OpiA, a Type Six Secretion System Substrate, Localizes to the Cell Pole and Plays a Role in Bacterial Growth and Viability in Francisella tularensis LVS

F. tularensis is a pathogenic intracellular pathogen that is of importance for public health and strategic defense. This study characterizes the opiA gene of F. tularensis LVS, an attenuated strain that has been used as a live vaccine but that also shares significant genetic similarity to related Francisella strains that cause human disease. The data presented here provide the first evidence of a T6SS effector protein that affects the physiology of F. tularensis, namely, the growth, cell size, viability, and aminoglycoside resistance of F. tularensis LVS. This study also adds insight into our understanding of OpiA as a determinant of virulence. Finally, the fluorescence fusion constructs presented here will be useful tools for dissecting the role of OpiA in infection.

ABSTRACT

Francisella tularensis is an intracellular pathogen and the causative agent of tularemia. The F. tularensis type six secretion system (T6SS) is required for a number of host-pathogen interactions, including phagolysosomal escape and invasion of erythrocytes. One known effector of the T6SS, OpiA, has recently been shown to be a phosphatidylinositol-3 kinase. To investigate the role of OpiA in erythrocyte invasion, we constructed an opiA-null mutant in the live vaccine strain, F. tularensis LVS. OpiA was not required for erythrocyte invasion; however, deletion of opiA affected growth of F. tularensis LVS in broth cultures in a medium-dependent manner. We also found that opiA influenced cell size, gentamicin sensitivity, bacterial viability, and the lipid content of F. tularensis. A fluorescently tagged OpiA (OpiA–emerald-green fluorescent protein [EmGFP]) accumulated at the cell poles of F. tularensis, which is consistent with the location of the T6SS. However, OpiA-EmGFP also exhibited a highly dynamic localization, and this fusion protein was detected in erythrocytes and THP-1 cells in vitro, further supporting that OpiA is secreted. Similar to previous reports with F. novicida, our data demonstrated that opiA had a minimal effect on intracellular replication of F. tularensis in host immune cells in vitro. However, THP-1 cells infected with the opiA mutant produced modestly (but significantly) higher levels of the proinflammatory cytokine tumor necrosis factor alpha compared to these host cells infected with wild-type bacteria. We conclude that, in addition to its role in host-pathogen interactions, our results reveal that the function of opiA is central to the biology of F. tularensis bacteria.

IMPORTANCE F. tularensis is a pathogenic intracellular pathogen that is of importance for public health and strategic defense. This study characterizes the opiA gene of F. tularensis LVS, an attenuated strain that has been used as a live vaccine but that also shares significant genetic similarity to related Francisella strains that cause human disease. The data presented here provide the first evidence of a T6SS effector protein that affects the physiology of F. tularensis, namely, the growth, cell size, viability, and aminoglycoside resistance of F. tularensis LVS. This study also adds insight into our understanding of OpiA as a determinant of virulence. Finally, the fluorescence fusion constructs presented here will be useful tools for dissecting the role of OpiA in infection.

INTRODUCTION

Francisella tularensis is a Gram-negative, intracellular pathogen and is the causative agent of the potentially lethal disease tularemia (1, 2). Due to its low infectious dose and the mortality rate associated with the pneumonic form of this disease, F. tularensis has been classified by the Centers for Disease Control and Prevention as a category A biodefense agent (1, 3). In addition, F. tularensis subspecies occur naturally and cause a variety of human infections (4). Human infection usually occurs from exposure to contaminated animals or from blood-sucking arthropods (1, 2, 4). Francisella spp. have a broad host range and have been identified in numerous mammalian hosts (1, 5) and from birds and fish (6). Various arthropod species, such as biting flies and ticks in North America and ticks and mosquitoes in Eurasia, are recognized as significant vectors of the transmission of tularemia (6, 7). Free-living amoebae have also been identified as potential environmental reservoirs of F. tularensis (8). However, the ecology and natural history of F. tularensis, in particular the sources of environmental persistence and processes that drive infectious outbreaks, are not well understood. Moreover, there is growing concern over the reemergence of tularemia, with an increase in reported outbreaks in the last two decades, particularly in Europe (5, 9–13). Environmental change and the encroachment of human settlement into wilderness areas may continue to exacerbate this trend (11, 14, 15).

F. tularensis can infect a broad range of host cells, including lung alveolar cells, kidney cells, hepatocytes, fibroblasts, and erythrocytes (16–23). Moreover, F. tularensis can survive and replicate within host phagocytes, including macrophages, wherein the bacteria block maturation of the phagosome and subsequently escape this compartment (24–27). Critical to this, and therefore to pathogenicity, is the Francisella type six secretion system (T6SS) (reviewed in reference 28). The T6SS is a contractile injection molecular machine that is produced by many Gram-negative bacteria and is required for the secretion of effector proteins across the inner and outer membranes into a variety of target cells, including other bacteria or eukaryotic hosts. However, the T6SS of Francisella is distinct from those found in other bacteria, representing a noncanonical group (T6SSⁱⁱ) unique to Francisella spp. (29, 30). The Francisella T6SSⁱⁱ is required specifically for phagosomal escape, intracellular replication, and pathogenicity in host organisms (29–33). The protein machinery of this secretion system is encoded by a cluster of genes in the Francisella pathogenicity island (FPI) (reviewed in reference 34) which is part of the regulon of MglA (35), a master regulator of genes required for virulence and intracellular growth (36).

The T6SSⁱⁱ of F. tularensis has been shown to be required for invasion of erythrocytes (21). However, F. tularensis does not appear to replicate within red blood cells (RBCs). Our laboratory is interested in understanding the mechanisms that allow F. tularensis to invade and persist within erythrocytes since it is likely that erythrocyte invasion increases the colonization of ticks following a blood meal, a phenomenon that could contribute to the transmission of F. tularensis (22). To learn more about the role of the T6SSⁱⁱ in erythrocyte invasion, we have turned our attention to the effector proteins of this apparatus that may facilitate invasion.

Recently, a proteomic approach was undertaken to identify substrates of the Francisella T6SSⁱⁱ system, identifying eight proteins that are substrates of the T6SSⁱⁱ in F. novicida (37). While five of these proteins were encoded within the FPI, three substrates were encoded outside the FPI, and one of these, OpiA, has been shown to be a phosphatidylinositol 3-kinase, which generates phosphotidylinositol-3-phosphate [PI(3)P] on late-endosome-like Francisella containing phagosomes (FCPs) facilitating the escape of F. novicida into the cytoplasm of infected host cells (38).

In this study, we present a characterization of opiA in F. tularensis LVS, an attenuated strain that can be worked with safely in BSL2 conditions but retains some pathogenicity in animal models (39–41). We did not find any evidence to suggest that opiA was required for RBC invasion; however, we present evidence that shows that opiA is involved in the growth and morphology of F. tularensis LVS during bacterial cultivation. We also show that opiA-null mutant bacteria stimulate higher levels of the proinflammatory cytokine, tumor necrosis factor alpha (TNF-α), from THP-1 monocytes compared to wild-type bacteria. In addition, data presented here suggest that opiA-null mutant bacteria are slightly (yet significantly) attenuated for pathogenesis in a chicken embryo tularemia model, consistent with previous reports that this gene plays a role during infection. Using a fusion of the fluorescent protein emerald-green fluorescent protein (EmGFP) to OpiA, we have localized OpiA in both broth- and agar-grown cultures of F. tularensis LVS and have shown that OpiA-EmGFP can be detected in host cells during in vitro cell infection assays, further supporting that this protein is secreted.

RESULTS

OpiA is not required for erythrocyte invasion.

Our laboratory has previously elucidated the requirement of the T6SSⁱⁱ in the invasion of erythrocytes by F. tularensis (21). Expression of mglA is upregulated in Francisella incubated with human erythrocytes and both dotU and iglC (both part of the mglA regulon and genes encoding structural components of the T6SSⁱⁱ) are required for invasion of erythrocytes by F. tularensis (21). Since OpiA is an effector of this secretion system, we generated a F. tularensis LVS opiA-null mutant (see Fig. S1A and B in the supplemental material) to determine the role of this gene in erythrocyte invasion. However, when we performed standard gentamicin protection assays, we observed confounding results that suggested that the ΔopiA strain may have altered viability and sensitivity to antibiotics under the conditions tested (data not shown). Specifically, the ΔopiA strain was resistant to gentamicin at the level and duration used in conventional erythrocyte invasion assays (21). Therefore, in order to determine whether opiA is required for erythrocyte invasion, we utilized fluorescence microscopy as an alternative to the gentamicin protection assay. Erythrocytes were incubated together for 24 h with either wild-type or ΔopiA F. tularensis LVS strains containing a plasmid that expressed emgfp from a constitutively active promoter at a multiplicity of infection (MOI) of 12.5. For both strains, the frequency of invasion was quite low, with only ∼2% of cells containing bacteria (Fig. 1A). However, this is consistent with previous observations of invasion and with the fact that F. tularensis LVS can invade but does not replicate within RBCs (21, 22). The fluorescently labeled ΔopiA strain was consistently observed invading RBCs at the same rate as wild-type bacteria, and similar results were obtained when DAPI (4′,6′-diamidino-2-phenylindole) staining or double-immunofluorescence microscopy (DIFM) (24) were used to assay invasion (data not shown). We conclude therefore that, while the T6SSⁱⁱ is required for RBC invasion (21), opiA does not play a role and that the requirement of the T6SSⁱⁱ for invasion is mediated by other effectors or via another mechanism altogether.

FIG 1.

In vitro and in vivo analysis of the effect of the opiA deletion on erythrocyte invasion, intracellular replication, and pathogenicity. (A) Deletion of opiA does not affect the ability of F. tularensis LVS to invade erythrocytes. The left panel shows fluorescently labeled wild-type and ΔopiA strains invading human erythrocytes. Scale bar, 5 μm. In the right panel, the frequency of invasion of erythrocytes was not significantly different between the wild-type and mutant strains (∼2% for both). For each strain, five fields of view were scored, and the means ± the standard errors (SE), along with the total number of scored erythrocytes, are shown. (B) Quantification of intracellular replication of F. tularensis LVS strains in host cells, THP-1 (left) and RAW 264.7 (right), by gentamicin protection assay. The means ± the SE from three independent experiments are shown. Deletion of opiA had no significant effect on intracellular growth in either THP-1 or RAW 264.7 cells compared to either the wild-type or complemented strains (as determined by two-way ANOVA analysis). (C) Deletion of opiA affects TNF-α production in THP- 1 cells in an in vitro infection assay. THP-1 cells were incubated for 24 h with the strains indicated. Cells incubated without bacteria were used as a negative control. Means ± the SE of three experiments are shown. The difference between the means for the wild type and the ΔopiA mutant was significant, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.0001). (D) The ΔopiA mutant is attenuated in a chicken embryo infection assay. Kaplan-Meier survival curves are shown for eggs infected with 1 × 10⁶ wild-type F. tularensis LVS cells (n = 51) or the ΔopiA mutant (n = 52). The curves represent the sum from five individual experiments, and the ΔopiA mutant showed a modest but statistically significant attenuation compared to the wild type (P = 0.390 using a log rank [Mantel-Cox] test).

OpiA is not required for replication within host cells in vitro.

The T6SSⁱⁱ is required for cytoplasmic replication and virulence for Francisella, including pathogenic species and F. tularensis LVS (29, 31–33,) and the recent description of opiA from F. novicida showed that it plays a role in infection by facilitating phagosomal escape from host cells (38). To test whether opiA in F. tularensis LVS had any effect on infection and intracellular replication, we used gentamicin protection assays using either THP-1 leukemic monocytes (42) or RAW 264.7 murine macrophages (43). Wild-type and ΔopiA strains grown overnight in Chamberlain’s defined medium (CDM) were incubated with host cells at an MOI of 500 for 3 h and then treated for 1 h with 100 μg ml⁻¹ gentamicin to kill any bacteria that had not been phagocytosed. The infected cells were spun down and resuspended in THP-1 growth medium to remove the gentamicin and any extracellular bacteria. Host cells were lysed at 4, 24, and 48 h postinfection, and the numbers of bacteria were enumerated. For both host cell types tested, no significant difference was observed between the wild-type and ΔopiA strains (Fig. 1B).

THP-1 cells incubated with an opiA deletion mutant produce higher levels of TNF-α.

Evasion of the innate host cell immunity is important for Francisella infection. Both gamma interferon (IFN-γ) and tumor necrosis factor alpha (TNF-α) are required to block intracellular replication of Francisella (44, 45), and Francisella inhibits TNF-α production during infection (20, 46). We wondered whether opiA played any role in evasion of innate immunity in host cells and, we therefore performed enzyme-linked immunosorbent assays (ELISAs) to measure TNF-α in THP-1 and RAW 264.7 cells after incubation with wild-type and ΔopiA strains. Bacteria were grown overnight in CDM and incubated with host cells at an MOI of 10. Supernatants were sampled after 2 and 24 h and assayed for TNF-α by ELISA. Consistently, higher levels of TNF-α were seen in supernatants from THP-1 cells incubated with the ΔopiA strain for 24 h (P < 0.0001 determined by two-way analysis of variance (ANOVA) with Dunnett’s multiple-comparison test) (Fig. 1C), suggesting that opiA may be involved in restraining the production of the proinflammatory cytokine, TNF-α, during infection. There was also a trend toward more TNF-α in supernatants from RAW 264.7 cells at the same time point; however, the difference was not significant (data not shown). Despite the increase of TNF-α in THP-1 cells during infection with the opiA mutant, deletion of opiA was not accompanied by a significant impact on intracellular proliferation of F. tularensis LVS (Fig. 1B).

opiA contributes to pathogenesis during infection of chicken embryos.

To determine whether deletion of opiA affects pathogenicity in vivo, we used a chicken embryo infection model (47). In total, five separate experiments were conducted using between 7 and 21 eggs per experiment. For each experiment, wild-type and ΔopiA strains were grown in tryptic soy broth supplemented with 0.1% cysteine (TSBc) to an optical density at 600 nm (OD₆₀₀) of ∼1.0 and were then diluted in phosphate-buffered saline (PBS) for inoculation at a dose of 5 × 10⁵ bacteria per egg. In four of five experiments, there was a slight yet significant difference between the wild type and the opiA mutant. When the results from all five experiments are pooled, with a total of 52 eggs for the opiA mutant and 51 eggs for the wild-type condition, there was a modest but significant attenuation of pathogenicity in the ΔopiA strain (P = 0.039, as determined by log rank [Mantel-Cox] test) (Fig. 1D). These data suggest that opiA contributes to the pathogenesis of F. tularensis LVS.

The opiA deletion mutant exhibits a medium-dependent growth defect.

Our investigation did not reveal any role for opiA in either RBC invasion or host cell replication, and the deletion of opiA had only a mild influence on pathogenicity in a chicken embryo infection model. These results are consistent with observations in the closely related organism, F. novicida, for which opiA influences intracellular growth during in vitro infection assays; however, this effect was seen in the context of infection and only when the opiA deletion was combined with a deletion of pdpC, which encodes another T6SSⁱⁱ effector protein (38). Given that opiA does not have a strong effect on host-cell interactions, we wanted to investigate the role of opiA in the bacterial physiology of F. tularensis LVS.

To determine the effect of the deletion opiA on in vitro growth of F. tularensis LVS, stationary-phase cultures consisting of wild-type, mutant, and complemented strains were diluted to a starting OD₆₀₀ of 0.05 and incubated at 37°C with shaking for 72 h; the optical density was then measured every 15 min. For strains grown in Chamberlain’s defined medium (CDM; a chemically defined medium) there was no significant difference between the wild type, the opiA mutant, or the complemented mutant (determined by two-way ANOVA with Dunnett’s multiple-comparison test) (Fig. 2A). Interestingly, a strain that contained an additional copy of opiA on a complementation plasmid grew more robustly than the other strains in CDM at 24-, 48-, and 72-h time points (P < 0.05, as determined by two-way ANOVA with Dunnett’s multiple-comparison test) (Fig. 2A). In brain heart infusion medium supplemented with 0.2% cysteine (BHIc medium), the opiA deletion mutant did not achieve the same densities observed for the wild type or the complemented opiA-null mutant at 24- and 36-h time points (P < 0.05, as determined by two-way ANOVA with Dunnett’s multiple-comparison test) (Fig. 2A). However, after 48 h of growth there was no significant difference in the OD₆₀₀ between the opiA mutant and the other strains, but at the 60- and 72-h time points the OD₆₀₀ for the wild type and complementation strains had begun to fall, whereas for the ΔopiA strain the OD₆₀₀ was significantly higher (P < 0.0001, as determined by two-way ANOVA and Dunnett’s multiple-comparison test). A visual analysis of the BHIc medium growth curves also suggest that the ΔopiA mutant strain is entering stationary phase at ∼48 h, while all bacteria expressing at least one intact copy of this gene are entering death phase at this time (Fig. 2B).

FIG 2.

The F. tularensis LVS opiA deletion mutant has a medium-dependent growth phenotype. (A) Wild-type and ΔopiA strains of F. tularensis LVS harboring either a complementation vector or an empty vector were cultured in 96-well plates in either CDM (left) or BHIc medium (right), and the OD₆₀₀ was measured at the time points indicated. Growth curves were determined at least three times, and a representative curve is shown. Error bars represent the SE calculated from six technical replicates; however, in some instances the error bars are smaller than the graph symbols. There was no significant difference in growth between the mutant and the wild-type strain grown in CDM. A strain containing an additional copy of opiA grew significantly better than the wild type at the 24-, 48-, and 72-h time points, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.05). For strains grown in BHIc medium, the ΔopiA mutant initially grew to a lower optical density than the other strains tested at the 24- and 36-h time points, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.05). At 48 h there was no significant difference in the OD₆₀₀ between any of the strains; however, at the 60- and 72-h time points the OD₆₀₀ for the ΔopiA mutant was significantly higher than that of the other strains, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.0001). (B) CFU per milliliter for the wild-type and mutant strains was enumerated by drip plating after growth in either CDM (left) or BHIc medium (right) at the time points indicated. All strains grown in CDM showed a reduction in CFU per milliliter at the 72-h time point, and this reduction was significantly greater for the ΔopiA strain, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.05). For cells grown in BHIc medium there were significantly more CFU per milliliter for the ΔopiA strain, as determined by two-way ANOVA with Dunnett’s multiple-comparison test (P < 0.05). For panels A and B, the means ± the SE from three independent experiments are shown. For most data points, the SE was smaller than the graph symbols. (C) Cells of the opiA mutant are significantly smaller than the wild type when grown in BHIc medium. Wild-type and ΔopiA strains of F. tularensis LVS harboring either a complementation vector or an empty vector were grown in BHIc medium at 37°C with shaking. At the indicated time points, samples were spotted directly onto pads of 1% agarose dissolved in PBS and visualized by phase-contrast microscopy. Scale bar, 10 μm. (D) The areas of individual cells were analyzed by measurement using ImageJ. The mean area of opiA mutant cells was significantly less than that of either the wild type or the complemented mutant strain (P < 0.0001, as determined by two-way ANOVA with Tukey’s multiple-comparison test). For each strain and time point, three fields of view from a representative experiment were analyzed. At least 267 cells were measured for each condition at each time point. Bars show the standard error.

To investigate the medium dependent phenotype of the ΔopiA strain further, we used drip plating to enumerate the CFU per milliliter for each strain at the 24-, 48-, and 72-h time points (Fig. 2B). For cultures grown in CDM, there was a decrease in the CFU per milliliter for all strains at the 72-h time point, but for the ΔopiA strain the reduction in CFU per milliliter was greater (P < 0.05, as determined by two-way ANOVA with Dunnett’s multiple-comparison test). In contrast, for cultures grown in BHIc medium there was no decrease in CFU per milliliter after 72 h and, in fact, at the 48- and 72-h time points the ΔopiA strain had a significantly higher CFU per milliliter than any of the other strains (P < 0.05, as determined by two-way ANOVA with Dunnett’s multiple-comparison test).

The morphology of the opiA deletion mutant is altered in BHIc medium.

Next, we monitored the morphology of wild-type and mutant F. tularensis LVS strains at 2 and 72 h postinoculation using phase-contrast microscopy. For bacteria grown in BHIc medium at both time points, the opiA mutant bacteria were significantly smaller than for either the wild type or a complemented mutant strain (P < 0.0001, as determined by standard two-way ANOVA) (Fig. 2C and D). The wild-type strain containing an additional copy of opiA was also analyzed but did not appear to be morphologically different from the wild type at any time point monitored (data not shown). There was also no discernible difference in size or morphology for any of the bacterial strains grown in CDM (data not shown).

Taken together, our results show that deletion of opiA by itself can significantly affect bacterial growth, morphology, and viability in a medium-dependent manner. This suggests a broader role for opiA outside the context of infection.

The opiA mutant has an increased resistance to aminoglycoside antibiotics.

Some of our preliminary experiments had suggested that sensitivity to antibiotics might be altered in the ΔopiA strain. To test this further, we conducted Kirby-Bauer disk diffusion assays, which confirmed that the opiA mutant had a higher resistance to gentamicin than the wild type (P < 0.05, as determined by one-way ANOVA with Dunnett’s multiple-comparison test), while there was no difference in sensitivity to tetracycline (Fig. 3A). Next, we used a time-to-kill assay with concentrations of gentamicin from 0 to 200 μg ml⁻¹, and the opiA mutant was insensitive to the highest concentrations of gentamicin tested (100 to 200 μg ml⁻¹) after 1 h, which was sufficient to kill the wild type or the complemented mutant strain (Fig. 3B). We also tested two other aminoglycosides, streptomycin and hygromycin, in the same way and saw increased resistance compared to the wild type (Fig. 3C and D). These data suggest that, for the opiA mutant, there is an increase in resistance specifically to aminoglycoside antibiotics. The observed resistance to gentamicin was specific to opiA because a pdpC mutant strain (pdpC encodes another T6SSⁱⁱ effector protein) did not show the same resistance to gentamicin in similar assays (S. Cantlay and J. Horzempa, unpublished data).

FIG 3.

The opiA-null mutant has increased resistance to aminoglycoside antibiotics. (A) Kirby-Bauer disk diffusion assays. Zones of inhibition were measured after 24 h of incubation with either 1 μg of gentamicin (left) or tetracycline (right). The opiA mutant showed a significant reduction in sensitivity to gentamicin compared to the wild type (determined by one-way ANOVA with Dunnett’s multiple-comparison test, P < 0.05). There was no difference in sensitivity to tetracycline. The means ± the SE for three independent experiments are shown. (B) Multiblot time-kill gentamicin resistance assay showing gentamicin resistance of the ΔopiA mutant. F. tularensis LVS strains (as indicated) were incubated with 200, 150, 100, 50, 25, or 0 μg ml⁻¹ of gentamicin. At the time intervals shown, bacteria were spotted in duplicate onto chocolate II agar plates and incubated overnight at 37°C with 5% CO₂. After 3 h of incubation, wild-type and complemented opiA mutant strains were completely killed by concentrations of gentamicin between 50 and 200 μg ml⁻¹, whereas the ΔopiA strain was affected only at the highest concentration. (C) F. tularensis LVS strains (as indicated) were incubated with 200, 150, 100, 50, 25, or 0 μg ml⁻¹ of hygromycin. After 24 h, the opiA mutant showed increased resistance to concentrations of hygromycin of 50 to 100 μg ml⁻¹. (D) The indicated F. tularensis LVS strains were incubated with 200, 150, 100, 50, 25, or 0 μg ml⁻¹ of streptomycin. At 3 h, the opiA mutant showed increased resistance to concentrations of streptomycin of 150 to 200 μg ml⁻¹.

OpiA affects endogenous PI(3)P levels in F. tularensis LVS.

Given that OpiA is a phosphatidylinositol-3-phosphate [PI(3)P] kinase (38), we wondered whether deletion of opiA would affect lipid metabolism in F. tularensis LVS, especially because changes in the lipid membrane could affect both permeability (and therefore sensitivity to antibiotics) and growth and cell morphology. Francisella has not previously been shown to form phosphatidylinositol (PI) (48, 49); however, other intracellular pathogens, such as Mycobacterium tuberculosis and Listeria monocytogenes, have been shown to generate PI (50, 51). We used a PI(3)P mass ELISA kit (Echelon Biosciences) to determine whether we could detect PI(3)P in F. tularensis LVS and whether it would be affected by deletion of opiA. Cultures were grown for 24 h in CDM, and lipid extraction was carried out according to the protocol supplied by the manufacturers of the kit. Lipids were also extracted from Escherichia coli, which does not generate PI(3)P, and results were normalized to this negative control. Surprisingly, we were able to detect PI(3)P in the wild type and the complemented ΔopiA strain at a concentration of ∼34 pmol ml⁻¹, which was significantly higher than either the E. coli negative-control strain or the ΔopiA strain (P < 0.01 determined by two-way ANOVA with Dunnett’s multiple-comparison test) (Fig. 4). There was no significant difference between the ΔopiA strain and E. coli (determined by two-way ANOVA with Dunnett’s multiple-comparison test). We conclude, therefore, that F. tularensis is able to generate PI(3)P and that opiA is required to do so. A wild-type strain containing an additional copy of opiA (on the complementation plasmid pSC9) produced higher levels of PI(3)P (∼46 pmol ml⁻¹); however, the difference from the wild type was not statistically significant.

FIG 4.

Endogenous PI(3)P levels are affected in a ΔopiA strain. Lipid extracts from the indicated strains grown for 24 h were assayed for PI(3)P using an ELISA kit (Echelon Biosciences). Compared to samples from an E. coli negative control, PI(3)P was detected in the wild-type and complemented ΔopiA strains but was absent from the ΔopiA strain (P < 0.01, as determined by two-way ANOVA with Dunnett’s multiple-comparison test). The means ± the SE of three technical replicates from a representative experiment are shown.

Fluorescently tagged OpiA localizes to cell poles in F. tularensis.

We wanted to learn more about the role of OpiA as an effector protein of the T6SSⁱⁱ and, in particular, we wondered whether we could investigate secretion of OpiA during infection. As a first step to address this question, we constructed a fluorescence reporter by fusing opiA with emgfp (Fig. S1D). To do this, the same fragment of the F. tularensis LVS genome that was used for complementation of the mutant was cloned into a plasmid containing a promoterless emgfp. This construct was introduced into F. tularensis LVS to create a strain that produces both wild-type OpiA and OpiA-EmGFP controlled by a native promoter element. For five independent cultures grown in BHIc broth for 24 h, OpiA-EmGFP could be seen as bright fluorescent foci that were mostly localized to cell poles but could also be seen in the middle of cells and, less frequently, diffuse fluorescence was observed (Fig. 5A; see also Fig. S2A in the supplemental material). We also obtained similar results for cultures grown in CDM or TSBc (data not shown). Fluorescent foci could occasionally be seen in earlier cultures, although not reproducibly, and no foci were observed in cultures that were grown for less than 18 h. Interestingly, the growth defect observed for the opiA mutant grown in BHIc medium becomes pronounced at around 20 h of growth (Fig. 2A). To test the functionality of the OpiA-EmGFP fusion protein, we introduced the opiA-emgfp construct into the ΔopiA strain. Even in the absence of the wild-type copy of opiA, bright foci of OpiA-EmGFP were observed (Fig. 5B), and the growth defect observed for BHIc broth cultures was rescued (Fig. S2B), suggesting that the fusion protein was functional.

FIG 5.

Localization of OpiA-EmGFP in F. tularensis cultures. (A) F. tularensis LVS containing opiA-emgfp, in trans, on pSC21, grown for 24 h in BHIc medium. Over five experiments, OpiA-EmGFP foci were consistently observed at a low frequency. In 34 cells, OpiA-EmGFP specific fluorescence was observed as either polar foci, foci in the middle of cells, or diffuse fluorescence, and the percentage of each category observed is indicated. (B) OpiA-EmGFP foci can be observed in an opiA mutant containing pSC21 grown for 24 h in BHIc medium. (C) Localization of OpiA-EmGFP is not dependent on iglC. OpiA-EmGFP was observed localized to cell poles in an F. tularensis LVS iglC-null mutant containing opiA-emgfp, in trans, on pSC21, grown for 24 h in BHIc medium. (D) The localization of OpiA-EmGFP can be dynamic. F. tularensis LVS containing FGRp opiA-emgfp, in trans, on pSC22, grown for 18 h in BHIc medium at 37°C and spotted onto a pad of 1% agarose dissolved in CDM and imaged at room temperature. Images were captured over 3 h at the indicated times. Arrowheads indicate initial foci of OpiA-EmGFP, and an arrow indicates an OpiA-EmGFP that localizes to the pole of a newly divided cell. (E) F. tularensis LVS containing opiA-emgfp grown on pads of 1% agarose dissolved in CDM and incubated at 37°C with 5% CO₂. OpiA-EmGFP fluorescence was monitored at the indicated time points, and arrowheads indicate discrete OpiA-EmGFP foci. For panels A to E, the images at the top show a merge of fluorescence and phase-contrast images, while the panels at the bottom show OpiA-EmGFP fluorescence. Scale bars, 5 μm. (F) For one representative experiment, bacteria from five fields of view were counted and scored for the presence of OpiA-EmGFP foci. The average frequency of OpiA-EmGFP foci per cell at each time point is shown. There was a significant increase in OpiA-EmGFP foci per cell at the 336-h time point (determined by one-way ANOVA with Tukey’s multiple-comparison test; P > 0.0001). From the left, the total numbers of bacteria were 2,496, 4,440, 794, and 902; the error bars represent the standard errors of the mean.

OpiA-EmGFP foci were consistently observed in broth cultures grown for 18 to 24 h; however, the frequency of foci present in any given field of view was extremely low (typically 0.1 to 1.0% of cells had OpiA-EmGFP foci). To increase the expression of opiA-emgfp, we cloned the complementing fragment into a vector which contained emgfp under the control of an additional active Francisella promoter FGRp (Fig. S1E) (52). Overexpression of opiA-emgfp led to an increase in fluorescent foci, with 57% of cells containing one or more fluorescent foci (Fig. S2C). The localization of OpiA-EmGFP was not altered in the overproduction strain, with 67% of foci localizing to cell poles, 26% localizing at the middle of cells, and 6% of cells having twin foci (Fig. S2C). The temporal pattern of OpiA-EmGFP fluorescence was not altered in the overproduction strain, with foci being observed in cultures grown between 18 and 24 h but not at earlier time points. This overproduction construct, therefore, is a useful tool for analyzing OpiA-EmGFP in cells because it does not alter the localization patterns of OpiA-EmGFP but increases the frequency at which localization can be observed.

When the overproduction construct was introduced into the ΔopiA strain, a similar increase in OpiA-EmGFP foci was observed and the pattern of localization was similar to that seen for the wild type; however, fewer cells (∼26%) had fluorescent foci (Fig. S2D). Our results show that the recombinant OpiA-EmGFP fusion protein can rescue the opiA mutant growth defect and that it is able to localize spontaneously to the cell poles in the absence of wild-type OpiA.

Localization of OpiA-EmGFP is not dependent on iglC, which encodes a structural component of the T6SSⁱⁱ.

Assembly of the T6SSⁱⁱ has been shown to occur at the cell poles in F. novicida (53). Given the predominantly polar localization of OpiA-EmGFP and the fact that OpiA is a substrate of the T6SSⁱⁱ, we wanted to test whether the localization was dependent on T6SSⁱⁱ assembly. To do this, we introduced opiA-emgfp into an iglC-null mutant (21) that is presumably defective in assembly of the T6SSⁱⁱ and has been shown to be attenuated for erythrocyte invasion (21). For three independent experiments in which a iglC-null opiA-emgfp strain was cultured for 18 to 48 h in BHIc broth, similar patterns of OpiA-EmGFP localization were observed, with a majority of foci having a polar localization (Fig. 5C). Overproduction of OpiA-EmGFP from the FGRp promoter in the iglC-null mutant also produced similar results as for the wild type (Fig. S2E). These results indicate that localization of OpiA-EmGFP is not dependent on iglC, a structural component of the T6SSⁱⁱ.

The localization of OpiA-EmGFP is dynamic.

The presence of either one or two OpiA-EmGFP foci at cell poles and, occasionally, foci at the midcell raised the possibility that the localization of OpiA-EmGFP could be dynamic. To investigate this, time-lapse imaging experiments were conducted. Cells grown in BHIc broth for 24 h were spotted onto pads of 1% agarose dissolved in CDM in 35-mm dishes (Ibidi Biosciences) and were incubated at room temperature. Two experiments were conducted with a strain containing opiA-emgfp controlled by the presumed native promoter, and two experiments were conducted where opiA-emgfp was also driven from the FGRp promoter. Experiments were conducted over the course of 1 to 3 h, and images were taken at 15- to 45-min intervals. Most OpiA-EmGFP foci in any given field of view remained static over the course of these experiments, suggesting that these foci, once formed, can be very stable. However, for each experiment, dynamic localization of OpiA-EmGFP foci was observed in at least one cell, with foci shifting position within cells or new foci appearing in a cell that contained an existing focus (Fig. 5D). These events are relatively rare and may reflect the fact that in most cells the foci of OpiA-EmGFP localized prior to the start of the experiment and remain stable throughout. In at least one case (Fig. 5D) the dynamic localization occurred in an actively dividing cell, which shows that the pattern of localization observed is not as a result of degradation or mislocalization in a dying cell. F. tularensis LVS cultures placed on agarose pads in 35-mm dishes and incubated at room temperature grow robustly, producing large microcolonies over the course of several days (data not shown), which also supports a conclusion that the localizations observed here are not artifacts of cell death or loss of viability during imaging. Even though foci of OpiA-EmGFP can be seen in a dividing cell, there were many other cells that were undergoing division which did not have any detectable fluorescence. We can conclude, therefore, that OpiA-EmGFP localization is not dependent on cell division.

OpiA-EmGFP foci become more abundant in older, surface-grown cultures of F. tularensis LVS.

Because deletion of opiA alters the viability of cells after prolonged cultivation on either plates (data not shown) or in broth, we were interested to see whether the localization of OpiA-EmGFP changes over time. The localization of OpiA-EmGFP in F. tularensis LVS cells grown overnight in broth culture occurs in a relatively small subset of cells, and prolonged culture of 48 and 72 h did not lead to an increase in the relative abundance of OpiA-EmGFP foci (data not shown). For cultures grown on solid medium, we examined cells expressing opiA-emgfp from its presumed native promoter that were grown overnight in broth culture and then were spotted onto pads of 1% agarose in CDM in 35-mm dishes that were incubated at 37°C and analyzed at intervals by microscopy. For cultures incubated between 24 and 96 h, OpiA-EmGFP was present at roughly similar frequencies to those observed for broth cultures with 0.1 to 0.5% of cells containing detectable OpiA-EmGFP foci. Interestingly, diffuse fluorescence was commonly observed for cells that were incubated on agarose pads for between 24 and 48 h; however, by 96 h, no diffuse fluorescence was detected. For cultures incubated for 2 weeks, there was significantly more OpiA-EmGFP with ∼0.82 foci per cell on average (P > 0.0001 determined by two-way ANOVA with Tukey’s multiple-comparison test) (Fig. 5F). The increase in frequency of OpiA-EmGFP foci is also coincident with considerable morphological heterogeneity of the F. tularensis cells.

OpiA-EmGFP fluorescence can be detected in THP-1 cells during infection by F. tularensis LVS.

Because the T6SSⁱⁱ has been shown to be involved in replication of Francisella spp. in phagocytic cells (29, 31–33) and since OpiA has been shown to be secreted by the T6SSⁱⁱ and is involved in pathogenesis (37, 38), we wanted to determine whether we could visualize OpiA-EmGFP in phagocytic cells during infection. To do this, we incubated THP-1 cells (having identified that opiA was involved in the attenuation of cytokine signaling in THP-1 cells) with opiA-emgfp-expressing strains of F. tularensis LVS at an MOI of 100 and monitored the fluorescence. THP-1 cells incubated with wild-type bacteria or a strain containing a vector with emgfp but no promoter were used as controls. After 3 h of incubation, diffuse OpiA-EmGFP fluorescence was detected in ∼10% of THP-1 cells incubated with F. tularensis LVS expressing opiA-emgfp from the presumed native promoter (Fig. 6A and B). The diffuse OpiA-EmGFP-specific fluorescence observed within THP-1 cells would be consistent with OpiA-EmGFP being secreted from the invading bacteria into the host cells. We monitored fluorescence in cells after 48 and 72 h of incubation; however, no significant differences between either strains or time points was observed (determined by two-way ANOVA with Holm-Sidak’s multiple-comparison test) (Fig. 6B). To eliminate the possibility that the diffuse fluorescence was the result of degradation of F. tularensis by the host cells, we utilized F. tularensis LVS strains expressing either emgfp or tdTomato from the FGRp promoter. For these cells, incubated with THP-1 cells under the same conditions, individual fluorescently labeled bacteria could be observed (Fig. S3A and B) at approximately the same frequency in 15% of cells (Fig. S3C) after 24 h of incubation. Even after prolonged incubation periods of 72 and 96 h (data not shown), fluorescently labeled F. tularensis appeared as discrete cells. To further test this, we incubated THP-1 cells with opiA-emgfp-expressing strains of F. tularensis LVS at an MOI of 100 for 24 h. The cells were washed once to remove extracellular bacteria, and the fluorescence of the sterilized cell lysates was measured (Fig. 6C). The fluorescence observed for the lysates from THP-1 cells incubated with F. tularensis LVS containing either opiA-emgfp under native promoter conditions, or with the FGRp in addition, was significantly higher than that for cells incubated with a control strain containing emgfp on a promoterless vector (P < 0.05 determined by one-way ANOVA with Dunnett’s multiple-comparison test). These data therefore suggest that the diffuse fluorescence shown in Fig. 6A is consistent with secreted OpiA-EmGFP and not OpiA-EmGFP fluorescence associated with individual cells or released from degraded bacterial cells.

FIG 6.

Localization of OpiA-EmGFP during THP-1 infection. (A) THP-1 cells incubated with F. tularensis LVS containing either opiA-emgfp expressed from its native promoter (pSC21) or with an additional constitutive promoter element (pSC22) for 24 h at an MOI of 100. Diffuse fluorescence was observed in cells (examples are indicated by arrowheads) incubated with bacteria expressing opiA-emgfp but not in a control strain. Scale bar, 25 μm. (B) THP-1 cells containing OpiA-EmGFP fluorescence were counted at the indicated time points. The averages from three independent experiments are shown; at least 129 cells were counted for each time point and experimental condition. There was no significant difference between strains or time points monitored (determined by two-way ANOVA with Holm-Sidak’s multiple-comparison test). The error bars represent the standard errors. (C) OpiA-EmGFP fluorescence can be detected in cell lysates. THP-1 cells were incubated with the indicated F. tularensis LVS strains at an MOI of 100 for 24 h. Fluorescence was significantly higher in lysates from cells incubated with F. tularensis LVS strains containing a copy of opiA-emgfp compared to a negative control (P < 0.05 determined by one-way ANOVA with Dunnett’s multiple-comparison test). The means ± the SE from four independent experiments are shown.

OpiA-EmGFP fluorescence can be detected in F. tularensis cells invading human erythrocytes.

We were also interested to determine whether we could detect OpiA-EmGFP during erythrocyte invasion. For this, F. tularensis LVS expressing opiA-emgfp was incubated with human erythrocytes in 35-mm dishes and EmGFP fluorescence was monitored at regular intervals. No fluorescence was detected in control samples with bacteria containing a vector with emgfp but no promoter. However, after 24 h of incubation, OpiA-EmGFP fluorescence could be detected in ∼2% of RBCs for opiA-emgfp driven by the presumed native promoter and in ∼2.5% of RBCs incubated with the OpiA-EmGFP overproduction strain (Fig. 7). No difference in native expression and overproduction strains was seen (determined by unpaired t test); however, after prolonged incubation (72 h) the proportion of erythrocytes containing OpiA-EmGFP fluorescence was significantly higher (∼4%) for both strains (determined by two-way ANOVA with Tukey’s multiple-comparison test) (Fig. 7B). We attempted to stain the DNA of F. tularensis cells with a membrane permeable Hoechst stain. However, in unfixed samples the labeling of bacteria was uneven; in fixed samples the DNA labeling was better but still far from complete, and the fixation process quenched EmGFP fluorescence. We are not therefore able to state with confidence whether the fluorescence observed in the erythrocytes was in individual bacteria or was secreted OpiA-EmGFP. However, we can conclude that OpiA-EmGFP-specific fluorescence was detected in erythrocytes and that the number of erythrocytes with OpiA-EmGFP fluorescence increased over time. Although opiA is not required for invasion of human erythrocytes, the presence of OpiA-EmGFP-specific fluorescence subsequent to invasion suggests that opiA may be involved in processes that occur after invasion of erythrocytes by F. tularensis LVS.

FIG 7.

Localization of OpiA-EmGFP during erythrocyte invasion. (A) Erythrocytes incubated with F. tularensis LVS containing either opiA-emgfp expressed from its native promoter (pSC21) or with an additional constitutive promoter element (pSC22) for 24 h at an MOI of 100. Discrete foci of fluorescence were observed in erythrocytes (examples are indicated by arrowheads) incubated with bacteria expressing opiA-emgfp but not in a control strain. Scale bar, 10 μm. (B) Erythrocytes containing OpiA-EmGFP foci were counted at the indicated time points. There were significantly more erythrocytes with OpiA-EMGFP foci at 72 h (P < 0.0001, as determined by two-way ANOVA with Tukey’s multiple-comparison test). The averages from five fields of view from a representative experiment are shown. At least 139 cells were counted for each time point and experimental condition. The error bars represent the standard errors.

DISCUSSION

We set out to investigate the role of opiA in erythrocyte invasion by F. tularensis. Our investigation showed that opiA was not required for erythrocyte invasion, raising the possibility that other T6SSⁱⁱ effector proteins mediate this phenomenon (21). We also showed that a ΔopiA strain was attenuated in a chicken embryo infection assay, consistent with the role previously identified for opiA in other Francisella spp. in pathogenesis (37, 38). However, our data also demonstrated that opiA influences the growth, cell size, viability, and aminoglycoside resistance of F. tularensis LVS, identifying a novel role for opiA, outside the context of infection. We also present here a fluorescent fusion to OpiA which localizes predominantly to the cell poles in broth-grown cultures and localizes more frequently in surface-grown cultures after prolonged incubation. The fluorescent fusion to OpiA was also used to investigate localization of OpiA-EmGFP during in vitro cell infection or erythrocyte invasion assays. OpiA-EmGFP fluorescence was seen in both erythrocytes and THP-1 cells; however, the patterns of localization were different in both host cell types with a punctate localization in the former and a diffuse localization consistent with secretion in the latter.

Of the known secreted effector proteins of the T6SSⁱⁱ in Francisella (PdpC, PdpD, OpiA, OpiB, IglD, and IglE), all have so far been shown be involved in virulence in host cells (37, 38, 54–58). In F. tularensis LVS, there was a slight, but statistically significant, attenuation of virulence in an in vivo chicken embryo infection model. We did not observe any attenuation in the two in vitro cell lines tested; however, it is possible that, as shown previously (37, 38), functional redundancy with other T6SSⁱⁱ effectors, or even other virulence factors, masks the effects of the deletion of opiA under the specific conditions tested.

Infection of host macrophages and subsequent intracellular replication of F. tularensis spp. relies on the early inhibition of, among others, the proinflammatory cytokine TNF-α (46, 59). Upregulation of TNF-α in THP-1 cells infected with a ΔopiA mutant suggests that opiA may exert an effect on virulence, in part, through suppression of host immune signaling.

Little is known about whether the T6SSⁱⁱ of Francisella spp., and its secreted effectors can influence bacterial physiology outside infection models. We report the first evidence of a secreted effector protein that affects both growth and cell morphology of F. tularensis LVS in broth cultures. The observed effect of opiA deletion was less pronounced in CDM. There was no obvious difference in growth or in cell morphology between the mutant and the wild-type strain, and while the ΔopiA strain had fewer CFU per milliliter after 72 h of growth, all of the strains tested showed a decline in plating efficiency at this time point, and the loss in viability observed for the opiA mutant in CDM was not accompanied by any loss in cell density. All of the strains grew more slowly in CDM than in BHIc, and this lower growth rate may have masked any effects of the deletion of opiA or allowed cells to compensate. For BHIc medium, there was a stronger effect. As measured by the optical density, the growth of the mutant was slower initially but was able to catch up by 48 h. At later time points the density of wild-type cultures declined, but this decline was much less obvious for the ΔopiA strain.

Differences in the pH of culture media has been shown to influence growth; for example, a previous study showed that deletion of ripA in F. tularensis LVS resulted in a growth defect in CDM in which the starting pH was adjusted to 7.5, whereas no defect was observed when this medium was set to pH 6.5 (60). We measured the starting pH of BHIc medium (pH 6), and it was slightly lower than that of CDM (pH 6.3); it is possible that this difference could have contributed to the observed phenotypes. We did not monitor changes in pH of media during growth, so we cannot rule out pH as a factor, especially at later time points in cultures that had reached stationary phase.

In addition, the composition of the respective media may have affected the contrasting patterns of growth we observed for the opiA mutant. The amino acid content of BHI compared to other culture media, for example, can lead to gene expression changes that may influence molecular cascades that maintain cell growth, morphology, and viability (61).

In addition to the growth defect in BHIc, the morphology of the cells was clearly affected in this medium, with cells of the opiA mutant being smaller on average than the wild type. It is not clear why the CFU per milliliter was higher for the ΔopiA strain by 48 and 72 h of growth in BHIc medium. One possibility is that the smaller cell size relates to more CFU per milliliter at any given optical density, and this would be consistent with the observed data. A smaller cell size would also explain why there were differences in measured optical density for the BHIc cultures but not for cultures grown in CDM. However, the reduction in viability observed for the mutant grown in CDM could not be explained by any change in cell size.

The question of viability is also complicated by the fact that Francisella spp. are known to enter into a viable but nonculturable (VBNC) condition (62, 63). Francisella readily enters this VBNC state in laboratory cultures; however, virtually nothing is known about the mechanisms that control this. VBNC is a state associated with a broad range of bacteria, many of them pathogens, and entry into VBNC has been associated with changes in membrane fluidity (64, 65) and antibiotic tolerance and persistence (66) and as a response to a variety of stressors (67).

The loss in viability of the opiA mutant in CDM, not accompanied by any loss in cell density, could be consistent with an increase in the proportion of VBNC cells. Conversely, the higher CFU per milliliter observed ΔopiA strain grown in BHIc, at least at the 48-h time point where the culture density is similar to that of the wild type, could indicate that, in the absence of opiA, fewer cells transition into an unculturable state. Interestingly, morphological changes in F. tularensis LVS, specifically a change from the rodlike to a more coccoid morphology, are coincident with the transition into a VBNC state (Cantlay and Horzempa, unpublished). In BHIc medium, the ΔopiA strain does not appear to undergo the same morphological changes seen for the wild type, and this coincides with an increased CFU per milliliter. Therefore, in BHIc medium (nutrient-rich broth), the presence of opiA is associated with transition into either an unculturable state or induced cell death, as well as with a larger cell size, suggesting that the product of this gene may mediate these phenomena. Perplexingly, in CDM the opposite effect on viability is observed. In this medium the absence of opiA eventually leads to reduced viability. This further suggests a role for OpiA in regulation of the cell biology of Francisella in a medium-dependent manner. However, more research is needed to definitively establish whether there is any relationship between the observed morphological and physiological phenotypes and entry into a VBNC state.

Overall, the defect in growth, viability, and morphology was medium dependent, specifically being seen in BHIc medium, which has been shown to promote upregulation of the mglA regulon in a fashion that more closely resembles expression observed during infection (61). Also, opiA has been shown to be regulated by mglA in F. novicida and upregulated during infection of macrophages (35, 68). However, deletion of mglA has not previously been shown to influence bacterial growth (35) in culture. These data were from experiments performed with F. novicida grown in tryptic soy broth, however. We also measured growth of the opiA mutant in TSBc compared to the wild type using fluorescently labeled strains and did not see any differences (data not shown), indicating that the growth defects observed were specific to BHIc. Also, we cannot rule out any species-specific effects; the generally faster growth of F. novicida compared to F. tularensis LVS, for example, might mask any growth defects of an opiA mutant in the former. In a 2018 analysis of OpiA from F. novicida, Ledvina et al. (18) highlighted the limitations of specific culture conditions and in vitro and in vivo models that may result in incomplete pictures of complex phenotypes resulting from the interactions of multiple factors. Our result showing medium-specific growth defects in F. tularensis LVS is one example of this.

The opiA-dependent resistance to aminoglycosides is an additional phenotype relating to the physiology of F. tularensis LVS, rather than to host cell interactions, identified in this study. It has previously been demonstrated that resistance to gentamicin is temperature dependent for both F. tularensis LVS and the pathogenic Schu S4 strain, with a reduction in gentamicin uptake at lower temperatures (26°C), presumably due to a decrease in permeability of the bacterial membrane (69). Differences in membrane composition between the opiA mutant and wild-type bacteria may be responsible for the observed resistance; however, this has yet to be determined.

There are many examples of T6SS affecting both growth and antibiotic resistance in bacteria. For example, strains of Acinetobacter baumannii, mutant for vgrG (which encodes a spike complex protein of the T6SS), showed reduced growth and altered resistance to chloramphenicol and ampicillin-sulbactam (70). The T6SS has been shown to be involved in quorum and cell density sensing in Vibrio spp. (71, 72) and Pseudomonas spp. and in a wide range of stress responses in Yersinia spp. (73).

The most direct mechanism through which OpiA could affect physiology of F. tularensis LVS is via its activity as a phosphatidylinositol kinase (PI-kinase). Bacterial PI-kinases are known to play a role in virulence for a range of pathogenic bacteria, including Francisella spp. (38, 74). The presence of phosphatidylinositol (PI) in bacterial membranes has been described or predicted in fewer species: Listeria monocytogenes, Myxococcus xanthus, and actinomycetes such as Mycobacterium tuberculosis (48, 50, 75, 76). Despite the fact that PI had not previously been reported in Francisella (48, 49), we decided to employ an ELISA that had been used to detect PI(3)P from host cells during infection or while expressing opiA (38). We were surprised to be able to consistently detect PI(3)P in lipid extracts from F. tularensis LVS since Francisella has not previously been shown to have PI(3)P in its membrane. The lack of detectable PI(3)P in the ΔopiA strain suggests that OpiA does play a role in the lipid metabolism of Francisella outside its established role in host cell infection. We searched the F. tularensis LVS genome for potential homologs of opiA and specifically looked for genes that shared the OpiA family protein motif identified by Ledvina et al. (38). One open reading frame, FTL_0131, is annotated as encoding a branched-chain amino acid synthetase, which shares 19.25% identity with OpiA and has 2 of the conserved IDH residues in the catalytic domain characteristic of OpiA family proteins (Fig. S4A). There were also genes that shared some broad homology with opiA (not shown) and, given that opiA itself does not encode a protein that shares strong similarity with PI-kinases from other bacteria, it is possible that there are other proteins encoded by the F. tularensis LVS genome that have a function similar to that of OpiA. However, our data suggest that opiA alone is sufficient for production of PI(3)P in F. tularensis LVS.

We saw a consistent increase in PI(3)P levels observed when an extra copy of opiA is present, although it was not statistically significant. While further investigation will be required to clarify exactly how, we speculate that an additional copy of opiA may lead to more OpiA being present in the cell and therefore more PI-kinase activity. Interestingly, the strain which contained an extra copy of opiA also grew to higher densities, particularly in CDM, raising the possibility that the increased PI(3)P content of the membrane positively influences growth under these conditions.

An important consideration here, however, is the source of the substrate PI from which OpiA or other proteins generate PI(3)P. Generation of PI has not been reported for F. tularensis or for the Gammaproteobacteria as a group (48, 49). However, homologs of the M. tuberculosis PI-synthetase, PgsA, are present in Francisella genomes. Notably, one gene from the F. tularensis LVS genome, FTL_0278 (AJI58288.1), encodes a protein which shares 25.62% identity (E = 3e⁻⁰⁸) with PgsA and has a conserved CDP-alcohol synthetase motif that includes two of four conserved residues that are characteristic specifically of PI-synthetases (Fig. S4B), so F. tularensis LVS may have a similar capacity to generate PI.

T6SSⁱⁱ sheath proteins, IglA and IglB, of F. novicida have been shown to localize at the cell poles (53, 77); the largely polar localization of OpiA-EmGFP, therefore, would be consistent with that of the T6SSⁱⁱ. In broth cultures we observed OpiA-EmGFP foci at a very low frequency, and this is also similar to what has been reported for T6SSⁱⁱ sheath proteins. For fluorescently labeled variants of IglA in F. novicida, assembly of fluorescent foci was stimulated after uptake by macrophages or by addition of KCl in broth cultures to mimic the intracellular environment of host cells (77). Interestingly, placing bacteria under sealed cover glass or under cover glass on agarose pads also stimulated assembly of IglA (53, 77).

We tried different culture media (TSBc, Mueller-Hinton broth, CDM, and BHIc medium) and the addition of KCl and Casamino Acids to determine whether we could find culture conditions under which OpiA-EmGFP fluorescence would be more frequent. However, none of the conditions we tested stimulated OpiA-EmGFP formation of foci at frequencies higher than those reported here for BHIc medium.

The potential that the cloned fragment containing opiA-emgfp and the entire upstream intergenic region does not contain all the regulatory elements of the native locus could not be ruled out. Therefore, it is possible that this could contribute to the observed fluorescence phenotype. However, the same fragment was able to complement the growth, culturability, and gentamicin resistance phenotypes in the mutant. This suggests that the low frequency of OpiA-EmGFP observed is the result of tight regulation of expression under the conditions tested. Placing an additional promoter element, FGRp, which is presumably not regulated by the same pathways as the native opiA promoter, upstream of opiA-emgfp and its presumed native promoter led to more OpiA-EmGFP foci in all strains tested.

Morphological plasticity is a known feature of Francisella (78, 79), but the molecular mechanisms that drive these morphological changes are not well understood. However, in liquid broth culture a wide range of morphological heterogeneity is observed, and under all conditions tested we saw no correlation between either cell shape and size on the presence or positioning of OpiA-EmGFP foci. We conclude, therefore, that localization of OpiA-EmGFP is not, per se, dependent on morphology. However, for cultures grown on agar, more OpiA-EmGFP foci were observed in cells after 2 weeks, a time point which was also coincident with an increased heterogeneity of cell morphology. Taken together with the observation that cell morphology is affected in the opiA mutant in BHIc broth culture, these data further suggest that opiA may be involved in pathways relating to cell shape and morphological plasticity in F. tularensis LVS.

The localization of OpiA-EmGFP was not affected in a strain mutant for iglC, which encodes the stacking subunit of the T6SSⁱⁱ tube in Francisella (80). iglC mutants lack a functional T6SSⁱⁱ; however, assembly of the T6SS occurs in a hierarchical fashion (reviewed in reference 81), so OpiA localization may be dependent on membrane complex or baseplate proteins. Time-lapse imaging revealed that a very small subset of OpiA-EmGFP foci localized dynamically and that any changes in localization were relatively slow, occurring over 2 to 3 h. This is in contrast to the very fast assembly and dynamic localization observed for T6SSⁱⁱ components in F. novicida (53). Consequently, even though the predominantly polar localization observed for OpiA-EmGFP is consistent with an interaction with a polar localized T6SSⁱⁱ, the differences in dynamic assembly and lack of dependence on iglC suggest that OpiA foci are not strictly dependent on a fully intact T6SSⁱⁱ. These data would be consistent with a dual role for opiA in both infection and bacterial physiology.

While it appears that the fusion protein is functional, based on the ability to localize in an opiA mutant and its ability to rescue the growth phenotype of the opiA mutant, it is possible that the addition of a fluorescent tag could affect secretion. Previously undertaken studies of putative substrates of the T6SSⁱⁱ have had mixed results due to the effects of tags used in the analyses (37, 82, 83). We could not detect direct evidence of secretion in broth culture; fluorescence was too low to be measured by a plate reader, and Western blots with whole-cell extracts and an antibody against green fluorescent protein (GFP) showed only a faint band of the expected size and only when expression of opiA-emgfp was upregulated by the FGRp promoter (data not shown).

Fluorescently tagged components of the T6SSⁱⁱ and putative secreted effectors have been localized in host cells during infection. Clemens et al. showed that in THP-1 macrophage infection assays IglA and IglB tagged with split GFP could be seen as bright foci in F. novicida cells as soon as 15 min postinfection, and by 22 h up to 70% of intracellular F. novicida had bright IglA-GFP foci (77). A recent study by Shimizu et al. reported that in, 293T cell infection assays, while IglH, IglI, and IglC tagged with GFP localized diffusely in host cells, PdPA and IglE had a punctate localization (56). For THP-1 cells incubated with F. tularensis LVS strains containing opiA-emgfp, increased fluorescence was observed under the microscope and increased fluorescence was observed in cell lysates, a finding consistent with the secretion of OpiA-EmGFP.

In conclusion, data presented here are consistent with previous findings suggesting a role for opiA during Francisella infection. Our findings extend this work by further identifying a novel role for opiA in the growth, viability, and morphology of F. tularensis LVS. The fluorescent fusion of OpiA to EmGFP has shown a subcellular localization for OpiA that is consistent with its role as a secreted effector of the T6SS and has shown that OpiA-EmGFP can be detected in both THP-1 cells and human erythrocytes.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains used in this study are listed in Table 1. For cultivation of F. tularensis LVS strains, frozen stock cultures of bacteria were streaked onto chocolate II agar (22) and incubated at 37°C with 5% CO₂ for 3 to 7 days. For broth cultures, bacteria from plates were used to inoculate either Chamberlain’s chemically defined medium (CDM; pH 6.3) (84), tryptic soy broth (Becton, Dickinson, and Co.) supplemented with 0.1% cysteine hydrochloride (TSBc; Fisher Scientific), or brain heart infusion broth (BHI; Oxoid, Ltd.) supplemented with 0.2% cysteine hydrochloride (53) (pH 6) and grown to stationary phase by overnight incubation at 37°C with agitation. Where required, media were supplemented with the following antibiotics: kanamycin (10 μg ml⁻¹), hygromycin (200 μg ml⁻¹), or polymyxin B (100 μg ml⁻¹). For E. coli, bacteria were grown at 37°C on Luria-Bertani (LB) agar plates or in LB broth (Fisher Scientific) with the following antibiotics where appropriate: kanamycin (50 μg ml⁻¹), hygromycin (200 μg ml⁻¹), ampicillin (100 μg ml⁻¹), or polymyxin B (100 μg ml⁻¹).

TABLE 1.

Bacterial strains, plasmids, and oligonucleotides used in this study

Strain, plasmid, or oligonucleotide	Description	Source or reference
Strains
F. tularensis
LVS	F. tularensis subsp. holarctica live vaccine strain	Karen Elkins
ΔopiA strain	LVS with opiA deleted	This study
iglC-null strain	LVS with both copies of iglC deleted	21
LVS/pFNLTP8	LVS containing the F. tularensis shuttle vector pFNLTP8	This study
ΔopiA/pFNLTP8 strain	LVS with opiA deleted containing pFNLTP8	This study
iglC-null/pFNLTP8 strain	LVS with both copies of iglC deleted containing pFNLTP8	This study
LVS/pSC9	LVS containing the opiA complementation vector pSC9	This study
ΔopiA/pSC9 strain	ΔopiA strain containing the opiA complementation vector pSC9	This study
LVS/pSC13	LVS containing promoterless emgfp on pSC13	This study
ΔopiA/pSC13 strain	ΔopiA strain containing promoterless emgfp on pSC13	This study
iglC-null/pSC13 strain	LVS with both copies of iglC deleted containing pSC13	This study
LVS/pSC21	LVS containing opiA-emgfp on pSC21	This study
ΔopiA/pSC21 strain	ΔopiA strain containing opiA-emgfp on pSC21	This study
iglC-null/pSC21 strain	LVS with both copies of iglC deleted containing pSC21	This study
LVS/pSC22	LVS containing FGRp-opiA-emgfp on pSC22	This study
ΔopiA/pSC22 strain	ΔopiA strain containing FGRp-opiA-emgfp on pSC22	This study
iglC-null/pSC22 strain	LVS with both copies of iglC deleted containing pSC22	This study
LVS/pKHEG	LVS expressing emgfp	This study
LVS/pTC3D	LVS expressing tdTomato	52
E. coli DH5α	fhuA2 Δ(argF-lacZ)U169 phoA glnV44 ϕ80Δ(lacZ)M15 gyrA96 recA1 relA1 endA1 thi-1 hsdR17	NEB
Plasmids
pJH1	An integrating suicide vector with an I-SceI restriction site	85
pGUTS	A stably replicating plasmid with I-SceI under the control of FGRp	85
pSC2	pJH1 with the left and right 500-bp flanking regions of opiA cloned into the BamHI-PstI site of pJH1	This study
pFNLTP8	Francisella shuttle plasmid	87
pSC9	The opiA complementation vector created by cloning a 1,283-bp fragment into the EcoRI-SalI site of pFNLTP8	This study
pRSET/EmGFP	The source of emgfp	Invitrogen
pSC13	emgfp cloned into the NdeI-BamHI site of pFNLTP8	This study
pSC21	opiA-emgfp under its presumed native promoter created by cloning the entire coding sequence of opiA and 650 bp upstream into the EcoRI-NdeI site of pSC13	This study
pSC11	The 51-bp Francisella FGRp cloned into the KpnI-EcoRI site of pFNLTP8	This study
pSC18	emgfp cloned into the NdeI-BamHI site of pSC11	This study
pSC22	FGRp opiA-emgfp created by cloning the entire coding sequence of opiA and 650 bp upstream into the EcoRI-NdeI site of pSC18	This study
pTC3D	pF8tdTomato with DNA (nucleotides 557161 to 558861) from the LVS chromosome cloned upstream of tdTomato	52, 86
pKHEG	326 nucleotides, including FGRp, from pTC3D cloned into the KpnI-NdeI site of pSC13	This study
Oligonucleotides
0136F1	CATGGGATCCGTCTCTAAAATCTTGGCTATATGATG	IDTa
0136R1	TAATAATAAGTCGACGGTACCACCGGTAATCAATGCGTTTTAGTCCTGTTAAAG	IDT
0136F2	ACCGGTGGTACCGTCGACTTATTATTAAGAACTTGGATGCTTATTTCCTCCC	IDT
0136R2	CATGCTGCAGGTCTATAGATGAAGTTTCTTAAGCAAAATTC	IDT
SC6	GGTGGTGAATTCGAGCTTAATTTTATAAGAATG	IDT
SC7	GGTGGTGTCGACCTAAGATTCTTGAATTTGC	IDT
SC9	GGTGGTCATATGCTCCGGGCCCGGCAGAAAGAAGAAATAAACTAGAC	IDT
Sig1	CTATCTCTTGGTATGTATAAAGTTAAGTTATAAATCAAAAATTGCTAGGAGCATTTAAG	IDT
Sig2	AATTCTTAAATGCTCCTAGCAATTTTTGATTTATAACTTAACTTTATACATACCAAGAGATAGGTAC	IDT
KH1	CATGGGTACCCTATAGTTCTTCATCATACTTATGCTCTTG	IDT
KH2	CATGCATATGGGTGGCGGGATCTTCAT	IDT

^aIDT, Integrated DNA Technologies.

Construction of mutant strain, complement, and opiA-emgfp constructs.

The plasmids and oligonucleotides used in this study are listed in Table 1. Restriction endonucleases and Phusion polymerase were used according to the manufacturer’s instructions (New England Biolabs). FTL_0136 has previously been identified as the gene encoding OpiA in F. tularensis LVS (37, 38). We used the unstable, integrating suicide vector, pJH1, and I-SceI endonuclease (85) to create an unmarked deletion of opiA. The 500-bp flanking either side of FTL_0136 were amplified by PCR using the primer pairs 0136F1/0136R1 and 0136F2/0136R2, and then extension overlap PCR with the primer pair 0136F1/0136R2 was used to combine these two amplification products to generate a disruption cassette that was cloned into pJHI to create pSC2 (see Fig. S1A in the supplemental material). This plasmid was introduced into F. tularensis LVS by triparental mating (86) and was excised by I-SceI restriction following the introduction of a plasmid (pGUTS) encoding I-SceI. Clones in which the wild-type opiA allele had been completely removed were identified by PCR using the primer pair 0136F1 and 0136R2 (Fig. S1B). A complementation vector was constructed by amplifying a 1,283-bp fragment from F. tularensis LVS chromosomal DNA using the primer pair SC6/SC7 that contained the entire 633-bp coding sequence of FTL_0136 and the 650 bp upstream that presumably contains the native promoter. This amplified fragment was cloned into the EcoRI-SalI site of the Francisella shuttle vector pFNLTP8 (87) to create pSC9 (Fig. S1C).

To create an opiA-emgfp construct, emgfp was first amplified from pRSET-Emgfp (Life Technologies) using the primer pair SC18/SC19 and cloned into the NdeI-BamHI site of pFNLTP8 to generate pSC13. A 1,280-bp fragment corresponding to the 650 bp upstream of opiA and the full coding sequencing of opiA minus the stop codon was amplified with the primer pair SC6/SC9. Primer SC9 included a sequence encoding an LGPGE linker motif (88) that was incorporated into the sequence of opiA during amplification. This fragment was introduced into the EcoRI-NdeI site of pSC13 to create an in-frame fusion of opiA with emgfp generating pSC21 (Fig. S1D).

To create a construct in which the opiA-emgfp allele was overexpressed, the Francisella promoter FGRp (52) was cloned into pFNLTP8. To do this, oligonucleotides Sig1 and Sig2 were mixed in equimolar concentrations and placed at 4°C for 10 min to anneal. The resulting double-stranded DNA molecule was cut with KpnI and EcoRI cloned into the KpnI-EcoRI site of pFNLTP8 to create pSC11. Next, emgfp (amplified as above) was cloned into the NdeI-BamHI site of pSC11 to create pSC18. Finally, the SC6/SC9 amplification product (detailed above) was cloned into the EcoRI-NdeI site of pSC18. This resulted in pSC22 (Fig. S1E), which has an in-frame fusion of opiA and emgfp with the presumed native promoter element and FGRp upstream.

The complementation vector (pSC9), the opiA-emgfp vectors (pSC21 and 22), and the parent vectors used as controls (pFNLTP8 and pSC13) were introduced into the wild-type and mutant strains by electroporation and selection on chocolate II agar supplemented with 10 μg ml⁻¹ kanamycin.

Determination of medium-dependent growth curves.

To determine growth curves, F. tularensis LVS strains were grown to stationary phase by an overnight incubation at 37°C in either CDM or BHIc. These cultures were diluted to an OD₆₀₀ of 0.05 and were used to inoculate 96-well plates with six technical replicates for each strain. The plates were incubated at 37°C in a Synergy H1 microplate reader (BioTek), and the OD₆₀₀ was measured at 15-min intervals for at least 48 h. For the opiA mutant and complementation studies, experiments were conducted independently at least six times with very similar results. For the complementation of the opiA mutant with the opiA-emgfp allele, in trans on pSC21, the experiment was repeated twice.

Gentamicin protection assay to determine intracellular growth of F. tularensis LVS strains.

Gentamicin protection assays were used to measure the intracellular growth of F. tularensis LVS strains in in vitro experiments. THP-1 cells (ATCC) were cultivated in RPMI 1640 medium (Gibco) supplemented with 25 mM HEPES (Corning), 1× Glutagro (Corning), 1× sodium pyruvate (Gibco), 0.05 mM β-mercaptoethanol (Gibco), and 10% fetal bovine serum (Gemini Bioproducts) in T-75 flasks (Corning) at 37°C with 5% CO₂. Cells were subcultured when they reached a density of 5 × 10⁸ cells ml⁻¹. RAW 264.7 murine macrophage cells (ATCC) were cultivated in Dulbecco modified Eagle medium (DMEM; Gibco) supplemented with 25 mM HEPES (Corning), 1× Glutagro (Corning), 1× sodium pyruvate (Gibco), and 10% fetal bovine serum (Gemini Bioproducts) in T-75 flasks (Corning) at 37°C with 5% CO₂. RAW 264.7 cells were released by scraping when they reached ∼80% confluence and were subcultured by a 1:30 dilution into fresh medium. For gentamicin protection assays, strains were inoculated from plates into CDM and grown to stationary phase by an overnight incubation at 37°C with agitation. These cultures were diluted to an OD of 0.3 (∼1 × 10⁹ CFU ml⁻¹ as determined by serial dilution and drip plating) and were further diluted to achieve an MOI of 500 (89, 90). Dilutions were made in prewarmed RPMI 1640 or DMEM as appropriate. RAW 264.7 cells were cultured to ∼80% confluence and were released by scraping; THP-1 cells were grown to a density of 5 × 10⁸ cells ml⁻¹ and pelleted by centrifugation at 400 × g for 5 min. Both RAW 264.7 and THP-1 cells washed once in warm medium and used to seed Primaria-coated 96-well plates (Becton, Dickinson, and Co.) at a density of 5 × 10⁴ cells per well and then incubated with F. tularensis LVS strains for 3 h to allow invasion. Cells were washed once in warm Hanks balanced salt solution (HBSS) and then incubated for 1 h at 37°C with 5% CO₂ in fresh medium with 100 μg ml⁻¹ gentamicin (US Biological) to remove any extracellular bacteria. The cells were then washed twice in warm HBSS and then incubated for 48 h at 37°C with 5% CO₂ in fresh medium. Cells were lysed with 0.02 SDS (Bio-Rad) at 0-, 24-, and 48-h time points, and intracellular bacteria were enumerated by serial dilution of the lysates in PBS (Corning) and drip plating. For each time point and condition, wells were set up in triplicate, and three independent replications of the experiments were performed.

ELISA analysis of TNF-α production in THP-1 cells challenged with wild-type and ΔopiA F. tularensis LVS strains in vitro.

To assay production of TNF-α, THP-1 cells were cultured to a density of 5 × 10⁸ cells ml⁻¹, pelleted at 400 × g for 5 min, washed once in warm medium, and incubated in Primaria-coated 96-well plates with opiA mutant and wild-type F. tularensis LVS strains (grown in CDM) containing either the complementation vector, pSC9, or the empty parent vector, pFNLTP8, at an MOI of 10 for 24 h at 37°C with 5% CO₂. Supernatants were analyzed using a human TNF-α DuoSet ELISA kit (R&D Systems) according to the manufacturer’s instructions. Standards and samples were assayed in triplicate, and three independent experiments were performed.

Chicken embryo infection assay.

Chicken embryos (Charles River) were infected with either wild-type or ΔopiA F. tularensis LVS as previously described (33, 47, 86). In total, five experiments were conducted with between 7 and 21 eggs per experiment with a total of 50 eggs for the wild type and 51 eggs for the opiA mutant. In each experiment, 1 × 10⁶ bacteria suspended in PBS were injected into the egg beneath the chorioallantoic membrane, and the infected embryos were incubated with high humidity and gentle rocking at 37°C. Eggs were candled daily to monitor viability for 8 days. Any embryos that expired within 48 h of the initial infection were considered to have suffered lethal trauma during inoculation and were removed from the experiment (four for the wild type and six for the opiA mutant).

Antibiotic resistance assays.

F. tularensis LVS strains were inoculated from plates into CDM and grown to stationary phase by an overnight incubation at 37°C with agitation. These cultures were diluted to an OD of 0.3 (∼1 × 10⁹ CFU ml⁻¹) in PBS. For disk diffusion assays, 1 × 10⁸ bacteria in 100 μl of PBS was spread plated onto solid chocolate II agar. Sterile Whatman filter disks were placed onto the plated bacteria and spotted with 1 μg of either gentamicin (US Biological) or tetracycline in 10 μl of PBS, and the plates were incubated overnight at 37°C with 5% CO₂. The diameters of the zones of inhibition were measured using a metric ruler. Three technical replicates were included for each bacterial strain, and the experiments were repeated three times. For multiblot time-kill assays, 1 × 10⁸ bacteria were incubated with 200, 150, 100, 50, 25, or 0 μg ml⁻¹ of either gentamicin (US Biological), streptomycin, or hygromycin in PBS at 37°C with 5% CO₂ in a 96-well plate (Corning). A 96-pin multiblot replicator was used to spot bacteria onto chocolate II agar plates at 1-, 2-, and 3-h intervals. Overnight incubations were also carried out in some cases. All experiments were performed at least two times, and each experiment had two technical replicates.

Bacterial viability assays.

To measure changes in viability over time, F. tularensis LVS strains were inoculated from plates into CDM or BHIc and grown to stationary phase by an overnight incubation at 37°C with agitation. These cultures were diluted to an OD₆₀₀ of 0.3 for CDM and 0.05 for BHIc in fresh medium and incubated for 3 days. At 1-, 2-, and 3-day time points, a 10-μl aliquot was taken, and serial dilutions were made in PBS to enumerate the CFU per milliliter by drip plating onto chocolate II agar. The experiment was repeated with biological replicates three times.

PI(3)P kinase assay.

PI(3)P from F. tularensis LVS cells was assayed using a PI(3)P mass ELISA kit (Echelon Biosciences, Inc.) according to the manufacturer’s instructions.

Strains were inoculated from plates into CDM and grown to stationary phase by an overnight incubation at 37°C with agitation. These cultures were diluted to an OD of 0.3 and inoculated into fresh CDM medium and then incubated for 24 h. Cultures were normalized by dilution to an OD₆₀₀ of 1.0, and cells from 50 ml of culture for each strain were pelleted by centrifugation at 3,900 rpm for 60 min at 4°C and washed once in ice-cold PBS. Pellets were stored overnight at –80°C. PI(3)P was extracted from cells according to the instructions provided with the PI(3)P ELISA kit, with minor modifications. Cell pellets were suspended in ice cold 0.5 M trichloroacetic acid (TCA; Fisher Scientific), vortexed briefly, and incubated on ice for 5 min. Cells were centrifuged for 10 min at a 3,900 rpm at 4°C. The remaining steps were conducted at room temperature. The pellet was washed twice by vortexing for 30 s in 3 ml of 5% TCA–1 mM EDTA and pelleting for 5 min at 3,000 rpm. Neutral lipids were extracted by two rounds of incubation with 3 ml of methanol (MeOH)-CHCl₃ (2:1; VWR) for 10 min with vortexing, and the pellet was spun down for 5 min at 3,000 rpm between rounds. Acidic lipids were extracted by incubation with 2.25 ml of MeOH–CHCl₃–12 N HCl (80:40:1) for 30 min with vortexing. After centrifugation for 5 min at 3,000 rpm, the supernatant was transferred to fresh tubes and mixed with 0.650 ml of CHCl₃ and 1.35 ml of 0.1 N HCl by vortexing for 30 s. After centrifugation for 5 min at 3,000 rpm, the lower organic phase was transferred to a fresh tube and air dried overnight.

Dried acidic lipid extracts were rehydrated in 245 μl of PBS-Tween buffer with 3% protein stabilizer [supplied with the Echelon Biosciences PI(3)P ELISA kit] by sonication for 10 min in a CPXH 2800 sonicating water bath (Fisher Scientific). Samples and standards were loaded in triplicate, and the assay was performed according to the manufacturer’s instructions. Lipid extracts from E. coli cells were used as negative controls. The experiments were performed independently an additional three times on a smaller scale using one-fifth of the normalized culture medium, and similar results were obtained.

Fluorescence microscopy.

Imaging in all experiments was performed using an Olympus IX73 microscope equipped with a 100×, 1.30-numerical-aperture (NA) phase objective and an ORCA-Flash4.0 LT+ digital CC11440-42U CMOS camera (Hamamatsu). For imaging of bacterial cells, F. tularensis LVS strains were inoculated from plates in BHIc medium (53) at 37°C with agitation for, typically, between 18 and 28 h and spotted directly onto pads of 1% agarose in PBS. Exposure times of 50 ms were used for each channel, and images were processed and analyzed using ImageJ (91) and FIJI (92). All fluorescence images were adjusted for contrast and brightness using wild-type strains with or without an empty vector (pSC13) as a reference to account for natural background fluorescence.

Fluorescence microscopy of the opiA mutant in erythrocytes.

For erythrocyte invasions, F. tularensis LVS strains expressing emgfp on a self-replicating plasmid, pKHEG (created by introducing a 326-bp fragment amplified from pTC3D [52], using the primer pair KH1/KH2 into pSC13), were incubated for 24 h at 37°C with agitation. Bacteria were washed in M5A complete medium and incubated with erythrocytes prepared as described previously (22) at an MOI of 100 for 24 h at 37°C with 5% CO₂. Prior to imaging, the cells were washed once by aspiration and replacement of the culture medium. Bacteria and erythrocytes were incubated in 35-mm dishes with glass bottoms (Ibidi), which allowed direct imaging of cells in the culture dishes. Cells were focused in phase contrast first to ensure that any in focus fluorescence would not be the result of bacteria above or below the RBCs. In fact, any fluorescent bacteria that were not inside erythrocytes were typically moving freely in the culture medium and, as such, could not be imaged at all. Cells were imaged unfixed, since fixation, either by alcohol or paraformaldehyde, introduced significant quenching of the fluorescence signal. We were careful to ensure that fluorescing bacteria were actually within erythrocytes, and we confirmed this by focusing above and below cells to ensure that fluorescence did not originate outside the focal plane of the RBCs. We also conducted experiments using fixed cells, staining with Hoechst 33342 (Life Technologies), and double immunofluorescence microscopy (DIFM) as described previously (22) with F. tularensis-specific antibodies that confirmed the invasion of erythrocytes by the F. tularensis LVS opiA mutant (not shown). Experiments with fluorescently labeled bacteria expressing emgfp were performed independently two times. To quantify the number of fluorescently labeled bacteria inside erythrocytes, images were set to contrast and brightness levels in ImageJ for which no fluorescence in the green channel was detected in control experiments using wild-type F. tularensis LVS containing pSC13 or erythrocytes without bacteria to account for the autofluorescence of both bacteria and erythrocytes. Erythrocytes in five fields of views were scored manually for invasion by fluorescently labeled bacteria for the wild-type and the opiA mutant.

Fluorescence microscopy of OpiA-EmGFP in F. tularensis LVS.

For localization of OpiA-EmGFP produced by expression from the presumed native promoter (on pSC21), strains were grown for between 18 and 24 h in BHIc medium at 37°C with agitation and spotted directly onto pads of 1% agarose in PBS for live imaging. Five independent experiments were performed, and for each experiment at least five randomly selected fields of view were imaged. Consistently, OpiA-EmGFP foci were observed in a very small subset of cells, typically between one and three cells in a given field of view contained a fluorescent focus. Some fields of view contained no foci at all, but foci were detected in every experiment conducted. For localization of OpiA-EmGFP in ΔopiA and iglC-null backgrounds, strains were grown in the same way and imaged as described above. Three independent experiments were conducted in which at least five randomly selected fields of view were imaged for each condition. For the iglC-null mutant expressing opiA-emgfp, identical results were obtained as those seen for the wild type. For the opiA mutant expressing opiA-emgfp OpiA-EmGFP foci were observed in each of the three independent experiments, but the number of foci observed was typically lower, with a majority of fields of view containing no foci. For all experiments, F. tularensis LVS containing a control vector (pSC13) which has a copy of emgfp but no promoter element was grown and imaged under identical conditions and used to determine autofluorescence.

For localization of OpiA-EmGFP produced by expression from the presumed native promoter and the additional constitutive FGRp promoter, strains were grown for between 18 and 24 h in BHIc medium at 37°C with agitation and spotted directly onto pads of 1% agarose in PBS for live imaging. For each condition, the experiment was repeated at least three times with similar results. To compare the frequency of OpiA-EmGFP in strains with either pSC21 or pSC22, a minimum of 700 bacteria were scored for each strain. Wild-type F. tularensis LVS containing promoterless emgfp on pSC13 was used as a fluorescence control.

To visualize dynamic localization of OpiA-EmGFP, F. tularensis LVS containing opiA-emgfp with the presumed native promoter and the constitutive FGRp promoter (on pSC22) were grown for between 18 and 24 h in BHIc medium at 37°C with agitation and spotted directly onto pads of 1% agarose in CDM for live imaging. Cells were left on the microscope at room temperature, and images were taken at 30-min intervals for up to 3 h. Experiments were repeated independently three times, and for each experiment at least two dynamically localizing OpiA-EmGFP foci were observed.

For localization of OpiA-EmGFP on surface-grown cultures, strains were grown for between 18 and 24 h in BHIc medium at 37°C with agitation, and 30 μl was spotted onto a 35-mm dish. A pad of 1% agarose in CDM was then placed on top of the bacteria. Bacteria adhered uniformly to the agarose pad, and the dishes were incubated at 37°C with 5% CO₂. Dishes were removed to room temperature briefly for imaging, with at least five random fields of view being imaged for each experimental condition. The experiment was repeated twice for a strain containing opiA-emgfp expressed from the presumed native promoter (on pSC21) and twice with a strain containing opiA-emgfp with the presumed native promoter and the constitutive FGRp promoter (on pSC22). Similar results were observed for both strains. Wild-type F. tularensis LVS containing promoterless emgfp on pSC13 was used as a fluorescence control. As an additional control, strains containing ftsZ_Ft-emgfp either with only the native promoter or with FGRp and the native promoter (pSC25 and pSC26 [Cantlay and Horzempa, unpublished]) were imaged and showed no accumulation of fluorescent foci in older cultures (data not shown). To quantify the OpiA-EmGFP foci, bacteria from one representative experiment were counted, and the foci were scored manually. At each time point, a subset of cells from each field of view was counted. At least 700 bacteria were counted at each time point.

Fluorescence microscopy of OpiA-EmGFP in THP-1 cells.

To analyze OpiA-EmGFP-specific fluorescence during in vitro infection assays, THP-1 cells were cultured to a density of 5 × 10⁸ cells ml⁻¹, washed once in warm medium, and incubated in 35-mm dishes at 37°C with 5% CO₂ with F. tularensis LVS containing opiA-emgfp expressed from the presumed native promoter (on pSC21) at an MOI of 100. Samples were washed once with warm medium and imaged in the 35-mm dishes at 3-, 24-, and 96-h time points. At least five fields of view were imaged in each experiment, and the experiments were repeated three times. All of the cells imaged for each experiment and at each time point were scored for OpiA-EmGFP-specific fluorescence. At least 129 cells were counted for each time point across the three experiments. Uninfected THP-1 cells or THP-1 cells infected with wild-type bacteria or F. tularensis LVS containing the empty vector, pSC13, were used as negative controls for fluorescence.

OpiA-EMGFP secretion assay.

To analyze OpiA-EmGFP fluorescence in THP-1 cell lysates, cells grown to a density of 1 × 10⁶ ml^–1 were washed once, and 1 × 10⁶ cells were mixed with F. tularensis LVS strains that had been grown to stationary phase overnight in CDM at an MOI of 100. THP-1 cells and bacteria were incubated for 24 h at 37°C with 5% CO₂. Cells were spun down at 300 × g for 5 min and washed to remove any extracellular bacteria. The pellets were resuspended in 1 ml of 0.02% SDS in PBS to lyse the cells. The cell lysates were filtered through a 0.22-μm syringe filter (Millipore) to remove intracellular bacteria. The fluorescence of 200-μl volumes of lysate was measured in an Eppendorf AF2200 plate reader with excitation/emission set to 485/535 nm with a 10-nm bandwidth for both. The experiment was repeated independently with four biological replicates, and THP-1 cells infected with F. tularensis LVS containing the empty vector, pSC13, were used as a negative control.

Fluorescence microscopy of OpiA-EmGFP in erythrocytes.

For localization of OpiA-EmGFP in erythrocytes, both bacterial strains and erythrocytes were prepared as described previously (22). Bacteria and erythrocytes were incubated in 35-mm dishes at 37°C with 5% CO₂. The experiment was repeated twice for F. tularensis LVS containing opiA-emgfp expressed from the presumed native promoter (on pSC21) and three times with a strain containing opiA-emgfp with the presumed native promoter and the constitutive FGRp promoter (on pSC22), with similar results. Wild-type F. tularensis LVS containing promoterless emgfp on pSC13 was used as a fluorescence control. Erythrocytes were imaged in the 35-mm dishes at 3-, 24-, and 72-h time points. Erythrocytes were not washed prior to imaging for these experiments, and no fluorescence was observed for extracellular bacteria. To quantify OpiA-EmGFP specific fluorescence in RBCs, for each time point, four fields of view from a representative experiment for each condition were scored for fluorescence. At least 139 erythrocytes were scored for each condition at each time point.

Statistical analyses.

Data were analyzed using GraphPad Prism software. The tests used and the P values obtained are presented in Results and the figure legends. Bacterial cell measurements were made using ImageJ.