Abstract
Because NF-κB signaling pathways are highly conserved in evolution, the fruit fly Drosophila melanogaster provides a good model to study these cascades. We carried out an RNA interference (RNAi)-based genome-wide in vitro reporter assay screen in Drosophila for components of NF-κB pathways. We analyzed 16,025 dsRNA-treatments and identified 10 novel NF-κB regulators. Of these, nine dsRNA-treatments affect primarily the Toll pathway. G protein-coupled receptor kinase (Gprk)2, CG15737/Toll pathway activation mediating protein, and u-shaped were required for normal Drosomycin response in vivo. Interaction studies revealed that Gprk2 interacts with the Drosophila IκB homolog Cactus, but is not required in Cactus degradation, indicating a novel mechanism for NF-κB regulation. Morpholino silencing of the zebrafish ortholog of Gprk2 in fish embryos caused impaired cytokine expression after Escherichia coli infection, indicating a conserved role in NF-κB signaling. Moreover, small interfering RNA silencing of the human ortholog GRK5 in HeLa cells impaired NF-κB reporter activity. Gprk2 RNAi flies are susceptible to infection with Enterococcus faecalis and Gprk2 RNAi rescues Toll10b-induced blood cell activation in Drosophila larvae in vivo. We conclude that Gprk2/GRK5 has an evolutionarily conserved role in regulating NF-κB signaling.
Nuclear factor-κB signaling is involved in a variety of cellular processes, including control of both the innate and adaptive immune systems. The NF-κB/Rel family of transcription factors consists of five members in humans. These proteins control the expression of hundreds of target genes, including various cytokines and chemokines, in a tightly regulated manner (1). In mammals, immune-related NF-κB activation mainly occurs via two signaling pathways, the TNFR pathway and the TLR pathway.
NF-κB signaling pathways are highly conserved in evolution, and therefore, similar signaling cascades are found in lower eukaryotes, such as the fruit fly Drosophila melanogaster. Drosophila systemic immune response is largely mediated by two NF-κB signaling cascades, the Toll and the immune deficiency (Imd) pathway, which closely resemble mammalian TLR and TNFR signaling cascades, respectively (2). Both signaling cascades lead to activation and nuclear localization of Drosophila NF-κB family protein and expression of a distinct but overlapping set of antimicrobial peptide genes (3–5). Thus, lacking adaptive immunity, the fruit fly makes a useful and simpler model to study the signaling cascades and their involvement in innate immune responses.
The Drosophila Imd pathway is activated in response to Gram-negative bacterial infection. After ligand binding to the receptor peptidoglycan recognition protein (PGRP)-LC (6–8), the signal proceeds via downstream components, which include the death-domain protein Imd (9), the MAPK kinase kinase TGF-β–activated kinase 1 (Tak1) (10), and a Drosophila homolog of Fas-associated death domain protein (FADD) (11). The signaling leads to activation of Drosophila IκB kinases Kenny and immune response deficient 5 (12), which phosphorylate the inhibitory domain of the NF-κB family transcription factor Relish (13), resulting in Relish cleavage by the caspase Dredd (14–16). Subsequently, the activated N-terminal 68-kDa Relish is translocated into the nucleus, where it activates transcription of antimicrobial peptide genes. In addition, inhibitor of apoptosis 2 and TGF-β–activated kinase 1-associated binding protein 2 (Tab2) are shown to play a key part in the regulation of Relish activity (17–20).
The Toll pathway is activated by Gram-positive bacteria and fungi (3) recognized by several pattern recognition receptors (2) including PGRP-SA (21), leading to proteolytic cleavage and activation of the cytokine Spätzle and its binding to the Toll receptor (22, 23). Intracellular components of Toll pathway include the death domain proteins Drosophila MyD88 (24), Tube, and Pelle (3). Finally, the signaling leads to degradation of the NF-κB inhibitory protein Cactus and nuclear localization of Dorsal-related immunity factor (Dif) and/or Dorsal (25). It has also been shown that Toll signaling is involved in the activation of the cellular immune system (26, 27). It is likely that components yet to be found are involved in regulating the cascade.
To identify novel gene products involved in Drosophila NF-κB signaling, we carried out a genome-wide screen for 16,025 dsRNAs using a Drosomycin luciferase reporter-based assay that enables us to monitor both the Toll and Imd pathways. We identified 10 novel NF-κB regulators, of which 9 act primarily on the Toll pathway. Furthermore, we identified G protein-coupled receptor kinase (Gprk)2/GRK5 as an evolutionarily conserved regulator of NF-κB signaling.
Materials and Methods
Drosophila dsRNA libraries and synthesis of targeted dsRNAs
The dsRNAs used in the RNA interference (RNAi) screen (16,025) were produced from a commercial Drosophila
GFP dsRNA.S2 cell treatments and reporter assays
S2 cell culture, transfections, dsRNA treatments, and reporter assays for the Toll and Imd pathways were performed essentially as previously described (17, 28). For the genome-wide screen, S2 cells were transfected with 0.1 μg Drosomycin luciferase (29) and 0.1 μg Actin 5C–β-galactosidase (Act5C–β-gal) reporter plasmids. In addition, the cells were treated with 0.5 μg dsRNAs. To screen both Toll and Imd pathways simultaneously, the Toll pathway was first activated by transfecting S2 cells with 0.1 μg Toll10b construct. Forty-eight hours posttransfection and 24 h prior to measurements, the Imd pathway was activated by adding heat-killed Escherichia coli.
S2 cell transfections for quantitative RT-PCR
For quantitative RT-PCR (qRT-PCR) experiments, S2 cells were seeded on 24-well plates and transfected with 0.1 μg Toll10b construct and 0.5 μg dsRNA. Seventy-two hours later, cells were harvested and lysed in TRIsure reagent (Bioline, London, U.K.) by pipetting up and down 10 times. Total RNAs were extracted according to the manufacturer’s instructions and RNAs subjected to quantitative RT-PCR analysis as detailed below.
HeLa cell culture and transfections
HeLa cells were grown in DMEM plus GlutaMAX (Gibco/Life Technologies, Carlsbad, CA) with 10% FBS, 1% nonessential amino acids (Sigma-Aldrich, St. Louis, MO), 100 U/ml penicillin, and 100 μg/ml streptomycin. For transfection, 6 × 104 cells per well were seeded onto 24-well plate. Twenty-four hours later, the cells were transfected with 0.1 μg NF-κB luciferase, 0.05 μg CMV–β-galactosidase reporter plasmid, and 50 pmol small interfering RNAs (siRNAs) (Ambion, Austin, TX) using Lipofectamine transfection reagent (Invitrogen/Life Technologies) and Opti-MEM medium (Life Technologies). siRNAs used were as follows: GFP siRNA (Silencer GFP [eGFP]; catalog number AM4626, negative control), GRK5 siRNA (catalog number AM16704; ID 110898), and RelA (catalog number AM16704; ID 216912, positive control). Forty-eight hours posttransfection, NF-κB signaling was induced with 10 ng/ml TNF-α (Sigma-Aldrich), and 6 h later, luciferase and β-galactosidase activities were measured from the cell lysates.
Immunohistochemistry with Gprk2/GRK5 constructs
To create a Gprk2-GFP fusion protein, the full-length cDNA for Drosophila Gprk2 gene was amplified by PCR and cloned into the KpnI site of the modified Drosophila expression vector pMT/GFP/V5/His, a kind gift from Dr. I. Kleino (University of Helsinki, Helsinki, Finland). S2 cells were transfected with the Gprk2-GFP fusion construct essentially as described previously (30). Overexpression of Gprk2-GFP fusion protein in S2 cells was induced with 350 μM CuSO4 for 36 h. For HeLa cell transfection, cells were seeded onto coverslips on six-well plates, and 24 h later, the cells were transfected with 0.1 μg GRK5-GFP construct. Thirty-six hours later, the coverslips were mounted to slides with Vectashield mounting medium for fluorescence with 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA). Cells were imaged with an Olympus IX70 confocal microscope (Olympus, Tokyo, Japan) and analyzed with Andor iQ software (Andor Technology, Belfast, U.K.).
Zebrafish maintenance and morpholino gene silencing
AB wild-type zebrafish strain was maintained according to standard protocols (31). Translation-blocking morpholinos targeting zebrafish GRK5 (ZDB-GENE-060929-1198) (5′-GGCCACGATATTCTCAATCTCCATT-3′) and MyD88 (ZDB-GENE-040219-3) (5′-GGTCTATACTTAACTTTGATGCCAT-3′), as well as a mismatch-GRK5 control morpholino (5′-GCCCACCATATTGTCAATGTCGATT-3′), were obtained from GeneTools (Philomath, OR). A total of 1 nl 250 μM morpholinos in 0.2 M KCl was injected into the yolk sacs of AB wild-type embryos at the 1-2 cell stage.
Zebrafish infections
For infection experiments, E. coli was grown in Luria-Bertani broth until OD of 0.3 at 600 nm. Bacterial cells were pelleted by centrifugation (10,000 × g, 5 min), washed with 0.2 M KCl, pelleted, and diluted 1:2 in 0.2 M KCl. Prior to injection, 70 kDa rhodamine dextran tracer was added to the bacterial suspension. Morpholino-injected zebrafish larvae were manually dechorionated at 24–28 h postfertilization, after which 1 nl prepared E. coli suspension was injected into the yolk. Before and after the injections with each needle, one injection dose was plated for checking the bacterial quantity. Infected larvae were kept at 28°C for 2–24 h postinfection, after which they were snap-frozen for total RNA extraction. Total RNAs were extracted according to standard procedures.
Coimmunoprecipitation
S2 cells were transfected with Gprk2-V5 full-length or deletion constructs and Cactus-myc constructs in pMT/V5/HisA vector (Invitrogen/Life Technologies) and coimmunoprecipitated, separated, transferred on the membrane, and detected essentially as described (30).
Stable S2/epidermal growth factor receptor-Toll cells, Western blotting, and quantification
S2 cells with stable integration of a chimeric epidermal growth factor receptor (EGFR)-Toll construct were made according to Ref. 32, and the response to Toll signaling by EGF was verified with Drosomycin luciferase construct. Stable S2/EGFR-Toll cells were grown in six-well culture dishes and treated with 15 μg dsRNA in a total volume of 3 ml medium for 4 d. Induction of the Toll pathway was done by addition of EGF (0.5 μg/ml) (Molecular Probes) for 30 min. Cytoplasmic extracts of S2-EGF/Toll cells were separated by electrophoresis, transferred to a Hybond-P membrane (GE Healthcare Life Sciences), and blocked. Cactus protein on the membrane was detected with polyclonal rabbit anti-Cactus Ab and HRP-linked donkey anti-rabbit IgG (GE Healthcare Life Sciences, Uppsala, Sweden). Rabbit polyclonal anti-GM130 Ab (Abcam, Cambridge, MA) targeting a Drosophila Golgi protein GM130 was used as a loading control. Band quantifications were done with Adobe Photoshop 7 software (Adobe Systems, San Jose, CA) as follows: to obtain the absolute intensity, the mean value of each band was multiplied by the pixel value. The relative intensity was calculated by normalizing absolute intensities with the absolute intensity of the negative control, which was set to the value of 1. Quantifications were carried out on three separate Western blots.
Fly stocks and maintenance
Drosophila stocks were kept on a standard mashed potato diet at room temperature or at 25°C. C564-GAL4 flies express GAL4 in the adult fatbody; the P{UAS-Tl10b:11} stock carries a Toll10b insert on the X chromosome and the hemolectinΔ (hmlΔ)-GAL4, UAS-GFP stock constitutively expresses GFP in the majority of blood cells (33). The upstream activating sequence (UAS)-RNAi fly stocks listed in Supplemental Table I were obtained from the Vienna Drosophila RNAi Center [VDRC; Vienna, Austria (34)] or the Kyoto Fly Stocks of the National Institute of Genetics (NIG-FLY) (Drosophila Genetic Resource Center, Kyoto Institute of Technology, Kyoto, Japan). The C564-GAL4 flies were crossed with UAS-RNAi flies, and the adult flies carrying one copy of the UAS-RNAi construct and one copy of the GAL4 driver were used in infections. UAS-RNAi flies crossed to w1118 were used as controls in infection experiments.
Fly infection and RNA extraction
To produce induced Drosomycin expression via septic injury, flies were pricked with a thin tungsten needle previously dipped in a concentrated culture of Micrococcus luteus and grown at 25°C. Twenty-four hours later, five flies per sample were collected and snap-frozen in dry ice. Alternatively, expression of Toll pathway target genes was induced by natural fungal infection with Beauveria bassiana at 29°C for 48 h as previously described (35), after which the flies were collected as mentioned above. Total RNAs were extracted according to standard procedures and RNAs subjected to qRT-PCR analysis as detailed below.
qRT-PCR
Extracted total RNAs from S2 cells, zebrafish embryos, or flies were used in qRT-PCR experiments. qRT-PCR for expression levels of chosen genes was carried out from dilutions of the extracted RNAs using the QuantiTect SYBR Green RT-PCR Kit (Qiagen, Hilden, Germany) and the ABI7000 instrument (Applied Biosystems, Foster City, CA). Primers and sizes of PCR products are listed in Supplemental Table II.
Fly survival experiments
To assess Toll pathway-mediated immunity, flies were first immunized by pricking them with M. luteus as described above (M. luteus infection activates the Toll pathway). Twenty-four hours later, the flies were infected with Enterococcus faecalis by pricking as above. The infected flies were kept at room temperature, and their survival was monitored for 24 h. For Imd pathway-mediated immunity, flies were pricked with a thin tungsten needle previously dipped in a concentrated culture of Enterobacter cloacae (a Gram-negative bacterium), and their survival was monitored for 48 h.
Fly larvae in vivo experiments
To assess the distribution of blood cells in Drosophila larvae in vivo, parental crosses were kept for 2 d at 29°C in stained mashed potato food, which permits the staging of larval progeny as described previously (27). The experiment was blinded by assigning arbitrary numbers to the fly bottles. Collected larvae were gently washed and embedded with the dorsal side up in 50% chilled glycerol between a glass slide and a coverslip. For immobilization, slides were kept at −20°C for 18 min before examining under UV light on an Axioplan microscope (Carl Zeiss, Jena, Germany). Digital pictures were taken with a Hamamatsu C4742-95 video unit (Hamamatsu Photonics, K.K., Hamamatsu City, Japan), controlled by the Openlap program (Improvision, Coventry, U.K.).
For each cross, 20 F1 progeny larvae were graded for the percentage of their segments showing a band formed by islets of sessile hemocytes under the epidermis. In the grading system, grade 1 larvae showed sessile hemocyte bands in 100% of the segments. Larvae receiving grade 2 or 3 showed bands in <75 or 50% of their segments, respectively. Larvae showing no bands or bands only in the most posterior 25% of their segments received grade 4. All crosses were repeated three times and the average grades of three independent experiments calculated.
Data analysis
Statistical analyses of reporter assays, qRT-PCR, and Western blot band quantification were carried out using one-way ANOVA. Statistical analysis of fly larvae in vivo experiments was performed using one-way ANOVA and Bonferroni as post hoc method. Statistical analysis of fly survival experiments was carried out using the log-rank (Mantel-Cox) test; p < 0.05 was considered to be statistically significant.
Results
Drosomycin expression is controlled by both the Toll and Imd pathways in Drosophila S2 cells
Drosomycin promoter-driven luciferase activity has been used to monitor the Drosophila Toll pathway activity (29). However, we have observed earlier (17) that in S2 cells, RNAi targeting components of the Imd pathway decreased the Drosomycin luciferase expression induced by the constitutively active form of Toll (Toll10b). To investigate the respective roles of the Toll and Imd pathways to the regulation of Drosomycin, we analyzed the Drosomycin luciferase activity in S2 cells activated with both Toll10b overexpression and heat-killed E. coli. As shown in Fig. 1A, Drosomycin expression was induced by Toll10b, and this induction was further enhanced if the Imd pathway was also activated by E. coli. This induction, stimulated by both pathways, can be drastically decreased by silencing known components of the Toll pathway, namely Toll, MyD88, or dorsal. Imd pathway components, namely PGRP-LC, Imd, Tab2, and Relish, are also required for normal induction in this assay. If both pathways are silenced by targeting both Relish and MyD88, Drosomycin activation is totally blocked (Fig. 1A). These results are in line with previous studies indicating that Drosomycin expression can be induced by both the Toll and Imd pathways in vivo and in vitro (5, 36). These results also show that Drosomycin luciferase activity can be used in screening for components of both the Toll and Imd signaling cascades in Drosophila S2 cells.
Genome-wide RNAi screen to identify genes required for NF-κB signaling in Drosophila S2 cells. A, Drosomycin luciferase reporter activity is regulated by both the Toll and Imd pathways in S2 cells. S2 cells were transfected with Drosomycin luciferase and Act5C–β-gal reporters. Toll pathway was induced by overexpression of Toll10b and the Imd pathway by heat-killed E. coli treatment for 24 h. RNAi targeting known components of the Toll pathway (Toll, MyD88, or dorsal) or components of the Imd pathway (Tab2, PGRP-LC, imd, or Relish) caused a strong reduction in the Drosomycin luciferase/Act5C–β-gal activity. Relish + MyD88 RNAi completely blocked the relative Drosomycin activation. Data are shown as mean ± SD; n ≥ 3. B, Results of the genome-wide RNAi screen of the Drosophila NF-κB signaling. The 16,025 independent dsRNA treatments were analyzed for effect on Drosomycin luciferase reporter activity induced via both the Toll and Imd signaling pathways. S2 cells were transfected with Drosomycin luciferase and Act5C–β-gal reporters, Toll10b, and dsRNAs and treated with heat-killed E. coli. Luciferase and β-galactosidase values were plotted on a log-scale. Negative control samples (GFP RNAi) are illustrated with light blue crosses. Toll pathway positive controls (MyD88 RNAi) are shown in purple, and Imd pathway positive controls (Relish RNAi) are shown in yellow. Samples with values in the top left corner of the plot include the most potential regulators of NF-κB signaling in Drosophila S2 cells. C, Targeted dsRNA treatments of potential regulators of the Drosophila NF-κB signaling confirm 23 dsRNA treatments that decrease Drosomycin luciferase reporter activity by >40%. Drosomycin luciferase and Act5C–β-gal values of induced GFP dsRNA-treated cells were set to 1. Data are shown as mean ± SD; n ≥ 3. Statistics refer to Drosomycin luciferase values.
Genome-wide analysis of the Toll and Imd pathways in Drosophila S2 cells
To identify gene products required for signaling via the Toll and Imd pathways, we examined the effect of 16, 025 dsRNA treatments for Drosomycin luciferase reporter activity in response to induction with Toll10b and E. coli in Drosophila S2 cells. The dsRNA collection was obtained by transcribing PCR products from the commercial Drosophila genome-wide library (MRC Geneservice) to dsRNAs (13,607) and transcribing dsRNAs (2,418) from S2 cell-derived cDNA library (37).
Drosophila S2 cells were transfected with Drosomycin luciferase and Act5C–β-gal reporter constructs, Toll10b, and dsRNAs. dsRNAs targeting MyD88 and Relish were used as positive controls and GFP as a negative control in each experiment. E. coli was added 24 h prior to luciferase and β-galactosidase measurements. Out of 16,025 dsRNA treatments, 23 repeatedly decreased the Drosomycin luciferase reporter activity >50% without considerably affecting the cell viability as measured by Act5C–β-gal reporter activity (Fig. 1B). Corresponding templates were sequenced and targeted PCR primers for dsRNA synthesis designed to confirm that the effect had been due to dsRNA according to the library data and not due to contaminating dsRNAs. As shown in Fig. 1C, five dsRNA treatments representing known components of the Toll pathway (Toll, MyD88, tube, pelle, and dorsal), and eight dsRNA treatments representing known Imd pathway components (Relish, kenny, FADD, Tak1, imd, Tab2, Ird5, and inhibitor of apoptosis 2) were identified, indicating that our screen effectively found components of both of the Drosophila NF-κB signaling pathways. Importantly, 10 novel regulators of NF-κB signaling were identified (Fig. 1C). Corresponding genes were subjected to further studies.
Nine of the identified regulators are required for signaling via the Toll pathway
Our RNAi screen effectively identified components of both the Toll and Imd signaling cascades. To find out which pathway is affected by these regulators, we carried out separate assays for the Toll and Imd pathways in S2 cells with targeted dsRNAs. Imd pathway activity was measured with a reporter assay in which Attacin A-driven luciferase (AttA-luc) construct, Act5C–β-gal, and dsRNAs were transfected into S2 cells, cells treated with E. coli, and reporter activities measured (Fig. 2A). Out of the 10 novel candidate genes, mediator complex subunit 25 (MED25) RNAi was shown to affect Imd pathway at the same level as RNAi to known Imd pathway components (21% of pathway activity left; Fig. 2A). Also, u-shaped RNAi decreased significantly the Imd pathway activity (48% of activity left). Spt6, CG31660, and CG15737 RNAi had little effect on Imd pathway activity, whereas RNAi targeting achaete, Gprk2, pannier, CG4325, and CG32133 resulted in hyperactivation of the pathway (Fig. 2A, on the right with a separate scale).
MED25 affects mainly the Imd signaling cascade, U-shaped affects both Toll and Imd signaling, and the other eight identified novel NF-κB regulators affect mainly Toll signaling. Data are shown as mean ± SD; n ≥ 3. In each panel, statistics refer to white bars. A, For Imd pathway, S2 cells were transfected with AttA-luc reporter, Act5C–β-gal reporter, and indicated dsRNAs and induced with heat-killed E. coli. In addition to known Imd pathway components (kenny, Tab2, Imd, Tak1, FADD, and Relish), MED25 RNAi decreased the AttA-luc reporter activity comparably to the known components. u-shaped RNAi also significantly affected the AttA-luc reporter activity. B, For Toll pathway activity, S2 cells were transfected with Toll10b, Drosomycin luciferase reporter, Act5C–β-gal reporter, and indicated dsRNAs. Of the novel regulators, pannier, CG4325, CG32133, u-shaped, CG15737, achaete, Gprk2, CG31660, and Spt6 dsRNAs considerably decreased the Drosomycin luciferase reporter activity. Also, most of the Imd pathway components/regulators (kenny, Tab2, imd, Tak1, MED25, and Relish) affected the Drosomycin luciferase activity. C, Drosomycin expression is inhibited by dsRNA treatments targeting all of the known tested components of Toll pathway and 10 dsRNAs identified from our collections. Endogenous Drosomycin expression in Toll10b-transfected dsRNA-treated S2 cells was analyzed using qRT-PCR and normalized to Act5C expression values. D, Schematic representation of the Drosophila Toll pathway. E, Epistasis analysis of the identified regulators of Toll signaling. Drosomycin luciferase expression was induced by cactus RNAi. Ten dsRNA treatments (dorsal, pannier, CG15737, Relish, Spt6, Gprk2, CG4325, u-shaped, CG32133, and achaete) blocked this induction and therefore appear to act downstream or independently of Cactus. Five dsRNA treatments (Toll, tube, pelle, CG31660, and MyD88) did not considerably affect this induction, indicating that these gene products act upstream of Cactus.
Out of the 10 novel candidate genes, 9 dsRNA treatments decreased the Drosomycin luciferase activity of Toll10b-induced S2 cells >60% (Fig. 2B). The Drosomycin luciferase activity of S2 cells treated with these dsRNAs was also reduced by at least 50% when the Toll pathway was activated by overexpression of the cleaved, active Spätzle ligand (SpzC106; data not shown). MED25 dsRNA treatment did not have as strong an effect on Toll pathway alone (Fig. 2B), so it was omitted from further analyses.
To ensure that our results were not due to an artifact related to the use of a reporter construct, we analyzed the relative endogenous Drosomycin expression levels of Toll10b-induced S2 cells by qRT-PCR. In this assay, Toll10b transfection to S2 cells induced the relative endogenous Drosomycin expression (Drosomycin/Act5C) in S2 cells ∼25-fold (Fig. 2C). Treating cells with dsRNAs targeting the known components of the Toll pathway, Toll, MyD88, tube, pelle, or dorsal, decreased the pathway activity by >55%. Similarly, RNAi targeting all of the nine novel Toll pathway regulators identified in the reporter assay, namely u-shaped, pannier, CG4325, Gprk2, CG15737, CG32133, CG31660, achaete, and Spt6 caused a statistically significant reduction in endogenous Drosomycin expression. Of note, RNAi targeting Relish, the NF-κB factor in the Imd pathway, also caused a statistically significant reduction in Toll10b-induced Drosomycin expression.
To gain more insight into the mechanism of how the novel regulators are functioning on the Toll pathway, we silenced cactus, the Drosophila homolog of human IκB, by RNAi (Fig. 2D, 2E). Silencing cactus results in Dif/Dorsal translocation into the nucleus (Fig. 2D) and >40-fold induction of the Toll pathway in a Drosomycin luciferase reporter assay (Fig. 2E). RNAi targeting known components of the pathway upstream of Cactus, namely Toll, tube, pelle, and MyD88 have no or very little effect on Cactus RNAi-induced Drosomycin luciferase activity. Conversely, RNAi targeting dorsal, the Drosophila NF-κB homolog in the Toll pathway (downstream of Cactus), blocks this induction completely (Fig. 2E). Results indicate that CG31660 appears to act upstream of Cactus, whereas pannier, CG15737, Spt6, Gprk2, CG4325, u-shaped, CG32133, and achaete appear to act downstream or independently of Cactus. Relish also acts downstream of Cactus.
Gprk2,CG15737/Toll pathway activation mediating protein, and u-shaped RNAi flies have reduced Drosomycin expression in Drosophila in vivo
To investigate whether the identified nine genes are important for the Toll pathway signaling in vivo, we carried out in vivo RNAi experiments with fly lines carrying UAS-RNAi constructs targeting these genes. RNAi flies (Supplemental Table I) were crossed with the C564-GAL4 driver line, which drives expression of the dsRNA in the fatbody. Fly strains without the driver (i.e., RNAi strains over w1118 flies) were used as controls. The Toll pathway was activated by M. luteus septic injury for 24 h, after which total RNAs were isolated. Relative Drosomycin expression in RNA samples was measured by qRT-PCR (Fig. 3). MyD88 RNAi crossed with C564-GAL4 flies were used as positive controls (Fig. 3A). Two Gprk2 RNAi fly lines, namely Gprk2 R-1 and Gprk2 R-3, crossed with C564-GAL4 showed a significant decrease in Drosomycin expression (Fig. 3B, 3C). Moreover, the fly line expressing the CG15737 RNAi construct showed a statistically significant reduction in Drosomycin expression compared with control flies, so we decided to name the CG15737 gene TAMP (Fig. 3D). Also, flies expressing the u-shaped RNAi construct showed a statistically significant reduction in Drosomycin expression compared with control flies (Fig. 3E). In vivo RNAi targeting other identified Toll pathway candidate genes did not significantly decrease Drosomycin expression (Supplemental Fig. 1).
RNAi targeting Gprk2, CG15737/TAMP, and u-shaped reduces Drosomycin expression in Drosophila in vivo. Fly lines containing indicated UAS-RNAi constructs crossed over the C564-GAL4-driver flies, and controls were infected with M. luteus by pricking and collected 24 h later. Total RNAs were extracted and Drosomycin expression levels measured by qRT-PCR. Results were normalized to Act5C expression values. In each experiment, the relative Drosomycin expression value of the control flies was set to 1. A, MyD88 RNAi flies were used as a positive control. Gprk2 (B, C), TAMP (D), and u-shaped (E) RNAi flies show impaired Drosomycin expression (p < 0.05) compared with controls without the driver. Data are shown as mean ± SD.
Drosophila Gprk2 is homologous to GRK5 from other organisms
Because of the strong phenotype obtained in both in vitro and in vivo Drosophila RNAi assays and its evolutionary conservation, we decided to subject Gprk2 to further studies. Drosophila Gprk2 (CG17998) is well conserved and has high sequence similarity with human, mouse, and zebrafish GRK5 (Supplemental Fig. 2). Gprk2 codes for a 714-aa protein that has three known domain structures: a regulator of G protein signaling (RGS) domain, a serine/threonine protein kinase catalytic domain, and an extension to kinase domain (Fig. 4A). It belongs to a protein family, the members of which are multifunctional, GTPase-accelerating proteins (38). When the Gprk2-GFP construct was overexpressed in Drosophila S2 cells, it was shown that Gprk2 is localized on the cell membrane or cytoplasm (Supplemental Fig. 3A). Similarly, the human GRK5-GFP construct was located on the cell membrane or cytoplasm when overexpressed in HeLa cells (Supplemental Fig. 3B).
Gprk2/GRK5 has an evolutionarily conserved role in NF-κB–mediated immune signaling in human HeLa cells in vitro and in zebrafish embryos in vivo. A, Schematic diagram of Drosophila Gprk2, Homo sapiens GRK5, and Danio rerio GRK5 proteins. B, GRK5 is required for TNF-α–triggered NF-κB signaling in HeLa cells. HeLa cells were transfected with NF-κB–luciferase and CMV–β-galactosidase reporters and with GRK5 or control siRNAs. Cells were induced with TNF-α 6 h prior to luciferase and β-galactosidase measurements. C and D, GRK5 is required for E. coli-induced activation of NF-κB signaling in zebrafish embryos. Zebrafish embryos were injected with GRK5 and control morpholinos, and 24 h later, embryos were infected by E. coli injection. After 18 h incubation, embryos were collected and total RNAs extracted. Relative expression of TNF-α (C) or IL-1β (D) in infected and control embryos was analyzed by qRT-PCR. Data are shown as mean ± SD; n ≥ 5. C, CaM binding site; EXT, extension to kinase domain; kinase, kinase domain.
Gprk2/GRK5 has an evolutionarily conserved role in NF-κB signaling
To investigate whether Gprk2/GRK5 has an evolutionarily conserved role in NF-κB signaling, we examined the role of GRK5 in NF-κB signaling in human HeLa cells in vitro. HeLa cells were transfected with NF-κB luciferase and CMV–β-galactosidase reporters and GRK5 or control siRNAs. Six hours prior to measurements, NF-κB signaling was induced with TNF-α. When HeLa cells are treated with GRK5 siRNA, the relative NF-κB-luc activity is reduced >60% (Fig. 4B). This indicates that GRK5 is an important regulator of human NF-κB signaling in vitro.
To study the role of Gprk2/GRK5 for vertebrate innate immune response in vivo, we silenced the zebrafish GRK5 in embryos with a translation-blocking morpholino. E. coli was injected into GRK5 morphant zebrafish larvae at 48 h postfertilization and proinflammatory cytokine levels were monitored 18 h postinfection. TNF-α mRNA expression was induced ∼600-fold and IL-1β 300-fold (data not shown). In larvae lacking GRK5, the relative TNF-α expression (Fig. 4C) and IL-1β (Fig. 4D) was significantly reduced from that of control morpholino-treated larvae. Blocking the translation of MyD88 also resulted in reduction of TNF-α and IL-1β expression levels (Fig. 4C, 4D). These results indicate that GRK5 is essential for NF-κB signaling in vertebrate immune system in vivo.
Gprk2 interacts with Cactus but is not required for its degradation upon signaling
Because Gprk2 acts at the level or downstream Cactus in the cactus dsRNA epistasis experiment (Fig. 2E), and because of reports of mammalian GRK5 interaction with members of the IκB family (39–41), we decided to investigate the interaction of Gprk2 with Cactus and Dorsal. V5-tagged full-length Gprk2 and deletion constructs were coimmunoprecipitated with myc-tagged Cactus and Dorsal in S2 cells. The full-length Gprk2, Calmodulin (CaM) binding-site deletion (ΔCaM1), and RGS-domain deletion (ΔRGS) constructs interact with Cactus protein, indicating that RGS and CaM1 domains are not needed in Gprk2 and Cactus interaction. In the kinase deletion (Δkinase) construct, this interaction is virtually not detectable anymore, which suggests that the kinase domain is important for the interaction, or that the protein, lacking a large domain, is not correctly folded anymore, resulting in loss of the interaction (Fig. 5A, Supplemental Fig. 4). V5-tagged Gprk2 proteins did not coimmunoprecipitate with Dorsal-myc (data not shown).
Gprk2 interacts with Cactus, but is not required for Cactus degradation upon signaling. A, The full-length Gprk2, CaM binding site deletion (ΔCaM1) and RGS-domain deletion (ΔRGS) constructs interact with Cactus protein. In the kinase deletion (Δkinase) construct, this interaction is virtually nondetectable. Protein expression was induced with CuSO4 (250 μM). Proteins were coimmunoprecipitated with anti-myc Ab and detected with anti-V5 Ab. B, Expression of pMT-Gprk2-V5 and pMT-Cactus-myc constructs in S2 cells induced with CuSO4 (250 μM) detected with anti-V5 and anti-myc Abs, respectively. C, Gprk2 RNAi does not affect Cactus degradation upon signaling. S2 cells expressing a chimeric EGFR-Toll construct were treated with GFP, Gprk2, and MyD88 dsRNAs. Toll pathway was activated with EGF, and Cactus protein in cytoplasmic extracts was detected by SDS-PAGE and Western blotting with anti-Cactus Ab. Gel loading was controlled using anti-GM130 Ab targeting a Drosophila Golgi protein. D, Cactus band intensities were quantified from three separate Western blots and normalized to the loading control GM130 band intensities. Data are shown as mean ± SD; n = 3.
To investigate the functional significance of Cactus-Gprk2 interaction, we used an established EGFR-Toll pathway induction system (32) to monitor Cactus degradation. S2 cells expressing a chimeric EGFR-Toll construct were treated with GFP, Gprk2, and MyD88 dsRNAs and Cactus degradation was monitored on a Western blot (Fig. 5B). Also, a loading control was carried out with anti-GM130 Ab (Abcam) targeting a Drosophila Golgi protein (Fig. 5B). Cactus band intensities were quantified from three separate Western blots with Adobe Photoshop 7 software (Adobe Systems) and normalized to the loading control GM130 band intensities (Fig. 5C). Gprk2 RNAi did not affect degradation of Cactus. Furthermore, we carried out kinase experiments with coimmunoprecipitated Cactus and Gprk2, but were not able to show Gprk2-mediated phosphorylation of the Cactus protein (data not shown). We conclude that Gprk2 interacts directly or indirectly with Cactus, but is not required for Cactus degradation upon signaling.
Gprk2 RNAi flies infected with B. bassiana have reduced expression of the Toll pathway target genes in Drosophila in vivo
To investigate the role of Gprk2 on Toll pathway-mediated immunity in vivo, we carried out an experiment in which Grpk2 RNAi flies and controls were subjected to natural fungal infection with an insect pathogen B. bassiana at +29°C for 48 h, after which total RNAs were isolated. RNAs from noninfected flies were isolated as a control for the infection. Expression of Toll pathway target genes, namely Drosomycin, IM1, and IM2 was measured by qRT-PCR (Fig. 6A–C, respectively). Results were normalized to Act5C expression values. Both Gprk2 R-1 and Gprk2 R-3 RNAi lines crossed with the driver C564-GAL4 showed a reduced expression of Toll pathway target genes. w1118 flies over the C564-GAL4 driver and MyD88 RNAi flies over the C564-GAL4 driver were used as negative and positive controls, respectively. These results indicate that in Gprk2-silenced flies, Toll pathway-induced genes are poorly activated after fungal infection.
Gprk2 RNAi flies have impaired expression of Toll pathway target genes when subjected to B. bassiana natural fungal infection. Gprk2 RNAi flies crossed with C564-GAL4 driver and controls were infected with B. bassiana, incubated for 48 h at 29°C, and collected. Total RNAs extracted from flies were subjected to qRT-PCR analysis. The relative Drosomycin, IM1, or IM2 value of the control flies (w1118/C564) was set to 1. A, Relative Drosomycin expression. Relative IM1 expression (B) and relative IM2 expression (C) in infected flies. Data are shown as mean ± SD, n = 3.
Gprk2 RNAi flies are susceptible to infection with Gram-positive bacteria E. faecalis
To examine whether the effect of Gprk2 silencing on Toll pathway is sufficient to impair the fly’s survival, we used septic injury with Gram-positive bacteria. The Toll pathway-mediated immune response was first induced by pricking flies with a needle dipped into a culture of M. luteus. M. luteus infection activates the Toll pathway response including Drosomycin expression. Twenty-four hours later, the flies were infected with E. faecalis by pricking as above. Both Gprk2 RNAi lines crossed with the C564-GAL4 driver show a statistically significant reduction in survival compared with the control line without the driver (Fig. 7A, 7B). When infected with the Gram-negative bacterium E. cloacae, there was no difference between the Gprk2 RNAi flies and controls (Supplemental Fig. 5A, 5B). In conclusion, Gprk2 is needed for normal defense against Gram-positive bacteria E. faecalis.
Gprk2 RNAi flies are susceptible to infection with E. faecalis, and Gprk2 RNAi rescues Toll10b-induced blood cell activation in Drosophila larvae in vivo. A and B, The Toll pathway was activated by pricking flies with M. luteus, and 24 h later, the flies were pricked with E. faecalis. The survival of the flies was monitored for 24 h. MyD88 RNAi flies were used as a positive control. Both Gprk2 RNAi lines (Gprk2 R-1 and Gprk2 R-3) crossed over the C564-GAL4 driver show a statistically significant reduction in survival compared with controls. Gprk2 R-1 × C564-GAL4 (n = 132); Gprk2 R-1 control without the driver (n = 117); Gprk2 R-3 × C564-GAL4 (n = 105); Gprk2 R-3 control without the driver (n = 99). C, Sessile hemocyte banding pattern (i), lost upon constitutive activation of the Toll signaling pathway in blood cells (ii), could be rescued by Gprk2 (iii) or MyD88 RNAi (iv). D, The average grades of three independent crosses. n = 20 larvae per cross, ± SEM. Grade 1, sessile hemocyte bands in 100% of segments; grade 2, bands in <75% of segments; grade 3, bands in <50% of segments; and grade 4, no bands or bands only in the most posterior 25% of segments.
Gprk2 RNAi construct can rescue UAS-Toll10b blood cell activation in Drosophila larvae in vivo
To examine if Gprk2 RNAi can inhibit blood cell activation caused by a constitutively activated Toll pathway in vivo, transgenic RNAi fly lines of Gprk2 R-3, and MyD88 as a control, were combined with blood cell-specific hmlΔ-GAL4, UAS-GFP driver. Males originating from these stocks were crossed to females of the UAS-Toll10b line. Males from the original driver line were crossed to females of either the UAS-Toll10b line (negative control) or w1118 (treatment control). Progeny third-instar larvae were graded for the percentage of their segments showing bands formed by islets of sessile hemocytes under the epidermis. Offspring of the treatment control showed bands of sessile cells in all segments, indicating that larval handling had little effect on the blood cell distribution (Fig. 7Ci). In contrast, larvae of the negative control showed a largely disturbed sessile hemocyte banding pattern caused by the constitutive activation of the Toll signaling pathway in blood cells (Fig. 7Cii) (27). This difference is reproducible and reflected in the average grades calculated, with significantly higher average grades for crosses of the negative control compared with the treatment control (Fig. 7D). The loss of sessile hemocyte banding pattern could be rescued by introducing the RNAi constructs targeting Gprk2 or MyD88 (Fig. 7Ciii, 7Civ, 7D). These results indicate that RNAi targeting Gprk2 can inhibit hemocyte activation caused by UAS-Toll10b.
Discussion
Large-scale in vitro RNAi screening has become a commonly used method to identify gene products involved in numerous cellular processes. In this study, we used a luciferase-based reporter assay, with which we were able to monitor both the Toll and the Imd pathway activities simultaneously. Based on careful setup of the assay, biologically meaningful cutoffs and assessment of the general well-being of the cells using Act5C–β-gal reporter, we obtained a sensible hit list of 23 genes. These included 5 known components of the Toll signaling pathway, 8 known components of the Imd pathway, and 10 previously uncharacterized novel regulators of Drosophila NF-κB signaling. Noteworthy, only one new regulator (MED25) strongly affected the Imd pathway, whereas there were nine dsRNA treatments that primarily decreased the Toll pathway activity. This is in accordance with the notion that several RNAi screens have already been carried out for the Imd pathway, whereas the Toll pathway is less thoroughly studied using RNAi. The results related to Toll pathway activity were further confirmed by secondary (SpzC106-based) and tertiary (qRT-PCR for endogenous Drosomycin) assays. After these confirmation and validation steps, we ended up with a solid hit list of nine novel modifiers of the Toll signaling pathway, namely u-shaped, pannier, achaete, TAMP, CG4325, CG32133, CG31660, Spt6, and Gprk2.
Our screen failed to identify one known positive regulator of the Toll pathway downstream of Toll receptor (Dif) and two components of the Imd pathway (PGRP-LC and Dredd). Targeted RNAi for Dif did not decrease the Drosomycin luciferase reporter activity either when induced by Toll10b or Toll10b together with E. coli. This suggests that in S2 cells, dorsal has a more important role in Toll pathway-mediated signaling than Dif. As for PGRP-LC and Dredd, we conclude that these two were not successfully targeted by our dsRNA libraries. In fact, there were several instances in which there was more than one PCR product, occasionally none of them corresponding to the indicated gene, in a single well of the MRC Geneservice PCR product library. Therefore, we found it imperative to TA-clone and sequence every PCR template corresponding to interesting RNAi phenotypes, to design gene-specific primers, and to carry out independent RNAi with targeted dsRNAs. If there were multiple PCR products, all corresponding targeted dsRNAs were tested to identify the one that caused the observed phenotype.
U-shaped is the Drosophila Friend of GATA homolog with a known important role in hemocytes. U-shaped has been shown to interact with and negatively regulate pannier, a Drosophila GATA transcription factor (42). U-shaped and pannier together with achaete (and scute) also regulate the bristle formation in Drosophila (43). This suggests that u-shaped, pannier, and achaete may act together in the process of Toll pathway regulation. TAMP (CG15737) encodes a protein with an N-terminal domain homologous to poly(A) polymerase proteins. CG4325 is a small protein with a RING finger domain, which is likely to bear E3 ubiquitin-protein ligase activity and is often involved in mediating protein-protein interactions. CG32133 is a large protein with postulated molecular functions in transcription factor binding, but the biological processes it mediates are unknown. Drosophila Spt6 has homology to Saccharomyces cerevisiae Spt6p, which has been implicated in transcription initiation and maintaining normal chromatin structure during transcription elongation (44). CG31660 bears homology to human G protein-coupled receptor 158, and it contains a domain typical for a metabotropic glutamate family. Metabotropic glutamate receptors are coupled to G proteins and stimulate the inositol phosphate/Ca2+ intracellular signaling pathway (45). It is likely that understanding the exact molecular functions of these genes will reveal novel levels and means to delicately control NF-κB pathway-mediated immune response.
Drosophila RNAi fly collections, namely the VDRC (Vienna, Austria) and NIG-FLY (Kyoto, Japan), provide a tool to study the importance of a selected gene product to a chosen function in the whole organism scale. Crossing RNAi flies with an appropriate GAL4 driver fly line results in silencing of the targeted gene in the chosen tissue in the progeny. However, it is recognized that as much as 35–40% of fly lines may give a false-negative result, which may be due to multiple reasons related to RNAi, driver GAL4 strain, and/or the assay chosen (34). In our in vivo infection assays, we found a phenotype with a statistical difference to controls in 5 out of 11 strains (Fig. 3, Supplemental Fig. 1). In addition, two other strains (Spt6 and pannier RNAi strains; Supplemental Fig. 1) showed a similar but nonsignificant trend. Therefore, our results are in line with the estimates presented by Dietzl and coworkers (34). Of note, the false-positive rate is estimated to be <2%, which means that it is very likely the reduction in Drosomycin expression in our driver-induced strains is due to silencing of the gene in question (34).
Importantly, we identified a novel, evolutionarily conserved regulator of NF-κB signaling, Gprk2, in our screen in S2 cells. Gprk2 is very well conserved and has high sequence similarity at the amino acid level with vertebrate GRK5. GRKs are known to phosphorylate G proteins, thus causing receptor desensitization and switching off of the G protein-coupled receptor signaling pathway (38). In addition to G proteins, GRKs are known to phosphorylate various other substrates and to modulate cellular responses in a phosphorylation-independent manner (46). Although mostly membrane-bound, GRK5 has been shown to contain a functional nuclear localization sequence (47), and a function as a histone deacetylase kinase in the nucleus of cardiomyocytes has been reported (48).
Recently, there have been implications as to the involvement of the human GRK5 to NF-κB–mediated immunity: in a recent report, the human GRK5 has been shown to participate in TNF-α–induced NF-κB signaling via direct interaction with and phosphorylation of IκBα (39). Also, effects on LPS-induced ERK1/2 signaling (40) and NF-κB transcriptional activity (41) have been proposed. In Drosophila, Gprk2 has been shown to regulate hedgehog signaling (49), but no involvement in innate immunity has previously been reported. In this study, we have shown that Gprk2 is an evolutionarily conserved regulator of innate immune signaling. Furthermore, we were able to show that Gprk2 is required for normal microbial resistance in vivo. Interestingly, although Gprk2 physically interacts with Cactus, it is not required for signal-induced Cactus degradation. It will be of great interest in the future to investigate the exact role of the Gprk2-Cactus interactions.
NF-κB signaling is of paramount importance for regulating immune response both in flies and vertebrates. The power of the Drosophila model includes the possibility of combining large-scale RNAi screening with sophisticated in vivo tools. In this study, we carried out a genome-wide RNAi screen in cultured Drosophila cells and identified 10 novel regulators of Drosophila NF-κB signaling. The evolutionarily conserved role for Gprk2/GRK5 in NF-κB pathway activation was shown using human HeLa cells in vitro and zebrafish embryos in vivo. Finally, the importance of Gprk2 for Drosophila NF-κB signaling was demonstrated both in vitro and in vivo.
Acknowledgments
We thank other members of our laboratory for help and insightful discussions, L. Mäkinen for technical help with Drosophila lines and zebrafish work, and M. Ovaska for creation of the cDNA library and dsRNA synthesis. We also thank J.-L. Imler, J.-M. Reichhart, and S. Stöven for reporter, Toll10b, SpzC106, and Relish plasmid constructs, S. Wasserman and P. Towb for the Cactus Ab and the EGFR-Toll construct, B. Lemaitre for C564-GAL4 driver flies and B. bassiana, and P. Wedegaertner for the GRK5-GFP construct. The P{UAS-Tl10b:11} stock was a kind gift of J.-M. Reichhart. We also thank S. Sinenko for the hmlΔ-GAL4, UAS-GFP fly stock and Kyoto NIG-FLY (Kyoto, Japan) and VDRC (Vienna, Austria) for RNAi fly lines.
Disclosures The authors have no financial conflicts of interest.
Footnotes
This work was supported by grants from the Academy of Finland (to M.R, M.P. and D.H.), the Foundation for Pediatric Research, Sigrid Juselius Foundation, and Emil Aaltonen Foundation (to M.R.), Competitive Research Funding of the Pirkanmaa Hospital District (to M.R., S.V., A.K., and M.P.), and The Swedish Cancer Society and The Swedish Research Council (to Y.E. and D.H.). Zebrafish work was done at the University of Tampere Zebrafish Core Facility.
The online version of this article contains supplemental material.
Abbreviations used in this paper:
- Act5C–β-gal
- Actin 5C–β-galactosidase
- C
- CaM binding site
- CaM
- Calmodulin
- Dif
- Dorsal-related immunity factor
- EGF
- epidermal growth factor
- EGFR
- EGF receptor
- EXT
- extension to kinase domain
- FADD
- Fas-associated death domain protein
- Gprk
- G protein-coupled receptor kinase
- GRK
- G protein-coupled receptor kinase
- hmlΔ
- hemolectinΔ
- IM
- immune-induced molecule
- Imd
- immune deficiency
- kinase
- kinase domain
- MED25
- mediator complex subunit 25
- MRC
- Medical Research Council
- NIG-FLY
- Fly Stocks of the National Institute of Genetics
- PGRP
- peptidoglycan recognition protein
- qRT-PCR
- quantitative RT-PCR
- RGS
- regulator of G protein signaling
- RNAi
- RNA interference
- siRNA
- small interfering RNA
- Tab2
- TGF-β–activated kinase 1-associated binding protein 2
- Tak1
- TGF-β–activated kinase 1
- UAS
- upstream activating sequence
- VDRC
- Vienna Drosophila RNAi Center.
- Received January 26, 2010.
- Accepted March 22, 2010.
- Copyright © 2010 by The American Association of Immunologists, Inc.