Rapid and simple SNP genotyping for Bordetella pertussis epidemic strain MT27 based on a multiplexed single-base extension assay

Multilocus variable-number tandem repeat analysis (MLVA) is widely used for genotyping of Bordetella pertussis, the causative bacteria for pertussis. However, MLVA genotyping is losing its discriminate power because prevalence of the epidemic MT27 strain (MLVA-27) is increasing worldwide. To address this, we developed a single nucleotide polymorphism (SNP) genotyping method for MT27 based on multiplexed single-base extension (SBE) assay. A total of 237 MT27 isolates collected in Japan during 1999–2018 were genotyped and classified into ten SNP genotypes (SG1 to SG10) with a Simpson’s diversity index (DI) of 0.79 (95% CI 0.76–0.82). Temporal trends showed a marked increase in the genotypic diversity in the 2010s: Simpson’s DI was zero in 1999–2004, 0.16 in 2005–2009, 0.83 in 2010–2014, and 0.76 in 2015–2018. This indicates that the SNP genotyping is applicable to the recently circulating MT27 strain. Additionally, almost all outbreak-associated MT27 isolates were classified into the same SNP genotypes for each outbreak. Multiplexed SBE assay allows for rapid and simple genotyping, indicating that the SNP genotyping can potentially be a useful tool for subtyping the B. pertussis MT27 strain in routine surveillance and outbreak investigations.


Scientific Reports
| (2021) 11:4823 | https://doi.org/10.1038/s41598-021-84409-0 www.nature.com/scientificreports/ Previously, a simple SNP genotyping method with 38 SNP targets was developed for B. pertussis using a single-base extension (SBE) assay 28 . The SBE assay is based on the incorporation of fluorescently labeled dideoxynucleotides (ddNTPs) into the 3′ end of allele-specific extension primers with a distinct length, and subsequent analysis with a capillary DNA sequencer. This assay allows for rapid and high-throughput genotyping. However, the SBE-based SNP genotyping had low discriminatory power for B. pertussis epidemic strain MT27, since the 38 SNP targets were selected based on the WGS data of various B. pertussis strains 28 . This limitation identifies the need for selecting polymorphic SNP markers among MT27 isolates in order to improve the discriminatory power for the MT27 strain.
In the present study, in order to easily subtype B. pertussis epidemic MT27 strain, we screened for polymorphic SNPs among MT27 isolates using their genome data and developed a novel SNP genotyping with 20 SNP targets based on multiplexed SBE assay. This genotyping system was evaluated with 237 B. pertussis MT27 isolates, and its applicability to outbreak investigations was assessed with outbreak-associated MT27 isolates.

Methods
Isolates and DNA preparation. We studied 237 B. pertussis MT27 isolates collected in Japan during 1999-2018. They were all of the MT27 isolates stored in the National Institute of Infectious Diseases (NIID) strain collection, which included epidemiologically related outbreak-associated isolates (Supplementary Table S1). The isolates were cultured on cyclodextrin solid medium (CSM) agar 29 and incubated at 36 °C for 2-3 days. DNAs were extracted from the isolates by boiling and stored at − 20 °C. For whole-genome sequencing (WGS), DNAs were purified using the NucleoSpin Tissue kit (Macherey-Nagel, Germany). Novogene (Beijing, China) performed the whole-genome sequencing experiment.  Table S1). The average coverage depth of the sequencing was more than 200 × for each isolate. The sequence data were submitted to the DDBJ Sequence Read Archive (DRA) (accession no. DRA007914). All the isolates were epidemiologically unrelated cases of pertussis.

WGS. Twenty
Identification and selection of SNPs. WGS reads were mapped to the reference genome sequence of B.
pertussis Tohama I (accession no. NC_002929.2) by the Burrows-Wheeler Aligner (BWA) software, and SNPs were detected by using the SAMtools software. A total of 269 SNPs were identified in coding sequences (including pseudogenes) between 20 sequenced isolates and the reference strain Tohama. Among the 269 SNPs, we selected a set of 20 informative SNPs (representing SNP frequencies of 7.8-49.0%) for single-base extension (SBE) assay based on 51 genome sequences of MT27 isolates: 20 sequences from this study and 31 from a previous study 17 (Supplementary Table S2). General information on the 20 SNPs is given in Table 1.
Of the 20 SNP targets, 10 were coding SNPs, 9 were silent SNPs, and the remaining one was a genome SNP located in the pseudogene BP1610 (Table 1). SNP11 located in gyrA was associated with quinolone resistance in B. pertussis 30 , while SNP16 and SNP26 were previously identified as unique SNPs to epidemic isolates of Australian B. pertussis 25 . All 20 SNP targets were found in previous studies 17,18,25,28 . SNP genotyping with SBE assay. Twenty selected SNPs were divided into two groups and typed in two 10-plex PCR assays termed SNP20A and SNP20B panels . The SNP20A panel targeted SNP2, SNP11, SNP12,  SNP14, SNP16, SNP18, SNP19, SNP20, SNP25, and SNP26, whereas the SNP20B targeted SNP8, SNP15, SNP17,  SNP22, SNP24, SNP28, SNP30, SNP31, SNP32, and SNP34 (Table 2). Each 10-plex PCR was performed in a 15-μl reaction volume containing 7.5 μl of 2 × PCR buffer for KOD FX, 0.33 U of KOD-FX DNA polymerase (TOYOBO, Co., Ltd., Japan), 3.3 µl of 2 mM dNTPs, 1 µl of DNA sample, and 1.5 µl of primer mixture (each one at 2 µM). The primer sequences for SNP20A and SNP20B panels are shown in Table 2. PCR conditions were as follows: denaturation for 2 min at 94 °C; 35 cycles with denaturation at 98 °C for 10 s, primer annealing for 30 s at 60 °C (SNP20A) or 63 °C (SNP20B), and extension at 68 °C for 30 s; and final extension at 68 °C for 5 min. Following the PCR assay, the PCR products were treated with 5 U of shrimp alkaline phosphatase (New England Biolabs) and 2 U of exonuclease I (New England Biolabs) at 37 °C for 60 min, followed by incubation at 75 °C for 15 min for enzyme inactivation.
Each 10-plex SBE assay was carried out in a 10-µl reaction volume containing 2.5 µl of SNaPshot Multiplex Ready Reaction Mix (Applied Biosystems), 2 µl of 5 × sequencing buffer (Applied Biosystems), 0.7 µl of cleaned PCR products, and optimized concentrations of ten SBE primers. The sequences and concentrations of SBE primers for SNP20A and SNP20B panels are listed in Table 3. Each SBE reaction was performed for 25 cycles of denaturation at 94 °C for 10 s, primer annealing at 50 °C for 5 s, and extension at 60 °C for 30 s. Unincorporated fluorescently labeled ddNTPs were removed by addition of 1 U of shrimp alkaline phosphatase and incubation at 37 °C for 60 min, followed by incubation at 75 °C for 15 min to inactivate the enzyme. The SBE products (0.5 µl) were mixed with 9 µl of Hi-Di formamide and 0.5 µl of GeneScan 120 LIZ size standard (Applied Biosystems). After heat denaturation for 5 min at 95 °C and rapid cooling on ice, the SBE products were separated using the Applied Biosystems 3130xl Genetic Analyzer with dye set E5. The data analysis was performed with GeneMapper software (Applied Biosystems), and the resulting data were concatenated to produce a 20-position SNP profile for each isolate tested.
In reference to the 51 sequenced isolates, all 20-position SNP profiles identified by the SNP genotyping were identical to those obtained from WGS data. The SBE-based SNP genotyping showed high reliability.

Relationship between SNP genotypes and virulence-associated allelic genes. Seven allele pro-
files were identified in the MT27 isolates tested. The isolate carrying ptxP3/ptxA1/prn2/fim3A was found to be predominant in all SGs ( Table 4). The isolates carrying other allele profiles were in SG1, SG7, and SG10, but the numbers of isolates were much smaller compared with isolates carrying the predominant allele profile. There were no SGs predominantly related to the minor allele profiles; however, all isolates carrying fim3B were in SG1.
Application of SNP genotyping to outbreak-associated isolates. MT27 isolates from pertussis outbreaks were genotyped (Table 5). Fifteen outbreak-associated isolates were collected in Miyazaki prefecture 36 , and all exhibited the same genotype of SG2. Similarly, all isolates from Toyama and Niigata prefectures (n = 4 each) belonged to the same SG10. Of the 12 isolates from Nagano prefecture, 1 and 11 were SG5 and SG7, respectively. The SG5 strain was not closely related to the SG7 strain since the SG5 strain had eight different SNPs in the 20-position SNP profile as compared with the SG7 strain. Together, all except one isolate were classified into the same SNP genotypes for each outbreak.  Table S4). Ninetytwo isolates (89%) belonged to SG1 and the remainder (11%) to SG3, SG7, and SG10. The isolates belonging to SG3, SG7, and SG10 were recent isolates collected in the 2010s. Overall, the genotypic diversity was much lower than that of MT27 isolates.   Fig. S1). The genotypic diversity of the US isolates markedly increased in the 2010s, similar to that of Japanese MT27 isolates.

Discussion
We developed an SBE-based SNP genotyping for the B. pertussis epidemic strain MT27 and evaluated its applicability. The data presented here show that Japanese MT27 isolates were subdivided into ten SNP genotypes and that the genotypic diversity of MT27 isolates markedly increased in the 2010s. Moreover, almost all outbreakassociated MT27 isolates were classified into the same SNP genotypes for each outbreak. The SNP genotyping method allows for subtyping of the recently circulating MT27 strain in routine surveillance and outbreak investigations. This is supported by analyses of Taiwanese and US isolates.
In this study, we demonstrated that the genotypic diversity with 20 SNP targets rapidly increased among Japanese MT27 isolates in the 2010s. One possible cause for the increased diversity is the pertussis epidemic. In Japan, a nationwide epidemic of pertussis occurred between 2008 and 2010, and the frequency of the MT27 strain increased during the epidemic period 4 . In this study, the SG1-MT27 strain was predominant in 2005-2010, and the genotypic diversity of the MT27 strain increased markedly in the 2010s (Fig. 3). These observations indicate that the SG1-MT27 strain was rapidly replaced with other genotypes (non-SG1) during and after the epidemic. We therefore speculate that B. pertussis has evolved by increasing diversity during pertussis epidemics. A previous study showed that the B. pertussis population (genotype prevalence) increased with changes in vaccine usage (coverage and schedule) 17 . In Japan, acellular pertussis vaccines (ACVs) were introduced in 1981 instead of whole-cell pertussis vaccines, and there were no changes in the vaccine usage after 1996. Therefore, ACVs were not directly associated with the increased genotypic diversity.  www.nature.com/scientificreports/ Interestingly, the rapid increase in SNP diversity was also observed among the US MT27 isolates in the early 2010s (Supplementary Fig. S1). In Japan and the US, the frequency of the SG1 strain decreased with time, whereas that of the SG7 strain markedly increased. The Japanese SG7 isolates (n = 53) were collected from 7 out of 8 districts, indicating that the emergence of the SG7 strain occurred nationwide. Similarly, the US SG7 isolates (n = 20) were collected from 13 states 16 . These observations suggest that the SBE-based SNP genotyping may also apply to recently circulating MT27 strain collected in countries other than Japan. In fact, we confirmed here that the SBE-based SNP genotyping was applicable to recent Taiwanese MT27 isolates. For the Japanese and US SG7 strains, our preliminary analysis with whole-genome SNPs shows that most US SG7 isolates are slightly different from Japanese SG7 isolates (inter-clade SNP distance, 4 SNPs), but one isolate was found to be 100% identical to Japanese SG7 isolates. Further genetic studies are needed to characterize the increasing non-SG1 strains, especially SG7.
The B. pertussis population has significantly changed worldwide in the last 60 years 18 . Strains carrying the virulence-associated allelic genes ptxP3 and fim3A have emerged and expanded, and those carrying ptxP1 and fim3A had decreased. More recently, ptxP3 strains carrying fim3B (alias fim3-2) have expanded. In this study, most Japanese MT27 isolates carried ptxP3 and fim3A and were classified into 10 SGs based on the 20-position SNP profile (Supplementary Fig. S2). In contrast, all MT27 isolates carrying fim3B (n = 21) were interestingly grouped into only SG1, implying that the fim3B strains are not included in non-SG1 groups. The fim3B strain has increased globally and has the potential to cause recent pertussis epidemics 4,23 . Our SBE-based SNP genotyping could be helpful to identify the emerging fim3B strain.
Here, whole-genome SNP genotyping showed that non-SG1 isolates clustered into each group of the SBEbased SNP genotypes, but SG1 isolates did not (Fig. 2). Most SG1 isolates were collected in the 2000s, whereas non-SG1 isolates were in the 2010s (Fig. 3). This suggests that the new non-SG1 strains have clonally expanded, but their genetic diversity may further increase with time, similar to the old SG1 strain. Thus, continuous genome surveillance is required for accurate subtyping of B. pertussis MT27 strain, especially for non-SG1 strains.
In the present study, we applied the SBE-based SNP genotyping to outbreak-associated MT27 isolates. Among 12 isolates from the Nagano outbreak, one isolate had a genotype (SG5) different from those of the other 11 isolates (SG7) ( Table 5). In the 20-position SNP profile, 8 SNPs were different between the SG5 and SG7 isolates. This indicates that the SG5 isolate was collected from a sporadic case that was not associated with the outbreak. In outbreak investigations, strain typing contributes to understanding bacterial transmission route(s), and rapid typing is important to take countermeasures to prevent further spread of the bacteria. Our SBE-based SNP genotyping is amenable to high-throughput analysis using 96-well plates. Starting from DNA extracts, the SBE-based SNP genotyping was able to analyze 96 isolates within two days, contributing to rapid genotyping, which would be critical in outbreak investigations involving large numbers of isolates.
Culture of B. pertussis has limited sensitivity for previously vaccinated persons, older children, adolescents, and adults 37,38 . Therefore, molecular strain typing (MLVA and/or MLST) is performed not only on the bacterial isolates, but also on DNA extracts from clinical specimens such as nasopharyngeal swabs 22,[39][40][41] . Our SBE-based SNP genotyping includes an initial PCR step that amplifies the region around each SNP target. We, therefore, tested the applicability of the SNP genotyping to clinical specimens. Of the 20 clinical specimens that were positive for B. pertussis by a nucleic acid amplification test, 13 (65%) were genotyped by the direct SNP genotyping (Supplementary Table S6). The remaining seven clinical specimens had lower bacterial loads than those of the genotyped specimens (mean Ct values, 23.7 versus 20.4; P < 0.01, Mann Whitney U test), indicating that improvements in the initial PCR conditions (including primer design) are required to increase the analytical power. Although there is room for improvement, SBE-based SNP genotyping has the potential to be directly applicable to clinical specimens.
In this study, we also tested the applicability of the SBE-based SNP genotyping to B. pertussis non-MT27 strains. Our data demonstrated that most Japanese non-MT27 isolates (89%) belonged to SG1, showing low genotypic diversity (Simpson's DI, 0.21) (Supplementary Table S4). In contrast, high genotypic diversity was seen for Taiwanese non-MT27 isolates (Simpson's DI, 0.73), although the number of isolates tested was small (Supplementary Table S5). Further analyses are needed for the applicability of the SBE-based SNP genotyping to non-MT27 isolates. A limitation of our SNP genotyping targeting 20 SNPs is that its discriminatory power is lower than that of whole-genome SNP genotyping (targeting over 250 SNPs). Therefore, our simple SNP genotyping cannot provide true genetic relationships among MT27 isolates, especially for SG1 strain of MT27; whole-genome SNP genotyping is required for this purpose.
In conclusion, the present study describes the successful development of a simple and rapid SNP genotyping for the subtyping of B. pertussis MT27 isolates. Since the MT27 strain is the predominant strain in most industrialized countries, our SNP genotyping can serve as a novel alternative to whole-genome SNP genotyping in routine surveillance and outbreak investigations.

Data availability
All data generated or analysed during this study are included in this published article and its Supplementary Information files.