Molecular profiling and antimicrobial resistance of Shiga toxin-producing Escherichia coli O26, O45, O103, O121, O145 and O157 isolates from cattle on cow-calf operations in South Africa

In this study, 140 cattle STEC isolates belonging to serogroups O157, O26, O145, O121, O103 and O45 were characterized for 38 virulence-associated genes, antimicrobial resistance profiles and genotyped by PFGE. The majority of isolates carried both stx1 and stx2 concurrently, stx2c, and stx2d; plasmid-encoded genes ehxA, espP, subA and saa but lacked katP and etpD and eaeA. Possession of eaeA was significantly associated with the presence of nle genes, katP, etpD, ureC and terC. However, saa and subA, stx1c and stx1d were only detected in eaeA negative isolates. A complete OI-122 and most non-LEE effector genes were detected in only two eaeA positive serotypes, including STEC O157:H7 and O103:H2. The eaeA gene was detected in STEC serotypes that are commonly implicated in severe humans disease and outbreaks including STEC O157:H7, STEC O145:H28 and O103:H2. PFGE revealed that the isolates were highly diverse with very low rates of antimicrobial resistance. In conclusion, only a small number of cattle STEC serotypes that possessed eaeA, had the highest number of virulence-associated genes, indicative of their high virulence. Further characterization of STEC O157:H7, STEC O145:H28 and O103:H2 using whole genome sequencing will be needed to fully understand their virulence potential for humans.

cycle, the supernatant was discarded. The pellet was re-suspended in 500 µL of sterile distilled water and boiled (heating block) for 20 min and cooled on ice for 10 min. Finally, the suspension was centrifuged at 12,000 rpm for 5 min, and DNA was stored at −20 °C for further use in PCR reactions.
Detection of stx1, stx2, eaeA and ehxA genes by PCR. A multiplex polymerase chain reaction (mPCR) was carried out to detect the stx1, stx2, eaeA and ehxA genes using previously described primers (Table 1) and cycling conditions 42 . Briefly, each PCR reaction (25 µL) contained 2.5 μL of 10X Thermopol reaction buffer, 2.0 μl of 2.5 mM dNTPs, 0.25 μl of 100 mM MgCl 2 , 0.3 µM of each primer, 1U of Taq DNA Polymerase (New England BioLabs ® Inc.) and 5 μl of DNA template. Sterile distilled water was used to top up the reaction volume to 25 µL. STEC O157:H7 EDL933 and sterile water were used as positive and negative controls, respectively. All PCR reagents were supplied by New England BioLabs (NEB, USA) except for primers, which were obtained from Inqaba Biotec (South Africa). PCR reactions were carried out in a C1000 Touch TM (Bio-Rad, USA) or Veriti 96-well Thermal Cycler (Applied Biosystems, USA). PCR amplicons were electrophoresed in 2% (w/v) agarose gels in TAE (Tris-acetate-ethylenediamine tetraacetic acid) buffer, stained with ethidium bromide (0.05 mg/μl) and visualized under ultraviolet (UV) light with a Gel Doc system (Bio-Rad, USA).
Pulsed-field gel electrophoresis. To subtype STEC isolates, DNA was extracted, digested with the XbaI restriction enzyme and subjected to PFGE according to the CDC/PulseNet protocol (https://www.cdc.gov/ pulsenet/pdf/ecoli-shigella-salmonella-pfge-protocol-508c.pdf). Salmonella enterica serotype Braenderup (strain H9812; American Type Culture Collection, BAA-664) DNA was used as the molecular weight marker. PFGE fingerprints were analyzed for similarity, and a dendrogram was generated by Bionumerics software, version 6.6 (Applied Maths, Sint Martens-Latem, Belgium) with the Dice similarity indices (complete linkage; optimization, 1.5%; position tolerance, 1.5%) and the unweighted-pair group method with arithmetic means (UPGMA). Statistical analysis. Descriptive statistical analysis was performed using the statistical package for social sciences (SPSS) software version 21 (IBM ® SPSS ® Statistics 21). Fisher's exact test was used to determine if there were statistically significant differences and associations between gene proportions. P values of < 0.05 were considered statistically significant.
The following proportions were observed for non-LEE effector (nle) genes: ent/espL2, 34 (Table S2 and Fig. 2). More than 10 nle genes were observed in 12.9% of the isolates, which were mainly eaeA positive, eight to nine nle genes were present in 6.4%, and one to seven nle genes were detected in 30.7% of the isolates. The remaining 50% of isolates did not carry a non-LEE effector gene.
Antimicrobial resistance. Of the 140 STEC isolates, 97.9% were susceptible to all the 15 antimicrobials.
Only 2.1% of STEC isolates were antimicrobial resistant, including one STEC O26:H11 isolate which was resistant to tetracycline, one STEC O26:H4 which was resistant to ampicillin and tetracycline and one STEC O45:H21 isolate which was resistant ampicillin, tetracycline and cephalothin.
Pulsed field gel electrophoresis. PFGE was conducted to investigate genetic relatedness among the STEC isolates. Six dendograms (Figs 3-6) that displayed relationships among individual serogroups were generated. All the 140 isolates yielded 101 distinct pulsotypes, including 43 for STEC O26, 27 for STEC O45, 2 for STEC O103, 6 for STEC O121, 7 for STEC O157, and 16 for O145 suggesting a high diversity (Dice similarity index < 70%) among STEC isolates in different serogroups. Most of the pulsotypes represented single isolates. The 39 isolates which shared identical PFGE profiles (100% similarity) in different serogroups either belonged to the same serotype or were recovered from the same animal or farm.    www.nature.com/scientificreports www.nature.com/scientificreports/ remain scanty. Cattle are a major reservoir of STEC. Molecular risk assessment studies on STEC isolates from cattle have contributed to a better understanding of the virulence potential cattle STEC present to humans and made it possible to differentiate low virulence from highly virulent STEC isolates. In this study, 140 STEC isolates from apparently healthy cattle on five cow-calf operations in South Africa were characterized for a panel 38 virulence-associated genes, antimicrobial resistance and PFGE profiles. The collection of isolates under study was a subset of a larger collection of STEC strains belonging 33 serotypes associated with serogroups O157, O45, O103, O121, O26 and O145.
The majority of STEC isolates carried both stx1 and stx2 concurrently. Almost all stx2-positive isolates (95.7%) harbored stx2a, stx2c and stx2d. Only a small fraction of isolates carried stx1 only. The stx2d subtype identified in this study was the stx2d-activatable variant 45 . The widespread distribution of stx2a, stx2c and stx2d subtypes in cattle isolates is in agreement with previous studies which have reported high rates of stx2a, stx2c and stx2d subtypes among STEC isolates from cattle in comparison to different stx1 subtypes [46][47][48][49][50] . STEC isolates that carry stx2 are more virulent compared to strains that possess stx1 alone or both stx1 and stx2 concurrently, and are frequently incriminated in outbreaks and severe human disease manifestations such as HC and HUS 45,51,52 . Furthermore, some studies have shown that Stx2, Stx2c and Stx2d subtypes are more potent than Stx1 45,53 . In addition, Rasooly and Do 54 reported that Stx2 was heat stable and not inactivated at currently approved pasteurization temperatures, making Stx2-producing isolates more likely to be implicated in human STEC disease outbreaks involving pasteurized cattle dairy products.
Less than 50% of isolates carried stx1c and stx1d subtypes. The stx1c subtype was significantly more frequent than stx1d, in agreement with a number of studies which have reported that STEC isolates of cattle origin are mainly stx1c positive 47,49 . STEC isolates that possess stx1c and/or stx1d subtypes have been mostly implicated in asymptomatic or mild diarrhea in humans 52,55,56 . However, some reports have implicated stx1c positive isolates in cases of human disease showing bloody diarrhea [56][57][58] . Interestingly, STEC O45:H2 and STEC O45:H11 isolates possessed stx1c, stx1d, stx2, stx2c, and stx2d concurrently. While STEC O45:H2 is a recognized enterohemorrhagic E. coli 59 , so far, there are no reports that have associated STEC O45:H11 isolates with severe human disease. Although STEC O45:H11 is not a recognised enterohemorrhagic E. coli, it is possible that the presence of numerous stx variants in this STEC serotype may be indicative of high virulence, assuming that all toxin encoding genes are maximally expressed during STEC infection in humans.
The majority of STEC isolates lacked eaeA, consistent with previous reports on cattle STEC, which have shown that only a subset of cattle STEC are eaeA positive 46,48,50 . The eaeA gene was present in seropathotypes A and B STEC strains (STEC O157:H7, STEC O103:H2, STEC O26:H2, STEC O145:H28) that were also stx2c and/or stx2d positive but lacked stx1c and stx1d genes. Possession of stx2 and eaeA genes by a STEC strain is indicative of highly virulent STEC strains (EFSA, 2013). STEC seropathotypes A and B strains are highly pathogenic for humans, and commonly implicated in outbreaks and severe disease, including HC or HUS 18,60 worldwide.
Karmali et al. 18 suggested that possession of OI-122 marker genes pagC, sen, efa1 (Z4332), efa1 (Z4333) is indicative of a complete OI-122. A complete OI-122 was observed in only 7.1%. that were mostly stx-2/eaeA-positive. Isolates which possessed the full complement of OI-122 marker genes belonged to serotypes which are commonly associated with STEC disease outbreaks, including HUS in humans (STEC O157:H7 and STEC O103:H2). Reports that have documented a complete OI-122 among clinically relevant isolates, including STEC O157:H7 and STEC O103:H2 have suggested that the presence of a complete OI-122 and stx2 in eaeA-positive strains is indicative of highly virulent STEC strains 18,68,71,72 .
Most STEC isolates that had an incomplete OI-122 were seropathotypes B and C STEC strains that are usually incriminated in mild or uncomplicated diarrhea (STEC O26:H2, O26:H8, O26:H21, O103:H21, O45:H2) 18 . However, isolates that were negative for all OI-122 markers were mainly seropathotypes D or E strains that are very rare in human disease or have never been incriminated in human disease 18 .
Fifty percent (50%) of isolates did not carry any nle-encoding genes. Isolates which did not posess any nle-encoding genes belonged to serotypes that have been incriminated in mild or uncomplicated diarrhea (STEC O26:H2, STEC O26:H8, STEC O26:H21, STEC O121:H8), hemorrhagic colitis (STEC O26:H7, STEC O145:H7) and HUS (STEC O26:H11) in humans 59,60 , and serotypes that have never been associated with human illness 2,78,79 . The lack of nle-encoding genes in STEC serotypes that have been previously implicated in mild to severe disease in humans, suggests that the capacity of these strains to cause disease in humans may not be dependent upon currently known non-LEE effectors. However, the absence of known non-LEE effector genes in STEC isolates that have never been implicated in human disease may also explain why these isolates have never been incriminated in human disease.
Antimicrobial resistance profiling showed that almost all (97.9%) the STEC isolates were susceptible to all the 15 tested antimicrobials except for three STEC isolates that were antimicrobial resistant. The three resistant isolates belonged to STEC O26:H11 (tetracycline), STEC O26:H4 (tetracycline and ampicillin) and STEC O45:H21 (amoxicillin-clavulanic acid and cephalothin). Similar findings were made by Dong et al. 49 , who also reported resistance to ampicillin and tetracycline among cattle STEC isolates. However, higher antimicrobial resistance levels to tetracycline, ampicillin, cephalothin and amoxicillin-clavulanic acid have been previously reported in a number of studies in STEC isolates 35,36,80 . The very low antimicrobial resistance rates observed in this study suggest that the selection pressure exerted on cattle farms from which the STEC isolates were recovered is negligible. Cattle on cow-calf operations in South Africa graze on pastureland all year round and are not supplemented with feed containing antimicrobial promoters that usually exert selective pressure on intestinal flora such as STEC and facilitate proliferation and development of antimicrobial resistant strains.
PFGE revealed that the STEC isolates under study were highly diverse and only a few isolates had identical fingerprints in individual serogroups. Isolates with identical fingerprints either belonged to the same serotype or were recovered from the same animal or farm. The high diversity observed among the STEC isolates under study is reflection of the high genetic flow occurring among STEC isolates through gene acquisition, shuffling and loss, particularly genes that are encoded on mobile genetic elements including plasmids, bacteriophages and pathogenicity islands.
In conclusion, the majority of STEC isolates were stx1, stx2a, stx2c and stx2d positive but lacked eaeA. Plasmid-encoded genes (hlyA, saa, subA and espP) were detected in most of the isolates but katP and etpD genes were only observed in a very small number of isolates that were also eaeA-positive. A complete plasmid, (ehxA, etpD, katP and espP) was observed in STEC O157:H7 isolates mainly. O island and nle marker genes were absent in most isolates, except for OI-43/48-associated genes (terC and iha), which were prevalent in more than 80% of isolates. STEC O157:H7, STEC O145:H28 and STEC O103:H2 and some STEC O26:H2 isolates possessed the highest number of virulence-associated genes. These serotypes which are frequently implicated in severe STEC disease in humans carried nle marker genes, such as nleB, nleE, ent/espL2, nleG2-3, nleG5-2, nleG9, nleG2-1 and nleB2, which are considered a "hallmark" of highly virulent STEC strains 19 .
To our knowledge, this is the first detailed characterization of a large number of cattle STEC isolates from South Africa. This study provides much needed data on the molecular characteristics of STEC serotypes from beef cattle in South Africa. Further studies using whole-genome sequencing (WGS) will be needed to fully assess the virulence potential of these cattle STEC isolates for humans.