Seasonal Prevalence of Shiga Toxin-Producing Escherichia coli on Pork Carcasses for Three Steps of the Harvest Process at Two Commercial Processing Plants in the United States

Seven serogroups of STEC are responsible for most (>75%) cases of severe illnesses caused by STEC and are considered adulterants of beef. However, some STEC outbreaks have been attributed to pork products, although the same E. coli are not considered adulterants in pork because little is known of their prevalence along the pork chain. The significance of the work presented here is that it identifies disease-causing STEC, EHEC, demonstrating that these same organisms are a food safety hazard in pork as well as beef. The results show that most STEC isolated from pork are not likely to cause severe disease in humans and that processes used in pork harvest, such as scalding, offer a significant control point to reduce contamination. The results will assist the pork processing industry and regulatory agencies to optimize interventions to improve the safety of pork products.

ABSTRACT

Shiga toxin-producing Escherichia coli (STEC) is a foodborne pathogen that has a significant impact on public health, with strains possessing the attachment factor intimin referred to as enterohemorrhagic E. coli (EHEC) and associated with life-threatening illnesses. Cattle and beef are considered typical sources of STEC, but their presence in pork products is a growing concern. Therefore, carcasses (n = 1,536) at two U.S. pork processors were sampled once per season at three stages of harvest (poststunning skins, postscald carcasses, and chilled carcasses) and then examined using PCR for Shiga toxin genes (stx), intimin genes (eae), aerobic plate count (APC), and Enterobacteriaceae counts (EBC). The prevalence of stx on skins, postscald, and chilled carcasses was 85.3, 17.5, and 5.4%, respectively, with 82.3, 7.8, and 1.7% of swabs, respectively, having stx and eae present. All stx-positive samples were subjected to culture isolation that resulted in 368 STEC and 46 EHEC isolates. The most frequently identified STEC were serogroups O121, O8, and O91 (63, 6.7, and 6.0% of total STEC, respectively). The most frequently isolated EHEC was serotype O157:H7 (63% of total EHEC). Results showed that scalding significantly reduced (P < 0.05) carcass APC and EBC by 3.00- and 2.50-log₁₀ CFU/100 cm², respectively. A seasonal effect was observed, with STEC prevalence lower (P < 0.05) in winter. The data from this study show significant (P < 0.05) reduction in the incidence of STEC (stx) from 85.3% to 5.4% and of EHEC (stx plus eae) from 82.3% to 1.7% within the slaughter-to-chilling continuum, respectively, and that potential EHEC can be confirmed present throughout using culture isolation.

IMPORTANCE Seven serogroups of STEC are responsible for most (>75%) cases of severe illnesses caused by STEC and are considered adulterants of beef. However, some STEC outbreaks have been attributed to pork products, although the same E. coli are not considered adulterants in pork because little is known of their prevalence along the pork chain. The significance of the work presented here is that it identifies disease-causing STEC, EHEC, demonstrating that these same organisms are a food safety hazard in pork as well as beef. The results show that most STEC isolated from pork are not likely to cause severe disease in humans and that processes used in pork harvest, such as scalding, offer a significant control point to reduce contamination. The results will assist the pork processing industry and regulatory agencies to optimize interventions to improve the safety of pork products.

INTRODUCTION

Shiga toxin-producing Escherichia coli (STEC) are potential foodborne pathogens that, after ingestion, can cause severe damage to the intestinal mucosa and, in some cases, other internal organs of the human host (1–3). Certain STEC possess adherence systems, the most commonly observed being the attaching and effacing (A/E) lesion of enteropathogenic E. coli, which possess the intimin gene (eae) or the fimbria of enteroaggregative E. coli. By adhering to the intestinal lining and expressing Shiga toxin, these organisms can cause enterohemorrhagic diseases such as hemorrhagic colitis (HC) or the life-threatening condition of hemolytic uremic syndrome (HUS). There have been strains involved in HUS, however, that lack either of these adherence mechanisms; thus, there are other genes (not fully appreciated) that likely contribute to the virulence associated with severe foodborne illness caused by STEC. In this study, we distinguish enterohemorrhagic E. coli (EHEC) that possess eae from other STEC because these strains are responsible for most (>75%) cases of severe illnesses caused by STEC (3).

Since the early 1980s, E. coli O157:H7 has emerged as the EHEC serotype of the most significant public health relevance, not because of the incidence of the illness, which is much lower than that of other foodborne pathogens, e.g., Campylobacter or Salmonella, but because of the severity of the symptoms, the low infectious dose, and potential sequelae. Although the major source of STEC and EHEC is healthy ruminants, predominantly cattle, the increasing trend of foodborne outbreaks associated with E. coli O157:H7 (O157 EHEC) and non-O157 EHEC that were reported over recent years, both in the United States and European Union, were attributed to the consumption of pork (4, 5; https://www.foodsafetynews.com/2018/04/e-coli-outbreak-linked-to-edmonton-area-meat-shop).

In the United States, annual testing of meat and meat products by the U.S. Department of Agriculture (USDA) Food Safety and Inspection Service (FSIS) is designed to allow regular testing of products produced in domestic establishments, imported products, and raw ground beef in retail; the presence of O157 EHEC in samples of raw nonintact ground beef products and raw beef intended for raw nonintact products, including ground beef, raw ground beef components, and beef trimmings, is carried out on a regular basis (6). The annual testing scheme also includes testing of raw pork meat for the presence of O157 EHEC, non-O157 EHEC, and indicator microorganisms; 3,800 samples of raw pork meat were tested in 2018, e.g., comminuted pork, intact pork cuts, and nonintact pork cuts (6). In a recent report, of 1,395 pork samples examined by FSIS for STEC, 309 (22%) screened positive for the presence of stx and eae, but only 3 (0.2%) were confirmed by culture isolation (7). Unlike U.S. beef processors, U.S. pork processors do not conduct their own testing of products for E. coli O157:H7. At the moment in the European Union, the only existing microbiological criterion for STEC in a food commodity is defined in European Commission regulation (EC) no. 209/2013 amending regulation (EC) no. 2073/2005 regarding microbiological criteria for sprouts (8). The monitoring data on STEC in foods, other than sprouts and in animals, originate from the reporting obligations of the EU member states (9), which stipulates that member states must investigate the presence of STEC at the “most appropriate stage” of the food chain. Currently, harmonized epidemiological indicators (HEI) at the EU level do not exist, allowing EU member states to carry out sampling, testing, data analysis, and interpretation of results in a consistent manner.

In addition, the epidemiology and virulence factors of STEC and EHEC carried by on-farm pigs remain largely unknown. It is known that healthy pigs are important reservoirs of STEC (10), and some isolated strains were reported as potential human pathogens (11, 12). Since certain outbreaks of STEC and EHEC were associated with pork consumption (13–16; https://www.foodsafetynews.com/2018/04/e-coli-outbreak-linked-to-edmonton-area-meat-shop), it is important to obtain additional scientific evidence on pathways of pork contamination by serogroups able to infect humans (17).

Thus, the aims of this study were (i) to determine the seasonal prevalence of STEC and EHEC, as well as aerobic plate count (APC) bacteria and Enterobacteriaceae counts (EBC), on pork carcasses at three different steps of harvest; (ii) to further characterize isolated STEC and EHEC strains; and (iii) to discuss the results obtained with their relevance to food safety and to propose the most effective control options for prevention/minimization of pork carcass contamination.

RESULTS

APC and EBC.

Differences in the levels of APC and EBC of pork carcasses along the processing line at three points were observed between plants A and B (Table 1). During slaughter, the APC was higher (6.50-log₁₀ CFU/100 cm² in plant A and 6.93-log₁₀ CFU/100 cm² in plant B, respectively) on the carcass skin, while their numbers were significantly decreased (P < 0.05) following the scalding process (3.91-log₁₀ CFU/100 cm² in plant A and 3.53-log₁₀ CFU/100 cm² in plant B, respectively) and following final interventions when measured on chilled carcasses (2.48-log₁₀ CFU/100 cm² in plant A and 2.22-log₁₀ CFU/100 cm² in plant B, respectively). Carcass skin samples from plants A and B had EBC of 4.41 and 4.37-log₁₀ CFU/100 cm², respectively, while the carcasses showed significantly lower numbers of EBC after scalding (2.28-log₁₀ CFU/100 cm² in plant A and 1.50-log₁₀ CFU/100 cm² in plant B), and again in the chiller (0.88-log₁₀ CFU/100 cm² in plant A and 0.49 log₁₀-CFU/100 cm² in plant B) (P < 0.05).

TABLE 1.

APC and EBC^a on pork carcasses by sample site, processing plant, and season

Seasonb	Plant	APC count (log10 CFU/100 cm2)			EBC count (log10 CFU/100 cm2)
		Skinc	Postscaldd	Finale	Skin	Postscald	Final
	A	6.50b	3.91a	2.48a	4.41a	2.28a	0.88a
	B	6.93a	3.53b	2.22b	4.37a	1.50b	0.49b
Winter		6.27y	3.28x	1.92y	4.06y	1.66y	0.49y
Spring		6.79x	2.85z	1.80y	4.51x	1.85x	0.51y
Summer		7.85w	5.59w	3.15w	5.01w	2.56w	1.02w
Fall		5.95z	3.05y	2.53x	3.99z	1.77xy	0.73x

^aValues represent the mean log₁₀ CFU/100 cm² (n = 768 by plant and n = 384 by season); those followed by the same letter within the column for plant or season are not different (P > 0.05).

^bSeasons include winter from December to February, spring from March to May, summer from June to August, and fall from September to November.

^cSkin of stunned exsanguinated pigs sampled along belly midline.

^dPostscald pre-evisceration pig carcasses sampled along midline from ham to breast, including foreshank and jowl. Carcass samples are not matched to other samples.

^eFinal represents chilled finished pig carcasses, sampled along the split midline from ham collar to jowl and foreshank. Carcass samples are not matched to other samples.

Season significantly influenced (P < 0.05) skin contamination. Significantly higher APC and EBC were measured on carcass surfaces during summer (7.85- and 5.01-log₁₀ CFU/cm², respectively) than all other seasons, followed by spring (6.79- and 4.51-log₁₀ CFU/cm²) and winter (6.27- and 4.06-log₁₀ CFU/cm²), while the lowest number of these bacteria were found during fall (5.95- and 3.99-log₁₀ CFU/cm²). Although scalding significantly decreased the numbers of these bacterial groups, seasonal variations remained significant (P < 0.05). After all interventions, carcasses in the chiller had the lowest numbers of APC and EBC recorded during winter (1.92- and 0.49-log₁₀ CFU/cm², respectively) and spring (1.80- and 0.51-log₁₀ CFU/cm²), with no significant differences (P > 0.05) observed between these two seasons.

PCR screening of pork carcasses for STEC (stx) and EHEC (stx plus eae).

All samples were enriched then screened by PCR for Shiga toxin (stx) and intimin (eae). The presence of stx was considered to indicate the presence of STEC, while the concomitant presence of eae identified samples that potentially contained EHEC. Therefore, a sample that was PCR positive for stx and eae was included in both the potential STEC-positive and the potential EHEC-positive groups. In regard to STEC and EHEC screening of skins, postscald pre-evisceration carcasses, and final carcasses, seasonal and plant differences were observed (Table 2).

TABLE 2.

Prevalence^a,i of STEC^b and EHEC^c in samples collected from pork processing as determined by PCR^d

Seasone	Plant	No. of samples	% of STEC-positive samples			% of EHEC-positive samples
			Skinf	Postscaldg	Finalh	Skin	Postscald	Final
All		1,536	85.3	17.5	5.4	82.3	7.8	1.7
	A	768	81.3y	13.8y	8.2x	76.3y	7.7x	3.1x
	B	768	89.3x	21.2x	2.6y	88.3x	7.9x	0.3y
Winter		384	41.7r	20.3q	3.6qr	29.7r	9.6q	0.8qr
Spring		384	100.0q	11.2r	3.4r	100.0q	2.9r	0.0r
Summer		384	99.5q	19.0q	7.6q	99.5q	8.3q	3.4q
Fall		384	100.0q	19.5q	7.0qr	100.0q	10.4q	2.6qr
Winter	A	192	26.0c	12.5gf	5.2ih	6.3de	7.8cde	1.6gfh
	B	192	57.3b	28.1c	2.1i	53.1b	11.5c	0.0h
Spring	A	192	100.0a	8.3gh	5.2ih	100.0a	4.2efg	0.0h
	B	192	100.0a	14.1gef	1.6i	100.0a	1.6fgh	0.0h
Summer	A	192	99.0a	18.2def	13.0gf	99.0a	8.3 cd	5.7de
	B	192	100.0a	19.8de	2.1i	100.0a	8.3 cd	1.0gh
Fall	A	192	100.0a	16.2ef	9.4gh	100.0a	10.4c	5.2dfe
	B	192	100.0a	22.9dc	4.7ih	100.0a	10.4c	0.0h

^aValues represent percentages of each sample type in each category found positive.

^bSTEC are Shiga toxin-producing E. coli indicated by the presence of stx₁ and or stx₂ gene(s) in the sample.

^cEHEC are enterohemorrhagic E. coli indicated by the presence of Shiga toxin (stx) and intimin (eae) genes in the sample.

^dThe screening PCR identified stx₁, stx₂, and eae in the enriched samples.

^eSeasons include winter from December to February, spring from March to May, summer from June to August, and fall from September to November.

^fSkin of stunned exsanguinated pigs sampled along belly midline.

^gPostscald pre-evisceration pig carcasses sampled along midline from ham to breast, including foreshank and jowl. Carcass samples are not matched to other samples.

^hFinal represents chilled finished pig carcasses, sampled along the split midline from ham collar to jowl and foreshank. Samples are not matched to other samples.

ⁱValues within a group, STEC or EHEC, plant (columns), season (columns), or plant by season (columns and rows) followed by the same letter are not different (P > 0.05).

Overall, 85.3% of skin samples were positive for STEC, with plant A having a lower rate (P < 0.05) than plant B. Seasonally, nearly 100% of skin samples were positive year-round for STEC, except for the winter months when STEC prevalence was 41.7% (P < 0.05). During the winter, the prevalence of STEC at plant A was 26.0%, half that of plant B (57.3%). This winter difference was responsible for all other differences observed on skins.

Following scalding and singeing but before any further processing, 17.5% of the pre-evisceration carcasses were STEC positive. Again, plant A had a lower rate (13.8%) and was different (P < 0.05) from plant B (21.2%). The seasonal effect observed on these carcasses was different, however, from that of the incoming skins. While winter-month skins screened lower for STEC, spring postscald carcasses (11.2%) were lower (P < 0.05) than the other seasons (19 to 20%). The lowest postscald carcass STEC screen rate was observed at plant A in the spring (8.3%), while the highest was observed at plant B in the winter (28.1%). Just 5.4% of the final carcasses in the chillers at plants A and B combined were positive for STEC, with plant A having approximately a 3-fold greater STEC prevalence (P < 0.05) than plant B. Seasonally, summer final carcasses possessed the greatest number of STEC positives (7.6%), with the lowest (P < 0.05) number of STEC positives in the spring (3.4%). However, rates in the winter and fall, 3.6% and 7.0%, respectively, were not different (P > 0.05) from the summer and spring levels, respectively. The seasonally observed rates of STEC positive final carcasses at plant A ranged from 5.2 to 13.0%, while at plant B, they ranged from 1.6 to 4.7%.

Since potential EHEC-positive samples represent a subset of all STEC-positive samples, the prevalence of potential EHEC on skins and the carcasses was lower; however, the plant and seasonal differences were generally maintained. Pork skins that screened positive for both stx and eae were 82.3%, plant A (76.3%) and plant B (88.3%) being different (P < 0.05), and winter skins (29.7%) less (P < 0.05) than the other seasons (99.5 to 100%). Nearly all skin samples were positive for both markers, indicating the presence of potential EHEC, except in the winter, where only 6.3% of plant A and 53.1% of plant B skin samples screened positive for potential EHEC.

Of all postscald carcasses, 7.8% were positive for potential EHEC, with no difference observed (P > 0.05) between the two plants (7.7 and 7.9%). There was a seasonal effect that followed the STEC screening, with spring lower (2.9%; P < 0.05) than the three other seasons, which were not different (P > 0.05) from one another, ranging from 8.3 to 10.4% of samples positive for potential EHEC.

The EHEC prevalence for final carcasses was very low at only 1.7%, but with significant differences (P < 0.05) between plant A at 3.1% and plant B at 0.3%. No final carcasses were positive for EHEC in the spring months, whereas 3.4% of final carcasses did so in the summer months. This was the only seasonal effect observed among final carcasses. In a season-by-plant analysis, in plant B, only 1.0% of final carcasses were positive for EHEC in the summer, whereas 1.6% were EHEC positive in plant A during the winter, which was less than the summer rate of 5.7% and significantly less (P < 0.05) than the fall rate (5.2%).

Isolation of STEC and EHEC from pork processing samples.

The presence of an EHEC exclusive of STEC could only be confirmed by culture isolation, as the samples could have been cocontaminated by a STEC strain (possessing an stx gene) and an atypical enteropathogenic E. coli (EPEC strain possessing an eae gene). Therefore, all stx-positive samples were subjected to culture confirmation. In total, 405 samples were culture confirmed. Three hundred sixty of the samples yielded 368 different STEC isolates (Table 3), while 46 samples yielded 46 EHEC isolates (Table 4). One sample was culture confirmed to harbor both STEC and EHEC isolates. Most isolates were found in samples collected in the spring and summer months, 120 and 135, respectively, whereas only 67 winter samples and 92 fall samples were culture confirmed. O121 was the most common STEC serotype on skin and postscald carcasses, and O157 was the most common EHEC serotype.

TABLE 3.

Summary^a of STEC^b strains (n = 368) isolated from pork processing plants by sample type, season,^c and processing plant

Sample type	Season	Plant	No. of strains in STEC serogroup:
			O2	O5	O8	O20	O32	O55	O74	O86	O91	O103	O110	O112	O121	O139	O141	O146	ONTg
Skind
	Winter	A	1		15	7							1		15	1			2
		B					1				10				1	4		1	4
	Spring	A			1									1	26				5
		B			4			7			5				50	5	2		13
	Summer	A			1	1			1		3				17	1			1
		B		1	1	1					3	1			68			2	7
	Fall	A				1									6		1		1
		B			1										42	3			3
Postscald carcasse
	Summer	A			1										1
		B			1					1					1				3
	Fall	A								1					2
		B									1				2
Final carcassf
	Summer	A													1	1			3
		B
Total			1	1	25	10	1	7	1	2	22	1	1	1	232	15	3	3	42

^aValues represent the number of isolates recovered from samples within each category; only seasons with results are presented.

^bSTEC are Shiga toxin-producing E. coli lacking the intimin (eae) gene.

^cSeasons include winter from December to February, spring from March to May, summer from June to August, and fall from September to November.

^dSkin of stunned exsanguinated pigs sampled along belly midline.

^ePostscald pre-evisceration pig carcasses sampled along midline from ham to breast, including foreshank and jowl. Carcasses are not matched to other samples.

^fFinal represents chilled finished pig carcasses, along the split midline from ham collar to jowl and foreshank. Carcasses are not matched to other samples.

^gONT serogroup was not typeable using limited antisera sets available.

TABLE 4.

Summary^a of EHEC^b (n = 46) isolated from pork processing plants by season^c and processing plant

Season	Plant	No. of strains in STEC serogroup:
		O8	O26	O103	O121	O157	ONTd
Winter	A			2
	B						2
Spring	A						1
	B
Summer	A		2e
	B				2	8	1
Fall	A	1			3	15	1e
	B		1		1	6e
Total		1	3	2	6	29	5

^aValues represent number of EHEC isolates of the given serogroup recovered from samples that screened positive for Shiga toxin genes by PCR.

^bEHEC are enterohemorrhagic E. coli possessing Shiga toxin (stx) and intimin (eae) genes.

^cSeasons include winter from December to February, spring from March to May, summer from June to August, and fall from September to November.

^dONT serogroup was not typeable using limited antisera sets available.

^eAll isolates were recovered from pork skin swab samples except the 2 EHEC O26 (plant A, summer) that were recovered from final pork carcasses, 1 EHEC ONT (plant A, fall) recovered from a preintervention carcass, and 1 EHEC O157 (plant B, fall) recovered from a preintervention carcass.

As suggested by the PCR screening results, samples collected from skins yielded the most STEC and EHEC isolates (Tables 3 and 4). Plant B had about twice as many skin samples culture confirmed with STEC (n = 240) compared to plant A (n = 109), but both plants had a similar number of skin samples culture confirmed as EHEC (25 and 21 for plants A and B, respectively). Samples collected in the spring and winter months only yielded 4 and 1 as EHEC, respectively, with the bulk of the isolated EHEC being found in the summer and fall (Table 4).

Nearly two-thirds (64.4%) of the STEC isolated from skins were STEC O121. STEC with nontypeable serogroups were second most common (10.5%). These two groups of STEC were the only ones found at both plants every season. Other STEC identified at both plants and/or during every season were STEC O8, O91, O139, and O20 (Table 3). The most common EHEC isolated from skins was EHEC O157:H7, which made up 63.0% of the EHEC isolates from skins. EHEC O157:H7 was found at plant B in the summer and both plants in the fall. The next most common EHEC isolated from skin samples was EHEC O121. It, too, was isolated in a similar pattern as that of EHEC O157:H7. Other EHEC isolated from skins were O8, O26, O103, and O nontypeable (Table 4).

For postscald pre-evisceration carcasses, 17.5% were PCR positive for STEC and culture confirmed at a rate of 0.9%, while 1.7% were PCR positive for EHEC, but only 0.1% were culture confirmed to carry EHEC. All isolates from postscald carcasses were only recovered from samples collected in the summer and fall months. These were the seasons with some of the highest PCR-positive rates. A third fewer STEC were found at plant A in the summer than at plant B. However, STEC O8 and STEC O121 were present at both plants in the summer. Similar numbers of STEC isolates were found at each plant in the fall, again with STEC O121 being the most common. One EHEC was isolated from the postscald carcasses at each plant in the fall. These isolates were an EHEC O157:H7 at plant B and an EHEC ONT at plant A.

Only 5 STEC isolates were recovered from final carcasses: STEC O121, STEC O139, and 3 STEC ONT isolates recovered from plant A during the summer. Only 2 EHEC O26 isolates were culture confirmed from final carcasses, similar to results from plant A during the summer. No isolates were recovered from final carcasses at plant B during the summer or during any other season. The recovery of isolates agrees with the PCR screening results, being highest for plant A in the summer at 13.0% and 5.7% for STEC and potential EHEC, respectively.

Characterization of STEC isolates.

Of the 367 STEC isolates, 6 were recovered from postscald carcasses and 1 from a final carcass, while the remaining 360 isolates were found on prescald carcass skins. STEC O121 made up 63% of the isolates (Table S1 in the supplemental material). Eighteen variations were observed based on the presence of the different virulence factors examined. Seven of the genotypes were unique isolates, whereas multiple isolates of similar genotypes numbered in groups of 2 to 163. In the case of 6 genotypes, the identical isolates were found across plants and seasons. However, 1 genotype represented by 163 isolates was recovered from skin samples at plant A during the spring. All but 7 of the STEC O121 isolates (6 from skin and 1 from postscald carcass) possessed Shiga toxin 2 subtype e (stx_2e). Two isolates carried a Shiga toxin subtype 1a (stx_1a) allele in addition to the stx_2e allele. Only 5 STEC O121 possessed what appeared to be incomplete pO157 plasmids. All five carried katP, while two also possessed etpD, with one of those also having espP. Most of the STEC O121 isolates carried an allele of eastA, and a small number also possessed iron acquisition genes. Two STEC O121 isolates possessed the adherence factor gene saa; these were found at plant B in the fall and plant A in the winter.

The remaining STEC isolates (n = 134) were of 15 serogroups and a large group (n = 41) of nonidentified serogroups (this was due to our limited serotyping antisera). The identified serogroups included O2, O5, O8, O20, O32, O55, O74, O86, O91, O103 (an intimin lacking STEC), O110, O112, O139, O141, and O146. These STEC non-O121 isolates (Tables S2 and S3) also predominantly had stx_2e. stx_1a was the lone Shiga toxin in 21 isolates of serogroups O20, O32, O91, O110, O112, and ONT. Shiga toxin subtypes 2a (stx_2a) and 2c (stx_2c) were uncommon, observed in only 2 isolates, a STEC O8 and a STEC ONT, respectively. Six isolates had stx₂ of nonidentifiable subtypes. In most cases, stx occurred as a single allele except for a STEC O8 possessing stx_2e and stx_2a, a STEC O32 with stx_1a and stx_2x, and STEC ONTs that possessed combinations of stx_1a with stx_2x, stx_2c with stx_2x, and stx_1a with stx_2e.

Incomplete variations of the pO157 plasmid were observed in multiple isolates. Eight STEC O91 isolates possessed the pO157 markers hlyA and katP, and these were the two most common of the pO157 markers identified in the STEC isolates (30 had katP and 11 had hlyA). One STEC O8 isolate had three pO157 markers present (katP, espP, and etpD) and represented the most complete pO157 plasmid within the non-O121 STEC isolates. In regard to other virulence factors, 2 isolates, a STEC O8 and a STEC O86, possessed the gene for cytotoxic necrotizing factor (cnf). Multiple strains had alleles of eastA, while iron acquisition genes iha and chuA were observed in isolates of STEC O8, O20, O55, O86, O91, and O139. Fourteen of the STEC ONT lacked these additional factors, while the rest possessed 2 or more of them.

Characterization of EHEC isolates.

The EHEC isolates were divided into E. coli O157:H7 (n = 29; Table S4) and non-O157 EHEC (n = 17; Table S5). The 29 E. coli O157:H7 isolates, compared for Shiga toxin types, nle effectors, composition of the pO157 plasmid, and other toxin, adherence, and iron utilization genes, all impacting virulence, resulted in 12 different genotypes (Table S4).

Twelve of the 29 E. coli O157:H7 isolates possessed identical gene patterns and were found across seasons and between the two plants. All the E. coli O157:H7 possessed stx₁ and stx_2a, but 3 isolates also carried the stx_2e allele. All E. coli O157:H7 isolates appeared to possess an intact pO157 plasmid as evidenced by the presence of hylA, katP, espP, and etpD, which are spaced around the plasmid. The iron utilization genes chuA and iha were also present in all of the E. coli O157:H7 isolates. The primary differences between the E. coli O157:H7 strains involved differences in the presence of the nle genes nleA, nleG2-3, and nleG9, as well as cytotoxic necrotizing factor (present in 3) and E. coli heat-stable enterotoxin 1.

Non-O157 EHEC (n = 17) were of 4 identifiable serogroups (O8, O26, O103, and O121), with 5 isolates having a nontypeable serogroup (Table S5). The non-O157 EHEC is divided into 15 groups based on genetic composition. These EHEC isolates possessed different complements of Shiga toxin alleles, stx_1a, stx_2a, stx_2c, and stx_2e. Three of the most frequent non-O157 STEC serogroups recognized by the CDC (1) and FSIS (18) were identified (O26, O103, and O121), each possessing the expected eae subtypes of β1 and ε; however, 2 of the EHEC O121 isolates had an eae gene that could not be subtyped using our primer sets, suggesting that it may be something other than eae-ε. Intimin γ was observed in one EHEC ONT. This isolate may be an EHEC O145 that lacks the chromosomal region our serogrouping PCR identifies. This strain did not appear to have rfb_O157 or fli_CH7 by PCR and was a sorbitol fermenter (data not shown), suggesting it is not likely E. coli O157:H7.

Variable numbers of nle genes were observed in the EHEC isolates, with EHEC O8 and 2 of the EHEC ONT possessing only 1 to 3 of the effectors (Table S5). The 2 EHEC O103 lacked many of the nle genes in comparison to the EHEC O26s. Two of the EHEC O121 and one of the EHEC ONT possessed nearly all of the nle genes. Intact and partial pO157 plasmids were identified in the non-O157 EHEC. An EHEC O26, 4 O121, and an ONT all appeared to possess a complete plasmid, while other isolates had incomplete versions. One EHEC ONT lacked all markers for the pO157 plasmid. In regard to other factors, the lifA gene was only present in one EHEC O26 found at plant A during the summer. Cytotoxic-necrotizing factor, E. coli heat-stable enterotoxin, and iron acquisition factors (iha and chuA) were variably present in all but four of the non-O157 EHEC isolated from pork carcasses.

DISCUSSION

The present study identified STEC and potential EHEC on the skins of prescald pork carcasses in two U.S. commercial hog processing plants. Contamination of pigs with pathogenic EHEC O157 and non-O157 may have occurred at farms (feed, water, or manure), during transport, or in lairage. Available data show that some EHEC O157 strains may persist for more than 2 years in the farm environment (19). In addition, the tonsils of some pigs have been reported to be colonized by significant levels of E. coli O157:H7 (20). The significantly higher (P < 0.05) STEC and EHEC prevalence on prescald carcasses sampled at plant B could be due to higher contamination at any of the steps prior to slaughter, or potentially the “all in-all out” method of pork production where each farm empties a full facility for slaughter. However, determination of the source of this contamination was not the aim of the present study.

The results obtained in our study showed a very high prevalence of the stx gene(s), indicating STEC (85.3%), and stx and eae indicating EHEC (82.3%) on the skin of pigs at slaughter. Nevertheless, a significant decrease in prevalence of these genetic markers was observed after scalding in the present study. Other authors reported the effectiveness of the scalding stage on reducing E. coli and coliform counts on pork carcasses (21, 22). This important step is usually a critical control point within a risk-based food safety management system (hazard analysis and critical control points [HACCP]) and reduces both bacterial numbers and the prevalence of pathogens (21).

APC bacteria are generally used to assess the hygiene of meat processing (23), and EBC are also used as indicators of fecal contamination (24, 25). The results of the present study showed that scalding is effective in reducing bacterial contamination on the carcass. Furthermore, our results are in line with previous reports showing that scalding (59 to 62°C) of pork carcasses resulted in a reduction of APC (21, 26, 27). In other experiments, scalding reduced APC and EBC by 3.1- to 3.8- and 1.7- and 3.3-log₁₀ CFU/100 cm², respectively (21, 26) which is similar to results found here (up to 3.4-log₁₀ CFU/100 cm² and 2.87-log₁₀ CFU/100 cm²).

Unfortunately, epidemiological data on STEC prevalence in different regions and studies are not always comparable due to differences in study designs, sampling, and methods applied for detection and isolation, as well as the season in which the study was performed (10, 17, 28). In Italy, Ercoli et al. (10) reported a STEC prevalence of 13.8% on pork carcasses before chilling, while in Belgium, the prevalence of this pathogen was 12.8% on carcasses after cutting and before chilling (29). In the present study, the prevalence of STEC after scalding ranged between 13.8% (plant A) and 21.2% (plant B). Moreover, the data from the present study also showed a significant (P < 0.05) reduction in the incidence of STEC, indicated by the stx gene(s), from 85.3% to 5.4% and of EHEC, indicated by stx and eae genes, from 82.3% to 1.7% within the slaughter-to-chilling continuum, respectively. Colello et al. (28) found that 4.08% of pork carcasses sampled were stx positive in a study carried out in Argentina. A similar prevalence of STEC as in the present study (5.4%) was also found in carcasses after cooling in a Canadian study (4.8%) (30).

Since the complete elimination of carcass surface bacteria is not possible, chilling as a standard operating procedure has the objective, in general, to reduce carcass surface temperature, thereby preventing and slowing microorganism growth (31, 32). In the present experiment, significant differences (P < 0.05) in carcass APC and EBC after chilling were observed between the two plants. These findings may be attributed to differences in chilling systems used by the plants. Although the incoming microorganism load on skins was higher at the beginning of harvest, at the end, lower levels of APC and EBC and a lower incidence of STEC were found in plant B (2.22- and 0.49-log₁₀ CFU/100 cm², 0.3%, respectively) where blast chilling was used, compared to conventional chilling in plant A (2.48- and 0.88-log₁₀ CFU/100 cm²; 3.1%, respectively). Blast chilling, in comparison with conventional chilling, lowers the carcass temperature at a high rate, resulting in the arrest of bacterial growth when the population is smaller. In addition, blast chilling may provoke cold shock, especially in particularly sensitive Gram-negative microorganisms, including E. coli and other Enterobacteriaceae species, whereas, with conventional chilling, microorganisms may have the opportunity to adapt to lower temperatures and avoid cold shock (33). However, the final carcasses that were sampled were not linked to the postscald carcasses and were, in fact, from hogs harvested the previous days. The average reduction of APC from postscald to final carcasses was not different (P > 0.05) between the two plants, while the reduction of EBC between these two points was significantly greater (P < 0.05) at plant A (data not shown). Therefore, the significantly different microbial counts observed on carcasses in the chiller was likely a combination of the interventions applied as carcasses entered the chiller and the chilling process itself.

A lactic acid treatment following the final carcass water wash was applied as carcasses entered the chiller. It is well-known that the combination of water and lactic acid treatment provides the greatest microbial reduction without large negative effects on quality attributes of pork meat (34, 35). As mentioned, in the present study, carcasses in both plants were treated with 2% lactic acid (ambient-temperature water, 10 to 30 s) before the cooling step. If the initial counts are higher, as in the present study, the effect of lactic acid decontamination treatment is more evident (35). Ba et al. (27) observed that significantly higher reductions in all bacterial species on pork carcasses were achieved when sprayed with 4% lactic acid. Kalchayanand et al. (36) reported a significant decrease of STEC O26, O45, O103, O111, O121, O145, and O157 in inoculated fresh beef after lactic acid treatment.

Results regarding seasonal effect observed in the present study should be interpreted with caution because the visits to the plants were only carried out on two consecutive days during each period. It was observed that there were significant increases (P < 0.05) in APC and EBC during the summer and spring compared to winter and fall. However, STEC prevalence indicated by stx genes on the skin of pigs at harvest was high (99 to 100%) and did not differ between spring, summer, and fall (P > 0.05). Only during winter was there a significantly lower prevalence (P < 0.05) of this pathogen indicator (stx) than during other seasons. Essendoubi et al. (25) also found a higher prevalence of STEC on beef carcasses during warmer months (from June to November), while Dawson et al. (37) reported higher E. coli O157:H7 colonization in cattle during warmer months than during cooler times of the year in various cattle production systems. One possible explanation may be that animals are dirtier during summer months due to soil and fecal contamination (32, 38, 39). In contrast, Cha et al. (40) reported higher STEC prevalence in pigs during fall and winter months (36.16% and 19.72%, respectively), suggesting that low temperatures may contribute to increased stress in pigs, leading to lower immunity and increased susceptibility to new STEC infections. The seasonal variations observed require further investigation as in the United States. Pigs are finished indoors in temperature-controlled facilities and not directly exposed to colder temperatures in winter.

EHEC are important pathogens of public health significance because these isolates possess not only stx₁ and/or stx₂ but also eae, the gene for the adherence factor intimin. Intimin, an integral outer membrane protein, is required for adherence to enterocytes, inducing a characteristic histopathological A/E lesion and has been considered a risk factor for disease in humans (28, 41). Although the presence of the eae gene is an aggravating factor, this virulence factor is not always essential for severe illness, suggesting that there may be alternative mechanisms for attachment (3). One such additional adherence factor we observed in a small number of STEC was the STEC autoagglutinating adhesin (indicated by presence of the saa gene), which has been identified in STEC isolated from humans with HUS or diarrhea (42).

The strains that possess stx₁ and stx₂ genes are often associated with HUS (43, 44). In the present study, the strains possessing stx₂ accounted for 88.74% of the total STEC isolates and 59.58% of all isolates (data not shown). While most stx₂ genes were subtype 2e, there were isolates the possessed stx_2a and stx_2c, both major subtypes produced by E. coli strains associated with HUS (44). Strains that have stx_2e do not consistently provoke foodborne illness in humans (45), but other data have confirmed the isolation of stx_2e-associated STEC from an HUS patient (46). With the exception of 8 STEC O121 that had an unidentified stx₂ subtype, the remaining STEC O121 only possessed stx_2e. STEC containing subtype stx_2e are typical swine-adapted STEC and present the most frequently reported Shiga toxin subtype from pigs (40, 47). This subtype is responsible for porcine edema disease in pigs (45) and, consequently, economic losses in production (12, 28). The significance of the unidentified stx₂ subtypes (as well as eae subtypes) upon the virulence of the isolates is unknown. We used previously validated subtyping PCRs (48); however, alternate approaches utilizing whole-genome sequencing (WGS) could likely resolve this issue and are an avenue for future work.

EHEC serogroups isolated in the present study included O26 (3), O103 (2), O121 (5), and O157 (28). The USDA FSIS has declared the so-called “big six” non-O157 serogroups (O26, O45, O103, O111, O121, and O145) as adulterants in beef (18). These serotypes present a public health burden because they are linked to a significant number of HC and HUS cases (1, 49, 50). The European Food Safety Authority (3) has made a similar declaration for serogroups with a high pathogenicity potential (O157, O26, O103, O145, O111, and O145). Therefore, in the present study, the STEC serogroups of public health importance that were isolated were O157 and O103 (3) and O157, O26, O103, and O121 (18). Our approach to STEC and EHEC isolation did not use immunomagnetic separation (IMS), which could have concentrated these select serogroups and potentially increased their isolation rate. We avoided this method in favor of direct plating onto washed sheep blood agar containing mitomycin (wSBAm), a STEC and EHEC indicator medium that allowed us to focus on isolation of all possible STEC and identify the relative abundance of EHEC among the STEC.

Most of the EHEC isolates found in the present study were O157:H7 (28) and were isolated from both plants during summer and fall. Serotype O157:H7 causes the most severe clinical symptoms in humans. Although pork is not a common vehicle of EHEC O157, some outbreaks in the United States, Canada (14–16, 51), and Italy (52) have been linked to consumption of roasted pork meat and salami containing pork. Serogroup O121 was the most prevalent non-O157 serotype found among pork carcasses. STEC O121 was previously linked with many outbreaks (4). Before the advent of WGS, a common tool used for tracking E. coli O157:H7 and the non-O157 STEC had been pulsed-field gel electrophoresis (PFGE). Using PFGE may have allowed us to identify strains with similar restriction digest patterns (RDPs), while using WGS analysis would allow identification of related strains based on single-nucleotide polymorphisms. Further investigation of all the EHEC isolated in the current study using WGS is warranted.

The potential of other strains isolated in our study to cause illness in humans should not be excluded. Serotypes O8 (1 EHEC- and 25 STEC-containing samples), O91 (22 STEC-containing samples), O139 (15 STEC-containing samples), O20 (9 STEC-containing samples), and O55 (7 STEC-containing samples) were recovered. E. coli O8 possessing stx_2e has been reported to cause acute diarrhea (53), while O91 STEC strains can cause HUS or HC, although they are eae negative (54). In addition, O8, and O91 were included in the 20 most frequent serogroups reported in confirmed cases of human STEC infections in the European Union/European Economic Area from 2015 to 2017 (3).

The results of the present study, observed with sampling only in two plants in the central part of the United States, showed that pigs carry a variety of different STEC and EHEC serotypes. Some of those serotypes are of high public health importance (e.g., O157 and O121); cross-contamination can occur during processing and dressing, and interventions applied before chilling have an important role in the reduction of microbial loads (APC and EBC) and prevalence of STEC and EHEC.

The presence of different STEC and EHEC serogroups on market pigs in this study was found in decreasing order (O157, O121, O8, O91, O139, O20, and O55), indicating that this could be the way of introducing them into the processing plant environment. Results showed that pork skin may be a significant source of EHEC and STEC in pork meat. The highest APC and EBC levels on pork skins were found during spring and summer, while the prevalence of genetic markers indicating the presence of STEC and EHEC was significantly less during winter. Hygienic processing at both plants significantly reduced contamination on carcasses, regardless of season. Postscald carcasses showed that STEC prevalence (indicated by the presence of stx) was significantly decreased by 80 to 90%, which makes this processing step key to contaminant reduction. Important control measures included decontamination of pork carcasses with 2% lactic acid applied before chilling. Since the results from the present study showed a higher prevalence of STEC and EHEC during spring, summer, and fall than winter, a risk-based food safety management system should be implemented during these three seasons to achieve beneficial effects in reducing the pathogen prevalence on pork carcasses. Further in-depth studies are needed to understand the sources of STEC and EHEC carried by pigs presented for harvest, cross-contamination of pork carcasses in the processing plant, and the impact of blast chilling on arresting the growth of bacterial contaminants on pork carcasses.

MATERIALS AND METHODS

Meat establishments.

Sample collection was conducted in two establishments (plant A and plant B) approved for export of pork meat and deli meat products to foreign markets by the USDA Food Safety and Inspection Service (FSIS). The selected meat establishments were two large U.S. commercial hog processing plants that harvested 11,000 to 17,000 hogs/day. The harvest process and dressing operations followed standard procedures of stunning, exsanguination, prescalding wash, scalding at 60°C, dehairing, singeing, polishing, pre-evisceration wash, evisceration, carcass splitting, trimming, final wash, and chilling (final carcass and cooler temperature was 4°C for 16 to 24 h) (Fig. 1). Plants A and B had different chilling systems, conventional and blast chilling systems, respectively.

FIG 1.

Standard operational procedures (SOPs) and sampling sites of a pork slaughter line.

Sample collection.

The sampling protocol targeted the incoming contaminants on skins and then examined carcasses at two relevant locations: postscald preintervention carcasses and finished carcasses after chilling, thereby identifying along the harvest line where pork carcasses may have been cross-contaminated with microbes, including STEC and EHEC (Fig. 1).

The sampling was carried out quarterly throughout the year, covering four seasons, e.g., QI, winter (December to February), QII, spring (March to May); QIII, summer (June to August); and QIV, fall (September to November). Each plant (designated plant A and plant B) was visited once per season, and carcass samples were collected over two consecutive days on each trip, totaling eight sampling days per plant per year, for a total of 16 sampling days per year for two plants. On each sampling day, 95 samples were taken from three sampling points along the harvest line, including skin of stunned exsanguinated prescald carcass, postscald pre-evisceration carcass, and chilled final carcass. In total, 1,536 samples were collected over the course of the study, with 384 samples in each season (winter through fall).

Samples were collected as described previously using moistened cellulose sponges (Whirl-Pak; Nasco, Fort Atkinson, WI), prewetted with 20 ml of buffered peptone water (BPW; Difco, Becton, Dickinson, Franklin Lakes, NJ) (55). To prevent cross-contamination, gloves were worn during sampling and were changed following each sample.

Samples from the skin of prescald carcass surfaces were obtained by using both sides of the prewetted sponge to swab an area of approximately 1,500 cm² along the belly midline. After scalding, singeing, and polishing of the carcass, pre-evisceration postscald carcass samples were obtained by using both sides of the prewetted sponge to swab approximately 4,000 cm² of the carcass surface along the midline from ham to sternum, including foreshank and jowl. Final carcass samples were obtained from carcasses that had been chilled at least overnight in coolers at 4°C by using both sides of the sponge to swab approximately 4,000 cm² of the carcass surface along the split midline from ham collar to jowl and foreshank. Due to the intense processing speed, in-plant operations, and safety considerations for personnel collecting the samples, only a convenience sample was collected; therefore, samples taken from each point were not matched to specific animals or groups of animals at other points. Skin and postscald carcass samples were collected at the same time, while final carcass samples were collected after 24 h of chilling from carcasses harvested on the previous day. All samples were transported in coolers with ice packs (at <4°C), received, and processed at the U.S. Meat Animal Research Center (Clay Center, NE, USA) within 24 h of collection according to the protocol described by Schmidt et al. (55). The levels of APC (56) as hygiene-level indicators, EBC (57) as indicators of fecal contamination, and STEC non-O157 as a foodborne pathogen (58) were determined.

Sample processing.

Each sponge swab was massaged by hand to ensure it was thoroughly mixed; then 1 ml was removed for APC and EBC. Eighty milliliters of tryptic soy broth (TSB; Difco, BD) was added to the remainder of the sample and sponge to enrich the samples for STEC. Enrichment consisted of incubation in a programmable incubator at 25°C for 2 h and 42°C for 6 h, then held at 4°C until processed. After enrichment, two 1-ml portions of each sample were removed for STEC screening and analysis, with one of the portions archived as a frozen (−70°C) 30% glycerol stock.

Screening for Shiga toxin genes.

One hundred microliters of an enrichment were placed in a microcentrifuge tube and used to prepare a crude DNA boil prep lysis (59). Two microliters of the DNA preparation were placed into separate 25-μl multiplex PCRs that detected stx₁, stx₂, eae, and ehx, which was performed as previously described (60). Products of the PCR amplifications were separated by agarose gel electrophoresis, stained using ethidium bromide, and then photographed and interpreted for the presence of the four possible reaction products. Enrichments that had stx₁ and/or stx₂ were considered positive for STEC, while enrichments that had eae and stx₁ and/or stx₂ were considered positive for EHEC for use in prevalence calculations.

Isolation of STEC and EHEC.

The sample enrichments determined by PCR to contain stx₁ and/or stx₂ were assayed by spiral plating of samples onto plates of washed sheep blood agar containing mitomycin (wSBAm) (61). Each enrichment was serially diluted to 1:500 and 1:5,000 in cold (4°C) BPW. Fifty microliters of each dilution were spiral plated onto wSBAm plates. The plates were incubated overnight at 37°C and then viewed on a white light box for the suspect enterohemolytic phenotype as a thin zone (≤1 mm) surrounding the colony (62). In addition, if other hemolytic phenotypes such as alpha, beta, or gamma hemolysis were present, additional colonies representative of each hemolytic phenotype were picked for screening. A minimum of 4 colonies (if colonies were present) and a maximum of 6 colonies per sample were picked and placed into individual wells of 96-well screening plates containing 100 μl TSB per well. After suspect colonies were picked, the wSBAm plates were placed at 4°C. The 96-well screening plate was incubated at 37°C overnight and then screened by PCR as described above. If at least one suspect colony from a sample did not contain stx₁ and/or stx₂, the wSBAm plates were removed from 4°C and subjected to another round of suspect colony picking. All stx-containing isolates were checked for purity by streaking for isolation on sorbitol MacConkey agar containing 5-bromo-4-chloro-3-indolyl-β-d-glucuronide (SMAC-BCIG; Oxoid-CM0981; Remel Inc., Lenexa, KS) and then transferred to tryptic soy agar (TSA; Difco, BD) plates for characterization.

Characterization of isolates.

All stx-containing isolates (STEC) and stx- and eae- containing isolates (EHEC) were confirmed to be E. coli by biochemical assays using Fluorocult LMX broth (Merck KGaA, Darmstadt, Germany) and API 20E strips (bioMérieux Inc., Hazelwood, MO), both used according to the recommendations of the manufacturers. Once an isolate was established as being a STEC or EHEC, its serotype was determined by molecular and serologic identification of the O serogroup. PCR was used for molecular identification of O groups O26, O45, O55, O103, O111, O113, O117, O121, O126, O145, and O146 as described previously (48). E. coli antisera (Cedarlane, Burlington, NC) were used to confirm the PCR results and identify other O serogroups. Virulence genes of each STEC or EHEC isolate were determined by PCR as described previously (48). Shiga toxin subtypes of the isolates were identified to be stx_1a, stx_1c, stx_2a, stx_2c, stx_2d, and stx_2e. If an stx subtype could not be identified, the isolate was simply identified as “stx₁” or “stx₂.” Intimin (eae) subtypes α1, α2, β1, β2, γ, δ, ε, θ, and ζ were identified by PCR as described previously (48), and if an eae subtype could not be identified for an isolate, it was referred to as eae. The presence of four genes associated with the large 60-MDa virulence plasmid, toxB, espP, katP, and etpD; additional toxin-encoding genes (subA, lifA, cnf, and astA), adherence-encoding genes (iha and saa), and hemolysin genes (hylA and chuA) were identified among the isolates by PCR as described previously (48). Lastly, genes described for molecular risk assessment associated with E. coli O157:H7 O-islands 36, 57, 71, and 122 (nleB, nleE, entG2-3, G5-2, G6-2, nleC, H1-1, nleB2, nleG, nleG9, nleF, H1-2, nleA, and G2-1) were identified by PCR as described previously (48).

Statistical analysis.

Results from the enumeration (APC and Enterobacteriaceae count) of bacterial groups were analyzed for each sample type (skin, postscald carcass, and final carcass) using analysis of variance with the GLM procedure of SAS. The model included main effects of season and plant. For significant main effects (P ≤ 0.05), least-squares means separation was carried out with the PDIFF option (a pairwise t test). The data for enumerations were log transformed before the analysis of variance. Pairwise comparisons of frequencies were made using the PROC FREQ and Mantel-Haenszel chi-square analysis of SAS.

Sample enrichments were sorted according to serotype and screening PCR-positive reaction patterns (stx₁, stx₂, and eae), and comparisons of prevalence were examined using a one-way analysis of variance (ANOVA) and the Bonferroni multiple-comparison posttest. Comparisons of median values of the data sets were made using the Kruskal-Wallis test for nonparametric data and Dunn’s multiple-comparison posttest. For data sets with only two groups of values, comparisons were made using either a two-tailed unpaired t test or the Mann-Whitney U test for nonparametric data. For cases when pairwise differences were made, the DIFFER procedure of PEPI software (USD, Inc., Stone Mountain, GA) was used. In all cases, significance was defined at a P value of  ≤0.05.