RpkA, a Highly Conserved GPCR with a Lipid Kinase Domain, Has a Role in Phagocytosis and Anti-Bacterial Defense

RpkA (Receptor phosphatidylinositol kinase A) is an unusual seven-helix transmembrane protein of Dictyostelium discoideum with a G protein coupled receptor (GPCR) signature and a C-terminal lipid kinase domain (GPCR-PIPK) predicted as a phosphatidylinositol-4-phosphate 5-kinase. RpkA-homologs are present in all so far sequenced Dictyostelidae as well as in several other lower eukaryotes like the oomycete Phytophthora, and in the Legionella host Acanthamoeba castellani. Here we show by immunofluorescence that RpkA localizes to endosomal membranes and is specifically recruited to phagosomes. RpkA interacts with the phagosomal protein complex V-ATPase as proteins of this complex co-precipitate with RpkA-GFP as well as with the GST-tagged PIPK domain of RpkA. Loss of RpkA leads to a defect in phagocytosis as measured by yeast particle uptake. The uptake of the pathogenic bacterium Legionella pneumophila was however unaltered whereas its intra-cellular replication was significantly enhanced in rpkA-. The difference between wild type and rpkA- was even more prominent when L. hackeliae was used. When we investigated the reason for the enhanced susceptibility for L. pneumophila of rpkA- we could not detect a difference in endosomal pH but rpkA- showed depletion of phosphoinositides (PIP and PIP2) when we compared metabolically labeled phosphoinositides from wild type and rpkA-. Furthermore rpkA- exhibited reduced nitrogen starvation tolerance, an indicator for a reduced autophagy rate. Our results indicate that RpkA is a component of the defense system of D. discoideum as well as other lower eukaryotes.


Introduction
Receptors are known to play important roles in phagocytosis and immunity in mammals. For example, Fc receptors, mannose receptors and scavenger receptors all reside at the cell surface of professional phagocytes and trigger phagocytosis upon binding their specific ligand [1]. Other receptors play a major role in immunity processes, e.g. Toll-like receptors, which serve as recognition receptors of pathogen-associated molecular patterns (PAMPs). These receptors are also present on maturing phagosomes. They participate in analyzing the content of the phagosome , trigger immune reactions upon stimulation and may influence the association of phagosome-binding proteins, as well as the maturation state of the phagosome, although this is still under debate [2,3]. In the professional phagocyte D. discoideum specific receptors for phagocytosis still remain unknown if they exist at all [4]. The same holds true for receptors which are involved in analyzing the phagosomal content. Until now only a few proteins have been identified in D. discoideum which are involved in phagocytosis and bacterial defense like the nine-transmembrane protein Phg1p or TirA a protein containing a Toll-Interleukin receptor domain [5,6].
Macrophages as well as Dictyostelium phagocytose a number of bacteria, but not all of them are effectively destroyed. Pathogenic bacteria like Listeria monocytogenes and Shigella flexneri are specialized to escape the phagosome [7,8], whereas others like Mycobacterium tuberculosis and Legionella pneumonia, staying inside the phagosome evade degradation by subverting maturation of the phagosome. One way to inhibit the phagosome maturation is to conceal its identity, which at least partially depends upon phosphoinositide composition of the membrane [9,10,11].
In recent years the social amoeba D. discoideum emerged as a suitable host to study infections with L. pneumophila [12,13]. As a professional phagocyte feeding on a variety of bacteria, D. discoideum is an ideal macrophage model. The gram-negative bacterium L. pneumophila is the pathogenic agent of Legionnaire's disease. Upon inhalation of contaminated aerosols it hijacks pulmonary macrophages in human hosts by reprogramming their phagosomes to become Legionella-containing vacuoles (LCVs). In this niche the bacteria undergo replication which ultimately leads to destruction of the macrophages and eventually to the clinical picture of pneumonia. In the US ,8,000-18,000 projected cases of hospitalized Legionnaires' disease occur per year [14]. Still, humans are only one possible host and a ''dead end street'' for the bacteria because transmission from human to human does not occur [15]. In fact, the primary targets of L. pneumophila are free living amoebae (FLA) in which the bacterium lives, divides and foremost is able to switch to a highly infectious mature intracellular form (MIF) which only occurs when grown in amoeba [16]. Inside its natural host, L. pneumophila is shielded from the surroundings and can survive even in environments usually hostile to bacteria, like artificial water supply systems [17,18]. FLA colonize water systems where they pose a threat to human health by hosting Legionella. Thus, understanding the molecular aspects of Legionella infection in amoebae as well as amoebal defense mechanisms provides a clear and present research goal [18,19,20].
We report on RpkA, a seven-helix transmembrane protein with a GPCR signature and a C-terminal lipid kinase domain predicted as a phosphatidylinositol-4-phosphate 5-kinase (GPCR-PIPK) localized in internal membranes. The RpkA gene is expressed throughout development and its loss is associated with a developmental defect [21]. The results presented here show that RpkA is specifically transported to maturing phagosomes with similar kinetics as V-ATPase and interacts with this protein complex. The overall pH however remains unaffected in the rpkA -. Loss of RpkA leads to a phagocytosis defect and results in an enhanced survival of L. pneumophila which could originate from a reduced phosphoinositide turnover and/or a reduced autophagy rate, making RpkA a component of the defense system of D. discoideum.

RpkA homologs are highly conserved in lower eukaryotes
In our initial studies we found RpkA-related proteins in the genomes of Phytophthora sojae and P. ramorum each encoding twelve RpkA homologs [22]. Recently additional genome data became available which allow a more detailed assessment of the occurrence of RpkA homologs during evolution. We have identified RpkA homologs in the closely related D. purpureum [23] as well as in D. fasciculatum and Polysphondylium pallidum which belong to different groups of the dictyostelids [24,25]. D. purpureum and D. fasciculatum harbor one rpkA gene each like D. discoideum, P. pallidum has two copies. The RpkAs in dictyostelids are highly homologous and share between 44 and 58% amino-acid identity (Table 1).
RpkA homologs are also present in other more distantly related species such as in the amoebozoa Acanthamoeba castellani, in Capsaspora owczarzaki, an amoeboid symbiont of a pulmonate snail [26], in Albugo laibachi, a blister rust parasitic to Arabidopsis thaliana and in the sponge Amphimedon queenslandica [27] (Table 1). Phytophthora, Albugo, Capsaspora and Amphimedon are opisthokonts whereas dictyostelids and Acanthamoeba are not. Thus, RpkA is a phylogenetically ancient protein, which is also present in ancient animals, the sponges, but seems to be absent in higher eukaryotes.
RpkA-GFP is present on acidic endosomal vesicles, on phagosomes and co-localizes with V-ATPase Previously we reported that carboxy-terminal GFP-tagged fusions of RpkA from D. discoideum (RpkA-GFP) localize to intracellular vesicles [21]. To exclude that the GFP-tag influenced the subcellular localization, we produced a variant harboring a Cterminal HA-tag (RpkA-HA). RpkA-HA was absent from the plasma membrane and present on intracellular vesicles of different size ( Figure S1A) resembling the distribution previously reported for RpkA-GFP [21]. Furthermore, RpkA-HA completely rescued the developmental phenotype of the mutant ( Figure S1B). We thus conclude that the C-terminal tag is unlikely to interfere with the function or distribution of RpkA.
In D. discoideum the V-ATPase is part of the contractile vacuole, an organelle which is responsible for osmoregulation, as well as of acidic endosomes. Since we detected RpkA-GFP in the phagosomal membrane upon incubation of cells expressing RpkA-GFP with TRITC labeled yeast we evaluated the possible colocalization of RpkA-GFP with VatA ( Figure 1, Figure S1A). Additionally, we co-expressed VatM-GFP, the membrane-spanning subunit of the V-ATPase complex, together with RpkA-RFP and observed areas of distinct co-localization in both cases ( Figure 1). Consistent with these findings RpkA-GFP containing membranes enclose LysoTracker positive compartments which have an acidic pH, namely late endosomes, lysosomes and maturing phagosomes. Furthermore, the common lysosomal antigen CLA, a carbohydrate epitope detected by mAb 173-185-1 [28], is present in several RpkA-GFP positive vesicles. Moreover, RpkA has recently been detected as part of the phagosome in a proteomic approach [29]. Little co-localization was observed between RpkA-GFP and vacuolin stained compartments, that represent post-lysosomal endosomes of neutral pH [30]. In contrast, there was a high degree of co-localization on internal membranes with p80, a putative copper transporter known to reside at the plasma membrane as well as throughout the whole endocytic transit (Figure 1) [31]. Thus, we conclude that RpkA is not a component of the plasma membrane but is rather present on phagosomes and a subpopulation of vesicles that are frequently acidic as well as positive for p80 and to some degree for V-ATPase.

RpkA is recruited to phagosomes
Since RpkA-GFP locates to phagosomes, we analyzed the timing of RpkA-GFP association with the phagosome during the uptake of yeast particles. Specifically, we wanted to determine whether the protein becomes part of the plasma membrane as a component of the phagocytic cup or if it is acquired later during maturation of the phagosome. AX2 (wild type) cells expressing RpkA-GFP were incubated with TRITC-labeled yeast and progress of phagocytosis was analyzed by confocal microscopy. We observed RpkA-GFP on vesicles of different diameters. These vesicles can be highly dynamic and approach the plasma membrane region forming the phagocytic cup, but they do not detectably fuse with the phagocytic cup ( Figure 2, 0 and 2.3 sec, white arrow heads). From 91 sec onward RpkA-GFP is detectable in the phagosomal membrane. After approximately 60 seconds (Figure 2, 62 and 91 sec) RpkA-GFP containing vesicles (white arrow heads) start to fuse with the phagosome, thus suggesting a mechanism of directed RpkA-GFP delivery to the phagosomal membrane. A similar mode has been described for the V-ATPase [32]. At later time points RpkA-GFP staining is enhanced and the protein remains on the phagosome until the end of the image recordings (48 min). Thus, RpkA is specifically acquired by the phagosome during its maturation process.

Loss of RpkA leads to a reduced phagocytosis rate
To study the impact of RpkA on phagocytosis we quantified the uptake of yeast cells in AX2, rpkA 2 , and rpkA 2 expressing RpkA-HA (D. discoideum rpkA 2 rescue strains 1D9 and 1E7) over time using TRITC-labeled yeast. We found that the uptake of yeast cells in rpkA 2 cells was reduced at every time point when compared with wild type cells. On average rpkAcells had taken up less than two yeast particles at 45 or 60 min ( Figure 3). Also, after 45 min rpkA 2 cells did not take up any further yeast cells, whereas AX2 cells engulfed one more cell on average. The rescue strains show an improved phagocytosis compared to rpkA 2 , incorporating ,2.3 yeast particles per cell compared to 1.5 for rpkA 2 at 60 min. However they reached only ,74% of the wild type level which could be due to differing levels of RpkA-HA protein.

Loss of RpkA affects survival of Legionella
Since RpkA is a component of phagosomal membranes which is acquired during the maturation process of the phagosome we wanted to know if RpkA has an impact on innate immunity related aspects like infection with L. pneumophila or autophagy.
To get insight into a possible role of RpkA in L. pneumophila infection, we investigated whether dead TRITC-labeled L. pneumophila co-localize with RpkA positive phagosomes. Indeed, after incubation of AX2 cells expressing RpkA-GFP with rhodamine-labeled L. pneumophila, bacteria are found in phagosomes positive for RpkA-GFP ( Figure 4A). Next we tested if live L. pneumophila are taken up into RpkA-positive vesicles. Therefore we incubated rescue strain 1E7 expressing RpkA-HA with unlabeled wild type L. pneumophila, fixed and stained for the HA-tag and VatA. We observed that whenever a L. pneumophila containing phagosomes was positive for RpkA-HA it was also positive for VatA and vice versa ( Figure 4B). Next we wanted to know if the loss of RpkA influences the uptake and/or replication of L. pneumophila by carrying out infection studies with live L. pneumophila. Wild type, rpkA 2 and rescue strains 1D9 and 1E7 were infected with L. pneumophila. After removal of extracellular bacteria, internalized Legionella were quantified. The quantification was done at 0h, 24h and 48h post infection. No initial difference was seen for uptake of bacteria between strains ( Figure 4C, 0h). After 48h the L. pneumophila content in rpkA 2 was 13 times higher than in AX2 ( Figure 4C, 48 h). The rescue strains again showed an intermediate behavior. Thus the loss of RpkA does not influence the uptake of L. pneumophila, but the engulfed bacteria reach significantly higher titers in the absence of RpkA.
This difference becomes even more prominent if L. hackeliae is employed which is less virulent compared with L. pneumophila [33].
In human macrophages L. hackeliae replicates and causes pneumonia, whereas in amoebae it does not replicate and is killed. In AX2 within 34 hours the killing of L. hackeliae is completed, whereas in rpkA 2 it is significantly delayed and bacteria are still alive after 48 hours ( Figure 4D).

RpkA-GFP interacts with the V-ATPase complex
Since RpkA is recruited to maturing phagosomes with the same kinetics as the V-ATPase complex we wanted to investigate if RpkA not only co-localizes with the V-ATPase but also interacts with this complex. Therefore, RpkA-GFP was immunoprecipitated from cell lysates, obtained proteins were separated by SDS-PAGE, and analyzed by mass spectrometry. We identified the subunits C and M of the V-ATPase in the immunoprecipitate ( Figure 5A). The interaction with V-ATPase was further verified since we found VatA and VatM-GFP to co-precipitate with GST-PIPK 343 -805 (residues 343 -805) and GST-PIPK 370 -828 (residues 370 -828). Thus, we conclude that the PIPK domain of RpkA is responsible and sufficient for this interaction ( Figure 5B).

RpkA does not affect the overall endosomal pH
Since RpkA co-localizes and directly interacts with the V-ATPase, we investigated the influence of the loss of RpkA on the endosomal pH and incubated AX2, rpkA 2 and rescue strain 1E7 cells with FITC-Dextran and measured the endosomal pH. We observed an average pH of 5.3 for AX2 which is in agreement with published values [34]. The pH determined for rpkA 2 and 1E7 cells was similar (pH 5.2) which indicates that RpkA does not affect the overall endocytic pH ( Figure 5C). Furthermore, L. pneumophila is known to inhibit the acquisition of V-ATPase which is responsible for establishing low pH values.

RpkA affects the phosphoinositide metabolism of the cell
A characterization of the PIPK of RpkA as for any other PIPK of a GPCR-PIPK is lacking. Neither the substrates nor the products are known. The PI-kinase activity of RpkA might be one factor determining the resistance of D. discoideum towards L. pneumophila infection as phosphoinositides are also known to play a major role in L. pneumophila infection and previous work showed that the bacteria can subvert the host's phosphoinositide turnover [35]. We wanted to approach the function of the PIPK of RpkA by in vitro and in vivo studies. First we tested the ability of the PIPKdomain of RpkA to bind to different phosphoinositides in vitro. In this study we expressed GST-PIPK 370-828 in E. coli and performed a dot blot overlay assay to assess its binding ability to lipids (PIP-Strip), enabling the detection of the PI-kinase substrate [36]. GST-PIPK 370 -828 bound preferentially to monophosphorylated PIs especially to PI3P and PI4P, consistent with a role in the generation of PIP 2 from PI3P or PI4P ( Figure 6A). GST-PIPK 370-828 was also able to bind to phosphatidylserine. GST-PIPK 343-805 , on the contrary did not exhibit any lipid binding (data not shown), implicating that the last 23 residues are important for the lipid binding. With an isoelectric point of 4.0 these amino acids do not contribute to a general affinity to the negatively charged PIPs but they might stabilize the PIP binding domain of the PIPK. It is on the other hand as well conceivable but less probable that the additional 27 residues on the N-terminus of GST-PIPK 343 -805 inhibit the binding to the PIPs.
GST alone did not bind to any of the tested lipids indicating that the binding of the PIPK 370 -828 to the lipids is specific ( Figure 6A).

Loss of RpkA leads to reduced levels of phosphoinositides
To get an impression of the relevance of RpkA for the phosphoinositide metabolism of the cell we investigated the consequences of the loss of RpkA on phosphoinositide turnover by metabolic labeling of phospholipids using [c-32P] ATP in vivo [37]. In rpkA 2 cells the turnover of monophosphorylated phosphoinositides (PIP) as well as bisphosphorylated phosphoinositides (PIP 2 ) was reduced to 70% and 44% of the wild type (AX2) cells, respectively ( Figure 6B). This was surprising because, assuming that RpkA is a PI4P5K, we expected that a loss of this enzyme would simply lead to an increase of the amount of PI4P (substrate) and a decrease of PI(4,5)P 2 (product).

Nitrogen starvation tolerance is reduced in rpkA 2 cells
Autophagy is a pathway which is involved in cell autonomous defense and helps to eliminate pathogenic bacteria that reside in the cytosol of the host cell through lysosomal degradation [38]. One of the earliest steps in autophagy is the activation of a specific class III phosphatidylinositol-3-OH kinase (PI3K) complex and the formation of phosphatidylinositol-3-phosphate (PI3P) in ER membranes which recruits proteins required for the formation of the autophagosome [39,40]. Based on our findings that the PI metabolism is altered in the rpkA 2 strain and on the observation that mutants deficient for autophagy related genes show similar defects in development [41,42], we assessed autophagy by testing the ability of the mutant to survive in the absence of an exogenous nitrogen source (nitrogen starvation assay) as autophagy is also strongly induced by nitrogen starvation [43,44]. We found that under such conditions the cell numbers of AX2 stayed nearly constant over three to four days and then decreased slowly. In contrast, cell numbers of the rpkA 2 strain significantly decreased from day three onward. AX2 and rpkA 2 started with approximately the same cell density of ,2.8610 6 cells/ml at day 0. After 6 days AX2 cultures had a density of 1.25610 6 cells/ml whereas rpkA 2 cultures had a 21-fold lower density (6610 4 cells/ml). Thus, the rpkA 2 mutant apparently cannot survive an extended period of nitrogen starvation (Figure 7).

Discussion
GPCRs are generally known to be transported along the secretory pathway to the plasma membrane where they are active.   Upon stimulation they can be internalized and can either cycle back to the plasma membrane or are sorted to late endosomes and can further be degraded within lysosomes.
We show that RpkA is delivered to phagosomes with similar kinetics as it has been published for the V-ATPase [32]. RpkA and the V-ATPase complex do not completely overlap in their localization, e.g. they do not co-localize in the contractile vacuole. In this compartment we only find the V-ATPase and RpkA is present on vesicles which are free of V-ATPase. However there are membranous compartments where both of them come together and can interact and one of these compartments is the maturing phagosome. To our knowledge RpkA is so far the first GPCR which is specifically associated with the maturing phagosome.
RpkA is also an interaction partner of the V-ATPase. The V-ATPase complex interacts with full length RpkA as well as with the PIPK 343 -805 and PIPK 370 -828 domain alone.
Loss of RpkA results in a reduced phagocytosis rate of yeast cells, whereas the uptake of Legionella does not differ in rpkAand AX2 cells. This might be due to the difference in the uptake mechanism of yeast and Legionella. Both can be phagocytosed, however Legionella might be taken up primarily via macropinocytosis by D. discoideum [45]. Furthermore, the bacteria are taken up by coiling phagocytosis and foremost can induce their uptake since they can also infect HeLa cells and other nonprofessional phagocytes [46,47,48,49].
Although the uptake of L. pneumophila is comparable in AX2 and rpkA 2 replication is significantly altered. L. pneumophila reaches 13 times higher numbers in rpkA 2 compared to AX2. The difference between wild type and mutant is even more intriguing regarding the less virulent L. hackeliae which survives far longer in rpkA 2 .
One reason for this effect may be that L. pneumophila, the most pathogenic Legionella strain, is able to manipulate even wild type D. discoideum so drastically that the difference between AX2 and a mutant is less obvious as it is in the case of a less harmful strain like L. hackeliae. Here the difference between AX2 and the mutant rpkA 2 becomes more prominent. AX2 is able to sustain the manipulation of L. hackeliae, whereas rpkA 2 has major problems to kill the bacteria. In macrophages which are more susceptible than rpkA 2 L. hackeliae can even replicate [33]. The intermediate behavior of the rescue strains is most probably due to heterogeneous expression of RpkA-HA. Although 1E7 and 1D9 are of single clone origin the expression pattern varies within the cell population as observed by immunofluorescence.
Since RpkA interacts with the V-ATPase at the phagosome one of the reasons for the significantly higher Legionella titer in the rpkA 2 might be an elevated phagosomal or endosomal pH which we however did not detect. Although the overall endosomal pH is unaltered in rpkA 2 we cannot rule out that RpkA has an influence on early events of pH changes during phagocytosis like a slight retardation of the pH drop.
L. pneumophila inhibits normal phagosomal maturation by translocating effector molecules through the Dot/Icm type IV secretion system from the LCV into the host cell's cytosol [50]. However, the effector molecules responsible for the arrest of  phagosomal maturation are still poorly understood. Recently, an effector protein has been discovered that inhibits the V-ATPase. If lysosomal proteins like the V-ATPase are inhibited solely by effector molecules secreted by L. pneumophila or if they are excluded from the LCV is still under debate [51,52,53,54]. Phosphoinositides are instrumental in the deployment of the phagosomal antimicrobial defense [55]. They are involved in determining the identity of membranous compartments. The plasma membrane predominantly contains PI(4,5)P 2 , whilst the Golgi apparatus holds PI4P, late endosomes and lysosomes harbor PI(3,5)P 2, whereas early endosomes and phagosomes contain PI3P [56]. These phosphoinositides serve as anchors for specific proteins [57]. The phosphoinositide-bound proteins can enable the recruitment of further proteins or catalyze enzymatic reactions to establish a compartment-specific environment. Thus, the true nature of a phagosome can be concealed by a pathogen through interfering with the phosphoinositide composition by either prohibiting the acquisition or synthesis of PI3P, by promoting the degradation of PI3P or by binding PI3P with other proteins. This process of organelle disguise, by pathogens like L. pneumophila, has been designated ''identity theft'' [10].
One example for such a phosphoinositide is PI3P, a marker of early phagosomes. Legionella can replicate better in D. discoideum cells upon inhibition of PI3Ks or when PI3K genes are disrupted or PI3Ks inactivated [45,58] indicating Legionella can more easily camouflage the identity of the phagosome if less PI3P is present on the phagosome. L. pneumophila secretes several proteins that bind PI3P and PI4P. LpnE for instance, is a secreted protein that binds to PI3P, and is involved in inhibiting the acquisition of lysosomal markers and phagosomal maturation, while SidC binds to PI(4)P on the LCV membrane [35,59].
Based on our findings that loss of RpkA significantly lowers the PIP and PIP 2 levels in the mutant, phosphoinositide composition of the phagosome as well as vesicle trafficking towards the phagosome may be crucially altered in the mutant allowing facilitated establishment of a replicative vacuole. GST-PIPK 370 -828 is able to bind to phosphoinositides and phosphatidylserine as has been shown for other PIPKs [36,60].
However, which phosphoinositide species are generated by GPCR-PIPKs has not been shown to date. It has been shown that the PIPKs of all the GPCR-PIPKs do not cluster with the known PIPKs ( Type I/II or III) but instead cluster in a new group (Type IV) [22]. It might well be that RpkA is directly involved in the generation of PIP and PIP2 as both species are diminished in rpkA 2 .
Autophagy is well known to play a role in innate immunity against cytoplasmic pathogens which are internalized into autophagosomes similar to mitophagy. The role of autophagy in infection with Legionella has been discussed controversially. Autophagy was initially assumed to be a favorable pathway exploited by L. pneumophila [61]. Later it was thought to play no role at all since in D. discoideum loss of ATG7 and ATG8 had no consequences for the replication of L. pneumophila [62]. Recently it has been reported that ATG7 and ATG8 localize to the LCV. The acquisition of these autophagy proteins to the LCV is elevated in mouse macrophages that are restrictive to L. pneumophila infection, in contrast to their acquisition in permissive mouse macrophages [63]. This suggests that autophagy is a mechanism of defense against the establishment of LCVs which is also supported by recent results in D. discoideum showing that loss of ATG9 leads to a lower clearance and higher replication rates of L. pneumophila [64].
Based on our findings RpkA is an endosomal GPCR-PIPK which is recruited to phagosomes during their maturation. It interacts with the V-ATPase and is involved in phagocytic processes. Its loss leads to a reduced resistance against the pathogen L. pneumophila. The protein seems to be ancient since it is conserved in several phylogenetically distant species. Till now RpkA homologs seem to be restricted to lower eukaryotes. The presence of an RpkA homolog in A. castellani is intriguing as this organism is an established primary host of L. pneumophila. Further analysis of the role of RpkA in A. castellani may provide us with tools for interfering with the environmental reservoir of L. pneumophila.

Phagocytosis assays and Legionella infection
Phagocytosis was assayed on a substratum where the cells were allowed to settle on coverslips and yeast cells (,20 yeast cells/ Dictyostelium cell) were added. After the indicated times the cells were fixed in methanol, and embedded. Approximately 150 cells per strain and time point were analyzed for uptake of yeast particles in two independent experiments [68]. Infection with L. pneumophila was done as described with the exception that L. pneumophila JR32 Phil was used for the assays [69]. L. pneumophila JR32 Phil was cultured on BCYE plates (buffered charcoal yeast extract agar for 3 days at 37uC and a CO 2 concentration of 5%. The bacteria were harvested in 1 ml of Soerensen buffer and adjusted to a density of 5610 6 colony forming units/ml. D. discoideum cells of a 3 day old culture were harvested (200 g, 7 min, RT) and resuspended in the same volume of infection medium (Soerensen buffer/HL5 1:1). Cells were seeded into 25 cm 2 culture flask and the volume was adjusted to 5 ml with freshly mixed infection medium. The final cell density was 5610 5 cells/ml. Before infection the cells were allowed to adhere for 30 min. The medium was removed from the cells and replaced by 5 ml of infection medium with bacteria, multiplicity of infection (MOI) of 10. Following an invasion period of 3 hours (infection time), the remaining extracellular bacteria were killed by a gentamicin treatment (100 mg/ml). After 50 min incubation at 25.5uC, the Dictyostelium cells were washed with 5 ml of Soerensen buffer. Then 5 ml of infection medium was added to each flask. For each time point cells were resuspended and 300 ml of the suspension was lysed by centrifugation (7 min, 20,0006 g) and vigorous shaking. Serial dilutions of these lysates were plated on BCYE agar.
For the L. hackeliae infection serotype 1 was used (ATCC 35250) [70]. The infection assays were done four times in triplicates.

GST-fusion protein expression and purification
For protein expression E. coli BL21 (DE) and XL1 blue were used. Induction of protein expression was induced with 0.25 mM isopropyl b-D-thio-galactoside (IPTG) when an OD 600 of 0.8 was reached. Cells were further cultured at 30uC for 3 hours. They were harvested, lysed in 50 mM Tris-HCl, pH 7.4 to 8.0, 100 mM NaCl, supplemented with Protease inhibitors (0.5 mM PMSF, 1 mM Benzamidine and Complete (Roche) and 1 mM DTT with an EmulsiFlex cell homogenizer. Lysates were separated into soluble and insoluble fractions by centrifugation at 18,000 g. The fusion proteins from the soluble fraction were purified using GST-Sepharose beads (GE Healthcare).

Phosphoinositide-binding assay
Phosphoinositide-binding assay using lipid strips supplied by Echelon Biosciences, Inc. (Salt Lake City, Utah, USA) was performed following the protocol of Echelon. Briefly, GST and GST-fusion proteins were eluted from the beads using 20 mM glutathione in TBS-T (50 mM Tris/HCl pH7.2, 100 mM NaCl with 0.2% Tween-20).
The membranes were blocked with 0.1% ovalbumin (Sigma # A-5253) in TBS for one hour at room temperature. After discarding the blocking solution membranes were incubated with 1 mg/ml protein (GST-PIPK 370 -828 or GST) in TBS-T at 4uC over night. Then the protein solution was discarded and membranes were washed with TBS-T three times 10 minutes each. Protein binding was detected by western blot analysis with polyclonal GST antibodies as primary and anti-rabbit IgG (Sigma # A-6154) as secondary antibody.

Nitrogen starvation assay
Nitrogen starvation assay was done as described [71]. Briefly, strains were incubated submerged for one day in FM Medium (ForMedium Ltd, UK). After washing away dead cells the cells were transferred to shaking culture flasks and incubated for two days in FM Medium. Then cells were harvested and washed two times with amino acid free FM Medium. Cells were adjusted to ,2610 6 cells/ml in 20 ml amino acid free FM Medium. Samples were taken at the indicated time points, diluted in Soerensen with 20 mM EDTA and incubated on ice until they were present as single cells. Then serial dilutions were plated on SM plates with Klebsiella. After 5 days the D. discoideum colonies were counted.

Determination of the endosomal pH:
Endosomal pH was determined according to [34]. Briefly, cells were grown to 2-5610 5 cells/ml, harvested and resuspended at a concentration of 3610 6 cells/mL in fresh axenic medium and loaded with FITC-dextran (2 mg/ml) (70 000 Mr, Sigma-Aldrich). Basal endosomal pH was measured after loading for 3 h. Cells were collected by centrifugation, washed in 50 mM MES buffer, pH 6.5, then resuspended in 206MES buffer and the fluorescence intensity was measured using an infinite M 1000 device (Tecan) equipped with Tecan i-control (version 1.6.19.2). The fluorescence excitation ratio (I495/I450) was calculated after subtraction of the background fluorescence. The endosomal pH was then determined from a standard curve.

Pull down and immunoprecipitation assays
For each pull down and immunoprecipitation experiments 5610 7 cells were lysed in 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 0.5% NP40, supplemented with Protease inhibitors (0.5 mM PMSF, 1 mM Benzamidine and Complete (Roche) by passing them 10 times through a 27 G syringe and 2620 sec incubation in a sonication bath. Then cells were incubated in agitation (1000 rpm/min) for 15 min at 4uC followed by a centrifugation step at 8,0006 g for 5 min. The supernatant was pre-cleared by incubation with protein A beads for 45 min. Precleared lysates were incubated with the indicated antibodies coupled to protein A beads or with GST and GST fusion proteins. After incubation for 3 h or overnight the beads were washed 36 with lysis buffer and the supernatant was completely removed with a Hamilton syringe. The beads were resuspended in 50 ml of SDSbuffer and after incubation for 5 min at 95uC the proteins were separated via SDS-PAGE. Mass spectrometry analysis of coimmunoprecipitated proteins by LC-MS/MS was performed by the CMMC service facilities. Figure S1 Characteristics of RpkA-HA. (A) RpkA-HA localizes to yeast phagosomes. 1E7 cells were incubated with TRITC labeled yeast for 15 min and fixed with methanol (220uC) for 25 min. The cells were incubated with anti-HA-tag antibody 3F10. As secondary antibody goat-anti-rat-IgG conjugated to Alexa 488 was used. Scale bar, 5 mm. (B) RpkA-HA rescues the developmental phenotype of rpkA 2 cells. 5610 7 cells of Ax2, rpkA 2 and of the two rescue strains 1D9 and 1E7 (rpkA 2 expressing RpkA-HA) were plated on plates with Klebsiella lawn and photographed after 5 days. Scale bar, 1 mm. (TIF)