Deletion of Braun lipoprotein gene (lpp) and curing of plasmid pPCP1 dramatically alter the virulence of Yersinia pestis CO92 in a mouse model of pneumonic plague

Abstract

Deletion of the murein (Braun) lipoprotein gene, lpp, attenuates the Yersinia pestis CO92 strain in mouse models of bubonic and pneumonic plague. In this report, we characterized the virulence of strains from which the plasminogen activating protease (pla)-encoding pPCP1 plasmid was cured from either the wild-type (WT) or the Δlpp mutant strain of Y. pestis CO92 in the mouse model of pneumonic infection. We noted a significantly increased survival rate in mice infected with the Y. pestis pPCP⁻/Δlpp mutant strain up to a dose of 5000 LD₅₀. Additionally, mice challenged with the pPCP⁻/Δlpp strain had substantially less tissue injury and a strong decrease in the levels of most cytokines and chemokines in tissue homogenates and sera when compared with the WT-infected group. Importantly, the Y. pestis pPCP⁻/Δlpp mutant strain was detectable in high numbers in the livers and spleens of some of the infected mice. In the lungs of pPCP⁻/Δlpp mutant-challenged animals, however, bacterial numbers dropped at 48 h after infection when compared with tissue homogenates from 1 h post-infection. Similarly, we noted that this mutant was unable to survive within murine macrophages in an in vitro assay, whereas survivability of the pPCP⁻ mutant within the macrophage environment was similar to that of the WT. Taken together, our data indicated that a significant and possibly synergistic attenuation in bacterial virulence occurred in a mouse model of pneumonic plague when both the lpp gene and the virulence plasmid pPCP1 encoding the pla gene were deleted from Y. pestis.

INTRODUCTION

Yersinia pestis is one of three species of yersiniae that cause serious infections in humans, and the disease it manifests, plague, has historically killed populations to disastrous proportions (Perry & Fetherston, 1997). All three human pathogenic species of Yersinia (Y. pestis, Yersinia pseudotuberculosis and Yersinia enterocolitica) share the presence of a 70 kb virulence plasmid, pCD1, that harbours genes encoding the type 3 secretion system (T3SS) and its effectors, referred to as Yersinia outer-membrane proteins (Yops). Yops are delivered into the cytoplasm following contact of the bacterium with a host cell and are required to cause disease in animals and humans (Perry & Fetherston, 1997). Y. pestis has two additional plasmids encoding virulence factors for the organism. Plasmid pMT1 has genes encoding the antiphagocytic capsular antigen F1 (Du et al., 2002) and murine toxin, the latter of which is important in establishing Y. pestis colonization in the midguts of fleas (Hinnebusch et al., 1998). In all, at least 115 genes are present on the 110 kb pMT1 plasmid in the KIM5 strain of Y. pestis (Lindler et al., 1998).

The smallest of the three virulence plasmids, pPCP1, is 9.5 kb in size and contains three genes encoding proteins with distinctly different functions. Pesticin, encoded by the pst gene, causes cell wall lysis and the death of neighbouring bacteria by hydrolysing the bond between N-acetylglucosamine and N-acetylmuramic acid, thus degrading the glycan backbone of murein (Braun) lipoprotein (Ferber & Brubaker, 1979). It has been suggested that Y. pestis uses this bacteriocin to kill neighbouring cells in which the pPCP1 plasmid is absent, while at the same time protecting itself with pesticin immunity protein, encoded by the pim gene on the pPCP1 plasmid (Sodeinde et al., 1992).

Also present on the pPCP1 plasmid is plasminogen activating protease (Pla), a surface protein consisting of two subunits, α- and β-, with molecular sizes of 37 and 35 kDa, respectively (Kutyrev et al., 1999; Sodeinde & Goguen, 1988). Pla is a member of the omptin family of aspartate proteases expressed on the surface of many members of the family Enterobacteriaceae (McDonough & Falkow, 1989; Sodeinde & Goguen, 1989). The function of Pla has been well characterized, and its role in the invasiveness and dissemination of Y. pestis extensively studied in vitro and in vivo (Beesley et al., 1967; Brubaker et al., 1965; Lathem et al., 2007; Sebbane et al., 2006). Pla subverts the host innate immune response by degrading complement proteins, thus reducing the ability of phagocytes to opsonize and phagocytize Y. pestis, and by preventing the chemotaxis of other inflammatory cells to the site of infection (Sodeinde et al., 1992). Pla also proteolytically activates plasminogen and anchors it to the bacterial surface (Beesley et al., 1967). Human plasminogen activators convert plasminogen into plasmin, which Y. pestis uses to gain access to deeper tissue layers and to colonize and disseminate through the host (Sodeinde et al., 1992).

Since Pla plays an important role in human infections and avoidance of the immune system, significant research has been performed to better understand the role of both the pla gene and the pPCP1 plasmid in the pathogenesis of Y. pestis infections. For example, the loss of the pPCP1 plasmid or the pla gene attenuates Y. pestis administered by the subcutaneous route (Brubaker et al., 1965; Samoilova et al., 1996; Sodeinde et al., 1992; Welkos et al., 2002, 1997) in mice or guinea pigs, and by the intradermal (Sebbane et al., 2006) and intranasal (i.n.) (Lathem et al., 2007) routes in mice, although the bacterial dissemination that occurred in intranasally infected mice was not noted in subcutaneously infected animals. No change in the LD₅₀ was observed in mice inoculated with the wild-type (WT) versus the pla-minus mutant by either the intraperitoneal (Welkos et al., 2002, 1997) or the intravenous (Sodeinde et al., 1992) route of infection. Reports on the virulence of strains lacking this plasmid and delivered via the aerosol route of infection are not consistent and are possibly dependent upon the strains of Y. pestis used in the studies (Samoilova et al., 1996; Welkos et al., 2002, 1997). Other studies have indicated that pla mutant strains evoke an earlier inflammatory response than the WT, and animals are able to clear the infection sooner, resulting in a higher survival rate in mutant versus WT bacteria-infected mice (Lathem et al., 2007; Sodeinde et al., 1992; Welkos et al., 1997).

We recently published a detailed characterization of how the absence of Braun lipoprotein (Lpp) attenuates Y. pestis CO92 in the i.n. mouse model of infection and how bacterial dissemination, host tissue damage and cytokine/chemokine production are decreased in mice infected with the Δlpp mutant of Y. pestis compared with those infected with the WT bacterium (Sha et al., 2008). In the light of the other studies describing the attenuation of pPCP⁻ and Δpla mutants of Y. pestis (Brubaker et al., 1965; Samoilova et al., 1996; Sodeinde et al., 1992; Welkos et al., 2002, 1997) and recent research characterizing the importance of Pla in establishing pneumonic infection in mice (Lathem et al., 2007), we undertook studies to delineate how the absence of the pPCP1 plasmid would attenuate our Δlpp mutant of Y. pestis CO92. We noted a significant increase in the LD₅₀ in mice that were infected with the double mutant pPCP⁻/Δlpp of Y. pestis, and a dramatic decrease in cytokine/chemokine release compared with animals that were infected with either the pPCP⁻ or the Δlpp mutant strain of Y. pestis alone. Our data suggested that Pla and Lpp synergistically affect the survivability and differentially affect the disseminative capacity of Y. pestis in a mouse model of pneumonic plague.

METHODS

Bacterial strains.

Virulent WT Y. pestis CO92 strain was acquired from the Centers for Disease Control and Prevention (CDC, Atlanta, GA, USA). The Y. pestis CO92 strain deficient in the lpp gene was created by us previously and is described elsewhere (Sha et al., 2008). Strains lacking the virulence plasmid pPCP1 (pPCP⁻) were created by growing either WT or Δlpp Y. pestis at 4 °C on trypticase soy agar plates with 5 % sheep blood (SBA, Becton, Dickinson and Company) for five passages (Straley & Brubaker, 1982). Single colonies were tested via PCR for the presence of genes on each of the three virulence plasmids: lcrV (981 bp, low calcium response antigen present on pCD1); pla (939 bp, plasminogen activating protease present on pPCP1); and caf1 (513 bp, capsular antigen present on pMT1). Colonies cured of the pla-encoding plasmid only were those in which PCR indicated that pla was absent, while verifying the presence of lcrV and caf1 genes. The presence or absence of plasmids in the cured strains was also ascertained by agarose gel electrophoresis (Sha et al., 2008). Y. pestis strains were grown in heart infusion broth (HIB) medium (Difco, Voigt Global Distribution) at 28 °C with constant shaking at 180 r.p.m. in our CDC-approved, restricted-entry, biosafety level (BSL)-2 laboratory.

Complementation of the Δlpp mutant strain of Y. pestis CO92.

The Tn7 transposon-based broad-range vector pUC18R6K-mini-Tn7T was used in this study (Choi et al., 2005). The transposon Tn7 can insert at a high frequency into bacterial chromosomes in a site- and orientation-specific manner at the Tn7 attachment (attTn7) site. This site is located downstream of the highly conserved glmS gene, which encodes an essential glucosamine-6-phosphate synthetase (Craig, 1996; Peters & Craig, 2001). This system provides a site-specific insertion of a single copy of the target gene into the genome without interrupting any other genes, and is an ideal system for complementation studies. Briefly, primers lpp5 (5′ ACGAATTCGCGCACCTTACAAATAAGAC 3′, EcoRI site underlined) and lpp3 (5′ AGCTGCAGTTACTTCTTGTAAGCTTGAG 3′, PstI site underlined) were used to PCR-amplify the lpp coding and promoter (200 bp upstream of the start codon) regions from the WT Y. pestis genome. The amplified DNA fragment was cloned into the pUC18R6K-mini-Tn7T vector at the EcoRI/PstI sites, which generated the recombinant plasmid pUC18R6K-mini-Tn7T/lpp. Subsequently, a kanamycin-resistance (Km^r) gene cassette [with Flippase (FLP) recombinase recognition target site at both ends], which was PCR-amplified from plasmid pKD13 by using specific primers Km5 (5′ ATTCCGGGGATCCGTCGACC 3′, BamHI site underlined) and Km3 (5′ TTGGATCCGTGTAGGCTGGAGCTGCTTC 3′, BamHI site underlined), was inserted into the plasmid pUC18R6K-mini-Tn7T/lpp at the BamHI site, thus generating the recombinant plasmid pUC18R6K-mini-Tn7T-Km/lpp.

This newly generated recombinant plasmid and the helper plasmid pTNS2 (Choi et al., 2005), which expresses the transposase complex to facilitate efficient transposition, were mixed at a ratio of 1 : 1 and transformed into the Δlpp Y. pestis mutant strain via electroporation by using a Gene Pulse Xcell (Bio-Rad). The transformants were selected on the kanamycin plate, and the antibiotic-resistant colonies were picked up. As the helper plasmid pTNS2 also contains the R6K origin, it is not able to replicate inside Y. pestis and was lost during bacterial propagation. The site-specific Tn7 insertion in these Y. pestis colonies was subjected to further PCR analyses with specific primer sets as described by Choi et al. (2005). The Y. pestis colony that contained the insertion of the Km^r and lpp gene with its promoter downstream of the glmS gene in the genome due to Tn7 site-specific integration was considered to be the intermediate form of the Δlpp Y. pestis complemented strain. Subsequently, the Km^r gene cassette was removed by transforming this intermediate form strain with plasmid pFlp2, which expresses the FLP recombinase. Plasmid pFlp2 was then cured with 5 % sucrose in the growth medium, and the final form of the Δlpp Y. pestis complemented strain was sensitive to kanamycin and free of plasmids pTNS2 and pFlp2. The expression of the lpp gene in the complemented strain was further confirmed by Western blot analysis using antibodies to Lpp (Sha et al., 2008).

Animals.

All of our mice studies were performed in the animal BSL (ABSL)-3 facility under the protocol approved by the UTMB Institutional Animal Care and Use Committee. Five- to six-week-old female Swiss-Webster mice were purchased from Charles River Laboratories and challenged via the i.n. route with various doses of WT Y. pestis or the Δlpp, pPCP⁻ and pPCP⁻/Δlpp mutant strains, as previously described (Sha et al., 2008). To generate the survival curves, mice were assessed twice daily for morbidity or mortality over a period of 40 days. For the histopathology, cytokine/chemokine and bacterial dissemination analyses, the sera and organs of animals were harvested at the indicated time points.

Histopathology.

Mice (five per group) infected with 8.5×10⁴ (250 LD₅₀) c.f.u. by the i.n. route, with either WT Y. pestis or one of the mutant strains, were euthanized at 1 and 48 h post-infection (p.i.) using a mixture of ketamine (90 mg kg⁻¹) and xylazine (10 mg kg⁻¹). At these time points, portions of the lung, liver, spleen and heart tissues were placed in screw-capped containers and immersion-fixed in 10 % neutral buffered formalin, as described previously (Agar et al., 2008a, b; Sha et al., 2008). The tissues were transferred to new containers with fresh formalin 3 days later, and then processed and sectioned at 5 μm. Tissue sections were mounted on glass slides, stained with haematoxylin and eosin (H&E), and evaluated by light microscopy in a blinded fashion. Tissue lesions were scored based on a severity scale of minimal, mild, moderate and marked; the scale correlated with estimates of lesion distribution and extent of tissue involvement (i.e. minimal = 2–10 %; mild >10–20 %, moderate >20–50 %, severe >50 %), as previously described (Agar et al., 2008a, b; Sha et al., 2008). Inflammation indicated the presence of neutrophils usually associated with oedema and fibrin. In some sections, bacteria consistent with Y. pestis infection were present, although bacteria-specific staining was not performed.

Measurements of cytokines and chemokines in tissue homogenates and sera.

Concurrently with the collection of tissues for histopathology from infected mice, at 1 and 48 h p.i., groups of animals were killed, blood was collected and portions of the organs were homogenized. Uninfected mice served as a control group. Blood was centrifuged for 5 min at 1500 r.p.m. and the sera were extracted and filtered using 0.22 μm pore-size centrifugal filters (Millipore). The organs were weighed and homogenized by using tissue grinders (Kendall), and tissue homogenates were filtered using both 0.45 and 0.22 μm pore-size syringe filters. The levels of 32 cytokines and chemokines were simultaneously measured using a Milliplex multiplex assay (Millipore).

Survival of WT and mutant strains in the peripheral organs of mice.

Using the tissue homogenates and blood collected as described above, serial dilutions were made and cultured on SBA plates to assess the dissemination of the bacteria. Bacterial colonies were counted following 28 °C incubation for 48–72 h.

Survival of WT and mutant strains in murine RAW 264.7 macrophages.

WT and mutant Y. pestis strains were grown in HIB overnight at 28 °C. RAW 264.7 cells cultured in Dulbecco's Modified Eagle's Medium (DMEM) with 10 % fetal calf serum (FCS) and 1 % l-glutamine (Cellgro) were seeded in six-well plates, so that after 24 h they were 60–70 % confluent. Bacteria at an m.o.i. of 1 were centrifuged, and the pellets were resuspended in DMEM before being transferred into duplicate wells of RAW cells, as previously described (Sha et al., 2008). In brief, plates were incubated at 37 °C in 5 % CO₂ for 30 min, after which the wells were washed twice with PBS. Extracellular bacteria were killed by adding 100 μg gentamicin ml⁻¹ to each well for 1 h. One set of plates was harvested immediately (0 h time point), while DMEM containing 10 μg gentamicin ml⁻¹ was added to the other wells for the duration of the experiment. At each time point (4, 8 and 24 h p.i.), RAW cells were washed twice with PBS before being lysed with 300 μl sterile water. Lysed host cells were serially diluted 10-fold and cultured on SBA plates. After 48 h at 28 °C, bacterial colonies were counted to determine the percentage survival of each Y. pestis strain at the corresponding time point.

Statistical analyses.

To generate the survival curves, 10 mice per group were infected with multiple doses of WT and mutant strains of Y. pestis, and the Kaplan–Meier survival estimate was used for data analyses. Either Student's t test or ANOVA with Bonferroni correction were used for cytokine/chemokine analysis, bacterial dissemination studies in mice, and intracellular survival of bacteria in macrophages.

RESULTS

The Y. pestis pPCP⁻/Δlpp mutant strain is less lethal than the WT or its single mutants

Mice were inoculated intranasally with various doses of WT, Δlpp, pPCP⁻ or pPCP⁻/Δlpp Y. pestis and were monitored for morbidity and mortality (Fig. 1). Animals infected with a 250 LD₅₀ dose [1 LD₅₀ was calculated to be 340 bacteria (Sha et al., 2008)] of either WT or the Δlpp mutant of Y. pestis died by days 3 and 5, respectively. As expected, the mean time to death was increased in mice infected with the Δlpp mutant (Sha et al., 2008). Notably, animals infected with the pPCP⁻/Δlpp mutant strain showed much higher survival rates at all doses tested than did those animals infected with the pPCP⁻ mutant strain alone at the same doses. For example, at the lowest tested dose, those mice infected with a 250 LD₅₀ dose of the pPCP⁻/Δlpp mutant strain showed 100 % survival, while those infected with the pPCP⁻ mutant strain exhibited only a 30 % survival rate. In the 2500 LD₅₀ dose group, 100 % of the animals infected with the pPCP⁻/Δlpp mutant strain survived, while 20 % of mice infected with the pPCP⁻ mutant strain survived. Finally, at the highest dose of 5000 LD₅₀, 50 % of the animals infected with the pPCP⁻/Δlpp mutant strain survived. All mice infected with the pPCP⁻ mutant strain at this dose died by day 5 p.i. This indicated to us that curing of the pPCP1 plasmid and deleting the lpp gene synergistically attenuated the virulent Y. pestis CO92 strain.

Fig. 1.

Survival analysis of mice infected with WT and mutant Y. pestis strains. Following i.n. inoculation with various doses of WT, Δlpp, pPCP⁻ and pPCP⁻/Δlpp mutants of Y. pestis, animals were monitored for morbidity and mortality. Those infected with a 250 LD₅₀ dose of either WT (•) or Δlpp (○) Y. pestis died by days 3 and 5, respectively. However, mice infected with either a 250 LD₅₀ (▵) or a 2500 LD₅₀ (□) dose of the pPCP⁻/Δlpp strain showed a 100 % survival rate, while those infected with the pPCP⁻ strain showed 30 % (▾) and 20 % (▪) survival rates, respectively. At the highest dose tested, 5000 LD₅₀, 50 % of the mice infected with the pPCP⁻/Δlpp strain survived (◊), while no mice infected with the pPCP⁻ strain survived (⧫). By Kaplan–Meier analysis, only the group infected with a 5000 LD₅₀ dose of the pPCP⁻ strain was not significantly different from the WT- and Δlpp-infected groups. All other treatment groups were significantly different (P<0.001).

WT-infected mice have more severe tissue injury than do mice infected with the mutant strains

Based on the data presented in Fig. 1, we chose a time point of 48 h for collection of tissues and blood for our subsequent analyses, as by day 3 all of the animals infected with the WT bacterium died due to the high dose (250 LD₅₀) used in this study. Portions of the mouse tissues were processed and stained with H&E to assess the level of tissue damage associated with the bacterial infection (Fig. 2 and Supplementary Figs S1 and S2). At 1 h p.i., no significant tissue damage was noted in the lungs, livers, spleens or hearts of mice infected with WT, Δlpp, pPCP⁻ or pPCP⁻/Δlpp mutant strains (data not shown). These tissues exhibited histopathology similar to those sections from uninfected mice (anaesthetized and given PBS) serving as negative controls (data not shown).

Fig. 2.

Histopathology of mouse tissues following infection with WT and mutant strains. Groups of five mice were infected with WT Y. pestis or Δlpp, pPCP⁻ or pPCP⁻/Δlpp mutant strains and euthanized at 1 and 48 h p.i. The lungs (a), livers (b), and spleens or hearts (Supplementary Figures S1–S2 and Supplementary Tables S2–S5) were harvested at these time points. No significant tissue damage was noted in mice infected with WT Y. pestis at 1 h p.i. (results not shown). Tissue sections of WT-infected mice at 48 h p.i. are shown in (A) and (E), those from Δlpp-infected mice are shown in (B) and (F), those from pPCP⁻-infected mice are shown in (C) and (G), and those from pPCP⁻/Δlpp-infected mice are shown in (D) and (H). Bacteria in these sections are indicated by arrows, oedema is indicated by asterisks, inflammation is outlined by chevrons, and necrosis is indicated with an arrowhead. All sections were stained with H&E and scale bars are present on each image.

At 48 h p.i., the lungs (Fig. 2a) of WT-infected mice (A and E) showed mild alveolar oedema (asterisks), as well as mild neutrophilic inflammation. Bacteria (arrows) were present in the alveoli of all mice and in the perivascular region in two out of five mice. The lungs of all five Δlpp-infected mice (B and F) showed moderate oedema (asterisks) and mild neutrophilic inflammation in the alveoli. Bacteria (arrows) were present in most alveoli associated with fibrin. The interstitium had small areas of necrosis in two of the five mice, and one mouse had multiple areas of moderate perivascular oedema with mild perivascular fibrin. The lungs of mice infected with the pPCP⁻ mutant strain (C and G) showed mild neutrophilic inflammation (outlined by chevrons) in both the alveolar and perivascular regions. Three of five mice had mild alveolar fibrin, one of five mice had alveolar oedema, and one of five had mild haemorrhaging. The lungs of all five mice infected with the pPCP⁻/Δlpp mutant (D and H) had perivascular neutrophilic inflammation, while only three had perivascular oedema. Two of five mice had mild peribronchiolar inflammation (outlined by chevrons), and one mouse had mild neutrophilic inflammation in the bronchioles.

The livers (Fig. 2b) of two of the five WT-infected mice (A and E) showed small areas of necrosis (arrowhead) and minimal neutrophilic inflammation in the hepatic sinusoids, which also contained occasional bacteria (arrows) and fibrin. Two of five Δlpp-infected mice had minimal neutrophilic inflammation in the sinusoids of the liver (B and F), while one mouse had bacteria (arrows) in the sinusoids. One of five mice infected with the pPCP⁻ mutant strain had bacteria in the liver sinusoids (C and G, arrow). Two of five mice had minimal and mild neutrophilic liver inflammation (outlined by chevrons). Three of five mice infected with the pPCP⁻/Δlpp mutant (D and H) had mild neutrophilic inflammation (outlined by chevrons), and one of these had a small area of necrosis in the sinusoids. The livers of the other two mice in this group were within normal limits.

Four out of five spleens (Supplementary Fig. S1) in the WT-infected mice group (A and E) had bacteria (arrows) and fibrin in the red pulp, while one mouse had lymphoid depletion in the white pulp. One of five mice infected with the Δlpp mutant had bacteria (arrows) and fibrin in the red pulp and moderate lymphoid depletion in the white pulp of the spleen (B and F). No mice infected with the Y. pestis pPCP⁻ or pPCP⁻/Δlpp mutant strains exhibited any lesions in the spleen (C, G and D, H, respectively). These histopathology data corresponded with gradual and reduced mortality in mice infected with the pPCP⁻ or pPCP⁻/lpp mutant strains compared with that found in the WT and Δlpp mutant strains (Fig. 1).

The histopathology for the hearts of mice is also shown in Supplementary Fig. S2, and appeared unremarkable. Overall, the inflammatory lesions and bacteria seen in the WT-infected mouse organs were consistent with Y. pestis infection. In general, lesions tended to diminish in occurrence, severity and/or the amount of bacteria present in the mice infected with the Δlpp, pPCP⁻ and pPCP⁻/Δlpp mutants as compared with the WT-infected mice, with the most minimal changes occurring in the latter two groups. The pathology seen in various organs of mice infected with the WT and the Δlpp mutant strain of Y. pestis appeared similar. However, we noted that in the organs of mice infected with the Δlpp mutant, there were less severe histopathological changes when compared with those tissues from WT-infected animals at 48 h p.i., which coincided with a longer mean time to death with the Δlpp mutant (Fig. 1).

Curing of the pPCP1 plasmid or deleting the lpp gene results in differential bacterial survivability in mouse peripheral organs

Organs retrieved for histopathological analysis were also homogenized and cultured to assess the survivability of the two pPCP mutant strains (pPCP⁻ and pPCP⁻/Δlpp) as compared with the WT and Δlpp mutant strains. The organs of five mice infected with each of the four bacterial strains were homogenized at 1 and 48 h p.i. and cultured on SBA plates. Colonies were counted and data plotted on a log scale to easily visualize the increases and decreases in bacterial numbers. Importantly, the numbers of bacteria in the lungs (Fig. 3a) of WT- and Δlpp mutant-infected mice increased by 1–2 logs at 48 h p.i. when compared with 1 h p.i. control lung tissues. However, the number of bacteria cultured from WT- and Δlpp mutant-infected mice was not statistically different at 48 h p.i. Conversely, in mice infected with either the pPCP⁻ or the pPCP⁻/Δlpp mutant, the levels of bacteria present in the lungs at 48 h p.i. dropped when compared with the levels in the lungs at 1 h p.i. This drop was significant (P<0.05) when we compared the findings with those in the WT-infected mice, but not when compared with the levels in the Δlpp mutant-infected mice at 48 h p.i. This indicated an inability of these mutant bacteria (pPCP⁻ or pPCP⁻/Δlpp mutant) to survive within the lung.

Fig. 3.

Survival of WT and mutant Y. pestis strains at 1 and 48 h p.i. in peripheral organs of mice. Groups of five animals were infected with a 250 LD₅₀ dose of WT and mutant Y. pestis strains. At 1 and 48 h p.i., mice were euthanized, blood was drawn and organs were harvested, homogenized, serially diluted and cultured on SBA plates. Bacterial counts per gram of lung (a), liver (b), spleen (c) and heart (d) and per millilitre of blood (e) were determined for 1 h p.i. (•) and 48 h p.i. (○). Solid bars indicate the average value at 1 h p.i., while hatched bars indicate the average value at 48 h p.i. The dashed line parallel to the x axis is the limit of bacterial detection. Significant differences (Student's t test) between groups are noted.

Bacteria recovered from the livers of mice (Fig. 3b) infected with either the WT or its Δlpp mutant strain increased by 4–5 logs from 1 to 48 h p.i. These increases in the numbers of bacteria recovered from the livers of Δlpp-infected mice were similar to what we reported previously (Sha et al., 2008). However, only one of five mice in either the pPCP⁻- or the pPCP⁻/Δlpp mutant-infected group had an appreciable number of bacteria in the liver. In the remaining four animals in each group, no detectable numbers of bacteria were noted in the livers when compared with those findings in the WT-infected mice. The limit of detection of bacteria in various organs in mice was ∼80 c.f.u.

In the spleen (Fig. 3c), there were no significant differences between the numbers of bacteria that disseminated in WT-infected mice compared with findings in the Δlpp-, pPCP⁻- or pPCP⁻/Δlpp mutant-infected animals at 48 h p.i. The trend, however, indicated an inability of Δlpp mutant bacteria (only one out of five mice having an appreciable number of bacteria) to survive within spleen cells, as we previously reported (Sha et al., 2008). Interestingly, in three and two out of five mice, respectively, appreciable numbers of the pPCP⁻ and pPCP⁻/Δlpp mutant bacteria were noted in the spleen. These data were in contrast to what was observed in the lungs and livers of mice infected with the pPCP⁻ and pPCP⁻/Δlpp mutant strains, as they did not survive in these tissues after 48 h p.i. (Fig. 3a, b). These data suggested to us that both of these mutants could disseminate to the liver and spleen but were possibly differentially killed in these two organs.

No pPCP⁻ or pPCP⁻/Δlpp mutant bacteria were detected in the hearts of mice (Fig. 3d), whereas significantly (P<0.05) higher levels of WT and Δlpp strains were noted in the heart at 48 h p.i. Likewise, WT bacteria were detected in the blood of mice (Fig. 3e) at 48 h p.i. Only one out of five mice showed appreciable numbers of Δlpp mutant bacteria in the blood. Although no pPCP⁻/Δlpp mutant was detected in the blood at 48 h p.i., three out of five mice had low levels of the pPCP⁻ mutant in the blood.

Mice infected with strains cured for the pPCP1 plasmid show lower levels of cytokines and chemokines

In parallel with the histopathology and dissemination experiments, blood was drawn via cardiac puncture from mice at 1 and 48 h after i.n. inoculation with either WT Y. pestis or one of the mutant strains. The sera from uninfected mice were used as controls. The levels of 32 cytokines and chemokines were evaluated by using the Millipore Milliplex assay. A maximum of 23 of these cytokines/chemokines showed a response and are listed and defined in Supplementary Table S1.

The levels of cytokines and chemokines (pg ml⁻¹) for all tissue homogenates and in the sera can be found in Supplementary Tables S2–S6. Most remarkable were the drops in the levels of cytokines and chemokines noted in the pPCP⁻/Δlpp-infected animals compared with the WT-infected mice. These data have been summarized based on their percentage differences in Table 1. Generally speaking, WT-infected mice had an overwhelming inflammatory response at 48 h p.i., and the levels of cytokines and chemokines from mice infected with the Δlpp mutant were similar to those for WT-infected animals. This is not surprising given the high dose used for this study. The levels of cytokines/chemokines in mice infected with the pPCP⁻ mutant strain, however, had dropped by 48 h p.i. when we compared them with the levels in WT-infected animals. For most of these cytokines and chemokines, the decrease was substantially greater in mice infected with the pPCP⁻/Δlpp mutant when compared with either of the single mutants. This indicated to us that a synergistic effect occurred in these cytokine/chemokine responses attributable to deletion of the lpp gene and curing of the pPCP1 plasmid in the mouse model of infection. In fact, in the lungs of mice (Supplementary Table S2), 16 of these cytokines and chemokines decreased between 90 and 100 % as compared with those levels from the WT-infected animals, and five more decreased between 50 and 89 % at 48 h p.i. Eotaxin levels decreased by 17 % when compared with the levels in WT-infected mice, while IL-10 levels decreased by 32 %.

Table 1.

Percentage changes in cytokine and chemokine levels in tissue homogenates 48 h p.i. with Y. pestis pPCP⁻/Δlpp when compared with WT CO92

bdl, Below detectable limit.

*Significant changes as compared with WT-infected mice (Student's t test, P<0.05).

†Increase in this cytokine or chemokine above the level in the WT.

In mice infected with the pPCP⁻/Δlpp mutant strain, significant decreases (P<0.05, indicated by an asterisk) were noted in the levels of 14 cytokines and chemokines in the liver (Supplementary Table S3), with G-CSF, IFN-γ, IL-1α, KC, LIF, MCP-1, M-CSF and MIP-2 dropping between 90 and 100 %, when compared with the levels in WT-infected animals. IL-1β, IL-6, MIG, MIP-1α, and TNF-α levels decreased between 50 and 90 %, while LIX decreased by only 16 % compared with levels in the livers of WT-infected mice at 48 h p.i. The levels of Eotaxin and IL-12p70 increased compared with those for the WT at 48 h p.i., while neither the IL-10 nor MIP-1β levels changed at 48 h p.i in the livers of animals infected with the pPCP⁻/Δlpp mutant strain. We again noted a synergistic effect in the decrease of many of the cytokine and chemokine levels in the livers of mice infected with the pPCP⁻/Δlpp mutant as compared with mice infected with either of the single mutant strains by 48 h p.i.

In the spleen (Supplementary Table S4) at 48 h p.i., we noted changes in the levels of 17 cytokines and chemokines in mice infected with the pPCP⁻/Δlpp mutant strain compared with the levels in WT-infected animals, with 13 of these showing a decrease between 90 and 100 %, three between 50 and 89 %, and one with a 21 % decrease. Many of these showed a synergistic decrease when compared with mice infected with the single mutants. Only RANTES increased significantly (P<0.05, indicated by a dagger) in mice infected with the pPCP⁻/Δlpp mutant strain, compared with RANTES levels found in WT-infected animals at 48 h p.i.

We noted significant decreases in the levels of eight cytokines and chemokines in the heart homogenates (Supplementary Table S5) of mice infected with the pPCP⁻/Δlpp mutant strain at 48 h p.i. Levels of five of these cytokines and chemokines were synergistically lower than the levels seen in either of the single mutant-infected animals (pPCP⁻ or Δlpp). The levels of LIX and M-CSF remained unchanged in the hearts of pPCP⁻/Δlpp mutant-infected mice.

In the sera (Supplementary Table S6) of mice infected with the pPCP⁻/Δlpp mutant strain at 48 h p.i., the levels of G-CSF, GM-CSF, IL-1α, IL-6, IL-9, IL-12p70, IP-10, MIP-1α, MIP-1β, RANTES and TNF-α were all significantly decreased by 80–100 % compared with those in WT-infected animals. Many of these were also synergistically lower than the levels seen in the sera of single mutant-infected mice groups. Eotaxin levels decreased by 67 %, and no cytokine or chemokine levels increased in the sera of mice infected with the pPCP⁻/Δlpp mutant compared with those in WT-infected mice.

Survivability of Y. pestis mutants in murine macrophages is dependent upon the lpp gene rather than the pPCP1 plasmid

RAW 264.7 murine macrophages were infected with each of the Yersinia strains, and infected macrophages were harvested after a 1 h gentamicin treatment (0 h time point), and at 4, 8 and 24 h post treatment. At each time point, macrophages were lysed and cultured on SBA plates to determine the percentage intracellular survival of Y. pestis strains (Fig. 4). Data were normalized to the 0 h time points.

Fig. 4.

Survival of WT and mutant Y. pestis strains in vitro. Murine RAW 264.7 macrophages were infected at an m.o.i. of 1 with WT and mutant Y. pestis strains for 30 min. Monolayers were treated with 100 μg gentamicin ml⁻¹ for 1 h, and at the indicated time points (post gentamicin), macrophages were lysed. Lysates were serially diluted and cultured on SBA plates. Colonies were counted and data were normalized to the 0 h time point. Significant differences (ANOVA, P<0.001) between groups are noted. (b) Comparison of WT- and Δlpp mutant-infected macrophages with the lpp-complemented strain of Y. pestis CO92. Significant differences (Student's t test, P<0.05) between groups are noted. The inset in (b) shows Western blot analysis data showing the presence or absence of Lpp (6.3 kDa) using specific antibodies in various strains.

At 4 h p.i. (Fig. 4a), the numbers of WT cells recovered from macrophages decreased slightly, compared with those at the 0 h time point, while the numbers of mutant bacteria recovered from macrophages were significantly decreased (P<0.001). The numbers of the pPCP⁻/Δlpp mutant strain recovered from macrophages decreased the most, however, to a level significantly different (P<0.001) from that of the pPCP⁻ mutant strain, while maintaining a level of survival similar to that of the Δlpp mutant strain. At 4 h p.i., the absence of Lpp appeared to be the determining factor for the intracellular survival of these Y. pestis strains within macrophages.

At 8 h p.i. (Fig. 4a), the number of WT Y. pestis that survived within the macrophages decreased only slightly from the 4 h time point and again showed a significantly higher (P<0.001) level of survivability when compared to findings in the other three mutant strains. A similar pattern with the mutant strains emerged at this time point, except that the number of Δlpp mutant bacteria that survived decreased to a level significantly (P<0.001) below that of the pPCP⁻ or pPCP⁻/Δlpp mutant strains recovered from macrophages.

At 24 h p.i. (Fig. 4a), the number of WT bacteria that survived inside the macrophages did not change compared with their numbers at the 8 h time point. Although the number of pPCP⁻ mutant bacteria increased from the 8 h time point, the numbers of bacteria retrieved from macrophages were significantly (P<0.001) lower than that for the WT bacteria. The survivabilities of the Δlpp and pPCP⁻/Δlpp mutant strains dropped dramatically compared with the survivability of either the WT or pPCP⁻ mutant strains and were not significantly different from each other at 24 h p.i. In addition, the strain in which the lpp gene was complemented showed statistically similar survivability to that of the WT bacteria in macrophages 24 h p.i. (Fig. 4b), and the complementation of the Δlpp mutant was also confirmed by Western blot analysis (inset, Fig. 4b). A 6.3 kDa band corresponding to Lpp could clearly be detected in the WT and lpp complemented strains but not in the Δlpp mutant of Y. pestis. Taken together, these data indicated that the lack of survivability of Y. pestis within macrophages seemed to be more dependent on the presence of Lpp rather than the pPCP1 plasmid.

DISCUSSION

We recently showed that deletion of the lpp gene from Y. pestis CO92 affects the virulence of the organism in the mouse model of pneumonic plague (Sha et al., 2008). Other investigators have also reported attenuation in Y. pestis strains lacking either the pPCP1 plasmid or the pla gene (Brubaker et al., 1965; Lathem et al., 2007; Samoilova et al., 1996; Sodeinde et al., 1992; Welkos et al., 2002, 1997). In fact, Lathem and co-workers concluded that Pla was responsible for establishing infection in the lung when bacteria were administered intranasally but was not necessary for Y. pestis to disseminate from the lungs to peripheral organs, e.g. the spleen. For this study, we passaged bacteria at a low temperature to cold-cure the smallest virulence plasmid, pPCP1, from the WT and the Δlpp mutant strain. Plasmid pCD1-dependent cytotoxicity, measured as previously described (Sha et al., 2008), did not differ between the WT and the mutant strains (Supplementary Fig. S3), and we were able to isolate the two remaining virulence plasmids and their associated virulence genes from the mutants (results not shown). Additionally, we noted that a 250 LD₅₀ dose of the pPCP⁻ strain, when administered into mice, provided minimal protection when compared with the WT bacteria, suggesting that cold-curing did not affect virulence of the mutant. Most interesting, however, was the observation that Y. pestis strains lacking both the lpp gene and the pPCP1 plasmid were highly attenuated in the mouse model of pneumonic plague, when compared with the single lpp gene deletion or the pPCP⁻ strain. Furthermore, such a pPCP⁻/Δlpp mutant strain exhibited drastically reduced tissue injury and a muted cytokine and chemokine induction in tissue homogenates and sera when compared with equivalent findings in WT, Δlpp and pPCP⁻ mutant strains. We opted to cure the pPCP1 plasmid from WT and Δlpp Y. pestis strains rather than selectively delete the pla gene, as pesticin and pesticin-immunity protein do not contribute to the virulence of Y. pestis in the animal model (Anisimov et al., 2009).

We used a much higher dose (250 LD₅₀) of bacteria in the current report with similar results to a previous study (Sha et al., 2008). Based on our previous report at a dose of 5 LD₅₀, WT-infected mice began to die between 60 and 72 h p.i., and we noted appropriate cytokine and chemokine responses and definitive pathology by 48 h p.i. (Agar et al., 2008a, b; Sha et al., 2008). Additionally, multiple survival/mortality experiments using the pPCP⁻ and pPCP⁻/Δlpp mutant strains indicated that we needed very high doses to determine the LD₅₀ of these two mutant strains. Therefore, in order to use WT- and Δlpp-infected mice as a comparison with the two pPCP⁻ mutant strains with such a large amount of bacteria, we chose the latest time point as 48 h p.i. in this study.

In terms of dissemination and survivability of WT Y. pestis CO92 and its various mutants in a mouse model, we concluded that the Δlpp mutant could leave the confines of the lungs to replicate within the liver and the heart, but was unable to do so in the spleen or blood and hence cleared from these organs by 48 h p.i. This confirmed our previous findings in which we demonstrated a high bacterial (Δlpp) load in the lungs and liver of mice but not in the spleen and blood at 48 h p.i. (Sha et al., 2008). The inability to recover mutant bacteria from the spleen could be explained by the uptake and subsequent killing of the mutant by splenic cells or by the migration of immune cells from the spleen following uptake of the mutant bacteria (Marketon et al., 2005). In the lungs, the ability of Δlpp mutant bacteria to replicate indicated either that the alveolar macrophages were unable to kill the mutant efficiently or that the bacteria were also taken up by lung epithelial cells, where they survived effectively and replicated. It is also possible, as was noted above, that these bacteria in the lungs resisted phagocytosis and efficiently grew extracellularly within the lung.

The pPCP⁻ and pPCP⁻/Δlpp mutant strains were not able to survive in the lung, unlike the WT and Δlpp mutant strains, and those bacteria that left the lung either showed a tropism for the spleen over the liver, heart or blood, or were cleared by the last three organs by 48 h. This was confirmed by our histopathology data, with pPCP⁻ and pPCP⁻/Δlpp mutant strains completely clearing from the tissues 48 h p.i. Clearly, Pla was required to establish an infection in the lungs of mice and possibly dissemination of bacteria from the lungs to other organs. However, the presence of pPCP⁻ and pPCP⁻/Δlpp mutant bacteria at 48 h p.i. in the spleens of 40–60 % of the mice indicated that these mutant bacteria did disseminate, albeit in lower numbers, or that these mutant bacteria were efficiently killed in the lungs, liver, heart and blood, but not in the spleen. Alternatively, the pPCP⁻ mutant might have a tropism for the spleen, while the Δlpp mutant migrated mostly to the liver. These data are very provocative and require further studies.

These data corroborated the cytokine and chemokine induction data from the tissue homogenates and the sera. Mice infected with the Δlpp mutant had significantly lower levels of MIG and MIP-2 in the liver homogenates, IP-10 and MIG in the spleen homogenates, and IL-9 and MIG in the heart homogenates when compared with WT-infected mice. However, the levels of the above-mentioned cytokines and chemokines were lowered even further in those mice infected with either of the two pPCP⁻ mutants (pPCP⁻ and pPCP⁻/Δlpp) when compared with the levels in the Δlpp mutant-infected mice. Interestingly, MIG and IP-10 are induced by IFN-γ (Brice et al., 2001), the cytokine with immunoregulatory functions important in the host response to infection. In the case of the WT- and Δlpp-infected mice, however, the overwhelming induction of cytokines and chemokines led to their demise. Animals infected with either pPCP⁻ mutant, on the other hand, showed lower cytokine and chemokine responses, and evidence of tissue repair by 48 h p.i., which indicated to us that by this time point the mice had resolved the infection. In fact, a study by Lathem and co-workers showed that mice infected with a pla isogenic mutant had little elevation of the quantities of cytokine transcripts (IL-10, IL-1α, IL-17, IL-6, MIP-2 and TNF-α) at 24 h p.i. (Lathem et al., 2007), which indicated that these mice and those in our study avoided the massive proinflammatory response that more than likely contributed to the demise of our WT- and Δlpp-infected animals. Taken together, these data suggest to us that Lpp and Pla contribute to bacterial virulence by different mechanisms, but synergistically result in a more efficient clearance of the bacteria from mice and hence their higher survival rate.

These in vivo data corroborated our in vitro studies, in which we infected murine RAW 264.7 macrophages with various mutant strains. As noted, Δlpp and pPCP⁻/Δlpp mutant bacteria died rapidly within the macrophages, when compared with the WT or pPCP⁻ strain of Y. pestis. These data indicated that Lpp and not Pla contributed to bacterial survival in macrophages. This supported an earlier study which showed that neither the pCD1 nor the pPCP1 virulence plasmid is required for Yersinia to efficiently grow within macrophages (Straley & Harmon, 1984). In contrast, in the liver, both the WT and Δlpp mutant bacteria could efficiently survive. Y. pestis produces an antiphagocytic capsule and Yops at 37 °C, and as a result efficiently grows extracellularly (Cavanaugh & Randall, 1959; Pujol & Bliska, 2005; Viboud & Bliska, 2005). Another alternative is that the WT and Δlpp mutant bacteria escaped killing by resisting phagocytosis and grew extracellularly in the liver.

Importantly, deletion of Lpp and Pla resulted in a greater ability of mice to recover from infection following high doses (250–2500 LD₅₀), as compared with the single-mutant strains, and this effect was synergistic. Because of the absence of Pla, chemotaxis of inflammatory cells was not impeded and complement C3 was not inactivated, resulting in a vigorous enough immune response for the mice to survive infection (Sodeinde et al., 1992).

Abbreviations

• H&E, haematoxylin and eosin

• i.n., intranasal

• p.i., post-infection

• WT, wild-type

• Yops, Yersinia outer-membrane proteins