of Pt-deficient isolate J365 contained a 28-kb deletion surrounding both the Pt biosynthetic (ptx) and transport (ptl) operons. The same deletion was reported in Pt-deficient isolates from New York (I979) and France (FR3749) (45). In the genomes of B199, J042, and J365, copies of IS481 were observed flanking both ends of the duplication or deletion, suggesting that these large structural mutations were the result of recombination events.

Phylogenetic and structural ancestry.

The phylogenetic placement of identified Fha-deficient B. pertussis isolates among a representative subset of circulating and reference isolates was reconstructed from 584 variable core positions using maximum parsimony (Fig. 4). Both of the recent Fha-deficient clinical isolates J014 and J199 belonged to a monophyletic clade sharing pulsed-field gel electrophoresis (PFGE) profile CDC237. All sequenced individuals in this clade share a conserved IS481 disruption of prn at position 1613 and a common chromosome structure, without rearrangement (36). Historic laboratory strains B199, J042, and J043 did not share SNP profiles or genome structures with recent circulating isolates; rather, they more closely matched the vaccine reference strains. Specifically, J042 and J043 were related to Tohama I (E476) and B199 was closest to vaccine strain 10536 (Fig. 4). Previously reported Fha-deficient Swedish isolates B3582 and B3585 were not related, by SNP phylogeny or genome structure, to deficient isolates recovered in the United States or each other. Pt-deficient isolate J365 did share a SNP phylogeny, genome structure, and prn mutation with I979, which harbors the same ptx-ptl deletion (45).

FIG 4.

SNP phylogeny and molecular typing of Fha-deficient isolates. Phylogenetic reconstruction of Fha-deficient isolates and select references from 584 variable nucleotides using maximum parsimony. Vaccine immunogen production, gene alleles, and PFGE profile are listed next to the tree as indicated in the key. Deficient isolates sequenced as part of the current study are highlighted in red. The scale bar indicates the number of substitutions per site.

DISCUSSION

In this study, we utilized an ECL assay to measure the production of the aP vaccine immunogen Fha and screened 787 isolates, many collected recently through disease surveillance. The vast majority of recent B. pertussis isolates produced Fha, consistent with existing knowledge of its central role in persistent infection, and only two deficient clinical isolates were observed. A genome sequence comparison of the Fha-deficient isolates identified here to a large collection of Fha-producing controls revealed putative underlying mutations. The results here demonstrate the potential for Fha deficiency and illustrate parallels to the recent emergence of Prn deficiency.

Mutations identified here underlying Fha deficiency included types also known to confer Prn deficiency, such as homopolymeric tract variation and IS481 insertion, indicating that both phenotypes can derive from a common mutation spectrum, as expected. Homopolymeric tracts are prone to strand slippage during replication, and changes in repeat length modulate phenotypic phase variation in diverse bacterial species (46, 47). In B. pertussis, homopolymeric tracts appear in coding regions or promoters of numerous genes, including virulence determinants, and some polymorphisms have been linked to variable expression (48, 49). J199 and Swedish isolate B3582 (35) exhibited mutations to the same homopolymeric tract within fhaB. A similar mutation to bvgS was observed in J043 and has been reported to control phenotypic switching (50). Fha deficiency resulting from such mutations may be reversible, potentially yielding a mixed population of Fha-producing and -deficient bacteria within the infected host. Although the recovery of single isolates by culture makes this impossible to determine, polymorphisms were observed among sequencing reads from J199 and J043, suggesting that reversions to the wild type were present during laboratory growth.

Genomes of B. pertussis harbor >240 copies of IS481, and approximately 250 additional target sites for insertion have been predicted, including three each within fhaB and prn (36). Insertion at any of the positions in prn constitutes a significant source of Prn deficiency, occurring repeatedly within the circulating population (13, 36). In fhaB, insertion at either of two sites present in the β-helical shaft domain would in all likelihood abolish Fha function, consistent with the observed Fha deficiency in B199. However, the third insertion site is located within the C-terminal prodomain that is degraded during secretion (23). Although deletion of this prodomain has previously been shown to not impede production of mature Fha or prevent Fha-mediated adherence of B. bronchiseptica to the mouse respiratory tract, full-length FhaB may play other roles in persistent infection (51). Therefore, while the presence of multiple target sites suggests that fhaB is vulnerable to frequent IS481 disruption, like that to prn seen in Prn deficiency, insertion may confer varied phenotypes depending on its position. The ECL assay here detected Fha production in J365, despite IS481 disruption of the C-terminal prodomain, highlighting its limitation to screening for mature Fha production only, not all FhaB-related activity.

Some Fha-deficient isolates lacked mutations in genes in the known Fha biosynthesis pathway, such as fhaB and bvgS, but did contain specific mutations elsewhere. In contrast, all identified mutations responsible for Prn deficiency occur within the prn coding region or promoter (13, 14, 36). Fha biosynthesis is complex, regulated transcriptionally by BvgAS (21, 22) and posttranslationally through a series of degradation and translocation steps (23, 24), and so additional opportunities for indirect disruption likely exist. In J014, mutation was observed to an undescribed ECF sigma factor, a small regulator from a family of proteins that often coordinate transcriptional responses to extracellular signals (38). Such regulators have previously been shown to control virulence gene expression, such as type III secretion, in Bordetella spp. (52, 53) but not Fha production. Likewise, a unique IS481 insertion in J042 was observed within the 5′ UTR of bipA, disrupting the intricate promoter architecture (39, 40). The function of BipA remains undetermined, but the predicted protein sequence includes secretion signals and surface-associated domains (54). Although transcription of both fhaB and bipA increases rapidly upon Bvg⁺ phase activation (55), no interaction between Fha and BipA has been reported. Mutations in this novel ECF sigma factor and bipA, whose specific functions are unknown, suggest that Fha regulation is more complex than currently understood and perhaps so too are Fha deficiency phenotypes, at least compared to Prn deficiency.

A phylogenetic reconstruction of Fha-deficient isolates corroborated that each arose through independent mutation rather than transmission of common genotypes, much like the emergence of Prn deficiency. Only recent clinical isolates J014 and J199 were closely related and belonged to a monophyletic clade that shares the predominant PFGE profile CDC237 (36). Recent whole-genome sequence comparison indicated that isolates with profile CDC237 share SNP patterns, a conserved IS481 insertion into prn, and a common chromosome structure (36). Given the rarity of Fha deficiency, it is unclear whether this genetic background facilitated the loss of Fha production or, alternatively, the shared ancestry of J014 and J199 simply reflected the higher abundance of CDC237 in the circulating U.S. population of isolates (56). Analogously, the emergence and spread of Prn deficiency has primarily occurred in the predominant genetic background defined by prn2-ptxP3-ptxA1-ptxB2 (13). Previous SNP analyses of large isolate collections have indicated that the recent genetic history of B. pertussis has been punctuated by successive selective sweeps separated by periods of allelic radiation (10, 11). Perhaps the observed independent mutations in J014 and J199 represent another period of diverging variation following the sweeping expansion of CDC237. Determining whether such mutations toward Fha deficiency result from vaccine-driven immune selection will depend on future observation of additional deficient clinical isolates.

Genome reduction has been credited with B. pertussis speciation (57) and IS481 continues to reshape the genome through deletion (58, 59), a process exemplified by J365. This isolate exhibited defects in aP vaccine immunogens Prn and Pt and potentially full-length FhaB activity, which were all acquired through successive IS481-mediated mutations. However, the application of complete genome assemblies for analyses here serves as a reminder that IS481-mediated structural variation also includes large duplications, albeit more rarely. The chromosomes of historic Fha-deficient laboratory strains B199 and J042, as well as some Fha-producing isolates, contained previously reported (42, 44, 60), yet phenotypically uncharacterized, duplication of the flagellar and ACT biosynthesis loci, respectively. Importantly, such insight into the genome evolution of circulating B. pertussis through gene deletion and duplication, potentially impacting vaccine efficacy, currently depends on bacterial culture, a diagnostic practice often omitted in favor of PCR.

Conclusions.

The rapid emergence and spread of Prn deficiency raise concerns about the stable production of other immunogenic proteins used in current aP vaccines. The ECL assay described here for Fha builds on previous applications of the same platform for measuring Prn and Pt, which together enable rapid, sensitive screening to identify deficient isolates. Clinical Fha-deficient mutants are currently rare; whether this is due to the significance of Fha for persistent infection, to the absence of selection pressure elicited by aP vaccines, or to the initially limited circulation remains unanswered, given the available data. However, their observed emergence by parallel mutation within a predominant genetic background mirrors trends underlying the recent spread of Prn deficiency, which was recognized only through retrospective analyses. These results emphasize the importance of screening for production of all aP vaccine immunogens during disease surveillance.

MATERIALS AND METHODS

Bacterial strains and characterization.

Recent U.S. Bordetella pertussis isolates screened for Fha production were collected by state public health laboratories during 2010 to 2016 either through routine surveillance and outbreaks or through the Enhanced Pertussis Surveillance/Emerging Infections Program, which combines systematic case ascertainment and augmented data collection with clinical isolate submission (61), and were forwarded to the Centers for Disease Control and Prevention (CDC) in Atlanta, GA. Historic B. pertussis clinical isolates and laboratory strains were taken from the CDC collection, dating back to 1935. Laboratory strain B199 was provided by Bruce Meade of the U.S. Food and Drug Administration (FDA). Laboratory strains J042 and J043, originally from the U.S. National Institutes of Health (NIH), were given by the University of Michigan. Basic characteristics of identified Fha-deficient isolates and select references are listed in Tables 1 and 2. Additional Fha-producing isolates sequenced for mutation identification are listed in Table S1.

Pulsed-field gel electrophoresis was performed by using restriction enzyme XbaI (62), based on the method developed by Gautom et al. (63) and similar to that currently used by U.S. state health departments participating in CDC's PulseNet for the typing of foodborne pathogens (64). PFGE patterns were compared to a database of B. pertussis isolate profiles maintained at CDC, and profiles were assigned on the basis of bands in the 125- to 450-kb range using BioNumerics v5.01 (Applied Maths, Austin, TX, USA).

Electrochemiluminescent assay.

Isolates were cultured on Regan-Lowe agar without cephalexin for 72 h at 37°C. Cells were collected and suspended in 0.01 M phosphate-buffered saline (PBS) to a turbidity of 1.0, measured using a Dade Behring MicroScan Turbidity Meter (Siemens Global, Munich, Germany). Cells were pelleted by centrifugation at 10,000 × g and resuspended in Xtractor lysis buffer (Clontech, California, USA) containing 25 U of Benzonase (Sigma-Aldrich, Missouri, USA) at a 1:1 ratio (vol/vol). The mixture was vortexed gently for 10 min at room temperature and centrifuged, and lysis buffer supernatant was discarded. Cell lysate pellets were resuspended in PBS for analysis.

The electrochemiluminescent antibody capture (ECL) assay was performed as previously described (34), using the Sector S6000 platform (Meso-Scale Discovery, Maryland, USA). Briefly, 96-well plates (Meso-Scale Discovery) were coated with 30 μl of capture sheep Fha antiserum (NIBSC, UK), at a 1/33,000 working dilution and incubated overnight at 4°C. Plates were washed with PBS containing 0.05% Tween 20 and blocked with PBS containing 5% dry skim milk powder (Fisher Scientific, Pennsylvania, USA), 2% goat serum (Sigma-Aldrich, Missouri, USA), and 2% Blocker A (Meso-Scale Discovery). The blocking solution was discarded and replaced with 25 μl cell lysate. Plates were washed again, and detection was done with 25 μl of mouse anti-Fha detection antibody (NIBSC, United Kingdom), at a 1/5,000 dilution. Goat anti-mouse Sulfotag secondary antibody (Meso-Scale Discovery) was then added at 1 μg/ml. Luminescence was measured with a Sector Reader (Meso-Scale Discovery). Purified Fha protein (GlaxoSmithKline, UK) at 1 μg/ml was used as a positive control. A 2-fold dilution series with 2 μg/ml purified Fha was performed and fit with a four-parameter logistic (4-PL) regression using the instrument software to confirm the absence of nonspecific cross-reactivity. Isolates were qualitatively defined as either Fha producing or Fha deficient based on their signal intensity compared to positive and negative controls measured on the same plate to account for variability between plates (34). For any isolate producing <50% signal intensity compared to the control, the assay was repeated for confirmation.

Western blotting.

Isolates were cultured on agar medium and prepared in a manner similar to that described above. However, final cell lysate pellets were instead resuspended in SDS Sample Buffer (Bio-Rad, California, USA) and both the lysis buffer supernatant and pellet suspension were used for SDS-PAGE. Cells from Regan-Lowe agar plates were also inoculated into Stainer-Scholte broth (65) and incubated at 37°C to an optical density at 650 nm (OD₆₅₀) of approximately 2.3. Liquid cultures were then diluted 1:100 into fresh medium, incubated at 37°C to OD₆₅₀ of 0.5 to 1.0, and centrifuged at 10,000 × g. Liquid culture cell pellets were lysed in the same manner as described above, while potential proteins remaining in the broth supernatant were precipitated by addition of 100% (wt/vol) trichloroacetic acid (TCA) according to Green and Sambrook (66). TCA was removed by centrifugation at 10,000 × g, and the pellet was washed three times with acetone before resuspension in SDS Sample Buffer. Sample preparations derived from the lysis buffer supernatant, cell pellet, and liquid culture broth precipitate were all used for SDS-PAGE.

The protein concentrations of sample preparations were quantified using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE). Ten micrograms of protein was denatured by boiling for 5 min in SDS Sample Buffer, loaded to a 4 to 20% SDS polyacrylamide gradient gel (Bio-Rad, California, USA), and run for 1 h at 200 V. Separated proteins were transferred to a polyvinylidene difluoride (PVDF) membrane using the Mini Trans-Blot Cell (Bio-Rad). The ∼220-kDa Fha protein was detected using a mouse anti-Fha antibody (NIBSC, UK) at a 1:1,000 dilution in nonfat dry milk (5%) followed by 1:5,000 horseradish peroxidase (HRP)-conjugated anti-mouse antibody. The blots were applied to Clarity Wester chemiluminescent (Bio-Rad) substrate, and images were captured using an X-OMAT 2000A (Kodak, New York, USA).

Genomic DNA preparation.

Isolates were cultured on Regan-Lowe agar without cephalexin for 72 h at 37°C. Genomic DNA isolation and purification were conducted according to the Qiagen Gentra Puregene Yeast/Bacteria kit's standard protocol with slight modification (Qiagen, Valencia, CA). Briefly, two aliquots of approximately 1 × 10⁹ bacterial cells were harvested and resuspended in 500 μl of 0.85% sterile saline and then pelleted by centrifugation for 1 min at 16,000 × g. Recovered genomic DNA was resuspended in 100 μl of DNA hydration solution. Aliquots were quantified and qualified using a Nanodrop 2000 instrument (Thermo Fisher Scientific Inc., Wilmington, DE).

Genome sequencing and assembly.

Whole-genome shotgun sequencing was performed using a combination of the PacBio RSII (Pacific Biosciences, Menlo Park, CA), Illumina HiSeq/MiSeq (Illumina, San Diego, CA), and Argus (OpGen, Gaithersburg, MD) platforms as described previously (12). Briefly, genomic DNA libraries were prepared for PacBio sequencing runs using the SMRTbell Template Prep kit 1.0 and Polymerase Binding kit P4 or P6, while Illumina libraries were prepared using the NEB Ultra Library Prep kit (New England BioLabs, Ipswich, MA). De novo genome assembly of filtered reads was performed using the Hierarchical Genome Assembly Process (HGAP, v3; Pacific Biosciences) (67) and, when necessary, the A-Hybrid-Assembly workflow (AHA; Pacific Biosciences). The resulting consensus sequences were determined with Quiver (v1), manually checked for circularity, and then reordered to start at the coding region for glucose-inhibited cell division protein A (gidA), consistent with the available genome sequences of B. pertussis. Assemblies were confirmed by comparison to restriction digest optical maps using the Argus system (OpGen) with MapSolver (v.2.1.1; OpGen). Final PacBio assemblies were further polished by mapping either Illumina HiSeq PE-100, MiSeq PE-150, or MiSeq PE-300 reads using CLC Genomics Workbench (v10.0.1; CLC bio, Boston, MA) and annotated using the NCBI automated Prokaryotic Genome Annotation Pipeline (PGAP). Whole-genome shotgun sequencing of Fha-producing isolates was performed using Illumina HiSeq/MiSeq only, and reads were quality trimmed and filtered with CLC Genomics Workbench (CLC bio) as described previously (12).

Duplication assembly in J042 was resolved manually with the aid of BridgeMapper (v1; Pacific Biosiences) and MapSolver (OpGen). The more complex duplication in B199 was manually assembled based on Nabsys HD-Mapping (Nabsys 2.0 LLC, Providence, RI). Following culture as described above, genomic DNA isolation was performed at CDC according to a Nabsys solution-based protocol modified from the bacterial DNA protocol for AXG 20 columns and the NucleoBond Buffer Set III (Macherey-Nagel, Bethlehem, PA). Briefly, modifications to reduce DNA shearing included increasing lysate incubation to 60°C for 60 min, reducing vortex mixing to 5 s, and using wide-bore pipette tips. Purified DNA was sent to Nabsys for nicking, tagging, coating, and processing on an HD-Mapping instrument using the enzymes Nb.BsmI, Nt.Bpu10I, and Nt.BspQI. The resulting de novo-assembled HD maps, raw data, and data remapped to PacBio de novo assemblies were provided by Nabsys for further analyses and sequence assembly confirmation at CDC with NPS Analysis (v1.2.1661; Nabsys). Duplication assembly in both J042 and B199 was further confirmed by successfully mapping filtered PacBio sequencing reads of >10 kb across each of the novel duplication boundaries using BLASR (v1) (68) and visualized with SMRT View (v2.3.0; Pacific Biosciences).

Genome sequence and structure variation.

Phylogenetic reconstruction of assembled genomes and related references was calculated using kSNP3 (69), with k of 23 and maximum parsimony of core SNPs after masking the pertactin gene (prn) and all IS element sequences with ‘N′s. Internal nodes with <50% bootstrap support were collapsed into multifurcations in Archaeopteryx v0.9901 (70), and tree annotation was performed with iTOL v3.0 (71). Variation in genome structure was detected by pairwise alignment of assembled genomes with closely related reference strains using progressiveMauve (72) with optimized parameters (−seed-weight = 16; −hmm-identity = 0.85).

Sequence variation was detected by mapping filtered Illumina reads from Fha-deficient and Fha-producing isolates to the reference strain E476 (CP010964) using the Basic Variant Detection tool in CLC Genomics Workbench (CLC bio) as described previously (12). Mutations putatively responsible for disrupting Fha production were identified by comparing the variant profile of each Fha-deficient mutant to those found in 148 Fha-producing isolates. Predicted proteins containing nonsynonymous mutations were further annotated with a betaproteobacterium-specific subset of EggNOG v4.1 (73) using HMMER v3.1bs (http://hmmer.org) as well as manual query of Swiss-Prot (74) and the Conserved Domain Database (75) using Delta-BLAST (76) via the NCBI web interface.

Mutations to the pertactin gene were identified in assembled genomes by alignment to a curated set of previously reported wild-type and deficient alleles (13) using BLASTn (77) with cutoffs of minimum sequence identity of 100% and query coverage of >99%. Alleles of fimH (fim3) were assigned in a similar manner.

Accession number(s).

The complete genome sequences determined in this study have been deposited at DDBJ/EMBL/GenBank under the accession numbers CP011245, CP012135, CP013867, CP016887, CP019869, and CP022361. The versions described in this paper are the first versions. Raw sequencing data have been deposited under the accession numbers listed in Table S1 in the supplemental material, organized under a BioProject with accession number PRJNA279196.