The Prevalence and Characterization of Fecal Extended-Spectrum-Beta-Lactamase-Producing Escherichia coli Isolated from Pigs on Farms of Different Sizes in Latvia

The aim of this study was to determine the prevalence of fecal ESBL-producing Escherichia coli (E. coli) in pigs on large and small farms in Latvia, to characterize beta-lactamase genes and establish an antimicrobial resistance profile. Fecal samples (n = 615) were collected from 4-week, 5-week, 6-week, 8-week, 12-week and 20-week-old piglets, pigs and sows on four large farms (L1, L2, L3, L4) and three small farms (S1, S2, S3) in Latvia. ChromArt ESBL agar and combination disc tests were used for the screening and confirmation of ESBL-producing E. coli. The antimicrobial resistance was determined by the disc diffusion method and ESBL genes were determined by polymerase chain reaction (PCR). Subsequently, ESBL-producing E. coli was confirmed on three large farms, L1 (64.3%), L2 (29.9%), L3 (10.7%) and one small farm, S1 (47.5%); n = 144 (23.4%). The prevalence of ESBL-producing E. coli differed considerably between the large and small farm groups (26.9% vs. 12.7%). Of ESBL E. coli isolates, 96% were multidrug-resistant (MDR), demonstrating there were more extensive MDR phenotypes on large farms. The distribution of ESBL genes was blaTEM (94%), blaCTX-M (86%) and blaSHV (48%). On the small farm, blaSHV dominated, thus demonstrating a positive association with resistance to amoxicillin-clavulanic acid, ceftazidime and cefixime, while on the large farms, blaCTX-M with a positive association to cephalexin and several non-beta lactam antibiotics dominated. The results indicated the prevalence of a broad variety of ESBL-producing E. coli among the small and large farms, putting the larger farms at a higher risk. Individual monitoring of ESBL and their antimicrobial resistance could be an important step in revealing hazardous MDR ESBL-producing E. coli strains and reviewing the management of antibiotic use.


Introduction
Antimicrobial resistance is one of the major global public health concerns currently facing humanity. The alarming situation has been exacerbated by the fact that the "pipeline of new antibiotics is dry". In Europe, an estimated 70% of antibiotics are used in the animal sector, thus increasing the interest in antibiotic consumption in this sector [1]. According to global trends in antibiotic use for food animals, the highest consumption of antimicrobials per kilogram of an animal is in pigs, compared to chickens and cattle [2]. In high-income countries, intensive pig production dominates and pigs are generally kept in indoor systems in which there is limited space, few opportunities to express natural behaviors, and many stressors such as weaning, moving, mixing, surgical procedures, and environmental conditions that increase the risk of production diseases and infections. Digestive problems in early-weaned piglets is the main reason why piglets receive antibiotics around weaning and post-weaning periods [3]. Most of the antibiotics used in pigs are the same or belong to the same classes as those used in human medicine [4]. According to a report

Prevalence of ESBL-Producing E. coli
Out of seven pig farms, the presence of fecal ESBL-producing E. coli was confirmed on four farms (L1, L2, L3 and S1; n = 144 (23.4%)). Among the large farms, the highest prevalence of ESBL-producing E. coli was observed on Farm L1 (64.3%), while among the small farms the prevalence on Farm S1 was 47.5%. The prevalence of ESBL-producing E. coli differed significantly between the large and small farm groups (26.9% vs. 12.7%, p = 0.029). Moreover, a notable difference (p < 0.05) was found between the large and small farms (Table 1).
Comparing the prevalence of fecal ESBL-producing E. coli in pigs of different ages, the presence of ESBL-producing E. coli was significantly higher in 4-week and 6-week-old piglets (28.2% and 35.1%, respectively) and in sows (41.6%), while the lowest prevalence of ESBL-producing E. coli was observed in pigs at 20 weeks (8.2%) ( Table 1).
Forty-eight of 50 (96%) fecal ESBL-producing E. coli isolates were classified as MDR according to Magiorakos et al., [14]. In a large proportion of the weaning-nursery piglets (47.6%) and sows (37.5%), ESBL-producing E. coli isolates were resistant to eight or more different antimicrobial categories, while there were less in the growing-finishing pigs (14.2%). ESBL-producing E. coli isolates from the large farms demonstrated more extensive MDR phenotypes; 79.0% of isolates were resistant to six or more antimicrobials. However, Antibiotics 2021, 10, 1099 4 of 14 on the small farm (S1), 75.0% of ESBL-producing E. coli isolates were classified as MDR5, at the same time, none of them were resistant to six or more antimicrobials. Furthermore, 16.7% of ESBL-producing E. coli isolates were classified as non-MRD ( Figure 2). different antimicrobial categories, while there were less in the growing-finishing pigs (14.2%). ESBL-producing E. coli isolates from the large farms demonstrated more extensive MDR phenotypes; 79.0% of isolates were resistant to six or more antimicrobials. However, on the small farm (S1), 75.0% of ESBL-producing E. coli isolates were classified as MDR5, at the same time, none of them were resistant to six or more antimicrobials. Furthermore, 16.7% of ESBL-producing E. coli isolates were classified as non-MRD ( Figure 2).
The distribution of beta-lactamase gene(s) on the large and small farms was different. On the small farm (S1), the blaSHV combination with other beta-lactamase genes was detected considerably more frequently compared to the large farms (83.3% vs. 36.8%, p = 0.007), while blaCTX-M alone, or in combination with other beta-lactamase genes, was more frequently detected on the large farms compared to the small one (97.4% vs. 50.0%, p < 0.001). On the large farms, the most prevalent beta-lactamase genes were blaTEM + blaCTX-M (52.6%, n = 20), compared to blaTEM + blaSHV (41.7%, n= 5) on the small farm (S1) ( Figure 3).
The distribution of beta-lactamase gene(s) on the large and small farms was different. On the small farm (S1), the bla SHV combination with other beta-lactamase genes was detected considerably more frequently compared to the large farms (83.3% vs. 36.8%, p = 0.007), while bla CTX-M alone, or in combination with other beta-lactamase genes, was more frequently detected on the large farms compared to the small one (97.4% vs. 50.0%, p < 0.001). On the large farms, the most prevalent beta-lactamase genes were bla TEM + bla CTX-M (52.6%, n = 20), compared to bla TEM + bla SHV (41.7%, n= 5) on the small farm (S1) ( Figure 3).
Antibiotics 2021, 10, x FOR PEER REVIEW 4 of 14 different antimicrobial categories, while there were less in the growing-finishing pigs (14.2%). ESBL-producing E. coli isolates from the large farms demonstrated more extensive MDR phenotypes; 79.0% of isolates were resistant to six or more antimicrobials. However, on the small farm (S1), 75.0% of ESBL-producing E. coli isolates were classified as MDR5, at the same time, none of them were resistant to six or more antimicrobials. Furthermore, 16.7% of ESBL-producing E. coli isolates were classified as non-MRD ( Figure 2).
The distribution of beta-lactamase gene(s) on the large and small farms was different. On the small farm (S1), the blaSHV combination with other beta-lactamase genes was detected considerably more frequently compared to the large farms (83.3% vs. 36.8%, p = 0.007), while blaCTX-M alone, or in combination with other beta-lactamase genes, was more frequently detected on the large farms compared to the small one (97.4% vs. 50.0%, p < 0.001). On the large farms, the most prevalent beta-lactamase genes were blaTEM + blaCTX-M (52.6%, n = 20), compared to blaTEM + blaSHV (41.7%, n= 5) on the small farm (S1) ( Figure 3).

Associations of Beta-Lactamase Gene(s) to Phenotypic Antimicrobial
The phenotype of AM-TIC-AMC-CL-CZ-CTX-CFM-FEP was the most commonly observed pattern of resistance to beta-lactams on the large farms (Table 2). This pattern most often contained all three beta-lactamase genes bla TEM + bla SHV + bla CTX-M (52.9%, n = 9) or the combination of bla TEM + bla CTX-M (35, 3%, n = 6). The next most common phenotype that was observed was similar to the one above, but without resistance to cefixime. The phenotypes of AM-TIC-AMC-CZ-CAZ-CTX-CFM-FEP and AM-TIC-AMC-CZ-CAZ-CTX-CFM were the most commonly observed patterns on the small farm (S1). Interestingly, the phenotype AM-TIC-CZ was observed in two cases, one of these contained the bla TEM gene, while the other contained bla TEM + bla SHV + bla CTX-M . Table 2. Occurrence of pattern for resistance to beta lactams antibiotics and beta-lactamase gene(s) detected in ESBLproducing E. coli. Note: AM-ampicillin, TIC-ticarcillin, AMC-amoxicillin-clavulanic acid, CL-cefalexin, CZ-cefazolin, FOX-cefoxitin, CAZ-ceftazidime, CTX-cefotaxime, CFM-cefixime, FEP-cefepime.

Discussion
According to the program set up to monitor AMR in the EU Member states, the prevalence of the ESBL indicator E. coli was reported to be low (1.4%) in fecal samples from fattening pigs at slaughter. However, the specific monitoring program with supplementary testing discovered a high prevalence of presumptive ESBL-producing E. coli in the samples of fattening pigs (34.3%). It should be noted that the level of prevalence varies greatly (0.3-80.3%) among the EU countries [4]. In our study, the prevalence of ESBL-producing E. coli in pigs decreased during the period of growing and fattening. The highest amount of ESBL-producing E. coli was present in sows (41.6%), followed by weaned piglets (35.1%), and was much lower in fattening pigs (8.2%). It was previously reported that ESBLproducing E. coli carriage considerably decreases over the production cycle of pigs [15,16]. Interestingly, in the above study by the monitoring program of the EU, a much higher prevalence of ESBL-producing E. coli in fattening pigs was reported in Latvia, where the prevalence was 42.3% [4] in comparison with our results. One of the reasons could be the differences in the ESBL screening method. The method employed by us did not include a non-selective pre-enrichment step, which could affect the sensitivity of the method. A research study from Germany has shown that by including the enrichment procedure, ESBL-producing E. coli could be isolated twice as often [17]. In addition, a German study reported that during the transportation and the waiting period at the slaughterhouse, the probability of transmission of ESBL-producing Enterobacteriacae among pigs increased to nearly 30% [18]. In our study, samples were obtained on farms; thus, the testing method could result in a higher prevalence of ESBL-producing E. coli.
Our results demonstrated that farm size could be one of the risk factors for an increased occurrence of ESBL-producing E. coli. At the same time, this conclusion should be treated with caution, because farms with the highest prevalence represented both sizes: large (L1) and small (S1). It is more important to highlight the considerable differences observed on farms in general. Several studies on the risk factors of increased prevalence of ESBLproducing E. coli on pig farms have been carried out. The use of private water sources for pigs, lack of well-organized biosecurity (entrance hygiene, pest control, presence of pets, etc.) and hygiene (disinfection, detergent use), the high use of antibiotics in pig production, and in particular, the recent use of cephalosporins [19,20] are serious risk factors. Although our study did not include a detailed evaluation of risk factors for the spread of ESBLproducing E. coli due to the limited number of the farms available for our study, a causal relationship was observed between the increased occurrence of diarrhea and the wide use and misuse of beta-lactam antibiotics in pigs, resulting in an increased prevalence of ESBL. Based on the survey data, on farms L1 and S1, diarrhea in piglets of different ages has been a common problem for a long time; moreover, overuse of amoxicillin and enrofloxacin could be responsible for the selection of ESBL-producing bacteria. Amoxicillin is usually added to feed, but as it has low absorption and bioavailability [21], it creates high selective pressure on gut microbiota, which can promote antibiotic resistance [22]. A correlation between the use of beta-lactam antibiotics (amoxicillin) and the high prevalence rate of ESBL producers was also reported in another study [23]. In contrast, on the sampled two small farms (S2 and S3) and on one large farm (Farm L4), diarrhea was rarely observed. On Farms S2 and S3, antibiotics, and non-beta lactam antibiotics, were used rarely; a similar situation was observed on Farm L4, where antibiotics from the broad-spectrum penicillin group were only used sporadically. The role of fluoroquinolone in increasing the spread of ESBL has also been confirmed [24].
The phenotypic detection of antimicrobial resistance in ESBL-producing E. coli revealed significant differences in resistance between isolates from the large and small farms. With regard to beta-lactam antibiotics, the resistance to cephalexin, cefotaxime and cefepime was more commonly observed in pigs from the large farms, while resistance to ceftazidime was found in pigs from the small farm. Furthermore, resistance to the non-beta lactam antibiotics: trimethoprim, sulfametaxazole-trimetoprim, gentamicin, tetracycline, enrofloxacin was observed only in isolates obtained from the large farms. Phenotypic differences in resistance to beta-lactam antibiotics were discovered by the presence of different beta-lactamases coding genes. As known, ESBL-producing bacteria produce ESBLs enzymes that hydrolyze most third-and fourth-generation cephalosporins, thus explaining the observation of the high prevalence of resistance to cefotaxime (100%) and cefepime (100%). At the same time, a high prevalence of resistance to amoxicillin-clavulanic acid (87%) was observed in fecal ESBL-producing E. coli isolates. One of reasons for this phenomenon could be the production of OXA-1 beta lactamase [25]. Further investigations are needed to confirm this; in addition, this could also be due to the hyperproduction of target beta-lactamases or relative permeability [26]. Extended-spectrum SHV and TEM betalactamases show higher levels of hydrolytic activity for ceftazidime than cefotaxime [27], while a large number of CTX-M beta lactamases show only cefotaxime activity [28]. The co-existence of beta-lactamase genes, especially bla CTX-M, with other genes, was explained by MDR phenotypes [29]. Resistance to mecillinam and imipenem was not observed in our study. Sporadic cases of ESBL-producing E. coli resistance to imipenem (carbapenems) obtained from pigs or pork in the EU have been reported in Germany [4] and Belgium [30]. Mecillinam belongs to the group of broad-spectrum penicillins and historically it has not been widely used in veterinary medicine, possibly due to its poor oral absorption; however, some authors believe that mecillinam has potential in veterinary medicine if administered parenterally [31]. No data are available on the ESBL-producing E. coli resistance to mecillinam in pigs, but several authors have reported the high susceptibility of ESBL-producing E. coli (96.9.2%) and MRD E. coli (96.5%) to mecillinam in humans [32]; therefore, since 2019, mecillinam has been included in category A, which is not authorized for use in veterinary medicine [33].
In contrast to ESBL-producing E. coli on the small farm, all of the ESBL-producing E. coli isolates from pigs on the large farms were classified as MDR; moreover, there were more extensive MDR phenotypes. This is in accordance with another study where these results were positively associated with the farm size, and a higher rate of MDR in E. coli on pig farms was reported [13]. No detailed information on the treatment management on the pig farms was available, but the higher usage of antibiotics in feed or water on the large farms in comparison to parenteral treatment on the small farm could be a credible explanation for higher and more extensive MDR isolates on the large farms. Moreover, the inclusion of zinc oxide as a feed additive could be an important factor in increasing extensive MDR phenotypes. High levels of zinc oxide supplementation may promote antimicrobial resistance and MDR in the pig's gut [34,35]. In our study, the group of growing-finishing pigs showed a tendency to dominate over narrower phenotypes of MDR. Indeed, narrower phenotypes of MDR in growing-finishing pigs were associated with a decrease in external selection pressure due to considerably reduced or completely discontinued antibiotic use during the fattening period. Several studies have reported a significant reduction in MDR E. coli in finishing pigs than in weaned or suckling pigs [36,37]. A higher prevalence of MDR in weaned piglets is usually explained by the wider use of antibiotics for treatment and prophylaxis because younger animals have an increased risk of enteric infections, especially when stressed by weaning and mixing litters [36]. Sows received antibiotics more often by parenteral administration [38], which would be a positive factor for the spread of resistance, but, at the same time, different types of active ingredients of antimicrobials were chosen more often [39], which could contribute to the high prevalence of MRD bacteria.
Among ESBL-producing E. coli isolates, bla TEM was found to be the most predominant, although it was only 8% more present than bla CTX-M . The distribution of ESBL genes in isolates from food-production animals in Europe has been reported by Silva et al. [6]. In pigs, the most frequently observed bla CTX-M genes were bla CTX-M-1 and bla CTX-M-15 from bla CTX-M-group1 . The genes bla CTX-M-group4 and bla CTX-M-group2 were relatively less common with 5% (bla CTX-M-14 ) and 3% (bla CTX-M-2 ), respectively [6]. As bla CTX-M-group2 and bla CTX-M-group4 were not identified in our study, the prevalence of the bla CTX-M genes could be slightly higher. A high prevalence of bla CTX-M-group1 and bla TEM in pigs was reported by other authors from diverse European countries, and similarly to our results, the genes bla TEM and bla CTX-M were most commonly observed in combination [40][41][42].
Our results demonstrated the differences in the distribution of beta-lactamase genes on the large and small farms. The gene of bla CTX-M in combination with other beta-lactamase genes were observed considerably more frequently on the large farms, while the gene of bla SHV in combination with other beta lactamase genes was dominant on the small farm. The main reason for the higher prevalence of bla CTX-M on the large farms could be the zinc oxide added to the feed. A recent study showed that the addition of zinc to fecal suspension in vitro increased the proportion of a plasmid-encoded bla CTX-M-1 E. coli compared with the total microbiota. In addition, zinc may induce the expression of resistance genes in the E. coli strains [43]. On the large farms, zinc was probably responsible for the spread of plasmid-encoded ESBL strains and inducing the expression of the bla CTX-M gene. This may explain the characteristic phenotype of bla CTX-M against several beta lactam antibiotics-cefazolin, cephalexin, cefotaxime and cefepime.
We observed a positive association between the gene bla SHV and the resistance of amoxicillin-clavulanic acid, ceftazidime and cefixime. However, the gene bla CTX-M had a positive association with the resistance of cephalexin and several non-beta-lactam antibiotics-enrofloxacin, tetracycline, gentamicin, trimethoprim and sulfamethoxazoletrimethoprim. Some TEM and SHV mutants have a single amino acid change at the amber position, therefore they overtake the first generation (e.g., clavulanic acid) of beta-lactamase inhibitors [44,45]. Other authors have reported the positive association of SHV with resistance to ceftazidime due to the higher hydrolytic activity of SHV against ceftazidime than against cefotaxime, which is attributed to the Glu 240 Lys substitution [27]. It has not been clearly defined whether the presence of bla SHV specifically causes resistance to cefixime. Moreover, Naziri at al. [46] have reported a high resistance to cefixime equally often in both isolates that contain bla CTX-M or bla SHV . It should be noted that Livermore at al. [47] reported that the level of resistance to the oral cephalosporin (cefixime) depends on the specific enzyme variant. A high degree of resistance was observed in SHV-4, SHV-5, CTX-M-15, but low resistance was observed in SHV-2 and CTX-M-9. This suggests the presence of specific SHV enzymes with an increased activity against cefixime; further in-depth studies would be required for confirmation. In our study, resistance to cephalexin showed a positive association with bla CTX-M , but a negative association with bla SHV . Broad spectrum beta-lactamases are usually especially active against the firstgeneration cephalosporins [48], which explained the positive association with the resistance of cephalexin. Interestingly, our results showed that cephalexin was effective against SHV. Similarly, Bedenic at al. [49] have reported good activity of cephalexin against SHV-2. Observation of the positive association of bla CTX-M with several non-beta lactam antibiotics could indicate the prevalence of specific MDR ESBL clones. Although all of ESBL genes may be located on the mobile genetic elements and are often associated with co-resistance phenotypes with aminoglycosides, fluoroquinolones, trimethoprim-sulphamethoxazole, tetracycline and other antibiotics [50,51], bla CTX-M is most commonly associated with broad MDR strains [29]. Clonal dispersion and the co-selection processes of CTX-M clones promote their spread globally as they are easily transmitted among family members in a colonization [52]. In recent years, there has been concern about the coexistence of bla CTX-M and plasmid-mediated-quinolone-resistance (PMQR), therefore non-discriminatory use of fluoroquinolone may result in the selection and rapid dissemination of MDR ESBLs in commensal E. coli [24]. Furthermore, a strong association of PMQR and bla CTX-Mgroup1 (especially bla CTX-M-15 ) was observed among ESBL-producing E. coli [53,54]. Dietary zinc for weaned piglets also promotes the abundance of MDR and might promote the selective growth and expression of plasmid-encoded ESBL-producing E. coli [43]. The above factors were responsible for the spread of MDR CTX-M strains on the large farms.

Screening of Extended-Spectrum-Beta-Lactamase-Producing E. coli
The screening of ESBL-producing E. coli was performed according to the following method: a directly inoculated sample was placed on the selected media [55,56] Briefly, using a sterile cotton swab, the fecal specimen was gently spread onto the ChromArt ESBL agar (Biolife, Milano, Italy) surface. The plates were incubated at 36 • C for 20-24 h. If the plates were negative, they were re-incubated for an additional 24 h. After the incubation, the large pink colonies were noted and regarded as presumptive ESBL producing E. coli. One typical presumptive ESBL colony was sub-cultured on a Tryptic soy agar (Biolife, IT) and incubated overnight at 36 • C to obtain pure culture for the identification and phenotypic confirmatory test. The oxidase negative, indole positive, urease and citrate negative colonies were considered to be E. coli, and the isolates were confirmed by VITEK MS (bioMerieux SA, Marcy L'Etoile, France) which uses MALDI-TOF technology.

Phenotypic Confirmatory Test for ESBLs in E. coli
According to EUCAST recommendations [57], the combination disc test (CDT) was chosen for the confirmation of ESBL. The CDT was performed in accordance with the Clinical and Laboratory Standards Institute guidelines [58]. A standardized inoculum (0,5McF) of the presumptive ESBL-producing E. coli isolate was swabbed on the surface of Muller Hinton agar II (Biolife, IT). Cefotaxime 30 µg (CTX-30, BD BBL) and ceftazidime 30 µg (CAZ-30, BD BBL) discs alone and in combination with clavulanic acid 10 µg (CTX CLA and CAZ CLA, Biolife) were placed on the plate. The results were obtained after 18 h incubation at 36 • C. ESBL-producing E. coli was confirmed as positive if there was a ≥5 mm increase in the inhibition zone. This was detected by clavulanic acid around either the cefotaxime or the ceftazidime disc, compared to the diameter around the disc containing cefotaxime or ceftazidime alone. E. coli ATCC 25922 (Bioscience, Botolph Claydon, United Kingdom) was used for the quality control when the ESBL confirmatory test was carried out.
The amplification reactions were performed in the Applied Biosystems 2720 thermal cycler under the following conditions: initial denaturation at 95 • C for 5 min, denaturation at 94 • C for 1 min of 35 cycles, annealing at 55 • C for 1 min for bla TEM , at 53 • C for 1 min for bla CTX-M , at 47 • C for 1 min for bla SHV , then extension at 72 • C for 1 min, followed by the final extension at 72 • C for 10 min.
After the amplification, PCR products were separated by a 2% agarose gel. After electrophoresis, the amplified DNA fragments were visualized by UV transilluminator and the gels were photographed. The amplicon size for bla TEM was 431 bp, for bla SHV it was 214 bp and for bla CTX-M it was 621 bp.

Statistical Analysis
The statistical analysis was conducted by R Studio software (version 1.1.463). The Chi square test was used to compare the significance of differences in the prevalence of ESBL-producing E. coli between the large and small farms. The pairwise comparisons from the Chi-squared test were used for post hoc identification of significant differences in the prevalence of ESBL-producing E. coli in different farms and in different age groups of pigs. Fifty of ESBL-producing E. coli isolates were selected by the stratified random sampling method for in-depth studies of antimicrobial resistance and beta-lactamase genes. The Fisher's exact test was used to compare the distribution and proportion of antimicrobial resistance and beta-lactamase genes. Odds ratios and 95% confidence intervals were used to evaluate the associations between beta-lactamase genes and antibiotic resistance. Statistical significance was considered as p < 0.05

Conclusions
The prevalence of ESBL-producing E. coli varies widely among farms and although higher prevalence was observed in the large farms, small farms are also of concern. The screening of ESBL-producing E. coli demonstrated a high prevalence of bla CTX-M with extended MDR phenotypes on large farms, but the dominance of bla SHV may also cause trouble due to the high resistance to ceftazidime and cefixime on small farms. Various contributing factors such as piglet diarrhea, the broad use of amoxicillin, fluoroquinolones and zinc included as a feed additive may contribute to a number of hazardous ESBL clones on farms. Individual analysis of the antibiotic resistance situation on small farms could be an important step for revealing hazardous MDR ESBL strains and to review the antimicrobial therapy in use. Large farms are more exposed to the alarming spread of MDR ESBL, which emphasizes the need to limit the use of antimicrobial therapy and the urgency of finding alternative strategies.

Data Availability Statement:
The data used to support the findings of this study are available from the corresponding author upon request.