Serotypes, Antibiotic Susceptibility, Genotypic Virulence Profiles and SpaA Variants of Erysipelothrix rhusiopathiae Strains Isolated from Pigs in Poland

The aim of the study was phenotypic and genotypic characterization of Erysipelothrix rhusiopathiae strains isolated from diseased pigs in Poland and comparison of the SpaA (Surface protective antigen A) sequence of wild-type strains with the sequence of the R32E11 vaccine strain. The antibiotic susceptibility of the isolates was assessed using the broth microdilution method. Resistance genes, virulence genes, and serotype determinants were detected using PCR. The gyrA and spaA amplicons were sequenced to determine nonsynonymous mutations. The E. rhusiopathiae isolates (n = 14) represented serotypes 1b (42.8%), 2 (21.4%), 5 (14.3%), 6 (7.1%), 8 (7.1%), and N (7.1%). All strains were susceptible to β-lactams, macrolides and florfenicol. One isolate showed resistance to lincosamides and tiamulin, and most strains were resistant to tetracycline and enrofloxacin. High MIC values of gentamicin, kanamycin, neomycin, trimethoprim, trimethoprim/sulfadiazine, and rifampicin were recorded for all isolates. Phenotypic resistance was correlated with the presence of the tetM, int-Tn, lasE, and lnuB genes. Resistance to enrofloxacin was due to a mutation in the gyrA gene. All strains contained the spaA gene and several other genes putatively involved in pathogenesis (nanH.1, nanH.2, intl, sub, hlyA, fbpA, ERH_1356, cpsA, algI, rspA and rspB) Seven variants of the SpaA protein were found in the tested strains, and a relationship between the structure of SpaA and the serotype was noted. E. rhusiopathiae strains occurring in pigs in Poland are diverse in terms of serotype and SpaA variant and differ antigenically from the R32E11 vaccine strain. Beta-lactam antibiotics, macrolides, or phenicols should be the first choice for treatment of swine erysipelas in Poland. However, due to the small number of tested strains, this conclusion should be approached with caution.


Introduction
Erysipelothrix rhusiopathiae is the aetiological agent of erysipelas, one of the most wellknown infectious diseases in pigs. The pathogen can also infect poultry and other groups of animals, as well as humans. Despite advanced veterinary care, cases of porcine erysipelas still occur on farms worldwide, causing significant economic losses [1]. The disease affects both growing pigs over 3 months of age and adult pigs. It can be acute, subacute, or chronic, but subclinical infections without signs of disease occur as well. In the acute course of the disease, sepsis and sudden unexpected deaths occur in the herd. Other pigs may have high fever, depression, and mobility problems. The subacute form is also septicaemic but strains and the R32E11 vaccine strain in terms of the structure of the SpaA immunogen. Effective infection control requires knowledge of the characteristics of the etiological agent, and currently there are no reports of E. rhusiopathiae strains from pigs in Poland. During disease outbreaks, correctly identifying phenotypic and genetic differences between isolates is critical to better understanding disease epidemiology.

Isolation of Erysipelothrix rhusiopathiae Strains
E. rhusiopathiae strains were isolated post-mortem from tissues from 14 pigs bred on various farms located in three provinces in Poland between 2017 and 2022. The age of the animals ranged from 4 to 10 months (Table S2). Nine pigs from small backyard farms (No. 1,3,4,6,7,10,11,12,13, and 14, corresponding to E. rhusiopathiae strains 1S, 3S, 4S, 6S, 7S, 10S, 11S, 12S, 13S, and 14S, respectively) were not vaccinated against erysipelas, while 4 pigs from large pig farms (No. 2, 5, 8, and 9, corresponding to strains 2S, 5S, 8S, and 9S, respectively) were vaccinated (information obtained from veterinarians or breeders shows that the ERYSENG ® PARVO vaccine, Hipra, was used most often, but precise information in this regard is not known). Most pigs showed no obvious signs of disease. However, some animals admitted to the slaughterhouse developed skin lesions, probably due to post-transport stress ( Figure S1). The veterinary history showed that several pigs had previously suffered from erysipelas and been treated with antibiotics. Samples were taken at the slaughterhouse from pork half-carcasses. The post-mortem examination revealed skin lesions or thickening in the subcutaneous tissue (Table S2). Carcasses of all individuals with symptoms of erysipelas were disposed of. Bacteria were isolated from nodular lesions located deep in the subcutaneous tissue, pathologically altered heart valves, muscular fascia, and lymph nodes located closest to the affected sites or synovial fluid. The collected tissues were homogenized in glass mortars and then suspended in TKT Edwards modified broth with 5% haemolysed sheep blood (Biomaxima, Lublin, Poland). After 48 h incubation at 37 • C, the material was plated on blood agar (GRASO Biotech, Owidz, Poland) and incubated at 37 • C for 48 h. Cultures growing as small, greyish translucent colonies and having the morphology of slender gram-positive rods were considered E. rhusiopathiae. Alpha haemolysis was observed for most strains ( Figure S1). Pure cultures were propagated on BHI broth with the addition of 0.1% Tween 80 (Merck, Warsaw, Poland).

Identification of E. rhusiopathiae Strains
The collected isolates grown on Columbia agar supplemented with 5% blood (BTL, Łódź, Poland) were identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (MS) using a standard ethanol/formic acid extraction method [23]. The mass spectra obtained from each isolate were processed with the MALDI Biotyper ® 3.1 database (Bruker Daltonics, Bremen, Germany), which contains 8468 mass spectra of reference strains, including 10 strains of E. rhusiopathiae. Identification to species level was considered reliable with log(score) ≥ 2.000 [23].

Serotyping of E. rhusiopathiae Strains
Serotyping was performed based on four multiplex PCR protocols [7,24]. The sequence of primers, annealing temperature, and size of PCR products are listed in Table S3. PCR reactions were performed using DreamTaq Green DNA Polymerase (Thermo Fisher Scientific Baltics UAB, Vilnius, Lithuania). Reference E. rhusiopathiae strains (Fujisawa serotype 1a, ATCC 19414 serotype 2, R32E11 serotype 2, Tuzok serotype 6, and Bano serotype 21) were used as positive controls. Strain ATCC 19414 was obtained from Argenta (Poznań, Poland) and other reference strains were provided in the form of genomic DNA by Dr. Shimoji, National Institute of Animal Health, Japan.
Inocula were prepared by suspending bacteria in 0.85% NaCl to obtain a density of 0.5 McFarland. A 150 µL volume of the inoculum was added to 10 mL of BHI broth (BTL, Łódź, Poland) containing 0.1% Tween 80. Microdilution plates were inoculated with 50 µL of bacterial suspension and 50 µL of the appropriate antibiotic concentration. Plates were incubated at 36 • C in 5% CO 2 for 45 h, and MIC values were read as the lowest concentration of an antimicrobial agent at which visible growth was inhibited. The reference strain ATCC 19414 was analysed in parallel with the wild strains. The quality control of antimicrobial substances was carried out using E. coli strain ATCC 25922 and Müller-Hinton broth [25].
Categorization of E. rhusiopathiae strains as susceptible, intermediately resistant, and resistant was carried out based on CLSI guidelines (document Vet06, 2017) [26]. For some antimicrobials not included in this guide, breakpoints recommended for other antibiotics of the same class or for other types of gram-positive bacteria have been adopted (Table 1). In the case of aminoglycoside antibiotics (gentamicin, kanamycin, neomycin, streptomycin, and spectinomycin), folic acid inhibitors (trimethoprim and trimethoprim/sulfadiazine), and rifampicin, no cut-off points were proposed due to high MIC values or/and their unimodal distribution.

Isolation of DNA
Whole-genome DNA was extracted from E. rhusiopathiae strains using the Gene MA-TRIX Bacterial and Yeast Genomic DNA Purification Kit (Eurx, Gdańsk, Poland) following the manufacturer's protocol, modified to extend the incubation time of bacteria in the lysis buffer from 15 min to 1 h (gram-positive bacteria, unlike gram-negative bacteria, have a thick cell wall, the removal of which requires a longer incubation of the bacteria with lytic enzymes). DNA concentration was determined using the NanoDrop Lite spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), and the quality of DNA was checked by agarose (1.5% w/v) gel electrophoresis. The final DNA concentration was~18 ng/µL.

Sequence Analysis of the gyrA Gene
The sequencing of the PCR products (613 bp fragment of the gyrA gene) of representative enrofloxacin-susceptible and enrofloxacin-resistant strains was carried out using the Sanger method in the external service laboratory of Nexbio Sp. z o.o. (Lublin, Poland). Chromatograms were analysed using Chromas Lite (ver. 2.6.6, Technelysium Pty Ltd., South Brisbane, Australia), and DNA sequences of gyrA genes were deposited in Gen-Bank (Accession Nos. OP921301-OP921305). Amino acid (aa) sequences were predicted using the NCBI translate tool ORF finder (https://www.ncbi.nlm.nih.gov/orffinder/, accessed on 26 November 2022). The ClustalW Multiple Alignment tool (MEGA X software, https://www.megasoftware.net/, accessed on 26 November 2022) was used to align predicted aa sequences. The gyrA sequence of reference strain E. rhusiopathiae ATCC 19414 (enrofloxacin-susceptible) was retrieved from the NCBI GenBank database (Acc. No. LR134439.1).

Whole spaA gene Amplification and Sequence Analysis
Amplification of the entire spaA gene was performed using primers Spa-fw (5 -ATGAAAAAGAAAAAACACCTA-3 ) and Spa-rv (5 -CTATTTTAAACTTCCATCGTT-3 ) [16] and DreamTaq polymerase (Thermo Fisher Scientific Baltics UAB, Vilnius, Lithuania) at an annealing temperature of 49 • C. In addition to the wild-type strains (n=14), the R32E11 strain, that is used in the production of the ERYSENG ® and ERYSENG ® PARVO vaccines (Laboratorios Hipra S.A., Amer, Spain), was also used in the study. DNA of the R32E11 strain was provided by Dr. Shimoji, National Institute of Animal Health, Japan. PCR products were sequenced using the Sanger method in the external service laboratory of Nexbio Sp. z o.o. (Lublin, Poland). The DNA SequenceReverse and Complement Online Tool (http://www.cellbiol.com, accessed on 24 November 2022) was used to determine the consensus sequence of the spaA gene, and aa sequences were predicted using the NCBI translate tool ORF finder (https://www.ncbi.nlm.nih.gov/orffinder/, accessed on 26 November 2022). DNA sequences of spaA genes were deposited in GenBank (Accession Nos. OP822679-OP822691, OQ054982 and MZ448116).
The ClustalW Multiple Alignment tool (MEGA X software) was used to align the predicted aa sequences of the SpaA of different isolates. To determine the evolutionary distance between the strains, a comparative analysis of the 447-aa N-terminal fragment (covering 29 aa-signal sequence + 384 aa-immunoprotective domain + 34 aa-proline-rich domain) of the SpaA protein sequence was performed. The spaA sequences of reference E. rhusiopathiae strains and several wild-type strains isolated from pigs (in various countries) and water poultry (in Poland) included in the analysis (n = 17) were retrieved from the NCBI GenBank database (Table S7). Clustering was conducted in MEGA X using the maximum likelihood method with a bootstrap support value of 500. All positions containing missing data were eliminated (complete deletion option). Due to the cut-off of illegible initial segments of the chromatograms, there were a total of 435 positions in the final dataset (the first 12 amino acids of the SpaA corresponding to the signal domain were missing).

Identification of E. rhusiopathiae Isolates
The bacteria isolated from porcine tissues were identified by mass spectrometry as E. rhusiopathiae (n = 14), and the log(score) for all samples was in the range of 2.028 to 2.312.

Genotypic Resistance Profiles
Genotypic resistance profiles were compatible with phenotypic resistance. All tetracycline-resistant isolates contained the tetM gene (coding for ribosomal protection protein, which catalyses the release of tetracycline from ribosomes in a reaction dependent on GTP) and the Tn916/Tn1 transposon integrase gene. The 8S strain resistant to lincosamides and tiamulin contained genes lnuB (coding for lincosamide nucleotidyltransferase) and lsaE (coding for ABC-F efflux pump) ( Table 2). The aadK gene encoding aminoglycoside nucleotidyltransferase was detected in all isolates; however, based on the results (unimodal distribution of MIC values), its role in the resistance of E. rhusiopathiae strains to aminoglycoside antibiotics remains unclear. None of the seven other considered genes determining resistance to aminoglycosides were detected in the isolates tested (Table 2). Sequence analysis of the gyrA gene showed that resistance of E. rhusiopathiae strains to enrofloxacin (MICs 8-16 µg/µL) is due to a mutation at position 257 (ACA→ATA or ACA→AAA), which translates into a change in the aa sequence, Thr86→Ile or Thr86→Lys86 (Table 3).

Virulence-Associated Genes
The genotypic virulence profiles of the tested strains of E. rhusiopathiae were identical. All isolates contained the spaA gene encoding the immunogenic SpaA protein, the nanH.1 and nanH.2 neuraminidase genes, and other genes whose products may enable invasion of host tissues (intl, sub, hlyA, fbpA, ERH_1356, rspA, and rspB) or determine resistance to attack by complement and phagocytic cells (cpsA and algI) ( Table 4).

SpaA Sequence Analysis
The size of the spaA amplicon in the tested E. rhusiopathiae isolates ranged from 1761 to 1881 bp (corresponding to 587-626 aa), while in the case of the reference R32E11 strain it was~2200 bp. This variation in the size of the PCR products was noticeable in the electropherogram (data not shown) and resulted from the different number of 60-nucleotide tandem repeats of the DNA sequence at the 3 end of the spaA gene.
In most of the tested strains (9/14; 64.3%, serotypes 1b, 2, 5, 6, and N), there were nine 20-aa repeats containing a GW module (corresponding to a spaA gene length of 1881 nt) in the predicted SpaA sequence, as in the reference Fujisawa and ATCC 19414 strains. Four strains (4/14; 28.6%) representing serotype 1b (4S, 11S, 12S, and 13S) contained 8 tandems (corresponding to a total spaA gene length of 1818 nt) and one strain (1/14; 7.1%) (8S, serotype 8) contained 7 tandems (total spaA gene length 1761 nt). Signal sequences spanning 29 aa were conserved among the SpaA protein of all isolates. However, it should be noted that the SpaA sequences obtained in this work were missing the first 11 or 12 aa (hence the analysed signal sequences contained only 17 aa) due to the need to remove the illegible ends of the chromatograms.
Following a comparative analysis of wild-type E. rhusiopathiae strains and R32E11 vaccine strain, as well as the highly virulent Fujisawa strain [9] and the reference strain ATCC 19414, 15 nonsynonymous mutations were found within the SpaA hypervariable domain (384-aa in length; region 30-413 aa) and the proline-rich region (34-aa in length; 414-447 aa). On this basis, 7 variants of the SpaA protein were distinguished in the wildtype strains (Table 5).
SpaA hypervariable region sequences (immunoprotective domain, SpaA antigen) of all wild-type E. rhusiopathiae strains differed from those of the R32E11 vaccine strain. The difference concerned one (strain 14S), two (1S), six (2S), seven (3S, 4S, 5S, 6S, 7S, 9S, 10S, 11S, and 13S), eight (8S) or even nine (12S) aa; taking into account the proline-rich domain as well, the difference increases to 8 aa in the case of the 5S and 9S strains. All strains representing serotypes 1b and 5, except the 12S strain, had an identical aa sequence in the SpaA immunoprotective domain. The variation at position 38 (Pro38→Gln) was unique to the strain 1S serotype N, the variation at position 54 (Gly54→Ala) was found only in the strain 8S serotype 8, polymorphism at positions 139 (Gln139→Lys) and 232 (Ile232→Thr) reported only in the 12S strain, and the variation at position 423 (Pro423→Gln) was unique to the 5S and 9S strains of serotype 2 ( Table 5).
As with the R32E11 vaccine strain, none of the wild-type isolates showed 100% homology to the Fujisawa strain in terms of immunoprotective domain and proline-rich region sequences. The highest homology was found in the 2S serotype 6 strain, in which variation was noted in only three aa positions (257, 426, 435). The remaining isolates differed by 4-6 aa from the Fujisawa strain (Table 5).
The sequence of the 14S strain was identical to that of the reference strain ATCC 19414 (isolated from the spleen of a pig with endocarditis in 1950), and the 1S strain serotype N differed from the ATCC 19414 strain in only one aa (Gln38) ( Table 5).
Comparative analysis of 31 strains of E. rhusiopathiae (14 strains tested in this study and 17 additional strains) showed that the sequence of the N-terminal 447-aa region of SpaA (including signal, immunoprotective, and proline-rich domains) in 7 of the 14 tested strains, representing serotype 1b or 5, was homologous with the sequence of E. rhusiopathiae strains isolated from pigs in China (AQ 150414 serotype 1a) and Japan (Ireland serotype 6), as well as from geese in Poland (1023 serotype 2, 48W serotype 5). The clustering revealed a relationship between the serotype and the sequence of the immunoprotective domain of the SpaA protein. The strain 8S serotype 8 clustered with the two E. rhusiopathiae goose strains (759W and 1092) representing serotype 8, while the 2S serotype 6 strain had an identical aa sequence to that of strain 846 serotype 6 isolated from geese in Poland. The two serotype 2 strains (5S and 9S) formed a common cluster with the goose serotype 2 strain 219. On the other hand, two other serotype 2 strains, 14S and 1023, formed clades with strains of different serotypes. The sequence of the strain 1S serotype N was homologous with the sequence of strain 49W serotype N isolated from domestic goose in Poland. The dendrogram also shows that, among the total 31 analysed strains representing 7 serotypes (1a, 1b, 2, 5, 6, 8, and N), 7 and 8 tandem repeats in the C-terminal region of SpaA were present only in strains of serotype 8 (2/3) and 1b (4/6), respectively ( Figure 2). Table 5. Nonsynonymous mutations in the N-terminal hypervariable and proline-rich region of the spaA gene and number of C-terminal tandem repeats in the wild-type E. rhusiopathiae strains compared with the corresponding sequence of the E. rhusiopathiae R32E11 vaccine strain as well as the Fujisawa and ATCC 19414 reference strains. Tandem Repeats  38  54  55  70  101  109  139  178  195  232  257  303  423  2 strain 219. On the other hand, two other serotype 2 strains, 14S and 1023, formed clades with strains of different serotypes. The sequence of the strain 1S serotype N was homologous with the sequence of strain 49W serotype N isolated from domestic goose in Poland. The dendrogram also shows that, among the total 31 analysed strains representing 7 serotypes (1a, 1b, 2, 5, 6, 8, and N), 7 and 8 tandem repeats in the C-terminal region of SpaA were present only in strains of serotype 8 (2/3) and 1b (4/6), respectively ( Figure 2).

Occurrence of E. rhusiopathiae Infections in Pigs
Most of the cases of erysipelas reported in this study were detected in unvaccinated pigs from backyard farms. The animals apparently showed no symptoms, or only mild symptoms, of the disease, and the infection was found only during the post-mortem examination. Some of the pigs included in the study had previously been treated for symptoms suggestive of erysipelas. It is therefore possible that antibiotic therapy did not have a completely curative effect in these individuals, and the disease became chronic. Vaccination against swine erysipelas in Poland is not mandatory and is usually not performed on backyard farms for economic reasons. The cost of the smallest package of vaccine can be higher than the cost of antibiotic therapy. The occurrence of erysipelas in small pig houses may be associated with the use of straw as bedding, unpaved paddocks, and insufficient removal of manure, which may be sources of E. rhusiopathiae [3]. Erysipelas is less common in large piggeries with a closed production cycle, equipped with structures for automatic removal of animal excrement. However, it sometimes occurs even in single vaccinated pigs bred on farms with proper biosecurity rules. The onset of the disease in such cases may be the result of human error during vaccination (inappropriate vaccine handling), or individual characteristics associated with the failure to develop an immune response following immunization [31,32]. It should also be considered that commercial vaccines may not be fully effective against field strains representing a different serotype and variant Spa antigen than the vaccine strains. In addition, field strains may differ from vaccine strains in terms of other surface antigens. The cases of erysipelas in vaccinated pigs reported in this study were caused by strains of serotypes 2 (5S, 9S), 6 (2S) and 8 (8S). A case of infection in a vaccinated pig caused by a SpaA-positive strain of serotype 6 was also reported by Shimoji et al. [33]. The results presented in this paper may be important in epidemiological analyses and in the selection of epidemiologically relevant candidates for vaccines.
Serotyping may be helpful in epidemiological studies to assess the spread of strains in the environment, especially locally between farms. Several studies have shown that immunization of pigs with inactivated (bacterin) or attenuated serotype 2 strains confers immunity against serotype 2 and 1 strains and, to varying degrees, against strains of other serotypes, including 5, 6, 8, and N, detected in this study [37][38][39]. Significant variation in cross-protection was noted in an experiment using mice. Animals vaccinated with attenuated strain Koganei 65-0.15 serotype 2 survived when challenged with serotypes 1b, 2, 8, and type N, but mortality occurred in mice exposed to serotypes 1a, 11, 12, 15, 16, or 21 (20-30% mortality), 4, 5, 6, 7, or 8 (40-50% mortality), and 9, 10, 18, or 19 (60-80% mortality) [38]. However, Forde et al. [36] suggest a cautious approach to the results of these provocation studies since they included only one representative strain of serotype 2 and that the results may have been influenced by immunogenic characteristics other than the serotype. It seems, therefore, that the question of whether the serotype is an important feature determining cross-protection requires more thorough research.

Antibiotic Susceptibility and Genotypic Resistance Profiles
The widespread susceptibility of E. rhusiopathiae strains to ampicillin and ceftiofur demonstrated in this study is fully consistent with several previous reports [1,11,28,40,41]. It is surprising that, in the era of rapidly increasing drug resistance among bacteria and the high use of beta-lactam antibiotics in animal husbandry [42], strains of E. rhusiopathiae isolated in various geographic regions of the world are still susceptible to this group of antimicrobial substances. This is good news for veterinarians and farmers, as penicillin is currently the drug of choice for swine erysipelas [2]. In pig medicine in Poland, amoxicillin and amoxicillin with clavulanic acid can also be used, as well as two other antimicrobial substances to which all E. rhusiopathiae strains were sensitive (tylosin and florfenicol). The reported susceptibility to florfenicol is consistent with the previous studies by Fidalgo et al. [43], who showed the MIC of chloramphenicol in E. rhusiopathiae strains (n = 60) ranged from 8 to 16 µg/mL. The widespread susceptibility of E. rhusiopathiae isolates to macrolides noted in this study is in contrast to the results of Wu et al. [28], who showed that as many as 53.3% of strains from pigs in China (collected between 2012 and 2018) were resistant to erythromycin. Interestingly, Ding et al. [1], examining isolates also collected in China (between 2012 and 2013), did not record erythromycin MIC values exceeding 1 µg/mL in any strain (all tested isolates were sensitive).
According to the adopted criteria, only 1 of 14 (7.1%) of the tested isolates was resistant to lincomycin, clindamycin, and tiamulin. Previously, widespread susceptibility to tiamulin (≤6.25 µg/mL) was recorded in E. rhusiopathiae strains isolated from pigs in Japan [41] and in Brazil [40]. The lsaE resistance gene detected in a tiamulin-resistant strain has previously been identified as part of plasmid-borne or chromosomal multi-drug resistance gene clusters in other gram-positive bacteria [44,45], and has recently been found in E. rhusiopathiae strains from pigs [28]. The results obtained in this work are consistent with previous reports from Japan showing that the vast majority of E. rhusiopathiae strains (obtained between 1988 and 1998) were susceptible to lincomycin and clindamycin [41]. Completely opposite results were recorded in China, where the percentage of strains resistant to lincosamides ranged from 64% to 72% [1,28]. The presence of the lincosamide nucleotidyltransferase lnuB gene in a phenotypically-resistant strain has previously been reported in various gram-positive bacteria from pigs, including E. rhusiopathiae [46][47][48].
The prevalence of tetracycline-resistant strains (71.4%) reported in this study is higher than that reported in China and Japan, where 50.8-60.4% of E. rhusiopathiae strains were tetracycline-resistant and 38% were doxycycline-resistant [1,28,41]. The reason for such a high prevalence of tetracycline-resistant strains may be the frequent use of tetracyclines in pig farming in Poland [49]. The correlation observed between tetracycline resistance and the presence of the tetM gene is consistent with the findings of Wu et al. [28], who showed the presence of tetM in 57.4% of tetracycline-resistant strains. The coexistence of the tetM gene and the transposon integrase Tn916 indicates the involvement of this mobile genomic element in the spread of the tetM gene in E. rhusiopathiae strains.
The percentage of strains resistant to enrofloxacin (57.1%) reported in this study is similar to the result obtained by Ding et al. [1] in China (~70% of E. rhusiopathiae strains showed resistance to norfloxacin and levofloxacin), but much lower than the percentage of fluoroquinolone-resistant E. rhusiopathiae strains recorded in other studies [28,40]. The reported mechanism of resistance associated with the mutation at position 86 (Thr86→Ile) of the GyrA subunit of DNA gyrase is consistent with the previous findings of Wu et al. [28]. However, it should be noted that these authors also reported mutations at position 90 of GyrA and position 82 of the ParC subunit of DNA topoisomerase IV in E. rhusiopathiae strains resistant to ciprofloxacin [28].
The high MICs of gentamicin, kanamycin, neomycin, trimethoprim, trimethoprim/ sulfadiazine, and rifampicin reported in this study, together with the unimodal MIC distribution and the absence of resistance genes, indicate that E. rhusiopathiae strains are inherently resistant to these antimicrobials. Wild unimodal MIC distributions were noted for streptomycin and spectinomycin, but their MIC values were lower than those of the other aminoglycosides. Interestingly, the aadK gene, that encodes aminoglycoside 6-adenylyltransferase, the streptomycin-modifying enzyme, was detected in all strains tested. It should be noted, however, that the sequence of the aadK gene of the Fujisawa strain (GenBank Acc. No. AP012027.1, locus_tag="ERH_1545) is not homologous with the sequence of the aadK gene found in other bacteria, i.e., in the strain Bacillus subtilis 168 (Acc. No. NG_047379.1) [50] and in strain 4300STDY6542365 Klebsiella pneumoniae (Acc. No. UFEG01000012.1) (data not shown, obtained from BLAST analysis). The widespread occurrence of high MIC values of kanamycin (MIC >100 µg/mL) in E. rhusiopathiae strains has also been reported by other authors [1,11,41]. However, the gentamicin susceptibility results obtained in this study differ slightly from those of Ding et al. [1], who demonstrated a wide range of gentamicin MICs (2->128 µg/mL) among E. rhusiopathiae strains from China. The resistance of E. rhusiopathiae strains to folic acid inhibitors (trimethoprim, trimethoprim/sulfadiazine) is consistent with the results of previous studies [1,11,40]. The high rifampicin MIC values (MIC >128 µg/mL) indicative of intrinsic resistance have not been previously reported in E. rhusiopathiae. Other gram-positive bacteria are generally susceptible (MIC ≤ 2 µg/mL) to this RNA polymerase inhibitor [51,52].

Virulence Genes
The common occurrence of the spaA gene in E. rhusiopathiae isolates recorded in this study is consistent with the results of other studies on both porcine and poultry strains [16,34]. The exclusive discovery of spaA and the lack of spaB and spaC in the collected E. rhusiopathiae isolates may be because the strains belong to serotypes for which the presence of spaA is characteristic (1b, 2, 5, and 8) [17]. The presence of the spaA gene was recorded even in the 2S strain of serotype 6, although earlier reports showed that strains of this serotype are usually spaB-positive [17]. The prevalence of other selected potential virulence genes in E. rhusiopathiae strains noted in this work is also consistent with several other reports [1,16,53]. Janßen et al. [16] demonstrated the presence of the genes nanH.1, ERH_1356, intl-like, rspA, rspB, algI, sub, hlyA, fbpA, and hlyIII in all tested E. rhusiopathiae strains (n = 165). Only the intl gene encoding internalin-like protein was absent in 15% of the analysed isolates, primarily from poultry and sheep. Zhu et al. [15] showed that the product of the intI gene (ERH_1472) acts as an adhesin enabling specific adherence of E. rhusiopathiae to the surface of pig iliac arterial endothelial cells. In Listeria monocytogenes, internalin contributes to the invasion of the bacteria into epithelial cells [54].
A study based on analysis of whole genome sequences of eight virulent strains of E. rhusiopathiae (Fujisawa, NCTC8163/ACTC 19414, WH13013, ZJ, ML101, GXBY-1, SY1027, and KC-Sb-R1) showed that genes encoding enzymes involved in synthesis of the bacterial capsule (cpsA, cpsB, and cpsC), neuraminidase (nanH), hyaluronidase (hylA, hylB, and hylC), and surface proteins (spaA, rspA, and rspB) are core genes [53]. The results of our study differ somewhat from those of Ding et al. [1], who failed to detect the hylA (ERH-0150), nanH.1, and ERH-1356 gene in 8.3%, 6.25%, and 22.9% of E. rhusiopathiae strains from pigs, respectively. However, the negative result was probably due to the fact that the primers designed by these authors were complementary to the sequences outside these genes (this conclusion is based on the BLAST analysis of primer annealing sites to the sequence of the Fujisawa strain, Acc. No. AP012027.1). The reported discrepancies may also be due to the small number of strains tested in these studies.
It should be emphasized that the role of only some of the putative virulence genes detected in this work has been confirmed in the pathogenesis of erysipelas. Zhu et al. [22], in a study with recombinant E. rhusiopathiae SpaA (rSpaA), demonstrated that this protein adheres to porcine endothelial cells, and its plasminogen-binding activity is highly likely to play a role in this adhesion. Li et al. [29], based on comparative analyses of gene expression in the highly virulent HX130709 and its isogenic avirulent derivative HX130709a, showed that SpaA and neuraminidase are key virulence factors of E. rhusiopathiae. Earlier research by Shimoji et al. [55] demonstrated that the virulence of E. rhusiopathiae is largely dependent on the presence of the capsular antigen, but not on hyaluronidase. Literature reports on the role of RspA and RspB proteins in the pathogenesis of E. rhusiopathiae infections are ambiguous. Initially, they indicated that these proteins are exposed on the cell surface of E. rhusiopathiae and participate in biofilm formation. Moreover, recombinant RspA, but not RspB, elicitated protection in mice against experimental challenge [56]. However, the results of a recent study by Li et al. [29] indicate no relationship between RspA and RspB and the virulence of E. rhusiopathiae.

SpaA Variants
Several authors have studied spaA gene sequences in E. rhusiopathiae strains, analysing nonsynonymous mutations in the region corresponding to the N-terminal immunoprotective domain (corresponding to 30-413 aa) and the number of tandem repeats in the C-terminal segment of the gene (corresponding to 448-626 aa in SpaA proteins containing 9 tandem repetitions) [16,17,19]. The 100% sequence homology between isolates in the signal region of SpaA (1-29 aa) reported in this work is consistent with previous findings [16,17].
Our results of analyses of the C-terminal region of SpaA, that is responsible for binding the protein to the bacterial cell surface, are consistent with the research of Janßen et al. [16], who showed that SpaA of E. rhusiopathiae strains from poultry and pigs in Germany most often contains 9 tandem repeats (89.7% strains), but their number varies from 7 to 13. In this work, the relationship between the number of tandem repeats and the serotype was demonstrated for the first time (8 tandems were found only in strains of serotype 1b, and 7 tandems were specific for serotype 8). A recent study by Wu et al. [57] showed that the number of tandem repeats in the SpaA chain affects the adhesive properties of E. rhusiopathiae. The ∆spaA mutant strain (mutated SE38 strain virulent in pigs) producing the SpaA protein truncated by 2 tandem repeats (120 nt deletion) displayed attenuated virulence in mice and decreased adhesion to porcine endothelial cells [57]. Thus, the strains tested in this work that have 7 or 8 tandems in the C-terminal region of SpaA can be expected to be less virulent than strains producing SpaA with 9 repeats.
Based on polymorphisms in the N-terminal 447-aa region of SpaA, including a 384aa hypervariable region (30-413 aa) and a 34-aa proline-rich region (414-447 aa), the 14 isolates were classified into 7 groups. The aa substitutions recorded at positions 55,70,101,178,195,257, and 303 of the SpaA protein chain were previously observed in E. rhusiopathiae strains [16,19], while the polymorphisms at positions 38, 54, 109, 139, and 232 (hypervariable domain) and 423, 426, and 435 (proline-rich domain) have not previously been described. The His109 mutation was unique to the R32E11 vaccine strain. Janßen et al. [16], based on the aa substitutions in the N-terminal protective region of SpaA, divided the strains tested (n = 165) into five groups (I-V). However, in contrast to our results, these authors did not observe a relationship between the serotype and the SpaA variant [16]. Most of our strains (9/14, 64.2%) (serotype 1b, 2 and 5) correspond to their group II SpaA (Ser101 and Ile257). However, unlike the prevalence of this SpaA variant in our strains, Janßen et al. [16] included only one of 36 isolates (2.8%) from pigs and 21.1% of isolates from other hosts, mainly chickens and turkeys, in group II. The SpaA sequence of the 14S strain (Ile55, Asn70, Asp178, Asn195, Ile257 and Glu303) is homologous with the sequence of the reference strain ATCC 19414 and with SpaA variant I, to which Janßen et al. [16] assigned the majority (52.7%) of tested strains, including 18 (50%) of 36 from pigs. It should be noted that Janßen et al. [16] assigned an incorrect aa to the GAG codon, which resulted in the Glu303 polymorphism being designated Gln303. The SpaA sequence of our strain 2S serotype 6 is homologous to SpaA variant III (Ile257) distinguished by Janßen et al. [16]; this SpaA variant, unique to our isolates, was found in 22.4% of E. rhusiopathiae strains tested in Germany, including 13 of 36 porcine strains (36.1%). The strains 1S, 8S, and 12S represent new, as yet undescribed, variants of SpaA; however, the high homology of the 1S strain with group I distinguished by Janßen et al. [16] should be noted (the difference concerns only Gln38). Among isolates (n = 34) from pigs in Japan, three SpaA variants were distinguished, based on substitutions at aa positions 195, 203, and 257 [19].
Sequence analyses of the SpaA of the R32E11 vaccine strain against the sequence of the wild-type E. rhusiopathiae strains have thus far not been performed. Of the 14 pig isolates, none showed 100% homology to the sequence of the immunoprotective domain of strain R32E11 (differences ranged from 1 to 8 aa). It is also worth mentioning that the number of tandem repeats in the C-terminal section of SpaA in this strain (13) is much higher than in field isolates (7-9). It is not known whether the structural variations in the SpaA immunogen of the R32E11 strain translate into a protective effect of the vaccine against infections with field E. rhusiopathiae strains. Only studies in animal models could clarify this issue.

Limitations of the Study
The main limitation of the current study was the small sample size. The analyses carried out on 14 strains, do not give a full view of the antibiotic susceptibility, serotypes or SpaA variants of E. rhusiopathiae strains causing swine erysipelas in Poland despite the fact that they came from three regions of the country. There is a need for further research in this area on a larger scale. We also regret that, due to the lack of public access to the strains based on which vaccines against erysipelas are prepared, we could not include them in the comparative analyses. It should be noted that the DNA of the R32E11 strain was not isolated directly from the vaccine (previous attempts to isolate DNA from the ERYSENG ® , Hipra, vaccine failed [36]), but from the E. rhusiopathiae R32E11 strain deposited in the collection of the National Institute of Animal Health in Japan.

Conclusions
The research results presented in this paper are the first reports on the occurrence of porcine erysipelas in Poland and the characteristics of E. rhusiopathiae strains. Despite the use of a small number of samples, we have shown that E. rhusiopathiae strains causing swine erysipelas in Poland are diverse in terms of antibiotic susceptibility, serotypes, and variants of the SpaA immunogen. Analyses involving the R32E11 vaccine strain provided valuable information on its relationship with circulating field strains and created space for possible research on the development of a new swine erysipelas vaccine. Due to the lack of diversity of E. rhusiopathiae strains in terms of the occurrence of potential virulence genes, their detection does not provide significant information on the virulence of these bacteria. Assessment of the participation of individual genes in the pathogenesis of erysipelas requires more in-depth research based on gene expression analysis or the use of knockout strains.
Our research shows that in the treatment of swine erysipelas in Poland, apart from the commonly used beta-lactam antibiotics, macrolides, and phenicols, can also be effective. Aminoglycoside antibiotics, folic acid inhibitors, tetracyclines, and fluoroquinolones should not be considered. The high prevalence of resistance to enrofloxacin and tetracycline indicates the need to limit the use of these antibiotics in pig farming in Poland.