Flagella Facilitate Escape of Salmonella from Oncotic Macrophages

Abstract

The intracellular parasite Salmonella enterica serovar Typhimurium causes a typhoid-like systemic disease in mice. Whereas the survival of Salmonella in phagocytes is well understood, little has been documented about the exit of intracellular Salmonella from host cells. Here we report that in a population of infected macrophages Salmonella induces “oncosis,” an irreversible progression to eukaryotic cell death characterized by swelling of the entire cell body. Oncotic macrophages (OnMφs) are terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling negative and lack actin filaments (F-actin). The plasma membrane of OnMφs filled with bacilli remains impermeable, and intracellular Salmonella bacilli move vigorously using flagella. Eventually, intracellular Salmonella bacilli intermittently exit host cells in a flagellum-dependent manner. These results suggest that induction of macrophage oncosis and intracellular accumulation of flagellated bacilli constitute a strategy whereby Salmonella escapes from host macrophages.

Macrophages detect and internalize bacterial pathogens in order to eliminate them from the host. Bacteria captured in phagosomes are usually killed by reactive oxygen species and nitrogen intermediates, acidification, starvation, lysosomal enzymes, antimicrobial peptides, or other activities. However, a subset of bacteria has evolved to survive within various types of vacuoles in macrophages.

The gram-negative bacterium Salmonella enterica serovar Typhimurium causes enterocolitis in humans and a systemic typhoid-like disease in mice. Intracellular Salmonella bacilli are found in Salmonella-containing vacuoles (SCVs), which localize around the endoplasmic reticulum (3, 18). Although SCVs are acidified, Salmonella bacilli can survive at low pH by remodeling their outer membrane (13). SCVs do not contain bactericidal enzymes such as NADPH oxidase, lysosomal enzymes, and inducible nitric oxide synthase (7, 57, 58). By contrast, Mycobacterium avium-containing vacuoles are not acidified, since the vacuolar proton ATPase is excluded from the membrane, and they do not mature into phagolysosomes (16). Legionella pneumophila-containing vacuoles are converted into rough endoplasmic reticulum-like structures and avoid recognition as phagosomes (30, 53). Little is known, however, about how any of these intracellular pathogens escapes from host phagocytes.

Genes in Salmonella pathogenicity island 1 (SPI1) are required mainly for host cell invasion and are down-regulated inside Salmonella-infected host cells (12). Within SCVs in macrophages, Salmonella also down-regulates synthesis of flagellar proteins in response to low pH (1, 12). Instead, intracellular survival and replication of Salmonella require genes located in a second locus, SPI2 (51), whose expression in host cells is activated by low osmolarity and acidic pH (11, 36).

While surviving in macrophages, Salmonella induces host cell death via diverse mechanisms at different times after infection (23, 31). Salmonella invasion protein B (SipB) is a major bacterial effector protein that induces macrophage cell death. SipB is encoded by SPI1 (24) and induces rapid (∼1 h) macrophage cell death through activation of the host protein caspase-1, resulting in chromatin fragmentation and plasma membrane blebbing (5, 8, 29, 44). Although apoptosis and necrosis are terms that are widely used to define most forms of eukaryotic cell death, such terminology may be an oversimplification. Recently, a new concept has emerged, that apoptosis and necrosis are not mutually exclusive but rather necrosis is the end stage of any cell death, including apoptosis (15, 23, 31, 39). Salmonella induces cell death through caspase-3-dependent apoptosis (32), autophagy (via degradation of cytosolic components) (28), and pyroptosis, which is caspase-1 dependent but caspase-3 independent (4). Other host factors, such as c-Fos/AP-1, also affect Salmonella-induced macrophage cell death (40).

Here, we report that intracellular Salmonella triggers swelling of macrophages, an event termed oncosis (from “onkos,” meaning swelling) (15, 39). We observed that “oncotic” macrophages (OnMφs) are often packed with motile Salmonella bacilli and that later, flagellated Salmonella bacilli escape intermittently from OnMφs, which then undergo necrotic cell death. These results reveal a novel strategy by which Salmonella survives in, accumulates in, and escapes from macrophages.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Bacterial strains were derived from the wild-type S. enterica serovar Typhimurium SR-11 strain χ3181 or χ3306 (Nal^r) (25). Strain MS005 expressing green fluorescent protein was generated by introducing the pGFPmut3.1 plasmid (Clontech) into χ3306. Strain MS300 lacking flagella (ΔfliA) was constructed using bacteriophage P22 mutants (50) for transduction of fliA::Tn10 in the Salmonella LT2 strain KK2091 (35) into χ3181. Strain MS007 lacking phoP (phoP::aphT) has been described previously (41). Bacteria were grown in L broth or on L agar (Difco Laboratories) supplemented with appropriate antibiotics, including 15 μg/ml tetracycline, 100 μg/ml ampicillin (Sigma-Aldrich), and 25 μg/ml nalidixic acid (Nacalai Tesque).

Anti-Salmonella mouse serum.

Female 6-week-old BALB/c mice (Charles River Japan) were orally inoculated with 1 × 10⁸ CFU of an exponential-phase nonvirulent Salmonella ΔphoP strain in 20 μl phosphate-buffered saline (PBS) containing 0.01% (wt/vol) gelatin, as described previously (41). Mice were inoculated using three 10-day cycles, and sera were prepared 2 weeks later. Increases in the level of Salmonella lipopolysaccharide-specific immunoglobulin G (IgG) in serum (11.6 ± 8.2 μg/ml) were confirmed using an enzyme-linked immunosorbent assay (42). All experiments were performed according to institutional guidelines for animal experiments.

Cell culture and infection.

RAW 264.7 (= ATCC TIB-71) and J774A.1 (= ATCC TIB-67) macrophages were seeded in Dulbecco modified Eagle medium (DMEM) (Sigma-Aldrich) supplemented with 10% fetal calf serum (Gibco) at a density of 3 × 10⁵ cells/well in 24-well plates (Falcon) or on glass coverslips (Fischer Science). Peritoneal macrophages were prepared from female 6-week-old BALB/c mice as described previously (48). Salmonella strains were cultured at 37°C overnight without agitation in L broth supplemented with antibiotics. The following day, bacteria were diluted 1:30 in L broth containing antibiotics and agitated for about 2 h at 37°C until the optical density at 600 nm reached 0.3. Unless otherwise indicated, Salmonella strains were then opsonized by incubation with 0.01 volume of anti-Salmonella mouse serum for 15 min at 37°C. Macrophages were infected with opsonized bacteria at a multiplicity of infection (MOI) of 100 unless otherwise indicated. One hour later, cells were gently washed twice with PBS and once with medium and then immersed in complete medium supplemented with 10 μg/ml gentamicin (Gibco). Apoptotic RAW 264.7 cells were prepared by incubating cells in 0.1% H₂O₂-DMEM for 30 min at room temperature.

Transmission electron microscopy.

Four hours after infection, RAW 264.7 cells were fixed for 1 h with 2.5% glutaraldehyde-PBS, washed three times in PBS, and postfixed in 1% OsO₄ in Sorensen's buffer for 1 h. Samples were subsequently dehydrated in ethanol and flat embedded in epoxy resin (Agar100; Agar Scientific). Thin sections (60 to 80 nm) mounted on copper grids were stained with uranyl acetate and lead citrate and viewed at 120 kV with an electron microscope (JEM1230; JEOL).

Video analysis.

Microscopy images (DMIL; Leica) were video recorded with a micro camera (GP-KS1000; Panasonic) using digital imaging software (Ulead VideoStudio version 6 and Adobe Premiere Pro 2.0). Bacterial velocity was calculated using an image-analyzing system (Move-tr/2D; Library).

Staining procedures.

For immunostaining, DMEM was gently removed from cultures and cells were fixed for 5 min at room temperature in 4% paraformaldehyde-PBS. After two washes with PBS, cells were incubated in 0.1% Triton X-100-PBS for 5 min, washed twice with PBS, and stained. The primary antibodies, all diluted 1:100, were monoclonal anti-Salmonella goat antibody CSA-1 (Accurate Chemical & Scientific), anti-Salmonella flagellum H-i rabbit sera (Denka Seiken), and monoclonal anti-CD18 (M18/2) rat antibody (Pharmingen). The secondary antibodies, which were diluted 1:200, were Alexa Fluor 647-conjugated anti-rat IgG chicken antibody, Alexa Fluor 488-conjugated anti-rabbit IgG chicken antibody, and Alexa Fluor 647-conjugated anti-goat IgG rabbit antibody (Molecular Probes). F-actin was detected using Alexa Fluor 546 phalloidin (Molecular Probes) at a 1:100 dilution. Nuclei were stained with 4′,6′-diamidino-2-phenylindole (DAPI) (Vector). A fluorescence microscope (Axiovert 135; Zeiss) and a confocal laser scanning microscope (FV1000; Olympus) were used for imaging. Giemsa staining was performed using Giemsa's solution (Nacalai Tesque). Caspase activities were detected by using APO LOGIX carboxyfluorescein caspase detection kits (Cell Technology) as described previously (2). The terminal deoxynucleotidyltransferase-mediated biotin-dUTP nick end labeling (TUNEL) assay was performed using a DeadEnd colorimetric TUNEL system kit (Promega). Plasma membrane integrity was analyzed using a LIVE/DEAD viability/cytotoxicity kit for mammalian cells (Molecular Probes).

Measurement of the intracellular Ca²⁺ concentration ([Ca²⁺]_i).

RAW 264.7 cells were infected with Salmonella in 35-mm glass bottom dishes (IWAKI) for 5 h. The culture medium was replaced with a solution containing 125 mM NaCl, 5 mM KCl, 1.2 mM MgSO₄, 2 mM CaCl₂, 1.2 mM KH₂PO₄, 6 mM glucose, 25 mM HEPES (pH 7.4), and the acetoxymethyl ester form of the radiometric fluorescent Ca²⁺ indicator fura-2 (fura-2 AM; Molecular Probes) at a concentration of 5 μM. Cells were incubated for 30 min, washed twice in the same buffer without fura-2 AM, and incubated in DMEM for 30 min. Fura-2 imaging was performed as described previously (20, 33).

Purification of CD11b-positive cells.

Bacterial escape was analyzed after infected RAW 264.7 cells were separated from extracellular Salmonella bacilli by incubation with streptavidin-conjugated anti-mouse CD11b (M1/70; BD Biosciences) in DMEM supplemented with 10% fetal calf serum for 15 min at 37°C. Cells were purified using IMag streptavidin Particle Plus-DM and the IMag cell separation system (BD Biosciences). Bacterial escape was monitored in the absence of gentamicin.

Statistical analysis.

Statistical comparisons were performed using Student's t test.

RESULTS

Formation of OnMφs after Salmonella infection.

We first infected RAW 264.7 cells with wild-type Salmonella strain χ3306 at an MOI of 100 after bacteria were opsonized with anti-Salmonella antisera, and then we examined infected cells over time by light microscopy. From 4 to 6 h after infection, we observed that 10 to 20% of macrophages resembled inflated balloons and had flexible and translucent cell membranes characteristic of OnMφs (15, 39). Moreover, bacteria were vigorously moving within 10 to 30% of the OnMφs (Fig. 1A; see Movie S1 in the supplemental material). By contrast, in mock infection controls, we did not observe OnMφs, indicating that Salmonella triggers OnMφ formation (data not shown). Formation of OnMφs, as judged by swelling and the lack of a sharp boundary of the cell body, was also observed with J774A.1 and peritoneal macrophages (Fig. 1B, C, and D). A fraction of the OnMφs in all three cell types contained motile Salmonella bacilli (Fig. 1D). Since vigorous movement of intracellular Salmonella bacilli was most readily observed with RAW 264.7 cells by light microscopy, we chose these cells for further analysis.

FIG. 1.

OnMφ formation. (A to C) OnMφs (arrowheads) formed in cultures of RAW 264.7 (A) and J774A.1 (B) cells and peritoneal macrophages (C) 4 h after infection at an MOI of 100 with wild-type Salmonella strain χ3306 opsonized with anti-Salmonella antisera. Scale bars = 20 μm. Also see Movie S1 in the supplemental material. (D) Percentage of OnMφs (dotted and filled bars) based on the total number of macrophages. The filled bars indicate the fraction of OnMφs with vigorously moving intracellular Salmonella bacilli observed with the light microscope. The data are means ± standard errors of the means. (E) Effects of opsonization and MOI on OnMφ formation. RAW 264.7 cells were infected with wild-type Salmonella strain χ3306 opsonized with anti-Salmonella antisera (AS) or normal mouse sera (N) or not opsonized (−) at MOIs of 5 and 100. (F) Effects of IFN-γ on OnMφ formation. RAW 264.7 cells were preincubated with 0, 1, 10, and 100 nM IFN-γ for 24 h and then infected with wild-type Salmonella strain χ3306. Infected cells were analyzed 4 h after infection. (G to J) Transmission electron micrographs of uninfected RAW 264.7 cells (G) and infected RAW 264.7 cells containing intracellular Salmonella bacilli (H to J). Scale bars = 2 μm. Salmonella bacilli are located in discrete SCVs (H), in a large common SCV (I), or in the diluted cytosol (J).

We quantified OnMφ formation with and without opsonization and at MOIs of 5 and 100. Infection at the higher MOI, as well as opsonization, enhanced OnMφ formation, although even without opsonization and at an MOI of 5 OnMφs consistently formed (Fig. 1E). We also asked whether pretreatment of RAW 264.7 cells with gamma interferon (IFN-γ) enhanced OnMφ formation. IFN-γ pretreatment followed by mock infection did not induce OnMφ formation (data not shown). By contrast, IFN-γ pretreatment of RAW 264.7 cells for 24 h increased the proportion of OnMφs (Fig. 1F) even though the number of OnMφs containing motile Salmonella bacilli was not increased by IFN-γ pretreatment (Fig. 1F). Since formation of OnMφs containing motile Salmonella bacilli was most frequently observed without IFN-γ pretreatment, with opsonization, and at an MOI of 100, we used these conditions in experiments described below. Using transmission electron microscopy, we next observed control uninfected RAW 264.7 cells (Fig. 1G) and infected RAW 264.7 cells containing abundant intracellular Salmonella bacilli (Fig. 1H to J). Salmonella bacilli were located either in discrete SCVs (Fig. 1H), in a large common SCV (Fig. 1I), or in the diluted cytosol, which was indicative of necrotic cell death (Fig. 1J).

Cytoskeletal reorganization.

We next monitored OnMφ formation by time-lapse video microscopy 4 to 6 h after infection. Prior to swelling, Salmonella movement was limited to a small region of the cytoplasm. Swelling was often complete in 5 min (Fig. 2A). Such morphological changes enabling formation of gigantic SCVs suggest that there is cytoskeletal reorganization (59).

FIG. 2.

F-actin dissociation in OnMφs. (A) Time-lapse video microscopy of RAW 264.7 cells forming an OnMφ 4 h after infection. The white dots indicate the cell boundary. (B to D) Confocal microscopy of uninfected (left panels) and infected (right panels) RAW 264.7 cells 4 h after infection. OnMφs are indicated by arrowheads. Samples were stained for F-actin (red) (B to D), Salmonella (green) (B to D), and CD18 (blue) (B), β-tubulin (blue) (C), or vimentin (blue) (D). (E) Confocal laser microscopy analysis of infected RAW 264.7 cells. Four hours after wild-type Salmonella strain χ3306 infection, infected cells were fixed and stained for TUNEL (pink), Salmonella (green), and F-actin (red). The arrowheads indicate TUNEL-negative, Salmonella-positive, and F-actin-negative macrophages (oncotic). The arrows indicate TUNEL-positive macrophages (apoptotic). There were few TUNEL-negative and F-actin-negative cells lacking intracellular Salmonella bacilli (asterisks). Scale bars = 20 μm.

Using confocal microscopy, we observed that while F-actin was located beneath the plasma membrane in uninfected macrophages, OnMφs were uniformly negative for F-actin (Fig. 2B) but remained positive for the macrophage marker CD18 (Fig. 2B). To determine whether other cytoskeletal filaments were reorganized in OnMφs, we stained cells for microtubules and intermediate filament proteins and detected no obvious changes in β-tubulin and vimentin in OnMφs containing gigantic SCVs (Fig. 2C and D). Taken together, these data indicate that while OnMφ formation has little effect on the overall cytoskeletal organization, it is accompanied by F-actin dissociation. Confocal microscopy analysis showed that most F-actin-negative OnMφs contained intracellular Salmonella bacilli (Fig. 2E) but a few OnMφs contained no detectable intracellular bacilli (Fig. 2E), suggesting that intracellular rather than extracellular Salmonella bacilli trigger OnMφ formation (for TUNEL staining, see below).

Lack of DNA damage in Salmonella-induced oncosis.

By 4 h after infection, more than 20 bacilli were often seen in a single OnMφ, as revealed by Giemsa staining (Fig. 3A). Staining also showed that such macrophages lacked chromatin condensation and DNA damage even in the presence of numerous Salmonella bacilli (Fig. 3A).

FIG. 3.

Biochemical features of Salmonella-induced OnMφs. (A) Giemsa staining of OnMφs containing more than 50 bacilli (arrowhead) 4 h after infection of RAW 264.7 cells. (B) Calcium levels in RAW 264.7 cells. The [Ca²⁺]_i of uninfected macrophages (n = 25), OnMφs containing motile Salmonella (n = 21), and H₂O₂-treated apoptotic macrophages (n = 12) were determined. Asterisk, P < 0.0001 for a comparison with uninfected cells. (C) Caspase-1 and caspase-3 activities in OnMφs. Uninfected, Salmonella-infected, and H₂O₂-treated apoptotic macrophages (control) were examined. OnMφs containing motile Salmonella bacilli are indicated by arrowheads. (D) TUNEL-positive macrophages (left panel, red nuclei, arrows) containing a few GFP-expressing Salmonella bacilli (green) and a TUNEL-negative OnMφ (right panel, arrowhead) containing numerous bacilli (green) 4 h after infection. Note that the images in the left and right panels are from the same field and are the same magnification. DAPI staining is blue. (E) Permeability of the plasma membrane of OnMφs (arrowheads). Live macrophages with intact membranes are stained green. Dead macrophages with permeable membranes are stained red (arrows). BF, bright field; DIC, differential interference contrast; IF, immunofluorescence. Scale bars = 20 μm.

To reveal biochemical features of Salmonella-induced OnMφs, we first measured the [Ca²⁺]_i using fluorescent imaging. Both OnMφs and apoptotic cells showed higher [Ca²⁺]_i than uninfected cells, but there was no significant difference between oncotic and H₂O₂-treated apoptotic macrophages (Fig. 3B). Since Salmonella cells produce proapoptotic effectors, such as SipB (5, 8, 29, 44), we measured caspase-1 and caspase-3 activities and found that OnMφs with motile Salmonella bacilli showed both caspase-1 activity (six of six OnMφs) and caspase-3 activity (eight of nine OnMφs) (Fig. 3C). To determine whether DNA damage occurs in OnMφs, we employed a TUNEL assay. Although we observed TUNEL-positive cells that did not show cellular swelling, OnMφs containing numerous bacilli were TUNEL negative (Fig. 3D), supporting the idea that OnMφs do not undergo apoptosis. Confocal microscopy also revealed both TUNEL-negative OnMφs (Fig. 2E) and TUNEL-positive apoptotic macrophages (Fig. 2E).

To analyze the integrity of the host cell plasma membrane, we stained cells with two fluorescent compounds, calcein AM, which passes through intact membranes and detects esterase activity in the cytosol (green), and ethidium homodimer-1, which stains dead cells with permeable membranes (red). Six of 10 swollen macrophages containing motile Salmonella bacilli showed weak green staining and were either negative for or only faintly stained with ethidium homodimer-1, indicating that their plasma membranes were largely intact (Fig. 3E); the other four cells in this group were stained only with the red stain, indicating that these cells underwent cell death (Fig. 3E). These data indicate that over one-half of the swollen macrophages containing motile bacteria had impermeable membranes and showed esterase activity.

Salmonella motility in OnMφs.

Although we observed vigorously moving Salmonella bacilli in OnMφs (Fig. 1A, B, and C; see Movie S1 in the supplemental material), flagellum expression is believed to be suppressed inside host cells (12). To determine whether Salmonella motility is flagellum dependent, we performed immunofluorescence staining with an antiflagellum antibody combined with F-actin and TUNEL staining. Confocal laser microscopy revealed that F-actin- and TUNEL-negative OnMφs contained Salmonella bacilli that stained positive for flagella (Fig. 4A).

FIG. 4.

Motility of Salmonella bacilli in OnMφs is driven by flagella. (A) Confocal microscopy of OnMφs (arrowheads) and apoptotic macrophages (arrow) derived from RAW 264.7 cells. Four hours after infection, macrophages were fixed and stained for F-actin (red) and flagella (green). Apoptotic cells are TUNEL positive (pink). (B) OnMφ formation by flagellum-deficient Salmonella. RAW 264.7 cells were infected with the χ3181-derived mutant lacking flagella (ΔfliA mutant). The bars indicate the percentages of OnMφs; vigorously moving intracellular Salmonella bacilli were not observed. The data are means ± standard errors of the means. AS, anti-Salmonella antisera; N, normal mouse sera. (C) Traces of Salmonella (lines) were recorded and analyzed for the wild-type strain (wt) and the ΔfliA mutant by time-lapse microscopy in a liquid medium (left panels) and in OnMφs 4 h after infection (right panels). The white dots indicate cell boundaries. Each trace was analyzed for 5 s. Scale bars = 20 μm. (D) Maximal velocity (V_max) of wild-type and ΔfliA Salmonella strains in liquid medium and in OnMφs. The data are means ± standard deviations from five traces shown in panel C. Two asterisks, P < 0.001 for a comparison with wild-type controls.

To further determine whether Salmonella motility in OnMφs was flagellum driven, we infected RAW 264.7 cells with a Salmonella ΔfliA mutant which lacked flagella due to a mutation in an alternative sigma factor required for flagellum synthesis (35). Under the various conditions tested, the efficiency of OnMφ formation by the Salmonella ΔfliA mutant (Fig. 4B) was not significantly different from that of wild-type Salmonella (Fig. 1E). Nonetheless, we did not observe vigorously moving intracellular Salmonella ΔfliA mutant bacilli by light microscopy (Fig. 4C). Consistently, time-lapse video microscopy showed that the maximum velocities of wild-type Salmonella bacilli in culture medium (68 μm/s) and in OnMφs (37 μm/s) were significantly greater than that observed for the Salmonella ΔfliA mutant (12 μm/s in culture medium and 3 μm/s in OnMφs) (Fig. 4C and D). We also observed that the intracellular wild-type Salmonella but not the Salmonella ΔfliA mutant rebounded from the host cell membrane in OnMφs. These data indicate that a subset of intracellular Salmonella bacilli is flagellated and that their motility is flagellum driven. Furthermore, induction of OnMφ formation per se does not require a flagellum.

Exit of Salmonella from OnMφs.

We next performed time-lapse video microscopy of RAW 264.7 cells 4 to 6 h postinfection after we separated the cells from extracellular Salmonella bacilli using CD11b-conjugated magnet beads and cultured them in the absence of gentamicin. We observed swollen macrophages, which released wild-type Salmonella from one or a few loci in the plasma membrane (Fig. 5A). We found that one Salmonella bacillus exited from a cell per minute over more than 1 h of observation (Fig. 5B). These data suggest that after F-actin dissociation, some Salmonella bacilli moving at random penetrate the host cell membrane, potentially making the locus more vulnerable to penetration by other bacilli. We determined the number of flagellated Salmonella bacilli in each host cell 6 h after infection using antisera against flagella. In contrast to the numerous Salmonella bacilli detected by Giemsa staining 4 h after infection (Fig. 3A), most (68%) swollen macrophages lacking F-actin contained no or fewer than six flagellated Salmonella bacilli 6 h after infection (Fig. 5C). This discrepancy may have been due to degradation of intracellular Salmonella in OnMφs, to the presence of flagellum-free Salmonella not detectable by this assay, or to the escape of flagellated Salmonella bacilli from host cells.

FIG. 5.

Escape of Salmonella bacilli from host cells. (A) Time-lapse microscopy showing Salmonella bacilli (arrowhead) exiting from a swollen macrophage at 5 h postinfection. The white dots indicate the host cell boundary. (B) Numbers of Salmonella bacteria exiting from a single host cell in panel A per minute for 70 min. (C) Histogram showing the percentages of the total macrophages containing different numbers of flagellated Salmonella bacilli per cell 6 h after infection. Cross-hatched bars, macrophages with intact F-actin; filled bars, macrophages with F-actin dissociation. (D) Confocal microscopy of RAW 264.7 cells infected with the wild-type Salmonella strain (wt) or the mutant lacking flagella (ΔfliA). Two hours after infection, infected RAW 264.7 cells were transferred to gentamicin-free medium, cultured for an additional 6 h, fixed, and stained for F-actin (red), Salmonella (green), and DAPI (blue). Scale bar = 20 μm.

To determine whether intracellular Salmonella bacilli escape from host macrophages in a flagellum-dependent manner, we observed RAW 264.7 cells infected with the wild-type strain or the Salmonella ΔfliA mutant lacking flagella for an extended period of time in the absence of gentamicin (46). Eight hours after infection with wild-type Salmonella, most macrophages lacked motile intracellular bacteria. By contrast, the Salmonella ΔfliA mutant remained abundant in host cells (Fig. 5D). This observation strongly suggests that after inducing OnMφ formation, intracellular Salmonella bacilli exit from the host cell in a flagellum-dependent manner.

DISCUSSION

In this study, we showed that intracellular Salmonella bacilli escape from macrophages after induction of OnMφ formation (Fig. 6). Salmonella-induced OnMφs have three attributes: (i) F-actin is dissociated beneath the cell membrane, (ii) they are TUNEL negative, and (iii) they often contain abundant, highly motile, flagellated Salmonella bacilli. OnMφs in which there is vigorous movement of Salmonella bacilli in gigantic SCVs are still viable because they preserve plasma membrane impermeability. Later, flagellated Salmonella bacilli escape intermittently from the host cell over an extended period of time.

FIG. 6.

Model for escape of Salmonella from OnMφs. Salmonella bacilli are confined to SCVs after they are engulfed by macrophages. Some infected cells undergo apoptosis, pyroptosis, or autophagy. Other infected cells become OnMφs (undergo oncosis), contain flagellated motile Salmonella bacilli, and lack F-actin, but they contain intact DNA. Intracellular Salmonella bacilli continuously exit from the cell. N, host cell nucleus.

Oncosis is defined as eukaryotic cell death accompanied by cellular swelling (15, 39). Bacteria such as Shigella flexneri, Pseudomonas aeruginosa, and Brucella abortus reportedly induce host cell oncosis (10, 14, 45, 47). Salmonella-induced delayed cell death is also associated with enlarged macrophages (5, 8, 22).

Although the mechanism of oncosis is unclear, it has been suggested that depleted ATP or increased levels of intracellular calcium ions lead to inefficient functioning of ion channels and resultant cellular swelling (56). Salmonella-induced oncosis is characterized by F-actin dissociation and is flagellum independent. It is currently not known whether Salmonella induces oncosis through ATP depletion or by acting directly on F-actin. Salmonella expresses various proteins functioning in F-actin regulation during host cell invasion (17, 26, 27, 37, 43, 52, 60). In addition, deletion of SpvB (Salmonella plasmid virulence B), which is secreted by intracellular Salmonella bacilli, exhibits ADP-ribosyltransferase activity, and modifies actin (34, 37, 54), resulted in reduced OnMφ formation in vitro and impaired intracellular proliferation of Salmonella and survival in mice (unpublished observations). Additional studies are needed to determine precise roles of SpvB and other F-actin regulatory proteins in induction of macrophage oncosis. Activation of macrophages by IFN-γ enhanced Salmonella-induced oncosis but not the proportion of OnMφs containing motile bacteria, presumably due to the increased bactericidal activity in host cells. Therefore, OnMφs packed with motile Salmonella bacilli might be formed only during a narrow window of survival in activated host macrophages.

OnMφs exhibited caspase-1 and -3 activities. However, Salmonella-induced OnMφs were not TUNEL positive, suggesting that Salmonella-induced OnMφs do not undergo DNA damage characteristic of apoptosis. Similarly, in Pseudomonas-induced oncosis, host cells also do not exhibit DNA damage (10).

Our data indicate that intracellular motility of Salmonella is flagellum driven. The velocity of intracellular wild-type Salmonella (37 μm/s) was much greater than that of Listeria, Shigella, or Rickettsia due to formation of “comet tails,” which require actin polymerization (up to 0.5 μm/s) (9, 19, 49, 55) or microtubule motors (0.02 μm/s) (21). Salmonella ΔfliA mutants lacking flagella also showed significantly reduced motility outside and inside host cells. Third, intracellular Salmonella bacteria in OnMφs were positive for an antiflagellum antibody, although we cannot exclude the possibility that flagella were detached from Salmonella cells (38). Finally, highly motile Salmonella bacilli escaped from host cells, while flagellumless ΔfliA mutants did not.

Intracellular flagellation was unanticipated, since it has been reported that intracellular Salmonella bacilli down-regulate flagellum synthesis (12). Salmonella bacilli in OnMφs do not retain flagella through the early stages of infection; we observed that most intracellular Salmonella bacilli were flagellum negative 2 h after infection (data not shown). However, some bacilli were flagellum positive 4 h later, suggesting that flagella are regenerated. We cannot, however, exclude the possibility that some Salmonella bacilli retain flagella throughout infection. Interestingly, flagellum reexpression by intracellular bacteria has also been reported in Legionella (6). Since flagellated Salmonella bacilli exit from host cells by 8 h after infection, it is likely that Salmonella bacilli inside OnMφs down-regulate genes required for flagellum synthesis, as has been reported previously (12), but then up-regulate them prior to exiting from host cells.

Our results indicate that Salmonella induces macrophage oncosis and accumulates in OnMφs, which is followed by intermittent escape of Salmonella bacilli from host cells via flagella. As escape of phagocytosed bacteria from macrophages is critical for establishment of infectious diseases caused by intracellular pathogens, investigation of the possible link between pathogenesis and mechanisms by which pathogens escape from macrophages is warranted.