Divergent Evolution of Legionella RCC1 Repeat Effectors Defines the Range of Ran GTPase Cycle Targets

Distribution and structural comparison of L. pneumophila RCC1 repeat effectors.A bioinformatics analysis of Legionella genomes revealed that RCC1 repeat genes are present in many Legionella species (24 of 58 analyzed) and are conserved in all 59 L. pneumophila strains examined in this study (Fig. 1). While some of the L. pneumophila strains contain only one RCC1 repeat effector gene (e.g., strains Lens and Paris: pieG/lpp1959), others contain two single genes (e.g., strains Philadelphia-1 and C7O: legG1/lpg1976 and ppgA/lpg2224) or, additionally, a duplicated ppgA gene. Indeed, the RCC1 repeat genes in L. pneumophila were found to be distributed in two main clusters, namely, the “Lens-Paris” cluster and the “Philadelphia-C7O” cluster, respectively. Most of the strains in the Lens-Paris cluster harbor only the pieG gene, the exceptions being a subcluster of seven strains containing pieG and two or three ppgA genes. The strains in the Philadelphia-C7O cluster are more closely related to each other and contain a split pieG gene (yielding lpg1975 and legG1; see below) as well as one ppgA gene. As an exception, strain E6N harbors single, intact pieG and ppgA genes.

• Open in new tab
• Download powerpoint

FIG 1

Distribution of L. pneumophila RCC1 repeat genes. The phylogenetic tree of 59 L. pneumophila strains was inferred using the fast core genome multialigner Parsnp (https://github.com/marbl/parsnp). Circles at nodes represent bootstrap support, and the size of each circle is proportional to the corresponding bootstrap value. The scale bar represents the estimated number of substitutions per site. For each strain, the presence or absence of the RCC1 repeat effector genes lpp1959 (pieG) and lpg2224 (ppgA) was assessed by OrthoMCL (https://orthomcl.org), and all of the strains were found to be distributed in two main clusters. In strain C4S, a split fragment of gene lpp1959 is a pseudogene. Strain NCTC11404 does not harbor an lpg2224 gene (verified by blast search), and one of three lpg2224 copies in strain D-7630 is a pseudogene.

The phylogenetic trees indicate that the RCC1 repeat genes have mostly followed the evolution of the strain (see Fig. S1 in the supplemental material). Strains E6N and NCTC11286 are exceptions, as they do not localize with the same strains in the phylogenetic trees of the strains and lpp1959 genes (Fig. S1A). This suggests that horizontal gene transfer took place and explains why strain E6N groups with the strains where lpp1959 is usually split. If present in two or more copies, lpg2224 genes from the same strain do not group together (Fig. S1B). This finding suggests that one or several duplication events originated in a common ancestor, followed by posterior sequence divergence (and perhaps functional diversification) among the different duplications and posterior emergence of new strains from the common ancestor. Taken together, the data show that RCC1 repeat genes are conserved in L. pneumophila and occur with a distinct pattern comprising two different strain clusters.

The RCC1 repeat genes of L. pneumophila strain Philadelphia-1 and Paris contain 2 to 3 RCC1 repeats and encode the effectors PpgA (66 kDa) and LegG1/Lpg1976 (31 kDa), or PieG (53 kDa), respectively (Fig. S2). Previously, the acronyms “LegG1” and “PieG” were used synonymously (51). However, given the distinct features of the proteins (see below), here we use LegG1 and PieG exclusively for the orthologs of strains Philadelphia-1 and Paris, respectively. The positions of the predicted RCC1 repeats in the L. pneumophila effectors vary, but the level of homology with the RCC1 repeat of the human Ran GEF RCC1 is very high, and the homologous region includes the conserved GQLGLGE/D motif but lacks amino acids essential for GEF activity (Fig. S2). Moreover, structural predictions indicate that the L. pneumophila RCC1 repeat effectors adopt β-propeller structures like human RCC1 (55).

Role of L. pneumophila RCC1 repeat effectors in pathogen-host interactions.To analyze the function of L. pneumophila RCC1 repeat effectors, strains lacking ppgA (LS03, ΔppgA), legG1 (ΔlegG1), ppgA and legG1 (LS01, ΔppgA-ΔlegG1), or pieG (LS08, ΔpieG) were constructed by deleting the genes from the chromosome by double homologous recombination (see Table S1A in the supplemental material). L. pneumophila ΔppgA, ΔlegG1, ΔppgA-ΔlegG1 (strain Philadelphia-1), and ΔpieG (strain Paris) grew at the same rate as the parental strains in broth (Fig. S3A and B). Moreover, the overproduction of PieG (Paris) in strain Philadelphia-1 or of PpgA (Philadelphia-1) in strain Paris did not impair growth in broth (Fig. S3C and D).

The ΔppgA, ΔlegG1, and ΔppgA-ΔlegG1 mutant strains were slightly but significantly impaired for intracellular replication in murine RAW 264.7 macrophages (Fig. 2A), but not in Acanthamoeba castellanii (Fig. S3E). Upon infection of D. discoideum, the ΔppgA-ΔlegG1 mutant strain was killed more efficiently at 24 h (Fig. S3F) but was impaired only for intracellular replication in a “single-round infection” assay (Fig. 2B). The ΔpieG mutant strain was slightly but significantly impaired for growth also in a multiple-round infection assay over several days (Fig. 2C). Upon coinfection of the ΔppgA, ΔlegG1, ΔppgA-ΔlegG1, or ΔpieG mutant strains with the parental strain at a 1:1 ratio in A. castellanii, the mutant strains were efficiently outcompeted by wild-type (WT) bacteria and eradicated within 6 to 9 days (Fig. S3G). Hence, PpgA—alone or in combination with LegG1—and PieG were essential for competition against wild-type L. pneumophila upon coinfection of amoebae.

• Open in new tab
• Download powerpoint

FIG 2

L. pneumophila RCC1 repeat effectors and their role in pathogen-host interactions. (A) Murine RAW 264.7 macrophages were infected (multiplicity of infection [MOI] 0.1) with L. pneumophila strain JR32, ΔicmT, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1, the cells were lysed, and intracellular replication at 37°C was assessed by CFU counting. (B) D. discoideum was infected (MOI 10) with L. pneumophila strain JR32, ΔicmT, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 producing GFP (pNT28), and intracellular replication at 25°C was assessed by GFP fluorescence increase. Data show means and standard deviations of results from triplicates (one-way analysis of variance [ANOVA]; ***, P < 0.001). RFU, relative fluorescence units. (C) D. discoideum was infected (MOI 0.1) with wild-type L. pneumophila strain Paris or with the ΔdotA or ΔpieG mutant strain, and intracellular replication at 25°C was assessed by CFU counting in lysates of infected cells (3 independent experiments; two-way ANOVA; *, P < 0.1; **, P < 0.01; ***, P < 0.001). (D and E) Real-time fluorescence microscopy of LCV motility in D. discoideum producing calnexin-GFP (pCaln-GFP) infected (MOI 5, 1 to 2 h) with L. pneumophila JR32, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 producing DsRed alone (pCR077) or together with M45-LegG1 (pER005) or M45-PpgA (pLS008) (D), or with the Paris wild-type strain or the ΔpieG mutant strain producing DsRed alone (pCR077) or together with M45-PieG (pLS033), M45-LegG1 (pER005), or M45-PpgA (pLS008) (E). LCV motility was recorded for 180 s with images taken every 10 s and quantified using ImageJ/Fiji software with the manual tracking plugin (n > 50/strain; 3 independent experiments; one-way ANOVA; **, P < 0.01; ***, P < 0.001; compared to the wild-type {D} or ΔpieG {E} strain). (F) D. discoideum producing GFP (pDM317) was infected (MOI 5, 1 h) with L. pneumophila JR32, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 producing DsRed alone (pCR077) or together with M45-LegG1 (pER005) or M45-PpgA (pLS008) and was seeded in a culture inset 2-well dish (Ibidi) for 2 h. After removal of the inset, cell migration was analyzed by confocal microscopy at 0 and 3 h. Bars, 200 μm. (G) Cell migration was quantified using ImageJ/Fiji software. The cell density ratio represents the average fluorescent signal intensity in the 500 μm gap divided by the average fluorescent signal intensity in 250 μm on each side of the gap center (3 independent experiments; two-way ANOVA; **, P < 0.01; ***, P < 0.001; all groups compared to wild-type strain at 3 h [t3]). (H) D. discoideum producing GFP-PpgA (pLS078), GFP-LegG1 (pER017), or GFP-LegG1_ΔCAAX (pER016) was infected (MOI 5, 1 h) with L. pneumophila JR32 or ΔicmT producing DsRed (pSW001), and localization of RCC1 repeat effectors was analyzed by confocal microscopy. Bars, 1 μm.

L. pneumophila strains lacking lpg1975 (LS06, Δlpg1975) or ppgA, legG1 and lpg1975 (LS05, ΔΔΔ) were also outcompeted by wild-type bacteria in the amoeba competition assay (Fig. S3G), but strain Δlpg1975 was not impaired for intracellular replication in RAW 264.7 macrophages or A. castellanii (Fig. S3H). The competition phenotype of strain Δlpg1975 was not simply due to the absence of the downstream legG1/lpg1976 gene, since in the mutant strain LegG1 was still produced, as established by mass spectrometry (Table S2A). Finally, lpg1975 was not required for the Icm/Dot-dependent translocation of LegG1 or PpgA into host cells (Fig. S3I) and is not an Icm/Dot substrate itself (51). In summary, the L. pneumophila RCC1 repeat effector PpgA (in particular, together with LegG1) and PieG contribute to pathogen-host cell interactions.

RCC1 repeat effectors promote LCV motility and cell migration.LegG1 has been implicated in microtubule stability and LCV motility as well as in host cell motility (47, 53). To assess whether the L. pneumophila RCC1 repeat effector PpgA also affects pathogen vacuole motility, we infected D. discoideum producing the endoplasmic reticulum (ER)/LCV marker calnexin-GFP (calnexin-green fluorescent protein) (12) with red fluorescent L. pneumophila wild-type, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 or with the complemented strains and analyzed pathogen vacuole dynamics by live-cell fluorescence microscopy (Fig. S4A). Compared to D. discoideum infected for 1 to 2 h with wild-type L. pneumophila, LCV motility was significantly reduced by ∼50% in amoebae infected with the ΔppgA mutant strain (or, as a control, the ΔlegG1 strain) (Fig. 2D). The LCV motility phenotype was complemented by providing the corresponding ppgA or legG1 gene on a plasmid under the control of the P_tac promoter. Interestingly, the phenotype of the ΔppgA strain was reversed by plasmid-borne legG1, while the phenotype of the ΔlegG1 strain was not reversed by ppgA. This finding suggests that while both RCC1 repeat effectors promote pathogen vacuole motility, their functions are not interchangeable. Similarly, upon infection of calnexin-GFP-producing D. discoideum with L. pneumophila strain Paris or the ΔpieG mutant, LCV motility was reduced by approximately 25% (Fig. 2E). The LCV motility phenotype was reversed by plasmid-encoded PieG or LegG1, but not by PpgA. Taken together, the L. pneumophila RCC1 repeat effectors PpgA, PieG and LegG1 promote LCV motility, and the functions of PieG and LegG1, at least regarding LCV motility, seem to be more closely related to one another than to PpgA.

To determine whether PpgA controls cell motility, we performed “gap closure” assays, where cell migration occurs to repopulate an area depleted of cells. To this end, D. discoideum producing GFP was infected with red fluorescent L. pneumophila strain JR32, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 or with the complemented strains. The infected amoebae were seeded in a culture inset 2-well dish, and after removal of the inset, cell migration was analyzed by confocal microscopy at 0 and 3 h (Fig. 2F). D. discoideum infected with the ΔicmT mutant as well as uninfected amoebae spread efficiently and increased the cell density in the insertion area by 70% to 90% in the course of the experiment (Fig. 2G). In contrast, D. discoideum infected with L. pneumophila strain JR32, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1 barely migrated, corresponding to an increase of cell density in the insertion area of merely 25% to 35% at 3 h after the start of the assay. Upon expression of plasmid-borne legG1 or ppgA, the inhibition of cell migration was reduced, in agreement with the notion that overproduction of the RCC1 repeat effectors has a positive effect on cell migration. Accordingly, the infection of amoebae with the ΔppgA-ΔlegG1 mutant led to a slight (statistically not significant) “hyperinhibition” of cell migration compared to the infection with parental strain JR32. In summary, these results indicate that the RCC1 repeat effectors PpgA and LegG1 (53) promote LCV dynamics and cell motility. The effects observed for the ΔppgA and ΔlegG1 mutant strains seem similar, but interestingly, in some cases they appear additive.

PpgA is dispensable for Ran activation on LCVs and localizes to the plasma membrane.To test whether RCC1 repeat effectors play a role for early steps of pathogen vacuole formation, D. discoideum amoebae producing the PtdIns(4)P probe P4C_SidC-GFP or calnexin-GFP were infected with red fluorescent L. pneumophila strain JR32, ΔicmT, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1, and GFP-positive LCVs were quantified by imaging flow cytometry (IFC) (Fig. S4B). The IFC colocalization scores were identical for the parental and the RCC1 repeat effector mutant strains, indicating that PpgA and LegG1 do not affect the early steps of LCV formation.

Next, we assessed a possible role of the RCC1 repeat effectors in the activation of Ran on LCVs. To this end, we employed D. discoideum producing GFP-RanBP1, which binds only to active Ran(GTP) (47). The amoebae were infected with red fluorescent L. pneumophila strain JR32, ΔppgA, ΔlegG1, or ΔppgA-ΔlegG1, or with the complemented mutants. The ER/LCV marker calnexin was immunostained in LCV isolates of the infected cells, and the localization of GFP-RanBP1 was analyzed by confocal fluorescence microscopy (Fig. S5A). Due to a low signal-to-noise ratio in intact infected cells, the recruitment of GFP-RanBP1 was observed only in LCV isolates of infected amoebae. While around 70% of the LCVs containing JR32 or the ΔppgA mutant stained positively for GFP-RanBP1, only approximately 40% of LCVs harboring the ΔlegG1 or ΔppgA-ΔlegG1 mutant strain were decorated with GFP-RanBP1 (Fig. S5B). For LCVs harboring the ΔlegG1 mutant strain, the localization of GFP-RanBP1 was restored to wild-type levels by providing legG1 on a plasmid under the control of the P_tac promoter. Taken together, these findings confirm that LegG1 promotes Ran activation on LCVs (47) and indicate that PpgA is dispensable for RanBP1 accumulation on LCVs. Hence, PpgA might be targeted to another cellular compartment.

To test the hypothesis that PpgA localizes to cellular compartments other than LCVs, we used D. discoideum ectopically producing GFP-PpgA, GFP-LegG1, or GFP-LegG1_ΔCAAX (lacking the C-terminal prenylation motif). The amoebae were left uninfected or were infected with red fluorescent L. pneumophila JR32 or the ΔicmT mutant strain, and localization of RCC1 repeat effectors was analyzed by microscopy. GFP-PpgA was produced by fewer cells than GFP-LegG1 (data not shown) but localized clearly and almost exclusively to the plasma membrane in uninfected D. discoideum, as well as in amoebae infected with virulent or avirulent L. pneumophila (Fig. 2H). PpgA-dependent activation of Ran on the plasma membrane through recruitment of GFP-RanBP1 could not be observed, likely because of the low signal-to-noise ratio, which also precluded visualization of the marker on LCVs in intact infected cells (Fig. S5A). In contrast to GFP-PpgA, GFP-LegG1 showed punctate staining in the cytoplasm and localized to pathogen vacuoles harboring JR32 but not ΔicmT mutant bacteria, as shown previously (47). Upon removal of the C-terminal prenylation motif, GFP-LegG1_ΔCAAX no longer localized to LCVs in D. discoideum (Fig. 2H), in agreement with the C-terminal signature being required for the localization to specific subcellular membranes in mammalian cells (51, 52). Finally, in homogenates of infected D. discoideum, V5-tagged LegG1 and PieG accumulated on LCVs, while V5-tagged PpgA did not localize to the pathogen vacuoles (Fig. S5C). Taking the results together, while LegG1 localizes to LCVs and promotes RanBP1 accumulation (Ran activation), PpgA localizes to the eukaryotic plasma membrane rather than to LCVs and is dispensable for RanBP1 accumulation on LCVs.

PpgA requires functional RanGAP1 to inhibit yeast growth.For further studies on targets and the mode of action of L. pneumophila RCC1 repeat effectors, we sought to employ the yeast Saccharomyces cerevisiae, a model which has previously been successfully used to assess effects of L. pneumophila effectors on eukaryotic processes (56, 57). In particular, the use of yeast lethality suppressors has been instrumental to characterize the mode of action of some effector proteins (58). Many L. pneumophila effectors target host proteins, which are conserved among eukaryotes. The yeast small GTPase Gsp1, e.g., is 83% and 72% identical to human Ran and D. discoideum RanA, respectively.

To validate the use of S. cerevisiae as a model to analyze RCC1 repeat effectors, we first analyzed the localization of ectopically produced GFP fusion proteins of the effectors by confocal fluorescence microscopy (Fig. 3A). GFP-PpgA predominantly localized to the plasma membrane in S. cerevisiae. In contrast, GFP-LegG1 and GFP-PieG were mostly distributed in the cytoplasm. Thus, the RCC1 repeat effectors localized to the same (PpgA) or similar (LegG1, PieG) subcellular compartments in amoebae and yeast, and accordingly, the latter might be a valid model system to functionally characterize the effectors.

• Open in new tab
• Download powerpoint

FIG 3

PpgA requires functional RanGAP1 to inhibit yeast growth. (A) The localization of L. pneumophila RCC1 repeat effectors was analyzed by confocal microscopy in S. cerevisiae wild-type strain BY4741 producing GFP-PpgA (pLS120), GFP-LegG1 (pLS118), or GFP-PieG (pLS113). Bars, 1 μm. (B) Growth of yeast producing RCC1 repeat proteins was tested by dot spot assays. S. cerevisiae wild-type strain BY4741 containing an empty plasmid (pYEP351gal) or producing FLAG-PpgA (pLS085), FLAG-LegG1 (pLS084), or FLAG-PieG (pLS086) was spotted in 10-fold dilutions on SG plates without leucine and grown at the indicated temperatures for 5 (30°C and 37°C), 6 (25°C), or 7 (20°C) days (left panel). Colony size (at 30°C) was measured using ImageJ/Fiji (right panel; n = 20; one-way ANOVA; ***, P < 0.001). (C) S. cerevisiae mutant strains rna1-1 (RanGAP1), prp20-1 (RCC1), yrb1-51 (RanBP1), and yrb2Δ (RanBP2) containing an empty plasmid (pYEP351gal) or a plasmid producing FLAG-PpgA (pLS085), FLAG-LegG1 (pLS084), or FLAG-PieG (pLS086) were spotted in 10-fold dilutions on SG plates without leucine and grown at the indicated temperatures for 5 (30°C), 6 (25°C and 37°C), or 7 (20°C) days (left panel). Colony size was measured using ImageJ/Fiji (right panel) (n > 10; one-way ANOVA; ***, P < 0.001). (D) S. cerevisiae strain BY4741 or mutant strain rna1-1 containing the empty plasmids pYEP351gal and pRS316, or pYEP351gal and pRS316-RNA1 (RNA1), pRS316 and pLS085 (FLAG-PpgA), or pRS316-RNA1 (RNA1) and pLS085 (FLAG-PpgA) was spotted in 10-fold dilutions on SG plates without leucine and uracil and grown at the indicated temperatures for 3 to 6 days. Colony size was measured using ImageJ/Fiji (one-way ANOVA; n > 9; ***, P < 0.001). (E) S. cerevisiae wild-type strain BY4741 or mutant strain rna1-1 containing the empty backbone (pYEP351gal) or producing FLAG-PpgA (pLS085), FLAG-LegG1 (pLS084), or FLAG-PieG (pLS086) was grown in SG medium without leucine for 7 days at 20°C, and OD₆₀₀ was measured every hour. Data show means and standard deviations (for simplicity, only the standard deviations at 50 h, 100 h, 150 h, and 167 h are shown) of results from three independent experiments (two-way ANOVA; **, P < 0.01; ***, P < 0.001).

Next, we tested the growth of yeast producing RCC1 repeat effectors by dot spot assays. S. cerevisiae wild-type strain BY4741 containing an empty plasmid or producing FLAG-PpgA, FLAG-LegG1, or FLAG-PieG was spotted in 10-fold dilutions on SG plates without leucine and grown at 20 to 37°C for 3 to 6 days (Fig. 3B; see also Fig. S6A). Under these conditions, production of FLAG-PpgA inhibited growth of the yeast cells, at lower temperatures in particular, while production of FLAG-LegG1 or FLAG-PieG did not affect yeast growth.

To obtain insights into the cellular pathways mediating growth inhibition by PpgA, we employed S. cerevisiae mutant strains defective for one of the following components in the Ran GTPase cycle: rna1-1 (RanGAP1), prp20-1 (RCC1), yrb1-51 (RanBP1), or yrb2Δ (RanBP2). FLAG-PpgA, FLAG-LegG1, or FLAG-PieG was produced in these mutant strains, and growth was assessed by dot spot assays (Fig. 3C). Interestingly, the production of FLAG-PpgA did not inhibit the growth of S. cerevisiae rna1-1, a temperature-sensitive mutant strain producing a nonfunctional RanGAP1 protein. Growth inhibition by FLAG-PpgA was restored upon production of wild-type RNA1 in the rna1-1 mutant strain (Fig. 3D; see also Fig. S6B). This result indicates that growth inhibition by PpgA is mediated partially by the Ran GTPase cycle and, more specifically, requires functional RanGAP1 (RNA1). In order to further confirm the role of RNA1 for the inhibitory effect of PpgA, growth of S. cerevisiae BY4741 or rna1-1 producing FLAG-PpgA, FLAG-LegG1, or FLAG-PieG in SG medium without leucine was monitored for 7 days at 20°C, and the optical density at 600 nm (OD₆₀₀) was measured every hour (Fig. 3E). Under these conditions, PpgA but not LegG1 or PieG significantly slowed yeast growth in the wild-type strain, and the growth inhibition mediated by PpgA was partially alleviated in the rna1-1 mutant strain.

Intriguingly, the production of FLAG-PieG or FLAG-PpgA also inhibited the growth of the S. cerevisiae yrb1-51 (RanBP1) strain (Fig. 3C), suggesting that PieG and PpgA might target a RanBP1-dependent pathway. In summary, the L. pneumophila RCC1 repeat effector PpgA inhibited the growth of wild-type yeast in a RanGAP1-dependent manner, and PieG as well as PpgA inhibited the growth of a strain lacking functional RanBP1. These findings are in agreement with the notion that different components of the Ran GTPase cycle are the targets of L. pneumophila RCC1 repeat effectors.

PpgA and LegG1 target distinct Ran GTPase cycle components.To test the hypothesis that the RNA1-dependent growth inhibition of yeast by PpgA is due to a direct interaction between the RCC1 repeat effector and the RanGAP1 RNA1, the putative interaction partners were ectopically produced in S. cerevisiae. Specifically, we produced Strep-tagged RNA1 or the mutant protein RNA1-1 and FLAG-tagged PpgA or LegG1 in strain BY4741. The rna1-1 allele contains two single-base-pair substitutions, resulting in two amino acid changes (S17F and A194V), which impairs enzyme activity and renders the strain temperature sensitive due to failure of RNA processing and nuclear export at the nonpermissive temperature (59). The interactions with RCC1 repeat effectors were tested by performing an anti-FLAG co-immunoprecipitation (co-IP) assay in yeast lysates, followed by anti-Strep Western blotting (Fig. 4A). Using this approach, FLAG-PpgA bound to RNA1 but not to RNA1-1, while FLAG-LegG1 did not bind to either host protein. These results indicate that PpgA but not LegG1 interacts with the RanGAP1 RNA1 and that the RNA1-1 mutations abolish the interaction. Hence, the PpgA-RNA1 interaction appears to be required for effector function.

• Open in new tab
• Download powerpoint

FIG 4

PpgA and LegG1 target distinct Ran GTPase cycle components and promote Ran activation. (A) S. cerevisiae BY4741 ectopically producing Strep-tagged RNA1 (pLS128) or RNA1-1 (pLS127) and FLAG-tagged PpgA (pLS085) or LegG1 (pLS084) was lysed, a co-IP with anti-FLAG was performed, and interactions were revealed by anti-Strep Western blotting. Yeast lysates were analyzed by Western blotting with anti-Strep (input control) and anti-FLAG (co-IP control). (B) Lysates of HEK 293T cells were incubated with Strep-LegG1_33–286 bound to Strep-Tactin resin, and interaction partners were identified by mass spectrometry. (C) Yeast two-hybrid assays using S. cerevisiae reporter strain AH109 containing plasmids encoding the GAL4 DNA-binding domain (BD; pGBKT7) alone or fused to RanBP10 (pLS213), RanGAP1 (pM1593), Ran_WT (pM1057), Ran_T24N (pM1059), Ran_Q69L (pM1058), or Ran_N122D (pM1286) and the GAL4 activation domain (AD) fused to LegG1 (pLS211). Transformants were spotted onto SD lacking His (−His) or SD plus His (+His) and incubated at 30°C for 5 days. (D) Yeast two-hybrid assay in reporter strain AH109 containing plasmids encoding the GAL4 DNA-binding domain (BD; pGBKT7) alone or fused to RanBP10 (pLS213) and the GAL4 activation domain (AD; pGADT7) alone or fused to LegG1 (pLS211), PpgA (pM1028) or PieG (pM1027). Transformants were spotted in 10-fold serial dilutions onto SD lacking His (−His) or SD plus His (SD+) and incubated at 30°C for 5 days. (E) Human A549 cells transfected for 48 h with 10 nM siRNA oligonucleotides targeting RanGAP1, RanBP10, or Arf1 (positive control) or with AllStars siRNA (“scrambled”) were infected (MOI 10) with GFP-producing L. pneumophila JR32 (pNT28). Intracellular bacterial replication was assessed by fluorescence increase with a fluorescence plate reader after 24 h and compared to the levels seen at 1 h. Means and standard deviations of results from three independent experiments are shown (one-way ANOVA; **, P < 0.01; all groups compared to scrambled). (F) The depletion efficiency of siRNA oligonucleotides targeting RanGAP1 or RanBP10 upon transfection of A549 epithelial cells for 48 h was assessed by Western blotting with the antibodies indicated. For each target protein four different oligonucleotides were used. Untreated cells were used as a negative control (“Untr.”), Qiagen AllStars unspecific oligonucleotides (“Scrambled,” “Scr.”) were used to control for off-target effects, and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) served as the loading control. Data are representative of results from two independent experiments. (G) HEK 293T cells were transfected for 24 h with constructs producing Strep-tagged PpgA (pLS229), LegG1 (pLS226), PieG (pLS230), or RCC1 (pLS231). Nontransfected cells (-) were taken along, and lysates were treated with GTPγS (positive control) or GDP (negative control). Ran(GTP) was precipitated with RanBP1-coupled agarose beads in cell lysates, and the amount of Ran(GTP), total Ran, and GAPDH (loading control) was assessed by Western blotting (top panel). The amount of Ran(GTP) relative to total Ran was quantified using ImageQuant TL (bottom panel). Means and standard deviations of results from three independent experiments are shown for the first four bands (one-way ANOVA; *, P < 0.05; **, P < 0.01; ***, P < 0.001; all groups compared to the nontransfected cells [-]).

In order to identify possible interaction partners and targets of LegG1, we incubated purified Strep-tagged LegG1_33–286 (LegG1 lacking only the 32 N-terminal amino acids) with HEK 293T cell lysates, treated these with Strep-Tactin resin, and separated the eluate by SDS-PAGE (Fig. 4B). Under the conditions used, several proteins eluted from the beads and were identified by mass spectrometry (Table S2B and C). Among the proteins identified were Ran GTPase, RanGAP1, RanBP1, RanBP9, and RanBP10, a cytoplasmic Ran GEF implicated in the modulation of noncentrosomal microtubules (45).

To assess the interaction between L. pneumophila RCC1 repeat effectors and potential interaction partners, we employed yeast two-hybrid assays (Fig. 4C and D). To this end, we fused LegG1 to the GAL4 activation domain (AD) and RanBP10, RanGAP1, Ran (WT), or different Ran mutants to the GAL4 DNA binding domain (BD). Moreover, we fused LegG1, PpgA, or PieG to the GAL4 activation domain and RanBP10 to the GAL4 DNA binding domain. Transformants of the AH109 reporter strain were spotted in 10-fold serial dilutions onto SD lacking His (−His) or SD plus His (+His) and incubated at 30°C for 5 days. While all of the yeast strains grew in the presence of His (growth control), the strain harboring AD-LegG1 and BD-RanBP10 but not BD-RanGAP1 or BD-Ran (wild type, constitutively active, dominant negative) grew in the absence of His (Fig. 4C and D). Hence, among the putative interactions of LegG1 with Ran cycle components detected by the sensitive mass spectrometry approach, only the robust interaction with RanBP10 was validated. Under selective conditions, no growth was observed for yeast strains producing AD-PpgA or AD-PieG or strains harboring the AD or BD empty vectors. Collectively, these data confirm that LegG1 specifically binds to the cytoplasmic Ran GEF, RanBP10, but neither PpgA nor PieG does so.

RanGAP1 is implicated in intracellular L. pneumophila replication.Previously, we found that the depletion of Ran or RanBP1 by RNA interference reduces intracellular growth of L. pneumophila (47). To assess whether RanGAP1 and RanBP10 play a role in intracellular replication of the pathogen, the proteins were individually depleted. To this end, human A549 lung epithelial cells were treated with small interfering RNA (siRNA) oligonucleotides targeting RanGAP1, RanBP10, or Arf1 (positive control), and intracellular replication of GFP-producing L. pneumophila was monitored by fluorescence after 24 h (Fig. 4E), as previously described (37, 47, 60). Under the conditions described and in the medium used, L. pneumophila grows only intracellularly, and the GFP production (fluorescence intensity) is proportional to the number of bacteria (expressed in CFU counts). Upon depletion of RanGAP1, the fluorescence intensity of intracellular L. pneumophila was reduced approximately 2-fold, indicating that Ran inactivation plays a role in efficient intracellular growth of L. pneumophila. In contrast, the depletion of RanBP10 did not have a significant effect. The depletion of Arf1 reduced intracellular growth of L. pneumophila to a similar extent as depletion of RanGAP1. The treatment with siRNA oligonucleotides efficiently depleted RanGAP1 or RanBP10 (Fig. 4F), but had no effect on host cell viability (data not shown). Taken together, the results indicate that the depletion of RanGAP1 reduces the intracellular replication of L. pneumophila without affecting the host cell physiology, while the Ran GEF RanBP10 is dispensable or plays a redundant role for intracellular growth of the pathogen.

Ectopic production of RCC1 repeat effectors promotes Ran activation.Given that the L. pneumophila RCC1 repeat effectors target different components of the Ran GTPase cycle, we assessed their effect on Ran activation. To this end, we ectopically produced Strep-tagged PpgA, LegG1, PieG, or RCC1 in HEK 293T cells. After cell lysis, active Ran was precipitated with RanBP1-coupled agarose beads, and the amount of Ran(GTP) was determined by Western blotting (Fig. 4G). Upon production of the L. pneumophila RCC1 repeat proteins, the cellular amount of Ran(GTP) was increased by 2-fold to 3-fold compared to the control. As expected, overproduction of Strep-RCC1 significantly increased the amount of active Ran GTPase, as did incubation of the cleared lysate with a slowly hydrolyzable form of GTP, GTPγS, but not GDP. These results demonstrate that while the bacterial effectors target distinct components of the Ran GTPase cycle, they all promote the activation of the small GTPase, albeit less efficiently than ectopically produced RCC1.

The Lpg1975-LegG1 fusion protein adopts PieG substrate specificity.An interesting pattern seen among all but one of the sequenced L. pneumophila genomes was that if two RCC1 paralogs were present in a genome, one of them was split (Fig. 1). Thus, the single RCC1 repeat gene in strain Paris, pieG, was split in strain Philadelphia-1 into legG1 and the upstream gene, lpg1975. A single point mutation in the gene corresponding to pieG (deletion of a thymidine phosphate) caused a frameshift and resulted in the two open reading frames, lpg1975 and legG1 (Fig. 5A). This pattern is conserved and was observed for at least 31 L. pneumophila strains (Fig. 1). Reverting the frameshift mutation, we created the lpg1975-legG1 fusion by insertion of a thymidine phosphate. The corresponding Lpg1975-LegG1 fusion protein (strain Philadelphia-1) is approximately 90% identical to PieG (strain Paris) (Fig. S7). In order to characterize potential functional consequences of the point mutation leading to split PieG, we assessed the host cell localization and targets of the L. pneumophila RCC1 repeat effectors.

• Open in new tab
• Download powerpoint

FIG 5

PieG and the fusion protein Lpg1975-LegG1 interact with Ran and RanGAP1. (A) A point mutation (insertion of thymidine phosphate) in L. pneumophila lpg1975 (strain Philadelphia-1) causes a frameshift and fuses the open reading frames of lpg1975 and legG1 (lpg1976), resulting in a fusion gene with high homology to pieG (strain Paris). (B) A co-IP screen was performed with yeast coproducing FLAG-PpgA (pLS085), FLAG-LegG1 (pLS084), FLAG-PieG (pLS086), or FLAG-Lpg1975-LegG1 fusion (pLS087) and RNA1-Strep (pLS128), Strep-PRP20 (pLS185), Strep-YRB1 (pLS186), or Strep-GSP1 (pLS173). Yeast lysates were incubated with FLAG beads, and eluates were analyzed for co-IP of Strep-proteins by anti-Strep Western blotting. Yeast lysates were analyzed by Western blotting with anti-Strep (input control) and anti-FLAG (co-IP control). (C) Lysates of yeast harboring pYEP351gal-FLAG-ppgA (pLS085), pYEP351gal-FLAG-legG1 (pLS084), pYEP351gal-FLAG-lpg1975-legG1 fusion (pLS087), or pYEP351gal-FLAG-pieG (pLS086) were incubated with FLAG-agarose beads, and eluates were analyzed for the co-IP of endogenous GSP1 using a specific antibody. (D) HEK 293T cells were transfected for 24 h with plasmids harboring GFP-Lpg1975 (pLS233), GFP-LegG1 (pLS227), GFP-Lpg1975-LegG1 (Fusion) (pLS234), GFP-PpgA (pLS095), GFP-PieG (pLS093), or GFP-PieG_split (pLS235) and Strep-RanBP10 (pLS242). Cell lysates were analyzed by anti-GFP Western blotting (input control), or incubated with anti-Strep antibody and A/G agarose beads, and eluates were analyzed by Western blotting for co-IP of GFP fusion proteins using anti-GFP and anti-Strep (co-IP control). (E) Real-time fluorescence microscopy of LCV motility in D. discoideum producing calnexin-GFP (pCaln-GFP) (green) infected (MOI 5, 1 to 2 h) with L. pneumophila JR32 (WT) or with the ΔlegG1-ΔppgA mutant producing DsRed alone (pCR077) or together with M45-tagged PieG (pLS033), Lpg1975-LegG1 fusion protein (pLS026), LegG1 (pER005), or PpgA (pLS008). LCV motility was recorded for 180 s with images taken every 10 s and quantified using ImageJ/Fiji software with the manual tracking plugin (n > 50/strain; 3 independent experiments; one-way ANOVA; ***, P < 0.001; all groups compared to ΔlegG1-ΔppgA).

Similarly to LegG1 from strain Philadelphia-1, PieG from strain Paris localizes to LCVs (Fig. S5C). In order to identify potential interaction partners and targets of the RCC1 repeat effectors, a co-IP screen was performed with yeast producing FLAG-tagged PpgA, LegG1, PieG, or Lpg1975-LegG1 (Fusion) and RNA1-Strep (RanGAP1), Strep-PRP20 (RCC1), Strep-YRB1 (RanBP1), or Strep-GSP1 (Ran). Yeast lysates were incubated with anti-FLAG beads, and eluates were analyzed for co-IP of Strep-tagged host proteins by anti-Strep Western blotting (Fig. 5B). Strikingly, FLAG-PieG as well as the FLAG-Lpg1975-LegG1 fusion protein, but not FLAG-LegG1, bound both RNA1-Strep (RanGAP1) and Strep-GSP1 (Ran). As observed before, FLAG-PpgA interacted with RNA1-Strep (RanGAP1). Under these conditions, LegG1 did not bind to any of the Ran GTPase cycle components tested. These results indicate that the Lpg1975-LegG1 fusion protein created upon insertion of a single nucleotide targets the same Ran GTPase cycle components as PieG.

To confirm these results with endogenous GSP1 (Ran), we produced FLAG-tagged PpgA, LegG1, Lpg1975-LegG1 (Fusion), or PieG in the S. cerevisiae wild-type strain and incubated lysates with FLAG-agarose beads. Eluates from the beads were analyzed for the co-IP of endogenous GSP1 using an anti-GSP1 antibody (Fig. 5C). Under these conditions, FLAG-PieG as well as the FLAG-Lpg1975-LegG1 fusion protein precipitated GSP1 (Ran), while LegG1 did not.

In order to assess the RCC1 repeat effector targets in mammalian cells, we performed co-IP experiments in HEK 293T cells (Fig. 5D). To this end, the cells were transfected with a plasmid producing Strep-RanBP10 and plasmids producing GFP-Lpg1975, GFP-LegG1, GFP-Lpg1975-LegG1 (Fusion), GFP-PpgA, GFP-PieG, or GFP-PieG_split (corresponding to LegG1). Cell lysates were incubated with anti-Strep antibody and protein A/G agarose beads, and eluates were analyzed for co-IP of GFP using an anti-GFP antibody. Under these conditions, RanBP10 was bound by LegG1, validating the binding studies performed in yeast (Fig. 4). Moreover, RanBP10 was weakly bound by Lpg1975 as well as by the Lpg1975-LegG1 fusion protein, but not at all by PieG or the C-terminal part of PieG (PieG_split), which corresponds to but is not identical with LegG1.

Finally, to assess whether the Lpg1975-LegG1 fusion protein can functionally replace PieG, we sought to complement the LCV motility phenotype of the ΔlegG1-ΔppgA double mutant strain (Fig. 5E). D. discoideum producing the ER/LCV marker calnexin-GFP was infected with the red fluorescent L. pneumophila wild-type strain or with the ΔlegG1-ΔppgA mutant producing M45-tagged PieG, Lpg1975-LegG1 fusion protein, LegG1, or PpgA, and LCV motility was recorded by fluorescence microscopy. The LCV motility defect of the ΔlegG1-ΔppgA double mutant strain was restored to similar extents by Lpg1975-LegG1, PieG, and LegG1, indicating that the fusion protein is functional.

In summary, the results of this study indicate that the divergent evolution of L. pneumophila RCC1 repeat effectors defines the range of Ran GTPase cycle targets. Accordingly, PpgA targets RanGAP1, LegG1 binds RanBP10, and PieG as well as the Lpg1975-LegG1 fusion interact with Ran and RanGAP1 (Fig. 6; see also Table 1). The L. pneumophila RCC1 repeat effectors promote the activation of Ran GTPase, and thus, our working model stipulates that LegG1 activates the GEF RanBP10, PpgA inhibits RanGAP1, and PieG stabilizes Ran(GTP) and/or inhibits RanGAP1.

• Open in new tab
• Download powerpoint

FIG 6

Ran GTPase cycle targets of L. pneumophila RCC1 repeat effectors. The different Ran GTPase cycle targets of the L. pneumophila RCC1 repeat effectors LegG1, PpgA and PieG (red) are indicated. Divergent evolution of L. pneumophila RCC1 repeat effectors defines the range of target components of the Ran GTPase cycle. Regardless of the distinct target, the L. pneumophila RCC1 repeat effectors promote the activation of Ran. Putative activation (>), inhibition (|), or binding (×) of Ran GTPase cycle components by L. pneumophila RCC1 repeat effectors is indicated with different arrow endings. For details see text.