|Home | About | Journals | Submit | Contact Us | Français|
Yersinia pestis evolved from Y. pseudotuberculosis to become the causative agent of bubonic and pneumonic plague. We identified a homolog of the Salmonella enterica serovar Typhimurium lipoprotein (lpp) gene in Yersinia species and prepared lpp gene deletion mutants of Y. pseudotuberculosis YPIII, Y. pestis KIM/D27 (pigmentation locus minus), and Y. pestis CO92 with reduced virulence. Mice injected via the intraperitoneal route with 5 × 107 CFU of the Δlpp KIM/D27 mutant survived a month, even though this would have constituted a lethal dose for the parental KIM/D27 strain. Subsequently, these Δlpp KIM/D27-injected mice were solidly protected against an intranasally administered, highly virulent Y. pestis CO92 strain when it was given as five 50% lethal doses (LD50). In a parallel study with the pneumonic plague mouse model, after 72 h postinfection, the lungs of animals infected with wild-type (WT) Y. pestis CO92 and given a subinhibitory dose of levofloxacin had acute inflammation, edema, and masses of bacteria, while the lung tissue appeared essentially normal in mice inoculated with the Δlpp mutant of CO92 and given the same dose of levofloxacin. Importantly, while WT Y. pestis CO92 could be detected in the bloodstreams and spleens of infected mice at 72 h postinfection, the Δlpp mutant of CO92 could not be detected in those organs. Furthermore, the levels of cytokines/chemokines detected in the sera were significantly lower in animals infected with the Δlpp mutant than in those infected with WT CO92. Additionally, the Δlpp mutant was more rapidly killed by macrophages than was the WT CO92 strain. These data provided evidence that the Δlpp mutants of yersiniae were significantly attenuated and could be useful tools in the development of new vaccines.
Yersiniae are gram-negative bacilli in the family Enterobacteriaceae, and three species in the genus Yersinia are human pathogens. Yersinia pseudotuberculosis and Y. enterocolitica cause enteropathogenic infections that are typically self-limiting and are characterized by diarrhea, fever, and abdominal pain (52). Y. pestis is transmitted to humans through the bite of an infected flea or inhalation of the organisms, resulting in bubonic, pneumonic, or septicemic forms of plague. Y. pestis is historically credited for 200 million deaths worldwide and is also a potential agent of biowarfare or bioterrorism (18, 31, 34).
The pathogenesis of yersinia infections is complex and multifactorial. A number of virulence-associated systems (mechanisms) are common among the three human-pathogenic yersiniae. For example, the type three secretion system (T3SS), encoded on a 70-kb plasmid, exists in all three species, and this plasmid is designated pCD1 in Y. pestis, pYVe in Y. enterocolitica, and pYV in Y. pseudotuberculosis (30, 45). Through the T3SS, a variety of effector proteins called Yops (for Yersinia outer membrane proteins) are translocated into the host cytosol and are absolutely required for the virulence of the yersiniae (30). On the other hand, each Yersinia species possesses its own unique set of virulence factors (30, 45). In the two enteropathogenic species, the virulence determinants Yad (for Yersinia adherence protein, formerly designated YopA) and invasin protein (Inv) play crucial roles in bacterial colonization and invasion of the intestinal epithelium but are not functional in Y. pestis (45). In addition to plasmid pCD1, Y. pestis harbors another two plasmids that are absent in the enteropathogenic yersiniae. Plasmid pPCP1 encodes the virulence determinant plasminogen activator (Pla), while pMT1, or pFra, encodes capsular antigen F1 and the murine toxin (45).
It has been shown that the deletion of the 102-kb pigmentation (pgm) locus from the chromosome of Y. pestis attenuates its virulence, while this pgm locus is usually silent in Y. pseudotuberculosis (30). Recently, a transcriptional regulator, RovA, was identified in all three Yersinia species. RovA regulates a variety of genes and is specifically required for the development of bubonic plague in mice (8). Interestingly, a new set of T3SS-dependent proteins has been identified in Y. pestis and includes insecticidal-like proteins that are thought to function as transmission factors, contributing to flea morbidity by promoting colonization of the midgut (26). Therefore, the continued investigation of new virulence factors and virulence-associated systems is warranted to fully elucidate the mechanisms of yersinia pathogenesis.
The outer membrane of gram-negative bacteria is comprised of many different proteins that help maintain the structural integrity of the bacterial cell envelope. Some of these proteins are covalently modified by the addition of a lipid, N-acyl-S-diacylglyceryl cysteine, and such modified proteins are collectively known as lipoproteins (37). One particularly abundant lipoprotein, designated murein (or Braun) lipoprotein (Lpp), is associated with the outer membranes of bacteria within the family Enterobacteriaceae. Originally identified in Escherichia coli, Lpp links the murein (peptidoglycan) layer to the outer bacterial membrane (29). Lpp has previously been demonstrated to play a role in the host's immune response against infections with some gram-negative enteric pathogens, such as E. coli, Salmonella enterica serovar Typhimurium, and Y. enterocolitica (65). Our earlier studies indicated that Lpp (6.3 kDa) purified from E. coli and Y. enterocolitica not only synergized with lipopolysaccharide (LPS) to induce septic shock, but also evoked the production of tumor necrosis factor alpha (TNF-α) and interleukin 6 (IL-6) in both LPS-responsive and LPS-nonresponsive mice and in mouse peritoneal exudate macrophages, suggesting an alternative signaling mechanism for Lpp (66). In fact, a subsequent study showed that Lpp signals through Toll-like receptor 2 (TLR-2) and not TLR-4 (2), which LPS utilizes for cell signaling.
In our most recent studies, we provided evidence that S. enterica serovar Typhimurium harbors two copies of the lipoprotein gene, lppA and lppB, separated by 82 bp on the chromosome (22). Both single and double lpp gene deletions resulted in attenuation of bacteria in terms of their motility, invasiveness, and cytokine/chemokine production, and such mutants were avirulent in mice when administered by systemic and oral routes compared to the wild-type (WT) bacterium (21, 22, 58).
By searching the National Center for Biotechnology Information database, we identified a homolog of the lpp gene in the genomes of all three pathogenic yersiniae (10, 15, 44). However unlike in S. enterica serovar Typhimurium, only a single copy of the lpp gene is present on the chromosome in Yersinia species. Consequently, in this study, we generated lpp mutants of WT/parental yersiniae (Y. pseudotuberculosis YPIII and Y. pestis KIM/D27 [pgm locus minus] and Y. pestis CO92) and evaluated them in mouse models of infection. Blood chemistries and blood cell counts, as well as organ pathology, were examined in specimens from the pneumonic and bubonic plague mouse models after infection with WT and Δlpp mutant strains of Y. pestis. We also evaluated the abilities of WT and mutant yersiniae to disseminate to the distal organs after infection. Additionally, the intracellular survival of WT and mutant Y. pestis CO92 bacteria was assessed in murine macrophages. Our data provided evidence that lpp mutants of Y. pseudotuberculosis and Y. pestis were attenuated in mice, an effect that could be complemented. More importantly, deletion of the lpp gene from the Y. pestis KIM/D27 mutant further attenuated the strain, and immunization of mice with this mutant provided protection to animals against pneumonic plague invoked by intranasal (i.n.) inoculation of Y. pestis CO92. To our knowledge, this is the first systematic study illustrating the role of Lpp in the pathogenesis of yersinia infection.
The bacterial strains and plasmids (8, 16, 19, 27, 57) used in this study are listed in Table Table1.1. Yersiniae were grown in brain heart infusion (BHI) or heart infusion broth (HIB) medium (Difco, Voigt Global Distribution Inc., Lawrence, KS) at 26 to 28°C with constant shaking (180 rpm) and/or on BHI/HIB agar plates, while Luria-Bertani medium was used for growing other bacterial cultures (e.g., recombinant E. coli) at 37°C with shaking. Restriction endonucleases and T4 DNA ligase were obtained from Promega (Madison, WI). The Advantage cDNA PCR Kit was purchased from Clontech (Palo Alto, CA). The digested plasmid DNA or DNA fragments from agarose gels were purified using QIAquick kits (Qiagen, Inc., Valencia, CA).
Based on the genome sequences of Y. pestis (KIM and CO92) and Y. pseudotuberculosis, the lpp gene coding region (237 bp) and its upstream flanking DNA sequences were identical in the two genomes (GenBank accession numbers NC_003143, NC_004088, and NC_006155) (10, 29). However, the downstream flanking DNA sequences of the lpp gene were somewhat different in the two Y. pestis strains (KIM and CO92), as well as in Y. pseudotuberculosis. By using different primer sets (Table (Table2),2), the lpp-flanking DNA fragments from each of the corresponding yersiniae were PCR amplified and cloned into the pDMS197 suicide vector (19), resulting in the recombinant plasmids pDMS197KIMUD, pDMS197CO92UD, and pDMS197pseudoUD, respectively (Table (Table1).1). By electroporation (Genepulser Xcell; Bio-Rad, Hercules, CA), the recombinant plasmids were transformed into their corresponding WT/parental strains for homologous recombination. The transformants were plated onto BHI agar plates containing 5% sucrose and screened by colony blot hybridization (49) with the lpp gene probe, which was PCR amplified with the specific primers lppF/lppR (Table (Table2)2) from the genomic DNA of Y. pestis CO92. The colonies that did not react with the lpp gene probe were potential lpp mutants, and their identities were confirmed by Southern blot analysis using the lpp gene probe and the suicide vector pDMS197 plasmid as a probe (58). The resulting mutants were designated ΔlppD27, ΔlppCO92, and Δlpppseudo (Table (Table11).
We used specific primers (lpp-N/PvuI and lpp-C/PstI) (Table (Table2)2) to amplify a DNA fragment that contained the coding region of the lpp gene (237 bp), as well as its potential promoter region (200 bp upstream of the start codon of the lpp gene), from the genomic DNA of Y. pseudotuberculosis. The amplified DNA fragment was cloned into the pBR322 vector, generating the recombinant plasmid pBRYlpp (Table (Table1).1). By electroporation, the plasmid pBRYlpp was transformed into the lpp mutants of Y. pestis KIM/D27 and Y. pseudotuberculosis for complementation. The pBR322 vector alone was also introduced into the parental strains and into the lpp mutants to serve as proper controls. These yersiniae were grown in BHI or HIB medium in the presence of tetracycline (15 μg/ml).
WT/parental Y. pestis KIM/D27 and CO92, as well as WT Y. pseudotuberculosis YPIII, their corresponding Δlpp mutants, and complemented strains, were grown in 3 ml of BHI medium at 28°C overnight with shaking. The bacteria were harvested, and Western blot analysis was performed with specific primary anti-Lpp monoclonal antibodies (1:2,000 dilution), followed by horseradish peroxidase-labeled goat anti-mouse secondary antibodies. The blots were developed using a SuperSignal West Pico chemiluminescent substrate (12, 58). All of the sample preparations were conducted in our approved, restricted-entry biosafety level 2 laboratory.
To detect the secreted form of LcrV (low-calcium response antigen), the above-mentioned overnight-grown yersinia cultures were diluted (in a total volume of 10 ml of the medium) to an optical density at 600 nm (OD600) of 0.2, and EGTA was added to a final concentration of 5 mM. After 4 to 6 h of growth at 37°C, the culture supernatants (10 ml each) were collected, filter sterilized (0.22 μm), and precipitated with trichloroacetic acid (TCA) (10% final concentration). The precipitated proteins were resuspended in 100 μl of sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer, and an aliquot (25 μl) from each of these samples was used for Western blot analysis with specific anti-LcrV antibodies (39).
The WT/parental yersiniae and their Δlpp mutant cultures were grown at 28°C overnight. Approximately 10 to 15 ml of each culture was used to isolate the plasmid DNA (pCD1, pYV, pMT1, and pPCP1) using the Qiagen plasmid isolation kit with slight modifications. Briefly, after the addition of neutralizing solution P3, the supernatants were separated from the pellet and extracted with phenol. The plasmid DNA was then directly precipitated with an equal volume of isopropyl alcohol and resolved using 0.7% agarose gel electrophoresis with Tris-acetate buffer (38).
A modified protease protection and digitonin extraction assay (41) was used to evaluate the translocation abilities of T3SS effectors in Yersinia Δlpp mutants. Briefly, duplicate wells of HeLa cells (in a six-well plate) were infected with different Yersinia strains (WT/parental and their Δlpp mutant strains) at a multiplicity of infection (MOI) of 10. After 2 to 3 h of infection at 37°C, the cells were treated with 1 ml of proteinase K (100 μg/ml in phosphate-buffered saline [PBS]), followed by the addition of 500 μl of freshly prepared phenylmethylsulfonyl fluoride (4 mM in PBS) to block protease activity. One set of the cells was lysed with 200 μl of digitonin (1% in PBS), and the other set was incubated with 200 μl of PBS without digitonin. The cells were dislodged from the surface of the plate, collected, and centrifuged. The supernatants were collected and subjected to Western blot analyses using antibodies to YopE (Santa Cruz Biotechnology, Santa Cruz, CA) and YopH (Agrisera, Stockholm, Sweden).
Overnight-grown WT Y. pseudotuberculosis, the Δlpp mutant, and its complemented strain were adjusted to similar ODs, and equal numbers of bacteria (1 × 106 CFU) were stabbed into 0.35% BHI-agar plates (58). The migration of organisms from the centers to the peripheries of the plates was measured after incubation for 48 h at 28°C.
Different methods were used to measure the membrane integrity of yersinia Δlpp mutant strains, as described below.
WT/parental yersinia strains and their corresponding Δlpp mutants were grown overnight and then diluted to an OD600 of 0.2. An aliquot (10 μl) of each diluted culture was spotted in triplicate onto BHI agar plates containing 1.5% Torula yeast RNA (Sigma-Aldrich, St. Louis, MO). The plates were incubated overnight at 26 to 28°C, after which 10% TCA was poured into the plates. As RNase I degraded the yeast RNA substrate on the plate, a clear zone around the bacterial growth was formed after application of TCA. Therefore, the size of the clear zone represented the amount of RNase I released from the bacterial periplasmic space and was indicative of cell membrane integrity (5).
The plasmid pBR322, a source of β-lactamase, was transformed into WT/parental Y. pestis KIM/D27, Y. pseudotuberculosis YPIII, and their corresponding Δlpp mutant strains. The membrane integrity of the Δlpp mutants was assessed by measuring the percentage of β-lactamase activity in the culture supernatant/cell extract and comparing it to the corresponding parental strains, as we previously described (58).
Triton X-100 (TX-100) was used in the detergent sensitivity assay as previously described (58). Briefly, WT/parental yersiniae and their Δlpp mutant strains were grown to an OD600 of 0.4 to 0.5 and diluted 50-fold. TX-100 was then added at various final concentrations (0, 1, 2, and 5%), and the cultures were incubated at 26 to 28°C for 3 h with shaking, after which the OD600 was measured. TX-100-treated cultures were compared to corresponding cultures without TX-100 in three independent experiments. A 50% reduction in the OD indicated sensitivity of the bacterium to detergent treatment and a breach in membrane integrity (7).
The WT/parental yersiniae and their Δlpp mutant strains were grown at 28°C or 37°C overnight. The cultures were then pelleted and resuspended in BHI medium with different pH (3.0 or 5.0) and TX-100 (0.5% or 2.0%) combinations for 5 min at room temperature (3). The normal BHI medium (pH 7.0 without the addition of TX-100) served as a control. The bacterial samples were utilized in triplicate in each treatment, and the surviving bacteria were counted by serial dilution and plating. The bacterial survival rates were calculated as the percentages of viable bacteria from the control groups.
Cultures of WT/parental yersiniae and their Δlpp mutant strains were grown overnight at 28°C. A 1:4 dilution was made of the overnight cultures in fresh broth, and incubation was continued at 28°C for 30 min and then at 37°C for 1 h. HeLa cells were infected with yersiniae at various MOIs (0.5, 1, 10, and 20) for 2 to 3 h at 37°C. The cell morphology (cytotoxicity) was monitored and recorded by using light microscopy.
All animal experiments were performed using protocols approved by the University of Texas Medical Branch Institutional Animal Care and Use and Institutional Biological Safety Committees in animal biosafety level 2/3 laboratories. Female 5- to 6-week-old Swiss-Webster or BALB/c mice were purchased from Charles River Laboratories (Wilmington, MA). The mice were challenged intraperitoneally (i.p.) (22), i.n. (46, 47), or subcutaneously (s.c.) (47) with various doses of different species and strains of WT/parental Yersinia and their Δlpp mutant and complemented strains. In some experiments, mice received a subinhibitory dose of levofloxacin (0.3 to 0.5 mg/kg of body weight/day) (unpublished data) starting 24 h postinfection (p.i.) by the i.p. route for a maximum period of 6 days. Levofloxacin is a homolog of ciprofloxacin with a longer half-life and has recently been approved by the Food and Drug Administration for use against anthrax (23). The mice were assessed for morbidity or mortality over the duration of each experiment or were sacrificed, and the sera and organs were harvested at the indicated time points. In animal experiments in which complemented strains were used, mice were given tetracycline in water (5 mg/ml) 48 h prior to challenge and the antibiotic treatment was continued throughout the experiment to maintain the plasmid in bacteria.
The MICs of levofloxacin against WT Y. pestis CO92 and its Δlpp mutant were determined by using the Etest (AB Biodisk North America Inc., Culver City, CA) (42). Briefly, the overnight-grown Yersinia cultures were diluted (1:4) with fresh BHI medium and continued to grow at 28°C for 2 h (OD600, 0.6). The bacterial cultures were then spread evenly onto Mueller-Hinton agar plates (Becton Dickinson, Cockeysville, MD) or on 5% sheep blood agar (SBA) plates (Teknova, Hollister, CA), and the predefined levofloxacin (range, 0.002 to 32 μg/ml) E-strips were placed onto the plates. The plates were incubated for 48 h at 28°C, and the MICs were recorded.
Swiss-Webster mice were inoculated via the i.n. route with five 50% lethal doses (LD50) (1.7 × 103 CFU) of WT Y. pestis CO92. After 24, 48, and 96 h p.i., blood was drawn via cardiac puncture, and three to five animals were used for each time point. The blood chemistries and cell counts were analyzed by using a Drew Scientific ProChem blood analyzer and a Drew Scientific Hemavet 950 hematology system (Drew Scientific, Inc., Dallas, TX), respectively. The results were compared to both the mouse baseline clinical chemistry values published by Charles River and the 0-h control. Data for all animals at each time point were averaged, and a standard deviation was calculated.
The lungs, livers, spleens, and heart tissues from Y. pestis-infected mice were collected at necropsy at various time points and were immersion fixed in 10% neutral buffered formalin in screw-cap containers. After 3 days, the tissues were moved to new containers with fresh formalin, processed, sectioned at 5 μm, mounted on glass slides, and stained with hematoxylin and eosin (HE). The sterility of the fixed tissues was confirmed by plate counts before histopathological examination. Lesions considered distinctive for Y. pestis infection included (i) the presence of massive numbers of organisms; (ii) inflammatory changes, such as edema and cellular infiltrates; and (iii) tissue and/or blood vessel necrosis, with congestion and hemorrhage. Lesions were graded based on a severity scale of minimal, mild, moderate, and severe. The scale correlated with estimates of lesion distribution and extent of tissue involvement (minimal, 2 to 10%; mild, >10 to 20%; moderate, >20 to 50%; severe, >50%).
In general, acute inflammation was characterized by the presence of neutrophils, while subacute inflammation indicated a mixture of neutrophils and mononuclear cells (macrophages and lymphocytes). Edema and/or necrosis was present in some acute and subacute lesions, particularly if moderate to severe lesions were observed. Bacteria seen in tissues were consistent with Y. pestis based on their morphology, although special staining was not performed, except in some cases where bacteria were stained with Giemsa. We analyzed tissues from three animals per group for histopathology, and data from representative tissues are presented.
Swiss-Webster mice were inoculated via the i.n. route with 5 × 104 or 8 × 104 CFU of either the WT or Δlpp mutant of Y. pestis CO92. Some of the mice received a subinhibitory dose of levofloxacin after 24 h p.i. At 1, 24, 48, and 72 h p.i., three mice from each group at each time point were euthanized using a mixture of ketamine (90 mg/kg) and xylazine (10 mg/kg). Blood was collected via cardiac puncture, and the lungs, liver, and spleen were removed immediately after death of the mice was ensured by cervical dislocation. The organs were weighed and homogenized, using tissue grinders (Kendall, Mansfield, MA), in 1 ml of PBS containing 25 μg/ml of polymyxin B to allow specific growth of Y. pestis CO92. The blood and tissue homogenates were serially diluted and cultured on SBA plates. After incubation of the plates at 28°C for 48 to 72 h, bacterial colonies were counted, and the CFU per gram of tissue or per ml of blood were calculated.
Swiss-Webster mice were inoculated via the i.n. route with 5 LD50 of either WT Y. pestis CO92 or its Δlpp mutant strain, and a group of uninfected mice served as controls. Starting at 24 h p.i., a subinhibitory dose of levofloxacin (0.3 to 0.5 mg/kg/day) was administered, and at 48, 72, and 96 h p.i., three to five mice from each time point and group were euthanized and blood was collected. The sera were filtered through 0.22-μm syringe filters, and the levels of cytokines and chemokines were measured using a multiplex assay with BioPlex (Bio-Rad), which detects 23 cytokines/chemokines. The blood samples were cultured on SBA plates to confirm their sterility.
RAW 264.7 murine macrophages were seeded in six-well plates and incubated until they reached 60 to 70% confluence (approximately 3 × 106 cells/well) (21, 22, 58). Different yersiniae, grown in HIB medium at either 28°C or 37°C, were used to infect macrophages at an MOI of 1. The plates were centrifuged at 1,500 rpm for 10 min to facilitate binding of the bacteria to the host cells and were then incubated at 37°C in 5% CO2 for the duration of the experiment. After 30 min of incubation, the medium was aspirated and the cells were carefully washed twice with PBS. Dulbecco's modified Eagle's medium supplemented with 200 μg/ml gentamicin was added to each well to kill extracellular bacteria. After 1 h, the medium was removed, and an aliquot (100 μl) from each well was spread onto SBA plates to determine whether all of the extracellular bacteria were killed by the gentamicin treatment. Subsequently, the infected RAW cells were carefully washed twice with PBS before Dulbecco's modified Eagle's medium containing 10 μg/ml of gentamicin was added to each well. The 0-h plates, representing host cells that were infected with bacteria for 30 min and then treated with gentamicin for 1 h, were harvested at this time to determine the number of initially invaded/phagocytosed bacteria. At the designated time points (between 0 and 24 h), host cell viability was assessed using a light microscope, after which the monolayers were carefully washed twice with PBS. The macrophages were then lysed with 300 μl of sterile water and shaken to ensure complete disruption of the host cells. Suspensions of lysed macrophages were serially diluted 10-fold and cultured on SBA plates for determining colony counts after incubation at 28°C for 24 to 48 h. The original bacterial suspension was also serially diluted and plated in order to calculate the percentage of invasion/phagocytosis.
Whenever appropriate, Student's t test, analysis of variance with Bonferroni correction, and a Kaplan-Meier survival estimate (for animal studies) were used for data analyses. For animal mortality studies, we used 10 mice per group. Three independent experiments were performed for statistical analysis of the data.
Primary pneumonic plague is highly contagious and almost always fatal (35, 45). To better understand the progression of the disease and the host responses, we used an established mouse model in which animals were challenged i.n. with WT Y. pestis CO92 (35). The LD50 of WT Y. pestis varies, depending on bacterial growth conditions (such as temperature) and routes of infection (45). In our study, WT Y. pestis CO92 was grown at 28°C in BHI or HIB medium. The i.n. LD50 of BHI-grown cells (53) was determined to be 340 bacteria for Swiss-Webster mice and approximately 100 CFU in BALB/c mice. Similar LD50 in mice were reported in the literature for this highly virulent strain of Y. pestis by the i.n. route (35).
To study systemic effects in mice during the course of Y. pestis infection, animals were inoculated via the i.n. route with 5 LD50 of WT Y. pestis CO92. Blood chemistry analysis of sera from infected mice showed minimal changes during the first 48 h following challenge; however, marked changes were observed late in the infection at 60 h (Table (Table3).3). The albumin, globulin, and total protein values gradually increased through the course of infection. A similar increasing trend was seen with electrolytes, such as calcium, chloride, potassium, and magnesium, although in some cases the values remained within the mouse normal range but above the average values (e.g., potassium). The bicarbonate values, on the other hand, decreased through the course of infection (Table (Table3).3). The triglyceride, creatinine (a breakdown product of creatine), alanine transaminase (ALT), and aspartate aminotransferase (AST) levels also increased (Table (Table3).3). These changes indicated advancing kidney and liver damage, edema, and dehydration of mice due to WT Y. pestis infection. For example, the albumin level increased from 3.1 to 5.3 g/dl (the average normal value is 3.3) from 0 h to 60 h p.i. Likewise, AST increased from 133 to 669 (U/liter) and creatinine from 0.4 to 4.8, while their normal values were 137 and 0.6, respectively.
The hematology results (Table (Table4)4) showed an increase in white (WBC) and red blood cells (RBC) and in hemoglobin (Hb) and hematocrit (Hct) levels. Likewise, platelets increased from 276 to 687, 702, and 417 at 0, 36, 48, and 60 h p.i., respectively, while the mean platelet volume (MPV) decreased from 15.5 to 5.3 to 6.7 over the period of infection. These changes in blood cell counts began at 36 h p.i., with maximum changes occurring at 60 h p.i. Further, alterations in blood chemistries and blood cell counts were also correlated with the development of behavioral changes in the infected mice; the animals became listless, hunched up, and huddled together at 48 to 60 h p.i. before eventually dying by 72 to 96 h p.i.
Uninfected control mice had no lesions in the lungs, liver, or spleen (Fig. 1A, F, and K, respectively). However, at 12 h p.i. with 5 LD50 of WT Y. pestis CO92 administered by the i.n. route, the bronchial epithelia of mice revealed mild vacuolation with minute amounts of polymorphonuclear leukocytes (PMNs) seen in peribronchial areas, focally, while the liver and spleen did not show any histopathological changes (data not shown). At 24 h p.i., the lungs revealed a mild infiltration of alveolar septa by PMNs and capillary dilation (data not shown). No histopathological changes were noticed in the liver, while the lymphoid follicles in the spleen showed mildly increased apoptotic bodies in the germinal centers. At 36 h p.i., the lungs revealed more prominent peribronchial infiltrates composed of PMNs, accompanied by patchy edema in the alveolar spaces (data not shown). The alveolar septa were also more congested and more heavily infiltrated by PMNs, involving more distal airways, than at earlier time points. The liver revealed occasional PMNs and lymphocytes in the sinusoids, while the spleen changes were essentially the same as at 24 h p.i.
At 48 h p.i. with WT Y. pestis CO92, the lungs (Fig. 1B and C) revealed diffuse vascular congestion in alveolar septa, with more PMNs in the interstitium, and patchy areas of bronchopneumonia with microabscess formation and necrosis. The liver (Fig. 1G and H) revealed the early formation of microabscesses, located randomly in the parenchyma, with focal hepatocyte death. The spleen (Fig. (Fig.1L)1L) showed prominent apoptotic bodies in germinal centers and a more distinct marginal layer in the lymphoid follicles. The red pulp showed mild increases in PMNs. At 60 h p.i., the lungs (Fig. 1D and E) revealed extensive pulmonary edema, more extensive bronchopneumonia, and visualization of bacterial clumps in the areas of necrosis. Bronchi also showed luminal secretions with PMNs. The liver (Fig. 1I and J) showed well-formed microabscesses with necrosis of hepatic parenchyma. The changes in the spleen (Fig. (Fig.1M)1M) were more prominent than at 48 h.
Compared with the two functional copies of the lpp gene in the genome of S. enterica serovar Typhimurium (58), Lpp of Y. pestis exhibited 89% and 91% identity and homology at the DNA and amino acid levels, respectively, with LppA of S. enterica serovar Typhimurium, while the corresponding identity and homology were 85% and 84% with LppB of S. enterica serovar Typhimurium (Fig. (Fig.2).2). To study the role of Lpp in the yersinia infections, we deleted the lpp gene from the Y. pestis KIM/D27 and CO92 strains, as well as from Y. pseudotuberculosis YPIII, thus generating ΔlppD27, ΔlppCO92, and Δlpppseudo mutants (Table (Table1).1). Western blot analysis with anti-Lpp specific antibodies showed that the expression of the lpp gene was correspondingly abrogated in the Δlpp mutants but restored in their complemented strains (Fig. (Fig.3)3) (58).
To further characterize Yersinia Δlpp mutants, we performed plasmid isolation and PCR analyses with specific primers for the lcrV, pla, and caf1 genes carried on the plasmids (Table (Table2).2). There is always a concern that mutagenesis procedures could result in the loss of plasmids or that a portion of the plasmid could be integrated into the chromosome (45). As shown in Fig. Fig.4,4, the plasmids harbored in the Δlpp mutant strains were intact and exhibited sizes similar to those seen in their corresponding WT/parental strains (45). As can be seen from lanes 3 and 4, Y. pseudotuberculosis harbored only the pYV plasmid, while Y. pestis KIM/D27 (lanes 1 and 2) and Y. pestis CO92 (lanes 5 and 6) contained two additional plasmids, representing pMT1 and pPCP1. Likewise, we could successfully PCR amplify lcrV (981 bp), pla (939 bp), and caf1 (513 bp) genes from the total DNA of Y. pestis WT/parental and mutant strains, as well as the lcrV gene from the WT and the Δlpp mutant of Y. pseudotuberculosis (data not shown).
We detected similar amounts of the secreted form of LcrV in the Δlpp mutants and in the corresponding WT/parental strains of yersiniae by Western blot analysis (data not shown). However, to determine that the T3SS was indeed intact in the Δlpp mutants of yersiniae, we assessed the abilities of YopE and YopH effectors to be translocated into HeLa cells. As shown in Fig. Fig.5,5, the Δlpp mutants (lanes 3, 7, and 11) delivered amounts of Yops (i.e., YopE and YopH) into the cytosol of infected HeLa cells similar to those of their corresponding WT/parental strains (lanes 1, 5, and 9). As can also be seen from the figure, minimal levels of YopE and YopH were detected in HeLa cells infected with the WT/parental strains (lanes 2, 6, and 10) or the Δlpp mutants (lanes 4, 8, and 12) of yersiniae that were not treated with digitonin. These data indicated that we indeed measured the translocated effectors, as digitonin lysed the host cells without affecting bacterial-cell integrity. More importantly, the HeLa cells infected with Yersinia Δlpp mutant strains exhibited similar T3SS-associated cytotoxicity, as noted for the HeLa cells infected with the WT/parental Yersinia strains (Fig. (Fig.6).6). We examined different MOIs and times of incubation for evaluating cytotoxicity with essentially similar data for the WT/parental versus the mutant strains. These data further confirmed the proper translocation of the T3SS effectors in the Δlpp mutants of yersiniae.
We further assessed the integrity of the bacterial membrane in the Δlpp mutants versus WT/parental yersiniae by measuring the release of periplasmic RNase I and β-lactamase, as well as by testing their sensitivities to the detergent TX-100. No obvious differences were noted between WT/parental strains and their corresponding Δlpp mutant strains in any of these assays (data not shown). We then analyzed these Δlpp mutants under low-pH conditions with or without 0.5 to 2% TX-100 to mimic the harsh environment in the phagosomes and observed similar survival rates in the Δlpp mutants and their corresponding WT/parental strains (data not shown). These data indicated that the membrane integrity in these Yersinia Δlpp mutants was not affected.
Unlike the nonmotile bacterium Y. pestis (45), the motility of the Y. pseudotuberculosis Δlpp mutant was reduced to 59% (8.5 ± 0.7 mm) of that of the corresponding WT bacterium (14.5 ± 2.1 mm), and this reduction in the motility phenotype could be fully complemented (15.5 ± 0.7 mm). As we reported previously, the lppAB mutant of S. enterica serovar Typhimurium had no motility (58), thus correlating with our observation in Y. pseudotuberculosis. However, we observed earlier that deletion of the lpp genes from S. enterica serovar Typhimurium significantly reduced bacterial invasion in the human colonic epithelial cell line T-84 (21, 22, 58), but the invasive capacity of the Y. pseudotuberculosis Δlpp mutant remained unaltered compared to that of the WT bacterium in both T-84 and HeLa cells (data not shown). These data suggested some diversity in the functions of Lpp in these two related pathogens.
To assess the overall virulence of the Δlpp mutant strain, mice were infected via the i.p. route with the WT, the Δlpp mutant, and/or the Δlpp-complemented strain of Y. pseudotuberculosis using various doses. As shown in Fig. Fig.7,7, all mice inoculated with the highest challenge dose (5 × 106 CFU) of the WT strain and its Δlpp mutant strain died from infection, but a somewhat longer time to death was observed in the Δlpp mutant-inoculated group. At lower challenge doses of 1 × 106 and 1 × 105 CFU, consistent and statistically significant higher survival rates (30 to 40%) were observed in the Δlpp-infected group than in the respective WT-infected mice. Further, at a dose of 1 × 105 CFU, the mice infected with the complemented-strain group had a mortality rate (30%) similar to that of the WT-infected group (40%), while no deaths were observed in the Δlpp mutant-infected mice at this dose, indicating a specific role of Lpp in the virulence of Y. pseudotuberculosis.
Although Y. pestis has evolved from Y. pseudotuberculosis (1), the diseases caused by the two bacteria differ tremendously (34). Consequently, we first evaluated the virulence potentials of the parental Y. pestis KIM/D27 strain (57), its Δlpp mutant, and/or the Δlpp-complemented strain in the mouse model of infection. As shown in Fig. Fig.8A,8A, Swiss-Webster mice infected by the i.p. route with doses of 5 × 107 and 1 × 108 CFU of the parental KIM/D27 bacterium died within 5 to 6 days, while 60% and 100% of the animals inoculated with 1 × 108 CFU and 5 × 107 CFU of the Δlpp mutant strain, respectively, survived the initial 30 days of observation. Importantly, the complemented Δlpp mutant strain of KIM/D27 was as virulent as the WT bacterium in the mouse model of lethality.
A similar pattern was noticed when BALB/c mice were used for infection (Fig. (Fig.8B).8B). In this experiment, 100% of the mice from the parental-strain-infected groups at doses of 5 × 106 and 1 × 106 CFU died, while an 80 to 100% survival rate was seen in the corresponding Δlpp mutant-infected groups (Fig. (Fig.8B).8B). In the lowest-dose group (5 × 105 CFU), 60% of the mice survived in the parental-strain-infected group and 100% in the Δlpp mutant-infected group (Fig. (Fig.8B).8B). Twenty-four to 30 days following the initial inoculation, we rechallenged the survivors (Swiss-Webster [Fig. [Fig.8A]8A] and BALB/c [Fig. [Fig.8B]8B] mice) from both the WT and Δlpp mutant groups with 5 LD50 of the highly virulent WT Y. pestis CO92 strain via the i.n. route. As can be seen in Fig. 8A and B, 70 to 100% of the mice survived rechallenge, indicating that immunity was established in the mice following the initial infection. A group of mice that were given PBS served as an appropriate age-matched control, and these mice were challenged with 5 LD50 of Y. pestis CO92 via the i.n. route at the same time immunized mice (with the parental strain or the Δlpp mutant KIM/D27) were subjected to challenge with Y. pestis CO92 (Fig. 8A and B). These control mice showed the usual death curve: 90 to 100% mortality within 3 to 4 days.
To confirm the role of Lpp in the virulence of Yersinia, a Δlpp mutant of the highly virulent Y. pestis CO92 strain was generated. Mice were challenged i.n. with various doses of the WT or the Δlpp mutant to mimic pneumonic plague. We could not demonstrate any differences in virulence between the WT and the Δlpp mutant of Y. pestis CO92 in terms of the death rate when tested at doses of 5 × 102 and 5 × 103 CFU, which represented close to 1 and 10 LD50 (Fig. (Fig.9).9). These results were not entirely surprising, as the Y. pestis CO92 strain is highly virulent and produces many other virulence factors. In an effort to increase the sensitivity of our animal assay for evaluation of virulence factors, we dosed some of the mice with a subinhibitory dose of levofloxacin and repeated the challenge study with 1 and 10 LD50 of WT Y. pestis CO92 and its Δlpp mutant. The rationale for this experiment was to demonstrate whether Lpp contributes to bacterial virulence in a pneumonic plague model.
As shown in Fig. Fig.9,9, the mice infected with Δlpp mutant bacteria and receiving a subinhibitory dose of levofloxacin for 6 days beginning 24 h p.i. had significantly higher survival rates—100% and 80%—at the challenge doses of 5 × 102 and 5 × 103 CFU, respectively. In contrast, animals infected with the WT Y. pestis CO92 experienced a higher rate of mortality, as 60% and only 10% of the mice infected with 5 × 102 and 5 × 103 CFU, respectively, survived infection (Fig. (Fig.9)9) when given the same regimen of antibiotic treatment.
As bubonic plague is most commonly encountered in humans (54), we assessed the role of Lpp in the development of this form of the disease. Mice were challenged s.c. with either WT Y. pestis CO92 or its Δlpp mutant. Based on the literature and our own studies, the LD50 of Y. pestis CO92 via the s.c. route was between 2.8 and 9.2 bacteria (45). Therefore, two doses (5 and 20 CFU) were used in this experiment. As shown in Fig. Fig.10,10, 30 to 50% of the mice inoculated with the Δlpp mutant of Y. pestis CO92 survived, while only 10% of the mice from the WT-infected group survived 30 days following infection. As also noted in this study, the mean time to death increased in the Δlpp mutant- versus WT-infected mice. In this set of experiments, no antibiotic treatment was given to the animals. Taken together, these data indicated that Lpp plays a role in the development of both pneumonic and bubonic plague; however, the dose of bacteria or their virulence is critical in discerning the effect of Lpp in overall bacterial-virulence attenuation.
We compared the differences in histopathology of tissues from the WT Y. pestis-infected animals with those infected with the Δlpp mutant at 48 and 72 h p.i. (5 LD50) in the presence of a subinhibitory dose of levofloxacin (Fig. (Fig.11).11). Uninfected control mice had no significant lesions in the lungs either in the absence (Fig. 11A) or in the presence (Fig. 11B) of levofloxacin. At 48 h p.i., all of the WT-infected mice had lesions characteristic of Y. pestis infection. We noted mild to moderate edema and acute inflammation, with masses of bacteria in the lungs of the mice (Fig. 11C, inset). The livers of the WT-infected mice showed signs of minimal inflammation and necrosis, while the spleens exhibited mild inflammation in the red pulp and lymphoid cell necrosis in the white pulp. No heart lesions were present at 48 h p.i. (data not shown).
Although at 48 h p.i. with the Δlpp mutant strain of Y. pestis CO92, the animals showed signs of mild to moderate inflammation in the lungs (Fig. 11D), there was neither edema nor the presence of masses of bacteria in most of the sections examined, and hence, these bacteria (which were more dispersed) could not be easily visualized in histopathology pictures, as could be seen in the lungs of WT bacterium-infected mice (Fig. 11C). In the Δlpp mutant-infected mice, we observed minimal liver necrosis, while the red pulp in the spleen was mildly inflamed, as was also noted in the WT bacterium-infected mice (data not shown). Although lymphoid cell necrosis was noted in the spleens’ white pulp in WT bacteria-infected mice, only mild lymphoid depletion was observed in the white pulp of the Δlpp mutant-infected animals. There were no significant lesions in the hearts of the Δlpp mutant-infected animals (data not shown).
At 72 h p.i., the mice infected with WT Y. pestis CO92 had severe inflammation in the lung and lining surface (pleura), with edema and masses of bacteria (Fig. 11E, inset), as well as extensive bronchopneumonia. The liver was acutely inflamed, with clumps of bacteria and necrosis. The red pulp of the spleen was moderately inflamed and had bacteria, and there was lymphoid depletion in the white pulp. The heart had low numbers of intravascular bacteria and minimal neutrophilic infiltrates.
Importantly, tissue damage was minimal in mice at 72 h p.i. with the Δlpp mutant of Y. pestis CO92. The lungs (Fig. 11F and inset) had essentially normal architecture, with no edema and no masses of bacteria (although more dispersed bacteria could be seen under high-power magnification, they were not easily visualized in histopathology pictures) or inflammation in most of the sections examined, while the spleen was mildly inflamed. There were no significant lesions consistent with Y. pestis infection in the liver or the heart.
Since mice succumbed to infection by day 4 or 5 after challenge with Y. pestis CO92 by the s.c. route (Fig. (Fig.10),10), we collected tissues at 120 h p.i. from uninfected control mice and animals infected with 5 LD50 of either WT Y. pestis CO92 or its Δlpp mutant (Fig. (Fig.12).12). The control mice had no significant lesions in any tissue, although data from the spleen are shown in the figure (Fig. 12A). Minimal to mild inflammation with neutrophilic infiltrates was observed in the lungs of the mice infected with WT Y. pestis CO92. The liver was mildly inflamed, and bacteria were present. The spleen showed lymphoid depletion (Fig. 12B, inset), and in concert with bacteria (Fig. 12C), the red pulp was acutely inflamed and lymphoids were moderately depleted in the white pulp (Fig. 12B and C). These data indicated that systemic infection also damaged the lungs, as well as the liver.
In mice infected with the Δlpp mutant strain of Y. pestis CO92, the lungs had minimal to mild inflammation with neutrophilic infiltrates. The liver was minimally inflamed, and no bacteria were present. The red pulp of the spleen was mildly inflamed, and the white pulp had moderate lymphoid depletion, without the presence of bacteria (Fig. 12D and inset).
Previous studies from our laboratory indicated that 90 to 100% of the mice inoculated with 1 × 107 CFU of the parental strain or the Δlpp mutant of Y. pestis KIM/D27 survived for the duration of the experiment, which was 21 days (data not shown). Therefore, the tissue sections were subjected to histopathology 7 days p.i.
At a magnification of ×100, the livers of animals infected with the parental Y. pestis KIM/D27 strain showed parenchyma with widespread, poorly formed granulomas composed of activated macrophages, scattered lymphocytes, and occasional neutrophils. Sinusoids also revealed increased numbers of inflammatory cells. Under high-power magnification (×400), liver granulomas showed macrophages with prominent eosinophilic cytoplasm and nuclei with vesicular chromatin. Likewise, the red pulp in the spleens displayed lymphoid hyperplasia and extensive necrosis (×100) with karyorrhectic nuclei (×400) (data not shown).
On the other hand, animals infected with the Δlpp mutant of KIM/D27 showed occasional poorly formed granulomas in the liver (×100 to ×400). The spleens of Δlpp mutant-infected mice exhibited parenchyma with prominent hematopoiesis, lymphoid hyperplasia, and some infiltration of red pulp by macrophages, but there was a complete absence of necrosis (×100 to ×400) (data not shown). Taken together, these data indicated that the Δlpp mutant of KIM/D27 induced less tissue damage than its corresponding WT bacterium.
The rationale for these studies was to assess the role of proinflammatory mediators in host tissue damage after infection with either WT Y. pestis CO92 or its Δlpp mutant by the i.n. route at 5 LD50. The blood was drawn at various time points (24, 48, 72, and 96 h) p.i. and analyzed for cytokine and chemokine production. When we compared at 24 h p.i. the findings in uninfected controls (given levofloxacin) to those in infected mice from either group (WT or the Δlpp mutant), we saw no significant changes in the levels of cytokines or chemokines (data not shown). At 48 h p.i., elevated levels of TNF-α, IL-1β, IL-1α, monocyte chemoattractant protein 1 (MCP-1), keratinocyte-derived chemokine (KC), IL-6, and granulocyte colony-stimulating factor (G-CSF) were detected in the WT-infected mice, and these continued to rise until the 96-h time point, except in the case of TNF-α, which peaked within the first 48 h and then declined (Table (Table5).5). The macrophage inflammatory protein (MIP)-1β levels remained unchanged until 96 h p.i. (Table (Table55).
In the Δlpp mutant-infected group, cytokine and chemokine levels considerably lower than those in the corresponding WT-infected group were detected. This trend was most obvious at 96 h p.i. (Table (Table5).5). However, levels of KC, G-CSF, and IL-6 slightly higher (although not statistically significant) than those in WT-infected animals were observed in the Δlpp mutant-infected group of mice 48 h p.i. These levels dropped to lower than those in the WT-infected group at later time points (72 and 96 h) (Table (Table5).5). Compared to uninfected control mice, only MCP-1 and G-CSF in the Δlpp mutant-infected group maintained higher levels during the course of infection, while the other tested cytokines and chemokines either increased only at the early time point (48 h p.i.) (IL-1β, TNF-α, KC, and IL-6) or remained unchanged (MIP-1β) up to 96 h p.i. (Table (Table5).5). These data typified the host response to the more virulent WT Y. pestis CO92 strain and generally correlated with host inflammatory responses, resulting in better animal survival following infection with the mutant strain. The other 15 cytokines tested (e.g., IL-10, RANTES, IL-12 p70, and granulocyte-macrophage colony-stimulating factor) were not significantly altered in WT- versus Δlpp mutant-infected mice.
We also measured cytokine/chemokine responses, using enzyme-linked immunosorbent assays, in the sera of mice infected via the i.p. route with the parental strain of Y. pestis KIM/D27 and its Δlpp mutant (1 × 107 CFU) on days 1 and 3 p.i. Our data indicated that the levels of IL-6, KC (equivalent to IL-8 in humans), and MCP-1 were much reduced in the sera of animals infected with the Δlpp mutant compared to those infected with the parental bacterium. Likewise, murine RAW 264.7 macrophages infected with the parental strain of Y. pestis KIM/D27 at an MOI of 10 for 6 h produced much higher levels of TNF-α, IL-1β, IL-6, IL-10, MCP-1, granulocyte-macrophage colony-stimulating factor, and G-CSF than host cells infected with the Δlpp mutant, as measured by Bio-Rad's BioPlex assay (data not shown).
To further investigate Y. pestis infection in the context of pneumonic plague, we monitored the growth of both WT Y. pestis CO92 and its Δlpp mutant in the lungs of infected mice and evaluated the ability of the bacteria to disseminate to the bloodstream and the distal organs (e.g., spleen and liver).
As shown in Table Table6,6, Yersinia bacilli (both the WT and the Δlpp mutant) were mainly localized at the site of infection (lung tissue) at early time points (1 to 24 h p.i.), and the number of bacteria recovered from the lungs (at 1 h p.i.; reported as CFU/g of tissue) was similar to that of the bacteria initially used to dose the animals (5 × 104 to 8 × 104 CFU). These numbers decreased by 1 log unit 24 h p.i. in WT-infected mice but remained stable in Δlpp mutant-infected mice. At 48 h p.i., no significant bacterial multiplication was noted in the lungs in both the WT- and Δlpp mutant-infected mice; however, the bacilli (both WT and Δlpp mutant) started to spread to the livers, but not to the spleens. Interestingly, bacilli were detected in the bloodstreams of mice infected only with WT Y. pestis CO92, but not with its Δlpp mutant strain (Table (Table6).6). At 72 h p.i., the number of bacteria observed in lungs increased by 1 log unit in both the WT- and Δlpp mutant-infected groups compared to the levels seen at early time points (1 to 24 h p.i.) (Table (Table6).6). All of the livers from infected mice (both the WT and Δlpp mutant groups) were infiltrated with a substantial number of bacteria. In the WT-infected group, bacilli were also detected in spleens and the bloodstream. However, there were no detectable bacteria in the blood or spleens of Δlpp mutant-infected mice (Table (Table6).6). These data thus indicated an impaired ability of the Δlpp mutant for broad dissemination in the mice.
We also noted that the overall bacterial replication and dissemination were very similar between the levofloxacin-treated groups of mice and those left untreated with levofloxacin. Moreover, the Δlpp mutant of Y. pestis CO92 multiplied in a manner similar to that of its WT bacterium in mice that were given the subinhibitory dose of levofloxacin (Table (Table6).6). These data suggested similar antibiotic susceptibilities of both the WT and its Δlpp mutant of Y. pestis CO92 in vivo and agreed with our in vitro levofloxacin susceptibility data, as evaluated by using the Etest, in which WT Y. pestis CO92 and its Δlpp mutant exhibited the same MIC of 0.032 μg/ml.
In the initial experiment, the bacteria were grown at 28°C prior to infection of macrophages. We observed occurrence in macrophages at nearly a 100% rate of invasion/phagocytosis of both WT bacteria and the Δlpp mutant (11). From 0 to 4 h p.i., the number of WT bacteria within the macrophages increased slightly from 3.0 × 106 to 3.6 × 106 CFU (Fig. (Fig.13).13). On the other hand, the number of Δlpp mutant bacteria within the macrophages decreased from 3.5 × 106 to 1.8 × 106 CFU/ml (P < 0.05) at this time point. At 8 h p.i., the host cells were still viable as assessed by light microscopy, and there was no obvious change in the numbers of viable bacteria (WT or the lpp mutant) measured in the macrophages. The difference between the numbers of WT (3.1 × 106 CFU/ml) and Δlpp mutant (1.7 × 106 CFU/ml) bacteria measured, however, was statistically significant at this time point (P < 0.05).
By 24 h p.i., the host cells still seemed healthy and adhered to the wells of the tissue culture plate. However, the number of viable Δlpp mutant bacteria decreased by a log unit to 2.8 × 105 CFU/ml between 8 and 24 h p.i., in comparison to the numbers of WT bacteria, which were 2.2 × 106 CFU/ml (P < 0.01) and decreased only marginally between the 8- and 24-h time points. These data indicated that although both the WT and mutant bacteria were phagocytosed equally, the intracellular viability of the bacteria was significantly affected, with the Δlpp mutant dying off rapidly compared to its WT counterpart. Importantly, neither the WT nor its Δlpp mutant increased in number under the tested conditions in macrophages. An essentially similar trend was seen when the bacteria were grown overnight at 37°C (data not shown) instead of at 28°C before the infection of macrophages. Because the production of the antiphagocytic capsule (17) and Yops was induced at 37°C, the percentage of phagocytosis by macrophages decreased to 20% for both the WT and its Δlpp mutant. Finally, we also showed that the intracellular survival of the Y. pestis KIM/D27 Δlpp mutant was reduced by 50% compared to the WT bacterium in macrophages 4 h p.i. Importantly, we could fully complement the survival defect in macrophages when the corresponding gene was supplied to the Δlpp mutant in trans (data not shown). For this experiment, various bacterial strains were grown at 28°C and the infection of macrophages occurred at 37°C.
Currently, the pathogenic mechanisms underlying the development of plague are understudied, and the host's response to the infection is largely unexplored. We have provided here, for the first time, detailed analyses of the blood chemistries and blood cell counts during the course of the development of pneumonic plague in mice infected with Y. pestis strain CO92. Increases in WBC, RBC, Hb, and Hct indicated overwhelming bacterial infection, the loss of fluid following organ failure, and a breach in blood vessel integrity. Furthermore, alterations in platelets and MPV suggested abnormal blood coagulation that would eventually lead to disseminated intravascular coagulation and extensive hemorrhage, which are hallmarks of severe bacterial infection (54). Changes in tissue histopathology were also observed in multiple organs of mice infected via the i.n. route, and the lungs were the first and most severely affected tissue. Respiratory failure, as a result of severe congestion and edema in the lungs, was believed to be the main cause of death in mice with pneumonic plague. However, multiorgan failure would occur if the animals survived long enough, as bacteria could be seen in the liver, spleen, and bloodstream.
It has been reported that mice with pneumonic plague develop a striking biphasic syndrome in which the infection begins with an anti-inflammatory state in the first 24 to 36 h, which rapidly progresses to a highly proinflammatory state by 48 h and death by 3 days (8, 35). Our observations are in agreement with these earlier findings. The asymptomatic state of infection in the first 24 to 36 h was thought likely to be due to the potent anti-inflammatory activity of the pCD1-based T3SS (35, 40). However, in our studies and in those of others, IL-10 (an anti-inflammatory cytokine) levels were not elevated in Y. pestis-infected mice (35), and the most recent study has shown that Y. pestis LcrV does not efficiently activate TLR-2 signaling. Therefore, TLR-mediated immunomodulation is unlikely to play a significant role in plague (50).
In searching for new virulence factors and virulence-associated mechanisms in yersiniae, we focused on a 6.3-kDa Lpp (21, 22, 58). We noted that the membrane integrity and T3SS in the yersinia Δlpp mutants were not affected, and virulence plasmids in the yersinia Δlpp mutants were normally maintained. Compared to WT Y. pseudotuberculosis, we observed that its Δlpp mutant had reduced motility and was significantly less virulent in a mouse model of infection. We obtained similar results with the S. enterica serovar Typhimurium Δlpp mutants (58). It has been reported that the nonmotile Salmonella strains (i.e., the flagellar mutants) were less virulent in mice (6), while the pathogenicity of Y. enterocolitica was not affected in a flagellar mutant (ΔfliA) (32). Therefore, the exact mechanism(s) by which the Δlpp mutant of Y. pseudotuberculosis was attenuated needs to be further explored.
In Y. pestis, deletion of the lpp gene from the KIM/D27 background strain further attenuated its virulence, but surprisingly, no obvious difference in pathogenicity was observed between the WT and Δlpp mutant strains of CO92 in a pneumonic plague mouse model. Interestingly, when groups of mice infected with either WT CO92 or its Δlpp mutant were given a subinhibitory dose of levofloxacin, we observed a significantly higher survival rate, less severe histopathological changes, and reduced cytokine and chemokine levels in the Δlpp mutant-infected group compared to these parameters in the WT-infected mice. The above-mentioned differences were not due to variation in the susceptibilities of the WT and the Δlpp mutant to the antibiotic, as we observed similar bacterial loads in the lungs and liver at 48 and 72 h p.i. in both groups. Furthermore, our in vitro susceptibility assessment also revealed no detectable differences in the effect of levofloxacin between the WT and its Δlpp mutant.
The mechanism of attenuation elicited by the Δlpp mutant under levofloxacin treatment in Y. pestis CO92 is not clear. However, levofloxacin is a fluoroquinolone antibiotic that inhibits bacterial DNA replication and transcription. Therefore, with the subinhibitory dose of the antibiotic, we might have reduced the number of bacteria to a critical threshold level that allowed us to better discern the contribution of Lpp to bacterial virulence. It has also been reported that a subinhibitory dose of fluoroquinolones inhibits the adherence of uropathogenic E. coli strains in vitro (4). Therefore, it is possible that Y. pestis, under treatment with levofloxacin, might have a similar postantibiotic effect in that the virulence of both the WT and its Δlpp mutant was reduced.
Importantly, in the relatively nonsevere bubonic plague mouse model, we clearly demonstrated attenuation of the Δlpp mutant of Y. pestis CO92. In fact, the histopathological data indicated that the mutant bacteria were rapidly cleared from the system, resulting in less severe pathology in the tissue and better survival of the animals. In contrast, the WT bacteria persisted in the tissues longer, resulting in severe pathology and in animals succumbing to infection.
Interestingly, in the pneumonic plague model, we observed that WT Y. pestis CO92 appeared in the bloodstreams of infected mice 48 h p.i. and in the spleens and blood 72 h p.i. However, the Δlpp mutant could not be detected in the spleens or the bloodstreams, indicating that the ability of the Δlpp mutant to disseminate was impaired. In addition, although the number of mutant bacteria increased in the lungs 72 h p.i., the tissues appeared normal, indicating an important role of Lpp in inducing histopathological changes. In corroboration of these data, we also demonstrated reduced cytokine/chemokine responses in the sera of mice infected with the Δlpp mutants of both Y. pestis CO92 and KIM/D27 compared to that in the WT/parental-strain-infected mice.
Additionally, we found that the liver was the most susceptible organ for the colonization of Y. pestis by the i.n. route, as bacteria were detected in the liver between 24 and 48 h p.i. in both WT- and Δlpp mutant-infected groups of mice. The susceptibility of the liver may be due to its relatively large physical size and anatomic position; however, the existence of a receptor(s) on liver cells that preferentially recruits the bacilli cannot be ruled out. Unlike the liver, the colonization of Y. pestis in the spleen happened only at later stages of infection (72 h p.i.) in the WT-infected group of mice. This differs from a recent report stating that Yersinia bacilli were detected in the spleen as early as 36 h p.i. (35), but this divergence may be due to the amounts of the challenge dose, the use of a different strain of mouse (outbred Swiss-Webster versus inbred C57BL/6), or differences in growth conditions during the preparation of the inoculum (e.g., 26°C versus 37°C) (34).
Much-reduced tissue pathology, decreases in cytokine and chemokine levels, and limited dissemination of the Δlpp mutant, compared to WT Y. pestis CO92, led us to investigate the abilities of these mutants to survive in vitro in the intracellular environment of murine macrophages. Although it is clear that most yersinia replication occurs extracellularly, there is evidence suggesting that all three pathogenic Yersinia species survive and multiply in macrophages (9, 13, 51, 52). The ability of yersiniae to proliferate in macrophages not only provides a replicative niche for the bacteria while they are protected from neutrophils, but also allows Yersinia to avoid antigen presentation and therefore delays the development of a specific immune response. Finally, macrophages may provide the bacterium with a vehicle for safe transport from the initial site of infection to deeper lymph tissues (52).
Based on our data, significantly fewer Δlpp mutant bacteria were present within macrophages 24 h p.i. compared to the number of WT bacteria. This correlated well with our bacterial-dissemination data, in which the Δlpp mutant displayed an impaired ability to disseminate to the bloodstreams and spleens of infected mice. Therefore, the mechanism of attenuation of the Y. pestis Δlpp mutant may be due to its inability to survive within the hostile macrophage environment. Data presented in our study indicated that the Δlpp mutant had no obvious defect in T3SS function but that such mutants exhibited a defect in intracellular survival in macrophages. These results correlated with a previous study that indicated that the T3SS played no significant role in this niche (51).
Finally, the primary difference between Y. pestis KIM/D27 and CO92 strains is the presence of an additional iron transport system (the yersiniabactin system) in CO92. Since the Δlpp mutant of KIM/D27 showed a significant defect in virulence compared to its parental strain, while the virulence of the CO92 Δlpp mutant via the i.n. route was essentially identical to that of the WT CO92 strain, it seems possible that the Δlpp mutant may affect iron transport in some manner, and this possibility will be explored further.
We did not observe an increased number of viable bacteria during the course of infection, even in the WT-infected macrophages, unlike in other studies (11, 33, 59, 60). This discrepancy may be due to the fact that yersiniae, when growing intracellularly, form tight aggregates that are not easily dispersed, resulting in an underestimation of bacterial numbers (51, 52). The exact mechanism by which Lpp affects the survival of Y. pestis in macrophages is not clear; however, earlier studies indicated that PhoP, a response regulator of the two-component PhoP/PhoQ sensory transduction system, is involved in regulating one or more genes important for the intracellular survival of Y. pestis in macrophages (20, 28, 43). Therefore, it is possible that the expression of the lpp gene in Y. pestis is under the control of PhoP or vice versa. Our future studies will focus on investigating the mechanism of preferential intracellular killing of the Δlpp mutant compared to the WT bacteria in macrophages.
Although Y. pestis and Y. pseudotuberculosis are genetically related bacilli, the diseases they cause are different. We showed that deletion of the lpp gene from either Y. pseudotuberculosis YPIII or the Y. pestis strains attenuated their virulence. However, Y. pseudotuberculosis YPIII is a phoP mutant strain and hence is unable to grow in macrophages (27), unlike Y. pestis. Therefore, Lpp might attenuate these species of yersiniae by different mechanism(s), and this possibility needs further study.
In comparison to plasmid-encoded virulence determinants (i.e., the T3SS), the contribution of the chromosomally encoded Lpp to the virulence of yersiniae is subtle; however, it provides an interesting insight into the complex mechanism of Yersinia pathogenesis. To our knowledge, this is the first systematic study illustrating the role of Lpp in the pathogenesis of yersinia infections. Currently, there is no vaccine licensed in the United States against plague. Although recent studies have indicated that the subunit vaccines containing Y. pestis antigens (e.g., LcrV and F1) have immunogenic properties in mice and macaques (14, 63, 64) and appear very promising, the effectiveness of these subunit vaccines against human pneumonic plague transmission has not yet been determined (48).
Until 1999, a formaldehyde-killed, whole-cell vaccine, prepared using a highly virulent Y. pestis strain, 195/P, was used for military personnel, but its production was halted because of severe side effects and its short-lived protection in humans (55). In addition, the above-mentioned vaccine provided protection against bubonic plague but not against pneumonic plague. Although the use of a live attenuated plague vaccine is not approved in the United States and many other countries because of safety reasons (61), a pgm locus-minus mutant of Y. pestis strain EV76 has been used in the countries of the former Soviet Union (62). This vaccine was found to be effective against both bubonic and pneumonic plague (24, 56). However, most of the EV76 derivatives can cause severe local and systemic reactions and gross tissue changes independent of their route of inoculation, thus limiting their use worldwide (62).
In an effort to further attenuate Y. pestis EV76, the lpxM gene (encoding lipid A myristoyl transferase) was deleted, and the generated ΔlpxM mutant synthesized a penta-acylated LPS less toxic than the hexa-acylated LPS in the parental strain (24). More importantly, the overall virulence potential of the ΔlpxM mutant of Y. pestis EV76 was reduced, and the mutant strain even provided a better level of immunity in immunized hosts (24). Lpp has a synergistic effect with LPS in bacterial virulence, and we have shown that deletion of the lpp gene from Y. pestis KIM/D27 further attenuates this strain while retaining its immunogenicity. Therefore, our current study not only provides an additional target for yersinia attenuation, but also indicates that deletion of both lpp and the biosynthetic genes for LPS might further attenuate the virulence of yersiniae. Such mutants could be excellent candidates for possible live-vaccine development. Indeed, our recent study indicated that deletion of the lpp and msbB (a homolog of lpxM) genes from S. enterica serovar Typhimurium resulted in the generation of cell-mediated and humoral immune responses in mice superior to those in animals infected with the WT bacterium (36).
Since live-vaccine strains are potentially released into the environment by vaccinated individuals, safety issues concerning the medical, as well as environmental, aspects must be considered (25). These concerns include a reversion to virulence by acquisition of complementation genes and exchange of genetic information with other vaccine or WT strains of the carrier organism (25). These safety issues must be thoroughly evaluated, especially the safety of such vaccines in immunocompromised hosts (25).
This research was supported by NIH/NIAID grants AI064389 and N01-AI-30065. S. L. Agar is a predoctoral fellow and was funded by an NIH T32 training grant (AI07526) in Emerging and Tropical Infectious Diseases.
We thank Mardelle Susman for carefully editing the manuscript.
Editor: S. R. Blanke
Published ahead of print on 28 January 2008.