Multidrug Resistance and Plasmid Profiles of Escherichia coli Isolated from Lebanese Broiler Farms

The present study was undertaken to determine the antimicrobial resistance patterns and plasmid fingerprints of commensal Escherichia coli isolated from Lebanese broiler chickens. To that end, a total of 30 E. coli isolates were collected from 15 semi-open broiler farms from North Lebanon and Bekaa Valley. Results showed that all the isolates were resistant to at least nine out of 18 evaluated antimicrobial agents. The best-performing antibiotic families were Carbapenems (Imipenem) and Quinolones (Ciprofloxacin and Norfloxacin) to which only 0.0 and 8.3% of the isolates were resistant, respectively. Fifteen various plasmid profiles were depicted, and all the isolates were found to possess one or multiple plasmids. The plasmid sizes varied from 1.2 to 21.0 kbp, and the most commonly detected plasmid had a size of 5.7 kbp (23.3% of the isolates). There was no significant association between the number of plasmids per isolate and resistance to a particular drug. Nevertheless, the presence of specific plasmids, namely, the 2.2 or 7.7 kbp sized ones, was strongly correlated to Quinolones or Trimethoprim resistance, respectively. Both the 7.7 and 6.8 kbp plasmids showed mild correlation to Amikacin resistance, and the 5.7 kbp plasmid was mildly correlated to Piperacillin-Tazobactam resistance. Our findings highlight the need to revise the list of antimicrobials currently used in Lebanese poultry and associate the presence of specific plasmids to antimicrobial resistance patterns in E. coli isolates. The revealed plasmid profiles could also serve any future epidemiological investigation of poultry disease outbreaks in the country.


Introduction
Growing rates of multidrug-resistant (MDR) bacteria pose a serious economic threat to the animal production sector and have major implications for public health. Te increasing resistance to common antibiotics is mostly due to the chaotic utilization of antimicrobial agents to treat diseases and to promote the growth of farm animals [1,2]. Exposure to antimicrobials of various groups can lead to cross-resistance, and the antibiotic resistance genes may spread among bacteria through horizontal gene transfer (HGT). Tat is, the antimicrobial resistance of commensal bacteria, such as Escherichia coli, is equally important as they constitute a reservoir and vector of resistance [3,4]. When E. coli bacteria are in the presence of antibiotic pressure, they are forced to develop alternative ways to survive and grow in a harmful environment. Tat is why the resistance in E. coli has been increasing at a faster rate among livestock isolates than it did among human clinical isolates [5].
In fact, E. coli acquires, through HGT, genes that confer resistance to broad-spectrum Cephalosporins, Carbapenems, Aminoglycosides, Fluoroquinolones, and Macrolides among other antibiotic classes. Several studies have recently shown that the acquisition of resistance could be encoded by chromosomal or plasmid-mediated genes in bacteria. More specifcally, plasmids are considered as the main vector in the dissemination of multidrug resistance through HGT [6,7]. Plasmids carrying resistance genes can be interchangeably transferred among bacteria within the same or diferent species and genera by conjugation and natural transformation [7]. As a result, the treatment of E. coli infections has been complicated by the emergence and dissemination of plasmid-mediated resistance to Fluoroquinolones, Aminoglycosides, broad-spectrum Cephalosporin, Polymyxins, Sulfonamides, Tetracyclines, etc. [6,8].
Several studies have identifed multidrug-resistant E. coli isolated from poultry in Lebanon. Tey have also reported that E. coli originating from animal husbandry is a major biological contaminant of the Lebanese marine environment and water resources used for irrigation and drinking. In most of these studies, the major tools for genetic analysis were (1) the end point PCR, whereby selected genes were targeted, and (2) sequencing [9][10][11][12][13][14][15][16][17][18][19]. Although the said techniques are highly specifc, plasmid fngerprinting (profling) ofers a relatively simple molecular typing method with low setup costs and provides a more practical tool for an insight on the role of plasmids in the transmission of antibiotic resistance in E. coli, among other bacteria [7,20,21]. However, the utility of plasmid profling depends on multiple factors such as the variability of plasmid patterns within a species, the frequency of plasmid-free isolates, the stability of plasmids, and the reproducibility of plasmid patterns [22].
Tis study describes the antimicrobial resistance patterns of thirty commensal E. coli isolates isolated from selected Lebanese broiler farms and characterizes, for the frst time, their plasmid fngerprints. It determines also the correlation between plasmid profles and specifc antimicrobial resistance patterns of the E. coli isolates. Moreover, the fgures reported in this paper provide baseline data that would serve any epidemiological study to be carried out in the future.  [23]. Te antimicrobial disks (Oxoid Ltd, Basingstoke, Hampshire, England) used in the study were Ampicillin (AMP-10 μg), Amoxicillin (AMC-20 μg), Piperacillin- Tazobactam (TZP-100  . Te plates were incubated overnight at 37°C, and the inhibition zone was subsequently observed and recorded. Interpretation charts were used to determine the susceptibility of E. coli isolate to each of the tested antimicrobials based on the inhibition zone diameter recorded.

Plasmid
Profling. Plasmid profling of the E. coli isolates was performed as per the method described by Lonkar et al. [24]. Briefy, E. coli isolates were harvested at Log phase from 5 ml TPB culture by centrifugation at 2000 rpm for 10 min. Te pellet was resuspended in 300 μl of cold solution I (10 mmol·L −1 EDTA, 50 mmol·L −1 Tris, pH 7.5, 100 μg·L −1 RNase A) and transferred in a 2 mL capacity microcentrifuge tube and vortexed for 1 min, at maximum speed. 300 μl of solution II (0.2 mol·L −1 NaOH, 1% SDS) was added. After 5 min, 300 μl of solution III (2.55 mol·L −1 potassium acetate, pH 4.8) was added. Te tube was then centrifuged at 10,000 g for 15 min. Supernatant was transferred to a second microcentrifuge tube, and proteins were extracted with 600 μl of a mixture of 25 : 24 : 1 (v/v) phenol-chloroformisoamyl alcohol. After gentle shaking, the mixture was centrifuged at room temperature for 5 min at 10,000 g. Te aqueous phase above the white protein interface was collected into a microcentrifuge tube, and the plasmid DNA was precipitated with one volume of isopropanol and pelleted by centrifugation at 4°C for 10 min at 15,000 g. Te isopropanol was carefully removed, and the DNA pellets were dried. Plasmid DNA was dissolved in TE bufer composed of 20 μl of 10 mmol·L −1 Tris, 1 mmol·L −1 EDTA pH 7.5. Te DNA was examined by 1% agarose gel electrophoresis at 80 V for 2 hours. Te kbp length of each plasmid was determined using the Quantity One software (BioRad).

Statistical Analysis.
Pearson's correlation was used to test signifcant trends in the linear association of E. coli plasmid counts and antimicrobial resistance over a 95% confdence interval by using SPSS v. 25 system software (SPSS Inc). Te same test was used to assess the strength and direction of the linear relationship between plasmids of specifc kbp length and resistance to each of the evaluated antimicrobials. Strong positive correlation was determined based on P value and Pearson's coefcient (R) whereby P should be less than 0.001 and R should be greater than 0.7.

Antimicrobial Susceptibility Testing (AST).
In this study, a total of thirty diferent E. coli isolates were retrieved from intestinal samples collected from 15 broiler farms. Te resistance of these isolates was tested against 18 antibiotics belonging to 10 diferent major antimicrobial classes. Each isolate was resistant to at least 11 diferent antimicrobial agents out of 18. As shown in Figure 1, the highest resistance rate was recorded against three antimicrobial classes, namely, Sulfonamides, Tetracyclines, and Macrolides (100.0%) followed by Cephalosporins (96.7%), Monobactams (90.0%), Penicillins (83.3%), and Aminoglycosides (81.7%). Te least resistance rate was observed against Fluoroquinolones (5.0%) and Carbapenems (0.0%). When taken individually, the best active antibiotics were Imipenem, Norfoxacin, and Ciprofoxacin to which the E. coli isolates exhibited 0.0, 3.3, and 6.7% resistance, respectively, while the least active were Penicillin G, Cephalothin, Fusidic Acid, Sulfamethoxazole, Tetracycline Erythromycin, Gentamicin, and Clindamycin with 100.0% E. coli resistance (Table 1).

Plasmid Profling and Its Correlation with Antimicrobial
Resistance. At least one plasmid was detected in all the E. coli isolated in this study. Tirty six percent of the isolates had one plasmid while 33.3% harbored two. Te rest exhibited three (10%), four (13.3%), or fve (3.3%) plasmids. Te depicted plasmids showed a length that was varying between 1.2 and 21.0 kbp (Table 2), and the most commonly detected plasmid (23.3% of the isolates) had a size of 5.7 kbp. Overall, 15 diferent plasmid profles were detected as shown in Table 2 and Figure 2 and were labelled 1 through 15.
Remarkably, even the isolates that had the same plasmid profle did not always demonstrate the same antimicrobial resistance pattern as it is the case of isolate 1 vs. isolates 2, 3, and 4. Te same applies for isolates 15 through 19 that had the same plasmid fngerprinting profle, yet only isolates 18 and 19 had similar antimicrobial resistance pattern in comparison to the others (isolates 15, 16, and 17). Moreover, and despite the diversity of the plasmid profles recorded, all the E. coli isolates were resistant to 9 diferent antibiotics, namely, Ampicillin (AMP), Cephalothin (KF), Sulfamethoxazole (SMX), Tetracycline (T), Erythromycin (E), Fusidic Acid (FC), Gentamicin (GM), Penicillin G (PG), and Clindamycin (CD).
Te antimicrobial resistance profle was not correlated to the number of plasmids detected in the commensal E. coli isolates obtained in this study (R � −0.231, P > 0.05). In addition, most of the detected plasmids were not signifcantly correlated to resistance to any specifc antibiotic among the evaluated ones. Most importantly, few plasmids exhibited strong or mild correlation to specifc antimicrobial resistance (Table 3). For instance, resistance to Fluoroquinolones (CIP and NOR) was strongly correlated to the presence of the 2.2 kbp plasmid (R � 1, P ≤ 0.001) while resistance to TM was strongly correlated to the presence of the 7.7 kbp plasmid (R � 0.882, P ≤ 0.001). On the other hand, resistance to AK was mildly associated to the presence of either 6.8 or 7.7 kbp plasmids (R � 0.363, P ≤ 0.05). One plasmid was also mildly correlated to TZP resistance, namely, the 5.7 kbp with an R value of 0.471 (P ≤ 0.05).

Discussion
Te rapid surge in the development of bacterial resistance to antibiotics is a major concern and an issue of public health interest. In fact, the excessive use of antimicrobial agents in animal husbandry is one of the major contributors to the emergence of resistant strains worldwide whereby antibiotics are used for therapeutic treatment and as growth promoter feed additives. Te alarming results obtained in this study made no exception. In fact, all the commensal E. coli isolated from broilers in this work are multidrugresistant (MDR) bacteria which means that they are resistant to antibiotics that belong to at least three diferent antimicrobial classes [25]. Tis refects the deleterious impact of continuous overuse of antibiotics in the poultry sector in Lebanon including Ampicillin, Gentamicin, Tetracycline, Erythromycin, and Doxycycline. Our fndings agree with abundant studies on multidrug-resistant bacteria isolated from Lebanese poultry. For instance, many authors [10,[12][13][14]16] reported resistance of commensal E. coli isolated from broilers to Penicillin, Ampicillin, Cefepime, Cefotaxime, Levofoxacin, Doripenem, Cefxime, Gentamicin, Kanamycin, Streptomycin, Tetracycline, and Sulfamethoxazole/Trimethoprim among other antibiotics. Remarkably, the same authors also documented a frequent bacterial resistance to Fluoroquinolones (CIP and NOR) which oppose the fndings of this study. Te fuctuating results could be linked to the complexity of resistance acquisition process against Fluoroquinolones. It is well known that the development of bacterial resistance to Quinolones is a multifactorial process whereby genes of both chromosomal and plasmid origin are involved [26]. As a matter of fact, the resistance of Enterobacteriaceae to Quinolones needs a sequence of several mutations: (1) a single mutation in the gyrA gene confers low-level Quinolone resistance; (2) the acquisition of a second mutation either in the amino acid codon Ser-80 or in the amino acid codon Glu-84 of the parC gene confers a moderate level of Ciprofoxacin resistance; (3) a third mutation, the second in gyrA, leads to a high level of Ciprofoxacin resistance; and (4) a fourth mutation, the second in parC, confers the highest level of resistance [27]. It is worth noting that Quinolones, namely, Ciprofoxacin, had the highest occurrence percentage (32.5%) in chicken carcasses collected from the Lebanese market among four antibiotics families, as revealed in the work of Jammoul and El Darra [28]. Tis is one of the indications that Quinolones are among the most frequently used antibiotics in Lebanese poultry husbandry.
All the isolates were susceptible to Imipenem. Similar results were obtained by Mikhayel et al. [14] where bacteria of the Enterobacteriaceae family isolated from rectal swabs of International Journal of Microbiology poultry did not exhibit resistance to Imipenem. Kassem et al. [10] reported inconsistent susceptibility of E. coli isolated from poultry to Carbapenems, and the isolates that were resistant to Doripenem carried a plasmid-mediated gene (blaCMY-2) that confers resistance to Carbapenems in porin-defcient strains. Te absence of resistance to Imipenem in this study and in the works of Mikhayel et al. [14] might be reassuring for the time being as it is considered, along with Colistin, one of the last resort antibiotics for the treatment of human infection with MDR Gram-negative Table 1: Resistance pattern of E. coli isolated from poultry intestinal specimens to various antimicrobials evaluated in this study.

Antibiotic
Antibiotic bacteria [29]. However, E. coli resistance to Doripenem, reported in the work of Kassem et al. [10], might raise a red fag vis-a-vis the use of Carbapenems in controlling Gramnegative bacterial infections in the future.
In regard to plasmid profles, the number of plasmids/ isolate varied between one and fve, with kbp length varying between 1.2 and 21.0. A total of 15 plasmid profle patterns were recorded for the 30 commensal E. coli isolates. Te     perfect situation would be to show that all of the isolates that have a specifc antibiogram profle would have the same plasmid number and fngerprint. However, the number of plasmids/isolate was found to have no signifcant correlation to antimicrobial resistance. Tere is abundant literature that reports the same fndings whereby resistance to antibiotics is related more to the presence of specifc plasmids rather than their number [30][31][32][33]. Te reason could be that plasmids have numerous functions besides antimicrobial resistance such as fertility plasmids (F), virulence plasmids, degradative plasmids, and Col plasmids responsible for the production of bacteriocin (colicin). Tat is, a higher number of plasmids would not entail automatically higher resistance to antibiotics, especially in Gram-negative bacteria where most of the said plasmids have been reported [34]. Te presence of multiple resistance on a single plasmid or multiple plasmids in the same organism might also explain the lack of signifcant correlation between the plasmid number and the E. coli resistance pattern [35]. In addition, the role of integrons in bacterial exchange of genes between its chromosome and plasmids is vital in the acquisition and dissemination of resistance genes. Consequently, the dynamics of these integrons might further explain the absence of a signifcant association between the multidrug resistance of E. coli isolated in this study and the number of plasmids per isolate [36]. Remarkably, all of the isolates carried at least one plasmid which prohibited the comparison of the resistance pattern between plasmid-bearing and plasmid-free E. coli isolates. Tis study reveals a solid relationship between the presence of particular plasmids and resistance to antimicrobial agent(s). Te strong correlation between the presence of the 2.2 kbp plasmid and the resistance to Quinolones evaluated in this study (CIP and NOR) indicates that even small plasmids could be carriers of specifc resistance patterns [37][38][39]. Although these small plasmids are most likely not self-transmissible due to the lack of conjugative genes, the presence of other large plasmids could fll the gap and help transfer small plasmids and other short transmissible elements (STEs) horizontally from one bacterium to another [40]. Tis study also reported a strong correlation between the presence of a 7.7 kbp plasmid in E. coli and resistance to Trimethoprim (TM). Several studies have linked TM resistance in E. coli isolates to large plasmids that could have a size of up to 180 kbp. A single large plasmid can carry resistance genes to many other antibiotics at the same time such as Chloramphenicol, Tetracyclines, Aminoglycosides, Quinolones, and Sulfonamides [8,37]. Apparently, smaller plasmids would carry resistance traits against very few antibiotics as in the case of the 7.7 kbp revealed in this study and which is also mildly correlated to AK resistance along with the 6.8 kbp one. Actually, there is abundant literature correlating AK resistance in E. coli to the presence of specifc genes in small plasmids (less than 13 kbp) that carry genes encoding for Amikacin phosphotransferases and adenylyltransferases. Moreover, and despite the fact that small plasmids carry very few genes, their presence play a significant role in plasmid evolution through processes mediated by mobile elements and mechanisms of recombination with other plasmids leading to an increased resistance to several antibiotics including Amikacin [41][42][43]. Piperacillin/Tazobactam (TZP) resistance in the studied E. coli isolates was mildly correlated to another small plasmid having a length of 5.7 kbp. In general, the predominant cause of resistance to β-lactam antibiotics, including TZP, in Gram-negative bacteria is mostly related to the hyperproduction of plasmid-mediated TEM-1 β-lactamases, production of extended-spectrum beta-lactamases (ESBLs), production of AmpC enzymes which are encoded by plasmid genes, and Carbapenem-hydrolyzing β-lactamases [44][45][46]. Te plasmid-mediated TZP resistance does not necessarily dictate the presence of large plasmids, as revealed in this study. Tese results are in agreement with those of Hubbard et al. [47] who reported that a small circular DNA of around 10.9 kbp was found in TZP-resistant E. coli isolates. Te said authors revealed that the large plasmids that were present in the TZP-resistant strains did not contain any antimicrobial or heavy metal resistance genes. However, the characterization of the small circular DNA molecule showed that it contained the missing TZP resistance genes along with several putative transposable elements.
Te correlation results reported in this study indicate a strong association between specifc plasmids in commensal E. coli isolated from poultry and resistance to particular antibiotics. However, this can be also very useful when conducting epidemiological surveillance and consequently developing strategies to curb the spread of plasmid-borne bacterial resistance. Te fndings could have been confrmed through conjugation studies whereby a better understanding of resistance would have been achieved by predicting the transfer of antimicrobial resistance-mediated conjugative plasmids. Moreover, plasmid sequencing could have unveiled key plasmid-specifc functions such as conjugative ability, replication, and mobility. Tis would have enabled the classifcation of the plasmids reported in this study into various categories based on their phylogenetic relatedness and provided insight into the epidemiology of plasmidmediated antimicrobial resistance of commensal E. coli isolated from poultry.

Data Availability
Te datasets used and/or analyzed during this study are available from the corresponding author on reasonable request. collected data. PA contributed in data organization and manuscript formatting. All authors have read and approved the fnal manuscript.