Assessment of the Antibiotic Resistance Profile, Genetic Heterogeneity and Biofilm Production of Methicillin-Resistant Staphylococcus aureus (MRSA) Isolated from The Italian Swine Production Chain

The main aim of the present study was to evaluate the level of antibiotic resistance, prevalence and virulence features of methicillin-resistant Staphylococcus aureus (MRSA) isolated from heavy swine at abattoir level and farming environments in Lombardy (Northern Italy). With this scope, 88 different heavy swine farms were surveyed, obtaining a total of n = 440 animal swabs and n = 150 environmental swabs. A total of n = 87 MRSA isolates were obtained, with an overall MRSA incidence of 17.50% (n = 77) among animal samples and a 6.67% (n = 10) among environmental. Molecular characterisation using multilocus sequence typing (MLST) plus spa-typing showed that sequence type ST398/t899 and ST398/t011 were the most commonly isolated genotypes, although other relevant sequence types such as ST1 or ST97 were also found. A lack of susceptibility to penicillins, tetracycline and ceftiofur was detected in >91.95, 85.05 and 48.28% of the isolates, respectively. Resistance to doxycycline (32.18%), enrofloxacin (27.59%) and gentamicin (25.29%) was also observed. Additionally, a remarkable level of antibiotic multiresistance (AMR) was observed representing a 77.01% (n = 67) of the obtained isolates. Genetic analysis revealed that 97.70% and 77.01% of the isolates harboured at least one antibiotic resistance or enterotoxin gene, respectively, pointing out a high isolate virulence potential. Lastly, 55.17% (n = 48) were able to produce measurable amounts of biofilm after 24 h. In spite of the current programmes for antibiotic reduction in intensively farming, a still on-going high level of AMR and virulence potential in MRSA was demonstrated, making this pathogen a serious risk in swine production chain, highlighting once more the need to develop efficient, pathogen-specific control strategies.


Sampling Design
A survey was carried out in a total of 88 heavy pig (Sus scrofa domesticus) farms (industrial livestock farming) located in Lombardy (Northern Italy) between 2016 and 2018. Overall, these farms represent about 20% of the total Italian heavy swine production and slaughtered animals (19.93% in 2018 and 19.75% according to latest Eurostat data [29]). From each farm, 5 different animals were randomly selected and animal samples were collected at abattoir level after stunning by rubbing a neck area of 50 × 50 cm using a sterile cotton swab, 1 per animal, thus obtaining a total number of n = 440 samples.
In addition to this, in 50 pig farms, different from those previously sampled, environmental samples were collected on a voluntary basis. Specifically, sterile cotton swabs were used to pick up samples from 3 different locations, i.e., barn, aeration device and feeder, obtaining a total number of samples of n = 150.
In all cases, swabs were then stored at 4 • C and processed within 6 h after sampling.

MRSA Confirmation
Confirmation of suspected colonies was performed via detection of nuc and mecA genes using a duplex colony PCR following the protocol reported by Pichon et al. [31]. Amplicons were resolved in a 1.5% agarose gel (Agarose Multi-Purpose, Roche) stained with GelRed (Biotium, Hayward, CA, USA) using 1 X Tris-Acetate-EDTA (TAE) buffer (Merck KGaA, Darmstadt, Germany). GeneRuler 100 bp Plus DNA Ladder (ThermoFisher Scientific, Vilnius, Lithuania) was used as a molecular marker and finally visualisation was carried out under UV light.
In the case of positive PCR confirmation, working cultures were prepared by mixing a loopful from TSA plates into a 2 mL sterile cryogenic vial containing 1 mL of sterile 50% glycerol. Cultures were stored at −20 • C for further analysis.

Disk Diffusion Assays
Isolates identified as MRSA were tested for susceptibility to a panel of 14 antimicrobial agents using the disk diffusion method following the Clinical & Laboratory Standards Institute indications [32].

Enterotoxin Gene Determination
The presence of genes related with staphylococcal enterotoxins A (sea), B (seb), C (sec), D (sed) and E (see) [42] was determined using the primer sets listed in Table S1. A multiplex PCR protocol was followed as described by Bianchi et al. [43]. In all cases, PCR reaction products were resolved as described in Section 2.3.

MRSA Molecular Subtyping and Phylogenetic Analysis
Samples identified as MRSA were subtyped with multilocus sequence typing (MLST) as described by Enright et al. [44]. To determine the sequence types (STs), data available on the Staphylococcus aureus MLST database [45] were used. For the spa typing, the spa gene of MRSA strains was amplified by PCR as described by Shopsin et al. [46], and spa subtypes were determined with Ridom StaphType software (Ridom GmbH Würzburg, Münster, Germany). A 3500xL genetic analyser (Applied Biosystems, Foster City, CA, USA) was used to obtain all the DNA sequences. Following this, phylogenetic analysis was conducted following the protocol described previously by Filipello et al. [19].

Quantification of Biofilm Production
The ability of MRSA isolates to produce biofilms on flat-bottomed polystyrene microplates was determined according to the protocol described by Stepanović et al. [47] with slight modifications. In all cases, all experiments were repeated in triplicate.
First, reactivation of the cells was carried out by transferring 100 µL of working cultures into 5 mL of TSB and incubated overnight at 37 • C. Next, cultures were adjusted to Abs 700 = 0.1 ± 0.001 in sterile saline using UV visible spectrophotometer (Varian SII Scan Cary 100 Spectrophotometer-Agilent Technologies, Santa Clara, CA, USA), corresponding to a concentration about 10 8 CFU/mL, according to previous calibrations. Adjusted cultures were 1:100 diluted in sterile fresh TSB + 1% glucose (Merck, Darmsatdt, Germany) reaching a final concentration of about 10 6 CFU/mL. In all cases, inocula cellular density was checked by plating onto TSA.
Following this, 200 µL of inoculum was transferred to seven wells of a sterile 96-well flat-bottomed polystyrene plate (ThermoFisher Scientific). In all cases 200 µL of sterile TSB + 1% glucose was used as negative control. Then plates were covered and incubated for 24 h at 37 • C in static conditions. After incubation, bulk culture was pipetted out, and each well was carefully washed three times with 300 µL of sterile phosphate-buffered saline (PBS, Oxoid) in order to remove loosely attached cells. Samples were then fixed at 60 • C for 1 h and stained with 150 µL of 2% crystal violet solution (CV; Condalab, Madrid, Spain) for 15 min. Excess staining was removed by washing wells with running tap water and allowed to air dry. Finally, 150 µL of 95% ethanol was added to each well to de-stain the wells and let to dwell for 30 min at room temperature.
Quantitative analysis was performed by measuring the absorbance at 540 nm using a microplate reader (Microscan FC, Thermo Scientific, Boston, MA, USA). All results were expressed by calculating the biofilm production index (BPI) as follows: OD (OD mean sample-OD negative control), whereas OD for the negative control was measured as OD nc : OD negative control + (3 × standard deviation of the negative control). Strains showing ability to produce biofilm were classified as weak (OD nc < OD ≤ 2 × OD nc ), moderate (2 × OD n c < OD ≤ 4 × OD nc ) or strong (OD > 4 × OD nc ) biofilm producers [47].

Statistical Analysis
A two-sided chi-square (X 2 ) test using IBM SPSS Statistics for Windows, Version 23.0 (IBM Corp. Armonk, NY, USA) was conducted to determine the correlation between the biofilm formation ability and the antibiotic multiresistance of the MRSA isolates. A p-value of <0.05 was deemed to be significant.

MRSA Detection
A total of 440 animal samples (i.e., swabs) obtained from 88 different fattening units at abattoir level were screened in order to determine the presence of MRSA by means of enrichment culture and isolation using agar plating with PCR confirmation of mecA gene. Overall, 42 (47.72%) resulted positive for MRSA. In samples from animals after stunning, positivity was detected in 17.50% (n = 77) of the analysed samples Regarding environmental samples (n = 150) from 50 different fattening units, positivity was found in 14% (n = 7) of the units surveyed. Molecular confirmation via mecA detection gave MRSA positivity in 6.67% (n = 10) of the environmental samples collected. Consequently, the overall incidence of MRSA-positive samples was 14.75% (n = 87).

Phenotypic Resistance
High resistance in β-lactams, especially penicillins (>91% of the isolates), 3rd generation cephalosporin (i.e., ceftiofur, 48.28%), and tetracycline (85.05%), was observed in MRSA isolates ( Figure 1). With regard to the rest of the antibiotics tested, percentage of resistance among MRSA isolates ranged between 36.78% (n = 32) in tiamulin (pleuromutilin) and 10.34% (n = 9) in the case of trimethoprim/sulfamethoxazole ( Figure 1). Of note, almost all MRSA isolates (85 out of 87) tested displayed resistance to at least three different antibiotics. The only two exceptions were isolate SA14 exhibiting only resistance to oxacillin and isolate SA4 with no phenotypic antibiotic resistance detected (Table S2).
Antibiotic multiresistance (AMR), i.e., resistance to three or more classes of antibiotics [33], was observed in 77.01% (n = 67) of the MRSA isolates obtained. Taking a closer look, some of the antibiotic patterns were repeated among the isolates. Specifically, a 62.69% (n = 42) of the isolates presented an antibiotic resistance profile that was present in at least two of the isolates (Table 1). It is important to highlight that these observed patterns were not exclusively associated to a given genotype (i.e., ST/spa-type) but a genetic heterogeneity was observed (Table 1). Of note, among those that belonged to the same genotype, differences either in the antimicrobial genes, in the enterotoxin genes or both were detected, indicating that even though closely related, the isolates were different to one another (results not shown).   Table 2, Table S2).

Minimum Inhibitory Concentration (MIC) Assays
Determination of MIC values in AMR isolates to antibiotics, considered as highly important or critical for antimicrobial chemotherapy in human (CN), veterinary (DXT, TET) or both (AMC, ENR, SxT, FFC), was determined in multiresistant isolates by means of miniaturised systems (MIC strips). Results showed the highest MIC values in TET (MIC TET ≥ 6 mg/L), followed by DXT (MIC DXT ≥ 8 mg/L), SxT (MIC SxT ≥ 4 mg/L) and CN (MIC CN > 1.5 mg/L), which are all above the available EUCAST resistance breakpoint values [36]. The range of MIC values for those antibiotics in which EUCAST does not provide a specific breakpoint value were as follows: MIC AMC = 0.19-24 mg/L; MIC ENR = 0.2->32 mg/L; MIC FFC = 3->256 mg/L).

Molecular Subtyping of MRSA Isolates
MRSA isolates' subtypes were ascertained by means of multilocus sequence typing (MLST) and spa-typing the isolates whose combination gave a total of 14 different MRSA genotypes (Table 3). On
The full eBURST algorithm was used to describe the different STs based on their genetic proximity. As depicted in Figure 2, MRSA isolates were clustered into six different STs and, in particular, ST398 was the most variable, grouping eight different spa-types.

Discussion
According to the latest report of The European Food Safety Authority (EFSA), MRSA prevalence in pig herds among European member states varied between 0.4 and 90.4% in 2017 with a slight reduction in 2018 with prevalence values ranging from 0 to 89.2% [49]. In the present work, 88

Biofilm Production Assays
In order to obtain a better picture of the phenotypic characteristics of the acquired isolates, biofilm production in polystyrene plates was determined through crystal violet staining [48]. Based on the outcomes obtained, 34.48% (n = 30) of the isolates were classified as non-biofilm producers, whereas 65.52% (n = 57) of the isolates were able to produce quantifiable amounts of biofilm at 24 h. Specifically, 44.83% (n = 39), were weak producers, 13.79% (n = 12), were moderate producers and 6.90% (n = 6), and were classified as strong producers ( Table 2). The latter group included three environmental isolates (SA2, SA15 and SA18) and three animal isolates (SA7, SA8 and SA10) ( Table 2).
Statistical analysis showed no correlation (p > 0.05) between the biofilm forming capacity (non-producer, weak, moderate and strong producer) and the antimicrobial multiresistance among the confirmed MRSA isolates.

Discussion
According to the latest report of The European Food Safety Authority (EFSA), MRSA prevalence in pig herds among European member states varied between 0.4 and 90.4% in 2017 with a slight reduction in 2018 with prevalence values ranging from 0 to 89.2% [49]. In the present work, 88 different swine farms at abattoir level were sampled obtaining a total of 440 samples. A total of 47.72% of the fattening units (n = 42) showed MRSA-positive results. It is important to note that this study was exclusively focused on heavy swine reared in Lombardy, the Italian region, with the main national pig production [50].
Similar to the outcomes obtained in the present work, results of positivity among pig farms in Italy have been also observed by Battisti et al. [22], in which a 38.1% of the sampled pig farms were positive for MRSA. Contrarily, recent studies performed in Italy showed a slightly higher incidence of MRSA-positive farms compared to the results obtained here. As a matter of illustration, the study of Parisi et al. [18] demonstrated that 64.7% of the pig farms turned out to be positive for MRSA. Similarly, Pirolo et al. [51] observed a positivity of MRSA in 66.8% of the swine farms sampled. These variations in results can be of a diverse nature, such as the use of different sampling, identification and characterisation methods that add an extra difficulty for prevalence data comparison [49,52].
Over the last years, the increasing resistance to antimicrobials among MRSA present in swine has been considered an issue of serious concern [21,53]. Antibiotic misuse/overuse in animal fattening units has posed an environmental pressure leading to the selection of a subpopulation of MRSA resistant not only to methicillin, but presenting cross-resistance to several other molecules [2][3][4]7]. This was observed by Alba et al. [54] in MRSA isolated from cattle farms in different European countries, demonstrating not only a high level of AMR, but the carriage of several virulence genes, conferring the fully invasive potential of MRSA to humans. In swine production chain, Lopes et al. [55] observed an unusual high level of resistance to clindamycin in the ST398 isolate, establishing a direct link between the use of antibiotics in a given farming unit and the level of resistance of the MRSA population present. Moreover, recently published data report a resistance to veterinary relevant antibiotics such as tiamulin of up to 50% among MRSA isolated from fattening pigs, and up to 100% among those isolated from pig meat [49]. This fact is in contrast with the obtained results, in which a slightly lower tiamulin level of resistance was observed ( Figure 2).
As displayed by the results, a remarkable level of resistance to TET ( Figure 2) and the presence of TET resistance-associated genes (Table 3), reported to be virtually present in all LA-MRSA [56,57], was demonstrated. This further corroborated data previously reported by several authors regarding TET resistance among LA-MRSA from pigs [18,22,23,51,57]. The main reason behind such TET-resistance levels is the intensive use of this antibiotic in farming [58][59][60] that may favour a cohort selection of TET-resistant bacteria; this is significantly notable in ST398 [21,53]. Taking a closer look, most of the obtained ST398 (CC398) isolates present TET resistance, which is in agreement with the latest EFSA data that claim that TET resistance is commonly associated with CC398 variants [49]. Nevertheless, among MRSA populations, a dynamic pattern is expected to be found due to natural evolution of this microorganism together with continuous exposure to antimicrobials and other environmental insults, thus modelling the overall phenotypic resistance characteristics of LA-MRSA in Italian abattoirs.
Multidrug resistance (MDR) was detected in 76.62% (n = 59) of the analysed MRSA (Table S2). The most observed resistance pattern was non-susceptibility to penicillin, amoxicillin/clavulanic acid, oxacillin, ampicillin, and ceftiofur. These observed high levels of resistance to penicillins and cephalosporins were in consonance with previously published data pointing out β-lactams resistance of MRSA in food systems [18,61]. It is important to remark that in this study two isolates, namely, SA63 (ST398/t899) and SA78 (ST1/t127), were found to display no susceptibility to 12 different molecules, respectively, representing 6 different classes. This is in line with previous results published in which the same MRSA genotypes isolated from the pork supply chain also displayed high levels of MDR [18]. This level of MDR encountered is of a special relevance when also considering the amount of resistance genes present among isolates and the fact that some of them are encoded in mobile elements that could further enhance horizontal and vertical resistance gene transfer in swine-related premises [56].
According to the EFSA report on antimicrobial monitoring on MRSA, the main route of transmission of LA-MRSA into the population is direct contact with animals, e.g., at abattoir level, and then dissemination [61]. In line with this, the study carried out by Pirolo et al. [62] involving pig farm workers demonstrated that 21.6% of the workers were colonised by livestock-associated MRSA (LA-MRSA) and that this colonisation was significantly higher if direct animal-human contact occurred. Additionally, Cuny et al. [63] detected a 30% antibiotic multiresistance to penicillins, tetracyclines and macrolides in ST398 MRSA isolated from humans in contact with farm animals, a fact that poses a therapeutic challenge if infection occurs. However, contamination routes within pig production environments still remain obscure considering that clones classified as community-acquired MRSA (CA-MRSA), such as ST1/t127 and ST30/t318, were also found among the MRSA isolated (Table 3), which is in agreement with previously reported data [64]. Regarding ST30, is important to remark that despite being commonly categorised as methicillin-sensitive S. aureus (MSSA) [65], MRSA variants have been isolated from pigs in Australia [66] and Portugal [67] that indicate not only a worldwide dissemination of this specific ST among pig herds, but tangible evidence of the dissemination of methicillin resistance via integration of the staphylococcal cassette chromosome mec (SCCmec) among sensitive strains [68]. Additionally, ST30 showed a remarkably high level of MDR (Table S2), harboured several antibiotic resistance and enterotoxin genes, and was identified as a strong biofilm producer (Table 2), which is highly relevant in the food production chain due to its potential harmfulness and environmental recalcitrance.
In European animal production systems, most efforts to track down MRSA routes of transmission have been mainly posed in regard to the swine production chain since it has been demonstrated that pigs are the major reservoir of several livestock-associated MRSA (LA-MRSA) genotypes such as ST97 [17,18,23], but especially ST398 [21,69]. Results obtained in this survey demonstrated that most of the MRSA isolates obtained belonged to the ST398, which is in agreement with the EFSA data and previous studies performed in Italy that highlight that this ST is the most prevalent LA-MRSA among pig herds [18,21,23,49]. Additionally, molecular characterisation detected eight different spa-types within ST398, including t011 and t899 with the latter being the most predominant, which is in line with previously reported data [18,22,53,69,70]. Of note, ST398/t899 have been reported to be among the major genotypes isolated in fattening pigs at abattoir level in Finland and Spain [49], indicative of a wide distribution regardless of the European zone. Additionally, in line with our results, recent data show that a high prevalence of ST398/t011 in Europe has been reported due to an increase in recent years among Swiss fattening pig herds [49]. Furthermore, a recent study conducted by Larsen et al. [71] demonstrated a high prevalence of ST398 in humans regardless of their direct contact with livestock animals, thus indicating that humans themselves could also act as further reservoirs for this specific lineage. It is important to highlight that ST398 was also present in 9 out of 10 of the environmental samples (Table 3). ST97 was also found among the analysed samples (Table 2). In Italy, ST97 (t1730 and t4795) is considered to be the most prevalent swine-associated lineage after ST398 [22,72]. This LA-MRSA is commonly associated with infection in cattle, although spreading among pig herds in intensive farming have been also described [72]. It has been hypothesised that in pig farming, ST97 can acquire high levels of antibiotic resistance, as observed in our results ( Table 1, Table S2), due to the selective pressure, and they can then be transmitted to humans, causing severe infections [72]. This capability of interspecies spreading together with the antimicrobial resistance makes ST97 an issue of serious concern for public health.
Genotype ST1/t127 was also detected ( Figure 2, Table 3), which is in line with previously published data [21,22,53]. Although firstly categorised as community acquired (CA) MRSA [49,73] responsible for various infections in human [63,74,75], it has been also isolated from livestock animals [18,22,23]. With this regard, Franco et al. [76] carried out a phylogenetic characterisation of human and pig MRSA ST1/t127 isolates from Italy, Spain and Demark. The results obtained demonstrated that despite belonging to the same genotype, the homology between the two groups differed by 25%. Moreover, it was demonstrated that genotype ST1/t127 displayed a high resistance to major antimicrobials, which is in accordance with the results obtained in the present study (Table 3).
Staphylococcal enterotoxins (SEs) are considered one of the major virulence factors in S. aureus in the context of food safety [77]. This highly heat-resistant family of proteins can cause severe food poisoning outbreaks with symptoms including vomiting, abdominal cramping, and even fatality due to hypersensitivity induction [78], and their synthesis is dependent on the staphylococcal cellular density [79]. The latter aspect is considered within the EC Regulation 2073/2005 of microbiological safety of foodstuffs as a safety criterion in some food products of animal origin [80]. In this study, the enterotoxigenic potential of MRSA was evaluated, demonstrating that 77.01% of the isolates possessed at least one gene relate to SE synthesis (Table 3). This is in contrast with a previous study carried out by Normanno et al. [23], where no MRSA isolated from swine carried any SE gene. Similarly, Kluytmans [10] reported that most MRSA ST398 seldom carry SE genes, which clearly contrasts with the outcomes in this work since most of the isolates obtained carried at least one ( Table 2). It is important to remark that among the isolated MRSA, a small proportion lacked on genes while presenting an MDR phenotype, and one of them, SA8, showed a high capacity of biofilm formation (Table 2). This aspect is important to take into account and highlights once more that the hazardous potential of MRSA strains found in swine production chain must be assessed from different angles, as a non-SE producer can harbour several resistance genes and become a persistent strain forming highly stable biofilms.
In scientific literature, it is widely accepted that S. aureus is able to embed itself in a polysaccharide matrix and form biofilms that adhere onto a wide variety of food-related surfaces, contributing to its survival to harsh environmental conditions (i.e., presence of antibiotics) alone or in multispecies communities [24,81]. The mechanisms beneath such resistance are diverse and lack antimicrobial diffusion into the biofilm matrix and the presence of recalcitrant cells, such as those belonging to the viable but not cultivable phenotype [82]. In S. aureus, the ability to form biofilms is directly related to the presence of virulence factors such as adhesins [83], and this structure contributes to the dissemination of resistance genes in the cells therein [5]. In this study, 65.52% of the resulting MRSA isolates were classified as biofilm-producing at 24 h. However, different levels of biofilm production were detected, in which most of the producers displayed low levels of biofilm production (i.e., weak producers, 44.83%) (see Section 3.4. for details). Similar results were previously reported by Di Ciccio et al. [84] in S. aureus strains from food and food-related environments. Of note, the quantification and subsequent classification of the biofilm capacity of the MRSA is only valid in the experimental conditions tested since, as demonstrated by Zhang et al. [83], biomass production, and subsequent classification, is directly dependent on the age of the biofilm. Based on our results, there was no correlation between the MDR phenotype and the capacity of a given MRSA of producing a biofilm. With this regard, a recent study carried out by Ou et al. [85] shows that there is a certain relationship between the level of antibiotic resistance and the biofilm forming ability in MRSA isolated from food from animals, but that the biofilm capacity of them is somehow impaired in those displaying the highest level of antibiotic resistance. Since MRSA biofilms can become recalcitrant contamination foci and a reservoir of several resistance genes, information about their adhesion capability is especially relevant to assess the potential hazard of a given subpopulation of MRSA, especially in those related with the food production chain.

Conclusions
Methicillin-resistant Staphylococcus aureus (MRSA) is among the major pathogens in the pig supply chain, and it is detected in both live and slaughtered animals as well as in meat products that are currently considered a potential MRSA transmission vector into the population [86,87]. In the present work, a high diversity of MRSA genotypes present in Italian pork supply chain was demonstrated. In addition to this, antimicrobial susceptibility evaluation highlighted high levels β-lactams and tetracyclines resistances among the isolated bacteria that may indicate an environmental selection of resistant MRSA variants due to the use of antibiotics in intensive animal farming. Additionally, we demonstrated capability of these MRSA to form biofilms, which gives these bacteria higher capability to survive in food-related premises. This could be translated into eventual recalcitrance and subsequent spreading of some genetic subtypes and antimicrobial resistance genes in the swine supply chain.
Lowering the level of antibiotic usage in order to improve animal and human welfare and thus avoid the selection process due to environmental pressure and dissemination of antibiotic resistances is an issue of extreme importance. Due to the high prevalence of some MRSA clones, such as ST398, it is of paramount importance to intensify specifically control measures regarding MRSA colonisation and transmission in farms and food-processing environments in order to avoid MRSA direct animal-human or foodborne spreading, through improvement of biosecurity practises, i.e., personal protection equipment, process environment and cleaning and disinfection procedures.