Shiga toxin producing Escherichia coli (STEC) are major foodborne pathogens responsible for human illnesses, characterized by non-bloody to bloody diarrhea, sometimes leading to complications of hemolytic uremic syndrome (HUS), particularly in children (Gyles, 2007). Escherichia coli O157:H7 is the major serotype responsible for many of the STEC illness outbreaks in humans. However, there is increasing incidence of outbreaks associated with non-O157 STEC in recent years, particularly O26, O45, O103, O111, O121, and O145, referred to as top six non-O157 STEC. According to FoodNet sites, incidence of top six non-O157 STEC infections increased from 0.12 per 100,000 population in 2,000 to 0.95 per 100,000 population in 2010 (Gould et al., 2013). Non-O157 STEC associated illnesses range from cases of sporadic to major outbreaks, and clinically, from mild watery diarrhea to life threatening complications of HUS, similar to STEC O157 infections (Johnson et al., 2006). Cattle are a major reservoir of O157 and non-O157 STEC, which harbor the organisms in the hindgut and shed in the feces. Consumption of water, beef and fresh produce contaminated with cattle feces leads to human illnesses. In addition to O157 and the six top non-O157, cattle do harbor and shed in the feces a number of other serogroups of STEC (Bettelheim, 2007; Hussein, 2007).

Shiga toxins (Stx) are the major virulence factors of STEC. Shiga toxins (Stx) belong to the AB₅ family of protein toxins, with an enzymatically active A moiety and a B moiety involved in binding to the host cell receptor. The A subunit is responsible for the cleavage of N-glycosidic bond in the 28 s rRNA of 60 s ribosomal subunit, which leads to cytotoxicity (Endo et al., 1988; Fraser et al., 1994). The two antigenically distinct Stx types, Stx1 and Stx2, encoded by stx1 and stx2 genes, share approximately 56% amino acid sequence similarity (Strockbine et al., 1986; Weinstein et al., 1988). Although Stx1 and Stx2 are structurally similar, they differ in cellular distribution and cytotoxicity. Shiga toxin 1 is located in the periplasmic space of the bacterial cell, whereas Stx2 is in the extracellular fraction (Shimizu et al., 2009). Basu et al. (2016) have shown that A₁ subunit of Stx2 has a higher affinity for ribosomes and a higher catalytic activity compared to A₁ subunit of Stx1 (Basu et al., 2016), which makes Stx2 more cytotoxic than Stx1. Shiga toxin 2 is reported to be 400 times more toxic than Stx1 in a murine infection model (Tesh et al., 1993). Shiga toxin 2 is more commonly associated with complications of HUS than Stx1 (Ethelberg et al., 2004; Brooks et al., 2005). Variants exist within stx1 (stx1a, stx1c, and stx1d) and stx2 (stx2a, stx2b, stx2c, stx2d, stx2e, stx2f, and stx2g) families based on differences in amino acid compositions and in the degree of cytotoxicity. The outcome of human illness associated with STEC strains has been shown to be influenced by Stx subtypes (Friedrich et al., 2002). Shiga toxin 2a and Stx2c are the major subtypes produced by E. coli O157:H7 strains associated with HUS in humans (Persson et al., 2007). Therefore, identifying the subtypes of (Stx) is important to assess the potential risk for human illnesses associated with STEC infections. Subtyping method based on restriction fragment length polymorphism of PCR products (PCR-RFLP) has been developed to identify stx1 (stx1a, stx1c, and stx1d) and stx2 (stx2, stx2v-ha, stx2v-hb, stx2g, stx2-NV206, and stx2-EC1586) genotypes at the nucleotide level based on their unique restriction patterns (Gobius et al., 2003; Beutin et al., 2007). However, the demerits of this method are the lack of consistency in the nomenclature, and misinterpretation of stx subtypes due to single nucleotide changes (Scheutz et al., 2012). Scheutz et al. (2012) standardized the Stx nomenclature by designating stx1 subtypes as stx1a, stx1c and stx1d, and stx2 subtypes as stx2a, stx2b, stx2c, stx2d, stx2e, stx2f, and stx2g based on amino acid sequence similarity (Scheutz et al., 2012). There is a paucity of data on Shiga toxin subtypes carried by non-O157 E. coli serogroups isolated from cattle feces in the United States. The objective of our study was to determine the subtypes of stx1 and stx2 in non-O157 E. coli serogroups isolated from cattle feces.

Of the total 192 non-O157 STEC strains (170 cattle and 22 human strains) belonging to 27 serogroups tested in the study, 93 (48.4%) were positive for stx1 only, 43 (22.4%) for stx2 only, and 56 (29.2%) for both stx1 and stx2. A total of 149 strains belonging to 23 serogroups were positive for stx1, and of those 132 (88.6%) were stx1a and 17 (11.4%) were stx1c (Table 2). Of the 140 stx1 positive strains originating from cattle feces, 123 (87.9%) were stx1a subtype, and 17 (12.1%) were stx1c. Shiga toxin 1a was the only stx1 subtype found in human STEC strains (9/9; 100%). The stx1a was also the most common subtype of stx1 identified in top six non-O157 E. coli serogroups (Table 2). None of the strains were positive for more than one stx1 subtype.

A total of 99 strains belonging to 25 serogroups were positive for stx2, and of those 79 were stx2a (79.8%), 11 (11.1%) were stx2c, 12 (12.1%) were stx2d. Three stx2 positive-strains belonging to O96, O113 and O130 serogroups of cattle origin revealed double peaks in the chromatogram of stx2 sequences, suggesting the presence of more than one stx2 subtype in the same strain. Subsequent cloning and sequencing revealed that all three strains carried a combination of stx2a and stx2d. All human clinical strains (n = 22), except one, were of subtype stx2a. Only one strain belonging to serogroup O145 carried stx2d. All strains positive for stx2c were of bovine origin (Table 2, Figure 1). Shiga toxin subtypes 2e, 2f, and 2g were not detected in any of the strains tested. A majority of cattle strains (n = 170) possessed stx1a subtype (123/170; 72.4%) followed by stx2a (58/170; 34.1%; Figure 1). However, a majority of human strains carried stx2a (21/22; 95.5%) followed by stx1a (9/22; 40.9%). Shiga toxin 1a was most commonly found in combination with stx2a in strains (n = 46) belonging to 18 serogroups. Six strains belonging to O8, O103, O117, and O130 carried a combination of stx1a and stx2d. Two strains of serogroup O74 carried a combination of stx1a and stx2c and a strain of serogroup O15 was positive for a combination of stx1c and stx2c. A strain of serogroup O130 was positive for three subtypes, stx1a, stx2a, and stx2d (Table 2).

The sequences from a subset of strains (n = 68) were also analyzed by in silico RFLP. The sizes of fragments generated by restriction enzyme digestion are shown in Table 3. The two stx1 genotypes (stx1 and stx1c) determined by in silico RFLP corresponded to the two subtypes (stx1a and stx1c) determined based on the amino acid motifs. For stx2, genotypes, such as stx2, stx2-NV206, and stx2v-ha determined based on the restriction patterns corresponded to stx2a, stx2d, and stx2c subtypes, respectively, determined based on amino acid sequence. However, some strains positive for stx2v-hb gene (based on restriction patterns) corresponded to stx2c and some to stx2d subtype (based on amino acid sequence motifs). Restriction patterns of stx1 and stx2 genotypes determined by in silico RFLP are shown in Figure 2.

Shiga toxins are the major virulence factors of STEC, which are responsible for foodborne illnesses, including life threatening complications of HUS in humans. Shiga toxins exist as several subtypes, which vary in their cytotoxicity, and therefore in the extent of their involvement in human illness (Friedrich et al., 2002; Persson et al., 2007; Fuller et al., 2011). Fuller et al. (2011) demonstrated that stx2a and stx2d subtypes were more potent than stx2b, stx2c, and stx1 based on in vitro (using primary human renal proximal tubule epithelial cells and Vero cells) and in vivo (using mice) experiments (Fuller et al., 2011). Karve and Weiss (2014) demonstrated stronger binding of stx2a, stx2c, and stx2d to a mixture of Gb3 and glycolipids when compared to stx1 and stx2b. Determining the Shiga toxin subtype carried by the STEC strains is important to estimate the risk of human illness associated with specific serotype or source of transmission. Studies have shown that STEC strains carrying stx2a and stx2c are most commonly associated with HUS in humans (Friedrich et al., 2002; Persson et al., 2007; Iyoda et al., 2014). Production of elastase activatable Stx2d subtype in STEC strains has been reported to be a predictor of severity of clinical illness (Bielaszewska et al., 2006). The cleavage of C-terminal amino acids of A₂ peptide of Stx2d by elastase has been reported to increase the cytotoxicity of this Shiga toxin subtype (Kokai-Kun et al., 2000; Melton-Celsa et al., 2002). The Stx2e subtype is most commonly associated with porcine STEC, however, it is also associated with STEC from asymptomatic humans and fresh produce (Friedrich et al., 2002; Beutin et al., 2008; Feng and Reddy, 2013). Although pigeons are a primary reservoir of stx2f carrying E. coli strains, this subtype has also been isolated from human diarrheic patients (Schmidt et al., 2000; Prager et al., 2009). There are studies reporting the distribution of Shiga toxin subtypes in STEC strains isolated from humans and fresh produce (Friedrich et al., 2002; Feng and Reddy, 2013). There are very limited studies on the distribution of Shiga toxin subtypes in cattle, particularly of non-O157 serogroups, in the United States.

In this study, we identified Shiga toxin subtypes associated with 27 serogroups of non-O157 STEC strains of cattle (n = 170) and human (n = 22) origin based on the amino acid sequences deduced from nucleotide sequences. Additionally, we also identified stx genotypes by in silico RFLP, based on the restriction patterns obtained after in silico digestion of nucleotide sequences with specific restriction enzymes. The most common subtype of stx1 carried by non-O157 STEC strains of cattle and human origin was stx1a. A similar finding was also reported in a study on cattle and human STEC strains in Canada (Chui et al., 2015). Tostes et al. (2017) reported that stx1a and stx2a are the most common subtypes in both E. coli O157 and non-O157 E. coli strains of human and bovine origin in Alberta (Tostes et al., 2017). In the present study, a majority of the cattle STEC strains (21.8%; 37/170) and human STEC strains (40.9%; 9/22) carried a combination of stx1a and stx2a. A majority of E. coli O157:H7 strains isolated from outbreaks and sporadic cases in Canada carried a combination of stx1a and stx2a (Chui et al., 2015). In our study, 17 (10%) cattle STEC strains carried stx1c, while none of the human STEC strains carried this subtype. STEC strains carrying stx1c have been isolated from asymptomatic humans and patients with uncomplicated diarrhea (Friedrich et al., 2003). However, stx1c carrying E. coli O78 was isolated from a 2-week old boy suffering from bacteremia and HUS in Finland (Lienemann et al., 2012).

Shiga toxin subtype 2a was the most common stx2 subtype (41.1%; 79/192) found in cattle and human sources. It was the most common subtype detected in STEC strains isolated from patients suffering from HUS (Persson et al., 2007) and those isolated from fresh produce (Feng and Reddy, 2013). It was also the most common subtype of stx2 in STEC strains isolated from cattle in Australia (Brett et al., 2003) and France (Bertin et al., 2001). The second most frequent subtype of stx2 associated with STEC strains isolated from cattle was stx2c, however, none of the human STEC strains included in this study contained stx2c. The stx2c subtype has been reported in STEC strains isolated from human patients suffering from HUS (Persson et al., 2007). It was the third most common subtype (next to stx2d) associated with STEC strains isolated from fresh produce (Feng and Reddy, 2013). Tostes et al. (2017) have reported the distribution of stx1 and stx2 subtypes associated with E. coli O157 and non-O157 E. coli serogroups of bovine and human origin in Alberta, and found that the stx2c subtype was found only in E. coli O157 strains (Tostes et al., 2017). Shiga toxin 2c was the most common subtype associated with STEC O157 strains of cattle origin in Italy, however, none of the STEC O157 strains harbored stx2a (Bonardi et al., 2015). However, in our study, 6.5% of the non-O157 STEC strains isolated from cattle carried stx2c subtype. In the present study, only a small proportion (6.5%) of the bovine strains carried stx2d subtype, and only one human STEC strain carried stx2d. Tasara et al. (2008) have reported stx2d carrying STEC strains isolated from cattle and sheep (Tasara et al., 2008). Presence of stx2d carrying STEC strains in human patients was reported to be significantly associated with HUS (Bielaszewska et al., 2006). In our study, three cattle strains carried two different stx2 subtypes (stx2a and stx2d). STEC strains carrying more than one stx2 subtypes have been reported in previous studies (Persson et al., 2007; Scheutz et al., 2012). The significance of multiple stx2 subtypes with regard to virulence of STEC has not been studied yet. The presence of multiple stx subtypes in STEC is assumed to be the result of recombination of stx phages (Ashton, 2015).

We also identified Shiga toxin genotypes by in silico RFLP. Some strains positive for stx2v-hb gene (based on restriction patterns; Beutin et al., 2007) corresponded to stx2c and some to stx2d subtype (based on amino acid sequence motifs). Single nucleotide changes within the restriction sites could lead to misinterpretation of stx subtypes based on the restriction pattern (Scheutz et al., 2012). In conclusion, a number of non-O157 STEC serogroups isolated from cattle possessed a wide variety of Shiga toxin subtypes, with stx1a and stx2a being the most predominant stx subtypes occurring individually or in combination. Cattle harbor a number of non-O157 STEC serogroups and identification of the Shiga toxin subtypes is useful in assessing the potential risk to cause human illnesses.

Conceived and designed the experiments: JB, TN; Performed the experiments: PS, CS, XS, LN; Contributed reagents/materials/analysis tools: JB, TN; Wrote the paper: PS, TN, JB.

Research was supported by the Agriculture and Food Research Initiative Grant No. 2012-68003-30155 from the USDA National Institute of Food and Agriculture.