Components of the endocytic and recycling trafficking pathways interfere with the integrity of the Legionella‐containing vacuole

Abstract Legionella pneumophila requires the Dot/Icm translocation system to replicate in a vacuolar compartment within host cells. Strains lacking the translocated substrate SdhA form a permeable vacuole during residence in the host cell, exposing bacteria to the host cytoplasm. In primary macrophages, mutants are defective for intracellular growth, with a pyroptotic cell death response mounted due to bacterial exposure to the cytosol. To understand how SdhA maintains vacuole integrity during intracellular growth, we performed high‐throughput RNAi screens against host membrane trafficking genes to identify factors that antagonise vacuole integrity in the absence of SdhA. Depletion of host proteins involved in endocytic uptake and recycling resulted in enhanced intracellular growth and lower levels of permeable vacuoles surrounding the ΔsdhA mutant. Of interest were three different Rab GTPases involved in these processes: Rab11b, Rab8b and Rab5 isoforms, that when depleted resulted in enhanced vacuole integrity surrounding the sdhA mutant. Proteins regulated by these Rabs are responsible for interfering with proper vacuole membrane maintenance, as depletion of the downstream effectors EEA1, Rab11FIP1, or VAMP3 rescued vacuole integrity and intracellular growth of the sdhA mutant. To test the model that specific vesicular components associated with these effectors could act to destabilise the replication vacuole, EEA1 and Rab11FIP1 showed increased density about the sdhA mutant vacuole compared with the wild type (WT) vacuole. Depletion of Rab5 isoforms or Rab11b reduced this aberrant redistribution. These findings are consistent with SdhA interfering with both endocytic and recycling membrane trafficking events that act to destabilise vacuole integrity during infection.

macrophages with its most severe manifestation being Legionnaires' disease pneumonia (Horwitz & Maxfield, 1984;Horwitz & Silverstein, 1980). In both amoebae and macrophages, the ability of L. pneumophila to replicate intracellularly is dependent on the Icm/Dot type IV secretion system (T4SS) (Segal, Feldman, & Zusman, 2005). This complex introduces more than 300 translocated substrates into the host cell cytosol, which promote establishment of the Legionella-containing vacuole (LCV) and exert a variety of regulatory controls on the host cell (Asrat, de Jesus, Hempstead, Ramabhadran, & Isberg, 2014;Burstein et al., 2016).
Establishment and maintenance of LCV membrane integrity are essential for protection of L. pneumophila from innate immune cytosolic sensing (Creasey & Isberg, 2014;Liu et al., 2018). The inability of a bacterial mutant to properly form the LCV or maintain its integrity results in a severe intracellular growth defect and pathogen clearance (Creasey & Isberg, 2012;Wiater, Dunn, Maxfield, & Shuman, 1998).
The identification of the Icm/Dot-translocated substrate SdhA has greatly contributed to our understanding of the consequences of failure to properly maintain pathogen replication vacuoles (Laguna, Creasey, Li, Valtz, & Isberg, 2006). In the absence of SdhA, the LCV becomes disrupted and L. pneumophila is detected in the cytoplasm of the infected macrophage (Creasey & Isberg, 2012;Laguna et al., 2006). Bacteria exposed in this fashion to the cytosol are susceptible to recruitment of the interferon-regulated guanylate-binding proteins (GBP) family of anti-microbial proteins, resulting in leakage of lipopolysaccharide (LPS) and activation of caspase-11 (Creasey & Isberg, 2012;Liu et al., 2018;Piro et al., 2017). As a consequence, gasdermin D is activated and pyroptotic cell death ensues (Aachoui et al., 2013;Pilla et al., 2014;Shi et al., 2015).
We previously demonstrated that mutations that removed the bacterial phospholipase PlaA rescued the defect in vacuole integrity observed with the ΔsdhA mutant, resulting in increased numbers of intracellular bacteria, thus demonstrating that PlaA promotes vacuole disruption (Creasey & Isberg, 2012). PlaA bears homology to the translocated phospholipase SseJ from Salmonella typhimurium, which is also involved in vacuole disruption in the absence of another Salmonella translocated substrate, SifA (Akoh, Lee, Liaw, Huang, & Shaw, 2004;Flieger, Neumeister, & Cianciotto, 2002;Ruiz-Albert et al., 2002). During infection, SifA binds host factor SKIP and sequesters Rab9 to prevent delivery of M6PR cargo to the vacuole (McGourty et al., 2012). The resemblance between SdhA/PlaA and SifA/SseJ suggests that SdhA may regulate host membrane trafficking events to maintain LCV integrity.
In addition to Salmonella, other pathogens have been reported to interface with host membrane trafficking pathway as a strategy for maintaining vacuolar integrity. The Chlamydia protein IncD was shown to recruit the host CERT-VAP complex to the inclusion and enable acquisition of host lipids that are essential for inclusion membrane stability (Derre, Swiss, & Agaisse, 2011). Recently, the Shigella vacuole has been reported to interact with endocytic Rab5-positive vesicles and recycling Rab11-positive vesicles to facilitate vacuole membrane rupture, indicating that vacuole integrity can be disrupted by interfacing with appropriate cellular compartments (Mellouk et al., 2014;Weiner et al., 2016). In this study, we determine if the disruption of LCV integrity that results from loss of L. pneumophila SdhA function can be attributed to a subset of host membrane trafficking pathways. Using high-throughput screening strategies, we identified specific endocytic and recycling Rab GTPase isoforms and their downstream effectors as playing a role in vacuole disruption resulting from challenge with the ΔsdhA strain. The similarity between Legionella and Shigella interactions with a common endocytic-recycling pathway provides an example of how a host process can either promote or interfere with pathogen replication, depending on the strategy used for intracellular growth.
2 | RESULTS 2.1 | Identification of host proteins that contribute to disruption of L. pneumophila ΔsdhA intracellular growth To identify host membrane trafficking pathways that interfere with growth of the L. pneumophila ΔsdhA mutant, we carried out a screen using an siRNA library targeting genes known to participate in membrane trafficking and remodelling. RAW 264.7 macrophages are known to be defective in the activation of pyroptosis and clearance in response to cytoplasmic exposure of L. pneumophila pattern recognition molecules, so growth of the mutant could be followed without premature host cell death downstream of cytoplasmic bacterial exposure (Pelegrin, Barroso-Gutierrez, & Surprenant, 2008). This allowed subtle differences in intracellular growth to be detected. RAW 264.7 macrophages were seeded in 96-well plates and transfected with an arrayed siRNA library against 112 mouse genes involved in membrane trafficking (Tables S1 and S2). Each well contained 4-pooled siRNAs per gene target and each plate contained six non-targeting siRNA control wells. After 48 hr of transfection, cells were challenged with a ΔsdhA Lux + strain and luminescence was measured as a proxy for intracellular growth, performing the assay in at least triplicate samples for each gene target ( Figure 1a). Candidates were identified by comparing the luminescence from each siRNA-treated well to the average luminescence of the non-targeting siRNA controls. The median absolute deviation score (ZMAD) was then determined for each gene target. We focused our analysis on early time points 12hpi and 24hpi because siRNA knockdown effects diminish after 72 hr of transfection, to identify candidates that stimulate ΔsdhA intracellular growth after depletion.
A total of 20 genes were identified as hits in the siRNA screen based on increased intracellular growth relative to the non-targeting siRNA controls with a cut-off of ZMAD ≥1.5. Candidate genes spanned functional roles in endocytosis, endocytic recycling and exocytosis ( Figure 1c) 2.2 | Identification of shRNAs that partially rescue ΔsdhA vacuole integrity The assay for increased bacterial yields allowed identification of knockdown candidates that could improve intracellular growth of the ΔsdhA mutant, but they may not increase the frequency of intact F I G U R E 1 Legend on next page.
vacuoles. To directly test if improved growth was due to increased vacuole integrity, we selected hits from the siRNA screen and additional related targets for analysis in a secondary screen using a direct assay for vacuole integrity in bone marrow-derived macrophages (BMDM) from the mouse. The secondary screen took advantage of a fluorescence readout we previously described to detect disrupted ΔsdhA vacuoles in knockdown macrophages combined with shRNA knockdown in these primary cells (Creasey & Isberg, 2012). Terminally differentiated BMDMs were used to facilitate microscopic readout, and to avoid the complication of having a portion of the cells in S phase, which results in a high proportion of unstable vacuoles (de Jesus-Diaz, Murphy, Sol, Dorer, & Isberg, 2017). This strategy allowed subtle differences in vacuole integrity to be detected.
BMDMs were transduced with a lentiviral shRNA library ( Figure 1b) targeting 23 membrane trafficking genes arrayed as four individual shRNAs assayed separately (Table S2). Lentivirus encoding sh-LacZ was used as a control. After allowing differentiation and 6 days knockdown, BMDMs were challenged in triplicate with the ΔsdhA strain for 6 hr, fixed and probed for permeable vacuoles using the antibody accessibility assay (Creasey & Isberg, 2012). After capturing multiple images per well and subjecting them to image analysis to quantitate permeable vacuoles, shRNAs that increased the integrity of the ΔsdhA vacuole relative to the shRNA-LacZ negative control were identified ( Figure 1d).
In this fashion, eight shRNAs were identified that significantly reduced the number of permeable ΔsdhA vacuoles detected in macrophages (Table 1, t test, p = .0048-.0346). The candidate genes included five members of the Rab family of GTPases that are associated with endocytic recycling, as well as ROCK1, and SYT1. Of this set, Rab5b, Rab11b and Rab8b were of particular interest due to their robust phenotype and their tight association with recycling, based on STRING analysis (Franceschini et al., 2013). Strikingly, two hits from the screen were shRNAs directed against Rab11b (Table 1), so this result was pursued further. Using the same screening procedures as described above, we performed a lentiviral shRNA library screen against gene targets  (Table S3). ShRNAs against SH3BP5, ZFYVE27 and RAB11FIP1 significantly reduced the number of disrupted ΔsdhA vacuoles detected in macrophages (Table 1; p < .0041 for RAB11FIP1). The gene target Rab11FIP1 was of particular interest, since this downstream Rab11 effector is known to F I G U R E 1 RNAi screens identify membrane trafficking proteins that antagonise L. pneumophila ΔsdhA intracellular growth and vacuole integrity. (a) Identification of siRNA that enhance intracellular growth. RAW 264.7 macrophages were seeded in 96-well plates and transfected the following day with a siRNA library directed against transcripts encoding membrane trafficking proteins. Transfected cells were challenged with L. pneumophila ΔsdhA Lux + and luminescence was measured at 12 and 24 hr post infection (hpi). (b) Identification of shRNAs that decrease vacuole permeability in primary macrophages. A/J bone marrow-derived progenitors were seeded in 96-well plates. Cells were transduced with shRNAs targeting genes identified in the siRNA screen (a). Terminally differentiated BMDM transductants were challenged with ΔsdhA for 6 hr. Plates were fixed and stained for cytosol-detected L.p. and total L.p.(c) Results of high-throughput siRNA screen for enhanced intracellular growth of L. pneumophila ΔsdhA Lux + . Experimental design, Figure 1a, performing luminescence readings at 12 or 24 hpi, with least three replicate measurements for each gene target being performed Luminescence from each experimental well was normalised to the average luminescence of the non-targeting siRNA-treated control wells of the same plate. Normalised ZMAD values are shown colour-coded for each siRNA of the library. A cut-off of ZMAD ≥1.5 was used to select potential candidates. (d) Candidates identified in Figure 2a and additional targets not covered in the original screen were tested in a high-throughput shRNA screen, performing assay detailed in Figure 1b, detecting cytosolic exposure of bacteria. Three replicates for each knockdown condition were performed, capturing 16 images/well with automated microscopy. Image capture, analysis and normalisation of data were performed as described (Experimental Procedures). (e) Depletion of transcripts for a subset of Rab11 effectors results in increased vacuole integrity during L. pneumophila ΔsdhA intracellular growth. Assay conditions as in Figure 2b. Statistical analyses were performed on normalised data by unpaired t test (*p < .05) (Experimental Procedures) T A B L E 1 Hits from shRNA lentiviral screens  [Luo & Isberg, 2004]). (c) Depletion of downstream Rab5 effector EEA1 rescues ΔsdhA vacuole integrity and growth defects. Nucleofection, vacuole integrity and bacterial yields determined as in panels (a and b). Statistical analyses were performed on normalised data by unpaired t test (*<.05; **<.01; ***<.001; Experimental Procedures) knockdown efficiency was determined by immunoblot ( Figure 2). Individual or group depletion of Rab5a, Rab5b, or Rab5c resulted in enhanced vacuole integrity in BMDM challenged with L. pneumophila ΔsdhA ( Figure 2a). Furthermore, each individual knockdown resulted in enhanced intracellular growth of the ΔsdhA strain. In particular, initiation of intracellular replication was clearly revived in a subset of cells treated with the siRNA, although it was clear that growth restoration was far from complete ( Figure 2b). This result was consistent with our previous experiments showing that increasing vacuole integrity in the absence of SdhA function only allows partial restoration of intracellular growth (Creasey & Isberg, 2012).
We next tested the effects on BMDMs that were treated with siRNA against EEA1, a Rab5 effector involved in promoting early endosome vesicle tethering, to determine if depletion of a well- Legend on next page.  LCVs, arguing that the effects are highly isoform-specific (Figure 4b).
Similar to the results with Rab5 and other strategies that increase the integrity of the ΔsdhA mutant-containing LCV, the stimulation of growth was relatively limited (Creasey & Isberg, 2012).
From the secondary shRNA screen in BMDMs, we found that Based on these findings, we expect that membrane material driven by these GTPases will be closely associated with the ΔsdhA LCV, contributing to vacuole disruption. In the presence of SdhA function, in contrast, access to these compartments should be limited. To this end, the localisation of the Rab11 effector FIP1 was analysed, using the same assay employed to detect EEA1 localisation (Figure 3). BMDM challenged for 4 hr with either WT or ΔsdhA strain was fixed and probed for vacuole permeability, and the association of Rab11FIP1 with the LCV was determined

| DISCUSSION
In this study, screens were performed that identified host membrane trafficking factors responsible for causing vacuole disruption and interfering with intracellular growth of an L. pneumophila ΔsdhA strain. We initially used a luciferase reporter assay to positively select for siRNAdepleted cells that showed increased intracellular growth of the mutant strain. Overall, 20 candidate genes were identified after screening 112 genes. To demonstrate specificity for knockdowns that result in increased LCV integrity relative to that observed in BMDMs, we turned to high-throughput microscopy analysis to identify factors specifically involved in antagonising vacuole integrity of the L. pneumophila ΔsdhA mutant. From a primary shRNA screen introduced into BMDMs by lentivirus, we identified eight candidate genes, with the majority being Rab GTPases involved in endocytic and membrane recycling. Genes of particular interest were Rab5, Rab8 and Rab11. Strikingly, Rab11a had been previously shown to regulate vacuole rupture as an early event that supports intracellular growth of Shigella after uptake into cultured cells (Mellouk et al., 2014;Weiner et al., 2016). This raises the possibility that trafficking events that support intracellular growth of a cytosolic pathogen (S. flexneri) interfere with the biogenesis of the membrane surrounding an intravacuolar organism (L. pneumophila).
Validation of screen hits revealed that rescue of ΔsdhA vacuole integrity and growth was specific to depletion of individual isoforms.
One exception to this observation was silencing Rab5a, Rab5b or Rab5c, in which we found that individual depletion of each isoform was sufficient to reduce vacuole disruption (Figure 2a). This phenotype is consistent with previous studies demonstrating that the three isoforms localise to the same endocytic compartment and cooperatively regulate endocytic events (Bucci et al., 1995). Depletion of all three isoforms rescued the ΔsdhA phenotype to a similar degree as the depletion of individual isoforms, indicating that the isoforms function collectively rather than separately during trafficking.
In contrast to the situation with the Rab5 isoforms, specific silencing of Rab11b, but not Rab11a, enhanced the integrity of the ΔsdhA vacuole ( Figure 4a). Furthermore, knockdown of both Rab11 isoforms produced a similar phenotype as individual knockdown of Rab11b, confirming that only Rab11b is required for ΔsdhA vacuole disruption. It has been shown in other cell types that these two isoforms can localise to distinct compartments within the cell, consistent with a division of function between the two isoforms that is linked to their association with different compartments (Kelly, Horgan, & McCaffrey, 2012;Lapierre et al., 2003). Of particular note in this regard is that the Rab11a isoform specifically supports growth of S. flexneri. The isoform differences between the two pathogens may reflect spatial and temporal differences in the biogenesis and degradation of the bacteriumcontaining compartments. After uptake into host cells, depletion of  (Figures 2c and 4c). We also found that depletion of VAMP3, a downstream regulator fusion event controlled by both Rab8b and Rab11b, could result in an increased number of intact vacuoles harbouring the ΔsdhA strain (Figure 5c) (Banerjee et al., 2017;Finetti et al., 2015;Wilcke et al., 2000). VAMP3 is SNARE protein that drives membrane fusion, thus pointing to a model in which disruption of membrane integrity is a consequence of fusion with a compartment that destabilises the LCV (Hu, Hardee, & Minnear, 2007 (Christoforidis et al., 1999;Lindsay & McCaffrey, 2004). Therefore, the nature of the lipid components could define different compartments, or these proteins could define tethering (EEA1), docking (Rab11Fip1) and fusion (Vamp3) events. Consistent with either model, we found that EEA1 and Rab11FIP1 localised to the ΔsdhA-containing vacuole with was a higher density compared to the WT-containing vacuole, indicating that they may play a direct role in destabilising the vacuole harbouring the mutant (Figures 3 and 6). One of the striking properties of the LCV is that it appears to be walled off from the host early endosomal trafficking system . The presence of EEA1 associated with events leading to vacuole rupture indicates that constructing of such a firewall is essential for preserving vacuole integrity. and it is possible that this cargo could be delivered to the ΔsdhA vacuole resulting in oxidative destabilisation (Casbon, Allen, Dunn, & Dinauer, 2009). It is interesting to note that vacuole degradation is an essential step necessary to release S. flexneri into the cytosol and initiate intracellular replication. These events appear to require recruitment of Rab11A-containing vesicles that could carry destabilising cargo (Mellouk et al., 2014;Weiner et al., 2016). Therefore, an event that for one pathogen is toxic, for another is an essential step in the biogenesis process leading to intracellular growth.
Altogether, this study describes a genetic screen that identified a specific set of host factors that are required for ΔsdhA vacuole disruption. Future studies should elucidate the exact mechanism by which these proteins facilitate vacuole disruption and the nature of the cargo being delivered to the vacuole that destabilises the vacuole.

| Bacterial culture and media
All L. pneumophila strains used in this study are derived from the Philadelphia 1 isolate and are described in Table 2 (Berger & Isberg, 1993).

| Secondary shRNA library screen
Selected hits from the primary siRNA library screen and additional related targets were tested in a screen using shRNA constructions intro-
The next day, cells were lifted in cold PBS (Gibco) and resuspended in RPMI medium containing FBS (Swanson & Isberg, 1995   I.S.A., W.Y.C. and R.R.I. wrote the paper.