FOOD BIOLOGICAL CONTAMINANTS

Rapid Enzyme-Linked Immunoassay for Detection ofSalmonellain Food and Feed Products: Performance Testing Program

BOLTON ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 2, 2000FREDERICK J. BOLTON, ERICA FRITZ , and SHARON POYNTON

Royal Preston Hospital, Public Health Laboratory, PO Box 202, Sharon Green Ln, Preston, PR2 9HG, United KingdomTORBEN JENSEN

1

Bioline, Polarvej 60, DK-7100, Denmark

The BIOLINE Salmonella ELISA Test for Salmonellaspp., which is a rapid, easy, and convenient assaywas evaluated for use in detecting Salmonella infoods and feeds. Each food matrix or feed was artifi-cially contaminated with low levels of Salmonella .Twenty different matrixes were studied and 20 dif-ferent Salmonella strains from a broad variety ofserogroups (B, C, D, E, F, G, H, I, M, O, P, and U)were used. The ELISA Test kit detected levels aslow as 1 cfu/25 g sample with at least 4 of the 20 ma-trixes tested. The test kit is applicable to all sampletypes tested. The BIOLINE Salmonella ELISA Testkit has been granted AOAC–RI performance testedstatus.

The detection ofSalmonellain food and feed using conven-tional techniques is a process that takes 3–5 days. Ap-proaches to decrease the detection time have included

shortening both the enrichment phase and the detection phase.However, most attempts to shorten the enrichment phase havenot led to significantly shortened enrichment protocols of any ofthe standard methods such as theFDA Bacteriological Analyti-cal Manual(BAM), International Standards Organization (ISO),and the Nordic Committee on Food Analysis (NMKL). Attemptsto shorten the detection phase using specific antibodies or spe-cific DNA techniques have been more successful (1). Most fastdetection methods have not been approved or even validated.Those approved have only been adopted on a limited scale due tosignificantly higher costs, limited usability on difficult matrixes,and questionable advantages.

The BIOLINESalmonellaELISA Test 96 is an ELISA testkit for determining the presence ofSalmonellain foods. Thekit is optimized for all sample types with a simple, efficient,and reliable enrichment protocol forSalmonella, which re-quires only 1 preenrichment and 1 selective enrichment step.Samples from the selective enrichment buffer are processedwith reagents provided by the BIOLINESalmonellaELISA

Test 96 kit. Antibody-coated wells capture theSalmonellaflagellar proteins, which are then exposed to an en-zyme-conjugated antibody followed by substrate after un-bound conjugate is removed. The color of the substrate turnsblue in the presence of bound enzyme, thereby signaling thepresence ofSalmonella flagellar proteins. The enzymecolorimetric reaction is stopped by adding dilute sulfuric acid.

The purpose of this trial was to evaluate the BIOLINESal-monellaELISA Test 96 against the BAM method in accor-dance with the validation requirements of the AOAC Re-search Institute. The BIOLINESalmonellaELISA Test uses astandard ELISA format with dividable microtiter strips. Allreagents are liquid, stable for 1 year, and ready to use, and theprotocol may be adopted on any automatic platform compati-ble with standard ELISA format. Manually, more than500 ELISA samples per day may be processed, and automat-ing may process 2000 ELISA samples a day. The results maybe obtained after 36 h, and are at least as reliable as the refer-ence protocol used (2).

Experimental

Media

The media and components were purchased from Oxoid,Ltd., Basingstoke, Hampshire, UK; Difco Laboratories, Ltd.,West Molesey, Surrey, UK; BDH/Merck, Merck, Ltd., Poole,Dorset, UK; Mast Diagnostic, Bootle, Merseyside, UK;bioMeriéux UK, Ltd., Basingstoke, Hampshire, UK.

The following media were used for the BAM method andfor confirmation of positive isolates: lactose broth (LB),Oxoid Cat. No. CM137; trypticase soy broth (TSB), Difco;lauryl tryptose broth (LST), Oxoid Cat. No. CM451; skim milkpowder, Oxoid Cat. No. L31; selenite cystine broth (SCB),Oxoid Cat. No. CM699; tetrathionate (TT), Oxoid Cat.No. CM671; bismuth sulfite agar (BS), Oxoid Cat. No. CM201;Hektoen enteric agar (HE), Oxoid Cat. No. CM419; xyloselysine desoxycholate (XLD), Oxoid Cat. No. CM469; lysineiron agar (LIA), Oxoid Cat. No. CM381; MacConkey agar(MA), Oxoid Cat. No. CM813; nutrient agar (NA), Oxoid Cat.No. CM3; triple sugar iron (TSI), Oxoid Cat. No. CM277.Rappaport-Vassiliadis medium (RVM) for the BAM procedurewas prepared by using a broth base containing tryptone 5 g/L,Oxoid Cat. No. L42; NaCl 8 g/L, BDH Cat. No. 30123;

BOLTON ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 2, 2000 299

Received August 28, 1998. Accepted by AH June 19, 1999.1 Author to whom correspondence should be addressed, e-mail

contact@bioline.dk.

300B

OLT

ON

ET

AL.:JO

UR

NA

LO

FA

OA

CIN

TE

RN

AT

ION

AL

VO

L.83,NO.2,2000

Table 1. Inoculation data and results of BAM and BIOLINE Salmonella immunoassay comparative study

Food group Food product Salmonella serovarSerotype

group Serotype Source MPN/25 g

Total No. of positives

BAM method BIOLINE assay

Presumptivepositives Confirmed

Presumptivepositives Confirmed

1 Animal feed Panama D 1,9,12:1v:1,5 Bonemeal 1 2 1 1 1

1 Animal feeda Panama D 1,9,12:1v:1,5 Bonemeal 5 9 9 9 9

1 Animal feeda Panama D 1,9,12:1v:1,5 Bonemeal 50 10 10 10 10

1 Cheese Tel-hashomer F 11:z10:n,x NCTC 8704 50 10 10 10 10

1 Dried whole eggs Lindi P 38:r:1,5 Human 5 8 5 10 9

1 Macaroni Nottingham I 16:d:e,n,z15 Human 2.25 7 6 9 9

1 Cake mix Ahuza U 43:k:1,5 Human 6 9 8 6 6

1 Milk powder Poona G 1,13,22:z:1,6 NCTC 8440 5.3 7 6 8 8

1 Coconut Hadar C 6,8:z10:e,n,x Human 5.3 10 10 9 9

1 Soy flour London E 3,10,[15]:1v:1,6 Human 1 6 5 5 4

1 Peanut butter Warragul H [1],6,14,[25]:gm:- NCTC 9942 1 8 8 4 4

2 Fish Agona B 1,4[5],12:f,g,s:[1,2] Seawater 20.5 0 0 10 10

2 Raw chicken meat Enteritidis D 1,9,12:[f],gm,[p]:[1,7] Human 10 10 9 10 10

2 Pork sausage Virchow C 6,7:r:1,2 Human 5.25 10 8 10 8

2 Raw ground beef Typhimurium B 1,4,[5],12:i:1,2 Human 50 10 9 10 9

2 Shrimp/prawns Kedougou G 1,13,23:i:l,w Seawater 17.5 10 10 10 10

2 Frog legs Anecho O 35:g,s,t:- Human 5 9 8 6 6

3 Black pepper Give E 3,10,15,[15,34]:[d]lv:1,7 Chicken 1 6 6 8 8

3 Onion powder Hull I 16:b:1,2 Human 10 10 7 8 8

3 Liquid whole eggs Pomona M 28:y:1,7 Food 2.1 10 10 8 8

4 Dried yeast Rubislaw F 11:r:e,n,x Food 10 10 9 9 9

5 Milk chocolate Uzaramo H 1,6,14,25:z4z24:- NCTC 8493 1.75 7 6 3 3

Totals 178 160 173 168

a The matrix animal feed was repeated because the first setup only yielded 1 confirmed positive with the BAM and the Bioline Salmonella Immunoassay.

KH2PO4 1.6 g/L, BDH Cat. No. 34045; and combining 1 L ofthis broth base with 100 mL magnesium chloride solution(400 g MgCl2⋅6H2O dissolved in 1 L distilled water) and10 mL malachite green oxalate solution (0.4 g malachite greenoxalate, BDH Cat. No. 34045, dissolved in 100 mL distilledwater). This solution was dispensed in 10 mL volumes in glasstubes and sterilized by autoclaving at 115EC for 15 min.

The following media were used for the immunoassay proce-dure: buffered peptone water (BPW), Merck Cat. No. 7228; andRappaport-Vassiliadis soya broth (RVS), Mast Cat. No. DM269.

Reagents and Chemicals

Brilliant green dye, BDH Cat. No. 34015; a solution of10% ferric chloride, BDH Cat. No. 10110; iodine, BDH Cat.No. 28564; a solution of 40% KOH, BDH Cat. No. 10210;Kovacs reagent prepared with concentrated HCl, BDH Cat.No. 2850734; amyl alcohol, BDH Cat. No. 272126W;para-dimethyl-amino benzaldehyde, BDH Cat. No. 28188N;α-naphthol, BDH Cat. No. 101623M; potassium iodide,BDH Cat. No. 29631; sodium biselenite, Oxoid Cat. No. L121;and saline, Oxoid Cat. No. BR053G. API 20E kit, bioMeriéuxCat. No. 20100.

Bacterial Strains

Strains used in the comparative study and the exclusivitystudy are listed in Tables 1 and 2, respectively. Cultures werekept on nutrient agar slopes at 4EC and were subcultured tofresh slopes monthly.

Artificial Inoculation of Salmonella Strains into FoodMatrixes

Depending on the food type, 1 of 4 procedures was used toinoculate the food matrixes with theSalmonellastrains. Ani-mal feed, cheese, dried whole eggs, macaroni, cake mix, milkpowder, soy flour, dried yeast, and milk chocolate were each

artificially inoculated with the serovars listed in Table 1. Forthese samples theSalmonellastrain was inoculated into10 mL BPW and incubated at 37EC for 20–24 h. This mixturewas added to sterile solution of milk, prepared by adding 25 gmilk powder to 225 mL BPW.These cultures were incubated at37EC for 20–24 h, after which 15–20 mL were added to severalsterile Petri dishes. The Petri dishes were stored at –80EC for6–24 h, and were then freeze-dried in an Edwards Modulyo 4Kfreeze dryer for 24 h, according to the manufacturer’s instruc-tions (Edwards High Vacuum International, Crawley, West Sus-sex, UK). The contaminated dried milk powders were stored in adesiccator. Prior to inoculation, each of the freeze-dried cultureswas ground to a fine powder with a pestle and mortar.

Coconut, black peppercorns, and onion powder were artifi-cially inoculated with the serovars designated in Table 1. Forthese samples, theSalmonellastrain was inoculated into 10 mLBPW, incubated at 37EC for 20–24 h, and then added to 25 g ofeach product in 225 mL BPW. Because these food products donot dissolve, the inoculated suspended products were mixedthoroughly by swirling and incubated at 37EC for 20–24 h. Theexcess supernatant fluid was removed with a sterile 10 mLgraduated pipet, and the contaminated sediment (food product)was transferred to a suspension of the food product prepared byadding 25 g of each product to 225 mL BPW. These cultureswere incubated at 37EC for 20–24 h; the excess liquid was thencarefully removed. The remaining contaminated food productwas transferred to several sterile Petri dishes to give a layer ofnot more than 10 mm of food substrate. The Petri dishes werestored at –80EC for 6–24 h, and then freeze-dried for 24 h. Thecontaminated food products were stored in a desiccator.

A sample of each of the freeze-driedSalmonellacultures,in the various food matrixes, was then used to determine thenumber of viable organisms. This number was used to deter-mine the dilution needed to obtain inoculum levels ofca 1–10 cfu/25 g food matrix. The contaminated dried milk

BOLTON ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 2, 2000 301

Table 2. Summary of media and test protocol for the BAM method and the BIOLINE immunoassay a

Steps in procedures

BAM method BIOLINE assay

Food groups

1 2 3 4 5 All

Preenrichment broth Lactose broth Lactose broth TSB TSB Milk BPW

Selective enrichment broth SCB + TT BRV + TT SCB + TT LST + TT SCB + TT RVM + SCB

ELISA ND ND ND ND ND All products

Subculture agars XLD, HE, BS XLD, HE, BS XLD, HE, BS XLD, HE, BS XLD, HE, BS XLDb, HE, BS

Confirmation TSI, LIA TSI, LIA TSI, LIA TSI, LIA TSI, LIA TSI, LIA

Serology PSO/PSH PSO/PSH PSO/PSH PSO/PSH PSO/PSH PSO/PSH

Biochemistry API 20E API 20E API 20E API 20E API 20E API 20E

a Abbreviations: SCB = selenite crystine broth, TT = tetrathionate, ND = not determined, XLD = xylose lysine desoxycholate agar,HE = Hektoen enteric agar, BS = bismuth sulfite agar, TSI = triple sugar iron agar, LIA = lysine iron agar, PSO = polyvalent serum againstO-antigen, PSH = polyvalent serum against H-antigen, API 20E = biochemical confirmation test from BioMeriéux, BRV = BAMRappaport-Vassiliadis, LST = lauryl tryptose broth, RVM = Rappaport-Vassiliadis medium, TSB = trypticase soy broth, BPW = bufferedpeptone water.

b All ELISA-positive samples were confirmed by subculture.

powder samples and the contaminated food products were di-luted in the appropriate food matrix as a dried mix. These mix-tures were stored in air-tight containers at ambient tempera-ture for 2–3 weeks before testing.

The high moisture food products, fish, raw chicken, porksausage, raw ground beef, shrimp, and frog legs were artifi-cially inoculated with theSalmonellastrains designated in Ta-ble 1. TheSalmonellastrain was grown in BPW at 37EC over-night. The number of viable organisms per gram of each of thefreeze-dried, contaminated food products was determined byinoculating plates of nutrient agar (Oxoid) with a spiral plater(Don Whitley Scientific Ltd., Shipley, West Yorkshire, UK)and incubating them at 37EC for 24 h. The cultures werestored overnight at 4EC and then diluted to 4–40 cfu/mL; 1 mLwas added to 10 sublots of 100 g food product. Each inocu-lated sublot was mixed, and all sublots were combined andmixed again. This constituted the artificially inoculated sam-ple which was stored at 4EC for 2–3 days before testing.

Peanut butter was artificially inoculated with theSalmonellastrain designated in Table 1. TheSalmonellastrain was grown inBPW at 37EC overnight and the viable count was determined.The culture was stored overnight at 4EC and then diluted to4–40 cfu/mL in BPW containing 15% glycerol, and 1 mL wasadded to sublots of 100 g peanut butter. Each inoculated sublotwas mixed, and all sublots were combined and mixed again. Thisconstituted the artificially inoculated sample which was stored atambient temperature for 2–3 weeks before testing.

Comparative Study

Procedure

Ten replicates of each artificially inoculated food sampleand 2 negative control samples were tested by the BAM methodand the BIOLINESalmonellaimmunoassay method accordingto the protocol set by the AOAC–RI. The actual count of eachSalmonellastrain in each of the contaminated food matrixeswas determined on the day of testing by a most probable num-ber (MPN) technique. The MPN method was performed byadding 3 × 100 g food product to each of 3 × 900 mL BPW,3 × 10 g to each 90 mL BPW, 3 × 1 g toeach of 3 × 9 mLBPW,and 3 × 1 mL of a1:10 dilution to each of 3 × 9 mL BPW.These preenrichment cultures were incubated at 35EC for 24 hand then inoculated into enrichment broths, subcultured, andconfirmed by the BAM protocol.

BAM Salmonella Method

Briefly, the BAM method (3) requires that 25 g product ispreenriched in a nonselective broth, which is dependent on thedifferent food types (Table 2). These preenrichment cultureswere incubated for 24 ± 2 h at 35EC. The specified selective en-richment broths were incubated for 24± 2 h at 35°C. These se-lective enrichment cultures were then subcultured to XLD, HE,and BS agars, which were incubated at 35EC for 24 ± 2 h. Fromeach set of cultures, up to 6 presumptive colonies were con-firmed by the serological and biochemical methods specified inthe BAM protocol and further confirmed with the API 20E kit.

302 BOLTON ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 2, 2000

Table 3. Exclusivity study a

No. Organism Source Test concentration Absorbance at 450 nm Presumptive

1 Escherichia coli NCTC 9001 1.1 × 106 0.044/0.048 Negative

2 Proteus rettgeri NCTC 7475 1.9 × 106 0.039/0.049 Negative

3 Yersinia enterocolitca NCTC 10460 1.5 × 106 0.045/0.049 Negative

4 Enterobacter aerogenes NCTC 10006 5.9 × 106 0.054/0.060 Negative

5 Shigella flexneri Human isolate 3.7 × 106 0.032/0.024 Negative

6 Shigella sonnei Human isolate 2.7 × 106 0.028/0.028 Negative

7 Yersinia enterocolitica Environmental 1.1 × 106 0.037/0.036 Negative

8 Escherichia coli 0157 VTEC NCTC 12079 1.7 × 106 0.045/0.028 Negative

9 Proteus mirabilis NCTC 10975 8.9 × 106 0.027/0.028 Negative

10 Enterobacter cloacae NCTC 11936 1.1 × 107 0.170/0.174 Negative

11 Hafnia alvei NCTC 8105 1.0 × 106 0.048/0.062 Negative

12 Hafnia alvei (environmental) Human isolate 9.1 × 106 0.078/0.031 Negative

13 Edwardsiella tarda NCTC 11934 8.6 × 106 0.043/0.029 Negative

14 Edwardsiella tarda NCTC 9750 1.6 × 106 0.033/0.027 Negative

15 Citrobacter freundii NCTC 9750 5.8 × 106 0.068/0.066 Negative

16 Citrobacter freundii Environmental 1.8 × 106 0.063/0.037 Negative

17 Klebsiella aerogenes NCTC 9528 3.3 × 106 0.032/0.04 Negative

18 Klebsiella pneumoniae Human isolate 3.9 × 106 0.037/0.034 Negative

19 Proteus vulgaris Human isolate 3.0 × 106 0.038/0.031 Negative

20 Enterobacter gergoviae Human isolate 7.8 × 106 0.111/0.117 Negative

a Strains were enriched in BPW, diluted 100-fold; cfu/mL determined by spiral plating and absorbance measurement after ELISA test.

BIOLINE Immunoassay Method

(a) Sample preparation.—Tests on 10 replicates of eachspiked food sample and 2 negative control samples wereperformed according to the instructions ofSalmonellaELISAKit from BIOLINE (Vejle, Denmark). A 25 g portion of thesample was homogenized in 225 mL BPW and incubated at37EC for 16–24 h; 0.1 mL of the culture was then added to10 mL RVS broth, which was incubated at 42°C for 20–22 h.About 3 mL of the RVS broth cultures were boiled for15–20 min and allowed to cool to room temperature.

(b) Immunoassay.—From each boiled RVS culture,100 µL was dispensed into microtiter wells of the BIOLINEELISA test kit. A positive control, a negative control, and ablank were included with each batch of tests. The microtiterwells were covered with plastic adhesive film and incubated at37EC for 30 min. Unbound material was removed by washingthe microtiter plates 5–7 times with the dilute washing buffer.A 100 µL portion of Salmonella antibody–horseradishperoxidase conjugate was dispensed into all test and controlwells except the blank. The microtiter wells were coveredwith plastic adhesive film and incubated at 37EC for 30 min.The microtiter plates were washed 5–7 times with dilutedwashing buffer, and then 100 µL monocomponent tetramethylbenzidine (TMB) substrate was added to all wells. After15 min incubation at room temperature, the enzyme reactionwas stopped by adding 100 µL 0.2M sulfuric acid stop solu-tion to each well. The optical densities of the microtiter plateswere read on a microtiter plate reader (Bio-Rad, Hercules,CA, Model 3550) set at a wavelength of 450 nm. All positiveresults, i.e., absorbance readings > 0.200 were confirmed bythe BAM confirmation method (Table 2).

Exclusivity Study

Twenty cultures of non-Salmonellaorganisms (Table 3)were grown in BPW at 37EC overnight and 10-fold serial dilu-tions were prepared from 10–1to 10–5in 10 mL BPW. The via-ble counts were determined by a surface counting method us-ing a spiral plater. The ELISA tests were performed directly induplicate on all dilutions.

Statistical Analysis

Overall results were analyzed by the SAS logistic regres-sion; individual food/inoculum results were determined by theSAS Fisher extract test (SAS, Cary, NC).

Results and Discussion

Table 1 shows the overall results of the inoculation of20 different serotypes into 20 different food and feed ma-trixes. The inoculation count was determined by MPN/25 gand was, in most cases, low. With inclusion of the repeat ex-periment for animal feed with higher inoculation levels, theBAM method indicated 178 presumptive positives, butonly 160 were verified by culture. The BIOLINESalmonellaELISA Test indicated 173 presumptive positives, with168 verified by culture. Thus a statistically significant

4.8% more culturally verified positive samples were found bythe BIOLINE immunoassay. But this difference disappearedif the fish results were omitted. The only food demonstrating astatistically significant difference between the 2 methods wasfish. Five immunoassay-positive samples were not verified byculture. The 5 samples yielded high absorbances in the rangeof 1.000–2.000, strongly indicating that the 5 samples werelikely true positives, but they were not culturally verified forunknown reasons. Because a second reculturing was not per-formed in this study, it is not possible to know the true statusof these 5 samples. Studies at BIOLINE and trial laboratorieson human fecal samples (to be published) have shown bettersensitivities than the culture method when ELISA-positive,but culture-negative, samples were recultured.

Surprisingly, the BAM reference method did not find anypositives for the fish samples. The experiment was repeatedfor this matrix with identical negative results of the BAMmethod. The same reagents and enrichment media were usedfor other food items, ruling out the possibility of errors in thepreparation of the media.

The experiment with frog legs had to be repeated becausethe uninoculated samples showed significant positive re-sponse with the BIOLINESalmonellaELISA Test and subse-quently was culturally confirmed as positive.

Twenty strains from theEnterobacteriaceaegroup weretested from pure cultures in the range of 106–107 cfu/mL.None of the tested strains gave absorbances above the cutoffrange, indicating positivity of the samples (Table 3). TheBIOLINE SalmonellaELISA Test is designed to work withsamples enriched in BPW and subsequently inRappaport-Vassiliadis selective buffer.Salmonellawill ac-tively grow in Rappaport-Vassiliadis buffer, whereas thegrowth of otherEnterobacteriaceaewill be strongly inhibitedin this buffer.

The test kit is designed to react with flagellar proteins. Thereaction pattern is thus not dependent on the serogroup butsolely on the presence of flagellar proteins. Under specifiedconditions, the BIOLINESalmonellaELISA Test yieldedvery high and significant absorbances in the range of 100-foldhigher than typical negative values and significantly higherthan the specified cutoff value of 0.200 (data not presented).This made the test easy and reliable to interpret.

Acknowledgment

The statistical assistance of Stuart L. Chirtel (U.S. Foodand Drug Administration, Center for Food Safety and AppliedNutrition, Washington, DC) is gratefully acknowledged.

References

(1) Blackburn, C. de W. (1993)J. Appl. Bacteriol.75, 199–214

(2) Bolton, E., & Franklin, J. (1995)Int. Food Hyg.6, 11–13

(3) Andrews, W., Bruce, V.R., June, G., Satchell, F., & Sherrod,P. (1992)FDA Bacteriological Analytical Manual, 7th Ed.,AOAC INTERNATIONAL, Arlington, VA

BOLTON ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 2, 2000 303

304 JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 2, 2000