Vol. 17, No. 1JOURNAL OF CLINICAL MICROBIOLOGY, Jan. 1983, p. 128-1380095-1137/83/010128-11$02.00/0

Virulence and Phenotypic Characterization of Yersiniaenterocolitica Isolated from Humans in the United States

BRADFORD A. KAY,1 KAYE WACHSMUTH,1* PETER GEMSKI,2 JAMES C. FEELEY,1 THOMAS J.QUAN,3 AND DON J. BRENNER1

Division of Bacterial Diseases, Center for Infectious Diseases, Centers for Disease Control, Atlanta, Georgia303331; Department of Biological Chemistry, Walter Reed Army Institute of Research, Washington, DC200122; and Division of Vector-Borne Viral Diseases, Center for Infectious Diseases, Centers for Disease

Control, Fort Collins, Colorado 805223

Received 22 July 1982/Accepted 10 October 1982

Yersinia enterocolitica was recently reclassified into Yersinia enterocoliticasensu stricto and three additional species. With this new classification, it was ofinterest to reexamine pathogenicity previously ascribed to Y. enterocolitica. Allavailable clinical isolates of Y. enterocolitica sent to the Centers for DiseaseControl from 1970 through 1980 were selected for characterization and compari-son. One-hundred such strains had been submitted, from 21 states. Most (85%)were biotype 1, and 0:8 was the most common of the 24 serotypes encountered.All strains were examined by several virulence assays. Two strains causedconjunctivitis in guinea pigs, 7 were lethal for mice, 54 invaded HEp2 cells, 18produced a heat-stable enterotoxin, 9 were calcium dependent, 20 autoagglutinat-ed, and 34 had a distinctive colonial morphology at 37°C. Ten isolates of each ofthe new species that had previously been grouped with Y. enterocolitica (Y.kristensenii, Y. intermedia, and Y. frederiksenii) were characterized and weregenerally negative in all assays. This study points out pathogenicity differencesamong Yersinia species, confirms the complex nature of virulence in Y. enteroco-litica, and confirms that no single current assay correlates with virulence in Y.enterocolitica .

A series of publications in 1980 defined thespecies Yersinia enterocolitica and proposed thecreation of three new Yersinia species: Y. inter-media, Y. kristensenii, and Y. frederiksenii (3, 4,8, 9, 28). These proposals were made afterextensive studies of DNA relatedness and bio-chemical characteristics. The newly describedspecies reflect the diversity of biochemical andepidemiological observations of the former atyp-ical or Y. enterocolitica-like organisms. Themore precise definition of Y. enterocoliticasensu stricto may clarify the diverse, and oftenperplexing, characteristics previously ascribedto this organism.The clinical and epidemiological significance

of the three newly described Yersinia species isuncertain at present. Past reports have principal-ly associated strains of these three species withanimal and environmental origins, and their rela-tionships to human disease are unknown (5, 6,26). Although these strains are usually consid-ered to be nonpathogenic for humans, it hasbeen (5) suggested that they may become hostadapted for humans and be clinically significantin the future. Currently, however, Y. enterocoli-tica alone is recognized as the significant human

pathogen of these four species (5).In 1980, we initiated a study of all of the

available (100) Y. enterocolitica sensu strictostrains isolated from humans in the UnitedStates and sent to the Centers For DiseaseControl (CDC) for identification or confirmationover the past 10 years. This study was undertak-en in an attempt to characterize these strains bybiotype, serotype, antibiotic susceptibility, andpotential virulence. Representatives of each ofthe three new species were to be similarly char-acterized and compared with Y. enterocoliticastrains. One of the primary goals of this researchwas to establish the potential pathogenicity ofthese Y. enterocolitica isolates by using well-documented assays that have been reported tobe indicators of virulence (12, 15-17, 24). Ourinitial problem was to relate the results of viru-lence assays to virulence in humans.

In this investigation, we used epidemiological-ly incriminated strains of Y. enterocolitica aspositive control cultures for the virulence assays(12, 14, 32). These strains were analyzed previ-ously for genes governing their virulence formice (12). Our avirulent controls were spontane-ous derivatives of the virulent strains which no

128

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

VIRULENCE AND PHENOTYPES OF Y. ENTEROCOLITICA 129

TABLE 1. Source and geographical origin of 100 strains of Y. enterocolitica

eCnterc Source of Origin of specimen enteric Source of Origin of specimennumber specimen number specimen

U.S.A.GeorgiaGeorigaMarylandU.S.A.U.S.A.U.S.A.North CarolinaNorth CarolinaPennsylvaniaCaliforniaCaliforniaNew JerseyWisconsinWisconsinWsiconsinWisconsinWisconsinWisconsinWisconsinWisconsinWisconsinConnecticutU.S.A.WisconsinWisconsinMississippiNew JerseyNew HampshireMassachusettsU.S.A.U.S.A.MichiganWashingtonWisconsinWisconsinWisconsinWisconsinWisconsinWisconsinConnecticutWisconsinWisconsinWisconsinConnecticutNew JerseyNew JerseyNew JerseyPennsylvaniaConnecticut

1338-771339-771693-779300-789302-789303-789304-789311-789312-789314-789315-789316-789317-789319-789322-789323-789324-789286-789291-789292-78T9294-789295-789297-789299-789369-789381-789383-789385-789387-789394-789395-789468-789484-78256-79

9326-789327-789328-789332-789333-789334-789366-78962-79

1209-792492-792519-792570-79

43-80603-80A2627A2628

UnknownUnknownStoolUnknownStoolBloodBloodStoolStoolSpleenBloodUnknownUnknownBloodUnknownUnknownUnknownStoolStoolStoolStoolStoolUnknownWoundUnknownUnknownUnknownUnknownUnknownWoundUnknownStoolStoolTissueUnknownStoolAbscessStoolStoolStoolTissue

TissueBloodStoolStoolStoolSputumStoolStoolStool

WisconsinWisconsinLouisianaNew YorkCaliforniaNew YorkNew YorkGeorgiaGeorgiaGeorgiaU.S.A.CaliforniaNew YorkNew YorkKentuckyVirginiaAlabamaKentuckyKentuckyKentuckyKentuckyKentuckyFloridaNew YorkCaliforniaCaliforniaNew JerseyNorth CarolinaMarylandNorth CarolinaWashingtonNew YorkConnecticutTexasLouisianaNorth CarolinaTennesseeKentuckyKentuckyKentuckyGeorgiaTexasNorth CarolinaMichiganNew HampshireOregonNew YorkMassachusettsNew YorkNew York

longer contained the implicated genes and whichwere no longer virulent for mice (12). A better ormore accurate definition of virulence for humansmight be achieved by feeding Y. enterocoliticastrains to volunteer human subjects; however,we were obliged to accept the epidemiologicaland microbiological definition.

MATERIALS AND METHODS

The abbreviations used for tests described in thispaper are: CM37, temperature-dependent colony mor-phology; AA, temperature-dependent autoagglutina-tion; CAD, temperature-dependent calcium require-ment; ST, heat-stable toxin production; TC, epithelialcell penetration; EYE, conjunctivitis in guinea pig eye;

VOL. 17, 1983

503-702138-722139-72815-73905-74906-74907-74

1223-751224-751529-752164-752165-752338-753%8-763971-763972-763973-763974-763975-763976-761811-771812-772339-772648-772649-772650-77

67-78200-78

1746-782214-789139-789155-789283-789284-783977-763978-763979-76183-77184-77185-77453-77678-77679-77681-77744-77802-77874-77940-77

1137-771324-77

UnknownUnknownUnknownUnknownUnknownUnknownUnknownStoolStoolStoolStoolStoolBloodStoolStoolStoolStoolStoolStoolStoolUnknownUnknownBloodUnknownUnknownUnknownBloodSputumStoolHuman milkUnknownUnknownStoolStoolStoolStoolStoolStoolStoolStoolGall bladderUnknownUnknownUnknownSputumStoolStoolStoolLymph nodeSputum

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

TABLE 2. Phenotypic characteristics of 100 strains of Y. enterocolitica

Serotype Biotype' Antibiotic resistancesb

0:8NTC'0:340:5,270:13,18NTRdO:TACOMAe 0:40:200:7,80:5,270:7,8,130:7,8,130:5NTX9NTX0:60:4NTR0:7,8NTRNTR0:50:50:8NTR0:50:5NT0:5NTR0:7,8,130:80:80:5NTRNTR0:8NTRNTR0:190:60:6, 0:280:19, 0:160:80:80:200:60:8NTR0:60:60:11,230:8NTR0:180:180:130:13,180:5,270:80:80:8

1

1

1

21

1

1

1

1

1

1

1

NBf1

1

1

1

11

21

1

2

1

1

1

2

1

2

1

1

1

1

NB1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

NB

NB

3

1

1

1

SD,CFSD,CF,AMCB,AMSD,CFCB,AMSD,CF,AMSXT,SDSusceptibleCBCB,CF,AMCB,CF,AMCB,CF,AMCB,SD,CF,AMCB,SD,CF,AMCB,CF,AMCB,CF,AMSDCB,CF,AMCB,SD,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMSD,AMCB,CF,AMCB,CF,AMCB,CF,AMCFCB,SD,CF,AMSD,CF,AMCB,AMSusceptibleSusceptibleSusceptibleCB,SXT,SD,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMCB,CF,AM,SCB,SD,CF,AMCB,CF,AMCB,CF,AMCB,SD,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMSDSD,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMCB,AMCB,SD,CF,AMCB,CF,AMCBCB,CF,AMCB,SD,CF,AMCB,SD,CF,AMSD,CFCFSD,CF

CDCentericnumber

503-702138-722139-72815-73905-74906-74907-74

1223-751224-751529-752164-752165-752338-753968-763971-763972-763973-763974-763975-763976-761811-771812-772339-772648-772649-772650-77

67-78200-78

1746-782214-789139-789155-789283-783977-763978-763979-76183-77184-77185-77453-77678-77679-77681-77744-77802-77874-77940-77

1137-771324-771338-771339-771693-779300-789302-789303-789304-789311-789212-789314-789315-789316-789317-78

130 KAY ET AL. J. CLIN. MICROBIOL.

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

VIRULENCE AND PHENOTYPES OF Y. ENTEROCOLITICA 131

TABLE 2-ContinuedCDCenteric Serotype Biotype' Antibiotic resistancesbnumber

0:80:200:13,180:80:200:6,300:6,31NTR0:180:210:80:6,31NT0:200:180:140:6,310:8O:TACOMA, 0:400:50:180:80:80:200:80:11,230:190:11,130:80:80:130:7,130:30:200:200:60:80:8

1

NB

1

21

1

1

1

1

1

1

1

1

1

1

1

2

1

1

1

1

1

1

NB1

11

1

1

NB1

1

1

1

1

CB,SD,CF,AMCFCB,CF,AMSD,CF,AMCBCB,CF,AMCF,AMCB,CFCB,AMCB,SD,CF,AMCBCB,CF,AMCB,CF,AMCB,SD,CF,AMCB,CF,AMCB,CF,AMCB,CF,AMCF,AMCB,SD,CF,AMSD,CF,AMCB,SD,CF,AMCFSD,CF,AMCB,CFSD,CFCB,CF,AMCB,SD,CF,AMCB,CF,AMCB,SD,CFCB,CFCB,SD,CF,CL,AMCB,CF,AMCB,SD,CF,AMCB,CF,AMCB,SXT,SD,CF,AMCB,SD,CF,AMCFCF

I Biotyping scheme of Bercovier et al. (3).b CF, Cephalothin; AM, ampicillin; CB, carbenicillin; SD, sulfadiazine; S, streptomycin; SXT, trimethoprim-

sulfamethoxazole; CL, colistin.I NT, Not typable, nonreactive.d NTR, Not typable, rough.' New unnumbered serotype.f NB, Not biotypable by the scheme of Bercovier et al.NTX, Not typable, multiple cross-reactions.

and LETH, mouse lethality by intraperitoneal inocula-tion. Positive and negative test results are indicated bya + or -, respectively, after the test abbreviation.Study strains. Human strains of Y. enterocolitica

isolated in the United States were selected as a studypopulation. Isolates were obtained from the Entero-bacteriology Section stock cotlection and had beenreceived by CDC between 1970 and 1980. A total of100 strains met the project criteria (viz., Y. enterocoli-tica of human origin from persons in the UnitedStates). These strains along with their sources andgeographical origins are listed in Table 1.

Virulence control strains. Three previously de-

scribed, outbreak-associated strains of Y. enterocoli-tica were chosen as virulent control cultures: (i) strainWA, originally isolated from a patient with septicemia(14); (ii) strain A2635 recovered from chocolate milkthat caused a foodborne outbreak of gastroenteritis(32); and (iii) strain Y7P, likewise recovered from a

patient with gastroenteritis (12). Spontaneously occur-

ring, avirulent (as measured in three assays for viru-lence) (12) derivatives of each of these strains were

chosen as negative control cultures.Yersiua species controls. Ten each of the newly

described Yersinia species were selected as speciescontrols of unknown virulence. These strains originat-

9319-789322-789323-789284-789286-789291-789292-78T9294-789295-789297-789299-789369-789381-789383-789385-789387-789394-789395-789468-789484-78256-79

9324-789326-789327-789328-789332-789333-789334-789366-78962-79

1209-792492-792519-792570-79

43-80603-80A2627A2628

VOL. 17, 1983

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

J. CLIN. MICROBIOL.

ed from a variety of sources and locations and aregenerally considered to be nonpathogenic, or at leastnot associated with disease typical of that caused by Y.enterocolitica (4, 9, 28).

Culture storage. All study and control cultures werestocked at room temperature on blood agar base slantswith paraffin-coated corks before June 1981. During1981 and 1982, all study and control strains weresubcultured and also stocked at -70°C in 50o glycer-ol-50% brain heart infusion broth.

Biotype. All Yersinia strains were biochemicallycharacterized, and the Y. enterocolitica strains wereassigned to one of the five biotypes defined by Berco-vier et al. (3). The API 20E Enteric Identificationsystem (Analytab Products, Plainview, N.Y.) wasused for biochemical identification. Tests were doneaccording to the manufacturer's instructions, and thecultures were incubated at 25°C.

Serotype. Serotyping was done by a modification ofthe Widal whole cell agglutination method (30). Themodification and a description of various antigens andantisera have been published elsewhere (21). Refer-ence cultures to confirm the specificity of antiserawere obtained for serotypes 0:1 through 0:23 and0:25 through 0:34 from H. Bercovier, Pasteur Insti-tute, Paris, France.

Antibiotic susceptibilities. Antimicrobial susceptibil-ity tests were performed according to standard proce-dures (1, 2, 10). Antibiotics were chosen on the basisof their clinical, epidemiological, and genetic appropri-ateness. Inoculated plates were incubated for 18 to 24h at 25°C. Zones of bacterial inhibition were measuredto the nearest millimeter and recorded.CM37. A unique colonial morphology designated

CM37+, was used to select Y. enterocolitica coloniesassociated with virulence for mice (J. Lazere and P.Gemski, Abstr. Annu. Meet. Am. Soc. Microbiol.1980, Bli, p. 19). All study and control strains werestreaked for isolated colonies on tryptic soy agarplates. The plates were then incubated at 37°C for 18 to24 h. After incubation, plates were examined with abinocular dissecting microscope, using obliquelytransmitted light. Small, smooth, opaque colonies witha somewhat raised crown were isolated as presump-tive virulent clones (CM37+). Presumptively avirulentcolonies were larger, flatter, and more translucent.AA. AA of Y. enterocolitica has been reported to be

associated with bacterial virulence in mice (15). Meth-ods are described elsewhere (15).CAD. Magnesium oxalate agar was prepared as

described elsewhere (12). Test strains of Yersinia weregrown in brain heart infusion broth overnight at 25°Cand diluted with sterile saline (0.85% NaCl) to aconcentration of approximately 104 cells per ml. Dupli-cate samples containing 10 ,ul of the diluted bacterialsuspension were spread over the entire surface of twomagnesium oxalate agar plates. Both plates were incu-bated for 48 h, one plate at 25°C and one plate at 37°C.Growth on the two plates was compared. The 25°Cplate served as a growth standard (uninhibited). Inhibi-tion of bacterial growth at 37°C (fewer numbers ormarkedly smaller colonies) was evidence of a positivetest (CAD+). Equal numbers of similarly sized colo-nies on magnesium oxalate agar plates at both tem-peratures were indicative of a negative test (CAD-).Toxin production. The production of a heat-stable

toxin by study strains was determined by the infant

mouse assay (11). Strains were grown at 25°C for 48 hwith aeration in 3 ml of tryptic soy broth with 0.6%yeast extract. Filtrates containing Evans blue dyewere injected intragastrically in 2- to 4-day-old mice.After 2.5 h of incubation at room temperature, micewere sacrificed, and a ratio of the intestinal weight tothe remaining body weight was determined. A ratio ofgreater than 0.083 was considered ST+ (7, 13). Theenzyme-linked immunosorbent assay procedure de-scribed by Yolken et al. (31) was used to assay for theproduction of labile toxin.HEp-2 penetration and monolayer detachment. HEp-

2 was maintained by the CDC Tissue Culture Labora-tory. The following modifications were made of previ-ously described assays of TC (16, 20, 27). Tissueculture flasks (25 cm2; Corning Glass Works, Corning,N.Y.) were seeded with approximately 1.5 x 10'Mycoplasma-free HEp-2 cells and grown to near con-fluency. Before bacterial inoculation, monolayerswere washed three times with phosphate-buffered sa-line, pH 7.4. A broth culture of approximately 108 cellswas prepared by inoculating 5 ml of tryptic soy brothand incubating at 25°C for 18 h. After incubation, cellswere pelleted and then suspended in 2 ml of Dulbeccomodified minimum essential medium (GIBCO Labora-tories, Grand Island, N.Y.) with 2% fetal calf serum.This inoculum was placed on the washed monolayerand allowed to remain for 90 min at room temperature.Monolayers were then washed three times with phos-phate-buffered saline fixed in absolute methanol,stained for 15 min with undiluted Giemsa stain, andobserved microscopically. Two parameters were ex-amined: the ability of strains to penetrate HEp-2 cellsand the ability of strains to cause toxic detachment ofHEp-2 monolayers. The monolayer detachment assaywas performed as above with modifications describedelsewhere (20).

Electron microscopy. Electron microscopic exami-nation of HEp-2 cells incubated with Y. enterocoliticawas accomplished by fixing the monolayers in situwith 2% (vol/vol) glutaraldehyde in 0.1 M S-collidinebuffer for 1 h and then rinsing with 0.1 M S-collidinebuffer. Cells were subsequently removed from theflask, centrifuged, and pelleted. After postfixation for45 min in 1% (voIlvol) OS04 in S-collidine buffer,pellets were stained in a saturated solution of uranylacetate in 70% ethanol for 3 h, dehydrated throughgraded concentrations of ethanol, and embedded inresin. Thin sections were cut, mounted on bare coppergrids, and examined with a Philips 200 electron micro-scope (Philips Electronic Instruments, Mt. Vernon,N.Y.) at 40 kV accelerating voltage.EYE. Due to the expense and support systems

necessary for guinea pig maintenance, we selected 46Y. enterocolitica strains for testing in experimentalanimals (24, 32). All strains giving positive reactions inany other assay were chosen, and all strains of sero-type 0:8 were chosen because this serotype has beenclosely associated with invasive human disease in theUnited States (14, 20, 32). Additionally, two represen-tatives of each Yersinia species control were tested(total of six). Therefore, 52 cultures were assayed.

Tryptic soy agar plates were inoculated for maxi-mum growth by swabbing with an overnight brothculture (brain heart infusion broth, 25°C) of the strainto be tested. After incubation at 25°C for 48 h, growthfrom the tryptic soy agar plates was harvested and

132 KAY ET AL.

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

VIRULENCE AND PHENOTYPES OF Y. ENTEROCOLITICA 133

8 8

0:5 0:20 0:6 0:18 0:5,27 0:19 0:3 0:6,280:6,31 0:13 0:14 0:6,300:11,23 0:7,8 0:21 0:7,130:13,18 0:7,8,13 0:34 0:16,19

0:40 0:Tacoma,40:Tacoma,40

FIG. 1. Serotypes of 100 strains of Y. enterocolitica. NT*, Not typable; **, per serotype.

emulsified in 2 to 3 drops of sterile saline. One dropwas introduced into the right eye of an adult femaleguinea pig (CDC Animal Facility, Lawrenceville, Ga.).Animals were examined daily for a period of up to 7days for evidence of eye infection characterized byswelling, eyeball depression, accumulation of fluid,mucous, blood, or corneal opacity. Failure to developevidence of infection within 7 days was regarded as anegative test (EYE-). Transient irritation (redness,lacrimation) was not considered significant.LETH. Pairs of adult ICR mice were inoculated

intraperitoneally with approximately 5 x 106 bacterialcells in 0.5 ml. The inoculum was prepared by growingstrains in 5 ml of brain heart infusion broth at 25°C for18 h, pelleting cells, and suspending in sterile saline tothe turbidity of a 0.5 McFarland standard (1). Theinoculum was confirmed by agar plate counts. Afterintraperitoneal inoculation, mice were observed dailyfor up to 21 days for signs of illness or death. Deathsoccurring within 24 h of inoculation were attributed toendotoxic shock, in which case a second pair of micewas inoculated.

RESULTSBiotyping results are presented in Table 2. As

shown, 85 strains (85%) were of biotype 1, 7%were of biotype 2, and 1 strain (1%) was ofbiotype 3. No strains were biotype 4 or 5. Of the100 strains examined, 7 were not biotypable bythe scheme of Bercovier et al. (3).

The Y. enterocolitica study strains contained24 serotypes. These results are summarized inTable 2 and Fig. 1. Serotype 0:8 occurred mostfrequently (24%) followed by serotypes 0:5(8%), 0:20 (8%), 0:6 (6%), and 0:18 (5%).Eighteen strains were not typable. The remain-ing strains were distributed among 19 serotypicpatterns with low frequency.

Antimicrobial resistances for Y. enterocoliticastrains are given in Table 2. Most strains of allfour species were resistant to cephalothin, ampi-cillin, and carbenicillin. With the exception of Y.frederiksenii, all strains showed some degree ofresistance to sulfadiazine: Y. enterocolitica,37%; Y. kristensenii, 20o; and Y. intermedia,10%. Y. enterocolitica alone showed resistanceto trimethoprim-sulfamethoxazole (3%), colistin(1%), and streptomycin (1%). One strain of Y.intermedia was resistant to chloramphenicol(10%).The experimental results of all virulence as-

says for Y. enterocolitica control strains arepresented in Table 3, and those for study strainsare presented in Table 4. Data for Yersiniaspecies control strains are presented in Table 5.A total of 34 Y. enterocolitica strains represent-ing 12 serotypes were CM37+. Approximately50% of colonies from CM37+ strains had the

VOL. 17, 1983

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

134 KAY ET AL.

TABLE 3. Results of seven assays for virulence-associated characteristics of Y. enterocolitica control

strainsTest

StrainCM37 AA CAD ST TC EYE LETH

WA+ + + + + + + +WA_ a _+ +

A2635 + + + - + + +TAMU-75b - - - - + - -

a Cured derivative of WA+.b Cured derivative of A2635.

virulence-associated morphology. In CM37-cultures, no colony variations were observed.All virulent controls were CM37+, and all aviru-lent controls were CM37- (Table 3). Strains ofthe other Yersinia species were uniformly nega-tive (Table 5).

Overall, 20 strains of Y. enterocolitica repre-senting six serotypes (Table 4) were AA+. Viru-lent control strains were positive, and avirulentcontrol strains were negative. The other Yer-sinia species were uniformly AA-.

Results of the CAD assay are also presentedin Table 4. Nine strains representing nine sero-types of Y. enterocolitica were CAD+ by thistest. All virulent controls were positive, andavirulent controls were negative. No strains ofthe other Yersinia species were CAD+.

Eighteen Y. enterocolitica study strains repre-senting nine different serotypes were ST+ (Ta-ble 4). One virulent control strain was positive;its avirulent derivative was also ST+ (Table 3).Additionally, seven strains (23%) of the otherYersinia species were ST+. Five of these sevenwere strains of Y. kristensenii, and the remainingtwo strains were one strain each of Y. interme-dia and Y. frederiksenii (see Table 5). All strainsof the four Yersinia species tested were negativefor labile toxin production as measured by en-zyme-linked immunosorbent assay.A total of 54 Y. enterocolitica strains repre-

senting 14 serotypes were positive for HEp-2cell penetration (TC+) (Table 4). Virulent con-trols were TC+, but avirulent controls were alsoTC+ (Table 3). Two strains of Y. intermediawere TC+; no strain of Y. frederiksenii or Y.kristensenii was positive in this assay. The pene-

tration results obtained with light microscopywere confirmed by electron microscopy, whichrevealed intracellular bacteria with TC+ Yer-sinia strains. The monolayer detachment assay

was negative with both study and virulencecontrol strains.LETH determination showed that seven

strains representing four serotypes wereLETH+ (Table 4). All virulent control strainskilled mice, and no avirulent derivative killedmice (by definition). No environment or Yer-

sinia species control strain killed mice.Two study and both virulence control strains

of Y. enterocolitica were EYE+; all of thesestrains were serotype 0:8. None of the otherYersinia species were EYE+.

DISCUSSIONAs stated above, the Y. enterocolitica strains

of known human virulence isolated in the UnitedStates are those virulent control cultures pre-sented in Table 3. There is an unavoidablecircularity involved in selecting avirulent deriva-tives of these by an assay (LETH) rather than byusing human volunteers. Absence of mouse le-thality, in our analysis, is given greater weightthan all other assays postulated as measures ofyersinial virulence for humans (12, 13, 15, 17,19-21, 23, 24). There was no strain of Y. entero-colitica or other Yersinia species which wascomparable in all assays to the virulence controlcultures, i.e., CM37+, AA+, CAD+, ST+/-,TC+, EYE+, LETH+. As a result, we consid-ered each virulence assay separately.

In the past, serotyping of Y. enterocolitica hasbeen the primary tool for the identification ofthose clinically and epidemiologically significantstrains (21). Although the recent definition of Y.enterocolitica sensu stricto has eliminated atleast 14 of the 53 previously described serofac-tors, recent clinical references fail to reflect this(22, 29). Additionally, our experience and datasuggest that the numerous and high-titer cross-reactions (indicating multiple serofactors perstrain in many circumstances) and lack of reac-tions (18%) make interpretation difficult. Sero-type does appear to be a useful strain character-istic for clinical and epidemiological purposes,but it is not easily adapted to the routine labora-tory and may not always predict virulence.Examples of this are the eight serotype 0:8strains which do not invade the guinea pig eye,the two which do, and the two serotype 0:8strains which are unreactive in any of the viru-lence-associated assays. There are also fivestrains containing the serofactor 0:8 plus otherfactors, and these give equally ambiguous re-sults, although none invade the eye tissues.Although serotype 0:9 is the second most

common human serotype on a global basis (e.g.,Europe, South Africa, Japan, and Canada), noneof our U.S. study strains were 0:9 (25). Similar-ly, only one strain was identified as serotype0:3, which is the most common cause of diseasein neighboring Canada and in much of Europe(25). This occurrence of specific serotypes in aparticular, well-defined geographical area seemsto be a characteristic of Y. enterocolitica in thisstudy, and as previously reported.

Before our study, the CM37 procedure hadbeen used exclusively with Y. enterocolitica

J. CLIN. MICROBIOL.

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

VIRULENCE AND PHENOTYPES OF Y. ENTEROCOLITICA 135

TABLE 4. Results of seven assays for virulence-associated characteristics of 100 strains of Y. enterocoliticaStrain Testa Strain Testaidenti- identi-fication CM37 AA CAD ST TC EYEb LETH fication CM37 AA CAD ST TC EYEb LETH

+ ++ + +

+ ++ + + +

+ +

+ + + + +

+

+ + + _ +

+ + -

+ -

+ -

+ -

+ + + -

+

+

+

+ +

LI

9155-789283-789284-789286-789291-789292-78T9294-789295-789297-789299-789300-789302-789303-789304-789311-789312-789314-789315-789316-789317-789319-789322-789323-789324-789326-789327-789328-789332-789333-789334-789366-789369-789381-789383-789385-789387-789394-789395-789468-789484-78256-79962-79

1209-792492-792519-792570-79

43-80603-80A2627A2628

++ +

++

+

++

++++

+

+++

+

+

+

+

+

+

+

++

+

+++++

+

++ +

++

+

+

+

+

+ -

+ ±

+ -

+ +

+++

+ + + +

++

++ + + + ++ + + + +

a Only positive reactions are listed for assays other than EYE; all strains were tested in each assay otherthan EYE.

b Only 46 of the 100 strains were tested in EYE (see the text).

+

serotypes 0:3 and 0:8. Our data indicate that nofewer than 12 serotypes have CM37+ strains,suggesting that the colony morphology associat-ed with mouse virulence may be widespreadwithin Y. enterocolitica, but is not present in the

newly designated Yersinia species. This associa-tion with virulence is strengthened by the occur-rence of CM37-avirulent control cultures.

Laird and Cavanaugh (15) first described theAA phenomenon and reported that 25 of 180 Y.

+

+

+

503-702138-722139-72815-73905-74906-74907-74

1223-751224-751529-752164-752165-752338-753968-763971-763972-763973-763974-763975-763976-763977-763978-763979-76183-77184-77185-77453-77678-77679-77681-77744-77802-77874-77940-77

1137-771324-771338-771339-771693-771811-771812-772339-772648-772649-772650-77

67-78200-78

1746-782214-789139-78

+

++

+

+

VOL . 17, 1983

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

136 KAY ET AL.

TABLE 5. Results of seven assays for virulence-associated characteristics of 30 Yersinia (not Y.

enterocolitica) control strainsNo. of isolates positive for:

CM37 AA CAD ST TC EYEa LETH

Y. frederikseniib 0 0 0 1 0 0 0Y. kristenseniib 0 0 0 5 0 0 0Y. intermediab 0 0 0 1 2 0 0

a Six strains, two from each species.b n = 10.

enterocolitica-complex (before the 1980 taxo-nomic division) strains were AA+. All 25 oftheir AA+ strains were reported to be virulentfor mice by the oral route. None of their AA-strains were virulent by this method. They feltthat this test enabled them to accurately distin-guish virulent and avirulent strains of Yersinia.Our avirulent controls were accordingly AA-.Selecting by colonial morphology at 37°C andtesting by AA reaction revealed that 14 Y.enterocolitica strains were positive by both as-says. Nineteen CM37+ strains were AA-, andsix AA+ strains were CM37-. Although bothassays have been reported as measures of mousevirulence, only 7 of our 14 CM37+, AA+ strainswere lethal for mice. All LETH+ strains wereCM37+ and AA+.AA+ reactions have been reported for strains

of the presumptive pathogenic serotypes 0:3,0:8, and 0:9 and, more recently, for strains ofserotypes 0:5,27, 0:4,32, and 0:21 (23). Ourdata suggest four previously unreported AA+ Y.enterocolitica strains of serotype 0:20, 0:13,0:19, and 0:40. Many of these serotypes arepoorly, or not at all, associated with humanillness. One of the most interesting aspects ofthese data is that all of the 30 Yersinia speciescontrol strains were both CM37- and AA-.Only nine of the Y. enterocolitica strains

expressed calcium dependency, and thesestrains were generally positive in other assays;each strain represented a different serotype.Despite the fact that virulence control strainsand other 0:8 study strains gave very clearlydefined reactions as described in the literaturefor these and 0:3 serotype strains (12) and allstrains of other Yersinia species were clearlyCAD-, our findings indicate that use of this testas a screen for virulence may have limitations.For example, one strain, 9292-78, was CAD+and negative for all other assays of virulence. Inaddition, some of these data from our isolateswere equivocal, requiring subjective interpreta-tion of growth that occurred on magnesiumoxalate agar plates at 370C.Although there have been published reports

that the majority of human isolates of Y. entero-

colitica are ST+ (16-18), the clinical presenta-tion of yersiniosis and our data indicate that aminority of human clinical isolates of Y. entero-colitica are ST+. One virulent control culturewas ST-, and although one virulent controlculture was ST+, the avirulent derivative wasalso ST+. Additionally, our "implied avirulent"group of Yersinia species control strains con-tained a higher percentage of ST+ than theclinical isolates. In general, these data do notsupport the previously observed association ofST+ strains with human, rather than environ-mental, isolates of yersiniae.As mentioned above, there is no clinical indi-

cation of enterotoxin production in yersiniosis.The labile toxin data confirm this. Labile toxinproduction has not been reported for yersiniae inthe past, but has been found for Vibrio cholerae,V. mimicus, Aeromonas, Escherichia coli, Ci-trobacter diversus, and other enteric organisms(22). Other investigators have probably foundsimilar and uniformly negative results for yersin-iae, which were not submitted for publication.The use of tissue culture cell penetration as an

experimental model of in vivo invasive disease isconsistent with the usual illness produced by Y.enterocolitica. However, epithelioid cells in cul-ture may be more susceptible to penetrationthan intact tissues in vivo, or penetration maynot reflect true invasion. Accordingly, Formal-in- or UV-killed Yersinia have been reported topenetrate (internalize) HeLa cells (19), and theavirulent controls (Table 5) in this study re-mained TC+ (19). Conversely, Yersinia speciescontrol strains were significantly (93%) TC-.We speculate that Yersinia strains which areunable to penetrate cells in culture are probablynot invasive for intact human tissues. The TCassay, which is simple and rapid to perform,may measure an incomplete but prerequisitecharacteristic of virulence. Results in almost allcases were unequivocal.The tissue culture detachment assay was un-

satisfactory in our hands; one strain which wasreported positive by another laboratory usingthis procedure (20) gave negative results withour HEp2 cell line. Neither of our pathogeniccontrol strains caused monolayer detachmentafter repeated examinations. Although promis-ing in other laboratories, these uniformly nega-tive results with known TC+ control strainsdiscouraged our use of this assay with otherstudy strains.An animal model assaying for the actual inva-

sion of tissues with subsequent in vivo systemicdisease may best reflect the clinical presentationassociated with human yersiniosis (ergo, yersin-ial virulence?). For this reason, we tested allstudy and control strains for mouse lethality. Bydefinition, virulence control strains were

J. CLIN. MICROBIOL.

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

VIRULENCE AND PHENOTYPES OF Y. ENTEROCOLITICA 137

LETH+, and avirulent derivatives as well asYersinia species control strains were LETH-.The seven LETH+ study strains were CM37+,AA+, and TC+. These data indicate that theLETH characteristic may require or depend onthe presence of other strain characteristics suchas cell wall properties (AA) and the ability tocolonize or penetrate host epithelial cells or both(TC). Two of the seven LETH+ strains wereEYE+; these were both serotype 0:8. Howev-er, no other LETH+ or 0:8 strain was EYE+.This was true for virulence control cultures also.We can say that, in our study, having the 0:8serofactor and being LETH+ appear to be nec-essary for EYE+ strains, but that relationship isnot reciprocal.Using the three United States Y. enterocoliti-

ca strains obtained during outbreaks of humanillness as a standard against which to measurevirulence, no study strain is fully virulent. Wepropose that this is not so much a reflection ofthe study population as it is of the positivecontrol population. More outbreak-associatedstrains are needed, especially strains of sero-types known to cause disease in other countries(23, 25). Based on data from this study andconclusions drawn by other investigators (12,23), we feel that the LETH assay best predictsyersinial virulence as measured in other assaysand that its invasive disease may partially reflecthuman disease. It is less costly and less variablethan most animal models and has an unequivocalendpoint.Our data confirm the uniqueness of Y. entero-

colitica sensu strictu, but indicate that diversityremains within the species, especially regardingthe strain characteristics studied here. It is cer-tainly not possible to say that all strains of Y.enterocolitica are pathogenic for humans andthat Y. intermedia, Y. kristensenii, and Y. fre-deriksenii are nonpathogenic.

ACKNOWLEDGMENTS

We thank Carey Callaway for the excellent electron micros-copy support, David Fraser and Roger Feldman for sugges-tions regarding the preparation of the manuscript, Dan Port-noy for sharing bacterial strains, and Patsy Bellamy for typingthe manuscript.

This research was supported in part by the CDC and theUniversity of North Carolina, cooperative agreement no. U20/CCU400277-03.

LITERATURE CITED

1. Barry, A. L., and C. Thornsberry. 1980. Susceptibilitytesting: diffusion test procedures, p. 463-474. In E. H.Lennette, A. Balows, W. J. Hausler, Jr., and J. P. Truant(ed.). Manual of clinical microbiology, 3rd ed.. AmericanSociety for Microbiology, Washington, D.C.

2. Bauer, A. W., W. M. M. Kirby, J. C. Shenis, and M.Turek. 1966. Antibiotic susceptibility testing by a stan-dardized single disk method. Am. J. Clin. Pathol. 45:493-496.

3. Bercovier, H., D. J. Brenner, J. Ursing, A. G. Stelgerwalt,G. R. Fanning, J. M. Alonso, G. P. Carter, and H. H.

Mollaret. 1980. Characterization of Yersinia enterocoliticasensu stricto. Curr. Microbiol. 4:201-206.

4. Bercovier, H., J. Ursing, D. J. Brenner, A. G. Steigerwalt,G. R. Fanning, G. P. Carter, and H. H. Mollaret. 1980.Yersinia kristensenii: a new species of Enterobacteriaceaecomposed of sucrose-negative strains formerly called Yer-sinia enterocolitica or Yersinia enterocolitica-like. Curr.Microbiol. 4:219-224.

5. Bottone, E. J. 1978. Atypical Yersinia enterocolitica: clin-ical and epidemiological parameters. J. Clin. Microbiol.7:562-567.

6. Bottone, E. J., B. Chester, M. S. Malowany, and J. Aller-hand. 1974. Unusual Yersinia enterocolitica isolates notassociated with mesentery lymphadenitis. Appl. Microbi-ol. 27:858-861.

7. Boyce, J. M., D. J. Evans, Jr., D. G. Evans, and H. L.DuPont. 1979. Production of heat-stable, methanol-solu-ble enterotoxin by Yersinia enterocolitica. Infect. Immun.25:532-537.

8. Brenner, D. J., H. Bercovier, J. Ursing, J. M. Alonso,A. G. Steigerwalt, G. R. Fanning, G. P. Carter, and H. H.Mollaret. 1980. Yersinia intermedia: a new species ofEnterobacteriaceae composed of rhamnose-positive, me-libiose-positive, raffinose-positive strains formerly calledatypical Yersinia enterocolitica or Yersinia enterocolitica-like. Curr. Microbiol. 4:207-212.

9. Brenner, D. J., J. Ursing, H. Bercovier, A. G. Steigerwalt,G. R. Fanning, J. M. Alonso, and H. H. Mollaret. 1980.Deoxyribonucleic acid-relatedness in Yersinia enterocoli-tica and Yersinia enterocolitica-like organisms. Curr. Mi-crobiol. 4:195-200.

10. Cornelis, G. 1981. Antibiotic resistance in Yersinia entero-colitica, p. 55-71. In E. Bottone (ed.), Yersinia enterocoli-tica. CRC Press, Boca Raton, Fla.

11. Dean, A. G., Y. Ching, R. G. Williams, and L. B. Harden.1972. Test for Escherichia coli enterotoxin using infantmice: application in a study of diarrhea in children inHonolulu. J. Infect. Dis. 125:407-411.

12. Gemski, P., J. R. Lazere, and T. Casey. 1980. Plasmidassociated with pathogenicity and calcium dependency ofYersinia enterocolitica. Infect. Immun. 27:682-685.

13. Kapperud, G., B. Berdal, and T. Omiand. 1980. Briefreport: enterotoxin production by Yersinia enterocoliticaand Yersinia enterocolitica-like microbes at 22'C and37rC. Acta Pathol. Microbiol. Scand. Sect. B 88:65-67.

14. Keet, E. E. 1974. Yersinia enterocolitica septicemia:source of infection and incubation period identified. N.Y.State J. Med. 74:2226-2230.

15. Laird, W. J., and D. C. Cavanaugh. 1980. Correlation ofautoagglutination and virulence of yersiniae. J. Clin.Microbiol. 11:430-432.

16. Lee, W. H. 1979. Testing for the recovery of Yersiniaenterocolitica in foods and their ability to invade HeLacells, p. 228-233. In P. B. Carter, L. Lafleur, and S. Toma(ed.), Contributions to microbiology and immunology,vol. 5. S. Karger, Basel.

17. Pai, C. H., and V. Moors. 1978. Production of enterotoxinby Yersinia enterocolitica. Infect. Immun. 19:908-911.

18. Pai, C. H., V. Mors, and S. Toma. 1978. Prevalence ofenterotoxigenicity in human and nonhuman isolates ofYersinia enterocolitica. Infect. Immun. 22:334-338.

19. Pedersen, K., S. Winblad, and V. Bitsch. 1979. Studies onthe interaction between different 0-serotypes of Yersiniaenterocolitica and HeLa cells. Acta Pathol. Microbiol.Scand. Sect. B 87:141-145.

20. Portnoy, D. A., S. L. Moseley, and S. Falkow. 1981.Characterization of plasmids and plasmid-associated de-terminants of Yersinia enterocolitica pathogenesis. Infect.Immun. 31:775-782.

21. Quan, T. J. 1979. Biotypic and serotypic profiles of 367Yersinia enterocolitica cultures of human and environ-mental origin in the United States, p. 83-87. In P. B.Carter, L. Lafleur, and S. Toma (ed.), Contributions tomicrobiology and immunology, vol. 5. S. Karger, Basel.

22. Sack, R. B., R. Tilton, and A. S. Weissfield. 1980. Cumi-

VOL. 17, 1983

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

J. CLIN. MICROBIOL.

tech 12, Laboratory diagnosis of bacterial diarrhea. Coor-dinating ed., S. J. Rubin. American Society for Microbiol-ogy, Washington, D.C.

23. Schiemann, D., J. Devenish, and S. Toma. 1981. Charac-teristics of virulence in human isolates of Yersinia entero-colitica. Infect. Immun. 32:400-403.

24. Sereny, B. 1955. Experimental Shigella keratoconjunctivi-tis. Acta Microbiol. Acad. Sci. Hung. 2:293-296.

25. Toma, S. 1973. Survey of the incidence of Yersiniaenterocolitica in the Province of Ontario. Can. J. PublicHealth 64:477-487.

26. Toma, S., L. Lafleur, and V. Deidrick. 1979. Canadianexperience with Yersinia enterocolitica (1966-1977), p.144-149. In P. B. Carter, L. Lafleur, and S. Toma (ed.),Contributions in microbiology and immunology. vol. 5. S.Karger, Basel.

27. Une, T. 1977. Studies on the pathogenicity of Yersiniaenterocolitica. II. Interaction with cultured cells in vitro.Microbiol. Immunol. 21:365-377.

28. Ursing, J.. D. J. Brenner, H. Bercovier, G. R. Fanning,A. G. Steigerwalt, J. M. Alonso, and H. H. Mollaret. 1980.Yersinia frederiksenii: a new species of Enterobac-teriaceae composed of rhamnose-positive strains (former-ly called atypical Yersinia enterocolitica or Yersinia en-terocolitica-like). Curr. Microbiol. 4:213-218.

29. Weissfield, A. P. 1981. Yersinia enterocolitica. Clin. Mi-crobiol. Newsletter 3:91-93.

30. Winblad, S. 1968. Studies on 0-antigen factors of Yersiniaenterocolitica. In Symposia Series in ImmunobiologicalStandardization vol. 9. S. Karger, Basel.

31. Yolken, R. H., H. Greenberg, M. Merson, R. Sack, and A.Kapikian. 1977. An enzyme-linked immunosorbent assayfor detection of Escherichia coli heat-labile enterotoxin. J.Clin. Microbiol. 6:439-444.

32. Zink, D. L., J. C. Feeley, C. Vanderzant, J. C. Vickery,and G. A. O'Donovan. 1978. Possible plasmid-mediatedvirulence in Yersinia enterocolitica. Trans. Gulf CoastMol. Biol. Conf. 3:156-163.

138 KAY ET AL.

on April 21, 2020 by guest

http://jcm.asm

.org/D

ownloaded from