Bacteriocin-Producing Escherichia coli Isolated from the Gastrointestinal Tract of Farm Animals: Prevalence, Molecular Characterization and Potential for Application

Due to the spread of antibiotic-resistant bacteria, new alternatives to antibiotics and ways to prevent infections are being sought. Bacteriocin-producing bacteria are therefore attracting attention due to their probiotic potential as a safe alternative to antimicrobial drugs. The aim of this work was to determine the prevalence of bacteriocin-encoded genes among Escherichia coli strains from healthy farm animals and to characterize the presence of virulence-associated genes, the possibility of prophage induction, and hemolytic and bacterial antagonistic activity of the bacteriocin-producing E. coli in order to reveal their potential for application. It was found that 17 of 72 E. coli strains (23.6%) produced bacteriocins. Among them, 18 out of 30 bacteriocin genes were detected: the most prevalent genes were those for microcin M (58.8%), colicin E1 (52.9%), and colicin M (35.3%). Colicin Ia (29.4%), colicin E9, colicin Ib, colicin B (23.5%), and colicin E9 (17.7%) genes were also frequent, while the prevalence of genes encoding microcins V, B17, and H47 and colicins E3, K, N, U, Y, 5, and 10 did not exceed 11.8%. At least two different bacteriocin genes were detected in all 17 bacteriocinogenic strains; the highest number of different bacteriocin genes detected in one strain was seven genes. E. coli strains with combinations of colicin E1 and E or microcin M and colicin E1 genes were more prevalent than others (17.7%). Among the 17 bacteriocin-producing E. coli strains, 5.9% were hemolytic, 47.1% contained prophages, and 58.8% carried genes encoding toxins. Cell-free supernatants of bacteriocin-producing strains were shown to inhibit the growth of pathogenic E. coli strains belonging to the APEC, STEC, and ETEC pathotypes. Thus, among the studied bacteriocin-producing E. coli isolated from the gastrointestinal tract of farm animals, three strains with high antagonistic bacterial activity and the absence of pathogenicity genes, prophages, and hemolytic activity were identified and therefore have potential for application.


Introduction
The intestinal microbiota of animals is a dynamic community of microorganisms, among which bacteria of the species Escherichia coli are of particular importance. Pathogenic strains of E. coli cause a wide range of diseases in livestock: intestinal pathogenic E. coli (IPEC) cause outbreaks of intestinal infections and uropathogenic E. coli (UPEC), sepsiscausing E. coli (SePEC), and neonatal meningitis-associated E. coli (NMEC) lead to infectious processes in extra-intestinal areas [1,2]. Of special significance are avian pathogenic E. coli (APEC) which cause colibacillosis [3,4]. Commensal E. coli produce vitamins, are involved in modulating the host immune system, strengthen the intestinal barrier, and provide Microorganisms 2022, 10, 1558 2 of 12 colonization resistance of the intestinal mucosa against pathogenic and opportunistic bacteria competing for a place in this ecological niche [5,6].
Since the widespread use of antimicrobials, including in agriculture, has led to the spread of antibiotic resistance, studies are being conducted on the antimicrobial activity of bacteriocins and bacteriocin-producing strains as an alternative to antibiotic therapy [7,[12][13][14]. On the one hand, bacteriocin-producing strains can be used as producers of new antimicrobial substances/bacteriocins; on the other hand, bacteriocin-producing strains themselves can be used as probiotics. The main staples in the search for new useful bacteriocin-producing E. coli strains are the isolation of bacteriocin-producing bacteria from natural sources, the comparative assessment of their activity, and the selection of the most promising strains [15,16]. E. coli from the normal gut microbiota of healthy animals are one of the safe sources of bacteriocin-producing strains.
The aim of the study was to determine the prevalence of bacteriocin-encoded genes among E. coli strains from the feces of healthy farm animals and to characterize the presence of virulence-associated genes, the possibility of prophage induction, and the hemolytic and bacterial antagonistic activity of the bacteriocin-producing E. coli in order to reveal their potential for application.

Bacterial Strains
For this study, 97 E. coli isolates obtained from fecal samples of healthy animals (chicken, turkeys, pigs, rabbits, cows) at agricultural enterprises and private farms in Perm Krai (Russia) between 2019 and 2021 were used ( Table 1). The diluted animal feces was sown directly onto the surface of the MacConkey agar (Sigma-Aldrich, St. Louis, MO, USA), plates were incubated overnight at 37 • C, and the next day typical E. coli colonies (pink colored, transparent smooth and raised colonies or dark red colonies with pronounced metallic luster) were taken from the MacConkey agar and subcultured onto MacConkey agar again. Gram staining (Gram-negative rods) and biochemical tests: Kligler Iron Agar (decomposes glucose and lactose to acid and gas, without hydrogen sulfide-yellow color with interruptions); Simmons Citrate Agar (does not use citrate-not change the color); and catalase (positive) and oxidase tests (negative) were used for preliminary identification of bacteria. Finally, for confirmation of E. coli by the ENTEROtest 16 diagnostic system (Erba Lachema s.r.o., Czech Republic) with the Russian version of the Micro-La-test "Microb-2" (Microbiological system monitoring "Microbe-2", SMMM-2) computer program, one randomly chosen isolate from each animal, thus 97 isolates in total, were taken. Isolates with an "acceptable" rating and a match rate of more than 94% were confirmed as E. coli. All 97 by ENTEROtest 16 analyzed isolates were confirmed to be E. coli, so all were taken for the final differentiation between clonal and non-clonal E. coli with ERIC-PCR [17]. Finally, 72 non-clonal E. coli strains (6 strains from chicken, 3 from turkey, 7 from quail, 3 from pigs, 4 from rabbits, 49 from cows) were collected and further analyzed in this study.
To assess the antagonistic effect, the following strains were used: APEC strains (BR4, BR35, and BR37) were previously isolated from poultry with colibacillosis at Perm Krai breeding farms [18] and deposited in the Ex culture collection of the University of Ljubljana (Univerza v Ljubljana, Slovenia) under the numbers L-5838, L-5865, and L-5868, respectively. IPEC strains (CA29, CA43, CA46) were obtained from cattle at agricultural enterprises in the Perm Krai. The belonging of test-strains to IPEC pathotypes was determined by detecting STEC and ETEC marker genes: stx1 (strain CA29), est1 (strain CA43), stx2+est1 (strain CA46). The resistance of test strains to antibiotics was determined by the disc-diffusion method [19].

Prevalence of Bacteriocin-Producing Strains
The detection of bacteriocin-producing strains was performed with the agar overlay method according to M. Budič et al., 2011 [20]. E. coli DH5α (SCPM-Obolensk, State Collection of Pathogenic Microorganisms and Cell Cultures of the State Research Center for Applied Microbiology and Biotechnology) was used as the sensitive indicator strain and E. coli ŽP, carrying the colicin E7 gene, constructed at the Biotechnical Faculty of the University of Ljubljana on the basis of the probiotic strain E. coli Nissle 1917 [21], was used as the bacteriocin-producing positive control strain. Strains that induced a lysis zone around the spotted tested isolated E. coli colony in the overlaid indicator strain were recognized as bacteriocin-producing. The diameters of the lysis zones of the indicator strain were measured (mm).

DNA Extraction
To obtain matrix DNA for PCR amplification, a loop of bacterial biomass was resuspended into 100 µL of ultrapure water, heated for 15 min at 97 • C in a solid-state thermostat with a timer TT-2 "Termite" (Russia), and centrifuged for 5 min at 13,000 rpm. The supernatants were transferred to fresh Eppendorf tubes and stored at −20 • C until further usage.

Evaluation of Antagonistic Activity of CFS of Bacteriocin-Producing E. coli
The antagonistic effect of bacteriocin-producing strains was assessed by evaluating the bacterial growth of pathogenic E. coli (APEC and IPEC) in the presence of metabolites of the studied E. coli strains in the culture medium. Overnight LB broth ("Difco", Le Pont de Claix Saint-Ferréol, France) cultures of APEC and IPEC were adjusted to 10 6 cell/mL with fresh medium, and 100 µL were added to the wells of 96-well microtiter plates containing either 100 µL CFS from 24 h cultures of bacteriocin-producing E. coli (antagonist strains) or 100 µL of fresh LB (control wells). The microtiter plates were incubated without shaking for 24 h at 35 • C. Subsequetly, the OD 600 of cultures was measured using the plate reader INFINITE M1000 (Tecan Austria GmbH, Grödig, Austria, and the percentage of growth inhibition after 22 h of co-cultivation was calculated, taking as 100% the optical density of the culture grown in the control wells.

Bacteriophage Induction
Bacterial overnight cultures were diluted in saline phosphate buffer (PBS) in order to obtain a concentration from 1 × 10 5 to 1 × 10 6 bacteria per ml, and 20 mL of such diluted overnight cultures were transferred into standard Petri dishes for exposure to the continuous UV-light treatment (λ = 260) for 70 s. After the UV exposure, the cultures were incubated for 1 h at 37 • C and then mixed with a culture of the sensitive strain E. coli DH5α and added to melted 0.6% agar (46 • C), mixed, and poured onto LB agar plates. After a 24 h incubation at 37 • C the presence of lysis zones of the sensitive strain was screened for.

Hemolytic Activities
The hemolytic activities of farm strains were determined by streak-plating on ram blood agar 5% (v/v; "Biomed" a branch of FSUE "Microgen", Russia). After a 24 h incubation at 37 • C the hemolytic zones were observed and the diameter (mm) of the lysis zone was measured.

Statistical Analysis
Data are presented as the arithmetic mean and its standard deviation (M ± SE). The statistical significance was tested with Fisher's exact test or Student's test (t-test), and p-values < 0.05 were considered statistically significant. Data processing was carried out using computer programs Microsoft Office XP Excel and STATISTICA 10.0.

Prevalence of Bacteriocin-Producing E. coli
Among the 72 E. coli strains studied, 17 (23.6%) were capable of producing lysis zones in the overlaid indicator strain DH5α, and hence these strains were recognized as bacteriocin-producing strains. The majority of these strains were isolated from fecal samples of cows (26.5%), while bacteriocin producers were not detected among fecal isolates from turkeys, rabbits, and pigs. It should be noted that among the tested strains, six bacteriocin-Microorganisms 2022, 10, 1558 5 of 12 producing strains had lysis zone diameters larger than the bacteriocin-producing positive control strain ŽP (Table 2). Table 2. Screening for the bacteriocin-producing E. coli by the agar overlay method.
With PCR, 30 different bacteriocin genes were screened for. Among the recognized 17 bacteriocin-producing E. coli strains, 18 of the screened 30 bacteriocin genes were found (Table 3). In all, in 17 strains at least one bacteriocin gene was found, and 47.1% of strains possessed two genes, 17.7% possessed three genes, and 29.4% of strains possessed four or more genes. One strain possessed six and one strain possessed seven different bacteriocin genes. 41.2% of the strains contained only colicin genes, and 58.8% contained both colicins and microcins.
Among the studied bacteriocin-producing strains, the most prevalent were microcin M (58.8%) and colicin E1 (52.9%) genes, followed by colicin E9, colicin Ib, colicin B genes (all three found in 23.5% of strains), colicin Ia genes (found in 29.4%), and colicin M genes (in 35.3% of bacteriocin-producing strains). The genes of microcins B17 and V and colicins K, E8, U, and Y were found each only in one strain. The genes of microcins J25, C7, and L and colicins E2, E4, E5, E6, E7, A, D, S4, and Js were not detected. Of the 17 strains, 12 (70.6%) had unique profiles of bacteriocin genes (not repeated in any other analyzed strain). E. coli strains with combinations of colicin E1 and E9 genes or microcin M and colicin E1 genes were each found in 17.7% of strains.
All analyzed bacteriocin-producing strains encoded at least one pore-forming bacteriocin ( Figure 1). The gene of colicin M with the mode of action of peptidoglycan degradation was found in 35.3% of strains. Strains encoding colicins and microcins with two different modes of action were found in 58.8% of strains, followed by those encoding bacteriocins with three different modes of action and four different modes of action, found in 23.5%, and 11.8% of strains, respectively. The probability of detecting strains containing genes of several bacteriocins with the same mode of action was only 5.9%; 35.3% of strains encoded simultaneously bacteriocins with a pore-forming mechanism and with the mode of action of degrading the peptidoglycan, and 29.4% of strains encoded simultaneously bacteriocins with a pore-forming mechanism and bacteriocins degrading nucleic acids. Strains that had E. coli DH5a lysis zone diameters of 10 mm or more often contained combinations of four or more bacteriocin genes (Fisher's exact test: p = 0.02). Poreforming Inhibition of RNA polymerase Inhibition of Asp-tRNA synthesis Other mechanisms Note: "+"-the gene was detected; "−"-the gene was not detected.

The Effect of CFS of Bacteriocin-Producing E. coli on the Growth of Pathogenic E. coli
All CFS obtained from the 17 bacteriocin-producing E. coli strains inhibited the growth of pathogenic E. coli after 22 h of cultivation, albeit with different levels of efficiency (Table 4). It should be noted that the antagonistic activity of most of the studied strains was higher than that of the control ŽP strain against APEC strains and CA29.
23.5%, and 11.8% of strains, respectively. The probability of detecting strains containing genes of several bacteriocins with the same mode of action was only 5.9%; 35.3% of strains encoded simultaneously bacteriocins with a pore-forming mechanism and with the mode of action of degrading the peptidoglycan, and 29.4% of strains encoded simultaneously bacteriocins with a pore-forming mechanism and bacteriocins degrading nucleic acids. Strains that had E. coli DH5a lysis zone diameters of 10 mm or more often contained combinations of four or more bacteriocin genes (Fisher's exact test: p=0.02).

The Effect of CFS of Bacteriocin-Producing E. coli on the Growth of Pathogenic E. coli
All CFS obtained from the 17 bacteriocin-producing E. coli strains inhibited the growth of pathogenic E. coli after 22 h of cultivation, albeit with different levels of efficiency (Table 4). It should be noted that the antagonistic activity of most of the studied strains was higher than that of the control ŽP strain against APEC strains and CA29.   The maximum value of inhibition index was 58.6% (CFS obtained from strain C23 on BR4 and CFS obtained from C41 on multidrug resistant BR35). All CFS effectively suppressed the growth of the multidrug resistant BR37; however, the maximum inhibition index reached 47.6% (suspension of strain Q8). In the presence of CFS, growth reduction of IPEC strains isolated from cattle was also observed, although the obtained inhibition indexes were smaller than for APEC: for most strains the difference was statistically significant (with the exception of strains C41, C48, and C61 (t-test)). For the hetero-pathogenic strain CA29, the maximum inhibition index was 36.6% (CFS of strain C61). The addition of CFS of strain CA49 did not inhibit the growth of pathogenic strains C43 and CA46. The CFS of strain CA61 had the maximum inhibition index (42.0% for the C43 and 42.1% for the C46). It should be noted that the antagonistic effect was observed against strains of both pathotypes, APEC and IPEC, and also against multidrug resistant strains (Table 5). Note: Amk-amikacin, Amp-ampicillin, Azt-aztreonam, Cfo-cefotaxime, Cfp-cefepim, Cfr-ceftriaxone, Cft-ceftazidim, Cpf-ciprofloxacin, Gen-gentamycin, Lfc-levofloxacin, Tet-tetracycline.

Biological and Molecular-Genetic Properties of Bacteriocin-Producing E. coli Strains
The biological characteristics of bacteriocin-producing E. coli strains are presented in Table 6. Only the strain C45 was detected to be hemolytic, while 47.1% of strains possessed prophages that could be activated by UV. Virulence-associated genes encoding toxins were carried by 58.8% of strains. Three strains (Q5, C23, and C41) did not contain any virulence-associated genes, nor bacteriophages. Table 6. Virulence-associated genes, presence of bacteriophages and hemolytic activity of bacteriocinproducing E. coli strains.

Discussion
In this work, screening of bacteriocin-producing E. coli strains isolated from healthy farm animals was performed as well as further characterization of the bacteriocin-producing E. coli strains found: the detection of bacteriocin genes and virulence-associated genes. In addition, CFS of the bacteriocin-producing strains were tested for antagonistic activity against APEC and IPEC, including multidrug-resistant strains.
The percentage of bacteriocin-producing E. coli strains isolated from different biotopes and ecological niches varied in different studies. In the study by Cameron et al., 2019 [15], 15.6% of isolates isolated from wastewater and feces of cattle inhibited the growth of the test culture. Mazurek-Popczyk et al., 2020 [7], showed that among isolates obtained from the feces of healthy humans, 37.1% were bacteriocinogenic. Micenková et al., 2016 [25], reported much higher numbers: 54.2% of all tested fecal E. coli isolates were found to be bacteriocin producers. An even higher prevalence of bacteriocin producers was shown by Budič et al., 2011 [20], in the population of E. coli isolated from patients with bacteremia: 61% of the strains carried at least one bacteriocin. In our study, the prevalence of bacteriocinproducing E. coli strains among representatives of the microbiota of healthy animals was 23.6%, while among cattle the frequency of occurrence of such strains was 26.5%. The relatively low prevalence of bacteriocinogenity in the population of E. coli obtained from healthy farm animals might be explained by the conditions in agricultural enterprises-the animals are artificially isolated from the environment (mostly kept indoors, use of standard feeding). Indirectly, this assumption could be confirmed by the fact that mainly bacteriocinproducing strains were isolated from cows, grazing outdoors in the summer-autumn period, while bacteriocin-producing strains were not isolated from turkeys, rabbits, and pigs living in boxes/cages all year round. Another fact that limits the biodiversity of strains, including bacteriocinogenity, may be the use of multiple antibiotic therapy regimens at the enterprises.
It is known that encoding multiple bacteriocins gives the cell a selective advantage since such strains have a broader inhibitory effect spectrum on competing organisms [11], including bacteria with multiple resistance to bacteriocins [26,27]. In a study by Gordon O'Brien, 2006 [8], it was shown that among the bacteriocin-producing strains, 42% produced one type of bacteriocin, 41% produced two, 16% produced three, and one strain produced four different types of bacteriocins. In our study, more than one bacteriocin gene was detected in all 17 strains, while about 30% of the strains had four or more bacteriocin genes.
Among the E. coli strains isolated from different sources, most often genes of the following colicins E1, Ib, K, and M and microcins H47, M, and V were found [7,15,20]. A number of researchers have reported correlations of encoding microcins H47 and M as well as colicins B and M [8,28]. In our study, the majority of strains (94.1%) encoded several bacteriocins with different modes of action. The most prevalent were strains encoding pore-forming bacteriocins (colicin E1, Ia, B) and those encoding colicin M with the mode of action of peptidoglycan degradation as well as microcin M (Table 2, Figure 1). It should be noted that out of 30 bacteriocin genes, 12 were not detected, including the E7 colicin gene, which, according to Mazurek-Popczyk et al., 2020 [7], is a rare colicin among the commensal microbiota.
According to some studies, E. coli bacteriocins are effective against IPEC strains, including the O157:H7 serotype [29]. The development of probiotics based on bacteriocinproducing strains that eliminate STEC in farm animals can prevent zoonotic transmission of resistant bacteria to humans. Taking into account the problem of bacterial antibiotic resistance, bacteriocins are considered promising candidates for use in agriculture, especially in relation to multidrug-resistant bacteria [14]. Mazurek-Popczyk et al., 2020, showed that bacteriocin-producing E. coli strains showed the same antagonistic activity against antibiotic-resistant and antibiotic-sensitive zoonotic E. coli strains [7]. Our study also showed that CFS of bacteriocin-producing strains inhibited the growth of both sensitive and antibacterial-drug-resistant, even multidrug-resistant, pathogenic E. coli belonging to different pathotypes.
Interestingly, we did not find a significant difference in antagonistic activity against APEC strains that differ in sensitivity to various antibiotics and bacteriocins ( Table 5). The CFS of E. coli strains C32, C40, and C41, encoding the colicin E9 with the DNase activity, inhibited growth of all APEC strains by 50% or more. The growth of strains BR4, BR35, and BR37, which differed in sensitivity to bacteriocins, was significantly suppressed in the presence of CFS of bacteriocin-producing E. coli from farm animals. This could be due to the action of other bacteriocins with different modes of action preventing the appearance of insensitive strains. It is known that the use of only one bacteriocin can quickly give rise to insensitive strains [20].
The main primary requirements for the selection and verification of probiotic strains of microorganisms are the absence of virulence-associated genes (the strain should not produce enzymes related to virulence) and mechanisms associated with horizontal transmission of genetic information [30]. Preliminary characterization of the strains in terms of hemolytic activity, lysogenicity, and the presence of virulence-associated genes allowed us to select three potentially probiotic strains with high antagonistic activity against pathogenic and conditionally pathogenic E. coli from the 17 bacteriocin-producing strains, i.e., studied strains Q5, C23, and C41, whose probiotic potential can be further explored. For the development of a probiotic product, this primary study should be followed up in the future with the evaluation of the strains' ability to actively colonize and persist long-term in the biotope as well as their biological safety. It is also necessary to select the optimal growth medium for the biotechnological process of producing the potential probiotic drug and to evaluate the survival of the microorganisms in the form in which the drug will be stored.

Conclusions
Increasing resistance of microorganisms to antibiotics in different areas such as medicine and veterinary medicine has encouraged researchers to seek alternative antimicrobial therapies. The potential of commensal strains from healthy animals to defend against zoonotic pathogenic strains was estimated in this study. The study revealed that bacteriocinogenic commensal E. coli strains show antagonistic activity against IPEC and APEC strains and inhibited the growth of both sensitive and antimicrobial-drug-resistant pathogenic E. coli of animal origin belonging to different pathotypes.