Phenotypic and Genotypic Characterization of Virulent Yersinia enterocolitica Strains Unable To Ferment Sucrose

Abstract

Several atypical sucrose-negative Yersinia strains, isolated from clinical samples and sometimes associated with symptoms, proved to have full virulence potential in in vitro and in vivo testings. DNA-relatedness studies revealed that they were authentic Yersinia enterocolitica strains. Therefore, atypical sucrose-negative Yersinia isolates should be analyzed for their virulence potential.

Some bioserotypes of Yersinia enterocolitica correlate with potential pathogenicity in humans (6, 10). Most commonly isolated in gastrointestinal infections are bioserotypes 4/O:3, 2/O:9, and 2-3/O:5-5,27 (5, 9). The Y. enterocolitica-related species can share antigenic reactivities with Y. enterocolitica sensu stricto, but they are avirulent.

Among the biochemical traits that differentially characterize these Y. enterocolitica-related and -sensu stricto species, lack of sucrose fermentation has been a distinct feature of Y. kristensenii (2, 3, 17). Consequently, sucrose-negative Y. enterocolitica-like organisms isolated from clinical specimens have been regarded as avirulent, and in routine practice, they are not characterized further. The present report, however, describes a series of sucrose-negative, Voges-Proskauer (VP)-positive Y. enterocolitica-like isolates that proved, upon thorough characterization, to be fully pathogenic strains of Y. enterocolitica.

Detection of atypical Y. enterocolitica-like clinical isolates.

A Yersinia strain, IP22109, was recovered in large numbers on CIN agar base (Difco) from the feces of a 24-year-old man hospitalized at the La Croix St. Simon Hospital, Paris, France, for high fever (39°C), diarrhea, and pain in the right iliac fossa. This case was one of several cases of acute gastroenteritis traced to a common meal, suggesting a food-borne outbreak although this small outbreak was not further investigated. No other potentially pathogenic microorganism was isolated from the stool samples. IP22109 was of serotype O:5. The patient’s serum was able to agglutinate the Y. enterocolitica O:5 reference strain as well as its homologous strain, indicating that the latter had triggered an immune response. IP22109 displayed an atypical biochemical profile, since sucrose negativity, typical of Y. kristensenii (3), was associated with a positive VP test which is negative in this species but is positive in Y. enterocolitica (3). The association between sucrose negativity and VP positivity has been occasionally observed in biotype 5 of Y. enterocolitica. This very uncommon biotype, however, is associated with serotype O:2,3, as opposed to serotype O:5 of IP22109.

A search of the 22,000-strain collection of the Yersinia Reference Laboratory and WHO Collaborating Center, Pasteur Institute, for sucrose-negative, VP-positive Yersinia isolates of serotype O:5 yielded four such strains. They all had been isolated from human stools, and the presence of concomitant digestive symptoms had been recorded for two of them (Table 1). They were, therefore, further characterized along with IP22109.

TABLE 1.

Characteristics of the sucrose-negative, VP-positive Y. enterocolitica strains used in this study^a

Strain	Result by test:		Geographic origin	Sampling origin	Symptom(s)	Result by test:				LD50 in mice	Species assignment	Biotype	Serotype
	Sucrose	VP				Autoagglutination	MOX	pYVb	Pyrazinamidase
IP22228	+	+	Australia	Human	Not known	+	+	+	−	8.8 × 104	Y. enterocolitica (control strain)	2	O:5
IP7230	−	−	England	Not known	Not known	−	−	−	+	>108	Y. kristensenii (control strain)	NAc	O:46
IP22109	−	+	France	Human stools	Abdominal pain	+	+	+	−	4.4 × 104	Y. enterocolitica	2	O:5
IP10664	−	+	France	Human stools	Not known	−	−	−	−	NDd	Y. enterocolitica	2	O:5
IP17381	−	+	France	Human stools	Not known	+	+	+	−	13.5 × 104	Y. enterocolitica	2	O:5
IP19817	−	+	France	Human stools	Abdominal pain	+	+	+	−	1.9 × 104	Y. enterocolitica	2	O:5
IP22914	−	+	France	Human stools	Diarrhea	+	+	+	−	4.2 × 104	Y. enterocolitica	2	O:5
IP25686	−	+	France	Human stools	Not known	+	+	+	−	ND	Y. enterocolitica	2	O:9

^aThe enteropathogenic Y. enterocolitica strain IP22228 and the nonpathogenic Y. kristensenii strain IP7230 were used as controls. 

^bDirect detection of pYV was determined by alkaline extraction followed by restriction analysis in all cases and by colony hybridization with a yadA-specific probe for all strains except IP25686. 

^cNA, not applicable. 

^dND, not determined. 

In vitro and in vivo virulence tests.

Pathogenic Yersinia strains carry the 70-kb pYV plasmid, which encodes essential virulence determinants. Four in vitro tests were used to address the presence of pYV in the five atypical isolates. Indirect assessment of the plasmid presence was obtained by (i) plating the strains on magnesium-oxalate agar (MOX test) in search for the growth inhibition triggered by the pYV plasmid under low-calcium conditions at 37°C (11, 16) and (ii) checking bacterial autoagglutination mediated by the pYV-encoded YadA adhesin upon incubation at 37°C in RPMI medium (1, 13). Direct detection of pYV was then performed by colony hybridization with a yadA-specific probe and by plasmid extraction. The MOX and autoagglutination tests were done as described previously (11, 13). The yadA-specific probe used for colony hybridization was obtained by nested PCR of plasmid DNA from Y. enterocolitica IP864 with the pairs of primers 5′-CTGCAAATAAGCTATACCGAT-3′ and 5′-ATGCCTGACTAGAACGATAT-3′ for the first reaction, and 5′-GTGACTGTAAGTAGTTCGACT-3′ and 5′-CCGACACCTGCAGTAAAGTT-3′ for the second reaction. PCRs and digoxigenin labeling during the second PCR were performed as published previously (8), except that the first PCR was done with 10 ng of template DNA and with primers at a final concentration of 1 μM each, and in the second PCR, each unlabeled nucleotide was at a final concentration of 40 μM. The first PCR mixture was denatured at 94°C for 2 min and then amplified for 24 cycles at 94°C for 1 min, 54°C for 1 min 30 s, and 72°C for 2 min. The last cycle was followed by a 10-min incubation at 72°C. Amplification products were reamplified with the second primer set and the same PCR parameters as above, except for the priming step, which was carried out at 58°C. Colony transfer and hybridization were performed as published previously (8), and immunological detection of the probe was done according to the manufacturer’s instructions (Boehringer). Plasmid extraction was performed according to the method of Birnboim and Doly (4).

As shown in Table 1 and Fig. 1, the results of the four pYV detection assays were in agreement and showed that all of the atypical strains except IP10664 harbored the 70-kb virulence plasmid. In the IP7230 Y. kristensenii control strain, these tests consistently failed to detect the plasmid.

FIG. 1.

Plasmid fingerprinting of the atypical isolates. EcoRI-restricted plasmids were resolved on a 0.8% agarose gel, ethidium bromide stained, and visualized under UV illumination. Comparison of the plasmid restriction patterns shows that most of the atypical strains harbor the pYV plasmid. Within the O:5 strains (lanes 1 to 6), the only detectable difference in the restriction profiles is an apparent size increase in a low-molecular-weight fragment of strains IP22914 and IP22109 (arrow). Lanes: 1, IP17381; 2, IP19817; 3, IP10664; 4, IP22109; 5, IP22914; 6, IP22228 (control Y. enterocolitica O:5 strain known to harbor pYV). The atypical strain of serotype O:9 (IP25686, lane 8) had a plasmid profile identical to that of the Y. enterocolitica O:9 control strain, IP23073 (lane 7). λ and M, λ/HindIII and 1-kb-ladder molecular weight markers, respectively (GibcoBRL). Sizes are in kilobases.

The atypical isolates were also assayed for pyrazinamidase activity as a plasmid-independent virulence test (12). Absence of detectable pyrazinamidase activity, which correlates with potential pathogenicity in Y. enterocolitica and related organisms (12), was found to be a feature of all atypical isolates (Table 1). This result indicated that the atypical Yersinia strains were all potentially pathogenic, although one of them had lost its pYV plasmid.

The pathogenicity of the atypical strains was confirmed in vivo by using the iron-overloaded mouse model (15). All 50% lethal doses (LD₅₀s) were determined as described previously (7) for groups of five intravenously infected OF1 female mice (Iffa Credo, France). The plasmid-harboring atypical strains proved to be as virulent as the pathogenic Y. enterocolitica IP22228 control, with LD₅₀s ranging between 2 × 10⁴ and 2 × 10⁵ CFU (Table 1). The LD₅₀ of the avirulent Y. kristensenii reference strain, IP7230, was markedly higher, with a more than 5,000-fold increase relative to that of the pathogenic Y. enterocolitica strains.

The virulence potential of the atypical strains demonstrated by the in vitro and in vivo testings was in keeping with the symptomatology associated with these strains (Table 1). It also strongly suggested that the atypical isolates belonged to Y. enterocolitica rather than to Y. kristensenii, since the pYV virulence plasmid is exclusively found in the three pathogenic species of Yersinia, i.e., Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica. Definitive species assignment was achieved by DNA-relatedness measurements.

Species assignment.

DNA-DNA hybridizations were performed with the atypical strains IP10664 and IP19817 and the control strains Y. enterocolitica IP22228 and Y. kristensenii IP105, according to the protocol published by Popoff and Coynault (14). DNA labeling of the control strains, assessment of the degree of hybridization, and normalization to the homologous reaction were as described previously (14). As shown in Table 2, the atypical organisms shared a high degree of relatedness with the Y. enterocolitica control strain, IP22228 (>93%), and a low degree of relatedness (≤49%) with the Y. kristensenii reference strain, IP105. Therefore, the DNA hybridization studies conclusively ascribed the atypical strains to the Y. enterocolitica species.

TABLE 2.

DNA relatedness of two atypical Yersinia strains with a control Y. enterocolitica O:5 strain and the Y. kristensenii reference strain

Strain	Species	% Hybridization with probe:
		IP22228	IP105
IP22228	Y. enterocolitica	100	48.6
IP105	Y. kristensenii	40.7	100
IP10664	U.I.a	93.6	49
IP19817	U.I.	100	37.7

^aU.I., strain under investigation in the present study. 

Detection of an atypical Y. enterocolitica strain of bioserotype 2/O:9.

One additional sucrose-negative, VP-positive Yersinia strain, IP25686, was identified in the Yersinia Reference Laboratory collection upon screening of the entire database. This strain was of bioserotype 2/O:9. IP25686 had the virulence characteristics of enteropathogenic Y. enterocolitica, because it belonged to a pathogenic bioserotype, possessed the pYV plasmid (Fig. 1), and lacked detectable pyrazinamidase activity (Table 1).

All of the sucrose-negative Y. enterocolitica strains, whatever their serotype, were of biotype 2 (Table 1). Since this biotype is much less common in France than biotype 4, its constant association with the loss of sucrose fermentation in this series might reflect a genetic or a biochemical linkage between the two phenotypic traits.

The results presented above demonstrate the existence of fully pathogenic sucrose-negative Y. enterocolitica strains within two of the most common pathogenic bioserotypes. The pYV plasmid-containing atypical Y. enterocolitica strains were pathogenic for mice to the same degree as a sucrose-positive control, indicating that full virulence of Y. enterocolitica in mice is not dependent on sucrose fermentation. Furthermore, the cluster of acute clinical cases associated with one atypical strain suggests that pathogenic potential in humans is retained upon loss of sucrose fermentation. Although loss of sucrose fermentation does not alter the virulence of Y. enterocolitica, the fact that this biochemical property is almost perfectly retained in the species suggests that it is beneficial for the bacterium, possibly when it is outside of its host.

An important clinical implication of this work is that sucrose-negative yersiniae can be virulent and can therefore be held responsible for acute and delayed pathologic manifestations typical of enteropathogenic yersiniae, e.g., enteritis, erythema nodosum, and arthropathies. Any sucrose-negative Y. enterocolitica-like organism with associated symptomatology should be characterized further, especially if the VP reaction is positive. Determination of the atypical isolate biotype, serotype, and, in certain cases, phagetype, along with the pyrazinamidase test may be simple means to orient the bacteriological diagnosis. If necessary, these tests should be completed by in vitro and in vivo virulence evaluations.