lntemationai Journal ofFoodMic&&gy

ELSEVIER International Journal of Food Microbiology 35 (1997) 239-250

The evaluation of a fluorogenic polymerase chain reaction assay for the detection of Salmonella species in food commodities

Shu Chen a,b, Arlene Yee b, Manse1 Griffiths a, Carolyn Larkin b, Carl T. Yamashiro ‘, Richa Behari ‘, Christine Paszko-Kolva ‘, Kris Rahn ‘, Stephanie A. De Grandis b,*

’ University of’ Guelph, Department of Food Science, Guelph, Ontario NIG 2WI. Canada

b Ontario Ministry of‘ A~riculturr, Food and Rural Affairs, Agriculturr and Food Laboratory Seroices Centrr, Guelph,

Ontario NlH 857, Canada

’ Perkin-Elmer Applied Biosystems, Foster City, CA 94404, USA

’ Agriculture and Agri-Food Canada, Health of’ Animals Laboratory, Guelph, Ontario NI G 3 W4, Canada

Received 5 July 1996; received in revised form 4 December 1996; accepted 27 December 1996

Abstract

The TaqManTM LS-SOB PCR Detection System facilitates the automated and direct detection of polymerase chain reaction (PCR) products. The system employs the 5’ nuclease activity of Tay DNA polymerase to hydrolyse a Salmonella specific internal fluorogenic probe for monitoring the amplification of a 287-bp region of the Salmonella incA gene. Using the fluorogenic 5’ nuclease assay, 164 Salmonella strains representing all the subspecies of Salmonella enterica were detected while over 50 non-Salmonella strains were not detected. The detection limit of the assay was two colony forming units (cfu) per PCR reaction when a pure culture of S. typhimurium was used. Six protocols for the isolation of PCR-amplifiable DNA were evaluated using chicken carcass rinses, ground beef, ground pork and raw milk contaminated with Salmonella. Of the six DNA isolation protocols, a modified sample preparation protocol using the EnviroAmp kit was chosen for subsequent studies because it was reliable, easy to use and efficient for the isolation of PCR-amplifiable DNA from foods. A detection limit of 3-7 cfu per PCR reaction was obtained using food samples that were pre-enriched overnight and then inoculated with Salmonella. The detection limit was below 3 cfui25 g or 25 ml when foods inoculated with Salmonella were pre-enriched overnight. Naturally contaminated foods (50 chicken carcass rinses and 60 raw milk samples) were examined using both the fluorogenic 5’ nuclease assay and a modified semi-solid rappaport vassiliadis (MSRV) culture method. Thirty four of the 110 samples tested were Salmonella-positive and 74 were Salmonella-negative by both the 5’ nuclease assay and the MSRV method. Two samples were Salmonella-positive by the 5’ nuclease assay, but negative by the MSRV method. The correlation between the 5’ nuclease assay and the MSRV method was over 98%. 0 1997 Elsevier Science B.V.

Keywordy: Polymerase chain reaction; Salmonella; Food-borne; DNA isolation; DNA detection

* Corresponding author: Tel.: + 1 5 19 767631 I; fax: + 1 519 7676240.

0168-1605/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved.

PIISO168-1605(97)01241-5

1. Introduction

The widespread increase in the number of cases of Sc~lrnotwll~~ infection in recent years has had major social and economic consequences which underlines the benefit of rapid pathogen monitor- ing (Tauxe. 1991 and Sockett. 1991). The detec- tion of Sulruonrllu by regulatory agencies is still primarily based on traditional culture methods that take several days to complete (Association of Official Analytical Chemists, 1984 and FDA, 1984). These culture methods are too time-con- suming to be used in regulatory food control and food production control. New molecular-based detection systems such as the polymerase chain reaction (PCR) have been increasingly used for rapid, sensitive and specific detection of food- borne pathogens (Harris and Griffiths, 1992 and Olsen et al., 1995). A number of PCR assays specific for different Sulmoncll~~ genes such as the irzrA gene (Rahn et al., 1992 and Wang et al., 1995), the insertion sequence IS200 (Cano et al., 1993) the 16 S ribosomal RNA gene (Iida et al.. 1993), the &A gene (Doran et al., 1993). the virulence-associated plasmid genes (Mahon J. and Lax, 1993 and Rexach et al.. 1994) and the ViaB sequence (Hashimoto et al., 1995) have been re- ported. Many of the assays employ either visual scoring of ethidium bromide-stained agarose gels or post-PCR hybridization-capture methods that are labour intensive, time-consuming and difficult to automate. Successful application of the PCR assays to food samples has been hindered by the lack of a convenient and relatively simple method for direct detection of PCR products, as well as by the lack of a rapid and efficient method for the preparation of PCR-amplifiable DNA from foods (Lantz ct al.. 1994). Other hindrances include technical complexity of PCR technology and in- sufficient data on its performance in the analysis of naturally contaminated foods.

A new fluorogenic PCR-based format employ- ing the homogeneous and automated direct detec- tion of PCR products has been developed (Livak et al., 1995) and applied to the detection of Listcj- I.~N /Ilonoc:rtoRenc.r (Bassler et al., 1995) and Shiga- like toxin (SLT)-producing E. coli (Witham et al., 1996). The assay utilizes the 5’ nuclease activity of

TLI(I DNA polymerase to hydrolyse an internal fluorogenic probe for monitoring amplification of DNA targets. The TaqManT” probe is labelled with both a quencher dye (rhodamine derivative) and a reporter dye (fluorescein derivative) allow- ing the quencher to absorb the ‘excited state’ energy of the reporter. When the probe anneals to a target sequence, it is subsequently cleaved by the 5’ nuclease activity of Tuq DNA polymerase dur- ing amplification. The separation of the two dyes due to hydrolysis of the probe allows an increase in reporter dye emission in proportion to amplifi- cation of DNA. This increase is recorded as a ARQ value which is a normalized ratio of reporter emission to quencher emission. The higher the ARQ value. the more the amplification of DNA: a ARQ value close to zero indicates no amplifica- tion of DNA.

In this paper, the development of the fluoro- genie 5’ nuclease assay for the detection of Strhmnell~r in food samples is described. Studies to elucidate the specificity and the sensirivity of the assay for pure cultures of S~~lnlorzc~lla and for .SLl//llonc~IILI-contaminated food samples were per- formed. Different protocols for the isolation of DNA from foods were evaluated. Foods such as chicken, red meat and milk, determined to be the main sources of food-borne salmonellosis (Tauxe, 1991), were tested. The applicability of the method to food samples was demonstrated.

2. Materials and methods

A total of 164 Suln~onrll~~ strains comprising seven subspecies of Salnwneh entrric’u and 52 nol~-S~~l~~~orzell~~ strains were used. These strains were provided by Drs Lucy Tompkins, Stanley Falkow and Bruce Stocker (Stanford University); Dr Mark Rodgers (EPA, Cincinnati); Dr Melissa Newman (Q laboratories, Cincinnati); Dr Manoushehr Shahamat (University of Maryland); Dr Raul Cano (California Polytechnic University, San Luis Obispo); Dr Joseph Chiu (FSIS, Alameda) and Dr Lee Ann Thomas (NVSL-DBL,

S. Chn et al. ilnternational Journal of Food Microbiology 35 (1997) 239-250 241

Ames). Cultures were maintained on appropriate isolation, and these food samples were desig- agar plates at 4°C. nated as post-enrichment-spiked samples.

A nalidixic acid-resistant (NalR) strain of Sulmonelb typhimurium (Health of Animals Laboratory collection, Agriculture and Agri- Food Canada) was used in the sensitivity tests. The cells were grown at 37°C overnight in Brain Heart Infusion (BHI) (Becton Dickinson) broth containing 30 pgg/ml nalidixic acid. The enumer- ation of Salmonellu was accomplished by plating dilutions (100 ~1) of overnight cultures on Xy- lose Lysine Desoxycholate (XLD) (Becton Dick- inson) plates containing 30 pgg/ml nalidixic acid and incubation overnight at 37°C.

2.3. DNA isolation

BPW pre-enriched food samples (1 ml) were centrifuged for 5 min at 16 000 x g to obtain pellets. DNA was prepared from the pellets using each of the following protocols:

2.2. Prepurution of food samples

Chicken carcass samples were collected from several local poultry plants in Ontario, Canada. Raw milk samples were obtained from the provincial bulk tanks. Red meat samples were purchased from a local retail store. Samples were stored on ice during shipping and pro- cessed immediately upon arrival.

Protocol 1. The sample pellet was washed with 1 ml of sterile distilled water and then resus-

pended in 200 ~1 of the InstaGene Matrix (Bio- Rad). The suspension was incubated at 56°C for 20 min, vortexed for 10 s and then boiled for 10

min in a water bath. The mixture was placed on ice for 10 min and then centrifuged at 16 000 x g for 5 min. The supernatant was used as the tem- plate.

Food samples (25 g or 25 ml) were homoge- nized in 225 ml of Buffered Peptone Water (BPW) (Oxoid) using a stomacher and incubated for 16- 18 h at 37°C. Chicken carcass rinses (CCR) were prepared by the method of Read et al. (1994) and pre-enriched overnight at 37°C. The pre-enriched samples were tested for Sulmonellu using the modified semi-solid rappa- port vassiliadis (MSRV) (Difco) culture method (Read et al., 1994) and were also used in the 5’ nuclease assays.

Protocol 2. The washed sample pellet (as de- scribed in protocol 1) was resuspended in 200 ,~l of an 1:4 dilution of DNA Extraction Reagent in TE buffer (Perkin-Elmer) and boiled for 5 min in a water bath. The mixture was then centrifuged at 16 000 x g for 5 min and the supernatant used as

the template.

In the artificial inoculation experiments, only those food samples that were confirmed Sulmonellu negative by both the MSRV method and the conventional PCR method (Chen et al., 1995) were used. Food samples were inoculated in two ways: (1) Food samples (25 g or 25 ml) were inoculated with O-100 cfu of S. fy- phimurium NalR prior to homogenization and pre-enrichment in BPW, and these food samples were designated as pre-enrichment-spiked sam- ples. (2) Food homogenate (250 ml) was pre-en- riched and then inoculated with 0-1.5~10~ cfu/ml of S. typhimurium NalR prior to DNA

Protocol 3. The protocol was developed from the recommended protocol of EnviroAmpm Le-

gionellu Sample Preparation Kit (Perkin-Elmer). The sample pellet was resuspended in 500 ~1 of the EnviroAmp DNA Extraction Reagent and boiled for 20 min in a water bath. The lysate was centrifuged at 16 000 x g for 30 s to remove cell and food debris. DNA was then precipitated from 400 ,~l of the supernatant using 400 ,~l of 100% isopropanol, and washed once with 500 ,ul of 75% isopropanol. The DNA pellet was resuspended in

160 ,~l of sterile distilled water and used as the template.

Protocol 4. The protocol was adapted from the recommended protocol of Nucleon I (ScotLab). The sample pellet was resuspended in 340 ,~l of Reagent B and 100 ~1 of sodium perchlorate, and incubated at 37°C for 20 min and then at 65°C for 20 min. The lysate was extracted once with 580 ,~l of chloroform in a Nucleon microtube with a silica plug. The DNA was precipitated with 880 ,~l of 100% ethanol, and the pellet washed with 1 ml

of 70% ethanol and resuspended in 160 ~1 of sterile distilled water.

Protocol 5. The template from protocol 1 was further treated with a Geneclean spin column (Bio/Can Scientific) according to the manufactur- er’s protocol.

Protocol 6. The protocol was modified from protocol 3 as follows: After the boiling step, the lysate was placed on ice for 5 min and then centrifuged at 16 000 x g for 3 min instead of 30 s to remove cell and food debris; prior to use. the template DNA solution was centrifuged for 5 min at maximum speed to further remove water- insoluble impurities.

Purified DNA was isolated from Salmonella and non-SaZmonella strains (Table 1) using tradi- tional phenol-chloroform procedures (Sambrook et al., 1989) and used in the specificity tests. DNA was also isolated from S. typhimurium NalK using 10 ~1 of the serial dilutions of the overnight culture and 200 ~1 of InstaGene ma- trix according to protocol 1 and used in the sensitivity tests.

2.4. DNA ampliJication and PCR product detection

The target sequence for the detection of Salmonella was a 287-bp region of the invA gene as described by Rahn et al. (1992). The fluores- cently labelled probe was designed and synthe- sized according to the guidelines of Bassler et al. (1995). PCR reaction mixes were prepared as described by Bassler et al. (1995) with the follow- ing modifications in concentrations of MgClz (6 mM), primers (100 nM) and probe (25 nM). In order to minimize contamination due to PCR product carry-over, uracil-N-glycosylase (0.5 unit, AmpEraseTM UNG, Perkin-Elmer) was in- cluded, and dUTP was substituted for dTTP in the reaction mixes. For each reaction (50 PI), 5 or 10 yl of DNA template was added. DNA from the pure culture of S. typhimurium NalR was used as a positive control. DNA from E. coli ATCC 25922 was used as a negative control and sterile distilled water as a no template control. Thermal cycling conditions (for the Perkin-Elmer GeneAmpTM PCR System 9600) were 50°C 2

min; 95°C. 5 min; 40 cycles of 95”C, 20 s and 65°C 1 min; 72°C 2 7 min.

Specific PCR products produced by the 5’ nu-

clease assay were detected using the TaqManTM LS-SOB PCR Detection System, an integrated system that includes a plate reader accessory (Perkin-Elmer). Forty microlitres of each sample, as well as negative and no template controls, were transferred to a 96-well microtiter plate (Perkin-Elmer) and read by the instrument. The equation ARQ = RQ + - RQ was used to ana- lyze the data. The variables are defined as fol- lows: RQ, reporter dye emission divided by quencher dye emission, RQ+ is the RQ value of a sample, and RQ- is the average RQ value of the no template controls. The PCR products (IO ~11) were also visualized using ethidium bromide- stained 2% agarose gels.

All experiments were done in duplicate. The average values of the duplicates or replicates are shown where appropriate.

3. Results

3.1. Spec(ficity and sensitivity of the 5’ nuclease assay using pure cultures of’ Salmonelkl

The specificity of the assay for Salmonella spe- cies was determined using 164 Salmonella and 52 non-Salmonella strains (Tables 1 and 2). These Salmonella strains represent common Salmonellu

serovars isolated from humans, animals and food products. All reactions containing Salmonellu re- sulted in ARQ values between 5 and 7. PCR products of the expected size (287-bp) were visu- alized on the agarose gels (results not shown). All reactions containing non-Salnlonellu had ARQ values close to zero and no PCR product of the expected size was observed on the gels.

The sensitivity of the assay was studied using dilutions of overnight pure cultures of S. ty- phimurium NalR. Three sets of Sulmonellu culture dilutions varying from 4 x lo4 to 4 x lo7 cfu/ml corresponding to 1.9 to 1900 cfu/PCR reaction were prepared. Assays were conducted in dupli- cate and then replicated on different days. Fig. 1

S. Chm rt al. /International Journal oJ’ Food Microbiology 35 (1997) 239-250 243

Table I

Salmonella serovars tested by PCR -

Serovar No. of strains

S. 18:242.32/

S. abaetetuha

S. abortus-equi

S. abortus-ocis

S. adrleide

S. agona

S. alhay)

S. anatwn

S. arinonae

S. austin

S. hai&

S. herta

S. bin-u

S. boukrr

S. horum

S. bovi.vnorb$cans

S. hraerrdwup

S. brandenhurg

S. hredeney

S. c’arrau

S. cerro

S. chandans

S. choleraesius

S. decatur

S. derb),

S. mrleagridis

S. memphis

S. maimi

S. monterideo

S. montgomer.~

S. muewhm

S. muenster

S. narstved

S. nebv&ton

S. nebvport

S. nienstedten

S. ohio

S. oranienburg

S. panarna

S. parat)phi A

S. paratyphi B

S. parat.vphi C

S. poona

S. pullorum

S. puttett

1

I 1

1

2

2

7

1

1

I 2

1

1 I I I

2

2

2

I 2

2

2

2

2

Serovar No. of strains

S. drypool

S. dublin

S. duishurg

S. eastbourne

S. rimsbutter

S. enterica

S. entrritidis

S. gallinarum

S. give

S. hadar

S. ha@

S. haoana

S. heidelberg

S. hemo

S. indiana

S. it+ntis

S. isangi

S. istanhul

S. jaoa

S. .jaoiana

S. ,johannesburg

S. kentucky

S. madelia

S. marogrosso

S. mbandaka

S. raw

S. reading

S. rubislatl

S. saintpaul

S. schrvarzmgrund

S. sendai

S. senftenberg

S. serenban

S. taksony

S. tennessee

S. thompson

S. typhi

S. typhimurium

S. typhimurium copc~nhagen

S. typhisuis

S. Uganda

S. wassena,

S. \corthington

s. spp.

2

4

I

2

2

I 1

2

2

2

1

8

15

1

8

shows a linear relationship between the log num- ber of cells and ARQ values. The detection limit was two CFU per PCR reaction which was also

confirmed using ethidium bromide/gel elec- trophoresis (lane d, Fig. 1). Good reproducibility was observed between the replicates.

Table 2

Non-Su/mm~//rr bacterial strains tested by PCR

Non-Sdnon~lla sp. No. of Strains

In our three preliminary studies in which DNA

was prepared from spiked chicken carcass rinses

(CCR) or pork samples using protocol 1 as de-

scribed in Section 2, amplification was inhibited

(results not shown). To obtain PCR-amplifiable

DNA, six protocols were used to isolate DNA

from post-enrichment-spiked food samples. and

these preparations were evaluated for use in the 5’

nuclease assay.

Fig. I Detection of Su/m~rrrl/rr in a pure culture using the 5’

nuclease assay and gel electrophoresis of the PCR products. The symbols: n . Run 1; + , Run 2; 1, Run 3. Lanes (cfu’

PCR): (a) 1900; (b) 190; (c) 19; (d) 1.9: (e) 0; M. IOO-bp ladder (Gibco BRL life Technologies).

Fig. 2(a) illustrates the relationship between the log number of target cells and ARQ values ob- tained using protocols l-5 for isolation of DNA from a S~ln~clrzrllu post-enrichment-spiked CCR. Fig. 2(b) shows the corresponding ethidium bro- mide stained gels of the PCR products. The DNA preparations using protocol 3 (EnviroAmp) and 4 (Nucleon I) resulted in an increase in ARQ values with the increase of the number of cells, while the DNA preparations using protocols 1, 2 and 5 resulted in relatively constant ARQ values irre- spective of the number of cells. These ‘constant’ low ARQ values were the background fluores- cence readings of the samples as the fluorescence readings of the unamplified reaction mixtures for unspiked samples were found to be about 1.0 as measured by ARQ values.

Fig. 3(a.b) summarize the results from the eval- uation of DNA isolation protocols l-5 using postenrichment-spiked raw milk. Protocol 4 re- sulted in amplification detection at a spiking level of 0.6 cfu:‘PCR or higher as seen by the linear rise of ARQ values relative to the logarithmic increase of Sc~lr~~ot~c~ll~~ cells applied to the samples. In comparison. protocol 5 resulted in some amplifi- cation detection at a spiking level of about 600 cfu/PCR and no amplification was detected with the DNA preparations using protocols 1, 2 and 3. The ARQ values for protocols 1 to 3, as well as for 0.06 and 0 cfu per reaction for protocol 4 were between 2 and 2.5. Yet the gel results showed that no amplification occurred at all for protocols 1, 2 and 3. as well as with 0.06 and 0 cfu for protocol 4. However, when the unspiked food samples prepared using these methods were added to the PCR mix and analyzed prior to thermal cycling (unamplified), the ARQ values were between 2 and 2.5. This indicates that the significant ARQ value is a result of fluorescent contaminants in the food sample itself that were not removed by the sample preparation method. The lack of amplifi- cation at 21 detectable level for protocols I. 2 and 3 also suggests that PCR inhibitors were not removed from the sample or the DNA was not recovered efficiently by these methods. These re- sults show that these methods need further opti- mization before they can be used effectively to prepare food samples for PCR and/or 5’ nuclease assay analysis.

S. Cllen c’t al. Untrmational Journal of Food Microbiology 35 (1997) 239-250 245

J 0.075 0.75 7.5 75 754

Spiking level (cfu/PCR)

Mabcd e f abcdef Mabcdefgh abcdef ab cd e f

Fig. 2. (A) Detection of Su/monrllcr in a post-spiked chicken carcass rinse using the 5’ nuclease assay: comparison of five DNA

isolation protocols. The symbols: 0, protocol I; + , protocol 2; W , protocol 3; A, protocol 4; x, protocol 5. (B) Gel electrophoresis of PCR products amplified using the 5’ nuclease assay with different DNA preparations from a post-spiked chicken carcass rinse.

Gels: I. protocol I; II, protocol 2: III, protocol 3; IV, protocol 4; V, protocol 5. Lanes (cfu;PCR): (a) 750; (b) 75; (c) 7.5: (d) 0.75;

(e) 0.075; (f) 0; (g) positive control (Salmonella, 950 cfu:PCR); (h) negative control (no DNA); M, lOO-bp ladder.

Protocol 6, a modified version of protocol 3,

was also tested using post-enrichment-spiked raw

milk because protocol 3 was rapid and easy to

use, but its performance was food-dependent (Fig. 2(a,b) for a chicken carcass rinse vs. Fig. 3(a,b)

for raw milk). Fig. 3(c) shows a linear relationship

between the log number of cells and the ARQ values, and gel electrophoresis of the PCR prod-

ucts using raw milk. As demonstrated in Fig. 3(c),

DNA isolated from the post-enrichment-spiked

raw milk using protocol 6 was amplifiable by the

fluorogenic 5’ nuclease assay. In addition, the DNA preparations using protocol 6 had low

fluorescence values when analyzed prior to ther-

mal cycling (ARQ < 0.5) and the amplification

detection limit was at 0.35 cfu/PCR, suggesting this method was better at removing PCR in-

hibitors and fluorescent contaminants than previ- ous methods.

Of the six DNA isolation protocols evaluated for the chicken carcass rinses and raw milk, pro-

tocol 4 and 6 seemed to be potentially suitable for DNA isolation from these foods. These two pro-

tocols were further evaluated using post-enrich- ment-spiked ground beef and ground pork samples. Both protocols resulted in PCR-amplifi- able DNA with low background ARQ values ( < 0.5) from both food commodities in the PCR assay (Figs. 4 and 5). The 287-bp PCR product was also observed on the agarose gels for the reactions that contained as low as 0.7 cfu/PCR (results not shown). Since protocol 6 resulted in PCR-amplifiable DNA for all the food commodi- ties tested, and it was efficient, easy and reliable to use, it was chosen for the subsequent studies.

3.3. Sensiticity of the assuy on pre-enrichment -spiked foods

The sensitivity of the assay was evaluated using raw milk, ground beef and ground pork that were inoculated with known numbers of S. ty- phimuriurn NalR cells and pre-enriched overnight. These samples were confirmed initially negative for Salmonellu as demonstrated by the conven- tional PCR and the MSRV method. Table 3 and

c MdhCd it ah‘d cf abc d rf Mabcdrf ah cd e1gh

I II III I\ \

Fig. 3. (A) Detection of Su/nzonella in post-spiked raw milk using the 5’ nuclease assay: comparison of five DNA isolation protocols.

The symbols are same as those in Fig. 2(a). (B) Gel electrophoresis of PCR products amplified using the 5’ nuclease assay with

different DNA preparations from post-spiked raw milk. Gels: I, protocol I; II, protocol 2; III. protocol 3: IV, protocol 4; V,

protocol 5. Lanes (cfu/PCR): (a) 600; (b) 60; (c) 6.0; (d) 0.6; (e) 0.06: (f) 0; (g) positive control (Salnmdla, 950 cfu,/PCR): (h)

negative control (no DNA): M. IOO-bp ladder. (C) Detection of S~/rr~vzr//r in post-spiked raw milk using the 5’ nuclease assay and

gel electrophoresis of the PCR products. Protocol 6 was used for DNA isolation. Lanes (cfu:PCR): (a) 350; (b) 35; (c) 3.5: (d) 0.35; (e) 0.035; (f) 0: (g) positive control (Salmonella, 950 cfu:PCR); (h) negative control (no DNA); M. IOO-bp ladder.

Fig. 6 show the results of sensitivity tests on the pre-enrichment-spiked foods. All positive samples had ARQ values above 6.5 and a PCR product of 287-bp on gels while all negative samples had ARQ values close to zero and no PCR product of

1 -1 0 ‘A 0075 0.75

Spiking level (cfu/PCR)

Fig. 4. Detection of S~bnonellu in post-spiked ground beef using the 5’ nuclease assay: comparison of two DNA isolation

protocols. The symbols: A, protocol 4; *. protocol 6.

the expected size observed on gels. In the pre-en- richment-spiked raw milk and ground beef sam- ples the detection limit was below two CFU per 25 g or 25 ml of food; in the pre-enrichment- spiked pork, the limit was three CFU per 25 g.

b O.b7 0:7 + A 760 Spiking level (CfulPCR)

Fig. 5. Detection of Sahonrlla in post-spiked ground pork using the 5’ nuclease assay: comparison of two DNA isolation

protocols. The symbols are the same as those in Fig. 4.

S. Chrn et al. /International Journal of’ Food Microbiology 35 (1997) 239-250 241

Table 3

ARQ values of pre-spiked foods as detected using the fluorogenic 5’ nuclease assay and DNA extraction protocol 6

Spiking level (cfu25g or 25 mL) 0 1.5 2 3 4 10 100

Foods

Raw milk

Ground beef

Ground pork

-0.11 ND* 7.10 ND* 6.64 7.40 7.04

-0.05 7.69 ND* 7.44 ND” 7.97 8.14

-0.20 -0.16 ND” 7.28 ND” 7.42 7.88

Controls

Salmonella (950 cfu/PCR)

Ecoli (8800 cfu/PCR)

7.65

-0.03

il Not determined.

3.4. Evcrluation of the 5’ nuclease ussay on naturally-contaminuted chicken CARCASS rinses (CCR) and ruw milk

The applicability of the TaqManTM method to naturally-contaminated food samples was evalu- ated using 50 CCR and 60 raw milk samples, with protocol 6 used for DNA isolation. The MSRV method was used as the standard culture method for the comparative study. Of the 110 samples, 34 (33 CCR and one raw milk) were positive and 74 (15 CCR and 59 raw milk) were negative for Sulmonellu by both methods; two CCR were posi- tive by the fluorogenic 5’ nuclease assay only (Table 4). There was a 98% agreement between the fluorogenic 5’ nuclease assay and the MSRV culture method.

I II III

Fig. 6. Gel electrophoresis of PCR products amplified using

the 5’ nuclease assay with DNA preparations from (25 mL or

g) pre-spiked raw milk (Gel I), ground beef (Gel II) and ground pork (Gel III) using protocol 6. Lanes (cfu/25 mL or

g): (a) 100; (b) 10; (c) 4; (c’) 3; (d) 2: (d’) 1.5; (e) 0; (I) positive control (Salmonella, 950 cfu/PCR); (g) negative control (E.

coli, 8800 cfu/PCR); M, IOO-bp ladder.

4. Discussion

4.1. SpeciJicity and sensitivity of‘ the 5’ nucleuse assay

Rahn et al. (1992) demonstrated that the inuA

gene contains sequences unique to Salmonella and is a suitable PCR target. Their primer set was shown to be specific for Sulmonellu based on extensive tests using 630 Salmonella and 142 non- Sulmonellu strains (Rahn et al., 1992). In this study the primer set was modified by the addition of two nucleotides on the reverse primer, and this modification was found not to adversely affect the specificity of the primer. The results obtained from our study demonstrate the specificity of the modified primer set and the probe for Salmonella species.

The primers used in this study were designed to produce a 287-bp fragment from the Salmonella

invA gene, but some non-specific amplifications were observed in some of the experiments (e.g. Fig. 3(c), gel). These non-specific amplifications, however, did not contribute to the ARQ values (Fig. 3(c), graph) obtained using the 5’ nuclease assay, indicating that the TaqManTM probe al- lowed detection of specific amplification only.

The detection limit of the fluorogenic 5’ nucle- ase assay developed in this study was two cfu/ PCR using a pure culture of Salmonella which correlates with the detection limit using agarose

Table 4

Detection of S~ln~onella in naturally-contaminated chicken carcass rinses (CCR) and raw milk using the fluorogenic 5’

nuclease assay and the MSRV culture method.

Sample” ARQ Gelh MSRV

CCR 01 6.55

CCR 02 6.56

CCR 03 4.00

CCR 04 5.08 + CCR 05 5.72

CCR 06 5.84

CCR 07 4.25

CCR 08 6.4 I + CCR 09 6.92

CCR 010 4.57

CCR 01 I 4.02

CCR 012 6.52

CCR 013 6.20

CCR 014 6.21

CCR 015 6.25

CCR 016 6.09 + CCR 017 6.15

CCR 018 5.96

CCR 019 5.89

CCR 020 5.69

CCR 021 6.19 + CCR 022 6.00 + CCR 023 4.59

CCR 024 6.24 + CCR 025 6.40

CCR 026 6.15

CCR 027 6.02

CCR 028 5.98

CCR 029 5.97

CCR 030 5.62

CCR 031 5.99

CCR 032 6.30

CCR 033 6.15

CCR 034 6.02

CCR 035 6.04

Milk l-15 3.50

SUII?l. 6.36

E. di 0.00

+ + + + f + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +

j’ All the remaining 74 food samples had ARQ values close to

zero and were Subnonrll~-negative by the MSRV

method.” Gel+ (or gel + ). presence (or absence) of a 287-bp PCR product upon agarose gel electrophoresia.

gel electrophoresis. This detection level is 2 logs lower than that (300 cfu/PCR) reported by Rahn et al. (1992). This could be mainly due to im- proved DNA isolation methods and optimized PCR conditions.

Application of PCR for detection of pathogens in food samples is often limited by the presence of substances that inhibit the PCR (Rossen et al., 1992). The effect of DNA quality on the perfor- mance of the 5’ nuclease assay was elucidated in this study. When DNA was not sufficiently purified, such as with protocol 1 and 2, amplifica- tion was inhibited. In addition, impurities, when carried over to the 5’ nuclease assay may con- tribute to background fluorescence readings (ARQ values). resulting in false-positives if pre-set cutoff values are employed for interpretation of results. One strategy to circumvent background fluores- cence is by taking pre and post reads of the samples and subtracting any inherent background fluorescence.

There has been no standard DNA isolation method which is universally applicable for foods although many methods have been postulated (Lantz et al., 1994). Many reported methods use extensive dilution of DNA (or small sample size) to ensure the removal of polymerase inhibitors, resulting in a loss of sensitivity (Chen et al.. 1995). This problem is usually counteracted by extensive DNA purification although this may affect accu- racy, sensitivity and reliability as DNA is lost at every stage of purification, and increase the chances of DNA contamination. In this study the inhibitory problems encountered with food com- modities were resolved after extensive evaluation of six different DNA isolation protocols. It was found that protocol 6, the modified EnviroAmp protocol, resulted in PCR-amplifiable DNA with low background fluorescence readings, and it was easy. efficient and reliable for DNA isolation

from the food commodities selected for the study. A 1 ml sample size for DNA isolation was used, and approximately 5% (50 /ll) of the I ml sample was carried over to each PCR reaction. This resulted in detection of about 100 cfuiml of Sdnmwllu in post-enrichment-spiked foods (equivalent to five cful’PCR).

The effectiveness of DNA preparation and PCR procedures is often commodity-dependent due to varying amounts and types of PCR in-

S. Chm et ul. / Intrrnarional Journul qf Food Micwhiology 35 (1997) 239-250 249

hibitors in different food samples. In this study stronger inhibition and higher background fluorescence readings were found with raw milk samples than with chicken carcass rinses. This may be due to the high protein and fat content of raw milk. By using more effective conditions for centrifugation following cell lysis in the Envi- roAmp protocol, more cell debris and protein particles could be removed from the raw milk samples, preventing their co-precipitation with DNA in the late stages of purification. By adding a final centrifugation step in the Envi- roAmp protocol, residual water-insoluble sub- stances could be further removed. These modifications had a very significant effect on the results obtained. Our study revealed that it is very important to maximize the purification effi- ciency for every step of DNA isolation when a DNA isolation protocol is developed for differ- ent food commodities.

4.3. Applicution of the 5’ nuclrus~ assuy to jtiod

samples

Since the presence of one culturable Salmonellu cell in 25 g or 25 ml of food samples must be detected, it is imperative that a pre-en- richment step be undertaken when food samples are examined. Based on the results obtained with post-enrichment-spiked food, pre-enrichment which increases the target cell number above 100 cfu/ml is necessary. In this study, detection of less than two to three cfu Sulmonellu in 25 g or 25 ml of foods was achieved following overnight pre-enrichment. The high values of ARQ ( > 7) obtained in this study with the pre-enrichment- spiked food samples indicated that overnight pre- enrichment is sufficient to ensure the sensitivity of the assay.

For pre-enrichment-spiked ground pork, a spiking level of 3 cfu125 g was detected but a level of 1.5 cfu/25 g was not detected using the fluorogenic 5’ nuclease assay. The main reasons for this could be: (1) no target cells were actually inoculated into the sample due to the low con- centration ( < 2 cfu/25 g); or (2) the target cells were injured during homogenization and did not grow to a detectable level.

Once a PCR method is developed, the method must be verified with a standard culture method. In this study, the fluorogenic 5’ nuclease assay was compared with the MSRV culture method using 110 naturally-contaminated food samples. The PCR assay was able to detect all of the 34 Salmonella-positive samples that were confirmed positive by the MSRV method. In addition, the assay detected two more Sulmonellu-positive chicken samples as compared with the culture method. The main reasons for this could be that the target cells were injured or dead, i.e were non-culturable but detectable by the PCR method, or that the MSRV method detected only the motile strains of Sulmonellu.

5. Conclusions

The 5’ nuclease assay evaluated in this research was specific, sensitive and rapid for the detection of Sulmonellu in food samples including chicken carcass rinses, raw milk, ground beef and ground pork. By using the modified EnviroAmp DNA preparation protocol and the 5’ nuclease assay, less than three CFU of Sulmonellu in 25 g or 25 ml of food were detected after overnight pre-en- richment. The procedures were also successfully applied to the detection of Sulmonellu in 110 naturally-contaminated raw milk and chicken carcass rinse samples. The total turn-around time was less than 24 h. The TaqManTM LS-SOB PCR Detection System allows automated and direct detection of the PCR products for more than 96 samples in 10 min. The 5’ nuclease assay utilizes uracil-N-glycosylase (AmpEraseTM UNG) to pre- vent PCR product carry-over so that a potential PCR product contamination problem is mini- mized. This 5’ nuclease assay system facilitated the evaluation of several different sample prepa- ration methods for PCR in an analytical manner. Due to this automatable and direct means of analyzing PCR products, this system represents a significant step towards the application of PCR in food microbiology. However, before this sys- tem is ready to be implemented in routine labs, it is suggested to further simplify the DNA isola- tion procedures.

Acknowledgements

This research was supported by Ontario Food Quality and Safety Research Fund. We are also grateful to federal and provincial food inspectors for collecting the food samples, to Janet Harris and Marina Steele for performing the MSRV assay and to Claudia Thio and Susan Flood for their valuable suggestions and support.

References

Association of Official Analytical Chemists. (1984) Official

methods of analysis of the Association of Official Analyti-

cal Chemists. 14th edn.. Arlington. V.A.: Association of

Official Analytical Chemists Inc.

Bassler, H.A.. Flood. S.J.A., Livak. K.J.. Marmaro, J., Knorr.

R. and Batt. C. (1995) Use of a Fluorogenic Probe in a

PCR-Based Assay for the Detection of Li.srrr-icr mor?oc:i~ro-

genes. Appl. Environ. Microbial. 61. 3724 3728.

Cano, R.J., Rasmussen, S.R., Fraga, G.S. and Palomares. J.C.

(1993) Fluorescent detection-polymerase chain reaction

(FD-PCR) assay on microwell plates as a screening test for

salmonellas in foods. J. Appl. Bacterial. 75, 2477253.

Chen. S., Yee. A.. Griffiths. M.W.. Rahn. K.. Johnson. R..

Read, S.. Gyles. C.. De Grandis. S., Martin. A., Bielawski.

D. and Harris, L. i 1995) Sample processing techniques to

enhance PCR detection of SCII~OII~IILI in poultry. meat, and

milk products. Poster 150. 45th Annual Canadian Society

of Microbiologists (CSM). June I I 15. 1995. Kingston,

Canada.

Doran. J.L.. Collinson, S.K.. Burian. J., Sarlos, Cr., Todd,

E.C.D., Munro. C.K.. Kay. C.M.. Banser. P.A.. Peterkin.

P.I. and Kay. W.W. (1993) DNA-based diagnostic tests for

Su/~nor~rllu species targeting ~/g/.4, the structural gene for

thin aggregative fimbriae. J. Clin. Microbial. 31. 2263

2273. Food and Drug Administration. (1984) Bacteriological analyt-

ical manual. 6th cdn, Arlington, VA.: Association of

Official Analytical Chemists Inc.

Harris, L.J. and Griffiths. M.W. (1992) The detection of

foodborne pathogens by the polymerase chain reaction.

Food Res. Int. 25. 4577469.

Hashimoto. Y.. Itho. Y.. Fujinaga. Y.. Khan, A.Q., Sultana. F.. Miyake, M., Hirose, K., Yamamoto, H. and Ezaki, T. (1995) Development of nested PCR based on the I’icrB

sequence to detect Sulnionrlltr r,tphi. J. Clin. Microbial. 3.1.

775 777.

lida. K.. Abe, A.. Matsui, H., Danbard, H.. Wakayama, S.

and Kawahara, K. (1993) Rapid and sensitive method for detection of Salmonella strains using a combination of polymerase chain reaction and reverse dot-blot hybridiza-

tion. FEMS Microbial. Lett. 114, 167-172. Lantz, P.-G., Hahn Hagerdal, B. and Radstrom, P. (1994)

Sample preparation methods in PCR-based detection of

food pathogens. Trends Food Sci. Technol. 5. 384 389. Livak, K.J.. Flood, S.J.A., Marmaro, J., Giusti. W. and Dectz.

K. (1995) Oligonucleotides wjith fluorescent dyes at oppo- site ends provide a quenched probe system useful for detecting PCR product and nucleic acid hybridization.

PCR Methods Appl. 4. 3577362. Mahon J. and Lax, A. (1993) A quantitative polymerase chain

reaction method for the detection in avian faeces of

salmonellas carrying the spcR gene. Epidemiol. Infect. I I I.

455-464. Olsen. J.E.. Aabo, S., Hill, W., Notermans, S.. Wernars. K..

Granum, P.E., Popovic, T., Rasmussen, H.N. and Olsvik,

0. (1995) Probes and polymerase chain reaction for detec- tion of food-borne bacterial pathogens. Int. J. Food Mi-

crobiol. 28, l-78. Rahn. K.. De Grandis, S.A., Clarke, R.C.. McEwen, S.A..

Galan, J.E., Ginocchio, C., Curtiss 111, R. and Gyles, C. L. (1992) Amplification of an inrA gene sequence of

&ln?o,~rllu f~phimurium by polymerase chain reaction as a specific method of detection of .Sn/~non~//cr. Mol. Cell. Probes 6. 271-279.

Read, S.. Iwjin. R.J.. Poppe, C. and Harris. J. (1994) A Comparison of two methods for isolation of Srrbnonellrr

from poultry litter samples. Poultry Sci. 73, 1617~1621. Rexach. L.. Dilasser. F. and Fach, P. (1994) Polymerase chain

reaction for Saltnc~nellu virulence- associated plasmid genes detection: a new tool in Salmonella epidemiology. Epi-

demiol. Infect. 112, 33343. Rossen. L., Norskov. P., Holmstrom, K. and Rasmussen, 0.

(1992) Inhibition of PCR by components of food samples, microbial diagnostic assays and DNA-extraction solutions.

Int. J. Food Microbial. 22. 37745. Sambrook. J.. Fritsch, E.F. and Maniatis, T. (1989) Molecular

Cloning: A Laboratory Manual, second edition. Cold Spring Harbor Laboratory Press, Cold Spring Harbour.

Sockett, P.N. (1991) The economic implications of human salmonella infection. J. Appl. Bacterial. 71. 289 295.

Tauxe, R.V. (1991) Sulmonellrr: A postmodern pathogen. J. Food Protection. 54, 563 568.

Wang H.. Blais. B.W. and Yamazaki. H. (1995) Rapid confir- mation of polymyxin-cloth enzyme immunoassay for group

D salmonellae including Salmondlu enteritidis in eggs by polymerase chain reaction. Food Control 6. 205-209.

Witham, P.K., Yamashiro, CT., Livak. K.J. and Batt, C.A. (I 996) A PCR-based assay for the detection of E.w/rrrichirr

c~olr Shiga-like toxin genes in ground beef. Appl. Environ. Microbial. 62, 134771353.