In Vivo Conditions to Identify Prkci Phosphorylation Targets Using the Analog-Sensitive Kinase Method in Zebrafish

Protein kinase C iota is required for various cell biological processes including epithelial tissue polarity and organ morphogenesis. To gain mechanistic insight into different roles of this kinase, it is essential to identify specific substrate proteins in their cellular context. The analog-sensitive kinase method provides a powerful tool for the identification of kinase substrates under in vivo conditions. However, it has remained a major challenge to establish screens based on this method in multicellular model organisms. Here, we report the methodology for in vivo conditions using the analog-sensitive kinase method in a genetically-tractable vertebrate model organism, the zebrafish. With this approach, kinase substrates can uniquely be labeled in the developing zebrafish embryo using bulky ATPγS analogs which results in the thiophosphorylation of substrates. The labeling of kinase substrates with a thiophosphoester epitope differs from phosphoesters that are generated by all other kinases and allows for an enrichment of thiophosphopeptides by immunoaffinity purification. This study provides the foundation for using the analog-sensitive kinase method in the context of complex vertebrate development, physiology, or disease.


Introduction
Phosphorylation is a protein modification that is essential for almost all aspects of cell biology. Kinases that catalyze this posttranslational modification are an abundant group of enzymes with promiscuous substrate specificity and a common requirement for ATP. For this reason, the identification of specific substrate proteins of particular kinases has remained a tedious challenge and traditionally involves in vitro phosphorylation assays with candidate substrates. However, in vitro assays often generate false-positive results and are inferior to in vivo screening conditions in which the kinase of interest localizes within the correct subcellular compartment and is associated with endogenous binding partners that modulate its activity and affinity towards substrate proteins. The analog-sensitive kinase method utilizes in vivo conditions for substrate identification but has never been employed in a multicellular model organism [1].
The atypical protein kinase C (aPKC) family consists of serine/ threonine kinases with essential cellular functions in cell polarity and organ morphogenesis, cell migration, apoptosis and proliferation (reviewed in [2][3][4][5]). Increasing evidence also points at an involvement of aPKCs in the promotion of carcinogenesis in vitro and in vivo [reviewed in [6]]. Among other proteins, aPKCs are core components of the apical Partition defective 6 (Pard6)-aPKC protein complex which is composed of several PDZ domain containing proteins and is required for the establishment of epithelial apicobasal polarity in many systems (reviewed in [4,7]). The unique N-terminal regulatory domain of aPKCs which contains a Phox Bem1 (PB1) domain mediates direct interactions with the polarity protein Pard6 which in turn modulates aPKC activity, or with the small GTPases Rac1 and Cdc42 [8]. The zebrafish heart and soul locus encodes Prkci, one of two aPKCs expressed in this organism [9,10]. Consistent with a function in apicobasal cell polarity, zebrafish mutants lacking Prkci show defective formation and maintenance of several embryonic epithelia and abnormal heart morphogenesis [9,11,12]. Prkci function in cellular polarity and organ morphogenesis requires its catalytic activity [12].
To date, only a small number of substrate proteins of Prkci in the context of cellular polarity, proliferation control or apoptosis have been identified in any organism, and the molecular mechanisms of their interaction with aPKCs have been convincingly demonstrated for only a few of them [13][14][15][16][17][18][19]. In addition to its auto-phosphorylation [3], potentially relevant substrate proteins of aPKC activity in the context of cellular polarity are Par3 [17,20,21], Numb [22,23], Miranda [24], Frizzled 1 [25], Partner of inscuteable [26], and GSK-3b [8]. To further elucidate the mechanistic relevance of aPKCs during development and in various cell biological processes it is necessary to identify phosphorylation targets in an unbiased manner and under in vivo conditions.
We have established conditions for a chemical genetics screen using the analog-sensitive kinase method to identify phosphorylation targets of Prkci during zebrafish development. This methodology opens the way for the identification and functional characterization of specific substrates in their normal subcellular context which is essential for a mechanistic understanding of how Prkci affects divergent cellular processes.

Design of an Analog-sensitive Prkci
We established a screening approach with the aim to identify Prkci phosphorylation targets in a multicellular model organism. The straightforward screening method using analog-sensitive kinases utilizes an environment that is similar to the in vivo state of the kinase. In principle, the method is based on engineering a mutant kinase that accepts bulky ATP or ATPcS analogs such as N 6 -benzyl ATP or N 6 -benzyl ATPcS. The ATP analog-sensitive kinase is generated by replacing a large hydrophobic group within the ATP binding pocket (the ''gatekeeper'' residue) with a smaller residue thereby enlarging the binding pocket which allows the kinase to accept bulkier ATP analogs (Fig. 1A). The specificity of this approach has been demonstrated for various kinases, e.g. by utilizing a modified v-Src, c-Raf-1, or AMPKa2 for the identification of novel substrates [1,27,28].
To identify an ATP binding site ''gatekeeper'' residue within the Prkci ATP binding pocket, we selected suitable residues for sitedirected mutagenesis based on the high evolutionary conservation of the ATP binding site in different protein kinases, including v-Src [1] and c-Raf-1 [28] that had successfully been engineered to utilize N 6 -benzyl ATP. The comparison of the primary sequence of Prkci, c-Src, and c-Raf-1 kinase domains suggested Val 300 and Ile 316 of Prkci to be the most likely ''gatekeeper'' residues and therefore to be the most suitable targets for site-directed mutagenesis (Fig. 1B).

Mutant Prkci I316A has Normal in Vivo Biological Activity
One stringent requirement for an analog-sensitive kinase is the conservation of its biological properties which includes normal in vivo functionality and substrate specificity. We therefore tested whether the Prkci V300A or Prkci I316A mutants with altered ATP binding pockets had normal biological activity. Loss of Prkci has well-characterized epithelial and organ morphogenesis defects that involve cardiac malformations and a defective neuroepithelium [9,11,12]. In functional rescue experiments, we co-injected at the one-cell stage mRNA encoding either Prkci WT or one of the two mutant Prkci proteins together with an antisense oligonucleotide morpholino (MO) for knockdown of endogenous Prkci. Whereas expression of Prkci I316A allowed normal cardiac development (63% of prkci morphant embryos rescued, n = 183) which was almost as efficient as Prkci WT expression (83% of prkci morphants rescued, n = 166), the Prkci V300A mutant was not biologically active based on the appearance of prkci morphant cardiac phenotypes among all injected embryos ( Fig. 2A) [9].
Because the Prkci I316A mutant was biologically fully functional within the heart, we anticipated that this mutant kinase should correctly localize within myocardial cells. Indeed, expression by injection at the one-cell stage of mRNA encoding Myc-tagged Prkci I316A resulted in the correct localization of the mutant protein to the cell membrane, similar to Myc-tagged Prkci WT or endogenous Prkci (Fig. 2B). That overexpression of Prkci WT or Prkci I316A did not cause any ectopic phenotypes strongly suggests substrate specificity (see below). Together, our results indicated that Prkci I316A is biologically functional and predominantly localizes to the correct subcellular compartment at the cell membrane, which makes this mutant kinase a strong candidate for a chemical genetic screen.

Mutant Prkci I316A can use Bulky ATP Analogs
We next tested whether Prkci I316A could utilize bulky ATP analogs by assaying the phosphorylation efficiency and specificity of several alkylated ATPcS analogs in in vitro kinase assays. ATPcS was utilized instead of regular ATP to ensure a transfer of a phosphorothioate moiety to the phosphoacceptor hydroxyl groups of respective substrates. The substitution of sulfur in place of oxygen generates unique thiophosphorylated epitopes that, when alkylated with p-nitrobenzyl mesylate (PNBM), generate thiophosphoester epitopes which can be recognized by a specific monoclonal antibody [29,30]. Using the baculovirus system in Sf9 insect cells, we produced the recombinant kinases and first confirmed that Prkci WT and Prkci I316A could utilize ATPcS to phosphorylate Myelin Basic Protein (MBP) as substrate and that this modification could efficiently be detected with the antithiophosphoester antibody (Fig. 3). However, whereas Prkci WT accepted ATPcS, it could not utilize any of the tested N 6 -alkylated ATPcS analogs (N 6 -benzyl-, N 6 -phenethyl-, or N 6 -cyclopentyl-ATPcS) [30] as assessed on Western blot upon the in vitro kinase assay. In contrast, Prkci I316A most efficiently utilized N 6 -benzyl ATPcS (Fig. 3) whereas the other two bulky ATPcS analogs were apparently not efficiently utilized (data not shown). Taken together, Prkci I316A had a normal biological function in the in vivo context and exerted catalytic activity using a bulky N 6 -benzyl-ATP analog. Hence, Prkci I316A fulfilled the basic requirements required to identify Prkci phosphorylation targets. Evidence for Prkci I316A -mediated Thiophosphorylation in the Zebrafish Embryo Two principal methods have been used for the enrichment of thiophosphorylated proteins in in vitro phosphorylation assays and in cell culture systems but not yet in any model organism. The ''covalent capture'' approach is based on the enrichment of thiophosphate-tagged substrates with iodoacetyl-agarose and subsequent analysis by mass spectrometry [31]. An alternative approach is based on ''immunoaffinity purification'' enrichment using the anti-thiophosphoester antibody, followed by mass spectrometry of enriched peptides [27,30]. Both methods require the utilization of ATPcS analogs for the unique labelling of substrate proteins.
One preeminent challenge of using ATPcS analogs in in vivo approaches is the potential toxicity of thiophosphates since such protein modifications cannot be removed by phosphatases. We first tested the toxicity of N 6 -benzyl ATPcS analog by injecting different concentrations (,1 nL injection volume) into wild-type embryos and found that a concentration of 200 mM of N 6 -benzyl ATPcS was the maximal concentration that could be injected into one-cell stage embryos without affecting development (200 mM: 93.3% of embryos developed normally, n = 45; 250 mM: 47.8% of embryos developed normally, n = 23; 500 mM: 37.2% of embryos developed normally, n = 43). We next tested the feasibility of applying this approach in the zebrafish embryo by optimizing the conditions for introducing N 6 -benzyl ATPcS together with mRNA encoding Prkci I316A by microinjection at the one-cell stage (Fig. 4A). With the concentration of 200 mM N 6 -benzyl ATPcS ( , 1 nL injection volume), we observed that most embryos survived up to 32 hours post fertilization (hpf) (97.2% of Prkci I316A  expressing embryos developed normally, n = 320), which was comparable with embryos co-injected with N 6 -benzyl ATPcS and mRNA encoding Prkci WT (96.5% of embryos developed normally, n = 261), or with non-injected control embryos (96.7% of embryos developed normally, n = 390). Therefore, N 6 -benzyl ATPcS in the range of physiological ATP levels is compatible with normal zebrafish embryogenesis.
To assess whether thiophosphorylation had occurred in the developing zebrafish embryo, such in vivo thiophosphorylated 8 hpf extracts were alkylated using PNBM, resolved by SDS-PAGE, and Western blots probed with the anti-thiophosphoester antibody. This analysis revealed that thiophosphorylation of substrates had indeed occurred in the zebrafish embryo and that Prkci I316A had catalyzed the selective labelling of at least one putative substrate (Fig. 4B).

Discussion
This study outlines the methodology required for an in vivo screening approach using an analog-sensitive kinase in a multicellular model organism. Our work demonstrates the conditions for in vivo thiophosphoester labeling of substrates using an analogsensitive kinase and bulky N 6 -benzyl ATPcS analogs. Injection of physiological levels of bulky N 6 -benzyl ATPcS does not interfere with zebrafish development even though thiophosphorylations are largely irreversible. Viability of zebrafish embryos under such conditions indicates that only a fraction of substrate proteins is modified by the analog-sensitive kinase and such in vivo thiopho-sphorylations are detectable on Western blots. That thiophosphorylations are also detected in Prkci WT samples is due to unspecific utilization of N 6 -benzyl ATPcS by other enzymes [31] and to the partial degradation of N 6 -benzyl ATPcS to ATPcS which can be utilized by other kinases. These contaminations with unspecific thiophosphoylations highlight the need to immunopurify embryonic extracts and to perform comparative mass spectrometric analyses for substrate identification.
The method for the enrichment and identification of novel putative kinase targets by immunoaffinity purification with an antithiophosphoester antibody is well-established [27,30]. Combining this methodology with the working protocol outlined in our study will soon provide unprecedented insight into kinase signaling in multi-tissue encompassing developmental processes, the regulation of different physiological conditions, or in disease processes involving aberrant kinase signaling. Taken together, our work provides the ground for similar approaches using the analogsensitive kinase method in this and other multicellular model organisms.

RNA and Antisense Oligonucleotide Morpholino Injections
Constructs were transcribed using the SP6 mMessage mMachine kit (Ambion). Tg[cmlc2:GFP] twu34 embryos were injected with 2.5 ng of prkci MO [12]. For rescue experiments 100 pg of mRNAs were injected. The heart morphology was assessed at 24hpf. Data presented are the means of at least 2 independent experiments. For in vivo labeling, 200 pg of mRNA encoding HisMyc-Prkci WT or HisMyc-Prkci I316A mRNA were injected at the one-cell stage.

DNA Constructs and Site-directed Mutagenesis
Both constructs encoding wt and mutant forms of Prkci were produced by PCR amplification from a full length cDNA template, pCS2+ HisMyc::prkci [12]. Site directed mutagenesis was performed using the QuickChange TM XL Site-Directed mutagenesis kit (Stratagene). Primer sequences are available upon request.

Protein Extraction from Zebrafish Embryos
Zebrafish embryos protein extraction was performed as previously described [35]. Embryos were dechorionated with pronase solution in E2 medium in Petri dishes coated with 1% agarose. After washes with E2 medium, embryos were transferred to 1.5 ml tubes. The yolk was disrupted by pipetting with a 1 ml pipet and vortexing for 30 seconds at 1100 rpm in deyolking buffer. Embryos were pelleted several times after washes in washing buffer at 3000 rpm for 30 seconds. Pelleted embryos were homogenized in the appropriated lysis buffer depending on the following experiment.

Recombinant Protein Expression in Insect Cells
To express recombinant Prcki WT or Prkci I316A in Sf9 insect cells (Sigma), Bac-to-BacH Baculovirus Expression System (Invitrogen) was used according to the manufacturers' protocol. HisMyc::prkci WT and HisMyc::prkci I316 cDNAs were cloned into pFastBac TM 1. Primer sequences are available upon request.

In Vivo Thiophosphorylation in Zebrafish Embryos
For in vivo thiophosphorylation 200 pg of mRNA encoding HisMyc-Prkci WT or HisMyc-Prkci I316A were injected into the yolk at the one-cell stage together with approximately 1 nL of 200 mM N 6 -benzyl ATPcS. Embryonic protein extracts were prepared at 8 hpf, and pelleted embryos were homogenized in RIPA buffer. Alkylation was perfomed for 2 hours at RT with 2.5 mM PNBM [30]. Samples were analyzed by Western blot using anti-thiophosphoester rabbit monoclonal C51-8 antibody (1:5000 Epitomics).