Intrabacterial Regulation of a Cytotoxic Effector by Its Cognate Metaeffector Promotes Legionella pneumophila Virulence

ABSTRACT Legionella pneumophila is a natural pathogen of unicellular protozoa that can opportunistically infect macrophages and cause Legionnaires’ Disease. Intracellular replication is driven by hundreds of bacterial effector proteins that are translocated into infected host cells by a Dot/Icm type IV secretion system. L. pneumophila effectors are temporally regulated in part by a unique family of translocated regulatory effectors, termed metaeffectors, which bind and modulate the function of a cognate effector in host cells. Regulation of the cytotoxic effector SidI by its cognate metaeffector, MesI, is critical for L. pneumophila virulence in natural and opportunistic hosts. MesI binds and negatively regulates SidI activity in vitro, but how impaired regulation of SidI impairs L. pneumophila intracellular replication is unclear. Using a chromosomally encoded inducible expression system, we found that SidI was toxic to L. pneumophila when uncoupled from MesI. SidI enzymatic activity was required for intrabacterial toxicity since L. pneumophila growth was unaffected by induced expression of a catalytically inactive sidI allele. We also found that MesI translocation into host cells was dispensable for intracellular replication and that MesI-deficient bacteria were rapidly degraded within host cells. These data suggest that MesI promotes L. pneumophila intracellular replication by regulating SidI within the bacterium and reveal a unique role for intrabacterial effector regulation by a translocated metaeffector in L. pneumophila virulence. IMPORTANCE Legionella pneumophila replicates within phagocytic host cells using hundreds of effector protein virulence factors, which canonically subvert the function of host proteins and pathways. L. pneumophila encodes a unique family of translocated effectors called metaeffectors, which bind and regulate the function of a cognate effector in host cells. The metaeffector MesI promotes L. pneumophila virulence by regulating the cytotoxic effector SidI; however, the MesI regulatory mechanism is poorly understood. We discovered a unique intrabacterial role for MesI in L. pneumophila virulence. When uncoupled from MesI, SidI was toxic to L. pneumophila in vitro and triggered robust bacterial degradation in host cells. Furthermore, translocation of MesI was dispensable for intracellular replication, demonstrating that intrabacterial regulation of SidI contributes to L. pneumophila virulence. These data show a novel and important role for translocated effector activity within the bacterium, which challenges the dogma that L. pneumophila effectors function exclusively within host cells.

300 effector proteins, bacterial virulence factors translocated into host cells through a Dot/ Icm type IV secretion system (T4SS) (6). Dot/Icm-translocated effectors are critical for L. pneumophila virulence in natural and opportunistic hosts (7); however, the role of most effectors remains poorly understood due primarily to functional redundancy and rarity of virulence phenotypes in laboratory infection models.
Many Gram-negative bacterial pathogens employ effector proteins as part of their virulence strategy. Canonically, effector proteins function by targeting host proteins and subverting cellular pathways to the benefit of the pathogen; however, L. pneumophila encodes a unique family of translocated effectors termed metaeffectors, which bind and regulate other L. pneumophila effectors within host cells (8,9). Metaeffectors are a key component of L. pneumophila pathogenicity since several are required for intracellular replication, but how most metaeffectors promote virulence is poorly understood.
We discovered that regulation of the effector SidI (Lpg2504) by its cognate metaeffector, MesI (Lpg2505), is critical for L. pneumophila virulence (10). Like most L. pneumophila effectors, loss of SidI has no effect on virulence in laboratory infection models; however, loss-of-function mutation in mesI (DmesI) uniquely attenuates L. pneumophila virulence in amoebae, macrophages, and mouse models of Legionnaires' Disease only when SidI is produced (10). SidI is a cytotoxic GDP-mannose-dependent glycosyl hydrolase that potently blocks eukaryotic mRNA translation (11,12). MesI binds SidI with high affinity, suppresses its toxicity, and abrogates SidI-mediated translation inhibition in vitro (12). The requirement for MesI is relieved when sidI is deleted (DsidI); however, a single amino acid substitution at Arg453, which renders SidI nontoxic and catalytically inactive (11,12), is also sufficient to restore replication of MesI-deficient L. pneumophila (10). These data have provided key biochemical insights into the MesI regulatory mechanism, but how MesI regulation of SidI promotes intracellular replication is still unclear.
In this study, we made the surprising discovery that MesI promotes L. pneumophila virulence by regulating SidI intrabacterially. SidI was toxic to L. pneumophila when uncoupled from MesI in vitro. We found that MesI translocation was dispensable for L. pneumophila intracellular replication and MesI-deficient bacteria are rapidly degraded in host cells. These data suggest a unique and important role for interkingdom effector regulation in L. pneumophila virulence.

RESULTS
Overexpression of sidI impairs L. pneumophila growth in vitro. We discovered that the L. pneumophila metaeffector MesI promotes bacterial intracellular replication by regulating its cognate effector SidI (10), but how loss of SidI regulation attenuates L. pneumophila virulence is unclear. The L. pneumophila DmesI strain is severely impaired for intracellular replication, but virulence is restored by sidI chromosomal deletion (DsidI DmesI) or by a single amino acid substitution (R453P) that renders SidI nontoxic and severely attenuates catalytic activity (sidI R453P DmesI) (10)(11)(12). We initially validated that excess SidI activity impairs L. pneumophila intracellular replication by plasmid-based genetic complementation of the DsidI mutation. We infected primary bone marrow-derived macrophages (BMDMs) from Nlrc4 2/2 mice, which are permissive to flagellated L. pneumophila (13,14), and quantified CFU over 72 h of infection. Indeed, induced expression of sidI from a complementing plasmid (psidI) severely attenuated L. pneumophila replication within BMDMs, similar to what we observed for the L. pneumophila DmesI strain (Fig. 1A). Interestingly, we had to induce sidI expression at the time of infection since we were unable to culture the L. pneumophila DsidI DmesI (psidI) strain on solid inducing media. Inducible plasmid-based genetic complementation is an established technique to fulfill molecular Koch's postulates, and we routinely culture L. pneumophila on inducing media (10,12,(15)(16)(17). We hypothesized that sidI adversely affects L. pneumophila fitness and found that replication of the L. pneumophila DsidI DmesI strain harboring psidI was unable to replicate in broth under inducing conditions (1isopropylb-D-thiogalactopyranoside [IPTG]) compared to both wild-type (WT) L. pneumophila and the DsidI DmesI (psidI) strain grown under noninducing conditions (Fig. 1B). Bacteria harboring psidI grown under noninducing conditions were slightly attenuated for replication, but the differences did not reach statistical significance (Fig. 1B). Induced expression of the sidI R453P allele also significantly impaired L. pneumophila growth (Fig. 1C); however, growth attenuation was far less severe than that observed for bacteria expressing the wild-type sidI allele (Fig. 1B). SidI R453P retains a small amount activity (11,12), which is likely responsible for the attenuation observed in Fig. 1C. These data support previous observations that SidI activity is dose dependent. These data also correlate with our previous observation that the L. pneumophila DmesI strain is not attenuated for growth in broth (10). Since endogenous sidI expression is >3-fold lower in broth compared to bacteria grown within host cells (18), the amount of endogenous SidI produced in broth is likely insufficient to impair L. pneumophila growth. Plasmid-based overexpression of an effector gene has not previously been associated with impaired bacterial growth in vitro, and these data suggest that SidI confers a dose-dependent fitness disadvantage on L. pneumophila when stoichiometrically uncoupled from MesI.
MesI rescues the SidI-mediated growth defect in vitro. MesI suppresses SidI toxicity in yeast (10); thus, we hypothesized that MesI also suppresses SidI toxicity in L. pneumophila. To test this hypothesis, we generated L. pneumophila strains that enable tightly controlled inducible expression of the sidI-mesI locus from its endogenous position in the chromosome. We inserted the araC-araBAD (P BAD ) promoter and an in-frame 3Âflag epitope tag directly upstream of sidI in the chromosome of the L. pneumophila WT, DmesI, and sidI R453P DmesI (B, C) Optical density at 600 nm (OD 600 ) of L. pneumophila strains grown in AYE broth. Where indicated, bacteria were grown in the presence (1) or absence (2) of 1 mM IPTG. WT growth from the same experiment shown in both panels B and C for comparison. Data shown are mean 6 SD on samples in triplicate, and asterisks denote statistical significance by one-way analysis of variance (ANOVA) (**, P , 0.01; ***, P , 0.0001). Data shown are representative of at least three independent experiments. strains ( Fig. 2A). Kinetics and abundance of 3xFLAG-SidI fusion protein production were consistent between strains and only observed under arabinose-inducing conditions (Fig. 2B). We observed a tightly controlled increase in P BAD -mediated sidI expression relative to endogenous levels (Fig. 2C). The ;4-fold increase in sidI expression from the P BAD promoter is more physiologically relevant than the >10-fold increase in expression from the Ptac promoter ( Fig. 2C) (18). Thus, we have established L. pneumophila strains in which we can control expression of sidI and mesI in their native stoichiometry and endogenous locus in the chromosome.
We leveraged our L. pneumophila P BAD strains ( Fig. 2A) to test the hypothesis that stoichiometric production of MesI abrogates SidI-mediated growth attenuation. We quantified replication our L. pneumophila P BAD strains under inducing and noninducing conditions and found that induced expression of sidI, but not the inactive sidI R453P allele, significantly attenuated bacterial growth only when uncoupled from mesI in broth culture (Fig. 3A) and within BMDMs (Fig. 3B). Thus, the SidI-mediated growth defect is suppressed by MesI.
MesI suppresses SidI intrabacterial toxicity. SidI impairs L. pneumophila growth, but it is unclear whether SidI is bacteriostatic or bactericidal. SidI is toxic to yeast (11); thus, we tested the hypothesis that SidI is bactericidal when uncoupled from MesI. L. pneumophila P BAD strains ( Fig. 2A) were grown for 24 h under arabinose-inducing or noninducing conditions in broth, and viability was evaluated using LIVE/DEAD viability staining and confocal microscopy (Fig. 4). Very few dead bacteria were observed under noninducing conditions or when induced sidI expression was coupled with mesI (P BAD sidI) (Fig. 4). However, significantly more dead bacteria were observed when enzymatically active sidI was uncoupled from mesI under inducing conditions (Fig. 4). We also observed that bacterial growth was rapidly blocked and CFU decreased over time when sidI expression was induced in exponentially growing L. pneumophila P BAD sidIDmesI strain cultures (see Fig. S1 in the supplemental material). These data suggest that SidI toxicity is conserved across the kingdoms of life and suppressed by MesI.
Intrabacterial regulation of SidI by MesI is sufficient for L. pneumophila virulence. Based on our observation that SidI is toxic to L. pneumophila when uncoupled from MesI, we  hypothesized that MesI intrabacterial regulation of SidI is sufficient for L. pneumophila virulence. We tested this by evaluating whether an intrabacterially retained MesI mutant could complement the DmesI mutation. We generated an allele of mesI lacking 10 amino acids from its extreme carboxy terminus; these amino acids are dispensable for SidI binding and contain the putative E-block Dot/Icm translocation signal (19,20). We cloned this allele into a complementing plasmid downstream of a 3Âflag epitope tag (pmesI D10 ) and transformed the plasmid into the L. pneumophila DmesI strain. We evaluated 3xFLAG-MesID10 translocation using Western blotting to visualize its abundance in saponin-soluble or -insoluble lysate fractions from L. pneumophila-infected cells, a technique routinely used to evaluate effector secretion, since saponin solubilizes eukaryotic membranes but minimally lyses L. pneumophila (21,22). HEK293 Fcg RII cells were infected with antibody-opsonized L. pneumophila harboring either pmesI or pmesI D10 . 3xFLAG-MesID10 was not translocated since it was retained in the insoluble fraction (pellet) and not present in saponin-soluble lysate fractions (Fig. 5A). As expected, wild-type MesI was secreted into host cells since it was present within saponin-soluble lysate fractions (Fig. 5A). Isocitrate dehydrogenase (ICDH) and b-actin were used as bacterial lysis and loading controls, respectively. We also confirmed that MesID10 interacts with SidI since it was retained by SidI on beads similarly to full-length MesI (see Fig. S2 in the supplemental material).
To evaluate a role for intrabacterial SidI regulation by MesI in L. pneumophila virulence, we tested whether the mesI D10 allele genetically complements DmesI mutation. We infected BMDMs and quantified replication of wild-type L. pneumophila and DmesI strains harboring pmesI, pmesI D10 , or empty plasmid vector (pEV). We were able to genetically complement the DmesI strain with the mesI D10 allele, demonstrating that MesI translocation into host cells is not necessary for L. pneumophila intracellular replication (Fig. 5B). These data suggest that intrabacterial regulation of SidI by MesI is sufficient for L. pneumophila virulence.
MesI-deficient bacteria are degraded within host cells. We found that MesI suppresses intrabacterial SidI toxicity and that MesI translocation into host cells is dispensable for L. pneumophila intracellular replication. Since SidI is also toxic to eukaryotic cells, we initially postulated that the virulence defect associated with loss of MesI results from SidI toxicity in host cells. However, we were unable to detect SidI-mediated cytotoxicity in L. pneumophila-infected BMDMs (see Fig. S3 in the supplemental material), and our new data suggest a potential role for MesI suppression of intrabacterial SidI toxicity in L. pneumophila's virulence strategy.
Our data show that SidI is toxic to L. pneumophila and suggest that MesI regulation of SidI within the bacterium is sufficient for intracellular replication. However, our SidI toxicity phenotypes were observed exclusively in vitro under inducing conditions, and it is unclear whether intrabacterial SidI toxicity contributes to the DmesI strain virulence defect. Since Intrabacterial SidI Regulation by MesI mSphere dead and dying bacteria are rapidly degraded by host cells, we rationalized that MesI-deficient bacteria would be degraded more robustly within host cells compared to L. pneumophila control strains. Transcriptomic analysis and in vitro expression data suggest that sidI is constitutively expressed at low levels but that expression increases from 4 to 18 h postinfection and that sidI expression is ;3-fold higher in intracellularly growing bacteria compared to brothgrown bacteria (11,18). Thus, we quantified degraded bacteria by immunofluorescence microscopy and blinded scoring at 8 h postinfection. Degraded L. pneumophila are easily identified using immunofluorescence microscopy since they have lost their rod shape and appear instead as diffuse puncta when cells are stained with a Legionella-specific antibody (23,24). We also evaluated whether MesI-deficient bacteria are able to subvert host endocytic maturation by quantifying association of the lysosomal marker LAMP1 with LCVs (25). We infected BMDMs with the L. pneumophila DmesI strain and compared degradation and LAMP1 localization to those of virulent L. pneumophila and the L. pneumophila DdotA strain, which is unable to evade lysosomal fusion (25). Cells were fixed and stained at 8 h postinfection, at which time sidI expression has increased and virulent bacteria have begun to replicate in a mature LCV (25,26). We found that significantly more L. pneumophila DmesI bacteria were degraded than both wild-type and DdotA control strains ( Fig. 6A and B). It is unlikely that bacterial degradation was accompanied by host cell death since macrophages harboring degraded bacteria had normal nuclear morphology (Fig. 6A, arrows). Furthermore, LCVs harboring in-tact L. pneumophila DmesI bacilli were largely devoid of LAMP1, similar to L. pneumophila WT and in contrast to the avirulent DdotA strain ( Fig. 6A and C). We did observe LAMP1 staining around degraded bacterial puncta, suggesting that bacteria are degraded in lysosomal compartments (Fig. 6A, DmesI [deg]; blue arrowheads). Bacterial degradation was significantly decreased when the DmesI mutation was complemented, and as expected, this strain retained the ability to avoid lysosomal fusion ( Fig. 6A and B). The robust degradation of the L. pneumophila DmesI strain and absence of obvious lysosomal targeting suggests a mechanism of bacterial attenuation distinct from the inability to subvert endocytic maturation.

DISCUSSION
This study supports a unique model whereby the L. pneumophila metaeffector MesI drives virulence by intrabacterial regulation of its cognate effector SidI (Fig. 7). Several L. pneumophila effectors are toxic to eukaryotic cells; however, effector bactericidal activity in L. pneumophila has not been previously observed. Furthermore, our data challenge the dogma that metaeffectors promote L. pneumophila virulence by functioning exclusively within host cells. SidI and MesI bear resemblance to canonical toxin-antitoxin (TA) and effector-immunity (E-I) pairs (27)(28)(29); however, they are distinct since antitoxin/immunity proteins, by definition, are not secreted (30). This also distinguishes MesI from canonical chaperones, which remain intrabacterial (31). Since MesI is a translocated substrate of the Dot/Icm T4SS, regulation of SidI within the host cell is also likely important for L. pneumophila virulence. Current data show that sidI expression is upregulated during the exponential growth phase and that expression of mesI is temporally delayed relative to sidI (18,19). These data suggest that MesI, like other metaeffectors, temporally regulates its cognate effector (8). We postulate that the temporal delay in mesI expression allows the bacterium to both "turn off" SidI activity in the host cell and prevent intrinsic collateral damage. Our data suggest that intrabacterial MesI activity drives L. pneumophila virulence in laboratory infection models; however, it is also important to define the role and mechanism by which Dot/Icm-translocated MesI functions within the host cell cytosol.
Our data suggest that MesI promotes L. pneumophila virulence by suppressing intrabacterial SidI toxicity and challenge the initial assumption that impaired replication was due to SidImediated host cell death. We found that Dot/Icm-translocated SidI did not increase host cell death and that macrophages harboring both in-tact and degraded L. pneumophila DmesI had normal morphology. Moreover, our data suggest that L. pneumophila DmesI strains are killed prior to degradation in host lysosomes since very few of the remaining in-tact bacilli localized with LAMP1. SidI activity is dose dependent, and we think that LCV biogenesis occurs normally until sidI expression increases after ;6 h of growth (11,18). It is very likely that viable L. pneumophila DmesI strains devoid of LAMP1 at 8 h postinfection may have genomic variations that suppress SidI toxicity. Further investigation is required to fully define the kinetics of LCV biogenesis and bacterial degradation when SidI is uncoupled from MesI.
Interkingdom SidI toxicity suggests that its target is highly conserved between L. pneumophila and host cells. SidI is one of at least seven L. pneumophila effectors that subvert host protein synthesis (11,32), one of the most conserved cellular processes across the kingdoms of life (33). Effector-mediated suppression of host mRNA translation is important for L. pneumophila's acquisition of host-derived amino acids, which are essential for intracellular replication and the main driver of bacterial phase switching within host cells (23,34). The mechanism by which SidI inhibits protein synthesis is unclear; however, SidI binds eukaryotic elongation factor 1A (eEF1A) and likely functions as a mannosyltransferase, so it is tempting to speculate that SidI blocks translation by modifying eEF1A (11,12). However, the functional significance of SidI-eEF1A binding remains unclear since it is unaffected by MesI and insufficient for translation inhibition (11,12). L. pneumophila has evolved strategies to prevent intrabacterial suppression of translation. For example, the effector Lgt1 blocks eukaryotic translation by glycosylating eEF1A on Ser53, which is conserved in eukaryotes but not the bacterial eEF1A homolog, EF-Tu (35,36). The mechanistic underpinnings of intrabacterial MesI activity is unclear; however, it is tempting to speculate that MesI may function in part to preserve protein synthesis within the bacterium.
The mechanism of intrabacterial SidI toxicity is unknown. SidI cleaves GDP-mannose and has predicted structural homology to mannosyltransferase enzymes (12). We think that SidI toxicity results from mannosylation of translation factors conserved between the host and L. pneumophila. However, hydrolysis of GDP-mannose may have other deleterious effects within the bacterium. For example, mannose comprises 4% of the core oligosaccharide in L. pneumophila lipopolysaccharide (LPS), and depletion of GDP-mannose by excess SidI activity may adversely affect cell envelope biosynthesis as has been suggested for MutT family enzymes in Escherichia coli (37,38). Interestingly, SidI shares predicted structural homology with phosphatidyl-myo-inositol mannosyltransferases that contribute to cell envelope biosynthesis in mycolated pathogens, including Mycobacterium and Corynebacterium spp. (12,39,40). The biological relevance of this activity in L. pneumophila, which does not have a mycolated cell envelope, is unclear. However, whether impaired LPS biosynthesis contributes to SidI intrabacterial toxicity is unknown and warrants investigation.
Recent work has challenged the dogma that bacterial effectors function exclusively within host cells. Hardwidge and colleagues revealed that the pathogenic E. coli effector NleB and Salmonella enterica serovar Typhimurium effector SseK1 function intrabacterially. NleB confers resistance to oxidative stress and SseK1 confers resistance to methylglyoxal and regulates UDP-GlcNAc biosynthesis (41)(42)(43). The pathogenic E. coli effector NleC can also function intrabacterially, but the endogenous, biologically relevant substrates are unknown (44). Our data suggest that intrabacterial MesI activity confers a fitness advantage on L. pneumophila by suppressing intrabacterial SidI activity. Interestingly, SidI appears to be constitutively produced at low levels when L. pneumophila is grown in broth (19), suggesting that low levels of SidI activity within the bacterium may be beneficial at early stages of the L. pneumophila life cycle. Indeed, there is emerging evidence that bacterial TA pairs can shape pathogen physiology by modulating posttranscriptional gene expression (45). The Escherichia coli toxin RelE potently blocks translation and is neutralized by the antitoxin RelB when the bacteria are grown in nutrient rich environments (46)(47)(48). However, under limiting conditions, RelE activity contributes to adaptation to nutrient starvation by suppressing bacterial translation, and when nutrients are restored, RelB production increases and translation resumes (48). Since L. pneumophila requires exogenous amino acids to replicate, it is tempting to speculate that SidI may suppress translation until its replicative niche within the host cell is established. Thus, SidI and MesI may be an ancient TA pair that have been co-opted by the T4SS to modulate both host and pathogen physiology.
Our data suggest that intrabacterial SidI toxicity is responsible for impaired L. pneumophila DmesI strain intracellular replication. However, McCloskey et al. recently proposed a model whereby MesI negatively regulates SidI translocation into host cells (19). MesI binds the extreme C terminus of SidI, which encodes the canonical E-block Dot/Icm translocation signal (12,19,49) however, the authors found no differences in SidI secretion when the relative abundance of MesI was increased (19). Furthermore, we did not detect any MesI-mediated differences in SidI translocation from L. pneumophila using an established adenylate cyclase (CyaA) reporter (12). This discrepancy may be a consequence of differences in experimental conditions. Further investigation is required to define the impact of MesI on SidI translocation and the potential role of this activity in the MesI regulatory mechanism and L. pneumophila virulence.
Together, this study revealed a unique role for intrabacterial regulation of a translocated effector by its cognate metaeffector in L. pneumophila virulence. Our data, in the context of previously published studies (11,12), suggest that SidI's target is conserved between host and pathogen. Recent phylogenetic analysis suggests ancient and extensive coevolution between the order Legionellales, which includes Legionella spp., and unicellular eukaryotes (50). Thus, metaeffectors may represent an ancient mechanism evolved by pathogenic bacteria for adaptation to eukaryotic hosts.
Plasmids and oligonucleotide primers used in this study are listed in Table 1 and Table 2, respectively. Molecular cloning and strain construction. L. pneumophila genomic DNA (gDNA) was isolated using the Illustra genomicPrep DNA spin kit (GE Healthcare) and used as a template for cloning sidI and mesI into the indicated plasmid vectors. sidI was amplified from L. pneumophila gDNA using SidIBamHI-F/SidISphI-R, which includes 60 base pairs (bp) downstream of the sidI open reading frame, and cloned as a BamHI/SphI fragment into pJB1806 (53) or downstream of an in-frame 3Âflag epitope tag in pSN85 (54) to generate pJB1806::sidI (psidI) and pSN85::sidI. To generate psidI R453P , psidI was mutagenized using site-directed mutagenesis with primer pairs SidIR453P-sense/SidIR453P-asense (10). To generate pmesI complementation plasmids, mesI and mesI D10 open reading frames were amplified from L. pneumophila gDNA using MesIBglII-F/MesISphI-R or MesIBglII-F/MesID10SphI-R primer pairs, respectively, and cloned into pSN85 in-frame with the 3Âflag epitope tag. Ligations were transformed into chemically competent E. coli Top10 and sequences confirmed by Sanger sequencing (Eton Biosciences). L. pneumophila complementation strains were generated by electroporation of plasmids into competent L. pneumophila strains using a Bio-Rad Gene Pulser at 2.4 kV, 200 X, and 0.25 mF and plated on CYE supplemented with 10 mg mL 21 chloramphenicol.
For production of recombinant His 6 -Myc-MesID10, mesID10 was amplified from L. pneumophila gDNA using MesIT7SalI-F/MesID10NotI-R and cloned as a SalI/NotI fragment into pT7HMT (55) downstream of an in-frame His 6 -Myc epitope tag. pT7HMT::mesI and pGEX::sidI were generated previously ( Table 2) (12). Ligation reactions were transformed into chemically competent E. coli Top10, and sequences were confirmed by Sanger sequencing (Eton Biosciences). The dotA open reading was deleted from the L. pneumophila chromosome using allelic exchange as described (56). To generate a clean in-frame deletion of dotA, 59 and 39 regions flanking the dotA open reading frame (726 bp upstream and 1,012 bp downstream) were amplified using DotAKOSacI-F/DotAKONotI-R and DotAKONotI-F/DotAKOSalI-R to generate SacI/NotI and NotI/SalI fragments, which were ligated into SacI/SalI digested pSR47s to generate pSR47s::DdotA, which was conjugated into L. pneumophila SRS43 for selection of double crossover events, as described. Sequences were confirmed by Sanger sequencing (Eton Biosciences) and the DdotA phenotype was confirmed by comparison with the established SRS43 dotA::Tn strain (10).
For allelic exchange to insert the araC-araBAD (P BAD ) promoter and a 3Âflag epitope tag upstream of sidI in the L. pneumophila chromosome, overlapping primers were used to generate a fusion construct consisting of 1,000 bp upstream of sidI (SidIUTR-F/SidIUTR-R; from L. pneumophila gDNA), P BAD promoter (araC-pBAD-F/araC-pBAD-R; from pMRBAD::z-CspGFP [a gift from Brian McNaughton; Addgene number 40730] [57]), and 3Âflag fused to the first 1,200 nucleotides of sidI (3FLAG-SidI-F/3FLAG-SidI-R; from pSN85::sidI), which was ligated as a SacI/SalI fragment into pSR47s to generate pSR47S::P BAD . Plasmids were transformed into chemically competent E. coli DH5alpir, and sequences were confirmed by Sanger sequencing (Eton Biosciences). To generate P BAD strains, pSR47s::P BAD was conjugated into the L. pneumophila SRS43 wild-type, DmesI, and sidI R453P DmesI strains (10) and selection of double crossover events was performed as described (56). Sucrose resistant, kanamycin-sensitive colonies were screened by PCR to verify gene insertion.
Mice. C57BL/6 Nlrc4 2/2 mice (a gift from Craig Roy) have been described (58). In-house colonies were maintained under specific pathogen-free conditions at Kansas State University. Bone marrow was harvested from 8-to 15-week-old mice as previously described (59). All experiments involving animals were approved by the Kansas State University Institutional Animal Care and Use Committee (IACUC-4501 and -4022) and performed in compliance with the Animal Welfare Act and National Institutes of Health guidelines.
Macrophage growth curves. BMDMs were seeded into 24-well tissue culture dishes at 2.5 Â 10 5 per well 1 day prior to infection. L. pneumophila were cultured on CYE agar and heavy patch-grown bacteria were used to infect BMDMs at an multiplicity of infection (MOI) of 1 in triplicates, and CFU were enumerated as previously described (10,15). Fold growth was calculated by normalizing CFU at 24 h, 48 h, and 72 h to internalized bacteria at 1 h postinfection. For genetic complementation, either 1 mM IPTG or 2% L-arabinose (wt/vol) was added to the medium as indicated at the time of infection and maintained throughout.
In vitro growth curves. L. pneumophila heavy patches were resuspended in fresh AYE broth and subcultured for consistent 600 nm optical density (OD 600 ). Cultures were split into triplicate wells of a 96-well round-bottom plate and incubated at 37°C with continuous orbital shaking using an Agilent BioTek Epoch2 plate reader. OD 600 from triplicate wells was read every 2 h for 36 h. Plasmids were maintained with 10 mg mL 21 chloramphenicol and IPTG (1 mM) or 1% L-arabinose (wt/vol) were added to the media to induce gene expression as indicated.
To quantify bacterial growth by CFU assay, L. pneumophila heavy patches were resuspended in fresh AYE broth and subcultured (OD 600 = 0.2), and 3 mL was added to each of six culture tubes. Bacteria were grown at 37°C with shaking. After 24 h, sidI expression was induced in three culture tubes with 1% L-arabinose ([vol/vol] from 30% [wt/vol] stock), a volume equivalent of sterile water (vehicle) was added to the other three culture tubes, and bacteria were grown for an additional 24 h. Samples were taken from culture tubes for plating and CFU enumeration after 3,6,10,24,27,30,34, and 48 h of growth.
Quantitative RT-PCR. L. pneumophila strains were grown for 1 or 6 h in AYE broth of incubation with and without arabinose (2%), and total RNA was purified with the Direct-zol RNA miniprep kit with TRI reagent (Zymo Research) following the manufacturer's instructions. RNA samples were treated with DNase (Sigma) before reverse transcription with SuperScript III (Invitrogen). Quantitative reverse transcriptase PCR (RT-PCR) was performed using the Invitrogen SuperScript III Platinum SYBR green one-step qRT-PCR kit. Transcript abundance was quantified using sidIRT-F/sidIRT-R (sidI) and 16SRT-F/16SRT-R (16S rRNA) primer pairs on a Bio-Rad CFX96 real-time PCR machine. Fold expression (2 DDCT ) was calculated by normalizing sidI transcript abundance to 16S rRNA and standardizing values to wild-type L. pneumophila.
Bacterial viability and LIVE/DEAD staining. L. pneumophila P BAD strains were resuspended in AYE media from a heavy patch (48 h). Strains were grown for 24 h at 37°C in the presence or absence of 0.6% L-arabinose (wt/vol). Cell viability was quantified using a LIVE/DEAD BacLight bacterial viability kit according to manufacturer's instructions. Bacteria were stained with 5 mM SYTO 9 (live; green) and 10 mM propidium iodide (dead; red) (Thermo Fisher) and incubated for 15 min at room temperature in the dark. To determine the baseline threshold for dead cells, a negative control was used, where cells were treated with 90% ethanol for 1 h. After staining, cells were washed with sterile water, and 5 mL of resuspended cells were loaded on a glass slide with a coverslip. Images were acquired at the KSU Division of Biology Microscopy Facility using a Zeiss LSM5 laser scanning confocal microscope using a 100Â oil-immersion objective. Percent dead bacteria was calculated by normalizing dead bacteria (red) to total bacteria using ImageJ software. At least 500 cells were evaluated for each strain and culture condition, and the researcher performing the analysis was blinded to sample identity.
Effector secretion assay. HEK293 Fcg RII cells (60) were seeded in 10-cm poly-L-lysine-coated tissue culture dishes (4 Â 10 6 ) 1 day prior to infection. L. pneumophila strains were patched from fresh single colonies onto CYE agar supplemented with 10 mg mL 21 chloramphenicol and 1 mM IPTG to induce gene expression. Bacteria were opsonized with a-L. pneumophila antibody (1:1,000 [Invitrogen; PA17227]) in DMEM 10% HIFBS supplemented with 1 mM IPTG for 20 min at room temperature (RT) with rotation. Cells were infected with opsonized bacteria at an MOI of 50 in for 2 h in 2 mL DMEM 10% HIFBS supplemented with 1 mM IPTG. Cells were washed 3Â with ice-cold phosphate-buffered saline (PBS) to remove extracellular bacteria and lysed for 10 min in 500 mL ice-cold Hank's balanced salt solution (HBSS) supplemented with 0.2% saponin and ProBlock gold mammalian protease inhibitor cocktail (GoldBio). Lysates were treated with RNase A (10 mg mL 21 ) and DNase I (10 mg mL 21 ), incubated at RT for 15 min, and centrifuged for 15 min at 17,000 relative centrifugal force (rcf) at 4°C. Saponin-soluble supernatants were filtered using a 0.22 mM syringe filter and transferred to a fresh 1.5-mL microcentrifuge tube. Saponin-insoluble pellets were resuspended in 50 mL TE buffer (10 mM Tris-HCl, 1 mM EDTA). Fifty microliters of supernatant and pellet fractions were diluted in 3Â Laemmli sample buffer and boiled for 10 min, and the whole sample was loaded into wells of a 1.5-mm 15% SDS-PAGE gel for Western blotting.
Affinity chromatography. Affinity chromatography from E. coli lysates was performed as described (12). Briefly, E. coli BL21(DE3) harboring pT7HMT::mesI, pT7HMT::mesI D10 , pGEX6P1::sidI, or pGEX6P1 were grown overnight with shaking at 37°C and subcultured at 1:100 in LB media. Subcultures were grown for 3 h followed by induction with 1 mM IPTG and growth overnight at 16°C. Clarified lysates from bacteria producing glutathione S-transferase (GST) fusion proteins were incubated with pre-equilibrated glutathione magnetic agarose beads (Pierce) for 1 h at 4°C with rotation. Beads were washed twice in washing buffer (125 mM Tris-Cl, 150 mM NaCl, 1 mM dithiothreitol [DTT], 1 mM EDTA, pH 7.4) and incubated with clarified lysates from bacteria producing His 6 -Myc fusion proteins at 4°C for 1 h with rotation. Beads were washed twice in washing buffer transferred to a fresh 1.5-mL microcentrifuge tube and boiled in 25 mM 3Â Laemmli sample buffer. Proteins were separated by SDS-PAGE and visualized using Coomassie brilliant blue or Western blotting as indicated.
Cytotoxicity assay. Cytotoxicity was evaluated by quantifying lactate dehydrogenase (LDH) activity in supernatants of L. pneumophila-infected BMDMs. BMDMs were seeded in 24-well tissue culture dishes at 2.5 Â 10 5 in seeding medium 1 day prior to infection. Cells were infected with L. pneumophila strains at an MOI of 50 for 1 h, washed 3 times with sterile phosphate-buffered saline (PBS 2/2 ), and incubated in seeding media for additional 5 h or 9 h. At the indicated times, plates were centrifuged at 250 rcf, and supernatants were transferred to a sterile non-tissue-culture-treated 96-well plate. For the 1 h time point, supernatants were collected from cells without washing. LDH was quantified using the CytoTox 96 nonradioactive cytotoxicity assay (Promega) according to manufacturer's instructions. Absorbance at 490 nm was read on a BioTek Epoch2 microplate reader, and percent cytotoxicity was calculated by normalizing absorbance values to cells treated with lysis buffer.
Immunofluorescence microscopy and scoring. BMDMs (1 Â 10 5 ) were seeded on poly-L-lysinecoated glass coverslips in 24-well tissue culture dishes 1 day prior to infection. BMDMs were infected in triplicate wells with the indicated L. pneumophila strains at an MOI of 30 for 1 h or 8 h. For the 8-h time point, extracellular bacteria were removed after 1 h by washing coverslips three times in PBS 2/2 and incubated with fresh media for 7 h. Coverslips were fixed in 4% paraformaldehyde for 15 min and permeabilized with ice-cold methanol. Coverslips were stained with 1:1,000 rabbit a-L. pneumophila primary antibody (Invitrogen; PA17227) and 1:500 Alexa488-conjugated goat a-rabbit secondary antibody (Thermo Fisher). Where indicated, BMDMs were also stained with 1:1,000 rat a-LAMP1 antibody (ID4B; Developmental Studies Hybridoma Bank) and 1:500 Alexa594conjugated goat a-rat secondary antibody (Thermo Fisher). Nuclei were stained with 1:10,000 Hoechst (Thermo Fisher), and coverslips were mounted on slides with ProLong gold antifade mountant (Invitrogen). Degraded bacteria and LAMP1 1 LCVs harboring in-tact bacilli were scored blind on a Leica DMiL LED inverted epifluorescence microscope (n = 300). Representative images were acquired at the KSU College of Veterinary Medicine Confocal Core using a Zeiss LSM 880 inverted confocal microscope and processed using Fiji ImageJ and Adobe Photoshop software.
Statistical analysis. Statistical analysis was performed with GraphPad Prism 9 software using an unpaired two-tailed t test with a 95% confidence interval. Unless otherwise indicated, data are presented as mean 6 standard deviation (SD), and statistical analysis was performed on triplicate biological replicates.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.