Molecular characterization and epidemiology of Streptococcus pneumoniae serotype 24F in Denmark

Since 2012, have we in Denmark observed an increase of invasive pneumococcal infections (IPD) due to Streptococcus pneumoniae serotype 24F. We here present epidemiological data on 24F IPD cases, and characterization of 48 24F clinical isolates based on clonal relationship, antimicrobial resistance (AMR) determinants and virulence factors. IPD surveillance data from (1999–2016) were used to calculate the incidence and age-distribution of serotype 24F IPD and the effect of pneumococcal conjugated vaccines (PCV). Characterization of forty-eight 24F isolates (14.7% of all 24F isolates from the period) was based on whole-genome sequencing analysis (WGS). The IPD cases of serotype 24F showed a significant increase (p < 0.05) for all age groups after the PCV-13 introduction in 2010. The majority of tested 24F isolates consisted of two MLST types, i.e. the ST72 and the ST162. Serotype 24F IPD increased in Denmark after the PCV-13 introduction in parallel with an increase of the ST162 clone. The genotypic penicillin binding protein (PBP) profile agreed with the phenotypical penicillin susceptibility. The virulence genes lytA, ply, piaA, piaB, piaC, rspB and the cpsA/wzg were detected in all 24F isolates, while the pspA and zmpC genes were absent.

to monitor the epidemiology and susceptibility of serotype 24F, and to provide information to the international community that measures against serotype 24F are needed.
The intention of this study was to present epidemiological data based on IPD cases in Denmark in the period from 1999 to 2016. Furthermore, to present a detailed characterization of forty-eight 24F isolates (representing 14.7% of all 24F isolates from the period) using whole-genome sequencing analysis (WGS). Thus, the serotype 24F's clonal relationship over the years and antimicrobial resistance determinants are presented. Finally, the presence of capsular, toxin and surface related genes are suggested as pneumococcal species.
Material and Methods strain collection. All invasive pneumococcal isolates of serotype 24F from 1999 to 2016 were retrieved from the Danish laboratory surveillance system at the national Neisseria and Streptococcus Reference Laboratory (NSR), Statens Serum Institut (SSI) (Supplementary Table 1). All the IPD serotype 24F cases reported were isolated form patients diagnosed with either bacteraemia (from blood) or meningitis (from cerebrospinal fluid). An IPD case was defined as the occurrence of S. pneumoniae in cerebrospinal fluid, blood or other normally sterile sites 9 . Incidence and Incidence Rate Ratio (IRR) data from 24F from 1999 to 2014 was previously presented in Slotved et al. 1 including information on age, sex, serotype and origin of the pneumococcal isolate (blood, cerebrospinal fluid etc.). Data on the total IPD cases for all age groups in Denmark are presented in Fig. 1 to compare with the serotype 24F. Detailed data on the total IPD cases for all age has previously been presented 1,10 . Identification of pneumococcal isolates. The pneumococci isolates were phenotypically identified by optochin susceptibility and bile solubility tests 1,11 . All isolates were serotyped either by the Quellung reaction alone or by the Pneumotest Latex kit (SSI-Diagnostica, Copenhagen, Denmark) combined with the Quellung reaction using type-specific pneumococcal rabbit-antisera as previously described (SSIDiagnostica, Copenhagen, Denmark) 11 .
Characterization of 48 clinical selected isolates. Forty-eight Streptococcus pneumoniae 24F isolates were selected for detailed characterization using whole genome sequencing (Table 1). Forty-seven 24F isolates were isolated from blood and spinal fluid in the period 1999 to 2017, and one historical isolate 24F was cultured from the trachea of an eight-year-old boy in 1943. The 47 isolates were selected to represent 26 children below five years of age and 21 elderly persons above 64 years of age. Otherwise the isolates were randomly selected during the period 1999-2017 to provide a molecular picture of the serotype 24F isolates in Denmark. The isolates represent 14.7% of the total number of 24F IPD cases during the period.
Molecular species identification. The isolates were sequenced by paired-end Illumina sequencing.
All 48 isolates were species identified according to the description by Scholz et al. 12 . Briefly, the 16S rRNA sequence (accession number: AY485600) identified by Arbique et al. 13 was used to identify the nucleotide position at 203 12 . The S. pneumoniae identification was based on the location of cytosine at the 203 position, while the existence of adenine-residues suggested that the species belonged to another Streptococcus species 12 .
The presence/absence of a gene was based on a cut-off of 80% coverage and a 95% identity for a positive gene detection in this study 14 .
The genomic sequence data for the 48 isolates are deposited in the Genbank (https://www.ebi.ac.uk/ena) (ENA project no. PRJEB31691).

Molecular characterization of their capsular genes.
Sequences from all 48 isolates were checked for the 92 capsular polysaccharide genes (CPS genes) by BLAST. The FASTA files for the capsular locus sequences were retrieved from the NCBI database, using the accession numbers CR931632-CR931722, JF911515.1 and HV580364.1 17,18 .
Identification of serogroup/type 24F was performed according to the presence/absence of genes using a cut-off of 80% coverage and a 95% identity as described by Sheppard et al. 14 and Kapatai et al. 18 . However, differentiating further into a specific serotype within the group 24 can be difficult 18 . We therefore only presented genotypes, where the coverage and identity clearly indicated a specific genotype.

Multilocus sequence typing (MLST).
MLST was performed using the PubMLST DataBase (https:// pubmlst.org/spneumoniae/) to identify the sequence type (ST) for each of the 24F S. pneumoniae strains. Analysis of the STs and assignment to CC was performed using PHYLOViZ 2.0 programme (http://phyloviz.readthedocs. io/en/latest/#). The STs that shared at least six of seven allelic variants composed a CC (clonal complex) 19 .
A phylogenetic tree based on single nucleotide polymorphisms (SNP's) analysis of the core genome was performed on the 48 isolates. Identification of SNP's' was performed using BWA-mem for mapping and GATK with filtering set to remove positions with less than 10-fold depth and 90% unambiguous variant calls as implemented in NASP 20 against isolate 0100-2014's chromosome, which was used as a reference strain in the SNP alignment after removal of duplicated regions using NUCmer. The resulting SNP matrix was purged for recombination using Gubbins 21

Antibiotic susceptibility testing (phenotypic tests). Screening of antibiotic susceptibility was per-
formed on all 48 isolates by disk diffusion using Mueller-Hinton 5% blood agar with NAD (Oxoid, Denmark) incubated in ambient air with 5% CO 2 at 35 °C and oxacillin, penicillin, erythromycin and clindamycin discs (Oxoid, Denmark). Some isolates were also tested using the E-test randomly, however not on a regularly basis.
MIC determination of penicillin G was done by using either a gradient test (Etest; bioMérieux), before 2010, or a broth microdilution method (Sensititre, Streptococcus species MIC Plate, STP6F, Trek Diagnostic System, USA), after 2010. S. pneumoniae ATCC 49619 was used as a quality control strain. Interpretation of susceptibility was done according to the breakpoints described in EUCAST (http://www.eucast.org/clinical_breakpoints/).
We do not routinely perform phenotypical screening of tetracycline on the isolates.
Genotypic antibiotic resistance profile. Penicillin (PEN) susceptibility in pneumococci is associated with penicillin-binding proteins (PBP) which in penicillin non-susceptible strains of pneumococci are modified to low-binding-affinity versions of the native PBP1A, PBP2B and PBP2X. The 48 isolates were analyzed for their PBP signature, based on a genotyping proposal and algorithm described for PBP1A, PBP2B and PBP2X 22 , where the combination of the three PBP signatures determines the level of beta-lactam resistance. The 48 isolates were tested by BLAST with the published types of predictive mutations vs. resistance levels of PBP1A, PBP2B and PBP2X proteins as described in Li et al. 22 (Table 1). Virulence gene profile. Presence of capsular, toxin and surface related genes used for pneumococcal virulence characterization 25 and presence of genes with potential species discrimination: The presence of the virulence genes lytA and ply 26 for the identification of S. pneumoniae were tested as recommended by Centers for Disease Control (CDC) 12,27 .
Genes zmpB 28 and zmpC (AE005672:75858-76420) 29 , which are paralogous zinc metalloproteases 30 . Genes for the pneumococcal surface proteins (pspA) (Genbank: AF516671) 31 and psrp 32 . The partial capsular gene cpsA, also known as the wzg gene, was tested for its presence/absence (Genbank: AF057294:2134-2473) 33 . ethical considerations. The study was a retrospective, population-based study based on national laboratory surveillance data on isolates from patients with IPD. Since data and samples from patients were collected routinely for national surveillance purposes, no ethical approval or informed consent from patients or guardians were required. The study was approved by the Danish Data Protection Agency (record number 2007-41-0229).

Results
Prevalence/incidence of invasive pneumococcal disease in 1999-2016 due to S. pneumoniae serotype 24F. Data from the incidence rates of serotype 24F IPD cases from 1999 until 2016 are presented in Fig. 1 and Table 2. The incidence of serotype 24F IPD cases was low in the period from 1999-2007 (0. 22  did not affect the 24F incidence in any of the age groups, while the introduction of PCV-13 in 2010 showed a significant increase in prevalence, two years after the introduction. The increase in the IRR of serotype 24F IPD varied with an IRR of 3.69 (CI: 1.30-10.53) in infants (P = 0.0083), 3.78 (CI: 1.73-8.32) in the age group from 5 to 64 years ((P < 0.001) and 1.74 (1.74 CI: 1.08-2.80) in the age group +65 (P = 0.025), although the general mean incidence was very low.
The general IPD incidence for all age groups and serotypes in Denmark was reduced since the introduction of PCV-7 (Fig. 1).

Identification of the capsular genes for serotype 24F.
While all isolates were confirmed phenotypically to be serotype 24F, it was only possible to correctly identify the genotype to group level for 20/48 serotype 24F isolates. For the 28 remaining isolates the 24F capsular genes were correctly identified (Table 1).
Phylogenetic analysis by MLST. The ST72 belonging to Clonal complex (CC) 72 prevailed among the others STs and included 23 isolates. This was followed by ST162 belonging to CC156 and consisting of 14 isolates.
MLST sequence types correlated to clade relationships depicted in the core SNP phylogeny (Fig. 2) except for isolate 1186-2004.
Characterization of pneumococcal virulence genes. The virulence genes lytA and ply, the combined genes piaA, piaB, and piaC for membrane and ATP-binding proteins, the rpsB gene encoding for ribosomal protein S2, and the partial capsular cpsA (wzg) gene were detected in all 48 isolates.
The zinc metalloprotease related gene zmpB was only detected in isolates from 2007 and forwards. The surface protein related gene psrp was most commonly found in isolates before 2012. Regarding the presence of zmpB and psrp, forty-one isolates harbored only one of the genes while seven isolates did not show any of the genes, and none harbored both genes.
The pneumococcal surface protein A gene (pspA) and the zinc metalloprotease zmpC gene were absent from all isolates.

Discussion
With the introduction of PCV-7 in 2007 in Denmark in the children's vaccination program, a reduction in IPD with PCV-included serotypes was observed. However, as noted in other countries, non-vaccine serotypes have emerged, and it is therefore important to monitor the appearance of replacement serotypes 1,5,7,34 . Serotype 24F has been observed to be one of the emerging non-PCV serotypes 5,7,28 . Although serotype 24F is only among the 20 most common cause of IPD in Denmark, it is one of the serotypes which have increased most after the PCV-13 introduction in Denmark 1 . Serotype 24F is described as a type with a high potential for invasive disease 5 , and it is therefore an important serotype to keep under surveillance. Figure 1 shows that the total incidence of 24F has been relatively steady until 2012, when a significant increase in serotype 24F IPD cases were observed with a peak in 2014 (Table 2 and Fig. 1). In general, the total IPD in Denmark has been reduced since the introduction of the PCV (Fig. 1) 1,10 .
The analyzed MLST data show, that the majority of the isolates belong to ST72 and ST162. According to the MLST database, ST72 is a well-known clone of serotype 24F. The distribution between the previously described susceptible ST72 and resistant ST162 7 was also observed in this study (Table 1). ST72 belonged to the penicillin susceptible lineages of CC72, while ST162 (belonging to CC156) also penicillin-susceptible was observed to be cotrimoxazole-resistant and found also among 9 V isolates 7 . The CC156 has emerged after the PCV-13 era 7 , which is also in accordance with our observation of the increase in ST162 ( Table 1). The ST162/CC156 are furthermore described as the PMEN3 clone (Spain 9V -156), particular related to the PCV included serotype 9V and serotype 14 35 . It has been suggested that the appearance of PMEN3 clone of serotype 14 might be due to capsular switching 35 , which also might be the situation with the serotype 24F ST162/CC156 isolates observed in this study. www.nature.com/scientificreports www.nature.com/scientificreports/ The ST162 lineage is described as a more successful lineage than the CC72 and the multidrug resistant lineage CC230 7 (Janoir et al. 2016).
Four of the 48 isolates were CC230, one ST230, two ST4253 and a new ST profile (isolate 0373-2003). As described by Janoir et al. 7 , CC230 is known as a highly resistant clonal complex, which we can confirm in this study ( Table 1). The CC230 has previously been described in relation to a Danish penicillin resistant serotype 14, and are referred to as Denmark 14 -32 PMEN clone. It has also been suggested that the CC230 serotype 24F might be due to capsular switching from a serotype 14 (https://www.pneumogen.net/pmen/, accessed 24th September 2018) 36 . Figure 2 shows that the clustering of isolates based on the SNP site location corresponded well with the MLST.
The historical trachea isolate 24F-1943 was found to harbor a new PBP signature (0-0-0) 22 . The isolate did not show any clonal relationship with the other forty-seven 24F isolates from the period 1999-2017, and it only showed some relation to one other single isolate (isolate 1853-2011) based on the SNP site location (Fig. 2). It is well-known that capsular switching occurs regularly among pneumococcal isolates 36 , and it is therefore a possible explanation that the 24F isolates we see today in Denmark are due to capsular switching in other serotypes 7,36 . In the study by Li et al. 22 , penicillin susceptibility could be predicted by the signature of the three penicillin-binding proteins PBP1a, PBP2b, and PBP2x. We also found an excellent correlation between the signature of the three PBPs and the phenotypical penicillin susceptibility. Interestingly, the four penicillin non-susceptible isolates all had the 17-15-x signature, while all susceptible isolates had low-numbered PBP signatures.
Phenotypic resistance to erythromycin/clindamycin was in general agreement with the genotypic resistance and presence of relevant genes ( Table 1). The presence of the ermB gene was linked to resistance toward erythromycin and clindamycin and not to the mefA and mefE genes. The 12 isolates which were positive for the ermB gene, also harbored the tetM gene. Of note, two isolates harboring both the ermB gene and the tetM gene were erythromycin susceptible, which has been seen by others 37 .
An interesting observation was made regarding the two genes pdrp and zmpB ( Table 3). None of the 48 isolates harbored both the psrp gene and the zmpB gene; most of the isolates had only one of the genes, while for seven isolates the two genes were absent. Interestingly, psrp was detected in the isolates before 2012. The psrp gene has been described as a gene found primarily in antibiotic susceptible strains 36 , which we also observed. Only the MLST ST7179 and the well-known susceptible MLST ST72 strain harbored the psrp gene (Table 3). Although the zinc metalloprotease related gene zmpB is described as being widespread in S. pneumoniae 30 , we only detected the gene in Danish isolates from 2007 and onwards. The zmpB gene was only detected in the non-susceptible strains, particularly in ST162 (Table 3). We have not been able to find any studies showing that the zmpB gene is linked to antibiotic susceptibility. A 24F isolate with the zmpB gene was first detected in 2007 and became more common in the 24F isolates with the increase in the 24F incidence in 2013 (Fig. 1).

Conclusion
We have seen an increase in serotype 24F IPD in Denmark after the introduction of the PCV-13 vaccine in 2010. It was not significantly associated with an increase in antibiotic resistance or virulence determinants, but was observed in parallel with an increase of the ST162 clone.