Conjugative Transfer by the Virulence System of Legionella pneumophila

See allHide authors and affiliations

Science  06 Feb 1998:
Vol. 279, Issue 5352, pp. 873-876
DOI: 10.1126/science.279.5352.873


Legionella pneumophila, the causative agent of Legionnaires' pneumonia, replicates within alveolar macrophages by preventing phagosome-lysosome fusion. Here, a large number of mutants called dot (defective for organelle trafficking) that were unable to replicate intracellularly because of an inability of the bacteria to alter the endocytic pathway of macrophages were isolated. The dot virulence genes encoded a large putative membrane complex that functioned as a secretion system that was able to transfer plasmid DNA from one cell to another.

A number of intracellular bacterial pathogens, such as Chlamydia trachomatis,Mycobacterium tuberculosis, and Legionella pneumophila, grow within membrane-bound compartments diverted from the normal endocytic pathway of host cells (1).Legionella pneumophila replicates within alveolar macrophages by preventing acidification of the nascent phagosome and subsequent fusion with lysosomes (2). Several L. pneumophila genes (dotA and icmWXYZ) that are required for growth in macrophages have been identified (3). Mutations in these genes allow bacteria to be internalized into compartments that fuse with lysosomal components (3, 4).

To understand how this organism prevents phagosome-lysosome fusion, we isolated a large collection of additional mutants that were defective for intracellular growth. Twenty-six spontaneous mutants were isolated on the basis of the observation that L. pneumophila strains resistant to low amounts of sodium chloride are often unable to replicate in macrophages (5, 6). We independently isolated six additional mutants by screening mutagenized L. pneumophila for a lack of intracellular growth (7). Complementation of these mutants revealed two 20-kb regions on theL. pneumophila chromosome that contain a large number of genes required for growth in macrophages (Fig.1). Region I contains three genes,dotDCB, located about 10 kb from the previously identifieddotA-icmWXYZ locus. Region II contains 11 genes in three potential operons (dotML, dotKJIHGFE, and dotNO). The majority of the dot andicm genes identified to date, 14 of 19, are predicted to encode proteins that are membrane-associated. Although most of these proteins are not homologous to any known open reading frames (ORFs), four Dot proteins have limited homology to components of bacterial conjugation systems (Fig. 1). The COOH-terminus of DotG is homologous to Trb I, a protein required for conjugation of the IncP plasmid RP4 (23% identity over the COOH-terminal 442 amino acids of DotG) (Fig. 1) (8). DotM and DotL have homology to TrbA and TrbC, respectively, from the Inc I plasmid R64 (23% identity for DotM and TrbA and 26% identity for DotL and TrbC) (9). Finally, DotB is homologous to a large family of nucleotide-binding proteins (for example, TrbB) that includes members of various conjugal-transfer systems (6).

Figure 1

Legionella pneumophilagenes required for growth in macrophages. dot genes are located in two regions, I and II, each consisting of about 20 kb of DNA. dotDCB (GenBank accession number AF026533) is located 10 kb from a previously identified locus required for intracellular growth consisting of the genes dotA andicmWXYZ (3). Insertion mutagenesis of the intervening region between dotA anddotB revealed no additional genes required for growth in macrophages. Region II contains 11 dot genes in three potential operons (dotML, dotKJIHGFE, anddotNO) (Genbank accession number AF026534). Region II is flanked on one side by the L. pneumophila homolog ofcitA, a plasmid-encoded citrate-proton symporter found in certain cit+E. coli strains (21). dotB, dotE, anddotG were identified by the salt selection (6),dotH and dotO by the mutagenesis screen (7), and the remaining genes by sequencing the surrounding region. Proteins encoding the genes that are shaded black have homology to known conjugation proteins, and proteins encoding the genes that are shaded gray are predicted to be membrane associated.

To examine the role of one of the dot genes with homology to conjugation genes, we constructed a large in-frame deletion indotG (10). The dotG deletion mutant was assayed for survival and replication in the monocytic cell line U937, which supports intracellular growth of virulent L. pneumophila (11). Wild-type L. pneumophilacontaining the vector pKB5 showed an increase in viable counts of 103 to 104 cells in 72 hours, whereas thedotG mutant with the same vector showed no growth during this time frame (Fig. 2A). Introduction of a plasmid encoding the dotG ORF (12) restored growth of the dotG deletion mutant to wild-type amounts. ThedotG mutant was also defective in altering the endocytic pathway, as exhibited by colocalization of phagosomes containing thedotG mutants and a late endocytic marker, LAMP-1 (Fig. 2B). In contrast, wild-type L. pneumophila are normally able to prevent phagosome-lysosome fusion, and therefore phagosomes containing these bacteria are relatively devoid of endocytic and lysosomal marker proteins (2, 13). This lack of proteins can be seen by a lack of colocalization of wild-type L. pneumophila phagosomes with LAMP-1 (Fig. 2B) (13). ThedotG mutant showed a targeting defect similar in magnitude to that of a previously characterized dotA mutant (80% LAMP-1 positive) in comparison with wild-type cells (20% LAMP-1 positive) (3). Thus, loss of a dot gene homologous to a conjugation gene resulted in a failure to replicate intracellularly because of an inability to alter targeting similar to that seen with the original dotA-icmWXYZ mutants (3, 4).

Figure 2

(A) DotG is required for intracellular growth. Phorbol ester–transformed U937 cells were challenged with a multiplicity of between 1 and 10 bacteria per macrophage, and growth was assayed as previously described (3). Samples from individual microtiter wells were taken daily in triplicate, and viable counts were titered on bacteriological plates. Growth was assayed for Lp02/pKB5 (○), a wild-type strain containing the vector pKB5 (3); Lp02/pJB480 (▵), a wild-type strain containing a plasmid expressing DotG+(15); JV573/pKB5 (◊), a ΔdotG mutant strain with vector pKB5 (13); and JV573/pJB480 (▪), a ΔdotG mutant strain with DotG+ ORF. (B) Phagosomes containing the dotG mutant colocalize with the late endocytic marker LAMP-1. Mouse bone marrow macrophages were challenged at a multiplicity of infection between 1 and 10 bacteria per macrophage to determine the intracellular localization of strains lacking functional dotG. After a 1-hour incubation at 37°, samples were fixed, permeabilized, and stained with antibodies to L. pneumophila to localize intracellular bacteria and antibody to LAMP-1 to stain the endocytic compartment (16). The top panels (a to c) are the wild-type L. pneumophila strain Lp02, and the bottom panels (d to f) are the dotG mutant strain JV573. LAMP-1 staining is shown in (a) and (d), L. pneumophilastaining in (b) and (e), and a composite of the LAMP-1 and bacterial staining in (c) and (f), consisting of LAMP-1 in green,L. pneumophila in red, and areas of colocalization in orange. The colocalization was performed with Color Merge of IP-Lab Spectrum (Signal Analytics, Vienna, Virginia). The efficiency of intracellular trafficking was assayed by quantification of the number of phagosomes containing bacteria that colocalized with LAMP-1 (n > 200 phagosomes assayed in three separate experiments).

To test if the homology of DotG to a protein required for conjugation was relevant, we attempted to determine if the Dot proteins could have evolved from a DNA transfer system. We assayed transfer of the mobilizable IncQ plasmid RSF1010, which codes for products involved in conjugative DNA processing but lacks the proteins involved in conjugal-pair formation and therefore requires those functions to be provided in trans (14). Legionella pneumophilawas able to mobilize RSF1010 to another strain of L. pneumophila at a frequency of about 10−6 conjugants per donor (Table 1, top).Legionella pneumophila was also able to transfer RSF1010 to two different strains of Escherichia coli (ER1793 and MM294) at about the same rate (15). Transfer required a cis-acting site on the plasmid, the origin of transfer (oriT), as normally seen with conjugation (ΔoriT in Table 1, top) (14). Moreover, the presence of deoxyribonuclease I (DNase I) had no effect, indicating that mobilization was not due to transformation by free DNA. Transfer required functional DotG protein because the ΔdotG strain characterized above was unable to transfer the RSF1010 plasmid pKB5, whereas providing thedotG ORF on pKB5 restored transfer to wild-type amounts (Table 1, bottom). Transfer also required the donor strain to have a functional copy of dotB, one of the other genes with homology to a conjugation gene. In addition, mobilization was dependent on genes with no homologies to known conjugation genes, including the previously characterized dotA and icmWXYZ genes as well as one of the genes we identified, dotE.

Table 1

The L. pneumophila dot virulence loci are required to transfer RSF1010 plasmid into recipient bacteria. Mating was assayed by mixing 1.0 × 109 L. pneumophila containing noted plasmids with a tenfold excess of a recipient bacteria strain. Matings were performed in triplicate by allowing the mixed bacterial cultures to incubate for 2 hours at 37°C on 45-mm Millipore hemagglutinin filters (HAWP 047 S0) placed onto prewarmed charcoal-yeast extract media with thymidine (CYET) plates (3).

View this table:

Because the Dot proteins are capable of mobilizing RSF1010, they likely constitute a secretion system that is capable of transferring a substrate across the outer membrane. During intracellular growth, this system could deposit a factor or inhibitor into macrophages to subvert the endocytic pathway. Legionella pneumophila may have acquired the dot secretion system by adaptation of the conjugation system of an integrated plasmid and may be another example of a pathogenicity island (16).

The L. pneumophila dot virulence system may be distantly related to specialized secretion systems, termed type IV, found in several other pathogens (17). Agrobacterium tumefaciens contains an operon of 12 genes, virB, which has extensive homology to a traditional plasmid transfer system and is used to inject oncogenic transferred DNA into plant cells (18). Agrobacterium tumefaciens is also able to transfer RSF1010 plasmid from one cell to another (19).Bordetella pertussis contains a related operon,ptl, which is used to secrete pertussis toxin (20). In contrast to these systems, L. pneumophila contains only two conserved proteins (DotB has homology to VirB11 and DotG has homology to VirB10). Moreover, these genes are not found in a single large operon of multiple conjugation genes as seen with the virB and ptl operons (17).

The actual substrate transferred by L. pneumophila into macrophages is presently unknown. However, it would seem unlikely that it injects “pathogenic” DNA into mammalian cells early in infection as A. tumefaciens does to plant cells because the endocytic pathway is altered extremely rapidly within minutes of uptake (4). In contrast, it is more likely that L. pneumophila transfers a protein that acts as an inhibitor or modifier of the endocytic pathway. The discovery that L. pneumophila dot genes are likely to form a secretion machinery provides the first functional indication of how L. pneumophila subverts the endocytic pathway of a macrophage. Understanding how this pathogen exploits a conjugal-transfer system for intracellular growth may shed light on how other clinically important pathogens, such as Chlamydia and Mycobacterium,cause disease.

Note added in proof: After completion of the refereeing of this manuscript, Segal and Shuman (22) reportedicmO and icmP, which are identical todotL and dotM, and indicated the presence of conjugal transfer.

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


View Abstract

Navigate This Article