Received 15 October 1997/Returned for modification 20 November 1997/Accepted 12 May 1998

Transcription

1 CLINICAL AND DIAGNOSTIC LABORATORY IMMUNOLOGY, July 1998, p Vol. 5, No X/98/$ Copyright 1998, American Society for Microbiology. All Rights Reserved. Interpretations of Antibody Responses to Salmonella enterica Serotype Enteritidis gm Flagellin in Poultry Flocks Are Enhanced by a Kinetics-Based Enzyme-Linked Immunosorbent Assay PATRICK L. MCDONOUGH, 1 * RICHARD H. JACOBSON, 1 JOHN F. TIMONEY, 2 AHMED MUTALIB, 3 DAVID C. KRADEL, 4 YUNG-FU CHANG, 1 SANG J. SHIN, 1 DONALD H. LEIN, 1 SUSAN TROCK, 5 AND KAYE WHEELER 5 Diagnostic Laboratory, 1 Department of Microbiology, Immunology and Parasitology, 2 and Department of Avian and Aquatic Animal Medicine, 3 College of Veterinary Medicine, Cornell University, Ithaca, New York 14853; Department of Veterinary Science, The Pennsylvania State University, State College, Pennsylvania ; and Veterinary Service, Animal and Plant Health Inspection Service, U.S. Department of Agriculture, Albany, New York Received 15 October 1997/Returned for modification 20 November 1997/Accepted 12 May 1998 Many regulatory and diagnostic programs for the detection of Salmonella enterica serotype Enteritidis infection in commercial poultry flocks have relied on rapid Pullorum agglutination tests to screen birds because of the shared antigens of S. enterica Enteritidis and S. enterica Pullorum and Gallinarum; however, the use of the enzyme-linked immunosorbent assay (ELISA) format affords better analytical sensitivity than crude agglutination tests. In this study, we adapted our earlier conventional indirect ELISA, using gm flagellin as the antigen, to a kinetics-based, computer-controlled ELISA (KELA). The KELA was used to screen for flagellin antibody from three commercial flocks: (i) a large flock involved in a U.S. Department of Agriculture trace back from a human S. enterica Enteritidis foodborne outbreak (n 3,209), (ii) a flock infected with the endemic S. enterica Enteritidis serotype but which also had multiple other salmonella serotypes (n 65), and (iii) an S. enterica Pullorum-infected flock (n 12). The first flock (S. enterica Enteritidis prevalence of 2.45% based on culture) provided a field test of the KELA and allowed the calculation of diagnostic sensitivity (D-Sn) and diagnostic specificity (D-Sp). With a cutoff of 10 (used for screening flocks [i.e., high sensitivity]), the KELA has a D-Sn of 95.2% and a D-Sp of 18.5%; with a cutoff of 140 (used in confirmatory flock testing [i.e., high specificity]), the KELA has a D-Sn of 28.0% and a D-Sp of 99.1%. We found that with a cutoff of 60 (D-Sn 63.1%; D-Sp 91.6%), we could eliminate reactions in the KELA caused by other non-s. enterica Enteritidis salmonellae. The KELA was also compared to two commercial rapid Pullorum tests, the Solvay (D-Sn 94.9%; D-Sp 55.5%) and the Vineland (D-Sn 62.0%; D-Sp 75.3%). Infections in domestic poultry with Salmonella enterica serotype Enteritidis of many phage types have become increasingly important primarily from the public health standpoint due to the vertical transmission of S. enterica Enteritidis in grade A shell eggs (14, 29 31). Since 1990, the National Poultry Improvement Plan has included a provision requiring the mandatory testing of poultry breeder flocks for Salmonella and specifically for S. enterica Enteritidis infection (35). A number of states also require routine surveillance of egg layer flocks and broilers for S. enterica Enteritidis (25, 28). The ability to detect S. enterica Enteritidis infection on a flockwide basis, and thus to reduce the potential of contaminated eggs reaching the consumer, has become a priority for the poultry industry, for veterinary diagnostic laboratories, and for government authorities dealing with foodborne diseases (14). Many diagnostic tests have been developed for the detection * Corresponding author. Mailing address: Diagnostic Laboratory, College of Veterinary Medicine, Cornell University, Ithaca, NY Phone: (607) Fax: (607) plm2@cornell.edu. Present address: The Gluck Equine Research Center, Department of Veterinary Science, University of Kentucky, Lexington, KY Present address: College of Veterinary Medicine, Mississippi State University, Mississippi State, MS Present address: PA Poultry Federation, Harrisburg, PA of salmonella infections in poultry. Antemortem bacterial cultures of cloacal swabs, eggs, and environmental swabs and postmortem cultures of internal tissues are fairly sensitive techniques for S. enterica Enteritidis detection on a flockwide basis (33, 34). However, culture is expensive and time-consuming and also suffers because individual birds excrete S. enterica Enteritidis intermittently or may eliminate the infection altogether. Serology is the other component of field and laboratory testing of flocks to establish their S. enterica Enteritidis status. Because of decreased costs and rapid turnaround time, serology has developed into a promising screening tool for flocks. S. enterica Enteritidis is an invasive serotype, and immunoglobulin G (IgG) responses persist in birds that have been infected with S. enterica Enteritidis. Therefore, serology would be a superior method to culture for screening birds that are intermittently culture positive or that have eliminated S. enterica Enteritidis infection (7 9). In pilot studies, we developed the S. enterica Enteritidis gm flagellin indirect enzyme-linked immunosorbent assay (ELISA) to screen experimentally infected birds and some commercial layer hens (8, 32). gm flagellin is the purified flagellar protein from S. enterica Enteritidis (19). In the present study, our goals were (i) to adapt the gm flagellin indirect ELISA to a computer-controlled and very reproducible kinetics-based ELISA (KELA), (ii) to compare the serological re- 550

2 VOL. 5, 1998 ELISA FOR S. ENTERICA ENTERITIDIS gm FLAGELLIN 551 sponses of birds naturally infected with S. enterica Enteritidis using two commercially available rapid Pullorum agglutination test antigens versus gm flagellin antigen of the KELA, and (iii) to evaluate the gm flagellin KELA in a field trial using sera from a large S. enterica Enteritidis-infected commercial flock involved in a trace back from an outbreak of human infection, sera from a commercial flock infected with multiple salmonella serotypes, and finally sera from a group of birds naturally infected with Salmonella enterica Pullorum. MATERIALS AND METHODS Bacterial strains. S. enterica Enteritidis strain 41 (obtained from M. Opitz [University of Maine] and R. Baker [Cornell University Poultry Science Department]), originally isolated from a laying hen, was used for the production of gm flagellin. This strain had the antigenic formula (1,9,12:gm: ) that is characteristic of S. enterica Enteritidis. Bird sera. Sera from three different groups of birds were tested. The first group consisted of sera from 3,207 commercial layer hens collected by the U.S. Department of Agriculture (USDA) Salmonella Enteretidis (SE) Task Force. This flock was traced as a possible source for a human foodborne salmonella outbreak. The second group of 65 sera came from a Northeastern United States commercial layer flock naturally infected with multiple serotypes of salmonellae including S. enterica Enteritidis (27 birds were chosen that were positive by the rapid Pullorum plate test, and 38 birds were chosen that were negative in the rapid Pullorum plate test). The last group of sera was obtained by the USDA from 12 hens culled from a small commercial layer flock shown to be infected with S. enterica Pullorum. All serum samples were shipped to the laboratory on ice packs and stored at 20 C until tested. Bacterial cultures of organ tissues. For the first group of 3,207 layer hens, the National Veterinary Services Laboratories (NVSL), Veterinary Service, Animal and Plant Health Inspection Service, USDA, Ames, Iowa, cultured the organs of all birds for Salmonella following the USDA SE Task Force protocol, i.e., ovary, oviduct, liver, expressed gall bladder, heart, pericardial sac, but not intestine or cecum (33). For the second group of 65 birds, the Task Force s protocol was followed (33), but in addition the intestinal ceca were added to the list and cultured for salmonellae. For the third group of 12 birds, the NVSL followed the National Poultry Improvement Plan protocol for culture of internal organs of the affected birds (34, 35). Pullorum serological testing. Two commercially available pullorum antigen kits were used to test bird sera, i.e., the S-Solvay-Pullorum Stained Antigen K Polyvalent (Salsbury Laboratories, Charles City, Iowa) and the Vineland Antigen (Vineland Laboratories, Inc., Vineland, N.J.). Testing was performed on glass plates (macroscopic agglutination) and read according to the manufacturers directions. gm flagellin production. The method of Ibrahim et al. (19) was used to produce the S. enterica Enteritidis gm flagellin antigen for the KELA. The purity of the flagellin protein was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (12% polyacrylamide) by the Laemmli method (23) followed by silver staining (Hi-Ho Silver Stain Kit; Accurate Chemical and Scientific Corp., Westbury, N.Y.). The protein concentration of the flagellin was determined by the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, Calif.). Purified flagellin was stored at 70 C until used. KELA. We adapted our original single-read ELISA to the KELA (32). Our laboratory has a long history of using the KELA format for routine bacterial and viral serologies (1, 3, 21, 24). The reagent concentrations were optimized in block titrations in preliminary experiments. Each well of a 96-well microtiter plate (Maxisorb; Nunc, Roskilde, Denmark) was coated with 0.1 ml of a solution of 0.5 g of gm flagellin antigen per ml diluted in carbonate-bicarbonate buffer, shaken for 10 min, incubated for 3hinahumid chamber at 37 C, and then frozen and stored in sealed bags at 70 C until used. On the day of testing, the microtiter plates and their contents were brought to room temperature and wells were washed four times with 0.1 M phosphate-buffered saline (PBS) containing 0.05% Tween 80 (PBST). Test sera (100 l) diluted 1:100 in PBST containing 2% nonfat dry milk and control sera similarly diluted (described below) were placed in triplicate wells, incubated for 1 h at 37 C in a humid chamber, and then washed four times with PBST. One hundred microliters of conjugate (rabbit anti-chicken immunoglobulin G [IgG], heavy and light chain specific), linked to horseradish peroxidase (Cappel Research Products, Durham, N.C.) diluted 1:800, were added to each well and incubated for 30 min at room temperature. Plates were then washed with PBST, followed by the addition of 100 l of substrate solution (3,3,5,5 -tetramethylbenzidine at 0.4 mg/ml, mixed with an equal volume of 0.02% H 2 O 2 in a citric acid buffer) per well. Plates were then read three times at 650 nm at 1-min intervals, starting 1 min after the addition of substrate solution (BioTek EL-312; BioTek Instruments, Inc., Winoski, Vt.) with continuous shaking between reads to detect color development kinetically. The results were calculated by the KELA computer program and expressed as the slope of the reaction rate between enzyme and substrate solution for each sample (3, 4, 20). (Slopes are directly proportional to the amount of antibody bound.) KELA standard control sera. We prepared four control serum pools, each from approximately 10 birds. Each pool was diluted 1:10 in PBS (ph 7.2), aliquoted in 0.5-ml volumes, and stored in sealed vials at 20 C until used. Each pool represented a standard for use in all subsequent runs of the assay. The positive sera originated from birds of a known S. enterica Enteritidis infection status from a commercial flock, and negative sera were from Salmonella-free flocks based on culture. The pools varied in the amount of antibody activity to gm flagellin based on negative, low, medium, and high positive slope values acquired in the KELA. Each standard was run 10 to 15 times in the KELA to determine an average expected activity. Standardization of data. The expected slopes for the standards were used in development of a standard curve for each run of the assay. The standard curve was generated by linear regression of the expected slope values for each standard against the slope values obtained in that run for the respective standards. A normalized slope for each test sample was then derived from the standard curve, allowing direct comparison of data from run to run (3, 20). KELA cutoff. Because the frequency distribution of the organ culture-positive and -negative birds for the first group of birds (flock 1) was not normally distributed, the selection of an optimal cutoff point between positive and negative sera was not intuitively obvious. Diagnostic sensitivity (D-Sn) and specificity (D-Sp) estimates were determined for various cutoffs in the KELA based on the gold standard of 3,128 culture-negative and 79 culture-positive birds (flock 1). A cutoff at the lower end of the scale of KELA slope values gives a high sensitivity estimate that is particularly useful in a screening assay. Conversely, a cutoff at the upper end of the KELA slope scale gives a high specificity estimate that is useful in a confirmatory assay (13). Predictive values. For the first flock of 3,207 birds, the positive predictive value (PPV) and the negative predictive value (NPV) were calculated from the test s sensitivity, specificity, and disease prevalence levels according to the following formulas: PPV (Prev)(D-Sn)/[(Prev)(D-Sn) (1 Prev)(1 D-Sp)] and NPV (1 Prev)(D-Sp)/[(1 Prev)(D-Sp) (Prev)(1 D-Sn)], where Prev is prevalence. RESULTS gm flagellin. Silver staining the sodium dodecyl sulfate-polyacrylamide gels revealed a single flagellin band of about 60 kda. The reported size of gm flagellin is 58.4 kda, and it is made up of polymers of a single protein (19). Several smaller bands were found at 43 and 15 kda which were believed to be either minor dissociated proteins of flagellin (19) or to be outer membrane proteins. No lipopolysaccharide (LPS) band of 100- kda size was found in any of the gels. The protein concentration of the gm flagellin was 750 g/ml. Serology and organ culture. The results of the Solvay and Vineland Pullorum serologies are shown in Table 1 for the first group of 3,223 commercial birds (note that 3,223 birds were tested versus just 3,207 in the KELA because not all sera tested in the Pullorum serologies were available for the KELA testing). The NVSL found that 79 birds were organ culture positive and 3,144 were culture negative, for an overall prevalence of 2.45%; internal organs but not ceca were cultured from these birds. The Solvay assay had a D-Sn and D-Sp of 94.9% and 55.5%, respectively, while the Vineland assay had corresponding results of 62.0 and 75.3%, respectively. For the KELA, multiple cutoffs were evaluated. A cutoff of 10 slope units resulted in a D-Sn of 95.2% and a D-Sp of 18.5%, whereas a cutoff of 140 units resulted in a D-Sn of 28.0% and a D-Sp of 99.1%. Based on the culture results, the prevalence of infection in the flocks tested was 2.45%. For this prevalence, the PPVs and NPVs are given in Table 2. The Pullorum serology for the second group of 65 birds from a breeder flock (flock 2) infected with the endemic serotype S. enterica Enteritidis but which also had other salmonella serotypes revealed that 27 birds were Pullorum serology positive, i.e., 26 birds reacted in the Solvay assay, 12 birds reacted in the Vineland assay, and 11 birds reacted in both assays. Internal organs and ceca were cultured from these birds. The salmonella serotypes recovered along with Pullorum and KELA serologies are shown in Table 3 for culture-positive birds only. Within this group, the slopes of the Pullorum serology-positive

3 552 MCDONOUGH ET AL. CLIN. DIAGN. LAB. IMMUNOL. TABLE 1. Organ culture results compared to Pullorum (Vineland and Solvay) serology for group 1 birds Serology and result Organ culture status (test result) a No. of positive birds No. of negative birds Total no. of birds D-Sn (%) D-Sp (%) Disease prevalence (%) Solvay 94.9 b 55.5 c 2.45 d Positive 75 (TP) 1,399 (FP) 1,474 Negative 4 (FN) 1,745 (TN) 1,749 Total 79 3,144 Vineland 62.0 e 75.3 f 2.45 g Positive 49 (TP) 776 (FP) 825 Negative 30 (FN) 2,367 (TN) 2,397 Total 79 3,143 a Test result abbreviations: TP, true positive; FP, false positive; FN, false negative; TN, true negative. b D-Sn calculated as follows: (75/79) %. c D-Sp calculated as follows: (1,745/3,144) %. d Disease prevalence calculated as follows: (79/3,223) %. e D-Sn calculated as follows: (49/79) %. f D-Sp calculated as follows: (2,367/3,143) %. g Disease prevalence calculated as follows: (79/3,222) birds (n 27) ranged from 0 to 51 (mean 13) and the slopes from the Pullorum-negative birds (n 38) ranged from 0 to 39 (mean 7). The Pullorum and KELA serologies for the third group of 12 birds (flock 3) of unknown culture status in the Pulloruminfected flock (flock 3) are shown in Table 4. DISCUSSION Serological diagnoses of Pullorum disease (S. enterica Pullorum) and fowl typhoid disease (S. enterica Gallinarum) in poultry have utilized agglutination tests, i.e., the serum plate agglutination test (Pullorum test) and macroscopic tube agglutination tests; these screening tests detect predominantly IgM antibody to cell wall (LPS) antigens (10, 38). For the serological screening of other Paratyphoid salmonella infections in poultry, e.g., S. enterica Enteritidis infections, surveillance programs have taken advantage of the fact that S. enterica Pullorum and S. enterica Enteritidis share cell wall antigens (both are found in Salmonella serogroup D and share antigens 9 and 12); the use of the Pullorum rapid serum plate agglutination test to detect cross-reacting IgM antibody in S. enterica Enteritidis-infected flocks was used in early S. enterica Enteritidis surveillance programs (15, 22). However, IgM responses are short-lived compared to IgG responses, and diagnosticians quickly learned that there was poor correlation between the Pullorum serum plate test to cell wall antigens and S. enterica Enteritidis-positive cultures (22). Consequently, there has been controversy over the utility of the Pullorum rapid plate tests for S. enterica Enteritidis diagnosis (7, 12, 15, 22, 29, 38). Titers to S. enterica Enteritidis cell wall antigens may persist for many weeks (8, 16, 26). Salmonella cell wall antigens have the disadvantage of lacking analytical specificity, i.e., they cross-react with antigens shared by related gram-negative bacteria, e.g., the serogroup B S. enterica Typhimurium and the serogroup D S. enterica Pullorum and Gallinarum and S. enterica Enteritidis. In contrast, researchers have shown that the antibody response to salmonella gm flagellin occurs earlier than responses to cell wall antigens (40), peaks early, and within 10 weeks reverts to low levels (2). The advantage of the gm flagellin antigen is that it is shared by no other salmonellae invasive for chickens (i.e., possesses analytical specificity) and is found on only a few salmonella serotypes (Menston, Derby, and Montevideo) that occasionally cause transient intestinal infections (32). Thus, gm flagellin would be useful to differentiate infection by related serogroup B and D infections in poultry (8). Also, gm flagellin is highly antigenic and is readily recovered in relatively pure form from the surfaces of the salmonella bacteria; it would be inexpensive to produce for widespread use. After the rapid Pullorum assays to screen flocks for infection with S. enterica Enteritidis, the next improvement in S. enterica Enteritidis diagnosis was the IgG ELISA (8, 11); ELISAs are inherently more analytically sensitive than macroscopic agglutination tests, such as the Pullorum test for detecting IgG responses. Other researchers have shown that gm flagellin is a useful antigen in the ELISA. van Zijdeveld et al. showed that an indirect gm flagellin ELISA was better than an indirect ELISA, using S. enterica Enteritidis LPS to detect S. enterica Enteritidis-infected chicks and adult hens (36). In contrast to our work, they used only experimentally infected birds and thus had no results available from any field studies; field studies would be needed for D-Sn and D-Sp determinations in a real-world situation. They also tested a blocking ELISA using gm flagellin which had better analytical specificity than the indirect gm flagellin ELISA. A field study in Holland by this group further supported the usefulness of the flagellin ELISA over culture for the detection of S. enterica Enteritidis infection in commercial flocks (37). Using the gm flagellin KELA, we took part in a World Health Organization-sponsored interlaboratory comparison of different ELISA antigens to determine if birds were S. enterica Enteritidis infected (11); specificpathogen-free chicken sera were tested from S. enterica Enteritidis-infected birds having a high antibody titer to S. enterica Enteritidis cell wall antigens, medium S. enterica Enteritidis titer, or specific-pathogen-free, uninfected birds with low titers to S. enterica Enteritidis. Also, sera from S. enterica Enteritidis-infected commercial birds having high S. enterica Enteritidis cell wall titers and uninfected commercial birds with low titers were tested in the gm flagellin KELA. In all cases, the KELA accurately determined the infection history of the birds (11). We used the gm flagellin antigen coupled with a rigorous method in the KELA to normalize data and thus were able to reduce the effects of variables, such as ambient temperature,

4 VOL. 5, 1998 ELISA FOR S. ENTERICA ENTERITIDIS gm FLAGELLIN 553 TABLE 2. D-Sn and D-Sp of the KELA based on the gold standard of organ culture (flock 1) KELA cutoff a D-Sp (%) D-Sn (%) PPV (%) b NPV (%) b a Slope units 1,000. b PPV and NPV based on 2.5% prevalence. instrumentation, and test parameters. We calculated results from a standard curve generated by four serum controls using linear regression analysis. This approach is more precise than that used in most indirect ELISA assays because it does not rely on the use of only a single serum control to which all test sample values are normalized. KELA data are thus more consistent from day to day and from run to run. Table 3 shows the high level of analytical specificity (lack of cross-reactivity) of the gm flagellin KELA compared with the Pullorum serologies. This commercial flock, flock 2, was infected with endemic S. enterica Enteritidis but also had other salmonella serotypes present in the flock. The Heidelberg and Typhimurium serotypes share antigen 12 with S. enterica Enteritidis, as does the nonmotile serotype 4,12: :, but yet birds infected with these serotypes had undetectable levels of gm flagellin response at a cutoff of 60 in the KELA (NPV 98.9% [Table 2]). The one S. enterica Enteritidis culture-positive bird failed to react in the KELA; it is assumed that this Bird TABLE 3. Organ and cecal culture results versus Pullorum and KELA serologies for flock 2 (Salmonella culture positive only) Salmonella serotype a recovered Pullorum serology b Solvay Vineland KELA result c 10 S. enterica Enteritidis 36 Salmonella 4,12: : 37 S. enterica Heidelberg 38 S. enterica Heidelberg 39 S. enterica Heidelberg 41 S. enterica Heidelberg 42 S. enterica Heidelberg 48 S. enterica Heidelberg 54 S. enterica Typhimurium 61 S. enterica Infantis a Antigenic formulas of salmonella serotypes are shown in parentheses: S. enterica Enteritidis, (1,9,12:gm: ); Salmonella spp., (4,12: : ); S. enterica Heidelberg, (1,4,[5], 12:r:1,2); S. enterica Typhimurium, (1,4,[5],12:i:1,2); S. enterica Infantis, (6,7,14:r:1,5). b Solvay Pullorum-stained antigen (Salsbury Laboratories) and Vineland Pullorum-stained antigen (Vineland Laboratories) used. c With a cutoff of 60 KELA slope units (D-Sn 63%; D-Sp 91.6%), all responses were classified as negative due to the high specificity of the KELA. lack of KELA response is related to the stage of infection of the bird, i.e., perhaps the bird had just been exposed to S. enterica Enteritidis and had not yet developed an immune response to the S. enterica Enteritidis gm flagellin. It is also possible that this bird may not have become infected, depending on the S. enterica Enteritidis strain s inherent virulence and infectious dose, etc. Rather, the culture isolation of S. enterica Enteritidis from this bird s cecum may have reflected merely environmental exposure. The Pullorum serology from this bird would appear to somewhat support this latter assumption, i.e., perhaps no Pullorum response reflects no infection, whereas a chronically infected bird would be expected to have developed some long-lasting S. enterica Enteritidis cell wall antibody that would be detectable at least at a low level in the Pullorum assay. The six positive samples found on the Pullorum serologies reflect the lack of D-Sp for these assays. Table 4 records the results of the third flock of birds infected with the endemic Pullorum serotype (of unknown S. enterica Enteritidis status) and tells us more about the analytical specificity of the gm flagellin assay. Birds 3, 9, 11, and 12 all had low Bird TABLE 4. Pullorum and KELA serologies for the S. enterica Pullorum-infected flock 3 Solvay Pullorum serology a Vineland KELA result b a See footnote b of Table 3 for description of Pullorum test antigens. b With a cutoff of 60 KELA slope units (D-Sn 63%; D-Sp 91.6%), all responses except for bird 4 were classified as negative due to the high specificity of the KELA.

5 554 MCDONOUGH ET AL. CLIN. DIAGN. LAB. IMMUNOL. KELA slopes and negative Pullorum serologies, suggesting lack of antibody to gm flagellin in Pullorum-infected birds; with a cutoff of 60 in the KELA, all but one bird (bird 4) would be considered gm flagellin negative. Bird 4 also did not react in the Pullorum test but was positive in the KELA. It is unknown if S. enterica Enteritidis was recovered from bird 4 to account for the positive KELA result. It is also unknown if S. enterica Pullorum was ultimately recovered from every one of the 12 birds at the NVSL. False-negative and false-positive Pullorum test results are problematic (38). The other birds tested with the KELA from this flock all had low KELA slopes in the negative range. The usefulness of the KELA depends on the selection of a cutoff point that is suitable for the intended purpose: a low cutoff leads to a high D-Sn with a low D-Sp (useful for a screening assay), while a high cutoff results in a low D-Sn and a high D-Sp (confirmatory assay). The gm flagellin KELA was designed for use on a flockwide basis (not for individual birds), such as when a flock is involved in a trace back from a human outbreak of foodborne disease. In light of the diagnostic sensitivity and specificity problems of the Pullorum tests, a more sensitive and specific assay was needed for assigning a flock infection status. The gm flagellin KELA with two cutoffs, one for screening (we typically use a slope of 10) and another for confirmation (use a slope of 140), was thus conceived. Also, if one uses a single cutoff of 60 slope units, the KELA gives a better D-Sp (91.6%) than currently is available in the Solvay and Vineland assays (55.5% and 75.3%, respectively). This enhanced specificity results in a predictive value for a positive KELA test result that is two to three times higher than that of the other assays (at a 2.45% prevalence of infection). Table 2 shows that birds with KELA slope units greater than about 200 units have a PPV of 44%, whereas the Pullorum assays have PPVs of from about 5 to 8% (data not shown in Table 1 but can be calculated). The D-Sn of KELA with a cutoff of 10 slope units (95.2%) is about equivalent with that of the Solvay assay (94.9%) but much better than the Vineland assay (62.0%). With this cutoff, the NPV of a KELA test result, with a prevalence of 2.45%, is 100%. With a cutoff of 60, we have shown that the gm flagellin KELA has a greater D-Sp than either of the two commercially available rapid Pullorum slide tests (Tables 1 and 2); the KELA also has a greater D-Sn than the Vineland but not the Solvay Pullorum test (the Solvay test s increased sensitivity may be attributed to its higher false-positive rate); this agrees with earlier work by Timoney et al. (32). It also corroborates the work of Gast and Beard (18) and Mutalib et al. (25) who showed that the Solvay antigen was more sensitive than the Vineland antigen, but the Solvay antigen also had a higher false-positive rate than the Vineland antigen (44.5% versus 24.7% in this study). Even though serology is a quicker method to establish the infection status of a flock, national regulatory agencies usually require bacterial culture to provide the definitive evidence of flock status. Antemortem bacterial cultures of cloacal swabs, eggs, and environmental swabs and postmortem cultures of internal tissues (ovary, oviduct, liver, gall bladder, heart, pericardial sac, and sometimes cecum) are fairly sensitive techniques for S. enterica Enteritidis detection on a flockwide basis (33, 34). S. enterica Enteritidis is often found in extraintestinal sites in the bird, e.g., ovary, oviduct, liver, spleen (5, 27, 29) in addition to the intestinal tract. However, some birds ceca (intestines) are colonized by S. enterica Enteritidis, while other internal tissues (e.g., ovary and oviduct) are not (6, 7, 38). This fact has consequences for regulatory programs such as that conducted by the USDA SE Task Force (begun in 1990 by Veterinary Service, Animal and Plant Health Inspection Service, USDA) whose field protocol does not require the testing of any intestinal or cecal tissues (33). Spent hen surveys from areas of the country where S. enterica Enteritidis is endemic have shown a high prevalence of S. enterica Enteritidis cecal carriage in hens (17, 39). Truly infected birds having antibody to S. enterica Enteritidis but culture negative (by the SE Task Force protocol), could account for some of the relatively high KELA slopes among culture-negative birds. Misclassification of bird infection status slightly reduces the calculated D-Sp of the KELA and also has an impact on the D-Sn. Other possible reasons for negative salmonella cultures are that individual birds excrete S. enterica Enteritidis intermittently or birds eliminate the infection altogether; in all of these cases, culture results may appear negative but bird sera may actually possess antibodies to gm flagellin and to other cell wall antigens of S. enterica Enteritidis. Were the birds classified correctly for infection status, the performance characteristics of the KELA would improve. In summary, the gm flagellin KELA is a useful flock screening tool for use in commercial poultry flocks to detect S. enterica Enteritidis infection. The antigen is relatively easy to produce, and the KELA format provides excellent within-run and between-run reproducible testing. The antigen also confers improved analytical specificity to serology, making it useful to distinguish S. enterica Enteritidis from other salmonella infections. With the use of a competitive KELA format, perhaps that analytical specificity can be enhanced even further. The gm flagellin KELA will be a useful complement for state flock surveillance and testing programs and for screening flocks involved in trace backs from human foodborne outbreaks. ACKNOWLEDGMENTS This work was funded by USDA/CSRS grant number awarded to P.L.M. We thank Linda Hendrichs and Rebecca Franklin for excellent technical support during this project. REFERENCES 1. Appel, M. J. G., S. Allan, R. H. Jacobson, T. L. Lauderdale, Y. F. Chang, S. J. Shin, J. W. Thomford, R. J. Todhunter, and B. A. Summers Experimental Lyme disease in dogs produces arthritis and persistent infection. J. Infect. Dis. 167: Baay, M. F., and J. H. J. Huis int Veld Alternative antigens reduce cross-reactions in an ELISA for the detection of Salmonella enteritidis in poultry. J. Appl. Bacteriol. 74: Barlough, J. E., R. H. Jacobson, D. R. Downing, T. J. Lynch, and F. W. Scott The kinetics-based enzyme-linked immunosorbent assay for coronavirus antibodies in cats: calibration to the indirect immunofluorescence assay and computerized standardization of results through normalization to control values. Can. J. Vet. Res. 51: Barlough, J. E., R. H. Jacobson, G. P. Sorresso, T. J. Lynch, and F. W. Scott Coronavirus antibody detection in cats by computer-assisted kineticsbased enzyme-linked immunosorbent assay (KELA): field studies. Cornell Vet. 76: Barnhart, H. M., D. W. Dreesen, and J. L. Burke Isolation of Salmonella from ovaries and oviducts from whole carcasses of spent hens. Avian Dis. 37: Barrow, P. A Experimental infection of chickens with Salmonella enteritidis. Avian Pathol. 20: Barrow, P. A Further observations on the serological response to experimental Salmonella typhimurium in chickens measured by ELISA. Epidemiol. Infect. 108: Barrow, P. A ELISAs and the serological analysis of salmonella infections in poultry: a review. Epidemiol. Infect. 109: Barrow, P. A Use of ELISAs for monitoring Salmonella in poultry. Vet. Rec. 134: Barrow, P. A., A. Berchieri, Jr., and O. Al-Haddad Serological response of chickens to infection with Salmonella gallinarum-s. pullorum detected by enzyme-linked immunosorbent assay. Avian Dis. 36: Barrow, P. A., M. Desmidt, R. Ducatelle, M. Guittet, H. M. J. F. vander

6 VOL. 5, 1998 ELISA FOR S. ENTERICA ENTERITIDIS gm FLAGELLIN 555 Heijden, P. S. Holt, J. H. J. Huis int Velt, P. McDonough, K. V. Nagaraja, R. E. Porter, K. Proux, F. Sisak, C. Staak, G. Steinbach, C. J. Thorns, C. Wray, and F. van Zijderveld World Health Organization-supervised interlaboratory comparison of ELISAs for the serological detection of Salmonella enterica serotype Enteritidis in chickens. Epidemiol. Infect. 117: Barrow, P. A., and M. A. Lovell Experimental infection of egg-laying hens with Salmonella enteritidis phage type 4. Avian Pathol. 20: Bech-Nielson, S., J. B. Jorgensen, P. Ahrens, and N. C. Feld Diagnostic accuracy of a Mycobacterium phlei-absorbed serum enzyme-linked immunosorbent assay for diagnosis of bovine paratuberculosis in dairy cows. J. Clin. Microbiol. 30: Centers for Disease Control and Prevention Outbreaks of Salmonella serotype Enteritidis infection associated with consumption of raw shell eggs United States, Morbid. Mortal. Weekly Rep. 45: Chart, H., B. Rowe, A. Baskerville, and T. J. Humphrey Serological tests for Salmonella enteritidis in chickens. Vet. Rec. 126: Cooper, G. L., R. A. Nicholas, and C. D. Racewell Serological and bacteriological investigations of chickens from flocks naturally infected with Salmonella enteritidis. Vet Rec. 125: Dreesen, D. W., H. M. Barnhart, J. L. Burke, T. Chen, and D. C. Johnson Frequency of Salmonella enteritidis and other salmonellae in the caeca of spent hens at time of slaughter. Avian Dis. 36: Gast, R. K., and C. W. Beard Serologic detection of experimental Salmonella enteritidis infections in laying hens. Avian Dis. 34: Ibrahim, G. F., G. H. Fleet, M. J. Lyons, and R. A. Walker Method for the isolation of highly purified salmonella flagellins. J. Clin. Microbiol. 22: Jacobson, R. H., and D. R. Downing (ed.) KELA. Acquisition, management, and analysis of ELISA data. Cornell University Research Foundation, Ithaca, N.Y. 21. Jacobson, R. H., D. R. Downing, and T. J. Lynch Computer-assisted enzyme immunoassays and simplified immunofluorescence assays: applications for the diagnostic laboratory and the veterinarian s office. J. Am. Vet. Med. Assoc. 181: Kim, C. J., K. V. Nagaraja, and B. S. Pomeroy Enzyme-linked immunosorbent assay for the detection of Salmonella enteritidis infection in chickens. Am. J. Vet. Res. 52: Laemmli, U. K Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227: McDonough, P. L., R. H. Jacobson, and J. F. Timoney Virulence determinants of Salmonella typhimurium. Am. J. Vet. Res. 50: Mutalib, A., P. McDonough, S. Shin, V. Patten, and D. Lein Salmonella enteritidis in commercial layer farms in New York state: environmental survey results and significance of available monitoring tests. J. Vet. Diagn. Invest. 4: Nicholas, R. A. J., and G. A. Cullen Development and application of an ELISA for detecting antibodies to Salmonella enteritidis in chicken flocks. Vet. Rec. 128: Poppe, C., R. P. Johnson, C. M. Forsberg, and R. J. Irwin Salmonella enteritidis and other Salmonella in laying hens and eggs from flocks with Salmonella in their environment. Can. J. Vet. Res. 56: Schlosser, W. D., D. J. Henzler, J. Mason, D. C. Kradel, and S. Hurd The Salmonella enteritidis pilot project and the Pennsylvania egg quality assurance program, p In Proceedings of the 98th Annual Conference of the U.S. Animal Health Association. U. S. Animal Health Association, Grand Rapids, Mich. 29. Shivaprasad, H. L., J. F. Timoney, S. Morales, B. Lucio, and R. C. Baker Pathogenesis of Salmonella enteritidis infection in laying chickens. I. Studies on egg transmission, clinical signs, fecal shedding, and serological responses. Avian Dis. 34: St. Louis, M. E., D. L. Morse, M. E. Potter, T. M. DeMelfi, J. J. Guzewich, R. V. Tauxe, and P. A. Blake The emergence of grade A shell eggs as a major source of Salmonella enteritidis infections. JAMA 259: Timoney, J. F., H. L. Shivaprasad, R. C. Baker, and B. Rowe Egg transmission after infection of hens with Salmonella enteritidis phage type 4. Vet. Rec. 125: Timoney, J. F., N. Sikora, H. L. Shivaprasad, and M. Opitz Detection of antibody to Salmonella enteritidis by a gm-flagellin-based ELISA. Vet. Rec. 127: U.S. Department of Agriculture Necropsy, p In Salmonella enteritidis field manual. SE Task Force, Animal and Plant Health Inspection Service, U.S. Department of Agriculture, Hyattsville, Md. 34. U.S. Department of Agriculture Procedures for collecting environmental samples and cloacal swabs for bacteriological examination, part 147, subpart B, , p In National Poultry Improvement Plan and Auxiliary Provisions. Animal and Plant Health Inspection Service, U.S. Department of Agriculture, Washington, D.C. 35. U.S. Department of Agriculture U.S. S. enteritidis monitored, part 145, subpart B, (d), p In National Poultry Improvement Plan and Auxiliary Provisions: special provisions for egg-type chicken breeding flocks and products. Animal and Plant Health Inspection Service, U.S. Department of Agriculture, Washington, D.C. 36. van Zijderveld, F. G., A. M. van Zijderveld-van Bemmel, and J. Anakotta Comparison of four different enzyme-linked immunosorbent assays for serological diagnosis of Salmonella enteritidis infections in experimentally infected chickens. J. Clin. Microbiol. 30: van Zijderveld, F. G., A. M. van Zijderveld-van Bemmel, R. A. M. Brouwers, T. S. de Vries, W. J. M. Landman, and W. A. de Jong Serological detection of chicken flocks naturally infected with Salmonella enteritidis, using an enzyme-linked immunosorbent assay based on monoclonal antibodies against the flagellar antigen. Vet. Q. 15: Waltman, W. D., and A. M. Horne Isolation of Salmonella from chickens reacting in the pullorum-typhoid agglutination test. Avian Dis. 37: Waltman, W. D., A. M. Horne, C. Pirkle, and D. C. Johnson Prevalence of Salmonella enteritidis in spent hens. Avian Dis. 36: Wray, C., R. A. Nicholas, C. J. Thorns, and U. M. Minga Studies on the serological diagnosis of Salmonella enteritidis infection in poultry, p In Proceedings of the symposium on the diagnosis and control of salmonellosis. U.S. Animal Health Association, San Diego, Calif.