JOURNAL OF CLINICAL MICROBIOLOGY, June 1994, p. 1444-1448 Vol. 32, No. 60095-1 137/94/$04.00+0Copyright C) 1994, American Society for Microbiology

Enzyme Capture Assay for Rapid Identification of Escherichiacoli in Blood Cultures

SHIU WEN HUANG,' JIUNN JONG WU,2 AND TSUNG CHAIN CHANG'*Food Industry Research and Development Institute, Hsinchu, Taiwan 300,1 and Department of Medical Technology,

National Cheng-Kung University Medical College, Tainan, Taiwan 700,2 Republic of China

Received 17 December 1993/Returned for modification 3 February 1994/Accepted 2 March 1994

An enzyme capture assay (ECA) for rapid identification of Escherichia coli in blood cultures by usingP-D-glucuronidase as a marker was developed. Microdilution plates coated with antiglucuronidase were usedto capture this enzyme from the cell lysates of blood cultures which showed growth of gram-negative bacteria.The assay, using 4-methylumbelliferyl-j3-D-glucuronide as a fluorogenic substrate, had a detection limit of 0.1ng/ml (3 x 10-13 M) for the enzyme; this was approximately equal to a cell concentration of 106 CFU ofE. coliper ml. Among 212 blood cultures showing growth of gram-negative bacteria, 77 specimens were found tocontain E. coli by conventional culture procedures and 73 samples were positive by ECA. Among the 135 bloodcultures from which E. coli was not isolated, ECA gave one false-positive (Salmonella enteritidis) reaction. Thus,the sensitivity and specificity for the identification of E. coli in blood cultures by ECA were 94.8% (73/77) and99.3% (134/135), respectively. From the finding of positive growth in the culture bottle, the assay can becompleted within 4 h. In view of the high rate of isolation of E. coli from bacteremic patients, the test can beperformed in parallel with conventional culture protocols; this may shorten the identification time for E. coli,and proper antimicrobial treatments may be started 24 h earlier than when results of conventionalidentification systems are used.

The isolation of any significant microorganism from a bloodculture is an occurrence that requires careful evaluation by theclinicians, and prompt action is usually necessary. The inci-dence of bacteremia and fungemia has been reported to be 3.4to 28 per 1,000 hospital admissions and was estimated toaverage 10 per 1,000 admissions (1%) in the United States(31). The five most common isolates from blood cultures wereEscherichia coli, Staphylococcus aureus, Streptococcus pneu-moniae, Klebsiella pneumoniae, and Pseudomonas aeruginosa (2).

E. coli is one of the more important gram-negative bacteriarecovered in clinical microbiology laboratories. It is isolatedfrequently from specimens from bacteremic episodes (2, 10, 19,21, 25), either as a pure culture or as part of a mixed culture.The isolation rate for E. coli can be as high as 40% of thegram-negative bacteria found in bacteremia (25). During 1991and 1992, isolates of E. coli accounted for 21.7% of all bacteriacausing bloodstream infections at National Cheng-Kung Uni-versity Hospital, Tainan, Taiwan (unpublished data). In an-other study, E. coli was found to account for 28.4% of allisolates from bacteremic episodes in the pediatric departmentof another hospital (5). In view of the high incidence andmortality (2), a rapid identification method for E. coli in bloodspecimens is of clinical importance.

Since 1976, when Kilian and Bulow (17) described theactivity of a 3-D-glucuronidase (GUD) as being restricted toEscherichia, Shigella, and Salmonella spp., this property hasbeen widely used for the detection and identification of E. coli(6, 9, 11, 16, 30) in food and in clinical microbiology labora-tories. Most studies incorporated 4-methylumbelliferyl-,3-D-glucuronide (MUG) as a fluorogenic substrate of GUD in themedium. The native GUD of E. coli may be a tetramer havinga molecular weight of 296,000 (1). Some strains of Yersinia,

* Corresponding author. Mailing address: Food Industry Researchand Development Institute, P.O. Box 246, Hsinchu, Taiwan 300,Republic of China. Phone: 886-35-223191 ext. 271. Fax: 886-35-214016.

Flavobacterium, staphylococci, and streptococci are also posi-tive for this enzyme (12, 22, 26), and false-positive reactionsare sometimes caused by animal tissues which contain theenzyme. For this reason, an enzyme capture assay (ECA) fordetection of E. coli in oysters was developed (12). Anti-GUDantibodies coated on microdilution plates were used to capturethe enzyme produced by E. coli present in food samples; thiswas followed by the addition of fluorogenic or chromogenicsubstrates to demonstrate GUD activity. Under these condi-tions, cross-reactions are caused only by members of the familyEnterobacteriaceae.

Although the direct incorporation of a fluorogenic substrateof GUD in the culture medium is a convenient and fast way todetect E. coli present in specimens (9, 30), the practice isimpossible for blood cultures because of the deep red color ofblood, which masks the fluorescence generated by GUD on thesubstrate. ECA seems to be a good choice for solving thisproblem. Husson et al. (14) have described an alkaline phos-phatase capture test for the identification of E. coli and Shigellaspecies in blood cultures and urine specimens. The sensitivityof the alkaline phosphatase capture assay for detecting E. coliin blood cultures is 91%.We describe here a rapid and sensitive ECA to identify E.

coli in positive blood cultures showing growth of gram-negativebacteria, including mixed cultures containing gram-negativebacteria. For the preparation of anti-GUD used in the ECA, arapid procedure for the purification of GUD by preparativepolyacrylamide gel electrophoresis is also described.

MATERIALS AND METHODS

Bacterial strains and culture conditions. Among the 108strains of E. coli tested for the production of GUD by ECA,102 were clinical isolates obtained from the Triservice GeneralHospital (Taipei, Taiwan). The remaining 6 strains of E. coliand 60 other strains of Enterobacteriaceae were from theCulture Collection and Research Center (CCRC), Hsinchu,

1444

on October 24, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

RAPID IDENTIFICATION OF E. COLI IN BLOOD CULTURES 1445

Taiwan). The identities of all E. coli isolates were reconfirmedby the Micro-ID diagnostic kit (Organon Teknika Corp.,Durham, N.C.). To test for the production of GUD by variousbacteria under blood culture conditions, a single colony of eachstrain grown overnight on tryptic soy agar was suspended in 0.5ml of sterile saline. Following the injection (with a syringe) ofthis bacterial suspension into a BACTEC NR6A (BectonDickinson, Sparks, Md.) aerobic blood culture vial, 3 ml ofhuman blood (from healthy donors) was added to the culturebottle. The culture bottles were incubated at 35°C for 24 hbefore the ECA was performed.

Purification of GUD. Partially purified GUD from E. coliK-12 (Boehringer, Mannheim, Germany) were further purifiedby disc polyacrylamide gel (10%, 3-mm thickness) electro-phoresis with a vertical apparatus (model V16; BethesdaResearch Laboratories, Gaithersburg, Md.). After electro-phoresis, the gel was activity stained with 5-bromo-4-chloro-3-indolyl-p-D-glucuronide (Sigma, St. Louis, Mo.) (50 ,ug/ml)dissolved in 0.2 M phosphate buffer, pH 6.5. After 10 to 15 minof incubation at room temperature, the blue band correspond-ing to the position of GUD was visible (6). The band was cutfrom the gel with a razor and used for antibody production.

Preparation and purification of antibody. The gel stripcontaining GUD was crushed to fine particles and used forimmunizing rabbits according to the procedure of Chang et al.(4). The specificities of the antisera were tested by Ouchterlonydouble diffusion (24), while the titers of the antisera weredetermined by an enzyme-linked immunosorbent assay (4)using microdilution plates coated with GUD. The immuno-globulin G fractions of the antisera were purified by DEAEion-exchange chromatography (18) and were further purifiedby affinity chromatography (13).ECA. The ECA was performed as previously described (16)

with some modifications. Wells of microdilution plates (Mi-croFluor; Dynatech, Alexandria, Va.) were coated with affin-ity-purified anti-GUD and blocked with bovine serum albumin.Serially diluted GUD solution or cell lysate was then added tothe wells and incubated at 37°C for 1.5 h. After the wells werewashed, MUG solution (200 ,ug/ml) was added and incubatedat 37°C for 1.5 h. The reaction was stopped with 0.2 N NaOH,and the results were read with a MicroFluor reader (Dynatech)at a wavelength of 365 nm. As a negative control, phosphate-buffered saline was used instead of GUD. A reading offluorescence units greater than that of the negative controlplus three standard errors was considered positive.

Culture broth (0.5 ml) obtained after 24 h of growth for purecultures or obtained from blood cultures positive for gram-negative bacteria in the BACTEC NR vials was asepticallydrawn into to a microcentrifuge tube, which was capped andcentrifuged at 5,000 x g for 10 min. The cell pellet was lysed at35°C for 15 min in 0.5 ml of 0.1 M phosphate buffer containing8% (wt/vol) sucrose, 25 mM EDTA, 0.05% Triton X-100, andlysozyme (Sigma) (1.2 mg/ml), pH 7.0. The activity of GUD inthe lysate was assayed by ECA.

Detection limit for E. coli. To determine the minimalnumber of E. coli cells in the blood culture vials necessary togive a positive ECA result, 3 ml of blood and 0.1 ml of a dilutedsuspension of E. coli (05-8 and CCRC 10316) were injectedinto the BACTEC NR6A bottle to give an inoculum level ofapproximately 2 to 10 CFU/ml. The vials were incubated at35°C, sampled at intervals to determine the cell numbers byplate count, and tested for GUD activity by ECA.

Clinical specimens. Blood specimens were collected at theNational Cheng-Kung University Hospital. The BACTECNR6A and NR7A vials were normally inoculated with 3 to 5 mlof blood from the patients, inserted into a BACTEC NR660

TABLE 1. Production of GUD by 168 strains of Enterobacteriaceaeas analyzed by ECA

Organism No. of No. (%) of strainsstrains tested ECA positive

E. coli 108 104 (96.3)Salmonella spp. 28 2 (7)Shigella spp. 8 2 (25)Citrobacter spp. 3 0Enterobacter spp. 2 0Klebsiella spp. 7 0Proteus spp. 11 0Serratia sp. 1 0

instrument (Johnson Laboratories Inc., Towson, Md.), andincubated at 35°C. Broth from positive vials was Gram stainedand subcultured onto MacConkey, chocolate, and blood agarplates. All isolates were identified by conventional microbio-logical techniques. The culture vials showing growth of gram-negative bacteria, including mixed cultures containing gram-negative bacteria, were tested by ECA according to theprocedures described above.

Sensitivity and specificity. Sensitivity and specificity were

determined as described by McClure (20).Induction of GUD. Since GUD is an inducible enzyme (23),

four P-D-glucuronides were compared for their abilities toinduce the enzyme in two strains of E. coli (05-8 and CCRC10316). After the addition of 0.5 ml of the bacterial suspensionand 3 ml of blood to the BACTEC NR6A vial, each stocksolution (16.5 mg/ml) of ,-D-glucuronides (5-bromo-4-chloro-3-indolyl-13-D-glucuronide, methyl-,B-D-glucuronide, MUG, andp-nitrophenyl-p3-D-glucuronide [all from Sigma]) was added toeach culture vial to reach a final concentration of 50 ,ug/ml.The culture vials were then incubated at 35°C for 24 h beforethe ECA was performed.

RESULTS

Preparation of anti-GUD. The antisera were specific toGUD, since only one precipitation line was found in theOuchterlony double diffusion test (data not shown). The titersof the antisera were found to be around 107 as determined byan enzyme-linked immunosorbent assay (4).

Bacteria producing GUD. Among the 108 strains of E. colitested, 104 (96.3%) produced GUD as detected by ECA(Table 1). Of the 28 strains of Salmonella, two strains (7%)gave positive ECA results. However, among the eight strains ofShigella, two strains (25%) were ECA positive. None of theother 24 strains of Enterobacteriaceae produced GUD as

determined by ECA.Detection limit for GUD and E. coli. For microdilution

plates coated with affinity-purified anti-GUD, the detectionlimit for GUD by ECA was about 0.1 ng/ml (3 x 10-13 M)(Fig. 1). This enzyme concentration was approximately equalto a cell density of about 106 CFU/ml, as determined with twoE. coli strains (05-8 and CCRC 10316) in the inoculationexperiments. The assay was about 5 to 10 times less sensitivewhen DEAE-purified anti-GUD was used for the coating ofmicrodilution plates (data not shown).

Detection of E. coli in blood cultures. The bacteria isolatedfrom the 212 blood cultures which showed growth of gram-negative bacteria were E. coli (77 isolates), K pneumoniae (33isolates), P. aeruginosa (26 isolates), Acinetobacter anitratus (11isolates), Enterobacter cloacae (10 isolates), and other bacteria,with a total of 247 isolates. Of the 212 positive blood cultures,

VOL. 32, 1994

on October 24, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

1446 HUANG ET AL.

C~)0

cnDa)

c)0C1)0

U)(.

a:

4

3

1

o

0.001 0.01 0.1 1 10

R-D-Glucuronidase Concentration (ng/ml)

FIG. 1. Dose response of ECA for GUD. The microdilution platewas coated with affinity-purified anti-GUD, and MUG was used as a

fluorogenic substrate for the enzyme. Each point represents the meanof quadruplicate determinations. The negative control had a responseof 15 relative fluorescence units.

20 were mixed cultures. Among the 77 blood samples fromwhich E. coli was isolated by conventional procedures, 73 were

also positive by ECA. There were 134 blood specimens whichwere E. coli negative by both the conventional methods andECA. ECA produced four false negatives and one falsepositive. Among the four false negatives, three were polymi-crobial infections with the E. coli isolates being GUD positiveas confirmed by subculturing the isolates in tryptic soy brothcontaining MUG (9) and retesting with ECA. The fourth falsenegative was a blood sample from which only E. coli was

isolated, and this isolate was confirmed to be GUD negative.The only false positive was caused by Salmonella enteritidis.Therefore, the sensitivity and specificity of ECA for thedetection of E. coli in blood cultures were 94.8% (73/77) and99.3% (134/135), respectively. The positive and negative pre-

dictive values of ECA were 98.6% (73/74) and 97.1% (134/138), respectively. The agreement between the conventionalculture methods and ECA was 97.6% (207/212).

Induction of GUD. Although most strains of E. coli can

produce GUD in the BACTEC NR6A vials without addedinducers, the captured GUD activity was about 5 to 64 timeshigher in the presence of inducers added at a concentration of50 pLg/ml (Fig. 2). Both E. coli CCRC 10316 and 05-8 were

induced to approximately the same extent by each f-glucu-ronide. However, p-nitrophenyl-p3-D-glucuronide was found tobe the best inducer among the four compounds tested; thecaptured enzyme activity was 45 and 64 times higher in E. coliCCRC 10316 and 05-8, respectively, in the presence of thiscompound than in vials without any inducers added. Methyl-3-D-glucuronide was the second most effective inducer among

the four 3-glucuronides (Fig. 2).

DISCUSSION

Of 108 isolates of E. coli tested with ECA, 104 (96.3%) were

positive (Table 1). This value is slightly higher than that (94%)reported by Holt et al. (12) and is very close to that (97%)reported by Kilian and Bulow (17). False positives caused by

pure cultures of Shigella and Salmonella spp. were 25 and 7%(Table 1), respectively. However, only a small percentage ofbloodstream infections are caused by Salmonella spp., andbacteremia caused by Shigella spp. is rarely found (10, 19, 25,27, 31). In this study, among the 212 blood cultures, containinggram-negative bacteria, Salmonella spp. were found in sixspecimens (2.8%), and only one (S. enteritidis) of the sixisolates displayed an ECA-positive result. However, Salmo-nella bacteremia is being identified with increasing frequencyin AIDS patient (3, 7, 28, 29). For this reason, the ECA may benot suitable for rapid identification of E. coli in AIDS patientshaving bloodstream infections. No Shigella spp. were isolatedfrom the 212 positive blood samples.The sensitivity (94.8%) and specificity (99.3%) of ECA for

identifying E. coli in blood samples were high. Of the fourfalse-negative results by ECA, three were polymicrobial infec-tions (E. coli and K pneumoniae, E. coli and P. aeruginosa, andE. coli, K pneumoniae, and P. aeruginosa). The three E. coliisolates from the polymicrobial bacteremia were GUD positiveas confirmed by subculturing these isolates in tryptic soy brothcontaining MUG (9) and retesting with ECA. The fourthfalse-negative specimen was unimicrobial, with the isolated E.coli being GUD negative. Therefore, most false-negative reac-tions were due to mixed cultures present in the blood samples.The reason why ECA could not detect GUD in a polymicrobialenvironment is not clear at present; one possibility might be thatthe cell density of E. coli was not high enough to produce adetectable amount (0.1 ng/ml or 10-l3 M [Fig. 1]) of GUD withthe competition of other bacteria in the blood broth environment.Although the detection of GUD with ECA was rather

sensitive, E. coli must grow to a cell concentration of about 106CFU/ml in the blood cultures to produce positive ECA results.If competitive microbes or residual antibiotics are present inthe blood specimens, false-negative reactions may occur be-

12-

E. coli CCRC 10316o010 mii E.coliO5-8

C: 8 -

ZI)Cl)

6 -

E~~~~..... ...

0

.a)

.: KI, U-],:>:~-...me BCIG MG MUG PNPG

13-D-Glucuronidase InducersFIG. 2. Induction of GUD by 5-bromo-4-chloro-3-indolyl-3-D-glu-

curonide (BCIG), methyl-f-D-glucuronide (MG), MUG, and p-nitro-phenyl-4-D-glucuronide (PNPG) in E. coli CCRC 10316 and 05-8. TheBACTEC NR6A blood culture vial was supplemented with the differ-ent inducers at a final concentration of 50 ,ug/ml. E. coli was inoculatedinto the culture vials, incubated at 35°C for 24 h, and tested for GUDactivity by ECA. E. coli CCRC 10316 and 05-8 showed 2,600 and 1,700relative fluorescence units, respectively, when no inducer was added tothe culture vial.

2F

J. CLIN. MICROBIOL.

on October 24, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

RAPID IDENTIFICATION OF E. COLI IN BLOOD CULTURES 1447

cause of decreased E. coli growth. A possible method to reducefalse negatives is to include an inducer(,B-D-glucuronide) in theculture broth to enhance the production of GUD by E. coli.Among the four inducers tested, p-nitrophenyl-,B-D-glucu-ronide had the ability to strongly promote enzyme production,by a factor of 45 to 64 times (Fig. 2). The effect of p-nitrophe-nyl-13-D-glucuronide on the growth of nine strains of gram-positive and gram-negative bacteria was tested, and no inhibi-tion was found (data not shown). The feasibility of adding anyinducer, which may cause contamination, before or after bloodis introduced into the culture bottles needs further study.Edberg and Trepeta (8) reported a high sensitivity (93%) for

identification of E. coli isolates within 1 h by GUD assay in abuffer containing p-nitrophenyl-p3-D-glucuronide. However,Jackson et al. (15) recently found that some E. coli isolatesneed 24 to 48 h to develop visible fluorescence in a brothcontaining MUG. We had similar findings in tests of more than100 pure cultures of E. coli from food and clinical sources (datanot shown). In addition, to obtain pure cultures, a subculturingstep which normally requires overnight incubation is needed.Some commercial products (e.g., RAPIDEC coli [AnalytabProducts, Montalieu Vercieu, France] and MUG Disk [Remel,Lenexa, Kans.]) are GUD-based rapid identification tests forE. coli, but pure cultures are required for these tests. Anotheradvantage of the ECA would be that GUD cross-reactionscaused by gram-positive bacteria (staphylococci or strepto-cocci) (12, 22, 26) other than Enterobacteriaceae are not likelyto occur. False positives caused by gram-positive bacteriapresent in the 12 mixed cultures were not found in this study.Husson et al. (14) reported an alkaline phosphatase ECA for

the detection of E. coli in urine and blood cultures. Since thesubstrate (p-nitrophenyl phosphate) used was a chromogeniccompound, a lower sensitivity of the assay could be anticipated.The detection limit of the assay for E. coli was about 107CFU/ml, which is about 10 times less sensitive than the presentECA using MUG as a fluorogenic substrate. In addition, aconcentrating step using 5 ml of the cultured blood is neededin the alkaline phosphatase immunocapture test, and to sepa-rate bacterial cells from blood cells, differential centrifugationsteps were used in that study (14). However, the ECA de-scribed in the present report requires only 0.5 ml of the culturebroth, and after a simple lysis step with lysozyme, the samplecan be directly analyzed.

In view of the high mortality and high isolation rate for E.coli from bacteremic episodes, a rapid identification methodfor this bacterium is important. The common practice used inhospitals is to subculture blood specimens at intervals after anincubation period of 12 h to 2 weeks (10, 32) and at the timewhen Gram stain is positive or bacterial growth is apparent(e.g., by turbidity, gas production, or hemolysis). The subcul-ture and identification steps normally take at least 24 h.Therefore, in addition to the conventional culture protocols,we propose that when growth of gram-negative bacteria isfound, the immunocapture assay for GUD be performed inparallel to rapidly identify E. coli in the blood cultures. Thetime required for ECA is 4 h, and multiple samples can besemiautomatically analyzed in microdilution plates. This couldallow proper antimicrobial treatment 24 h before results ofother commonly used systems are available and may helpreduce the mortality from bacteremia caused by E. coli.

ACKNOWLEDGMENTS

We thank C. H. Chen, Triservice General Hospital, Taipei, Taiwan,for his supply of clinical isolates of E. coli. We are also grateful to theDivision of Bacteriology, Department of Pathology, National Cheng-Kung University Hospital, for providing clinical blood culture samples.

This work was partly supported by a grant (82-EC-2-A-15-0072)from the Ministry of Economic Affairs to T.C.C. and by a grant(NSC82-0412-B006-034) from the National Science Council, Republicof China, to J.J.W.

REFERENCES1. Blanco, C., and G. Nemoz. 1987. One step purification of Esche-

richia coli l-glucuronidase. Biochimie 69:157-161.2. Brauner, A., B. Kaijser, B. Wretlind, andI. Kfihn. 1991. Charac-

terization of Escherichia coli isolated in blood, urine and faecesfrom bacteremic patients and possible spread of infection. ActaPathol. Microbiol. Immunol. Scand. 99:381-386.

3. Chaisson, R. E. 1988. Infections due to encapsulated bacteria,Salmonella, Shigella, and Campylobacter. Infect. Dis. Clin. N. Am.2:475-484.

4. Chang, T. C., C. H. Chen, and H. C. Chen. 1994. A latexagglutination test for the rapid identification of Vibrio parahaemo-lyticus. J. Food Prot. 57:31-36.

5. Chiang, T. M., and T. Y. Chang. 1991. Pediatric bacteremia strainsgrow in blood culture media. Chin. Med. J. (Taipei) 47:39-44.

6. Delisle, G. J., and A. Ley. 1989. Rapid detection of Escherichia coliin urine samples by a new chromogenic,3-glucuronidase assay. J.Clin. Microbiol. 27:778-779.

7. De Wit, S., H. Taelman, P. Van de Perre, D. Rouvroy, and N.ClumecL 1988. Salmonella bacteremia in African patients withhuman immunodeficiency virus infection. Eur. J. Clin. Microbiol.Infect. Dis. 7:45-47.

8. Edberg, S. C., and R. W. Trepeta. 1983. Rapid and economicalidentification and antimicrobial susceptibility test methodology forurinary tract pathogens. J. Clin. Microbiol. 18:1287-1291.

9. Feng, P. C. S., and P. A. Hartman. 1982. Fluorogenic assays forimmediate confirmation of Escherichia coli. Appl. Environ. Micro-biol. 43:1320-1329.

10. Filice, G. A., L. L. Van Etta, C. P. Darby, and D. W. Fraser. 1986.Bacteremia in Charleston County, South Carolina. Am. J. Epide-miol. 123:128-136.

11. Heizmann, W., P. C. Doller, B. Gutbrod, and H. Werner. 1988.Rapid identification of Escherichia coli by Fluorocult media andpositive indole reaction. J. Clin. Microbiol. 26:2682-2684.

12. Holt, S. M., P. A. Hartman, and C. W. Kaspar. 1989. Enzyme-capture assay for rapid detection of Escherichia coli in oysters.Appl. Environ. Microbiol. 55:229-232.

13. Hudson, L., and F. C. Hay. 1989. Practical immunology, 3rd ed., p.322-329. Blackwell Scientific Publications, London.

14. Husson, M. O., C. Mielcarek, D. Izard, and H. Leclerc. 1989.Alkaline phosphatase capture test for the rapid identification ofEscherichia coli and Shigella species based on a specific monoclo-nal antibody. J. Clin. Microbiol. 27:1518-1521.

15. Jackson, L., B. E. Langlois, and K. A. Dawson. 1992.,3-Glucuron-idase activities of fecal isolates from healthy swine. J. Clin.Microbiol. 30:2113-2117.

16. Kaspar, C. W., P. A. Hartman, and A. K. Benson. 1987. Coagglu-tination and enzyme capture tests for detection of Escherichia coli0-galactosidase, ,-glucuronidase, and glutamate decarboxylase.Appl. Environ. Microbiol. 53:1073-1077.

17. Kilian, M., and P. Bulow. 1976. Rapid diagnosis of Enterobacteri-aceae. I. Detection of bacterial glycosidases. Acta Pathol. Micro-biol. Scand. Sect. B 84:245-251.

18. Linn, T. G., A. L. Greenleaf, R. G. Shorenstein, and R. Losick.1973. Loss of the sigma activity of RNA polymerase of Bacillussubtilis during sporulation. Proc. Natl. Acad. Sci. USA 70:1865-1869.

19. Ljungman, P., A. S. Malmborg, B. Nystrom, and A. Tillegard.1984. Bacteremia in a Swedish university hospital: a one-yearprospective study in 1981 and a comparison with 1975-76. Infec-tion 12:243-247.

20. McClure, F. D. 1990. Design and analysis of quantitative collabo-rative studies: minimum collaborative program. J. Assoc. Off.Anal. Chem. 73:953-960.

21. McGowan, J. E. 1985. Changing etiology of nosocomial bacteremiaand fungemia and other hospital acquired infections. Rev. Infect.Dis. 7:S357-S370.

22. Moberg, L. H. 1985. Fluorogenic assay for rapid detection of

VOL. 32, 1994

on October 24, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

1448 HUANG ET AL. J. CLIN. MICROBIOL.

Escherichia coli in food. Appi. Environ. Microbiol. 50:1383-1387.23. Novel, G., M. L. Didier-Fichet, and F. Stoeber. 1974. Inducibility of

f-glucuronidase in wild-type and hexuronate-negative mutants ofEscherichia coli K-12. J. Bacteriol. 120:89-95.

24. Ouchterlony, 0. 1949. Antigen antibody reactions in gels. ActaPathol. Microbiol. Scand. 26:507-515.

25. Roberts, F. J. 1980. A review of positive blood cultures: identifi-cation and source of microorganisms and patterns of sensitivity toantibiotics. Rev. Infect. Dis. 2:329-339.

26. Robison, B. J. 1984. Evaluation of a fluorogenic assay for detectionof Escherichia coli in foods. Appl. Environ. Microbiol. 48:285-288.

27. Speer, C. P., D. Hauptmann, P. Stubbe, and M. Gahr. 1985.Neonatal septicemia and meningitis in Gottingen, West Germany.Pediatr. Infect. Dis. 4:36-41.

28. Sperber, S. J., and C. J. Schleupner. 1987. Salmonellosis duringinfection with human immunodeficiency virus. Rev. Infect. Dis.9:925-934.

29. Tocalli, L., G. Nardi, A. Mammino, A. Salvaggio, and L. Salvaggio.1991. Salmonellosis diagnosed by the laboratory of the 'L. Sacco'Hospital of Milan (Italy) in patients with HIV disease. Eur. J.Epidemiol. 7:690-695.

30. Trepeta, R. W., and S. C. Edberg. 1984. Methylumbelliferyl-,-D-glucuronide-based medium for rapid isolation and identification ofEscherichia coli. J. Clin. Microbiol. 19:172-174.

31. Weinstein, M. P., J. R. Murphy, L. B. Reller, and K. A. Lichten-stein. 1983. The clinical significance of positive blood cultures: acomprehensive analysis of 500 episodes of bacteremia and funge-mia in adults. II. Clinical observations, with special reference tofactors influencing prognosis. Rev. Infect. Dis. 5:54-70.

32. Weinstein, M. P., L. B. Reller, J. R. Murphy, and K. A. Lichten-stein. 1983. The clinical significance of positive blood cultures: acomprehensive analysis of 500 episodes of bacteremia and funge-mia in adults. I. Laboratory and epidemiologic observations. Rev.Infect. Dis. 5:35-53.

on October 24, 2020 by guest

http://jcm.asm

.org/D

ownloaded from