Functional characterization of human Homeodomain-interacting protein kinases (HIPKs) in Drosophila melanogaster reveal both conserved functions and differential induction of HOX gene expression

Homeodomain-interacting protein kinases (Hipks) are a family of conserved proteins that are necessary for development in both invertebrate and vertebrate organisms. Vertebrates have four paralogues, Hipks 1-4. Mice lacking Hipk1 or Hipk2 are viable, however loss of both is lethal during early embryonic development, with embryos exhibiting homeotic skeletal transformations and incorrect HOX gene expression. While these results suggest Hipks have a role in regulating HOX genes, a regulatory mechanism has not been characterized, and further comparisons of the roles of Hipks in development has not progressed. One challenge with characterizing developmental regulators in vertebrates is the extensive redundancy of genes. For this reason, we used Drosophila melanogaster, which has reduced genetic redundancy, to study the functions of the four human HIPKs (hHIPKs). In D. melanogaster, zygotic loss of the single ortholog dhipk results in lethality with distinct eye and head defects. We used a dhipk mutant background to compare the ability of each hHIPK protein to rescue the phenotypes caused by the loss of dHipk. In these humanized flies, both hHIPK1 and hHIPK2 rescued lethality, while hHIPK3 and hHIPK4 only rescued minor dhipk mutant patterning phenotypes. This evidence for conserved functions of hHIPKs in D. melanogaster directed our efforts to identify and compare the developmental potential of hHIPKs by expressing them in well-defined tissue domains and monitoring changes in phenotypes. We observed unique patterns of homeotic transformations in flies expressing hHIPK1, hHIPK2, or hHIPK3 caused by ectopic induction of Hox proteins. These results were indicative of inhibited Polycomb-group complex (PcG) components, suggesting that hHIPKs play a role in regulating its activity. Furthermore, knockdown of PcG components phenocopied hHIPK and dHipk expression phenotypes. Together, this data shows that hHIPKs function in D. melanogaster, where they appear to have variable ability to inhibit PcG, which may reflect their roles in development. Author summary The redundancy of vertebrate genes often makes identifying their functions difficult, and Hipks are no exception. Individually, each of the four vertebrate Hipks are expendable for development, but together they are essential. The reason Hipks are necessary for development is unclear and comparing their developmental functions in a vertebrate model is difficult. However, the invertebrate fruit fly has a single essential dhipk gene that can be effectively removed and replaced with the individual vertebrate orthologs. We used this technique in the fruit fly to compare the developmental capacity of the four human HIPKs (hHIPKs). We found that hHIPK1 and hHIPK2 are each able to rescue the lethality caused by loss of dhipk, while hHIPK3 and hHIPK4 rescue minor patterning defects, but not lethality. We then leveraged the extensive adult phenotypes associated with genetic mutants in the fruit fly to detect altered developmental pathways when hHIPKs are mis-expressed. We found that expression of hHIPKs 1-3 or dhipk each produce phenotypes that mimic loss of function of components of the Polycomb-group complex, which are needed to regulate expression of key developmental transcription factors. We therefore propose that Hipks inhibit Polycomb components in normal development, though details of this interaction remain uncharacterized.

As a first step in characterizing hHIPKs in D. melanogaster, we tested whether 100 expression of hHIPKs individually could rescue dhipk mutant phenotypes. To do this, we 101 combined two existing dhipk mutant alleles to generate a transheterozygous (heteroallelic) 102 knockout which gives rise to a severe zygotic loss of function phenotype (Lee et al., 2009) were set up at both temperatures and the degree of adult survival was determined ( Fig 1B). In 123 cases of pupal lethality, we quantified the stage at which lethality occurred based on morphology 124 of the fly within their pupal case, as shown in Fig 1C.   Combined, these are the most obvious external phenotypes of pharate dhipk mutant flies. 146 Therefore, we assessed the ability of the UAS-hHIPKs to rescue the reduced eye size, and loss of 147 ocelli and bristles. As with the dhipk mutant lethality rescue experiments, we carried out these 148 crosses at both 18°C and 25°C (Figs 2, S3).

149
Overexpression of UAS-dhipk was able to significantly rescue each dhipk mutant 150 phenotype when raised at 25°C and rescued all but the anterior bristle loss at 18°C (  Collectively these data revealed that within the same developmental context, the four hHIPKs 161 can exert both shared and distinct effects that may reveal unique roles in development. 184 Therefore, we stained 3 rd instar larval wing imaginal discs expressing the individual hHIPKs or 185 dHipk to determine if ectopic Ubx expression was occurring. We found that expression of either 186 hHIPK1 or hHIPK2 caused induction of Ubx in the wing pouch, but not in other wing imaginal 187 disc regions where dpp-Gal4 is expressed (Fig 3D). The degree of Ubx induction was greater in 188 wing imaginal discs expressing hHIPK1 compared to those expressing hHIPK2, which matches 189 the severity of the adult wing notching phenotypes. We also stained the same wing imaginal  We have previously showed that using a different UAS-dhipk insertion strain (UAS-212 Hipk 3M ) that has higher expression levels than the attP40 strain used in this work promotes cell  Thus, we found that hHIPKs can variably induce proliferation in a tissue-dependent manner.  Table). Additionally, we found that UAS-hHIPK1 and UAS-hHIPK3 each caused ectopic sex 235 comb formation on the middle and rear legs of males ( Fig 4A, arrows, S1 Table). The dpp-Gal4 236 domain is expressed in the region that produces sex combs in the leg-imaginal discs (Fig 4B). 12

237
Because the ectopic sex comb phenotype is strongly associated with expression of the Hox 238 protein Sex combs reduced (Scr, Fig 4C), we stained the larval imaginal discs that give rise to 239 the middle legs with anti-Scr antibodies and found that those expressing UAS-hHIPK1 or UAS-240 hHIPK3 consistently showed ectopic Scr expression (Fig 4D).

UAS-dhipk and UAS-hHIPK1-3 expression phenocopies loss of Polycomb components 242
When staining for Ubx and Scr expression to determine a molecular cause for the adult 243 wing and leg phenotypes, respectively, we also stained other larval tissues using each antibody. surprising. Historically, the majority of phenotypes associated with altered PcG were obtained 266 using genetic mutants for various PcG components, so it is possible that the 267 overgrown/malformed leg phenotype that UAS-dhipk overexpression produced is due to reduced 268 PcG activity but was not observed in PcG mutants due to impaired earlier development.  the whole organism, which is key to identifying the necessity of these proteins in development. 295 We thought the fly would be an excellent model to investigate this question of developmental suggests their roles are more divergent from those of hHIPK1 and hHIPK2. This is not surprising 307 for hHIPK4, since it lacks nearly all similarities to hHIPK1-3 and dHipk outside of the kinase 308 domain, and even within the kinase domain it shares the least amount of similarity between them 309 (see Fig 1). However, hHIPK3 shares nearly as much amino acid similarity with hHIPK2 as 310 hHIPK1 does, so its inability to rescue dhipk mutant lethality may warrant further investigation 311 into the significance of the amino acid sequence discrepancies between these proteins. The  The well-established dpp-Gal4 driver was selected for its common use among D. hHIPK4 had little to no effect on the adult wing structure and did not induce Ubx expression.

339
Only wing imaginal discs expressing hHIPK1 showed a noticeable decrease in Wg staining at the   As this study investigates human proteins expressed in D. melanogaster, it was necessary 425 to clearly indicate which species of protein is specified in each experiment. Throughout this 426 paper, D. melanogaster Hipk protein is written "dHipk" while mutants or DNA are referred to as 427 dhipk, human HIPKs are written as "hHIPKs", and in cases where reference is made to proteins 428 from both species, "Hipks" is used. were performed with 24-hour egg lays, and all non-Tubby pupal cases were collected 5 days 449 after flies were expected to have eclosed. Pupal cases were scored into 5 categories: 1) "eclosed" 450 flies were counted when pupal cases were empty. 2) Flies were scored as "pharate" when the 451 adult head, thorax, and abdomen were fully developed and pigmented, but they were unable to 452 eclose. 3) "Pupal lethal 1" was assigned to pupae that had defined head, thorax, and abdomen 453 within the pupal case, but only had partial pigmentation. 4) "Pupal lethal 2" was assigned to 454 pupae that had defined head, thorax, and abdomen, but no pigmentation. 5) "Pupal lethal 3" was 455 assigned to pupae that had no defined head, thorax, or abdomen.

456
The pharate pupae and viable adults from the dhipk mutant viability rescue experiment 457 were collected, and if necessary, gently removed from their pupal cases with dissecting tweezers 458 before being immediately placed in 70% ethanol and stored at -20°C for preservation until they

523
RNA that was used to confirm reduced dHipk mRNA in dhipk mutant and rescue crosses, as well 524 as verify the correct hHIPK expression in the rescue crosses, was collected from four combined 525 wandering 3 rd instar larvae (two male and two female) for each cross. Larvae were washed in 526 PBS before being spot dried on a clean paper towel and transferred to 300µl buffer RLT Plus, 527 supplemented with freshly added β-mercaptoethanol to 1%. Larvae were homogenized with 528 pestles by hand in 1.6mL tubes before being centrifuged for 3 minutes at maximum speed to 529 pellet debris. Supernatant was transferred to a gDNA Eliminator spin column, with the remaining 530 RNA extraction steps following the manufacturer's instructions.