Proteomic approaches for protein kinase substrate identification in Apicomplexa

Abstract

Apicomplexa is a phylum of protist parasites, notable for causing life-threatening diseases including malaria, toxoplasmosis, cryptosporidiosis, and babesiosis. Apicomplexan pathogenesis is generally a function of lytic replication, dissemination, persistence, host cell modification, and immune subversion. Decades of research have revealed essential roles for apicomplexan protein kinases in establishing infections and promoting pathogenesis. Protein kinases modify their substrates by phosphorylating serine, threonine, or tyrosine residues, resulting in rapid functional changes in the target protein. Post-translational modification by phosphorylation can activate or inhibit a substrate, alter its localization, or promote interactions with other proteins or ligands. Deciphering direct kinase substrates is crucial to understand mechanisms of kinase signaling, yet can be challenging due to the transient nature of kinase phosphorylation and potential for downstream indirect phosphorylation events. However, with recent advances in proteomic approaches, our understanding of kinase function in Apicomplexa has improved dramatically. Here, we discuss methods that have been used to identify kinase substrates in apicomplexan parasites, classifying them into three main categories: i) kinase interactome, ii) indirect phosphoproteomics and iii) direct labeling. We briefly discuss each approach, including their advantages and limitations, and highlight representative examples from the Apicomplexa literature. Finally, we conclude each main category by introducing prospective approaches from other fields that would benefit kinase substrate identification in Apicomplexa.

1. Introduction

Phosphorylation is one of the most common and well-studied post-translational modifications (PTM) of proteins, influencing every aspect of cellular growth, metabolism, and homeostasis [1]. Protein kinases act by catalyzing the transfer of a gamma phosphate group from adenosine triphosphate (ATP) onto a residue of a protein substrate (usually serine, threonine, or tyrosine) [2, 3]. Other molecules, such as lipids, carbohydrates, and nucleotides can also be phosphorylated by specialized kinases [4, 5], but our focus here is to discuss protein kinase phosphorylation of protein substrates. The apicomplexan kinome (i.e. the complete set of kinases) is considered one of the smallest among eukaryotes [6]. A kinome search of the current catalog of 128 sequenced apicomplexan genomes (VEuPathDB Release 67; PFAM: PF00069 or PF14531) revealed that apicomplexan kinomes range from 40 genes (Theileria orientalis) to 181 genes (Besnoitia besnoiti). These kinomes include both putative active kinases and putative inactive kinases (pseudokinases). For example, the Toxoplasma gondii (T. gondii) kinome consists of 108 kinases and 51 pseudokinases [7]. Parasite kinases have been shown to be essential for a variety of processes including such as lytic growth, development, egress, motility, host modulation, and pathogenesis, as nicely reviewed previously [6, 8-11]. While several kinase clades are conserved between apicomplexans and humans, apicomplexans also possess many kinase clades that are only partially conserved or absent in humans [12], making them especially attractive targets for therapeutic development.

Recent advances in proteomic and phosphoproteomic fields have contributed to a better understanding of the diverse roles kinases fulfill in apicomplexan parasites [13, 14]. To date, kinase substrate identification in Apicomplexa can be categorized by three general proteomic approaches: i) kinase interactome, ii) indirect phosphoproteomics and iii) direct labeling. These approaches can be combined to identify kinase protein substrates and map specific phosphorylation motifs. Here, we organized our discussion around these three proteomic approaches for kinases substrate identification in Apicomplexa (Figure 1), highlighting their main advantages and limitations (Table 1). We briefly discuss each method and highlight studies that have implemented the method for kinase substrate discovery in Apicomplexa (Figure 2). Finally, we provide an outlook of complementary or alternative approaches that have been used to successfully identify kinase substrates in other organisms that could be adapted for apicomplexan research.

Figure 1 –

Proteomic approaches for apicomplexan kinase substrate identification Approaches used for kinase substrate identification in Apicomplexa can be classified in three general categories: i) Kinase interactome, ii) Indirect phosphoproteomics, and iii) Direct labeling.

Table 1.

Advantages and limitations of proteomic approaches used to identify kinase substrates in Apicomplexa.

	Approach	Advantages	Limitations
Kinase interactome	BioID	Allows full protein solubilization; Provides a snapshot of protein complexes in live cells; Transient interactions are more likely to be labeled	Slow tagging kinetics; Reduced Activity below 37°C (BioID2 has improved efficiency and temperature range); False positive non-interacting neighboring proteins
	TurboID	Fast labeling kinetics; Active between 20-37°C; Transient interactions are more likely to be labeled	Higher background labeling than BioID; Less control of labeling window; False positive non-interacting neighboring proteins
	co-IP	Interacting proteins are captured in the native conformation; More likely to find direct interactors	False positive interactions created post lysis; False negatives created by low affinity or transient interactions
	XL-co-IP	Stabilizes weak or transient interactions	False negatives from solubility issues or peptic digest issues
Quantitative phosphopro	Label-free	Simple and fast workflow; Lower cost when compared to label-based approaches	Samples cannot be combined and analyzed simultaneously; Increased post-acquisition data analysis time; semi-quantitative
	SILAC	Limited sample multiplexing; Quantitative	Unintended interconversion of amino acids; Some cell lines may not be stable in SILAC media
	iTRAQ	Allows sample multiplexing; Labelling at peptide level allows flexibility in experiment design; Quantitative	Reduced peptide identification efficiency when compared to label-free
	TMT	Allows sample multiplexing; Labelling at peptide level allows flexibility in experiment design; High precision in phosphopeptide quantification	Reduced peptide identification efficiency when compared to label-free
Direct labeling	In vitro kinase assay	Gold standard for kinase substrate validation	Use of radioactive ATP; Low throughput
	AS kinase	Unbiased labeling of direct targets; Thiophosphate group is resistant to dephosphorylation by phosphatases	Not all kinases can be mutated to use ATPγS
	Peptide library and Microarray	Straightforward screening of kinase substrates; Identifies motif preference of the target kinase	Complex folded protein motifs are not represented by linear synthetic peptides

Figure 2 –

Overview of kinase substrate identification in Apicomplexa

Representative kinase substrate discovery in Apicomplexa organized by category and approach. The approaches can be classified in three general categories: i) Kinase interactome (BioID, TurboID, PerTurboID, co-IP and XL-co-IP) highlighted in red; ii) Indirect phosphoproteomics (Label-free, SILAC, iTRAQ and TMT) highlighted in blue; iii) Direct labeling (in vitro kinase assay, AS-kinase and Peptide library and microarray) highlighted in green. The kinase of interest and putative interactors/substrates identified in each study is addressed.

2. Kinase interactome

Protein kinases physically interact with proteins or protein complexes that they phosphorylate since close contact is required for the phosphate transfer. Perhaps the most straightforward approach to identifying potential kinase substrates is to determine which proteins kinases specifically interact with (i.e. kinase interactome). Two general proteomic approaches have been used to define apicomplexan kinase interactomes: proximity labeling and affinity purification [15]. Each has their own advantages and limitations but can also be combined to enhance confidence in the enriched hits.

2.1. Proximity labeling

2.1.1. BioID

Proximity-dependent biotinylation using promiscuous biotin ligases is an interactome profiling tool applied in live cells for mapping protein-protein interactions [16]. Protein complexes have pivotal roles in cellular functions such as signal transduction and cellular growth [17]. Therefore, mapping protein complexes provides a better understanding of the dynamic aspects of cellular functions. Proximity-dependent biotin identification (BioID) is an approach for interactome profiling that uses a highly promiscuous biotin ligase (BirA*) fused to a protein of interest (POI; bait) to biotinylate proteins in close proximity (prey) [18]. It is important to include negative controls to establish the baseline biotinylated proteome, such as the parental cell line (no BirA*) treated with biotin. Additional controls such as fusion of BirA* to an irrelevant target such as GFP (non-specific control), or fusion of BirA* to a protein in the same compartment as the protein of interest (spatial control), are useful for data analysis to identify proteins specifically biotinylated and enriched by the POI-BirA* fusion. Following several hours of biotin treatment (usually overnight), biotinylated proteins are pulled-down using streptavidin- or neutravidin-coated beads, eluted, proteolytically digested into peptides, and identified by liquid chromatography tandem mass spectrometry (LC-MS/MS) [18]. There are several advantages to BioID. First, stable and transient kinase interactors can potentially get labelled for capture. Second, proteins can be fully solubilized after lysis with no need to maintain protein complexes. Third, BioID provides a snapshot of protein complexes in living cells, making it less susceptible to false-positive or false-negative results that may occur post-lysis under native conditions used with co-immunoprecipitation (co-IP). Limitations of this approach include the relatively slow labeling kinetics of BirA*, caused by its low catalytic activity, potential for false positive interactions caused by nearby bystander proteins getting labeled, and inherent exclusion of endogenously-biotinylated proteins [18]. A smaller version of BioID called BioID2 has an increased temperature range and increased affinity for biotin so that less is needed [19]. BioID2 has been used to map proteins of the apical annuli of T. gondii [20], suggesting that the approach could be adapted for apicomplexan kinase interactome analysis as well.

In Apicomplexa, BioID has been used to identify putative TgCDPK3 substrates in T. gondii [21]. TgCDPK3 is a calcium-dependent protein kinase (CDPK) that controls microneme secretion and host cell egress [22-25]. A total of fourteen proteins were exclusively captured and detected from biotin-treated TgCDPK3-BirA* parasites compared to the negative control, including the essential glideosome protein TgMyoA. TgCDPK3 phosphorylation of a TgMyoA peptide array was confirmed using an in vitro kinase assay [21]. Mutational analysis of TgMyoA demonstrated that phosphorylation of serine S21 and S743 residues by TgCDPK3 is important for parasite motility and egress [21]. In a separate phosphoproteomics study, TgMyoA was less phosphorylated at serine S20 and S21 residues in TgCDPK3 mutant parasites when compared to wild-type (WT) control [26], further indicating that TgMyoA is regulated by TgCDPK3 in vivo (discussed further in section 3.2 SILAC). Additional roles for TgCDPK3 have also been identified, indicating that it may regulate multiple substrates. Following egress, TgCDPK3 is able to sense extracellular stress to promote extracellular survival by phosphorylating TgeEF2 to repress translation [27]. In this study, TgCDPK3 was shown to interact with TgeEF2 in the cytosol under stress conditions using co-IP with immunoblotting and also immunofluorescence microscopy [27]. Therefore, it is important to consider that each parasite kinase may regulate multiple protein substrates in distinct signaling pathways.

BioID was also used to identify putative interactors and potential substrates of TgERK7 in T. gondii [28]. TgERK7 is a mitogen activated protein kinase (MAPK) [29] that localizes to the parasite apical cap and is required for maturation of the apical complex [30]. The TgERK7 BioID approach identified a number of apical cap proteins including TgAC9 and TgAC10 [28], which are known to recruit TgERK7 to the apical cap [31, 32]. To validate these hits and further explore direct TgERK7 protein interactions, a yeast-two-hybrid system was implemented using TgERK7 as bait to capture protein products coexpressed in yeast from a T. gondii cDNA library [28]. A total of 21 candidate interactors of TgERK7 were identified by yeast-two-hybrid screening, 8 of which were also identified through BioID, including TgAC9 and TgAC10, suggesting a high confidence overlap of candidates in both methods [28]. Interestingly, yeast-two-hybrid screening also identified TgCSAR1, an E3 ubiquitin ligase, which was investigated as a downstream substrate of TgERK7. TgCSAR1 was found to be essential for functional maturation of the apical complex and regulated by TgERK7 [28].

In P. falciparum, BioID was used to gain insights into how PfCRK4, a cyclin-dependent kinase related kinase (CRK), regulates S-phase in blood stage parasites [33]. Cyclin-dependent kinases (CDKs) are activated by cyclins to precisely control cell cycle progression in eukaryotes [34]. Until cyclin-dependent activation has been established, CDKs are typically referred to as CDK-related kinases (CRK)s. Previous studies demonstrated that loss of PfCRK4 suppresses DNA synthesis [35, 36], likely by phosphorylating effector proteins involved in origin of replication firing [36]. Using BioID, PfCRK4 was shown to potentially interact with 44 proteins, including four proteins known to be hypophosphorylated (PF3D7_0624600, PF3D7_0209800, PF3D7_1357400, and PF3D7_0303500; PfSPB) in the absence of PfCRK4 [33]. Additional experiments revealed that PfCRK4 regulates both microtubule rearrangement and DNA replication during schizogony [33]. Interestingly, PfCRK4 clades closely with TgCrk6 and TgCrk4 by phylogenetic analyses, which also regulate replication in T. gondii [37] (discussed further in section 3.1 Label-free). In summary, BioID is useful for defining kinase substrate interactions in apicomplexan parasites. However, due to lengthy labeling times, it is suboptimal for characterizing rapid kinase substrate interactions.

2.1.2. TurboID

To enhance the efficiency and speed of BioID, Branon and colleagues developed TurboID by mutating 15 amino acids of E.coli BirA [38]. TurboID fusions permit kinetic interactome analyses at minute to sub-minute resolution. TurboID is also active at broad temperature range (20-37°C), allowing its use in a variety of organisms [38]. Since TurboID is more efficient than BioID, it is important to minimize available biotin to reduce background labeling.

TurboID has been used in various fields for identification of kinase substrates, including apicomplexan parasites. Using TurboID, Davies et al. [39] first defined the dynamic interactome of FIKK4.1 at two timepoints in the P. falciparum asexual blood stage. FIKK4.1 is a kinase belonging to the FIKK family and is involved in the modulation of the infected red blood cell (iRBC) rigidity and efficient translocation of PfEMP1 to the iRBC surface [40]. PfEMP1 is a key virulence factor in P. falciparum responsible for the cytoadhesion of the iRBC to the host vascular endothelium system [41]. Using FIKK4.1-TurboID, 101 parasite proteins were enriched over the WT controls [39]. The majority of these proteins are predicted to be exported and include likely direct targets of FIKK4.1 that regulate PfEMP1 presentation on the iRBC surface such as PfEMP1-trafficking protein 4 (PTP4) and knob-associated histidine-rich protein (KAHRP) [40, 42, 43]. To evaluate the local protein environment and interactors of PTP4 and KAHRP, and how it changes after FIKK4.1 deletion, PTP4-TurboID and KAHRP-TurboID fusions were generated in FIKK4.1 conditional knockout (KO) parasites, respectively. This approach, named PerTurboID, revealed interactors of PTP4 and KAHRP, based on the enrichment and detection biotinylated peptides, that are modulated by the presence of FIKK4.1 [39]. These examples demonstrate the power and adaptability of proximity labeling for revealing kinase substrate interactions in Apicomplexa.

2.2. Affinity purification

2.2.1. Co-immunoprecipitation

Immunoprecipitation (IP) and co-immunoprecipitation (co-IP) utilize antibodies for affinity purification of target proteins and protein complexes from complex mixtures, respectively. Co-IP coupled with peptide identification by LC-MS/MS is a common and straightforward approach to define protein interactomes [44]. In a typical co-IP experiment, native cell lysate is incubated with beads coated with a target-specific antibody to capture a target protein complex, unbound material is removed by washing, then captured proteins are trypsin digested for peptide identification by LC-MS/MS [44]. In comparison to proximity labeling, co-IP requires that protein interactors stably interact with the bait protein of interest throughout lysis and washing steps, increasing the chances of finding direct interactors [44, 45]. The main limitations of this approach stem from false positive interactions that can occur post lysis or from off target antibody capture, and false negative results from low affinity or transient protein-protein interactions, or weak antibody interactions, that can be lost during washing steps [46]. Thus, it is important to include negative controls such as antibody isotype control beads or cells lacking the target of target-specific antibody beads to establish a baseline for non-specific binding.

Co-IP with LC-MS/MS has been used to define the interactomes of several kinases in Apicomplexa. For example, this approach has been used to define signaling pathways that regulate Plasmodium blood stages. P. falciparum casein kinase 2 (PfCK2), consisting of two regulatory subunits (β1 and β2) and a catalytic subunit (α), is required for asexual proliferation and sexual development of blood-stage parasites [47-49]. PfCK2 is thought to function by regulating several other kinases [50]. Interactome analysis by co-IP with LC-MS/MS of PfCK2β1 and PfCK2β2 regulatory subunits revealed a large number of components from parasite chromatin assembly pathway, suggesting the involvement of PfCK2 in chromatin dynamics. In support, recombinant PfCK2α was capable of phosphorylating nucleosome assembly proteins, histones, and two members of the Alba family proteins using in vitro kinase assays [48], which is a common direct-labeling approach.

In T. gondii, co-IP with LC-MS/MS analysis was used to identify putative substrates of TgARK1. Aurora kinases (ARK) are a conserved family of serine/threonine kinases that regulate chromosomal segregation and cytoskeletal remodeling during cell division [51]. TgARK1 regulates daughter budding during endodyogeny by organizing the mitotic spindle as part of the chromosomal passenger complex (CPC) [52]. The interactome of TgARK1 suggested that this kinase may phosphorylate cytoskeleton-associated proteins, nuclear proteins, and other known cell cycle regulators [52]. Interestingly, the interactomes of WT and kinase dead TgARK1 were similar but with some unique hits to each. The use of enzyme dead mutants for co-IP is known as substrate trapping, since dead enzymes tend to interact with their substrates longer than usual [52].

Kinase interactome analysis can also be used to define kinase complexes or upstream kinase regulators. For instance, co-IP was used to define the pantothenate kinase (PanK) complex in P. falciparum and T. gondii [53]. PanK catalyzes the first step in the conversion of pantothenate to coenzyme A [54], an essential cofactor in most organisms [55]. Co-IP with LC-MS/MS analysis revealed that PfPanK1 and PfPanK2 associate and form a heteromeric pantothenate kinase complex with Pf14-3-3I in the intraerythrocytic stage of P. falciparum [53]. In T. gondii, TgPank1 and TgPank2 also associate to form a PanK complex to phosphorylate pantothenate, but Tg14-3-3I was not detected by TgPank1 or TgPank2 co-IP by immunoblotting [53]. Unlike other eukaryotic PanK complexes that are homodimeric, this work suggested that apicomplexans encode multiple PanK genes to generate heteromeric PanK complexes. A similar strategy was used define the AMP Kinase (AMPK) complex in T. gondii [56]. AMPK is a serine/threonine kinase composed of one catalytic subunit (AMPKα) and two regulatory subunits (AMPKβ and AMPKγ), that regulates cellular metabolism by monitoring the levels of AMP and ATP in the cell [57]. AMPKβ acts as the structural core of the AMPK complex by binding both AMPKα and AMPKγ. AMPKγ senses AMP to phosphorylate and activate AMPKα, which phosphorylates a number of different protein substrates important for cellular metabolism [58]. T. gondii encodes orthologs of AMPK, including one putative catalytic subunit (TGGT1_233905, TgAMPKα), and two putative regulatory subunits (TGGT1_268960, TgAMPKβ; TGGT1_239870, TgAMPKγ) [56]. To define the AMPK composition and interactions in T. gondii, putative TgAMPK subunits were tagged with Ty or HA, then captured by co-IP and analyzed by LC-MS/MS. When using TgAMPKα, TgAMPKβ, or TgAMPKγ as bait, each identified other AMPK subunits as top hits, strongly suggesting that these three proteins form the TgAMPK complex [56]. Potential regulators or substrates were also found to associate with the TgAMPK complex, including the kinases TgCDPK6, TgCDPK2a, TgPKG, TgCDPK1 and TgCDPK3 [56]. Follow-up experiments demonstrated that TgAMPKγ depletion drastically reduces TgAMPKα phosphorylation, which is expected to inactivate TgAMPKα and prevent downstream signaling. To test this possibility, the TgAMPKγ-dependent phosphoproteome was defined using indirect phosphoproteomics [56]. In total, 170 proteins were differently phosphorylated upon TgAMPKγ depletion, many of which that are known or expected to regulate cellular metabolism T. gondii [56]. Thus, these studies nicely represent the use of multiple approaches to identify mechanisms of kinase function in Apicomplexa.

Kinase interactome analyses using co-IP have also been performed in Eimeria tenella, the major causative agent of avian coccidiosis in chickens. Co-IP of EtCDPK4, a kinase important for invasion of host cells [59], and subsequent LC-MS/MS analysis identified 91 putative interactors of EtCDPK4 not found in the negative control [60]. To narrow the list of potential EtCDPK4 substrates, recombinant HIS-tagged EtCDPK4 was immobilized to Co²⁺ resin, then incubated with sporulated oocyst proteins to pull down interacting proteins. A total of 8 proteins pulled down with rHIS-EtCDPK4 as detected by LC-MS/MS, with EteIF-5A also found by co-IP. EteIF-5A and EtCDPK4 were confirmed to interact by bimolecular fluorescence complementation and GST pulldown. EteIF-5A colocalizes with EtCDPK4 in the cytoplasm of sporozoites and trophozoites, suggesting that both proteins may work together for invasion [60]. Studies from the same group also determined that EtCDPK4 interacts with a serine protease inhibitor (EtSerpin) [61] and Et14-3-3 [62]. These interactions may contribute to EtCDPK4’s role in host cell invasion but has not yet been demonstrated. In a separate study, co-IP with LC-MS/MS protein analysis revealed putative interactors of EtROP1 kinase in E. tenella [63]. Among the 29 putative interactors of EtROP1 was the host protein p53. An in vitro kinase assay confirmed that EtROP1 kinase phosphorylates host p53 at threonine T387 [63]. Additional experiments showed that phosphorylation of host p53 by EtROP1 causes inhibition of host cell apoptosis during intracellular development of E. tenella and G0/G1 host cell cycle arrest [63]. These studies highlight the utility of co-IP to identify stable interactors of target kinases but may miss transient kinase interactions in the absence of protein crosslinkers.

2.2.2. Crosslinking co-immunoprecipitation

Due to the transient nature of kinase substrate interactions, it can be beneficial to crosslink proteins in whole cells prior to lysis for co-IP interactome analysis. Chemical crosslinking co-IP (XL co-IP) has been used to stabilize and capture weak or transient interactions in a protein complex prior to LC-MS/MS analysis [64-66]. There are many chemical crosslinkers available, but formaldehyde has been extensively used because of its many advantages [65-67].

Formaldehyde is a very small molecule with a short spacer arm ranging from 2.3-2.7 Å [67], but formaldehyde crosslinks may actually be the dimerization product of two formaldehyde-modified amino acids (usually lysine or arginine), effectively doubling the crosslinking diameter [68]. Formaldehyde strongly stabilizes protein-protein interactions through covalent bonds between proteins and other molecules, which allows the samples to be washed in more stringent conditions for removal of non-specific binders compared to native co-IP conditions. Chemical crosslinking has a few limitations that should be considered when using this method for proteomic analysis. Chemical crosslinking can decrease protein solubility so it is important to optimize lysis conditions to insure that the target cross-linked protein complex is retained in the soluble fraction that will be used for co-IP [67]. Also, crosslinking proteins is dependent on the number and the position of the certain residues like lysine and arginine, therefore, there is a risk false negative results for real interactors that lack the necessary residues for crosslinking [66]. Protein crosslinking can also interfere with tryptic digestion of XL-co-IP samples, which may prevent non-digested peptides from being identified by LC-MS/MS [69]. Despite these limitations, XL-co-IP has a better chance of capturing transient kinase interactions than traditional co-IP.

In Apicomplexa, XL-co-IP and LC-MS/MS analysis has been used to identify PbCDPK4 and PbCDPK1 interactomes in P. berghei [70]. In malaria parasites, CDPK4 is an essential regulator of cell cycle progression during male gametogenesis, and subsequent transmission to the mosquito vector [70-73], while CDPK1 is involved in phosphorylation of multiple inner membrane complex proteins, suggesting a role in inner membrane complex (IMC) assembly or stability [74]. Furthermore, CDPK4 and CDPK1 functionally interact and can complement each other, working synergistically with protein kinase G (PKG), an essential kinase that controls key calcium signaling events in malaria parasites [75], to promote ookinete gliding and control asexual growth [70]. Interactome analysis of XL-co-IP PbCDPK4-3HA tagged parasites identified 19 interactors enriched over the WT control, including PbGAP40, PbMyoA, and PbGAC [70]. These proteins are associated with IMC biogenesis or parasite motility [76, 77], strongly suggesting that PbCDPK4 regulates the molecular machinery responsible for parasite motility, host cell invasion, and egress. To validate the putative interaction between CDPK4 and GAP40, reciprocal XL-co-IP of GAP40-3HA parasites identified CDPK4 among the 20 interactors, supporting the model that CDPK4 and GAP40 are interactors [70]. Indirect phosphoproteomics-based approaches have also been utilized to characterize CDPK4 function in Plasmodium sexual blood stages (further discussed in section 3.1 Label-free).

To define the interactome of PbPKG, XL-co-IP was used to capture PbPKG-3HA complexes and identify co-purifying proteins by LC-MS/MS [78]. The top hit was Important for Calcium Mobilization 1 (PbICM1), which was validated by reciprocal XL-co-IP and LC-MS/MS. Similar results were obtained in P. falciparum using standard co-IP and LC-MS/MS and further confirmed using a chemoproteomic approach [78]. P. falciparum protein extracts were incubated with kinobeads (sepharose beads conjugated broad spectrum kinase inhibitors), washed, and treated with ML10 (PKG inhibitor) to displace PfPKG and its interactors from the kinobeads. Based on LC-MS/MS of the kinobead assay, PfPKG and PfICM1 were both displaced by ML10, providing additional evidence of interaction [78]. Furthermore, indirect phosphoproteomic analysis identified PKG-dependent phosphosites on ICM1 in P. falciparum and P. berghei, suggesting that it is likely a substrate of PKG in multiple species of Plasmodium. Functionally, ICM1 was required for PKG-dependent Ca²⁺ mobilization in Plasmodium. This study nicely represents how kinase interactome approaches can identify candidate substrates to define mechanisms of kinase function in apicomplexan parasites.

In a recent study, XL-co-IP and LC-MS/MS analysis was applied to identify the aurora related kinase 2 (ARK2) interactome on the sexual stages of P. berghei parasites [79]. To better understand the function of PbARK2, interactome analysis was performed using XL-co-IP followed by LC-MS/MS analysis in PbARK2-GFP P. berghei gametocytes, identifying a network of microtubule associated proteins, such as PbEB1, PbMISFIT and PbMyoK [79]. These data suggests that PbARK2 may interact with these proteins to form a complex that is important for the spindle organization and chromosome segregation during cell division of the malaria parasite. This hypothesis is further supported by the fact that reciprocal XL-co-IP of EB1-GFP parasites also captured and identified ARK2 [79].

2.2.3. Kinase interactome outlook

Proximity labeling (BioID, TurboID, PerTurboID) and affinity purification (co-IP, XL-co-IP) strategies have been useful in defining apicomplexan kinase interactomes as potential substrates. However, other useful proximity labeling and affinity purification techniques could be adapted to define apicomplexan kinase interactomes. Such techniques include ascorbate peroxidase (APEX) or APEX2 proximity labeling, high throughput kinase pull down, and kinobead based assays.

APEX is a proximity labeling approach used to biotinylate proteins in close proximity to an APEX-tagged bait protein in live cells for proteomic analysis [80]. In the presence of hydrogen peroxide, APEX catalyzes the oxidation of biotin-phenol to reactive biotin-phenoxyl radicals, labeling nearby proteins for streptavidin capture and identification by LC-MS/MS [80]. APEX has a limited sensitivity when expressed at low levels, possibly due to protein misfolding or stability. A new mutant version of APEX with improved sensitivity was generated, named APEX2 [81]. This new version has been successfully applied to identify MAPKs interactome in cardiomyocytes [82], and secreted T. gondii effectors that modulates host cell immune response [83]. Therefore, APEX or APEX2 are feasible alternatives to BioID or TurboID for kinase proximity labeling in Apicomplexa.

Similar to co-IP, pull down assays are useful for investigating protein-protein interactions [84, 85]. Unlike co-IP, which relies on antibody capture, pull down assays rely on affinity tagged proteins (bait) bound to immobilized affinity ligands (affinity support) to capture interactors (prey). Captured prey can be detected by SDS-PAGE and analyzed by LC-MS/MS [84, 85]. This approach identified an interaction network of kinases involved in human diseases using a library of affinity tagged kinases as bait [86], and therefore may be applied to identify kinase interactors in Apicomplexa.

A newer approach known as kinobead competition and correlation analysis (kiCCA) has been developed for profiling kinase interactomes in cancer cell lines [87]. This high throughput method is based on the incubation of cell and tissue lysates with kinobeads, for enrichment of kinase along with its interactor partners [88, 89], and ATP-competitive kinase inhibitors, to displace the kinase-interactor complex from the kinobead [90, 91]. However, instead of using a single selective kinase inhibitor, a panel of multiple kinase inhibitors probes [88, 92, 93] is applied in the kiCCA approach for displacement of hundreds of kinase-associated protein complex that are further analyzed through LC-MS/MS [87]. Then, the intensity values for the kinases are analyzed using Pearson correlation, where the highest Pearson r value represents the most likely kinase interactor. This approach identified thousands of protein-protein interactions in different cancer cell lines [87]. Adaptation of kiCCA for use in apicomplexan parasites would facilitate high throughput multiplexed kinase interactome discovery.

3. Indirect phosphoproteomics

In addition to kinase interactome analysis, kinase substrates can be identified from indirect phosphoproteomics-based approaches and direct labeling approaches. The phosphoproteome is the subset of the total proteome that contains phosphorylated proteins [94]. Indirect phosphoproteomics based approaches can be used to measure kinase-dependent changes in the phosphoproteome to identify candidate kinase substrates. The approach is considered indirect because it cannot distinguish between primary and secondary phosphorylation events. A general strategy to identify putative kinase substrates through phosphoproteomics is to perturb the target kinase activity or expression in vivo using chemical or genetic approaches, then quantify changes in the phosphoproteome compared to the wild-type kinase control to assign specific kinase-dependent phosphosites. Usually, a bottom-up phosphoproteomics workflow strategy is applied, involving protein extraction from biological samples, protein digestion with a protease (usually trypsin), label-free or label-based quantitative strategies, phosphopeptide enrichment and fractionation, and LC-MS/MS analysis [95].

Strategies have been developed to improve accuracy and precision in protein and peptide quantification by mass spectrometry, including protein phosphorylation [96]. Among the label-based strategies, stable isotope labeling by amino acids in cell culture (SILAC), isobaric tags for relative and absolute quantification (iTRAQ), and tandem mass tag (TMT) has been extensively used in quantitative phosphoproteomics. Following the label-free or label-based strategy, the enrichment of the phosphorylated serine, threonine, or tyrosine by affinity approaches is a key step in quantitative phosphoproteomics analysis [97, 98]. Immobilized metal affinity chromatography (IMAC) [99] and metal oxide affinity chromatography (MOAC) have been extensively used in recent years for purifying phosphopeptides [100]. In IMAC, negatively charged phosphopeptides bind to positively charged metal ions such as Fe³⁺ or Zr⁴⁺, that can be immobilized using magnetic beads or silica resins to retain the phosphopeptides. In MOAC, titanium dioxide (TiO₂) is usually used to bind phosphopeptides in metal oxide matrixes. Furthermore, fractionation of phosphopeptides based on charge attraction or repulsion using strong cation exchange chromatography (SCX) [101], or strong anion exchange chromatography (SAX) [102], prior to or after the phosphopeptide enrichment step, can also be used to enhance phosphopeptide discovery. In the following section, we discuss the current quantitative phosphoproteomic strategies that have been used to identify putative kinase substrates in apicomplexan parasites [13, 75, 103].

3.1. Label-free

Label-free phosphoproteomics is useful for quantifying phosphoproteomes without the use of isotopic labels based on normalization methods. In this approach, proteins are extracted from cells or tissues (test vs control), enzymatically digested, enriched for phosphopeptide discovery and analyzed by LC-MS/MS. This method is popular in due to flexible experimental design, straightforward analysis with simple and fast workflow, and low cost when compared to labeling methods [104]. However, samples in label-free approach cannot be combined and analyzed simultaneously, which increases acquisition runtime to accommodate individual samples. Robust data acquisition and analysis software, such as MaxLFQ integrated in MaXQuant [105], and fast workflow have made the label-free an attractive approach for quantitative phosphoproteomics in apicomplexan parasites [72, 106].

Label-free phosphoproteomics has been used to study kinase signaling in P. falciparum. Label-free phosphoproteomics with IMAC enrichment identified 58 phosphorylation sites on 50 proteins that were significantly reduced after PfCDPK5 knockdown in P. falciparum schizonts [106]. PfCDPK5 is a key regulator of parasite egress since blood-stage parasites lacking PfCDPK5 develop into mature schizonts but become reversibly trapped within host cells [107]. Among the hits, novel putative transporter (NPT) PfNPT1 serine S259 residue was significantly less phosphorylated after PfCDPK5 knockdown, suggesting that PfNPT1 is a potential PfCDPK5 substrate in P. falciparum schizont. This interaction was further confirmed using an in vitro kinase assay, which is a direct labeling approach, and both co-localize at the plasma membrane [106]. Phosphorylation of PfNPT1 by PfCDPK5 may regulate parasite egress but warrants further testing. A similar approach was used to characterize the signaling network controlled by PfCDPK4 in P. falciparum gametocytes, a protein critical for male gametogenesis [72]. A total of 193 phosphosites in 170 proteins were quantified in WT parasites, but were not detected in parasites lacking PfCDPK4. Among the putative substrates of PfCDPK4, synthetic peptides from an uncharacterized protein (PF3D7_0417600), PfCDPK1, PfSOC3, PfSOC7, and PfPFK9 were directly phosphorylated by recombinant PfCDPK4 using an in intro kinase assay [72]. These PfCDPK4 substrates have important and diverse roles in Plasmodium gametocytogenesis. In P. berghei, PbSOC3 is involved in parasite exflagellation [73], while PfSOC7 is involved in DNA synthesis in P. falciparum [108]. A recent study found that PfPFK9 regulates glycolysis, representing a key source of energy for flagellar motion in male gametes [109]. In P. falciparum, PfCDPK1 is involved in parasite exflagellation [110], indicating that crosstalk between PfCDPK1 and PfCDPK4 may occur. These data support a model in which CDPK4 regulates Plasmodium gametocyte biology by phosphorylating multiple substrates.

The role of kinases in T. gondii has also been investigated using label-free phosphoproteomics. RON13 is a rhoptry-resident kinase and a key virulence factor in T. gondii that plays a crucial role in host cell invasion by phosphorylating components of the RON complex, stabilizing it at the moving junction [111]. Comparative phosphoproteomics of WT and TgRON13-KD parasites revealed a total of 63 phosphosites in 42 proteins as highly probable targets of TgRON13. TgRON4 plays a critical role in moving junction formation and was differently phosphorylated in parasites lacking TgRON13. Further analysis through in vitro kinase assays confirmed that TgRON13 directly phosphorylates TgRON4. These data suggest that phosphorylation of TgRON4 and other members of the RON complex, such as TgRON2/4/5/8 [112], is a crucial requirement for proper assembly of the moving junction to ensure host cell invasion [111]. Using a similar approach, substrates of TgCrk4 were identified T. gondii [113]. Ten Crks and seven atypical cyclins have been described in T. gondii, and several of them are required for cell cycle progression [37]. Specifically, TgCrk1, TgCrk2, TgCrk4 and TgCrk6 were found to be required or essential for T. gondii replication [37]. Label-free phosphoproteome analysis of TgCrk4-arrested tachyzoites identified a total of 1112 up- and 883 downregulated phosphosites from multiple proteins, including nucleic acid binding proteins, kinases, and IMC proteins [113]. The TgCrk4 label-free phosphoproteome analysis led to the identification of its substrates, including TgORC4, TgCdc20, TgGCP2 and TgPP2ACA [113]. Further analysis through co-IP and LC-MS/MS demonstrated that TgCrk4 and TgCyc4 form a complex to regulate the G2 phase in T. gondii, and this complex interacts with TgiRD1, a protein that controls DNA and centrosome reduplication in tachyzoites [113]. Taken together, label-free technologies have been successfully used to define kinase-dependent phosphoproteomes in multiple apicomplexan parasites.

3.2. SILAC

Label-based approaches allow for improved quantification of proteomic and phosphoproteomic approaches. Stable isotope labeling with amino acids in cell culture (SILAC) is a straightforward metabolic labeling approach that allows the relative and absolute sample quantification by LC-MS/MS [114]. One cell population grows in culture media containing “light” (normal) amino acids, while the other grows in culture media containing “heavy” (isotopes) amino acids. After a certain period of cell divisions or passages, “heavy” amino acids such as arginine or lysine (¹³C instead of ¹²C, or ¹⁵N instead of ¹⁴N) are incorporated into peptides leading to a known mass shift when compared to the normal “light” amino acids. Samples from both populations are then mixed in a 1:1 ratio and analyzed by LC-MS/MS for protein identification and quantification [114, 115]. Each peptide appears as a pair in the mass spectra, and as the light and heavy amino acids are chemically identical, except by their mass, the ratio of peak intensity between the light and heavy peptides reflects the relative protein abundance between the two cell populations. One of the main advantages of using SILAC in quantitative phosphoproteomics is its application at more upstream level (i.e. labeling in cell culture), when compared to chemical labeling (i.e. labeling after protein digestion), reducing false positive hits due to experimental errors in downstream steps [115]. SILAC also demands attention regarding the cell culture and the type of amino acid used for the metabolic labeling, as arginine can be metabolic converted to proline via the arginase pathway in some cell lines, compromising the data acquisition and quantification accuracy [116]. Furthermore, limitations regarding the cell line also need to be considered, as some cell lines do not divide in culture or may not be stable in SILAC media.

SILAC has been successfully applied in quantitative phosphoproteomics to identify downstream targets of kinases in apicomplexan parasites by comparing the changes in phosphorylation signaling events between WT and kinase perturbed parasites [26, 117]. SILAC-based quantitative phosphoproteomics was used to identify TgWNG1 substrates in T. gondii, a kinase that localizes to the parasitophorous vacuole (PV) lumen and is critical for the PV intravacuolar network biogenesis through phosphorylation of resident proteins [117]. A total of 10 proteins had downregulated phosphosites in TgWNG1 KO parasites when compared to WT parasites, including TgGRA2, TgGRA6, and TgGRA7. The TgGRA2 and TgGRA6 proteins are key players in the intravacuolar network morphology [118]. In vitro kinase assay using recombinant TgWNG1 confirmed that TgGRA2, TgGRA6, and TgGRA7 are direct substrates of TgWNG1 [117].

Treeck and colleagues used SILAC-based quantitative phosphoproteomics and SCX/IMAC for phosphopeptide enrichment, to identify phosphoregulatory events dependent on TgCDPK3 [26]. As described earlier, TgCDPK3 is an essential kinase that controls microneme secretion and host cell egress in T. gondii [23-25]. By comparing WT and TgCDPK3-KO parasites, TgCDPK3-dependent phosphosites were identified in TgCDPK1, TgCDPK2a, a putative calmodulin (TGGT1_042450) and a putative calcium-transporting ATPase (TGGT1_103910), providing strong evidence that TgCDPK3 controls calcium-regulated processes [26]. The only member of the glideosome with significant TgCDPK3-dependent phosphorylation was TgMyoA as discussed earlier (section 2.1.1 BioID). Taken together, SILAC has been shown to facilitate quantitative phosphoproteomics in multiple apicomplexan species.

3.3. iTRAQ

Isobaric tag for relative and absolute quantification (iTRAQ) is a technique developed by AB SCIEX company that uses isobaric reagents for protein identification and quantification by LC-MS/MS [119]. The isobaric tag is composed of a reporter group, a balance group, and an amine-reactive group. The reporter group contains a specific mass for peptide quantification. The balance group contains a mass to make all tags isobaric. The amine-reactive group covalently attaches the tag to the peptide which upon fragmentation, releases low mass reporter ions. The iTRAQ approach is based on protein labeling by isobaric tags in the N-terminus and side chain amine groups, where peptides can be distinguished based on the reporter ion mass and quantified based on the reporter ion intensity. One of the main advantages of iTRAQ is sample multiplexing, allowing high-throughput quantitative proteomics [119]. Furthermore, as iTRAQ is used at peptide level, it can be applied to all biological samples which allow flexibility in the experimental design using well-established labeling kits. When compared to label-free methods, iTRAQ may lead to more variations in protein sample caused by sample manipulation during the protein digestion process when compared to label-free or SILAC methods. Furthermore, the number of identified proteins and reproducibility may be lower in iTRAQ approach when compared to label-free methods [120], as well as more expensive and time consuming.

The use of iTRAQ and phosphopeptide enrichment with TiO₂ combined with LC-MS/MS successfully identified potential PfCDPK1 and PfCDPK7 substrates in P. falciparum [74, 121]. Conditional knockdown of PfCDPK1 using an FKBP destabilization domain system and iTRAQ identified 79 phosphosites from 64 proteins were downregulated after PfCDPK1 knockdown, including PfGAP45 and PfIMC1g [74]. In vitro kinase assay using recombinant PfCDPK1 further confirmed that GAP45, PfIMC1g and PfPKA-R are direct substrates of PfCDPK1, which may represent mechanisms by which PfCDPK1 controls invasion of host erythrocytes [74].

iTRAQ was also used to identify potential PfCDPK7 substrates [121], a critical kinase for P. falciparum asexual development [122]. iTRAQ quantitative phosphoproteomics in PfCDPK7-KO parasites revealed that 82 phosphosites from 68 proteins were hypophosphorylated, including phosphoethanolamine-N-methyltransferase on serine S49 residue and ethanolamine kinase on serine S37 residue, indicating a potential role of PfCDPK7 regulating phosphatidylcholine synthesis in Plasmodium [121].

3.4. TMT

Tandem mass tags (TMTs) are isobaric chemical tags that were first introduced by Thompson et al. [123] for identification and quantification of peptides by LC-MS/MS. TMTs general structure is composed of an amine reactive group, a cleavable unique mass reporter and a fragmentation-vulnerable spacer arm. Upon fragmentation in LC-MS/MS, each tag generates a low molecular mass ion reporter that is used for relative peptide quantification, as the intensity of the ion reporter represents the relative abundance of the tagged molecule. This method allows high-throughput analysis by multiplexing up to 18 samples in a single LC-MS/MS run, reducing instrument time, cost per sample, and experiment variability. Furthermore, as the chemical tags are isobaric and chemically identical, this ensures that identical peptides labeled with different tags will have the same elution profile and the same ionization condition in the LC-MS/MS run, reducing data distortion. Systematic comparison between different quantitative phosphoproteomic approaches have shown that TMT labeling have the highest precision in phosphopeptide quantification when compared to label-free and SILAC methods [124]. Despite the greater advantages, TMT labeling cannot be applied in vivo and is usually more expensive than other quantitative proteomic methods, such as label-free.

This approach has been increasingly used in large-scale phosphoproteomic profiling of complex peptide mixtures [124], and has also allowed the identification of putative kinase substrates in apicomplexan parasites [22, 40, 103, 125-128]. By using TMT and phosphopeptide enrichment with IMAC/TiO₂, Alam et al. [103] identified PfPKG substrates in Plasmodium schizonts by comparing WT and PfPKG_T618Q parasites treated with compound 2, a selective PKG inhibitor [129]. PfPKG_T618Q parasites harbors a substitution of threonine gate-keeper with a bulkier glutamine residue, reducing PKG sensitivity to compound 2 by 2,000-fold [130]. Therefore, by comparing the phosphoproteomes of WT and PfPKG_T618Q parasites treated with compound 2, PKG-dependent phosphosites can be assessed between the two samples. Phosphoproteomics identified 107 phosphorylation sites on 69 proteins that were dependent on PfPKG activity, including the serine S64 residue on PfCDPK1. An in vitro kinase assay using recombinant PfPKG and a catalytic inactive version of PfCDPK1 showed that PfPKG directly phosphorylates S64 of PfCDPK1, and this phosphorylation is important for maintaining PfCDPK1 as a high molecular weight complex [103]. Similarly, TMT-based phosphoproteomics has been used to define Ca²⁺- and cGMP-dependent phosphoproteomes in T. gondii at sub-minute resolution [13, 131].

TMT-based phosphoproteomics has also been employed to investigate the function of several kinases in T. gondii. For instance, loss of TgCDPK7 modulated the abundance of 301 phosphosites on 215 proteins based on TMT-based phosphoproteomics [125]. Interestingly, TgRab11a possessed TgCDPK7-dependent phosphorylation, a protein previously shown to be involved in vesicular trafficking [132]. An in vitro kinase assay demonstrated that TgRab11a S207 is directly phosphorylated by TgCDPK7. Furthermore, this phosphosite was shown to be important for its localization in a punctate endosome-like compartment [125].

TMT-based phosphoproteomics was also applied to gain insights into the pathways controlled by TgPKA, a cAMP-dependent kinase, in T. gondii [133]. A total of 191 proteins had TgPKA-dependent phosphosites, including proteins thought to regulate cell cycle, as well as 83 proteins with no functional annotation [133]. Interestingly, inhibition of TgPKA increased phosphorylation of phosphodiesterase 2 (TgPDE2), which degrades cAMP, suggesting a feedback loop exists between TgPKA and cAMP levels [133]. TgPDE2 has been shown to regulate several steps of the lytic cycle, likely by modulating TgPKA activity through cAMP regulation [134, 135]. Further analysis showed that TgPKAc1 acts as a balancing regulator of cGMP metabolism to control T. gondii egress by acting as a repressor of PKG-mediated signaling and downstream Ca²⁺ signaling [133, 136]. In addition to cAMP regulation, TgPKA may also be regulated by kinase phosphorylation. A recent study investigated the role of Store Potentiating/Activating Regulatory Kinase (SPARK) in regulating other kinases in T. gondii [137]. SPARK is an ortholog of the phosphoinositide-dependent protein kinase 1, a protein that in metazoans regulates the second messenger signaling through the regulation of AGC kinases [138]. In T. gondii, recent studies revealed that TgSPARK has a key role in host cell invasion and egress by regulating the calcium release from intracellular stores [139]. In Plasmodium, an ortholog of SPARK called 3-phosphoinositide–dependent protein kinase-1 (PfPDK1) plays a key role in host cell invasion and parasite proliferation by regulating the PfPKA pathway [140], suggesting that TgSPARK may also regulate TgPKA in T. gondii. TMT-based proteomics and phosphoproteomics revealed 138 phosphopeptides were downregulated after TgSPARK knockdown compared to untreated controls [137]. Enrichment analysis of the TgSPARK-dependent phosphopeptides revealed that TgSPARK may regulate AGC kinases (including TgPKA and TgPKG), phosphatases, ubiquitin ligases, nucleic acid binding proteins, and transporters [137]. Co-IP and TurboID interactome analyses also revealed further interactors and consequently the pathways that TgSPARK might be regulating in T. gondii [137]. To gain further insights into how TgSPARK regulates AGC kinases in T. gondii, the TgPKG-, TgPKAc1-, and TgPKAc3-dependent phosphoproteomes were also analyzed using TMT-based phosphoproteomics [137]. By comparing these kinase-dependent phosphoproteomes, a model was proposed that TgSPARK regulates the stability and activity TgPKG and TgPKA [137].

3.5. Indirect phosphoproteomics outlook

Large-scale analysis of protein phosphorylation through indirect phosphoproteomics is a powerful tool for identifying apicomplexan kinase substrates. As the name suggests, the main limitation of indirect phosphoproteomics is discerning direct kinase targets from indirect kinase targets. New technologies have been developed that can help discern direct targets in phosphoproteomic data.

Kinase assay linked with phosphoproteomics (KALIP) is an approach that combines in vitro kinase reaction with kinase-dependent phosphoproteomics [141]. For this approach, it is important that the kinase and substrates are derived from the same cell type. First, an in vitro kinase assay is performed using a purified kinase of interest and dephosphorylated peptides derived from cell lysates as substrates. The resulting phosphopeptides are enriched using polymer based metal ion affinity capture (PolyMac) [142] and identified by LC-MS/MS. Second, the in vivo kinase of interest-dependent phosphoproteome (kinase+ vs kinase−) is defined by indirect phosphoproteomics analysis. Phosphopeptides enriched in both sets of experiments are more likely to be direct substrates of the kinase of interest. This approach successfully identified 64 candidate substrates of spleen tyrosine kinase in B cells and 23 in breast cancer cells [141]. Thus, KALIP represents a new approach that should be readily adaptable for kinase substrate discovery in Apicomplexa.

4. Direct labeling

Kinase interactome and phosphoproteomics can reveal candidate kinase substrates, especially when combined. The third main approach to identifying kinase substrates is to screen for proteins that are directly labeled (phosphorylated) by a kinase of interest.

4.1. In vitro kinase assay

The in vitro kinase assay is the simplest and most straightforward direct labeling approach to identify kinase substrates or phosphorylation motifs [143, 144]. Substrate phosphorylation demands a direct interaction between the kinase and the substrate protein or protein complex [145-147]. In a typical in vitro kinase assay, the recombinant purified kinase and potential substrate are incubated with radio-labeled ATP (γ-32P), and the transfer of the γ-32P from ATP to the substrate can be visualized through autoradiogram or quantified using a scintillation counter. Despite the in vitro kinase assay being considered a gold standard for kinase substrate identification, the use of radio-labeled ATP requires proper handling of the samples and further correct disposal of radioactive material. Furthermore, artificial labeling of substrates may occur by using high concentration of purified kinase/substrate in the reaction [148-150].

Several apicomplexan kinases have evolved to phosphorylate host proteins to promote parasite survival and virulence. For instance, certain rhoptry (ROP) kinases/pseudokinases are injected into host cells during invasion and modify host proteins to inactivate immune defenses and modulate host cell transcription [11, 151-154]. TgROP16 is a polymorphic tyrosine kinase and a key virulence factor that plays a crucial role in host cell immunity evasion by phosphorylating host STATs [155-158]. TgROP16 traffics to the host nucleus and directly phosphorylates host STAT3 at tyrosine Tyr705 residue [155] and STAT6 at tyrosine Tyr641 residue [156], as confirmed using in vitro kinase assays. Phosphorylation of STAT3/6 by TgROP16 plays a key role in host immune response evasion by suppressing proinflammatory cytokines [154] and controlling parasite replication by induction of arginase-1 in the host [158].

In vitro kinase assays also provided crucial findings regarding the substrates of T. gondii ROP18 kinase. TgROP18 is a serine/threonine kinase injected in the host cell during invasion that forms a complex with TgROP5, a pseudokinase that controls T. gondii virulence by regulating TgROP18 activity [159]. In virulent T. gondii type I strains, both TgROP5 and TgROP18 interacts and decorates the parasitophorous vacuole membrane to phosphorylate immunity-related GTPases (IRGs) [160, 161], important effectors induced by IFN-γ that restrict T. gondii growth [162-167]. In vitro kinase assays and immunoprecipitation experiments showed that TgROP18 binds to and directly phosphorylates IRGs family members, such as Irgb6, Irga6 and Irgb10, but not when TgROP18 was mutated to a kinase-dead version [160]. This data confirmed that TgROP18 is a key virulence factor by blocking the accumulation of IRGs family members on the parasitophorous vacuole membrane and subsequent parasite clearance. In a different study, a western blot using anti-pT102 and anti-pT108 antibodies for Irga6 in parallel with in vitro kinase assay demonstrated that TgROP18 directly phosphorylates Irga6, confirming the crucial of TgROP18 in host immunity evasion [161].

In vitro kinase assays have been used in a number of Plasmodium kinase studies. For instance, these assays have been used to validate the target Plasmodium PKA, which plays a key role in merozoite invasion in human erythrocytes [168, 169]. In vitro kinase assays confirmed that PfPKA directly phosphorylates S610 in the cytoplasmic tail of PfAMA1, an event that is essential for red blood cells invasion [169]. Overall, direct-labeling approaches through in vitro kinase assays have been extremely useful to identify putative kinase substrates in Apicomplexa, revealing important aspects of parasite biology [170-179].

4.2. Analog-sensitive kinases

Eukaryotic protein kinases are highly conserved in their catalytic site structure and basically all use the same phosphorylation mechanism to transfer the γ-phosphate of ATP to a substrate [180]. Allen and colleagues developed an innovative method that uses a genetically engineered analog-sensitive (AS) kinase and atypical ATP analogs to identify kinase substrates [181]. AS kinases are genetically engineered to harbor a mutation in the ATP-binding pocket creating a new active site that allows the AS kinase to accept bulky side chains of ATP analogs. This creates a “lock and key” system that allows only AS kinase to use ATP analogs to phosphorylate substrates, creating an unbiased labeling of direct targets of the kinase on a global scale. A variety of kinase substrate identification methods using AS kinases and ATP analogs have been described [182], including the use of an ATP analog that carries a γ-thiophosphate group (ATP-γ-S). The thiophosphate group works as a unique phosphate mimetic tag that is resistant to dephosphorylation by serine/threonine phosphatases [183, 184] and can be covalent captured by iodoacetyl-agarose beads (Sulfolink) or affinity purified using specific antibodies to reveal direct targets of kinases by LC-MS/MS [181, 185-187]. One disadvantage of this approach is that it is not universally applicable, as not all kinases can be mutated to efficiently use ATP-γ-S, limiting its applicability [185]. Furthermore, as thiophosphate group is resistant to phosphatases activity, its application in vivo can be problematic caused by cell toxicity.

The AS kinase approach has been used successfully to identify TgCDPK1 substrates in T. gondii by comparing TgCDPK1 WT and conditional knockout parasites [188]. Two proteins were found to be exclusively thiophosphorylated in TgCDPK1 WT parasites: dynamin-related protein B and a phosphorylated repeat protein (TGME49_205320, formerly TGME49_005320), suggesting that both substrates are phosphorylated in a TgCDPK1-dependent manner [188]. This approach was further developed to identify additional targets of TgCDPK1 by thiophosphorylation [131]. In total, 104 proteins were identified as direct substrates of TgCDPK1 including TgHOOK which regulates microneme secretion and host cell invasion, partially recapitulating the function of TgCDPK1 [131].

4.3. Peptide library and microarray

Protein kinase assays using peptide and microarray libraries allows a straightforward screening of putative kinase substrates by incubating the purified kinase of interest with a library of putative substrates [189, 190]. Different library types have been developed over the last decades for large-scale analysis of kinase substrates and can be summarized in two general groups: knowledge-based libraries and random libraries [189]. The knowledge-based libraries are generated using natural occurring proteins and allows the identification of the phospho-acceptor residue by scanning overlapping peptides or by use of sequences that only surrounds the potential phospho-acceptor residue [189, 191]. Random libraries consist of arbitrarily generated single or mixed peptides by combinatorial approaches and are particularly convenient for use with orphan kinases which the orthologs are not known [189, 192]. Peptide libraries can be immobilized on a solid support to build high-density peptide microarrays that allows substrate screening in a high throughput manner, identifying protein phosphorylation and motif preference of a target kinase [193]. Using small amounts of reagents, the kinase of interest is incubated with the arrayed peptide library and phosphorylated peptides scanned by distinct detection methods such as radiolabeling, immunoblotting, and fluorescence [194]. Despite the many advantages for kinase substrate discovery, there are important limitations regarding this approach that need to be considered. The use of concentrated purified kinase for the in vitro reaction may led to artificial labeling of substrates, which may reduce reaction specificity [148-150]. Furthermore, as the kinase-substrate reaction occurs in vitro, some important regulatory mechanisms that occurs physiologically may be lost [149, 150]. Therefore, is critical to confirm that the putative substrate found in vitro is physiologically relevant through in vivo studies [195].

Positional scanning peptide library [196, 197] approaches have been extensively used in Apicomplexa to determine kinase substrates and its preferred consensus motif. For TgROP18 in T. gondii, an optimal substrate phosphorylation motif was determined, and shows a broad sequence selectivity that can be generalized as (X)-(X, not E)-(X)-(E)-(H)-(T)-(R/mixed, not P and not negatively charged)-(Ar)-(Ar)-(Ar), where the phospho-acceptor is underlined, X denotes any amino acid residue, and Ar denotes an aromatic residue [198]. This approach was also used to explore the peptide preference of TgCDPK1, which showed a strong preference for serine over threonine, for arginine at the −3 position, and for hydrophobic residues at the −5 position [188]. In T. gondii, TgROP17 is a serine/threonine kinase that accumulates in the external face of the parasitophorous vacuole membrane (PVM) and forms a complex with TgROP5 to phosphorylates IRGs, therefore, developing a crucial role in parasite pathogenesis [199]. Positional scanning peptide library analysis showed that TgROP17 has a strong preference for threonine over serine as phospho-acceptor, with a slight preference for hydrophobic amino acid surrounding the phosphorylation site [199]. Furthermore, an in vitro kinase assay using recombinant Irgb6 and Irga6 showed that TgROP17 prefers Irgb6 as substrate, and phosphorylates Irgb6 at threonine Thr102 residue, as demonstrated by semiquantitative LC-MC/MS [199]

In Plasmodium spp., the positional-scanning peptide library approach identified the PfPKG substrate specificity, that showed a strong preference for basic residues, in particular, Arg, at the −5 and −3 positions relative to the phosphorylation site, basic residues at −2 position and hydrophobic residues at the +1 position [200]. Furthermore, PfPKG generally deselect acidic residues at both upstream and downstream of the phospho-acceptor residue and display a strong preference for threonine over serine [200]. Further analysis by in vitro kinase assay showed that PfPKG directly phosphorylates RPT1, a subunit of the 19S proteasome and PBANKA_090940, a hypothetical protein [200]. In Plasmodium, PfPK7 is an “orphan” kinase with no human homologs [201]. Using peptide microarray library, Merckx and colleagues identified the PfPK7 consensus peptide substrate sequence as (r-R-R/K-K/R-S/T-P-K/R-K-R).

4.4. Direct labeling outlook

We have provided examples of how direct labeling assays have been successfully used to identify kinases substrates in Apicomplexa [188, 198, 200, 202]. Additional direct labeling approaches could also prove useful for identifying kinase substrates in Apicomplexa. For instance, the kinase-catalyzed biotinylation with inactivated lysates for discovery of substrates (K-BILDS) has been used to identify kinase substrates in cell lysates [203]. In this method, kinases in lysates are inactivated with FSBA, a covalent inhibitor that reacts with lysine in the kinase active site and inhibits the kinase activity irreversibly [204]. Next, FSBA is removed, and exogenous active kinase of interest is added with ATP-biotin in the lysates, biotinylated proteins are enriched with streptavidin resin, separated by SDS-PAGE and analyzed by LC-MS/MS [203]. As proof of concept, this approach was applied for discovery of PKA substrates in HeLa cells lysates, identifying 279 candidate substrates, from which 56 was previously known PKA substrates [203]. The fact that this approach can be used to basically all kinases [205] and the putative substrates can be analyzed by high-throughput LC-MS/MS, makes this an promising approach for kinase substrate identification in Apicomplexa.

Another useful direct labeling approach is called Kinase-interacting substrate screening (KISS) [206]. In this approach, cell or tissue lysates containing affinity beads coated with a GST-tagged catalytic fragment of the kinase of interest are loaded through a glutathione-sepharose affinity column. Kinase-bound proteins are incubated with ATP/Mg²⁺ (phosphorylated samples) or without ATP/Mg²⁺ (non-phosphorylated samples) in a reaction mixture to promote on-bead phosphorylation, followed by tryptic digestion. Subsequently, phosphopeptides are enriched using titanium oxide column and analyzed by LC-MS/MS [206].

Using KISS, Amano and colleagues [206] identified 356 phosphorylation sites in 140 proteins as candidate substrates of Rho-associated kinase, including previously known and new candidates. Furthermore, the KISS method was also applied to other kinases, efficiently identifying its substrates and phosphorylation sites, confirming it broad range applicability. This approach may represent a new interesting and broad applicable method for kinase substrate discovery in apicomplexan parasites.

5. Conclusions

Protein phosphorylation is one of the most well studied post-translational modifications, and controls many aspects of cell biology [1]. Advances in proteomic technologies have facilitated apicomplexan kinase substrate identification. Here, we have summarized and provided examples of the three categories of proteomic approaches for apicomplexan kinase substrate discovery and validation: i) kinase interactome, ii) indirect phosphoproteomics, and iii) direct labeling. New technologies emerge each year that make kinase substrate identification even more robust, efficient, and feasible. Each approach has advantages and limitations that affect kinase substrate identification. Several studies have combined two or more of these approaches to enhance confidence in apicomplexan kinase substrate identification [73, 78, 131]. Kinase substrate screens can generate dozens of candidate substrates, so prioritization is key. For instance, fitness screens can help determine which candidate substrates are essential or likely to as important downstream mediators of the kinase of interest [207-209]. It is also important to validate the putative kinase substrates identified using in vitro kinase assays of purified kinases and substrates. Once direct substrates are confirmed, mutational analysis of the phosphorylation site through genetic complementation is useful for determining its biological significance in vivo [21, 103, 106, 121, 125]. Identifying the critical post-translational modifications from each essential apicomplexan kinase will reveal regulatory features of essential substrates that could be targeted for novel therapeutic intervention.

Highlights.

• Multiple mass spectrometry-based proteomic approaches are currently available for protein kinase substrate discovery in Apicomplexa

• Kinase interactome analysis using proximity labeling, co-immunoprecipitation, or pull down can reveal candidate kinase substrates and potential regulators

• Indirect phosphoproteomics using label free or label-based technologies can reveal kinase-dependent phosphorylation sites

• Direct labeling approaches such as analog sensitive kinases can identify direct kinase substrates