ABSTRACT
Yersinia pestis, the etiologic agent of plague, has only recently evolved from Yersinia pseudotuberculosis. hfq deletion caused severe growth restriction at 37°C in Y. pestis but not in Y. pseudotuberculosis. Strains from all epidemic plague biovars were similarly affected, implicating Hfq, and likely small RNAs (sRNAs), in the unique biology of the plague bacillus.
Yersinia pestis, the etiologic agent of plague, poses a continuing natural threat to human health, and recent concerns about its potential use as a biological threat agent have increased our need to understand the biology of this deadly bacterial pathogen. Molecular evidence indicates that Y. pestis evolved from Yersinia pseudotuberculosis within the last 20,000 years (1), and three-fourths of their genes remain at least 97% identical (8). The diseases caused by these two bacteria differ greatly, and only Y. pestis has the capacity for the complex process of flea-borne transmission (15, 24, 31). The extent to which biological differences between these distinct bacterial pathogens are caused by altered expression of conserved genes, rather than overt genetic differences, is not known.
Hfq functions as a posttranscriptional regulator by stabilizing small RNAs (sRNAs) and facilitating their interactions with mRNA targets (18, 20, 21, 35, 41, 43, 45, 48). Hfq's pleiotropic role in gene regulation is well established (29, 42, 43), and its importance as a regulator of virulence in pathogenic bacteria is being increasingly recognized (3, 9, 10, 12, 25, 27, 28, 33, 37, 38, 47). Hfq positively regulates the Y. enterocolitica-specific yst toxin gene (30) and is required for virulence of Y. pseudotuberculosis (34) and a Pestoides variant of Y. pestis (19). However, the role of Hfq in the epidemic Y. pestis biovars has not been explored.
Our interest in the regulation of differential gene expression led us to generate hfq deletion mutants of Y. pestis and Y. pseudotuberculosis. Preliminary studies suggested species-specific differences in the growth properties of these mutants. In this study, we report that epidemic Y. pestis strains are strongly dependent on Hfq for in vitro growth at 37°C but that Y. pseudotuberculosis strains are not. These results indicate the potential for significant differences in the roles of sRNAs in the physiology of these genetically related, but biologically distinct, bacterial pathogens.
Growth and cell morphology of the hfq mutant strains.
The Hfq amino acid sequences are identical among eight sequenced strains of Y. pestis and four of Y. pseudotuberculosis (http://www/ncbi.nlm.nih.gov/Blast.cgi ). This protein shares 98.6% identity with Escherichia coli Hfq over a 73-amino-acid (aa) N-terminal portion that is associated with RNA chaperone function (39). This shared region is also highly conserved in Pseudomonas aeruginosa Hfq, which contains only 82 aa (40).
Deletion of hfq has a moderate effect on E. coli cell morphology, tolerance of stress conditions, and growth rate in rich media, which is more pronounced at 25°C than at 37°C (42). We initially explored Hfq's role in the growth of Y. pestis and Y. pseudotuberculosis by generating hfq knockout mutations in Y. pestis KIM6+ (Pgm+ pCD1− pMT1+ pPst+) and KIM10+ (Pgm+ pCD1− pMT1+ pPst−), as well as Y. pseudotuberculosis PTB52c (pYV−; serotype IB) and PTB54c (pYV−; serotype III). Mutations were generated by using homologous recombination to replace the entire hfq open reading frame (ORF) (306 bp) with a cat (chloramphenicol acetyltransferase) gene cassette amplified from pKD3 (11). Mutations were initially complemented with a single-copy hfq gene expressed from its native promoter and integrated into the chromosome 3′ to one of three tRNAAsn genes using pSPC471, which contains the YP-HPI pathogenicity island's attachment and integration sequences (6, 23) subcloned from pKR529 (generously provided by A. Rakin).
Bacterial growth in brain heart infusion (BHI) media was measured visually on agar plates or by optical density (OD) in liquid cultures at 28 and 37°C. Cell morphology was also examined by microscopy at selected time points. Growth rates and morphology of the Y. pseudotuberculosis hfq mutants were similar to those of wild-type (WT) cells at either 28°C or 37°C (Fig. 1 and 2). In contrast, the Y. pestis mutants exhibited significant temperature-dependent phenotypes. Growth of the Y. pestis KIM6+ and KIM10+ hfq mutants was only moderately reduced at 28°C relative to that of the WT, but they showed little to no growth on either solid or liquid BHI medium at 37°C (Fig. 1). The Y. pestis hfq mutants also formed lengthened rods, which were significantly different from the sphere and short-rod morphology of WT, particularly at 37°C (Fig. 2).
Comparison of growth of hfq mutants with solid (A) and liquid (B) media. (A) WT, hfq null mutant (hfq), and complemented hfq mutant (com) strains of Y. pestis KIM6+, and Y. pseudotuberculosis PTB52c were streaked on BHI plates and incubated for 2 days at 28°C or 37°C, as indicated. (B) Growth curves of WT, hfq mutant (hfq), and complemented mutant (hfq-C) strains of Y. pestis KIM6+ and Y. pseudotuberculosis PTB52c in BHI broth at 28°C or 37°C, measured by optical density. With both solid and liquid media, Y. pestis KIM10+ results were similar to those for Y. pestis KIM6+ strains, while results for Y. pseudotuberculosis PTB54c were similar to those for PTB52c (not shown).
Morphological comparison of Y. pestis KIM6+ (A) and Y. pseudotuberculosis PTB52c hfq mutants (B). Cell membranes and DNA of WT, mutant (hfq), and complemented mutant (hfq-C) strains were visualized by fluorescence microscopy. Overnight bacterial cultures were diluted 1:100 into BHI broth without antibiotics and grown at either 28°C or 37°C with agitation. Heat-fixed smears were made with 10 μl of each culture at 8 h and 24 h postinoculation. Bacteria were stained by addition of 3 μl phosphate-buffered saline (PBS) containing 1 μg/ml FM1-43 (Molecular Probes) for membranes and 2 μg/ml DAPI (4′,6′-diamidino-2-phenylindole) (Sigma) for DNA, covered with a poly-l-lysine treated coverslip, and imaged by fluorescence microscopy as described previously (13).
Growth and morphology of the Y. pestis hfq mutants were largely restored by addition of a single-copy WT hfq expressed from its immediate upstream promoter (Fig. 1 and 2). However, we noted that complementation in Y. pestis was incomplete, and the slight growth defect in Y. pseudotuberculosis PTB52c was not affected by hfq addition. Similar difficulties in achieving complete hfq complementation have been reported previously for E. coli (42), but the basis of this defect is not known. We investigated whether complementation was limited by either inadequate levels of Hfq or lack of hfq coexpression with adjacent genes.
Transcription start sites (TSS) for hfq in Y. pestis and Y. pseudotuberculosis were determined by primer extension analyses. The TSS mapped to a position 90 nucleotides (nt) upstream of the translation start site, confirming that hfq is expressed from its immediate upstream promoter (Fig. 3 A and B). However, hfq and hflX mRNA levels in the complemented strain were lower than those in the WT (Fig. 3C). Further reverse transcription-PCR (RT-PCR) analyses showed that hfq was also transcribed as part of a larger overlapping operon with miaA and hflX (Fig. 3B and C), as it is in E. coli (42). We reasoned that expression from this additional promoter upstream of miaA could increase hfq levels in WT cells. A new hfq-complementing strain was generated using a multiple-copy pCRII-based plasmid (Invitrogen), and increased hfq levels were confirmed by Northern blot analysis (data not shown). This multicopy construct fully restored the growth defect of Y. pestis hfq mutants at both 28°C and 37°C (Fig. 3D and E), indicating that Hfq alone is sufficient for complementation, provided that it is expressed at adequate levels.
Genetic analysis of the hfq locus in Y. pestis and Y. pseudotuberculosis. (A) Primer extension was used to map the TSS of hfq in Y. pestis (lane 1) and Y. pseudotuberculosis (lane 2). Total RNA of Y. pestis and Y. pseudotuberculosis was prepared from cultures grown at 28°C in BHI broth. Primer KM2517 (5′GGATCTTGCAAAGATTGCCC) was end labeled with [γ-32P]ATP (MP Biochemicals) using T4 polynucleotide kinase, and the reactions were performed as described previously (44). The TSS is indicated with an asterisk. A putative −10 sequence (TACAAT) is indicated with a black bar. The DNA ladder is the hfq upstream sequence generated with the same primer. (B) Gene arrangement comparison of hfq in E. coli and Y. pestis. Numbers indicate the size of each intergenic region (in bp). Negative numbers indicate overlap of adjacent ORFs. (C) Semiquantitative RT-PCR using cDNA synthesized from WT, hfq mutant, and complemented Y. pestis strains. DNA contamination was ruled out, prior to all RT-PCR analyses, by PCR without reverse transcriptase using crr primers and 10 ng of RNA as the template (not shown). Note that hflX levels were reduced in the hfq deletion mutant and complemented strains. Regions spanning hfq to the mia ORF (Up), and hfq to the hflX ORF (Down), were amplified by RT-PCR. The miaA-hfq (Up) amplification product from the WT strain indicates that hfq is transcribed with miaA as part of a larger operon, in addition to its expression from the immediate upstream promoter shown by primer extension. PCR with crr primers was used as a control for normalization of cDNA levels for different strains. (D) Growth curves of the Y. pestis KIM6+ hfq mutant and complemented strains in BHI at 28°C. hfq-Csc, single-copy complementation; hfq-Cmc, multiple-copy complementation. (E) Growth curves of the Y. pestis KIM6+ hfq mutant and complemented strains in BHI broth at 37°C.
Generality of the plague bacillus' dependence on Hfq for growth.
We were struck by the biological differences between Y. pestis and Y. pseudotuberculosis with respect to their dependence on Hfq for growth at 37°C. Other investigators recently reported less dependence on Hfq for growth using an enzootic Pestoides (sometimes called Microtus [49]) Y. pestis strain (19). The previous study did not measure growth at late time points or on solid media, where we observed the most dramatic growth restriction. However, strain differences may also be relevant. The enzootic Pestoides strains, which are virulent only for small rodents, are thought to represent an evolutionary intermediate between Y. pseudotuberculosis and epidemic Y. pestis strains (14, 17). The phenotypic, biochemical, and genetic profiles of Pestoides strains combine features that are otherwise considered specific for either Y. pseudotuberculosis or Y. pestis lineages (2, 4). KIM, the strain that we used, belongs to the Mediavalis biovar, one of three Y. pestis biovars associated with epidemic plague and virulence for large mammals (including humans) as well as small rodents. We therefore extended our study to determine whether KIM's strong physiological dependence on Hfq also occurred in other epidemic plague biovars. hfq mutants were generated from representative strains from the remaining two epidemic biovars, CO92 R73 (Pgm+ Lcr− pFra+ pPst+) (biovar Orientalis) and Kuma D11 (Pgm+ Lcr− Pst+ Fra+ Gly+ pMT2+ pPCP2+) (biovar Antigua).
Growth of all Y. pestis mutants was severely restricted on BHI plates at 37°C (Fig. 4). As with the KIM mutants, the Kuma and CO92 mutants showed little to no growth on BHI plates at 37°C (Fig. 4C). The Y. pestis Kuma hfq mutant, from the oldest epidemic biovar, grew slowly even at 28°C (Fig. 4A), taking up to 24 h to reach WT stationary-phase ODs at 600 nm (OD600) (not shown). The Kuma hfq mutant's limited growth in liquid BHI medium at 37°C was similar to that of the KIM mutants (Fig. 4B), and the OD600 of the KIM and Kuma mutants remained at 0.2 to 0.3 for up to 24 h (data not shown). However, the Y. pestis CO92 hfq mutant showed a significant increase in OD600 after 5 h at 37°C and formed sporadic colonies on plates. Regrowth experiments with recovered bacteria (not shown) suggested that this increased OD600 was due to outgrowth of spontaneous suppressor mutants, consistent with the suppressor colonies that we observed on BHI plates (Fig. 4C).
Growth of hfq mutants generated from different Y. pestis biovars. (A and B) Growth curves of Y. pestis hfq mutant strains compared with their respective WT strains at 28°C (A) and 37°C (B). (C) Growth comparison of different strains on BHI plates for 48 h at 28°C or 37°C. Note that the growth of the CO92 hfq mutant at 37°C is actually due to sporadic suppressor colonies. (D) Relative suppressor frequencies for different Y. pestis mutants following overnight growth in BHI broth. Diluted cultures were spread on BHI plates in duplicate, followed by incubation at either 28°C or 37°C. The frequency of suppressor mutants was calculated as the ratio of CFU at 37°C to that at 28°C. All data shown are representative of two biological repeats.
We further explored this suppressor phenomenon by quantitating the relative frequencies of spontaneous suppressor mutations for each of the mutants after overnight growth at 28°C. Y. pestis CO92 hfq mutant suppressor clones were present at levels ∼100-fold (Kuma) to ∼1,000-fold (KIM) higher than with the other mutants (Fig. 4D). These results show that the essential nature of hfq for Y. pestis' multiplication at 37°C is a general phenomenon that occurs across all three epidemic plague biovars. However, cross-biovar differences, particularly with respect to suppressor frequencies, suggest that there is continued genetic plasticity associated with this phenotype.
Biological implications of the plague bacillus' dependence on Hfq for growth.
Y. pestis' dependence on Hfq for in vitro growth at 37°C was surprising to us, particularly compared with Hfq's relative lack of importance for Y. pseudotuberculosis growth. In this respect, Y. pseudotuberculosis more closely resembles other bacterial pathogens, in which significant phenotypic defects due to hfq deletion manifest only under host-associated stress conditions rather than standard in vitro laboratory growth (for a review, see reference 9). For example, a Salmonella hfq deletion strain is highly attenuated during mouse infection and defective for intracellular growth, but it had only a slight in vitro growth restriction at 37°C (37). Likewise, Vibrio cholerae and Brucella abortus hfq mutants are significantly compromised for growth in vivo but not in vitro (12, 33).
The Y. pestis hfq mutant's limited growth at 37°C suggests that Hfq is essential for plague infection of a mammalian host, which is consistent with recent studies in which hfq null Pestoides Y. pestis and Y. pseudotuberculosis showed decreased survival in macrophages and mice (19, 34). However, the different Hfq-dependent phenotypes observed in this study strongly suggest that sRNAs also play a unique role in Y. pestis biology and possibly in plague evolution. RNA regulators are associated with “evolvability” (46), and it is tempting to speculate that they have played a role in the divergence of Y. pestis from Y. pseudotuberculosis. Hfq interacts with sRNAs and mRNAs associated with both core genome and laterally transferred gene sequences, facilitating cross-regulation and incorporation of newly acquired genes into existing regulatory interactions (9, 37). For example, it has been suggested that Hfq stabilization of the SPI-1-encoded InvR sRNA facilitated the early establishment of this pathogenicity island in the Salmonella lineage by downregulating expression of the core genome-encoded OmpD porin (32, 36). Lateral gene transfer was also critical for Y. pestis evolution (16), and Hfq has been implicated in the regulation of genes on all three of its plasmids (19). The possible role of Hfq in the establishment of newly acquired genes in Y. pestis warrants further investigation.
The mechanism underlying Y. pestis' dependence on Hfq for growth at 37°C is not yet clear, although it is likely to involve core chromosomal genes. The similarity of the KIM6+ and KIM10+ results indicates that these hfq-associated phenotypes are not caused by the Y. pestis-specific pesticin plasmid. Likewise, the presence of the YP-HPI pathogenicity island in all Y. pestis strains and Y. pseudotuberculosis PTB52c, but not PTB54c (22), argues against YP-HPI being the differentiating factor. All strains also lacked the low-calcium-response (Lcr) plasmid (pCD in Y. pestis or pYV in Y. pseudotuberculosis), which encodes the type 3 secretion system (T3SS) and is required for the in vitro growth restriction that occurs with all pathogenic Yersinia strains when subjected to low calcium levels at 37°C (5, 26, 31). Therefore, the hfq-associated growth restriction that we observed is distinct from Lcr-based growth cessation.
Nonetheless, Y. pestis undergoes Lcr-associated growth restriction much more abruptly than Y. pseudotuberculosis, despite the presence of nearly identical Lcr plasmids (5, 7). This indicates that there are other important growth-associated physiological differences between these bacteria. This is consistent with functional inactivation of more than 10% of the Y. pestis genome relative to Y. pseudotuberculosis, including at least one mutation in aspA that further distinguishes epidemic from Pestoides Y. pestis strains (4, 8). We propose that Hfq is involved in some of the compensatory gene regulation associated with these gene inactivations.
Elucidation of Hfq's unusual role in Y. pestis growth will provide new insights into the unique biology of the plague bacillus and the physiological factors that distinguish this deadly pathogen from its enteric relatives and progenitors.
ACKNOWLEDGMENTS
We thank R. Losick and his lab for generously providing K.A.M. with the use of their imaging facility and many helpful discussions. We also thank A. Rakin for his kind gift of pKR29 and R. Perry for providing Yersinia pestis CO92 and Kuma WT strains. We are grateful to R. Lease and K. Spaeth for useful discussions, D. Schaak for technical assistance, H. Vasudeva-Rao and G. Knapp for critical reading of the manuscript. We acknowledge the Wadsworth Center Molecular Genetics Core facility for DNA sequencing and the David Axelrod Institute Imaging facility for use of microscopy resources.
This work was supported by grant AI06160602 from the National Institutes of Health.
FOOTNOTES
- Received 3 May 2010.
- Accepted 22 May 2010.
- Copyright © 2010 American Society for Microbiology