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

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 C-terminal 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 co-localization 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⁻, and rpkA⁻ expressing RpkA-HA (D. discoideum rpkA⁻ rescue strains 1D9 and 1E7) over time using TRITC-labeled yeast. We found that the uptake of yeast cells in rpkA⁻ cells was reduced at every time point when compared with wild type cells. On average rpkA^- cells had taken up less than two yeast particles at 45 or 60 min (Figure 3). Also, after 45 min rpkA⁻ 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⁻, incorporating ∼2.3 yeast particles per cell compared to 1.5 for rpkA⁻ 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⁻ 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⁻ 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⁻ 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⁻ 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⁻ 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 PIPK-domain 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 PIP₂ 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 [γ-32P] ATP in vivo [37]. In rpkA⁻ cells the turnover of monophosphorylated phosphoinositides (PIP) as well as bisphosphorylated phosphoinositides (PIP₂) 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)P₂ (product).

Nitrogen starvation tolerance is reduced in rpkA⁻ 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⁻ 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⁻ strain significantly decreased from day three onward. AX2 and rpkA⁻ started with approximately the same cell density of ∼2.8×10⁶ cells/ml at day 0. After 6 days AX2 cultures had a density of 1.25×10⁶ cells/ml whereas rpkA⁻ cultures had a 21-fold lower density (6×10⁴ cells/ml). Thus, the rpkA⁻ mutant apparently cannot survive an extended period of nitrogen starvation (Figure 7).

Growth, Transformation and Development

Cells were either grown on a lawn of K. aerogenes on SM agar plates or cultivated in shaking suspension (160 rpm) or in a submerged culture at 21–23°C in axenic medium [65]. Development was initiated by plating 5×10⁷ cells which were washed twice with Soerensen phosphate buffer (17 mM Na⁺/K⁺ phosphate, pH 6.0) on phosphate agar plates and monitored. Mutants were maintained in the presence of appropriate antibiotics (2–4 µg/ml G418 (Roche Applied Science) or 3–5 µg/ml blasticidin (MP Biomedicals Inc., Eschwege, Germany)). The following strains have been used; AX2-214 (wild type) [66], AX2 expressing GFP-tagged RpkA (RpkA-GFP, GFP fused to the C-terminus of RpkA) or HA-tagged RpkA (RpkA-HA, HA-tag fused to the C-terminus of RpkA), rpkA^- [21] and rpkA^- rescue strains 1E7 and 1D9 expressing RpkA-HA, AX2 expressing RFP-tagged RpkA (RpkA-RFP, mRFPmars [67] fused to the C-terminus of RpkA).

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 37°C and a CO₂ concentration of 5%. The bacteria were harvested in 1 ml of Soerensen buffer and adjusted to a density of 5×10⁶ 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 cm² culture flask and the volume was adjusted to 5 ml with freshly mixed infection medium. The final cell density was 5×10⁵ 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 µg/ml). After 50 min incubation at 25.5°C, 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 µl of the suspension was lysed by centrifugation (7 min, 20,000× 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 β-D-thio-galactoside (IPTG) when an OD₆₀₀ of 0.8 was reached. Cells were further cultured at 30°C 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).

Immunofluorescence analysis

Antibodies have been listed in Bakthavatsalam et al. (2007) except for HA-tag antibody 3F10 (Roche Diagnostics, Mannheim, Germany) and mAb 173-185-1 [28]. Fixation of cells was done with methanol (–20°C for 25 min). For labeling acidic compartments LysoTracker Red was used (Invitrogen-Molecular Probes). For live cell imaging cells were monitored using a confocal microscope Leica TCS SP5 (Leica, Wetzlar, Germany).

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 µg/ml protein (GST- PIPK_{370 – 828} or GST) in TBS-T at 4°C 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 ∼2×10⁶ 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.

Phospholipid labeling

A saponin-based cell permeabilization protocol for Dictyostelium was adapted to measure phospholipid labeling in wild type (AX2) and rpkA⁻ cells [37], [72]. Briefly, AX2 cells were developed for 5 h as previously described [73], transferred to still dishes (2.5 cm), allowed to settle to give a confluent monolayer in KK2 (20 mM potassium phosphate buffer, pH 6.1). At regular time intervals, buffer was replaced with labeling solution (139 mM sodium glutamate, 5 mM glucose, 5 mM EDTA, 20 mM PIPES pH 6.6, 1 mM MgSO₄·2H₂O, 0,25% (w/v) saponin, 1× phosphatase inhibitor cocktails 1 and 2 (Roche Ltd.), and 1 µCi/ml [γ-³²P]ATP (Perkin Elmer Ltd.). Following a 6 min incubation, labeling solution was removed, cells were lysed in acidified methanol and phospholipids were separated as previously detailed [74]. Phospholipid labeling was quantified using a Typhoon PhosphorImager. Even loading was determined using total lipid stain with copper sulphate.

Determination of the endosomal pH:

Endosomal pH was determined according to [34]. Briefly, cells were grown to 2–5×10⁵ cells/ml, harvested and resuspended at a concentration of 3×10⁶ 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 20× MES 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 5×10⁷ 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 2×20 sec incubation in a sonication bath. Then cells were incubated in agitation (1000 rpm/min) for 15 min at 4°C followed by a centrifugation step at 8,000× g for 5 min. The supernatant was pre-cleared by incubation with protein A beads for 45 min. Pre-cleared 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 3× with lysis buffer and the supernatant was completely removed with a Hamilton syringe. The beads were resuspended in 50 µl of SDS-buffer and after incubation for 5 min at 95°C the proteins were separated via SDS-PAGE. Mass spectrometry analysis of co-immunoprecipitated proteins by LC-MS/MS was performed by the CMMC service facilities.