The putative arthritogenic cationic 19-kilodalton antigen of Yersinia enterocolitica is a urease beta-subunit

Molecular and biochemical characterization of urease and survival of Yersinia enterocolitica biovar 1A in acidic pH in vitro

Background

Yersinia enterocolitica, an important food- and water-borne enteric pathogen is represented by six biovars viz. 1A, 1B, 2, 3, 4 and 5. Despite the lack of recognized virulence determinants, some biovar 1A strains have been reported to produce disease symptoms resembling that produced by known pathogenic biovars (1B, 2-5). It is therefore imperative to identify determinants that might contribute to the pathogenicity of Y. enterocolitica biovar 1A strains. Y. enterocolitica invariably produces urease and the role of this enzyme in the virulence of biovar 1B and biovar 4 strains has been reported recently. The objective of this work was to study genetic organization of the urease (ure) gene complex of Y. enterocolitica biovar 1A, biochemical characterization of the urease, and the survival of these strains under acidic conditions in vitro.

Results

The ure gene complex (ureABCEFGD) of Y. enterocolitica biovar 1A included three structural and four accessory genes, which were contiguous and was flanked by a urea transport (yut) gene on the 3' side. Differences were identified in ure gene complex of biovar 1A strain compared to biovar 1B and 4 strains. This included a smaller ureB gene and larger intergenic regions between the structural genes. The crude urease preparation exhibited optimal pH and temperature of 5.5 and 65°C respectively, and Michaelis-Menten kinetics with a K_m of 1.7 ± 0.4 mM urea and V_max of 7.29 ± 0.42 μmol of ammonia released/min/mg protein. The urease activity was dependent on growth temperature and growth phase of Y. enterocolitica biovar 1A, and the presence of nickel in the medium. The molecular mass of the enzyme was > 545 kDa and an isoelectric point of 5.2. The number of viable Y. enterocolitica biovar 1A decreased significantly when incubated at pH 2.5 for 2 h. However, no such decrease was observed at this pH in the presence of urea.

Conclusions

The ure gene cluster of biovar 1A strains though similar to biovar 1B and 4 strains, exhibited important differences. The study also showed the ability of biovar 1A strains of Y. enterocolitica to survive at highly acidic pH in vitro in the presence of urea.

Characterization of the urease operon of Brucella abortus and assessment of its role in virulence of the bacterium

Most members of the genus Brucella show strong urease activity. However, the role of this enzyme in the pathogenesis of Brucella infections is poorly understood. We isolated several Tn5 insertion mutants deficient in urease activity from Brucella abortus strain 2308. The mutations of most of these mutants mapped to a 5.7-kbp DNA region essential for urease activity. Sequencing of this region, designated ure1, revealed the presence of seven open reading frames corresponding to the urease structural proteins (UreA, UreB, and UreC) and the accessory proteins (UreD, UreE, UreF, and UreG). In addition to the urease genes, another gene (cobT) was identified, and inactivation of this gene affected urease activity in Brucella. Subsequent analysis of the previously described sequences of the genomes of Brucella spp. revealed the presence of a second urease cluster, ure2, in all them. The ure2 locus was apparently inactive in B. abortus 2308. Urease-deficient mutants were used to evaluate the role of urease in Brucella pathogenesis. The urease-producing strains were found to be resistant in vitro to strong acid conditions in the presence of urea, while urease-negative mutants were susceptible to acid treatment. Similarly, the urease-negative mutants were killed more efficiently than the urease-producing strains during transit through the stomach. These results suggested that urease protects brucellae during their passage through the stomach when the bacteria are acquired by the oral route, which is the major route of infection in human brucellosis.

Detection of a novel repeated sequence useful for epidemiological typing of pathogenic Yersinia enterocolitica

Strains (n = 203) of Yersinia species were used in genotyping and PCR experiments in order to evaluate the genotyping potential of the YeO:3RS probe. This probe comprises a 12.5 kb genomic fragment of the Y. enterocolitica O:3 lipopolysaccharide O-antigen gene cluster cloned into plasmid pBR322. The genotyping potential of YeO:3RS was shown to reside in the region upstream of the O-antigen gene cluster, i.e., in the first 1.65 kb of the cloned genomic fragment that contains a repeated sequence (RS) present in multiple copies in the genome. In genotyping, the YeO:3RS probe was hybridised to DNA of Yersinia enterocolitica isolates (n = 112) from humans, animals and food, along with strains of other Yersinia species (n = 5) and Salmonella enterica strains (n = 3). The YeO:3RS probe efficiently detected and subtyped all European pathogenic Yersinia enterocolitica isolates of the serobiotypes O:3/4, O:9/2 and O:5,27/2 studied (n = 87), whereas it hybridised only weakly or not at all with the other strains. Within Yersinia enterocolitica serobiotype O:3/4 strains, YeO:3RS genotyping was as discriminatory as genotyping by pulsed-field gel electrophoresis (PFGE) of XbaI-NotI digested genomic DNA. When these two methods were combined, YeO:3RS genotyping divided both of the two predominant PFGE types into six subtypes, thus increasing the discrimination. In PCR screening of additional 86 Yersinia strains, the 1.65 kb region was detected in European pathogenic serotypes O:1 and O:2 in addition to serotypes O:3, O:5,27 and O:9, indicating that it can be exploited in detecting and typing of European pathogenic serotypes in general.

Genetic characterization of DNA region containing the trh and ure genes of Vibrio parahaemolyticus

We have demonstrated that possession of the gene for thermostable direct hemolysin-related hemolysin (trh) coincides with the presence of the urease gene among clinical Vibrio parahaemolyticus strains and that the location of the two genes are in close proximity on the chromosome. Here, we cloned and sequenced the 15,754-bp DNA region containing the trh gene and the gene cluster for urease production from the chromosome of clinical V. parahaemolyticus (TH3996). We found 16 open reading frames (ORFs) and a lower G+C content (41%) compared with the total genome of this bacterium (46 to 47%). The ure cluster consisted of eight genes, namely, ureDABCEFG and ureR. ureR was located 5.2 kb upstream of the other seven genes in the opposite direction. The genetic organization and sequences of the ure genes resembled those found in Proteus mirabilis. Between ureR and the other ure genes, there were five ORFs, which are homologous with the nickel transport operon (nik) of Escherichia coli. We disrupted each of the ureR, ureC, and nikD genes in TH3996 by homologous recombination and analyzed the phenotype of the mutants. In the presence of urea these mutant strains had dramatically less urease activity than the strain they were derived from. Disruption of ureR, nikD, or ureC, however, had no effect on TRH production. The DNA region containing the trh, nik, and ure genes was found in only trh-positive strains and not in Kanagawa phenomenon-positive and environmental V. parahaemolyticus strains. At the end of the region, an insertion sequence-like element existed. These results suggest that the DNA region was introduced into V. parahaemolyticus in the past through a mechanism mediated by insertion sequences. This is the first reported case that the genes for an ATP-binding cassette-type nickel transport system, which may play a role in nickel transport through bacterial cytoplasmic membrane, are located adjacent to the ure cluster on the genome of an organism.

Reactive arthritis

During the past year, no big advances were reported in understanding the pathogenesis or treatment of reactive arthritis (ReA). The need for generally accepted diagnostic criteria has become a central issue. Regarding pathogenesis, attention has been drawn to a similarity between ReA and the experimental antigen-induced arthritis. Molecular mechanisms of the HLA-B27-associated pathogenesis have remained, in spite of intensive research, so far a mystery. It is apparent that antibiotics have no effect on the fully developed reactive arthritis, with the exception of patients with Chlamydia-triggered ReA, who might benefit from a course of antibiotics.

Construction of urease-negative mutants of Yersinia enterocolitica serotypes O:3 and o:8: role of urease in virulence and arthritogenicity

Yersinia enterocolitica serotype O:3 and O:8 urease-negative mutants unable to express the 19-kDa beta subunit of urease were constructed and tested for virulence and arthritogenicity. Our results indicate that urease is needed for full virulence in oral infections and that it is not an arthritogenic factor in the rat model.

Yersinia enterocolitica: overview and epidemiologic correlates

Yersinia enterocolitica comprises both pathogenic and nonpathogenic members. Distinguished by biogrouping, serogrouping, and ecological distribution, commonly occurring pathogenic serobiogroups, e.g., O:3/4; O:5,27/2; O:8/1b; O:9/2, possess both chromosomal and plasmid-mediated virulence traits. Studies have revealed several (oral, blood transfusion) modes of acquisition, elucidated the putative role of chromosomal and plasmid-encoded virulence factors in the pathogenesis of human infection, and have identified major animal reservoirs, e.g., the pig. Diagnosis has been refined though use of selective media, monoclonal antibodies directed against outer membrane proteins, and of purified yersiniae outer membrane proteins for antibody detection. Epidemiological investigations of foodborne outbreaks have been advanced through the use of molecular biology techniques such as ribotyping and pulsed-field gel electrophoresis.

Characterization of the synovial T cell response to various recombinant Yersinia antigens in Yersinia enterocolitica-triggered reactive arthritis. Heat-shock protein 60 drives a major immune response

OBJECTIVE

In Yersinia enterocolitica-triggered reactive arthritis (Yersinia ReA), the synovial T cell response is primarily directed against bacterial components, which are mostly unknown. This study was performed to investigate the synovial proliferative T cell response to a panel of recombinant Yersinia antigens in patients with Yersinia ReA and in controls.

METHODS

Synovial fluid mononuclear cells (SFMC) were obtained from 4 patients with Yersinia ReA and from 14 patients with arthritides of different etiology. SFMC were stimulated with 5 recombinant Yersinia antigens (the 19-kd urease beta subunit, 13-kd ribosomal L23 protein, 32-kd ribosomal L2 protein, 18-kd outer membrane protein H, and Y. enterocolitica heat-shock protein 60 [hsp60]), and with human, Chlamydia trachomatis, and Borrelia burgdorferi hsp60. Three T cell clones specific for Y. enterocolitica hsp60 were generated from 1 patient with Yersinia ReA. Antigen-induced cytokine release was measured by enzyme-linked immunosorbent assay.

RESULTS

SFMC from all 4 patients with Yersinia ReA responded to each of the Yersinia antigens except the 13-kd protein. These antigens were also recognized by SFMC from a subgroup of patients with undifferentiated arthritis (n = 4), but not by SFMC from other patients with arthritis of different etiology (n = 10). Y. enterocolitica hsp60 induced the strongest proliferative response in all cases. Two types of hsp60-reactive T cell clones could be obtained. One clone responded to all hsp60 variants, including the human variant, and showed a type 2 T helper (Th2)-like cytokine-secretion pattern. In contrast, another clone with specificity for the bacterial hsp60 proteins, but not the human equivalent, reacted with a more Th1-like pattern.

CONCLUSION

In Y. enterocolitica-triggered ReA, at least 4 immunodominant T cell antigens exist, which might be used in lymphocyte proliferation assays to identify patients with Yersinia ReA. The hsp60 is a strong antigen, inducing both bacteria-specific and potentially autoreactive CD4+ T cells of both the Th1 and Th2 type.

Urease is not involved in the virulence of Yersinia pseudotuberculosis in mice

A chromosomal locus (ure) involved in the production of urease activity in the bacterial pathogen Yersinia pseudotuberculosis was characterized. The genetic organization of the Y. pseudotuberculosis ure locus closely resembles that of the related ureolytic Yersinia species Y. enterocolitica. This locus encompasses seven open reading frames encoding polypeptides with predicted molecular weights of 10,894 (UreA), 15,820 (UreB), 61,001 (UreC), 25,801 (UreE), 24,551 (UreF), 20,330 (UreG), and 31,308 (UreD). The polypeptides have 85 to 96% identity with the corresponding Ure polypeptides of Y. enterocolitica serotype 0:8. Restriction fragment length polymorphisms of the ure loci from 12 unrelated Y. pseudotuberculosis strains produced by HaeIII and MboI indicate a low level of genetic variability of this locus in this species. The role of urease in the pathogenicity of Y. pseudotuberculosis was studied by constructing an isogenic urease-negative mutant obtained by disruption of structural gene ureB by aphA-3', which encodes kanamycin resistance. Experimental infection of mice with this mutant demonstrates that urease is not essential for Y. pseudotuberculosis virulence. Urease might be required mostly during the saprophytic life of this pathogen.

Yersinia enterocolitica: the charisma continues

Yersinia enterocolitica, a gram-negative coccobacillus, comprises a heterogeneous group of bacterial strains recovered from animal and environmental reservoirs. The majority of human pathogenic strains are found among distinct serogroups (e.g. O:3, O:5,27, O:8, O:9) and contain both chromosome- and plasmid (60 to 75 kb)-mediated virulence factors that are absent in "avirulent" strains. While Y. enterocolitica is primarily a gastrointestinal tract pathogen, it may produce extraintestinal infections in hosts with underlying predisposing factors. Postinfection sequelae include arthritis and erythema nodosum, which are seen mainly in Europe among patients with serogroups O:3 and O:9 infection and HLA-B27 antigen. Y. enterocolitica is acquired through the oral route and is epidemiologically linked to porcine sources. Bacteremia is prominent in the setting of immunosuppression or in patients with iron overload or those being treated with desferrioxamine. metastatic foci following bacteremia are common and often involve the liver and spleen. Of particular concern is blood transfusion-related bacteremia. Evidence has accumulated substantiating the role of Y. enterocolitica as a food-borne pathogen that has caused six major outbreaks in the United States. The diagnosis of Y. enterocolitica gastroenteritis is best achieved through isolation of the bacterium on routine or selective bacteriologic media. When necessary, serogrouping, biogrouping, and assessment for plasmid-encoded virulence traits may aid in distinguishing virulent from "avirulent" strains. Epidemiologically, outside of identified food-borne outbreaks, the source (reservoir) of Y. enterocolitica in sporadic cases is speculative. Therefore, prevention and control measures are difficult to institute.

Analysis of the urease gene complex of members of the genus Yersinia

The urease gene complex of Yersinia enterocolitica is relatively conserved within the species, although this conservation may not extend to other members of the genus. Spontaneous urease-negative isolates of Y. enterocolitica appear to have arisen as a result of large deletions within this complex, while Y. pestis shows no significant deletions within the complex, despite being urease negative.

Aetiological agents and immune mechanisms in enterogenic reactive arthritis

Reactive arthritis is triggered by an infection, either of the genitourinary or gastrointestinal tracts; the common triggering bacteria in enteric ReA include salmonella, shigella, yersinia, and campylobacter. It is still not clear how such different bacteria can lead to a similar clinical picture and have a similar association with the MHC class I antigen HLA-B27. Common both to enterogenic and urogenic bacteria is the type of peripheral joint involvement. However, this is not so different from other bacteria-associated arthritides and is probably the consequence of bacteria persistent inside the joint. What is unique to these bacteria is the HLA-B27-association and the nearly exclusively B27-linked clinical manifestations as sacroiliitis and iritis. Shigella-induced ReA has the highest B27-association while in salmonella- and chlamydia-induced ReA a lower association can be found. Mucosal entry of enterogenic bacteria give easy access to macrophages which might be important for the transport into the joint. Although bacteria-specific antibodies are of diagnostic value, the humoral immune response does not explain the immunopathogenesis and MHC-association of this disease. Bacteria-specific T-cells have been constantly found in the synovial fluid from ReA patients and have been further analysed. The identification of immunodominant antigens of these bacteria is of great importance to understand the pathogenesis. Although an antigen shared by all bacteria has not been identified until now progress is being made in this field. We have also to consider the possibility that these bacteria are not only driving the immune response themselves but rather work as a trigger for autoimmunity.

Purification and characterization of urease from schizosaccharomyces pombe

The urease from the ascomycetous fission yeast Schizosaccharomyces pombe was purified about 4000-fold (34% yield) to homogeneity by acetone precipitation, ammonium sulfate precipitation, DEAE-Sepharose ion-exchange column chromatography, and if required, Mono-Q ion-exchange fast protein liquid chromatography. The enzyme was intracellular and only one species of urease was detected by nondenaturing polyacrylamide gel electrophoresis (PAGE). The native enzyme had a M(r) of 212 kDa (Sepharose CL6B-200 gel filtration) and a single subunit was detected with a M(r) of 102 kDa (PAGE with sodium dodecyl sulfate). The subunit stoichiometry was not specifically determined, but the molecular mass estimations indicate that the undissociated enzyme may be a dimer of identical subunits. The specific activity was 700-800 micromols urea.min-1.mg protein-1, the optimum pH for activity was 8.0, and the Km for urea was 1.03 mM. The sequence of the amino terminus was Met-Gln-Pro-Arg-Glu-Leu-His-Lys-Leu-Thr-Leu-His-Gln-Leu-Gly-Ser-Leu-Ala and the sequence of two tryptic peptides of the enzyme were Phe-Ile-Glu-Thr-Asn-Glu-Lys and Leu-Tyr-Ala-Pro-Glu-Asn-Ser-Pro-Gly-Phe-Val-Glu-Val-Leu-Glu-Gly-Glu-Ile- Glu- Leu-Leu-Pro-Asn-Leu-Pro. The N-terminal sequence and physical and kinetic properties indicated that S. pombe urease was more like the plant enzymes than the bacterial ureases.

HLA-B27 and bacteria

Most HLA-B27 positive individuals live without noticing the risk factor they carry. Only when certain bacteria come into play, usually in the form of intestinal or genital infection, some of the HLA-B27 positive individuals will develop a reactive disease. Such an association is so well established that it has formed a basis to study the relation of autoimmune reactions to microbial infections. Early epidemiological investigations have been followed by studies on immunological cross-reactivity and recently by studies on presentation of bacterial peptides by the HLA-B27 molecule. This introductory review briefly summarizes the evolution of the research, the goal of which still remains to clarify the etiopathogenetic mechanisms of HLA-B27 associated diseases.

Molecular biology of microbial ureases

Urease (urea amidohydrolase; EC 3.5.1.5) catalyzes the hydrolysis of urea to yield ammonia and carbamate. The latter compound spontaneously decomposes to yield another molecule of ammonia and carbonic acid. The urease phenotype is widely distributed across the bacterial kingdom, and the gene clusters encoding this enzyme have been cloned from numerous bacterial species. The complete nucleotide sequence, ranging from 5.15 to 6.45 kb, has been determined for five species including Bacillus sp. strain TB-90, Klebsiella aerogenes, Proteus mirabilis, Helicobacter pylori, and Yersinia enterocolitica. Sequences for selected genes have been determined for at least 10 other bacterial species and the jack bean enzyme. Urease synthesis can be nitrogen regulated, urea inducible, or constitutive. The crystal structure of the K. aerogenes enzyme has been determined. When combined with chemical modification studies, biophysical and spectroscopic analyses, site-directed mutagenesis results, and kinetic inhibition experiments, the structure provides important insight into the mechanism of catalysis. Synthesis of active enzyme requires incorporation of both carbon dioxide and nickel ions into the protein. Accessory genes have been shown to be required for activation of urease apoprotein, and roles for the accessory proteins in metallocenter assembly have been proposed. Urease is central to the virulence of P. mirabilis and H. pylori. Urea hydrolysis by P. mirabilis in the urinary tract leads directly to urolithiasis (stone formation) and contributes to the development of acute pyelonephritis. The urease of H. pylori is necessary for colonization of the gastric mucosa in experimental animal models of gastritis and serves as the major antigen and diagnostic marker for gastritis and peptic ulcer disease in humans. In addition, the urease of Y. enterocolitica has been implicated as an arthritogenic factor in the development of infection-induced reactive arthritis. The significant progress in our understanding of the molecular biology of microbial ureases is reviewed.

Role of YadA in arthritogenicity of Yersinia enterocolitica serotype O:8: experimental studies with rats

Outer membrane protein YadA, the Yersinia adhesin, is one of the plasmid-encoded virulence factors of yersiniae. To evaluate the role of YadA in the pathogenesis of reactive arthritis experimentally, we used YadA- strain YeO8-116, a kanamycin GenBlock insertion mutant derived from Yersinia enterocolitica O:8 wild-type strain 8081. As control strains, a plasmid-cured derivative (8081-c) of 8081 and a YopH- mutant (8081-yoph) were used. In addition, YeO8-116, with the yadA mutation transcomplemented with plasmid pMW10, was used. YeO8-116 induced arthritis to a considerably lesser extent than did wild-type strain 8081 when inoculated intravenously into Lewis rats. In rats surviving for over 14 days after the bacterial inoculation, the arthritis incidences were 6% (4 of 72) among those inoculated with the yadA mutant and 51% (33 of 65) among those inoculated with wild-type strain 8081. When the yadA gene was transcomplemented back to YeO8-116, YeO8-116/pMW10 induced arthritis in 47% (9 of 19) of the inoculated rats. Plasmid-cured strain 8081-c did not induce arthritis in any of the 24 inoculated rats, whereas YopH- mutant 8081-yoph induced arthritis in 20% (5 of 25) of the rats inoculated. Although the 50% lethal dose of YeO8-116 was about sixfold higher than that of 8081, the kinetics of bacterial elimination from the spleen and mesenteric lymph nodes were about the same with both strains. Antibody responses in rats infected with the two strains were also indistinguishable. Our results indicate that YadA contributes to the arthritogenicity of Y. enterocolitica in the rat model.

The evolutionarily conserved ribosomal protein L23 and the cationic urease beta-subunit of Yersinia enterocolitica O:3 belong to the immunodominant antigens in Yersinia-triggered reactive arthritis: implications for autoimmunity

BACKGROUND

Reactive arthritis (ReA) is a T cell mediated inflammatory process. The immune response is primarily directed against a triggering organism, although autoimmunity has been invoked in long-lasting, antibiotic-resistant disease. Although a variety of different species are known to trigger Reactive arthritis, the clinical manifestations are strikingly similar as well as closely associated to the HLA-B27 (70%).

MATERIALS AND METHODS

Various antigenic fractions and single antigens of Yersinia enterocolitica were prepared, and their immunological activity was assessed by proliferation of synovial fluid mononuclear cells from 10 Reactive arthritis patients. The gene encoding one hitherto unknown antigen has been sequenced. Nonapeptides deduced from sequences of the target antigens were tested in an assembly assay.

RESULTS

Two immunodominant proteins of Yersinia enterocolitica were found, one being the urease beta-subunit and the other the 50 S ribosomal protein L23. The latter has been sequenced and belongs to the evolutionarily conserved ribosomal proteins with homology to procaryotes and eucaryotes. One nonapeptide derived from the urease beta-subunit was identified as a possible epitope for HLA-B27-restricted cytotoxic T cells by its high affinity. This epitope is also highly conserved.

CONCLUSION

Sharing of conserved immunodominant proteins between different disease triggering microorganisms could provide an explanation of the shared clinical picture in Reactive arthritis. Moreover, autoimmunity in Reactive arthritis might be mediated by antigen mimicry between evolutionarily conserved epitopes of ribosomal proteins and their host analogs.

Role of bacteria-specific T cells in the immunopathogenesis of reactive arthritis

Reactive arthritis is a usually self-limited sterile inflammation of joints that follows certain bacterial gastrointestinal or urogenital infections. The immunopathogenesis involves CD4+ T cells, which mediate an antigen-specific TH1 response to bacterial constituents within the joint. Properties of the arthritogenic bacteria and the physicochemical characteristics of the bacterial antigens may contribute to the development of reactive arthritis.

Characterisation of the urease-encoding gene complex of Yersinia enterocolitica

A cosmid gene library of chromosomal DNA from Yersinia enterocolitica A2635 (serogroup O:8) was constructed in Escherichia coli. Subcloning of a urease-positive (Ure+) clone revealed a region of 6.6 kb that was sufficient for expression of Ure activity in E. coli. Sequencing of this fragment disclosed seven ORFs transcribed in the same direction. On the basis of homology to known Ure, these were designated ureA, ureB, ureC, ureE, ureF, ureG and ureD, which are predicted to encode polypeptides of 11.1, 17.9, 61.0, 29.5, 25.0, 24.1 and 36.4 kDa, respectively. The polypeptides encoded by the ure gene complex of Y. enterocolitica are significantly divergent from those encoded by the ure operons of other Enterobacteriaceae, which appear to be closely related to each other. This suggests that the ure genes were acquired by Y. enterocolitica from an unrelated organism or alternatively, that they diverged from those of other Enterobacteriaceae some considerable time ago.

Identification of the Yersinia enterocolitica urease beta subunit as a target antigen for human synovial T lymphocytes in reactive arthritis

The local T-cell response to bacterial antigens is involved in the pathogenesis of reactive arthritis (ReA). Here, we have identified a 19-kDa antigen of Yersinia enterocolitica O:9 recognized by Yersinia-specific synovial fluid CD4+ T cells in two patients with Yersinia-induced ReA. N-terminal amino acid sequencing of this protein revealed that it was identical to the 19-kDa urease beta subunit of Y. enterocolitica O:9. This protein has previously been shown to be arthritogenic in preimmunized rats after intra-articular injection. Analysis of the T-cell response to this protein showed that it contains several T-cell epitopes, one of which cross-reacts with other enterobacteria not able to induce ReA. This indicates that the arthritogenicity of the 19-kDa antigen is not a property of the 19-kDa protein alone but is dependent on its expression in bacteria able to induce ReA.