International Journal of Food Microbiology 170 (2014) 48–54

Contents lists available at ScienceDirect

International Journal of Food Microbiology

j ourna l homepage: www.e lsev ie r .com/ locate / i j foodmicro

Detection of viable Escherichia coli O157:H7 in ground beef by propidiummonoazide real-time PCR

Yarui Liu, Azlin Mustapha ⁎Food Science Program, Division of Food Systems and Bioengineering, University of Missouri, Columbia, United States

⁎ Corresponding author at: Food Science Program,University of Missouri, Columbia, MO 65211, United Sfax: +1 573 884 7964.

E-mail address: MustaphaA@missouri.edu (A. Mustap

0168-1605/$ – see front matter © 2013 Elsevier B.V. All rhttp://dx.doi.org/10.1016/j.ijfoodmicro.2013.10.026

a b s t r a c t

a r t i c l e i n f o

Article history:Received 1 July 2013Received in revised form 6 October 2013Accepted 27 October 2013Available online 6 November 2013

Keywords:PropidiummonoazideReal-time PCRViable Escherichia coli O157:H7Ground beef

Escherichia coli O157:H7 associated with food has caused many serious public health problems in recent years.However, only viable cells of this pathogen can cause infections, and false-positive detection caused by deadcells can lead to unnecessary product recalls. The objective of this study was to develop and optimize a methodthat combines propidiummonoazide (PMA) stainingwith real-time PCR to detect only viable cells of E. coliO157:H7 in ground beef. PMA is a DNA intercalating dye that can penetrate compromisedmembranes of dead cells andbind to cellular DNA, preventing its amplification via a subsequent PCR. Three strains of E. coli O157:H7 (505B,G5310 and C7927) at concentrations of 100 to 108 CFU/mL were used as live cells. Dead cells were obtained byheating cell suspensions at 85 °C for 15 min. Suspensions were treated with PMA and the optimized assay wasapplied to artificially contaminated ground beef with two different fat contents (10% and 27%). DNAwas extract-ed and amplified by TaqMan® real-time PCR assay targeting the uidA gene for detection of E. coli O157:H7. Plas-mid pUC19was added as an internal amplification control (IAC). A treatment of 25 μM PMAwith a 10-min lightexposure on ice was sufficient to eliminate DNA from 108 dead E. coli O157:H7 cells/mL. The optimized assaycould detect as low as 102 CFU/mL viable E. coli O157:H7 in pure culture and 105 CFU/g in ground beef, in thepresence of 106/mL or g of dead cells. With an 8-h enrichment, 1 CFU/g viable E. coli O157:H7 in ground beefwas detectablewithout interference from106 dead cells/g. In conclusion, the PMA real-timePCR could effectivelydetect viable E. coli O157:H7 without being compromised by dead cells.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Escherichia coli O157:H7 is one of the most notorious foodbornepathogens, with an infectious dose of as low as a few hundred cells(Karmali, 2004). Beef and dairy products, juices and fresh produce arefoods that are often associated with E. coli O157:H7 outbreaks. E. coliO157:H7 infections can lead to nonspecific diarrhea, hemorrhagic colitisand even hemolytic uremic syndrome (HUS) (Banatvala et al., 2001).According to the Centers for Disease Control and Prevention (CDC),E. coli O157:H7 leads to an estimated 73,480 illnesses each year,resulting in more than 2000 hospitalizations and 60 deaths (Meadet al., 1999). It has been estimated that the annual cost of E. coli O157:H7 infections was $405 million from 1996 to 2004 (Frenzen et al.,2005). Thus, accurate and sensitive methods to detect E. coli O157:H7in food products are urgently needed.

Conventional culture-based methods, involving enrichment, isola-tion and confirmation steps, have been used for over a century due totheir sensitivity, low cost, ease of use, and ability tomonitor cell viability(Murakami, 2012). However, they require four to five days to achieve

256 WCS Wing, Eckles Hall,tates. Tel.: +1 573 882 2649;

ha).

ights reserved.

confirmatory results. On the other hand, the polymerase chain reaction(PCR), a widely used nucleic acid-based technique, can identify targetspecies within 3 h. However, conventional PCR cannot differentiate via-ble cells from dead cells (Wang and Levin, 2006). DNA from dead cellscan lead to false-positive PCR results, leading to unnecessary product re-calls and economic losses.

Membrane integrity is considered the most important criterion fordistinguishing between viable and irreversibly damaged cells (Nockeret al., 2006). Propidium monoazide (PMA), a DNA intercalating dye,can only penetrate dead or membrane-compromised cells and cova-lently bind to cellular DNA through photolysis. Consequently, the cova-lent link will render the DNA insoluble and inhibit PCR amplification ofDNA from dead or membrane-compromised cells (Nocker et al., 2006).PMA is less likely to penetrate viable but nonculturable (VBNC) cellswith intact membranes. Xiao et al. (2013) combined PMA stainingwith real-time PCR and successfully detected as low as 100 CFU/mL ofE. coli O157:H7 in a VBNC state with this assay. PMA has been success-fully used in combination with real-time PCR to detect viable lacticacid bacteria (García-Cayuela et al., 2009), T4 phage (Fittipaldi et al.,2010), Campylobacter (Josefsen et al., 2010), Salmonella (Liang et al.,2011) and E. coli (Yang et al., 2011; Taskin et al., 2011) in environmentaland food samples. Ethidiummonoazide (EMA) is another dye that cova-lently binds to cellular DNA. Both PMA and EMA have specific advan-tages and disadvantages. Compared with PMA, EMA penetrates dead

49Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

cellsmore effectively but exerts a greater inhibitory effect on viable cells(Lee and Levin, 2006). EMA, at high concentrations, penetrates viablecells and leads to false-negative PCR results (Nocker et al., 2006).Wang et al. (2009) developed an EMA real-time PCR assay to detect vi-able E. coliO157:H7 cells in ground beef. According to their results, EMA,at a concentration of 100 μg/mL in the cell suspension (approximately120 μM), completely prevented DNA amplification from 108 CFU viableE. coli O157:H7 cells via real-time PCR, while a 12-h enrichment stepwas necessary to detect 101 CFU/g of viable E. coli O157:H7 in spikedground beef samples.

Real-time PCR is a rapid and sensitive method that can be used forthe quantification of microorganisms of interest. However, substancesnaturally found in environmental and food samplesmay inhibit the am-plification of target DNA in real-time PCR. To solve this problem, an in-ternal amplification control (IAC) is included in a real-time PCR assay tomonitor the PCR's efficiency and to prevent false-negative results due toPCR inhibitors present in food samples (Murphy et al., 2007). The objec-tives of this study were to optimize a PMA staining procedure, and tocombine it with real-time PCR for the detection and quantification ofonly viable E. coli O157:H7 cells in contaminated ground beef samples.

2. Materials and methods

2.1. Preparation of viable and dead E. coli O157:H7 cells

E. coli O157:H7 strains G5310, C7927, and 505B were from ourown culture collection at the Food Microbiology Laboratory, Universityof Missouri, Columbia, MO. Each of the three strains was grown intryptic soy broth supplemented with 0.5% yeast extract (TSBY; DifcoLabs., BD Diagnostic Systems, Sparks, MD, USA) at 37 °C overnight(~109 CFU/mL). One milliliter of each strain was then mixed together,cells were harvested by centrifugation at 13,400 ×g for 5 min, andwashed and serially diluted in 0.1% peptone water to yield cell suspen-sions ranging from 100 to 108 CFU/mL. To obtain dead cells, the cell sus-pensions were heated at 85 °C for 35 min. Cell viability was checked byplating in plate count agar (PCA; Difco Labs.).

2.2. Minimum PMA amount needed to bind dead cell DNA

Propidium monoazide (PMA) was purchased from Biotium Inc.(Hayward, CA, USA), and a 20 mMstock solutionwas prepared in steriledistilled water and stored in the dark at −20 °C. Six dead E. coli O157:H7 samples (108 cells/mL each) were prepared as described above. Dif-ferent amounts of PMA (0, 0.5, 1.25, 2.5, 5, 10 μL)were added to 1 mL ofeach sample to achieve PMA concentrations of 0, 10, 25, 50, 100, and200 μM, respectively. Following a 5-min incubation period in the darkat room temperature on a Labquake Rotisserie (Barnstead International,Dubuque, IA, USA), samples were exposed to a 650-W halogen light for10 min. The samples were placed on ice at a distance of 20 cm from thelight source to avoid excessive heating. After the photo-induced cross-linking, cells were collected by centrifugation at 13,400 ×g for 5 min,andwashed in sterile distilledwater under the same centrifugation con-ditions prior to DNA extraction.

2.3. Influence of PMA concentration on amplification of DNA from viablecells

Fresh viable E. coliO157:H7 cells (108 CFU/mL)were prepared as de-scribed previously. Varying amounts of PMA (0, 10, 25, 50, 100, and200 μM) were added to 1 mL of each cell suspension. PMA stainingwas conducted as described earlier (5 min in the dark followed by a10-min light exposure). Cell pellets were collected by centrifugation at13,400 ×g for 5 min and washed with sterile distilled water under thesame centrifugation conditions. After DNA extraction, the influence ofPMA concentration on DNA amplification from viable cells was evaluat-ed by real-time PCR.

2.4. Influence of light exposure time on DNA amplification from viable cells

Six viable E. coli O157:H7 samples (108 CFU/mL each) were pre-pared as described previously. PMA stock solution in the amount of1.25 μLwas added to each 1 mLof cell suspension to achieve afinal con-centration of 25 μM. Samples were incubated in the dark at room tem-perature for 5 min and exposed to a 650-halogen light for varyinglengths of time (0, 2, 5, 10, 15, 20 min). After PMA staining, cell pelletswere collected by centrifugation at 13,400 ×g for 5 min and washedwith sterile distilled water. The influence of light exposure time onDNA amplification from viable cells was evaluated by real-time PCRafter DNA extraction.

2.5. DNA extraction

Cell pellets were resuspended in 100 μL of PrepMan® Ultra SamplePreparationReagent (Applied Biosystems, Foster City, CA, USA), vortexedfor 10 s and boiled for 25 min. Boiled cell suspension was centrifuged at13,400 ×g for 3 min and 5 μL of the supernatant was used as templateDNA for the real-time PCR assay.

2.6. Real-time PCR

Primers and probes targeting E. coli O157:H7 and pUC 19were usedas previously described byWang et al. (2007, 2009) withminormodifi-cations. The sequence of E. coli O157:H7 primer 1 is 5′-TTGACCCACACTTTGCCGTAA-3′, and that of E. coli O157:H7 primer 2 is 5′-GCGAAAACTGTGGAATTGGG-3′. The sequence of E. coli O157:H7 probe is 5′-5HEX-TGACCGCATCGAAACGCAGCT-3BHQ_1-3′. pUC 19 was used asan IAC. The sequence of the IAC primer 1 is 5′-GCAGCCACTGGTAACAGGAT-3′ and that of the IAC primer 2 is 5′-GCAGAGCGCAGATACCAAAT-3′, and the sequence of the IAC probe is 5′-56FAM-AGAGCGAGGTATGTATGTAGGCGG-3BHQ_1-3′ (Fricker et al., 2007).

A 7500 real-time PCR system (Applied Biosystems) was used. A PCRreaction of 50 μL contained 25 μL of 2× TaqMan™Universal PCRMasterMix (Applied Biosystems), 0.5 μM of each E. coli O157:H7 primer,0.4 μM of each IAC primer, 0.2 μM of E. coli and IAC probe, 0.25 pg ofpUC19 (8.62 × 104 copies; Promega, Madison, WI, USA), and 3 μLDNA extract. Nuclease-free water (Promega) was used to adjust the re-action volume to 50 μL. The real-time PCR program consisted of 50 °Cfor 2 min and 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 sand 60 °C for 1 min.

2.7. Application of PMA real-time PCR to viable E. coli O157:H7 cells

Onemilliliter of viable E. coliO157:H7 cells (101 to 108 CFU/mL)wastreated or untreated with PMA. For PMA staining, 1.25 μL of PMA stocksolution was added to each sample to achieve a final concentration of25 μM. Samples were incubated in the dark for 5 min and exposed toa 650-W halogen light for 10 min. DNA extraction and real-time PCRwith IAC were conducted as described earlier. Standard curves wereconstructed by plotting Ct values generated from real-time PCR againstE. coli O157:H7 cell concentrations (log CFU/mL).

2.8. Application of PMA real-time PCR to mixed viable and dead E. coliO157:H7 cells

One milliliter of dead E. coli O157:H7 cells (106/mL) was mixedwith 1 mL of fresh viable cells with different concentrations (101

to 108 CFU/mL) and centrifuged at 13,400 ×g for 5 min. The mixedcell pellets were resuspended in 1 mL of 0.1% peptone water and un-treated or treated with PMA, as was optimized previously. DNA ex-traction and real-time PCR with IAC were conducted as describedearlier. Real-time PCR with IAC was conducted to check results.

UD UD UD UD UD0

5

10

15

20

25

30

35

40

0 10 25 50 100 200 IAC

Ct

PMA concentration (µM)

E. coli O157

IAC

Fig. 1. Minimum PMA concentration necessary for binding DNA from 108 cells/mL ofdead E. coli O157:H7. Results were from two repeated experiments. UD, undetectedafter 40 cycles in real-time PCR.

0

5

10

15

20

25

30

35

40

0 10 25 50 100 200 IAC

Ct

PMA concentration (µM)

E. coli O157

IAC

Fig. 2. Influence of PMA concentration on amplification of DNA from 108 CFU/mL of viableE. coli O157:H7 cells. Results were from two repeated experiments. UD, undetected after40 cycles in real-time PCR.

50 Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

2.9. Application of PMA real-time PCR to artificially contaminated groundbeef

Ground beef with two different fat contents (10% and 27%) was pur-chased from a local grocery store. It was determined to be free of E. coliO157:H7 using standard culturemethods (FDA, 1995) and the real-timePCR method used in this study. Twenty-five grams of each ground beefsample was placed in sterile stomacher bags containing filter mem-branes (Filtra-Bag®,VWR International, Edmonton, AB, Canada). Twosets of samples were prepared. The first set of nine samples was pre-pared by adding only viable E. coli O157:H7 cells at different concentra-tions (100 to 108 CFU/g) into each sample. Considering it unlikely thatas high as 108 or 107/mL dead cells of E. coli O157:H7 would be presentin food products in reality, the second set of nine samples was preparedby adding 106/g dead E. coliO157:H7 cells and various concentrations ofviable E. coli O157:H7 cells (100 to 108 CFU/g) into each 25 g of groundbeef sample. The spiked beef samples were massaged by hands for2 min to allow for adequate distribution and attachment of cells to themeat. Samples were individually mixed with 225 mL TSBY and homog-enized for 2 min by stomaching. Onemilliliter of homogenized suspen-sion from each sample was removed for DNA extractionwith prior PMAstaining, as was optimized earlier. Real-time PCR was conducted as de-scribed previously.

2.10. Detection of low concentrations of viable E. coli O157:H7 in groundbeef samples with enrichment by PMA real-time PCR

Twenty-five grams of each ground beef sample was inoculated with106/g of dead E. coliO157:H7 cells, as well as viable E. coli O157:H7 cellsat concentrations of 100, 101, 102, 103 and 104 CFU/g. Each sample wasadded to 225 mL TSBY and homogenized for 2 min by stomaching.Samples were incubated in a shaker incubator set at approximately200 rpm at a temperature of 37 °C. Ten milliliters of each sample werecollected at different enrichment times (0, 2, 4, 6, and 8 h). The collectedsuspensions were first centrifuged at 2000 ×g for 2 min to precipitatethemeat tissues and fat. Then, cell pelletswere collected by centrifugingat 8000 ×g for 15 min. Cell pellets were washed and resuspended in1 mL of 0.1% peptone water. PMA staining, DNA extraction and real-time PCR were conducted as described previously.

2.11. Statistical data analysis

The SAS GLM procedure (SAS 9.2, Copyright 2002–2007; SAS Insti-tute Inc., Cary, NC, USA) was used to evaluate the effect of PMA stainingon dead and viable cells. Tukey's test was applied to determine differ-ences between various PMA staining and light exposure treatments. Dif-ferences were compared at a significance level of 0.05.

3. Results and discussion

3.1. Minimum PMA amount needed to bind dead cell DNA

As shown in Fig. 1, DNA from dead cells that were not stained withPMA generated false-positive results in real-time PCR with a Ct valueof 18.82, confirming the fact that DNA persists after cell death and PCRcannot differentiate live fromdead cells. By staining dead cells with var-ious concentrations of PMA, it was shown that a minimum of 25 μMPMA was able to completely inhibit real-time PCR amplification ofDNA from 108 cells/mL of dead E. coli O157:H7 (Fig. 1). The amplifica-tion of the IAC (pUC 19) in each reaction successfully monitored the ef-ficiency of the PCR reaction.

According to Wang et al. (2009), a treatment of 12 μM EMA with a10-min light exposure on ice was sufficient to eliminate DNA from108/mL dead E. coli O157:H7 cells. In this study, a treatment of 25 μMPMAwith the same light exposure conditionswas found to be optimum

for binding DNA obtained from the same concentration of dead E. coliO157:H7 cells and preventing their subsequent PCR amplification.

3.2. Influence of PMA concentration on amplification of DNA from viablecells

Fig. 2 shows that when the PMA concentration was≤50 μM, no sig-nificant differenceswere foundbetween Ct values generated fromPMA-treated and those from PMA-untreated viable cells (P N 0.05). However,when the PMA concentration was ≥100 μM, the Ct values generatedfrom 108 CFU/mL of PMA-treated viable cells were significantly higherthan those generated from PMA-untreated cells (P ≤ 0.05). The resultsindicated that high concentrations of PMA (≥100 μM) could also inhibitthe amplification of DNA from viable cells, causing an underestimationof viable cells in real-time PCR. Data from Figs. 1 and 2 confirmed thata PMA concentration of 25 μM could completely inhibit PCR amplifica-tion of DNA from dead cells, without influencing DNA amplificationfrom viable cells. Thus, 25 μMof PMAwas decided to be used for furtherstudies.

EMA is another DNA binding dye which has also shown to havesimilar inhibitory effects at higher concentrations. Wang et al. (2009)reported that the higher the EMA concentration, the greater its inhibi-tion of DNA amplification from viable E. coli O157:H7. EMA, at aconcentration higher than 120 μM in combination with a 10-min

20

25

30

35

40

Ct

51Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

light exposure, completely prevented DNA amplification from108 CFU/mL viable E. coli O157:H7 cells, generating false-negativeresults. However in this study, unlike EMA, higher concentrations(≥100 μM) of PMA only lead to an underestimation of viable E. coliO157:H7 cells while positive results could still be obtained in real-time PCR. By comparing these results, PMA was found to be less toxicto viable E. coli O157:H7 cells than similar concentrations of EMA.

y = -3.6488x + 47.518R² = 0.9473

y = -3.5991x + 48.825R² = 0.9635

0

5

10

15

0 2 4 6 8 10Cell count (Log CFU/mL)

a. without PMA

b. with PMA

Fig. 4. Standard curves, without PMA (a) and with PMA (b) for detection of viable E. coliO157:H7 cells, in the absence of dead cells, by real-time PCR. Results are from two repeat-ed experiments.

3.3. Influence of light exposure time on DNA amplification from viable cells

Fig. 3 shows that the length of light exposure time did not signifi-cantly influence the Ct values generated from 108 CFU/mL viable E. coliO157:H7 cells (P N 0.05). The light exposure step is necessary to acti-vate the PMA that is bound to DNA from dead cells and to inactivateany excess PMA that have not entered cells (Fittipaldi et al., 2012). Pre-viously published studies on PMA staining varied on the optimal lightexposure time for DNA amplification. This difference might be becauseof the different light sources and bacterial species used in different stud-ies. In this study, a light exposure time of 10 min was found to be effec-tive for cells treated with 25 μM of PMA to completely prevent DNAamplification from 108/mL dead E. coli O157:H7 cells (Fig. 1). Fig. 3demonstrates that viable cells treated with a 10-min light exposureshowed a Ct value close to that of untreated controls (P N 0.05). There-fore, a 10-min light exposure was used for further development of theassay.

3.4. Application of PMA real-time PCR to viable E. coli O157:H7 cells

Without PMA treatment, the detection range of the real-time PCRassay was from 102 to 108 CFU/mL viable E. coli O157:H7 cells (Fig. 4).With PMA staining, the lowest detection limit of 102 CFU/mL did notchange. However, a slight increase (approximate 1.5 cycles) in the Ctvalues from PMA-treated samples was observed. The increase in Ctvalues might be because a small portion of E. coli O157:H7 cells is lostduring the washing step after PMA treatment (Liang et al., 2011). An-other possible reason was that there could be a small portion of deadcells present in the overnight culture used in this experiment. The com-promised membranes of dead cells in the overnight culture were pene-trated by PMA and the amplification of their DNA was then inhibited inthe following PCR reactions, resulting in larger Ct values of PMA-treatedsamples. The PMA treatment, as optimized previously, had no adverseeffect on detection of viable E. coli O157:H7 by real-time PCR. To com-pare PMA with EMA, another set of viable cells was treated by EMA fol-lowingWang et al. (2009) prior to real-time PCR. It turned out that thedetection limit of viable E. coli O157:H7 in pure culture using EMA real-

UD0

5

10

15

20

25

30

35

40

0 2 5 10 15 20 IAC

Ct

Light exposure time (min)

E. coli O157

IAC

Fig. 3. Influence of light exposure time on amplification of DNA from108 CFU/mL of viableE. coli O157:H7 cells. Results were from two repeated experiments. UD, undetected.

time PCR assay was also 102 CFU/mL, which was the same as that ofPMA real-time PCR (data not shown).

3.5. Application of PMA real-time PCR to mixed viable and dead E. coliO157:H7 cells

Without PMA treatment, the real-time PCRwas unable to differenti-ate between DNA from viable E. coli O157:H7 cells and that from deadcells (Fig. 5). Because DNA from both viable and dead cells served astemplates for real-time PCR, the number of viable cells was significantlyoverestimated. For example, when 101 CFU/mL of viable cells werepresent together with 106/mL of dead cells, the real-time PCR withoutPMA treatment generated a Ct value of 25.8, whereas the PMA real-time PCR assay could not detect any E. coli O157:H7 (neither viablenor dead) in the mixture (Fig. 5). The Ct value obtained abovecorresponded to the Ct value yielded from 106 CFU/mL of viable cellsby using real-time PCR without PMA treatment (Fig. 4). These resultsconfirm that PMA staining prior to DNA extraction can effectively pre-vent false-positive results in real-time PCR that was caused by the pres-ence of DNA from dead cells.

With PMA treatment, this real-time PCR assay could detect a rangeof 102 to 108 CFU/mL of viable E. coliO157:H7 cells even in the presenceof 106/mL dead cells (Fig. 5). Comparing Fig. 5 with Fig. 4, the detectionrange of PMA real-time PCR for viable cells, when both viable and deadcells were present, was the same aswhen only viable cells were present.Although similar slopes of the two standard curves were observed, theCt values from viable and dead cell mixtures (y = −3.2767x + 46.42;R2 = 0.9368) were slightly smaller than that from only viable cells

y = -3.2767x + 46.42R² = 0.9368

0

5

10

15

20

25

30

35

40

0 2 4 6 8 10

Ct

Cell count (Log CFU/mL)

a. without PMA

b. with PMA

Fig. 5. Detection range of E. coli O157:H7 cells without PMA (a) and with PMA (b), whenviable and 106 dead cells/mL are present, by real-time PCR. Results are from two repeatedexperiments.

y = -4.1025x + 55.727R² = 0.9362

y = -4.5275x + 58.147R² = 0.9568

0

5

10

15

20

25

30

35

40

4 5 6 7 8 9

Ct

(a)

a. only viable

b. viable and dead mixture

y = -3.8588x + 54.412R² = 0.9253

y = -3.7335x + 52.407R² = 0.9946

0

5

10

15

20

25

30

35

40

4 5 6 7 8 9

Cell count (Log CFU/g)

(b)

a. only viable

b. viable and dead mixture

Fig. 6. Application of PMA real-time PCR to artificially contaminated 10% ground beef(a) and 27% fat ground beef (b), and its detection range for viable or viable and deadE. coli O157:H7 cells. Results are from two repeated experiments.

52 Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

(y = −3.5991x + 48.825; R2 = 0.9635), especially when the concen-trations of viable cellswere low (≤103 CFU/mL), indicating a slight over-estimation of viable cells in the mixture. For example, the average Ctvalue yielded from 102 CFU/mL viable and 106/mL dead cells was 36.4,whereas that generated from102 CFU/mL viable alonewas 38.5. Howev-er, no Ct valuewas yielded from the control when only 106/mL dead cellswere present (data not shown), which again proved that the PMA treat-ment was sufficient to eliminate DNA from dead cells when only deadcells were present. These indicate that the PMA treatment was not ableto completely remove signals from a high number of dead E. coli O157:H7 cells when both viable and dead cells were present in the sample. In-complete suppression of dead cell signals using PMA real-time PCR hasbeen reported previously (Pan and Breidt, 2007; Kralik et al., 2010;Løvdal et al., 2011). Pan and Breidt (2007) stated that the number of vi-able Listeria monocytogenes was significantly overestimated when theratio of dead cells to live cells exceeded 104 and the concentration oflive cells was less than 103 CFU/mL. We found that the detection limitof viable E. coli H157:H7 in a mixture of viable and dead cells usingEMA real-time PCR, was the same as that of PMA real-time PCR. Howev-er, the Ct values from EMA-treated viable and dead cell mixtures werelarger than those from only viable cells (data not shown), which indi-cates that EMA also penetrates viable cells, thus influencing the Ct valuesgenerated by real-time PCR. Therefore, based on the results, it was con-cluded that PMA overestimates viable E. coli O157:H7 cells in viableand dead cell mixture, whereas EMA underestimates viable cells in themixture.

3.6. Application of PMA real-time PCR to artificially contaminated groundbeef

Both culture-based and real time PCR-based tests confirmed that theground beef was originally free of E. coli O157:H7. The PMA real-timePCR assay was applied to ground beef samples inoculated with only vi-able E. coliO157:H7 from101 to 108 CFU/g. It was also applied to groundbeef samples whichwere simultaneously inoculated with 106/g of deadE. coli O157:H7 cells and different concentrations of viable E. coli O157:H7 cells (100 to 108 CFU/g).

As shown in Fig. 6, the PMA real-time PCR could detect a range from105 to 108 CFU/g of viable E. coli O157:H7, in ground beef samples re-gardless of whether dead cells were present or not. Again, the PMAtreatment was proven to be effective at penetrating dead E. coli O157:H7 cells without being negatively influenced by meat tissues and back-ground microflora. The relatively high detection limit of viable E. coliO157:H7 cells in ground beef (105 cells/g) could be caused by: 1) thehigh counts of background microflora in the ground beef interferingwith the detection of target E. coli O157:H7 cells (the aerobic platecount of the 10% fat content ground beef was 1.7 × 106 CFU/g, andthat of the 27% fat content ground beef was 1.5 × 105 CFU/g); 2) thefood samples which contain many organic and inorganic substances,such as phenolic compounds, fat, enzymes, polysaccharides, proteinsand salts, all ofwhich can either inhibit PCR amplification or lead to a re-duction in amplification efficiency of PCR reactions (Španová et al.,2000); and 3) a small portion of viable cells thatmay not have recoveredafter homogenization by stomaching and the intactmembranes of someviable cells that might have been injured during homogenization andpenetrated by PMA, leading to fewer DNA templates in the followingPCR reactions.

Standard curves of ground beefwith 10% and 27% fat contents exhib-ited very similar linear relationships between Ct values and the concen-trations of viable E. coli O157:H7 in ground beef. Nonetheless, thedetection limit of 105 CFU/g in ground beef was still not satisfactory,since the infectious dose of E. coli O157:H7 is low. Therefore, a pre-enrichment step was needed to increase the number of viable cellsto a detectable level when the initial concentration is low. Yanget al. (2013) reported that 103 CFU/g of viable E. coli O157:H7 inground beef could be detected by using magnetic nanobead-based

immunomagnetic separation (IMS) combined with PMA multiplexPCR (mPCR). Although their IMS–PMA–mPCR assay achieved a betterdetection limit of E. coli O157:H7 in ground beef (103 cells/g) than thisstudy (105 cells/g) did, it was still not sensitive enough to detect thepathogen at its infectious dose. On the other hand, with an 8-h enrich-ment step, the PMA real-time PCR assay described in this study coulddetect as low as 1 CFU/g of viable E. coli O157:H7 in ground beef, whichis even lower than its infectious dose. Moreover, a real-time PCR assayis more sensitive than traditional PCR and can be quantitative for patho-genic bacterial detection in food. By plugging the Ct value of an unknownsample into the standard curve (Fig. 6), the concentration of viable E. coliO157:H7 in the original sample could be estimated. Continuation of thisstudy by combining IMSwith PMA real-time PCRwould be an interestingnext step.

3.7. Detection of low concentrations of viable E. coli O157:H7 in ground beefsamples by PMA real-time PCR with enrichment

To detect low concentrations of viable E. coliO157:H7 cells in groundbeef, enrichment in TSBY was conducted prior to the PMA real-timePCR. A 6-h enrichment was required to detect initial concentrations of102, 103 and 104 CFU/g viable E. coli O157:H7. After an 8-h enrichment,PMA real-time PCR could detect as low as 1 CFU/g of viable E. coli O157:H7 in ground beef (Table 1). The results from 10% fat content and thosefrom 27% fat content ground beef were the same. In addition, it was alsofound that the sample inoculated with only 106/g dead cells generatednegative results during the entire enrichment process using PMA real-time PCR, while the same dead cell samples generated false-positive re-sults using real-time PCR without PMA treatment (Table 1). These re-sults demonstrated that the addition of PMA prior to DNA extractioneffectively eliminated theDNA fromdead E. coliO157:H7 cells in groundbeef. Blank food samples without any artificial inoculation of E. coliO157:H7 were also enriched, showing that there was no E. coli O157:

Table 1Application of PMA real-time PCR to groundbeef (10% and27% fat contents) contaminatedwith low concentrations of viable and 106 dead cells/g E. coli O157:H7. Results are fromtwo repeated experiments.

Enrichmenttime (h)

Concentrationof viable cells(CFU/g)

Concentrationof dead cells(cells/g)

PMA treatment No PMAtreatment

E. coliO157:H7

IAC E. coliO157:H7

IAC

2 h 104 106 − + + +103 106 − + + +102 106 − + + +101 106 − + + +100 106 − + + +0 106 − + + +

4 h 104 106 − + + +103 106 − + + +102 106 − + + +101 106 − + + +100 106 − + + +0 106 − + + +

6 h 104 106 + + + +103 106 + + + +102 106 + + + +101 106 − + + +100 106 − + + +0 106 − + + +

8 h 104 106 + + + +103 106 + + + +102 106 + + + +101 106 + + + +100 106 + + + +0 106 − + + +

53Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

H7 contamination in the original samples. The addition of an IAC inthis study successfully monitored the efficiency of the real-time PCR,preventing the occurrence of false-negative results.

Both viable and dead cells can be present in food products. However,only viable pathogens pose a risk to consumers' health. Therefore, effec-tive methods for the detection of only viable bacterial pathogens areneeded. A pre-enrichment step is commonly included in pathogen de-tection methods, allowing the ratio of live cells to dead cells to increaseand decreasing the influence of DNA from dead cells in a subsequentPCR. However, as shown in this study, enrichment steps alone cannotcompletely prevent the false-positive result generated from 106/gdead E. coli O157:H7 cells after the 8-h enrichment period. With PMAstaining, DNA amplification from dead cells in the subsequent PCR andthe generation of false-positive results yielded fromdead cellswere suc-cessfully prevented. Comparedwith the literature, a shorter enrichmenttime (8 h) and a lower detection limit (1 CFU/g) was achieved in thisstudy. Thismight be because a shaker incubatorwasused for the enrich-ment period in this study, which facilitated faster growth of E. coliO157:H7. Additionally, instead of 1 mL of the enriched cell suspension, 10 mLof sample was used for DNA isolation. This modification resulted in ahigher yield offinal DNAwhich, in turn, facilitated and improved its am-plification in the subsequent real-time PCR. In September 2011, the U.S.Department of Agriculture (USDA) announced that serogroups O26,O45, O103, O111, O121, and O145 of Shiga toxin producing E. coliwould also be considered adulterants when present in raw or non-intact beef products (USDA, 2011). The PMA-real time PCR assay devel-oped in this study shows great promise and could be optimized for de-tection of viable cells of these other pathogenic serogroups in groundbeef in future studies.

4. Conclusions

Overall, a PMA treatment can effectively prevent the amplificationof DNA from dead E. coli O157:H7 cells in ground beef. The PMA real-time PCR assay described in this study can selectively detect only viable

E. coli O157:H7 with a sensitivity of 102 CFU/mL in pure culture, and105 CFU/g in ground beef, in the presence of 106 dead cells/mL or g.With an 8-h enrichment, the assay could detect as low as 1 CFU/g viableE. coli O157:H7 in ground beef. The whole process can easily be com-pleted in about 5 h after an 8-h enrichment step. Hence, PMA real-time PCR assay has the potential to be employed as a rapid, yet selectiveand sensitive, detection method for the food industry to detect onlyviable E. coli O157:H7 in food products.

Acknowledgments

We sincerely thank Dr. Luxin Wang and Prashant Singh for theirinvaluable assistance and suggestions in this research.

References

Banatvala, N., Griffin, P.M., Greene, K.D., Barrett, T.J., Bibb, W.F., Green, J.H., Wells, J.G.,Siegler, R.L., Pavia, A., Hedberg, K., MacDonald, K., Belongia, E., Kulhanjian, J., Janner,D., Morrell, R.E., Miller, K., Atkin, B., De Beukelaer, M., Waldo, B., Houser, M.T.,Wead, T., Wendell, P., Knupp, C., Gerber, R.A., Frohlich, H., Bunchman, T., Wood, E.,Ornt, D., Loveys, A., Powell, K., Baetz-Greenwalt, B., Bastian, J., Blumberg, D., Cherry,J.D., Foreman, J., Gauthier, B., Trachtman, H., Frank, R., Rogers, J., Lettau, L., Manian,F.A., Alejos, J., Calk, T.E., Cunningham, C., Guyton, D., Ross, S., Hanevold, C., 2001.The United States national prospective hemolytic uremic syndrome study: microbio-logic, serologic, clinical, and epidemiologic findings. J. Infect. Dis. 183 (7), 1063–1070.

Fittipaldi, M., Rodriguez, N.J.P., Codony, F., Adrados, B., Peñuela, G.A., Morató, J., 2010. Dis-crimination of infectious bacteriophage T4 virus by propidium monoazide real-timePCR. J. Virol. Methods 168 (1–2), 228–232.

Fittipaldi, M., Nocker, A., Codony, F., 2012. Progress in understanding preferential detec-tion of live cells using viability dyes in combination with DNA amplification.J. Microbiol. Methods 91 (2), 276–289.

Food and Drug Administration, 1995. Bacteriological Analytical Manual, eighth ed. AOACInternational, Gaithersburg, MD.

Frenzen, P.D., Drake, A., Angulo, F.J., 2005. Economic cost of illness due to Escherichia coliO157 infections in the United States. J. Food Prot. 68 (12), 2623–2630.

Fricker, M., Messelhäußer, U., Busch, U., Scherer, S., Ehling-Schulz, M., 2007. Diagnosticreal-time PCR assays for the detection of emetic Bacillus cereus strains in foods andrecent food-borne outbreaks. Appl. Environ. Microbiol. 73 (6), 1892–1898.

García-Cayuela, T., Tabasco, R., Peláez, C., Requena, T., 2009. Simultaneous detection andenumeration of viable lactic acid bacteria and bifidobacteria in fermented milk byusing propidium monoazide and real-time PCR. Int. Dairy J. 19 (6–7), 405–409.

Josefsen, M.H., Löfström, C., Hansen, T.B., Christensen, L.S., Olsen, J.E., Hoorfar, J., 2010.Rapid quantification of viable Campylobacter bacteria on chicken carcasses, usingreal-time PCR and propidiummonoazide treatment, as a tool for quantitative risk as-sessment. Appl. Environ. Microbiol. 76 (15), 5097–5104.

Karmali, M.A., 2004. Infection by shiga toxin-producing Escherichia coli: an overview.Appl. Biochem. Biotechnol. B Mol. Biotechnol. 26 (2), 117–122.

Kralik, P., Nocker, A., Pavlik, I., 2010. Mycobacterium avium subsp. paratuberculosis viabil-ity determination using F57 quantitative PCR in combination with propidiummonoazide treatment. Int. J. Food Microbiol. 141, S80–S86 (Suppl.).

Lee, J.L., Levin, R.E., 2006. Use of ethidium bromide monoazide for quantification of viableand dead mixed bacterial flora from fish fillets by polymerase chain reaction.J. Microbiol. Methods 67 (3), 456–462.

Liang, N., Dong, J., Luo, L., Li, Y., 2011. Detection of viable Salmonella in lettuce bypropidium monoazide real-time PCR. J. Food Sci. 76 (4), M234–M237.

Løvdal, T., Hovda, M.B., Björkblom, B., Møller, S.G., 2011. Propidium monoazide com-bined with real-time quantitative PCR underestimates heat-killed Listeria innocua.J. Microbiol. Methods 85 (2), 164–169.

Mead, P.S., Slutsker, L., Dietz, V., McCaig, L.F., Bresee, J.S., Shapiro, C., Griffin, P.M., Tauxe,R.V., 1999. Food-related illness and death in the United States. Emerg. Infect. Dis. 5(5), 607–625.

Murakami, T., 2012. Filter-based pathogen enrichment technology for detection of multi-ple viable foodborne pathogens in one day. J. Food Prot. 75 (9), 1603–1610.

Murphy, N.M., McLauchlin, J., Ohai, C., Grant, K.A., 2007. Construction and evaluation of amicrobiological positive process internal control for PCR-based examination of foodsamples for Listeria monocytogenes and Salmonella enterica. Int. J. Food Microbiol.120 (1–2), 110–119.

Nocker, A., Cheung, C.Y., Camper, A.K., 2006. Comparison of propidium monoazide withethidium monoazide for differentiation of live vs. dead bacteria by selective removalof DNA from dead cells. J. Microbiol. Methods 67 (2), 310–320.

Pan, Y., Breidt Jr., F., 2007. Enumeration of viable Listeria monocytogenes cells by real-timePCR with propidium monoazide and ethidium monoazide in the presence of deadcells. Appl. Environ. Microbiol. 73 (24), 8028–8031.

Španová, A., Rittich, B., Karpíšková, R., Čechová, L., Škapová, D., 2000. PCR identification ofSalmonella cells in food and stool samples after immunomagnetic separation.Bioseparation 9 (6), 379–384.

Taskin, B., Gozen, A.G., Duran, M., 2011. Selective quantification of viable Escherichia colibacteria in biosolids by quantitative PCR with propidium monoazide modification.Appl. Environ. Microbiol. 77 (13), 4329–4335.

U.S. Department of Agriculture, Food Safety and Inspection Service, 2011. USDA TakesNew Steps to Fight E. coli, Protect the Food Supply. Designation Extends Zero

54 Y. Liu, A. Mustapha / International Journal of Food Microbiology 170 (2014) 48–54

Tolerance Policy for E. coli O157:H7 to Six Additional E. coli Serogroups. Avail-able at: http://www.usda.gov/wps/portal/usda/usdamediafb?contentid=2011/09/0400.xml&printable=true&contentidonly=true (Accessed 8 September 2013).

Wang, S., Levin, R.E., 2006. Discrimination of viable Vibrio vulnificus cells from dead cells inreal-time PCR. J. Microbiol. Methods 64 (1), 1–8.

Wang, L., Li, Y., Mustapha, A., 2007. Rapid and simultaneous quantitationof Escherichia coli O157:H7, Salmonella, and Shigella in ground beef by multi-plex real-time PCR and immunomagnetic separation. J. Food Prot. 70 (6),1366–1372.

Wang, L., Li, Y., Mustapha, A., 2009. Detection of viable Escherichia coli O157:H7 byethidium monoazide real-time PCR. J. Appl. Microbiol. 107 (5), 1719–1728.

Xiao, X.L., Tian, C., Yu, Y.G., Wu, H., 2013. Detection of viable but nonculturable Escherichiacoli O157:H7 using propidium monoazide treatments and qPCR. Can. J. Microbiol. 59(3), 157–163.

Yang, X., Badoni, M., Gill, C.O., 2011. Use of propidium monoazide and quantitative PCRfor differentiation of viable Escherichia coli from E. coli killed by mild or pasteurizingheat treatments. Food Microbiol. 28 (8), 1478–1482.

Yang, Y., Xu, F., Xu, H., Aguilar, Z.P., Niu, R., Yuan, Y., Sun, J., You, X., Lai, W., Xiong, Y., Wan,C., Wei, H., 2013. Magnetic nano-beads based separation combined with propidiummonoazide treatment and multiplex PCR assay for simultaneous detection of viableSalmonella Typhimurium, Escherichia coli O157:H7 and Listeria monocytogenes infood products. Food Microbiol. 34 (2), 418–424.