Identification, Shiga toxin subtypes and prevalence of minor serogroups of Shiga toxin-producing Escherichia coli in feedlot cattle feces

Shiga toxin-producing Escherichia coli (STEC) are foodborne pathogens that cause illnesses in humans ranging from mild to hemorrhagic enteritis with complications of hemolytic uremic syndrome and even death. Cattle are a major reservoir of STEC, which reside in the hindgut and are shed in the feces, a major source of food and water contaminations. Seven serogroups, O26, O45, O103, O111, O121, O145 and O157, called ‘top-7’, are responsible for the majority of human STEC infections in North America. Additionally, 151 serogroups of E. coli are known to carry Shiga toxin genes (stx). Not much is known about fecal shedding and prevalence and virulence potential of STEC other than the top-7. Our primary objectives were to identify serogroups of STEC strains, other than the top-7, isolated from cattle feces and subtype stx genes to assess their virulence potential. Additional objective was to develop and validate a novel multiplex PCR assay to detect and determine prevalence of six serogroups, O2, O74, O109, O131, O168, and O171, in cattle feces. A total of 351 strains, positive for stx gene and negative for the top-7 serogroups, isolated from feedlot cattle feces were used in the study. Of the 351 strains, 291 belonged to 16 serogroups and 60 could not be serogrouped. Among the 351 strains, 63 (17.9%) carried stx1 gene and 300 (82.1%) carried stx2, including 12 strains positive for both. The majority of the stx1 and stx2 were of stx1a (47/63; 74.6%) and stx2a subtypes (234/300; 78%), respectively, which are often associated with human infections. A novel multiplex PCR assay developed and validated to detect six serogroups, O2, O74, O109, O131, O168, and O171, which accounted for 86.9% of the STEC strains identified, was utilized to determine their prevalence in fecal samples (n = 576) collected from a commercial feedlot. Four serogroups, O2, O109, O168, and O171 were identified as the dominant serogroups prevalent in cattle feces. In conclusion, cattle shed in the feces a number of STEC serogroups, other than the top-7, and the majority of the strains isolated possessed stx2, particularly of the subtype 2a, suggesting their potential risk to cause human infections.

Prevalence of O2, O74, O109, O131, O168, and O171 serogroups in feedlot cattle feces. A total of 576 fecal samples collected from cattle in a commercial feedlot were subjected to the mPCR assay. Modeladjusted mean prevalence and 95% confidence intervals of test positive samples for O serogroups and virulence genes are shown in Table 5. Of the six serogroups, O109 (91.6%), O171 (87.5%), O168 (79.5%), and O2 (59.5%) were the most predominant. A high proportion of the samples positive for the serogroups tested positive for the three virulence genes, stx1, stx2 and eae. The majority of the fecal samples (80.5%) tested positive for 3 to 5 serogroups of the six non-top-7 STEC (Fig. 3).

Discussion
Serogrouping of E. coli is based on the chemical composition of the O antigen of the lipopolysaccharide located in the outer membrane of the cell envelope 30,31 . A total of 187 serogroups of E. coli have been identified 29 . Shiga toxin-producing E. coli (STEC) constitute a major pathotype and includes as many as 158 serogroups 16,[19][20][21][22][23] . A number of studies have reported on the isolation and prevalence of the O157 and the top-6 non-O157 serogroups in cattle feces and other sample matrices by culture method 24,26,[32][33][34][35][36] . Serogroup identification by agglutination reactions carried out in microtiter plates with a panel of antisera generated by immunization of rabbits with Table 4. Comparison of serogroup identification by PCR and serology of Shiga toxin gene-positive Escherichia coli strains (n = 351) isolated from feedlot cattle feces. a O8 strain was positive for O2 by PCR with primers from Iguchi et al. 38 and DebRoy et al. 29,37 . b Strains O2 (1), O152 (1), and O156 (26) were negative by PCR for O2, O152 and O156 with primers from Iguchi et al. 38 and DebRoy et al. 29,37 . c Strain O156 (1) was positive for O109 by PCR with primers for O109 from Iguchi et al. 38 and DebRoy et al. 29,37 . d Strains O152 (30) and O11 (1) were negative by PCR for O152 and O11 with primers from DebRoy et al. 29,37 .
No. of strains Serogrouping by PCR (no. of strains) Serogrouping by serology (no. of strains)  29 . Two serogroups, O14 and O57, were not included because neither contain O-antigen biosynthesis gene clusters 49,50 . We have developed and validated 14 sets of mPCR assays, each targeting seven to 12 serogroups, to detect 137 STEC serogroups that have been detected in cattle feces 23 .
Using the 14 sets of mPCR assays, serogroups of 291 strains out of 351 (82.9%) were identified as belonging to 16 serogroups, and the remaining 60 (17.1%) were unidentified. However, of the 60 strains unidentified by our PCR, serology identified 31 strains as O11 (1) and O152 (30) and the remaining 29 as untypeable. The strains that were identified differently by serology (O2, O11, O131, O109, O152, and O156) also tested negative by PCR with primers described by DebRoy et al 37 and Iguchi et al 38 . The discrepancy between PCR and serology has been www.nature.com/scientificreports/ reported previously 50 . In certain serogroups with similar nucleotide sequences, serology may not show any cross reactivity, which could be due to posttranslational modifications of the proteins resulting in changed epitopes in antigens 50,51 . In silico serogrouping based on assembled whole genome sequence (WGS) 51 or raw short read WGS data 52 may identify the serogroup and likely provide reasons as to why PCR was not able to identify them. Based on both methods of serogrouping, only 19 STEC serogroups were identified; 16 by PCR and 3 additional serogroups by serology. It is important to recognize that although as many 151 non-top-7 serogroups of STEC have been isolated from cattle feces, only 19 STEC serogroups were obtained from fecal samples collected from 9 feedlots. Because the isolates were obtained from immunomagnetic beads (IMS) specific for the top-7 STEC, the serogroups may not represent the true distribution of STEC serogroups in cattle feces.
Interestingly, the majority of the non-top-7 STEC strains possessed stx2 (300/351; 85.5%), which is in contrast to what has been observed for the six serogroups of non-O157 STEC, which primarily possess stx1 13,34,53,54 . The predominance of stx2 suggests the potential risk of the non-top-7 STEC to cause human infections. Shiga toxin 2 was about 400 times more toxic in a mouse infection model 5 and was more commonly associated with complications of human STEC illnesses than Shiga toxin 1 4, 55, 56 . Shiga toxin 2, particularly in association with intimin, results in a higher risk for severe infections 57 , although Shiga toxin 2 without intimin can cause severe infection   58 . The EHEC pathotype, a subset of STEC, was once considered to be associated more often with severe STEC infections. In a scientific opinion agreed upon by the European Food Safety Authority Panel, the EHEC terminology is considered obsolete and the recommendation was to use STEC for all stx-positive strains 8 . The predominance of the subtypes stx1a and stx2a in the non-top-7 STEC identified in this study is similar to previous reports of their dominance among O157 and top-6 non-O157 strains of human clinical origin 56, 59, 60 . The stx1a is often produced by LEE-positive strains of STEC and have the potential to cause severe infections 13 . Epidemiological data from human infections indicate a stronger association of stx2a-and stx2d-positive strains with severe hemorrhagic enteritis, including HUS 56, 61, 62 . These two subtypes were more cytotoxic than stx2b and stx2c in an in vitro potency assay 63 .
Of the total 16 serogroups identified in the study, seven serogroups, O2, O74, O104, O109, O131, O168, and O171, accounted for 76.9% (270/351) and 92.8% (270/291) of the total and serogroup-identified strains, respectively. Because the 351 strains used in the study were from immunomagnetic beads that targeted the six non-O157 serogroups, the dominance of a few serogroups in isolated strains is not indicative of their prevalence in cattle feces. Also, IMS beads are not available for these serogroups, which rule out culture method to selectively isolate, identify and determine their prevalence. Therefore, a novel mPCR assay of the dominant serogroups was designed, validated, and utilized to determine the prevalence in feces of feedlot cattle. The PCR assay did not include O104 because we had previously developed a mPCR assay for the top-7 STEC and O104 64 and determined the prevalence of O104 serogroups and characterized the isolated serotypes in cattle feces 65,66 . The reason for including O104 with the top-7 STEC at the time was because O104:H4, a hybrid pathotype of STEC and enteroaggregative E. coli, was involved in a major foodborne outbreak in Germany in 2011 67 . Of the six serogroups, prevalence of four serogroups, O2 (59.2%), O109 (91%), O171 (86.5%), and O168 (78.1%), were higher than the other two serogroups. It should be noted that this is only a preliminary finding, based on PCR assay from fecal samples collected from one feedlot and additional studies, possibly including culture methods, are needed. In the majority of the fecal samples (80.8%), multiple serogroups (three or more) were present, likely because of the high prevalence of three serogroups (> 78% of O168, O171, and O109). This is in contrast to the prevalence of the six major non-O157 serogroups, in which the majority of the samples (68.1%) were positive for one or two serogroups 26 . The four serogroups, O2, O109, O171 and O168, have been frequently isolated from feces of healthy cattle [68][69][70][71] . However, this is the first study that provides prevalence estimates of these six groups in feces from commercial feedlot cattle (with natural shedding).
At least 130 of the 151 serogroups of non-top-7 STEC have been reported to be associated with clinical cases of diarrhea, and a few serogroups and serotypes have been associated with severe forms of infections, including complication of HUS 16,20,22,34,[72][73][74][75][76][77][78] . Certain serogroups, such as O2, O8, and O113, and specifically certain serotypes within these serogroups, have been reported to cause outbreaks associated with consumption of contaminated beef in the US, European countries, and Australia 20,77,79 . Serogroup O113 (mostly the H21 serotype) has been associated with severe cases of hemorrhagic colitis and HUS in the US and other countries [79][80][81] .
In conclusion, cattle harbor and shed in feces a number of serogroups of STEC other than the top-7 responsible for the majority of foodborne STEC infections. The majority of the non-top-7 strains isolated and serogrouped possessed stx2 and were of the subtype stx2a, suggesting their potential to cause severe infections in humans. Although a majority of the non-top-7 STEC have been shown to cause sporadic infections, a few serogroups, notably O2, O8, O91, and O113 have been implicated in outbreaks and serious infections. The fecal prevalence of a few serogroups, namely O2, O109, O168, and O171, was high in feedlot cattle. The importance of these nontop-7 STEC as foodborne pathogens in humans is not known. Not much is known about the prevalence of these STEC serogroups on cattle hides and carcass surfaces and in beef products and other food matrices in the USA, largely because detection strategies have not been developed and validated. Our study provides information on the detection and prevalence of major serogroups of non-top-7 STEC in cattle. Because the chromogenic colonies of the six non-O157 serogroups on MP medium were phenotypically indistinguishable, a pool of 10 randomly picked chromogenic colonies from the plate inoculated with IMS beads was prepared and tested by mPCR assay targeting seven serogroups of STEC 45 . If positive for any of the seven serogroups, then each of the 10 colonies was tested individually by a mPCR assay targeting the seven serogroups (O26, O45, O103, O111, O121, O145, and O157) and three virulence genes (stx1, stx2, and eae) 26,39 . Isolates positive for stx1 and/ or stx2 and negative for any of the seven serogroups were considered as STEC other than the top-7 (non-top-7). The 117 isolates from the 2013 study were from a total of 576 fecal samples collected from 24 pens in a single commercial feedlot. The 234 isolates from the 2014 study were from a total of 1,886 fecal samples collected from 64 pens in eight commercial feedlots located in two major U. S. beef cattle states. The isolates stored in cryobeads (CryoCare, Key Scientific Products, Round Rock, TX) at -80° C were streaked onto blood agar plates (BAP; Remel, Lenexa, KS) and incubated overnight at 37° C. A single colony of each strain was suspended in 50 µl distilled water, boiled for 10 min, centrifuged, and the supernatant was used as the template in the 14 sets of mPCR assays designed to detect 137 serogroups of non-top-7 STEC 23 . The serological tests for O-group determination, based on agglutination 31 , were conducted at the E. coli Reference Center (Pennsylvania State University).
Subtyping of stx genes. The subtypes of stx1 and stx2 genes of the 351 STEC strains were determined according to the protocol described by Shridhar et al 54 . Briefly, a colony from BAP was suspended in distilled water, boiled, centrifuged and the lysate was used to amplify stx1 and stx2 genes by touchdown PCR. PCR products were purified using a QIAquick PCR purification kit (Qiagen, Valencia, CA) and shipped to Genewiz, Inc., (South Plainfield, NJ) for nucleotide sequencing. The chromatogram data of each sequence was individually analyzed for conflicts and secondary peaks, and consensus sequences were produced using the CLC Main Workbench software (Qiagen, Valencia, CA). The nucleotide sequences were conceptually translated to amino acid sequences and Shiga toxin subtypes were determined based on the amino acid motifs that define each stx subtype 7 .
Development and validation of a mPCR assay targeting O2, O74, O109, O131, O168, and O171 serogroups.. Primers design. The serogroup-specific wzx gene, which encodes for the transmembrane lipid transporter enzyme or flippase, required for the O-polysaccharide export, was targeted in this assay. The primers were designed based on the nucleotide sequences of the target gene for each of the six serogroups obtained from the GenBank database. The sequences for each serogroup were aligned using ClustalX version 2.0 82 , and the primers that amplify the targets with distinct amplicon sizes that can be differentiated by capillary gel electrophoresis were chosen for the study.
Assay running conditions. The reaction consisted of 10 μl of BioRad iQ multiplex powermix, 1 µl of six pairs of primer mix (8 pM/µl for each primer), 2 µl of template, and 7 µl of water (total reaction volume = 20 µl). The PCR running conditions consisted of an initial denaturation at 94° C for 5 min followed by 30 cycles of denaturation at 94° C for 30 s, annealing at 72° C for 30 s, extension at 68° C for 75 s, and a final extension step at 68° C for 7 min. The primer sequences and amplicon sizes are provided in Table 6.
Applicability of the six-plex PCR assay to detect O2, O74, O109, O131, O168, and O171 serogroups in cattle feces. Fecal samples collected from a commercial feedlot in the 2013 study to determine the prevalence of top-7 STEC serogroups 26,27 were used. Fecal samples were collected weekly for 12 weeks in the summer (June-August). Each week, 24 pen-floor fecal samples were collected from each of two pens of finishing cattle a day before transport of cattle for slaughter. A total of 576 fecal samples from 24 pens were collected. Fecal samples were enriched in E. coli broth 45 by incubating at 40 °C for 6 h and stored at − 80 °C. Enriched fecal samples were thawed and the DNA extracted and purified (as described above) was used as the template for the six-plex PCR assay.
Statistical analysis. Descriptive statistics (frequency tables [number and %]) were computed to describe the cumulative fecal prevalence for serogroup, STEC, and EHEC O groups. A sample was considered serogroup positive, if it tested positive for the serogroup only (disregarding presence or absence of virulence genes). Modeladjusted cumulative sample-level prevalence estimates and their 95% confidence intervals of test positive samples for O serogroups and virulence genes were estimated from model intercepts using generalized linear mixed models. Outcomes consisted of: 1) sample-level serogroup prevalence of O2, O74, O109, O131, O168, and O171 groups, 2) sample-level STEC prevalence of O2, O74, O109, O131, O168, and O171 (samples positive for serogroup and stx1 and or stx2), and 3) sample-level EHEC prevalence of O2, O74, O109, O131, O168, and O171 (samples positive for serogroup and stx1 and or stx2 and eae). Statistical models were fitted in Proc Glimmix (SAS 9.4; SAS Institute Inc., Cary, NC) using a binary distribution, logit link, residual pseudo-likelihood estimation, Kenward-Rogers degrees of freedom approximation, and random intercepts for pen and week to account for the clustering of pens nested within sampling week.