Phylogenetic diversity, antimicrobial susceptibility and virulence gene profiles of Brachyspira hyodysenteriae isolates from pigs in Germany

Swine dysentery (SD) is an economically important diarrheal disease in pigs caused by different strongly hemolytic Brachyspira (B.) species, such as B. hyodysenteriae, B. suanatina and B. hampsonii. Possible associations of epidemiologic data, such as multilocus sequence types (STs) to virulence gene profiles and antimicrobial susceptibility are rather scarce, particularly for B. hyodysenteriae isolates from Germany. In this study, B. hyodysenteriae (n = 116) isolated from diarrheic pigs between 1990 and 2016 in Germany were investigated for their STs, susceptibility to the major drugs used for treatment of SD (tiamulin and valnemulin) and genes that were previously linked with virulence and encode for hemolysins (tlyA, tlyB, tlyC, hlyA, BHWA1_RS02885, BHWA1_RS09085, BHWA1_RS04705, and BHWA1_RS02195), outer membrane proteins (OMPs) (bhlp16, bhlp17.6, bhlp29.7, bhmp39f, and bhmp39h) as well as iron acquisition factors (ftnA and bitC). Multilocus sequence typing (MLST) revealed that 79.4% of the isolates belonged to only three STs, namely ST52 (41.4%), ST8 (12.1%), and ST112 (25.9%) which have been observed in other European countries before. Another 24 isolates belonged to twelve new STs (ST113-118, ST120-123, ST131, and ST193). The temporal distribution of STs revealed the presence of new STs as well as the regular presence of ST52 over three decades (1990s–2000s). The proportion of strains that showed resistance to both tiamulin und valnemulin (39.1%) varied considerably among the most frequent STs ranging from 0% (0/14 isolates resistant) in ST8 isolates to 46.7% (14/30), 52.1% (25/48), and 85.7% (6/7) in isolates belonging to ST112, ST52, and ST114, respectively. All hemolysin genes as well as the iron-related gene ftnA and the OMP gene bhlp29.7 were regularly present in the isolates, while the OMP genes bhlp17.6 and bhmp39h could not be detected. Sequence analysis of hemolysin genes of selected isolates revealed co-evolution of tlyB, BHWA1_RS02885, BHWA1_RS09085, and BHWA1_RS02195 with the core genome and suggested independent evolution of tlyA, tlyC, and hlyA. Our data indicate that in Germany, swine dysentery might be caused by a limited number of B. hyodysenteriae clonal groups. Major STs (ST8, ST52, and ST112) are shared with other countries in Europe suggesting a possible role of the European intra-Community trade of pigs in the dissemination of certain clones. The identification of several novel STs, some of which are single or double locus variants of ST52, may on the other hand hint towards an ongoing diversification of the pathogen in the studied area. The linkage of pleuromutilin susceptibility and sequence type of an isolate might reflect a clonal expansion of the underlying resistance mechanism, namely mutations in the ribosomal RNA genes. A linkage between single virulence-associated genes (VAGs) or even VAG patterns and the phylogenetic background of the isolates could not be established, since almost all VAGs were regularly present in the isolates.


Introduction
Swine dysentery (SD) belongs to one of the most challenging diarrheal infections in pigs worldwide with mortality up to 30% [1]. The disease is caused by the strongly hemolytic Brachyspira (B.) species B. hyodysenteriae, B. suanatina and B. hampsonii with B. hyodysenteriae being the most relevant species [2]. However, B. hyodysenteriae can also be isolated from apparently healthy pigs [3]. Hitherto, the novel strongly hemolytic B. suanatina has been only isolated from pigs in Scandinavia where this pathogen caused damages consistent with mild SD [2,4]. Until now, B. suanatina has no clinical relevance in other pig-producing countries. In contrast, B. hampsonii, which was initially described in the United States of America and Canada [5], has already been detected in pigs from Europe, either during routine diagnostic screening or as part of monitoring programs, but without any linkage to SD so far [6,7]. To differentiate these newly emerged strongly hemolytic species from B. hyodysenteriae, a PCR method was established since biochemical tests failed for specific species identification [8,9].
Some of the major issues of research in the field of swine dysentery currently touch aspects of the population structure and diversity of B. hyodysenteriae, the presence and distribution of virulence-associated genes (VAGs), and the antimicrobial susceptibility of the pathogen to pleuromutilins [2]. Recent studies employing multilocus sequence typing (MLST) have identified different major clonal groups among B. hyodysenteriae isolates [10,11,12,13,14] that sometimes reflected the geographical origin of the isolates. Furthermore, combining MLST with data of antimicrobial susceptibility revealed the accumulation of pleuromutilin-susceptible isolates in specific clonal complexes [14,15]. Antibiotics such as pleuromutilins, tylosin, and lincomycin are the drugs most commonly used for the treatment of SD [1]. However, due to their wide use increasing resistance of B. hyodysenteriae isolates to these substances has become a global matter of concern. The pleuromutilin antibiotics tiamulin and valnemulin target the peptidyl transferase center (PTC), including parts of the 23S rRNA and the ribosomal protein L3. Previous studies demonstrated a significant association between pleuromutilin susceptibility and a single nucleotide change in the ribosomal protein L3 gene at position 443 (amino acid change Asn148Ser) [16,17].
B. hyodysenteriae isolates show differences in their virulence which may be due to the presence of certain VAGs [18,19]. Genes encoding hemolysins, outer membrane proteins, iron uptake systems, flagellin or NADH-oxidase have been associated with virulence [20,21,22,23]. A pathogenic relevance has also been suggested for six genes located on a recently identified 36 kb circular plasmid [3]. However, recent comparative studies including genomes of virulent and avirulent strains indicated that the relevance of the plasmid genes for virulence remains elusive [3]. So far, the presence of plasmid genes and other VAGs of B. hyodysenteriae could not unalterably be linked to the sequence type [3,24].
The present study aimed to determine the clonal origin, pleuromutilin susceptibility, and virulence gene profile of B. hyodysenteriae isolates and to investigate possible associations between these features to broaden the knowledge about the epidemiology of this important swine pathogen.

Multilocus sequence typing (MLST)
MLST was performed according to previously published protocols [10,11] targeting genes that encode for the following enzymes: alcohol dehydrogenase (adh), alkaline phosphatase (alp), esterase (est), glutamate dehydrogenase (gdh), glucose kinase (glpk), phosphoglucomutase (pgm), and acetyl-CoA acetyltransferase (thi). PCR amplicons were sequenced by LGC Genomics (Berlin, Germany) and sequences were analyzed using Ridom seqsphere version 1.0.1 (http://www.ridom.de/seqsphere/). Isolates were assigned to sequence types (STs) according to the allele profiles provided on the Brachyspira MLST website (http://pubmlst.org/brachyspira/ ). New STs and alleles were submitted to the curator of the website for assignments. To put our isolates into a global and temporal context, we included 421 B. hyodysenteriae-isolates from the MLST website, 52 isolates previously published from Spain and Portugal [12] as well as 160 additional isolates from Italy [14]. Isolates that shared at least six of the seven loci have been defined as a clonal complex (CC). eBURST analysis (http://eburst.mlst.net/) has been employed to determine the putative founder of a clonal complex, which determines the CC number.

Antimicrobial susceptibility testing and identification of single nucleotide polymorphisms in the ribosomal protein L3 gene
Antimicrobial susceptibility testing of B. hyodysenteriae isolates to tiamulin and valnemulin (Sandoz, Kundl, Austria) was performed as recently described [17]. Briefly, with 100 μl of the antimicrobial substances two-fold serial dilutions in BHI were prepared in micro titer plates (MTP; Greiner Bio-One, Frickenhausen, Germany). From each isolate, a bacterial suspension with an optical density of McFarland standard 1, which is equivalent to approximately 3 x 10 7 CFU (Colony Forming Units) was prepared [29]. One milliliter was diluted in 14 ml Brain-Heart-Infusion broth (BHI; Oxoid, Wesel, Germany) supplemented with 20% fetal calf serum (FCS; Biochrom, Berlin, Germany), and 100 μl each were distributed on the wells of a 96-well MTP, resulting in 2 x 10 5 CFU (absolute number) per well. MTPs were incubated under anaerobic conditions at 37˚C on a shaker (Edmund Bühler, Hechingen, Germany) with 125 circular motions per minute. The last dilution that inhibited bacterial growth was recorded as the minimal inhibitory concentration (MIC). MIC 50 and MIC 90 are defined as those concentrations at which 50% and 90% of the isolates are inhibited in growth. MIC 90 values were chosen to compare STs, whereas the MIC 50 value was chosen to analyze the antimicrobial susceptibility of the isolates over time. Type strain B204 was used as an internal control on each MTP.
As approved standards for antimicrobial susceptibility testing of B. hyodysenteriae are not provided by the Clinical and Laboratory Standards Institute (CLSI) or by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) MIC values were interpreted according to a previous publication [30] which was in line with the Swedish Veterinary Antimicrobial Resistance Monitoring program (SVARM).
To unveil pleuromutilin resistance mechanisms, i.e. previously reported single nucleotide polymorphisms inter alia in the ribosomal protein L3 gene (GenBank accession no. AF114845), we analyzed nucleotide sequences of the ribosomal protein L3 [17] from 31 isolates representing the most common STs of our study. For comparisons, the ribosomal protein L3 gene of the whole genome sequenced susceptible strain G21 (ST8 [31]) served as reference.

Determination of a possible link between the sequence type of B. hyodysenteriae and hemolysin gene sequences
For the concatenated sequences of the seven housekeeping genes and the sequences of seven hemolysin genes (excluding BHWA1_RS04705) of selected B. hyodysenteriae isolates distance matrices were created by use of the software Geneious version 8.1.8 (Biomatters, Auckland, New Zealand) based on the neighbour-joining-algorithm. The statistical calculation of the correlation between hemolysin genes and housekeeping genes was carried out with the Mantel test [40] in the open-source software R version 3.

Statistics and data analysis
Data were analyzed using the software BioNumerics version 6.6 (Applied Maths, Sint-Martens-Latem, Belgium), SPSS Statistics version 22 (IBM, Armonk, United States) as well as Geneious version 8.1.8 (Biomatters, Auckland, New Zealand). The BURST algorithm of eBURST version 3 (http://eburst.mlst.net/) was used to determine the affiliation of isolates into CCs.

MLST of B. hyodysenteriae isolates
Among the 116 B. hyodysenteriae isolates, three STs, namely ST52 (41.4%), ST8 (12.1%), and the newly assigned ST112 (25.9%), occurred predominantly. The remaining isolates were distributed among twelve STs (ST113-118, ST120-123, ST131, and ST193), which were previously unknown and have been deposited into the MLST database. Based on the nucleotide allelic profiles of our B. hyodysenteriae isolates, 421 isolates deposited to the MLST database (http:// pubmlst.org/brachyspira; last accessed 09/08/2017) and 212 isolates previously published but not submitted to the MLST database the eBURST analysis (S1 Fig The new ST112 is a DLV of ST20, which was described from Australian isolates and differs in the glpK and pgm alleles. The distribution of the 746 B. hyodysenteriae isolates is shown according to the country of origin ( Fig 1A) and the year of isolation ( Fig 1B) in the MSTs. Most (99.3%) of the 746 isolates originated from pigs, while few isolates were from birds (n = 3) and mice (n = 2). It becomes evident that the prominent STs identified in our study, namely ST8, ST52, and ST112, belong to separate clonal complexes. While the novel ST112 is limited to German and Belgian isolates only, the other two prominent STs also contain isolates from other European countries, i.e. Italy, Spain, Belgium, and Austria in the case of ST52, and Italy, the United Kingdom, Spain, Belgium, Poland, and Serbia in case of ST8. Except for ST120 that has been detected in Italy as well, the twelve novel STs appear unique to Germany. B. hyodysenteriae isolates from Europe, the United States of America, and from Australia hardly showed any overlap in their sequence types. Interestingly, ST104 contains isolates from Germany [3] and the United States of America. In addition, some countries, such as Italy, revealed the predominance of specific STs, such as ST75, ST77, and ST76 [15]. Regarding the temporal distribution of the 746 B. hyodysenteriae isolates the MST (Fig 1B) illustrates that a high number of isolates was obtained in the 2010s (n = 466) while from the 1990s, only eight isolates had been submitted to the MLST database. There are differences regarding the time of isolation among different STs. First, single STs or even CCs were only observed in the last century. This applies to isolates belonging to CC36 and CC30 which were only detected from the 1980s until the 1990s. Second, there are certain STs that were only obtained after the turn of century, like isolates of CC64 which were only detected in 2006/07. Furthermore, ST77, ST76, and ST83 were first detected in 2004. These three STs were continuously isolated until the end of our study period (2015/16). The novel sequence types ST94 and ST93 were only isolated between 2010 and 2013. Regarding the major STs in Germany, ST112 was not detected before the 2000s while the first ST8 and ST52 isolates were already present in the 1990s. While the detection rate of ST8 and ST52 isolates increased over time, the frequency of ST52 isolates peaked in the 2000s (35 isolates) and decreased in the 2010s (25 isolates).

Antimicrobial susceptibility of B. hyodysenteriae and identification of single nucleotide polymorphisms in the ribosomal protein L3 gene
MIC values ranged from 0.016 μg/ml to >16 μg/ml for tiamulin (MIC 50 : 2 μg/ml) and 0.004 μg/ml to >4 μg/ml for valnemulin (MIC 50 : 0.5 μg/ml) (Table 1). Overall, 39.1% of our   Table). The analysis of the mutation of the ribosomal protein L3 gene in 31 B. hyodysenteriae isolates of different STs at nucleotide position 443 (amino acid position 148), which has recently been linked with susceptibility to pleuromutilins [16,17], revealed that all eight tested ST112 isolates lacked this mutation, but only four of them were resistant to at least one pleuromutilin (Table 2). Among seven L3 gene sequenced ST8 isolates, four isolates showed the specific Asn148Ser mutation, with three isolates being classified as susceptible to both pleuromutilins. Of 16 ST52 isolates only two revealed a non-synonymous mutation at the specific position in the L3 gene. Isolates of this ST that lacked the mutation were classified as intermediate or resistant.

Distribution of virulence-associated genes (VAGs) among B. hyodysenteriae isolates
Most of the 15 VAGs tested by PCR were almost regularly present in the 116 B. hyodysenteriae field isolates. The iron-transporter genes ftnA and bitC were detected in 100% and 95.7% of the isolates, respectively. OMP gene bhlp16 was present in 70 isolates (60.3%), while bhlp17. 6 and bhmp39h could not be amplified from any of the field isolates. Bhlp29.7 was present in all isolates, whereas 87.9% of the isolates possessed bhmp39f. The eight hemolysin genes tlyA, tlyB, tlyC, hlyA, BHWA1_RS04705, BHWA1_RS09085, BHWA1_RS02195, and BHWA1_RS02885 were regularly detected in all isolates. VAG typing of the reference strains revealed comparable results, with both strains being positive for all genes but bhlp17.6 (B204) and bhlp16 as well as bhmp39h (B8044). Overall, we could determine five different VAG patterns among our B. hyodysenteriae field isolates ( Table 3). The most prevalent pattern (type 1) was represented by 62 isolates that each harbored 13 out of 15 VAGs. Interestingly, the presence of bhmp39f varied among isolates recovered from four of the six farms that had submitted samples repeatedly. In two farms (no. 31 and 81) the initial isolates did not possess the bhmp39f gene whereas it was present in those isolates recovered 12 and 13 months later. In the two other farms (no. 40 and 43) an exactly opposite situation was observed with initial isolates harboring bhmp39f and late isolates being negative for the gene (S2 Table). Links between sequence types, virulence genes and antibiotic resistance of B. hyodysenteriae

VAGs and antimicrobial resistance in relation to sequence types of B. hyodysenteriae
Due to the overall presence (hemolysin genes, iron-transferring genes, and OMP encoding genes bhlp29.7 and bhmp39f) or absence (bhlp17.6 and bhmp39h) of VAGs in most of our isolates, to establish a significant link between VAGs and STs was not possible. Merely the OMP gene bhlp16 showed a ST-related distribution ((p<0.0001), χ2 test), with all ST8 and ST112 isolates harboring this gene, while it was only present in 14.6% of ST52 isolates.
Comprehensive data of the isolates including the year of isolation, farm of origin, antimicrobial susceptibility, and sequence type are provided in S2 Table.

Determination of a possible linkage between the sequence type of B. hyodysenteriae and hemolysin gene sequences
Based on their nucleotide sequences (NS) four of the seven genes, namely the putative hemolysin genes BHWA1_RS09085, BHWA1_RS02885, BHWA1_RS02195 (yplQ) as well as tlyB, showed a significant (p<0.05) correlation towards the concatenated NS of the seven

Discussion
This study determined the clonal diversity of B. hyodysenteriae isolates from different regions of Germany and explored possible links between the phylogenetic background and the susceptibility of these isolates to pleuromutilins as well as to the presence of virulence gene profiles compiled from data of 15 VAGs.
We observed a predominance of ST8, ST52, and ST112 among B. hyodysenteriae isolates in Germany. Of the 421 B. hyodysenteriae isolates currently listed on the MLST website (http:// pubmlst.org/brachyspira/) (accessed at 09/08/2017) only one of the 196 non-European isolates has been assigned to one of these STs. In contrast, all three STs were frequently present among isolates circulating in different European countries, such as Italy, Spain, Belgium, and Poland. Hence, ST8, ST52, and ST112 seem to be geographically restricted, with the only exception of one ST52 isolate that was submitted from Japan (data from this study). The emergence of "European" B. hyodysenteriae STs may result from the enormous pig trade movement among European countries. This has recently been discussed as a probable reason for the wide dissemination of antimicrobial resistant B. hyodysenteriae isolates in Europe [14,15]. On the other hand, trade restrictions may lead to uneven geographical developments. Australia, for example, has banned the import of pigs in the mid-1980s [41]. This reasonably explains why STs observed among Australian B. hyodysenteriae isolates differ from those of other countries ( Fig  1A). Similarly, of isolates of B. hampsonii clade II and I only those from clade I (KC35 and EB106, GenBank accession no. JX197410 and JX197409) have been isolated in Europe [6,7].
The distribution of most of the German isolates amongst only a few STs (ST8, ST52, and ST112) is a finding that has been observed in other countries as well. In Italy (ST77, ST79 and ST83), the United States (ST93), and Spain (ST8 and ST71), certain STs were predominant  [13], and the development of different STs or MLVA (multilocus variable number tandem repeat analysis) types within the same CC over time in Italy [14]. In recent studies from Italy and Australia [15,41] different STs in the same farm were observed, but most of them represented single locus variants that belonged to the same clonal complex, suggesting a high relatedness between the isolates. In our study isolates originating from the same farm showed the same ST over longer time periods, such as ST120, that has been isolated on farm no. 95 in an interval of 19 months. This finding might indicate a long-term persistence of certain B. hyodysenteriae genotypes in pig farms. Deeper clonal analyses of such isolates would be necessary to determine, if even identical strains do persist over a certain time on pig farms. Re-entry of isolates through purchase of pigs from other farms and/or countries or through other sources such as mice, which have been suggested as a potential reservoir of B. hyodysenteriae [13], would also give reasonable explanations for the maintenance of STs in pig farms.
Although we included all published or submitted B. hyodysenteriae isolates for which STs have been published and/or submitted to the MLST website in our MST, the current picture of the population structure of B. hyodysenteriae as well as that for the temporal distribution of STs is likely biased by different study designs and by the lack of data from certain countries with relevant pig populations, such as The Netherlands, Norway, France, and several non-European countries. The availability of a representative MLST dataset of global B. hyodysenteriae isolates would be of immense value to identify possible transmission routes between pig herds on a national and international level [2,10,11,12,15]. Also the MST depicting the temporal distribution of STs (Fig 1B) must be interpreted with caution as it can only consider strains for which MLST and whole genome sequence data are available. Thus, it cannot be determined whether distinct STs such as ST112, ST83, ST77, ST76, ST93, and ST94 are indeed genotypes that have evolved most recently. To this time, it seems the there is an ongoing diversification of the pathogen B. hyodysenteriae. This has also been hypothesized in a recent study, where only six of 69 STs from B. hyodysenteriae isolates from Australia were maintained over time (1980-2000 until 2014-2016). This may either indicate an ongoing evolution of the pathogen or a source of isolates with highly diverse genotypes [41]. When combining the data of both MSTs (Fig 1A and 1B) presented in our study, ST8 and ST52 seem to play a key role in European countries. The first time of isolation for both STs was in the 1990s with both being continuously detected in the 2000s and the 2010s.
Regarding the VAGs of B. hyodysenteriae, this is one of the first studies that analyzed a representative number of isolates for all known and putative hemolysin genes. The almost regular detection of the eight hemolysin genes in our isolates confirms data from whole genome sequencing, where B. hyodysenteriae isolates also contained all these genes [3,31]. Besides the fact that the hemolysin genes seem to be ubiquitous, they have a strikingly high homology rate independent of the ST of the isolate. In our study, the hemolysin genes BHWA1_RS02195 and BHWA1_RS02885 were those with the highest variety in the nucleotide sequences (identity 97.8% and 99.0%, respectively). This is in accordance with previously published findings, where both genes showed the highest amount of variety [3,24]. By comparing the nucleotide sequences of hemolysin genes with the concatenated housekeeping genes, largely reflecting the core genome of the isolates, we observed that the putative hemolysin genes most likely followed the phylogeny of the core genome. On the AA level this correlation was only seen for one of them, namely BHWA1_RS02195 (YplQ), indicating a higher presence of non-synonymous mutations. Other hemolysin genes, such as tlyA, tlyC, and hlyA seem to be less conserved than the putative genes and thus might have undergone an independent evolution. This is likely controlled by other factors, which are unknown so far, than those mediating the phylogeny. Although this is rather speculative, it seems to be a most reasonable explanation for these observations.
Of the other VAGs the OMP encoding bhlp29.7 was present in all 116 isolates. This is in accordance with a similar study demonstrating the gene in all (n = 23) examined isolates [3]. Due to the wide distribution of this gene, whose function has not finally been resolved [33], it sounds convincing that Bhlp29.7 was previously recommended as suitable antigen for the serological diagnosis of SD [42]. Of the bhmp39a-h, forming a set of paralogous genes, bhmp39f was present in almost all isolates, while bhmp39h could not be detected at all. This is in contrast to former studies that demonstrated a wide distribution of both genes [43]. The remaining OMPs encoding genes (bhlp16 and bhlp17.6) are suggested to be located on the same gene locus, so that one isolate either harbors one or the other gene [44]. Interestingly, 46 of our isolates neither contained bhlp16 nor bhlp17.6 which has also been reported in other studies [3,32]. The absence of both genes was verified by use of additional primers (S1 Table) in the flanking region of the genes. Considering the sequence type, it is notable that only a few isolates (7 of 52) of the CC52 (i.e. ST52, ST116, ST131, and ST193 of this study) harbored the bhlp16 gene. Conclusive studies about the role of bhlp16 are not available so far, but since this gene is assumed to play a role in immune evasion [45] genetically different B. hyodysenteriae isolates must operate with different OMPs or pathways. Interestingly, also the occurrence of bhmp39f varied among isolates originating from the same farm. A reasonable explanation implicates that bhmp39f is not essential for the viability of the pathogen as it has recently been concluded for some plasmid genes possibly associated with virulence [41]. It should also be emphasized, that any negative finding could be the result of a false negative PCR due to variations in primer binding sites, which may particularly be the case in OMP genes that are involved in antigenic variability [43]. Whether mismatches result in an altered functionality of the protein cannot be determined at this moment.
Genes encoding iron-transfer proteins were also confirmed to be widely present in B. hyodysenteriae, which might be due to their essential role for the bacteria's vitality and virulence. Accordingly, the Bit system is absent in apathogenic B. innocens or less virulent B. pilosicoli [46]. In summary, most of the investigated VAGs, except for bhlp16, were widespread among B. hyodysenteriae isolates, which made it impossible to determine any linkages between the sequence type and possession of distinct virulence genes. However, we could demonstrate that isolates belonging to the same ST did not necessarily share the same virulence gene pattern. For example, the predominant ST52 contained all six virulence patterns found in this study. Nevertheless, it must be considered that PCR has a lower sensitivity than whole genome sequencing (WGS). Recently published data about WGS could show an accordance of STs with VAGs [3].
Data of antimicrobial susceptibility testing (40.9% pleuromutilin resistant isolates) confirmed the trend towards a reduced antimicrobial susceptibility of B. hyodysenteriae isolates from Germany to tiamulin and valnemulin in recent years [29,47]. Pleuromutilin resistance of B. hyodysenteriae is not limited to German isolates but has been observed in other countries as well. A recent study from Japan revealed high MIC 90 -values of tiamulin (6,25 μg/ml) and valnemulin (6,25 μg/ml) of 29 B. hyodysenteriae-isolates [48]. Also in Spain and the Czech Republic trends towards a reduced antimicrobial susceptibility of B. hyodysenteriae isolates were reported [49,50]. In contrast, Kirchgässner et al. (2016) [51] reported full susceptibility of B. hyodysenteriae isolates from Switzerland towards pleuromutilins, which might be due to limited trade activities of this non-EU country with the EU countries. Since the first detection of B. hyodysenteriae in fattening pig farms was in 2008 and in multiplier herds in 2016 [52] it can be assumed that selected therapy was rather uncommon, which leads to a low selection pressure. Our data indicate a correlation between the ST and pleuromutilin resistance of B. hyodysenteriae. Resistant isolates were predominantly found among ST52 and ST112, and isolates of ST52 with reduced susceptibility to pleuromutilins very likely emerged during the past 25 years (MIC 50 for valnemulin increased from 0,125 μg/ml to 4 μg/ml) ( Table 4). In contrast, isolates belonging to ST112 showed decreased MIC 50 -values for both pleuromutilins throughout the study period with a change from resistant to intermediate MIC 50 values. However, these findings do not necessarily reflect a general trend in resistance of B. hyodysenteriae to pleuromutilins since these data only refer to STs that contained a statistically representative number of isolates. If the observed differences in antimicrobial susceptibility may be due to increased hygienic efforts and infection control programs as well as an appropriate feeding and management, probably accompanied by a reduced usage of antibiotics, as recently reported, remains elusive [1,53]. Furthermore, mutations in ribosomal proteins can lead to increased susceptibility of B. hyodysenteriae isolates to pleuromutilins, like it was already demonstrated for the ribosomal protein L3 [17]. However, in two of six pig farms (no. 43 and 95) isolates recovered in intervals of three weeks and 19 months, respectively, changed in their antimicrobial susceptibility from susceptible to resistant. A fast shift towards resistance, as it was the case in farm no. 43, probably indicates the presence of already resistant strains in the pig population. Wide use of antibiotics in pig production including inappropriate dosages, incorrect application intervals and poor housing conditions as well as entry of resistant isolates by animal trade are plausible reasons for the increase of antimicrobial resistance [1,54,55].
Analyses of the ribosomal protein L3 encoding gene among isolates of the predominant STs, however, were not conclusive, except that the relevant SNP occurred more often in ST8 than in ST52 or ST112 isolates. The fact that changes in antimicrobial susceptibility over time caused by mutations in ribosomal genes coincide with unchanged STs, is very likely explainable by differences in the location and mutation rates of the genes that predict resistance and ST, respectively.
The proposed CC3 of Rugna et al. (2015), which resembles CC52 in our study, contains several STs, such as ST52, ST116, ST131, and ST193 that comprise a large number of pleuromutilin resistant isolates, with only marginal differences between countries [15]. Referring to another study from Italy, CC4 isolates (CC8 in our study) were largely susceptible to valnemulin [14], which was also true for our ST8 (CC8) isolates. This is also in accordance with a recent study from Belgium where ST8 isolates were mostly susceptible to pleuromutilins [56].

Conclusion
Our study revealed the presence of 15 STs of B. hyodysenteriae in Germany with three of them being dominant (ST8, ST52, and ST112). They are present in several European countries which may be due to the vast trading activities with pigs within the EU. More than one third of our isolates were resistant to both pleuromutilins that currently represent the most effective antibiotic substances used for the control of SD. Pleuromutilin resistance as well as susceptibility could be linked with CCs or even STs, suggesting a predictive role of MLST for the resistance phenotype of an isolate.
Regarding VAGs, only the OMP gene bhlp16 was negatively correlated with a certain group of isolates (CC52). All other VAGs including the hemolysin genes were either regular present or completely absent among the investigated isolates and a linkage to their STs could not be determined. Comparing nucleotide and amino acid sequences, we could show that the putative hemolysin genes most likely followed the phylogeny of the core genome whereas the other hemolysin genes, such as tlyA, tlyC, and hlyA, were less conserved and might have undergone an independent evolution, most probably due to other kinds of selective pressure. Extending the set of global, well-characterized isolates would be very helpful in broadening our understanding of the molecular epidemiology of B. hyodysenteriae isolates. Comprehensive MLST and virulence data could help to select epidemiologically relevant isolates for further analysis, such as whole genome sequencing, transcriptome analysis or targeted pathogenicity studies, and finally for the selection of vaccine candidates.