Legionella Effectors That Promote Nonlytic Release from Protozoa

See allHide authors and affiliations

Science  27 Feb 2004:
Vol. 303, Issue 5662, pp. 1358-1361
DOI: 10.1126/science.1094226


Legionella pneumophila, the bacterial agent of legionnaires' disease, replicates intracellularly within a specialized vacuole of mammalian and protozoan host cells. Little is known about the specialized vacuole except that the Icm/Dot type IVsecretion system is essential for its formation and maintenance. The Legionella genome database contains two open reading frames encoding polypeptides (LepA and LepB) with predicted coiled-coil regions and weak homology to SNAREs; these are delivered to host cells by an Icm/Dot–dependent mechanism. Analysis of mutant strains suggests that the Lep proteins may enable the Legionella to commandeer a protozoan exocytic pathway for dissemination of the pathogen.

Several intracellular pathogens including Legionella pneumophila, Mycobacterium avium (1), Chlamydia spp. (2), and Francisella tularensis (3) are able to replicate within protozoan trophozoites. Thus, free-living amoebae may serve as a significant reservoir for pathogens in the environment, perhaps even as a “training environment” for the selection of virulence-related traits in these pathogens (4). L. pneumophila, the causative agent of legionnaires' disease, is frequently detected in association with Hartmanella vermiformis at the sources of infection during outbreaks (5). Under experimental conditions, L. pneumophila can multiply within and kill a variety of phylogenetically unrelated protozoa ranging from Acanthamoeba castellani to the genetically well characterized social amoeba Dictyostelium discoideum (6, 7).

Intracellular pathogens have evolved three distinct strategies for surviving phagosomelysosome fusion. Two of these mechanisms, tolerance of the toxic environment and escape from the phago-lysosome into the cytosol, are used by a variety of pathogens such as Salmonella enterica and Listeria monocytogenes, respectively. L. pneumophila and several other prokaryotic and eukaryotic intracellular pathogens use a third strategy— prevention of phagosome-lysosome fusion. After uptake of L. pneumophila by macrophages or protozoa, the bacteria are found within a specialized vacuole that does not fuse with lysosomes and does not acidify (8, 9); this allows replication to proceed. The specialized vacuole associates with endoplasmic reticulum (ER)–derived secretory vesicles (10), mitochondria, and rough ER (11, 12) and, near the end of the replicative cycle, acquires late endosomal markers (13). These observations strongly suggest that the bacteria play an active and continuous role in modulating organelle trafficking events from within the confines of their specialized vacuole.

A group of 24 genes called icm (intracellular multiplication) (14) or dot (defect in organelle trafficking) (15) are required for intracellular multiplication of L. pneumophila. Sequence similarity between several Icm/Dot proteins and those of the conjugative system of IncI plasmids colIb-9 and R64 (16) indicates that the Icm/Dot proteins form a type IV secretion system (TFSS) that delivers effectors to host cells. However, L. pneumophila mutants lacking the previously identified RalF or LidA effectors do not display obvious defects in organelle trafficking or intracellular replication (17, 18). Because direct biochemical observation of infected cells has not led to the identification of additional effectors in L. pneumophila, we searched the genome sequence database for candidate genes.

As the modulation of organelle trafficking events appears to be important during intracellular multiplication, we looked for L. pneumophila effectors that resemble components of the SNARE system (19) and that might somehow disable or alter its function. Two such open reading frames (ORFs) were identified. Both exhibit limited sequence similarity to mammalian EEA1 and yeast USO1, proteins known to be involved in vesicle trafficking. Both Legionella ORFs are predicted to encode large regions of α-helical coiled-coils, structures present in EEA1 and USO1 and also commonly found in SNAREs (table S1).

As there are no functional assays for these putative effectors, we investigated whether L. pneumophila delivers the ORF products to host cells. We used a gene fusion approach that takes advantage of the calmodulin-sensitive adenylate cyclase of Bordetella pertussis. Adenylate cyclase hybrid proteins have very low activity in the bacterial cells in which they are synthesized but are strongly activated if they are translocated to the cytosol of eukaryotes and interact with calmodulin (20). We constructed gene fusions that encoded chimeras of the Legionella effector proteins LepA and LepB with the catalytic domain encoded by cyaA. We then tested the ability of L. pneumophila to deliver the hybrid proteins to host cells by measuring the levels of cAMP in J774 macrophages infected by different L. pneumophila strains. When wild-type L. pneumophila expressing LepA-CyaA and LepB-CyaA under control of the Ptac promoter were used to infect J774 cells, large amounts of adenosine 3′,5′-monophosphate (cAMP) were produced within 30 min after infection (table S2). Controls showed that cAMP production depended on the presence of the LepA or LepB sequences. Translocation also required the DotA and IcmT proteins, which indicates that the proteins are delivered via the Icm/Dot TFSS.

To evaluate whether LepA or LepB plays any role during intracellular multiplication of L. pneumophila, we constructed strains lacking either or both genes. We used a standard assay for intracellular replication to examine the abilities of the strains to replicate in both human and protozoan hosts. This assay detects colonyforming units (CFUs) released into the supernatant at different times following infection. None of the three strains showed a detectable replication defect in several mammalian phagocytic cells, including human macrophages, murine J774 macrophages, and human macrophage-like U937cells (fig. S1). In contrast, all three mutant strains were recovered in smaller numbers than wild-type L. pneumophila during cocultivation with the protozoan host A. castellani. Much smaller numbers of CFUs were recovered for strain JC51 in which both genes were deleted (Fig. 1A). To verify that the defect is due to the absence of the LepA and LepB proteins, we used plasmids encoding either LepA or LepB. As expected, either plasmid could reverse the defect in JC51 (fig. S2). Similar results were obtained with another protozoan, the social amoeba D. discoideum (21). We conclude that although LepA and LepB are not required for replication in mammalian cells, they each play a partially redundant role in protozoan host cells.

Fig. 1.

L. pneumophila lepA and lepB mutants are defective for release from A. castellanii, but can multiply intracellularly. A. castellanii trophozooites were seeded into 24-well tissue culture plates, infected with L. pneumophila strains, and the numbers of CFUs released into the supernatants were measured. (A) The lepA or lepB single mutants are partially defective for release of Legionella when cocultured with A. castellanii. The lepA, lepB double mutant is severely defective for release of Legionella. (B) CFUs from a single step of growth for wild-type and the lepA and lepB mutants were measured in cell lysates. The single or double lep mutants exhibited levels of intracellular replication similar to wild type's in the first round of infection. Values represent the mean of at least three independent samples ± SD. (C and D) TEM images of D. discoideum infected with L. pneumophila strain JC51, showing examples of material remaining attached to the plastic surface of a tissue culture well 24 hours after infection. (C), An intact cell containing several L. pneumophila inside vacuoles; (D), remnants of a lysed cell including several membrane-bound L. pneumophila. See Materials and Methods for details. Scale bars, 10 μm.

To test whether the bacteria were capable of intracellular growth but were not released from the host cells, we performed single-step growth experiments in which the total bacterial contents of the host cells and the extra-cellular medium were counted after artificial lysis of the infected amoebae (Fig. 1B). These experiments revealed that the lepA and lepB mutations, either singly or together, had little effect on intracellular multiplication; even the JC51 double mutant replicated almost as well as the JR32 wild-type strain. Therefore, the lower number of CFUs recovered for the mutants in the standard assay depicted in Fig. 1A must be due to a defect in release from the protozoan cells, magnified over the course of several infection cycles.

To confirm that the bacteria are able to replicate inside protozoa and to observe their intracellular location, D. discoideum cells infected with either the wild-type JR32 or the JC51 mutant strain were examined by transmission electron microscopy (TEM) at 0.5, 5, and 24 hours after infection. After 0.5 hour and 5 hours, the samples contained similar numbers of bacteria in membrane-bound compartments (fig. S3). By 24 hours, there were few intact cells remaining for either sample. A rare intact D. discoideum cell infected with JC51 24 hours after infection is shown in Fig. 1C; there are several membrane-bound compartments containing bacteria. In the samples infected with JC51, numerous vesicles containing bacteria were observed with the attached remains of lysed cells (Fig. 1D). Such structures were not observed in samples infected with wild-type L. pneumophila, presumably because the vesicles had already been released into the medium. These results are entirely consistent with the view that the LepA and LepB proteins allow nonlytic release of L. pneumophila. Others have documented that L. pneumophila are released from protozoa enclosed within “faecal vesicles” (6) or “respirable vesicles” (22), which are believed to be highly infectious to humans. JC51 exhibits a defect in release from protozoan but not mammalian host cells; this suggests that the bacteria are normally released via an exocytic pathway in protozoa that is either absent or unused in mammalian cells.

Because release of Legionella from host cells has been suggested to involve poreforming cytotoxicity (23), we evaluated the alternative possibility that the LepA and LepB products facilitate bacterial release by lysing host cells. We used a vital dye to measure the viability of amoebae after infection by wild-type and mutant strains of L. pneumophila. Infection of A. castellani with a dotA mutant had no effect on the ability of the amoebae to reduce the dye, because this strain is incapable of intracellular growth or host cell killing. In contrast, the wild-type and the lep mutant strains completely prevented dye reduction above an input multiplicity of infection of 1 bacterium per amoeba (fig. S4). Similar results were obtained with D. discoideum (21). We conclude that the Lep proteins play no role in cytotoxicity or host cell killing.

The ability of the JC51 strain to grow within and to kill protozoa suggests that it retains the ability to prevent phagosomelysosome fusion. To verify this, we used D. discoideum cells that express a fusion of green fluorescent protein (GFP) to VatM, the 100-kD transmembrane subunit of the vacuolar H+-translocating adenosine triphosphatase (V-ATPase). Earlier studies using VatM-GFP revealed that the V-ATPase is delivered to the membrane of new phagosomes by fusion with preexisting endolysosomes (24). Amoebae expressing VatM-GFP were infected with different L. pneumophila strains expressing DsRed-Express. Living infected cells were observed by confocal fluorescence microscopy. Representative images of the VatM-GFP expressing D. discoideum cells infected with wild-type (JR32), icmT (GS3011), or ΔlepA, lepB (JC51) are shown in Fig. 2. The red signal of the icmT mutant bacteria is clearly surrounded by the green VatM-GFP signal, indicating fusion of the phagosomes with endo-lysosomes (Fig. 2, C, D, and E). In contrast, neither the red signal of JR32 nor that of JC51 associates with the green signal of the VatM-GFP (Fig. 2, A and B). Thus, the V-ATPase is delivered to the membranes of phagosomes containing the icmT mutant but not to phagosomes containing either the wild-type or the lepA, lepB JC51 double mutant. We conclude that neither LepA nor LepB is required for preventing fusion of the Legionella phagosome with lysosomes.

Fig. 2.

Trafficking of vacuolar proton pumps to phagosomes containing three strains of L. pneumophila. (A), JR32 (wild-type); (B), JC51 (lepA::Gm, lepB::Km); (C to E), GS3011 (icmT::Km). (A and B), cells 3 to 4 hours after infection; (C), cells after 50 min of infection; and (D), the same preparation 24 hours later, still in the presence of L. pneumophila. (E) The same cell as in (D), but minus the red channel, to reveal more clearly VatM-GFP in the phagosome membrane. Scale bars, 10 μm. (A and B), same magnification as (C); (D) same as (E.)

Although we have no information about the mechanism used by the Lep proteins to promote release of Legionella from protozoa, the availability of D. discoideum mutants with defects in the late stages of endosomal transit including exocytosis (25) may provide a way to test whether a known exocytic pathway is involved. In addition, ectopic expression of the Lep proteins in amoebae may provide additional information about their function at the molecular level.

The ability of L. pneumophila and other bacteria to prey on protozoa is becoming more widely recognized. Several bacterial species, some of which may be phylogenetically related to L. pneumophila (26), have been found to exist as symbionts within protozoa (27). The genetic and physiologic factors that contribute to the development of host-pathogen versus host-symbiont relations can best be studied using both hosts and bacteria that are genetically tractable, like those described here.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 and S2


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article