INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Yersinia enterocolitica infects a wide range of animal hosts, including humans, and is transmitted through ingestion of contaminated food and water (4). In humans, the infection generally leads to an acute gastroenteritis which is self-limiting and manifests as fever, diarrhea, and abdominal pain. However, in patients who are iron overloaded or immunocompromised, a systemic infection, which frequently is fatal, may ensue. Y. enterocolitica is a lymphotropic organism. After ingestion, the bacteria are able to cross the intestinal epithelium by invasion of specialized M cells and colonize the underlying lymphoid follicles, called Peyer's patches (PP), that line the small intestine (1). From the PP, the bacteria can progress to the mesenteric lymph nodes (MLN) and eventually establish a systemic infection (5).

A large number of virulence factors reside on a well-characterized 70-kb virulence plasmid (7). Several of these genes encode a type III secretion system that exports factors, also encoded on the virulence plasmid, into host cells which disrupt cellular function. Other factors are chromosomally encoded, such as invasin, which has been shown to bind to ₁ integrins on the surface of M cells and in Y. enterocolitica is important in the initial steps of invasion (13, 21, 38). Previously, an in vivo expression technology (IVET) screen was performed in order to identify additional chromosomally encoded factors which are expressed early during infection (36). In that study, random fragments of Y. enterocolitica DNA were transcriptionally fused to a promoterless cat gene and integrated onto the chromosome of wild-type Y. enterocolitica. Fusion of cat to an active promoter confers to the bacterium resistance to chloramphenicol (CHL). Strains containing cat fusions were used to orally infect BALB/c mice. Y. enterocolitica strains in which the promoter upstream of cat was transcriptionally active during infection (i.e., in vivo) were enriched for by administration of CHL to the mice. Fusions that were enriched in the animal were then screened for the absence of expression under standard laboratory conditions. One of the genes discovered by that screen was designated hre20. Hre20 was found by BLAST analysis to be 67% identical to YeiE from Escherichia coli, which is a hypothetical protein with homology to the LysR family of transcriptional regulators. An insertion in hre20 resulted in increased splenic dissemination of bacteria during infection in a BALB/c mouse model, indicating a role for the hre20 locus during infection. Based on the observed in vivo phenotype, hre20 was designated rscR for reduced splenic colonization regulator.

Due to the apparent role in the progression of Y. enterocolitica infection, rscR was further characterized in the present study. Since RscR is proposed to be a transcriptional regulator, the genes which it regulates are likely to encode the effectors of the observed in vivo phenotype. A screen was undertaken to identify genes regulated by RscR in order to further understand the effect that these factors have on the course of infection.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth conditions. All cultures were grown in Luria-Bertani (LB) broth unless otherwise noted. E. coli was grown at 37°C and Y. enterocolitica was grown at 26°C, with aeration on a roller drum. The following antibiotics were used at the indicated concentrations: ampicillin, 100 µg/ml; CHL, 12.5 µg/ml; kanamycin (KAN), 100 µg/ml; nalidixic acid (NAL), 20 µg/ml; streptomycin, 50 µg/ml; and spectinomycin, 50 µg/ml. As a chromogenic substrate for -galactosidase, 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) was used at 40 µg/ml. For regulation of the P_BAD promoter, arabinose and glucose were used at 0.2% in LB.

Bacterial strains and plasmids. Bacterial strains and plasmids used in this study are described in Table 1. A designation of "v" indicates the presence of the 70-kb virulence plasmid. Electroporations were performed as described previously (25). Restriction enzymes and DNA ligase were purchased from New England Biolabs. PCR was carried out using recombinant PFU (Stratagene). Sequencing was performed with BigDye Terminator (Amersham) and reactions were processed by the Protein and Nucleic Acid Chemistry Lab at Washington University.

Animal experiments. Six- to seven-week-old BALB/c mice were used. Oral infections were performed using a 1-ml syringe with 2 in. of intramedic tubing encompassing a 21-guage needle. Bacteria were grown 16 to 18 h at 26°C and then resuspended in sterile phosphate-buffered saline to the desired concentration. Bacterial suspension (200 µl) was administered. On the indicated day postinfection, the tissues of interest were harvested and homogenized in sterile phosphate-buffered saline. The homogenates were diluted and plated to determine viable cell counts. Y. enterocolitica was selected for using NAL. Bacterial load was reported as CFU per gram of tissue. The significance of the results was assessed using the Mann-Whitney test.

Mutagenesis. The strain YVM618 was mutagenized by delivery of mTn5Km2-lacZY from pRev10. Conjugation was carried out by filter mating of YVM617 [rscR::(Sm/Sp)] carrying pKN14 and E. coli S17-1pir containing pRev10. An aliquot (100 µl) of a 16- to 18-h culture of each strain was added to 3 ml of sterile 10 mM MgSO₄ and was pushed through a 0.45-µm-pore-size filter. The filter was placed on minimal medium without glucose to prevent outgrowth and was incubated overnight at 26°C. Filters were vortexed in 1 ml of MgSO₄, and 100 µl of a 1:4 dilution was plated on LB containing NAL (to select against the E. coli donor strain), KAN (to select for transposition), and CHL (to select for pKN14). Plates were incubated at 26°C for 30 h, and then each was replica plated to two platesone containing NAL, KAN, CHL, X-Gal, and 0.2% arabinose and one containing NAL, KAN, CHL, X-Gal, and 0.2% glucose. Colonies were compared visually, with inspection for differences in color after growth on arabinose compared to glucose. Colonies displaying potential differential -galactosidase expression were purified, and enzyme assays were performed.

Enzyme assays. -Galactosidase assays were performed as previously described (19). Cultures (1 ml) were grown for 16 to 18 h at 26°C in LB containing appropriate antibiotics and either 0.2% arabinose or 0.2% glucose as described above for the plate assays. In initial screening, single assays were performed with each mutant, and those displaying differential expression were then repeated and assays were performed in duplicate. Fusions selected for further characterization displayed fivefold or more induction in arabinose or glucose in the repeated assays.

Screen controls. To identify fusions to genes encoding products involved in arabinose utilization, bacteria containing fusions were patched onto minimal medium containing arabinose as the only carbon source. To identify fusions responding to glucose or arabinose rather than the presence or absence of RscR, the rscR expression plasmid, pKN14, was cured by culturing without selection for 3 days. These cultures were diluted and plated on LB agar containing no antibiotics. After 2 days of growth, colonies were patched to two plates, one containing KAN and one containing CHL. Clones which retained Kan^r and lost Cm^r were selected. If both resistances were lost, this indicated a transposon insertion on the plasmid. Selected plasmid-cured fusions were analyzed as above for differential -galactosidase levels in arabinose and glucose. Fusions displaying induction after the loss of the plasmid were removed from further consideration as it was likely that the differential expression observed was due to the carbon source rather than the presence or absence of RscR.

Cloning and sequencing. Chromosome-transposon junctions were cloned as follows. Total genomic DNA preparations of the mutant strains were digested with restriction enzymes known to have a unique restriction site between the KAN resistance gene and the lacZY genes of the transposon. Using the KAN resistance gene as a probe, Southern hybridization analysis was used to determine which fragment contained the fusion. After identification of potential fragments for cloning, the genomic DNA was digested with the selected enzyme and separated on an agarose gel. The approximately sized fragments, as determined from the Southern blot, were excised from the gel and the DNA was purified by Gene-Clean. These fragments were then ligated into the cloning vector pHG329 which had been linearized with the same enzyme. After the clones were obtained, these were sequenced using a primer, P6, designed from the transposon end which allows sequencing through the junction and into the disrupted gene (12).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The rscR mutation is responsible for increased dissemination to the spleen. The nucleotide sequence of rscR, formerly hre20, and the surrounding DNA was determined (Fig. 1). The predicted amino acid sequence of RscR is 68% identical to the hypothetical protein YeiE and 24% identical to LysR of E. coli. Upstream of rscR is a divergently transcribed ORF with 55% identity to YeiH, the hypothetical protein encoded upstream of yeiE in E. coli. YeiH has no homology to any other known proteins in the database. The potential translational start site of the yeiH homologue is 560 bp from the rscR initiation codon. Downstream of rscR is an ORF with 80% identity to LysP of E. coli, which is also located downstream of yeiE in E. coli. The translational start site of the LysP homologue is 244 bp from the stop codon of rscR.

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Schematic of the rscR locus showing the gene arrangement. DNA fragments and restriction sites are indicated for relevant constructs. Only partial gene length is shown for the yeiH and lysP homologues. Numbering begins at left from position 0. The fragments (Frag.) subcloned to generate pKN7, pKN14, and pKN15 are shown. In addition the region deleted to generate the in-frame rscR mutation is indicated.

Screening for rscR-regulated genes. rscR has been demonstrated to play a role in the course of infection by Y. enterocolitica. Since RscR has homology to the LysR family of transcriptional regulators, it was presumed that RscR does not have a direct effect on the course of infection but instead may regulate one or more genes involved in pathogenesis. Therefore, it was of interest to determine what genes are regulated by RscR, because these would be expected to be the effectors of the virulence phenotype. The screen for RscR-regulated genes utilized transposon mutagenesis (Fig. 2). The transposon used was an mTn5 derivative containing a KAN resistance gene and a promoterless lacZY on the plasmid designated pRev10. Transposition into a given gene in the correct orientation will generate a transcriptional fusion to lacZY.

View larger version (47K): [in this window] [in a new window]   FIG. 2.   Flowchart depicting the screen for RscR-regulated genes.

Identification and sequencing of RscR-regulated genes. Southern blot analysis indicated that the eighteen mutants belong to four allelic groups (Fig. 3). Southern blotting was performed using the KAN resistance gene as a probe. Allelic groups were determined on the basis of the same size fragments containing the KAN resistance gene for at least two different restriction enzyme digests (data not shown). To identify the potential RscR-regulated genes, transposon-chromosome junctions were cloned. Sequence data were obtained by using a primer that annealed to sequences at the transposon end. This allowed the DNA sequence of the cloned junction to be determined. The sequence was then analyzed for potential ORFs and used to search available databases for sequence homology. Other than the mutant 39A5, representative sequence data were obtained for each allelic group.

View larger version (31K): [in this window] [in a new window]   FIG. 3.   Chart displaying levels of induction of putative RscR-regulated fusions. Fold induction represents -galactosidase activity in 0.2% arabinose divided by -galactosidase activity in 0.2% glucose. The allelic groups are represented by braces, and identified loci are indicated. Representative assays are shown. Assays were performed three times on at least two different days.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Comparison of the H. influenzae hmwABC locus and the Y. enterocolitica rscBAC locus. Nucleotides are numbered starting from the putative translational start sites of hmwA and rscB. The fragment utilized for creation of (rscA::lacZYA) and the fragment deleted in rscA are indicated.

Expression of the rsc locus. The 13 insertions in the rscBAC locus all map to the hxuB homologue, rscB, although the gene arrangement suggests an operon structure in which insertions could have been isolated from each of the three ORFs. To begin to address the question of why transposon insertions were not isolated from rscA and rscC, a chromosomal lacZYA fusion to rscA, the putative effector, was generated by integration of a suicide plasmid, pFUSE, carrying a fragment of rscA. (rscA::lacZYA) was generated in JB580v, rscR, and rovA::erm (YVM641), to create strains YVM796, YVM797, and YVM885, respectively. RovA regulates the expression of inv and is expected to regulate other genes in Y. enterocolitica (24). Levels of expression from the rscA promoter as determined by monitoring -galactosidase activity from (rscA::lacZYA) from these three strains were equivalent (141 ± 2, 140 ± 3, and 141 ± 2 Miller units, respectively [results are given as means ± standard deviations]). To recreate the conditions of the screen, plasmid pKN15 containing an arabinose inducible rscR was moved into the rscR mutant containing (rscA::lacZYA). When rscR was induced with 0.2% arabinose the -galactosidase activity from (rscA::lacZYA) was 318 ± 31 Miller units. When rscR expression was repressed by 0.2% glucose, the -galactosidase activity from (rscA::lacZYA) was 84 ± 2 Miller units. In the presence of RscR, rscA was induced fourfold, which is an induction level that would have been excluded in the original screen. As a control, plasmid pBAD18 was also moved into the rscR mutant containing (rscA::lacZYA). No induction was observed with the control plasmid; -galactosidase activity from (rscA::lacZYA) was 84 ± 3 Miller units in arabinose and 100 ± 3 Miller units in glucose. This suggests that transposon insertions in rscA or rscC may not have displayed sufficient differential expression to be selected in the screen, and this could explain why none were identified.

Analysis of an rscA mutant. In order to determine if the rscBAC locus encodes the factor responsible for the increased splenic colonization observed for an rscR mutant, a deletion was made in the rscA gene, designated rscA (YVM776). This gene was chosen because by analogy to H. influenzae RscA would be the predicted effector and RscB and RscC would be accessory proteins. The ORF of rscA is 6,195 bp, and the deletion removes 5,131 bp to create the rscA mutant (see Materials and Methods). BALB/c mice were infected orally with 10⁸ CFU of the rscA mutant, the rscR mutant, or JB580v; the rscR strain and JB580v were tested in parallel for comparison. For each strain, 12 mice were infected. On days 4 and 6 postinfection, six mice per strain were sacrificed, and the MLN and spleen were harvested and viable cell counts were assessed for each tissue. On day 4, mice infected with the parental strain looked similar to mice infected with the rscA mutant or the rscR mutant (Fig. 5A). By day 6, more mice infected with either the rscA mutant or the rscR mutant had a high bacterial load in the spleen compared to mice infected with the parental strain (Fig. 5B). A second experiment again demonstrated increased bacterial dissemination to the spleen of mice infected with the rscA mutant or the rscR mutant compared to JB580v at day 6 postinfection (Fig. 5C). These results were significantly different compared to those from wild-type-infected mice, with P values of 0.008 and 0.03 for rscA and rscR, respectively. In both experiments, it appeared that the rscA mutation caused a more severe phenotype than the rscR mutation (P = 0.06). These results demonstrate that a mutation in rscA mimicked the phenotype observed when mice were infected with the rscR mutant and potentially a mutation in rscA causes an even more severe phenotype than a mutation in rscR.

View larger version (56K): [in this window] [in a new window]   FIG. 5.   Kinetics of infection of wild type, rscR, and rscA. Viable cell counts recovered from the MLN and spleen are indicated (log scale). Each bar represents data from one animal. ND, no data. (A) Viable cell counts recovered from mice at day 4 postinfection. (B) Viable cell counts recovered from mice at day 6 postinfection. Results are from the same experiment as shown in panel A. (C) Viable cell counts recovered from mice at day 6 postinfection. These results are from a separate experiment from that shown in panels A and B.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A previous study by Young and Miller examined the effect on infection of an insertion in rscR (formerly called hre20) and found that the mutant spread more rapidly to the liver and spleen of infected mice than did the parental strain (36). Sequencing of the region surrounding rscR showed a gene arrangement that is similar to the organization of yeiE and lysP in E. coli, with a homologue of lysP downstream of rscR. In this study, we demonstrated that an in-frame deletion in the rscR gene also results in increased dissemination of bacteria to the spleen after oral infection of BALB/c mice. This suggests that the phenotype of the previously described rscR mutant was not a result of polar effects on the downstream gene. This was confirmed by comparing the kinetics of infection for mice infected with wild-type bacteria to mice infected with a mutant containing the nonpolar rscR allele.

After establishing that RscR plays a role in the course of infection, we wanted to identify genes that were regulated by RscR, because RscR is predicted to be a LysR-type transcriptional regulator. Members of the LysR family of transcriptional regulators typically utilize a coinducer for optimal function (26). Hence, it was of interest to determine in vitro conditions which could produce optimal levels of RscR and would more likely be permissive for a coinducer to be present if one is required. In vitro conditions which allowed expression of rscR from its own promoter were not identified, and the presence or absence of a necessary coinducer can only be speculated. This made it necessary to artificially induce expression of rscR from a P_BAD promoter for the purpose of screening for genes regulated by RscR. A strain containing an inducible rscR was mutagenized with the mTn5Km2-lacZY transposon. However, because RscR levels may not be physiological or a necessary coinducer may not be present, some genes regulated by RscR may have been missed in the screen and the level of regulation observed is not likely to be representative of true in vivo values.

Several RscR-regulated fusions were identified in the screen and placed into four allelic groups based on Southern blot analysis. The 39A5 mutant was not cloned or sequenced. The second group also contained only one fusion, 24A2, and although this junction was cloned and sequenced, the sequence did not display significant similarity to any known proteins in the databases. The third group contained three fusions to a gene exhibiting homology to ompF. These fusions demonstrated differential expression in response to RscR levels (10-fold induction in arabinose) but also displayed differential expression to arabinose and glucose in the absence of RscR (threefold induction in arabinose). Studies in other systems have shown ompF to be regulated by many factors, including the carbon source (23). The fourth group of 13 insertions was located in a 12-kb region of the chromosome. This region was sequenced, and three ORFs were identified which were designated rscBAC (for reduced splenic colonization).

Although indirect, the observed in vivo phenotype of the rscA mutant indicates that RscR regulates expression of rscBAC in the host environment. If the presence of RscR is necessary for proper expression of the rscBAC locus, then one would predict that a mutation in the regulator would have an effect on infection similar to that of a deletion of one of the genes in the locus. A comparison of the kinetics of infection of the parental strain, the rscR mutant, and the rscA mutant demonstrated that for both mutants there were increased numbers of mice with bacterial colonization of the spleen, with the rscA mutant colonizing the largest number of spleens by day 6 postinfection. The phenotype of a rscA mutant appears slightly more severe than that of a rscR mutant, which could be explained if there is a basal level of transcription of the rscBAC locus in the absence of RscR. If this were the case, then a rscR mutant would be producing a low level of RscA during infection and this could prevent the phenotype from being as severe as a rscA mutant which expresses no functional RscA. These data along with the reporter fusion studies suggest that RscR, directly or indirectly, activates rscA expression.

The rscBAC locus is most similar to the hmwABC locus of H. influenzae. HmwA is a surface-expressed protein which promotes adherence to various epithelial cell lines in vitro and has been demonstrated to be expressed in clinical isolates from patients with acute otitis media (17). HmwA is expressed as a 150-kDa protein which contains a nontraditional signal sequence of 68 amino acids (9). These 68 amino acids are cleaved after secretion across the inner membrane by the general secretory pathway. A second cleavage event takes place after amino acid 441, concurrent with secretion across the outer membrane, to produce the mature 125-kDa protein, though this processing has been shown not to be essential for adherence. HmwB is essential for surface expression and the second cleavage of HmwA (30). Mutations in hmwB lead to the presence of unprocessed HmwA and probable degradation in the periplasm. Mutations in hmwC result in reduced levels of HmwA, and the protein is present in the unprocessed form. Null mutations in both hmwB or hmwC result in loss of adherence.

RscA has a putative nontraditional signal sequence which contains a number of charged residues in the first 47 amino acids followed by a typical signal sequence with a predicted cleavage site after residue 69. This is the same distribution of residues present in the HmwA 68-amino-acid signal sequence (9). Additionally, the second cleavage event of HmwA at amino acid 441 contains WLLDP, and this motif is present in RscA at residues 460 to 464, with predicted cleavage after residue 463. Another motif, NPNGI, is present in the amino terminus of HmwA (residues 150 to 154) as well as three related proteins: ShlA of S. marcescens, HpmA of Proteus mirabilis, and HhdA of Haemophilus ducreyi (20, 27, 30, 33). This family of proteins is grouped together based upon the mechanism of processing and secretion. Each protein has an associated outer membrane protein that functions in processing the protein to its mature form and in transport across the outer membrane. The NPNGI sequence is suggested to be a point of contact with the outer membrane transporter, which is identified as HmwB for HmwA (14, 28). The NPNGI motif is also present in RscA at residues 154 to 158, suggesting that RscA is a new member of this family of proteins. The hmwABC locus is unique in that it encodes a third functional protein, HmwC, which does not have homologues in the other related loci (2). RscC is the first identified homologue of HmwC, and this could suggest that HmwABC and RscBAC are functionally similar. RscA has the highest similarity to the adhesin HmwA, and the similarity spans the entire protein. However, RscA does have significant homology to other outer membrane and some secreted proteins with diverse functions and activities. For example, RscA is similar to a filamentous hemagglutinin-like protein of H. ducreyi which is suspected to be involved in forming lesions in the rabbit model of infection (35). RscA is also similar to the ShdA protein of Salmonella enterica, which is involved in fecal shedding of the bacteria and believed to be important in the spread of the bacteria among livestock and domestic fowl populations (16).

Adhesins have been shown to be important in the pathogenesis of numerous bacteria. Initial colonization of a pathogen at a particular site often requires a surface molecule which promotes adherence to the host cells. From these experiments it does not appear that RscA would function in initial colonization. Kinetic analysis of the rscR and rscA did not demonstrate a defect in the colonization of the PP, the initial site of a Y. enterocolitica infection (data not shown). In contrast, a mutation in inv, the gene encoding the surface protein invasin, results in greatly decreased colonization of the PP at early time points postinfection (21). Adhesins can function in later steps of infection as well. The YadA adhesin of Y. enterocolitica has been demonstrated to be important in the persistence of bacterial colonization in the PP (22). It is possible that RscA could function as an adhesin for a particular cell type during infection, and loss of adherence in an rscA mutant results in the altered kinetics seen in this study. Future experiments will examine the ability of the rscBAC locus to promote adherence to different cell types in vitro.

This study demonstrated that the RscR protein affects the normal progression of disease in the BALB/c mouse model of Y. enterocolitica infection. The rscBAC locus was identified in a transposon mutagenesis screen to be regulated by RscR, and a deletion in the rscA gene results in an alteration of the normal infection kinetics of Y. enterocolitica similar to that of an rscR mutant. The rapid dissemination to the spleen of the rscR and rscA mutants is an unusual phenotype. This alteration of infection kinetics could impact the virulence of Y. enterocolitica. Further study should elucidate more specifically the role that the rscBAC locus and related loci in other pathogenic bacteria play during infection.