Vol. 44, No. 6APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 1982, p. 1349-13550099-2240/82/121349-07$02.00/0Copyright © 1982, American Society for Microbiology

Enzyme-Linked Immunosorbent Assay for Detection ofStaphylococcal Enterotoxins in Foods

ROBERT C. FREED, MARY L. EVENSON, RAOUL F. REISER, AND MERLIN S. BERGDOLL*

Food Research Institute, University of Wisconsin-Madison, Madison, Wisconsin 53706

Received 26 May 1982/Accepted 11 August 1982

An enzyme-linked immunosorbent assay (ELISA) was developed for detectionof staphylococcal enterotoxins in foods. The "double-antibody sandwich" proto-col combines parts of several procedures reported previously. Horseradishperoxidase was conjugated to antibody specific for an enterotoxin, and theantibody-enzyme conjugate was assayed with a 2,2'-azino-di-(3-ethylbenzthiazo-line sulfonic acid)-H202 substrate solution. Enterotoxins were added to a varietyof foods that were representative of those implicated in staphylococcal foodpoisoning outbreaks. Extracts of the foods were assayed by the ELISA andradioimmunoassay. Enterotoxin levels below 1 ng/g of food were consistentlydetectable by the ELISA. These results compared favorably with those of theradioimmunoassay. Experiments confirmed the interference of protein A indouble-antibody sandwich ELISAs. Although protein A interference has not beendemonstrated to be a problem in food extracts, we suggest a screen for protein Ainterference in which immunoglobulin G from nonimmunized rabbits is used. Allof the known staphylococcal enterotoxins could be detected by this method.Analysis of a food product for entertoxin by the ELISA can be completed in an 8-h working day.

Many methods have been developed for thedetection of staphylococcal enterotoxins infoods. The most widely used is the extraction-concentration-microslide method. This involvesan immunodiffusion detection technique and issensitive to 100 ng of enterotoxin per 100 g ofsample (3, 13). A method of this type is currentlyin use in the U.S. Food and Drug AdministrationLaboratories (3). An extract from a 100-g foodsample must be concentrated to 0.2 ml in orderto detect 1 ng of enterotoxin per g of sample.The minimum amount of enterotoxin requiredfor development offood poisoning is consideredto be 100 ng. This is a cumbersome and time-consuming process, requiring 3 to 6 days tocomplete.

Several radioimmunoassay (RIA) methodshave been proposed for the detection of entero-toxins in foods (1). These methods are equal insensitivity to the extraction-concentration-mi-croslide method but require only 1 to 2 days for afood analysis. They involve simple extractionand require only a small sample offood. The useof RIA is limited because of the need for radio-active materials, which require government li-censing. These methods require purified entero-toxins, which are not generally available.The enzyme-linked immunosorbent assay

(ELISA), also referred to as enzyme immunoas-say (5, 17), has been developed for the detection

of enterotoxins (5, 7, 8, 10, 12, 17-19). TheELISA can be completed as quickly as can theRIA and is as sensitive but does not require theuse of radioactive materials. Two types ofELISA methods have been proposed. In the"double-antibody sandwich" method (12, 17),the enzyme is coupled to the specific antibody,whereas in the competitive methods (5, 7, 8, 10,18, 19), it is coupled to the enterotoxin. Thecoupling of the enzyme to the antibody elimi-nates the need for purified enterotoxins. Thismethod also results in absorbance readings thatrelate directly to the amount of enterotoxin inthe samples.The ELISA method reported here is a solid-

phase double-antibody sandwich procedure inwhich a combination of parts of previously re-ported methods are used. Horseradish peroxi-dase is coupled to antibody specific for anenterotoxin. The substrate is H202 in solutionwith 2,2'-azino-di-(3-ethylbenzthiazoline sulfon-ic acid) (ABTS). Polystyrene balls and microti-ter plates are used as solid-phase supports. Thismethod is sensitive to .1 ng of enterotoxin per gin a variety of foods. An analysis of a foodproduct can be completed in 1 working day.

MATERIALS AND METHODSPurified enterotoidns and antisera. The purified en-

terotoxins and antisera were prepared in this labora-tory. The immunoglobulin G (IgG) fractions of the

1349

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

APPL. ENVIRON. MICROBIOL.

antisera were purified by ammonium sulfate precipita-tion (4), lyophilized, and stored under vacuum at 0°C.The IgG was redissolved in 0.01 M sodium phosphate-buffered saline (PBS) at pH 7.2 and further purified bygel filtration through a column of Sephadex G-200(Pharmacia Fine Chemicals, Piscataway, N.J.). Thepurified IgG fractions were combined and dialyzed(three changes) against 0.01 M sodium carbonate buff-er (Na2CO3-NaHCO3) at pH 9.6. The purified IgG wasconcentrated by ultrafiltration (model 75 Ultrafiltra-tion Device; Amicon Corp., Lexington, Mass.) with aPTGC 043 10 filter (Millipore Corp., Bedford, Mass.),frozen with liquid nitrogen and stored at 0°C (-70°Cfor long-term storage).Antibody-enzyme conjugate. Horseradish peroxidase

(type IV; Sigma Chemical Co., St. Louis, Mo.) wasconjugated to IgG by a method similar to that ofNakane and Kawaoi (11). After NaBH4 treatment ofthe antibody-enzyme conjugate, the solution was dilut-ed with 0.01 M PBS and dialyzed by ultrafiltration withmodel 75 stir cell equipped with an XM 100A filter(both from Amicon) to remove unreacted horseradishperoxidase and NaBH4. This process was repeatedthree times, and the conjugate was taken up in 3 ml ofPBS, passed through a 0.2-p.m sterilizing filter (Acro-disc; Gelman Sciences, Inc., Ann Arbor, Mich.), andstored at 40C for daily use or lyophilized and stored at00C.

Solid-phase coating. Polystyrene balls (6.5 mm; specfinish; Precision Plastic Ball Co., Chicago, Ill.) werewashed by agitation in deionized water for severalhours. The same process was repeated in 0.01 Msodium carbonate buffer at pH 9.6. The washed ballswere agitated in 10 jig of IgG per ml of carbonatebuffer at room temperature (20 to 25°C) overnight. Thecoated balls were placed in 0.01 M PBS containing0.1% Tween 20 (PBS-TWN) and agitated for 30 min.They were stored in PBS-TWN.Each well of nonprocessed, nonsterile, flat-bottom

Linbro polystyrene microtiter plates (Flow Labora-tories, Inc., McLean, Va.) was coated with 50 R1 of anIgG-carbonate buffer solution (10 ,ug/ml, pH 9.6), andthe plates were incubated at 4°C for 3 h or overnight.The coating solution was removed, and each well waswashed twice with PBS-TWN, filled with the samebuffer, and allowed to stand for 30 min before use.

Preparation of food extracts. Food extracts wereprepared by a procedure similar to that describedpreviously (9). Solid food was ground to a homoge-neous slurry with an Omnimixer (DuPont Instruments,Newton, Conn.). We added 1 ml of water per g offood.When the slurry was very viscous, 1.5 ml of water wasadded per g of food. The pH of the slurry was adjustedto 4.5 with 6 N HCI. Each slurry was centrifuged for 20min at 20,000 x g and 4°C. The pH of the supernatantwas adjusted to 7.5 with 5 N NaOH. Chloroform (1ml/10 ml of extract) was added, and the extract wasstirred for 3 to 5 min, centrifuged as described above,and filtered through water-saturated Miracloth (Chico-pee Mills, Inc., New York) to remove the chloroform.Enterotoxin was added to foods before extraction togenerate positive controls. Extracts of foods to whichno enterotoxin was added were used as blanks ornegative controls. Serial dilutions of enterotoxin add-ed to negative controls were used as standards. Theextracts were assayed by RIA in duplicate. No extrac-tion was required for ELISAs of milk. Tween 20

(0.5%) was added to extracts and milk assayed byELISA. Triplicate samples were run.ELISA procedure. We placed 1-ml test solutions in

culture tubes (18 by 150 mm), each of which containeda polystyrene ball coated with antibody specific for anenterotoxin. Each tube was agitated for up to 2 h.Each ball was washed in 4 ml of PBS-TWN in a secondtube, placed in a third tube with 50 p.1 of diluteantibody-enzyme conjugate in PBS-TWN, and gentlyagitated for up to 1 h. Each ball was washed again andplaced in 2 ml of ABTS-H202 substrate solution (0.6nM ABTS [Sigma], 1.2 nM H202, 0.05 M citric acidbuffer, pH 4.0) (16). After the addition of 1 ml ofstopping solution (0.2 M HF, 0.012 M NaOH, 2.6 nMEDTA) (16), the absorbance of the reaction mixturewas determined at 414 nm with standard cuvettes in amodel 25 spectrophotometer (Beckman Instruments,Inc., Fullerton, Calif.).The same protocol was used for microtiter plate

assays. We added 100 ,ul of the sample to eachantibody-coated well and agitated the plates for up to 2h. The plates were washed with PBS-TWN and agitat-ed for 1 h after 50 p1 of dilute antibody-enzymeconjugate was added to each well. The plates werewashed, and 100 p.1 of ABTS-H202 solution was addedto each well. The plates were agitated for 20 minbefore the enzyme reaction was quenched by theaddition of 50 p.1 of stopping solution to each well. Theabsorbance was read at 410 nm with a model MR-590Microelisa Minireader (Dynatech Laboratories, Inc.,Alexandria, Va.).

Serial dilutions of enterotoxin solutions in PBS-TWN were generated for use as standards. A bufferblank or negative control was assayed to account forany nonspecific color reaction. Absorbance readingsof all replicates were averaged, and the average blankvalues were subtracted. The residual values wereplotted against enterotoxin concentrations, using alogit logarithm transformation (14).RIA. The standard curve for the RIA was generated

by assaying a solution containing the following: 1.0 mlof RIA buffer (0.5% bovine serum albumin, 0.01 Mphosphate buffer, 0.10% NaN3, pH 7.6), 100 p.1 ofantiserum (dilution that binds 50% of the labeledenterotoxin in the absence of unlabeled enterotoxin),0.10 ml of unlabeled enterotoxins (0.0, 6.25, 12.5, 25.0,50.0, and 100.0 ng/ml), and 0.9 ml of extract from thenegative control. Nonspecific binding and inhibitionwere estimated by assaying tubes containing no unla-beled enterotoxin and no antiserum. Another standardcurve was generated, substituting 0.9 ml of RIA bufferfor the food extract. Positive control samples (knownamounts of enterotoxin added before extraction) wereassayed as described above. Duplicates were incubat-ed in polystyrene tubes (10 by 75 mm) at 4°C over-night. The assays were completed as described previ-ously (9).

Statistical evaluation of data. Mean values for allELISA replicates were calculated, and mean blankvalues were subtracted. The remaining values forstandards and enterotoxin extractions were linearized,using a logit logarithm transformation (14) where the Yvalue was logit absorbance (log absorbance/1-absor-bance) and X was logarithm concentration. Least-squares regression analysis of the transformed datawas done with a computer (Minitab Program, Univer-sity of Toledo version PDP-11 81.1, 1981; Brief 5

1350 FREED ET AL.

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

DETECTION OF ENTEROTOXINS IN FOODS 1351

regression analysis, Pennsylvania State University,University Park). This analysis produces a regressionequation standard deviation of Y about the regressionline, a predicted Y value generated for every measuredY value, and a standard deviation for the predictedvalue. The standard deviation for the predicted valuewas multiplied by the appropriate t value (2), and theproduct was subtracted from the predicted Y value tocalculate the lower 95% confidence limit for anyparticular point. If the lower confidence limit wasgreater than the Y value for an absorbance of 0.01, thepoint was considered significant. Although data pointsare usually determined to be significantly differentfrom zero, 0.01 was used since it is impossible tocalculate the logarithm of zero. Standard deviationsfor predicted Y values, based on regression analysis,were generated for concentration levels down to 0.1 ngof enterotoxin per ml. The detection limit was estab-lished as the lowest enterotoxin concentration with apredicted Y value significantly different from the Yvalue for an absorbance of 0.01.The amount of enterotoxin detected in positive

controls was determined by comparison of the absor-bance values to the regression line generated from thestandards. A corresponding enterotoxin concentrationwas determined for each absorbance value and multi-plied by the volume of the extract (over the grams offood extracted) to determine the amount of enterotox-in extracted.

Protein A interference. Serial dilutions of protein Ain PBS-TWN were assayed with anti-staphylococcalenterotoxin B (SEB) antibody reagents in microtiterplates as follows: (i) a rabbit antibody coating andrabbit antibody-enzyme conjugate, (ii) a sheep anti-body coating and rabbit antibody-enzyme conjugate,(iii) a rabbit antibody coating and sheep antibody-enzyme conjugate, and (iv) a sheep antibody coatingand sheep antibody-enzyme conjugate. The assayswere completed in triplicate. Rabbit antibody reagentsspecific for the other enterotoxins were also used toassay for protein A. The IgG fractions of sera fromnonimmunized rabbits were isolated as describedabove and used to coat microtiter plates. Serial dilu-tions of protein A and all of the known enterotoxinswere assayed in these plates with appropriate anti-body-enzyme conjugates.

RESULTSEffect of sample incubation time on enterotoxin

analysis. Samples (1 ml) containing 0.0, 0.63,1.25 and 2.5 ng of SEA per ml of PBS-TWNwere incubated with coated balls for 0.5, 0.75,1.0, 1.5, and 2.0 h (Fig. 1). We used 2-h incuba-tions for subsequent ELISAs. The results ofincubating veal patty extracts for 2 and 18 h withpolystyrene balls and in microtiter plates weregiven in Table 1.

Effect of extract volume on enterotoxin analy-sis. Solutions (0.5, 1.0, 2.0, 3.0, and 5.0 ml)containing 0.0, 0.3, 0.6, 1.0, and 3.0 ng of SEAper ml of PBS-TWN were incubated for 2 h(Table 2). On the basis of the results, we chose 1ml as the sample volume for subsequent analy-sis.

0.700.60 -

0.50E 0.40 > ,2 hrs.-'1 A1_ .5 hr.Z 0.30 -1 hr.

0.20 -0.75 hr.Z -0.50 hr.co 0.10

lr~~~~OCNRAINl4SA/i

0

40.01.-0.03-

Q02002 0.312 0J625 1.25 2.5 51)

CONCENTRATION (ng SEA/ml)

FIG. 1. Effect of varying sample incubation times.Logit logarithm plots of absorbance versus concentra-tion for ELISAs of SEA standards in PBS-TWNincubated with balls for 0.5, 0.75, 1.0, 1.5, and 2.0 hare shown.

Enzyme-substrate reaction time. Various con-centrations of SEA were assayed by the ballsystem. The enzyme-substrate reaction wasquenched at various intervals (Fig. 2). The reac-tion rate was constant for 35 to 45 min. A similarstudy was conducted with microtiter plates.Readings were taken periodically without theaddition of stopping solution. The reaction ratewas constant for 25 to 30 min. In subsequentELISAs, 25 min was allowed for the enzyme-substrate reaction in microtiter plates, and 30min was allowed in the ball system.

Analysis of foods with known amounts of en-terotoxin added. The results of ELISAs andRIAs of foods to which known amounts ofenterotoxin were added are given in Table 3.The amounts of enterotoxin detected were cal-culated from the standard curve. The resultsobtained for ham are given in Table 4. Less than1 ng of enterotoxin per g of food was detectablein all foods tested by the ELISA. For each food,detection limits generated from regression anal-ysis of standards and enterotoxin extractions,along with the lowest amount detected in eachcase, are presented in Table 5.

Effect of protein A on the double-antibodysandwich ELISA. The results from ELISAs ofserial dilutions of protein A with rabbit andsheep anti-SEB reagents in the combinationsdescribed above are presented in Fig. 3. Rabbitantibody-enzyme conjugate and correspondingcoating with IgG specific for enterotoxins otherthan SEB gave results for protein A similar tothose shown in Fig. 3. The same reaction withprotein A was obtained when, instead of entero-toxin-specific IgG, IgG from nonimmunized rab-bits was used for coating the plates. All of theenterotoxins were tested with IgG from nonim-munized rabbits, and none gave a positive reac-tion.

VOL. 44, 1982

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

TABLE 1. Detection of SEA per gram of veal patties: effect of incubation timeBall system with incubation of: Microtiter plate system with incubation of:

Amt ofSEA 2 ha 18 hb 2 hc 18 hdadded (ng)

A414e SEA (ng) A414f SEA (ng) A410' SEA (ng) A410h SEA (ng)

0.63 0.028 0.54 0.086 0.64 0.000 0.00 0.000 0.001.25 0.062 1.24 0.138 1.19 0.000 0.00 0.040 0.432.50 0.073 1.47 0.209 2.15 0.000 0.00 0.090 0.965.00 0.146 3.26 0.279 3.38 0.020 2.25 0.260 3.11

a Detection limit, 0.50 ng.b Detection limit, 0.10 ng.c No analysis based on one data point was possible.d Detection limit, 1.25 ng.e A414, Absorbance at 414 nm; a blank of 0.666 was subtracted.f A blank of 0.120 was subtracted.g A blank of 0.130 was subtracted.h A blank of 0.150 was subtracted.

Detection of enterotoxin in ham implicated in afood poisoning outbreak. The analysis of hamimplicated in a food poisoning outbreak showedthe presence of SEA. This was the only entero-toxin produced by the organisms isolated fromthe ham. The extracts gave negative resultswhen balls coated with IgG from nonimmunizedrabbits were used to screen for protein A inter-ference. The enterotoxin was detectable by RIAand by ELISA with either balls or microtiterplates.

DISCUSSIONIn this investigation, foods were analyzed for

staphylococcal enterotoxins by ELISAs witheither polystyrene balls (19) or microtiter platesas the solid-phase support. The microtiter platesystem is more efficient if many assays are to berun routinely, provided a multichannel pipetterand a microelisa plate reader can be obtained. Inmany laboratories, this equipment is not avail-able, and the number of assays to be done maynot justify its cost. Saunders and Barlett (17)pipetted the reaction product from microtiter

.300

E

.200-

L

10 20 30 40 50 60 70 80ENZYME-SUBSTRATE REACTION TIME (MIN.)

FIG. 2. Enzyme-substrate reaction. Absorbance isplotted against enzyme-substrate reaction time forELISAs (ball system) of various SEA concentrations.

plates into tubes and diluted the product 10-fold.This is time-consuming and increases the possi-bilities of error. With the ball system, the absor-bance can be read with any colorimetric devicewithout dilution. Visual evaluation of results ispossible in both systems when qualitative resultsare sufficient. The ball system is more sensitivethan the microtiter plate system even when theplates have been incubated with extracts for 18 h(Table 1). The ball system is therefore recom-mended when greater sensitivity is desired.We incubated 1 ml samples of extract for 2 h,

using the ball system. Although the samplevolume had virtually no effect on absorbancevalues (Table 2), the absorbance values for the1-ml samples were slightly higher than the val-ues for the other sample sizes, and 1 ml wasselected as the sample volume for this proce-dure. If samples are incubated for longer peri-ods, lower detection levels are possible (Table1), but incubation periods longer than 2 h makeit difficult to complete a food analysis in one 8-hworking day. The 2-h incubation period wasconsidered satisfactory because the required de-tection level of 1 ng of enterotoxin per g of foodis easily achieved.The rate of the enzyme-substrate reaction

TABLE 2. Effect of sample volume on recovery ofSEA with polystyrene balls

Sample A414 with enterotoxin concn (ng/ml) of a:vol (ml) 0.3 0.6 1.0 3.0

0.5 0.031 0.076 0.130 0.3571.0 0.040 0.087 0.153 0.4072.0 0.043 0.076 0.121 0.3253.0 0.042 0.074 0.127 0.3255.0 0.029 0.068 0.115 0.302a Average absorbance at 414 nm (A414) minus a

blank of 0.016.

1352 FREED ET AL. APPL. ENVIRON. MICROBIOL.

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

DETECTION OF ENTEROTOXINS IN FOODS 1353

TABLE 3. Detection of staphylococcal enterotoxinsin foods

AmtEntrotxin Amt detectable

FOOd EnterXadded added (ng/g) by:(ng/g) ELISA RIA

Cheese SEA 1.25 0.59SED 0.63 0.15

1.25 0.64SEE 0.63 0.38

1.25 0.93

Cheese food SEA 0.63 0.36 0.001.25 1.01 0.53

Genoa sausage SEA 0.63 0.36 0.031.25 0.34 1.13

Veal patties SEA 0.63 0.54 0.001.25 1.24 0.54

Hard-boiled eggs SEB 0.63 0.24 0.541.25 0.44 1.34

Potato salad SEB 0.63 0.18 0.001.25 0.44 0.17

Spaghetti SEA 0.63 0.54 0.391.25 0.96 0.91

Ham SEA 0.63 0.341.25 0.55

Milk SEA 0.631.25

remained constant for 35 to 45 min in the ballsystem (Fig. 2). When the ball procedure wasused in the investigation, 30 min was the timeallowed; however, a 40-min incubation period isrecommended because this resulted in a 30%increase in color intensity (Fig. 2).The foods selected for this study are repre-

sentative of those frequently involved in food

poisoning outbreaks. The amounts of enterotox-ins that can be detected after extractions fromvarious foods are shown in Table 4. The ELISAball system compared favorably with the RIA inthe amount of enterotoxin that could be detect-ed.The detection limits calculated from the stan-

dard curve and from the positive controls foreach food extract were consistently below the 1-ng/g level (Table 5). When analyzing a sample offood suspected of being implicated in a foodpoisoning outbreak, it is important to use acontrol as similar to the suspect food as possi-ble. One then generates the standard curvewhich is best for predicting the presence or

absence of enterotoxin. Positive controls towhich known amounts of enterotoxin have beenadded also must be analyzed to determine theeffectiveness of this combination of extractionand detection methods.The double-antibody sandwich ELISA has

been shown to give false-positive results whenprotein A is present in high concentrations (12).This was confirmed when concentrations as lowas 31 ng of protein A per ml gave a positivereaction when rabbit antibody reagents wereused (Fig. 3). This problem can be resolved bytreating the IgG with pepsin to remove the Fcportion of the molecule to which the protein Abinds (12). Unfortunately, this procedure is la-borious and results in a loss of sensitivity.Another approach is to use antibodies preparedin sheep because sheep IgG has a low affinity forprotein A (6; S. Notermans, personal communi-cation). A comparison of the protein A-sheepantibody reagent reaction with the protein A-rabbit antibody reagent reaction (Fig. 3) con-firms this; however, sheep antibodies specificfor enterotoxins are not widely available. Pro-tein A interference has been observed to be aproblem in the examination of culture superna-tants but has not been demQnstrated to be aproblem in the examination of food extracts.

TABLE 4. Analysis of SEA in ham and ham extract with the ball systemA414b

SEAadded Added to ham extract replicate: Added to ham replicate:(ng)a 1 2 3 Avg Netc 1 2 3 Avg Net'0.00 0.039 0.040 0.039 0.0390.31 0.076 0.071 0.075 0.074 0.035 0.058 0.052 0.054 0.055 0.0160.63 0.118 0.115 0.100 0.111 0.072 0.071 0.078 0.087 0.079 0.0401.25 0.184 0.179 0.168 0.177 0.138 0.095 0.112 0.100 0.102 0.0632.50 0.255 0.272 0.295 0.274 0.235 0.150 0.158 0.155 0.154 0.1155.00 0.404 0.367 0.386 0.386 0.347 0.252 0.238 0.241 0.244 0.205

a For ham extract, amount added per milliliter of extract; for ham before extraction, amount added per gram ofham.

b A414, Absorbance of 414 nm.c Average minus blank.

VOL. 44, 1982

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

TABLE 5. Levels of detection of staphylococcal enterotoxin in foodsStandards (in food extract)a Enterotoxin extractionsb

Food Enterotoxin Detection Lowest amt Detection Lowest amtadded limit detected limit detected

(ng/ml) (ng/ml) (ng/g) (ng/g)

Cheese SEA 0.20c 0.31c 0.40a 0.63aCheese SED 0.20 0.31 d 0.63dCheese SEE 0.20 0.31 0.60e 0.63eCheese food SEA 0.35 0.63 0.63 0.63Genoa sausage SEA 0.20 0.31 0.56 0.63Ham SEA 0.20' 0.31c 0.30' 0.31'Hard-boiled eggs SEB 0.10 0.31 0.30 0.63Milk SEAf 0.10 0.16Milk SECf 0.309 0.31Milk SECf.h 0.30' 0.63'Pepperoni SEA 0.25 0.25Potato salad SEB 0.10 0.31 0.22 0.63Spaghetti SEA 0.15 0.31 0.15 0.63Veal patties with soy extender SEA 0.30 0.31 0.50 0.63

Six values and four degrees of freedom.b Four values and two degrees of freedom.Seven values and five degrees of freedom.

d -, Only two values; no regression is meaningful.e Three values and one degree of freedom.f Enterotoxin added directly to milk; no extraction.g Eight values and six degrees of freedom.h Assays were performed in microtiter plates.i Five values and three degrees of freedom.

Simpler methods are available for the detectionof enterotoxigenic strains (15). If the ELISA isto be used, 10- to 50-fold dilutions of culturesupernatants can be assayed with sufficient sen-sitivity to classify strains as enterotoxigenic.This assay was designed for analysis of foodproducts. A special effort to solve this problemis not warranted because protein A has not beendemonstrated to be present in food extracts inlarge enough concentrations to cause interfer-ence. A control in which IgG from a nonimmu-nized rabbit is used was incorporated into theanalysis to screen for interfering levels of pro-

E 1.000c

w .750

z

cocr- .500-0

v)co

.250F

5 10 20 31.25 62.5 125 250 500 1000

PROTEIN A CONCENTRATION (ng /ml)

FIG. 3. Protein A interference in the double-anti-body sandwich ELISA with rabbit IgG and sheep IgGreagents. Absorbance is plotted against logarithm ofprotein A concentration.

tein A. If interference does occur, it will beapparent, and the protein A can be removed bytreatment of the extract with IgG from a nonim-munized rabbit.The ELISA presented in this report was a

fast, reliable, and sensitive method for the detec-tion of staphylococcal enterotoxins in a varietyof foods. In addition, the method is simple andrequires equipment that is standard in mostlaboratories. These advantages make the ELISAthe best available method for analysis of foodsfor staphylococcal enterotoxins.

ACKNOWLEDGMENTSWe thank Ruth N. Robbins and Boniface Aliu for the

preparation of antibodies specific for the staphylococcal en-terotoxins; S. Notermans for the sheep antibody reagents andadvice; and Andy Kirsch, College of Agricultural and LifeSciences Computing Service, for assistance with the statisticalwork.

This research was suported by the College of Agriculturaland Life Sciences, University of Wisconsin-Madison; theAmerican Meat Institute; Public Health Service training grantEF 07015 from the National Institute of Environmental HealthSciences; and various companies and associations of the foodindustries.

LITERATURE CITED1. Bergdoll, M. S., and R. Reiser. 1980. Application of

radioimmunoassay for detection of staphylococcal entero-toxins in foods. J. Food. Prot. 43:68-72.

2. Bhattacharyya, G. K., and R. A. Johnson. 1977. Percent-age points of t distribution, p. 599. In G. K. Bhattachar-yya and R. A. Johnson (ed.), Statistical concepts andmethods. J. Wiley & Sons, Inc., New York.

* RABBIT COATING-RABBIT CONJUGATE

O SHEEP COATING-RABBIT CONJUGATE

A RABBIT COATING-SHEEP CONJUGATE

A SHEEP COATING-SHEEP CONJUGATE -

i~~~~~~~~~

1354 FREED ET AL. APPL. ENVIRON. MICROBIOL.

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from

DETECTION OF ENTEROTOXINS IN FOODS 1355

3. Casman, E. P., and R. W. Bennett. 1965. Detection ofstaphylococcal enterotoxins in food. Appi. Microbiol.13:181-189.

4. Herbert, G. A., P. L. Pelham, and B. Pittman. 1973.Determination of the optimal ammonium sulfate concen-tration for the fractionation of rabbit, sheep, horse, andgoat antisera. AppI. Microbiol. 25:26-36.

5. Kauffman, P. E. 1980. Enzyme immunoassay for staphy-lococcal enterotoxin A. J. Assoc. Off. Anal. Chem.63:1138-1143.

6. Kessler, S. W. 1976. Cell membrane antigen solution withstaphylococcal protein A-antibody adsorbent. J. Immu-nol. 117:1482-1490.

7. Koper, J. W., A. M. Hagenaars, and S. Notermans. 1980.Prevention of cross-reactions in the enzyme linked im-munosorbent assay (ELISA) for the detection of Staphy-lococcus aureus enterotoxin type B in culture filtrates andfoods. J. Food Saf. 2:35-45.

8. Kuo, J. K. S., and G. J. Silverman. 1980. Application ofenzyme-linked immunosorbant assay for detection ofstaphylococcal enterotoxins in foods. J. Food Prot.43:404-407.

9. Miller, B. A., R. F. Reiser, and M. S. Bergdoll. 1978.Detection of staphylococcal enterotoxins A, B, C, D, andE in foods by radioimmunoassay, using staphylococcalcells containing protein A as immunosorbent. Appl. Envi-ron. Microbiol. 36:421-426.

10. Morita, T. N., and M. J. Woodburn. 1978. Homogenousenzyme immune assay for staphylococcal enterotoxintype B. Infect. Immun. 21:666-668.

11. Nakane, P. K., and A. Kawaoi. 1974. Peroxidase-labeledantibody. A new method of conjugation. J. Histochem.

Cytochem. 22:1084-1091.12. Notermans, S., H. L. Verjans, J. Bol, and M. van Schoth-

orst. 1978. Enzyme linked immunosorbent assay (ELISA)for determination of Staphylococcus aureus enterotoxintype B. Health Lab. Sci. 15:28-31.

13. Reiser, R., D. Conaway, and M. S. Bergdoll. 1974. Detec-tion of staphylococcal enterotoxin in foods. Appl. Micro-biol. 27:83-85.

14. Ritchie, D. G., J. M. Nickerson, and G. M. Fuller. 1981.Two simple programs for the analysis of data from en-zyme-linked immunosorbent (ELISA) assays on a pro-grammable desk-top calculator. Anal. Biochem. 110:281-290.

15. Robbins, R., S. Gould, and M. Bergdoll. 1974. Detectingthe enterotoxigenicity of Staphylococcus aureus strains.Appl. Microbiol. 28:946-950.

16. Saunders, G. C. 1979. The art of solid enzyme immunoas-say including selected protocols, p. 98-119. In R. M.Nakamurs, R. W. Dito, and E. S. Tucker (ed.), Immuno-assays in the clinical laboratory. Alan R. Liss, Inc., NewYork.

17. Saunders, G. C., and M. L. Bartlett. 1977. Double-anti-body solid-phase enzyme immunoassay for the detectionof staphylococcal enterotoxin A. Appl. Environ. Microbi-ol. 34:518-522.

18. Simon, E., and G. Terplan. 1977. Nachweis von staphylo-kokken Enterotoxin B mittels ELISA-test. Zentralbl.Veterinaermed. Reihe B 24:842-844.

19. Stiffier-Rosenberg, G., and H. Fey. 1978. Simple assay forstaphylococcal enterotoxins A, B, and C: modification ofenzyme-linked immunosorbent assay. J. Clin. Microbiol.8:473-479.

VOL . 44, 1982

on Novem

ber 18, 2020 by guesthttp://aem

.asm.org/

Dow

nloaded from