JOURNAL OF CLINICAL MICROBIOLOGY, OCt. 1992, p. 2589-25940095-1137/92/102589-06$02.00/0Copyright © 1992, American Society for Microbiology

Vol. 30, No. 10

Pyrazinamidase, CR-MOX Agar, Salicin Fermentation-EsculinHydrolysis, and D-Xylose Fermentation for Identifying

Pathogenic Serotypes of Yersinia enterocoliticaJ. J. FARMER 111,1* G. P. CARTER,' V. L. MILLER,2t S. FALKOW,2 AND I. K. WACHSMUTH'

Enteric Diseases Laboratory Section, Enteric Diseases Branch, Division ofBacterial and Mycotic Diseases,National Centerfor Infectious Diseases, Centers for Disease Control, Atlanta, Georgia 30333,1 and

Department ofMedical Microbiology, Stanford University, Stanford, California 943052

Received 30 March 1992/Accepted 26 June 1992

We evaluated several simple laboratory tests that have been used to identify pathogenic serotypes of Yersiniaenterocolitica or to indicate the pathogenic potential of individual strains. A total of 100 strains of Y.enterocolitica were studied, including 25 isolated during five outbreak investigations, 63 from sporadic cases,and 12 from stock cultures. The pyrazinamidase test, which does not depend on the Yersinia virulence plasmid,correctly identified 60 of 63 (95% sensitivity) strains of pathogenic serotypes and 34 of 37 (92% specificity)strains of nonpathogenic serotypes. Salicin fermentation-esculin hydrolysis (25°C, 48 h) correctly identified all63 (100%Y sensitivity) strains of the pathogenic serotypes and 34 of 37 (92% specificity) strains of thenonpathogenic serotypes. The results of the pyrazinamidase and salicin-esculin tests disagreed for only 7 of the100 strains of Y. enterocolitica, and these would require additional testing. Congo red-magnesium oxalate(CR-MOX) agar determines Congo red dye uptake and calcium-dependent growth at 36°C, and small redcolonies are present only if the strain contains the Yersinia virulence plasmid. This test has proven to beextremely useful for freshly isolated cultures, but only 15 of 62 strains of pathogenic serotypes that had beenstored for 1 to 10 years were CR-MOX positive. None of the 16 strains of Y. enterocolitica serotype 03fermented D-xylose, so this test easily differentiated strains of this serotype, which now appears to be the mostcommon in the United States. Although antisera that can actually be used to serotype strains of Y. enterocoliticaare not readily available, the four simple tests described above can be used to screen for pathogenic serotypes.

Since it was first described in 1939, Yersinia enterocoliticahas become well established as an enteric pathogen (4-7).However, there is considerable confusion in the literaturebecause not all strains of this species can cause intestinalinfections. Unlike Salmonella and Shigella species, whichare intrinsic pathogens (essentially all strains can causeenteric infections), there is strain-to-strain variation in thepathogenicity of Y enterocolitica (4, 6, 7, 12, 14, 19). Anumber of studies have shown an excellent correlationbetween the serotype and biotype ofY enterocolitica and itsability to cause infections of the intestinal tract and to invadetissue (4, 6, 7, 12, 14, 21).Many different tests and methods have been used over the

years to define the pathogenic potential of Y enterocoliticastrains, but it has only recently been realized that animportant component of virulence in this species is deter-mined by a plasmid that is easily lost (10, 11, 19, 21). For thisreason, there is confusion in the literature on how well thevarious methods identify pathogenic serotypes and deter-mine pathogenic potential.

In 1985, Kandolo and Wauters (10) described the pyra-zinamidase test as a way of differentiating pathogenic fromnonpathogenic serotypes of Y enterocolitica. In 1989, Rileyand Toma (16) described Congo red-magnesium oxalate(CR-MOX) agar, a medium that allows easy recognition ofYenterocolitica colonies that contain the Yersinia virulenceplasmid. This medium determines two properties that are

* Corresponding author.t Present address: Department of Microbiology and Molecular

Genetics, University of California, Los Angeles, Los Angeles, CA90024.

plasmid dependent, Congo red dye uptake and calcium-dependent growth at 36°C (see Fig. 1). In this report, wedescribe our experience with these two simple tests usingstrains isolated in the United States and show their correla-tion with serotype, salicin fermentation-esculin hydrolysis(salicin-esculin), and some other simple laboratory testsused to evaluate pathogenic potential (Table 1).

MATERIALS AND METHODS

Nomenclature. In this report, we use the term "pathogenicserotype" to refer to 11 0-antigen groups of the Y entero-colitica serotyping schema. Strains belonging to these sero-types are generally recognized as being enteric pathogens onthe basis of their occurrence in outbreaks and sporadiccases, pathogenicity in animal models, or the ability toinvade tissue or animal cells in tissue culture (4, 6, 7, 14).Strains of serotypes 03, 09, and 05,27 are well documentedas enteric pathogens in humans and are the most importantserotypes found in Europe and much of the rest of the world(4, 6, 7). Until recently, these three serotypes were rare inthe United States (2, 5, 13), where the pathogenic serotypes04,32; 08; 013a,13b; 018; 020; and 021 are usually found(2-6, 12, 13). These six serotypes have been designated thepathogenic American serotypes (7, 10). Strains of two addi-tional serotypes cause disease in animals: serotype 01,2,3causes disease in chinchillas and serotype 02,3 causesdisease in hares (4, 7). Serotype 01,2,3 has also beenisolated from human clinical infections (16).

Strains and stock cultures. The strains of Yersinia spp.studied are described in more detail in Tables 2 through 4.Stock cultures were prepared at the time the culture was

2589

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

on M

arch 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

on M

arch 6, 2020 by guesthttp://jcm

.asm.org/

Dow

nloaded from

2590 FARMER ET AL.

TABLE 1. Properties of Y enterocolitica strainsa

Pathogenicserotypes with

Property virulence Nonpathogenicplasmid: serotypes

Present Absent'

Small red colonies on CR-MOX + - -agar

Temp-dependent autoagglutination +Invasiveness (cell culture) + +Pyrazinamidase production - - +Salicin fermentation - - +Esculin hydrolysis - - +Belong to serotypes 01,2,3; 02,3; + +

03; 04,32; 05,27; 08; 09;013a,13b; 018; 020; or 021

Belong to other serotypes - - +

Based on published reports (10, 14-16, 20).b Presumably, the virulence plasmid was originally present, but it was lost

during subculture or storage.

received (1978 to 1983 for outbreaks, 1982 to 1987 forsporadic cases) by touching a well-isolated colony with astraight wire and inoculating a tube of stock culture medium(either nutrient agar or semisolid trypticase soy agar). Thissingle-colony pick, as well as any other single-colony picksdone by other laboratories, may have selected a populationthat has lost the Yersinia virulence plasmid. Stock cultureswere stored at room temperature without transfer, with aparaffin-coated cork or rubber stopper used to seal them.Beginning in 1985, we also made a frozen stock of the wholeculture by harvesting cells from the heavy area of growth ona 1- to 3-day-old blood agar plate (25°C) into 10% skim milk.This was quick-frozen and stored at about -70°C. Unlessotherwise specified, test were done at the Centers forDisease Control on fresh subcultures of stock cultures thathad been stored at room temperature for 1 to 10 years.

Outbreaks. Table 2 lists 25 strains that were isolatedduring outbreak investigations. Some of these were nonpath-ogenic serotypes and were considered incidental to theoutbreak itself. Outbreak 1 was an outbreak caused by Yenterocolitica serotype 01,2,3 in chinchillas at several farmsin California. Outbreak 2 occurred in a family group in LeeCounty, Ky., in 1974 (8). Outbreak 3 occurred in Septemberand October 1975 in Oneida County, N.Y. (3), and was dueto contaminated chocolate milk distributed as part of theschool lunch program. Outbreak 4 occurred in 1981 and 1982in Seattle, Wash., and was associated with eating tofucontaminated with Y enterocolitica (17). A number ofdifferent serotypes, both pathogenic and nonpathogenic,were recovered during the outbreak investigations, but themajority from patients were serogroup 08 (data not shown).Outbreak 5 occurred during June and July 1982 in Tennes-see, Arkansas, and Mississippi (18) and was due to Yenterocolitica 013a,13b that contaminated pasteurized milk.

Strains from sporadic cases, old stock cultures, and otherYersinia species. Strains from sporadic cases, old stockcultures, and other Yersinia species (see Table 3) included 63consecutive isolates that were referred to the Enteric Dis-eases Laboratory for one of our diagnostic services (sero-typing, serodiagnosis, identification, etc.). Ten stock cul-tures of strains from the culture collection of theInternational Yersinia Centre at the Institut Pasteur, Paris,France, were also included, because these were more than

10 years old and presumably had been transferred manytimes, making it likely that they had lost the Yersiniavirulence plasmid. Two cultures of serotype 09 isolated inBelgium were also included. Serotype 09 is rarely, if ever,isolated in the United States. For comparison, some otherYersinia species were also studied (see Table 4).

Serotyping. The 0 antigen was determined by microtubeagglutination in 96-well plastic plates or by slide agglutina-tion (3, 4). Absorbed sera were used when necessary. Strainsthat did not agglutinate in any of the sera used were codedNR, for no reaction. Strains that reacted in multiple serawere coded UND, for undefined. A few strains were roughand agglutinated in the saline control. All strains weredefined as belonging to pathogenic or nonpathogenic sero-types on the basis of these serotyping results.

Tissue culture invasiveness. All 100 strains were tested atStanford University for their ability to invade HEp-2 cells inculture (14). Briefly, about 2 x 107 bacterial cells were addedto a monolayer of HEp-2 cells in a 24-well plate. Afterseveral incubation and handling steps, the HEp-2 cells werelysed to release intracellular bacteria, which were thenrecovered and counted on plating media. The study of Milleret al. (14) also gives the results for the strains listed in Tables1 to 3 that were obtained with several DNA probes designedto measure chromosomal virulence. Cell invasion appears tobe independent of the Yersinia virulence plasmid and iscoded by genes on the chromosome of Y enterocolitica (7,14). Strains of pathogenic serotypes (as defined by serotyp-ing) were positive for cell invasion, as measured by a tissueculture assay and by DNA probe assays (14), but strains ofnonpathogenic serotypes were negative.

Evaluation of simple tests used to define pathogenic sero-types of Y. enterocolitica. Since antisera for typing clinicalisolates of Y enterocolitica are not generally available,several simple tests that can be used to identify pathogenicserotypes have been proposed. Table 1 summarizes theexpected results for pathogenic and nonpathogenic sero-types in these tests.

Pyrazinamidase. The pyrazinamidase test was done asdescribed by Kandolo and Wauters (10). Strains were inoc-ulated over the entire slant of pyrazinamide agar, incubatedat 25°C for 48 h, and tested with 1 ml of a freshly prepared1% solution of ferrous ammonium sulfate. A positive pyra-zinamidase reaction was indicated by a pink to brown colorthat developed on the slant.

Salicin-esculin. Salicin and esculin are both 13-glucosidesthat vary in the chemical structure of the compound attachedto D-glucose by a beta linkage. Both compounds are hydro-lyzed by the enzyme 3-glucosidase. Fermentation of salicinwas determined in enteric fermentation base (catalog no.1828-17-9; Difco Laboratories, Detroit, Mich.) with 1%salicin and Andrade indicator (all autoclaved at 121°C for 10min). Esculin hydrolysis was determined in esculin broth (5g of peptone, 1 g of KH2PO4, 5 g of esculin, 0.5 g of ferriccitrate, and 10 ml of Andrade indicator with 990 ml ofdistilled water; all components were autoclaved at 121°C for10 min). Strains were inoculated; incubated at 25°C; and readat days 1, 2, 3 (or 4 or 5), and 7.

D-Xylose fermentation. Strains of Y enterocolitica sero-type 03 do not ferment D-xylose (20). This is in contrast toother serotypes of this species and most of the other speciesof Yersinia that are xylose positive after 1 to 2 days ofincubation at 25°C. Xylose fermentation was determined at25°C in enteric fermentation base with 1% xylose (filtersterilized) and Andrade indicator.CR-MOX agar. In 1989, Riley and Toma (16) described

J. CLIN. MICROBIOL.

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

SIMPLE TESTS FOR PATHOGENIC Y ENTEROCOLITIC4 2591

TABLE 2. Reactions of 25 Y enterocolitica strains isolated during outbreak investigations

Strain no. Source Outbreak 0 Pathogenic Invasiveness .z . Salicin Esculin CR-MOX Autoagglu-no. group serotype (cell culture) J fermentation hydrolysis agar tination

9341-78 (JCF Chinchilla 1 1,2,3 + + + (Weak)A1343)

9286-78 (JCF Human (feces) 2 20 + +A386)

9287-78 (JCF Dog no. (feces) 2 20 + +A387)

9291-78 (JCF Human (cousin) 2 6 - - + + +A400)

9292-78 (JCF Human (aunt) 2 22 - - +A528)

9294-78 (JCF Human (cousin) 2 22 - - +A0530)

9293-78 (JCF Human (grandfa- 2 Unda - + + +A529) ther)

JCF A2650 Human (throat) 3 8 + +9149-79 (JCF Chocolate milk 3 8 + NDb

A2635)9134-79 (JCF 1932 isolate 3 8 + +A3070)

9141-79 (JCF Human (sputum) 3 4,32 + +A4186)

9142-79 (JCF Human (feces) 3 6 - - + + +A4182)

1460-82 (A7525) Human 4 8 + + - - -

1466-82 (A7443) Human 4 8 + + - - -

1464-82 (A7534) Human 4 4,32 + + - - - + +1463-82 (A7532) Human 4 21 + + - - - - -1461-82 (A7527) Human 4 6 - - + + + - -

1459-82 (A7523) Human 4 7,13 - - + + + - -

1467-82 (A7458) Human 4 7,13 - - + + + - -

1458-825 (A7513) Tofu 4 Rough - - + + + - -

1462-82 (A7528) Human 4 Und - - + + + - -

1176-82 Human (stool) 5 13a,13b + + - - - + +1211-82 Human (appendix) 5 13a,13b + + - - -

1568-82 Human (stool) 5 13a,13b + + - - -

8-83 Human (stool) 5 13a,13b + + - - -

a Und, undetermined (no reaction in any of the 0 antisera).b ND, Not done.

CR-MOX agar, which allows visualization of calcium-depen-dent growth and uptake of Congo red dye on the same plate.Cultures were streaked onto CR-MOX agar plates, incu-bated at 36°C, and observed for the presence of small redcolonies at 24 and 48 h. Freshly isolated strains of patho-genic serotypes contain the Yersinia virulence plasmid andare CR-MOX positive (see Fig. 1). When the plates are keptat 36°C for several additional days, it is often possible toobserve the development of a large, colorless colony from asmall red colony.

Autoagglutination in broth that is temperature dependent.Cultures were inoculated into two tubes of MR-VP broth(Difco). One tube was incubated at 36°C and the other wasincubated at 25°C. After 18 to 24 h, the tubes were observedfor agglutination, with care taken not to shake or disturb thesediment at the bottom and along the sides of the tube.Strains of pathogenic serotypes that contain the Yersiniavirulence plasmid agglutinate at 36°C but not 25°C. Strainsthat lack the virulence plasmid do not agglutinate at eithertemperature. Strains that agglutinated at both temperatureswere considered "rough."

RESULTS AND DISCUSSION

Comparison of all results. Table 2 gives the results forindividual strains isolated during outbreak investigations.Table 3 summarizes the results for strains from sporadiccases and stock cultures. Table 4 gives the results for theother species of Yersinia.

Pyrazinamidase test. The pyrizinamidase test was easy todo routinely. Most strains produced a strong pink-browncolor or no color. However, three strains produced a slightamount of color (Table 5) that made it difficult to code themas positive or negative. These were read as "positive weak"and were considered positive in the tabulations. A negativepyrazinamidase test correctly identified 60 of 63 strains ofpathogenic serotypes (95% sensitivity), and a positivepyrazinamidase test correctly identified 34 of the 37 strainsof nonpathogenic serotypes (92% specificity). Thus, thepyrazinamidase test correctly identified 94% of the isolates.If the weak reactions had been coded negative, these per-centages would have changed slightly. In their originaldescription of the pyrazinamidase test, Kandolo and Wau-

VOL. 30, 1992

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

2592 FARMER ET AL.

TABLE 3. Reactions of Y enterocolitica strains isolated from sporadic cases'

% Positive for:

Serotype No. Pathogenicstudied serotype Invasiveness Salicin . . CR-MOX

(tissue culture) Pyraziamidase fermentation Esculm hydrolysis (small red colonies) Autoagglutiation1,2,3 1 Yes 100 100 0 0 100 1002,3 1 Yes 0 0 0 0 03 16 Yes 100 0 0 0 38 194,32 2 Yes 100 0 0 0 0 05,27 6 Yes 100 17 0 0 33 08 9 Yes 100 0 0 0 11 119 3 Yes 100 0 0 0 100 10013a,13b 3 Yes 100 0 0 0 0 018 1 Yes 100 0 0 0 0 020 4 Yes 100 0 0 0 0 021 2 Yes 100 0 0 0 50 50

5 3 No 0 100 100 100 0 06 5 No 0 100 100 100 0 07,13 2 No 0 100 100 100 0 07,19 1 No 0 0 100 100 0 07,8 3 No 0 67 100 100 0 010 2 No 0 50 50 50 0 022 1 No 0 100 100 100 0 027 4 No 0 100 100 100 0 028 1 No 0 100 100 100 0 0Otherb 5 No 0 100 100 100 0 0

a Includes 12 stock cultures: 3 of 09 and 1 each of 02,3; 03; 04,32; 05,27; 08; 018; 020; 021; and 027.b Includes three rough strains and two strains that did not react with any of the antisera used.

ters (10) found 100% accuracy in differentiating pathogenic at 2 days) for strains of Y. enterocolitica (Tables 2 and 3).and nonpathogenic serotypes. However, Riley and Toma Salicin fermentation was easy to read as positive or negative,(16) noted that 5 of their 21 strains of Y enterocolitica but it was difficult to decide how much blackening wasserotype 01,2,3 were pyrazinamidase positive, indicating a required for a positive esculin reaction. A negative reactionnonpathogenic serotype. We also noted that this serotype for salicin-esculin correctly identified all 63 strains of thecan be pyrazinamidase positive (Tables 2 and 3). Our data pathogenic serotypes (100% sensitivity), and a positive re-confirm that the pyrazinamidase test is a simple and reliable sult correctly identified 34 of 37 strains of nonpathogenicway of differentiating pathogenic and nonpathogenic sero- serotypes (92% specificity). Thus, salicin-esculin correctlytypes (10, 16). identified 97% of the isolates, a percentage that was much

Salicin-esculin. There was 100% agreement between sali- better than expected. Wauters et al. (20) did not considercin fermentation and esculin hydrolysis (25°C, final reading esculin hydrolysis to be a particularly useful test and cau-

TABLE 4. Reactions of six Yersinia species other than Y enterocolitica

Species and strain no. Source' Invasiveness Pyrazinamidase Salicin Esculin CR-MOX Autoagglutination(tissue culture) Pyaiaiae fermentation hydrolysis agar

Yersinia aldovae0669-83 Water - +0670-83 Water - + - - - -

Yersinia frederiksenii1461-81 - + + + - -2581-77 Chocolate milk - + + +

Yersinia intermedia0031-83 Human urine - +3208-86 Chocolate milk - + + +

Yersinia kristensenni1458-81 - + - - - -1465-82 Tofu - + - - - -

Yersinia rohdei3022-85 Dog stool - - (Weak) - - - -

3435-85 Human stool - - (Weak) - - - -

Yersinia "group xl"9022-87 - + (Weak) - - - -

9023-87 - + (Weak) - - - -

a A blank indicates that the source is not known.

J. CLIN. MICROBIOL.

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

SIMPLE TESTS FOR PATHOGENIC Y ENTEROCOLITICA 2593

TABLE 5. Strains of Y enterocolitica with results forpyrazinamidase differing from those for salicin-esculin

Serotype and Serotype Invasiveness Pyrazinamidase Salicin-strain (cell culture) esculin

Pathogenic2455-87 01,2,3 + + -9341-78 01,2,3 + + (Weak)3308-85 05,27 + + (Weak)

Nonpathogenic3355-86 07,8 - - +0065-84 07,19 - - +1200-84 010 - - +9292-78 022 - + (Weak)9294-78 022 - + -2062-82 010 - + -

tioned that a positive reaction is dependent on both temper-ature and time of incubation. We found that incubation at25°C with a final reading at 2 days accurately differentiatedpathogenic from nonpathogenic serotypes. The results forsalicin-esculin at 36°C were not as helpful in identifyingstrains of nonpathogenic serotypes (data not shown) becausepositive reactions often occurred after the 48-h cutoff. It maynot be necessary to do esculin hydrolysis routinely since it ismore difficult to read and correlates 100% with salicinfermentation.

Correlation of pyrazinamidase with salicin-esculin. Therewas 91% agreement between the pyrazinamidase test andsalicin-esculin (Table 5). Pyrazinamidase incorrectly identi-fied strains of three pathogenic and three nonpathogenicserotypes. Salicin-esculin incorrectly identified strains ofthree nonpathogenic serotypes (Table 5). However, none ofthe nine strains were incorrectly identified by both thepyrazinamidase and the salicin-esculin tests. Both tests can

be used routinely, and strains that provide conflicting resultscan be studied by other methods.CR-MOX agar. Only 15 of the 63 strains of pathogenic

serotypes produced small red colonies on CR-MOX agar.These results are in agreement with those of others (10, 12,16, 19), who have noted that strains that have been stored orsubcultured several times have often lost the Yersinia viru-lence plasmid. After 24 to 30 h of incubation, it is usuallypossible to identify a pathogenic serotype and determine thepercentage of the population (colonies) that still contains theplasmid (Fig. 1).

Autoagglutination. Autoagglutination in MR-VP broth waseasy to do and was easily adaptable to the daily routine sinceMR-VP broth is readily available in most laboratories. Aswith CR-MOX agar, old strains of pathogenic serotypeswere usually negative. Only 8 of 63 strains were positive.Autoagglutination was usually clear and was easy to read;however, the tubes should be handled gently, because theagglutinated cells are easily resuspended, giving a uniformturbidity that would be read as negative. Although autoag-glutination can be a useful test for fresh isolates, we nolonger use it routinely, because CR-MOX agar provides thesame information and has several advantages.

D-Xylose fermentation. All 16 strains of Y enterocoliticaserotype 03 were xylose negative, in contrast to otherserotypes, which were uniformly positive. Thus, this simpletest was extremely useful in quickly identifying this sero-

type, which is now the most common in the United States (2,5, 13).

Reactions of other Yersinia species. Although the pyrazin-

~~ ~ ~ ~ ~ ~ 'y .......

40.~~~~~~~._vsu@

b b k ,*

st~ **~~~~~3W~~ ~ ~~W

FIG. 1. A pathogenic serotype of Y. enterocolitica on CR-MOXagar (incubated at 36°C for 24 h). The colonies that appear black aredark red on the actual plate; some white colonies can be seen to beemerging from the tiny black colonies.

amidase and salicin-esculin tests were useful in differentiat-ing pathogenic and nonpathogenic serotypes of Y enteroco-litica, they should not be used until a strain has beenidentified to the species level. All strains of the other sixYersinia species were negative for tissue culture invasive-ness and for the presence of the Yersinia virulence plasmid,as evidenced by their colonies on CR-MOX agar and lack ofautoagglutination in MR-VP medium. These tests and manyother reports indicate that strains of the six Yersinia specieslisted in Table 4 are not enteric pathogens (6, 7). However,four strains had very weak pyrazinamidase reactions andmight be read as negative (Table 4). Nine strains weresalicin-esculin negative, a property that is usually found inpathogenic serotypes ofY enterocolitica. Thus, biochemicaltesting is essential for differentiating these other Yersiniaspecies from Y enterocolitica. Noble et al. (15) found thatstrains of some of the other Yersinia species had one or moreproperties perhaps suggestive of pathogenicity, but theseresults have not been confirmed. At present, there is noconvincing evidence that strains of these other Yersiniaspecies are enteric pathogens or are highly invasive fortissue. Further studies are needed to determine whether theyhave a role in human disease other than as opportunisticpathogens.

Until recently it has been very difficult for clinical andpublic health laboratories to determine the pathogenic po-tential of Y enterocolitica strains isolated from feces (1, 9).Improvements in isolation methods for Yersinia such as CIN(cefsulodin-irgasan-novobiocin) medium and cold enrich-

VOL. 30, 1992

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

2594 FARMER ET AL.

ment have made it possible to detect small numbers ofYersinia organisms mixed with normal enteric flora. Al-though this has greatly increased the isolation rate forYersinia strains, many of these stool isolates will not beenteric pathogens. The first step in work with Yersiniastrains is to identify the isolate to the species level, since nofurther testing is needed for species other than Y enteroco-litica. The four simple tests evaluated in the study describedhere can then be used to identify the isolate as a pathogenicor nonpathogenic serotype of Y enterocolitica.

ACKNOWLEDGMENT

V. L. Miller was supported by fellowship DRG-885 from theDamon Runyon-Walter Winchell Cancer Fund.

REFERENCES1. Barteluk, R. L., and M. A. Noble. 1988. Routine culturing of

stool specimens for Yersinia enterocolitica. J. Clin. Microbiol.26:1616. (Letter.)

2. Bissett, M. L., C. Powers, S. L. Abbott, and J. M. Janda. 1990.Epidemiologic investigations of Yersinia enterocolitica and re-lated species: sources, frequency, and serogroup distribution. J.Clin. Microbiol. 28:910-912.

3. Black, R. E., R. J. Jackson, T. Tsai, M. Medvesky, M. Shaye-gani, J. C. Feeley, K. I. E. MacLeod, and A. M. Wakelee. 1978.Epidemic Yersinia enterocolitica infection due to contaminatedchocolate milk. N. Engl. J. Med. 298:76-79.

4. Bottone, E. J. (ed.). 1981. Yersinia enterocolitica. CRC Press,Inc., Boca Raton, Fla.

5. Bottone, E. J. 1983. Current trends of Yersinia enterocoliticaisolates in the New York City area. J. Clin. Microbiol. 17:63-67.

6. Butler, T. 1983. Plague and other Yersinia infections. PlenumMedical Book Company, New York.

7. Cornelis, G., Y. Laroche, G. Ballingad, M.-P. Sory, and G.Wauters. 1987. Yersinia enterocolitica, a primary model forbacterial invasiveness. Rev. Infect. Dis. 9:64-87.

8. Gutman, L. T., E. A. Ottesen, T. J. Quan, P. S. Noce, and S. L.Katz. 1973. An inter-familial outbreak of Yersinia enterocoliticaenteritis. N. Engl. J. Med. 288:1372-1377.

9. Kachoris, M., K. L. Ruoff, K. Welch, W. Kallas, and M. J.Ferraro. 1988. Routine culturing of stool specimens for Yersiniaenterocolitica. J. Clin. Microbiol. 26:1616-1617. (Letter.)

10. Kandolo, K., and G. Wauters. 1985. Pyrazinamidase activity in

Yersinia enterocolitica and related organisms. J. Clin. Micro-biol. 21:980-982.

11. Kay, B. A., K. Wachsmuth, and P. Gemski. 1982. New viru-lence-associated plasmid in Yersinia enterocolitica. J. Clin.Microbiol. 15:1161-1163.

12. Kay, B. A., K. Wachsmuth, P. Gemski, J. C. Feeley, T. J. Quan,and D. J. Brenner. 1983. Virulence and phenotypic characteri-zation of Yersinia enterocolitica isolated from humans in theUnited States. J. Clin. Microbiol. 17:128-138.

13. Lee, L. A., A. R. Gerber, D. R. Lonsway, J. D. Smith, G. P.Carter, N. D. Puhr, C. M. Parrish, R. K. Sikes, R. J. Finton, andR. V. Tauxe. 1990. Yersinia enterocolitica 0:3 infections ininfants and children, associated with the household preparationof chitterlings. N. Engl. J. Med. 322:984-987.

14. Miller, V. L., J. J. Farmer m, W. E. Hill, and S. Falkow. 1989.The ail locus is found uniquely in Yersinia enterocolitica sero-types commonly associated with disease. Infect. Immun. 57:121-131.

15. Noble, M. A., R. L. Barteluk, H. J. Freeman, R. Subramaniam,and J. B. Hudson. 1987. Clinical significance ofvirulence-relatedassay of Yersinia species. J. Clin. Microbiol. 25:802-807.

16. Riley, G., and S. Toma. 1989. Detection of pathogenic Yersiniaenterocolitica by using congo red-magnesium oxalate agar me-dium. J. Clin. Microbiol. 27:213-214.

17. Tacket, C. O., J. Ballard, N. Harris, J. Allard, C. Nolan, T.Quan, and M. L. Cohen. 1985. An outbreak of Yersinia entero-colitica infections caused by contaminated tofu (soybean curd).Am. J. Epidemiol. 121:705-711.

18. Tacket, C. O., J. P. Narain, R. Sattin, J. P. Lofgren, C.Konigsberg, R. C. Redtorff, A. Rausa, B. R. Davis, and M. L.Cohen. 1984. A multistate outbreak of infections caused byYersinia enterocolitica transmitted by pasteurized milk. JAMA251:483-486.

19. Wachsmuth, K., B. A. Kay, and K. A. Birkness. 1984. Diagnos-tic value of plasmid analyses and assays for virulence inYersinia enterocolitica. Diagn. Microbiol. Infect. Dis. 2:219-228.

20. Wauters, G., K. Kandolo, and M. Janssens. 1987. Revisedbiogrouping scheme of Yersinia enterocolitica. Contrib. Micro-biol. Immunol. 9:14-21.

21. Zink, D. L., J. C. Feeley, J. G. Wells, C. Vanderzant, J. C.Vickery, W. D. Roof, and G. A. O'Donovan. 1980. Plasmid-mediated tissue invasiveness in Yersinia enterocolitica. Nature(London) 283:224-226.

J. CLIN. MICROBIOL.

on March 6, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 1993, p. 2248-22500095-1137/93/082248-03$02.00/0

Letters to the EditorIdentification of Pathogenic Serotypes of Yersinia enterocolitica

We were interested to read the report by Farmer et al. (3)on the combined use of pyrazinamidase activity (PYZ),growth on Congo red-magnesium oxalate agar (CR-MOX),D-xylose fermentation (XYL), and salicin fermentation-es-culin hydrolysis (SAL-ESC) for identifying pathogenic sero-groups of Yersinia enterocolitica. We wish to present datafrom a similar, but prospective, study carried out in Italyfrom 1987 to 1992.

During the study period, a total of 1,619 strains of Yenterocolitica of human and nonhuman origin in Italy weretested for PYZ, XYL, and SAL-ESC (all at 25°C for 48 h).No strains had been epidemiologically incriminated in out-breaks. Not included in this study are strains formerlyreferred to as biogroup 3A and 3B of Y enterocolitica andnow classified as Y mollaretii and Y bercovieri, respec-tively.Of the 1,619 strains, 603 represented unquestionably

pathogenic phenotypes as they fit into biogroup 4, serogroup03, phage type VIII or IXa; biogroup 2, serogroup 09, phagetype X3; biogroup 1B, serogroup 08, phage type X,; bio-group 2, serogroup 05,27; or biogroup 3, serogroup 01,2,3.These isolates were uniformly PYZ- and SAL-ESC-. All423 strains belonging to biogroup 4, serogroup 03, wereXYL-, in contrast to other pathogenic and nonpathogenic(see below) phenotypes of Y enterocolitica, which wereuniformly positive. Thus, we agree with the authors thatD-xylose fermentation is very useful for identifying thisbio-serogroup.The remaining 1,016 strains were positive for XYL, PYZ,

and SAL-ESC, with the exception of 19 isolates which wereXYL+ and PYZ+ but SAL-ESC- after 48 h of incubation at25°C. All of these strains were biochemically more likebiogroup 1A (lipase+, indole+, xylose+, trehalose+, ni-trate', pyrazinamidase+, beta-glucosidase+, Voges-Pros-kauer+, proline peptidase negative or positive) (6). From thestandpoint of clinical and public health, we do not considerstrains of biogroup 1A to be enteric pathogens. This conclu-sion was based on (i) their widespread presence in theenvironment, (ii) their high isolation rate from healthy indi-viduals and patients whose histories were not compatiblewith yersiniosis, (iii) lack of their recovery upon primaryplating, regardless of source of isolation, and (iv) lack ofantibody response in serum samples from 34 examinedpatients from whose stools a biogroup 1A strain had beenisolated. Surprisingly, while 0 serotyping isolates of bio-group 1A, we found that single 0 factors representative ofthe main pathogenic strains of Y. enterocolitica often oc-

curred in these environmental strains. We noted this in 65 of997 PYZ+ and SAL-ESC+ strains which included 04,32 (28strains), 03 (19 strains), 021 (9 strains), 08 (5 strains), 020(3 strains), and 018 (1 strain) serogroups. It is also notewor-thy that three nonhuman 08 isolates were PYZ+ and SAL-ESC+ but nevertheless contained plasmid DNA of a sizesimilar to that of plasmid DNA from the well-known humanvirulent 08 strains WA and 8081. Further characterization ofthese plasmids by Southern hybridization after BamHI re-

striction showed the lack of plasmid homology between our

study strains and the virulent 08 strains.

Farmer et al. (3) state that PYZ, XYL, SAL-ESC, andCR-MOX can be used to screen for pathogenic serotypes ofY. enterocolitica since 0 antisera are not generally available.Our results reaffirm that the use of 0 antisera without priorcomplete biochemical characterization of the Y. enterocolit-ica isolates could lead to considerable overestimation ofpathogenic serogroups (1, 2, 4, 5). We agree that simple testssuch as PYZ and SAL-ESC are a primary tool to identify Yenterocolitica isolates as a pathogenic or nonpathogenicphenotype, while 0 typing is important in understanding theepidemiology and ecology of this species.

This work was supported in part by Consiglio Nazionale delleRicerche-Progetto Finalizzato "Fattori di Rischio di Malattia"(FATMA).

REFERENCES1. Chiesa, C., L. Pacifico, V. Cianfrano, and M. Midulla. 1987.

Italian experience with yersiniosis (1978-1985). Contrib. Micro-biol. Immunol. 9:76-88.

2. Chiesa, C., L. Pacifico, A. Guiyoule, G. Ravagnan, and H. H.Mollaret. 1991. Phenotypic characterization and virulence of 08Yersinia strains isolated in Europe. Contrib. Microbiol. Immu-nol. 12:182-191.

3. Farmer, J. J., III, G. P. Carter, V. L. Miller, S. Falkow, and I. K.Wachsmuth. 1992. Pyrazinamidase, CR-MOX agar, salicin fer-mentation-esculin hydrolysis, and D-xylose fermentation foridentifying pathogenic serotypes of Yersinia enterocolitica. J.Clin. Microbiol. 30:2589-2594.

4. Wauters, G. 1981. Antigens of Yersinia enterocolitica, p. 41-53.In E. Bottone (ed.), Yersinia enterocolitica. CRC Press, BocaRaton, Fla.

5. Wauters, G., S. Aleksic, G. Charlier, and G. Schulze. 1991.Somatic and flagellar antigens of Yersinia enterocolitica andrelated species. Contrib. Microbiol. Immunol. 12:239-243.

6. Wauters, G., K. Kandolo, and M. Janssens. 1987. Revised bio-grouping scheme of Yersinia enterocolitica. Contrib. Microbiol.Immunol. 9:14-21.

Claudio ChiesaLucia PacificoGiampietro RavagnanNational Centerfor YersiniosisInstitute of Experimental MedicineCNR-Institute of PediatricsLa Sapienza University ofRomeViale R. Elena 32400161-RomeItaly

Authors' ReplyWe were encouraged by the results of the extensive

survey of human and environmental Y enterocolitica strainsfrom the Italian National Center for Yersiniosis. The factthat only 19 of 1,619 strains had discordant results for thepyrazinamidase and salicin-esculin tests is much more en-

couraging than the 9% we observed in our smaller series ofstrains from human patients in the United States. Theauthors point out that strains with these discordant resultscan be biotyped to determine whether they belong to bio-group 1A, which contains only nonpathogenic strains, butwe are not optimistic that many clinical microbiology labo-

2248

Vol. 31, No. 8

LETTERS TO THE EDITOR 2249

ratories will undertake this specialized testing. We wouldlike to reemphasize that Yersinia strains can be immediatelysubcultured from primary plates to CR-MOX agar. Mostpathogenic strains will be apparent after overnight incuba-tion at 36°C because of the characteristic small red coloniesthey will produce.We used the term pathogenic serotypes because it has

been widely used in the literature to indicate strains that arepotential enteric pathogens. Chiesa et al. encounteredstrains of Y enterocolitica (usually from environmentalsources) that agglutinated in 6 of the 11 sera used to definepathogenic serotypes: 03; 04,32; 08; 018; 020; and 021.However, these strains had other phenotypic propertiesindicating that they are not enteric pathogens. For example,they found 19 strains that agglutinated in 03 antiserum butwere pyrazinamidase, salicin-esculin, and xylose positive.These strains would be classified as serotype 03, a patho-genic serotype, yet they would not be considered entericpathogens. This conflicting nomenclature could easily be

confusing, particularly to those not familiar with the subtle-ties of pathogenicity in the genus Yersinia. We agree withChiesa et al., who used the term pathogenic phenotype, thatthe term pathogenic serotype can occasionally be verymisleading. Perhaps it is time to replace it with a moreprecise term such as pathogenic phenotype, pathogenicbio-serotype, or pathogenic bio-serogroup.

J. J. Farmer IIIG. P. CarterI. K. WachsmuthDivision of Bacterial and Mycotic DiseaseNational Centerfor Infectious DiseasesCenters for Disease Control and PreventionAtlanta, Georgia 30333

V. L. MillerS. FalkowDepartment ofMedical MicrobiologySanford UniversitySanford, Califomnia 94305

Wellcolex Colour Salmonella Test and Selenite-F Broth

I read with interest a recent publication entitled "Evalu-ation of the Wellcolex Colour Salmonella Test for Detectionof Salmonella spp. in Enrichment Broths" (6). However, wewould like to make a number of comments on the results ofthe Wellcolex Colour Salmonella Test in the above publica-tion, as they are significantly different from our own expe-rience with the test (5).

In the study by Rohner et al., the sensitivity of theWellcolex test is lower, even with the Selenite-F broth, thanthose in other studies (1, 3, 5). A number of key factors wereomitted in this study which could explain the lower perfor-mance of the Wellcolex Salmonella Test.

(i) Amount of inoculum. The manufacturer gives preciseinstructions regarding the amount of inoculum to be used fora specific volume of Selenite broth, as it is added at a criticalstage for optimum recovery of Salmonellae. Seeding Selenitebroth with too little or too much would lead to poor growthof salmonellae.

(ii) Emulsification. Emulsification of fecal specimens priorto inoculation is also recommended by the manufacturer, as

this should liberate Salmonella spp. and allow maximumgrowth in this selective environment.

(iii) GN broth. The recommended incubation time for GNbroth is 6 to 8 h (2). In the study by Rohner et al. (6), theincubation time was 18 to 24 h, which would indicate that thelaboratory procedure did not use culture conditions foroptimal recovery of Salmonella spp. This appears to beconfirmed by the fact that the subcultures from the GN brothmissed five Salmonella spp. which were isolated with theprimary plates. There was therefore no point in evaluatingthe performance of the Wellcolex Colour Salmonella Test onthis GN broth. Furthermore, the manufacturer recommendstesting only on Selenite-F Broth.

(iv) Quality of the Selenite-F broth. Lastly, the quality ofthe Selenite-F broth is obviously critical in the recovery ofSalmonella spp. The above study (6) only compared GN andSelenite-F broths, without comparing different Selenitebroths. Our own evaluation (5) has shown quality differencesbetween Selenite-F broths from two manufacturers.

By changing the methodology as described above, wefound that the sensitivity of the Wellcolex Colour SalmonellaTest was 99% on Selenite-F broth (5). It should also be notedthat the salmonella incidence in our study was over 20%,compared with 4% in the study by Rohner et al.We wish to publish this letter in reply to Dr. Rohner's

publication, as we feel that the results of their study areincomplete and do not represent the true performance of theWellcolex Colour Salmonella Test.

REFERENCES1. Crotti, D., M. L. D'Annibale, and G. Fonzo. 1990. Evaluation of

a commercial kit for the rapid diagnosis of salmonella in stoolspecimens, p. 83, abstr. 148. Abstr. 6th Int. Congr. RapidMethods Automation Microbiol. Immunol. Helsinki, 1990.

2. Gilligan, P. H., J. M. Janda, M. A. Karmali, and J. M. Miller.1992. Cumitech 12A, Laboratory diagnosis of bacterial diarrhea.Coordinating ed., F. S. Nolte. American Society for Microbiol-ogy, Washington, D.C.

3. Hansen, W., Y. Glyoczynsky, and E. Yourassowsky. 1990. Evalua-tion of a latex agglutination test for the detection of Salmonella inenrichment broth and comparison with traditional culture methods,abstr. C-223, p. 381. Abstr. 90th Annu. Meet. Am. Soc. Microbiol.1990. American Society for Microbiology, Washington, D.C.

4. Orden, B., and M. A. Amerigo. 1991. Screening of Salmonellaspp. in Selenite broth by a new latex colour test, abstr. C-274, p.387. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. AmericanSociety for Microbiology, Washington.

5. Orden, B., A. Gonzalez, E. Juarez, D. Caravaca, and A. Franco.1991. Wellcolex Colour Salmonella Test for detection of Salmo-nella spp. from selenite broth an/or enteric differential agars.Abstr. 17th Int. Congr. Chemother., Berlin, 1991.

6. Rohner, P., S. Dharan, and R. Auckenthaler. 1992. Evaluation ofthe Wellcolex Colour Salmonella Test for detection of Salmo-nella spp. in enrichment broths. J. Clin. Microbiol. 30:3274-3276.

Beatriz OrdenAna FrancoSecci6n de MicrobiologiaServicio de Analisis ClinicosAmbulatorio "Arguelles"Quintana 1128008 MadridSpain

VOL. 31, 1993