1828  Gao et al.: Journal of AOAC International Vol. 101, No.
6, 2018
Food Biological Contaminants
Evaluation of a Multiplex PCR for Detection of the Top
Seven Shiga Toxin-Producing Escherichia coli Serogroups
­in Ready-to-Eat Meats, Fruits, and Vegetables
Anli Gao, Jennifer Fischer-Jenssen, Colin Cooper, Honghong Li, Jiping Li, Shu Chen,
and Perry ­Martos
University of Guelph, Laboratory Services Division, 95 Stone Rd W, Guelph, ON N1H8J7 Canada
Background: Ready-to-eat (RTE) meats, fruits, and
vegetables contaminated by Shiga toxin producing
Escherichia coli (STEC) raise serious concerns
because they are often consumed directly without
further processing. Objective: To evaluate a multiplex
PCR for the detection of STEC across food categories.
Methods: Samples (25 g) from seven RTE meat and
nine fruit and vegetable matrices were inoculated with
each of seven STEC (O157:H7, O26, O121, O145, O45,
O103, O111) strains targeting 10 CFU/25 g, enriched
in 225 mL of modified tryptone soya broth (mTSB),
and tested by a multiplex real-time PCR for stx and
eae genes, following U.S. Department of Agriculture
(USDA) Food Safety and Inspection Service (FSIS)
Microbiology Laboratory Guidebook (MLG) 5B,
which was originally validated for meat products
and environmental sponge. Results: The mTSB was
successful at enriching for STEC in RTE meat, fruit,
and vegetable matrices, except for sprouts; however,
mEHEC resulted in successful enrichment of target
organisms in mung bean sprouts. Suppression of eae
results by stx in PCR was observed in six fruit and
vegetable matrices. Conversely, suppression of stx
gene by eae was not observed. PCR solely targeting
eae is recommended if a fruit or vegetable sample
tested positive for stx and negative for eae. Despite
the significant effect from food matrix, strain, and
experimental batch, the cycle threshold of PCR was
<30 in inoculated samples, and mostly 30–42 and up
in uninoculated samples. Conclusions: The multiplex
PCR can be adopted for detection of all seven
regulated STEC in RTE meat, fruit and vegetable
matrices after validation with cut off value selected
and justified based on real samples.
C
ontamination from Shiga toxin producing Escherichia
coli (STEC) is a significant and highly relevant concern
in food safety. The development of methods for the
detection of STEC has been the topic of significant interest in
Received January 9, 2018. Accepted by AH March 9, 2018.
This research is funded by Ontario Ministry of Agriculture, Food
and Rural Affairs (OMAFRA) via University of Guelph–OMAFRA
Partnership.
Corresponding author’s e-mail: agao@uoguelph.ca
DOI: https://doi.org/10.5740/jaoacint.18-0010
recent decades, especially as STEC serotype, O157:H7, has
become recognized as a food-borne pathogen (1). Additionally,
as non-O157 strains are increasingly detected as important
food-borne pathogens in outbreaks, six other STEC serogroups
(O26, O45, O103, O111, O121, and O145) are now being
regulated by the U.S. Department of Agriculture (USDA), Food
Safety and Inspection Service (FSIS) in raw beef products (2).
A number of methods have been developed and evaluated by
research groups, such as for the detection of STEC O104 in
sprouts (3); O157:H7 and non-O157 in leafy green vegetables,
cilantro, sprouts and milk (4, 5); O157 in various food samples
(6); O157:H7 in alfalfa sprouts (7); non-O157 in pure culture and
food (8), and the USDA Microbiology Laboratory Guidebook
(MLG) 5B method for raw meat (9). However, despite these
efforts, information on the evaluation of methods for the
detection of these seven STEC serogroups in different food
categories remains limited (10). To our knowledge, methods for
the detection of both O157 and non-O157 STEC across broad
food categories have not been reported. Fruits, vegetables, and
ready-to-eat (RTE) meats contaminated by STEC raise a more
serious concern to public health, unlike raw beef, which is
cooked prior to consumption. The objective of this study was
to develop a screening method for the detection of these STEC
serogroups in fruits, vegetables, and RTE meats prior to bacterial
culture. This work presents an adopted multiplex real-time PCR
screening method that can be used to detect all seven regulated
STEC serogroups across different food categories.
Materials and Methods
Experimental Design
Seven RTE meats (cooked chicken, cooked deli, cooked
ham, cooked sausage, dry cured ham, salami, and sausage) and
nine fruit and vegetable (apple cider, broccoli, carrot, lettuce,
raspberry, spinach, mung bean sprouts, strawberry, and tomato)
samples were inoculated with each of the seven STEC strains
(Table 1) in a factorial design. The experiments were carried
out in eight batches. Each batch consisted of one blank control
sample and two inoculated replicate samples from each of the
16 food matrices, with only one STEC strain employed for
artificial inoculation. Seven batches were conducted to cover
each of the seven STEC strains. An additional batch of mung
bean sprout samples was inoculated with the seven STEC
strains individually and enriched in a proprietary media mEHEC
(BioControl) when it was observed that modified tryptone soya
broth (mTSB; Oxoid) did not provide successful enrichment of
 Gao et al.: Journal of AOAC International Vol. 101, No. 6, 2018  1829

Table 1.  Shiga toxin-producing Escherichia coli strains of a Ct value, “undetermined” is given to a PCR reaction result
used for inoculation of food samples when there is not enough target DNA present. A value of 45 was
assigned to the Ct value of stx gene when it was “undetermined”
No. Serotype Strain eae stx1 stx2
in PCR. Otherwise, the samples that were “undetermined”
1 O26:H11 EC910114 + + – by PCR would be excluded by SAS in data analysis because
2 O145:NM EC19960923 + – + “undetermined” is not a recognizable quantitative value. The
3 O45:H2 EC19940040 + + – effect of the food matrix and STEC strain and their interaction
on the Ct value in PCR was analyzed by mixed-model analysis
4 O103:H2 EC930004 + + –
of variance (ANOVA) in SAS 9.3 with Enterprise Guide 4.3
5 O111:NM STXX + + +
(SAS Institute Inc., Cary, NC). Uninoculated samples and
6 O121:H19 EC19990161 + – + inoculated samples were analyzed separately. The effect of the
7 O157:H7 ATCC 43889 + + + food matrix and strains was based on inoculated samples only.
A significant difference was accepted at P < 0.05. Scatterplots
of stx gene Ct value from inoculated and uninoculated samples
STEC in sprouts. Randomly selected samples including both
for individual matrix and for individual STEC strains, as well
uninoculated and inoculated samples were tested by PCR in
as the scatterplot of stx gene Ct value between duplicate tests,
duplicate to determine the uncertainty of PCR detection.
were created in Microsoft Excel 2010. The uncertainty of PCR
detection, or the repeatability SD of Ct value, was estimated
Sample Preparation
based on the samples tested in duplicate following a procedure
described in P19-CALA Measurement Uncertainty Policy (12).
Samples were purchased from retail stores and artificially
inoculated by targeting 10 CFU/25 g sample. The actual
Results
inoculation levels turned out to be from 10 to 40 CFU/25 g, or
0.25 to 1.6 CFU/g of sample, as monitored by quantitative plate
counts of the inoculum preparation. Inoculated and uninoculated Method for DNA Extraction
samples (25 g) were enriched with 225 mL of mTSB containing
mitomycin C (100 ng/mL), homogenized for 30 s in a The Ct values of stx gene amplification in multiplex PCR
stomacher, and incubated at 35 ± 2°C for 20 to 24 h. Mung were compared between the template DNA obtained following
bean sprout samples were also enriched in mEHEC, followed the MLG 5B protocol and the DNA prepared using InstaGene
by incubation at 42 ± 1°C for 48 h in the last batch following matrix (Figure 1). The Ct values of stx gene ranged from 13 to 25
the manufacturer’s instructions after it was realized that mTSB in 80 inoculated RTE meat, fruit, and vegetable samples (except
did not work for sprouts, as mentioned above. All uninoculated sprouts) and from 25 to “undetermined” in 40 uninoculated
samples were negative by enzyme-linked immunosorbent assay samples. The template DNA samples prepared using InstaGene
and immunoblotting following the protocol described (11). resulted in lower Ct values for inoculated samples. The gap of
Ct values between uninoculated and inoculated samples seemed
DNA Extraction and PCR larger when DNA was prepared by InstaGene than the gap when
the DNA was prepared following MLG 5B. When a Chi-square
The enrichment cultures were subjected to DNA extraction as test was conducted using cutoff values between 28 and 30, both
described in USDA FSIS MLG 5B (9). Subsamples of cultures methods detected a similar number of positive samples (data
without large debris (1.4 ± 0.1 mL) were centrifuged at 10 000 × g not shown). This suggests that for screening purposes, these two
for 5 min. The pellet was then suspended in 500 ± 50 μL of DNA extraction methods are not significantly different.
0.85% saline. Samples were then again centrifuged at 10  000 × g
for 3 min. The pellet was suspended in 90 ± 9 μL TE buffer
Ct value of stx gene from DNA following MLG 5B

(10 mM Tris, 1 mM EDTA, pH 7.4) and heated to 97 ± 2°C 45

for 15 ± 1 min. Samples were centrifuged at 16 000 × g for
4 min. Supernatants were then transferred to a new tube for 40 y = 0.847x + 3.97
R2 = 0.878
PCR analysis. In order to optimize the protocol, the first two
batches of sample DNA were extracted using InstaGene Matrix 35
(BioRad) in parallel with the USDA centrifugation and heating
method as described above. Then, either the Instagene Matrix or 30
the centrifugation and heating protocol was employed alone after
25
it was found that there was no significant difference between
the two methods. The PCR reaction for stx and eae genes was Uninoculated
20
carried out as described in USDA FSIS 5B Appendix 1 for Inoculated
primers, probes, and reagents and Appendix 3 for PCR platform 15 Linear (Regression)
instruction, data analysis, and control results interpretation (9).
10
Data Analysis 10 15 20 25 30 35 40 45
Ct value of stx gene from DNA using InstaGene
While the multiplex PCR provides cycle threshold (Ct) values
Figure 1.  Scatterplot of cycle threshold (Ct) values between
for both stx and eae genes, this study focused on stx gene Ct for the DNA templates prepared following USDA MLG 5B and using
STEC screening given the specificity of the gene to target. Instead InstaGene Matrix (BioRad) from identical samples.
1830  Gao et al.: Journal of AOAC International Vol. 101, No. 6, 2018

Effect of Food Matrix and STEC Strain
The stx gene Ct value was affected by the type of food matrix,
the STEC strain, and separate batches (Table 2). The larger an
F value, the smaller a P value, and the higher confidence to accept
a significant difference. The F value corresponding to a P = 0.05
for the effect of matrix in Table 2 ranges from 2.01 to 2.07, and
the F value for P = 0.05 for strain/batch ranges from 3.67 to 3.70.
The calculated F values for the effect of matrix and strain/batch
(152.25, 85.79) are much larger than their F values at P = 0.05,
resulting in P values (<0.0001) much less than 0.05. The interaction
between food matrix and STEC strain is also significant; however,
the mean square explained by the interaction (8.78) is much
smaller than that explained by either matrix (323.86) or strain/
batch (182.49; Table 2). The scatterplot of stx gene Ct values
against food matrix in both inoculated and uninoculated samples
is presented in Figure 2, and that against strain is in Figure 3.
A cutoff value of 30 was selected for all matrices and yielded 4
false-positives and 0 false-negative (Figures 2 and 3).
Table 2. ANOVA for the effect of food matrix and STEC
strain/batch on stx cycle threshold (Ct) value in PCR based
on inoculated samples
Sum of
Source DFa squares
Matrix 16 5181.83
Strain/batch 6 1094.92
Matrix × strain/batch 96 843.05
Residual 273 580.73
a
  Degree of freedom.
Strain/batch, instead of strain alone, is defined in ANOVA
(Table 2) because our experiment was conducted in batches,
in which samples from each of the 16 food matrices were
inoculated with one STEC strain as mentioned above.
Therefore, variation between batches also contributed to
the significance. In other words, strain/batch covers both the
variation of strains and the variation of experimental batches.
There is no way to mathematically separate them in this study.
Although a level of 10 CFU/25 g sample had been targeted at
inoculation, the difference of inoculum level between strains/
batches had been noticed within a range of 10 to 40 CFU/25 g
(Figure 3). Hopefully, the difference in inoculum level within
such a range will not introduce significant variation in the
PCR Ct value after a 20 to 24 h enrichment. If any, this type
of variation should have been caught in the variation of strain/
batch in Table 2.
Amplification of eae Gene in Multiplex PCR
Multiplex PCR yielded Ct values for the amplification
of both stx and eae genes. The Ct value of stx and eae genes
correlated significantly with each other among the samples
yielded a Ct value < 45 from both genes (y = 0.96x + 1.38;
R2 = 0.98, where y is the Ct value of eae and x is of stx; N = 501
tests). At the same time, among the tests of inoculated fruit and
Mean vegetable samples (n = 221), 19 (9%; 4 carrot, 3 lettuce, 1 sprout,
square F value P value 3 raspberry, 5 strawberry, and 3 tomato) produced stx Ct values
323.86 152.25 <0.0001 from 13 to 25, while their eae gene was “undetermined.” The 19
182.49 85.79 <0.0001 samples consisted of eight PCR tests of samples inoculated with
O157:H7, seven tests with O111:NM, two with O145:NM, one
8.78 4.13 <0.0001
with O45:H2, and one with O121:H19. All of the seven STEC
2.13
strains are known carriers of both stx and eae genes (Table 1).
Further PCR reactions targeting only the eae gene revealed

45

40

35

30
stx gene Ct Value

25

20
Sprouts- in mEHEC**

15
Cooked Ham
Cooked Deli

Apple Cider

Strawberry
Raspberry
Dry Cured

10
Sausage
Sausage

Sprouts*
Spinach
Chicken

Broccoli
Cooked

Tomato
Cooked

Lettuce
Salami

Carrot
Ham

5
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Food Matrix Uninoculated Inoculated Proposed Cutoff

Figure 2.  The stx gene cycle threshold (Ct) value in ready-to-eat meat, fruit, and vegetable matrices enriched in mTSB except where
indicated. *mTSB was ineffective for mung bean sprouts. **mEHEC worked significantly better than mTSB for mung bean sprouts.
 Gao et al.: Journal of AOAC International Vol. 101, No. 6, 2018  1831

45

40

35 Uninoculated RTE
meat

Uninoculated fruit
30 & vegetable
stx gene Ct value

Inoculated RTE
meat
25
Inoculated fruit &
vegetable

20
Proposed Cutoff

15
USDA & InstaGene

USDA & InstaGene

10
InstaGene

InstaGene
O121:H19
O111:NM

O145:NM
O103:H2

O26:H11
O15:H7

O45:H2
USDA

USDA

USDA
5
0 1 2 3 4 5 6 7
STEC strain and batch DNA extraction method

Figure 3.  The stx gene cycle threshold (Ct) value, DNA extraction protocol, and inoculum level of STEC strain in samples from
ready-to-eat meat, fruit, and vegetable matrices (except sprouts) enriched in mTSB. Strain/batch inoculum at CFU/25 g: O103:H2, 31; O111:NM,
10; O121:H19, 31; O145:NM, 31; O157:H7, 29; O26:H11, 11; and O45:H2, 40. USDA, DNA extracted following USDA FSIS MLG 5B. InstaGene,
DNA extracted using InstaGene Matrix (BioRad).

that these samples were actually positive for the eae gene as 45
well. Suppression of eae gene in multiplex PCR reaction was
stx gene Ct value of replicate 2

not found in any of the tests of RTE meat samples (n = 171 40

tests). Conversely, the suppression of the stx gene by eae was
35
not observed in our study.
30
Uncertainty of PCR Detection
25

The scatterplot of stx gene Ct value between duplicate 20

tests are presented in Figure 4. Uncertainty was calculated Inoculated Uninoculated
separately for the inoculated samples and the uninoculated 15
because of an apparently larger variation between the duplicate
tests for samples with a high Ct value. The repeatability SD 10
10 15 20 25 30 35 40 45
of stx Ct value is 0.74, with a 95% CI of ±1.5 Ct value for the stx gene Ct value of replicate 1
tests with Ct values less than 30 (n = 69). The SD and 95% CI
are 1.62 ± 3.2 Ct value for the tests with Ct values larger than Figure 4.  Scatterplot illustrating repeatability of stx gene cycle
30 (n = 35; Figure 4). threshold (Ct) value between duplicate tests.

Discussion
Unlike clinical samples, pathogens in foods are generally
present at much lower levels and are frequently injured during
food processing, which demands improved detection sensitivity
and pre-enrichment to resuscitate injured cells against
competition from high levels of normal flora (13). Therefore,
a successful enrichment is critical to selectively increase target
organisms to levels above the LOD of a method. The effect of
1832  Gao et al.: Journal of AOAC International Vol. 101, No. 6, 2018
the food matrix is one of the important factors to consider in
method development. Although mTSB resulted in successful
STEC enrichment for almost all RTE meat, fruit, and vegetable
samples, it was not successful for mung bean sprouts, likely
because of its specific high background flora. Therefore, the
use of a specialized broth for STEC enrichment was chosen,
and mEHEC resulted in successful enrichment for mung bean
sprouts. When mEHEC was used for alfalfa sprouts, a secondary
enrichment had to be employed, which might be due to the
dilution of bactericidal compounds, such as hypochlorite, used
to treat sprouts prior to packaging. To test whether hypochlorite
from sprout samples was inhibiting spiked STEC growth,
hypochlorite (100 ppm) was added to the spiked samples;
however, we found that the addition of hypochlorite did not
affect the growth and detection of STEC (data not shown). The
real reason is still to be investigated.
The multiplex PCR for stx and eae genes as described in
USDA FSIS MLG 5B for raw meat and environmental samples
worked for the detection of all seven STEC serogroups,
including O157:H7 in RTE meat, fruit, and vegetable samples;
however, given the possible suppression of the eae gene in
multiplex PCR when applied to fruit and vegetable samples,
caution should be used in the interpretation of test results when
a sample is positive for the stx gene and negative for eae.
Additional PCR solely targeting the eae gene is recommended
when such information is required for confirmation. Samples
for the eae results were compromised over five of the seven
STEC strains and six of the nine fruit and vegetable matrices.
This suggests that it might not be associated with particular
STEC strains or with specific matrices. It did not seem to
be a PCR inhibitor issue either because stx has never been
compromised. This also implies the requirement of further
optimization of the multiplex PCR as a screening method for
fruits and vegetables. It demonstrates that a method designed
for given food matrices should only be adopted with proper
study and validation.
The multiplex PCR originally designated for meat products
and environmental samples can be adopted for detection of
STEC in RTE meat, fruit, and vegetable matrices after proper
study and validation. Although uninoculated food samples with
a normal microbial background could have an “undetermined”
Ct value in PCR, most of them did yield Ct values from 30 to 42
(Figure 2 and Figure 3). It is recommended that a cutoff value be
established and justified based on the representative matrices to
be tested rather than pure cultures (14), and validation samples
should be selected and created under defined and controlled
conditions.
Acknowledgments
The authors thank Roger Johnson from the Public Health
Agency of Canada (Guelph, ON, Canada) for providing STEC
strains, and Susan Lee and Carlos Leon-Verlarde from the
Laboratory Services Division, University of Guelph (Guelph,
ON, Canada) for technical advice.
References
(1) Riley, L.W., Remis, R.S., Helgerson, S.D., Mcgee, H.B.,
Wells, J.G., Davis, B.R., Hebert, R.J., Olcott, E.S.,
Johnson, L.M., Hargrett, N.T., Blake, P.A., & Cohen,
M.L. (1983) N. Engl. J. Med. 308, 681–685. doi:10.1056/
nejm198303243081203
(2) U.S. Department of Agriculture (2012) Federal Register, 77,
pp 31975–31981
(3) Jinneman, K.C., Waite-Cusic, J.G., & Yoshitomi, K.J. (2012)
Food Microbiol. 30, 321–328. doi:10.1016/j.fm.2011.12.007
(4) Kase, J.A., Maounounen-Laasri, A., Son, I., Deer, D.M.,
Borenstein, S., Prezioso, S., & Hammack, T.S. (2012) Food
Microbiol. 32, 423–426. doi:10.1016/j.fm.2012.08.006
(5) Kase, J.A., Maounounen-Laasri, A., Son, I., Lin, A., &
Hammack, T.S. (2015) Food Microbiol. 46, 280–287.
doi:10.1016/j.fm.2014.08.020
(6) Weagant, S.D., Bryant, J.L., & Jinneman, K.G. (1995) J. Food
Prot. 58, 7–12. doi:10.4315/0362-028x-58.1.7
(7) Weagant, S.D., Jinneman, K.C., Yoshitomi, K.J., Zapata, R., &
Fedio, W.M. (2011) Int. J. Food Microbiol. 149, 209–217.
doi:10.1016/j.ijfoodmicro.2011.06.008
(8) Kanki, M., Seto, K., Harada, T., Yonogi, S., & Kumeda, Y.
(2011) Lett. Appl. Microbiol. 53, 167–173. doi:10.1111/j.1472-
765X.2011.03085.x
(9) U.S. Department of Agriculture (2014) Detection & Isolation
of non-O157 Shiga Toxin-Producing Escherichia coli (STEC)
from Meat Products & Carcass and Environmental Sponges,
Microbiology Laboratory Guide 5B.05
(10) Tzschoppe, M., Martin, A., & Beutin, L. (2012) Int. J. Food
Microbiol. 152, 19–30. doi:10.1016/j.ijfoodmicro.2011.10.009
(11) Atalla, H.N., Johnson, R., McEwen, S., Usborne, R.W., &
Gyles, C.L. (2000) J. Food Prot. 63, 1167–1172.
doi:10.4315/0362-028X-63.9.1167
(12) Canadian Association for Laboratory Accreditation Inc. (2010)
P19–CALA Measurement Uncertainty Policy, Revision 1.10,
Ottawa, ON, Canada
(13) Ge, B.L., & Meng, J.H. (2009) Jala 14, 235–241. doi:10.1016/j.
jala.2008.12.012
(14) Caraguel, C.G.B., Stryhn, H., Gagné, N., Dohoo, I.R., &
Hammell, K.L. (2011) J. Vet. Diag. Invest. 23, 2–15.
doi:10.1177/1040638711023001022
Copyright of Journal of AOAC International is the property of AOAC International and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.