Numerous Serine/Threonine Kinases Affect Blood Cell Homeostasis in Drosophila melanogaster

Blood cells in Drosophila serve primarily innate immune responses. Various stressors influence blood cell homeostasis regarding both numbers and the proportion of blood cell types. The principle molecular mechanisms governing hematopoiesis are conserved amongst species and involve major signaling pathways like Notch, Toll, JNK, JAK/Stat or RTK. Albeit signaling pathways generally rely on the activity of protein kinases, their specific contribution to hematopoiesis remains understudied. Here, we assess the role of Serine/Threonine kinases with the potential to phosphorylate the transcription factor Su(H) in crystal cell homeostasis. Su(H) is central to Notch signal transduction, and its inhibition by phosphorylation impedes crystal cell formation. Overall, nearly twenty percent of all Drosophila Serine/Threonine kinases were studied in two assays, global and hemocyte-specific overexpression and downregulation, respectively. Unexpectedly, the majority of kinases influenced crystal cell numbers, albeit only a few were related to hematopoiesis so far. Four kinases appeared essential for crystal cell formation, whereas most kinases restrained crystal cell development. This group comprises all kinase classes, indicative of the complex regulatory network underlying blood cell homeostasis. The rather indiscriminative response we observed opens the possibility that blood cells measure their overall phospho-status as a proxy for stress-signals, and activate an adaptive immune response accordingly.


Introduction
The hematopoietic system in Drosophila melanogaster serves primarily immune responses, as oxygen disperses from the trachea via the hemolymph to the various organs.There are three distinct blood cell types or hemocytes fulfilling the immunity function, plasmatocytes, crystal cells and lamellocytes.Plasmatocytes constitute the majority of blood cells with more than 90% in a healthy animal.They are bone fide phagocytes resembling mammalian macrophages, able to engulf and destroy pathogens and cellular debris, cells infected with viruses and apoptotic cells.Moreover, they secrete antimicrobial peptides and produce extracellular matrix during wound healing (for review [1][2][3]).In addition, plasmatocytes are central to blood cell homeostasis.They have the capability of self-renewal, i.e., regenerating and increasing the blood cell pool, thus serving as the source for pro-hemocytes.Moreover, the other two cell types may arise by trans-differentiation directly from plasmatocytes (for review [4]).
Crystal cells make up 2-5% of the total hemocytes, whereas lamellocytes differentiate particularly in response to endo-parasitic wasp infestation of larvae.Crystal cells share similarities with mammalian megakaryocytes.They also combat microorganisms, are Ser/Thr kinases with the potential to target Su(H) Ser269.We identified Pkc53E involved in Su(H) phosphorylation in response to larval parasitization, however, regulating blood cell homeostasis in normal conditions as well [47].In the course of the work presented here, about 20% of the known Drosophila Ser/Thr kinases were examined for their role in blood cell formation.Two assays were employed, global and blood cell-specific overexpression and downregulation, respectively.Unexpectedly, the vast majority of the investigated Ser/Thr kinases of any kinase class influenced crystal cell homeostasis.The contribution of protein kinases to hematopoiesis has remained fairly understudied to date, despite their defined roles in the various hematopoietic signaling pathways.However, not only kinases that are central core components of respective signaling cascades influenced blood cell formation.Instead, the rather general involvement of many kinases points to a much more complex regulation of blood cell homeostasis than anticipated so far.Perhaps, blood cells measure their overall phospho-status as a proxy for stress-signals, and activate an adaptive immune response accordingly.Precursors, called pro-hemocytes, are determined in the embryonic head mesoderm by the activity of Serpent (Srp) and U-shaped (Ush).Pro-hemocytes are able to proliferate, and to differentiate into predominant plasmatocytes under the influence of Glial cells missing transcription factors (Gcm/Glide, Gcm2).The alternative crystal cell fate is induced by Notch activity via Lozenge (Lz); it can also arise by trans-differentiation of plasmatocytes [3,4].Lamellocytes differentiate in response to parasitism, involving the activity of several signaling pathways, including JNK, Toll, JAK/STAT and the repression of Notch activity [3].(b-d) Sessile crystal cells can be detected through the cuticle of heated larvae [48,49].Control larvae (Su(H) gwt ) display around 100 crystal cells in the dorsal hematopoietic pockets of the last two posterior segments (b).The phospho-mimetic allele Su(H) S269D  barely develops any crystal cells (c), whereas the phospho-mutant Su(H) S269A displays increased numbers (d) [36].Representative images are shown.Scale bar, 250 µm.

Fly Work
Fly strains used in this work are listed in the Supplemental Table S1.Flies were raised on standard fly food at ambient temperature (25 °C).Comparative crosses were set up in parallel.For the ectopic expression of the kinase constructs, the Gal4/UAS system was applied [50].We either used the ubiquitous driver line da-Gal4 (BL55849) [51] or the hemocyte-specific driver line HmlΔ-Gal4 (BL30141, FBrf0210198; hml-Gal4), which induces transgene expression in hemocytes prior to or at the stage of crystal-cell commitment [34].

Fly Work
Fly strains used in this work are listed in the Supplemental Table S1.Flies were raised on standard fly food at ambient temperature (25 • C).Comparative crosses were set up in parallel.For the ectopic expression of the kinase constructs, the Gal4/UAS system was applied [50].We either used the ubiquitous driver line da-Gal4 (BL55849) [51] or the hemocyte-specific driver line Hml∆-Gal4 (BL30141, FBrf0210198; hml-Gal4), which induces transgene expression in hemocytes prior to or at the stage of crystal-cell commitment [34].

Cloning of UAS-Kinase Constructs
The cDNAs encoding the respective kinases were obtained from the Drosophila Genome Resource Center (DGRC), cdk7 (FMO01629), Dyrk3 (RE60792), Fray (RE53265) and Gskt (FMO04376).The same scheme was used for tagging and cloning either cdk7 or Gskt.Firstly, the kinase cDNA was amplified from its original vector using oligonucleotides with Xho I (5 ′ ), and Xba I (3 ′ ) overlaps.Secondly, it was cloned into a likewise opened pBT vector (Stratagene) harboring 3xHA tags, inserted as annealed respective oligonucleotides into the Acc 65I/Xho I sites, named pBT-HA-STaRT.The HA-tagged cDNA was then released by Acc 65I/Xba I digest, to be reinserted into likewise opened pUASt-attB vector [58].Dyrk3 cDNA was amplified with primers containing Sal I/Bam HI overhangs, and inserted into likewise opened pBT-HA-STaRT.The HA-tagged Dyrk3 cDNA was then released with Acc 65I and Xba I and cloned into pUASt-attB [58] as above.In case of Fray, the primers contained Sal I/Eco RI overhangs.The amplicon was inserted into Xho I/Eco RI opened pBT-HA-STaRT, released with Acc 65I and Eco RI and shuttled into pEGFP-N1 (Clontech).After a second release with Bam HI/Xho I, the insert was shuttled into the Bgl II/Xho I sites of pUASt-attB [58].Constructs were verified by diagnostic digests and by sequencing before generating transgenic lines.Primers used for cloning are listed in the Supplemental Table S2.

Quantification of Crystal Cells
Crystal cells were visualized through the larval cuticle as described before [48,49].Larvae from the respective crosses were reared on normal fly food.Overcrowding was strictly avoided to synchronize the larval stages that were heated in batches as described previously [47].Dorsal views from the larval posterior end were taken with a Pixera Pro 120ES camera (Pixera, Santa Clara, CA, USA), mounted onto a Wild M3Z stereo microscope (Leica, Wetzlar, Germany) using Pixera Viewfinder 2.5 software.Melanized crystal cells were counted in the last to segments using the 'Cell Counter' Plugin of Image J 1.51 (Fiji) software [60].At least two biological replicates were performed; the total number of analyzed larvae of each genotype is given within the figure.

Selection of Ser/Thr Kinases
Ser/Thr kinases were selected by their potential to recognize the Su(H) S269 target sequence FNRLRSQTVSTR by the following two approaches.First, an in silico search with the GPS3.1 software tool yielded 20 Drosophila kinase candidates [47,61].Second, of 245 human kinases, 62 had the capability to in vitro phosphorylate the Su(H) beta-trefoil domain overlapping the S269 target site (ProQinase, Freiburg, Germany) [47].These human kinases correspond to 40 Drosophila kinase candidates, containing ten of the first group.Altogether, 46 kinase candidates were tested in gain and loss of function screens.

Statistical Analyses
For statistic evaluation of the data, we used Microsoft Excel to calculate mean, standard deviation (SD) and median; statistical significance was evaluated by using a two-tailed analysis of variance (ANOVA) approach for multiple comparisons according to Dunnett's test relative to the control; values were *** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, and p > 0.05 not significant (n.s.).Statistical graphs were created with Excel 2016 (MS Office) and Origin R© 2018b software (OriginLab Corporation, Northampton, MA, USA).

Experimental Design
Serine/Threonine (Ser/Thr) kinases constitute the largest group of protein kinases.They have been further classified according to their structure, function and targets (Figure 2a) [62][63][64][65][66][67].In this work, Ser/Thr kinases were selected in silico and in vitro by their potential ability to target the sequence FNRLRSQTVSTR corresponding to the Su(H) S269 target sequence [47,61].This way, 46 kinases were selected, representing roughly 20% of all Drosophila Ser/Thr kinases (Table 1).
1 Kinases used in both, gain and loss-of function assays. 2 Abbreviation of classes as in Figure 2.
Two assays were employed, global and hemocyte-specific (1) overexpression and (2) downregulation of a particular kinase, respectively, thereby considering adequate overall class representation (Figure 2b,c).The assays were based on the observation that the phospho-mimetic Su(H) S269D allele largely fails to produce crystal cells, in contrast to the phospho-deficient Su(H) S269A allele developing more crystal cells than controls (Figure 1b,c) [36].Hence, we expected a substantial impact on crystal cell formation upon the activation or the inhibition of a given kinase involved in hematopoiesis.Sessile crystal cells can be visualized in a coarse experiment: crystal cells rupture upon heating of third instar larvae, resulting in a melanization reaction [3,48,49].The blackened cells are visible through the larval cuticle; they were recorded as an approximation for crystal cell numbers, allowing the quantitative comparison of multiple genotypes (Figure 1b-d).
resulting in a melanization reaction [3,48,49].The blackened cells are visible thro larval cuticle; they were recorded as an approximation for crystal cell numbers, a the quantitative comparison of multiple genotypes (Figure 1b-d).

Influence of Ubiquitous and Hemocyte-Specific induction of Ser/Thr Protein Kinase Blood Cell Homeostasis
In the first line of experiments, the effects of an overexpression of Ser/Thr ki crystal cell numbers were investigated.In order to determine the specific impact o cells and distinguish it from overall effects, we wanted to compare the outcome eral overexpression and a cell type-specific overexpression.Using the Gal4/UAS the respective UAS lines are expressed in a temporally and/or spatially restricted [50].The da-Gal4 line is a ubiquitously expressing driver line, which allows ind general expression throughout embryogenesis and early larval life [51].The Hm (hml-Gal4) driver line, however, drives UAS constructs specifically within differ hemocytes prior to or at the stage of crystal cell commitment [34].Whereas da expected to induce kinase expression in all tissues, including blood cell precurso Gal4 is highly specific to determined blood cells.Hence, kinase expression may o the path of differentiation but not induce de novo differentiation.A total of 22 kinases were assayed in the overexpression experiments.Three EP lines and fifte lines were used in this study (see Supplemental Table S1).In addition, four UAS pression lines were established in the course of this work (Cdk7, Dyrk3, Gskt, an To this end, the respective cDNAs were PCR-amplified and cloned into a suitab attB transformation vector [58].Transgenic lines were established by site-specifi bination, integrating the constructs at the identical chromosomal position to av position effects [58].
For the assay, control and experimental crosses were set up in parallel; the re UAS-kinase line itself served as control to be compared with the effects of a glo hemocyte-specific ectopic expression.We hypothesized that the effects of global pression may reflect a more general activity, e.g., on proliferation, whereas hemoc cific effects may reflect a more specific role during hematopoiesis and during h differentiation, respectively.The results are shown in Figure 3, allowing to cla

Influence of Ubiquitous and Hemocyte-Specific Induction of Ser/Thr Protein Kinases on Blood Cell Homeostasis
In the first line of experiments, the effects of an overexpression of Ser/Thr kinases on crystal cell numbers were investigated.In order to determine the specific impact on blood cells and distinguish it from overall effects, we wanted to compare the outcome of a general overexpression and a cell type-specific overexpression.Using the Gal4/UAS system, the respective UAS lines are expressed in a temporally and/or spatially restricted manner [50].The da-Gal4 line is a ubiquitously expressing driver line, which allows inducing a general expression throughout embryogenesis and early larval life [51].The Hml∆-Gal4 (hml-Gal4) driver line, however, drives UAS constructs specifically within differentiated hemocytes prior to or at the stage of crystal cell commitment [34].Whereas da-Gal4 is expected to induce kinase expression in all tissues, including blood cell precursors, hml-Gal4 is highly specific to determined blood cells.Hence, kinase expression may only alter the path of differentiation but not induce de novo differentiation.A total of 22 different kinases were assayed in the overexpression experiments.Three EP lines and fifteen UAS lines were used in this study (see Supplemental Table S1).In addition, four UAS-overexpression lines were established in the course of this work (Cdk7, Dyrk3, Gskt, and Fray).To this end, the respective cDNAs were PCR-amplified and cloned into a suitable UAS-attB transformation vector [58].Transgenic lines were established by site-specific recombination, integrating the constructs at the identical chromosomal position to avoid any position effects [58].
For the assay, control and experimental crosses were set up in parallel; the respective UAS-kinase line itself served as control to be compared with the effects of a global or a hemocyte-specific ectopic expression.We hypothesized that the effects of global overexpression may reflect a more general activity, e.g., on proliferation, whereas hemocyte-specific effects may reflect a more specific role during hematopoiesis and during hemocyte differentiation, respectively.The results are shown in Figure 3, allowing to classify the kinases into inhibitors or mediators of crystal cell formation, and/or kinases with a more general role.S1).Crystal cells in the dors hematopoietic pockets were recorded in the last two segments; every dot represents one larva (n, shown below the X-axis).UAS-lacZ served as a control.Crystal cell increase is indicated in red an decrease in blue; coloration reflects significance.Whiskers show standard deviation, center line th median and center dot the average.ANOVA for multiple comparisons according to Dunnett's te relative to the respective UAS line was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, n.s.p > 0.05 n significant).
Only two kinases impeded crystal cell formation significantly when specifically e pressed in hemocytes, BubR1 and the activated isoform of Pkc53E EDDD .We recent Figure 3. Crystal cell formation in response to ectopic expression of Ser/Thr kinases.Expression of the indicated kinases was induced ubiquitously with da-Gal4, or specifically within hemocytes using hml-Gal4.UAS kinase strains are in alphabetical order (see Table S1).Crystal cells in the dorsal hematopoietic pockets were recorded in the last two segments; every dot represents one larva (n, as shown below the X-axis).UAS-lacZ served as a control.Crystal cell increase is indicated in red and decrease in blue; coloration reflects significance.Whiskers show standard deviation, center line the median and center dot the average.ANOVA for multiple comparisons according to Dunnett's test relative to the respective UAS line was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, n.s.p > 0.05 not significant).
Only two kinases impeded crystal cell formation significantly when specifically expressed in hemocytes, BubR1 and the activated isoform of Pkc53E EDDD .We recently showed that Pkc53E is indeed involved in the phosphorylation of Su(H) at S269, in accordance with the observed inhibition of crystal cell formation [47].BubR1, however, has several essential functions during mitosis, and the overexpression may affect mitotic timing of hemocytes [68].A loss of crystal cells was also seen upon the ubiquitous but not the hemocyte-specific induction of Gskt and Sik2 S1032A .The former shares similarities with GSK3 kinase, a core component of the Wnt signaling pathway, whereas the latter regulates energy homeostasis as well as Notch signal transduction [57,69,70].Hence, both may have a more general role in proliferation and development.The mild effects of a global overexpression of Akt1, Cdk7 and Wnk may be explained likewise [71][72][73][74][75][76][77].In contrast, the activated Pdk1 A467V and Par-1 kinases caused additional crystal cells in both the global and the hemocyte-specific overexpression situations, whereas three kinases, Asator, CaMKII act and Dyrk3, induced a hemocyte-specific increase in crystal cell numbers only (Figure 3).Pdk1 is a master kinase with a crucial role in cell growth [56,78,79], easily explaining the general increase in cell numbers.Par-1, however, plays an important role in cell polarization and tumor suppression, and balances proliferation by influencing the Hippo signaling pathway [80][81][82].Asator is involved in regulating microtubule spindle function, and may hence relate to cell cycle control [66,83], whereas both Dyrk3 and CaMKII have been connected to calcium signaling as well as to Drosophila hematopoiesis [22,84,85].A special case is represented by the activated Tkv CA construct; the global overexpression induced a reduction in crystal cells, whereas the hemocyte-specific expression resulted in increased numbers (Figure 3).Tkv encodes a Dpp receptor subunit, i.e., acts as a core component within the TGF-beta signaling cascade involved in pattern formation and development [86][87][88].Perhaps the global overexpression somehow interferes with the development of the embryonic head mesoderm or with cell migration, whereas the hemocyte-specific activation in the larvae acts agonistically [89][90][91][92].The global overexpression of Raf and Wee1, however, was lethal, whereas hemocyte-specific overexpression of the former increased crystal cell numbers.This result is expected since Raf is the central core component of RTK signaling pathways and hence expected to profoundly affect cell growth and survival, and induce proliferation when overactive [74,[93][94][95][96].The Wee kinase, in contrast, blocks entry into mitosis.Accordingly, a general overexpression may block development altogether, whereas induction in specified hemocytes appears without consequences [74,[97][98][99].In contrast, neither global nor hemocyte-specific overexpression of the UAS-lacZ control had an impact on crystal cell numbers (Figure 3).Together, these results show that a majority of kinases somehow influenced crystal cell formation, albeit only a few had been directly associated with hematopoiesis in the past.Presumably, these effects are largely indirect and not a result of a Su(H) S269 phosphorylation, rather reflecting the intricate cross-talk amongst signaling pathways via kinase activity.

Downregulation of Kinase Activity Reveals a General Involvement of Ser/Thr Kinases in Crystal Cell Formation
Surprised by the broad impact of the overexpression of Ser/Thr kinases on crystal cell development, we sought to analyze the effects of a knockdown of kinase activity.The rationale was that an overexpression of Ser/Thr kinases might be not very specific since kinases are known to cross-react and phosphorylate targets with a decent similarity [64][65][66][67]100].Moreover, not all of the overexpressed kinases were present in an activated form; hence, we may have missed some in the previous assay.Accordingly, we doubled the number of kinases to be analyzed in a loss of function screen to a total of 46 (see Table 1).
We were able to screen 17 kinase mutants for crystal cell formation (Figure 4).However, not all developed into third instar larvae, and five could only be investigated as heterozygotes (alc Ad2 , msn 102 , par-1 k06334 , Pkc98E f06221 , S6K I-1 ).The kinase mutants displayed significant changes in crystal cell numbers compared to the control.Not unexpectedly, reduced crystal cell numbers were observed in the cell cycle mutants cdk1 E1-23 and cdk7 del as well as in the sgg M1-1 and the raf 12 alleles, known to affect cellular growth and homeostasis [74,77,[95][96][97]99,101].However, larvae mutant for the other kinases developed far too many crystal cells (Figure 4).There was only one exception: the heterozygous msn 102 /+ mutant displayed nearly normal numbers, demonstrating the recessive character of the allele.In addition, blocking the activity of BubR1 and CamKII kinases, respectively, by inducing dominant negative forms in hemocytes, again caused a significant increase in crystal cell numbers (Figure 4).The downregulation of Akt1, CamKII and S6k caused an increase in crystal cell numbers similar to Pkc53E.We have shown previously that human Cells 2024, 13, 576 9 of 19 PKCα, representing Pkc53E, phosphorylates Ser269 in a Su(H) peptide, while the three other human homologues piloted the neighboring Thr271 [47].To date, we do not know whether Thr271 phosphorylation affects blood cell formation.However, all three kinases matched Pkc53E in their activity on crystal cell formation in support of this idea.Whether all the other kinases displaying similar phenotypes (i.e., Alc, Drak, Limk1, Mei-41, Msn, Par-1, Pkc98E, PKD, Slpr, Tkv and BubR1) also phosphorylate Su(H) or perhaps activate Pkc53E, Akt1, CamKII, S6k or upstream kinases, or whether they act independently in signaling pathways other than Notch, remains to be investigated.
allele.In addition, blocking the activity of BubR1 and CamKII kinases, respectively, by inducing dominant negative forms in hemocytes, again caused a significant increase in crystal cell numbers (Figure 4).The downregulation of Akt1, CamKII and S6k caused an increase in crystal cell numbers similar to Pkc53E.We have shown previously that human PKCα, representing Pkc53E, phosphorylates Ser269 in a Su(H) peptide, while the three other human homologues piloted the neighboring Thr271 [47].To date, we do not know whether Thr271 phosphorylation affects blood cell formation.However, all three kinases matched Pkc53E in their activity on crystal cell formation in support of this idea.Whether all the other kinases displaying similar phenotypes (i.e., Alc, Drak, Limk1, Mei-41, Msn, Par-1, Pkc98E, PKD, Slpr, Tkv and BubR1) also phosphorylate Su(H) or perhaps activate Pkc53E, Akt1, CamKII, S6k or upstream kinases, or whether they act independently in signaling pathways other than Notch, remains to be investigated.In order to confirm and extend the above result, we next employed a hemocyte-specific knockdown via RNAi, as a complete loss of a kinase activity may compromise development in general, or more specifically, mesoderm and/or hemocyte determination.To this end, we induced a total of 41 respective UAS-RNAi-lines with the HmlΔ-Gal4 driver In order to confirm and extend the above result, we next employed a hemocyte-specific knockdown via RNAi, as a complete loss of a kinase activity may compromise development in general, or more specifically, mesoderm and/or hemocyte determination.To this end, we induced a total of 41 respective UAS-RNAi-lines with the Hml∆-Gal4 driver covering an additional 27 Ser/Thr kinases (Figure 5).A complete list of the fly stocks used in these assays is presented in Table S1.
Confirming the above results, a knockdown of either kinase CamKII, Pkc53E, Pkc98E, PKD or Slpr, resulted in significantly higher crystal cell numbers.Moreover, a marked increase in crystal cell numbers was also observed, when either kinase Bsk, Cdk8, CG8173, CG14305, CKIIα, Doa, Dyrk3, Gskt, Msn, Niki, Pdk1, Pll, Put, or Wnk was downregulated.Albeit several of these kinases may play a role in JNK, TLR or WNT signaling, only Bsk and Msn as integral members of the JNK pathway, as well as Pll as central component of the TLR cascade, have ascribed roles during hematopoiesis [3,[101][102][103].Others like Pdk1 and Cdk8 are required in the broader context of apoptosis, proliferation and transcriptional regulation, or like Put for cardiac mesoderm development [3,56,66,77,79,104].Notably, Pdk1 acts as a master kinase upstream of many kinases including Akt1, S6k and Pkc53E, easily explaining the observed phenotypes by kinase cross-talk [56,78,79].Albeit it is conceivable that a network of cross-reacting kinases acts upstream of Pkc53E, hence indirectly regulating Su(H) phosphorylation, not all of them do.For example, whilst downregulation of Gskt induced crystal cell gain in the range of Pkc53E, the human homologue GSK3B was not able to phosphorylate a Su(H) peptide [47].Instead, Gskt, similar to Wnk, may play a role in Wnt signaling, which regulates hemocyte precursor development [3,69,76,101].Similarly, independent roles are expected for members of the JNK and TLR pathways, albeit cross-talk of signaling pathways during immune responses is well established [3,102,105].
In fact, 28 of the tested lines, representing 22 kinases, developed significantly more crystal cells when downregulated in hemocytes, i.e., nearly 70% of all tested lines, suggesting their involvement in blood cell homeostasis by restricting crystal cell numbers (Figure 5).In contrast, the knockdown of Dsor1 in hemocytes impaired crystal cell formation substantially (Figure 5).Apparently, Dsor1, the downstream kinase of Raf1 (MEK1), is critically required in hematopoiesis, in agreement with the results obtained for the raf 12 mutant (Figure 4) and earlier reports on the role of RTK signaling in hemocyte proliferation [96,[105][106][107].
Tissue-specific induction of RNAi did not induce significantly altered crystal cell numbers in about one third of the crosses (Figure 5).In some cases, this may be due to ineffectiveness of the respective RNAi-line, e.g., for CamKI, Cdk2, CG5790 and Tefu [108].UAS-RNAi lines affecting kinase activity of Asator, Fray, Hpo, Lic, Mnb and Sik2, however, have been reported before to be effective, suggesting that the respective kinases are not important for crystal cell development (Figure 5) [108,109].
A combined analysis of the loss of function screens reveals that in sum 30 of 46 tested kinases, i.e., about 65%, appear to limit crystal cell formation, whereas only five promoted it.For an overview, kinases were sorted by their effect on crystal cell numbers into five classes with little effect (+/−20% deviation), moderate (up to 50% deviation) or strong increase or decrease (over 50%), respectively, when compared to the reference (Figure 6).
We observed some discrepancies with regard to the overexpression analyses.For example, RNAi-mediated downregulation of Asator or Sik2 did not affect crystal cell numbers, whereas the respective overexpression caused significant changes, perhaps reflecting inefficient RNA interference, or cross-talk of these kinases in the overexpression context.Similarly, whereas the downregulation of the kinases Alc, Cdk8 and CG8173 increased crystal cell numbers, no effect was seen in response to an overexpression, presumably due to inactivity of the respective kinase.Likewise, whereas crystal cell numbers were increased in the S6K mutant, neither ubiquitous nor hemocyte-specific expression of the activated S6K impaired crystal cell formation.Moreover, both the overexpression as well as the downregulation of either Dyrk3, Par-1 or Pdk1 caused a substantial increase in crystal cell numbers, suggesting a major impact in a regulatory network rather than a specific role in hematopoiesis.RNAi-mediated downregulation of Ser/Thr kinases affects crystal cell homeostasis.Numbers represent crystal cells in dorsal hematopoietic pockets from the last two segments in larvae of the indicated genotype (in alphabetical order).Every dot represents one larva (n, as shown above Xaxis).Crystal cell increase is indicated in red and decrease in blue; coloration reflects significance.Whiskers, standard deviation; center line, median; center dot, average.For statistical analysis, ANOVA for multiple comparisons according to Dunnett's test relative to control was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, and n.s.p > 0.05 not significant).The respective UAS-RNAi line served as control.
bers represent crystal cells in dorsal hematopoietic pockets from the last two segments in larvae of the indicated genotype (in alphabetical order).Every dot represents one larva (n, as shown above X-axis).Crystal cell increase is indicated in red and decrease in blue; coloration reflects significance.Whiskers, standard deviation; center line, median; center dot, average.For statistical analysis, ANOVA for multiple comparisons according to Dunnett's test relative to control was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, and n.s.p > 0.05 not significant).The respective UAS-RNAi line served as control.We observed some discrepancies with regard to the overexpression analyses.For example, RNAi-mediated downregulation of Asator or Sik2 did not affect crystal cell numbers, whereas the respective overexpression caused significant changes, perhaps reflecting inefficient RNA interference, or cross-talk of these kinases in the overexpression context.Similarly, whereas the downregulation of the kinases Alc, Cdk8 and CG8173 increased crystal cell numbers, no effect was seen in response to an overexpression, presumably due to inactivity of the respective kinase.Likewise, whereas crystal cell numbers were increased in the S6K mutant, neither ubiquitous nor hemocyte-specific expression of the activated S6K impaired crystal cell formation.Moreover, both the overexpression as well as the downregulation of either Dyrk3, Par-1 or Pdk1 caused a substantial increase in crystal cell numbers, suggesting a major impact in a regulatory network rather than a specific role in hematopoiesis.

Discussion
In this work, we have addressed the role of Ser/Thr kinases in larval crystal cell formation of Drosophila.In our search for the kinase specifically piloting Su(H) during hematopoiesis, we observed that the large majority of kinases we screened affected crystal cell formation.This result was unexpected, because our screens did not specifically aim at kinases known to be involved in the hematopoiesis of Drosophila but rather at those predicted to recognize Ser269 in the beta-trefoil domain of Su(H) [45,47].Based on our results, most of these kinases have a role in restricting crystal cell development, whereas only a handful of kinases appeared to be required for crystal cell formation.
Previous screens applying RNA interference at large scale missed the apparently ubiquitous role of Ser/Thr kinases during hematopoiesis, which we attribute largely to the particular focus as well as the screening procedure.For example, one screen aimed at the identification of factors regulating the numbers of hemocytes deficient for the PDGF/VEGF receptor, coming up with components of the EcR pathway as suppressors and RTK signaling pathways as enhancers, including the downstream kinases, Akt1, S6K, Dsor and MAPK, respectively [106].Indeed, the MAPK cascade, apart from regulating hemocyte proliferation, specifically inhibits IMD signaling in larval hemocytes, thereby preventing spurious immune activation and limiting the immune response [107].In a recent large-scale screen, larval hemocytes were labelled with GFP, and the changes in fluorescent signal strength or distribution upon RNAi-mediated downregulation of gene activity was evaluated [110].This screen again picked up Akt kinase's relevance for hemocyte proliferation [110].A specific contribution of Ser/Thr kinases to crystal cell formation, however, was not uncovered although the screening procedure included crystal cells as well.Presumably, more subtle changes went unnoticed.Taking into account that crystal cells normally make up just about 5% of the larval hemocytes, a doubling or halving of their numbers would not stand out in the bulk of plasmatocytes.Other recent RNAi screens addressed lamellocytebased tumor formation without taking crystal cells into account [111,112].Screening the Drosophila genome for tumor suppressors, only two Ser/Thr kinases with established roles in immunity were picked up, accounting for roughly three percent of the screened Ser/Thr kinases in total [111].Likewise, any specific contributions of kinases to hemocyte differentiation or to specific immune responses were missed in the recent genome-wide analyses concentrating on the transcriptome rather than the proteome or the kinome [113][114][115].
Specifically aiming at crystal cell formation, our gain-and loss-of-function screens, however, uncovered an unexpected major contribution for Ser/Thr kinases.Nearly 75% of the investigated Ser/Thr kinases influenced blood cell homeostasis (35 of 46).Only the kinases Sgg, Cdk1, Cdk7, Raf and Dsor1 appeared to be required for the formation of crystal cells, since the respective mutant larvae, sgg M1-1 , cdk1 E1-23 , cdk7 del and raf 12 , as well as animals with a Dsor1 knockdown, lacked crystal cells, or developed significantly less than the controls (Figures 4 and 5).A loss-of-function of the majority of the studied kinases, however, entailed a significant increase in crystal cell numbers.Interestingly, all kinase classes are represented, pointing to a much more complex regulation of blood cell homeostasis than anticipated so far (Figure 7).Hence, any class contains representatives that may be involved in blood cell homeostasis.However, members belonging to the AGC, the OPK and the PKL classes appear particularly important, since most of their mutants displayed increased crystal cell numbers.In contrast, CMGC members give a mixed picture, with roughly half-and-half support vs. restriction of crystal cell formation, in accord with this group comprising kinases with a rather general role in development and cellular homeostasis.Quite intriguingly, our screens revealed an apparently general requirement o Ser/Thr kinases in blood cell homeostasis, and specifically in crystal cell formation.W did not prescreen kinase candidates for a likely involvement in hematopoiesis, but rathe for their competence to phosphorylate Su(H) at Ser269, thereby assessing just about 20% of Drosophila Ser/Thr kinases.Hence, the group of kinases whose downregulation in creases crystal cell numbers may be even larger.That all these kinases directly phosphor ylate Su(H) is out of the question.Given the intricate network of kinase cross-reactivity is conceivable that a fair number of them pilot a few kinases that either directly phosphor Quite intriguingly, our screens revealed an apparently general requirement of Ser/Thr kinases in blood cell homeostasis, and specifically in crystal cell formation.We did not prescreen kinase candidates for a likely involvement in hematopoiesis, but rather for their competence to phosphorylate Su(H) at Ser269, thereby assessing just about 20% of Drosophila Ser/Thr kinases.Hence, the group of kinases whose downregulation increases crystal cell numbers may be even larger.That all these kinases directly phosphorylate Su(H) is out of the question.Given the intricate network of kinase cross-reactivity it is conceivable that a fair number of them pilot a few kinases that either directly phosphorylate Su(H), like Pkc53E, or act upstream.Presumably, the majority of these kinases functions in signaling pathways unrelated to Su(H) activity.This may also explain why the downregulation of some kinases induced crystal cell numbers exceeding those caused by a loss of Pkc53E.For example, the rise in crystal cell numbers may reflect a general expansion of hemocytes or derive from the final differentiation of plasmatocytes into crystal cells.Excessive differentiation of plasmatocytes and crystal cells in response to JNK activation has been observed in the lymph gland [116], but neither in the hemolymph nor in the sessile compartment.In our hands, however, a knockdown of JNK activity caused an increase rather than a decrease in the numbers of sessile crystal cells.Crystal cells play various roles in innate immunity.Importantly, via the melanization cascade, they help in wound healing and combat bacterial and fungal infections by the production of cytotoxic reactive oxygen species (ROS) [3,102,117,118].Apart from its function in redox signaling and oxidative stress, ROS is an important signaling molecule during Drosophila hematopoiesis [116,119,120].Moreover, ample reports on redox-sensitivity of kinases and phosphatases point to a comprehensive cross-talk of several signaling pathways and the cellular redox signaling (reviewed in [119][120][121]).Redox-triggered mechanisms, cystein oxidation in particular, can alter the catalytic properties of kinases and phosphatases.Notably, redox regulation of protein tyrosine phosphatases and protein tyrosine kinases is well documented, and may alter the overall cellular phospho-status by also influencing Ser/Thr kinases.However, the activity of several Ser/Thr kinases is regulated by redox modifications as well.This has been demonstrated already for Akt1, MAPKs (e.g., JNK and p38), ATM, PKC and CAMKs (reviewed in: [119,121,122]).Clearly, signaling pathways regulating the cellular redox state are themselves subjected to redox regulation, arguing for an intensive cross-talk between the systems.
While our findings may reconcile the activity and cross-talk of the multitude of signaling pathways influencing blood cell development, proliferation and maintenance, we speculate that the innate immune responses of the Drosophila larva may in addition rely on the overall phospho-status within the larval blood cells.Hence, tinkering with kinases activities, e.g., by a hemocyte-specific knockdown, may lower the intracellular phospho-status, thereby promoting crystal cell differentiation and hence, increasing crystal cell numbers.This hypothesis could explain the rather indiscriminative response we observed.In this case, protein phosphorylation may act as a general measure for cellular stress, resulting in the activation of adaptive responses.

Conclusions
Blood cell homeostasis is under the influence of many signaling pathways granting proper immune responses to external stressors and infections.However, an unexpectedly large fraction of Ser/Thr kinases investigated in this work affected crystal cell formation, indicating their potential involvement beyond established hematopoietic signaling pathways.Some of these kinases may, directly or indirectly via cross-talk, cause the phosphorylation of Su(H) and thereby influence crystal cell numbers.Others, however, may be involved in blood cell formation through their activity in other signaling pathways unrelated to Notch signaling.Based on the rather indiscriminative response to the activity changes of Ser/Thr kinase in general, however, we speculate that the overall intracellular phospho-status in hemocytes serves as a fundamental measure for stress signals to adapt innate immune responses accordingly.

Figure 1 .
Figure 1.Blood cell homeostasis in Drosophila.(a) Scheme of larval blood cell development.Precursors, called pro-hemocytes, are determined in the embryonic head mesoderm by the activity of Serpent (Srp) and U-shaped (Ush).Pro-hemocytes are able to proliferate, and to differentiate into predominant plasmatocytes under the influence of Glial cells missing transcription factors (Gcm/Glide, Gcm2).The alternative crystal cell fate is induced by Notch activity via Lozenge (Lz); it can also arise by trans-differentiation of plasmatocytes[3,4].Lamellocytes differentiate in response to parasitism, involving the activity of several signaling pathways, including JNK, Toll, JAK/STAT and the repression of Notch activity[3].(b-d) Sessile crystal cells can be detected through the cuticle of heated larvae[48,49].Control larvae (Su(H) gwt ) display around 100 crystal cells in the dorsal hematopoietic pockets of the last two posterior segments (b).The phospho-mimetic allele Su(H) S269D barely develops any crystal cells (c), whereas the phospho-mutant Su(H) S269A displays increased numbers (d)[36].Representative images are shown.Scale bar, 250 µm.

Figure 1 .
Figure 1.Blood cell homeostasis in Drosophila.(a) Scheme of larval blood cell development.Precursors, called pro-hemocytes, are determined in the embryonic head mesoderm by the activity of Serpent (Srp) and U-shaped (Ush).Pro-hemocytes are able to proliferate, and to differentiate into predominant plasmatocytes under the influence of Glial cells missing transcription factors (Gcm/Glide, Gcm2).The alternative crystal cell fate is induced by Notch activity via Lozenge (Lz); it can also arise by trans-differentiation of plasmatocytes[3,4].Lamellocytes differentiate in response to parasitism, involving the activity of several signaling pathways, including JNK, Toll, JAK/STAT and the repression of Notch activity[3].(b-d) Sessile crystal cells can be detected through the cuticle of heated larvae[48,49].Control larvae (Su(H) gwt ) display around 100 crystal cells in the dorsal hematopoietic pockets of the last two posterior segments (b).The phospho-mimetic allele Su(H) S269D barely develops any crystal cells (c), whereas the phospho-mutant Su(H) S269A displays increased numbers (d)[36].Representative images are shown.Scale bar, 250 µm.

Figure 2 .
Figure 2. Classification of tested kinases.(a) A total of 184 Drosophila Ser/Thr kinases are gr the seven indicated classes.(b) Classification of the 22 kinases tested by overexpressio slight overrepresentation of other protein kinases (OPK).(c) Classification of the 46 kinases in loss of function assays.Note the overrepresentation of STE kinases.Abbreviations: AGC dependent, cGMP-dependent and protein kinase C; CAMK, Ca/calmodulin-dependent p nase; CKI, Casein Kinase 1 family; CMGC, CDK, MAPK, GCK3 and CLK members; O protein kinases; STE, STE7,11,20 gene homologues; PKL, protein kinase like.

Figure 2 .
Figure 2. Classification of tested kinases.(a) A total of 184 Drosophila Ser/Thr kinases are grouped in the seven indicated classes.(b) Classification of the 22 kinases tested by overexpression.Note a slight overrepresentation of other protein kinases (OPK).(c) Classification of the 46 kinases analyzed in loss of function assays.Note the overrepresentation of STE kinases.Abbreviations: AGC, cAMPdependent, cGMP-dependent and protein kinase C; CAMK, Ca/calmodulin-dependent protein kinase; CKI, Casein Kinase 1 family; CMGC, CDK, MAPK, GCK3 and CLK members; OPK, other protein kinases; STE, STE7,11,20 gene homologues; PKL, protein kinase like.
Cells 2024, 13, 576 7 of 19 Cells 2024, 13, x FOR PEER REVIEW 7 of kinases into inhibitors or mediators of crystal cell formation, and/or kinases with a mor general role.

Figure 3 .
Figure3.Crystal cell formation in response to ectopic expression of Ser/Thr kinases.Expression the indicated kinases was induced ubiquitously with da-Gal4, or specifically within hemocytes u ing hml-Gal4.UAS kinase strains are in alphabetical order (see TableS1).Crystal cells in the dors hematopoietic pockets were recorded in the last two segments; every dot represents one larva (n, shown below the X-axis).UAS-lacZ served as a control.Crystal cell increase is indicated in red an decrease in blue; coloration reflects significance.Whiskers show standard deviation, center line th median and center dot the average.ANOVA for multiple comparisons according to Dunnett's te relative to the respective UAS line was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, n.s.p > 0.05 n significant).

Figure 4 .
Figure 4. Crystal cell numbers are altered in kinase mutants.(a) Crystal cell numbers registered in larvae mutant for the indicated kinase or with kinase activity blocked in hemocytes.Numbers represent crystal cells in dorsal hematopoietic pockets from the last two segments.Every dot represents one larva (n, as shown below X-axis).Crystal cell increase is indicated in red and decrease in blue; intensity reflects significance. .Whiskers, standard deviation; center line, median; center dot, average.For statistical analysis, ANOVA for multiple comparisons according to Dunnett's test relative to control was employed (*** p ≤ 0.001, * p ≤ 0.05).Su(H) gwt served as a control in case of the mutants, and the respective UAS strain in case of the dominant negative lines.(b) Representative examples for mutant larvae of the indicated genotype.Left to the Su(H) gwt control are mutants with reduced, and to the right mutants with increased crystal cell numbers in alphabetical order.

Figure 4 .
Figure 4. Crystal cell numbers are altered in kinase mutants.(a) Crystal cell numbers registered in larvae mutant for the indicated kinase or with kinase activity blocked in hemocytes.Numbers represent crystal cells in dorsal hematopoietic pockets from the last two segments.Every dot represents one larva (n, as shown below X-axis).Crystal cell increase is indicated in red and decrease in blue; intensity reflects significance. .Whiskers, standard deviation; center line, median; center dot, average.For statistical analysis, ANOVA for multiple comparisons according to Dunnett's test relative to control was employed (*** p ≤ 0.001, * p ≤ 0.05).Su(H) gwt served as a control in case of the mutants, and the respective UAS strain in case of the dominant negative lines.(b) Representative examples for mutant larvae of the indicated genotype.Left to the Su(H) gwt control are mutants with reduced, and to the right mutants with increased crystal cell numbers in alphabetical order.

Figure 5 .
Figure 5. RNAi-mediated downregulation of Ser/Thr kinases affects crystal cell homeostasis.Numbers represent crystal cells in dorsal hematopoietic pockets from the last two segments in larvae of the indicated genotype (in alphabetical order).Every dot represents one larva (n, as shown above Xaxis).Crystal cell increase is indicated in red and decrease in blue; coloration reflects significance.Whiskers, standard deviation; center line, median; center dot, average.For statistical analysis, ANOVA for multiple comparisons according to Dunnett's test relative to control was employed (*** p ≤ 0.001, ** p ≤ 0.01, * p ≤ 0.05, and n.s.p > 0.05 not significant).The respective UAS-RNAi line served as control.

Table 1 .
List of Drosophila Ser/Thr kinases assayed, and their human homologues.