Molecular characterization of Escherichia coli isolated from milk samples with regard to virulence factors and antibiotic resistance

Abstract Background and Aim: Raw milk is considered an essential source of nutrition during all stages of human life because it offers a valuable supply of protein and minerals. Importantly, milk is considered a good media for the growth and contamination of many pathogenic bacteria, especially food-borne pathogens such as Escherichia coli. Thus, the objective of this study was to characterize E. coli and detect its virulence factors and antibiotic resistance from raw milk samples. Materials and Methods: Raw milk samples (n=100) were collected from different localities in Qena, Egypt, and investigated for the presence of E. coli using different biochemical tests, IMViC tests, serotyping to detect somatic antigen type, and molecularly by polymerase chain reaction (PCR) tests. The presence of different virulence and antimicrobial genes (hly, eae, stx1, stx2, blaTEM, tetA(A), and tetB genes) in E. coli isolates was evaluated using PCR. Results: The results demonstrated that 10 out of 100 milk samples were contaminated with E. coli. Depending on serology, the isolates were classified as O114 (one isolate), O27 (two isolates), O111 (one isolate), O125 (two isolates), and untypeable (five isolates) E. coli. The sequencing of partially amplified 16S rRNA of the untypeable isolates resulted in one isolate, which was initially misidentified as untypeable E. coli but later proved as Enterobacter hormaechei. Moreover, antibacterial susceptibility analysis revealed that nearly all isolates were resistant to more than 3 families of antibiotics, particularly to b-lactams, clindamycin, and rifampin. PCR results demonstrated that all E. coli isolates showed an accurate amplicon for the blaTEM and tetA(A) genes, four isolates harbored eae gene, other four harbored tetB gene, and only one isolate exhibited a positive stx2 gene. Conclusion: Our study explored vital methods for identifying E. coli as a harmful pathogen of raw milk using 16S rRNA sequencing, phylogenetic analysis, and detection of virulence factors and antibiotic-resistant genes.


Introduction
Escherichia coli is a facultative anaerobe and one of the normal inhabitants in the human and animal intestinal tracts [1]. However, the pathogenic strains of E. coli cause many diseases [2]. Recently, pathogenic E. coli strains have been categorized using different antibodies for perceiving surface antigens correlated to "183 O-groups (lipopolysaccharide) and 53 H-types (flagellar antigen)" [3]. E. coli is differentiated into several pathotypes such as enterotoxigenic E. coli (ETEC) (O27), enterohemorrhagic E.coli (EHEC) (O111), and enteropathogenic E.coli (EPEC)" [4]. In general, these pathotypes encode genes for specific virulence factors that are associated with the attachment and secretion of hemolysins and enterotoxins; however, there is a significant polymorphism presenting the nucleotide sequences of these genes [5].
Dissimilar strains of pathogenic E. coli produce several effective toxins, such as Shiga-like toxins. These Shiga toxin-producing E. coli (STEC) contribute toward gastroenteritis, bloody diarrhea, and uremic syndrome in infected humans [6]. Other virulence factors elevate the pathogenicity of E. coli combined with plasmid-encoded enterohemolysin (hlyA), which is frequently associated with acute sickness in individuals [7], and also with intimin (eaeA), which is responsible for attachment, adherence, and clustering epithelial cell surfaces [8]. Polymerase chain reaction (PCR) analysis has detected that all strains of E. coli possess these virulence genes. However, the procurement of one or more virulence genes does not place bacteria in a harmful grade if that strain has not harbored the proper virulence gene that initiates disease in specific species [9]. "The dispersal of antibacterial resistance genes has been noticed between E. coli isolates from human, animal, and environmental sources." Antimicrobial resistance has been observed as a severe problem worldwide [10]. However, the incidence rates for antibiotic-resistant E. coli strains Available at www.veterinaryworld.org/Vol.14/September-2021/16.pdf diverge in different environments [10]. Many authors have suggested that the extensive administration of penicillin, tetracyclines, and other sulfa drugs has contributed to the spread of antimicrobial-resistant E. coli, specifically those from animal sources [11].
Therefore, the main objectives of this study are to apply a broader array of virulence and antimicrobial-resistant genes that are known to arise in different pathotypes of E. coli while causing several diseases and to optimize uniplex PCR assays for their detection in milk samples.

Ethical approval
The study was approved by the Animal Ethics Committee for Veterinary Research, Faculty of Veterinary Medicine, South Valley University, Qena, Egypt.

Study period, location, and detection of E. coli in raw milk samples
The study was conducted from February to August 2019. A total of 100 milk samples were collected from different sources (47 samples from dairy farms, 27 samples from retail markets, and 26 samples from farmers' houses) in Qena, Egypt. These samples were collected under aseptic conditions in a clean, sterile 15 mL falcon tube and transferred instantly in an icebox to the bacteriological laboratory in the Department of Microbiology, Faculty of Veterinary Medicine, South Valley University. One milliliter of milk samples were fed in 9 mL buffer peptone water (Oxoid) and incubated at 37°C for 18-24 h. Subsequently, a sterile loop was used to transfer bacteria from the inoculated buffer peptone water and was inoculated on a MacConkey agar plate (Oxoid). Plates were incubated at 37°C for 24 h. The suspected colonies were inoculated in eosin methylene blue (Oxoid). After the procedure, green metallic sheen colonies were selected for biochemical identification using the IMViC reaction and triple sugar iron test [12].

Serology of E. coli isolates
E. coli isolates were serogrouped according to a previous study [13]. The serotyping of E. coli isolates was performed by commercially available kits (E. coli antisera set 1 for O-antigen, Denka Seiken, Japan), which combined 8 polyvalent sera and 43 monovalent sera.

DNA extraction
DNA extraction was performed using the QIAamp DNA Mini kit (Qiagen GmbH, Germany) with changes to the manufacturer's recommendations. Briefly, 10 µL of proteinase K and 200 µL of lysis buffer were added to 200 µL of the bacterial suspension, and this mixture was then incubated at 56°C for 10 min. After incubation, the lysate was mixed with 200 µL of 100% ethanol. The mixture was then rinsed and centrifuged after the manufacturer's recommendations. Nucleic acid was washed with 100 µL of elution buffer accompanied within the kit.

Detection of 16S rRNA gene
The extracted bacterial DNA and controls were amplified with 12.5 µL of Emerald Amp Max PCR Master Mix (Takara, Japan), 1 µL of each primer (20 pmol concentration), 6 µL of DNA template, and 4.5 µL of water to reach a final volume of 25 µL. DNA was amplified (Table-1) in a thermal cycler (Applied Biosystem 2720, California, USA). PCR products were isolated on 1.5% of agarose gel, and DNA amplicons were visualized with ethidium bromide.

The sequence of partial amplified 16S rRNA gene of untypeable E. coli strains
Purified PCR products of partially amplified 16S rRNA of untypeable E. coli strains were sequenced in the forward and reverse directions, as well as in separate reactions, using primers of 16S rRNA, according to the manufacturer's protocol using the following kits: Big Dye Terminator V.3.1 Cycle sequencing kit (Applied Biosystems), the sequencing cycle purification kit, DyeEx 2.0 Spin kit (Qiagen), and Hi-Diionized formamide (Applied Biosystems). DNA sequences were obtained from the Applied Biosystems 3130 (Tokyo, Japan). All the obtained 16S rRNA gene sequences were submitted to the GenBank. The multiple alignment algorithms in MegAlign (Dnastar, Window version 3.12e) were used to perform the sequence alignment of isolates.

Phylogenetic analysis
A phylogenetic tree, which relied on the 16S rRNA gene nucleotide sequence, was a tribute for the untypeable isolates to demonstrate these isolates' identities and reference strains logged in GenBank using MegAlign from the Lasergene package version 7 (Dnastar).

Uniplex PCR
Primers for the following genes hly, eaeA, blaTEM, tetA(A), and tetB were used in a 25 µL reaction containing 12.5 µL of Emerald Amp Max PCR Master Mix (Japan), 1 µL of each primer, 4.5 µL of water, and 6 µL of DNA template. The reaction was conducted in an Applied Biosystem 2720 thermal cycler (Table-1).

Investigation of the PCR products
The uniplex and duplex PCR products were separated by electrophoresis on 1.5% of agarose gel (AppliChem GmbH, Germany) in 1× TBE buffer at room temperature (30ºC) using a gradient of 5 V/cm. For gel analysis, 20 µL of each product was loaded in each gel well. A 100 bp DNA ladder (Qiagen) was used to find out the amplicon sizes. The gel was visualized by a gel documentation system (Alpha Innotech, Biometra, Germany).

Results
We isolated from 10 out of 100 raw milk samples (10%). Four samples were brought from markets, three samples were brought from dairy farms, and three samples were obtained from farmers' houses. The serology results of the 10 E. coli isolates from the raw milk samples, using antisera against the O-antigen, exhibited that one isolate was O111, one isolate was O27, one isolate was O114, two isolates were O125, and the other five isolates were untypeable (Table-2).
The antibiotic susceptibility profile of the 10 E. coli isolates was determined against 12 different antibiotics according to Clinical Laboratory Standards Institute [14]. All 10 E. coli isolates did not show any sensitivity for oxacillin, penicillin, rifampin, and clindamycin. A total of nine isolates were resistant to erythromycin, and six isolates were resistant to trimethoprim. Moreover, five isolates were resistant to trimethoprim-sulfamethoxazole, chloramphenicol, nitrofurantoin, and gentamicin; furthermore, seven isolates did not have any zone of inhibition against tetracycline and four isolates were resistant to nalidixic acid. Four isolates were sensitive to both trimethoprim and trimethoprim-sulfamethoxazole. Three isolates were sensitive to chloramphenicol and nitrofurantoin.
Two isolates were resistant to tetracycline and gentamicin, whereas only one isolate was sensitive to erythromycin. Six isolates cleared intermediary susceptibility toward nalidixic acid, three isolates exhibited intermediate susceptibility to gentamicin, two isolates showed intermediary susceptibility toward chloramphenicol and nitrofurantoin, and only one isolate exhibited intermediary susceptibility toward trimethoprim-sulfamethoxazole and tetracycline (Table-2). PCR results confirmed the presence of E. coli DNA in 10 isolates using housekeeping gene primers (16S rRNA). The presence of different virulence genes (hly, eaeA, stx1, and stx2) and antibiotic-resistant genes (blaTEM, tetA(A), and tetB) was then evaluated in the 10 isolates (Table-3). Available at www.veterinaryworld.org/Vol.14/September-2021/16.pdf  R  S  I  I  I  S  I  R  R  I  R  R  2  Retail market  O27  R  S  S  R  R  S  I  R  R  I  R  R  3  Retail market  Untypeable  R  R  R  S  I  R  S  R  R  I  R  R  4 Dairy farm Untypeable We found that two E. coli isolates (O125) and two untypeable E. coli isolates were positive for the eaeA gene (Figure-1). Only one E. coli isolate (O125) was positive for the hly gene (Figure-2). All 10 E. coli isolates were negative for the presence of stx1 and stx2 genes, except for one untypeable E. coli isolate that was positive for stx2 ( Figure-3). As it pertains to the presence of specific antibiotic-resistant genes, we found that all 10 isolates of E. coli were positive for the blaTEM (Figure-4) and tetA (A) (Figure-5). In addition, four untypeable E. coli isolates were positive for the tetB (Figure-6).
The partially amplified 16S rRNA gene sequencing results confirmed that four untypeable isolates were of E. coli (Table-4). Interestingly, one isolate (number 5) originally classified as E. coli was determined to be a strain of Enterobacter hormaechei (Table-4) after subjecting this isolate's sequence to an NCBI Blast search. The neighbor-joining algorithm was used to generate a phylogenetic analysis using MegAlign from the Lasergene package version 7 (Dnastar). The phylogenetic trees, reconstructed from partially amplified 16S rRNA gene sequences from the five untypeable E. coli isolates, were mainly the same altogether. Nevertheless, a slight variety was noticed  in branching, as shown in Figure-7. The phylogenetic tree identified groups corresponding to species belonging to the same genus. Our isolates exhibited a high degree of similarity with one other (Figure-8), with similarity percentages ranging between 96.5% and 100%. These isolates were registered on GenBank by accession number (Table-4).

Discussion
The phenotypic identification of raw milk samples revealed a concern with maintaining proper hygienic conditions at the sampling sites. Among the 100 milk samples collected, 10 (10%) were contaminated with E. coli. These results are nearly similar to those of Kumar and Prasad [21], who found that 8.14% of the milk samples were contaminated with E. coli. da Silva et al. [22] isolated E. coli from 22.1% of milk samples in Brazil. Disassa et al. [23] isolated E. coli from 33.9% of cow milk samples in Ethiopia. These studies share the same concern as our research that E. coli contamination of raw milk samples is a growing challenge.
The outer layer of E. coli features many lipopolysaccharides distinguished by the O-antigen. The enucleating of the O-antigen from the wall of bacterial cell attenuate bacterial pathogenicity in suggesting the O-antigen, which plays a key role in host-pathogen interactions [24]. Correspondingly, O-antigens caused antigenic diversity among bacteria and besieged to be the biomarkers for the identification of E. coli since the 1940s [25]. Based on the serological analysis targeting the O-antigen of the 10 E. coli isolates sampled from raw milk sources in our study, five isolates were serotyped as O111, O27, O114, and O125, and five strains were untypeable by agglutination. The identified isolates were differentiated according to a previous study [26] as enterotoxigenic E. coli (ETEC) (O27), enterohemorrhagic E. coli (O111), and enteropathogenic E. coli (EPEC) (O114 and O125). Our results are related to the results obtained by the previous studies [27,28]. Several authors have recorded a wide range of E. coli serogroups in cattle [26,29,30]. Pathogencity of EPEC isolates obtained from milk samples represented a potential risk for public health due to their linkage with infant diarrhea [31][32][33]. As stated earlier, in this study, five E. coli isolates were untypeable in the agglutination test. This highlights one of the challenges of serological methods, as discussed by Delannoy et al. [34] who found that the serological agglutination of the O-antigens is difficult, time-consuming, and expensive. Besides, many samples remain untypeable by agglutination; consequently, molecular methods are better alternatives for O-typing [35].
The DNA sequencing of 16S rRNA can aid in the identification of E. coli [36], especially in the cases of phenotypic misidentification as occurred with one isolate from the milk samples in our study. The serological evaluation indicated that one isolate was an untypeable E. coli strain but was later confirmed as an isolate of E. hormaechei (accession no. MN066412). Garbaj et al. [37] found that out of 27 E. coli strains identified using conventional methods, only 11 strains were confirmed to be E. coli using 16S rRNA sequencing. This result suggests the importance of using molecular methods to validate the isolate information obtained from conventional serological techniques.
The additional appraisal of the 16S rRNA sequences of E. coli strains earned from milk samples by the neighbor-joining tree found that there was a high degree of similarity between our isolates (similarity of 95.5-100%), as all of these species originated from the Enterobacteriaceae family [38]. The sequencing of 16S rRNA further assisted in identifying the five untypeable strains by grouping these strains with GenBank-based reference strains. This analysis revealed that E. coli (MN064835) was related to two reference E. coli strains (MN006357 and KY711198). Both strains are uropathogenic E. coli (UPEC) strains [39]. Moreover, E. coli isolates MN066319, MN064861, and MN06483 were found to be related to a reference strain (E. coli AP019675), which is an extended-spectrum β-lactamase E. coli (ESBL) strain [40]. Most of the reference strains were isolated from patients and environmental samples (e.g., water and soil) [39,40]. Thus, finding similar E. coli isolates in raw milk are unsurprising. Nevertheless, this observation highlights the importance of performing a microbiological analysis of milk to verify that there is no public health risk to consume the milk. This is particularly important to curb the outbreaks of food-borne illnesses, which were formerly linked to the consumption of milk and dairy products that had been contaminated with pathogenic bacteria or their toxins [41]. STEC (Stx) strains were identified from cattle farms more than 3 decades ago. Studies implemented in several regions have shown that 10-80% of cattle may harbor STEC [29,42,43]. In this study, the stx2 gene was found in only one isolate (untypeable E. coli), whereas the stx1 gene was not harbored by any of the 10 isolates. Our result matches the previous studies [28,44] but differs from the results obtained by other studies [45,46]. Almost the equal distribution of stx1 and stx2 was noted by Zschock et al. [47] in the fecal samples of E. coli obtained from cows. This may be attributed to the repeated turnover of serotypes of E. coli, with many being isolated only sporadically from a herd [48]. Furthermore, Jenkins et al. [49] confirmed that there is a dissimilarity in the percentage of stx genes detected, according to the distribution of STEC.
As it pertains to pathogenesis, E. coli utilizes various virulence factors to colonize and infect host cells. One factor is intimin (encoded by the eaeA gene), which plays an important role in the pathogenesis of attaching-and-effacing lesion, which facilitates the colonization of host tissues [50]. In our study, the eaeA gene was detected in four E. coli isolates (40%). Two isolates were EPEC of serotype O125 and O114, and the other two isolates were untypeable E. coli. Many authors have previously revealed that EPEC strains harbor the eaeA gene [51,52]. The eaeA gene was detected in 4% of E.
A second key virulence factor used by E. coli to promote pathogenesis is the liberation of hemolysin, a pore-forming cytolysin (encoded by the hlyCABD gene). In our study, we illustrated that only one isolate belonging to EPEC harbored the hly gene. EPEC strains, UPEC strains, ETEC strains, and STEC strains carried the hly gene, which is encoded on a plasmid reported by Burgos [54].
After evaluating the existence of genes encoding key virulence factors, we next assessed the antibiotic resistance profile of the 10 E. coli isolates obtained from the raw milk samples. The propagation of antibacterial resistance is the leading cause of many health problems worldwide. In this study, we found that a high percentage (100%) of E. coli isolates were resistant to penicillin, oxytetracycline, doxycycline, rifampicin, trimethoprim/sulfamethoxazole, and tetracycline. Multidrug resistance (resistance to more than 3 classes of antibiotics) was also observed in several of our E. coli isolates. The isolation of multidrug-resistant E. coli isolates is not surprising as this phenotype has been reported for E. coli in other studies [55,56]. In addition, two studies have found that E. coli isolates obtained from milk samples were resistant to antibiotics, mainly to β-lactam antibiotics [57,58].
Bacterial resistance to antibiotics may occur by an impulsive mutation in the target gene(s) due to using antibiotics in therapy or enhancing growth in animals. These antibiotic-resistant bacteria can be spread in the environment when infected animals defecate, thus contaminating soil and agricultural products. If the raw products are consumed, then the resistant bacteria may consequently colonize and infect humans [59]. Among the bacterial isolates obtained from the raw milk samples in this study, we found that genes conferring resistance to tetracycline (tetA(A) and tetB) were the most prevalent. This finding is concerning because tetracycline is one of the most important antibiotics used in Egypt [60,61].
In addition to resistance to tetracycline, the presence of ESBL's confers resistance to many antibiotics, such as aztreonam, related oxyimino-β-lactams, cephalosporins, and penicillin. There is a wide variety of ESBLs, including plasmid-mediated resistant to ampicillin, oxacillin, and cefotaxime [62]. The mechanism of drug efflux pumps and the existence of the outer membrane further obstruct the entry of antibiotics in Gram-negative bacterium such as E. coli. Focusing on ESBLs, in this study, the blaTEM gene was detected in all E. coli isolates. The prevalence of blaTEM found in our E. coli isolates matches results obtained by the previous studies [63,64], who found that most of their E. coli isolate harbored blaTEM.
Our study demonstrated a relationship between the resistance to drugs and the presence of virulence factors. The attainment of antibiotic resistance and virulence factors in E. coli can arise through gene transfer in the environment or the human and animal guts. The genes can be found either on the same plasmid or separately on bacterial chromosomal DNA and plasmids. These genes might be coselected in response to antibiotic pressure, hence aggravating the hazard imposed by E. coli on food safety [65,66].

Conclusion
According to this study's findings, E. coli is considered one of the most common food-borne pathogens that can be transmitted to humans by drinking raw milk. Some strains of E. coli harbor some virulent genes eae, hyl stx2 genes, antibiotic-resistant genes blaTEM, tetA(A), and tetB. Furthermore, they showed marked resistance to several antibiotics commonly used in animals and humans in Egypt (MDR). Therefore, much more attention should be paid to hygienic measures during the milking process in dairy farms and the prudent use of antibiotics supported by antibiogram tests before drug administration in dairy farms.

Authors' Contributions
WY and HMAM: Designed the study, drafted and critically revised the manuscript, collected samples, and did laboratory works. SH: Revised the manuscript and interpretation of the data. All authors read and approved the final manuscript.