Loop-Mediated Isothermal Amplification Assays for Detecting Shiga Toxin-Producing Escherichia coli in Ground Beef and Human Stools

Abstract

Shiga toxin-producing Escherichia coli (STEC), encompassing E. coli O157 and non-O157 STEC, is a significant cause of food-borne illnesses and deaths in the United States and worldwide. Shiga toxins (encoded by stx) and intimin (encoded by eae) are important virulence factors for STEC strains linked to severe human illnesses such as hemorrhagic colitis and hemolytic-uremic syndrome. In this study, the stx₁, stx₂, and eae genes were chosen as targets to design loop-mediated isothermal amplification (LAMP) assays for the rapid, specific, sensitive, and quantitative detection of STEC strains. The assay performances in pure culture and spiked ground beef and human stools were evaluated and compared with those of quantitative PCR (qPCR). No false-positive or false-negative results were observed among 90 bacterial strains used to evaluate assay specificity. The limits of detection for seven STEC strains of various serogroups (O26, O45, O103, O111, O121, O145, and O157) were approximately 1 to 20 CFU/reaction in pure culture and 10³ to 10⁴ CFU/g in spiked ground beef, which were comparable to the results of qPCR. Standard curves generated suggested good linear relationships between STEC cell numbers and LAMP turbidity signals. When applied in ground beef samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of STEC cultures, the LAMP assays achieved accurate detection after 6 to 8 h enrichment. The assays also consistently detected STEC in human stool specimens spiked with 10³ or 10⁴ CFU/0.5 g stool after 4 h enrichment, while qPCR required 4 to 6 h. In conclusion, the LAMP assays developed in this study may facilitate rapid and reliable identification of STEC contaminations in high-risk food commodities and also facilitate prompt diagnosis of STEC infections in clinical laboratories.

INTRODUCTION

Shiga toxin-producing Escherichia coli (STEC) is a zoonotic food-borne pathogen of significant public health concern due to its frequent involvement in outbreaks of hemorrhagic colitis (HC) and ability to cause life-threatening complications such as hemolytic-uremic syndrome (HUS) and thrombotic thrombocytopenic purpura (TTP) (38). In the United States, STEC causes an estimated 176,000 illnesses, 2,400 hospitalizations, and 20 deaths annually (36). Ruminants, particularly cattle, are the major reservoirs for STEC strains (23). STEC transmission commonly occurs through consumption of contaminated food (ground beef, produce, milk, juice) and water, through contact with animals, and from person to person (15). Less than 100 organisms of some STEC serotypes can lead to human illness (38).

First recognized as a food-borne pathogen in 1982 (27), E. coli O157:H7 remains the most common STEC serotype causing human illness (36). However, the clinical significance of non-O157 STEC is on the rise worldwide, with well over 100 serotypes associated with sporadic and epidemic human infections (25). For the first time since 2000, FoodNet in the United States actually reported a higher incidence of laboratory-confirmed non-O157 STEC infections than STEC O157 infections in 2010 (7). O26, O45, O103, O111, O121, and O145 are the top 6 non-O157 serogroups in the United States (4), whereas additional ones are more prevalent in other countries (25). Since May 2011, an unprecedented large outbreak of E. coli O104:H4 in Germany has resulted in a total of 4,075 cases (including 908 cases of HUS) and 50 deaths as of 21 July 2011 (41). Given this emerging and evolving nature of STEC serotypes involved in human illness, it is crucial that rapid and reliable detection methods be available to screen for all STEC serotypes in food and clinical samples so that proper control and treatment can be implemented promptly.

By definition, all STEC serotypes are capable of producing at least one Shiga toxin (Stx1 or Stx2), the major virulence factor contributing to STEC pathogenicity (38). Stx1 is identical (or with only a single amino acid difference) to Shiga toxin produced by Shigella dysenteriae type 1 (32), whereas Stx2 shares 55 to 60% homology with Stx1 and is immunologically distinct (24). In addition to Stx, many STEC strains carry a large chromosomal pathogenicity island, termed the locus of enterocyte effacement (LEE), which is responsible for producing attaching and effacing (A/E) lesions on enterocytes (32). Within the LEE region, an outer membrane protein intimin (encoded by eae) mediates the intimate attachment of bacteria to the enterocyte membrane (32). Although STEC virulence factors have yet to be fully elucidated, epidemiological data suggest that strains harboring both stx₂ and eae are strongly associated with severe human illnesses such as HC and HUS (3, 4, 10).

For STEC detection, three broad categories of assays are available. First, while traditional culture methods using sorbitol-containing selective media can readily identify E. coli O157:H7, currently no selective and differential media exist to culture non-O157 STEC strains (14). Second, enzyme immunoassays (EIAs) for Shiga toxins and a few STEC serogroups are commercially available (14). However, false-positive results have been reported (5, 6). Further, it is recommended that Shiga toxin EIA be performed on overnight (16 to 24 h) enrichment cultures of stools rather than direct examination (14), stretching the total assay time to days rather than hours (13). Third, rapid, specific, and sensitive nucleic acid amplification tests (NAATs) such as PCR and quantitative PCR (qPCR) have been developed to detect STEC by targeting genes coding for major STEC virulence factors such as Stx, intimin, or hemolysin (12, 35). Nonetheless, a sophisticated thermal cycling instrument is an indispensable requirement of such tests, limiting their wide applicability.

Recently, a novel NAAT technology termed loop-mediated isothermal amplification (LAMP) has attracted great attention as a rapid, accurate, and cost-effective pathogen detection method in both food testing and clinical diagnostics (31, 33). LAMP employs four to six specially designed primers and a strand-displacing Bst DNA polymerase to amplify up to 10⁹ target DNA copies under isothermal conditions (60 to 65°C) within an hour (31). Since it is isothermal, LAMP can be performed in much simpler instruments, such as a heater or water bath. To date, several LAMP assays targeting STEC Shiga toxin genes (stx₁ and stx₂) have been developed and evaluated in food samples (19–21, 28, 29, 34, 46, 47), as have a few others targeting the rfbE gene (encoding perosamine synthetase) specific for the O157 antigen of STEC O157 (40, 46, 47). However, to our knowledge, there are no LAMP assays currently available for the E. coli intimin gene (eae). Due to the importance of STEC intimin in causing severe human illnesses (3, 4, 10), screening for both stx and eae using qPCR is currently recommended by the U.S. Department of Agriculture's (USDA's) Food Safety and Inspection Service (39). Additionally, none of the LAMP studies have evaluated the assay applicability in clinical samples.

The objectives of this study were to develop rapid and reliable LAMP detection assays for STEC by targeting stx₁, stx₂, and eae and evaluate the assay performance with ground beef and human stools experimentally contaminated with low levels of STEC strains of seven major serogroups, i.e., O26, O45, O103, O111, O121, O145, and O157.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

A total of 90 strains (50 STEC and 40 non-STEC strains; see the table in the supplemental material) were used for specificity testing. Among them, seven STEC strains belonging to serogroups O26, O45, O103, O111, O121, O145, and O157 were used for sensitivity and ground beef testing. STEC O157 strain EDL933 (BEI Resources, Manassas, VA) was also used for assay optimization and application in human stools. The strains were examined for the presence of target genes (stx₁, stx₂, and eae) using PCR assays described previously (42). STEC and other Enterobacteriaceae were cultured at 35°C overnight on Trypticase soy agar or broth (TSA or TSB; BD Diagnostic Systems, Sparks, MD). Non-Enterobacteriaceae strains were grown on blood agar, except for Vibrio strains, for which TSA supplemented with 2% NaCl was used. Campylobacter strains were grown under microaerophilic conditions (85% N₂, 10% CO₂, and 5% O₂).

LAMP primers and reaction conditions.

The STEC stx₁, stx₂, and eae genes (GenBank accession numbers M19473, X07865, and Z11541, respectively) were selected as targets for designing LAMP primers (Table 1). A set of six primers, two outer (F3 and B3), two inner (FIP and BIP), and two loop (LF and LB), which recognize eight distinct regions of the target gene was designed for each target using the PrimerExplorer program (version 4; Fujitsu Limited, Japan).

Table 1.

LAMP and qPCR primers used in this study to detect STEC strains by targeting three genes (stx₁, stx₂, and eae)

Assay type	Primer/probe name	Sequencea (5′–3′)	Positiona,b	Amplicon sizec (bp)	Source or reference
stx1-LAMP	stx1-F3	TGATTTTTCACATGTTACCTTTC	507–529		This study
	stx1-B3	TAACATCGCTCTTGCCAC	688–705
	stx1-FIP	CCTGCAACACGCTGTAACGT-CAGGTACAACAGCGGTTA	574–593, 530–547
	stx1-BIP	AGTCGTACGGGGATGCAGAT-AGTGAGGTTCCACTATGC	598–617, 660–677
	stx1-LF	GTATAGCTACTGTCACCAGACAATG	548–572
	stx1-LB	AAATCGCCATTCGTTGACTACTTCT	618–642
stx2-LAMP	stx2-F3	CGCTTCAGGCAGATACAGAG	812–831		This study
	stx2-B3	CCCCCTGATGATGGCAATT	1022–1040
	stx2-FIP	TTCGCCCCCAGTTCAGAGTGA-GTCAGGCACTGTCTGAAACT	897–917, 840–859
	stx2-BIP	TGCTTCCGGAGTATCGGGGAG-CAGTCCCCAGTATCGCTGA	927–947, 989–1007
	stx2-LF	GCGTCATCGTATACACAGGAGC	860–881
	stx2-LB	GATGGTGTCAGAGTGGGGAGAA	950–971
eae-LAMP	eae-F3	TGACTAAAATGTCCCCGG	502–519		This study
	eae-B3	CGTTCCATAATGTTGTAACCAG	683–704
	eae-FIP	GAAGCTGGCTACCGAGACTC-CCAAAAGCAACATGACCGA	581–600, 526-544
	eae-BIP	GCGATCTCTGAACGGCGATT-CCTGCAACTGTGACGAAG	605–624, 664-681
	eae-LF	GCCGCATAATTTAATGCCTTGTCA	545–568
	eae-LB	ACGCGAAAGATACCGCTCT	625–643
stx1-qPCR	stx1-150-F	GACTGCAAAGACGTATGTAGATTCG	252–276	151	12
	stx1-150-R	ATCTATCCCTCTGACATCAACTGC	379–402
	stx1-150-P	FAM-TGAATGTCATTCGCTCTGCAATAGGTACTC-Iowa Black FQ	278–307
stx2-qPCR	stx2-200-F	ATTAACCACACCCCACCG	425–442	206	12
	stx2-200-R	GTCATGGAAACCGTTGTCAC	611-630
	stx2-200-P	FAM-CAGTTATTTTGCTGTGGATATACGAGGGCTTG-Iowa Black FQ	445–476
eae-qPCR	eae-170-F	CTTTGACGGTAGTTCACTGGAC	734–755	170	12
	eae-170-R	CAATGAAGACGTTATAGCCCAAC	811–903
	eae-170-P	FAM-CTGGCATTTGGTCAGGTCGGGGCG-Iowa Black FQ	789–812

^aUnderlining corresponds to the F2 or B2 region of the FIP or BIP primer, respectively.

^bThe positions are numbered based on the coding sequences of STEC stx₁, stx₂, and eae genes with GenBank accession numbers M19473, X07865, and Z11541, respectively.

^cLadder-like bands with variable sizes were generated by all three LAMP assays.

The LAMP prototypic conditions were those recommended by the manufacturer (Eiken Chemical Co., Ltd., Tokyo, Japan). Following optimization, the final LAMP reaction mix (25 μl) for stx₁ and stx₂ consisted of 1× ThermoPol reaction buffer (New England BioLabs, Ipswich, MA), 6 mM MgSO₄, 1.2 mM each deoxynucleoside triphosphate (dNTP), 0.1 μM F3 and B3 (Integrated DNA Technologies, Coralville, IA), 1.8 μM FIP and BIP, 1 μM LF and LB, 10 U of Bst DNA polymerase (New England BioLabs), and 2 μl of DNA template. The optimized eae reaction mix differed from those described above for the following parameters: MgSO₄ (8 mM), dNTP (1.8 mM each), F3 and B3 (0.3 μM each), FIP and BIP (2 μM each), and LF and LB (1.2 μM each). One positive control and one negative control were included in each LAMP run.

LAMP reactions were carried out at 65°C for 1 h and terminated at 80°C for 5 min in an LA-320C real-time turbidimeter (Eiken Chemical Co., Ltd.) with turbidity readings at 650 nm every 6 s. The time threshold (T_t; in min) was determined when the turbidity increase measurement (differential value of moving averages of turbidity) exceeded a threshold of 0.1.

qPCR assays.

In comparison, qPCR assays described previously (12) for STEC stx₁, stx₂, and eae were carried out. The mix (25 μl) contained 1× PCR buffer, 0.2 mM each dNTP, 4 mM MgCl₂, 0.25 μM each primer (Table 1), 0.1875 μM probe, 1.5 U of GoTaq Hot Start polymerase (Promega, Madison, WI), and 2 μl of DNA template. The assays were conducted using 40 cycles of denaturation at 94°C for 20 s, annealing at 60°C for 30 s, and extension at 72°C for 50 s in a SmartCycler II system (Cepheid, Sunnyvale, CA). Fluorescence readings were acquired using the 6-carboxyfluorescein (FAM) channel, and the cycle threshold (C_T; number of cycles) was obtained when the readings crossed 30 units.

LAMP specificity and sensitivity.

For LAMP specificity, DNA templates of 90 bacterial strains (see the table in the supplemental material) were prepared by heating at 95°C for 10 min as described previously (9). Aliquots (2 μl) of each template were subjected to LAMP amplification and repeated twice.

LAMP sensitivity (limit of detection) was determined by using 10-fold serial dilutions of seven STEC strains. Briefly, 3 to 5 single colonies of each strain were inoculated separately into 8 ml of fresh TSB and incubated at 35°C for 16 h to reach the stationary phase (optical density at 600 nm = 1, approximately 10⁹ CFU/ml). The cultures were 10-fold serially diluted in 0.1% peptone water, and aliquots (500 μl) of each dilution were used to prepare DNA templates similarly by heating. The exact cell numbers were determined by standard plate counting. Aliquots (2 μl) of each template were tested by LAMP and qPCR and repeated three times each.

LAMP evaluation in ground beef.

Ground beef (23% fat) samples were obtained from a local grocery store and analyzed within 2 h of collection. To determine LAMP sensitivity in ground beef, each test sample (25 g) was inoculated with 2 ml of 10-fold serially diluted individual overnight STEC cultures, resulting in spiking levels of between 10⁹ and 10⁵ CFU/25 g. Another sample was included as the uninoculated control. The samples were homogenized with 225 ml of buffered peptone water (BPW; BD Diagnostic Systems) in a food stomacher (model 400; Tekmar Company, Cincinnati, OH) for 1 min. Aliquots (1 ml) of the homogenates were centrifuged at 16,000 × g for 3 min, and pellets were suspended in 100 μl of PrepMan Ultra sample preparation reagent (Applied Biosystems, Foster City, CA). The mixtures were heated at 95°C for 10 min and centrifuged again at 12,000 × g for 2 min. The supernatants (2 μl) were used for both LAMP and qPCR, which were repeated three times each. Aerobic plate counts were performed for the uninoculated control by standard pour plate method.

Additionally, the capability of LAMP to detect low levels of seven STEC strains in ground beef was evaluated. For this application, ground beef samples were spiked with individual STEC cultures at two levels: 1 to 2 and 10 to 20 CFU/25 g. Another sample was included as the uninoculated control. The samples were homogenized with 225 ml of prewarmed BPW supplemented with 8 mg/liter vancomycin (Sigma-Aldrich, St. Louis, MO) for 1 min in the food stomacher, followed by incubation at 42°C for up to 24 h. Aliquots (1 ml) of the enrichment broth were removed at 4, 6, 8, 10, 12, and 24 h and processed similarly by PrepMan Ultra sample preparation reagents. Two microliters of the sample DNA extracts was subjected to both LAMP and qPCR. This experiment was independently repeated twice.

LAMP application in human stools.

Human stool specimens were obtained from a healthy donor and processed immediately. Each stool sample (0.5 g) was inoculated with 1 ml of 10-fold serially diluted STEC O157 strain EDL933 overnight culture, resulting in spiking levels of 10³ and 10⁴ CFU/0.5 g stool. The samples were mixed with 5 ml of TSA, and aliquots (1 ml) were removed for direct testing. The remaining mixtures were incubated at 35°C, and aliquots (1 ml) were removed at 4, 6, and 8 h for further analysis. For both direct stool testing and testing after enrichment, the samples were treated similarly with PrepMan Ultra sample preparation reagents and detected by LAMP and qPCR. This experiment was independently repeated twice.

Data analysis.

Means and standard deviations of T_t for LAMP or C_T for qPCR were calculated by Microsoft Excel software (Seattle, WA). The detection limits (numbers of CFU/reaction in pure culture or numbers of CFU/g in spiked ground beef) were presented as the lowest numbers of STEC cells that could be detected by the assays. In spiked ground beef, the number of CFU/reaction was calculated by using the number of CFU/g × 25 g/250 × 10 × 2 × 10⁻³, i.e., number of CFU/g × 2 × 10⁻³. Similarly, in spiked human stools, the number of CFU/reaction was converted by using the number of CFU/g × 2 × 10⁻³. Standard curves to quantify STEC in pure culture and spiked ground beef were generated by plotting T_t values against log number of CFU/reaction or log number of CFU/g, respectively, and the quantitative capabilities of the LAMP assays were derived on the basis of the correlation coefficient (R²) values from the standard curves.

In spiked ground beef and human stool experiments, T_t and C_T values sorted by target gene, spiking level, and enrichment time were compared by using the analysis of variance (ANOVA; SAS for Windows, version 9; SAS Institute Inc., Cary, NC). Differences between the mean values were considered significant when P was <0.05.

RESULTS

LAMP specificity.

Among 90 bacterial strains used to determine the specificity of the three LAMP assays (stx₁-LAMP, stx₂-LAMP, and eae-LAMP), false-positive or false-negative results were not observed; i.e., LAMP results matched 100% with known strain characteristics for the three target genes. Using stx₁-LAMP, mean T_t values for 30 STEC strains harboring the stx₁ gene ranged from 11.4 to 14.5 min, and one stx₁-positive Shigella dysenteriae strain (NCTC 4837) also gave a positive LAMP result with a mean T_t value of 13.2 min (data not shown). Similarly, by stx₂-LAMP, mean T_t values for 35 STEC strains containing stx₂ fell between 13.1 and 19.7 min, whereas for 47 eae-positive E. coli strains, mean T_t values by eae-LAMP were between 13 and 25.2 min. In contrast, for strains lacking any or all of the three target genes, no T_t value was obtained by corresponding LAMP assays, suggesting negative LAMP results.

LAMP sensitivity and quantitative capability.

Table 2 summarizes LAMP sensitivity when testing 10-fold serial dilutions of individual STEC strains of seven serogroups in three repeats. In pure-culture testing, all three LAMP assays consistently detected down to 10¹ CFU/reaction of the seven STEC strains, except for eae-LAMP when STEC O26 strain 97-3250 and O111 strain 3215-99 were tested. Further, in one to two out of three repeats, all three LAMP assays detected several STEC strains at concentrations 10-fold lower (i.e., 10⁰ CFU/reaction). It is noteworthy that LAMP assays proceeded faster in some strains than others. For example, for the 10⁵-CFU/reaction templates tested by stx₁-LAMP, the mean T_t values ranged from 15 min for STEC O103 strain MT#80 to 19.6 min for O157 strain EDL933 (data not shown). Similar variations in amplification speed among the seven STEC strains were also observed for stx₂-LAMP and eae-LAMP (data not shown). Regardless of target genes, the detection limits for qPCR fell between 10⁰ and 10¹ CFU/reaction (data not shown).

Table 2.

Sensitivity of the three LAMP assays when testing 10-fold serial dilutions of individual STEC strains of seven serogroups in pure culture and spiked ground beef samples

Strain identifier	Serotype	Stx	Intimin	Detection limit (no. of CFU/reaction or no. of CFU/gc)
				stx1-LAMP		stx2-LAMP		eae-LAMP
				Culture	Ground beef	Culture	Ground beef	Culture	Ground beef
97-3250	O26:H11	1, 2	+	10	4 × 103–4 × 104b	1–10b	4 × 103–4 × 104b	10–100b	4 × 104–4 × 105b
MI01-88	O45:H2	1	+	1.6–16a	7 × 103	NAd	NA	1.6–16b	7 × 103–7 × 104a
MT#80	O103:H2	1	+	1.6–16b	6.5 × 103	NA	NA	16	6.5 × 104
3215–99	O111:H8	1, 2	+	1.1–11a	5 × 103	1.1–11b	5 × 103	110–1,100a	5 × 104–5 × 105a
MDCH-4	O121:H19	2	−	NA	NA	1.2–12b	4 × 103	NA	NA
GS G5578620	O145:NMe	1	+	17	4 × 103	NA	NA	1.7–17b	4 × 103
EDL933	O157:H7	1, 2	+	1.6–16a	6.5 × 103	1.6–16b	6.5 × 103	1.6–16a	6.5 × 103

^aOne out of three repeats was positive for the lower detection limit.

^bTwo out of three repeats were positive for the lower detection limit.

^cIn ground beef testing, the number of CFU/reaction equals the number of CFU/g × 2 × 10⁻³.

^dNA, not applicable.

^eNM, nonmotile.

Figure 1 shows a typical LAMP amplification graph and a standard curve generated for pure-culture sensitivity testing of STEC O157 strain EDL933 by stx₂-LAMP. The T_t values ranged from 21.4 to 45.2 min for cell concentrations between 1.6 × 10⁵ and 1.6 CFU/reaction. Excluding data for 1.6 and 16 CFU/reaction, the quantification equation for this assay was determined to be y = −2x + 31.2 with an R² value of 0.997. Similar quantification equations were obtained for other assay/strain pairs, and the overall R² values ranged from 0.933 to 0.997 (data not shown).

LAMP sensitivity in spiked ground beef is also summarized in Table 2. For the uninoculated control sample, aerobic plate counts averaged 2 × 10⁵ CFU/g, and all three target genes tested negative by LAMP and qPCR (data not shown). Using stx₁-LAMP and stx₂-LAMP, the lower limits of detection were at the 10³-CFU/g level, equivalent to 8 to 14 CFU/reaction, while by eae-LAMP, at least 10-fold higher cell concentrations (i.e., 10⁴ CFU/g) were needed in three strains (Table 2). In comparison, the majority of qPCR assays had detection limits of 10⁴ CFU/g for stx₁ and stx₂ and 10³ CFU/g for eae in spiked ground beef (data not shown). Similar to pure-culture testing, quantification equations were generated on the basis of ground beef sensitivity data, and R² ranged from 0.904 to 0.994 (data not shown).

Rapid detection of low levels of STEC in ground beef.

Table 3 summarizes LAMP and qPCR results in ground beef samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of individual STEC strains of seven serogroups after various enrichment periods. A typical LAMP judgment graph generated for ground beef enrichment samples is shown in Fig. 2. Regardless of spiking levels, none of the 4-h-enrichment samples tested positive by either LAMP or qPCR. Positive LAMP results appeared at 6 h with significantly higher T_t values (P < 0.05), and for samples enriched for 8, 10, 12, and 24 h, stable and lower T_t values were observed, with no significant differences among enrichment periods (P > 0.05) (Table 3). A similar trend of detection was observed for qPCR (Table 3). At the 6-h enrichment point, the only LAMP-negative sample was the one spiked with STEC O157 strain EDL933 and tested by stx₁-LAMP, and the result was confirmed by qPCR. However, STEC O45 strain MI01-88 tested positive by LAMP but was negative by qPCR (Table 3). Additionally, qPCR results were presented by cycles, which were approximately 2 min/cycle. Therefore, an additional 30 to 55 min of amplification time was needed for qPCR when testing the same enrichment sample.

Table 3.

Comparison of LAMP and qPCR assays in ground beef samples spiked with low levels of individual STEC strains of seven serogroups

Cell level (no. of CFU/25 g)	Target gene	Avg LAMP Tt (min) after enrichment ofa:					Avg qPCR CT (cycles) after enrichment ofa:
		6 h	8 h	10 h	12 h	24 h	6 h	8 h	10 h	12 h	24 h
1–2	stx1	20.5 ± 4.3A	17.0 ± 3.5AB	15.8 ± 3.1B	15.9 ± 2.7B	16.1 ± 3.3B	37.4 ± 2.2A	32.5 ± 3.3B	29.4 ± 2.2B	30.2 ± 1.9B	30.5 ± 3.3B
	stx2	28.2 ± 3.3A	22.4 ± 0.8B	20.4 ± 0.6B	20.3 ± 0.3B	20.5 ± 1.1B	37.7 ± 1.5A	30.1 ± 3.4B	28.4 ± 1.6B	29.4 ± 1.4B	29.0 ± 2.6B
	eae	27.7 ± 12.4A	19.6 ± 4.8B	18.8 ± 4.2B	18.6 ± 4.0B	18.9 ± 4.2B	34.7 ± 2.2A	28.3 ± 2.9B	26.1 ± 2.0B	27.3 ± 2.1B	27.1 ± 2.6B
10–20	stx1	22.0 ± 7.7A	15.9 ± 3.4B	15.1 ± 3.4B	15.1 ± 3.0B	15.2 ± 2.8B	36.0 ± 3.5A	28.9 ± 3.6B	26.4 ± 3.5B	27.2 ± 2.8B	26.2 ± 3.4B
	stx2	23.4 ± 0.6A	19.4 ± 0.4B	18.5 ± 0.6B	19.1 ± 0.7B	19.2 ± 0.7B	35.4 ± 2.3A	27.8 ± 3.9B	25.9 ± 3.0B	26.8 ± 2.1B	26.1 ± 3.0B
	eae	23.2 ± 7.1A	18.5 ± 4.2AB	18.0 ± 4.1AB	17.2 ± 4.0B	17.6 ± 4.0AB	31.8 ± 4.4A	26.0 ± 2.8B	24.0 ± 3.0B	24.4 ± 3.3B	23.7 ± 3.4B

^aNone of the 4-h-enrichment samples tested positive by either LAMP or qPCR. After 6 h enrichment, one out of six stx₁-positive strains was negative for LAMP, whereas two were negative for qPCR. In each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters are statistically significant (P < 0.05).

Fig 2.

A typical LAMP amplification graph generated when testing ground beef samples spiked with two low levels of individual STEC strains of seven serogroups after various enrichment periods (4, 6, 8, 10, 12, and 24 h). In this graph, the ground beef samples were spiked with 1.2 CFU of STEC O111 strain 3215-99 and the enrichment samples were tested by stx₂-LAMP.

Rapid diagnosis of STEC in human stools.

Table 4 shows LAMP and qPCR results in human stool specimens spiked with 10³ and 10⁴ CFU/0.5 g of STEC O157 EDL933 cultures based on two independent repeats. For direct stool testing, all samples were negative except for one with the 10⁴-CFU/0.5 g level tested by eae-LAMP in one repeat. Regardless of spiking levels, after 4, 6, and 8 h of enrichment, all samples were positive by LAMP. However, several negative qPCR results were observed at the 4-h enrichment point for the 10³-CFU/0.5 g spiking level (Table 4). Noticeably, both T_t and C_T values decreased as the enrichment proceeded, with significantly higher T_t values observed at the 4-h enrichment point than the other two (P < 0.05). Similar to ground beef testing, qPCR (C_T values of approximately 30 cycles) required an additional 40 min to generate positive results compared to LAMP (T_t values of about 20 min).

Table 4.

Comparison of LAMP and qPCR assays in human stool specimens spiked with 10³ and 10⁴ CFU/0.5 g of STEC O157 strain EDL933 based on two independent repeats

Cell level (no. of CFU/0.5 g)	Target gene	LAMP Tt (min) after enrichment ofb:				qPCR CT (cycles) after enrichment ofb:
		0 h	4 h	6 h	8 h	0 h	4 h	6 h	8 h
103	stx1	NAc	32.0 ± 2.6A	20.9 ± 1.5B	17.3 ± 1.6B	NA	37.9a	34.1 ± 1.5A	29.1 ± 1.4A
	stx2	NA	33.9 ± 2.3A	23.9 ± 1.1B	21.0 ± 1.3B	NA	NA	34.3 ± 1.5A	29.4 ± 1.5A
	eae	NA	27.8 ± 2.1A	19.0 ± 2.0B	15.9 ± 1.8B	NA	37.2a	32.3 ± 1.4A	26.6 ± 1.5A
104	stx1	NA	22.9 ± 0.8A	18.9 ± 0.4B	18.2 ± 0.5B	NA	34.5 ± 1.4A	31.3 ± 1.5A	30.2 ± 1.4A
	stx2	NA	26.6 ± 1.3A	23.2 ± 0.5AB	21.5 ± 1.3B	NA	34.5 ± 1.2A	31.2 ± 1.4AB	30.3 ± 1.3B
	eae	26.9a	19.8 ± 3.0A	17.5 ± 1.4A	17.0 ± 1.1A	NA	33.5 ± 1.8A	29.7 ± 1.5A	28.8 ± 1.1A

^aOnly one repeat generated a positive result.

^bIn each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters are statistically significant (P < 0.05).

^cNA, not available.

DISCUSSION

The three LAMP assays (stx₁-LAMP, stx₂-LAMP, and eae-LAMP) developed in the present study were rapid (11 to 45 min), specific (100% inclusivity and 100% exclusivity among 90 strains tested), sensitive (1 to 20 CFU/reaction in pure culture and 10³ to 10⁴ CFU/g in spiked ground beef), and quantitative (R² = 0.904 to 0.997). With 6 to 8 h of enrichment, the assays accurately detected two low levels (1 to 2 and 10 to 20 CFU/25 g) of STEC in ground beef samples. In human stool specimens, the assays also consistently detected STEC spiked at 10³ or 10⁴ CFU/0.5 g stool after 4 h enrichment. To our knowledge, this is the first study applying the novel LAMP NAAT technology to detect STEC in food and clinical samples by targeting both stx and eae.

Previously, LAMP assays have been developed for the detection of generic (22) and pathogenic E. coli, including enteroaggregative E. coli (44), enteroinvasive E. coli (37), enterotoxigenic E. coli (43), STEC (20, 28, 29), and more specifically, E. coli O157:H7 (40, 46, 47). With 35 min to 1 h of reaction time, these LAMP assays were capable of detecting between 0.7 and 100 CFU of E. coli per reaction, 10- to 100-fold more sensitive than conventional PCR (20, 22, 28, 37, 40, 44, 47). The three LAMP assays developed here fell within these detection ranges in terms of speed and sensitivity. Numerous other studies also reported the superior sensitivity of LAMP in comparison with PCR (9, 16, 18); however, few comparisons were made between LAMP and qPCR. Similar to findings of the present study, a recent study on LAMP detection of Salmonella also reported comparable sensitivities between LAMP and qPCR (9). It is noteworthy that the LAMP assays reported in the present study were markedly faster than qPCR assays developed by USDA scientists (12) by at least 30 min, therefore significantly shortening the total assay time.

Among the three target genes, stx₁ does not possess sequence heterogeneity, but multiple distinct variants of either stx₂ or eae have been identified (15, 45). In this study, sequence alignments of several Stx2 and intimin variants were conducted before suitable regions were chosen for LAMP primer design. Consequently, all of the three LAMP assays possessed 100% inclusivity and 100% exclusivity among the 50 STEC and 40 non-STEC strains tested, a specificity similar to that reported previously for LAMP assays targeting stx₁ and stx₂ (20). Noticeably, eae-LAMP showed inferior sensitivity in detecting two strains (STEC O26 97-3250 and O111 3215-99) compared to the others (Table 2), which may be partially explained by sequence variations of the eae gene (45).

LAMP-positive reactions are commonly detected by gel electrophoresis, visual endpoint judgment of turbidity or color change, and real-time turbidity/fluorescence analysis (31). Through real-time turbidity analysis, the quantitative capability of LAMP has been demonstrated previously (8, 9, 17, 18, 30). Other studies also showed LAMP to be quantitative using fluorescence-based platforms (1, 8, 17). In the present study, R² fell between 0.933 and 0.997 for STEC cells present at levels ranging from 10⁵ to 10² CFU/reaction in pure culture and 0.904 to 0.988 for cells present at levels ranging from 10⁷ to 10⁴ CFU/g (10⁵ to 10² CFU/reaction) in spiked ground beef, suggesting good quantitative capabilities. However, for STEC cells present at less than 10² CFU/reaction, the quantitative capability of LAMP was poor, indicated by much delayed T_t values (Fig. 1A). Similar findings regarding the poor quantification of LAMP at low cell levels were reported previously (2, 11). It is important to note that whenever enrichment is incorporated in the detection steps, quantification is no longer relevant.

Fig 1.

A typical LAMP amplification graph (A) and a standard curve generated for pure-culture sensitivity testing of STEC O157 strain EDL933 by stx₂-LAMP (B). Samples 1 to 6 correspond to 10-fold serial dilutions of E. coli O157:H7 EDL933 cells ranging from 1.6 × 10⁵ to 1.6 CFU/reaction; sample 7 is water. The standard curve was drawn on the basis of three independent repeats and excluding data for cell concentrations of 1.6 and 16 CFU/reaction.

To date, application of LAMP assays for the detection of STEC and E. coli O157:H7 has been reported exclusively in food samples (19–21, 34, 40). Only STEC O157 and O26 strains have been used for inoculation, and the spiked samples were usually enriched overnight without characterizing the effects of different enrichment times on the detection outcomes (19–21, 34, 40). For instance, a recent study reported that 45 to 50% of liver samples inoculated with 1 to 4 CFU/25 g of E. coli O157:H7 strains tested positive by LAMP after overnight enrichment, compared to a 10 to 35% detection rate by culture (34). Two earlier studies by the same group reported that for ground beef samples inoculated with approximately 10 CFU/25 g of E. coli O157 or O26 strains, 100% detection rates were observed after 24 h enrichment (19, 20), whereas culture methods detected 100% of ground beef samples spiked with STEC O157 but only 50 to 80% of those spiked with STEC O26 (19). In raw milk, a detection limit of 4.1 × 10⁴ CFU/ml of E. coli O157 was reported (40).

In the present study, STEC strains of seven major serogroups were used in ground beef experiments, and STEC O157 EDL933 was also used to spike human stool specimens. The three LAMP assays had 10³- to 10⁴-CFU/g detection limits in ground beef, which were comparable to those of previously reported LAMP and qPCR assays (12, 40). For the ground beef samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of STEC, positive detection occurred at 6 h enrichment and consistently thereafter, which were results superior to those obtained in the liver study (34) mentioned above. We also found that LAMP performed better than qPCR in terms of positive detection rate and assay speed in spiked ground beef. In human stool experiments, consistent detection of samples spiked with 10³ and 10⁴ CFU/0.5 g of STEC O157 EDL933 culture after 4 h enrichment was observed by LAMP in the present study. Again, qPCR failed to detect several LAMP-positive samples at 4 h enrichment. In general, molecular-based detection methods such as PCR and LAMP are subjected to various inhibitors present in food and clinical samples. However, corroborating with our findings, LAMP has previously been confirmed to be more robust than PCR with regard to tolerance to inhibitors in clinical samples and other biological substances (11, 26).

Currently, E. coli O157:H7 is regulated as an adulterant in raw beef in the United States. The growing clinical importance of non-O157 E. coli also warrants the development of rapid, sensitive, and specific methods for detection. However, to meet the goal of detecting very low levels of these pathogens in food, an enrichment step is essential (13). For example, in the newly updated USDA protocol for E. coli O157:H7 and non-O157 STEC detection in ground beef and beef trimmings, enrichment is an indispensable step and is followed by initial screening of Shiga toxins and intimin by qPCR and a second screening of O157 and the top six STEC serogroups by another set of qPCR assays (39). Given the rapidity, sensitivity, specificity, and robustness of LAMP assays demonstrated in the present study, these assays may effectively serve as a means of serogroup-independent screening of STEC strains in ground beef samples, which is to be followed with serogroup-specific tests and virulence characterizations to ascertain the food safety and public health relevance of the STEC-positive samples. The three STEC assays also had superior performance to qPCR in human stools. In conclusion, the LAMP assays developed in this study may facilitate rapid and reliable identification of STEC contaminations in high-risk food commodities and also facilitate prompt diagnosis of STEC infections in clinical laboratories.