The secreted protein kinase CstK from Coxiella burnetii influences vacuole development and interacts with the GTPase-activating host protein TBC1D5

The intracellular bacterial pathogen Coxiella burnetii is the etiological agent of the emerging zoonosis Q fever. Crucial to its pathogenesis is type 4b secretion system–mediated secretion of bacterial effectors into host cells that subvert host cell membrane trafficking, leading to the biogenesis of a parasitophorous vacuole for intracellular replication. The characterization of prokaryotic serine/threonine protein kinases in bacterial pathogens is emerging as an important strategy to better understand host–pathogen interactions. In this study, we investigated CstK (for Coxiella Ser/Thr kinase), a protein kinase identified in C. burnetii by in silico analysis. We demonstrate that this putative protein kinase undergoes autophosphorylation on Thr and Tyr residues and phosphorylates a classical eukaryotic protein kinase substrate in vitro. This dual Thr-Tyr kinase activity is also observed for a eukaryotic dual-specificity Tyr phosphorylation-regulated kinase class. We found that CstK is translocated during infections and localizes to Coxiella-containing vacuoles (CCVs). Moreover, a CstK-overexpressing C. burnetii strain displayed a severe CCV development phenotype, suggesting that CstK fine-tunes CCV biogenesis during the infection. Protein–protein interaction experiments identified the Rab7 GTPase-activating protein TBC1D5 as a candidate CstK-specific target, suggesting a role for this host GTPase-activating protein in Coxiella infections. Indeed, CstK co-localized with TBC1D5 in noninfected cells, and TBC1D5 was recruited to CCVs in infected cells. Accordingly, TBC1D5 depletion from infected cells significantly affected CCV development. Our results indicate that CstK functions as a bacterial effector protein that interacts with the host protein TBC1D5 during vacuole biogenesis and intracellular replication.

Signal transduction is an essential and universal function that allows all cells, from prokaryotes to eukaryotes, to translate environmental signals to adaptive changes. By this mechanism, extracellular inputs propagate through complex signaling networks whose activity is often regulated by reversible protein phosphorylation. Signaling mediated by serine/threonine/tyrosine protein phosphorylation has been extensively studied in eukaryotes; however, its relevance in prokaryotes has only begun to be appreciated. The recent discovery that bacteria also use Ser/Thr/Tyr kinase-based signaling pathways has opened new perspectives to study environmental adaptation, especially in the case of bacterial pathogens, with respect to host infection (1). Thus, advances in genetic strategies and genome sequencing have revealed the existence of "eukaryotic-like" serine/threonine protein kinases (STPKs) 4 and phosphatases in a number of prokaryotic organisms (2), including pathogens such as Streptococcus spp. (3)(4)(5)(6), Mycobacteria (7)(8)(9)(10)(11)(12), Yersinia spp. (13,14), Listeria monocytogenes (15,16), Pseudomonas aeruginosa (17), Enterococcus faecalis (18), or Staphylococcus aureus (19,20). Consequently, the study of STPKs in human bacterial pathogens is emerging as an important strategy to better understand host-pathogen interactions and develop new, targeted antimicrobial therapies. However, if on one hand it is clear that STPKs and phosphatases regulate important functions in bacterial pathogens, their signal transduction mechanism remains ill-defined and restricted to a limited number of microbes. Importantly, STPKs expressed by pathogenic bacteria can either act as key regulators of important microbial processes or be translocated by secretion systems to interact with host substrates, thereby subverting essential host functions including the immune response, cell shape, and integrity (21). Phosphorylation of host substrates has been demonstrated for some bacterial STPKs, whereas others seem to require their kinase activity, but their phosphorylated substrates remain to be identified (21). Therefore, biochemical mechanisms of these pathogendirected targeted perturbations in the host cell-signaling network are being actively investigated, and STPKs are proving to be molecular switches that play key roles in host-pathogen interactions (21).
Among emerging human pathogens, Coxiella burnetii is a highly infectious bacterium, responsible for the zoonosis Q fever, a debilitating flu-like disease leading to large outbreaks with a severe health and economic burden (22)(23)(24). The efficiency of infections by C. burnetii is likely associated with the remarkable capacity of this bacterium to adapt to environmental as well as intracellular stress. Indeed, outside the host, C. burnetii generates pseudospores that facilitate its airborne dissemination. C. burnetii has developed a unique adaptation to the host, being the only bacterium that thrives in an acidic compartment containing active lysosomal enzymes. Upon host cell invasion, bacteria reside within membrane-bound compartments that passively traffic through the endocytic maturation pathway, progressively acquiring early and late endocytic markers such as Rab5 and Rab7, respectively (25). Fusion of Coxiella-containing vacuoles (CCVs) with late endosomes and lysosomes is accompanied by the acidification of the endosomal environment, which is required to activate the translocation of bacterial effector proteins by a Dot/Icm type 4b secretion system (26). Some of these effectors modulate important signaling pathways of infected cells, including apoptosis and inflammasome activation (27)(28)(29), whereas others are essential for the development of the intracellular replicative niche. Among these, CvpB and CvpF have been recently implicated in the manipulation of autophagy for optimal vacuole development (30 -33). C. burnetii genome analysis revealed a close homology to the facultative intracellular pathogen Legionella pneumophila, in particular at the level of Dot/Icm core genes (34). In silico analysis identified over 100 candidate effector proteins encoded in the C. burnetii genome, some of which have been validated for secretion using either C. burnetii or L. pneumophila as a surrogate model (26,35,36).
In this study, we investigated the candidate effector CBU_0175, which encodes a unique putative Coxiella Ser/Thr kinase (CstK). We demonstrated CstK translocation by C. burnetii during infection, and we reported its localization at CCVs. In vitro kinase assays revealed that CstK undergoes autophosphorylation on Thr and Tyr residues and displays a bona fide kinase activity toward a test substrate of eukaryotic protein kinases. Furthermore, the identification of the Rab7 GTP-activating protein TBC1D5 as a CstK interactor suggests that this protein might be involved during infection to facilitate CCVs biogenesis. Indeed, TBC1D5 is actively recruited at CCVs during Coxiella infections, and TBC1D5-targeting siRNAs significantly affect CCVs development. Our data provide the first evidence that a C. burnetii secreted kinase might control host cell infection.

C. burnetii genome encodes a single putative protein kinase
In silico analysis of the virulent C. burnetii strain RSA493 NMI genome revealed only one gene encoding a putative STPK. To date, no STPKs have been characterized in this organism. This gene was named cstK for C. burnetii serine threonine kinase and encodes a 246-amino acid protein with an estimated molecular mass of 31 kDa. The gene coding for cstK is flanked by genes CBU_0174 (which encodes an hypothetical protein) and CBU_0176, a gene coding for the serine protease domaincontaining protein degP.1. Of note, these genes are not part of an operon (Fig. 1A). InterProScan analysis of CstK revealed the presence of most of the essential amino acids and sequence subdomains characterizing the Hanks family of eukaryotic-like protein kinases (37). CstK shares a common eukaryotic protein kinase superfamily fold with two lobes and a Gly-rich loop. These protein kinases include the central core of the catalytic domain and the invariant lysine residue in the consensus motif within subdomain II, which is usually involved in the phosphate transfer reaction and required for the autophosphorylating activity of eukaryotic STPKs (Fig. 1A) (37)(38)(39). The activation loop in the catalytic domain is particularly short in CstK, and the DFG motif is substituted by a GLG motif. Interestingly, the transmembrane domain usually present in classical prokaryotic STPKs is lacking in CstK; thus, it is a so-called cytoplasmic STPK.

CstK is a Dot/Icm effector protein
Bioinformatics analysis using the prediction software S4TE 2.0 (40) indicated that CstK harbors features corresponding to secreted effector proteins, including a promoter motif typically found in effector proteins from intravacuolar bacterial pathogens, suggesting that CstK is indeed a Coxiella effector protein (Fig. 1A). Consistently, previous studies by Chen et al. (36) have shown that CstK is secreted in a type 4b secretion system-dependent manner by the surrogate host L. pneumophila, albeit with low efficiency. To validate CstK secretion in C. burnetii, we engineered plasmids encoding, either CstK or CvpB (a known C. burnetii effector protein) (30), fused to ␤-lactamase (BLAM) and expressed in WT Coxiella Tn1832 (a C. burnetii transposon mutant expressing GFP and that phenocopies WT C. burnetii) or the Dot/Icm-defective dotA::Tn mutant, also expressing GFP. By means of a BLAM secretion assay, we could observe that BLAM-CstK was secreted by WT Coxiella at 48 and 72 h, but not at 24 h postinfection (Fig. 1B). Secretion of BLAM-CvpB or BLAM-CstK was not detectable in cells infected with the dotA::Tn strain, indicating that both CvpB and CstK are C. burnetii Dot/Icm substrates (Fig. 1B). Next, the intracellular localization of Cstk was investigated by ectopically expressing HA-tagged CstK either in noninfected or WT C. burnetii-infected U2OS cells. In noninfected cells, CstK localized at intracellular compartments that were negative for the lysosomal marker LAMP1, whereas it was recruited at CCVs (as revealed by the co-localization with LAMP1) in infected cells (Fig. 1C).

CstK displays autokinase and protein kinase activities
To determine whether CstK is a functional protein kinase, this protein was overproduced in Escherichia coli and purified as a recombinant protein fused to glutathione S-transferase (GST) tag. The purified tagged CstK protein ( Fig. 2A, upper panel) was then assayed for autokinase activity in the presence of the phosphate donor [␥-33 P]ATP. As shown in Fig. 2A

CstK affects C. burnetii vacuole biogenesis
the expected size of the protein isoform, strongly suggesting that this kinase undergoes autophosphorylation. To confirm CstK autophosphorylation and exclude the possibility that contaminant kinase activities from E. coli extracts might phosphorylate CstK, we mutated the conserved Lys 55 residue present in subdomain II into CstK by site-directed mutagenesis. Indeed protein sequence analysis revealed that Lys 55 in CstK is similarly positioned as a conserved Lys residue usually involved in the phosphotransfer reaction and also required for the autophosphorylating activity of eukaryotic-like STPKs (37,38). Thus, Lys 55 was substituted by a Met residue, the mutated form of CstK, CstK_K55M, was purified as described above ( Fig. 2A,  upper panel), and it was then tested for autophosphorylation in the presence of [␥-33 P]ATP. As expected, no radioactive signal could be detected ( Fig. 2A, lower panel), thus establishing that CstK displayed autophosphorylation activity. A kinetic analysis of CstK phosphorylation was next performed to determine the initial CstK phosphorylation rate (Fig. 2B). Incorporation of ␥-phosphate occurred rapidly, reaching ϳ50% of its maximum rate within 5 min of reaction. This autokinase activity depended on bivalent cations such as Mg 2ϩ and Mn 2ϩ in the range of 5 mM, thus in correlation with concentrations required for canonical STPK activity, as shown in Fig. 2C, and abolished by addition of 20 mM EDTA chelating all the divalent cations available (data not shown).
The recombinant CstK protein was further characterized by studying its ability to phosphorylate exogenous proteins and was thus assayed for in vitro phosphorylation of the general eukaryotic protein kinase substrate, myelin basic protein (MBP), in the presence of [␥-33 P]ATP. MBP is a commonly used substrate for both Ser/Thr and Tyr kinases. A radiolabeled signal at the expected 18-kDa molecular mass of MBP was detected, thus demonstrating that CstK phosphorylates protein substrates such as MBP ( Fig. 2A). As expected, the CstK_K55M mutant did not phosphorylated MBP. Altogether, these data indicate that in vitro, CstK possesses intrinsic autophosphorylation activity and displays kinase functions for exogenous substrates.

Identification of CstK autophosphorylation sites
To determine the specificity of this kinase, we next identified its autophosphorylation sites. A MS approach was used because this technique allows precise characterization of post-translational modifications including phosphorylation (41,42). NanoLC/nanospray/tandem MS (LC-ESI/MS/MS) was applied for the identification of phosphorylated peptides and for localization of phosphorylation sites in CstK. This approach led to 97% of sequence coverage, whereas the remaining residues uncovered did not include Ser, Thr, or Tyr residues.

CstK affects C. burnetii vacuole biogenesis
As detailed in Table 1, analysis of tryptic digests allowed the characterization of three phosphorylation sites in CstK. Surprisingly, unlike classical Ser/Thr or Tyr kinases, CstK was phos-phorylated on two Tyr residues (Tyr 14 and Tyr 209 ), in addition to one Thr site (Thr 232 ). Because protein sequence analysis did not reveal a classical activation loop in this kinase, the contri-

CstK affects C. burnetii vacuole biogenesis
bution of Thr 232 , Tyr 14 , and Tyr 209 to CstK kinase activity was individually assessed. Hence, these residues were mutated either to phenylalanine to replace tyrosine residues or alanine to replace threonine residue, generating the single mutants CstK_Y14F, CstK_Y209F, and CstK_T232A, as well as the CstK_Y14F/Y209F/T232A triple mutant (CstK_FFA). Next, in vitro kinase assays with [␥-33 P]ATP were carried out and revealed that maximum loss in CstK autophosphorylation activity was observed in the CstK_Y14F mutant ( Fig. 2D), suggesting that this site is central for CstK activation. In contrast, the CstK_Y209F mutant exhibited a slight hyperphosphorylation, which might indicate that Tyr 209 only plays an accessory role in controlling CstK autophosphorylation (Fig. 2D). Finally, the CstK_T232A mutant showed a reduced CstK phosphorylation and displayed diminished kinase activity toward the exogenous substrate MBP (Fig. 2D). Note that mutating all three autophosphorylation sites fully abrogated CstK kinase activity (Fig. 2D). These results indicate that Tyr 14 and Thr 232 are the major phosphorylation sites in CstK and strongly suggest that CstK might be a dual specificity (Thr/Tyr) kinase.

CstK activity and phosphorylated state affect its intracellular localization
Next, we ectopically expressed HA-tagged CstK, CstK_ K55M, and CstK_FFA derivatives in noninfected and C. burnetii-infected U2OS cells to investigate its intracellular localization. CstK mainly localized at vesicular compartments in noninfected cells and accumulated at CCVs upon C. burnetii infection, suggesting an active role in the biogenesis of this compartment (Fig. 1C). Interestingly, the inactive CstK_K55M mutant localized at vesicular structures positive for the lysosomal marker LAMP1 but not at CCVs, whereas the nonphosphorylated CstK_FFA displayed a diffuse localization in the cytosol of transfected cells (Fig. 2E). Overall, these data suggest that the kinase activity and phosphorylated state of CstK play an important role in its localization in cells.

CstK regulates vacuole development and C. burnetii replication within infected cells
As a first step toward the understanding of CstK functions in the course of infection and to appreciate the extent to which this kinase is required for growth and viability of C. burnetii, we attempted to inactivate the corresponding chromosomal gene. Unfortunately, after several attempts we were unable to generate a null mutant, suggesting that cstK might be essential. However, we had previously isolated a C. burnetii mutant (Tn2496) carrying a transposon insertion allowing GFP expression at position 156,783, 32 bp upstream of the starting codon of cstK (43) (Fig. 1A). To determine the effect of this transposon insertion on cstK gene expression, we assessed the expression level of cstK mRNA from WT C. burnetii and Tn2496 strains. Surprisingly, cstK expression was significantly up-regulated in the mutant strain, suggesting that the transposon insertion may have released a transcriptional negative regulation (Fig. 3A). This suggested that a putative transcriptional regulator might bind the cstK promoter and control its activity during host invasion.
We next examined the effects of CstK overexpression on C. burnetii infections by challenging Vero cells with WT C. burnetii, the Dot/Icm-defective dotA::Tn mutant, or the Tn2496 mutant. Intracellular growth of the CstK-overexpressing strain was significantly reduced over 7 days of infection with an intermediate phenotype between WT and the dotA::Tn mutant (Fig. 3B). Accordingly, multiparametric phenotypic profile analysis of the Tn2496 mutant indicated that this strain exhibited a major defect in CCV development as compared with WT C. burnetii (Fig. 3, C and D). To further investigate the effects of CstK overexpression on C. burnetii infections, GFPexpressing C. burnetii were transformed with plasmids expressing HA-tagged WT CstK or its corresponding mutants (CstK_K55M and CstK_FFA) under the control of an IPTG promoter. U2OS cells expressing cytoplasmic mCherry were challenged with the three C. burnetii strains in the presence or absence of IPTG. After 6 days of infection, the cells were fixed, labeled with Hoechst and anti-LAMP1 antibody to visualize host cells nuclei and CCVs, respectively, and processed for automated image analysis. In all cases, the overexpression of CstK was detrimental for CCVs biogenesis and bacterial replication (Fig. 3E). Next, U2OS cells were challenged with WT C. burnetii, the GFP-expressing Tn2496 mutant strain, or a combination of the two for 6 days (Fig. 3F). The cells were then fixed and labeled with an anti-C. burnetii antibody to label both bacteria strains and incubated with Hoechst to visualize host cells nuclei (Fig. 3F). Automated image analysis was then used to determine the effects of CstK overexpression on the replication of WT bacteria, in trans. Co-infections resulted in a significant increase in the size of Tn2496 colonies, indicating that WT C. burnetii can partially restore the growth of the CstK-overexpressing strain (Fig. 3G). However, a significant decrease in the size of bacterial colonies labeled by the anti-C. burnetii antibodies indicated that CstK overexpression has a detrimental effect in trans on the development of WT bacteria (Fig. 3G). Of note, vacuoles harboring WT or mutant colonies alone were never observed in co-infection experiments. Therefore, we concluded that CstK participates in the formation of the C. burnetii replicative vacuole and that its expression must be finely tuned for optimal intracellular replication.

CstK specifically interacts with host cell proteins
Because CstK is a secreted protein, we assume that this kinase would interfere with host cell signal transduction pathways to subvert host cell defenses to the benefit of the bacteria. To identify host cell proteins that could interact with CstK, we made use of the model amoeba Dictyostelium discoideum. D. discoideum is a eukaryotic professional phagocyte amenable to genetic and biochemical studies. Lysate from cells overex-

CstK affects C. burnetii vacuole biogenesis
pressing CstK tagged with a C-terminal FLAG epitope (CstK-FLAG) was incubated with beads coupled to an anti-FLAG antibody. The beads were extensively washed, and bound proteins were separated by SDS-PAGE before MS analysis. Among the putative interactants of CstK identified by this approach, some were discarded on the basis of their

CstK affects C. burnetii vacuole biogenesis
intracellular localization, whereas other retained candidates were mostly involved in the endocytic pathway (Table S1). Among these, the Rab GTPase-activating protein/TBC domain-containing protein, DDB_G0280253 (UniProtKB Q54VM3), is a 136.4-kDa protein homologous to mammalian TBC1D5 (dictyBase), a GTPase-activating protein for Rab7a and Rab7b (44 -46) that acts as a molecular switch between the endosomal and the autophagy pathway (47). Given the recently reported implication of TBC1D5 in the biogenesis of L. pneumophila-containing vacuoles (48) and the role of autophagy in the biogenesis of CCVs (25,31,49), we aimed at validating the interaction between human TBC1D5 (Hs-TBC1D5) and CstK in HEK-293T cells. The cells co-expressing Hs-TBC1D5-GFP and HA-CstK or CstK mutants were used for immunoprecipitation using anti-HA beads. WT and CstK derivatives were detected as co-immunoprecipitated in the presence of Hs-TBC1D5-GFP, thus confirming that Hs-TBC1D5 is a bona fide CstK interactant (Fig. 4A). Significantly higher levels of TBC1D5 were co-immunoprecipitated by the CstK mutants, suggesting that the interaction might be increased in the absence of phosphorylation turnover of the kinase (Fig. 4A, bottom panel). Interestingly, the interaction is not dependent on the phosphorylation status of CstK because the K55M mutant and the triple FFA mutant are still able to interact. Other candidates identified by MS are currently being investigated.

TBC1D5 is recruited at CCVs and regulates their biogenesis
Given its recently reported role in the development of L. pneumophila-containing vacuoles, we investigated the localization of TBC1D5 in U2OS cells infected with WT C. burnetii, the CstK-overexpressing strain Tn2496, or the Dot/Icmdefective mutant dotA::Tn. Consistent with previous studies demonstrating a role in the activation of Rab7a and b, TBC1D5 seems to accumulate specifically at CCVs in a Dot/Icm-independent manner because the eukaryotic protein was found at CCVs generated by all C. burnetii (WT, the Tn2496, and the dotA::Tn; Fig. 4B and Fig. S1). Next, we used siRNA to deplete  Figure 4. CstK interacts with TBC1D5. A, The interaction between CstK and the human TBC1D5 has been confirmed after cotransfection of HEK-293T cells to express each HA-tagged CstK derivatives with GFP-tagged Hs-TBC1D5. Cells have been lysed and HA-tagged CstK derivatives have been trapped on anti-HA magnetic beads (Pierce). Beads were washed, eluted by boiling, and bound proteins were revealed by Western blot analysis. Anti-HA antibody confirms the immunoprecipitation of CstK derivatives. For the densitometry graph, Regions of Interest (ROIs) were obtained from each band of interest and the intensity was measured (ctrl for "control" and WCL for "whole cell lysates"). B, Hs_TBC1D5 localization during infection was monitored in U2OS cells expressing mCherry-tagged Hs-TBC1D5 (red) challenged for 4 days either with wt C. burnetii, the Dot/Icm-defective dotA::Tn mutant or the CstK-overexpressing mutant Tn2496, all expressing GFP (green). Lysosomal compartments were labelled with anti-LAMP1 antibodies (blue). Scale bars 10 m. Inset magnification scale bars 4 m. C, The role of TBC1D5 in C. burnetii infections was investigated using siRNA to deplete Hs-TBC1D5 in U2OS prior to challenge with GFP-expressing wt C. burnetii. The size of CCVs was automatically calculated over an average of 150000 cells per condition. Red bars indicate medians. ***, p Ͻ 0.0001; **, p Ͻ 0.001; *, p Ͻ 0.01; one-way ANOVA, Dunnett's multiple comparison test. Scale bars 10 m.

CstK affects C. burnetii vacuole biogenesis
cells of TBC1D5 prior to C. burnetii infection, to investigate a possible role in CCVs development and intracellular bacterial replication. Indeed, vacuole development was significantly reduced in cells exposed to Hs-TBC1D5-targeted siRNAs as opposed to cells treated with nontargeting siRNA oligonucleotides (Fig. 4C).

TBC1D5 is not phosphorylated in vitro by recombinant CstK
We assessed whether CstK might phosphorylate the recombinant Hs-TBC1D5. Despite the in silico prediction of several Ser/Thr and Tyr phosphorylatable residues in Hs-TBC1D5, we failed to detect Hs-TBC1D5 phosphorylation using several in vitro kinases assays (Fig. S2). In addition, Hs-TBC1D5 phosphorylation status was also investigated upon transfection with CstK or its inactive derivative (K55M) followed by Hs-TBC1D5 immunoprecipitation. No phosphorylation could be detected in our experimental conditions.

Discussion
Bacterial Ser/Thr/Tyr kinases expressed by pathogenic bacteria can either act as key regulators of important microbial processes or be translocated by secretion systems to interact with host substrates; thereby our results provide the first biochemical analysis of the secreted C. burnetii kinase CstK and its involvement in the process of infection and CCVs development. Importantly, CstK presents important differences as compared with classical Ser/Thr kinases. In particular, we provided evidence that CstK is a dual kinase able to autophosphorylate on Thr and Tyr residues. Moreover, the observation that a transposon insertion 32 bp upstream of the cstK starting codon leads to an increase in the levels of cstK mRNAs was indicative of the presence of a negative transcriptional regulation of gene expression, suggesting a fine-tuning of the levels of CstK. Indeed, the Tn2496 mutant displays a severe CCV biogenesis defect when used to challenge U2OS cells, highlighting the importance of regulating cstK expression during C. burnetii infections. Accordingly, inducing the expression of WT CstK in WT C. burnetii severely impairs CCVs development and bacterial replication. Co-infection experiments demonstrated that CstK overexpression can also act in trans, by perturbing the intracellular replication of WT C. burnetii. The identification of candidate eukaryotic interactors of CstK further corroborated a role of the bacterial kinase in subverting host functions during infection. Here we confirmed that CstK interacts with TBC1D5, but we failed to detect phosphorylation of the eukaryotic target by CstK. However, we cannot exclude that TBC1D5 is a genuine CstK substrate in vivo because the lack of phosphorylation of host interactors of bacterial STPKs is not uncommon. Interaction between STPKs and host proteins might well perturb protein interaction networks at play in host cells (21). Indeed, the induced overexpression of CstK mutants lacking kinase activity in WT C. burnetii impaired CCVs development to the same extent as the overexpression of WT CstK. The biochemical mechanisms of these pathogen-directed targeted perturbations of host cell-signaling networks are being actively investigated. Regardless, siRNA depletion of TBC1D5 in C. burnetii-infected cells points at a role of the eukaryotic protein in CCVs development. In mammals, TBC1D5 was sug-gested to function as a molecular switch between endosomal and autophagy pathways. Indeed TBC1D5 associates the retromer VPS29 subunit involved in endosomal trafficking, and upon autophagy induction, the autophagy ubiquitin-like protein LC3 can displace VPS29, thus orienting TBC1D5 functions toward autophagy instead of endosomal functions (47). It is thus tempting to propose that CstK might interfere with this tight regulation between TBC1D5, LC3, and VPS29 and redirect TBC1D5 functions to support efficient C. burnetii intracellular replication. Further work will need to be carried out to decipher how CstK recognizes these host substrates and how they participate in the establishment of C. burnetii parasitophorous vacuoles. Another perspective of this work is the opening of a new field of investigation for future drug development to fight this pathogen. Because CstK seems to be essential, specific inhibitors of this kinase preventing C. burnetii growth would be extremely useful for the development of new therapies.

Bacterial strains and growth conditions
Bacterial strains and plasmids are described in Table 2. Strains used for cloning and expression of recombinant proteins were E. coli TOP10 (Invitrogen) and E. coli BL21 (DE3)Star (Stratagene), respectively. E. coli cells were grown and maintained at 25°C in LB medium supplemented with 100 g/ml ampicillin when required. C. burnetii RSA439 NMII and transposon mutants Tn1832, Tn2496, and dotA::Tn were grown in ACCM-2 (45) supplemented with chloramphenicol (3 mg/ml) in a humidified atmosphere of 5% CO 2 and 2.5% O 2 at 37°C.

Cloning, expression, and purification of CstK derivatives
The cstK (CBU_0175) gene was amplified by PCR using C. burnetii RSA439 NMII chromosomal DNA as a template with the primers listed in Table 3 containing a BamHI and Hin-dIII restriction site, respectively. The corresponding amplified product was digested with BamHI and HindIII and ligated into the bacterial pGEX(M) plasmid, which includes a N-terminal GST tag, thus generating pGEX(M)_cstK. pGEX(M)_cstK derivatives harboring different mutations were generated by using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and resulted in the construction of plasmids detailed in Table 3. For overexpression assays, cstK and its derivatives were cloned in the pJA-LACO-4xHA vector (30) using KpnI and BamHI restriction sites. All constructs were verified by DNA sequencing. Transformed E. coli BL21 Star cells with pGEX(M)_cstK derivatives were grown at 16°C in LB medium containing 1 g/liter of glucose and 100 g/ml of ampicillin and protein synthesis induced with 0.5 mM IPTG overnight. Bacteria were disrupted by sonication (Branson, digital sonifier) and centrifuged at 14,000 rpm for 25 min. Purifications of the GST-tagged recombinants were performed as described by the manufacturer (GE Healthcare). cstK coding sequence was also optimized for mammalian cell expression (GenScript), amplified by PCR, and cloned into pDXA-3C (50) containing a FLAG tag for C-terminal fusion. After sequencing, the plasmid was linearized by the restriction

CstK affects C. burnetii vacuole biogenesis
enzyme ScaI and transfected in D. discoideum as described (51). Clone selection was made with 10 mg/ml of G418, and protein expression was assayed by Western blotting analysis of D. discoideum crude extract with an anti-FLAG rabbit polyclonal antibody (GenScript). For ectopic expression assays, cstK, cstK_K55M, and cstK_FFA with optimized codons (IDT) were cloned in pRK5-HA using the primer pair CstKopt-BamHI-Fw/CstKopt-EcoRI-Rv.

Cloning, expression, and purification of TBC1D5 derivatives
The D. discoideum GFP-tagged TBC1D5 was previously generated (48). Cells were grown at 22°C in HL5 medium as previously described (51). Human TBC1D5 coding sequence was obtained from the hORFeome v8.1 (ORF 2659, Q92609, fully sequenced cloned human ORFs in Gateway Entry clones ready for transfer to Gateway-compatible expression vectors). HsTBC1D5 coding sequence has been recombined into pEGFP-N1 RfC Destination vector by GateWay reaction (MGC Platform Montpellier), thus generating pEGFP-N1_HsTBC1D5 coding for HsTBC1D5 with a C-terminal GFP tag. pmCH_Hs-TBC1D5-mCherry has been generated by the same method (MGC Platform Montpellier).

RNA extraction and quantitative RT-PCR (qRT-PCR)
50 ml of C. burnetii culture was harvested, resuspended in 600 l of RNA protect reagent (Qiagen) and incubated for 5 min at room temperature. Bacteria were centrifuged and resuspended in 200 l of TE (10 mM Tris-HCl, 1 mM EDTA, pH 8) containing 1 mg/ml lysozyme. Bacterial suspension was incubated at room temperature for 5 min, and bacteria were disrupted by vigorous vortexing for 10 s every 2 min. 700 l of lysis buffer from RNA easy kit (Qiagen) were added to the bacterial  Table 3. The expression level of cstK in the WT strain was set at 100, and the expression levels of cstK in the Tn2496 mutant were normalized to the WT levels.

In vitro kinase assays
In vitro phosphorylation was performed with 4 g of WT CstK or CstK derivatives in 20 l of buffer P (25 mM Tris-HCl, pH 7.0, 1 mM DTT, 5 mM MnCl 2 , 1 mM EDTA, 50 M ATP) with 200 Ci ml Ϫ1 (65 nM; ␥-33 P]ATP; PerkinElmer, ref: NEG 602H250UC, 3000 Ci mmol Ϫ1 ) for 30 min at 37°C. For substrate phosphorylation, 4 g of MBP (Sigma) and 4 g of CstK were used. Each reaction mixture was stopped by addition of an equal volume of Laemmli buffer, and the mixture was heated at 100°C for 5 min. After electrophoresis, the gels were soaked in 16% TCA for 10 min at 90°C and dried. Radioactive proteins were visualized by autoradiography using direct exposure to films.

Mass spectrometry analysis
For MS analysis, CstK was phosphorylated as described above, except that [␥-33 P]ATP was replaced with 5 mM cold ATP. Subsequent MS analyses were performed as previously reported (52,53). Briefly, the samples were submitted to trypsin digestion and analyzed using an Ultimate 3000 nano-RSLC (Thermo Scientific, San Jose, CA) coupled on line with a quadrupole Orbitrap Q Exactive HF mass spectrometer via a nanoelectrospray ionization source (Thermo Scientific). The samples were injected and loaded on a C18 Acclaim PepMap100 trap-column (Thermo Scientific) and separated on a C18 Acclaim Pepmap100 nano-column (Thermo Scientific). MS data were acquired in a data-dependent strategy selecting the fragmentation events based on the 20 most abundant precursor ions in the survey scan (350 -1600 Th). The resolution of the survey scan was 60,000 at m/z 200 Th, and for MS/MS scan the resolution was set to 15,000 at m/z 200 Th. Peptides selected for MS/MS acquisition were then placed on an exclusion list for 30 s using the dynamic exclusion mode to limit duplicate spectra. Data files were then analyzed with Proteome Discover 2.2 using the SEQUEST HT algorithm against the Uniprot D. discoideum, which included the sequence of CstK.

C. burnetii infections
U2OS epithelial cells were challenged with C. burnetii RSA439 NMII, the transposon mutants Tn1832, dotA::Tn, or Tn2496 as previously described (38,49). For co-infection experiments, the cells were challenged with a 1:1 ratio of C. burnetii RSA439 NMII and Tn2496 transposon mutant. For gene silencing, U2OS cells were seeded at 2,000 cells/well in black, clear-bottomed, 96-well plates in triplicate and transfected with siRNA oligonucleotides 24 h later by using the RNAiMAX transfection reagent (Thermo Fisher Scientific) according to the manufacturer's recommendations. At 24 h post-transfection, the cells were challenged with C. burnetii (MOI of 100) and further incubated for 5 days. The cells were then fixed and processed for immunofluorescence. Where appropriate, anti-LAMP1 antibodies were used to label lysosomes and CCVs as previously described (54). Samples were imaged with a Zeiss Axio Imager Z1 epifluorescence microscope (Carl Zeiss) connected to a CoolSNAP HQ 2 CCD camera (Teledyne Photometrics, Tucson, AZ). Images were acquired with 40ϫ oil immersion objectives and processed with Metamorph (Molecular Devices, San Jose, CA). For phenotypic screening, the samples were imaged with an ArrayScan VTI Live epifluorescence automated microscope (Cellomics) equipped with an ORCA-ER CCD camera (Hamamatsu). 25 fields/well were acquired for image analysis. ImageJ and ICY software were used for image analysis and quantifications. Phenotypic profiles (expressed as z scores) were calculated using CellProfiler, from triplicate experiments as previously described (54) following median based normalization of 96-well plates. Plates effects were cor-

CstK affects C. burnetii vacuole biogenesis
rected by the median value across wells that are annotated as control.

␤-Lactamase translocation assay
Effector proteins translocation assays were performed as previously described (30). Briefly, C. burnetii Tn1832 (WT) and dotA::Tn were transformed with pXDC-Blam (negative control), pXDC-Blam-CvpB (positive control), or pXDC-Blam-CstK. Each strain was used to infect U2OS epithelial cells. After 24, 48, or 72 h of infection, the cells were loaded with the fluorescent substrate CCF4/AM (LiveBLAzer-FRET B/G loading kit; Invitrogen) in Hanks' balanced salt solution (20 mM HEPES, pH 7.3) containing 15 mM probenecid (Sigma). The cells were incubated in the dark for 1 h at room temperature and imaged using an EVOS inverted fluorescence microscope. Images were acquired using 4Ј,6Ј-diamino-2-phenylindole and GFP filter cubes. The image analysis software CellProfiler was used to segment and identify all cells in the sample (GFP) and positive cells (4Ј,6Ј-diamino-2-phenylindole) and to calculate the intensity of fluorescence in each channel. The percentage of positive cells versus the total number of infected cells was then calculated and used to evaluate effector translocation.

Immunoprecipitation from D. discoideum lysates
For immunoprecipitation assays, 2 ϫ 10 7 cells were lysed in lysis buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 0.5% Nonidet P-40, protease inhibitors (Roche)) and cleared by centrifugation for 15 min at 14,000 rpm in a microfuge. Lysate supernatants were incubated overnight at 4°C with anti-FLAG mAb coated on agarose beads (Genscript). The beads were then washed five times in wash buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 0.1% Nonidet P-40) and once in PBS. Bound proteins were migrated on SDS-PAGE and analyzed by LC-MS/MS.

Cell culture, heterologous expression, and anti-HA immunoprecipitation
HEK-293T cells were grown in DMEM (Gibco) containing 10% (v/v) FBS, 1% GlutaMAX (Gibco, 200 mM stock), 0.5% penicillin/streptomycin (Gibco, 10,000 units/ml stock) and maintained under standard conditions at 37°C in a humidified atmosphere containing 5% CO 2 . The cells were transiently transfected using the jetPEI transfection reagent (Polyplus-Transfection Inc.) to express either Hs-TBC1D5-GFP, Cst-K_HA derivatives, or each CstK derivatives with TBC1D5_GFP protein. The cells were used 24 h after transfection for immunoprecipitation assay. Transfected cells were washed two times in cold PBS and lysed in lysis buffer (50 mM Tris, 150 mM NaCl, 0.5 mM EDTA, 0.5% Nonidet P-40, protease, and phosphatases inhibitors (Roche)). Cleared lysate (950 l, ϳ1 mg of total proteins) were incubated with anti HA magnetic Beads (Pierce) for 30 min at room temperature under gentle rotation. The beads were washed three times in lysis buffer, boiled in 2ϫ Laemmli sample buffer and loaded on ExpressPlus TM PAGE gels (Gen-Script). The eluted proteins were visualized by Western blotting with the following antibodies: anti-HA from Chromotek, anti-GFP from Torrey Pines, donkey anti-rat, or anti-rabbit from Jackson ImmunoResearch.

Densitometry
Regions of Interest (ROIs) were obtained from each band of interest, and the intensity was measured using ImageLab (From Bio-Rad). For each band, the same ROI was used for background calculation and removal from areas adjacent to each band.