A Bacterial Guanine Nucleotide Exchange Factor Activates ARF on Legionella Phagosomes

See allHide authors and affiliations

Science  25 Jan 2002:
Vol. 295, Issue 5555, pp. 679-682
DOI: 10.1126/science.1067025


The intracellular pathogen Legionella pneumophilasubverts vesicle traffic in eukaryotic host cells to create a vacuole that supports replication. The dot/icm genes encode a protein secretion apparatus that L. pneumophila require for biogenesis of this vacuole. Here we show that L. pneumophilaproduce a protein called RalF that functions as an exchange factor for the ADP ribosylation factor (ARF) family of guanosine triphosphatases (GTPases). The RalF protein is required for the localization of ARF on phagosomes containing L. pneumophila. Translocation of RalF protein through the phagosomal membrane is adot/icm-dependent process. Thus, RalF is a substrate of the Dot/Icm secretion apparatus.

Legionella pneumophila are aquatic bacteria that infect and grow within protozoan hosts in most freshwater ecosystems (1). When these bacteria are inhaled by humans, L. pneumophila will replicate in alveolar macrophages, resulting in a severe pneumonia known as Legionnaires' disease (2, 3).Legionella pneumophila replicate within phagocytes by first creating a specialized vacuole that is similar morphologically to the endoplasmic reticulum (ER) of its host (4, 5). Biogenesis of this replicative vacuole requires the Dot/Icm transporter (6), which is a type IV protein secretion apparatus (7, 8). Pathogens such as Agrobacterium tumefaciens and Helicobacter pylori use type IV transporters to inject bacterial proteins directly into the cytosol of eukaryotic host cells (9–11). It is thought that the Dot/Icm transporter is used by L. pneumophila to inject proteins into host cells in order to control the biogenesis of a replicative organelle by modulating the activity of host factors involved in vesicle traffic. However, genetic screens that have been successful in isolating virulence determinants required for growth ofL. pneumophila in host cells, including the genes encoding the Dot/Icm secretion apparatus, have not revealed any injected proteins (7, 8, 12, 13).

The host protein ADP ribosylation factor–1 (ARF1) is found on phagosomes containing wild-type L. pneumophila but is not localized to phagosomes containing L. pneumophila dot/icm mutants (14). ARF1 is a highly conserved small GTP-binding protein that acts as a key regulator of vesicle traffic from the ER and Golgi [reviewed in (15)]. Because ARF1 localization on phagosomes containing L. pneumophila requires the Dot/Icm transporter, an injected bacterial protein may be required for ARF1 recruitment. To find proteins that are injected into host cells by the Dot/Icm transporter, we focused on bacterial gene products that may play a direct role in localization of ARF1 to phagosomes containing L. pneumophila. Because the association of cytosolic ARF onto vesicle membranes is coincident with GTP activation, we searched the genome ofL. pneumophila for proteins that had homology to ARF-specific guanine nucleotide exchange factors (GEFs) (16).

We identified a L. pneumophila gene that encodes a 374 amino acid protein with a Sec7-homology domain (Fig. 1A) (17). Sec7-homology domains are found in a diverse family of eukaryotic ARF-GEFs and are sufficient to stimulate the exchange of GDP for GTP [reviewed in (18)]. This L. pneumophila gene was called ralF. A gene from the intracellular pathogenRickettsia prowazekii (19) that is predicted to encode a protein that has 42% identity to the full-length RalF protein was identified using the RalF protein sequence as a BLASTP query (Fig. 1A). Currently, the RalF protein and this R. prowazekiiprotein are the only two prokaryotic gene products known to contain a Sec7-homology domain.

Figure 1

Legionella pneumophila produce a protein that contains a Sec7-homology domain. (A) The Sec7-homology domains from several eukaryotic and prokaryotic proteins are aligned to reveal regions of amino acid identity and similarity. The L. pneumophila RalF protein contains an NH2-terminal domain of 200 amino acids that is 41% identical to the Sec7-homology domains found in members of the eukaryotic family of ARF-GEFs. A ralF xenolog inRickettsia prowazekii (Rickettsia) also encodes a protein with an NH2-terminal Sec7-homology domain. GenBank accession numbers for proteins containing Sec7-homology domains are as follows: ARNO (Q99418), Cytohesin 1 (Q15438), GRP1 (CAA06434), Sec7 (P11075), p200 (AAD43651), GBF1 (AAD15903), Gea1 (P47102), Legionella (AY056455), and Rickettsia (G71694). (B) A 42-kDralF product was identified in L. pneumophila by immunoblot analysis with the use of an antibody specific for the RalF protein. The RalF protein was produced by a laboratory strain of L. pneumophila (Lp01), an isogenic mutant defective in the Dot/Icm transporter (Lp01 ΔdotA), and a clinical isolate of L. pneumophila (Lp philadelphia-1). The RalF protein was not detected in a mutant strain that had theralF gene deleted (Lp01 ΔralF). (C)Legionella pneumophila growing logarithmically and in stationary phase were isolated. Whole cell lysates were prepared, and cellular levels of the RalF protein were determined by immunoblot analysis (36). (D) Total RNA was extracted fromL. pneumophila cultures in logarithmic and stationary phase. The level of ralF mRNA was measured by quantitative slot blot hybridization (36).

The RalF protein was detected (20) in wild-type L. pneumophila (Fig. 1B, Lp01) and in an isogenic mutant that has a defective Dot/Icm transporter (Fig. 1B, Lp01 ΔdotA) but was not detected after the ralF gene was deleted from theL. pneumophila chromosome (Fig. 1B, Lp01 ΔralF). The RalF protein was also produced by L. pneumophila Philadelphila-1 (Fig. 1B, Lp philadelphia-1), which is a clinical isolate obtained from the first documented L. pneumophila disease outbreak (3). These data demonstrate that a protein containing a Sec7-homology domain is produced by L. pneumophila.

It has been hypothesized that proteins injected into host cells by the Dot/Icm apparatus are up-regulated as exponentially growing bacteria enter stationary phase (21). Immunoblot analysis showed that the cellular concentration of the RalF protein is greater in stationary phase bacteria than in bacteria growing exponentially (Fig. 1C). To examine whether the ralF gene is growth-phase regulated, mRNA levels from bacteria in exponential and stationary phase were measured by slot-blot hybridization (Fig. 1D). There was a threefold increase in ralF expression as exponentially growing bacteria enter stationary phase. Thus, the RalF protein has an expression profile predicted for proteins injected into host cells by the Dot/Icm transporter.

To investigate the in vivo role of the RalF protein, the recruitment of ARF1 to phagosomes containing L. pneumophila was examined (22). ARF1-GFP co-localization was observed on phagosomes containing wild-type L. pneumophila but was not detected on phagosomes containing ΔdotA mutants of L. pneumophila (Fig. 2, A and B). ARF1-GFP was not observed on phagosomes containing ΔralFmutants, indicating that this protein facilitates localization of ARF1 to phagosomes containing L. pneumophila. Thus, RalF protein and the Dot/Icm transporter are both required for the recruitment of ARF to the Legionella phagosome accounting for the acronym RALF.

Figure 2

The RalF protein is a guanine nucleotide exchange factor that recruits ARF1 to phagosomes containing L. pneumophila. (A) Mouse bone marrow–derived macrophages expressing ARF1-GFP were infected with wild-type L. pneumophila (Lp01) and an isogenic mutant that has the gene encoding RalF deleted (ΔralF). Macrophages were fixed 1 hour after infection. Bacteria were stained with antibody to L. pneumophila and secondary antibody labeled with Texas Red. Stacked confocal images show that ARF1-GFP is found on phagosomes containing wild-type L. pneumophila but is not located on phagosomes containing the ΔralF mutant. Bar, 5 μm. (B) Localization of ARF1-GFP was measured for phagosomes containing wild-type L. pneumophila (Lp01), mutants defective in Dot/Icm transporter function (ΔdotA), and ralFmutants (ΔralF). The proportion of phagosomes that stain positive for ARF1-GFP at 60 and 120 min after infection is shown for each strain of L. pneumophila. Results are the average of two independent experiments in which at least 50 phagosomes were scored. The numbers obtained in each experiment varied by less than 20%. (C) ARF guanine nucleotide exchange activity of the GST-RalF protein was determined with the use of myristoylated ARF3. Exchange activity is presented as the stoichiometry of [35S]GTPγS bound to ARF3 (1 μM) after a 10-min incubation with varying concentrations of GST-RalF (RalF + ARF3). The minimal Sec7 homology domain from Saccharomyces cerevisiae Sec7p was used as a positive control (ySec7 domain + ARF3), and reactions containing GST-RalF without ARF3 served as a negative control (RalF).

It was predicted that the RalF protein could function as an ARF-GEF by virtue of its Sec7-homology domain. The six mammalian ARFs fall into three classes that have overlapping functions in regulating vesicle transport (15). Guanine nucleotide exchange assays (23) revealed that RalF was active on ARF1 and ARF3, which are class I proteins (Fig. 2C) (24). RalF was also active on the class III protein ARF6, but was less active on the class II protein ARF5 (24). These data are consistent with previous studies of eukaryotic proteins containing Sec7-homology domains, which often have guanine nucleotide exchange activity for more than one ARF protein family member (18). Thus, the RalF protein from L. pneumophila functions as an exchange factor that activates members of the ARF protein family.

To interact with ARF and to activate it directly, the RalF protein must be translocated out of the bacterium and across the phagosome membrane. Because bacteria use type IV transporters to inject proteins into foreign cells and because dot/icm mutants ofL. pneumophila are unable to recruit ARF to their phagosomes, we predicted that the Dot/Icm apparatus was necessary for RalF export during host cell infection. This hypothesis was tested by immunofluorescent staining of L. pneumophila phagosomes with an affinity-purified antibody specific for the RalF protein (25). Co-localization of RalF and ARF1-GFP was apparent on phagosomes containing wild-type L. pneumophila (Fig. 3A). When RalF staining on L. pneumophila phagosomes was measured 30 min after internalization, we found that 38 ± 5.5% of phagosomes containing wild-typeL. pneumophila stained positive for RalF (24). Specific staining with the RalF antibody was not observed on phagosomes containing ΔdotA mutants (2.0 ± 1.4%) or on phagosomes containing ΔralFmutants (3.7 ± 3.6%). Thus, localization of RalF protein on phagosomes containing L. pneumophila requires a functional Dot/Icm transporter, which indicates that RalF is translocated into eukaryotic host cells by the Dot/Icm transporter. There are no obvious reasons why RalF protein staining remains localized to phagosomes containing L. pneumophila. Transmembrane or membrane interaction domains are not apparent in the RalF protein sequence. This may mean that RalF interacts with another protein or a lipid on the phagosome membrane or perhaps RalF is rapidly degraded in the host cytoplasm.

Figure 3

The RalF protein is a substrate translocated into host cells by the Dot/Icm transporter. (A) Mouse bone marrow–derived macrophages expressing ARF1-GFP were infected with wild-type L. pneumophila and fixed 1 hour after uptake. Localization of the RalF protein was visualized by staining with an affinity-purified antibody specific for RalF and a secondary antibody conjugated to Cy5. DNA was stained with propidium iodide, which labels both the macrophage nucleus and bacterial cells. A projection of stacked confocal images shows that RalF (blue) and ARF1-GFP (green) co-localize on phagosomes containing L. pneumophila (red). Bar, 5 μm. (B) Intracellular growth of L. pneumophila strain Lp01 (closed circles) and the isogenicralF mutant (open circles) was measured in eukaryotic host cells. The intracellular growth kinetics of these L. pneumophila strains were determined in mouse bone marrow–derived macrophages, the human macrophage-like cell line U937, and the protozoan host A. castellanii. Graphs show colony-forming units ± standard deviation of each strain recovered from infected host cells over 48 hours. (C) Southern analysis indicates that the ralF gene is present in most serogroups of L. pneumophila but is not detected in other Legionellaspecies examined, which included isolates of L. micdadei(Lm), L. bozemanii (Lb), L. gratiana (Lg), and L. longbeachae (Ll). The strains ofLegionella examined were obtained either as clinical specimens from infected patients (C) or were isolated from environmental sources (E). Blue type is used to highlight theralF deletion mutant constructed in this study (Lp01 ΔralF) and serogroup 2 strain of L. pneumophila (ATCC 33154) lacking the ralF gene is highlighted with red type. The film in the far right panel was exposed longer (12 hours) than films in the left and middle panels (4 hours).

We were interested in whether L. pneumophila require RalF for replication inside eukaryotic cells. Growth of wild-type L. pneumophila and an isogenic ΔralF mutant was measured in murine bone marrow–derived macrophages (26), in the human macrophage-like cell line U937 (27), and in the protozoan host Acanthamoeba castellanii(28). There was no intracellular growth defect observed for the ΔralF mutant in these three eukaryotic hosts (Fig. 3B). Thus, the RalF protein is not required for growth of L. pneumophila within these hosts, which explains why ralFhad not been isolated previously in genetic screens for L. pneumophila intracellular growth mutants.

The high degree of sequence identity between the Sec7-homology domains in RalF and eukaryotic ARF-GEFs in addition to the observation that proteins with Sec7-homology domains have been found in only two prokaryotic organisms suggests that horizontal gene transfer of a eukaryotic gene gave rise to the ralF gene. IfralF were acquired recently by horizontal gene transfer, it might not be present in all L. pneumophila serogroups or in other Legionella species. To address this question, genomic DNA from several L. pneumophila serogroups was analyzed by Southern hybridization to determine the genetic distribution ofralF (29). The ralF gene was present in all L. pneumophila serogroups with the exception of the American Type Culture Collection (ATCC) L. pneumophilaserogroup 2 strain #33154 (Fig. 3C, red type). The ralF gene was not detected in any other species of Legionellaexamined. However, the dotA gene was detected in these other species (24) and we have shown previously that the IcmX protein is produced by these bacteria (30), which means these species of Legionella have essential components of the Dot/Icm transporter but do not have the ralF gene. Although a gene similar to ralF is found in R. prowazekii, comparative genomic analysis indicates that there are no gene products containing a Sec7-homology domain in the related species R. conorii (31). Thus, the ralF gene was most likely acquired by L. pneumophila and R. prowazekii after speciation occurred within each genus.

Legionella pneumophila are found ubiquitously in freshwater environments where they parasitize protozoan hosts (1). It is reasonable to assume that natural conditions exist where L. pneumophila containing the ralFgene have a selective advantage. For instance, ralF may allow L. pneumophila to infect protozoan hosts that restrict the growth of Legionella lacking ralF, or perhapsralF enables L. pneumophila to infect permissive protozoan host cells more efficiently during periods of environmental stress. There are over 35 different species of Legionella, yet most large outbreaks of community-acquired Legionnaires' disease worldwide are caused by L. pneumophila (32). This raises the question of whether acquisition of ralF makesL. pneumophila a more virulent human pathogen compared with the other Legionella species that are missingralF. It is likely that genes acquired recently encoding proteins that are secreted by the Dot/Icm transporter may not only enhance replication of Legionella in new environments but could coincidentally increase virulence of those Legionellafor humans. For example, a newly acquired substrate of the Dot/Icm transporter may allow L. pneumophila to evade host immune responses more effectively or permit replication of L. pneumophila in human cells that are not permissive for bacteria which lack this protein. Identifying and characterizing additional substrates of the Dot/Icm transporter, in combination with a comparative analysis of Legionella genomes, will provide valuable information on how an environmental organism has become a human pathogen.

  • * To whom correspondence should be addressed. E-mail: craig.roy{at}


View Abstract

Navigate This Article