L. pneumophila and E. coli strains (listed in Table S1 ) used in this work were grown as previously described (Shohdy et al., 2005; Franco et al., 2012). For construction of deletion mutants Δlpp1450, Δlpp3070 and Δvfx10045, L. pneumophila strains were transformed with DNA fragments containing a kanamycin resistance cassette (Kan^R) flanked by ~1kb regions upstream or downstream from the corresponding genes. To facilitate the attainment of this DNA fragment, each of these three regions (“upstream”, “Kan^R” and “downstream”) was independently amplified by PCR using oligos with restriction sites that allowed their progressive cloning at high copy number plasmid pUC18 (Thermo Fisher Scientific). The PCR product subsequently used for transformation was then obtained by amplification with the flanking primers (see Table S2 ).

Plasmids and oligonucleotides used in this study are listed, respectively, in Tables S2 and S3, as well as details of how plasmids were constructed. For general cloning procedures, restriction enzymes (Thermo Fisher Scientific), T4 DNA ligase (Thermo Fisher Scientific), Phusion polymerase (Thermo Fisher Scientific) were used according to the manufacturer’s instructions. For site-directed mutagenesis, amino acid substitutions in VFX10045 were made by overlap PCR with oligonucleotides carrying appropriate modified nucleotides. The accuracy of the nucleotide sequence of the constructs was confirmed by DNA sequencing.

CHO FcγRII cells (Joiner et al., 1990) were grown in Dulbecco’s Modified Eagle Medium (DMEM; Corning) and 10% (v/v) heat-inactivated fetal bovine serum (FBS; Thermo Fisher Scientific), at 37°C in a 5% (v/v) CO₂ incubator. For immunoblot experiments, CHO cells were seeded on 24-well plates at 1x10⁵ cells/well, for microscopy experiments at 5x10⁴ cells/well. CHO cells were transfected using the jetPEI™ reagent (Polyplus) according to manufacturer’s protocol for 24 hours. THP-1 human monocyte-like cells were grown in RPMI 1640 (Thermo Fisher Scientific) and 10% FBS (Thermo Fisher Scientific), 1 mM L-Glutamine (Thermo Fisher Scientific), 10 mM HEPES (Thermo Fisher Scientific), Sodium Pyruvate 1 mM (Thermo Fisher Scientific) and 0.05 mM β-mercaptoethanol at 37°C in a 5% (v/v) CO₂ incubator. Infection of THP-1 cells was carried out as described below. For infection experiments, THP-1 cells were seeded at 5x10⁵ cells/well and allowed to differentiate for 24 hours in the presence of phorbol 12-myristate 13-acetate (PMA), after which they were incubated with fresh RPMI for additional 24 hours.

Transfected CHO cells or infected THP-1 cells were fixed and permeabilized for immunofluorescence microscopy as described previously, using Triton X 0.1% for cell permeabilization. For labelings, we used mouse anti-FLAG (Sigma; 1:200), mouse anti-myc (Calbiochem; 1:200), rat anti-HA (Sigma; 1:200), followed by appropriate fluorophore-conjugated anti-mouse antibodies (Jackson ImmunoResearch; 1:200). 4′,6-Diamidino-2-phenylindole (DAPI; 1:30.000) was used to label DNA, and actin staining was carried out by incubating CHO cells with Phalloidin-Alexa555 (Thermo Fisher Scientific, 1:200) during 30 min. Images were acquired on an Axio Imager D2 (Zeiss) and processed with ZEN or Fiji software. For each enhanced GFP (EGFP) fusion protein, the proportion of protein in the nucleus was determined by calculating the ratio between the average GFP fluorescence in the nucleus and the average GFP fluorescence in the cytosol. Quantification of these values was made in Fiji, using the DAPI stain to delineate the nucleus and the F-actin stain for the cell outline. For each protein construct, fluorescence was quantified for at least 60 cells from 3 independent experiments. Statistical significance was assessed with Student’s T-test.

For the preparation of bacterial cell extracts, L. pneumophila strains were grown on charcoal yeast extract (CYE) plates supplemented with appropriate antibiotics for 4-5 days at 37°C. Bacteria were patched on identical plates with isopropyl β-D-1-thiogalactopyranoside (IPTG) 1 mM and grown for additional 24 hours. From these plates, a loop of bacteria was removed and resuspended in sterile water, the optical density at 600 nm (OD₆₀₀) measured and resuspended in appropriate volumes of SDS-PAGE loading buffer in order to obtain the same concentration of bacteria. For the preparation of mammalian cell extracts, transfected CHO cells were washed with PBS and trypsinized with 100 µl Tryple Express (Thermo Fisher Scientific), resuspended in 900 µl DMEM, collected by centrifugation at 720 g for 5 min and washed in phosphate-buffered saline (PBS) twice. The pellets were resuspended in SDS-PAGE loading buffer, boiled for 5 min and loaded for SDS-PAGE, after which gels were processed for immunoblotting using Trans-Blot Turbo Transfer System (BioRad) and 0.2 μm pore-size nitrocellulose membranes (BioRad). The following antibodies were used for immunoblotting: goat anti-GFP (SICGEN; 1:1000), mouse anti-α-Tubulin (Sigma; 1:1000), mouse anti-FLAG (Sigma-Aldrich; 1:1000), rat anti-HA (Sigma; 1:1000), rabbit anti-myc (Cell Signalling Technologies), mouse anti-TEM (QEDBiosciences Inc; 1:500), rabbit anti-Legionella pneumophila MOMP (1:1000) followed by appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies (GE Healthcare or Jackson ImmunoResearch; 1:10.000). Immunoblot detection was done with SuperSignal West Pico ECL (Thermo Scientific) and exposure to Amersham Hyperfilm ECL (GE Healthcare).

Assays were performed essentially as described (Allombert et al., 2017) using a LiveBLAzer™ FRET - B/G Loading CCF2/AM kit (Thermo Fisher Scientific). L. pneumophila strains harboring pXDC61 derivatives encoding TEM fusions to putative effectors (wildtype or dotA background; see Table S1 and S2) were grown in ACES-buffered yeast extract (AYE) supplemented with 1 mM IPTG and required antibiotics at 37°C overnight with agitation. THP-1 human monocyte-like cells were differentiated on coverslips and infected for 1 hour using an MOI=50. Cells were loaded with CCF2/AM, incubated for 2 hours in the dark, washed with PBS and fixed for 20 min with 4% PFA, followed by mounting. For quantification of protein delivery (translocation) into host cells images were acquired from the DAPI and FITC channels using fixed exposure times and ZEN software (Zeiss). Background fluorescence was imaged from coverslips without cells. To obtain the ratio of blue/green fluorescence, total blue and total green fluorescence in each image was measured (Fiji software), the background blue and green fluorescence was subtracted, and blue/green ratio calculated. Data obtained represent the mean value of blue/green ratios from three independent experiments, each with 11 analysed images for each strain.

THP-1 cells were seeded at 5x10⁵ cells/well in 24-well plates and differentiated as described above. L. pneumophila strains were grown on CYE plates for 4-5 days at 37°C. From these plates, a loop of bacteria was removed and resuspended in sterile water, the OD₆₀₀ measured and appropriate dilutions made to obtain an infection stock at 5x10⁵ bacteria/ml. A volume of 100 µl of the infection stock was added to each well, corresponding to a multiplicity of infection of 0.1. Infection was synchronized by centrifuging at 800 g for 10 min. Infection was allowed to occur for 1.5 hours in a CO₂ incubator, after which cells were washed with PBS and RPMI containing gentamicin (100 µg/ml) was added to each well. Following an additional incubation for 1 hour, cells were washed with PBS and fresh RPMI was added (T₀). To calculate CFUs at appropriate time-points, lysis was carried out in the following manner: cells were washed with PBS, incubated with 1 ml of sterile water at room temperature for 15 min and vigorously resuspended. The lysate was subjected to appropriate serial dilutions and 100 µl plated on CYE, incubated for 5 days at 37°C and colony forming unites (CFUs) counted.

To search for previously unidentified Icm/Dot substrates/effectors in the novel strain L. pneumophila Pt/VFX2014 (draft genome accession available from https://www.ncbi.nlm.nih.gov/nuccore/LORH00000000.1), we first screened the Pt/VFX2014 predicted protein sequences (https://sra-download.ncbi.nlm.nih.gov/traces/wgs03/wgs_aux/LO/RH/LORH01/LORH01.1.fsa_aa.gz) using the web EffectiveDB platform (https://effectors.csb.univie.ac.at/; accessed on 25-26 July 2016) (Jehl et al., 2011; Eichinger et al., 2016), which provides several online tools for detecting putative bacterial effectors and predicting eukaryotic-like domains likely to interact with host proteins. Specifically, the following tools (and settings) were applied: i) T4SEpre (minimal score: 0.9999), which predicts Type IV secreted proteins based on amino acid composition in C-termini; ii) EffectiveELD (minimal score: 5), which predicts secreted proteins based on eukaryotic-like domains; iii) Predotar (model: human/animal), which predicts subcellular localization of secreted proteins in the host cell; and iv) EffectiveS346 (enabled), which predicts Type III, IV, VI secretion systems for protein sequences from (nearly) complete genomes. From the 128 out of the 2991 proteins predicted as potential T4SS effectors by T4SEpre, 23 and two proteins also had a hit in Predotar and EffectiveELD, respectively. Of these 25 candidates, only four (PtVFX2014_03805, PtVFX2014_06065, PtVFX2014_12350 and PtVFX2014_13425) have not previously been described as putative effectors in literature after cross-checking against a custom database of the repertoire of known Dot/Icm substrates in Pt/VFX2014 strain (summarized in Supplementary Table 2 in Borges et al., 2016 and additional literature (Zhu et al., 2011). In parallel to this in silico prediction, we navigated through the Pt/VFX2014 genome annotation to identify (clusters of) genes (not previously unidentified as Icm/Dot effectors) that are encoded adjacently to known T4SS substrates. From this search, we selected these additional candidates: i) PtVFX2014_05045 and PtVFX2014_05055 (absent in Philadelphia-1 strain, but encoded in a region - lpg1491-lpg1496 - encoding known T4SS substrates); ii) PtVFX2014_09510 (absent in Philadelphia-1 strain, but encoded adjacently to the T4SS substrate lpg0096/ceg4; and, iii) PtVFX2014_10045 (absent in Philadelphia-1 strain, but encoded adjacently to the T4SS substrated lpg2999 and lpg3000). In summary, eight candidate proteins (PtVFX2014_03805, PtVFX2014_06065, PtVFX2014_12350, PtVFX2014_13425, PtVFX2014_05045 [GenBank KZX34370.1], PtVFX2014_05055, PtVFX2014_09510 and PtVFX2014_10045 [GenBank KZX33632.1]) were selected to proceed to experimental assays. For the sake of simplicity, the prefix of GenBank locus tags “PtVFX2014_” is referred as “VFX” throughout the text.

Based on in silico predictions (as detailed in Materials and Methods), eight proteins of the novel strain L. pneumophila Pt/VFX2014 were selected as candidate Icm/Dot substrates: VFX03805, VFX05045, VFX05055, VFX06065, VFX09510, VFX12350, VFX13425 and VFX10045. To assess if these putative effectors were indeed delivered (i.e., translocated) into host cells by the Icm/Dot T4SS secretion system, we used the TEM-1 β-lactamase FRET-based reporter system (Allombert et al., 2017). In this methodology, fusion proteins containing N-terminal TEM-1 are produced in L. pneumophila under the control of the Ptac IPTG-inducible promoter. Their transport into to the cytosol of macrophages previously loaded with the compound CCF2/AM should lead to its cleavage by the TEM-1 β-lactamase, resulting in the emission of blue fluorescence which contrasts to the intrinsic green fluorescence of CCF2/AM. Plasmids encoding fusion proteins of TEM-1 to the eight putative effectors ( Table S2 ) were transformed into L. pneumophila JR32 wild-type, or into an Icm/Dot-deficient derivative (dotA mutant; used as negative control for type 4 secretion mediated transport into host cells). As additional controls, we used strains expressing TEM-1 fusions to FabI, a non-translocated L. pneumophila protein, or to the characterized effector LepA (de Felipe et al., 2008). Production of proteins TEM-VFX06065 and TEM-VFX09510 was not detected by immunoblotting, thus the analyses were performed with the remaining six candidates. Production of these six TEM-1 fusion proteins by L. pneumophila was tested by immunoblot using an anti-TEM antibody ( Figure S1A ). Infection of THP-1 macrophages was carried out for 2 hours, after which CCF2/AM was added and cleavage of the compound allowed to occur for 1 hour. The presence of the candidate effectors in the macrophage cytosol was then assessed by fluorescence microscopy ( Figure 1A ). Only L. pneumophila strains expressing protein fusions TEM-1-VFX05045 and VFX10045 gave rise to a blue fluorescence in macrophages, identical to the TEM-LepA positive control. Determination of total blue and green fluorescence and calculation of the blue/green (B/G) ratio allowed the quantification of translocated TEM-1 fusion proteins for each strain. Comparison of the B/G ratio obtained for the positive and negative controls confirmed the translocation of VFX05045 and VFX10045 via the Icm/Dot secretion system ( Figure 1B ). Therefore, proteins VFX05045 and VFX10045 encoded by the outbreak strain L. pneumophila Pt/VFX2014 are newly discovered T4SS substrates transported into the host cell cytosol during infection.

A search in the genomes of other L. pneumophila strains revealed the presence of coding regions for homologs of VFX05045 and VFX10045 in L. pneumophila Paris, Alcoy and Corby, but missing in L. pneumophila Philadelphia-1 and in other Legionella sp. ( Figure 2 ). The genomic region surrounding vfx10045 is conserved in strains L. pneumophila Paris, Alcoy and Corby, and the genome of strain Philadelphia-1 contains an identical region except for the absence of the vfx10045 locus. In contrast, for vfx05045 the genomic context is only partially conserved in L. pneumophila Paris and distinct in L. pneumophila Alcoy and Corby. For both vfx05045 and vfx10045, their expression may be regulated by predicted antisense non-coding RNAs that target the effector transcripts, lppnc0353 and lppnc0712, respectively (Sahr et al., 2012). Evaluation of the amino acid sequence of the VFX05045 and VFX10045 homologs revealed that L. pneumophila Paris proteins Lpp1450 and Lpp3070 shared the highest degree of identity, 90% and 94% respectively ( Figure S2 and Figure 2 ). To confirm that Lpp1450 and Lpp3070 are also bona fide substrates of the Icm/Dot system, studies with TEM-1 β-lactamase fusions were performed, as described above. Using this methodology, the presence of both proteins was detected in the macrophage cytosol, showing that L. pneumophila Paris Lpp1450 and Lpp3070 are also Icm/Dot substrates ( Figure 1C ; Figure S1B ).

To examine the importance of L. pneumophila Lpp1450 and Lpp3070 proteins for an efficient infection of macrophages, null mutants were constructed by allelic replacement with a Kan^R cassette. THP-1 macrophages were then infected with wild-type L. pneumophila Paris or derivatives ΔdotA, Δlpp3070 and Δlpp1450, and replication inside these cells was monitored by CFU counting at 0 h, 24 h and 48 h after infection ( Figure 2B ). The results showed no significant differences in uptake or intracellular replication resulting from deletion of either lpp1450 or lpp3070. Similarly, loss of vfx10045 did not yield any defect in this assay when compared to wild-type L. pneumophila Pt/VFX2014. This lack of phenotype is not surprising, as deletion of most L. pneumophila effector genes has no obvious repercussion on replication rate. Thus, effectors Lpp1450, Lpp3070 and VFX10045 are not essential for L. pneumophila replication inside macrophages.

To gain insight into the function of these newly discovered Icm/Dot substrates, we started by assessing their subcellular localization as EGFP fusion proteins when expressed ectopically in mammalian CHO cells, a cell line commonly used as a model for these studies. For this, plasmids encoding EGFP, EGFP-VFX05045, EGFP-Lpp1450, EGFP-VFX10045, or EGFP-Lpp3070 were used to transiently transfect CHO cells, which were further analysed by fluorescence microscopy after fixation and fluorescence staining of the nucleus and actin cytoskeleton. Localization of EGFP-VFX10045 was identical to the one displayed by cognate EGFP-Lpp1450, being both homogeneously dispersed in the cell cytosol ( Figure 3 ). However, EGFP-VFX10045 showed a striking localization in the cell nucleus, which was identical when using other tags in this protein (myc-VFX10045, VFX10045-myc, and 3xFLAG-VFX10045; Figure S3 ). Surprisingly, its homolog Lpp3070 showed a different distribution pattern, being mostly spread in the cytosol with some occasional enriched patches (EGFP-Lpp3070 and Lpp3070-myc; Figure S3 ). To rule out the possibility of an artifactual localization of the fusion proteins due to their degradation or cleavage in the cell, we confirmed by western blot that they were being produced as the expected full-length versions (approximately 68 kDa for VFX05045/Lpp1450 and 95 kDa for VFX10045/Lpp3070; Figures S1D , S3B ).

To analyse the localization of these proteins in the context of macrophage infection, L. pneumophila Paris derivative strains were used that expressed plasmid-encoded fusions of VFX05045, VFX10045, Lpp1450 and Lpp3070 to an N-terminal 4xHA tag. Infection of THP-1 macrophages was carried out for 24 hours, a time-point at which all translocated effectors were clearly visualized within host cells. After fixation and permeabilization, fluorescence labeling was performed for F-actin, DNA (cell nucleus and bacteria) and the 4HA-tagged proteins. Homologs VFX05045 and Lpp1450 dispersed essentially throughout the host cell cytosol ( Figure 4 , upper panels), in agreement with what was observed in the transfection experiments ( Figure 3 ). 4HA-VFX10045 accumulated in the nucleus of most infected cells, while in contrast homolog 4HA-Lpp3070 essentially distributed in the cytosol with occasional enrichments, namely in cortical cell-cell contact regions ( Figure 4 ). All four 4HA-tagged proteins migrated on SDS-PAGE according to their expected molecular mass, when analyzed by immunoblotting ( Figure S1C ). Taken together, subcellular localization analyses of these L. pneumophila Icm/Dot substrates were consistent in transfection and infection experiments and revealed a nuclear tropism for VFX10045.

Transport of proteins into the eukaryotic nucleus may be accomplished by passive diffusion or by transport factors that recognize specific nuclear localization sequence(s) (NLSs). The migration pattern observed in SDS-PAGE for the previously analysed tagged versions of effector VFX10045 ( Figures S1 and S3B ) is in accordance with the predicted molecular mass of ~68 kDa for the untagged protein, which precludes its localization to the nucleus via diffusion, as this does not occur in macromolecules over 50-60 kDa.

To identify the mechanism responsible for the nuclear localization of VFX10045 (583 amino acid residues), we initially searched for NLS(s) using the softwares NLStradamus (Nguyen Ba et al., 2009), NLS MAPPER (Kosugi et al., 2009) and SeqNLS (Lin et al., 2012). Two highly scored putative canonical NLSs were identified that consisted of regions enriched in positively charged residues and containing the consensus sequence K-K/R-x-K/R between residues 42-45 (KKIK), and between residues 550-555 (KRKNKK) ( Figure S4 ). To verify if these corresponded to the signal(s) directing the protein to the nuclear compartment, we performed site-directed mutagenesis on the plasmid encoding EGFP-VFX10045 followed by analyses of the localization of the mutant proteins in transfected CHO cells ( Figure 5 and Figure S5 ). In each case, to measure the impact of the engineered amino acid replacements on the nuclear targeting of EGFP-VFX10045, we calculated the amount of protein in the nucleus relatively to the protein in the cytosol. This was expressed as the ratio nucleus/cytosol, or N/C (see Materials and Methods for details), which in the case of EGFP-VFX10045 is approximately 4 ( Figure 5A ; representative microscopy images in Figure S5 and immunoblots in Figure S6 ).

Nuclear localization of the mutant EGFP-VFX10045 proteins with single replacements K₄₃A or R₅₅₁A (expected to disrupt each putative NLSs), multiple replacements KRKNKK_550-555AAAAAA or K₄₃A combined with R₅₅₁A was assessed, but no significant changes were observed relative to wild-type EGFP-VFX10045 ( Figure 5 and Figure S5 ). To search for non-canonical nuclear targeting region(s), we then examined the effect of deletions, starting by constructs containing only the N-terminal half of VFX10045 (residues M₁-K₂₉₃) or the C-terminal half region (residues I₂₉₄-I₅₈₃). Visualization of transfected cells showed that the C-terminal region (I₂₉₄-I₅₈₃) is essential and sufficient to target the protein to the nucleus, in contrast to the nonessential role of the N-terminal region ( Figure 5 ). By analysing cells transfected with plasmids encoding increasingly larger sections from the C-terminus of VFX10045, we observed that a truncated EGFP-VFX10045_1-534 retained the nuclear localization ( Figure 5A ) but smaller proteins were not localized in the nucleus (VFX10045_1-480, VFX10045_1-430, VFX10045_1-380 and VFX10045_1-331). This excluded the last ~50 residues of VFX10045 from a relevant role in its nuclear targeting and attributed a fundamental role to the region between residues 294 and 534. Analysis of the localization in transfected cells of additional truncated EGFP-VFX10045 proteins, showed that the region between residues 380 and 534 has an essential role in the nuclear tropism of VFX10045 ( Figure 5 ). In some of these experiments, to prevent unwanted diffusion to the nucleus because of a small protein size, we included in the constructs the N-terminal region (VFX10045_1-293; Figure 5A ), shown to be non-relevant for the nuclear targeting (see above).

To help pinpointing the residues enabling nuclear targeting of VFX10045, we aimed to analyze the localization in transfected CHO cells of chimeras between VFX10045 and its 94% identical homolog Lpp3070, which is mostly cytosolic ( Figures S2, 3 and 4 ). In initial experiments with truncated versions of Lpp3070 fused to EGFP we observed, as expected, that EGFP- Lpp3070_1-534 does not localize to the nucleus but, surprisingly, that EGFP-Lpp3070_380-586 or EGFP-Lpp3070_380-534 concentrate in the nucleus ( Figure 5 , Figures S5 and S6 ). This suggests that the C-terminal region of Lpp3070 contains a cryptic nuclear targeting signal that is nonfunctional in the context of full-length Lpp3070. As regions between residues 380 and 584 of VFX10045 and Lpp3070 are 85% identical, the capacity of VFX10045 to migrate to the nucleus must be related to discrete differences in the nucleotide sequence of VFX10045. When we analyzed the localization of chimeras VFX10045-Lpp3070, this revealed that VFX10045_1-380-Lpp3070_381-586 did not localize to the nucleus, whereas Lpp3070_1-380-VFX10045_381-583 did ( Figures 5 , S5 and S6 ). This further confirmed that the C-terminal region of VFX10045 (from residue 381) can drive nuclear transport. Analysing the localization of additional chimeras with a shorter C-terminal of VFX10045 and longer portions of Lpp3070 (Lpp3070_1-410-VFX_411-583, Lpp3070_1-430-VFX_431-583 and Lpp3070_1-480-VFX_481-583) showed the relevance of residues 430 to 480 from VFX10045 in nuclear targeting ( Figure 5 ). A comparison of the primary structure of the two proteins in this region ( Figure S4 ) revealed two pairs of amino acids with non-conserved substitutions: R440 and I441 in VFX10045, corresponding to S443 and V444 in Lpp3070; and T463 and S464, corresponding to S466 and N467 in Lpp3070. Therefore, we performed additional site-directed mutagenesis on the plasmid encoding EGFP-VFX10045 generating vectors encoding EGFP-VFX10045_R440S+I441V and EGFP-VFX_T463S+S464N mutant proteins. While EGFP-VFX_T463S+S464N localized in the nucleus, the exchanges R440S and I440V abrogated transport of the mutant EGFP-VFX10045 protein to the nucleus, also observed in the corresponding 4HA tagged mutant during infection ( Figure 5B and Figure 4 ). Subsequent independent replacements R440A and I441A confirmed and pinpointed the importance of both residues, particularly of R440, in the nuclear targeting of VFX10045 ( Figure 5 and Figure S5 ).