Antibiotic Resistance Patterns and Molecular Characterization of Streptococcus suis Isolates from Swine and Humans in China

ABSTRACT Streptococcus suis is a zoonotic pathogen that causes disease in humans after exposure to infected pigs or pig-derived food products. In this study, we examined the serotype distribution, antimicrobial resistance phenotypes and genotypes, integrative and conjugative elements (ICEs), and associated genomic environments of S. suis isolates from humans and pigs in China from 2008 to 2019. We identified isolates of 13 serotypes, predominated by serotype 2 (40/96; 41.7%), serotype 3 (10/96; 10.4%), and serotype 1 (6/96; 6.3%). Whole-genome sequencing analysis revealed that these isolates possessed 36 different sequence types (STs), and ST242 and ST117 were the most prevalent. Phylogenetic analysis revealed possible animal and human clonal transmission, while antimicrobial susceptibility testing indicated high-level resistance to macrolides, tetracyclines, and aminoglycosides. These isolates carried 24 antibiotic resistance genes (ARGs) that conferred resistance to 7 antibiotic classes. The antibiotic resistance genotypes were directly correlated with the observed phenotypes. We also identified ICEs in 10 isolates, which were present in 4 different genetic environments and possessed differing ARG combinations. We also predicted and confirmed by PCR analysis the existence of a translocatable unit (TU) in which the oxazolidinone resistance gene optrA was flanked by IS1216E elements. One-half (5/10) of the ICE-carrying strains could be mobilized by conjugation. A comparison of the parental recipient with an ICE-carrying transconjugant in a mouse in vivo thigh infection model indicated that the ICE strain could not be eliminated with tetracycline treatment. S. suis therefore poses a significant challenge to global public health and requires continuous monitoring, especially for the presence of ICEs and associated ARGs that can be transferred via conjugation. IMPORTANCE S. suis is a serious zoonotic pathogen. In this study, we investigated the epidemiological and molecular characteristics of 96 S. suis isolates from 10 different provinces of China from 2008 to 2019. A subset of these isolates (10) carried ICEs that were able to be horizontally transferred among isolates of different S. suis serotypes. A mouse thigh infection model revealed that ICE-facilitated ARG transfer promoted resistance development. S. suis requires continuous monitoring, especially for the presence of ICEs and associated ARGs that can be transferred via conjugation.

levels of mortality, while survival can be accompanied by lifelong sequelae such as deafness (4). The numbers of human cases of S. suis infection were low (1,600) and were concentrated primarily in Asia and Europe (4,5). Two notable large-scale human infections occurred in China in 1998 and 2005, resulting in 14 and 35 deaths, respectively (6)(7)(8). Infection sources for humans have been pork and pork-derived food products, and these are considered high-risk factors (9). Since S. suis is a major swine pathogen, the monitoring of S. suis isolates on farms is necessary for animal as well as human health protection (10).
There are currently 29 serotypes of S. suis, and serotype 2 is the most common for both human and pig clinical isolates (4,11). Clinical treatment relies on antibiotics since there is no available commercial vaccine (12,13). The primary treatment modalities are penicillins alone or in combination with aminoglycosides, macrolides and lincosamides, fluoroquinolones, and tetracyclines (14). However, unreasonable antibiotic therapy has resulted in the increasing antibiotic resistance of S. suis (14)(15)(16). Thus, monitoring resistance in clinical isolates is strongly recommended.
Globally, the frequency of the acquisition of antibiotic resistance in S. suis has been lower than for common Gram-negative Enterobacteriaceae members, possibly due to the more limited horizontal transmission of antibiotic resistance genes (ARGs) in Grampositive bacteria that rely on integrative and conjugative elements (ICEs) located in the chromosome (14). These elements are mobilized via circularization and can be horizontally self-transferred via conjugation (17). The primary functions of ICEs are in host bacterial evolution, and ICEs can transfer favorable fitness genes that enhance virulence, promote resistance to antibiotics and heavy metals, regulate biofilm formation, and encode toxin-antitoxin systems (18). S. suis isolates are also important antimicrobial resistance reservoirs that facilitate the spread of ARGs to major streptococcal pathogens, including Streptococcus pyogenes, Streptococcus pneumoniae, and Streptococcus agalactiae, and between isolates of different S. suis serotypes (19)(20)(21). Therefore, ICEs are primary mediators of ARG transfer, and a more complete understanding of their function is necessary to understand resistance development in S. suis.

RESULTS
Phylogenetic analysis of S. suis and typing of isolates. We examined all 96 of the S. suis isolates by whole-genome sequencing (WGS), and 79/96 (77.5%) included 13 known serotypes, including serotype 2 (40/96; 41.7%), serotype 3 (10/96; 10.4%), and serotype 1 (6/ 96; 6.3%). The remaining 10 serotypes involved fewer than 5% of the isolates. These 96 isolates contained a total of 36 sequence types (STs), and ST242 and ST117 predominated; interestingly, most ST242 isolates were of serotype 2, and all ST117 isolates were of serotype 3, while ST1714 to -1719 and ST1892 to -1908 were newly assigned STs (Fig. 1). A phylogenetic tree indicated the presence of 4 clades, and the major lineage I included 55 (57.3%) of the isolates. Further analysis revealed that lineage I included the 3 human isolates and possessed the broadest distributions by year and location, suggesting clonal spread. In particular, human isolate SS18-7 and porcine isolate 110 from Guangdong shared only 9 single nucleotide polymorphisms (SNPs), indicating a high probability of clonal transmission between pigs and humans ( Fig. 1; see also Fig. S1 in the supplemental material). The phylogenetic tree demonstrated the molecular evolutionary relationship between the 40 S. suis serotype 2 (SS2) isolates in this study and the other 232 SS2 isolates in the NCBI database. The population structure analysis showed four distinct sequence lineages (named lineages I to IV), ranging in size from 22 to 152 isolates (Fig. 2). The majority of the 40 isolates in this study were assigned to lineage II. With the exception of 113 undetermined isolates, SS2 was the most abundant in eastern China but not in the northeastern and northwest regions. From 2013 to 2017, SS2 was the most abundant and was distributed across 4 lineages.
Analysis of resistance profiles. The group of 96 S. suis isolates possessed a total of 24 ARG types that conferred resistance to 7 classes of antibiotics, including macrolidelincosamide-streptogramin B (MLS B ), tetracycline, aminoglycosides, lincosamides, pleu-  The maximum likelihood tree outlines the phylogenetic structures of 40 S. suis serotype 2 isolates unique to this study (highlighted by red points) combined with 232 Chinese S. suis serotype 2 isolates. The year of isolation and the location from which the isolates were obtained are indicated in the inner ring and the outer ring, respectively. romutilin-lincosamide-streptogramin A, chloramphenicol, and oxazolidinone. These resistance genotypes were consistent with the resistance phenotypes, and the MLS genes erm(B) and tet(O) were the most prevalent, at .80% (Fig. 1). These results suggest that this resistance profile might have been acquired by horizontal ARG transfer and could potentially mediate the development of multidrug resistance (MDR).
ICE identification. ICEs and multidrug resistance have been previously reported for S. suis (20,21,24). In our group of 96 S. suis isolates, 10 were confirmed to be ICE carriers using WGS. These elements ranged in size from 64 to 124 kb, with GC contents of 35 to 38%. Each ICE was flanked by the 15-bp repeat 59-TTATTTAAGAGTAAC-39, indicative of attL and attR sites. The sites of integration for these 10 strains were located between the intergenic region extending from hyd (hydrolase, predicted) and rplL (50S large ribosomal protein subunit) ( Table 1). A total of 4 genetic environments (types I to IV) were identified ( and optrA genes, which were flanked by two copies of IS1216E in the same orientation, as well as some other genes (Fig. 5). These two copies of IS1216E could recombine and loop out a translocatable unit (TU) that consisted of one IS1216E copy along with the sequence located between the two IS1216E elements. A 4,845-bp TU was formed in ICESsuS71-1 and contained one IS1216E copy and the ARGs Derm(A) and optrA, along with several other genes (Fig. 5).
Detection of the transferability of ICEs and difficult elimination of the recipient in vitro and vivo. Conjugation experiments were employed to assess the transferability of the ICEs in 10 S. suis isolates, and 5 were successfully transferred to the recipient S. suis P1/7 at frequencies ranging from 1.31 Â 10 28 to 5.14 Â 10 28 . MIC testing indicated that the transconjugants possessed resistance phenotypes consistent with that of the donor (Table 2). Time-kill curves indicated that bacterial reductions of ;3 to 4 log CFU were observed for treatment with tetracycline or erythromycin against P1/7. However, the same treatment failed to inhibit P1/7-ICESsuS82 ( Fig. 6A and B). A transconjugant carrying tet(O) and erm(B) on ICESsuS82 and unmodified recipient strain P1/7 were examined for pathogenicity using a mouse thigh infection model. Tetracycline or erythromycin treatment of mice infected with control strain P1/7 resulted in a significant 10-fold or 100-fold decrease in the CFU per gram of tissue, respectively (P , 0.05). Conversely, tetracycline did not alter (P . 0.05) the levels of the transconjugant P1/7-ICESsuS82, and the CFU were higher than those of the parental strain P1/7 (Fig. 6C). These results indicated that the ICE-carrying transconjugant possessed drug resistance that was difficult to eliminate in vivo.

DISCUSSION
S. suis isolates are increasingly being recognized as the causes of preventable and emerging zoonotic infections in humans, with global distributions that are tracked using serotyping (25,26). The S. suis serotypes identified in this study revealed substantial diversity, and for 83/96 isolates tested, we found 13 different capsular serotypes and 13.5% that were nontypeable. The nontypeable isolates may represent new serotypes of S. suis isolates, or they may be mutant variants of known serotypes with deletions and insertions in genes of the capsular polysaccharide locus (27). The Chinese isolates of highly pathogenic S. suis serotype 2 caused two large-scale human epidemics and massive pig deaths in China (28)(29)(30). Serotype 2 is also the most virulent serotype of S. suis isolates (31) and constitutes the bulk of published genomic sequences for S. suis (32). We also found that serotype 2 predominated in our isolates (40/96). We also identified ST1714 to -1719 and ST1892 to -1908 as newly assigned STs. However, the serotypes and STs did not exactly correspond, and a Molecular Characterization and Resistance of S. suis Microbiology Spectrum particular serotype could correspond to multiple STs and vice versa. However, all ST117 isolates were of serotype 3, and the majority of the ST242 isolates were of serotype 2. All isolates were classified into a 4-clade lineage, with lineage I covering the majority of the members across 7 years and 7 regions, suggesting clonal spread. In pigs, S. suis is transmitted primarily via aerosol, and airborne transmission between pigs has been clearly demonstrated (33). Human S. suis infections were almost always related to pig handling, pig slaughtering, or pork consumption (34) and might account for the presence of the 3 human isolates and the majority of the swine isolates in lineage I. Interestingly, a human isolate from Guangdong and a porcine isolate from Guangdong shared only 9 SNPs, and this was indicative of clonal transmission between animals and humans. Notably, Guangdong human isolate SS18-6, isolated in 2019, and 05ZYH33 (35), which caused the second large-scale outbreak of S. suis infection in China, were both part of lineage I. The harm of S. suis to humans should not be underestimated, and the spread of S. suis should not be ignored. Antibiotics have long been used in the swine industry for disease treatment and prevention (14). However, antibiotics should be used with caution to reduce the selection of resistant S. suis isolates. MIC testing of our isolates indicated that the level of resistance of S. suis to commonly used antibiotics was relatively high. All isolates were resistant to at least one class of antibiotics, and high frequencies of resistance to amikacin, tetracycline, erythromycin, tilmicosin, azithromycin, clindamycin, and lincomycin were observed, consistent with the results of previous studies (15,22). The resistance phenotypes of most of the isolates perfectly correlated with the presence of a corresponding acquired ARG. We found high rates of carriage of MLS B , tetracycline, and aminoglycoside resistance genes in our isolates. In veterinary medicine, macrolides and lincosamides can be used as feed and injection preparations for treating swine respiratory tract infections (14,36). In addition, tetracycline, florfenicol, and penicillins are frequently applied separately or in combination for treating S. suis infections (37-39). Notably, we found linezolid and vancomycin resistance (LVR), and these are last-resort antimicrobial agents for the treatment of multidrug-resistant Gram-positive bacterial infections, which poses a severe threat to public health. LVR S. suis isolates have been previously described (40,41). Therefore, close surveillance is urgently needed to monitor the antibiotic resistance of S. suis.
WGS analysis of our isolates further revealed that the macrolide resistance gene erm and the tetracycline resistance gene tet were carried at high frequencies, and this pattern was previously reported for S. suis (42,43). Notably, resistance to the last-resort agent linezolid was carried by 11 of our isolates, indicated by the presence of optrA, which may have occurred under florfenicol selection on farms (40). Previous studies have established correlations between the presence of an ICE and multidrug resistance in S. suis (20,24). We found 10 isolates that fit this pattern, and the type I and II resistance genes and the genetic environment were similar to those reported previously (20,21,44,45); that is, the integration site was located in the intergenic region between hyd and rplL, similar to most reported ICE integration sites in S. suis (24,46,47). ICEs were inserted at the rplL locus, which is one of the common insertion hot spots for mobile genetic elements (MGEs) in S. suis, forming perfect 15-bp target site duplications at both termini . In addition to the rplL locus, rpsI (S9 ribosomal protein), lysS (lysyl-tRNA), rpmG (L33 ribosomal protein), and guaA (GMP synthetase) have also been reported in Streptococcus spp. (46). We also identified additional resistance gene clusters for type III (ICESsuS102-4) and type IV (ICESsuS71-1). The lsa(E) gene has been found primarily as a part of a large MDR gene cluster often associated with lnu(B) (48). IS26-containing transposes have been shown to move via an excised circular element called a translocatable unit (TU) (49). The IS26 mechanism also explains the properties of IS1216, which belongs to the same insertion sequence (IS) family and mobilizes resistance genes in Gram-positive bacteria (50,51). Importantly, we found optrA, encoding lincosamide resistance, associated with the TU. This location will potentially aid in horizontal transfer among S. suis isolates and to other Gram-positive bacteria. We were able to successfully transfer the ICEs of isolates of 5 different serotypes to S. suis serotype 2 strain P1/7 by conjugation. Multiple attempts were unsuccessful for the remaining 5 isolates, and this failure was most likely associated with a lack of transfer elements in the donor strains, as previously reported (20,45). Our in vitro and in vivo experiments further confirmed that ICEs carrying ARGs were difficult to eliminate with antibiotic treatment. As movable genetic elements, the ICEs in S. suis can be horizontally transferred to isolates of the same serotype and different serotypes of S. suis by conjugation. This can greatly accelerate ARG spread between different species and genera and could bring great difficulties and challenges for clinical treatment protocols. Conclusions. We isolated 96 S. suis strains from humans and swine in China from 2008 to 2019, and serotype 2 predominated. WGS analysis verified that ST242 and ST117 were the most prevalent STs. Phylogenetic analysis of the core genomes of these isolates showed that S. suis most likely clonally spread between pigs and humans in China. WGS further revealed that the erm(B) and tet(O) genes were detected in more than 80% of the isolates. ICEs were detected in 10 S. suis isolates, which were divided into 4 types based on the genomic environment. ICEs can horizontally transfer ARGs between S. suis isolates of different serotypes. Our in vivo experiments further confirmed that an ICE strain possessing ARGs was difficult to eliminate clinically. Therefore, monitoring for ICEs as well as ARGs is necessary to ensure continued treatment efficacy for S. suis isolates.

MATERIALS AND METHODS
S. suis isolates and growth conditions. We collected 96 S. suis isolates for the current study, including 3 human isolates, from the Guangdong Provincial Center for Disease Control and Prevention, and 93 isolates from swine farms in 10 provinces of China from 2008 to 2019 (see Table S1 in the supplemental material). Suspected S. suis isolates were grown, subcultured, and quantified using tryptic soy broth (TSB) and agar containing 5% newborn calf serum. All isolates were identified to the species level using matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry and 16S rRNA gene sequencing.
Whole-genome sequencing and analysis. DNA was extracted from S. suis isolates using a commercial genomic DNA purification kit according to the manufacturer's instructions (Tiangen, Beijing, China). Whole-   (54). The phylogenetic tree was built with RAxML using the GTRGAMMA substitution model and 100 bootstraps and visualized using the iTOL online platform (55,56). ARG homologs were identified using the ResFinder server (http://www.genomicepidemiology.org/) (57). Sequence types (STs) were identified using the PubMLST database (https://pubmlst.org/) (58). ICEs were predicted using ICEfinder (https:// bioinfo-mml.sjtu.edu.cn/ICEberg2/index.php) (17), and complete ICE sequences on different contigs were determined using Sanger sequencing following PCR amplification (Table S2). Genomes were annotated using the Rapid Annotation of Microbial Genomes Using Subsystems Technology (RAST) annotation server (https:// rast.nmpdr.org/rast.cgi). ICEs were identified by comparison with other MGEs from GenBank and were visualized using Easyfig 2.2.2 (59). Conjugation transfer experiment. ARG transferability was assessed by conjugation using the rifampicin-resistant tetracycline-sensitive S. suis serotype 2 strain P1/7 as the recipient. Transconjugants were selected on tryptic soy agar (TSA) plates containing 10 mg/mL each of erythromycin and tetracycline and 30 mg/mL of rifampicin. The transconjugants were further examined for MIC phenotypes and by specific PCR amplification of ARGs and the unique serotype 2 gene sequences (Table S2) (60).
Murine thigh infection model. Six-week-old specific-pathogen-free ICR mice (23 to 29 g) were used for the in vivo studies (Guangdong Medical Lab Animal Center, Guangzhou, China). Mice were rendered neutropenic by the injection of cyclophosphamide intraperitoneally 4 days (150 mg/kg of body weight) and 1 day (100 mg/kg) prior to infection (61). Thigh infections were produced by the intramuscular injection of 0.10 mL of the inoculum (10 8 CFU/mL of exponential-phase cells) into the thighs of isoflurane-anesthetized neutropenic mice. Two hours after thigh infection, the mice were randomized to receive (i) control treatment with the vehicle, (ii) tetracycline at 10 mg/kg by oral gavage twice a day, or (iii) erythromycin at 20 mg/kg by oral gavage twice a day. The dose of tetracycline was chosen according to a previous study with a standard oral dose in nursery pigs (62). The erythromycin dose was chosen to obtain pharmacokinetic values similar to those achieved with the clinical dose of 500 mg in humans (63)(64)(65). After 24 h of therapy, the thighs (six per group) were aseptically removed, homogenized, and processed for bacterial CFU determination.
Ethics statement. Animals were maintained according to national standards for laboratory animals in China (GB 14925-2010) (66). The Animal Research Committee (IACUC) of the South China Agricultural University (SCAU) approved these studies (approval number 2022C014).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.4 MB.