The Legionella effector kinase LegK7 hijacks the host Hippo pathway to promote infection

Summary

The intracellular pathogen Legionella pneumophila encodes translocated effector proteins that modify host cell processes to support bacterial survival and growth. Here, we show that the L. pneumophila effector protein LegK7 hijacks the conserved Hippo signaling pathway by molecularly mimicking host Hippo kinase (MST1 in mammals), which is the key regulator of pathway activation. LegK7, like Hippo/MST1, phosphorylates the scaffolding protein MOB1, which triggers a signaling cascade resulting in the degradation of the transcriptional regulators TAZ and YAP1. Transcriptome analysis revealed that LegK7-mediated targeting of TAZ and YAP1 alters the transcriptional profile of mammalian macrophages – a key cellular target of L. pneumophila infection. Specifically, genes targeted by the transcription factor PPARγ, which is regulated by TAZ, displayed altered expression, and continuous interference with PPARγ activity rendered macrophages less permissive to L. pneumophila intracellular growth. Thus, a conserved L. pneumophila effector kinase exploits the Hippo pathway to promote bacterial growth and infection.

Introduction

Microbial pathogens use a variety of strategies to exploit their hosts and cause disease. Many bacteria deliver proteins, called effectors, into host cells where they alter signaling pathways to establish conditions favorable for bacterial survival and growth (Alto and Orth, 2012). Signaling pathways are often controlled by protein phosphorylation, an important post-translational modification catalyzed by kinases. By transferring the gamma-phosphate group from adenosine triphosphate (ATP) onto substrate residues, kinases control protein function and signal transduction in all living cells. Hundreds of kinases and thousands of kinase substrates have been identified within the human proteome (Manning et al., 2002; Safaei et al., 2011), highlighting the importance and complexity of kinase signaling for cell homeostasis and human health.

Bacterial pathogens can exploit host phosphorylation networks by encoding effectors that possess kinase activity and that phosphorylate host targets during infection. For example, the Yersinia effector YopO/YpkA phosphorylates vasodilator-stimulated phosphoprotein (VASP) and other host proteins that regulate actin polymerization, thereby obstructing phagocytosis and escaping the host innate immune response (Ke et al., 2015; Navarro et al., 2007). Likewise, Salmonella SteC alters the host cell actin cytoskeleton by phosphorylating the mitogen-activated protein (MAP) kinase MEK (Odendall et al., 2012), and Shigella OspG modulates host inflammation by disrupting the nuclear factor kappa B (NFkB) pathway (Kim et al., 2005). Despite their apparent importance for microbial virulence, knowledge about bacterial effector kinases and their role during infection has remained limited.

Legionella pneumophila, the causative agent of Legionnaires’ pneumonia, is another bacterium that encodes effector kinases to exploit its host. This opportunistic pathogen creates a replicative niche within free-living protozoans in the environment or within human alveolar macrophages during infection (Horwitz and Silverstein, 1980; Rowbotham, 1980). Virulence of L. pneumophila relies on the function of close to 300 effector proteins that are delivered into host cells by the Defective in organelle trafficking/Intracellular multiplication (Dot/Icm) type IV secretion system (T4SS) (Ensminger and Isberg, 2009; Zhu et al., 2011). However, the host targets of many L. pneumophila effectors have remained unidentified, and only few of the effectors have been functionally characterized.

Among the large number of L. pneumophila effectors are four proteins, named LegK1 to LegK4, that show primary amino-acid sequence homology to eukaryotic kinases (Hervet et al., 2011). LegK1 activates the host cell NFκB pathway by phosphorylating the IκB family of inhibitors (Ge et al., 2009), while LegK2 from L. pneumophila strain Paris was suggested to phosphorylate the ARPC1B and ARP3 subunits of the ARP2/3 complex to alter host cytoskeletal dynamics (Michard et al., 2015). Although the host substrates of LegK3 and LegK4 have yet to be determined, it has become clear that targeting host signaling pathways through effector kinases is an important virulence tactic of L. pneumophila.

In this study, we discovered and experimentally validated a previously unrecognized L. pneumophila effector kinase and show that it promotes intracellular bacterial growth by targeting the host cell Hippo pathway, an ancient signaling cascade that until now has not been directly associated with microbial virulence processes.

Results

LegK7 is an effector kinase from L. pneumophila

A major obstacle for deciphering the function of L. pneumophila effectors has been that their encoding genes often lack homologs in other non-Legionella genomes, both bacterial and eukaryotic, making similarity-based predictions of their molecular function challenging. To bypass this limitation, we used Profile Hidden Markov Model-based protein structure prediction (Alva et al., 2016) to identify cryptic catalytic domains within L. pneumophila effectors. We discovered that the L. pneumophila protein Lpg1924 (LegK7 hereafter) contains a central domain (residues 183–462) that has folding homology to eukaryotic protein kinases, such as PKAca (Figure 1A, Figure S1A). Despite sharing significant secondary structure similarity (probability > 99.9 %) over an area of 280 residues, the primary sequence identity was below 12%, explaining why the kinase domain in LegK7 has been missed in earlier genome annotations (Chien et al., 2004).

Figure 1: L. pneumophila LegK7 is an effector kinase.

(A) Schematic representation of LegK7 with the predicted kinase domain shown in magenta. Conserved catalytic residues are highlighted in red. Numbers represent amino acid positions within each kinase. NIK, human NF-κB inducing kinase; PKAca, mouse cAMP-dependent protein kinase catalytic subunit a. (B) Human U937 macrophages were challenged with a T4SS-competent L. pneumophila strain, Lp02, or a T4SS-deficient L. pneumophila strain, Lp03, carrying plasmids encoding the reporter protein β-lactamase fused to either LegK7 or RalF, a known translocated effector (positive control). Translocation of β-lactamase fusion proteins into host cells via the T4SS was detected by the CCF4-AM-based FRET assay. (C) Schematic cartoon depicting the detection of thiophosphorylated proteins by the thiophosphate labeling assay. (D) Thiophosphate labeling assay with purified His6-LegK7 wild-type (WT) or His6-LegK7 (D307A). Thiophosphorylated (p^s) proteins were detected by immunoblot. Total protein was detected by silver-staining.

The presence of eukaryotic-like motifs within bacterial proteins indicates that the encoding genes are of eukaryotic evolutionary origin and play a role as translocated effectors during infection (Bruggemann et al., 2006). Consistent with earlier findings (Zhu et al., 2011), we confirmed that LegK7 is a genuine effector translocated into host cells by L. pneumophila that encode a functional T4SS (Figure 1B). Not surprisingly, a L. pneumophila legK7 deletion strain (Lp02ΔlegK7) replicated as proficiently as the parental strain (Lp02) within human U937 macrophages (Figure S1B), consistent with mounting evidence for the existence of functionally redundant effectors that can compensate for the loss of individual effector-encoding genes (O’Connor et al., 2012).

The predicted LegK7 kinase domain contains several putative catalytic residues (D307, K309, N312) that are conserved in the catalytic center of its closest eukaryotic homologs (Figure 1A). To determine whether LegK7 possesses kinase activity, we affinity-purified LegK7 and LegK7(D307A) as hexahistidine (His6)-tagged proteins from Escherichia coli and monitored their auto-phosphorylation, a phenomenon common among kinases. As phosphate donor we used adenosine 5’-O-(3-thiotriphosphate) (ATPγS). Upon conjugation of the γ-thiophosphate onto amino acid residues by kinases, the thiophosphate moiety can be alkylated with p-nitrobenzyl mesylate (PNBM) and detected by a thiophosphate ester-specific antibody (Figure 1C) (Allen et al., 2007). Using this labeling technique, we observed robust auto-thiophosphorylation for His6-LegK7 but not for the mutant protein His6-LegK7(D307A) (Figure 1D). His6-LegK7 also efficiently thiophosphorylated myelin basic protein (Myelin A1) (Figure 1D), a frequently used pseudo-substrate for both eukaryotic kinases and eukaryotic-like kinases from bacteria (Haubrich and Swinney, 2016). Together, these results demonstrate that LegK7 is indeed a L. pneumophila effector kinase, and that D307 is critical for its catalytic activity. Primary sequence homology searches identified 37 LegK7 orthologs in 32 out of 49 Legionella species that have a sequenced genome (Figure S1C). The kinase domain showed a high degree of conservation with the catalytic motif preserved in all orthologs (Figure S1C), suggesting that LegK7 is important for Legionella cell biology.

LegK7 phosphorylates human MOB1A

To learn more about the function of LegK7, we aimed to identify its host substrate(s). Discovering substrates of protein kinases has remained a major challenge because of the transient nature of kinase-substrate interactions and the complexity of the phosphoproteome even at steady-state. To bypass these obstacles, we developed a screening platform that combined the previously described thiophosphate labeling technique (Allen et al., 2007) with a high-density human protein microarray (Figure 2A). Using this platform, we found that thiophosphorylation of Mps one binder kinase activator 1A (MOB1A) was dramatically greater on microarrays incubated with His6-LegK7 than on microarrays incubated with His6-LegK7(D307A) (Figure 2B and 2C). Although several other candidate substrates also showed quantitatively significant though much lower signal ratio scores (Figure 2C), MOB1A emerged as the most promising LegK7 substrate from this screen and was selected for further analysis.

Figure 2: L. pneumophila LegK7 phosphorylates human MOB1.

(A) Schematic overview of kinase substrate identification on protein microarrays. (B) Identification of LegK7 substrates on the protein microarrays. Representative images of microarrays incubated with LegK7 or LegK7(D307A). Regions containing MOB1A protein spots (printed in duplicate) are magnified. (C) Candidate substrates of LegK7 identified by high-throughput protein array screens. Representative images of spots containing candidate substrates (printed in duplicate) on the microarrays incubated with LegK7 or the LegK7(D307A) are shown. (D) Validation of MOB1A as substrate of LegK7. After in vitro thiophosphate labeling reaction, thiophosphorylated p^s-GST-MOB1A (Top), GST-MOB1A or GST (Middle), and total MBP-LegK7 (Bottom) were detected by immunoblot. The arrow marks the expected position of GST in the thiophosphorylation blot. (E) Overview of LegK7 phosphorylation sites on MOB1A detected by LC-MS/MS analysis. (F) Schematic representation of the canonical Hippo pathway in mammalian cells.

To confirm MOB1A as a direct substrate of LegK7, we performed thiophosphate labeling assays using purified glutathione S-transferase (GST)-tagged human MOB1A and maltose-binding protein (MBP)-tagged LegK7. We found that MBP-LegK7 catalyzed efficient thiophosphorylation of GST-MOB1A but not of GST, whereas MBP-LegK7(D307A) did not thiophosphorylate either protein (Figure 2D and Figure S2A). To confirm proper folding of recombinant GST-MOB1A, we included purified MST1, a known kinase of MOB1A, into our reconstitution assays. As expected, both MST1 and LegK7 catalyzed a continuous increase in thiophosphorylation on GST-MOB1A over time (Figure S2A), indicating that structural irregularities in MOB1A were not responsible for its recognition as a substrate by LegK7.

Mammalian cells encode two conserved MOB1 paralogs, MOB1A and MOB1B, which share 95% amino acid sequence identity and are considered functionally indistinguishable (Praskova et al., 2008). MOB1 is a key scaffold protein within the Hippo kinase signaling pathway. First identified in Drosophila, the conserved Hippo pathway controls cell cycle progression, cell proliferation, differentiation, and apoptosis in eukaryotes (Figure 2F) (Hergovich, 2011; Meng et al., 2016). The mammalian Ste20-like kinases 1/2 (MST1/2), orthologs of the Drosophila Hippo kinase (Hpo), phosphorylate MOB1 on threonine-12 (T12) and T35 (Praskova et al., 2008), while T74, targeted by MST2, and T181 are involved in the activation of Nuclear Dbf2-related kinase 1 (NDR1) (Hirabayashi et al., 2008). To determine the residues of MOB1A that are phosphorylated by LegK7, we performed an in vitro kinase assay using purified GST-MOB1A and MBP-LegK7 and analyzed the samples by liquid chromatography-tandem mass spectrometry (LC-MS/MS). We obtained clear signals for phosphorylation on T12 and T35 of MOB1A upon incubation with MBP-LegK7 (Figure 2E and Figure S2B–D), while no phosphorylation signal was detected for T181. Despite several attempts, the peptide containing T74 was undetectable by LC-MS/MS under any of the conditions tested here. These results demonstrate that, similar to human MST1, L. pneumophila LegK7 phosphorylates MOB1A (Figure 2D and 2E) on T12 and T35. Interestingly, T12 and T35 are conserved in MOB1A homologs from humans and amoeba, the disease host and environmental host, respectively, of L. pneumophila (Figure S2B).

LegK7 phosphorylates MOB1 and promotes LATS1 activation within mammalian cells.

We next examined whether LegK7 phosphorylates endogenous MOB1 in living cells. Green fluorescent protein (GFP)-tagged LegK7 or GFP-LegK7(D307A) were produced in transiently transfected human embryonic kidney (HEK) 293T cells, and cell lysate was probed for phosphorylation of endogenous MOB1 using antibodies directly against either phospho-T12 or phospho-T35 of MOB1. Compared to non-transfected cells or cells producing GFP, HEK293T cells producing GFP-LegK7 showed robust increases in MOB1 phosphorylation at both T12 and T35 (Figure 3A), accompanied by a noticeable decrease in the total level of MOB1. Production of the catalytic mutant GFP-LegK7(D307A) caused no significant phosphorylation or degradation of MOB1 (Figure 3A), demonstrating that phosphorylation of MOB1 depended on the kinase activity of LegK7.

Figure 3. LegK7 triggers the Hippo signaling cascade within mammalian cells.

(A) GFP or GFP-LegK7 (WT or D307A) were produced in HEK293T cells, and phosphorylation of T35 or T12 of MOB1, total MOB1, GFP (for GFP-LegK7 or GFP) and GAPDH within lysate were detected by immunoblot. Numbers represent the signal ratio (total MOB1/GAPDH) relative to GFP control (set to 100 %). (B) GFP-LegK7 or GFP (control) were produced in HEK293T cells. Phosphorylation of S909 of LATS1, total LATS1, and GAPDH within lysate were detected by immunoblot. Relative phosphorylation levels were calculated as the ratio of p-S909 LATS1/total LATS1 and are normalized to the GFP control (arbitrarily set to 1). Data shown as mean ± SD (n=3, **: p<0.01; *: p<0.05; two-tailed, unpaired t-test). (C) Human U937 macrophages were challenged with the indicated L. pneumophila strains, and macrophage lysate was probed for MOB1 phosphorylation by immunoblot. Relative phosphorylation levels were calculated as the ratio of p-T35 MOB1/total MOB1 and normalized to the control (No infection sample; set as 1). Data shown as mean ± SD (n=3, **: p<0.01; *: p<0.05; two-tailed, unpaired t-test). (D) RAW264.7 macrophages were challenged for 3 hours or 4.5 hours with the indicated L. pneumophila strains, and the level of YAP1, TAZ, and GAPDH was determined by immunoblot.

In the canonical Hippo pathway, phosphorylated MOB1 forms a complex with large tumor suppressor kinase 1 (LATS1) and promotes the activation of LATS1 by stimulating LATS1 autophosphorylation on serine-909 (Ni et al., 2015) (Figure 2F). Consistent with the observed LegK7-mediated elevation in MOB1 phosphorylation (Figure 3A), we detected increased phosphorylation on S909 of LATS1 in HEK293T cells producing GFP-LegK7 but not in cells producing GFP-LegK7(D307A) or GFP (Figure 3B). Together, these findings suggest that LegK7 is a functional mimic of MST1/2 and that LegK7 modifies MOB1 to promote LATS1 activation.

L. pneumophila LegK7 triggers Hippo pathway signaling during infection.

To determine whether L. pneumophila targets the host Hippo pathway, we examined phosphorylation of MOB1 in macrophages during infection. Challenge of human U937 macrophages with the virulent L. pneumophila strain Lp02 caused a significant increase (>2-fold) in the relative levels of phosphorylated MOB1 compared to cells challenged with the avirulent T4SS-defective strain Lp03 (Figure 3C). Macrophages infected with a legK7 mutant (Lp02DlegK7) showed only moderate MOB1 phosphorylation levels (Figure 3C). These results confirmed that the level of phosphorylated MOB1 was elevated upon L. pneumophila infection and that LegK7 was partially though not solely responsible for this effect. Increased levels of phospho-MOB1 were also detectable in Lp02-infected mouse RAW264.7 macrophages and human blood monocytes-derived macrophages (Figure S3A and S3B) with a profound reduction in the level of total MOB1. Notably, it has been shown that MOB1 degradation is mediated by the 26S proteasome in glioblastoma cells (Lignitto et al., 2013). Accordingly, we found that treatment of RAW264.7 macrophages with the proteasome inhibitor MG132 restored the levels of MOB1 during challenge with virulent L. pneumophila (Figure S3C), showing that MOB1 is targeted for proteasomal degradation during infection.

In the mammalian Hippo pathway, phosphorylated MOB1 activates the LATS1 kinase which subsequently phosphorylates the co-transcriptional regulators yes-associated protein 1 (YAP1) and its homolog transcriptional coactivator with PDZ-binding motif (TAZ). Once phosphorylated, YAP1 and TAZ are either sequestered in the cytosol by binding to 14-3-3 proteins or targeted for proteolytic degradation (Meng et al., 2016; Ni et al., 2015) (Figure 2F). Given that LegK7 activity resulted in enhanced LATS1 phosphorylation (Figure 3B), we examined the status of YAP1 and TAZ during L. pneumophila infection. While uninfected RAW264.7 macrophages and macrophages challenged with Lp03 showed similar YAP1 levels, the amount of YAP1 in Lp02-infected cells was reduced after 3 hours of infection and, even more so, 4.5 hours post infection (Figure 3D). Macrophages challenged with Lp02ΔlegK7 showed intermediate YAP1 levels at both time points (Figure 3D), indicating that LegK7 is required for efficient YAP1 degradation. Similar results were obtained in U937 macrophages upon challenge with L. pneumophila (Figure S3D). We also observed a noticeable decline in the levels of TAZ in RAW264.7 macrophages upon infection with Lp02 but not Lp03 (Figure 3D). These results demonstrate that virulent L. pneumophila promotes YAP1/TAZ degradation during infection.

L. pneumophila LegK7 modulates host gene expression.

The main outcome of signaling through the Hippo pathway are changes in gene expression due to the cytoplasmic sequestration or degradation of the co-transcriptional regulators YAP1/TAZ (Meng et al., 2016). Since challenge of host cells by L. pneumophila triggers Hippo signaling (Figure 3), we determined whether LegK7 alters the transcriptional landscape of host cells during infection. RAW264.7 macrophages were challenged with either Lp02 or Lp02ΔlegK7, and the infection was allowed to proceed for 3 hours or 5 hours, the time period when the observed difference between these two strains with respect to MOB1 phosphorylation and YAP1/TAZ degradation in macrophages was most pronounced. Total RNA was isolated from macrophages, and differences in host gene expression were determined by whole-transcriptome shotgun sequencing (RNAseq). After accounting for biological variability (see STAR Methods), 135 genes emerged that were differentially expressed (Figure 4A), of which 66 genes had a known function in regulating cell development and differentiation, metabolism, or immunity (Figure S4A). Interestingly, genes involved in immunity were overrepresented among those differentially regulated during the early time point (3 hpi), while genes that control cell development/differentiation were most abundant at the later infection stage (5 hpi) (Figure 4A).

Figure 4. LegK7 promotes intracellular bacterial replication by altering expression of PPARγ-regulated genes.

(A) Summary of genes differentially regulated in mouse RAW264.7 macrophages challenged with either Lp02 or Lp02ΔlegK7. The presence of predicted transcription factor binding motifs is indicated. Characterized genes are sorted by biological function. Uncharacterized genes are sorted by fold-change in expression. (B) Alteration of host gene expression by LegK7 is confirmed by reverse transcription (RT)-realtime PCR. Mouse RAW264.7 macrophages were challenged with Lp02ΔlegK7 complemented with a plasmid carrying wild type legK7 (pflag-legK7) or the empty plasmid backbone (pflag) for 4.5 hours, and host gene expression levels (top ten genes (bold) from (A) based on padj value) were analyzed by RT-realtime PCR. Data shown are the mean of two independent experiments. (C) Mouse RAW264.7 macrophages treated with the PPARγ antagonist GW9662 or vehicle (DMSO) at the indicated concentrations were challenged with Lp02 or Lp03, and intracellular growth of the L. pneumophila strains was determined by measuring colony forming units (CFU) via a plating assay. Data shown as mean ± SD of log2-fold change in CFU compared to initial CFU at 2 hour-post infection (n=3, *: p<0.05; two-tailed, unpaired t-test).

To confirm that the alteration of host gene expression was due to the lack of LegK7, we challenged mouse RAW264.7 macrophages with the Lp02ΔlegK7 strain that had been complemented either with a plasmid encoding legK7 (pflag-legK7) or the empty vector (pflag; control), and monitored expression of a set of host genes using reverse transcription realtime PCR (Figure 4B). We found that legK7 complementation reversed the relative changes in host gene expression, demonstrating that the alterations observed in the RNAseq analysis were indeed LegK7-dependent.

Using gene group functional profiling (Reimand et al., 2016) (see STAR Methods), we discovered binding motifs for several transcription factors, including peroxisome proliferator-activated receptor gamma (PPARγ), lymphocyte function-associated antigen 1 (LF-A1), zinc finger of the cerebellum 1 (Zic1), and myoblast determination protein (MyoD) among the differentially regulated genes (Figure 4A and Figure S4B), suggesting that those transcription factors may operate as downstream mediators in LegK7-dependent host gene expression. Both, MyoD and PPARγ were shown to physically interact with and be regulated by TAZ (Hong et al., 2005; Jeong et al., 2010), indicating a possible molecular link between them and the Hippo signaling pathway.

PPARγ activity increases macrophage susceptibility to L. pneumophila.

The transcription factor PPARγ plays a critical role in modulating gene expression during differentiation of adipocytes and in regulating inflammatory responses in macrophages (Croasdell et al., 2015; Jones et al., 2005). Upon ligand-mediated activation, PPARγ forms a heterodimeric complex with other transcription factors, such as retinoid X receptor α (RXRα), to regulate transcription of target genes (Croasdell et al., 2015). Unlike LF-A1, Zic1, and MyoD, PPARγ has been the target of therapeutic intervention methods, and small molecule inhibitors are readily available. Since PPARγ-regulated genes were differentially regulated during L. pneumophila infection in a LegK7-dependent manner and the role of PPARγ in L. pneumophila pathogenesis was unclear, we pharmacologically blocked PPARγ activity in host cells and determined the effect on intracellular replication of L. pneumophila. Upon treatment of RAW267.4 macrophages with the PPARγ antagonist GW9662, intracellular replication of Lp02 was significantly reduced at both 48-hour and 72-hour post-infection compared to that in vehicle-treated host cells (Figure 4C). GW9662 treatments had no effect on the numbers of intracellular Lp03, showing that the reduction in colony-forming units of Lp02 had not been caused by general impairment of macrophages or their loss from the cell monolayer. These findings indicate that PPARγ-regulated genes are important for maximal intracellular replication of virulent L. pneumophila, and suggest that LegK7-mediated Hippo pathway signaling participates in the manipulation of PPARγ.

Discussion

In this study, we provide evidence that the Hippo signaling pathway, in addition to its well-described role in regulating organ growth and development in metazoans, plays an unexpected role during L. pneumophila pathogenesis. We found that the L. pneumophila effector LegK7, like the Hippo kinase MST1/2, directly phosphorylates T12 and T35 on MOB1, and that LegK7 triggers a signaling cascasde that alters the transcriptional landscape of host cells to improve conditions for intracellular L. pneumophila growth.

Given the lack of primary sequence similarity to eukaryotic kinases, the detection of a kinase domain within LegK7 by in silico analyses was unexpected (Figure S1A). Using a thiophosphorylation-based labeling approach, we subsequently confirmed the existence of kinase activity within LegK7 (Figure 1C, 1D) and, in addition, discovered MOB1 as a target of LegK7 (Figure 2). Similar to its eukaryotic counterpart Hpo/MST1, LegK7 phosphorylated MOB1 on T12 and T35 (Figure 2E), positions critical for MOB1-mediated LATS1 recruitment and activation (Ni et al., 2015). Accordingly, we detected activation of LATS1 within mammalian cells upon MOB1 phosphorylation by LegK7 (Figure 3B), as well as the degradation of YAP1/TAZ, downstream targets of LATS1 (Figure 3D). These data demonstrated that L. pneumophila, by translocating LegK7 into host cells, can hijack the Hippo signaling cascade during infection, and provided a remarkable example of molecular mimicry aimed at exploiting a signaling pathway that is hardwired into the developmental program of all eukaryotic cells.

Our transcriptome analysis revealed that a define set of genes that are under the control of several transcription factors, namely PPARγ, MyoD, Zic1 and LF-A1, were differentially regulated in a LegK7-dependent manner, suggesting that these transcription factors are downstream mediators of the LegK7-mediated Hippo signaling cascade. In fact, in mesenchymal stem cells TAZ was shown to interact with PPARγ and negatively control PPARγ-dependent gene transcription (Hong et al., 2005), while TAZ was reported to regulate MyoD-dependent gene expression in myoblasts through intraction with MyoD (Jeong et al., 2010). Thus, it is reasonable to assume that LegK7 manipulates the activity of these transcription factors, at least in part, by hijacking the Hippo pathway and reducing the cellular levels of YAP1/TAZ (Figure 3D). Notably, it cannot be excluded that the regions of unknown function flanking the LegK7 kinase domain contributed, either in a Hippo/YAP/TAZ-dependent or -independent manner, to the differential regulation of host gene expression. Future studies will need to determine what the molecular functions of the non-kinase domains of LegK7 are and how they affect host cells during infection.

We also showed that pharmacological interference with PPARγ activation in infected macrophages reduced intracellular replication of L. pneumophila (Figure 4C). Notably, the numbers of intracellular Lp03 remained unaffected by GW9662 treatment, suggesting that pharmacological inhibition of PPARγ did not cause an overall decrease in host cell viability. Instead, these data indicate that manipulation of PPARγ is an important step in the virulence strategy of this intracellular pathogen. Our findings are in line with a recent report demonstrating that activation of PPARγ plays a similarly important role in promoting Brucella abortus infection (Xavier et al., 2013). It remains to be determined which of the PPARγ-controlled genes that were differentially regulated in response to LegK7 impact, either positively or negatively, intracellular L. pneumophila growth, and whether other host pathways targeted by L. pneumophila contribute to the manipulation of PPARγ. It is worth mentioning though that many of the genes differentially regulated at the early stage of infection were related to immune functions, a time point in the infection when L. pneumophila has to actively antagonize the host immune response, while genes involved in development and differentiation were overrepresented at a later time point (5 hpi), the infection stage when intracellular L. pneumophila prepare for replication.

Our data provide strong evidence that the Hippo pathway, in addition to its well-established functions in developmental biology and cancer, may play an underappreciated role in innate immunity. In fact, mutations in MST1 were shown to be associated with recurrent infections in humans (Abdollahpour et al., 2012), and mice with a conditional MST1/2 knockout in hematopoietic cells or Drosophila with Hpo knockdown showed decreased anti-bacterial activity (Geng et al., 2015; Liu et al., 2016). We propose that LegK7, by functionally mimicking Hpo/MST1, antagonizes host innate immune processes, thereby creating conditions that favor intracellular replication of L. pneumophila. The choice of MOB1 as LegK7 target is supported by the fact that this protein is the most conserved of all core components of the Hippo pathway (Figure S4C), such that targeting MOB1 instead of less well-conserved downstream signaling components allows L. pneumophila to proliferate within a wider range of host species, including humans.

STAR Methods

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Matthias Machner (machnerm@nih.gov)

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Bacterial strains

Bacterial strains used in this study are listed in the Key Resource Table. Escherichia coli strains were cultured in LB (Luria-Bertani) or 2xYT media at 37 °C. L. pneumophila strains were patched on CAYET agar plates (2 g/L activated charcoal, 10 g/L ACES, 10 g/L yeast extract, 400 μg/mL cysteine, 100 μg/mL thymidine and 135 μg/mL ferric nitrate) for two days at 37 °C. Bacterial patches were scraped off from CAYET plates and cultured overnight at 37 °C in AYE medium (10 g/L ACES, 10 g/L yeast extract) supplemented with 400 μg/mL cysteine, 100 μg/mL thymidine and 135 μg/mL ferric nitrate. For L. pneumophila strains carrying pJB908 plasmid derivatives, the bacteria were patched on CAYE agar plates (2 g/L activated charcoal, 10 g/L ACES, 10 g/L yeast extract, 400 μg/mL cysteine, and 135 μg/mL ferric nitrate) for two days at 37 °C. Patches were scraped off from the plates and cultured overnight at 37 °C in AYE medium supplemented with 400 μg/mL cysteine, and 135 μg/mL ferric nitrate. For L. pneumophila strains carrying pXDC61 or pflag plasmid derivatives, the bacteria were patched on CAYET agar plates containing 5 μg/ml choloramphenicol for two days at 37 °C. Patches were scraped off from the plates and cultured overnight at 37 °C in AYE medium supplemented with 400 μg/mL cysteine, 100 μg/mL thymidine, 135 μg/mL ferric nitrate and 5 μg/ml choloramphenicol.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-thiophosphate ester	Abcam	ab92570; RRID:AB 10562142
Anti-maltose binding protein (MBP)	New England BioLabs	E8032S
Anti-MOB1	Cell Signaling	#13730
Anti-phospho-T35 MOB1	Cell Signaling	#8699S; RRID:AB_11139998
Anti-phospho-T12 MOB1	Cell Signaling	#8843; RRID:AB 10971644
Anti-LATS1	Cell Signaling	#3477; RRID:AB 2133513
Anti-phospho-S909 LATS1	Cell Signaling	#9157; RRID:AB 2133515
Anti-YAP1	Santa Cruz Biotechnology	sc-101199; RRID:AB 1131430
Bacterial Strains
BL21(DE3) E. coli B F− dcm ompT hsdS(rB− mB−) gal lon λ (DE3[lacl lacUV5-T7 gene 1 ind1 sam7 nin5])	Agilent	Cat # 200131
L. pneumophila-Philadelphia-1, Lp02 (thyAQ33/stop, rpsLK88R, hsdR−)	(Berger and Isberg, 1993)	N/A
L. pneumophila-Philadelphia-1, Lp03 (thyAS167frameshift, rpsLK88R, hsdR−, dotAQ188/stop)	(Berger and Isberg, 1993)	N/A
L. pneumophila-Philadelphia-1, Lp02ΔlegK7 (thyAQ33/stop, rpsLK88R, hsdR−, ΔlegK7)	This study	N/A
Chemicals, Peptides, and Recombinant Proteins
ATPγS (Adenosine 5’-(3-thiotriphosphate) tetralithium salt)	Abcam	ab138911
p-nitrobenzyl mesylate (PNBM)	Abcam	ab138910
GW9662	Santa Cruz Biotechnology	sc-202641
MG-132	Sigma-Aldrich	M7449
myelin basic protein (bovine)	Sigma-Aldrich	M1891
Digitonin	Calbiochem	Cat. 300410
Protease inhibitor cocktail (cOmplete, EDTA-free)	Roche	Cat. 11873580001
PowerUp™ SYBR™ green master mix	Applied Biosystems	A25742
SuperScript™ VILO™ cDNA synthesis	Invitrogen	11754050
Protoarray® human protein microarrays	Invitrogen	N/A
LiveBLAzer FRET-B/G loading kit with CCF4-AM	Invitrogen	K1095
RNeasy® mini kit	Qiagen	74104
Recombinant GST-MST1	Invitrogen	PV3854
Deposited Data
RNA sequencing fastq files	Sequence Read Archive (SRA) https://www.ncbi.nlm.nih.gov/sra/	Accession number: SRP131889
Experimental Models: Cell Lines
RAW264.7 mouse macrophages	ATCC	TIB-71
U937 human macrophages	ATCC	CRL1593.2
HEK293T Human embryonic kidney cells	ATCC	CRL-3216
Oligonucleotides
Sequences of oligonucleotides are listed in supplementary table 2	This manuscript	N/A
Recombinant DNA
pXDC61	(Charpentier et al., 2009)	N/A
pXDC61-ralF	(Charpentier et al., 2009)	N/A
pXDC61-legK7	This manuscript	N/A
pDonor221-legK7	This manuscript	N/A
pDEST17	ThermoFisher Scientific	11803012
pDEST17-legK7	This manuscript	N/A
pDEST17-legK7D307A	This manuscript	N/A
pMal-c5x	New England BioLabs	N8108S
pMal-legK7	This manuscript	N/A
pMal-legK7D307A	This manuscript	N/A
pGEX6P1	GE Healthcare	GE28-9546-48
pGEX6P1-MOB1A	This manuscript	N/A
pcDNA6.2-N-EmGFP-DEST	ThermoFisher Scientific	V35620
pcDNA6.2-N-EmGFP-legK7	This manuscript	N/A
pcDNA6.2-N-EmGFP-legK7D307A	This manuscript	N/A
pSR47S	(Merriam et al., 1997)	N/A
pSR47S-ΔlegK7	This paper	N/A
pJB908	(Sexton et al., 2004)	N/A
pflag	This manuscript	N/A
pflag-legK7	This manuscript	N/A
Software and Algorithms
Prism	GraphPad software	https://www.graphpad.com/
Invitrogen Prospector	Invitrogen	https://www.thermofisher.com/
Genepix Pro 6	Genepix	http://mdc.custhelp.com/
Image Lab	BioRad	http://www.bio-rad.com

Human and murine cell lines

Human (male) U937 cells were cultured in RPMI1640 medium supplemented with 10% FBS at 37 °C under 5% CO₂ in humidified incubators. Mouse (male) RAW264.7 cells and human (female) embryonic kidney 293T (HEK293T) cells were cultured in DMEM medium supplemented with 10% FBS at 37 °C under 5% CO₂ in humidified incubators.

Human primary cells

Human blood monocytes-derived primary macrophages were generously provided by Drs. Yu-Ting Su and Jing Wu (Neuro-oncology branch, NCI). Human monocytes elutriated from blood of healthy donors were isolated following an Institutional Review Board-approved protocol from the NIH Blood Bank (Bethesda, MD). Human monocytes were spun down, washed with 10 ml of ACK lysis buffer (Invitrogen), resuspended in RPMI1640 medium with 2 mM L-glutamine and 10% FBS, and incubated at 37 °C overnight. The human monocytes were differentiated to macrophages by seeding 1×10⁶ monocytes/well in 12-well plates in IMDM (Iscove’s modified Dulbeccos’s medium) with 10 % heat-inactivated FBS and 20 ng/ml of M-CSF for 5 days at 37 °C under 5% CO₂ in humidified incubators. Blood monocytes were isolated from adult male donors (age 57–61 year-old; one African American, one Asian, and one Caucasian) and each independent experiment used blood monocytes from one single donor.

METHOD DETAILS

Plasmids, reagents and antibodies

Plasmids and oligonucleotides used in this study are listed in the Key Resource Table.

Recombinant protein production and purification

For the purification of His6-LegK7 and MBP-LegK7 proteins, E. coli strain BL21(DE3) carrying plasmids encoding the recombinant proteins were cultured in 2xYT medium at 37 °C for 2–3 hours, recombinant protein production was induced by adding 500 μM IPTG (Isopropyl β-D-1-thiogalactopyranoside). After 1.5–2 hours at 37 °C, the bacterial cells were pelleted, resuspended in TALON lysis buffer (1xPBS pH 7.4, 1 mM b-mercaptoethanol and 1x protease inhibitor cocktail) for His6-LegK7 or MBP lysis buffer (20 mM Tris, pH 7.4, 200 mM NaCl, 1 mM EDTA, 1 mM β-mercaptoethanol, and 1x protease inhibitor cocktail) for MBP-LegK7, and lysed using a Microfluidics (M110P) cell breaker. The recombinant proteins were affinity-purified from cleared lysate using TALON metal affinity beads (Clontech) or amylose resin (New England BioLabs) according to manufacturer guidelines. For GST and GST-MOB1A proteins, E. coli strains BL21(DE3) carrying plasmids encoding GST or GST-MOB1A were cultured in LB medium at 37 °C for 2–3 hours, production of the recombinant proteins was induced with 500 μM IPTG, and cells were grown overnight at 20 °C. Bacterial cells were pelleted, resuspended in GST lysis buffer (1x PBS pH 7.4, 1 mM β-mercaptoethanol, and 1x protease inhibitor cocktail), and mechanically lysed using a Microfluidics cell breaker. GST-tagged proteins were purified from the lysate using glutathione sepharose resin (GE Healthcare) according to the manufacturer’s manual. All purified proteins were dialyzed in the kinase assay buffer (10 mM HEPES pH 7.3, 150 mM NaCl) by using Snakeskin dialysis tubing membranes (3,500 Da pore size, ThermoFisher Scientific) prior to use in the in vitro kinase assay.

In vitro kinase assay

0.4 μg of purified MBP-LegK7 and 2 μg of purified GST or GST-MOB1A (wild-type or mutants) were mixed in 10 mM HEPES pH7.3, 150 mM NaCl, 10 mM MgCl_2, and 1 mM ATPγS, the volume was adjusted to 30 μl, and incubated at 30 °C. After 45 minutes, 1.5 μL of 50 mM PNBM (final concentration 2.5 mM) was added, and the alkylation reaction on thiophosphorylated proteins was allowed to occur for 2 hours at room temperature. Alkylation was stopped by adding 31.5 μl of 2x SDS protein sample buffer (100 mM Tris-HCl pH 6.8, 4% SDS, 0.2% bromophenol blue, 20% glycerol and 200 mM β–mercaptoenthanol), and the samples were heated at 95 °C for 10 minutes. Proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes, and protein thiophosphorylation was detected using an anti-thiophosphate ester-specific primary antibody and a HRP-conjugated secondary antibody. Blots were developed using an enhanced chemiluminescence (ECL) assay, and images were acquired by X-ray films or digitally using a ChemiDoc™-MP imaging system (BioRad). All blots are representatives of at least three independent experiments.

Kinase substrate screen on protein microarrays

Protoarray® human microarrays (Invitrogen) were incubated in blocking buffer containing 10 mM HEPES pH 7.3, 150 mM NaCl, and 1% BSA for 1 hour at 4°C. A kinase reaction mixture containing 1 μg of purified His6-LegK7 or His6-LegK7(D307A) in 10 mM HEPES pH 7.3, 150 mM NaCl, 10 mM MgCl_2, and 1 mM ATPγS (final concentrations) was applied to each microarray, and the microarrays were incubated for 45 minutes at 30 °C. Microarrays were then rinsed and incubated with alkylation mixture (10 mM HEPES pH 7.3, 150 mM NaCl, 2.5 mM PNBM) overnight at 4 °C. Microarrays were washed with detection blocking buffer (50 mM HEPES pH 7.5, 200 mM NaCl, 0.08% Triton X-100, 25% glycerol, 20 mM reduced glutathione, 1x Synthetic block (Invitrogen), and 1 mM DTT) and blocked with the detection blocking buffer for 1 hour at 4 °C. Thiophosphorylated proteins on the microarray were detected by incubation with anti-thiophoshpate ester primary antibody (1:1,000 dilution) in detection washing buffer (1x PBS, 1x Synthetic block, 0.1% Tween 20), followed by Alexa® Red Fluor 647-conjugated secondary antibody (1:2,000 dilution). Images of the microarrays were acquired in a GenePix 4200AL microarray scanner (Molecular Devices). The fluorescence intensity of the protein spots on the microarrays was analyzed with the GenePix Pro 6 software (GenePix) and Prospector® (Invitrogen) software, and Z scores (# of standard deviations higher than the mean spot intensity) for each protein spot were calculated and ranked as described (Ramani et al., 2012). Protein spots with a confidence (Z) score difference ratio of ≥0.3 [(Z score of protein spot on microarray incubated with LegK7(WT) - Z score of protein spot on microarray incubated with LegK7(D307A))/ Z score of protein spot on microarray incubated with LegK7(WT)] and an adjusted signal ratio score of ≥1.5 [adjusted signal intensity of protein spot on microarray incubated with LegK7(WT)/adjusted signal intensity of protein spot on microarray incubated with LegK7(D307A)] were selected as potential substrates. Z score difference ratio and adjusted signal ratio presented in Figure 2C are the mean from two independent experiments.

Identifying phosphorylation sites on MOB1A

2.5 μg of purified MBP-LegK7 and 100 μg of purified GST-MOB1A were mixed in 10 mM HEPES pH7.3, 150 mM NaCl, 10 mM MgCl_2, and 1 mM ATP, the volume was adjusted to 60 μl, and the reaction was incubated at 30 °C for 1 hour. Proteins were precipitated in 10% trichoroacetic acid (TCA) and the pellet was washed with acetone. Precipitated protein pellets were dissolved in 6 M Urea and digested by trypsin or chymotrypsin (at enzyme:substrate ratio=1:30) overnight at 30 °C. After digestion, peptides were acidified by adding formic acid and purified by Ziptips (Millipore). 4 μg of purified peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) to identify phosphorylation sites on MOB1A. LC-MS/MS was performed using a Dionex UltiMate 3000 rapid separation nano UHPLC system (Thermo Scientific) coupled online to an Orbitrap Fusion Lumos tribrid mass spectrometer (Thermo). Protein digest was first loaded onto a nano trap column (Acclaim PepMap100 C18, 3 μm, 100Å, 75 μm i.d. × 2 cm, Thermo), and then separated on a reversed-phase EASY-Spray analytical column (PepMap RSLC C18, 2 μm, 75 μm i.d. × 50 cm, Thermo) using a linear gradient of 4–32% B (buffer A: 0.1% formic acid in water; buffer B: 0.1% formic acid in acetonitrile) for 100 min. The mass spectrometer was equipped with a nano EASY-Spray ionization source, and eluted peptides were brought into gas-phase ions by electrospray ionization and analyzed in the orbitrap. High-resolution survey MS scans and HCD fragment MS/MS spectra were acquired in a data dependent manner with a cycle time of 3 s. Dynamic exclusion was enabled. Raw data files generated from LC-MS/MS were analyzed using a Proteome Discoverer v2.2 software package (Thermo) and the Sequest HT search engine. Fusion protein sequences were appended to a SwissProt human protein fasta database. The following database search criteria were set to: enzyme, trypsin or chymotrypsin; max miscleavages, 2; variable modifications, phosphorylation (STY), oxidation (M), deamidation (NQ); peptide precursor mass tolerance, 20 ppm; MS/MS fragment mass tolerance, 0.03 Da. Peptide-spectrum matches (PSMs) were filtered to achieve an estimated false discovery rate (FDR) of 1% based on a target-decoy database search strategy. The relative abundance of a phosphorylated peptide in different samples was estimated by its peak areas.

Mammalian cell transfection

HEK293T cells were seeded in 6-well plates (4×10⁵ cells/well) 24 hours prior to transfection. 3 μg of pcDNA6.2 plasmids encoding GFP or GFP-LegK7 were mixed with 6 μl of Lipofectamine 2000 transfection reagent (Invitrogen) and added to each well of HEK293T cells. After 24 hours, cell monolayers were rinsed, cells were lysed by adding 300 μl of 1x SDS protein sample buffer, and lysate was heated at 95 °C for 10 minutes. Protein levels and protein phosphorylation were analyzed by immunoblotting with specific antibodies as described in the main text. Blots shown are representatives of five independent experiments.

Macrophage infection

Human U937 macrophages were cultured in RPMI1640 medium containing 10% FBS, and differentiated in the presence of 10 ng/mL phorbol 12-myristate 13-acetate (TPA, Sigma-Aldrich) for 48 hours. The TPA-containing media was then replaced with fresh RPMI1640 containing 10% FBS, and the differentiated U937 cells were cultured for an additional 24 hours prior to being challenged with L. pneumophila. Mouse RAW264.7 macrophages were cultured in DMEM containing 10% FBS and reseeded 48 hours prior to challenge with L. pneumophila. Human blood monocytes-derived primary macrophages were prepared as described in the “Experimental model and subject detail” section prior to infection. Overnight cultures of L. pneumophila were diluted in cell culture media (RPMI or DMEM) and added to the macrophage monolayers at a multiplicity of infection (MOI) of 25. The culture plates were then centrifuged at 200× g for 5 minutes at room temperature. After incubating at 37 °C for the indicated time points, macrophages were washed twice with cold 1x PBS and harvested by scraping. Macrophages were pelleted, the supernatant was removed, and cells were resuspended in 1x SDS protein sample buffer and heated at 95 °C for 10 minutes. Protein level and phosphorylation of specific proteins in the lysates were analyzed by immunoblot using antigen-specific antibodies as described in the main text. Blots shown are representatives of three independent experiments.

FRET translocation assay

Human U937 macrophages were seeded in 96-well plates (1.25×10⁵ cells/well) and differentiated as described in the previous section. L. pneumophila strains containing pXDC61 that encodes β-lactamase fused to the N-terminus of either LegK7 or RalF were cultured in AYE medium supplemented with 5 μg/mL chloramphenicol, 400 μg/mL cysteine, 100 μg/mL thymidine and 135 μg/mL ferric nitrate. β-lactamase fusion protein production was induced with 500 μM IPTG. After overnight growth, the L. pneumophila strains were diluted in RPMI1640 medium, and 100 μL of the suspension was added to macrophages at an MOI of 100. Macrophage culture plates were centrifuged at 200×g for 5 minutes at room temperature and incubated for an additional 3 hours at 37 °C. Translocation of β-lactamase fusion proteins into macrophages was detected by CCF4-AM-based fluorescence resonance energy transfer (FRET) assay using the LiveBLAzer FRET-B/G loading kit (ThermoFisher Scientific) according to the manufacturer’s manual. CCF4-AM is cleaved in the presence of β-lactamase, resulting in a shift in fluorescence emission light from 530 nm (green) to 460 nm (blue). Images are representatives of two independent experiments.

Chromosomal gene deletion in L. pneumophila

legK7 was deleted in frame from the L. pneumophila chromosome (Lp02DlegK7) by allelic exchange as described (Merriam et al., 1997). Briefly, the allelic exchange plasmid pSR47s carrying ~500 bp-long upstream and downstream flanking DNA regions of legK7 (pSR47s-DlegK7) was delivered into Lp02 by tri-parental mating, and Lp02 recipients were enriched on CAYET agar plates supplemented with 20 μg/mL kanamycin and 50 μg/mL streptomycin. Emerging colonies were cultured without kanamycin overnight and then re-plated on CAYET plates supplemented with 5% sucrose. Sucrose-insensitive colonies were re-streaked onto CAYET agar plates to obtain single colonies. Chromosomal deletion of legK7 in the resulting single colonies was confirmed by PCR analysis using primers to amplify the flanking regions of the legK7 gene (see Key Resource Table).

RNA sequencing and bioinformatic analysis

Mouse RAW264.7 macrophages seeded in 6-well plates (1.5×10⁶ cells/well, 2 days prior to challenge) were challenged at an MOI of 12.5 with Lp02 or Lp02DlegK7 and incubated at 37 °C for the indicated time points (3 or 5 hours). Macrophage monolayers were rinsed twice with cold 1x PBS, cells were harvested by scraping, and pelleted by centrifugation at 200×g for 5 minutes at 4 °C. Total macrophage RNA was extracted using a RNeasy® mini kit (Qiagen) with DNase treatment according to the manufacturer’s manual. RNA samples of four replicates for each condition were processed by poly-A enrichment, and transcriptome libraries were generated using Illumina’s TruSeq Stranded Total RNA Library Prep Kit and sequenced by Illumina HiSeq systems. RNA-STAR was used to align reads against the mouse mm10 build. Quantitation of transcripts was performed with the subread package featureCounts. DESeq2 was used to analyze gene expression. Gene expression of macrophages challenged with Lp02DlegK7 was compared to that of macrophages challenged with Lp02. Two approaches were used to select and analyze differentially expressed genes. In the first approach, all differentially regulated genes (up-regulated or down-regulated > 2-fold) with a p-value < 0.01 were selected from the 3-hour and 5-hour infection samples (Figure 4A). Of the 135 genes, 65 have been characterized, and functional categories of these genes were assigned by the authors (Figure S4A). The 135 differentially regulated genes were then analyzed using the gProfiler web-tool (https://biit.cs.ut.ee/gprofiler/) (Reimand et al., 2016) to identify regulatory DNA motifs enriched in the gene set with statistical domain size restricted to annotated genes only (Figure S4B). In the second approach, differentially regulated genes (up-regulated or down-regulated > 2-fold) with a p-adjust value < 0.1 were selected from the 3-hour and 5-hour infection samples, and the emerging 13 genes were analyzed by gProfiler with the statistical domain size set as “all known genes”. Regulatory DNA motifs identified by the second approach are listed in Supplementary Excel spreadsheets. Among the identified motifs, PPAR regulatory elements were most abundant with three types of PPARγ-RXRα motifs, two types of PPAR motifs, and three types of RXRα motifs. The 13 genes selected by the second approach overlapped with those from the first approach, and the PPARγ-RXRα motif M02262_1 was identified as being enriched in both approaches. The complete lists of differentially regulated genes and enriched motifs identified by gProfiler analysis from both approaches are shown in supplementary table 1.

Reverse transcription-Realtime PCR

L. pneumophila Lp02ΔlegK7 strains were transformed with a plasmid containing legK7 (pflag-legK7) or the empty plasmid (pflag), and production of LegK7 was induced by adding IPTG (200 μM final concentration) overnight in AYE liquid culture. Mouse RAW264.7 macrophages seeded in 6-well plates (1.5×10⁶ cells/well, 2 days prior to challenge) were challenged at an MOI of 25 with Lp02DlegK7/pflag-legK7 or Lp02DlegK7/pflag and incubated at 37 °C for 4.5 hours. Macrophage monolayers were rinsed twice with cold 1x PBS, cells were harvested by scraping, and pelleted by centrifugation at 200×g for 5 minutes at 4 °C. Total macrophage RNA was extracted using a RNeasyÒ mini kit (Qiagen) with DNase treatment according to the manufacturer’s manual, and RNA expression level of host genes was determined by reverse transcription-realtime PCR. 2 μg of total RNA from the legK7 complementation experiment was reverse transcribed to complementary DNA (cDNA) using SuperScript™ VILO™ cDNA synthesis kit (Invitrogen) following the manufacturer’s manual. The cDNA templates were diluted and mixed with PowerUp™ SYBR™ green master mix (Applied Biosystems) and primers that specifically amplify the top 10 differentially regulated genes from the RNA sequencing analysis based on the padj value (Dbn1, Trim69, Ankrd55, Srcin1, Tas1r3, Bpifb6, Gprc5a, Pygm, Tmem54, and Lrfn1) and mouse 18S rRNA gene (as internal control). Gene expression level was determined by realtime PCR using QuantStudioÔ 7 Flex System (Applied Biosystems). Changes of gene expression are presented as mean log2-fold change from two independent experiments.

L. pneumophila replication in macrophages

Human U937 macrophages were seeded in 24-well plates (5×10⁵ cells/well) and differentiated as described in the previous section. Overnight cultures of L. pneumophila were diluted in RPMI medium and added to the macrophage monolayers at an MOI of 0.05. The culture plates were then centrifuged at 200×g for 5 minutes at room temperature. After incubation at 37 °C for 2 hours, macrophages were washed three times with 1x PBS, and fresh RPMI medium supplemented with 10% FBS and 100 μg/mL thymidine was added to the wells. U937 macrophages were cultured for an additional 24, 48, or 72 hours at 37°C. Mouse RAW264.7 macrophages were seeded at 1×10⁵ cells/well in 24-well plates and cultured for 2 days prior to challenge. The macrophages were treated with the indicated concentrations of GW9662 or DMSO (vehicle control), and challenged with L. pneumophila Lp02 (pJB908) or Lp03/pJB908 at an MOI of 1. After incubation for 2 hours at 37 °C, the macrophages were washed 3 times with 1x PBS, and fresh DMEM with the indicated concentrations of GW9662 and 10% FBS was added to the wells. The macrophages were further cultured for an additional 72 hours to allow intracellular replication of L. pneumophila. Digitonin (0.01%) was added to the wells to permeablize the plasma membrane of the macrophages, and bacteria were harvested and serial dilutions were spot-plated on CYET agar plates. The number of L. pneumophila in each well was determined by counting colony-forming units and presented as the mean ± SD of 3 independent experiments. p-values were calculated by Student’s t-test (two-tailed, unpaired).

For the experiments shown, no randomization was employed, no data were excluded, and no blinding was applied.

QUANTIFICATION AND STATISTICAL ANALYSIS

Band intensities in immunoblots were quantitated using Image Lab software (BioRad), and the ratio of the band intensities was presented as the mean ± SD of three independent experiments (n=3) unless otherwise specified. p-values were calculated by Student’s t-test (two-tailed, unpaired) using Microsoft Excel and Graphpad Prism softwares, and p-value<0.05 were considered as statistical significant. For L. pneumophila intracellular replication assays in mouse RAW264.7 macrophages, data presented are the mean ± SD log2-fold change from three independent experiments (n=3) shwoing CFU at the indicated time points relative to initial CFU at 2 hour-post infection. p-values were calculated by Student’s t-test (two-tailed, unpaired) using Microsoft Excel and Graphpad Prism software, and p-values<0.05 were considered as statistical significant.

DATA AND SOFTWARE AVAILABILITY

Data generated by RNA sequencing have been deposited in the Sequence Read Archive (SRA; https://www.ncbi.nlm.nih.gov/sra/) under SRA accession number SRP131889. Source information for softwares used in this study is provided in the Key Resource Table. No in-house built software was used in this study.

Supplementary Material

Supplementary Table 1

Table S1. Lists of differentially regulated genes in macrophages infected with L. pneumophila and of enriched transcription factor binding motifs within the gene sets (related to STAR Methods) are provided online as supplementary table 1.

Supplementary Table 2

Table S2. Oligonucleotide sequences (related to Key Resources Table) are provided online as supplementary table 2.

1

Figure S1A. Comparative analysis of kinase predicted secondary structures (related to Figure 1).

Profile Hidden Markov Model-based protein homology search (HHPred, https://toolkit.tuebingen.mpg.de/hhpred/) predicted that L. pneumophila LegK7 kinase domain shares high secondary structural homology with more than 250 kinases (probability > 99.9 %). The predicted secondary structure alignment of PKAcα (mouse cAMP-dependent protein kinase catalytic subunit α) and LegK7 is shown as a representative example. Beta strands are shown as blue arrows, helices as magenta jagged lines, and loops as straight lines. The catalytic motif is highlighted in yellow. Red letters indicate identical residues. Numbers indicate amino acid positions.

Figure S1B. LegK7 is not essential for L. pneumophila growth in human U937 macrophages (related to Figure 1). Human U937 macrophages were challenged with the virulent, T4SS-competent L. pneumophila strain, Lp02, the non-virulent, T4SS-deficient L. pneumophila strain, Lp03, or a Lp02 strain with chromosomal legK7 deletion, Lp02ΔlegK7. Intracellular bacterial replication over 24, 48 and 72 hours of incubation was determined by measuring colony forming units in a plating assay. Data are representative of two independent experiments.

Figure S1C. LegK7 orthologs present in Legionella species (related to Figure 1).

A Basic Local Alignment Search Tool (Blast) analysis (https://blast.ncbi.nlm.nih.gov/Blast.cgi) identified a total of 37 LegK7 orthologs among 32 of 49 sequenced Legionella species.

Figure S1D. Sequence alignment of LegK7 and orthologs from four representative Legionella species known to cause Legionnaires’ diseases (related to Figure 1). L. pneumophila: YP_095941.1, L. longbeachae: WP_003636120.1, L. anisa: KTC76241.1, L. bozemanae: WP_064115092.1 and L. dumoffii: WP_010654035.1. Sequences were aligned by ClustalOmega and processed by ESPript 3.0. Identical residues are white on red background, similar residues are red on white background. Numbers indicate amino acid positions in LegK7; Black lines indicate the region of the predicted kinase domain in LegK7. Asterisks indicate catalytic residues in the kinase motif.

Figure S1E. Sequence alignment of the catalytic motifs of LegK7 and its orthologs (related to Figure 1). Kinase catalytic motifs (DxKxxN; single letter code wherein x is any residue) of LegK7 (YP_095941) and its orthologs from the indicated Legionella species were aligned. Sequences were aligned using ClustalOmega and processed by ESPript 3.0. Identical residues are white on a red background. Similar residues are red on a white background. Numbers indicate amino acid positions. Asterisks indicate catalytic residues in the kinase motif.

Figure S2. Phosphorylation sites and sequence conservation of human MOB1A (related to Figure 2). (A) LegK7- and MST1-catalyzed phosphorylation of MOB1 in vitro. Purified GST-MOB1A (1.6 μM) was incubated with either MBP-LegK7 (40 nM) or GST-MST1 (40 nM), and thiophosphorylated ps-GST-MOB1A or total protein levels were detected by immunoblot. (B) Alignment of MOB1 sequences. Human MOB1A (NP_060691.2) and MOB1B (NP_775739.1), Drosophila melanogaster Mats (NP_651041.3), and MOB1 homologs from protozoa (Dictyostelium discoideum (DdMobA, XP_647813.1), Acanthamoeba castellanii (AcMob4B, XP_004356628.1)) were aligned using ClustalOmega and processed by ESPript 3.0. Identical residues are white on a red background, similar residues are red on a white background. Numbers indicate amino acid positions in MOB1A. Asterisks indicate amino acid residues that are phosphorylated by LegK7. Green lines indicate the peptide sequence detected by the liquid chromatography-tandem mass spectrometry (LC-MS/MS, total coverage: 88.4%). (C and D) Phosphorylation sites on GST-MOB1A detected by LC-MS/MS analysis. Purified GST-MOB1A was incubated with or without MBP-LegK7, and an in vitro kinase reaction using ATP as phosphate donor was performed. The LC trace for peptides containing T12 and T35 is presented. The peptides containing un-phosphorylated and phosphorylated T12 in (C) showed the same retention time. Phosphorylated T12 was only detected in the GST-MOB1A+MBP-LegK7 sample.

Figure S3. L. pneumophila manipulates components of the Hippo pathway during infection (Related to Figure 3). (A) MOB1 is phosphorylated during L. pneumophila infection in mouse macrophages. Mouse RAW264.7 macrophages were challenged with the indicated L. pneumophila strains for 4.5 hours. Phosphorylation of MOB1 and total MOB1 in the macrophage lysate was determined by immunoblot using antibodies against phospho-T35 MOB1 (Top) or MOB1 (Middle) or GAPDH (Bottom). *; the GAPDH blot is the same as shown in Figure 3D (GAPDH blot for YAP1 4.5 hr) because the samples originated from the same experiment and the immunoblot was performed at the same time and subsequently cropped. The graph shows the ratio of p-T35 MOB1/total MOB1 normalized to the control (No infection sample; arbitrarily set as 1). Data shown as mean ± SD (n=3, **: p<0.01; *: p<0.05; two-tailed, unpaired t-test). (B) MOB1 is phosphorylated during L. pneumophila infection in human blood monocytes-derived macrophages. Human blood monocytes-derived macrophages were challenged with the indicated L. pneumophila strains for 4 hours. Phosphorylation of MOB1 and total MOB1 in the macrophage lysate was determined by immunoblot using antibodies against phospho-T35 MOB1 (Top), MOB1 (Middle) or GAPDH (Bottom). The graph shows the ratio of p-T35 MOB1/total MOB1 normalized to the control (No infection sample; arbitrarily set as 1). Data shown as mean ± SD (n=3, **: p<0.01; *: p<0.05; two-tailed, unpaired t-test). (C) L. pneumophila promotes MOB1 degradation via the host cell 26S proteasome. RAW264.7 macrophages were challenged with the indicated L. pneumophila strains and treated for 4.5 hours with the proteasome inhibitor MG132 at the indicated concentrations. MOB1 and GAPDH were detected in the macrophage lysate using specific antibodies. The graph shows the ratio of MOB1/GAPDH. Data were normalized to the control (No infection samples) within each MG132 treatment group (arbitrarily set to 1) and are the mean of two independent experiments.(D) YAP1 is degraded in human macrophages challenged with virulent L. pneumophila. Human U937 macrophages were challenged with the indicated L. pneumophila strains, and the level of YAP1 was determined by immunoblot as described in Figure 3D.

Figure S4. Bioinformatic analysis of LegK7-dependent host gene expression and conservation of the core components of the Hippo pathway (Related to Figure 4). (A) Functional analysis of genes differentially expressed in response to LegK7. Of the 135 genes differentially regulated in mouse RAW264.7 macrophages upon challenge with either Lp02 or Lp02ΔlegK7, 78 genes were up-regulated > 2-fold and 57 genes were down-regulated > 2-fold (Lp02ΔlegK7 vs. Lp02, p<0.01) at either 3-hour or 5-hour post infection. 66 of those genes have been characterized and were categorized based on their biological function. The number of genes in each functional group is indicated in brackets. (B) Transcription factor binding motifs enriched in macrophage genes differentially regulated by LegK7. Of the 135 differentially regulated genes shown in Figure 4A, 87 genes (number of query genes) have entries in the TRANSFAC TFBS database. The number of common genes that have each of the predicted motifs is shown. Peroxisome proliferator-activated receptor gamma: Retinoid X receptor alpha (PPARγ:RXRα), Lymphocyte function-associated antigen 1 (LF-A1), Zinc finger of the cerebellum 1 (Zic1), Myoblast determination protein (MyoD). (C) Overview of core components of the Hippo pathway and their conservation level. Amino acid identity between human proteins and their orthologs from the indicated organisms was determined by pairwise alignment using Blast analysis. Protein coverage was calculated as the length of the homologous region relative to the full length of the ortholog. No ortholog of human YAP1 was identified in either Dictyostelium or Acanthamoeba.