S. Typhimurium SR-11 x3181 carrying an mCherry-expressing plasmid (Hubber et al., 2014) was used for infection throughout the study. Legionella strains used in this study were derivatives of L. pneumophila strains Philadelphia-1 (Lp01) (Berger and Isberg, 1993). Salmonella was grown in Luria-Bertani (LB) medium containing 20 μg/ml of chloramphenicol to maintain the mCherry expressing plasmid. Legionella strains were grown in liquid N-(2-acetamido)-2-aminoethanesulfonic acid (ACES) buffered yeast extract (AYE) media (Horwitz and Silverstein, 1983). The ravZ and dotA deletion strains were constructed by allelic exchange as described previously (Zuckman et al., 1999). Eukaryotic expression plasmids for RavZ were constructed by cloning the PCR-amplified wild-type or mutant (C258A) ravZ into the p3xFLAG-CMV-10 expression vector (Sigma). Site-directed mutagenesis was conducted using the Quickchange II Site-Directed Mutagenesis Kit according to the manufacturer's recommendations (Agilent Technologies).

HeLa-FcγRII and HEK293T-FcγRII cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum. Bone marrow derived macrophages (BMDMs) were prepared from A/J mice (Japan SLC, Inc.). The cells were plated onto cover glass in 24-well tissue culture plates 1 day before infection or transfection. Transfection was conducted using Lipofectamine 2000 (Invitrogen) or Fugene HD (Promega) according to the manufacturers' recommendations.

Mouse antibody against mono- and polyubiquitinylated conjugates (FK2) was purchased from Enzo (BML-PW8810). Mouse antibody against ubiquitin (P4D1) was purchased from Cell Signaling (#3936). Mouse antibody against α-tubulin was purchased from Sigma (T6074). Rabbit anti-Legionella antibody was purchased from Biodesign (B65051G). Secondary antibodies were purchased; Peroxidase-conjugated goat anti-mouse antibody (62–6520, Invitrogen), Alexa Fluor 488-conjugated goat anti-mouse antibody (A-11029, Thermo Fisher Scientific) and Rhodamine RedX-conjugated goat anti-rabbit antibody (R6394, Thermo Fisher Scientific).

Overnight Salmonella culture was diluted 1:30 with fresh LB medium containing 20 μg/ml of Chloramphenicol, grown to an OD₆₀₀ value of about 2.0, and used for infection. Overnight liquid culture of Legionella was used for infection. The bacterial cultures were spun down by microfuge, the culture media were replaced with phosphate-buffered saline (PBS), and the suspension was immediately used for infection. HeLa cells were infected with Salmonella with multiplicity of infection (MOI) 300 for 1 h. After 20 min of infection, the cell culture medium was replaced with fresh medium containing 50 μg/ml of gentamycin to kill extracellular Salmonella. For the co-infection, a pre-mixed solution of Salmonella and Legionella was prepared and immediately used for infection with MOI 300 of both bacteria for 1 h. The time course study of Legionella infection in BMDMs was conducted with MOI 50. After 1 h of infection, the cells were washed three times with PBS to remove extracellular Legionella and the culture medium was replaced with fresh medium.

Infected cells were fixed with 4% (w/v) of paraformaldehyde (PFA) after three washes with PBS and permeabilized with chilled methanol or with 0.1% (w/v) of Triton X-100 in PBS. After blocking with 2% (w/v) of goat serum in PBS for 20 min, the cells on the coverslips were reacted with a primary antibody for 1 h, followed by a secondary antibody for 30 min, at concentrations recommended by the manufacturers.

BL21(DE3) carrying pET15b-ravZ (pNH1678) or pET15b-ravZ_C258A (pNH1679) was grown to logarithmic phase at 37°C in 1 l of L-broth. After addition of IPTG to a final concentration of 0.4 mM to induce the production of His-RavZ, cells were cultured overnight at 16°C. E. coli cells were recovered by centrifugation and suspended in buffer A (20 mM Tris HCl pH 7.5, 5 mM EDTA) containing complete protease inhibitor cocktail (Roche). After treatment with 250 μg/ml lysozyme for 30 min, cells were disrupted by sonication. Lysate was centrifuged (20,000 g, 20 min) to remove insoluble materials and applied to Q-sepharose Fast Flow (~25 ml bed volume; GE Healthcase) equilibrated with buffer B (20 mM Tris HCl pH 7.5, 10 mM β-mercaptoethanol) in an open column format. After washing with buffer B containing 200 mM NaCl, the fraction containing His-RavZ was eluted with buffer B containing 300 mM NaCl. This fraction was further applied to a HisSelect column (1 ml bed volume; Sigma-Aldridge) equilibrated with buffer C (20 mM Tris HCl pH 7.5, 300 mM NaCl, 10 mM β-mercaptoethanol). After washing with buffer C containing 10 mM imidazole, fractions were eluted with a linear gradient of 10–100 mM imidazole in buffer C. Fractions containing His-RavZ were pooled and concentrated and subjected to a size exclusion column (Superdex 200 16/60 pg; GE healthcare). His-RavZ was eluted in 20 mM Tris HCl pH 7.5, 300 mM NaCl, 1 mM DTT.

After 1 h of infection with Salmonella, HeLa cells were washed twice with semi-permeabilization buffer [125 mM K(OAc), 2.5 mM Mg(OAc)₂, 25 mM Hepes-KOH pH 7.4, 1 mg/ml Glucose, 1 mM DTT]. The cells were treated with 30 μg/ml digitonin in the semi-permeabilization buffer for 4 min at room temperature and washed another three times with the buffer. The permeabilized cells were treated with purified His-tagged wild-type or mutant RavZ proteins with gentle agitation for 1 h at room temperature. After two washes, the cells were fixed with 4% (w/v) PFA and subjected to immunostaining.

Immunofluorescence experiments were conducted with at least 100 bacterial vacuoles counted per single experiment. Values were compared using paired Student's t-tests on three independent experiments.

All animal experiments were performed in accordance with the institutional guidelines and were approved by the Animal Care and Use Committee of the Research Institute for Microbial Diseases, Osaka University, Japan (Biken-AP-H26-10-0).

As previously described, Legionella RavZ is sufficient to inhibit macroautophagy in mammalian cells, but it is not the only determinant preventing LC3 recruitment to LCVs (Choy et al., 2012). In an attempt to identify L. pneumophila protein factors responsible for the prevention of LC3 recruitment, we used an S. Typhimurium and L. pneumophila co-infection system. Selective autophagy targets a fraction (~25%) of Salmonella-containing vacuoles (SCVs) at 1 h post-infection, as evidenced by the LC3-positive SCVs observed in GFP-LC3 expressing HeLa cells challenged with S. Typhimurium (Figure 1 SR-11). When HeLa cells were co-infected with wild-type L. pneumophila, LC3 recruitment to SCVs was reduced (Figure 1 SR-11+Lp01). In contrast, in HeLa cells co-infected with L. pneumophila strains lacking functional Dot/Icm T4SS or RavZ, LC3 recruitment to SCVs was not affected (Figure 1 SR-11+Lp01 ΔdotA/ΔravZ). These results illustrate that the co-infection experiment can be used to identify a L. pneumophila effector responsible for the restriction of LC3 recruitment to bacteria-containing vacuoles. These results also demonstrate that the restriction of LC3 recruitment to SCVs in the experimental condition can be explained solely by the function of the L. pneumophila effector RavZ, and any other effectors play only a limited role.

Adaptor proteins such as p62 and NDP52 carry both LC3- and ubiquitin-binding domains and are proposed to play a critical role in the recognition of invading microorganisms and the induction of selective autophagy. We thus examined the effect of co-infecting L. pneumophila on the recruitment of these adaptor proteins to SCVs. A fraction of SCVs acquired GFP-p62 or GFP-NDP52 in HeLa cells producing GFP-p62 or GFP-NDP52 respectively at 1 h post-infection (Figure 2 SR-11). Intriguingly, the recruitment to SCVs of both GFP-p62 and GFP-NDP52 was severely restricted by co-infecting L. pneumophila, as in the case of LC3 (Figures 2A,B,C SR-11+Lp01). The restriction was dependent on the presence of functional Dot/Icm T4SS in the co-infecting L. pneumophila (Figures 2A,B,C SR-11+Lp01 ΔdotA). RavZ disruption restored the GFP-p62 and GFP-NDP52 recruitment to extents similar to or somewhat less than those observed in HeLa cells infected solely with S. Typhimurium (Figures 2A,B,C SR-11+ΔravZ). We also examined the recruitment to SCVs of Tecpr1-GFP, another autophagy adaptor protein that does not possess ubiquitin-binding activity. In contrast to GFP-p62 and GFP-NDP52, Tecpr1-GFP recruitment to SCVs was not affected by co-infecting L. pneumophila (Figures 2A,D). These results prompted us to examine ubiquitin recruitment to SCVs, an upstream event before adaptor recruitment, in the co-infection system. Using immunofluorescence microscopy with anti-conjugated-ubiquitin antibody (FK2), we found that the fraction of ubiquitin-positive SCVs was significantly reduced in the presence of co-infecting L. pneumophila, (Figures 3A,B SR-11+Lp01). As in the case of LC3, p62, and NDP52, the restriction required a functional Dot/Icm T4SS and the effector RavZ of co-infecting L. pneumophila (Figures 3A,B SR-11+Lp01 ΔdotA/ΔravZ).

The L. pneumophila effector RavZ appears to be necessary for the reduction of ubiquitin-positive SCVs in the co-infection condition. To test whether RavZ was sufficient for this reduction, we examined ubiquitin recruitment to SCVs in HeLa cells ectopically expressing RavZ (Figure 4A). The fraction of ubiquitin-positive SCVs in 3xFLAG-RavZ producing HeLa cells was lower (~5%) than that in vector-transfected cells (~30%), indicating that RavZ is sufficient for the restriction. RavZ deconjugates LC3, and the RavZ_C258A mutation disrupts this enzymatic activity. Ectopic expression of 3xFLAG-RavZ_C258A did not affect ubiquitin recruitment to SCVs, demonstrating that the same enzymatic activity is required for the restriction of ubiquitin recruitment to SCVs (Figure 4A).

To further characterize the role of RavZ in decreasing ubiquitin-positive SCVs, we used semi-permeabilized Salmonella-infected HeLa cells. The semi-permeabilized cells were washed to remove cytoplasmic proteins and compounds, and purified His-RavZ or His-RavZ_C258A was added back at various concentrations. LC3 and ubiquitin levels on SCVs were determined by immunofluorescence staining using specific antibodies. Treatment with His-RavZ at a saturating concentration of 25 μg/ml (430 μM) decreased LC3 detectable on SCVs (Figure 4B, Figure S1A). In the same condition, the fraction of ubiquitin-positive SCVs was reduced two-fold (Figure 4C). The added His-RavZ significantly reduced ubiquitin and LC3 levels on SCVs at a concentration as low as 0.008 μg/ml (140 nM), and the reactions were readily saturated as His-RavZ concentration increased (Figure S1), indicating the RavZ activity toward ubiquitin had an effective concentration comparable to that of its activity toward LC3. There was a residual fraction of ubiquitin-positive SCVs even at the saturated concentration of RavZ, suggesting that a certain fraction or species of ubiquitin cannot be removed by the action of RavZ (Figure S1B). Collectively, these results demonstrated that the activity of RavZ is necessary and sufficient for the reduction of ubiquitin levels on SCVs, presumably by enzymatically removing ubiquitin from unknown substrates on Salmonella or SCVs.

We examined whether RavZ inhibits ubiquitin recruitment to LCVs in L. pneumophila-infected cells. In HeLa cells infected with a L. pneumophila ΔravZ strain, the frequency of ubiquitin-positive LCVs was not significantly affected by ectopically expressed 3xFLAG-RavZ or its catalytic mutant (Figure 4D). In mouse bone marrow-derived macrophages (BMDMs), infection with L. pneumophila resulted in the appearance of LCVs highly decorated with ubiquitin in a T4SS-dependent manner (Figure 5). However, up to 10 h post-infection, there was no significant difference in the fractions of ubiquitin-positive LCVs containing wild-type and ΔravZ strains. It has been shown that ubiquitin recruitment to LCVs requires bacterial protein synthesis (Dorer et al., 2006). To examine the possibility that the persistent ubiquitin recruitment to LCVs obscures RavZ-dependent ubiquitin reduction, we repeated this experiment in the presence of chloramphenicol, a bacterial translation inhibitor, in order to restrict de novo supply of ubiquitin to LCVs (Figure S2). The fraction of ubiquitin-positive LCVs was strongly reduced in the presence of chloramphenicol, while again there was no significant difference in the fractions of ubiquitin-positive LCVs containing wild-type and ΔravZ strains. These results do not support a potential contribution of RavZ to removing ubiquitin from LCVs.

Autophagy adaptor proteins p62 and NDP52 can bind both to ubiquitin and to LC3. RavZ possesses the enzymatic activity to irreversibly deconjugate LC3 from PE. Therefore, one of the possible explanations for the reduction of ubiquitin-positive SCVs in the presence of functional RavZ is that RavZ-dependent deconjugation of LC3 leads to destabilization or dissociation of ubiquitinated complexes containing adaptor proteins in the vicinity of SCVs. Alternatively, the apparent decrease in ubiquitin could be attributed to loss of ubiquitinated LC3 from SCVs. These feedback scenarios predict that the reduction of LC3-positive SCVs is a prerequisite for the reduction of ubiquitin-positive SCVs. To test this possibility, we examined the ubiquitin levels on SCVs in Atg7 knockout (KO) MEF cells, in which LC3 conjugation to PE does not take place (Komatsu et al., 2005) (Figure 6). We found that the RavZ-dependent reduction of ubiquitin-positive SCVs was observed both in wild-type and Atg7 KO MEF cells (Figure 6). This strongly suggests that the RavZ-dependent restriction of ubiquitin-positive SCVs does not result from the RavZ-mediated deconjugation of LC3 from PE.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.