Characterization of pig tonsils as niches for the generation of Streptococcus suis diversity

Streptococcus suis is a gram-positive bacterium that causes meningitis, septicemia, endocarditis, and other disorders in pigs and humans. We obtained 42 and 50 S. suis isolates from lesions of porcine endocarditis and palatine tonsils, respectively, of clinically healthy pigs in Japan; we then determined their sequence types (STs) by multilocus sequence typing (MLST), cps genotypes, serotypes, and presence of classical major virulence-associated marker genes (mrp, epf, and sly). The 42 isolates from endocarditis lesions were assigned to a limited number of STs and clonal complexes (CCs). On the other hand, the 50 isolates from tonsils were diverse in these traits and seemingly in the degree of virulence, suggesting that tonsils can accommodate a variety of S. suis isolates. The goeBURST full algorithm using tonsil isolates obtained in this study and those retrieved from the database showed that major CCs as well as many other clusters were composed of isolates originating from different countries, and some of the STs were very similar to each other despite the difference in country of origin. These findings indicate that S. suis with not only different but also similar mutations in the genome have survived in tonsils independently across different geographical locations. Therefore, unlike the lesions of endocarditis, the tonsils of pigs seemingly accommodate various S. suis lineages. The present study suggests that S. suis acquired its diversity by natural mutations during colonization and persistence in the tonsils of pigs. Supplementary Information The online version contains supplementary material available at 10.1186/s13567-024-01270-5.


Introduction
Streptococcus suis (S. suis), a Gram-positive bacterium, is an important swine pathogen that causes meningitis, septicemia, sudden death, pneumoniae, and arthritis in farmed pigs.In contrast, clinically healthy pigs are often found to carry endocarditis during slaughter [1].S. suis showed affinity to pig oral cavities, as 100% of examined pigs carried S. suis in their saliva [2,3].S. suis may also affect humans and other animals [4].S. suis strains were classified into 35 serotypes (serotypes 1 to 34 and 1/2 which reacts to antisera type 1 and 2) on the basis of the antigenicity of capsular polysaccharide antigens [5,6].However, the serotypes 20, 22, 26, 32, 33 and 34 were moved from S. suis on the basis of molecular phylogenetic analyses, and the remaining 29 serotypes have so far been officially recognized as S. suis [7][8][9].
The capsule of S. suis is encoded by a cluster of capsular polysaccharide synthesis (cps) genes located in a single locus of the genome [10].Typing of the cps gene for all serotypes by two-step multiplex PCR has been developed [11].Conventional serotyping using serum agglutination is time-consuming and it is costly to purchase or prepare all the typing antisera; thus, it is ideal to conduct genotyping by PCR (referred to as cps genotyping) in advance, followed by serum agglutination test using expected antisera.From investigations of field isolates of S. suis, most isolates from diseased pigs belonged to limited serotypes (serotypes 2, 3, 7, and 9); in particular, the isolates of serotype 2 were highly virulent and most prevalent amongst all serotypes [1,4,5,12].
The capsule of S. suis acts as an important virulence factor by escaping from phagocytosis by macrophages and neutrophils of the host defense system [13][14][15].Since isogenic un-encapsulated mutants showed a low degree of virulence in pigs and mice, un-encapsulated S. suis is believed to be avirulent [16][17][18].Furthermore, recent experiments of several serotypes and replacing the total cps gene cluster showed that the capsule itself altered the degree of virulence in S. suis [19].In contrast, both encapsulated and un-encapsulated S. suis have been found to persist in lesions of porcine endocarditis [9].Un-encapsulated isolates showed increased ability to adhere to porcine and human platelets and intercellular matrix proteins, to invade to cultured porcine cells, and to form biofilms [20][21][22][23].In addition, the loss of the capsule caused by spontaneous mutations in one or more of cps genes possibly occurred during persistent infection in pig bodies [8,20,24].
In addition to the capsule, many other potential virulence factors have been described, including muramidase-released protein (MRP, encoded by mrp) [25], extracellular factor (EF, encoded by epf) [26], and suilysin (SLY, encoded by sly) [27].The precise roles of MRP and EF have not been identified; however, these putative virulence factors have been frequently found to be associated with highly virulent serotype 2 isolates [28,29].Although many additional virulence-associated markers have been proposed, the three markers of MRP, EF and SLY still commonly serve as classical virulence markers.
To date, genetic heterogeneity and the phylogenetic diversity of S. suis strains have been described using various molecular tools [30][31][32][33][34][35][36].However, multilocus sequence typing (MLST) is currently one of the most valuable tools for examining the population structure and global distribution of S. suis [37][38][39][40][41][42].Notably, the results of MLST analysis can be visualized by a bioinformatic tool, goeBURST [43].S. suis serotype 2 can be classified by MLST into more than 10 sequence types (STs), among which some closely related STs can be grouped as ST clonal complexes (CCs) by goeBURST.For example, strains of S. suis serotype 2 with high virulence were classified into ST1 or ST7, which belonged to CC1.In contrast, some low or non-virulent strains were allotted to many other STs and could not be assigned to a certain CC [38,40], indicating that there is higher diversity of low or non-virulent strains than highly virulent strains.
Through the use of the typing tools described above, many studies have been performed to elucidate the determinant for the virulence of S. suis, utilizing isolates from both diseased and healthy pigs as well as human cases [28,33,[44][45][46][47].On the contrary, studies using comparison or population analysis of the low or non-virulent isolates of S. suis are rare.In the present study, we focused on low or non-virulent isolates of S. suis from healthy pigs to examine their population structure and relationship in order to clarify S. suis diversity.The total design of this study is depicted in Figure 1.

Isolation and identification of S. suis
Organ samples were collected from 110 pigs shipped to a meat sanitary inspection station in Tokyo from 22 farms in Tohoku and Kanto areas of Japan between 2015 and 2021.All the pigs were considered clinically healthy before slaughter; however, at the time of inspection, 50 pigs of marketing ages (approximately 6 months of age), reproductive ages (2 to 3 years of age), and age unknown were condemned because of endocarditis.The rest of 60 pigs were marketing age.We defined the 60 pigs as healthy pigs in the present study.Samples of heart valve vegetations from 42 pigs and of palatine tonsils from 68 pigs (8 with endocarditis and 60 healthy) were stamped onto either Trypticase soy agar (Becton, Dickinson and Company, Sparks, MD, USA) supplemented with 5% horse blood (TSA-HB) or Columbia agar (OXOID, Kanto Chemical Co., Tokyo, Japan) supplemented with Streptococcus selective supplement (Thermo Fisher Scientific, Waltham, USA) and 5% horse blood (COBA), and then incubated at 37 °C for 48 h in air, and if necessary, passaged to TSA-HB or COBA and incubated for 24 h in air or air plus 5% CO 2 , respectively.Colonies grown on the agar with α-hemolysis were selected and screened by Gram staining, catalase test, and rapid identification kit (Rapid ID32 Strep API, BioMérieux, Marcy-l'Étoile, France).The suspected isolates were confirmed to be S. suis by species-specific PCR for S. suis (recN PCR) [48].The S. suis isolates were frozen in skim milk (Becton Dickinson) at −80 °C until use.

DNA extraction and PCR
The S. suis isolates grown on TSA-HB at 37 °C for 24 h in air plus 5% CO 2 were used for bacterial DNA extraction following the alkaline boiling method [49].Briefly, the bacteria were suspended in 25 mM NaOH, heated at 95 °C for 10 min, and then neutralized by adding an equal volume of 80 mM Tris-HCl (pH 8.0), followed by centrifugation at 13 000 × g for 3 min.The supernatants were used for PCR.AmpliTaq Gold 360 Master Mix (Thermo Fisher Scientific) was used for PCR amplification according to the manufacturer's instructions.

MLST
As described previously, MLST was performed by direct sequencing of seven PCR-amplified house-keeping genes [37].The allele numbers and sequence type (ST) of the isolates were determined by comparing their sequences with those in the PubMLST database [63] accessed in May 2023.Novel alleles and STs were assigned by submitting the respective data to the database administrator.The goeBURST (v1.2) algorithm from Phyloviz 2.0 software [50] was used to visualize CCs by creating an MLST-based minimal spanning tree (MST) with all the isolates obtained in this study and all strains retrieved from the PubMLST database (2273 STs on May 11, 2023) at the triple-locus variant level (TLV).CCs were composed of STs with at least 6 identical alleles, except that ST117, ST1528, and ST1529 were included in CCs as they were double-locus variants (DLVs) or TLVs of major CCs.Among the STs that could not be assigned to any CC, those that differed in 4 or more loci from their closest counterparts were defined as singletons.For visualizing relationships among STs of the tonsil isolates, the 50 isolates obtained in this study and 559 isolates with tonsil origin noted in the database were linked by the goe-BURST full algorithm in Phyloviz 2.0.From the database, MLST profiles of 9 or more deposited from the same country were selected for this analysis and the differences in countries were color-coded.

Capsular polysaccharide synthesis (cps) gene profiles and serotyping
Genotypes of cps genes were determined by the 2-step multiplex PCR assay previously described [11].If the PCR fragments could not be obtained or the sizes of the fragments were different from those described in the original study, such cases were classified as untypeable and were not subject to serum agglutinations.A mismatch amplification mutation assay-PCR [51] was used to distinguish

MLST Visualizing CCs by goeBURST
Visualizing relationships among STs of the tonsil isolates

Cps typing Serotyping
Virulence-associated gene profiling

Antimicrobial susceptibility testing
The S. suis isolates from vegetations

• porcine endocarditis
The S. suis isolates from porcine tonsils

Methods for characterization
Figure 1 The scheme of the present study design type 1 and 14, and type 2 and 1/2.If the cps genotypes could be determined, the serotypes were determined using an expected type of commercially available antisera by either co-agglutination tests or slide agglutinations [20].In cases where the positive agglutination was not observed, such isolates were defined as untypeable.

Antimicrobial susceptibility testing
The minimum inhibitory concentrations (MICs) of the tested antibiotics were determined using a broth microdilution method with a commercially available kit (Eiken Chemical, Tokyo, Japan) for eight antimicrobial agents (penicillin

Characterization of S. suis isolates by MLST and goeBURST
Forty-two S. suis isolates were identified from all the vegetation samples examined, and 5 and 45 S. suis isolates were identified from the palatine tonsils of 8 pigs with endocarditis and that of 60 healthy pigs, respectively.The details of the isolates are listed in Additional file 1.A total of 92 isolates were assigned to 36 STs by MLST analysis (Table 1).Among them, 31 isolates were assigned to 27 novel STs (ST1524-ST1539 and ST1675-ST1685).The goeBURST algorithm (v1.2) at the TLV level showed that the 42 isolates from endocarditis lesions were assigned to either CC1 (ST1 and ST1526) or CC28 (ST28) (Table 1, Figure 2A (panels A-1 and A-3)].On the contrary, among the 50 isolates from palatine tonsils, three and twelve isolates were assigned to CC1 and CC28, respectively, nine isolates were assigned to various previously identified CCs, and among them, six isolates belonged to novel STs (Table 1, Figures 2A and B).The remaining 26 isolates from palatine tonsils were not assigned to any CCs; two isolates were assigned to previously identified STs (ST54 and ST664), and 24 isolates belonged to novel STs (Table 1).
On the other hand, among the 27 novel STs, four (ST1526, ST1528, ST1529, and ST1679) were assigned to previously described CCs (Table 1, Figure 2A (panels A-2 and A-3)).Fifteen novel STs were linked to previously identified or novel STs at least at the TLV level (Table 1, Figure 2B).Remarkably, two small clusters, one comprising ST1536, ST1534, ST1683, and ST1682, and the other comprising ST1678 and ST1680, consisted of only novel STs (Figure 2B).Notably, ST54 and ST1539 were close to STs that were previously assigned to CC231/1109 (Figure 2B) [40].However, eight remaining novel STs were considered to be singletons.The closest STs to these singletons were quadruple-locus variants or quintuple-locus variants (Additional file 2).

Relationship among the 50 tonsil isolates and other previously identified tonsil isolates
MLST-based MST calculated by the goeBURST full algorithm in Phyloviz 2.0 could link all the examined tonsil isolates (Figure 3).Except for two STs, ST1354 and ST2230, which were linked to ST1 at the septuple-locus variant level (Figure 3B), all the STs showed at least one locus that was identical to their counterparts.The STs of tonsil isolates were clustered with not only Japanese isolates but also those from many other countries.In the CC1, CC17, CC28, and CC94 clusters, the representative STs were surrounded by other STs with isolates originating from other countries (Figure 3B).In addition to the clusters involved in and close to CCs, isolates from different countries formed many clusters.Although most STs differed in more than 3 loci from their counterparts, some were single-locus variants (SLVs) or DLVs of other STs despite the difference in country of origin (Additional file 3).

Characterization of S. suis isolates by cps typing and serotyping
The cps types and serotypes are summarized in Table 2. Cps typing of the 42 isolates from vegetations of endocarditis identified only cps2 except for an untypeable isolate.Among them, 20 isolates were serotype 2, and the expected antiserum did not agglutinate the remaining 21 isolates and were untypeable serotypes.Among the 50 isolates from tonsils, nine were cps untypeable.The remaining 41 isolates were assigned to 14 different cps types.Three of them, belonging to CC1, were cps2.Among them, two were serotype 2, and one was untypeable serotype.The eight and four isolates belonging to CC28 were cps2 and cps3, respectively; the former was serotype 2 and the later was serotype 3.Among the remaining 35 isolates, cps types could be determined for 26 isolates and serotypes for 23 isolates (Table 2).Most cps types harbored one or two STs, whereas cps16 and cps31 harbored three and five STs, respectively, in which one and three were singletons.Cps2 were commonly found in the isolates from both vegetations of endocarditis and tonsils, and STs in the isolates from  vegetations and tonsils were similar.On the other hand, other cps types were found only in the isolates from tonsils (Table 2).

Discussion
The MLST analyses of S. suis isolated from diseased pigs and clinically healthy pigs have been described in many previous studies and there have been many efforts to discriminate the highly virulent S. suis isolates from low or non-virulent isolates [40,46,47,[54][55][56][57][58].Although highly virulent strains that may affect mortality of nursery piglets have been spotlighted, low or non-virulent strains that occupy the majority of S. suis may play a part of the whole ecosystem.This implies that the low or nonvirulent stains may affect the healthy status of not only nursery piglets but also pigs of other ages.In the present study, we isolated and examined S. suis from clinically healthy pigs to elucidate the diversity of and relationships among presumably low virulent S. suis isolates using MLST analysis.In a previous study, we showed that prevalence of S. suis in Japan was almost similar from north to south parts of Japan [2], this suggests that the results obtained in this study can be extrapolated to other areas because the samples were collected in the meat inspection center in Tokyo, the biggest consumption area in Japan.As has been previously described [8,9,20], the isolates from endocarditis lesions belonged to a limited number of CCs, CC1 and CC28, which are reported to involve highly virulent isolates [29, 37-39, 46, 47, 56].On the other hand, fifteen isolates from tonsils belonged to CC1 or CC28, consistent with previous studies reporting that potentially hazardous S. suis persists in asymptomatic pigs [59,60].To date, in addition to CC1 and CC28, CC16, CC17, CC20, CC25, CC94, CC104, CC233/379, and CC221/234 have been reported to involve highly virulent or potentially hazardous S. suis [56].ST28 was predominant in the isolates from endocarditis and found most in those from tonsils.However, as far as the typing of MLST, serotypes, and virulence-associated genes, we could not discriminate between the ST28 isolates from endocarditis and those from tonsils.On the contrary, in the present study, the isolates from tonsils showed extensive ST diversity; in particular, 30 of the 50 isolates from tonsils were assigned to 27 novel STs.Four isolates belonging to novel STs (ST1526, ST1528, ST1529, and ST1679) were assigned to previously described CCs that may involve potentially hazardous S. suis.In particular, ST1526 is SLV of ST1 and was found in isolates from both tonsils and endocarditis vegetations, suggesting that the isolates belonging to ST1526 were diverged from ST1 and were virulent.However, the remaining 26 isolates belonging to novel STs and two isolates belonging to previously described STs (ST54 and ST664) could not be assigned to any CCs or were considered singletons (Table 1).Although some of these isolates showed relationships to previously described STs or other novel STs at least at the TLV level (Figure 2B), these STs did not seem to involve highly virulent S. suis.Furthermore, eight singletons showed no close relationships to other STs (Additional file 2), indicating that the isolates belonging to these novel STs were low or non-virulent S. suis.These observations suggest that tonsils can accommodate potentially hazardous S. suis and permit a variety of low or non-virulent S. suis strains to persistently colonize in this niche by protecting them from the host immune system.
To characterize the potential of tonsils as niches accommodating low or non-virulent S. suis, we made a schema of an MLST-based MST calculated by the goe-BURST full algorithm using S. suis isolates from tonsils of healthy pigs obtained in this study and those retrieved from the database (Figure 3).Since the algorithm was able to connect all the STs even though there are differences in 7 loci from other STs, with only two exceptions, all the examined STs were connected to form a large tree with at least the sextuple-locus variant level (Figure 3).Although the number of the isolates from each country was different, many of the clusters that appeared in the MST were composed of STs whose isolates originated from different countries.Furthermore, many STs did not belong to major CCs, that is, most STs seemed to contain low or non-virulent S. suis strains.Notably, although most of the STs were linked at lower than TLV levels, some STs whose isolates originated from different countries were connected at SLV or DLV levels (Additional file 3), indicating that similar mutations affecting the ST have occurred across different countries.Imports of live pigs into Asian countries from either the EU or North American countries for the purpose of breeding take place frequently.It is difficult to determine whether the international trading of pigs shaped the MST clusters  of isolates from different countries or whether a selective pressure under the same environment in the tonsils resulted in similar progenies surviving in such environment and forming the clusters in the MST.In the case of humans, frequent and long-distance moving by globalization can rapidly and globally spread the same clone of pathogens, as exemplified by a lineage 4 of Mycobacterium tuberculosis [61].In contrast to humans, pigs in farms are not frequently transported between different countries and regions.Therefore, such a high degree of diversity found in tonsil isolates and survivals of the same or related STs may be caused by natural mutations followed by selective pressures that occurred during multiplications of the bacteria in each country, presumably in pig bodies, rather than carryover of S. suis from exporting countries.Although the ST28 strains, which were predominant in this study, could be imported by the international trading from other ST28 endemic countries (e.g., North America [55]), these strains have already been established and circulating in Japan [59].
All the isolates except 1 (untypeable) from endocarditis lesions were cps type 2, as reported previously [9,20,59].Among them, 21 isolates did not show agglutination against serotype 2 antiserum, suggesting that these isolates were un-encapsulated, as observed previously [8,9,20] (Table 2).On the other hand, only 11 isolates from tonsils were cps type 2, of which 10 isolates were serotype 2. Although we do not have any experimental evidence, the one isolate that did not show positive agglutination seemed to be un-encapsulated.Interestingly, one isolate each in cps types 1, 11, and 31 did not show positive agglutination against expected types of antisera, suggesting that these isolates also seemed to be un-encapsulated.The un-encapsulated S. suis is believed to be avirulent because of weakness against phagocytosis by the host defense systems [1,13,14,16,17].However, un-encapsulated S. suis exhibits increased levels of adhesion to host cells and biofilm formation that may confer upon the bacteria an ability to persistently colonize and resist clearance by host immune systems [14,20,22].Assuming that the isolates that did not show agglutination were un-encapsulated, they may have persisted in tonsils through the same mechanism of persistence in endocarditis lesions.Although we cannot determine the degree of virulence on the basis of the serotypes, the serotypes in conjunction with STs suggest that, in addition to unencapsulated isolates, most of the isolates from tonsils seemed to be low or non-virulent.
The types of virulence-associated genes varied among previous reports.However, the major three virulenceassociated genes examined in the present study were commonly appeared in those studies [38,40,59] and easily compare the results among such reports.The isolates from endocarditis lesions showed typical patterns of virulence-associated marker genotypes.On the other hand, only a few isolates from the tonsils showed the typical patterns above; in particular, there was no isolates that showed mrp+epf+sly+, which was frequently found in highly virulent ST1 strains [29,38].The fact that various patterns of virulence-associated genes were found and the most frequently found pattern was mrp− epf− sly− as well as the findings on MLST and serotypes strongly suggest that most of the isolates from tonsils were low or non-virulent.
The proportions of resistance to macrolides (AZM and CAM), lincomycins (CLDM), and tetracyclines (TC) were high (approximately 70%) regardless of the origins, with the majority of all isolates presenting multidrug resistance to these four agents.Compared to previous investigations in the Tokai region of Japan, where antimicrobial resistances to CAM, CLDM, and TC were 56.1%, 65.8%, and 80.7%, respectively [62], our results showed similar but higher resistance rates for macrolides and lincomycins.
On the basis of MLST, cps genotypes, serotypes, and virulence-associated marker genotypes, the isolates from endocarditis lesions showed similar patterns, indicating that these isolates were highly clonal.On the other hand, the isolates from tonsils showed an extensively high degree of diversity in all the features examined, and most of them seemed to be low or non-virulent; moreover, a variety of low or non-virulent isolates could, in fact, persist in tonsils of pigs across many countries.
Characteristics of endocarditis and tonsils isolates summarized in Table 4 clearly showed their difference, indicating the degree of virulence was apparently different between them.
Contrary to the isolates from endocarditis lesions, which were highly clonal, the isolates from the tonsils of healthy pigs were extensively diverse in MLST analyses, cps gene types, serotyping, and virulence-associated marker genotypes.Although most of the isolates from the tonsils seemed to be low or non-virulent, such isolates may colonize and persist in the tonsils.A variety of STs found in the isolates from tonsils suggests that many mutations in the genome have occurred.Some of the STs with isolates obtained in different countries showed a close relationship.Such mutations occurred randomly and incidentally; however, in vivo conditions in tonsils may affect the survival of such mutants; as a result, STs that showed a close relationship with each other could be found in isolates that originated in different countries.Although it is generally difficult to determine whether an isolate is virulent or low or non-virulent, our applied method comparing MLST/serotypes/virulent-associated genes with previous reports can be useful to predict virulent, low or non-virulent strains.Taken all together, the present study suggests that S. suis acquired its diversity through natural mutations during colonization and persistence in pig tonsils.

Figure 2 Figure 3
Figure2goeBURST analysis using all S. suis isolates described in this study and all strains retrieved from the PubMLST database (2273 STs) at the triple-locus variant level visualizing clonal complexes.A-1-A-3 Major clonal complexes related to each other.ST117 is related to ST28 via ST27 (i.e., a double-locus variant of ST28) and is included in CC28.ST1528 is linked to ST87 at the double-locus variant level, and is included in CC87.ST1529 is related to ST94 via ST108 (i.e., a triple-locus variant of ST94) and is included in CC94.B Independent CCs and Singletons

Table 1 MLST analysis, cps type and profiles of virulence-associated genes of the 92 S. suis isolates
Variants of mrp are shown as *** (2400 bp), * (1556 bp), and S (747 bp).A large-size variant of epf is shown as *.Single-to triple-locus variants of other STs; however, clonal complex could not be assigned.More than 4 loci were different from other STs.
a Novel STs.b c

Table 2 Cps types and serotypes of 92 S. suis isolates, and their clonal complexes (CCs) and sequence types (STs)
a Not determined.bNo clonal complex could be assigned.c Singleton.

Table 3 Number of isolates and resistance rates (in brackets) of each antimicrobial agents by origin and sequence types (STs)
, and breakpoints were taken from the Clinical and Laboratory Standard Institute (CLSI) criteria in M100-ED33 for the Streptococcus spp.viridans group (CLSI, 2023).There was no isolate resistant for ABPC, CFPM, CTRX, and LVFX.