Prevalence of Escherichia coli O-Types and Shiga Toxin Genes in Fecal Samples from Feedlot Cattle

Abstract

While efforts to control foodborne illness associated with the Shiga toxin–producing Escherichia coli (E. coli) O157 through processes and procedures implemented at harvest facilities have been very successful, there is concern about the burden of illness associated with other Shiga toxin–producing E. coli. The U.S. Department of Agriculture Food Safety and Inspection Service announced plans to classify an additional six non-O157 Shiga toxin–producing E. coli as adulterants. Little is known about the prevalence and distribution of these E. coli in the animal production environment. An investigation of the prevalence of O157 and the six major non-O157 E. coli serogroups was conducted in 21 feedlots over the period July 2011 to October 2011. Individual fecal swabs were collected from cattle approximately 60 days after their arrival in the feedlot and were pooled for evaluation using a polymerase chain reaction assay to identify the presence of seven E. coli O-types (O157, O45, O103, O121, O145, O26, and O111) and four virulence genes (stx1, stx2, eaeA, and ehxA). Overall, 1145 fecal pools were evaluated, with 506 (44.2%) being positive for one or more of the E. coli O-serogroups. The pool prevalences for E. coli O157, O45, O26, O103, O121, O145, and O111 were 19.7%, 13.8%, 9.9%, 9.3%, 5.5%, 1.1%, and 0.5%, respectively. Nearly all pools were positive for ehxA (99.7%) or stx2 (98.6%). The pool level prevalence for stx1 and eae was 65.5% and 69.3%, respectively. Pools that were positive for one or more of the other E. coli O-serogroups were 1.37 times more likely to be positive for E. coli O157. Conversely, pools that were positive for E. coli O157 were 1.43 times more likely to be positive for at least one of the other E. coli O-serogroups evaluated. These data will be useful to understand the expected prevalence of potential Shiga toxin–producing E. coli in cattle feedlots.

Introduction

E ach year an estimated 9.4 million cases of foodborne illness occur in the United States, resulting in an estimated 1351 deaths (Scallan et al., 2011). Of these, 63,153 cases and 20 deaths are attributed to domestically acquired foodborne Escherichia coli (E. coli) O157 infections. The Centers for Disease Control and Prevention (CDC) has set goals for reducing the incidence of E. coli O157 through the Healthy People initiative (CDC, 2012). The Healthy People 2010 goal for the incidence of E. coli O157 cases was met early, following steady declines over the preceding years (CDC, 2000; NCHS, 2012). The beef industry has adopted procedures and processes in harvest facilities to reduce the occurrence of E. coli O157 on carcasses and in trim meat used for grinding. These procedures comprise part of the required Hazard Analysis and Critical Control Point plans for all harvest facilities as mandated by the United States Department of Agriculture Food Safety and Inspection Service. Despite the steady progress in reducing the occurrence of E. coli O157 on carcasses and trim, cases of hemorrhagic diarrhea and hemolytic uremic syndrome continue to occur in people. Some of these cases of hemorrhagic diarrhea and hemolytic uremic syndrome cases have been attributed to non-O157 Shiga toxin–producing E. coli (STEC) (Scallan et al., 2011). The estimated number of E. coli O157 and non-O157 STEC cases attributed to domestically acquired foodborne transmission in the United States are 63,153 (∼36%) and 112,753 (∼64%), respectively. The United States Department of Agriculture Food Safety and Inspection Service has announced plans to classify six O-serogroups of STEC as adulterants, in addition to E. coli O157 (FSIS, 2011). The additional O-serogroups include O26, O103, O111, O121, O45, and O145. Little is known about the prevalence of these O-serogroups of E. coli in the beef production environment. The objective of this study was to use an 11-gene polymerase chain reaction (PCR) assay to evaluate pooled fecal samples from feedlot cattle for the presence of E. coli O157 and the six additional O-serogroups, as well as to evaluate samples for genes coding for production of Shiga toxins 1 and 2 (stx1 and stx2), intimin (eae) and enterohemorrhagic hemolysin (ehxA, or EHEC-hlyA).

Materials and Methods

Operation and pen selection

A convenience sample of 21 feedlots from four states (Iowa, Kansas, Oklahoma, and Texas) was recruited to participate in the study. Feedlots from 6000- to 60,000-head capacity were visited for sampling approximately monthly. On the day of sampling, pens that were scheduled for re-implant processing were eligible for sampling. On any given sampling day, between one and five pens were sampled for that particular feedlot.

Sample collection and processing

Samples were collected from July 11, 2011 through September 29, 2011. For each pen, a fecal swab (BD BBL, Franklin Lakes, NJ) was collected per rectum from each of up to 40 animals. Pools for testing were designated at the time of collection by placing five swabs (in their individual transport vials) into a single plastic bag. Pooled samples were shipped on ice for overnight delivery and held at 4°C until processing. All samples were processed within 72 h of collection. At the time of laboratory processing, each of the five swabs within a plastic bag (designated pool) was placed in a single E. coli enrichment broth (EC Broth, Oxoid Ltd., Hampshire, England) tube by cutting the swabs from midshaft. The broth containing the pooled swabs was incubated for 15–18 h at 37°C prior to DNA extraction.

Sample evaluation

Laboratory methods for the evaluation of the samples have been described elsewhere (Bai et al., 2012). A brief overview of the methods follows.

Primer design

All primers, except the pair for O111, were from Bai et al. (2012). Primers for O111 were redesigned in this study to achieve more even intensity in amplification (Forward [O111F3]: ACAAGAGTGCTCTGGGCTTC; Reverse [O111R3]: AAACTAAGTGAGACGCCACCA). The wzx gene, which encodes for a flippase required for O-polysaccharide export, was used to design this new pair of primers using Primer3 (http://biotools.umassmed.edu/bioapps/primer3_www.cgi), a free online software. Other primer sequences were given in Table 1 of Bai et al. (2012). The amplicon sizes of the 11 genes are: 890, 740, 655, 587, 523, 477, 417, 375, 296, 230, and 199 bp for genes encoding O45, O103, stx1, O121, O145, stx2, O26, eae, O157, O111, and ehxA, respectively. Specificity of the test has been validated with 185 E. coli strains of the “Top-7” serogroups (O26, O103, O111, O121, O45, O145, and O157), 65 E. coli strains belonging to 26 other serogroups, and 33 strain representing 12 species of non-E. coli Gram-negative, and Gram-positive bacteria. Sensitivity was evaluated with pure culture, and culture-spiked negative cattle feces (Bai et al., 2012).

Table 1.

Sample Distribution by Pen and Feedlot

Feedlot number	Number sampling dates	Pens sampled	Pools created
1	2	6	48
2	3	9	67
3	4	10	72
4	3	9	72
5	3	11	72
6	3	9	81
7	4	12	105
8	4	11	80
9	1	3	24
10	1	3	24
11	1	3	24
12	2	6	48
13	2	6	50
14	2	6	48
15	2	5	40
16	2	6	52
17	2	6	48
18	2	6	48
19	2	6	48
20	2	4	22
21	3	9	72
Total	50	146	1145

Template DNA preparation

Following the overnight enrichment, 250 μL of the broth was transferred to a 96-well plate, spun down, and DNA was extracted using the BioSprint 96 DNA Blood Kit with the following modifications. The pellet of the overnight culture was resuspended with 180 μL of ATL buffer (purchased separately) and then 20 μL of protease was added. The mixture was incubated in a 70°C waterbath for 1 h, and processed with an automated DNA extraction system, BioSprint 96, following instructions from the BioSprint 96 DNA Handbook (Qiagen, Valencia, CA). The elution volume used was 100 μL for each sample.

Multiplex PCR reaction and visualization

The PCR procedure was similar to that of Bai et al. (2012). Briefly, all primer stocks were prepared in 1X TE buffer (Integrated DNA Technologies, Inc., Coralville, IA) at concentrations of 100 pM/μL. Equal volumes of the 11-primer pairs (22 primers) were mixed together. In a 25-μL PCR reaction, 1.2 μL of primer mix was used, resulting in a final primer concentration of 0.22 μM for each primer. Each reaction also contained 12.5 μL of BioRad iQ Multiplex Powermix, and 5 μL of extracted DNA from the enriched fecal swab. The PCR amplification program included a 5-min denaturation at 94°C, followed by 35 cycles of 94°C for 30 s, and 67°C for 80 s. The PCR products were separated and visualized with a screening cartridge using a capillary electrophoresis system (QIAxcel, Qiagen, Valencia, CA).

Statistical analysis

All descriptive analyses were performed using SAS (SAS 9.3; SAS Institute, Cary, NC). Associations between the presence of PCR product for E. coli O157 and the other non-O157 STECs were done by Proc Genmod in SAS using a repeated statement in order to account for the clustering of sample pools within pens. The variance component analysis was done using STATA (Intercooled STATA 9; StataCorp., College Station, TX). Sample pools were considered to be nested within pens that were nested within feedlots. For the purpose of evaluating seasonal trends, the sampling dates were collapsed into 3 months and the prevalence was determined for each month's sample pools. Statistical significance was determined by p<0.05.

Results

Samples were collected on one to four dates in 21 feedlots (Table 1). From 1 to 12 pens were sampled in each feedlot during the course of the study. Sampling resulted in the creation and evaluation of 1145 fecal pools from 146 pens.

All of the 21 feedlots had one or more of the seven O-serogroups of E. coli detected at some point in the study. The frequency of the E. coli O-serogroups and virulence factors are shown in Table 2. All feedlots had positive pools for O45 and O157 over the course of the study. Only four feedlots (19.1%) had one or more pools positive for O111 over the sampling period.

Table 2.

Percent Feedlots from Four U.S. States, Pens, and Sample Pools with Positive Polymerase Chain Reaction (PCR) Products

PCR product	Feedlots (n=21)	Pens (n=146)	Sample pools (n=1145)
O157	100	55.5	19.7
O45	100	46.6	13.8
O26	90.5	39.7	9.9
O103	85.7	27.4	9.3
O121	76.2	21.2	5.5
O145	33.3	6.9	1.1
O111	19.1	3.4	0.5
Any O-type	100	86.3	44.2
stx1	100	97.9	65.5
stx2	100	99.3	98.6
eaeA	100	95.2	69.3
hlyA	100	100	99.7

Overall, 86.3% of pens were positive for one or more of the O-serogroups. The most common O-serogroup detected at the pen level was O157 with 81 (55.5%) positives. Least common was O-serogroup O111 with only five (3.4%) positives. Among the 126 pens that were positive for one or more O-serogroups, 41 (32.5%) were positive for only one of the seven O-serogroups.

At the sample pool level, 44.2% of the pools had one or more of the O-serogroups detected. The most commonly detected O-serogroup among the sample pools was O157, with 19.7% of pools positive. The least commonly detected O-serogroup among pools was for O111, with only six (0.5%) being positive. Among the 506 pools that were positive for one or more O-serogroups, 70.4% were positive for only one of the seven O-serogroups.

The number of pools that were positive for a single O-serogroup and concurrently positive for the virulence factors is shown in Table 3. All of the pools that were positive for one of the seven O-serogroups were positive for ehxA. The percentage of pools that were positive for stx2 ranged from 85.7% for those positive for only O145 to 100% for those positive for only O157, O103, or O121.

Table 3.

Percentage of Sample Pools Positive for a Single O-Serogroup That Were Positive for Various Virulence Factors

		Percent positive
O-serogroup	Number sample pools	stx1	stx2	eaeA
O157	137	74.5	100	90.5
O45	83	80.7	97.6	75.9
O26	69	59.4	98.5	79.7
O103	37	81.1	100	89.2
O121	21	80.9	100	57.1
O145	7	57.1	85.7	71.4
O111	2	100	100	100

In slightly over half of the sample pools (n=639, 55.8% of all pools), none of the seven O-serogroups were identified; however, many were still positive for one or more of the virulence factors. All but three pools (99.5%) were positive for ehxA. Of the pools that were negative for all O-serogroups evaluated, the percentage of pools positive for stx1, stx2, and eae were 58.1%, 98.1%, and 56.5%, respectively.

Pools that were positive for one or more of the other E. coli O-serogroups were more likely to be positive for E. coli O157 (odds ratio=1.37, p=0.03) than those that were negative for all of the other E. coli O-serogroups (Table 4). There was also an increased likelihood of a pool testing positive for another O-serogroup when E. coli O157 was identified in the pool (odds ratio=1.43, p=0.02).

Table 4.

Odds Ratio for a Pool Testing Positive for O157 by Status for Other O-Serogroups

O-serogroup	Odds ratio	p-Value
O45	1.41	0.08
O26	1.06	0.82
O103	1.22	0.36
O121	1.14	0.67
O145	0.94	0.87
O111	1.15	0.91
Any	1.37	0.03

The prevalence of positive sample pools by month generally increased from July to September (Table 5). The largest increase in the percentage of sample pools that were positive was for E. coli O157 going from 5.1% of sample pools positive in July to 40.9% positive in September.

Table 5.

Percent of Sample Pools Positive by Sampling Month

	O-serogroup
Month	O45	O103	O121	O145	O26	O157	O111
July (n=157)	3.8	0.6	0.6	0.6	14.6	5.1	0
August (n=663)	14.6	10.7	4.5	0.6	8.1	12.8	0.3
September (n=325)	16.9	10.5	9.8	2.5	11.1	40.9	1.2

Most of the variation in the percentage of positive sample pools was observed at the sample pool and pen levels rather than the feedlot level (Table 6).

Table 6.

Percent of Variation in Positive Sample Pools Attributed to Study Design Variables

	O-serogroup
Level	O45	O103	O121	O145	O26	O157	O111
Feedlot	16.5	37.9	13.7	15.5	10.7	11.5	8.2
Pen	27.4	28.3	47.1	27.6	31.3	42.6	43.9
Sample pool	56.1	33.8	39.2	56.8	57.9	45.9	47.9

Discussion

All of the E. coli O-serogroups and virulence genes evaluated were widely distributed across the feedlots participating in the study. The feedlot prevalence by O-serogroup varied from 33.3% (O145) to 100% (O45, O157). Although primarily focused on O157, previous studies have found similarly high group (feedlot or pen) prevalence rates (Hancock et al., 1997b; Kobayashi et al., 2001; Smith et al., 2001; Woerner et al., 2006). All feedlots were positive for each of the four virulence genes evaluated in this study. The pen-level prevalence of O-serogroups was more moderate, ranging from approximately one fourth to one half of pens having one or more positive pools. Still, nearly all pens had positive pools for the virulence genes evaluated. At the pool level, prevalence was much lower, ranging from 0.5% (O111) to nearly 20% (O157). At least two thirds of the pools were positive for the virulence genes evaluated. These data suggest that these O-serogroups may be widely distributed throughout the feedlot industry but that at any one time relatively few animals will be positive. On the other hand, the high prevalence of the virulence genes at all levels of analysis (feedlot, pen, and pool) suggests that organisms with one or more of these genes are common inhabitants of the cattle gastrointestinal tract, and screening only these genes may not be useful as a sole screening method to identify animals or groups of animals that present a high risk of transmitting potential foodborne pathogens.

There was generally a rise in the prevalence of all the E. coli O-serogroup-positive pools even over the relatively short duration of the study. For most of the O-serogroups, the prevalence in September was higher than the prevalence in July, and for some of the O-serogroups (e.g., O157) this rise in prevalence was substantial. The seasonal increase in E. coli O157 prevalence has been observed by others (Chapman et al., 1997; Hancock et al., 1997a; VanDonkersgoed et al., 1999). Though not of the same magnitude, the rise in prevalence for the other O-serogroups suggests that this apparent seasonal pattern may be under similar control for the entire group of organisms but that the seasonal peaks in prevalence may not be as dramatic as for E. coli O157.

Attributing the variation in observed positive and negative findings to the different levels of the study design hierarchy gives some clues about where investigation of intervention strategies may be appropriate. The low amount of variability attributed to the feedlot level stems from the fact that most of the feedlots were positive for most of the E. coli O-serogroups. The proportion of variation attributed to the individual pools was highest for each of the O-serogroups with the exception of O121 and O103. This suggests that there is greater diversity (mix of positives and negatives) at this level than at other levels. Understanding the factors responsible for the relatively large proportion of variation observed may suggest control strategies.

There are some limitations associated with this study. The feedlots participating in this study were not a random sample of feedlots. Since the status of an animal or pen or feedlot with regard to the presence of various O-serogroups of E. coli is not readily apparent without testing, it is unlikely that these operations would be a biased population with either increased or decreased prevalence of the O-serogroups evaluated leading to their participation in the study.

The evaluation of the samples by PCR for specific genes does not provide information on the viability of the organisms that were the source of the genes, the numbers of organisms present, or the simultaneous presence of the genes within the same organism. For example, the concurrent detection of an O157 strain and a stx2 gene does not necessarily mean that the O157 strain carries the stx2 gene; it could be carried by other strains in the same sample pool. This was demonstrated in the current study by the presence of pools that were O-serogroup negative, but virulence gene positive. Because of these limitations, the results should be interpreted with caution and the positive pools should only be considered as “potential” STEC positives. Isolation and further testing would be required to confirm the presence of a STEC in the fecal pool. The importance of this was illustrated in a study by Galland et al. (2001), where they demonstrated that approximately one half of E. coli O157:H7 isolates did not possess eae, hly, stx1, or stx2 genes. Still, it is important to know the overall prevalence of each of the O-serogroups and the potential virulence genes in the feedlot production environment. The prevalence information gives at least a qualitative indicator of how common these potentially pathogenic organisms may be among the sampled population. The data for those pools where only one O-serogroup was identified give a general indication of the potential prevalence of virulence genes for the respective O-serogroups. By also looking at the pools where none of the seven O-serogroups was identified, the “background” prevalence of virulence genes can be assessed. Again, since the linkage of these virulence genes within the same organisms cannot be assessed, it remains unknown what combinations of virulence genes may be present for various organisms.

Conclusions

The data from this study will be helpful in designing future studies to evaluate the ecology of these and other E. coli O-serogroups, as they give some information about possible expected prevalence and distribution of the organism in the beef production environment, specifically, cattle feedlots. Information on expected prevalence will help in setting sample sizes for future studies. The limited information on seasonal patterns will also be critical for designing future studies to assess risk and evaluate appropriate mitigation strategies either preharvest or at harvest.

Footnotes

Acknowledgments

We gratefully acknowledge the technical assistance of Tanya Purvis, Amy Burklund, Lindsay Beardall, and Taghreed Mahmood of the Kansas State Veterinary Diagnostic Laboratory.

Disclosure Statement

No competing financial interests exist.

References

Bai

, Paddock

, Shi

, Li

, An

, Nagaraja

. Applicability of a multiplex PCR to detect the seven major Shiga toxin-producing Escherichia coli based on genes that code for serogroup-specific O-antigens and major virulence factors in cattle feces. Foodborne Pathog Dis, 2012; 9:541–548.

[CDC] U.S. Centers for Disease Control and Prevention, Healthy People 2020. 2012. http://www.healthypeople.gov/2020/topicsobjectives2020/default.aspx. 2012 May 1.

CDC. Healthy People 2010: Understanding and Improving Health, 2 ^nd. Washington, DC: U.S. Government Printing Office, 2000.

Chapman

, Siddons

, Cerdan Malo

et al. A 1-year study of Escherichia coli O157 in cattle, sheep, pigs and poultry. Epidemiol Infect, 1997; 119:245–250.

[NCHS] National Center for Health Statistics. Health People 2010 Final Review, 2012. http://www.cdc.gov/nchs/data/hpdata2010/hp2010_final_review.pdf. 2012 February 16.

[FSIS] Food Safety and Inspection Service. Federal Register 58157, 76,182 September 20, 2011.

Galland

, Hyatt

, Crupper

, Acheson

. Prevalence, antibiotic susceptibility, and diversity of Escherichia coli O157:H7 isolates from a longitudinal study of beef cattle feedlots. Appl Env Microbiol, 2001; 67:1619–1627.

Hancock

, Besser

, Rice

et al. A longitudinal study of Escherichia coli O157 in fourteen cattle herds. Epidemiol Infect, 1997a. 118:193–195.

Hancock

, Rice

, Thomas

, Dargatz

, Besser

. Epidemiology of Escherichia coli O157 in feedlot cattle. J Food Prot, 1997b. 60:462–465.

10.

Kobayashi

, Shimada

, Nakazawa

, Morozumi

, Pohjanvirta

, Pelkonen

, Yamamoto

. Prevalence and characteristics of Shiga toxin-producing Escherichia coli from healthy cattle in Japan. Appl Env Microbiol, 2001; 67:484–489.

11.

Scallan

, Hoekstra

, Angulo

, Tauxe

, Widdowson

M-A

, Roy

et al. Foodborne illness acquired in the United States: Major pathogens. Emerg Infect Dis, 2011. https://dx-doi-org.web.bisu.edu.cn/10.3201/eid1701.P11101. 2013 February 24.

12.

Smith

, Blackford

, Younts

, Moxley

, Gray

, Hungerford

, Milton

, Klopfenstein

. Ecological relationships between the prevalence of cattle shedding Escherchia coli O157:H7 and characteristics of the cattle or conditions of the feedlot pen. J Food Prot, 2001; 64:1899–1903.

13.

VanDonkersgoed

, Graham

, Gannon

. The prevalence of verotoxins, Escherichia coli O157:H7 and Salmonella in the feces and rumen of cattle at processing. Can Vet J, 1999; 40:332–338.

14.

Woerner

, Ransom

, Sofos

, Dewell

, Smith

, Salman

, Belk

. Determining the prevalence of Escherichia coli O157 in cattle and beef from the feedlot to the cooler. J Food Prot, 2006; 69:2824–2827.