INFECrION AND IMMUNITY, Sept. 1977, p. 541-545 Copyright 0 1977 American Society for Microbiology
Vol. 17, No. 3 Printed in U.S.A.
Microtiter Solid-Phase Radioimmunoassay for Detection of Escherichia coli Heat-Labile Enterotoxin HARRY B. GREENBERG,`* DAVID A. SACK,2 WILLIAM RODRIGUEZ,3 R BRADLEY SACK 2 RICHARD G. WYATT,1 ANTHONY R. KALICA,1 ROBERT L. HORSWOOD,1 ROBERT M. CHANOCK,1 AND ALBERT Z. KAPIKIAN'
Laboratory of Infectious Diseases, National Institute ofAllergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20014'; Johns Hopkins Hospital, Baltimore, Maryland 212102; and Children's Hospital National Medical Center of the District of Columbia, Washington, D.C. 200103
Received for publication 21 April 1977
The development of a microtiter solid-phase radioimmunoassay for the detection of Vibrio cholerae enterotoxin and heat-labile Escherichia coli enterotoxin is described. The test is based on the immunological similarity between V. cholerae toxin and E. coli heat-labile toxin. The assay is easy to perforn, quantitative, and at least as sensitive and specific as the Y-1 adrenal cell system. In recent years enterotoxigenic strains of Escherichia coli have been shown to be an important cause of acute diarrheal disease in both humans and animals (16). These strains elaborate at least two toxins, both of which appear to be implicated in disease production. One form of toxin is heat-stable, nonantigenic, and of low molecular weight (13). The other toxin is heat-labile (LT) (8), has a molecular weight of greater than 30,000 (8), is antigenic, and appears to act through stimulation of the adenyl cyclase system (16). Neither toxin has been completely purified (1, 4, 7, 8). The LT E. coli enterotoxin is related immunologically to the enterotoxin of Vibrio cholerae (18). Several workers have shown that antitoxin against V. cholerae enterotoxin will neutralize not only the homologous cholera enterotoxin but also the E. coli LT (12). The immunological similarity between the two toxins is great enough to allow partial purification of LT with a V. cholerae antitoxin affinity column (1). Until very recently only bioassay systems were available to. measure E. coli LT enterotoxin. These systems include the rabbit ileal loop (2), infant rabbit (9), rabbit skin (6), mouse Y-1 adrenal cells (13), and Chinese hamster ovary cell assays (10). In addition a serological test for detecting E. coli LT involving inhibition of immune hemolysis has recently been described (5). At present, tissue culture assays are the most widely used screening procedures for detection of E. coli LT. While these assays appear to be specific and sensitive, they are often difficult to perform, expensive, and not easily quantitated. We have used the immunological cross-reactivity of E. coli LT and V. cholerae
toxin to develop a solid-phase radioimmunoassay for E. coli enterotoxin. MATERLALS AND METHODS Reagents. High-titer burro antiserum to V. cholerae enterotoxin (kindly provided by John B. Robbins, Bureau of Biologics, Food and Drug Administration) was produced as previously described (1). Immunoglobulin G was purified from this serum by a combination of ammonium sulfate precipitation followed by diethylaminoethyl-celiulose chromatography (14). Purified cholera toxin was purchased from Schwarz/ Mann (Orangeburg, N. Y.). Clinical specimens. E. coli colonies were isolated from stool specimens from Peace Corps volunteers in Kenya as previously described (D. Sack et al., Johns Hopkins Med. J., in press). Similarly, E. coli and other coliforms were isolated from children with diarrheal disease seen at the Children's Hospital National Medical Center, Washington, D.C. Unless otherwise stated, randomly selected colonies were grown in syncaseglucose or Trypticase soy broth. After a 24-h growth period, the cultures were centrifuged and the supernatants were assayed for toxin activity in the Y-1 adrenal cell assay (15). The same supernatant fluids were also tested by radioimmunoassay (RIA). Lyophilized dialyzed filtrates from broth cultures of toxigenic E. coli that had been isolated in various parts of the world were also tested by RIA. These filtrates had been tested previously for toxin by a rabbit ileal loop assay (17). RIA reagents. Purified burro immunoglobulin G anti-cholera toxin was labeled with '25I (ca. 20 LCi/4g of protein) by a modification of the Hunter-Greenwood method (14). Chromatography on Sephadex G75 was used to separate unreacted "5I from labeled protein. The labeled immunoglobulin was diluted 1:1 in fetal calf serum and stored at 4°C. RIA test. A solid-phase microtiter RIA test was used. The test was similar to those used for detecting 541
542
INFECT. IMMUN.
GREENBERG ET AL.
hepatitis B surface antigen (14) and human rotavirus (13a). Briefly, polyvinyl microtiter plates (Cooke) were precoated with burro anti-cholera serum (100 gid per well) diluted in phosphate-buffered saline (PBS), pH 7.4. The optimal dilution of antiserum (1:2,000 in this instance) was determined by checkerboard titration. After incubating the plates at room temperature for 12 h, the wells were washed five times with PBS and then filled with PBS containing 1% bovine serum albumin (BSA) and again incubated for 12 h. The PBS-BSA was removed, and the plates were washed twice with PBS. Fifty microliters of the appropriate antigen, which consisted of E. coli culture supernatant, dilutions of lyophilized E. coli culture, or purified cholera toxin diluted in PBS with 1% BSA, was added to the wells and allowed to incubate overnight at room temperature. The wells were then washed five times with PBS, and 50 pl of l"I-labeled burro anti-cholera toxin was added to each well. After a 4-h incubation at 370C, the plates were again washed five times with PBS and cut up with a scissors, and the separate wells were placed in gamma-counting tubes for analysis. Results were expressed as a ratio of residual counts in the sample well to the mean of residual counts in wells that received either PBS-BSA or sterile syncase-glucose broth (positive/negative, P/ N). A P/N ratio of 1.7 or greater was considered
25
Z
a. a
j
15
S
18
I
1/125
1525
1/25
1/125
SERIAL DILUTION OF EscA
c
LT
1/5
11
STANDARD
FIG. 2. Titration of crude broth supernatant from a 24-h culture of a toxigenic E. coli strain. P/N ratio represents counts per minute of '25I-labeled immunoglobulin G burro anti-cholera toxin bound to the solid phase of an antigen-containing well divided by counts per minute bound in a weU containing broth.
E. coli toxin had not yet been purified, it was not possible to quantitate the sensitivity of the assay system in terms of the absolute amount of
LT that could be detected. However, it should be noted that the P/N ratios in this heterologous system were considerably lower than those obtained with cholera toxin in an homologous positive. system. The RIA reactivity of both toxins was reduced by greater than 99% when they were RESULTS incubated at 1000C for 20 min prior to testing. When serial 100-fold dilutions of purified Seven dialyzed lyophilized broth filtrates of cholera toxin were used as test antigen, a typical toxigenic strains of E. coli isolated in various sigmoidal binding curve was observed (Fig. 1). parts of the world were assayed quantitatively Of interest was the sensitivity of the technique. in both the rabbit ileal loop test and RIA (Fig. Toxin was detected at the 0.01-pg level. Serial 3). Both assays were done with serial dilutions, fivefold dilutions of a standard crude broth su- and both were based on the dry weight of the pernatant of a potent toxigenic E. coli strain crude lyophilized filtrate. There was a signifiwere tested in a similar manner (Fig. 2). Again, cant rank-order correlation between the two a typical binding curve was obtained. Because systems (Spearman rank correlation coefficient 1.0, P < 0.001). The RIA was approximately 100fold more sensitive than the rabbit ileal loop, and, thus, the RIA was roughly equivalent to the sensitivity of the cell culture systems (16). Three hundred and fifty-nine separate broth supernatants from 38 Peace Corps volunteers in Kenya and 4 other patients with diarrhea were tested simultaneously by both the Y-1 adrenal cell assay and RIA. The RIA was performed under code (Fig. 4). The correlation between the results of the two tests was extremely high (V = +0.9607, Kendall coefficient of association; X2= 331.4; P < 0.0001). False positives were not encountered, but three false negatives were found. The adrenal cell assay-positive orga100 nisms were derived from selected E. coli cololDtERA TOXI CMNNTRATAN nies isolated from 15 of the 42 patients. FIG. 1. Titration of purified cholera toxin by miForty-seven pediatric patients with acute crotiter solid-phase RIA: P/N ratio represents diarrhea seen at Children's Hospital National counts per minute of '25I-labeled immunoglobulin G burro anti-cholera toxin bound to the solid phase of Medical Center, Washington, D.C., were studied for evidence of LT toxin-producing organisms. an antigen-containing well divided by counts per minute bound in a well containing PBS and 1% BSA. The results of both the Y-1 adrenal cell assay IIDIp/Al
0p,m
a "W
..g/d
RIA FOR E. COLI LT
VOi. 17, 1977
and RIA are seen in Table 1. The correlation between results of the two tests was highly significant (V = +0.9306, Kendall coefficient of association; x2 = 40.7; P < 0.0001). In only one patient studied did the results of the two assays differ. In this case the RIA was repeatedly positive, whereas the adrenal cell assay repeatedly yielded equivocal results (reported as negative in Table 1). As in the tests performed on specimens from Peace Corps volunteers, the RIA was performed under code. The sensitivity of the Y-1 adrenal cell assay and RIA was compared by testing serial fivefold dilutions of five positive E. coli broths. The results of this experiment are seen in Table 2. The assays appeared to be roughly comparable in sensitivity.
-
A series of experiments was done to define the best growth conditions for producing RIAdetectable E. coli LT enterotoxin (Table 3). Generally, the best medium appeared to be syncase-glucose broth, and brain heart infusion broth was definitely inferior. Of interest was the finding that 48-h broth cultures that were grown in a microtiter "mini"-plate and that were not filtered or centrifuged appeared to have more toxin activity than culture supematants. Thus far in preliminary studies it does not appear that the use of such broth specimens containing whole organisms alters the specificity of the test. In addition, it appears feasible to actually grow the E. coli cultures in a microtiter plate that TABLE 1. Comparison of detection of E. coli LT by RIA and an adrenal cell assay No. of patients tested by RIA with indicated
a
m..
I*_
543
No. of patients tested by adrenal
cell assay with indicated result
result
Positive
Negative
Positive Negative
8 0
1 38
TABLE 2. Comparison of sensitivity of RLA and Y-1 adrenal ceU assay for detection of E. coli LT
a*I
I 1
E. coli isolate F
TITE A
Strain
Reciprocal titera of toxin activity in: RIA
fygI
FIG. 3. Comparison of toxin activity of seven broth filtrates from various toxigenic E. coli isolates by the rabbit ikal loop assay and RIA. Titers were determined by serial dilutions and based on the dry weight of the lyophilized broth filtrates. The country of orWgin of the filtrate is indicated in parentheses.
Adrenal cell
125 125 1 1 337cl 10407 625 125 5 25 349c5 25 5 375c4 a Serial fivefold dilutions of broth supernatants. 1 2 3 4 5
408-3
ADRENAL CELL CULTURE ASSAY
POSITIVE
NEGATIVE
iE
a
S
FIG. 4. Comparison of RIA and adrenal cell assay for detection of E. coli LT enterotoxin in broth filtrates. Broth filtrates were derived from E. coli isolated from Peace Corps volunteers in Kenya and other patients with diarrhea. The RIA P/N values were calculated using an N value that was the counts per minute bound by wells inoculated with sterile broth.
544
INFECT. IMMUN.
GREENBERG ET AL.
TABLE 3. P/N ratios observed in RIA for E. coli LT under indicated cultural conditions Culture conditions
Organism (strain) 1 (337c0) 2 (408-3) 3 (10407) 4 (375c0) 5 (349c0) a
Roller rum for 24 ha Flask shaking for 24 Flask stationary for 24 Flask stationary for 48 Mini-plate stationha ha ha ary for 48 hb
Sc 8.2 9.6 5.8 7.0
Td 3.0 13 14
9.2
0.9 3.6 4.9 3.2
6.2
3.8
2.0
Be
S 14 14 18 11 16
T
B
S
T
10 23 23 10 9.2
3.5 9.0 6.0 5.0
3.6 13 5.8 23
0.9 7.7 20 28
4.6
6.8
2.2
B 1
3.5 6.6 15 1.0
S
T
B
S
T
B
6.0 5.4 37 8.1
1.4 15 13 10 3.5
1.3 4.0 7.4 7.7
21 28 74 33 24
20 40 42 46
9 18 28
62
1.8
13
28 8.2
Supernatant fluid tested in assay.
b Whole c S, Glucose d
culture, organisms, and broth were tested in assay. syncase. T, Trypticase soy broth with 0.6% yeast extract. e B, Brain heart infusion.
has previously been precoated with anti-cholera The radioimmunoassay should aid in the purification of E. coli LT since large numbers of toxin, thereby simplifying the assay further. fractions derived from various separation procedures can be tested and quantitative data obDISCUSSION tained. The use of RIA may offer an additional Recent studies have clearly demonstrated advantage in that immunological activity may that certain strains of E. coli produce LT enter- be preserved better than biological activity durotoxin that is very similar both immunologically ing purification procedures such as affinity coland physiologically to the enterotoxin produced umn chromatography. Preliminary studies indiby V. cholerae. Taking advantage of this immu- cate that the RIA can be used for detection and nological relationship, we devised a RIA for de- quantitation of antibodies for LT in human setection of E. coli LT that used antibody raised rum. Finally, it should be noted that this type of against highly purified cholera toxin. Our find- solid-phase RIA also provides an extremely senings indicate that LT toxin can be reliably de- sitive assay system for the detection and quantitected by a solid-phase RIA. The specificity and tation of V. cholerae enterotoxin (S. J. Cryz et sensitivity of the RIA are equivalent to or al., Program Abstr. Intersci. Conf. Antimicrob. greater than the Y-1 adrenal cell assay, one of Agents Chemother. 16th, Chicago, Ill., Abstr. the most sensitive bioassay systems currently no. 367, 1976). available to identify toxin production. HyperimLITERATURE CITED mune monospecific antiserum to E. coli LT, 1. Dafni, Z., and J. B. Robbins. 1976. Purification of heatlabile enterotoxin from Escherichia coli 078:H11 by when available, should increase the sensitivity affinity chromatography with antisera to Vibrio choland specificity of this assay further. In data not erae toxin. J. Infect. Dis. 133:S138-S141. shown we found that non-toxin-producing enter- 2. De, S. W., K. Bhattacharya, and S. Sarkar. 1956. A study of the pathogenicity of strains of bacterium coli. opathogenic serotypes and 12 E. coli isolates J. Pathol. Bacteriol. 71:201-209. producing only heat-stable toxin were unreacS. T., H. W. Moon, and S. C. Wipp. 1974. Detive in this test. This would be expected since 3. Donta, tection of heat labile Escherichia coli enterotoxin with the antibody was raised against highly purified the use of adrenal cells in tissue culture. Science cholera toxin. 183:334-336. There are several advantages of the RIA over 4. Dorner, F., H. Jaksche, and W. Stockl. 1976. Escherichia coli enterotoxin: purification partial characteriexisting detection systems. The RIA is easy to zation and immunological observation. J. Infect. Dis. standardize and automate. Problems with cell 133:S142-S156. culture contamination and variability are 5. Evans, D. J., and D. G. Evans. 1977. Inhibition of immune hemolysis: a serological assay for heat-labile enavoided, and the results obtained are quantitaterotoxin of Escherichia coli. J. Clin. Microbiol. tive. Whether this system can be used to detect 5:100-105. toxin directly in stool or intestinal wash speci- 6. Evans, D. J., D. G. Evans, and S. L. Gorbach. 1973. mens remains to be seen. In addition, since the Production of vascular permeability factor by enterotoxigenic Escherichia coli isolated from man. Infect. assay is based on a specific immunological reacImmun. 8:725-730. tion rather than a biological effect, it can be D. J., D. G. Evans, S. H. Richardson, and S. used more reliably to evaluate possible V. chol- 7. Evans, L Gorbach. 1976. Purification of polymixin-released erae-like LT production in non-E. coli coliforms heat labile enterotoxin of Escherichia coli. J. Infect. (11). Dis. 133:S97-S102.
VOL. 17, 1977 8. Finkelstein, R. A., M. K. LaRue, D. W. Johnston, M. L Vasil, G. J. Cho, and J. R. Jones. 1976. Isolation and properties of heat-labile enterotoxins from enterotoxigenic Escherichia coli. J. Infect. Dis. 133: S120-S137. 9. Gorbach, S. L., and C. M. Kharana. 1972. Toxigenic Escherichia coli. N. Engl. J. Med. 287:791-795. 10. Guerrant, R. L., L L Branton, T. C. Schartman, L. I. Rebhun, and A. G. Gilman. 1974. Cyclic adenosine monophosphate and alteration of the Chinese hamster ovary cell morphology: a rapid sensitive in vitro assay for the enterotoxins of Vibrio cholerae and Escherichia coli. Infect. Immun. 10:320-327. 11. Guerrant, R. L, M. D. Dickens, R. P. Wenzel, and A. Z. Kapikian. 1976. Toxigenic bacterial diarrhea: nursery involving multiple bacterial strains. J. Pediatr. 89:885-891. 12. Gyles, C. L 1974. Immunological study of the heat-labile enterotoxin of Escherichia coli and Vibrio cholerae. Infect. Immun. 9:564-570. 13. Jacks, T. M., and B. J. Wu. 1974. Biochemical properties of Escherichia coli low-molecular-weight heat-sta-
RIA FOR E. COLI LT
545
ble enterotoxin. Infect. Immun. 9:342-347. 13a. Kalica, A. R., R. H. Purcell, M. M. Sereno, R. G. Wyatt, H. W. Kim, R. M. Chanock, and A. Z. Kapildan. 1977. A microtiter solid phase radioimmunoassay for detection of human reovirus-like agent in stools. J. Immunol. 118:1275-1279. 14. Purcell, R. H., D. C. Wong, H. J. Alter, and P. V. Holland. 1973. Microtiter solid-phase radioimmunoassay for hepatitis B antigen. Appl. Microbiol. 26: 478-484. 15. Sach, D. A., and R. B. Sach. 1975. Test for enterotoxigenic Escherichia coli using Y1 adrenal cells in miniculture. Infect. Immun. 11:334-336. 16. Sack, R. B. 1975. Human diarrheal disease caused by enterotoxigenic Escherichia coli. Annu. Rev. Microbiol. 29:333-353. 17. Sack, R. B., and J. Froehlich. 1977. Antigenic similarity of heat-labile enterotoxins of diverse strains of Escherichia coli. J. Clin. Microbiol. 5:570-572. 18. Smith, N. W., and R. B. Sack. 1973. Immunologic crossreactions of enterotoxins from Escherichia coli and Vibrio cholerae. J. Infect. Dis. 127:164-170.
Vol. 17, No. 3 Printed in U.S.A.
Microtiter Solid-Phase Radioimmunoassay for Detection of Escherichia coli Heat-Labile Enterotoxin HARRY B. GREENBERG,`* DAVID A. SACK,2 WILLIAM RODRIGUEZ,3 R BRADLEY SACK 2 RICHARD G. WYATT,1 ANTHONY R. KALICA,1 ROBERT L. HORSWOOD,1 ROBERT M. CHANOCK,1 AND ALBERT Z. KAPIKIAN'
Laboratory of Infectious Diseases, National Institute ofAllergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland 20014'; Johns Hopkins Hospital, Baltimore, Maryland 212102; and Children's Hospital National Medical Center of the District of Columbia, Washington, D.C. 200103
Received for publication 21 April 1977
The development of a microtiter solid-phase radioimmunoassay for the detection of Vibrio cholerae enterotoxin and heat-labile Escherichia coli enterotoxin is described. The test is based on the immunological similarity between V. cholerae toxin and E. coli heat-labile toxin. The assay is easy to perforn, quantitative, and at least as sensitive and specific as the Y-1 adrenal cell system. In recent years enterotoxigenic strains of Escherichia coli have been shown to be an important cause of acute diarrheal disease in both humans and animals (16). These strains elaborate at least two toxins, both of which appear to be implicated in disease production. One form of toxin is heat-stable, nonantigenic, and of low molecular weight (13). The other toxin is heat-labile (LT) (8), has a molecular weight of greater than 30,000 (8), is antigenic, and appears to act through stimulation of the adenyl cyclase system (16). Neither toxin has been completely purified (1, 4, 7, 8). The LT E. coli enterotoxin is related immunologically to the enterotoxin of Vibrio cholerae (18). Several workers have shown that antitoxin against V. cholerae enterotoxin will neutralize not only the homologous cholera enterotoxin but also the E. coli LT (12). The immunological similarity between the two toxins is great enough to allow partial purification of LT with a V. cholerae antitoxin affinity column (1). Until very recently only bioassay systems were available to. measure E. coli LT enterotoxin. These systems include the rabbit ileal loop (2), infant rabbit (9), rabbit skin (6), mouse Y-1 adrenal cells (13), and Chinese hamster ovary cell assays (10). In addition a serological test for detecting E. coli LT involving inhibition of immune hemolysis has recently been described (5). At present, tissue culture assays are the most widely used screening procedures for detection of E. coli LT. While these assays appear to be specific and sensitive, they are often difficult to perform, expensive, and not easily quantitated. We have used the immunological cross-reactivity of E. coli LT and V. cholerae
toxin to develop a solid-phase radioimmunoassay for E. coli enterotoxin. MATERLALS AND METHODS Reagents. High-titer burro antiserum to V. cholerae enterotoxin (kindly provided by John B. Robbins, Bureau of Biologics, Food and Drug Administration) was produced as previously described (1). Immunoglobulin G was purified from this serum by a combination of ammonium sulfate precipitation followed by diethylaminoethyl-celiulose chromatography (14). Purified cholera toxin was purchased from Schwarz/ Mann (Orangeburg, N. Y.). Clinical specimens. E. coli colonies were isolated from stool specimens from Peace Corps volunteers in Kenya as previously described (D. Sack et al., Johns Hopkins Med. J., in press). Similarly, E. coli and other coliforms were isolated from children with diarrheal disease seen at the Children's Hospital National Medical Center, Washington, D.C. Unless otherwise stated, randomly selected colonies were grown in syncaseglucose or Trypticase soy broth. After a 24-h growth period, the cultures were centrifuged and the supernatants were assayed for toxin activity in the Y-1 adrenal cell assay (15). The same supernatant fluids were also tested by radioimmunoassay (RIA). Lyophilized dialyzed filtrates from broth cultures of toxigenic E. coli that had been isolated in various parts of the world were also tested by RIA. These filtrates had been tested previously for toxin by a rabbit ileal loop assay (17). RIA reagents. Purified burro immunoglobulin G anti-cholera toxin was labeled with '25I (ca. 20 LCi/4g of protein) by a modification of the Hunter-Greenwood method (14). Chromatography on Sephadex G75 was used to separate unreacted "5I from labeled protein. The labeled immunoglobulin was diluted 1:1 in fetal calf serum and stored at 4°C. RIA test. A solid-phase microtiter RIA test was used. The test was similar to those used for detecting 541
542
INFECT. IMMUN.
GREENBERG ET AL.
hepatitis B surface antigen (14) and human rotavirus (13a). Briefly, polyvinyl microtiter plates (Cooke) were precoated with burro anti-cholera serum (100 gid per well) diluted in phosphate-buffered saline (PBS), pH 7.4. The optimal dilution of antiserum (1:2,000 in this instance) was determined by checkerboard titration. After incubating the plates at room temperature for 12 h, the wells were washed five times with PBS and then filled with PBS containing 1% bovine serum albumin (BSA) and again incubated for 12 h. The PBS-BSA was removed, and the plates were washed twice with PBS. Fifty microliters of the appropriate antigen, which consisted of E. coli culture supernatant, dilutions of lyophilized E. coli culture, or purified cholera toxin diluted in PBS with 1% BSA, was added to the wells and allowed to incubate overnight at room temperature. The wells were then washed five times with PBS, and 50 pl of l"I-labeled burro anti-cholera toxin was added to each well. After a 4-h incubation at 370C, the plates were again washed five times with PBS and cut up with a scissors, and the separate wells were placed in gamma-counting tubes for analysis. Results were expressed as a ratio of residual counts in the sample well to the mean of residual counts in wells that received either PBS-BSA or sterile syncase-glucose broth (positive/negative, P/ N). A P/N ratio of 1.7 or greater was considered
25
Z
a. a
j
15
S
18
I
1/125
1525
1/25
1/125
SERIAL DILUTION OF EscA
c
LT
1/5
11
STANDARD
FIG. 2. Titration of crude broth supernatant from a 24-h culture of a toxigenic E. coli strain. P/N ratio represents counts per minute of '25I-labeled immunoglobulin G burro anti-cholera toxin bound to the solid phase of an antigen-containing well divided by counts per minute bound in a weU containing broth.
E. coli toxin had not yet been purified, it was not possible to quantitate the sensitivity of the assay system in terms of the absolute amount of
LT that could be detected. However, it should be noted that the P/N ratios in this heterologous system were considerably lower than those obtained with cholera toxin in an homologous positive. system. The RIA reactivity of both toxins was reduced by greater than 99% when they were RESULTS incubated at 1000C for 20 min prior to testing. When serial 100-fold dilutions of purified Seven dialyzed lyophilized broth filtrates of cholera toxin were used as test antigen, a typical toxigenic strains of E. coli isolated in various sigmoidal binding curve was observed (Fig. 1). parts of the world were assayed quantitatively Of interest was the sensitivity of the technique. in both the rabbit ileal loop test and RIA (Fig. Toxin was detected at the 0.01-pg level. Serial 3). Both assays were done with serial dilutions, fivefold dilutions of a standard crude broth su- and both were based on the dry weight of the pernatant of a potent toxigenic E. coli strain crude lyophilized filtrate. There was a signifiwere tested in a similar manner (Fig. 2). Again, cant rank-order correlation between the two a typical binding curve was obtained. Because systems (Spearman rank correlation coefficient 1.0, P < 0.001). The RIA was approximately 100fold more sensitive than the rabbit ileal loop, and, thus, the RIA was roughly equivalent to the sensitivity of the cell culture systems (16). Three hundred and fifty-nine separate broth supernatants from 38 Peace Corps volunteers in Kenya and 4 other patients with diarrhea were tested simultaneously by both the Y-1 adrenal cell assay and RIA. The RIA was performed under code (Fig. 4). The correlation between the results of the two tests was extremely high (V = +0.9607, Kendall coefficient of association; X2= 331.4; P < 0.0001). False positives were not encountered, but three false negatives were found. The adrenal cell assay-positive orga100 nisms were derived from selected E. coli cololDtERA TOXI CMNNTRATAN nies isolated from 15 of the 42 patients. FIG. 1. Titration of purified cholera toxin by miForty-seven pediatric patients with acute crotiter solid-phase RIA: P/N ratio represents diarrhea seen at Children's Hospital National counts per minute of '25I-labeled immunoglobulin G burro anti-cholera toxin bound to the solid phase of Medical Center, Washington, D.C., were studied for evidence of LT toxin-producing organisms. an antigen-containing well divided by counts per minute bound in a well containing PBS and 1% BSA. The results of both the Y-1 adrenal cell assay IIDIp/Al
0p,m
a "W
..g/d
RIA FOR E. COLI LT
VOi. 17, 1977
and RIA are seen in Table 1. The correlation between results of the two tests was highly significant (V = +0.9306, Kendall coefficient of association; x2 = 40.7; P < 0.0001). In only one patient studied did the results of the two assays differ. In this case the RIA was repeatedly positive, whereas the adrenal cell assay repeatedly yielded equivocal results (reported as negative in Table 1). As in the tests performed on specimens from Peace Corps volunteers, the RIA was performed under code. The sensitivity of the Y-1 adrenal cell assay and RIA was compared by testing serial fivefold dilutions of five positive E. coli broths. The results of this experiment are seen in Table 2. The assays appeared to be roughly comparable in sensitivity.
-
A series of experiments was done to define the best growth conditions for producing RIAdetectable E. coli LT enterotoxin (Table 3). Generally, the best medium appeared to be syncase-glucose broth, and brain heart infusion broth was definitely inferior. Of interest was the finding that 48-h broth cultures that were grown in a microtiter "mini"-plate and that were not filtered or centrifuged appeared to have more toxin activity than culture supematants. Thus far in preliminary studies it does not appear that the use of such broth specimens containing whole organisms alters the specificity of the test. In addition, it appears feasible to actually grow the E. coli cultures in a microtiter plate that TABLE 1. Comparison of detection of E. coli LT by RIA and an adrenal cell assay No. of patients tested by RIA with indicated
a
m..
I*_
543
No. of patients tested by adrenal
cell assay with indicated result
result
Positive
Negative
Positive Negative
8 0
1 38
TABLE 2. Comparison of sensitivity of RLA and Y-1 adrenal ceU assay for detection of E. coli LT
a*I
I 1
E. coli isolate F
TITE A
Strain
Reciprocal titera of toxin activity in: RIA
fygI
FIG. 3. Comparison of toxin activity of seven broth filtrates from various toxigenic E. coli isolates by the rabbit ikal loop assay and RIA. Titers were determined by serial dilutions and based on the dry weight of the lyophilized broth filtrates. The country of orWgin of the filtrate is indicated in parentheses.
Adrenal cell
125 125 1 1 337cl 10407 625 125 5 25 349c5 25 5 375c4 a Serial fivefold dilutions of broth supernatants. 1 2 3 4 5
408-3
ADRENAL CELL CULTURE ASSAY
POSITIVE
NEGATIVE
iE
a
S
FIG. 4. Comparison of RIA and adrenal cell assay for detection of E. coli LT enterotoxin in broth filtrates. Broth filtrates were derived from E. coli isolated from Peace Corps volunteers in Kenya and other patients with diarrhea. The RIA P/N values were calculated using an N value that was the counts per minute bound by wells inoculated with sterile broth.
544
INFECT. IMMUN.
GREENBERG ET AL.
TABLE 3. P/N ratios observed in RIA for E. coli LT under indicated cultural conditions Culture conditions
Organism (strain) 1 (337c0) 2 (408-3) 3 (10407) 4 (375c0) 5 (349c0) a
Roller rum for 24 ha Flask shaking for 24 Flask stationary for 24 Flask stationary for 48 Mini-plate stationha ha ha ary for 48 hb
Sc 8.2 9.6 5.8 7.0
Td 3.0 13 14
9.2
0.9 3.6 4.9 3.2
6.2
3.8
2.0
Be
S 14 14 18 11 16
T
B
S
T
10 23 23 10 9.2
3.5 9.0 6.0 5.0
3.6 13 5.8 23
0.9 7.7 20 28
4.6
6.8
2.2
B 1
3.5 6.6 15 1.0
S
T
B
S
T
B
6.0 5.4 37 8.1
1.4 15 13 10 3.5
1.3 4.0 7.4 7.7
21 28 74 33 24
20 40 42 46
9 18 28
62
1.8
13
28 8.2
Supernatant fluid tested in assay.
b Whole c S, Glucose d
culture, organisms, and broth were tested in assay. syncase. T, Trypticase soy broth with 0.6% yeast extract. e B, Brain heart infusion.
has previously been precoated with anti-cholera The radioimmunoassay should aid in the purification of E. coli LT since large numbers of toxin, thereby simplifying the assay further. fractions derived from various separation procedures can be tested and quantitative data obDISCUSSION tained. The use of RIA may offer an additional Recent studies have clearly demonstrated advantage in that immunological activity may that certain strains of E. coli produce LT enter- be preserved better than biological activity durotoxin that is very similar both immunologically ing purification procedures such as affinity coland physiologically to the enterotoxin produced umn chromatography. Preliminary studies indiby V. cholerae. Taking advantage of this immu- cate that the RIA can be used for detection and nological relationship, we devised a RIA for de- quantitation of antibodies for LT in human setection of E. coli LT that used antibody raised rum. Finally, it should be noted that this type of against highly purified cholera toxin. Our find- solid-phase RIA also provides an extremely senings indicate that LT toxin can be reliably de- sitive assay system for the detection and quantitected by a solid-phase RIA. The specificity and tation of V. cholerae enterotoxin (S. J. Cryz et sensitivity of the RIA are equivalent to or al., Program Abstr. Intersci. Conf. Antimicrob. greater than the Y-1 adrenal cell assay, one of Agents Chemother. 16th, Chicago, Ill., Abstr. the most sensitive bioassay systems currently no. 367, 1976). available to identify toxin production. HyperimLITERATURE CITED mune monospecific antiserum to E. coli LT, 1. Dafni, Z., and J. B. Robbins. 1976. Purification of heatlabile enterotoxin from Escherichia coli 078:H11 by when available, should increase the sensitivity affinity chromatography with antisera to Vibrio choland specificity of this assay further. In data not erae toxin. J. Infect. Dis. 133:S138-S141. shown we found that non-toxin-producing enter- 2. De, S. W., K. Bhattacharya, and S. Sarkar. 1956. A study of the pathogenicity of strains of bacterium coli. opathogenic serotypes and 12 E. coli isolates J. Pathol. Bacteriol. 71:201-209. producing only heat-stable toxin were unreacS. T., H. W. Moon, and S. C. Wipp. 1974. Detive in this test. This would be expected since 3. Donta, tection of heat labile Escherichia coli enterotoxin with the antibody was raised against highly purified the use of adrenal cells in tissue culture. Science cholera toxin. 183:334-336. There are several advantages of the RIA over 4. Dorner, F., H. Jaksche, and W. Stockl. 1976. Escherichia coli enterotoxin: purification partial characteriexisting detection systems. The RIA is easy to zation and immunological observation. J. Infect. Dis. standardize and automate. Problems with cell 133:S142-S156. culture contamination and variability are 5. Evans, D. J., and D. G. Evans. 1977. Inhibition of immune hemolysis: a serological assay for heat-labile enavoided, and the results obtained are quantitaterotoxin of Escherichia coli. J. Clin. Microbiol. tive. Whether this system can be used to detect 5:100-105. toxin directly in stool or intestinal wash speci- 6. Evans, D. J., D. G. Evans, and S. L. Gorbach. 1973. mens remains to be seen. In addition, since the Production of vascular permeability factor by enterotoxigenic Escherichia coli isolated from man. Infect. assay is based on a specific immunological reacImmun. 8:725-730. tion rather than a biological effect, it can be D. J., D. G. Evans, S. H. Richardson, and S. used more reliably to evaluate possible V. chol- 7. Evans, L Gorbach. 1976. Purification of polymixin-released erae-like LT production in non-E. coli coliforms heat labile enterotoxin of Escherichia coli. J. Infect. (11). Dis. 133:S97-S102.
VOL. 17, 1977 8. Finkelstein, R. A., M. K. LaRue, D. W. Johnston, M. L Vasil, G. J. Cho, and J. R. Jones. 1976. Isolation and properties of heat-labile enterotoxins from enterotoxigenic Escherichia coli. J. Infect. Dis. 133: S120-S137. 9. Gorbach, S. L., and C. M. Kharana. 1972. Toxigenic Escherichia coli. N. Engl. J. Med. 287:791-795. 10. Guerrant, R. L., L L Branton, T. C. Schartman, L. I. Rebhun, and A. G. Gilman. 1974. Cyclic adenosine monophosphate and alteration of the Chinese hamster ovary cell morphology: a rapid sensitive in vitro assay for the enterotoxins of Vibrio cholerae and Escherichia coli. Infect. Immun. 10:320-327. 11. Guerrant, R. L, M. D. Dickens, R. P. Wenzel, and A. Z. Kapikian. 1976. Toxigenic bacterial diarrhea: nursery involving multiple bacterial strains. J. Pediatr. 89:885-891. 12. Gyles, C. L 1974. Immunological study of the heat-labile enterotoxin of Escherichia coli and Vibrio cholerae. Infect. Immun. 9:564-570. 13. Jacks, T. M., and B. J. Wu. 1974. Biochemical properties of Escherichia coli low-molecular-weight heat-sta-
RIA FOR E. COLI LT
545
ble enterotoxin. Infect. Immun. 9:342-347. 13a. Kalica, A. R., R. H. Purcell, M. M. Sereno, R. G. Wyatt, H. W. Kim, R. M. Chanock, and A. Z. Kapildan. 1977. A microtiter solid phase radioimmunoassay for detection of human reovirus-like agent in stools. J. Immunol. 118:1275-1279. 14. Purcell, R. H., D. C. Wong, H. J. Alter, and P. V. Holland. 1973. Microtiter solid-phase radioimmunoassay for hepatitis B antigen. Appl. Microbiol. 26: 478-484. 15. Sach, D. A., and R. B. Sach. 1975. Test for enterotoxigenic Escherichia coli using Y1 adrenal cells in miniculture. Infect. Immun. 11:334-336. 16. Sack, R. B. 1975. Human diarrheal disease caused by enterotoxigenic Escherichia coli. Annu. Rev. Microbiol. 29:333-353. 17. Sack, R. B., and J. Froehlich. 1977. Antigenic similarity of heat-labile enterotoxins of diverse strains of Escherichia coli. J. Clin. Microbiol. 5:570-572. 18. Smith, N. W., and R. B. Sack. 1973. Immunologic crossreactions of enterotoxins from Escherichia coli and Vibrio cholerae. J. Infect. Dis. 127:164-170.
