Discussion
Ruminants, particularly cattle and sheep, have long been considered as important reservoirs for STEC strains. To our knowledge, this is the first study that has characterized cattle and sheep STEC strains isolated from Xinjiang Province, China, a region well known for animal husbandry and grazing. Cattle and sheep can harbor a wide range of STEC serotypes in their gut (Hussein & Bollinger, 2005; Zweifel, Blanco, Blanco, Blanco, & Stephan, 2004). In line with previous findings, our isolates demonstrate a high diversity of serotypes. Serotypes identified in this report were compared to those isolated in other studies from various sources. Isolates in this study belong to 23 O serogroups and 29 O:H serotypes (Fig. 1), of which 12 were isolated from cattle only and 13 from sheep only. Minimal overlap between isolates of two different animal origins was observed. However, many studies indicate that all sheep isolate serogroups found here were isolated from bovine feces and/or carcasses (Arthur, Barkocy-Gallagher, Rivera-Betancourt, & Koohmaraie, 2002; Hussein & Bollinger, 2005). Notably, the O serotypes O3, O15, O81, O116, O129, and O140 of cattle isolates here were rarely reported to be present in sheep isolates (Blanco et al., 2003; Martins et al., 2015; Oporto, Esteban, Aduriz, Juste, & Hurtado, 2008). O129, which was formerly suggested to be EPEC, was not reported in cattle isolates in previous studies, although it was earlier isolated from human feces (Blanco et al., 2006; Schwaiger, Grif, Pierard, Karch, & Allerberger, 1999). Some serogroups share a common ancestor with other pathotypes such as EPEC and EAEC, and integration of astx -containing bacteriophage may convert these into a more virulent variant (Steyert et al., 2012). All serogroups here were previously shown to be associated with human STEC infections, but with different degrees of correlation (CDC, 2017). For example, O5 and O76 isolates caused numerous human infection cases in recent years, but O128 and O113 resulted in a moderate number of cases, whereas O66 and O81 were responsible for a very limited number of infections (CDC, 2017). Among all the serotypes, O8:H21, O76:H19, O104:H7, O113:H4, and O128:H2 have been linked to diarrhea or HUS cases (CDC, 2017; Islam et al., 2007; Monaghan et al., 2012; Mora et al., 2007). Twenty-two serotypes found in this study were reported elsewhere among isolates from various origins such as cattle, goat, and human patients. Serotypes O66:H45 and O74:H39 have not previously been reported in STEC isolates of sheep; O15:H10, O21:H11, and O6:H21 have not previously been reported in STEC isolates of cattle. Compared to a study of 126 beef cattle STEC isolates from a farm in Sichuan Province, China, we share 7 serogroups but only 1 serotype, O81:H31, suggesting high diversity of O serotypes among different regions of isolation.
Aside from the O:H serotypes, the presence of a single stx 1 orstx 2 or a combination of these and stx subtypes has been recognized as potential indicator of STEC pathogenicity. STEC strains that produce stx 1 only are generally mild in pathogenicity, whereas strains that produce stx 2 alone are more frequently associated with severe diseases such as bloody diarrhea (BD) or HUS (Johura et al., 2016; Melton-Celsa, 2014). The presence ofstx 1+stx 2 may suggest medium pathogenicity between that ofstx 1 only and stx 2 only (Arthur et al., 2002; Brandal et al., 2015). In a study of 361 non-O157 STEC isolates from cattle carcasses, the prevalence ofstx 2 only, stx 1 only, and stx 1+stx 2 isolates was shown to be 36.6%, 50.2%, and 12.2%, respectively (Arthur et al., 2002). For the Sichuan cattle isolates from China, the corresponding prevalence is 35.7%, 24.6%, and 39.7%, respectively (Fan et al., 2019). For our cattle isolates, the corresponding prevalence is 7.4%, 31.4%, and 61.2%, respectively. The lower percentage ofstx 2-only isolates and higher percentage ofstx 1+stx 2 isolates suggest a lower possibility of highly virulent strains but a higher possibility of moderate-to-high virulent isolates from cattle. In a study of 379 sheep non-O157 STEC fecal swab isolates, the prevalence of stx 2 only, stx 1 only, andstx 1+stx 2 isolates was shown to be 1.3%, 56.2%, and 42.5%, respectively (Blanco et al., 2003). In a study of 70 ovine isolates, the corresponding numbers are 14.3%, 52.8%, and 32.9%, respectively (Martins et al., 2015). The corresponding prevalence in our sheep isolates is 3.4%, 79.4%, and 17.2%, respectively. Thus, compared to sheep isolates, cattle isolates are more likely to contain stx 2 alone or stx 1+stx 2, implying cattle isolates have a high probability of causing severe disease.
For cattle isolates, all stx 1 are stx 1a subtype, and a majority of stx 2 subtypes is stx 2a (62.5%), with others being stx 2c. For sheep isolates, stx 1 subtypes arestx 1a (35.7%) and stx 1c (64.3%); stx 2 subtypes are mostly stx 2b [83.3%, similar to the reported 84.8% (Martins et al., 2015)], withstx 2a being 16.7%. High prevalence of stx 1c-positive (but not other subtypes) sheep isolates was also found in other studies [57% stx 1c among the stx 1-positive strains (Zweifel et al., 2004)], supporting the notion that sheep represent the main reservoir of stx 1c-carrying strains (Brett et al., 2003).Stx 1c-positive strains are not frequently associated with HC or HUS but tend to trigger asymptomatic infection or mild diarrhea (Brandal et al., 2015). STECs carryingstx 2b have been associated with sporadic HUS cases, indicating its potential to cause severe infections in humans, although thestx 2b subtype in most cases is linked to lower virulence (de Boer et al., 2015) (Buvens et al., 2012). The stx 1a,stx 2a, stx 2c, and stx 2d subtypes are most often associated with HC and HUS (Fao/Who Stec Expert, 2019; Persson, Olsen, Ethelberg, & Scheutz, 2007). Stx2a is more potent than Stx2b, Stx2c, and Stx1 in cell cultures and mouse models (Fuller, Pellino, Flagler, Strasser, & Weiss, 2011). Therefore, by judging from the stx subtypes, cattle isolates tend to have high virulence potential, whereas sheep isolates may be assumed to be of low virulence.
None of the isolates in this study contain the eae gene. A low occurrence of eae in STEC isolates from domestic ruminants was also observed elsewhere (Amezquita-Lopez, Quinones, Lee, & Chaidez, 2014; Schilling et al., 2012). In a study of 521 STEC sheep isolates, eae was detected only in 0.8% of the isolates (S. Sanchez, 2010). In contrast, other adhesin genes such as iha , saa , and subA were present in many strains but with different prevalence rates. It was reported iha and saa are carried by more than half of the STEC strains isolated from yak and cattle (Bai et al., 2013; Bosilevac & Koohmaraie, 2011). For our cattle isolates, iha and saa prevalence was 42.6% and 40.7%, respectively. It has been suggested that saa is correlated with the presence of the large STEC virulence plasmid (Toma et al., 2004). Indeed, the isolates positive with saa also harbored ehxA , which is usually encoded on the virulence plasmid (Blanco et al., 2003; Martins et al., 2015; Oporto et al., 2008). In many isolates,ehxA and espP are genetically linked (Fan et al., 2019), particularly the O157 serogroup (Islam et al., 2008), but in our study, these do not always coexist. In particular,ehxA -positive cattle isolates were all tested negative inespP , suggesting that these may not always encode on the same plasmid. We found low prevalence of katP (3.7%) and subA(11.1%) in cattle isolates, which is comparable to another study of China isolates (Fan et al., 2019). In a study of 60 ovine isolates, 67% of these are saa -positive, but our ovine isolates did not harbor saa , thereby suggesting a unique regional trait. There were few studies reporting the virulence profile of non-O157 sheep isolates; thus our study provides comprehensive and valuable information on virulence genes of non-O157 sheep isolates.
The genetic relatedness of our isolates to the human HUSEC collection was explored. The O76:H19 serotype is very common in sheep isolates (Zweifel et al., 2004). The four ST675 isolates belonging to O76:H19 are similar to HUSEC039. Although lackingeae and containing a mild-virulence subtype stx 1c, these contain ehxA , subA , and iha , which could increase their virulence potential. Another human disease-associated serotype O128:H2 frequently isolated from sheep (Zweifel et al., 2004) contains isolate SG18-6-2 and HUSEC028. For HUSEC023 and the 4 sheep isolates of ST40, the former contains a virulent stx 2d, but the 4 isolates containstx 1c only. In the case of ST101, two cattle isolates clustered with HUSEC025, but the former contains stx 2a, whereas HUSEC025 isstx 2-negative. The ST43 cattle strain CD15-213 clustered with HUSEC001 contains two virulent stx subtypes, stx 1a andstx 2c, as well as the key virulence genes ehxA ,espP , subA , saa , iha , andfyuA /irp2 . Thus, CD15-213 may exhibit the highest pathogenicity. Remarkably, this serotype was shown to be present in Morocco market meat (Badri, Fassouane, Filliol, Hassar, & Cohen, 2011), raising significant food safety concerns. Taken together, the aforementioned cattle isolates may pose the greatest risk. In addition, they should be paid much attention in future surveillance work.
In conclusion, we used WGS analysis to characterize the serotype, virulence gene profile, and genetic diversity, and relationship of a collection of non-O157 STEC isolates from cattle and sheep. This method enables us to reveal much more detail of bacterial isolates, with the highest discriminatory power possible. Several sheep and cattle grouped with HUSEC strains, particularly strain CD15-213, may have the potential to cause serious human infections. Local human HC and HUS cases should be closely monitored, and its relatedness to STEC isolates from various sources should be investigated in order for better food-borne disease prevention.