Protein Kinase D Is Dispensable for Development and Survival of Drosophila melanogaster

Members of the Protein Kinase D (PKD) family are involved in numerous cellular processes in mammals, including cell survival after oxidative stress, polarized transport of Golgi vesicles, as well as cell migration and invasion. PKD proteins belong to the PKC/CAMK class of serine/threonine kinases, and transmit diacylglycerol-regulated signals. Whereas three PKD isoforms are known in mammals, Drosophila melanogaster contains a single PKD homolog. Previous analyses using overexpression and RNAi studies indicated likewise multi-facetted roles for Drosophila PKD, including the regulation of secretory transport and actin-cytoskeletal dynamics. Recently, involvement in growth regulation has been proposed based on the hypomorphic dPKDH allele. We have generated PKD null alleles that are homozygous viable without apparent phenotype. They largely match control flies regarding fertility, developmental timing and weight. Males, but not females, are slightly shorter lived and starvation sensitive. Furthermore, migration of pole cells in embryos and border cells in oocytes appears normal. PKD mutants tolerate heat, cold and osmotic stress like the control but are sensitive to oxidative stress, conforming to the described role for mammalian PKDs. A candidate screen to identify functionally redundant kinases uncovered genetic interactions of PKD with Pkcδ, sqa and Drak mutants, further supporting the role of PKD in oxidative stress response, and suggesting its involvement in starvation induced autophagy and regulation of cytoskeletal dynamics. Overall, PKD appears dispensable for fly development and survival presumably due to redundancy, but influences environmental responses.


INTRODUCTION
Protein Kinase D (PKD) isoforms are serine/threonine kinases of the Protein Kinase C family typified by a long N-terminal regulatory region followed by a catalytic kinase domain ( Figure 1A). The regulatory region contains two cysteine-rich domains that bind to diacylglycerol and phorbolesters, and a Pleckstrin-homology module, that auto-inhibits the catalytic domain ( Figure 1A) (reviewed in Fu and Rubin 2011). Once recruited to the membrane by diacylglycerol, PKD is activated by members of the PKC family through the phosphorylation of two serine residues in the activation loop of the kinase domain (Fu and Rubin 2011). Based on sequence similarity of the kinase domain, PKD has been classified as member of the Ca 2+ /Calmodulin-dependent serine/threonine protein kinases (CAMK) Rubin 2011, Cobbaut andVan Lint 2018).
Three PKD isoforms exist in mammals, where they are involved in various processes, including the regulation of a cell survival response upon oxidative stress, cell proliferation, cell differentiation, cell motility and invasion, as well as secretory transport from the trans-Golgi compartment to the plasma membrane (reviewed in Fu and Rubin 2011, Olayioye et al. 2013, Cobbaut and Van Lint 2018. The extent of overlap in their biological function is still a matter of debate, but it is evident that all three act as multi-functional kinases with a major role in structural integrity and function of the Golgi complex as well as in the regulation of actin-cytoskeletal dynamics (Fu and Rubin 2011, affecting oxidative stress response in embryonic fibroblasts (Zhang et al. 2015). Whereas this result may be taken as indication for functional redundancy of the three PKD isoforms, it may also be interpreted as a highly (perhaps cell type) specific, non-lethal function of the respective kinase. Alternatively, only laboratory, non-stressed conditions may allow normal development (Zhang et al. 2015).
Drosophila melanogaster harbors a single PKD homolog with a similarity of 67% and identity of nearly 60% to any of the three human PKD kinases (Maier et al. 2006). Accordingly, Drosophila has been used as an in vivo model to study the biological roles of PKD. The Drosophila PKD gene is broadly expressed throughout development. Whereas PKD mRNA is uniform in imaginal tissues, it strongly accumulates in ectodermal derivatives in the late embryo, detected for example in the epithelia of the epidermis, the salivary glands, the hind-and the foregut (Maier et al. 2006). A ubiquitously expressed PKD-GFP fusion protein was present in the cytosol and along cell membranes, and in the trans-Golgi compartment of secretory tissues like the salivary glands (Maier et al. 2006). The biological function of PKD in the fly was assessed by the overexpression of presumptive activated and dominant negative isoforms of PKD, as well as by RNA interference experiments (Maier et al. 2006, Maier et al. 2007, Nagel et al. 2010. Whereas the overexpression of a dominant negative PKD isoform interfered with pattern formation in the wing, the activated PKD-SE isoform affected development more broadly (Maier et al. 2007). RNAi-mediated knockdown of PKD activity effected tissue loss primarily through apoptosis. Interestingly, a light-dependent degeneration of the adult retina was observed in flies overexpressing activated PKD-SE (Maier et al. 2007). Being a typical consequence of rhodopsin maturation or trafficking defects (Colley et al. 1995), this phenotype points to a role of PKD in secretory transport and in cytoskeletal dynamics (Maier et al. 2007). Figure 1 Generation of PKD mutants by homologous recombination. (A) Scheme of the PKD protein; it comprises 836 amino acids. The protein is subdivided in a regulatory and a catalytic kinase domain (KD). Characteristics of the regulatory domain are tandem zinc-finger motifs (C1a, C1b), and a pleckstrin homology domain (PH). Numbering depicts codons. The deletion generated in the mutants starts within codon 91, and hence affects all relevant domains of the PKD protein.
(B) Genomic map of the PKD locus and the proximal genes CG7126, CG18600 and Dlc90F (numbers give sequence location according to flybase). Exons are shown as boxes, coding exons are shaded darker. Underneath, the two DNA fragments (Phl, Phr) used for homologous recombination are shown. P1-P6 depicts roughly the position of primers used in confirmation experiments (not to scale). No function has yet been assigned to CG7126 or CG18600. (C) Map of PKD 26 and PKD cl4 mutants. In the PKD 26 mutant, most of the PKD locus was replaced by the white + marker gene, which was deleted in PKD cl4 by Cre-loxP mediated recombination. P5-P8 depicts roughly primer positions (not to scale). (D) Confirmation of the successful homologous recombination event in the PKD 26 allele. In situ hybridization with a white + probe on salivary gland chromosomes from PKD 26 mutants highlights the white locus at the tip of the X-chromosome in the upper panel (open arrowhead), and shows insertion of the w + -bearing transgene at position 91A on the right arm of the third chromosome (closed arrow). Numbers correspond to chromosomal sections (Lindsley and Zimm 1992). The whole spread is shown in supplemental Figure S1. (E) In situ hybridization of a PKD probe on whole mount wild type (wt, upper panel) and homozygous PKD 26 embryos (lower panel) confirms absence of PKD transcripts in the mutant. Scale bar corresponds to 100 mm for both panels. (F) RT-PCR was performed on RNA of PKD 26 and PKD cl4 mutant and y 1 w 67c23 control flies as indicated. (+) with reverse transcriptase and (-) no-RT control. Primer pairs P1/P2 (278 nt) and P3/P4 (282 nt) (see (A)) overlap the fifth and the third introns, respectively. No transcripts beyond exon four are detected in the PKD mutant animals. Tubulin primers served as control for intact mRNA (Tub, expected 299 nt). NEB 100 bp DNA-ladder served as size standard (500 bp reference, 300 bp and 200 bp are labeled). The picture was inverted for better visibility.
The latter aspect was corroborated by the finding that Drosophila PKD, like its mammalian counterpart PKD1, impacts actin remodelling by the regulation of cofilin activity through its phosphatase Slingshot (Bari sić et al. 2011). Accordingly, accumulation of F-actin and phosphorylated cofilin was likewise observed in cell clones either overexpressing activated PKD-SE or lacking Slingshot (Nagel et al. 2010, Bari sić et al. 2011. These investigations provided evidence for a possible role of Drosophila PKD in cell motility by modulating actin dynamics (Bari sić et al. 2011). Recently, a role of Drosophila PKD in the secretion of insulin like peptide ILP2 was uncovered, thereby influencing metabolism and growth of the developing animal (Ashe et al. 2018). In sum, Drosophila PKD appears to be a multifunctional kinase like its mammalian homologs Rubin 2011, Olayioye et al. 2013).
In order to address the role of Drosophila PKD in further detail, we generated null alleles by ends-out directed homologous recombination (PKD 26 , PKD cl4 ). Both alleles turned out to be homozygous viable without apparent phenotypes. With regard to developmental timing, life span, fertility, weight and fat content, the PKD null mutants were largely within the range of the control. Whereas the PKD null mutants tolerated various stress factors, sensitivity toward oxidative stress was uncovered. In a candidate kinase screen, we obtained evidence for redundant kinase function regarding oxidative stress response, starvation induced autophagy and regulation of cytoskeletal dynamics. Overall, our data indicate that PKD is largely dispensable for development and survival of Drosophila but is required for combatting oxidative stress.

Generation and confirmation of PKD mutant alleles
To generate PKD null mutants, we followed the 'ends-out' homologous recombination protocol developed by Gong and Golic (2003). To this end genomic fragments covering 59 and 39 regions of the locus were PCR-isolated from lambda phage ED clones (Maier et al. 1992): a 5 kb genomic Acc65I/AscI fragment starting within the first intron and covering the second coding exon (Phl; Figure 1B), and a 3.4 kb Acc65I/SphI fragment overlapping the downstream CG18600 locus (Phr; Figure 1B Fragments were cloned into pW25 transformation vector using compatible restriction sites (Gong and Golic 2003) to generate transgenic starter line T15-2 (second chromosomal insertion) by classical P-element mediated germ line transformation of y 1 w 67c23 flies (Rubin and Spradling 1982). Homologous recombination was performed as described before (Gong andGolic 2003, Fischer et al. 2015), starting with 592 G1 single-crosses resulting in the line PKD 26 . The inserted white marker gene was excised with help of I-CreI exactly as described before (Rong et al. 2002) to yield allele PKD cl4 .
Homologous recombination at the PKD locus at 92E was confirmed for PKD 26 by chromosome in situ hybridization, by Southern blotting as described earlier (Preiss et al. 1988), and by PCR with primer pairs P1/2, P3/4, P5/6, P5/7 and P6/8. Loss of mRNA expression in PKD 26 was confirmed by in situ hybridization on whole mount embryos according to standard protocols (Tautz and Pfeifle 1989) and by RT-PCR. The breakpoint fragment generated by PCR from PKD cl4 total DNA with primer pair P5/P6 was sequence confirmed. The following primers were used (position is schematically shown in Figure 1B) In situ hybridization Polytene chromosomes from salivary glands of PKD 26 homozygous mutant third instar larvae were prepared for in situ hybridization as outlined in Ashburner (1989). The probe, labeled with DIG-dUTP by random priming (Tautz and Pfeifle 1989), was generated using CaSpeR-vector as template as it contains the white + minigene (Pirrotta 1988), using the DIG DNA Labeling and Detection kit (Roche; Merck). Hybridization was as outlined in Ashburner (1989), and detection according to the manufacturer's protocol. In situ hybridization on whole mount embryos was as described earlier (Tautz and Pfeifle 1989;Maier et al. 2006). Embryos from y 1 w 67c23 control flies and PKD 26 mutant flies were collected overnight. As probe, we used PKD cDNA (pOT GH26429) (Maier et al. 2006), labeled with DIG-dUTP as above.
Transcriptional analysis by RT-PCR Poly(A) + RNA was isolated with the PolyATract System 1000 kit (Promega Mannheim, Germany) from 25 PKD 26 , PKD cl4 and y 1 w 67c23 male flies each according the supplier's protocol. cDNA was produced with qScriber cDNA Synthesis Kit (highQu, Kraichtal, Germany) according to the manufacturer's protocol. Amplification was with primer pairs P1/P2 and P3/P4, both overlapping introns: Genomic DNA should yield a 352 bp and a 1662 bp amplificate, respectively, whereas 278 bp and 282 bp are expected from cDNA. Tubulin 56D primers served as positive control (see primer list above). Absence of genomic DNA was tested in a non-RT control. NEB quick-load 100 bp DNA ladder was used as size standard. Real time qRT-PCR was conducted as outlined in Praxenthaler et al. (2017) using Blue S'Green qPCR kit (Biozym, Hessisch-Oldendorf, Germany) on 10 ng of cDNA from PKD 26 , PKD cl4 and OreR flies in 10ml end volume using MIC magnetic induction cycler (bms, Pots Point, Australia) including target, no-template and non-RT controls. As internal references for PKD or SOD expression, bTub56D and tbp were used. Primers were selected from the DRSC FlyPrimer bank (Hu et al. 2013): SOD1, PP70435; SOD2, PP70435; tbp, PP1556. Primers for Tub56D and PKD (P1/P2) are listed above. Relative quantification of three biological and two technical replicates was performed with micPCR software Version 2.6.5 based on REST taking target efficiency into account (Pfaffl et al. 2002).

Phenotypic analyses
Flies were raised under non-crowded conditions on standard agarcorn-molasses food (per liter 18g dry yeast, 10g soy flour, 22g molasses, 80g malt extract, 80g cornmeal, 6.25ml propionic acid, 8g agar-agar) at constant 25°and 78% humidity. Analyses were performed on one to five days old flies. To investigate developmental timing, offspring from parallel inter se crosses of 5 females and 3 males each was counted at days 8 to 18 (Wang et al. 2003, Fischer et al. 2015. To determine fertility, 4-8 virgin females (one to four days old) were kept in a vial with wild type males for three days, and then put on fresh food for 4-5 days at 18°; the number of offspring was recorded and calculated per female per day on food. For longevity experiments, animals separated by sex at the day of eclosion were transferred in batches of 25-30 to fresh food every third day; dead animals were recorded daily (Clancy et al. 2001, Wang et al. 2003, Tettweiler et al. 2005. High sucrose medium contained additional 10% sucrose (Wang and Clark 1995;Magwere et al. 2004). Using a precision balance adult weight was determined in batches of five animals 1-2 days after hatching (Fischer et al. 2015). Relative fat content was determined as percentage of dry weight as outlined in Vermeulen et al. (2006). The larval floating test was applied on fully fed, third instar wandering larvae in 8%, 10% and 12% sucrose in PBS, respectively (Reis et al. 2010;Fischer et al. 2015). Resistance to starvation was recorded either in the absence of food (wet starvation) or absence of all (dry starvation). To this end, sexed flies (2-3 days old) were starved in batches of 25 in empty vials (dry starvation) or in vials supplemented daily with a wet filter paper (wet starvation); dead animals were recorded regularly (Craig et al. 2004;Wang et al. 2004). Thermotolerance of flies, collected 2 days after eclosion, was tested in dry heat, wet heat and cold. Survival was determined by incubating flies in batches of 25-30 at 37°in an incubator on pre-warmed food for 2-3 hr (dry heat), or alternatively by submersing vials in a 37°degree water bath and recording dead animals regularly (Craig et al. 2004, Nielsen et al. 2005. To test cold sensitivity, two days old animals were cooled to 4°for 2 hr and recovery time until walking was recorded (Nielsen et al. 2005). To generate oxidative stress, 20 males 3-5 days old were first starved for four hours on 1% agarose/PBS medium, and then transferred to vials with filter paper soaked with a solution of 20 mM paraquat in 5% sucrose at 25°in the dark; dead flies were recorded twice a day, and live flies then transferred to a fresh paraquat containing vial (Clancy et al. 2001;Wang et al. 2003). Osmotic stress was applied by rearing flies on food containing additional 0.5 M NaCl; death toll was recorded daily (Craig et al. 2004).
Border cell migration was studied by staining ovaries with rhodaminecoupled phalloidin (Molecular Probes, Eugene OR, USA) as outlined before (Nagel et al. 2012), analyzed by confocal microscopy using a BioRad MRC1024 coupled to a Zeiss Axiophot and LaserSharp 2000 imaging software (Carl Zeiss AG, Oberkochen, Germany). Pole cells were stained in embryos using anti-vasa antibodies (developed by A. C. Spradling and D. Williams, obtained from the Developmental Studies Hybridoma Bank (DSHB) developed under the auspices of the NICHD and maintained by the Univ. of Iowa, Dept of Biology, Iowa, USA) as outlined before (Hay et al. 1988). Goat secondary antibody coupled to alkaline phosphatase was obtained from Jackson Immuno-Research Laboratories (Dianova, Hamburg, Germany). Microphotographs of chromosomes, embryos, larvae or adults were taken with a Pixera ES120 digital camera (Optronics) coupled to a Zeiss Axiophot or to a Leica Wild M3C stereomicroscope using the Pixera Viewfinder Version 2.0 software. Figures were assembled using Corel Photo Paint, Corel Draw, Exel, and BoxPlotR software. Statistical significance of probes was determined by ANOVA two-tailed test for multiple comparisons using Dunnet's approach with raw p-values: P . 0.05 (not significant); Ã P , 0.05; ÃÃ P , 0.01; ÃÃÃ P , 0.001.
Data availability PKD mutant strains are available upon request. Supplemental data comprise 4 Supplemental Figures S1 to S4 in one file and one Supplemental Table S1. Figure S1 shows the complete chromosome spread of Figure 1D. Figure S2 contains the comparison of the developmental timing of PKD mutants and several additional controls. Figure S3 contains the comparison of the lifespan of PKD mutants and several additional controls. Figure S4 contains the comparison of the starvation resistance of PKD mutants and several additional controls. Table S1 contains details on the kinase mutants used in the candidate screen. Supplemental material available at FigShare: https://doi.org/10.25387/ g3.7667078.

Generation of PKD null mutant alleles by ends-out homologous recombination
The Drosophila PKD locus has been mapped to the right arm of chromosome 3 at position 91A2 (https://flybase.org). In order to generate specific PKD mutant alleles, we employed the technique of ends-out homologous recombination ( Figure 1B, C) (Gong and Golic 2003). In allele PKD 26 , all relevant coding regions of PKD were replaced by the white + gene used for selection, i.e., the regulatory and the catalytic domain of the kinase ( Figure 1A-C). Replacement of the PKD locus by the white + gene in PKD 26 was confirmed by chromosomal in situ hybridization ( Figure 1D, supplemental Figure S1). In allele PKD cl4 , the white gene was deleted by Cre I-mediated recombination as outlined before (Rong et al. 2002). The resultant deletion was confirmed by sequence analysis of a PCR amplificate overlapping the breakpoint ( Figure 1B, C). Absence of PKD transcripts was verified by in situ hybridization on whole mount PKD 26 embryos ( Figure 1E) and by RT-PCR ( Figure 1F) and qRT-PCR for both alleles, respectively.
PKD null mutants are homozygous viable without apparent phenotype Based on our earlier studies we expected loss of PKD to affect growth and perhaps cause lethality (Maier et al. 2007). Yet, homozygous PKD null mutants were viable without apparent phenotype with respect to size or the overall pattern of the external organs ( Figure 2A). Moreover, developmental timing, i.e., emergence of adult flies ( Figure 2B) was similar to y 1 w 67c23 animals used for control, as this was the parental origin of the starting line. As the PKD 26 and PKD cl4 mutant strains were not isogenic, however, we used additional controls to account for possible variance of the genetic background. Indeed, a great variance between the y 1 w 67c23 genotype and other control strains, i.e., w 1118 , Canton-S and notably Oregon R was observed (Figure 2, supplemental Figure S2). Still, PKD 26 mutants were within the normal range, and neither pupae nor adults developed particularly slower than the controls (Figure 2, supplemental Figure S2). We also addressed female fertility, which was indistinguishable between the homozygous PKD mutants and their heterozygous siblings ( Figure 2C). As observed for fly hatching time, the life span also differed remarkably between several control strains ( Figure 2D, D'; supplemental Figure S3). It was shortest in PKD 26 mutant males, in contrast to that of the PKD 26 females that lived longer than the y 1 w 67c23 control ( Figure 2D, D'; supplemental Figure S3).
As it was recently suggested that PKD may play a role in weight control and fat homeostasis (Ashe et al. 2018), we measured body weight and fat content of our mutants ( Figure 3A, B). Overall, there was not a great difference between PKD 26 mutants and y 1 w 67c23 flies, which were, however, of lower weight compared to the wild type OreR but not compared to w 1118 used for further control ( Figure 3A, B). Moreover, OreR males, but not the females, had a significantly lower relative fat content compared to PKD 26 mutants and y 1 w 67c23 ( Figure  3B). The specific larval weight was measured in a buoyancy assay ( Figure 3C): here the PKD 26 mutants matched the y 1 w 67c23 control, whereas PKD cl4 were more similar to the OreR wild type. We conclude that the PKD mutants lie in between the two control strains. In addition, we tested sensitivity of PKD mutants toward starvation under wet ( Figure 3D) and dry conditions ( Figure S4): males were slightly more sensitive than the y 1 w 67c23 control, whereas females appeared more resistant but less than OreR ( Figure 3D and supplemental Figure S4). Again, great variability toward dry starvation was observed among several strains: y 1 w 67c23 and the wild type strain Canton-S were most sensitive (supplemental Figure S4). The life-shortening effect of a highsugar diet (Magwere et al. 2004, Al Saud et al. 2015 was also observed for the PKD 26 mutant. When flies were raised on high sucrose, life span of PKD 26 mutants did not differ from y 1 w 67c23 control, irrespective of sex ( Figure 3E).
PKD null mutants are sensitive for oxidative stress As PKD is apparently not strictly required for fly development, life span and fertility, we wondered whether this kinase might be involved in stress regulation. We assayed sensitivity of PKD 26 mutants toward a variety of Figure 2 Phenotypic analyses of homozygous PKD mutant animals. (A) Comparison of control animals (OreR and y 1 w 67c23 ) with PKD 26 and PKD cl4 mutant animal at first (L1), second (L2) and third (L3) larval stage, of pupae and of adult males. Typical examples are shown, no apparent size difference was observed. Scale bar is 1 mm in all panels. (B) Developmental timing: PKD 26 mutants and control flies y 1 w 67c23 and OreR were reared under identical conditions at 25°C. Eclosion of adults was monitored over time (in days after egg deposition), and is shown as the fraction of the total number of animals (given in the legend). Four experiments were sampled: standard deviation is depicted as bars. Note similar timing of PKD 26 mutants and controls (additional controls are within supplemental Figure S2). (C) Fertility of female flies is given as offspring per female per day at 18°C (n $ 12 experiments as indicated; PKD cl4 n = 4 experiments). Note similar fertility of homo-and heterozygous PKD mutants; doubly balanced flies produced least offspring. Box-Plots depict medians as Center lines; box limits indicate the 25 th and 75 th percentiles; whiskers extend 1.5 times the interquartile range. Sample points are given for each genotype in parentheses. Statistical analysis was performed with a two-tailed ANOVA test relative to TM3 Sb /TM6B using Dunnet's approach (not significant). (D, D') Lifespan of male and female control flies (OreR and y 1 w 67c23 ) vs. PKD 26 mutant flies is depicted. Surviving flies over time (in days after egg deposition) is shown as fraction of the total (given in the respective legend). Whereas PKD 26 males have a reduced life span (D), the PKD 26 females live slightly longer than the control y 1 w 67c23 (D'). OreR flies live longest. See also supplemental Figure S3 for additional controls and statistics.
stressors. PKD 26 mutants resisted the application of heat in a dry oven or in a water bath similar to controls ( Figure 4A, B), whereas p38a 1 mutants were more and bsk 1 mutants slightly less sensitive, as described in the literature (Wang et al. 2003, Craig et al. 2004. Moreover, no particular cold sensitivity was observed ( Figure 4C). Likewise, PKD 26 flies tolerated osmotic stress similar to the control ( Figure 4D). Oxidative stress, however, was less tolerated by the PKD null mutant alleles compared to the y 1 w 67c23 and OreR controls ( Figure 4E). This is in line with the described role of mammalian PKDs to protect cells from oxidative stress mediated apoptosis. Here, PKD is involved in mitochondrial ROS detoxification by driving the expression of Manganese superoxide dismutase (MnSOD) (Storz 2007;Cobbaut and Van Lint 2018). To address a likewise role for Drosophila PKD, expression of superoxide dismutase was measured by quantitative RT-PCR in the two mutant PKD alleles. We addressed both, the expression of MnSOD (SOD2) as well as of Cu/ZnSOD (SOD1) that act in mitochondria and the cytosol, respectively, to detoxify the cell from superoxide released from mitochondria. In the absence of PKD, SOD1 and SOD2 levels were very similar those of the OreR control: a slight but not significant decrease was observed for SOD1 expression (0.72 -0.87 fold), whereas expression of SOD2 was slightly increased (1.35-1.38 fold) ( Figure 4F). Apparently, SOD regulation in Drosophila is largely independent of PKD activity in unstressed conditions. Figure 3 Influence of PKD on body weight and starvation resistance. (A) BoxPlot representation of the fresh weight of animals of the given genotype weighed in batches of five given in mg. The left panel shows the weight range of males, the right panel that of females. (B) The relative fat content of male and female flies of the given genotype was determined on 25-30 animals per experiment relative to the dry weight. PKD 26 mutants are not different from y 1 w 67c23 control, whereas OreR males have a significantly larger body weight, and hence, a relatively low fat content. BoxPlots in (A) and (B) depict medians as Center lines; box limits indicate the 25 th and 75 th percentiles; whiskers extend 1.5 times the interquartile range; outliers are represented by dots. Sample points are given for each genotype in parentheses. Statistical analysis was performed with a two-tailed ANOVA test relative to y 1 w 67c23 using Dunnet's approach with ÃÃÃ , P , 0.001 and ÃÃ , P , 0.01. (C) Larval specific weight was determined in a floating assay with sucrose of different density. Number of floating larvae is indicated for 12% sucrose (blue, 7 experiments with 10 larvae each) and 10% sucrose (orange, 8 experiments with 10 larvae each). At 8% sucrose, all larvae sank (70 animals tested per genotype). Note that PKD 26 resembles y 1 w 67c23 control, and PKD cl4 the OreR control. (D) Sensitivity to starvation stress was measured as survival of male and female flies on wet filter paper over time; dead flies were counted regularly. Whereas males of different genotype were similar, female OreR flies were highly resistant and female y 1 w 67c23 very sensitive toward starvation (4-6 experiments each; the total number of tested animals is given in the legend for each genotype). (E) Lifespan was determined on medium with normal and high sucrose content (+s). High sucrose medium shortened life span of PKD 26 mutants as well as y 1 w 67c23 control irrespective of sex to similar values.

Migration of border cells and pole cells appears normal in PKD mutants
Human PKDs are pivotal to cell motility by regulating cytoskeletal dynamics. Specifically, human PKDs act as negative regulator of Slingshot-phosphatase, thereby influencing cofilin availability and actin filament de/polymerization (reviewed in Olayioye et al. 2013). We have shown before that overexpression of active Drosophila PKD-SE negatively regulates Slingshot activity, suggesting a likewise involvement of PKD in cytoskeletal dynamics in the fly (Nagel et al. 2010, Bari sić et al. 2011. Accordingly, we might expect an impact of a loss of Drosophila PKD on cell migration. We studied two well-characterized processes of cell migration during Drosophila development. First, we monitored migration of border cells during oogenesis. Border cells are specifically determined follicle cells that actively migrate from the anterior tip of the follicle in between the nurse cells to the anterior border of the oocyte (reviewed in Montell 2006). The timing of the migration can be followed by the columnar follicle cells that retract in parallel toward the posterior of the oocyte (reviewed in Montell 2006). No difference was seen in border cell migration behavior between PKD cl4 and control ( Figure 5). Second, we studied pole cell migration during embryogenesis. Pole cells are the primordial germ cells of Drosophila. They arise at the posterior of the embryo and migrate through the posterior midgut and along the germ band to populate the gonadal mesoderm (Hay et al. 1988;reviewed in Montell 2006). This process appears normal in PKD mutant embryos that moreover, display the same number of pole cells than the wild type ( Figure 6). We conclude that PKD is not required for these two processes of cell migration in Drosophila. As the overexpression of an activated isoform PKD-SE can influence the cytoskeleton (Nagel et al. 2010, Bari sić et al. 2011, the most likely explanation for a lack of migration defects is the presence of redundant kinases that adopt the function of PKD in its absence.

A small candidate screen on potentially redundant kinases
The fact that a loss of Drosophila PKD appears to be without major phenotypic consequences suggests that it may act redundantly to some other kinase, most likely members of the family of PKC or CAMK kinases. Candidates located on the first or second chromosome were selected by the availability of mutant alleles, as shown in Supplemental  Table S1, thereby covering most of the Drosophila PKC family members (four out of five), and about one quarter of all CAMK family members (7 of 30) (Morrison et al. 2000). The mutants were combined with PKD cl4 to record survival rate of the offspring. The fraction of balanced vs. homozygous PKD cl4 mutants was determined in the heterozygous background of the relevant kinase mutant. Survival rate of PKD cl4 homozygotes was reduced to 70-80% in a sqa f01512 and Pkcdelta e04408 Figure 4 Stress resistance. Flies of the given genotype were subjected to several different stressors, exposure to dry heat (A) or wet heat (B), to cold (C), to high NaCl content for osmotic stress (D), and to oxidative stress by paraquat exposure (E). Percent of flies surviving the treatment was determined in (A) and over time in (B, D, E), respectively, whereas in (C) recovery time from treatment was measured. (A) BoxPlot representation of survivors of heat treatment is shown. Center lines show the medians; box limits indicate the 25 th and 75 th percentiles; whiskers extend 1.5 times the interquartile range; dots represent outliers. Numbers in parentheses indicate experiments with each approximately 30 flies. Statistical analysis was performed with a two-tailed ANOVA test relative to y 1 w 67c23 using Dunnet's approach with P , 0.001, ÃÃÃ . (B-E) Total number of animals tested for each genotype is given in the legend. (F) Expression levels of Cu/ZnSOD (SOD1) and of MnSOD (SOD2) were quantified by qRT-PCR in the PKD mutants relative to OreR control. Data were assembled from three biological and two technical replicates. Minimax depicts 95% confidence, median corresponds to expression ratio. Tub56D and Tbp were used as reference genes. Amplification efficiencies for SOD1 (0.92), SOD2 (0.98), for Tub56D (0.97) and Tbp (0.96) were accounted for in determining relative quantities by REST (Pfaffl et al. 2002).
heterozygous background (Table 1). A likewise increased mortality of the PKD cl4 homozygotes was observed when either Pkcdelta e04408 or Drak BG00876 were homozygous in addition (Table 1). Unfortunately, double mutants could only be tested for these two, as most of the kinase candidate mutants are homozygous lethal. As most kinase mutants are fully recessive, however, heterozygotes are unlikely to be influenced by a loss of PKD activity. Pronounced synthetic lethality, as we might have expected if one of the kinases required PKD to supplement its activity, was not observed.

DISCUSSION
We have generated specific knock out alleles of Drosophila PKD, which to our surprise were homozygous viable without apparent phenotype. Based on our own previous experiments, we expected conspicuous phenotypes, resulting for example from defects in the cytoskeleton or in protein secretion (Maier et al. 2007;Nagel et al. 2010;Bari sić et al. 2011). Lack of drastic phenotypes, however, suggests redundancy: presumably other serine-threonine protein kinases of the PKC/CAMK family act in place of PKD. Our tentative candidate kinase screen indeed uncovered three kinases, Drak, Sqa and Pkcd that may be linked with PKD activity, since respective hypomorphic alleles impeded fly viability in the absence of PKD (Table 1). The former two belong to the CAMK and the latter to the PKC family of kinases (Morrison et al. 2000). These results are rather intriguing in light of the known roles for mammalian PKDs in oxidative stress response and in the regulation of cell motility and invasion (Fu and Rubin 2011, Cobbaut and Van Lint 2018, Olayioye et al. 2013. It is well known that mammalian PKDs oppose the apoptotic effects of oxidative stress in a variety of cells (reviewed in Storz 2007;Cobbaut and Van Lint 2018). In this process, PKD1 is activated by PKCd in response to phospholipase D activation at the mitochondrial membrane. Consequently, active PKD1 inhibits mitochondrial depolarization and decreases the release of cytochrome C, thereby protecting cells from apoptosis, and more generally from oxidative damage. In addition, PKD1 mediates expression of MnSOD, the superoxide dismutase that detoxifies the cell from superoxides released from the mitochondria. Altogether, PKD1 activity results in pro-survival signals in oxidative stress (reviewed in Storz 2007;Cobbaut and Van Lint 2018). We have observed that PKD 26 mutants display an increased sensitivity toward oxidative stress ( Figure 4E), whereas in unstressed conditions transcription levels of MnSOD and Cu/ZnSOD were similar to wild type ( Figure 4F). In addition, downregulating the activity of the Drosophila PKDd homolog strongly increased the mortality of PKD cl4 homozygotes (Table 1). Together, these data strongly support a likewise protective role for Drosophila PKD in combating oxidative stress in the fly.
In mammals, PKDs play an important role in actin cytoskeletal dynamics via the regulation of the phosphatase Slingshot (reviewed in Olayioye et al. 2013). Using overexpression experiments, we have shown earlier that Drosophila PKD likewise affects Slingshot activity, and consequently the dynamics of filamentous actin turnover (Bari sić et al. 2011, Nagel et al. 2010. However, the complete absence of PKD had no apparent influence on the migration of border or pole cells, indicating redundancy for PKD activity in the context of cell motility. A possible candidate for a redundant kinase might be Drak, which has been involved in actomyosin contractility and dynamics (Neubueser andHipfner 2010, Chougule et al. 2016). Our work shows a genetic interaction between PKD and Drak, since a downregulation of Drak activity in the hypomorphic allele Drak BG00876 increased mortality of PKD cl4 mutants considerably (Table 1). We thus conclude that Drak and PKD may act together in the regulation of cell motility.
RNAi mediated knock down of PKD activity affected cell growth and differentiation (Maier et al. 2007). In accordance with these data, Ashe and co-workers reported growth defects resulting from PKD depletion (Ashe et al. 2018). Moreover, a specific role for PKD in the release of Drosophila insulin like peptide ILP2 was reported, explaining the reduced weight as well as starvation sensitivity of the dPKD H allele used in their study. The mutant dPKD H allele has a 70% reduced PKD mRNA level in third instar larvae (Ashe et al. 2018). One might have expected an even stronger phenotype in the complete absence of PKD, which Figure 5 Border cell migration. (A-A'') Sketch of border cell migration at stages 9 to 10 during Drosophila oogenesis. Early in stage 9, a cluster of border cells dispatches from the anterior tip of the follicle (A, red arrow), to migrate in between the large, polyploid nurse cells (A') to finally reach the anterior of the oocyte in stage 10 (A''). At the same time, the columnar follicle cells retract posteriorly; they are a useful marker for orderly migration (read arrowheads). Border cells are depicted in red, and highlighted by a red frame and a red arrow. (B-C'') Cell outlines were visualized with phalloidin staining (black). Confocal images taken from the respective stages of control (B-B'') and PKD cl4 mutant females (C-C'') are shown inverse for better visibility. Border cells are marked by red frame and arrow; columnar follicle cell margins by read arrowheads, respectively, as in (A-A''). The migratory behavior of border cells appears not different between the two genotypes.
however, was not observed. PKD null mutants were largely normal with respect to growth and weight. How can we explain the different results obtained from the null alleles PKD 26 and PKD cl4 compared to dPKD H ? First, dPKD H might rather match a null allele based on the low residual PKD expression levels. In this case, the null alleles may not show dramatically stronger phenotypes. The observed differences may then be attributed to external factors, i.e., rearing conditions (food composition, relative humidity, temperature etc.) or internal factors, i.e., genetic background. The latter appears more likely, as we have seen a high variance among various control strains regarding several tested parameters (Figures 2, 3, S2-S4). Still, none of the control strains displayed such strong and specific defects as the dPKD H allele in Ashe et al. (2018), which was developmentally retarded and underweight. We therefore favor the possibility of a second site hit in the dPKD H allele that influences its phenotype. PKD null mutants are evidently without or very little phenotype on their own, presumably due to the function of redundant kinase(s). Mutation in (any) one of these kinases is without conspicuous phenotype, at least when heterozygous. The absence of PKD, however, may uncover a growth defect in such a mutant, if the respective kinase is involved in the regulation for example of TOR signaling activity. A rescue of this growth defect by addition of PKDfor example by ubiquitous overexpression like Act . dPKD (Ashe et al. 2018) -is to be expected, since in this case, PKD can take over the redundant function. A possible candidate for such a redundant kinase Figure 6 Pole cell migration. (A) Embryos were stained with anti-vasa antibodies to mark the pole cells. Anterior is to the left, embryonic stages are indicated (st5-st15). Pole cells originate at the posterior pole (stage 5), and migrate through the posterior midgut (stage 8) to move posteriorly during the following stages (9-14), before they conglomerate with mesodermal cells (stage 15) to form the round gonadal anlagen in the embryo. No apparent differences are seen between the control Oregon R (OreR, upper row) and PKD cl4 mutants (lower row). (B) Moreover, the number of pole cells was not significantly different between the two PKD alleles, PKD 26 and PKD cl4 , and the control OreR. Center lines of the BoxPlots show the medians; box limits indicate the 25 th and 75 th percentiles; whiskers extend 1.5 times the interquartile range; dot represents outlier. Number of data points is given in parentheses for each genotype.
n Table 1 Candidate kinase screen Expected fraction of homozygous PKD cl4 animals inferred from the number heterozygous PKD cl4 siblings in a heterozygous background of the respective candidate kinase mutant is given in percent %. The final inter se cross was of the following genotype: M or Y/FM7; PKD cl4 /TM3Sb and M/CyO; PKD cl4 / TM6B, respectively, with M representing mutant allele or deletion might be spaghetti squash activator (sqa), which encodes a novel myosin light chain kinase with a role in starvation induced autophagy and the regulation of TOR signaling activity (Tang et al. 2011, Findlay et al. 2007). The sqa f01512 allele is fully recessive, and the heterozygotes are without apparent phenotype. Still, the sqa f01512 heterozygous background caused a dramatic lowered survival rate of less than 70% of PKD cl14 flies (Table 1). It is hence conceivable that PKD plays indeed a role in growth regulation, however, overlaps functionally with other kinases in this process, for example with Sqa. Overall, our results suggest that presumably PKD's absence can be replaced in the respective context by one or several other kinases with overlapping function. For example, Pkcd has not yet been assigned a specific role, as the mutants are viable without phenotype. Still, the mutant allele Pkcdelta e04408 markedly impedes fly viability in the absence of PKD activity, strongly indicating functional redundancy. For technical reasons, we could test only a small subset of the Drosophila CAMK family. We expect overlap of PKD with further protein kinases not yet included in this test.
To date the functions of PKDs have been studied primarily in model cell-culture systems, and only few data exist on PKDs' functions in the context of normal cells and tissues of intact organisms (Fu and Rubin 2011). Three models have been studied by now, D. melanogaster (this work), C. elegans and the mouse. Null mutants in either of the two PKD isoforms from C. elegans, named DKF-1 and DKF-2, are viable. Animals lacking DKF-1 display locomotory defects, whereas DKF-2 mutants impede adult life span by affecting stress and innate immunity responses (Feng et al. 2006, Feng et al. 2007, Ren et al. 2009). PKD1 mutant mice are viable and fertile, but appear semilethal with only half of the expected offspring. Embryonic fibroblasts derived from these mice are highly susceptible to apoptosis induced by oxidative stress or by starvation, which is mediated by PKCd activity (Zhang et al. 2015). In these cells, PKD1 is the key regulator in determining the threshold of mitochondrial depolarization affecting Bcl2-Bax fractions, implicating a role for PKD1 during aging and nutrient deprivation (Zhang et al. 2015). As neither PKD2 nor PKD3 can compensate the loss of PKD1, viability of the PKD1 mutants is most likely not explained by redundancy with the other PKD isoforms. Rather oxidative stress responses may be tissue specific, or defects are observed only when animals are subjected to non-laboratory, stressful conditions (Zhang et al. 2015). This work shows that Drosophila PKD null alleles are viable and fertile, however, are sensitive toward oxidative stress. Moreover, we uncovered genetic interactions of Drosophila PKD with three kinases PKCd, Drak and Sqa. Most likely overlapping functions exist between PKD and other kinases of the PKC/CAMK family, explaining the lack of apparent phenotypes in the null mutants. Whether the overlap is restricted to these three kinases or extends to a larger kinase family, and whether it applies only to certain tissues or processes, will require further investigations in the future.

ACKNOWLEDGMENTS
We are indebted to A. Cerny for floxing and initial phenotypic analyses of the PKD mutants. We greatly acknowledge S. Peter, J. Schmid, I. Wech and M. Zimmermann for invaluable technical assistance, and the B07 student course for data collection. We thank A. Hausser for critical reading and edition of the manuscript. The work was supported by a grant from the German science foundation to DM (MA 1328/8-1).