ABSTRACT
Yersinia pestis is the causative agent of bubonic, pneumonic, and septicemic plague. We demonstrate that Toll-like receptor 2-deficient (TLR2−/−) mice are resistant to septicemic infection by the KIM5 strain of Y. pestis but not to infection by the CO92 Δpgm strain. This resistance is dependent on TLR2, the route of infection, and the isoform of YopJ. Elevated bacterial burdens were found in the spleens of CO92 Δpgm-infected animals by 24 h postinfection and in the livers by 4 days. The YopJ isoform present contributed directly to cytotoxicity and inflammatory cytokine production of bone marrow-derived macrophages from TLR2−/− mice. Immune cell trafficking is altered in CO92 Δpgm infections, with an increased neutrophil infiltration to the spleen 5 days postinfection. Immune cell infiltration to the liver was greater and earlier in KIM5-infected TLR2−/− mice. The functionality of the immune cells was assessed by the ability to develop reactive oxygen and nitrogen species. Our data suggest an inhibition of granulocytes in forming these species in CO92 Δpgm-infected TLR2−/− mice. These findings suggest that resistance to KIM5 in TLR2−/− mice is dependent on early immune cell trafficking and functionality.
INTRODUCTION
Three of the 11 species found in the genus Yersinia are pathogenic to humans: Yersinia pseudotuberculosis, Yersinia enterocolitica, and Yersinia pestis (1). Y. pseudotuberculosis and Y. enterocolitica cause gastrointestinal disorders (2), whereas Y. pestis is the causative agent of bubonic, pneumonic, and septicemic plague (3). Plague is transmitted to mammalian hosts via infected flea bites. After transmission to a mammalian host, the bacteria evade the host innate immune response and colonize the proximal lymph nodes (4). The organism will proliferate in the lymph nodes, leading to enlarged and inflamed lymph nodes called buboes, a manifestation of bubonic plague. From the lymph nodes, the organism spreads systemically to spleen, liver, and lungs. Colonization of bacteria in lungs can lead to secondary pneumonia and may result in person-to-person transmission, causing primary pneumonic plague (5).
Y. pestis has been classified into the following 4 biovars depending on their ability to reduce nitrate and ferment glycerol: Antiqua (positive for both metabolic functions), Medievalis (cannot reduce nitrate but can ferment glycerol), Orientalis (can reduce nitrate but cannot ferment glycerol), and the non-human-pathogenic biovar Microtus. Y. pestis strain CO92 (Orientalis) is an American isolate from the third plague pandemic (6), while Y. pestis strain KIM5 (Medievalis) is an isolate from an endemic plague focus (7). These strains share 95% sequence similarity between their 2 genomes (7, 8). The CO92 genome is ∼50 kbp larger than the KIM5 genome, due to the presence of an 11-kbp insertion and several other small insertions in the CO92 genome relative to that of KIM5. About 27 kbp of the differences are due to insertion sequence (IS) elements present in CO92 (7). Strain CO92 has one less rRNA operon than strain KIM5. Both strains of Y. pestis harbor an ∼70-kbp highly conserved plasmid, pCD1 (calcium dependence), containing ∼50 virulence genes encoding the type III secretion system (T3SS). The T3SS delivers several effector molecules (YpkA, YopH, YopE, YopJ, YopT, YopK, and YopM) into the cytosol of host cells (2). YopJ, an effector molecule secreted by the T3SS, inhibits the mitogen-activated protein kinase (MAPK) and nuclear factor-kappa B (NF-κB) pathways. Delivery of YopJ to host cells prevents the formation of antiapoptotic factors, and Yersinia-infected macrophages thereby undergo apoptosis that requires Toll-like receptor 4 (TLR4)-dependent activation of caspases (9–12). YopJ is homologous to cysteine proteases (13) and also functions as a deubiquitinase (14–16). More recent studies identify YopJ as an acetyl transferase that is activated by a host-specific factor, inositol hexakisphosphate (17). YopJ acetylates Ser and Thr residues important for activation of MAPK kinase (MKK) and IκB kinase complex (IKKβ) pathways (17, 18). YopJ-mediated acetylation of MKKs and IKKβ directly competes with phosphorylation of these residues, thus blocking the signal transduction required for activation of MAPK and NF-κβ transcription factors (18).
Y. pestis-infected hosts have a lack of early inflammatory responses with immune cell depletion at sites of bacterial replication and dampening of proinflammatory cytokine responses (1). Although Y. pestis causes deadly infections in hosts, there exist a percentage of hosts that are intrinsically resistant to plague (19, 20). Multiple laboratory strains of inbred mice are resistant to plague, but the mechanisms of resistance and susceptibility are poorly understood. Susceptible C57BL/6 mice infected with the KIM5 strain of Y. pestis have a 50% lethal dose (LD50) of 20 to 50 CFU (19), whereas KIM5-infected 129S2/SV.Hsd mice have an LD50 of 2 × 106 CFU (19). Some BALB/c strains of mice are susceptible to plague infection; however, the BALB/cJ substrain is resistant to plague infection (21). The resistance was mapped to a region that coincided with the major histocompatibility complex on chromosome 17 (20). Previous research from our laboratory has demonstrated differences in susceptibility of B10.T (6R) mice systemically infected with the KIM5 strain of Y. pestis. This difference was age dependent, where 2- month-old mice were more resistant than 5- to 12-month-old mice, with a difference in LD50 of over 2 log (22).
Like TLR4, TLR2, upon stimulation, will induce proinflammatory cytokines such as tumor necrosis factor alpha (TNF-α) and interleukin-6 (IL-6). TLR2 can also act on a dual axis for the induction of the anti-inflammatory cytokines IL-10 and IL-4. The production of these cytokines is biphasic; the proinflammatory cytokines are produced first, followed later by the anti-inflammatory cytokines (23, 24). The role of TLR2 in Y. pestis infections has not been extensively studied. Recombinant LcrV has been demonstrated to be a potent TLR2 agonist only when present in highly aggregated forms. The dimer and trimer forms of LcrV that form the majority of the LcrV present do not stimulate TLR2 robustly. Differences in the pathogenicity of subcutaneous infections in wild-type and TLR2−/− mice utilizing the KIM1001 and GB strains were not apparent. These data have led researchers to conclude that TLR2 does not play a significant role in Y. pestis pathogenesis (25, 26).
In this paper, the resistance of TLR2−/− mice to intravenous (i.v.) Y. pestis KIM5 challenge was examined. Interestingly, TLR2−/− mice that are resistant to the KIM5 strain were susceptible to the CO92 Δpgm strain. To our knowledge, this is the first report to document such a difference in the virulence of plague strains. The infectious route was altered from i.v. to subcutaneous, and the difference in strain virulence was ablated. These results recapitulate earlier findings of TLR2 not playing a significant role in pathogenesis. The site where the i.v. challenge was administered did not affect the virulence difference of the two strains. The cytotoxicity of bone marrow-derived macrophages (BMDMs) and the level of cytokine production were increased in KIM5-treated cells and reduced in CO92 Δpgm-treated cells, indicating a mechanism of inflammatory reduction by the CO92 Δpgm strain. This phenotype was mirrored in the KIM5 strain when the effector molecule YopJ was replaced with the YopJ isoform from CO92 Δpgm. Immune cell profiling indicated an increase in neutrophils 5 days postinfection in CO92 Δpgm-challenged mice, but the granulocytic cells 5 days postinfection had reduced amounts of reactive nitrogen and oxygen species generated in CO92 Δpgm-challenged mice, indicating a dysfunction of the immune cells when encountering the pathogen. Finally, liver histology revealed increased inflammatory foci in mice challenged with KIM5 3 days postinfection. These results suggest that pathogenic virulence differences between CO92 Δpgm and KIM5 can be attributed to the amount of early inflammation being produced, as well as the trafficking timing and functionality of immune cells recruited to the site of infection.
RESULTS
TLR2−/− mice are resistant to KIM5 infection.Groups of inbred C57BL/6 (B6) mice and TLR2−/− mice were challenged i.v. with KIM5 and CO92 Δpgm. Both B6 groups succumbed to infection 4 days after challenge. TLR2−/− mice challenged with CO92 Δpgm all succumbed to infection 6 days postinfection. TLR2−/− mice challenged with KIM5 had no mortality out to 21 days postinfection (Fig. 1A). TLR4−/− and TLR2−/− mice were challenged with KIM5, and once again, no mortality was seen in the TLR2−/− group while 100% mortality was seen in the TLR4−/− mouse group (Fig. 1B). One of the differences between Y. pestis KIM5 and Y. pestis CO92 is a variance in YopJ. The YopJ isoform in Y. pestis KIM5 has amino acid differences at two sites, F177L (L instead of F at position 177) and K206E, compared to YopJ of Y. pestis C092 (27). The differences in YopJ isoforms do not appear to affect the translocation of other effector molecules (27). A recent paper demonstrated that YopJ is essential for the efficient spread of Y. pestis KIM5 within the lymphatic network, resulting in necrotic death of infected macrophages and the promotion of septic infection (28). This led us to determine whether the different isoforms of YopJ were responsible for the difference in infection outcomes. TLR2−/− mice were infected with mutants expressing different isoforms of YopJ. Mutations that allowed production of the Y. pestis CO92 YopJ isoform in Y. pestis KIM5 (KIM5-YopJCO92) increased the virulence of Y. pestis KIM5 to approximately the same value as shown by the Y. pestis CO92 Δpgm strain (Fig. 1C). These results indicate that KIM5 is still pathogenic in B6 mice and is not pathogenic in the resistant TLR2−/− mice and that the phenotype may be dependent on the YopJ isoform.
TLR2−/− mice are resistant to Y. pestis strain KIM5 infection. Differing survival patterns of mice challenged with Y. pestis strains KIM5 and CO92 Δpgm. (A, B) Survival curves of C57BL/6 (B6) and TLR2−/− female mice (A) and of TLR4−/− and TLR2−/− female mice (B) challenged with 1,000 CFU of strain KIM5 or CO92 Δpgm as indicated in the key. (C) Survival curves of TLR2−/− female mice challenged with 1,000 CFU of strain KIM5 or KIM5-YopJCO92. ***, P < 0.005.
Route of infection but not site of infection ablates the resistance phenotype.Previous data have demonstrated no change in pathogenicity in subcutaneous infections in TLR2−/− mice compared to B6 mice (26). We sought to recapitulate those results and obtained similar findings. To determine if the resistance phenotype was a result of i.v. infections in the retro-orbital sinus, we also challenged TLR2−/− mice i.v. via tail vein injections. CO92 Δpgm-challenged mice all succumbed to infection by 6 days postinfection, while 100% of KIM5-challenged mice survived infection (Fig. 2A). TLR2−/− mice infected subcutaneously with KIM5 all succumbed to infection 4 days postinfection, and similar results were found with CO92 Δpgm, which resulted in 100% lethality 5 days postinfection (Fig. 2B). These results suggest that the route of infection dictates the resistance phenotype of TLR2−/− mice.
Route of infection but not site of infection ablates resistance phenotype. (A) Survival curves of TLR2−/− female mice challenged with 1,000 CFU of KIM5 or CO92 Δpgm via tail vein or retro-orbital sinus. (B) Survival curves of TLR2−/− female mice challenged with 1,000 CFU of KIM5 and CO92 Δpgm subcutaneously. ****, P < 0.0001; ns, not significant.
Differences in bacterial burden in CO92 Δpgm-infected TLR2−/− mice.Eventually, Y. pestis infection spreads systemically throughout the visceral organs (29). Bacterial burdens were assessed in the spleens and livers of TLR2−/− mice infected with 1 × 103 CFU of KIM5 or CO92 Δpgm 1, 3, 4, and 5 days postinfection (Fig. 3A to H). A significant increase in bacterial load was noted in the spleens of CO92 Δpgm-infected animals 1 day postinfection, and the trend remained throughout the infectious timeline. The bacterial load in the liver 1 day postinfection was significantly increased in CO92 Δpgm-infected animals (Fig. 3B). This increase diminished 3 days postinfection and then returned 4 and 5 days postinfection (Fig. 3D, F, and H). These data suggest that TLR2−/− mice infected with CO92 Δpgm are unable to successfully inhibit the systemic spread to the spleen that in turn spreads to the liver and overwhelms the immune response there.
Differences in bacterial burdens in CO92 Δpgm-infected TLR2−/− mice. Bacterial dissemination kinetics in the spleen and liver. Bacterial loads in the spleen and liver 24 h postinfection (A, B), 3 days postinfection (C, D), 4 days postinfection (E, F), and 5 days postinfection (G, H). *, P < 0.05; **, P < 0.01; ***, P < 0.005; ****, P < 0.0001; ns, not significant.
YopJ modulation in TLR2−/− BMDMs.To determine the effects that the different isoforms of YopJ could potentially have on the role it plays in pathogenicity, we measured cytotoxicity and inflammatory cytokine production in BMDMs isolated from TLR2−/− mice. We did a time course measurement of lactate dehydrogenase (LDH) release by BMDMs infected with KIM5, CO92 Δpgm, KIM5-YopJCO92, and a mutant with catalytically inactivated YopJ, KIM5-YopJC172A (Fig. 4A). Our data suggest that KIM5 is the only strain to produce significant amounts of cytotoxicity in TLR2−/− BMDMs. Supernatants collected during the time course were assayed for IL-1β, IL-1α, and TNF-α. These data suggest that CO92 Δpgm YopJ can inhibit the production of IL-1β and IL-1α and that both YopJ isoforms have the ability to inhibit the production of TNF-α (Fig. 4B to D). The isoform of YopJ that is present directly affects the ability of the bacteria to cause cellular lysis of macrophages and inhibit the production of proinflammatory cytokines.
YopJ modulation in TLR2−/− BMDMs. (A) Cellular cytotoxicity in BMDMs infected with an MOI of 10. (B to D) Cytokine production from BMDMs 6, 12, and 24 h postinfection. *, P < 0.05; **, P < 0.01; ***, P < 0.005; ****, P < 0.0001.
TLR2−/− mice infected with CO92 Δpgm have a late increase in neutrophils and an attenuated cytokine response.Early neutrophil infiltration and activation at the site of infection is critical for the control of Y. pestis infection. Immune cell phenotyping demonstrated an increase in splenic neutrophils 5 days postinfection with CO92 Δpgm (Fig. 5A). T cell phenotyping and activation statuses were also assessed, but no statistical differences were found (Fig. S2 in the supplemental material). Serum cytokine analysis revealed that IL-6 is more highly expressed at 3 days postinfection and gamma interferon (IFN-γ) at 5 days postinfection in KIM5-challenged mice (Fig. 5B and C). These data suggest that early neutrophil function is abrogated in CO92 Δpgm-infected mice and that proinflammatory mediators are present in KIM5-infected animals to enable the immune system to clear the remaining infection and prime the adaptive immune response.
Neutrophils are increased 5 days after CO92 Δpgm infection. (A) Spleens were harvested 3 and 5 days postinfection. Cells were stained and examined via flow cytometry, excluding debris, doublets, and dead cells. CD11b+ LY6G+ neutrophils were increased 5 days postinfection in CO92 Δpgm-infected TLR2−/− mice. Control mice were noninfected. (B) Serum levels of IL-6 3 days postinfection, demonstrating early inflammation for clearance, and serum levels of IFN-γ 5 days postinfection, demonstrating continued inflammation for clearance. Representative data shown are from 3 independent experiments (n = 3 or 4). *, P < 0.05; ***, P < 0.005; n.s., not significant.
TLR2−/− mice infected with CO92 Δpgm have reduced levels of ROS and RNS in their granulocyte population.It has previously been established that the production of reactive oxygen species (ROS) and reactive nitrogen species (RNS) is critical for the elimination of Y. pestis once internalized in neutrophils (30). We sought to investigate the levels of ROS and RNS present in our infectious model systems. Our data suggest that at 3 days postinfection, mice infected with CO92 Δpgm have decreased amounts of oxidative stress and superoxide in their granulocyte population (Fig. 6A and E). This phenomenon appears to be independent of the YopJ isotype present. At 5 days postinfection, CO92 Δpgm-infected mice have significantly decreased amounts of oxidative stress, nitric oxide, and superoxide in their granulocyte population compared to the results for KIM5-infected mice (Fig. 6B, D, and F). These data suggest a potential mechanism of granulocyte dysfunction in CO92 Δpgm-infected TLR2−/− mice compared to the granulocyte function in KIM5-infected mice.
TLR2−/− mice infected with CO92 Δpgm have reduced levels of ROS and RNS in their granulocyte population. (A, C, E) Reactive oxygen and nitrogen species measured 3 days postinfection. (B, D, F) Reactive oxygen and nitrogen species measured 5 days postinfection. MFI, mean fluorescence intensity; FITC, fluorescein isothiocyanate; APC, allophycocyanin. *, P < 0.05; ***, P < 0.005.
Immune cell infiltrates in livers of challenged TLR2−/− mice.Differences in the immune cell recruitment to the livers were observed in infected TLR2−/− mice. Liver sections of mice 3 and 5 days postinfection were stained with hematoxylin and eosin (H&E). At 3 days postinfection, the number of inflammatory foci was trending higher in KIM5-infected mice and the areas of the foci were significantly larger than in CO92 Δpgm-challenged mice (Fig. 7A and B). At 5 days postinfection, the inflammatory foci of CO92 Δpgm-challenged mice were significantly larger than those of KIM5-challenged mice; however, at 5 days postinfection, the KIM5-challenged mice had already cleared the bacteria from the liver (Fig. 7C and D), as previously demonstrated by the experiment whose results are shown in Fig. 3. These data suggest a delayed immune cell response to the liver of the CO92 Δpgm-challenged mice.
Immune cell infiltrates in livers of challenged TLR2−/− mice. TLR2−/− mice were infected with 1,000 CFU of Y. pestis KIM5 or CO92 Δpgm, and livers were harvested at 3 and 5 days and were stained with H&E. (A, C) Numbers of immune cell infiltrates per field 3 days (A) and 5 days (C) postinfection. (B, D) Relative areas of immune cell infiltrates 3 days (B) and 5 days (D) postinfection. Black arrows indicate examples of inflammatory foci. *, P < 0.05; ****, P < 0.0001; ns, not significant.
DISCUSSION
Yersinia pestis has developed many ways to manipulate and evade host immune responses. Y. pestis, like other Gram-negative bacteria, has lipopolysaccharide (LPS) in its outer membrane. LPS is a ligand for Toll-like receptor 4 (TLR4). Based on temperature, Y. pestis synthesizes two forms of LPS: tetra- and hexa-acylated LPS (31). At 26°C, bacteria grown in the flea gut produce a hexa-acylated LPS, which elicits TLR4-mediated immune signaling to induce proinflammatory cytokines, such as TNF-α, IL-1β, IL-6, and IL-8 (31). After infection of the mammalian host, the temperature shift from 26°C to 37°C lets Y. pestis synthesize a tetra-acylated form of LPS that is inhibitory for TLR4-mediated immune signaling, prevention of macrophage and dendritic cell activation, and repression of proinflammatory cytokines (31). During early stages of infection, the bacteria are phagocytosed by macrophages and neutrophils at the site of infection. Flow cytometry data and histological evidence suggest that Y. pestis bacteria are killed by neutrophils and that neutrophils can control Y. pestis growth up to 2 days postinfection (31). However, in the phagolysosome of the macrophages, Y. pestis bacteria can survive and proliferate (32). This intracellular growth is important in Y. pestis pathogenesis, allowing the bacteria to express various virulence determinants, such as F1 antigen, V antigen, and various effector molecules secreted by the type 3 secretion system (T3SS) (27).
YopJ, one of the effector molecules of the T3SS, is an acetyltransferase and inhibits the NF-κβ and MAPK signaling pathways. Additionally, the isoform of YopJ in the KIM5 strain, YopJKIM5, has a higher affinity for IKKβ and triggers apoptosis, caspase-1 activation, and IL-1β secretion more than other isoforms of YopJ (27). Two amino acid substitutions at two sites, F177L and E206K, are present in YopJ from Y. pestis KIM5 compared to the sequence of YopJ from Y. pestis CO92, and these substitutions account for the various phenotypes seen in BMDMs (27). Once we had demonstrated differences in cytotoxicity and inflammatory cytokine levels due to YopJ isotypes by utilizing TLR2−/− BMDMs, we sought to take a broader approach to determine the changes in the immunological phenotypes present in our challenge systems.
Previous reports have demonstrated no differences in pathogenicity in TLR2−/− mice infected with Y. pestis (25, 26). Both of these studies utilized subcutaneous infection for the route of challenge. Our laboratory has investigated Y. pestis pathogenicity using a septicemic model. We feel that understanding differences in the route of infection is an important area of research due to the rapid progression of plague and the increased case-fatality rate of septicemic plague infections (34). In the United States, the case-fatality rate of septicemic plague infections is 28%, which is three times higher than that of bubonic plague (35). In this study, we sought to determine if there was a difference in pathogenicity between KIM5 and CO92 Δpgm strains of Y. pestis in TLR2−/− mice. To our knowledge, this is the first study to demonstrate a difference in pathogenicity between strains of Y. pestis and a KIM5 resistance phenotype in TLR2−/− mice. To address this issue, we challenged C57BL/6 and TLR2−/− mice with equivalent doses of KIM5 and CO92 Δpgm. Strikingly, the TLR2−/− mice challenged with KIM5 were resistant to infection, while the TLR2−/− mice infected with CO92 Δpgm succumbed to infection by day 6. All the C57BL/6 mice succumbed to infection. These results were the first indication of a difference in strain pathogenicities. We then sought to determine if the resistance phenotype was TLR2−/− dependent. We challenged TLR4−/− mice with an equivalent dose of KIM5, and the result was 100% lethality, while the TLR2−/− mice continued to be resistant. To ensure our data would also reflect the findings of previous reports, we infected TLR2−/− mice subcutaneously and observed the same phenotype previously reported. These data indicated that a septicemic model, not surprisingly, has different pathogenic kinetics but can harbor unforeseen pathogenic outcomes. We went on to determine if the site of infection was relevant in the resistance phenotype. We challenged TLR2−/− mice intravenously via the tail vein and the retro-orbital sinus. The resistance phenotype remained utilizing both challenge sites for KIM5. Changing the infection site did not alter the susceptibility of TLR2−/− mice to CO92 Δpgm. As mentioned above, a major difference between KIM5 and CO92 Δpgm is the isoform of the effector molecule YopJ. We sought to determine if the replacement of KIM5 YopJ with CO92 YopJ would restore the susceptible phenotype. We determined that KIM5-YopJCO92 restored the susceptibility of TLR2−/− mice to Y. pestis infection. We then wanted to further explore the host-pathogen interactions contributing to KIM5 resistance in TLR2−/− mice.
The kinetics in our study suggest that the difference in bacterial colonization occurs early in infection, with a significantly higher bacterial burden in the spleens and livers of TLR2−/− mice infected with CO92 Δpgm. The difference in bacterial loads in the spleen remains constant throughout infection. CO92 Δpgm demonstrates the ability to spread to the liver at a higher rate than KIM5 demonstrated at the 24-h time point, but by 3 days postinfection, this difference no longer exists. By day 4, the bacteria are nearly cleared from the spleens and livers of TLR2−/− mice infected with KIM5, but mice infected with CO92 Δpgm demonstrate uncontrolled bacterial replication in the spleens and livers. Due to the unchecked bacterial replication occurring in TLR2−/− mice infected with CO92 Δpgm, we sought to determine the immunological differences occurring.
A large fraction of the first cells to encounter Y. pestis once it enters the host are neutrophils and macrophages. Differences in cytotoxicity and cytokine production in BMDMs have previously been noted between KIM5 and CO92 Δpgm (36). We wanted to determine if the difference remained in TLR2−/− BMDMs and if the YopJ isotype could play a role. Our data suggest that KIM5 induced significantly more cytotoxicity and IL-1β and IL-1α production than CO92 Δpgm. This difference was then eliminated when TLR2−/− BMDMs were infected with KIM5-YopJCO92, indicating that the YopJ isotype can directly affect early cytokine production and whether the bacteria have the ability to remain within macrophages to multiply. Immune modulation differences were also observed in the in vivo system through immune cell phenotyping. Neutrophil infiltration was increased at 5 days postinfection with CO92 Δpgm, indicating the inability of the initial innate response to control the bacterial infection. The lack of proinflammatory cytokines detected in CO92 Δpgm-infected mice suggests enhanced immunomodulatory functionality in CO92 Δpgm. Not only do the immune cells need to traffic quickly and to the correct location to combat the invading bacteria, they need to be functional as well. The generation of reactive oxygen and nitrogen species by neutrophils to kill internalized bacteria is critical for the initial control of Y. pestis infection. Our data demonstrate that granulocytes from CO92 Δpgm-infected TLR2−/− mice have smaller amounts of oxidative stress and superoxide production 3 days postinfection. The trend continues at 5 days postinfection, with significantly smaller amounts of oxidative stress and nitric oxide and superoxide production in granulocytes from CO92 Δpgm-infected TLR2−/− mice. Dysfunctionality of neutrophils and the ability to inhibit macrophage inflammatory cytokine production would allow CO92 Δpgm to overcome many of the innate immune responses necessary to control Y. pestis infection, leading to the uncontrolled bacterial growth demonstrated in the experiments described above. Histological evidence from the livers of infected mice also demonstrated the inability of CO92 Δpgm-infected TLR2−/− mice to mount an adequate early immune response. The size and amount of early inflammatory foci in the KIM5-infected TLR2−/− mice allow early clearance of bacteria, while the later influx in CO92 Δpgm-infected mice does not allow complete clearance of bacteria.
Taken together, our results demonstrated that TLR2−/− mice are resistant to KIM5 challenge and susceptible to CO92 Δpgm in a septicemic model system. Additionally, we determined that the isotype of the effector molecule YopJ influences the pathogenic susceptibility of TLR2−/− mice, the amount of cytotoxicity, and the cytokine profile of macrophages. YopJ is not the sole reason for pathogenic differences, however; ROS and RNS were attenuated in CO92 Δpgm-infected TLR2−/− mice in a YopJ-independent manner. The development of inflammatory foci in the liver suggests that a reduced early immune response in CO92 Δpgm-infected TLR2−/− mice could also contribute to susceptibility.
MATERIALS AND METHODS
Mice.Inbred C57BL/6 (B6) mice were purchased from Jackson Laboratories (JAX stock number 000664). TLR2−/− and TLR4−/− mice were a generous gift from Joytika Sharma (University of North Dakota, Grand Forks, ND). Mice were bred in a laminar flow containment system and were maintained in a clean conventional area within the Center for Biomedical Research at the University of North Dakota. All animal studies were approved by the University of North Dakota IACUC.
Bacterial strains.The bacterial strains used were Yersinia pestis KIM5 biovar 2.MED (pCD1+, pMT1+, pPCP1+ Δpgm [37]), KIM5-YopJCO92 (pCD1Apr yopJL177F E206K pMT1+, pPCP1+ Δpgm [27]), KIM5-YopJC172A (pCD1Apr yopJC172A pMT1+, pPCP1+ Δpgm [27]), and CO92 Δpgm biovar 1.ORI (pCD1+, pMT1+, pPCP1+ Δpgm). Strains KIM5-YopJCO92, KIM5-YopJC172A, and CO92 Δpgm biovar 1.ORI were a kind gift from James B. Bliska. All strains were stored at −80°C in 25% glycerol (vol/vol). Y. pestis strains lacking the pigmentation locus are exempt from select agent guidelines.
Bacterial challenge.Female mice were infected as previously described (38). Briefly, the challenge inoculum strains were grown at 26°C with shaking overnight in heart infusion broth (HIB; BD Difco Laboratories, Sparks, MD), followed by subculture to 0.1 A620 and incubation at 26°C with shaking to an A620 of 1.0. Bacteria were centrifuged at 3,220 × g for 5 min, washed twice in sterile phosphate-buffered saline (PBS), and resuspended in PBS. The infectious dose was determined by plating triplicate serial dilutions of bacteria onto tryptose blood agar plates (TBA; Difco). The challenge dose for this study was 1 × 103 CFU. Groups of mice consisted of at least 6 mice. Mice that were challenged i.v. were infected via the retro-orbital sinus unless otherwise noted. Mice infected subcutaneously were treated intraperitoneally (i.p.) with 4 mg of iron dextran 1 day prior to infection and every alternating day thereafter, in order to recapitulate disease pathogenesis utilizing the attenuated Δpgm strains (37). Our laboratory has previously demonstrated that this dose is nontoxic compared to ferrous chloride (unpublished data). Mice were monitored for survival twice daily for 21 days.
Measurement of bacterial burden.Livers and spleens were harvested and placed in preweighed bullet blender tubes containing 1 ml of PBS. Organs were weighed and homogenized for 5 min. The resulting supernatants were serially diluted in PBS and plated onto TBA plates in triplicate to determine bacterial loads.
BMDM isolation and culture conditions.BMDMs were isolated from femurs of 5- to 8-week-old TLR2−/− female mice as previously described (33).
Macrophage infections.Twenty-four hours prior to infection, BMDMs were seeded into 24-well plates at a density of 1 × 105 cells/ml as described previously (33). Bacteria were grown as described above and then used to infect at a multiplicity of infection (MOI) of 10. After the addition of bacteria to the plate, the plates were centrifuged for 5 min at 95 × g to promote contact between the bacteria and the macrophages. After incubation at 37°C with 5% CO2 for 30 min, BMDMs were washed once with PBS. Fresh infection medium containing 8 μg/ml of gentamicin was added for 1 h at 37°C with 5% CO2 to kill extracellular bacteria. The BMDMs were washed with PBS, and infection medium was added with a lower concentration of gentamicin (4 μg/ml) for the times indicated in Fig. 4 to inhibit growth of extracellular bacteria (33).
LDH release.Cell supernatants were removed from infected BMDMs, and a Pierce LDH cytotoxicity assay kit (Thermo Scientific, Rockford, IL) was utilized to determine cytotoxicity according to the manufacturer’s instructions. Supernatants from three replicate wells per infection condition were collected and centrifuged to remove cellular debris. LDH levels in each replicate were measured in triplicate. Spontaneous LDH release from uninfected cells was measured, while total LDH release from cells treated with lysis buffer 45 min prior to the assay was measured. The values for the optical density at 490 nm (OD490) and OD680 of the nine measurements were averaged, and the percent LDH release was calculated utilizing the equation provided by the manufacturer.
Cytokine measurement.BMDM cell supernatants and serum samples collected from infected mice 3 and 5 days postinfection were analyzed for proinflammatory cytokines utilizing a Legendplex inflammation panel (Biolegend, San Diego, CA). The manufacturer’s instructions were followed utilizing a V-bottom plate for cell culture supernatant and serum samples. Briefly, 25 μl of assay buffer, 25 μl of beads, and 25 μl of either sample or standard were added to the wells and the plate was covered and set on a shaker for 2 h at 800 rpm at room temperature. Serum samples differ from the cell culture supernatant in that the standards for the serum samples need the addition of 25 μl Matrix C. The plate was centrifuged for 5 min at 250 × g, and the supernatant was decanted. The plate was washed with wash buffer, and then 25 μl of detection antibody was added. The plate was covered and set on a shaker (800 rpm) for 1 h at room temperature, and then 25 μl of streptavidin-phycoerythrin (SA-PE) antibody was added directly and the plate incubated for an additional 30 min under the same conditions. The plate was then centrifuged and decanted a second time, followed by the addition of 100 μl of wash buffer. The plate was returned to the shaker (800 rpm) for 2 min, and then the samples were read on the flow cytometer.
Immune cell phenotyping.Splenocytes were isolated by passing mouse spleens through a 70-μm nylon strainer (Falcon) using the plunger from a 5-ml syringe. Cells were washed with Hanks balanced salt solution (HBSS; Gibco), lysed for 5 min in ammonium-chloride-potassium (ACK) lysis buffer, and subsequently washed with HBSS containing 2% heat-inactivated fetal bovine serum (FBS; Atlanta Biologicals) for flow cytometry analysis. In brief, ∼1 × 106 cells were blocked using FC block (BioLegend) and stained with Tonbo Ghost Dye, and cell surface expression was interrogated with antibodies to CD45, CD3, CD4, CD8, CD69, Ly6G, and CD11b (BioLegend). Cells were analyzed using a BD Symphony A3 flow cytometer in the North Dakota Flow Cytometry and Cell Sorting Core. Data were analyzed using FlowJo (version 10.5.3).
Measuring reactive oxygen and nitrogen species.Splenocytes were isolated into single-cell suspensions as described above. We utilized Abcam’s (Cambridge, UK) cellular ROS/RNS detection assay kit (Abcam) for the detection of nitric oxide, hydrogen peroxide, peroxynitrite, hydroxyl radicals, and superoxide. The manufacturer’s suggested procedure was followed for cell suspensions, and then flow cytometry analysis was performed. Briefly, 1 × 105 splenocytes were seeded in a 24-well plate. The cells were washed and then resuspended with the 3-plex detection mixture and incubated for 2.5 h at 37°C with 5% CO2. Cells were washed and resuspended in Dulbecco modified Eagle medium (DMEM), and positive-control wells were treated to stimulate individual reactive species according to the manufacturer’s instructions. We then washed the cells once with 1× wash buffer. Cells were then transferred to a 96-well V-bottom plate in a final volume of 100 μl and analyzed by flow cytometry. The gating strategy for the separation of splenocytes is depicted in Fig. S1 in the supplemental material.
Histology.Liver samples were harvested from infected mice at 3 days and 5 days postinfection. Organs were fixed in 10% buffered formalin, sectioned, and stained with H&E. Images were obtained and were analyzed utilizing ImageJ 1.52a as previously described to measure inflammatory focus areas (39).
Statistical analysis.Differences in CFU/mg, cytotoxicity, cytokine amounts, and inflammatory focus number and area between strains and controls were analyzed using analysis of variance (ANOVA) with a Bonferroni posttest for multiple comparisons. Survival was measured using Kaplan-Meier and Mantel-Cox tests. Reactive oxygen and nitrogen species were analyzed by two-way ANOVA with Tukey’s posttest. Flow cytometry data were compared using one-way ANOVA corrected with Tukey’s posttest. GraphPad Prism (version 7.0a; GraphPad Software, Inc., San Diego, CA) was used for statistical analysis, with P values of <0.05 considered significant.
FOOTNOTES
- Received 5 October 2019.
- Returned for modification 29 October 2019.
- Accepted 16 December 2019.
- Accepted manuscript posted online 6 January 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
