Rapid and Specific Detection of Escherichia coli Serogroups O26, O45, O103, O111, O121, O145, and O157 in Ground Beef, Beef Trim, and Produce by Loop-Mediated Isothermal Amplification

Abstract

Escherichia coli O157 and six additional serogroups of Shiga toxin-producing E. coli (STEC) (O26, O45, O103, O111, O121, and O145) account for the majority of STEC infections in the United States. In this study, O serogroup-specific genes (wzx or wzy) were used to design loop-mediated isothermal amplification (LAMP) assays for the rapid and specific detection of these leading STEC serogroups. The assays were evaluated in pure culture and spiked food samples (ground beef, beef trim, lettuce, and spinach) and compared with real-time quantitative PCR (qPCR). No false-positive or false-negative results were observed among 120 bacterial strains used to evaluate assay specificity. The limits of detection of various STEC strains belonging to these target serogroups were approximately 1 to 20 CFU/reaction mixture in pure culture and 10³ to 10⁴ CFU/g in spiked food samples, which were comparable to those of qPCR. Standard curves generated suggested good linear relationships between STEC cell numbers and LAMP turbidity signals. In various beef and produce samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of respective STEC strains, the LAMP assays consistently achieved accurate detection after 6 to 8 h of enrichment. In conclusion, these newly developed LAMP assays may facilitate rapid and reliable detection of the seven major STEC serogroups in ground beef, beef trim, and produce during routine sample testing.

INTRODUCTION

Shiga toxin-producing Escherichia coli (STEC) strains are zoonotic agents responsible for many large-scale food-borne outbreaks worldwide (4, 5, 33). Disease symptoms range from watery diarrhea to hemorrhagic colitis (HC), and to potentially fatal hemolytic-uremic syndrome (HUS) (36). Annually, STEC strains cause approximately 176,000 illnesses, 2,400 hospitalizations, and 20 deaths in the United States (40). More than one-third of the total illnesses and all of the deaths are caused by E. coli O157:H7, the most common and notorious STEC serotype (40). Nonetheless, well over 100 other STEC serotypes have been associated with outbreaks and sporadic cases, and their clinical significance is mounting in many countries (27). The recent massive German outbreak of HUS in May 2011 was attributable to a rare STEC serotype, O104:H4, in sprouts (4). In the United States, for the first time since surveillance for non-O157 STEC began in 2000, FoodNet reported a slightly higher incidence of non-O157 STEC infections than O157 cases in 2010 (6). The most prevalent pathogenic non-O157 STEC serogroups in the United States are O26, O45, O103, O111, O121, and O145, causing over 70 to 83% of the total non-O157 STEC illnesses (3, 44).

These top six non-O157 STEC serogroups share many epidemiological and virulence features with E. coli O157:H7. For instance, ruminants, particularly cattle, are a major reservoir (19). Transmission routes include contaminated food and water, animal contact, and person-to-person contact (19). Food commodities frequently implicated in the outbreaks are beef, cheese, milk, juice, and produce (5, 31). Similar to E. coli O157:H7, all of these non-O157 STEC strains can cause HC, and all except O45 have been shown to cause HUS (44). The infectious doses are a few hundred cells or even lower, comparable to that of E. coli O157:H7 (44). Virulence factors for both E. coli O157:H7 and non-O157 STEC include, but are not limited to, Shiga toxins 1 and/or 2 (Stx1, Stx2) and intimin (encoded by eae) (36). Since 1994, E. coli O157:H7 has been regulated by the U.S. Department of Agriculture (USDA) Food Safety and Inspection Service (FSIS) as an adulterant in raw beef (42). Very recently, the FSIS announced that beginning on 5 March 2012 (updated on 4 June 2012), the top six non-O157 STEC serogroups will be included in this zero tolerance policy concerning nonintact raw beef products (45). With this forthcoming regulation, it is imperative that rapid and reliable detection methods be available to test specifically for these STEC serogroups of significant public health concern in beef and other high-risk foods.

Unlike E. coli O157:H7, effective detection and isolation of non-O157 STEC using traditional culture methods remain problematic due to the lack of phenotypic characteristics (e.g., sorbitol fermentation) distinguishing them from generic E. coli (18). In clinical diagnostics, the U.S. Centers for Disease Control and Prevention recommends that all stool samples submitted from patients with acute diarrhea be simultaneously cultured for E. coli O157:H7 and tested by a nonculture method (Shiga toxin enzyme immunoassay or PCR) for non-O157 STEC (18). Additional culture isolation and immunological and molecular characterizations are needed to determine specific non-O157 STEC serogroups (18). The recently updated FSIS method for non-O157 STEC in meat products consists of a stepwise screening procedure using real-time quantitative PCR (qPCR) first for stx and eae genes and then for serogroup-specific wzx genes (encoding O-unit flippase) of the top six non-O157 STEC serogroups (43). Samples screened positive are subjected to culture isolation using corresponding antibody-coated immunomagnetic-separation (IMS) beads and further confirmed by immunological, qPCR, and biochemical assays (43). Antibodies are commercially available for STEC O26, O103, O111, O145, and O157 but not for O45 and O121 (43). Also, unreliable IMS results have been reported previously when STEC cell numbers were low (20). Besides wzx, other serogroup-specific genes in the O-antigen gene cluster, such as wzy (encoding O-antigen polymerase) and wbd (encoding O-antigen transferase), have been used as targets to design qPCR assays for the specific detection of various STEC serogroups (13, 30, 39). Although reported to be rapid, specific, and sensitive, qPCR assays require a sophisticated thermal cycling instrument with real-time fluorescence monitoring, limiting their wide applicability.

Recently, a novel nucleic acid amplification technology termed loop-mediated isothermal amplification (LAMP) has attracted great attention as a rapid, accurate, and cost-effective alternative to the detection of bacterial and viral agents in food and clinical samples (35, 37). LAMP differs from PCR in that four to six specially designed primers and a strand-displacing Bst DNA polymerase are used to efficiently amplify the target DNA at a single temperature (60 to 65°C) (37). Since it is isothermal, LAMP can be performed with much simpler instruments such as a heater or water bath. LAMP is also advantageous over PCR in that positive results can be directly detected through visual observation of turbidity changes (34). We recently developed and evaluated a set of serogroup-independent STEC LAMP assays by targeting common STEC virulence genes (stx₁, stx₂, and eae), and the assays were rapid, sensitive, specific, and robust with ground beef and human stool samples (47). Additionally, several STEC O157 LAMP assays targeting the rfbE gene (encoding perosamine synthetase) have been reported (46, 52, 53). However, to our knowledge, there are no LAMP assays currently available for the top six non-O157 STEC serogroups.

This study aimed to develop rapid and reliable LAMP detection assays specifically for the seven leading STEC serogroups (O26, O45, O103, O111, O121, O145, and O157) by targeting respective serogroup-specific wzx or wzy genes and to evaluate the assays' performance in comparison with that of qPCR with food samples (ground beef, beef trim, lettuce, and spinach) spiked with low levels of STEC strains belonging to these serogroups.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

A total of 91 E. coli strains of 20 serogroups (Table 1) and 29 non-E. coli strains as described in our recent study (47) were used for specificity testing. The E. coli serogroup information was taken directly from the source database (Table 1). Among them, seven STEC strains (97-3250, MI01-88, MT#80, 3215-99, MT#2, GS G5578620, and EDL933, belonging to serogroups O26, O45, O103, O111, O121, O145, and O157, respectively) were also used for sensitivity testing and spiked-food experiments. The non-E. coli strains tested included Campylobacter, Listeria, Salmonella, Shigella, and Vibrio, among others (47). E. coli and other Enterobacteriaceae strains were cultured at 35°C overnight on Trypticase soy agar (TSA; BD Diagnostic Systems, Sparks, MD) or in Trypticase soy broth. 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).

Table 1.

Ninety-one E. coli strains used in this study to evaluate LAMP specificity and sensitivity

Strain category (total no. of strains) and serogroup	No. of strains	Strain identifier(s)a	Origin	Sourceb
Target (n = 76)
O26	11	97-3250, 3047-86, EH1534, MT#10, TB352A, [DEC9F, DEC10C, EK29, H19, TB285A, VP30]	Human	BEI Resources, STEC Center
O45	10	DA-21, MI01-88, MI03-19, MI05-14, [2566-58, 4309-65, 5431-72, B8026-C1, D88-28058, DEC11C]	Human or animal	STEC Center
O103	9	8419, MT#80, PT91-24, TB154A, [87-2931, 107-226, EK30, MT#82, RW1372]	Human or animal	STEC Center
O111	11	0201 9611, 3007-85, 3215-99, RD8, [412/55, C412, CL-37, DEC8C, ED-31, EK35, TB226A]	Human or animal	STEC Center
O121	9	DA-5, MT#2, MT#11, MT#18, [3-524, 87-2914, DA-1, DA-69, F6173]	Human	STEC Center
O145	8	4865/96, EH1533, GS G5578620, IH 16, [02-3422, BCL73, MT#66, TB269C]	Human or animal	BEI Resources, STEC Center
O157	18	86-24, 93-111, 94-G7771, 493/89, A, BDMS 770, CoGen002096, E32511, EDL931, EDL932, EDL933, G5101, MDL 3562, MDL 4444, MDL 4445, MDL 4572, OK-1, RIMD 509952	Human or food	BEI Resources, STEC Center
Nontarget (n = 15)
O3	1	NCDC U14-41	Human	BEI Resources
O6	1	CFT073	Human	BEI Resources
O9	1	HS	Human	BEI Resources
O15	1	88-1509	Human	STEC Center
O25	1	E2539-C1	Human	BEI Resources
O28	1	NCDC 909-51	Human	BEI Resources
O29	1	1885-77	Human	BEI Resources
O55	2	5905, DEC5D	Human or food	STEC Center
O78	1	H10407	Human	BEI Resources
O91	2	B2F1, H414-36/89	Human	BEI Resources
O104	1	G5506	Human	STEC Center
O126	1	ATCC 12807	Human	BEI Resources
OR	1	K-12	Laboratory	BEI Resources

^aStrains in brackets were newly added ones not used in our recent study (47) that developed and evaluated serogroup-independent STEC LAMP assays. Underlined strains were used in the present study to evaluate both the specificity and the sensitivity of the seven serogroup-specific LAMP assays, whereas others were used for specificity testing alone. Strains in bold within serogroups O26, O45, O103, O111, O121, O145, and O157 do not produce Shiga toxins.

^bThe STEC Center is based at Michigan State University, East Lansing, MI, and BEI Resources is located in Manassas, VA. Additional information about the strains may be obtained at http://www.shigatox.net/stec/cgi-bin/powersearch and http://beiresources.org/Catalog/tabid/248/Default.aspx, respectively.

LAMP primers and reaction conditions.

Sequences of O-antigen gene clusters of E. coli serogroups O26, O45, O103, O111, O121, O145, and O157 were retrieved from GenBank using accession numbers AF529080, AY771223, AY532664, AF078736, AY208937, AY647260, and AF061251, respectively. Within each gene cluster, serogroup-specific wzx (for O103 and O145) or wzy (for O26, O45, O111, O121, and O157) genes were selected as targets to design LAMP primers using the PrimerExplorer program (version 4; Fujitsu Limited, Tokyo, Japan). For each target, a set of five or six LAMP primers, two outer (F3 and B3), two inner (FIP and BIP), and one or two loop (LF and LB) primers, which recognize seven or eight distinct regions of the target gene sequence was selected (Table 2).

Table 2.

LAMP primers used in this study to detect seven specific STEC serogroups (O26, O45, O103, O111, O121, O145, and O157) by targeting respective wzx or wzy genes

Target (GenBank accession no.) and primer name	Sequence (5′–3′)	Positiona
O26-wzy (AF529080)
O26-F3	GACTATGAAGCGTATGTTGAT	136–156
O26-B3	TCCTGATTTGAACAATGTCAAT	352–373
O26-FIP	ACCGCCTAAATACTTAACACCATAA-TTAATGTCAATGAACTTTATGCC	207–231, 161–183
O26-BIP	TTCCTTGGGACCACATTCCT-ACATGTAAAGCAGCAAACC	265–284, 319–337
O26-LF	ACCAGCGATAACCAATCTC	184–202
O26-LB	TACAATACAGTAAGTATACAGCATT	293–317
O45-wzy (AY771223)
O45-F3	AATGTCCCCAGGGTTTGT	15–32
O45-B3	TTTAGTCGCTCGCCAAGA	217–234
O45-FIP	AGCGGGCTAATATTAGTAGTCACTC-GTATGCTTCAATTTGGCTGT	77–101, 33–52
O45-BIP	ACTCTGGGTTTGATTTTTTCACTTC-ATAATTTCATCCAGACGAACG	139–163, 192–212
O45-LB	TTATTACTCCTGGCAGTATTAATCG	167–191
O103-wzx (AY532664)
O103-F3	ACTCAGTGGTGTAGTAACATG	33–53
O103-B3	TCACCTTGATTTTCTGCTGA	205–224
O103-FIP	ATTTGCTATTCCAATTGGACCAGTA-CTTTAGACTAATTTGTGGCCTTC	102–126, 54–76
O103-BIP	TTGGGACAATTGCAAAATTTTGTGG-ATCTATTAACTCCTTGTGAAACTTG	127–151, 178–202
O103-LF	AATTGCAACAACTTTTGAAATAA	77–99
O103-LB	CCTTTATAAATGGATTCATTTCATC	152–176
O111-wzy (AF078736)
O111-F3	AAGGCGTAACTTTTTTTGAAC	623–643
O111-B3	TCATGAGGGTCATTAGGAATT	786–806
O111-FIP	TCACCAAGCTGTGAAACCAAA-CTACAGCAAGTAATATTGAACGT	684–704, 644–666
O111-BIP	TCCATGGTATGGGGACATTAAATTT-TGATGGAAGTCCATATAACGT	713–737, 763–783
O111-LB	CTTAAATAACGGCGGACAAT	738–757
O121-wzy (AY208937)
O121-F3	GCTCAGCTTTTATCTTGTTCAA	864–885
O121-B3	ATAGGCTCCCAACCATCC	1087–1104
O121-FIP	ACGCAAAAAGTATGGATTCATACCT-GATATAACAGAACCGACTTGG	955–979, 895–915
O121-BIP	TGTTGCTGGTTCCTTATTATGTAGT-AAAAGCAAGCCAAAACACTC	995–1019, 1047–1066
O121-LF	TAAAGCCATCCAACCACGC	929–947
O145-wzx (AY647260)
O145-F3	TTTGTAAGACAAGGTGTATGG	433–453
O145-B3	GCATTGGTACAGACAGCTTTA	632–652
O145-FIP	CACAGTACCACCAAACCAAAAAATA-TTGGTTAGCTATAGCTGTGA	516–540, 456–475
O145-BIP	AGTGTGCTTGGAGTGGCTTA-CAATCCCAGTTTGTAATATCGC	547–566, 590–611
O145-LF	TTCTTAAGTTCGGATACACTAGCA	476–499
O157-wzy (AF061251)
O157-F3	TCCCTTTAGGGATATATATACCTT	935–958
O157-B3	ATAACTGATATTTTCATTTCGTGAT	1146–1170
O157-FIP	TTCCCAGCCACTAAGTATTGCAATA-TGAAAAAAACCCATAGCTCGA	1034–1058, 977–997
O157-BIP	TGCATCGGCCTTCTTTTTTGG-AACGTATCATGCAATAAGATCA	1059–1079, 1115–1136
O157-LF	ATAATGATATATGAATAGAATGCGC	1004–1028
O157-LB	TCCTTTTCTCTCCGTATTGAT	1080–1100

^aPositions are numbered based on the coding sequences of respective wzx or wzy genes in each O-antigen gene cluster. Underlining corresponds to the F2 or B2 region of the FIP or BIP primer, respectively.

The LAMP reaction mixture (25 μl) 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 template DNA. One positive control and one negative control were included in each LAMP run. LAMP reaction mixtures were heated at 65°C (63°C for O157) for 50 min and terminated by heating at 80°C for 5 min in an LA-320C real-time turbidimeter (Eiken Chemical Co., Ltd., Tokyo, Japan). Turbidity readings at 650 nm were obtained every 6 s, and the time threshold (T_t; in min) was determined when the turbidity increase measurements (differential values of moving averages of turbidity) exceeded 0.1.

qPCR conditions.

In comparison, qPCR assays (13, 15) described previously for serogroup-specific detection of STEC O26, O45, O103, O111, O121, O145, and O157 were carried out with some modifications. The qPCR reagent mixture (25 μl) contained 1× PCR buffer, 4 mM MgCl₂, 0.2 mM each dNTP, 0.25 μM each primer, 0.1875 μM probe (Integrated DNA Technologies), 1.5 U of GoTaq Hot Start Polymerase (Promega, Madison, WI), and 2 μl of template DNA. 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 channel, and the cycle threshold (C_T; in number of cycles) was obtained when the readings crossed 30 units.

LAMP specificity and sensitivity.

For LAMP specificity, template DNAs of the 120 bacterial strains (Table 1) (47) were prepared by heating at 95°C for 10 min as described previously (8). Aliquots (2 μl) of each template were subjected to the seven LAMP assays.

LAMP sensitivity (limit of detection) was determined by using 10-fold serial dilutions of the seven STEC strains mentioned above. Template DNAs were prepared from stationary-phase cultures as described previously (47). Aliquots (2 μl) of each template were tested by both LAMP and qPCR and repeated five times each.

LAMP evaluation in spiked food samples.

Ground beef (23% fat) and produce (lettuce and spinach) samples were obtained from a local grocery store and analyzed within 2 h of collection. Beef trim samples were obtained from the School of Animal Sciences, Louisiana State University, where the animals were slaughtered in a laboratory class. First, LAMP sensitivities in these food matrices were determined. Briefly, lettuce and spinach leaves were cut into 4-cm² pieces using sterile scissors and weighed 25 g/sample. Ground beef and beef trim samples were also divided into 25-g test portions. Each test sample (25 g) was then 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 the LAMP and qPCR assays, which were repeated three times each. Aerobic plate counts of the uninoculated samples were determined by the standard pour plate method.

Second, the capability of LAMP to specifically detect low levels of STEC strains of these target serogroups in spiked food samples (ground beef, beef trim, lettuce, and spinach) was evaluated. For this application, food samples were similarly 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 with 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.

Data analysis.

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

In spiked-food experiments, T_t and C_T values sorted by target serogroup, spiking level, and enrichment time were compared by using the analysis of variance (SAS for Windows, version 9.2; SAS Institute Inc., Cary, NC). Differences between the mean values were significant when the P value was <0.05.

RESULTS

LAMP specificity.

Among 91 E. coli strains (Table 1) used to evaluate the specificity of the seven LAMP assays each targeting one of the leading STEC serogroups (O26, O45, O103, O111, O121, O145, or O157), false-positive or false-negative results were not observed; i.e., LAMP results agreed 100% with the strain serogroup information supplied by the source database. By the O26 LAMP assay, the mean T_t values for 11 O26 strains ranged from 16.1 to 19.8 min (data not shown), whereas no T_t value was generated when these strains were tested by the other six LAMP assays. Similarly, as determined by respective serogroup-specific LAMP assays, the mean T_t values for the E. coli O45 (n = 10), O103 (n = 9), O111 (n = 11), O121 (n = 9), O145 (n = 8), and O157 (n = 18) strains all fell between 16 and 23 min (data not shown). In contrast, for the other 44 strains consisting of 15 E. coli strains of various nontarget serogroups and 29 non-E. coli strains, no T_t value was obtained with any of the seven LAMP assays, indicating 100% exclusivity.

LAMP sensitivity and quantitative capability.

Table 3 summarizes the sensitivities of the seven LAMP assays when testing 10-fold serial dilutions of STEC strains belonging to the target serogroups. In pure-culture testing, all of the LAMP assays consistently detected down to approximately 10 to 20 cells per reaction mixture, except for O45 LAMP, which had consistent positive results further down to the 1.6 cells/reaction mixture level. In one or two out of five repeats, the O26, O145, and O157 LAMP assays were also capable of detecting the respective STEC strains at this lower concentration, i.e., 1 to 2 cells/reaction mixture (Table 3). Similarly, the detection limits for serogroup-specific qPCR assays fell between 1 and 20 cells (Table 3). For several serogroups (O103, O111, and O121), qPCR had better sensitivities than LAMP, whereas LAMP assays were slightly more sensitive than qPCR for serogroups O145 and O157.

Table 3.

Comparison of the sensitivities of seven LAMP and seven qPCR assays when testing 10-fold serial dilutions of individual STEC strains of respective serogroups in pure culture and various food matrices

Strain ID	Serotype	Detection limit (no. of CFU/reaction mixture in pure culture or CFU/g in spiked food samples)d
		Pure culture		Ground beef		Beef trim		Lettuce		Spinach
		LAMP	qPCR	LAMP	qPCR	LAMP	qPCR	LAMP	qPCR	LAMP	qPCR
97-3250	O26:H11	1b	1b	4 × 103a	4 × 103b	4 × 103b	4 × 103	4 × 103b	4 × 103	4 × 103	4 × 103
MI01-88	O45:H2	1.6	1.6	7 × 103	7 × 103	6 × 103	6 × 103	6 × 103	6 × 103	6 × 103	6 × 103
MT#80	O103:H2	16	1.6c	7 × 103	7 × 103	9 × 104	9 × 103	9 × 103b	9 × 103	9 × 103b	9 × 103
3215-99	O111:H8	11	1.1b	5 × 104	5 × 104	6 × 103b	6 × 103	6 × 103	6 × 103	6 × 104	6 × 103
MT#2	O121:H19	18	1.8	9 × 103	9 × 103	7 × 103a	7 × 103a	7 × 103	7 × 103	7 × 103b	7 × 103
GS G5578620	O145:NM	1.7a	17	4 × 103b	4 × 104	7 × 103a	7 × 103b	7 × 103b	7 × 103a	7 × 104	7 × 103a
EDL933	O157:H7	1.6b	1.6a	7 × 103a	7 × 103b	7 ×104	7 × 103	7 × 103b	7 × 103	7 × 103b	7 × 103

^aOne repeat was positive for this detection limit.

^bTwo repeats were positive for this detection limit.

^cThree repeats out of five were positive for this detection limit.

^dIn spiked food experiments, the number of CFU per reaction mixture equals the number of CFU per gram times 2 × 10⁻³.

Figure 1 shows a typical LAMP amplification graph and a standard curve generated in pure-culture sensitivity testing of STEC O26 strain 97-3250 by O26 LAMP. The average T_t values based on five repeats ranged from 20.9 to 35.2 min for cell concentrations spanning between 1 × 10⁵ and 10 CFU/reaction mixture. In two out of five repeats, amplification of the 1-CFU/reaction mixture template DNA also occurred (data not shown). Excluding data for this 1-CFU/reaction mixture level, the quantification equation of this LAMP assay was determined to be y = −3.556x + 37.636 with an R² value of 0.964 (Fig. 1). Similar quantification equations were obtained for the other six LAMP assays, and the overall R² values ranged from 0.945 to 0.996 (data not shown).

Fig 1.

A typical LAMP amplification graph (A) and a standard curve (B) generated in pure-culture sensitivity testing of STEC O26 strain 97-3250 by O26 LAMP. Samples 1 to 6 correspond to 10-fold serial dilutions of E. coli O26:H11 strain 97-3250 cells ranging from 1 × 10⁵ to 1 CFU/reaction mixture; sample 7 is water. The standard curve was drawn based on five independent repeats, and data for cell concentrations of 1 CFU/reaction mixture were excluded.

LAMP sensitivities in various food matrices are also summarized in Table 3. All of the uninoculated control samples tested negative for serogroup-specific wzx or wzy genes by LAMP and qPCR (data not shown). The mean aerobic plate cell counts for ground beef, beef trim, lettuce, and spinach control samples were 5 × 10⁵, 4 × 10³, 1 × 10³, and 5× 10⁴ CFU/g, respectively. Regardless of the matrix types, all of the LAMP assays consistently detected the corresponding STEC strains down to 10⁴ CFU/g. Consistent with pure-culture sensitivity data, O45 LAMP possessed a detection limit of 10³ CFU/g in all types of food matrices, equivalent to approximately 10 CFU/reaction mixture. This level of sensitivity was also consistently observed for the other six LAMP assays in some food matrices, such as O103 and O121 LAMP in ground beef, O111 and O121 LAMP in lettuce, and O26 in spinach (Table 3). In comparison, the majority of qPCR assays had detection limits of 10³ CFU/g in beef trim, lettuce, and spinach and 10⁴ CFU/g for serogroups O111 and O145 in ground beef (Table 3). Similar to pure-culture testing, quantification equations were generated based on sensitivity data in food matrices, with R² ranging from 0.867 to 0.999 (data not shown).

Rapid and specific detection of low levels of STEC in ground beef, beef trim, lettuce, and spinach.

Table 4 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 target serogroups and tested after various enrichment periods. A typical LAMP judgment graph generated for this application is shown in Fig. 2. Regardless of target serogroups or 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 the enrichment periods (P > 0.05). A similar trend of detection was observed for qPCR (Table 4). It is noteworthy that qPCR results were presented in cycles, which were approximately 2 min/cycle. Therefore, an additional 25 to 40 min of amplification time was needed for qPCR when the same enrichment sample was tested.

Table 4.

Comparison of LAMP and qPCR assays of ground beef samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of individual STEC strains of seven target serogroups and tested after various enrichment periods

No. of cells/25 g and target group	Avg LAMP Tt (min)a after enrichment for:					Avg qPCR CT (cycles)a after enrichment for:
	6 h	8 h	10 h	12 h	24 h	6 h	8 h	10 h	12 h	24 h
1–2
O26	29.3 ± 3.3 (A)	18.9 ± 0.5 (B)	17.4 ± 0.9 (B)	17.7 ± 1.3 (B)	18.5 ± 0.3 (B)	34.8 ± 0.7 (A)	27.4 ± 1.1 (B)	23.8 ± 3.4 (B)	24.2 ± 3.0 (B)	25.5 ± 1.1 (B)
O45	36.6 ± 0.2 (A)	26.2 ± 0.2 (B)	23.5 ± 0.1 (C)	22.6 ± 0.1 (C)	24.5 ± 2.1 (BC)	35.2 ± 1.1 (A)	28.0 ± 2.3 (B)	23.2 ± 2.6 (B)	23.2 ± 3.3 (B)	24.8 ± 2.3 (B)
O103	31.7 ± 7.6 (A)	20.6 ± 1.3 (B)	18.6 ± 0 (B)	18.5 ± 0.8 (B)	19.2 ± 0.3 (B)	33.6 ± 3.5 (A)	25.8 ± 2.5 (B)	21.0 ± 2.1 (B)	21.3 ± 3.0 (B)	22.1 ± 0.4 (B)
O111	30.1 ± 1.9 (A)	25.7 ± 0.8 (B)	24.4 ± 0.1 (B)	24.1 ± 0.9 (B)	26.3 ± 1.5 (B)	33.9 ± 0.3 (A)	25.6 ± 0 (B)	21.8 ± 2.2 (B)	22.5 ± 4.5 (B)	24.9 ± 0.6 (B)
O121	27.6 ± 0.6 (A)	21.2 ± 0.6 (BC)	19.2 ± 0.4 (C)	19.0 ± 0.7 (C)	21.7 ± 1.6 (B)	32.5 ± 0.6 (A)	24.8 ± 0.4 (B)	19.6 ± 0.1 (C)	19.4 ± 0.1 (C)	24.2 ± 0.1 (B)
O145	32.7 ± 3.9 (A)	22.8 ± 3.2 (B)	21.1 ± 3.0 (B)	21.6 ± 4.5 (B)	22.1 ± 1.1 (B)	35.3 ± 1.6 (A)	28.4 ± 0.3 (B)	26.3 ± 1.7 (B)	26.0 ± 2.8 (B)	27.4 ± 1.4 (B)
O157	34.2 ± 0.1 (A)	26.7 ± 0 (B)	21.2 ± 3.2 (C)	20.7 ± 3.5 (C)	23.5 ± 0.3 (BC)	37.2 ± 0.8 (A)	30.8 ± 1.1 (B)	27.1 ± 2.4 (B)	26.1 ± 4.0 (B)	29.0 ± 1.8 (B)
10–20
O26	23.6 ± 0.9 (A)	18.0 ± 1.4 (B)	17.0 ± 1.5 (B)	16.9 ± 1.1 (B)	17.7 ± 0.7 (B)	33.0 ± 1.0 (A)	25.4 ± 2.9 (AB)	22.5 ± 3.8 (B)	22.2 ± 4.3 (B)	23.8 ± 2.7 (B)
O45	30.6 ± 0.9 (A)	24.4 ± 0.1 (B)	21.2 ± 1.0 (C)	21.0 ± 0.6 (C)	22.6 ± 1.4 (BC)	31.2 ± 0.9 (A)	24.3 ± 2.2 (B)	19.8 ± 2.0 (B)	19.9 ± 2.8 (B)	22.0 ± 0.2 (B)
O103	29.2 ± 8.7 (A)	20.1 ± 1.6 (AB)	17.7 ± 0 (B)	17.5 ± 1.0 (B)	17.9 ± 0.3 (B)	32.1 ± 4.1 (A)	24.7 ± 3.8 (B)	19.3 ± 1.6 (B)	18.9 ± 2.6 (B)	19.6 ± 0.3 (B)
O111	28.7 ± 1.3 (A)	24.2 ± 0.9 (B)	22.9 ± 0.1 (B)	23.2 ± 0.5 (B)	25.0 ± 2.1 (B)	31.6 ± 0.4 (A)	24.2 ± 0.7 (B)	20.4 ± 1.8 (B)	21.1 ± 3.8 (B)	23.2 ± 1.9 (B)
O121	25.4 ± 1.6 (A)	20.7 ± 0.5 (BC)	18.8 ± 0.7 (C)	18.7 ± 1.3 (C)	21.9 ± 0.6 (B)	30.9 ± 0.2 (A)	24.1 ± 0.5 (B)	19.1 ± 0 (C)	18.3 ± 0.2 (D)	23.6 ± 0.2 (B)
O145	27.5 ± 1.6 (A)	20.7 ± 2.2 (B)	19.1 ± 2.3 (B)	19.3 ± 2.1 (B)	20.9 ± 0.4 (B)	32.7 ± 1.2 (A)	25.5 ± 0.5 (B)	22.3 ± 1.4 (B)	22.1 ± 2.6 (B)	24.3 ± 1.2 (B)
O157	31.3 ± 2.0 (A)	20.4 ± 1.9 (B)	17.9 ± 3.3 (B)	17.9 ± 3.3 (B)	20.9 ± 1.1 (B)	33.4 ± 2.4 (A)	27.2 ± 3.3 (AB)	24.4 ± 4.5 (AB)	23.6 ± 5.5 (B)	25.5 ± 1.9 (AB)

^aNone of the 4-h enrichment samples tested positive by either LAMP or qPCR. Average T_t or C_T values were calculated based on two independent repeats. In each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters in parentheses are statistically significantly different (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 experiment, ground beef samples were spiked with 10 CFU of STEC O26 strain 97-3250 and the enrichment samples were tested by O26 LAMP.

In other food matrices tested (Tables 5 to 7), again, none of the 4-h enrichment samples tested positive by either LAMP or qPCR. Positive results generally appeared at 6 h with significantly higher T_t or C_T values (P < 0.05), which decreased and stabilized as enrichment proceeded. In beef trim (Table 5), after a 6-h enrichment of samples spiked with 1 to 2 CFU/25 g of STEC, LAMP tested negative in one or both repeats (in the case of O111) for all serogroups except for O45, whereas qPCR was negative for five serogroups, excluding O45 and O103. On the other hand, qPCR test results were negative for samples spiked with STEC O157 strain EDL933 at both levels (2 and 20 CFU/25 g) in two repeats, while LAMP was positive in one repeat each (Table 5). Also in beef trim, stable and insignificant T_t values (P > 0.05) for LAMP were observed after 8 to 10 h of enrichment while qPCR almost always required 10 h to gain stable C_T values (Table 5). In lettuce and spinach (Tables 6 and 7), after 6 h of enrichment, all samples tested positive by LAMP whereas by qPCR, samples inoculated with the lower level (1 to 2 CFU/25 g) of O145 in spinach and O157 in lettuce were negative in one or both repeats. Similar to beef trim testing, stable and insignificant T_t values for LAMP were observed after 8 to 10 h of enrichment while qPCR required 10 h of enrichment (Tables 6 and 7). Consistent with ground beef data, up to 40 min more of amplification time was needed for qPCR than for LAMP when the same enrichment sample of beef trim, lettuce, or spinach was tested (Tables 5 to 7).

Table 5.

Comparison of LAMP and qPCR assays of beef trim samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of individual STEC strains of seven target serogroups and tested after various enrichment periods

No. of cells/25 g and target group	Avg LAMP Tt (min)a after enrichment for:					Avg qPCR CT (cycles)a after enrichment for:
	6 h	8 h	10 h	12 h	24 h	6 h	8 h	10 h	12 h	24 h
1–2
O26	38.9	27.3 ± 3.8 (A)	22.4 ± 2.1 (AB)	20.6 ± 1.2 (AB)	18.9 ± 2.1 (B)	39.2	30.3 ± 1.4 (A)	23.4 ± 3.3 (B)	17.3 ± 0.2 (C)	16.0 ± 0.6 (C)
O45	33 ± 3.2 (A)	23.6 ± 0.7 (B)	20.5 ± 0.4 (BC)	20.1 ± 0.6 (BC)	18.4 ± 0.1 (C)	36.2 ± 2.4 (A)	28.3 ± 0.3 (B)	20.6 ± 0.4 (C)	15.8 ± 1.4 (D)	16.0 ± 0.1 (D)
O103	37.9	31.9 ± 4.2 (A)	25.7 ± 0.8 (B)	25.3 ± 0.1 (B)	25.1 ± 6.2 (B)	36.1 ± 2.3 (A)	29 ± 3.2 (B)	20.0 ± 1.8 (C)	16.8 ± 0.6 (C)	15.9 ± 1.5 (C)
O111	NDb	33.1 ± 3.3 (A)	26.6 ± 0.6 (B)	24.4 ± 0.3 (B)	23.6 ± 0.1 (B)	38.0	31.8 ± 2.9 (A)	25.1 ± 3.2 (B)	19.6 ± 0.1 (BC)	18.0 ± 0.2 (C)
O121	33.4	30.2 ± 5.7 (A)	25.4 ± 2.8 (A)	24.9 ± 1.8 (A)	23.3 ± 0.1 (B)	35.1	29.2 ± 4.1 (A)	23.1 ± 4.1 (A)	19.2 ± 1.3 (B)	17.0 ± 0.8 (B)
O145	37.2	31.4 ± 8.8 (AB)	22.6 ± 1.2 (B)	21.2 ± 0 (B)	20.3 ± 0.8 (B)	36.1	31.2 ± 2.8 (A)	24.0 ± 0.8 (B)	20.3 ± 0.3 (BC)	18.9 ± 0.7 (C)
O157	41.7	38.5 ± 1.5 (A)	29.1 ± 4.1 (B)	22.6 ± 0.6 (BC)	17.3 ± 0.1 (C)	ND	34.8 ± 0.5 (A)	28.7 ± 1.4 (B)	23.9 ± 1.1 (C)	16.1 ± 1 (D)
10–20
O26	33.8 ± 5.3 (A)	25.1 ± 4.3 (B)	20.5 ± 0.8 (B)	19.7 ± 0.8 (B)	18.8 ± 2.1 (B)	34.4 ± 2.3 (A)	27.1 ± 4.2 (B)	18.6 ± 1.3 (C)	16.3 ± 0.7 (C)	16.2 ± 1.0 (C)
O45	31.4 ± 0.7 (A)	23.6 ± 1.3 (B)	20.3 ± 0.1 (C)	20.5 ± 0.4 (C)	19 ± 0.8 (C)	33.5 ± 0.4 (A)	25.6 ± 0.1 (B)	19.3 ± 0.1 (C)	16.3 ± 0.5 (D)	15.9 ± 0.8 (D)
O103	37.9 ± 5.2 (A)	28.4 ± 1.7 (B)	25.6 ± 0.1 (B)	24.7 ± 0.8 (B)	21.8 ± 2.3 (B)	33.4 ± 2 (A)	25 ± 1.9 (B)	18.0 ± 1 (C)	16.2 ± 0.4 (C)	15.5 ± 0.4 (C)
O111	44.0 ± 1.0 (A)	31.9 ± 0 (B)	27.1 ± 0.5 (C)	24.2 ± 0.2 (D)	22.3 ± 0.9 (E)	35 ± 0.7 (A)	28.4 ± 0.9 (B)	21.3 ± 0.8 (C)	17.8 ± 0.1 (D)	16.1 ± 0.4 (E)
O121	34.1 ± 4.9 (A)	28.5 ± 4 (AB)	25.3 ± 2.1 (B)	24.5 ± 1.8 (B)	22.1 ± 1.5 (B)	33.9 ± 0.3 (A)	26.7 ± 2 (B)	21.2 ± 2.5 (C)	17.1 ± 0.1 (C)	16.9 ± 2.1 (C)
O145	37.4 ± 3.3 (A)	25.8 ± 3.5 (B)	21.4 ± 0.6 (BC)	19.5 ± 0.1 (C)	18.9 ± 1.1 (C)	34.9	29.0 ± 2.2 (A)	21.0 ± 0.5 (B)	18.4 ± 0.1 (B)	18.3 ± 1.6 (B)
O157	43.8	35.2 ± 0.1 (A)	26.2 ± 1.7 (B)	21.9 ± 1.0 (C)	16.2 ± 1.3 (D)	ND	33.6 ± 0.7 (A)	28.1 ± 2.8 (B)	24.8 ± 0.7 (B)	15.9 ± 1.4 (C)

^aNone of the 4-h enrichment samples tested positive by either LAMP or qPCR. Average T_t or C_T values were calculated based on two independent repeats. When there is no standard deviation, only one repeat generated T_t or C_T values. In each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters in parentheses are statistically significantly different (P < 0.05).

^bND, not detected.

Table 7.

Comparison of LAMP and qPCR assays of spinach samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of individual STEC strains of seven target serogroups and tested after various enrichment periods

No. of cells/25 g and target group	Avg LAMP Tt (min)a after enrichment for:					Avg qPCR CT (cycles)a after enrichment for:
	6 h	8 h	10 h	12 h	24 h	6 h	8 h	10 h	12 h	24 h
1–2
O26	32.4 ± 0.3 (A)	25.6 ± 0.6 (B)	23.6 ± 0.1 (C)	21.4 ± 0.3 (CD)	22.1 ± 1.3 (D)	34.8 ± 0.5 (A)	25.4 ± 1.8 (B)	22.8 ± 0.6 (BC)	20.4 ± 0.4 (C)	17.4 ± 1.4 (D)
O45	26.8 ± 1.0 (A)	21.4 ± 0.1 (B)	20.9 ± 1.8 (B)	18.7 ± 1.4 (B)	19.1 ± 0.5 (B)	33.0 ± 1.0 (A)	24.6 ± 0.6 (B)	19.5 ± 0.1 (C)	18.1 ± 0.4 (C)	16.4 ± 0.4 (D)
O103	36.8 ± 0.3 (A)	27.1 ± 1.0 (B)	25.8 ± 0.3 (BC)	22.9 ± 2.3 (CD)	22.3 ± 1.8 (D)	31.6 ± 0.5 (A)	23.8 ± 0.5 (B)	17.7 ± 1.6 (C)	16.5 ± 0.6 (C)	16.0 ± 1.3 (C)
O111	41.2 ± 5.7 (A)	27.7 ± 1.8 (B)	25.4 ± 0.8 (B)	23.6 ± 1.1 (B)	23.6 ± 1.3 (B)	35.1 ± 0.8 (A)	27.4 ± 2.1 (B)	20.7 ± 2.3 (C)	18.7 ± 2.0 (C)	18.2 ± 0.2 (C)
O121	36.4 ± 1.2 (A)	27.0 ± 1.8 (B)	24.8 ± 0.1 (BC)	24.0 ± 1.1 (BC)	24.3 ± 0.5 (C)	32.2 ± 1.9 (A)	24.9 ± 1.6 (B)	19.7 ± 0.7 (C)	17.8 ± 0.9 (C)	17.9 ± 0.5 (C)
O145	33.8 ± 3.1 (A)	23.8 ± 0.9 (B)	20.7 ± 0.8 (B)	18.2 ± 0.6 (B)	20.6 ± 5.1 (B)	34.7	27.6 ± 0.5 (A)	20.6 ± 2.1 (B)	19.2 ± 2 (B)	18.6 ± 0.9 (B)
O157	37.4 ± 1.9 (A)	23.0 ± 0.4 (B)	18.7 ± 0.3 (C)	16.6 ± 0.4 (C)	16.5 ± 0.5 (C)	34.5 ± 0.4 (A)	27.3 ± 0.6 (B)	20.7 ± 0.4 (C)	17.2 ± 0.5 (D)	15.4 ± 1.3 (E)
10–20
O26	27.2 ± 0.6 (A)	20.7 ± 1.0 (B)	19.5 ± 0.5 (B)	18.4 ± 0.4 (B)	18.4 ± 3.0 (B)	30.6 ± 0.2 (A)	23.3 ± 1.1 (B)	17.9 ± 0.7 (C)	16.9 ± 0.9 (C)	16.1 ± 1.6 (C)
O45	23.8 ± 0.3 (A)	20.3 ± 0.8 (B)	19.7 ± 1.2 (B)	18.6 ± 1.3 (B)	18.6 ± 0.4 (B)	29.2 ± 0.2 (A)	21.6 ± 0.5 (B)	17.9 ± 0.8 (C)	17.2 ± 0 (C)	16.7 ± 0.7 (C)
O103	31.2 ± 2.4 (A)	26.3 ± 1.8 (AB)	23.9 ± 1.1 (B)	23.2 ± 2.4 (B)	23.8 ± 3.3 (B)	29.0 ± 0.2 (A)	21.0 ± 0.9 (B)	17.5 ± 0.1 (C)	16.8 ± 0.6 (C)	16.0 ± 1.9 (C)
O111	33.3 ± 0.6 (A)	26.7 ± 1.1 (B)	25.6 ± 1.4 (B)	23.5 ± 0.7 (B)	23.8 ± 3.7 (B)	31.1 ± 1.3 (A)	23.3 ± 1.5 (B)	18.6 ± 1.8 (C)	17.7 ± 0.3 (C)	16.8 ± 1.8 (C)
O121	33.6 ± 0.6 (A)	25.3 ± 1.4 (B)	24.1 ± 1.5 (B)	24.3 ± 1.4 (B)	22.5 ± 0.8 (B)	30.4 ± 0.5 (A)	22.3 ± 0.9 (B)	18.5 ± 0.5 (C)	16.9 ± 0.5 (CD)	16.6 ± 0.8 (D)
O145	28.8 ± 3.7 (A)	24.7 ± 1.9 (AB)	20.9 ± 3 (BC)	20.2 ± 1.0 (BC)	18.1 ± 1.2 (C)	32.2 ± 1.4 (A)	24.8 ± 0.9 (B)	20.1 ± 1.8 (C)	19.1 ± 1.4 (C)	18.8 ± 2.3 (C)
O157	27.2 ± 0.1 (A)	20.8 ± 1.1 (B)	17.5 ± 0.4 (C)	16.5 ± 1.4 (C)	16.5 ± 0.4 (C)	31.3 ± 2.3 (A)	23.8 ± 2.3 (B)	18.4 ± 1.5 (C)	16.9 ± 2.1 (C)	16.3 ± 0.2 (C)

^aNone of the 4-h enrichment samples tested positive by either LAMP or qPCR. Average T_t or C_T values were calculated based on two independent repeats. When there is no standard deviation, only one repeat generated T_t or C_T values. In each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters in parentheses are statistically significantly different (P < 0.05).

Table 6.

Comparison of LAMP and qPCR assays of lettuce samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of individual STEC strains of seven target serogroups and tested after various enrichment periods

No. of cells/25 g and target group	Avg LAMP Tt (min)a after enrichment for:					Avg qPCR CT (cycles)a after enrichment for:
	6 h	8 h	10 h	12 h	24 h	6 h	8 h	10 h	12 h	24 h
1–2
O26	33.8 ± 0.4 (A)	27.2 ± 0.5 (B)	24.9 ± 1.1 (BC)	23.8 ± 2.4 (C)	23.8 ± 0.8 (C)	33.8 ± 0.5 (A)	24.5 ± 0.4 (B)	18.8 ± 0.6 (C)	17.8 ± 0.7 (C)	18.4 ± 0.1 (C)
O45	26.8 ± 0.4 (A)	21.9 ± 0.5 (B)	19.0 ± 0.1 (C)	19.3 ± 0.3 (C)	18.8 ± 0.4 (C)	32.7 ± 0.6 (A)	24.2 ± 1.3 (B)	18.5 ± 0.8 (C)	16.0 ± 0.2 (D)	16.3 ± 0.1 (D)
O103	31.7 ± 2.0 (A)	23.9 ± 1.8 (B)	18.8 ± 5.2 (B)	22.0 ± 0.6 (B)	23.0 ± 0.2 (B)	30.7 ± 0.2 (A)	23.8 ± 0.6 (B)	17.2 ± 0.5 (C)	14.6 ± 0.2 (E)	15.8 ± 0.1 (D)
O111	33.1 ± 1.3 (A)	25.9 ± 0.2 (B)	21.8 ± 0.3 (C)	20.9 ± 0.4 (C)	21.2 ± 0.5 (C)	32.4 ± 1.4 (A)	23.9 ± 0.4 (B)	18.2 ± 0.1 (C)	15.6 ± 0.3 (D)	16.4 ± 0.1 (D)
O121	40.5 ± 1.3 (A)	29.1 ± 0.3 (B)	26.2 ± 0.9 (BC)	28.2 ± 1.6 (CD)	24.9 ± 0.3 (D)	30.9 ± 0.1 (A)	23.6 ± 0.7 (B)	17.1 ± 0.8 (C)	15.3 ± 0.6 (D)	15.3 ± 0.2 (D)
O145	34.9 ± 2.2 (A)	25.8 ± 4.4 (B)	22.4 ± 1.1 (B)	21.7 ± 0.1 (B)	22.1 ± 1.9 (B)	33.2 ± 0.7 (A)	26.2 ± 0.4 (B)	20.9 ± 0.7 (C)	17.2 ± 0.7 (D)	18.4 ± 0.7 (D)
O157	35.0 ± 0.4 (A)	22.9 ± 0.4 (B)	19.7 ± 0.6 (C)	16.7 ± 0.8 (D)	14.7 ± 0.6 (E)	NDb	29.8 ± 0.1 (A)	23.5 ± 0.4 (B)	17.8 ± 0.7 (C)	16.0 ± 0.2 (D)
10–20
O26	30.0 ± 1.5 (A)	24.9 ± 1.3 (B)	23.6 ± 2.2 (B)	23.3 ± 2.8 (B)	22.9 ± 1.0 (B)	28.4 ± 0.2 (A)	20.5 ± 0.1 (B)	16.0 ± 1.0 (C)	16.2 ± 0.7 (C)	17.4 ± 0.2 (C)
O45	23.0 ± 0.2 (A)	20.1 ± 0.4 (B)	18.7 ± 0.2 (C)	19.1 ± 0.6 (CD)	18.0 ± 0.1 (D)	28.2 ± 0.6 (A)	20.7 ± 0.7 (B)	16.7 ± 0.1 (C)	16.2 ± 0.6 (C)	16.4 ± 0.2 (C)
O103	28.7 ± 2.3 (A)	22.5 ± 0.2 (B)	21.3 ± 1.3 (B)	22.0 ± 0.6 (B)	19.9 ± 1.1 (B)	27.7 ± 0.5 (A)	19.6 ± 0.2 (C)	15.3 ± 1.4 (C)	14.4 ± 0.2 (C)	16.0 ± 0.5 (C)
O111	31.3 ± 3.5 (A)	24.5 ± 1.5 (B)	21.5 ± 0.6 (B)	21.3 ± 0.2 (B)	21.2 ± 0.1 (B)	30.4 ± 0.9 (A)	22.4 ± 0.7 (B)	17.0 ± 0.2 (CD)	16.0 ± 0.4 (D)	17.6 ± 0.4 (C)
O121	32.3 ± 1.1 (A)	28.6 ± 0 (B)	26.4 ± 0.6 (BC)	26.7 ± 0.8 (CD)	24.6 ± 1.1 (D)	28.0 ± 0.8 (A)	20.9 ± 0.3 (B)	16.1 ± 0.1 (C)	15.1 ± 0.1 (C)	15.5 ± 0.1 (C)
O145	31.5 ± 3.7 (A)	24.9 ± 1.0 (B)	23.4 ± 0.4 (B)	22.7 ± 0.6 (B)	22.3 ± 0.1 (B)	30.3 ± 1.1 (A)	23.0 ± 0.5 (B)	18.1 ± 0.8 (C)	17.4 ± 0.2 (C)	18.7 ± 0.3 (C)
O157	29.6 ± 0.7 (A)	22.0 ± 0.8 (B)	16.4 ± 0.6 (C)	14.9 ± 0.1 (D)	15.0 ± 0.1 (D)	32.7 ± 1.0 (A)	26.8 ± 0.1 (B)	19.8 ± 0.7 (C)	16.0 ± 0.4 (D)	15.2 ± 0.1 (D)

^aNone of the 4-h enrichment samples tested positive by either LAMP or qPCR. Average T_t or C_T values were calculated based on two independent repeats. In each row within LAMP or qPCR, T_t or C_T values followed by different uppercase letters in parentheses are statistically significantly different (P < 0.05).

^bND, not detected.

DISCUSSION

The seven LAMP assays developed in the present study, each targeting one of the leading STEC serogroups (O26, O45, O103, O111, O121, O145, and O157), were rapid (15 to 45 min), specific (100% inclusivity and 100% exclusivity among 120 strains tested), sensitive (1 to 20 CFU/reaction mixture in pure culture and 10³ to 10⁴ CFU/g in spiked ground beef, beef trim, lettuce, and spinach), and quantitative (R² = 0.867 to 0.999). With 6 to 8 h of enrichment, these assays accurately detected two low levels (1 to 2 and 10 to 20 CFU/25 g) of respective STEC strains in the four food matrices tested. To our knowledge, this is the first study applying the novel LAMP technology to specifically detect STEC serogroups of significant public health concern in multiple high-risk foods.

Parallel to the work presented here, we recently developed and evaluated a set of serogroup-independent STEC LAMP assays by targeting common STEC virulence genes (stx₁, stx₂, and eae), and the performance of those assays in pure cultures and ground beef closely mimicked that observed in the present study (47). Additional LAMP assays for STEC (24, 29, 32) and other pathogenic E. coli groups (41, 50, 51) have been reported. In direct contrast, the development of serogroup-specific LAMP assays was confined to E. coli O157 only (46, 52, 53). The detection limits of the seven serogroup-specific LAMP assays developed in the present study (1 to 20 CFU/reaction mixture) fell well within the limits reported for previous E. coli LAMP studies (24, 26, 29, 41, 46, 50–53), i.e., 0.7 to 100 CFU/reaction mixture. In addition, the findings of this study corroborated those of two others (8, 47) that suggested comparable sensitivities of LAMP and qPCR. It is notable that LAMP assays took less time to run (50 min) than qPCR assays (at least 1.5 h) recently developed by USDA scientists (13, 15), therefore markedly shortening the total assay time.

In the present study, serogroup-specific wzx (for O103 and O145 only) or wzy genes within the O-antigen gene clusters of the seven STEC serogroups were chosen as targets to design LAMP primers. Initial attempts were made to use wzy genes of O103 and O145, but suitable LAMP primers were not obtained (data not shown). Based on sequence alignments performed previously and here, all of the wzx and wzy genes were highly specific for each STEC serogroup (9–11, 14, 16, 48, 49). Consequently, all of the seven LAMP assays possessed 100% inclusivity and 100% exclusivity among the 120 bacterial strains tested, a specificity similar to that reported previously for STEC O157 LAMP assays targeting the rfbE gene (46, 52). Notably, in the E. coli O157 O-antigen gene cluster (GenBank accession number AF061251), the rfbE gene was immediately downstream of wzx and the two genes overlapped by 4 bp (49).

LAMP positive reactions are commonly detected by gel electrophoresis, visual endpoint judgment of turbidity or color change, or real-time monitoring of turbidity or fluorescence signals (35). Using the last approach, the ability of LAMP to quantitatively detect Vibrio spp. and Salmonella were demonstrated on turbidity-based (7, 8, 21, 22) and fluorescence-based (1, 7, 21) platforms. However, the quantitative detection of E. coli O157 by LAMP has not been examined previously. In this study, the R² values fell between 0.945 and 0.996 for respective STEC serogroup cells ranging from 10⁵ to 10¹ (10⁰ for O45) CFU/reaction mixture in pure culture and 0.867 to 0.999 for cells between 10⁷ and 10⁴ (10³ for O45) CFU/g in spiked food samples, suggesting good quantitative capabilities. However, much-delayed T_t values, sometimes negative LAMP results, were observed when the STEC serogroup cells fell below 10¹ CFU/reaction mixture (data not shown). This poor quantification of LAMP at low cell levels have been reported previously (2, 8). Nonetheless, it is important to note that whenever enrichment is incorporated in the detection procedure, quantification is no longer relevant (17).

LAMP assays have been applied for detecting STEC or E. coli O157:H7 in food samples, usually after overnight enrichment (23–25, 38, 46, 47). A recent study (38) reported that 45 to 50% of liver samples inoculated with 1 to 4 CFU/25 g of E. coli O157:H7 tested positive by LAMP after overnight enrichment but only 10 to 35% of them were positive by culture. Two earlier reports (23, 24) by the same group stated that among ground beef samples inoculated with approximately 10 CFU/25 g of E. coli O157 or O26 strains, LAMP had 100% detection rates after 24 h of enrichment, whereas culture methods detected 100% of the samples spiked with STEC O157 but only 50 to 80% of those spiked with STEC O26 (23). Without enrichment, a LAMP detection limit of 4.1 × 10⁴ CFU/ml for E. coli O157 in raw milk was reported (46). LAMP assays developed in the present study had 10³ to 10⁴ CFU/g detection limits in spiked food samples (ground beef, beef trim, lettuce, and spinach), comparable to those reported for our recently developed STEC LAMP assays (47) and the raw-milk study (46). In ground beef samples spiked with two low levels (1 to 2 and 10 to 20 CFU/25 g) of the respective STEC strains, positive LAMP results occurred after 6 h of enrichment and consistently thereafter, which were superior to the results obtained in the liver study (38). We also found that LAMP performed better than qPCR in obtaining consistent and stable results with spiked beef trim, lettuce, and spinach, as LAMP could generate stable data after 8 h of enrichment while qPCR almost always required 10 h of enrichment. LAMP has been reported previously to be more robust than PCR-type assays with regard to tolerance to inhibitors in clinical samples and other biological substances (12, 28) and therefore may be more advantageous in food testing.

With the current and anticipated zero-tolerance policy for E. coli O157:H7 and the top six non-O157 STEC serotypes in raw beef products, it is critical that available methods are capable of rapidly and accurately detecting very low levels of these pathogens in food, usually after an indispensable enrichment step (17). Given the rapidity, sensitivity, specificity, and robustness of LAMP demonstrated by serogroup-specific LAMP assays developed in the present study and serogroup-independent LAMP assays developed in our recent study (47), these assays may effectively serve as two-step method of screening for STEC strains in ground beef, beef trim, or produce samples during routine sample testing. Also, considering the emerging and evolving nature of STEC serogroups involved in human illnesses, similar serogroup-specific LAMP assays targeting other emerging serogroups may be developed in the future and add to the panel of available assays. In conclusion, the rapid, specific, sensitive, and simple platform of LAMP assays may present another valuable tool for the meat and produce industry and regulatory agencies to better control potential STEC risks associated with the consumption of these high-risk foods.