Vol. 59, No. 1

INFECTION AND IMMUNITY, Jan. 1991, p. 162-167

0019-9567/91/010162-06$02.00/0 Copyright C) 1991, American Society for Microbiology

Pathogenicity of Yersinia kristensenii for Mice ROY M.

ROBINS-BROWNE,'* SAM CIANCIOSI,l ANNE-MARIE BORDUN,1 AND

GEORGES WAUTERS2

Department of Microbiology, University of Melbourne, Parkville, Victoria 3052, Australia,1 and Unite de Microbiologie, Universite de Louvain, Brussels, Belgium2 Received 31 July 1990/Accepted 8 October 1990

Forty-seven strains of Yersinia kristensenii from widely differing sources, representing all known 0 of this species, were investigated for virulence with a variety of animal and in vitro assays. Twenty-four (51%) of the isolates were lethal for mice pretreated with iron dextran. Mouse-lethal strains occurred predominantly within 0 serogroups 0:11, 0:12,25, and 0:16. Virulent Y. kristensenii strains generally did not express the virulence-associated phenotype (Ca2" dependence and binding of Congo red and crystal violet) which characterizes virulent strains of Y. enterocolitica, nor did they carry the Yersinia virulence plasmid. Although all strains hybridized with a DNA probe derived from the inv (invasin) gene of Y. enterocolitica, none was able to invade HEp-2 epithelial cell culture. Y. kristensenii strains were virulent only when inoculated parenterally into iron-loaded mice. Animals infected in this way succumbed rapidly to infection, generally within 24 h. This finding suggested that the pathogenicity of these bacteria may be attributable to a secreted toxin, but a search for such a substance and for other in vitro correlates of pathogenicity was unsuccessful. These observations indicate that some strains of Y. kristensenii kill mice by a mechanism not previously recognized in yersiniae.

serogroups

respectively, have been described previously (25, 26). Y. enterocolitica A2635c is a plasmidless derivative of A2635. For virulence studies, bacteria were cultivated on Trypticase soy agar (BBL, Cockeysville, Md.) at 28°C overnight. Bacteria were suspended in isotonic phosphate buffer (pH 7.4) to an optical density equal to a McFarland no. 3 barium sulfate nephelometric standard (19). This suspension (equivalent to approximately 2 x 109 CFU/ml) was used to inoculate mice by gavage. For screening of virulence by intraperitoneal (i.p.) inoculation, the suspension was diluted 1 in 50 in phosphate buffer. The number of viable bacteria in each inoculum was determined by plating 10-fold dilutions on duplicate Trypticase soy agar plates. Preparation of culture filtrates, killed bacteria, and homogenates. Bacteria were grown on Trypticase soy agar at 28 or 37°C for 48 h. Bacterial cells were removed by centrifugation. Culture supernatants were sterilized by filtration through 0.2-pum-pore-size cellulose acetate filters (Schleicher and Schuell, Dassell, Germany) and tested immediately in mice. For some experiments, bacteria were killed by being heated at 60°C for 5 min or by being incubated in 1% formaldehyde or 1% glutaraldehyde at room temperature for 2 h. Cells killed in this way were washed three times in phosphate buffer and resuspended at an optical density equal to a McFarland no. 3 standard. Homogenates of bacteria were prepared by ballistic disintegration in an MSK homogenizer (B. Braun, Melsungen, Germany) set at 4,000 rpm with 0.1-mm-diameter glass beads for 1 min under liquid CO2. This treatment reduced the number of viable bacteria by approximately 90%. The homogenates were sterilized by filtration. The sterility of each of these preparations was verified before they were inoculated into animals. Virulence testing. Unless otherwise stated, virulence studies were performed with 6- to 8-week-old female BALB/c mice in groups of five. Twenty-four hours before infection, mice were given an i.p. injection of 5 mg of iron as iron dextran (Imferon; Fisons Pty. Ltd., Sydney, Australia) and/or 5 mg of desferrioxamine B mesylate (Ciba Geigy Ltd., Sydney, Australia) (26).

The genus Yersinia includes three species, Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica, which are pathogenic for mammals. The virulence mechanisms of these bacteria have been extensively studied and exhaustively reviewed (8, 17, 21). Briefly, the pathogenicity of yersiniae is determined by both chromosomal and plasmid-borne genes. Chromosomal genes encoding virulence include inv and ail, which are believed to specify initial attachment to and invasion of host tissues (17). In addition, all virulent strains of Yersinia carry a 70-kb virulence plasmid which contributes to the survival and multiplication of bacteria in host tissues (8, 21). These plasmids share a considerable degree of DNA homology, irrespective of the Yersinia species in which they originate (8, 25). Nonpathogenic biotypes of Y. enterocolitica and other Yersinia species, such as Y. frederiksenii, Y. intermedia, and Y. kristensenii, carry chromosomal DNA homologous with the inv determinant but generally lack the ail and plasmidborne determinants of virulence (16, 25). Nonetheless, some of these species are occasionally implicated in disease (3-5, 11, 18). During a recent investigation of virulence of yersiniae for mice, we discovered a strain of Y. kristensenii which caused a fatal infection in iron-loaded mice. This observation prompted us to examine a collection of Y. kristensenii strains for virulence attributes. MATERIALS AND METHODS

Bacteria. The 47 strains of Y. kristensenii used in this investigation are listed in Table 1. The bacteria were isolated in widely dispersed areas from a variety of sources. A total of 15 strains were of human origin, 14 were from animals, 14 were from food (usually pork), and 3 were from water. The origin of one strain was unknown. Y. enterocolitica 30.42.67 and A2635, virulent strains of serogroups 0:3 and 0:8,

*

Corresponding author. 162

VOL. 59, 1991

PATHOGENICITY OF YERSINIA KRISTENSENII

163

TABLE 1. Characteristics of Yersinia strains used in this study Result Strain

CDC1 IP105 IP1420 IP841 P1223 WA122 WA32a/89 WA336 WA396 WE89/79 WS60/88 IP490 IP6048 P1494 WATe167 WATe168 WE17/90 GK11047 IP103 WATe169 WATe171 WATe172 WATe173 WATe174 WS45/89 CDC2 WA948 WA31a/89 WA758 WAT120 WE211/80 GK9036 GKG2 IP1474 WA596 IP7230 IP7229 IP7209 WA599 WA936 WA139 WA17/87 WA30a/89 WA584 WA590 WA595 WA987 30.42.67g A26359 A2635cg

Serogroup 0:11 0:11 0:11 0:11 0:11 0:11 0:11 0:11 0:11 0:11 0:11 0:12,25 0:12,25 0:12,25 0:12,25 0:12,25 0:12,25 0:12,26 0:12,26 0:12,26 0:12,26 0:12,26 0: 12,26 0:12,26 0:12,26 0:16 0:16 0:16,29 0:16,29 0:16,29 0:16,29 0:28,50 0:28,50 0:28,50 0:28,50 0:46 0:50 0:52 0:59 0:61e NTf NT NT NT NT NT NT 0:3 0:8 0:8

Source

Food Human urine Fox Human bile Human wound Human urine Rat Carrots Pork Human feces Human feces Hare Not known Pork Monkey Monkey Human feces Rodent Sheep Monkey Pork Pork Pork Pork Human feces Food Human feces Rat Human feces Pork Human feces Rodent Water Water Pork Mouse Rodent Human feces Pork Water Pig Human feces Rodent Human feces Human feces Pork Meat Human feces Milk Laboratory

Country of

origin

Hybridization with

ofv~:

Lethality'~

United States Denmark Switzerland United States Denmark Czechoslovakia Japan France United States Belgium Belgium France Not known Czechoslovakia

Japan Japan Belgium Norway Faroe Islands Japan Japan Japan Japan Japan

CAD

CV

Invasion'

CR

+ + + + + + + + + + + + + + + + +

Belgium United States Finland Japan Finland Canada Belgium Norway Norway Norway Norway United Kingdom Czechoslovakia Belgium Norway Germany Netherlands Netherlands Japan Finland Finland Norway Austria Sweden United States

+ + + +

9d

-

_

+ +

+ + +

+

+

+

HYD

+ + + + + + + -

0.003 0.005 0.004 0.002 0.004 0.001 0.004 0.003 0.006 0.002 0.009 0.007 0.005 0.007 0.003 0.003 0.003 0.010 0.009 0.002 0.007 0.004 0.003 0.006 0.001 0.009 0.004 0.005 0.003 0.005 0.001 0.002 0.005 0.002 0.004 0.003 0.003 0.002 0.005 0.002 0.007 0.003 0.004 0.004 0.005 0.008 0.004 6.5 ND 14.0

probe from: Chromosome Plasmid (iC

+

+ + +

+ +

For mice treated with iron and desferrioxamine and inoculated i.p. +, Lethal; -, not lethal. CAD, Calcium dependence; CR, binding of Congo red; CV, binding of crystal violet; HYD, production of hydroxamate. Results are the percentage of a bacterial inoculum recovered from HEp-2 cells after killing of the extracellular bacteria with gentamicin (16). ND, Not determined. d Results uninterpretable because bacteria produced large and small colonies on calcium-deficient media at both 28 and 37°C. Provisional serotype designation. f NT, Nontypeable. g Y. enterocolitica strains (all others are Y. kristensenii). a

b

c

For most experiments, mice were inoculated i.p. with 0.5 ml of a bacterial suspension and observed for 14 days. In some experiments, bacteria were inoculated perorally (by gavage; 0.5 ml), subcutaneously (0.5 ml), or intravenously (0.2 ml). For determination of the 50% lethal dose (LD50),

mice were given 10-fold serial dilutions of bacteria. The LD50 was calculated according to the method of Reed and Muench (23). The keratoconjunctivitis (Sereny) test was performed with guinea pigs pretreated with 50 mg of iron and 50 mg of desferrioxamine as described previously (26).

164

INFECT. IMMUN.

ROBINS-BROWNE ET AL.

Assay for enterotoxin. Bacteria were grown in Trypticase broth supplemented with 0.6% (wt/vol) yeast extract (BBL) at 26°C for 48 h with constant shaking. Bacteria were removed by centrifugation, and 0.1 ml of the culture supernatant was inoculated by gavage into each of three 3- to 4-day-old infant mice. The presence of enterotoxic activity was determined as described previously (24). Mouse protection studies. Mice in groups of six or more were inoculated i.p. with various doses of live or killed bacteria. Three weeks or more later, one mouse from each group was sacrificed by cervical dislocation. Cardiac blood, the entire spleen, and approximately half of the liver were collected aseptically. The organs were homogenized in 5 ml of Trypticase soy broth, which was incubated at 28°C for 7 days and then subcultured onto Trypticase soy agar plates. These studies were performed to ensure that no viable bacteria remained in the mice at the time of challenge. The remaining animals in each group were inoculated i.p. with 5 mg of iron dextran and desferrioxamine B. The next day each mouse was inoculated with approximately 100 LD50s of a virulent Yersinia strain. The animals were observed for up to 14 days. In vitro studies. 32P-labeled single-stranded DNA probes were prepared from the cloned inv and ail chromosomal genes of Y. enterocolitica 8081 (serogroup 0:8) and from the virulence plasmid of Y. enterocolitica A2635 as described previously (25). These probes were used to examine colonies of Y. kristensenii for homologous DNA sequences at high stringency (25). The ability of bacteria to invade HEp-2 cells was determined by the method of Miller et al. (16). Tests for calcium dependence and binding of Congo red and crystal violet were performed as described previously (22, 25). The ability of bacteria to produce hydroxamate siderophores was determined by the hydroxylamine assay as described previously (29). soy

RESULTS Virulence of Y. kristensenii strains for mice. In a preliminary study, we showed that the type strain of Y. kristensenii, IP105 (serogroup 0:11), was lethal when injected i.p. into mice pretreated with iron and desferrioxamine B. We subsequently investigated the pathogenicity of a wider range of strains in this animal model. The results showed that all 11 serogroup 0:11 isolates in our collection were lethal for mice (Table 1). Other serogroups with a large proportion of virulent strains were 0:12,25 (five of six strains), 0:16 (two of two strains), 0:16,29 (two of four strains), 0:59 (one strain), and 0:61 (one strain) (Table 1). By contrast, 0 serogroups 0:12,26 (one of eight strains) and 0:28,50 (none of four strains) were typically avirulent. Of seven strains whose serogroup could not be determined, only one was virulent. The LD50s of i.p. inoculated Y. kristensenii for mice given iron and desferrioxamine B were determined for three strains of different serotypes. The values obtained ranged from 2 x 106 CFU for strain IP105 (serogroup 0:11) to 8 x 106 for WE17/90 (0:12,25) and 6 x 107 for WA948 (0:16). There was no significant correlation between virulence for mice and the origin of the bacteria. Of 15 strains isolated from humans, 9 (60%) were virulent for mice, as were 6 (43%) of 14 strains from animals and 8 (47%) of 17 strains from food or water. Some inbred strains of mice are inherently susceptible to infection with yersiniae (12). In case the virulence of Y. kristensenii was confined to BALB/c mice, we examined the susceptibilities of two other mouse strains, CBA and C57BL/

TABLE 2. Effect of pretreatment with iron dextran and desferrioxamine B on the susceptibility of mice to infection with Yersinia species LD50 (CFU) of: Pretreatment"

None Iron Desferrioxamine B Iron and desferrioxamine B

kristensenii IP105

Y. enterocolitica 30.42.67

>5x 108 1x107 >5 x 108 2 x 106

>5x 108 3x107 2 x 101 2 x 109; i.p., 3 x 106; intravenous, 8 x 106; and subcutaneous, 5 x 108. Animals not pretreated with iron were resistant to the lethal effects of Y. kristensenii regardless of the route by which the bacteria were administered (data not shown). Kinetics of infection. During the course of these studies, we noted that mice which survived infection with Y. kristensenii beyond 2 days almost always recovered completely. This appeared to differ from what we had previously observed with respect to Y. enterocolitica. We therefore undertook a study in which mice pretreated with iron dextran were inoculated i.p. with approximately five LD50s of Y. kristensenii IP105 or Y. enterocolitica 30.42.67 organisms. The results demonstrated that mice given a lethal dose of Y. kristensenii succumbed more rapidly to infection than those given an equivalent number of Y. enterocolitica organisms (Fig. 1). Effects of culture filtrates, killed bacteria, and bacterial homogenates. Twenty-four hours after receiving an i.p. inoculation of a sterile culture filtrate, bacterial homogenate, or a killed preparation of Y. kristensenii, mice were hunched, lethargic, and reluctant to feed. These signs gradually improved over the next 72 h, by which time the animals appeared to have recovered completely. The effects resem-

VOL. 59, 1991

PATHOGENICITY OF YERSINIA KRISTENSENII

100

-

80

-

0 f>'

X

60-

0 E 0 .

40 -

E *

20

o

0

0

' 1

' 2

3

Y. kristensenii Y. enterocolitica

4

' 5

' 6

Days after inoculation mortality FIG. 1. Cumu mice infected with Y. kristensenii and Y. enterrliative Mice followingroupsday, they wereinjecultedi dextran. The following inoculated with 5 mg of ironlidextran. i.p. with approxi mately 5 LD50s fY. kristensenii (serogroup 0:11) or Y. enter*ocolitica 30.42.67 (serogroup 0:3) organisms. Mice were observed ftDr 14 days, but no deaths occurred after day 6. were

o

preparations of killed bacteria contained as much as the equivalent of 200 times the LD50 of living yersiniae. Assay for enterotoxin. Some strains of Y. kristensenii produce a heat-stable enterotoxin which is reactive in infant mice (13). Because this substance may conceivably be a virulence determinant of Y. kristensenii, we examined the ability of 10 strains (5 mouse lethal and 5 nonlethal, from a range of serogroups) to produce this substance. The results showed that only one strain, GK9036 (0:28,50), produced an enterotoxin which was reactive in infant mice. This strain is not virulent in adult iron-treated mice. By contrast, none of the five mouse-lethal strains investigated (CDC1, CDC2, IP105, WA758, and WE17/90) was positive in this assay. Mouse protection experiments. In order to determine whether mouse-lethal strains of Y. kristensenii possessed common

' 7

*P105

bled those in nnice given sublethal amounts of lipopolysaccharide, and eflFects obtained with preparations derived from mouse-lethal sitrains (IP105 and WE17/90) were similar to those obtainedI with preparations derived from nonlethal strains (IP1474 and WS45/89). The effects of the culture filtrates were uindiminished when preparations were boiled for 5 min and Mvere only partially reduced when preparations were heated at 121°C for 5 min (data not shown). None of the mice died as a result of these experiments, even though the

165

protective antigens,

we

inoculated mice with living

or killed Y. kristensenii organisms according to various schedules and examined their susceptibilities to subsequent infection with Y. kristensenii of homologous and heterologous serogroups and to Y. enterocolitica. The results showed that prior inoculation with Y. kristensenii conferred solid protection against infection with bacteria of the same or a related 0 serogroup (Table 3). Prior infection with a strain of an unrelated serogroup (irrespective of virulence) provided partial protection against lethal infection with

Y.

kristensenii. Similar degrees of protection were conferred by sterile culture filtrates of virulent or avirulent Y. kristensenii and by 25 p.g of purified lipopolysaccharide derived from Escherichia coli (Sigma Chemical Co., St. Louis, Mo.) (data not shown). Infection with Y. kristensenii even afforded mice partial protection against infection with Y. enterocolitica

(Table 3). Guinea pig keratoconjunctivitis test. Four mouse-virulent strains of Y. kristensenii of different serogroups were examined in this assay. None gave a positive reaction. In vitro studies. The results of these investigations are summarized in Table 1. Almost all of the Y. kristensenii isolates were negative in the assays for calcium dependence and binding of Congo red and crystal violet. The few strains which did give a positive reaction in one of these assays included both mouse-virulent and avirulent varieties.

TABLE 3. Effect of prior exposure to Y. kristensenii on the susceptibility of mice to subsequent infection with Yersinia species Strain (serogroup) No. of Intervalc used for initial doses (wk) CFU/doseb adss(k exposure

(0:11) (0:11) (0:11) (0:11) (0:11) (0:11) (0:11) IP105 (0:11) IP105 (0:11) WATe167 (0:12,25) WS45/89 (0:12,26) WS45/89 (0:12,26) IP105 IP105 IP105 IP105 IP105 IP105 IP105

5 x 107 (1) 5 x 105 (1) 5 x 107 (1) 5 x 107 (1) 5 x 107 (1) 5 x 107 (1) 5 x 107 (1 ) 2 x 108 (g) 2 x 108 (h) 3 x 107 (1) 4 x 107 (1) 4 x 107 (1 )

1 1 1 1 1 39 3 1 1 1 1 3

3 12 3 3 3 3 3 3 3 3 3 3

Strain (serogroup) used for subsequent

infection'

Potective efficacy (%)

IP105 (0:11) IP105 (0:11) IP841 (0:11) WE17/90 (0:12,25) 30.42.67 (0:3}f 30.42.67 (0:3)f WE17/90 (0:12,25) IP105 (0:11) IP105 (0:11) WE17/90 (0:12,25) WE17/90 (0:12,25) IP105 (0:11)

100 100 100 40 0 60 60 100 100 100 100 40

Given i.p. to mice not pretreated with iron or desferrioxamine B. b 1, Living bacteria; g, glutaraldehyde-killed; h, heat-killed. c Between the final dose of the initial exposure and subsequent challenge. d Approximately 100 LD50s were given i.p. to mice pretreated with iron and desferrioxamine B. ' Determined from the following equation: [(mortality rate in control mice - mortality rate in test mice)/mortality rate in control mice] x 100. f Virulent strain of Y. enterocolitica. g Given at weekly intervals.

166

ROBINS-BROWNE ET AL.

We have previously shown that Y. kristensenii may produce the hydroxamate siderophore aerobactin (29). In case the production of such substances was associated with virulence, we examined all the bacteria for their ability to synthesize hydroxamates under iron-limiting conditions. The results showed that production of these compounds was limited to bacteria in serogroups 0:16, 0:16,29, 0:46, and 0:52 (Table 1). Thus, the production of hydroxamates was not associated with virulence. The DNA hybridization studies showed that the Inv probe of Y. enterocolitica recognized all the bacteria in this study (data not shown). This result was not unexpected, because this probe detects all members of the genus Yersinia, regardless of virulence (20, 25). The HEp-2 invasion assay, however, showed that none of the Y. kristensenii strains produced active invasin, as they all invaded HEp-2 cells to a negligible extent compared with strains of Y. enterocolitica which are known to express this gene (Table 1). Only one strain (IP7229) hybridized with the Ail probe of Y. enterocolitica, but this strain did not invade HEp-2 cells and was avirulent for mice. None of the Y. kristensenii strains examined hybridized with any of three DNA probes (BamHI fragments 2, 6, and 7) prepared from widely dispersed regions of the virulence plasmid of Y. enterocolitica A2635

(25). DISCUSSION Before 1980, Y. kristensenii was considered to be part of a broad Y. enterocolitica complex. It is distinguishable from Y. enterocolitica sensu stricto and other yersiniae, however, on the bases of DNA relatedness and biochemical reactivity, being positive for trehalose and ornithine decarboxylase and negative for sucrose and rhamnose (1, 2). Little is known of the epidemiology of infections with this organism. In surveys of fecal carriage of yersiniae in humans, Y. kristensenii is found infrequently compared with other yersiniae (6, 10, 14, 28, 32). Nonetheless, when it has been recovered from human material, it generally has been regarded to be of clinical significance (3-5, 11, 18). As far as can be ascertained, these strains usually belong to those 0 serogroups which are virulent for mice (3-5). The isolation of Y. kristensenii from the feces of a young woman with acute enteritis, in conjunction with detectable serum agglutinins, prompted Bottone to speculate that Y. kristensenii is intermediate in virulence between Y. enterocolitica on the one hand and Y. frederiksenii and Y. intermedia on the other (4). This suggestion has been borne out by the present study, in which the virulence of Y. kristensenii for animals has been shown for the first time. Although Y. kristensenii shares few of the phenotypic or genetic characteristics associated with virulence in Y. enterocolitica or other Yersinia species (Table 1), the lethal effect of Y. kristensenii for mice was specific. This was evidenced by the findings that only certain serogroups were virulent in this model and that solid serogroup-specific immunity was induced after a single exposure to a sublethal dose of a homologous strain (Table 3). Moreover, we have previously examined several hundred isolates of Yersinia species, including tissue culture-invasive, plasmid-cured strains of Y. enterocolitica and Y. pseudotuberculosis, in iron- and desferrioxamine-treated mice and, until now, have found none that gave a false-positive result (22, 25). The mechanisms underlying the pathogenicity of Y. kristensenii for mice are not clear. The short time course of the illness compared with that caused by Y. enterocolitica

INFECT. IMMUN.

suggested that a bacterial toxin may be involved. In order to examine this possibility, we inoculated bacterial lysates, culture filtrates, and whole killed organisms into mice. None of these preparations was lethal, suggesting that bacterial multiplication in vivo was required for Y. kristensenii to express virulence. Additional support for this suggestion came from animal protection experiments in which prior exposure to Y. kristensenii WS45/89, a nonlethal serogroup 0:12,26 strain which was presumed to lack putative virulence determinants, conferred solid immunity against infection with WE17/90, a virulent strain of serogroup 0:12,25 (Table 3). The immunity in this instance seemed likely to be mediated by antibodies to the shared 0:12 antigen. On the other hand, exposure to a virulent or avirulent strain of Y. kristensenii afforded only partial protection against subsequent infection with a virulent strain of an unrelated serogroup (Table 3). Similar degrees of protection were conferred by prior exposure to sterile culture filtrates of Y. kristensenii, heat-treated filtrates, and purified endotoxin from E. coli (data not shown), indicating that this nonspecific immunity was likely to be mediated by lipopolysaccharide. Nonspecific endotoxin-mediated immunity of this type to a variety of infectious agents is well documented (7, 31). New evidence that Y. kristensenii may share a unique virulence determinant with Y. enterocolitica has emerged from a recent study by Delor et al., who used a DNA probe derived from the chromosomal gene for the heat-stable enterotoxin of Y. enterocolitica to examine a heterogeneous collection of Yersinia species (9). They found that the probe hybridized specifically with DNA from virulent strains of Y. enterocolitica and selected Y. kristensenii isolates but not with DNA from other strains (9). In the present study, however, there was no correlation between the ability of Y. kristensenii strains to produce enterotoxin and their virulence for iron-stressed mice. Although Y. kristensenii is frequently isolated from animals, it does not appear to be a natural pathogen of mice. Peroral administration of large numbers of bacteria to ironand desferrioxamine-treated mice caused no ill effects, presumably because the bacteria were unable to penetrate the intestinal mucosa. This suggestion was corroborated by experiments which demonstrated the inability of Y. kristensenii strains to invade HEp-2 cell culture (Table 1) or the conjunctival epithelium of guinea pigs. By contrast, Y. enterocolitica strains, which are virulent for mice by the oral route, readily invade HEp-2 monolayers. The observation that Y. kristensenii strains do not invade cell culture despite bearing DNA homologous to the inv locus of Y. enterocolitica indicates that the inv gene in these bacteria is nonfunctional, as is the case with nonpathogenic strains of Y. enterocolitica (20). One strain in this study, IP7229, evidently also carried a nonfunctional ail gene. Y. kristensenii was pathogenic for mice only when administered parenterally, suggesting that under normal circumstances blood-sucking arthropods, such as fleas, may transmit these bacteria. Two factors argue against this possibility, however: (i) Y. kristensenii has seldom been isolated from arthropods, and (ii) Y. kristensenii caused no illness and did not persist in the tissues of mice not pretreated with iron (data not shown). The precise mechanism whereby treatment with iron increases the susceptibility of mice to infection with yersiniae is unresolved. Possible explanations include the supply of an essential, growth-limiting nutrient to the bacteria or the impairment of immune responsiveness of the host (27). A finding analogous to that reported here is the observation

VOL. 59, 1991

PATHOGENICITY OF YERSINIA KRISTENSENII

that certain mutants of Y. pestis are lethal for iron-stressed mice only when given by i.p. injection (15). These findings suggest that iron overload may interfere with a critical host defense mechanism, such as macrophage effectiveness, in the peritoneal cavity. In terms of iron responsiveness, Y. kristensenii behaved similarly to Y. enterocolitica (Table 2). An important difference between these two species, however, concerned the effect of desferrioxamine B, which markedly enhanced the susceptibility of mice to infection with Y. enterocolitica but had only a moderate effect on the pathogenicity of Y. kristensenii, and then only when administered together with iron (Table 2). Of interest in this regard is the finding that although some strains of Y. kristensenii produced an ironregulated hydroxamate siderophore, this property did not correlate with virulence (Table 1). In summary, this study has shown that Y. kristensenii is pathogenic for iron-treated mice. The mechanisms of virulence of this species are obscure but are clearly different from those which operate in other species of Yersinia. Further studies are required to characterize the virulence determinants of Y. kristensenii.

coproantibody secretory immunoglobulin A response to Yersinia species. J. Clin. Microbiol. 26:287-292. Hancock, G. E., R. W. Schaedler, and T. T. MacDonald. 1986. Yersinia enterocolitica infection in resistant and susceptible strains of mice. Infect. Immun. 53:26-31. Kapperud, G. 1982. Enterotoxin production at 40, 220, and 370 among Yersinia enterocolitica and Y. enterocolitica-like bacteria. Acta Pathol. Microbiol. Immunol. Scand. Sect. B. 90:185189. Lewis, A. M., and B. Chattopadhyay. 1986. Faecal carriage of Yersinia species. J. Hyg. 97:281-287. Mehigh, R. J., A. K. Sample, and R. R. Brubaker. 1989. Expression of the low calcium response in Yersinia pestis. Microb. Pathog. 6:203-217. Miller, V. L., J. J. Farmer III, W. E. Hill, and S. Falkow. 1989. The ail locus is found uniquely in Yersinia enterocolitica serotypes commonly associated with disease. Infect. Immun. 57:

ACKNOWLEDGMENTS We are grateful to Vicki Bennett-Wood for assistance with cell culture; to Stanley Falkow and Virginia Miller, Stanford University, for providing the Inv and Ail probes; and to Simon Stuart, LaTrobe University, for helpful discussion. This study was supported by a research grant from the Australian National Health and Medical Research Council. REFERENCES 1. Bercovier, H., J. Brault, N. Barre, M. Treignier, J. M. Alonso, and H. H. Mollaret. 1978. Biochemical, serological and phage typing characteristics of 459 Yersinia strains isolated from a terrestrial ecosystem. Curr. Microbiol. 1:353-357. 2. Bercovier, H., J. Ursing, D. J. Brenner, A. G. Steigerwalt, G. R. Fanning, G. P. Carter, and H. H. MoHaret. 1980. Yersinia kristensenii: a new species of Enterobacteriaceae composed of sucrose-negative strains (formerly called atypical Yersinia enterocolitica or Yersinia enterocolitica-like). Curr. Microbiol. 4:219-224. 3. Bissett, M. L. 1976. Yersinia enterocolitica isolates from humans in California, 1968-1975. J. Clin. Microbiol. 4:137-144. 4. Bottone, E. J. 1978. Atypical Yersinia enterocolitica: clinical and epidemiological parameters. J. Clin. Microbiol. 7:562-567. 5. Bottone, E. J., and T. Robin. 1977. Yersinia enterocolitica: recovery and characterization of two unusual isolates from a case of acute enteritis. J. Chin. Microbiol. 5:341-345. 6. Chiesa, C., L. Pacifico, V. Cianfrano, and M. Miduila. 1987. Italian experience with yersiniosis. Contrib. Microbiol. Immunol. 9:76-88. 7. Cluff, L. E. 1971. Effects of lipopolysaccharides (endotoxins) on susceptibility to infections, p. 399-413. In S. Kadis, G. Weinbaum, and S. J. Ajl (ed.), Microbial toxins, vol. 5. Academic Press, Inc., New York. 8. Cornelis, G., Y. Laroche, G. Balligand, M.-P. Sory, and G. Wauters. 1987. Yersinia enterocolitica, a primary model for bacterial invasiveness. Rev. Infect. Dis. 9:64-87. 9. Delor, I., A. Kaeckenbeeck, G. Wauters, and G. R. Cornelis. 1990. Nucleotide sequence of yst, the Yersinia enterocolitica gene encoding the heat-stable enterotoxin, and prevalence of the gene among pathogenic and nonpathogenic yersiniae. Infect. Immun. 58:2983-2988. 10. Falcfio, D. P. 1987. Yersiniosis in Brazil. Contrib. Microbiol. Immunol. 9:68-75. 11. Fletcher, K. M., C. M. Morris, and M. A. Noble. 1988. Human

12. 13.

14.

15.

16.

167

121-131. 17. Miller, V. L., B. B. Finlay, and S. Falkow. 1988. Factors essential for the penetration of mammalian cells by Yersinia. Curr. Top. Microbiol. Immunol. 138:15-39. 18. Noble, M. A., R. L. Barteluk, H. J. Freeman, R. Subramaniam, and J. B. Hudson. 1987. Clinical significance of virulence-related assay of Yersinia species. J. Clin. Microbiol. 25:802-807. 19. Paik, G. 1980. Reagents, stains, and miscellaneous test procedures, p. 1000-1024. In E. H. Lennette, A. Balows, W. J. Hausler, Jr., and J. P. Truant (ed.), Manual of clinical microbiology, 3rd ed. American Society for Microbiology, Washington,

D.C. 20. Pierson, D. E., and S. Falkow. 1990. Nonpathogenic isolates of Yersinia enterocolitica do not contain functional inv-homologous sequences. Infect. Immnun. 58:1059-1064. 21. Portnoy, D. A., and R. J. Martinez. 1985. Role of a plasmid in the pathogenicity of Yersinia species. Curr. Top. Microbiol. Immunol. 118:29-51. 22. Prpic, J. K., R. M. Robins-Browne, and R. B. Davey. 1985. In vitro assessment of virulence in Yersinia enterocolitica and related species. J. Clin. Microbiol. 22:105-110. 23. Reed, L. J., and H. Muench. 1935. A simple method of estimating fifty per cent endpoints. Am. J. Hyg. 27:493-497. 24. Robins-Browne, R. M., and M. M. Levine. 1981. Effect of chlorpromazine on intestinal secretion mediated by Escherichia coli heat-stable enterotoxin and $-Br-cyclic GMP in infant mice. Gastroenterology 80:321-326. 25. Robins-Browne, R. M., M. D. Miliotis, S. Cianciosi, V. L. Miller, S. Falkow, and J. G. Morris, Jr. 1989. Evaluation of DNA colony hybridization and other techniques for detection of virulence in Yersinia species. J. Clin. Microbiol. 27:644-650. 26. Robins-Browne, R. M., and J. K. Prpic. 1985. Effects of iron and desferrioxamine on infections with Yersinia enterocolitica. Infect. Immun. 47:774-779. 27. Robins-Browne, R. M., J. K. Prpic, and S. J. Stuart. 1987. Yersiniae and iron. A study in host-parasite relationships. Contrib. Microbiol. Immunol. 9:254-258. 28. Shayegani, M., and L. M. Parsons. 1987. Epidemiology and pathogenicity of Yersinia enterocolitica in New York State. Contrib. Microbiol. Immunol. 9:41-47. 29. Stuart, S. J., J. K. Prpic, and R. M. Robins-Browne. 1986. Production of aerobactin by some species of the genus Yersinia. J. Bacteriol. 166:1131-1133. 30. Une, T., and R. R. Brubaker. 1984. In vivo comparison of avirulent Vwa- and Pgm- or Pstr phenotypes of yersiniae. Infect. Immun. 43:895-900. 31. Urbaschek, R., and B. Urbaschek. 1987. Tumor necrosis factor and interleukin 1 as mediators of endotoxin induced beneficial effects. Rev. Infect. Dis. 9:S607-S615. 32. Van Noyen, R., R. Selderslaghs, G. Wauters, and J. Vandepitte. 1987. Comparative epidemiology of Yersinia enterocolitica and related species in patients and healthy controls. Contrib. Microbiol. Immunol. 9:61-67.