The bacterial strains used in this study are listed in Table 1. The strain HN105 was isolated from the knee joint fluid of a diseased pig in Henan province, China in 2014. The SS2 virulent strain ZY05719 was isolated in 2005 from a pig with acute septicemia during an epidemic in Ziyang, China (Zhang and Lu, 2007). The reference virulent strain P1/7 was originally isolated from a diseased pig with meningitis (Wu et al., 2014b). All strains were grown in Todd-Hewitt broth (THB, BD) or on agar at 37°C with 5% CO₂. Chloramphenicol (5 μg/ml) was used, when necessary. Total genomic DNA was extracted using an Omega Bacteria DNA Kit (OMEGA, China) according to the manufacturer’s instructions. All primers used in this study are listed in Supplementary Table S1.

Zebrafish and BABL/c mice infections were performed according to previous studies (Wu et al., 2014a; Dong et al., 2015). The S. suis strains were grown to mid-log phase (about OD₆₀₀ = 0.6), washed three times with 1× PBS (pH 7.4), and diluted to different doses in PBS before injection. Eight BALB/c (5-week-old) mice or 15 zebrafish for each infection group were injected intraperitoneally with the bacterial suspensions and mortality was recorded for 7 days post-infection. The mice were injected with 5 × 10⁸ CFU and 2 × 10⁸ CFU, then monitored three times a day; loss of mobility was set as a predictor of death during observation. Zebrafish were injected with 1 × 10⁵ CFU, 1 × 10⁶ CFU, 1 × 10⁷ CFU, and 1 × 10⁸ CFU. The moribund mice, recovered mice, and the mice in the control group were anesthetized at the end of the 7-day observation period with pelltobarbitalum natricum (30 mg/kg), and sacrificed by cervical dislocation. The zebrafish received lethal anesthesia with 3-aminobenzoic acid ethyl ester methanesulfonate MS-222 (90 mg/L).

The overnight S. suis cultures in THB medium, grown to an OD₆₀₀ of 0.6, were sub-cultured again into 50 ml THB at a 1:100 dilution. The optical densities at OD₆₀₀ of bacterial cultures were measured periodically until the late stationary phase (15 h after seeding, 1 h intervals).

An antibiotic susceptibility test was performed by assessing the minimum inhibitory concentration (MIC) of strain HN105, P1/7 and ZY05719. Eighteen antibiotics were used, including some important antibiotics such as erythromycin, vancomycin, tetracycline, cephalosporins, chloramphenicol and kanamycin. The test was performed by the broth microdilution method, according to the standard methods in the Clinical and Laboratory Standards Institute (CLSI) guidelines (2016). Briefly, the overnight bacterial culture was diluted (1:1000). Equal volumes of the bacterial culture were added in the 96-well plates and mixed with serially diluted antibiotics. Following 16 h incubation at 37°C with CO₂, the growth in each well was recorded. The highest dilution of the antibiotic that inhibited bacterial growth, compared to the media control, was recorded as the MIC.

The whole genome sequencing of HN105 was performed on the Illumina Miseq and Pacbio RSII platforms (Personalbio Bio-Technology Co., Ltd.). Using SPAdes (version 3.7.1), and the second generation and third generation sequencing data were assembled. Using the MUMmer software for co-linearity analysis with the reference genome, the positional relationship of contigs on the genome was determined. The gaps between the contigs were repaired using the original data from the third-generation sequencing. Finally, the complete chromosomal sequence was corrected using Pilon software. For plasmids, the contigs were compared to the nucleotide library in the NCBI database, the plasmid sequence was picked out, and the third-generation sequencing data was used to fill the gaps as noted above in order to obtain a complete plasmid sequence. A total of 2,196,724 bp for chromosomal DNA were generated by the combination of the two libraries (PE400 and S10K). The open reading frames (ORFs) were identified using Glimmer 3.0 based on the HN105 genome sequence (Delcher et al., 1999). The tRNAs and rRNAs were predicted by tRNAscan-SE 1.31 and RNAmmer 1.2, respectively (Lowe and Eddy, 1997), whereas ncRNA was BLAST-analyzed to the Rfam database (Lagesen et al., 2007). The CRISPR recognition tool (CRT) was used to predict DRs (forward repeats) and spacers (Bland et al., 2007). The GIs of HN105 were determined with Island Viewer¹. The sequences and annotations of HN105 were deposited with GenBank² under accession no. CP029398 and no. CP029399 for the plasmid.

By comparing the sequences of seven housekeeping genes (aroA, cpn60, dpr, gki, mutS, recA, and thrA) in S. suis, MLST provides a discriminatory genotyping method for different strains (Maiden et al., 2013). The allele numbers and sequence type (ST) of strains used in this study were downloaded from the MLST database³ and analyzed with eBURST⁴.

Strains P1/7 and ZY05719 were representative clinical and virulent SS2 strains, respectively (Table 1). A comparison of HN105 with the above SS2 strains were performed using the program progressiveMauve (Darling et al., 2010), displaying multiple genome alignments.

A comparison of HN105 with P1/7 and ZY05719 was performed using progressiveMauve (Darling et al., 2010) and a major genomic island discovered was determined to be an ICE element.

Annotation of the complete ICE sequence was conducted with RAST (Aziz et al., 2008) and the results were manually checked by comparison with the NCBI database⁵. Comparative analysis was carried out with the BLAST program⁶.

A recent study identified 30 core genes of ICEs in S. suis, primarily those encoding integrase, excisionase, relaxase, plasmid mobilization relaxosome protein, type IV secretion systems, DNA primase, replication initiator protein A, DNA 5′-methyltransferase and hypothetical proteins (Huang J. et al., 2016). Using this data, we compared the HN105 ICE genes with the corresponding genes of ICESa2603 using BLAST analysis and the ACT software in order to build phylogenetic trees. In all, 13 ICEs were used to generate the phylogenetic tree on MEGA6, using the maximum-likelihood method with bootstrapping.

The unique 80K genomic island was knocked out by natural transformation, according to a recent study (Zaccaria et al., 2014). The forward and reverse homologous sequences of the target gene were fused with the chloramphenicol marker by overlap PCR. Then the DNA products were mixed with the peptide, incubated, and selected on THB agar (Cm^R). MICs were then determined for kanamycin, gentamicin, erythromycin, spectinomycin, tetracycline, lincomycin, and neomycin.

Genomic DNA was extracted from HN105 using the Omega genomic DNA purification kit (OMEGA, China). The transformants were generated by natural transformation of HN105 DNA into ZY05719 (SS2) on THB agar, as described previously (Zaccaria et al., 2014). The receptor strain ZY05719 was cultured in THB until OD about 0.04, and HN105 DNA with pheromones peptide were mixed with 100 μl THY bacteria. After 2 h of incubation at 37°C, samples were spread on plates (Kan⁺ and Spc⁺). Of note, HN105 is multidrug-resistant (spectinomycin, kanamycin, tetracycline and erythromycin) and conferred chromosomally, whereas ZY05719 is susceptible to all three. The antibiotic resistance of transformants was tested on THB agar with 100 μg/ml spectinomycin and 50 μg/ml kanamycin. The transformants were cultured and the DNA extracted for genome sequencing.

The integrated form and extrachromosomal circular form of the ICEs in S. suis were detected by combination primers, P1–P4. The amplicons were sequenced by Sanger sequencing and aligned with the attL and attR of ICESsuHN105.

For mating experiments, the mutant strain P1/7 (Cm^R, Tet^S, and Ery^S) was used as a recipient and the multi-drug resistant strain HN105 as the donor. The filter mating assays were performed as previously described (Huang J. et al., 2016;Huang K. et al., 2016). The cultured donor and recipient strains, at an OD₆₀₀ of 0.4–0.6, were mixed at a ratio of 1:5. The cell mixtures were resuspended, spotted on filter membranes, and were grown on THB agar plates overnight. The transconjugants were selected using the antibiotics tetracycline/erythromycin and chloramphenicol. Finally, the transconjugants were further confirmed by PCR for the ICESsuHN105 genes recN, gdh, and cps2.

The lantibiotic cluster deletion mutant strain ΔSss (Cm^R) was constructed as described above. For antimicrobial susceptibility testing, 5 μL of OD₆₀₀ 0.6 Streptococcus culture was placed on the surface of THB agar (1.5%, w/v) plates and incubated at 37°C under 5% CO₂ for 24 h. Melting agar (0.5% agar, w/v) medium was prepared and mixed with the indicator bacteria at about 50°C. The mixing agar medium (5 mL) containing 10⁶ CFU indicator bacteria was overlaid on the plate uniformly and then incubated at 37°C for 24 h. The antimicrobial susceptibility was measured by the appearance and diameter of the inhibition zone. The lantibiotic extraction was performed as previously described (Vaillancourt et al., 2015) and details are provided in the Supplementary Material.

The bacterial strain isolated from a diseased pig with acute arthritis was identified as S. suis according to species-specific PCR (detecting gdh and recN) and 16S rRNA sequencing (Ishida et al., 2014), and named HN105. A remarkable self-agglutination resulted in a negative serologic agglutination test. Serotype-specific PCR identified HN105 as a type 5 strain of S. suis in genotype and showed α-hemolysis on 5% (v/v) sheep blood agar (Supplementary Figure S1). As HN105 was not serologically identified in the phenotype, we marked it as SS5 (genotype) here.

The in vitro growth characteristics of HN105 were compared with two other S. suis strains (P1/7 and ZY05719) in THB. The results indicated that the SS5 (genotype) strain HN105 grew faster at early logarithmic phase (Supplementary Figure S2).

To evaluate the virulence of HN105 in vivo, we first used an established zebrafish model to compare its pathogenicity with two well-known SS2 virulent strains, P1/7 and ZY05719. The LD50 values for strains HN105, P1/7 and ZY05719 were 7.65 × 10⁵ CFU/fish, 1.08 × 10⁶ CFU/fish, and 3.52 × 10⁵ CFU/fish, respectively (Figure 1A). This result implied that the virulence of strain HN105 is close to that of ZY05719 and higher than that of P1/7 in the zebrafish model (Wu et al., 2014a).

Next, we used the BABL/c mouse infection model to reassess the virulence of HN105. As shown in the survival curves (Figure 1B), the HN105 strain was more pathogenic to BABL/c mice than the two SS2 strains, further supporting the idea that HN105 should be classified as part of the virulent group of S. suis (Dong et al., 2015). Furthermore, some of the mice infected with HN105 developed typical disease symptoms, including shivering and dyspnea.

The results of the MIC test were tabulated (Table 2). The data demonstrated that HN105 was resistant to many antibiotics such as tetracycline, kanamycin, gentamicin, erythromycin, lincomycin, neomycin, and spectinomycin. The S. suis strains P1/7 and ZY05719 were susceptible to these antibiotics, except for tetracycline, gentamicin and neomycin. This multidrug resistance, coupled with the high virulence of HN105 in animal infection assays, prompted a comprehensive genomics analysis of the SS5 (genotype) strain, HN105.

The genomic sequencing of strain HN105 revealed a 2,131,882 bp circular chromosome and a 20,752 bp plasmid (Figure 2). The general features of the genome are summarized in Supplementary Table S2. In total, 2077 coding sequences (CDSs) were identified in the chromosome and 27 coding sequences on the plasmid. The average G+C contents were 41.4 and 33.22% for the chromosome and plasmid, respectively. Additionally, the tRNA and rRNA cluster copy numbers were similar to the other SS2 strains, and only one type of CRISPR was identified. MLST analysis identified that HN105 belongs to sequence type 498, which harbors another SS5 strain, Sdly140101.

Comparative genomic analysis of HN105, the reference virulent SS2 strain P1/7 and ZY05719 was performed to better understand the multidrug resistance of the SS5 (genotype) strain. Genome alignment showed large-scale genomic rearrangements in HN105, including inversions, insertions and deletions. Although there was no typical 89K PAI encoded in HN105, a unique 80K genomic island (GI) was seen inserted in the chromosome. This GI was 79,639 bp in size with an average GC content of 38.04%, significantly different from that of the whole genome (41.4%). Genes encoding transposase and integrase were found in its flanking region. Although a lot of genes in 80K GI encode hypothetical proteins, we identified some antibiotic resistance-related genes or regulators, which will be interesting for further analysis (Figure 3A,B). In addition, there were two another GIs (43K and 40K) inserted into the chromosome, composed primarily of accessory secretion systems, glycosyltransferase and prophage elements.

Despite the large-scale genomic differences and the absence of important SS2-associated virulence genes such as mrp or sly in the chromosome of HN105, we identified some virulence-related factors such as fbp, srtA, dippIV, enolase, an orphan regulator, covR and the nadR regulator. In addition, the capsule synthesis gene cps and the capsule regulatory genes ccpA were also identified in HN105. Expanding the search to genes coding for products potentially involved in bacteremia-enhancing S. suis virulence revealed the presence of important virulence-related genes (Supplementary Table S3).

Further analysis of the 80K genomic island confirmed that it was a novel ICE element, possessing a conserved integrase closely related to intICESa2603 (>99% amino acid identity). Moreover, alignment of the core genes with ICESa2603 showed >60% DNA sequence identity. Both the above indexes are specific features of the ICESa2603 family⁷. Accordingly, the HN105 ICE was renamed ICESsuHN105 and classified as a member of the ICESa2603 family.

In ICESsuHN105, integration between the rpsL and hydrolase genes yielded two 15 bp att sites at the left and right ends (attL/R: 5′-TTATTTAAGAGTAAC-3′). A putative oriT sequence was identified upstream of the mobC gene, at the genomic locus 1015147 to 1015183 (GGGATATTGTGGACACAATATCTGAGCTCGCAAAGAC), identical to ICESa2603. In addition to possessing genes involved in conjugation, which were annotated as a tyrosine family integrase, Tn5252 orf4 relaxase, Tn5252 orf9 mobC and Tn5252 orf10, ICESsuHN105 also contained some accessory genes conferring growth advantages (Figure 3C). Furthermore, ORFs predicted to encode the PezAT toxin–antitoxin system were verified to be functional and transcribed as an operon mode (Figure 3C).

As expected, the analysis results showed that ICESsuHN105 contains several antibiotic resistance-associated determinants. Examples include HN015_04855 encoding an aminoglycoside O-phosphotransferase (APH), which could confer resistance to multiple antibiotics including kanamycin, lincomycin, neomycin, and amikacin; two identical genes (HN015_04830 and HN015_05090) encode rRNA adenine N-6-methyltransferases for erythromycin resistance; genes HN015_05015 and HN015_04965 encode ribosomal protection proteins, which could protect the ribosome from the translation inhibition of tetracycline; and spectinomycin adenylyltransferase (HN015_04990) could result in spectinomycin resistance. Notably, a complete lantibiotic locus was encoded in this ICE (Figure 3D).

To analyze the evolution of the ICE core cluster in S. suis, we compared the DNA sequence identity of each of the ICE core genes to the corresponding genes of ICESa2603 (Figure 4). The phylogenetic trees based on the core genes demonstrated that the ICEs were further divided into three subgroups, with the divergent branch (group III) containing only one member, ICESsuHN136. Group I included 89K-like ICEs, most of them containing critical virulence-related factors, and tetracycline and macrolide resistance genes (Tang et al., 2006; Li et al., 2011). Three novel ICEs comprised group II, and included ICESsuHN105, ICESsuAH681 and ICESsuCZ130302. ICESsuCZ130302 and ICESsuAH681 were detected in novel serotype Chz strains. In addition, ICESsuCZ130302 has been demonstrated to be excisable from the donor cell chromosome, cyclized and transferred to the S. suis serotype 2 virulent reference strain P1/7 (Pan et al., 2019). These ICEs carried multiple antibiotic resistance genes instead of VFs.

The resistance profiles of the wild-type and mutant Δ80K strains were compared for 7 antibiotics, to which HN105 and ZY05719 had been identified to be resistant and sensitive, respectively (Table 2). We found that the mutant strain was not resistant to any of these antibiotics (Table 3). Based on the data, we could preliminarily conclude that ICESsuHN105 is effective in conferring antibiotic resistance to HN105.

The transmissibility of multidrug resistance from HN105 to the different S. suis strains was determined. Here we devised an experimental strategy to identify the genetic changes causing antibiotic resistance (Supplementary Figure S4). This strategy was involved to induce genomic hybrids by natural transformation between two S. suis strains with different phenotypic antibiotic sensitivity. Accordingly, purified HN105 genomic DNA was used as the donor DNA, with the SS2 ZY05719 being the recipient strain. The positive clones screened on THB plates containing multiple antibiotics were named ZY05719_(HN105) transformants. The antibiotic resistance of ZY05719_(HN105) was further investigated by MIC assays. The results indicated an increase in the MICs of spectinomycin, kanamycin, neomycin, tetracycline, and erythromycin for the ZY05719_(HN105) transformants compared to the parental strain (Table 4). This result indicated the antibiotic resistant genes could easily be horizontally transferred between clinical isolates from different serotypes under environmental antibiotic stress.

To explore whether horizontal transfer of antibiotic genes occurred by an insertion of the whole ICESsuHN105 box in the ZY05719 genome, one ZY05719_(HN105) transformant was sequenced and the draft genome was analyzed (GanBank ID: RZIC00000000). The genomic data suggested that the different antibiotic resistance elements from ICESsuHN105 were dispersedly inserted as single genes at different genomic locations. Although the acquisition of novel antibiotic resistance genes was confirmed by the contribution of the individual HN105-encoded genes, the experiments did not identify the horizontal transfer of intact ICESsuHN105 by natural transformation in Streptococcus species.

Active ICEs can be excised from the chromosome with the aid of the integrase and excisionase to form circular extrachromosomal ICEs. Four specific primers (P1, P2, P3, and P4) targeting HN105 were designed in order to detect ICESsuHN105. PCR with different combinations of the four primers can be used to identify both the chromosomally integrated (P1–P2 and P3–P4 positive; P2–P3 and P4–P1 negative) and excised-circularized (P1–P2 and P3–P4 negative; P2–P3 and P4–P1 positive) forms of ICESsuHN105 in S. suis HN105 grown to logarithmic phase. The PCR analysis showed that DNA amplicons were obtained with all these four pairs of primers; however, the bands of P2–P3 and P4–P1 were shallower. The results indicated that although the probability of occurrence is relatively low, both the forms of ICESsuHN105 (chromosomally integrated, and excised-circularized) occurred when strain HN105 was cultured. This suggested that ICESsuHN105 had the ability to form an annular structure (Figure 5). However, successful conjugants between donor S. suis HN105 and recipient strains P1/7 (Cm^R) have not been obtained in our in vitro conjugation assays.

According to the comparative genomics analysis, there is a type B lantibiotic locus at the end of the 80K ICESsuHN105; an identical gene order to the nisin locus in S. equi subsp. strain H70 (Holden et al., 2009) and S. suis 65 (Vaillancourt et al., 2015). Based on the annotation, this complete lantibiotic locus (suicin locus) contains 10 coding genes, including a sensor histidine kinase (sssK), a response regulator (sssR), three precursors (sssA1, sssA2, and sssA3), a lantibiotic modification (sssM), an ABC transporter (sssT), and three immunity proteins (sssF, sssE, and sssG) (Figure 3D). Therefore, we redesignated this locus as ‘locus suicin 105.’

The plate diffusion assay was used to investigate the antimicrobial ability of HN105 to other streptococcal strains. The data showed that S. agalactiae A909, S. equi subsp. zooepidemicus ATCC35246, S. suis P1/7, and S. pneumoniae RX1 were sensitive to HN105, but S. agalactiae GD1008-001 and S. suis ZY05719 were not (Figure 6A). Therefore, S. agalactiae A909 was selected as the indicator strain in further studies.

To verify that the observed antimicrobial ability of HN105 was conferred by suicin 105, a mutant strain ΔSss was constructed. Compared to the wild type, ΔSss displayed significantly smaller inhibitory zones in plate diffusion assays using S. agalactiae A909 as the indicator (Figure 6B). This result indicated that suicin 105 contributes to antimicrobial activity of HN105 against other streptococcal strains.

A previous study identified that bacteriocins were not secreted in a liquid medium (Vaillancourt et al., 2015). Accordingly, we purified suicin 105 after growth on THB plates using cationic exchange and reversed-phase MS, and a bacteriocin enrichment fraction (Thiede et al., 2014). Mass spectrometry identified the type of polypeptide that shared 92.3% identity with the predicted mature suicin peptide (Supplementary Table S4). This finding confirmed the secretion of suicin 105 in HN105.