The Yersinia pestis gcvB gene encodes two small regulatory RNA molecules

Background In recent years it has become clear that small non-coding RNAs function as regulatory elements in bacterial virulence and bacterial stress responses. We tested for the presence of the small non-coding GcvB RNAs in Y. pestis as possible regulators of gene expression in this organism. Results In this study, we report that the Yersinia pestis KIM6 gcvB gene encodes two small RNAs. Transcription of gcvB is activated by the GcvA protein and repressed by the GcvR protein. The gcvB-encoded RNAs are required for repression of the Y. pestis dppA gene, encoding the periplasmic-binding protein component of the dipeptide transport system, showing that the GcvB RNAs have regulatory activity. A deletion of the gcvB gene from the Y. pestis KIM6 chromosome results in a decrease in the generation time of the organism as well as a change in colony morphology. Conclusion The results of this study indicate that the Y. pestis gcvB gene encodes two small non-coding regulatory RNAs that repress dppA expression. A gcvB deletion is pleiotropic, suggesting that the sRNAs are likely involved in controlling genes in addition to dppA.


Background
Yersinia pestis is the causative agent of plague, an infectious disease that results in lymphatic and blood infections [1]. The Y. pestis genome has been sequenced [2,3]. Y. pestis carries three plasmids of approximately 9.5, 70, and 100 kilobasepairs and each carries genes necessary for or that contribute to the pathogenicity of the bacterium [1]. The 70 kilobasepair plasmid encodes the low-calcium response stimulon (LCRS). Components of the LCRS include Yops (secreted anti-host proteins) and a type III secretion apparatus, or Ysc. The type III secretion apparatus is responsible for the translocation of the Yops to host cells that in turn down-regulate the response of the host phagocytic cells to infection [4]. Natural LCRS-negative mutants of Y. pestis occur, resulting in avirulence of the bacteria [1]. Besides the three plasmids, another pathogenicity factor is pigmentation. Cells of Y. pestis adsorb hemin at 26°C but not at 37°C and are pigmented (Pgm + ) and virulent. Spontaneous nonpigmented (Pgm -) mutants of Y. pestis have been isolated. The Yersiniabactin iron transport system is part of the pgm locus, and its loss results in a Pgmmutant that is avirulent in mice unless hemin, ferrous sulfate, or ferric chloride is injected into mice along with the bacterial challenge [1].
Recently, a new class of molecules has been shown to regulate gene expression in bacteria, small non-coding regulatory RNAs (sRNAs). These sRNAs have gained much attention as recent genome-wide studies have identified sRNAs in a wide variety of organisms [5]. Most of these bacterial sRNAs are between 50 and 400 nucleotides (nts) in length and play important roles in global regulation [6,7]. Hfq is a small RNA binding protein and sRNAs in particular are targets for Hfq [6]. Binding of these sRNAs by Hfq in some way facilitates base pairing between the sRNAs and their respective target RNAs [8,9]. In Vibrio cholerae, sRNAs (Qrr RNAs) have been shown to regulate virulence genes [10] and in Brucella abortus an hfq mutation is lethal [9]. These results suggest that sRNAs and Hfq likely play important roles in the virulence of certain Gram-negative pathogens.
The E. coli gcvB gene encodes sRNAs that are not translated in vivo [11]. A strain carrying a deletion of gcvB has constitutive synthesis of OppA and DppA, periplasmic binding proteins of the two major peptide transport systems nor-mally repressed in cells grown in rich medium [12][13][14]. In addition to OppA and DppA, several other proteins were shown to increase or decrease in response to GcvB RNA levels, but the specific proteins were not identified [11]. Nevertheless, the results show that the GcvB RNAs are regulatory and possibly serve as global regulators. A computer search of the Y. pestis sequence showed that Y. pestis has a gcvB gene that shares considerable sequence homology with the E. coli gcvB sequence (Fig. 1). Thus, the GcvB RNAs from E. coli likely have functional counterparts in Y. pestis. The results of this study show that the Y. pestis gcvB gene encodes two sRNAs that, in turn, have regulatory activity. In addition, a deletion of the gcvB gene from the Y. pestis chromosome alters growth rate and colony morphology.

Results and discussion
Identification of the Y. pestis gcvB gene The E. coli gcvB gene is divergently transcribed from gcvA, which encodes the activator protein for gcvB expression ( Fig. 1) [11]. Thus, we used computer searches of genome Comparison of the E. coli and Y. pestis gcvA/gcvB control regions and gcvB genes Figure 1 Comparison of the E. coli and Y. pestis gcvA/gcvB control regions and gcvB genes. Ec, E. coli; Yp, Y. pestis. Bases that are identical are boxed in gray. The E. coli promoter -10 and -35 sequences are underlined for gcvA [29] and overlined for gcvB [11]. Arrows indicate transcription start sites and directions of transcription of gcvA and gcvB. The GcvA binding region is indicated above the sequence [30]. The deduced Y. pestis -10 and -35 promoter sequences are underlined for gcvA and overlined for gcvB, and the deduced GcvA binding site is indicated above the sequence. Two Rho-independent transcription terminators for the E. coli and Y. pestis gcvB genes are indicated by inverted arrows. The fusion points for transcriptional fusions gcvB +53 ::lacZ (TF-1), gcvB +164 ::lacZ (TF-2) and gcvB +251 ::lacZ (TF-3) are indicated by vertical arrows. The gcvB gene from Y. pestis possesses two possible Rhoindependent transcription terminators, which if functional, would allow the production of two sRNAs of about 130 nts and 206 nts (Fig. 1). Three transcriptional gene fusions of the Y. pestis gcvB gene to lacZ were created to determine if these putative terminator sites function as transcription terminators in vivo. The three fusions, designated λgcvB Yp+53 ::lacZ, λgcvB Yp+164 ::lacZ and λgcvB Yp+251 ::lacZ, were used to lysogenize E. coli strain GS162. The lysogens were then grown in Luria-Bertani broth (LB) [16] to mid-log phase of growth and the cultures assayed for β-galactosidase activity. About 44% of the β-galactosidase activity seen when the fusion precedes both terminators (λgcvB Yp+53 ::lacZ) is lost when the fusion point follows terminator t1 (λgcvB Yp+164 ::lacZ), implicating t1 as a site of transcription termination in vivo ( Table  1). The remaining activity that escapes termination by t1 is not seen in GS162λgcvB Yp+251 ::lacZ, indicating t2 also functions as a terminator in vivo (Table 1). When the 206 nts preceding terminator t2 for gcvB were analyzed, there were only short open reading frames (ORFs) that could encode polypeptides of 36 amino acids or less. These ORFs all lack good translational start sites and were not tested to determine if they encode small polypeptides. The E. coli GcvB RNAs are not translated into polypeptides [11]. Thus, we conclude that the products of the gcvB gene in Y. pestis are two sRNAs that are not translated, although

E. coli Y. pestis V. cholerae
the results do not completely rule out the possibility that the Y. pestis GcvB sRNAs encode small peptides. In E. coli, about 90% of the transcripts that initiate at the gcvB promoter terminate at terminator t1, and the remaining 10% terminate at terminator t2 [11]. A comparison of the E. coli and Y. pestis t1 sites shows that an additional 2 bps occur between the predicted GC-rich stem-loop structure and the run of T residues in the Y. pestis t1 site that are not present in the E. coli sequence, suggesting that the Y. pestis t1 site is likely less functional as a transcription terminator than the E. coli t1 site ( Fig. 1). Nevertheless, the results are consistent with the Y. pestis gcvB gene encoding two sRNA molecules of about 130 and 206 nts and in roughly equal amounts.
We used Northern blotting to confirm that the gcvB locus in Y. pestis encodes sRNA transcripts of about 130 and 206 nts. Two small RNA molecules were detected in RNA isolated from Y. pestis KIM6 grown in HIB medium using a probe specific for the gcvB locus (Fig. 3). These results are consistent with the in vivo results with the Y. pestis gcvB transcriptional fusions.

Regulation of the Y. pestis gcvB gene
The E. coli gcvB gene is activated by GcvA in the presence of glycine and repressed by GcvA + GcvR in its absence; this repression is enhanced by the addition of purines [11]. The regulation of the Y. pestis gcvB gene was tested with respect to the effects of glycine and purine supplementation to the growth medium and with respect to the GcvR and GcvA proteins and the GcvB RNAs. For these experiments we used the λgcvB Yp+53 ::lacZ fusion to lysogenize appropriate E. coli host strains. The lysogens were grown in glucose minimal (GM) or GM supplemented with glycine or inosine to mid-log phase of growth and assayed for β-galactosidase levels. In the wild-type (WT) GS162λgcvB Yp+53 ::lacZ lysogen, the addition of glycine to GM growth medium resulted in an 11.5-fold induction of β-galactosidase expression, whereas the addition of the purine inosine resulted in a 2.5-fold repression below the unsupplemented GM level ( Table 2, line 1). In the gcvA mutant lysogen GS1118λgcvB Yp+53 ::lacZ, the β-galactosidase levels were low and non-inducible by glycine (  [17,18] awaits further investigation. Furthermore, there appears to be no autoregulation of gcvB by its own sRNA products as the gcvB mutant lysogen GS1144λgcvB Yp+53 ::lacZ shows normal regulation of the gcvB Yp+53 ::lacZ fusion ( Table 2, line 4).

Y. pestis gcvA encodes an activator protein for gcvB expression
Since activation of the Y. pestis gcvB Yp+53 fusion in E. coli was dependent on GcvA (Table 2), we determined if the Y. pestis gcvA gene also encodes an activator protein for gcvB expression. We assumed this would be the case, as the E. coli and Y. pestis GcvA proteins are 88% identical at the Northern blot analysis of GcvB from Y. pestis strain KIM6 Figure 3 Northern blot analysis of GcvB from Y. pestis strain KIM6. Total cell RNA was isolated from strain KIM6 grown in HIB at 30°C to an O.D. 600 of 0.7 and probed with a 32 P-labeled GcvB specific DNA probe as described in Methods. Two gcvB transcripts of about 206 and 130 nucleotides identified are indicated with arrows. Their sizes were determined based on their mobilities relative to the mobility of the E. coli GcvB RNA and 5S rRNAs (not shown). amino acid sequence level. The Y. pestis gcvA gene was cloned into plasmid pACYC177 and tested for its ability to complement an E. coli gcvA mutant. The E. coli strain GS1132 carries a deletion of the gcvA gene [11]. This strain was lysogenized with an E. coli λgcvB::lacZ transcriptional gene fusion and subsequently transformed with the control plasmid pACYC177, or pACYC177 carrying either the E. coli or the Y. pestis gcvA gene. The cells were grown in LB to mid-log phase of growth and assayed for β-galactosidase activity. As reported previously [11], expression of the E. coli gcvB::lacZ fusion was increased about 400-fold in the presence of the E. coli gcvA gene (   Whether the Y. pestis GcvA and GcvR proteins also regulate the Y. pestis gcvTHP operon, or have additional regulatory roles, awaits further investigation.

The Y. pestis GcvB RNAs regulate the E. coli and Y. pestis dppA genes
The E. coli gcvB gene negatively regulates the dppA and oppA genes [11]. In addition, many other genes were shown to be either negatively or positively regulated by the GcvB RNAs [11]. Thus, the E. coli GcvB RNAs are likely global regulators of gene expression. Y. pestis has homologs of dppA and oppA. To determine if the Y. pestis GcvB RNAs are regulatory, we transformed an E. coli Δ gcvB λdppA::lacZ lysogen with pgcvB Yp-p322 , the transformant and the parent lysogen were grown in LB to mid-log phase of growth and assayed for β-galactosidase levels. As expected, deletion of gcvB caused an increase in dppA::lacZ expression (  [11]. We are constructing a plasmid that will only produce the 130 nucleotide Y. pestis GcvB RNA to determine whether the 130 or 206 nucleotide RNA species is required for activity in Y. pestis.

Deletion of the Y. pestis gcvB gene slows growth rate and alters colony morphology
The KIM6Δ gcvB strain routinely gave smaller colonies on HIB plates than the parent KIM6 strain. Thus, we investigated the growth of KIM6Δ gcvB to determine the effect of the ΔgcvB mutation on growth rate. The parent strain KIM6, KIM6ΔgcvB and KIM6ΔgcvB [pgcvB Yp-sc ] were grown in HIB broth at 37°C. The generation times were then calculated. The KIM6 generation time at 37°C was 135 ± 15 minutes whereas KIM6ΔgcvB had a generation time of 194 ± 20 minutes (Fig. 4). The presence of pgcvB Yp-sc in KIM6ΔgcvB complemented the gcvB deletion, as the generation time was reduced to 155 ± 8 minutes, close to the generation time of strain KIM6. Thus, deletion of the gcvB gene impairs the ability of Y. pestis to grow as well as the parent strain on either solid media or in liquid media. This is in contrast to E. coli gcvB deletion mutants that  In E. coli, many genes respond to the GcvB RNAs [11]. The pleiotropic nature of the Y. pestis gcvB deletion suggests that the Y. pestis GcvB RNAs are likely global regulators as well. Identification of the specific genes regulated by the GcvB RNAs that are responsible for the altered phenotype will allow us to test directly their involvement in virulence of the organism. In addition, the GcvB sequences and regulatory regions from bp -90 to +1, which include the putative GcvA binding sites for activation of gcvB, are 100% identical in all Yersina pestis strains, and greater than 92% identical in other Yersinia species. Thus, expression of gcvB and the regulatory mechanisms of the GcvB RNAs are likely similar in all Yersinia species.

Conclusion
In summary, the Y. pestis gcvB gene is activated by the GcvA protein and repressed by the GcvR protein. The gcvB gene encodes two sRNAs that have regulatory activity, repressing dppA expression. A gcvB deletion is pleiotropic, suggesting that the GcvB RNAs possibly serve as global regulators in Y. pestis.

Bacterial strains, plasmids and phage
Bacterial strains, plasmids and phage used in this study are listed in Table 6 or are described in the text.

Media
For E. coli strains, the complex medium used was LB [16] and the defined medium used was the minimal salts of Vogel and Bonner [20] supplemented with 0.4% glucose. GM media was always supplemented with 50 μg ml -1 of phenylalanine and 1 μg ml -1 of vitamin B1, since all E. coli strains carry pheA, thi mutations. Where indicated, glycine and inosine were added at 300 μg ml -1 and 50 μg ml -1 , respectively. For Y. pestis strains, HIB was used [21]. Agar was added at 1.5% to make solid media. Antibiotics were added at the following concentrations: AP, 150 μg ml -1 for multi-copy plasmids and 50 μg ml -1 for single-copy plasmids; chloramphenicol (CM), 20μg ml -1 ; tetracycline (TC), 10 μg ml -1 .
β-galactosidase assays β-galactosidase assays were performed on mid-log phase cells (OD 600~0 .5) as described by Miller [16]. Each experiment was repeated at least twice, with each sample assayed in triplicate.

DNA manipulation
Plasmid DNA was isolated using Qiagen Miniprep kits as described by the manufacturer (Qiagen). Restriction enzyme digestions and DNA ligations were carried out according to the manufacturer (New England Biolabs). DNA sequencing was performed by the University of Iowa DNA Core Facility.

PCR
PCR reactions were performed in 100 μl volumes. Each reaction mixture contained 10 μl 10 × polymerase buffer, 10 μl 10 × dNTPs (0.2 mM each), 5 μl Y. pestis DNA (~15 ng), 100 pmoles of forward and reverse primers designed specifically for each reaction, 1 μl of vent polymerase, and sterile water to bring the volume to 100 μl. PCR reactions were carried out under the following conditions: 5 min pre-incubation at 95°C, and then 30 cycles of 95°C for 30 sec, 45°C for 30 sec, and 72°C for 2 min.

Klett units
Membrane (ISC BioExpress). The blot was hybridized with a PCR generated DNA fragment from bp +1 to +198 of the Y. pestis gcvB gene and 32 P-labeled using the Redip-rimeTM II Random Prime Labeling System (Amersham Biosciences). Hybridization of the blot was at 58°C as described [22].

Construction of lacZ gene fusions
Three different transcriptional gene fusions of gcvB to the lacZ gene were constructed by PCR synthesis of fragments with common BamHI termini 128 bp upstream of the gcvB transcription start site and 3 different fusion points within gcvB. In plasmid pB Yp+53 ::lacZ, the downstream PCR primer hybridized to the gcvB sequence beginning at bp +53 relative to the predicted transcription start site (+1) of gcvB (Fig. 1). A synthetic HindIII site was included at the end of the primer to allow the cloning of the 202 bpBamHI-HindIII fragment into the BamHI-HindIII sites of the lacZ transcriptional reporter plasmid pQF50 [25]. Plasmids pB Yp+164 ::lacZ and pB Yp+251 ::lacZ were constructed similarly except that the downstream primers used for PCR synthesis hybridized to the gcvB sequence beginning at bp +164 and +251 (Fig. 1), and the 313 and 400 bp fragments produced were cloned into the BamHI-HindIII sites of pQF50. Each fusion was sequenced at the University of Iowa DNA Core Facility to verify that the fusions were at the correct sites and that no bp changes were introduced during the PCR amplification procedure. Each gcvB transcriptional fusion was then subcloned into plasmid pMC1403 [26], generating plasmids pgcvB Yp+53 ::lacZ, pgcvB Yp+164 ::lacZ and pgcvB Yp+251 ::lacZ, and subsequently transferred to phage λgt2 [27] as described [11], generating phage λgcvB Yp+53 ::lacZ, λgcvB Yp+164 ::lacZ and λgcvB Yp+251 ::lacZ, respectively.
A single-copy Y. pestis dppA Yp ::lacZ translational fusion was constructed in two steps. First, a dppA Yp ::lacZ translational fusion was constructed using an upstream PCR primer with an EcoRI site complementary to the Y. pestis DNA sequence beginning 300 bps upstream of the dppA transcription initiation site and a downstream primer that contains an artificial SmaI site and that hybridizes to the dppA sequence after the 15 th codon relative to the translation initiation site. The 611 bp dppA fragment was cloned into the EcoRI and SmaI sites of the lacZ translational reporter plasmid pMC1403. The fusion was sequenced at the University of Iowa DNA Core Facility to verify that the fusion was at the correct site and that no bp changes were introduced during the PCR amplification procedure. The dppA Yp ::lacZ fusion, along with the lacY and lacA genes, was then cloned into the single-copy plasmid pGS366, designated pdppA Yp ::lacZ.

Chromosomal deletion of gcvB
A gcvB deletion was constructed on the Y. pestis chromosome essentially as described [28]. Y. pestis strain KIM6 was transformed with plasmid pKD46, which encodes the Red recombinase of phage λ [28]. PCR products were then generated using two primers with 50 nt extensions that are complementary to sequences that flank the gcvB gene and 20 nt priming sequences that are complementary to the template plasmid pKD32 and that flank the CM R gene and the FLP recognition sequence [28]. The PCR fragment was gel purified and used to transform Y. pestis KIM6 [pKD46]. The cells were plated on HIB plates with CM and CM R recombinants were selected. One CM R recombinant was single colony purified, chromosomal DNA was prepared, and PCR analysis was used to verify that the gcvB gene was deleted and replaced with the CM R marker. The pKD46 plasmid is a temperature sensitive replicon and was cured by growth at 37°C [28]. The strain was designated KIM6Δ gcvB.

Authors' contributions
SM carried out most of the genetic experiments and wrote the first draft of the manuscript. SP carried out the genetic experiments with gcvR and also performed the Northern analysis. GS carried out the computer search to identify putative gcvB genes in other organisms and was the principal investigator and supervised the project.