Microbial contamination, antimicrobial resistance and biofilm formation of bacteria isolated from a high-throughput pig abattoir.

The aim of this work was to assess the level of microbial contamination and resistance of bacteria isolated from a highthroughput heavy pig slaughterhouse (approx. 4600 pigs/day) towards antimicrobials considered as critical for human, veterinary or both chemotherapies. Samples, pre-operative and operative, were obtained in 4 different surveys. These comprised environmental sampling, i.e. air (ntotal = 192) and surfaces (ntotal = 32), in four different locations. Moreover, a total of 40 carcasses were sampled in two different moments of slaughtering following Reg. (CE) 2073/2005. Overall, 60 different colonies were randomly selected from VRBGA plates belonging to 20 species, 15 genera and 10 families being Enterobacteriaceae, Moraxellaceae and Pseudomonadaceae the most represented ones. Thirty-seven isolates presented resistance to at least one molecule and seventeen were classified as multi-drug resistant. Enterobacteriaceae, particularly E. coli, displayed high MIC values towards trimethoprim, ampicillin, tetracycline and sulphametoxazole with MICmax of 16, 32, 32 and 512 mg/L, respectively. Moreover, isolated Pseudomonas spp. showed high MIC values in critical antibiotics such as ampicillin and azithromycin with MICmax of 32 and 64 mg/L, respectively. Additionally, in vitro biofilm formation assays demonstrated that fifteen of these isolates can be classified as strong biofilm formers. Results demonstrated that a high diversity of bacteria containing antibiotic resistant and multiresistant species is present in the sampled abattoir. Considering these findings, it could be hypothesised that the processing environment could be a potential diffusion determinant of antibiotic resistant bacteria through the food chain and operators. ©Copyright: the Author(s).

Introduction

In northern Italy, the breeding of heavy pigs (slaughtered at an approximate live weight of 170 kg) represents nearly the 90% of the national production with nearly 1 million slaughtered animals per month (I.Stat, 2021). At abattoir level, carcasses can be contaminated by bacteria in any of the different stages of processing (Pearce et al., 2006). In order to control such hazards, it is compulsory that Food Business Operators (FBOs) implement Good Manufacturing and Hygienic Practices (GMP/GHP), proper Sanitation Standard Operating Procedures (SSOP) and Hazard Analysis and Critical Control Point (HACCP) based procedures along the entire slaughtering chain (Blagojevic et al., 2021) so as to meet the hygienic criteria established by the Reg. (CE) 2073/2005 (European Commission, 2005).

The microbiota present in the abattoir environment is considered a major risk for carcass contamination mainly due to transfer of bacteria from food and non-food contact surfaces (Di Ciccio et al., 2016) and from bioaerosols (Cosenza-Sutton, 2004). The first one is referred to the inherent ability of bacteria to form multispecies biofilms that can persist through time (Rodriguez- Lopez et al., 2020). Furthermore, these structures can facilitate horizontal transfer of antibiotic resistance genes (ARGs) raising concerns regarding the spreading of antimicrobial-resistant bacteria (ARB) through the food chain (EFSA, 2021).

On the other hand, it is nowadays accepted that air plays a critical role as a reservoir for ARB in pig abattoirs (EFSA, 2021; Masotti et al., 2019). For instance, high levels of resistance to tetracycline, trimethoprim-sulfamethoxazole and ampicillin in two hundred forty-three aerobic bacteria isolated from a pig slaughterhouse were found (Li et al., 2016). Similarly, a high occurrence of colistin-resistant Escherichia coli and Klebsiella pneumoniae isolated from process water and wastewater taken in pig slaughterhouses was reported (Savin et al., 2020). Additionally, recently published data report that Gram-negative bacteria play a central role in the transmission of the genes responsible for the onset of antimicrobial resistance (EFSA, 2021). Regarding this, the study carried out by Cosenza-Sutton, (2004) has shown that bacteria such as Pseudomonas spp. or Enterobacteriaceae are part of the normal air bioburden in both wet and clean zones of pig abattoirs (Cosenza-Sutton, 2004).

Therefore, the main aim of the present study was to investigate the level of contamination and the antimicrobial resistance and biofilm formation capability of Gramnegative bacteria present in a high-throughput heavy pig abattoir located in Emilia- Romagna (northern Italy).

Materials and Methods

Rationale of the study

This study was performed in a pig slaughterhouse located in Emilia-Romagna, one of the regions included in the so-called Italian Food Valley together with Lombardy, Piedmont and Veneto. This abattoir was characterised by a daily slaughtering capacity of about 480 fatteners per hour destined for Protected Designation of Origin (PDO) production. The premise had a linear design and was divided into two zones for sampling purposes: a ‘wet’ room (slaughter) and a ‘clean’ room (dressing). Sampling was performed weekly over a month.

Sampling

Air

Air samples were collected from four different locations: two in the wet room (WR), next to the gambrelling table (A1) and the polisher (A2), and two in the clean room (CR), near evisceration (A3) and precutting (A4) areas. At each survey, samples were collected one hour before (pre-operative) and five hours after the commencement of the slaughtering process (operative) using a portable SAS Super ISO-100 sampler (PBI International, Milan, Italy) placed at 1.5 m distance from the floor and 1.5 m distance from the processing line. Sampler’s head was autoclaved at 121°C for 15 min before and after each sampling day and decontaminated with 70% ethanol between samples. Sampled volume in each location was determined in a preliminary survey (data not shown).

Samples were collected in triplicate on Plate Count Agar (PCA, Oxoid, Hampshire, UK) for enumeration of the total aerobic count (TVC; ntotal = 96 plates) and on Violet Red Bile Glucose Agar (VRBGA, Oxoid) for the enumeration of Gram-negative bacteria (GNB; ntotal = 96 plates).

Surfaces

Four different surfaces before (preoperative) and after the commencement (operative) of the plant activity were sampled. Specifically, from the WR the gambrelling table (S1) and the polisher (S2), and from the CR the saw (S3) and the Fat- O-Meter (FOM, S4) were chosen (ntotal = 32). In each location, an area of 100 cm2 was sampled by thoroughly rubbing the surface using a sterile sponge (Whirl-PakR, Madison, WI) moistened in 10 mL of Buffered Peptone Water (BPW, Oxoid) and kept refrigerated until analysis.

Carcasses

en carcasses were sampled in each sampling day as described in Reg. (CE) 2073/2005 (European Commission, 2005). In each sampling day, carcasses were sampled in two different rounds, at the beginning of the slaughtering activities, first batch of the day, and five hours after the initiation of the plant activity (ntotal = 40).

Bacterial enumeration

Air samples were directly incubated. Sponges were mixed with 90 mL of BPW and stomached for 2 min. Resulting suspensions were serially diluted in phosphate buffered saline (PBS, Oxoid) and a 1 mL aliquot was spread onto PCA and VRBGA plates. In all cases, PCA plates were incubated at 30°C for 48 h and VRBGA plates at 37°C for 24 h for enumeration of TVC and GNB, respectively.

For air sampling, results were expressed in log CFU/m3 representing the mean value of three plates whereas for surfaces and carcasses outcomes were expressed in log CFU/cm2.

Isolation and identification of GNB

Bacterial colonies were randomly selected from VRBGA plates, stroked on Trypticase Soy Agar (TSA, Oxoid) and incubated at 37°C for 24 h. To ensure purity, the latter was repeated twice.

Next, colonies were subjected to oxidase test using Oxidase strips (Oxoid). Oxidase negatives were identified using the MicrogenR GN-ID A/B system (Microgen bioproducts, Camberley, UK) whereas for oxidase positive isolates the API 20 NE system (bioMerieux, Marcy-l’etoile, France) was preferred.

Stock cultures of each isolate were kept at −80 °C in Brain-Heart Infusion (BHI, Oxoid) containing 50% glycerol 1:1 (v/v) mixed. Working cultures were maintained at −20 °C in Trypticase Soy Broth (TSB, Oxoid) containing 50% glycerol 1:1 (v/v) mixed.

Minimum inhibitory concentration (MIC) assays

The antimicrobial susceptibility of the GNB isolates was assessed via determination of MIC using the SensititreTM – VIZIONTM system (ThermoFisher Scientific, Waltham, MA, USA) using the EUVSEC3 panel, containing antibiotics considered as critical for antimicrobial chemotherapy in veterinary (Cloramphenicol, C; Sulphametoxazole, SMX; Tetracycline, TET; Trimethoprim, TRI), human (Nalidixic Acid, NA; Colistin, COL; Meropenem, MER) or both (Amikazin, AMI; Ampicillin, AMP; Azitromycin, AZI; Ceftazidime, CAZ; Ciprofloxacin, CIP, Cefoxitin, CTX; Gentamicin, GEN; Tigecycline, TIG) following manufacturer’s instructions.

Biofilm assays

The production of biofilm was assessed by the method described by Stepanovic et al. (2007). Strains were classified as non-producers or weak, moderate, strong formers.

Statistical analyses

Statistical tests were performed using OriginPro 2021 software for Windows, v. 9.8.0.200 (OriginLab, Northampton, MA, USA) expressing significance at the 95% confidence level (α=0.05) or greater.

Specifically, to statistically compare TVC and GNB values between locations a one-way ANOVA with Tukey’s post hoc test was chosen. Contrarily, a two-tailed Student’s t-test was preferred to determine significance between pre-operative and operative samples.

Lastly, a two-sided chi-square (Χ2) test was used to determine the correlation between the biofilm formation and antimicrobial multiresistance of GNB isolated.

Results

Microbial counts

In pre-operative air samples, total viable count (TVC) presented mean values ranging from 2.19±0.11 and 3.67±0.27 log CFU/m3 in locations A4 and A1, respectively (Table 1). One-way ANOVA analysis showed significance (P<0.05) in all sampling sites but those belonging to the CR, i.e. locations A3 and A4. Conversely, operative samples TVC values ranged from 2.91±0.13 and 4.33±0.08 log CFU/m3 belonging to locations A4 and A1, respectively. No significance among operative samples (P>0.05) in TVC mean values was observed between the two locations of the WR (i.e. A1 and A2) whereas TVC values of the rest of sampling locations were significantly different (P<0.05; Table 1).

Table 1.

Mean values (n = 3 agar plates, each) expressed in log CFU/m3 for total viable count (TVC) and Gram-negative bacteria (GNB) obtained in air active sampling at the selected locations at a pig abattoir. In each column, values with a different superscript indicate statistically significant differences between mean values (One-way ANOVA; α= 0.05). Additionally, a two-tailed Student’s t test (α= 0.05) was performed to assess the significance between PreOperative and Operative samples.

Zone	Location	Operation	TVC (log CFU/m3)			GNB (log CFU/m3)
			PreOperative	Operative	Dif Pre/OP	PreOperative	Operative	Dif Pre/OP
Wet room	A1	Gambrelling	3.67 ± 0.27A	4.33 ± 0.08A	Yes	1.42 ± 0.55A	1.93 ± 0.18A	No
	A2	Polishing	3.15 ± 0.19B	4.26 ± 0.22A	Yes	1.06 ± 0.13A	1.36 ± 0.37A,B	No
Clean room	A3	Evisceration	2.38 ± 0.45C	3.17 ± 0.14B	Yes	1.10 ± 0.17A	1.16 ± 0.22B	No
	A4	Pre-cut	2.19 ± 0.11C	2.91 ± 0.13C	Yes	0.92 ± 0.37A	1.39 ± 0.25B	No

Dif Pre/OP: Significance between pre-operative and operative samples.

Of note, initiation of the normal activity of the slaughterhouse increased all TVC mean values of 0.65, 1.11, 0.79 and 0.71 log CFU/m3 in locations A1, A2, A3 and A4, respectively, which represented a significant increase (P<0.05; two-tailed t Student) if TVC values were compared individually in terms pre-operative and operative sampling momenta (Table 1).

Overall, Gram-negative bacteria (GNB) counts in air pre-operative samples were lower compared to TVC values (Table 1). Additionally, GNB outcomes in some of the preoperative samples, from both wet and clean areas, demonstrated that the amount of GNB were below the limit of detection (10 CFU/m3) considering the sampled volume (results not shown). Moreover, no significance (P>0.05) in GNB values were observed between any location among preoperative samples (Table 1).

Slaughtering activities increased GNB counts in 0.51, 0.3, 0.06 and 0.47 log CFU/m3 in locations A1 to A4, respectively. Taking a closer look, among operative samples GNB in location A1 (WR) was of 1.93±0.48 log CFU/m3 which was significantly different (P<0.05) to locations A3 and A4 (CR). Of note, location A2 (WR) presented no significance compared to the rest of the locations (Table 1).

Among pre-operative samples, mean TVC ranged from 1.42±0.60 to 5.05±0.62 log CFU/cm2 corresponding to locations S3 and S1, respectively (Table 2). Of note, in locations A2, A3 and A4, collected samples had TVC that were below the limit of detection (results not shown). In all cases, normal slaughter activity of the plant increased the TVC mean values, being especially relevant in samples coming from the cutting saw presenting a significant increment of 2.23 log CFU/cm2. For the rest of the locations the observed increase was of 1.06, 1.7 and 0.6 log CFU/cm2 corresponding to locations S1, S2 and S4, respectively (Table 2). Moreover, a sharp decrease in TVC between pre and operative samples of 2.53 and 1.99 log CFU/cm2 was observed in locations S2 and S3, respectively (Table 2).

Table 2.

Mean values in log CFU/cm2 for total viable count (TVC) and Gram-negative bacteria (GNB) obtained in surface sampling at the selected locations at a pig abattoir. In each column, values with a different superscript indicate statistically significant differences between mean values (One-way ANOVA; α= 0.05). Additionally, a two-tailed Student’s t test (α= 0.05) was performed to assess the significance between PreOperative and Operative samples.

Zone	Location	Operation	TVC (log CFU/cm2)			GNB (log CFU/cm2)
			PreOperative	Operative	Dif Pre/OP	PreOperative	Operative	Dif Pre/OP
Wet room	S1	Gambrelling	5.05 ± 0.62A	6.11 ± 0.23A	Yes	3.02 ± 0.78	4.01 ± 0.33A	Yes
	S2	Polishing	3.95 ± 0.24A	5.65 ± 0.59A	Yes	N.D.	3.45 ± 0.85A,B	N.A.
Clean room	S3	Saw	1.42 ± 0.60A	3.66 ± 0.66B	Yes	N.D.	2.02 ± 0.63B,C	N.A.
	S4	FOM	3.76 ± 0.68B	4.37 ± 0.49B	No	N.D.	2.74 ± 0.13C	N.A.

Dif Pre/OP: Significance between pre-operative and operative samples; N.D.: Not detected; N.A.: Not applicable.

GNB in preoperative samples were detected only among those coming from location S1. Mean value was significantly lower (P>0.05) compared to the GNB mean obtained in operative samples (4.01±0.33 log CFU/cm2). In all the other locations, samples collected during the slaughtering activities, presented detectable GNB counts ranging from 2.02 to 3.45 log CFU/cm2 (Table 2).

Mean TVC coming from pig carcasses did not present variations in mean values comparing both sampling times. Similarly, GNB outcomes neither presented statistically significant differences (two-tailed Student’s t test; P>0.05) comparing the GNB values obtained in the first and the second moment of sampling, respectively. Of note, considering both counts, it can be inferred that more than 50 % of the microbiota isolated from the carcasses, was mainly due to the presence of GNB (Table 3).

Table 3.

Mean values in log CFU/cm2 for total viable count (TVC) and Gram-negative bacteria (GNB) obtained from pig carcasses in both sampling times sampling using the method described in Reg. (CE) 2073/2005. Of note, there was no significance between the two different moments of sampling (two-tailed Student’s t test, α= 0.05).

TVC		GNB
Sampling 1	Sampling 2	Sampling 1	Sampling 2
3.81 ± 0.38	3.95 ± 0.29	2.45 ± 0.49	2.09 ± 0.42

Diversity in GNB

Overall, 20 different species belonging to 10 families coming from air (n=15 preoperative, n=15 operative), surfaces (n=6 pre-operative, n=16 operative) and carcasses (n=8) were identified (Table 4). Regarding the families, Enterobacteriaceae (n=22), especially E. coli (n=15) followed by Serratia liquefaciens (n=3), was the most represented one (Table 4). Major non- Enterobacteriaceae isolated families were Pseudomonadaceae (n=9) and Moraxellaceae, (n=9) with Pseudomonas and Moraxella as representatives, respectively. Additionally, minor families such as Aeromonoadaceae, Burkholdeliaceae and Brucellaceae, were also identified (Table 4).

Table 4.

List of families, species, isolation sources, antibiotic resistance pattern and biofilm production of the Gram-negative isolates obtained in this study. For a complete list of MIC values for each antibiotic molecule tested, the reader is kindly referred to Table S1.

Isolate ID	Family	Species	Source	Sampling time	Room	Resistance Pattern	Biofilm production
GN1	Enterobacteriaceae	Escherichia coli	Air	PreOperative	Wet	AMP-SMX-TET-TRI	Weak
GN2		Escherichia coli	Air	PreOperative	Wet	AMP-SMX-TRI	Non producer
GN3		Escherichia coli	Air	PreOperative	Clean	AMP-C-SMX-TET-TRI	Weak
GN4		Escherichia coli	Air	PreOperative	Clean	AMP-CAZ-COL-TRI	Weak
GN5		Escherichia coli	Air	PreOperative	Clean	n.d.	Weak
GN6		Escherichia coli	Air	Operative	Wet	n.d.	Strong
GN7		Escherichia coli	Air	Operative	Wet	CAZ-COL-TRI	Moderate
GN8		Escherichia coli	Air	Operative	Clean	AMP-SMX-TET-TRI	Moderate
GN9		Escherichia coli	Air	Operative	Clean	AMP-SMX-TRI	Weak
GN10		Escherichia coli	Surface	Operative	Wet	n.d.	Weak
GN11		Escherichia coli	Surface	Operative	Clean	TRI	Strong
GN12		Escherichia coli	Surface	Operative	Clean	n.d.	Moderate
GN13		Escherichia coli	Surface	Operative	Clean	AMP-C-SMX-TET-TRI	Weak
GN14		Escherichia coli	Carcass	n.a.	n.a.	AMP-NA-SMX-TRI	Moderate
GN15		Escherichia coli	Carcass	n.a.	n.a.	TET	Weak
GN16		Enterobacter gergoviae	Surface	Operative	Wet	AMP	Moderate
GN17		Pantoea agglomerans	Air	PreOperative	Clean	n.d.	Moderate
GN18		Pantoea agglomerans	Air	Operative	Wet	AMP-SMX-TET-TRI	Moderate
GN19		Serratia liquefaciens	Air	Operative	Clean	AMP-C-SMX-TET-TRI	Moderate
GN20		Serratia liquefaciens	Carcass	n.a.	n.a.	AMP-TET	Weak
GN21		Serratia liquefaciens	Carcass	n.a.	n.a.	AMP-C-GEN-SMX-TET-TRI	Weak
GN22		Klebsiella oxytoca	Surface	Operative	Wet	n.d.	Non producer
GN23	Morganellaceae	Morganella spp.	Surface	Operative	Wet	AZI-COL	Weak
GN24	Comamonaceae	Comamonas spp.	Air	Operative	Wet	n.d.	Weak
GN25	Moraxellaceae	Acinetobacter baumannii	Air	Operative	Wet	n.d.	Strong
GN26		Acinetobacter baumannii	Surface	PreOperative	Wet	AMP-AZI-C	Strong
GN27		Acinetobacter baumanniii	Surface	Operative	Wet	n.d.	Moderate
GN28		Acinetobacter haemolyticus	Air	PreOperative	Clean	n.d.	Weak
GN29		Acinetobacter lwoffii	Surface	Operative	Wet	COL	Moderate
GN30		Acinetobacter spp.	Air	PreOperative	Wet	n.d.	Weak
GN31		Acinetobacter spp.	Air	Operative	Wet	n.d.	Weak
GN32		Moraxella spp.	Air	PreOperative	Wet	n.d.	Non producer
GN33		Moraxella spp.	Air	PreOperative	Wet	n.d.	Weak
GN34		Moraxella spp.	Air	Operative	Clean	AMP-C	Strong
GN35		Moraxella spp.	Surface	Operative	Clean	n.d.	Strong
GN36		Moraxella spp.	Surface	Operative	Clean	n.d.	Strong
GN37		Moraxella spp.	Surface	Operative	Wet	AMP-CIP	Strong
GN38		Moraxella spp.	Carcass	n.a.	n.a.	n.d.	Strong
GN39	Brucellaceae	Ochrobactrum anthropi	Air	Operative	Clean	AMP	Weak
GN40		Ochrobactrum anthropi	Surface	PreOperative	Clean	n.d.	Moderate
GN41	Burkholdeliaceae	Ralstonia pickettii	Air	Operative	Wet	n.d.	Moderate
GN42		Ralstonia pickettii	Surface	PreOperative	Wet	n.d.	Weak
GN43		Ralstonia pickettii	Surface	PreOperative	Wet	n.d.	Weak
GN44	Alcaligenaceae	Alcaligenes spp.	Surface	Operative	Clean	n.d.	Strong
GN45	Pseudomonadaceae	Pseudomonas fluorescens	Air	PreOperative	Wet	AMP-AZI-CTX-C-COL-TIG	Strong
GN46		Pseudomonas fluorescens	Carcass	n.a.	n.a.	AMP-AZI-C-TET	Strong
GN47		Pseudomonas orzyhabitans	Air	PreOperative	Clean	AMP	Moderate
GN48		Pseudomonas putida	Air	Operative	Wet	AMP-C	Weak
GN49		Pseudomonas putida	Carcass	n.a.	n.a.	AMP-AZI-C	Strong
GN50		Pseudomonas stutzeri	Air	PreOperative	Wet	n.d.	Strong
GN51		Pseudomonas stutzeri	Air	Operative	Clean	AMP	Weak
GN52		Pseudomonas stutzeri	Surface	PreOperative	Wet	AMP-CIP	Strong
GN53		Pseudomonas stutzeri	Surface	Operative	Clean	AMP-CIP	Moderate
GN54	Xanthomonadaceae	Stenotrophomonas maltophilia	Carcass	n.a.	n.a.	AMP-CTX	Non producer
GN55	Aeromononadaceae	Aeromonas hydrophila	Air	PreOperative	Wet	AMP-CTX-C-TIG	Weak
GN56		Aeromonas hydrophila	Surface	Operative	Clean	C	Weak
GN57		Aeromonas sobria	Air	PreOperative	Clean	AMP-TET	Non producer
GN58		Aeromonas sobria	Air	Operative	Clean	AMP-TET	Moderate
GN59		Aeromonas sobria	Suface	PreOperative	Wet	AMP-TET-TIG	Weak
GN60		Aeromonas spp.	Surface	Operative	Wet	AMP	Weak

AMP: Ampicillin; AZI: Azitromycin; CTX: Cefotaxime; CAZ: Ceftazidime; C: Chloramphenicol; CIP: Ciprofloxacin; COL: Colistin; GENT: Gentamicin; NA: Nalidixic acid; SMX: Sulfamethoxazole; TET: Tetracycline; TIG: Tigecycline; TRI: Trimethroprim; n.a.: Not applicable; n.d.: Not detected.

Antimicrobial resistance

Out of the 60 isolates assayed, 37 (61.67%) were resistant to at least one antibiotic. As displayed in Table 4 and Supplementary Table S1, overall, 59.09% of the Enterobacteriaceae isolates presented resistance to AMP (n=13, MICmax = 32 mg/L), 59.09 to % TRI (n=13, MICmax = 16 mg/L), 45.05% to SMX (n=10, MICmax = 512 mg/L), 40.91% to TET (n=9, MICmax = 32 mg/L) and 22.72% to C (n=5, MICmax = 64 mg/L). Two E. coli isolates, namely GN4 and GN7, were the only that displayed resistance to. CAZ (MICmax = 8 mg/L) and to COL (MICmax = 16 mg/L). Of note, one of the identified S. liquefaciens, namely GN21, was the only Enterobacteriaceae member to display resistance to GEN (MIC = 4 mg/L). No resistance to AZI, CTX, CIP, MER and TIG was detected in any of the Enterobacteriaceae isolated.

Twelve of these isolates were classified as multidrug resistant (MDR), i.e. presenting resistance towards 3 or more antibiotic classes (Magiorakos et al., 2012). Among E. coli, most MDR were isolated from air samples either from preoperative (isolates GN1, GN2, GN3 and GN4) and operative (isolates GN7, GN8 and GN9) sampling times. This MDR phenotype was also observed in E. coli from surfaces (isolate GN13) and carcass (isolate GN14). Other assayed Enterobacteriaceae members displaying the MDR phenotype were S. liquefaciens from air (isolate GN19) and carcass (isolate GN21) samples, and Pantoea agglomerans from air samples (isolate GN18) (Table 4).

Among non-Enterobacteriaceae members identified, the level of antimicrobial resistance was generally lower compared to Enterobacteriaceae with a 44.74% (n=17) presenting no resistance to any of the antibiotics (Table 4). In addition to this, 47.37% of the isolates presented resistance to AMP (n=18; MICmax = 32 mg/L), followed by a 21.05% resistant to C (n=8; MICmax = 64 mg/L), a 13.06% resistant to AZI (n=5; MICmax = 64 mg/L), and a 10.53% resistant to TIG (MICmax = 4 mg/L). Of note, most of these resistances were displayed by members of the genus Pseudomonas (Table 4, Supplementary Table S1). MDR was determined in 5 of the non-Enterobacteriaceae isolates assayed. Thus, the species displaying this phenotype were Acinetobacter baumanii (isolate GN26), Pseuodomonas fluorescens (isolates GN45 and GN46), Pseudomonas putida (isolate GN49), and Aeromonas hydrophila (isolate GN55) (Table 4).

Overall, maximum level of resistance was displayed by isolates S. liquefaciens G21 (carcass) and P. fluorescens GN45 (air) showing lack of susceptibility upon 6 out of the 15 molecules tested, representing 6 and 5 antibiotic classes, respectively (Table 4).

Biofilm formation

Most isolates (91.66%; n=55) were able to adhere to produce quantifiable biomass levels after 24 h of incubation. More specifically, 41.67% (n=25) were weak producers, 25% (n=15) were moderate producers and 25% (n=15) and were classified as strong producers (Table 4).

In the latter category, it is important to remark that the isolates were obtained both from environmental sources (i.e. air and surfaces) and from swine carcasses. At a species level, outcomes demonstrated a wide variability in terms of strong biofilm formers, showing that this phenotype is not limited to one particular species (Table 4). Specifically, most of them belonged to Moraxella spp. (n=5). However, this strong biofilm production was not concomitant with resistance towards the antibiotics tested (Table 4).

Isolates namely GN26 (A. baumannii), GN45, GN46 (P. fluorescens) and GN49 (P. putida) were categorised as MDR (Table 4). Statistical analysis (Χ2 test) showed no correlation (P>0.05) between the biofilm forming capacity and MDR phenotype (results not shown).

Discussion

Bioaerosols represent a major factor of bacterial spread in food-related premises (EFSA, 2021; Jericho et al., 2000). Consequently, knowing the aerobiology of a given environment would provide key information to avoid microorganism dissemination by including air dynamics in GHP, GMP, SSOP and HACCP schemes (EFSA, 2021; Haddrell and Thomas, 2017). In this line, our outcomes demonstrated that among pre-operative samples, highest TVC were found in the WR with mean values of 3.67±0.27 log CFU/m3 that progressively diminished as the slaughter line advanced reaching minimum mean TVC of 2.19±0.11 log CFU/m3 in the clean area which significantly increased (P<0.05) after the commencement of the abattoir activity (Table 1). The latter can be attributable to an increase or suspended water droplets loaded with microorganism clumps potentially mobilised by FBOs inside the premise (Haddrell and Thomas, 2017; Pearce et al., 2006). Similar TVC values in WR and CR and the increase in aerial bacterial counts due to the plant activity have been previously reported by Pearce et al., (2006) in an Irish pig abattoir although lower counts were obtained in pre-operative ones due to differences in clumping and physiological state of bacteria affecting their growth onto agar plates (Haddrell and Thomas, 2017; Okraszewska-Lasica et al., 2014). Additionally, results showed a significant decrease of TVC between WR and CR indicating that, at the moment of sampling, physical barriers such as walls and doors and controlled ventilation systems, caused an interruption in the air flow in areas containing a higher bioburden thus impairing its spread (Pearce et al., 2006; Prendergast et al., 2004).

Airborne Gram-negative bacteria (GNB) are commonly associated with animal skin, soil and manure (Okraszewska- Lasica et al., 2012) and disseminated within abattoirs by the incoming animals, suspended dust particles, and faecal matter present in workers’ boots. (Cosenza-Sutton, 2004). A variety of members of Enterobacteriaceae, Moraxellaceae and Pseudomonadaceae were mainly detected in air samples (Table 4). This is in line with the study performed by Cosenza-Sutton, (2004) in three pork abattoirs, concluding that these families are part of the normal microbiota of a swine abattoir. In the light of this, monitoring of aerosolised GNB could be a useful tool in HACCP schemes for hygienic status assessment in abattoirs in terms of contamination from faecal origin (Okraszewska-Lasica et al., 2012). Of note, airborne GNB represented less than half of the TVC, indicating that, in this scenario, aerial microbiota may also harbour microorganisms other than GNB such as Gram-positive bacteria, fungi and yeasts as previously reported (Fernstrom and Goldblatt, 2013).

Regardless of the moment of sampling, recovered Enterobacteriaceae displayed a remarkable level of antimicrobial resistance to β-lactams, SMX, C, COL, TET and TRI (Table 4). Resistance to these antibiotics have been previously detected in E. coli, Serratia spp., Pantoea spp., Klebsiella spp., Enterobacter spp. and other Enterobacteriaceae recovered from samples in a Chinese abattoir (Li et al., 2016). In a similar way, Amador et al. (2019) carried out an study dealing with the antimicrobial resistance of Enterobacteriaceae present in pork and beef slaughterhouses samples in Portugal demonstrating high levels of resistance to β-lactams, C, CIP, TET and sulphametoxazole-trimethoprim (SxT), being in line with our results. Plausible origins of these bacteria could be aerosolised wastewater containing manure considered as one of the major sources of antimicrobicresistant Enterobacteriaceae in pig slaughterhouses (Homeier-Bachmann et al., 2021; Savin et al., 2020).

A remarkable number of airborne Enterobacteriaceae isolated were classified as MDR (Table 4) showing resistance towards AMP (MICmax = 32 mg/L), SMX (MICmax = 512 mg/L), TET (MICmax = 32 mg/L) and TRI (MICmax = 16 mg/L) somehow forming a sort of core-resistance pattern. This phenomenon has been previously described by Schwaiger et al. (2012) in pork samples where isolated Enterobacteriaceae showed common resistance to penicillins, tetracyclines and SxT, indicating not only a transmission and recirculation of genes associated to antimicrobial resistance but also that environmental pressures (humidity, improper use of sanitizers, etc.) selecting for a bacterial subset are present (EFSA, 2021). Regarding this, recent data demonstrates that E. coli exposure to sub-lethal concentrations of disinfectants such as benzalkonium chloride or hydrogen peroxide, induces cross-resistance to AMP and C which is linked to modifications in the membrane composition and subsequent changes in cell surface charge, as well as the selection of mutant variants (Nordholt et al., 2021; Pereira et al., 2021).

Moreover, five isolates were classified as strong biofilm formers (Table 4). Among them, isolate GN45 (P. fluorescens) was also considered as MDR. In abattoirs, MDR Pseudomonas spp. with outstanding capacity to form biofilms have been described (Lerma et al., 2014). These structures consisting of bacteria embedded in a self-produced polymeric extracellular matrix providing species within high endurance to insults such as antimicrobials (Flemming, 2011). Consequently, isolate GN45 represent a potential hazard due to its antimicrobial resistance profile and the spreading capacity as airborne bacteria suspended into dust particles or water droplets, accumulating and adhering onto other surfaces generating new foci of contamination harbouring pathogenic and non-pathogenic bacteria that can be further transferred to carcasses (EFSA, 2021; Rodriguez-Lopez et al., 2020).

Among pre-operative samples, identified bacteria belonged to Aeromonadaceae, Brucellaceae, Burkholdeliaceae, Moraxellaceae and Pseudomonadaceae (Table 4), previously described as part of the microbiota in food-processing environments (Moretro and Langsrud, 2017). Moreover, pre-operative samples give information about the surfaces’ resident microbiota providing the first step for aggregation of further bacterial species coming from different sources (incoming animals, water, dust particles, etc.) during the normal activity of the plant (Moretro and Langsrud, 2017). Among them, isolate GN26 (A. baumannii) is of especial relevance both because its adhesion capacity and its MDR characteristics (Table 4). Regarding this, a study carried out by Li et al. (2016) demonstrated that among Acinetobacter spp. isolated from pork at slaughter, high resistances to AMP and C were found and that more than 50% of them displayed resistance to multiple antibiotics. Similarly, Hrenovic et al. (2019) detected high resistance to AMI (MIC ≥ 256 mg/L) and reduced sensibility to carbapenems and AMP in A. baumannii isolated from swine manure. Considering the potential risk of these species, isolates displaying GN26 features should be carefully monitored due to its capability to adhere forming subsequent contamination foci and also becoming a potential reservoir of ARGs eventually transmitted into the food chain (EFSA, 2021).

Outcomes also demonstrated that not all bacteria isolated from surfaces in pre-operative samples were strong biofilm formers (Table 4). Therefore, the fact that one species is present in a given surface is not only dependent on the adhesion capacity of the bacterium itself but also that, in that particular spot, a lack of proper sanitation could be present. This highlights the importance of proper monitoring of environmental contamination for bacterial control carried out by FBOs and continuous verification of the SSOs efficacy or their modifications whenever needed.

Higher GNB counts concomitant with an increase in the bacterial diversity was observed among operative samples (Table 4). The latter was mainly due to species belonging to Enterobacteriaceae and Moraxellaceae found in both WR and CR (Table 4). Of note, among Enterobacteriaceae an isolate of Klebsiella oxytoca was found which is especially relevant since it is considered nowadays as an emerging pathogen causing nosocomial infections (Neog et al., 2021). Moreover, among values showed that the more separated the locations were, the more different they were which was expected considering that sampled surfaces in the wet area (locations S1 and S2), are in more contact with the slaughtered animals compared with the locations in the clean one (locations S3 and S4) (Table 2). Previous published data described how members of these families have been associated with pigs (Abuoun et al., 2017; Li et al., 2016), fact that can be linked to our results thus indicating that incoming pigs would be one of the main sources of contamination/recontamination in the sampled abattoir.

Conclusions

In this work, the presence of antimicrobial- resistant bacteria in the environmental of a high-throughput pig abattoir was demonstrated. This fact raises concerns in terms and assurance of safety and quality of meat products mainly in two respects: firstly, the diversity in the Gram-negative microbiota found and the levels of resistance, are indicative of a presumable higher burden of ARGs. These can be easily spread horizontally thus generating MDR variants, subsequently disseminated though air directly to foodstuffs or environmentally persist as biofilms. Secondly, it highlights once more the upmost importance for FBOs to follow a “One-Health” approach for the monitoring of hygienic conditions considering the different ecological niches in a given food-related scenario.

Therefore, it is of paramount importance that FBOs implement specific control measures regarding ARB colonisation and transmission in abattoirs and other related food-processing environments in order to avoid dissemination between animals, the environment and FBOs themselves. These objectives can be achieved through improvement of good hygienic practises during the normal activity of the plant as well as the biosecurity practises, i.e., individual protection of personnel, monitoring of environmental contamination and efficacy of sanitation methodologies.