Journal of Microbiological Methods 55 (2003) 141 – 147
Confirmation of viable E. coli O157:H7 by enrichment and
PCR after rapid biosensor detection
T. Bryan Tims, Daniel V. Lim *
Department of Biology, University of South Florida, 4202 East Fowler Ave, SCA 110, Tampa, FL 33620-5200, USA
Received 6 January 2003; received in revised form 28 February 2003; accepted 7 April 2003
Many rapid tests have been developed for the detection of Escherichia coli O157:H7 from complex matrices such as food
and water. However, many of these methods rely on traditional culture steps for confirmation, which can take an extra 24 – 48 h.
The fiber optic biosensor has been used to rapidly detect pathogens from complex matrices. In this paper, we demonstrate a
method using a rapid biosensor assay, recovery through a short enrichment, and PCR to detect and confirm the presence of at
least 103 CFU/ml of E. coli O157:H7 in a sample in less than 10 h.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Detection; Biosensor; PCR; E. coli O157:H7; Immunoassay
1. Introduction methods with selective media such as Sorbitol Mac-
Conkey agar and Rainbow agar (Manafi and Krems-
Escherichia coli O157:H7 is a facultative gram- maier, 2001), immunomagnetic separation of the
negative bacillus that has been implicated in outbreaks organism from the sample matrix followed by PCR
of illness due to ingestion of meats (Chapman et al., identification (Chapman et al., 2001; Uyttendaele et
2000), water, and uncooked fruits and vegetables al., 1999), and immunological techniques such as
(Pebody et al., 1999). This pathogen is known to ELISA and fiber optic biosensors (DeMarco and
cause diarrhea and hemolytic uremic syndrome (HUS) Lim, 2001; DeMarco et al., 1999; DeMarco and
in humans (DeCludt et al., 2000). It is the most Lim, 2002). Each of these methods has its own unique
common strain of Shiga toxin-producing E. coli set of drawbacks. Enrichment and plating often take
(STEC) in the United States, Canada, and the United 24 – 48 h to identify the organism. Immunological
Kingdom (Kaper, 1998). Many detection methods techniques cannot differentiate viable and nonviable
have been employed to rapidly detect low levels of cells and often require enrichments. PCR is more
organisms in complex matrices such as food. Current rapid than plating techniques, but it can be inhibited
techniques include traditional enrichment and plating by components of the sample matrix and cannot be
used to determine viability. PCR also requires DNA
* Corresponding author. Tel.: +1-813-974-1618; fax: +1-813-
purification from sample matrices. The solution to
974-3263. these problems lies in a combination of all of these
E-mail address: email@example.com (D.V. Lim). techniques.
0167-7012/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
142 T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147
We have previously reported a rapid fiber optic stained with ethidium bromide, destained with water,
biosensor assay that can detect low levels of E. coli and viewed under a UV light. Samples were consid-
O157:H7 in ground beef and apple cider in 30 min ered positive if a band at approximately 450 bp was
(DeMarco and Lim, 2001; DeMarco et al., 1999; visible with the naked eye.
DeMarco and Lim, 2002). We have also been able
to recover isolated colonies of the target organism 2.3. Antibodies
directly from the biosensor through enrichment after
completion of the fiber optic biosensor assay from a Lyophilized affinity-purified biotin-labeled anti-
complex matrix like ground beef (Kramer et al., body to E. coli O157:H7 was purchased from Kirke-
2002). In this study, we describe a 10 h procedure gaard and Perry Laboratories (KPL, Gaithersburg,
for selective enrichment of low levels of E. coli MD). This antibody was rehydrated using 0.1 M
O157:H7 recovered from fiber optic waveguides fol- phosphate buffer, pH 7.4.
lowed by selective enrichment and plating to confirm
viability and molecular identification by PCR. 2.4. Instrument and waveguide preparation
The Analyte 2000 is a portable, fiber optic wave
2. Materials and methods biosensor manufactured by Research International
(Woodinville, WA). Tapered polystyrene waveguides
2.1. Bacterial strains and culture conditions produced by Research International were used with
the Analyte 2000. These waveguides were prepared as
E. coli O157:H7 was obtained from Dr. Harvey described by DeMarco and Lim (2001) and coated
George (Massachusetts Department of Public State with the biotinylated antibodies from KPL.
Laboratory Institute, Jamaica Plains, MA). The
strain chosen was isolated from hamburger meat 2.5. Biosensor sample incubation and preparation
used at a taco stand implicated in a county fair
outbreak. All cultures were maintained on tryptic After waveguides were coated with biotinylated
soy agar (TSA, Remel, Lenexa, KS) plates at 4 jC. antibody, 1 ml of blocking buffer was added and
Cultures for assays and viable counts were grown incubated for 20 min at 25 jC (PBS, 2 mg/ml casein,
on TSA for 18 h in a 37 jC incubator and serially 2 mg/ml BSA). Waveguides were then rinsed with
diluted in sterile 0.01 M phosphate-buffered saline, 1 ml PBS with 0.01% Tween 20 (PBST). One milli-
0.85% NaCl, pH 7.4 (PBS) for use in experiments. liter of sample of E. coli O157:H7 at various concen-
trations was added and incubated for 10 min.
2.2. PCR amplification Waveguides were then rinsed with 1 ml PBST. After
rinsing, waveguides were either cut into approxi-
PCR was performed with primers specific to the 3V mately 1-mm sections and added directly to the
end of the E. coli O157:H7 eaeA gene (Louie et al., PCR reaction or incubated for 10 min with 0.5 M
1994). Each 50 Al reaction volume contained the glycine, 0.5 M NaCl, pH 3.2. When the waveguides
following: eaeA primer set as defined by Louie et al. were treated with glycine, the glycine-treated material
(1994) at a concentration of 0.4 AM, 30.25 Al sterile was used directly as the PCR sample for non-enriched
water, 200 AM each deoxynucleoside triphosphate, samples, or the cells on the waveguide and in the
1 Â iTaq buffer (20 mM Tris –HCl, 50 mM KCl, pH glycine buffer were enriched in 5 ml modified Luria –
8.4), 1.5 mM MgCl2, 10 Al sample, and 1.25U of iTaq Bertani broth with acriflavin (mLB) as described by
DNA polymerase (iTaq DNA polymerase, Bio-Rad Kramer et al. (2002). For those experiments where the
Laboratories, Hercules, CA). Samples were run on a cells from the waveguides were enriched in mLB,
Bio-Rad iCycler with the following conditions: 95 jC samples were cultured for 6 h at 42 jC and centri-
for 3 min; 40 cycles of 94 jC for 30 s, 59 jC for 30 s, fuged at 450 Â g for 20 min in a IEC clinical cen-
and 72 jC for 45 s; and extension at 72 jC for 10 min. trifuge (International Equipment Company, Needham
PCR products were run on a 1% agarose gel at 80 V, Hts., MA). The supernatant fluid was then decanted,
T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147 143
Fig. 1. Procedure for confirming results after rapid biosensor assay.
and the sample was resuspended in 100 Al of sterile results illustrate the PCR sensitivity in an ideal sample
water. Ten microliters of the resuspended cells were of sterile water.
used as the sample for PCR. For the growth studies,
100 Al of the unconcentrated mLB enrichment broth 3.2. PCR directly from waveguides
were serially diluted and plated on TSA every hour to
determine the number of viable cells at each time We attempted to remove organisms directly to
point. Fig. 1 illustrates the final method for recovering allow a more direct and rapid preparation of cells
a sample containing E. coli O157:H7 from the wave- for PCR confirmation. PCR was initially attempted
guide, confirming its presence by PCR, and testing
cell viability by plating.
3.1. PCR sensitivity in water
Sterile water was seeded with concentrations of
E. coli O157:H7 and a 10 Al sample was used for the
PCR reaction as described. The lowest number of
cells that could be visualized after the PCR from water
seeded with E. coli O157:H7 was 6.9 Â 104 CFU/ml,
which was equivalent to 6.9 Â 102 total cells per PCR
reaction mixture (Fig. 2). Strong bands were observed
with a positive control of 10 ng total purified E. coli Fig. 2. PCR sensitivity in water. Lane 1, 100 bp marker; lane 2,
6.9 Â 104 CFU per reaction; lane 3, 6.9 Â 103 CFU per reaction;
O157:H7 DNA. This experiment allowed comparison lane 4, 6.9 Â 102 CFU per reaction; lane 5, 69 CFU per reaction;
between ideal PCR conditions and those conditions lane 6, 6.9 CFU per reaction; lane 7, positive control, 10 ng purified
present after capture, and after enrichment. These DNA per reaction; lane 8, negative control, no DNA template.
144 T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147
directly from the waveguide as a more rapid method
of recovery. Small pieces of an unused waveguide
were placed directly into the PCR mixture containing
10 ng total purified E. coli O157:H7 DNA. Amplified
product was present in the samples that contained
DNA template only in the reaction mixture, but not in
the samples containing DNA template and waveguide.
Other attempts to recover organisms from the wave-
guide such as boiling were also unsuccessful (data not
shown). The PCR was inhibited when the sterile water
used for the positive control samples was boiled for
10 min with an unused waveguide (data not shown). It
appears that heating of the waveguide releases sub-
stances that are inhibitory to PCR.
Glycine buffer (0.5 M glycine, 0.5 M NaCl, pH Fig. 3. Inhibition of PCR with glycine buffer. Lane 1, 100 bp marker;
3.2) was also used in an attempt to directly recover lane 2, 1.7 Â 104 CFU per reaction; lane 3, 1.7 Â 103 CFU per
organism for PCR. Glycine buffer at a low pH has reaction; lane 4, 1.7 Â 102 CFU per reaction; lane 5, positive control;
previously been used to dissociate antigen/antibody lane 6, 1.7 Â 104 CFU per reaction; lane 7, 1.7 Â 103 CFU per
reaction; lane 8, 1.7 Â 102 CFU per reaction; lane 9, positive control;
interactions (Suzuki et al., 2002). After cells were lane 10, 1.7 Â 104 CFU per reaction; lane 11, 1.7 Â 103 CFU per
captured on the biosensor waveguides by immobilized reaction; lane 12, 1.7 Â 102 CFU per reaction. Lanes 2 – 4 are in
antibodies, glycine buffer was added to dissociate the glycine buffer minus 0.5 M NaCl; lanes 6 – 8 are in glycine buffer
cells. Viable and direct counts were performed on the with 1 Â PCR buffer minus KCl; lanes 10 – 12 are run in water.
buffer suspension to measure the removal of cells
from the waveguide. Recovery of E. coli O157:H7 with various concentrations of E. coli O157:H7, and
from the waveguide with glycine buffers of varying the lower limit of detection of PCR was determined.
molarities of glycine and NaCl and varying pH was Five milliliters of mLB were seeded with various
tested. Approximately 0.02% of cells injected into the concentrations of E. coli O157:H7 and centrifuged
biosensor were recovered on TSA for each different for 20 min at 450 Â g. The pellet was resuspended in
buffer formulation. Only slightly more cells were 100 Al of sterile water, and 10 Al were removed and
observed when cells were enumerated by direct count. used as the PCR template. Detection was achieved
In order to determine the inhibition of PCR by the when the mLB was seeded with 2.9 Â 103 CFU/ml,
different glycine buffer components, cells were resus- which was 1.5 Â 104 CFU per PCR reaction mixture
pended in glycine buffer, and PCR was run using PCR (Fig. 4). Therefore, the mLB with acriflavin did not
buffer with no KCl. Samples were also resuspended in totally inhibit the PCR reaction, and reasonable sensi-
glycine buffer without 0.5 M NaCl, and PCR was run tivity was obtained after the centrifugation and resus-
with unchanged PCR buffer. Fig. 3 shows the inhib- pension step that removed inhibitors such as acriflavin
ition in both cases, suggesting that samples were (Scheu et al., 1998). Centrifugation also concentrated
inhibited in these experiments by the presence of the the cells so that more cells could be included in the
glycine or the low pH of the buffer. This inhibition, smaller volume required for the PCR reaction.
coupled with the poor recovery rates of glycine-
dissociated cells, led to enrichment as the preferred 3.4. Enrichment times from waveguides
method of recovery.
For enrichment studies, one ml of E. coli O157:H7
3.3. PCR amplification of E. coli O157:H7 in mLB at a concentration of at least 103 CFU/ml was captured
broth with the biosensor waveguide, treated with glycine
buffer, and added to the enrichment broth. A detection
In order to determine if the enrichment media would step was not performed prior to recovery and enrich-
inhibit PCR, mLB broth with acriflavin was spiked ment as detection has been documented to have no
T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147 145
Fig. 4. PCR sensitivity in enrichment media after centrifugation and Fig. 5. PCR after recovery and enrichment. Lane 1, 100 bp marker;
resuspension. Lane 1, 100 bp marker; lane 2, 1.5 Â 102 CFU per lane 2, negative control PBS; lane 3, 6.5 Â 102 CFU/ml injected into
reaction; lane 3, 1.5 Â 103 CFU per reaction; lane 4, 1.5 Â 104 CFU biosensor; lane 4, 6.5 Â 103 CFU/ml injected into biosensor; lane 5,
per reaction; lane 5, 1.5 Â 105 CFU per reaction; lane 6, 1.5 Â 106 6.5 Â 104 CFU/ml injected into biosensor; lane 6, positive control,
CFU per reaction; lane 7, positive control, 10 ng purified DNA per 10 ng purified DNA added to sample from lane 2.
reaction; lane 8, negative control, no DNA template.
time required for detection of E. coli O157:H7 by
effect on recovery (Kramer et al., 2002). Viable counts PCR. Samples of PBS containing varying starting
were performed at 1-h intervals. In order to determine concentrations of E. coli O157:H7 were injected and
the enrichment time needed to reach the lower limit of captured on the biosensor waveguides. These wave-
at least 2.9 Â 103 CFU/ml for PCR detection from guides were then treated with glycine buffer, enriched
enriched cultures, it was necessary to enrich the in mLB, and confirmed with PCR. Fig. 5 illustrates a
bacteria in mLB broth at 42 jC for 5 – 6 h after representative experiment in which 1 ml samples of
capture on the biosensor waveguide. Concentrations E. coli O157:H7 were resuspended in PBS at concen-
of E. coli O157:H7 after enrichment at various times trations of 0 CFU/ml (PBS buffer), 6.5 Â 10 2 ,
and with different concentrations injected into the 6.5 Â 103, or 6.5 Â 104 CFU/ml and were injected
biosensor are shown in Table 1. into the biosensor. Positive samples were confirmed
by PCR at the 6.5 Â 103 and 6.5 Â 104 CFU/ml
3.5. Final results of capture, enrichment, and PCR concentrations. The inability to confirm a positive
from different starting concentrations sample at the 6.5 Â 102 CFU/ml concentration is
likely due to the inability to recover viable organisms
Finally, all steps were performed in sequence to at this concentration and is consistent with our pre-
confirm the starting concentration and enrichment viously reported data (Kramer et al., 2002).
Concentration of E. coli O157:H7 in enrichment culture after 4. Discussion
recovery from biosensor waveguides
Biosensor Concentration after enrichment (CFU/ml) E. coli O157:H7 is an important food pathogen that
capture 1h 2h 3h 4h 5h 6h
can cause serious human illness. Rapid detection and
3 2 2
identification of the pathogen allows physicians to
4.1 Â 10 0 0 0 4.0 Â 10 7.3 Â 10 1.4 Â 104
begin appropriate treatment in a more timely fashion
4.1 Â 104 0 0 60 1.3 Â 103 1.1 Â 104 1.2 Â 105 and prevents costly recalls and potential illness from
CFU/ml food products. However, new rapid methods such as
Biosensor capture is the concentration of bacteria injected into the real-time PCR do not allow for rapid follow-up
biosensor. confirmatory tests. In fact, some diagnostic tools
146 T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147
allow for no confirmation at all without splitting or of boiling or glycine treatment were not as high as for
diluting the sample. For example, after real-time PCR, enrichment. These methods were also found to inhibit
there is no whole cell left for either culture confirma- the PCR reaction. In addition, whereas neither the
tion or immunological testing. The biosensor has been biosensor nor PCR alone can determine viability,
shown to be an efficient method to detect low levels of viability can be determined with the enrichment
pathogens from complex matrices (DeMarco and Lim, culture step. We found that the PCR is not sensitive
2001; DeMarco et al., 1999; DeMarco and Lim, enough to detect E. coli O157:H7 directly from the
2002). However, even the most reliable tests require waveguide when present at low concentrations. There-
some type of confirmation. Therefore, we have devel- fore, at these low initial concentrations, the PCR
oped culture techniques for recovery of live organisms product will be present only when viable cells that
for traditional confirmational tests (Kramer et al., can be enriched from the waveguide are in the sample.
2002). Culture also allows for confirmation of viability by
Because the biosensor relies on immunological traditional plating methods as previously demonstra-
techniques, the potential for cross-reaction with epit- ted (Kramer et al., 2002). In addition, it provides a
opes from other organisms exists. In fact, immuno- viable culture for performance of other tests such as
logical cross reaction between E. coli O157:H7 and antimicrobial susceptibility or for epidemiological and
some strains of Citrobacter freundii and Escherichia criminal investigations. The importance of a viable
hermanii, among others, has been documented with culture, even in the context of real time PCR, has been
other immunological procedures (Bettelheim et al., documented in the literature (Bell et al., 2002).
1993; Perry et al., 1986). PCR detection relies on a PCR amplification of recovered cells in this study
specific DNA sequence, and primers to the eaeA gene indicates the ability to perform additional molecular
are specific for E. coli O157:H7 and O55:H7 (Louie et techniques after a rapid biosensor detection and re-
al., 1994). Therefore, there is only a small chance of covery. In these experiments, PCR was used as a
cross-reactivity in any PCR-based test. Thus, by using molecular confirmatory step. However, other techni-
two rapid tests (immunological and PCR) that are ques such as ribotyping, pulsed field gel electropho-
based on different biological interactions, confirma- resis, or sequencing could be performed to compare
tion of the specific organism is assured. Because both the sample strain to a known database to help deter-
the detection step and the confirmatory steps are rapid, mine the source of the organism. Thus, the results of
the organism can be tested and identification can be this study indicate that it is possible to detect and
confirmed in one working day. identify a target microorganism such as E. coli
Many PCR assays depend on immunomagnetic O157:H7, recover the captured microorganism and
separation (IMS) to remove inhibitory substances that culture it in enrichment broth, and confirm the identity
are present in complex matrices. The biosensor assay of the microbe by PCR, all within 10 h.
can be used in a similar manner as a solid-phase
immunoseparation instrument to capture target organ-
isms from complex matrices. However, unlike IMS, References
the biosensor assay adds a rapid detection step as well.
Because organisms can be recovered after the assay Bell, C.A., Uhl, J.R., Hadfield, T.L., David, J.C., Meyer, R.F.,
with inhibitory substances removed, PCR can subse- Smith, T.F., Cockerill III, F.R., 2002. Detection of Bacillus an-
quently be run on recovered captured microorganisms. thracis DNA by LightCycler PCR. J. Clin. Microbiol. 40,
2897 – 2902.
Thus, without much additional time, the biosensor Bettelheim, K.A., Evangelids, H., Pearce, J.I., Sowers, E., Strock-
assay can perform the same function as IMS but with bine, N.A., 1993. Isolation of a Citrobacter freundii strain which
an added preliminary detection step. carries the Escherichia coli O157, antigen. J. Clin. Microbiol.
We found that recovery of E. coli O157:H7 from 31, 760 – 761.
the biosensor waveguide through culture enrichment Chapman, P.A., Siddons, C.A., Cerdan Malo, A.T., Harkin, M.A.,
2000. A one year study of Escherichia coli O157 in raw beef
may be somewhat slower than the other methods that and lamb products. Epidemiol. Infect. 124, 207 – 213.
were tested, but it provided the greatest sensitivity. Chapman, P.A., Ellin, M., Ashton, R., Shafique, W., 2001. Com-
Recovery rates for the more direct recovery methods parison of culture, PCR and immunoassays for detecting Es-
T.B. Tims, D.V. Lim / Journal of Microbiological Methods 55 (2003) 141–147 147
cherichia coli O157 following enrichment culture and immuno- coli using serotype-specific primers. Epidemiol. Infect. 112,
magnetic separation performed on naturally contaminated raw 449 – 461.
meat products. Int. J. Food Microbiol. 68, 11 – 20. Manafi, M., Kremsmaier, B., 2001. Comparative evaluation of dif-
DeCludt, B., Bouvet, P., Mariani-Kurkdjian, P., Grimont, F., Gri- ferent chromogenic/fluorogenic media for detecting Escherichia
mont, P.A.D., Hubert, B., Loirat, C., 2000. Haemolytic uraemic coli O157:H7 in food. Int. J. Food Microbiol. 71, 257 – 262.
syndrome and Shiga toxin-producing Escherichia coli infection Perry, M.B., MacLean, L., Griffith, D.W., 1986. Structure of the O-
in children in France. Epidemiol. Infect. 124, 215 – 220. chain polysaccharide of the phenol-phase soluble lipopolysac-
DeMarco, D.R., Lim, D.V., 2001. Direct detection of Escherichia charide of Escherichia coli O157:H7. Biochem. Cell. Biol. 4,
coli O157:H7 in unpasteurized apple juice with an evanescent 21 – 28.
wave biosensor. J. Rapid Methods Autom. Microbiol. 9, Pebody, R.G., Furtado, C., Rojas, A., McCarthy, N., Nylen, G.,
241 – 257. Ruutu, P., Leino, T., Chalmers, R., de Jong, B., Donnelly, M.,
DeMarco, D.R., Lim, D.V., 2002. Detection of Escherichia coli Fisher, I., Gilham, C., Graverson, L., Cheasty, T., Willshaw, G.,
O157:H7 in 10- and 25-gram ground beef samples with an Navarro, M., Salmon, R., Leinikki, P., Wall, P., Bartlett, C.,
evanescent wave biosensor with silica and polystyrene wave- 1999. An international outbreak of Vero cytotoxin-producing
guides. J. Food Prot. 65, 596 – 602. Escherichia coli O157 infection amongst tourists; a challenge
DeMarco, D.R., Saaski, E.W., McCrae, D.A., Lim, D.V., 1999. for the European infectious disease surveillance network. Epi-
Rapid detection of Escherichia coli O157:H7 in ground beef demiol. Infect. 123, 217 – 223.
using a fiber-optic biosensor. J. Food Prot. 62, 711 – 716. Scheu, P.M., Berghof, K., Stahl, U., 1998. Detection of pathogenic
Kaper, J.B., 1998. Enterohemorrhagic Escherichia coli. Curr. Opin. and spoilage microorganisms in food with the polymerase chain
Microbiol. 1, 103 – 108. reaction. Food Microbiol. 15, 13 – 31.
Kramer, M.F., Tims, T.B., DeMarco, D.R., Lim, D.V., 2002. Re- Suzuki, M., Ozawa, F., Sugimoto, W., Aso, S., 2002. Miniature
covery of Escherichia coli O157:H7 from fiber optic wave- surface-plasmon resonance immunosensors—rapid and repeti-
guides used for rapid biosensor detection. J. Rapid Methods tive procedure. Anal. Bioanal. Chem. 372, 301 – 304.
Autom. Microbiol. 10, 93 – 106. Uyttendaele, M., van Boxstael, S., Debevere, J., 1999. PCR assay
Louie, M., de Azavedo, J., Clarke, R., Borczyk, A., Lior, H., for detection of the E. coli O157:H7 eae-gene and effect on the
Richter, M., Brunton, J., 1994. Sequence heterogeneity of sample preparation method on PCR detection of heat-killed E.
the eae gene and detection of verotoxin-producing Escherichia coli O157:H7 in ground beef. Int. J. Food Microbiol. 52, 85 – 95.