(See the editorial commentary by Conlan, on pages 6–8.)
Francisella tularensis is the causative agent of tularemia and is classified as a category A biodefense agent by the Centers for Disease Control and Prevention because of its highly infectious nature. F. tularensis infects leukocytes and exhibits an extracellular phase in the blood of the host. It is unknown, however, whether F. tularensis can infect erythrocytes; thus, we examined this possibility in vivo and in vitro. In the murine model of pulmonary type A tularemia, we showed the presence of intraerythrocytic bacteria by double-immunofluorescence microscopy and ex vivo gentamicin protection of the purified erythrocyte fraction. In vitro, F. tularensis invaded human erythrocytes, as shown in the gentamicin protection assays, double-immunofluorescence microscopy, flow cytometry, scanning electron microscopy, and transmission electron microscopy with immunogold labeling of the bacteria. Additional in vitro tests indicated that serum complement-dependent and complement-independent mechanisms contribute to erythrocyte invasion. Our results reveal a novel intraerythrocytic phase during F. tularensis infection.
Francisella tularensis is a highly infectious microorganism; <10 bacteria can result in the severe disease tularemia . This bacterium has been weaponized and could be used for bioterrorism, prompting the Center for Disease Control and Prevention (CDC) to classify F. tularensis as a category A biodefense agent . In addition, F. tularensis causes a variety of naturally occurring human infections that can be acquired by inhalation, arthropod bites, oropharyngeal exposure, or contact [3–5].
The virulence of F. tularensis has principally been associated with this organism’s ability to replicate in phagocytic cells of the innate immune system, such as macrophages [6, 7]. However, the network of host-microbe interactions that contribute to Francisella pathogenesis is much more complex [8–10]. F. tularensis exhibits an extracellular phase in the blood during mouse infection , an environment where this pathogen likely uses its ability to resist complement [12, 13]. In addition to macrophages, F. tularensis can invade and replicate in a range of nonphagocytic host cells, such as alveolar epithelial cells, kidney epithelial cells, hepatocytes, and fibroblasts [9, 14–19]. Recent work has shown that replication in nonmacrophages is sufficient for F. tularensis pathogenesis, emphasizing the importance of these other cell types during infection . Uptake of F. tularensis in both macrophages and nonphagocytes is mediated by the host cell’s endocytic machinery [16, 20]. It is unknown, however, whether this pathogen invades cells incapable of endocytosis, such as erythrocytes .
Erythrocytes are the most abundant cell type in the blood of mammals and are the primary means by which oxygen is transported throughout the body [22, 23]. In addition, these cells hold most of the body’s iron . Erythrocytes are parasitized by Plasmodium falciparum and other protozoa, which enhances evasion of host immune defenses and transmission by mosquito vectors . In addition, the bacterial pathogens Bartonella species, Anaplasma marginale, Mycoplasma suis, and M. gallisepticum establish intraerythrocytic infections [26–29].
In this study, we provide evidence that erythrocyte invasion by F. tularensis is a feature of tularemia. Here, we report that F. tularensis invades erythrocytes in vivo during infection of mice, and this phenomenon was modeled in vitro with use of human erythrocytes. Mechanistic studies indicate that complement and heat-stable components of serum contribute to erythrocyte invasion.
MATERIALS AND METHODS
F. tularensis holarctica live vaccine strain (LVS) was provided by Dr. Karen Elkins of the US Food and Drug Administration. Listeria monocytogenes, strain EGD was provided by Doug Drevets, University of Oklahoma. F. tularensis, strain SCHU S4 (FSC237), NR-643 was obtained through the National Institutes of Health (NIH) Biodefense and Emerging Infections Research Resources Repository, National Institute of Allergy and Infectious Diseases, NIH. Frozen stock cultures of bacteria were streaked onto chocolate II agar plates and incubated at 37°C with 5% carbon dioxide for 1–3 days (F. tularensis) or overnight (L. monocytogenes). Liquid cultures, inoculated with bacteria grown on chocolate II agar, were generated in BBL Trypticase Soy Broth (BD Biosciences) supplemented with 0.1% cysteine hydrochloride incubated at 37°C with agitation to stationary phase. All work with F. tularensis Schu S4 was conducted under BSL-3 conditions at the University of Pittsburgh with approval from the CDC Select Agent Program.
Human erythrocytes were obtained from donated buffy coats (Pittsburgh Central Blood Bank), with approval of the institutional review board. Plasma was removed by centrifugation at 200 x g for 10 minutes. The pellet was suspended in phosphate-buffered saline (PBS) and was separated by density gradient centrifugation using Ficoll . Erythrocytes were washed and stored in Hanks Balanced Salts Solution (HBSS; Gibco) at 4°C. Erythrocyte populations were >99.9% pure based on microscopy and flow cytometry. Before infection, erythrocytes were washed, counted on a hemocytometer, and suspended to the desired concentration in complete medium: either Dulbecco’s Modified Eagle Medium (DMEM) or McCoy’s 5A Medium (M5A), both supplemented with 2 mM glutamine, 25 mM HEPES, and 10% human AB serum (Gemini).
Intratracheal Mouse Infection
All experiments using mice were approved by the University of Pittsburgh’s Institutional Animal Care and Use Committee. C57BL/6J mice (Jackson Laboratory; female, 6–8 weeks of age) were housed under ABSL-3 conditions. These mice were infected intratracheally via oropharyngeal instillation with ∼102 colony-forming units (CFU) F. tularensis Schu S4 . Blood samples, which were obtained in a heparinized syringe via cardiac puncture, were either diluted and plated on chocolate II agar to enumerate CFUs, used to generate thin smears for Wright-Giemsa staining (LeukoStat; Fisher Scientific) or fractionated for ex vivo gentamicin protection assays.
Gentamicin Protection Assays and Intravenous Infections
Blood samples were obtained from intratracheally infected mice 4 days after infection. Leukocytes and erythrocytes were separated from the plasma by centrifugation at 200 × g. The cell fraction was suspended in complete DMEM containing 10% fetal bovine serum with gentamicin (100 μg/mL; mock treatment lacked gentamicin) and incubated at 37°C, 5% carbon dioxide for 1 hour to kill extracellular bacteria. Erythrocytes and leukocytes were then separated using a Ficoll gradient . To enumerate intraerythrocytic CFUs, each fraction was washed with PBS, lysed in 0.02% sodium dodecyl sulfate (SDS), and diluted and plated. For transfer experiments, erythrocytes were washed and suspended in PBS to the original volume of blood samples. Naive mice were infected intravenously by tail vein injection with 100 μL of these erythrocytes. CFUs administered were 2.4 × 106 for the mock-treated erythrocytes and 2.6 × 103 for the gentamicin-treated group. Additional mice received intravenous delivery of F. tularensis Schu S4 prepared from broth culture  (1.5 × 102 CFU) or uninfected erythrocytes from naive mice. Mice were monitored for morbidity and mortality .
For the in vitro gentamicin protection assays , erythrocytes were suspended in complete DMEM or complete M5A. We tested a range of erythrocyte concentrations for invasion by F. tularensis with varying multiplicities of infection (MOI; data not shown). We found that a concentration of 3 × 105 cells per well in a 96-well V-bottom microtiter dish with a MOI of 50–100 yielded consistent invasion data, and therefore, we used these values throughout this study. After incubation at 37°C with 5% carbon dioxide for 2–3 hour, the plate was centrifuged at 400 × g to collect erythrocytes and reduce the number of extracellular bacteria. The erythrocytes were resuspended in gentamicin (100 μg/mL) for 1 hour at 37°C with 5% carbon dioxide to kill remaining extracellular bacteria. The cells were pelleted and washed with complete medium and pelleted again, and the erythrocytes were lysed in 0.02% SDS. This suspension was diluted and plated to enumerate CFUs. The prelysed control was treated with SDS before gentamicin treatment to show that antibiotic levels were sufficient for bacterial killing. Additional controls included wells containing bacteria but not erythrocytes.
Double-Immunofluorescence Microscopy (DIFM)
This method can distinguish between intra- and extracellular bacteria . To analyze in vivo samples, plasma was removed from the heparinized mouse blood by centrifugation, and the cell pellet was suspended in fixative (2.5% glutaraldehyde or 2% paraformaldehyde plus 0.1% glutaraldehyde in PBS). After 1 hour of incubation at room temperature, the cells were suspended in PBS, diluted, and added to chamber slides (LAB-TEK). Additional fixative was used to link the cells to the slide’s surface. After 3 washes with PBS, the wells were blocked with 2.5% bovine serum albumin (BSA) for 30 minutes. Subsequently, the cells were probed with the primary antibody (polyclonal rabbit anti–F. tularensis; BD Biosciences); phycoerythrin-conjugated rat anti-mouse TER119 (Ly76, erythroid marker; eBioscience) in 2.5% BSA overnight at 4°C. After 3 washes in PBS with 0.2% Tween 20, the slides were probed with the secondary antibodies (Alexa Fluor 350 donkey anti-rabbit IgG and Alexa Fluor 555 goat anti-rat IgG) for at least 1 hours at room temperature. After 3 washes with 0.2% Tween 20 in PBS, the erythrocytes were treated with 0.1% Triton X-100 for 10 seconds to permeabilize. The slide was washed in PBS and blocked with 2.5% BSA for 30 minutes at room temperature followed by treatment with rabbit anti–F. tularensis for at least 1 hour. After 3 washes with 0.2% Tween 20 in PBS, the cells were probed with Alexa 488 donkey anti-rabbit for 1 hours. The cells were washed 3 times with 0.2% Tween 20 in PBS, PBS alone, and then mounted in ProLong Gold antifade reagent (Invitrogen).
The DIFM of in vitro assays was performed similarly as described for the in vivo. After the coincubation of bacteria and human erythrocytes, the cells were washed and treated as we described above, except the primary antibody mixture was rabbit anti–F. tularensis and PE-conjugated mouse anti-human glycophorin A (CD235a). The secondary antibody mixture was Alexa Fluor 350 donkey anti-rabbit IgG and Alexa Fluor 555 donkey anti-mouse IgG.
A Zeiss Axiovert 200 microscope was used for the DIFM. When using the high-power 63 × Neofluar oil immersion objective (Zeiss), an ApoTome was used to enhance contrast and increase axial resolution. AxioVision software was used to view the captured images and to make modifications in brightness and contrast uniformly across all images. Adobe Photoshop was used to pseudocolor channels to allow for green-red merge to yellow.
Bacteria were treated with Syto 9 (Invitrogen) and coincubated with erythrocytes for 2 hour at 37°C with 5% carbon dioxide. After washing, erythrocytes were treated with Fc blocking reagent and probed with PE-conjugated mouse anti-human glycophorin A. After 3 washes with 1% BSA in PBS, the cells were analyzed by flow cytometry using a BD FacsAria by Timothy Sturgeon at the University of Pittsburgh Center for Vaccine Research.
Scanning Electron Microscopy (SEM)
Human erythrocytes and F. tularensis were cocultured as before, washed extensively, and added to 12-mm round glass coverslips, then fixed in 2.5% glutaraldehyde in PBS at 4°C. Samples were washed 3 times in PBS, postfixed for 1 hour in aqueous 1% osmium tetroxide, and washed 3 times in PBS. Cells were dehydrated through a graded ethanol series, washed 3 times with absolute ethanol, chemically dehydrated using hexamethyldisilazane, and air-dried. Dried cells were mounted onto aluminum stubs, grounded with silver paint, and then sputter coated with 3.5 nm gold/palladium. Samples were viewed in a JEOL JSM-6330F scanning electron microscope at 3 kV.
Transmission Electron Microscopy (TEM) With Immunogold Labeling
Human erythrocytes and F. tularensis Schu S4 were cocultured as previously described; however, we extended the incubation time to ∼4 hours to increase the likelihood of observing intraerythrocytic bacteria. Subsequently, cells were washed and fixed in 2% paraformaldehyde plus 0.1% glutaraldehyde in PBS at room temperature for 1 hour. After the cells were washed with HBSS, they were stored at 4°C for 1 hour in this buffer. Cells were pelleted and resuspended in a small amount of 3% gelatin in PBS, solidified at 4°C, and fixed for an additional 15 minutes in the same fixative. Gelatin-cell block was cryoprotected in PVP cryoprotectant overnight at 4°C (25% polyvinylpyrrolidone, 2.3 M sucrose, 0.055M Na2CO3; pH, 7.4). Erythrocyte blocks were frozen on ultracryotome stubs under liquid nitrogen and stored in liquid nitrogen until use. Ultrathin sections (70–100 nm) were cut using a Reichert Ultracut UC7 ultramicrotome with cryo-attachment, lifted on a small drop of 2.3 M sucrose, and mounted on Formvar-coated copper grids. Sections were washed 3 times with PBS and 3 times with PBS containing 0.5% bovine serum albumin and 0.15% glycine (PBG buffer), followed by 30 minutes blocking with 5% normal goat serum in PBG. Sections were labeled with rabbit anti–F. tularensis (1:500) in PBG for 1 hour. Sections were washed 4 times in PBG and labeled with goat anti-rabbit (5 nm) gold conjugated secondary antibodies (1:25; Amersham) for 1 hour. Sections were washed 3 times in PBG, and 3 times in PBS, then fixed in 2.5% glutaraldehyde in PBS for 5 minutes, washed 2 times in PBS, and washed 6 times in ddH2O. Sections were poststained in 2% neutral uranyl acetate for 7 minutes, washed 3 times in ddH2O, stained for 2 minutes in 4% uranyl acetate, and embedded in 1.25% methyl cellulose. Labeling was observed on a JEOL JEM 1011 electron microscope at 80 kV fitted with a side mount AMT 2k digital camera.
In Vivo Erythrocyte Invasion
Mice were infected intratracheally with ∼102 CFU F. tularensis Schu S4 as a model for inhalational tularemia [19, 32]. We detected bacteria in blood samples from mice 2 days after inoculation (Figure 1A), similar to our previous work . To visualize the F. tularensis bacteremia, blood smears from infected mice were subjected to Wright-Giemsa staining. Four days after infection, we observed darkly stained bodies, consistent with the size and shape of F. tularensis that appeared to be both extra- and intraerythrocytic (Figure 1B).
Erythrocyte invasion by F. tularensis has not been reported previously. To examine this possibility further, we analyzed blood samples from infected mice by DIFM (Figure 2A) . This method can distinguish between intra- and extracellular bacteria by successive incubation of red blood cells with antibodies before and after permeabilization of the erythrocyte membrane . Bacteria that stained with only one color by DIFM clearly showed intraerythrocytic F. tularensis Schu S4 in blood samples from infected mice (Figure 2A). We also observed bacteria on the surface of erythrocytes, as indicated by fluorescence staining before and after permeabilization (Figure 2B). Neither uninfected erythrocytes nor infected cells treated with isotype control antibodies showed fluorescence (data not shown), validating the specificity of the staining in Figure 2A and 2B.
To confirm the microscopy data, we conducted a gentamicin protection assay on the erythrocyte fraction of blood from infected mice. In this assay, intracellular bacteria are protected from the gentamicin treatment because aminoglycosides do not penetrate host cell membranes . Blood samples were obtained at 4 days after infection from mice that had been infected intratracheally with F. tularensis Schu S4. Plasma was removed by centrifugation, and cells were incubated in completed DMEM with 10% BSA and gentamicin at 100 times the 90% minimum inhibitory concentration for Francisella species . Erythrocytes and leukocytes were then separated , and each fraction was washed, lysed, and diluted and plated to enumerate CFUs. Using this ex vivo gentamicin protection assay, we quantified 2 × 104 intraerythrocytic bacteria per mL blood in the infected mice. The intracellular CFUs in circulating erythrocytes versus leukocytes were not statistically different (data not shown; P = .15 and P = .24 for 2 experiments). Transfer of these infected erythrocytes (intravenously) to naive mice resulted in rapid morbidity (Figure 1C), indicating that intraerythrocytic bacteria maintain virulence.
In Vitro Erythrocyte Invasion
We next studied human erythrocyte invasion in vitro with use of F. tularensis Schu S4 and LVS. We first conducted DIFM after bacteria were cocultivated with human erythrocytes. In some instances, bacteria stained before and after permeabilization, showing that these bacteria were outside the erythrocytes (Figure 2C and 2D). We also detected F. tularensis Schu S4 (Figure 2C) and LVS (Figure 2D) only after permeabilization of the erythrocyte, indicating that these bacteria were intracellular. Cells treated with isotype control antibodies did not stain, verifying the antibody specificity (data not shown).
Using a gentamicin protection assay , we also consistently observed invasion of human erythrocytes by both LVS and Schu S4 (Figure 3A). L. monocytogenes (Figure 3A), an intracellular pathogen that is not believed to invade erythrocytes, and E. coli (data not shown) were not protected from the gentamicin treatment. To confirm that the gentamicin was active in killing the extracellular Francisella species, we lysed erythrocytes before antibiotic treatment and saw no significant survival (Figure 3A). In addition, bacteria did not survive gentamicin treatment in wells that excluded erythrocytes (Figure 3B).
To determine whether F. tularensis replicated in erythrocytes, we replenished the medium after gentamicin treatment. The intraerythrocytic growth of F. tularensis was limited (Figure 3C), a finding consistent with Bartonella species and the parasitic protozoan Theileria parva [34, 35]. Although not replicating, we observed that F. tularensis persisted in erythrocytes in vitro after extended incubation times following gentamicin treatment (72 hours; data not shown), suggesting that red blood cells may be a long-term intracellular niche.
We further analyzed cocultures of human erythrocytes and F. tularensis LVS with use of flow cytometry. Before infection, F. tularensis LVS or L. monocytogenes were treated with the green fluorescent vital stain, Syto 9. After 2 hours coincubation, cells were washed and probed with PE-conjugated anti–glycophorin A. This resulted in a complete shift in red (PE) fluorescence of the erythrocyte population, as expected (Figure 4A and data not shown). Erythrocytes that were cocultured with F. tularensis LVS but not with L. monocytogenes had a subpopulation of red blood cells with green (FITC) fluorescence (Figure 4B and C) despite equal labeling of the bacteria with Syto 9. Green fluorescence was not detected in uninfected erythrocytes (Figure 4A). These data indicated that F. tularensis, not L. monocytogenes, associated with erythrocytes in this assay—results consistent with erythrocyte invasion.
We also evaluated human erythrocyte invasion by SEM (Figure 5). Uninfected erythrocytes showed a smooth contour with use of this technique (Figure 5A). After cocultivation with F. tularensis Schu S4, we observed that bacteria were closely associated with erythrocytes (Figure 5B) or partially embedded in the erythrocyte membrane (Figure 5C).
To confirm the presence of intraerythrocytic F. tularensis, we used TEM with immunogold staining (Figure 6). After incubation of F. tularensis Schu S4 with human erythrocytes, cell suspensions were fixed and prepared for sectioning. Ultrathin sections were probed with rabbit anti–F. tularensis, followed by anti-rabbit secondary antibodies labeled with 5-nm gold particles. Sections treated with the primary and secondary antibodies, but not secondary alone, showed the presence of gold-staining bacteria in the cytosol delineated by the bacterium’s inner and outer membranes and an irregularly staining surface , consistent with its capsule (Figure 6 and data not shown). Together, results of the in vitro gentamicin protection assays, flow cytometry, DIFM, SEM, and TEM confirmed the in vivo data and established that F. tularensis invades erythrocytes (Figures 2–6).
Involvement of Complement in Erythrocyte Invasion by F. tularensis
Francisella inactivates C3b to C3bi and deposits this substance on its surface . We therefore hypothesized that C3bi may interact with CR1 (CD35), the erythrocyte complement receptor, to facilitate a bacterial-host cell interaction. To test whether serum complement was involved in erythrocyte invasion, F. tularensis Schu S4 and LVS were incubated in media with 1% or 10% normal human serum or 10% heat-inactivated serum before cocultivation with erythrocytes (Figure 7). Higher serum concentration enhanced invasion by both F. tularensis strains (Figure 7). Reduced invasion was seen after incubation in heat-inactivated serum (Figure 7). However, heat inactivation did not abolish invasion, because intraerythrocytic Francisella species were still detected in this treatment group (Figure 7). Equivalent numbers of erythrocytes were recovered under all serum conditions after incubation with bacteria, indicating that the data in Figure 7 could not be attributed to differences in red blood cell count. This suggests that both complement and heat-stabile components of serum contribute to erythrocyte invasion.
Using in vivo and in vitro approaches, we present evidence that F. tularensis invades erythrocytes—a novel feature of tularemia. The TEM revealed that F. tularensis exists directly in the cytosol inside the erythrocyte, similar to part of its lifecycle in macrophages. Because iron concentration modulates virulence factor expression of this organism , it is likely that the high content of hemoglobin in the erythrocyte influences similar changes, some of which may adapt F. tularensis to this environment.
Relapse after antibiotic treatment is well documented for tularemia, ranging from 6% for gentamicin to 21% for chloramphenicol . Duration of gentamicin therapy is a factor in determining successful treatment . Because erythrocytes have a long lifespan of ∼120 days , it is feasible that inhabiting erythrocyte cytoplasm contributes to persistence. This is consistent with our data that F. tularensis does not replicate optimally in the cytosol of the erythrocyte, suggesting that the bacterium may be dormant in this environment. Moreover, erythrocytes are unlikely to concentrate aminoglycosides in their cytoplasm, because mature, adult erythrocytes lack pinocytosis exhibited by leukocytes [40, 41]. We presented evidence that infected erythrocytes treated with gentamicin produce rapid morbidity when transferred to naive mice intravenously, indicating that intraerythrocytic bacteria maintain virulence. Persistence in an erythrocyte could therefore facilitate F. tularensis relapse after inadequate antimicrobial treatment, such as an abbreviated gentamicin regimen. In addition to persistence, erythrocyte invasion may enhance Francisella dissemination and provide asylum from the immune response, thereby contributing to pathogenesis.
The results of our gentamicin protection assays suggest a mechanistic model of invasion involving serum complement. On the basis of these data and our understanding that Francisella species inactivates C3b to C3bi and deposits this substance on its surface , we propose that the surface-deposited C3bi interacts with the erythrocyte complement receptor to facilitate an initial bacteria-host cell interaction. This initial interaction is unlikely to result in entry, because adult erythrocytes are defective in receptor-mediated endocytosis . Therefore, other bacterial and host molecules are likely to be responsible for entry in the erythrocyte, including heat-stabile, non-complement serum components.
Entering erythrocytes may provide an additional benefit to F. tularensis beyond infections of mammals. Previous studies have shown that invasion of erythrocytes enhances arthropod transmission of Bartonella species and T. parva. An intraerythrocytic phase may protect these pathogens from digestive enzymes, antibacterial peptides, and oxidative stress [42, 43] after entering the arthropod gut [34, 35]. It is therefore plausible that accessing the intracellular space of circulating blood cells may enhance F. tularensis survival in the tick midgut and subsequent transmission to mammals.
This work was supported by Immunology of Infectious Disease, National Institute of Allergy and Infectious Diseases. (T32 AI060525 to J. H.)