Forward genetic screen for L.p-encoded TORC1-activating factors identified the T4bSS effector BinA as a putative regulator

One mechanism mediating bacteria-induced TORC1 activation in L.p infected macrophages is driven by the accumulation of intracellular amino acids caused by a L.p -induced blockade in host protein translation [48] (Fig 1A). Previous data also supports the idea of T4bSS effectors-mediated TORC1 activation [4,48], thus we carried out an immunofluorescence-based forward genetic screen to identify putative bacterial mTOR regulators using a cellular model of infection (Fig 1A-C). To this end, Myd88^-/- bone marrow-derived macrophages (BMDMs) were selected as host cell model because previous data demonstrates that under these conditions TORC1 activation (i) is confined to cells harboring bacteria and (ii) is triggered solely by L.p in a T4bSS-dependent manner [4]. Thus, putative bacterial inducers of mTOR signaling can be identified in this setting. Phosphorylation of the ribosomal S6 protein (rS6p) at Ser235/236 by the S6 protein kinase (S6K) is a canonical downstream readout for TORC1 activation, which was incorporated in the screen. In Legionella infection assays, phospho-rS6p abundance can be monitored via immunoblot analysis at the population level (Fig 1A) [4] and via immunofluorescence microscopy at the single cell level (Fig 1C). We screened an arrayed T4bSS effectors library of 177 single gene transposon mutants [52] in synchronized cellular infection assays lasting 8 hours, where each infection was carried out in a technical triplicate (Fig 1B). The p-rS6p signal in each infected cell was determined via immunofluorescence where ‘ON’ state was set at p-rS6p mean fluorescent signal intensity > 400 based on data from bystander uninfected cells as well as cells infected by a mutant strain lacking a functional T4bSS – L.p ∆dotA (Fig 1C) [4]. Because deletion of T4bSS effectors might produce intracellular colonization defects due to failure in LCV trafficking or remodeling, immunostaining with anti-Ubiquitin (FK2) antibody was also included in the analysis so that LCV defects can be measured as an experimental variable. Ubiquitin accumulation at the LCV coincides with initiation of LCV biogenesis and persists for the lifespan of the vacuole [53]. Representative BMDMs with distinct TORC1 signaling states (as defined for the screen) that harbor remodeled Ub + LCVs as well as uninfected bystander cells are shown in Fig 1C. For each set of experiments, infections with wild-type L.p were included as a baseline for the percentage of infected cells that trigger rS6p phosphorylation, which was determined to be 79.5 ± 9.7% (Fig 1C). Two z-scores were calculated for each mutant strain taking into account all infected cells or only BMDMs harboring Ub + LCVs, which allowed for identification of false positives due to intracellular colonization defects (Fig 1D). The top hit in both categories was a mutant strain with an insertion in the binA gene (lpg0393) (Fig 1D-E). BinA has been shown to be a bona fide substrate of the Legionella Dot/Icm T4bSS [54] that contains a eukaryotic VPS9 domain typically found in GEFs that activate Rab5 family of small GTPases [55]. BinA can interact with Rab5 as well as exhibits GEF activity in vitro towards Rab5, Rab21 and Rab22, where D41 and Y84 represent key catalytic residues in the VPS9 domain required for nucleotide exchange [56]. The role of BinA in Legionella infections remains to be elucidated.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Identification of Lp T4bSS effector regulating TORC1 signaling.

(A) TORC1 signaling induced by Lp results in phosphorylation of the host ribosomal S6 protein (rS6p). (A, left panel) Lp can activate TORC1 by suppressing host translation, which increases the intracellular amino acids concentration [4,48]. (A, right panel) Immunoblot (IB) analysis showing rS6p phosphorylation at Ser235/236 in cell lysates from Myd88 KO iBMDMs infected with Lp01 ∆flaA (MOI = 15) for the indicated time periods in the presence or absence of the MTOR inhibitor Torin2 or were left uninfected (UN). (B) Schematic depiction of the performed loss-of-function screen for TORC1 activators with an arrayed library of single-gene transposon mutants strains lacking individual T4bSS effectors. Myd88 KO BMDMs were seeded on glass cover slips and were infected for 8hrs, one strain at a time where each plate included Lp01 ∆flaA baseline infection (MOI = 15). Cover slips were immunostained for p-rS6p (S235/236) and ubiquitin followed by quantitative microscopy analysis of the percentage of infected p-rS6p+ macrophages. (C) Representative micrograph of Lp01 ∆flaA infection showing (*) an infected p-rS6p+ cell, (**) an infected p-rS6p− cell, and (***) an uninfected p-rS6p− cell. The average baseline percentage of p-rS6p+ cells in Lp01 ∆flaA infections was 79.5 ± 9.7%. (D) Data shows the distribution of the standard deviations away from the mean (z-score) for each mutant strain tested. Data is parsed out for macrophages housing remodeled Ub + LCVs as well as for all infected cells. The results for the top hit binA::Tm are highlighted in both categories. The complete dataset is provided in S1 Table. (E) Domain architecture of BinA is shown. Residues D41 and Y84 are critical for the GEF activity of the VPS9-like domain.

https://doi.org/10.1371/journal.ppat.1012998.g001

To investigate BinA as a putative TORC1 regulator we generated a binA clean deletion strain, which was subsequently complemented with a plasmid-borne 3xFLAG-tagged copy of binA under the control of an IPTG-inducible promoter (Fig 2A). The capacity of those strains to trigger rS6p phosphorylation in Myd88^-/- BMDMs was assessed at distinct time-points after infection (Fig 2B-D). To this end, the p-rS6p signal intensity was measured in synchronized infections at the single-cell level via immunofluorescence where time points were selected to capture TORC1 signaling at the time of initiation of bacterial replication (4hpi), after 1–2 doubling events (6hpi) or after several rounds of replication (10hpi) have taken place (Fig 2B-D). Macrophages harboring binA– bacteria across all time point produced significantly lower p-rS6p signal at the population level as compared to BMDMs infected by the wild-type or the complemented strains (Fig 2D). The p-rS6p signal in some cells infected by binA– bacteria was comparable to cells harboring binA+ strains, however in majority of the cell population the signal was lower indicating that either BinA is required for optimal TORC1 signaling or functional redundancy from other bacterial factors partially compensates for the loss of binA. The percentage of Ub + LCVs harboring binA– bacteria at 6hpi was comparable to LCVs containing binA+ bacteria (82.4 ± 7% vs 77.8 ± 9%) indicating that loss of binA did not produce LCV maturation or trafficking defects. Importantly, the induction of p-rS6p was restored in macrophages infected by the binA complemented strain under IPTG-inducing conditions to a level indistinguishable from the wild type parental strain (Fig 2B-D). Taken together, these data indicate a role for BinA as a putative bacterial TORC1 regulatory factor.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Deletion of BinA compromises TORC1 activation by Lp.

(A) Anti-Flag immunoblot analysis of various Lp strains expressing the indicated plasmid borne IPTG-inducible 3XFlag-BinA alleles. Ponceau S staining is used to control for lysate loading. (B) Representative micrographs of TORC1 activation in Myd88KO iBMDMs infected by Lp strains lacking or expressing BinA at 6hpi. Phosphorylation of rS6p was used as a readout for TORC1 signaling and is shown as pseudocolor in (C). (D) Quantitative microscopy of the p-rS6p signal from uninfected (UN) and infected Myd88KO iBMDMs. Log₁₀ transformed data of p-rS6p mean fluorescent intensity (MFI) is shown where each circle represents data from a single cell. The number for cells analyzed for each condition is noted and the mean for each condition is indicated with a red bar. Statistical analysis was performed with multiple comparisons Kruskal-Wallis one-way ANOVA test. The respective p- values across the different comparisons are shown. (B-D) One representative of three biological replicates is shown.

https://doi.org/10.1371/journal.ppat.1012998.g002

BinA promotes expansion of the Legionella-occupied organelle in an mTOR-dependent manner

One consequence of mTOR induction by Legionella is the stimulation of host lipogenesis which satisfies the need for membrane phospholipids as the LCV expands and the bacteria replicate within it. As a result, lipogenesis prevents premature rupture of the intracellular niche and extends the LCV lifespan [4,5]. Thus, we investigated whether binA deletion impacts the mTOR-dependent LCV homeostasis processes. To this end, we first generated a MTOR^−/− U937 human monocytic cell line using CRISPR/Cas methodology, which can be terminally differentiated into macrophages following phorbol ester treatment [57] (Fig 3A). To validate the loss of mTOR signaling in the knockout cell line, monocytes were stimulated with E.coli LPS (a canonical TORC1 inducer) and the amount of p-rS6p was measured via immunoblot analysis (Fig 3B). In this setting, LPS triggered rS6p phosphorylation in the parental U937 cells but not in the MTOR^−/− cells, which is consistent with a loss of mTORC1 activity in the knockout cell line (Fig 3B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. BinA promotes LCV expansion in a MTOR-dependent manner.

(A) Immunoblot analysis of MTOR abundance in U937 Mtor ^−/− monocytes and its respective parental WT cell line. (B) Loss of TORC1 signaling in U937 Mtor ^−/− monocytes stimulated with E.coli LPS (100 ng/ml) as indicated. (C-E) Representative micrograph time-series from live-cell imaging of different GFP + Lp strains showing LCV dynamics of vacuoles containing either Lp01 (C) or Lp01 ∆binA (D) within U937 macrophages (MFs) or vacuoles occupied by Lp01 inside U937 Mtor ^−/− MFs (E) (MOI = 2.5). Individual LCVs are circled by a solid line prior to and by a dashed line after rupture. Multiple LCVs in the same micrograph are indicated by different line color. (F) Population level analysis of LCV size during infection of U937 MFs that express or lack MTOR. The changes in the average LCV sizes ± StDev from three technical replicates are shown. Some conditions were supplemented with hiFBS (5% v/v) to complement the loss of MTOR-dependent metabolite production by the host cells. (G-H) Quantitative analysis of LCV dynamics at the level of single LCVs. For each condition, the parameters of at least 100 individually tracked LCVs were analyzed, where each data point is derived from a single vacuole. (G) The maximum size reached by individual LCVs under the indicated infection conditions are shown, where the bar indicates population mean value. (H) The graph shows the average ± SEM duration of the expansion phase of the pathogen-occupied vacuole for each infection condition. (F, G-H) Statistical analysis was completed via two-way (F) or one-way (G-H) ANOVA with Dunnett’s multiple comparison test using the Lp infection as baseline control. The respective p-values are indicated for each panel. (B-H) Data is from one representative of three biological replicates.

https://doi.org/10.1371/journal.ppat.1012998.g003

Next, U937 macrophages expressing or lacking mTOR were infected with different GFP+ Legionella strains and the lifecycle parameters of multiple LCVs from infected macrophages were measured via live-cell imaging (Fig 3C-E). In these infection assays, cells were imaged every 4hrs over several days, which allowed for determination of changes in the LCV size over time as well as the LCV lifespan. LCVs from U937 macrophages infected with L.p ∆binA GFP + are shown in Fig 3D, whereas L.p GFP+ infections of MTOR+ or MTOR– macrophages are shown in Fig 3C and 3E respectively. In mTOR sufficient macrophages, the LCV expansion rate of vacuoles harboring binA− or binA+ were identical as demonstrated by the slopes of the LCV growth curves (Fig 3F, left panel). However, the growth of LCVs containing binA− bacteria plateaued at a vacuole size that is significantly smaller as compared to LCVs harboring binA+ bacteria, indicative of premature organelle growth arrest (Fig 3F, left panel). Conversely, LCVs harboring binA− or binA+ bacteria terminated expansion at similar sizes in mTOR deficient macrophages (Fig 3F, right panel) indicating that the BinA-dependent regulation of LCV expansion requires the mTOR kinase. Supplementation of exogenous serum-derived lipids during infection has been shown to rescue the LCV biogenesis defects caused by the loss of mTOR-dependent de novo lipogenesis by directly providing lipids for membrane synthesis [4,5]. Similarly, the expansion of LCVs harboring binA− bacteria in both mTOR-sufficient and mTOR-deficient macrophages was restored upon hiFBS supplementation further supporting the notion of phenotypic link between the absence of BinA and mTOR activities (Fig 3F). Single vacuole analysis data showed that on average the maximum size achieved by individual bacteria-occupied organelles was significantly lower when either MTOR or binA were knocked out (Fig 3G) and organelle expansion terminated earlier as well (Fig 3H). Importantly, both of these LCV biogenesis parameters were indistinguishable between infections with binA− and binA+ bacteria in the absence of mTOR activity (Fig 3G-H).

Taken together these data indicate a mechanistic connection between BinA and mTOR signaling that phenotypically manifests specifically in premature termination of LCV expansion and an overall lower bacterial load per organelle that can be rescued by exogenous serum-derived lipids supplementation.

BinA sensitizes mTOR signaling in response to amino acid stimulation upstream of the heterodimeric Rag GTPase complex

To gain insight into the mechanism by which BinA regulates mTOR, we first tested whether BinA is sufficient to trigger TORC1 activation upon amino acids stimulation when produced by eukaryotic cells independent of infection. To this end, HEK293 cells ectopically producing GFP or GFP-tagged BinA were assayed for their capacity to trigger TORC1 signaling in a prototypical amino-acid withdrawal/re-stimulation assay where cells were cultured in amino acid-free media for 4 hours and subsequently were treated for 30min with various amounts of arginine (Fig 4A) or leucine (Fig 4B). As expected, treatment with either amino acid triggered phosphorylation of S6K (a TORC1 substrate) and its immediate downstream substrate rS6p (Fig 4A-B) indicative of TORC1 activation. Notably, expression of GFP-BinA but not GFP resulted in TORC1 activation at lower concentrations of Arg (10–50µM vs. 1000µM) (Fig 4A) and Leu (10 – 100µM vs. 200 – 400µM) (Fig 4B). In addition, BinA expression alone did not trigger phosphorylation of either S6K or rS6p (Fig 4A-B) in the absence of Arg/Leu indicating that the TORC1 pathway retains stimulus-dependency despite the BinA-dependent sensitization. Furthermore, kinetic analysis of TORC1 activation by different doses of Arg in the presence of BinA revealed increased p-rS6p signal at early time points following stimulation, which is consistent with an accelerated and a sensitized response (Fig 5A). Similar results were observed following Leu stimulation (Fig 5B). Based on these data we conclude that BinA directly or indirectly lowers the threshold for TORC1 activation upon stimulation with either Arg or Leu in the absence of other bacterial factors.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. BinA sensitizes TORC1 signaling downstream of arginine and leucine stimulation.

(A-B) Ectopic expression of BinA in eukaryotic cells is sufficient to sensitize TORC1 signaling induced by arginine or leucine. Data shows dose-dependent TORC1 activation by Arg (A) or Leu (B) at 30 min post-treatment (mpt) in amino-acid starvation/refeeding experiments using HEK293 cells transfected with either GFP or GFP-BinA. Band signal intensity in the phospho-immunoblots for S6K and rS6p for each condition was quantified and is presented below each Fig panel as fold change from untreated cells. The data shown is from one experiment out of at least 3 biological replicates.

https://doi.org/10.1371/journal.ppat.1012998.g004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Sensitization of TORC1 signaling by BinA is stimulus-dependent and requires the Rag GTPase complex.

Comparison of TORC1 signaling kinetics in response to stimulation with different doses of arginine (A), leucine (B), glutamine (C) or methionine (D) in HEK293 cells that ectopically produce either GFP or GFP-BinA. (E-F) Ectopic expression of the GDP-locked RagA T21N allele inhibits arginine-induced TORC1 signaling in HEK293 cells lacking (E) or producing GFP-BinA (F) in amino-acid starvation/refeeding experiments. Treatments with either 1mM or 250 µM Arg are shown respectively in (E) and (F). (G) The GTP-locked RagA Q66L allele augments TORC1 signaling in BinA-expressing HEK293 treated with low-dose 100 µM Arg for the indicated time periods in a starvation/refeeding experiment. (A-G) Immunoblot data from amino acid starvation/refeeding experiments are shown, where band signal intensity in the phospho-immunoblots for S6K (A-B) and rS6p (A-G) for each condition was quantified and is presented below each Fig panel as fold change from untreated cells. The data shown is from one experiment out of three biological replicates.

https://doi.org/10.1371/journal.ppat.1012998.g005

The regulatory network coordinating TORC1 signaling involves multiple amino acid-specific transporters, sensors and signaling auxiliary intermediates [20,58]. We exploited the inherent network complexity to begin addressing the BinA-driven TORC1 sensitization mechanism. First, we tested whether BinA function is broadly prevalent in multiple TORC1-activation pathways, which would likely suggest direct regulation of the TORC1 complex. To this end, we took advantage of two TORC1 amino acid agonists – Met and Gln - that are sensed via mechanisms distinct from Leu and Arg [58–61]. Broadly speaking, sufficiency of different amino acid is detected and relayed to the TORC1 complex via multiple pathways that incorporate distinct small GTPases (RagA/B/C/D, Rab1a or Arf1) which function as molecular switches to recruit TORC1 to either lysosomal or Golgi membranes for activation [20,21,58,60,62,63]. S-adenosylmethionine (SAM) – a methionine-derived metabolite – stimulates TORC1 in a Rag GTPase complex-dependent manner similarly to Arg and Leu [59]. Conversely, glutamine stimulates TORC1 through an Arf1-dependent mechanism [60]; however, glutamine breakdown via glutaminolysis also produces α-ketoglutarate, which can stimulate TORC1 in Rag GTPase complex-dependent manner [61]. Thus, we reasoned that TORC1 stimulation by Met and Gln in the context of BinA could provide mechanistic insight into the sensitization process. To this end, HEK293 cells expressing GFP or GFP-BinA were stimulated with different amounts of Gln (Fig 5C) or Met (Fig 5D). Unlike Leu and Arg stimulation, TORC1 signaling was not sensitized by BinA after treatment with either Gln or Met (Fig 5C-D) indicating that BinA unlikely targets the TORC1 complex directly which would result in sensitization to all amino acid treatments. These data also indicate that BinA likely sensitizes the Rag GTPase complex-dependent TORC1 activation pathway, idea that was tested in the next set of experiments.

The heterodimeric Rag GTPase complex coordinates TORC1 signaling by recruiting TORC1 to the lysosomal membrane based on amino acids availability where the complex is directly activated by the small GTPase Rheb [20,58]. The primary components of the Rag complex are RagA, RagB, RagC and RagD which form a heterodimer of a small subunit (RagA or RagB) and a large subunit (RagC or RagD) [21,63]. Under amino acid replete conditions, the complex consists of a GTP-bound RagA/B and a GDP-bound RagC/D, whereas upon amino acid starvation the nucleotide-bound state of the GTPases is reversed [20,58]. Ectopic expression of different point mutant RagA/B/C/D alleles that lock the GTPase in distinct nucleotide bounds states promotes or interferes with TORC1 signaling depending on the allele [21,63]. As expected, expression of the GDP-locked point mutant allele RagA T21N interfered with Arg-induced TORC1 activation (Fig 5E). Cells co-expressing RagA T21N and BinA exhibited ~ 3-fold reduction in the p-rS6p signal in low dose Arg starvation/refeeding experiments as compared to cells expressing BinA alone (Fig 5F) indicating that TORC1 sensitization by BinA is a Rag GTPase complex-dependent event. Consistent with that idea, cells producing BinA together with the GTP-locked RagA Q66L allele hyperactivated TORC1 when stimulated with low dose Arg in comparison to cells producing RagA Q66L alone (Fig 5G).

Based on these data, we conclude that (i) BinA sensitizes TORC1 signaling specifically to Arg/Leu stimulation by targeting the major Rag GTPase complex-dependent arm of the pathway at a point upstream of the TORC1 complex and (ii) BinA can enhance TORC1 signaling independent of other bacterial factors.

BinA enhances amino acid uptake in infected macrophages

We reasoned that BinA likely regulates a factor or a process common to the pathways relaying arginine and leucine sufficiency to TORC1 because response sensitization was specific for Arg and Leu stimulation but not Met or Gln treatment (Fig 5A-D). The methionine-derived metabolite SAM triggers TORC1 activation through the sensory proteins SAMTOR (S-adenosylmethionine sensor upstream of mTORC1) and the Rag GTPase complex [59], thus it is unlikely BinA targets the following proximal upstream TORC1 regulators that are shared by the Arg, Leu and Met sufficiency response pathways – (i) the pentameric Ragulator complex that tethers the Rag GTPase complex to lysosomes [23]; (ii) the GATOR complex (GTPase activating proteins toward Rags), which has GAP activity towards the small RagA/B GTPases [64]; (iii) the Rag GTPase complex, which recruits TORC1 to lysosomes [21,63]. Therefore, we hypothesized that BinA might regulate Leu/Arg transporters or the respective amino acids cytosolic sensors.

TORC1 activation through the amino acid sufficiency pathway can be triggered independently of amino acid transport pathways by the lysosomotropic agent LLoMe (l-leucyl-l-leucine methyl ester). LLOMe is internalized via receptor mediated endocytosis and in lysosomes is converted into a hydrophobic polymer which subsequently transverses the lysosomal membrane and triggers TORC1 signaling via the amino acid sensing pathway [65]. BinA expression did not increase nor accelerate TORC1 signaling in HEK293 cells upon stimulation with different doses of LLoMe (Fig 6), which would be expected if BinA functions downstream of amino acid import. Because BinA cannot sensitize TORC1 when amino acid transport is eliminated as a variable, the data suggests BinA may function by regulating Arg/Leu import.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. BinA does not sensitize TORC1 signaling in response to L-Leucyl-L-Leucine methyl ester.

MTOR sensitization by BinA is not evident when HEK293 cells expressing either GFP or GFP-BinA are treated with low doses of the membrane permeant TORC1 signaling agonist L-Leucyl-L-Leucine methyl ester (LLOMe). Immunoblot data from amino acid starvation/LLOMe refeeding experiment are shown, where band signal intensity in the phospho-immunoblot for rS6p for each condition was quantified and is presented below the respective Fig panel as fold change from untreated cells. The data shown is from one experiment out of three biological replicates.

https://doi.org/10.1371/journal.ppat.1012998.g006

Therefore, BinA capacity to modulate amino acid import was investigated in the context of infection. To this end, we utilized a fluorescent sensor to measure specifically the uptake of the amino acid analogue L-BPA (L-boronophenylalanine) which is imported by the heterodimeric branched amino acid transporter LAT1 (L-type amino acid transporter 1) (Fig 7A) [66]. For these experiments, Myd88KO immortalized BMDMs (iBMDMs) were infected with binA+ or binA− strains and L-BPA import by infected cells was measured for 5min at 4 hpi (Fig 7B). Macrophages infected by the L.p ∆binA imported significantly lower amounts of L-BPA as compared to the parental wild type strain (Fig 7B). Importantly, plasmid-borne BinA complemented the L-BPA import defect demonstrating that BinA secreted by L.p during infection can alters amino acid transport through LAT-1 (Fig 7B). As expected, treatment with the LAT-1 inhibitor BCH (2-aminobicyclo-(2,2,1)-heptane-2-carboxylic acid) reduced L-BPA import (Fig 7B). Together, these data implicate BinA as a bacterial regulator used by L.p to enhance amino acid import in the cytosol of infected cells which can account for BinA capacity to sensitize TORC1 signaling.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. BinA is required for optimal amino acid import through the L-type amino acid transporter 1 (LAT1) in macrophages infected with Lp.

(A) Schematic depiction of the LAT1-dependent uptake assay for import of the amino acid analog ρ-Boronophenylalanine (L-BPA). Cell were treated with L-BPA for 5min (1) after which cells were extensively washed, followed by the addition of a specific membrane permeable probe for additional 5 min (2), which emits fluorescent signal when bound to L-BPA (3). (B) Graph shows L-BPA uptake by Myd88 KO iBMDMs infected with the indicated Lp strains at MOI = 20 or left uninfected (UN). In some control conditions, the LAT1 inhibitor 2-Aminobicyclo [2.2.1] heptane-2-carboxylic acid (BCH) at 1mM was added 10 min prior to and for the duration of the L-BPA uptake assay. BPA uptake lasted 5min and was measured at 4 hpi, after which fluorescence output was measured, the cells were lysed and total protein content was measured. Mean fluorescence data was normalized to the total protein content. Averages ± SEM of data from three biological repeats are shown for each condition. Statistical analysis was completed via one-way ANOVA Kruskal-Wallis test with Dunn’s multiple comparison and the respective p-values among the conditions compared are indicated.

https://doi.org/10.1371/journal.ppat.1012998.g007

Previous data indicates that Legionella-induced inhibition of host protein synthesis triggers TORC1 signaling in infected cells by increasing the cytosolic amino acid pool [48]; however, it is unclear if import of extracellular amino acids pool contributes to TORC1 activation by L.p. Thus, we measured TORC1 signaling under infection conditions in the presence or absence of extracellular amino acids. To this end, Myd88KO iBMDMs were infected for 4hrs in medium containing amino acids (serum free RPMI), which resulted in robust 15.3-fold induction of p-rS6p as compared to uninfected cells (Fig 8A). Subsequently, infections were continued either in the presence (serum free RPMI) or absence (HBSS) of extracellular amino acids for additional 4 hours (Fig 8A). In this setting TORC1 signaling gradually decreased upon withdrawal of extracellular amino acids reaching the level observed in uninfected cells within 2 hours of starvation demonstrating that the non-cytosolic amino acid pool is needed from sustained TORC1 activation in L.p infections (Fig 8A). In agreement, TORC1 signaling was maintained in the presence of exogenous amino acids (Fig 8A). The phosphorylation of eIF2α (Eukaryotic initiation factor-2α), which is triggered by cellular stress responses, did not decline under the same conditions indicating that amino acid withdrawal did not reduce macrophage capacity to carry out signal transduction in general (Fig 8A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Extracellular amino acids sustain TORC1 activity in Lp infected cells.

(A) Immunoblot analysis of TORC1 (p-rS6p) and stress response (p-eIF2α) signaling in Lp infected macrophages in the presence or absence of extracellular amino acids (MOI = 15). Data is derived from Myd88 KO iBMDMs that were infected with Lp for 4 hours after which RPMI-derived amino acids were either kept or withdrawn by replacing the RPMI medium with amino-acid free Hank’s Balanced Salt Solution (HBSS) for the indicated time periods. Band signal intensity for p-rS6p and p-eIF2α for each condition was quantified and is presented below the respective Fig panels as fold change from uninfected (UN) cells. (B) BinA expression prolongs TORC1 activation caused by protein translation inhibitors. Changes in TORC1 signaling upon amino-acid withdrawal (HBSS treatment) from HEK293 cells producing either GFP or GFP-BinA in the presence or absence of 350µM cycloheximide (CHX). CHX was added 30min prior to and was retained during the HBSS treatment. Band signal intensity for p-S6K and p-rS6p for each condition was quantified and is presented below the respective Fig panels as percentage of signal from cells that were not treated with HBSS. The data shown is from one experiment out of three biological replicates.

https://doi.org/10.1371/journal.ppat.1012998.g008

Next, we investigated the interplay between the cytosolic and the extracellular amino acid pool in TORC1 signaling in the context of BinA expression. Treatment of HEK293 cells expressing either GFP or GFP-BinA with the protein translation inhibitor cycloheximide, which increases the cytosolic amino acid pool [51], prolonged TORC1 signaling upon withdrawal of extracellular amino acids (Fig 8B). However, under those conditions the p-rS6p signal in cell lacking BinA was reduced by >90% within 90 mpt with HBSS, whereas in cell expressing BinA the p-rS6p signal started declining only after 150 mpt (Fig 8B). Conversely, TORC1 signaling was rapidly shut off when extracellular amino acids were withdrawn in the absence of cycloheximide treatment regardless of BinA expression (Fig 8B). Taken together, these data indicate BinA can prolong TORC1 signaling upon withdrawal of extracellular amino acids providing the cytosolic amino acid pool is not depleted by the protein translational machinery. Because (i) amino acid transporters localize at the plasma membrane as well as along the endolysosomal continuum [67–69] and (ii) amino acid transport is in part linked through antiporter activities which couples import of one amino acid with the export of another [70,71], the BinA-enhanced import of TORC1 stimulating amino acids, such as Leu and Arg, might depend on increased cytosolic amino acid pool for sustained bi-directional amino acid transport.

BinA sensitizes the TORC1 response to arginine and leucine though the function of the small GTPases Rab21 and Rab22

BinA has been shown in vitro to activate the Rab5 GTPase family members – Rab5, Rab21 and Rab22 – by catalyzing nucleotide exchange via its VPS9-like domain [56]. Rab5 GTPase family members specialize in the regulation of multiple membrane trafficking events along the endocytic compartment [72]. Rab22 intercepts and rapidly redirects endocytosed CIE (Clathrin independent endocytosis) cargo – including CD98hc-containing heterodimeric transporters – to recycling endosomes, whereas Rab21 controls recycling of CIE cargo from recycling endosomes to the cell surface [73–76]; conversely, Rab5 promotes endosomal maturation [72]. Genetic evidence indicates that TORC1 responses to amino acids sufficiency requires Rab5 presumably because of endocytic maturation [77]. Given the role of Rab5 GTPase family members in trafficking of amino acid transporters and BinA capacity to promote amino acid import to sensitize TORC1, we investigated whether Rab5, Rab21 or Rab22 are involved. To this end, TORC1 signaling was assessed after arginine stimulation in HEK293 cells expressing BinA alone or in combination with different Rab5/21/22 alleles, including constitutively active GTP-locked as well as dominant negative GDP-locked mutants. For Rab5b, neither the GTP-locked Q79L nor the GDP-locked S34N mutant impacted BinA capacity for TORC1 sensitization in response to Arg and Leu (Fig 9A–B and S1A–S1B Fig). Conversely, expression of either the GDP-locked Rab21 T31N or Rab22a S19N inhibited TORC1 sensitization by BinA in response to Arg (Fig 9C-F) and Leu (S1C–S1F Fig). Based on these finding, we conclude that Rab21 and Rab22 but not Rab5 mediate TORC1 sensitization by BinA after Arg or Leu stimulation. Both Rab21 and Rab22 dominant negative alleles independently ablated TORC1 signaling indicating that these GTPase integrate linearly within the same regulatory pathway controlled by BinA. Because Rab21 and Rab22 function in endocytic cargo transport to and from the recycling endosomal compartment, including amino acid transporters, it is likely that BinA-dependent mTOR sensitization activity also originates at that location.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. Sensitization of arginine-dependent TORC1 signaling by BinA is regulated by Rab21 and Rab22a but not Rab5.

Kinetics of TORC1 activation triggered by starvation/refeeding stimulation with 500µM Arg in HEK293 cells producing BinA in the presence or absence of different Rab5 (A-B), Rab21 (C-D), or Rab22a (E-F) alleles. (B, D and F) Quantitative analyses of band signal intensities show Averages ± StDev from at least three biological repeats (left panels) and the respective area-under-the curve (AUC) analyses are presented in the right panels. Statistical analyses were completed with one-way ANOVA with Dunnett’s multiple comparison test using the ‘BinA+GFP’ as control group and p-values are indicated in the respective data panels.

https://doi.org/10.1371/journal.ppat.1012998.g009

These data were surprising because previous study on the role of different Rabs in amino acid-induced TORC1 activation uncovered requirement for Rab5 but not Rab21 or Rab22 [77]. To investigate whether BinA dictates the observed mechanistic differences – i.e., the switch from a Rab5 dependency to Rab21/22 dependency – we took advantage of the RagA GTP-locked mutant Q66L as an alternative eukaryotic TORC1 sensitization factor. Co-expression of RagA Q66L with Rab5b S34N significantly reduced TORC1 activation by either Arg (Fig 10A) or Leu (Fig 10B). Conversely, no significant phenotype was observed upon amino acid stimulation in cells expressing RagA Q66L together with either Rab21 T31N or Rab22a S19N (Fig 10A-B). Thus, BinA and RagA Q66L sensitize TORC1 signaling via distinct mechanisms, whereby different Rab5 family members participate as regulatory factors in each case. Interestingly, when Rab-dependency studies utilizing the Rab5/21/22 GDP-locked mutants were repeated in HEK293 cells that did not produce either BinA or RagA G66L, TORC1 activation by Leu (S2A–S2C Fig) or Arg (S3A–S3C Fig) was not affected. A functional redundancy in the amino sufficiency detection and response pathways downstream of Arg and Leu can reconcile the data, where parallel Rab5-dependent and Rab21/22-dependent cascades generate amino acid sufficiency signaling input to the Rag GTPase complex.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 10. Sensitization of TORC1 signaling by the RagA Q66L allele is regulated by Rab5 but not Rab21 or Rab22a.

Kinetics of TORC1 activation triggered by starvation/refeeding stimulation with either 500µM Arg (A) or 100 µM Leu (B) in HEK293 cells producing RagA Q66L in the presence or absence of different GDP-locked Rab5, Rab21 or Rab22a alleles. (A-B) Immunoblots from representative experiments are shown in the left panels and AUC analyses from three biological repeats are graphed in the right panels. Statistical analyses were completed with one-way ANOVA with Dunnett’s multiple comparison test using the ‘GFP-RagA Q66L’ as control group and p-values are indicated in the respective data panels.

https://doi.org/10.1371/journal.ppat.1012998.g010

In amino acid starvation/refeeding experiments which included Rab22 alleles the amount of the Rab22a S19N mutant detected via immunoblotting was consistently lower compared to Rab22a or Rab22a Q64L (Figs 9E & S1E, S2C, S3C Figs). This phenomenon was not observed when lysates from cells cultured in nutrient rich conditions were prepared (S4 Fig). Direct comparison between Rab22a S19N amounts in HEK293 cells prior to and 4 hours post amino acid withdrawal confirmed a reduction by more than 70% (S4 Fig), indicating a potential link between amino acids sufficiency sensors and the Rab22 nucleotide bound state, where inability to cycle from the GDP-bound state under amino acids starvation targets the protein for degradation.

BinA causes steady state accumulation of LAT-1 at the plasma membrane

Given that our data linked amino acid transport and the endosomal trafficking regulators Rab21/Rab22 with BinA-driven TORC1 sensitization, we investigated whether altered LAT-1 subcellular localization could be responsible for the phenotype because LAT-1, similar to other CIE cargos, transverses the recycling endosomal compartment, which is regulated by Rab21/Rab22 [73–75,78]. For these trafficking experiments we used the human A549 type II pneumocyte cell line because those cells express high levels of LAT-1, thus facilitating investigation of the endogenous protein via immunofluorescence. Distinct LAT-1 localization patterns were observed when A549 cells were cultured under amino acid replete vs. starvation conditions. Majority of cells (~ 80%) had punctate LAT-1 distribution when exogenous amino acids were provided and obvious plasma membrane accumulation upon amino acids withdrawal (Fig 11A-C). Phorbol ester treatment has been shown to trigger internalization of LAT-1 from the plasma membrane in Hela cells [79] and PMA stimulation in the absence of exogenous amino acids caused redistribution of LAT-1 from the cell surface into an intracellular vesicular compartment in A549 cell as well (Fig 11A-C). Both Rab21 (Fig 11D & F) and Rab22 (Fig 11E & F) decorated the LAT-1 + vesicles under those conditions indicating that LAT-1 can cycle from the recycling endosomal compartment to the cell surface and back based on amino acid availability and/or PMA stimulation. Upon amino acids starvation, cell surface re-localization of LAT-1 was disrupted when cells express the GDP-locked Rab21 T31N allele which is consistent with the notion of Rab21-dependent transport from the recycling endosomal compartment to the cell surface (Fig 11G-I). Taken together, these data demonstrate that cells respond to scarcity of exogenous amino acids by re-distribution of LAT-1 from intracellular vesicles to the cell surface – a process that is regulated, at least part through Rab21.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 11. Rab21-dependent LAT-1 plasma membrane recycling in response to exogenous amino acids.

(A-C) LAT-1 cycles between the plasma membrane and intracellular vesicles in response to exogenous amino acids. Subcellular localization of LAT-1 in A549 cells cultured in amino acid replete medium or following depletion of exogenous amino acids (HBSS treatment for 5hrs) in the presence or absence of phorbol ester stimulation for the last 2hrs of the HBSS treatment (depletion + PMA). Micrographs from merged and their respective individual channel images are shown in (A), where arrows indicate vesicular and arrowheads indicate plasma membrane localization of LAT-1. The LAT-1 signal is also shown as pseudocolor in (B). (C) Quantitative analysis of cells with surface localized LAT-1 under the indicated conditions. (D-F) LAT-1 transverses the recycling endocytic compartment marked by Rab21 and Rab22. Representative micrographs showing LAT-1 within Rab21+ (D) or Rab22+ (E) vesicles following internalization from the plasma membrane after treatment with HBSS+PMA. Insets highlight cellular regions with LAT-1 + vesicles accumulating Rab21/Rab22 (arrow) or devoid of the small GTPases (arrowhead). (F) Quantitative analysis of LAT-1 + vesicles recruiting Rab21 or Rab22. (G-I) Plasma membrane redistribution of LAT-1 upon withdrawal of exogenous amino acids is mediated by Rab21. Micrographs showing LAT-1 localization 5 hours after withdrawal of exogenous amino acids in A549 cells expressing either GFP-Rab21 (G) or GFP-Rab21 T31N (H). (I) Quantitative analysis of cells with plasma membrane localized LAT-1 following 5 hour-long HBSS treatment. (C,F & I) Each graphed data point represents a single biological replicate where >100 cells (C & I) or >50 cells (F) were scored. Statistical analyses were completed with one-way (C) or two-way (I) ordinary ANOVA and p-values are indicated in the respective data panels. Scale bars = 10µm (A,B,D,E,G & H) and inset scale bars = 1µm (D & E).

https://doi.org/10.1371/journal.ppat.1012998.g011

Remarkably, GFP-BinA expression significantly increased the number of cells with plasma membrane localized LAT-1 under amino acids replete conditions as compared to GFP expression (Fig 12A & E). Because amino acid permeases have immediate access to substrates at the plasma membrane, unlike in endosomes where micropinocytosis and trafficking must occur first, these data provide a mechanistic explanation for the increase in amino acid import (Fig 7) and consequently TORC1 signaling (Fig 4) caused by BinA. On the other hand, expression of RagA Q66L allele did not cause LAT-1 surface re-localization under the same conditions (Fig 12B & E), demonstrates that hyper TORC1 signaling itself did not promote LAT-1 anterograde transport. Under amino acid starvation LAT-1 re-localized to the plasma membrane in majority of cells expressing either BinA, RagA QL or GFP alone; however, this phenotype was less pronounced in the presence of the RagA QL allele (Fig 12C-E). Together, these data demonstrate that exogenous amino acids sufficiency dampens the Rab21-dependent anterograde trafficking of LAT-1 to the plasma membrane which is overcome when cells produce BinA resulting LAT-1 surface accumulation.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 12. BinA promotes LAT-1 cell surface localization.

(A-D) Representative micrographs showing LAT-1 localization in A549 cells expressing either GFP-BinA (A & C) or GFP-RagA Q66L (B & D) under replete conditions (+AA) (A-B) or in the absence of exogenous amino acids (-AA) (C-D). (E) Quantitative analysis of cells with surface localized LAT-1 under the indicated conditions where each data point is derived from a separate experiment scoring >100 cells. Statistical significance was determined by ordinary two-way ANOVA and p-values are indicated. Scale bars = 10µm (A-D).

https://doi.org/10.1371/journal.ppat.1012998.g012

A central domain of BinA is required for TORC1 sensitization and LAT-1 surface localization

Given BinA GEF activity towards Rab21/Rab22 [56] and the dependence of its TORC1 sensitization capacity on Rab21 and Rab22 we investigated if BinA GEF activity is required for TORC1 sensitization. The BinA D41A allele contains a loss-of-function mutation in the VPS9 domain at a residue required for the nucleotide exchange activity [56]. Indeed, HEK293 cells expressing BinA D41A lacked enlarged Rab5-decorated endosomes unlike cells expressing the wild type BinA allele (S5A–S5C Fig). Expression of BinA D41A by HEK293 cells augmented TORC1 signaling in response to Arg comparably if not slightly better than the wild type allele indicating that the GEF activity of the VPS9 domain is dispensable for TORC1 sensitization (Fig 13A). To determine whether that was the case under infection conditions as well, the L.p ∆binA mutant strain was complemented with an IPTG-inducible plasmid-borne copy of BinA D41A (Fig 2A). In macrophage infections, binA deletion strains complemented with either the GEF-sufficient or the GEF-deficient D41A allele restored TORC1 signaling as indicated by the robust p-rS6p signal measured by quantitative microscopy (Fig 13B-D). In those assays, higher p-rS6p signal was detected on average in cells harboring bacteria expressing the D41A mutation as compared to the parental allele even though protein expression was similar (Fig 2A). Based on our data, we conclude that TORC1 sensitization by BinA is independent of the GEF activity of the VPS9 domain towards the Rab5 GTPase family members.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 13. A central BinA domain is required for LAT-1 surface retention and TORC1 sensitization.

(A) Kinetics of TORC1 activation triggered by starvation/refeeding stimulation with 500µM Arg in HEK293 cells producing GFP, GFP-BinA or GFP-BinA D41A alleles. Band signal intensity in the phospho-immunoblots for rS6p for each condition was quantified and is presented below the Fig panel as fold change from untreated cells. (B) Representative micrographs of TORC1 activation at 6hpi in Myd88KO iBMDMs infected by Lp01 ∆binA strains complemented with plasmid borne BinA or BinA D41A alleles. Micrographs from merged and their respective individual channel images are shown, where arrows indicate remodeled bacteria-occupied vacuoles decorated with ubiquitin. Phosphorylation of rS6p was used as a readout for TORC1 signaling and is shown as pseudocolor in (C). (D) Quantitative microscopy of the p-rS6p signal at 6hpi from individual Myd88KO iBMDMs infected by Lp01 ∆binA strains complemented with empty vector (EV) or with the indicated plasmid borne BinA alleles. Log₁₀ transformed data of p-rS6p mean fluorescent intensity (MFI) is shown where each circle represents data from a single cell. The number of cells analyzed for each condition is noted and the mean for each condition is indicated with a horizontal bar. Statistical analysis was performed with multiple comparisons Kruskal-Wallis one-way ANOVA test. The respective p- values across the different comparisons are shown. (A-D) One representative of three biological replicates is shown. (E) BinA domain architecture showing the six truncation mutants generated in the study. (F) TORC1 activation triggered by starvation/refeeding stimulation with 500µM Leu for 30min in HEK293 cells producing full-length (FL) GFP-BinA or the truncated BinA alleles (as denoted in E). Band signal intensity in the phospho-immunoblots for rS6p for each condition was quantified and is presented below the Fig panel as fold change from untreated cells expressing full-length BinA. (G) Quantitative analysis of A549 cells expressing the indicated GFP or GFP-BinA alleles showing surface localized LAT-1 in the presence of exogenous amino acids where each data point is derived from a separate experiment scoring >100 cells. Statistical significance was determined by ordinary two-way ANOVA and p-values are indicated.

https://doi.org/10.1371/journal.ppat.1012998.g013

To identify region(s) required for TORC1 sensitization we generated a series of BinA N- and C-terminal truncation alleles (Fig 13E) and assessed their activity in Leu starvation/refeeding experiments. A BinA (1–130aa) mutant allele encoding only the VPS9 domain was sufficient to sensitize TORC1 suggesting that while its GEF activity is dispensable the domain itself can promote TORC1 signaling (Fig 13F). Remarkably, the BinA (84–287aa) mutant which lacks 65% of the VPS9 domain, including the critical catalytic D41 residue, could sensitize TORC1 responses to Leu (Fig 13F). Only mutants lacking a central region spanning residues 84–130 lost the capacity to sensitize TORC1 (Fig 13F). Thus, our data defines a central region of BinA containing residues 84–130 as a domain required for TORC1 sensitization.

If TORC1 sensitization by BinA is downstream of altered LAT-1 anterograde transport, one would expect the same molecular determinants would govern both processes. Therefore, the ability to trigger LAT-1 surface re-localization under amino acids sufficiency in A549 cells was investigated for the BinA truncation mutants for which we determined TORC1 sensitization capacity. Similar to full length BinA, the VPS9-only mutant (1–130aa) enhanced LAT-1 abundance at the plasma membrane (Fig 13G). However, in cells expressing mutants lacking the central domain required for TORC1 sensitization LAT-1 localized mostly within intracellular vesicles (Fig 13G). Taken together, these data demonstrate that a central BinA domain (84–130aa) is required for both subversion of LAT-1 trafficking and TORC1 sensitization.

Ethics statement

Ethical approval for animal procedures for experiments in this study was granted by the Institutional Animal Care and Use Committee at LSUHSC-Shreveport (protocol# P-15–026)

Reagents

The pharmacological agents used in the study were obtained from the following suppliers - Cayman Chemical Company (Torin2); Alfa Aesar (Leucine, Arginine, Glutamine, Cysteine, and iron nitrate); Cayman Chemicals (L-Leucyl-L-Leucine methyl ester). Primary antibodies were purchased from: (i) Cell Signaling – α-Akt (pan) (C67E7), α-prS6p (S235/236, D57.2.2E), mTOR (7C10), α-pS6K1 (T389, 108D12), α-p70 S6Kinase (E8K6T), α-LAT1/SLC7A5 (E9O4D) and rS6p (5G10); (ii) Santa Cruz - α-Actin (C-2); (iii) Cayman Chemicals - α-multiubiquitin chain monoclonal antibody (Clone FK2). The following antibody was custom produced by Cocalico Biologicals against formalin-killed bacterial strains– α-L. pneumophila (chicken IgY). Secondary antibodies and Hoechst 33342 (cat #H3570) were purchased from ThermoFisher Scientific as well as α-mouse-Alexa488 (cat #A11001), α-mouse-Alexa647 (cat #A21235), α-rabbit-Alexa647 (cat #A21244), α-chicken-IgY-FITC (cat #A16055), α-chicken-IgY-TRITC (cat #A16059).

Bacterial strains

All L. pneumophila strain used in this study are detailed in S2 Table and were derived from the L. pneumophila serogroup 1, strain Lp01 [83] and have a clean deletion of the flaA gene to avoid NLRC4-mediated pyroptotic cell death response triggered by flagellin when BMMs from C57BL/6J mice are infected with flagellated Legionella.

Bacterial culture conditions: Legionella strains were cultured either on charcoal yeast extract (CYE) plates [1% yeast extract, 1%N-(2-acetamido)-2-aminoethanesulphonic acid (ACES; pH6.9), 3.3mM l-cysteine, 0.33mM Fe(NO₃)₃, 1.5% bacto-agar, 0.2% activated charcoal] or in complete ACES-buffered yeast extract (AYE) broth (10mg/ml ACES: pH6.9, 10mg/ml yeast extract, 400mg/l L-cysteine, 135mg/l Fe(NO₃)₃) supplemented with 100µg/ml streptomycin. As needed, the media were supplemented with the following: 25 μg/ml kanamycin (Kam), 100 μg/ml streptomycin, 10 μg/ml chloramphenicol, or 1 mM Isopropyl β-D-1-thiogalactopyranoside (IPTG).

Bacterial culture conditions for inoculum preparation: For all infection experiments, Legionella strains obtained from dense patches grown on CYE plates for two days were resuspended to OD₆₀₀ of 0.5 U in 1 mL AYE, placed in 15-ml glass culture tubes, and cultivated aerobically 24–26 hours with continuous shaking (175 rpm) at 37°C until early stationary phase was reached (OD₆₀₀ range of 2.0 to 3.0 U). The AYE broth cultures of Lp01 GFP+ strains harboring a plasmid carrying the gfp gene under the control of the P_tac contained 10μg/ml chloramphenicol and were supplemented with 0.1 mM IPTG between 18–20 h post seeding in AYE broth.

Inactivation of Lp genes via clean deletion (CD) mutagenesis: The Lp01 ΔbinA allele was generated by fusing ~1kb regions upstream and downstream of the binA gene. To this end the upstream region was PCR amplified from Lp01 genomic DNA using the primers UpF_CD_BamHI_binA and UpR_CD_EcoRI_binA. The downstream region was similarly produced with the primers DownF_CD_EcoRI_binA and DownR_CD_SacI_binA. The resulting fragments were linked by an EcoRI site and cloned into the BamHI/SacI sites of the gene replacement vector pSR47s creating pSR47s-CDLpg0393. The pSR47s-CD was introduced in the Lp01 ∆flaA strain via tri-parental mating for the generation of the ΔflaA ∆binA strain by allelic exchange of binA with ΔbinA via double homologous recombination. The allelic exchange was confirmed by genotyping clonal isolates with the primers SP1_CD_binA, SP2_CD_binA SP3_CD_binA in a single PCR reaction that produces either a fragment of 324nt for binA or a fragment of 525nt for ΔbinA. Detailed information on the primers used in this study is included in S3 Table.

Generation of plasmid-bearing Lp strains: Plasmids were introduced in an electrocompetent Lp01 ∆flaA ∆binA strain via electroporation (voltage-1800, capacitance- 25µF and resistance- 200ohms). Plasmid bearing clones were isolated after growth selection on CYE plates supplemented with 10μg/ml chloramphenicol after recovery in AYE broth cultures for 6 hours. The GFP expressing strain was generated by introduction of pAM239 plasmid. Production of BinA or BinA D41A was restored by electroporation of clean deletion binA strain with the pJB1806–3XFlag-BinA or the pJB1806–3XFlag-BinA D41A plasmid respectively. The pJB1806–3XFlag plasmid was introduced to generate the control Lp01 ∆flaA ∆binA + EV strain.

Construction of plasmid-borne alleles

Detailed information on the primers and plasmids used in this study is included in S3 and S4 Table respectively. A DNA fragment containing the binA gene flanked by BamHI and SbFI sites was produced in a high-fidelity PCR reaction with the primers F_BamHI_ + 2_lpg0393-binA and R_SbFI_lpg0393-binA using Lp01 gDNA as a template. The BamHI-binA-SbFI amplicon was cloned in pJB1806–3XFlag (BamHI/PstI) for bacterial expression and in pEGFP-C2 (BglII/PstI) for expression in eukaryotic cells. The GFP fusion allele was subsequently moved from pEGFP-C2 into pLB using the AgeI/EcoRI RE sites. The same cloning strategy was used for the cloning of binA truncation mutants into the pLB lentiviral plasmid – all primers are listed in the S3 Table.

The BinA D41A allele was generated via Quick Change Point mutagenesis using pEGFP-BinA as a template with the primers QC_F_lpg0393_D41A and QC_R_lpg0393_D41A. Colony PCR screen was used to identify clones carrying the D41A mutation with the QC_FSP_lpg0393_D41A and the R_SbFI_lpg0393-binA primers. Putative D41A mutant alleles were sequenced. The BinA D41A allele was PCR amplified with the primers F_BamHI_ + 2_lpg0393-binA and R_SbFI_lpg0393-binA and cloned in pJB1806–3XFlag (BamHI/PstI) to generate the pJB1806–3XFlag-BinA D41A plasmid.

The human alleles Rab5b, Rab5b Q79L, Rab5b S34N, Rab22a, Rab22a Q64L, Rab22a S19N, RagA Q66L, and RagA T21N were synthesized as gBlocks with flanking BglII/EcoRI restriction sites and were cloned in pEGFP-C2 plasmid digested with BglII/EcoRI. All alleles were confirmed by sequencing. The murine Rab21, Rab21 Q76L and Rab21 T31N alleles were a gift from Johanna Ivaska (Addgene plasmids # 83421, 83422, 83423; RRID:Addgene_83421, RRID:Addgene_83422, RRID:Addgene_83423) [84]. All GFP-tagged alleles used for the LAT-1 trafficking experiments were relocated from their respective pEGFP-C2 backbones into the lentiviral pLB plasmid through excision and re-cloning with the AgeI/EcoRI RE.

Murine strains and generation of primary murine bone marrow-derived macrophages

C57BL/6J Myd88^-/- mice (cat #009088) were purchased from Jackson Laboratories and housed at LSUH-Shreveport animal facility. Ethical approval for animal procedures for experiments in this study was granted by the Institutional Animal Care and Use Committee at LSUHSC-Shreveport (protocol# P-15–026). Primary BMMs were derived from bone marrow progenitors isolated from C57BL/6J mice were cultured on 10 cm petri dishes in RPMI 1640 with L-glutamine (BI Biologics, cat #01–100-1A) supplemented with 10% v/v FBS (Atlas Biologics, cat #FS-0500-AD), conditioned medium from L929 fibroblast cells (ATCC, CCL-1) (30% final volume) and penicillin/streptomycin (Thermo Scientific, cat#15140163) at 37°C with 5% CO₂. Additional media was added at days 3, 5 and 7 from the start of differentiation. Macrophages were collected and seeded for infections on day 9.

Cell lines derivation and culture conditions

Derivation of immortalized Myd88 KO BMDMs: Bone marrow cells from C57BL/6J Myd88^-/- mice were differentiated into macrophages with L-929 conditioned (30% v/v) complete DMEM medium (High glucose DMEM with L-glutamine and sodium pyruvate (Thermo Cat. 11995073); 10% hiFBS, supplemented with sodium pyruvate to 2mM final concentration; supplemented with L-glutamine to 6mM final concentration 5.5 ml of 100X L-glut; 1X Penicillin/Streptomycin) for 7 days and were immortalized with the J2 murine retrovirus. J2 producing cells were a kind gift from Dr. Jonathan Kagan (Harvard Medical School) [85]. The J2 virus-containing complete DMEM medium was produced by collecting the culture medium from J2 producing cells that were cultured for 24 hours after reaching 100% confluency. The virus-containing medium was centrifuged (400x g, 5 min) and the supernatant was passed through a 0.2µm filter. BMDMs were transduced with J2 virus-containing supernatant fortified with 30% L-929 conditioned medium on consecutive days. Twelve hours after the second round of transduction the medium was replaced with complete DMEM containing L-929 conditioned medium (30% v/v). For the completion of the immortalization process BMDMs were split weekly (1:2 or 1:3) and the amount of L-929 conditioned medium was progressively reduced after every split (from 30% → 25% → 20% → 15% → 10% → 5% → 4% → 3% → 2% → 1% → 0%). Once weaned from L-929 conditioned medium iBMDMs were cultured in complete DMEM medium.

Generation of MTOR^−/−U937 cell line: U937 cell line stably expressing Cas9 were previously described [57]. Lentiviral plasmids containing guide RNAs targeting human MTOR gRNA1 (BRDN0001145082), gRNA2 (BRDN0001147813), gRNA3 (BRDN0001145868) were a gift from John Doench & David Root (Addgene plasmids # 76782, #76783, #76784; RRID:Addgene_76782; RRID:Addgene_76783; RRID:Addgene_76784) [86]. HEK293E cells were used for lentivirus production. Briefly, HEK293E cells were plated in DMEM (10% hiFBS) at confluency in six-well plates one day and 24 hours later were transfected using X-tremeGENE HP DNA transfection reagent transfection reagent according to manufacturer protocol with the retroviral envelope plasmids (pPAX2 and pCMV-VSVG) and 1µg of the gRNA lentiviral plasmid. Media was collected at 24 hours post-transfection and fresh DMEM (30% hiFBS) was added for additional 24 hours. The collected supernatants were combined, centrifuged at 500x g to remove cellular debris and were filtered through 0.45µm filter before they were used for transduction. Cas9-expressing U937 monocytes were plated at 0.7x10⁵ in 6-well plates and were transduced with combined supernatants from all three guide RNAs in DMEM containing 0.5µg/ml polybrene and 250mM HEPES buffer. Plates were centrifuged at 500x g for 30min. After 4 hours, the transduction was repeated. At 24 hours post transduction the culture medium was replaced with fresh DMEM (10% hiFBS).

Cell lines culture conditions: Immortalized BMMs were propagated in complete High-Glucose DMEM supplemented with L-glutamine to 6mM final concentration (ThermoScientific, cat# BP379–100), 10% v/v heat-inactivated FBS (Atlas Biologics, cat# F-0500-DR), 2mM sodium pyruvate (Quality Biologicals, cat#116-079-721), 100 U/ml penicillin and 100 µg/ml streptomycin. The HEK293 and HEK293E (purchased from ATCC) cell lines were propagated in DMEM supplemented with 10% v/v hiFBS, 100 U/ml penicillin and 100 µg/ml streptomycin. U937 monocytes (purchased from ATCC) were propagated in RPMI-1640 with L-glutamine (BI Biologics, cat #01–100-1A), 10% v/v hiFBS, 100 U/ml penicillin and 100 µg/ml streptomycin. The human alveolar type II pneumocyte A549 cell line (purchased from ATCC) was propagated in Ham’s F-12 (Gibco, cat# 11-765-054), 10% v/v hiFBS, 100 U/ml penicillin and 100 µg/ml streptomycin. For macrophage differentiation, U937 monocytes were treated with 10 ng/ml Phorbol 12-myristate 13-acetate (PMA)(Adipogen) for the first 24 hours after which the cells were cultured for additional 48 hours without PMA and antibiotics. All cell lines were cultured in a tissue-culture incubator at a temperature of 37°C in the presence of 5% CO₂.

Amino acid starvation/refeeding in transfected HEK293 cells and iBMDMs

HEK293T cells were plated in poly-L-Lysine coated 24-well plates at 0.8x10⁵ cells/well in DMEM (10% hiFBS). After 4 hours cells were transiently transfected with various plasmids (300ng total DNA/well) using X-tremeGENE HP DNA transfection reagent (Roche, cat # 06366236001) according with manufacturer protocol. The medium was replaced after 48 hours and treatments were carried out at 72 hours post-transfection. The starvation/refeeding treatments in experiments with HEK293 were performed as follows: (i) DMEM medium was replaced with the amino acid-free Hank’s balance salt solution (HBSS, Quality Biological, cat # 119-065-101); (ii) after 4 hour of amino acid starvation, cells were stimulated with different amino acids or LLoMe diluted in HBSS for the indicated time periods; (iii) cell were then lysed and processed for immunoblot analysis. For iBMDM treatments, 1x10⁶ cells/well were seeded in 6-well plates and were cultured in serum free (SF) RPMI overnight. Next, macrophages were infected with Lp01 ∆flaA bacteria in SF-RPMI (MOI = 15). After 4 hours, cells were washed 2X with pre-warmed PBS and HBSS (amino acids −) or SF-RPMI (amino acids +) was added to the cells for the indicated time periods.

Immunoblot analysis

For cell lysis preparation, culture medium was removed, cells were washed with ice-cold PBS and scrapped in complete RIPA buffer (50mM Tris HCl, pH = 8.0 with 150mM NaCl, 1% Igepal CA-630 (NP40), 0.5% sodium deoxycholate and 0.1% sodium dodecyl sulfate (SDS)) containing protease and phosphatase inhibitors (Thermo Scientific, cat #A3296). The total protein concentrations for each lysate was measured with a bicinchoninic acid (BCA) protein assay (Thermo Scientific, cat #23225) and 20µg were resolved by SDS-PAGE, transferred onto nitrocellulose membrane, blocked with 5% milk and incubated overnight with primary antibodies at 4⁰C. Secondary horseradish peroxidase-conjugated antibodies were used at 1:5000 for 2h at room temperature. Chemiluminescence signals were detected and digitalized with GE Amersham imager 680 instrument and images were analyzed with ImageJ version 1.53t (NIH). For each experiment, samples were run in parallel duplicates, where one membrane was immunoblotted with phospho-specific antibodies and the other membrane was probed with antibodies against the total protein.

Immunofluorescence analysis of LAT-1 trafficking in A549

A549 cells transduced with various GFP tagged alleles were seeded on coverslips in 24-well plates at 2x10⁵ cells/well in Ham’s F12 medium (10% hiFBS). After the cells attached, medium was replaced with serum-free F12 for 14 hours. Next, cells were either incubated with fresh serum-free F12 medium (amino acids replete) or HBSS (amino acids depleted) for 5 hours. To some HBSS-treated cells, PMA was added for the last 2 hours of the treatment at 1µM final concentration. For immunostaining, coverslips containing cells were washed with 3x PBS and fixed with 2% paraformaldehyde for 20 min, permeabilized with cold methanol for 30 seconds, blocked with 2% goat serum in PBS for 60 min. Coverslips were incubated with α-LAT1/SLC7A5 (E9O4D) rabbit monoclonal antibody (Cell Signaling, cat #32683S) at 1:250 dilution in PBS containing 1% BSA overnight at 4°C. The secondary antibody was used at 1:500 dilution and Hoechst 33342 at 1:2000 for 60 min at room temperature. Coverslips were mounted with ProLong Gold antifade reagent (ThermoFisher, cat# 36934) onto slides and examined by fluorescence microscopy.

Immunofluorescence analysis of TORC1 activation in infected cells

For infections, BMDMs or iBMDMs (2.5x10⁵ per well) were seed on cover slips in complete DMEM for 8 hours prior to serum starvation. For all experiments cell were serum starved for 10 hours by culturing in RPMI 1640 with L-glutamine (SF-RPMI). Cell were infected with Legionella at MOIs as indicated for each experiment and at 60 min post infection extracellular bacteria were removed by washing 5X with pre-warm PBS. Cells were cultured in SF-RPMI for the duration of the infection.

For immunostaining, coverslips containing cells were washed with 3x PBS and fixed with 2% paraformaldehyde for 20 min, permeabilized with cold methanol for 30 seconds, blocked with 2% BSA in PBS for 60 min. Primary antibodies dilutions were 1:500 for α-p-rS6p, 1:200 for α-ubiquitin (FK2), and 1:1000 for all α-Legionella antibodies. Primary antibodies were incubated in PBS containing 1% BSA overnight at 4°C. Secondary antibodies were used at 1:500 dilution and Hoechst 33342 at 1:2000 for 60 min at room temperature. Coverslips were mounted with ProLong Gold antifade reagent (ThermoFisher, cat# 36934) onto slides and examined by fluorescence microscopy.

Microscopy analyses of infected cells

Images were acquired with inverted wide-field microscope (Nikon Eclipse Ti) controlled by NES Elements v4.3 imaging software (Nikon) using a 60X/1.40 oil objective (Nikon Plan Apo λ), LED illumination (Lumencor) and CoolSNAP MYO CCD camera. Image acquisition and analysis was completed with NES Elements v4.3 imaging software. Only linear image corrections in brightness or contrast were completed. For all analyses, three-dimensional images of randomly selected fields were acquired, and image acquisition parameters were kept constant for all the cover slips from the same experiment. For phospho-rS6p analysis in cells, the ubiquitin signal was utilized to define a binary mask for each cell from which the mean fluorescence intensity of p-rS6p was obtained. Background fluorescence signal from each acquired field was individually determined from a cell-size mask positioned in an unoccupied area and subsequently subtracted from the MFI of each cell in the field.

Automated time-lapse live-cell imaging of Legionella growth with host cells

U937 monocytes were seeded at 1x10⁵ cells/well in 96-well black-wall clear-bottom plates (Corning, cat# 3904) and differentiated into macrophages in RPMI media supplemented with PMA for 24 hours followed by 48 hours culture in PMA-free RPMI medium. All infections were carried out at MOI = 2.5 with AYE broth grown bacteria in the presence of 1mM IPTG. Final media volume was 150µl per well. In each experiment, all conditions were performed in technical triplicates. Plates were centrifuged (500 rpms, 5min) and were loaded into the IncuCyte™ S3 microscope housing module. The IncuCyte™ S3 HD live-cell imaging platform (Sartorius) is a wide-filed microscope mounted inside a tissue culture incubator and is controlled by IncuCyte™ software. For each well, four single plane images in bright field and green channel (ExW 440–480nm/EmW 504–544nm) were automatically acquired with S Plan Fluor 20X/0.45 objective every four hours over several days. Images were analyzed with the IncuCyte™ Analysis software. For quantitation of bacterial replication, GFP signal was used for the generation of a binary mask that defined total signal integrated intensity (GCU x µm²) in each image or for each GFP+ object when LCV expansion was measured. For LCV analysis, bacterial fluorescence was used to generate a binary mask to define individual LCV objects and to measure the object’s area size (GCU x µm²) and the number of objects per image. For single object LCV tracking, the LCV area was recorded for ~100 LCVs per condition for every timeframe from appearance of the LCV to its disappearance. The Incucyte S3 imaging analysis was performed in the Innovative North Louisiana Experimental Therapeutics program (INLET) core facility at LSU Health-Shreveport (RRID: SCR_024990).

Loss-of-function screen for TORC1 regulating Lp T4bSS effectors

Arrayed effector mutant transposon library used in this study has been previously described in detail [52]. TORC1 activation in primary MyD88 KO BMDMs infected by Lp was analyzed via immunofluorescence for p-rS6p (Ser235/236) as detailed in the ‘Immunofluorescence analysis of TORC1 activation in infected cells’ methods section. To this end, Myd88 KO BMDMs were seeded at 2.5x10⁵ cells/well on 1.5mm glass coverslips placed in each well of a 24-well plate in replating medium for 8 hours prior to serum starvation. For all experiments cell were serum starved for 16 hours by culturing in RPMI 1640 with L-glutamine (SF-RPMI). Transposon mutants were grown on CYE agar plates for 48 hours from which liquid cultures in AYE medium were grown overnight as detailed in the ‘Bacterial culture conditions for inoculum preparation’ methods section. Infections were performed in technical triplicates for bacterial strain (~ MOI = 15) and in each set of infections a wild-type control strain (Lp01 ∆flaA) as well as uninfected macrophages were used to provide baselines for TORC1 signaling. All infections were synchronized at 2hpi and were stopped at 8hpi.

Imaging of immunolabelled infected macrophages was completed as detailed in the ‘Microscopy analyses of infected cells’ methods section. For each infection the percentage of infected macrophages with moderate/high p-rS6p signal (MFI > 400) was scored taking into account LCV remodeling (i.e., presence of Ub signal at the LCV) as an additional criterion. More than 20 mutant strains were analyzed in each experimental set of infections, where the mean percentage of p-rS6p+ infected cells as well as the standard deviation of the data was calculated by averaging the data from all the bacterial strains in the set. The z-score (standard deviations away from the mean) was calculated for each bacterial strain. Data is provided in S1 Table.

Assay of boronophenylalanine (BPA) uptake via LAT1-mediated transport

BPA - a phenylalanine derivative - is internalized via the amino acid transporter LAT-1. BPA uptake assay (Dojindo Labratories, cat# UP04) was used to measure BPA transport in Lp-infected macrophages as per the suggested manufacturer’s protocol. Briefly, 1x10⁵ Myd88 KO iBMDMs were seeded in each well of a 96-well black wall microplate and were cultured at 37°C overnight in a 5% CO2 incubator in complete DMEM medium. Macrophages were infected in SF-RPMI medium with the indicated strains grown in overnight AYE liquid cultures at MOI = 20. Infections were synchronized at 2 hpi by several washing cycles with pre-warmed PBS. At 4 hpi, the SF-RPMI culture medium was removed and the cells were washed three times with pre-warmed HBSS. Next, pre-warmed HBSS containing BPA was added to each well and macrophages were allowed to accumulate L-BPA for 5 min in a tissue culture incubator at 37°C. BPA was omitted from the ‘blank’ control. Next, culture supernatants were removed, and cell were washed three times with pre-warmed HBSS, after which HBSS containing the BPA-specific fluorescent probe was added to each well for 5 minutes. LAT-1 transport activity was inhibited with 1 mM 2-aminobicyclo [2.2.1]-heptane-2 carboxylic acid (BCH), which was added 10 min prior to L-BPA addition and was maintained for the duration of the experiment. Fluorescence was measured with a Tecan Spark plate reader using the DAPI filter (Excitation 360 ± 30nm/ Emission 480 ± 30nm) and the gain was set at 75. For each well, the data represents the mean fluorescent signal from 25 tiled area scans that cover the entire surface of the well. Immediately after a plate was imaged the cell were lysed with RIPA cell lysis buffer and the total protein content was measured against a BSA standard. Every experimental condition was run in four technical replicates and the data is presented as fluorescence units/mg protein.