OF CLINICAL MICROBIOLOGY, Dec. 1979, 0095-1137/79/12-0791/05$02.00/0
JOURNAL
Vol. 10, No. 6
p. 791-795
Evaluation of a Ganglioside Immunosorbent Assay for Detection of Escherichia coli Heat-Labile Enterotoxin BACK,3* ANN-MARI SVENNERHOLM,2 JAN HOLMGREN,2 AND ROLAND MOLLBY3 Department of Infectious Diseases, Karolinska Institutet, Roslagstull Hospital, S-114 89 Stockholm, Sweden'; Institute of Medical Microbiology, Guldhedsgatan 10, S-413 46 Goteborg, Sweden2; and Department of Bacteriology, National Bacteriological Laboratory, S-105 21 Stockholm, Sweden3
ERIK
Received for publication 10 September 1979
The GM1 ganglioside enzyme-linked immunosorbent assay (GM1-ELISA), an immunological method for detection of Escherichia coli heat-labile enterotoxin (LT), was quantitatively and qualitatively compared with the conventional adrenal cell test for the identification of LT-producing strains. A micromodification model of the assay was developed. Enterotoxin preparations from 120 E. coli isolates from individuals with diarrhea, which had been previously shown to be enterotoxigenic by the adrenal cell test, and from 44 control strains of E. coli were compared in parallel by the two methods. Quantitatively the covariation of the enterotoxin titers was highly significant (Rs = 0.98, P < 0.001), the GM1ELISA being somewhat more sensitive than the adrenal cell test. The methodological error was less than 5% in both tests. Qualitatively the overall agreement for positive and negative reactions for the two methods was 89%. The GM1ELISA is practical for routine use in the diagnosis of enterotoxigenic E. coli, especially in laboratories without facilities for cell culture. Production of heat-labile enterotoxin (LT) by Escherichia coli has mainly been assayed on cultures of mouse Y1 adrenal cells (YMAc) or Chinese hamster ovary cells (3, 9). These methods require equipment and experience in handling cell cultures and interpretation of the results (14), which make them less suitable for smaller laboratories. Efforts have, therefore, been put into developing immunological methods for detection of E. coli LT (2, 5, 8, 15). However, these methods have been hampered by the lack of purified E. coli enterotoxin, which is why they have had to depend on indirect means of identifying LT. A GM1 ganglioside enzyme-linked immunosorbent assay (GM1ELISA) method, using the ganglioside GM1 as a specific sorbent for enterotoxin, has been developed as an alternative approach (13). In this study, GM1-ELISA was compared with the Y1 adrenal cell test for detection of E. coli LT, using culture supernatants from a large number of previously enterotoxigenic E. coli isolates representing more than 25 different serotypes and presumptive non-enterotoxigenic control strains. The aim was to evaluate the usefulness of the GM1-ELISA for routine detection of enterotoxigenic E. coli. MATERIALS AND METHODS Bacterial strains. One hundred sixty-four strains of E. coli were studied. Of these, 120 were isolated
from fecal specimens either of Swedish patients with diarrhea, in principle as described by Back et al. (1), or of Ethiopian children with diarrhea. The Ethiopian strains were collected as part of a research project in Addis Ababa, and details on these strains will be published elsewhere (E. Back, R. Mollby, B. Kaijser, S. Stintzing, T. Wadstrom, and D. Habte, manuscript in preparation). All of these strains had been, by means of the adrenal cell test, previously shown to produce LT on at least two occasions. The strains had been kept as stab cultures in 1% nutrient broth agar at room temperature for up to 1 year since their isolation. The strains had been kindly serotyped by B. Kaijser, Goteborg (11), and represented more than 25 different O groups. Twenty-nine freshly isolated E. coli strains collected from Swedish persons who had not been abroad or been on antibiotic therapy recently and who had normal stools and no abdominal complaints were used as presumptive non-enterotoxigenic controls. In addition, 15 urinary isolates of E. coli were used as non-enterotoxigenic control strains. Cultivation. A loopful of bacteria from a 5% horse blood agar plate was transferred to each tube with 10 ml of a medium containing 20 g of Casamino Acids (Difco Laboratories, Detroit, Mich.), 6 g of yeast extract (Difco), 2.5 g of NaCl, 8.7 g of K2HPO4, 50 mg of MgSO4, 50 mg of MnCl2, and 5 mg of FeCl3 per liter (pH 8.5) (7). The strains were grown on a shaker at 37°C overnight. The tubes were centrifuged at 4,000 x g for 20 min, and the supernatants were stored at -20°C as two samples until use (within
JOURNAL
Vol. 10, No. 6
p. 791-795
Evaluation of a Ganglioside Immunosorbent Assay for Detection of Escherichia coli Heat-Labile Enterotoxin BACK,3* ANN-MARI SVENNERHOLM,2 JAN HOLMGREN,2 AND ROLAND MOLLBY3 Department of Infectious Diseases, Karolinska Institutet, Roslagstull Hospital, S-114 89 Stockholm, Sweden'; Institute of Medical Microbiology, Guldhedsgatan 10, S-413 46 Goteborg, Sweden2; and Department of Bacteriology, National Bacteriological Laboratory, S-105 21 Stockholm, Sweden3
ERIK
Received for publication 10 September 1979
The GM1 ganglioside enzyme-linked immunosorbent assay (GM1-ELISA), an immunological method for detection of Escherichia coli heat-labile enterotoxin (LT), was quantitatively and qualitatively compared with the conventional adrenal cell test for the identification of LT-producing strains. A micromodification model of the assay was developed. Enterotoxin preparations from 120 E. coli isolates from individuals with diarrhea, which had been previously shown to be enterotoxigenic by the adrenal cell test, and from 44 control strains of E. coli were compared in parallel by the two methods. Quantitatively the covariation of the enterotoxin titers was highly significant (Rs = 0.98, P < 0.001), the GM1ELISA being somewhat more sensitive than the adrenal cell test. The methodological error was less than 5% in both tests. Qualitatively the overall agreement for positive and negative reactions for the two methods was 89%. The GM1ELISA is practical for routine use in the diagnosis of enterotoxigenic E. coli, especially in laboratories without facilities for cell culture. Production of heat-labile enterotoxin (LT) by Escherichia coli has mainly been assayed on cultures of mouse Y1 adrenal cells (YMAc) or Chinese hamster ovary cells (3, 9). These methods require equipment and experience in handling cell cultures and interpretation of the results (14), which make them less suitable for smaller laboratories. Efforts have, therefore, been put into developing immunological methods for detection of E. coli LT (2, 5, 8, 15). However, these methods have been hampered by the lack of purified E. coli enterotoxin, which is why they have had to depend on indirect means of identifying LT. A GM1 ganglioside enzyme-linked immunosorbent assay (GM1ELISA) method, using the ganglioside GM1 as a specific sorbent for enterotoxin, has been developed as an alternative approach (13). In this study, GM1-ELISA was compared with the Y1 adrenal cell test for detection of E. coli LT, using culture supernatants from a large number of previously enterotoxigenic E. coli isolates representing more than 25 different serotypes and presumptive non-enterotoxigenic control strains. The aim was to evaluate the usefulness of the GM1-ELISA for routine detection of enterotoxigenic E. coli. MATERIALS AND METHODS Bacterial strains. One hundred sixty-four strains of E. coli were studied. Of these, 120 were isolated
from fecal specimens either of Swedish patients with diarrhea, in principle as described by Back et al. (1), or of Ethiopian children with diarrhea. The Ethiopian strains were collected as part of a research project in Addis Ababa, and details on these strains will be published elsewhere (E. Back, R. Mollby, B. Kaijser, S. Stintzing, T. Wadstrom, and D. Habte, manuscript in preparation). All of these strains had been, by means of the adrenal cell test, previously shown to produce LT on at least two occasions. The strains had been kept as stab cultures in 1% nutrient broth agar at room temperature for up to 1 year since their isolation. The strains had been kindly serotyped by B. Kaijser, Goteborg (11), and represented more than 25 different O groups. Twenty-nine freshly isolated E. coli strains collected from Swedish persons who had not been abroad or been on antibiotic therapy recently and who had normal stools and no abdominal complaints were used as presumptive non-enterotoxigenic controls. In addition, 15 urinary isolates of E. coli were used as non-enterotoxigenic control strains. Cultivation. A loopful of bacteria from a 5% horse blood agar plate was transferred to each tube with 10 ml of a medium containing 20 g of Casamino Acids (Difco Laboratories, Detroit, Mich.), 6 g of yeast extract (Difco), 2.5 g of NaCl, 8.7 g of K2HPO4, 50 mg of MgSO4, 50 mg of MnCl2, and 5 mg of FeCl3 per liter (pH 8.5) (7). The strains were grown on a shaker at 37°C overnight. The tubes were centrifuged at 4,000 x g for 20 min, and the supernatants were stored at -20°C as two samples until use (within
