Flagellin-deficient Legionella mutants evade caspase-1- and Naip5-mediated macrophage immunity

Abstract

Macrophages from C57BL/6J (B6) mice restrict growth of the intracellular bacterial pathogen Legionella pneumophila. Restriction of bacterial growth requires caspase-1 and the leucine-rich repeat-containing protein Naip5 (Birc1e). We identified mutants of L. pneumophila that evade macrophage innate immunity. All mutants were deficient in expression of flagellin, the primary flagellar subunit, and failed to induce caspase-1-mediated macrophage death. Interestingly, a previously isolated flagellar mutant (fliI) that expresses, but does not assemble, flagellin did not replicate in macrophages, and induced macrophage death. Thus, flagellin itself, not flagella or motility, is required to initiate macrophage innate immunity. Immunity to Legionella did not require MyD88, an essential adaptor for toll-like receptor 5 (TLR5) signaling. Moreover, flagellin of Legionella and Salmonella induced cytotoxicity when delivered to the macrophage cytosol using Escherichia coli as a heterologous host. It thus appears that macrophages sense cytosolic flagellin via a TLR5-independent pathway that leads to rapid caspase-1-dependent cell death and provides defense against intracellular bacterial pathogens.

Figure 1. Characterization of Legionella Mutants Harboring Transposon Insertions in Flagellin (flaA)

(A) Nucleotide positions of transposon insertions in flaA. (B) Western blotting for flagellin (FlaA). Proteins from pelleted bacteria (top) or TCA-precipitated culture supernatants (bottom) obtained from stationary-phase cultures were separated by SDS-PAGE, blotted, and probed with an anti-FlaA monoclonal antibody. ΔflaA ahpC::flaA is the ΔflaA mutant which has been complemented with a copy of flaA inserted on the Legionella chromosome just after the ahpC locus. (C) Electron microscopy showing the presence or absence of flagella on various Legionella strains. Representative bacteria (of >100 surveyed) are shown. No flagellated bacteria were seen for any of the strains shown as lacking flagella. Tn, transposon.

Figure 2. Growth of Legionella in Bone Marrow–Derived Macrophages

Growth of Legionella strains in bone marrow–derived macrophages from (A) B6 mice or (B) B6.A-Chr13 mice (which carry the permissive A/J allele of Naip5). A confluent layer of macrophages were grown in 24-well plates and were infected with the indicated Legionella strains at an MOI of 0.02. Growth of wild-type Legionella (LP02) in B6-backcrossed MyD88^−/− macrophages is also depicted in (A). After addition of bacteria, the plate was spun at 400 g for 10 min. One hour after infection, the media was changed. At daily intervals (starting with 1 h post-infection), a timepoint was taken. Macrophages were lysed with sterile water and the lysate combined with supernatant from the same well. Colony-forming units per well were determined by plating dilutions onto BCYE plates.

Figure 3. Rapid Lysis of Macrophages in Response to Flagellin-Expressing Legionella

In all experiments shown, bacteria were added at an MOI of 2, and were spun onto the macrophages at 400 g for 10 min. Cells were assayed 4 h after infection. LP02 is the wild-type strain and is isogenic with the mutants utilized in these experiments. (A) Release of the intracellular enzyme LDH by B6 and MyD88^−/− macrophages with indicated Legionella strains, including four mutants with transposon insertions in flaA. One hundred percent release is set as the amount of LDH released by detergent-treated macrophages (minus spontaneous release). (B) LDH release by B6, B6.A-Chr13, and B6 caspase-1-deficient macrophages infected with the indicated Legionella strains. FlaAΔ56 is an LP02-derived strain that contains an EMS-induced point mutation in flaA, resulting in a truncation of the last 56 amino acids of FlaA. (C) Cell death of macrophages was assessed by degree of failure to take up neutral red in a 4-h assay. At least 100 macrophages were counted for each condition; one representative experiment of two is shown. (D) Cell death was quantified by assessing permeability to ethidium bromide-2 homodimer. At least 400 macrophages were counted for each condition; one representative experiment of two is shown.

Figure 4. Flagellin-Induced Cytotoxicity of Macrophages Is Independent of Motility but Requires Type IV Secretion

(A and B) Centrifugation is required to promote cytotoxicity of a fliI mutant, but does not restore cytotoxicity of a flaA mutant. The calcium ionophore A23187 (1 μM, or a vehicle control, DMSO) was added to some wells simultaneously with the addition of bacteria. Infection and LDH release were assayed as in Figure 3, with or without centrifugation immediately after infection, as indicated. (C) Coinfection of macrophages with a ΔflaA and ΔdotA mutant does not restore cytotoxicity. Infection and LDH release were assayed as in Figure 3, except that higher MOIs were used and chilled macrophages were infected, as indicated.

Figure 5. Salmonella and Legionella Flagellin Induce Macrophage Death

(A) Salmonella flagellin complements a ΔflaA mutant of Legionella. The LP02 ΔflaA mutant was complemented with plasmids expressing flagellin of L. pneumophila (Lp flaA), S. typhimurium (St fliC), S. flexneri (Sf fliC), or E. coli (Ec fliC). A 48-amino acid C-terminal truncation mutant of FlaA (Lp flaAΔ48) was also expressed as a control. Infection and LDH release of B6 macrophages were assayed as in Figure 3. An MOI of 2 was used in this experiment. Asterisks denote p < 0.02 (two-tailed student's t-test) versus ΔflaA + Lp flaAΔ48. (B) Flagellin from Legionella and Salmonella induce macrophage death when delivered to the cytosol by E. coli. A flagellin-deficient strain of E. coli (CM735ΔfliC) was transformed with the same plasmids as in (A), or an empty vector as a control. In some cases, the E. coli also expressed a non-secreted form of the pore-forming toxin LLO. In this system, E. coli is degraded in the phagolysosome, releasing cytoplasmic LLO. LLO forms pores in the phagosome, giving phagosomal contents access to the cytosol. Infection and LDH release of B6 macrophages was assayed as in Figure 3. An MOI of 2 was used in this experiment. Asterisks denote p < 0.02 (two-tailed student's t-test) versus LLO + empty vector (gray bar). The p-value for LLO + Sf fliC was 0.05; the p-value for LLO + Ec fliC was 0.05; and the p-value for LLO + Lp FlaAΔ48 was 0.155. (C) The E. coli CM735ΔfliC-derived strains from (B) were tested for the ability to swim through soft agar (motility) plates. Nonmotile strains remain at the point of inoculation, whereas motile strains show a halo of bacteria swimming outward from the point of inoculation.