Shiga Toxin-Producing Escherichia coli in Yaks (Bos grunniens) from the Qinghai-Tibetan Plateau, China

Shiga toxin (Stx)-producing Escherichia coli (STEC) are recognized as important human pathogens of public health concern. Many animals are the sources of STEC. In this study we determined the occurrence and characteristics of the STEC in yaks (Bos grunniens) from the Qinghai-Tibetan plateau, China. A total of 728 yak fecal samples was collected from June to August, 2012 and was screened for the presence of the stx 1 and stx 2 genes by TaqMan real-time PCR after the sample was enriched in modified Tryptone Soya Broth. Of the 138 (18.96%) stx 1 and/or stx 2-positive samples, 85 (61.59%) were confirmed to have at least 1 STEC isolate present by culture isolation, from which 128 STEC isolates were recovered. All STEC isolates were serotyped, genotyped by pulsed-field gel electrophoresis (PFGE) and characterized for the presence of 16 known virulence factors. Fifteen different O serogroups and 36 different O:H serotypes were identified in the 128 STEC isolates with 21 and 4 untypable for the O and H antigens respectively. One stx 1 subtype (stx 1a) and 5 stx 2 subtypes (stx 2a, stx 2b, stx 2c, stx 2d and stx 2g) were present in these STEC isolates. Apart from lpfA O157/OI-141, lpfA O157/OI-154, lpfA O113, katP and toxB which were all absent, other virulence factors screened (eaeA, iha, efa1, saa, paa, cnf1, cnf2, astA, subA, exhA and espP) were variably present in the 128 STEC isolates. PFGE were successful for all except 5 isolates and separated them into 67 different PFGE patterns. For the 18 serotypes with 2 or more isolates, isolates of the same serotypes had the same or closely related PFGE patterns, demonstrating clonality of these serotypes. This study was the first report on occurrence and characteristics of STEC isolated from yaks (Bos grunniens) from the Qinghai-Tibetan plateau, China, and extended the genetic diversity and reservoir host range of STEC.


Introduction
Shiga toxin-producing Escherichia coli (STEC) are recognized globally as major food-borne pathogens. Clinical manifestations of STEC infections in humans range from non-bloody diarrhea to hemorrhagic colitis (HC) and sometimes fatal hemolytic uremic syndrome (HUS) complications. There are more than 200 known STEC serotypes associated with human illness [1]. O157:H7 is the most frequently encountered STEC in human infections [2]. Many outbreaks and sporadic infections caused by STEC O157:H7/NM have been reported in different regions of the world [3,4,5,6,7,8]. However, non-O157 STEC isolates have been increasingly associated with human infections and outbreaks. In 2011, Germany experienced the largest outbreak of non-O157 STEC, O104:H4, ever recorded with 3,816 cases including 845 HUS cases and 54 deaths, similar outbreaks were reported in France and other counties in Europe subsequently [9,10,11,12]. Non-O157 STEC infections are likely to be under-reported due to awareness and difficulties in isolation and identification in clinical laboratories.
STEC possesse a number of virulence factors, with the production of Shiga toxins (Stxs) being the most critical which leads to the damage of the endothelial cells and potential HUS [13]. The Stx family can be categorized into two major types, Stx1 and Stx2 [14], which differ in their effects on the endothelial cells [15]. Stx1 and Stx2 are further divided into 3 subtypes (Stx1a, Stx1c and Stx1d) and 7 subtypes (Stx2a to Stx2g) respectively [14]. The different Stx types and/or subtypes may be associated with differences in the severity of illness [16,17]. Other factors are purported to increase virulence in STEC isolates. Cytotoxic necrotizing factor 1 (CNF1) and its isoform CNF2 are cytotoxins that activate Rho GTPases leading to tissue damage, perturb the epithelial barrier and impair the function of immune cells [18]. EAST-1 is a genetically distinct toxin structurally related to heatstable enterotoxin (STa) of enterotoxigenic E. coli [19]. Subtilase cytotoxin (SubAB) is the prototype of a new AB 5 toxin family produced by a subset of STEC strains [20]. SubAB is lethal for mice and induces pathological features overlapping those seen in HUS [21].
Typically STEC also possesse the locus of enterocyte effacement (LEE), which encodes proteins necessary for the formation of attaching and effacing (A/E) lesions including the intimin, a translocated intimin receptor (Tir), a type III secretion apparatus, and effector proteins translocated by the secretion system [22]. In the absence of intimin, other adherence factors may increase adherence and virulence in STEC. These include Iha (IrgA homologue adhesin) which is a STEC adherence-conferring molecule conferring the adherence phenotype upon nonadherent laboratory E. coli [23]; Efa1 (EHEC factor for adherence 1) which was shown to be essential for the adherence of the bacteria to cultured epithelial cells, hemagglutination and autoaggregation [24]; LPF (long polar fimbriae) which is closely related to the LPF of Salmonella enterica serovar Typhimurium [25]; and Saa (STEC autoagglutinating adhesin) which is an autoagglutinating adhesin produced by LEE-negative STEC strains [26]. Paa (porcine A/E associated protein), which was first discovered in porcine enteropathogenic E. coli, contributes to the early stages of the development of the A/E lesions and is also present in O157:H7 [27]. Additionally, many STEC strains contain the heterologous 60-MDa virulence plasmid, which contains a number of virulence genes: an enterohemolysin (ehxA), a catalase-peroxidase (katP), an extracellular serine protease (espP) and a adhesin (toxB) [28].
Domestic or wild animals are the primary sources of STEC, such as cattle, pig, sheep, dog, cat, horse, deer and wild boars [29,30,31,32,33], with cattle being regarded as the main natural reservoirs [34]. Humans are the accidental host of STEC through the ingestion of contaminated meat, milk, vegetables, fruits and water. The yak (Bos grunniens) lives at high altitude (above 3,000 m) in China, India, Nepal and other countries. There are more than 14 million yaks on the Qinghai-Tibetan plateau, which represent more than 90% of the world yak population. Yaks are adapted to the harsh environments of severe cold, less atmospheric oxygen, strong ultra-violet radiation and poor forage resources. Domestic yaks are of economic importance (such as meat, milk for food, hide for leather and dung for fuel) to Tibetans and other nomadic pastoralists in high-altitude environments [35,36]. In this study we determined the occurrence and characteristics of STEC from yaks from the Qinghai-Tibetan plateau, China.

Collection of Samples and Enrichment of Fecal Samples
The investigation was carried out in Yushu tibetan autonomous prefecture, Qinghai province, China. Four big herds (more than one thousand free-ranging yaks) were chosen during June to August, 2012. The sites are Jielachong (3,970 m above msl (mean sea level), latitude of 33u489 and longitude of 96u519), Gandacun (4,322 m above msl, latitude of 33u139 and longitude of 96u739), Batangtan (3,987 m above msl, latitude of 32u519 and longitude of 96u569) and Batang (3,871 m above msl, latitude of 32u849 and longitude of 97u119) respectively. Fresh fecal samples of yaks were collected in 2 ml sterile tubes containing Luria-Bertani (LB) medium (30% glycerol added). Collected samples were stored at 220uC immediately and transported to the laboratory in National Institute of Communicable Disease Control and Prevention, China CDC in ice cold conditions. A total of 728 fecal samples was collected for the present study.
Each fecal sample was inoculated into modified Tryptone Soya Broth (mTSB) supplemented with novobiocin (10 mg/ml) (Oxoid, UK) and incubated at 37uC for 18 to 24 h with shaking at 200 rpm.
stx Screening by TaqMan Real-time PCR The enriched samples were investigated for stx 1 /stx 2 genes by TaqMan duplex real-time PCR assay developed in this study (Probe-1, primers Stx1Fr and Stx1Rr for stx 1 ; Probe-2, primers Stx2Fr and Stx2Rr for stx 2 ) ( Table 1). Briefly, 1.5 ml of each enrichment sample was centrifuged at 13,0006g for 2 min, the pellet was suspended in 150 ml of the rapid lysis buffer (100 mM NaCl, 10 mM Tris-HCl [pH 8.3], 1 mM EDTA [pH 9.0], 1% Triton X-100), then boiled for 10 min, and centrifuged at 13,0006g for 2 min. The supernatant was then used as template. Real-time PCR was performed with the Rotor-Gene Q Real-Time PCR system (Qiagen, Germany) using oligonucleotide primers and fluorescent probes targeting stx 1 and stx 2 . The amplification conditions were as follows: initial denaturation at 95uC for 10 s and then 40 cycles of 95uC for 5 s and 60uC for 20 s.
The performance of the TaqMan real-time PCR was validated with reference plasmid constructs, pMD18-stx1 and pMD18-stx2, containing a copy of stx 1 and stx 2 respectively and O157:H7 EDL933 spiked human stools. The limit of detection for stx 1 or stx 2 was 1610 2 copies per reaction using the reference plasmids and 3.55610 3

Isolation of Shiga Toxin-producing Escherichia coli
Enriched fecal samples tested positive by the TaqMan real-time PCR assay for stx 1 and/or stx 2 genes were plated onto CHROMagar TM ECC agar (CHROMagar, Paris, France ), and incubated at 37uC overnight. Ten presumptive colonies (blue or colorless, round moist colonies, but the colonial morphology may be variable) on each plate were picked and screened for the presence of stx 1 and/or stx 2 genes by single colony duplex PCR assay (primers Stx1F and Stx1R for stx 1 , primers Stx2F and Stx2R for stx 2 ) ( Table 1). The stx-positive colonies were plated onto LB and incubated overnight to obtain single colonies for further identification. If all 10 colonies were negative for stx, another 10 colonies were picked and screened. Finally, 1 to 3 stx-positive isolates from each sample were collected for further investigation.

Biochemical Test and Serotyping of STEC Isolates
stx-positive isolates were confirmed to be E. coli by biochemical identification using the API 20E system (bioMérieux, France). The O serogroups were screened by PCR using O antigen specific primers in DebRoy et al. [37]. E. coli O antisera (Statens Serum Institute, Denmark) were used to confirm the O group PCR results. The H type of each isolate was determined by amplifying and sequencing the fliC gene and comparing sequences in GenBank as previously described [34].

Pulsed-field Gel Electrophoresis (PFGE)
Pulsed-field gel electrophoresis was performed using the non-O157 STEC subtyping protocol from PulseNet, USA with some modifications. The genomic DNA was digested with 45 U of XbaI (Takara, Dalian, China) at 37uC for 2 h. A contour-clamped homogenous electric field apparatus CHEF-Mapper (Bio-Rad, USA) was used. The pulse time was ramped from 6.76 s to 35.38 s over 18 h at 6.0 V/cm. The image was captured with a Gel Documentation 2000 software (Bio-Rad, USA) and exported to Bionumerics (Version 4.0, Applied Maths BVBA, Belgium) for analysis of the PFGE patterns. An UPGMA dendrogram was drawn using the BioNumerics software.

Ethics Statement
Fecal samples of free-ranging yaks were acquired with the consent of the owners of the lands and animals. The study was approved by the ethics committee of National Institute for Communicable Disease Control and Prevention, China CDC, according to the medical research regulations of Ministry of Health, China.

Prevalence of STEC in Yak Fecal Samples
Out of 728 yak fecal samples analyzed in this study, 138 (18.96%) were positive for stx 1 and/or stx 2 genes using TaqMan real-time PCR assay. The four herds showed different stx 1 and/or stx 2 positive rates ranging from 14% to 29%. One hundred and twenty eight STEC isolates were isolated from 85 of the 138 stx positive fecal samples giving a culture positive STEC rate of 61.59% for stx PCR positive samples and 11.68% for all samples ( Table 2). A single isolate was obtained from 44 fecal samples, two isolates per sample were recovered from 39 fecal samples, and three isolates each were obtained from two samples.  Table 3).

Presence of stx Genes and stx Subtypes and Other Virulence Factor Genes
Among the 128 STEC isolates, 33 were tested positive for stx 1 only, 75 for stx 2 only and 20 positive for both stx 1 and stx 2 ( Table 4 and Table S1). All of the 53 stx 1 -positive STEC isolates were stx 1a subtype. Of the 95 stx 2 -positive STEC isolates, 5 stx 2 subtypes were identified with 20 isolates of stx 2a , 50 of stx 2b , 6 of stx 2c , 21 of stx 2d and 6 of stx 2g . Four isolates carried both stx 2a and stx 2b , and another 4 isolates carried both stx 2a and stx 2c ( Table 4 and Table  S1).

PFGE
The 128 non-O157 STEC isolates were analyzed by PFGE to investigate their genetic relationship. Five isolates failed to produce distinctive patterns. The remaining 123 isolates were divided into  Table S1). For the 41 fecal samples with two or three isolates, the multiple isolates for 28 samples showed identical PFGE banding pattern, serotype and virulence gene profile ( Figure 1 and Table  S1), suggesting that the multiple isolates from the same sample belong to the same STEC strain. However, 10 (Figure 1 and Table S1). These data suggest that some yaks were colonized by more than one STEC strain. There were also 3 samples (samples 114, 369, 525) with multiple isolates having the same PFGE type but different serotypes (Figure 1 and Table S1).
The PFGE patterns were used to construct an UPGMA dendrogram (Figure 1) which shows that the STEC isolates were genetically diverse with nodes linking single isolates or groups of isolates at less than 80% similarity. Interestingly many isolates were grouped together with similarity at 90% or greater suggesting close genetic relationships. In particular, isolates of the same serotype had the tendency to cluster together and also carried identical stx or stx subtypes. These includes O117:H21, O8:H9, O8:H16, O2:H45, O22:H8 and O78:H45 carrying stx 2b , stx 1a +stx 2d , stx 2b , stx 1a , stx 2b and stx 2g respectively. The O117:H21 isolates were in 3 related nodes while the others were in their own single node. Isolates showing identical PFGE patterns came from the same yak herd with the exception of EZKX01005   **  23  15  14  1  22  1  22  0  1  1  1  13  23  8   Total  128  53  95  2  87  2  66  7  13  13  24  44  which contained 5 isolates from herd 2 and 1 isolate from herd 4. The herd 4 isolate (MN1287-15) also shared the same H antigen, stx 1a and the presence of iha, saa and ehxA as the herd 2 isolates. However this PFGE type displayed higher heterogeneity with 2 different O and 2 different H antigens and 1 isolate also carrying astA. The two isolates from another 2 samples (samples 114 and 525) each showed the same PFGE type but different serotypes. The virulence gene profiles also showed a clustered distribution but less pronounced than the serotypes and stx subtypes. The main nodes containing the following serotypes were uniformly positive for some of the virulence factors: O117:H21 was positive for both and only cnf1 and cnf2; O2:H45 for astA only, O8:H9 for iha, saa,ehxA and espA; O8:H16 for iha only; O78:H8 for iha, saa, subA and ehxA; and O78:H45 for both paa and astA. Interestingly O22:H8 carried none of the virulence factors tested (Table S1).

Discussion
Ruminants, especially cattle, are the major reservoirs of STEC. The prevalence of STEC in beef cattle ranged from 0.2 to 27.8% for O157 STEC, and 2.1 to 70.1% for non-O157 STEC [38]. Bandyopadhyay et al. [39] recovered 42 STEC isolates from 273 rectal swab samples (15.38%) in an STEC study of the yak Poephagus grunniens. Our results showed a similar rate of STEC isolation. Of the 728 yak (Bos grunniens) fecal samples screened, 18.96% of the samples (138/728) were positive for the stx genes by PCR and 11.68% (85/728) by culture. Interestingly nearly 40% of the STEC positive samples by PCR were negative by culture. It seems that either the STEC cell numbers were low in the fecal sample or stx-positive non-E. coli was present in the feces. Bosilevac et al. [34] screened ground beef samples for STEC and only recovered STEC by culture from 300 out of the 1006 stx-positive samples with a success rate of just 30%.
Bandyopadhyay et al. reported the occurrence and characteristics of STEC from feces, milk and milk products of Poephagus grunniens, another species of the yaks, in India [39,46,47]. The STEC strain isolated from the feces of Poephagus grunniens belonged to 22 divergent O serogroups [39]. Among these serogroups, only three (O2, O22 and O158) were also present in our study, suggesting that there are diverse STEC strains of overlapping O serogroups present in these two species of the yaks.
Since the carriage of combinations of the stx genes and stx subtypes has been associated with disease severity, the profile of the stx genes gives us an overview of the pathogenic potential of these STEC isolates from the yaks. In this study, 1 stx 1 subtype,5 stx 2 subtypes and 12 different combinations of stx 1 /stx 2 subtypes were found in the 128 STEC isolates analyzed ( Table 4). Manning et al. found that clade 8 O157:H7 strains which were significantly more likely to infect patients with HUS and more likely to have both the stx 2a and stx 2c genes, implying that the carriage of both the Stx2a and Stx2c phages contributes in part to the greater virulence of clade 8 strains [48]. Mellmann et al. analyzed a collection of 524 EHEC isolated from HUS patients and found that 169 (32.3%) belonged to 34 non-O157 serotypes and that profiles of stx 1a only, stx 1c only, stx 1a +stx 2a , stx 1c +stx 2a , stx 1c +stx 2d , stx 2a only, stx 2c only, stx 2d only and stx 2a +stx 2c were present in these non-O157 STEC isolates [2]. In our STEC isolates, several of above profiles were also present including stx 1a only, stx 1a +stx 2a , stx 2a , stx 2c only, stx 2d only and stx 2a +stx 2c . These results suggest that the non-O157 STEC isolates from the yaks have the potential to cause human illness and there is a need to monitor the local human population for STEC infections.
Non-O157 STEC isolates that carry both stx 2 and eae genes were more often associated with severe disease [49,50]. In this study only 2 eae positive isolates (MN1208-22 and MN1208-34) were isolated from the yaks, both of which carried stx 2a, indicating their virulence potential. The 2 isolates were on the same PFGE node, but interspersed by other eae negative isolates (Figure 1), suggesting independent acquisition of the LEE locus by the two isolates. However, although eae encoded on the LEE pathogenicity island is absent from almost all of the STEC isolates, non-LEEencoded effector proteins potentially involved in virulence have been demonstrated in some serotypes identified in this study [51].
Since the majority of the STEC isolates were eae negative, we investigated other factors associated with adherence including Iha and Saa, both of which have been reported to be correlated with increased adherence in the eae negative strains [23,26,52]. We found that a high proportion of the yak STEC isolates contained iha (67.97%) and saa (51.56%). This finding is similar to that reported by Bosilevac et al. [34] in the cattle where 88% and 73% of the STEC isolates were positive for iha and saa respectively. The study of Bosilevac et al. further found that the saa gene was always present in the absence of eae and was more often present in isolates that were also positive for the large virulence plasmid [34]. Our yak STEC isolates also showed a similar correlation. Interestingly our two eae positive isolates also harbored the iha and saa genes.
Of the two other adhesin genes, lpfA and paa screened, all 128 isolates were negative for lpfA while only 7 isolates were positive for paa, 5 of which belonged to the same serotype (O78:H45). Thirty four isolates were negative for all the adherence -associated genes tested (eae, iha, efa1, lpfA O157/OI-154 , lpfA O157/OI-141 , lpfA O113 , saa,  paa and toxB), suggesting that other novel adherence genes must exist in these STEC isolates, which warrants further investigation. Of the additional putative virulence factors (cnf1, cnf2, astA, subA) screened, the prevalence of cnf1 and cnf2 was low (10.16%), with both being present in the same 13 STEC isolates. The astA gene was also present at a low frequency of 18.75%. However, the subA gene was more prevalent with 34.38% isolates positive. Interestingly, the subA gene was present in none of STEC isolates that harbored the astA gene. The significance of this mutually exclusive presence is unknown. The prevalence of subA in the yaks is similar to that observed in some cattle populations [53]. In contrast, a much smaller proportion of the STEC isolates isolated from human infections were positive for the subAB genes, which ranged from 2% in the USA to 10% in Australia [54].
Of the 4 60-MDa large plasmid encoded virulence genes [55], ehxA, katP, espP and toxB screened, none of the STEC isolates carried katP and toxB, whereas 51.56% and 28.13% STEC isolates were positive for ehxA and espP respectively. All STEC isolates that were positive for espP were also positive for ehxA. Our findings are similar to that in the cattle population based on isolation from ground beef in which ehxA and espP were more commonly present [34].
PFGE analysis showed that there is quite high genetic diversity of the STEC in the yaks and it seems that there is no separation of the STEC population between the yak herds. Eight (O8:H16, O8:H45, O22:H8, O52:H2, O78:H8, O78:H21, O78:H45 and O117:H21) of the 36 serotypes were present in 2 different herds.
The serotype data also suggested that some clones are wide spread. The yaks may harbor more than one type of STEC isolates since 11.76% (10/85) of the samples contained 2 or more isolates of different PFGE patterns.
In conclusion, this study was the first report on the occurrence and characteristics of STEC isolated from yaks (Bos grunniens). We isolated 128 STEC isolates of different serotypes, stx subtypes and virulence gene profiles from the yaks (Bos grunniens) from the Qinghai-Tibetan plateau, revealing that Bos grunniens are natural reservoirs of STEC. This study further extends our knowledge of the genetic diversity and reservoir host range of STEC. The serotypes and stx subtypes identified were partially reported in human infections, pointing to the potential of these STEC isolates to cause disease in humans. Further investigations are needed to assess their public health significance in Tibetans and other nomadic pastoralists in this region.

Supporting Information
Table S1 Profiles of the 128 STEC isolates. (XLS) Figure 1. PFGE profiles of non-O157 STEC isolates from the yaks. The corresponding isolate names, PFGE patterns, no. of samples and herds, serotypes and stx 1 and/or stx 2 subtypes are listed on the right. For samples with more than 1 isolates, the numbers (x/y) in brackets in the sample column are number of strains (x) belonging to that PFGE pattern out of the total number of y isolates from that sample. Note that 5 isolates failed to produce a PFGE pattern and were not on the tree. For full list of isolates and their PFGE patterns and other data, see Table S1. doi:10.1371/journal.pone.0065537.g001